{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.5608481868469576,
  "eval_steps": 500,
  "global_step": 7300,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00015365703749231714,
      "grad_norm": 54.03105926513672,
      "learning_rate": 1.9999385371850035e-05,
      "loss": 5.8486,
      "step": 2
    },
    {
      "epoch": 0.00030731407498463427,
      "grad_norm": 60.26078414916992,
      "learning_rate": 1.999877074370006e-05,
      "loss": 5.5476,
      "step": 4
    },
    {
      "epoch": 0.00046097111247695143,
      "grad_norm": 172.71730041503906,
      "learning_rate": 1.9998156115550094e-05,
      "loss": 4.8213,
      "step": 6
    },
    {
      "epoch": 0.0006146281499692685,
      "grad_norm": 80.93025970458984,
      "learning_rate": 1.9997541487400124e-05,
      "loss": 4.8849,
      "step": 8
    },
    {
      "epoch": 0.0007682851874615857,
      "grad_norm": 118.46244049072266,
      "learning_rate": 1.9996926859250153e-05,
      "loss": 4.0143,
      "step": 10
    },
    {
      "epoch": 0.0009219422249539029,
      "grad_norm": 58.34870910644531,
      "learning_rate": 1.9996312231100187e-05,
      "loss": 4.0411,
      "step": 12
    },
    {
      "epoch": 0.0010755992624462201,
      "grad_norm": 35.50437927246094,
      "learning_rate": 1.9995697602950216e-05,
      "loss": 3.9182,
      "step": 14
    },
    {
      "epoch": 0.001229256299938537,
      "grad_norm": 60.75965118408203,
      "learning_rate": 1.999508297480025e-05,
      "loss": 3.7527,
      "step": 16
    },
    {
      "epoch": 0.0013829133374308542,
      "grad_norm": 53.328765869140625,
      "learning_rate": 1.999446834665028e-05,
      "loss": 3.5424,
      "step": 18
    },
    {
      "epoch": 0.0015365703749231714,
      "grad_norm": 40.73623275756836,
      "learning_rate": 1.999385371850031e-05,
      "loss": 3.7179,
      "step": 20
    },
    {
      "epoch": 0.0016902274124154886,
      "grad_norm": 15.511174201965332,
      "learning_rate": 1.9993239090350342e-05,
      "loss": 3.3177,
      "step": 22
    },
    {
      "epoch": 0.0018438844499078057,
      "grad_norm": 62.24359130859375,
      "learning_rate": 1.9992624462200368e-05,
      "loss": 3.5207,
      "step": 24
    },
    {
      "epoch": 0.001997541487400123,
      "grad_norm": 69.20399475097656,
      "learning_rate": 1.99920098340504e-05,
      "loss": 3.6051,
      "step": 26
    },
    {
      "epoch": 0.0021511985248924403,
      "grad_norm": 39.98881530761719,
      "learning_rate": 1.999139520590043e-05,
      "loss": 3.3193,
      "step": 28
    },
    {
      "epoch": 0.0023048555623847574,
      "grad_norm": 10.113142013549805,
      "learning_rate": 1.999078057775046e-05,
      "loss": 3.2187,
      "step": 30
    },
    {
      "epoch": 0.002458512599877074,
      "grad_norm": 28.31175422668457,
      "learning_rate": 1.9990165949600494e-05,
      "loss": 3.4458,
      "step": 32
    },
    {
      "epoch": 0.0026121696373693913,
      "grad_norm": 21.829612731933594,
      "learning_rate": 1.9989551321450523e-05,
      "loss": 3.5843,
      "step": 34
    },
    {
      "epoch": 0.0027658266748617085,
      "grad_norm": 18.00796127319336,
      "learning_rate": 1.9988936693300556e-05,
      "loss": 3.2827,
      "step": 36
    },
    {
      "epoch": 0.0029194837123540257,
      "grad_norm": 16.2840576171875,
      "learning_rate": 1.9988322065150586e-05,
      "loss": 3.2059,
      "step": 38
    },
    {
      "epoch": 0.003073140749846343,
      "grad_norm": 11.987384796142578,
      "learning_rate": 1.9987707437000616e-05,
      "loss": 3.1313,
      "step": 40
    },
    {
      "epoch": 0.00322679778733866,
      "grad_norm": 6.873617649078369,
      "learning_rate": 1.998709280885065e-05,
      "loss": 2.8568,
      "step": 42
    },
    {
      "epoch": 0.003380454824830977,
      "grad_norm": 6.603146076202393,
      "learning_rate": 1.998647818070068e-05,
      "loss": 3.0781,
      "step": 44
    },
    {
      "epoch": 0.0035341118623232943,
      "grad_norm": 25.308164596557617,
      "learning_rate": 1.9985863552550708e-05,
      "loss": 3.0764,
      "step": 46
    },
    {
      "epoch": 0.0036877688998156115,
      "grad_norm": 15.176654815673828,
      "learning_rate": 1.998524892440074e-05,
      "loss": 3.0356,
      "step": 48
    },
    {
      "epoch": 0.0038414259373079286,
      "grad_norm": 7.444390773773193,
      "learning_rate": 1.9984634296250767e-05,
      "loss": 2.9488,
      "step": 50
    },
    {
      "epoch": 0.003995082974800246,
      "grad_norm": 18.565139770507812,
      "learning_rate": 1.99840196681008e-05,
      "loss": 2.7179,
      "step": 52
    },
    {
      "epoch": 0.004148740012292563,
      "grad_norm": 10.658416748046875,
      "learning_rate": 1.998340503995083e-05,
      "loss": 2.716,
      "step": 54
    },
    {
      "epoch": 0.0043023970497848806,
      "grad_norm": 9.682657241821289,
      "learning_rate": 1.9982790411800863e-05,
      "loss": 2.9189,
      "step": 56
    },
    {
      "epoch": 0.004456054087277198,
      "grad_norm": 20.967639923095703,
      "learning_rate": 1.9982175783650893e-05,
      "loss": 2.8078,
      "step": 58
    },
    {
      "epoch": 0.004609711124769515,
      "grad_norm": 16.931556701660156,
      "learning_rate": 1.9981561155500923e-05,
      "loss": 2.837,
      "step": 60
    },
    {
      "epoch": 0.004763368162261831,
      "grad_norm": 12.055686950683594,
      "learning_rate": 1.9980946527350956e-05,
      "loss": 2.7392,
      "step": 62
    },
    {
      "epoch": 0.004917025199754148,
      "grad_norm": 7.959167957305908,
      "learning_rate": 1.9980331899200985e-05,
      "loss": 2.8915,
      "step": 64
    },
    {
      "epoch": 0.0050706822372464655,
      "grad_norm": 9.24318790435791,
      "learning_rate": 1.9979717271051015e-05,
      "loss": 2.5171,
      "step": 66
    },
    {
      "epoch": 0.005224339274738783,
      "grad_norm": 20.02304458618164,
      "learning_rate": 1.9979102642901048e-05,
      "loss": 2.7947,
      "step": 68
    },
    {
      "epoch": 0.0053779963122311,
      "grad_norm": 8.09688663482666,
      "learning_rate": 1.9978488014751078e-05,
      "loss": 2.7323,
      "step": 70
    },
    {
      "epoch": 0.005531653349723417,
      "grad_norm": 8.636987686157227,
      "learning_rate": 1.9977873386601108e-05,
      "loss": 2.6309,
      "step": 72
    },
    {
      "epoch": 0.005685310387215734,
      "grad_norm": 6.815808296203613,
      "learning_rate": 1.997725875845114e-05,
      "loss": 2.4741,
      "step": 74
    },
    {
      "epoch": 0.005838967424708051,
      "grad_norm": 7.532662868499756,
      "learning_rate": 1.997664413030117e-05,
      "loss": 2.5875,
      "step": 76
    },
    {
      "epoch": 0.0059926244622003685,
      "grad_norm": 6.733164310455322,
      "learning_rate": 1.99760295021512e-05,
      "loss": 2.6103,
      "step": 78
    },
    {
      "epoch": 0.006146281499692686,
      "grad_norm": 6.442116737365723,
      "learning_rate": 1.997541487400123e-05,
      "loss": 2.6208,
      "step": 80
    },
    {
      "epoch": 0.006299938537185003,
      "grad_norm": 6.882765769958496,
      "learning_rate": 1.9974800245851263e-05,
      "loss": 2.5554,
      "step": 82
    },
    {
      "epoch": 0.00645359557467732,
      "grad_norm": 6.64527702331543,
      "learning_rate": 1.9974185617701292e-05,
      "loss": 2.5667,
      "step": 84
    },
    {
      "epoch": 0.006607252612169637,
      "grad_norm": 7.69775390625,
      "learning_rate": 1.9973570989551322e-05,
      "loss": 2.6117,
      "step": 86
    },
    {
      "epoch": 0.006760909649661954,
      "grad_norm": 7.077218532562256,
      "learning_rate": 1.9972956361401355e-05,
      "loss": 2.4501,
      "step": 88
    },
    {
      "epoch": 0.0069145666871542714,
      "grad_norm": 5.539189338684082,
      "learning_rate": 1.9972341733251385e-05,
      "loss": 2.4775,
      "step": 90
    },
    {
      "epoch": 0.007068223724646589,
      "grad_norm": 6.602914333343506,
      "learning_rate": 1.9971727105101415e-05,
      "loss": 2.3944,
      "step": 92
    },
    {
      "epoch": 0.007221880762138906,
      "grad_norm": 5.995626449584961,
      "learning_rate": 1.9971112476951448e-05,
      "loss": 2.4826,
      "step": 94
    },
    {
      "epoch": 0.007375537799631223,
      "grad_norm": 6.836587429046631,
      "learning_rate": 1.9970497848801477e-05,
      "loss": 2.468,
      "step": 96
    },
    {
      "epoch": 0.00752919483712354,
      "grad_norm": 6.4697651863098145,
      "learning_rate": 1.9969883220651507e-05,
      "loss": 2.2833,
      "step": 98
    },
    {
      "epoch": 0.007682851874615857,
      "grad_norm": 8.081903457641602,
      "learning_rate": 1.996926859250154e-05,
      "loss": 2.6544,
      "step": 100
    },
    {
      "epoch": 0.007836508912108174,
      "grad_norm": 6.688724517822266,
      "learning_rate": 1.996865396435157e-05,
      "loss": 2.55,
      "step": 102
    },
    {
      "epoch": 0.007990165949600492,
      "grad_norm": 6.878283977508545,
      "learning_rate": 1.99680393362016e-05,
      "loss": 2.2658,
      "step": 104
    },
    {
      "epoch": 0.008143822987092809,
      "grad_norm": 7.079164505004883,
      "learning_rate": 1.996742470805163e-05,
      "loss": 2.4793,
      "step": 106
    },
    {
      "epoch": 0.008297480024585127,
      "grad_norm": 6.391737461090088,
      "learning_rate": 1.9966810079901662e-05,
      "loss": 2.4154,
      "step": 108
    },
    {
      "epoch": 0.008451137062077443,
      "grad_norm": 7.503854274749756,
      "learning_rate": 1.9966195451751692e-05,
      "loss": 2.3137,
      "step": 110
    },
    {
      "epoch": 0.008604794099569761,
      "grad_norm": 6.10397481918335,
      "learning_rate": 1.996558082360172e-05,
      "loss": 2.4306,
      "step": 112
    },
    {
      "epoch": 0.008758451137062077,
      "grad_norm": 6.1603264808654785,
      "learning_rate": 1.9964966195451755e-05,
      "loss": 2.2859,
      "step": 114
    },
    {
      "epoch": 0.008912108174554395,
      "grad_norm": 7.389194011688232,
      "learning_rate": 1.9964351567301784e-05,
      "loss": 2.3876,
      "step": 116
    },
    {
      "epoch": 0.009065765212046712,
      "grad_norm": 6.887446403503418,
      "learning_rate": 1.9963736939151814e-05,
      "loss": 2.5139,
      "step": 118
    },
    {
      "epoch": 0.00921942224953903,
      "grad_norm": 7.2416768074035645,
      "learning_rate": 1.9963122311001847e-05,
      "loss": 2.5653,
      "step": 120
    },
    {
      "epoch": 0.009373079287031346,
      "grad_norm": 7.454037189483643,
      "learning_rate": 1.9962507682851877e-05,
      "loss": 2.3707,
      "step": 122
    },
    {
      "epoch": 0.009526736324523662,
      "grad_norm": 6.9176483154296875,
      "learning_rate": 1.9961893054701906e-05,
      "loss": 2.4158,
      "step": 124
    },
    {
      "epoch": 0.00968039336201598,
      "grad_norm": 7.838490009307861,
      "learning_rate": 1.9961278426551936e-05,
      "loss": 2.45,
      "step": 126
    },
    {
      "epoch": 0.009834050399508297,
      "grad_norm": 6.680061340332031,
      "learning_rate": 1.996066379840197e-05,
      "loss": 2.1975,
      "step": 128
    },
    {
      "epoch": 0.009987707437000615,
      "grad_norm": 7.567671775817871,
      "learning_rate": 1.9960049170252e-05,
      "loss": 2.1892,
      "step": 130
    },
    {
      "epoch": 0.010141364474492931,
      "grad_norm": 6.0987396240234375,
      "learning_rate": 1.995943454210203e-05,
      "loss": 2.2957,
      "step": 132
    },
    {
      "epoch": 0.010295021511985249,
      "grad_norm": 6.579552173614502,
      "learning_rate": 1.995881991395206e-05,
      "loss": 2.4326,
      "step": 134
    },
    {
      "epoch": 0.010448678549477565,
      "grad_norm": 7.131938934326172,
      "learning_rate": 1.995820528580209e-05,
      "loss": 2.4767,
      "step": 136
    },
    {
      "epoch": 0.010602335586969883,
      "grad_norm": 6.883522033691406,
      "learning_rate": 1.995759065765212e-05,
      "loss": 2.3893,
      "step": 138
    },
    {
      "epoch": 0.0107559926244622,
      "grad_norm": 5.52859354019165,
      "learning_rate": 1.9956976029502154e-05,
      "loss": 2.1851,
      "step": 140
    },
    {
      "epoch": 0.010909649661954518,
      "grad_norm": 6.14478874206543,
      "learning_rate": 1.9956361401352184e-05,
      "loss": 2.2019,
      "step": 142
    },
    {
      "epoch": 0.011063306699446834,
      "grad_norm": 6.477922439575195,
      "learning_rate": 1.9955746773202213e-05,
      "loss": 2.2746,
      "step": 144
    },
    {
      "epoch": 0.011216963736939152,
      "grad_norm": 7.661022186279297,
      "learning_rate": 1.9955132145052246e-05,
      "loss": 2.3499,
      "step": 146
    },
    {
      "epoch": 0.011370620774431468,
      "grad_norm": 7.439324378967285,
      "learning_rate": 1.9954517516902276e-05,
      "loss": 2.1848,
      "step": 148
    },
    {
      "epoch": 0.011524277811923786,
      "grad_norm": 7.070183753967285,
      "learning_rate": 1.9953902888752306e-05,
      "loss": 2.2816,
      "step": 150
    },
    {
      "epoch": 0.011677934849416103,
      "grad_norm": 5.912161350250244,
      "learning_rate": 1.9953288260602336e-05,
      "loss": 2.3688,
      "step": 152
    },
    {
      "epoch": 0.01183159188690842,
      "grad_norm": 6.827462673187256,
      "learning_rate": 1.995267363245237e-05,
      "loss": 2.3945,
      "step": 154
    },
    {
      "epoch": 0.011985248924400737,
      "grad_norm": 5.7712082862854,
      "learning_rate": 1.9952059004302398e-05,
      "loss": 2.1618,
      "step": 156
    },
    {
      "epoch": 0.012138905961893055,
      "grad_norm": 5.9169020652771,
      "learning_rate": 1.9951444376152428e-05,
      "loss": 2.1781,
      "step": 158
    },
    {
      "epoch": 0.012292562999385371,
      "grad_norm": 5.994232177734375,
      "learning_rate": 1.995082974800246e-05,
      "loss": 2.1474,
      "step": 160
    },
    {
      "epoch": 0.01244622003687769,
      "grad_norm": 6.10550594329834,
      "learning_rate": 1.995021511985249e-05,
      "loss": 2.2227,
      "step": 162
    },
    {
      "epoch": 0.012599877074370006,
      "grad_norm": 7.107779502868652,
      "learning_rate": 1.994960049170252e-05,
      "loss": 2.334,
      "step": 164
    },
    {
      "epoch": 0.012753534111862324,
      "grad_norm": 4.990610122680664,
      "learning_rate": 1.9948985863552553e-05,
      "loss": 2.2313,
      "step": 166
    },
    {
      "epoch": 0.01290719114935464,
      "grad_norm": 8.93641185760498,
      "learning_rate": 1.9948371235402583e-05,
      "loss": 2.1062,
      "step": 168
    },
    {
      "epoch": 0.013060848186846958,
      "grad_norm": 5.389564037322998,
      "learning_rate": 1.9947756607252613e-05,
      "loss": 2.1729,
      "step": 170
    },
    {
      "epoch": 0.013214505224339274,
      "grad_norm": 5.347591400146484,
      "learning_rate": 1.9947141979102646e-05,
      "loss": 2.0474,
      "step": 172
    },
    {
      "epoch": 0.013368162261831592,
      "grad_norm": 6.475700378417969,
      "learning_rate": 1.9946527350952676e-05,
      "loss": 2.1939,
      "step": 174
    },
    {
      "epoch": 0.013521819299323909,
      "grad_norm": 6.144668102264404,
      "learning_rate": 1.9945912722802705e-05,
      "loss": 2.217,
      "step": 176
    },
    {
      "epoch": 0.013675476336816227,
      "grad_norm": 6.778875350952148,
      "learning_rate": 1.9945298094652735e-05,
      "loss": 2.139,
      "step": 178
    },
    {
      "epoch": 0.013829133374308543,
      "grad_norm": 7.560453414916992,
      "learning_rate": 1.9944683466502768e-05,
      "loss": 2.1931,
      "step": 180
    },
    {
      "epoch": 0.013982790411800861,
      "grad_norm": 5.251035690307617,
      "learning_rate": 1.9944068838352798e-05,
      "loss": 2.1596,
      "step": 182
    },
    {
      "epoch": 0.014136447449293177,
      "grad_norm": 5.9772162437438965,
      "learning_rate": 1.9943454210202827e-05,
      "loss": 2.232,
      "step": 184
    },
    {
      "epoch": 0.014290104486785495,
      "grad_norm": 7.088453769683838,
      "learning_rate": 1.994283958205286e-05,
      "loss": 2.3468,
      "step": 186
    },
    {
      "epoch": 0.014443761524277812,
      "grad_norm": 6.209799289703369,
      "learning_rate": 1.994222495390289e-05,
      "loss": 2.3158,
      "step": 188
    },
    {
      "epoch": 0.01459741856177013,
      "grad_norm": 6.048709392547607,
      "learning_rate": 1.994161032575292e-05,
      "loss": 1.9986,
      "step": 190
    },
    {
      "epoch": 0.014751075599262446,
      "grad_norm": 5.292468070983887,
      "learning_rate": 1.9940995697602953e-05,
      "loss": 2.1564,
      "step": 192
    },
    {
      "epoch": 0.014904732636754764,
      "grad_norm": 6.045801639556885,
      "learning_rate": 1.9940381069452983e-05,
      "loss": 2.2064,
      "step": 194
    },
    {
      "epoch": 0.01505838967424708,
      "grad_norm": 6.204288482666016,
      "learning_rate": 1.9939766441303012e-05,
      "loss": 2.2869,
      "step": 196
    },
    {
      "epoch": 0.015212046711739398,
      "grad_norm": 6.579591274261475,
      "learning_rate": 1.9939151813153045e-05,
      "loss": 2.1494,
      "step": 198
    },
    {
      "epoch": 0.015365703749231715,
      "grad_norm": 6.20919942855835,
      "learning_rate": 1.9938537185003075e-05,
      "loss": 2.0245,
      "step": 200
    },
    {
      "epoch": 0.015519360786724033,
      "grad_norm": 6.129773139953613,
      "learning_rate": 1.9937922556853108e-05,
      "loss": 2.0684,
      "step": 202
    },
    {
      "epoch": 0.01567301782421635,
      "grad_norm": 7.500084400177002,
      "learning_rate": 1.9937307928703134e-05,
      "loss": 2.1818,
      "step": 204
    },
    {
      "epoch": 0.015826674861708665,
      "grad_norm": 6.189898490905762,
      "learning_rate": 1.9936693300553167e-05,
      "loss": 2.1377,
      "step": 206
    },
    {
      "epoch": 0.015980331899200985,
      "grad_norm": 5.788628101348877,
      "learning_rate": 1.9936078672403197e-05,
      "loss": 2.1195,
      "step": 208
    },
    {
      "epoch": 0.0161339889366933,
      "grad_norm": 6.9061055183410645,
      "learning_rate": 1.9935464044253227e-05,
      "loss": 2.1815,
      "step": 210
    },
    {
      "epoch": 0.016287645974185617,
      "grad_norm": 7.366201877593994,
      "learning_rate": 1.993484941610326e-05,
      "loss": 2.2884,
      "step": 212
    },
    {
      "epoch": 0.016441303011677934,
      "grad_norm": 5.979190826416016,
      "learning_rate": 1.993423478795329e-05,
      "loss": 2.3251,
      "step": 214
    },
    {
      "epoch": 0.016594960049170254,
      "grad_norm": 6.170030117034912,
      "learning_rate": 1.993362015980332e-05,
      "loss": 2.2108,
      "step": 216
    },
    {
      "epoch": 0.01674861708666257,
      "grad_norm": 6.819857120513916,
      "learning_rate": 1.9933005531653352e-05,
      "loss": 2.2231,
      "step": 218
    },
    {
      "epoch": 0.016902274124154886,
      "grad_norm": 7.386382579803467,
      "learning_rate": 1.9932390903503382e-05,
      "loss": 2.1647,
      "step": 220
    },
    {
      "epoch": 0.017055931161647202,
      "grad_norm": 5.797331809997559,
      "learning_rate": 1.9931776275353415e-05,
      "loss": 2.2092,
      "step": 222
    },
    {
      "epoch": 0.017209588199139522,
      "grad_norm": 5.605097770690918,
      "learning_rate": 1.993116164720344e-05,
      "loss": 2.2266,
      "step": 224
    },
    {
      "epoch": 0.01736324523663184,
      "grad_norm": 5.865804672241211,
      "learning_rate": 1.9930547019053474e-05,
      "loss": 2.0874,
      "step": 226
    },
    {
      "epoch": 0.017516902274124155,
      "grad_norm": 7.769106864929199,
      "learning_rate": 1.9929932390903508e-05,
      "loss": 2.1032,
      "step": 228
    },
    {
      "epoch": 0.01767055931161647,
      "grad_norm": 6.673518180847168,
      "learning_rate": 1.9929317762753534e-05,
      "loss": 2.0957,
      "step": 230
    },
    {
      "epoch": 0.01782421634910879,
      "grad_norm": 6.331215858459473,
      "learning_rate": 1.9928703134603567e-05,
      "loss": 2.143,
      "step": 232
    },
    {
      "epoch": 0.017977873386601107,
      "grad_norm": 5.792760848999023,
      "learning_rate": 1.9928088506453597e-05,
      "loss": 2.0157,
      "step": 234
    },
    {
      "epoch": 0.018131530424093423,
      "grad_norm": 6.460434436798096,
      "learning_rate": 1.9927473878303626e-05,
      "loss": 2.0018,
      "step": 236
    },
    {
      "epoch": 0.01828518746158574,
      "grad_norm": 6.339091777801514,
      "learning_rate": 1.992685925015366e-05,
      "loss": 2.2635,
      "step": 238
    },
    {
      "epoch": 0.01843884449907806,
      "grad_norm": 5.446582317352295,
      "learning_rate": 1.992624462200369e-05,
      "loss": 2.036,
      "step": 240
    },
    {
      "epoch": 0.018592501536570376,
      "grad_norm": 6.4099273681640625,
      "learning_rate": 1.9925629993853722e-05,
      "loss": 2.0031,
      "step": 242
    },
    {
      "epoch": 0.018746158574062692,
      "grad_norm": 7.307748794555664,
      "learning_rate": 1.9925015365703752e-05,
      "loss": 2.0746,
      "step": 244
    },
    {
      "epoch": 0.01889981561155501,
      "grad_norm": 5.755754470825195,
      "learning_rate": 1.992440073755378e-05,
      "loss": 2.1756,
      "step": 246
    },
    {
      "epoch": 0.019053472649047325,
      "grad_norm": 5.9470744132995605,
      "learning_rate": 1.9923786109403815e-05,
      "loss": 2.1579,
      "step": 248
    },
    {
      "epoch": 0.019207129686539644,
      "grad_norm": 5.4200873374938965,
      "learning_rate": 1.992317148125384e-05,
      "loss": 2.1868,
      "step": 250
    },
    {
      "epoch": 0.01936078672403196,
      "grad_norm": 6.8247175216674805,
      "learning_rate": 1.9922556853103874e-05,
      "loss": 2.1525,
      "step": 252
    },
    {
      "epoch": 0.019514443761524277,
      "grad_norm": 6.334802627563477,
      "learning_rate": 1.9921942224953904e-05,
      "loss": 2.1261,
      "step": 254
    },
    {
      "epoch": 0.019668100799016593,
      "grad_norm": 7.025927543640137,
      "learning_rate": 1.9921327596803933e-05,
      "loss": 2.2474,
      "step": 256
    },
    {
      "epoch": 0.019821757836508913,
      "grad_norm": 6.594686508178711,
      "learning_rate": 1.9920712968653966e-05,
      "loss": 1.9885,
      "step": 258
    },
    {
      "epoch": 0.01997541487400123,
      "grad_norm": 6.713582992553711,
      "learning_rate": 1.9920098340503996e-05,
      "loss": 2.3728,
      "step": 260
    },
    {
      "epoch": 0.020129071911493546,
      "grad_norm": 5.78023099899292,
      "learning_rate": 1.9919483712354026e-05,
      "loss": 2.0887,
      "step": 262
    },
    {
      "epoch": 0.020282728948985862,
      "grad_norm": 5.462549686431885,
      "learning_rate": 1.991886908420406e-05,
      "loss": 2.0673,
      "step": 264
    },
    {
      "epoch": 0.020436385986478182,
      "grad_norm": 6.792922019958496,
      "learning_rate": 1.991825445605409e-05,
      "loss": 2.177,
      "step": 266
    },
    {
      "epoch": 0.020590043023970498,
      "grad_norm": 6.281880855560303,
      "learning_rate": 1.991763982790412e-05,
      "loss": 1.9686,
      "step": 268
    },
    {
      "epoch": 0.020743700061462814,
      "grad_norm": 5.745354175567627,
      "learning_rate": 1.991702519975415e-05,
      "loss": 2.1414,
      "step": 270
    },
    {
      "epoch": 0.02089735709895513,
      "grad_norm": 6.046512126922607,
      "learning_rate": 1.991641057160418e-05,
      "loss": 2.1541,
      "step": 272
    },
    {
      "epoch": 0.02105101413644745,
      "grad_norm": 7.513150691986084,
      "learning_rate": 1.9915795943454214e-05,
      "loss": 2.1383,
      "step": 274
    },
    {
      "epoch": 0.021204671173939767,
      "grad_norm": 8.351797103881836,
      "learning_rate": 1.991518131530424e-05,
      "loss": 2.209,
      "step": 276
    },
    {
      "epoch": 0.021358328211432083,
      "grad_norm": 6.781789302825928,
      "learning_rate": 1.9914566687154273e-05,
      "loss": 1.9494,
      "step": 278
    },
    {
      "epoch": 0.0215119852489244,
      "grad_norm": 5.912288188934326,
      "learning_rate": 1.9913952059004303e-05,
      "loss": 1.9871,
      "step": 280
    },
    {
      "epoch": 0.02166564228641672,
      "grad_norm": 5.441234111785889,
      "learning_rate": 1.9913337430854333e-05,
      "loss": 2.118,
      "step": 282
    },
    {
      "epoch": 0.021819299323909035,
      "grad_norm": 6.041057109832764,
      "learning_rate": 1.9912722802704366e-05,
      "loss": 2.0064,
      "step": 284
    },
    {
      "epoch": 0.02197295636140135,
      "grad_norm": 6.26601505279541,
      "learning_rate": 1.9912108174554395e-05,
      "loss": 1.9593,
      "step": 286
    },
    {
      "epoch": 0.022126613398893668,
      "grad_norm": 6.992424488067627,
      "learning_rate": 1.991149354640443e-05,
      "loss": 2.1785,
      "step": 288
    },
    {
      "epoch": 0.022280270436385988,
      "grad_norm": 7.048946857452393,
      "learning_rate": 1.9910878918254458e-05,
      "loss": 2.0809,
      "step": 290
    },
    {
      "epoch": 0.022433927473878304,
      "grad_norm": 7.00367546081543,
      "learning_rate": 1.9910264290104488e-05,
      "loss": 2.0688,
      "step": 292
    },
    {
      "epoch": 0.02258758451137062,
      "grad_norm": 6.326030731201172,
      "learning_rate": 1.990964966195452e-05,
      "loss": 2.1279,
      "step": 294
    },
    {
      "epoch": 0.022741241548862937,
      "grad_norm": 5.886343002319336,
      "learning_rate": 1.990903503380455e-05,
      "loss": 1.9146,
      "step": 296
    },
    {
      "epoch": 0.022894898586355256,
      "grad_norm": 6.407416820526123,
      "learning_rate": 1.990842040565458e-05,
      "loss": 2.073,
      "step": 298
    },
    {
      "epoch": 0.023048555623847573,
      "grad_norm": 5.35817289352417,
      "learning_rate": 1.9907805777504613e-05,
      "loss": 2.064,
      "step": 300
    },
    {
      "epoch": 0.02320221266133989,
      "grad_norm": 5.71148157119751,
      "learning_rate": 1.990719114935464e-05,
      "loss": 2.2207,
      "step": 302
    },
    {
      "epoch": 0.023355869698832205,
      "grad_norm": 7.2422051429748535,
      "learning_rate": 1.9906576521204673e-05,
      "loss": 2.1518,
      "step": 304
    },
    {
      "epoch": 0.023509526736324525,
      "grad_norm": 7.267468452453613,
      "learning_rate": 1.9905961893054702e-05,
      "loss": 2.0082,
      "step": 306
    },
    {
      "epoch": 0.02366318377381684,
      "grad_norm": 6.504114627838135,
      "learning_rate": 1.9905347264904736e-05,
      "loss": 1.9722,
      "step": 308
    },
    {
      "epoch": 0.023816840811309158,
      "grad_norm": 7.074812889099121,
      "learning_rate": 1.9904732636754765e-05,
      "loss": 2.1789,
      "step": 310
    },
    {
      "epoch": 0.023970497848801474,
      "grad_norm": 6.774876117706299,
      "learning_rate": 1.9904118008604795e-05,
      "loss": 2.219,
      "step": 312
    },
    {
      "epoch": 0.024124154886293794,
      "grad_norm": 5.666469097137451,
      "learning_rate": 1.9903503380454828e-05,
      "loss": 1.8294,
      "step": 314
    },
    {
      "epoch": 0.02427781192378611,
      "grad_norm": 6.548127174377441,
      "learning_rate": 1.9902888752304858e-05,
      "loss": 2.0859,
      "step": 316
    },
    {
      "epoch": 0.024431468961278426,
      "grad_norm": 5.174642562866211,
      "learning_rate": 1.9902274124154887e-05,
      "loss": 1.989,
      "step": 318
    },
    {
      "epoch": 0.024585125998770743,
      "grad_norm": 5.891490936279297,
      "learning_rate": 1.990165949600492e-05,
      "loss": 2.0776,
      "step": 320
    },
    {
      "epoch": 0.024738783036263062,
      "grad_norm": 5.7647504806518555,
      "learning_rate": 1.9901044867854947e-05,
      "loss": 1.9681,
      "step": 322
    },
    {
      "epoch": 0.02489244007375538,
      "grad_norm": 5.61868143081665,
      "learning_rate": 1.990043023970498e-05,
      "loss": 1.8923,
      "step": 324
    },
    {
      "epoch": 0.025046097111247695,
      "grad_norm": 7.358055114746094,
      "learning_rate": 1.9899815611555013e-05,
      "loss": 1.9859,
      "step": 326
    },
    {
      "epoch": 0.02519975414874001,
      "grad_norm": 5.265814781188965,
      "learning_rate": 1.9899200983405043e-05,
      "loss": 1.939,
      "step": 328
    },
    {
      "epoch": 0.02535341118623233,
      "grad_norm": 9.370257377624512,
      "learning_rate": 1.9898586355255072e-05,
      "loss": 1.9538,
      "step": 330
    },
    {
      "epoch": 0.025507068223724647,
      "grad_norm": 7.504848003387451,
      "learning_rate": 1.9897971727105102e-05,
      "loss": 2.0802,
      "step": 332
    },
    {
      "epoch": 0.025660725261216964,
      "grad_norm": 5.975841045379639,
      "learning_rate": 1.9897357098955135e-05,
      "loss": 1.853,
      "step": 334
    },
    {
      "epoch": 0.02581438229870928,
      "grad_norm": 6.099985122680664,
      "learning_rate": 1.9896742470805165e-05,
      "loss": 2.0014,
      "step": 336
    },
    {
      "epoch": 0.0259680393362016,
      "grad_norm": 6.825030326843262,
      "learning_rate": 1.9896127842655194e-05,
      "loss": 1.9608,
      "step": 338
    },
    {
      "epoch": 0.026121696373693916,
      "grad_norm": 6.16441535949707,
      "learning_rate": 1.9895513214505227e-05,
      "loss": 2.0848,
      "step": 340
    },
    {
      "epoch": 0.026275353411186232,
      "grad_norm": 6.392692565917969,
      "learning_rate": 1.9894898586355257e-05,
      "loss": 1.9651,
      "step": 342
    },
    {
      "epoch": 0.02642901044867855,
      "grad_norm": 5.567882537841797,
      "learning_rate": 1.9894283958205287e-05,
      "loss": 2.1211,
      "step": 344
    },
    {
      "epoch": 0.026582667486170868,
      "grad_norm": 10.182480812072754,
      "learning_rate": 1.989366933005532e-05,
      "loss": 1.9924,
      "step": 346
    },
    {
      "epoch": 0.026736324523663185,
      "grad_norm": 5.608663558959961,
      "learning_rate": 1.989305470190535e-05,
      "loss": 1.936,
      "step": 348
    },
    {
      "epoch": 0.0268899815611555,
      "grad_norm": 5.883683204650879,
      "learning_rate": 1.989244007375538e-05,
      "loss": 2.0998,
      "step": 350
    },
    {
      "epoch": 0.027043638598647817,
      "grad_norm": 8.584614753723145,
      "learning_rate": 1.989182544560541e-05,
      "loss": 2.1266,
      "step": 352
    },
    {
      "epoch": 0.027197295636140137,
      "grad_norm": 6.828667640686035,
      "learning_rate": 1.9891210817455442e-05,
      "loss": 1.8693,
      "step": 354
    },
    {
      "epoch": 0.027350952673632453,
      "grad_norm": 7.0278449058532715,
      "learning_rate": 1.989059618930547e-05,
      "loss": 1.9258,
      "step": 356
    },
    {
      "epoch": 0.02750460971112477,
      "grad_norm": 5.643075466156006,
      "learning_rate": 1.98899815611555e-05,
      "loss": 2.071,
      "step": 358
    },
    {
      "epoch": 0.027658266748617086,
      "grad_norm": 6.685908794403076,
      "learning_rate": 1.9889366933005534e-05,
      "loss": 1.8658,
      "step": 360
    },
    {
      "epoch": 0.027811923786109402,
      "grad_norm": 5.766722679138184,
      "learning_rate": 1.9888752304855564e-05,
      "loss": 2.0608,
      "step": 362
    },
    {
      "epoch": 0.027965580823601722,
      "grad_norm": 6.229999542236328,
      "learning_rate": 1.9888137676705594e-05,
      "loss": 1.8478,
      "step": 364
    },
    {
      "epoch": 0.028119237861094038,
      "grad_norm": 14.6449613571167,
      "learning_rate": 1.9887523048555627e-05,
      "loss": 2.0233,
      "step": 366
    },
    {
      "epoch": 0.028272894898586354,
      "grad_norm": 5.458970069885254,
      "learning_rate": 1.9886908420405657e-05,
      "loss": 1.8742,
      "step": 368
    },
    {
      "epoch": 0.02842655193607867,
      "grad_norm": 9.708429336547852,
      "learning_rate": 1.9886293792255686e-05,
      "loss": 2.0435,
      "step": 370
    },
    {
      "epoch": 0.02858020897357099,
      "grad_norm": 8.345685958862305,
      "learning_rate": 1.988567916410572e-05,
      "loss": 1.9448,
      "step": 372
    },
    {
      "epoch": 0.028733866011063307,
      "grad_norm": 5.213901519775391,
      "learning_rate": 1.988506453595575e-05,
      "loss": 1.8902,
      "step": 374
    },
    {
      "epoch": 0.028887523048555623,
      "grad_norm": 6.842494964599609,
      "learning_rate": 1.988444990780578e-05,
      "loss": 2.0958,
      "step": 376
    },
    {
      "epoch": 0.02904118008604794,
      "grad_norm": 6.533809185028076,
      "learning_rate": 1.988383527965581e-05,
      "loss": 2.0073,
      "step": 378
    },
    {
      "epoch": 0.02919483712354026,
      "grad_norm": 5.832721710205078,
      "learning_rate": 1.988322065150584e-05,
      "loss": 1.9462,
      "step": 380
    },
    {
      "epoch": 0.029348494161032575,
      "grad_norm": 6.040827751159668,
      "learning_rate": 1.988260602335587e-05,
      "loss": 2.0111,
      "step": 382
    },
    {
      "epoch": 0.02950215119852489,
      "grad_norm": 6.082043647766113,
      "learning_rate": 1.98819913952059e-05,
      "loss": 2.0088,
      "step": 384
    },
    {
      "epoch": 0.029655808236017208,
      "grad_norm": 4.5363383293151855,
      "learning_rate": 1.9881376767055934e-05,
      "loss": 1.9059,
      "step": 386
    },
    {
      "epoch": 0.029809465273509528,
      "grad_norm": 4.769321918487549,
      "learning_rate": 1.9880762138905964e-05,
      "loss": 1.8781,
      "step": 388
    },
    {
      "epoch": 0.029963122311001844,
      "grad_norm": 6.1424994468688965,
      "learning_rate": 1.9880147510755993e-05,
      "loss": 2.0232,
      "step": 390
    },
    {
      "epoch": 0.03011677934849416,
      "grad_norm": 6.081544399261475,
      "learning_rate": 1.9879532882606026e-05,
      "loss": 1.8908,
      "step": 392
    },
    {
      "epoch": 0.030270436385986477,
      "grad_norm": 6.146285057067871,
      "learning_rate": 1.9878918254456056e-05,
      "loss": 2.0144,
      "step": 394
    },
    {
      "epoch": 0.030424093423478796,
      "grad_norm": 5.401834011077881,
      "learning_rate": 1.9878303626306086e-05,
      "loss": 1.8258,
      "step": 396
    },
    {
      "epoch": 0.030577750460971113,
      "grad_norm": 6.835007667541504,
      "learning_rate": 1.987768899815612e-05,
      "loss": 2.0515,
      "step": 398
    },
    {
      "epoch": 0.03073140749846343,
      "grad_norm": 7.031691551208496,
      "learning_rate": 1.987707437000615e-05,
      "loss": 2.0362,
      "step": 400
    },
    {
      "epoch": 0.030885064535955745,
      "grad_norm": 5.733877182006836,
      "learning_rate": 1.9876459741856178e-05,
      "loss": 2.099,
      "step": 402
    },
    {
      "epoch": 0.031038721573448065,
      "grad_norm": 6.152698516845703,
      "learning_rate": 1.9875845113706208e-05,
      "loss": 1.9393,
      "step": 404
    },
    {
      "epoch": 0.03119237861094038,
      "grad_norm": 5.859741687774658,
      "learning_rate": 1.987523048555624e-05,
      "loss": 1.995,
      "step": 406
    },
    {
      "epoch": 0.0313460356484327,
      "grad_norm": 6.834084510803223,
      "learning_rate": 1.987461585740627e-05,
      "loss": 2.0035,
      "step": 408
    },
    {
      "epoch": 0.031499692685925014,
      "grad_norm": 6.169229030609131,
      "learning_rate": 1.98740012292563e-05,
      "loss": 2.1276,
      "step": 410
    },
    {
      "epoch": 0.03165334972341733,
      "grad_norm": 5.270079135894775,
      "learning_rate": 1.9873386601106333e-05,
      "loss": 2.0073,
      "step": 412
    },
    {
      "epoch": 0.03180700676090965,
      "grad_norm": 5.952144145965576,
      "learning_rate": 1.9872771972956363e-05,
      "loss": 1.8817,
      "step": 414
    },
    {
      "epoch": 0.03196066379840197,
      "grad_norm": 6.3290019035339355,
      "learning_rate": 1.9872157344806393e-05,
      "loss": 1.9526,
      "step": 416
    },
    {
      "epoch": 0.032114320835894286,
      "grad_norm": 5.712306499481201,
      "learning_rate": 1.9871542716656426e-05,
      "loss": 2.0793,
      "step": 418
    },
    {
      "epoch": 0.0322679778733866,
      "grad_norm": 5.497166156768799,
      "learning_rate": 1.9870928088506455e-05,
      "loss": 1.9061,
      "step": 420
    },
    {
      "epoch": 0.03242163491087892,
      "grad_norm": 6.435750484466553,
      "learning_rate": 1.9870313460356485e-05,
      "loss": 1.8971,
      "step": 422
    },
    {
      "epoch": 0.032575291948371235,
      "grad_norm": 5.9519734382629395,
      "learning_rate": 1.9869698832206518e-05,
      "loss": 2.0295,
      "step": 424
    },
    {
      "epoch": 0.03272894898586355,
      "grad_norm": 6.359841823577881,
      "learning_rate": 1.9869084204056548e-05,
      "loss": 1.9017,
      "step": 426
    },
    {
      "epoch": 0.03288260602335587,
      "grad_norm": 6.195022106170654,
      "learning_rate": 1.9868469575906578e-05,
      "loss": 2.0663,
      "step": 428
    },
    {
      "epoch": 0.033036263060848184,
      "grad_norm": 5.500522613525391,
      "learning_rate": 1.9867854947756607e-05,
      "loss": 1.9694,
      "step": 430
    },
    {
      "epoch": 0.03318992009834051,
      "grad_norm": 7.16880464553833,
      "learning_rate": 1.986724031960664e-05,
      "loss": 1.918,
      "step": 432
    },
    {
      "epoch": 0.03334357713583282,
      "grad_norm": 6.0987348556518555,
      "learning_rate": 1.986662569145667e-05,
      "loss": 1.8705,
      "step": 434
    },
    {
      "epoch": 0.03349723417332514,
      "grad_norm": 6.8652753829956055,
      "learning_rate": 1.98660110633067e-05,
      "loss": 1.9383,
      "step": 436
    },
    {
      "epoch": 0.033650891210817456,
      "grad_norm": 5.421166896820068,
      "learning_rate": 1.9865396435156733e-05,
      "loss": 1.879,
      "step": 438
    },
    {
      "epoch": 0.03380454824830977,
      "grad_norm": 5.929842948913574,
      "learning_rate": 1.9864781807006762e-05,
      "loss": 1.7183,
      "step": 440
    },
    {
      "epoch": 0.03395820528580209,
      "grad_norm": 5.500015735626221,
      "learning_rate": 1.9864167178856792e-05,
      "loss": 1.9168,
      "step": 442
    },
    {
      "epoch": 0.034111862323294405,
      "grad_norm": 6.267481327056885,
      "learning_rate": 1.9863552550706825e-05,
      "loss": 1.8126,
      "step": 444
    },
    {
      "epoch": 0.03426551936078672,
      "grad_norm": 6.300197124481201,
      "learning_rate": 1.9862937922556855e-05,
      "loss": 2.0519,
      "step": 446
    },
    {
      "epoch": 0.034419176398279044,
      "grad_norm": 8.094818115234375,
      "learning_rate": 1.9862323294406885e-05,
      "loss": 1.8122,
      "step": 448
    },
    {
      "epoch": 0.03457283343577136,
      "grad_norm": 5.738587379455566,
      "learning_rate": 1.9861708666256914e-05,
      "loss": 1.8155,
      "step": 450
    },
    {
      "epoch": 0.03472649047326368,
      "grad_norm": 5.194686412811279,
      "learning_rate": 1.9861094038106947e-05,
      "loss": 1.9198,
      "step": 452
    },
    {
      "epoch": 0.03488014751075599,
      "grad_norm": 4.97174072265625,
      "learning_rate": 1.9860479409956977e-05,
      "loss": 1.9955,
      "step": 454
    },
    {
      "epoch": 0.03503380454824831,
      "grad_norm": 5.790378570556641,
      "learning_rate": 1.9859864781807007e-05,
      "loss": 1.8218,
      "step": 456
    },
    {
      "epoch": 0.035187461585740626,
      "grad_norm": 5.287135124206543,
      "learning_rate": 1.985925015365704e-05,
      "loss": 1.9169,
      "step": 458
    },
    {
      "epoch": 0.03534111862323294,
      "grad_norm": 8.098136901855469,
      "learning_rate": 1.985863552550707e-05,
      "loss": 1.9039,
      "step": 460
    },
    {
      "epoch": 0.03549477566072526,
      "grad_norm": 6.957726955413818,
      "learning_rate": 1.98580208973571e-05,
      "loss": 2.0036,
      "step": 462
    },
    {
      "epoch": 0.03564843269821758,
      "grad_norm": 4.368841171264648,
      "learning_rate": 1.9857406269207132e-05,
      "loss": 1.8883,
      "step": 464
    },
    {
      "epoch": 0.0358020897357099,
      "grad_norm": 5.95673131942749,
      "learning_rate": 1.9856791641057162e-05,
      "loss": 1.8977,
      "step": 466
    },
    {
      "epoch": 0.035955746773202214,
      "grad_norm": 7.365513324737549,
      "learning_rate": 1.985617701290719e-05,
      "loss": 1.9865,
      "step": 468
    },
    {
      "epoch": 0.03610940381069453,
      "grad_norm": 5.386063098907471,
      "learning_rate": 1.9855562384757225e-05,
      "loss": 1.8164,
      "step": 470
    },
    {
      "epoch": 0.03626306084818685,
      "grad_norm": 6.155988693237305,
      "learning_rate": 1.9854947756607254e-05,
      "loss": 2.0083,
      "step": 472
    },
    {
      "epoch": 0.03641671788567916,
      "grad_norm": 6.110922336578369,
      "learning_rate": 1.9854333128457287e-05,
      "loss": 1.8688,
      "step": 474
    },
    {
      "epoch": 0.03657037492317148,
      "grad_norm": 5.692699909210205,
      "learning_rate": 1.9853718500307314e-05,
      "loss": 1.8501,
      "step": 476
    },
    {
      "epoch": 0.036724031960663796,
      "grad_norm": 6.044013977050781,
      "learning_rate": 1.9853103872157347e-05,
      "loss": 1.8486,
      "step": 478
    },
    {
      "epoch": 0.03687768899815612,
      "grad_norm": 6.102372169494629,
      "learning_rate": 1.9852489244007376e-05,
      "loss": 2.0873,
      "step": 480
    },
    {
      "epoch": 0.037031346035648435,
      "grad_norm": 5.4327239990234375,
      "learning_rate": 1.9851874615857406e-05,
      "loss": 1.8635,
      "step": 482
    },
    {
      "epoch": 0.03718500307314075,
      "grad_norm": 5.779347896575928,
      "learning_rate": 1.985125998770744e-05,
      "loss": 2.0413,
      "step": 484
    },
    {
      "epoch": 0.03733866011063307,
      "grad_norm": 5.000186920166016,
      "learning_rate": 1.985064535955747e-05,
      "loss": 2.0214,
      "step": 486
    },
    {
      "epoch": 0.037492317148125384,
      "grad_norm": 6.581515312194824,
      "learning_rate": 1.98500307314075e-05,
      "loss": 1.9141,
      "step": 488
    },
    {
      "epoch": 0.0376459741856177,
      "grad_norm": 6.037952423095703,
      "learning_rate": 1.984941610325753e-05,
      "loss": 1.9475,
      "step": 490
    },
    {
      "epoch": 0.03779963122311002,
      "grad_norm": 4.99038553237915,
      "learning_rate": 1.984880147510756e-05,
      "loss": 1.8296,
      "step": 492
    },
    {
      "epoch": 0.03795328826060233,
      "grad_norm": 5.351291656494141,
      "learning_rate": 1.9848186846957594e-05,
      "loss": 1.9845,
      "step": 494
    },
    {
      "epoch": 0.03810694529809465,
      "grad_norm": 6.249404430389404,
      "learning_rate": 1.9847572218807624e-05,
      "loss": 1.8824,
      "step": 496
    },
    {
      "epoch": 0.03826060233558697,
      "grad_norm": 5.460664749145508,
      "learning_rate": 1.9846957590657654e-05,
      "loss": 1.9348,
      "step": 498
    },
    {
      "epoch": 0.03841425937307929,
      "grad_norm": 5.399702072143555,
      "learning_rate": 1.9846342962507687e-05,
      "loss": 1.8646,
      "step": 500
    },
    {
      "epoch": 0.038567916410571605,
      "grad_norm": 6.00943660736084,
      "learning_rate": 1.9845728334357713e-05,
      "loss": 1.8804,
      "step": 502
    },
    {
      "epoch": 0.03872157344806392,
      "grad_norm": 6.057244300842285,
      "learning_rate": 1.9845113706207746e-05,
      "loss": 1.8876,
      "step": 504
    },
    {
      "epoch": 0.03887523048555624,
      "grad_norm": 5.178292274475098,
      "learning_rate": 1.9844499078057776e-05,
      "loss": 1.8163,
      "step": 506
    },
    {
      "epoch": 0.039028887523048554,
      "grad_norm": 5.430099964141846,
      "learning_rate": 1.9843884449907806e-05,
      "loss": 1.9221,
      "step": 508
    },
    {
      "epoch": 0.03918254456054087,
      "grad_norm": 5.2391791343688965,
      "learning_rate": 1.984326982175784e-05,
      "loss": 1.9671,
      "step": 510
    },
    {
      "epoch": 0.03933620159803319,
      "grad_norm": 6.54328727722168,
      "learning_rate": 1.9842655193607868e-05,
      "loss": 1.9916,
      "step": 512
    },
    {
      "epoch": 0.03948985863552551,
      "grad_norm": 5.6781134605407715,
      "learning_rate": 1.98420405654579e-05,
      "loss": 1.9,
      "step": 514
    },
    {
      "epoch": 0.039643515673017826,
      "grad_norm": 5.34329891204834,
      "learning_rate": 1.984142593730793e-05,
      "loss": 1.7433,
      "step": 516
    },
    {
      "epoch": 0.03979717271051014,
      "grad_norm": 6.142169952392578,
      "learning_rate": 1.984081130915796e-05,
      "loss": 1.8559,
      "step": 518
    },
    {
      "epoch": 0.03995082974800246,
      "grad_norm": 5.825856685638428,
      "learning_rate": 1.9840196681007994e-05,
      "loss": 1.7434,
      "step": 520
    },
    {
      "epoch": 0.040104486785494775,
      "grad_norm": 4.883429050445557,
      "learning_rate": 1.9839582052858023e-05,
      "loss": 1.8403,
      "step": 522
    },
    {
      "epoch": 0.04025814382298709,
      "grad_norm": 5.759003162384033,
      "learning_rate": 1.9838967424708053e-05,
      "loss": 1.872,
      "step": 524
    },
    {
      "epoch": 0.04041180086047941,
      "grad_norm": 5.845025539398193,
      "learning_rate": 1.9838352796558086e-05,
      "loss": 1.818,
      "step": 526
    },
    {
      "epoch": 0.040565457897971724,
      "grad_norm": 6.238631248474121,
      "learning_rate": 1.9837738168408113e-05,
      "loss": 1.9553,
      "step": 528
    },
    {
      "epoch": 0.04071911493546405,
      "grad_norm": 5.450825214385986,
      "learning_rate": 1.9837123540258146e-05,
      "loss": 1.9314,
      "step": 530
    },
    {
      "epoch": 0.040872771972956363,
      "grad_norm": 5.4290385246276855,
      "learning_rate": 1.9836508912108175e-05,
      "loss": 1.8316,
      "step": 532
    },
    {
      "epoch": 0.04102642901044868,
      "grad_norm": 6.243955612182617,
      "learning_rate": 1.9835894283958205e-05,
      "loss": 1.9605,
      "step": 534
    },
    {
      "epoch": 0.041180086047940996,
      "grad_norm": 5.5207672119140625,
      "learning_rate": 1.9835279655808238e-05,
      "loss": 1.9377,
      "step": 536
    },
    {
      "epoch": 0.04133374308543331,
      "grad_norm": 5.570779323577881,
      "learning_rate": 1.9834665027658268e-05,
      "loss": 1.9706,
      "step": 538
    },
    {
      "epoch": 0.04148740012292563,
      "grad_norm": 4.921234130859375,
      "learning_rate": 1.98340503995083e-05,
      "loss": 1.8666,
      "step": 540
    },
    {
      "epoch": 0.041641057160417945,
      "grad_norm": 6.029317855834961,
      "learning_rate": 1.983343577135833e-05,
      "loss": 1.8431,
      "step": 542
    },
    {
      "epoch": 0.04179471419791026,
      "grad_norm": 5.6237664222717285,
      "learning_rate": 1.983282114320836e-05,
      "loss": 2.0265,
      "step": 544
    },
    {
      "epoch": 0.041948371235402585,
      "grad_norm": 4.848809719085693,
      "learning_rate": 1.9832206515058393e-05,
      "loss": 1.851,
      "step": 546
    },
    {
      "epoch": 0.0421020282728949,
      "grad_norm": 6.06104040145874,
      "learning_rate": 1.983159188690842e-05,
      "loss": 1.9252,
      "step": 548
    },
    {
      "epoch": 0.04225568531038722,
      "grad_norm": 6.721662521362305,
      "learning_rate": 1.9830977258758453e-05,
      "loss": 1.9046,
      "step": 550
    },
    {
      "epoch": 0.04240934234787953,
      "grad_norm": 5.039158821105957,
      "learning_rate": 1.9830362630608482e-05,
      "loss": 1.9457,
      "step": 552
    },
    {
      "epoch": 0.04256299938537185,
      "grad_norm": 4.985758304595947,
      "learning_rate": 1.9829748002458512e-05,
      "loss": 1.7706,
      "step": 554
    },
    {
      "epoch": 0.042716656422864166,
      "grad_norm": 5.59445858001709,
      "learning_rate": 1.9829133374308545e-05,
      "loss": 1.9232,
      "step": 556
    },
    {
      "epoch": 0.04287031346035648,
      "grad_norm": 5.786518573760986,
      "learning_rate": 1.9828518746158575e-05,
      "loss": 1.9535,
      "step": 558
    },
    {
      "epoch": 0.0430239704978488,
      "grad_norm": 5.362064838409424,
      "learning_rate": 1.9827904118008608e-05,
      "loss": 1.774,
      "step": 560
    },
    {
      "epoch": 0.04317762753534112,
      "grad_norm": 6.807535171508789,
      "learning_rate": 1.9827289489858637e-05,
      "loss": 1.9963,
      "step": 562
    },
    {
      "epoch": 0.04333128457283344,
      "grad_norm": 4.927182197570801,
      "learning_rate": 1.9826674861708667e-05,
      "loss": 1.8839,
      "step": 564
    },
    {
      "epoch": 0.043484941610325754,
      "grad_norm": 7.077647686004639,
      "learning_rate": 1.98260602335587e-05,
      "loss": 1.8577,
      "step": 566
    },
    {
      "epoch": 0.04363859864781807,
      "grad_norm": 4.930956840515137,
      "learning_rate": 1.982544560540873e-05,
      "loss": 1.9032,
      "step": 568
    },
    {
      "epoch": 0.04379225568531039,
      "grad_norm": 5.537839889526367,
      "learning_rate": 1.982483097725876e-05,
      "loss": 1.8599,
      "step": 570
    },
    {
      "epoch": 0.0439459127228027,
      "grad_norm": 4.91294527053833,
      "learning_rate": 1.9824216349108793e-05,
      "loss": 1.8962,
      "step": 572
    },
    {
      "epoch": 0.04409956976029502,
      "grad_norm": 7.946929931640625,
      "learning_rate": 1.982360172095882e-05,
      "loss": 2.0401,
      "step": 574
    },
    {
      "epoch": 0.044253226797787336,
      "grad_norm": 5.566417217254639,
      "learning_rate": 1.9822987092808852e-05,
      "loss": 1.7317,
      "step": 576
    },
    {
      "epoch": 0.04440688383527966,
      "grad_norm": 6.196030616760254,
      "learning_rate": 1.9822372464658882e-05,
      "loss": 1.9818,
      "step": 578
    },
    {
      "epoch": 0.044560540872771975,
      "grad_norm": 5.8990888595581055,
      "learning_rate": 1.9821757836508915e-05,
      "loss": 1.9209,
      "step": 580
    },
    {
      "epoch": 0.04471419791026429,
      "grad_norm": 4.752439022064209,
      "learning_rate": 1.9821143208358944e-05,
      "loss": 1.8661,
      "step": 582
    },
    {
      "epoch": 0.04486785494775661,
      "grad_norm": 5.3692121505737305,
      "learning_rate": 1.9820528580208974e-05,
      "loss": 1.8574,
      "step": 584
    },
    {
      "epoch": 0.045021511985248924,
      "grad_norm": 4.94577169418335,
      "learning_rate": 1.9819913952059007e-05,
      "loss": 1.76,
      "step": 586
    },
    {
      "epoch": 0.04517516902274124,
      "grad_norm": 5.1533708572387695,
      "learning_rate": 1.9819299323909037e-05,
      "loss": 1.8634,
      "step": 588
    },
    {
      "epoch": 0.04532882606023356,
      "grad_norm": 5.460253715515137,
      "learning_rate": 1.9818684695759067e-05,
      "loss": 1.7615,
      "step": 590
    },
    {
      "epoch": 0.04548248309772587,
      "grad_norm": 6.106910705566406,
      "learning_rate": 1.98180700676091e-05,
      "loss": 1.8658,
      "step": 592
    },
    {
      "epoch": 0.045636140135218196,
      "grad_norm": 8.604896545410156,
      "learning_rate": 1.981745543945913e-05,
      "loss": 1.8234,
      "step": 594
    },
    {
      "epoch": 0.04578979717271051,
      "grad_norm": 5.533381938934326,
      "learning_rate": 1.981684081130916e-05,
      "loss": 1.8133,
      "step": 596
    },
    {
      "epoch": 0.04594345421020283,
      "grad_norm": 5.140172481536865,
      "learning_rate": 1.9816226183159192e-05,
      "loss": 1.7655,
      "step": 598
    },
    {
      "epoch": 0.046097111247695145,
      "grad_norm": 5.633389472961426,
      "learning_rate": 1.9815611555009222e-05,
      "loss": 1.8804,
      "step": 600
    },
    {
      "epoch": 0.04625076828518746,
      "grad_norm": 5.397654056549072,
      "learning_rate": 1.981499692685925e-05,
      "loss": 1.9422,
      "step": 602
    },
    {
      "epoch": 0.04640442532267978,
      "grad_norm": 5.916885852813721,
      "learning_rate": 1.981438229870928e-05,
      "loss": 1.9222,
      "step": 604
    },
    {
      "epoch": 0.046558082360172094,
      "grad_norm": 4.4198198318481445,
      "learning_rate": 1.9813767670559314e-05,
      "loss": 1.8088,
      "step": 606
    },
    {
      "epoch": 0.04671173939766441,
      "grad_norm": 6.035666465759277,
      "learning_rate": 1.9813153042409344e-05,
      "loss": 1.9505,
      "step": 608
    },
    {
      "epoch": 0.04686539643515673,
      "grad_norm": 5.293002605438232,
      "learning_rate": 1.9812538414259374e-05,
      "loss": 1.9354,
      "step": 610
    },
    {
      "epoch": 0.04701905347264905,
      "grad_norm": 5.066743850708008,
      "learning_rate": 1.9811923786109407e-05,
      "loss": 2.001,
      "step": 612
    },
    {
      "epoch": 0.047172710510141366,
      "grad_norm": 6.867171764373779,
      "learning_rate": 1.9811309157959436e-05,
      "loss": 1.86,
      "step": 614
    },
    {
      "epoch": 0.04732636754763368,
      "grad_norm": 4.908615589141846,
      "learning_rate": 1.9810694529809466e-05,
      "loss": 1.8855,
      "step": 616
    },
    {
      "epoch": 0.047480024585126,
      "grad_norm": 5.6588006019592285,
      "learning_rate": 1.98100799016595e-05,
      "loss": 1.8047,
      "step": 618
    },
    {
      "epoch": 0.047633681622618315,
      "grad_norm": 5.6555304527282715,
      "learning_rate": 1.980946527350953e-05,
      "loss": 1.7656,
      "step": 620
    },
    {
      "epoch": 0.04778733866011063,
      "grad_norm": 4.742602348327637,
      "learning_rate": 1.980885064535956e-05,
      "loss": 1.976,
      "step": 622
    },
    {
      "epoch": 0.04794099569760295,
      "grad_norm": 5.0910868644714355,
      "learning_rate": 1.980823601720959e-05,
      "loss": 1.8894,
      "step": 624
    },
    {
      "epoch": 0.048094652735095264,
      "grad_norm": 5.279669761657715,
      "learning_rate": 1.980762138905962e-05,
      "loss": 1.9323,
      "step": 626
    },
    {
      "epoch": 0.04824830977258759,
      "grad_norm": 5.603051662445068,
      "learning_rate": 1.980700676090965e-05,
      "loss": 1.9327,
      "step": 628
    },
    {
      "epoch": 0.048401966810079904,
      "grad_norm": 5.823456764221191,
      "learning_rate": 1.980639213275968e-05,
      "loss": 1.9087,
      "step": 630
    },
    {
      "epoch": 0.04855562384757222,
      "grad_norm": 4.226296424865723,
      "learning_rate": 1.9805777504609714e-05,
      "loss": 1.7298,
      "step": 632
    },
    {
      "epoch": 0.048709280885064536,
      "grad_norm": 4.537020683288574,
      "learning_rate": 1.9805162876459743e-05,
      "loss": 1.8588,
      "step": 634
    },
    {
      "epoch": 0.04886293792255685,
      "grad_norm": 5.843430519104004,
      "learning_rate": 1.9804548248309773e-05,
      "loss": 1.8581,
      "step": 636
    },
    {
      "epoch": 0.04901659496004917,
      "grad_norm": 5.234043598175049,
      "learning_rate": 1.9803933620159806e-05,
      "loss": 1.8016,
      "step": 638
    },
    {
      "epoch": 0.049170251997541485,
      "grad_norm": 6.091218948364258,
      "learning_rate": 1.9803318992009836e-05,
      "loss": 1.8419,
      "step": 640
    },
    {
      "epoch": 0.0493239090350338,
      "grad_norm": 5.473825454711914,
      "learning_rate": 1.9802704363859865e-05,
      "loss": 1.8742,
      "step": 642
    },
    {
      "epoch": 0.049477566072526125,
      "grad_norm": 5.018134117126465,
      "learning_rate": 1.98020897357099e-05,
      "loss": 1.9246,
      "step": 644
    },
    {
      "epoch": 0.04963122311001844,
      "grad_norm": 5.1250505447387695,
      "learning_rate": 1.9801475107559928e-05,
      "loss": 1.8988,
      "step": 646
    },
    {
      "epoch": 0.04978488014751076,
      "grad_norm": 5.310157299041748,
      "learning_rate": 1.9800860479409958e-05,
      "loss": 1.9718,
      "step": 648
    },
    {
      "epoch": 0.049938537185003073,
      "grad_norm": 5.5490570068359375,
      "learning_rate": 1.980024585125999e-05,
      "loss": 1.8779,
      "step": 650
    },
    {
      "epoch": 0.05009219422249539,
      "grad_norm": 5.242208480834961,
      "learning_rate": 1.979963122311002e-05,
      "loss": 1.8712,
      "step": 652
    },
    {
      "epoch": 0.050245851259987706,
      "grad_norm": 4.680446624755859,
      "learning_rate": 1.979901659496005e-05,
      "loss": 1.9475,
      "step": 654
    },
    {
      "epoch": 0.05039950829748002,
      "grad_norm": 12.400496482849121,
      "learning_rate": 1.979840196681008e-05,
      "loss": 1.9387,
      "step": 656
    },
    {
      "epoch": 0.05055316533497234,
      "grad_norm": 4.818700313568115,
      "learning_rate": 1.9797787338660113e-05,
      "loss": 1.7356,
      "step": 658
    },
    {
      "epoch": 0.05070682237246466,
      "grad_norm": 4.733686923980713,
      "learning_rate": 1.9797172710510143e-05,
      "loss": 1.7161,
      "step": 660
    },
    {
      "epoch": 0.05086047940995698,
      "grad_norm": 5.9219865798950195,
      "learning_rate": 1.9796558082360172e-05,
      "loss": 1.9821,
      "step": 662
    },
    {
      "epoch": 0.051014136447449294,
      "grad_norm": 4.954675197601318,
      "learning_rate": 1.9795943454210206e-05,
      "loss": 1.9392,
      "step": 664
    },
    {
      "epoch": 0.05116779348494161,
      "grad_norm": 4.482631206512451,
      "learning_rate": 1.9795328826060235e-05,
      "loss": 1.9687,
      "step": 666
    },
    {
      "epoch": 0.05132145052243393,
      "grad_norm": 6.749068737030029,
      "learning_rate": 1.9794714197910265e-05,
      "loss": 1.8242,
      "step": 668
    },
    {
      "epoch": 0.05147510755992624,
      "grad_norm": 4.532095909118652,
      "learning_rate": 1.9794099569760298e-05,
      "loss": 1.7596,
      "step": 670
    },
    {
      "epoch": 0.05162876459741856,
      "grad_norm": 5.727676868438721,
      "learning_rate": 1.9793484941610328e-05,
      "loss": 1.94,
      "step": 672
    },
    {
      "epoch": 0.051782421634910876,
      "grad_norm": 5.493950843811035,
      "learning_rate": 1.9792870313460357e-05,
      "loss": 1.9243,
      "step": 674
    },
    {
      "epoch": 0.0519360786724032,
      "grad_norm": 5.48468017578125,
      "learning_rate": 1.9792255685310387e-05,
      "loss": 1.7862,
      "step": 676
    },
    {
      "epoch": 0.052089735709895515,
      "grad_norm": 5.862773895263672,
      "learning_rate": 1.979164105716042e-05,
      "loss": 1.843,
      "step": 678
    },
    {
      "epoch": 0.05224339274738783,
      "grad_norm": 5.505096912384033,
      "learning_rate": 1.979102642901045e-05,
      "loss": 1.8366,
      "step": 680
    },
    {
      "epoch": 0.05239704978488015,
      "grad_norm": 5.697121620178223,
      "learning_rate": 1.979041180086048e-05,
      "loss": 1.9764,
      "step": 682
    },
    {
      "epoch": 0.052550706822372464,
      "grad_norm": 4.900547027587891,
      "learning_rate": 1.9789797172710513e-05,
      "loss": 1.9252,
      "step": 684
    },
    {
      "epoch": 0.05270436385986478,
      "grad_norm": 5.347836017608643,
      "learning_rate": 1.9789182544560542e-05,
      "loss": 1.8527,
      "step": 686
    },
    {
      "epoch": 0.0528580208973571,
      "grad_norm": 5.393474102020264,
      "learning_rate": 1.9788567916410572e-05,
      "loss": 1.8422,
      "step": 688
    },
    {
      "epoch": 0.05301167793484941,
      "grad_norm": 5.27833366394043,
      "learning_rate": 1.9787953288260605e-05,
      "loss": 1.8933,
      "step": 690
    },
    {
      "epoch": 0.053165334972341736,
      "grad_norm": 5.38336181640625,
      "learning_rate": 1.9787338660110635e-05,
      "loss": 1.9456,
      "step": 692
    },
    {
      "epoch": 0.05331899200983405,
      "grad_norm": 5.273176193237305,
      "learning_rate": 1.9786724031960664e-05,
      "loss": 1.8142,
      "step": 694
    },
    {
      "epoch": 0.05347264904732637,
      "grad_norm": 5.413751125335693,
      "learning_rate": 1.9786109403810697e-05,
      "loss": 1.7652,
      "step": 696
    },
    {
      "epoch": 0.053626306084818685,
      "grad_norm": 5.373195648193359,
      "learning_rate": 1.9785494775660727e-05,
      "loss": 1.883,
      "step": 698
    },
    {
      "epoch": 0.053779963122311,
      "grad_norm": 4.942586421966553,
      "learning_rate": 1.9784880147510757e-05,
      "loss": 1.76,
      "step": 700
    },
    {
      "epoch": 0.05393362015980332,
      "grad_norm": 5.6196980476379395,
      "learning_rate": 1.9784265519360786e-05,
      "loss": 1.7673,
      "step": 702
    },
    {
      "epoch": 0.054087277197295634,
      "grad_norm": 5.702764987945557,
      "learning_rate": 1.978365089121082e-05,
      "loss": 1.8437,
      "step": 704
    },
    {
      "epoch": 0.05424093423478795,
      "grad_norm": 4.99530553817749,
      "learning_rate": 1.978303626306085e-05,
      "loss": 1.8747,
      "step": 706
    },
    {
      "epoch": 0.054394591272280274,
      "grad_norm": 5.105679035186768,
      "learning_rate": 1.978242163491088e-05,
      "loss": 1.6632,
      "step": 708
    },
    {
      "epoch": 0.05454824830977259,
      "grad_norm": 4.710418701171875,
      "learning_rate": 1.9781807006760912e-05,
      "loss": 1.8736,
      "step": 710
    },
    {
      "epoch": 0.054701905347264906,
      "grad_norm": 4.792379856109619,
      "learning_rate": 1.978119237861094e-05,
      "loss": 1.8016,
      "step": 712
    },
    {
      "epoch": 0.05485556238475722,
      "grad_norm": 4.937024116516113,
      "learning_rate": 1.978057775046097e-05,
      "loss": 1.7436,
      "step": 714
    },
    {
      "epoch": 0.05500921942224954,
      "grad_norm": 5.5544867515563965,
      "learning_rate": 1.9779963122311004e-05,
      "loss": 1.926,
      "step": 716
    },
    {
      "epoch": 0.055162876459741855,
      "grad_norm": 6.484194278717041,
      "learning_rate": 1.9779348494161034e-05,
      "loss": 1.9102,
      "step": 718
    },
    {
      "epoch": 0.05531653349723417,
      "grad_norm": 5.408361434936523,
      "learning_rate": 1.9778733866011064e-05,
      "loss": 1.7786,
      "step": 720
    },
    {
      "epoch": 0.05547019053472649,
      "grad_norm": 5.705206394195557,
      "learning_rate": 1.9778119237861097e-05,
      "loss": 1.779,
      "step": 722
    },
    {
      "epoch": 0.055623847572218804,
      "grad_norm": 6.138594627380371,
      "learning_rate": 1.9777504609711127e-05,
      "loss": 1.6926,
      "step": 724
    },
    {
      "epoch": 0.05577750460971113,
      "grad_norm": 5.507882595062256,
      "learning_rate": 1.977688998156116e-05,
      "loss": 2.0119,
      "step": 726
    },
    {
      "epoch": 0.055931161647203444,
      "grad_norm": 5.1471710205078125,
      "learning_rate": 1.9776275353411186e-05,
      "loss": 1.7674,
      "step": 728
    },
    {
      "epoch": 0.05608481868469576,
      "grad_norm": 5.558322906494141,
      "learning_rate": 1.977566072526122e-05,
      "loss": 1.7262,
      "step": 730
    },
    {
      "epoch": 0.056238475722188076,
      "grad_norm": 5.859812259674072,
      "learning_rate": 1.977504609711125e-05,
      "loss": 2.0273,
      "step": 732
    },
    {
      "epoch": 0.05639213275968039,
      "grad_norm": 4.931456565856934,
      "learning_rate": 1.977443146896128e-05,
      "loss": 1.7008,
      "step": 734
    },
    {
      "epoch": 0.05654578979717271,
      "grad_norm": 4.835200786590576,
      "learning_rate": 1.977381684081131e-05,
      "loss": 1.7615,
      "step": 736
    },
    {
      "epoch": 0.056699446834665025,
      "grad_norm": 5.542105674743652,
      "learning_rate": 1.977320221266134e-05,
      "loss": 1.7991,
      "step": 738
    },
    {
      "epoch": 0.05685310387215734,
      "grad_norm": 5.737773895263672,
      "learning_rate": 1.977258758451137e-05,
      "loss": 1.7067,
      "step": 740
    },
    {
      "epoch": 0.057006760909649665,
      "grad_norm": 4.556394100189209,
      "learning_rate": 1.9771972956361404e-05,
      "loss": 1.8418,
      "step": 742
    },
    {
      "epoch": 0.05716041794714198,
      "grad_norm": 4.682400226593018,
      "learning_rate": 1.9771358328211434e-05,
      "loss": 1.7565,
      "step": 744
    },
    {
      "epoch": 0.0573140749846343,
      "grad_norm": 5.617753982543945,
      "learning_rate": 1.9770743700061467e-05,
      "loss": 1.8314,
      "step": 746
    },
    {
      "epoch": 0.057467732022126614,
      "grad_norm": 4.796401500701904,
      "learning_rate": 1.9770129071911496e-05,
      "loss": 1.6892,
      "step": 748
    },
    {
      "epoch": 0.05762138905961893,
      "grad_norm": 5.084446430206299,
      "learning_rate": 1.9769514443761526e-05,
      "loss": 1.6731,
      "step": 750
    },
    {
      "epoch": 0.057775046097111246,
      "grad_norm": 5.344216823577881,
      "learning_rate": 1.976889981561156e-05,
      "loss": 1.8096,
      "step": 752
    },
    {
      "epoch": 0.05792870313460356,
      "grad_norm": 4.87506103515625,
      "learning_rate": 1.9768285187461585e-05,
      "loss": 1.9015,
      "step": 754
    },
    {
      "epoch": 0.05808236017209588,
      "grad_norm": 5.019058704376221,
      "learning_rate": 1.976767055931162e-05,
      "loss": 1.9467,
      "step": 756
    },
    {
      "epoch": 0.0582360172095882,
      "grad_norm": 5.275008678436279,
      "learning_rate": 1.9767055931161648e-05,
      "loss": 1.6159,
      "step": 758
    },
    {
      "epoch": 0.05838967424708052,
      "grad_norm": 5.17955207824707,
      "learning_rate": 1.9766441303011678e-05,
      "loss": 1.6819,
      "step": 760
    },
    {
      "epoch": 0.058543331284572835,
      "grad_norm": 5.578658580780029,
      "learning_rate": 1.976582667486171e-05,
      "loss": 1.8369,
      "step": 762
    },
    {
      "epoch": 0.05869698832206515,
      "grad_norm": 4.934607982635498,
      "learning_rate": 1.976521204671174e-05,
      "loss": 1.8909,
      "step": 764
    },
    {
      "epoch": 0.05885064535955747,
      "grad_norm": 5.5896759033203125,
      "learning_rate": 1.9764597418561774e-05,
      "loss": 1.7238,
      "step": 766
    },
    {
      "epoch": 0.05900430239704978,
      "grad_norm": 5.263469696044922,
      "learning_rate": 1.9763982790411803e-05,
      "loss": 1.7776,
      "step": 768
    },
    {
      "epoch": 0.0591579594345421,
      "grad_norm": 4.459990978240967,
      "learning_rate": 1.9763368162261833e-05,
      "loss": 1.7082,
      "step": 770
    },
    {
      "epoch": 0.059311616472034416,
      "grad_norm": 5.528759002685547,
      "learning_rate": 1.9762753534111866e-05,
      "loss": 1.9014,
      "step": 772
    },
    {
      "epoch": 0.05946527350952674,
      "grad_norm": 5.372073650360107,
      "learning_rate": 1.9762138905961892e-05,
      "loss": 1.691,
      "step": 774
    },
    {
      "epoch": 0.059618930547019056,
      "grad_norm": 5.765900135040283,
      "learning_rate": 1.9761524277811925e-05,
      "loss": 1.9243,
      "step": 776
    },
    {
      "epoch": 0.05977258758451137,
      "grad_norm": 5.123989105224609,
      "learning_rate": 1.9760909649661955e-05,
      "loss": 2.0223,
      "step": 778
    },
    {
      "epoch": 0.05992624462200369,
      "grad_norm": 5.149808406829834,
      "learning_rate": 1.9760295021511985e-05,
      "loss": 1.8063,
      "step": 780
    },
    {
      "epoch": 0.060079901659496004,
      "grad_norm": 5.047703266143799,
      "learning_rate": 1.9759680393362018e-05,
      "loss": 1.9353,
      "step": 782
    },
    {
      "epoch": 0.06023355869698832,
      "grad_norm": 5.555423259735107,
      "learning_rate": 1.9759065765212048e-05,
      "loss": 1.695,
      "step": 784
    },
    {
      "epoch": 0.06038721573448064,
      "grad_norm": 5.100247859954834,
      "learning_rate": 1.9758451137062077e-05,
      "loss": 1.6758,
      "step": 786
    },
    {
      "epoch": 0.06054087277197295,
      "grad_norm": 4.941176891326904,
      "learning_rate": 1.975783650891211e-05,
      "loss": 1.7391,
      "step": 788
    },
    {
      "epoch": 0.06069452980946528,
      "grad_norm": 5.3119964599609375,
      "learning_rate": 1.975722188076214e-05,
      "loss": 2.0366,
      "step": 790
    },
    {
      "epoch": 0.06084818684695759,
      "grad_norm": 6.0686235427856445,
      "learning_rate": 1.9756607252612173e-05,
      "loss": 1.7149,
      "step": 792
    },
    {
      "epoch": 0.06100184388444991,
      "grad_norm": 6.141575336456299,
      "learning_rate": 1.9755992624462203e-05,
      "loss": 1.7808,
      "step": 794
    },
    {
      "epoch": 0.061155500921942225,
      "grad_norm": 5.157688140869141,
      "learning_rate": 1.9755377996312232e-05,
      "loss": 1.8069,
      "step": 796
    },
    {
      "epoch": 0.06130915795943454,
      "grad_norm": 5.358695983886719,
      "learning_rate": 1.9754763368162266e-05,
      "loss": 1.9632,
      "step": 798
    },
    {
      "epoch": 0.06146281499692686,
      "grad_norm": 5.3423261642456055,
      "learning_rate": 1.9754148740012292e-05,
      "loss": 1.6711,
      "step": 800
    },
    {
      "epoch": 0.061616472034419174,
      "grad_norm": 5.9911980628967285,
      "learning_rate": 1.9753534111862325e-05,
      "loss": 1.6384,
      "step": 802
    },
    {
      "epoch": 0.06177012907191149,
      "grad_norm": 5.021694183349609,
      "learning_rate": 1.9752919483712355e-05,
      "loss": 1.7412,
      "step": 804
    },
    {
      "epoch": 0.061923786109403814,
      "grad_norm": 5.38372802734375,
      "learning_rate": 1.9752304855562384e-05,
      "loss": 1.5895,
      "step": 806
    },
    {
      "epoch": 0.06207744314689613,
      "grad_norm": 5.618641376495361,
      "learning_rate": 1.9751690227412417e-05,
      "loss": 1.8097,
      "step": 808
    },
    {
      "epoch": 0.062231100184388446,
      "grad_norm": 5.081387519836426,
      "learning_rate": 1.9751075599262447e-05,
      "loss": 1.7789,
      "step": 810
    },
    {
      "epoch": 0.06238475722188076,
      "grad_norm": 5.361464500427246,
      "learning_rate": 1.975046097111248e-05,
      "loss": 1.7821,
      "step": 812
    },
    {
      "epoch": 0.06253841425937308,
      "grad_norm": 5.113397598266602,
      "learning_rate": 1.974984634296251e-05,
      "loss": 1.7292,
      "step": 814
    },
    {
      "epoch": 0.0626920712968654,
      "grad_norm": 5.261277198791504,
      "learning_rate": 1.974923171481254e-05,
      "loss": 1.6956,
      "step": 816
    },
    {
      "epoch": 0.06284572833435771,
      "grad_norm": 7.06874942779541,
      "learning_rate": 1.9748617086662573e-05,
      "loss": 1.865,
      "step": 818
    },
    {
      "epoch": 0.06299938537185003,
      "grad_norm": 4.949324131011963,
      "learning_rate": 1.9748002458512602e-05,
      "loss": 1.7885,
      "step": 820
    },
    {
      "epoch": 0.06315304240934234,
      "grad_norm": 6.291264533996582,
      "learning_rate": 1.9747387830362632e-05,
      "loss": 1.8685,
      "step": 822
    },
    {
      "epoch": 0.06330669944683466,
      "grad_norm": 4.500913143157959,
      "learning_rate": 1.9746773202212665e-05,
      "loss": 1.6422,
      "step": 824
    },
    {
      "epoch": 0.06346035648432698,
      "grad_norm": 5.313440322875977,
      "learning_rate": 1.974615857406269e-05,
      "loss": 1.8328,
      "step": 826
    },
    {
      "epoch": 0.0636140135218193,
      "grad_norm": 5.809798240661621,
      "learning_rate": 1.9745543945912724e-05,
      "loss": 1.8912,
      "step": 828
    },
    {
      "epoch": 0.06376767055931162,
      "grad_norm": 4.69051456451416,
      "learning_rate": 1.9744929317762754e-05,
      "loss": 1.6786,
      "step": 830
    },
    {
      "epoch": 0.06392132759680394,
      "grad_norm": 5.292459487915039,
      "learning_rate": 1.9744314689612787e-05,
      "loss": 1.8497,
      "step": 832
    },
    {
      "epoch": 0.06407498463429626,
      "grad_norm": 4.772144794464111,
      "learning_rate": 1.9743700061462817e-05,
      "loss": 1.6261,
      "step": 834
    },
    {
      "epoch": 0.06422864167178857,
      "grad_norm": 4.984364032745361,
      "learning_rate": 1.9743085433312846e-05,
      "loss": 1.7354,
      "step": 836
    },
    {
      "epoch": 0.06438229870928089,
      "grad_norm": 4.450577735900879,
      "learning_rate": 1.974247080516288e-05,
      "loss": 1.7349,
      "step": 838
    },
    {
      "epoch": 0.0645359557467732,
      "grad_norm": 5.341747760772705,
      "learning_rate": 1.974185617701291e-05,
      "loss": 1.8332,
      "step": 840
    },
    {
      "epoch": 0.06468961278426552,
      "grad_norm": 5.368303298950195,
      "learning_rate": 1.974124154886294e-05,
      "loss": 1.8219,
      "step": 842
    },
    {
      "epoch": 0.06484326982175784,
      "grad_norm": 4.4096360206604,
      "learning_rate": 1.9740626920712972e-05,
      "loss": 1.8158,
      "step": 844
    },
    {
      "epoch": 0.06499692685925015,
      "grad_norm": 6.098479270935059,
      "learning_rate": 1.9740012292563e-05,
      "loss": 1.7323,
      "step": 846
    },
    {
      "epoch": 0.06515058389674247,
      "grad_norm": 4.606769561767578,
      "learning_rate": 1.973939766441303e-05,
      "loss": 1.7527,
      "step": 848
    },
    {
      "epoch": 0.06530424093423479,
      "grad_norm": 6.082258701324463,
      "learning_rate": 1.9738783036263064e-05,
      "loss": 1.8471,
      "step": 850
    },
    {
      "epoch": 0.0654578979717271,
      "grad_norm": 5.389958381652832,
      "learning_rate": 1.9738168408113094e-05,
      "loss": 1.8564,
      "step": 852
    },
    {
      "epoch": 0.06561155500921942,
      "grad_norm": 5.574385643005371,
      "learning_rate": 1.9737553779963124e-05,
      "loss": 1.784,
      "step": 854
    },
    {
      "epoch": 0.06576521204671174,
      "grad_norm": 5.1567487716674805,
      "learning_rate": 1.9736939151813153e-05,
      "loss": 1.713,
      "step": 856
    },
    {
      "epoch": 0.06591886908420405,
      "grad_norm": 5.475706577301025,
      "learning_rate": 1.9736324523663187e-05,
      "loss": 1.7555,
      "step": 858
    },
    {
      "epoch": 0.06607252612169637,
      "grad_norm": 4.831605434417725,
      "learning_rate": 1.9735709895513216e-05,
      "loss": 1.8695,
      "step": 860
    },
    {
      "epoch": 0.06622618315918868,
      "grad_norm": 6.022873878479004,
      "learning_rate": 1.9735095267363246e-05,
      "loss": 1.7765,
      "step": 862
    },
    {
      "epoch": 0.06637984019668101,
      "grad_norm": 4.874941825866699,
      "learning_rate": 1.973448063921328e-05,
      "loss": 1.757,
      "step": 864
    },
    {
      "epoch": 0.06653349723417333,
      "grad_norm": 4.488655090332031,
      "learning_rate": 1.973386601106331e-05,
      "loss": 1.8837,
      "step": 866
    },
    {
      "epoch": 0.06668715427166565,
      "grad_norm": 4.5713090896606445,
      "learning_rate": 1.973325138291334e-05,
      "loss": 1.6627,
      "step": 868
    },
    {
      "epoch": 0.06684081130915796,
      "grad_norm": 5.312070846557617,
      "learning_rate": 1.973263675476337e-05,
      "loss": 1.7551,
      "step": 870
    },
    {
      "epoch": 0.06699446834665028,
      "grad_norm": 5.104644775390625,
      "learning_rate": 1.97320221266134e-05,
      "loss": 1.8789,
      "step": 872
    },
    {
      "epoch": 0.0671481253841426,
      "grad_norm": 4.595895290374756,
      "learning_rate": 1.973140749846343e-05,
      "loss": 1.8319,
      "step": 874
    },
    {
      "epoch": 0.06730178242163491,
      "grad_norm": 5.223522186279297,
      "learning_rate": 1.973079287031346e-05,
      "loss": 1.6661,
      "step": 876
    },
    {
      "epoch": 0.06745543945912723,
      "grad_norm": 4.466522693634033,
      "learning_rate": 1.9730178242163494e-05,
      "loss": 1.8338,
      "step": 878
    },
    {
      "epoch": 0.06760909649661954,
      "grad_norm": 4.613927841186523,
      "learning_rate": 1.9729563614013523e-05,
      "loss": 1.7562,
      "step": 880
    },
    {
      "epoch": 0.06776275353411186,
      "grad_norm": 5.868101119995117,
      "learning_rate": 1.9728948985863553e-05,
      "loss": 1.9181,
      "step": 882
    },
    {
      "epoch": 0.06791641057160418,
      "grad_norm": 5.407005786895752,
      "learning_rate": 1.9728334357713586e-05,
      "loss": 1.8462,
      "step": 884
    },
    {
      "epoch": 0.0680700676090965,
      "grad_norm": 6.075082778930664,
      "learning_rate": 1.9727719729563616e-05,
      "loss": 1.816,
      "step": 886
    },
    {
      "epoch": 0.06822372464658881,
      "grad_norm": 5.164397716522217,
      "learning_rate": 1.9727105101413645e-05,
      "loss": 1.9236,
      "step": 888
    },
    {
      "epoch": 0.06837738168408113,
      "grad_norm": 5.688714981079102,
      "learning_rate": 1.972649047326368e-05,
      "loss": 1.7171,
      "step": 890
    },
    {
      "epoch": 0.06853103872157344,
      "grad_norm": 4.9518842697143555,
      "learning_rate": 1.9725875845113708e-05,
      "loss": 1.7829,
      "step": 892
    },
    {
      "epoch": 0.06868469575906576,
      "grad_norm": 5.185763835906982,
      "learning_rate": 1.9725261216963738e-05,
      "loss": 1.7146,
      "step": 894
    },
    {
      "epoch": 0.06883835279655809,
      "grad_norm": 5.043625354766846,
      "learning_rate": 1.972464658881377e-05,
      "loss": 1.8594,
      "step": 896
    },
    {
      "epoch": 0.0689920098340504,
      "grad_norm": 4.783642292022705,
      "learning_rate": 1.97240319606638e-05,
      "loss": 1.7053,
      "step": 898
    },
    {
      "epoch": 0.06914566687154272,
      "grad_norm": 4.887513637542725,
      "learning_rate": 1.972341733251383e-05,
      "loss": 1.7929,
      "step": 900
    },
    {
      "epoch": 0.06929932390903504,
      "grad_norm": 4.4108123779296875,
      "learning_rate": 1.972280270436386e-05,
      "loss": 1.7296,
      "step": 902
    },
    {
      "epoch": 0.06945298094652735,
      "grad_norm": 5.531246662139893,
      "learning_rate": 1.9722188076213893e-05,
      "loss": 1.7792,
      "step": 904
    },
    {
      "epoch": 0.06960663798401967,
      "grad_norm": 4.462593078613281,
      "learning_rate": 1.9721573448063923e-05,
      "loss": 1.7121,
      "step": 906
    },
    {
      "epoch": 0.06976029502151199,
      "grad_norm": 4.543118000030518,
      "learning_rate": 1.9720958819913952e-05,
      "loss": 1.7871,
      "step": 908
    },
    {
      "epoch": 0.0699139520590043,
      "grad_norm": 5.9536638259887695,
      "learning_rate": 1.9720344191763985e-05,
      "loss": 1.7956,
      "step": 910
    },
    {
      "epoch": 0.07006760909649662,
      "grad_norm": 4.735901832580566,
      "learning_rate": 1.9719729563614015e-05,
      "loss": 1.7511,
      "step": 912
    },
    {
      "epoch": 0.07022126613398894,
      "grad_norm": 4.490820407867432,
      "learning_rate": 1.9719114935464045e-05,
      "loss": 1.7364,
      "step": 914
    },
    {
      "epoch": 0.07037492317148125,
      "grad_norm": 4.837772846221924,
      "learning_rate": 1.9718500307314078e-05,
      "loss": 1.8204,
      "step": 916
    },
    {
      "epoch": 0.07052858020897357,
      "grad_norm": 4.400464057922363,
      "learning_rate": 1.9717885679164108e-05,
      "loss": 1.8637,
      "step": 918
    },
    {
      "epoch": 0.07068223724646588,
      "grad_norm": 4.991397857666016,
      "learning_rate": 1.9717271051014137e-05,
      "loss": 1.9175,
      "step": 920
    },
    {
      "epoch": 0.0708358942839582,
      "grad_norm": 4.983181953430176,
      "learning_rate": 1.971665642286417e-05,
      "loss": 1.7065,
      "step": 922
    },
    {
      "epoch": 0.07098955132145052,
      "grad_norm": 4.5055341720581055,
      "learning_rate": 1.97160417947142e-05,
      "loss": 1.7113,
      "step": 924
    },
    {
      "epoch": 0.07114320835894283,
      "grad_norm": 5.021308422088623,
      "learning_rate": 1.971542716656423e-05,
      "loss": 1.8903,
      "step": 926
    },
    {
      "epoch": 0.07129686539643516,
      "grad_norm": 6.712131023406982,
      "learning_rate": 1.971481253841426e-05,
      "loss": 1.7608,
      "step": 928
    },
    {
      "epoch": 0.07145052243392748,
      "grad_norm": 5.711028575897217,
      "learning_rate": 1.9714197910264292e-05,
      "loss": 1.7826,
      "step": 930
    },
    {
      "epoch": 0.0716041794714198,
      "grad_norm": 5.202549457550049,
      "learning_rate": 1.9713583282114322e-05,
      "loss": 1.7201,
      "step": 932
    },
    {
      "epoch": 0.07175783650891211,
      "grad_norm": 4.809873580932617,
      "learning_rate": 1.9712968653964352e-05,
      "loss": 1.7824,
      "step": 934
    },
    {
      "epoch": 0.07191149354640443,
      "grad_norm": 4.417870998382568,
      "learning_rate": 1.9712354025814385e-05,
      "loss": 1.8275,
      "step": 936
    },
    {
      "epoch": 0.07206515058389674,
      "grad_norm": 4.823970794677734,
      "learning_rate": 1.9711739397664415e-05,
      "loss": 1.7293,
      "step": 938
    },
    {
      "epoch": 0.07221880762138906,
      "grad_norm": 5.289034843444824,
      "learning_rate": 1.9711124769514444e-05,
      "loss": 1.7507,
      "step": 940
    },
    {
      "epoch": 0.07237246465888138,
      "grad_norm": 4.538127422332764,
      "learning_rate": 1.9710510141364477e-05,
      "loss": 1.6391,
      "step": 942
    },
    {
      "epoch": 0.0725261216963737,
      "grad_norm": 4.500412464141846,
      "learning_rate": 1.9709895513214507e-05,
      "loss": 1.6816,
      "step": 944
    },
    {
      "epoch": 0.07267977873386601,
      "grad_norm": 5.149693012237549,
      "learning_rate": 1.9709280885064537e-05,
      "loss": 1.6366,
      "step": 946
    },
    {
      "epoch": 0.07283343577135833,
      "grad_norm": 4.554830074310303,
      "learning_rate": 1.970866625691457e-05,
      "loss": 1.7712,
      "step": 948
    },
    {
      "epoch": 0.07298709280885064,
      "grad_norm": 5.352977752685547,
      "learning_rate": 1.97080516287646e-05,
      "loss": 1.5819,
      "step": 950
    },
    {
      "epoch": 0.07314074984634296,
      "grad_norm": 5.3114094734191895,
      "learning_rate": 1.970743700061463e-05,
      "loss": 1.802,
      "step": 952
    },
    {
      "epoch": 0.07329440688383528,
      "grad_norm": 4.19597053527832,
      "learning_rate": 1.970682237246466e-05,
      "loss": 1.5855,
      "step": 954
    },
    {
      "epoch": 0.07344806392132759,
      "grad_norm": 4.087234020233154,
      "learning_rate": 1.9706207744314692e-05,
      "loss": 1.7158,
      "step": 956
    },
    {
      "epoch": 0.07360172095881991,
      "grad_norm": 5.064235210418701,
      "learning_rate": 1.970559311616472e-05,
      "loss": 1.7579,
      "step": 958
    },
    {
      "epoch": 0.07375537799631224,
      "grad_norm": 5.033870697021484,
      "learning_rate": 1.970497848801475e-05,
      "loss": 1.8105,
      "step": 960
    },
    {
      "epoch": 0.07390903503380455,
      "grad_norm": 4.546055793762207,
      "learning_rate": 1.9704363859864784e-05,
      "loss": 1.6964,
      "step": 962
    },
    {
      "epoch": 0.07406269207129687,
      "grad_norm": 5.068551540374756,
      "learning_rate": 1.9703749231714814e-05,
      "loss": 1.6664,
      "step": 964
    },
    {
      "epoch": 0.07421634910878919,
      "grad_norm": 5.1122050285339355,
      "learning_rate": 1.9703134603564844e-05,
      "loss": 1.819,
      "step": 966
    },
    {
      "epoch": 0.0743700061462815,
      "grad_norm": 4.410829544067383,
      "learning_rate": 1.9702519975414877e-05,
      "loss": 1.7872,
      "step": 968
    },
    {
      "epoch": 0.07452366318377382,
      "grad_norm": 4.5425190925598145,
      "learning_rate": 1.9701905347264906e-05,
      "loss": 1.6679,
      "step": 970
    },
    {
      "epoch": 0.07467732022126614,
      "grad_norm": 4.571905612945557,
      "learning_rate": 1.9701290719114936e-05,
      "loss": 1.7135,
      "step": 972
    },
    {
      "epoch": 0.07483097725875845,
      "grad_norm": 4.853597640991211,
      "learning_rate": 1.970067609096497e-05,
      "loss": 1.6162,
      "step": 974
    },
    {
      "epoch": 0.07498463429625077,
      "grad_norm": 5.288270473480225,
      "learning_rate": 1.9700061462815e-05,
      "loss": 1.7473,
      "step": 976
    },
    {
      "epoch": 0.07513829133374308,
      "grad_norm": 4.418172359466553,
      "learning_rate": 1.9699446834665032e-05,
      "loss": 1.6986,
      "step": 978
    },
    {
      "epoch": 0.0752919483712354,
      "grad_norm": 4.6486496925354,
      "learning_rate": 1.9698832206515058e-05,
      "loss": 1.6914,
      "step": 980
    },
    {
      "epoch": 0.07544560540872772,
      "grad_norm": 5.205509185791016,
      "learning_rate": 1.969821757836509e-05,
      "loss": 1.6129,
      "step": 982
    },
    {
      "epoch": 0.07559926244622003,
      "grad_norm": 4.540061950683594,
      "learning_rate": 1.969760295021512e-05,
      "loss": 1.7026,
      "step": 984
    },
    {
      "epoch": 0.07575291948371235,
      "grad_norm": 4.7810845375061035,
      "learning_rate": 1.969698832206515e-05,
      "loss": 1.686,
      "step": 986
    },
    {
      "epoch": 0.07590657652120467,
      "grad_norm": 4.475192546844482,
      "learning_rate": 1.9696373693915184e-05,
      "loss": 1.739,
      "step": 988
    },
    {
      "epoch": 0.07606023355869698,
      "grad_norm": 5.317092418670654,
      "learning_rate": 1.9695759065765213e-05,
      "loss": 2.0029,
      "step": 990
    },
    {
      "epoch": 0.0762138905961893,
      "grad_norm": 5.178996562957764,
      "learning_rate": 1.9695144437615243e-05,
      "loss": 1.7278,
      "step": 992
    },
    {
      "epoch": 0.07636754763368163,
      "grad_norm": 5.976894855499268,
      "learning_rate": 1.9694529809465276e-05,
      "loss": 1.753,
      "step": 994
    },
    {
      "epoch": 0.07652120467117395,
      "grad_norm": 4.536045551300049,
      "learning_rate": 1.9693915181315306e-05,
      "loss": 1.6081,
      "step": 996
    },
    {
      "epoch": 0.07667486170866626,
      "grad_norm": 4.751937389373779,
      "learning_rate": 1.969330055316534e-05,
      "loss": 1.6765,
      "step": 998
    },
    {
      "epoch": 0.07682851874615858,
      "grad_norm": 5.145371437072754,
      "learning_rate": 1.9692685925015365e-05,
      "loss": 1.9088,
      "step": 1000
    },
    {
      "epoch": 0.0769821757836509,
      "grad_norm": 5.149151802062988,
      "learning_rate": 1.9692071296865398e-05,
      "loss": 1.7855,
      "step": 1002
    },
    {
      "epoch": 0.07713583282114321,
      "grad_norm": 4.331295490264893,
      "learning_rate": 1.9691456668715428e-05,
      "loss": 1.6398,
      "step": 1004
    },
    {
      "epoch": 0.07728948985863553,
      "grad_norm": 5.288293361663818,
      "learning_rate": 1.9690842040565458e-05,
      "loss": 1.8451,
      "step": 1006
    },
    {
      "epoch": 0.07744314689612784,
      "grad_norm": 4.709437370300293,
      "learning_rate": 1.969022741241549e-05,
      "loss": 1.8546,
      "step": 1008
    },
    {
      "epoch": 0.07759680393362016,
      "grad_norm": 5.373138427734375,
      "learning_rate": 1.968961278426552e-05,
      "loss": 1.7561,
      "step": 1010
    },
    {
      "epoch": 0.07775046097111248,
      "grad_norm": 6.308657169342041,
      "learning_rate": 1.968899815611555e-05,
      "loss": 1.8271,
      "step": 1012
    },
    {
      "epoch": 0.07790411800860479,
      "grad_norm": 5.380505561828613,
      "learning_rate": 1.9688383527965583e-05,
      "loss": 1.7852,
      "step": 1014
    },
    {
      "epoch": 0.07805777504609711,
      "grad_norm": 5.468803405761719,
      "learning_rate": 1.9687768899815613e-05,
      "loss": 1.7797,
      "step": 1016
    },
    {
      "epoch": 0.07821143208358942,
      "grad_norm": 4.73374080657959,
      "learning_rate": 1.9687154271665646e-05,
      "loss": 1.6201,
      "step": 1018
    },
    {
      "epoch": 0.07836508912108174,
      "grad_norm": 4.9570631980896,
      "learning_rate": 1.9686539643515676e-05,
      "loss": 1.7538,
      "step": 1020
    },
    {
      "epoch": 0.07851874615857406,
      "grad_norm": 4.574160575866699,
      "learning_rate": 1.9685925015365705e-05,
      "loss": 1.7656,
      "step": 1022
    },
    {
      "epoch": 0.07867240319606637,
      "grad_norm": 5.078851222991943,
      "learning_rate": 1.968531038721574e-05,
      "loss": 1.806,
      "step": 1024
    },
    {
      "epoch": 0.0788260602335587,
      "grad_norm": 5.474549293518066,
      "learning_rate": 1.9684695759065765e-05,
      "loss": 1.8226,
      "step": 1026
    },
    {
      "epoch": 0.07897971727105102,
      "grad_norm": 6.257920265197754,
      "learning_rate": 1.9684081130915798e-05,
      "loss": 1.7714,
      "step": 1028
    },
    {
      "epoch": 0.07913337430854334,
      "grad_norm": 4.811653137207031,
      "learning_rate": 1.9683466502765827e-05,
      "loss": 1.7903,
      "step": 1030
    },
    {
      "epoch": 0.07928703134603565,
      "grad_norm": 4.90382194519043,
      "learning_rate": 1.9682851874615857e-05,
      "loss": 1.9185,
      "step": 1032
    },
    {
      "epoch": 0.07944068838352797,
      "grad_norm": 5.112819194793701,
      "learning_rate": 1.968223724646589e-05,
      "loss": 1.8098,
      "step": 1034
    },
    {
      "epoch": 0.07959434542102028,
      "grad_norm": 4.832859039306641,
      "learning_rate": 1.968162261831592e-05,
      "loss": 1.7965,
      "step": 1036
    },
    {
      "epoch": 0.0797480024585126,
      "grad_norm": 5.124858379364014,
      "learning_rate": 1.9681007990165953e-05,
      "loss": 1.6745,
      "step": 1038
    },
    {
      "epoch": 0.07990165949600492,
      "grad_norm": 4.530187606811523,
      "learning_rate": 1.9680393362015983e-05,
      "loss": 1.7152,
      "step": 1040
    },
    {
      "epoch": 0.08005531653349723,
      "grad_norm": 4.918298244476318,
      "learning_rate": 1.9679778733866012e-05,
      "loss": 1.7445,
      "step": 1042
    },
    {
      "epoch": 0.08020897357098955,
      "grad_norm": 4.637542247772217,
      "learning_rate": 1.9679164105716045e-05,
      "loss": 1.6802,
      "step": 1044
    },
    {
      "epoch": 0.08036263060848187,
      "grad_norm": 5.314078330993652,
      "learning_rate": 1.9678549477566075e-05,
      "loss": 1.7547,
      "step": 1046
    },
    {
      "epoch": 0.08051628764597418,
      "grad_norm": 5.02266263961792,
      "learning_rate": 1.9677934849416105e-05,
      "loss": 1.7107,
      "step": 1048
    },
    {
      "epoch": 0.0806699446834665,
      "grad_norm": 4.683084487915039,
      "learning_rate": 1.9677320221266138e-05,
      "loss": 1.7415,
      "step": 1050
    },
    {
      "epoch": 0.08082360172095882,
      "grad_norm": 4.539281845092773,
      "learning_rate": 1.9676705593116164e-05,
      "loss": 1.9356,
      "step": 1052
    },
    {
      "epoch": 0.08097725875845113,
      "grad_norm": 4.526500701904297,
      "learning_rate": 1.9676090964966197e-05,
      "loss": 1.6487,
      "step": 1054
    },
    {
      "epoch": 0.08113091579594345,
      "grad_norm": 4.272531509399414,
      "learning_rate": 1.9675476336816227e-05,
      "loss": 1.6478,
      "step": 1056
    },
    {
      "epoch": 0.08128457283343578,
      "grad_norm": 4.874919414520264,
      "learning_rate": 1.9674861708666257e-05,
      "loss": 1.715,
      "step": 1058
    },
    {
      "epoch": 0.0814382298709281,
      "grad_norm": 5.863064765930176,
      "learning_rate": 1.967424708051629e-05,
      "loss": 1.8635,
      "step": 1060
    },
    {
      "epoch": 0.08159188690842041,
      "grad_norm": 5.5292205810546875,
      "learning_rate": 1.967363245236632e-05,
      "loss": 1.7227,
      "step": 1062
    },
    {
      "epoch": 0.08174554394591273,
      "grad_norm": 4.343425273895264,
      "learning_rate": 1.9673017824216352e-05,
      "loss": 1.8257,
      "step": 1064
    },
    {
      "epoch": 0.08189920098340504,
      "grad_norm": 5.900411128997803,
      "learning_rate": 1.9672403196066382e-05,
      "loss": 1.8205,
      "step": 1066
    },
    {
      "epoch": 0.08205285802089736,
      "grad_norm": 4.439347743988037,
      "learning_rate": 1.9671788567916412e-05,
      "loss": 1.6958,
      "step": 1068
    },
    {
      "epoch": 0.08220651505838968,
      "grad_norm": 5.694681644439697,
      "learning_rate": 1.9671173939766445e-05,
      "loss": 1.8959,
      "step": 1070
    },
    {
      "epoch": 0.08236017209588199,
      "grad_norm": 4.767448425292969,
      "learning_rate": 1.9670559311616474e-05,
      "loss": 1.6721,
      "step": 1072
    },
    {
      "epoch": 0.08251382913337431,
      "grad_norm": 4.426461696624756,
      "learning_rate": 1.9669944683466504e-05,
      "loss": 1.607,
      "step": 1074
    },
    {
      "epoch": 0.08266748617086662,
      "grad_norm": 4.794045925140381,
      "learning_rate": 1.9669330055316537e-05,
      "loss": 1.6254,
      "step": 1076
    },
    {
      "epoch": 0.08282114320835894,
      "grad_norm": 4.544212341308594,
      "learning_rate": 1.9668715427166564e-05,
      "loss": 1.6082,
      "step": 1078
    },
    {
      "epoch": 0.08297480024585126,
      "grad_norm": 4.256971836090088,
      "learning_rate": 1.9668100799016597e-05,
      "loss": 1.8001,
      "step": 1080
    },
    {
      "epoch": 0.08312845728334357,
      "grad_norm": 6.058056831359863,
      "learning_rate": 1.9667486170866626e-05,
      "loss": 1.7258,
      "step": 1082
    },
    {
      "epoch": 0.08328211432083589,
      "grad_norm": 4.815703868865967,
      "learning_rate": 1.966687154271666e-05,
      "loss": 1.7528,
      "step": 1084
    },
    {
      "epoch": 0.0834357713583282,
      "grad_norm": 4.661309719085693,
      "learning_rate": 1.966625691456669e-05,
      "loss": 1.7316,
      "step": 1086
    },
    {
      "epoch": 0.08358942839582052,
      "grad_norm": 4.863770961761475,
      "learning_rate": 1.966564228641672e-05,
      "loss": 1.8824,
      "step": 1088
    },
    {
      "epoch": 0.08374308543331284,
      "grad_norm": 5.061347961425781,
      "learning_rate": 1.9665027658266752e-05,
      "loss": 1.6873,
      "step": 1090
    },
    {
      "epoch": 0.08389674247080517,
      "grad_norm": 4.252581596374512,
      "learning_rate": 1.966441303011678e-05,
      "loss": 1.8093,
      "step": 1092
    },
    {
      "epoch": 0.08405039950829749,
      "grad_norm": 5.112542152404785,
      "learning_rate": 1.966379840196681e-05,
      "loss": 1.7366,
      "step": 1094
    },
    {
      "epoch": 0.0842040565457898,
      "grad_norm": 5.147494792938232,
      "learning_rate": 1.9663183773816844e-05,
      "loss": 1.8676,
      "step": 1096
    },
    {
      "epoch": 0.08435771358328212,
      "grad_norm": 4.757107257843018,
      "learning_rate": 1.966256914566687e-05,
      "loss": 1.5888,
      "step": 1098
    },
    {
      "epoch": 0.08451137062077443,
      "grad_norm": 5.5242600440979,
      "learning_rate": 1.9661954517516904e-05,
      "loss": 1.7018,
      "step": 1100
    },
    {
      "epoch": 0.08466502765826675,
      "grad_norm": 5.5325117111206055,
      "learning_rate": 1.9661339889366933e-05,
      "loss": 1.6928,
      "step": 1102
    },
    {
      "epoch": 0.08481868469575907,
      "grad_norm": 5.308017253875732,
      "learning_rate": 1.9660725261216966e-05,
      "loss": 1.7695,
      "step": 1104
    },
    {
      "epoch": 0.08497234173325138,
      "grad_norm": 4.460916519165039,
      "learning_rate": 1.9660110633066996e-05,
      "loss": 1.649,
      "step": 1106
    },
    {
      "epoch": 0.0851259987707437,
      "grad_norm": 5.222771644592285,
      "learning_rate": 1.9659496004917026e-05,
      "loss": 1.8017,
      "step": 1108
    },
    {
      "epoch": 0.08527965580823602,
      "grad_norm": 4.484593391418457,
      "learning_rate": 1.965888137676706e-05,
      "loss": 1.697,
      "step": 1110
    },
    {
      "epoch": 0.08543331284572833,
      "grad_norm": 4.95808219909668,
      "learning_rate": 1.965826674861709e-05,
      "loss": 1.6171,
      "step": 1112
    },
    {
      "epoch": 0.08558696988322065,
      "grad_norm": 5.313704967498779,
      "learning_rate": 1.9657652120467118e-05,
      "loss": 1.665,
      "step": 1114
    },
    {
      "epoch": 0.08574062692071296,
      "grad_norm": 4.555895805358887,
      "learning_rate": 1.965703749231715e-05,
      "loss": 1.7004,
      "step": 1116
    },
    {
      "epoch": 0.08589428395820528,
      "grad_norm": 4.939544677734375,
      "learning_rate": 1.965642286416718e-05,
      "loss": 1.7204,
      "step": 1118
    },
    {
      "epoch": 0.0860479409956976,
      "grad_norm": 4.291268348693848,
      "learning_rate": 1.965580823601721e-05,
      "loss": 1.6592,
      "step": 1120
    },
    {
      "epoch": 0.08620159803318991,
      "grad_norm": 5.050233840942383,
      "learning_rate": 1.9655193607867244e-05,
      "loss": 1.7656,
      "step": 1122
    },
    {
      "epoch": 0.08635525507068224,
      "grad_norm": 4.5748090744018555,
      "learning_rate": 1.9654578979717273e-05,
      "loss": 1.8491,
      "step": 1124
    },
    {
      "epoch": 0.08650891210817456,
      "grad_norm": 4.3803391456604,
      "learning_rate": 1.9653964351567303e-05,
      "loss": 1.697,
      "step": 1126
    },
    {
      "epoch": 0.08666256914566688,
      "grad_norm": 5.346717834472656,
      "learning_rate": 1.9653349723417333e-05,
      "loss": 1.817,
      "step": 1128
    },
    {
      "epoch": 0.08681622618315919,
      "grad_norm": 7.8311944007873535,
      "learning_rate": 1.9652735095267366e-05,
      "loss": 1.7637,
      "step": 1130
    },
    {
      "epoch": 0.08696988322065151,
      "grad_norm": 4.9822845458984375,
      "learning_rate": 1.9652120467117395e-05,
      "loss": 1.6673,
      "step": 1132
    },
    {
      "epoch": 0.08712354025814383,
      "grad_norm": 5.492576599121094,
      "learning_rate": 1.9651505838967425e-05,
      "loss": 1.7831,
      "step": 1134
    },
    {
      "epoch": 0.08727719729563614,
      "grad_norm": 5.493293285369873,
      "learning_rate": 1.9650891210817458e-05,
      "loss": 1.698,
      "step": 1136
    },
    {
      "epoch": 0.08743085433312846,
      "grad_norm": 6.50536584854126,
      "learning_rate": 1.9650276582667488e-05,
      "loss": 1.7632,
      "step": 1138
    },
    {
      "epoch": 0.08758451137062077,
      "grad_norm": 4.753702640533447,
      "learning_rate": 1.9649661954517518e-05,
      "loss": 1.64,
      "step": 1140
    },
    {
      "epoch": 0.08773816840811309,
      "grad_norm": 4.77907657623291,
      "learning_rate": 1.964904732636755e-05,
      "loss": 1.9073,
      "step": 1142
    },
    {
      "epoch": 0.0878918254456054,
      "grad_norm": 4.242752552032471,
      "learning_rate": 1.964843269821758e-05,
      "loss": 1.7625,
      "step": 1144
    },
    {
      "epoch": 0.08804548248309772,
      "grad_norm": 5.127992630004883,
      "learning_rate": 1.964781807006761e-05,
      "loss": 1.6271,
      "step": 1146
    },
    {
      "epoch": 0.08819913952059004,
      "grad_norm": 4.7297749519348145,
      "learning_rate": 1.9647203441917643e-05,
      "loss": 1.6252,
      "step": 1148
    },
    {
      "epoch": 0.08835279655808236,
      "grad_norm": 5.08091926574707,
      "learning_rate": 1.9646588813767673e-05,
      "loss": 1.7145,
      "step": 1150
    },
    {
      "epoch": 0.08850645359557467,
      "grad_norm": 4.986452102661133,
      "learning_rate": 1.9645974185617702e-05,
      "loss": 1.8065,
      "step": 1152
    },
    {
      "epoch": 0.08866011063306699,
      "grad_norm": 5.4139204025268555,
      "learning_rate": 1.9645359557467732e-05,
      "loss": 1.6284,
      "step": 1154
    },
    {
      "epoch": 0.08881376767055932,
      "grad_norm": 5.257991313934326,
      "learning_rate": 1.9644744929317765e-05,
      "loss": 1.8338,
      "step": 1156
    },
    {
      "epoch": 0.08896742470805163,
      "grad_norm": 4.872262001037598,
      "learning_rate": 1.9644130301167795e-05,
      "loss": 1.8488,
      "step": 1158
    },
    {
      "epoch": 0.08912108174554395,
      "grad_norm": 5.82914924621582,
      "learning_rate": 1.9643515673017825e-05,
      "loss": 1.7731,
      "step": 1160
    },
    {
      "epoch": 0.08927473878303627,
      "grad_norm": 4.288515090942383,
      "learning_rate": 1.9642901044867858e-05,
      "loss": 1.6748,
      "step": 1162
    },
    {
      "epoch": 0.08942839582052858,
      "grad_norm": 4.828827381134033,
      "learning_rate": 1.9642286416717887e-05,
      "loss": 1.7893,
      "step": 1164
    },
    {
      "epoch": 0.0895820528580209,
      "grad_norm": 4.2209577560424805,
      "learning_rate": 1.9641671788567917e-05,
      "loss": 1.6139,
      "step": 1166
    },
    {
      "epoch": 0.08973570989551322,
      "grad_norm": 5.542986869812012,
      "learning_rate": 1.964105716041795e-05,
      "loss": 1.5873,
      "step": 1168
    },
    {
      "epoch": 0.08988936693300553,
      "grad_norm": 4.995078086853027,
      "learning_rate": 1.964044253226798e-05,
      "loss": 1.5976,
      "step": 1170
    },
    {
      "epoch": 0.09004302397049785,
      "grad_norm": 4.493627071380615,
      "learning_rate": 1.963982790411801e-05,
      "loss": 1.6423,
      "step": 1172
    },
    {
      "epoch": 0.09019668100799016,
      "grad_norm": 5.041021823883057,
      "learning_rate": 1.9639213275968043e-05,
      "loss": 1.6996,
      "step": 1174
    },
    {
      "epoch": 0.09035033804548248,
      "grad_norm": 5.527616024017334,
      "learning_rate": 1.9638598647818072e-05,
      "loss": 1.7338,
      "step": 1176
    },
    {
      "epoch": 0.0905039950829748,
      "grad_norm": 4.937674045562744,
      "learning_rate": 1.9637984019668102e-05,
      "loss": 1.8472,
      "step": 1178
    },
    {
      "epoch": 0.09065765212046711,
      "grad_norm": 5.515735149383545,
      "learning_rate": 1.963736939151813e-05,
      "loss": 1.8828,
      "step": 1180
    },
    {
      "epoch": 0.09081130915795943,
      "grad_norm": 5.273295879364014,
      "learning_rate": 1.9636754763368165e-05,
      "loss": 1.6819,
      "step": 1182
    },
    {
      "epoch": 0.09096496619545175,
      "grad_norm": 4.745762825012207,
      "learning_rate": 1.9636140135218194e-05,
      "loss": 1.7752,
      "step": 1184
    },
    {
      "epoch": 0.09111862323294406,
      "grad_norm": 4.8165130615234375,
      "learning_rate": 1.9635525507068224e-05,
      "loss": 1.6133,
      "step": 1186
    },
    {
      "epoch": 0.09127228027043639,
      "grad_norm": 4.156392574310303,
      "learning_rate": 1.9634910878918257e-05,
      "loss": 1.7126,
      "step": 1188
    },
    {
      "epoch": 0.09142593730792871,
      "grad_norm": 4.523599624633789,
      "learning_rate": 1.9634296250768287e-05,
      "loss": 1.616,
      "step": 1190
    },
    {
      "epoch": 0.09157959434542103,
      "grad_norm": 4.08304500579834,
      "learning_rate": 1.9633681622618316e-05,
      "loss": 1.6602,
      "step": 1192
    },
    {
      "epoch": 0.09173325138291334,
      "grad_norm": 4.6315202713012695,
      "learning_rate": 1.963306699446835e-05,
      "loss": 1.7004,
      "step": 1194
    },
    {
      "epoch": 0.09188690842040566,
      "grad_norm": 4.613175868988037,
      "learning_rate": 1.963245236631838e-05,
      "loss": 1.7607,
      "step": 1196
    },
    {
      "epoch": 0.09204056545789797,
      "grad_norm": 5.473268508911133,
      "learning_rate": 1.963183773816841e-05,
      "loss": 1.7788,
      "step": 1198
    },
    {
      "epoch": 0.09219422249539029,
      "grad_norm": 4.436158657073975,
      "learning_rate": 1.963122311001844e-05,
      "loss": 1.507,
      "step": 1200
    },
    {
      "epoch": 0.0923478795328826,
      "grad_norm": 4.081661701202393,
      "learning_rate": 1.963060848186847e-05,
      "loss": 1.6724,
      "step": 1202
    },
    {
      "epoch": 0.09250153657037492,
      "grad_norm": 4.4669318199157715,
      "learning_rate": 1.96299938537185e-05,
      "loss": 1.6776,
      "step": 1204
    },
    {
      "epoch": 0.09265519360786724,
      "grad_norm": 4.446565628051758,
      "learning_rate": 1.962937922556853e-05,
      "loss": 1.7225,
      "step": 1206
    },
    {
      "epoch": 0.09280885064535956,
      "grad_norm": 4.1054816246032715,
      "learning_rate": 1.9628764597418564e-05,
      "loss": 1.5964,
      "step": 1208
    },
    {
      "epoch": 0.09296250768285187,
      "grad_norm": 3.9810330867767334,
      "learning_rate": 1.9628149969268594e-05,
      "loss": 1.6933,
      "step": 1210
    },
    {
      "epoch": 0.09311616472034419,
      "grad_norm": 5.808743476867676,
      "learning_rate": 1.9627535341118623e-05,
      "loss": 1.796,
      "step": 1212
    },
    {
      "epoch": 0.0932698217578365,
      "grad_norm": 5.4757795333862305,
      "learning_rate": 1.9626920712968657e-05,
      "loss": 1.6707,
      "step": 1214
    },
    {
      "epoch": 0.09342347879532882,
      "grad_norm": 4.4285430908203125,
      "learning_rate": 1.9626306084818686e-05,
      "loss": 1.6214,
      "step": 1216
    },
    {
      "epoch": 0.09357713583282114,
      "grad_norm": 4.347642421722412,
      "learning_rate": 1.9625691456668716e-05,
      "loss": 1.7012,
      "step": 1218
    },
    {
      "epoch": 0.09373079287031345,
      "grad_norm": 4.711369037628174,
      "learning_rate": 1.962507682851875e-05,
      "loss": 1.6307,
      "step": 1220
    },
    {
      "epoch": 0.09388444990780578,
      "grad_norm": 4.427557945251465,
      "learning_rate": 1.962446220036878e-05,
      "loss": 1.7988,
      "step": 1222
    },
    {
      "epoch": 0.0940381069452981,
      "grad_norm": 4.078681945800781,
      "learning_rate": 1.962384757221881e-05,
      "loss": 1.6596,
      "step": 1224
    },
    {
      "epoch": 0.09419176398279042,
      "grad_norm": 4.403939247131348,
      "learning_rate": 1.9623232944068838e-05,
      "loss": 1.7761,
      "step": 1226
    },
    {
      "epoch": 0.09434542102028273,
      "grad_norm": 4.222808361053467,
      "learning_rate": 1.962261831591887e-05,
      "loss": 1.7145,
      "step": 1228
    },
    {
      "epoch": 0.09449907805777505,
      "grad_norm": 4.754458904266357,
      "learning_rate": 1.96220036877689e-05,
      "loss": 1.6256,
      "step": 1230
    },
    {
      "epoch": 0.09465273509526737,
      "grad_norm": 4.7824201583862305,
      "learning_rate": 1.962138905961893e-05,
      "loss": 1.7846,
      "step": 1232
    },
    {
      "epoch": 0.09480639213275968,
      "grad_norm": 4.751527786254883,
      "learning_rate": 1.9620774431468964e-05,
      "loss": 1.6602,
      "step": 1234
    },
    {
      "epoch": 0.094960049170252,
      "grad_norm": 4.74616813659668,
      "learning_rate": 1.9620159803318993e-05,
      "loss": 1.6707,
      "step": 1236
    },
    {
      "epoch": 0.09511370620774431,
      "grad_norm": 5.099863052368164,
      "learning_rate": 1.9619545175169023e-05,
      "loss": 1.7293,
      "step": 1238
    },
    {
      "epoch": 0.09526736324523663,
      "grad_norm": 5.293537616729736,
      "learning_rate": 1.9618930547019056e-05,
      "loss": 1.7172,
      "step": 1240
    },
    {
      "epoch": 0.09542102028272895,
      "grad_norm": 5.443637847900391,
      "learning_rate": 1.9618315918869086e-05,
      "loss": 1.7516,
      "step": 1242
    },
    {
      "epoch": 0.09557467732022126,
      "grad_norm": 4.447843551635742,
      "learning_rate": 1.9617701290719115e-05,
      "loss": 1.7449,
      "step": 1244
    },
    {
      "epoch": 0.09572833435771358,
      "grad_norm": 4.490113258361816,
      "learning_rate": 1.961708666256915e-05,
      "loss": 1.6253,
      "step": 1246
    },
    {
      "epoch": 0.0958819913952059,
      "grad_norm": 4.979306221008301,
      "learning_rate": 1.9616472034419178e-05,
      "loss": 1.6743,
      "step": 1248
    },
    {
      "epoch": 0.09603564843269821,
      "grad_norm": 4.146381855010986,
      "learning_rate": 1.961585740626921e-05,
      "loss": 1.7014,
      "step": 1250
    },
    {
      "epoch": 0.09618930547019053,
      "grad_norm": 4.571809768676758,
      "learning_rate": 1.9615242778119237e-05,
      "loss": 1.7867,
      "step": 1252
    },
    {
      "epoch": 0.09634296250768286,
      "grad_norm": 4.7382988929748535,
      "learning_rate": 1.961462814996927e-05,
      "loss": 1.7669,
      "step": 1254
    },
    {
      "epoch": 0.09649661954517517,
      "grad_norm": 4.332629203796387,
      "learning_rate": 1.96140135218193e-05,
      "loss": 1.8072,
      "step": 1256
    },
    {
      "epoch": 0.09665027658266749,
      "grad_norm": 4.376523494720459,
      "learning_rate": 1.961339889366933e-05,
      "loss": 1.7595,
      "step": 1258
    },
    {
      "epoch": 0.09680393362015981,
      "grad_norm": 4.876426696777344,
      "learning_rate": 1.9612784265519363e-05,
      "loss": 1.5174,
      "step": 1260
    },
    {
      "epoch": 0.09695759065765212,
      "grad_norm": 4.8033905029296875,
      "learning_rate": 1.9612169637369393e-05,
      "loss": 1.7284,
      "step": 1262
    },
    {
      "epoch": 0.09711124769514444,
      "grad_norm": 4.221518039703369,
      "learning_rate": 1.9611555009219422e-05,
      "loss": 1.7409,
      "step": 1264
    },
    {
      "epoch": 0.09726490473263676,
      "grad_norm": 4.445687294006348,
      "learning_rate": 1.9610940381069455e-05,
      "loss": 1.6211,
      "step": 1266
    },
    {
      "epoch": 0.09741856177012907,
      "grad_norm": 4.590234279632568,
      "learning_rate": 1.9610325752919485e-05,
      "loss": 1.5695,
      "step": 1268
    },
    {
      "epoch": 0.09757221880762139,
      "grad_norm": 5.60252571105957,
      "learning_rate": 1.9609711124769518e-05,
      "loss": 1.6369,
      "step": 1270
    },
    {
      "epoch": 0.0977258758451137,
      "grad_norm": 4.938035011291504,
      "learning_rate": 1.9609096496619548e-05,
      "loss": 1.7329,
      "step": 1272
    },
    {
      "epoch": 0.09787953288260602,
      "grad_norm": 5.1367106437683105,
      "learning_rate": 1.9608481868469578e-05,
      "loss": 1.8568,
      "step": 1274
    },
    {
      "epoch": 0.09803318992009834,
      "grad_norm": 4.405098915100098,
      "learning_rate": 1.960786724031961e-05,
      "loss": 1.6226,
      "step": 1276
    },
    {
      "epoch": 0.09818684695759065,
      "grad_norm": 5.822478771209717,
      "learning_rate": 1.9607252612169637e-05,
      "loss": 1.7561,
      "step": 1278
    },
    {
      "epoch": 0.09834050399508297,
      "grad_norm": 4.770538806915283,
      "learning_rate": 1.960663798401967e-05,
      "loss": 1.6214,
      "step": 1280
    },
    {
      "epoch": 0.09849416103257529,
      "grad_norm": 6.316437244415283,
      "learning_rate": 1.96060233558697e-05,
      "loss": 1.8268,
      "step": 1282
    },
    {
      "epoch": 0.0986478180700676,
      "grad_norm": 4.640567302703857,
      "learning_rate": 1.960540872771973e-05,
      "loss": 1.7484,
      "step": 1284
    },
    {
      "epoch": 0.09880147510755993,
      "grad_norm": 4.596543312072754,
      "learning_rate": 1.9604794099569762e-05,
      "loss": 1.6986,
      "step": 1286
    },
    {
      "epoch": 0.09895513214505225,
      "grad_norm": 4.36724328994751,
      "learning_rate": 1.9604179471419792e-05,
      "loss": 1.6108,
      "step": 1288
    },
    {
      "epoch": 0.09910878918254457,
      "grad_norm": 5.017337322235107,
      "learning_rate": 1.9603564843269825e-05,
      "loss": 1.7047,
      "step": 1290
    },
    {
      "epoch": 0.09926244622003688,
      "grad_norm": 4.327188491821289,
      "learning_rate": 1.9602950215119855e-05,
      "loss": 1.6083,
      "step": 1292
    },
    {
      "epoch": 0.0994161032575292,
      "grad_norm": 5.734022617340088,
      "learning_rate": 1.9602335586969885e-05,
      "loss": 1.7501,
      "step": 1294
    },
    {
      "epoch": 0.09956976029502151,
      "grad_norm": 4.524082183837891,
      "learning_rate": 1.9601720958819918e-05,
      "loss": 1.7556,
      "step": 1296
    },
    {
      "epoch": 0.09972341733251383,
      "grad_norm": 4.61295747756958,
      "learning_rate": 1.9601106330669947e-05,
      "loss": 1.6598,
      "step": 1298
    },
    {
      "epoch": 0.09987707437000615,
      "grad_norm": 4.453684329986572,
      "learning_rate": 1.9600491702519977e-05,
      "loss": 1.6554,
      "step": 1300
    },
    {
      "epoch": 0.10003073140749846,
      "grad_norm": 4.732148170471191,
      "learning_rate": 1.959987707437001e-05,
      "loss": 1.7181,
      "step": 1302
    },
    {
      "epoch": 0.10018438844499078,
      "grad_norm": 4.715574741363525,
      "learning_rate": 1.9599262446220036e-05,
      "loss": 1.6849,
      "step": 1304
    },
    {
      "epoch": 0.1003380454824831,
      "grad_norm": 4.356414318084717,
      "learning_rate": 1.959864781807007e-05,
      "loss": 1.687,
      "step": 1306
    },
    {
      "epoch": 0.10049170251997541,
      "grad_norm": 4.813374996185303,
      "learning_rate": 1.95980331899201e-05,
      "loss": 1.5262,
      "step": 1308
    },
    {
      "epoch": 0.10064535955746773,
      "grad_norm": 4.9926981925964355,
      "learning_rate": 1.959741856177013e-05,
      "loss": 1.8137,
      "step": 1310
    },
    {
      "epoch": 0.10079901659496004,
      "grad_norm": 5.103787422180176,
      "learning_rate": 1.9596803933620162e-05,
      "loss": 1.64,
      "step": 1312
    },
    {
      "epoch": 0.10095267363245236,
      "grad_norm": 4.895768165588379,
      "learning_rate": 1.959618930547019e-05,
      "loss": 1.693,
      "step": 1314
    },
    {
      "epoch": 0.10110633066994468,
      "grad_norm": 4.513470649719238,
      "learning_rate": 1.9595574677320225e-05,
      "loss": 1.5745,
      "step": 1316
    },
    {
      "epoch": 0.10125998770743701,
      "grad_norm": 5.475149154663086,
      "learning_rate": 1.9594960049170254e-05,
      "loss": 1.6189,
      "step": 1318
    },
    {
      "epoch": 0.10141364474492932,
      "grad_norm": 4.828972339630127,
      "learning_rate": 1.9594345421020284e-05,
      "loss": 1.6547,
      "step": 1320
    },
    {
      "epoch": 0.10156730178242164,
      "grad_norm": 5.218929290771484,
      "learning_rate": 1.9593730792870317e-05,
      "loss": 1.6494,
      "step": 1322
    },
    {
      "epoch": 0.10172095881991396,
      "grad_norm": 4.358766078948975,
      "learning_rate": 1.9593116164720343e-05,
      "loss": 1.7283,
      "step": 1324
    },
    {
      "epoch": 0.10187461585740627,
      "grad_norm": 4.14285135269165,
      "learning_rate": 1.9592501536570376e-05,
      "loss": 1.7769,
      "step": 1326
    },
    {
      "epoch": 0.10202827289489859,
      "grad_norm": 4.319285869598389,
      "learning_rate": 1.9591886908420406e-05,
      "loss": 1.4707,
      "step": 1328
    },
    {
      "epoch": 0.1021819299323909,
      "grad_norm": 5.230128288269043,
      "learning_rate": 1.9591272280270436e-05,
      "loss": 1.4906,
      "step": 1330
    },
    {
      "epoch": 0.10233558696988322,
      "grad_norm": 5.243448257446289,
      "learning_rate": 1.959065765212047e-05,
      "loss": 1.8825,
      "step": 1332
    },
    {
      "epoch": 0.10248924400737554,
      "grad_norm": 4.784072399139404,
      "learning_rate": 1.95900430239705e-05,
      "loss": 1.7553,
      "step": 1334
    },
    {
      "epoch": 0.10264290104486785,
      "grad_norm": 5.595427513122559,
      "learning_rate": 1.958942839582053e-05,
      "loss": 1.6913,
      "step": 1336
    },
    {
      "epoch": 0.10279655808236017,
      "grad_norm": 4.856276512145996,
      "learning_rate": 1.958881376767056e-05,
      "loss": 1.7659,
      "step": 1338
    },
    {
      "epoch": 0.10295021511985249,
      "grad_norm": 5.188042640686035,
      "learning_rate": 1.958819913952059e-05,
      "loss": 1.8205,
      "step": 1340
    },
    {
      "epoch": 0.1031038721573448,
      "grad_norm": 4.261306285858154,
      "learning_rate": 1.9587584511370624e-05,
      "loss": 1.5887,
      "step": 1342
    },
    {
      "epoch": 0.10325752919483712,
      "grad_norm": 4.269975185394287,
      "learning_rate": 1.9586969883220654e-05,
      "loss": 1.5869,
      "step": 1344
    },
    {
      "epoch": 0.10341118623232944,
      "grad_norm": 5.029308795928955,
      "learning_rate": 1.9586355255070683e-05,
      "loss": 1.8049,
      "step": 1346
    },
    {
      "epoch": 0.10356484326982175,
      "grad_norm": 4.857789516448975,
      "learning_rate": 1.9585740626920716e-05,
      "loss": 1.5154,
      "step": 1348
    },
    {
      "epoch": 0.10371850030731407,
      "grad_norm": 4.701939582824707,
      "learning_rate": 1.9585125998770743e-05,
      "loss": 1.6822,
      "step": 1350
    },
    {
      "epoch": 0.1038721573448064,
      "grad_norm": 4.069787979125977,
      "learning_rate": 1.9584511370620776e-05,
      "loss": 1.7238,
      "step": 1352
    },
    {
      "epoch": 0.10402581438229871,
      "grad_norm": 4.703420162200928,
      "learning_rate": 1.9583896742470806e-05,
      "loss": 1.6951,
      "step": 1354
    },
    {
      "epoch": 0.10417947141979103,
      "grad_norm": 4.920733451843262,
      "learning_rate": 1.958328211432084e-05,
      "loss": 1.6335,
      "step": 1356
    },
    {
      "epoch": 0.10433312845728335,
      "grad_norm": 4.38323974609375,
      "learning_rate": 1.9582667486170868e-05,
      "loss": 1.7094,
      "step": 1358
    },
    {
      "epoch": 0.10448678549477566,
      "grad_norm": 4.646501541137695,
      "learning_rate": 1.9582052858020898e-05,
      "loss": 1.6878,
      "step": 1360
    },
    {
      "epoch": 0.10464044253226798,
      "grad_norm": 4.569819450378418,
      "learning_rate": 1.958143822987093e-05,
      "loss": 1.7198,
      "step": 1362
    },
    {
      "epoch": 0.1047940995697603,
      "grad_norm": 4.552595615386963,
      "learning_rate": 1.958082360172096e-05,
      "loss": 1.7335,
      "step": 1364
    },
    {
      "epoch": 0.10494775660725261,
      "grad_norm": 3.9051506519317627,
      "learning_rate": 1.958020897357099e-05,
      "loss": 1.5694,
      "step": 1366
    },
    {
      "epoch": 0.10510141364474493,
      "grad_norm": 3.9420411586761475,
      "learning_rate": 1.9579594345421023e-05,
      "loss": 1.7298,
      "step": 1368
    },
    {
      "epoch": 0.10525507068223725,
      "grad_norm": 4.996294021606445,
      "learning_rate": 1.9578979717271053e-05,
      "loss": 1.7769,
      "step": 1370
    },
    {
      "epoch": 0.10540872771972956,
      "grad_norm": 4.845794677734375,
      "learning_rate": 1.9578365089121083e-05,
      "loss": 1.694,
      "step": 1372
    },
    {
      "epoch": 0.10556238475722188,
      "grad_norm": 4.156089782714844,
      "learning_rate": 1.9577750460971116e-05,
      "loss": 1.5621,
      "step": 1374
    },
    {
      "epoch": 0.1057160417947142,
      "grad_norm": 5.298906326293945,
      "learning_rate": 1.9577135832821146e-05,
      "loss": 1.6016,
      "step": 1376
    },
    {
      "epoch": 0.10586969883220651,
      "grad_norm": 4.974923610687256,
      "learning_rate": 1.9576521204671175e-05,
      "loss": 1.9024,
      "step": 1378
    },
    {
      "epoch": 0.10602335586969883,
      "grad_norm": 4.5802998542785645,
      "learning_rate": 1.9575906576521205e-05,
      "loss": 1.8249,
      "step": 1380
    },
    {
      "epoch": 0.10617701290719114,
      "grad_norm": 5.364488124847412,
      "learning_rate": 1.9575291948371238e-05,
      "loss": 1.6205,
      "step": 1382
    },
    {
      "epoch": 0.10633066994468347,
      "grad_norm": 4.810891151428223,
      "learning_rate": 1.9574677320221268e-05,
      "loss": 1.6702,
      "step": 1384
    },
    {
      "epoch": 0.10648432698217579,
      "grad_norm": 5.155327320098877,
      "learning_rate": 1.9574062692071297e-05,
      "loss": 1.6251,
      "step": 1386
    },
    {
      "epoch": 0.1066379840196681,
      "grad_norm": 4.292688369750977,
      "learning_rate": 1.957344806392133e-05,
      "loss": 1.505,
      "step": 1388
    },
    {
      "epoch": 0.10679164105716042,
      "grad_norm": 4.611319541931152,
      "learning_rate": 1.957283343577136e-05,
      "loss": 1.7301,
      "step": 1390
    },
    {
      "epoch": 0.10694529809465274,
      "grad_norm": 4.324422359466553,
      "learning_rate": 1.957221880762139e-05,
      "loss": 1.6551,
      "step": 1392
    },
    {
      "epoch": 0.10709895513214505,
      "grad_norm": 4.826112747192383,
      "learning_rate": 1.9571604179471423e-05,
      "loss": 1.607,
      "step": 1394
    },
    {
      "epoch": 0.10725261216963737,
      "grad_norm": 4.303924560546875,
      "learning_rate": 1.9570989551321453e-05,
      "loss": 1.6128,
      "step": 1396
    },
    {
      "epoch": 0.10740626920712969,
      "grad_norm": 5.093891620635986,
      "learning_rate": 1.9570374923171482e-05,
      "loss": 1.6747,
      "step": 1398
    },
    {
      "epoch": 0.107559926244622,
      "grad_norm": 4.303253650665283,
      "learning_rate": 1.9569760295021515e-05,
      "loss": 1.5981,
      "step": 1400
    },
    {
      "epoch": 0.10771358328211432,
      "grad_norm": 4.165480613708496,
      "learning_rate": 1.9569145666871545e-05,
      "loss": 1.5993,
      "step": 1402
    },
    {
      "epoch": 0.10786724031960664,
      "grad_norm": 4.655346393585205,
      "learning_rate": 1.9568531038721575e-05,
      "loss": 1.6806,
      "step": 1404
    },
    {
      "epoch": 0.10802089735709895,
      "grad_norm": 4.743736743927002,
      "learning_rate": 1.9567916410571604e-05,
      "loss": 1.6624,
      "step": 1406
    },
    {
      "epoch": 0.10817455439459127,
      "grad_norm": 4.2791643142700195,
      "learning_rate": 1.9567301782421637e-05,
      "loss": 1.6776,
      "step": 1408
    },
    {
      "epoch": 0.10832821143208358,
      "grad_norm": 5.005465030670166,
      "learning_rate": 1.9566687154271667e-05,
      "loss": 1.5962,
      "step": 1410
    },
    {
      "epoch": 0.1084818684695759,
      "grad_norm": 4.345304012298584,
      "learning_rate": 1.9566072526121697e-05,
      "loss": 1.5865,
      "step": 1412
    },
    {
      "epoch": 0.10863552550706822,
      "grad_norm": 3.8712103366851807,
      "learning_rate": 1.956545789797173e-05,
      "loss": 1.6583,
      "step": 1414
    },
    {
      "epoch": 0.10878918254456055,
      "grad_norm": 4.381411075592041,
      "learning_rate": 1.956484326982176e-05,
      "loss": 1.6203,
      "step": 1416
    },
    {
      "epoch": 0.10894283958205286,
      "grad_norm": 3.933609962463379,
      "learning_rate": 1.956422864167179e-05,
      "loss": 1.7631,
      "step": 1418
    },
    {
      "epoch": 0.10909649661954518,
      "grad_norm": 5.570189952850342,
      "learning_rate": 1.9563614013521822e-05,
      "loss": 1.7173,
      "step": 1420
    },
    {
      "epoch": 0.1092501536570375,
      "grad_norm": 4.816314220428467,
      "learning_rate": 1.9562999385371852e-05,
      "loss": 1.6941,
      "step": 1422
    },
    {
      "epoch": 0.10940381069452981,
      "grad_norm": 4.110052585601807,
      "learning_rate": 1.9562384757221882e-05,
      "loss": 1.6782,
      "step": 1424
    },
    {
      "epoch": 0.10955746773202213,
      "grad_norm": 4.069727420806885,
      "learning_rate": 1.956177012907191e-05,
      "loss": 1.657,
      "step": 1426
    },
    {
      "epoch": 0.10971112476951445,
      "grad_norm": 5.244446277618408,
      "learning_rate": 1.9561155500921944e-05,
      "loss": 1.683,
      "step": 1428
    },
    {
      "epoch": 0.10986478180700676,
      "grad_norm": 5.359142780303955,
      "learning_rate": 1.9560540872771974e-05,
      "loss": 1.6764,
      "step": 1430
    },
    {
      "epoch": 0.11001843884449908,
      "grad_norm": 5.057417869567871,
      "learning_rate": 1.9559926244622004e-05,
      "loss": 1.787,
      "step": 1432
    },
    {
      "epoch": 0.1101720958819914,
      "grad_norm": 4.59893274307251,
      "learning_rate": 1.9559311616472037e-05,
      "loss": 1.6119,
      "step": 1434
    },
    {
      "epoch": 0.11032575291948371,
      "grad_norm": 4.8417744636535645,
      "learning_rate": 1.9558696988322067e-05,
      "loss": 1.7965,
      "step": 1436
    },
    {
      "epoch": 0.11047940995697603,
      "grad_norm": 4.829365253448486,
      "learning_rate": 1.9558082360172096e-05,
      "loss": 1.6125,
      "step": 1438
    },
    {
      "epoch": 0.11063306699446834,
      "grad_norm": 4.74966287612915,
      "learning_rate": 1.955746773202213e-05,
      "loss": 1.5761,
      "step": 1440
    },
    {
      "epoch": 0.11078672403196066,
      "grad_norm": 4.8681535720825195,
      "learning_rate": 1.955685310387216e-05,
      "loss": 1.5977,
      "step": 1442
    },
    {
      "epoch": 0.11094038106945298,
      "grad_norm": 4.576766014099121,
      "learning_rate": 1.955623847572219e-05,
      "loss": 1.6297,
      "step": 1444
    },
    {
      "epoch": 0.11109403810694529,
      "grad_norm": 4.206700325012207,
      "learning_rate": 1.9555623847572222e-05,
      "loss": 1.6246,
      "step": 1446
    },
    {
      "epoch": 0.11124769514443761,
      "grad_norm": 4.753570079803467,
      "learning_rate": 1.955500921942225e-05,
      "loss": 1.6627,
      "step": 1448
    },
    {
      "epoch": 0.11140135218192994,
      "grad_norm": 4.992982864379883,
      "learning_rate": 1.955439459127228e-05,
      "loss": 1.7223,
      "step": 1450
    },
    {
      "epoch": 0.11155500921942225,
      "grad_norm": 4.912965297698975,
      "learning_rate": 1.955377996312231e-05,
      "loss": 1.5616,
      "step": 1452
    },
    {
      "epoch": 0.11170866625691457,
      "grad_norm": 4.4759840965271,
      "learning_rate": 1.9553165334972344e-05,
      "loss": 1.5403,
      "step": 1454
    },
    {
      "epoch": 0.11186232329440689,
      "grad_norm": 5.181031703948975,
      "learning_rate": 1.9552550706822374e-05,
      "loss": 1.6365,
      "step": 1456
    },
    {
      "epoch": 0.1120159803318992,
      "grad_norm": 4.845396518707275,
      "learning_rate": 1.9551936078672403e-05,
      "loss": 1.6279,
      "step": 1458
    },
    {
      "epoch": 0.11216963736939152,
      "grad_norm": 4.756799221038818,
      "learning_rate": 1.9551321450522436e-05,
      "loss": 1.6919,
      "step": 1460
    },
    {
      "epoch": 0.11232329440688384,
      "grad_norm": 5.1768107414245605,
      "learning_rate": 1.9550706822372466e-05,
      "loss": 1.5765,
      "step": 1462
    },
    {
      "epoch": 0.11247695144437615,
      "grad_norm": 4.743069648742676,
      "learning_rate": 1.9550092194222496e-05,
      "loss": 1.8469,
      "step": 1464
    },
    {
      "epoch": 0.11263060848186847,
      "grad_norm": 4.831038951873779,
      "learning_rate": 1.954947756607253e-05,
      "loss": 1.7746,
      "step": 1466
    },
    {
      "epoch": 0.11278426551936079,
      "grad_norm": 4.309507846832275,
      "learning_rate": 1.954886293792256e-05,
      "loss": 1.5142,
      "step": 1468
    },
    {
      "epoch": 0.1129379225568531,
      "grad_norm": 55.4850959777832,
      "learning_rate": 1.9548248309772588e-05,
      "loss": 1.683,
      "step": 1470
    },
    {
      "epoch": 0.11309157959434542,
      "grad_norm": 4.364781856536865,
      "learning_rate": 1.954763368162262e-05,
      "loss": 1.8977,
      "step": 1472
    },
    {
      "epoch": 0.11324523663183773,
      "grad_norm": 4.795863151550293,
      "learning_rate": 1.954701905347265e-05,
      "loss": 1.6436,
      "step": 1474
    },
    {
      "epoch": 0.11339889366933005,
      "grad_norm": 4.47898530960083,
      "learning_rate": 1.954640442532268e-05,
      "loss": 1.7043,
      "step": 1476
    },
    {
      "epoch": 0.11355255070682237,
      "grad_norm": 5.761251926422119,
      "learning_rate": 1.954578979717271e-05,
      "loss": 1.6943,
      "step": 1478
    },
    {
      "epoch": 0.11370620774431468,
      "grad_norm": 5.274534225463867,
      "learning_rate": 1.9545175169022743e-05,
      "loss": 1.6005,
      "step": 1480
    },
    {
      "epoch": 0.11385986478180701,
      "grad_norm": 4.412993431091309,
      "learning_rate": 1.9544560540872773e-05,
      "loss": 1.4977,
      "step": 1482
    },
    {
      "epoch": 0.11401352181929933,
      "grad_norm": 4.08578634262085,
      "learning_rate": 1.9543945912722803e-05,
      "loss": 1.6459,
      "step": 1484
    },
    {
      "epoch": 0.11416717885679165,
      "grad_norm": 3.7015979290008545,
      "learning_rate": 1.9543331284572836e-05,
      "loss": 1.5978,
      "step": 1486
    },
    {
      "epoch": 0.11432083589428396,
      "grad_norm": 4.919078826904297,
      "learning_rate": 1.9542716656422865e-05,
      "loss": 1.5456,
      "step": 1488
    },
    {
      "epoch": 0.11447449293177628,
      "grad_norm": 4.756066799163818,
      "learning_rate": 1.9542102028272895e-05,
      "loss": 1.7221,
      "step": 1490
    },
    {
      "epoch": 0.1146281499692686,
      "grad_norm": 4.4432525634765625,
      "learning_rate": 1.9541487400122928e-05,
      "loss": 1.5918,
      "step": 1492
    },
    {
      "epoch": 0.11478180700676091,
      "grad_norm": 5.371875286102295,
      "learning_rate": 1.9540872771972958e-05,
      "loss": 1.7009,
      "step": 1494
    },
    {
      "epoch": 0.11493546404425323,
      "grad_norm": 3.5335211753845215,
      "learning_rate": 1.9540258143822988e-05,
      "loss": 1.6021,
      "step": 1496
    },
    {
      "epoch": 0.11508912108174554,
      "grad_norm": 4.77205753326416,
      "learning_rate": 1.953964351567302e-05,
      "loss": 1.5721,
      "step": 1498
    },
    {
      "epoch": 0.11524277811923786,
      "grad_norm": 5.020537376403809,
      "learning_rate": 1.953902888752305e-05,
      "loss": 1.6367,
      "step": 1500
    },
    {
      "epoch": 0.11539643515673018,
      "grad_norm": 4.866142272949219,
      "learning_rate": 1.9538414259373083e-05,
      "loss": 1.6035,
      "step": 1502
    },
    {
      "epoch": 0.11555009219422249,
      "grad_norm": 3.647397756576538,
      "learning_rate": 1.953779963122311e-05,
      "loss": 1.5274,
      "step": 1504
    },
    {
      "epoch": 0.11570374923171481,
      "grad_norm": 3.999390125274658,
      "learning_rate": 1.9537185003073143e-05,
      "loss": 1.7818,
      "step": 1506
    },
    {
      "epoch": 0.11585740626920712,
      "grad_norm": 4.787381172180176,
      "learning_rate": 1.9536570374923172e-05,
      "loss": 1.7033,
      "step": 1508
    },
    {
      "epoch": 0.11601106330669944,
      "grad_norm": 4.415989398956299,
      "learning_rate": 1.9535955746773202e-05,
      "loss": 1.831,
      "step": 1510
    },
    {
      "epoch": 0.11616472034419176,
      "grad_norm": 4.548354148864746,
      "learning_rate": 1.9535341118623235e-05,
      "loss": 1.5145,
      "step": 1512
    },
    {
      "epoch": 0.11631837738168409,
      "grad_norm": 5.4493560791015625,
      "learning_rate": 1.9534726490473265e-05,
      "loss": 1.876,
      "step": 1514
    },
    {
      "epoch": 0.1164720344191764,
      "grad_norm": 3.9988834857940674,
      "learning_rate": 1.9534111862323295e-05,
      "loss": 1.6325,
      "step": 1516
    },
    {
      "epoch": 0.11662569145666872,
      "grad_norm": 4.861139297485352,
      "learning_rate": 1.9533497234173328e-05,
      "loss": 1.6743,
      "step": 1518
    },
    {
      "epoch": 0.11677934849416104,
      "grad_norm": 5.388833522796631,
      "learning_rate": 1.9532882606023357e-05,
      "loss": 1.5854,
      "step": 1520
    },
    {
      "epoch": 0.11693300553165335,
      "grad_norm": 4.772726058959961,
      "learning_rate": 1.953226797787339e-05,
      "loss": 1.6481,
      "step": 1522
    },
    {
      "epoch": 0.11708666256914567,
      "grad_norm": 4.285337924957275,
      "learning_rate": 1.9531653349723417e-05,
      "loss": 1.7054,
      "step": 1524
    },
    {
      "epoch": 0.11724031960663799,
      "grad_norm": 4.5872626304626465,
      "learning_rate": 1.953103872157345e-05,
      "loss": 1.6569,
      "step": 1526
    },
    {
      "epoch": 0.1173939766441303,
      "grad_norm": 4.3280463218688965,
      "learning_rate": 1.9530424093423483e-05,
      "loss": 1.5277,
      "step": 1528
    },
    {
      "epoch": 0.11754763368162262,
      "grad_norm": 4.480382919311523,
      "learning_rate": 1.952980946527351e-05,
      "loss": 1.7289,
      "step": 1530
    },
    {
      "epoch": 0.11770129071911493,
      "grad_norm": 4.207196235656738,
      "learning_rate": 1.9529194837123542e-05,
      "loss": 1.803,
      "step": 1532
    },
    {
      "epoch": 0.11785494775660725,
      "grad_norm": 4.125123023986816,
      "learning_rate": 1.9528580208973572e-05,
      "loss": 1.6289,
      "step": 1534
    },
    {
      "epoch": 0.11800860479409957,
      "grad_norm": 6.329103469848633,
      "learning_rate": 1.95279655808236e-05,
      "loss": 1.6592,
      "step": 1536
    },
    {
      "epoch": 0.11816226183159188,
      "grad_norm": 4.436602592468262,
      "learning_rate": 1.9527350952673635e-05,
      "loss": 1.5904,
      "step": 1538
    },
    {
      "epoch": 0.1183159188690842,
      "grad_norm": 4.564888954162598,
      "learning_rate": 1.9526736324523664e-05,
      "loss": 1.7565,
      "step": 1540
    },
    {
      "epoch": 0.11846957590657652,
      "grad_norm": 4.3771514892578125,
      "learning_rate": 1.9526121696373697e-05,
      "loss": 1.6746,
      "step": 1542
    },
    {
      "epoch": 0.11862323294406883,
      "grad_norm": 4.449161529541016,
      "learning_rate": 1.9525507068223727e-05,
      "loss": 1.6377,
      "step": 1544
    },
    {
      "epoch": 0.11877688998156116,
      "grad_norm": 4.770364761352539,
      "learning_rate": 1.9524892440073757e-05,
      "loss": 1.5953,
      "step": 1546
    },
    {
      "epoch": 0.11893054701905348,
      "grad_norm": 4.0749640464782715,
      "learning_rate": 1.952427781192379e-05,
      "loss": 1.5511,
      "step": 1548
    },
    {
      "epoch": 0.1190842040565458,
      "grad_norm": 4.361663341522217,
      "learning_rate": 1.9523663183773816e-05,
      "loss": 1.5562,
      "step": 1550
    },
    {
      "epoch": 0.11923786109403811,
      "grad_norm": 4.269155025482178,
      "learning_rate": 1.952304855562385e-05,
      "loss": 1.5725,
      "step": 1552
    },
    {
      "epoch": 0.11939151813153043,
      "grad_norm": 4.128551483154297,
      "learning_rate": 1.952243392747388e-05,
      "loss": 1.5,
      "step": 1554
    },
    {
      "epoch": 0.11954517516902274,
      "grad_norm": 4.763240814208984,
      "learning_rate": 1.952181929932391e-05,
      "loss": 1.7572,
      "step": 1556
    },
    {
      "epoch": 0.11969883220651506,
      "grad_norm": 4.871914386749268,
      "learning_rate": 1.952120467117394e-05,
      "loss": 1.609,
      "step": 1558
    },
    {
      "epoch": 0.11985248924400738,
      "grad_norm": 4.267725467681885,
      "learning_rate": 1.952059004302397e-05,
      "loss": 1.5832,
      "step": 1560
    },
    {
      "epoch": 0.12000614628149969,
      "grad_norm": 4.569482326507568,
      "learning_rate": 1.9519975414874e-05,
      "loss": 1.6262,
      "step": 1562
    },
    {
      "epoch": 0.12015980331899201,
      "grad_norm": 4.285094261169434,
      "learning_rate": 1.9519360786724034e-05,
      "loss": 1.7479,
      "step": 1564
    },
    {
      "epoch": 0.12031346035648433,
      "grad_norm": 4.529351234436035,
      "learning_rate": 1.9518746158574064e-05,
      "loss": 1.6297,
      "step": 1566
    },
    {
      "epoch": 0.12046711739397664,
      "grad_norm": 4.966389179229736,
      "learning_rate": 1.9518131530424097e-05,
      "loss": 1.7544,
      "step": 1568
    },
    {
      "epoch": 0.12062077443146896,
      "grad_norm": 4.608340263366699,
      "learning_rate": 1.9517516902274127e-05,
      "loss": 1.4635,
      "step": 1570
    },
    {
      "epoch": 0.12077443146896127,
      "grad_norm": 3.8790552616119385,
      "learning_rate": 1.9516902274124156e-05,
      "loss": 1.6345,
      "step": 1572
    },
    {
      "epoch": 0.12092808850645359,
      "grad_norm": 5.229369163513184,
      "learning_rate": 1.951628764597419e-05,
      "loss": 1.6829,
      "step": 1574
    },
    {
      "epoch": 0.1210817455439459,
      "grad_norm": 4.269663333892822,
      "learning_rate": 1.9515673017824216e-05,
      "loss": 1.568,
      "step": 1576
    },
    {
      "epoch": 0.12123540258143822,
      "grad_norm": 4.905238151550293,
      "learning_rate": 1.951505838967425e-05,
      "loss": 1.671,
      "step": 1578
    },
    {
      "epoch": 0.12138905961893055,
      "grad_norm": 4.5513596534729,
      "learning_rate": 1.951444376152428e-05,
      "loss": 1.6636,
      "step": 1580
    },
    {
      "epoch": 0.12154271665642287,
      "grad_norm": 4.586058616638184,
      "learning_rate": 1.9513829133374308e-05,
      "loss": 1.7669,
      "step": 1582
    },
    {
      "epoch": 0.12169637369391519,
      "grad_norm": 5.4855170249938965,
      "learning_rate": 1.951321450522434e-05,
      "loss": 1.5033,
      "step": 1584
    },
    {
      "epoch": 0.1218500307314075,
      "grad_norm": 4.668776035308838,
      "learning_rate": 1.951259987707437e-05,
      "loss": 1.5859,
      "step": 1586
    },
    {
      "epoch": 0.12200368776889982,
      "grad_norm": 3.9210376739501953,
      "learning_rate": 1.9511985248924404e-05,
      "loss": 1.5757,
      "step": 1588
    },
    {
      "epoch": 0.12215734480639213,
      "grad_norm": 4.558568000793457,
      "learning_rate": 1.9511370620774434e-05,
      "loss": 1.5945,
      "step": 1590
    },
    {
      "epoch": 0.12231100184388445,
      "grad_norm": 4.247246265411377,
      "learning_rate": 1.9510755992624463e-05,
      "loss": 1.624,
      "step": 1592
    },
    {
      "epoch": 0.12246465888137677,
      "grad_norm": 4.2471604347229,
      "learning_rate": 1.9510141364474496e-05,
      "loss": 1.5873,
      "step": 1594
    },
    {
      "epoch": 0.12261831591886908,
      "grad_norm": 4.362886428833008,
      "learning_rate": 1.9509526736324526e-05,
      "loss": 1.7448,
      "step": 1596
    },
    {
      "epoch": 0.1227719729563614,
      "grad_norm": 5.111678123474121,
      "learning_rate": 1.9508912108174556e-05,
      "loss": 1.8134,
      "step": 1598
    },
    {
      "epoch": 0.12292562999385372,
      "grad_norm": 4.4582624435424805,
      "learning_rate": 1.950829748002459e-05,
      "loss": 1.7155,
      "step": 1600
    },
    {
      "epoch": 0.12307928703134603,
      "grad_norm": 3.796780586242676,
      "learning_rate": 1.9507682851874615e-05,
      "loss": 1.5636,
      "step": 1602
    },
    {
      "epoch": 0.12323294406883835,
      "grad_norm": 4.517824649810791,
      "learning_rate": 1.9507068223724648e-05,
      "loss": 1.6092,
      "step": 1604
    },
    {
      "epoch": 0.12338660110633067,
      "grad_norm": 4.659684181213379,
      "learning_rate": 1.9506453595574678e-05,
      "loss": 1.641,
      "step": 1606
    },
    {
      "epoch": 0.12354025814382298,
      "grad_norm": 4.470782279968262,
      "learning_rate": 1.950583896742471e-05,
      "loss": 1.618,
      "step": 1608
    },
    {
      "epoch": 0.1236939151813153,
      "grad_norm": 4.486400604248047,
      "learning_rate": 1.950522433927474e-05,
      "loss": 1.6912,
      "step": 1610
    },
    {
      "epoch": 0.12384757221880763,
      "grad_norm": 4.459258556365967,
      "learning_rate": 1.950460971112477e-05,
      "loss": 1.5627,
      "step": 1612
    },
    {
      "epoch": 0.12400122925629994,
      "grad_norm": 4.486885070800781,
      "learning_rate": 1.9503995082974803e-05,
      "loss": 1.8642,
      "step": 1614
    },
    {
      "epoch": 0.12415488629379226,
      "grad_norm": 4.576472282409668,
      "learning_rate": 1.9503380454824833e-05,
      "loss": 1.6411,
      "step": 1616
    },
    {
      "epoch": 0.12430854333128458,
      "grad_norm": 4.349391460418701,
      "learning_rate": 1.9502765826674863e-05,
      "loss": 1.6382,
      "step": 1618
    },
    {
      "epoch": 0.12446220036877689,
      "grad_norm": 4.264526844024658,
      "learning_rate": 1.9502151198524896e-05,
      "loss": 1.621,
      "step": 1620
    },
    {
      "epoch": 0.12461585740626921,
      "grad_norm": 4.798770904541016,
      "learning_rate": 1.9501536570374925e-05,
      "loss": 1.8124,
      "step": 1622
    },
    {
      "epoch": 0.12476951444376153,
      "grad_norm": 3.747992515563965,
      "learning_rate": 1.9500921942224955e-05,
      "loss": 1.516,
      "step": 1624
    },
    {
      "epoch": 0.12492317148125384,
      "grad_norm": 4.410411834716797,
      "learning_rate": 1.9500307314074988e-05,
      "loss": 1.5645,
      "step": 1626
    },
    {
      "epoch": 0.12507682851874616,
      "grad_norm": 4.139060020446777,
      "learning_rate": 1.9499692685925018e-05,
      "loss": 1.6217,
      "step": 1628
    },
    {
      "epoch": 0.12523048555623847,
      "grad_norm": 4.380125045776367,
      "learning_rate": 1.9499078057775048e-05,
      "loss": 1.6909,
      "step": 1630
    },
    {
      "epoch": 0.1253841425937308,
      "grad_norm": 4.449796676635742,
      "learning_rate": 1.9498463429625077e-05,
      "loss": 1.7215,
      "step": 1632
    },
    {
      "epoch": 0.1255377996312231,
      "grad_norm": 4.043376922607422,
      "learning_rate": 1.949784880147511e-05,
      "loss": 1.6326,
      "step": 1634
    },
    {
      "epoch": 0.12569145666871542,
      "grad_norm": 4.427875518798828,
      "learning_rate": 1.949723417332514e-05,
      "loss": 1.6962,
      "step": 1636
    },
    {
      "epoch": 0.12584511370620774,
      "grad_norm": 4.617554187774658,
      "learning_rate": 1.949661954517517e-05,
      "loss": 1.5711,
      "step": 1638
    },
    {
      "epoch": 0.12599877074370006,
      "grad_norm": 4.245482444763184,
      "learning_rate": 1.9496004917025203e-05,
      "loss": 1.6479,
      "step": 1640
    },
    {
      "epoch": 0.12615242778119237,
      "grad_norm": 4.876771926879883,
      "learning_rate": 1.9495390288875232e-05,
      "loss": 1.7238,
      "step": 1642
    },
    {
      "epoch": 0.1263060848186847,
      "grad_norm": 4.263737678527832,
      "learning_rate": 1.9494775660725262e-05,
      "loss": 1.5666,
      "step": 1644
    },
    {
      "epoch": 0.126459741856177,
      "grad_norm": 6.202945232391357,
      "learning_rate": 1.9494161032575295e-05,
      "loss": 1.6217,
      "step": 1646
    },
    {
      "epoch": 0.12661339889366932,
      "grad_norm": 4.307828426361084,
      "learning_rate": 1.9493546404425325e-05,
      "loss": 1.49,
      "step": 1648
    },
    {
      "epoch": 0.12676705593116164,
      "grad_norm": 4.122886657714844,
      "learning_rate": 1.9492931776275355e-05,
      "loss": 1.7121,
      "step": 1650
    },
    {
      "epoch": 0.12692071296865395,
      "grad_norm": 4.3632426261901855,
      "learning_rate": 1.9492317148125384e-05,
      "loss": 1.6835,
      "step": 1652
    },
    {
      "epoch": 0.12707437000614627,
      "grad_norm": 4.4186625480651855,
      "learning_rate": 1.9491702519975417e-05,
      "loss": 1.7579,
      "step": 1654
    },
    {
      "epoch": 0.1272280270436386,
      "grad_norm": 4.411682605743408,
      "learning_rate": 1.9491087891825447e-05,
      "loss": 1.5771,
      "step": 1656
    },
    {
      "epoch": 0.1273816840811309,
      "grad_norm": 4.259854316711426,
      "learning_rate": 1.9490473263675477e-05,
      "loss": 1.6239,
      "step": 1658
    },
    {
      "epoch": 0.12753534111862325,
      "grad_norm": 4.225386619567871,
      "learning_rate": 1.948985863552551e-05,
      "loss": 1.6777,
      "step": 1660
    },
    {
      "epoch": 0.12768899815611556,
      "grad_norm": 4.977676868438721,
      "learning_rate": 1.948924400737554e-05,
      "loss": 1.6166,
      "step": 1662
    },
    {
      "epoch": 0.12784265519360788,
      "grad_norm": 3.7306509017944336,
      "learning_rate": 1.948862937922557e-05,
      "loss": 1.5834,
      "step": 1664
    },
    {
      "epoch": 0.1279963122311002,
      "grad_norm": 4.451853275299072,
      "learning_rate": 1.9488014751075602e-05,
      "loss": 1.6464,
      "step": 1666
    },
    {
      "epoch": 0.1281499692685925,
      "grad_norm": 4.641234397888184,
      "learning_rate": 1.9487400122925632e-05,
      "loss": 1.6698,
      "step": 1668
    },
    {
      "epoch": 0.12830362630608483,
      "grad_norm": 5.218206882476807,
      "learning_rate": 1.948678549477566e-05,
      "loss": 1.6614,
      "step": 1670
    },
    {
      "epoch": 0.12845728334357714,
      "grad_norm": 4.623648166656494,
      "learning_rate": 1.9486170866625695e-05,
      "loss": 1.6586,
      "step": 1672
    },
    {
      "epoch": 0.12861094038106946,
      "grad_norm": 5.1708478927612305,
      "learning_rate": 1.9485556238475724e-05,
      "loss": 1.6275,
      "step": 1674
    },
    {
      "epoch": 0.12876459741856178,
      "grad_norm": 4.305856227874756,
      "learning_rate": 1.9484941610325754e-05,
      "loss": 1.6349,
      "step": 1676
    },
    {
      "epoch": 0.1289182544560541,
      "grad_norm": 4.788485050201416,
      "learning_rate": 1.9484326982175784e-05,
      "loss": 1.4676,
      "step": 1678
    },
    {
      "epoch": 0.1290719114935464,
      "grad_norm": 4.4581379890441895,
      "learning_rate": 1.9483712354025817e-05,
      "loss": 1.5062,
      "step": 1680
    },
    {
      "epoch": 0.12922556853103873,
      "grad_norm": 3.9021549224853516,
      "learning_rate": 1.9483097725875846e-05,
      "loss": 1.7848,
      "step": 1682
    },
    {
      "epoch": 0.12937922556853104,
      "grad_norm": 4.530584812164307,
      "learning_rate": 1.9482483097725876e-05,
      "loss": 1.6594,
      "step": 1684
    },
    {
      "epoch": 0.12953288260602336,
      "grad_norm": 4.8017497062683105,
      "learning_rate": 1.948186846957591e-05,
      "loss": 1.6167,
      "step": 1686
    },
    {
      "epoch": 0.12968653964351567,
      "grad_norm": 4.41823148727417,
      "learning_rate": 1.948125384142594e-05,
      "loss": 1.5293,
      "step": 1688
    },
    {
      "epoch": 0.129840196681008,
      "grad_norm": 4.470682144165039,
      "learning_rate": 1.948063921327597e-05,
      "loss": 1.6036,
      "step": 1690
    },
    {
      "epoch": 0.1299938537185003,
      "grad_norm": 3.947842597961426,
      "learning_rate": 1.9480024585126e-05,
      "loss": 1.7037,
      "step": 1692
    },
    {
      "epoch": 0.13014751075599262,
      "grad_norm": 4.953098297119141,
      "learning_rate": 1.947940995697603e-05,
      "loss": 1.6436,
      "step": 1694
    },
    {
      "epoch": 0.13030116779348494,
      "grad_norm": 4.112635135650635,
      "learning_rate": 1.947879532882606e-05,
      "loss": 1.4595,
      "step": 1696
    },
    {
      "epoch": 0.13045482483097726,
      "grad_norm": 4.197033882141113,
      "learning_rate": 1.9478180700676094e-05,
      "loss": 1.657,
      "step": 1698
    },
    {
      "epoch": 0.13060848186846957,
      "grad_norm": 4.02692985534668,
      "learning_rate": 1.9477566072526124e-05,
      "loss": 1.6321,
      "step": 1700
    },
    {
      "epoch": 0.1307621389059619,
      "grad_norm": 4.7861809730529785,
      "learning_rate": 1.9476951444376153e-05,
      "loss": 1.5609,
      "step": 1702
    },
    {
      "epoch": 0.1309157959434542,
      "grad_norm": 4.392903804779053,
      "learning_rate": 1.9476336816226183e-05,
      "loss": 1.7549,
      "step": 1704
    },
    {
      "epoch": 0.13106945298094652,
      "grad_norm": 4.314429759979248,
      "learning_rate": 1.9475722188076216e-05,
      "loss": 1.5698,
      "step": 1706
    },
    {
      "epoch": 0.13122311001843884,
      "grad_norm": 4.254858016967773,
      "learning_rate": 1.9475107559926246e-05,
      "loss": 1.6291,
      "step": 1708
    },
    {
      "epoch": 0.13137676705593115,
      "grad_norm": 4.288058757781982,
      "learning_rate": 1.9474492931776276e-05,
      "loss": 1.822,
      "step": 1710
    },
    {
      "epoch": 0.13153042409342347,
      "grad_norm": 4.206986904144287,
      "learning_rate": 1.947387830362631e-05,
      "loss": 1.6372,
      "step": 1712
    },
    {
      "epoch": 0.1316840811309158,
      "grad_norm": 3.9056224822998047,
      "learning_rate": 1.947326367547634e-05,
      "loss": 1.6141,
      "step": 1714
    },
    {
      "epoch": 0.1318377381684081,
      "grad_norm": 5.1152777671813965,
      "learning_rate": 1.9472649047326368e-05,
      "loss": 1.6361,
      "step": 1716
    },
    {
      "epoch": 0.13199139520590042,
      "grad_norm": 4.0903120040893555,
      "learning_rate": 1.94720344191764e-05,
      "loss": 1.5559,
      "step": 1718
    },
    {
      "epoch": 0.13214505224339274,
      "grad_norm": 4.825276851654053,
      "learning_rate": 1.947141979102643e-05,
      "loss": 1.732,
      "step": 1720
    },
    {
      "epoch": 0.13229870928088505,
      "grad_norm": 4.649293899536133,
      "learning_rate": 1.947080516287646e-05,
      "loss": 1.5941,
      "step": 1722
    },
    {
      "epoch": 0.13245236631837737,
      "grad_norm": 4.052992820739746,
      "learning_rate": 1.9470190534726494e-05,
      "loss": 1.5664,
      "step": 1724
    },
    {
      "epoch": 0.1326060233558697,
      "grad_norm": 4.36129903793335,
      "learning_rate": 1.9469575906576523e-05,
      "loss": 1.7345,
      "step": 1726
    },
    {
      "epoch": 0.13275968039336203,
      "grad_norm": 4.522770404815674,
      "learning_rate": 1.9468961278426553e-05,
      "loss": 1.6731,
      "step": 1728
    },
    {
      "epoch": 0.13291333743085434,
      "grad_norm": 4.922299385070801,
      "learning_rate": 1.9468346650276583e-05,
      "loss": 1.8072,
      "step": 1730
    },
    {
      "epoch": 0.13306699446834666,
      "grad_norm": 4.385134220123291,
      "learning_rate": 1.9467732022126616e-05,
      "loss": 1.5836,
      "step": 1732
    },
    {
      "epoch": 0.13322065150583898,
      "grad_norm": 4.031277179718018,
      "learning_rate": 1.9467117393976645e-05,
      "loss": 1.465,
      "step": 1734
    },
    {
      "epoch": 0.1333743085433313,
      "grad_norm": 4.437002182006836,
      "learning_rate": 1.9466502765826675e-05,
      "loss": 1.5624,
      "step": 1736
    },
    {
      "epoch": 0.1335279655808236,
      "grad_norm": 3.754696846008301,
      "learning_rate": 1.9465888137676708e-05,
      "loss": 1.6044,
      "step": 1738
    },
    {
      "epoch": 0.13368162261831593,
      "grad_norm": 3.967130661010742,
      "learning_rate": 1.9465273509526738e-05,
      "loss": 1.5832,
      "step": 1740
    },
    {
      "epoch": 0.13383527965580824,
      "grad_norm": 3.958448648452759,
      "learning_rate": 1.9464658881376767e-05,
      "loss": 1.6812,
      "step": 1742
    },
    {
      "epoch": 0.13398893669330056,
      "grad_norm": 5.2511982917785645,
      "learning_rate": 1.94640442532268e-05,
      "loss": 1.8565,
      "step": 1744
    },
    {
      "epoch": 0.13414259373079288,
      "grad_norm": 4.229193210601807,
      "learning_rate": 1.946342962507683e-05,
      "loss": 1.6365,
      "step": 1746
    },
    {
      "epoch": 0.1342962507682852,
      "grad_norm": 3.8518741130828857,
      "learning_rate": 1.946281499692686e-05,
      "loss": 1.571,
      "step": 1748
    },
    {
      "epoch": 0.1344499078057775,
      "grad_norm": 4.383627414703369,
      "learning_rate": 1.946220036877689e-05,
      "loss": 1.5181,
      "step": 1750
    },
    {
      "epoch": 0.13460356484326982,
      "grad_norm": 4.58341121673584,
      "learning_rate": 1.9461585740626923e-05,
      "loss": 1.6387,
      "step": 1752
    },
    {
      "epoch": 0.13475722188076214,
      "grad_norm": 4.656858921051025,
      "learning_rate": 1.9460971112476956e-05,
      "loss": 1.6474,
      "step": 1754
    },
    {
      "epoch": 0.13491087891825446,
      "grad_norm": 5.039700031280518,
      "learning_rate": 1.9460356484326982e-05,
      "loss": 1.6755,
      "step": 1756
    },
    {
      "epoch": 0.13506453595574677,
      "grad_norm": 4.46349573135376,
      "learning_rate": 1.9459741856177015e-05,
      "loss": 1.5728,
      "step": 1758
    },
    {
      "epoch": 0.1352181929932391,
      "grad_norm": 4.041154861450195,
      "learning_rate": 1.9459127228027045e-05,
      "loss": 1.5847,
      "step": 1760
    },
    {
      "epoch": 0.1353718500307314,
      "grad_norm": 4.126910209655762,
      "learning_rate": 1.9458512599877074e-05,
      "loss": 1.6807,
      "step": 1762
    },
    {
      "epoch": 0.13552550706822372,
      "grad_norm": 4.063604831695557,
      "learning_rate": 1.9457897971727108e-05,
      "loss": 1.5294,
      "step": 1764
    },
    {
      "epoch": 0.13567916410571604,
      "grad_norm": 4.1347150802612305,
      "learning_rate": 1.9457283343577137e-05,
      "loss": 1.5728,
      "step": 1766
    },
    {
      "epoch": 0.13583282114320835,
      "grad_norm": 4.593793869018555,
      "learning_rate": 1.9456668715427167e-05,
      "loss": 1.7155,
      "step": 1768
    },
    {
      "epoch": 0.13598647818070067,
      "grad_norm": 4.340649127960205,
      "learning_rate": 1.94560540872772e-05,
      "loss": 1.6996,
      "step": 1770
    },
    {
      "epoch": 0.136140135218193,
      "grad_norm": 4.278517246246338,
      "learning_rate": 1.945543945912723e-05,
      "loss": 1.6012,
      "step": 1772
    },
    {
      "epoch": 0.1362937922556853,
      "grad_norm": 4.626030445098877,
      "learning_rate": 1.9454824830977263e-05,
      "loss": 1.6195,
      "step": 1774
    },
    {
      "epoch": 0.13644744929317762,
      "grad_norm": 4.450915813446045,
      "learning_rate": 1.945421020282729e-05,
      "loss": 1.6398,
      "step": 1776
    },
    {
      "epoch": 0.13660110633066994,
      "grad_norm": 4.265727996826172,
      "learning_rate": 1.9453595574677322e-05,
      "loss": 1.5958,
      "step": 1778
    },
    {
      "epoch": 0.13675476336816225,
      "grad_norm": 4.036159038543701,
      "learning_rate": 1.9452980946527352e-05,
      "loss": 1.5647,
      "step": 1780
    },
    {
      "epoch": 0.13690842040565457,
      "grad_norm": 4.2282257080078125,
      "learning_rate": 1.945236631837738e-05,
      "loss": 1.6079,
      "step": 1782
    },
    {
      "epoch": 0.13706207744314688,
      "grad_norm": 4.005040645599365,
      "learning_rate": 1.9451751690227415e-05,
      "loss": 1.5044,
      "step": 1784
    },
    {
      "epoch": 0.1372157344806392,
      "grad_norm": 4.676270484924316,
      "learning_rate": 1.9451137062077444e-05,
      "loss": 1.6304,
      "step": 1786
    },
    {
      "epoch": 0.13736939151813152,
      "grad_norm": 4.598161697387695,
      "learning_rate": 1.9450522433927474e-05,
      "loss": 1.6887,
      "step": 1788
    },
    {
      "epoch": 0.13752304855562386,
      "grad_norm": 5.0116448402404785,
      "learning_rate": 1.9449907805777507e-05,
      "loss": 1.6998,
      "step": 1790
    },
    {
      "epoch": 0.13767670559311618,
      "grad_norm": 4.892838954925537,
      "learning_rate": 1.9449293177627537e-05,
      "loss": 1.6274,
      "step": 1792
    },
    {
      "epoch": 0.1378303626306085,
      "grad_norm": 5.293637752532959,
      "learning_rate": 1.944867854947757e-05,
      "loss": 1.628,
      "step": 1794
    },
    {
      "epoch": 0.1379840196681008,
      "grad_norm": 4.583549976348877,
      "learning_rate": 1.94480639213276e-05,
      "loss": 1.4792,
      "step": 1796
    },
    {
      "epoch": 0.13813767670559313,
      "grad_norm": 3.773277759552002,
      "learning_rate": 1.944744929317763e-05,
      "loss": 1.5219,
      "step": 1798
    },
    {
      "epoch": 0.13829133374308544,
      "grad_norm": 4.440420150756836,
      "learning_rate": 1.9446834665027662e-05,
      "loss": 1.6732,
      "step": 1800
    },
    {
      "epoch": 0.13844499078057776,
      "grad_norm": 4.711763858795166,
      "learning_rate": 1.944622003687769e-05,
      "loss": 1.5463,
      "step": 1802
    },
    {
      "epoch": 0.13859864781807008,
      "grad_norm": 5.035058498382568,
      "learning_rate": 1.944560540872772e-05,
      "loss": 1.6961,
      "step": 1804
    },
    {
      "epoch": 0.1387523048555624,
      "grad_norm": 3.963282346725464,
      "learning_rate": 1.944499078057775e-05,
      "loss": 1.5568,
      "step": 1806
    },
    {
      "epoch": 0.1389059618930547,
      "grad_norm": 4.577483654022217,
      "learning_rate": 1.944437615242778e-05,
      "loss": 1.5428,
      "step": 1808
    },
    {
      "epoch": 0.13905961893054702,
      "grad_norm": 4.509146690368652,
      "learning_rate": 1.9443761524277814e-05,
      "loss": 1.6397,
      "step": 1810
    },
    {
      "epoch": 0.13921327596803934,
      "grad_norm": 4.317050933837891,
      "learning_rate": 1.9443146896127844e-05,
      "loss": 1.7304,
      "step": 1812
    },
    {
      "epoch": 0.13936693300553166,
      "grad_norm": 4.572277069091797,
      "learning_rate": 1.9442532267977877e-05,
      "loss": 1.6759,
      "step": 1814
    },
    {
      "epoch": 0.13952059004302397,
      "grad_norm": 4.773606777191162,
      "learning_rate": 1.9441917639827906e-05,
      "loss": 1.6869,
      "step": 1816
    },
    {
      "epoch": 0.1396742470805163,
      "grad_norm": 4.57815408706665,
      "learning_rate": 1.9441303011677936e-05,
      "loss": 1.647,
      "step": 1818
    },
    {
      "epoch": 0.1398279041180086,
      "grad_norm": 4.822877407073975,
      "learning_rate": 1.944068838352797e-05,
      "loss": 1.6182,
      "step": 1820
    },
    {
      "epoch": 0.13998156115550092,
      "grad_norm": 4.272431373596191,
      "learning_rate": 1.9440073755378e-05,
      "loss": 1.6215,
      "step": 1822
    },
    {
      "epoch": 0.14013521819299324,
      "grad_norm": 4.476557731628418,
      "learning_rate": 1.943945912722803e-05,
      "loss": 1.6277,
      "step": 1824
    },
    {
      "epoch": 0.14028887523048555,
      "grad_norm": 4.522927284240723,
      "learning_rate": 1.943884449907806e-05,
      "loss": 1.526,
      "step": 1826
    },
    {
      "epoch": 0.14044253226797787,
      "grad_norm": 3.991070032119751,
      "learning_rate": 1.9438229870928088e-05,
      "loss": 1.6106,
      "step": 1828
    },
    {
      "epoch": 0.1405961893054702,
      "grad_norm": 4.189483165740967,
      "learning_rate": 1.943761524277812e-05,
      "loss": 1.619,
      "step": 1830
    },
    {
      "epoch": 0.1407498463429625,
      "grad_norm": 4.5693159103393555,
      "learning_rate": 1.943700061462815e-05,
      "loss": 1.7438,
      "step": 1832
    },
    {
      "epoch": 0.14090350338045482,
      "grad_norm": 3.8766119480133057,
      "learning_rate": 1.943638598647818e-05,
      "loss": 1.4953,
      "step": 1834
    },
    {
      "epoch": 0.14105716041794714,
      "grad_norm": 4.294021129608154,
      "learning_rate": 1.9435771358328213e-05,
      "loss": 1.5746,
      "step": 1836
    },
    {
      "epoch": 0.14121081745543945,
      "grad_norm": 4.195743083953857,
      "learning_rate": 1.9435156730178243e-05,
      "loss": 1.5049,
      "step": 1838
    },
    {
      "epoch": 0.14136447449293177,
      "grad_norm": 4.331358909606934,
      "learning_rate": 1.9434542102028276e-05,
      "loss": 1.8416,
      "step": 1840
    },
    {
      "epoch": 0.14151813153042408,
      "grad_norm": 4.328099727630615,
      "learning_rate": 1.9433927473878306e-05,
      "loss": 1.7201,
      "step": 1842
    },
    {
      "epoch": 0.1416717885679164,
      "grad_norm": 4.2462005615234375,
      "learning_rate": 1.9433312845728336e-05,
      "loss": 1.6308,
      "step": 1844
    },
    {
      "epoch": 0.14182544560540872,
      "grad_norm": 4.253352165222168,
      "learning_rate": 1.943269821757837e-05,
      "loss": 1.6856,
      "step": 1846
    },
    {
      "epoch": 0.14197910264290103,
      "grad_norm": 4.154186248779297,
      "learning_rate": 1.9432083589428395e-05,
      "loss": 1.6079,
      "step": 1848
    },
    {
      "epoch": 0.14213275968039335,
      "grad_norm": 6.227648735046387,
      "learning_rate": 1.9431468961278428e-05,
      "loss": 1.73,
      "step": 1850
    },
    {
      "epoch": 0.14228641671788567,
      "grad_norm": 4.038461208343506,
      "learning_rate": 1.943085433312846e-05,
      "loss": 1.7227,
      "step": 1852
    },
    {
      "epoch": 0.14244007375537798,
      "grad_norm": 4.844911098480225,
      "learning_rate": 1.9430239704978487e-05,
      "loss": 1.5841,
      "step": 1854
    },
    {
      "epoch": 0.14259373079287033,
      "grad_norm": 3.845120429992676,
      "learning_rate": 1.942962507682852e-05,
      "loss": 1.6545,
      "step": 1856
    },
    {
      "epoch": 0.14274738783036264,
      "grad_norm": 4.25357723236084,
      "learning_rate": 1.942901044867855e-05,
      "loss": 1.6038,
      "step": 1858
    },
    {
      "epoch": 0.14290104486785496,
      "grad_norm": 4.518612861633301,
      "learning_rate": 1.9428395820528583e-05,
      "loss": 1.703,
      "step": 1860
    },
    {
      "epoch": 0.14305470190534728,
      "grad_norm": 4.541075229644775,
      "learning_rate": 1.9427781192378613e-05,
      "loss": 1.6203,
      "step": 1862
    },
    {
      "epoch": 0.1432083589428396,
      "grad_norm": 4.06412935256958,
      "learning_rate": 1.9427166564228643e-05,
      "loss": 1.714,
      "step": 1864
    },
    {
      "epoch": 0.1433620159803319,
      "grad_norm": 4.289870738983154,
      "learning_rate": 1.9426551936078676e-05,
      "loss": 1.5524,
      "step": 1866
    },
    {
      "epoch": 0.14351567301782422,
      "grad_norm": 3.937469005584717,
      "learning_rate": 1.9425937307928705e-05,
      "loss": 1.5572,
      "step": 1868
    },
    {
      "epoch": 0.14366933005531654,
      "grad_norm": 4.361362457275391,
      "learning_rate": 1.9425322679778735e-05,
      "loss": 1.493,
      "step": 1870
    },
    {
      "epoch": 0.14382298709280886,
      "grad_norm": 3.9257559776306152,
      "learning_rate": 1.9424708051628768e-05,
      "loss": 1.502,
      "step": 1872
    },
    {
      "epoch": 0.14397664413030117,
      "grad_norm": 4.2765655517578125,
      "learning_rate": 1.9424093423478794e-05,
      "loss": 1.6001,
      "step": 1874
    },
    {
      "epoch": 0.1441303011677935,
      "grad_norm": 3.724155902862549,
      "learning_rate": 1.9423478795328827e-05,
      "loss": 1.5722,
      "step": 1876
    },
    {
      "epoch": 0.1442839582052858,
      "grad_norm": 4.135402679443359,
      "learning_rate": 1.9422864167178857e-05,
      "loss": 1.7137,
      "step": 1878
    },
    {
      "epoch": 0.14443761524277812,
      "grad_norm": 4.522433280944824,
      "learning_rate": 1.942224953902889e-05,
      "loss": 1.6161,
      "step": 1880
    },
    {
      "epoch": 0.14459127228027044,
      "grad_norm": 4.247946262359619,
      "learning_rate": 1.942163491087892e-05,
      "loss": 1.67,
      "step": 1882
    },
    {
      "epoch": 0.14474492931776275,
      "grad_norm": 4.322160243988037,
      "learning_rate": 1.942102028272895e-05,
      "loss": 1.5362,
      "step": 1884
    },
    {
      "epoch": 0.14489858635525507,
      "grad_norm": 4.25150728225708,
      "learning_rate": 1.9420405654578983e-05,
      "loss": 1.6026,
      "step": 1886
    },
    {
      "epoch": 0.1450522433927474,
      "grad_norm": 4.95831823348999,
      "learning_rate": 1.9419791026429012e-05,
      "loss": 1.6806,
      "step": 1888
    },
    {
      "epoch": 0.1452059004302397,
      "grad_norm": 4.125936031341553,
      "learning_rate": 1.9419176398279042e-05,
      "loss": 1.634,
      "step": 1890
    },
    {
      "epoch": 0.14535955746773202,
      "grad_norm": 3.6493210792541504,
      "learning_rate": 1.9418561770129075e-05,
      "loss": 1.4507,
      "step": 1892
    },
    {
      "epoch": 0.14551321450522434,
      "grad_norm": 4.338488578796387,
      "learning_rate": 1.9417947141979105e-05,
      "loss": 1.5127,
      "step": 1894
    },
    {
      "epoch": 0.14566687154271665,
      "grad_norm": 4.250901222229004,
      "learning_rate": 1.9417332513829134e-05,
      "loss": 1.4201,
      "step": 1896
    },
    {
      "epoch": 0.14582052858020897,
      "grad_norm": 3.8600480556488037,
      "learning_rate": 1.9416717885679167e-05,
      "loss": 1.5595,
      "step": 1898
    },
    {
      "epoch": 0.14597418561770129,
      "grad_norm": 4.317285537719727,
      "learning_rate": 1.9416103257529197e-05,
      "loss": 1.6094,
      "step": 1900
    },
    {
      "epoch": 0.1461278426551936,
      "grad_norm": 4.718072891235352,
      "learning_rate": 1.9415488629379227e-05,
      "loss": 1.768,
      "step": 1902
    },
    {
      "epoch": 0.14628149969268592,
      "grad_norm": 4.9370808601379395,
      "learning_rate": 1.9414874001229257e-05,
      "loss": 1.5801,
      "step": 1904
    },
    {
      "epoch": 0.14643515673017823,
      "grad_norm": 4.436810493469238,
      "learning_rate": 1.941425937307929e-05,
      "loss": 1.5847,
      "step": 1906
    },
    {
      "epoch": 0.14658881376767055,
      "grad_norm": 4.890700817108154,
      "learning_rate": 1.941364474492932e-05,
      "loss": 1.5092,
      "step": 1908
    },
    {
      "epoch": 0.14674247080516287,
      "grad_norm": 3.926815986633301,
      "learning_rate": 1.941303011677935e-05,
      "loss": 1.6612,
      "step": 1910
    },
    {
      "epoch": 0.14689612784265518,
      "grad_norm": 4.331315994262695,
      "learning_rate": 1.9412415488629382e-05,
      "loss": 1.7669,
      "step": 1912
    },
    {
      "epoch": 0.1470497848801475,
      "grad_norm": 5.178247928619385,
      "learning_rate": 1.9411800860479412e-05,
      "loss": 1.6749,
      "step": 1914
    },
    {
      "epoch": 0.14720344191763982,
      "grad_norm": 3.871377944946289,
      "learning_rate": 1.941118623232944e-05,
      "loss": 1.4508,
      "step": 1916
    },
    {
      "epoch": 0.14735709895513213,
      "grad_norm": 4.062928676605225,
      "learning_rate": 1.9410571604179474e-05,
      "loss": 1.458,
      "step": 1918
    },
    {
      "epoch": 0.14751075599262448,
      "grad_norm": 4.205310344696045,
      "learning_rate": 1.9409956976029504e-05,
      "loss": 1.6419,
      "step": 1920
    },
    {
      "epoch": 0.1476644130301168,
      "grad_norm": 3.822014093399048,
      "learning_rate": 1.9409342347879534e-05,
      "loss": 1.558,
      "step": 1922
    },
    {
      "epoch": 0.1478180700676091,
      "grad_norm": 4.9377593994140625,
      "learning_rate": 1.9408727719729567e-05,
      "loss": 1.687,
      "step": 1924
    },
    {
      "epoch": 0.14797172710510142,
      "grad_norm": 4.531102657318115,
      "learning_rate": 1.9408113091579597e-05,
      "loss": 1.7183,
      "step": 1926
    },
    {
      "epoch": 0.14812538414259374,
      "grad_norm": 4.2398881912231445,
      "learning_rate": 1.9407498463429626e-05,
      "loss": 1.5952,
      "step": 1928
    },
    {
      "epoch": 0.14827904118008606,
      "grad_norm": 4.426001071929932,
      "learning_rate": 1.9406883835279656e-05,
      "loss": 1.7844,
      "step": 1930
    },
    {
      "epoch": 0.14843269821757837,
      "grad_norm": 4.2123494148254395,
      "learning_rate": 1.940626920712969e-05,
      "loss": 1.6711,
      "step": 1932
    },
    {
      "epoch": 0.1485863552550707,
      "grad_norm": 4.681150913238525,
      "learning_rate": 1.940565457897972e-05,
      "loss": 1.6319,
      "step": 1934
    },
    {
      "epoch": 0.148740012292563,
      "grad_norm": 4.499131202697754,
      "learning_rate": 1.940503995082975e-05,
      "loss": 1.6108,
      "step": 1936
    },
    {
      "epoch": 0.14889366933005532,
      "grad_norm": 5.171452522277832,
      "learning_rate": 1.940442532267978e-05,
      "loss": 1.7528,
      "step": 1938
    },
    {
      "epoch": 0.14904732636754764,
      "grad_norm": 4.263631343841553,
      "learning_rate": 1.940381069452981e-05,
      "loss": 1.5191,
      "step": 1940
    },
    {
      "epoch": 0.14920098340503996,
      "grad_norm": 5.029245853424072,
      "learning_rate": 1.940319606637984e-05,
      "loss": 1.5623,
      "step": 1942
    },
    {
      "epoch": 0.14935464044253227,
      "grad_norm": 4.505579948425293,
      "learning_rate": 1.9402581438229874e-05,
      "loss": 1.4555,
      "step": 1944
    },
    {
      "epoch": 0.1495082974800246,
      "grad_norm": 4.269020080566406,
      "learning_rate": 1.9401966810079904e-05,
      "loss": 1.5765,
      "step": 1946
    },
    {
      "epoch": 0.1496619545175169,
      "grad_norm": 3.9689948558807373,
      "learning_rate": 1.9401352181929933e-05,
      "loss": 1.6277,
      "step": 1948
    },
    {
      "epoch": 0.14981561155500922,
      "grad_norm": 3.684130907058716,
      "learning_rate": 1.9400737553779966e-05,
      "loss": 1.5172,
      "step": 1950
    },
    {
      "epoch": 0.14996926859250154,
      "grad_norm": 3.713602066040039,
      "learning_rate": 1.9400122925629996e-05,
      "loss": 1.5211,
      "step": 1952
    },
    {
      "epoch": 0.15012292562999385,
      "grad_norm": 4.927131175994873,
      "learning_rate": 1.9399508297480026e-05,
      "loss": 1.6324,
      "step": 1954
    },
    {
      "epoch": 0.15027658266748617,
      "grad_norm": 4.4799299240112305,
      "learning_rate": 1.9398893669330055e-05,
      "loss": 1.5679,
      "step": 1956
    },
    {
      "epoch": 0.15043023970497849,
      "grad_norm": 4.372833251953125,
      "learning_rate": 1.939827904118009e-05,
      "loss": 1.5638,
      "step": 1958
    },
    {
      "epoch": 0.1505838967424708,
      "grad_norm": 4.321191787719727,
      "learning_rate": 1.9397664413030118e-05,
      "loss": 1.4329,
      "step": 1960
    },
    {
      "epoch": 0.15073755377996312,
      "grad_norm": 4.75023078918457,
      "learning_rate": 1.9397049784880148e-05,
      "loss": 1.6682,
      "step": 1962
    },
    {
      "epoch": 0.15089121081745543,
      "grad_norm": 4.172933101654053,
      "learning_rate": 1.939643515673018e-05,
      "loss": 1.7316,
      "step": 1964
    },
    {
      "epoch": 0.15104486785494775,
      "grad_norm": 4.329415321350098,
      "learning_rate": 1.939582052858021e-05,
      "loss": 1.7017,
      "step": 1966
    },
    {
      "epoch": 0.15119852489244007,
      "grad_norm": 4.249721527099609,
      "learning_rate": 1.939520590043024e-05,
      "loss": 1.5796,
      "step": 1968
    },
    {
      "epoch": 0.15135218192993238,
      "grad_norm": 4.071712970733643,
      "learning_rate": 1.9394591272280273e-05,
      "loss": 1.4623,
      "step": 1970
    },
    {
      "epoch": 0.1515058389674247,
      "grad_norm": 4.0507731437683105,
      "learning_rate": 1.9393976644130303e-05,
      "loss": 1.4821,
      "step": 1972
    },
    {
      "epoch": 0.15165949600491702,
      "grad_norm": 4.356963634490967,
      "learning_rate": 1.9393362015980333e-05,
      "loss": 1.547,
      "step": 1974
    },
    {
      "epoch": 0.15181315304240933,
      "grad_norm": 5.182737350463867,
      "learning_rate": 1.9392747387830362e-05,
      "loss": 1.6367,
      "step": 1976
    },
    {
      "epoch": 0.15196681007990165,
      "grad_norm": 3.9492363929748535,
      "learning_rate": 1.9392132759680395e-05,
      "loss": 1.679,
      "step": 1978
    },
    {
      "epoch": 0.15212046711739396,
      "grad_norm": 4.404160976409912,
      "learning_rate": 1.9391518131530425e-05,
      "loss": 1.5288,
      "step": 1980
    },
    {
      "epoch": 0.15227412415488628,
      "grad_norm": 4.500973701477051,
      "learning_rate": 1.9390903503380455e-05,
      "loss": 1.7715,
      "step": 1982
    },
    {
      "epoch": 0.1524277811923786,
      "grad_norm": 4.121068954467773,
      "learning_rate": 1.9390288875230488e-05,
      "loss": 1.5507,
      "step": 1984
    },
    {
      "epoch": 0.15258143822987094,
      "grad_norm": 3.7095515727996826,
      "learning_rate": 1.9389674247080518e-05,
      "loss": 1.616,
      "step": 1986
    },
    {
      "epoch": 0.15273509526736326,
      "grad_norm": 5.333407878875732,
      "learning_rate": 1.9389059618930547e-05,
      "loss": 1.7196,
      "step": 1988
    },
    {
      "epoch": 0.15288875230485557,
      "grad_norm": 4.188971042633057,
      "learning_rate": 1.938844499078058e-05,
      "loss": 1.6256,
      "step": 1990
    },
    {
      "epoch": 0.1530424093423479,
      "grad_norm": 4.126604080200195,
      "learning_rate": 1.938783036263061e-05,
      "loss": 1.5089,
      "step": 1992
    },
    {
      "epoch": 0.1531960663798402,
      "grad_norm": 4.127197742462158,
      "learning_rate": 1.938721573448064e-05,
      "loss": 1.5598,
      "step": 1994
    },
    {
      "epoch": 0.15334972341733252,
      "grad_norm": 4.481958389282227,
      "learning_rate": 1.9386601106330673e-05,
      "loss": 1.6238,
      "step": 1996
    },
    {
      "epoch": 0.15350338045482484,
      "grad_norm": 4.15784215927124,
      "learning_rate": 1.9385986478180702e-05,
      "loss": 1.6611,
      "step": 1998
    },
    {
      "epoch": 0.15365703749231716,
      "grad_norm": 4.322861194610596,
      "learning_rate": 1.9385371850030732e-05,
      "loss": 1.513,
      "step": 2000
    },
    {
      "epoch": 0.15381069452980947,
      "grad_norm": 3.9926345348358154,
      "learning_rate": 1.9384757221880762e-05,
      "loss": 1.5764,
      "step": 2002
    },
    {
      "epoch": 0.1539643515673018,
      "grad_norm": 5.112368583679199,
      "learning_rate": 1.9384142593730795e-05,
      "loss": 1.6181,
      "step": 2004
    },
    {
      "epoch": 0.1541180086047941,
      "grad_norm": 3.6655466556549072,
      "learning_rate": 1.9383527965580825e-05,
      "loss": 1.3721,
      "step": 2006
    },
    {
      "epoch": 0.15427166564228642,
      "grad_norm": 4.533908367156982,
      "learning_rate": 1.9382913337430854e-05,
      "loss": 1.5841,
      "step": 2008
    },
    {
      "epoch": 0.15442532267977874,
      "grad_norm": 3.722304344177246,
      "learning_rate": 1.9382298709280887e-05,
      "loss": 1.5965,
      "step": 2010
    },
    {
      "epoch": 0.15457897971727105,
      "grad_norm": 4.5336012840271,
      "learning_rate": 1.9381684081130917e-05,
      "loss": 1.5995,
      "step": 2012
    },
    {
      "epoch": 0.15473263675476337,
      "grad_norm": 4.144698619842529,
      "learning_rate": 1.9381069452980947e-05,
      "loss": 1.6343,
      "step": 2014
    },
    {
      "epoch": 0.15488629379225569,
      "grad_norm": 4.033304691314697,
      "learning_rate": 1.938045482483098e-05,
      "loss": 1.6888,
      "step": 2016
    },
    {
      "epoch": 0.155039950829748,
      "grad_norm": 4.00151252746582,
      "learning_rate": 1.937984019668101e-05,
      "loss": 1.5246,
      "step": 2018
    },
    {
      "epoch": 0.15519360786724032,
      "grad_norm": 4.766987323760986,
      "learning_rate": 1.937922556853104e-05,
      "loss": 1.7345,
      "step": 2020
    },
    {
      "epoch": 0.15534726490473263,
      "grad_norm": 4.144290924072266,
      "learning_rate": 1.9378610940381072e-05,
      "loss": 1.5848,
      "step": 2022
    },
    {
      "epoch": 0.15550092194222495,
      "grad_norm": 4.038874626159668,
      "learning_rate": 1.9377996312231102e-05,
      "loss": 1.6425,
      "step": 2024
    },
    {
      "epoch": 0.15565457897971727,
      "grad_norm": 3.832429885864258,
      "learning_rate": 1.9377381684081135e-05,
      "loss": 1.5454,
      "step": 2026
    },
    {
      "epoch": 0.15580823601720958,
      "grad_norm": 4.775087833404541,
      "learning_rate": 1.937676705593116e-05,
      "loss": 1.6767,
      "step": 2028
    },
    {
      "epoch": 0.1559618930547019,
      "grad_norm": 3.997192144393921,
      "learning_rate": 1.9376152427781194e-05,
      "loss": 1.5539,
      "step": 2030
    },
    {
      "epoch": 0.15611555009219422,
      "grad_norm": 3.764519453048706,
      "learning_rate": 1.9375537799631224e-05,
      "loss": 1.46,
      "step": 2032
    },
    {
      "epoch": 0.15626920712968653,
      "grad_norm": 4.074234962463379,
      "learning_rate": 1.9374923171481254e-05,
      "loss": 1.5064,
      "step": 2034
    },
    {
      "epoch": 0.15642286416717885,
      "grad_norm": 4.302229881286621,
      "learning_rate": 1.9374308543331287e-05,
      "loss": 1.6276,
      "step": 2036
    },
    {
      "epoch": 0.15657652120467117,
      "grad_norm": 4.602327346801758,
      "learning_rate": 1.9373693915181316e-05,
      "loss": 1.5551,
      "step": 2038
    },
    {
      "epoch": 0.15673017824216348,
      "grad_norm": 4.131155490875244,
      "learning_rate": 1.9373079287031346e-05,
      "loss": 1.4554,
      "step": 2040
    },
    {
      "epoch": 0.1568838352796558,
      "grad_norm": 4.661727428436279,
      "learning_rate": 1.937246465888138e-05,
      "loss": 1.6566,
      "step": 2042
    },
    {
      "epoch": 0.15703749231714811,
      "grad_norm": 4.23723030090332,
      "learning_rate": 1.937185003073141e-05,
      "loss": 1.6038,
      "step": 2044
    },
    {
      "epoch": 0.15719114935464043,
      "grad_norm": 3.995077133178711,
      "learning_rate": 1.9371235402581442e-05,
      "loss": 1.7374,
      "step": 2046
    },
    {
      "epoch": 0.15734480639213275,
      "grad_norm": 4.001912593841553,
      "learning_rate": 1.937062077443147e-05,
      "loss": 1.5991,
      "step": 2048
    },
    {
      "epoch": 0.15749846342962506,
      "grad_norm": 4.558352470397949,
      "learning_rate": 1.93700061462815e-05,
      "loss": 1.6115,
      "step": 2050
    },
    {
      "epoch": 0.1576521204671174,
      "grad_norm": 4.651041030883789,
      "learning_rate": 1.9369391518131534e-05,
      "loss": 1.5451,
      "step": 2052
    },
    {
      "epoch": 0.15780577750460972,
      "grad_norm": 4.203875541687012,
      "learning_rate": 1.936877688998156e-05,
      "loss": 1.5506,
      "step": 2054
    },
    {
      "epoch": 0.15795943454210204,
      "grad_norm": 4.153205394744873,
      "learning_rate": 1.9368162261831594e-05,
      "loss": 1.5209,
      "step": 2056
    },
    {
      "epoch": 0.15811309157959436,
      "grad_norm": 4.187285900115967,
      "learning_rate": 1.9367547633681623e-05,
      "loss": 1.7677,
      "step": 2058
    },
    {
      "epoch": 0.15826674861708667,
      "grad_norm": 4.654025554656982,
      "learning_rate": 1.9366933005531653e-05,
      "loss": 1.63,
      "step": 2060
    },
    {
      "epoch": 0.158420405654579,
      "grad_norm": 4.145837306976318,
      "learning_rate": 1.9366318377381686e-05,
      "loss": 1.7693,
      "step": 2062
    },
    {
      "epoch": 0.1585740626920713,
      "grad_norm": 4.076268196105957,
      "learning_rate": 1.9365703749231716e-05,
      "loss": 1.5436,
      "step": 2064
    },
    {
      "epoch": 0.15872771972956362,
      "grad_norm": 3.9100687503814697,
      "learning_rate": 1.936508912108175e-05,
      "loss": 1.5738,
      "step": 2066
    },
    {
      "epoch": 0.15888137676705594,
      "grad_norm": 4.173727989196777,
      "learning_rate": 1.936447449293178e-05,
      "loss": 1.6155,
      "step": 2068
    },
    {
      "epoch": 0.15903503380454825,
      "grad_norm": 4.555833339691162,
      "learning_rate": 1.936385986478181e-05,
      "loss": 1.6879,
      "step": 2070
    },
    {
      "epoch": 0.15918869084204057,
      "grad_norm": 4.008725166320801,
      "learning_rate": 1.936324523663184e-05,
      "loss": 1.5752,
      "step": 2072
    },
    {
      "epoch": 0.1593423478795329,
      "grad_norm": 5.628279209136963,
      "learning_rate": 1.9362630608481868e-05,
      "loss": 1.6574,
      "step": 2074
    },
    {
      "epoch": 0.1594960049170252,
      "grad_norm": 4.112339496612549,
      "learning_rate": 1.93620159803319e-05,
      "loss": 1.6789,
      "step": 2076
    },
    {
      "epoch": 0.15964966195451752,
      "grad_norm": 4.378570556640625,
      "learning_rate": 1.9361401352181934e-05,
      "loss": 1.4837,
      "step": 2078
    },
    {
      "epoch": 0.15980331899200984,
      "grad_norm": 3.819582223892212,
      "learning_rate": 1.936078672403196e-05,
      "loss": 1.573,
      "step": 2080
    },
    {
      "epoch": 0.15995697602950215,
      "grad_norm": 4.303280830383301,
      "learning_rate": 1.9360172095881993e-05,
      "loss": 1.4278,
      "step": 2082
    },
    {
      "epoch": 0.16011063306699447,
      "grad_norm": 3.9183764457702637,
      "learning_rate": 1.9359557467732023e-05,
      "loss": 1.4808,
      "step": 2084
    },
    {
      "epoch": 0.16026429010448678,
      "grad_norm": 4.484659671783447,
      "learning_rate": 1.9358942839582053e-05,
      "loss": 1.557,
      "step": 2086
    },
    {
      "epoch": 0.1604179471419791,
      "grad_norm": 4.600409984588623,
      "learning_rate": 1.9358328211432086e-05,
      "loss": 1.5848,
      "step": 2088
    },
    {
      "epoch": 0.16057160417947142,
      "grad_norm": 3.4427108764648438,
      "learning_rate": 1.9357713583282115e-05,
      "loss": 1.5589,
      "step": 2090
    },
    {
      "epoch": 0.16072526121696373,
      "grad_norm": 7.8328375816345215,
      "learning_rate": 1.935709895513215e-05,
      "loss": 1.5519,
      "step": 2092
    },
    {
      "epoch": 0.16087891825445605,
      "grad_norm": 4.630655765533447,
      "learning_rate": 1.9356484326982178e-05,
      "loss": 1.6441,
      "step": 2094
    },
    {
      "epoch": 0.16103257529194837,
      "grad_norm": 4.217165946960449,
      "learning_rate": 1.9355869698832208e-05,
      "loss": 1.5559,
      "step": 2096
    },
    {
      "epoch": 0.16118623232944068,
      "grad_norm": 4.941442012786865,
      "learning_rate": 1.935525507068224e-05,
      "loss": 1.5065,
      "step": 2098
    },
    {
      "epoch": 0.161339889366933,
      "grad_norm": 4.171699523925781,
      "learning_rate": 1.9354640442532267e-05,
      "loss": 1.4993,
      "step": 2100
    },
    {
      "epoch": 0.16149354640442531,
      "grad_norm": 3.8235647678375244,
      "learning_rate": 1.93540258143823e-05,
      "loss": 1.6035,
      "step": 2102
    },
    {
      "epoch": 0.16164720344191763,
      "grad_norm": 5.202590465545654,
      "learning_rate": 1.935341118623233e-05,
      "loss": 1.6397,
      "step": 2104
    },
    {
      "epoch": 0.16180086047940995,
      "grad_norm": 3.8114516735076904,
      "learning_rate": 1.935279655808236e-05,
      "loss": 1.5103,
      "step": 2106
    },
    {
      "epoch": 0.16195451751690226,
      "grad_norm": 4.666793346405029,
      "learning_rate": 1.9352181929932393e-05,
      "loss": 1.5363,
      "step": 2108
    },
    {
      "epoch": 0.16210817455439458,
      "grad_norm": 4.18889856338501,
      "learning_rate": 1.9351567301782422e-05,
      "loss": 1.662,
      "step": 2110
    },
    {
      "epoch": 0.1622618315918869,
      "grad_norm": 4.392826557159424,
      "learning_rate": 1.9350952673632455e-05,
      "loss": 1.5492,
      "step": 2112
    },
    {
      "epoch": 0.1624154886293792,
      "grad_norm": 4.0653839111328125,
      "learning_rate": 1.9350338045482485e-05,
      "loss": 1.7512,
      "step": 2114
    },
    {
      "epoch": 0.16256914566687156,
      "grad_norm": 4.020386219024658,
      "learning_rate": 1.9349723417332515e-05,
      "loss": 1.7604,
      "step": 2116
    },
    {
      "epoch": 0.16272280270436387,
      "grad_norm": 4.062155723571777,
      "learning_rate": 1.9349108789182548e-05,
      "loss": 1.7181,
      "step": 2118
    },
    {
      "epoch": 0.1628764597418562,
      "grad_norm": 4.4019999504089355,
      "learning_rate": 1.9348494161032578e-05,
      "loss": 1.639,
      "step": 2120
    },
    {
      "epoch": 0.1630301167793485,
      "grad_norm": 3.761319875717163,
      "learning_rate": 1.9347879532882607e-05,
      "loss": 1.4084,
      "step": 2122
    },
    {
      "epoch": 0.16318377381684082,
      "grad_norm": 4.566369533538818,
      "learning_rate": 1.934726490473264e-05,
      "loss": 1.5193,
      "step": 2124
    },
    {
      "epoch": 0.16333743085433314,
      "grad_norm": 4.366701602935791,
      "learning_rate": 1.9346650276582667e-05,
      "loss": 1.5211,
      "step": 2126
    },
    {
      "epoch": 0.16349108789182545,
      "grad_norm": 4.0437116622924805,
      "learning_rate": 1.93460356484327e-05,
      "loss": 1.5901,
      "step": 2128
    },
    {
      "epoch": 0.16364474492931777,
      "grad_norm": 3.9914474487304688,
      "learning_rate": 1.934542102028273e-05,
      "loss": 1.4975,
      "step": 2130
    },
    {
      "epoch": 0.1637984019668101,
      "grad_norm": 4.267744541168213,
      "learning_rate": 1.9344806392132762e-05,
      "loss": 1.6574,
      "step": 2132
    },
    {
      "epoch": 0.1639520590043024,
      "grad_norm": 3.843414068222046,
      "learning_rate": 1.9344191763982792e-05,
      "loss": 1.4976,
      "step": 2134
    },
    {
      "epoch": 0.16410571604179472,
      "grad_norm": 4.3155975341796875,
      "learning_rate": 1.9343577135832822e-05,
      "loss": 1.5333,
      "step": 2136
    },
    {
      "epoch": 0.16425937307928704,
      "grad_norm": 4.159292697906494,
      "learning_rate": 1.9342962507682855e-05,
      "loss": 1.5053,
      "step": 2138
    },
    {
      "epoch": 0.16441303011677935,
      "grad_norm": 4.296607971191406,
      "learning_rate": 1.9342347879532885e-05,
      "loss": 1.6112,
      "step": 2140
    },
    {
      "epoch": 0.16456668715427167,
      "grad_norm": 3.9782469272613525,
      "learning_rate": 1.9341733251382914e-05,
      "loss": 1.5241,
      "step": 2142
    },
    {
      "epoch": 0.16472034419176398,
      "grad_norm": 4.183950901031494,
      "learning_rate": 1.9341118623232947e-05,
      "loss": 1.5901,
      "step": 2144
    },
    {
      "epoch": 0.1648740012292563,
      "grad_norm": 3.9188575744628906,
      "learning_rate": 1.9340503995082977e-05,
      "loss": 1.621,
      "step": 2146
    },
    {
      "epoch": 0.16502765826674862,
      "grad_norm": 4.474676132202148,
      "learning_rate": 1.9339889366933007e-05,
      "loss": 1.487,
      "step": 2148
    },
    {
      "epoch": 0.16518131530424093,
      "grad_norm": 4.6410017013549805,
      "learning_rate": 1.933927473878304e-05,
      "loss": 1.6997,
      "step": 2150
    },
    {
      "epoch": 0.16533497234173325,
      "grad_norm": 10.933167457580566,
      "learning_rate": 1.933866011063307e-05,
      "loss": 1.4752,
      "step": 2152
    },
    {
      "epoch": 0.16548862937922557,
      "grad_norm": 4.885360240936279,
      "learning_rate": 1.93380454824831e-05,
      "loss": 1.5604,
      "step": 2154
    },
    {
      "epoch": 0.16564228641671788,
      "grad_norm": 3.9420063495635986,
      "learning_rate": 1.933743085433313e-05,
      "loss": 1.5909,
      "step": 2156
    },
    {
      "epoch": 0.1657959434542102,
      "grad_norm": 3.5523521900177,
      "learning_rate": 1.9336816226183162e-05,
      "loss": 1.6274,
      "step": 2158
    },
    {
      "epoch": 0.16594960049170251,
      "grad_norm": 3.6621317863464355,
      "learning_rate": 1.933620159803319e-05,
      "loss": 1.5111,
      "step": 2160
    },
    {
      "epoch": 0.16610325752919483,
      "grad_norm": 4.57207727432251,
      "learning_rate": 1.933558696988322e-05,
      "loss": 1.6702,
      "step": 2162
    },
    {
      "epoch": 0.16625691456668715,
      "grad_norm": 3.9489428997039795,
      "learning_rate": 1.9334972341733254e-05,
      "loss": 1.589,
      "step": 2164
    },
    {
      "epoch": 0.16641057160417946,
      "grad_norm": 3.8509654998779297,
      "learning_rate": 1.9334357713583284e-05,
      "loss": 1.5522,
      "step": 2166
    },
    {
      "epoch": 0.16656422864167178,
      "grad_norm": 3.6893975734710693,
      "learning_rate": 1.9333743085433314e-05,
      "loss": 1.4767,
      "step": 2168
    },
    {
      "epoch": 0.1667178856791641,
      "grad_norm": 4.064876079559326,
      "learning_rate": 1.9333128457283347e-05,
      "loss": 1.5515,
      "step": 2170
    },
    {
      "epoch": 0.1668715427166564,
      "grad_norm": 4.427343845367432,
      "learning_rate": 1.9332513829133376e-05,
      "loss": 1.7098,
      "step": 2172
    },
    {
      "epoch": 0.16702519975414873,
      "grad_norm": 3.724740505218506,
      "learning_rate": 1.9331899200983406e-05,
      "loss": 1.5612,
      "step": 2174
    },
    {
      "epoch": 0.16717885679164105,
      "grad_norm": 4.302028656005859,
      "learning_rate": 1.933128457283344e-05,
      "loss": 1.641,
      "step": 2176
    },
    {
      "epoch": 0.16733251382913336,
      "grad_norm": 4.149264335632324,
      "learning_rate": 1.933066994468347e-05,
      "loss": 1.6923,
      "step": 2178
    },
    {
      "epoch": 0.16748617086662568,
      "grad_norm": 4.091360092163086,
      "learning_rate": 1.93300553165335e-05,
      "loss": 1.4934,
      "step": 2180
    },
    {
      "epoch": 0.16763982790411802,
      "grad_norm": 4.653087139129639,
      "learning_rate": 1.9329440688383528e-05,
      "loss": 1.6434,
      "step": 2182
    },
    {
      "epoch": 0.16779348494161034,
      "grad_norm": 4.141650199890137,
      "learning_rate": 1.932882606023356e-05,
      "loss": 1.5228,
      "step": 2184
    },
    {
      "epoch": 0.16794714197910265,
      "grad_norm": 4.257270336151123,
      "learning_rate": 1.932821143208359e-05,
      "loss": 1.7463,
      "step": 2186
    },
    {
      "epoch": 0.16810079901659497,
      "grad_norm": 4.1163554191589355,
      "learning_rate": 1.932759680393362e-05,
      "loss": 1.4543,
      "step": 2188
    },
    {
      "epoch": 0.1682544560540873,
      "grad_norm": 3.6683640480041504,
      "learning_rate": 1.9326982175783654e-05,
      "loss": 1.4778,
      "step": 2190
    },
    {
      "epoch": 0.1684081130915796,
      "grad_norm": 3.734006881713867,
      "learning_rate": 1.9326367547633683e-05,
      "loss": 1.5738,
      "step": 2192
    },
    {
      "epoch": 0.16856177012907192,
      "grad_norm": 4.454776287078857,
      "learning_rate": 1.9325752919483713e-05,
      "loss": 1.6569,
      "step": 2194
    },
    {
      "epoch": 0.16871542716656424,
      "grad_norm": 4.1497883796691895,
      "learning_rate": 1.9325138291333746e-05,
      "loss": 1.5172,
      "step": 2196
    },
    {
      "epoch": 0.16886908420405655,
      "grad_norm": 4.288064479827881,
      "learning_rate": 1.9324523663183776e-05,
      "loss": 1.6416,
      "step": 2198
    },
    {
      "epoch": 0.16902274124154887,
      "grad_norm": 3.463115930557251,
      "learning_rate": 1.9323909035033806e-05,
      "loss": 1.4196,
      "step": 2200
    },
    {
      "epoch": 0.16917639827904118,
      "grad_norm": 5.139834403991699,
      "learning_rate": 1.9323294406883835e-05,
      "loss": 1.6058,
      "step": 2202
    },
    {
      "epoch": 0.1693300553165335,
      "grad_norm": 4.1170806884765625,
      "learning_rate": 1.9322679778733868e-05,
      "loss": 1.4493,
      "step": 2204
    },
    {
      "epoch": 0.16948371235402582,
      "grad_norm": 3.8090291023254395,
      "learning_rate": 1.9322065150583898e-05,
      "loss": 1.6205,
      "step": 2206
    },
    {
      "epoch": 0.16963736939151813,
      "grad_norm": 3.461530923843384,
      "learning_rate": 1.9321450522433928e-05,
      "loss": 1.4338,
      "step": 2208
    },
    {
      "epoch": 0.16979102642901045,
      "grad_norm": 4.355661392211914,
      "learning_rate": 1.932083589428396e-05,
      "loss": 1.6044,
      "step": 2210
    },
    {
      "epoch": 0.16994468346650277,
      "grad_norm": 3.4141671657562256,
      "learning_rate": 1.932022126613399e-05,
      "loss": 1.5096,
      "step": 2212
    },
    {
      "epoch": 0.17009834050399508,
      "grad_norm": 4.202045917510986,
      "learning_rate": 1.931960663798402e-05,
      "loss": 1.7529,
      "step": 2214
    },
    {
      "epoch": 0.1702519975414874,
      "grad_norm": 3.8714635372161865,
      "learning_rate": 1.9318992009834053e-05,
      "loss": 1.5085,
      "step": 2216
    },
    {
      "epoch": 0.17040565457897972,
      "grad_norm": 5.047643184661865,
      "learning_rate": 1.9318377381684083e-05,
      "loss": 1.6937,
      "step": 2218
    },
    {
      "epoch": 0.17055931161647203,
      "grad_norm": 4.094550132751465,
      "learning_rate": 1.9317762753534113e-05,
      "loss": 1.7053,
      "step": 2220
    },
    {
      "epoch": 0.17071296865396435,
      "grad_norm": 3.65474796295166,
      "learning_rate": 1.9317148125384146e-05,
      "loss": 1.5531,
      "step": 2222
    },
    {
      "epoch": 0.17086662569145666,
      "grad_norm": 4.38794469833374,
      "learning_rate": 1.9316533497234175e-05,
      "loss": 1.7151,
      "step": 2224
    },
    {
      "epoch": 0.17102028272894898,
      "grad_norm": 4.336912631988525,
      "learning_rate": 1.9315918869084205e-05,
      "loss": 1.4871,
      "step": 2226
    },
    {
      "epoch": 0.1711739397664413,
      "grad_norm": 3.9137933254241943,
      "learning_rate": 1.9315304240934235e-05,
      "loss": 1.3785,
      "step": 2228
    },
    {
      "epoch": 0.1713275968039336,
      "grad_norm": 4.241963863372803,
      "learning_rate": 1.9314689612784268e-05,
      "loss": 1.482,
      "step": 2230
    },
    {
      "epoch": 0.17148125384142593,
      "grad_norm": 4.047958850860596,
      "learning_rate": 1.9314074984634297e-05,
      "loss": 1.7098,
      "step": 2232
    },
    {
      "epoch": 0.17163491087891825,
      "grad_norm": 4.688525676727295,
      "learning_rate": 1.9313460356484327e-05,
      "loss": 1.5618,
      "step": 2234
    },
    {
      "epoch": 0.17178856791641056,
      "grad_norm": 4.020751953125,
      "learning_rate": 1.931284572833436e-05,
      "loss": 1.5526,
      "step": 2236
    },
    {
      "epoch": 0.17194222495390288,
      "grad_norm": 3.93445086479187,
      "learning_rate": 1.931223110018439e-05,
      "loss": 1.6595,
      "step": 2238
    },
    {
      "epoch": 0.1720958819913952,
      "grad_norm": 5.324620723724365,
      "learning_rate": 1.931161647203442e-05,
      "loss": 1.657,
      "step": 2240
    },
    {
      "epoch": 0.1722495390288875,
      "grad_norm": 3.6193902492523193,
      "learning_rate": 1.9311001843884453e-05,
      "loss": 1.5,
      "step": 2242
    },
    {
      "epoch": 0.17240319606637983,
      "grad_norm": 4.382716178894043,
      "learning_rate": 1.9310387215734482e-05,
      "loss": 1.6785,
      "step": 2244
    },
    {
      "epoch": 0.17255685310387217,
      "grad_norm": 3.7359304428100586,
      "learning_rate": 1.9309772587584512e-05,
      "loss": 1.5003,
      "step": 2246
    },
    {
      "epoch": 0.1727105101413645,
      "grad_norm": 4.570140838623047,
      "learning_rate": 1.9309157959434545e-05,
      "loss": 1.6353,
      "step": 2248
    },
    {
      "epoch": 0.1728641671788568,
      "grad_norm": 4.809631824493408,
      "learning_rate": 1.9308543331284575e-05,
      "loss": 1.5898,
      "step": 2250
    },
    {
      "epoch": 0.17301782421634912,
      "grad_norm": 4.994627475738525,
      "learning_rate": 1.9307928703134604e-05,
      "loss": 1.6614,
      "step": 2252
    },
    {
      "epoch": 0.17317148125384144,
      "grad_norm": 4.121060371398926,
      "learning_rate": 1.9307314074984634e-05,
      "loss": 1.6817,
      "step": 2254
    },
    {
      "epoch": 0.17332513829133375,
      "grad_norm": 4.009014129638672,
      "learning_rate": 1.9306699446834667e-05,
      "loss": 1.5645,
      "step": 2256
    },
    {
      "epoch": 0.17347879532882607,
      "grad_norm": 4.27223539352417,
      "learning_rate": 1.9306084818684697e-05,
      "loss": 1.6764,
      "step": 2258
    },
    {
      "epoch": 0.17363245236631838,
      "grad_norm": 4.074213027954102,
      "learning_rate": 1.9305470190534727e-05,
      "loss": 1.6095,
      "step": 2260
    },
    {
      "epoch": 0.1737861094038107,
      "grad_norm": 3.6030173301696777,
      "learning_rate": 1.930485556238476e-05,
      "loss": 1.5907,
      "step": 2262
    },
    {
      "epoch": 0.17393976644130302,
      "grad_norm": 4.34961462020874,
      "learning_rate": 1.930424093423479e-05,
      "loss": 1.611,
      "step": 2264
    },
    {
      "epoch": 0.17409342347879533,
      "grad_norm": 3.9723429679870605,
      "learning_rate": 1.930362630608482e-05,
      "loss": 1.5256,
      "step": 2266
    },
    {
      "epoch": 0.17424708051628765,
      "grad_norm": 3.7899746894836426,
      "learning_rate": 1.9303011677934852e-05,
      "loss": 1.6923,
      "step": 2268
    },
    {
      "epoch": 0.17440073755377997,
      "grad_norm": 4.415828227996826,
      "learning_rate": 1.9302397049784882e-05,
      "loss": 1.4681,
      "step": 2270
    },
    {
      "epoch": 0.17455439459127228,
      "grad_norm": 5.616640567779541,
      "learning_rate": 1.930178242163491e-05,
      "loss": 1.6252,
      "step": 2272
    },
    {
      "epoch": 0.1747080516287646,
      "grad_norm": 3.5017950534820557,
      "learning_rate": 1.9301167793484944e-05,
      "loss": 1.4621,
      "step": 2274
    },
    {
      "epoch": 0.17486170866625692,
      "grad_norm": 4.120169639587402,
      "learning_rate": 1.9300553165334974e-05,
      "loss": 1.5537,
      "step": 2276
    },
    {
      "epoch": 0.17501536570374923,
      "grad_norm": 4.489522457122803,
      "learning_rate": 1.9299938537185007e-05,
      "loss": 1.6915,
      "step": 2278
    },
    {
      "epoch": 0.17516902274124155,
      "grad_norm": 4.285830974578857,
      "learning_rate": 1.9299323909035034e-05,
      "loss": 1.5101,
      "step": 2280
    },
    {
      "epoch": 0.17532267977873386,
      "grad_norm": 3.9038150310516357,
      "learning_rate": 1.9298709280885067e-05,
      "loss": 1.6067,
      "step": 2282
    },
    {
      "epoch": 0.17547633681622618,
      "grad_norm": 3.8521156311035156,
      "learning_rate": 1.9298094652735096e-05,
      "loss": 1.4494,
      "step": 2284
    },
    {
      "epoch": 0.1756299938537185,
      "grad_norm": 7.329223155975342,
      "learning_rate": 1.9297480024585126e-05,
      "loss": 1.5588,
      "step": 2286
    },
    {
      "epoch": 0.1757836508912108,
      "grad_norm": 3.97939395904541,
      "learning_rate": 1.929686539643516e-05,
      "loss": 1.4802,
      "step": 2288
    },
    {
      "epoch": 0.17593730792870313,
      "grad_norm": 3.464115858078003,
      "learning_rate": 1.929625076828519e-05,
      "loss": 1.4913,
      "step": 2290
    },
    {
      "epoch": 0.17609096496619545,
      "grad_norm": 4.677506446838379,
      "learning_rate": 1.929563614013522e-05,
      "loss": 1.5254,
      "step": 2292
    },
    {
      "epoch": 0.17624462200368776,
      "grad_norm": 3.7886929512023926,
      "learning_rate": 1.929502151198525e-05,
      "loss": 1.6297,
      "step": 2294
    },
    {
      "epoch": 0.17639827904118008,
      "grad_norm": 3.5035488605499268,
      "learning_rate": 1.929440688383528e-05,
      "loss": 1.657,
      "step": 2296
    },
    {
      "epoch": 0.1765519360786724,
      "grad_norm": 4.172173976898193,
      "learning_rate": 1.9293792255685314e-05,
      "loss": 1.5612,
      "step": 2298
    },
    {
      "epoch": 0.1767055931161647,
      "grad_norm": 3.9481425285339355,
      "learning_rate": 1.929317762753534e-05,
      "loss": 1.4419,
      "step": 2300
    },
    {
      "epoch": 0.17685925015365703,
      "grad_norm": 3.922159433364868,
      "learning_rate": 1.9292562999385374e-05,
      "loss": 1.5459,
      "step": 2302
    },
    {
      "epoch": 0.17701290719114934,
      "grad_norm": 4.2247233390808105,
      "learning_rate": 1.9291948371235403e-05,
      "loss": 1.6869,
      "step": 2304
    },
    {
      "epoch": 0.17716656422864166,
      "grad_norm": 4.960201740264893,
      "learning_rate": 1.9291333743085433e-05,
      "loss": 1.4055,
      "step": 2306
    },
    {
      "epoch": 0.17732022126613398,
      "grad_norm": 4.675178527832031,
      "learning_rate": 1.9290719114935466e-05,
      "loss": 1.4635,
      "step": 2308
    },
    {
      "epoch": 0.1774738783036263,
      "grad_norm": 4.3724589347839355,
      "learning_rate": 1.9290104486785496e-05,
      "loss": 1.4423,
      "step": 2310
    },
    {
      "epoch": 0.17762753534111864,
      "grad_norm": 4.629543304443359,
      "learning_rate": 1.9289489858635525e-05,
      "loss": 1.6696,
      "step": 2312
    },
    {
      "epoch": 0.17778119237861095,
      "grad_norm": 3.8183395862579346,
      "learning_rate": 1.928887523048556e-05,
      "loss": 1.5558,
      "step": 2314
    },
    {
      "epoch": 0.17793484941610327,
      "grad_norm": 3.7984275817871094,
      "learning_rate": 1.9288260602335588e-05,
      "loss": 1.5884,
      "step": 2316
    },
    {
      "epoch": 0.17808850645359559,
      "grad_norm": 3.9068145751953125,
      "learning_rate": 1.928764597418562e-05,
      "loss": 1.6217,
      "step": 2318
    },
    {
      "epoch": 0.1782421634910879,
      "grad_norm": 4.159458160400391,
      "learning_rate": 1.928703134603565e-05,
      "loss": 1.5934,
      "step": 2320
    },
    {
      "epoch": 0.17839582052858022,
      "grad_norm": 4.013321876525879,
      "learning_rate": 1.928641671788568e-05,
      "loss": 1.5542,
      "step": 2322
    },
    {
      "epoch": 0.17854947756607253,
      "grad_norm": 4.504942893981934,
      "learning_rate": 1.9285802089735714e-05,
      "loss": 1.7811,
      "step": 2324
    },
    {
      "epoch": 0.17870313460356485,
      "grad_norm": 4.721269130706787,
      "learning_rate": 1.928518746158574e-05,
      "loss": 1.4974,
      "step": 2326
    },
    {
      "epoch": 0.17885679164105717,
      "grad_norm": 3.662440776824951,
      "learning_rate": 1.9284572833435773e-05,
      "loss": 1.6497,
      "step": 2328
    },
    {
      "epoch": 0.17901044867854948,
      "grad_norm": 3.8075759410858154,
      "learning_rate": 1.9283958205285803e-05,
      "loss": 1.4787,
      "step": 2330
    },
    {
      "epoch": 0.1791641057160418,
      "grad_norm": 4.013290882110596,
      "learning_rate": 1.9283343577135832e-05,
      "loss": 1.5371,
      "step": 2332
    },
    {
      "epoch": 0.17931776275353412,
      "grad_norm": 4.095331192016602,
      "learning_rate": 1.9282728948985865e-05,
      "loss": 1.6089,
      "step": 2334
    },
    {
      "epoch": 0.17947141979102643,
      "grad_norm": 4.137665748596191,
      "learning_rate": 1.9282114320835895e-05,
      "loss": 1.6971,
      "step": 2336
    },
    {
      "epoch": 0.17962507682851875,
      "grad_norm": 4.847195625305176,
      "learning_rate": 1.9281499692685928e-05,
      "loss": 1.365,
      "step": 2338
    },
    {
      "epoch": 0.17977873386601106,
      "grad_norm": 4.068114280700684,
      "learning_rate": 1.9280885064535958e-05,
      "loss": 1.7029,
      "step": 2340
    },
    {
      "epoch": 0.17993239090350338,
      "grad_norm": 4.104188442230225,
      "learning_rate": 1.9280270436385988e-05,
      "loss": 1.636,
      "step": 2342
    },
    {
      "epoch": 0.1800860479409957,
      "grad_norm": 4.033984661102295,
      "learning_rate": 1.927965580823602e-05,
      "loss": 1.5451,
      "step": 2344
    },
    {
      "epoch": 0.180239704978488,
      "grad_norm": 4.00771951675415,
      "learning_rate": 1.927904118008605e-05,
      "loss": 1.5162,
      "step": 2346
    },
    {
      "epoch": 0.18039336201598033,
      "grad_norm": 4.097219467163086,
      "learning_rate": 1.927842655193608e-05,
      "loss": 1.6558,
      "step": 2348
    },
    {
      "epoch": 0.18054701905347265,
      "grad_norm": 4.354104042053223,
      "learning_rate": 1.9277811923786113e-05,
      "loss": 1.61,
      "step": 2350
    },
    {
      "epoch": 0.18070067609096496,
      "grad_norm": 4.535645484924316,
      "learning_rate": 1.927719729563614e-05,
      "loss": 1.6289,
      "step": 2352
    },
    {
      "epoch": 0.18085433312845728,
      "grad_norm": 4.078785419464111,
      "learning_rate": 1.9276582667486172e-05,
      "loss": 1.7284,
      "step": 2354
    },
    {
      "epoch": 0.1810079901659496,
      "grad_norm": 4.275857448577881,
      "learning_rate": 1.9275968039336202e-05,
      "loss": 1.5461,
      "step": 2356
    },
    {
      "epoch": 0.1811616472034419,
      "grad_norm": 4.156821250915527,
      "learning_rate": 1.9275353411186232e-05,
      "loss": 1.5956,
      "step": 2358
    },
    {
      "epoch": 0.18131530424093423,
      "grad_norm": 3.6036367416381836,
      "learning_rate": 1.9274738783036265e-05,
      "loss": 1.5949,
      "step": 2360
    },
    {
      "epoch": 0.18146896127842654,
      "grad_norm": 4.079240798950195,
      "learning_rate": 1.9274124154886295e-05,
      "loss": 1.5671,
      "step": 2362
    },
    {
      "epoch": 0.18162261831591886,
      "grad_norm": 4.024125576019287,
      "learning_rate": 1.9273509526736328e-05,
      "loss": 1.4904,
      "step": 2364
    },
    {
      "epoch": 0.18177627535341118,
      "grad_norm": 3.7651987075805664,
      "learning_rate": 1.9272894898586357e-05,
      "loss": 1.5179,
      "step": 2366
    },
    {
      "epoch": 0.1819299323909035,
      "grad_norm": 3.8718831539154053,
      "learning_rate": 1.9272280270436387e-05,
      "loss": 1.4699,
      "step": 2368
    },
    {
      "epoch": 0.1820835894283958,
      "grad_norm": 4.548869609832764,
      "learning_rate": 1.927166564228642e-05,
      "loss": 1.6476,
      "step": 2370
    },
    {
      "epoch": 0.18223724646588813,
      "grad_norm": 4.528201580047607,
      "learning_rate": 1.927105101413645e-05,
      "loss": 1.6755,
      "step": 2372
    },
    {
      "epoch": 0.18239090350338044,
      "grad_norm": 4.0802388191223145,
      "learning_rate": 1.927043638598648e-05,
      "loss": 1.5722,
      "step": 2374
    },
    {
      "epoch": 0.18254456054087279,
      "grad_norm": 4.145775318145752,
      "learning_rate": 1.9269821757836513e-05,
      "loss": 1.5816,
      "step": 2376
    },
    {
      "epoch": 0.1826982175783651,
      "grad_norm": 3.925696611404419,
      "learning_rate": 1.926920712968654e-05,
      "loss": 1.625,
      "step": 2378
    },
    {
      "epoch": 0.18285187461585742,
      "grad_norm": 3.8499910831451416,
      "learning_rate": 1.9268592501536572e-05,
      "loss": 1.5522,
      "step": 2380
    },
    {
      "epoch": 0.18300553165334973,
      "grad_norm": 4.174738883972168,
      "learning_rate": 1.92679778733866e-05,
      "loss": 1.561,
      "step": 2382
    },
    {
      "epoch": 0.18315918869084205,
      "grad_norm": 3.801260232925415,
      "learning_rate": 1.9267363245236635e-05,
      "loss": 1.5159,
      "step": 2384
    },
    {
      "epoch": 0.18331284572833437,
      "grad_norm": 4.0040202140808105,
      "learning_rate": 1.9266748617086664e-05,
      "loss": 1.6021,
      "step": 2386
    },
    {
      "epoch": 0.18346650276582668,
      "grad_norm": 4.132852554321289,
      "learning_rate": 1.9266133988936694e-05,
      "loss": 1.6887,
      "step": 2388
    },
    {
      "epoch": 0.183620159803319,
      "grad_norm": 3.7313075065612793,
      "learning_rate": 1.9265519360786727e-05,
      "loss": 1.4582,
      "step": 2390
    },
    {
      "epoch": 0.18377381684081132,
      "grad_norm": 3.824453115463257,
      "learning_rate": 1.9264904732636757e-05,
      "loss": 1.4394,
      "step": 2392
    },
    {
      "epoch": 0.18392747387830363,
      "grad_norm": 4.368152141571045,
      "learning_rate": 1.9264290104486786e-05,
      "loss": 1.5397,
      "step": 2394
    },
    {
      "epoch": 0.18408113091579595,
      "grad_norm": 3.7525463104248047,
      "learning_rate": 1.926367547633682e-05,
      "loss": 1.6875,
      "step": 2396
    },
    {
      "epoch": 0.18423478795328826,
      "grad_norm": 4.229045391082764,
      "learning_rate": 1.9263060848186846e-05,
      "loss": 1.6248,
      "step": 2398
    },
    {
      "epoch": 0.18438844499078058,
      "grad_norm": 3.9596312046051025,
      "learning_rate": 1.926244622003688e-05,
      "loss": 1.5278,
      "step": 2400
    },
    {
      "epoch": 0.1845421020282729,
      "grad_norm": 3.97829270362854,
      "learning_rate": 1.9261831591886912e-05,
      "loss": 1.5206,
      "step": 2402
    },
    {
      "epoch": 0.1846957590657652,
      "grad_norm": 3.6935653686523438,
      "learning_rate": 1.926121696373694e-05,
      "loss": 1.5395,
      "step": 2404
    },
    {
      "epoch": 0.18484941610325753,
      "grad_norm": 3.869833469390869,
      "learning_rate": 1.926060233558697e-05,
      "loss": 1.4951,
      "step": 2406
    },
    {
      "epoch": 0.18500307314074985,
      "grad_norm": 3.3914384841918945,
      "learning_rate": 1.9259987707437e-05,
      "loss": 1.3401,
      "step": 2408
    },
    {
      "epoch": 0.18515673017824216,
      "grad_norm": 4.250161647796631,
      "learning_rate": 1.9259373079287034e-05,
      "loss": 1.6384,
      "step": 2410
    },
    {
      "epoch": 0.18531038721573448,
      "grad_norm": 4.362060070037842,
      "learning_rate": 1.9258758451137064e-05,
      "loss": 1.5895,
      "step": 2412
    },
    {
      "epoch": 0.1854640442532268,
      "grad_norm": 3.7243545055389404,
      "learning_rate": 1.9258143822987093e-05,
      "loss": 1.4715,
      "step": 2414
    },
    {
      "epoch": 0.1856177012907191,
      "grad_norm": 4.146886825561523,
      "learning_rate": 1.9257529194837127e-05,
      "loss": 1.5387,
      "step": 2416
    },
    {
      "epoch": 0.18577135832821143,
      "grad_norm": 3.769721746444702,
      "learning_rate": 1.9256914566687156e-05,
      "loss": 1.5978,
      "step": 2418
    },
    {
      "epoch": 0.18592501536570374,
      "grad_norm": 4.210140228271484,
      "learning_rate": 1.9256299938537186e-05,
      "loss": 1.4916,
      "step": 2420
    },
    {
      "epoch": 0.18607867240319606,
      "grad_norm": 3.8122193813323975,
      "learning_rate": 1.925568531038722e-05,
      "loss": 1.4786,
      "step": 2422
    },
    {
      "epoch": 0.18623232944068838,
      "grad_norm": 4.192070484161377,
      "learning_rate": 1.925507068223725e-05,
      "loss": 1.5604,
      "step": 2424
    },
    {
      "epoch": 0.1863859864781807,
      "grad_norm": 4.002556800842285,
      "learning_rate": 1.925445605408728e-05,
      "loss": 1.6205,
      "step": 2426
    },
    {
      "epoch": 0.186539643515673,
      "grad_norm": 4.130584716796875,
      "learning_rate": 1.9253841425937308e-05,
      "loss": 1.6588,
      "step": 2428
    },
    {
      "epoch": 0.18669330055316533,
      "grad_norm": 4.271806240081787,
      "learning_rate": 1.925322679778734e-05,
      "loss": 1.4771,
      "step": 2430
    },
    {
      "epoch": 0.18684695759065764,
      "grad_norm": 3.9257447719573975,
      "learning_rate": 1.925261216963737e-05,
      "loss": 1.4746,
      "step": 2432
    },
    {
      "epoch": 0.18700061462814996,
      "grad_norm": 3.5739989280700684,
      "learning_rate": 1.92519975414874e-05,
      "loss": 1.6682,
      "step": 2434
    },
    {
      "epoch": 0.18715427166564227,
      "grad_norm": 4.071542263031006,
      "learning_rate": 1.9251382913337434e-05,
      "loss": 1.7167,
      "step": 2436
    },
    {
      "epoch": 0.1873079287031346,
      "grad_norm": 3.6181557178497314,
      "learning_rate": 1.9250768285187463e-05,
      "loss": 1.4887,
      "step": 2438
    },
    {
      "epoch": 0.1874615857406269,
      "grad_norm": 3.9054434299468994,
      "learning_rate": 1.9250153657037493e-05,
      "loss": 1.5903,
      "step": 2440
    },
    {
      "epoch": 0.18761524277811925,
      "grad_norm": 3.884770631790161,
      "learning_rate": 1.9249539028887526e-05,
      "loss": 1.533,
      "step": 2442
    },
    {
      "epoch": 0.18776889981561157,
      "grad_norm": 4.007852554321289,
      "learning_rate": 1.9248924400737556e-05,
      "loss": 1.4776,
      "step": 2444
    },
    {
      "epoch": 0.18792255685310388,
      "grad_norm": 4.36082649230957,
      "learning_rate": 1.9248309772587585e-05,
      "loss": 1.6813,
      "step": 2446
    },
    {
      "epoch": 0.1880762138905962,
      "grad_norm": 3.7035789489746094,
      "learning_rate": 1.924769514443762e-05,
      "loss": 1.5808,
      "step": 2448
    },
    {
      "epoch": 0.18822987092808852,
      "grad_norm": 4.472562789916992,
      "learning_rate": 1.9247080516287648e-05,
      "loss": 1.6157,
      "step": 2450
    },
    {
      "epoch": 0.18838352796558083,
      "grad_norm": 3.8329532146453857,
      "learning_rate": 1.9246465888137678e-05,
      "loss": 1.6105,
      "step": 2452
    },
    {
      "epoch": 0.18853718500307315,
      "grad_norm": 3.5638465881347656,
      "learning_rate": 1.9245851259987707e-05,
      "loss": 1.3973,
      "step": 2454
    },
    {
      "epoch": 0.18869084204056547,
      "grad_norm": 3.769814968109131,
      "learning_rate": 1.924523663183774e-05,
      "loss": 1.436,
      "step": 2456
    },
    {
      "epoch": 0.18884449907805778,
      "grad_norm": 3.7624809741973877,
      "learning_rate": 1.924462200368777e-05,
      "loss": 1.6093,
      "step": 2458
    },
    {
      "epoch": 0.1889981561155501,
      "grad_norm": 3.934250593185425,
      "learning_rate": 1.92440073755378e-05,
      "loss": 1.5041,
      "step": 2460
    },
    {
      "epoch": 0.18915181315304241,
      "grad_norm": 4.421424865722656,
      "learning_rate": 1.9243392747387833e-05,
      "loss": 1.4708,
      "step": 2462
    },
    {
      "epoch": 0.18930547019053473,
      "grad_norm": 4.019033432006836,
      "learning_rate": 1.9242778119237863e-05,
      "loss": 1.5314,
      "step": 2464
    },
    {
      "epoch": 0.18945912722802705,
      "grad_norm": 4.172586917877197,
      "learning_rate": 1.9242163491087892e-05,
      "loss": 1.5259,
      "step": 2466
    },
    {
      "epoch": 0.18961278426551936,
      "grad_norm": 3.919337034225464,
      "learning_rate": 1.9241548862937925e-05,
      "loss": 1.6305,
      "step": 2468
    },
    {
      "epoch": 0.18976644130301168,
      "grad_norm": 4.666354656219482,
      "learning_rate": 1.9240934234787955e-05,
      "loss": 1.66,
      "step": 2470
    },
    {
      "epoch": 0.189920098340504,
      "grad_norm": 3.4628660678863525,
      "learning_rate": 1.9240319606637985e-05,
      "loss": 1.4733,
      "step": 2472
    },
    {
      "epoch": 0.1900737553779963,
      "grad_norm": 4.151192665100098,
      "learning_rate": 1.9239704978488018e-05,
      "loss": 1.6414,
      "step": 2474
    },
    {
      "epoch": 0.19022741241548863,
      "grad_norm": 3.8772740364074707,
      "learning_rate": 1.9239090350338048e-05,
      "loss": 1.6062,
      "step": 2476
    },
    {
      "epoch": 0.19038106945298094,
      "grad_norm": 4.541169166564941,
      "learning_rate": 1.9238475722188077e-05,
      "loss": 1.5332,
      "step": 2478
    },
    {
      "epoch": 0.19053472649047326,
      "grad_norm": 3.821932315826416,
      "learning_rate": 1.9237861094038107e-05,
      "loss": 1.4712,
      "step": 2480
    },
    {
      "epoch": 0.19068838352796558,
      "grad_norm": 4.174224853515625,
      "learning_rate": 1.923724646588814e-05,
      "loss": 1.5465,
      "step": 2482
    },
    {
      "epoch": 0.1908420405654579,
      "grad_norm": 3.4857685565948486,
      "learning_rate": 1.923663183773817e-05,
      "loss": 1.3857,
      "step": 2484
    },
    {
      "epoch": 0.1909956976029502,
      "grad_norm": 3.488931655883789,
      "learning_rate": 1.92360172095882e-05,
      "loss": 1.4848,
      "step": 2486
    },
    {
      "epoch": 0.19114935464044253,
      "grad_norm": 4.406091213226318,
      "learning_rate": 1.9235402581438232e-05,
      "loss": 1.5634,
      "step": 2488
    },
    {
      "epoch": 0.19130301167793484,
      "grad_norm": 4.208408832550049,
      "learning_rate": 1.9234787953288262e-05,
      "loss": 1.4935,
      "step": 2490
    },
    {
      "epoch": 0.19145666871542716,
      "grad_norm": 4.038366794586182,
      "learning_rate": 1.9234173325138292e-05,
      "loss": 1.435,
      "step": 2492
    },
    {
      "epoch": 0.19161032575291947,
      "grad_norm": 3.6954140663146973,
      "learning_rate": 1.9233558696988325e-05,
      "loss": 1.6045,
      "step": 2494
    },
    {
      "epoch": 0.1917639827904118,
      "grad_norm": 3.928567409515381,
      "learning_rate": 1.9232944068838355e-05,
      "loss": 1.6917,
      "step": 2496
    },
    {
      "epoch": 0.1919176398279041,
      "grad_norm": 3.8990769386291504,
      "learning_rate": 1.9232329440688384e-05,
      "loss": 1.6294,
      "step": 2498
    },
    {
      "epoch": 0.19207129686539642,
      "grad_norm": 4.251073360443115,
      "learning_rate": 1.9231714812538417e-05,
      "loss": 1.7429,
      "step": 2500
    },
    {
      "epoch": 0.19222495390288874,
      "grad_norm": 3.8998966217041016,
      "learning_rate": 1.9231100184388447e-05,
      "loss": 1.5395,
      "step": 2502
    },
    {
      "epoch": 0.19237861094038106,
      "grad_norm": 3.5932552814483643,
      "learning_rate": 1.9230485556238477e-05,
      "loss": 1.4188,
      "step": 2504
    },
    {
      "epoch": 0.1925322679778734,
      "grad_norm": 4.385290622711182,
      "learning_rate": 1.9229870928088506e-05,
      "loss": 1.5512,
      "step": 2506
    },
    {
      "epoch": 0.19268592501536572,
      "grad_norm": 4.489023685455322,
      "learning_rate": 1.922925629993854e-05,
      "loss": 1.5199,
      "step": 2508
    },
    {
      "epoch": 0.19283958205285803,
      "grad_norm": 3.464556932449341,
      "learning_rate": 1.922864167178857e-05,
      "loss": 1.6343,
      "step": 2510
    },
    {
      "epoch": 0.19299323909035035,
      "grad_norm": 3.967116594314575,
      "learning_rate": 1.92280270436386e-05,
      "loss": 1.5112,
      "step": 2512
    },
    {
      "epoch": 0.19314689612784267,
      "grad_norm": 3.6863479614257812,
      "learning_rate": 1.9227412415488632e-05,
      "loss": 1.4394,
      "step": 2514
    },
    {
      "epoch": 0.19330055316533498,
      "grad_norm": 3.8512773513793945,
      "learning_rate": 1.922679778733866e-05,
      "loss": 1.5762,
      "step": 2516
    },
    {
      "epoch": 0.1934542102028273,
      "grad_norm": 4.207437515258789,
      "learning_rate": 1.922618315918869e-05,
      "loss": 1.5081,
      "step": 2518
    },
    {
      "epoch": 0.19360786724031961,
      "grad_norm": 3.882289409637451,
      "learning_rate": 1.9225568531038724e-05,
      "loss": 1.542,
      "step": 2520
    },
    {
      "epoch": 0.19376152427781193,
      "grad_norm": 3.5873258113861084,
      "learning_rate": 1.9224953902888754e-05,
      "loss": 1.4955,
      "step": 2522
    },
    {
      "epoch": 0.19391518131530425,
      "grad_norm": 4.062012195587158,
      "learning_rate": 1.9224339274738784e-05,
      "loss": 1.6206,
      "step": 2524
    },
    {
      "epoch": 0.19406883835279656,
      "grad_norm": 4.009846210479736,
      "learning_rate": 1.9223724646588813e-05,
      "loss": 1.4736,
      "step": 2526
    },
    {
      "epoch": 0.19422249539028888,
      "grad_norm": 3.5200116634368896,
      "learning_rate": 1.9223110018438846e-05,
      "loss": 1.5089,
      "step": 2528
    },
    {
      "epoch": 0.1943761524277812,
      "grad_norm": 3.888718366622925,
      "learning_rate": 1.9222495390288876e-05,
      "loss": 1.5568,
      "step": 2530
    },
    {
      "epoch": 0.1945298094652735,
      "grad_norm": 4.041849136352539,
      "learning_rate": 1.9221880762138906e-05,
      "loss": 1.57,
      "step": 2532
    },
    {
      "epoch": 0.19468346650276583,
      "grad_norm": 3.235793352127075,
      "learning_rate": 1.922126613398894e-05,
      "loss": 1.562,
      "step": 2534
    },
    {
      "epoch": 0.19483712354025814,
      "grad_norm": 3.7582755088806152,
      "learning_rate": 1.922065150583897e-05,
      "loss": 1.6906,
      "step": 2536
    },
    {
      "epoch": 0.19499078057775046,
      "grad_norm": 4.138317108154297,
      "learning_rate": 1.9220036877688998e-05,
      "loss": 1.6045,
      "step": 2538
    },
    {
      "epoch": 0.19514443761524278,
      "grad_norm": 3.8657236099243164,
      "learning_rate": 1.921942224953903e-05,
      "loss": 1.6388,
      "step": 2540
    },
    {
      "epoch": 0.1952980946527351,
      "grad_norm": 4.225650310516357,
      "learning_rate": 1.921880762138906e-05,
      "loss": 1.5472,
      "step": 2542
    },
    {
      "epoch": 0.1954517516902274,
      "grad_norm": 4.000063896179199,
      "learning_rate": 1.921819299323909e-05,
      "loss": 1.5838,
      "step": 2544
    },
    {
      "epoch": 0.19560540872771973,
      "grad_norm": 3.2265849113464355,
      "learning_rate": 1.9217578365089124e-05,
      "loss": 1.401,
      "step": 2546
    },
    {
      "epoch": 0.19575906576521204,
      "grad_norm": 4.049646854400635,
      "learning_rate": 1.9216963736939153e-05,
      "loss": 1.6306,
      "step": 2548
    },
    {
      "epoch": 0.19591272280270436,
      "grad_norm": 4.136661052703857,
      "learning_rate": 1.9216349108789187e-05,
      "loss": 1.6809,
      "step": 2550
    },
    {
      "epoch": 0.19606637984019668,
      "grad_norm": 4.1102294921875,
      "learning_rate": 1.9215734480639213e-05,
      "loss": 1.5619,
      "step": 2552
    },
    {
      "epoch": 0.196220036877689,
      "grad_norm": 3.632099151611328,
      "learning_rate": 1.9215119852489246e-05,
      "loss": 1.4556,
      "step": 2554
    },
    {
      "epoch": 0.1963736939151813,
      "grad_norm": 3.7188732624053955,
      "learning_rate": 1.9214505224339276e-05,
      "loss": 1.6132,
      "step": 2556
    },
    {
      "epoch": 0.19652735095267362,
      "grad_norm": 3.4973299503326416,
      "learning_rate": 1.9213890596189305e-05,
      "loss": 1.5363,
      "step": 2558
    },
    {
      "epoch": 0.19668100799016594,
      "grad_norm": 4.495885848999023,
      "learning_rate": 1.921327596803934e-05,
      "loss": 1.4821,
      "step": 2560
    },
    {
      "epoch": 0.19683466502765826,
      "grad_norm": 4.349817752838135,
      "learning_rate": 1.9212661339889368e-05,
      "loss": 1.5153,
      "step": 2562
    },
    {
      "epoch": 0.19698832206515057,
      "grad_norm": 3.5700700283050537,
      "learning_rate": 1.9212046711739398e-05,
      "loss": 1.6255,
      "step": 2564
    },
    {
      "epoch": 0.1971419791026429,
      "grad_norm": 3.944878101348877,
      "learning_rate": 1.921143208358943e-05,
      "loss": 1.5505,
      "step": 2566
    },
    {
      "epoch": 0.1972956361401352,
      "grad_norm": 4.354259490966797,
      "learning_rate": 1.921081745543946e-05,
      "loss": 1.6005,
      "step": 2568
    },
    {
      "epoch": 0.19744929317762752,
      "grad_norm": 4.167532920837402,
      "learning_rate": 1.9210202827289494e-05,
      "loss": 1.6639,
      "step": 2570
    },
    {
      "epoch": 0.19760295021511987,
      "grad_norm": 4.1236958503723145,
      "learning_rate": 1.9209588199139523e-05,
      "loss": 1.6543,
      "step": 2572
    },
    {
      "epoch": 0.19775660725261218,
      "grad_norm": 4.3486328125,
      "learning_rate": 1.9208973570989553e-05,
      "loss": 1.6486,
      "step": 2574
    },
    {
      "epoch": 0.1979102642901045,
      "grad_norm": 4.074930191040039,
      "learning_rate": 1.9208358942839586e-05,
      "loss": 1.5789,
      "step": 2576
    },
    {
      "epoch": 0.19806392132759681,
      "grad_norm": 4.678352355957031,
      "learning_rate": 1.9207744314689612e-05,
      "loss": 1.5615,
      "step": 2578
    },
    {
      "epoch": 0.19821757836508913,
      "grad_norm": 3.689697504043579,
      "learning_rate": 1.9207129686539645e-05,
      "loss": 1.4933,
      "step": 2580
    },
    {
      "epoch": 0.19837123540258145,
      "grad_norm": 3.895163059234619,
      "learning_rate": 1.9206515058389675e-05,
      "loss": 1.5115,
      "step": 2582
    },
    {
      "epoch": 0.19852489244007376,
      "grad_norm": 4.155092239379883,
      "learning_rate": 1.9205900430239705e-05,
      "loss": 1.4538,
      "step": 2584
    },
    {
      "epoch": 0.19867854947756608,
      "grad_norm": 4.396788120269775,
      "learning_rate": 1.9205285802089738e-05,
      "loss": 1.5629,
      "step": 2586
    },
    {
      "epoch": 0.1988322065150584,
      "grad_norm": 4.8076934814453125,
      "learning_rate": 1.9204671173939767e-05,
      "loss": 1.4844,
      "step": 2588
    },
    {
      "epoch": 0.1989858635525507,
      "grad_norm": 3.6861722469329834,
      "learning_rate": 1.92040565457898e-05,
      "loss": 1.501,
      "step": 2590
    },
    {
      "epoch": 0.19913952059004303,
      "grad_norm": 4.285679817199707,
      "learning_rate": 1.920344191763983e-05,
      "loss": 1.6599,
      "step": 2592
    },
    {
      "epoch": 0.19929317762753535,
      "grad_norm": 4.052687644958496,
      "learning_rate": 1.920282728948986e-05,
      "loss": 1.4244,
      "step": 2594
    },
    {
      "epoch": 0.19944683466502766,
      "grad_norm": 3.6357898712158203,
      "learning_rate": 1.9202212661339893e-05,
      "loss": 1.5088,
      "step": 2596
    },
    {
      "epoch": 0.19960049170251998,
      "grad_norm": 3.789736270904541,
      "learning_rate": 1.9201598033189923e-05,
      "loss": 1.5779,
      "step": 2598
    },
    {
      "epoch": 0.1997541487400123,
      "grad_norm": 4.087263107299805,
      "learning_rate": 1.9200983405039952e-05,
      "loss": 1.6094,
      "step": 2600
    },
    {
      "epoch": 0.1999078057775046,
      "grad_norm": 4.693444728851318,
      "learning_rate": 1.9200368776889985e-05,
      "loss": 1.3526,
      "step": 2602
    },
    {
      "epoch": 0.20006146281499693,
      "grad_norm": 3.850968837738037,
      "learning_rate": 1.919975414874001e-05,
      "loss": 1.5124,
      "step": 2604
    },
    {
      "epoch": 0.20021511985248924,
      "grad_norm": 3.671071767807007,
      "learning_rate": 1.9199139520590045e-05,
      "loss": 1.5842,
      "step": 2606
    },
    {
      "epoch": 0.20036877688998156,
      "grad_norm": 3.6253607273101807,
      "learning_rate": 1.9198524892440074e-05,
      "loss": 1.5148,
      "step": 2608
    },
    {
      "epoch": 0.20052243392747388,
      "grad_norm": 4.358476638793945,
      "learning_rate": 1.9197910264290104e-05,
      "loss": 1.531,
      "step": 2610
    },
    {
      "epoch": 0.2006760909649662,
      "grad_norm": 3.9361257553100586,
      "learning_rate": 1.9197295636140137e-05,
      "loss": 1.5476,
      "step": 2612
    },
    {
      "epoch": 0.2008297480024585,
      "grad_norm": 4.064403533935547,
      "learning_rate": 1.9196681007990167e-05,
      "loss": 1.5514,
      "step": 2614
    },
    {
      "epoch": 0.20098340503995082,
      "grad_norm": 4.13236665725708,
      "learning_rate": 1.91960663798402e-05,
      "loss": 1.5569,
      "step": 2616
    },
    {
      "epoch": 0.20113706207744314,
      "grad_norm": 3.7941882610321045,
      "learning_rate": 1.919545175169023e-05,
      "loss": 1.511,
      "step": 2618
    },
    {
      "epoch": 0.20129071911493546,
      "grad_norm": 3.5626373291015625,
      "learning_rate": 1.919483712354026e-05,
      "loss": 1.4677,
      "step": 2620
    },
    {
      "epoch": 0.20144437615242777,
      "grad_norm": 4.068457126617432,
      "learning_rate": 1.9194222495390292e-05,
      "loss": 1.5312,
      "step": 2622
    },
    {
      "epoch": 0.2015980331899201,
      "grad_norm": 4.036095142364502,
      "learning_rate": 1.919360786724032e-05,
      "loss": 1.6681,
      "step": 2624
    },
    {
      "epoch": 0.2017516902274124,
      "grad_norm": 3.8620471954345703,
      "learning_rate": 1.9192993239090352e-05,
      "loss": 1.5411,
      "step": 2626
    },
    {
      "epoch": 0.20190534726490472,
      "grad_norm": 3.6075820922851562,
      "learning_rate": 1.919237861094038e-05,
      "loss": 1.5125,
      "step": 2628
    },
    {
      "epoch": 0.20205900430239704,
      "grad_norm": 3.6271731853485107,
      "learning_rate": 1.919176398279041e-05,
      "loss": 1.5795,
      "step": 2630
    },
    {
      "epoch": 0.20221266133988935,
      "grad_norm": 4.1661810874938965,
      "learning_rate": 1.9191149354640444e-05,
      "loss": 1.5523,
      "step": 2632
    },
    {
      "epoch": 0.20236631837738167,
      "grad_norm": 4.115680694580078,
      "learning_rate": 1.9190534726490474e-05,
      "loss": 1.5146,
      "step": 2634
    },
    {
      "epoch": 0.20251997541487401,
      "grad_norm": 3.4160807132720947,
      "learning_rate": 1.9189920098340507e-05,
      "loss": 1.5035,
      "step": 2636
    },
    {
      "epoch": 0.20267363245236633,
      "grad_norm": 4.1160969734191895,
      "learning_rate": 1.9189305470190537e-05,
      "loss": 1.481,
      "step": 2638
    },
    {
      "epoch": 0.20282728948985865,
      "grad_norm": 4.941380500793457,
      "learning_rate": 1.9188690842040566e-05,
      "loss": 1.7115,
      "step": 2640
    },
    {
      "epoch": 0.20298094652735096,
      "grad_norm": 3.743597984313965,
      "learning_rate": 1.91880762138906e-05,
      "loss": 1.5442,
      "step": 2642
    },
    {
      "epoch": 0.20313460356484328,
      "grad_norm": 4.364147186279297,
      "learning_rate": 1.918746158574063e-05,
      "loss": 1.3992,
      "step": 2644
    },
    {
      "epoch": 0.2032882606023356,
      "grad_norm": 3.69368052482605,
      "learning_rate": 1.918684695759066e-05,
      "loss": 1.4778,
      "step": 2646
    },
    {
      "epoch": 0.2034419176398279,
      "grad_norm": 4.028282642364502,
      "learning_rate": 1.9186232329440692e-05,
      "loss": 1.6028,
      "step": 2648
    },
    {
      "epoch": 0.20359557467732023,
      "grad_norm": 3.767077922821045,
      "learning_rate": 1.9185617701290718e-05,
      "loss": 1.4907,
      "step": 2650
    },
    {
      "epoch": 0.20374923171481255,
      "grad_norm": 3.511185646057129,
      "learning_rate": 1.918500307314075e-05,
      "loss": 1.6204,
      "step": 2652
    },
    {
      "epoch": 0.20390288875230486,
      "grad_norm": 3.6800639629364014,
      "learning_rate": 1.918438844499078e-05,
      "loss": 1.5762,
      "step": 2654
    },
    {
      "epoch": 0.20405654578979718,
      "grad_norm": 3.3730273246765137,
      "learning_rate": 1.9183773816840814e-05,
      "loss": 1.4401,
      "step": 2656
    },
    {
      "epoch": 0.2042102028272895,
      "grad_norm": 4.639964580535889,
      "learning_rate": 1.9183159188690844e-05,
      "loss": 1.5344,
      "step": 2658
    },
    {
      "epoch": 0.2043638598647818,
      "grad_norm": 3.624997138977051,
      "learning_rate": 1.9182544560540873e-05,
      "loss": 1.5386,
      "step": 2660
    },
    {
      "epoch": 0.20451751690227413,
      "grad_norm": 4.5834503173828125,
      "learning_rate": 1.9181929932390906e-05,
      "loss": 1.6546,
      "step": 2662
    },
    {
      "epoch": 0.20467117393976644,
      "grad_norm": 3.5726184844970703,
      "learning_rate": 1.9181315304240936e-05,
      "loss": 1.4267,
      "step": 2664
    },
    {
      "epoch": 0.20482483097725876,
      "grad_norm": 3.3437399864196777,
      "learning_rate": 1.9180700676090966e-05,
      "loss": 1.6013,
      "step": 2666
    },
    {
      "epoch": 0.20497848801475108,
      "grad_norm": 3.5842888355255127,
      "learning_rate": 1.9180086047941e-05,
      "loss": 1.4479,
      "step": 2668
    },
    {
      "epoch": 0.2051321450522434,
      "grad_norm": 4.096655368804932,
      "learning_rate": 1.917947141979103e-05,
      "loss": 1.4553,
      "step": 2670
    },
    {
      "epoch": 0.2052858020897357,
      "grad_norm": 4.347322940826416,
      "learning_rate": 1.9178856791641058e-05,
      "loss": 1.3843,
      "step": 2672
    },
    {
      "epoch": 0.20543945912722802,
      "grad_norm": 3.9682047367095947,
      "learning_rate": 1.917824216349109e-05,
      "loss": 1.6752,
      "step": 2674
    },
    {
      "epoch": 0.20559311616472034,
      "grad_norm": 4.238104343414307,
      "learning_rate": 1.917762753534112e-05,
      "loss": 1.5107,
      "step": 2676
    },
    {
      "epoch": 0.20574677320221266,
      "grad_norm": 3.6419684886932373,
      "learning_rate": 1.917701290719115e-05,
      "loss": 1.6268,
      "step": 2678
    },
    {
      "epoch": 0.20590043023970497,
      "grad_norm": 4.493232727050781,
      "learning_rate": 1.917639827904118e-05,
      "loss": 1.5479,
      "step": 2680
    },
    {
      "epoch": 0.2060540872771973,
      "grad_norm": 3.5530130863189697,
      "learning_rate": 1.9175783650891213e-05,
      "loss": 1.4865,
      "step": 2682
    },
    {
      "epoch": 0.2062077443146896,
      "grad_norm": 3.930048942565918,
      "learning_rate": 1.9175169022741243e-05,
      "loss": 1.4627,
      "step": 2684
    },
    {
      "epoch": 0.20636140135218192,
      "grad_norm": 4.193355560302734,
      "learning_rate": 1.9174554394591273e-05,
      "loss": 1.5918,
      "step": 2686
    },
    {
      "epoch": 0.20651505838967424,
      "grad_norm": 3.4783825874328613,
      "learning_rate": 1.9173939766441306e-05,
      "loss": 1.441,
      "step": 2688
    },
    {
      "epoch": 0.20666871542716655,
      "grad_norm": 3.6716883182525635,
      "learning_rate": 1.9173325138291336e-05,
      "loss": 1.5617,
      "step": 2690
    },
    {
      "epoch": 0.20682237246465887,
      "grad_norm": 3.806039810180664,
      "learning_rate": 1.9172710510141365e-05,
      "loss": 1.5566,
      "step": 2692
    },
    {
      "epoch": 0.2069760295021512,
      "grad_norm": 3.7331125736236572,
      "learning_rate": 1.9172095881991398e-05,
      "loss": 1.4934,
      "step": 2694
    },
    {
      "epoch": 0.2071296865396435,
      "grad_norm": 4.192966938018799,
      "learning_rate": 1.9171481253841428e-05,
      "loss": 1.5679,
      "step": 2696
    },
    {
      "epoch": 0.20728334357713582,
      "grad_norm": 3.590785503387451,
      "learning_rate": 1.9170866625691458e-05,
      "loss": 1.4876,
      "step": 2698
    },
    {
      "epoch": 0.20743700061462814,
      "grad_norm": 4.1026458740234375,
      "learning_rate": 1.917025199754149e-05,
      "loss": 1.5708,
      "step": 2700
    },
    {
      "epoch": 0.20759065765212048,
      "grad_norm": 3.8768205642700195,
      "learning_rate": 1.916963736939152e-05,
      "loss": 1.4715,
      "step": 2702
    },
    {
      "epoch": 0.2077443146896128,
      "grad_norm": 3.648543357849121,
      "learning_rate": 1.916902274124155e-05,
      "loss": 1.6127,
      "step": 2704
    },
    {
      "epoch": 0.2078979717271051,
      "grad_norm": 3.6755056381225586,
      "learning_rate": 1.916840811309158e-05,
      "loss": 1.6711,
      "step": 2706
    },
    {
      "epoch": 0.20805162876459743,
      "grad_norm": 3.4196550846099854,
      "learning_rate": 1.9167793484941613e-05,
      "loss": 1.4866,
      "step": 2708
    },
    {
      "epoch": 0.20820528580208975,
      "grad_norm": 3.9521472454071045,
      "learning_rate": 1.9167178856791643e-05,
      "loss": 1.5474,
      "step": 2710
    },
    {
      "epoch": 0.20835894283958206,
      "grad_norm": 4.077064037322998,
      "learning_rate": 1.9166564228641672e-05,
      "loss": 1.6824,
      "step": 2712
    },
    {
      "epoch": 0.20851259987707438,
      "grad_norm": 4.164952278137207,
      "learning_rate": 1.9165949600491705e-05,
      "loss": 1.4411,
      "step": 2714
    },
    {
      "epoch": 0.2086662569145667,
      "grad_norm": 3.788529396057129,
      "learning_rate": 1.9165334972341735e-05,
      "loss": 1.6062,
      "step": 2716
    },
    {
      "epoch": 0.208819913952059,
      "grad_norm": 4.074838638305664,
      "learning_rate": 1.9164720344191765e-05,
      "loss": 1.567,
      "step": 2718
    },
    {
      "epoch": 0.20897357098955133,
      "grad_norm": 3.398486852645874,
      "learning_rate": 1.9164105716041798e-05,
      "loss": 1.6186,
      "step": 2720
    },
    {
      "epoch": 0.20912722802704364,
      "grad_norm": 3.534534215927124,
      "learning_rate": 1.9163491087891827e-05,
      "loss": 1.4404,
      "step": 2722
    },
    {
      "epoch": 0.20928088506453596,
      "grad_norm": 3.3846664428710938,
      "learning_rate": 1.9162876459741857e-05,
      "loss": 1.4741,
      "step": 2724
    },
    {
      "epoch": 0.20943454210202828,
      "grad_norm": 3.8569769859313965,
      "learning_rate": 1.916226183159189e-05,
      "loss": 1.4941,
      "step": 2726
    },
    {
      "epoch": 0.2095881991395206,
      "grad_norm": 3.6158618927001953,
      "learning_rate": 1.916164720344192e-05,
      "loss": 1.5539,
      "step": 2728
    },
    {
      "epoch": 0.2097418561770129,
      "grad_norm": 4.233087062835693,
      "learning_rate": 1.916103257529195e-05,
      "loss": 1.5474,
      "step": 2730
    },
    {
      "epoch": 0.20989551321450522,
      "grad_norm": 3.693319797515869,
      "learning_rate": 1.916041794714198e-05,
      "loss": 1.5927,
      "step": 2732
    },
    {
      "epoch": 0.21004917025199754,
      "grad_norm": 3.5169527530670166,
      "learning_rate": 1.9159803318992012e-05,
      "loss": 1.3885,
      "step": 2734
    },
    {
      "epoch": 0.21020282728948986,
      "grad_norm": 3.507129430770874,
      "learning_rate": 1.9159188690842042e-05,
      "loss": 1.4943,
      "step": 2736
    },
    {
      "epoch": 0.21035648432698217,
      "grad_norm": 4.948350429534912,
      "learning_rate": 1.915857406269207e-05,
      "loss": 1.505,
      "step": 2738
    },
    {
      "epoch": 0.2105101413644745,
      "grad_norm": 3.834376573562622,
      "learning_rate": 1.9157959434542105e-05,
      "loss": 1.6174,
      "step": 2740
    },
    {
      "epoch": 0.2106637984019668,
      "grad_norm": 4.351836204528809,
      "learning_rate": 1.9157344806392134e-05,
      "loss": 1.5882,
      "step": 2742
    },
    {
      "epoch": 0.21081745543945912,
      "grad_norm": 3.8399126529693604,
      "learning_rate": 1.9156730178242164e-05,
      "loss": 1.5669,
      "step": 2744
    },
    {
      "epoch": 0.21097111247695144,
      "grad_norm": 4.092167377471924,
      "learning_rate": 1.9156115550092197e-05,
      "loss": 1.5737,
      "step": 2746
    },
    {
      "epoch": 0.21112476951444376,
      "grad_norm": 3.5932557582855225,
      "learning_rate": 1.9155500921942227e-05,
      "loss": 1.5082,
      "step": 2748
    },
    {
      "epoch": 0.21127842655193607,
      "grad_norm": 4.129446506500244,
      "learning_rate": 1.9154886293792257e-05,
      "loss": 1.6286,
      "step": 2750
    },
    {
      "epoch": 0.2114320835894284,
      "grad_norm": 3.489421844482422,
      "learning_rate": 1.9154271665642286e-05,
      "loss": 1.5005,
      "step": 2752
    },
    {
      "epoch": 0.2115857406269207,
      "grad_norm": 3.7528555393218994,
      "learning_rate": 1.915365703749232e-05,
      "loss": 1.5174,
      "step": 2754
    },
    {
      "epoch": 0.21173939766441302,
      "grad_norm": 3.6098313331604004,
      "learning_rate": 1.915304240934235e-05,
      "loss": 1.4932,
      "step": 2756
    },
    {
      "epoch": 0.21189305470190534,
      "grad_norm": 3.8902368545532227,
      "learning_rate": 1.915242778119238e-05,
      "loss": 1.4891,
      "step": 2758
    },
    {
      "epoch": 0.21204671173939765,
      "grad_norm": 4.011685371398926,
      "learning_rate": 1.9151813153042412e-05,
      "loss": 1.5252,
      "step": 2760
    },
    {
      "epoch": 0.21220036877688997,
      "grad_norm": 3.6917941570281982,
      "learning_rate": 1.915119852489244e-05,
      "loss": 1.5622,
      "step": 2762
    },
    {
      "epoch": 0.21235402581438229,
      "grad_norm": 3.5468204021453857,
      "learning_rate": 1.915058389674247e-05,
      "loss": 1.3541,
      "step": 2764
    },
    {
      "epoch": 0.21250768285187463,
      "grad_norm": 3.7897398471832275,
      "learning_rate": 1.9149969268592504e-05,
      "loss": 1.7065,
      "step": 2766
    },
    {
      "epoch": 0.21266133988936695,
      "grad_norm": 3.539816379547119,
      "learning_rate": 1.9149354640442534e-05,
      "loss": 1.4987,
      "step": 2768
    },
    {
      "epoch": 0.21281499692685926,
      "grad_norm": 3.866915702819824,
      "learning_rate": 1.9148740012292564e-05,
      "loss": 1.6211,
      "step": 2770
    },
    {
      "epoch": 0.21296865396435158,
      "grad_norm": 4.1955180168151855,
      "learning_rate": 1.9148125384142597e-05,
      "loss": 1.448,
      "step": 2772
    },
    {
      "epoch": 0.2131223110018439,
      "grad_norm": 3.6780476570129395,
      "learning_rate": 1.9147510755992626e-05,
      "loss": 1.5676,
      "step": 2774
    },
    {
      "epoch": 0.2132759680393362,
      "grad_norm": 4.097123622894287,
      "learning_rate": 1.9146896127842656e-05,
      "loss": 1.5146,
      "step": 2776
    },
    {
      "epoch": 0.21342962507682853,
      "grad_norm": 3.8207781314849854,
      "learning_rate": 1.9146281499692686e-05,
      "loss": 1.4983,
      "step": 2778
    },
    {
      "epoch": 0.21358328211432084,
      "grad_norm": 3.585047960281372,
      "learning_rate": 1.914566687154272e-05,
      "loss": 1.4271,
      "step": 2780
    },
    {
      "epoch": 0.21373693915181316,
      "grad_norm": 3.4344520568847656,
      "learning_rate": 1.914505224339275e-05,
      "loss": 1.5673,
      "step": 2782
    },
    {
      "epoch": 0.21389059618930548,
      "grad_norm": 3.5881597995758057,
      "learning_rate": 1.9144437615242778e-05,
      "loss": 1.4111,
      "step": 2784
    },
    {
      "epoch": 0.2140442532267978,
      "grad_norm": 3.65913724899292,
      "learning_rate": 1.914382298709281e-05,
      "loss": 1.5894,
      "step": 2786
    },
    {
      "epoch": 0.2141979102642901,
      "grad_norm": 3.761591911315918,
      "learning_rate": 1.914320835894284e-05,
      "loss": 1.4502,
      "step": 2788
    },
    {
      "epoch": 0.21435156730178243,
      "grad_norm": 3.943924903869629,
      "learning_rate": 1.914259373079287e-05,
      "loss": 1.4754,
      "step": 2790
    },
    {
      "epoch": 0.21450522433927474,
      "grad_norm": 4.111385345458984,
      "learning_rate": 1.9141979102642904e-05,
      "loss": 1.5946,
      "step": 2792
    },
    {
      "epoch": 0.21465888137676706,
      "grad_norm": 3.8199594020843506,
      "learning_rate": 1.9141364474492933e-05,
      "loss": 1.4817,
      "step": 2794
    },
    {
      "epoch": 0.21481253841425937,
      "grad_norm": 3.667856454849243,
      "learning_rate": 1.9140749846342963e-05,
      "loss": 1.5357,
      "step": 2796
    },
    {
      "epoch": 0.2149661954517517,
      "grad_norm": 3.980133295059204,
      "learning_rate": 1.9140135218192996e-05,
      "loss": 1.6328,
      "step": 2798
    },
    {
      "epoch": 0.215119852489244,
      "grad_norm": 4.220106601715088,
      "learning_rate": 1.9139520590043026e-05,
      "loss": 1.5825,
      "step": 2800
    },
    {
      "epoch": 0.21527350952673632,
      "grad_norm": 3.8425889015197754,
      "learning_rate": 1.913890596189306e-05,
      "loss": 1.5565,
      "step": 2802
    },
    {
      "epoch": 0.21542716656422864,
      "grad_norm": 3.5814743041992188,
      "learning_rate": 1.9138291333743085e-05,
      "loss": 1.5044,
      "step": 2804
    },
    {
      "epoch": 0.21558082360172096,
      "grad_norm": 4.015439510345459,
      "learning_rate": 1.9137676705593118e-05,
      "loss": 1.6277,
      "step": 2806
    },
    {
      "epoch": 0.21573448063921327,
      "grad_norm": 4.553494930267334,
      "learning_rate": 1.9137062077443148e-05,
      "loss": 1.4374,
      "step": 2808
    },
    {
      "epoch": 0.2158881376767056,
      "grad_norm": 3.7021353244781494,
      "learning_rate": 1.9136447449293178e-05,
      "loss": 1.5173,
      "step": 2810
    },
    {
      "epoch": 0.2160417947141979,
      "grad_norm": 4.46372652053833,
      "learning_rate": 1.913583282114321e-05,
      "loss": 1.7864,
      "step": 2812
    },
    {
      "epoch": 0.21619545175169022,
      "grad_norm": 3.551791191101074,
      "learning_rate": 1.913521819299324e-05,
      "loss": 1.3458,
      "step": 2814
    },
    {
      "epoch": 0.21634910878918254,
      "grad_norm": 3.224550724029541,
      "learning_rate": 1.913460356484327e-05,
      "loss": 1.4272,
      "step": 2816
    },
    {
      "epoch": 0.21650276582667485,
      "grad_norm": 4.008483409881592,
      "learning_rate": 1.9133988936693303e-05,
      "loss": 1.5502,
      "step": 2818
    },
    {
      "epoch": 0.21665642286416717,
      "grad_norm": 3.782654047012329,
      "learning_rate": 1.9133374308543333e-05,
      "loss": 1.413,
      "step": 2820
    },
    {
      "epoch": 0.21681007990165949,
      "grad_norm": 3.2302143573760986,
      "learning_rate": 1.9132759680393366e-05,
      "loss": 1.4368,
      "step": 2822
    },
    {
      "epoch": 0.2169637369391518,
      "grad_norm": 3.4619314670562744,
      "learning_rate": 1.9132145052243395e-05,
      "loss": 1.4042,
      "step": 2824
    },
    {
      "epoch": 0.21711739397664412,
      "grad_norm": 4.229294300079346,
      "learning_rate": 1.9131530424093425e-05,
      "loss": 1.5187,
      "step": 2826
    },
    {
      "epoch": 0.21727105101413643,
      "grad_norm": 3.9806225299835205,
      "learning_rate": 1.9130915795943458e-05,
      "loss": 1.4521,
      "step": 2828
    },
    {
      "epoch": 0.21742470805162875,
      "grad_norm": 3.7580087184906006,
      "learning_rate": 1.9130301167793485e-05,
      "loss": 1.5982,
      "step": 2830
    },
    {
      "epoch": 0.2175783650891211,
      "grad_norm": 3.9534852504730225,
      "learning_rate": 1.9129686539643518e-05,
      "loss": 1.6518,
      "step": 2832
    },
    {
      "epoch": 0.2177320221266134,
      "grad_norm": 3.770359992980957,
      "learning_rate": 1.9129071911493547e-05,
      "loss": 1.3771,
      "step": 2834
    },
    {
      "epoch": 0.21788567916410573,
      "grad_norm": 4.148795127868652,
      "learning_rate": 1.9128457283343577e-05,
      "loss": 1.6236,
      "step": 2836
    },
    {
      "epoch": 0.21803933620159804,
      "grad_norm": 4.049807548522949,
      "learning_rate": 1.912784265519361e-05,
      "loss": 1.6702,
      "step": 2838
    },
    {
      "epoch": 0.21819299323909036,
      "grad_norm": 3.7045743465423584,
      "learning_rate": 1.912722802704364e-05,
      "loss": 1.54,
      "step": 2840
    },
    {
      "epoch": 0.21834665027658268,
      "grad_norm": 3.9694302082061768,
      "learning_rate": 1.9126613398893673e-05,
      "loss": 1.6067,
      "step": 2842
    },
    {
      "epoch": 0.218500307314075,
      "grad_norm": 3.4496958255767822,
      "learning_rate": 1.9125998770743702e-05,
      "loss": 1.5128,
      "step": 2844
    },
    {
      "epoch": 0.2186539643515673,
      "grad_norm": 3.6536827087402344,
      "learning_rate": 1.9125384142593732e-05,
      "loss": 1.5447,
      "step": 2846
    },
    {
      "epoch": 0.21880762138905963,
      "grad_norm": 3.9362754821777344,
      "learning_rate": 1.9124769514443765e-05,
      "loss": 1.4012,
      "step": 2848
    },
    {
      "epoch": 0.21896127842655194,
      "grad_norm": 3.2885048389434814,
      "learning_rate": 1.912415488629379e-05,
      "loss": 1.4489,
      "step": 2850
    },
    {
      "epoch": 0.21911493546404426,
      "grad_norm": 3.519296884536743,
      "learning_rate": 1.9123540258143825e-05,
      "loss": 1.4015,
      "step": 2852
    },
    {
      "epoch": 0.21926859250153657,
      "grad_norm": 4.157144546508789,
      "learning_rate": 1.9122925629993854e-05,
      "loss": 1.5355,
      "step": 2854
    },
    {
      "epoch": 0.2194222495390289,
      "grad_norm": 3.6727182865142822,
      "learning_rate": 1.9122311001843884e-05,
      "loss": 1.5225,
      "step": 2856
    },
    {
      "epoch": 0.2195759065765212,
      "grad_norm": 3.801722288131714,
      "learning_rate": 1.9121696373693917e-05,
      "loss": 1.4482,
      "step": 2858
    },
    {
      "epoch": 0.21972956361401352,
      "grad_norm": 4.15856409072876,
      "learning_rate": 1.9121081745543947e-05,
      "loss": 1.5494,
      "step": 2860
    },
    {
      "epoch": 0.21988322065150584,
      "grad_norm": 4.247369289398193,
      "learning_rate": 1.912046711739398e-05,
      "loss": 1.4578,
      "step": 2862
    },
    {
      "epoch": 0.22003687768899816,
      "grad_norm": 4.413219928741455,
      "learning_rate": 1.911985248924401e-05,
      "loss": 1.5631,
      "step": 2864
    },
    {
      "epoch": 0.22019053472649047,
      "grad_norm": 3.894334077835083,
      "learning_rate": 1.911923786109404e-05,
      "loss": 1.5072,
      "step": 2866
    },
    {
      "epoch": 0.2203441917639828,
      "grad_norm": 3.5060431957244873,
      "learning_rate": 1.9118623232944072e-05,
      "loss": 1.5318,
      "step": 2868
    },
    {
      "epoch": 0.2204978488014751,
      "grad_norm": 3.312140464782715,
      "learning_rate": 1.9118008604794102e-05,
      "loss": 1.5173,
      "step": 2870
    },
    {
      "epoch": 0.22065150583896742,
      "grad_norm": 4.122204780578613,
      "learning_rate": 1.911739397664413e-05,
      "loss": 1.5222,
      "step": 2872
    },
    {
      "epoch": 0.22080516287645974,
      "grad_norm": 3.994349956512451,
      "learning_rate": 1.9116779348494165e-05,
      "loss": 1.6783,
      "step": 2874
    },
    {
      "epoch": 0.22095881991395205,
      "grad_norm": 3.831480026245117,
      "learning_rate": 1.911616472034419e-05,
      "loss": 1.4112,
      "step": 2876
    },
    {
      "epoch": 0.22111247695144437,
      "grad_norm": 3.1337056159973145,
      "learning_rate": 1.9115550092194224e-05,
      "loss": 1.4297,
      "step": 2878
    },
    {
      "epoch": 0.2212661339889367,
      "grad_norm": 3.5576846599578857,
      "learning_rate": 1.9114935464044254e-05,
      "loss": 1.5911,
      "step": 2880
    },
    {
      "epoch": 0.221419791026429,
      "grad_norm": 4.290283679962158,
      "learning_rate": 1.9114320835894283e-05,
      "loss": 1.381,
      "step": 2882
    },
    {
      "epoch": 0.22157344806392132,
      "grad_norm": 3.7715091705322266,
      "learning_rate": 1.9113706207744316e-05,
      "loss": 1.4368,
      "step": 2884
    },
    {
      "epoch": 0.22172710510141364,
      "grad_norm": 3.3459980487823486,
      "learning_rate": 1.9113091579594346e-05,
      "loss": 1.3408,
      "step": 2886
    },
    {
      "epoch": 0.22188076213890595,
      "grad_norm": 3.626512050628662,
      "learning_rate": 1.911247695144438e-05,
      "loss": 1.5328,
      "step": 2888
    },
    {
      "epoch": 0.22203441917639827,
      "grad_norm": 4.11099100112915,
      "learning_rate": 1.911186232329441e-05,
      "loss": 1.3911,
      "step": 2890
    },
    {
      "epoch": 0.22218807621389058,
      "grad_norm": 3.386157751083374,
      "learning_rate": 1.911124769514444e-05,
      "loss": 1.4783,
      "step": 2892
    },
    {
      "epoch": 0.2223417332513829,
      "grad_norm": 3.7897472381591797,
      "learning_rate": 1.911063306699447e-05,
      "loss": 1.723,
      "step": 2894
    },
    {
      "epoch": 0.22249539028887522,
      "grad_norm": 4.359883785247803,
      "learning_rate": 1.91100184388445e-05,
      "loss": 1.515,
      "step": 2896
    },
    {
      "epoch": 0.22264904732636756,
      "grad_norm": 3.69624924659729,
      "learning_rate": 1.910940381069453e-05,
      "loss": 1.4789,
      "step": 2898
    },
    {
      "epoch": 0.22280270436385988,
      "grad_norm": 3.6302242279052734,
      "learning_rate": 1.9108789182544564e-05,
      "loss": 1.4188,
      "step": 2900
    },
    {
      "epoch": 0.2229563614013522,
      "grad_norm": 3.765510082244873,
      "learning_rate": 1.910817455439459e-05,
      "loss": 1.5475,
      "step": 2902
    },
    {
      "epoch": 0.2231100184388445,
      "grad_norm": 3.6644978523254395,
      "learning_rate": 1.9107559926244623e-05,
      "loss": 1.3725,
      "step": 2904
    },
    {
      "epoch": 0.22326367547633683,
      "grad_norm": 3.9660632610321045,
      "learning_rate": 1.9106945298094653e-05,
      "loss": 1.6626,
      "step": 2906
    },
    {
      "epoch": 0.22341733251382914,
      "grad_norm": 3.578211545944214,
      "learning_rate": 1.9106330669944686e-05,
      "loss": 1.7112,
      "step": 2908
    },
    {
      "epoch": 0.22357098955132146,
      "grad_norm": 3.61378812789917,
      "learning_rate": 1.9105716041794716e-05,
      "loss": 1.4498,
      "step": 2910
    },
    {
      "epoch": 0.22372464658881377,
      "grad_norm": 3.705375909805298,
      "learning_rate": 1.9105101413644746e-05,
      "loss": 1.5779,
      "step": 2912
    },
    {
      "epoch": 0.2238783036263061,
      "grad_norm": 3.6211791038513184,
      "learning_rate": 1.910448678549478e-05,
      "loss": 1.5316,
      "step": 2914
    },
    {
      "epoch": 0.2240319606637984,
      "grad_norm": 3.84789776802063,
      "learning_rate": 1.910387215734481e-05,
      "loss": 1.4939,
      "step": 2916
    },
    {
      "epoch": 0.22418561770129072,
      "grad_norm": 3.532919406890869,
      "learning_rate": 1.9103257529194838e-05,
      "loss": 1.4652,
      "step": 2918
    },
    {
      "epoch": 0.22433927473878304,
      "grad_norm": 4.090033531188965,
      "learning_rate": 1.910264290104487e-05,
      "loss": 1.6013,
      "step": 2920
    },
    {
      "epoch": 0.22449293177627536,
      "grad_norm": 3.964073896408081,
      "learning_rate": 1.91020282728949e-05,
      "loss": 1.5672,
      "step": 2922
    },
    {
      "epoch": 0.22464658881376767,
      "grad_norm": 3.7312428951263428,
      "learning_rate": 1.910141364474493e-05,
      "loss": 1.4724,
      "step": 2924
    },
    {
      "epoch": 0.22480024585126,
      "grad_norm": 3.863544464111328,
      "learning_rate": 1.9100799016594964e-05,
      "loss": 1.4465,
      "step": 2926
    },
    {
      "epoch": 0.2249539028887523,
      "grad_norm": 3.6252338886260986,
      "learning_rate": 1.9100184388444993e-05,
      "loss": 1.5351,
      "step": 2928
    },
    {
      "epoch": 0.22510755992624462,
      "grad_norm": 3.6253201961517334,
      "learning_rate": 1.9099569760295023e-05,
      "loss": 1.487,
      "step": 2930
    },
    {
      "epoch": 0.22526121696373694,
      "grad_norm": 4.433292865753174,
      "learning_rate": 1.9098955132145053e-05,
      "loss": 1.5361,
      "step": 2932
    },
    {
      "epoch": 0.22541487400122925,
      "grad_norm": 3.5505495071411133,
      "learning_rate": 1.9098340503995086e-05,
      "loss": 1.5658,
      "step": 2934
    },
    {
      "epoch": 0.22556853103872157,
      "grad_norm": 3.8930625915527344,
      "learning_rate": 1.9097725875845115e-05,
      "loss": 1.6665,
      "step": 2936
    },
    {
      "epoch": 0.2257221880762139,
      "grad_norm": 3.6677374839782715,
      "learning_rate": 1.9097111247695145e-05,
      "loss": 1.4741,
      "step": 2938
    },
    {
      "epoch": 0.2258758451137062,
      "grad_norm": 3.793630838394165,
      "learning_rate": 1.9096496619545178e-05,
      "loss": 1.5319,
      "step": 2940
    },
    {
      "epoch": 0.22602950215119852,
      "grad_norm": 3.9335010051727295,
      "learning_rate": 1.9095881991395208e-05,
      "loss": 1.5558,
      "step": 2942
    },
    {
      "epoch": 0.22618315918869084,
      "grad_norm": 3.583728551864624,
      "learning_rate": 1.9095267363245237e-05,
      "loss": 1.4462,
      "step": 2944
    },
    {
      "epoch": 0.22633681622618315,
      "grad_norm": 3.9286651611328125,
      "learning_rate": 1.909465273509527e-05,
      "loss": 1.438,
      "step": 2946
    },
    {
      "epoch": 0.22649047326367547,
      "grad_norm": 3.526460647583008,
      "learning_rate": 1.90940381069453e-05,
      "loss": 1.3429,
      "step": 2948
    },
    {
      "epoch": 0.22664413030116778,
      "grad_norm": 4.004605293273926,
      "learning_rate": 1.909342347879533e-05,
      "loss": 1.6075,
      "step": 2950
    },
    {
      "epoch": 0.2267977873386601,
      "grad_norm": 3.9438345432281494,
      "learning_rate": 1.909280885064536e-05,
      "loss": 1.5637,
      "step": 2952
    },
    {
      "epoch": 0.22695144437615242,
      "grad_norm": 4.000015735626221,
      "learning_rate": 1.9092194222495393e-05,
      "loss": 1.4984,
      "step": 2954
    },
    {
      "epoch": 0.22710510141364473,
      "grad_norm": 3.9805843830108643,
      "learning_rate": 1.9091579594345422e-05,
      "loss": 1.5232,
      "step": 2956
    },
    {
      "epoch": 0.22725875845113705,
      "grad_norm": 3.8267416954040527,
      "learning_rate": 1.9090964966195452e-05,
      "loss": 1.496,
      "step": 2958
    },
    {
      "epoch": 0.22741241548862937,
      "grad_norm": 3.9185121059417725,
      "learning_rate": 1.9090350338045485e-05,
      "loss": 1.5375,
      "step": 2960
    },
    {
      "epoch": 0.2275660725261217,
      "grad_norm": 3.3180205821990967,
      "learning_rate": 1.9089735709895515e-05,
      "loss": 1.3771,
      "step": 2962
    },
    {
      "epoch": 0.22771972956361403,
      "grad_norm": 3.9198102951049805,
      "learning_rate": 1.9089121081745544e-05,
      "loss": 1.4476,
      "step": 2964
    },
    {
      "epoch": 0.22787338660110634,
      "grad_norm": 4.271195411682129,
      "learning_rate": 1.9088506453595578e-05,
      "loss": 1.5008,
      "step": 2966
    },
    {
      "epoch": 0.22802704363859866,
      "grad_norm": 3.8531033992767334,
      "learning_rate": 1.9087891825445607e-05,
      "loss": 1.5622,
      "step": 2968
    },
    {
      "epoch": 0.22818070067609098,
      "grad_norm": 3.83855938911438,
      "learning_rate": 1.9087277197295637e-05,
      "loss": 1.5225,
      "step": 2970
    },
    {
      "epoch": 0.2283343577135833,
      "grad_norm": 3.9290051460266113,
      "learning_rate": 1.908666256914567e-05,
      "loss": 1.5943,
      "step": 2972
    },
    {
      "epoch": 0.2284880147510756,
      "grad_norm": 3.9568498134613037,
      "learning_rate": 1.90860479409957e-05,
      "loss": 1.4396,
      "step": 2974
    },
    {
      "epoch": 0.22864167178856792,
      "grad_norm": 3.8864598274230957,
      "learning_rate": 1.908543331284573e-05,
      "loss": 1.5007,
      "step": 2976
    },
    {
      "epoch": 0.22879532882606024,
      "grad_norm": 3.8972291946411133,
      "learning_rate": 1.908481868469576e-05,
      "loss": 1.4453,
      "step": 2978
    },
    {
      "epoch": 0.22894898586355256,
      "grad_norm": 3.3269383907318115,
      "learning_rate": 1.9084204056545792e-05,
      "loss": 1.4875,
      "step": 2980
    },
    {
      "epoch": 0.22910264290104487,
      "grad_norm": 4.227247714996338,
      "learning_rate": 1.9083589428395822e-05,
      "loss": 1.4625,
      "step": 2982
    },
    {
      "epoch": 0.2292562999385372,
      "grad_norm": 4.22311544418335,
      "learning_rate": 1.908297480024585e-05,
      "loss": 1.6099,
      "step": 2984
    },
    {
      "epoch": 0.2294099569760295,
      "grad_norm": 3.618152141571045,
      "learning_rate": 1.9082360172095885e-05,
      "loss": 1.3739,
      "step": 2986
    },
    {
      "epoch": 0.22956361401352182,
      "grad_norm": 4.052188873291016,
      "learning_rate": 1.9081745543945914e-05,
      "loss": 1.4373,
      "step": 2988
    },
    {
      "epoch": 0.22971727105101414,
      "grad_norm": 3.879241943359375,
      "learning_rate": 1.9081130915795944e-05,
      "loss": 1.4802,
      "step": 2990
    },
    {
      "epoch": 0.22987092808850645,
      "grad_norm": 4.114073276519775,
      "learning_rate": 1.9080516287645977e-05,
      "loss": 1.4496,
      "step": 2992
    },
    {
      "epoch": 0.23002458512599877,
      "grad_norm": 3.420759677886963,
      "learning_rate": 1.9079901659496007e-05,
      "loss": 1.4345,
      "step": 2994
    },
    {
      "epoch": 0.2301782421634911,
      "grad_norm": 3.7326130867004395,
      "learning_rate": 1.9079287031346036e-05,
      "loss": 1.4949,
      "step": 2996
    },
    {
      "epoch": 0.2303318992009834,
      "grad_norm": 4.011677265167236,
      "learning_rate": 1.907867240319607e-05,
      "loss": 1.504,
      "step": 2998
    },
    {
      "epoch": 0.23048555623847572,
      "grad_norm": 4.186577796936035,
      "learning_rate": 1.90780577750461e-05,
      "loss": 1.5432,
      "step": 3000
    },
    {
      "epoch": 0.23063921327596804,
      "grad_norm": 3.768660306930542,
      "learning_rate": 1.907744314689613e-05,
      "loss": 1.403,
      "step": 3002
    },
    {
      "epoch": 0.23079287031346035,
      "grad_norm": 3.8193624019622803,
      "learning_rate": 1.907682851874616e-05,
      "loss": 1.401,
      "step": 3004
    },
    {
      "epoch": 0.23094652735095267,
      "grad_norm": 3.425020217895508,
      "learning_rate": 1.907621389059619e-05,
      "loss": 1.4576,
      "step": 3006
    },
    {
      "epoch": 0.23110018438844498,
      "grad_norm": 3.75453782081604,
      "learning_rate": 1.907559926244622e-05,
      "loss": 1.5563,
      "step": 3008
    },
    {
      "epoch": 0.2312538414259373,
      "grad_norm": 3.68115496635437,
      "learning_rate": 1.907498463429625e-05,
      "loss": 1.3805,
      "step": 3010
    },
    {
      "epoch": 0.23140749846342962,
      "grad_norm": 3.3006389141082764,
      "learning_rate": 1.9074370006146284e-05,
      "loss": 1.5252,
      "step": 3012
    },
    {
      "epoch": 0.23156115550092193,
      "grad_norm": 4.148028373718262,
      "learning_rate": 1.9073755377996314e-05,
      "loss": 1.5676,
      "step": 3014
    },
    {
      "epoch": 0.23171481253841425,
      "grad_norm": 3.5741395950317383,
      "learning_rate": 1.9073140749846343e-05,
      "loss": 1.5295,
      "step": 3016
    },
    {
      "epoch": 0.23186846957590657,
      "grad_norm": 3.9004082679748535,
      "learning_rate": 1.9072526121696376e-05,
      "loss": 1.4535,
      "step": 3018
    },
    {
      "epoch": 0.23202212661339888,
      "grad_norm": 4.176894664764404,
      "learning_rate": 1.9071911493546406e-05,
      "loss": 1.3629,
      "step": 3020
    },
    {
      "epoch": 0.2321757836508912,
      "grad_norm": 5.075558662414551,
      "learning_rate": 1.9071296865396436e-05,
      "loss": 1.5217,
      "step": 3022
    },
    {
      "epoch": 0.23232944068838352,
      "grad_norm": 3.4459147453308105,
      "learning_rate": 1.907068223724647e-05,
      "loss": 1.4933,
      "step": 3024
    },
    {
      "epoch": 0.23248309772587583,
      "grad_norm": 7.433279037475586,
      "learning_rate": 1.90700676090965e-05,
      "loss": 1.5533,
      "step": 3026
    },
    {
      "epoch": 0.23263675476336818,
      "grad_norm": 3.6367766857147217,
      "learning_rate": 1.9069452980946528e-05,
      "loss": 1.5066,
      "step": 3028
    },
    {
      "epoch": 0.2327904118008605,
      "grad_norm": 3.537909746170044,
      "learning_rate": 1.9068838352796558e-05,
      "loss": 1.337,
      "step": 3030
    },
    {
      "epoch": 0.2329440688383528,
      "grad_norm": 4.002386093139648,
      "learning_rate": 1.906822372464659e-05,
      "loss": 1.6041,
      "step": 3032
    },
    {
      "epoch": 0.23309772587584512,
      "grad_norm": 3.76108717918396,
      "learning_rate": 1.906760909649662e-05,
      "loss": 1.415,
      "step": 3034
    },
    {
      "epoch": 0.23325138291333744,
      "grad_norm": 3.783128023147583,
      "learning_rate": 1.906699446834665e-05,
      "loss": 1.3389,
      "step": 3036
    },
    {
      "epoch": 0.23340503995082976,
      "grad_norm": 3.628431558609009,
      "learning_rate": 1.9066379840196683e-05,
      "loss": 1.4391,
      "step": 3038
    },
    {
      "epoch": 0.23355869698832207,
      "grad_norm": 3.474383592605591,
      "learning_rate": 1.9065765212046713e-05,
      "loss": 1.6119,
      "step": 3040
    },
    {
      "epoch": 0.2337123540258144,
      "grad_norm": 3.4599947929382324,
      "learning_rate": 1.9065150583896743e-05,
      "loss": 1.5611,
      "step": 3042
    },
    {
      "epoch": 0.2338660110633067,
      "grad_norm": 3.5529699325561523,
      "learning_rate": 1.9064535955746776e-05,
      "loss": 1.5067,
      "step": 3044
    },
    {
      "epoch": 0.23401966810079902,
      "grad_norm": 3.7097628116607666,
      "learning_rate": 1.9063921327596806e-05,
      "loss": 1.5433,
      "step": 3046
    },
    {
      "epoch": 0.23417332513829134,
      "grad_norm": 3.5367672443389893,
      "learning_rate": 1.9063306699446835e-05,
      "loss": 1.5336,
      "step": 3048
    },
    {
      "epoch": 0.23432698217578365,
      "grad_norm": 3.9764819145202637,
      "learning_rate": 1.9062692071296868e-05,
      "loss": 1.4671,
      "step": 3050
    },
    {
      "epoch": 0.23448063921327597,
      "grad_norm": 3.8383781909942627,
      "learning_rate": 1.9062077443146898e-05,
      "loss": 1.5263,
      "step": 3052
    },
    {
      "epoch": 0.2346342962507683,
      "grad_norm": 3.899062395095825,
      "learning_rate": 1.906146281499693e-05,
      "loss": 1.4592,
      "step": 3054
    },
    {
      "epoch": 0.2347879532882606,
      "grad_norm": 3.9547131061553955,
      "learning_rate": 1.9060848186846957e-05,
      "loss": 1.4835,
      "step": 3056
    },
    {
      "epoch": 0.23494161032575292,
      "grad_norm": 4.053800582885742,
      "learning_rate": 1.906023355869699e-05,
      "loss": 1.3099,
      "step": 3058
    },
    {
      "epoch": 0.23509526736324524,
      "grad_norm": 3.65938401222229,
      "learning_rate": 1.905961893054702e-05,
      "loss": 1.5249,
      "step": 3060
    },
    {
      "epoch": 0.23524892440073755,
      "grad_norm": 3.514390707015991,
      "learning_rate": 1.905900430239705e-05,
      "loss": 1.4615,
      "step": 3062
    },
    {
      "epoch": 0.23540258143822987,
      "grad_norm": 3.5489819049835205,
      "learning_rate": 1.9058389674247083e-05,
      "loss": 1.3366,
      "step": 3064
    },
    {
      "epoch": 0.23555623847572218,
      "grad_norm": 3.88149356842041,
      "learning_rate": 1.9057775046097113e-05,
      "loss": 1.6467,
      "step": 3066
    },
    {
      "epoch": 0.2357098955132145,
      "grad_norm": 3.912346601486206,
      "learning_rate": 1.9057160417947142e-05,
      "loss": 1.6762,
      "step": 3068
    },
    {
      "epoch": 0.23586355255070682,
      "grad_norm": 3.9258928298950195,
      "learning_rate": 1.9056545789797175e-05,
      "loss": 1.6048,
      "step": 3070
    },
    {
      "epoch": 0.23601720958819913,
      "grad_norm": 4.021396636962891,
      "learning_rate": 1.9055931161647205e-05,
      "loss": 1.6012,
      "step": 3072
    },
    {
      "epoch": 0.23617086662569145,
      "grad_norm": 3.756319999694824,
      "learning_rate": 1.9055316533497238e-05,
      "loss": 1.505,
      "step": 3074
    },
    {
      "epoch": 0.23632452366318377,
      "grad_norm": 3.8204259872436523,
      "learning_rate": 1.9054701905347264e-05,
      "loss": 1.4867,
      "step": 3076
    },
    {
      "epoch": 0.23647818070067608,
      "grad_norm": 3.6957874298095703,
      "learning_rate": 1.9054087277197297e-05,
      "loss": 1.6277,
      "step": 3078
    },
    {
      "epoch": 0.2366318377381684,
      "grad_norm": 3.272287607192993,
      "learning_rate": 1.9053472649047327e-05,
      "loss": 1.276,
      "step": 3080
    },
    {
      "epoch": 0.23678549477566072,
      "grad_norm": 3.6933488845825195,
      "learning_rate": 1.9052858020897357e-05,
      "loss": 1.569,
      "step": 3082
    },
    {
      "epoch": 0.23693915181315303,
      "grad_norm": 4.391809940338135,
      "learning_rate": 1.905224339274739e-05,
      "loss": 1.4694,
      "step": 3084
    },
    {
      "epoch": 0.23709280885064535,
      "grad_norm": 3.9479763507843018,
      "learning_rate": 1.905162876459742e-05,
      "loss": 1.4931,
      "step": 3086
    },
    {
      "epoch": 0.23724646588813766,
      "grad_norm": 3.6774582862854004,
      "learning_rate": 1.905101413644745e-05,
      "loss": 1.6148,
      "step": 3088
    },
    {
      "epoch": 0.23740012292562998,
      "grad_norm": 3.2491109371185303,
      "learning_rate": 1.9050399508297482e-05,
      "loss": 1.629,
      "step": 3090
    },
    {
      "epoch": 0.23755377996312232,
      "grad_norm": 3.8977487087249756,
      "learning_rate": 1.9049784880147512e-05,
      "loss": 1.4423,
      "step": 3092
    },
    {
      "epoch": 0.23770743700061464,
      "grad_norm": 3.438864231109619,
      "learning_rate": 1.9049170251997545e-05,
      "loss": 1.3834,
      "step": 3094
    },
    {
      "epoch": 0.23786109403810696,
      "grad_norm": 3.799286365509033,
      "learning_rate": 1.9048555623847575e-05,
      "loss": 1.6176,
      "step": 3096
    },
    {
      "epoch": 0.23801475107559927,
      "grad_norm": 3.665806770324707,
      "learning_rate": 1.9047940995697604e-05,
      "loss": 1.6198,
      "step": 3098
    },
    {
      "epoch": 0.2381684081130916,
      "grad_norm": 4.014431476593018,
      "learning_rate": 1.9047326367547637e-05,
      "loss": 1.4675,
      "step": 3100
    },
    {
      "epoch": 0.2383220651505839,
      "grad_norm": 3.293631076812744,
      "learning_rate": 1.9046711739397664e-05,
      "loss": 1.4766,
      "step": 3102
    },
    {
      "epoch": 0.23847572218807622,
      "grad_norm": 3.3270959854125977,
      "learning_rate": 1.9046097111247697e-05,
      "loss": 1.3699,
      "step": 3104
    },
    {
      "epoch": 0.23862937922556854,
      "grad_norm": 3.4034950733184814,
      "learning_rate": 1.9045482483097727e-05,
      "loss": 1.3973,
      "step": 3106
    },
    {
      "epoch": 0.23878303626306085,
      "grad_norm": 4.333171367645264,
      "learning_rate": 1.9044867854947756e-05,
      "loss": 1.4756,
      "step": 3108
    },
    {
      "epoch": 0.23893669330055317,
      "grad_norm": 3.8026325702667236,
      "learning_rate": 1.904425322679779e-05,
      "loss": 1.5391,
      "step": 3110
    },
    {
      "epoch": 0.2390903503380455,
      "grad_norm": 4.236443996429443,
      "learning_rate": 1.904363859864782e-05,
      "loss": 1.5477,
      "step": 3112
    },
    {
      "epoch": 0.2392440073755378,
      "grad_norm": 3.979402780532837,
      "learning_rate": 1.9043023970497852e-05,
      "loss": 1.5373,
      "step": 3114
    },
    {
      "epoch": 0.23939766441303012,
      "grad_norm": 4.0372443199157715,
      "learning_rate": 1.9042409342347882e-05,
      "loss": 1.5543,
      "step": 3116
    },
    {
      "epoch": 0.23955132145052244,
      "grad_norm": 3.884770393371582,
      "learning_rate": 1.904179471419791e-05,
      "loss": 1.4172,
      "step": 3118
    },
    {
      "epoch": 0.23970497848801475,
      "grad_norm": 3.852644205093384,
      "learning_rate": 1.9041180086047944e-05,
      "loss": 1.5288,
      "step": 3120
    },
    {
      "epoch": 0.23985863552550707,
      "grad_norm": 3.6635940074920654,
      "learning_rate": 1.9040565457897974e-05,
      "loss": 1.4201,
      "step": 3122
    },
    {
      "epoch": 0.24001229256299939,
      "grad_norm": 3.9604082107543945,
      "learning_rate": 1.9039950829748004e-05,
      "loss": 1.5166,
      "step": 3124
    },
    {
      "epoch": 0.2401659496004917,
      "grad_norm": 3.084805965423584,
      "learning_rate": 1.9039336201598037e-05,
      "loss": 1.472,
      "step": 3126
    },
    {
      "epoch": 0.24031960663798402,
      "grad_norm": 3.741560697555542,
      "learning_rate": 1.9038721573448063e-05,
      "loss": 1.5564,
      "step": 3128
    },
    {
      "epoch": 0.24047326367547633,
      "grad_norm": 3.3543457984924316,
      "learning_rate": 1.9038106945298096e-05,
      "loss": 1.5025,
      "step": 3130
    },
    {
      "epoch": 0.24062692071296865,
      "grad_norm": 3.7456090450286865,
      "learning_rate": 1.9037492317148126e-05,
      "loss": 1.5189,
      "step": 3132
    },
    {
      "epoch": 0.24078057775046097,
      "grad_norm": 3.4162986278533936,
      "learning_rate": 1.9036877688998156e-05,
      "loss": 1.622,
      "step": 3134
    },
    {
      "epoch": 0.24093423478795328,
      "grad_norm": 4.069525241851807,
      "learning_rate": 1.903626306084819e-05,
      "loss": 1.6334,
      "step": 3136
    },
    {
      "epoch": 0.2410878918254456,
      "grad_norm": 3.6952528953552246,
      "learning_rate": 1.903564843269822e-05,
      "loss": 1.4343,
      "step": 3138
    },
    {
      "epoch": 0.24124154886293792,
      "grad_norm": 3.562882900238037,
      "learning_rate": 1.903503380454825e-05,
      "loss": 1.3561,
      "step": 3140
    },
    {
      "epoch": 0.24139520590043023,
      "grad_norm": 3.8922019004821777,
      "learning_rate": 1.903441917639828e-05,
      "loss": 1.4585,
      "step": 3142
    },
    {
      "epoch": 0.24154886293792255,
      "grad_norm": 3.665839910507202,
      "learning_rate": 1.903380454824831e-05,
      "loss": 1.7023,
      "step": 3144
    },
    {
      "epoch": 0.24170251997541486,
      "grad_norm": 3.4272682666778564,
      "learning_rate": 1.9033189920098344e-05,
      "loss": 1.5242,
      "step": 3146
    },
    {
      "epoch": 0.24185617701290718,
      "grad_norm": 3.8519303798675537,
      "learning_rate": 1.9032575291948374e-05,
      "loss": 1.4245,
      "step": 3148
    },
    {
      "epoch": 0.2420098340503995,
      "grad_norm": 3.6139211654663086,
      "learning_rate": 1.9031960663798403e-05,
      "loss": 1.3585,
      "step": 3150
    },
    {
      "epoch": 0.2421634910878918,
      "grad_norm": 4.0420966148376465,
      "learning_rate": 1.9031346035648436e-05,
      "loss": 1.577,
      "step": 3152
    },
    {
      "epoch": 0.24231714812538413,
      "grad_norm": 3.215604305267334,
      "learning_rate": 1.9030731407498463e-05,
      "loss": 1.4625,
      "step": 3154
    },
    {
      "epoch": 0.24247080516287645,
      "grad_norm": 4.21508264541626,
      "learning_rate": 1.9030116779348496e-05,
      "loss": 1.4881,
      "step": 3156
    },
    {
      "epoch": 0.2426244622003688,
      "grad_norm": 3.351759910583496,
      "learning_rate": 1.9029502151198525e-05,
      "loss": 1.5047,
      "step": 3158
    },
    {
      "epoch": 0.2427781192378611,
      "grad_norm": 3.681420087814331,
      "learning_rate": 1.902888752304856e-05,
      "loss": 1.3997,
      "step": 3160
    },
    {
      "epoch": 0.24293177627535342,
      "grad_norm": 3.751452922821045,
      "learning_rate": 1.9028272894898588e-05,
      "loss": 1.5441,
      "step": 3162
    },
    {
      "epoch": 0.24308543331284574,
      "grad_norm": 3.7483127117156982,
      "learning_rate": 1.9027658266748618e-05,
      "loss": 1.5719,
      "step": 3164
    },
    {
      "epoch": 0.24323909035033806,
      "grad_norm": 4.023811340332031,
      "learning_rate": 1.902704363859865e-05,
      "loss": 1.5686,
      "step": 3166
    },
    {
      "epoch": 0.24339274738783037,
      "grad_norm": 3.655235767364502,
      "learning_rate": 1.902642901044868e-05,
      "loss": 1.4781,
      "step": 3168
    },
    {
      "epoch": 0.2435464044253227,
      "grad_norm": 3.8731863498687744,
      "learning_rate": 1.902581438229871e-05,
      "loss": 1.4679,
      "step": 3170
    },
    {
      "epoch": 0.243700061462815,
      "grad_norm": 3.9271247386932373,
      "learning_rate": 1.9025199754148743e-05,
      "loss": 1.6756,
      "step": 3172
    },
    {
      "epoch": 0.24385371850030732,
      "grad_norm": 3.7917888164520264,
      "learning_rate": 1.902458512599877e-05,
      "loss": 1.5004,
      "step": 3174
    },
    {
      "epoch": 0.24400737553779964,
      "grad_norm": 3.341165781021118,
      "learning_rate": 1.9023970497848803e-05,
      "loss": 1.4226,
      "step": 3176
    },
    {
      "epoch": 0.24416103257529195,
      "grad_norm": 3.6708719730377197,
      "learning_rate": 1.9023355869698832e-05,
      "loss": 1.4637,
      "step": 3178
    },
    {
      "epoch": 0.24431468961278427,
      "grad_norm": 3.8046674728393555,
      "learning_rate": 1.9022741241548865e-05,
      "loss": 1.4156,
      "step": 3180
    },
    {
      "epoch": 0.24446834665027659,
      "grad_norm": 3.7367637157440186,
      "learning_rate": 1.9022126613398895e-05,
      "loss": 1.5416,
      "step": 3182
    },
    {
      "epoch": 0.2446220036877689,
      "grad_norm": 3.453035354614258,
      "learning_rate": 1.9021511985248925e-05,
      "loss": 1.459,
      "step": 3184
    },
    {
      "epoch": 0.24477566072526122,
      "grad_norm": 3.8223092555999756,
      "learning_rate": 1.9020897357098958e-05,
      "loss": 1.5786,
      "step": 3186
    },
    {
      "epoch": 0.24492931776275353,
      "grad_norm": 3.4199178218841553,
      "learning_rate": 1.9020282728948988e-05,
      "loss": 1.6042,
      "step": 3188
    },
    {
      "epoch": 0.24508297480024585,
      "grad_norm": 3.828610420227051,
      "learning_rate": 1.9019668100799017e-05,
      "loss": 1.4459,
      "step": 3190
    },
    {
      "epoch": 0.24523663183773817,
      "grad_norm": 3.5266432762145996,
      "learning_rate": 1.901905347264905e-05,
      "loss": 1.5675,
      "step": 3192
    },
    {
      "epoch": 0.24539028887523048,
      "grad_norm": 3.738628387451172,
      "learning_rate": 1.901843884449908e-05,
      "loss": 1.4532,
      "step": 3194
    },
    {
      "epoch": 0.2455439459127228,
      "grad_norm": 3.32661771774292,
      "learning_rate": 1.901782421634911e-05,
      "loss": 1.4821,
      "step": 3196
    },
    {
      "epoch": 0.24569760295021512,
      "grad_norm": 3.8040380477905273,
      "learning_rate": 1.9017209588199143e-05,
      "loss": 1.5357,
      "step": 3198
    },
    {
      "epoch": 0.24585125998770743,
      "grad_norm": 3.24855637550354,
      "learning_rate": 1.9016594960049172e-05,
      "loss": 1.36,
      "step": 3200
    },
    {
      "epoch": 0.24600491702519975,
      "grad_norm": 3.494410514831543,
      "learning_rate": 1.9015980331899202e-05,
      "loss": 1.705,
      "step": 3202
    },
    {
      "epoch": 0.24615857406269206,
      "grad_norm": 3.760875940322876,
      "learning_rate": 1.9015365703749232e-05,
      "loss": 1.4332,
      "step": 3204
    },
    {
      "epoch": 0.24631223110018438,
      "grad_norm": 3.2367358207702637,
      "learning_rate": 1.9014751075599265e-05,
      "loss": 1.429,
      "step": 3206
    },
    {
      "epoch": 0.2464658881376767,
      "grad_norm": 3.4821407794952393,
      "learning_rate": 1.9014136447449295e-05,
      "loss": 1.5069,
      "step": 3208
    },
    {
      "epoch": 0.246619545175169,
      "grad_norm": 3.640047073364258,
      "learning_rate": 1.9013521819299324e-05,
      "loss": 1.4913,
      "step": 3210
    },
    {
      "epoch": 0.24677320221266133,
      "grad_norm": 3.7262701988220215,
      "learning_rate": 1.9012907191149357e-05,
      "loss": 1.4896,
      "step": 3212
    },
    {
      "epoch": 0.24692685925015365,
      "grad_norm": 4.199042797088623,
      "learning_rate": 1.9012292562999387e-05,
      "loss": 1.5823,
      "step": 3214
    },
    {
      "epoch": 0.24708051628764596,
      "grad_norm": 3.3966314792633057,
      "learning_rate": 1.9011677934849417e-05,
      "loss": 1.4982,
      "step": 3216
    },
    {
      "epoch": 0.24723417332513828,
      "grad_norm": 3.9125616550445557,
      "learning_rate": 1.901106330669945e-05,
      "loss": 1.4921,
      "step": 3218
    },
    {
      "epoch": 0.2473878303626306,
      "grad_norm": 3.583458423614502,
      "learning_rate": 1.901044867854948e-05,
      "loss": 1.4831,
      "step": 3220
    },
    {
      "epoch": 0.24754148740012294,
      "grad_norm": 3.9417824745178223,
      "learning_rate": 1.900983405039951e-05,
      "loss": 1.533,
      "step": 3222
    },
    {
      "epoch": 0.24769514443761526,
      "grad_norm": 3.6411452293395996,
      "learning_rate": 1.9009219422249542e-05,
      "loss": 1.4955,
      "step": 3224
    },
    {
      "epoch": 0.24784880147510757,
      "grad_norm": 3.75683856010437,
      "learning_rate": 1.9008604794099572e-05,
      "loss": 1.5469,
      "step": 3226
    },
    {
      "epoch": 0.2480024585125999,
      "grad_norm": 4.777811050415039,
      "learning_rate": 1.90079901659496e-05,
      "loss": 1.4417,
      "step": 3228
    },
    {
      "epoch": 0.2481561155500922,
      "grad_norm": 3.5624022483825684,
      "learning_rate": 1.900737553779963e-05,
      "loss": 1.4786,
      "step": 3230
    },
    {
      "epoch": 0.24830977258758452,
      "grad_norm": 3.511544704437256,
      "learning_rate": 1.9006760909649664e-05,
      "loss": 1.4188,
      "step": 3232
    },
    {
      "epoch": 0.24846342962507684,
      "grad_norm": 3.966139078140259,
      "learning_rate": 1.9006146281499694e-05,
      "loss": 1.5502,
      "step": 3234
    },
    {
      "epoch": 0.24861708666256915,
      "grad_norm": 3.792802095413208,
      "learning_rate": 1.9005531653349724e-05,
      "loss": 1.4778,
      "step": 3236
    },
    {
      "epoch": 0.24877074370006147,
      "grad_norm": 4.079867839813232,
      "learning_rate": 1.9004917025199757e-05,
      "loss": 1.5482,
      "step": 3238
    },
    {
      "epoch": 0.24892440073755379,
      "grad_norm": 3.424563407897949,
      "learning_rate": 1.9004302397049786e-05,
      "loss": 1.5189,
      "step": 3240
    },
    {
      "epoch": 0.2490780577750461,
      "grad_norm": 3.5327744483947754,
      "learning_rate": 1.9003687768899816e-05,
      "loss": 1.4631,
      "step": 3242
    },
    {
      "epoch": 0.24923171481253842,
      "grad_norm": 3.8727915287017822,
      "learning_rate": 1.900307314074985e-05,
      "loss": 1.5833,
      "step": 3244
    },
    {
      "epoch": 0.24938537185003073,
      "grad_norm": 4.0103302001953125,
      "learning_rate": 1.900245851259988e-05,
      "loss": 1.4591,
      "step": 3246
    },
    {
      "epoch": 0.24953902888752305,
      "grad_norm": 4.012939453125,
      "learning_rate": 1.900184388444991e-05,
      "loss": 1.5609,
      "step": 3248
    },
    {
      "epoch": 0.24969268592501537,
      "grad_norm": 3.425726890563965,
      "learning_rate": 1.900122925629994e-05,
      "loss": 1.3594,
      "step": 3250
    },
    {
      "epoch": 0.24984634296250768,
      "grad_norm": 3.660951614379883,
      "learning_rate": 1.900061462814997e-05,
      "loss": 1.3977,
      "step": 3252
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.4327516555786133,
      "learning_rate": 1.9e-05,
      "loss": 1.5269,
      "step": 3254
    },
    {
      "epoch": 0.2501536570374923,
      "grad_norm": 3.792966365814209,
      "learning_rate": 1.899938537185003e-05,
      "loss": 1.4157,
      "step": 3256
    },
    {
      "epoch": 0.25030731407498463,
      "grad_norm": 3.777806282043457,
      "learning_rate": 1.8998770743700064e-05,
      "loss": 1.61,
      "step": 3258
    },
    {
      "epoch": 0.25046097111247695,
      "grad_norm": 3.6547727584838867,
      "learning_rate": 1.8998156115550093e-05,
      "loss": 1.578,
      "step": 3260
    },
    {
      "epoch": 0.25061462814996927,
      "grad_norm": 3.7142393589019775,
      "learning_rate": 1.8997541487400123e-05,
      "loss": 1.5387,
      "step": 3262
    },
    {
      "epoch": 0.2507682851874616,
      "grad_norm": 3.795875310897827,
      "learning_rate": 1.8996926859250156e-05,
      "loss": 1.4108,
      "step": 3264
    },
    {
      "epoch": 0.2509219422249539,
      "grad_norm": 4.018312931060791,
      "learning_rate": 1.8996312231100186e-05,
      "loss": 1.609,
      "step": 3266
    },
    {
      "epoch": 0.2510755992624462,
      "grad_norm": 3.576310157775879,
      "learning_rate": 1.8995697602950216e-05,
      "loss": 1.5313,
      "step": 3268
    },
    {
      "epoch": 0.25122925629993853,
      "grad_norm": 3.7783751487731934,
      "learning_rate": 1.899508297480025e-05,
      "loss": 1.5393,
      "step": 3270
    },
    {
      "epoch": 0.25138291333743085,
      "grad_norm": 3.8343939781188965,
      "learning_rate": 1.899446834665028e-05,
      "loss": 1.4593,
      "step": 3272
    },
    {
      "epoch": 0.25153657037492316,
      "grad_norm": 3.8293821811676025,
      "learning_rate": 1.8993853718500308e-05,
      "loss": 1.4991,
      "step": 3274
    },
    {
      "epoch": 0.2516902274124155,
      "grad_norm": 3.804919958114624,
      "learning_rate": 1.8993239090350338e-05,
      "loss": 1.5475,
      "step": 3276
    },
    {
      "epoch": 0.2518438844499078,
      "grad_norm": 3.3940484523773193,
      "learning_rate": 1.899262446220037e-05,
      "loss": 1.329,
      "step": 3278
    },
    {
      "epoch": 0.2519975414874001,
      "grad_norm": 4.106419086456299,
      "learning_rate": 1.89920098340504e-05,
      "loss": 1.6195,
      "step": 3280
    },
    {
      "epoch": 0.25215119852489243,
      "grad_norm": 3.56506609916687,
      "learning_rate": 1.899139520590043e-05,
      "loss": 1.6574,
      "step": 3282
    },
    {
      "epoch": 0.25230485556238474,
      "grad_norm": 3.6036717891693115,
      "learning_rate": 1.8990780577750463e-05,
      "loss": 1.4525,
      "step": 3284
    },
    {
      "epoch": 0.25245851259987706,
      "grad_norm": 4.163785934448242,
      "learning_rate": 1.8990165949600493e-05,
      "loss": 1.4881,
      "step": 3286
    },
    {
      "epoch": 0.2526121696373694,
      "grad_norm": 4.115818023681641,
      "learning_rate": 1.8989551321450523e-05,
      "loss": 1.5126,
      "step": 3288
    },
    {
      "epoch": 0.2527658266748617,
      "grad_norm": 3.6225428581237793,
      "learning_rate": 1.8988936693300556e-05,
      "loss": 1.4663,
      "step": 3290
    },
    {
      "epoch": 0.252919483712354,
      "grad_norm": 3.335973024368286,
      "learning_rate": 1.8988322065150585e-05,
      "loss": 1.3223,
      "step": 3292
    },
    {
      "epoch": 0.2530731407498463,
      "grad_norm": 3.5263774394989014,
      "learning_rate": 1.8987707437000615e-05,
      "loss": 1.4948,
      "step": 3294
    },
    {
      "epoch": 0.25322679778733864,
      "grad_norm": 3.5017597675323486,
      "learning_rate": 1.8987092808850648e-05,
      "loss": 1.2772,
      "step": 3296
    },
    {
      "epoch": 0.25338045482483096,
      "grad_norm": 3.744292736053467,
      "learning_rate": 1.8986478180700678e-05,
      "loss": 1.3346,
      "step": 3298
    },
    {
      "epoch": 0.2535341118623233,
      "grad_norm": 4.142531871795654,
      "learning_rate": 1.8985863552550707e-05,
      "loss": 1.4425,
      "step": 3300
    },
    {
      "epoch": 0.2536877688998156,
      "grad_norm": 3.734403371810913,
      "learning_rate": 1.8985248924400737e-05,
      "loss": 1.5465,
      "step": 3302
    },
    {
      "epoch": 0.2538414259373079,
      "grad_norm": 3.28890061378479,
      "learning_rate": 1.898463429625077e-05,
      "loss": 1.4063,
      "step": 3304
    },
    {
      "epoch": 0.2539950829748002,
      "grad_norm": 3.3916125297546387,
      "learning_rate": 1.89840196681008e-05,
      "loss": 1.5272,
      "step": 3306
    },
    {
      "epoch": 0.25414874001229254,
      "grad_norm": 3.462782382965088,
      "learning_rate": 1.898340503995083e-05,
      "loss": 1.5789,
      "step": 3308
    },
    {
      "epoch": 0.25430239704978486,
      "grad_norm": 3.189033031463623,
      "learning_rate": 1.8982790411800863e-05,
      "loss": 1.4295,
      "step": 3310
    },
    {
      "epoch": 0.2544560540872772,
      "grad_norm": 3.8161280155181885,
      "learning_rate": 1.8982175783650892e-05,
      "loss": 1.4154,
      "step": 3312
    },
    {
      "epoch": 0.2546097111247695,
      "grad_norm": 3.279613494873047,
      "learning_rate": 1.8981561155500922e-05,
      "loss": 1.3957,
      "step": 3314
    },
    {
      "epoch": 0.2547633681622618,
      "grad_norm": 3.655524492263794,
      "learning_rate": 1.8980946527350955e-05,
      "loss": 1.5173,
      "step": 3316
    },
    {
      "epoch": 0.2549170251997541,
      "grad_norm": 3.407174587249756,
      "learning_rate": 1.8980331899200985e-05,
      "loss": 1.4034,
      "step": 3318
    },
    {
      "epoch": 0.2550706822372465,
      "grad_norm": 3.6055874824523926,
      "learning_rate": 1.8979717271051014e-05,
      "loss": 1.4338,
      "step": 3320
    },
    {
      "epoch": 0.2552243392747388,
      "grad_norm": 3.903757333755493,
      "learning_rate": 1.8979102642901048e-05,
      "loss": 1.4843,
      "step": 3322
    },
    {
      "epoch": 0.2553779963122311,
      "grad_norm": 3.445702314376831,
      "learning_rate": 1.8978488014751077e-05,
      "loss": 1.5697,
      "step": 3324
    },
    {
      "epoch": 0.25553165334972344,
      "grad_norm": 3.888247013092041,
      "learning_rate": 1.897787338660111e-05,
      "loss": 1.3772,
      "step": 3326
    },
    {
      "epoch": 0.25568531038721576,
      "grad_norm": 3.8378067016601562,
      "learning_rate": 1.8977258758451137e-05,
      "loss": 1.461,
      "step": 3328
    },
    {
      "epoch": 0.2558389674247081,
      "grad_norm": 4.017933368682861,
      "learning_rate": 1.897664413030117e-05,
      "loss": 1.4029,
      "step": 3330
    },
    {
      "epoch": 0.2559926244622004,
      "grad_norm": 3.43816876411438,
      "learning_rate": 1.89760295021512e-05,
      "loss": 1.4058,
      "step": 3332
    },
    {
      "epoch": 0.2561462814996927,
      "grad_norm": 3.6177170276641846,
      "learning_rate": 1.897541487400123e-05,
      "loss": 1.5816,
      "step": 3334
    },
    {
      "epoch": 0.256299938537185,
      "grad_norm": 3.708869218826294,
      "learning_rate": 1.8974800245851262e-05,
      "loss": 1.5383,
      "step": 3336
    },
    {
      "epoch": 0.25645359557467734,
      "grad_norm": 3.4780690670013428,
      "learning_rate": 1.8974185617701292e-05,
      "loss": 1.4352,
      "step": 3338
    },
    {
      "epoch": 0.25660725261216966,
      "grad_norm": 3.964945077896118,
      "learning_rate": 1.897357098955132e-05,
      "loss": 1.4872,
      "step": 3340
    },
    {
      "epoch": 0.256760909649662,
      "grad_norm": 3.3842320442199707,
      "learning_rate": 1.8972956361401355e-05,
      "loss": 1.4933,
      "step": 3342
    },
    {
      "epoch": 0.2569145666871543,
      "grad_norm": 3.7447729110717773,
      "learning_rate": 1.8972341733251384e-05,
      "loss": 1.6113,
      "step": 3344
    },
    {
      "epoch": 0.2570682237246466,
      "grad_norm": 4.0971174240112305,
      "learning_rate": 1.8971727105101417e-05,
      "loss": 1.4749,
      "step": 3346
    },
    {
      "epoch": 0.2572218807621389,
      "grad_norm": 3.2165610790252686,
      "learning_rate": 1.8971112476951447e-05,
      "loss": 1.4361,
      "step": 3348
    },
    {
      "epoch": 0.25737553779963124,
      "grad_norm": 3.8490447998046875,
      "learning_rate": 1.8970497848801477e-05,
      "loss": 1.5455,
      "step": 3350
    },
    {
      "epoch": 0.25752919483712355,
      "grad_norm": 4.487706661224365,
      "learning_rate": 1.896988322065151e-05,
      "loss": 1.5265,
      "step": 3352
    },
    {
      "epoch": 0.25768285187461587,
      "grad_norm": 3.487905502319336,
      "learning_rate": 1.8969268592501536e-05,
      "loss": 1.4864,
      "step": 3354
    },
    {
      "epoch": 0.2578365089121082,
      "grad_norm": 4.049886703491211,
      "learning_rate": 1.896865396435157e-05,
      "loss": 1.3788,
      "step": 3356
    },
    {
      "epoch": 0.2579901659496005,
      "grad_norm": 3.7233083248138428,
      "learning_rate": 1.89680393362016e-05,
      "loss": 1.4227,
      "step": 3358
    },
    {
      "epoch": 0.2581438229870928,
      "grad_norm": 3.6753907203674316,
      "learning_rate": 1.896742470805163e-05,
      "loss": 1.5348,
      "step": 3360
    },
    {
      "epoch": 0.25829748002458514,
      "grad_norm": 3.5778372287750244,
      "learning_rate": 1.896681007990166e-05,
      "loss": 1.4769,
      "step": 3362
    },
    {
      "epoch": 0.25845113706207745,
      "grad_norm": 3.8472750186920166,
      "learning_rate": 1.896619545175169e-05,
      "loss": 1.4646,
      "step": 3364
    },
    {
      "epoch": 0.25860479409956977,
      "grad_norm": 3.4961729049682617,
      "learning_rate": 1.8965580823601724e-05,
      "loss": 1.4237,
      "step": 3366
    },
    {
      "epoch": 0.2587584511370621,
      "grad_norm": 3.384688377380371,
      "learning_rate": 1.8964966195451754e-05,
      "loss": 1.3757,
      "step": 3368
    },
    {
      "epoch": 0.2589121081745544,
      "grad_norm": 3.461256265640259,
      "learning_rate": 1.8964351567301784e-05,
      "loss": 1.6313,
      "step": 3370
    },
    {
      "epoch": 0.2590657652120467,
      "grad_norm": 3.5917470455169678,
      "learning_rate": 1.8963736939151817e-05,
      "loss": 1.6363,
      "step": 3372
    },
    {
      "epoch": 0.25921942224953903,
      "grad_norm": 4.257856845855713,
      "learning_rate": 1.8963122311001846e-05,
      "loss": 1.5657,
      "step": 3374
    },
    {
      "epoch": 0.25937307928703135,
      "grad_norm": 3.825030565261841,
      "learning_rate": 1.8962507682851876e-05,
      "loss": 1.4029,
      "step": 3376
    },
    {
      "epoch": 0.25952673632452367,
      "grad_norm": 3.698003053665161,
      "learning_rate": 1.896189305470191e-05,
      "loss": 1.4794,
      "step": 3378
    },
    {
      "epoch": 0.259680393362016,
      "grad_norm": 3.656878709793091,
      "learning_rate": 1.8961278426551935e-05,
      "loss": 1.4764,
      "step": 3380
    },
    {
      "epoch": 0.2598340503995083,
      "grad_norm": 3.9345102310180664,
      "learning_rate": 1.896066379840197e-05,
      "loss": 1.4978,
      "step": 3382
    },
    {
      "epoch": 0.2599877074370006,
      "grad_norm": 3.2846667766571045,
      "learning_rate": 1.8960049170251998e-05,
      "loss": 1.4882,
      "step": 3384
    },
    {
      "epoch": 0.26014136447449293,
      "grad_norm": 3.6656949520111084,
      "learning_rate": 1.8959434542102028e-05,
      "loss": 1.5028,
      "step": 3386
    },
    {
      "epoch": 0.26029502151198525,
      "grad_norm": 3.3560938835144043,
      "learning_rate": 1.895881991395206e-05,
      "loss": 1.462,
      "step": 3388
    },
    {
      "epoch": 0.26044867854947756,
      "grad_norm": 3.903484582901001,
      "learning_rate": 1.895820528580209e-05,
      "loss": 1.5055,
      "step": 3390
    },
    {
      "epoch": 0.2606023355869699,
      "grad_norm": 3.4434473514556885,
      "learning_rate": 1.8957590657652124e-05,
      "loss": 1.4052,
      "step": 3392
    },
    {
      "epoch": 0.2607559926244622,
      "grad_norm": 3.8662304878234863,
      "learning_rate": 1.8956976029502153e-05,
      "loss": 1.4433,
      "step": 3394
    },
    {
      "epoch": 0.2609096496619545,
      "grad_norm": 3.921478509902954,
      "learning_rate": 1.8956361401352183e-05,
      "loss": 1.4143,
      "step": 3396
    },
    {
      "epoch": 0.26106330669944683,
      "grad_norm": 4.5299601554870605,
      "learning_rate": 1.8955746773202216e-05,
      "loss": 1.5001,
      "step": 3398
    },
    {
      "epoch": 0.26121696373693915,
      "grad_norm": 4.293244361877441,
      "learning_rate": 1.8955132145052242e-05,
      "loss": 1.4988,
      "step": 3400
    },
    {
      "epoch": 0.26137062077443146,
      "grad_norm": 3.8071391582489014,
      "learning_rate": 1.8954517516902276e-05,
      "loss": 1.5376,
      "step": 3402
    },
    {
      "epoch": 0.2615242778119238,
      "grad_norm": 3.9262189865112305,
      "learning_rate": 1.8953902888752305e-05,
      "loss": 1.6795,
      "step": 3404
    },
    {
      "epoch": 0.2616779348494161,
      "grad_norm": 3.64919376373291,
      "learning_rate": 1.8953288260602335e-05,
      "loss": 1.4907,
      "step": 3406
    },
    {
      "epoch": 0.2618315918869084,
      "grad_norm": 3.7635629177093506,
      "learning_rate": 1.8952673632452368e-05,
      "loss": 1.6125,
      "step": 3408
    },
    {
      "epoch": 0.2619852489244007,
      "grad_norm": 3.243130683898926,
      "learning_rate": 1.8952059004302398e-05,
      "loss": 1.4673,
      "step": 3410
    },
    {
      "epoch": 0.26213890596189304,
      "grad_norm": 4.027158737182617,
      "learning_rate": 1.895144437615243e-05,
      "loss": 1.572,
      "step": 3412
    },
    {
      "epoch": 0.26229256299938536,
      "grad_norm": 3.7150518894195557,
      "learning_rate": 1.895082974800246e-05,
      "loss": 1.5544,
      "step": 3414
    },
    {
      "epoch": 0.2624462200368777,
      "grad_norm": 3.56063175201416,
      "learning_rate": 1.895021511985249e-05,
      "loss": 1.4073,
      "step": 3416
    },
    {
      "epoch": 0.26259987707437,
      "grad_norm": 3.2205371856689453,
      "learning_rate": 1.8949600491702523e-05,
      "loss": 1.3674,
      "step": 3418
    },
    {
      "epoch": 0.2627535341118623,
      "grad_norm": 3.5181539058685303,
      "learning_rate": 1.8948985863552553e-05,
      "loss": 1.5704,
      "step": 3420
    },
    {
      "epoch": 0.2629071911493546,
      "grad_norm": 3.9559669494628906,
      "learning_rate": 1.8948371235402583e-05,
      "loss": 1.4059,
      "step": 3422
    },
    {
      "epoch": 0.26306084818684694,
      "grad_norm": 3.9452176094055176,
      "learning_rate": 1.8947756607252616e-05,
      "loss": 1.5914,
      "step": 3424
    },
    {
      "epoch": 0.26321450522433926,
      "grad_norm": 3.5629098415374756,
      "learning_rate": 1.8947141979102642e-05,
      "loss": 1.4329,
      "step": 3426
    },
    {
      "epoch": 0.2633681622618316,
      "grad_norm": 3.84218430519104,
      "learning_rate": 1.8946527350952675e-05,
      "loss": 1.4207,
      "step": 3428
    },
    {
      "epoch": 0.2635218192993239,
      "grad_norm": 3.7253074645996094,
      "learning_rate": 1.8945912722802705e-05,
      "loss": 1.623,
      "step": 3430
    },
    {
      "epoch": 0.2636754763368162,
      "grad_norm": 3.367321729660034,
      "learning_rate": 1.8945298094652738e-05,
      "loss": 1.3962,
      "step": 3432
    },
    {
      "epoch": 0.2638291333743085,
      "grad_norm": 3.7452688217163086,
      "learning_rate": 1.8944683466502767e-05,
      "loss": 1.5708,
      "step": 3434
    },
    {
      "epoch": 0.26398279041180084,
      "grad_norm": 4.207433700561523,
      "learning_rate": 1.8944068838352797e-05,
      "loss": 1.5347,
      "step": 3436
    },
    {
      "epoch": 0.26413644744929315,
      "grad_norm": 3.2880661487579346,
      "learning_rate": 1.894345421020283e-05,
      "loss": 1.4852,
      "step": 3438
    },
    {
      "epoch": 0.26429010448678547,
      "grad_norm": 3.8437421321868896,
      "learning_rate": 1.894283958205286e-05,
      "loss": 1.4737,
      "step": 3440
    },
    {
      "epoch": 0.2644437615242778,
      "grad_norm": 4.380143642425537,
      "learning_rate": 1.894222495390289e-05,
      "loss": 1.4646,
      "step": 3442
    },
    {
      "epoch": 0.2645974185617701,
      "grad_norm": 3.446812391281128,
      "learning_rate": 1.8941610325752923e-05,
      "loss": 1.5911,
      "step": 3444
    },
    {
      "epoch": 0.2647510755992624,
      "grad_norm": 3.8308660984039307,
      "learning_rate": 1.8940995697602952e-05,
      "loss": 1.4695,
      "step": 3446
    },
    {
      "epoch": 0.26490473263675474,
      "grad_norm": 3.475552558898926,
      "learning_rate": 1.8940381069452982e-05,
      "loss": 1.6034,
      "step": 3448
    },
    {
      "epoch": 0.2650583896742471,
      "grad_norm": 3.3068044185638428,
      "learning_rate": 1.8939766441303015e-05,
      "loss": 1.4709,
      "step": 3450
    },
    {
      "epoch": 0.2652120467117394,
      "grad_norm": 3.57163143157959,
      "learning_rate": 1.8939151813153045e-05,
      "loss": 1.4525,
      "step": 3452
    },
    {
      "epoch": 0.26536570374923174,
      "grad_norm": 3.2639644145965576,
      "learning_rate": 1.8938537185003074e-05,
      "loss": 1.5822,
      "step": 3454
    },
    {
      "epoch": 0.26551936078672406,
      "grad_norm": 3.7956414222717285,
      "learning_rate": 1.8937922556853104e-05,
      "loss": 1.4841,
      "step": 3456
    },
    {
      "epoch": 0.2656730178242164,
      "grad_norm": 3.958355665206909,
      "learning_rate": 1.8937307928703137e-05,
      "loss": 1.6366,
      "step": 3458
    },
    {
      "epoch": 0.2658266748617087,
      "grad_norm": 4.118971347808838,
      "learning_rate": 1.8936693300553167e-05,
      "loss": 1.6656,
      "step": 3460
    },
    {
      "epoch": 0.265980331899201,
      "grad_norm": 3.3143534660339355,
      "learning_rate": 1.8936078672403197e-05,
      "loss": 1.5126,
      "step": 3462
    },
    {
      "epoch": 0.2661339889366933,
      "grad_norm": 3.237959384918213,
      "learning_rate": 1.893546404425323e-05,
      "loss": 1.4093,
      "step": 3464
    },
    {
      "epoch": 0.26628764597418564,
      "grad_norm": 3.183522939682007,
      "learning_rate": 1.893484941610326e-05,
      "loss": 1.3261,
      "step": 3466
    },
    {
      "epoch": 0.26644130301167795,
      "grad_norm": 3.7242038249969482,
      "learning_rate": 1.893423478795329e-05,
      "loss": 1.5207,
      "step": 3468
    },
    {
      "epoch": 0.26659496004917027,
      "grad_norm": 3.6746206283569336,
      "learning_rate": 1.8933620159803322e-05,
      "loss": 1.5599,
      "step": 3470
    },
    {
      "epoch": 0.2667486170866626,
      "grad_norm": 3.866694211959839,
      "learning_rate": 1.8933005531653352e-05,
      "loss": 1.4262,
      "step": 3472
    },
    {
      "epoch": 0.2669022741241549,
      "grad_norm": 3.508993148803711,
      "learning_rate": 1.893239090350338e-05,
      "loss": 1.4758,
      "step": 3474
    },
    {
      "epoch": 0.2670559311616472,
      "grad_norm": 3.6842212677001953,
      "learning_rate": 1.8931776275353415e-05,
      "loss": 1.5336,
      "step": 3476
    },
    {
      "epoch": 0.26720958819913954,
      "grad_norm": 3.6138060092926025,
      "learning_rate": 1.8931161647203444e-05,
      "loss": 1.4907,
      "step": 3478
    },
    {
      "epoch": 0.26736324523663185,
      "grad_norm": 3.9442241191864014,
      "learning_rate": 1.8930547019053474e-05,
      "loss": 1.4674,
      "step": 3480
    },
    {
      "epoch": 0.26751690227412417,
      "grad_norm": 3.853156089782715,
      "learning_rate": 1.8929932390903504e-05,
      "loss": 1.5482,
      "step": 3482
    },
    {
      "epoch": 0.2676705593116165,
      "grad_norm": 3.863734245300293,
      "learning_rate": 1.8929317762753537e-05,
      "loss": 1.3643,
      "step": 3484
    },
    {
      "epoch": 0.2678242163491088,
      "grad_norm": 3.796949625015259,
      "learning_rate": 1.8928703134603566e-05,
      "loss": 1.5439,
      "step": 3486
    },
    {
      "epoch": 0.2679778733866011,
      "grad_norm": 3.865708827972412,
      "learning_rate": 1.8928088506453596e-05,
      "loss": 1.3416,
      "step": 3488
    },
    {
      "epoch": 0.26813153042409343,
      "grad_norm": 3.347963571548462,
      "learning_rate": 1.892747387830363e-05,
      "loss": 1.495,
      "step": 3490
    },
    {
      "epoch": 0.26828518746158575,
      "grad_norm": 3.663733959197998,
      "learning_rate": 1.892685925015366e-05,
      "loss": 1.5101,
      "step": 3492
    },
    {
      "epoch": 0.26843884449907807,
      "grad_norm": 3.4801948070526123,
      "learning_rate": 1.892624462200369e-05,
      "loss": 1.5782,
      "step": 3494
    },
    {
      "epoch": 0.2685925015365704,
      "grad_norm": 3.6601758003234863,
      "learning_rate": 1.892562999385372e-05,
      "loss": 1.6072,
      "step": 3496
    },
    {
      "epoch": 0.2687461585740627,
      "grad_norm": 3.8066141605377197,
      "learning_rate": 1.892501536570375e-05,
      "loss": 1.4949,
      "step": 3498
    },
    {
      "epoch": 0.268899815611555,
      "grad_norm": 4.257297515869141,
      "learning_rate": 1.892440073755378e-05,
      "loss": 1.5364,
      "step": 3500
    },
    {
      "epoch": 0.26905347264904733,
      "grad_norm": 3.5987801551818848,
      "learning_rate": 1.892378610940381e-05,
      "loss": 1.4022,
      "step": 3502
    },
    {
      "epoch": 0.26920712968653965,
      "grad_norm": 3.342484474182129,
      "learning_rate": 1.8923171481253844e-05,
      "loss": 1.5268,
      "step": 3504
    },
    {
      "epoch": 0.26936078672403196,
      "grad_norm": 3.992051601409912,
      "learning_rate": 1.8922556853103873e-05,
      "loss": 1.4786,
      "step": 3506
    },
    {
      "epoch": 0.2695144437615243,
      "grad_norm": 3.877612829208374,
      "learning_rate": 1.8921942224953903e-05,
      "loss": 1.5529,
      "step": 3508
    },
    {
      "epoch": 0.2696681007990166,
      "grad_norm": 3.6077497005462646,
      "learning_rate": 1.8921327596803936e-05,
      "loss": 1.577,
      "step": 3510
    },
    {
      "epoch": 0.2698217578365089,
      "grad_norm": 3.497678279876709,
      "learning_rate": 1.8920712968653966e-05,
      "loss": 1.4717,
      "step": 3512
    },
    {
      "epoch": 0.26997541487400123,
      "grad_norm": 3.7486648559570312,
      "learning_rate": 1.8920098340503995e-05,
      "loss": 1.3829,
      "step": 3514
    },
    {
      "epoch": 0.27012907191149355,
      "grad_norm": 3.9724059104919434,
      "learning_rate": 1.891948371235403e-05,
      "loss": 1.6304,
      "step": 3516
    },
    {
      "epoch": 0.27028272894898586,
      "grad_norm": 3.657163381576538,
      "learning_rate": 1.8918869084204058e-05,
      "loss": 1.5072,
      "step": 3518
    },
    {
      "epoch": 0.2704363859864782,
      "grad_norm": 3.7878036499023438,
      "learning_rate": 1.8918254456054088e-05,
      "loss": 1.5066,
      "step": 3520
    },
    {
      "epoch": 0.2705900430239705,
      "grad_norm": 3.5070736408233643,
      "learning_rate": 1.891763982790412e-05,
      "loss": 1.5437,
      "step": 3522
    },
    {
      "epoch": 0.2707437000614628,
      "grad_norm": 3.5397820472717285,
      "learning_rate": 1.891702519975415e-05,
      "loss": 1.502,
      "step": 3524
    },
    {
      "epoch": 0.2708973570989551,
      "grad_norm": 3.4602363109588623,
      "learning_rate": 1.891641057160418e-05,
      "loss": 1.4634,
      "step": 3526
    },
    {
      "epoch": 0.27105101413644744,
      "grad_norm": 3.507657051086426,
      "learning_rate": 1.891579594345421e-05,
      "loss": 1.4342,
      "step": 3528
    },
    {
      "epoch": 0.27120467117393976,
      "grad_norm": 3.7547881603240967,
      "learning_rate": 1.8915181315304243e-05,
      "loss": 1.3555,
      "step": 3530
    },
    {
      "epoch": 0.2713583282114321,
      "grad_norm": 3.522979497909546,
      "learning_rate": 1.8914566687154273e-05,
      "loss": 1.4903,
      "step": 3532
    },
    {
      "epoch": 0.2715119852489244,
      "grad_norm": 3.673581123352051,
      "learning_rate": 1.8913952059004302e-05,
      "loss": 1.3561,
      "step": 3534
    },
    {
      "epoch": 0.2716656422864167,
      "grad_norm": 3.447091579437256,
      "learning_rate": 1.8913337430854336e-05,
      "loss": 1.5173,
      "step": 3536
    },
    {
      "epoch": 0.271819299323909,
      "grad_norm": 3.915567398071289,
      "learning_rate": 1.8912722802704365e-05,
      "loss": 1.502,
      "step": 3538
    },
    {
      "epoch": 0.27197295636140134,
      "grad_norm": 3.157322883605957,
      "learning_rate": 1.8912108174554395e-05,
      "loss": 1.4469,
      "step": 3540
    },
    {
      "epoch": 0.27212661339889366,
      "grad_norm": 3.3162407875061035,
      "learning_rate": 1.8911493546404428e-05,
      "loss": 1.3842,
      "step": 3542
    },
    {
      "epoch": 0.272280270436386,
      "grad_norm": 4.137020111083984,
      "learning_rate": 1.8910878918254458e-05,
      "loss": 1.4895,
      "step": 3544
    },
    {
      "epoch": 0.2724339274738783,
      "grad_norm": 3.5766658782958984,
      "learning_rate": 1.8910264290104487e-05,
      "loss": 1.4133,
      "step": 3546
    },
    {
      "epoch": 0.2725875845113706,
      "grad_norm": 3.5843989849090576,
      "learning_rate": 1.890964966195452e-05,
      "loss": 1.5026,
      "step": 3548
    },
    {
      "epoch": 0.2727412415488629,
      "grad_norm": 3.568018913269043,
      "learning_rate": 1.890903503380455e-05,
      "loss": 1.4175,
      "step": 3550
    },
    {
      "epoch": 0.27289489858635524,
      "grad_norm": 3.131871461868286,
      "learning_rate": 1.890842040565458e-05,
      "loss": 1.348,
      "step": 3552
    },
    {
      "epoch": 0.27304855562384756,
      "grad_norm": 4.097682476043701,
      "learning_rate": 1.890780577750461e-05,
      "loss": 1.4924,
      "step": 3554
    },
    {
      "epoch": 0.27320221266133987,
      "grad_norm": 3.9119415283203125,
      "learning_rate": 1.8907191149354643e-05,
      "loss": 1.431,
      "step": 3556
    },
    {
      "epoch": 0.2733558696988322,
      "grad_norm": 3.3313307762145996,
      "learning_rate": 1.8906576521204672e-05,
      "loss": 1.5508,
      "step": 3558
    },
    {
      "epoch": 0.2735095267363245,
      "grad_norm": 3.149726152420044,
      "learning_rate": 1.8905961893054702e-05,
      "loss": 1.4358,
      "step": 3560
    },
    {
      "epoch": 0.2736631837738168,
      "grad_norm": 3.852184534072876,
      "learning_rate": 1.8905347264904735e-05,
      "loss": 1.4537,
      "step": 3562
    },
    {
      "epoch": 0.27381684081130914,
      "grad_norm": 3.5580079555511475,
      "learning_rate": 1.8904732636754765e-05,
      "loss": 1.6419,
      "step": 3564
    },
    {
      "epoch": 0.27397049784880145,
      "grad_norm": 3.484915018081665,
      "learning_rate": 1.8904118008604794e-05,
      "loss": 1.4562,
      "step": 3566
    },
    {
      "epoch": 0.27412415488629377,
      "grad_norm": 4.093445301055908,
      "learning_rate": 1.8903503380454827e-05,
      "loss": 1.4078,
      "step": 3568
    },
    {
      "epoch": 0.2742778119237861,
      "grad_norm": 3.514277458190918,
      "learning_rate": 1.8902888752304857e-05,
      "loss": 1.4331,
      "step": 3570
    },
    {
      "epoch": 0.2744314689612784,
      "grad_norm": 4.0411810874938965,
      "learning_rate": 1.8902274124154887e-05,
      "loss": 1.4517,
      "step": 3572
    },
    {
      "epoch": 0.2745851259987707,
      "grad_norm": 3.505648374557495,
      "learning_rate": 1.890165949600492e-05,
      "loss": 1.415,
      "step": 3574
    },
    {
      "epoch": 0.27473878303626303,
      "grad_norm": 4.12168550491333,
      "learning_rate": 1.890104486785495e-05,
      "loss": 1.4809,
      "step": 3576
    },
    {
      "epoch": 0.27489244007375535,
      "grad_norm": 3.5654399394989014,
      "learning_rate": 1.8900430239704983e-05,
      "loss": 1.5533,
      "step": 3578
    },
    {
      "epoch": 0.2750460971112477,
      "grad_norm": 4.086021900177002,
      "learning_rate": 1.889981561155501e-05,
      "loss": 1.5845,
      "step": 3580
    },
    {
      "epoch": 0.27519975414874004,
      "grad_norm": 3.6145718097686768,
      "learning_rate": 1.8899200983405042e-05,
      "loss": 1.3608,
      "step": 3582
    },
    {
      "epoch": 0.27535341118623236,
      "grad_norm": 3.477968692779541,
      "learning_rate": 1.889858635525507e-05,
      "loss": 1.4685,
      "step": 3584
    },
    {
      "epoch": 0.27550706822372467,
      "grad_norm": 3.627117872238159,
      "learning_rate": 1.88979717271051e-05,
      "loss": 1.4376,
      "step": 3586
    },
    {
      "epoch": 0.275660725261217,
      "grad_norm": 3.631131649017334,
      "learning_rate": 1.8897357098955134e-05,
      "loss": 1.5543,
      "step": 3588
    },
    {
      "epoch": 0.2758143822987093,
      "grad_norm": 3.5396790504455566,
      "learning_rate": 1.8896742470805164e-05,
      "loss": 1.4909,
      "step": 3590
    },
    {
      "epoch": 0.2759680393362016,
      "grad_norm": 3.816535472869873,
      "learning_rate": 1.8896127842655194e-05,
      "loss": 1.399,
      "step": 3592
    },
    {
      "epoch": 0.27612169637369394,
      "grad_norm": 3.565800428390503,
      "learning_rate": 1.8895513214505227e-05,
      "loss": 1.2989,
      "step": 3594
    },
    {
      "epoch": 0.27627535341118625,
      "grad_norm": 4.057579040527344,
      "learning_rate": 1.8894898586355257e-05,
      "loss": 1.484,
      "step": 3596
    },
    {
      "epoch": 0.27642901044867857,
      "grad_norm": 3.517569065093994,
      "learning_rate": 1.889428395820529e-05,
      "loss": 1.5513,
      "step": 3598
    },
    {
      "epoch": 0.2765826674861709,
      "grad_norm": 3.6325526237487793,
      "learning_rate": 1.8893669330055316e-05,
      "loss": 1.4866,
      "step": 3600
    },
    {
      "epoch": 0.2767363245236632,
      "grad_norm": 3.697222948074341,
      "learning_rate": 1.889305470190535e-05,
      "loss": 1.5287,
      "step": 3602
    },
    {
      "epoch": 0.2768899815611555,
      "grad_norm": 3.3492066860198975,
      "learning_rate": 1.8892440073755382e-05,
      "loss": 1.5102,
      "step": 3604
    },
    {
      "epoch": 0.27704363859864783,
      "grad_norm": 3.629889726638794,
      "learning_rate": 1.889182544560541e-05,
      "loss": 1.5843,
      "step": 3606
    },
    {
      "epoch": 0.27719729563614015,
      "grad_norm": 4.252458095550537,
      "learning_rate": 1.889121081745544e-05,
      "loss": 1.4569,
      "step": 3608
    },
    {
      "epoch": 0.27735095267363247,
      "grad_norm": 3.4219071865081787,
      "learning_rate": 1.889059618930547e-05,
      "loss": 1.5789,
      "step": 3610
    },
    {
      "epoch": 0.2775046097111248,
      "grad_norm": 3.4524829387664795,
      "learning_rate": 1.88899815611555e-05,
      "loss": 1.5203,
      "step": 3612
    },
    {
      "epoch": 0.2776582667486171,
      "grad_norm": 3.130014181137085,
      "learning_rate": 1.8889366933005534e-05,
      "loss": 1.3825,
      "step": 3614
    },
    {
      "epoch": 0.2778119237861094,
      "grad_norm": 3.439162015914917,
      "learning_rate": 1.8888752304855564e-05,
      "loss": 1.3977,
      "step": 3616
    },
    {
      "epoch": 0.27796558082360173,
      "grad_norm": 3.3094050884246826,
      "learning_rate": 1.8888137676705597e-05,
      "loss": 1.4268,
      "step": 3618
    },
    {
      "epoch": 0.27811923786109405,
      "grad_norm": 3.6266913414001465,
      "learning_rate": 1.8887523048555626e-05,
      "loss": 1.7052,
      "step": 3620
    },
    {
      "epoch": 0.27827289489858636,
      "grad_norm": 3.5086283683776855,
      "learning_rate": 1.8886908420405656e-05,
      "loss": 1.4714,
      "step": 3622
    },
    {
      "epoch": 0.2784265519360787,
      "grad_norm": 3.7588820457458496,
      "learning_rate": 1.888629379225569e-05,
      "loss": 1.414,
      "step": 3624
    },
    {
      "epoch": 0.278580208973571,
      "grad_norm": 3.1712088584899902,
      "learning_rate": 1.8885679164105715e-05,
      "loss": 1.4258,
      "step": 3626
    },
    {
      "epoch": 0.2787338660110633,
      "grad_norm": 3.323568820953369,
      "learning_rate": 1.888506453595575e-05,
      "loss": 1.5158,
      "step": 3628
    },
    {
      "epoch": 0.27888752304855563,
      "grad_norm": 3.520214557647705,
      "learning_rate": 1.8884449907805778e-05,
      "loss": 1.5343,
      "step": 3630
    },
    {
      "epoch": 0.27904118008604795,
      "grad_norm": 3.6342597007751465,
      "learning_rate": 1.8883835279655808e-05,
      "loss": 1.5991,
      "step": 3632
    },
    {
      "epoch": 0.27919483712354026,
      "grad_norm": 4.191102027893066,
      "learning_rate": 1.888322065150584e-05,
      "loss": 1.4219,
      "step": 3634
    },
    {
      "epoch": 0.2793484941610326,
      "grad_norm": 4.2305779457092285,
      "learning_rate": 1.888260602335587e-05,
      "loss": 1.6116,
      "step": 3636
    },
    {
      "epoch": 0.2795021511985249,
      "grad_norm": 3.5002593994140625,
      "learning_rate": 1.8881991395205904e-05,
      "loss": 1.4798,
      "step": 3638
    },
    {
      "epoch": 0.2796558082360172,
      "grad_norm": 2.9989280700683594,
      "learning_rate": 1.8881376767055933e-05,
      "loss": 1.2869,
      "step": 3640
    },
    {
      "epoch": 0.27980946527350953,
      "grad_norm": 3.6314494609832764,
      "learning_rate": 1.8880762138905963e-05,
      "loss": 1.3329,
      "step": 3642
    },
    {
      "epoch": 0.27996312231100184,
      "grad_norm": 3.1968257427215576,
      "learning_rate": 1.8880147510755996e-05,
      "loss": 1.4851,
      "step": 3644
    },
    {
      "epoch": 0.28011677934849416,
      "grad_norm": 3.2353439331054688,
      "learning_rate": 1.8879532882606026e-05,
      "loss": 1.5025,
      "step": 3646
    },
    {
      "epoch": 0.2802704363859865,
      "grad_norm": 3.394658327102661,
      "learning_rate": 1.8878918254456055e-05,
      "loss": 1.4325,
      "step": 3648
    },
    {
      "epoch": 0.2804240934234788,
      "grad_norm": 3.303316831588745,
      "learning_rate": 1.887830362630609e-05,
      "loss": 1.482,
      "step": 3650
    },
    {
      "epoch": 0.2805777504609711,
      "grad_norm": 3.44572114944458,
      "learning_rate": 1.8877688998156115e-05,
      "loss": 1.4126,
      "step": 3652
    },
    {
      "epoch": 0.2807314074984634,
      "grad_norm": 3.64277982711792,
      "learning_rate": 1.8877074370006148e-05,
      "loss": 1.4845,
      "step": 3654
    },
    {
      "epoch": 0.28088506453595574,
      "grad_norm": 3.9562630653381348,
      "learning_rate": 1.8876459741856178e-05,
      "loss": 1.4596,
      "step": 3656
    },
    {
      "epoch": 0.28103872157344806,
      "grad_norm": 3.3409039974212646,
      "learning_rate": 1.8875845113706207e-05,
      "loss": 1.4672,
      "step": 3658
    },
    {
      "epoch": 0.2811923786109404,
      "grad_norm": 3.783766031265259,
      "learning_rate": 1.887523048555624e-05,
      "loss": 1.662,
      "step": 3660
    },
    {
      "epoch": 0.2813460356484327,
      "grad_norm": 4.031844139099121,
      "learning_rate": 1.887461585740627e-05,
      "loss": 1.5427,
      "step": 3662
    },
    {
      "epoch": 0.281499692685925,
      "grad_norm": 3.9957854747772217,
      "learning_rate": 1.8874001229256303e-05,
      "loss": 1.505,
      "step": 3664
    },
    {
      "epoch": 0.2816533497234173,
      "grad_norm": 3.1964595317840576,
      "learning_rate": 1.8873386601106333e-05,
      "loss": 1.3482,
      "step": 3666
    },
    {
      "epoch": 0.28180700676090964,
      "grad_norm": 4.197731018066406,
      "learning_rate": 1.8872771972956362e-05,
      "loss": 1.4717,
      "step": 3668
    },
    {
      "epoch": 0.28196066379840196,
      "grad_norm": 3.6347153186798096,
      "learning_rate": 1.8872157344806395e-05,
      "loss": 1.5804,
      "step": 3670
    },
    {
      "epoch": 0.28211432083589427,
      "grad_norm": 3.2941391468048096,
      "learning_rate": 1.8871542716656425e-05,
      "loss": 1.4578,
      "step": 3672
    },
    {
      "epoch": 0.2822679778733866,
      "grad_norm": 3.4472978115081787,
      "learning_rate": 1.8870928088506455e-05,
      "loss": 1.445,
      "step": 3674
    },
    {
      "epoch": 0.2824216349108789,
      "grad_norm": 3.385477304458618,
      "learning_rate": 1.8870313460356488e-05,
      "loss": 1.3812,
      "step": 3676
    },
    {
      "epoch": 0.2825752919483712,
      "grad_norm": 3.4385907649993896,
      "learning_rate": 1.8869698832206514e-05,
      "loss": 1.4371,
      "step": 3678
    },
    {
      "epoch": 0.28272894898586354,
      "grad_norm": 3.3431503772735596,
      "learning_rate": 1.8869084204056547e-05,
      "loss": 1.4938,
      "step": 3680
    },
    {
      "epoch": 0.28288260602335585,
      "grad_norm": 3.215934991836548,
      "learning_rate": 1.8868469575906577e-05,
      "loss": 1.4857,
      "step": 3682
    },
    {
      "epoch": 0.28303626306084817,
      "grad_norm": 3.9575302600860596,
      "learning_rate": 1.886785494775661e-05,
      "loss": 1.4792,
      "step": 3684
    },
    {
      "epoch": 0.2831899200983405,
      "grad_norm": 2.991589307785034,
      "learning_rate": 1.886724031960664e-05,
      "loss": 1.3701,
      "step": 3686
    },
    {
      "epoch": 0.2833435771358328,
      "grad_norm": 3.557347536087036,
      "learning_rate": 1.886662569145667e-05,
      "loss": 1.4585,
      "step": 3688
    },
    {
      "epoch": 0.2834972341733251,
      "grad_norm": 3.5175206661224365,
      "learning_rate": 1.8866011063306702e-05,
      "loss": 1.6076,
      "step": 3690
    },
    {
      "epoch": 0.28365089121081744,
      "grad_norm": 4.530017375946045,
      "learning_rate": 1.8865396435156732e-05,
      "loss": 1.6293,
      "step": 3692
    },
    {
      "epoch": 0.28380454824830975,
      "grad_norm": 4.1233320236206055,
      "learning_rate": 1.8864781807006762e-05,
      "loss": 1.3835,
      "step": 3694
    },
    {
      "epoch": 0.28395820528580207,
      "grad_norm": 3.4351844787597656,
      "learning_rate": 1.8864167178856795e-05,
      "loss": 1.3736,
      "step": 3696
    },
    {
      "epoch": 0.2841118623232944,
      "grad_norm": 3.3016061782836914,
      "learning_rate": 1.8863552550706825e-05,
      "loss": 1.5559,
      "step": 3698
    },
    {
      "epoch": 0.2842655193607867,
      "grad_norm": 4.053586006164551,
      "learning_rate": 1.8862937922556854e-05,
      "loss": 1.492,
      "step": 3700
    },
    {
      "epoch": 0.284419176398279,
      "grad_norm": 3.7303128242492676,
      "learning_rate": 1.8862323294406887e-05,
      "loss": 1.368,
      "step": 3702
    },
    {
      "epoch": 0.28457283343577133,
      "grad_norm": 3.206644296646118,
      "learning_rate": 1.8861708666256917e-05,
      "loss": 1.405,
      "step": 3704
    },
    {
      "epoch": 0.28472649047326365,
      "grad_norm": 3.58850359916687,
      "learning_rate": 1.8861094038106947e-05,
      "loss": 1.3416,
      "step": 3706
    },
    {
      "epoch": 0.28488014751075597,
      "grad_norm": 4.489683628082275,
      "learning_rate": 1.8860479409956976e-05,
      "loss": 1.315,
      "step": 3708
    },
    {
      "epoch": 0.28503380454824834,
      "grad_norm": 3.5646450519561768,
      "learning_rate": 1.885986478180701e-05,
      "loss": 1.4451,
      "step": 3710
    },
    {
      "epoch": 0.28518746158574065,
      "grad_norm": 3.7019946575164795,
      "learning_rate": 1.885925015365704e-05,
      "loss": 1.4403,
      "step": 3712
    },
    {
      "epoch": 0.28534111862323297,
      "grad_norm": 3.1615312099456787,
      "learning_rate": 1.885863552550707e-05,
      "loss": 1.4083,
      "step": 3714
    },
    {
      "epoch": 0.2854947756607253,
      "grad_norm": 3.486154556274414,
      "learning_rate": 1.8858020897357102e-05,
      "loss": 1.46,
      "step": 3716
    },
    {
      "epoch": 0.2856484326982176,
      "grad_norm": 4.051648139953613,
      "learning_rate": 1.885740626920713e-05,
      "loss": 1.4501,
      "step": 3718
    },
    {
      "epoch": 0.2858020897357099,
      "grad_norm": 3.6418347358703613,
      "learning_rate": 1.885679164105716e-05,
      "loss": 1.4339,
      "step": 3720
    },
    {
      "epoch": 0.28595574677320224,
      "grad_norm": 3.5506417751312256,
      "learning_rate": 1.8856177012907194e-05,
      "loss": 1.4697,
      "step": 3722
    },
    {
      "epoch": 0.28610940381069455,
      "grad_norm": 3.2061004638671875,
      "learning_rate": 1.8855562384757224e-05,
      "loss": 1.4207,
      "step": 3724
    },
    {
      "epoch": 0.28626306084818687,
      "grad_norm": 3.7919108867645264,
      "learning_rate": 1.8854947756607254e-05,
      "loss": 1.4708,
      "step": 3726
    },
    {
      "epoch": 0.2864167178856792,
      "grad_norm": 3.9891295433044434,
      "learning_rate": 1.8854333128457283e-05,
      "loss": 1.4744,
      "step": 3728
    },
    {
      "epoch": 0.2865703749231715,
      "grad_norm": 3.63608455657959,
      "learning_rate": 1.8853718500307316e-05,
      "loss": 1.5032,
      "step": 3730
    },
    {
      "epoch": 0.2867240319606638,
      "grad_norm": 3.5813047885894775,
      "learning_rate": 1.8853103872157346e-05,
      "loss": 1.581,
      "step": 3732
    },
    {
      "epoch": 0.28687768899815613,
      "grad_norm": 2.9888648986816406,
      "learning_rate": 1.8852489244007376e-05,
      "loss": 1.5092,
      "step": 3734
    },
    {
      "epoch": 0.28703134603564845,
      "grad_norm": 3.926024913787842,
      "learning_rate": 1.885187461585741e-05,
      "loss": 1.6346,
      "step": 3736
    },
    {
      "epoch": 0.28718500307314077,
      "grad_norm": 3.657963275909424,
      "learning_rate": 1.885125998770744e-05,
      "loss": 1.4397,
      "step": 3738
    },
    {
      "epoch": 0.2873386601106331,
      "grad_norm": 3.7995994091033936,
      "learning_rate": 1.8850645359557468e-05,
      "loss": 1.5096,
      "step": 3740
    },
    {
      "epoch": 0.2874923171481254,
      "grad_norm": 3.9522006511688232,
      "learning_rate": 1.88500307314075e-05,
      "loss": 1.3611,
      "step": 3742
    },
    {
      "epoch": 0.2876459741856177,
      "grad_norm": 4.043625831604004,
      "learning_rate": 1.884941610325753e-05,
      "loss": 1.5588,
      "step": 3744
    },
    {
      "epoch": 0.28779963122311003,
      "grad_norm": 3.6213269233703613,
      "learning_rate": 1.884880147510756e-05,
      "loss": 1.6376,
      "step": 3746
    },
    {
      "epoch": 0.28795328826060235,
      "grad_norm": 3.4300777912139893,
      "learning_rate": 1.8848186846957594e-05,
      "loss": 1.6227,
      "step": 3748
    },
    {
      "epoch": 0.28810694529809466,
      "grad_norm": 3.5119426250457764,
      "learning_rate": 1.8847572218807623e-05,
      "loss": 1.479,
      "step": 3750
    },
    {
      "epoch": 0.288260602335587,
      "grad_norm": 3.7524571418762207,
      "learning_rate": 1.8846957590657653e-05,
      "loss": 1.4244,
      "step": 3752
    },
    {
      "epoch": 0.2884142593730793,
      "grad_norm": 2.923687219619751,
      "learning_rate": 1.8846342962507683e-05,
      "loss": 1.4444,
      "step": 3754
    },
    {
      "epoch": 0.2885679164105716,
      "grad_norm": 4.0984296798706055,
      "learning_rate": 1.8845728334357716e-05,
      "loss": 1.5281,
      "step": 3756
    },
    {
      "epoch": 0.28872157344806393,
      "grad_norm": 3.296266794204712,
      "learning_rate": 1.8845113706207746e-05,
      "loss": 1.4179,
      "step": 3758
    },
    {
      "epoch": 0.28887523048555624,
      "grad_norm": 3.4929823875427246,
      "learning_rate": 1.8844499078057775e-05,
      "loss": 1.5606,
      "step": 3760
    },
    {
      "epoch": 0.28902888752304856,
      "grad_norm": 3.201611042022705,
      "learning_rate": 1.884388444990781e-05,
      "loss": 1.4212,
      "step": 3762
    },
    {
      "epoch": 0.2891825445605409,
      "grad_norm": 3.264183282852173,
      "learning_rate": 1.8843269821757838e-05,
      "loss": 1.3158,
      "step": 3764
    },
    {
      "epoch": 0.2893362015980332,
      "grad_norm": 2.94294810295105,
      "learning_rate": 1.8842655193607868e-05,
      "loss": 1.4594,
      "step": 3766
    },
    {
      "epoch": 0.2894898586355255,
      "grad_norm": 18.42613410949707,
      "learning_rate": 1.88420405654579e-05,
      "loss": 1.4853,
      "step": 3768
    },
    {
      "epoch": 0.2896435156730178,
      "grad_norm": 3.516327381134033,
      "learning_rate": 1.884142593730793e-05,
      "loss": 1.4182,
      "step": 3770
    },
    {
      "epoch": 0.28979717271051014,
      "grad_norm": 3.5198872089385986,
      "learning_rate": 1.884081130915796e-05,
      "loss": 1.3558,
      "step": 3772
    },
    {
      "epoch": 0.28995082974800246,
      "grad_norm": 3.396672487258911,
      "learning_rate": 1.8840196681007993e-05,
      "loss": 1.292,
      "step": 3774
    },
    {
      "epoch": 0.2901044867854948,
      "grad_norm": 3.304868698120117,
      "learning_rate": 1.8839582052858023e-05,
      "loss": 1.5348,
      "step": 3776
    },
    {
      "epoch": 0.2902581438229871,
      "grad_norm": 3.4228885173797607,
      "learning_rate": 1.8838967424708053e-05,
      "loss": 1.4141,
      "step": 3778
    },
    {
      "epoch": 0.2904118008604794,
      "grad_norm": 3.2031636238098145,
      "learning_rate": 1.8838352796558082e-05,
      "loss": 1.3834,
      "step": 3780
    },
    {
      "epoch": 0.2905654578979717,
      "grad_norm": 3.6728415489196777,
      "learning_rate": 1.8837738168408115e-05,
      "loss": 1.3677,
      "step": 3782
    },
    {
      "epoch": 0.29071911493546404,
      "grad_norm": 3.418509006500244,
      "learning_rate": 1.8837123540258145e-05,
      "loss": 1.5158,
      "step": 3784
    },
    {
      "epoch": 0.29087277197295636,
      "grad_norm": 3.4060676097869873,
      "learning_rate": 1.8836508912108175e-05,
      "loss": 1.561,
      "step": 3786
    },
    {
      "epoch": 0.2910264290104487,
      "grad_norm": 3.5852086544036865,
      "learning_rate": 1.8835894283958208e-05,
      "loss": 1.4347,
      "step": 3788
    },
    {
      "epoch": 0.291180086047941,
      "grad_norm": 3.987565517425537,
      "learning_rate": 1.8835279655808237e-05,
      "loss": 1.4233,
      "step": 3790
    },
    {
      "epoch": 0.2913337430854333,
      "grad_norm": 3.6521356105804443,
      "learning_rate": 1.8834665027658267e-05,
      "loss": 1.4408,
      "step": 3792
    },
    {
      "epoch": 0.2914874001229256,
      "grad_norm": 5.208387851715088,
      "learning_rate": 1.88340503995083e-05,
      "loss": 1.6046,
      "step": 3794
    },
    {
      "epoch": 0.29164105716041794,
      "grad_norm": 3.065742254257202,
      "learning_rate": 1.883343577135833e-05,
      "loss": 1.4079,
      "step": 3796
    },
    {
      "epoch": 0.29179471419791025,
      "grad_norm": 3.513312816619873,
      "learning_rate": 1.883282114320836e-05,
      "loss": 1.547,
      "step": 3798
    },
    {
      "epoch": 0.29194837123540257,
      "grad_norm": 2.8478775024414062,
      "learning_rate": 1.8832206515058393e-05,
      "loss": 1.3478,
      "step": 3800
    },
    {
      "epoch": 0.2921020282728949,
      "grad_norm": 3.3886022567749023,
      "learning_rate": 1.8831591886908422e-05,
      "loss": 1.3928,
      "step": 3802
    },
    {
      "epoch": 0.2922556853103872,
      "grad_norm": 3.5944290161132812,
      "learning_rate": 1.8830977258758452e-05,
      "loss": 1.3548,
      "step": 3804
    },
    {
      "epoch": 0.2924093423478795,
      "grad_norm": 4.22208309173584,
      "learning_rate": 1.8830362630608482e-05,
      "loss": 1.5221,
      "step": 3806
    },
    {
      "epoch": 0.29256299938537184,
      "grad_norm": 3.431133985519409,
      "learning_rate": 1.8829748002458515e-05,
      "loss": 1.5649,
      "step": 3808
    },
    {
      "epoch": 0.29271665642286415,
      "grad_norm": 3.4773547649383545,
      "learning_rate": 1.8829133374308544e-05,
      "loss": 1.3719,
      "step": 3810
    },
    {
      "epoch": 0.29287031346035647,
      "grad_norm": 3.4864048957824707,
      "learning_rate": 1.8828518746158574e-05,
      "loss": 1.3556,
      "step": 3812
    },
    {
      "epoch": 0.2930239704978488,
      "grad_norm": 3.3974108695983887,
      "learning_rate": 1.8827904118008607e-05,
      "loss": 1.6966,
      "step": 3814
    },
    {
      "epoch": 0.2931776275353411,
      "grad_norm": 3.71232533454895,
      "learning_rate": 1.8827289489858637e-05,
      "loss": 1.5968,
      "step": 3816
    },
    {
      "epoch": 0.2933312845728334,
      "grad_norm": 3.693206787109375,
      "learning_rate": 1.8826674861708667e-05,
      "loss": 1.5184,
      "step": 3818
    },
    {
      "epoch": 0.29348494161032573,
      "grad_norm": 3.710134744644165,
      "learning_rate": 1.88260602335587e-05,
      "loss": 1.5327,
      "step": 3820
    },
    {
      "epoch": 0.29363859864781805,
      "grad_norm": 3.417041540145874,
      "learning_rate": 1.882544560540873e-05,
      "loss": 1.5009,
      "step": 3822
    },
    {
      "epoch": 0.29379225568531037,
      "grad_norm": 3.0500283241271973,
      "learning_rate": 1.882483097725876e-05,
      "loss": 1.5554,
      "step": 3824
    },
    {
      "epoch": 0.2939459127228027,
      "grad_norm": 3.6560235023498535,
      "learning_rate": 1.882421634910879e-05,
      "loss": 1.4728,
      "step": 3826
    },
    {
      "epoch": 0.294099569760295,
      "grad_norm": 3.733311653137207,
      "learning_rate": 1.8823601720958822e-05,
      "loss": 1.291,
      "step": 3828
    },
    {
      "epoch": 0.2942532267977873,
      "grad_norm": 3.633963108062744,
      "learning_rate": 1.8822987092808855e-05,
      "loss": 1.6095,
      "step": 3830
    },
    {
      "epoch": 0.29440688383527963,
      "grad_norm": 3.664761781692505,
      "learning_rate": 1.882237246465888e-05,
      "loss": 1.3628,
      "step": 3832
    },
    {
      "epoch": 0.29456054087277195,
      "grad_norm": 4.00673246383667,
      "learning_rate": 1.8821757836508914e-05,
      "loss": 1.5943,
      "step": 3834
    },
    {
      "epoch": 0.29471419791026426,
      "grad_norm": 3.2937426567077637,
      "learning_rate": 1.8821143208358944e-05,
      "loss": 1.341,
      "step": 3836
    },
    {
      "epoch": 0.2948678549477566,
      "grad_norm": 2.9937968254089355,
      "learning_rate": 1.8820528580208974e-05,
      "loss": 1.5026,
      "step": 3838
    },
    {
      "epoch": 0.29502151198524895,
      "grad_norm": 3.309907913208008,
      "learning_rate": 1.8819913952059007e-05,
      "loss": 1.4908,
      "step": 3840
    },
    {
      "epoch": 0.29517516902274127,
      "grad_norm": 3.4216926097869873,
      "learning_rate": 1.8819299323909036e-05,
      "loss": 1.4074,
      "step": 3842
    },
    {
      "epoch": 0.2953288260602336,
      "grad_norm": 3.285968780517578,
      "learning_rate": 1.8818684695759066e-05,
      "loss": 1.4249,
      "step": 3844
    },
    {
      "epoch": 0.2954824830977259,
      "grad_norm": 3.103705644607544,
      "learning_rate": 1.88180700676091e-05,
      "loss": 1.3529,
      "step": 3846
    },
    {
      "epoch": 0.2956361401352182,
      "grad_norm": 3.3987631797790527,
      "learning_rate": 1.881745543945913e-05,
      "loss": 1.5761,
      "step": 3848
    },
    {
      "epoch": 0.29578979717271053,
      "grad_norm": 3.2349257469177246,
      "learning_rate": 1.8816840811309162e-05,
      "loss": 1.5087,
      "step": 3850
    },
    {
      "epoch": 0.29594345421020285,
      "grad_norm": 3.374030113220215,
      "learning_rate": 1.8816226183159188e-05,
      "loss": 1.4778,
      "step": 3852
    },
    {
      "epoch": 0.29609711124769517,
      "grad_norm": 3.8146629333496094,
      "learning_rate": 1.881561155500922e-05,
      "loss": 1.4556,
      "step": 3854
    },
    {
      "epoch": 0.2962507682851875,
      "grad_norm": 3.6150548458099365,
      "learning_rate": 1.881499692685925e-05,
      "loss": 1.2892,
      "step": 3856
    },
    {
      "epoch": 0.2964044253226798,
      "grad_norm": 3.345280170440674,
      "learning_rate": 1.881438229870928e-05,
      "loss": 1.4552,
      "step": 3858
    },
    {
      "epoch": 0.2965580823601721,
      "grad_norm": 2.790616273880005,
      "learning_rate": 1.8813767670559314e-05,
      "loss": 1.4375,
      "step": 3860
    },
    {
      "epoch": 0.29671173939766443,
      "grad_norm": 3.415255546569824,
      "learning_rate": 1.8813153042409343e-05,
      "loss": 1.5357,
      "step": 3862
    },
    {
      "epoch": 0.29686539643515675,
      "grad_norm": 3.4177236557006836,
      "learning_rate": 1.8812538414259373e-05,
      "loss": 1.6121,
      "step": 3864
    },
    {
      "epoch": 0.29701905347264906,
      "grad_norm": 3.5526952743530273,
      "learning_rate": 1.8811923786109406e-05,
      "loss": 1.4373,
      "step": 3866
    },
    {
      "epoch": 0.2971727105101414,
      "grad_norm": 3.729252576828003,
      "learning_rate": 1.8811309157959436e-05,
      "loss": 1.3217,
      "step": 3868
    },
    {
      "epoch": 0.2973263675476337,
      "grad_norm": 3.5976545810699463,
      "learning_rate": 1.881069452980947e-05,
      "loss": 1.4549,
      "step": 3870
    },
    {
      "epoch": 0.297480024585126,
      "grad_norm": 3.2108335494995117,
      "learning_rate": 1.88100799016595e-05,
      "loss": 1.5147,
      "step": 3872
    },
    {
      "epoch": 0.29763368162261833,
      "grad_norm": 3.279672622680664,
      "learning_rate": 1.8809465273509528e-05,
      "loss": 1.3839,
      "step": 3874
    },
    {
      "epoch": 0.29778733866011065,
      "grad_norm": 3.832663059234619,
      "learning_rate": 1.880885064535956e-05,
      "loss": 1.4755,
      "step": 3876
    },
    {
      "epoch": 0.29794099569760296,
      "grad_norm": 3.84242844581604,
      "learning_rate": 1.8808236017209588e-05,
      "loss": 1.4298,
      "step": 3878
    },
    {
      "epoch": 0.2980946527350953,
      "grad_norm": 3.9651095867156982,
      "learning_rate": 1.880762138905962e-05,
      "loss": 1.4646,
      "step": 3880
    },
    {
      "epoch": 0.2982483097725876,
      "grad_norm": 3.306433916091919,
      "learning_rate": 1.880700676090965e-05,
      "loss": 1.44,
      "step": 3882
    },
    {
      "epoch": 0.2984019668100799,
      "grad_norm": 3.3314905166625977,
      "learning_rate": 1.880639213275968e-05,
      "loss": 1.6509,
      "step": 3884
    },
    {
      "epoch": 0.2985556238475722,
      "grad_norm": 3.340235471725464,
      "learning_rate": 1.8805777504609713e-05,
      "loss": 1.4151,
      "step": 3886
    },
    {
      "epoch": 0.29870928088506454,
      "grad_norm": 4.01276159286499,
      "learning_rate": 1.8805162876459743e-05,
      "loss": 1.4507,
      "step": 3888
    },
    {
      "epoch": 0.29886293792255686,
      "grad_norm": 3.1373071670532227,
      "learning_rate": 1.8804548248309776e-05,
      "loss": 1.3502,
      "step": 3890
    },
    {
      "epoch": 0.2990165949600492,
      "grad_norm": 3.7189245223999023,
      "learning_rate": 1.8803933620159806e-05,
      "loss": 1.5203,
      "step": 3892
    },
    {
      "epoch": 0.2991702519975415,
      "grad_norm": 3.4619784355163574,
      "learning_rate": 1.8803318992009835e-05,
      "loss": 1.4551,
      "step": 3894
    },
    {
      "epoch": 0.2993239090350338,
      "grad_norm": 4.235640048980713,
      "learning_rate": 1.8802704363859868e-05,
      "loss": 1.4857,
      "step": 3896
    },
    {
      "epoch": 0.2994775660725261,
      "grad_norm": 3.1417219638824463,
      "learning_rate": 1.8802089735709898e-05,
      "loss": 1.4426,
      "step": 3898
    },
    {
      "epoch": 0.29963122311001844,
      "grad_norm": 3.5244154930114746,
      "learning_rate": 1.8801475107559928e-05,
      "loss": 1.4637,
      "step": 3900
    },
    {
      "epoch": 0.29978488014751076,
      "grad_norm": 3.3270487785339355,
      "learning_rate": 1.880086047940996e-05,
      "loss": 1.2798,
      "step": 3902
    },
    {
      "epoch": 0.2999385371850031,
      "grad_norm": 4.055757522583008,
      "learning_rate": 1.8800245851259987e-05,
      "loss": 1.4502,
      "step": 3904
    },
    {
      "epoch": 0.3000921942224954,
      "grad_norm": 3.4112069606781006,
      "learning_rate": 1.879963122311002e-05,
      "loss": 1.4785,
      "step": 3906
    },
    {
      "epoch": 0.3002458512599877,
      "grad_norm": 3.5137009620666504,
      "learning_rate": 1.879901659496005e-05,
      "loss": 1.4411,
      "step": 3908
    },
    {
      "epoch": 0.30039950829748,
      "grad_norm": 3.7082371711730957,
      "learning_rate": 1.879840196681008e-05,
      "loss": 1.4781,
      "step": 3910
    },
    {
      "epoch": 0.30055316533497234,
      "grad_norm": 3.384655237197876,
      "learning_rate": 1.8797787338660113e-05,
      "loss": 1.4317,
      "step": 3912
    },
    {
      "epoch": 0.30070682237246465,
      "grad_norm": 3.7674429416656494,
      "learning_rate": 1.8797172710510142e-05,
      "loss": 1.4593,
      "step": 3914
    },
    {
      "epoch": 0.30086047940995697,
      "grad_norm": 3.340071439743042,
      "learning_rate": 1.8796558082360175e-05,
      "loss": 1.5556,
      "step": 3916
    },
    {
      "epoch": 0.3010141364474493,
      "grad_norm": 4.023772239685059,
      "learning_rate": 1.8795943454210205e-05,
      "loss": 1.4398,
      "step": 3918
    },
    {
      "epoch": 0.3011677934849416,
      "grad_norm": 3.8660881519317627,
      "learning_rate": 1.8795328826060235e-05,
      "loss": 1.5777,
      "step": 3920
    },
    {
      "epoch": 0.3013214505224339,
      "grad_norm": 3.843226909637451,
      "learning_rate": 1.8794714197910268e-05,
      "loss": 1.5498,
      "step": 3922
    },
    {
      "epoch": 0.30147510755992624,
      "grad_norm": 3.1243956089019775,
      "learning_rate": 1.8794099569760294e-05,
      "loss": 1.4583,
      "step": 3924
    },
    {
      "epoch": 0.30162876459741855,
      "grad_norm": 5.0555219650268555,
      "learning_rate": 1.8793484941610327e-05,
      "loss": 1.5479,
      "step": 3926
    },
    {
      "epoch": 0.30178242163491087,
      "grad_norm": 3.7559964656829834,
      "learning_rate": 1.879287031346036e-05,
      "loss": 1.3887,
      "step": 3928
    },
    {
      "epoch": 0.3019360786724032,
      "grad_norm": 4.365975856781006,
      "learning_rate": 1.8792255685310386e-05,
      "loss": 1.5871,
      "step": 3930
    },
    {
      "epoch": 0.3020897357098955,
      "grad_norm": 3.261347770690918,
      "learning_rate": 1.879164105716042e-05,
      "loss": 1.3479,
      "step": 3932
    },
    {
      "epoch": 0.3022433927473878,
      "grad_norm": 3.1579368114471436,
      "learning_rate": 1.879102642901045e-05,
      "loss": 1.5464,
      "step": 3934
    },
    {
      "epoch": 0.30239704978488013,
      "grad_norm": 3.574812173843384,
      "learning_rate": 1.8790411800860482e-05,
      "loss": 1.4928,
      "step": 3936
    },
    {
      "epoch": 0.30255070682237245,
      "grad_norm": 3.7639217376708984,
      "learning_rate": 1.8789797172710512e-05,
      "loss": 1.4009,
      "step": 3938
    },
    {
      "epoch": 0.30270436385986477,
      "grad_norm": 3.222525119781494,
      "learning_rate": 1.878918254456054e-05,
      "loss": 1.3771,
      "step": 3940
    },
    {
      "epoch": 0.3028580208973571,
      "grad_norm": 3.6078226566314697,
      "learning_rate": 1.8788567916410575e-05,
      "loss": 1.4167,
      "step": 3942
    },
    {
      "epoch": 0.3030116779348494,
      "grad_norm": 3.52523136138916,
      "learning_rate": 1.8787953288260604e-05,
      "loss": 1.5286,
      "step": 3944
    },
    {
      "epoch": 0.3031653349723417,
      "grad_norm": 3.7852256298065186,
      "learning_rate": 1.8787338660110634e-05,
      "loss": 1.5492,
      "step": 3946
    },
    {
      "epoch": 0.30331899200983403,
      "grad_norm": 3.428528308868408,
      "learning_rate": 1.8786724031960667e-05,
      "loss": 1.5489,
      "step": 3948
    },
    {
      "epoch": 0.30347264904732635,
      "grad_norm": 3.4774372577667236,
      "learning_rate": 1.8786109403810693e-05,
      "loss": 1.4985,
      "step": 3950
    },
    {
      "epoch": 0.30362630608481866,
      "grad_norm": 3.802961587905884,
      "learning_rate": 1.8785494775660727e-05,
      "loss": 1.4524,
      "step": 3952
    },
    {
      "epoch": 0.303779963122311,
      "grad_norm": 3.172055959701538,
      "learning_rate": 1.8784880147510756e-05,
      "loss": 1.519,
      "step": 3954
    },
    {
      "epoch": 0.3039336201598033,
      "grad_norm": 3.412998676300049,
      "learning_rate": 1.878426551936079e-05,
      "loss": 1.2872,
      "step": 3956
    },
    {
      "epoch": 0.3040872771972956,
      "grad_norm": 3.683685302734375,
      "learning_rate": 1.878365089121082e-05,
      "loss": 1.5138,
      "step": 3958
    },
    {
      "epoch": 0.30424093423478793,
      "grad_norm": 3.2270660400390625,
      "learning_rate": 1.878303626306085e-05,
      "loss": 1.4364,
      "step": 3960
    },
    {
      "epoch": 0.30439459127228025,
      "grad_norm": 3.330509901046753,
      "learning_rate": 1.8782421634910882e-05,
      "loss": 1.5553,
      "step": 3962
    },
    {
      "epoch": 0.30454824830977256,
      "grad_norm": 4.157684803009033,
      "learning_rate": 1.878180700676091e-05,
      "loss": 1.5988,
      "step": 3964
    },
    {
      "epoch": 0.3047019053472649,
      "grad_norm": 3.3523051738739014,
      "learning_rate": 1.878119237861094e-05,
      "loss": 1.5015,
      "step": 3966
    },
    {
      "epoch": 0.3048555623847572,
      "grad_norm": 3.5124311447143555,
      "learning_rate": 1.8780577750460974e-05,
      "loss": 1.4289,
      "step": 3968
    },
    {
      "epoch": 0.30500921942224957,
      "grad_norm": 3.454429864883423,
      "learning_rate": 1.8779963122311004e-05,
      "loss": 1.4927,
      "step": 3970
    },
    {
      "epoch": 0.3051628764597419,
      "grad_norm": 3.9791088104248047,
      "learning_rate": 1.8779348494161034e-05,
      "loss": 1.6143,
      "step": 3972
    },
    {
      "epoch": 0.3053165334972342,
      "grad_norm": 3.819364547729492,
      "learning_rate": 1.8778733866011067e-05,
      "loss": 1.5797,
      "step": 3974
    },
    {
      "epoch": 0.3054701905347265,
      "grad_norm": 3.4424610137939453,
      "learning_rate": 1.8778119237861096e-05,
      "loss": 1.4364,
      "step": 3976
    },
    {
      "epoch": 0.30562384757221883,
      "grad_norm": 3.86603045463562,
      "learning_rate": 1.8777504609711126e-05,
      "loss": 1.5102,
      "step": 3978
    },
    {
      "epoch": 0.30577750460971115,
      "grad_norm": 3.8857436180114746,
      "learning_rate": 1.8776889981561156e-05,
      "loss": 1.5109,
      "step": 3980
    },
    {
      "epoch": 0.30593116164720346,
      "grad_norm": 3.8796331882476807,
      "learning_rate": 1.877627535341119e-05,
      "loss": 1.4598,
      "step": 3982
    },
    {
      "epoch": 0.3060848186846958,
      "grad_norm": 3.510911464691162,
      "learning_rate": 1.877566072526122e-05,
      "loss": 1.4632,
      "step": 3984
    },
    {
      "epoch": 0.3062384757221881,
      "grad_norm": 3.9845762252807617,
      "learning_rate": 1.8775046097111248e-05,
      "loss": 1.5021,
      "step": 3986
    },
    {
      "epoch": 0.3063921327596804,
      "grad_norm": 3.142535448074341,
      "learning_rate": 1.877443146896128e-05,
      "loss": 1.4725,
      "step": 3988
    },
    {
      "epoch": 0.30654578979717273,
      "grad_norm": 3.894141435623169,
      "learning_rate": 1.877381684081131e-05,
      "loss": 1.5427,
      "step": 3990
    },
    {
      "epoch": 0.30669944683466505,
      "grad_norm": 3.3294742107391357,
      "learning_rate": 1.877320221266134e-05,
      "loss": 1.5218,
      "step": 3992
    },
    {
      "epoch": 0.30685310387215736,
      "grad_norm": 3.9828765392303467,
      "learning_rate": 1.8772587584511374e-05,
      "loss": 1.5058,
      "step": 3994
    },
    {
      "epoch": 0.3070067609096497,
      "grad_norm": 3.469477415084839,
      "learning_rate": 1.8771972956361403e-05,
      "loss": 1.4868,
      "step": 3996
    },
    {
      "epoch": 0.307160417947142,
      "grad_norm": 3.239800453186035,
      "learning_rate": 1.8771358328211433e-05,
      "loss": 1.3784,
      "step": 3998
    },
    {
      "epoch": 0.3073140749846343,
      "grad_norm": 3.649010419845581,
      "learning_rate": 1.8770743700061466e-05,
      "loss": 1.3653,
      "step": 4000
    },
    {
      "epoch": 0.3074677320221266,
      "grad_norm": 3.0160863399505615,
      "learning_rate": 1.8770129071911496e-05,
      "loss": 1.2752,
      "step": 4002
    },
    {
      "epoch": 0.30762138905961894,
      "grad_norm": 3.026254892349243,
      "learning_rate": 1.8769514443761525e-05,
      "loss": 1.52,
      "step": 4004
    },
    {
      "epoch": 0.30777504609711126,
      "grad_norm": 3.2732443809509277,
      "learning_rate": 1.8768899815611555e-05,
      "loss": 1.4611,
      "step": 4006
    },
    {
      "epoch": 0.3079287031346036,
      "grad_norm": 3.575145721435547,
      "learning_rate": 1.8768285187461588e-05,
      "loss": 1.4099,
      "step": 4008
    },
    {
      "epoch": 0.3080823601720959,
      "grad_norm": 4.473994255065918,
      "learning_rate": 1.8767670559311618e-05,
      "loss": 1.5474,
      "step": 4010
    },
    {
      "epoch": 0.3082360172095882,
      "grad_norm": 3.5895843505859375,
      "learning_rate": 1.8767055931161648e-05,
      "loss": 1.3964,
      "step": 4012
    },
    {
      "epoch": 0.3083896742470805,
      "grad_norm": 3.707585334777832,
      "learning_rate": 1.876644130301168e-05,
      "loss": 1.4779,
      "step": 4014
    },
    {
      "epoch": 0.30854333128457284,
      "grad_norm": 2.8725762367248535,
      "learning_rate": 1.876582667486171e-05,
      "loss": 1.4191,
      "step": 4016
    },
    {
      "epoch": 0.30869698832206516,
      "grad_norm": 4.254179000854492,
      "learning_rate": 1.876521204671174e-05,
      "loss": 1.5689,
      "step": 4018
    },
    {
      "epoch": 0.3088506453595575,
      "grad_norm": 3.4411301612854004,
      "learning_rate": 1.8764597418561773e-05,
      "loss": 1.4418,
      "step": 4020
    },
    {
      "epoch": 0.3090043023970498,
      "grad_norm": 3.499048948287964,
      "learning_rate": 1.8763982790411803e-05,
      "loss": 1.4129,
      "step": 4022
    },
    {
      "epoch": 0.3091579594345421,
      "grad_norm": 3.3201608657836914,
      "learning_rate": 1.8763368162261832e-05,
      "loss": 1.4546,
      "step": 4024
    },
    {
      "epoch": 0.3093116164720344,
      "grad_norm": 3.6823911666870117,
      "learning_rate": 1.8762753534111865e-05,
      "loss": 1.3802,
      "step": 4026
    },
    {
      "epoch": 0.30946527350952674,
      "grad_norm": 3.8812990188598633,
      "learning_rate": 1.8762138905961895e-05,
      "loss": 1.4478,
      "step": 4028
    },
    {
      "epoch": 0.30961893054701906,
      "grad_norm": 3.55722713470459,
      "learning_rate": 1.8761524277811925e-05,
      "loss": 1.5156,
      "step": 4030
    },
    {
      "epoch": 0.30977258758451137,
      "grad_norm": 3.832512378692627,
      "learning_rate": 1.8760909649661955e-05,
      "loss": 1.4308,
      "step": 4032
    },
    {
      "epoch": 0.3099262446220037,
      "grad_norm": 3.5508055686950684,
      "learning_rate": 1.8760295021511988e-05,
      "loss": 1.3902,
      "step": 4034
    },
    {
      "epoch": 0.310079901659496,
      "grad_norm": 3.4884133338928223,
      "learning_rate": 1.8759680393362017e-05,
      "loss": 1.3769,
      "step": 4036
    },
    {
      "epoch": 0.3102335586969883,
      "grad_norm": 3.4577252864837646,
      "learning_rate": 1.8759065765212047e-05,
      "loss": 1.4883,
      "step": 4038
    },
    {
      "epoch": 0.31038721573448064,
      "grad_norm": 3.4539401531219482,
      "learning_rate": 1.875845113706208e-05,
      "loss": 1.4841,
      "step": 4040
    },
    {
      "epoch": 0.31054087277197295,
      "grad_norm": 3.1228976249694824,
      "learning_rate": 1.875783650891211e-05,
      "loss": 1.366,
      "step": 4042
    },
    {
      "epoch": 0.31069452980946527,
      "grad_norm": 3.5434763431549072,
      "learning_rate": 1.875722188076214e-05,
      "loss": 1.5336,
      "step": 4044
    },
    {
      "epoch": 0.3108481868469576,
      "grad_norm": 3.5703988075256348,
      "learning_rate": 1.8756607252612172e-05,
      "loss": 1.3806,
      "step": 4046
    },
    {
      "epoch": 0.3110018438844499,
      "grad_norm": 3.933476686477661,
      "learning_rate": 1.8755992624462202e-05,
      "loss": 1.4786,
      "step": 4048
    },
    {
      "epoch": 0.3111555009219422,
      "grad_norm": 3.1267027854919434,
      "learning_rate": 1.8755377996312232e-05,
      "loss": 1.3594,
      "step": 4050
    },
    {
      "epoch": 0.31130915795943453,
      "grad_norm": 3.2897634506225586,
      "learning_rate": 1.875476336816226e-05,
      "loss": 1.4306,
      "step": 4052
    },
    {
      "epoch": 0.31146281499692685,
      "grad_norm": 3.3133482933044434,
      "learning_rate": 1.8754148740012295e-05,
      "loss": 1.4541,
      "step": 4054
    },
    {
      "epoch": 0.31161647203441917,
      "grad_norm": 3.548305034637451,
      "learning_rate": 1.8753534111862324e-05,
      "loss": 1.3178,
      "step": 4056
    },
    {
      "epoch": 0.3117701290719115,
      "grad_norm": 3.642669200897217,
      "learning_rate": 1.8752919483712354e-05,
      "loss": 1.6155,
      "step": 4058
    },
    {
      "epoch": 0.3119237861094038,
      "grad_norm": 3.6712703704833984,
      "learning_rate": 1.8752304855562387e-05,
      "loss": 1.4287,
      "step": 4060
    },
    {
      "epoch": 0.3120774431468961,
      "grad_norm": 3.957775592803955,
      "learning_rate": 1.8751690227412417e-05,
      "loss": 1.5324,
      "step": 4062
    },
    {
      "epoch": 0.31223110018438843,
      "grad_norm": 3.234064817428589,
      "learning_rate": 1.8751075599262446e-05,
      "loss": 1.4191,
      "step": 4064
    },
    {
      "epoch": 0.31238475722188075,
      "grad_norm": 4.080970287322998,
      "learning_rate": 1.875046097111248e-05,
      "loss": 1.5567,
      "step": 4066
    },
    {
      "epoch": 0.31253841425937307,
      "grad_norm": 3.7865617275238037,
      "learning_rate": 1.874984634296251e-05,
      "loss": 1.3855,
      "step": 4068
    },
    {
      "epoch": 0.3126920712968654,
      "grad_norm": 3.3486435413360596,
      "learning_rate": 1.874923171481254e-05,
      "loss": 1.4618,
      "step": 4070
    },
    {
      "epoch": 0.3128457283343577,
      "grad_norm": 3.3620622158050537,
      "learning_rate": 1.8748617086662572e-05,
      "loss": 1.4594,
      "step": 4072
    },
    {
      "epoch": 0.31299938537185,
      "grad_norm": 3.3821537494659424,
      "learning_rate": 1.87480024585126e-05,
      "loss": 1.3137,
      "step": 4074
    },
    {
      "epoch": 0.31315304240934233,
      "grad_norm": 3.1241402626037598,
      "learning_rate": 1.874738783036263e-05,
      "loss": 1.5348,
      "step": 4076
    },
    {
      "epoch": 0.31330669944683465,
      "grad_norm": 3.696268081665039,
      "learning_rate": 1.874677320221266e-05,
      "loss": 1.5934,
      "step": 4078
    },
    {
      "epoch": 0.31346035648432696,
      "grad_norm": 3.3421854972839355,
      "learning_rate": 1.8746158574062694e-05,
      "loss": 1.4672,
      "step": 4080
    },
    {
      "epoch": 0.3136140135218193,
      "grad_norm": 4.347311496734619,
      "learning_rate": 1.8745543945912724e-05,
      "loss": 1.4414,
      "step": 4082
    },
    {
      "epoch": 0.3137676705593116,
      "grad_norm": 3.2006402015686035,
      "learning_rate": 1.8744929317762753e-05,
      "loss": 1.4057,
      "step": 4084
    },
    {
      "epoch": 0.3139213275968039,
      "grad_norm": 3.5505783557891846,
      "learning_rate": 1.8744314689612786e-05,
      "loss": 1.4324,
      "step": 4086
    },
    {
      "epoch": 0.31407498463429623,
      "grad_norm": 3.9366791248321533,
      "learning_rate": 1.8743700061462816e-05,
      "loss": 1.5607,
      "step": 4088
    },
    {
      "epoch": 0.31422864167178854,
      "grad_norm": 3.4869041442871094,
      "learning_rate": 1.8743085433312846e-05,
      "loss": 1.368,
      "step": 4090
    },
    {
      "epoch": 0.31438229870928086,
      "grad_norm": 3.363190174102783,
      "learning_rate": 1.874247080516288e-05,
      "loss": 1.4657,
      "step": 4092
    },
    {
      "epoch": 0.3145359557467732,
      "grad_norm": 3.5293331146240234,
      "learning_rate": 1.874185617701291e-05,
      "loss": 1.527,
      "step": 4094
    },
    {
      "epoch": 0.3146896127842655,
      "grad_norm": 3.2504899501800537,
      "learning_rate": 1.8741241548862938e-05,
      "loss": 1.4395,
      "step": 4096
    },
    {
      "epoch": 0.3148432698217578,
      "grad_norm": 3.309957981109619,
      "learning_rate": 1.874062692071297e-05,
      "loss": 1.4809,
      "step": 4098
    },
    {
      "epoch": 0.3149969268592501,
      "grad_norm": 3.5390985012054443,
      "learning_rate": 1.8740012292563e-05,
      "loss": 1.4781,
      "step": 4100
    },
    {
      "epoch": 0.3151505838967425,
      "grad_norm": 3.856665849685669,
      "learning_rate": 1.8739397664413034e-05,
      "loss": 1.569,
      "step": 4102
    },
    {
      "epoch": 0.3153042409342348,
      "grad_norm": 3.2815372943878174,
      "learning_rate": 1.873878303626306e-05,
      "loss": 1.4387,
      "step": 4104
    },
    {
      "epoch": 0.31545789797172713,
      "grad_norm": 3.4734060764312744,
      "learning_rate": 1.8738168408113093e-05,
      "loss": 1.5414,
      "step": 4106
    },
    {
      "epoch": 0.31561155500921945,
      "grad_norm": 3.0709571838378906,
      "learning_rate": 1.8737553779963123e-05,
      "loss": 1.3674,
      "step": 4108
    },
    {
      "epoch": 0.31576521204671176,
      "grad_norm": 3.460763692855835,
      "learning_rate": 1.8736939151813153e-05,
      "loss": 1.4759,
      "step": 4110
    },
    {
      "epoch": 0.3159188690842041,
      "grad_norm": 3.14346981048584,
      "learning_rate": 1.8736324523663186e-05,
      "loss": 1.4407,
      "step": 4112
    },
    {
      "epoch": 0.3160725261216964,
      "grad_norm": 3.1773383617401123,
      "learning_rate": 1.8735709895513216e-05,
      "loss": 1.4686,
      "step": 4114
    },
    {
      "epoch": 0.3162261831591887,
      "grad_norm": 3.2053041458129883,
      "learning_rate": 1.8735095267363245e-05,
      "loss": 1.4797,
      "step": 4116
    },
    {
      "epoch": 0.31637984019668103,
      "grad_norm": 3.4825661182403564,
      "learning_rate": 1.873448063921328e-05,
      "loss": 1.3502,
      "step": 4118
    },
    {
      "epoch": 0.31653349723417334,
      "grad_norm": 3.6239278316497803,
      "learning_rate": 1.8733866011063308e-05,
      "loss": 1.5613,
      "step": 4120
    },
    {
      "epoch": 0.31668715427166566,
      "grad_norm": 3.647029161453247,
      "learning_rate": 1.873325138291334e-05,
      "loss": 1.504,
      "step": 4122
    },
    {
      "epoch": 0.316840811309158,
      "grad_norm": 3.517030715942383,
      "learning_rate": 1.873263675476337e-05,
      "loss": 1.4534,
      "step": 4124
    },
    {
      "epoch": 0.3169944683466503,
      "grad_norm": 3.620820999145508,
      "learning_rate": 1.87320221266134e-05,
      "loss": 1.6111,
      "step": 4126
    },
    {
      "epoch": 0.3171481253841426,
      "grad_norm": 3.410433769226074,
      "learning_rate": 1.8731407498463434e-05,
      "loss": 1.6134,
      "step": 4128
    },
    {
      "epoch": 0.3173017824216349,
      "grad_norm": 3.1551625728607178,
      "learning_rate": 1.873079287031346e-05,
      "loss": 1.3274,
      "step": 4130
    },
    {
      "epoch": 0.31745543945912724,
      "grad_norm": 2.9773385524749756,
      "learning_rate": 1.8730178242163493e-05,
      "loss": 1.3743,
      "step": 4132
    },
    {
      "epoch": 0.31760909649661956,
      "grad_norm": 3.5205631256103516,
      "learning_rate": 1.8729563614013523e-05,
      "loss": 1.5177,
      "step": 4134
    },
    {
      "epoch": 0.3177627535341119,
      "grad_norm": 3.157318592071533,
      "learning_rate": 1.8728948985863552e-05,
      "loss": 1.3897,
      "step": 4136
    },
    {
      "epoch": 0.3179164105716042,
      "grad_norm": 3.2768332958221436,
      "learning_rate": 1.8728334357713585e-05,
      "loss": 1.5896,
      "step": 4138
    },
    {
      "epoch": 0.3180700676090965,
      "grad_norm": 3.3517439365386963,
      "learning_rate": 1.8727719729563615e-05,
      "loss": 1.4006,
      "step": 4140
    },
    {
      "epoch": 0.3182237246465888,
      "grad_norm": 3.4043068885803223,
      "learning_rate": 1.8727105101413648e-05,
      "loss": 1.4092,
      "step": 4142
    },
    {
      "epoch": 0.31837738168408114,
      "grad_norm": 4.253170490264893,
      "learning_rate": 1.8726490473263678e-05,
      "loss": 1.499,
      "step": 4144
    },
    {
      "epoch": 0.31853103872157346,
      "grad_norm": 3.3389856815338135,
      "learning_rate": 1.8725875845113707e-05,
      "loss": 1.3179,
      "step": 4146
    },
    {
      "epoch": 0.3186846957590658,
      "grad_norm": 3.3309342861175537,
      "learning_rate": 1.872526121696374e-05,
      "loss": 1.3684,
      "step": 4148
    },
    {
      "epoch": 0.3188383527965581,
      "grad_norm": 3.2987401485443115,
      "learning_rate": 1.8724646588813767e-05,
      "loss": 1.4179,
      "step": 4150
    },
    {
      "epoch": 0.3189920098340504,
      "grad_norm": 4.022191047668457,
      "learning_rate": 1.87240319606638e-05,
      "loss": 1.6518,
      "step": 4152
    },
    {
      "epoch": 0.3191456668715427,
      "grad_norm": 3.6432507038116455,
      "learning_rate": 1.8723417332513833e-05,
      "loss": 1.6108,
      "step": 4154
    },
    {
      "epoch": 0.31929932390903504,
      "grad_norm": 3.7931549549102783,
      "learning_rate": 1.872280270436386e-05,
      "loss": 1.4346,
      "step": 4156
    },
    {
      "epoch": 0.31945298094652735,
      "grad_norm": 3.829770088195801,
      "learning_rate": 1.8722188076213892e-05,
      "loss": 1.3766,
      "step": 4158
    },
    {
      "epoch": 0.31960663798401967,
      "grad_norm": 3.3749098777770996,
      "learning_rate": 1.8721573448063922e-05,
      "loss": 1.5031,
      "step": 4160
    },
    {
      "epoch": 0.319760295021512,
      "grad_norm": 3.3764944076538086,
      "learning_rate": 1.8720958819913955e-05,
      "loss": 1.4594,
      "step": 4162
    },
    {
      "epoch": 0.3199139520590043,
      "grad_norm": 3.541229248046875,
      "learning_rate": 1.8720344191763985e-05,
      "loss": 1.5586,
      "step": 4164
    },
    {
      "epoch": 0.3200676090964966,
      "grad_norm": 3.983872175216675,
      "learning_rate": 1.8719729563614014e-05,
      "loss": 1.4576,
      "step": 4166
    },
    {
      "epoch": 0.32022126613398894,
      "grad_norm": 3.3289988040924072,
      "learning_rate": 1.8719114935464048e-05,
      "loss": 1.5158,
      "step": 4168
    },
    {
      "epoch": 0.32037492317148125,
      "grad_norm": 3.6405656337738037,
      "learning_rate": 1.8718500307314077e-05,
      "loss": 1.3827,
      "step": 4170
    },
    {
      "epoch": 0.32052858020897357,
      "grad_norm": 3.8185787200927734,
      "learning_rate": 1.8717885679164107e-05,
      "loss": 1.4772,
      "step": 4172
    },
    {
      "epoch": 0.3206822372464659,
      "grad_norm": 2.896965980529785,
      "learning_rate": 1.871727105101414e-05,
      "loss": 1.4799,
      "step": 4174
    },
    {
      "epoch": 0.3208358942839582,
      "grad_norm": 3.191843271255493,
      "learning_rate": 1.8716656422864166e-05,
      "loss": 1.4894,
      "step": 4176
    },
    {
      "epoch": 0.3209895513214505,
      "grad_norm": 3.1855785846710205,
      "learning_rate": 1.87160417947142e-05,
      "loss": 1.3661,
      "step": 4178
    },
    {
      "epoch": 0.32114320835894283,
      "grad_norm": 3.66670298576355,
      "learning_rate": 1.871542716656423e-05,
      "loss": 1.4934,
      "step": 4180
    },
    {
      "epoch": 0.32129686539643515,
      "grad_norm": 2.9798126220703125,
      "learning_rate": 1.871481253841426e-05,
      "loss": 1.3584,
      "step": 4182
    },
    {
      "epoch": 0.32145052243392747,
      "grad_norm": 3.9265804290771484,
      "learning_rate": 1.8714197910264292e-05,
      "loss": 1.3534,
      "step": 4184
    },
    {
      "epoch": 0.3216041794714198,
      "grad_norm": 3.341581344604492,
      "learning_rate": 1.871358328211432e-05,
      "loss": 1.349,
      "step": 4186
    },
    {
      "epoch": 0.3217578365089121,
      "grad_norm": 3.206651449203491,
      "learning_rate": 1.8712968653964355e-05,
      "loss": 1.368,
      "step": 4188
    },
    {
      "epoch": 0.3219114935464044,
      "grad_norm": 3.7142248153686523,
      "learning_rate": 1.8712354025814384e-05,
      "loss": 1.4761,
      "step": 4190
    },
    {
      "epoch": 0.32206515058389673,
      "grad_norm": 3.654179096221924,
      "learning_rate": 1.8711739397664414e-05,
      "loss": 1.5724,
      "step": 4192
    },
    {
      "epoch": 0.32221880762138905,
      "grad_norm": 3.130634069442749,
      "learning_rate": 1.8711124769514447e-05,
      "loss": 1.3865,
      "step": 4194
    },
    {
      "epoch": 0.32237246465888136,
      "grad_norm": 3.973550319671631,
      "learning_rate": 1.8710510141364477e-05,
      "loss": 1.4493,
      "step": 4196
    },
    {
      "epoch": 0.3225261216963737,
      "grad_norm": 3.044903039932251,
      "learning_rate": 1.8709895513214506e-05,
      "loss": 1.4508,
      "step": 4198
    },
    {
      "epoch": 0.322679778733866,
      "grad_norm": 3.1042187213897705,
      "learning_rate": 1.870928088506454e-05,
      "loss": 1.5592,
      "step": 4200
    },
    {
      "epoch": 0.3228334357713583,
      "grad_norm": 3.174510955810547,
      "learning_rate": 1.8708666256914566e-05,
      "loss": 1.5262,
      "step": 4202
    },
    {
      "epoch": 0.32298709280885063,
      "grad_norm": 3.3536434173583984,
      "learning_rate": 1.87080516287646e-05,
      "loss": 1.3904,
      "step": 4204
    },
    {
      "epoch": 0.32314074984634295,
      "grad_norm": 3.7183215618133545,
      "learning_rate": 1.870743700061463e-05,
      "loss": 1.3558,
      "step": 4206
    },
    {
      "epoch": 0.32329440688383526,
      "grad_norm": 3.7605910301208496,
      "learning_rate": 1.870682237246466e-05,
      "loss": 1.4543,
      "step": 4208
    },
    {
      "epoch": 0.3234480639213276,
      "grad_norm": 3.1257007122039795,
      "learning_rate": 1.870620774431469e-05,
      "loss": 1.3025,
      "step": 4210
    },
    {
      "epoch": 0.3236017209588199,
      "grad_norm": 3.429586172103882,
      "learning_rate": 1.870559311616472e-05,
      "loss": 1.4547,
      "step": 4212
    },
    {
      "epoch": 0.3237553779963122,
      "grad_norm": 3.820537805557251,
      "learning_rate": 1.8704978488014754e-05,
      "loss": 1.5153,
      "step": 4214
    },
    {
      "epoch": 0.3239090350338045,
      "grad_norm": 3.532174825668335,
      "learning_rate": 1.8704363859864784e-05,
      "loss": 1.3529,
      "step": 4216
    },
    {
      "epoch": 0.32406269207129684,
      "grad_norm": 3.619643211364746,
      "learning_rate": 1.8703749231714813e-05,
      "loss": 1.4404,
      "step": 4218
    },
    {
      "epoch": 0.32421634910878916,
      "grad_norm": 3.474963665008545,
      "learning_rate": 1.8703134603564846e-05,
      "loss": 1.4391,
      "step": 4220
    },
    {
      "epoch": 0.3243700061462815,
      "grad_norm": 3.2901859283447266,
      "learning_rate": 1.8702519975414876e-05,
      "loss": 1.4308,
      "step": 4222
    },
    {
      "epoch": 0.3245236631837738,
      "grad_norm": 3.482391834259033,
      "learning_rate": 1.8701905347264906e-05,
      "loss": 1.4473,
      "step": 4224
    },
    {
      "epoch": 0.3246773202212661,
      "grad_norm": 3.2671616077423096,
      "learning_rate": 1.870129071911494e-05,
      "loss": 1.385,
      "step": 4226
    },
    {
      "epoch": 0.3248309772587584,
      "grad_norm": 3.6331212520599365,
      "learning_rate": 1.870067609096497e-05,
      "loss": 1.4433,
      "step": 4228
    },
    {
      "epoch": 0.32498463429625074,
      "grad_norm": 3.0199925899505615,
      "learning_rate": 1.8700061462814998e-05,
      "loss": 1.4552,
      "step": 4230
    },
    {
      "epoch": 0.3251382913337431,
      "grad_norm": 3.5435800552368164,
      "learning_rate": 1.8699446834665028e-05,
      "loss": 1.4098,
      "step": 4232
    },
    {
      "epoch": 0.32529194837123543,
      "grad_norm": 3.5522637367248535,
      "learning_rate": 1.869883220651506e-05,
      "loss": 1.3704,
      "step": 4234
    },
    {
      "epoch": 0.32544560540872775,
      "grad_norm": 3.2306251525878906,
      "learning_rate": 1.869821757836509e-05,
      "loss": 1.3251,
      "step": 4236
    },
    {
      "epoch": 0.32559926244622006,
      "grad_norm": 3.570638656616211,
      "learning_rate": 1.869760295021512e-05,
      "loss": 1.3477,
      "step": 4238
    },
    {
      "epoch": 0.3257529194837124,
      "grad_norm": 3.919713020324707,
      "learning_rate": 1.8696988322065153e-05,
      "loss": 1.455,
      "step": 4240
    },
    {
      "epoch": 0.3259065765212047,
      "grad_norm": 3.3150136470794678,
      "learning_rate": 1.8696373693915183e-05,
      "loss": 1.5392,
      "step": 4242
    },
    {
      "epoch": 0.326060233558697,
      "grad_norm": 3.362748146057129,
      "learning_rate": 1.8695759065765213e-05,
      "loss": 1.3796,
      "step": 4244
    },
    {
      "epoch": 0.3262138905961893,
      "grad_norm": 3.556650161743164,
      "learning_rate": 1.8695144437615246e-05,
      "loss": 1.3962,
      "step": 4246
    },
    {
      "epoch": 0.32636754763368164,
      "grad_norm": 3.651918649673462,
      "learning_rate": 1.8694529809465276e-05,
      "loss": 1.4432,
      "step": 4248
    },
    {
      "epoch": 0.32652120467117396,
      "grad_norm": 3.641026496887207,
      "learning_rate": 1.8693915181315305e-05,
      "loss": 1.3968,
      "step": 4250
    },
    {
      "epoch": 0.3266748617086663,
      "grad_norm": 3.573974370956421,
      "learning_rate": 1.869330055316534e-05,
      "loss": 1.5365,
      "step": 4252
    },
    {
      "epoch": 0.3268285187461586,
      "grad_norm": 3.4720635414123535,
      "learning_rate": 1.8692685925015368e-05,
      "loss": 1.4469,
      "step": 4254
    },
    {
      "epoch": 0.3269821757836509,
      "grad_norm": 3.5822393894195557,
      "learning_rate": 1.8692071296865398e-05,
      "loss": 1.5314,
      "step": 4256
    },
    {
      "epoch": 0.3271358328211432,
      "grad_norm": 3.3091721534729004,
      "learning_rate": 1.8691456668715427e-05,
      "loss": 1.5533,
      "step": 4258
    },
    {
      "epoch": 0.32728948985863554,
      "grad_norm": 3.3937950134277344,
      "learning_rate": 1.869084204056546e-05,
      "loss": 1.498,
      "step": 4260
    },
    {
      "epoch": 0.32744314689612786,
      "grad_norm": 3.1424973011016846,
      "learning_rate": 1.869022741241549e-05,
      "loss": 1.384,
      "step": 4262
    },
    {
      "epoch": 0.3275968039336202,
      "grad_norm": 3.1673672199249268,
      "learning_rate": 1.868961278426552e-05,
      "loss": 1.3467,
      "step": 4264
    },
    {
      "epoch": 0.3277504609711125,
      "grad_norm": 3.0541131496429443,
      "learning_rate": 1.8688998156115553e-05,
      "loss": 1.5197,
      "step": 4266
    },
    {
      "epoch": 0.3279041180086048,
      "grad_norm": 3.423753261566162,
      "learning_rate": 1.8688383527965583e-05,
      "loss": 1.4421,
      "step": 4268
    },
    {
      "epoch": 0.3280577750460971,
      "grad_norm": 3.221930980682373,
      "learning_rate": 1.8687768899815612e-05,
      "loss": 1.3816,
      "step": 4270
    },
    {
      "epoch": 0.32821143208358944,
      "grad_norm": 3.318981170654297,
      "learning_rate": 1.8687154271665645e-05,
      "loss": 1.3417,
      "step": 4272
    },
    {
      "epoch": 0.32836508912108175,
      "grad_norm": 3.644233226776123,
      "learning_rate": 1.8686539643515675e-05,
      "loss": 1.5216,
      "step": 4274
    },
    {
      "epoch": 0.32851874615857407,
      "grad_norm": 3.196216106414795,
      "learning_rate": 1.8685925015365705e-05,
      "loss": 1.313,
      "step": 4276
    },
    {
      "epoch": 0.3286724031960664,
      "grad_norm": 3.06890606880188,
      "learning_rate": 1.8685310387215734e-05,
      "loss": 1.446,
      "step": 4278
    },
    {
      "epoch": 0.3288260602335587,
      "grad_norm": 3.106513500213623,
      "learning_rate": 1.8684695759065767e-05,
      "loss": 1.3891,
      "step": 4280
    },
    {
      "epoch": 0.328979717271051,
      "grad_norm": 3.394026517868042,
      "learning_rate": 1.8684081130915797e-05,
      "loss": 1.3584,
      "step": 4282
    },
    {
      "epoch": 0.32913337430854334,
      "grad_norm": 3.0050766468048096,
      "learning_rate": 1.8683466502765827e-05,
      "loss": 1.4602,
      "step": 4284
    },
    {
      "epoch": 0.32928703134603565,
      "grad_norm": 3.0258212089538574,
      "learning_rate": 1.868285187461586e-05,
      "loss": 1.3084,
      "step": 4286
    },
    {
      "epoch": 0.32944068838352797,
      "grad_norm": 3.1330137252807617,
      "learning_rate": 1.868223724646589e-05,
      "loss": 1.367,
      "step": 4288
    },
    {
      "epoch": 0.3295943454210203,
      "grad_norm": 3.145908832550049,
      "learning_rate": 1.868162261831592e-05,
      "loss": 1.3691,
      "step": 4290
    },
    {
      "epoch": 0.3297480024585126,
      "grad_norm": 3.0322062969207764,
      "learning_rate": 1.8681007990165952e-05,
      "loss": 1.4524,
      "step": 4292
    },
    {
      "epoch": 0.3299016594960049,
      "grad_norm": 8.492423057556152,
      "learning_rate": 1.8680393362015982e-05,
      "loss": 1.4982,
      "step": 4294
    },
    {
      "epoch": 0.33005531653349723,
      "grad_norm": 3.640282154083252,
      "learning_rate": 1.867977873386601e-05,
      "loss": 1.4207,
      "step": 4296
    },
    {
      "epoch": 0.33020897357098955,
      "grad_norm": 3.1707777976989746,
      "learning_rate": 1.8679164105716045e-05,
      "loss": 1.4874,
      "step": 4298
    },
    {
      "epoch": 0.33036263060848187,
      "grad_norm": 3.1674551963806152,
      "learning_rate": 1.8678549477566074e-05,
      "loss": 1.5869,
      "step": 4300
    },
    {
      "epoch": 0.3305162876459742,
      "grad_norm": 3.2966535091400146,
      "learning_rate": 1.8677934849416104e-05,
      "loss": 1.4055,
      "step": 4302
    },
    {
      "epoch": 0.3306699446834665,
      "grad_norm": 3.6385233402252197,
      "learning_rate": 1.8677320221266134e-05,
      "loss": 1.6527,
      "step": 4304
    },
    {
      "epoch": 0.3308236017209588,
      "grad_norm": 3.1749191284179688,
      "learning_rate": 1.8676705593116167e-05,
      "loss": 1.3133,
      "step": 4306
    },
    {
      "epoch": 0.33097725875845113,
      "grad_norm": 3.473065137863159,
      "learning_rate": 1.8676090964966197e-05,
      "loss": 1.5052,
      "step": 4308
    },
    {
      "epoch": 0.33113091579594345,
      "grad_norm": 3.3100507259368896,
      "learning_rate": 1.8675476336816226e-05,
      "loss": 1.4369,
      "step": 4310
    },
    {
      "epoch": 0.33128457283343576,
      "grad_norm": 3.4080681800842285,
      "learning_rate": 1.867486170866626e-05,
      "loss": 1.5524,
      "step": 4312
    },
    {
      "epoch": 0.3314382298709281,
      "grad_norm": 3.419074296951294,
      "learning_rate": 1.867424708051629e-05,
      "loss": 1.515,
      "step": 4314
    },
    {
      "epoch": 0.3315918869084204,
      "grad_norm": 4.8144965171813965,
      "learning_rate": 1.867363245236632e-05,
      "loss": 1.425,
      "step": 4316
    },
    {
      "epoch": 0.3317455439459127,
      "grad_norm": 2.8420140743255615,
      "learning_rate": 1.8673017824216352e-05,
      "loss": 1.3018,
      "step": 4318
    },
    {
      "epoch": 0.33189920098340503,
      "grad_norm": 3.841867208480835,
      "learning_rate": 1.867240319606638e-05,
      "loss": 1.4571,
      "step": 4320
    },
    {
      "epoch": 0.33205285802089735,
      "grad_norm": 3.476323127746582,
      "learning_rate": 1.867178856791641e-05,
      "loss": 1.4041,
      "step": 4322
    },
    {
      "epoch": 0.33220651505838966,
      "grad_norm": 3.4778249263763428,
      "learning_rate": 1.8671173939766444e-05,
      "loss": 1.5176,
      "step": 4324
    },
    {
      "epoch": 0.332360172095882,
      "grad_norm": 3.6123509407043457,
      "learning_rate": 1.8670559311616474e-05,
      "loss": 1.4301,
      "step": 4326
    },
    {
      "epoch": 0.3325138291333743,
      "grad_norm": 3.1025562286376953,
      "learning_rate": 1.8669944683466504e-05,
      "loss": 1.3542,
      "step": 4328
    },
    {
      "epoch": 0.3326674861708666,
      "grad_norm": 3.7515599727630615,
      "learning_rate": 1.8669330055316533e-05,
      "loss": 1.5406,
      "step": 4330
    },
    {
      "epoch": 0.3328211432083589,
      "grad_norm": 3.2796072959899902,
      "learning_rate": 1.8668715427166566e-05,
      "loss": 1.4538,
      "step": 4332
    },
    {
      "epoch": 0.33297480024585124,
      "grad_norm": 3.37237548828125,
      "learning_rate": 1.8668100799016596e-05,
      "loss": 1.4781,
      "step": 4334
    },
    {
      "epoch": 0.33312845728334356,
      "grad_norm": 3.1982147693634033,
      "learning_rate": 1.8667486170866626e-05,
      "loss": 1.3791,
      "step": 4336
    },
    {
      "epoch": 0.3332821143208359,
      "grad_norm": 3.7179698944091797,
      "learning_rate": 1.866687154271666e-05,
      "loss": 1.4296,
      "step": 4338
    },
    {
      "epoch": 0.3334357713583282,
      "grad_norm": 3.3739073276519775,
      "learning_rate": 1.866625691456669e-05,
      "loss": 1.4229,
      "step": 4340
    },
    {
      "epoch": 0.3335894283958205,
      "grad_norm": 4.1158447265625,
      "learning_rate": 1.8665642286416718e-05,
      "loss": 1.4937,
      "step": 4342
    },
    {
      "epoch": 0.3337430854333128,
      "grad_norm": 3.25079345703125,
      "learning_rate": 1.866502765826675e-05,
      "loss": 1.3979,
      "step": 4344
    },
    {
      "epoch": 0.33389674247080514,
      "grad_norm": 3.4430036544799805,
      "learning_rate": 1.866441303011678e-05,
      "loss": 1.3599,
      "step": 4346
    },
    {
      "epoch": 0.33405039950829746,
      "grad_norm": 3.156423568725586,
      "learning_rate": 1.866379840196681e-05,
      "loss": 1.3951,
      "step": 4348
    },
    {
      "epoch": 0.3342040565457898,
      "grad_norm": 3.685227632522583,
      "learning_rate": 1.8663183773816844e-05,
      "loss": 1.4817,
      "step": 4350
    },
    {
      "epoch": 0.3343577135832821,
      "grad_norm": 3.6018106937408447,
      "learning_rate": 1.8662569145666873e-05,
      "loss": 1.4947,
      "step": 4352
    },
    {
      "epoch": 0.3345113706207744,
      "grad_norm": 3.826826572418213,
      "learning_rate": 1.8661954517516906e-05,
      "loss": 1.4161,
      "step": 4354
    },
    {
      "epoch": 0.3346650276582667,
      "grad_norm": 3.780466318130493,
      "learning_rate": 1.8661339889366933e-05,
      "loss": 1.4293,
      "step": 4356
    },
    {
      "epoch": 0.33481868469575904,
      "grad_norm": 3.5221405029296875,
      "learning_rate": 1.8660725261216966e-05,
      "loss": 1.359,
      "step": 4358
    },
    {
      "epoch": 0.33497234173325136,
      "grad_norm": 5.303091526031494,
      "learning_rate": 1.8660110633066995e-05,
      "loss": 1.5138,
      "step": 4360
    },
    {
      "epoch": 0.3351259987707437,
      "grad_norm": 4.311324596405029,
      "learning_rate": 1.8659496004917025e-05,
      "loss": 1.3831,
      "step": 4362
    },
    {
      "epoch": 0.33527965580823604,
      "grad_norm": 3.16186261177063,
      "learning_rate": 1.8658881376767058e-05,
      "loss": 1.4087,
      "step": 4364
    },
    {
      "epoch": 0.33543331284572836,
      "grad_norm": 3.3038687705993652,
      "learning_rate": 1.8658266748617088e-05,
      "loss": 1.4246,
      "step": 4366
    },
    {
      "epoch": 0.3355869698832207,
      "grad_norm": 3.750396490097046,
      "learning_rate": 1.8657652120467118e-05,
      "loss": 1.4069,
      "step": 4368
    },
    {
      "epoch": 0.335740626920713,
      "grad_norm": 3.6830053329467773,
      "learning_rate": 1.865703749231715e-05,
      "loss": 1.4182,
      "step": 4370
    },
    {
      "epoch": 0.3358942839582053,
      "grad_norm": 3.1328139305114746,
      "learning_rate": 1.865642286416718e-05,
      "loss": 1.573,
      "step": 4372
    },
    {
      "epoch": 0.3360479409956976,
      "grad_norm": 3.484109401702881,
      "learning_rate": 1.8655808236017213e-05,
      "loss": 1.4309,
      "step": 4374
    },
    {
      "epoch": 0.33620159803318994,
      "grad_norm": 3.408421516418457,
      "learning_rate": 1.865519360786724e-05,
      "loss": 1.4844,
      "step": 4376
    },
    {
      "epoch": 0.33635525507068226,
      "grad_norm": 3.305617094039917,
      "learning_rate": 1.8654578979717273e-05,
      "loss": 1.5163,
      "step": 4378
    },
    {
      "epoch": 0.3365089121081746,
      "grad_norm": 3.2654824256896973,
      "learning_rate": 1.8653964351567302e-05,
      "loss": 1.4274,
      "step": 4380
    },
    {
      "epoch": 0.3366625691456669,
      "grad_norm": 3.064093589782715,
      "learning_rate": 1.8653349723417332e-05,
      "loss": 1.504,
      "step": 4382
    },
    {
      "epoch": 0.3368162261831592,
      "grad_norm": 3.359511375427246,
      "learning_rate": 1.8652735095267365e-05,
      "loss": 1.486,
      "step": 4384
    },
    {
      "epoch": 0.3369698832206515,
      "grad_norm": 4.074959754943848,
      "learning_rate": 1.8652120467117395e-05,
      "loss": 1.5215,
      "step": 4386
    },
    {
      "epoch": 0.33712354025814384,
      "grad_norm": 3.4263994693756104,
      "learning_rate": 1.8651505838967425e-05,
      "loss": 1.2797,
      "step": 4388
    },
    {
      "epoch": 0.33727719729563616,
      "grad_norm": 3.5367822647094727,
      "learning_rate": 1.8650891210817458e-05,
      "loss": 1.3524,
      "step": 4390
    },
    {
      "epoch": 0.33743085433312847,
      "grad_norm": 5.375179290771484,
      "learning_rate": 1.8650276582667487e-05,
      "loss": 1.5596,
      "step": 4392
    },
    {
      "epoch": 0.3375845113706208,
      "grad_norm": 2.911515474319458,
      "learning_rate": 1.864966195451752e-05,
      "loss": 1.3833,
      "step": 4394
    },
    {
      "epoch": 0.3377381684081131,
      "grad_norm": 3.0453405380249023,
      "learning_rate": 1.864904732636755e-05,
      "loss": 1.4149,
      "step": 4396
    },
    {
      "epoch": 0.3378918254456054,
      "grad_norm": 5.521705150604248,
      "learning_rate": 1.864843269821758e-05,
      "loss": 1.4082,
      "step": 4398
    },
    {
      "epoch": 0.33804548248309774,
      "grad_norm": 3.640129804611206,
      "learning_rate": 1.8647818070067613e-05,
      "loss": 1.6099,
      "step": 4400
    },
    {
      "epoch": 0.33819913952059005,
      "grad_norm": 3.4472126960754395,
      "learning_rate": 1.864720344191764e-05,
      "loss": 1.4893,
      "step": 4402
    },
    {
      "epoch": 0.33835279655808237,
      "grad_norm": 3.440143585205078,
      "learning_rate": 1.8646588813767672e-05,
      "loss": 1.4867,
      "step": 4404
    },
    {
      "epoch": 0.3385064535955747,
      "grad_norm": 2.8615074157714844,
      "learning_rate": 1.8645974185617702e-05,
      "loss": 1.3287,
      "step": 4406
    },
    {
      "epoch": 0.338660110633067,
      "grad_norm": 3.6857492923736572,
      "learning_rate": 1.864535955746773e-05,
      "loss": 1.4894,
      "step": 4408
    },
    {
      "epoch": 0.3388137676705593,
      "grad_norm": 3.317317485809326,
      "learning_rate": 1.8644744929317765e-05,
      "loss": 1.4137,
      "step": 4410
    },
    {
      "epoch": 0.33896742470805163,
      "grad_norm": 3.293468952178955,
      "learning_rate": 1.8644130301167794e-05,
      "loss": 1.3174,
      "step": 4412
    },
    {
      "epoch": 0.33912108174554395,
      "grad_norm": 2.963331699371338,
      "learning_rate": 1.8643515673017827e-05,
      "loss": 1.396,
      "step": 4414
    },
    {
      "epoch": 0.33927473878303627,
      "grad_norm": 3.2345709800720215,
      "learning_rate": 1.8642901044867857e-05,
      "loss": 1.428,
      "step": 4416
    },
    {
      "epoch": 0.3394283958205286,
      "grad_norm": 3.399947166442871,
      "learning_rate": 1.8642286416717887e-05,
      "loss": 1.3047,
      "step": 4418
    },
    {
      "epoch": 0.3395820528580209,
      "grad_norm": 3.6432290077209473,
      "learning_rate": 1.864167178856792e-05,
      "loss": 1.4678,
      "step": 4420
    },
    {
      "epoch": 0.3397357098955132,
      "grad_norm": 3.2603604793548584,
      "learning_rate": 1.864105716041795e-05,
      "loss": 1.4103,
      "step": 4422
    },
    {
      "epoch": 0.33988936693300553,
      "grad_norm": 2.959282159805298,
      "learning_rate": 1.864044253226798e-05,
      "loss": 1.4762,
      "step": 4424
    },
    {
      "epoch": 0.34004302397049785,
      "grad_norm": 3.5913710594177246,
      "learning_rate": 1.8639827904118012e-05,
      "loss": 1.5115,
      "step": 4426
    },
    {
      "epoch": 0.34019668100799016,
      "grad_norm": 3.229787826538086,
      "learning_rate": 1.863921327596804e-05,
      "loss": 1.3089,
      "step": 4428
    },
    {
      "epoch": 0.3403503380454825,
      "grad_norm": 3.356717109680176,
      "learning_rate": 1.863859864781807e-05,
      "loss": 1.5121,
      "step": 4430
    },
    {
      "epoch": 0.3405039950829748,
      "grad_norm": 3.2696216106414795,
      "learning_rate": 1.86379840196681e-05,
      "loss": 1.5658,
      "step": 4432
    },
    {
      "epoch": 0.3406576521204671,
      "grad_norm": 3.3706064224243164,
      "learning_rate": 1.863736939151813e-05,
      "loss": 1.4288,
      "step": 4434
    },
    {
      "epoch": 0.34081130915795943,
      "grad_norm": 3.3257944583892822,
      "learning_rate": 1.8636754763368164e-05,
      "loss": 1.4619,
      "step": 4436
    },
    {
      "epoch": 0.34096496619545175,
      "grad_norm": 3.006544589996338,
      "learning_rate": 1.8636140135218194e-05,
      "loss": 1.3658,
      "step": 4438
    },
    {
      "epoch": 0.34111862323294406,
      "grad_norm": 3.39756441116333,
      "learning_rate": 1.8635525507068227e-05,
      "loss": 1.2458,
      "step": 4440
    },
    {
      "epoch": 0.3412722802704364,
      "grad_norm": 3.541398286819458,
      "learning_rate": 1.8634910878918257e-05,
      "loss": 1.4737,
      "step": 4442
    },
    {
      "epoch": 0.3414259373079287,
      "grad_norm": 3.466864824295044,
      "learning_rate": 1.8634296250768286e-05,
      "loss": 1.518,
      "step": 4444
    },
    {
      "epoch": 0.341579594345421,
      "grad_norm": 3.820302963256836,
      "learning_rate": 1.863368162261832e-05,
      "loss": 1.4187,
      "step": 4446
    },
    {
      "epoch": 0.34173325138291333,
      "grad_norm": 3.4347641468048096,
      "learning_rate": 1.863306699446835e-05,
      "loss": 1.3245,
      "step": 4448
    },
    {
      "epoch": 0.34188690842040564,
      "grad_norm": 3.3268067836761475,
      "learning_rate": 1.863245236631838e-05,
      "loss": 1.3987,
      "step": 4450
    },
    {
      "epoch": 0.34204056545789796,
      "grad_norm": 3.1883060932159424,
      "learning_rate": 1.8631837738168412e-05,
      "loss": 1.3047,
      "step": 4452
    },
    {
      "epoch": 0.3421942224953903,
      "grad_norm": 3.4428231716156006,
      "learning_rate": 1.8631223110018438e-05,
      "loss": 1.5088,
      "step": 4454
    },
    {
      "epoch": 0.3423478795328826,
      "grad_norm": 3.522672414779663,
      "learning_rate": 1.863060848186847e-05,
      "loss": 1.3722,
      "step": 4456
    },
    {
      "epoch": 0.3425015365703749,
      "grad_norm": 3.2964699268341064,
      "learning_rate": 1.86299938537185e-05,
      "loss": 1.4182,
      "step": 4458
    },
    {
      "epoch": 0.3426551936078672,
      "grad_norm": 3.4219284057617188,
      "learning_rate": 1.8629379225568534e-05,
      "loss": 1.3521,
      "step": 4460
    },
    {
      "epoch": 0.34280885064535954,
      "grad_norm": 3.896960496902466,
      "learning_rate": 1.8628764597418564e-05,
      "loss": 1.5334,
      "step": 4462
    },
    {
      "epoch": 0.34296250768285186,
      "grad_norm": 3.8269264698028564,
      "learning_rate": 1.8628149969268593e-05,
      "loss": 1.4035,
      "step": 4464
    },
    {
      "epoch": 0.3431161647203442,
      "grad_norm": 3.4882993698120117,
      "learning_rate": 1.8627535341118626e-05,
      "loss": 1.4092,
      "step": 4466
    },
    {
      "epoch": 0.3432698217578365,
      "grad_norm": 3.7473082542419434,
      "learning_rate": 1.8626920712968656e-05,
      "loss": 1.5193,
      "step": 4468
    },
    {
      "epoch": 0.3434234787953288,
      "grad_norm": 3.521932601928711,
      "learning_rate": 1.8626306084818686e-05,
      "loss": 1.3961,
      "step": 4470
    },
    {
      "epoch": 0.3435771358328211,
      "grad_norm": 3.807311534881592,
      "learning_rate": 1.862569145666872e-05,
      "loss": 1.5003,
      "step": 4472
    },
    {
      "epoch": 0.34373079287031344,
      "grad_norm": 3.610978603363037,
      "learning_rate": 1.8625076828518745e-05,
      "loss": 1.4156,
      "step": 4474
    },
    {
      "epoch": 0.34388444990780576,
      "grad_norm": 3.065843343734741,
      "learning_rate": 1.8624462200368778e-05,
      "loss": 1.5464,
      "step": 4476
    },
    {
      "epoch": 0.34403810694529807,
      "grad_norm": 3.488435745239258,
      "learning_rate": 1.862384757221881e-05,
      "loss": 1.4361,
      "step": 4478
    },
    {
      "epoch": 0.3441917639827904,
      "grad_norm": 4.204352378845215,
      "learning_rate": 1.862323294406884e-05,
      "loss": 1.5593,
      "step": 4480
    },
    {
      "epoch": 0.3443454210202827,
      "grad_norm": 3.2348999977111816,
      "learning_rate": 1.862261831591887e-05,
      "loss": 1.3608,
      "step": 4482
    },
    {
      "epoch": 0.344499078057775,
      "grad_norm": 3.1907153129577637,
      "learning_rate": 1.86220036877689e-05,
      "loss": 1.3123,
      "step": 4484
    },
    {
      "epoch": 0.34465273509526734,
      "grad_norm": 3.5538814067840576,
      "learning_rate": 1.8621389059618933e-05,
      "loss": 1.4566,
      "step": 4486
    },
    {
      "epoch": 0.34480639213275965,
      "grad_norm": 3.1382246017456055,
      "learning_rate": 1.8620774431468963e-05,
      "loss": 1.4596,
      "step": 4488
    },
    {
      "epoch": 0.34496004917025197,
      "grad_norm": 3.199798583984375,
      "learning_rate": 1.8620159803318993e-05,
      "loss": 1.313,
      "step": 4490
    },
    {
      "epoch": 0.34511370620774434,
      "grad_norm": 3.067171812057495,
      "learning_rate": 1.8619545175169026e-05,
      "loss": 1.3282,
      "step": 4492
    },
    {
      "epoch": 0.34526736324523666,
      "grad_norm": 3.06500506401062,
      "learning_rate": 1.8618930547019055e-05,
      "loss": 1.5624,
      "step": 4494
    },
    {
      "epoch": 0.345421020282729,
      "grad_norm": 3.439253091812134,
      "learning_rate": 1.8618315918869085e-05,
      "loss": 1.4413,
      "step": 4496
    },
    {
      "epoch": 0.3455746773202213,
      "grad_norm": 3.672922134399414,
      "learning_rate": 1.8617701290719118e-05,
      "loss": 1.2986,
      "step": 4498
    },
    {
      "epoch": 0.3457283343577136,
      "grad_norm": 3.6982719898223877,
      "learning_rate": 1.8617086662569148e-05,
      "loss": 1.503,
      "step": 4500
    },
    {
      "epoch": 0.3458819913952059,
      "grad_norm": 3.3970751762390137,
      "learning_rate": 1.8616472034419178e-05,
      "loss": 1.5047,
      "step": 4502
    },
    {
      "epoch": 0.34603564843269824,
      "grad_norm": 3.2691447734832764,
      "learning_rate": 1.8615857406269207e-05,
      "loss": 1.4006,
      "step": 4504
    },
    {
      "epoch": 0.34618930547019056,
      "grad_norm": 3.3114163875579834,
      "learning_rate": 1.861524277811924e-05,
      "loss": 1.4555,
      "step": 4506
    },
    {
      "epoch": 0.34634296250768287,
      "grad_norm": 3.086209535598755,
      "learning_rate": 1.861462814996927e-05,
      "loss": 1.2947,
      "step": 4508
    },
    {
      "epoch": 0.3464966195451752,
      "grad_norm": 3.334463596343994,
      "learning_rate": 1.86140135218193e-05,
      "loss": 1.4271,
      "step": 4510
    },
    {
      "epoch": 0.3466502765826675,
      "grad_norm": 3.682609796524048,
      "learning_rate": 1.8613398893669333e-05,
      "loss": 1.5643,
      "step": 4512
    },
    {
      "epoch": 0.3468039336201598,
      "grad_norm": 3.6021106243133545,
      "learning_rate": 1.8612784265519362e-05,
      "loss": 1.5502,
      "step": 4514
    },
    {
      "epoch": 0.34695759065765214,
      "grad_norm": 3.319687843322754,
      "learning_rate": 1.8612169637369392e-05,
      "loss": 1.5814,
      "step": 4516
    },
    {
      "epoch": 0.34711124769514445,
      "grad_norm": 3.140190362930298,
      "learning_rate": 1.8611555009219425e-05,
      "loss": 1.4551,
      "step": 4518
    },
    {
      "epoch": 0.34726490473263677,
      "grad_norm": 3.0607619285583496,
      "learning_rate": 1.8610940381069455e-05,
      "loss": 1.3777,
      "step": 4520
    },
    {
      "epoch": 0.3474185617701291,
      "grad_norm": 2.9984660148620605,
      "learning_rate": 1.8610325752919485e-05,
      "loss": 1.4872,
      "step": 4522
    },
    {
      "epoch": 0.3475722188076214,
      "grad_norm": 3.7872512340545654,
      "learning_rate": 1.8609711124769518e-05,
      "loss": 1.4314,
      "step": 4524
    },
    {
      "epoch": 0.3477258758451137,
      "grad_norm": 3.393026113510132,
      "learning_rate": 1.8609096496619547e-05,
      "loss": 1.5033,
      "step": 4526
    },
    {
      "epoch": 0.34787953288260604,
      "grad_norm": 3.257423162460327,
      "learning_rate": 1.8608481868469577e-05,
      "loss": 1.4024,
      "step": 4528
    },
    {
      "epoch": 0.34803318992009835,
      "grad_norm": 3.547748565673828,
      "learning_rate": 1.8607867240319607e-05,
      "loss": 1.5249,
      "step": 4530
    },
    {
      "epoch": 0.34818684695759067,
      "grad_norm": 3.3628032207489014,
      "learning_rate": 1.860725261216964e-05,
      "loss": 1.4415,
      "step": 4532
    },
    {
      "epoch": 0.348340503995083,
      "grad_norm": 3.198662519454956,
      "learning_rate": 1.860663798401967e-05,
      "loss": 1.4072,
      "step": 4534
    },
    {
      "epoch": 0.3484941610325753,
      "grad_norm": 4.2712721824646,
      "learning_rate": 1.86060233558697e-05,
      "loss": 1.4693,
      "step": 4536
    },
    {
      "epoch": 0.3486478180700676,
      "grad_norm": 2.8440463542938232,
      "learning_rate": 1.8605408727719732e-05,
      "loss": 1.4625,
      "step": 4538
    },
    {
      "epoch": 0.34880147510755993,
      "grad_norm": 3.3915908336639404,
      "learning_rate": 1.8604794099569762e-05,
      "loss": 1.2992,
      "step": 4540
    },
    {
      "epoch": 0.34895513214505225,
      "grad_norm": 3.731377601623535,
      "learning_rate": 1.860417947141979e-05,
      "loss": 1.4686,
      "step": 4542
    },
    {
      "epoch": 0.34910878918254457,
      "grad_norm": 3.094580888748169,
      "learning_rate": 1.8603564843269825e-05,
      "loss": 1.5452,
      "step": 4544
    },
    {
      "epoch": 0.3492624462200369,
      "grad_norm": 3.4237303733825684,
      "learning_rate": 1.8602950215119854e-05,
      "loss": 1.498,
      "step": 4546
    },
    {
      "epoch": 0.3494161032575292,
      "grad_norm": 3.4610507488250732,
      "learning_rate": 1.8602335586969884e-05,
      "loss": 1.3702,
      "step": 4548
    },
    {
      "epoch": 0.3495697602950215,
      "grad_norm": 3.4906504154205322,
      "learning_rate": 1.8601720958819917e-05,
      "loss": 1.3723,
      "step": 4550
    },
    {
      "epoch": 0.34972341733251383,
      "grad_norm": 3.214956521987915,
      "learning_rate": 1.8601106330669947e-05,
      "loss": 1.5002,
      "step": 4552
    },
    {
      "epoch": 0.34987707437000615,
      "grad_norm": 3.5659472942352295,
      "learning_rate": 1.8600491702519976e-05,
      "loss": 1.4491,
      "step": 4554
    },
    {
      "epoch": 0.35003073140749846,
      "grad_norm": 3.7536747455596924,
      "learning_rate": 1.8599877074370006e-05,
      "loss": 1.6745,
      "step": 4556
    },
    {
      "epoch": 0.3501843884449908,
      "grad_norm": 3.7435832023620605,
      "learning_rate": 1.859926244622004e-05,
      "loss": 1.3983,
      "step": 4558
    },
    {
      "epoch": 0.3503380454824831,
      "grad_norm": 3.1772406101226807,
      "learning_rate": 1.859864781807007e-05,
      "loss": 1.2819,
      "step": 4560
    },
    {
      "epoch": 0.3504917025199754,
      "grad_norm": 3.307325839996338,
      "learning_rate": 1.85980331899201e-05,
      "loss": 1.4716,
      "step": 4562
    },
    {
      "epoch": 0.35064535955746773,
      "grad_norm": 3.418013334274292,
      "learning_rate": 1.859741856177013e-05,
      "loss": 1.3982,
      "step": 4564
    },
    {
      "epoch": 0.35079901659496004,
      "grad_norm": 3.2819008827209473,
      "learning_rate": 1.859680393362016e-05,
      "loss": 1.468,
      "step": 4566
    },
    {
      "epoch": 0.35095267363245236,
      "grad_norm": 3.145008087158203,
      "learning_rate": 1.859618930547019e-05,
      "loss": 1.3179,
      "step": 4568
    },
    {
      "epoch": 0.3511063306699447,
      "grad_norm": 3.330179452896118,
      "learning_rate": 1.8595574677320224e-05,
      "loss": 1.3268,
      "step": 4570
    },
    {
      "epoch": 0.351259987707437,
      "grad_norm": 3.0777249336242676,
      "learning_rate": 1.8594960049170254e-05,
      "loss": 1.4471,
      "step": 4572
    },
    {
      "epoch": 0.3514136447449293,
      "grad_norm": 3.1006267070770264,
      "learning_rate": 1.8594345421020283e-05,
      "loss": 1.3894,
      "step": 4574
    },
    {
      "epoch": 0.3515673017824216,
      "grad_norm": 3.358553409576416,
      "learning_rate": 1.8593730792870316e-05,
      "loss": 1.4034,
      "step": 4576
    },
    {
      "epoch": 0.35172095881991394,
      "grad_norm": 3.211599349975586,
      "learning_rate": 1.8593116164720346e-05,
      "loss": 1.5581,
      "step": 4578
    },
    {
      "epoch": 0.35187461585740626,
      "grad_norm": 3.6108851432800293,
      "learning_rate": 1.8592501536570376e-05,
      "loss": 1.5294,
      "step": 4580
    },
    {
      "epoch": 0.3520282728948986,
      "grad_norm": 3.4219422340393066,
      "learning_rate": 1.8591886908420406e-05,
      "loss": 1.2853,
      "step": 4582
    },
    {
      "epoch": 0.3521819299323909,
      "grad_norm": 3.706087112426758,
      "learning_rate": 1.859127228027044e-05,
      "loss": 1.4577,
      "step": 4584
    },
    {
      "epoch": 0.3523355869698832,
      "grad_norm": 4.492853164672852,
      "learning_rate": 1.8590657652120468e-05,
      "loss": 1.5574,
      "step": 4586
    },
    {
      "epoch": 0.3524892440073755,
      "grad_norm": 2.998133420944214,
      "learning_rate": 1.8590043023970498e-05,
      "loss": 1.39,
      "step": 4588
    },
    {
      "epoch": 0.35264290104486784,
      "grad_norm": 4.351780414581299,
      "learning_rate": 1.858942839582053e-05,
      "loss": 1.4118,
      "step": 4590
    },
    {
      "epoch": 0.35279655808236016,
      "grad_norm": 3.8137764930725098,
      "learning_rate": 1.858881376767056e-05,
      "loss": 1.324,
      "step": 4592
    },
    {
      "epoch": 0.3529502151198525,
      "grad_norm": 3.1416056156158447,
      "learning_rate": 1.858819913952059e-05,
      "loss": 1.3979,
      "step": 4594
    },
    {
      "epoch": 0.3531038721573448,
      "grad_norm": 3.3871965408325195,
      "learning_rate": 1.8587584511370623e-05,
      "loss": 1.3873,
      "step": 4596
    },
    {
      "epoch": 0.3532575291948371,
      "grad_norm": 3.1268036365509033,
      "learning_rate": 1.8586969883220653e-05,
      "loss": 1.4496,
      "step": 4598
    },
    {
      "epoch": 0.3534111862323294,
      "grad_norm": 4.005764007568359,
      "learning_rate": 1.8586355255070683e-05,
      "loss": 1.5423,
      "step": 4600
    },
    {
      "epoch": 0.35356484326982174,
      "grad_norm": 3.157153367996216,
      "learning_rate": 1.8585740626920713e-05,
      "loss": 1.3039,
      "step": 4602
    },
    {
      "epoch": 0.35371850030731405,
      "grad_norm": 3.5315701961517334,
      "learning_rate": 1.8585125998770746e-05,
      "loss": 1.3771,
      "step": 4604
    },
    {
      "epoch": 0.35387215734480637,
      "grad_norm": 3.5096662044525146,
      "learning_rate": 1.8584511370620775e-05,
      "loss": 1.4756,
      "step": 4606
    },
    {
      "epoch": 0.3540258143822987,
      "grad_norm": 3.7469310760498047,
      "learning_rate": 1.8583896742470805e-05,
      "loss": 1.5494,
      "step": 4608
    },
    {
      "epoch": 0.354179471419791,
      "grad_norm": 3.916029691696167,
      "learning_rate": 1.8583282114320838e-05,
      "loss": 1.4195,
      "step": 4610
    },
    {
      "epoch": 0.3543331284572833,
      "grad_norm": 3.3973281383514404,
      "learning_rate": 1.8582667486170868e-05,
      "loss": 1.4924,
      "step": 4612
    },
    {
      "epoch": 0.35448678549477564,
      "grad_norm": 3.974994421005249,
      "learning_rate": 1.8582052858020897e-05,
      "loss": 1.3841,
      "step": 4614
    },
    {
      "epoch": 0.35464044253226795,
      "grad_norm": 3.317247152328491,
      "learning_rate": 1.858143822987093e-05,
      "loss": 1.3693,
      "step": 4616
    },
    {
      "epoch": 0.35479409956976027,
      "grad_norm": 3.2448232173919678,
      "learning_rate": 1.858082360172096e-05,
      "loss": 1.5438,
      "step": 4618
    },
    {
      "epoch": 0.3549477566072526,
      "grad_norm": 3.4983530044555664,
      "learning_rate": 1.858020897357099e-05,
      "loss": 1.5307,
      "step": 4620
    },
    {
      "epoch": 0.35510141364474496,
      "grad_norm": 4.106202602386475,
      "learning_rate": 1.8579594345421023e-05,
      "loss": 1.3608,
      "step": 4622
    },
    {
      "epoch": 0.3552550706822373,
      "grad_norm": 3.7495827674865723,
      "learning_rate": 1.8578979717271053e-05,
      "loss": 1.4775,
      "step": 4624
    },
    {
      "epoch": 0.3554087277197296,
      "grad_norm": 3.6079421043395996,
      "learning_rate": 1.8578365089121086e-05,
      "loss": 1.3453,
      "step": 4626
    },
    {
      "epoch": 0.3555623847572219,
      "grad_norm": 3.424074649810791,
      "learning_rate": 1.8577750460971112e-05,
      "loss": 1.4793,
      "step": 4628
    },
    {
      "epoch": 0.3557160417947142,
      "grad_norm": 3.2874410152435303,
      "learning_rate": 1.8577135832821145e-05,
      "loss": 1.4401,
      "step": 4630
    },
    {
      "epoch": 0.35586969883220654,
      "grad_norm": 3.3648531436920166,
      "learning_rate": 1.8576521204671175e-05,
      "loss": 1.3748,
      "step": 4632
    },
    {
      "epoch": 0.35602335586969885,
      "grad_norm": 3.250709056854248,
      "learning_rate": 1.8575906576521204e-05,
      "loss": 1.4623,
      "step": 4634
    },
    {
      "epoch": 0.35617701290719117,
      "grad_norm": 3.538073778152466,
      "learning_rate": 1.8575291948371237e-05,
      "loss": 1.5085,
      "step": 4636
    },
    {
      "epoch": 0.3563306699446835,
      "grad_norm": 3.82171630859375,
      "learning_rate": 1.8574677320221267e-05,
      "loss": 1.5804,
      "step": 4638
    },
    {
      "epoch": 0.3564843269821758,
      "grad_norm": 3.834066390991211,
      "learning_rate": 1.8574062692071297e-05,
      "loss": 1.4279,
      "step": 4640
    },
    {
      "epoch": 0.3566379840196681,
      "grad_norm": 3.344313144683838,
      "learning_rate": 1.857344806392133e-05,
      "loss": 1.3728,
      "step": 4642
    },
    {
      "epoch": 0.35679164105716044,
      "grad_norm": 3.3498642444610596,
      "learning_rate": 1.857283343577136e-05,
      "loss": 1.3697,
      "step": 4644
    },
    {
      "epoch": 0.35694529809465275,
      "grad_norm": 3.3379592895507812,
      "learning_rate": 1.8572218807621393e-05,
      "loss": 1.2771,
      "step": 4646
    },
    {
      "epoch": 0.35709895513214507,
      "grad_norm": 3.0588574409484863,
      "learning_rate": 1.8571604179471422e-05,
      "loss": 1.4029,
      "step": 4648
    },
    {
      "epoch": 0.3572526121696374,
      "grad_norm": 3.230083703994751,
      "learning_rate": 1.8570989551321452e-05,
      "loss": 1.5251,
      "step": 4650
    },
    {
      "epoch": 0.3574062692071297,
      "grad_norm": 3.0492098331451416,
      "learning_rate": 1.8570374923171485e-05,
      "loss": 1.4231,
      "step": 4652
    },
    {
      "epoch": 0.357559926244622,
      "grad_norm": 3.1024155616760254,
      "learning_rate": 1.856976029502151e-05,
      "loss": 1.3485,
      "step": 4654
    },
    {
      "epoch": 0.35771358328211433,
      "grad_norm": 3.091562509536743,
      "learning_rate": 1.8569145666871544e-05,
      "loss": 1.4922,
      "step": 4656
    },
    {
      "epoch": 0.35786724031960665,
      "grad_norm": 3.7615408897399902,
      "learning_rate": 1.8568531038721574e-05,
      "loss": 1.4149,
      "step": 4658
    },
    {
      "epoch": 0.35802089735709897,
      "grad_norm": 2.9202842712402344,
      "learning_rate": 1.8567916410571604e-05,
      "loss": 1.3638,
      "step": 4660
    },
    {
      "epoch": 0.3581745543945913,
      "grad_norm": 3.4049057960510254,
      "learning_rate": 1.8567301782421637e-05,
      "loss": 1.3405,
      "step": 4662
    },
    {
      "epoch": 0.3583282114320836,
      "grad_norm": 3.389155864715576,
      "learning_rate": 1.8566687154271667e-05,
      "loss": 1.4706,
      "step": 4664
    },
    {
      "epoch": 0.3584818684695759,
      "grad_norm": 3.2135565280914307,
      "learning_rate": 1.85660725261217e-05,
      "loss": 1.3997,
      "step": 4666
    },
    {
      "epoch": 0.35863552550706823,
      "grad_norm": 3.588104009628296,
      "learning_rate": 1.856545789797173e-05,
      "loss": 1.4596,
      "step": 4668
    },
    {
      "epoch": 0.35878918254456055,
      "grad_norm": 3.40895676612854,
      "learning_rate": 1.856484326982176e-05,
      "loss": 1.4499,
      "step": 4670
    },
    {
      "epoch": 0.35894283958205286,
      "grad_norm": 3.071735143661499,
      "learning_rate": 1.8564228641671792e-05,
      "loss": 1.369,
      "step": 4672
    },
    {
      "epoch": 0.3590964966195452,
      "grad_norm": 3.500002861022949,
      "learning_rate": 1.8563614013521822e-05,
      "loss": 1.3623,
      "step": 4674
    },
    {
      "epoch": 0.3592501536570375,
      "grad_norm": 3.3356125354766846,
      "learning_rate": 1.856299938537185e-05,
      "loss": 1.4402,
      "step": 4676
    },
    {
      "epoch": 0.3594038106945298,
      "grad_norm": 3.249798059463501,
      "learning_rate": 1.8562384757221885e-05,
      "loss": 1.335,
      "step": 4678
    },
    {
      "epoch": 0.35955746773202213,
      "grad_norm": 3.477461099624634,
      "learning_rate": 1.856177012907191e-05,
      "loss": 1.4934,
      "step": 4680
    },
    {
      "epoch": 0.35971112476951445,
      "grad_norm": 3.0536158084869385,
      "learning_rate": 1.8561155500921944e-05,
      "loss": 1.3237,
      "step": 4682
    },
    {
      "epoch": 0.35986478180700676,
      "grad_norm": 2.8353586196899414,
      "learning_rate": 1.8560540872771974e-05,
      "loss": 1.4192,
      "step": 4684
    },
    {
      "epoch": 0.3600184388444991,
      "grad_norm": 3.6377627849578857,
      "learning_rate": 1.8559926244622003e-05,
      "loss": 1.5224,
      "step": 4686
    },
    {
      "epoch": 0.3601720958819914,
      "grad_norm": 3.3342795372009277,
      "learning_rate": 1.8559311616472036e-05,
      "loss": 1.5283,
      "step": 4688
    },
    {
      "epoch": 0.3603257529194837,
      "grad_norm": 3.5192248821258545,
      "learning_rate": 1.8558696988322066e-05,
      "loss": 1.4173,
      "step": 4690
    },
    {
      "epoch": 0.360479409956976,
      "grad_norm": 3.1181795597076416,
      "learning_rate": 1.85580823601721e-05,
      "loss": 1.3981,
      "step": 4692
    },
    {
      "epoch": 0.36063306699446834,
      "grad_norm": 3.264928102493286,
      "learning_rate": 1.855746773202213e-05,
      "loss": 1.3971,
      "step": 4694
    },
    {
      "epoch": 0.36078672403196066,
      "grad_norm": 3.1449787616729736,
      "learning_rate": 1.855685310387216e-05,
      "loss": 1.3952,
      "step": 4696
    },
    {
      "epoch": 0.360940381069453,
      "grad_norm": 3.4378745555877686,
      "learning_rate": 1.855623847572219e-05,
      "loss": 1.4382,
      "step": 4698
    },
    {
      "epoch": 0.3610940381069453,
      "grad_norm": 3.424116373062134,
      "learning_rate": 1.8555623847572218e-05,
      "loss": 1.3557,
      "step": 4700
    },
    {
      "epoch": 0.3612476951444376,
      "grad_norm": 3.6776251792907715,
      "learning_rate": 1.855500921942225e-05,
      "loss": 1.5211,
      "step": 4702
    },
    {
      "epoch": 0.3614013521819299,
      "grad_norm": 3.3191077709198,
      "learning_rate": 1.855439459127228e-05,
      "loss": 1.2799,
      "step": 4704
    },
    {
      "epoch": 0.36155500921942224,
      "grad_norm": 3.217733383178711,
      "learning_rate": 1.855377996312231e-05,
      "loss": 1.3677,
      "step": 4706
    },
    {
      "epoch": 0.36170866625691456,
      "grad_norm": 3.2171108722686768,
      "learning_rate": 1.8553165334972343e-05,
      "loss": 1.5173,
      "step": 4708
    },
    {
      "epoch": 0.3618623232944069,
      "grad_norm": 3.5864431858062744,
      "learning_rate": 1.8552550706822373e-05,
      "loss": 1.4608,
      "step": 4710
    },
    {
      "epoch": 0.3620159803318992,
      "grad_norm": 3.2654552459716797,
      "learning_rate": 1.8551936078672406e-05,
      "loss": 1.3668,
      "step": 4712
    },
    {
      "epoch": 0.3621696373693915,
      "grad_norm": 3.540778160095215,
      "learning_rate": 1.8551321450522436e-05,
      "loss": 1.346,
      "step": 4714
    },
    {
      "epoch": 0.3623232944068838,
      "grad_norm": 3.4130847454071045,
      "learning_rate": 1.8550706822372465e-05,
      "loss": 1.3636,
      "step": 4716
    },
    {
      "epoch": 0.36247695144437614,
      "grad_norm": 3.9183061122894287,
      "learning_rate": 1.85500921942225e-05,
      "loss": 1.3392,
      "step": 4718
    },
    {
      "epoch": 0.36263060848186845,
      "grad_norm": 3.619670867919922,
      "learning_rate": 1.8549477566072528e-05,
      "loss": 1.4334,
      "step": 4720
    },
    {
      "epoch": 0.36278426551936077,
      "grad_norm": 3.4983813762664795,
      "learning_rate": 1.8548862937922558e-05,
      "loss": 1.3492,
      "step": 4722
    },
    {
      "epoch": 0.3629379225568531,
      "grad_norm": 3.3781394958496094,
      "learning_rate": 1.854824830977259e-05,
      "loss": 1.4709,
      "step": 4724
    },
    {
      "epoch": 0.3630915795943454,
      "grad_norm": 3.9609100818634033,
      "learning_rate": 1.8547633681622617e-05,
      "loss": 1.3665,
      "step": 4726
    },
    {
      "epoch": 0.3632452366318377,
      "grad_norm": 3.8073222637176514,
      "learning_rate": 1.854701905347265e-05,
      "loss": 1.441,
      "step": 4728
    },
    {
      "epoch": 0.36339889366933004,
      "grad_norm": 3.3414063453674316,
      "learning_rate": 1.854640442532268e-05,
      "loss": 1.3758,
      "step": 4730
    },
    {
      "epoch": 0.36355255070682235,
      "grad_norm": 3.2114644050598145,
      "learning_rate": 1.8545789797172713e-05,
      "loss": 1.3025,
      "step": 4732
    },
    {
      "epoch": 0.36370620774431467,
      "grad_norm": 3.299943447113037,
      "learning_rate": 1.8545175169022743e-05,
      "loss": 1.4549,
      "step": 4734
    },
    {
      "epoch": 0.363859864781807,
      "grad_norm": 3.959162473678589,
      "learning_rate": 1.8544560540872772e-05,
      "loss": 1.431,
      "step": 4736
    },
    {
      "epoch": 0.3640135218192993,
      "grad_norm": 3.4247779846191406,
      "learning_rate": 1.8543945912722806e-05,
      "loss": 1.5114,
      "step": 4738
    },
    {
      "epoch": 0.3641671788567916,
      "grad_norm": 3.407259941101074,
      "learning_rate": 1.8543331284572835e-05,
      "loss": 1.5001,
      "step": 4740
    },
    {
      "epoch": 0.36432083589428393,
      "grad_norm": 3.3463499546051025,
      "learning_rate": 1.8542716656422865e-05,
      "loss": 1.3259,
      "step": 4742
    },
    {
      "epoch": 0.36447449293177625,
      "grad_norm": 2.8831546306610107,
      "learning_rate": 1.8542102028272898e-05,
      "loss": 1.4593,
      "step": 4744
    },
    {
      "epoch": 0.36462814996926857,
      "grad_norm": 3.3898556232452393,
      "learning_rate": 1.8541487400122928e-05,
      "loss": 1.4173,
      "step": 4746
    },
    {
      "epoch": 0.3647818070067609,
      "grad_norm": 3.1895546913146973,
      "learning_rate": 1.8540872771972957e-05,
      "loss": 1.4014,
      "step": 4748
    },
    {
      "epoch": 0.3649354640442532,
      "grad_norm": 3.217935085296631,
      "learning_rate": 1.854025814382299e-05,
      "loss": 1.3577,
      "step": 4750
    },
    {
      "epoch": 0.36508912108174557,
      "grad_norm": 3.2080564498901367,
      "learning_rate": 1.853964351567302e-05,
      "loss": 1.2894,
      "step": 4752
    },
    {
      "epoch": 0.3652427781192379,
      "grad_norm": 3.4375319480895996,
      "learning_rate": 1.853902888752305e-05,
      "loss": 1.4444,
      "step": 4754
    },
    {
      "epoch": 0.3653964351567302,
      "grad_norm": 24.822725296020508,
      "learning_rate": 1.853841425937308e-05,
      "loss": 1.3199,
      "step": 4756
    },
    {
      "epoch": 0.3655500921942225,
      "grad_norm": 3.4176087379455566,
      "learning_rate": 1.8537799631223113e-05,
      "loss": 1.4149,
      "step": 4758
    },
    {
      "epoch": 0.36570374923171484,
      "grad_norm": 3.6621382236480713,
      "learning_rate": 1.8537185003073142e-05,
      "loss": 1.4023,
      "step": 4760
    },
    {
      "epoch": 0.36585740626920715,
      "grad_norm": 3.013288974761963,
      "learning_rate": 1.8536570374923172e-05,
      "loss": 1.4663,
      "step": 4762
    },
    {
      "epoch": 0.36601106330669947,
      "grad_norm": 3.411132335662842,
      "learning_rate": 1.8535955746773205e-05,
      "loss": 1.379,
      "step": 4764
    },
    {
      "epoch": 0.3661647203441918,
      "grad_norm": 3.7102386951446533,
      "learning_rate": 1.8535341118623235e-05,
      "loss": 1.4119,
      "step": 4766
    },
    {
      "epoch": 0.3663183773816841,
      "grad_norm": 3.6515626907348633,
      "learning_rate": 1.8534726490473264e-05,
      "loss": 1.4011,
      "step": 4768
    },
    {
      "epoch": 0.3664720344191764,
      "grad_norm": 2.929818630218506,
      "learning_rate": 1.8534111862323297e-05,
      "loss": 1.3097,
      "step": 4770
    },
    {
      "epoch": 0.36662569145666873,
      "grad_norm": 3.2975289821624756,
      "learning_rate": 1.8533497234173327e-05,
      "loss": 1.3629,
      "step": 4772
    },
    {
      "epoch": 0.36677934849416105,
      "grad_norm": 3.4439969062805176,
      "learning_rate": 1.8532882606023357e-05,
      "loss": 1.3984,
      "step": 4774
    },
    {
      "epoch": 0.36693300553165337,
      "grad_norm": 3.5669538974761963,
      "learning_rate": 1.853226797787339e-05,
      "loss": 1.4848,
      "step": 4776
    },
    {
      "epoch": 0.3670866625691457,
      "grad_norm": 3.3473563194274902,
      "learning_rate": 1.853165334972342e-05,
      "loss": 1.3792,
      "step": 4778
    },
    {
      "epoch": 0.367240319606638,
      "grad_norm": 3.0193610191345215,
      "learning_rate": 1.853103872157345e-05,
      "loss": 1.3099,
      "step": 4780
    },
    {
      "epoch": 0.3673939766441303,
      "grad_norm": 3.254918336868286,
      "learning_rate": 1.853042409342348e-05,
      "loss": 1.4501,
      "step": 4782
    },
    {
      "epoch": 0.36754763368162263,
      "grad_norm": 3.2965073585510254,
      "learning_rate": 1.8529809465273512e-05,
      "loss": 1.5442,
      "step": 4784
    },
    {
      "epoch": 0.36770129071911495,
      "grad_norm": 3.3364098072052,
      "learning_rate": 1.852919483712354e-05,
      "loss": 1.3758,
      "step": 4786
    },
    {
      "epoch": 0.36785494775660726,
      "grad_norm": 3.033813714981079,
      "learning_rate": 1.852858020897357e-05,
      "loss": 1.4798,
      "step": 4788
    },
    {
      "epoch": 0.3680086047940996,
      "grad_norm": 3.194025993347168,
      "learning_rate": 1.8527965580823604e-05,
      "loss": 1.2675,
      "step": 4790
    },
    {
      "epoch": 0.3681622618315919,
      "grad_norm": 3.411519765853882,
      "learning_rate": 1.8527350952673634e-05,
      "loss": 1.5546,
      "step": 4792
    },
    {
      "epoch": 0.3683159188690842,
      "grad_norm": 3.634873628616333,
      "learning_rate": 1.8526736324523664e-05,
      "loss": 1.5041,
      "step": 4794
    },
    {
      "epoch": 0.36846957590657653,
      "grad_norm": 2.9618473052978516,
      "learning_rate": 1.8526121696373697e-05,
      "loss": 1.4126,
      "step": 4796
    },
    {
      "epoch": 0.36862323294406885,
      "grad_norm": 3.2538132667541504,
      "learning_rate": 1.8525507068223727e-05,
      "loss": 1.4021,
      "step": 4798
    },
    {
      "epoch": 0.36877688998156116,
      "grad_norm": 3.3406293392181396,
      "learning_rate": 1.8524892440073756e-05,
      "loss": 1.349,
      "step": 4800
    },
    {
      "epoch": 0.3689305470190535,
      "grad_norm": 3.3935546875,
      "learning_rate": 1.852427781192379e-05,
      "loss": 1.2304,
      "step": 4802
    },
    {
      "epoch": 0.3690842040565458,
      "grad_norm": 3.685398578643799,
      "learning_rate": 1.852366318377382e-05,
      "loss": 1.3805,
      "step": 4804
    },
    {
      "epoch": 0.3692378610940381,
      "grad_norm": 3.6336817741394043,
      "learning_rate": 1.852304855562385e-05,
      "loss": 1.5106,
      "step": 4806
    },
    {
      "epoch": 0.3693915181315304,
      "grad_norm": 3.2699592113494873,
      "learning_rate": 1.852243392747388e-05,
      "loss": 1.3488,
      "step": 4808
    },
    {
      "epoch": 0.36954517516902274,
      "grad_norm": 3.5057644844055176,
      "learning_rate": 1.852181929932391e-05,
      "loss": 1.4388,
      "step": 4810
    },
    {
      "epoch": 0.36969883220651506,
      "grad_norm": 3.1006228923797607,
      "learning_rate": 1.852120467117394e-05,
      "loss": 1.6204,
      "step": 4812
    },
    {
      "epoch": 0.3698524892440074,
      "grad_norm": 3.2444615364074707,
      "learning_rate": 1.852059004302397e-05,
      "loss": 1.3704,
      "step": 4814
    },
    {
      "epoch": 0.3700061462814997,
      "grad_norm": 3.3723206520080566,
      "learning_rate": 1.8519975414874004e-05,
      "loss": 1.3931,
      "step": 4816
    },
    {
      "epoch": 0.370159803318992,
      "grad_norm": 3.3920679092407227,
      "learning_rate": 1.8519360786724034e-05,
      "loss": 1.2844,
      "step": 4818
    },
    {
      "epoch": 0.3703134603564843,
      "grad_norm": 3.732646942138672,
      "learning_rate": 1.8518746158574063e-05,
      "loss": 1.3408,
      "step": 4820
    },
    {
      "epoch": 0.37046711739397664,
      "grad_norm": 3.0246357917785645,
      "learning_rate": 1.8518131530424096e-05,
      "loss": 1.2567,
      "step": 4822
    },
    {
      "epoch": 0.37062077443146896,
      "grad_norm": 3.309549331665039,
      "learning_rate": 1.8517516902274126e-05,
      "loss": 1.3551,
      "step": 4824
    },
    {
      "epoch": 0.3707744314689613,
      "grad_norm": 3.1484642028808594,
      "learning_rate": 1.8516902274124156e-05,
      "loss": 1.4835,
      "step": 4826
    },
    {
      "epoch": 0.3709280885064536,
      "grad_norm": 3.240321636199951,
      "learning_rate": 1.8516287645974185e-05,
      "loss": 1.4054,
      "step": 4828
    },
    {
      "epoch": 0.3710817455439459,
      "grad_norm": 3.533933162689209,
      "learning_rate": 1.851567301782422e-05,
      "loss": 1.4685,
      "step": 4830
    },
    {
      "epoch": 0.3712354025814382,
      "grad_norm": 3.4116733074188232,
      "learning_rate": 1.8515058389674248e-05,
      "loss": 1.3839,
      "step": 4832
    },
    {
      "epoch": 0.37138905961893054,
      "grad_norm": 3.154917001724243,
      "learning_rate": 1.8514443761524278e-05,
      "loss": 1.4637,
      "step": 4834
    },
    {
      "epoch": 0.37154271665642286,
      "grad_norm": 2.889686107635498,
      "learning_rate": 1.851382913337431e-05,
      "loss": 1.38,
      "step": 4836
    },
    {
      "epoch": 0.37169637369391517,
      "grad_norm": 3.351166009902954,
      "learning_rate": 1.851321450522434e-05,
      "loss": 1.4498,
      "step": 4838
    },
    {
      "epoch": 0.3718500307314075,
      "grad_norm": 3.114797592163086,
      "learning_rate": 1.851259987707437e-05,
      "loss": 1.3205,
      "step": 4840
    },
    {
      "epoch": 0.3720036877688998,
      "grad_norm": 3.2657954692840576,
      "learning_rate": 1.8511985248924403e-05,
      "loss": 1.3892,
      "step": 4842
    },
    {
      "epoch": 0.3721573448063921,
      "grad_norm": 3.140523910522461,
      "learning_rate": 1.8511370620774433e-05,
      "loss": 1.5039,
      "step": 4844
    },
    {
      "epoch": 0.37231100184388444,
      "grad_norm": 3.459165334701538,
      "learning_rate": 1.8510755992624463e-05,
      "loss": 1.4855,
      "step": 4846
    },
    {
      "epoch": 0.37246465888137675,
      "grad_norm": 3.286754846572876,
      "learning_rate": 1.8510141364474496e-05,
      "loss": 1.4934,
      "step": 4848
    },
    {
      "epoch": 0.37261831591886907,
      "grad_norm": 3.208974599838257,
      "learning_rate": 1.8509526736324525e-05,
      "loss": 1.3993,
      "step": 4850
    },
    {
      "epoch": 0.3727719729563614,
      "grad_norm": 3.7249045372009277,
      "learning_rate": 1.8508912108174555e-05,
      "loss": 1.5347,
      "step": 4852
    },
    {
      "epoch": 0.3729256299938537,
      "grad_norm": 2.87788724899292,
      "learning_rate": 1.8508297480024585e-05,
      "loss": 1.2677,
      "step": 4854
    },
    {
      "epoch": 0.373079287031346,
      "grad_norm": 3.778860330581665,
      "learning_rate": 1.8507682851874618e-05,
      "loss": 1.3966,
      "step": 4856
    },
    {
      "epoch": 0.37323294406883833,
      "grad_norm": 3.1479995250701904,
      "learning_rate": 1.8507068223724648e-05,
      "loss": 1.268,
      "step": 4858
    },
    {
      "epoch": 0.37338660110633065,
      "grad_norm": 3.2577826976776123,
      "learning_rate": 1.8506453595574677e-05,
      "loss": 1.4929,
      "step": 4860
    },
    {
      "epoch": 0.37354025814382297,
      "grad_norm": 3.440727472305298,
      "learning_rate": 1.850583896742471e-05,
      "loss": 1.4138,
      "step": 4862
    },
    {
      "epoch": 0.3736939151813153,
      "grad_norm": 4.165016174316406,
      "learning_rate": 1.850522433927474e-05,
      "loss": 1.4426,
      "step": 4864
    },
    {
      "epoch": 0.3738475722188076,
      "grad_norm": 3.2028725147247314,
      "learning_rate": 1.850460971112477e-05,
      "loss": 1.2641,
      "step": 4866
    },
    {
      "epoch": 0.3740012292562999,
      "grad_norm": 2.9112865924835205,
      "learning_rate": 1.8503995082974803e-05,
      "loss": 1.4361,
      "step": 4868
    },
    {
      "epoch": 0.37415488629379223,
      "grad_norm": 3.539193868637085,
      "learning_rate": 1.8503380454824832e-05,
      "loss": 1.3404,
      "step": 4870
    },
    {
      "epoch": 0.37430854333128455,
      "grad_norm": 3.1732802391052246,
      "learning_rate": 1.8502765826674862e-05,
      "loss": 1.4955,
      "step": 4872
    },
    {
      "epoch": 0.37446220036877687,
      "grad_norm": 3.662275791168213,
      "learning_rate": 1.8502151198524895e-05,
      "loss": 1.4514,
      "step": 4874
    },
    {
      "epoch": 0.3746158574062692,
      "grad_norm": 3.213059186935425,
      "learning_rate": 1.8501536570374925e-05,
      "loss": 1.517,
      "step": 4876
    },
    {
      "epoch": 0.3747695144437615,
      "grad_norm": 3.332075834274292,
      "learning_rate": 1.8500921942224958e-05,
      "loss": 1.4251,
      "step": 4878
    },
    {
      "epoch": 0.3749231714812538,
      "grad_norm": 3.2899796962738037,
      "learning_rate": 1.8500307314074984e-05,
      "loss": 1.3975,
      "step": 4880
    },
    {
      "epoch": 0.3750768285187462,
      "grad_norm": 3.4021992683410645,
      "learning_rate": 1.8499692685925017e-05,
      "loss": 1.4664,
      "step": 4882
    },
    {
      "epoch": 0.3752304855562385,
      "grad_norm": 3.553896188735962,
      "learning_rate": 1.8499078057775047e-05,
      "loss": 1.356,
      "step": 4884
    },
    {
      "epoch": 0.3753841425937308,
      "grad_norm": 3.311842203140259,
      "learning_rate": 1.8498463429625077e-05,
      "loss": 1.4542,
      "step": 4886
    },
    {
      "epoch": 0.37553779963122313,
      "grad_norm": 3.7482736110687256,
      "learning_rate": 1.849784880147511e-05,
      "loss": 1.4163,
      "step": 4888
    },
    {
      "epoch": 0.37569145666871545,
      "grad_norm": 3.5204427242279053,
      "learning_rate": 1.849723417332514e-05,
      "loss": 1.3673,
      "step": 4890
    },
    {
      "epoch": 0.37584511370620777,
      "grad_norm": 3.4200925827026367,
      "learning_rate": 1.849661954517517e-05,
      "loss": 1.3482,
      "step": 4892
    },
    {
      "epoch": 0.3759987707437001,
      "grad_norm": 3.378998279571533,
      "learning_rate": 1.8496004917025202e-05,
      "loss": 1.5857,
      "step": 4894
    },
    {
      "epoch": 0.3761524277811924,
      "grad_norm": 3.0784049034118652,
      "learning_rate": 1.8495390288875232e-05,
      "loss": 1.4947,
      "step": 4896
    },
    {
      "epoch": 0.3763060848186847,
      "grad_norm": 3.1790988445281982,
      "learning_rate": 1.8494775660725265e-05,
      "loss": 1.4122,
      "step": 4898
    },
    {
      "epoch": 0.37645974185617703,
      "grad_norm": 3.191070318222046,
      "learning_rate": 1.8494161032575295e-05,
      "loss": 1.3629,
      "step": 4900
    },
    {
      "epoch": 0.37661339889366935,
      "grad_norm": 3.2839808464050293,
      "learning_rate": 1.8493546404425324e-05,
      "loss": 1.3496,
      "step": 4902
    },
    {
      "epoch": 0.37676705593116167,
      "grad_norm": 3.522803783416748,
      "learning_rate": 1.8492931776275357e-05,
      "loss": 1.4566,
      "step": 4904
    },
    {
      "epoch": 0.376920712968654,
      "grad_norm": 3.301510810852051,
      "learning_rate": 1.8492317148125384e-05,
      "loss": 1.4704,
      "step": 4906
    },
    {
      "epoch": 0.3770743700061463,
      "grad_norm": 3.3755650520324707,
      "learning_rate": 1.8491702519975417e-05,
      "loss": 1.33,
      "step": 4908
    },
    {
      "epoch": 0.3772280270436386,
      "grad_norm": 3.328902244567871,
      "learning_rate": 1.8491087891825446e-05,
      "loss": 1.2691,
      "step": 4910
    },
    {
      "epoch": 0.37738168408113093,
      "grad_norm": 3.257596254348755,
      "learning_rate": 1.8490473263675476e-05,
      "loss": 1.5409,
      "step": 4912
    },
    {
      "epoch": 0.37753534111862325,
      "grad_norm": 3.0310218334198,
      "learning_rate": 1.848985863552551e-05,
      "loss": 1.3193,
      "step": 4914
    },
    {
      "epoch": 0.37768899815611556,
      "grad_norm": 3.495089054107666,
      "learning_rate": 1.848924400737554e-05,
      "loss": 1.5036,
      "step": 4916
    },
    {
      "epoch": 0.3778426551936079,
      "grad_norm": 3.5136234760284424,
      "learning_rate": 1.8488629379225572e-05,
      "loss": 1.4914,
      "step": 4918
    },
    {
      "epoch": 0.3779963122311002,
      "grad_norm": 2.943639039993286,
      "learning_rate": 1.84880147510756e-05,
      "loss": 1.3518,
      "step": 4920
    },
    {
      "epoch": 0.3781499692685925,
      "grad_norm": 3.4134724140167236,
      "learning_rate": 1.848740012292563e-05,
      "loss": 1.3874,
      "step": 4922
    },
    {
      "epoch": 0.37830362630608483,
      "grad_norm": 3.4609227180480957,
      "learning_rate": 1.8486785494775664e-05,
      "loss": 1.4883,
      "step": 4924
    },
    {
      "epoch": 0.37845728334357714,
      "grad_norm": 3.085495710372925,
      "learning_rate": 1.848617086662569e-05,
      "loss": 1.429,
      "step": 4926
    },
    {
      "epoch": 0.37861094038106946,
      "grad_norm": 3.304070234298706,
      "learning_rate": 1.8485556238475724e-05,
      "loss": 1.5389,
      "step": 4928
    },
    {
      "epoch": 0.3787645974185618,
      "grad_norm": 2.85673189163208,
      "learning_rate": 1.8484941610325753e-05,
      "loss": 1.4052,
      "step": 4930
    },
    {
      "epoch": 0.3789182544560541,
      "grad_norm": 3.134815216064453,
      "learning_rate": 1.8484326982175783e-05,
      "loss": 1.4306,
      "step": 4932
    },
    {
      "epoch": 0.3790719114935464,
      "grad_norm": 3.489353656768799,
      "learning_rate": 1.8483712354025816e-05,
      "loss": 1.3598,
      "step": 4934
    },
    {
      "epoch": 0.3792255685310387,
      "grad_norm": 3.4628639221191406,
      "learning_rate": 1.8483097725875846e-05,
      "loss": 1.3958,
      "step": 4936
    },
    {
      "epoch": 0.37937922556853104,
      "grad_norm": 3.2604832649230957,
      "learning_rate": 1.848248309772588e-05,
      "loss": 1.3876,
      "step": 4938
    },
    {
      "epoch": 0.37953288260602336,
      "grad_norm": 3.188798189163208,
      "learning_rate": 1.848186846957591e-05,
      "loss": 1.3992,
      "step": 4940
    },
    {
      "epoch": 0.3796865396435157,
      "grad_norm": 3.7080070972442627,
      "learning_rate": 1.8481253841425938e-05,
      "loss": 1.4537,
      "step": 4942
    },
    {
      "epoch": 0.379840196681008,
      "grad_norm": 2.9099485874176025,
      "learning_rate": 1.848063921327597e-05,
      "loss": 1.4259,
      "step": 4944
    },
    {
      "epoch": 0.3799938537185003,
      "grad_norm": 3.4322965145111084,
      "learning_rate": 1.8480024585126e-05,
      "loss": 1.4752,
      "step": 4946
    },
    {
      "epoch": 0.3801475107559926,
      "grad_norm": 3.3011248111724854,
      "learning_rate": 1.847940995697603e-05,
      "loss": 1.4632,
      "step": 4948
    },
    {
      "epoch": 0.38030116779348494,
      "grad_norm": 2.9199092388153076,
      "learning_rate": 1.8478795328826064e-05,
      "loss": 1.3915,
      "step": 4950
    },
    {
      "epoch": 0.38045482483097726,
      "grad_norm": 2.939312219619751,
      "learning_rate": 1.847818070067609e-05,
      "loss": 1.3709,
      "step": 4952
    },
    {
      "epoch": 0.3806084818684696,
      "grad_norm": 3.100517749786377,
      "learning_rate": 1.8477566072526123e-05,
      "loss": 1.3626,
      "step": 4954
    },
    {
      "epoch": 0.3807621389059619,
      "grad_norm": 3.323359489440918,
      "learning_rate": 1.8476951444376153e-05,
      "loss": 1.388,
      "step": 4956
    },
    {
      "epoch": 0.3809157959434542,
      "grad_norm": 3.2481391429901123,
      "learning_rate": 1.8476336816226183e-05,
      "loss": 1.3469,
      "step": 4958
    },
    {
      "epoch": 0.3810694529809465,
      "grad_norm": 2.9566774368286133,
      "learning_rate": 1.8475722188076216e-05,
      "loss": 1.4155,
      "step": 4960
    },
    {
      "epoch": 0.38122311001843884,
      "grad_norm": 3.426717758178711,
      "learning_rate": 1.8475107559926245e-05,
      "loss": 1.509,
      "step": 4962
    },
    {
      "epoch": 0.38137676705593115,
      "grad_norm": 3.242187738418579,
      "learning_rate": 1.847449293177628e-05,
      "loss": 1.5115,
      "step": 4964
    },
    {
      "epoch": 0.38153042409342347,
      "grad_norm": 3.4956090450286865,
      "learning_rate": 1.8473878303626308e-05,
      "loss": 1.3576,
      "step": 4966
    },
    {
      "epoch": 0.3816840811309158,
      "grad_norm": 12.086480140686035,
      "learning_rate": 1.8473263675476338e-05,
      "loss": 1.3618,
      "step": 4968
    },
    {
      "epoch": 0.3818377381684081,
      "grad_norm": 3.5222816467285156,
      "learning_rate": 1.847264904732637e-05,
      "loss": 1.4648,
      "step": 4970
    },
    {
      "epoch": 0.3819913952059004,
      "grad_norm": 2.8597071170806885,
      "learning_rate": 1.84720344191764e-05,
      "loss": 1.4428,
      "step": 4972
    },
    {
      "epoch": 0.38214505224339274,
      "grad_norm": 2.9223949909210205,
      "learning_rate": 1.847141979102643e-05,
      "loss": 1.3655,
      "step": 4974
    },
    {
      "epoch": 0.38229870928088505,
      "grad_norm": 3.160203695297241,
      "learning_rate": 1.8470805162876463e-05,
      "loss": 1.5182,
      "step": 4976
    },
    {
      "epoch": 0.38245236631837737,
      "grad_norm": 3.2447376251220703,
      "learning_rate": 1.847019053472649e-05,
      "loss": 1.5215,
      "step": 4978
    },
    {
      "epoch": 0.3826060233558697,
      "grad_norm": 3.0082247257232666,
      "learning_rate": 1.8469575906576523e-05,
      "loss": 1.3198,
      "step": 4980
    },
    {
      "epoch": 0.382759680393362,
      "grad_norm": 3.1510775089263916,
      "learning_rate": 1.8468961278426552e-05,
      "loss": 1.3984,
      "step": 4982
    },
    {
      "epoch": 0.3829133374308543,
      "grad_norm": 3.150709629058838,
      "learning_rate": 1.8468346650276585e-05,
      "loss": 1.3621,
      "step": 4984
    },
    {
      "epoch": 0.38306699446834663,
      "grad_norm": 2.9365999698638916,
      "learning_rate": 1.8467732022126615e-05,
      "loss": 1.4187,
      "step": 4986
    },
    {
      "epoch": 0.38322065150583895,
      "grad_norm": 3.299741744995117,
      "learning_rate": 1.8467117393976645e-05,
      "loss": 1.4882,
      "step": 4988
    },
    {
      "epoch": 0.38337430854333127,
      "grad_norm": 3.2206766605377197,
      "learning_rate": 1.8466502765826678e-05,
      "loss": 1.4838,
      "step": 4990
    },
    {
      "epoch": 0.3835279655808236,
      "grad_norm": 3.3576810359954834,
      "learning_rate": 1.8465888137676707e-05,
      "loss": 1.3716,
      "step": 4992
    },
    {
      "epoch": 0.3836816226183159,
      "grad_norm": 3.3749849796295166,
      "learning_rate": 1.8465273509526737e-05,
      "loss": 1.5736,
      "step": 4994
    },
    {
      "epoch": 0.3838352796558082,
      "grad_norm": 3.3339991569519043,
      "learning_rate": 1.846465888137677e-05,
      "loss": 1.4127,
      "step": 4996
    },
    {
      "epoch": 0.38398893669330053,
      "grad_norm": 3.272212266921997,
      "learning_rate": 1.84640442532268e-05,
      "loss": 1.3956,
      "step": 4998
    },
    {
      "epoch": 0.38414259373079285,
      "grad_norm": 3.243849515914917,
      "learning_rate": 1.846342962507683e-05,
      "loss": 1.4803,
      "step": 5000
    },
    {
      "epoch": 0.38429625076828516,
      "grad_norm": 3.5325818061828613,
      "learning_rate": 1.8462814996926863e-05,
      "loss": 1.4568,
      "step": 5002
    },
    {
      "epoch": 0.3844499078057775,
      "grad_norm": 3.187891721725464,
      "learning_rate": 1.8462200368776892e-05,
      "loss": 1.4075,
      "step": 5004
    },
    {
      "epoch": 0.3846035648432698,
      "grad_norm": 3.035670757293701,
      "learning_rate": 1.8461585740626922e-05,
      "loss": 1.3194,
      "step": 5006
    },
    {
      "epoch": 0.3847572218807621,
      "grad_norm": 3.086202383041382,
      "learning_rate": 1.8460971112476952e-05,
      "loss": 1.5004,
      "step": 5008
    },
    {
      "epoch": 0.38491087891825443,
      "grad_norm": 3.264146327972412,
      "learning_rate": 1.8460356484326985e-05,
      "loss": 1.4502,
      "step": 5010
    },
    {
      "epoch": 0.3850645359557468,
      "grad_norm": 3.0813944339752197,
      "learning_rate": 1.8459741856177014e-05,
      "loss": 1.4018,
      "step": 5012
    },
    {
      "epoch": 0.3852181929932391,
      "grad_norm": 3.097153663635254,
      "learning_rate": 1.8459127228027044e-05,
      "loss": 1.3495,
      "step": 5014
    },
    {
      "epoch": 0.38537185003073143,
      "grad_norm": 3.213472366333008,
      "learning_rate": 1.8458512599877077e-05,
      "loss": 1.3695,
      "step": 5016
    },
    {
      "epoch": 0.38552550706822375,
      "grad_norm": 3.733430862426758,
      "learning_rate": 1.8457897971727107e-05,
      "loss": 1.5006,
      "step": 5018
    },
    {
      "epoch": 0.38567916410571607,
      "grad_norm": 3.3197145462036133,
      "learning_rate": 1.8457283343577137e-05,
      "loss": 1.4598,
      "step": 5020
    },
    {
      "epoch": 0.3858328211432084,
      "grad_norm": 2.798349618911743,
      "learning_rate": 1.845666871542717e-05,
      "loss": 1.2987,
      "step": 5022
    },
    {
      "epoch": 0.3859864781807007,
      "grad_norm": 3.056670665740967,
      "learning_rate": 1.84560540872772e-05,
      "loss": 1.5097,
      "step": 5024
    },
    {
      "epoch": 0.386140135218193,
      "grad_norm": 4.263172149658203,
      "learning_rate": 1.845543945912723e-05,
      "loss": 1.4077,
      "step": 5026
    },
    {
      "epoch": 0.38629379225568533,
      "grad_norm": 2.9781241416931152,
      "learning_rate": 1.845482483097726e-05,
      "loss": 1.4007,
      "step": 5028
    },
    {
      "epoch": 0.38644744929317765,
      "grad_norm": 3.4377338886260986,
      "learning_rate": 1.8454210202827292e-05,
      "loss": 1.2981,
      "step": 5030
    },
    {
      "epoch": 0.38660110633066996,
      "grad_norm": 2.9486420154571533,
      "learning_rate": 1.845359557467732e-05,
      "loss": 1.3136,
      "step": 5032
    },
    {
      "epoch": 0.3867547633681623,
      "grad_norm": 3.4080779552459717,
      "learning_rate": 1.845298094652735e-05,
      "loss": 1.4167,
      "step": 5034
    },
    {
      "epoch": 0.3869084204056546,
      "grad_norm": 4.366278648376465,
      "learning_rate": 1.8452366318377384e-05,
      "loss": 1.3073,
      "step": 5036
    },
    {
      "epoch": 0.3870620774431469,
      "grad_norm": 3.295130491256714,
      "learning_rate": 1.8451751690227414e-05,
      "loss": 1.391,
      "step": 5038
    },
    {
      "epoch": 0.38721573448063923,
      "grad_norm": 3.4234602451324463,
      "learning_rate": 1.8451137062077444e-05,
      "loss": 1.4918,
      "step": 5040
    },
    {
      "epoch": 0.38736939151813155,
      "grad_norm": 3.76176118850708,
      "learning_rate": 1.8450522433927477e-05,
      "loss": 1.3417,
      "step": 5042
    },
    {
      "epoch": 0.38752304855562386,
      "grad_norm": 2.9911139011383057,
      "learning_rate": 1.8449907805777506e-05,
      "loss": 1.3087,
      "step": 5044
    },
    {
      "epoch": 0.3876767055931162,
      "grad_norm": 3.2155096530914307,
      "learning_rate": 1.8449293177627536e-05,
      "loss": 1.2759,
      "step": 5046
    },
    {
      "epoch": 0.3878303626306085,
      "grad_norm": 3.25315523147583,
      "learning_rate": 1.844867854947757e-05,
      "loss": 1.4139,
      "step": 5048
    },
    {
      "epoch": 0.3879840196681008,
      "grad_norm": 3.077777862548828,
      "learning_rate": 1.84480639213276e-05,
      "loss": 1.4828,
      "step": 5050
    },
    {
      "epoch": 0.3881376767055931,
      "grad_norm": 3.0072484016418457,
      "learning_rate": 1.844744929317763e-05,
      "loss": 1.2987,
      "step": 5052
    },
    {
      "epoch": 0.38829133374308544,
      "grad_norm": 3.023391008377075,
      "learning_rate": 1.8446834665027658e-05,
      "loss": 1.4497,
      "step": 5054
    },
    {
      "epoch": 0.38844499078057776,
      "grad_norm": 2.8816518783569336,
      "learning_rate": 1.844622003687769e-05,
      "loss": 1.316,
      "step": 5056
    },
    {
      "epoch": 0.3885986478180701,
      "grad_norm": 3.7871954441070557,
      "learning_rate": 1.844560540872772e-05,
      "loss": 1.3195,
      "step": 5058
    },
    {
      "epoch": 0.3887523048555624,
      "grad_norm": 3.246473550796509,
      "learning_rate": 1.844499078057775e-05,
      "loss": 1.4516,
      "step": 5060
    },
    {
      "epoch": 0.3889059618930547,
      "grad_norm": 3.081164836883545,
      "learning_rate": 1.8444376152427784e-05,
      "loss": 1.4083,
      "step": 5062
    },
    {
      "epoch": 0.389059618930547,
      "grad_norm": 3.482865810394287,
      "learning_rate": 1.8443761524277813e-05,
      "loss": 1.4287,
      "step": 5064
    },
    {
      "epoch": 0.38921327596803934,
      "grad_norm": 3.7230827808380127,
      "learning_rate": 1.8443146896127843e-05,
      "loss": 1.4355,
      "step": 5066
    },
    {
      "epoch": 0.38936693300553166,
      "grad_norm": 2.9292538166046143,
      "learning_rate": 1.8442532267977876e-05,
      "loss": 1.3523,
      "step": 5068
    },
    {
      "epoch": 0.389520590043024,
      "grad_norm": 3.3166749477386475,
      "learning_rate": 1.8441917639827906e-05,
      "loss": 1.3417,
      "step": 5070
    },
    {
      "epoch": 0.3896742470805163,
      "grad_norm": 3.5913453102111816,
      "learning_rate": 1.8441303011677935e-05,
      "loss": 1.4755,
      "step": 5072
    },
    {
      "epoch": 0.3898279041180086,
      "grad_norm": 3.5193488597869873,
      "learning_rate": 1.844068838352797e-05,
      "loss": 1.5068,
      "step": 5074
    },
    {
      "epoch": 0.3899815611555009,
      "grad_norm": 3.6477646827697754,
      "learning_rate": 1.8440073755377998e-05,
      "loss": 1.4358,
      "step": 5076
    },
    {
      "epoch": 0.39013521819299324,
      "grad_norm": 3.212559700012207,
      "learning_rate": 1.8439459127228028e-05,
      "loss": 1.3363,
      "step": 5078
    },
    {
      "epoch": 0.39028887523048555,
      "grad_norm": 3.282552480697632,
      "learning_rate": 1.8438844499078058e-05,
      "loss": 1.4447,
      "step": 5080
    },
    {
      "epoch": 0.39044253226797787,
      "grad_norm": 3.2416465282440186,
      "learning_rate": 1.843822987092809e-05,
      "loss": 1.4563,
      "step": 5082
    },
    {
      "epoch": 0.3905961893054702,
      "grad_norm": 3.1679558753967285,
      "learning_rate": 1.843761524277812e-05,
      "loss": 1.4569,
      "step": 5084
    },
    {
      "epoch": 0.3907498463429625,
      "grad_norm": 2.9866411685943604,
      "learning_rate": 1.843700061462815e-05,
      "loss": 1.3765,
      "step": 5086
    },
    {
      "epoch": 0.3909035033804548,
      "grad_norm": 3.1638734340667725,
      "learning_rate": 1.8436385986478183e-05,
      "loss": 1.5417,
      "step": 5088
    },
    {
      "epoch": 0.39105716041794714,
      "grad_norm": 3.5159435272216797,
      "learning_rate": 1.8435771358328213e-05,
      "loss": 1.4869,
      "step": 5090
    },
    {
      "epoch": 0.39121081745543945,
      "grad_norm": 3.1329610347747803,
      "learning_rate": 1.8435156730178242e-05,
      "loss": 1.4025,
      "step": 5092
    },
    {
      "epoch": 0.39136447449293177,
      "grad_norm": 3.4398860931396484,
      "learning_rate": 1.8434542102028276e-05,
      "loss": 1.5109,
      "step": 5094
    },
    {
      "epoch": 0.3915181315304241,
      "grad_norm": 3.5402066707611084,
      "learning_rate": 1.8433927473878305e-05,
      "loss": 1.4089,
      "step": 5096
    },
    {
      "epoch": 0.3916717885679164,
      "grad_norm": 3.4249396324157715,
      "learning_rate": 1.8433312845728335e-05,
      "loss": 1.4015,
      "step": 5098
    },
    {
      "epoch": 0.3918254456054087,
      "grad_norm": 2.852504014968872,
      "learning_rate": 1.8432698217578368e-05,
      "loss": 1.4679,
      "step": 5100
    },
    {
      "epoch": 0.39197910264290103,
      "grad_norm": 3.3587238788604736,
      "learning_rate": 1.8432083589428398e-05,
      "loss": 1.5021,
      "step": 5102
    },
    {
      "epoch": 0.39213275968039335,
      "grad_norm": 3.091439723968506,
      "learning_rate": 1.8431468961278427e-05,
      "loss": 1.531,
      "step": 5104
    },
    {
      "epoch": 0.39228641671788567,
      "grad_norm": 3.6447200775146484,
      "learning_rate": 1.8430854333128457e-05,
      "loss": 1.2355,
      "step": 5106
    },
    {
      "epoch": 0.392440073755378,
      "grad_norm": 3.580310821533203,
      "learning_rate": 1.843023970497849e-05,
      "loss": 1.5096,
      "step": 5108
    },
    {
      "epoch": 0.3925937307928703,
      "grad_norm": 5.039089679718018,
      "learning_rate": 1.842962507682852e-05,
      "loss": 1.3975,
      "step": 5110
    },
    {
      "epoch": 0.3927473878303626,
      "grad_norm": 3.476210594177246,
      "learning_rate": 1.842901044867855e-05,
      "loss": 1.4081,
      "step": 5112
    },
    {
      "epoch": 0.39290104486785493,
      "grad_norm": 3.437044382095337,
      "learning_rate": 1.8428395820528583e-05,
      "loss": 1.4921,
      "step": 5114
    },
    {
      "epoch": 0.39305470190534725,
      "grad_norm": 3.2391202449798584,
      "learning_rate": 1.8427781192378612e-05,
      "loss": 1.3004,
      "step": 5116
    },
    {
      "epoch": 0.39320835894283956,
      "grad_norm": 3.0900301933288574,
      "learning_rate": 1.8427166564228642e-05,
      "loss": 1.3734,
      "step": 5118
    },
    {
      "epoch": 0.3933620159803319,
      "grad_norm": 3.2002086639404297,
      "learning_rate": 1.8426551936078675e-05,
      "loss": 1.3398,
      "step": 5120
    },
    {
      "epoch": 0.3935156730178242,
      "grad_norm": 3.1345863342285156,
      "learning_rate": 1.8425937307928705e-05,
      "loss": 1.4848,
      "step": 5122
    },
    {
      "epoch": 0.3936693300553165,
      "grad_norm": 3.710662603378296,
      "learning_rate": 1.8425322679778734e-05,
      "loss": 1.3935,
      "step": 5124
    },
    {
      "epoch": 0.39382298709280883,
      "grad_norm": 3.5913710594177246,
      "learning_rate": 1.8424708051628767e-05,
      "loss": 1.4266,
      "step": 5126
    },
    {
      "epoch": 0.39397664413030115,
      "grad_norm": 3.0973658561706543,
      "learning_rate": 1.8424093423478797e-05,
      "loss": 1.516,
      "step": 5128
    },
    {
      "epoch": 0.39413030116779346,
      "grad_norm": 3.180147171020508,
      "learning_rate": 1.842347879532883e-05,
      "loss": 1.3113,
      "step": 5130
    },
    {
      "epoch": 0.3942839582052858,
      "grad_norm": 3.239496946334839,
      "learning_rate": 1.8422864167178856e-05,
      "loss": 1.4985,
      "step": 5132
    },
    {
      "epoch": 0.3944376152427781,
      "grad_norm": 3.4438583850860596,
      "learning_rate": 1.842224953902889e-05,
      "loss": 1.4734,
      "step": 5134
    },
    {
      "epoch": 0.3945912722802704,
      "grad_norm": 3.0532734394073486,
      "learning_rate": 1.842163491087892e-05,
      "loss": 1.3405,
      "step": 5136
    },
    {
      "epoch": 0.3947449293177627,
      "grad_norm": 3.0784571170806885,
      "learning_rate": 1.842102028272895e-05,
      "loss": 1.3506,
      "step": 5138
    },
    {
      "epoch": 0.39489858635525504,
      "grad_norm": 3.235511064529419,
      "learning_rate": 1.8420405654578982e-05,
      "loss": 1.3963,
      "step": 5140
    },
    {
      "epoch": 0.3950522433927474,
      "grad_norm": 2.876587390899658,
      "learning_rate": 1.841979102642901e-05,
      "loss": 1.4846,
      "step": 5142
    },
    {
      "epoch": 0.39520590043023973,
      "grad_norm": 3.13354754447937,
      "learning_rate": 1.841917639827904e-05,
      "loss": 1.3671,
      "step": 5144
    },
    {
      "epoch": 0.39535955746773205,
      "grad_norm": 3.12070894241333,
      "learning_rate": 1.8418561770129074e-05,
      "loss": 1.4328,
      "step": 5146
    },
    {
      "epoch": 0.39551321450522436,
      "grad_norm": 3.660349130630493,
      "learning_rate": 1.8417947141979104e-05,
      "loss": 1.4445,
      "step": 5148
    },
    {
      "epoch": 0.3956668715427167,
      "grad_norm": 3.272484064102173,
      "learning_rate": 1.8417332513829137e-05,
      "loss": 1.3548,
      "step": 5150
    },
    {
      "epoch": 0.395820528580209,
      "grad_norm": 2.723353624343872,
      "learning_rate": 1.8416717885679163e-05,
      "loss": 1.3979,
      "step": 5152
    },
    {
      "epoch": 0.3959741856177013,
      "grad_norm": 3.0933709144592285,
      "learning_rate": 1.8416103257529197e-05,
      "loss": 1.4523,
      "step": 5154
    },
    {
      "epoch": 0.39612784265519363,
      "grad_norm": 3.3411343097686768,
      "learning_rate": 1.8415488629379226e-05,
      "loss": 1.5236,
      "step": 5156
    },
    {
      "epoch": 0.39628149969268595,
      "grad_norm": 3.618424415588379,
      "learning_rate": 1.8414874001229256e-05,
      "loss": 1.3047,
      "step": 5158
    },
    {
      "epoch": 0.39643515673017826,
      "grad_norm": 3.5820114612579346,
      "learning_rate": 1.841425937307929e-05,
      "loss": 1.4815,
      "step": 5160
    },
    {
      "epoch": 0.3965888137676706,
      "grad_norm": 3.064063549041748,
      "learning_rate": 1.841364474492932e-05,
      "loss": 1.4813,
      "step": 5162
    },
    {
      "epoch": 0.3967424708051629,
      "grad_norm": 3.082622528076172,
      "learning_rate": 1.841303011677935e-05,
      "loss": 1.3751,
      "step": 5164
    },
    {
      "epoch": 0.3968961278426552,
      "grad_norm": 3.184774398803711,
      "learning_rate": 1.841241548862938e-05,
      "loss": 1.3173,
      "step": 5166
    },
    {
      "epoch": 0.3970497848801475,
      "grad_norm": 3.2373135089874268,
      "learning_rate": 1.841180086047941e-05,
      "loss": 1.5307,
      "step": 5168
    },
    {
      "epoch": 0.39720344191763984,
      "grad_norm": 2.656855583190918,
      "learning_rate": 1.8411186232329444e-05,
      "loss": 1.2303,
      "step": 5170
    },
    {
      "epoch": 0.39735709895513216,
      "grad_norm": 3.1388304233551025,
      "learning_rate": 1.8410571604179474e-05,
      "loss": 1.4094,
      "step": 5172
    },
    {
      "epoch": 0.3975107559926245,
      "grad_norm": 3.365753173828125,
      "learning_rate": 1.8409956976029504e-05,
      "loss": 1.3659,
      "step": 5174
    },
    {
      "epoch": 0.3976644130301168,
      "grad_norm": 3.192204236984253,
      "learning_rate": 1.8409342347879537e-05,
      "loss": 1.3244,
      "step": 5176
    },
    {
      "epoch": 0.3978180700676091,
      "grad_norm": 3.1940553188323975,
      "learning_rate": 1.8408727719729563e-05,
      "loss": 1.4838,
      "step": 5178
    },
    {
      "epoch": 0.3979717271051014,
      "grad_norm": 3.3534750938415527,
      "learning_rate": 1.8408113091579596e-05,
      "loss": 1.3472,
      "step": 5180
    },
    {
      "epoch": 0.39812538414259374,
      "grad_norm": 3.4053544998168945,
      "learning_rate": 1.8407498463429626e-05,
      "loss": 1.4988,
      "step": 5182
    },
    {
      "epoch": 0.39827904118008606,
      "grad_norm": 3.329324960708618,
      "learning_rate": 1.8406883835279655e-05,
      "loss": 1.5343,
      "step": 5184
    },
    {
      "epoch": 0.3984326982175784,
      "grad_norm": 3.0776827335357666,
      "learning_rate": 1.840626920712969e-05,
      "loss": 1.3744,
      "step": 5186
    },
    {
      "epoch": 0.3985863552550707,
      "grad_norm": 3.1414880752563477,
      "learning_rate": 1.8405654578979718e-05,
      "loss": 1.3941,
      "step": 5188
    },
    {
      "epoch": 0.398740012292563,
      "grad_norm": 3.3760602474212646,
      "learning_rate": 1.840503995082975e-05,
      "loss": 1.4755,
      "step": 5190
    },
    {
      "epoch": 0.3988936693300553,
      "grad_norm": 3.2794857025146484,
      "learning_rate": 1.840442532267978e-05,
      "loss": 1.3525,
      "step": 5192
    },
    {
      "epoch": 0.39904732636754764,
      "grad_norm": 2.952638864517212,
      "learning_rate": 1.840381069452981e-05,
      "loss": 1.3713,
      "step": 5194
    },
    {
      "epoch": 0.39920098340503996,
      "grad_norm": 3.3649606704711914,
      "learning_rate": 1.8403196066379844e-05,
      "loss": 1.4609,
      "step": 5196
    },
    {
      "epoch": 0.39935464044253227,
      "grad_norm": 3.112147569656372,
      "learning_rate": 1.8402581438229873e-05,
      "loss": 1.4294,
      "step": 5198
    },
    {
      "epoch": 0.3995082974800246,
      "grad_norm": 3.05891489982605,
      "learning_rate": 1.8401966810079903e-05,
      "loss": 1.4269,
      "step": 5200
    },
    {
      "epoch": 0.3996619545175169,
      "grad_norm": 3.5872931480407715,
      "learning_rate": 1.8401352181929936e-05,
      "loss": 1.4834,
      "step": 5202
    },
    {
      "epoch": 0.3998156115550092,
      "grad_norm": 2.9866600036621094,
      "learning_rate": 1.8400737553779962e-05,
      "loss": 1.2641,
      "step": 5204
    },
    {
      "epoch": 0.39996926859250154,
      "grad_norm": 3.228916645050049,
      "learning_rate": 1.8400122925629995e-05,
      "loss": 1.3528,
      "step": 5206
    },
    {
      "epoch": 0.40012292562999385,
      "grad_norm": 3.571171998977661,
      "learning_rate": 1.8399508297480025e-05,
      "loss": 1.3518,
      "step": 5208
    },
    {
      "epoch": 0.40027658266748617,
      "grad_norm": 3.4012258052825928,
      "learning_rate": 1.8398893669330055e-05,
      "loss": 1.3364,
      "step": 5210
    },
    {
      "epoch": 0.4004302397049785,
      "grad_norm": 3.279327630996704,
      "learning_rate": 1.8398279041180088e-05,
      "loss": 1.3591,
      "step": 5212
    },
    {
      "epoch": 0.4005838967424708,
      "grad_norm": 3.180152416229248,
      "learning_rate": 1.8397664413030118e-05,
      "loss": 1.4357,
      "step": 5214
    },
    {
      "epoch": 0.4007375537799631,
      "grad_norm": 3.127516269683838,
      "learning_rate": 1.839704978488015e-05,
      "loss": 1.3258,
      "step": 5216
    },
    {
      "epoch": 0.40089121081745543,
      "grad_norm": 3.2782299518585205,
      "learning_rate": 1.839643515673018e-05,
      "loss": 1.3646,
      "step": 5218
    },
    {
      "epoch": 0.40104486785494775,
      "grad_norm": 3.528404712677002,
      "learning_rate": 1.839582052858021e-05,
      "loss": 1.3139,
      "step": 5220
    },
    {
      "epoch": 0.40119852489244007,
      "grad_norm": 3.2569220066070557,
      "learning_rate": 1.8395205900430243e-05,
      "loss": 1.4107,
      "step": 5222
    },
    {
      "epoch": 0.4013521819299324,
      "grad_norm": 3.212218761444092,
      "learning_rate": 1.8394591272280273e-05,
      "loss": 1.4875,
      "step": 5224
    },
    {
      "epoch": 0.4015058389674247,
      "grad_norm": 3.204953670501709,
      "learning_rate": 1.8393976644130302e-05,
      "loss": 1.4317,
      "step": 5226
    },
    {
      "epoch": 0.401659496004917,
      "grad_norm": 3.2656290531158447,
      "learning_rate": 1.8393362015980336e-05,
      "loss": 1.4636,
      "step": 5228
    },
    {
      "epoch": 0.40181315304240933,
      "grad_norm": 2.8223979473114014,
      "learning_rate": 1.8392747387830362e-05,
      "loss": 1.4513,
      "step": 5230
    },
    {
      "epoch": 0.40196681007990165,
      "grad_norm": 2.845686912536621,
      "learning_rate": 1.8392132759680395e-05,
      "loss": 1.3737,
      "step": 5232
    },
    {
      "epoch": 0.40212046711739396,
      "grad_norm": 2.8757717609405518,
      "learning_rate": 1.8391518131530425e-05,
      "loss": 1.3632,
      "step": 5234
    },
    {
      "epoch": 0.4022741241548863,
      "grad_norm": 3.4212610721588135,
      "learning_rate": 1.8390903503380458e-05,
      "loss": 1.357,
      "step": 5236
    },
    {
      "epoch": 0.4024277811923786,
      "grad_norm": 3.0520594120025635,
      "learning_rate": 1.8390288875230487e-05,
      "loss": 1.3139,
      "step": 5238
    },
    {
      "epoch": 0.4025814382298709,
      "grad_norm": 3.0812041759490967,
      "learning_rate": 1.8389674247080517e-05,
      "loss": 1.3224,
      "step": 5240
    },
    {
      "epoch": 0.40273509526736323,
      "grad_norm": 3.380028009414673,
      "learning_rate": 1.838905961893055e-05,
      "loss": 1.3579,
      "step": 5242
    },
    {
      "epoch": 0.40288875230485555,
      "grad_norm": 3.7895190715789795,
      "learning_rate": 1.838844499078058e-05,
      "loss": 1.3326,
      "step": 5244
    },
    {
      "epoch": 0.40304240934234786,
      "grad_norm": 3.471078395843506,
      "learning_rate": 1.838783036263061e-05,
      "loss": 1.2679,
      "step": 5246
    },
    {
      "epoch": 0.4031960663798402,
      "grad_norm": 3.2340338230133057,
      "learning_rate": 1.8387215734480643e-05,
      "loss": 1.4917,
      "step": 5248
    },
    {
      "epoch": 0.4033497234173325,
      "grad_norm": 3.610034942626953,
      "learning_rate": 1.838660110633067e-05,
      "loss": 1.549,
      "step": 5250
    },
    {
      "epoch": 0.4035033804548248,
      "grad_norm": 3.2047059535980225,
      "learning_rate": 1.8385986478180702e-05,
      "loss": 1.538,
      "step": 5252
    },
    {
      "epoch": 0.40365703749231713,
      "grad_norm": 3.190843105316162,
      "learning_rate": 1.838537185003073e-05,
      "loss": 1.5433,
      "step": 5254
    },
    {
      "epoch": 0.40381069452980944,
      "grad_norm": 3.0473310947418213,
      "learning_rate": 1.8384757221880765e-05,
      "loss": 1.4303,
      "step": 5256
    },
    {
      "epoch": 0.40396435156730176,
      "grad_norm": 3.305148124694824,
      "learning_rate": 1.8384142593730794e-05,
      "loss": 1.4647,
      "step": 5258
    },
    {
      "epoch": 0.4041180086047941,
      "grad_norm": 3.2546324729919434,
      "learning_rate": 1.8383527965580824e-05,
      "loss": 1.403,
      "step": 5260
    },
    {
      "epoch": 0.4042716656422864,
      "grad_norm": 3.26485276222229,
      "learning_rate": 1.8382913337430857e-05,
      "loss": 1.3281,
      "step": 5262
    },
    {
      "epoch": 0.4044253226797787,
      "grad_norm": 3.755387306213379,
      "learning_rate": 1.8382298709280887e-05,
      "loss": 1.3544,
      "step": 5264
    },
    {
      "epoch": 0.404578979717271,
      "grad_norm": 3.23287034034729,
      "learning_rate": 1.8381684081130916e-05,
      "loss": 1.4258,
      "step": 5266
    },
    {
      "epoch": 0.40473263675476334,
      "grad_norm": 3.2097320556640625,
      "learning_rate": 1.838106945298095e-05,
      "loss": 1.3425,
      "step": 5268
    },
    {
      "epoch": 0.40488629379225566,
      "grad_norm": 3.3704915046691895,
      "learning_rate": 1.838045482483098e-05,
      "loss": 1.5884,
      "step": 5270
    },
    {
      "epoch": 0.40503995082974803,
      "grad_norm": 2.9925537109375,
      "learning_rate": 1.837984019668101e-05,
      "loss": 1.3085,
      "step": 5272
    },
    {
      "epoch": 0.40519360786724035,
      "grad_norm": 2.810943841934204,
      "learning_rate": 1.8379225568531042e-05,
      "loss": 1.3259,
      "step": 5274
    },
    {
      "epoch": 0.40534726490473266,
      "grad_norm": 3.005441904067993,
      "learning_rate": 1.837861094038107e-05,
      "loss": 1.5093,
      "step": 5276
    },
    {
      "epoch": 0.405500921942225,
      "grad_norm": 3.141219139099121,
      "learning_rate": 1.83779963122311e-05,
      "loss": 1.3509,
      "step": 5278
    },
    {
      "epoch": 0.4056545789797173,
      "grad_norm": 3.446096658706665,
      "learning_rate": 1.837738168408113e-05,
      "loss": 1.4114,
      "step": 5280
    },
    {
      "epoch": 0.4058082360172096,
      "grad_norm": 3.5960264205932617,
      "learning_rate": 1.8376767055931164e-05,
      "loss": 1.5067,
      "step": 5282
    },
    {
      "epoch": 0.40596189305470193,
      "grad_norm": 3.441776990890503,
      "learning_rate": 1.8376152427781194e-05,
      "loss": 1.3281,
      "step": 5284
    },
    {
      "epoch": 0.40611555009219424,
      "grad_norm": 2.811274290084839,
      "learning_rate": 1.8375537799631223e-05,
      "loss": 1.4239,
      "step": 5286
    },
    {
      "epoch": 0.40626920712968656,
      "grad_norm": 2.918977737426758,
      "learning_rate": 1.8374923171481257e-05,
      "loss": 1.3116,
      "step": 5288
    },
    {
      "epoch": 0.4064228641671789,
      "grad_norm": 3.3775150775909424,
      "learning_rate": 1.8374308543331286e-05,
      "loss": 1.476,
      "step": 5290
    },
    {
      "epoch": 0.4065765212046712,
      "grad_norm": 3.5085885524749756,
      "learning_rate": 1.8373693915181316e-05,
      "loss": 1.4227,
      "step": 5292
    },
    {
      "epoch": 0.4067301782421635,
      "grad_norm": 3.496614933013916,
      "learning_rate": 1.837307928703135e-05,
      "loss": 1.3704,
      "step": 5294
    },
    {
      "epoch": 0.4068838352796558,
      "grad_norm": 3.1365761756896973,
      "learning_rate": 1.837246465888138e-05,
      "loss": 1.4542,
      "step": 5296
    },
    {
      "epoch": 0.40703749231714814,
      "grad_norm": 3.266559600830078,
      "learning_rate": 1.837185003073141e-05,
      "loss": 1.2663,
      "step": 5298
    },
    {
      "epoch": 0.40719114935464046,
      "grad_norm": 3.0944910049438477,
      "learning_rate": 1.837123540258144e-05,
      "loss": 1.3376,
      "step": 5300
    },
    {
      "epoch": 0.4073448063921328,
      "grad_norm": 3.3353304862976074,
      "learning_rate": 1.837062077443147e-05,
      "loss": 1.4512,
      "step": 5302
    },
    {
      "epoch": 0.4074984634296251,
      "grad_norm": 3.0446059703826904,
      "learning_rate": 1.83700061462815e-05,
      "loss": 1.3703,
      "step": 5304
    },
    {
      "epoch": 0.4076521204671174,
      "grad_norm": 2.9475607872009277,
      "learning_rate": 1.836939151813153e-05,
      "loss": 1.4063,
      "step": 5306
    },
    {
      "epoch": 0.4078057775046097,
      "grad_norm": 3.077463388442993,
      "learning_rate": 1.8368776889981564e-05,
      "loss": 1.378,
      "step": 5308
    },
    {
      "epoch": 0.40795943454210204,
      "grad_norm": 3.3104872703552246,
      "learning_rate": 1.8368162261831593e-05,
      "loss": 1.3273,
      "step": 5310
    },
    {
      "epoch": 0.40811309157959436,
      "grad_norm": 3.311328649520874,
      "learning_rate": 1.8367547633681623e-05,
      "loss": 1.3674,
      "step": 5312
    },
    {
      "epoch": 0.40826674861708667,
      "grad_norm": 3.025520086288452,
      "learning_rate": 1.8366933005531656e-05,
      "loss": 1.3716,
      "step": 5314
    },
    {
      "epoch": 0.408420405654579,
      "grad_norm": 3.5785486698150635,
      "learning_rate": 1.8366318377381686e-05,
      "loss": 1.4323,
      "step": 5316
    },
    {
      "epoch": 0.4085740626920713,
      "grad_norm": 3.210676670074463,
      "learning_rate": 1.8365703749231715e-05,
      "loss": 1.5345,
      "step": 5318
    },
    {
      "epoch": 0.4087277197295636,
      "grad_norm": 3.0286824703216553,
      "learning_rate": 1.836508912108175e-05,
      "loss": 1.423,
      "step": 5320
    },
    {
      "epoch": 0.40888137676705594,
      "grad_norm": 3.5791726112365723,
      "learning_rate": 1.8364474492931778e-05,
      "loss": 1.4609,
      "step": 5322
    },
    {
      "epoch": 0.40903503380454825,
      "grad_norm": 2.9176530838012695,
      "learning_rate": 1.8363859864781808e-05,
      "loss": 1.2732,
      "step": 5324
    },
    {
      "epoch": 0.40918869084204057,
      "grad_norm": 3.217237949371338,
      "learning_rate": 1.836324523663184e-05,
      "loss": 1.4485,
      "step": 5326
    },
    {
      "epoch": 0.4093423478795329,
      "grad_norm": 3.778691530227661,
      "learning_rate": 1.836263060848187e-05,
      "loss": 1.3841,
      "step": 5328
    },
    {
      "epoch": 0.4094960049170252,
      "grad_norm": 2.9347708225250244,
      "learning_rate": 1.83620159803319e-05,
      "loss": 1.3844,
      "step": 5330
    },
    {
      "epoch": 0.4096496619545175,
      "grad_norm": 3.1779749393463135,
      "learning_rate": 1.836140135218193e-05,
      "loss": 1.4284,
      "step": 5332
    },
    {
      "epoch": 0.40980331899200984,
      "grad_norm": 3.506448745727539,
      "learning_rate": 1.8360786724031963e-05,
      "loss": 1.4299,
      "step": 5334
    },
    {
      "epoch": 0.40995697602950215,
      "grad_norm": 2.704458713531494,
      "learning_rate": 1.8360172095881993e-05,
      "loss": 1.3064,
      "step": 5336
    },
    {
      "epoch": 0.41011063306699447,
      "grad_norm": 2.8528892993927,
      "learning_rate": 1.8359557467732022e-05,
      "loss": 1.2785,
      "step": 5338
    },
    {
      "epoch": 0.4102642901044868,
      "grad_norm": 3.1593194007873535,
      "learning_rate": 1.8358942839582055e-05,
      "loss": 1.4634,
      "step": 5340
    },
    {
      "epoch": 0.4104179471419791,
      "grad_norm": 4.361567974090576,
      "learning_rate": 1.8358328211432085e-05,
      "loss": 1.3619,
      "step": 5342
    },
    {
      "epoch": 0.4105716041794714,
      "grad_norm": 3.1610984802246094,
      "learning_rate": 1.8357713583282115e-05,
      "loss": 1.4343,
      "step": 5344
    },
    {
      "epoch": 0.41072526121696373,
      "grad_norm": 3.0249135494232178,
      "learning_rate": 1.8357098955132148e-05,
      "loss": 1.3613,
      "step": 5346
    },
    {
      "epoch": 0.41087891825445605,
      "grad_norm": 3.4013311862945557,
      "learning_rate": 1.8356484326982178e-05,
      "loss": 1.2754,
      "step": 5348
    },
    {
      "epoch": 0.41103257529194837,
      "grad_norm": 3.1088316440582275,
      "learning_rate": 1.8355869698832207e-05,
      "loss": 1.328,
      "step": 5350
    },
    {
      "epoch": 0.4111862323294407,
      "grad_norm": 2.942134141921997,
      "learning_rate": 1.8355255070682237e-05,
      "loss": 1.4541,
      "step": 5352
    },
    {
      "epoch": 0.411339889366933,
      "grad_norm": 3.704908847808838,
      "learning_rate": 1.835464044253227e-05,
      "loss": 1.477,
      "step": 5354
    },
    {
      "epoch": 0.4114935464044253,
      "grad_norm": 3.170875072479248,
      "learning_rate": 1.83540258143823e-05,
      "loss": 1.4399,
      "step": 5356
    },
    {
      "epoch": 0.41164720344191763,
      "grad_norm": 2.908482074737549,
      "learning_rate": 1.835341118623233e-05,
      "loss": 1.2963,
      "step": 5358
    },
    {
      "epoch": 0.41180086047940995,
      "grad_norm": 3.5838634967803955,
      "learning_rate": 1.8352796558082362e-05,
      "loss": 1.4518,
      "step": 5360
    },
    {
      "epoch": 0.41195451751690226,
      "grad_norm": 3.6148693561553955,
      "learning_rate": 1.8352181929932392e-05,
      "loss": 1.3671,
      "step": 5362
    },
    {
      "epoch": 0.4121081745543946,
      "grad_norm": 3.1762709617614746,
      "learning_rate": 1.8351567301782422e-05,
      "loss": 1.3727,
      "step": 5364
    },
    {
      "epoch": 0.4122618315918869,
      "grad_norm": 3.1139426231384277,
      "learning_rate": 1.8350952673632455e-05,
      "loss": 1.4239,
      "step": 5366
    },
    {
      "epoch": 0.4124154886293792,
      "grad_norm": 3.6204798221588135,
      "learning_rate": 1.8350338045482485e-05,
      "loss": 1.3177,
      "step": 5368
    },
    {
      "epoch": 0.41256914566687153,
      "grad_norm": 5.793487071990967,
      "learning_rate": 1.8349723417332514e-05,
      "loss": 1.4471,
      "step": 5370
    },
    {
      "epoch": 0.41272280270436384,
      "grad_norm": 3.2737765312194824,
      "learning_rate": 1.8349108789182547e-05,
      "loss": 1.4576,
      "step": 5372
    },
    {
      "epoch": 0.41287645974185616,
      "grad_norm": 3.0231080055236816,
      "learning_rate": 1.8348494161032577e-05,
      "loss": 1.3666,
      "step": 5374
    },
    {
      "epoch": 0.4130301167793485,
      "grad_norm": 2.8385233879089355,
      "learning_rate": 1.8347879532882607e-05,
      "loss": 1.3454,
      "step": 5376
    },
    {
      "epoch": 0.4131837738168408,
      "grad_norm": 3.1396775245666504,
      "learning_rate": 1.8347264904732636e-05,
      "loss": 1.4084,
      "step": 5378
    },
    {
      "epoch": 0.4133374308543331,
      "grad_norm": 3.143645763397217,
      "learning_rate": 1.834665027658267e-05,
      "loss": 1.3337,
      "step": 5380
    },
    {
      "epoch": 0.4134910878918254,
      "grad_norm": 2.823948383331299,
      "learning_rate": 1.83460356484327e-05,
      "loss": 1.3681,
      "step": 5382
    },
    {
      "epoch": 0.41364474492931774,
      "grad_norm": 3.2592108249664307,
      "learning_rate": 1.834542102028273e-05,
      "loss": 1.4719,
      "step": 5384
    },
    {
      "epoch": 0.41379840196681006,
      "grad_norm": 2.9613685607910156,
      "learning_rate": 1.8344806392132762e-05,
      "loss": 1.3567,
      "step": 5386
    },
    {
      "epoch": 0.4139520590043024,
      "grad_norm": 3.1878020763397217,
      "learning_rate": 1.834419176398279e-05,
      "loss": 1.4586,
      "step": 5388
    },
    {
      "epoch": 0.4141057160417947,
      "grad_norm": 3.22352933883667,
      "learning_rate": 1.834357713583282e-05,
      "loss": 1.3671,
      "step": 5390
    },
    {
      "epoch": 0.414259373079287,
      "grad_norm": 3.33892822265625,
      "learning_rate": 1.8342962507682854e-05,
      "loss": 1.3421,
      "step": 5392
    },
    {
      "epoch": 0.4144130301167793,
      "grad_norm": 3.0953426361083984,
      "learning_rate": 1.8342347879532884e-05,
      "loss": 1.4798,
      "step": 5394
    },
    {
      "epoch": 0.41456668715427164,
      "grad_norm": 3.559352397918701,
      "learning_rate": 1.8341733251382914e-05,
      "loss": 1.4564,
      "step": 5396
    },
    {
      "epoch": 0.41472034419176396,
      "grad_norm": 3.4220855236053467,
      "learning_rate": 1.8341118623232947e-05,
      "loss": 1.3482,
      "step": 5398
    },
    {
      "epoch": 0.4148740012292563,
      "grad_norm": 3.376953363418579,
      "learning_rate": 1.8340503995082976e-05,
      "loss": 1.3386,
      "step": 5400
    },
    {
      "epoch": 0.41502765826674864,
      "grad_norm": 3.256744861602783,
      "learning_rate": 1.833988936693301e-05,
      "loss": 1.4248,
      "step": 5402
    },
    {
      "epoch": 0.41518131530424096,
      "grad_norm": 3.20052170753479,
      "learning_rate": 1.8339274738783036e-05,
      "loss": 1.3001,
      "step": 5404
    },
    {
      "epoch": 0.4153349723417333,
      "grad_norm": 3.402547836303711,
      "learning_rate": 1.833866011063307e-05,
      "loss": 1.3549,
      "step": 5406
    },
    {
      "epoch": 0.4154886293792256,
      "grad_norm": 3.4516971111297607,
      "learning_rate": 1.83380454824831e-05,
      "loss": 1.4308,
      "step": 5408
    },
    {
      "epoch": 0.4156422864167179,
      "grad_norm": 3.404031753540039,
      "learning_rate": 1.8337430854333128e-05,
      "loss": 1.4279,
      "step": 5410
    },
    {
      "epoch": 0.4157959434542102,
      "grad_norm": 3.22731876373291,
      "learning_rate": 1.833681622618316e-05,
      "loss": 1.3986,
      "step": 5412
    },
    {
      "epoch": 0.41594960049170254,
      "grad_norm": 3.5908353328704834,
      "learning_rate": 1.833620159803319e-05,
      "loss": 1.4378,
      "step": 5414
    },
    {
      "epoch": 0.41610325752919486,
      "grad_norm": 2.8062829971313477,
      "learning_rate": 1.833558696988322e-05,
      "loss": 1.3553,
      "step": 5416
    },
    {
      "epoch": 0.4162569145666872,
      "grad_norm": 3.7070417404174805,
      "learning_rate": 1.8334972341733254e-05,
      "loss": 1.473,
      "step": 5418
    },
    {
      "epoch": 0.4164105716041795,
      "grad_norm": 3.7002766132354736,
      "learning_rate": 1.8334357713583283e-05,
      "loss": 1.5181,
      "step": 5420
    },
    {
      "epoch": 0.4165642286416718,
      "grad_norm": 3.2762959003448486,
      "learning_rate": 1.8333743085433316e-05,
      "loss": 1.2852,
      "step": 5422
    },
    {
      "epoch": 0.4167178856791641,
      "grad_norm": 3.2527976036071777,
      "learning_rate": 1.8333128457283346e-05,
      "loss": 1.5333,
      "step": 5424
    },
    {
      "epoch": 0.41687154271665644,
      "grad_norm": 3.0944836139678955,
      "learning_rate": 1.8332513829133376e-05,
      "loss": 1.3534,
      "step": 5426
    },
    {
      "epoch": 0.41702519975414876,
      "grad_norm": 2.9821205139160156,
      "learning_rate": 1.833189920098341e-05,
      "loss": 1.3932,
      "step": 5428
    },
    {
      "epoch": 0.4171788567916411,
      "grad_norm": 2.741713285446167,
      "learning_rate": 1.8331284572833435e-05,
      "loss": 1.3016,
      "step": 5430
    },
    {
      "epoch": 0.4173325138291334,
      "grad_norm": 2.732677459716797,
      "learning_rate": 1.8330669944683468e-05,
      "loss": 1.2872,
      "step": 5432
    },
    {
      "epoch": 0.4174861708666257,
      "grad_norm": 3.852531909942627,
      "learning_rate": 1.8330055316533498e-05,
      "loss": 1.4938,
      "step": 5434
    },
    {
      "epoch": 0.417639827904118,
      "grad_norm": 2.77213716506958,
      "learning_rate": 1.8329440688383528e-05,
      "loss": 1.2675,
      "step": 5436
    },
    {
      "epoch": 0.41779348494161034,
      "grad_norm": 3.0634377002716064,
      "learning_rate": 1.832882606023356e-05,
      "loss": 1.4362,
      "step": 5438
    },
    {
      "epoch": 0.41794714197910265,
      "grad_norm": 3.7038559913635254,
      "learning_rate": 1.832821143208359e-05,
      "loss": 1.5786,
      "step": 5440
    },
    {
      "epoch": 0.41810079901659497,
      "grad_norm": 2.922403335571289,
      "learning_rate": 1.8327596803933623e-05,
      "loss": 1.36,
      "step": 5442
    },
    {
      "epoch": 0.4182544560540873,
      "grad_norm": 3.7074828147888184,
      "learning_rate": 1.8326982175783653e-05,
      "loss": 1.4259,
      "step": 5444
    },
    {
      "epoch": 0.4184081130915796,
      "grad_norm": 3.401524305343628,
      "learning_rate": 1.8326367547633683e-05,
      "loss": 1.4031,
      "step": 5446
    },
    {
      "epoch": 0.4185617701290719,
      "grad_norm": 3.994004487991333,
      "learning_rate": 1.8325752919483716e-05,
      "loss": 1.3395,
      "step": 5448
    },
    {
      "epoch": 0.41871542716656424,
      "grad_norm": 3.46433162689209,
      "learning_rate": 1.8325138291333746e-05,
      "loss": 1.3823,
      "step": 5450
    },
    {
      "epoch": 0.41886908420405655,
      "grad_norm": 3.4628031253814697,
      "learning_rate": 1.8324523663183775e-05,
      "loss": 1.4543,
      "step": 5452
    },
    {
      "epoch": 0.41902274124154887,
      "grad_norm": 3.4159624576568604,
      "learning_rate": 1.832390903503381e-05,
      "loss": 1.3729,
      "step": 5454
    },
    {
      "epoch": 0.4191763982790412,
      "grad_norm": 3.102025270462036,
      "learning_rate": 1.8323294406883835e-05,
      "loss": 1.3801,
      "step": 5456
    },
    {
      "epoch": 0.4193300553165335,
      "grad_norm": 3.3577568531036377,
      "learning_rate": 1.8322679778733868e-05,
      "loss": 1.3856,
      "step": 5458
    },
    {
      "epoch": 0.4194837123540258,
      "grad_norm": 3.043654203414917,
      "learning_rate": 1.8322065150583897e-05,
      "loss": 1.5341,
      "step": 5460
    },
    {
      "epoch": 0.41963736939151813,
      "grad_norm": 3.0720834732055664,
      "learning_rate": 1.832145052243393e-05,
      "loss": 1.3789,
      "step": 5462
    },
    {
      "epoch": 0.41979102642901045,
      "grad_norm": 3.7414183616638184,
      "learning_rate": 1.832083589428396e-05,
      "loss": 1.3286,
      "step": 5464
    },
    {
      "epoch": 0.41994468346650277,
      "grad_norm": 3.149899482727051,
      "learning_rate": 1.832022126613399e-05,
      "loss": 1.5493,
      "step": 5466
    },
    {
      "epoch": 0.4200983405039951,
      "grad_norm": 3.5700736045837402,
      "learning_rate": 1.8319606637984023e-05,
      "loss": 1.4777,
      "step": 5468
    },
    {
      "epoch": 0.4202519975414874,
      "grad_norm": 2.8731086254119873,
      "learning_rate": 1.8318992009834053e-05,
      "loss": 1.3664,
      "step": 5470
    },
    {
      "epoch": 0.4204056545789797,
      "grad_norm": 3.048271417617798,
      "learning_rate": 1.8318377381684082e-05,
      "loss": 1.5034,
      "step": 5472
    },
    {
      "epoch": 0.42055931161647203,
      "grad_norm": 2.9073920249938965,
      "learning_rate": 1.8317762753534115e-05,
      "loss": 1.3296,
      "step": 5474
    },
    {
      "epoch": 0.42071296865396435,
      "grad_norm": 3.0303492546081543,
      "learning_rate": 1.831714812538414e-05,
      "loss": 1.3736,
      "step": 5476
    },
    {
      "epoch": 0.42086662569145666,
      "grad_norm": 3.4354753494262695,
      "learning_rate": 1.8316533497234175e-05,
      "loss": 1.4299,
      "step": 5478
    },
    {
      "epoch": 0.421020282728949,
      "grad_norm": 3.6933016777038574,
      "learning_rate": 1.8315918869084204e-05,
      "loss": 1.4725,
      "step": 5480
    },
    {
      "epoch": 0.4211739397664413,
      "grad_norm": 3.131046772003174,
      "learning_rate": 1.8315304240934234e-05,
      "loss": 1.3902,
      "step": 5482
    },
    {
      "epoch": 0.4213275968039336,
      "grad_norm": 3.8743679523468018,
      "learning_rate": 1.8314689612784267e-05,
      "loss": 1.3884,
      "step": 5484
    },
    {
      "epoch": 0.42148125384142593,
      "grad_norm": 3.1388769149780273,
      "learning_rate": 1.8314074984634297e-05,
      "loss": 1.3651,
      "step": 5486
    },
    {
      "epoch": 0.42163491087891825,
      "grad_norm": 3.010105609893799,
      "learning_rate": 1.831346035648433e-05,
      "loss": 1.3886,
      "step": 5488
    },
    {
      "epoch": 0.42178856791641056,
      "grad_norm": 3.1552770137786865,
      "learning_rate": 1.831284572833436e-05,
      "loss": 1.511,
      "step": 5490
    },
    {
      "epoch": 0.4219422249539029,
      "grad_norm": 3.428985834121704,
      "learning_rate": 1.831223110018439e-05,
      "loss": 1.3053,
      "step": 5492
    },
    {
      "epoch": 0.4220958819913952,
      "grad_norm": 3.2660410404205322,
      "learning_rate": 1.8311616472034422e-05,
      "loss": 1.4879,
      "step": 5494
    },
    {
      "epoch": 0.4222495390288875,
      "grad_norm": 3.157233238220215,
      "learning_rate": 1.8311001843884452e-05,
      "loss": 1.3789,
      "step": 5496
    },
    {
      "epoch": 0.4224031960663798,
      "grad_norm": 3.170839786529541,
      "learning_rate": 1.8310387215734482e-05,
      "loss": 1.2828,
      "step": 5498
    },
    {
      "epoch": 0.42255685310387214,
      "grad_norm": 4.559313774108887,
      "learning_rate": 1.8309772587584515e-05,
      "loss": 1.4493,
      "step": 5500
    },
    {
      "epoch": 0.42271051014136446,
      "grad_norm": 3.0980498790740967,
      "learning_rate": 1.830915795943454e-05,
      "loss": 1.2786,
      "step": 5502
    },
    {
      "epoch": 0.4228641671788568,
      "grad_norm": 3.488553047180176,
      "learning_rate": 1.8308543331284574e-05,
      "loss": 1.5031,
      "step": 5504
    },
    {
      "epoch": 0.4230178242163491,
      "grad_norm": 3.1194798946380615,
      "learning_rate": 1.8307928703134604e-05,
      "loss": 1.2781,
      "step": 5506
    },
    {
      "epoch": 0.4231714812538414,
      "grad_norm": 3.33617901802063,
      "learning_rate": 1.8307314074984637e-05,
      "loss": 1.3998,
      "step": 5508
    },
    {
      "epoch": 0.4233251382913337,
      "grad_norm": 3.321866273880005,
      "learning_rate": 1.8306699446834667e-05,
      "loss": 1.412,
      "step": 5510
    },
    {
      "epoch": 0.42347879532882604,
      "grad_norm": 3.68853497505188,
      "learning_rate": 1.8306084818684696e-05,
      "loss": 1.374,
      "step": 5512
    },
    {
      "epoch": 0.42363245236631836,
      "grad_norm": 3.372403383255005,
      "learning_rate": 1.830547019053473e-05,
      "loss": 1.3341,
      "step": 5514
    },
    {
      "epoch": 0.4237861094038107,
      "grad_norm": 3.3488733768463135,
      "learning_rate": 1.830485556238476e-05,
      "loss": 1.3271,
      "step": 5516
    },
    {
      "epoch": 0.423939766441303,
      "grad_norm": 3.3170018196105957,
      "learning_rate": 1.830424093423479e-05,
      "loss": 1.5108,
      "step": 5518
    },
    {
      "epoch": 0.4240934234787953,
      "grad_norm": 3.225128412246704,
      "learning_rate": 1.8303626306084822e-05,
      "loss": 1.5062,
      "step": 5520
    },
    {
      "epoch": 0.4242470805162876,
      "grad_norm": 3.2453081607818604,
      "learning_rate": 1.830301167793485e-05,
      "loss": 1.3055,
      "step": 5522
    },
    {
      "epoch": 0.42440073755377994,
      "grad_norm": 2.8871963024139404,
      "learning_rate": 1.830239704978488e-05,
      "loss": 1.4496,
      "step": 5524
    },
    {
      "epoch": 0.42455439459127225,
      "grad_norm": 3.092216730117798,
      "learning_rate": 1.8301782421634914e-05,
      "loss": 1.3905,
      "step": 5526
    },
    {
      "epoch": 0.42470805162876457,
      "grad_norm": 3.2610034942626953,
      "learning_rate": 1.8301167793484944e-05,
      "loss": 1.4187,
      "step": 5528
    },
    {
      "epoch": 0.4248617086662569,
      "grad_norm": 3.0660836696624756,
      "learning_rate": 1.8300553165334974e-05,
      "loss": 1.4302,
      "step": 5530
    },
    {
      "epoch": 0.42501536570374926,
      "grad_norm": 2.9052979946136475,
      "learning_rate": 1.8299938537185003e-05,
      "loss": 1.3238,
      "step": 5532
    },
    {
      "epoch": 0.4251690227412416,
      "grad_norm": 3.2881360054016113,
      "learning_rate": 1.8299323909035036e-05,
      "loss": 1.6043,
      "step": 5534
    },
    {
      "epoch": 0.4253226797787339,
      "grad_norm": 3.306642532348633,
      "learning_rate": 1.8298709280885066e-05,
      "loss": 1.398,
      "step": 5536
    },
    {
      "epoch": 0.4254763368162262,
      "grad_norm": 3.069509983062744,
      "learning_rate": 1.8298094652735096e-05,
      "loss": 1.2468,
      "step": 5538
    },
    {
      "epoch": 0.4256299938537185,
      "grad_norm": 3.1374897956848145,
      "learning_rate": 1.829748002458513e-05,
      "loss": 1.479,
      "step": 5540
    },
    {
      "epoch": 0.42578365089121084,
      "grad_norm": 8.134029388427734,
      "learning_rate": 1.829686539643516e-05,
      "loss": 1.4549,
      "step": 5542
    },
    {
      "epoch": 0.42593730792870316,
      "grad_norm": 3.1467490196228027,
      "learning_rate": 1.8296250768285188e-05,
      "loss": 1.3115,
      "step": 5544
    },
    {
      "epoch": 0.4260909649661955,
      "grad_norm": 2.9459738731384277,
      "learning_rate": 1.829563614013522e-05,
      "loss": 1.2946,
      "step": 5546
    },
    {
      "epoch": 0.4262446220036878,
      "grad_norm": 2.924170970916748,
      "learning_rate": 1.829502151198525e-05,
      "loss": 1.4114,
      "step": 5548
    },
    {
      "epoch": 0.4263982790411801,
      "grad_norm": 3.5444743633270264,
      "learning_rate": 1.829440688383528e-05,
      "loss": 1.413,
      "step": 5550
    },
    {
      "epoch": 0.4265519360786724,
      "grad_norm": 3.2071783542633057,
      "learning_rate": 1.8293792255685314e-05,
      "loss": 1.4042,
      "step": 5552
    },
    {
      "epoch": 0.42670559311616474,
      "grad_norm": 3.367691993713379,
      "learning_rate": 1.8293177627535343e-05,
      "loss": 1.3649,
      "step": 5554
    },
    {
      "epoch": 0.42685925015365705,
      "grad_norm": 3.16751766204834,
      "learning_rate": 1.8292562999385373e-05,
      "loss": 1.3487,
      "step": 5556
    },
    {
      "epoch": 0.42701290719114937,
      "grad_norm": 3.2429380416870117,
      "learning_rate": 1.8291948371235403e-05,
      "loss": 1.3935,
      "step": 5558
    },
    {
      "epoch": 0.4271665642286417,
      "grad_norm": 4.151123046875,
      "learning_rate": 1.8291333743085436e-05,
      "loss": 1.3883,
      "step": 5560
    },
    {
      "epoch": 0.427320221266134,
      "grad_norm": 3.531403064727783,
      "learning_rate": 1.8290719114935465e-05,
      "loss": 1.4176,
      "step": 5562
    },
    {
      "epoch": 0.4274738783036263,
      "grad_norm": 2.87608003616333,
      "learning_rate": 1.8290104486785495e-05,
      "loss": 1.3478,
      "step": 5564
    },
    {
      "epoch": 0.42762753534111864,
      "grad_norm": 3.214362621307373,
      "learning_rate": 1.8289489858635528e-05,
      "loss": 1.3697,
      "step": 5566
    },
    {
      "epoch": 0.42778119237861095,
      "grad_norm": 3.0433409214019775,
      "learning_rate": 1.8288875230485558e-05,
      "loss": 1.4356,
      "step": 5568
    },
    {
      "epoch": 0.42793484941610327,
      "grad_norm": 3.267775297164917,
      "learning_rate": 1.8288260602335588e-05,
      "loss": 1.4391,
      "step": 5570
    },
    {
      "epoch": 0.4280885064535956,
      "grad_norm": 3.465855121612549,
      "learning_rate": 1.828764597418562e-05,
      "loss": 1.3637,
      "step": 5572
    },
    {
      "epoch": 0.4282421634910879,
      "grad_norm": 2.8692171573638916,
      "learning_rate": 1.828703134603565e-05,
      "loss": 1.3866,
      "step": 5574
    },
    {
      "epoch": 0.4283958205285802,
      "grad_norm": 2.991672992706299,
      "learning_rate": 1.828641671788568e-05,
      "loss": 1.3646,
      "step": 5576
    },
    {
      "epoch": 0.42854947756607253,
      "grad_norm": 3.042018413543701,
      "learning_rate": 1.828580208973571e-05,
      "loss": 1.3627,
      "step": 5578
    },
    {
      "epoch": 0.42870313460356485,
      "grad_norm": 3.0607712268829346,
      "learning_rate": 1.8285187461585743e-05,
      "loss": 1.4435,
      "step": 5580
    },
    {
      "epoch": 0.42885679164105717,
      "grad_norm": 3.6141839027404785,
      "learning_rate": 1.8284572833435772e-05,
      "loss": 1.3652,
      "step": 5582
    },
    {
      "epoch": 0.4290104486785495,
      "grad_norm": 2.990736722946167,
      "learning_rate": 1.8283958205285802e-05,
      "loss": 1.3635,
      "step": 5584
    },
    {
      "epoch": 0.4291641057160418,
      "grad_norm": 3.5047953128814697,
      "learning_rate": 1.8283343577135835e-05,
      "loss": 1.3506,
      "step": 5586
    },
    {
      "epoch": 0.4293177627535341,
      "grad_norm": 3.17103910446167,
      "learning_rate": 1.8282728948985865e-05,
      "loss": 1.2826,
      "step": 5588
    },
    {
      "epoch": 0.42947141979102643,
      "grad_norm": 2.9763083457946777,
      "learning_rate": 1.8282114320835895e-05,
      "loss": 1.5198,
      "step": 5590
    },
    {
      "epoch": 0.42962507682851875,
      "grad_norm": 3.388418197631836,
      "learning_rate": 1.8281499692685928e-05,
      "loss": 1.3798,
      "step": 5592
    },
    {
      "epoch": 0.42977873386601106,
      "grad_norm": 3.1480460166931152,
      "learning_rate": 1.8280885064535957e-05,
      "loss": 1.3923,
      "step": 5594
    },
    {
      "epoch": 0.4299323909035034,
      "grad_norm": 3.1141629219055176,
      "learning_rate": 1.8280270436385987e-05,
      "loss": 1.4089,
      "step": 5596
    },
    {
      "epoch": 0.4300860479409957,
      "grad_norm": 2.6607182025909424,
      "learning_rate": 1.827965580823602e-05,
      "loss": 1.3245,
      "step": 5598
    },
    {
      "epoch": 0.430239704978488,
      "grad_norm": 3.3764638900756836,
      "learning_rate": 1.827904118008605e-05,
      "loss": 1.3798,
      "step": 5600
    },
    {
      "epoch": 0.43039336201598033,
      "grad_norm": 2.802183151245117,
      "learning_rate": 1.827842655193608e-05,
      "loss": 1.3138,
      "step": 5602
    },
    {
      "epoch": 0.43054701905347265,
      "grad_norm": 3.2850708961486816,
      "learning_rate": 1.827781192378611e-05,
      "loss": 1.3341,
      "step": 5604
    },
    {
      "epoch": 0.43070067609096496,
      "grad_norm": 3.2753207683563232,
      "learning_rate": 1.8277197295636142e-05,
      "loss": 1.3529,
      "step": 5606
    },
    {
      "epoch": 0.4308543331284573,
      "grad_norm": 3.2558462619781494,
      "learning_rate": 1.8276582667486172e-05,
      "loss": 1.4033,
      "step": 5608
    },
    {
      "epoch": 0.4310079901659496,
      "grad_norm": 2.962024211883545,
      "learning_rate": 1.82759680393362e-05,
      "loss": 1.3368,
      "step": 5610
    },
    {
      "epoch": 0.4311616472034419,
      "grad_norm": 3.1479384899139404,
      "learning_rate": 1.8275353411186235e-05,
      "loss": 1.3557,
      "step": 5612
    },
    {
      "epoch": 0.4313153042409342,
      "grad_norm": 3.0928778648376465,
      "learning_rate": 1.8274738783036264e-05,
      "loss": 1.4615,
      "step": 5614
    },
    {
      "epoch": 0.43146896127842654,
      "grad_norm": 3.5253334045410156,
      "learning_rate": 1.8274124154886294e-05,
      "loss": 1.3321,
      "step": 5616
    },
    {
      "epoch": 0.43162261831591886,
      "grad_norm": 3.1185035705566406,
      "learning_rate": 1.8273509526736327e-05,
      "loss": 1.4343,
      "step": 5618
    },
    {
      "epoch": 0.4317762753534112,
      "grad_norm": 3.1733176708221436,
      "learning_rate": 1.8272894898586357e-05,
      "loss": 1.4088,
      "step": 5620
    },
    {
      "epoch": 0.4319299323909035,
      "grad_norm": 3.129121780395508,
      "learning_rate": 1.8272280270436386e-05,
      "loss": 1.3631,
      "step": 5622
    },
    {
      "epoch": 0.4320835894283958,
      "grad_norm": 3.0119454860687256,
      "learning_rate": 1.827166564228642e-05,
      "loss": 1.3877,
      "step": 5624
    },
    {
      "epoch": 0.4322372464658881,
      "grad_norm": 3.1937038898468018,
      "learning_rate": 1.827105101413645e-05,
      "loss": 1.3881,
      "step": 5626
    },
    {
      "epoch": 0.43239090350338044,
      "grad_norm": 2.997652769088745,
      "learning_rate": 1.827043638598648e-05,
      "loss": 1.3087,
      "step": 5628
    },
    {
      "epoch": 0.43254456054087276,
      "grad_norm": 2.841203212738037,
      "learning_rate": 1.826982175783651e-05,
      "loss": 1.2979,
      "step": 5630
    },
    {
      "epoch": 0.4326982175783651,
      "grad_norm": 3.034036636352539,
      "learning_rate": 1.826920712968654e-05,
      "loss": 1.229,
      "step": 5632
    },
    {
      "epoch": 0.4328518746158574,
      "grad_norm": 2.866893768310547,
      "learning_rate": 1.826859250153657e-05,
      "loss": 1.4217,
      "step": 5634
    },
    {
      "epoch": 0.4330055316533497,
      "grad_norm": 2.9957258701324463,
      "learning_rate": 1.82679778733866e-05,
      "loss": 1.5528,
      "step": 5636
    },
    {
      "epoch": 0.433159188690842,
      "grad_norm": 2.990673303604126,
      "learning_rate": 1.8267363245236634e-05,
      "loss": 1.4186,
      "step": 5638
    },
    {
      "epoch": 0.43331284572833434,
      "grad_norm": 3.3053412437438965,
      "learning_rate": 1.8266748617086664e-05,
      "loss": 1.4479,
      "step": 5640
    },
    {
      "epoch": 0.43346650276582666,
      "grad_norm": 3.35445237159729,
      "learning_rate": 1.8266133988936693e-05,
      "loss": 1.3,
      "step": 5642
    },
    {
      "epoch": 0.43362015980331897,
      "grad_norm": 3.1358654499053955,
      "learning_rate": 1.8265519360786727e-05,
      "loss": 1.3337,
      "step": 5644
    },
    {
      "epoch": 0.4337738168408113,
      "grad_norm": 3.132446050643921,
      "learning_rate": 1.8264904732636756e-05,
      "loss": 1.5941,
      "step": 5646
    },
    {
      "epoch": 0.4339274738783036,
      "grad_norm": 3.0220460891723633,
      "learning_rate": 1.8264290104486786e-05,
      "loss": 1.336,
      "step": 5648
    },
    {
      "epoch": 0.4340811309157959,
      "grad_norm": 3.412433385848999,
      "learning_rate": 1.826367547633682e-05,
      "loss": 1.4985,
      "step": 5650
    },
    {
      "epoch": 0.43423478795328824,
      "grad_norm": 3.2753655910491943,
      "learning_rate": 1.826306084818685e-05,
      "loss": 1.3647,
      "step": 5652
    },
    {
      "epoch": 0.43438844499078055,
      "grad_norm": 3.021898031234741,
      "learning_rate": 1.8262446220036882e-05,
      "loss": 1.416,
      "step": 5654
    },
    {
      "epoch": 0.43454210202827287,
      "grad_norm": 2.817472457885742,
      "learning_rate": 1.8261831591886908e-05,
      "loss": 1.3628,
      "step": 5656
    },
    {
      "epoch": 0.4346957590657652,
      "grad_norm": 2.896252155303955,
      "learning_rate": 1.826121696373694e-05,
      "loss": 1.4208,
      "step": 5658
    },
    {
      "epoch": 0.4348494161032575,
      "grad_norm": 3.5040619373321533,
      "learning_rate": 1.826060233558697e-05,
      "loss": 1.4589,
      "step": 5660
    },
    {
      "epoch": 0.4350030731407499,
      "grad_norm": 2.9098873138427734,
      "learning_rate": 1.8259987707437e-05,
      "loss": 1.4208,
      "step": 5662
    },
    {
      "epoch": 0.4351567301782422,
      "grad_norm": 3.0636959075927734,
      "learning_rate": 1.8259373079287034e-05,
      "loss": 1.4636,
      "step": 5664
    },
    {
      "epoch": 0.4353103872157345,
      "grad_norm": 2.761655807495117,
      "learning_rate": 1.8258758451137063e-05,
      "loss": 1.2639,
      "step": 5666
    },
    {
      "epoch": 0.4354640442532268,
      "grad_norm": 2.907930374145508,
      "learning_rate": 1.8258143822987093e-05,
      "loss": 1.498,
      "step": 5668
    },
    {
      "epoch": 0.43561770129071914,
      "grad_norm": 3.2700185775756836,
      "learning_rate": 1.8257529194837126e-05,
      "loss": 1.3444,
      "step": 5670
    },
    {
      "epoch": 0.43577135832821146,
      "grad_norm": 3.1657397747039795,
      "learning_rate": 1.8256914566687156e-05,
      "loss": 1.4352,
      "step": 5672
    },
    {
      "epoch": 0.43592501536570377,
      "grad_norm": 3.0951035022735596,
      "learning_rate": 1.825629993853719e-05,
      "loss": 1.3414,
      "step": 5674
    },
    {
      "epoch": 0.4360786724031961,
      "grad_norm": 2.763122797012329,
      "learning_rate": 1.8255685310387215e-05,
      "loss": 1.3182,
      "step": 5676
    },
    {
      "epoch": 0.4362323294406884,
      "grad_norm": 3.504037618637085,
      "learning_rate": 1.8255070682237248e-05,
      "loss": 1.3866,
      "step": 5678
    },
    {
      "epoch": 0.4363859864781807,
      "grad_norm": 3.1605093479156494,
      "learning_rate": 1.825445605408728e-05,
      "loss": 1.4574,
      "step": 5680
    },
    {
      "epoch": 0.43653964351567304,
      "grad_norm": 3.2070469856262207,
      "learning_rate": 1.8253841425937307e-05,
      "loss": 1.2922,
      "step": 5682
    },
    {
      "epoch": 0.43669330055316535,
      "grad_norm": 3.452393054962158,
      "learning_rate": 1.825322679778734e-05,
      "loss": 1.4231,
      "step": 5684
    },
    {
      "epoch": 0.43684695759065767,
      "grad_norm": 2.997011184692383,
      "learning_rate": 1.825261216963737e-05,
      "loss": 1.2949,
      "step": 5686
    },
    {
      "epoch": 0.43700061462815,
      "grad_norm": 3.404040813446045,
      "learning_rate": 1.82519975414874e-05,
      "loss": 1.3552,
      "step": 5688
    },
    {
      "epoch": 0.4371542716656423,
      "grad_norm": 3.3398168087005615,
      "learning_rate": 1.8251382913337433e-05,
      "loss": 1.3336,
      "step": 5690
    },
    {
      "epoch": 0.4373079287031346,
      "grad_norm": 3.090134859085083,
      "learning_rate": 1.8250768285187463e-05,
      "loss": 1.351,
      "step": 5692
    },
    {
      "epoch": 0.43746158574062693,
      "grad_norm": 3.0853729248046875,
      "learning_rate": 1.8250153657037496e-05,
      "loss": 1.2877,
      "step": 5694
    },
    {
      "epoch": 0.43761524277811925,
      "grad_norm": 2.9759137630462646,
      "learning_rate": 1.8249539028887525e-05,
      "loss": 1.3217,
      "step": 5696
    },
    {
      "epoch": 0.43776889981561157,
      "grad_norm": 3.0432422161102295,
      "learning_rate": 1.8248924400737555e-05,
      "loss": 1.4036,
      "step": 5698
    },
    {
      "epoch": 0.4379225568531039,
      "grad_norm": 2.5743117332458496,
      "learning_rate": 1.8248309772587588e-05,
      "loss": 1.3252,
      "step": 5700
    },
    {
      "epoch": 0.4380762138905962,
      "grad_norm": 2.912013530731201,
      "learning_rate": 1.8247695144437614e-05,
      "loss": 1.3125,
      "step": 5702
    },
    {
      "epoch": 0.4382298709280885,
      "grad_norm": 3.1190555095672607,
      "learning_rate": 1.8247080516287648e-05,
      "loss": 1.5832,
      "step": 5704
    },
    {
      "epoch": 0.43838352796558083,
      "grad_norm": 3.2913010120391846,
      "learning_rate": 1.8246465888137677e-05,
      "loss": 1.4592,
      "step": 5706
    },
    {
      "epoch": 0.43853718500307315,
      "grad_norm": 3.0158445835113525,
      "learning_rate": 1.8245851259987707e-05,
      "loss": 1.258,
      "step": 5708
    },
    {
      "epoch": 0.43869084204056547,
      "grad_norm": 2.7523088455200195,
      "learning_rate": 1.824523663183774e-05,
      "loss": 1.3033,
      "step": 5710
    },
    {
      "epoch": 0.4388444990780578,
      "grad_norm": 3.2428712844848633,
      "learning_rate": 1.824462200368777e-05,
      "loss": 1.5227,
      "step": 5712
    },
    {
      "epoch": 0.4389981561155501,
      "grad_norm": 3.1375772953033447,
      "learning_rate": 1.8244007375537803e-05,
      "loss": 1.2947,
      "step": 5714
    },
    {
      "epoch": 0.4391518131530424,
      "grad_norm": 3.0907890796661377,
      "learning_rate": 1.8243392747387832e-05,
      "loss": 1.5986,
      "step": 5716
    },
    {
      "epoch": 0.43930547019053473,
      "grad_norm": 2.9191091060638428,
      "learning_rate": 1.8242778119237862e-05,
      "loss": 1.3305,
      "step": 5718
    },
    {
      "epoch": 0.43945912722802705,
      "grad_norm": 2.765789747238159,
      "learning_rate": 1.8242163491087895e-05,
      "loss": 1.3996,
      "step": 5720
    },
    {
      "epoch": 0.43961278426551936,
      "grad_norm": 3.196028470993042,
      "learning_rate": 1.8241548862937925e-05,
      "loss": 1.4643,
      "step": 5722
    },
    {
      "epoch": 0.4397664413030117,
      "grad_norm": 3.159723997116089,
      "learning_rate": 1.8240934234787955e-05,
      "loss": 1.2982,
      "step": 5724
    },
    {
      "epoch": 0.439920098340504,
      "grad_norm": 4.066509246826172,
      "learning_rate": 1.8240319606637988e-05,
      "loss": 1.4798,
      "step": 5726
    },
    {
      "epoch": 0.4400737553779963,
      "grad_norm": 3.310293436050415,
      "learning_rate": 1.8239704978488014e-05,
      "loss": 1.3133,
      "step": 5728
    },
    {
      "epoch": 0.44022741241548863,
      "grad_norm": 3.166078805923462,
      "learning_rate": 1.8239090350338047e-05,
      "loss": 1.3163,
      "step": 5730
    },
    {
      "epoch": 0.44038106945298094,
      "grad_norm": 3.0429303646087646,
      "learning_rate": 1.8238475722188077e-05,
      "loss": 1.3178,
      "step": 5732
    },
    {
      "epoch": 0.44053472649047326,
      "grad_norm": 2.984344959259033,
      "learning_rate": 1.8237861094038106e-05,
      "loss": 1.4449,
      "step": 5734
    },
    {
      "epoch": 0.4406883835279656,
      "grad_norm": 3.379456043243408,
      "learning_rate": 1.823724646588814e-05,
      "loss": 1.3448,
      "step": 5736
    },
    {
      "epoch": 0.4408420405654579,
      "grad_norm": 3.0715973377227783,
      "learning_rate": 1.823663183773817e-05,
      "loss": 1.4384,
      "step": 5738
    },
    {
      "epoch": 0.4409956976029502,
      "grad_norm": 3.0291173458099365,
      "learning_rate": 1.8236017209588202e-05,
      "loss": 1.452,
      "step": 5740
    },
    {
      "epoch": 0.4411493546404425,
      "grad_norm": 3.0475244522094727,
      "learning_rate": 1.8235402581438232e-05,
      "loss": 1.3113,
      "step": 5742
    },
    {
      "epoch": 0.44130301167793484,
      "grad_norm": 3.0464026927948,
      "learning_rate": 1.823478795328826e-05,
      "loss": 1.2598,
      "step": 5744
    },
    {
      "epoch": 0.44145666871542716,
      "grad_norm": 3.1016945838928223,
      "learning_rate": 1.8234173325138295e-05,
      "loss": 1.396,
      "step": 5746
    },
    {
      "epoch": 0.4416103257529195,
      "grad_norm": 3.356527328491211,
      "learning_rate": 1.8233558696988324e-05,
      "loss": 1.4148,
      "step": 5748
    },
    {
      "epoch": 0.4417639827904118,
      "grad_norm": 2.8270328044891357,
      "learning_rate": 1.8232944068838354e-05,
      "loss": 1.346,
      "step": 5750
    },
    {
      "epoch": 0.4419176398279041,
      "grad_norm": 3.364140510559082,
      "learning_rate": 1.8232329440688387e-05,
      "loss": 1.3203,
      "step": 5752
    },
    {
      "epoch": 0.4420712968653964,
      "grad_norm": 3.246849298477173,
      "learning_rate": 1.8231714812538413e-05,
      "loss": 1.5842,
      "step": 5754
    },
    {
      "epoch": 0.44222495390288874,
      "grad_norm": 3.0349695682525635,
      "learning_rate": 1.8231100184388446e-05,
      "loss": 1.3516,
      "step": 5756
    },
    {
      "epoch": 0.44237861094038106,
      "grad_norm": 3.3905465602874756,
      "learning_rate": 1.8230485556238476e-05,
      "loss": 1.3804,
      "step": 5758
    },
    {
      "epoch": 0.4425322679778734,
      "grad_norm": 3.100095510482788,
      "learning_rate": 1.822987092808851e-05,
      "loss": 1.3149,
      "step": 5760
    },
    {
      "epoch": 0.4426859250153657,
      "grad_norm": 2.685727596282959,
      "learning_rate": 1.822925629993854e-05,
      "loss": 1.4374,
      "step": 5762
    },
    {
      "epoch": 0.442839582052858,
      "grad_norm": 3.614854335784912,
      "learning_rate": 1.822864167178857e-05,
      "loss": 1.4598,
      "step": 5764
    },
    {
      "epoch": 0.4429932390903503,
      "grad_norm": 3.2616031169891357,
      "learning_rate": 1.82280270436386e-05,
      "loss": 1.3823,
      "step": 5766
    },
    {
      "epoch": 0.44314689612784264,
      "grad_norm": 3.1923673152923584,
      "learning_rate": 1.822741241548863e-05,
      "loss": 1.3283,
      "step": 5768
    },
    {
      "epoch": 0.44330055316533495,
      "grad_norm": 2.938978433609009,
      "learning_rate": 1.822679778733866e-05,
      "loss": 1.3428,
      "step": 5770
    },
    {
      "epoch": 0.44345421020282727,
      "grad_norm": 2.983696222305298,
      "learning_rate": 1.8226183159188694e-05,
      "loss": 1.4124,
      "step": 5772
    },
    {
      "epoch": 0.4436078672403196,
      "grad_norm": 2.893930673599243,
      "learning_rate": 1.8225568531038724e-05,
      "loss": 1.3289,
      "step": 5774
    },
    {
      "epoch": 0.4437615242778119,
      "grad_norm": 2.9682881832122803,
      "learning_rate": 1.8224953902888753e-05,
      "loss": 1.3341,
      "step": 5776
    },
    {
      "epoch": 0.4439151813153042,
      "grad_norm": 3.4092354774475098,
      "learning_rate": 1.8224339274738786e-05,
      "loss": 1.4816,
      "step": 5778
    },
    {
      "epoch": 0.44406883835279654,
      "grad_norm": 3.190753698348999,
      "learning_rate": 1.8223724646588816e-05,
      "loss": 1.4722,
      "step": 5780
    },
    {
      "epoch": 0.44422249539028885,
      "grad_norm": 3.21653413772583,
      "learning_rate": 1.8223110018438846e-05,
      "loss": 1.3672,
      "step": 5782
    },
    {
      "epoch": 0.44437615242778117,
      "grad_norm": 2.8916025161743164,
      "learning_rate": 1.8222495390288876e-05,
      "loss": 1.3928,
      "step": 5784
    },
    {
      "epoch": 0.4445298094652735,
      "grad_norm": 3.229156494140625,
      "learning_rate": 1.822188076213891e-05,
      "loss": 1.3865,
      "step": 5786
    },
    {
      "epoch": 0.4446834665027658,
      "grad_norm": 3.3504526615142822,
      "learning_rate": 1.8221266133988938e-05,
      "loss": 1.4361,
      "step": 5788
    },
    {
      "epoch": 0.4448371235402581,
      "grad_norm": 3.143098831176758,
      "learning_rate": 1.8220651505838968e-05,
      "loss": 1.3759,
      "step": 5790
    },
    {
      "epoch": 0.44499078057775043,
      "grad_norm": 3.4050183296203613,
      "learning_rate": 1.8220036877689e-05,
      "loss": 1.4189,
      "step": 5792
    },
    {
      "epoch": 0.4451444376152428,
      "grad_norm": 2.914482593536377,
      "learning_rate": 1.821942224953903e-05,
      "loss": 1.4096,
      "step": 5794
    },
    {
      "epoch": 0.4452980946527351,
      "grad_norm": 3.0142135620117188,
      "learning_rate": 1.821880762138906e-05,
      "loss": 1.3218,
      "step": 5796
    },
    {
      "epoch": 0.44545175169022744,
      "grad_norm": 3.1696937084198,
      "learning_rate": 1.8218192993239093e-05,
      "loss": 1.3977,
      "step": 5798
    },
    {
      "epoch": 0.44560540872771975,
      "grad_norm": 3.157047986984253,
      "learning_rate": 1.8217578365089123e-05,
      "loss": 1.4031,
      "step": 5800
    },
    {
      "epoch": 0.44575906576521207,
      "grad_norm": 3.1851911544799805,
      "learning_rate": 1.8216963736939153e-05,
      "loss": 1.4248,
      "step": 5802
    },
    {
      "epoch": 0.4459127228027044,
      "grad_norm": 2.895350456237793,
      "learning_rate": 1.8216349108789183e-05,
      "loss": 1.3526,
      "step": 5804
    },
    {
      "epoch": 0.4460663798401967,
      "grad_norm": 3.438236713409424,
      "learning_rate": 1.8215734480639216e-05,
      "loss": 1.4285,
      "step": 5806
    },
    {
      "epoch": 0.446220036877689,
      "grad_norm": 3.439058780670166,
      "learning_rate": 1.8215119852489245e-05,
      "loss": 1.3343,
      "step": 5808
    },
    {
      "epoch": 0.44637369391518134,
      "grad_norm": 3.126304864883423,
      "learning_rate": 1.8214505224339275e-05,
      "loss": 1.4201,
      "step": 5810
    },
    {
      "epoch": 0.44652735095267365,
      "grad_norm": 2.642670154571533,
      "learning_rate": 1.8213890596189308e-05,
      "loss": 1.3266,
      "step": 5812
    },
    {
      "epoch": 0.44668100799016597,
      "grad_norm": 3.0952022075653076,
      "learning_rate": 1.8213275968039338e-05,
      "loss": 1.4502,
      "step": 5814
    },
    {
      "epoch": 0.4468346650276583,
      "grad_norm": 3.7239139080047607,
      "learning_rate": 1.8212661339889367e-05,
      "loss": 1.5575,
      "step": 5816
    },
    {
      "epoch": 0.4469883220651506,
      "grad_norm": 3.190653085708618,
      "learning_rate": 1.82120467117394e-05,
      "loss": 1.4421,
      "step": 5818
    },
    {
      "epoch": 0.4471419791026429,
      "grad_norm": 3.4979770183563232,
      "learning_rate": 1.821143208358943e-05,
      "loss": 1.3664,
      "step": 5820
    },
    {
      "epoch": 0.44729563614013523,
      "grad_norm": 3.044233798980713,
      "learning_rate": 1.821081745543946e-05,
      "loss": 1.3876,
      "step": 5822
    },
    {
      "epoch": 0.44744929317762755,
      "grad_norm": 3.0839719772338867,
      "learning_rate": 1.8210202827289493e-05,
      "loss": 1.294,
      "step": 5824
    },
    {
      "epoch": 0.44760295021511987,
      "grad_norm": 3.3031883239746094,
      "learning_rate": 1.8209588199139523e-05,
      "loss": 1.4067,
      "step": 5826
    },
    {
      "epoch": 0.4477566072526122,
      "grad_norm": 3.392949104309082,
      "learning_rate": 1.8208973570989552e-05,
      "loss": 1.4638,
      "step": 5828
    },
    {
      "epoch": 0.4479102642901045,
      "grad_norm": 3.3899407386779785,
      "learning_rate": 1.8208358942839582e-05,
      "loss": 1.3866,
      "step": 5830
    },
    {
      "epoch": 0.4480639213275968,
      "grad_norm": 2.626540422439575,
      "learning_rate": 1.8207744314689615e-05,
      "loss": 1.2544,
      "step": 5832
    },
    {
      "epoch": 0.44821757836508913,
      "grad_norm": 3.6084189414978027,
      "learning_rate": 1.8207129686539645e-05,
      "loss": 1.3872,
      "step": 5834
    },
    {
      "epoch": 0.44837123540258145,
      "grad_norm": 2.835602045059204,
      "learning_rate": 1.8206515058389674e-05,
      "loss": 1.3862,
      "step": 5836
    },
    {
      "epoch": 0.44852489244007376,
      "grad_norm": 2.9570329189300537,
      "learning_rate": 1.8205900430239707e-05,
      "loss": 1.4073,
      "step": 5838
    },
    {
      "epoch": 0.4486785494775661,
      "grad_norm": 2.910385847091675,
      "learning_rate": 1.8205285802089737e-05,
      "loss": 1.3464,
      "step": 5840
    },
    {
      "epoch": 0.4488322065150584,
      "grad_norm": 3.0790252685546875,
      "learning_rate": 1.8204671173939767e-05,
      "loss": 1.3937,
      "step": 5842
    },
    {
      "epoch": 0.4489858635525507,
      "grad_norm": 3.0580132007598877,
      "learning_rate": 1.82040565457898e-05,
      "loss": 1.2775,
      "step": 5844
    },
    {
      "epoch": 0.44913952059004303,
      "grad_norm": 2.8806533813476562,
      "learning_rate": 1.820344191763983e-05,
      "loss": 1.4138,
      "step": 5846
    },
    {
      "epoch": 0.44929317762753535,
      "grad_norm": 3.8401284217834473,
      "learning_rate": 1.820282728948986e-05,
      "loss": 1.3532,
      "step": 5848
    },
    {
      "epoch": 0.44944683466502766,
      "grad_norm": 3.237717628479004,
      "learning_rate": 1.8202212661339892e-05,
      "loss": 1.3316,
      "step": 5850
    },
    {
      "epoch": 0.44960049170252,
      "grad_norm": 2.958611011505127,
      "learning_rate": 1.8201598033189922e-05,
      "loss": 1.2768,
      "step": 5852
    },
    {
      "epoch": 0.4497541487400123,
      "grad_norm": 3.4589743614196777,
      "learning_rate": 1.8200983405039952e-05,
      "loss": 1.5463,
      "step": 5854
    },
    {
      "epoch": 0.4499078057775046,
      "grad_norm": 2.9172708988189697,
      "learning_rate": 1.820036877688998e-05,
      "loss": 1.4517,
      "step": 5856
    },
    {
      "epoch": 0.4500614628149969,
      "grad_norm": 3.213388681411743,
      "learning_rate": 1.8199754148740014e-05,
      "loss": 1.2623,
      "step": 5858
    },
    {
      "epoch": 0.45021511985248924,
      "grad_norm": 2.901362180709839,
      "learning_rate": 1.8199139520590044e-05,
      "loss": 1.2491,
      "step": 5860
    },
    {
      "epoch": 0.45036877688998156,
      "grad_norm": 3.0259957313537598,
      "learning_rate": 1.8198524892440074e-05,
      "loss": 1.5275,
      "step": 5862
    },
    {
      "epoch": 0.4505224339274739,
      "grad_norm": 3.00325870513916,
      "learning_rate": 1.8197910264290107e-05,
      "loss": 1.2732,
      "step": 5864
    },
    {
      "epoch": 0.4506760909649662,
      "grad_norm": 3.229722023010254,
      "learning_rate": 1.8197295636140137e-05,
      "loss": 1.4167,
      "step": 5866
    },
    {
      "epoch": 0.4508297480024585,
      "grad_norm": 3.2484633922576904,
      "learning_rate": 1.8196681007990166e-05,
      "loss": 1.3736,
      "step": 5868
    },
    {
      "epoch": 0.4509834050399508,
      "grad_norm": 2.872192859649658,
      "learning_rate": 1.81960663798402e-05,
      "loss": 1.4017,
      "step": 5870
    },
    {
      "epoch": 0.45113706207744314,
      "grad_norm": 3.5792553424835205,
      "learning_rate": 1.819545175169023e-05,
      "loss": 1.3938,
      "step": 5872
    },
    {
      "epoch": 0.45129071911493546,
      "grad_norm": 3.1069889068603516,
      "learning_rate": 1.819483712354026e-05,
      "loss": 1.3554,
      "step": 5874
    },
    {
      "epoch": 0.4514443761524278,
      "grad_norm": 3.3849682807922363,
      "learning_rate": 1.8194222495390292e-05,
      "loss": 1.3308,
      "step": 5876
    },
    {
      "epoch": 0.4515980331899201,
      "grad_norm": 3.07407808303833,
      "learning_rate": 1.819360786724032e-05,
      "loss": 1.3343,
      "step": 5878
    },
    {
      "epoch": 0.4517516902274124,
      "grad_norm": 3.1967318058013916,
      "learning_rate": 1.819299323909035e-05,
      "loss": 1.3538,
      "step": 5880
    },
    {
      "epoch": 0.4519053472649047,
      "grad_norm": 3.133248805999756,
      "learning_rate": 1.819237861094038e-05,
      "loss": 1.4287,
      "step": 5882
    },
    {
      "epoch": 0.45205900430239704,
      "grad_norm": 3.287682056427002,
      "learning_rate": 1.8191763982790414e-05,
      "loss": 1.3651,
      "step": 5884
    },
    {
      "epoch": 0.45221266133988935,
      "grad_norm": 3.5104236602783203,
      "learning_rate": 1.8191149354640444e-05,
      "loss": 1.3389,
      "step": 5886
    },
    {
      "epoch": 0.45236631837738167,
      "grad_norm": 3.1581168174743652,
      "learning_rate": 1.8190534726490473e-05,
      "loss": 1.493,
      "step": 5888
    },
    {
      "epoch": 0.452519975414874,
      "grad_norm": 3.408336877822876,
      "learning_rate": 1.8189920098340506e-05,
      "loss": 1.3866,
      "step": 5890
    },
    {
      "epoch": 0.4526736324523663,
      "grad_norm": 3.840761661529541,
      "learning_rate": 1.8189305470190536e-05,
      "loss": 1.5172,
      "step": 5892
    },
    {
      "epoch": 0.4528272894898586,
      "grad_norm": 3.4090771675109863,
      "learning_rate": 1.8188690842040566e-05,
      "loss": 1.3251,
      "step": 5894
    },
    {
      "epoch": 0.45298094652735094,
      "grad_norm": 2.792132616043091,
      "learning_rate": 1.81880762138906e-05,
      "loss": 1.4536,
      "step": 5896
    },
    {
      "epoch": 0.45313460356484325,
      "grad_norm": 2.6732709407806396,
      "learning_rate": 1.818746158574063e-05,
      "loss": 1.3353,
      "step": 5898
    },
    {
      "epoch": 0.45328826060233557,
      "grad_norm": 3.2554478645324707,
      "learning_rate": 1.8186846957590658e-05,
      "loss": 1.3426,
      "step": 5900
    },
    {
      "epoch": 0.4534419176398279,
      "grad_norm": 3.0684661865234375,
      "learning_rate": 1.8186232329440688e-05,
      "loss": 1.2457,
      "step": 5902
    },
    {
      "epoch": 0.4535955746773202,
      "grad_norm": 2.9400198459625244,
      "learning_rate": 1.818561770129072e-05,
      "loss": 1.386,
      "step": 5904
    },
    {
      "epoch": 0.4537492317148125,
      "grad_norm": 3.3350679874420166,
      "learning_rate": 1.8185003073140754e-05,
      "loss": 1.4366,
      "step": 5906
    },
    {
      "epoch": 0.45390288875230483,
      "grad_norm": 2.902355670928955,
      "learning_rate": 1.818438844499078e-05,
      "loss": 1.2785,
      "step": 5908
    },
    {
      "epoch": 0.45405654578979715,
      "grad_norm": 2.7429327964782715,
      "learning_rate": 1.8183773816840813e-05,
      "loss": 1.2129,
      "step": 5910
    },
    {
      "epoch": 0.45421020282728947,
      "grad_norm": 3.4028749465942383,
      "learning_rate": 1.8183159188690843e-05,
      "loss": 1.5344,
      "step": 5912
    },
    {
      "epoch": 0.4543638598647818,
      "grad_norm": 3.4940731525421143,
      "learning_rate": 1.8182544560540873e-05,
      "loss": 1.3376,
      "step": 5914
    },
    {
      "epoch": 0.4545175169022741,
      "grad_norm": 3.099562644958496,
      "learning_rate": 1.8181929932390906e-05,
      "loss": 1.3071,
      "step": 5916
    },
    {
      "epoch": 0.4546711739397664,
      "grad_norm": 2.8082468509674072,
      "learning_rate": 1.8181315304240935e-05,
      "loss": 1.273,
      "step": 5918
    },
    {
      "epoch": 0.45482483097725873,
      "grad_norm": 2.951359510421753,
      "learning_rate": 1.8180700676090965e-05,
      "loss": 1.2602,
      "step": 5920
    },
    {
      "epoch": 0.45497848801475105,
      "grad_norm": 3.421891212463379,
      "learning_rate": 1.8180086047940998e-05,
      "loss": 1.4899,
      "step": 5922
    },
    {
      "epoch": 0.4551321450522434,
      "grad_norm": 3.0956802368164062,
      "learning_rate": 1.8179471419791028e-05,
      "loss": 1.4078,
      "step": 5924
    },
    {
      "epoch": 0.45528580208973574,
      "grad_norm": 3.26657772064209,
      "learning_rate": 1.817885679164106e-05,
      "loss": 1.4338,
      "step": 5926
    },
    {
      "epoch": 0.45543945912722805,
      "grad_norm": 3.3916943073272705,
      "learning_rate": 1.8178242163491087e-05,
      "loss": 1.4113,
      "step": 5928
    },
    {
      "epoch": 0.45559311616472037,
      "grad_norm": 3.281963348388672,
      "learning_rate": 1.817762753534112e-05,
      "loss": 1.3594,
      "step": 5930
    },
    {
      "epoch": 0.4557467732022127,
      "grad_norm": 2.989858627319336,
      "learning_rate": 1.817701290719115e-05,
      "loss": 1.4444,
      "step": 5932
    },
    {
      "epoch": 0.455900430239705,
      "grad_norm": 2.8269715309143066,
      "learning_rate": 1.817639827904118e-05,
      "loss": 1.3883,
      "step": 5934
    },
    {
      "epoch": 0.4560540872771973,
      "grad_norm": 3.383490800857544,
      "learning_rate": 1.8175783650891213e-05,
      "loss": 1.4248,
      "step": 5936
    },
    {
      "epoch": 0.45620774431468963,
      "grad_norm": 3.237833023071289,
      "learning_rate": 1.8175169022741242e-05,
      "loss": 1.3974,
      "step": 5938
    },
    {
      "epoch": 0.45636140135218195,
      "grad_norm": 3.240793466567993,
      "learning_rate": 1.8174554394591272e-05,
      "loss": 1.3285,
      "step": 5940
    },
    {
      "epoch": 0.45651505838967427,
      "grad_norm": 3.475192070007324,
      "learning_rate": 1.8173939766441305e-05,
      "loss": 1.384,
      "step": 5942
    },
    {
      "epoch": 0.4566687154271666,
      "grad_norm": 3.198943853378296,
      "learning_rate": 1.8173325138291335e-05,
      "loss": 1.4785,
      "step": 5944
    },
    {
      "epoch": 0.4568223724646589,
      "grad_norm": 3.021594524383545,
      "learning_rate": 1.8172710510141368e-05,
      "loss": 1.3124,
      "step": 5946
    },
    {
      "epoch": 0.4569760295021512,
      "grad_norm": 3.0245521068573,
      "learning_rate": 1.8172095881991398e-05,
      "loss": 1.439,
      "step": 5948
    },
    {
      "epoch": 0.45712968653964353,
      "grad_norm": 3.3448681831359863,
      "learning_rate": 1.8171481253841427e-05,
      "loss": 1.401,
      "step": 5950
    },
    {
      "epoch": 0.45728334357713585,
      "grad_norm": 3.0003669261932373,
      "learning_rate": 1.817086662569146e-05,
      "loss": 1.4555,
      "step": 5952
    },
    {
      "epoch": 0.45743700061462816,
      "grad_norm": 3.3074393272399902,
      "learning_rate": 1.8170251997541487e-05,
      "loss": 1.3984,
      "step": 5954
    },
    {
      "epoch": 0.4575906576521205,
      "grad_norm": 3.2424259185791016,
      "learning_rate": 1.816963736939152e-05,
      "loss": 1.3531,
      "step": 5956
    },
    {
      "epoch": 0.4577443146896128,
      "grad_norm": 3.414992094039917,
      "learning_rate": 1.816902274124155e-05,
      "loss": 1.4555,
      "step": 5958
    },
    {
      "epoch": 0.4578979717271051,
      "grad_norm": 3.0615060329437256,
      "learning_rate": 1.816840811309158e-05,
      "loss": 1.5183,
      "step": 5960
    },
    {
      "epoch": 0.45805162876459743,
      "grad_norm": 3.127685070037842,
      "learning_rate": 1.8167793484941612e-05,
      "loss": 1.4996,
      "step": 5962
    },
    {
      "epoch": 0.45820528580208975,
      "grad_norm": 2.8747687339782715,
      "learning_rate": 1.8167178856791642e-05,
      "loss": 1.363,
      "step": 5964
    },
    {
      "epoch": 0.45835894283958206,
      "grad_norm": 3.215275764465332,
      "learning_rate": 1.8166564228641675e-05,
      "loss": 1.32,
      "step": 5966
    },
    {
      "epoch": 0.4585125998770744,
      "grad_norm": 2.8502352237701416,
      "learning_rate": 1.8165949600491705e-05,
      "loss": 1.3669,
      "step": 5968
    },
    {
      "epoch": 0.4586662569145667,
      "grad_norm": 3.226792812347412,
      "learning_rate": 1.8165334972341734e-05,
      "loss": 1.4084,
      "step": 5970
    },
    {
      "epoch": 0.458819913952059,
      "grad_norm": 2.946282386779785,
      "learning_rate": 1.8164720344191767e-05,
      "loss": 1.3186,
      "step": 5972
    },
    {
      "epoch": 0.4589735709895513,
      "grad_norm": 2.6620097160339355,
      "learning_rate": 1.8164105716041797e-05,
      "loss": 1.219,
      "step": 5974
    },
    {
      "epoch": 0.45912722802704364,
      "grad_norm": 3.052964925765991,
      "learning_rate": 1.8163491087891827e-05,
      "loss": 1.2451,
      "step": 5976
    },
    {
      "epoch": 0.45928088506453596,
      "grad_norm": 2.9971115589141846,
      "learning_rate": 1.816287645974186e-05,
      "loss": 1.3349,
      "step": 5978
    },
    {
      "epoch": 0.4594345421020283,
      "grad_norm": 2.765615701675415,
      "learning_rate": 1.8162261831591886e-05,
      "loss": 1.3993,
      "step": 5980
    },
    {
      "epoch": 0.4595881991395206,
      "grad_norm": 3.110050916671753,
      "learning_rate": 1.816164720344192e-05,
      "loss": 1.3793,
      "step": 5982
    },
    {
      "epoch": 0.4597418561770129,
      "grad_norm": 3.004164934158325,
      "learning_rate": 1.816103257529195e-05,
      "loss": 1.3538,
      "step": 5984
    },
    {
      "epoch": 0.4598955132145052,
      "grad_norm": 2.9276111125946045,
      "learning_rate": 1.8160417947141982e-05,
      "loss": 1.4343,
      "step": 5986
    },
    {
      "epoch": 0.46004917025199754,
      "grad_norm": 3.2313802242279053,
      "learning_rate": 1.815980331899201e-05,
      "loss": 1.4701,
      "step": 5988
    },
    {
      "epoch": 0.46020282728948986,
      "grad_norm": 3.9364774227142334,
      "learning_rate": 1.815918869084204e-05,
      "loss": 1.4805,
      "step": 5990
    },
    {
      "epoch": 0.4603564843269822,
      "grad_norm": 3.274184465408325,
      "learning_rate": 1.8158574062692074e-05,
      "loss": 1.3289,
      "step": 5992
    },
    {
      "epoch": 0.4605101413644745,
      "grad_norm": 3.1217334270477295,
      "learning_rate": 1.8157959434542104e-05,
      "loss": 1.4466,
      "step": 5994
    },
    {
      "epoch": 0.4606637984019668,
      "grad_norm": 2.6427552700042725,
      "learning_rate": 1.8157344806392134e-05,
      "loss": 1.3293,
      "step": 5996
    },
    {
      "epoch": 0.4608174554394591,
      "grad_norm": 3.3423068523406982,
      "learning_rate": 1.8156730178242167e-05,
      "loss": 1.3001,
      "step": 5998
    },
    {
      "epoch": 0.46097111247695144,
      "grad_norm": 3.421719551086426,
      "learning_rate": 1.8156115550092193e-05,
      "loss": 1.2599,
      "step": 6000
    },
    {
      "epoch": 0.46112476951444376,
      "grad_norm": 2.9063069820404053,
      "learning_rate": 1.8155500921942226e-05,
      "loss": 1.2448,
      "step": 6002
    },
    {
      "epoch": 0.46127842655193607,
      "grad_norm": 3.389843225479126,
      "learning_rate": 1.815488629379226e-05,
      "loss": 1.465,
      "step": 6004
    },
    {
      "epoch": 0.4614320835894284,
      "grad_norm": 3.1748673915863037,
      "learning_rate": 1.8154271665642286e-05,
      "loss": 1.442,
      "step": 6006
    },
    {
      "epoch": 0.4615857406269207,
      "grad_norm": 3.1274852752685547,
      "learning_rate": 1.815365703749232e-05,
      "loss": 1.3625,
      "step": 6008
    },
    {
      "epoch": 0.461739397664413,
      "grad_norm": 2.8795769214630127,
      "learning_rate": 1.815304240934235e-05,
      "loss": 1.3289,
      "step": 6010
    },
    {
      "epoch": 0.46189305470190534,
      "grad_norm": 2.991797924041748,
      "learning_rate": 1.815242778119238e-05,
      "loss": 1.3398,
      "step": 6012
    },
    {
      "epoch": 0.46204671173939765,
      "grad_norm": 2.745926856994629,
      "learning_rate": 1.815181315304241e-05,
      "loss": 1.3411,
      "step": 6014
    },
    {
      "epoch": 0.46220036877688997,
      "grad_norm": 3.4233834743499756,
      "learning_rate": 1.815119852489244e-05,
      "loss": 1.476,
      "step": 6016
    },
    {
      "epoch": 0.4623540258143823,
      "grad_norm": 3.179094076156616,
      "learning_rate": 1.8150583896742474e-05,
      "loss": 1.3088,
      "step": 6018
    },
    {
      "epoch": 0.4625076828518746,
      "grad_norm": 3.251293897628784,
      "learning_rate": 1.8149969268592504e-05,
      "loss": 1.3724,
      "step": 6020
    },
    {
      "epoch": 0.4626613398893669,
      "grad_norm": 3.3762452602386475,
      "learning_rate": 1.8149354640442533e-05,
      "loss": 1.4215,
      "step": 6022
    },
    {
      "epoch": 0.46281499692685923,
      "grad_norm": 2.5178112983703613,
      "learning_rate": 1.8148740012292566e-05,
      "loss": 1.2476,
      "step": 6024
    },
    {
      "epoch": 0.46296865396435155,
      "grad_norm": 3.1534478664398193,
      "learning_rate": 1.8148125384142593e-05,
      "loss": 1.2818,
      "step": 6026
    },
    {
      "epoch": 0.46312231100184387,
      "grad_norm": 2.7992680072784424,
      "learning_rate": 1.8147510755992626e-05,
      "loss": 1.3746,
      "step": 6028
    },
    {
      "epoch": 0.4632759680393362,
      "grad_norm": 3.016869068145752,
      "learning_rate": 1.8146896127842655e-05,
      "loss": 1.1759,
      "step": 6030
    },
    {
      "epoch": 0.4634296250768285,
      "grad_norm": 3.1666297912597656,
      "learning_rate": 1.814628149969269e-05,
      "loss": 1.2996,
      "step": 6032
    },
    {
      "epoch": 0.4635832821143208,
      "grad_norm": 3.2005155086517334,
      "learning_rate": 1.8145666871542718e-05,
      "loss": 1.354,
      "step": 6034
    },
    {
      "epoch": 0.46373693915181313,
      "grad_norm": 2.9398839473724365,
      "learning_rate": 1.8145052243392748e-05,
      "loss": 1.3371,
      "step": 6036
    },
    {
      "epoch": 0.46389059618930545,
      "grad_norm": 2.675481081008911,
      "learning_rate": 1.814443761524278e-05,
      "loss": 1.4086,
      "step": 6038
    },
    {
      "epoch": 0.46404425322679776,
      "grad_norm": 2.6939587593078613,
      "learning_rate": 1.814382298709281e-05,
      "loss": 1.4703,
      "step": 6040
    },
    {
      "epoch": 0.4641979102642901,
      "grad_norm": 3.3314614295959473,
      "learning_rate": 1.814320835894284e-05,
      "loss": 1.3802,
      "step": 6042
    },
    {
      "epoch": 0.4643515673017824,
      "grad_norm": 2.7407829761505127,
      "learning_rate": 1.8142593730792873e-05,
      "loss": 1.2443,
      "step": 6044
    },
    {
      "epoch": 0.4645052243392747,
      "grad_norm": 3.5611183643341064,
      "learning_rate": 1.8141979102642903e-05,
      "loss": 1.3102,
      "step": 6046
    },
    {
      "epoch": 0.46465888137676703,
      "grad_norm": 3.135925054550171,
      "learning_rate": 1.8141364474492933e-05,
      "loss": 1.2848,
      "step": 6048
    },
    {
      "epoch": 0.46481253841425935,
      "grad_norm": 2.807861566543579,
      "learning_rate": 1.8140749846342966e-05,
      "loss": 1.498,
      "step": 6050
    },
    {
      "epoch": 0.46496619545175166,
      "grad_norm": 2.9039602279663086,
      "learning_rate": 1.8140135218192995e-05,
      "loss": 1.3336,
      "step": 6052
    },
    {
      "epoch": 0.46511985248924403,
      "grad_norm": 3.3446543216705322,
      "learning_rate": 1.8139520590043025e-05,
      "loss": 1.3786,
      "step": 6054
    },
    {
      "epoch": 0.46527350952673635,
      "grad_norm": 3.125364065170288,
      "learning_rate": 1.8138905961893055e-05,
      "loss": 1.3462,
      "step": 6056
    },
    {
      "epoch": 0.46542716656422867,
      "grad_norm": 3.407083511352539,
      "learning_rate": 1.8138291333743088e-05,
      "loss": 1.4261,
      "step": 6058
    },
    {
      "epoch": 0.465580823601721,
      "grad_norm": 2.6646227836608887,
      "learning_rate": 1.8137676705593118e-05,
      "loss": 1.2944,
      "step": 6060
    },
    {
      "epoch": 0.4657344806392133,
      "grad_norm": 3.1010098457336426,
      "learning_rate": 1.8137062077443147e-05,
      "loss": 1.4409,
      "step": 6062
    },
    {
      "epoch": 0.4658881376767056,
      "grad_norm": 2.994729518890381,
      "learning_rate": 1.813644744929318e-05,
      "loss": 1.2887,
      "step": 6064
    },
    {
      "epoch": 0.46604179471419793,
      "grad_norm": 2.98126482963562,
      "learning_rate": 1.813583282114321e-05,
      "loss": 1.4378,
      "step": 6066
    },
    {
      "epoch": 0.46619545175169025,
      "grad_norm": 3.0406136512756348,
      "learning_rate": 1.813521819299324e-05,
      "loss": 1.4089,
      "step": 6068
    },
    {
      "epoch": 0.46634910878918256,
      "grad_norm": 2.912886619567871,
      "learning_rate": 1.8134603564843273e-05,
      "loss": 1.297,
      "step": 6070
    },
    {
      "epoch": 0.4665027658266749,
      "grad_norm": 3.1055288314819336,
      "learning_rate": 1.8133988936693302e-05,
      "loss": 1.3387,
      "step": 6072
    },
    {
      "epoch": 0.4666564228641672,
      "grad_norm": 2.698050022125244,
      "learning_rate": 1.8133374308543332e-05,
      "loss": 1.3589,
      "step": 6074
    },
    {
      "epoch": 0.4668100799016595,
      "grad_norm": 3.304744243621826,
      "learning_rate": 1.8132759680393365e-05,
      "loss": 1.3344,
      "step": 6076
    },
    {
      "epoch": 0.46696373693915183,
      "grad_norm": 3.2374093532562256,
      "learning_rate": 1.8132145052243395e-05,
      "loss": 1.3627,
      "step": 6078
    },
    {
      "epoch": 0.46711739397664415,
      "grad_norm": 3.138913631439209,
      "learning_rate": 1.8131530424093425e-05,
      "loss": 1.3703,
      "step": 6080
    },
    {
      "epoch": 0.46727105101413646,
      "grad_norm": 3.517970085144043,
      "learning_rate": 1.8130915795943454e-05,
      "loss": 1.3284,
      "step": 6082
    },
    {
      "epoch": 0.4674247080516288,
      "grad_norm": 2.9830198287963867,
      "learning_rate": 1.8130301167793487e-05,
      "loss": 1.3917,
      "step": 6084
    },
    {
      "epoch": 0.4675783650891211,
      "grad_norm": 2.6390559673309326,
      "learning_rate": 1.8129686539643517e-05,
      "loss": 1.3801,
      "step": 6086
    },
    {
      "epoch": 0.4677320221266134,
      "grad_norm": 3.0859718322753906,
      "learning_rate": 1.8129071911493547e-05,
      "loss": 1.4656,
      "step": 6088
    },
    {
      "epoch": 0.46788567916410573,
      "grad_norm": 2.984755516052246,
      "learning_rate": 1.812845728334358e-05,
      "loss": 1.4782,
      "step": 6090
    },
    {
      "epoch": 0.46803933620159804,
      "grad_norm": 3.045100450515747,
      "learning_rate": 1.812784265519361e-05,
      "loss": 1.4161,
      "step": 6092
    },
    {
      "epoch": 0.46819299323909036,
      "grad_norm": 3.148865222930908,
      "learning_rate": 1.812722802704364e-05,
      "loss": 1.3084,
      "step": 6094
    },
    {
      "epoch": 0.4683466502765827,
      "grad_norm": 3.177959680557251,
      "learning_rate": 1.8126613398893672e-05,
      "loss": 1.3234,
      "step": 6096
    },
    {
      "epoch": 0.468500307314075,
      "grad_norm": 2.8266592025756836,
      "learning_rate": 1.8125998770743702e-05,
      "loss": 1.4313,
      "step": 6098
    },
    {
      "epoch": 0.4686539643515673,
      "grad_norm": 3.3804705142974854,
      "learning_rate": 1.812538414259373e-05,
      "loss": 1.3802,
      "step": 6100
    },
    {
      "epoch": 0.4688076213890596,
      "grad_norm": 3.3395400047302246,
      "learning_rate": 1.8124769514443765e-05,
      "loss": 1.2999,
      "step": 6102
    },
    {
      "epoch": 0.46896127842655194,
      "grad_norm": 2.902346134185791,
      "learning_rate": 1.8124154886293794e-05,
      "loss": 1.3411,
      "step": 6104
    },
    {
      "epoch": 0.46911493546404426,
      "grad_norm": 3.264467477798462,
      "learning_rate": 1.8123540258143824e-05,
      "loss": 1.4063,
      "step": 6106
    },
    {
      "epoch": 0.4692685925015366,
      "grad_norm": 2.926862955093384,
      "learning_rate": 1.8122925629993854e-05,
      "loss": 1.2848,
      "step": 6108
    },
    {
      "epoch": 0.4694222495390289,
      "grad_norm": 3.159416913986206,
      "learning_rate": 1.8122311001843887e-05,
      "loss": 1.526,
      "step": 6110
    },
    {
      "epoch": 0.4695759065765212,
      "grad_norm": 2.464360237121582,
      "learning_rate": 1.8121696373693916e-05,
      "loss": 1.2014,
      "step": 6112
    },
    {
      "epoch": 0.4697295636140135,
      "grad_norm": 3.0088672637939453,
      "learning_rate": 1.8121081745543946e-05,
      "loss": 1.4058,
      "step": 6114
    },
    {
      "epoch": 0.46988322065150584,
      "grad_norm": 2.786822557449341,
      "learning_rate": 1.812046711739398e-05,
      "loss": 1.2714,
      "step": 6116
    },
    {
      "epoch": 0.47003687768899816,
      "grad_norm": 2.730426549911499,
      "learning_rate": 1.811985248924401e-05,
      "loss": 1.1898,
      "step": 6118
    },
    {
      "epoch": 0.47019053472649047,
      "grad_norm": 2.9403066635131836,
      "learning_rate": 1.811923786109404e-05,
      "loss": 1.4679,
      "step": 6120
    },
    {
      "epoch": 0.4703441917639828,
      "grad_norm": 2.8969335556030273,
      "learning_rate": 1.811862323294407e-05,
      "loss": 1.3628,
      "step": 6122
    },
    {
      "epoch": 0.4704978488014751,
      "grad_norm": 3.1321475505828857,
      "learning_rate": 1.81180086047941e-05,
      "loss": 1.2599,
      "step": 6124
    },
    {
      "epoch": 0.4706515058389674,
      "grad_norm": 2.7310781478881836,
      "learning_rate": 1.811739397664413e-05,
      "loss": 1.343,
      "step": 6126
    },
    {
      "epoch": 0.47080516287645974,
      "grad_norm": 3.2368597984313965,
      "learning_rate": 1.811677934849416e-05,
      "loss": 1.452,
      "step": 6128
    },
    {
      "epoch": 0.47095881991395205,
      "grad_norm": 3.1757235527038574,
      "learning_rate": 1.8116164720344194e-05,
      "loss": 1.2629,
      "step": 6130
    },
    {
      "epoch": 0.47111247695144437,
      "grad_norm": 3.524467706680298,
      "learning_rate": 1.8115550092194223e-05,
      "loss": 1.3946,
      "step": 6132
    },
    {
      "epoch": 0.4712661339889367,
      "grad_norm": 2.9439032077789307,
      "learning_rate": 1.8114935464044253e-05,
      "loss": 1.3863,
      "step": 6134
    },
    {
      "epoch": 0.471419791026429,
      "grad_norm": 3.1204442977905273,
      "learning_rate": 1.8114320835894286e-05,
      "loss": 1.3305,
      "step": 6136
    },
    {
      "epoch": 0.4715734480639213,
      "grad_norm": 2.8491554260253906,
      "learning_rate": 1.8113706207744316e-05,
      "loss": 1.3718,
      "step": 6138
    },
    {
      "epoch": 0.47172710510141364,
      "grad_norm": 3.3984553813934326,
      "learning_rate": 1.8113091579594346e-05,
      "loss": 1.3786,
      "step": 6140
    },
    {
      "epoch": 0.47188076213890595,
      "grad_norm": 2.843414545059204,
      "learning_rate": 1.811247695144438e-05,
      "loss": 1.4403,
      "step": 6142
    },
    {
      "epoch": 0.47203441917639827,
      "grad_norm": 3.248155355453491,
      "learning_rate": 1.811186232329441e-05,
      "loss": 1.3445,
      "step": 6144
    },
    {
      "epoch": 0.4721880762138906,
      "grad_norm": 2.920718193054199,
      "learning_rate": 1.8111247695144438e-05,
      "loss": 1.4953,
      "step": 6146
    },
    {
      "epoch": 0.4723417332513829,
      "grad_norm": 2.642434597015381,
      "learning_rate": 1.811063306699447e-05,
      "loss": 1.4154,
      "step": 6148
    },
    {
      "epoch": 0.4724953902888752,
      "grad_norm": 2.8133106231689453,
      "learning_rate": 1.81100184388445e-05,
      "loss": 1.2154,
      "step": 6150
    },
    {
      "epoch": 0.47264904732636753,
      "grad_norm": 2.921673059463501,
      "learning_rate": 1.810940381069453e-05,
      "loss": 1.4372,
      "step": 6152
    },
    {
      "epoch": 0.47280270436385985,
      "grad_norm": 2.9900777339935303,
      "learning_rate": 1.810878918254456e-05,
      "loss": 1.3489,
      "step": 6154
    },
    {
      "epoch": 0.47295636140135217,
      "grad_norm": 3.1354458332061768,
      "learning_rate": 1.8108174554394593e-05,
      "loss": 1.3848,
      "step": 6156
    },
    {
      "epoch": 0.4731100184388445,
      "grad_norm": 2.98395037651062,
      "learning_rate": 1.8107559926244623e-05,
      "loss": 1.1782,
      "step": 6158
    },
    {
      "epoch": 0.4732636754763368,
      "grad_norm": 3.0558385848999023,
      "learning_rate": 1.8106945298094653e-05,
      "loss": 1.4362,
      "step": 6160
    },
    {
      "epoch": 0.4734173325138291,
      "grad_norm": 2.871683120727539,
      "learning_rate": 1.8106330669944686e-05,
      "loss": 1.361,
      "step": 6162
    },
    {
      "epoch": 0.47357098955132143,
      "grad_norm": 3.1952767372131348,
      "learning_rate": 1.8105716041794715e-05,
      "loss": 1.371,
      "step": 6164
    },
    {
      "epoch": 0.47372464658881375,
      "grad_norm": 3.256314754486084,
      "learning_rate": 1.8105101413644745e-05,
      "loss": 1.5385,
      "step": 6166
    },
    {
      "epoch": 0.47387830362630606,
      "grad_norm": 3.1399924755096436,
      "learning_rate": 1.8104486785494778e-05,
      "loss": 1.4099,
      "step": 6168
    },
    {
      "epoch": 0.4740319606637984,
      "grad_norm": 2.872492551803589,
      "learning_rate": 1.8103872157344808e-05,
      "loss": 1.3904,
      "step": 6170
    },
    {
      "epoch": 0.4741856177012907,
      "grad_norm": 3.3130695819854736,
      "learning_rate": 1.8103257529194837e-05,
      "loss": 1.4766,
      "step": 6172
    },
    {
      "epoch": 0.474339274738783,
      "grad_norm": 3.3041231632232666,
      "learning_rate": 1.810264290104487e-05,
      "loss": 1.3754,
      "step": 6174
    },
    {
      "epoch": 0.47449293177627533,
      "grad_norm": 2.94787859916687,
      "learning_rate": 1.81020282728949e-05,
      "loss": 1.407,
      "step": 6176
    },
    {
      "epoch": 0.47464658881376764,
      "grad_norm": 2.9895570278167725,
      "learning_rate": 1.8101413644744933e-05,
      "loss": 1.4253,
      "step": 6178
    },
    {
      "epoch": 0.47480024585125996,
      "grad_norm": 3.096012830734253,
      "learning_rate": 1.810079901659496e-05,
      "loss": 1.4598,
      "step": 6180
    },
    {
      "epoch": 0.4749539028887523,
      "grad_norm": 3.4266176223754883,
      "learning_rate": 1.8100184388444993e-05,
      "loss": 1.3014,
      "step": 6182
    },
    {
      "epoch": 0.47510755992624465,
      "grad_norm": 3.3502449989318848,
      "learning_rate": 1.8099569760295022e-05,
      "loss": 1.403,
      "step": 6184
    },
    {
      "epoch": 0.47526121696373697,
      "grad_norm": 2.791080951690674,
      "learning_rate": 1.8098955132145052e-05,
      "loss": 1.3161,
      "step": 6186
    },
    {
      "epoch": 0.4754148740012293,
      "grad_norm": 3.389315366744995,
      "learning_rate": 1.8098340503995085e-05,
      "loss": 1.3105,
      "step": 6188
    },
    {
      "epoch": 0.4755685310387216,
      "grad_norm": 3.1894092559814453,
      "learning_rate": 1.8097725875845115e-05,
      "loss": 1.334,
      "step": 6190
    },
    {
      "epoch": 0.4757221880762139,
      "grad_norm": 3.219374418258667,
      "learning_rate": 1.8097111247695144e-05,
      "loss": 1.2583,
      "step": 6192
    },
    {
      "epoch": 0.47587584511370623,
      "grad_norm": 2.9629125595092773,
      "learning_rate": 1.8096496619545178e-05,
      "loss": 1.472,
      "step": 6194
    },
    {
      "epoch": 0.47602950215119855,
      "grad_norm": 2.8309929370880127,
      "learning_rate": 1.8095881991395207e-05,
      "loss": 1.4128,
      "step": 6196
    },
    {
      "epoch": 0.47618315918869086,
      "grad_norm": 3.1029868125915527,
      "learning_rate": 1.809526736324524e-05,
      "loss": 1.4316,
      "step": 6198
    },
    {
      "epoch": 0.4763368162261832,
      "grad_norm": 3.910332679748535,
      "learning_rate": 1.809465273509527e-05,
      "loss": 1.4264,
      "step": 6200
    },
    {
      "epoch": 0.4764904732636755,
      "grad_norm": 3.1213109493255615,
      "learning_rate": 1.80940381069453e-05,
      "loss": 1.3166,
      "step": 6202
    },
    {
      "epoch": 0.4766441303011678,
      "grad_norm": 3.3485963344573975,
      "learning_rate": 1.8093423478795333e-05,
      "loss": 1.4312,
      "step": 6204
    },
    {
      "epoch": 0.47679778733866013,
      "grad_norm": 2.8326988220214844,
      "learning_rate": 1.809280885064536e-05,
      "loss": 1.3337,
      "step": 6206
    },
    {
      "epoch": 0.47695144437615244,
      "grad_norm": 3.0213944911956787,
      "learning_rate": 1.8092194222495392e-05,
      "loss": 1.3796,
      "step": 6208
    },
    {
      "epoch": 0.47710510141364476,
      "grad_norm": 3.090485095977783,
      "learning_rate": 1.8091579594345422e-05,
      "loss": 1.3794,
      "step": 6210
    },
    {
      "epoch": 0.4772587584511371,
      "grad_norm": 2.931671619415283,
      "learning_rate": 1.809096496619545e-05,
      "loss": 1.4677,
      "step": 6212
    },
    {
      "epoch": 0.4774124154886294,
      "grad_norm": 3.267240524291992,
      "learning_rate": 1.8090350338045485e-05,
      "loss": 1.3751,
      "step": 6214
    },
    {
      "epoch": 0.4775660725261217,
      "grad_norm": 2.553067207336426,
      "learning_rate": 1.8089735709895514e-05,
      "loss": 1.3013,
      "step": 6216
    },
    {
      "epoch": 0.477719729563614,
      "grad_norm": 3.1998229026794434,
      "learning_rate": 1.8089121081745547e-05,
      "loss": 1.3474,
      "step": 6218
    },
    {
      "epoch": 0.47787338660110634,
      "grad_norm": 11.610280990600586,
      "learning_rate": 1.8088506453595577e-05,
      "loss": 1.2955,
      "step": 6220
    },
    {
      "epoch": 0.47802704363859866,
      "grad_norm": 2.8859567642211914,
      "learning_rate": 1.8087891825445607e-05,
      "loss": 1.3495,
      "step": 6222
    },
    {
      "epoch": 0.478180700676091,
      "grad_norm": 2.9452872276306152,
      "learning_rate": 1.808727719729564e-05,
      "loss": 1.276,
      "step": 6224
    },
    {
      "epoch": 0.4783343577135833,
      "grad_norm": 3.28800892829895,
      "learning_rate": 1.8086662569145666e-05,
      "loss": 1.4918,
      "step": 6226
    },
    {
      "epoch": 0.4784880147510756,
      "grad_norm": 3.066276788711548,
      "learning_rate": 1.80860479409957e-05,
      "loss": 1.3771,
      "step": 6228
    },
    {
      "epoch": 0.4786416717885679,
      "grad_norm": 2.8848133087158203,
      "learning_rate": 1.8085433312845732e-05,
      "loss": 1.3354,
      "step": 6230
    },
    {
      "epoch": 0.47879532882606024,
      "grad_norm": 2.9447460174560547,
      "learning_rate": 1.808481868469576e-05,
      "loss": 1.4236,
      "step": 6232
    },
    {
      "epoch": 0.47894898586355256,
      "grad_norm": 3.061072826385498,
      "learning_rate": 1.808420405654579e-05,
      "loss": 1.2652,
      "step": 6234
    },
    {
      "epoch": 0.4791026429010449,
      "grad_norm": 3.3094377517700195,
      "learning_rate": 1.808358942839582e-05,
      "loss": 1.3274,
      "step": 6236
    },
    {
      "epoch": 0.4792562999385372,
      "grad_norm": 2.868401050567627,
      "learning_rate": 1.8082974800245854e-05,
      "loss": 1.361,
      "step": 6238
    },
    {
      "epoch": 0.4794099569760295,
      "grad_norm": 2.7703821659088135,
      "learning_rate": 1.8082360172095884e-05,
      "loss": 1.3177,
      "step": 6240
    },
    {
      "epoch": 0.4795636140135218,
      "grad_norm": 2.5703728199005127,
      "learning_rate": 1.8081745543945914e-05,
      "loss": 1.3978,
      "step": 6242
    },
    {
      "epoch": 0.47971727105101414,
      "grad_norm": 3.038760185241699,
      "learning_rate": 1.8081130915795947e-05,
      "loss": 1.3651,
      "step": 6244
    },
    {
      "epoch": 0.47987092808850645,
      "grad_norm": 3.6976478099823,
      "learning_rate": 1.8080516287645976e-05,
      "loss": 1.4703,
      "step": 6246
    },
    {
      "epoch": 0.48002458512599877,
      "grad_norm": 3.0694258213043213,
      "learning_rate": 1.8079901659496006e-05,
      "loss": 1.3212,
      "step": 6248
    },
    {
      "epoch": 0.4801782421634911,
      "grad_norm": 3.3709535598754883,
      "learning_rate": 1.807928703134604e-05,
      "loss": 1.36,
      "step": 6250
    },
    {
      "epoch": 0.4803318992009834,
      "grad_norm": 3.294551134109497,
      "learning_rate": 1.8078672403196065e-05,
      "loss": 1.5437,
      "step": 6252
    },
    {
      "epoch": 0.4804855562384757,
      "grad_norm": 3.8295395374298096,
      "learning_rate": 1.80780577750461e-05,
      "loss": 1.2844,
      "step": 6254
    },
    {
      "epoch": 0.48063921327596804,
      "grad_norm": 2.620631217956543,
      "learning_rate": 1.8077443146896128e-05,
      "loss": 1.2676,
      "step": 6256
    },
    {
      "epoch": 0.48079287031346035,
      "grad_norm": 2.7182457447052,
      "learning_rate": 1.8076828518746158e-05,
      "loss": 1.4544,
      "step": 6258
    },
    {
      "epoch": 0.48094652735095267,
      "grad_norm": 3.396521806716919,
      "learning_rate": 1.807621389059619e-05,
      "loss": 1.3121,
      "step": 6260
    },
    {
      "epoch": 0.481100184388445,
      "grad_norm": 2.63437557220459,
      "learning_rate": 1.807559926244622e-05,
      "loss": 1.234,
      "step": 6262
    },
    {
      "epoch": 0.4812538414259373,
      "grad_norm": 2.776506185531616,
      "learning_rate": 1.8074984634296254e-05,
      "loss": 1.4487,
      "step": 6264
    },
    {
      "epoch": 0.4814074984634296,
      "grad_norm": 3.227975845336914,
      "learning_rate": 1.8074370006146283e-05,
      "loss": 1.2383,
      "step": 6266
    },
    {
      "epoch": 0.48156115550092193,
      "grad_norm": 2.9529471397399902,
      "learning_rate": 1.8073755377996313e-05,
      "loss": 1.3515,
      "step": 6268
    },
    {
      "epoch": 0.48171481253841425,
      "grad_norm": 3.3522536754608154,
      "learning_rate": 1.8073140749846346e-05,
      "loss": 1.3238,
      "step": 6270
    },
    {
      "epoch": 0.48186846957590657,
      "grad_norm": 3.3435351848602295,
      "learning_rate": 1.8072526121696376e-05,
      "loss": 1.4359,
      "step": 6272
    },
    {
      "epoch": 0.4820221266133989,
      "grad_norm": 2.8637821674346924,
      "learning_rate": 1.8071911493546406e-05,
      "loss": 1.2188,
      "step": 6274
    },
    {
      "epoch": 0.4821757836508912,
      "grad_norm": 2.939358949661255,
      "learning_rate": 1.807129686539644e-05,
      "loss": 1.2918,
      "step": 6276
    },
    {
      "epoch": 0.4823294406883835,
      "grad_norm": 3.1831789016723633,
      "learning_rate": 1.8070682237246465e-05,
      "loss": 1.4176,
      "step": 6278
    },
    {
      "epoch": 0.48248309772587583,
      "grad_norm": 3.3042044639587402,
      "learning_rate": 1.8070067609096498e-05,
      "loss": 1.3979,
      "step": 6280
    },
    {
      "epoch": 0.48263675476336815,
      "grad_norm": 3.7356269359588623,
      "learning_rate": 1.8069452980946528e-05,
      "loss": 1.3986,
      "step": 6282
    },
    {
      "epoch": 0.48279041180086046,
      "grad_norm": 3.174906015396118,
      "learning_rate": 1.806883835279656e-05,
      "loss": 1.2884,
      "step": 6284
    },
    {
      "epoch": 0.4829440688383528,
      "grad_norm": 2.9945216178894043,
      "learning_rate": 1.806822372464659e-05,
      "loss": 1.4439,
      "step": 6286
    },
    {
      "epoch": 0.4830977258758451,
      "grad_norm": 3.113851308822632,
      "learning_rate": 1.806760909649662e-05,
      "loss": 1.3341,
      "step": 6288
    },
    {
      "epoch": 0.4832513829133374,
      "grad_norm": 3.578928232192993,
      "learning_rate": 1.8066994468346653e-05,
      "loss": 1.4566,
      "step": 6290
    },
    {
      "epoch": 0.48340503995082973,
      "grad_norm": 3.227860450744629,
      "learning_rate": 1.8066379840196683e-05,
      "loss": 1.3845,
      "step": 6292
    },
    {
      "epoch": 0.48355869698832205,
      "grad_norm": 3.249185562133789,
      "learning_rate": 1.8065765212046713e-05,
      "loss": 1.3953,
      "step": 6294
    },
    {
      "epoch": 0.48371235402581436,
      "grad_norm": 2.9868662357330322,
      "learning_rate": 1.8065150583896746e-05,
      "loss": 1.2176,
      "step": 6296
    },
    {
      "epoch": 0.4838660110633067,
      "grad_norm": 2.748054265975952,
      "learning_rate": 1.8064535955746775e-05,
      "loss": 1.251,
      "step": 6298
    },
    {
      "epoch": 0.484019668100799,
      "grad_norm": 3.2431583404541016,
      "learning_rate": 1.8063921327596805e-05,
      "loss": 1.3409,
      "step": 6300
    },
    {
      "epoch": 0.4841733251382913,
      "grad_norm": 3.484886407852173,
      "learning_rate": 1.8063306699446838e-05,
      "loss": 1.3708,
      "step": 6302
    },
    {
      "epoch": 0.4843269821757836,
      "grad_norm": 3.2029964923858643,
      "learning_rate": 1.8062692071296868e-05,
      "loss": 1.4806,
      "step": 6304
    },
    {
      "epoch": 0.48448063921327594,
      "grad_norm": 2.78397274017334,
      "learning_rate": 1.8062077443146897e-05,
      "loss": 1.2184,
      "step": 6306
    },
    {
      "epoch": 0.48463429625076826,
      "grad_norm": 2.9794437885284424,
      "learning_rate": 1.8061462814996927e-05,
      "loss": 1.4035,
      "step": 6308
    },
    {
      "epoch": 0.4847879532882606,
      "grad_norm": 2.6800053119659424,
      "learning_rate": 1.806084818684696e-05,
      "loss": 1.3018,
      "step": 6310
    },
    {
      "epoch": 0.4849416103257529,
      "grad_norm": 2.774409055709839,
      "learning_rate": 1.806023355869699e-05,
      "loss": 1.4247,
      "step": 6312
    },
    {
      "epoch": 0.48509526736324526,
      "grad_norm": 3.3923118114471436,
      "learning_rate": 1.805961893054702e-05,
      "loss": 1.1499,
      "step": 6314
    },
    {
      "epoch": 0.4852489244007376,
      "grad_norm": 2.7925145626068115,
      "learning_rate": 1.8059004302397053e-05,
      "loss": 1.2022,
      "step": 6316
    },
    {
      "epoch": 0.4854025814382299,
      "grad_norm": 3.3250386714935303,
      "learning_rate": 1.8058389674247082e-05,
      "loss": 1.3755,
      "step": 6318
    },
    {
      "epoch": 0.4855562384757222,
      "grad_norm": 3.531944990158081,
      "learning_rate": 1.8057775046097112e-05,
      "loss": 1.4049,
      "step": 6320
    },
    {
      "epoch": 0.48570989551321453,
      "grad_norm": 3.0159146785736084,
      "learning_rate": 1.8057160417947145e-05,
      "loss": 1.4002,
      "step": 6322
    },
    {
      "epoch": 0.48586355255070685,
      "grad_norm": 3.2475900650024414,
      "learning_rate": 1.8056545789797175e-05,
      "loss": 1.3624,
      "step": 6324
    },
    {
      "epoch": 0.48601720958819916,
      "grad_norm": 3.2522149085998535,
      "learning_rate": 1.8055931161647204e-05,
      "loss": 1.3565,
      "step": 6326
    },
    {
      "epoch": 0.4861708666256915,
      "grad_norm": 3.1061174869537354,
      "learning_rate": 1.8055316533497237e-05,
      "loss": 1.3549,
      "step": 6328
    },
    {
      "epoch": 0.4863245236631838,
      "grad_norm": 3.162954330444336,
      "learning_rate": 1.8054701905347267e-05,
      "loss": 1.4947,
      "step": 6330
    },
    {
      "epoch": 0.4864781807006761,
      "grad_norm": 3.179232597351074,
      "learning_rate": 1.8054087277197297e-05,
      "loss": 1.32,
      "step": 6332
    },
    {
      "epoch": 0.4866318377381684,
      "grad_norm": 2.8867878913879395,
      "learning_rate": 1.8053472649047327e-05,
      "loss": 1.2925,
      "step": 6334
    },
    {
      "epoch": 0.48678549477566074,
      "grad_norm": 3.4606575965881348,
      "learning_rate": 1.805285802089736e-05,
      "loss": 1.4834,
      "step": 6336
    },
    {
      "epoch": 0.48693915181315306,
      "grad_norm": 3.118943214416504,
      "learning_rate": 1.805224339274739e-05,
      "loss": 1.227,
      "step": 6338
    },
    {
      "epoch": 0.4870928088506454,
      "grad_norm": 3.1794466972351074,
      "learning_rate": 1.805162876459742e-05,
      "loss": 1.4092,
      "step": 6340
    },
    {
      "epoch": 0.4872464658881377,
      "grad_norm": 2.719515562057495,
      "learning_rate": 1.8051014136447452e-05,
      "loss": 1.3534,
      "step": 6342
    },
    {
      "epoch": 0.48740012292563,
      "grad_norm": 3.8740384578704834,
      "learning_rate": 1.8050399508297482e-05,
      "loss": 1.4607,
      "step": 6344
    },
    {
      "epoch": 0.4875537799631223,
      "grad_norm": 3.121920585632324,
      "learning_rate": 1.804978488014751e-05,
      "loss": 1.365,
      "step": 6346
    },
    {
      "epoch": 0.48770743700061464,
      "grad_norm": 2.8695993423461914,
      "learning_rate": 1.8049170251997544e-05,
      "loss": 1.2682,
      "step": 6348
    },
    {
      "epoch": 0.48786109403810696,
      "grad_norm": 2.9940059185028076,
      "learning_rate": 1.8048555623847574e-05,
      "loss": 1.3376,
      "step": 6350
    },
    {
      "epoch": 0.4880147510755993,
      "grad_norm": 3.0862960815429688,
      "learning_rate": 1.8047940995697604e-05,
      "loss": 1.3186,
      "step": 6352
    },
    {
      "epoch": 0.4881684081130916,
      "grad_norm": 3.2420318126678467,
      "learning_rate": 1.8047326367547634e-05,
      "loss": 1.3098,
      "step": 6354
    },
    {
      "epoch": 0.4883220651505839,
      "grad_norm": 2.9130239486694336,
      "learning_rate": 1.8046711739397667e-05,
      "loss": 1.3989,
      "step": 6356
    },
    {
      "epoch": 0.4884757221880762,
      "grad_norm": 3.1051688194274902,
      "learning_rate": 1.8046097111247696e-05,
      "loss": 1.3173,
      "step": 6358
    },
    {
      "epoch": 0.48862937922556854,
      "grad_norm": 2.9414143562316895,
      "learning_rate": 1.8045482483097726e-05,
      "loss": 1.3141,
      "step": 6360
    },
    {
      "epoch": 0.48878303626306085,
      "grad_norm": 2.926856279373169,
      "learning_rate": 1.804486785494776e-05,
      "loss": 1.3659,
      "step": 6362
    },
    {
      "epoch": 0.48893669330055317,
      "grad_norm": 3.0181140899658203,
      "learning_rate": 1.804425322679779e-05,
      "loss": 1.4083,
      "step": 6364
    },
    {
      "epoch": 0.4890903503380455,
      "grad_norm": 2.822953462600708,
      "learning_rate": 1.804363859864782e-05,
      "loss": 1.3084,
      "step": 6366
    },
    {
      "epoch": 0.4892440073755378,
      "grad_norm": 3.2265994548797607,
      "learning_rate": 1.804302397049785e-05,
      "loss": 1.4421,
      "step": 6368
    },
    {
      "epoch": 0.4893976644130301,
      "grad_norm": 2.932751417160034,
      "learning_rate": 1.804240934234788e-05,
      "loss": 1.3666,
      "step": 6370
    },
    {
      "epoch": 0.48955132145052244,
      "grad_norm": 3.190852403640747,
      "learning_rate": 1.804179471419791e-05,
      "loss": 1.4584,
      "step": 6372
    },
    {
      "epoch": 0.48970497848801475,
      "grad_norm": 2.9671835899353027,
      "learning_rate": 1.8041180086047944e-05,
      "loss": 1.4951,
      "step": 6374
    },
    {
      "epoch": 0.48985863552550707,
      "grad_norm": 3.1517322063446045,
      "learning_rate": 1.8040565457897974e-05,
      "loss": 1.4506,
      "step": 6376
    },
    {
      "epoch": 0.4900122925629994,
      "grad_norm": 3.0485892295837402,
      "learning_rate": 1.8039950829748003e-05,
      "loss": 1.3996,
      "step": 6378
    },
    {
      "epoch": 0.4901659496004917,
      "grad_norm": 2.753948450088501,
      "learning_rate": 1.8039336201598033e-05,
      "loss": 1.4296,
      "step": 6380
    },
    {
      "epoch": 0.490319606637984,
      "grad_norm": 2.8919365406036377,
      "learning_rate": 1.8038721573448066e-05,
      "loss": 1.3162,
      "step": 6382
    },
    {
      "epoch": 0.49047326367547633,
      "grad_norm": 2.782630205154419,
      "learning_rate": 1.8038106945298096e-05,
      "loss": 1.3588,
      "step": 6384
    },
    {
      "epoch": 0.49062692071296865,
      "grad_norm": 2.8316001892089844,
      "learning_rate": 1.8037492317148125e-05,
      "loss": 1.4375,
      "step": 6386
    },
    {
      "epoch": 0.49078057775046097,
      "grad_norm": 2.9996912479400635,
      "learning_rate": 1.803687768899816e-05,
      "loss": 1.3141,
      "step": 6388
    },
    {
      "epoch": 0.4909342347879533,
      "grad_norm": 2.8670809268951416,
      "learning_rate": 1.8036263060848188e-05,
      "loss": 1.424,
      "step": 6390
    },
    {
      "epoch": 0.4910878918254456,
      "grad_norm": 3.0059220790863037,
      "learning_rate": 1.8035648432698218e-05,
      "loss": 1.3933,
      "step": 6392
    },
    {
      "epoch": 0.4912415488629379,
      "grad_norm": 3.1974833011627197,
      "learning_rate": 1.803503380454825e-05,
      "loss": 1.3125,
      "step": 6394
    },
    {
      "epoch": 0.49139520590043023,
      "grad_norm": 2.683246612548828,
      "learning_rate": 1.803441917639828e-05,
      "loss": 1.258,
      "step": 6396
    },
    {
      "epoch": 0.49154886293792255,
      "grad_norm": 3.330538034439087,
      "learning_rate": 1.803380454824831e-05,
      "loss": 1.4212,
      "step": 6398
    },
    {
      "epoch": 0.49170251997541486,
      "grad_norm": 3.116828680038452,
      "learning_rate": 1.8033189920098343e-05,
      "loss": 1.3991,
      "step": 6400
    },
    {
      "epoch": 0.4918561770129072,
      "grad_norm": 3.45032000541687,
      "learning_rate": 1.8032575291948373e-05,
      "loss": 1.3591,
      "step": 6402
    },
    {
      "epoch": 0.4920098340503995,
      "grad_norm": 3.455242872238159,
      "learning_rate": 1.8031960663798403e-05,
      "loss": 1.3312,
      "step": 6404
    },
    {
      "epoch": 0.4921634910878918,
      "grad_norm": 2.821232557296753,
      "learning_rate": 1.8031346035648432e-05,
      "loss": 1.2594,
      "step": 6406
    },
    {
      "epoch": 0.49231714812538413,
      "grad_norm": 3.163733720779419,
      "learning_rate": 1.8030731407498465e-05,
      "loss": 1.3575,
      "step": 6408
    },
    {
      "epoch": 0.49247080516287645,
      "grad_norm": 3.2715537548065186,
      "learning_rate": 1.8030116779348495e-05,
      "loss": 1.4909,
      "step": 6410
    },
    {
      "epoch": 0.49262446220036876,
      "grad_norm": 3.1828911304473877,
      "learning_rate": 1.8029502151198525e-05,
      "loss": 1.3965,
      "step": 6412
    },
    {
      "epoch": 0.4927781192378611,
      "grad_norm": 3.2268998622894287,
      "learning_rate": 1.8028887523048558e-05,
      "loss": 1.3785,
      "step": 6414
    },
    {
      "epoch": 0.4929317762753534,
      "grad_norm": 3.106019973754883,
      "learning_rate": 1.8028272894898588e-05,
      "loss": 1.2841,
      "step": 6416
    },
    {
      "epoch": 0.4930854333128457,
      "grad_norm": 3.220978260040283,
      "learning_rate": 1.8027658266748617e-05,
      "loss": 1.5033,
      "step": 6418
    },
    {
      "epoch": 0.493239090350338,
      "grad_norm": 3.288722276687622,
      "learning_rate": 1.802704363859865e-05,
      "loss": 1.3628,
      "step": 6420
    },
    {
      "epoch": 0.49339274738783034,
      "grad_norm": 3.134910821914673,
      "learning_rate": 1.802642901044868e-05,
      "loss": 1.3892,
      "step": 6422
    },
    {
      "epoch": 0.49354640442532266,
      "grad_norm": 2.868943929672241,
      "learning_rate": 1.802581438229871e-05,
      "loss": 1.4807,
      "step": 6424
    },
    {
      "epoch": 0.493700061462815,
      "grad_norm": 3.257479190826416,
      "learning_rate": 1.8025199754148743e-05,
      "loss": 1.3037,
      "step": 6426
    },
    {
      "epoch": 0.4938537185003073,
      "grad_norm": 2.6661643981933594,
      "learning_rate": 1.8024585125998772e-05,
      "loss": 1.3603,
      "step": 6428
    },
    {
      "epoch": 0.4940073755377996,
      "grad_norm": 3.075345039367676,
      "learning_rate": 1.8023970497848806e-05,
      "loss": 1.3357,
      "step": 6430
    },
    {
      "epoch": 0.4941610325752919,
      "grad_norm": 3.3657965660095215,
      "learning_rate": 1.8023355869698832e-05,
      "loss": 1.3358,
      "step": 6432
    },
    {
      "epoch": 0.49431468961278424,
      "grad_norm": 3.102015495300293,
      "learning_rate": 1.8022741241548865e-05,
      "loss": 1.4151,
      "step": 6434
    },
    {
      "epoch": 0.49446834665027656,
      "grad_norm": 2.722320795059204,
      "learning_rate": 1.8022126613398895e-05,
      "loss": 1.3975,
      "step": 6436
    },
    {
      "epoch": 0.4946220036877689,
      "grad_norm": 3.373051166534424,
      "learning_rate": 1.8021511985248924e-05,
      "loss": 1.2497,
      "step": 6438
    },
    {
      "epoch": 0.4947756607252612,
      "grad_norm": 2.7113919258117676,
      "learning_rate": 1.8020897357098957e-05,
      "loss": 1.3809,
      "step": 6440
    },
    {
      "epoch": 0.4949293177627535,
      "grad_norm": 3.1061747074127197,
      "learning_rate": 1.8020282728948987e-05,
      "loss": 1.4414,
      "step": 6442
    },
    {
      "epoch": 0.4950829748002459,
      "grad_norm": 3.091012477874756,
      "learning_rate": 1.8019668100799017e-05,
      "loss": 1.3563,
      "step": 6444
    },
    {
      "epoch": 0.4952366318377382,
      "grad_norm": 2.845564126968384,
      "learning_rate": 1.801905347264905e-05,
      "loss": 1.3844,
      "step": 6446
    },
    {
      "epoch": 0.4953902888752305,
      "grad_norm": 3.5024054050445557,
      "learning_rate": 1.801843884449908e-05,
      "loss": 1.4408,
      "step": 6448
    },
    {
      "epoch": 0.4955439459127228,
      "grad_norm": 2.915093421936035,
      "learning_rate": 1.8017824216349113e-05,
      "loss": 1.2814,
      "step": 6450
    },
    {
      "epoch": 0.49569760295021514,
      "grad_norm": 3.042811632156372,
      "learning_rate": 1.801720958819914e-05,
      "loss": 1.5103,
      "step": 6452
    },
    {
      "epoch": 0.49585125998770746,
      "grad_norm": 2.851787805557251,
      "learning_rate": 1.8016594960049172e-05,
      "loss": 1.4586,
      "step": 6454
    },
    {
      "epoch": 0.4960049170251998,
      "grad_norm": 3.1319057941436768,
      "learning_rate": 1.80159803318992e-05,
      "loss": 1.2987,
      "step": 6456
    },
    {
      "epoch": 0.4961585740626921,
      "grad_norm": 2.650089740753174,
      "learning_rate": 1.801536570374923e-05,
      "loss": 1.3441,
      "step": 6458
    },
    {
      "epoch": 0.4963122311001844,
      "grad_norm": 3.493260383605957,
      "learning_rate": 1.8014751075599264e-05,
      "loss": 1.6013,
      "step": 6460
    },
    {
      "epoch": 0.4964658881376767,
      "grad_norm": 2.9392683506011963,
      "learning_rate": 1.8014136447449294e-05,
      "loss": 1.3622,
      "step": 6462
    },
    {
      "epoch": 0.49661954517516904,
      "grad_norm": 3.106326103210449,
      "learning_rate": 1.8013521819299324e-05,
      "loss": 1.3347,
      "step": 6464
    },
    {
      "epoch": 0.49677320221266136,
      "grad_norm": 2.914437770843506,
      "learning_rate": 1.8012907191149357e-05,
      "loss": 1.2599,
      "step": 6466
    },
    {
      "epoch": 0.4969268592501537,
      "grad_norm": 2.910841464996338,
      "learning_rate": 1.8012292562999386e-05,
      "loss": 1.3622,
      "step": 6468
    },
    {
      "epoch": 0.497080516287646,
      "grad_norm": 2.89884352684021,
      "learning_rate": 1.801167793484942e-05,
      "loss": 1.3582,
      "step": 6470
    },
    {
      "epoch": 0.4972341733251383,
      "grad_norm": 2.8821210861206055,
      "learning_rate": 1.801106330669945e-05,
      "loss": 1.4164,
      "step": 6472
    },
    {
      "epoch": 0.4973878303626306,
      "grad_norm": 2.601987838745117,
      "learning_rate": 1.801044867854948e-05,
      "loss": 1.5184,
      "step": 6474
    },
    {
      "epoch": 0.49754148740012294,
      "grad_norm": 2.958704710006714,
      "learning_rate": 1.8009834050399512e-05,
      "loss": 1.3903,
      "step": 6476
    },
    {
      "epoch": 0.49769514443761526,
      "grad_norm": 2.9979889392852783,
      "learning_rate": 1.8009219422249538e-05,
      "loss": 1.2592,
      "step": 6478
    },
    {
      "epoch": 0.49784880147510757,
      "grad_norm": 2.916813611984253,
      "learning_rate": 1.800860479409957e-05,
      "loss": 1.3302,
      "step": 6480
    },
    {
      "epoch": 0.4980024585125999,
      "grad_norm": 3.3828582763671875,
      "learning_rate": 1.80079901659496e-05,
      "loss": 1.4222,
      "step": 6482
    },
    {
      "epoch": 0.4981561155500922,
      "grad_norm": 3.1152589321136475,
      "learning_rate": 1.800737553779963e-05,
      "loss": 1.3731,
      "step": 6484
    },
    {
      "epoch": 0.4983097725875845,
      "grad_norm": 2.974968671798706,
      "learning_rate": 1.8006760909649664e-05,
      "loss": 1.2744,
      "step": 6486
    },
    {
      "epoch": 0.49846342962507684,
      "grad_norm": 3.652846336364746,
      "learning_rate": 1.8006146281499693e-05,
      "loss": 1.4291,
      "step": 6488
    },
    {
      "epoch": 0.49861708666256915,
      "grad_norm": 2.827791213989258,
      "learning_rate": 1.8005531653349727e-05,
      "loss": 1.2982,
      "step": 6490
    },
    {
      "epoch": 0.49877074370006147,
      "grad_norm": 3.1473135948181152,
      "learning_rate": 1.8004917025199756e-05,
      "loss": 1.3051,
      "step": 6492
    },
    {
      "epoch": 0.4989244007375538,
      "grad_norm": 3.156839370727539,
      "learning_rate": 1.8004302397049786e-05,
      "loss": 1.4366,
      "step": 6494
    },
    {
      "epoch": 0.4990780577750461,
      "grad_norm": 2.8511626720428467,
      "learning_rate": 1.800368776889982e-05,
      "loss": 1.3214,
      "step": 6496
    },
    {
      "epoch": 0.4992317148125384,
      "grad_norm": 2.7867062091827393,
      "learning_rate": 1.800307314074985e-05,
      "loss": 1.3791,
      "step": 6498
    },
    {
      "epoch": 0.49938537185003073,
      "grad_norm": 3.0567257404327393,
      "learning_rate": 1.800245851259988e-05,
      "loss": 1.3549,
      "step": 6500
    },
    {
      "epoch": 0.49953902888752305,
      "grad_norm": 2.9443788528442383,
      "learning_rate": 1.800184388444991e-05,
      "loss": 1.3798,
      "step": 6502
    },
    {
      "epoch": 0.49969268592501537,
      "grad_norm": 2.903205156326294,
      "learning_rate": 1.8001229256299938e-05,
      "loss": 1.3004,
      "step": 6504
    },
    {
      "epoch": 0.4998463429625077,
      "grad_norm": 3.2189865112304688,
      "learning_rate": 1.800061462814997e-05,
      "loss": 1.3023,
      "step": 6506
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.1087801456451416,
      "learning_rate": 1.8e-05,
      "loss": 1.508,
      "step": 6508
    },
    {
      "epoch": 0.5001536570374924,
      "grad_norm": 3.1676154136657715,
      "learning_rate": 1.799938537185003e-05,
      "loss": 1.4198,
      "step": 6510
    },
    {
      "epoch": 0.5003073140749846,
      "grad_norm": 2.717998504638672,
      "learning_rate": 1.7998770743700063e-05,
      "loss": 1.3673,
      "step": 6512
    },
    {
      "epoch": 0.500460971112477,
      "grad_norm": 2.779644012451172,
      "learning_rate": 1.7998156115550093e-05,
      "loss": 1.493,
      "step": 6514
    },
    {
      "epoch": 0.5006146281499693,
      "grad_norm": 2.706282377243042,
      "learning_rate": 1.7997541487400126e-05,
      "loss": 1.2036,
      "step": 6516
    },
    {
      "epoch": 0.5007682851874616,
      "grad_norm": 2.846618175506592,
      "learning_rate": 1.7996926859250156e-05,
      "loss": 1.431,
      "step": 6518
    },
    {
      "epoch": 0.5009219422249539,
      "grad_norm": 2.9443068504333496,
      "learning_rate": 1.7996312231100185e-05,
      "loss": 1.2797,
      "step": 6520
    },
    {
      "epoch": 0.5010755992624463,
      "grad_norm": 3.471630334854126,
      "learning_rate": 1.799569760295022e-05,
      "loss": 1.436,
      "step": 6522
    },
    {
      "epoch": 0.5012292562999385,
      "grad_norm": 3.2783944606781006,
      "learning_rate": 1.7995082974800248e-05,
      "loss": 1.4904,
      "step": 6524
    },
    {
      "epoch": 0.5013829133374309,
      "grad_norm": 3.51324725151062,
      "learning_rate": 1.7994468346650278e-05,
      "loss": 1.3418,
      "step": 6526
    },
    {
      "epoch": 0.5015365703749232,
      "grad_norm": 2.648725748062134,
      "learning_rate": 1.799385371850031e-05,
      "loss": 1.2583,
      "step": 6528
    },
    {
      "epoch": 0.5016902274124155,
      "grad_norm": 2.8676857948303223,
      "learning_rate": 1.7993239090350337e-05,
      "loss": 1.3348,
      "step": 6530
    },
    {
      "epoch": 0.5018438844499078,
      "grad_norm": 3.1224286556243896,
      "learning_rate": 1.799262446220037e-05,
      "loss": 1.3221,
      "step": 6532
    },
    {
      "epoch": 0.5019975414874002,
      "grad_norm": 3.003473997116089,
      "learning_rate": 1.79920098340504e-05,
      "loss": 1.3466,
      "step": 6534
    },
    {
      "epoch": 0.5021511985248924,
      "grad_norm": 2.8612453937530518,
      "learning_rate": 1.7991395205900433e-05,
      "loss": 1.2505,
      "step": 6536
    },
    {
      "epoch": 0.5023048555623848,
      "grad_norm": 2.9750072956085205,
      "learning_rate": 1.7990780577750463e-05,
      "loss": 1.3482,
      "step": 6538
    },
    {
      "epoch": 0.5024585125998771,
      "grad_norm": 2.6988656520843506,
      "learning_rate": 1.7990165949600492e-05,
      "loss": 1.2996,
      "step": 6540
    },
    {
      "epoch": 0.5026121696373694,
      "grad_norm": 3.0278279781341553,
      "learning_rate": 1.7989551321450525e-05,
      "loss": 1.3919,
      "step": 6542
    },
    {
      "epoch": 0.5027658266748617,
      "grad_norm": 2.9088683128356934,
      "learning_rate": 1.7988936693300555e-05,
      "loss": 1.4706,
      "step": 6544
    },
    {
      "epoch": 0.5029194837123541,
      "grad_norm": 2.8318252563476562,
      "learning_rate": 1.7988322065150585e-05,
      "loss": 1.4209,
      "step": 6546
    },
    {
      "epoch": 0.5030731407498463,
      "grad_norm": 2.994474411010742,
      "learning_rate": 1.7987707437000618e-05,
      "loss": 1.3159,
      "step": 6548
    },
    {
      "epoch": 0.5032267977873387,
      "grad_norm": 2.9801907539367676,
      "learning_rate": 1.7987092808850644e-05,
      "loss": 1.2645,
      "step": 6550
    },
    {
      "epoch": 0.503380454824831,
      "grad_norm": 3.4261326789855957,
      "learning_rate": 1.7986478180700677e-05,
      "loss": 1.4416,
      "step": 6552
    },
    {
      "epoch": 0.5035341118623233,
      "grad_norm": 2.65868878364563,
      "learning_rate": 1.798586355255071e-05,
      "loss": 1.2989,
      "step": 6554
    },
    {
      "epoch": 0.5036877688998156,
      "grad_norm": 3.106370210647583,
      "learning_rate": 1.798524892440074e-05,
      "loss": 1.3748,
      "step": 6556
    },
    {
      "epoch": 0.503841425937308,
      "grad_norm": 2.6410956382751465,
      "learning_rate": 1.798463429625077e-05,
      "loss": 1.3152,
      "step": 6558
    },
    {
      "epoch": 0.5039950829748002,
      "grad_norm": 2.8190648555755615,
      "learning_rate": 1.79840196681008e-05,
      "loss": 1.4078,
      "step": 6560
    },
    {
      "epoch": 0.5041487400122926,
      "grad_norm": 2.9152209758758545,
      "learning_rate": 1.7983405039950832e-05,
      "loss": 1.2513,
      "step": 6562
    },
    {
      "epoch": 0.5043023970497849,
      "grad_norm": 3.070107936859131,
      "learning_rate": 1.7982790411800862e-05,
      "loss": 1.4488,
      "step": 6564
    },
    {
      "epoch": 0.5044560540872772,
      "grad_norm": 2.9198601245880127,
      "learning_rate": 1.7982175783650892e-05,
      "loss": 1.2694,
      "step": 6566
    },
    {
      "epoch": 0.5046097111247695,
      "grad_norm": 3.084261417388916,
      "learning_rate": 1.7981561155500925e-05,
      "loss": 1.3638,
      "step": 6568
    },
    {
      "epoch": 0.5047633681622619,
      "grad_norm": 3.0732178688049316,
      "learning_rate": 1.7980946527350955e-05,
      "loss": 1.2477,
      "step": 6570
    },
    {
      "epoch": 0.5049170251997541,
      "grad_norm": 3.3027431964874268,
      "learning_rate": 1.7980331899200984e-05,
      "loss": 1.335,
      "step": 6572
    },
    {
      "epoch": 0.5050706822372465,
      "grad_norm": 3.3048105239868164,
      "learning_rate": 1.7979717271051017e-05,
      "loss": 1.4374,
      "step": 6574
    },
    {
      "epoch": 0.5052243392747388,
      "grad_norm": 3.292642831802368,
      "learning_rate": 1.7979102642901047e-05,
      "loss": 1.353,
      "step": 6576
    },
    {
      "epoch": 0.5053779963122311,
      "grad_norm": 3.257822036743164,
      "learning_rate": 1.7978488014751077e-05,
      "loss": 1.3081,
      "step": 6578
    },
    {
      "epoch": 0.5055316533497234,
      "grad_norm": 2.796616792678833,
      "learning_rate": 1.7977873386601106e-05,
      "loss": 1.3497,
      "step": 6580
    },
    {
      "epoch": 0.5056853103872158,
      "grad_norm": 3.194371223449707,
      "learning_rate": 1.797725875845114e-05,
      "loss": 1.3587,
      "step": 6582
    },
    {
      "epoch": 0.505838967424708,
      "grad_norm": 2.6798551082611084,
      "learning_rate": 1.797664413030117e-05,
      "loss": 1.3413,
      "step": 6584
    },
    {
      "epoch": 0.5059926244622004,
      "grad_norm": 3.123992681503296,
      "learning_rate": 1.79760295021512e-05,
      "loss": 1.2649,
      "step": 6586
    },
    {
      "epoch": 0.5061462814996927,
      "grad_norm": 3.01867413520813,
      "learning_rate": 1.7975414874001232e-05,
      "loss": 1.3729,
      "step": 6588
    },
    {
      "epoch": 0.506299938537185,
      "grad_norm": 3.5075738430023193,
      "learning_rate": 1.797480024585126e-05,
      "loss": 1.3458,
      "step": 6590
    },
    {
      "epoch": 0.5064535955746773,
      "grad_norm": 3.0404345989227295,
      "learning_rate": 1.797418561770129e-05,
      "loss": 1.4686,
      "step": 6592
    },
    {
      "epoch": 0.5066072526121697,
      "grad_norm": 2.7560205459594727,
      "learning_rate": 1.7973570989551324e-05,
      "loss": 1.3492,
      "step": 6594
    },
    {
      "epoch": 0.5067609096496619,
      "grad_norm": 3.127852439880371,
      "learning_rate": 1.7972956361401354e-05,
      "loss": 1.3995,
      "step": 6596
    },
    {
      "epoch": 0.5069145666871543,
      "grad_norm": 2.963434934616089,
      "learning_rate": 1.7972341733251384e-05,
      "loss": 1.3773,
      "step": 6598
    },
    {
      "epoch": 0.5070682237246465,
      "grad_norm": 3.146491050720215,
      "learning_rate": 1.7971727105101417e-05,
      "loss": 1.282,
      "step": 6600
    },
    {
      "epoch": 0.5072218807621389,
      "grad_norm": 2.8593227863311768,
      "learning_rate": 1.7971112476951446e-05,
      "loss": 1.2804,
      "step": 6602
    },
    {
      "epoch": 0.5073755377996312,
      "grad_norm": 3.011699914932251,
      "learning_rate": 1.7970497848801476e-05,
      "loss": 1.3739,
      "step": 6604
    },
    {
      "epoch": 0.5075291948371236,
      "grad_norm": 3.7797420024871826,
      "learning_rate": 1.7969883220651506e-05,
      "loss": 1.4324,
      "step": 6606
    },
    {
      "epoch": 0.5076828518746158,
      "grad_norm": 2.8261046409606934,
      "learning_rate": 1.796926859250154e-05,
      "loss": 1.3562,
      "step": 6608
    },
    {
      "epoch": 0.5078365089121082,
      "grad_norm": 2.9160585403442383,
      "learning_rate": 1.796865396435157e-05,
      "loss": 1.5061,
      "step": 6610
    },
    {
      "epoch": 0.5079901659496004,
      "grad_norm": 3.1277146339416504,
      "learning_rate": 1.7968039336201598e-05,
      "loss": 1.3759,
      "step": 6612
    },
    {
      "epoch": 0.5081438229870928,
      "grad_norm": 3.4111275672912598,
      "learning_rate": 1.796742470805163e-05,
      "loss": 1.3682,
      "step": 6614
    },
    {
      "epoch": 0.5082974800245851,
      "grad_norm": 3.055345296859741,
      "learning_rate": 1.796681007990166e-05,
      "loss": 1.4154,
      "step": 6616
    },
    {
      "epoch": 0.5084511370620775,
      "grad_norm": 3.3309273719787598,
      "learning_rate": 1.796619545175169e-05,
      "loss": 1.3364,
      "step": 6618
    },
    {
      "epoch": 0.5086047940995697,
      "grad_norm": 2.7727885246276855,
      "learning_rate": 1.7965580823601724e-05,
      "loss": 1.2019,
      "step": 6620
    },
    {
      "epoch": 0.5087584511370621,
      "grad_norm": 3.5600733757019043,
      "learning_rate": 1.7964966195451753e-05,
      "loss": 1.4411,
      "step": 6622
    },
    {
      "epoch": 0.5089121081745543,
      "grad_norm": 3.1073739528656006,
      "learning_rate": 1.7964351567301783e-05,
      "loss": 1.4255,
      "step": 6624
    },
    {
      "epoch": 0.5090657652120467,
      "grad_norm": 2.9542529582977295,
      "learning_rate": 1.7963736939151816e-05,
      "loss": 1.2744,
      "step": 6626
    },
    {
      "epoch": 0.509219422249539,
      "grad_norm": 3.2199599742889404,
      "learning_rate": 1.7963122311001846e-05,
      "loss": 1.289,
      "step": 6628
    },
    {
      "epoch": 0.5093730792870313,
      "grad_norm": 3.095400333404541,
      "learning_rate": 1.7962507682851876e-05,
      "loss": 1.3363,
      "step": 6630
    },
    {
      "epoch": 0.5095267363245236,
      "grad_norm": 3.337625503540039,
      "learning_rate": 1.7961893054701905e-05,
      "loss": 1.4096,
      "step": 6632
    },
    {
      "epoch": 0.509680393362016,
      "grad_norm": 3.3786509037017822,
      "learning_rate": 1.7961278426551938e-05,
      "loss": 1.3938,
      "step": 6634
    },
    {
      "epoch": 0.5098340503995082,
      "grad_norm": 2.652902364730835,
      "learning_rate": 1.7960663798401968e-05,
      "loss": 1.2738,
      "step": 6636
    },
    {
      "epoch": 0.5099877074370006,
      "grad_norm": 3.4588985443115234,
      "learning_rate": 1.7960049170251998e-05,
      "loss": 1.2758,
      "step": 6638
    },
    {
      "epoch": 0.510141364474493,
      "grad_norm": 3.1711127758026123,
      "learning_rate": 1.795943454210203e-05,
      "loss": 1.3747,
      "step": 6640
    },
    {
      "epoch": 0.5102950215119852,
      "grad_norm": 3.4093017578125,
      "learning_rate": 1.795881991395206e-05,
      "loss": 1.2861,
      "step": 6642
    },
    {
      "epoch": 0.5104486785494776,
      "grad_norm": 3.0328543186187744,
      "learning_rate": 1.795820528580209e-05,
      "loss": 1.5437,
      "step": 6644
    },
    {
      "epoch": 0.5106023355869699,
      "grad_norm": 2.8662827014923096,
      "learning_rate": 1.7957590657652123e-05,
      "loss": 1.3798,
      "step": 6646
    },
    {
      "epoch": 0.5107559926244623,
      "grad_norm": 2.6307129859924316,
      "learning_rate": 1.7956976029502153e-05,
      "loss": 1.1287,
      "step": 6648
    },
    {
      "epoch": 0.5109096496619545,
      "grad_norm": 3.1315627098083496,
      "learning_rate": 1.7956361401352183e-05,
      "loss": 1.4458,
      "step": 6650
    },
    {
      "epoch": 0.5110633066994469,
      "grad_norm": 3.4800686836242676,
      "learning_rate": 1.7955746773202216e-05,
      "loss": 1.5149,
      "step": 6652
    },
    {
      "epoch": 0.5112169637369391,
      "grad_norm": 3.0013492107391357,
      "learning_rate": 1.7955132145052245e-05,
      "loss": 1.3806,
      "step": 6654
    },
    {
      "epoch": 0.5113706207744315,
      "grad_norm": 3.450124502182007,
      "learning_rate": 1.7954517516902275e-05,
      "loss": 1.4322,
      "step": 6656
    },
    {
      "epoch": 0.5115242778119238,
      "grad_norm": 2.9068808555603027,
      "learning_rate": 1.7953902888752305e-05,
      "loss": 1.3998,
      "step": 6658
    },
    {
      "epoch": 0.5116779348494161,
      "grad_norm": 2.6469430923461914,
      "learning_rate": 1.7953288260602338e-05,
      "loss": 1.4466,
      "step": 6660
    },
    {
      "epoch": 0.5118315918869084,
      "grad_norm": 2.9947774410247803,
      "learning_rate": 1.7952673632452367e-05,
      "loss": 1.3114,
      "step": 6662
    },
    {
      "epoch": 0.5119852489244008,
      "grad_norm": 3.5101516246795654,
      "learning_rate": 1.7952059004302397e-05,
      "loss": 1.3693,
      "step": 6664
    },
    {
      "epoch": 0.512138905961893,
      "grad_norm": 3.2445108890533447,
      "learning_rate": 1.795144437615243e-05,
      "loss": 1.3274,
      "step": 6666
    },
    {
      "epoch": 0.5122925629993854,
      "grad_norm": 3.050546169281006,
      "learning_rate": 1.795082974800246e-05,
      "loss": 1.3715,
      "step": 6668
    },
    {
      "epoch": 0.5124462200368777,
      "grad_norm": 3.0119810104370117,
      "learning_rate": 1.795021511985249e-05,
      "loss": 1.3978,
      "step": 6670
    },
    {
      "epoch": 0.51259987707437,
      "grad_norm": 3.514958381652832,
      "learning_rate": 1.7949600491702523e-05,
      "loss": 1.4204,
      "step": 6672
    },
    {
      "epoch": 0.5127535341118623,
      "grad_norm": 2.9083685874938965,
      "learning_rate": 1.7948985863552552e-05,
      "loss": 1.2959,
      "step": 6674
    },
    {
      "epoch": 0.5129071911493547,
      "grad_norm": 2.6381547451019287,
      "learning_rate": 1.7948371235402582e-05,
      "loss": 1.3359,
      "step": 6676
    },
    {
      "epoch": 0.5130608481868469,
      "grad_norm": 3.478752374649048,
      "learning_rate": 1.794775660725261e-05,
      "loss": 1.4583,
      "step": 6678
    },
    {
      "epoch": 0.5132145052243393,
      "grad_norm": 2.834408760070801,
      "learning_rate": 1.7947141979102645e-05,
      "loss": 1.3429,
      "step": 6680
    },
    {
      "epoch": 0.5133681622618316,
      "grad_norm": 3.0341265201568604,
      "learning_rate": 1.7946527350952674e-05,
      "loss": 1.3716,
      "step": 6682
    },
    {
      "epoch": 0.513521819299324,
      "grad_norm": 2.8251383304595947,
      "learning_rate": 1.7945912722802704e-05,
      "loss": 1.5223,
      "step": 6684
    },
    {
      "epoch": 0.5136754763368162,
      "grad_norm": 3.265641212463379,
      "learning_rate": 1.7945298094652737e-05,
      "loss": 1.269,
      "step": 6686
    },
    {
      "epoch": 0.5138291333743086,
      "grad_norm": 3.247816562652588,
      "learning_rate": 1.7944683466502767e-05,
      "loss": 1.2121,
      "step": 6688
    },
    {
      "epoch": 0.5139827904118008,
      "grad_norm": 3.2081658840179443,
      "learning_rate": 1.7944068838352797e-05,
      "loss": 1.2807,
      "step": 6690
    },
    {
      "epoch": 0.5141364474492932,
      "grad_norm": 3.1236660480499268,
      "learning_rate": 1.794345421020283e-05,
      "loss": 1.5633,
      "step": 6692
    },
    {
      "epoch": 0.5142901044867855,
      "grad_norm": 2.951101064682007,
      "learning_rate": 1.794283958205286e-05,
      "loss": 1.3899,
      "step": 6694
    },
    {
      "epoch": 0.5144437615242778,
      "grad_norm": 3.1760647296905518,
      "learning_rate": 1.794222495390289e-05,
      "loss": 1.2689,
      "step": 6696
    },
    {
      "epoch": 0.5145974185617701,
      "grad_norm": 3.310821294784546,
      "learning_rate": 1.7941610325752922e-05,
      "loss": 1.4301,
      "step": 6698
    },
    {
      "epoch": 0.5147510755992625,
      "grad_norm": 3.1752665042877197,
      "learning_rate": 1.7940995697602952e-05,
      "loss": 1.2948,
      "step": 6700
    },
    {
      "epoch": 0.5149047326367547,
      "grad_norm": 3.2648708820343018,
      "learning_rate": 1.7940381069452985e-05,
      "loss": 1.3254,
      "step": 6702
    },
    {
      "epoch": 0.5150583896742471,
      "grad_norm": 3.0156726837158203,
      "learning_rate": 1.793976644130301e-05,
      "loss": 1.4047,
      "step": 6704
    },
    {
      "epoch": 0.5152120467117394,
      "grad_norm": 3.873260974884033,
      "learning_rate": 1.7939151813153044e-05,
      "loss": 1.4925,
      "step": 6706
    },
    {
      "epoch": 0.5153657037492317,
      "grad_norm": 3.176854372024536,
      "learning_rate": 1.7938537185003074e-05,
      "loss": 1.301,
      "step": 6708
    },
    {
      "epoch": 0.515519360786724,
      "grad_norm": 2.7950966358184814,
      "learning_rate": 1.7937922556853104e-05,
      "loss": 1.3525,
      "step": 6710
    },
    {
      "epoch": 0.5156730178242164,
      "grad_norm": 3.167912244796753,
      "learning_rate": 1.7937307928703137e-05,
      "loss": 1.5551,
      "step": 6712
    },
    {
      "epoch": 0.5158266748617086,
      "grad_norm": 3.4793012142181396,
      "learning_rate": 1.7936693300553166e-05,
      "loss": 1.5019,
      "step": 6714
    },
    {
      "epoch": 0.515980331899201,
      "grad_norm": 2.9153974056243896,
      "learning_rate": 1.7936078672403196e-05,
      "loss": 1.407,
      "step": 6716
    },
    {
      "epoch": 0.5161339889366933,
      "grad_norm": 2.8722434043884277,
      "learning_rate": 1.793546404425323e-05,
      "loss": 1.3882,
      "step": 6718
    },
    {
      "epoch": 0.5162876459741856,
      "grad_norm": 2.9194371700286865,
      "learning_rate": 1.793484941610326e-05,
      "loss": 1.4366,
      "step": 6720
    },
    {
      "epoch": 0.5164413030116779,
      "grad_norm": 3.2603633403778076,
      "learning_rate": 1.7934234787953292e-05,
      "loss": 1.3387,
      "step": 6722
    },
    {
      "epoch": 0.5165949600491703,
      "grad_norm": 3.0995850563049316,
      "learning_rate": 1.793362015980332e-05,
      "loss": 1.3719,
      "step": 6724
    },
    {
      "epoch": 0.5167486170866625,
      "grad_norm": 2.8780972957611084,
      "learning_rate": 1.793300553165335e-05,
      "loss": 1.3721,
      "step": 6726
    },
    {
      "epoch": 0.5169022741241549,
      "grad_norm": 3.1328043937683105,
      "learning_rate": 1.7932390903503384e-05,
      "loss": 1.2831,
      "step": 6728
    },
    {
      "epoch": 0.5170559311616472,
      "grad_norm": 3.0199432373046875,
      "learning_rate": 1.793177627535341e-05,
      "loss": 1.3505,
      "step": 6730
    },
    {
      "epoch": 0.5172095881991395,
      "grad_norm": 2.7379181385040283,
      "learning_rate": 1.7931161647203444e-05,
      "loss": 1.3088,
      "step": 6732
    },
    {
      "epoch": 0.5173632452366318,
      "grad_norm": 2.929868459701538,
      "learning_rate": 1.7930547019053473e-05,
      "loss": 1.3396,
      "step": 6734
    },
    {
      "epoch": 0.5175169022741242,
      "grad_norm": 2.787508249282837,
      "learning_rate": 1.7929932390903503e-05,
      "loss": 1.3644,
      "step": 6736
    },
    {
      "epoch": 0.5176705593116164,
      "grad_norm": 2.7956087589263916,
      "learning_rate": 1.7929317762753536e-05,
      "loss": 1.2359,
      "step": 6738
    },
    {
      "epoch": 0.5178242163491088,
      "grad_norm": 2.53226900100708,
      "learning_rate": 1.7928703134603566e-05,
      "loss": 1.307,
      "step": 6740
    },
    {
      "epoch": 0.5179778733866011,
      "grad_norm": 2.9932830333709717,
      "learning_rate": 1.79280885064536e-05,
      "loss": 1.3173,
      "step": 6742
    },
    {
      "epoch": 0.5181315304240934,
      "grad_norm": 2.900578260421753,
      "learning_rate": 1.792747387830363e-05,
      "loss": 1.4958,
      "step": 6744
    },
    {
      "epoch": 0.5182851874615857,
      "grad_norm": 3.190718173980713,
      "learning_rate": 1.7926859250153658e-05,
      "loss": 1.3313,
      "step": 6746
    },
    {
      "epoch": 0.5184388444990781,
      "grad_norm": 3.0104870796203613,
      "learning_rate": 1.792624462200369e-05,
      "loss": 1.3749,
      "step": 6748
    },
    {
      "epoch": 0.5185925015365703,
      "grad_norm": 3.335132598876953,
      "learning_rate": 1.792562999385372e-05,
      "loss": 1.3441,
      "step": 6750
    },
    {
      "epoch": 0.5187461585740627,
      "grad_norm": 2.9974780082702637,
      "learning_rate": 1.792501536570375e-05,
      "loss": 1.4476,
      "step": 6752
    },
    {
      "epoch": 0.518899815611555,
      "grad_norm": 2.8957419395446777,
      "learning_rate": 1.7924400737553784e-05,
      "loss": 1.3322,
      "step": 6754
    },
    {
      "epoch": 0.5190534726490473,
      "grad_norm": 3.018171787261963,
      "learning_rate": 1.792378610940381e-05,
      "loss": 1.3581,
      "step": 6756
    },
    {
      "epoch": 0.5192071296865396,
      "grad_norm": 2.8184151649475098,
      "learning_rate": 1.7923171481253843e-05,
      "loss": 1.4037,
      "step": 6758
    },
    {
      "epoch": 0.519360786724032,
      "grad_norm": 2.97739315032959,
      "learning_rate": 1.7922556853103873e-05,
      "loss": 1.2547,
      "step": 6760
    },
    {
      "epoch": 0.5195144437615242,
      "grad_norm": 3.4361817836761475,
      "learning_rate": 1.7921942224953906e-05,
      "loss": 1.3367,
      "step": 6762
    },
    {
      "epoch": 0.5196681007990166,
      "grad_norm": 25.288660049438477,
      "learning_rate": 1.7921327596803935e-05,
      "loss": 1.4986,
      "step": 6764
    },
    {
      "epoch": 0.5198217578365089,
      "grad_norm": 3.100538492202759,
      "learning_rate": 1.7920712968653965e-05,
      "loss": 1.3149,
      "step": 6766
    },
    {
      "epoch": 0.5199754148740012,
      "grad_norm": 2.8900961875915527,
      "learning_rate": 1.7920098340503998e-05,
      "loss": 1.352,
      "step": 6768
    },
    {
      "epoch": 0.5201290719114936,
      "grad_norm": 2.9886999130249023,
      "learning_rate": 1.7919483712354028e-05,
      "loss": 1.2732,
      "step": 6770
    },
    {
      "epoch": 0.5202827289489859,
      "grad_norm": 2.7704336643218994,
      "learning_rate": 1.7918869084204058e-05,
      "loss": 1.1976,
      "step": 6772
    },
    {
      "epoch": 0.5204363859864782,
      "grad_norm": 2.448591709136963,
      "learning_rate": 1.791825445605409e-05,
      "loss": 1.2853,
      "step": 6774
    },
    {
      "epoch": 0.5205900430239705,
      "grad_norm": 2.9498369693756104,
      "learning_rate": 1.7917639827904117e-05,
      "loss": 1.306,
      "step": 6776
    },
    {
      "epoch": 0.5207437000614629,
      "grad_norm": 3.2364501953125,
      "learning_rate": 1.791702519975415e-05,
      "loss": 1.379,
      "step": 6778
    },
    {
      "epoch": 0.5208973570989551,
      "grad_norm": 3.1900076866149902,
      "learning_rate": 1.791641057160418e-05,
      "loss": 1.4177,
      "step": 6780
    },
    {
      "epoch": 0.5210510141364475,
      "grad_norm": 3.0492565631866455,
      "learning_rate": 1.791579594345421e-05,
      "loss": 1.4582,
      "step": 6782
    },
    {
      "epoch": 0.5212046711739398,
      "grad_norm": 2.8004837036132812,
      "learning_rate": 1.7915181315304242e-05,
      "loss": 1.4414,
      "step": 6784
    },
    {
      "epoch": 0.5213583282114321,
      "grad_norm": 2.4796876907348633,
      "learning_rate": 1.7914566687154272e-05,
      "loss": 1.388,
      "step": 6786
    },
    {
      "epoch": 0.5215119852489244,
      "grad_norm": 2.892904281616211,
      "learning_rate": 1.7913952059004305e-05,
      "loss": 1.4414,
      "step": 6788
    },
    {
      "epoch": 0.5216656422864168,
      "grad_norm": 3.042837381362915,
      "learning_rate": 1.7913337430854335e-05,
      "loss": 1.4302,
      "step": 6790
    },
    {
      "epoch": 0.521819299323909,
      "grad_norm": 3.3126227855682373,
      "learning_rate": 1.7912722802704365e-05,
      "loss": 1.4458,
      "step": 6792
    },
    {
      "epoch": 0.5219729563614014,
      "grad_norm": 2.8981101512908936,
      "learning_rate": 1.7912108174554398e-05,
      "loss": 1.4403,
      "step": 6794
    },
    {
      "epoch": 0.5221266133988937,
      "grad_norm": 3.016014337539673,
      "learning_rate": 1.7911493546404427e-05,
      "loss": 1.235,
      "step": 6796
    },
    {
      "epoch": 0.522280270436386,
      "grad_norm": 2.693057060241699,
      "learning_rate": 1.7910878918254457e-05,
      "loss": 1.3,
      "step": 6798
    },
    {
      "epoch": 0.5224339274738783,
      "grad_norm": 2.8503472805023193,
      "learning_rate": 1.791026429010449e-05,
      "loss": 1.2262,
      "step": 6800
    },
    {
      "epoch": 0.5225875845113707,
      "grad_norm": 3.0016205310821533,
      "learning_rate": 1.7909649661954516e-05,
      "loss": 1.328,
      "step": 6802
    },
    {
      "epoch": 0.5227412415488629,
      "grad_norm": 2.7652969360351562,
      "learning_rate": 1.790903503380455e-05,
      "loss": 1.1953,
      "step": 6804
    },
    {
      "epoch": 0.5228948985863553,
      "grad_norm": 3.02223539352417,
      "learning_rate": 1.790842040565458e-05,
      "loss": 1.1621,
      "step": 6806
    },
    {
      "epoch": 0.5230485556238476,
      "grad_norm": 3.1197988986968994,
      "learning_rate": 1.7907805777504612e-05,
      "loss": 1.4014,
      "step": 6808
    },
    {
      "epoch": 0.5232022126613399,
      "grad_norm": 2.959120273590088,
      "learning_rate": 1.7907191149354642e-05,
      "loss": 1.4844,
      "step": 6810
    },
    {
      "epoch": 0.5233558696988322,
      "grad_norm": 3.0649282932281494,
      "learning_rate": 1.790657652120467e-05,
      "loss": 1.4225,
      "step": 6812
    },
    {
      "epoch": 0.5235095267363246,
      "grad_norm": 2.9492135047912598,
      "learning_rate": 1.7905961893054705e-05,
      "loss": 1.4206,
      "step": 6814
    },
    {
      "epoch": 0.5236631837738168,
      "grad_norm": 3.1265509128570557,
      "learning_rate": 1.7905347264904734e-05,
      "loss": 1.317,
      "step": 6816
    },
    {
      "epoch": 0.5238168408113092,
      "grad_norm": 2.831228256225586,
      "learning_rate": 1.7904732636754764e-05,
      "loss": 1.4686,
      "step": 6818
    },
    {
      "epoch": 0.5239704978488015,
      "grad_norm": 3.228123903274536,
      "learning_rate": 1.7904118008604797e-05,
      "loss": 1.4721,
      "step": 6820
    },
    {
      "epoch": 0.5241241548862938,
      "grad_norm": 3.5068886280059814,
      "learning_rate": 1.7903503380454827e-05,
      "loss": 1.4167,
      "step": 6822
    },
    {
      "epoch": 0.5242778119237861,
      "grad_norm": 2.8935916423797607,
      "learning_rate": 1.7902888752304856e-05,
      "loss": 1.362,
      "step": 6824
    },
    {
      "epoch": 0.5244314689612785,
      "grad_norm": 3.246994972229004,
      "learning_rate": 1.790227412415489e-05,
      "loss": 1.3156,
      "step": 6826
    },
    {
      "epoch": 0.5245851259987707,
      "grad_norm": 3.070075035095215,
      "learning_rate": 1.790165949600492e-05,
      "loss": 1.3155,
      "step": 6828
    },
    {
      "epoch": 0.5247387830362631,
      "grad_norm": 3.2165653705596924,
      "learning_rate": 1.790104486785495e-05,
      "loss": 1.4158,
      "step": 6830
    },
    {
      "epoch": 0.5248924400737554,
      "grad_norm": 3.122731924057007,
      "learning_rate": 1.790043023970498e-05,
      "loss": 1.312,
      "step": 6832
    },
    {
      "epoch": 0.5250460971112477,
      "grad_norm": 2.7937536239624023,
      "learning_rate": 1.789981561155501e-05,
      "loss": 1.4255,
      "step": 6834
    },
    {
      "epoch": 0.52519975414874,
      "grad_norm": 2.8714072704315186,
      "learning_rate": 1.789920098340504e-05,
      "loss": 1.3561,
      "step": 6836
    },
    {
      "epoch": 0.5253534111862324,
      "grad_norm": 3.3876631259918213,
      "learning_rate": 1.789858635525507e-05,
      "loss": 1.2653,
      "step": 6838
    },
    {
      "epoch": 0.5255070682237246,
      "grad_norm": 3.067852258682251,
      "learning_rate": 1.7897971727105104e-05,
      "loss": 1.462,
      "step": 6840
    },
    {
      "epoch": 0.525660725261217,
      "grad_norm": 2.623149871826172,
      "learning_rate": 1.7897357098955134e-05,
      "loss": 1.358,
      "step": 6842
    },
    {
      "epoch": 0.5258143822987092,
      "grad_norm": 3.001631021499634,
      "learning_rate": 1.7896742470805163e-05,
      "loss": 1.4515,
      "step": 6844
    },
    {
      "epoch": 0.5259680393362016,
      "grad_norm": 2.7457780838012695,
      "learning_rate": 1.7896127842655197e-05,
      "loss": 1.2601,
      "step": 6846
    },
    {
      "epoch": 0.5261216963736939,
      "grad_norm": 3.05543851852417,
      "learning_rate": 1.7895513214505226e-05,
      "loss": 1.3837,
      "step": 6848
    },
    {
      "epoch": 0.5262753534111863,
      "grad_norm": 2.7177581787109375,
      "learning_rate": 1.7894898586355256e-05,
      "loss": 1.2661,
      "step": 6850
    },
    {
      "epoch": 0.5264290104486785,
      "grad_norm": 2.9022562503814697,
      "learning_rate": 1.789428395820529e-05,
      "loss": 1.3782,
      "step": 6852
    },
    {
      "epoch": 0.5265826674861709,
      "grad_norm": 3.01700758934021,
      "learning_rate": 1.789366933005532e-05,
      "loss": 1.3188,
      "step": 6854
    },
    {
      "epoch": 0.5267363245236631,
      "grad_norm": 2.9770376682281494,
      "learning_rate": 1.789305470190535e-05,
      "loss": 1.4046,
      "step": 6856
    },
    {
      "epoch": 0.5268899815611555,
      "grad_norm": 3.635504961013794,
      "learning_rate": 1.7892440073755378e-05,
      "loss": 1.3997,
      "step": 6858
    },
    {
      "epoch": 0.5270436385986478,
      "grad_norm": 3.3790907859802246,
      "learning_rate": 1.789182544560541e-05,
      "loss": 1.3248,
      "step": 6860
    },
    {
      "epoch": 0.5271972956361402,
      "grad_norm": 3.0795631408691406,
      "learning_rate": 1.789121081745544e-05,
      "loss": 1.2888,
      "step": 6862
    },
    {
      "epoch": 0.5273509526736324,
      "grad_norm": 2.7538647651672363,
      "learning_rate": 1.789059618930547e-05,
      "loss": 1.5236,
      "step": 6864
    },
    {
      "epoch": 0.5275046097111248,
      "grad_norm": 3.4936375617980957,
      "learning_rate": 1.7889981561155504e-05,
      "loss": 1.4455,
      "step": 6866
    },
    {
      "epoch": 0.527658266748617,
      "grad_norm": 3.7572572231292725,
      "learning_rate": 1.7889366933005533e-05,
      "loss": 1.4181,
      "step": 6868
    },
    {
      "epoch": 0.5278119237861094,
      "grad_norm": 3.407961130142212,
      "learning_rate": 1.7888752304855563e-05,
      "loss": 1.3646,
      "step": 6870
    },
    {
      "epoch": 0.5279655808236017,
      "grad_norm": 2.7877607345581055,
      "learning_rate": 1.7888137676705596e-05,
      "loss": 1.259,
      "step": 6872
    },
    {
      "epoch": 0.528119237861094,
      "grad_norm": 2.999338150024414,
      "learning_rate": 1.7887523048555626e-05,
      "loss": 1.3156,
      "step": 6874
    },
    {
      "epoch": 0.5282728948985863,
      "grad_norm": 3.449953079223633,
      "learning_rate": 1.7886908420405655e-05,
      "loss": 1.4194,
      "step": 6876
    },
    {
      "epoch": 0.5284265519360787,
      "grad_norm": 3.0166075229644775,
      "learning_rate": 1.788629379225569e-05,
      "loss": 1.3544,
      "step": 6878
    },
    {
      "epoch": 0.5285802089735709,
      "grad_norm": 2.9438271522521973,
      "learning_rate": 1.7885679164105718e-05,
      "loss": 1.4669,
      "step": 6880
    },
    {
      "epoch": 0.5287338660110633,
      "grad_norm": 2.558922529220581,
      "learning_rate": 1.7885064535955748e-05,
      "loss": 1.2715,
      "step": 6882
    },
    {
      "epoch": 0.5288875230485556,
      "grad_norm": 2.997262954711914,
      "learning_rate": 1.7884449907805777e-05,
      "loss": 1.3229,
      "step": 6884
    },
    {
      "epoch": 0.529041180086048,
      "grad_norm": 2.7784371376037598,
      "learning_rate": 1.788383527965581e-05,
      "loss": 1.2801,
      "step": 6886
    },
    {
      "epoch": 0.5291948371235402,
      "grad_norm": 3.280125379562378,
      "learning_rate": 1.788322065150584e-05,
      "loss": 1.3505,
      "step": 6888
    },
    {
      "epoch": 0.5293484941610326,
      "grad_norm": 3.9201979637145996,
      "learning_rate": 1.788260602335587e-05,
      "loss": 1.3235,
      "step": 6890
    },
    {
      "epoch": 0.5295021511985248,
      "grad_norm": 3.0474581718444824,
      "learning_rate": 1.7881991395205903e-05,
      "loss": 1.3269,
      "step": 6892
    },
    {
      "epoch": 0.5296558082360172,
      "grad_norm": 2.956028938293457,
      "learning_rate": 1.7881376767055933e-05,
      "loss": 1.2897,
      "step": 6894
    },
    {
      "epoch": 0.5298094652735095,
      "grad_norm": 3.143683433532715,
      "learning_rate": 1.7880762138905962e-05,
      "loss": 1.3913,
      "step": 6896
    },
    {
      "epoch": 0.5299631223110018,
      "grad_norm": 2.942600727081299,
      "learning_rate": 1.7880147510755995e-05,
      "loss": 1.2993,
      "step": 6898
    },
    {
      "epoch": 0.5301167793484942,
      "grad_norm": 2.7133402824401855,
      "learning_rate": 1.7879532882606025e-05,
      "loss": 1.2878,
      "step": 6900
    },
    {
      "epoch": 0.5302704363859865,
      "grad_norm": 2.7319624423980713,
      "learning_rate": 1.7878918254456055e-05,
      "loss": 1.2711,
      "step": 6902
    },
    {
      "epoch": 0.5304240934234788,
      "grad_norm": 3.0787789821624756,
      "learning_rate": 1.7878303626306084e-05,
      "loss": 1.4048,
      "step": 6904
    },
    {
      "epoch": 0.5305777504609711,
      "grad_norm": 3.218059539794922,
      "learning_rate": 1.7877688998156118e-05,
      "loss": 1.346,
      "step": 6906
    },
    {
      "epoch": 0.5307314074984635,
      "grad_norm": 3.2375054359436035,
      "learning_rate": 1.7877074370006147e-05,
      "loss": 1.3984,
      "step": 6908
    },
    {
      "epoch": 0.5308850645359557,
      "grad_norm": 2.9513885974884033,
      "learning_rate": 1.7876459741856177e-05,
      "loss": 1.4062,
      "step": 6910
    },
    {
      "epoch": 0.5310387215734481,
      "grad_norm": 3.1400296688079834,
      "learning_rate": 1.787584511370621e-05,
      "loss": 1.4292,
      "step": 6912
    },
    {
      "epoch": 0.5311923786109404,
      "grad_norm": 3.114333152770996,
      "learning_rate": 1.787523048555624e-05,
      "loss": 1.5008,
      "step": 6914
    },
    {
      "epoch": 0.5313460356484327,
      "grad_norm": 2.8262038230895996,
      "learning_rate": 1.787461585740627e-05,
      "loss": 1.4744,
      "step": 6916
    },
    {
      "epoch": 0.531499692685925,
      "grad_norm": 3.2163949012756348,
      "learning_rate": 1.7874001229256302e-05,
      "loss": 1.3959,
      "step": 6918
    },
    {
      "epoch": 0.5316533497234174,
      "grad_norm": 2.7985939979553223,
      "learning_rate": 1.7873386601106332e-05,
      "loss": 1.4022,
      "step": 6920
    },
    {
      "epoch": 0.5318070067609096,
      "grad_norm": 2.731464147567749,
      "learning_rate": 1.7872771972956362e-05,
      "loss": 1.3438,
      "step": 6922
    },
    {
      "epoch": 0.531960663798402,
      "grad_norm": 3.187049627304077,
      "learning_rate": 1.7872157344806395e-05,
      "loss": 1.4009,
      "step": 6924
    },
    {
      "epoch": 0.5321143208358943,
      "grad_norm": 2.9336483478546143,
      "learning_rate": 1.7871542716656425e-05,
      "loss": 1.412,
      "step": 6926
    },
    {
      "epoch": 0.5322679778733866,
      "grad_norm": 3.132073402404785,
      "learning_rate": 1.7870928088506454e-05,
      "loss": 1.3688,
      "step": 6928
    },
    {
      "epoch": 0.5324216349108789,
      "grad_norm": 2.981800079345703,
      "learning_rate": 1.7870313460356484e-05,
      "loss": 1.3054,
      "step": 6930
    },
    {
      "epoch": 0.5325752919483713,
      "grad_norm": 3.3410542011260986,
      "learning_rate": 1.7869698832206517e-05,
      "loss": 1.3939,
      "step": 6932
    },
    {
      "epoch": 0.5327289489858635,
      "grad_norm": 2.8164236545562744,
      "learning_rate": 1.7869084204056547e-05,
      "loss": 1.3269,
      "step": 6934
    },
    {
      "epoch": 0.5328826060233559,
      "grad_norm": 3.8438332080841064,
      "learning_rate": 1.7868469575906576e-05,
      "loss": 1.3612,
      "step": 6936
    },
    {
      "epoch": 0.5330362630608482,
      "grad_norm": 2.7539196014404297,
      "learning_rate": 1.786785494775661e-05,
      "loss": 1.351,
      "step": 6938
    },
    {
      "epoch": 0.5331899200983405,
      "grad_norm": 3.0499660968780518,
      "learning_rate": 1.786724031960664e-05,
      "loss": 1.4173,
      "step": 6940
    },
    {
      "epoch": 0.5333435771358328,
      "grad_norm": 3.459529399871826,
      "learning_rate": 1.786662569145667e-05,
      "loss": 1.4837,
      "step": 6942
    },
    {
      "epoch": 0.5334972341733252,
      "grad_norm": 3.2329859733581543,
      "learning_rate": 1.7866011063306702e-05,
      "loss": 1.4251,
      "step": 6944
    },
    {
      "epoch": 0.5336508912108174,
      "grad_norm": 3.210688352584839,
      "learning_rate": 1.786539643515673e-05,
      "loss": 1.3004,
      "step": 6946
    },
    {
      "epoch": 0.5338045482483098,
      "grad_norm": 3.564739227294922,
      "learning_rate": 1.786478180700676e-05,
      "loss": 1.443,
      "step": 6948
    },
    {
      "epoch": 0.5339582052858021,
      "grad_norm": 2.8052964210510254,
      "learning_rate": 1.7864167178856794e-05,
      "loss": 1.2934,
      "step": 6950
    },
    {
      "epoch": 0.5341118623232944,
      "grad_norm": 3.189671039581299,
      "learning_rate": 1.7863552550706824e-05,
      "loss": 1.3459,
      "step": 6952
    },
    {
      "epoch": 0.5342655193607867,
      "grad_norm": 3.1696484088897705,
      "learning_rate": 1.7862937922556857e-05,
      "loss": 1.4234,
      "step": 6954
    },
    {
      "epoch": 0.5344191763982791,
      "grad_norm": 2.973815679550171,
      "learning_rate": 1.7862323294406883e-05,
      "loss": 1.2077,
      "step": 6956
    },
    {
      "epoch": 0.5345728334357713,
      "grad_norm": 3.508512258529663,
      "learning_rate": 1.7861708666256916e-05,
      "loss": 1.3832,
      "step": 6958
    },
    {
      "epoch": 0.5347264904732637,
      "grad_norm": 2.8029847145080566,
      "learning_rate": 1.7861094038106946e-05,
      "loss": 1.3648,
      "step": 6960
    },
    {
      "epoch": 0.534880147510756,
      "grad_norm": 3.183760404586792,
      "learning_rate": 1.7860479409956976e-05,
      "loss": 1.4012,
      "step": 6962
    },
    {
      "epoch": 0.5350338045482483,
      "grad_norm": 3.181417942047119,
      "learning_rate": 1.785986478180701e-05,
      "loss": 1.3642,
      "step": 6964
    },
    {
      "epoch": 0.5351874615857406,
      "grad_norm": 3.075376033782959,
      "learning_rate": 1.785925015365704e-05,
      "loss": 1.4268,
      "step": 6966
    },
    {
      "epoch": 0.535341118623233,
      "grad_norm": 3.069875478744507,
      "learning_rate": 1.7858635525507068e-05,
      "loss": 1.4777,
      "step": 6968
    },
    {
      "epoch": 0.5354947756607252,
      "grad_norm": 3.1056337356567383,
      "learning_rate": 1.78580208973571e-05,
      "loss": 1.4871,
      "step": 6970
    },
    {
      "epoch": 0.5356484326982176,
      "grad_norm": 3.0267107486724854,
      "learning_rate": 1.785740626920713e-05,
      "loss": 1.4342,
      "step": 6972
    },
    {
      "epoch": 0.5358020897357099,
      "grad_norm": 3.082707643508911,
      "learning_rate": 1.7856791641057164e-05,
      "loss": 1.4585,
      "step": 6974
    },
    {
      "epoch": 0.5359557467732022,
      "grad_norm": 3.154895544052124,
      "learning_rate": 1.7856177012907194e-05,
      "loss": 1.4232,
      "step": 6976
    },
    {
      "epoch": 0.5361094038106945,
      "grad_norm": 2.978407621383667,
      "learning_rate": 1.7855562384757223e-05,
      "loss": 1.3511,
      "step": 6978
    },
    {
      "epoch": 0.5362630608481869,
      "grad_norm": 2.940761089324951,
      "learning_rate": 1.7854947756607257e-05,
      "loss": 1.372,
      "step": 6980
    },
    {
      "epoch": 0.5364167178856791,
      "grad_norm": 2.9799039363861084,
      "learning_rate": 1.7854333128457283e-05,
      "loss": 1.5043,
      "step": 6982
    },
    {
      "epoch": 0.5365703749231715,
      "grad_norm": 2.7917842864990234,
      "learning_rate": 1.7853718500307316e-05,
      "loss": 1.2237,
      "step": 6984
    },
    {
      "epoch": 0.5367240319606638,
      "grad_norm": 2.8482396602630615,
      "learning_rate": 1.7853103872157346e-05,
      "loss": 1.4475,
      "step": 6986
    },
    {
      "epoch": 0.5368776889981561,
      "grad_norm": 3.9226222038269043,
      "learning_rate": 1.7852489244007375e-05,
      "loss": 1.3577,
      "step": 6988
    },
    {
      "epoch": 0.5370313460356484,
      "grad_norm": 2.828793525695801,
      "learning_rate": 1.785187461585741e-05,
      "loss": 1.3318,
      "step": 6990
    },
    {
      "epoch": 0.5371850030731408,
      "grad_norm": 2.9199106693267822,
      "learning_rate": 1.7851259987707438e-05,
      "loss": 1.3467,
      "step": 6992
    },
    {
      "epoch": 0.537338660110633,
      "grad_norm": 2.835475444793701,
      "learning_rate": 1.785064535955747e-05,
      "loss": 1.3634,
      "step": 6994
    },
    {
      "epoch": 0.5374923171481254,
      "grad_norm": 3.4660840034484863,
      "learning_rate": 1.78500307314075e-05,
      "loss": 1.2317,
      "step": 6996
    },
    {
      "epoch": 0.5376459741856177,
      "grad_norm": 4.105753421783447,
      "learning_rate": 1.784941610325753e-05,
      "loss": 1.3137,
      "step": 6998
    },
    {
      "epoch": 0.53779963122311,
      "grad_norm": 2.8999927043914795,
      "learning_rate": 1.7848801475107564e-05,
      "loss": 1.2568,
      "step": 7000
    },
    {
      "epoch": 0.5379532882606023,
      "grad_norm": 2.814363718032837,
      "learning_rate": 1.784818684695759e-05,
      "loss": 1.4887,
      "step": 7002
    },
    {
      "epoch": 0.5381069452980947,
      "grad_norm": 2.9154648780822754,
      "learning_rate": 1.7847572218807623e-05,
      "loss": 1.1712,
      "step": 7004
    },
    {
      "epoch": 0.5382606023355869,
      "grad_norm": 3.22001051902771,
      "learning_rate": 1.7846957590657653e-05,
      "loss": 1.3,
      "step": 7006
    },
    {
      "epoch": 0.5384142593730793,
      "grad_norm": 3.288647174835205,
      "learning_rate": 1.7846342962507682e-05,
      "loss": 1.3765,
      "step": 7008
    },
    {
      "epoch": 0.5385679164105716,
      "grad_norm": 3.0418827533721924,
      "learning_rate": 1.7845728334357715e-05,
      "loss": 1.454,
      "step": 7010
    },
    {
      "epoch": 0.5387215734480639,
      "grad_norm": 2.8727383613586426,
      "learning_rate": 1.7845113706207745e-05,
      "loss": 1.3323,
      "step": 7012
    },
    {
      "epoch": 0.5388752304855562,
      "grad_norm": 3.0164737701416016,
      "learning_rate": 1.7844499078057778e-05,
      "loss": 1.2651,
      "step": 7014
    },
    {
      "epoch": 0.5390288875230486,
      "grad_norm": 2.6829867362976074,
      "learning_rate": 1.7843884449907808e-05,
      "loss": 1.3604,
      "step": 7016
    },
    {
      "epoch": 0.5391825445605408,
      "grad_norm": 3.5748116970062256,
      "learning_rate": 1.7843269821757837e-05,
      "loss": 1.4137,
      "step": 7018
    },
    {
      "epoch": 0.5393362015980332,
      "grad_norm": 2.9325039386749268,
      "learning_rate": 1.784265519360787e-05,
      "loss": 1.3057,
      "step": 7020
    },
    {
      "epoch": 0.5394898586355255,
      "grad_norm": 2.7085931301116943,
      "learning_rate": 1.78420405654579e-05,
      "loss": 1.1975,
      "step": 7022
    },
    {
      "epoch": 0.5396435156730178,
      "grad_norm": 2.880764961242676,
      "learning_rate": 1.784142593730793e-05,
      "loss": 1.35,
      "step": 7024
    },
    {
      "epoch": 0.5397971727105101,
      "grad_norm": 3.0631964206695557,
      "learning_rate": 1.7840811309157963e-05,
      "loss": 1.2638,
      "step": 7026
    },
    {
      "epoch": 0.5399508297480025,
      "grad_norm": 4.236493110656738,
      "learning_rate": 1.784019668100799e-05,
      "loss": 1.2436,
      "step": 7028
    },
    {
      "epoch": 0.5401044867854948,
      "grad_norm": 2.839618682861328,
      "learning_rate": 1.7839582052858022e-05,
      "loss": 1.3736,
      "step": 7030
    },
    {
      "epoch": 0.5402581438229871,
      "grad_norm": 3.1497669219970703,
      "learning_rate": 1.7838967424708052e-05,
      "loss": 1.4248,
      "step": 7032
    },
    {
      "epoch": 0.5404118008604795,
      "grad_norm": 2.8350067138671875,
      "learning_rate": 1.783835279655808e-05,
      "loss": 1.2533,
      "step": 7034
    },
    {
      "epoch": 0.5405654578979717,
      "grad_norm": 33.07500457763672,
      "learning_rate": 1.7837738168408115e-05,
      "loss": 1.4906,
      "step": 7036
    },
    {
      "epoch": 0.5407191149354641,
      "grad_norm": 3.066897392272949,
      "learning_rate": 1.7837123540258144e-05,
      "loss": 1.4008,
      "step": 7038
    },
    {
      "epoch": 0.5408727719729564,
      "grad_norm": 3.2352633476257324,
      "learning_rate": 1.7836508912108178e-05,
      "loss": 1.3084,
      "step": 7040
    },
    {
      "epoch": 0.5410264290104487,
      "grad_norm": 3.37577486038208,
      "learning_rate": 1.7835894283958207e-05,
      "loss": 1.3395,
      "step": 7042
    },
    {
      "epoch": 0.541180086047941,
      "grad_norm": 2.8560941219329834,
      "learning_rate": 1.7835279655808237e-05,
      "loss": 1.3559,
      "step": 7044
    },
    {
      "epoch": 0.5413337430854334,
      "grad_norm": 3.0230565071105957,
      "learning_rate": 1.783466502765827e-05,
      "loss": 1.3866,
      "step": 7046
    },
    {
      "epoch": 0.5414874001229256,
      "grad_norm": 2.6992247104644775,
      "learning_rate": 1.78340503995083e-05,
      "loss": 1.2888,
      "step": 7048
    },
    {
      "epoch": 0.541641057160418,
      "grad_norm": 3.213545322418213,
      "learning_rate": 1.783343577135833e-05,
      "loss": 1.4569,
      "step": 7050
    },
    {
      "epoch": 0.5417947141979103,
      "grad_norm": 3.045323610305786,
      "learning_rate": 1.7832821143208362e-05,
      "loss": 1.406,
      "step": 7052
    },
    {
      "epoch": 0.5419483712354026,
      "grad_norm": 3.1526753902435303,
      "learning_rate": 1.783220651505839e-05,
      "loss": 1.4462,
      "step": 7054
    },
    {
      "epoch": 0.5421020282728949,
      "grad_norm": 2.5964386463165283,
      "learning_rate": 1.7831591886908422e-05,
      "loss": 1.4212,
      "step": 7056
    },
    {
      "epoch": 0.5422556853103873,
      "grad_norm": 3.107309341430664,
      "learning_rate": 1.783097725875845e-05,
      "loss": 1.2955,
      "step": 7058
    },
    {
      "epoch": 0.5424093423478795,
      "grad_norm": 3.1672708988189697,
      "learning_rate": 1.7830362630608485e-05,
      "loss": 1.282,
      "step": 7060
    },
    {
      "epoch": 0.5425629993853719,
      "grad_norm": 3.1407082080841064,
      "learning_rate": 1.7829748002458514e-05,
      "loss": 1.2697,
      "step": 7062
    },
    {
      "epoch": 0.5427166564228642,
      "grad_norm": 2.588823080062866,
      "learning_rate": 1.7829133374308544e-05,
      "loss": 1.2566,
      "step": 7064
    },
    {
      "epoch": 0.5428703134603565,
      "grad_norm": 2.709662914276123,
      "learning_rate": 1.7828518746158577e-05,
      "loss": 1.3653,
      "step": 7066
    },
    {
      "epoch": 0.5430239704978488,
      "grad_norm": 2.963088274002075,
      "learning_rate": 1.7827904118008607e-05,
      "loss": 1.3215,
      "step": 7068
    },
    {
      "epoch": 0.5431776275353412,
      "grad_norm": 3.0335934162139893,
      "learning_rate": 1.7827289489858636e-05,
      "loss": 1.4173,
      "step": 7070
    },
    {
      "epoch": 0.5433312845728334,
      "grad_norm": 3.156770944595337,
      "learning_rate": 1.782667486170867e-05,
      "loss": 1.3679,
      "step": 7072
    },
    {
      "epoch": 0.5434849416103258,
      "grad_norm": 3.3931353092193604,
      "learning_rate": 1.78260602335587e-05,
      "loss": 1.4272,
      "step": 7074
    },
    {
      "epoch": 0.543638598647818,
      "grad_norm": 2.5168139934539795,
      "learning_rate": 1.782544560540873e-05,
      "loss": 1.3466,
      "step": 7076
    },
    {
      "epoch": 0.5437922556853104,
      "grad_norm": 2.82444429397583,
      "learning_rate": 1.7824830977258762e-05,
      "loss": 1.3579,
      "step": 7078
    },
    {
      "epoch": 0.5439459127228027,
      "grad_norm": 3.1184284687042236,
      "learning_rate": 1.782421634910879e-05,
      "loss": 1.4413,
      "step": 7080
    },
    {
      "epoch": 0.544099569760295,
      "grad_norm": 3.5062296390533447,
      "learning_rate": 1.782360172095882e-05,
      "loss": 1.4573,
      "step": 7082
    },
    {
      "epoch": 0.5442532267977873,
      "grad_norm": 3.173992395401001,
      "learning_rate": 1.782298709280885e-05,
      "loss": 1.3502,
      "step": 7084
    },
    {
      "epoch": 0.5444068838352797,
      "grad_norm": 2.8979501724243164,
      "learning_rate": 1.7822372464658884e-05,
      "loss": 1.4154,
      "step": 7086
    },
    {
      "epoch": 0.544560540872772,
      "grad_norm": 3.195020914077759,
      "learning_rate": 1.7821757836508914e-05,
      "loss": 1.408,
      "step": 7088
    },
    {
      "epoch": 0.5447141979102643,
      "grad_norm": 2.883970022201538,
      "learning_rate": 1.7821143208358943e-05,
      "loss": 1.3705,
      "step": 7090
    },
    {
      "epoch": 0.5448678549477566,
      "grad_norm": 3.1312525272369385,
      "learning_rate": 1.7820528580208976e-05,
      "loss": 1.3827,
      "step": 7092
    },
    {
      "epoch": 0.545021511985249,
      "grad_norm": 2.855011224746704,
      "learning_rate": 1.7819913952059006e-05,
      "loss": 1.1705,
      "step": 7094
    },
    {
      "epoch": 0.5451751690227412,
      "grad_norm": 2.7898807525634766,
      "learning_rate": 1.7819299323909036e-05,
      "loss": 1.2252,
      "step": 7096
    },
    {
      "epoch": 0.5453288260602336,
      "grad_norm": 3.0114316940307617,
      "learning_rate": 1.781868469575907e-05,
      "loss": 1.3664,
      "step": 7098
    },
    {
      "epoch": 0.5454824830977258,
      "grad_norm": 3.1177165508270264,
      "learning_rate": 1.78180700676091e-05,
      "loss": 1.4363,
      "step": 7100
    },
    {
      "epoch": 0.5456361401352182,
      "grad_norm": 3.265270471572876,
      "learning_rate": 1.7817455439459128e-05,
      "loss": 1.4604,
      "step": 7102
    },
    {
      "epoch": 0.5457897971727105,
      "grad_norm": 2.9632492065429688,
      "learning_rate": 1.7816840811309158e-05,
      "loss": 1.4497,
      "step": 7104
    },
    {
      "epoch": 0.5459434542102028,
      "grad_norm": 2.721161127090454,
      "learning_rate": 1.781622618315919e-05,
      "loss": 1.3612,
      "step": 7106
    },
    {
      "epoch": 0.5460971112476951,
      "grad_norm": 2.503848075866699,
      "learning_rate": 1.781561155500922e-05,
      "loss": 1.2765,
      "step": 7108
    },
    {
      "epoch": 0.5462507682851875,
      "grad_norm": 3.5437700748443604,
      "learning_rate": 1.781499692685925e-05,
      "loss": 1.3824,
      "step": 7110
    },
    {
      "epoch": 0.5464044253226797,
      "grad_norm": 2.859767436981201,
      "learning_rate": 1.7814382298709283e-05,
      "loss": 1.3918,
      "step": 7112
    },
    {
      "epoch": 0.5465580823601721,
      "grad_norm": 2.664762496948242,
      "learning_rate": 1.7813767670559313e-05,
      "loss": 1.3302,
      "step": 7114
    },
    {
      "epoch": 0.5467117393976644,
      "grad_norm": 2.8141744136810303,
      "learning_rate": 1.7813153042409343e-05,
      "loss": 1.2234,
      "step": 7116
    },
    {
      "epoch": 0.5468653964351567,
      "grad_norm": 3.569658041000366,
      "learning_rate": 1.7812538414259376e-05,
      "loss": 1.3897,
      "step": 7118
    },
    {
      "epoch": 0.547019053472649,
      "grad_norm": 2.765789747238159,
      "learning_rate": 1.7811923786109406e-05,
      "loss": 1.3527,
      "step": 7120
    },
    {
      "epoch": 0.5471727105101414,
      "grad_norm": 3.1543784141540527,
      "learning_rate": 1.7811309157959435e-05,
      "loss": 1.3532,
      "step": 7122
    },
    {
      "epoch": 0.5473263675476336,
      "grad_norm": 3.019174337387085,
      "learning_rate": 1.7810694529809468e-05,
      "loss": 1.3695,
      "step": 7124
    },
    {
      "epoch": 0.547480024585126,
      "grad_norm": 3.103487968444824,
      "learning_rate": 1.7810079901659498e-05,
      "loss": 1.3873,
      "step": 7126
    },
    {
      "epoch": 0.5476336816226183,
      "grad_norm": 2.756798505783081,
      "learning_rate": 1.7809465273509528e-05,
      "loss": 1.2963,
      "step": 7128
    },
    {
      "epoch": 0.5477873386601106,
      "grad_norm": 3.0003297328948975,
      "learning_rate": 1.7808850645359557e-05,
      "loss": 1.3152,
      "step": 7130
    },
    {
      "epoch": 0.5479409956976029,
      "grad_norm": 2.684039831161499,
      "learning_rate": 1.780823601720959e-05,
      "loss": 1.4226,
      "step": 7132
    },
    {
      "epoch": 0.5480946527350953,
      "grad_norm": 2.9724183082580566,
      "learning_rate": 1.780762138905962e-05,
      "loss": 1.4356,
      "step": 7134
    },
    {
      "epoch": 0.5482483097725875,
      "grad_norm": 2.8482649326324463,
      "learning_rate": 1.780700676090965e-05,
      "loss": 1.1933,
      "step": 7136
    },
    {
      "epoch": 0.5484019668100799,
      "grad_norm": 3.3711745738983154,
      "learning_rate": 1.7806392132759683e-05,
      "loss": 1.3696,
      "step": 7138
    },
    {
      "epoch": 0.5485556238475722,
      "grad_norm": 3.0896215438842773,
      "learning_rate": 1.7805777504609713e-05,
      "loss": 1.4287,
      "step": 7140
    },
    {
      "epoch": 0.5487092808850645,
      "grad_norm": 3.777125120162964,
      "learning_rate": 1.7805162876459742e-05,
      "loss": 1.3143,
      "step": 7142
    },
    {
      "epoch": 0.5488629379225568,
      "grad_norm": 3.1980645656585693,
      "learning_rate": 1.7804548248309775e-05,
      "loss": 1.3388,
      "step": 7144
    },
    {
      "epoch": 0.5490165949600492,
      "grad_norm": 3.0263702869415283,
      "learning_rate": 1.7803933620159805e-05,
      "loss": 1.2497,
      "step": 7146
    },
    {
      "epoch": 0.5491702519975414,
      "grad_norm": 3.010733127593994,
      "learning_rate": 1.7803318992009835e-05,
      "loss": 1.1814,
      "step": 7148
    },
    {
      "epoch": 0.5493239090350338,
      "grad_norm": 2.8031928539276123,
      "learning_rate": 1.7802704363859868e-05,
      "loss": 1.251,
      "step": 7150
    },
    {
      "epoch": 0.5494775660725261,
      "grad_norm": 3.2427356243133545,
      "learning_rate": 1.7802089735709897e-05,
      "loss": 1.3079,
      "step": 7152
    },
    {
      "epoch": 0.5496312231100184,
      "grad_norm": 3.0464084148406982,
      "learning_rate": 1.7801475107559927e-05,
      "loss": 1.3989,
      "step": 7154
    },
    {
      "epoch": 0.5497848801475107,
      "grad_norm": 2.847355604171753,
      "learning_rate": 1.7800860479409957e-05,
      "loss": 1.3637,
      "step": 7156
    },
    {
      "epoch": 0.5499385371850031,
      "grad_norm": 3.1915597915649414,
      "learning_rate": 1.780024585125999e-05,
      "loss": 1.3317,
      "step": 7158
    },
    {
      "epoch": 0.5500921942224954,
      "grad_norm": 3.0718259811401367,
      "learning_rate": 1.779963122311002e-05,
      "loss": 1.3284,
      "step": 7160
    },
    {
      "epoch": 0.5502458512599877,
      "grad_norm": 3.015608310699463,
      "learning_rate": 1.779901659496005e-05,
      "loss": 1.3458,
      "step": 7162
    },
    {
      "epoch": 0.5503995082974801,
      "grad_norm": 3.1199285984039307,
      "learning_rate": 1.7798401966810082e-05,
      "loss": 1.3407,
      "step": 7164
    },
    {
      "epoch": 0.5505531653349723,
      "grad_norm": 3.2667245864868164,
      "learning_rate": 1.7797787338660112e-05,
      "loss": 1.3999,
      "step": 7166
    },
    {
      "epoch": 0.5507068223724647,
      "grad_norm": 2.711768865585327,
      "learning_rate": 1.779717271051014e-05,
      "loss": 1.2374,
      "step": 7168
    },
    {
      "epoch": 0.550860479409957,
      "grad_norm": 2.933833599090576,
      "learning_rate": 1.7796558082360175e-05,
      "loss": 1.2441,
      "step": 7170
    },
    {
      "epoch": 0.5510141364474493,
      "grad_norm": 2.5497310161590576,
      "learning_rate": 1.7795943454210204e-05,
      "loss": 1.3304,
      "step": 7172
    },
    {
      "epoch": 0.5511677934849416,
      "grad_norm": 2.6509580612182617,
      "learning_rate": 1.7795328826060234e-05,
      "loss": 1.3488,
      "step": 7174
    },
    {
      "epoch": 0.551321450522434,
      "grad_norm": 2.720896005630493,
      "learning_rate": 1.7794714197910267e-05,
      "loss": 1.2818,
      "step": 7176
    },
    {
      "epoch": 0.5514751075599262,
      "grad_norm": 3.513991355895996,
      "learning_rate": 1.7794099569760297e-05,
      "loss": 1.2761,
      "step": 7178
    },
    {
      "epoch": 0.5516287645974186,
      "grad_norm": 3.5131587982177734,
      "learning_rate": 1.7793484941610327e-05,
      "loss": 1.4,
      "step": 7180
    },
    {
      "epoch": 0.5517824216349109,
      "grad_norm": 3.281924247741699,
      "learning_rate": 1.7792870313460356e-05,
      "loss": 1.3619,
      "step": 7182
    },
    {
      "epoch": 0.5519360786724032,
      "grad_norm": 2.9389426708221436,
      "learning_rate": 1.779225568531039e-05,
      "loss": 1.6219,
      "step": 7184
    },
    {
      "epoch": 0.5520897357098955,
      "grad_norm": 2.482316493988037,
      "learning_rate": 1.779164105716042e-05,
      "loss": 1.2586,
      "step": 7186
    },
    {
      "epoch": 0.5522433927473879,
      "grad_norm": 3.0932509899139404,
      "learning_rate": 1.779102642901045e-05,
      "loss": 1.3397,
      "step": 7188
    },
    {
      "epoch": 0.5523970497848801,
      "grad_norm": 3.1350340843200684,
      "learning_rate": 1.7790411800860482e-05,
      "loss": 1.255,
      "step": 7190
    },
    {
      "epoch": 0.5525507068223725,
      "grad_norm": 2.5633490085601807,
      "learning_rate": 1.778979717271051e-05,
      "loss": 1.1744,
      "step": 7192
    },
    {
      "epoch": 0.5527043638598648,
      "grad_norm": 2.803712844848633,
      "learning_rate": 1.778918254456054e-05,
      "loss": 1.4449,
      "step": 7194
    },
    {
      "epoch": 0.5528580208973571,
      "grad_norm": 2.5207178592681885,
      "learning_rate": 1.7788567916410574e-05,
      "loss": 1.4056,
      "step": 7196
    },
    {
      "epoch": 0.5530116779348494,
      "grad_norm": 2.7499046325683594,
      "learning_rate": 1.7787953288260604e-05,
      "loss": 1.2398,
      "step": 7198
    },
    {
      "epoch": 0.5531653349723418,
      "grad_norm": 3.0889878273010254,
      "learning_rate": 1.7787338660110634e-05,
      "loss": 1.4251,
      "step": 7200
    },
    {
      "epoch": 0.553318992009834,
      "grad_norm": 2.5837948322296143,
      "learning_rate": 1.7786724031960667e-05,
      "loss": 1.345,
      "step": 7202
    },
    {
      "epoch": 0.5534726490473264,
      "grad_norm": 3.2794241905212402,
      "learning_rate": 1.7786109403810696e-05,
      "loss": 1.4219,
      "step": 7204
    },
    {
      "epoch": 0.5536263060848187,
      "grad_norm": 2.494079828262329,
      "learning_rate": 1.778549477566073e-05,
      "loss": 1.2795,
      "step": 7206
    },
    {
      "epoch": 0.553779963122311,
      "grad_norm": 3.3279178142547607,
      "learning_rate": 1.7784880147510756e-05,
      "loss": 1.3046,
      "step": 7208
    },
    {
      "epoch": 0.5539336201598033,
      "grad_norm": 2.7889404296875,
      "learning_rate": 1.778426551936079e-05,
      "loss": 1.2484,
      "step": 7210
    },
    {
      "epoch": 0.5540872771972957,
      "grad_norm": 2.8925845623016357,
      "learning_rate": 1.778365089121082e-05,
      "loss": 1.2489,
      "step": 7212
    },
    {
      "epoch": 0.5542409342347879,
      "grad_norm": 3.148367166519165,
      "learning_rate": 1.7783036263060848e-05,
      "loss": 1.4886,
      "step": 7214
    },
    {
      "epoch": 0.5543945912722803,
      "grad_norm": 3.505509376525879,
      "learning_rate": 1.778242163491088e-05,
      "loss": 1.4575,
      "step": 7216
    },
    {
      "epoch": 0.5545482483097726,
      "grad_norm": 2.391782760620117,
      "learning_rate": 1.778180700676091e-05,
      "loss": 1.2687,
      "step": 7218
    },
    {
      "epoch": 0.5547019053472649,
      "grad_norm": 2.8791768550872803,
      "learning_rate": 1.778119237861094e-05,
      "loss": 1.4009,
      "step": 7220
    },
    {
      "epoch": 0.5548555623847572,
      "grad_norm": 3.043830633163452,
      "learning_rate": 1.7780577750460974e-05,
      "loss": 1.3839,
      "step": 7222
    },
    {
      "epoch": 0.5550092194222496,
      "grad_norm": 3.7112839221954346,
      "learning_rate": 1.7779963122311003e-05,
      "loss": 1.3653,
      "step": 7224
    },
    {
      "epoch": 0.5551628764597418,
      "grad_norm": 2.8248283863067627,
      "learning_rate": 1.7779348494161036e-05,
      "loss": 1.3582,
      "step": 7226
    },
    {
      "epoch": 0.5553165334972342,
      "grad_norm": 2.891206979751587,
      "learning_rate": 1.7778733866011063e-05,
      "loss": 1.2634,
      "step": 7228
    },
    {
      "epoch": 0.5554701905347265,
      "grad_norm": 2.6088945865631104,
      "learning_rate": 1.7778119237861096e-05,
      "loss": 1.4858,
      "step": 7230
    },
    {
      "epoch": 0.5556238475722188,
      "grad_norm": 3.5376429557800293,
      "learning_rate": 1.7777504609711125e-05,
      "loss": 1.3223,
      "step": 7232
    },
    {
      "epoch": 0.5557775046097111,
      "grad_norm": 2.7143330574035645,
      "learning_rate": 1.7776889981561155e-05,
      "loss": 1.2609,
      "step": 7234
    },
    {
      "epoch": 0.5559311616472035,
      "grad_norm": 2.7578842639923096,
      "learning_rate": 1.7776275353411188e-05,
      "loss": 1.3574,
      "step": 7236
    },
    {
      "epoch": 0.5560848186846957,
      "grad_norm": 2.8860747814178467,
      "learning_rate": 1.7775660725261218e-05,
      "loss": 1.2957,
      "step": 7238
    },
    {
      "epoch": 0.5562384757221881,
      "grad_norm": 3.3761367797851562,
      "learning_rate": 1.7775046097111248e-05,
      "loss": 1.396,
      "step": 7240
    },
    {
      "epoch": 0.5563921327596804,
      "grad_norm": 2.8714826107025146,
      "learning_rate": 1.777443146896128e-05,
      "loss": 1.289,
      "step": 7242
    },
    {
      "epoch": 0.5565457897971727,
      "grad_norm": 2.9392268657684326,
      "learning_rate": 1.777381684081131e-05,
      "loss": 1.3034,
      "step": 7244
    },
    {
      "epoch": 0.556699446834665,
      "grad_norm": 2.9862611293792725,
      "learning_rate": 1.7773202212661343e-05,
      "loss": 1.2842,
      "step": 7246
    },
    {
      "epoch": 0.5568531038721574,
      "grad_norm": 3.157655954360962,
      "learning_rate": 1.7772587584511373e-05,
      "loss": 1.3206,
      "step": 7248
    },
    {
      "epoch": 0.5570067609096496,
      "grad_norm": 2.7796456813812256,
      "learning_rate": 1.7771972956361403e-05,
      "loss": 1.3415,
      "step": 7250
    },
    {
      "epoch": 0.557160417947142,
      "grad_norm": 2.800544023513794,
      "learning_rate": 1.7771358328211436e-05,
      "loss": 1.3417,
      "step": 7252
    },
    {
      "epoch": 0.5573140749846343,
      "grad_norm": 2.797663688659668,
      "learning_rate": 1.7770743700061462e-05,
      "loss": 1.2614,
      "step": 7254
    },
    {
      "epoch": 0.5574677320221266,
      "grad_norm": 2.899327278137207,
      "learning_rate": 1.7770129071911495e-05,
      "loss": 1.4148,
      "step": 7256
    },
    {
      "epoch": 0.5576213890596189,
      "grad_norm": 3.1357805728912354,
      "learning_rate": 1.7769514443761525e-05,
      "loss": 1.4304,
      "step": 7258
    },
    {
      "epoch": 0.5577750460971113,
      "grad_norm": 2.704547643661499,
      "learning_rate": 1.7768899815611555e-05,
      "loss": 1.3199,
      "step": 7260
    },
    {
      "epoch": 0.5579287031346035,
      "grad_norm": 3.0023064613342285,
      "learning_rate": 1.7768285187461588e-05,
      "loss": 1.3294,
      "step": 7262
    },
    {
      "epoch": 0.5580823601720959,
      "grad_norm": 2.6803040504455566,
      "learning_rate": 1.7767670559311617e-05,
      "loss": 1.3632,
      "step": 7264
    },
    {
      "epoch": 0.5582360172095882,
      "grad_norm": 3.395805597305298,
      "learning_rate": 1.776705593116165e-05,
      "loss": 1.513,
      "step": 7266
    },
    {
      "epoch": 0.5583896742470805,
      "grad_norm": 3.188375949859619,
      "learning_rate": 1.776644130301168e-05,
      "loss": 1.2284,
      "step": 7268
    },
    {
      "epoch": 0.5585433312845728,
      "grad_norm": 3.19944167137146,
      "learning_rate": 1.776582667486171e-05,
      "loss": 1.5306,
      "step": 7270
    },
    {
      "epoch": 0.5586969883220652,
      "grad_norm": 3.132715940475464,
      "learning_rate": 1.7765212046711743e-05,
      "loss": 1.5381,
      "step": 7272
    },
    {
      "epoch": 0.5588506453595574,
      "grad_norm": 2.777874231338501,
      "learning_rate": 1.7764597418561772e-05,
      "loss": 1.3303,
      "step": 7274
    },
    {
      "epoch": 0.5590043023970498,
      "grad_norm": 3.418839931488037,
      "learning_rate": 1.7763982790411802e-05,
      "loss": 1.3595,
      "step": 7276
    },
    {
      "epoch": 0.559157959434542,
      "grad_norm": 3.0690410137176514,
      "learning_rate": 1.7763368162261835e-05,
      "loss": 1.3164,
      "step": 7278
    },
    {
      "epoch": 0.5593116164720344,
      "grad_norm": 3.341252326965332,
      "learning_rate": 1.776275353411186e-05,
      "loss": 1.4019,
      "step": 7280
    },
    {
      "epoch": 0.5594652735095267,
      "grad_norm": 2.947756052017212,
      "learning_rate": 1.7762138905961895e-05,
      "loss": 1.1799,
      "step": 7282
    },
    {
      "epoch": 0.5596189305470191,
      "grad_norm": 3.1042239665985107,
      "learning_rate": 1.7761524277811924e-05,
      "loss": 1.3689,
      "step": 7284
    },
    {
      "epoch": 0.5597725875845113,
      "grad_norm": 3.0481724739074707,
      "learning_rate": 1.7760909649661957e-05,
      "loss": 1.3483,
      "step": 7286
    },
    {
      "epoch": 0.5599262446220037,
      "grad_norm": 3.041898250579834,
      "learning_rate": 1.7760295021511987e-05,
      "loss": 1.3574,
      "step": 7288
    },
    {
      "epoch": 0.5600799016594961,
      "grad_norm": 2.627877712249756,
      "learning_rate": 1.7759680393362017e-05,
      "loss": 1.3905,
      "step": 7290
    },
    {
      "epoch": 0.5602335586969883,
      "grad_norm": 2.628779649734497,
      "learning_rate": 1.775906576521205e-05,
      "loss": 1.3318,
      "step": 7292
    },
    {
      "epoch": 0.5603872157344807,
      "grad_norm": 2.788853406906128,
      "learning_rate": 1.775845113706208e-05,
      "loss": 1.3654,
      "step": 7294
    },
    {
      "epoch": 0.560540872771973,
      "grad_norm": 2.898160696029663,
      "learning_rate": 1.775783650891211e-05,
      "loss": 1.3316,
      "step": 7296
    },
    {
      "epoch": 0.5606945298094653,
      "grad_norm": 3.0796875953674316,
      "learning_rate": 1.7757221880762142e-05,
      "loss": 1.2504,
      "step": 7298
    },
    {
      "epoch": 0.5608481868469576,
      "grad_norm": 2.4467694759368896,
      "learning_rate": 1.7756607252612172e-05,
      "loss": 1.2919,
      "step": 7300
    }
  ],
  "logging_steps": 2,
  "max_steps": 65080,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 5,
  "save_steps": 100,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 4.6598612105927066e+19,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}