{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 55.945200155460554,
  "eval_steps": 1000,
  "global_step": 72000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0194325689856199,
      "grad_norm": 65.91597747802734,
      "learning_rate": 4.9e-07,
      "loss": 8.2687,
      "step": 50
    },
    {
      "epoch": 0.0388651379712398,
      "grad_norm": 51.2559700012207,
      "learning_rate": 9.9e-07,
      "loss": 7.9936,
      "step": 100
    },
    {
      "epoch": 0.05829770695685969,
      "grad_norm": 23.14748764038086,
      "learning_rate": 1.4900000000000001e-06,
      "loss": 7.527,
      "step": 150
    },
    {
      "epoch": 0.0777302759424796,
      "grad_norm": 22.350494384765625,
      "learning_rate": 1.99e-06,
      "loss": 6.3514,
      "step": 200
    },
    {
      "epoch": 0.0971628449280995,
      "grad_norm": 22.213350296020508,
      "learning_rate": 2.49e-06,
      "loss": 5.6544,
      "step": 250
    },
    {
      "epoch": 0.11659541391371939,
      "grad_norm": 12.773467063903809,
      "learning_rate": 2.99e-06,
      "loss": 4.6271,
      "step": 300
    },
    {
      "epoch": 0.1360279828993393,
      "grad_norm": 9.955669403076172,
      "learning_rate": 3.49e-06,
      "loss": 3.8173,
      "step": 350
    },
    {
      "epoch": 0.1554605518849592,
      "grad_norm": 6.667578220367432,
      "learning_rate": 3.99e-06,
      "loss": 3.4296,
      "step": 400
    },
    {
      "epoch": 0.1748931208705791,
      "grad_norm": 3.926382064819336,
      "learning_rate": 4.49e-06,
      "loss": 3.2801,
      "step": 450
    },
    {
      "epoch": 0.194325689856199,
      "grad_norm": 5.839285850524902,
      "learning_rate": 4.9900000000000005e-06,
      "loss": 3.1784,
      "step": 500
    },
    {
      "epoch": 0.2137582588418189,
      "grad_norm": 5.314439296722412,
      "learning_rate": 5.49e-06,
      "loss": 3.15,
      "step": 550
    },
    {
      "epoch": 0.23319082782743877,
      "grad_norm": 3.1964633464813232,
      "learning_rate": 5.99e-06,
      "loss": 3.0832,
      "step": 600
    },
    {
      "epoch": 0.2526233968130587,
      "grad_norm": 5.283780097961426,
      "learning_rate": 6.4900000000000005e-06,
      "loss": 3.099,
      "step": 650
    },
    {
      "epoch": 0.2720559657986786,
      "grad_norm": 4.881435394287109,
      "learning_rate": 6.990000000000001e-06,
      "loss": 3.068,
      "step": 700
    },
    {
      "epoch": 0.2914885347842985,
      "grad_norm": 6.7736945152282715,
      "learning_rate": 7.4899999999999994e-06,
      "loss": 3.0591,
      "step": 750
    },
    {
      "epoch": 0.3109211037699184,
      "grad_norm": 15.492425918579102,
      "learning_rate": 7.99e-06,
      "loss": 3.0675,
      "step": 800
    },
    {
      "epoch": 0.3303536727555383,
      "grad_norm": 31.496707916259766,
      "learning_rate": 8.49e-06,
      "loss": 3.0561,
      "step": 850
    },
    {
      "epoch": 0.3497862417411582,
      "grad_norm": 3.549668073654175,
      "learning_rate": 8.99e-06,
      "loss": 3.0493,
      "step": 900
    },
    {
      "epoch": 0.3692188107267781,
      "grad_norm": 4.400085926055908,
      "learning_rate": 9.49e-06,
      "loss": 3.0294,
      "step": 950
    },
    {
      "epoch": 0.388651379712398,
      "grad_norm": 7.091118335723877,
      "learning_rate": 9.990000000000001e-06,
      "loss": 3.053,
      "step": 1000
    },
    {
      "epoch": 0.388651379712398,
      "eval_accuracy": 0.0002429189136666181,
      "eval_runtime": 907.7535,
      "eval_samples_per_second": 22.675,
      "eval_steps_per_second": 1.418,
      "step": 1000
    },
    {
      "epoch": 0.4080839486980179,
      "grad_norm": 2.439570903778076,
      "learning_rate": 1.049e-05,
      "loss": 3.0036,
      "step": 1050
    },
    {
      "epoch": 0.4275165176836378,
      "grad_norm": 2.5228285789489746,
      "learning_rate": 1.099e-05,
      "loss": 3.0156,
      "step": 1100
    },
    {
      "epoch": 0.4469490866692577,
      "grad_norm": 3.5071604251861572,
      "learning_rate": 1.149e-05,
      "loss": 3.0285,
      "step": 1150
    },
    {
      "epoch": 0.46638165565487755,
      "grad_norm": 2.174527168273926,
      "learning_rate": 1.199e-05,
      "loss": 3.024,
      "step": 1200
    },
    {
      "epoch": 0.48581422464049745,
      "grad_norm": 2.910525321960449,
      "learning_rate": 1.249e-05,
      "loss": 2.9969,
      "step": 1250
    },
    {
      "epoch": 0.5052467936261174,
      "grad_norm": 2.5316076278686523,
      "learning_rate": 1.299e-05,
      "loss": 3.0125,
      "step": 1300
    },
    {
      "epoch": 0.5246793626117373,
      "grad_norm": 2.175851821899414,
      "learning_rate": 1.349e-05,
      "loss": 2.993,
      "step": 1350
    },
    {
      "epoch": 0.5441119315973572,
      "grad_norm": 2.5506649017333984,
      "learning_rate": 1.399e-05,
      "loss": 3.0089,
      "step": 1400
    },
    {
      "epoch": 0.5635445005829771,
      "grad_norm": 3.2838923931121826,
      "learning_rate": 1.449e-05,
      "loss": 2.9501,
      "step": 1450
    },
    {
      "epoch": 0.582977069568597,
      "grad_norm": 4.193533420562744,
      "learning_rate": 1.499e-05,
      "loss": 2.9575,
      "step": 1500
    },
    {
      "epoch": 0.6024096385542169,
      "grad_norm": 5.010601997375488,
      "learning_rate": 1.5490000000000002e-05,
      "loss": 2.9187,
      "step": 1550
    },
    {
      "epoch": 0.6218422075398368,
      "grad_norm": 4.640109539031982,
      "learning_rate": 1.599e-05,
      "loss": 2.9261,
      "step": 1600
    },
    {
      "epoch": 0.6412747765254566,
      "grad_norm": 4.375769138336182,
      "learning_rate": 1.649e-05,
      "loss": 2.9343,
      "step": 1650
    },
    {
      "epoch": 0.6607073455110766,
      "grad_norm": 3.256545066833496,
      "learning_rate": 1.699e-05,
      "loss": 2.8974,
      "step": 1700
    },
    {
      "epoch": 0.6801399144966964,
      "grad_norm": 8.876006126403809,
      "learning_rate": 1.749e-05,
      "loss": 2.8581,
      "step": 1750
    },
    {
      "epoch": 0.6995724834823164,
      "grad_norm": 5.627965927124023,
      "learning_rate": 1.7990000000000002e-05,
      "loss": 2.8551,
      "step": 1800
    },
    {
      "epoch": 0.7190050524679362,
      "grad_norm": 5.085724830627441,
      "learning_rate": 1.849e-05,
      "loss": 2.8319,
      "step": 1850
    },
    {
      "epoch": 0.7384376214535562,
      "grad_norm": 5.020870685577393,
      "learning_rate": 1.8990000000000003e-05,
      "loss": 2.7724,
      "step": 1900
    },
    {
      "epoch": 0.757870190439176,
      "grad_norm": 7.748109340667725,
      "learning_rate": 1.949e-05,
      "loss": 2.7172,
      "step": 1950
    },
    {
      "epoch": 0.777302759424796,
      "grad_norm": 3.8322463035583496,
      "learning_rate": 1.999e-05,
      "loss": 2.6999,
      "step": 2000
    },
    {
      "epoch": 0.777302759424796,
      "eval_accuracy": 0.0023320215711995335,
      "eval_runtime": 907.4599,
      "eval_samples_per_second": 22.682,
      "eval_steps_per_second": 1.418,
      "step": 2000
    },
    {
      "epoch": 0.7967353284104158,
      "grad_norm": 4.006162643432617,
      "learning_rate": 2.0490000000000002e-05,
      "loss": 2.7107,
      "step": 2050
    },
    {
      "epoch": 0.8161678973960358,
      "grad_norm": 5.7521233558654785,
      "learning_rate": 2.099e-05,
      "loss": 2.6041,
      "step": 2100
    },
    {
      "epoch": 0.8356004663816556,
      "grad_norm": 8.727128028869629,
      "learning_rate": 2.1490000000000003e-05,
      "loss": 2.6262,
      "step": 2150
    },
    {
      "epoch": 0.8550330353672756,
      "grad_norm": 5.179250717163086,
      "learning_rate": 2.199e-05,
      "loss": 2.6203,
      "step": 2200
    },
    {
      "epoch": 0.8744656043528954,
      "grad_norm": 4.94296407699585,
      "learning_rate": 2.249e-05,
      "loss": 2.533,
      "step": 2250
    },
    {
      "epoch": 0.8938981733385154,
      "grad_norm": 5.3395094871521,
      "learning_rate": 2.2990000000000002e-05,
      "loss": 2.6065,
      "step": 2300
    },
    {
      "epoch": 0.9133307423241352,
      "grad_norm": 3.9141852855682373,
      "learning_rate": 2.349e-05,
      "loss": 2.5368,
      "step": 2350
    },
    {
      "epoch": 0.9327633113097551,
      "grad_norm": 4.978011608123779,
      "learning_rate": 2.3990000000000002e-05,
      "loss": 2.5219,
      "step": 2400
    },
    {
      "epoch": 0.952195880295375,
      "grad_norm": 4.247427940368652,
      "learning_rate": 2.449e-05,
      "loss": 2.5095,
      "step": 2450
    },
    {
      "epoch": 0.9716284492809949,
      "grad_norm": 8.596626281738281,
      "learning_rate": 2.4990000000000003e-05,
      "loss": 2.536,
      "step": 2500
    },
    {
      "epoch": 0.9910610182666149,
      "grad_norm": 3.68994402885437,
      "learning_rate": 2.549e-05,
      "loss": 2.5272,
      "step": 2550
    },
    {
      "epoch": 1.0104935872522347,
      "grad_norm": 7.549562454223633,
      "learning_rate": 2.5990000000000004e-05,
      "loss": 2.4804,
      "step": 2600
    },
    {
      "epoch": 1.0299261562378546,
      "grad_norm": 6.586672306060791,
      "learning_rate": 2.6490000000000002e-05,
      "loss": 2.4526,
      "step": 2650
    },
    {
      "epoch": 1.0493587252234746,
      "grad_norm": 6.57544469833374,
      "learning_rate": 2.6989999999999997e-05,
      "loss": 2.4115,
      "step": 2700
    },
    {
      "epoch": 1.0687912942090945,
      "grad_norm": 3.6275458335876465,
      "learning_rate": 2.749e-05,
      "loss": 2.3916,
      "step": 2750
    },
    {
      "epoch": 1.0882238631947143,
      "grad_norm": 5.756579399108887,
      "learning_rate": 2.7989999999999998e-05,
      "loss": 2.4652,
      "step": 2800
    },
    {
      "epoch": 1.1076564321803342,
      "grad_norm": 8.164681434631348,
      "learning_rate": 2.849e-05,
      "loss": 2.4645,
      "step": 2850
    },
    {
      "epoch": 1.1270890011659542,
      "grad_norm": 4.305910110473633,
      "learning_rate": 2.8990000000000002e-05,
      "loss": 2.41,
      "step": 2900
    },
    {
      "epoch": 1.146521570151574,
      "grad_norm": 5.3087286949157715,
      "learning_rate": 2.949e-05,
      "loss": 2.4348,
      "step": 2950
    },
    {
      "epoch": 1.165954139137194,
      "grad_norm": 5.00079345703125,
      "learning_rate": 2.9990000000000003e-05,
      "loss": 2.3594,
      "step": 3000
    },
    {
      "epoch": 1.165954139137194,
      "eval_accuracy": 0.004809794490599038,
      "eval_runtime": 907.572,
      "eval_samples_per_second": 22.679,
      "eval_steps_per_second": 1.418,
      "step": 3000
    },
    {
      "epoch": 1.1853867081228138,
      "grad_norm": 4.513027191162109,
      "learning_rate": 3.049e-05,
      "loss": 2.3504,
      "step": 3050
    },
    {
      "epoch": 1.2048192771084336,
      "grad_norm": 5.349442005157471,
      "learning_rate": 3.099e-05,
      "loss": 2.3989,
      "step": 3100
    },
    {
      "epoch": 1.2242518460940537,
      "grad_norm": 5.596327304840088,
      "learning_rate": 3.1490000000000005e-05,
      "loss": 2.3555,
      "step": 3150
    },
    {
      "epoch": 1.2436844150796735,
      "grad_norm": 3.9062111377716064,
      "learning_rate": 3.1990000000000004e-05,
      "loss": 2.409,
      "step": 3200
    },
    {
      "epoch": 1.2631169840652934,
      "grad_norm": 4.693165302276611,
      "learning_rate": 3.249e-05,
      "loss": 2.3674,
      "step": 3250
    },
    {
      "epoch": 1.2825495530509134,
      "grad_norm": 5.275837421417236,
      "learning_rate": 3.299e-05,
      "loss": 2.3401,
      "step": 3300
    },
    {
      "epoch": 1.3019821220365333,
      "grad_norm": 3.579828977584839,
      "learning_rate": 3.349e-05,
      "loss": 2.3985,
      "step": 3350
    },
    {
      "epoch": 1.3214146910221531,
      "grad_norm": 5.416715145111084,
      "learning_rate": 3.399e-05,
      "loss": 2.3403,
      "step": 3400
    },
    {
      "epoch": 1.340847260007773,
      "grad_norm": 7.789376735687256,
      "learning_rate": 3.449e-05,
      "loss": 2.4155,
      "step": 3450
    },
    {
      "epoch": 1.3602798289933928,
      "grad_norm": 4.4100470542907715,
      "learning_rate": 3.499e-05,
      "loss": 2.3535,
      "step": 3500
    },
    {
      "epoch": 1.379712397979013,
      "grad_norm": 3.998701572418213,
      "learning_rate": 3.549e-05,
      "loss": 2.3238,
      "step": 3550
    },
    {
      "epoch": 1.3991449669646328,
      "grad_norm": 6.838342666625977,
      "learning_rate": 3.599e-05,
      "loss": 2.3312,
      "step": 3600
    },
    {
      "epoch": 1.4185775359502526,
      "grad_norm": 3.762887477874756,
      "learning_rate": 3.6490000000000005e-05,
      "loss": 2.2789,
      "step": 3650
    },
    {
      "epoch": 1.4380101049358724,
      "grad_norm": 7.700969219207764,
      "learning_rate": 3.699e-05,
      "loss": 2.3047,
      "step": 3700
    },
    {
      "epoch": 1.4574426739214923,
      "grad_norm": 4.673271179199219,
      "learning_rate": 3.749e-05,
      "loss": 2.3218,
      "step": 3750
    },
    {
      "epoch": 1.4768752429071124,
      "grad_norm": 4.929595947265625,
      "learning_rate": 3.799e-05,
      "loss": 2.2811,
      "step": 3800
    },
    {
      "epoch": 1.4963078118927322,
      "grad_norm": 6.298782825469971,
      "learning_rate": 3.8490000000000006e-05,
      "loss": 2.2838,
      "step": 3850
    },
    {
      "epoch": 1.515740380878352,
      "grad_norm": 7.514346122741699,
      "learning_rate": 3.8990000000000004e-05,
      "loss": 2.2797,
      "step": 3900
    },
    {
      "epoch": 1.5351729498639721,
      "grad_norm": 4.082162857055664,
      "learning_rate": 3.9489999999999996e-05,
      "loss": 2.3473,
      "step": 3950
    },
    {
      "epoch": 1.5546055188495917,
      "grad_norm": 3.844625473022461,
      "learning_rate": 3.999e-05,
      "loss": 2.2859,
      "step": 4000
    },
    {
      "epoch": 1.5546055188495917,
      "eval_accuracy": 0.005198464752465627,
      "eval_runtime": 907.1135,
      "eval_samples_per_second": 22.691,
      "eval_steps_per_second": 1.419,
      "step": 4000
    },
    {
      "epoch": 1.5740380878352118,
      "grad_norm": 3.8445560932159424,
      "learning_rate": 4.049e-05,
      "loss": 2.3033,
      "step": 4050
    },
    {
      "epoch": 1.5934706568208317,
      "grad_norm": 4.017306327819824,
      "learning_rate": 4.099e-05,
      "loss": 2.2842,
      "step": 4100
    },
    {
      "epoch": 1.6129032258064515,
      "grad_norm": 3.259737968444824,
      "learning_rate": 4.1490000000000004e-05,
      "loss": 2.2398,
      "step": 4150
    },
    {
      "epoch": 1.6323357947920716,
      "grad_norm": 3.228858470916748,
      "learning_rate": 4.199e-05,
      "loss": 2.2501,
      "step": 4200
    },
    {
      "epoch": 1.6517683637776914,
      "grad_norm": 9.149596214294434,
      "learning_rate": 4.249e-05,
      "loss": 2.3207,
      "step": 4250
    },
    {
      "epoch": 1.6712009327633113,
      "grad_norm": 5.323018550872803,
      "learning_rate": 4.299e-05,
      "loss": 2.2831,
      "step": 4300
    },
    {
      "epoch": 1.6906335017489313,
      "grad_norm": 2.7300260066986084,
      "learning_rate": 4.3490000000000005e-05,
      "loss": 2.2799,
      "step": 4350
    },
    {
      "epoch": 1.710066070734551,
      "grad_norm": 2.534250259399414,
      "learning_rate": 4.3990000000000004e-05,
      "loss": 2.2916,
      "step": 4400
    },
    {
      "epoch": 1.729498639720171,
      "grad_norm": 4.524829387664795,
      "learning_rate": 4.449e-05,
      "loss": 2.2706,
      "step": 4450
    },
    {
      "epoch": 1.7489312087057909,
      "grad_norm": 5.60699987411499,
      "learning_rate": 4.499e-05,
      "loss": 2.2751,
      "step": 4500
    },
    {
      "epoch": 1.7683637776914107,
      "grad_norm": 4.741508960723877,
      "learning_rate": 4.549000000000001e-05,
      "loss": 2.2387,
      "step": 4550
    },
    {
      "epoch": 1.7877963466770308,
      "grad_norm": 4.468010902404785,
      "learning_rate": 4.599e-05,
      "loss": 2.2431,
      "step": 4600
    },
    {
      "epoch": 1.8072289156626506,
      "grad_norm": 3.4687862396240234,
      "learning_rate": 4.649e-05,
      "loss": 2.2534,
      "step": 4650
    },
    {
      "epoch": 1.8266614846482705,
      "grad_norm": 3.7992916107177734,
      "learning_rate": 4.699e-05,
      "loss": 2.2387,
      "step": 4700
    },
    {
      "epoch": 1.8460940536338906,
      "grad_norm": 4.4966936111450195,
      "learning_rate": 4.749e-05,
      "loss": 2.2469,
      "step": 4750
    },
    {
      "epoch": 1.8655266226195102,
      "grad_norm": 3.7579238414764404,
      "learning_rate": 4.799e-05,
      "loss": 2.217,
      "step": 4800
    },
    {
      "epoch": 1.8849591916051303,
      "grad_norm": 5.868619441986084,
      "learning_rate": 4.8490000000000005e-05,
      "loss": 2.281,
      "step": 4850
    },
    {
      "epoch": 1.90439176059075,
      "grad_norm": 3.1073763370513916,
      "learning_rate": 4.8990000000000004e-05,
      "loss": 2.2435,
      "step": 4900
    },
    {
      "epoch": 1.92382432957637,
      "grad_norm": 5.91089391708374,
      "learning_rate": 4.949e-05,
      "loss": 2.228,
      "step": 4950
    },
    {
      "epoch": 1.94325689856199,
      "grad_norm": 4.082432270050049,
      "learning_rate": 4.999e-05,
      "loss": 2.1793,
      "step": 5000
    },
    {
      "epoch": 1.94325689856199,
      "eval_accuracy": 0.007724821454598455,
      "eval_runtime": 907.4609,
      "eval_samples_per_second": 22.682,
      "eval_steps_per_second": 1.418,
      "step": 5000
    },
    {
      "epoch": 1.9626894675476096,
      "grad_norm": 3.8252387046813965,
      "learning_rate": 5.0490000000000006e-05,
      "loss": 2.2716,
      "step": 5050
    },
    {
      "epoch": 1.9821220365332297,
      "grad_norm": 5.004796504974365,
      "learning_rate": 5.0990000000000005e-05,
      "loss": 2.2269,
      "step": 5100
    },
    {
      "epoch": 2.00155460551885,
      "grad_norm": 3.568448066711426,
      "learning_rate": 5.149e-05,
      "loss": 2.2113,
      "step": 5150
    },
    {
      "epoch": 2.0209871745044694,
      "grad_norm": 3.9365673065185547,
      "learning_rate": 5.199000000000001e-05,
      "loss": 2.1322,
      "step": 5200
    },
    {
      "epoch": 2.0404197434900895,
      "grad_norm": 4.7153401374816895,
      "learning_rate": 5.249000000000001e-05,
      "loss": 2.1687,
      "step": 5250
    },
    {
      "epoch": 2.059852312475709,
      "grad_norm": 3.279733896255493,
      "learning_rate": 5.2990000000000006e-05,
      "loss": 2.1613,
      "step": 5300
    },
    {
      "epoch": 2.079284881461329,
      "grad_norm": 8.057929039001465,
      "learning_rate": 5.3490000000000005e-05,
      "loss": 2.152,
      "step": 5350
    },
    {
      "epoch": 2.0987174504469492,
      "grad_norm": 2.202526330947876,
      "learning_rate": 5.399000000000001e-05,
      "loss": 2.1066,
      "step": 5400
    },
    {
      "epoch": 2.118150019432569,
      "grad_norm": 5.828799247741699,
      "learning_rate": 5.449000000000001e-05,
      "loss": 2.152,
      "step": 5450
    },
    {
      "epoch": 2.137582588418189,
      "grad_norm": 4.265638828277588,
      "learning_rate": 5.499000000000001e-05,
      "loss": 2.1494,
      "step": 5500
    },
    {
      "epoch": 2.1570151574038086,
      "grad_norm": 3.351078748703003,
      "learning_rate": 5.549e-05,
      "loss": 2.1429,
      "step": 5550
    },
    {
      "epoch": 2.1764477263894286,
      "grad_norm": 3.50518798828125,
      "learning_rate": 5.599e-05,
      "loss": 2.0642,
      "step": 5600
    },
    {
      "epoch": 2.1958802953750487,
      "grad_norm": 3.6676487922668457,
      "learning_rate": 5.6489999999999996e-05,
      "loss": 2.1673,
      "step": 5650
    },
    {
      "epoch": 2.2153128643606683,
      "grad_norm": 13.605622291564941,
      "learning_rate": 5.699e-05,
      "loss": 2.187,
      "step": 5700
    },
    {
      "epoch": 2.2347454333462884,
      "grad_norm": 2.574394702911377,
      "learning_rate": 5.749e-05,
      "loss": 2.1414,
      "step": 5750
    },
    {
      "epoch": 2.2541780023319085,
      "grad_norm": 2.6154654026031494,
      "learning_rate": 5.799e-05,
      "loss": 2.1965,
      "step": 5800
    },
    {
      "epoch": 2.273610571317528,
      "grad_norm": 4.553177356719971,
      "learning_rate": 5.849e-05,
      "loss": 2.1489,
      "step": 5850
    },
    {
      "epoch": 2.293043140303148,
      "grad_norm": 5.114558696746826,
      "learning_rate": 5.899e-05,
      "loss": 2.1779,
      "step": 5900
    },
    {
      "epoch": 2.312475709288768,
      "grad_norm": 4.845452785491943,
      "learning_rate": 5.949e-05,
      "loss": 2.1091,
      "step": 5950
    },
    {
      "epoch": 2.331908278274388,
      "grad_norm": 5.508037090301514,
      "learning_rate": 5.999e-05,
      "loss": 2.1407,
      "step": 6000
    },
    {
      "epoch": 2.331908278274388,
      "eval_accuracy": 0.011805859204197638,
      "eval_runtime": 907.658,
      "eval_samples_per_second": 22.677,
      "eval_steps_per_second": 1.418,
      "step": 6000
    },
    {
      "epoch": 2.351340847260008,
      "grad_norm": 3.4910573959350586,
      "learning_rate": 6.0490000000000005e-05,
      "loss": 2.0972,
      "step": 6050
    },
    {
      "epoch": 2.3707734162456275,
      "grad_norm": 11.334199905395508,
      "learning_rate": 6.0990000000000004e-05,
      "loss": 2.1085,
      "step": 6100
    },
    {
      "epoch": 2.3902059852312476,
      "grad_norm": 4.658063888549805,
      "learning_rate": 6.149000000000001e-05,
      "loss": 2.161,
      "step": 6150
    },
    {
      "epoch": 2.4096385542168672,
      "grad_norm": 4.726357460021973,
      "learning_rate": 6.199000000000001e-05,
      "loss": 2.0757,
      "step": 6200
    },
    {
      "epoch": 2.4290711232024873,
      "grad_norm": 7.879475116729736,
      "learning_rate": 6.249e-05,
      "loss": 2.1643,
      "step": 6250
    },
    {
      "epoch": 2.4485036921881074,
      "grad_norm": 4.898128986358643,
      "learning_rate": 6.299e-05,
      "loss": 2.1573,
      "step": 6300
    },
    {
      "epoch": 2.467936261173727,
      "grad_norm": 4.784426212310791,
      "learning_rate": 6.349e-05,
      "loss": 2.1296,
      "step": 6350
    },
    {
      "epoch": 2.487368830159347,
      "grad_norm": 4.347411155700684,
      "learning_rate": 6.399e-05,
      "loss": 2.1221,
      "step": 6400
    },
    {
      "epoch": 2.506801399144967,
      "grad_norm": 9.684266090393066,
      "learning_rate": 6.449e-05,
      "loss": 2.0595,
      "step": 6450
    },
    {
      "epoch": 2.5262339681305868,
      "grad_norm": 3.320681095123291,
      "learning_rate": 6.499000000000001e-05,
      "loss": 2.1178,
      "step": 6500
    },
    {
      "epoch": 2.545666537116207,
      "grad_norm": 5.14285945892334,
      "learning_rate": 6.549000000000001e-05,
      "loss": 2.1096,
      "step": 6550
    },
    {
      "epoch": 2.565099106101827,
      "grad_norm": 4.604893684387207,
      "learning_rate": 6.599000000000001e-05,
      "loss": 2.0953,
      "step": 6600
    },
    {
      "epoch": 2.5845316750874465,
      "grad_norm": 3.0445802211761475,
      "learning_rate": 6.649000000000001e-05,
      "loss": 2.1309,
      "step": 6650
    },
    {
      "epoch": 2.6039642440730666,
      "grad_norm": 5.163757801055908,
      "learning_rate": 6.699000000000001e-05,
      "loss": 2.1532,
      "step": 6700
    },
    {
      "epoch": 2.623396813058686,
      "grad_norm": 3.930135488510132,
      "learning_rate": 6.749e-05,
      "loss": 2.0524,
      "step": 6750
    },
    {
      "epoch": 2.6428293820443063,
      "grad_norm": 2.7163515090942383,
      "learning_rate": 6.799e-05,
      "loss": 2.1163,
      "step": 6800
    },
    {
      "epoch": 2.662261951029926,
      "grad_norm": 10.727341651916504,
      "learning_rate": 6.849e-05,
      "loss": 2.159,
      "step": 6850
    },
    {
      "epoch": 2.681694520015546,
      "grad_norm": 3.4097917079925537,
      "learning_rate": 6.899e-05,
      "loss": 2.093,
      "step": 6900
    },
    {
      "epoch": 2.701127089001166,
      "grad_norm": 3.5827713012695312,
      "learning_rate": 6.949e-05,
      "loss": 2.1316,
      "step": 6950
    },
    {
      "epoch": 2.7205596579867857,
      "grad_norm": 2.8892476558685303,
      "learning_rate": 6.999e-05,
      "loss": 2.0978,
      "step": 7000
    },
    {
      "epoch": 2.7205596579867857,
      "eval_accuracy": 0.018413253655929652,
      "eval_runtime": 906.6344,
      "eval_samples_per_second": 22.703,
      "eval_steps_per_second": 1.42,
      "step": 7000
    },
    {
      "epoch": 2.7399922269724057,
      "grad_norm": 3.845646619796753,
      "learning_rate": 7.049e-05,
      "loss": 2.0751,
      "step": 7050
    },
    {
      "epoch": 2.759424795958026,
      "grad_norm": 2.9022414684295654,
      "learning_rate": 7.099e-05,
      "loss": 2.112,
      "step": 7100
    },
    {
      "epoch": 2.7788573649436454,
      "grad_norm": 2.3950488567352295,
      "learning_rate": 7.149e-05,
      "loss": 2.0933,
      "step": 7150
    },
    {
      "epoch": 2.7982899339292655,
      "grad_norm": 4.62535285949707,
      "learning_rate": 7.199000000000001e-05,
      "loss": 2.1066,
      "step": 7200
    },
    {
      "epoch": 2.8177225029148856,
      "grad_norm": 1.856604814529419,
      "learning_rate": 7.249e-05,
      "loss": 2.1142,
      "step": 7250
    },
    {
      "epoch": 2.837155071900505,
      "grad_norm": 5.750202655792236,
      "learning_rate": 7.299e-05,
      "loss": 2.1047,
      "step": 7300
    },
    {
      "epoch": 2.8565876408861253,
      "grad_norm": 3.7773256301879883,
      "learning_rate": 7.349e-05,
      "loss": 2.0695,
      "step": 7350
    },
    {
      "epoch": 2.876020209871745,
      "grad_norm": 3.556065082550049,
      "learning_rate": 7.399e-05,
      "loss": 2.1637,
      "step": 7400
    },
    {
      "epoch": 2.895452778857365,
      "grad_norm": 6.621519565582275,
      "learning_rate": 7.449e-05,
      "loss": 2.0937,
      "step": 7450
    },
    {
      "epoch": 2.9148853478429846,
      "grad_norm": 7.399990558624268,
      "learning_rate": 7.499e-05,
      "loss": 2.1091,
      "step": 7500
    },
    {
      "epoch": 2.9343179168286047,
      "grad_norm": 5.250476360321045,
      "learning_rate": 7.549000000000001e-05,
      "loss": 2.0978,
      "step": 7550
    },
    {
      "epoch": 2.9537504858142247,
      "grad_norm": 2.741161823272705,
      "learning_rate": 7.599000000000001e-05,
      "loss": 2.093,
      "step": 7600
    },
    {
      "epoch": 2.9731830547998443,
      "grad_norm": 2.7134132385253906,
      "learning_rate": 7.649000000000001e-05,
      "loss": 2.0529,
      "step": 7650
    },
    {
      "epoch": 2.9926156237854644,
      "grad_norm": 4.2654194831848145,
      "learning_rate": 7.699e-05,
      "loss": 2.0865,
      "step": 7700
    },
    {
      "epoch": 3.0120481927710845,
      "grad_norm": 4.368113994598389,
      "learning_rate": 7.749e-05,
      "loss": 2.0073,
      "step": 7750
    },
    {
      "epoch": 3.031480761756704,
      "grad_norm": 3.039257526397705,
      "learning_rate": 7.799e-05,
      "loss": 1.981,
      "step": 7800
    },
    {
      "epoch": 3.050913330742324,
      "grad_norm": 2.098691463470459,
      "learning_rate": 7.849e-05,
      "loss": 1.9836,
      "step": 7850
    },
    {
      "epoch": 3.0703458997279442,
      "grad_norm": 3.731783151626587,
      "learning_rate": 7.899000000000001e-05,
      "loss": 1.978,
      "step": 7900
    },
    {
      "epoch": 3.089778468713564,
      "grad_norm": 2.487757682800293,
      "learning_rate": 7.949000000000001e-05,
      "loss": 1.9576,
      "step": 7950
    },
    {
      "epoch": 3.109211037699184,
      "grad_norm": 7.170228481292725,
      "learning_rate": 7.999000000000001e-05,
      "loss": 1.9401,
      "step": 8000
    },
    {
      "epoch": 3.109211037699184,
      "eval_accuracy": 0.036243501919059416,
      "eval_runtime": 907.4435,
      "eval_samples_per_second": 22.682,
      "eval_steps_per_second": 1.418,
      "step": 8000
    },
    {
      "epoch": 3.1286436066848036,
      "grad_norm": 3.2367658615112305,
      "learning_rate": 8.049e-05,
      "loss": 1.9768,
      "step": 8050
    },
    {
      "epoch": 3.1480761756704236,
      "grad_norm": 11.398568153381348,
      "learning_rate": 8.099e-05,
      "loss": 1.9919,
      "step": 8100
    },
    {
      "epoch": 3.1675087446560437,
      "grad_norm": 4.276248931884766,
      "learning_rate": 8.149e-05,
      "loss": 1.9771,
      "step": 8150
    },
    {
      "epoch": 3.1869413136416633,
      "grad_norm": 4.85842752456665,
      "learning_rate": 8.199e-05,
      "loss": 1.9818,
      "step": 8200
    },
    {
      "epoch": 3.2063738826272834,
      "grad_norm": 2.933837652206421,
      "learning_rate": 8.249e-05,
      "loss": 1.9883,
      "step": 8250
    },
    {
      "epoch": 3.225806451612903,
      "grad_norm": 4.235050678253174,
      "learning_rate": 8.299e-05,
      "loss": 1.9446,
      "step": 8300
    },
    {
      "epoch": 3.245239020598523,
      "grad_norm": 2.48342227935791,
      "learning_rate": 8.349e-05,
      "loss": 1.9867,
      "step": 8350
    },
    {
      "epoch": 3.264671589584143,
      "grad_norm": 6.0085530281066895,
      "learning_rate": 8.399e-05,
      "loss": 1.9291,
      "step": 8400
    },
    {
      "epoch": 3.284104158569763,
      "grad_norm": 3.1313884258270264,
      "learning_rate": 8.449e-05,
      "loss": 1.9675,
      "step": 8450
    },
    {
      "epoch": 3.303536727555383,
      "grad_norm": 5.40889835357666,
      "learning_rate": 8.499e-05,
      "loss": 1.9775,
      "step": 8500
    },
    {
      "epoch": 3.322969296541003,
      "grad_norm": 2.2098782062530518,
      "learning_rate": 8.549000000000001e-05,
      "loss": 1.9164,
      "step": 8550
    },
    {
      "epoch": 3.3424018655266226,
      "grad_norm": 3.1617531776428223,
      "learning_rate": 8.599000000000001e-05,
      "loss": 1.9841,
      "step": 8600
    },
    {
      "epoch": 3.3618344345122426,
      "grad_norm": 5.392375946044922,
      "learning_rate": 8.649000000000001e-05,
      "loss": 1.9923,
      "step": 8650
    },
    {
      "epoch": 3.3812670034978622,
      "grad_norm": 2.839928150177002,
      "learning_rate": 8.699e-05,
      "loss": 2.0026,
      "step": 8700
    },
    {
      "epoch": 3.4006995724834823,
      "grad_norm": 4.040922164916992,
      "learning_rate": 8.749e-05,
      "loss": 1.9445,
      "step": 8750
    },
    {
      "epoch": 3.4201321414691024,
      "grad_norm": 8.354971885681152,
      "learning_rate": 8.799e-05,
      "loss": 1.9008,
      "step": 8800
    },
    {
      "epoch": 3.439564710454722,
      "grad_norm": 3.485241174697876,
      "learning_rate": 8.849e-05,
      "loss": 1.9589,
      "step": 8850
    },
    {
      "epoch": 3.458997279440342,
      "grad_norm": 31.03635597229004,
      "learning_rate": 8.899e-05,
      "loss": 1.9488,
      "step": 8900
    },
    {
      "epoch": 3.4784298484259617,
      "grad_norm": 2.6878533363342285,
      "learning_rate": 8.949000000000001e-05,
      "loss": 1.9127,
      "step": 8950
    },
    {
      "epoch": 3.4978624174115818,
      "grad_norm": 2.712730646133423,
      "learning_rate": 8.999000000000001e-05,
      "loss": 1.907,
      "step": 9000
    },
    {
      "epoch": 3.4978624174115818,
      "eval_accuracy": 0.04727202059952388,
      "eval_runtime": 908.1635,
      "eval_samples_per_second": 22.664,
      "eval_steps_per_second": 1.417,
      "step": 9000
    },
    {
      "epoch": 3.517294986397202,
      "grad_norm": 2.0906474590301514,
      "learning_rate": 9.049000000000001e-05,
      "loss": 1.9679,
      "step": 9050
    },
    {
      "epoch": 3.5367275553828215,
      "grad_norm": 4.546792507171631,
      "learning_rate": 9.099000000000001e-05,
      "loss": 1.9406,
      "step": 9100
    },
    {
      "epoch": 3.5561601243684415,
      "grad_norm": 5.863422870635986,
      "learning_rate": 9.149e-05,
      "loss": 1.9497,
      "step": 9150
    },
    {
      "epoch": 3.5755926933540616,
      "grad_norm": 3.8730051517486572,
      "learning_rate": 9.199e-05,
      "loss": 1.949,
      "step": 9200
    },
    {
      "epoch": 3.5950252623396812,
      "grad_norm": 8.266348838806152,
      "learning_rate": 9.249e-05,
      "loss": 1.9245,
      "step": 9250
    },
    {
      "epoch": 3.6144578313253013,
      "grad_norm": 3.6021506786346436,
      "learning_rate": 9.299e-05,
      "loss": 1.9254,
      "step": 9300
    },
    {
      "epoch": 3.6338904003109214,
      "grad_norm": 3.5761210918426514,
      "learning_rate": 9.349e-05,
      "loss": 1.9584,
      "step": 9350
    },
    {
      "epoch": 3.653322969296541,
      "grad_norm": 3.0531485080718994,
      "learning_rate": 9.399e-05,
      "loss": 1.9022,
      "step": 9400
    },
    {
      "epoch": 3.672755538282161,
      "grad_norm": 4.91207218170166,
      "learning_rate": 9.449e-05,
      "loss": 1.9076,
      "step": 9450
    },
    {
      "epoch": 3.6921881072677807,
      "grad_norm": 3.6810998916625977,
      "learning_rate": 9.499e-05,
      "loss": 1.9406,
      "step": 9500
    },
    {
      "epoch": 3.7116206762534008,
      "grad_norm": 4.888444423675537,
      "learning_rate": 9.549e-05,
      "loss": 1.9407,
      "step": 9550
    },
    {
      "epoch": 3.7310532452390204,
      "grad_norm": 3.0702097415924072,
      "learning_rate": 9.599000000000001e-05,
      "loss": 1.9525,
      "step": 9600
    },
    {
      "epoch": 3.7504858142246404,
      "grad_norm": 3.8583812713623047,
      "learning_rate": 9.649e-05,
      "loss": 1.9311,
      "step": 9650
    },
    {
      "epoch": 3.7699183832102605,
      "grad_norm": 6.406527042388916,
      "learning_rate": 9.699e-05,
      "loss": 1.911,
      "step": 9700
    },
    {
      "epoch": 3.78935095219588,
      "grad_norm": 3.266839027404785,
      "learning_rate": 9.749e-05,
      "loss": 1.922,
      "step": 9750
    },
    {
      "epoch": 3.8087835211815,
      "grad_norm": 3.8981516361236572,
      "learning_rate": 9.799e-05,
      "loss": 1.9372,
      "step": 9800
    },
    {
      "epoch": 3.8282160901671203,
      "grad_norm": 2.969425916671753,
      "learning_rate": 9.849e-05,
      "loss": 1.9577,
      "step": 9850
    },
    {
      "epoch": 3.84764865915274,
      "grad_norm": 3.436954975128174,
      "learning_rate": 9.899e-05,
      "loss": 1.957,
      "step": 9900
    },
    {
      "epoch": 3.86708122813836,
      "grad_norm": 7.24133825302124,
      "learning_rate": 9.949000000000001e-05,
      "loss": 1.8949,
      "step": 9950
    },
    {
      "epoch": 3.88651379712398,
      "grad_norm": 3.6599984169006348,
      "learning_rate": 9.999000000000001e-05,
      "loss": 1.9428,
      "step": 10000
    },
    {
      "epoch": 3.88651379712398,
      "eval_accuracy": 0.07073798765971918,
      "eval_runtime": 907.1728,
      "eval_samples_per_second": 22.689,
      "eval_steps_per_second": 1.419,
      "step": 10000
    },
    {
      "epoch": 3.9059463661095997,
      "grad_norm": 3.687016010284424,
      "learning_rate": 9.99742105263158e-05,
      "loss": 1.9132,
      "step": 10050
    },
    {
      "epoch": 3.9253789350952197,
      "grad_norm": 4.09041166305542,
      "learning_rate": 9.994789473684211e-05,
      "loss": 1.9465,
      "step": 10100
    },
    {
      "epoch": 3.9448115040808394,
      "grad_norm": 3.4802134037017822,
      "learning_rate": 9.992157894736842e-05,
      "loss": 1.9263,
      "step": 10150
    },
    {
      "epoch": 3.9642440730664594,
      "grad_norm": 2.9187519550323486,
      "learning_rate": 9.989526315789473e-05,
      "loss": 1.954,
      "step": 10200
    },
    {
      "epoch": 3.983676642052079,
      "grad_norm": 4.464130878448486,
      "learning_rate": 9.986894736842106e-05,
      "loss": 1.8683,
      "step": 10250
    },
    {
      "epoch": 4.0031092110377,
      "grad_norm": 57.50473403930664,
      "learning_rate": 9.984263157894738e-05,
      "loss": 1.9317,
      "step": 10300
    },
    {
      "epoch": 4.022541780023319,
      "grad_norm": 3.390007734298706,
      "learning_rate": 9.98163157894737e-05,
      "loss": 1.7154,
      "step": 10350
    },
    {
      "epoch": 4.041974349008939,
      "grad_norm": 4.305722236633301,
      "learning_rate": 9.979e-05,
      "loss": 1.762,
      "step": 10400
    },
    {
      "epoch": 4.061406917994558,
      "grad_norm": 3.997655153274536,
      "learning_rate": 9.976368421052632e-05,
      "loss": 1.7644,
      "step": 10450
    },
    {
      "epoch": 4.080839486980179,
      "grad_norm": 4.27736759185791,
      "learning_rate": 9.973736842105263e-05,
      "loss": 1.7189,
      "step": 10500
    },
    {
      "epoch": 4.100272055965799,
      "grad_norm": 16.003461837768555,
      "learning_rate": 9.971105263157895e-05,
      "loss": 1.7184,
      "step": 10550
    },
    {
      "epoch": 4.119704624951418,
      "grad_norm": 4.10724401473999,
      "learning_rate": 9.968473684210526e-05,
      "loss": 1.7408,
      "step": 10600
    },
    {
      "epoch": 4.139137193937039,
      "grad_norm": 4.501052379608154,
      "learning_rate": 9.965842105263158e-05,
      "loss": 1.7309,
      "step": 10650
    },
    {
      "epoch": 4.158569762922658,
      "grad_norm": 4.458967208862305,
      "learning_rate": 9.96321052631579e-05,
      "loss": 1.7234,
      "step": 10700
    },
    {
      "epoch": 4.178002331908278,
      "grad_norm": 7.051727771759033,
      "learning_rate": 9.960578947368421e-05,
      "loss": 1.711,
      "step": 10750
    },
    {
      "epoch": 4.1974349008938985,
      "grad_norm": 4.20425271987915,
      "learning_rate": 9.957947368421054e-05,
      "loss": 1.7405,
      "step": 10800
    },
    {
      "epoch": 4.216867469879518,
      "grad_norm": 4.120607852935791,
      "learning_rate": 9.955315789473685e-05,
      "loss": 1.7183,
      "step": 10850
    },
    {
      "epoch": 4.236300038865138,
      "grad_norm": 3.240799903869629,
      "learning_rate": 9.952684210526316e-05,
      "loss": 1.7711,
      "step": 10900
    },
    {
      "epoch": 4.255732607850758,
      "grad_norm": 5.256502628326416,
      "learning_rate": 9.950052631578947e-05,
      "loss": 1.7256,
      "step": 10950
    },
    {
      "epoch": 4.275165176836378,
      "grad_norm": 6.0444560050964355,
      "learning_rate": 9.94742105263158e-05,
      "loss": 1.6995,
      "step": 11000
    },
    {
      "epoch": 4.275165176836378,
      "eval_accuracy": 0.13307098090657338,
      "eval_runtime": 906.647,
      "eval_samples_per_second": 22.702,
      "eval_steps_per_second": 1.42,
      "step": 11000
    },
    {
      "epoch": 4.2945977458219975,
      "grad_norm": 3.814336061477661,
      "learning_rate": 9.944789473684211e-05,
      "loss": 1.7494,
      "step": 11050
    },
    {
      "epoch": 4.314030314807617,
      "grad_norm": 5.616982460021973,
      "learning_rate": 9.942157894736842e-05,
      "loss": 1.7516,
      "step": 11100
    },
    {
      "epoch": 4.333462883793238,
      "grad_norm": 4.849693775177002,
      "learning_rate": 9.939526315789475e-05,
      "loss": 1.726,
      "step": 11150
    },
    {
      "epoch": 4.352895452778857,
      "grad_norm": 4.43636417388916,
      "learning_rate": 9.936894736842106e-05,
      "loss": 1.7132,
      "step": 11200
    },
    {
      "epoch": 4.372328021764477,
      "grad_norm": 4.7518439292907715,
      "learning_rate": 9.934263157894737e-05,
      "loss": 1.7164,
      "step": 11250
    },
    {
      "epoch": 4.391760590750097,
      "grad_norm": 5.752714157104492,
      "learning_rate": 9.93163157894737e-05,
      "loss": 1.7357,
      "step": 11300
    },
    {
      "epoch": 4.411193159735717,
      "grad_norm": 5.446049690246582,
      "learning_rate": 9.929e-05,
      "loss": 1.7636,
      "step": 11350
    },
    {
      "epoch": 4.430625728721337,
      "grad_norm": 5.5599775314331055,
      "learning_rate": 9.926368421052632e-05,
      "loss": 1.7316,
      "step": 11400
    },
    {
      "epoch": 4.450058297706957,
      "grad_norm": 4.994411945343018,
      "learning_rate": 9.923736842105263e-05,
      "loss": 1.6766,
      "step": 11450
    },
    {
      "epoch": 4.469490866692577,
      "grad_norm": 6.706633567810059,
      "learning_rate": 9.921105263157895e-05,
      "loss": 1.6803,
      "step": 11500
    },
    {
      "epoch": 4.488923435678196,
      "grad_norm": 5.894743919372559,
      "learning_rate": 9.918473684210527e-05,
      "loss": 1.7184,
      "step": 11550
    },
    {
      "epoch": 4.508356004663817,
      "grad_norm": 3.859015941619873,
      "learning_rate": 9.915842105263158e-05,
      "loss": 1.7238,
      "step": 11600
    },
    {
      "epoch": 4.5277885736494365,
      "grad_norm": 5.355344772338867,
      "learning_rate": 9.91321052631579e-05,
      "loss": 1.7641,
      "step": 11650
    },
    {
      "epoch": 4.547221142635056,
      "grad_norm": 9.785111427307129,
      "learning_rate": 9.910578947368421e-05,
      "loss": 1.705,
      "step": 11700
    },
    {
      "epoch": 4.566653711620676,
      "grad_norm": 3.365846633911133,
      "learning_rate": 9.907947368421054e-05,
      "loss": 1.6986,
      "step": 11750
    },
    {
      "epoch": 4.586086280606296,
      "grad_norm": 3.3706541061401367,
      "learning_rate": 9.905315789473685e-05,
      "loss": 1.7428,
      "step": 11800
    },
    {
      "epoch": 4.605518849591916,
      "grad_norm": 4.395911693572998,
      "learning_rate": 9.902684210526316e-05,
      "loss": 1.7401,
      "step": 11850
    },
    {
      "epoch": 4.624951418577536,
      "grad_norm": 3.5814549922943115,
      "learning_rate": 9.900052631578947e-05,
      "loss": 1.7055,
      "step": 11900
    },
    {
      "epoch": 4.644383987563156,
      "grad_norm": 6.163331031799316,
      "learning_rate": 9.897421052631579e-05,
      "loss": 1.7195,
      "step": 11950
    },
    {
      "epoch": 4.663816556548776,
      "grad_norm": 7.837599277496338,
      "learning_rate": 9.894789473684211e-05,
      "loss": 1.6419,
      "step": 12000
    },
    {
      "epoch": 4.663816556548776,
      "eval_accuracy": 0.1958412281980275,
      "eval_runtime": 912.2477,
      "eval_samples_per_second": 22.563,
      "eval_steps_per_second": 1.411,
      "step": 12000
    },
    {
      "epoch": 4.683249125534395,
      "grad_norm": 4.789027690887451,
      "learning_rate": 9.892157894736842e-05,
      "loss": 1.7383,
      "step": 12050
    },
    {
      "epoch": 4.702681694520016,
      "grad_norm": 3.7628960609436035,
      "learning_rate": 9.889526315789475e-05,
      "loss": 1.6448,
      "step": 12100
    },
    {
      "epoch": 4.7221142635056355,
      "grad_norm": 4.155508041381836,
      "learning_rate": 9.886894736842106e-05,
      "loss": 1.6326,
      "step": 12150
    },
    {
      "epoch": 4.741546832491255,
      "grad_norm": 4.6860785484313965,
      "learning_rate": 9.884263157894737e-05,
      "loss": 1.669,
      "step": 12200
    },
    {
      "epoch": 4.760979401476876,
      "grad_norm": 9.118622779846191,
      "learning_rate": 9.88163157894737e-05,
      "loss": 1.7108,
      "step": 12250
    },
    {
      "epoch": 4.780411970462495,
      "grad_norm": 5.970766067504883,
      "learning_rate": 9.879000000000001e-05,
      "loss": 1.6809,
      "step": 12300
    },
    {
      "epoch": 4.799844539448115,
      "grad_norm": 6.902002811431885,
      "learning_rate": 9.876368421052632e-05,
      "loss": 1.7195,
      "step": 12350
    },
    {
      "epoch": 4.8192771084337345,
      "grad_norm": 8.516134262084961,
      "learning_rate": 9.873736842105263e-05,
      "loss": 1.6597,
      "step": 12400
    },
    {
      "epoch": 4.838709677419355,
      "grad_norm": 5.790510654449463,
      "learning_rate": 9.871105263157894e-05,
      "loss": 1.6695,
      "step": 12450
    },
    {
      "epoch": 4.858142246404975,
      "grad_norm": 7.634342193603516,
      "learning_rate": 9.868473684210527e-05,
      "loss": 1.6527,
      "step": 12500
    },
    {
      "epoch": 4.877574815390595,
      "grad_norm": 4.296464920043945,
      "learning_rate": 9.865842105263159e-05,
      "loss": 1.6022,
      "step": 12550
    },
    {
      "epoch": 4.897007384376215,
      "grad_norm": 7.289546489715576,
      "learning_rate": 9.86321052631579e-05,
      "loss": 1.7196,
      "step": 12600
    },
    {
      "epoch": 4.916439953361834,
      "grad_norm": 4.080571174621582,
      "learning_rate": 9.860578947368422e-05,
      "loss": 1.6623,
      "step": 12650
    },
    {
      "epoch": 4.935872522347454,
      "grad_norm": 6.348351001739502,
      "learning_rate": 9.857947368421053e-05,
      "loss": 1.638,
      "step": 12700
    },
    {
      "epoch": 4.9553050913330745,
      "grad_norm": 5.687911033630371,
      "learning_rate": 9.855315789473685e-05,
      "loss": 1.6331,
      "step": 12750
    },
    {
      "epoch": 4.974737660318694,
      "grad_norm": 4.747588634490967,
      "learning_rate": 9.852684210526316e-05,
      "loss": 1.6322,
      "step": 12800
    },
    {
      "epoch": 4.994170229304314,
      "grad_norm": 3.4772491455078125,
      "learning_rate": 9.850052631578948e-05,
      "loss": 1.6759,
      "step": 12850
    },
    {
      "epoch": 5.013602798289934,
      "grad_norm": 7.125967502593994,
      "learning_rate": 9.847421052631579e-05,
      "loss": 1.5166,
      "step": 12900
    },
    {
      "epoch": 5.033035367275554,
      "grad_norm": 7.8761305809021,
      "learning_rate": 9.84478947368421e-05,
      "loss": 1.4225,
      "step": 12950
    },
    {
      "epoch": 5.0524679362611735,
      "grad_norm": 4.948407173156738,
      "learning_rate": 9.842157894736842e-05,
      "loss": 1.4377,
      "step": 13000
    },
    {
      "epoch": 5.0524679362611735,
      "eval_accuracy": 0.28732449108487584,
      "eval_runtime": 908.7257,
      "eval_samples_per_second": 22.65,
      "eval_steps_per_second": 1.416,
      "step": 13000
    },
    {
      "epoch": 5.071900505246794,
      "grad_norm": 4.546520709991455,
      "learning_rate": 9.839526315789475e-05,
      "loss": 1.462,
      "step": 13050
    },
    {
      "epoch": 5.091333074232414,
      "grad_norm": 3.945157051086426,
      "learning_rate": 9.836894736842106e-05,
      "loss": 1.4214,
      "step": 13100
    },
    {
      "epoch": 5.110765643218033,
      "grad_norm": 5.798865795135498,
      "learning_rate": 9.834263157894737e-05,
      "loss": 1.3673,
      "step": 13150
    },
    {
      "epoch": 5.130198212203653,
      "grad_norm": 5.186479568481445,
      "learning_rate": 9.831631578947368e-05,
      "loss": 1.425,
      "step": 13200
    },
    {
      "epoch": 5.149630781189273,
      "grad_norm": 6.136054992675781,
      "learning_rate": 9.829000000000001e-05,
      "loss": 1.399,
      "step": 13250
    },
    {
      "epoch": 5.169063350174893,
      "grad_norm": 5.4267191886901855,
      "learning_rate": 9.826368421052632e-05,
      "loss": 1.3612,
      "step": 13300
    },
    {
      "epoch": 5.188495919160513,
      "grad_norm": 6.774726390838623,
      "learning_rate": 9.823736842105263e-05,
      "loss": 1.4772,
      "step": 13350
    },
    {
      "epoch": 5.207928488146133,
      "grad_norm": 3.9114222526550293,
      "learning_rate": 9.821105263157894e-05,
      "loss": 1.4336,
      "step": 13400
    },
    {
      "epoch": 5.227361057131753,
      "grad_norm": 5.662858963012695,
      "learning_rate": 9.818473684210527e-05,
      "loss": 1.4467,
      "step": 13450
    },
    {
      "epoch": 5.246793626117372,
      "grad_norm": 4.636590957641602,
      "learning_rate": 9.81584210526316e-05,
      "loss": 1.4401,
      "step": 13500
    },
    {
      "epoch": 5.266226195102993,
      "grad_norm": 3.994364023208618,
      "learning_rate": 9.81321052631579e-05,
      "loss": 1.4181,
      "step": 13550
    },
    {
      "epoch": 5.285658764088613,
      "grad_norm": 7.642608642578125,
      "learning_rate": 9.810578947368422e-05,
      "loss": 1.3862,
      "step": 13600
    },
    {
      "epoch": 5.305091333074232,
      "grad_norm": 6.14906120300293,
      "learning_rate": 9.807947368421053e-05,
      "loss": 1.3952,
      "step": 13650
    },
    {
      "epoch": 5.324523902059853,
      "grad_norm": 6.9564127922058105,
      "learning_rate": 9.805315789473684e-05,
      "loss": 1.3662,
      "step": 13700
    },
    {
      "epoch": 5.343956471045472,
      "grad_norm": 5.573160648345947,
      "learning_rate": 9.802684210526317e-05,
      "loss": 1.408,
      "step": 13750
    },
    {
      "epoch": 5.363389040031092,
      "grad_norm": 8.71044921875,
      "learning_rate": 9.800052631578948e-05,
      "loss": 1.4109,
      "step": 13800
    },
    {
      "epoch": 5.3828216090167125,
      "grad_norm": 7.305654525756836,
      "learning_rate": 9.797421052631579e-05,
      "loss": 1.3941,
      "step": 13850
    },
    {
      "epoch": 5.402254178002332,
      "grad_norm": 11.20594596862793,
      "learning_rate": 9.79478947368421e-05,
      "loss": 1.3478,
      "step": 13900
    },
    {
      "epoch": 5.421686746987952,
      "grad_norm": 7.136280059814453,
      "learning_rate": 9.792157894736843e-05,
      "loss": 1.4619,
      "step": 13950
    },
    {
      "epoch": 5.441119315973571,
      "grad_norm": 16.48175048828125,
      "learning_rate": 9.789526315789475e-05,
      "loss": 1.3761,
      "step": 14000
    },
    {
      "epoch": 5.441119315973571,
      "eval_accuracy": 0.38327746198319,
      "eval_runtime": 910.388,
      "eval_samples_per_second": 22.609,
      "eval_steps_per_second": 1.414,
      "step": 14000
    },
    {
      "epoch": 5.460551884959192,
      "grad_norm": 11.419659614562988,
      "learning_rate": 9.786894736842106e-05,
      "loss": 1.3115,
      "step": 14050
    },
    {
      "epoch": 5.4799844539448115,
      "grad_norm": 5.9120869636535645,
      "learning_rate": 9.784263157894737e-05,
      "loss": 1.4261,
      "step": 14100
    },
    {
      "epoch": 5.499417022930431,
      "grad_norm": 7.246486663818359,
      "learning_rate": 9.781631578947369e-05,
      "loss": 1.374,
      "step": 14150
    },
    {
      "epoch": 5.518849591916052,
      "grad_norm": 7.775355815887451,
      "learning_rate": 9.779e-05,
      "loss": 1.3663,
      "step": 14200
    },
    {
      "epoch": 5.538282160901671,
      "grad_norm": 7.888273239135742,
      "learning_rate": 9.776368421052632e-05,
      "loss": 1.363,
      "step": 14250
    },
    {
      "epoch": 5.557714729887291,
      "grad_norm": 5.733011722564697,
      "learning_rate": 9.773736842105263e-05,
      "loss": 1.4194,
      "step": 14300
    },
    {
      "epoch": 5.577147298872911,
      "grad_norm": 10.086505889892578,
      "learning_rate": 9.771105263157895e-05,
      "loss": 1.3748,
      "step": 14350
    },
    {
      "epoch": 5.596579867858531,
      "grad_norm": 4.082585334777832,
      "learning_rate": 9.768473684210527e-05,
      "loss": 1.4055,
      "step": 14400
    },
    {
      "epoch": 5.616012436844151,
      "grad_norm": 8.12739086151123,
      "learning_rate": 9.765842105263158e-05,
      "loss": 1.371,
      "step": 14450
    },
    {
      "epoch": 5.635445005829771,
      "grad_norm": 5.28172492980957,
      "learning_rate": 9.763210526315791e-05,
      "loss": 1.3121,
      "step": 14500
    },
    {
      "epoch": 5.654877574815391,
      "grad_norm": 6.072118759155273,
      "learning_rate": 9.760578947368422e-05,
      "loss": 1.3478,
      "step": 14550
    },
    {
      "epoch": 5.67431014380101,
      "grad_norm": 4.58156681060791,
      "learning_rate": 9.757947368421053e-05,
      "loss": 1.3545,
      "step": 14600
    },
    {
      "epoch": 5.69374271278663,
      "grad_norm": 8.339735984802246,
      "learning_rate": 9.755315789473684e-05,
      "loss": 1.3544,
      "step": 14650
    },
    {
      "epoch": 5.7131752817722505,
      "grad_norm": 6.120683670043945,
      "learning_rate": 9.752684210526317e-05,
      "loss": 1.3693,
      "step": 14700
    },
    {
      "epoch": 5.73260785075787,
      "grad_norm": 5.467242240905762,
      "learning_rate": 9.750052631578948e-05,
      "loss": 1.3501,
      "step": 14750
    },
    {
      "epoch": 5.75204041974349,
      "grad_norm": 6.579073429107666,
      "learning_rate": 9.747421052631579e-05,
      "loss": 1.3322,
      "step": 14800
    },
    {
      "epoch": 5.77147298872911,
      "grad_norm": 6.063732624053955,
      "learning_rate": 9.74478947368421e-05,
      "loss": 1.353,
      "step": 14850
    },
    {
      "epoch": 5.79090555771473,
      "grad_norm": 6.175238132476807,
      "learning_rate": 9.742157894736843e-05,
      "loss": 1.3488,
      "step": 14900
    },
    {
      "epoch": 5.8103381267003495,
      "grad_norm": 9.881817817687988,
      "learning_rate": 9.739526315789474e-05,
      "loss": 1.2767,
      "step": 14950
    },
    {
      "epoch": 5.82977069568597,
      "grad_norm": 6.4397759437561035,
      "learning_rate": 9.736894736842106e-05,
      "loss": 1.3274,
      "step": 15000
    },
    {
      "epoch": 5.82977069568597,
      "eval_accuracy": 0.4773356653549045,
      "eval_runtime": 906.2985,
      "eval_samples_per_second": 22.711,
      "eval_steps_per_second": 1.42,
      "step": 15000
    },
    {
      "epoch": 5.84920326467159,
      "grad_norm": 7.608861446380615,
      "learning_rate": 9.734263157894738e-05,
      "loss": 1.2751,
      "step": 15050
    },
    {
      "epoch": 5.868635833657209,
      "grad_norm": 5.615549087524414,
      "learning_rate": 9.731631578947369e-05,
      "loss": 1.3654,
      "step": 15100
    },
    {
      "epoch": 5.88806840264283,
      "grad_norm": 7.904622554779053,
      "learning_rate": 9.729e-05,
      "loss": 1.3844,
      "step": 15150
    },
    {
      "epoch": 5.9075009716284494,
      "grad_norm": 4.8503851890563965,
      "learning_rate": 9.726368421052632e-05,
      "loss": 1.3207,
      "step": 15200
    },
    {
      "epoch": 5.926933540614069,
      "grad_norm": 5.832537651062012,
      "learning_rate": 9.723736842105264e-05,
      "loss": 1.3454,
      "step": 15250
    },
    {
      "epoch": 5.946366109599689,
      "grad_norm": 7.324341773986816,
      "learning_rate": 9.721105263157895e-05,
      "loss": 1.3522,
      "step": 15300
    },
    {
      "epoch": 5.965798678585309,
      "grad_norm": 7.22886848449707,
      "learning_rate": 9.718473684210527e-05,
      "loss": 1.3406,
      "step": 15350
    },
    {
      "epoch": 5.985231247570929,
      "grad_norm": 10.675936698913574,
      "learning_rate": 9.715842105263158e-05,
      "loss": 1.3592,
      "step": 15400
    },
    {
      "epoch": 6.0046638165565485,
      "grad_norm": 4.880441188812256,
      "learning_rate": 9.713210526315791e-05,
      "loss": 1.2014,
      "step": 15450
    },
    {
      "epoch": 6.024096385542169,
      "grad_norm": 9.532160758972168,
      "learning_rate": 9.710578947368422e-05,
      "loss": 1.0553,
      "step": 15500
    },
    {
      "epoch": 6.043528954527789,
      "grad_norm": 6.296610355377197,
      "learning_rate": 9.707947368421053e-05,
      "loss": 1.0809,
      "step": 15550
    },
    {
      "epoch": 6.062961523513408,
      "grad_norm": 6.660314559936523,
      "learning_rate": 9.705315789473684e-05,
      "loss": 1.0264,
      "step": 15600
    },
    {
      "epoch": 6.082394092499029,
      "grad_norm": 7.901432514190674,
      "learning_rate": 9.702684210526316e-05,
      "loss": 1.153,
      "step": 15650
    },
    {
      "epoch": 6.101826661484648,
      "grad_norm": 73.9762954711914,
      "learning_rate": 9.700052631578948e-05,
      "loss": 1.1122,
      "step": 15700
    },
    {
      "epoch": 6.121259230470268,
      "grad_norm": 6.655057430267334,
      "learning_rate": 9.697421052631579e-05,
      "loss": 1.0927,
      "step": 15750
    },
    {
      "epoch": 6.1406917994558885,
      "grad_norm": 6.886069297790527,
      "learning_rate": 9.694789473684212e-05,
      "loss": 0.9707,
      "step": 15800
    },
    {
      "epoch": 6.160124368441508,
      "grad_norm": 6.370534420013428,
      "learning_rate": 9.692157894736843e-05,
      "loss": 1.0316,
      "step": 15850
    },
    {
      "epoch": 6.179556937427128,
      "grad_norm": 8.026642799377441,
      "learning_rate": 9.689526315789474e-05,
      "loss": 1.0855,
      "step": 15900
    },
    {
      "epoch": 6.198989506412747,
      "grad_norm": 4.1472859382629395,
      "learning_rate": 9.686894736842107e-05,
      "loss": 1.0527,
      "step": 15950
    },
    {
      "epoch": 6.218422075398368,
      "grad_norm": 13.967721939086914,
      "learning_rate": 9.684263157894738e-05,
      "loss": 1.0484,
      "step": 16000
    },
    {
      "epoch": 6.218422075398368,
      "eval_accuracy": 0.564494971578487,
      "eval_runtime": 907.0139,
      "eval_samples_per_second": 22.693,
      "eval_steps_per_second": 1.419,
      "step": 16000
    },
    {
      "epoch": 6.2378546443839875,
      "grad_norm": 8.06767463684082,
      "learning_rate": 9.681631578947369e-05,
      "loss": 1.0868,
      "step": 16050
    },
    {
      "epoch": 6.257287213369607,
      "grad_norm": 5.721848487854004,
      "learning_rate": 9.679e-05,
      "loss": 1.0125,
      "step": 16100
    },
    {
      "epoch": 6.276719782355228,
      "grad_norm": 7.221185207366943,
      "learning_rate": 9.676368421052631e-05,
      "loss": 1.0789,
      "step": 16150
    },
    {
      "epoch": 6.296152351340847,
      "grad_norm": 6.3180670738220215,
      "learning_rate": 9.673736842105264e-05,
      "loss": 1.08,
      "step": 16200
    },
    {
      "epoch": 6.315584920326467,
      "grad_norm": 5.97916841506958,
      "learning_rate": 9.671105263157895e-05,
      "loss": 1.045,
      "step": 16250
    },
    {
      "epoch": 6.335017489312087,
      "grad_norm": 5.4424967765808105,
      "learning_rate": 9.668473684210527e-05,
      "loss": 1.0835,
      "step": 16300
    },
    {
      "epoch": 6.354450058297707,
      "grad_norm": 5.190217018127441,
      "learning_rate": 9.665842105263158e-05,
      "loss": 1.0738,
      "step": 16350
    },
    {
      "epoch": 6.373882627283327,
      "grad_norm": 6.745716571807861,
      "learning_rate": 9.66321052631579e-05,
      "loss": 1.0482,
      "step": 16400
    },
    {
      "epoch": 6.393315196268947,
      "grad_norm": 5.413162708282471,
      "learning_rate": 9.660578947368422e-05,
      "loss": 1.073,
      "step": 16450
    },
    {
      "epoch": 6.412747765254567,
      "grad_norm": 7.720066547393799,
      "learning_rate": 9.657947368421053e-05,
      "loss": 1.0818,
      "step": 16500
    },
    {
      "epoch": 6.432180334240186,
      "grad_norm": 7.573994159698486,
      "learning_rate": 9.655315789473684e-05,
      "loss": 1.0531,
      "step": 16550
    },
    {
      "epoch": 6.451612903225806,
      "grad_norm": 4.831798076629639,
      "learning_rate": 9.652684210526316e-05,
      "loss": 1.037,
      "step": 16600
    },
    {
      "epoch": 6.471045472211427,
      "grad_norm": 4.910946846008301,
      "learning_rate": 9.650052631578947e-05,
      "loss": 0.9816,
      "step": 16650
    },
    {
      "epoch": 6.490478041197046,
      "grad_norm": 3.0695698261260986,
      "learning_rate": 9.647421052631579e-05,
      "loss": 1.13,
      "step": 16700
    },
    {
      "epoch": 6.509910610182666,
      "grad_norm": 7.250815391540527,
      "learning_rate": 9.644789473684212e-05,
      "loss": 1.0495,
      "step": 16750
    },
    {
      "epoch": 6.529343179168286,
      "grad_norm": 7.248625755310059,
      "learning_rate": 9.642157894736843e-05,
      "loss": 1.0516,
      "step": 16800
    },
    {
      "epoch": 6.548775748153906,
      "grad_norm": 7.137625694274902,
      "learning_rate": 9.639526315789474e-05,
      "loss": 0.9927,
      "step": 16850
    },
    {
      "epoch": 6.568208317139526,
      "grad_norm": 8.5234956741333,
      "learning_rate": 9.636894736842105e-05,
      "loss": 1.0574,
      "step": 16900
    },
    {
      "epoch": 6.587640886125146,
      "grad_norm": 11.187993049621582,
      "learning_rate": 9.634263157894738e-05,
      "loss": 1.0173,
      "step": 16950
    },
    {
      "epoch": 6.607073455110766,
      "grad_norm": 7.501619815826416,
      "learning_rate": 9.631631578947369e-05,
      "loss": 1.0407,
      "step": 17000
    },
    {
      "epoch": 6.607073455110766,
      "eval_accuracy": 0.644949715784871,
      "eval_runtime": 908.2934,
      "eval_samples_per_second": 22.661,
      "eval_steps_per_second": 1.417,
      "step": 17000
    },
    {
      "epoch": 6.626506024096385,
      "grad_norm": 8.143706321716309,
      "learning_rate": 9.629e-05,
      "loss": 1.0366,
      "step": 17050
    },
    {
      "epoch": 6.645938593082006,
      "grad_norm": 7.184727191925049,
      "learning_rate": 9.626368421052631e-05,
      "loss": 1.0755,
      "step": 17100
    },
    {
      "epoch": 6.6653711620676255,
      "grad_norm": 9.023209571838379,
      "learning_rate": 9.623736842105264e-05,
      "loss": 1.0336,
      "step": 17150
    },
    {
      "epoch": 6.684803731053245,
      "grad_norm": 6.733196258544922,
      "learning_rate": 9.621105263157896e-05,
      "loss": 0.9945,
      "step": 17200
    },
    {
      "epoch": 6.704236300038865,
      "grad_norm": 8.987995147705078,
      "learning_rate": 9.618473684210527e-05,
      "loss": 1.0559,
      "step": 17250
    },
    {
      "epoch": 6.723668869024485,
      "grad_norm": 5.8680853843688965,
      "learning_rate": 9.615842105263159e-05,
      "loss": 1.0421,
      "step": 17300
    },
    {
      "epoch": 6.743101438010105,
      "grad_norm": 9.718525886535645,
      "learning_rate": 9.61321052631579e-05,
      "loss": 1.0841,
      "step": 17350
    },
    {
      "epoch": 6.7625340069957245,
      "grad_norm": 5.664957046508789,
      "learning_rate": 9.610578947368421e-05,
      "loss": 1.0099,
      "step": 17400
    },
    {
      "epoch": 6.781966575981345,
      "grad_norm": 5.648830413818359,
      "learning_rate": 9.607947368421053e-05,
      "loss": 0.9931,
      "step": 17450
    },
    {
      "epoch": 6.801399144966965,
      "grad_norm": 16.464553833007812,
      "learning_rate": 9.605315789473685e-05,
      "loss": 0.9833,
      "step": 17500
    },
    {
      "epoch": 6.820831713952584,
      "grad_norm": 9.624323844909668,
      "learning_rate": 9.602684210526316e-05,
      "loss": 1.0671,
      "step": 17550
    },
    {
      "epoch": 6.840264282938205,
      "grad_norm": 5.411578178405762,
      "learning_rate": 9.600052631578947e-05,
      "loss": 1.0591,
      "step": 17600
    },
    {
      "epoch": 6.859696851923824,
      "grad_norm": 6.365889549255371,
      "learning_rate": 9.59742105263158e-05,
      "loss": 1.0182,
      "step": 17650
    },
    {
      "epoch": 6.879129420909444,
      "grad_norm": 9.195286750793457,
      "learning_rate": 9.594789473684212e-05,
      "loss": 1.0284,
      "step": 17700
    },
    {
      "epoch": 6.8985619898950645,
      "grad_norm": 8.23979377746582,
      "learning_rate": 9.592157894736843e-05,
      "loss": 1.0397,
      "step": 17750
    },
    {
      "epoch": 6.917994558880684,
      "grad_norm": 6.512070178985596,
      "learning_rate": 9.589526315789474e-05,
      "loss": 1.0816,
      "step": 17800
    },
    {
      "epoch": 6.937427127866304,
      "grad_norm": 8.572783470153809,
      "learning_rate": 9.586894736842105e-05,
      "loss": 1.0445,
      "step": 17850
    },
    {
      "epoch": 6.956859696851923,
      "grad_norm": 7.494849681854248,
      "learning_rate": 9.584263157894737e-05,
      "loss": 0.9656,
      "step": 17900
    },
    {
      "epoch": 6.976292265837544,
      "grad_norm": 7.2482123374938965,
      "learning_rate": 9.581631578947369e-05,
      "loss": 0.9848,
      "step": 17950
    },
    {
      "epoch": 6.9957248348231635,
      "grad_norm": 8.3186674118042,
      "learning_rate": 9.579e-05,
      "loss": 0.9241,
      "step": 18000
    },
    {
      "epoch": 6.9957248348231635,
      "eval_accuracy": 0.7144245250935238,
      "eval_runtime": 908.8783,
      "eval_samples_per_second": 22.647,
      "eval_steps_per_second": 1.416,
      "step": 18000
    },
    {
      "epoch": 14.025650991061019,
      "grad_norm": 3.8534250259399414,
      "learning_rate": 9.576368421052631e-05,
      "loss": 0.7464,
      "step": 18050
    },
    {
      "epoch": 14.064516129032258,
      "grad_norm": 5.60610294342041,
      "learning_rate": 9.573736842105263e-05,
      "loss": 0.7384,
      "step": 18100
    },
    {
      "epoch": 14.103381267003497,
      "grad_norm": 4.310544967651367,
      "learning_rate": 9.571105263157895e-05,
      "loss": 0.7456,
      "step": 18150
    },
    {
      "epoch": 14.142246404974738,
      "grad_norm": 4.901150226593018,
      "learning_rate": 9.568473684210528e-05,
      "loss": 0.7099,
      "step": 18200
    },
    {
      "epoch": 14.181111542945978,
      "grad_norm": 7.738977909088135,
      "learning_rate": 9.565842105263159e-05,
      "loss": 0.7437,
      "step": 18250
    },
    {
      "epoch": 14.219976680917217,
      "grad_norm": 4.552745819091797,
      "learning_rate": 9.56321052631579e-05,
      "loss": 0.7383,
      "step": 18300
    },
    {
      "epoch": 14.258841818888458,
      "grad_norm": 5.0580034255981445,
      "learning_rate": 9.560578947368421e-05,
      "loss": 0.7134,
      "step": 18350
    },
    {
      "epoch": 14.297706956859697,
      "grad_norm": 4.803134441375732,
      "learning_rate": 9.557947368421054e-05,
      "loss": 0.7098,
      "step": 18400
    },
    {
      "epoch": 14.336572094830936,
      "grad_norm": 4.600313186645508,
      "learning_rate": 9.555315789473685e-05,
      "loss": 0.708,
      "step": 18450
    },
    {
      "epoch": 14.375437232802177,
      "grad_norm": 4.624382019042969,
      "learning_rate": 9.552684210526316e-05,
      "loss": 0.6778,
      "step": 18500
    },
    {
      "epoch": 14.414302370773417,
      "grad_norm": 4.699090957641602,
      "learning_rate": 9.550052631578947e-05,
      "loss": 0.695,
      "step": 18550
    },
    {
      "epoch": 14.453167508744656,
      "grad_norm": 4.733650207519531,
      "learning_rate": 9.54742105263158e-05,
      "loss": 0.6811,
      "step": 18600
    },
    {
      "epoch": 14.492032646715895,
      "grad_norm": 6.0096282958984375,
      "learning_rate": 9.544789473684211e-05,
      "loss": 0.6831,
      "step": 18650
    },
    {
      "epoch": 14.530897784687136,
      "grad_norm": 3.8526785373687744,
      "learning_rate": 9.542157894736843e-05,
      "loss": 0.7374,
      "step": 18700
    },
    {
      "epoch": 14.569762922658375,
      "grad_norm": 4.106224060058594,
      "learning_rate": 9.539526315789474e-05,
      "loss": 0.6959,
      "step": 18750
    },
    {
      "epoch": 14.608628060629615,
      "grad_norm": 5.415355205535889,
      "learning_rate": 9.536894736842106e-05,
      "loss": 0.6764,
      "step": 18800
    },
    {
      "epoch": 14.647493198600856,
      "grad_norm": 5.404593467712402,
      "learning_rate": 9.534263157894737e-05,
      "loss": 0.7101,
      "step": 18850
    },
    {
      "epoch": 14.686358336572095,
      "grad_norm": 5.305964469909668,
      "learning_rate": 9.531631578947369e-05,
      "loss": 0.6862,
      "step": 18900
    },
    {
      "epoch": 14.725223474543334,
      "grad_norm": 4.21513557434082,
      "learning_rate": 9.529e-05,
      "loss": 0.6621,
      "step": 18950
    },
    {
      "epoch": 14.764088612514575,
      "grad_norm": 5.1977715492248535,
      "learning_rate": 9.526368421052632e-05,
      "loss": 0.6832,
      "step": 19000
    },
    {
      "epoch": 14.764088612514575,
      "eval_accuracy": 0.8375844143224992,
      "eval_runtime": 859.0271,
      "eval_samples_per_second": 23.961,
      "eval_steps_per_second": 2.995,
      "step": 19000
    },
    {
      "epoch": 14.802953750485814,
      "grad_norm": 4.032940864562988,
      "learning_rate": 9.523736842105263e-05,
      "loss": 0.7262,
      "step": 19050
    },
    {
      "epoch": 14.841818888457054,
      "grad_norm": 6.659789562225342,
      "learning_rate": 9.521105263157895e-05,
      "loss": 0.6872,
      "step": 19100
    },
    {
      "epoch": 14.880684026428295,
      "grad_norm": 4.311520576477051,
      "learning_rate": 9.518473684210528e-05,
      "loss": 0.715,
      "step": 19150
    },
    {
      "epoch": 14.919549164399534,
      "grad_norm": 4.256176471710205,
      "learning_rate": 9.515842105263159e-05,
      "loss": 0.7551,
      "step": 19200
    },
    {
      "epoch": 14.958414302370773,
      "grad_norm": 5.675657749176025,
      "learning_rate": 9.51321052631579e-05,
      "loss": 0.6938,
      "step": 19250
    },
    {
      "epoch": 14.997279440342012,
      "grad_norm": 5.095183372497559,
      "learning_rate": 9.510578947368421e-05,
      "loss": 0.6902,
      "step": 19300
    },
    {
      "epoch": 15.035755926933541,
      "grad_norm": 3.1958320140838623,
      "learning_rate": 9.507947368421052e-05,
      "loss": 0.4941,
      "step": 19350
    },
    {
      "epoch": 15.07462106490478,
      "grad_norm": 4.101893901824951,
      "learning_rate": 9.505315789473685e-05,
      "loss": 0.5105,
      "step": 19400
    },
    {
      "epoch": 15.11348620287602,
      "grad_norm": 54.76054763793945,
      "learning_rate": 9.502684210526316e-05,
      "loss": 0.5441,
      "step": 19450
    },
    {
      "epoch": 15.15235134084726,
      "grad_norm": 4.233603477478027,
      "learning_rate": 9.500052631578947e-05,
      "loss": 0.4974,
      "step": 19500
    },
    {
      "epoch": 15.1912164788185,
      "grad_norm": 4.762860298156738,
      "learning_rate": 9.49742105263158e-05,
      "loss": 0.5109,
      "step": 19550
    },
    {
      "epoch": 15.23008161678974,
      "grad_norm": 5.9275054931640625,
      "learning_rate": 9.494789473684211e-05,
      "loss": 0.4692,
      "step": 19600
    },
    {
      "epoch": 15.268946754760979,
      "grad_norm": 4.715384006500244,
      "learning_rate": 9.492157894736843e-05,
      "loss": 0.507,
      "step": 19650
    },
    {
      "epoch": 15.30781189273222,
      "grad_norm": 6.866525650024414,
      "learning_rate": 9.489526315789475e-05,
      "loss": 0.5331,
      "step": 19700
    },
    {
      "epoch": 15.346677030703459,
      "grad_norm": 7.685290336608887,
      "learning_rate": 9.486894736842106e-05,
      "loss": 0.4853,
      "step": 19750
    },
    {
      "epoch": 15.385542168674698,
      "grad_norm": 5.017430782318115,
      "learning_rate": 9.484263157894737e-05,
      "loss": 0.49,
      "step": 19800
    },
    {
      "epoch": 15.42440730664594,
      "grad_norm": 4.645963191986084,
      "learning_rate": 9.481631578947368e-05,
      "loss": 0.536,
      "step": 19850
    },
    {
      "epoch": 15.463272444617179,
      "grad_norm": 3.9454667568206787,
      "learning_rate": 9.479e-05,
      "loss": 0.4746,
      "step": 19900
    },
    {
      "epoch": 15.502137582588418,
      "grad_norm": 9.802485466003418,
      "learning_rate": 9.476368421052632e-05,
      "loss": 0.488,
      "step": 19950
    },
    {
      "epoch": 15.541002720559659,
      "grad_norm": 3.2630667686462402,
      "learning_rate": 9.473736842105264e-05,
      "loss": 0.5281,
      "step": 20000
    },
    {
      "epoch": 15.541002720559659,
      "eval_accuracy": 0.8903464023708886,
      "eval_runtime": 856.9951,
      "eval_samples_per_second": 24.018,
      "eval_steps_per_second": 3.002,
      "step": 20000
    },
    {
      "epoch": 15.579867858530898,
      "grad_norm": 5.689331531524658,
      "learning_rate": 9.471105263157895e-05,
      "loss": 0.5053,
      "step": 20050
    },
    {
      "epoch": 15.618732996502137,
      "grad_norm": 9.707596778869629,
      "learning_rate": 9.468473684210527e-05,
      "loss": 0.4869,
      "step": 20100
    },
    {
      "epoch": 15.657598134473377,
      "grad_norm": 4.015494346618652,
      "learning_rate": 9.465842105263159e-05,
      "loss": 0.5089,
      "step": 20150
    },
    {
      "epoch": 15.696463272444618,
      "grad_norm": 6.2136616706848145,
      "learning_rate": 9.46321052631579e-05,
      "loss": 0.5071,
      "step": 20200
    },
    {
      "epoch": 15.735328410415857,
      "grad_norm": 4.501992702484131,
      "learning_rate": 9.460578947368421e-05,
      "loss": 0.5169,
      "step": 20250
    },
    {
      "epoch": 15.774193548387096,
      "grad_norm": 3.086958408355713,
      "learning_rate": 9.457947368421053e-05,
      "loss": 0.5076,
      "step": 20300
    },
    {
      "epoch": 15.813058686358337,
      "grad_norm": 5.842269420623779,
      "learning_rate": 9.455315789473684e-05,
      "loss": 0.4818,
      "step": 20350
    },
    {
      "epoch": 15.851923824329576,
      "grad_norm": 5.230940341949463,
      "learning_rate": 9.452684210526316e-05,
      "loss": 0.5153,
      "step": 20400
    },
    {
      "epoch": 15.890788962300816,
      "grad_norm": 5.823616027832031,
      "learning_rate": 9.450052631578947e-05,
      "loss": 0.4984,
      "step": 20450
    },
    {
      "epoch": 15.929654100272057,
      "grad_norm": 4.166293144226074,
      "learning_rate": 9.44742105263158e-05,
      "loss": 0.5255,
      "step": 20500
    },
    {
      "epoch": 15.968519238243296,
      "grad_norm": 5.066617488861084,
      "learning_rate": 9.444789473684211e-05,
      "loss": 0.4885,
      "step": 20550
    },
    {
      "epoch": 16.006995724834823,
      "grad_norm": 3.68914532661438,
      "learning_rate": 9.442157894736842e-05,
      "loss": 0.4843,
      "step": 20600
    },
    {
      "epoch": 16.045860862806062,
      "grad_norm": 3.0051112174987793,
      "learning_rate": 9.439526315789475e-05,
      "loss": 0.3863,
      "step": 20650
    },
    {
      "epoch": 16.0847260007773,
      "grad_norm": 4.9845404624938965,
      "learning_rate": 9.436894736842106e-05,
      "loss": 0.3352,
      "step": 20700
    },
    {
      "epoch": 16.12359113874854,
      "grad_norm": 6.085782527923584,
      "learning_rate": 9.434263157894737e-05,
      "loss": 0.3561,
      "step": 20750
    },
    {
      "epoch": 16.162456276719784,
      "grad_norm": 9.10416030883789,
      "learning_rate": 9.431631578947368e-05,
      "loss": 0.3884,
      "step": 20800
    },
    {
      "epoch": 16.201321414691023,
      "grad_norm": 4.070003509521484,
      "learning_rate": 9.429000000000001e-05,
      "loss": 0.3566,
      "step": 20850
    },
    {
      "epoch": 16.240186552662262,
      "grad_norm": 6.551778793334961,
      "learning_rate": 9.426368421052632e-05,
      "loss": 0.3936,
      "step": 20900
    },
    {
      "epoch": 16.2790516906335,
      "grad_norm": 2.916369676589966,
      "learning_rate": 9.423736842105264e-05,
      "loss": 0.3535,
      "step": 20950
    },
    {
      "epoch": 16.31791682860474,
      "grad_norm": 4.9708123207092285,
      "learning_rate": 9.421105263157896e-05,
      "loss": 0.3538,
      "step": 21000
    },
    {
      "epoch": 16.31791682860474,
      "eval_accuracy": 0.9259583151144148,
      "eval_runtime": 857.5854,
      "eval_samples_per_second": 24.001,
      "eval_steps_per_second": 3.0,
      "step": 21000
    },
    {
      "epoch": 16.35678196657598,
      "grad_norm": 5.022733688354492,
      "learning_rate": 9.418473684210527e-05,
      "loss": 0.3687,
      "step": 21050
    },
    {
      "epoch": 16.395647104547223,
      "grad_norm": 6.180454254150391,
      "learning_rate": 9.415842105263158e-05,
      "loss": 0.3892,
      "step": 21100
    },
    {
      "epoch": 16.434512242518462,
      "grad_norm": 5.216296195983887,
      "learning_rate": 9.41321052631579e-05,
      "loss": 0.3633,
      "step": 21150
    },
    {
      "epoch": 16.4733773804897,
      "grad_norm": 4.853186130523682,
      "learning_rate": 9.410578947368422e-05,
      "loss": 0.3367,
      "step": 21200
    },
    {
      "epoch": 16.51224251846094,
      "grad_norm": 4.97639274597168,
      "learning_rate": 9.407947368421053e-05,
      "loss": 0.3945,
      "step": 21250
    },
    {
      "epoch": 16.55110765643218,
      "grad_norm": 6.093092441558838,
      "learning_rate": 9.405315789473684e-05,
      "loss": 0.3445,
      "step": 21300
    },
    {
      "epoch": 16.58997279440342,
      "grad_norm": 5.588438034057617,
      "learning_rate": 9.402684210526316e-05,
      "loss": 0.3521,
      "step": 21350
    },
    {
      "epoch": 16.62883793237466,
      "grad_norm": 4.828138828277588,
      "learning_rate": 9.400052631578949e-05,
      "loss": 0.3503,
      "step": 21400
    },
    {
      "epoch": 16.6677030703459,
      "grad_norm": 4.060586929321289,
      "learning_rate": 9.39742105263158e-05,
      "loss": 0.3969,
      "step": 21450
    },
    {
      "epoch": 16.70656820831714,
      "grad_norm": 3.49153995513916,
      "learning_rate": 9.394789473684211e-05,
      "loss": 0.3768,
      "step": 21500
    },
    {
      "epoch": 16.74543334628838,
      "grad_norm": 4.94192361831665,
      "learning_rate": 9.392157894736842e-05,
      "loss": 0.3908,
      "step": 21550
    },
    {
      "epoch": 16.78429848425962,
      "grad_norm": 5.816671848297119,
      "learning_rate": 9.389526315789475e-05,
      "loss": 0.3491,
      "step": 21600
    },
    {
      "epoch": 16.823163622230858,
      "grad_norm": 4.488963603973389,
      "learning_rate": 9.386894736842106e-05,
      "loss": 0.3465,
      "step": 21650
    },
    {
      "epoch": 16.862028760202097,
      "grad_norm": 5.019749641418457,
      "learning_rate": 9.384263157894737e-05,
      "loss": 0.3675,
      "step": 21700
    },
    {
      "epoch": 16.90089389817334,
      "grad_norm": 8.628226280212402,
      "learning_rate": 9.381631578947368e-05,
      "loss": 0.3765,
      "step": 21750
    },
    {
      "epoch": 16.93975903614458,
      "grad_norm": 5.863889694213867,
      "learning_rate": 9.379e-05,
      "loss": 0.3939,
      "step": 21800
    },
    {
      "epoch": 16.97862417411582,
      "grad_norm": 3.1837944984436035,
      "learning_rate": 9.376368421052632e-05,
      "loss": 0.404,
      "step": 21850
    },
    {
      "epoch": 17.017100660707346,
      "grad_norm": 2.9766252040863037,
      "learning_rate": 9.373736842105265e-05,
      "loss": 0.3439,
      "step": 21900
    },
    {
      "epoch": 17.055965798678585,
      "grad_norm": 3.5172083377838135,
      "learning_rate": 9.371105263157896e-05,
      "loss": 0.258,
      "step": 21950
    },
    {
      "epoch": 17.094830936649824,
      "grad_norm": 4.440106391906738,
      "learning_rate": 9.368473684210527e-05,
      "loss": 0.275,
      "step": 22000
    },
    {
      "epoch": 17.094830936649824,
      "eval_accuracy": 0.9561774279745421,
      "eval_runtime": 854.4439,
      "eval_samples_per_second": 24.089,
      "eval_steps_per_second": 3.011,
      "step": 22000
    },
    {
      "epoch": 17.133696074621064,
      "grad_norm": 19.169038772583008,
      "learning_rate": 9.365842105263158e-05,
      "loss": 0.2566,
      "step": 22050
    },
    {
      "epoch": 17.172561212592306,
      "grad_norm": 5.666447639465332,
      "learning_rate": 9.36321052631579e-05,
      "loss": 0.2736,
      "step": 22100
    },
    {
      "epoch": 17.211426350563546,
      "grad_norm": 2.622077465057373,
      "learning_rate": 9.360578947368422e-05,
      "loss": 0.2408,
      "step": 22150
    },
    {
      "epoch": 17.250291488534785,
      "grad_norm": 3.8020524978637695,
      "learning_rate": 9.357947368421053e-05,
      "loss": 0.2865,
      "step": 22200
    },
    {
      "epoch": 17.289156626506024,
      "grad_norm": 3.419302225112915,
      "learning_rate": 9.355315789473684e-05,
      "loss": 0.2249,
      "step": 22250
    },
    {
      "epoch": 17.328021764477263,
      "grad_norm": 4.451815128326416,
      "learning_rate": 9.352684210526315e-05,
      "loss": 0.296,
      "step": 22300
    },
    {
      "epoch": 17.366886902448503,
      "grad_norm": 4.277288913726807,
      "learning_rate": 9.350052631578948e-05,
      "loss": 0.2647,
      "step": 22350
    },
    {
      "epoch": 17.405752040419742,
      "grad_norm": 3.892031669616699,
      "learning_rate": 9.34742105263158e-05,
      "loss": 0.2746,
      "step": 22400
    },
    {
      "epoch": 17.444617178390985,
      "grad_norm": 3.7854325771331787,
      "learning_rate": 9.344789473684211e-05,
      "loss": 0.2615,
      "step": 22450
    },
    {
      "epoch": 17.483482316362224,
      "grad_norm": 5.1615471839904785,
      "learning_rate": 9.342157894736842e-05,
      "loss": 0.27,
      "step": 22500
    },
    {
      "epoch": 17.522347454333463,
      "grad_norm": 7.163486003875732,
      "learning_rate": 9.339526315789474e-05,
      "loss": 0.2791,
      "step": 22550
    },
    {
      "epoch": 17.561212592304702,
      "grad_norm": 3.133782386779785,
      "learning_rate": 9.336894736842106e-05,
      "loss": 0.289,
      "step": 22600
    },
    {
      "epoch": 17.60007773027594,
      "grad_norm": 8.343432426452637,
      "learning_rate": 9.334263157894737e-05,
      "loss": 0.2562,
      "step": 22650
    },
    {
      "epoch": 17.63894286824718,
      "grad_norm": 4.588578701019287,
      "learning_rate": 9.331631578947368e-05,
      "loss": 0.2722,
      "step": 22700
    },
    {
      "epoch": 17.677808006218424,
      "grad_norm": 5.381346225738525,
      "learning_rate": 9.329e-05,
      "loss": 0.2841,
      "step": 22750
    },
    {
      "epoch": 17.716673144189663,
      "grad_norm": 5.521486759185791,
      "learning_rate": 9.326368421052632e-05,
      "loss": 0.2821,
      "step": 22800
    },
    {
      "epoch": 17.755538282160902,
      "grad_norm": 7.152676105499268,
      "learning_rate": 9.323736842105265e-05,
      "loss": 0.2991,
      "step": 22850
    },
    {
      "epoch": 17.79440342013214,
      "grad_norm": 2.277432918548584,
      "learning_rate": 9.321105263157896e-05,
      "loss": 0.2974,
      "step": 22900
    },
    {
      "epoch": 17.83326855810338,
      "grad_norm": 2.4336891174316406,
      "learning_rate": 9.318473684210527e-05,
      "loss": 0.2944,
      "step": 22950
    },
    {
      "epoch": 17.87213369607462,
      "grad_norm": 3.757892608642578,
      "learning_rate": 9.315842105263158e-05,
      "loss": 0.2839,
      "step": 23000
    },
    {
      "epoch": 17.87213369607462,
      "eval_accuracy": 0.9704610600981393,
      "eval_runtime": 853.279,
      "eval_samples_per_second": 24.122,
      "eval_steps_per_second": 3.015,
      "step": 23000
    },
    {
      "epoch": 17.91099883404586,
      "grad_norm": 2.538939952850342,
      "learning_rate": 9.313210526315789e-05,
      "loss": 0.2912,
      "step": 23050
    },
    {
      "epoch": 17.949863972017102,
      "grad_norm": 2.826970338821411,
      "learning_rate": 9.310578947368422e-05,
      "loss": 0.3091,
      "step": 23100
    },
    {
      "epoch": 17.98872910998834,
      "grad_norm": 6.114530086517334,
      "learning_rate": 9.307947368421053e-05,
      "loss": 0.3039,
      "step": 23150
    },
    {
      "epoch": 18.02720559657987,
      "grad_norm": 3.5452566146850586,
      "learning_rate": 9.305315789473684e-05,
      "loss": 0.2166,
      "step": 23200
    },
    {
      "epoch": 18.066070734551108,
      "grad_norm": 3.9908535480499268,
      "learning_rate": 9.302684210526317e-05,
      "loss": 0.1961,
      "step": 23250
    },
    {
      "epoch": 18.104935872522347,
      "grad_norm": 3.3680648803710938,
      "learning_rate": 9.300052631578948e-05,
      "loss": 0.1853,
      "step": 23300
    },
    {
      "epoch": 18.143801010493586,
      "grad_norm": 3.556689977645874,
      "learning_rate": 9.29742105263158e-05,
      "loss": 0.2204,
      "step": 23350
    },
    {
      "epoch": 18.182666148464826,
      "grad_norm": 3.8766796588897705,
      "learning_rate": 9.294789473684211e-05,
      "loss": 0.212,
      "step": 23400
    },
    {
      "epoch": 18.22153128643607,
      "grad_norm": 3.0015463829040527,
      "learning_rate": 9.292157894736843e-05,
      "loss": 0.2259,
      "step": 23450
    },
    {
      "epoch": 18.260396424407308,
      "grad_norm": 6.374223709106445,
      "learning_rate": 9.289526315789474e-05,
      "loss": 0.228,
      "step": 23500
    },
    {
      "epoch": 18.299261562378547,
      "grad_norm": 4.002296447753906,
      "learning_rate": 9.286894736842105e-05,
      "loss": 0.1872,
      "step": 23550
    },
    {
      "epoch": 18.338126700349786,
      "grad_norm": 3.892784357070923,
      "learning_rate": 9.284263157894737e-05,
      "loss": 0.2024,
      "step": 23600
    },
    {
      "epoch": 18.376991838321025,
      "grad_norm": 2.9542012214660645,
      "learning_rate": 9.281631578947369e-05,
      "loss": 0.2348,
      "step": 23650
    },
    {
      "epoch": 18.415856976292265,
      "grad_norm": 4.670549392700195,
      "learning_rate": 9.279e-05,
      "loss": 0.2231,
      "step": 23700
    },
    {
      "epoch": 18.454722114263504,
      "grad_norm": 4.850724220275879,
      "learning_rate": 9.276368421052632e-05,
      "loss": 0.2145,
      "step": 23750
    },
    {
      "epoch": 18.493587252234747,
      "grad_norm": 4.08035135269165,
      "learning_rate": 9.273736842105263e-05,
      "loss": 0.1965,
      "step": 23800
    },
    {
      "epoch": 18.532452390205986,
      "grad_norm": 2.9276678562164307,
      "learning_rate": 9.271105263157896e-05,
      "loss": 0.2289,
      "step": 23850
    },
    {
      "epoch": 18.571317528177225,
      "grad_norm": 3.3067493438720703,
      "learning_rate": 9.268473684210527e-05,
      "loss": 0.2138,
      "step": 23900
    },
    {
      "epoch": 18.610182666148464,
      "grad_norm": 6.159919261932373,
      "learning_rate": 9.265842105263158e-05,
      "loss": 0.2366,
      "step": 23950
    },
    {
      "epoch": 18.649047804119704,
      "grad_norm": 2.813751459121704,
      "learning_rate": 9.26321052631579e-05,
      "loss": 0.2365,
      "step": 24000
    },
    {
      "epoch": 18.649047804119704,
      "eval_accuracy": 0.9767769518534714,
      "eval_runtime": 853.479,
      "eval_samples_per_second": 24.117,
      "eval_steps_per_second": 3.015,
      "step": 24000
    },
    {
      "epoch": 18.687912942090943,
      "grad_norm": 4.694916248321533,
      "learning_rate": 9.26057894736842e-05,
      "loss": 0.2256,
      "step": 24050
    },
    {
      "epoch": 18.726778080062186,
      "grad_norm": 6.502670764923096,
      "learning_rate": 9.257947368421053e-05,
      "loss": 0.1918,
      "step": 24100
    },
    {
      "epoch": 18.765643218033425,
      "grad_norm": 4.842265605926514,
      "learning_rate": 9.255315789473684e-05,
      "loss": 0.2503,
      "step": 24150
    },
    {
      "epoch": 18.804508356004664,
      "grad_norm": 3.503544330596924,
      "learning_rate": 9.252684210526317e-05,
      "loss": 0.21,
      "step": 24200
    },
    {
      "epoch": 18.843373493975903,
      "grad_norm": 2.136115312576294,
      "learning_rate": 9.250052631578948e-05,
      "loss": 0.2161,
      "step": 24250
    },
    {
      "epoch": 18.882238631947143,
      "grad_norm": 2.678394317626953,
      "learning_rate": 9.247421052631579e-05,
      "loss": 0.1887,
      "step": 24300
    },
    {
      "epoch": 18.921103769918382,
      "grad_norm": 3.438103675842285,
      "learning_rate": 9.244789473684212e-05,
      "loss": 0.254,
      "step": 24350
    },
    {
      "epoch": 18.959968907889625,
      "grad_norm": 3.543240785598755,
      "learning_rate": 9.242157894736843e-05,
      "loss": 0.2246,
      "step": 24400
    },
    {
      "epoch": 18.998834045860864,
      "grad_norm": 4.896575450897217,
      "learning_rate": 9.239526315789474e-05,
      "loss": 0.2251,
      "step": 24450
    },
    {
      "epoch": 19.03731053245239,
      "grad_norm": 4.922858715057373,
      "learning_rate": 9.236894736842105e-05,
      "loss": 0.1551,
      "step": 24500
    },
    {
      "epoch": 19.07617567042363,
      "grad_norm": 12.493631362915039,
      "learning_rate": 9.234263157894738e-05,
      "loss": 0.1661,
      "step": 24550
    },
    {
      "epoch": 19.11504080839487,
      "grad_norm": 4.771761894226074,
      "learning_rate": 9.231631578947369e-05,
      "loss": 0.154,
      "step": 24600
    },
    {
      "epoch": 19.15390594636611,
      "grad_norm": 3.863640785217285,
      "learning_rate": 9.229000000000001e-05,
      "loss": 0.14,
      "step": 24650
    },
    {
      "epoch": 19.19277108433735,
      "grad_norm": 5.57180118560791,
      "learning_rate": 9.226368421052632e-05,
      "loss": 0.1585,
      "step": 24700
    },
    {
      "epoch": 19.231636222308587,
      "grad_norm": 4.381727695465088,
      "learning_rate": 9.223736842105264e-05,
      "loss": 0.1764,
      "step": 24750
    },
    {
      "epoch": 19.27050136027983,
      "grad_norm": 3.324845552444458,
      "learning_rate": 9.221105263157895e-05,
      "loss": 0.1605,
      "step": 24800
    },
    {
      "epoch": 19.30936649825107,
      "grad_norm": 3.4294657707214355,
      "learning_rate": 9.218473684210527e-05,
      "loss": 0.168,
      "step": 24850
    },
    {
      "epoch": 19.34823163622231,
      "grad_norm": 2.7688677310943604,
      "learning_rate": 9.215842105263158e-05,
      "loss": 0.1594,
      "step": 24900
    },
    {
      "epoch": 19.387096774193548,
      "grad_norm": 4.8695807456970215,
      "learning_rate": 9.21321052631579e-05,
      "loss": 0.1659,
      "step": 24950
    },
    {
      "epoch": 19.425961912164787,
      "grad_norm": 3.9514307975769043,
      "learning_rate": 9.210578947368421e-05,
      "loss": 0.1872,
      "step": 25000
    },
    {
      "epoch": 19.425961912164787,
      "eval_accuracy": 0.9811009085167371,
      "eval_runtime": 857.0247,
      "eval_samples_per_second": 24.017,
      "eval_steps_per_second": 3.002,
      "step": 25000
    },
    {
      "epoch": 19.464827050136027,
      "grad_norm": 2.4025089740753174,
      "learning_rate": 9.207947368421053e-05,
      "loss": 0.1627,
      "step": 25050
    },
    {
      "epoch": 19.50369218810727,
      "grad_norm": 4.841952323913574,
      "learning_rate": 9.205315789473684e-05,
      "loss": 0.1714,
      "step": 25100
    },
    {
      "epoch": 19.54255732607851,
      "grad_norm": 3.034189224243164,
      "learning_rate": 9.202684210526317e-05,
      "loss": 0.1652,
      "step": 25150
    },
    {
      "epoch": 19.581422464049748,
      "grad_norm": 3.2492823600769043,
      "learning_rate": 9.200052631578948e-05,
      "loss": 0.2033,
      "step": 25200
    },
    {
      "epoch": 19.620287602020987,
      "grad_norm": 7.616010665893555,
      "learning_rate": 9.197421052631579e-05,
      "loss": 0.2131,
      "step": 25250
    },
    {
      "epoch": 19.659152739992226,
      "grad_norm": 3.7748496532440186,
      "learning_rate": 9.194789473684212e-05,
      "loss": 0.1808,
      "step": 25300
    },
    {
      "epoch": 19.698017877963466,
      "grad_norm": 4.933774471282959,
      "learning_rate": 9.192157894736843e-05,
      "loss": 0.1586,
      "step": 25350
    },
    {
      "epoch": 19.73688301593471,
      "grad_norm": 6.590142250061035,
      "learning_rate": 9.189526315789474e-05,
      "loss": 0.2023,
      "step": 25400
    },
    {
      "epoch": 19.775748153905948,
      "grad_norm": 3.3198065757751465,
      "learning_rate": 9.186894736842105e-05,
      "loss": 0.1881,
      "step": 25450
    },
    {
      "epoch": 19.814613291877187,
      "grad_norm": 5.870887279510498,
      "learning_rate": 9.184263157894736e-05,
      "loss": 0.1717,
      "step": 25500
    },
    {
      "epoch": 19.853478429848426,
      "grad_norm": 2.5725080966949463,
      "learning_rate": 9.181631578947369e-05,
      "loss": 0.1939,
      "step": 25550
    },
    {
      "epoch": 19.892343567819665,
      "grad_norm": 3.569075345993042,
      "learning_rate": 9.179000000000001e-05,
      "loss": 0.1915,
      "step": 25600
    },
    {
      "epoch": 19.931208705790905,
      "grad_norm": 3.5029423236846924,
      "learning_rate": 9.176368421052633e-05,
      "loss": 0.1679,
      "step": 25650
    },
    {
      "epoch": 19.970073843762144,
      "grad_norm": 2.4752578735351562,
      "learning_rate": 9.173736842105264e-05,
      "loss": 0.168,
      "step": 25700
    },
    {
      "epoch": 20.00855033035367,
      "grad_norm": 9.303007125854492,
      "learning_rate": 9.171105263157895e-05,
      "loss": 0.1754,
      "step": 25750
    },
    {
      "epoch": 20.047415468324914,
      "grad_norm": 4.825704574584961,
      "learning_rate": 9.168473684210527e-05,
      "loss": 0.1179,
      "step": 25800
    },
    {
      "epoch": 20.086280606296153,
      "grad_norm": 4.486258506774902,
      "learning_rate": 9.165842105263159e-05,
      "loss": 0.1358,
      "step": 25850
    },
    {
      "epoch": 20.125145744267392,
      "grad_norm": 1.9208111763000488,
      "learning_rate": 9.16321052631579e-05,
      "loss": 0.1231,
      "step": 25900
    },
    {
      "epoch": 20.16401088223863,
      "grad_norm": 5.461341857910156,
      "learning_rate": 9.160578947368421e-05,
      "loss": 0.1262,
      "step": 25950
    },
    {
      "epoch": 20.20287602020987,
      "grad_norm": 0.8620373010635376,
      "learning_rate": 9.157947368421052e-05,
      "loss": 0.1263,
      "step": 26000
    },
    {
      "epoch": 20.20287602020987,
      "eval_accuracy": 0.9867366273138026,
      "eval_runtime": 884.9066,
      "eval_samples_per_second": 23.26,
      "eval_steps_per_second": 2.908,
      "step": 26000
    },
    {
      "epoch": 20.24174115818111,
      "grad_norm": 3.118957281112671,
      "learning_rate": 9.155315789473685e-05,
      "loss": 0.1281,
      "step": 26050
    },
    {
      "epoch": 20.280606296152353,
      "grad_norm": 2.739800214767456,
      "learning_rate": 9.152684210526317e-05,
      "loss": 0.1213,
      "step": 26100
    },
    {
      "epoch": 20.319471434123592,
      "grad_norm": 6.1789445877075195,
      "learning_rate": 9.150052631578948e-05,
      "loss": 0.1474,
      "step": 26150
    },
    {
      "epoch": 20.35833657209483,
      "grad_norm": 2.280268907546997,
      "learning_rate": 9.14742105263158e-05,
      "loss": 0.1275,
      "step": 26200
    },
    {
      "epoch": 20.39720171006607,
      "grad_norm": 4.79387903213501,
      "learning_rate": 9.14478947368421e-05,
      "loss": 0.1337,
      "step": 26250
    },
    {
      "epoch": 20.43606684803731,
      "grad_norm": 2.4207911491394043,
      "learning_rate": 9.142157894736843e-05,
      "loss": 0.1317,
      "step": 26300
    },
    {
      "epoch": 20.47493198600855,
      "grad_norm": 0.7470328211784363,
      "learning_rate": 9.139526315789474e-05,
      "loss": 0.1197,
      "step": 26350
    },
    {
      "epoch": 20.51379712397979,
      "grad_norm": 3.8738152980804443,
      "learning_rate": 9.136894736842105e-05,
      "loss": 0.1336,
      "step": 26400
    },
    {
      "epoch": 20.55266226195103,
      "grad_norm": 3.2332074642181396,
      "learning_rate": 9.134263157894737e-05,
      "loss": 0.1508,
      "step": 26450
    },
    {
      "epoch": 20.59152739992227,
      "grad_norm": 4.310563087463379,
      "learning_rate": 9.131631578947368e-05,
      "loss": 0.1238,
      "step": 26500
    },
    {
      "epoch": 20.63039253789351,
      "grad_norm": 4.384401321411133,
      "learning_rate": 9.129000000000002e-05,
      "loss": 0.1583,
      "step": 26550
    },
    {
      "epoch": 20.66925767586475,
      "grad_norm": 4.6766486167907715,
      "learning_rate": 9.126368421052633e-05,
      "loss": 0.1224,
      "step": 26600
    },
    {
      "epoch": 20.70812281383599,
      "grad_norm": 2.9925906658172607,
      "learning_rate": 9.123736842105264e-05,
      "loss": 0.1504,
      "step": 26650
    },
    {
      "epoch": 20.746987951807228,
      "grad_norm": 6.686166763305664,
      "learning_rate": 9.121105263157895e-05,
      "loss": 0.1324,
      "step": 26700
    },
    {
      "epoch": 20.78585308977847,
      "grad_norm": 4.77677059173584,
      "learning_rate": 9.118473684210526e-05,
      "loss": 0.1291,
      "step": 26750
    },
    {
      "epoch": 20.82471822774971,
      "grad_norm": 4.284395217895508,
      "learning_rate": 9.115842105263159e-05,
      "loss": 0.1421,
      "step": 26800
    },
    {
      "epoch": 20.86358336572095,
      "grad_norm": 3.2048757076263428,
      "learning_rate": 9.11321052631579e-05,
      "loss": 0.141,
      "step": 26850
    },
    {
      "epoch": 20.902448503692188,
      "grad_norm": 3.526207447052002,
      "learning_rate": 9.110578947368421e-05,
      "loss": 0.1471,
      "step": 26900
    },
    {
      "epoch": 20.941313641663427,
      "grad_norm": 2.7321910858154297,
      "learning_rate": 9.107947368421052e-05,
      "loss": 0.1307,
      "step": 26950
    },
    {
      "epoch": 20.980178779634667,
      "grad_norm": 3.099595546722412,
      "learning_rate": 9.105315789473685e-05,
      "loss": 0.1459,
      "step": 27000
    },
    {
      "epoch": 20.980178779634667,
      "eval_accuracy": 0.989943156974202,
      "eval_runtime": 880.3227,
      "eval_samples_per_second": 23.381,
      "eval_steps_per_second": 2.923,
      "step": 27000
    },
    {
      "epoch": 21.018655266226194,
      "grad_norm": 5.128775119781494,
      "learning_rate": 9.102684210526317e-05,
      "loss": 0.1232,
      "step": 27050
    },
    {
      "epoch": 21.057520404197437,
      "grad_norm": 3.6633059978485107,
      "learning_rate": 9.100052631578948e-05,
      "loss": 0.1005,
      "step": 27100
    },
    {
      "epoch": 21.096385542168676,
      "grad_norm": 2.404827117919922,
      "learning_rate": 9.09742105263158e-05,
      "loss": 0.1007,
      "step": 27150
    },
    {
      "epoch": 21.135250680139915,
      "grad_norm": 1.061808466911316,
      "learning_rate": 9.09478947368421e-05,
      "loss": 0.1129,
      "step": 27200
    },
    {
      "epoch": 21.174115818111154,
      "grad_norm": 4.0880513191223145,
      "learning_rate": 9.092157894736842e-05,
      "loss": 0.1165,
      "step": 27250
    },
    {
      "epoch": 21.212980956082394,
      "grad_norm": 9.207396507263184,
      "learning_rate": 9.089526315789474e-05,
      "loss": 0.1177,
      "step": 27300
    },
    {
      "epoch": 21.251846094053633,
      "grad_norm": 1.657961130142212,
      "learning_rate": 9.086894736842105e-05,
      "loss": 0.1053,
      "step": 27350
    },
    {
      "epoch": 21.290711232024872,
      "grad_norm": 3.722139596939087,
      "learning_rate": 9.084263157894737e-05,
      "loss": 0.1,
      "step": 27400
    },
    {
      "epoch": 21.329576369996115,
      "grad_norm": 3.5209944248199463,
      "learning_rate": 9.081631578947369e-05,
      "loss": 0.1073,
      "step": 27450
    },
    {
      "epoch": 21.368441507967354,
      "grad_norm": 2.708458662033081,
      "learning_rate": 9.079e-05,
      "loss": 0.1089,
      "step": 27500
    },
    {
      "epoch": 21.407306645938593,
      "grad_norm": 5.232219696044922,
      "learning_rate": 9.076368421052633e-05,
      "loss": 0.1182,
      "step": 27550
    },
    {
      "epoch": 21.446171783909833,
      "grad_norm": 4.402341365814209,
      "learning_rate": 9.073736842105264e-05,
      "loss": 0.1194,
      "step": 27600
    },
    {
      "epoch": 21.485036921881072,
      "grad_norm": 1.9735609292984009,
      "learning_rate": 9.071105263157895e-05,
      "loss": 0.1254,
      "step": 27650
    },
    {
      "epoch": 21.52390205985231,
      "grad_norm": 2.739964485168457,
      "learning_rate": 9.068473684210526e-05,
      "loss": 0.1152,
      "step": 27700
    },
    {
      "epoch": 21.562767197823554,
      "grad_norm": 4.691028594970703,
      "learning_rate": 9.065842105263157e-05,
      "loss": 0.1004,
      "step": 27750
    },
    {
      "epoch": 21.601632335794793,
      "grad_norm": 2.4163448810577393,
      "learning_rate": 9.06321052631579e-05,
      "loss": 0.1202,
      "step": 27800
    },
    {
      "epoch": 21.640497473766032,
      "grad_norm": 3.471006155014038,
      "learning_rate": 9.060578947368421e-05,
      "loss": 0.123,
      "step": 27850
    },
    {
      "epoch": 21.67936261173727,
      "grad_norm": 3.148470640182495,
      "learning_rate": 9.057947368421052e-05,
      "loss": 0.1041,
      "step": 27900
    },
    {
      "epoch": 21.71822774970851,
      "grad_norm": 1.7018976211547852,
      "learning_rate": 9.055315789473685e-05,
      "loss": 0.1223,
      "step": 27950
    },
    {
      "epoch": 21.75709288767975,
      "grad_norm": 1.1783331632614136,
      "learning_rate": 9.052684210526316e-05,
      "loss": 0.125,
      "step": 28000
    },
    {
      "epoch": 21.75709288767975,
      "eval_accuracy": 0.9914978380216684,
      "eval_runtime": 873.9179,
      "eval_samples_per_second": 23.553,
      "eval_steps_per_second": 2.944,
      "step": 28000
    },
    {
      "epoch": 21.79595802565099,
      "grad_norm": 2.917894124984741,
      "learning_rate": 9.050052631578948e-05,
      "loss": 0.1228,
      "step": 28050
    },
    {
      "epoch": 21.834823163622232,
      "grad_norm": 2.4652888774871826,
      "learning_rate": 9.04742105263158e-05,
      "loss": 0.1249,
      "step": 28100
    },
    {
      "epoch": 21.87368830159347,
      "grad_norm": 0.5699157118797302,
      "learning_rate": 9.044789473684211e-05,
      "loss": 0.1147,
      "step": 28150
    },
    {
      "epoch": 21.91255343956471,
      "grad_norm": 0.5322246551513672,
      "learning_rate": 9.042157894736842e-05,
      "loss": 0.1075,
      "step": 28200
    },
    {
      "epoch": 21.95141857753595,
      "grad_norm": 6.01038122177124,
      "learning_rate": 9.039526315789474e-05,
      "loss": 0.1283,
      "step": 28250
    },
    {
      "epoch": 21.99028371550719,
      "grad_norm": 2.6229259967803955,
      "learning_rate": 9.036894736842106e-05,
      "loss": 0.1176,
      "step": 28300
    },
    {
      "epoch": 22.028760202098717,
      "grad_norm": 0.6570966839790344,
      "learning_rate": 9.034263157894737e-05,
      "loss": 0.0926,
      "step": 28350
    },
    {
      "epoch": 22.067625340069956,
      "grad_norm": 3.900761127471924,
      "learning_rate": 9.031631578947369e-05,
      "loss": 0.0874,
      "step": 28400
    },
    {
      "epoch": 22.1064904780412,
      "grad_norm": 1.106946349143982,
      "learning_rate": 9.029e-05,
      "loss": 0.0904,
      "step": 28450
    },
    {
      "epoch": 22.145355616012438,
      "grad_norm": 2.706346035003662,
      "learning_rate": 9.026368421052632e-05,
      "loss": 0.0832,
      "step": 28500
    },
    {
      "epoch": 22.184220753983677,
      "grad_norm": 2.462101459503174,
      "learning_rate": 9.023736842105264e-05,
      "loss": 0.0866,
      "step": 28550
    },
    {
      "epoch": 22.223085891954916,
      "grad_norm": 4.171143054962158,
      "learning_rate": 9.021105263157895e-05,
      "loss": 0.0879,
      "step": 28600
    },
    {
      "epoch": 22.261951029926156,
      "grad_norm": 2.033679962158203,
      "learning_rate": 9.018473684210526e-05,
      "loss": 0.0885,
      "step": 28650
    },
    {
      "epoch": 22.300816167897395,
      "grad_norm": 4.7418365478515625,
      "learning_rate": 9.015842105263158e-05,
      "loss": 0.1195,
      "step": 28700
    },
    {
      "epoch": 22.339681305868634,
      "grad_norm": 2.649731397628784,
      "learning_rate": 9.01321052631579e-05,
      "loss": 0.0883,
      "step": 28750
    },
    {
      "epoch": 22.378546443839877,
      "grad_norm": 0.8859423398971558,
      "learning_rate": 9.010578947368421e-05,
      "loss": 0.0934,
      "step": 28800
    },
    {
      "epoch": 22.417411581811116,
      "grad_norm": 4.102853775024414,
      "learning_rate": 9.007947368421054e-05,
      "loss": 0.0906,
      "step": 28850
    },
    {
      "epoch": 22.456276719782355,
      "grad_norm": 4.6845598220825195,
      "learning_rate": 9.005315789473685e-05,
      "loss": 0.1131,
      "step": 28900
    },
    {
      "epoch": 22.495141857753595,
      "grad_norm": 3.274709463119507,
      "learning_rate": 9.002684210526316e-05,
      "loss": 0.0977,
      "step": 28950
    },
    {
      "epoch": 22.534006995724834,
      "grad_norm": 2.538916826248169,
      "learning_rate": 9.000052631578949e-05,
      "loss": 0.1073,
      "step": 29000
    },
    {
      "epoch": 22.534006995724834,
      "eval_accuracy": 0.9921294271972015,
      "eval_runtime": 876.8455,
      "eval_samples_per_second": 23.474,
      "eval_steps_per_second": 2.934,
      "step": 29000
    },
    {
      "epoch": 22.572872133696073,
      "grad_norm": 2.3162245750427246,
      "learning_rate": 8.99742105263158e-05,
      "loss": 0.0725,
      "step": 29050
    },
    {
      "epoch": 22.611737271667316,
      "grad_norm": 3.547755241394043,
      "learning_rate": 8.994789473684211e-05,
      "loss": 0.0853,
      "step": 29100
    },
    {
      "epoch": 22.650602409638555,
      "grad_norm": 0.8838712573051453,
      "learning_rate": 8.992157894736842e-05,
      "loss": 0.1115,
      "step": 29150
    },
    {
      "epoch": 22.689467547609794,
      "grad_norm": 3.248056650161743,
      "learning_rate": 8.989526315789473e-05,
      "loss": 0.0898,
      "step": 29200
    },
    {
      "epoch": 22.728332685581034,
      "grad_norm": 3.2503175735473633,
      "learning_rate": 8.986894736842106e-05,
      "loss": 0.094,
      "step": 29250
    },
    {
      "epoch": 22.767197823552273,
      "grad_norm": 1.3047415018081665,
      "learning_rate": 8.984263157894738e-05,
      "loss": 0.0974,
      "step": 29300
    },
    {
      "epoch": 22.806062961523512,
      "grad_norm": 3.866671085357666,
      "learning_rate": 8.98163157894737e-05,
      "loss": 0.0923,
      "step": 29350
    },
    {
      "epoch": 22.84492809949475,
      "grad_norm": 4.593476295471191,
      "learning_rate": 8.979e-05,
      "loss": 0.1018,
      "step": 29400
    },
    {
      "epoch": 22.883793237465994,
      "grad_norm": 5.606900691986084,
      "learning_rate": 8.976368421052632e-05,
      "loss": 0.1081,
      "step": 29450
    },
    {
      "epoch": 22.922658375437234,
      "grad_norm": 1.0974065065383911,
      "learning_rate": 8.973736842105264e-05,
      "loss": 0.0936,
      "step": 29500
    },
    {
      "epoch": 22.961523513408473,
      "grad_norm": 3.682856321334839,
      "learning_rate": 8.971105263157895e-05,
      "loss": 0.1205,
      "step": 29550
    },
    {
      "epoch": 23.0,
      "grad_norm": 3.5092780590057373,
      "learning_rate": 8.968473684210527e-05,
      "loss": 0.1111,
      "step": 29600
    },
    {
      "epoch": 23.03886513797124,
      "grad_norm": 0.9673867225646973,
      "learning_rate": 8.965842105263158e-05,
      "loss": 0.067,
      "step": 29650
    },
    {
      "epoch": 23.07773027594248,
      "grad_norm": 0.83358234167099,
      "learning_rate": 8.963210526315789e-05,
      "loss": 0.067,
      "step": 29700
    },
    {
      "epoch": 23.116595413913718,
      "grad_norm": 4.988703727722168,
      "learning_rate": 8.960578947368421e-05,
      "loss": 0.0618,
      "step": 29750
    },
    {
      "epoch": 23.15546055188496,
      "grad_norm": 1.1745929718017578,
      "learning_rate": 8.957947368421054e-05,
      "loss": 0.0752,
      "step": 29800
    },
    {
      "epoch": 23.1943256898562,
      "grad_norm": 3.4321532249450684,
      "learning_rate": 8.955315789473685e-05,
      "loss": 0.0709,
      "step": 29850
    },
    {
      "epoch": 23.23319082782744,
      "grad_norm": 1.3007441759109497,
      "learning_rate": 8.952684210526316e-05,
      "loss": 0.0713,
      "step": 29900
    },
    {
      "epoch": 23.27205596579868,
      "grad_norm": 0.5880366563796997,
      "learning_rate": 8.950052631578947e-05,
      "loss": 0.0781,
      "step": 29950
    },
    {
      "epoch": 23.310921103769918,
      "grad_norm": 0.8208589553833008,
      "learning_rate": 8.94742105263158e-05,
      "loss": 0.0734,
      "step": 30000
    },
    {
      "epoch": 23.310921103769918,
      "eval_accuracy": 0.9923723461108682,
      "eval_runtime": 872.4649,
      "eval_samples_per_second": 23.592,
      "eval_steps_per_second": 2.949,
      "step": 30000
    },
    {
      "epoch": 23.349786241741157,
      "grad_norm": 2.487136125564575,
      "learning_rate": 8.944789473684211e-05,
      "loss": 0.0715,
      "step": 30050
    },
    {
      "epoch": 23.3886513797124,
      "grad_norm": 3.428359031677246,
      "learning_rate": 8.942157894736842e-05,
      "loss": 0.0782,
      "step": 30100
    },
    {
      "epoch": 23.42751651768364,
      "grad_norm": 3.85115122795105,
      "learning_rate": 8.939526315789473e-05,
      "loss": 0.0823,
      "step": 30150
    },
    {
      "epoch": 23.466381655654878,
      "grad_norm": 4.094728469848633,
      "learning_rate": 8.936894736842105e-05,
      "loss": 0.1054,
      "step": 30200
    },
    {
      "epoch": 23.505246793626117,
      "grad_norm": 1.2569011449813843,
      "learning_rate": 8.934263157894738e-05,
      "loss": 0.0792,
      "step": 30250
    },
    {
      "epoch": 23.544111931597357,
      "grad_norm": 1.1779286861419678,
      "learning_rate": 8.93163157894737e-05,
      "loss": 0.0655,
      "step": 30300
    },
    {
      "epoch": 23.582977069568596,
      "grad_norm": 1.402733564376831,
      "learning_rate": 8.929000000000001e-05,
      "loss": 0.0877,
      "step": 30350
    },
    {
      "epoch": 23.621842207539835,
      "grad_norm": 5.030694007873535,
      "learning_rate": 8.926368421052632e-05,
      "loss": 0.0817,
      "step": 30400
    },
    {
      "epoch": 23.660707345511078,
      "grad_norm": 4.230234146118164,
      "learning_rate": 8.923736842105263e-05,
      "loss": 0.0656,
      "step": 30450
    },
    {
      "epoch": 23.699572483482317,
      "grad_norm": 3.24495792388916,
      "learning_rate": 8.921105263157896e-05,
      "loss": 0.0966,
      "step": 30500
    },
    {
      "epoch": 23.738437621453556,
      "grad_norm": 0.8058502078056335,
      "learning_rate": 8.918473684210527e-05,
      "loss": 0.0663,
      "step": 30550
    },
    {
      "epoch": 23.777302759424796,
      "grad_norm": 4.629317760467529,
      "learning_rate": 8.915842105263158e-05,
      "loss": 0.0804,
      "step": 30600
    },
    {
      "epoch": 23.816167897396035,
      "grad_norm": 0.6016934514045715,
      "learning_rate": 8.913210526315789e-05,
      "loss": 0.088,
      "step": 30650
    },
    {
      "epoch": 23.855033035367274,
      "grad_norm": 1.444543719291687,
      "learning_rate": 8.910578947368422e-05,
      "loss": 0.0704,
      "step": 30700
    },
    {
      "epoch": 23.893898173338517,
      "grad_norm": 4.45253849029541,
      "learning_rate": 8.907947368421054e-05,
      "loss": 0.0777,
      "step": 30750
    },
    {
      "epoch": 23.932763311309756,
      "grad_norm": 2.3076658248901367,
      "learning_rate": 8.905315789473685e-05,
      "loss": 0.0978,
      "step": 30800
    },
    {
      "epoch": 23.971628449280995,
      "grad_norm": 1.820786952972412,
      "learning_rate": 8.902684210526316e-05,
      "loss": 0.1152,
      "step": 30850
    },
    {
      "epoch": 24.010104935872523,
      "grad_norm": 1.6674726009368896,
      "learning_rate": 8.900052631578948e-05,
      "loss": 0.0788,
      "step": 30900
    },
    {
      "epoch": 24.048970073843762,
      "grad_norm": 0.5249277949333191,
      "learning_rate": 8.897421052631579e-05,
      "loss": 0.0662,
      "step": 30950
    },
    {
      "epoch": 24.087835211815,
      "grad_norm": 2.1410939693450928,
      "learning_rate": 8.894789473684211e-05,
      "loss": 0.0467,
      "step": 31000
    },
    {
      "epoch": 24.087835211815,
      "eval_accuracy": 0.9940241947238012,
      "eval_runtime": 875.8174,
      "eval_samples_per_second": 23.501,
      "eval_steps_per_second": 2.938,
      "step": 31000
    },
    {
      "epoch": 24.12670034978624,
      "grad_norm": 0.8823385834693909,
      "learning_rate": 8.892157894736842e-05,
      "loss": 0.063,
      "step": 31050
    },
    {
      "epoch": 24.165565487757483,
      "grad_norm": 1.1366829872131348,
      "learning_rate": 8.889526315789474e-05,
      "loss": 0.0589,
      "step": 31100
    },
    {
      "epoch": 24.204430625728723,
      "grad_norm": 0.5657198429107666,
      "learning_rate": 8.886894736842105e-05,
      "loss": 0.0675,
      "step": 31150
    },
    {
      "epoch": 24.24329576369996,
      "grad_norm": 5.198428630828857,
      "learning_rate": 8.884263157894737e-05,
      "loss": 0.071,
      "step": 31200
    },
    {
      "epoch": 24.2821609016712,
      "grad_norm": 2.458716869354248,
      "learning_rate": 8.88163157894737e-05,
      "loss": 0.0699,
      "step": 31250
    },
    {
      "epoch": 24.32102603964244,
      "grad_norm": 1.9143106937408447,
      "learning_rate": 8.879000000000001e-05,
      "loss": 0.0656,
      "step": 31300
    },
    {
      "epoch": 24.35989117761368,
      "grad_norm": 2.3596277236938477,
      "learning_rate": 8.876368421052632e-05,
      "loss": 0.0841,
      "step": 31350
    },
    {
      "epoch": 24.39875631558492,
      "grad_norm": 2.6902995109558105,
      "learning_rate": 8.873736842105263e-05,
      "loss": 0.0592,
      "step": 31400
    },
    {
      "epoch": 24.43762145355616,
      "grad_norm": 0.7977442145347595,
      "learning_rate": 8.871105263157894e-05,
      "loss": 0.066,
      "step": 31450
    },
    {
      "epoch": 24.4764865915274,
      "grad_norm": 1.8103632926940918,
      "learning_rate": 8.868473684210527e-05,
      "loss": 0.0816,
      "step": 31500
    },
    {
      "epoch": 24.51535172949864,
      "grad_norm": 3.2258715629577637,
      "learning_rate": 8.865842105263158e-05,
      "loss": 0.0613,
      "step": 31550
    },
    {
      "epoch": 24.55421686746988,
      "grad_norm": 1.8390077352523804,
      "learning_rate": 8.863210526315789e-05,
      "loss": 0.0788,
      "step": 31600
    },
    {
      "epoch": 24.59308200544112,
      "grad_norm": 1.1279451847076416,
      "learning_rate": 8.860578947368422e-05,
      "loss": 0.0843,
      "step": 31650
    },
    {
      "epoch": 24.631947143412358,
      "grad_norm": 3.265662670135498,
      "learning_rate": 8.857947368421053e-05,
      "loss": 0.0589,
      "step": 31700
    },
    {
      "epoch": 24.6708122813836,
      "grad_norm": 3.05871844291687,
      "learning_rate": 8.855315789473685e-05,
      "loss": 0.0767,
      "step": 31750
    },
    {
      "epoch": 24.70967741935484,
      "grad_norm": 3.3928933143615723,
      "learning_rate": 8.852684210526317e-05,
      "loss": 0.064,
      "step": 31800
    },
    {
      "epoch": 24.74854255732608,
      "grad_norm": 2.3093581199645996,
      "learning_rate": 8.850052631578948e-05,
      "loss": 0.0756,
      "step": 31850
    },
    {
      "epoch": 24.78740769529732,
      "grad_norm": 1.7362569570541382,
      "learning_rate": 8.847421052631579e-05,
      "loss": 0.0682,
      "step": 31900
    },
    {
      "epoch": 24.826272833268558,
      "grad_norm": 2.4637110233306885,
      "learning_rate": 8.844789473684211e-05,
      "loss": 0.0677,
      "step": 31950
    },
    {
      "epoch": 24.865137971239797,
      "grad_norm": 2.16152024269104,
      "learning_rate": 8.842157894736843e-05,
      "loss": 0.0687,
      "step": 32000
    },
    {
      "epoch": 24.865137971239797,
      "eval_accuracy": 0.9946557838993344,
      "eval_runtime": 876.7622,
      "eval_samples_per_second": 23.476,
      "eval_steps_per_second": 2.935,
      "step": 32000
    },
    {
      "epoch": 24.904003109211036,
      "grad_norm": 1.6575545072555542,
      "learning_rate": 8.839526315789474e-05,
      "loss": 0.0714,
      "step": 32050
    },
    {
      "epoch": 24.94286824718228,
      "grad_norm": 3.0430097579956055,
      "learning_rate": 8.836894736842106e-05,
      "loss": 0.0942,
      "step": 32100
    },
    {
      "epoch": 24.981733385153518,
      "grad_norm": 1.7112277746200562,
      "learning_rate": 8.834263157894737e-05,
      "loss": 0.0727,
      "step": 32150
    },
    {
      "epoch": 25.020209871745045,
      "grad_norm": 1.13712477684021,
      "learning_rate": 8.831631578947369e-05,
      "loss": 0.0634,
      "step": 32200
    },
    {
      "epoch": 25.059075009716285,
      "grad_norm": 0.15798869729042053,
      "learning_rate": 8.829000000000001e-05,
      "loss": 0.06,
      "step": 32250
    },
    {
      "epoch": 25.097940147687524,
      "grad_norm": 3.0741162300109863,
      "learning_rate": 8.826368421052632e-05,
      "loss": 0.0749,
      "step": 32300
    },
    {
      "epoch": 25.136805285658763,
      "grad_norm": 0.9993603825569153,
      "learning_rate": 8.823736842105263e-05,
      "loss": 0.0492,
      "step": 32350
    },
    {
      "epoch": 25.175670423630002,
      "grad_norm": 0.929404616355896,
      "learning_rate": 8.821105263157894e-05,
      "loss": 0.0507,
      "step": 32400
    },
    {
      "epoch": 25.214535561601245,
      "grad_norm": 1.5012816190719604,
      "learning_rate": 8.818473684210527e-05,
      "loss": 0.0566,
      "step": 32450
    },
    {
      "epoch": 25.253400699572484,
      "grad_norm": 0.46388107538223267,
      "learning_rate": 8.815842105263158e-05,
      "loss": 0.0594,
      "step": 32500
    },
    {
      "epoch": 25.292265837543724,
      "grad_norm": 2.559751272201538,
      "learning_rate": 8.81321052631579e-05,
      "loss": 0.0619,
      "step": 32550
    },
    {
      "epoch": 25.331130975514963,
      "grad_norm": 0.8673915863037109,
      "learning_rate": 8.810578947368422e-05,
      "loss": 0.053,
      "step": 32600
    },
    {
      "epoch": 25.369996113486202,
      "grad_norm": 4.02969217300415,
      "learning_rate": 8.807947368421053e-05,
      "loss": 0.0521,
      "step": 32650
    },
    {
      "epoch": 25.40886125145744,
      "grad_norm": 0.8806849718093872,
      "learning_rate": 8.805315789473686e-05,
      "loss": 0.0557,
      "step": 32700
    },
    {
      "epoch": 25.44772638942868,
      "grad_norm": 1.1096513271331787,
      "learning_rate": 8.802684210526317e-05,
      "loss": 0.0641,
      "step": 32750
    },
    {
      "epoch": 25.486591527399924,
      "grad_norm": 0.36731696128845215,
      "learning_rate": 8.800052631578948e-05,
      "loss": 0.0534,
      "step": 32800
    },
    {
      "epoch": 25.525456665371163,
      "grad_norm": 0.8237600922584534,
      "learning_rate": 8.797421052631579e-05,
      "loss": 0.0593,
      "step": 32850
    },
    {
      "epoch": 25.564321803342402,
      "grad_norm": 2.0420188903808594,
      "learning_rate": 8.79478947368421e-05,
      "loss": 0.0576,
      "step": 32900
    },
    {
      "epoch": 25.60318694131364,
      "grad_norm": 3.4013640880584717,
      "learning_rate": 8.792157894736843e-05,
      "loss": 0.0657,
      "step": 32950
    },
    {
      "epoch": 25.64205207928488,
      "grad_norm": 2.2952609062194824,
      "learning_rate": 8.789526315789474e-05,
      "loss": 0.0565,
      "step": 33000
    },
    {
      "epoch": 25.64205207928488,
      "eval_accuracy": 0.9952873730748676,
      "eval_runtime": 878.0538,
      "eval_samples_per_second": 23.442,
      "eval_steps_per_second": 2.93,
      "step": 33000
    },
    {
      "epoch": 25.68091721725612,
      "grad_norm": 4.091900825500488,
      "learning_rate": 8.786894736842106e-05,
      "loss": 0.0707,
      "step": 33050
    },
    {
      "epoch": 25.719782355227363,
      "grad_norm": 2.2345447540283203,
      "learning_rate": 8.784263157894737e-05,
      "loss": 0.0661,
      "step": 33100
    },
    {
      "epoch": 25.758647493198602,
      "grad_norm": 5.394659996032715,
      "learning_rate": 8.781631578947369e-05,
      "loss": 0.0703,
      "step": 33150
    },
    {
      "epoch": 25.79751263116984,
      "grad_norm": 2.6766786575317383,
      "learning_rate": 8.779000000000001e-05,
      "loss": 0.0531,
      "step": 33200
    },
    {
      "epoch": 25.83637776914108,
      "grad_norm": 0.757504403591156,
      "learning_rate": 8.776368421052632e-05,
      "loss": 0.0657,
      "step": 33250
    },
    {
      "epoch": 25.87524290711232,
      "grad_norm": 9.220613479614258,
      "learning_rate": 8.773736842105263e-05,
      "loss": 0.0639,
      "step": 33300
    },
    {
      "epoch": 25.91410804508356,
      "grad_norm": 0.38048455119132996,
      "learning_rate": 8.771105263157895e-05,
      "loss": 0.0583,
      "step": 33350
    },
    {
      "epoch": 25.9529731830548,
      "grad_norm": 0.569552481174469,
      "learning_rate": 8.768473684210526e-05,
      "loss": 0.0629,
      "step": 33400
    },
    {
      "epoch": 25.99183832102604,
      "grad_norm": 0.9764150977134705,
      "learning_rate": 8.765842105263158e-05,
      "loss": 0.0684,
      "step": 33450
    },
    {
      "epoch": 26.030314807617568,
      "grad_norm": 0.6928421854972839,
      "learning_rate": 8.763210526315791e-05,
      "loss": 0.0578,
      "step": 33500
    },
    {
      "epoch": 26.069179945588807,
      "grad_norm": 2.852064371109009,
      "learning_rate": 8.760578947368422e-05,
      "loss": 0.046,
      "step": 33550
    },
    {
      "epoch": 26.108045083560047,
      "grad_norm": 2.2197632789611816,
      "learning_rate": 8.757947368421053e-05,
      "loss": 0.0379,
      "step": 33600
    },
    {
      "epoch": 26.146910221531286,
      "grad_norm": 0.2494724988937378,
      "learning_rate": 8.755315789473684e-05,
      "loss": 0.0444,
      "step": 33650
    },
    {
      "epoch": 26.185775359502525,
      "grad_norm": 0.8166674971580505,
      "learning_rate": 8.752684210526317e-05,
      "loss": 0.0531,
      "step": 33700
    },
    {
      "epoch": 26.224640497473764,
      "grad_norm": 2.2353274822235107,
      "learning_rate": 8.750052631578948e-05,
      "loss": 0.0627,
      "step": 33750
    },
    {
      "epoch": 26.263505635445007,
      "grad_norm": 1.0710805654525757,
      "learning_rate": 8.747421052631579e-05,
      "loss": 0.047,
      "step": 33800
    },
    {
      "epoch": 26.302370773416246,
      "grad_norm": 1.9708255529403687,
      "learning_rate": 8.74478947368421e-05,
      "loss": 0.051,
      "step": 33850
    },
    {
      "epoch": 26.341235911387486,
      "grad_norm": 1.2529079914093018,
      "learning_rate": 8.742157894736841e-05,
      "loss": 0.0492,
      "step": 33900
    },
    {
      "epoch": 26.380101049358725,
      "grad_norm": 0.9265830516815186,
      "learning_rate": 8.739526315789474e-05,
      "loss": 0.0555,
      "step": 33950
    },
    {
      "epoch": 26.418966187329964,
      "grad_norm": 3.131944179534912,
      "learning_rate": 8.736894736842106e-05,
      "loss": 0.0473,
      "step": 34000
    },
    {
      "epoch": 26.418966187329964,
      "eval_accuracy": 0.995044454161201,
      "eval_runtime": 876.168,
      "eval_samples_per_second": 23.492,
      "eval_steps_per_second": 2.937,
      "step": 34000
    },
    {
      "epoch": 26.457831325301203,
      "grad_norm": 0.6346553564071655,
      "learning_rate": 8.734263157894738e-05,
      "loss": 0.0498,
      "step": 34050
    },
    {
      "epoch": 26.496696463272446,
      "grad_norm": 0.4876219928264618,
      "learning_rate": 8.731631578947369e-05,
      "loss": 0.0483,
      "step": 34100
    },
    {
      "epoch": 26.535561601243685,
      "grad_norm": 1.6699252128601074,
      "learning_rate": 8.729e-05,
      "loss": 0.0525,
      "step": 34150
    },
    {
      "epoch": 26.574426739214925,
      "grad_norm": 2.112086772918701,
      "learning_rate": 8.726368421052632e-05,
      "loss": 0.053,
      "step": 34200
    },
    {
      "epoch": 26.613291877186164,
      "grad_norm": 1.6960245370864868,
      "learning_rate": 8.723736842105264e-05,
      "loss": 0.048,
      "step": 34250
    },
    {
      "epoch": 26.652157015157403,
      "grad_norm": 3.9964354038238525,
      "learning_rate": 8.721105263157895e-05,
      "loss": 0.0606,
      "step": 34300
    },
    {
      "epoch": 26.691022153128642,
      "grad_norm": 4.299483299255371,
      "learning_rate": 8.718473684210526e-05,
      "loss": 0.0387,
      "step": 34350
    },
    {
      "epoch": 26.72988729109988,
      "grad_norm": 0.5646001100540161,
      "learning_rate": 8.715842105263158e-05,
      "loss": 0.0459,
      "step": 34400
    },
    {
      "epoch": 26.768752429071125,
      "grad_norm": 3.601149320602417,
      "learning_rate": 8.713210526315791e-05,
      "loss": 0.0762,
      "step": 34450
    },
    {
      "epoch": 26.807617567042364,
      "grad_norm": 5.873041152954102,
      "learning_rate": 8.710578947368422e-05,
      "loss": 0.0664,
      "step": 34500
    },
    {
      "epoch": 26.846482705013603,
      "grad_norm": 0.9744375348091125,
      "learning_rate": 8.707947368421053e-05,
      "loss": 0.0552,
      "step": 34550
    },
    {
      "epoch": 26.885347842984842,
      "grad_norm": 1.381766676902771,
      "learning_rate": 8.705315789473684e-05,
      "loss": 0.0645,
      "step": 34600
    },
    {
      "epoch": 26.92421298095608,
      "grad_norm": 3.265528678894043,
      "learning_rate": 8.702684210526316e-05,
      "loss": 0.0649,
      "step": 34650
    },
    {
      "epoch": 26.96307811892732,
      "grad_norm": 2.2487645149230957,
      "learning_rate": 8.700052631578948e-05,
      "loss": 0.0714,
      "step": 34700
    },
    {
      "epoch": 27.001554605518848,
      "grad_norm": 0.22899124026298523,
      "learning_rate": 8.697421052631579e-05,
      "loss": 0.0585,
      "step": 34750
    },
    {
      "epoch": 27.04041974349009,
      "grad_norm": 2.3547112941741943,
      "learning_rate": 8.69478947368421e-05,
      "loss": 0.0357,
      "step": 34800
    },
    {
      "epoch": 27.07928488146133,
      "grad_norm": 0.5145918130874634,
      "learning_rate": 8.692157894736842e-05,
      "loss": 0.0493,
      "step": 34850
    },
    {
      "epoch": 27.11815001943257,
      "grad_norm": 3.8266923427581787,
      "learning_rate": 8.689526315789474e-05,
      "loss": 0.0313,
      "step": 34900
    },
    {
      "epoch": 27.15701515740381,
      "grad_norm": 0.9876055717468262,
      "learning_rate": 8.686894736842107e-05,
      "loss": 0.0416,
      "step": 34950
    },
    {
      "epoch": 27.195880295375048,
      "grad_norm": 0.4796826243400574,
      "learning_rate": 8.684263157894738e-05,
      "loss": 0.0386,
      "step": 35000
    },
    {
      "epoch": 27.195880295375048,
      "eval_accuracy": 0.9959189622504008,
      "eval_runtime": 876.7261,
      "eval_samples_per_second": 23.477,
      "eval_steps_per_second": 2.935,
      "step": 35000
    },
    {
      "epoch": 27.234745433346287,
      "grad_norm": 0.3122335970401764,
      "learning_rate": 8.681631578947369e-05,
      "loss": 0.05,
      "step": 35050
    },
    {
      "epoch": 27.27361057131753,
      "grad_norm": 1.78782320022583,
      "learning_rate": 8.679e-05,
      "loss": 0.039,
      "step": 35100
    },
    {
      "epoch": 27.31247570928877,
      "grad_norm": 2.0089144706726074,
      "learning_rate": 8.676368421052633e-05,
      "loss": 0.0407,
      "step": 35150
    },
    {
      "epoch": 27.35134084726001,
      "grad_norm": 0.6602781414985657,
      "learning_rate": 8.673736842105264e-05,
      "loss": 0.0439,
      "step": 35200
    },
    {
      "epoch": 27.390205985231248,
      "grad_norm": 1.503353476524353,
      "learning_rate": 8.671105263157895e-05,
      "loss": 0.0458,
      "step": 35250
    },
    {
      "epoch": 27.429071123202487,
      "grad_norm": 3.133200168609619,
      "learning_rate": 8.668473684210526e-05,
      "loss": 0.0455,
      "step": 35300
    },
    {
      "epoch": 27.467936261173726,
      "grad_norm": 3.797276496887207,
      "learning_rate": 8.665842105263157e-05,
      "loss": 0.0444,
      "step": 35350
    },
    {
      "epoch": 27.506801399144965,
      "grad_norm": 1.8748810291290283,
      "learning_rate": 8.66321052631579e-05,
      "loss": 0.0445,
      "step": 35400
    },
    {
      "epoch": 27.545666537116208,
      "grad_norm": 1.198071837425232,
      "learning_rate": 8.660578947368422e-05,
      "loss": 0.0544,
      "step": 35450
    },
    {
      "epoch": 27.584531675087447,
      "grad_norm": 0.8296390175819397,
      "learning_rate": 8.657947368421053e-05,
      "loss": 0.0334,
      "step": 35500
    },
    {
      "epoch": 27.623396813058687,
      "grad_norm": 0.7118576169013977,
      "learning_rate": 8.655315789473685e-05,
      "loss": 0.0408,
      "step": 35550
    },
    {
      "epoch": 27.662261951029926,
      "grad_norm": 2.786017656326294,
      "learning_rate": 8.652684210526316e-05,
      "loss": 0.0474,
      "step": 35600
    },
    {
      "epoch": 27.701127089001165,
      "grad_norm": 0.6719661951065063,
      "learning_rate": 8.650052631578948e-05,
      "loss": 0.0586,
      "step": 35650
    },
    {
      "epoch": 27.739992226972404,
      "grad_norm": 1.9458692073822021,
      "learning_rate": 8.64742105263158e-05,
      "loss": 0.0403,
      "step": 35700
    },
    {
      "epoch": 27.778857364943647,
      "grad_norm": 2.8628056049346924,
      "learning_rate": 8.64478947368421e-05,
      "loss": 0.0491,
      "step": 35750
    },
    {
      "epoch": 27.817722502914886,
      "grad_norm": 3.354652166366577,
      "learning_rate": 8.642157894736842e-05,
      "loss": 0.0623,
      "step": 35800
    },
    {
      "epoch": 27.856587640886126,
      "grad_norm": 3.452547788619995,
      "learning_rate": 8.639526315789474e-05,
      "loss": 0.0564,
      "step": 35850
    },
    {
      "epoch": 27.895452778857365,
      "grad_norm": 2.5311927795410156,
      "learning_rate": 8.636894736842105e-05,
      "loss": 0.0505,
      "step": 35900
    },
    {
      "epoch": 27.934317916828604,
      "grad_norm": 2.357215642929077,
      "learning_rate": 8.634263157894738e-05,
      "loss": 0.0387,
      "step": 35950
    },
    {
      "epoch": 27.973183054799843,
      "grad_norm": 2.599184036254883,
      "learning_rate": 8.631631578947369e-05,
      "loss": 0.0565,
      "step": 36000
    },
    {
      "epoch": 27.973183054799843,
      "eval_accuracy": 0.9960161298158675,
      "eval_runtime": 876.7552,
      "eval_samples_per_second": 23.476,
      "eval_steps_per_second": 2.935,
      "step": 36000
    },
    {
      "epoch": 28.01165954139137,
      "grad_norm": 0.7705150246620178,
      "learning_rate": 8.629e-05,
      "loss": 0.0467,
      "step": 36050
    },
    {
      "epoch": 28.05052467936261,
      "grad_norm": 3.2739124298095703,
      "learning_rate": 8.626368421052631e-05,
      "loss": 0.0295,
      "step": 36100
    },
    {
      "epoch": 28.089389817333853,
      "grad_norm": 0.19782505929470062,
      "learning_rate": 8.623736842105264e-05,
      "loss": 0.0394,
      "step": 36150
    },
    {
      "epoch": 28.128254955305092,
      "grad_norm": 3.2549073696136475,
      "learning_rate": 8.621105263157895e-05,
      "loss": 0.0284,
      "step": 36200
    },
    {
      "epoch": 28.16712009327633,
      "grad_norm": 0.6249772310256958,
      "learning_rate": 8.618473684210526e-05,
      "loss": 0.0537,
      "step": 36250
    },
    {
      "epoch": 28.20598523124757,
      "grad_norm": 2.322044610977173,
      "learning_rate": 8.615842105263159e-05,
      "loss": 0.0506,
      "step": 36300
    },
    {
      "epoch": 28.24485036921881,
      "grad_norm": 1.3696391582489014,
      "learning_rate": 8.61321052631579e-05,
      "loss": 0.0373,
      "step": 36350
    },
    {
      "epoch": 28.28371550719005,
      "grad_norm": 0.34534499049186707,
      "learning_rate": 8.610578947368422e-05,
      "loss": 0.0431,
      "step": 36400
    },
    {
      "epoch": 28.322580645161292,
      "grad_norm": 1.8817704916000366,
      "learning_rate": 8.607947368421054e-05,
      "loss": 0.0459,
      "step": 36450
    },
    {
      "epoch": 28.36144578313253,
      "grad_norm": 3.955824851989746,
      "learning_rate": 8.605315789473685e-05,
      "loss": 0.0485,
      "step": 36500
    },
    {
      "epoch": 28.40031092110377,
      "grad_norm": 0.8568444848060608,
      "learning_rate": 8.602684210526316e-05,
      "loss": 0.0366,
      "step": 36550
    },
    {
      "epoch": 28.43917605907501,
      "grad_norm": 2.062861442565918,
      "learning_rate": 8.600052631578947e-05,
      "loss": 0.0382,
      "step": 36600
    },
    {
      "epoch": 28.47804119704625,
      "grad_norm": 0.24858438968658447,
      "learning_rate": 8.59742105263158e-05,
      "loss": 0.0453,
      "step": 36650
    },
    {
      "epoch": 28.516906335017488,
      "grad_norm": 1.0825531482696533,
      "learning_rate": 8.594789473684211e-05,
      "loss": 0.0384,
      "step": 36700
    },
    {
      "epoch": 28.55577147298873,
      "grad_norm": 1.907128095626831,
      "learning_rate": 8.592157894736842e-05,
      "loss": 0.0483,
      "step": 36750
    },
    {
      "epoch": 28.59463661095997,
      "grad_norm": 1.5113343000411987,
      "learning_rate": 8.589526315789474e-05,
      "loss": 0.0512,
      "step": 36800
    },
    {
      "epoch": 28.63350174893121,
      "grad_norm": 3.101874828338623,
      "learning_rate": 8.586894736842106e-05,
      "loss": 0.0401,
      "step": 36850
    },
    {
      "epoch": 28.67236688690245,
      "grad_norm": 2.6152641773223877,
      "learning_rate": 8.584263157894738e-05,
      "loss": 0.0415,
      "step": 36900
    },
    {
      "epoch": 28.711232024873688,
      "grad_norm": 4.213351726531982,
      "learning_rate": 8.581631578947369e-05,
      "loss": 0.0266,
      "step": 36950
    },
    {
      "epoch": 28.750097162844927,
      "grad_norm": 3.643454074859619,
      "learning_rate": 8.579e-05,
      "loss": 0.0584,
      "step": 37000
    },
    {
      "epoch": 28.750097162844927,
      "eval_accuracy": 0.9965505514259341,
      "eval_runtime": 878.5114,
      "eval_samples_per_second": 23.429,
      "eval_steps_per_second": 2.929,
      "step": 37000
    },
    {
      "epoch": 28.788962300816166,
      "grad_norm": 2.3031961917877197,
      "learning_rate": 8.576368421052632e-05,
      "loss": 0.0592,
      "step": 37050
    },
    {
      "epoch": 28.82782743878741,
      "grad_norm": 0.538669764995575,
      "learning_rate": 8.573736842105263e-05,
      "loss": 0.0423,
      "step": 37100
    },
    {
      "epoch": 28.86669257675865,
      "grad_norm": 4.017199993133545,
      "learning_rate": 8.571105263157895e-05,
      "loss": 0.0499,
      "step": 37150
    },
    {
      "epoch": 28.905557714729888,
      "grad_norm": 2.1706478595733643,
      "learning_rate": 8.568473684210526e-05,
      "loss": 0.0395,
      "step": 37200
    },
    {
      "epoch": 28.944422852701127,
      "grad_norm": 4.465363025665283,
      "learning_rate": 8.565842105263159e-05,
      "loss": 0.0491,
      "step": 37250
    },
    {
      "epoch": 28.983287990672366,
      "grad_norm": 4.176932334899902,
      "learning_rate": 8.56321052631579e-05,
      "loss": 0.0595,
      "step": 37300
    },
    {
      "epoch": 29.021764477263893,
      "grad_norm": 5.255528926849365,
      "learning_rate": 8.560578947368421e-05,
      "loss": 0.0385,
      "step": 37350
    },
    {
      "epoch": 29.060629615235133,
      "grad_norm": 3.7985963821411133,
      "learning_rate": 8.557947368421054e-05,
      "loss": 0.0308,
      "step": 37400
    },
    {
      "epoch": 29.099494753206375,
      "grad_norm": 1.006502628326416,
      "learning_rate": 8.555315789473685e-05,
      "loss": 0.0369,
      "step": 37450
    },
    {
      "epoch": 29.138359891177615,
      "grad_norm": 0.36928635835647583,
      "learning_rate": 8.552684210526316e-05,
      "loss": 0.0335,
      "step": 37500
    },
    {
      "epoch": 29.177225029148854,
      "grad_norm": 1.71954345703125,
      "learning_rate": 8.550052631578947e-05,
      "loss": 0.0383,
      "step": 37550
    },
    {
      "epoch": 29.216090167120093,
      "grad_norm": 0.19994154572486877,
      "learning_rate": 8.547421052631578e-05,
      "loss": 0.0379,
      "step": 37600
    },
    {
      "epoch": 29.254955305091332,
      "grad_norm": 1.6422315835952759,
      "learning_rate": 8.544789473684211e-05,
      "loss": 0.0365,
      "step": 37650
    },
    {
      "epoch": 29.29382044306257,
      "grad_norm": 0.8327766060829163,
      "learning_rate": 8.542157894736843e-05,
      "loss": 0.0383,
      "step": 37700
    },
    {
      "epoch": 29.33268558103381,
      "grad_norm": 0.69755619764328,
      "learning_rate": 8.539526315789475e-05,
      "loss": 0.039,
      "step": 37750
    },
    {
      "epoch": 29.371550719005054,
      "grad_norm": 0.969310462474823,
      "learning_rate": 8.536894736842106e-05,
      "loss": 0.0351,
      "step": 37800
    },
    {
      "epoch": 29.410415856976293,
      "grad_norm": 3.9037134647369385,
      "learning_rate": 8.534263157894737e-05,
      "loss": 0.0336,
      "step": 37850
    },
    {
      "epoch": 29.449280994947532,
      "grad_norm": 6.574547290802002,
      "learning_rate": 8.53163157894737e-05,
      "loss": 0.0318,
      "step": 37900
    },
    {
      "epoch": 29.48814613291877,
      "grad_norm": 0.7531656622886658,
      "learning_rate": 8.529e-05,
      "loss": 0.0372,
      "step": 37950
    },
    {
      "epoch": 29.52701127089001,
      "grad_norm": 3.7106435298919678,
      "learning_rate": 8.526368421052632e-05,
      "loss": 0.0369,
      "step": 38000
    },
    {
      "epoch": 29.52701127089001,
      "eval_accuracy": 0.9972307243842006,
      "eval_runtime": 877.9929,
      "eval_samples_per_second": 23.443,
      "eval_steps_per_second": 2.931,
      "step": 38000
    },
    {
      "epoch": 29.56587640886125,
      "grad_norm": 3.104320526123047,
      "learning_rate": 8.523736842105263e-05,
      "loss": 0.0403,
      "step": 38050
    },
    {
      "epoch": 29.604741546832493,
      "grad_norm": 0.4907873272895813,
      "learning_rate": 8.521105263157895e-05,
      "loss": 0.0412,
      "step": 38100
    },
    {
      "epoch": 29.643606684803732,
      "grad_norm": 0.19150598347187042,
      "learning_rate": 8.518473684210528e-05,
      "loss": 0.0417,
      "step": 38150
    },
    {
      "epoch": 29.68247182277497,
      "grad_norm": 1.1941745281219482,
      "learning_rate": 8.515842105263159e-05,
      "loss": 0.0335,
      "step": 38200
    },
    {
      "epoch": 29.72133696074621,
      "grad_norm": 0.36566030979156494,
      "learning_rate": 8.51321052631579e-05,
      "loss": 0.0461,
      "step": 38250
    },
    {
      "epoch": 29.76020209871745,
      "grad_norm": 0.7068200707435608,
      "learning_rate": 8.510578947368421e-05,
      "loss": 0.0462,
      "step": 38300
    },
    {
      "epoch": 29.79906723668869,
      "grad_norm": 0.6579626202583313,
      "learning_rate": 8.507947368421052e-05,
      "loss": 0.0383,
      "step": 38350
    },
    {
      "epoch": 29.83793237465993,
      "grad_norm": 2.1728909015655518,
      "learning_rate": 8.505315789473685e-05,
      "loss": 0.0398,
      "step": 38400
    },
    {
      "epoch": 29.87679751263117,
      "grad_norm": 0.7398192286491394,
      "learning_rate": 8.502684210526316e-05,
      "loss": 0.0483,
      "step": 38450
    },
    {
      "epoch": 29.91566265060241,
      "grad_norm": 2.1343936920166016,
      "learning_rate": 8.500052631578947e-05,
      "loss": 0.0519,
      "step": 38500
    },
    {
      "epoch": 29.95452778857365,
      "grad_norm": 0.2378653883934021,
      "learning_rate": 8.497421052631578e-05,
      "loss": 0.0384,
      "step": 38550
    },
    {
      "epoch": 29.99339292654489,
      "grad_norm": 2.9573421478271484,
      "learning_rate": 8.494789473684211e-05,
      "loss": 0.0541,
      "step": 38600
    },
    {
      "epoch": 30.031869413136416,
      "grad_norm": 0.5036262273788452,
      "learning_rate": 8.492157894736843e-05,
      "loss": 0.0295,
      "step": 38650
    },
    {
      "epoch": 30.070734551107655,
      "grad_norm": 1.123215913772583,
      "learning_rate": 8.489526315789475e-05,
      "loss": 0.0347,
      "step": 38700
    },
    {
      "epoch": 30.109599689078895,
      "grad_norm": 3.358325958251953,
      "learning_rate": 8.486894736842106e-05,
      "loss": 0.043,
      "step": 38750
    },
    {
      "epoch": 30.148464827050137,
      "grad_norm": 1.2945706844329834,
      "learning_rate": 8.484263157894737e-05,
      "loss": 0.0274,
      "step": 38800
    },
    {
      "epoch": 30.187329965021377,
      "grad_norm": 0.08089222759008408,
      "learning_rate": 8.48163157894737e-05,
      "loss": 0.026,
      "step": 38850
    },
    {
      "epoch": 30.226195102992616,
      "grad_norm": 0.9981313347816467,
      "learning_rate": 8.479e-05,
      "loss": 0.0283,
      "step": 38900
    },
    {
      "epoch": 30.265060240963855,
      "grad_norm": 2.40240740776062,
      "learning_rate": 8.476368421052632e-05,
      "loss": 0.0409,
      "step": 38950
    },
    {
      "epoch": 30.303925378935094,
      "grad_norm": 2.257906913757324,
      "learning_rate": 8.473736842105263e-05,
      "loss": 0.0336,
      "step": 39000
    },
    {
      "epoch": 30.303925378935094,
      "eval_accuracy": 0.996696302774134,
      "eval_runtime": 876.703,
      "eval_samples_per_second": 23.478,
      "eval_steps_per_second": 2.935,
      "step": 39000
    },
    {
      "epoch": 30.342790516906334,
      "grad_norm": 3.1069672107696533,
      "learning_rate": 8.471105263157894e-05,
      "loss": 0.0354,
      "step": 39050
    },
    {
      "epoch": 30.381655654877576,
      "grad_norm": 0.984092652797699,
      "learning_rate": 8.468473684210527e-05,
      "loss": 0.0283,
      "step": 39100
    },
    {
      "epoch": 30.420520792848816,
      "grad_norm": 1.6073585748672485,
      "learning_rate": 8.465842105263159e-05,
      "loss": 0.0329,
      "step": 39150
    },
    {
      "epoch": 30.459385930820055,
      "grad_norm": 0.9413264989852905,
      "learning_rate": 8.46321052631579e-05,
      "loss": 0.0355,
      "step": 39200
    },
    {
      "epoch": 30.498251068791294,
      "grad_norm": 0.16653196513652802,
      "learning_rate": 8.460578947368421e-05,
      "loss": 0.0349,
      "step": 39250
    },
    {
      "epoch": 30.537116206762533,
      "grad_norm": 2.087178945541382,
      "learning_rate": 8.457947368421053e-05,
      "loss": 0.0516,
      "step": 39300
    },
    {
      "epoch": 30.575981344733773,
      "grad_norm": 0.6503152847290039,
      "learning_rate": 8.455315789473685e-05,
      "loss": 0.0403,
      "step": 39350
    },
    {
      "epoch": 30.614846482705012,
      "grad_norm": 0.18071837723255157,
      "learning_rate": 8.452684210526316e-05,
      "loss": 0.0291,
      "step": 39400
    },
    {
      "epoch": 30.653711620676255,
      "grad_norm": 0.3294931650161743,
      "learning_rate": 8.450052631578947e-05,
      "loss": 0.0266,
      "step": 39450
    },
    {
      "epoch": 30.692576758647494,
      "grad_norm": 2.715810537338257,
      "learning_rate": 8.447421052631579e-05,
      "loss": 0.0331,
      "step": 39500
    },
    {
      "epoch": 30.731441896618733,
      "grad_norm": 0.7051801085472107,
      "learning_rate": 8.444789473684211e-05,
      "loss": 0.0353,
      "step": 39550
    },
    {
      "epoch": 30.770307034589973,
      "grad_norm": 0.311470627784729,
      "learning_rate": 8.442157894736842e-05,
      "loss": 0.0304,
      "step": 39600
    },
    {
      "epoch": 30.80917217256121,
      "grad_norm": 0.666933536529541,
      "learning_rate": 8.439526315789475e-05,
      "loss": 0.0304,
      "step": 39650
    },
    {
      "epoch": 30.84803731053245,
      "grad_norm": 3.9159634113311768,
      "learning_rate": 8.436894736842106e-05,
      "loss": 0.0424,
      "step": 39700
    },
    {
      "epoch": 30.886902448503694,
      "grad_norm": 1.1517620086669922,
      "learning_rate": 8.434263157894737e-05,
      "loss": 0.0389,
      "step": 39750
    },
    {
      "epoch": 30.925767586474933,
      "grad_norm": 2.5502254962921143,
      "learning_rate": 8.431631578947368e-05,
      "loss": 0.0405,
      "step": 39800
    },
    {
      "epoch": 30.964632724446172,
      "grad_norm": 2.4017560482025146,
      "learning_rate": 8.429000000000001e-05,
      "loss": 0.0317,
      "step": 39850
    },
    {
      "epoch": 31.0031092110377,
      "grad_norm": 1.4030591249465942,
      "learning_rate": 8.426368421052632e-05,
      "loss": 0.0405,
      "step": 39900
    },
    {
      "epoch": 31.04197434900894,
      "grad_norm": 3.136474847793579,
      "learning_rate": 8.423736842105263e-05,
      "loss": 0.0235,
      "step": 39950
    },
    {
      "epoch": 31.080839486980178,
      "grad_norm": 0.37211865186691284,
      "learning_rate": 8.421105263157894e-05,
      "loss": 0.0168,
      "step": 40000
    },
    {
      "epoch": 31.080839486980178,
      "eval_accuracy": 0.9970363892532672,
      "eval_runtime": 873.7751,
      "eval_samples_per_second": 23.556,
      "eval_steps_per_second": 2.945,
      "step": 40000
    },
    {
      "epoch": 31.119704624951417,
      "grad_norm": 1.3298871517181396,
      "learning_rate": 8.418473684210527e-05,
      "loss": 0.0258,
      "step": 40050
    },
    {
      "epoch": 31.15856976292266,
      "grad_norm": 0.12341593205928802,
      "learning_rate": 8.415842105263159e-05,
      "loss": 0.0387,
      "step": 40100
    },
    {
      "epoch": 31.1974349008939,
      "grad_norm": 0.411594420671463,
      "learning_rate": 8.41321052631579e-05,
      "loss": 0.0306,
      "step": 40150
    },
    {
      "epoch": 31.23630003886514,
      "grad_norm": 1.527314305305481,
      "learning_rate": 8.410578947368422e-05,
      "loss": 0.0275,
      "step": 40200
    },
    {
      "epoch": 31.275165176836378,
      "grad_norm": 1.303266167640686,
      "learning_rate": 8.407947368421053e-05,
      "loss": 0.0225,
      "step": 40250
    },
    {
      "epoch": 31.314030314807617,
      "grad_norm": 4.040980815887451,
      "learning_rate": 8.405315789473684e-05,
      "loss": 0.0302,
      "step": 40300
    },
    {
      "epoch": 31.352895452778856,
      "grad_norm": 95.21772766113281,
      "learning_rate": 8.402684210526316e-05,
      "loss": 0.0341,
      "step": 40350
    },
    {
      "epoch": 31.391760590750096,
      "grad_norm": 2.755831241607666,
      "learning_rate": 8.400052631578948e-05,
      "loss": 0.0219,
      "step": 40400
    },
    {
      "epoch": 31.43062572872134,
      "grad_norm": 0.8875411748886108,
      "learning_rate": 8.397421052631579e-05,
      "loss": 0.0312,
      "step": 40450
    },
    {
      "epoch": 31.469490866692578,
      "grad_norm": 0.5669600963592529,
      "learning_rate": 8.394789473684211e-05,
      "loss": 0.0319,
      "step": 40500
    },
    {
      "epoch": 31.508356004663817,
      "grad_norm": 2.629176139831543,
      "learning_rate": 8.392157894736842e-05,
      "loss": 0.0415,
      "step": 40550
    },
    {
      "epoch": 31.547221142635056,
      "grad_norm": 0.5514523983001709,
      "learning_rate": 8.389526315789475e-05,
      "loss": 0.0363,
      "step": 40600
    },
    {
      "epoch": 31.586086280606295,
      "grad_norm": 2.319906711578369,
      "learning_rate": 8.386894736842106e-05,
      "loss": 0.0284,
      "step": 40650
    },
    {
      "epoch": 31.624951418577535,
      "grad_norm": 3.822218894958496,
      "learning_rate": 8.384263157894737e-05,
      "loss": 0.0332,
      "step": 40700
    },
    {
      "epoch": 31.663816556548774,
      "grad_norm": 0.9227602481842041,
      "learning_rate": 8.381631578947368e-05,
      "loss": 0.0542,
      "step": 40750
    },
    {
      "epoch": 31.702681694520017,
      "grad_norm": 1.3786684274673462,
      "learning_rate": 8.379e-05,
      "loss": 0.0468,
      "step": 40800
    },
    {
      "epoch": 31.741546832491256,
      "grad_norm": 1.7518463134765625,
      "learning_rate": 8.376368421052632e-05,
      "loss": 0.0376,
      "step": 40850
    },
    {
      "epoch": 31.780411970462495,
      "grad_norm": 3.1545917987823486,
      "learning_rate": 8.373736842105263e-05,
      "loss": 0.0344,
      "step": 40900
    },
    {
      "epoch": 31.819277108433734,
      "grad_norm": 0.835984468460083,
      "learning_rate": 8.371105263157896e-05,
      "loss": 0.033,
      "step": 40950
    },
    {
      "epoch": 31.858142246404974,
      "grad_norm": 1.3513562679290771,
      "learning_rate": 8.368473684210527e-05,
      "loss": 0.0357,
      "step": 41000
    },
    {
      "epoch": 31.858142246404974,
      "eval_accuracy": 0.9974250595151338,
      "eval_runtime": 876.8345,
      "eval_samples_per_second": 23.474,
      "eval_steps_per_second": 2.934,
      "step": 41000
    },
    {
      "epoch": 31.897007384376213,
      "grad_norm": 1.5770503282546997,
      "learning_rate": 8.365842105263158e-05,
      "loss": 0.0305,
      "step": 41050
    },
    {
      "epoch": 31.935872522347456,
      "grad_norm": 2.7758595943450928,
      "learning_rate": 8.36321052631579e-05,
      "loss": 0.0329,
      "step": 41100
    },
    {
      "epoch": 31.974737660318695,
      "grad_norm": 0.14476031064987183,
      "learning_rate": 8.360578947368422e-05,
      "loss": 0.0364,
      "step": 41150
    },
    {
      "epoch": 32.01321414691022,
      "grad_norm": 2.3019137382507324,
      "learning_rate": 8.357947368421053e-05,
      "loss": 0.0389,
      "step": 41200
    },
    {
      "epoch": 32.05207928488146,
      "grad_norm": 1.0233415365219116,
      "learning_rate": 8.355315789473684e-05,
      "loss": 0.0246,
      "step": 41250
    },
    {
      "epoch": 32.0909444228527,
      "grad_norm": 0.19109289348125458,
      "learning_rate": 8.352684210526315e-05,
      "loss": 0.0248,
      "step": 41300
    },
    {
      "epoch": 32.129809560823944,
      "grad_norm": 0.2724582850933075,
      "learning_rate": 8.350052631578948e-05,
      "loss": 0.0193,
      "step": 41350
    },
    {
      "epoch": 32.16867469879518,
      "grad_norm": 0.41294848918914795,
      "learning_rate": 8.347421052631579e-05,
      "loss": 0.0345,
      "step": 41400
    },
    {
      "epoch": 32.20753983676642,
      "grad_norm": 0.17901839315891266,
      "learning_rate": 8.344789473684211e-05,
      "loss": 0.0325,
      "step": 41450
    },
    {
      "epoch": 32.24640497473766,
      "grad_norm": 0.3015025854110718,
      "learning_rate": 8.342157894736843e-05,
      "loss": 0.0226,
      "step": 41500
    },
    {
      "epoch": 32.2852701127089,
      "grad_norm": 1.5335288047790527,
      "learning_rate": 8.339526315789474e-05,
      "loss": 0.0297,
      "step": 41550
    },
    {
      "epoch": 32.32413525068014,
      "grad_norm": 0.370561420917511,
      "learning_rate": 8.336894736842106e-05,
      "loss": 0.0206,
      "step": 41600
    },
    {
      "epoch": 32.36300038865138,
      "grad_norm": 2.148465156555176,
      "learning_rate": 8.334263157894737e-05,
      "loss": 0.0278,
      "step": 41650
    },
    {
      "epoch": 32.40186552662262,
      "grad_norm": 0.1226087436079979,
      "learning_rate": 8.331631578947369e-05,
      "loss": 0.042,
      "step": 41700
    },
    {
      "epoch": 32.44073066459386,
      "grad_norm": 0.08631648868322372,
      "learning_rate": 8.329e-05,
      "loss": 0.0245,
      "step": 41750
    },
    {
      "epoch": 32.4795958025651,
      "grad_norm": 0.32861921191215515,
      "learning_rate": 8.326368421052632e-05,
      "loss": 0.035,
      "step": 41800
    },
    {
      "epoch": 32.518460940536336,
      "grad_norm": 0.6523399353027344,
      "learning_rate": 8.323736842105263e-05,
      "loss": 0.0235,
      "step": 41850
    },
    {
      "epoch": 32.55732607850758,
      "grad_norm": 5.079765796661377,
      "learning_rate": 8.321105263157896e-05,
      "loss": 0.0356,
      "step": 41900
    },
    {
      "epoch": 32.59619121647882,
      "grad_norm": 0.8504376411437988,
      "learning_rate": 8.318473684210527e-05,
      "loss": 0.035,
      "step": 41950
    },
    {
      "epoch": 32.63505635445006,
      "grad_norm": 0.7626635432243347,
      "learning_rate": 8.315842105263158e-05,
      "loss": 0.0365,
      "step": 42000
    },
    {
      "epoch": 32.63505635445006,
      "eval_accuracy": 0.9966477189914007,
      "eval_runtime": 949.1706,
      "eval_samples_per_second": 21.685,
      "eval_steps_per_second": 2.711,
      "step": 42000
    },
    {
      "epoch": 32.6739214924213,
      "grad_norm": 0.28948360681533813,
      "learning_rate": 8.31321052631579e-05,
      "loss": 0.0349,
      "step": 42050
    },
    {
      "epoch": 32.712786630392536,
      "grad_norm": 1.969033122062683,
      "learning_rate": 8.310578947368422e-05,
      "loss": 0.0341,
      "step": 42100
    },
    {
      "epoch": 32.75165176836378,
      "grad_norm": 0.21110635995864868,
      "learning_rate": 8.307947368421053e-05,
      "loss": 0.0275,
      "step": 42150
    },
    {
      "epoch": 32.790516906335014,
      "grad_norm": 1.5202889442443848,
      "learning_rate": 8.305315789473684e-05,
      "loss": 0.0414,
      "step": 42200
    },
    {
      "epoch": 32.82938204430626,
      "grad_norm": 1.3945695161819458,
      "learning_rate": 8.302684210526315e-05,
      "loss": 0.036,
      "step": 42250
    },
    {
      "epoch": 32.8682471822775,
      "grad_norm": 1.1253477334976196,
      "learning_rate": 8.300052631578948e-05,
      "loss": 0.0302,
      "step": 42300
    },
    {
      "epoch": 32.907112320248736,
      "grad_norm": 0.15563784539699554,
      "learning_rate": 8.29742105263158e-05,
      "loss": 0.0343,
      "step": 42350
    },
    {
      "epoch": 32.94597745821998,
      "grad_norm": 2.101106643676758,
      "learning_rate": 8.294789473684212e-05,
      "loss": 0.0321,
      "step": 42400
    },
    {
      "epoch": 32.984842596191214,
      "grad_norm": 3.4534518718719482,
      "learning_rate": 8.292157894736843e-05,
      "loss": 0.0265,
      "step": 42450
    },
    {
      "epoch": 33.02331908278274,
      "grad_norm": 1.4187448024749756,
      "learning_rate": 8.289526315789474e-05,
      "loss": 0.0294,
      "step": 42500
    },
    {
      "epoch": 33.062184220753984,
      "grad_norm": 4.081812381744385,
      "learning_rate": 8.286894736842106e-05,
      "loss": 0.0302,
      "step": 42550
    },
    {
      "epoch": 33.10104935872522,
      "grad_norm": 3.367576837539673,
      "learning_rate": 8.284263157894738e-05,
      "loss": 0.0214,
      "step": 42600
    },
    {
      "epoch": 33.13991449669646,
      "grad_norm": 0.43796464800834656,
      "learning_rate": 8.281631578947369e-05,
      "loss": 0.021,
      "step": 42650
    },
    {
      "epoch": 33.178779634667706,
      "grad_norm": 0.7556443214416504,
      "learning_rate": 8.279e-05,
      "loss": 0.0307,
      "step": 42700
    },
    {
      "epoch": 33.21764477263894,
      "grad_norm": 0.5198609232902527,
      "learning_rate": 8.276368421052631e-05,
      "loss": 0.0255,
      "step": 42750
    },
    {
      "epoch": 33.256509910610184,
      "grad_norm": 0.27239367365837097,
      "learning_rate": 8.273736842105264e-05,
      "loss": 0.0215,
      "step": 42800
    },
    {
      "epoch": 33.29537504858142,
      "grad_norm": 2.588319778442383,
      "learning_rate": 8.271105263157896e-05,
      "loss": 0.0422,
      "step": 42850
    },
    {
      "epoch": 33.33424018655266,
      "grad_norm": 0.2941960394382477,
      "learning_rate": 8.268473684210527e-05,
      "loss": 0.0274,
      "step": 42900
    },
    {
      "epoch": 33.373105324523905,
      "grad_norm": 2.5152249336242676,
      "learning_rate": 8.265842105263158e-05,
      "loss": 0.0167,
      "step": 42950
    },
    {
      "epoch": 33.41197046249514,
      "grad_norm": 2.8935768604278564,
      "learning_rate": 8.26321052631579e-05,
      "loss": 0.0391,
      "step": 43000
    },
    {
      "epoch": 33.41197046249514,
      "eval_accuracy": 0.9973764757324005,
      "eval_runtime": 876.8781,
      "eval_samples_per_second": 23.473,
      "eval_steps_per_second": 2.934,
      "step": 43000
    },
    {
      "epoch": 33.450835600466384,
      "grad_norm": 0.657610297203064,
      "learning_rate": 8.260578947368422e-05,
      "loss": 0.0223,
      "step": 43050
    },
    {
      "epoch": 33.48970073843762,
      "grad_norm": 0.24365423619747162,
      "learning_rate": 8.257947368421053e-05,
      "loss": 0.0204,
      "step": 43100
    },
    {
      "epoch": 33.52856587640886,
      "grad_norm": 0.0590318888425827,
      "learning_rate": 8.255315789473684e-05,
      "loss": 0.0239,
      "step": 43150
    },
    {
      "epoch": 33.5674310143801,
      "grad_norm": 0.2320161759853363,
      "learning_rate": 8.252684210526315e-05,
      "loss": 0.0308,
      "step": 43200
    },
    {
      "epoch": 33.60629615235134,
      "grad_norm": 2.4144468307495117,
      "learning_rate": 8.250052631578947e-05,
      "loss": 0.0246,
      "step": 43250
    },
    {
      "epoch": 33.645161290322584,
      "grad_norm": 0.04267783835530281,
      "learning_rate": 8.247421052631579e-05,
      "loss": 0.0399,
      "step": 43300
    },
    {
      "epoch": 33.68402642829382,
      "grad_norm": 0.4845680594444275,
      "learning_rate": 8.244789473684212e-05,
      "loss": 0.0372,
      "step": 43350
    },
    {
      "epoch": 33.72289156626506,
      "grad_norm": 0.2614389657974243,
      "learning_rate": 8.242157894736843e-05,
      "loss": 0.0319,
      "step": 43400
    },
    {
      "epoch": 33.7617567042363,
      "grad_norm": 1.213369607925415,
      "learning_rate": 8.239526315789474e-05,
      "loss": 0.0387,
      "step": 43450
    },
    {
      "epoch": 33.80062184220754,
      "grad_norm": 0.19858339428901672,
      "learning_rate": 8.236894736842105e-05,
      "loss": 0.0276,
      "step": 43500
    },
    {
      "epoch": 33.839486980178776,
      "grad_norm": 0.058305270969867706,
      "learning_rate": 8.234263157894738e-05,
      "loss": 0.0421,
      "step": 43550
    },
    {
      "epoch": 33.87835211815002,
      "grad_norm": 0.3969908356666565,
      "learning_rate": 8.231631578947369e-05,
      "loss": 0.0432,
      "step": 43600
    },
    {
      "epoch": 33.91721725612126,
      "grad_norm": 3.350163459777832,
      "learning_rate": 8.229e-05,
      "loss": 0.0373,
      "step": 43650
    },
    {
      "epoch": 33.9560823940925,
      "grad_norm": 1.675500750541687,
      "learning_rate": 8.226368421052631e-05,
      "loss": 0.0283,
      "step": 43700
    },
    {
      "epoch": 33.99494753206374,
      "grad_norm": 1.9610657691955566,
      "learning_rate": 8.223736842105264e-05,
      "loss": 0.0383,
      "step": 43750
    },
    {
      "epoch": 34.03342401865527,
      "grad_norm": 0.2761792540550232,
      "learning_rate": 8.221105263157896e-05,
      "loss": 0.0158,
      "step": 43800
    },
    {
      "epoch": 34.0722891566265,
      "grad_norm": 4.586607456207275,
      "learning_rate": 8.218473684210527e-05,
      "loss": 0.0273,
      "step": 43850
    },
    {
      "epoch": 34.111154294597746,
      "grad_norm": 0.15766066312789917,
      "learning_rate": 8.215842105263158e-05,
      "loss": 0.0237,
      "step": 43900
    },
    {
      "epoch": 34.15001943256899,
      "grad_norm": 0.7128147482872009,
      "learning_rate": 8.21321052631579e-05,
      "loss": 0.0289,
      "step": 43950
    },
    {
      "epoch": 34.188884570540225,
      "grad_norm": 0.08594869077205658,
      "learning_rate": 8.210578947368421e-05,
      "loss": 0.024,
      "step": 44000
    },
    {
      "epoch": 34.188884570540225,
      "eval_accuracy": 0.9972307243842006,
      "eval_runtime": 876.3484,
      "eval_samples_per_second": 23.487,
      "eval_steps_per_second": 2.936,
      "step": 44000
    },
    {
      "epoch": 34.22774970851147,
      "grad_norm": 2.7693264484405518,
      "learning_rate": 8.207947368421053e-05,
      "loss": 0.0224,
      "step": 44050
    },
    {
      "epoch": 34.2666148464827,
      "grad_norm": 1.722844123840332,
      "learning_rate": 8.205315789473684e-05,
      "loss": 0.0398,
      "step": 44100
    },
    {
      "epoch": 34.305479984453946,
      "grad_norm": 4.099954605102539,
      "learning_rate": 8.202684210526316e-05,
      "loss": 0.0328,
      "step": 44150
    },
    {
      "epoch": 34.34434512242518,
      "grad_norm": 5.51669979095459,
      "learning_rate": 8.200052631578947e-05,
      "loss": 0.0368,
      "step": 44200
    },
    {
      "epoch": 34.383210260396424,
      "grad_norm": 0.23224566876888275,
      "learning_rate": 8.197421052631579e-05,
      "loss": 0.0232,
      "step": 44250
    },
    {
      "epoch": 34.42207539836767,
      "grad_norm": 0.1644495576620102,
      "learning_rate": 8.194789473684212e-05,
      "loss": 0.019,
      "step": 44300
    },
    {
      "epoch": 34.4609405363389,
      "grad_norm": 1.9373985528945923,
      "learning_rate": 8.192157894736843e-05,
      "loss": 0.0294,
      "step": 44350
    },
    {
      "epoch": 34.499805674310146,
      "grad_norm": 1.882383942604065,
      "learning_rate": 8.189526315789474e-05,
      "loss": 0.0259,
      "step": 44400
    },
    {
      "epoch": 34.53867081228138,
      "grad_norm": 3.6029393672943115,
      "learning_rate": 8.186894736842105e-05,
      "loss": 0.0256,
      "step": 44450
    },
    {
      "epoch": 34.577535950252624,
      "grad_norm": 0.5291832089424133,
      "learning_rate": 8.184263157894736e-05,
      "loss": 0.0195,
      "step": 44500
    },
    {
      "epoch": 34.61640108822386,
      "grad_norm": 0.31490087509155273,
      "learning_rate": 8.181631578947369e-05,
      "loss": 0.0193,
      "step": 44550
    },
    {
      "epoch": 34.6552662261951,
      "grad_norm": 3.0802714824676514,
      "learning_rate": 8.179e-05,
      "loss": 0.0245,
      "step": 44600
    },
    {
      "epoch": 34.694131364166346,
      "grad_norm": 0.4628906846046448,
      "learning_rate": 8.176368421052631e-05,
      "loss": 0.0295,
      "step": 44650
    },
    {
      "epoch": 34.73299650213758,
      "grad_norm": 0.29779794812202454,
      "learning_rate": 8.173736842105264e-05,
      "loss": 0.0219,
      "step": 44700
    },
    {
      "epoch": 34.771861640108824,
      "grad_norm": 2.227691650390625,
      "learning_rate": 8.171105263157895e-05,
      "loss": 0.0282,
      "step": 44750
    },
    {
      "epoch": 34.81072677808006,
      "grad_norm": 0.709452211856842,
      "learning_rate": 8.168473684210527e-05,
      "loss": 0.0331,
      "step": 44800
    },
    {
      "epoch": 34.8495919160513,
      "grad_norm": 0.4563519358634949,
      "learning_rate": 8.165842105263159e-05,
      "loss": 0.0263,
      "step": 44850
    },
    {
      "epoch": 34.88845705402254,
      "grad_norm": 0.22809520363807678,
      "learning_rate": 8.16321052631579e-05,
      "loss": 0.0285,
      "step": 44900
    },
    {
      "epoch": 34.92732219199378,
      "grad_norm": 3.2207274436950684,
      "learning_rate": 8.160578947368421e-05,
      "loss": 0.0288,
      "step": 44950
    },
    {
      "epoch": 34.966187329965024,
      "grad_norm": 3.6163206100463867,
      "learning_rate": 8.157947368421053e-05,
      "loss": 0.0318,
      "step": 45000
    },
    {
      "epoch": 34.966187329965024,
      "eval_accuracy": 0.9970849730360006,
      "eval_runtime": 876.4574,
      "eval_samples_per_second": 23.484,
      "eval_steps_per_second": 2.936,
      "step": 45000
    },
    {
      "epoch": 35.00466381655655,
      "grad_norm": 0.31362563371658325,
      "learning_rate": 8.155315789473685e-05,
      "loss": 0.0215,
      "step": 45050
    },
    {
      "epoch": 35.04352895452779,
      "grad_norm": 1.6593172550201416,
      "learning_rate": 8.152684210526316e-05,
      "loss": 0.0203,
      "step": 45100
    },
    {
      "epoch": 35.08239409249903,
      "grad_norm": 0.22460521757602692,
      "learning_rate": 8.150052631578948e-05,
      "loss": 0.0212,
      "step": 45150
    },
    {
      "epoch": 35.121259230470265,
      "grad_norm": 0.0782230868935585,
      "learning_rate": 8.14742105263158e-05,
      "loss": 0.025,
      "step": 45200
    },
    {
      "epoch": 35.16012436844151,
      "grad_norm": 0.7649909257888794,
      "learning_rate": 8.14478947368421e-05,
      "loss": 0.0192,
      "step": 45250
    },
    {
      "epoch": 35.19898950641275,
      "grad_norm": 0.5605923533439636,
      "learning_rate": 8.142157894736843e-05,
      "loss": 0.0276,
      "step": 45300
    },
    {
      "epoch": 35.23785464438399,
      "grad_norm": 0.656554102897644,
      "learning_rate": 8.139526315789474e-05,
      "loss": 0.0293,
      "step": 45350
    },
    {
      "epoch": 35.27671978235523,
      "grad_norm": 0.15811707079410553,
      "learning_rate": 8.136894736842105e-05,
      "loss": 0.0222,
      "step": 45400
    },
    {
      "epoch": 35.315584920326465,
      "grad_norm": 0.06599489599466324,
      "learning_rate": 8.134263157894737e-05,
      "loss": 0.0213,
      "step": 45450
    },
    {
      "epoch": 35.35445005829771,
      "grad_norm": 2.348271608352661,
      "learning_rate": 8.131631578947369e-05,
      "loss": 0.0211,
      "step": 45500
    },
    {
      "epoch": 35.393315196268944,
      "grad_norm": 0.1668296754360199,
      "learning_rate": 8.129e-05,
      "loss": 0.0202,
      "step": 45550
    },
    {
      "epoch": 35.432180334240186,
      "grad_norm": 0.6717625856399536,
      "learning_rate": 8.126368421052633e-05,
      "loss": 0.0147,
      "step": 45600
    },
    {
      "epoch": 35.47104547221143,
      "grad_norm": 0.5778363943099976,
      "learning_rate": 8.123736842105264e-05,
      "loss": 0.0246,
      "step": 45650
    },
    {
      "epoch": 35.509910610182665,
      "grad_norm": 1.1340782642364502,
      "learning_rate": 8.121105263157895e-05,
      "loss": 0.0201,
      "step": 45700
    },
    {
      "epoch": 35.54877574815391,
      "grad_norm": 0.16294288635253906,
      "learning_rate": 8.118473684210526e-05,
      "loss": 0.0276,
      "step": 45750
    },
    {
      "epoch": 35.58764088612514,
      "grad_norm": 0.07922358065843582,
      "learning_rate": 8.115842105263159e-05,
      "loss": 0.0224,
      "step": 45800
    },
    {
      "epoch": 35.626506024096386,
      "grad_norm": 0.037852223962545395,
      "learning_rate": 8.11321052631579e-05,
      "loss": 0.0209,
      "step": 45850
    },
    {
      "epoch": 35.66537116206762,
      "grad_norm": 0.5991801619529724,
      "learning_rate": 8.110578947368421e-05,
      "loss": 0.0223,
      "step": 45900
    },
    {
      "epoch": 35.704236300038865,
      "grad_norm": 1.8638439178466797,
      "learning_rate": 8.107947368421052e-05,
      "loss": 0.0294,
      "step": 45950
    },
    {
      "epoch": 35.74310143801011,
      "grad_norm": 1.1038992404937744,
      "learning_rate": 8.105315789473685e-05,
      "loss": 0.0162,
      "step": 46000
    },
    {
      "epoch": 35.74310143801011,
      "eval_accuracy": 0.9976679784288005,
      "eval_runtime": 878.5451,
      "eval_samples_per_second": 23.429,
      "eval_steps_per_second": 2.929,
      "step": 46000
    },
    {
      "epoch": 35.78196657598134,
      "grad_norm": 0.5733603239059448,
      "learning_rate": 8.102684210526316e-05,
      "loss": 0.0336,
      "step": 46050
    },
    {
      "epoch": 35.820831713952586,
      "grad_norm": 5.950699329376221,
      "learning_rate": 8.100052631578948e-05,
      "loss": 0.0273,
      "step": 46100
    },
    {
      "epoch": 35.85969685192382,
      "grad_norm": 0.018481293693184853,
      "learning_rate": 8.09742105263158e-05,
      "loss": 0.0169,
      "step": 46150
    },
    {
      "epoch": 35.898561989895065,
      "grad_norm": 0.4097042977809906,
      "learning_rate": 8.094789473684211e-05,
      "loss": 0.0285,
      "step": 46200
    },
    {
      "epoch": 35.93742712786631,
      "grad_norm": 0.5013408660888672,
      "learning_rate": 8.092157894736843e-05,
      "loss": 0.0241,
      "step": 46250
    },
    {
      "epoch": 35.97629226583754,
      "grad_norm": 0.3590555489063263,
      "learning_rate": 8.089526315789474e-05,
      "loss": 0.0261,
      "step": 46300
    },
    {
      "epoch": 36.01476875242907,
      "grad_norm": 3.033891201019287,
      "learning_rate": 8.086894736842106e-05,
      "loss": 0.0256,
      "step": 46350
    },
    {
      "epoch": 36.05363389040031,
      "grad_norm": 2.469217300415039,
      "learning_rate": 8.084263157894737e-05,
      "loss": 0.0325,
      "step": 46400
    },
    {
      "epoch": 36.09249902837155,
      "grad_norm": 1.083754301071167,
      "learning_rate": 8.081631578947368e-05,
      "loss": 0.0166,
      "step": 46450
    },
    {
      "epoch": 36.13136416634279,
      "grad_norm": 0.434164434671402,
      "learning_rate": 8.079e-05,
      "loss": 0.0146,
      "step": 46500
    },
    {
      "epoch": 36.17022930431403,
      "grad_norm": 1.0752344131469727,
      "learning_rate": 8.076368421052633e-05,
      "loss": 0.015,
      "step": 46550
    },
    {
      "epoch": 36.20909444228527,
      "grad_norm": 1.8302538394927979,
      "learning_rate": 8.073736842105264e-05,
      "loss": 0.0243,
      "step": 46600
    },
    {
      "epoch": 36.24795958025651,
      "grad_norm": 0.20908156037330627,
      "learning_rate": 8.071105263157895e-05,
      "loss": 0.0236,
      "step": 46650
    },
    {
      "epoch": 36.28682471822775,
      "grad_norm": 0.074724480509758,
      "learning_rate": 8.068473684210526e-05,
      "loss": 0.0148,
      "step": 46700
    },
    {
      "epoch": 36.32568985619899,
      "grad_norm": 2.220067024230957,
      "learning_rate": 8.065842105263159e-05,
      "loss": 0.0237,
      "step": 46750
    },
    {
      "epoch": 36.36455499417023,
      "grad_norm": 1.991211175918579,
      "learning_rate": 8.06321052631579e-05,
      "loss": 0.0263,
      "step": 46800
    },
    {
      "epoch": 36.40342013214147,
      "grad_norm": 0.17283689975738525,
      "learning_rate": 8.060578947368421e-05,
      "loss": 0.0227,
      "step": 46850
    },
    {
      "epoch": 36.442285270112706,
      "grad_norm": 4.825707912445068,
      "learning_rate": 8.057947368421052e-05,
      "loss": 0.0203,
      "step": 46900
    },
    {
      "epoch": 36.48115040808395,
      "grad_norm": 2.0094900131225586,
      "learning_rate": 8.055315789473684e-05,
      "loss": 0.0253,
      "step": 46950
    },
    {
      "epoch": 36.52001554605519,
      "grad_norm": 1.4706486463546753,
      "learning_rate": 8.052684210526316e-05,
      "loss": 0.0355,
      "step": 47000
    },
    {
      "epoch": 36.52001554605519,
      "eval_accuracy": 0.9974250595151338,
      "eval_runtime": 880.3189,
      "eval_samples_per_second": 23.381,
      "eval_steps_per_second": 2.923,
      "step": 47000
    },
    {
      "epoch": 36.55888068402643,
      "grad_norm": 1.3879079818725586,
      "learning_rate": 8.050052631578949e-05,
      "loss": 0.0203,
      "step": 47050
    },
    {
      "epoch": 36.59774582199767,
      "grad_norm": 0.15365901589393616,
      "learning_rate": 8.04742105263158e-05,
      "loss": 0.0224,
      "step": 47100
    },
    {
      "epoch": 36.636610959968905,
      "grad_norm": 3.338381052017212,
      "learning_rate": 8.044789473684211e-05,
      "loss": 0.0238,
      "step": 47150
    },
    {
      "epoch": 36.67547609794015,
      "grad_norm": 0.1772722750902176,
      "learning_rate": 8.042157894736842e-05,
      "loss": 0.0268,
      "step": 47200
    },
    {
      "epoch": 36.71434123591139,
      "grad_norm": 1.338255763053894,
      "learning_rate": 8.039526315789475e-05,
      "loss": 0.027,
      "step": 47250
    },
    {
      "epoch": 36.75320637388263,
      "grad_norm": 3.2425434589385986,
      "learning_rate": 8.036894736842106e-05,
      "loss": 0.0209,
      "step": 47300
    },
    {
      "epoch": 36.79207151185387,
      "grad_norm": 2.552506923675537,
      "learning_rate": 8.034263157894737e-05,
      "loss": 0.034,
      "step": 47350
    },
    {
      "epoch": 36.830936649825105,
      "grad_norm": 0.42373234033584595,
      "learning_rate": 8.031631578947368e-05,
      "loss": 0.0203,
      "step": 47400
    },
    {
      "epoch": 36.86980178779635,
      "grad_norm": 2.9511702060699463,
      "learning_rate": 8.028999999999999e-05,
      "loss": 0.0199,
      "step": 47450
    },
    {
      "epoch": 36.908666925767584,
      "grad_norm": 0.06529057770967484,
      "learning_rate": 8.026368421052633e-05,
      "loss": 0.0216,
      "step": 47500
    },
    {
      "epoch": 36.94753206373883,
      "grad_norm": 1.3030205965042114,
      "learning_rate": 8.023736842105264e-05,
      "loss": 0.0406,
      "step": 47550
    },
    {
      "epoch": 36.98639720171007,
      "grad_norm": 0.37334632873535156,
      "learning_rate": 8.021105263157895e-05,
      "loss": 0.0285,
      "step": 47600
    },
    {
      "epoch": 37.0248736883016,
      "grad_norm": 1.9018858671188354,
      "learning_rate": 8.018473684210527e-05,
      "loss": 0.0219,
      "step": 47650
    },
    {
      "epoch": 37.06373882627283,
      "grad_norm": 0.15374009311199188,
      "learning_rate": 8.015842105263158e-05,
      "loss": 0.0205,
      "step": 47700
    },
    {
      "epoch": 37.102603964244075,
      "grad_norm": 0.3583906590938568,
      "learning_rate": 8.01321052631579e-05,
      "loss": 0.018,
      "step": 47750
    },
    {
      "epoch": 37.14146910221531,
      "grad_norm": 2.795750856399536,
      "learning_rate": 8.010578947368421e-05,
      "loss": 0.0178,
      "step": 47800
    },
    {
      "epoch": 37.18033424018655,
      "grad_norm": 0.7013401389122009,
      "learning_rate": 8.007947368421053e-05,
      "loss": 0.0303,
      "step": 47850
    },
    {
      "epoch": 37.21919937815779,
      "grad_norm": 1.7055522203445435,
      "learning_rate": 8.005315789473684e-05,
      "loss": 0.0174,
      "step": 47900
    },
    {
      "epoch": 37.25806451612903,
      "grad_norm": 0.1926261931657791,
      "learning_rate": 8.002684210526316e-05,
      "loss": 0.0275,
      "step": 47950
    },
    {
      "epoch": 37.296929654100275,
      "grad_norm": 3.2588279247283936,
      "learning_rate": 8.000052631578949e-05,
      "loss": 0.0146,
      "step": 48000
    },
    {
      "epoch": 37.296929654100275,
      "eval_accuracy": 0.9987368216489336,
      "eval_runtime": 877.0856,
      "eval_samples_per_second": 23.467,
      "eval_steps_per_second": 2.934,
      "step": 48000
    },
    {
      "epoch": 37.33579479207151,
      "grad_norm": 1.3797889947891235,
      "learning_rate": 7.99742105263158e-05,
      "loss": 0.0228,
      "step": 48050
    },
    {
      "epoch": 37.37465993004275,
      "grad_norm": 0.4642263352870941,
      "learning_rate": 7.994789473684211e-05,
      "loss": 0.0276,
      "step": 48100
    },
    {
      "epoch": 37.41352506801399,
      "grad_norm": 0.7000479102134705,
      "learning_rate": 7.992157894736842e-05,
      "loss": 0.0095,
      "step": 48150
    },
    {
      "epoch": 37.45239020598523,
      "grad_norm": 0.5434962511062622,
      "learning_rate": 7.989526315789473e-05,
      "loss": 0.0167,
      "step": 48200
    },
    {
      "epoch": 37.49125534395647,
      "grad_norm": 0.2139407992362976,
      "learning_rate": 7.986894736842106e-05,
      "loss": 0.0156,
      "step": 48250
    },
    {
      "epoch": 37.53012048192771,
      "grad_norm": 0.040264032781124115,
      "learning_rate": 7.984263157894737e-05,
      "loss": 0.0212,
      "step": 48300
    },
    {
      "epoch": 37.56898561989895,
      "grad_norm": 0.15313409268856049,
      "learning_rate": 7.981631578947368e-05,
      "loss": 0.0182,
      "step": 48350
    },
    {
      "epoch": 37.60785075787019,
      "grad_norm": 0.43257951736450195,
      "learning_rate": 7.979000000000001e-05,
      "loss": 0.0226,
      "step": 48400
    },
    {
      "epoch": 37.64671589584143,
      "grad_norm": 2.317460775375366,
      "learning_rate": 7.976368421052632e-05,
      "loss": 0.0164,
      "step": 48450
    },
    {
      "epoch": 37.68558103381267,
      "grad_norm": 0.07254145294427872,
      "learning_rate": 7.973736842105264e-05,
      "loss": 0.0194,
      "step": 48500
    },
    {
      "epoch": 37.72444617178391,
      "grad_norm": 0.2547185719013214,
      "learning_rate": 7.971105263157896e-05,
      "loss": 0.0266,
      "step": 48550
    },
    {
      "epoch": 37.76331130975515,
      "grad_norm": 3.119913101196289,
      "learning_rate": 7.968473684210527e-05,
      "loss": 0.0145,
      "step": 48600
    },
    {
      "epoch": 37.80217644772639,
      "grad_norm": 1.1772373914718628,
      "learning_rate": 7.965842105263158e-05,
      "loss": 0.0265,
      "step": 48650
    },
    {
      "epoch": 37.84104158569763,
      "grad_norm": 0.11725064367055893,
      "learning_rate": 7.96321052631579e-05,
      "loss": 0.0116,
      "step": 48700
    },
    {
      "epoch": 37.87990672366887,
      "grad_norm": 0.8524043560028076,
      "learning_rate": 7.960578947368422e-05,
      "loss": 0.0178,
      "step": 48750
    },
    {
      "epoch": 37.91877186164011,
      "grad_norm": 0.40145233273506165,
      "learning_rate": 7.957947368421053e-05,
      "loss": 0.0319,
      "step": 48800
    },
    {
      "epoch": 37.957636999611346,
      "grad_norm": 0.2777029275894165,
      "learning_rate": 7.955315789473684e-05,
      "loss": 0.0221,
      "step": 48850
    },
    {
      "epoch": 37.99650213758259,
      "grad_norm": 0.24182066321372986,
      "learning_rate": 7.952684210526316e-05,
      "loss": 0.0168,
      "step": 48900
    },
    {
      "epoch": 38.034978624174116,
      "grad_norm": 2.333148241043091,
      "learning_rate": 7.950052631578947e-05,
      "loss": 0.024,
      "step": 48950
    },
    {
      "epoch": 38.07384376214536,
      "grad_norm": 3.2772161960601807,
      "learning_rate": 7.94742105263158e-05,
      "loss": 0.0187,
      "step": 49000
    },
    {
      "epoch": 38.07384376214536,
      "eval_accuracy": 0.9978137297770004,
      "eval_runtime": 886.8812,
      "eval_samples_per_second": 23.208,
      "eval_steps_per_second": 2.901,
      "step": 49000
    },
    {
      "epoch": 38.112708900116594,
      "grad_norm": 1.3939074277877808,
      "learning_rate": 7.944789473684211e-05,
      "loss": 0.0214,
      "step": 49050
    },
    {
      "epoch": 38.15157403808784,
      "grad_norm": 0.7525760531425476,
      "learning_rate": 7.942157894736842e-05,
      "loss": 0.0252,
      "step": 49100
    },
    {
      "epoch": 38.19043917605907,
      "grad_norm": 4.617874622344971,
      "learning_rate": 7.939526315789473e-05,
      "loss": 0.028,
      "step": 49150
    },
    {
      "epoch": 38.229304314030315,
      "grad_norm": 0.4796071946620941,
      "learning_rate": 7.936894736842106e-05,
      "loss": 0.0257,
      "step": 49200
    },
    {
      "epoch": 38.26816945200155,
      "grad_norm": 0.20932449400424957,
      "learning_rate": 7.934263157894737e-05,
      "loss": 0.0174,
      "step": 49250
    },
    {
      "epoch": 38.307034589972794,
      "grad_norm": 2.9770264625549316,
      "learning_rate": 7.931631578947368e-05,
      "loss": 0.0174,
      "step": 49300
    },
    {
      "epoch": 38.34589972794404,
      "grad_norm": 2.455333948135376,
      "learning_rate": 7.929000000000001e-05,
      "loss": 0.0335,
      "step": 49350
    },
    {
      "epoch": 38.38476486591527,
      "grad_norm": 2.985612630844116,
      "learning_rate": 7.926368421052632e-05,
      "loss": 0.0205,
      "step": 49400
    },
    {
      "epoch": 38.423630003886515,
      "grad_norm": 2.100100517272949,
      "learning_rate": 7.923736842105263e-05,
      "loss": 0.0249,
      "step": 49450
    },
    {
      "epoch": 38.46249514185775,
      "grad_norm": 3.7981529235839844,
      "learning_rate": 7.921105263157896e-05,
      "loss": 0.0154,
      "step": 49500
    },
    {
      "epoch": 38.501360279828994,
      "grad_norm": 1.1639838218688965,
      "learning_rate": 7.918473684210527e-05,
      "loss": 0.0188,
      "step": 49550
    },
    {
      "epoch": 38.54022541780024,
      "grad_norm": 0.5902754068374634,
      "learning_rate": 7.915842105263158e-05,
      "loss": 0.0253,
      "step": 49600
    },
    {
      "epoch": 38.57909055577147,
      "grad_norm": 3.1052238941192627,
      "learning_rate": 7.913210526315789e-05,
      "loss": 0.0143,
      "step": 49650
    },
    {
      "epoch": 38.617955693742715,
      "grad_norm": 0.11219343543052673,
      "learning_rate": 7.910578947368422e-05,
      "loss": 0.0167,
      "step": 49700
    },
    {
      "epoch": 38.65682083171395,
      "grad_norm": 0.9959619045257568,
      "learning_rate": 7.907947368421053e-05,
      "loss": 0.0223,
      "step": 49750
    },
    {
      "epoch": 38.695685969685194,
      "grad_norm": 0.7064417600631714,
      "learning_rate": 7.905315789473685e-05,
      "loss": 0.0204,
      "step": 49800
    },
    {
      "epoch": 38.73455110765643,
      "grad_norm": 0.32734501361846924,
      "learning_rate": 7.902684210526316e-05,
      "loss": 0.0235,
      "step": 49850
    },
    {
      "epoch": 38.77341624562767,
      "grad_norm": 1.8845106363296509,
      "learning_rate": 7.900052631578948e-05,
      "loss": 0.0231,
      "step": 49900
    },
    {
      "epoch": 38.812281383598915,
      "grad_norm": 1.0510152578353882,
      "learning_rate": 7.89742105263158e-05,
      "loss": 0.0196,
      "step": 49950
    },
    {
      "epoch": 38.85114652157015,
      "grad_norm": 2.2157108783721924,
      "learning_rate": 7.894789473684211e-05,
      "loss": 0.021,
      "step": 50000
    },
    {
      "epoch": 38.85114652157015,
      "eval_accuracy": 0.9982024000388671,
      "eval_runtime": 880.8313,
      "eval_samples_per_second": 23.368,
      "eval_steps_per_second": 2.921,
      "step": 50000
    },
    {
      "epoch": 38.89001165954139,
      "grad_norm": 2.7347636222839355,
      "learning_rate": 7.892157894736842e-05,
      "loss": 0.0219,
      "step": 50050
    },
    {
      "epoch": 38.92887679751263,
      "grad_norm": 0.05592096224427223,
      "learning_rate": 7.889526315789474e-05,
      "loss": 0.0234,
      "step": 50100
    },
    {
      "epoch": 38.96774193548387,
      "grad_norm": 0.25417473912239075,
      "learning_rate": 7.886894736842105e-05,
      "loss": 0.0273,
      "step": 50150
    },
    {
      "epoch": 39.0062184220754,
      "grad_norm": 0.08906584233045578,
      "learning_rate": 7.884263157894737e-05,
      "loss": 0.0234,
      "step": 50200
    },
    {
      "epoch": 39.045083560046635,
      "grad_norm": 0.2883197069168091,
      "learning_rate": 7.881631578947368e-05,
      "loss": 0.0121,
      "step": 50250
    },
    {
      "epoch": 39.08394869801788,
      "grad_norm": 0.18966904282569885,
      "learning_rate": 7.879000000000001e-05,
      "loss": 0.0161,
      "step": 50300
    },
    {
      "epoch": 39.12281383598912,
      "grad_norm": 0.5769274830818176,
      "learning_rate": 7.876368421052632e-05,
      "loss": 0.0152,
      "step": 50350
    },
    {
      "epoch": 39.161678973960356,
      "grad_norm": 2.145620107650757,
      "learning_rate": 7.873736842105263e-05,
      "loss": 0.0169,
      "step": 50400
    },
    {
      "epoch": 39.2005441119316,
      "grad_norm": 0.7396944761276245,
      "learning_rate": 7.871105263157896e-05,
      "loss": 0.0099,
      "step": 50450
    },
    {
      "epoch": 39.239409249902835,
      "grad_norm": 0.2696094810962677,
      "learning_rate": 7.868473684210527e-05,
      "loss": 0.0126,
      "step": 50500
    },
    {
      "epoch": 39.27827438787408,
      "grad_norm": 0.7367541790008545,
      "learning_rate": 7.865842105263158e-05,
      "loss": 0.0152,
      "step": 50550
    },
    {
      "epoch": 39.31713952584532,
      "grad_norm": 0.13489407300949097,
      "learning_rate": 7.863210526315789e-05,
      "loss": 0.0144,
      "step": 50600
    },
    {
      "epoch": 39.356004663816556,
      "grad_norm": 3.5517187118530273,
      "learning_rate": 7.86057894736842e-05,
      "loss": 0.0159,
      "step": 50650
    },
    {
      "epoch": 39.3948698017878,
      "grad_norm": 0.4950302839279175,
      "learning_rate": 7.857947368421053e-05,
      "loss": 0.0157,
      "step": 50700
    },
    {
      "epoch": 39.433734939759034,
      "grad_norm": 0.3861680328845978,
      "learning_rate": 7.855315789473685e-05,
      "loss": 0.0148,
      "step": 50750
    },
    {
      "epoch": 39.47260007773028,
      "grad_norm": 0.625752866268158,
      "learning_rate": 7.852684210526317e-05,
      "loss": 0.0209,
      "step": 50800
    },
    {
      "epoch": 39.51146521570151,
      "grad_norm": 1.8318170309066772,
      "learning_rate": 7.850052631578948e-05,
      "loss": 0.017,
      "step": 50850
    },
    {
      "epoch": 39.550330353672756,
      "grad_norm": 6.518121719360352,
      "learning_rate": 7.847421052631579e-05,
      "loss": 0.0191,
      "step": 50900
    },
    {
      "epoch": 39.589195491644,
      "grad_norm": 0.056834232062101364,
      "learning_rate": 7.844789473684211e-05,
      "loss": 0.0212,
      "step": 50950
    },
    {
      "epoch": 39.628060629615234,
      "grad_norm": 3.313100576400757,
      "learning_rate": 7.842157894736843e-05,
      "loss": 0.0217,
      "step": 51000
    },
    {
      "epoch": 39.628060629615234,
      "eval_accuracy": 0.9980080649079337,
      "eval_runtime": 878.9892,
      "eval_samples_per_second": 23.417,
      "eval_steps_per_second": 2.927,
      "step": 51000
    },
    {
      "epoch": 39.66692576758648,
      "grad_norm": 0.2621047794818878,
      "learning_rate": 7.839526315789474e-05,
      "loss": 0.0124,
      "step": 51050
    },
    {
      "epoch": 39.70579090555771,
      "grad_norm": 4.195189476013184,
      "learning_rate": 7.836894736842105e-05,
      "loss": 0.0231,
      "step": 51100
    },
    {
      "epoch": 39.744656043528956,
      "grad_norm": 0.27412697672843933,
      "learning_rate": 7.834263157894736e-05,
      "loss": 0.0265,
      "step": 51150
    },
    {
      "epoch": 39.78352118150019,
      "grad_norm": 0.005160004366189241,
      "learning_rate": 7.83163157894737e-05,
      "loss": 0.0183,
      "step": 51200
    },
    {
      "epoch": 39.822386319471434,
      "grad_norm": 3.9593896865844727,
      "learning_rate": 7.829000000000001e-05,
      "loss": 0.0235,
      "step": 51250
    },
    {
      "epoch": 39.86125145744268,
      "grad_norm": 2.3815083503723145,
      "learning_rate": 7.826368421052632e-05,
      "loss": 0.0204,
      "step": 51300
    },
    {
      "epoch": 39.90011659541391,
      "grad_norm": 0.07994846999645233,
      "learning_rate": 7.823736842105263e-05,
      "loss": 0.0221,
      "step": 51350
    },
    {
      "epoch": 39.938981733385155,
      "grad_norm": 2.044116258621216,
      "learning_rate": 7.821105263157895e-05,
      "loss": 0.0215,
      "step": 51400
    },
    {
      "epoch": 39.97784687135639,
      "grad_norm": 0.45163819193840027,
      "learning_rate": 7.818473684210527e-05,
      "loss": 0.0197,
      "step": 51450
    },
    {
      "epoch": 40.01632335794792,
      "grad_norm": 0.2218233197927475,
      "learning_rate": 7.815842105263158e-05,
      "loss": 0.0113,
      "step": 51500
    },
    {
      "epoch": 40.05518849591916,
      "grad_norm": 0.23619444668293,
      "learning_rate": 7.81321052631579e-05,
      "loss": 0.0167,
      "step": 51550
    },
    {
      "epoch": 40.0940536338904,
      "grad_norm": 0.0961291566491127,
      "learning_rate": 7.81057894736842e-05,
      "loss": 0.0155,
      "step": 51600
    },
    {
      "epoch": 40.13291877186164,
      "grad_norm": 16.20973014831543,
      "learning_rate": 7.807947368421053e-05,
      "loss": 0.0176,
      "step": 51650
    },
    {
      "epoch": 40.17178390983288,
      "grad_norm": 0.22694605588912964,
      "learning_rate": 7.805315789473686e-05,
      "loss": 0.0153,
      "step": 51700
    },
    {
      "epoch": 40.21064904780412,
      "grad_norm": 0.20650620758533478,
      "learning_rate": 7.802684210526317e-05,
      "loss": 0.0157,
      "step": 51750
    },
    {
      "epoch": 40.24951418577536,
      "grad_norm": 2.3340983390808105,
      "learning_rate": 7.800052631578948e-05,
      "loss": 0.0097,
      "step": 51800
    },
    {
      "epoch": 40.2883793237466,
      "grad_norm": 0.4349310100078583,
      "learning_rate": 7.797421052631579e-05,
      "loss": 0.0186,
      "step": 51850
    },
    {
      "epoch": 40.32724446171784,
      "grad_norm": 1.1148430109024048,
      "learning_rate": 7.79478947368421e-05,
      "loss": 0.0182,
      "step": 51900
    },
    {
      "epoch": 40.36610959968908,
      "grad_norm": 0.6705226898193359,
      "learning_rate": 7.792157894736843e-05,
      "loss": 0.0237,
      "step": 51950
    },
    {
      "epoch": 40.40497473766032,
      "grad_norm": 0.337016224861145,
      "learning_rate": 7.789526315789474e-05,
      "loss": 0.0193,
      "step": 52000
    },
    {
      "epoch": 40.40497473766032,
      "eval_accuracy": 0.9984453189525336,
      "eval_runtime": 879.6611,
      "eval_samples_per_second": 23.399,
      "eval_steps_per_second": 2.925,
      "step": 52000
    },
    {
      "epoch": 40.44383987563156,
      "grad_norm": 1.561038851737976,
      "learning_rate": 7.786894736842105e-05,
      "loss": 0.0148,
      "step": 52050
    },
    {
      "epoch": 40.482705013602796,
      "grad_norm": 1.7603131532669067,
      "learning_rate": 7.784263157894736e-05,
      "loss": 0.0192,
      "step": 52100
    },
    {
      "epoch": 40.52157015157404,
      "grad_norm": 2.7705743312835693,
      "learning_rate": 7.781631578947369e-05,
      "loss": 0.0303,
      "step": 52150
    },
    {
      "epoch": 40.560435289545275,
      "grad_norm": 3.2688467502593994,
      "learning_rate": 7.779000000000001e-05,
      "loss": 0.0236,
      "step": 52200
    },
    {
      "epoch": 40.59930042751652,
      "grad_norm": 0.5117989778518677,
      "learning_rate": 7.776368421052632e-05,
      "loss": 0.0195,
      "step": 52250
    },
    {
      "epoch": 40.63816556548776,
      "grad_norm": 6.964997291564941,
      "learning_rate": 7.773736842105264e-05,
      "loss": 0.0195,
      "step": 52300
    },
    {
      "epoch": 40.677030703458996,
      "grad_norm": 2.0262928009033203,
      "learning_rate": 7.771105263157895e-05,
      "loss": 0.0143,
      "step": 52350
    },
    {
      "epoch": 40.71589584143024,
      "grad_norm": 0.025713708251714706,
      "learning_rate": 7.768473684210527e-05,
      "loss": 0.0191,
      "step": 52400
    },
    {
      "epoch": 40.754760979401475,
      "grad_norm": 0.850270688533783,
      "learning_rate": 7.765842105263158e-05,
      "loss": 0.0219,
      "step": 52450
    },
    {
      "epoch": 40.79362611737272,
      "grad_norm": 2.509840250015259,
      "learning_rate": 7.76321052631579e-05,
      "loss": 0.0203,
      "step": 52500
    },
    {
      "epoch": 40.83249125534395,
      "grad_norm": 0.4855981469154358,
      "learning_rate": 7.760578947368421e-05,
      "loss": 0.023,
      "step": 52550
    },
    {
      "epoch": 40.871356393315196,
      "grad_norm": 0.11746831238269806,
      "learning_rate": 7.757947368421053e-05,
      "loss": 0.0167,
      "step": 52600
    },
    {
      "epoch": 40.91022153128644,
      "grad_norm": 0.8593801856040955,
      "learning_rate": 7.755315789473684e-05,
      "loss": 0.0214,
      "step": 52650
    },
    {
      "epoch": 40.949086669257674,
      "grad_norm": 2.0908868312835693,
      "learning_rate": 7.752684210526317e-05,
      "loss": 0.0135,
      "step": 52700
    },
    {
      "epoch": 40.98795180722892,
      "grad_norm": 0.017740854993462563,
      "learning_rate": 7.750052631578948e-05,
      "loss": 0.018,
      "step": 52750
    },
    {
      "epoch": 41.026428293820445,
      "grad_norm": 0.7258341312408447,
      "learning_rate": 7.747421052631579e-05,
      "loss": 0.0123,
      "step": 52800
    },
    {
      "epoch": 41.06529343179168,
      "grad_norm": 1.297969102859497,
      "learning_rate": 7.74478947368421e-05,
      "loss": 0.0211,
      "step": 52850
    },
    {
      "epoch": 41.10415856976292,
      "grad_norm": 0.03711100295186043,
      "learning_rate": 7.742157894736843e-05,
      "loss": 0.0189,
      "step": 52900
    },
    {
      "epoch": 41.143023707734166,
      "grad_norm": 0.9266149401664734,
      "learning_rate": 7.739526315789474e-05,
      "loss": 0.0133,
      "step": 52950
    },
    {
      "epoch": 41.1818888457054,
      "grad_norm": 0.04078902676701546,
      "learning_rate": 7.736894736842105e-05,
      "loss": 0.0139,
      "step": 53000
    },
    {
      "epoch": 41.1818888457054,
      "eval_accuracy": 0.998493902735267,
      "eval_runtime": 881.642,
      "eval_samples_per_second": 23.346,
      "eval_steps_per_second": 2.918,
      "step": 53000
    },
    {
      "epoch": 41.220753983676644,
      "grad_norm": 2.0239853858947754,
      "learning_rate": 7.734263157894738e-05,
      "loss": 0.0211,
      "step": 53050
    },
    {
      "epoch": 41.25961912164788,
      "grad_norm": 0.8523136377334595,
      "learning_rate": 7.731631578947369e-05,
      "loss": 0.0183,
      "step": 53100
    },
    {
      "epoch": 41.29848425961912,
      "grad_norm": 0.05443975329399109,
      "learning_rate": 7.729e-05,
      "loss": 0.0141,
      "step": 53150
    },
    {
      "epoch": 41.33734939759036,
      "grad_norm": 3.1442885398864746,
      "learning_rate": 7.726368421052633e-05,
      "loss": 0.0169,
      "step": 53200
    },
    {
      "epoch": 41.3762145355616,
      "grad_norm": 1.146552562713623,
      "learning_rate": 7.723736842105264e-05,
      "loss": 0.0158,
      "step": 53250
    },
    {
      "epoch": 41.415079673532844,
      "grad_norm": 1.5513288974761963,
      "learning_rate": 7.721105263157895e-05,
      "loss": 0.0119,
      "step": 53300
    },
    {
      "epoch": 41.45394481150408,
      "grad_norm": 1.2826228141784668,
      "learning_rate": 7.718473684210526e-05,
      "loss": 0.0115,
      "step": 53350
    },
    {
      "epoch": 41.49280994947532,
      "grad_norm": 1.0318225622177124,
      "learning_rate": 7.715842105263159e-05,
      "loss": 0.0143,
      "step": 53400
    },
    {
      "epoch": 41.53167508744656,
      "grad_norm": 0.11040183156728745,
      "learning_rate": 7.71321052631579e-05,
      "loss": 0.0289,
      "step": 53450
    },
    {
      "epoch": 41.5705402254178,
      "grad_norm": 0.10635088384151459,
      "learning_rate": 7.710578947368421e-05,
      "loss": 0.0159,
      "step": 53500
    },
    {
      "epoch": 41.60940536338904,
      "grad_norm": 0.13474248349666595,
      "learning_rate": 7.707947368421053e-05,
      "loss": 0.0156,
      "step": 53550
    },
    {
      "epoch": 41.64827050136028,
      "grad_norm": 2.0040714740753174,
      "learning_rate": 7.705315789473685e-05,
      "loss": 0.0078,
      "step": 53600
    },
    {
      "epoch": 41.68713563933152,
      "grad_norm": 1.4266414642333984,
      "learning_rate": 7.702684210526317e-05,
      "loss": 0.0136,
      "step": 53650
    },
    {
      "epoch": 41.72600077730276,
      "grad_norm": 0.7016364336013794,
      "learning_rate": 7.700052631578948e-05,
      "loss": 0.018,
      "step": 53700
    },
    {
      "epoch": 41.764865915274,
      "grad_norm": 0.6079683899879456,
      "learning_rate": 7.69742105263158e-05,
      "loss": 0.0219,
      "step": 53750
    },
    {
      "epoch": 41.80373105324524,
      "grad_norm": 0.13080672919750214,
      "learning_rate": 7.69478947368421e-05,
      "loss": 0.0116,
      "step": 53800
    },
    {
      "epoch": 41.84259619121648,
      "grad_norm": 1.093205451965332,
      "learning_rate": 7.692157894736842e-05,
      "loss": 0.018,
      "step": 53850
    },
    {
      "epoch": 41.881461329187715,
      "grad_norm": 0.07690184563398361,
      "learning_rate": 7.689526315789474e-05,
      "loss": 0.0166,
      "step": 53900
    },
    {
      "epoch": 41.92032646715896,
      "grad_norm": 0.6435837745666504,
      "learning_rate": 7.686894736842105e-05,
      "loss": 0.0159,
      "step": 53950
    },
    {
      "epoch": 41.9591916051302,
      "grad_norm": 0.10160457342863083,
      "learning_rate": 7.684263157894738e-05,
      "loss": 0.0205,
      "step": 54000
    },
    {
      "epoch": 41.9591916051302,
      "eval_accuracy": 0.9978623135597338,
      "eval_runtime": 1074.1404,
      "eval_samples_per_second": 19.162,
      "eval_steps_per_second": 2.395,
      "step": 54000
    },
    {
      "epoch": 41.998056743101436,
      "grad_norm": 1.1743571758270264,
      "learning_rate": 7.681631578947369e-05,
      "loss": 0.0175,
      "step": 54050
    },
    {
      "epoch": 42.036533229692964,
      "grad_norm": 2.6207754611968994,
      "learning_rate": 7.679e-05,
      "loss": 0.0094,
      "step": 54100
    },
    {
      "epoch": 42.07539836766421,
      "grad_norm": 0.26519230008125305,
      "learning_rate": 7.676368421052633e-05,
      "loss": 0.0169,
      "step": 54150
    },
    {
      "epoch": 42.11426350563544,
      "grad_norm": 0.45864981412887573,
      "learning_rate": 7.673736842105264e-05,
      "loss": 0.0104,
      "step": 54200
    },
    {
      "epoch": 42.153128643606685,
      "grad_norm": 0.0494319349527359,
      "learning_rate": 7.671105263157895e-05,
      "loss": 0.006,
      "step": 54250
    },
    {
      "epoch": 42.19199378157793,
      "grad_norm": 0.43440911173820496,
      "learning_rate": 7.668473684210526e-05,
      "loss": 0.0183,
      "step": 54300
    },
    {
      "epoch": 42.23085891954916,
      "grad_norm": 0.030721886083483696,
      "learning_rate": 7.665842105263157e-05,
      "loss": 0.0129,
      "step": 54350
    },
    {
      "epoch": 42.269724057520406,
      "grad_norm": 0.8205466270446777,
      "learning_rate": 7.66321052631579e-05,
      "loss": 0.0185,
      "step": 54400
    },
    {
      "epoch": 42.30858919549164,
      "grad_norm": 0.013601802289485931,
      "learning_rate": 7.660578947368422e-05,
      "loss": 0.0137,
      "step": 54450
    },
    {
      "epoch": 42.347454333462885,
      "grad_norm": 3.7669804096221924,
      "learning_rate": 7.657947368421054e-05,
      "loss": 0.0153,
      "step": 54500
    },
    {
      "epoch": 42.38631947143412,
      "grad_norm": 0.4318236708641052,
      "learning_rate": 7.655315789473685e-05,
      "loss": 0.0182,
      "step": 54550
    },
    {
      "epoch": 42.42518460940536,
      "grad_norm": 0.08533640205860138,
      "learning_rate": 7.652684210526316e-05,
      "loss": 0.0211,
      "step": 54600
    },
    {
      "epoch": 42.464049747376606,
      "grad_norm": 0.10274555534124374,
      "learning_rate": 7.650052631578948e-05,
      "loss": 0.0142,
      "step": 54650
    },
    {
      "epoch": 42.50291488534784,
      "grad_norm": 6.137352466583252,
      "learning_rate": 7.64742105263158e-05,
      "loss": 0.0163,
      "step": 54700
    },
    {
      "epoch": 42.541780023319085,
      "grad_norm": 0.6145594716072083,
      "learning_rate": 7.64478947368421e-05,
      "loss": 0.0087,
      "step": 54750
    },
    {
      "epoch": 42.58064516129032,
      "grad_norm": 1.7177934646606445,
      "learning_rate": 7.642157894736842e-05,
      "loss": 0.018,
      "step": 54800
    },
    {
      "epoch": 42.61951029926156,
      "grad_norm": 0.0985284298658371,
      "learning_rate": 7.639526315789473e-05,
      "loss": 0.0068,
      "step": 54850
    },
    {
      "epoch": 42.6583754372328,
      "grad_norm": 0.36002910137176514,
      "learning_rate": 7.636894736842105e-05,
      "loss": 0.0286,
      "step": 54900
    },
    {
      "epoch": 42.69724057520404,
      "grad_norm": 0.5385025143623352,
      "learning_rate": 7.634263157894738e-05,
      "loss": 0.0175,
      "step": 54950
    },
    {
      "epoch": 42.736105713175284,
      "grad_norm": 0.27823692560195923,
      "learning_rate": 7.631631578947369e-05,
      "loss": 0.0107,
      "step": 55000
    },
    {
      "epoch": 42.736105713175284,
      "eval_accuracy": 0.9982509838216004,
      "eval_runtime": 881.3376,
      "eval_samples_per_second": 23.354,
      "eval_steps_per_second": 2.919,
      "step": 55000
    },
    {
      "epoch": 42.77497085114652,
      "grad_norm": 0.012415263801813126,
      "learning_rate": 7.629e-05,
      "loss": 0.0152,
      "step": 55050
    },
    {
      "epoch": 42.81383598911776,
      "grad_norm": 0.06934750825166702,
      "learning_rate": 7.626368421052631e-05,
      "loss": 0.0106,
      "step": 55100
    },
    {
      "epoch": 42.852701127089,
      "grad_norm": 2.376636505126953,
      "learning_rate": 7.623736842105264e-05,
      "loss": 0.0137,
      "step": 55150
    },
    {
      "epoch": 42.89156626506024,
      "grad_norm": 0.2681966722011566,
      "learning_rate": 7.621105263157895e-05,
      "loss": 0.0177,
      "step": 55200
    },
    {
      "epoch": 42.930431403031484,
      "grad_norm": 0.0518764927983284,
      "learning_rate": 7.618473684210526e-05,
      "loss": 0.0249,
      "step": 55250
    },
    {
      "epoch": 42.96929654100272,
      "grad_norm": 0.05432099476456642,
      "learning_rate": 7.615842105263157e-05,
      "loss": 0.021,
      "step": 55300
    },
    {
      "epoch": 43.00777302759425,
      "grad_norm": 0.03369888663291931,
      "learning_rate": 7.61321052631579e-05,
      "loss": 0.0252,
      "step": 55350
    },
    {
      "epoch": 43.04663816556549,
      "grad_norm": 0.7964507341384888,
      "learning_rate": 7.610578947368422e-05,
      "loss": 0.018,
      "step": 55400
    },
    {
      "epoch": 43.085503303536726,
      "grad_norm": 1.1792151927947998,
      "learning_rate": 7.607947368421054e-05,
      "loss": 0.0143,
      "step": 55450
    },
    {
      "epoch": 43.12436844150797,
      "grad_norm": 0.8480147123336792,
      "learning_rate": 7.605315789473685e-05,
      "loss": 0.013,
      "step": 55500
    },
    {
      "epoch": 43.163233579479204,
      "grad_norm": 1.0581895112991333,
      "learning_rate": 7.602684210526316e-05,
      "loss": 0.0221,
      "step": 55550
    },
    {
      "epoch": 43.20209871745045,
      "grad_norm": 0.1439225971698761,
      "learning_rate": 7.600052631578947e-05,
      "loss": 0.017,
      "step": 55600
    },
    {
      "epoch": 43.24096385542169,
      "grad_norm": 0.47069478034973145,
      "learning_rate": 7.59742105263158e-05,
      "loss": 0.0186,
      "step": 55650
    },
    {
      "epoch": 43.279828993392925,
      "grad_norm": 0.09937036782503128,
      "learning_rate": 7.594789473684211e-05,
      "loss": 0.0111,
      "step": 55700
    },
    {
      "epoch": 43.31869413136417,
      "grad_norm": 0.12931126356124878,
      "learning_rate": 7.592157894736842e-05,
      "loss": 0.0109,
      "step": 55750
    },
    {
      "epoch": 43.357559269335404,
      "grad_norm": 2.7805063724517822,
      "learning_rate": 7.589526315789473e-05,
      "loss": 0.0173,
      "step": 55800
    },
    {
      "epoch": 43.39642440730665,
      "grad_norm": 0.9066233038902283,
      "learning_rate": 7.586894736842106e-05,
      "loss": 0.0243,
      "step": 55850
    },
    {
      "epoch": 43.43528954527788,
      "grad_norm": 0.13462518155574799,
      "learning_rate": 7.584263157894738e-05,
      "loss": 0.0219,
      "step": 55900
    },
    {
      "epoch": 43.474154683249125,
      "grad_norm": 0.2093678116798401,
      "learning_rate": 7.581631578947369e-05,
      "loss": 0.0077,
      "step": 55950
    },
    {
      "epoch": 43.51301982122037,
      "grad_norm": 0.5262479782104492,
      "learning_rate": 7.579e-05,
      "loss": 0.0157,
      "step": 56000
    },
    {
      "epoch": 43.51301982122037,
      "eval_accuracy": 0.9981052324734003,
      "eval_runtime": 879.9359,
      "eval_samples_per_second": 23.391,
      "eval_steps_per_second": 2.924,
      "step": 56000
    },
    {
      "epoch": 43.551884959191604,
      "grad_norm": 0.41396597027778625,
      "learning_rate": 7.576368421052632e-05,
      "loss": 0.0172,
      "step": 56050
    },
    {
      "epoch": 43.59075009716285,
      "grad_norm": 0.23529602587223053,
      "learning_rate": 7.573736842105264e-05,
      "loss": 0.0208,
      "step": 56100
    },
    {
      "epoch": 43.62961523513408,
      "grad_norm": 0.17375200986862183,
      "learning_rate": 7.571105263157895e-05,
      "loss": 0.0168,
      "step": 56150
    },
    {
      "epoch": 43.668480373105325,
      "grad_norm": 0.03255872428417206,
      "learning_rate": 7.568473684210526e-05,
      "loss": 0.0168,
      "step": 56200
    },
    {
      "epoch": 43.70734551107657,
      "grad_norm": 1.1647653579711914,
      "learning_rate": 7.565842105263158e-05,
      "loss": 0.0143,
      "step": 56250
    },
    {
      "epoch": 43.7462106490478,
      "grad_norm": 0.39687833189964294,
      "learning_rate": 7.563210526315789e-05,
      "loss": 0.0087,
      "step": 56300
    },
    {
      "epoch": 43.785075787019046,
      "grad_norm": 1.4236061573028564,
      "learning_rate": 7.560578947368421e-05,
      "loss": 0.0127,
      "step": 56350
    },
    {
      "epoch": 43.82394092499028,
      "grad_norm": 1.305365800857544,
      "learning_rate": 7.557947368421054e-05,
      "loss": 0.0211,
      "step": 56400
    },
    {
      "epoch": 43.862806062961525,
      "grad_norm": 1.352987289428711,
      "learning_rate": 7.555315789473685e-05,
      "loss": 0.0149,
      "step": 56450
    },
    {
      "epoch": 43.90167120093276,
      "grad_norm": 0.7978178858757019,
      "learning_rate": 7.552684210526316e-05,
      "loss": 0.0251,
      "step": 56500
    },
    {
      "epoch": 43.940536338904,
      "grad_norm": 0.01514784898608923,
      "learning_rate": 7.550052631578947e-05,
      "loss": 0.0173,
      "step": 56550
    },
    {
      "epoch": 43.979401476875246,
      "grad_norm": 2.1784021854400635,
      "learning_rate": 7.54742105263158e-05,
      "loss": 0.0139,
      "step": 56600
    },
    {
      "epoch": 44.01787796346677,
      "grad_norm": 0.22825995087623596,
      "learning_rate": 7.544789473684211e-05,
      "loss": 0.0168,
      "step": 56650
    },
    {
      "epoch": 44.05674310143801,
      "grad_norm": 0.3057117760181427,
      "learning_rate": 7.542157894736842e-05,
      "loss": 0.014,
      "step": 56700
    },
    {
      "epoch": 44.09560823940925,
      "grad_norm": 1.0708446502685547,
      "learning_rate": 7.539526315789473e-05,
      "loss": 0.0114,
      "step": 56750
    },
    {
      "epoch": 44.13447337738049,
      "grad_norm": 0.020148010924458504,
      "learning_rate": 7.536894736842106e-05,
      "loss": 0.0203,
      "step": 56800
    },
    {
      "epoch": 44.17333851535173,
      "grad_norm": 1.150698184967041,
      "learning_rate": 7.534263157894737e-05,
      "loss": 0.0151,
      "step": 56850
    },
    {
      "epoch": 44.212203653322966,
      "grad_norm": 0.1435837298631668,
      "learning_rate": 7.53163157894737e-05,
      "loss": 0.0121,
      "step": 56900
    },
    {
      "epoch": 44.25106879129421,
      "grad_norm": 0.8915746808052063,
      "learning_rate": 7.529e-05,
      "loss": 0.0147,
      "step": 56950
    },
    {
      "epoch": 44.28993392926545,
      "grad_norm": 0.151080921292305,
      "learning_rate": 7.526368421052632e-05,
      "loss": 0.0154,
      "step": 57000
    },
    {
      "epoch": 44.28993392926545,
      "eval_accuracy": 0.9985910703007336,
      "eval_runtime": 877.799,
      "eval_samples_per_second": 23.448,
      "eval_steps_per_second": 2.931,
      "step": 57000
    },
    {
      "epoch": 44.32879906723669,
      "grad_norm": 1.8628959655761719,
      "learning_rate": 7.523736842105263e-05,
      "loss": 0.0115,
      "step": 57050
    },
    {
      "epoch": 44.36766420520793,
      "grad_norm": 0.06097853183746338,
      "learning_rate": 7.521105263157895e-05,
      "loss": 0.0117,
      "step": 57100
    },
    {
      "epoch": 44.406529343179166,
      "grad_norm": 0.1226363331079483,
      "learning_rate": 7.518473684210527e-05,
      "loss": 0.0121,
      "step": 57150
    },
    {
      "epoch": 44.44539448115041,
      "grad_norm": 0.30275824666023254,
      "learning_rate": 7.515842105263158e-05,
      "loss": 0.0129,
      "step": 57200
    },
    {
      "epoch": 44.484259619121644,
      "grad_norm": 0.03589918836951256,
      "learning_rate": 7.51321052631579e-05,
      "loss": 0.0141,
      "step": 57250
    },
    {
      "epoch": 44.52312475709289,
      "grad_norm": 0.1084870994091034,
      "learning_rate": 7.510578947368421e-05,
      "loss": 0.0175,
      "step": 57300
    },
    {
      "epoch": 44.56198989506413,
      "grad_norm": 0.06478150933980942,
      "learning_rate": 7.507947368421054e-05,
      "loss": 0.0201,
      "step": 57350
    },
    {
      "epoch": 44.600855033035366,
      "grad_norm": 0.11772127449512482,
      "learning_rate": 7.505315789473685e-05,
      "loss": 0.018,
      "step": 57400
    },
    {
      "epoch": 44.63972017100661,
      "grad_norm": 3.1865267753601074,
      "learning_rate": 7.502684210526316e-05,
      "loss": 0.009,
      "step": 57450
    },
    {
      "epoch": 44.678585308977844,
      "grad_norm": 0.5008084774017334,
      "learning_rate": 7.500052631578947e-05,
      "loss": 0.0209,
      "step": 57500
    },
    {
      "epoch": 44.71745044694909,
      "grad_norm": 0.014076965861022472,
      "learning_rate": 7.497421052631579e-05,
      "loss": 0.0149,
      "step": 57550
    },
    {
      "epoch": 44.75631558492033,
      "grad_norm": 3.251145362854004,
      "learning_rate": 7.494789473684211e-05,
      "loss": 0.0166,
      "step": 57600
    },
    {
      "epoch": 44.795180722891565,
      "grad_norm": 0.050258051604032516,
      "learning_rate": 7.492157894736842e-05,
      "loss": 0.0171,
      "step": 57650
    },
    {
      "epoch": 44.83404586086281,
      "grad_norm": 2.849292039871216,
      "learning_rate": 7.489526315789473e-05,
      "loss": 0.0145,
      "step": 57700
    },
    {
      "epoch": 44.872910998834044,
      "grad_norm": 0.8286195993423462,
      "learning_rate": 7.486894736842106e-05,
      "loss": 0.0158,
      "step": 57750
    },
    {
      "epoch": 44.91177613680529,
      "grad_norm": 1.5031063556671143,
      "learning_rate": 7.484263157894737e-05,
      "loss": 0.0197,
      "step": 57800
    },
    {
      "epoch": 44.95064127477652,
      "grad_norm": 0.2209365963935852,
      "learning_rate": 7.48163157894737e-05,
      "loss": 0.0165,
      "step": 57850
    },
    {
      "epoch": 44.989506412747765,
      "grad_norm": 4.55959415435791,
      "learning_rate": 7.479000000000001e-05,
      "loss": 0.0187,
      "step": 57900
    },
    {
      "epoch": 45.02798289933929,
      "grad_norm": 0.03095565363764763,
      "learning_rate": 7.476368421052632e-05,
      "loss": 0.0128,
      "step": 57950
    },
    {
      "epoch": 45.066848037310535,
      "grad_norm": 0.09888409823179245,
      "learning_rate": 7.473736842105263e-05,
      "loss": 0.0097,
      "step": 58000
    },
    {
      "epoch": 45.066848037310535,
      "eval_accuracy": 0.9982024000388671,
      "eval_runtime": 873.7797,
      "eval_samples_per_second": 23.556,
      "eval_steps_per_second": 2.945,
      "step": 58000
    },
    {
      "epoch": 45.10571317528177,
      "grad_norm": 0.46758630871772766,
      "learning_rate": 7.471105263157894e-05,
      "loss": 0.0111,
      "step": 58050
    },
    {
      "epoch": 45.144578313253014,
      "grad_norm": 3.0760531425476074,
      "learning_rate": 7.468473684210527e-05,
      "loss": 0.0125,
      "step": 58100
    },
    {
      "epoch": 45.18344345122425,
      "grad_norm": 0.02062395215034485,
      "learning_rate": 7.465842105263158e-05,
      "loss": 0.0092,
      "step": 58150
    },
    {
      "epoch": 45.22230858919549,
      "grad_norm": 0.43851378560066223,
      "learning_rate": 7.46321052631579e-05,
      "loss": 0.0143,
      "step": 58200
    },
    {
      "epoch": 45.26117372716673,
      "grad_norm": 2.340397357940674,
      "learning_rate": 7.460578947368422e-05,
      "loss": 0.019,
      "step": 58250
    },
    {
      "epoch": 45.30003886513797,
      "grad_norm": 1.865161418914795,
      "learning_rate": 7.457947368421053e-05,
      "loss": 0.0123,
      "step": 58300
    },
    {
      "epoch": 45.338904003109214,
      "grad_norm": 0.6069706678390503,
      "learning_rate": 7.455315789473685e-05,
      "loss": 0.0139,
      "step": 58350
    },
    {
      "epoch": 45.37776914108045,
      "grad_norm": 0.08736221492290497,
      "learning_rate": 7.452684210526316e-05,
      "loss": 0.0134,
      "step": 58400
    },
    {
      "epoch": 45.41663427905169,
      "grad_norm": 2.229518175125122,
      "learning_rate": 7.450052631578948e-05,
      "loss": 0.0114,
      "step": 58450
    },
    {
      "epoch": 45.45549941702293,
      "grad_norm": 0.29379507899284363,
      "learning_rate": 7.447421052631579e-05,
      "loss": 0.0139,
      "step": 58500
    },
    {
      "epoch": 45.49436455499417,
      "grad_norm": 2.3412671089172363,
      "learning_rate": 7.444789473684211e-05,
      "loss": 0.0157,
      "step": 58550
    },
    {
      "epoch": 45.53322969296541,
      "grad_norm": 0.7046128511428833,
      "learning_rate": 7.442157894736842e-05,
      "loss": 0.0171,
      "step": 58600
    },
    {
      "epoch": 45.57209483093665,
      "grad_norm": 0.04889620468020439,
      "learning_rate": 7.439526315789475e-05,
      "loss": 0.0204,
      "step": 58650
    },
    {
      "epoch": 45.61095996890789,
      "grad_norm": 0.0866706445813179,
      "learning_rate": 7.436894736842106e-05,
      "loss": 0.0178,
      "step": 58700
    },
    {
      "epoch": 45.64982510687913,
      "grad_norm": 0.6070784330368042,
      "learning_rate": 7.434263157894737e-05,
      "loss": 0.0193,
      "step": 58750
    },
    {
      "epoch": 45.68869024485037,
      "grad_norm": 2.2081806659698486,
      "learning_rate": 7.431631578947368e-05,
      "loss": 0.0142,
      "step": 58800
    },
    {
      "epoch": 45.727555382821606,
      "grad_norm": 2.7623023986816406,
      "learning_rate": 7.429000000000001e-05,
      "loss": 0.0136,
      "step": 58850
    },
    {
      "epoch": 45.76642052079285,
      "grad_norm": 0.286423921585083,
      "learning_rate": 7.426368421052632e-05,
      "loss": 0.015,
      "step": 58900
    },
    {
      "epoch": 45.80528565876409,
      "grad_norm": 0.11053986847400665,
      "learning_rate": 7.423736842105263e-05,
      "loss": 0.0131,
      "step": 58950
    },
    {
      "epoch": 45.84415079673533,
      "grad_norm": 0.2008861005306244,
      "learning_rate": 7.421105263157894e-05,
      "loss": 0.0082,
      "step": 59000
    },
    {
      "epoch": 45.84415079673533,
      "eval_accuracy": 0.9987854054316669,
      "eval_runtime": 869.9812,
      "eval_samples_per_second": 23.659,
      "eval_steps_per_second": 2.958,
      "step": 59000
    },
    {
      "epoch": 45.88301593470657,
      "grad_norm": 0.1004604771733284,
      "learning_rate": 7.418473684210527e-05,
      "loss": 0.0158,
      "step": 59050
    },
    {
      "epoch": 45.921881072677806,
      "grad_norm": 1.3480688333511353,
      "learning_rate": 7.415842105263158e-05,
      "loss": 0.0103,
      "step": 59100
    },
    {
      "epoch": 45.96074621064905,
      "grad_norm": 0.016539020463824272,
      "learning_rate": 7.41321052631579e-05,
      "loss": 0.0115,
      "step": 59150
    },
    {
      "epoch": 45.999611348620284,
      "grad_norm": 0.07502561807632446,
      "learning_rate": 7.410578947368422e-05,
      "loss": 0.0161,
      "step": 59200
    },
    {
      "epoch": 46.03808783521181,
      "grad_norm": 0.060839392244815826,
      "learning_rate": 7.407947368421053e-05,
      "loss": 0.0125,
      "step": 59250
    },
    {
      "epoch": 46.076952973183054,
      "grad_norm": 1.0475828647613525,
      "learning_rate": 7.405315789473684e-05,
      "loss": 0.0087,
      "step": 59300
    },
    {
      "epoch": 46.1158181111543,
      "grad_norm": 0.024058906361460686,
      "learning_rate": 7.402684210526317e-05,
      "loss": 0.0106,
      "step": 59350
    },
    {
      "epoch": 46.15468324912553,
      "grad_norm": 0.049573808908462524,
      "learning_rate": 7.400052631578948e-05,
      "loss": 0.01,
      "step": 59400
    },
    {
      "epoch": 46.193548387096776,
      "grad_norm": 1.1914864778518677,
      "learning_rate": 7.397421052631579e-05,
      "loss": 0.0111,
      "step": 59450
    },
    {
      "epoch": 46.23241352506801,
      "grad_norm": 0.13890480995178223,
      "learning_rate": 7.39478947368421e-05,
      "loss": 0.012,
      "step": 59500
    },
    {
      "epoch": 46.271278663039254,
      "grad_norm": 0.18326811492443085,
      "learning_rate": 7.392157894736843e-05,
      "loss": 0.0093,
      "step": 59550
    },
    {
      "epoch": 46.3101438010105,
      "grad_norm": 0.5677682161331177,
      "learning_rate": 7.389526315789475e-05,
      "loss": 0.0112,
      "step": 59600
    },
    {
      "epoch": 46.34900893898173,
      "grad_norm": 0.02287537045776844,
      "learning_rate": 7.386894736842106e-05,
      "loss": 0.0109,
      "step": 59650
    },
    {
      "epoch": 46.387874076952976,
      "grad_norm": 0.07677704095840454,
      "learning_rate": 7.384263157894737e-05,
      "loss": 0.0061,
      "step": 59700
    },
    {
      "epoch": 46.42673921492421,
      "grad_norm": 2.4414875507354736,
      "learning_rate": 7.381631578947368e-05,
      "loss": 0.0163,
      "step": 59750
    },
    {
      "epoch": 46.465604352895454,
      "grad_norm": 1.8416739702224731,
      "learning_rate": 7.379000000000001e-05,
      "loss": 0.0136,
      "step": 59800
    },
    {
      "epoch": 46.50446949086669,
      "grad_norm": 3.0992000102996826,
      "learning_rate": 7.376368421052632e-05,
      "loss": 0.0233,
      "step": 59850
    },
    {
      "epoch": 46.54333462883793,
      "grad_norm": 0.06737317144870758,
      "learning_rate": 7.373736842105263e-05,
      "loss": 0.0135,
      "step": 59900
    },
    {
      "epoch": 46.582199766809175,
      "grad_norm": 0.14803974330425262,
      "learning_rate": 7.371105263157894e-05,
      "loss": 0.0133,
      "step": 59950
    },
    {
      "epoch": 46.62106490478041,
      "grad_norm": 0.18518660962581635,
      "learning_rate": 7.368473684210526e-05,
      "loss": 0.0163,
      "step": 60000
    },
    {
      "epoch": 46.62106490478041,
      "eval_accuracy": 0.9984453189525336,
      "eval_runtime": 877.3612,
      "eval_samples_per_second": 23.46,
      "eval_steps_per_second": 2.933,
      "step": 60000
    },
    {
      "epoch": 46.659930042751654,
      "grad_norm": 0.07843168079853058,
      "learning_rate": 7.365842105263158e-05,
      "loss": 0.0164,
      "step": 60050
    },
    {
      "epoch": 46.69879518072289,
      "grad_norm": 0.48835334181785583,
      "learning_rate": 7.36321052631579e-05,
      "loss": 0.0095,
      "step": 60100
    },
    {
      "epoch": 46.73766031869413,
      "grad_norm": 0.03689779341220856,
      "learning_rate": 7.360578947368422e-05,
      "loss": 0.0089,
      "step": 60150
    },
    {
      "epoch": 46.77652545666537,
      "grad_norm": 0.12304223328828812,
      "learning_rate": 7.357947368421053e-05,
      "loss": 0.013,
      "step": 60200
    },
    {
      "epoch": 46.81539059463661,
      "grad_norm": 0.028829611837863922,
      "learning_rate": 7.355315789473684e-05,
      "loss": 0.0095,
      "step": 60250
    },
    {
      "epoch": 46.854255732607854,
      "grad_norm": 0.1779215931892395,
      "learning_rate": 7.352684210526317e-05,
      "loss": 0.0203,
      "step": 60300
    },
    {
      "epoch": 46.89312087057909,
      "grad_norm": 0.027951283380389214,
      "learning_rate": 7.350052631578948e-05,
      "loss": 0.0242,
      "step": 60350
    },
    {
      "epoch": 46.93198600855033,
      "grad_norm": 0.24708083271980286,
      "learning_rate": 7.347421052631579e-05,
      "loss": 0.0179,
      "step": 60400
    },
    {
      "epoch": 46.97085114652157,
      "grad_norm": 3.7258784770965576,
      "learning_rate": 7.34478947368421e-05,
      "loss": 0.0094,
      "step": 60450
    },
    {
      "epoch": 47.009327633113095,
      "grad_norm": 3.078702688217163,
      "learning_rate": 7.342157894736841e-05,
      "loss": 0.0184,
      "step": 60500
    },
    {
      "epoch": 47.04819277108434,
      "grad_norm": 0.09043591469526291,
      "learning_rate": 7.339526315789474e-05,
      "loss": 0.0201,
      "step": 60550
    },
    {
      "epoch": 47.087057909055574,
      "grad_norm": 0.01854565367102623,
      "learning_rate": 7.336894736842106e-05,
      "loss": 0.0093,
      "step": 60600
    },
    {
      "epoch": 47.12592304702682,
      "grad_norm": 0.008243966847658157,
      "learning_rate": 7.334263157894737e-05,
      "loss": 0.0173,
      "step": 60650
    },
    {
      "epoch": 47.16478818499806,
      "grad_norm": 0.0831562727689743,
      "learning_rate": 7.331631578947369e-05,
      "loss": 0.0094,
      "step": 60700
    },
    {
      "epoch": 47.203653322969295,
      "grad_norm": 2.855015516281128,
      "learning_rate": 7.329e-05,
      "loss": 0.0073,
      "step": 60750
    },
    {
      "epoch": 47.24251846094054,
      "grad_norm": 1.0864886045455933,
      "learning_rate": 7.326368421052632e-05,
      "loss": 0.0096,
      "step": 60800
    },
    {
      "epoch": 47.28138359891177,
      "grad_norm": 0.05669767037034035,
      "learning_rate": 7.323736842105263e-05,
      "loss": 0.0122,
      "step": 60850
    },
    {
      "epoch": 47.320248736883016,
      "grad_norm": 0.26644644141197205,
      "learning_rate": 7.321105263157895e-05,
      "loss": 0.0058,
      "step": 60900
    },
    {
      "epoch": 47.35911387485426,
      "grad_norm": 0.01469517033547163,
      "learning_rate": 7.318473684210526e-05,
      "loss": 0.0127,
      "step": 60950
    },
    {
      "epoch": 47.397979012825495,
      "grad_norm": 0.21100173890590668,
      "learning_rate": 7.315842105263158e-05,
      "loss": 0.0125,
      "step": 61000
    },
    {
      "epoch": 47.397979012825495,
      "eval_accuracy": 0.998493902735267,
      "eval_runtime": 849.6945,
      "eval_samples_per_second": 24.224,
      "eval_steps_per_second": 3.028,
      "step": 61000
    },
    {
      "epoch": 47.43684415079674,
      "grad_norm": 0.06516958773136139,
      "learning_rate": 7.313210526315791e-05,
      "loss": 0.0136,
      "step": 61050
    },
    {
      "epoch": 47.47570928876797,
      "grad_norm": 0.05053483694791794,
      "learning_rate": 7.310578947368422e-05,
      "loss": 0.0093,
      "step": 61100
    },
    {
      "epoch": 47.514574426739216,
      "grad_norm": 0.5198332667350769,
      "learning_rate": 7.307947368421053e-05,
      "loss": 0.0112,
      "step": 61150
    },
    {
      "epoch": 47.55343956471045,
      "grad_norm": 0.052330415695905685,
      "learning_rate": 7.305315789473684e-05,
      "loss": 0.011,
      "step": 61200
    },
    {
      "epoch": 47.592304702681695,
      "grad_norm": 0.7372391223907471,
      "learning_rate": 7.302684210526315e-05,
      "loss": 0.0115,
      "step": 61250
    },
    {
      "epoch": 47.63116984065294,
      "grad_norm": 0.9434056878089905,
      "learning_rate": 7.300052631578948e-05,
      "loss": 0.0136,
      "step": 61300
    },
    {
      "epoch": 47.67003497862417,
      "grad_norm": 0.547739565372467,
      "learning_rate": 7.297421052631579e-05,
      "loss": 0.0185,
      "step": 61350
    },
    {
      "epoch": 47.708900116595416,
      "grad_norm": 0.1716149002313614,
      "learning_rate": 7.29478947368421e-05,
      "loss": 0.0135,
      "step": 61400
    },
    {
      "epoch": 47.74776525456665,
      "grad_norm": 0.1340044140815735,
      "learning_rate": 7.292157894736843e-05,
      "loss": 0.012,
      "step": 61450
    },
    {
      "epoch": 47.786630392537894,
      "grad_norm": 1.5391032695770264,
      "learning_rate": 7.289526315789474e-05,
      "loss": 0.0108,
      "step": 61500
    },
    {
      "epoch": 47.82549553050913,
      "grad_norm": 0.813843846321106,
      "learning_rate": 7.286894736842106e-05,
      "loss": 0.0121,
      "step": 61550
    },
    {
      "epoch": 47.86436066848037,
      "grad_norm": 3.4990172386169434,
      "learning_rate": 7.284263157894738e-05,
      "loss": 0.013,
      "step": 61600
    },
    {
      "epoch": 47.903225806451616,
      "grad_norm": 0.053542912006378174,
      "learning_rate": 7.281631578947369e-05,
      "loss": 0.0074,
      "step": 61650
    },
    {
      "epoch": 47.94209094442285,
      "grad_norm": 0.01410434115678072,
      "learning_rate": 7.279e-05,
      "loss": 0.0189,
      "step": 61700
    },
    {
      "epoch": 47.980956082394094,
      "grad_norm": 0.09987429529428482,
      "learning_rate": 7.276368421052631e-05,
      "loss": 0.0109,
      "step": 61750
    },
    {
      "epoch": 48.01943256898562,
      "grad_norm": 0.6762978434562683,
      "learning_rate": 7.273736842105264e-05,
      "loss": 0.0156,
      "step": 61800
    },
    {
      "epoch": 48.05829770695686,
      "grad_norm": 0.04120064899325371,
      "learning_rate": 7.271105263157895e-05,
      "loss": 0.0163,
      "step": 61850
    },
    {
      "epoch": 48.0971628449281,
      "grad_norm": 0.0444943942129612,
      "learning_rate": 7.268473684210527e-05,
      "loss": 0.012,
      "step": 61900
    },
    {
      "epoch": 48.13602798289934,
      "grad_norm": 1.1446350812911987,
      "learning_rate": 7.265842105263158e-05,
      "loss": 0.0153,
      "step": 61950
    },
    {
      "epoch": 48.17489312087058,
      "grad_norm": 0.03808463364839554,
      "learning_rate": 7.26321052631579e-05,
      "loss": 0.014,
      "step": 62000
    },
    {
      "epoch": 48.17489312087058,
      "eval_accuracy": 0.9986396540834669,
      "eval_runtime": 851.9896,
      "eval_samples_per_second": 24.159,
      "eval_steps_per_second": 3.02,
      "step": 62000
    },
    {
      "epoch": 48.21375825884182,
      "grad_norm": 0.23002126812934875,
      "learning_rate": 7.260578947368422e-05,
      "loss": 0.0105,
      "step": 62050
    },
    {
      "epoch": 48.25262339681306,
      "grad_norm": 0.052425310015678406,
      "learning_rate": 7.257947368421053e-05,
      "loss": 0.0149,
      "step": 62100
    },
    {
      "epoch": 48.2914885347843,
      "grad_norm": 1.247965693473816,
      "learning_rate": 7.255315789473684e-05,
      "loss": 0.01,
      "step": 62150
    },
    {
      "epoch": 48.330353672755535,
      "grad_norm": 0.17689017951488495,
      "learning_rate": 7.252684210526316e-05,
      "loss": 0.0146,
      "step": 62200
    },
    {
      "epoch": 48.36921881072678,
      "grad_norm": 0.10547871887683868,
      "learning_rate": 7.250052631578948e-05,
      "loss": 0.0108,
      "step": 62250
    },
    {
      "epoch": 48.40808394869802,
      "grad_norm": 0.049669452011585236,
      "learning_rate": 7.247421052631579e-05,
      "loss": 0.0083,
      "step": 62300
    },
    {
      "epoch": 48.44694908666926,
      "grad_norm": 0.9575231075286865,
      "learning_rate": 7.24478947368421e-05,
      "loss": 0.0114,
      "step": 62350
    },
    {
      "epoch": 48.4858142246405,
      "grad_norm": 1.0744680166244507,
      "learning_rate": 7.242157894736843e-05,
      "loss": 0.0119,
      "step": 62400
    },
    {
      "epoch": 48.524679362611735,
      "grad_norm": 0.05026625469326973,
      "learning_rate": 7.239526315789474e-05,
      "loss": 0.0107,
      "step": 62450
    },
    {
      "epoch": 48.56354450058298,
      "grad_norm": 0.013055695220828056,
      "learning_rate": 7.236894736842105e-05,
      "loss": 0.0136,
      "step": 62500
    },
    {
      "epoch": 48.602409638554214,
      "grad_norm": 0.5503189563751221,
      "learning_rate": 7.234263157894738e-05,
      "loss": 0.0107,
      "step": 62550
    },
    {
      "epoch": 48.64127477652546,
      "grad_norm": 0.5869662165641785,
      "learning_rate": 7.231631578947369e-05,
      "loss": 0.0165,
      "step": 62600
    },
    {
      "epoch": 48.6801399144967,
      "grad_norm": 0.45683395862579346,
      "learning_rate": 7.229e-05,
      "loss": 0.0086,
      "step": 62650
    },
    {
      "epoch": 48.719005052467935,
      "grad_norm": 0.046745553612709045,
      "learning_rate": 7.226368421052631e-05,
      "loss": 0.0119,
      "step": 62700
    },
    {
      "epoch": 48.75787019043918,
      "grad_norm": 0.08209559321403503,
      "learning_rate": 7.223736842105264e-05,
      "loss": 0.0087,
      "step": 62750
    },
    {
      "epoch": 48.79673532841041,
      "grad_norm": 0.7342914938926697,
      "learning_rate": 7.221105263157895e-05,
      "loss": 0.0086,
      "step": 62800
    },
    {
      "epoch": 48.835600466381656,
      "grad_norm": 0.2686988413333893,
      "learning_rate": 7.218473684210527e-05,
      "loss": 0.0249,
      "step": 62850
    },
    {
      "epoch": 48.87446560435289,
      "grad_norm": 0.045273009687662125,
      "learning_rate": 7.215842105263159e-05,
      "loss": 0.01,
      "step": 62900
    },
    {
      "epoch": 48.913330742324135,
      "grad_norm": 0.5616592168807983,
      "learning_rate": 7.21321052631579e-05,
      "loss": 0.0189,
      "step": 62950
    },
    {
      "epoch": 48.95219588029538,
      "grad_norm": 0.11938033998012543,
      "learning_rate": 7.210578947368421e-05,
      "loss": 0.0142,
      "step": 63000
    },
    {
      "epoch": 48.95219588029538,
      "eval_accuracy": 0.9986396540834669,
      "eval_runtime": 852.2966,
      "eval_samples_per_second": 24.15,
      "eval_steps_per_second": 3.019,
      "step": 63000
    },
    {
      "epoch": 48.99106101826661,
      "grad_norm": 1.7056082487106323,
      "learning_rate": 7.207947368421053e-05,
      "loss": 0.0142,
      "step": 63050
    },
    {
      "epoch": 49.02953750485814,
      "grad_norm": 0.026881588622927666,
      "learning_rate": 7.205315789473685e-05,
      "loss": 0.0164,
      "step": 63100
    },
    {
      "epoch": 49.06840264282938,
      "grad_norm": 0.5080930590629578,
      "learning_rate": 7.202684210526316e-05,
      "loss": 0.0087,
      "step": 63150
    },
    {
      "epoch": 49.10726778080062,
      "grad_norm": 0.6311091780662537,
      "learning_rate": 7.200052631578947e-05,
      "loss": 0.0119,
      "step": 63200
    },
    {
      "epoch": 49.14613291877186,
      "grad_norm": 0.007207009941339493,
      "learning_rate": 7.19742105263158e-05,
      "loss": 0.0101,
      "step": 63250
    },
    {
      "epoch": 49.184998056743105,
      "grad_norm": 0.6172986626625061,
      "learning_rate": 7.194789473684212e-05,
      "loss": 0.0084,
      "step": 63300
    },
    {
      "epoch": 49.22386319471434,
      "grad_norm": 0.12769931554794312,
      "learning_rate": 7.192157894736843e-05,
      "loss": 0.0074,
      "step": 63350
    },
    {
      "epoch": 49.26272833268558,
      "grad_norm": 0.22103777527809143,
      "learning_rate": 7.189526315789474e-05,
      "loss": 0.0246,
      "step": 63400
    },
    {
      "epoch": 49.30159347065682,
      "grad_norm": 1.9517908096313477,
      "learning_rate": 7.186894736842105e-05,
      "loss": 0.0095,
      "step": 63450
    },
    {
      "epoch": 49.34045860862806,
      "grad_norm": 0.9385688304901123,
      "learning_rate": 7.184263157894738e-05,
      "loss": 0.0133,
      "step": 63500
    },
    {
      "epoch": 49.3793237465993,
      "grad_norm": 0.32060927152633667,
      "learning_rate": 7.181631578947369e-05,
      "loss": 0.0052,
      "step": 63550
    },
    {
      "epoch": 49.41818888457054,
      "grad_norm": 3.535792350769043,
      "learning_rate": 7.179e-05,
      "loss": 0.0113,
      "step": 63600
    },
    {
      "epoch": 49.45705402254178,
      "grad_norm": 0.07511692494153976,
      "learning_rate": 7.176368421052631e-05,
      "loss": 0.0097,
      "step": 63650
    },
    {
      "epoch": 49.49591916051302,
      "grad_norm": 0.19153301417827606,
      "learning_rate": 7.173736842105263e-05,
      "loss": 0.0159,
      "step": 63700
    },
    {
      "epoch": 49.53478429848426,
      "grad_norm": 0.2613302171230316,
      "learning_rate": 7.171105263157895e-05,
      "loss": 0.012,
      "step": 63750
    },
    {
      "epoch": 49.5736494364555,
      "grad_norm": 0.9980623722076416,
      "learning_rate": 7.168473684210528e-05,
      "loss": 0.0119,
      "step": 63800
    },
    {
      "epoch": 49.61251457442674,
      "grad_norm": 0.07133900374174118,
      "learning_rate": 7.165842105263159e-05,
      "loss": 0.0126,
      "step": 63850
    },
    {
      "epoch": 49.651379712397976,
      "grad_norm": 1.0189536809921265,
      "learning_rate": 7.16321052631579e-05,
      "loss": 0.0091,
      "step": 63900
    },
    {
      "epoch": 49.69024485036922,
      "grad_norm": 0.18489180505275726,
      "learning_rate": 7.160578947368421e-05,
      "loss": 0.0076,
      "step": 63950
    },
    {
      "epoch": 49.72910998834046,
      "grad_norm": 0.25034400820732117,
      "learning_rate": 7.157947368421054e-05,
      "loss": 0.0092,
      "step": 64000
    },
    {
      "epoch": 49.72910998834046,
      "eval_accuracy": 0.9987854054316669,
      "eval_runtime": 854.0151,
      "eval_samples_per_second": 24.101,
      "eval_steps_per_second": 3.013,
      "step": 64000
    },
    {
      "epoch": 49.7679751263117,
      "grad_norm": 0.03618240728974342,
      "learning_rate": 7.155315789473685e-05,
      "loss": 0.0132,
      "step": 64050
    },
    {
      "epoch": 49.80684026428294,
      "grad_norm": 0.04059286043047905,
      "learning_rate": 7.152684210526316e-05,
      "loss": 0.0127,
      "step": 64100
    },
    {
      "epoch": 49.845705402254175,
      "grad_norm": 0.31507959961891174,
      "learning_rate": 7.150052631578947e-05,
      "loss": 0.0099,
      "step": 64150
    },
    {
      "epoch": 49.88457054022542,
      "grad_norm": 0.14851006865501404,
      "learning_rate": 7.147421052631578e-05,
      "loss": 0.0092,
      "step": 64200
    },
    {
      "epoch": 49.92343567819666,
      "grad_norm": 0.2910158336162567,
      "learning_rate": 7.144789473684211e-05,
      "loss": 0.0091,
      "step": 64250
    },
    {
      "epoch": 49.9623008161679,
      "grad_norm": 0.9559641480445862,
      "learning_rate": 7.142157894736843e-05,
      "loss": 0.0156,
      "step": 64300
    },
    {
      "epoch": 50.000777302759424,
      "grad_norm": 0.012824985198676586,
      "learning_rate": 7.139526315789474e-05,
      "loss": 0.0085,
      "step": 64350
    },
    {
      "epoch": 50.03964244073067,
      "grad_norm": 0.12682408094406128,
      "learning_rate": 7.136894736842106e-05,
      "loss": 0.0129,
      "step": 64400
    },
    {
      "epoch": 50.0785075787019,
      "grad_norm": 0.025887005031108856,
      "learning_rate": 7.134263157894737e-05,
      "loss": 0.0063,
      "step": 64450
    },
    {
      "epoch": 50.117372716673145,
      "grad_norm": 0.06498356908559799,
      "learning_rate": 7.131631578947369e-05,
      "loss": 0.01,
      "step": 64500
    },
    {
      "epoch": 50.15623785464438,
      "grad_norm": 2.5489485263824463,
      "learning_rate": 7.129e-05,
      "loss": 0.0118,
      "step": 64550
    },
    {
      "epoch": 50.195102992615624,
      "grad_norm": 0.08452356606721878,
      "learning_rate": 7.126368421052632e-05,
      "loss": 0.0137,
      "step": 64600
    },
    {
      "epoch": 50.23396813058687,
      "grad_norm": 3.793992757797241,
      "learning_rate": 7.123736842105263e-05,
      "loss": 0.01,
      "step": 64650
    },
    {
      "epoch": 50.2728332685581,
      "grad_norm": 0.2957603335380554,
      "learning_rate": 7.121105263157895e-05,
      "loss": 0.0157,
      "step": 64700
    },
    {
      "epoch": 50.311698406529345,
      "grad_norm": 0.11913411319255829,
      "learning_rate": 7.118473684210528e-05,
      "loss": 0.0144,
      "step": 64750
    },
    {
      "epoch": 50.35056354450058,
      "grad_norm": 0.07215093076229095,
      "learning_rate": 7.115842105263159e-05,
      "loss": 0.0199,
      "step": 64800
    },
    {
      "epoch": 50.389428682471824,
      "grad_norm": 0.04829329997301102,
      "learning_rate": 7.11321052631579e-05,
      "loss": 0.0165,
      "step": 64850
    },
    {
      "epoch": 50.42829382044306,
      "grad_norm": 0.13153165578842163,
      "learning_rate": 7.110578947368421e-05,
      "loss": 0.0151,
      "step": 64900
    },
    {
      "epoch": 50.4671589584143,
      "grad_norm": 0.029550815001130104,
      "learning_rate": 7.107947368421052e-05,
      "loss": 0.0101,
      "step": 64950
    },
    {
      "epoch": 50.506024096385545,
      "grad_norm": 0.010725216940045357,
      "learning_rate": 7.105315789473685e-05,
      "loss": 0.0069,
      "step": 65000
    },
    {
      "epoch": 50.506024096385545,
      "eval_accuracy": 0.9988825729971336,
      "eval_runtime": 858.6368,
      "eval_samples_per_second": 23.972,
      "eval_steps_per_second": 2.997,
      "step": 65000
    },
    {
      "epoch": 50.54488923435678,
      "grad_norm": 0.3737189471721649,
      "learning_rate": 7.102684210526316e-05,
      "loss": 0.0088,
      "step": 65050
    },
    {
      "epoch": 50.58375437232802,
      "grad_norm": 0.07762659341096878,
      "learning_rate": 7.100052631578947e-05,
      "loss": 0.0122,
      "step": 65100
    },
    {
      "epoch": 50.62261951029926,
      "grad_norm": 0.008183227851986885,
      "learning_rate": 7.097421052631578e-05,
      "loss": 0.0075,
      "step": 65150
    },
    {
      "epoch": 50.6614846482705,
      "grad_norm": 1.5395113229751587,
      "learning_rate": 7.094789473684211e-05,
      "loss": 0.0127,
      "step": 65200
    },
    {
      "epoch": 50.700349786241745,
      "grad_norm": 1.8640962839126587,
      "learning_rate": 7.092157894736843e-05,
      "loss": 0.0159,
      "step": 65250
    },
    {
      "epoch": 50.73921492421298,
      "grad_norm": 0.3395310938358307,
      "learning_rate": 7.089526315789475e-05,
      "loss": 0.0115,
      "step": 65300
    },
    {
      "epoch": 50.77808006218422,
      "grad_norm": 0.5718401074409485,
      "learning_rate": 7.086894736842106e-05,
      "loss": 0.0149,
      "step": 65350
    },
    {
      "epoch": 50.81694520015546,
      "grad_norm": 0.6449897885322571,
      "learning_rate": 7.084263157894737e-05,
      "loss": 0.014,
      "step": 65400
    },
    {
      "epoch": 50.8558103381267,
      "grad_norm": 0.24561935663223267,
      "learning_rate": 7.081631578947368e-05,
      "loss": 0.0101,
      "step": 65450
    },
    {
      "epoch": 50.89467547609794,
      "grad_norm": 0.222761869430542,
      "learning_rate": 7.079e-05,
      "loss": 0.0092,
      "step": 65500
    },
    {
      "epoch": 50.93354061406918,
      "grad_norm": 3.4988646507263184,
      "learning_rate": 7.076368421052632e-05,
      "loss": 0.0148,
      "step": 65550
    },
    {
      "epoch": 50.97240575204042,
      "grad_norm": 1.750221610069275,
      "learning_rate": 7.073736842105263e-05,
      "loss": 0.0127,
      "step": 65600
    },
    {
      "epoch": 51.01088223863195,
      "grad_norm": 0.1166532039642334,
      "learning_rate": 7.071105263157895e-05,
      "loss": 0.0045,
      "step": 65650
    },
    {
      "epoch": 51.049747376603186,
      "grad_norm": 2.3461062908172607,
      "learning_rate": 7.068473684210526e-05,
      "loss": 0.0045,
      "step": 65700
    },
    {
      "epoch": 51.08861251457443,
      "grad_norm": 0.01984625868499279,
      "learning_rate": 7.065842105263159e-05,
      "loss": 0.0098,
      "step": 65750
    },
    {
      "epoch": 51.127477652545664,
      "grad_norm": 1.5265790224075317,
      "learning_rate": 7.06321052631579e-05,
      "loss": 0.0076,
      "step": 65800
    },
    {
      "epoch": 51.16634279051691,
      "grad_norm": 0.029023034498095512,
      "learning_rate": 7.060578947368421e-05,
      "loss": 0.0093,
      "step": 65850
    },
    {
      "epoch": 51.20520792848814,
      "grad_norm": 0.08802133798599243,
      "learning_rate": 7.057947368421052e-05,
      "loss": 0.0037,
      "step": 65900
    },
    {
      "epoch": 51.244073066459386,
      "grad_norm": 0.2991699278354645,
      "learning_rate": 7.055315789473685e-05,
      "loss": 0.0093,
      "step": 65950
    },
    {
      "epoch": 51.28293820443063,
      "grad_norm": 2.19167160987854,
      "learning_rate": 7.052684210526316e-05,
      "loss": 0.0087,
      "step": 66000
    },
    {
      "epoch": 51.28293820443063,
      "eval_accuracy": 0.9989311567798669,
      "eval_runtime": 854.2184,
      "eval_samples_per_second": 24.096,
      "eval_steps_per_second": 3.012,
      "step": 66000
    },
    {
      "epoch": 51.321803342401864,
      "grad_norm": 0.21082434058189392,
      "learning_rate": 7.050052631578947e-05,
      "loss": 0.0098,
      "step": 66050
    },
    {
      "epoch": 51.36066848037311,
      "grad_norm": 0.013236195780336857,
      "learning_rate": 7.04742105263158e-05,
      "loss": 0.0136,
      "step": 66100
    },
    {
      "epoch": 51.39953361834434,
      "grad_norm": 1.2647536993026733,
      "learning_rate": 7.044789473684211e-05,
      "loss": 0.0087,
      "step": 66150
    },
    {
      "epoch": 51.438398756315586,
      "grad_norm": 0.042822957038879395,
      "learning_rate": 7.042157894736842e-05,
      "loss": 0.0161,
      "step": 66200
    },
    {
      "epoch": 51.47726389428682,
      "grad_norm": 0.0055616190657019615,
      "learning_rate": 7.039526315789475e-05,
      "loss": 0.0163,
      "step": 66250
    },
    {
      "epoch": 51.516129032258064,
      "grad_norm": 0.6984388828277588,
      "learning_rate": 7.036894736842106e-05,
      "loss": 0.0066,
      "step": 66300
    },
    {
      "epoch": 51.55499417022931,
      "grad_norm": 0.17050936818122864,
      "learning_rate": 7.034263157894737e-05,
      "loss": 0.008,
      "step": 66350
    },
    {
      "epoch": 51.59385930820054,
      "grad_norm": 0.04073571413755417,
      "learning_rate": 7.031631578947368e-05,
      "loss": 0.012,
      "step": 66400
    },
    {
      "epoch": 51.632724446171785,
      "grad_norm": 2.6651530265808105,
      "learning_rate": 7.029e-05,
      "loss": 0.013,
      "step": 66450
    },
    {
      "epoch": 51.67158958414302,
      "grad_norm": 0.2084534913301468,
      "learning_rate": 7.026368421052632e-05,
      "loss": 0.0107,
      "step": 66500
    },
    {
      "epoch": 51.710454722114264,
      "grad_norm": 0.3147223889827728,
      "learning_rate": 7.023736842105263e-05,
      "loss": 0.0145,
      "step": 66550
    },
    {
      "epoch": 51.74931986008551,
      "grad_norm": 0.026732545346021652,
      "learning_rate": 7.021105263157895e-05,
      "loss": 0.0134,
      "step": 66600
    },
    {
      "epoch": 51.78818499805674,
      "grad_norm": 1.0873422622680664,
      "learning_rate": 7.018473684210527e-05,
      "loss": 0.0127,
      "step": 66650
    },
    {
      "epoch": 51.827050136027985,
      "grad_norm": 0.040288642048835754,
      "learning_rate": 7.015842105263158e-05,
      "loss": 0.0122,
      "step": 66700
    },
    {
      "epoch": 51.86591527399922,
      "grad_norm": 0.10090038180351257,
      "learning_rate": 7.01321052631579e-05,
      "loss": 0.0154,
      "step": 66750
    },
    {
      "epoch": 51.904780411970464,
      "grad_norm": 0.007838579826056957,
      "learning_rate": 7.010578947368421e-05,
      "loss": 0.0228,
      "step": 66800
    },
    {
      "epoch": 51.9436455499417,
      "grad_norm": 0.0807645320892334,
      "learning_rate": 7.007947368421053e-05,
      "loss": 0.0087,
      "step": 66850
    },
    {
      "epoch": 51.98251068791294,
      "grad_norm": 0.5056226849555969,
      "learning_rate": 7.005315789473684e-05,
      "loss": 0.0115,
      "step": 66900
    },
    {
      "epoch": 52.02098717450447,
      "grad_norm": 0.03711940720677376,
      "learning_rate": 7.002684210526316e-05,
      "loss": 0.0075,
      "step": 66950
    },
    {
      "epoch": 52.05985231247571,
      "grad_norm": 0.18952326476573944,
      "learning_rate": 7.000052631578947e-05,
      "loss": 0.0079,
      "step": 67000
    },
    {
      "epoch": 52.05985231247571,
      "eval_accuracy": 0.9989311567798669,
      "eval_runtime": 851.5324,
      "eval_samples_per_second": 24.172,
      "eval_steps_per_second": 3.022,
      "step": 67000
    },
    {
      "epoch": 52.09871745044695,
      "grad_norm": 0.048657167702913284,
      "learning_rate": 6.99742105263158e-05,
      "loss": 0.0162,
      "step": 67050
    },
    {
      "epoch": 52.13758258841819,
      "grad_norm": 0.28034019470214844,
      "learning_rate": 6.994789473684211e-05,
      "loss": 0.0044,
      "step": 67100
    },
    {
      "epoch": 52.176447726389426,
      "grad_norm": 0.060565035790205,
      "learning_rate": 6.992157894736842e-05,
      "loss": 0.0097,
      "step": 67150
    },
    {
      "epoch": 52.21531286436067,
      "grad_norm": 0.13445107638835907,
      "learning_rate": 6.989526315789475e-05,
      "loss": 0.011,
      "step": 67200
    },
    {
      "epoch": 52.254178002331905,
      "grad_norm": 0.060040175914764404,
      "learning_rate": 6.986894736842106e-05,
      "loss": 0.0065,
      "step": 67250
    },
    {
      "epoch": 52.29304314030315,
      "grad_norm": 0.11060953140258789,
      "learning_rate": 6.984263157894737e-05,
      "loss": 0.0151,
      "step": 67300
    },
    {
      "epoch": 52.33190827827439,
      "grad_norm": 0.6473913788795471,
      "learning_rate": 6.981631578947368e-05,
      "loss": 0.0085,
      "step": 67350
    },
    {
      "epoch": 52.370773416245626,
      "grad_norm": 0.02483026310801506,
      "learning_rate": 6.979e-05,
      "loss": 0.0083,
      "step": 67400
    },
    {
      "epoch": 52.40963855421687,
      "grad_norm": 0.03834308311343193,
      "learning_rate": 6.976368421052632e-05,
      "loss": 0.0057,
      "step": 67450
    },
    {
      "epoch": 52.448503692188105,
      "grad_norm": 0.16416214406490326,
      "learning_rate": 6.973736842105264e-05,
      "loss": 0.011,
      "step": 67500
    },
    {
      "epoch": 52.48736883015935,
      "grad_norm": 0.012599085457623005,
      "learning_rate": 6.971105263157896e-05,
      "loss": 0.0065,
      "step": 67550
    },
    {
      "epoch": 52.52623396813059,
      "grad_norm": 0.04111713543534279,
      "learning_rate": 6.968473684210527e-05,
      "loss": 0.0112,
      "step": 67600
    },
    {
      "epoch": 52.565099106101826,
      "grad_norm": 0.003991789184510708,
      "learning_rate": 6.965842105263158e-05,
      "loss": 0.0131,
      "step": 67650
    },
    {
      "epoch": 52.60396424407307,
      "grad_norm": 0.05732455104589462,
      "learning_rate": 6.96321052631579e-05,
      "loss": 0.0107,
      "step": 67700
    },
    {
      "epoch": 52.642829382044305,
      "grad_norm": 0.1286925971508026,
      "learning_rate": 6.960578947368422e-05,
      "loss": 0.0152,
      "step": 67750
    },
    {
      "epoch": 52.68169452001555,
      "grad_norm": 0.13254734873771667,
      "learning_rate": 6.957947368421053e-05,
      "loss": 0.011,
      "step": 67800
    },
    {
      "epoch": 52.72055965798678,
      "grad_norm": 1.788832187652588,
      "learning_rate": 6.955315789473684e-05,
      "loss": 0.008,
      "step": 67850
    },
    {
      "epoch": 52.759424795958026,
      "grad_norm": 0.010066797025501728,
      "learning_rate": 6.952684210526315e-05,
      "loss": 0.0107,
      "step": 67900
    },
    {
      "epoch": 52.79828993392927,
      "grad_norm": 0.12101956456899643,
      "learning_rate": 6.950052631578948e-05,
      "loss": 0.0095,
      "step": 67950
    },
    {
      "epoch": 52.837155071900504,
      "grad_norm": 0.42573440074920654,
      "learning_rate": 6.94742105263158e-05,
      "loss": 0.0088,
      "step": 68000
    },
    {
      "epoch": 52.837155071900504,
      "eval_accuracy": 0.9985910703007336,
      "eval_runtime": 855.217,
      "eval_samples_per_second": 24.068,
      "eval_steps_per_second": 3.009,
      "step": 68000
    },
    {
      "epoch": 52.87602020987175,
      "grad_norm": 0.5126527547836304,
      "learning_rate": 6.944789473684211e-05,
      "loss": 0.0071,
      "step": 68050
    },
    {
      "epoch": 52.91488534784298,
      "grad_norm": 0.46387553215026855,
      "learning_rate": 6.942157894736842e-05,
      "loss": 0.0107,
      "step": 68100
    },
    {
      "epoch": 52.953750485814226,
      "grad_norm": 0.572813868522644,
      "learning_rate": 6.939526315789474e-05,
      "loss": 0.0089,
      "step": 68150
    },
    {
      "epoch": 52.99261562378546,
      "grad_norm": 0.013004188425838947,
      "learning_rate": 6.936894736842106e-05,
      "loss": 0.0147,
      "step": 68200
    },
    {
      "epoch": 53.03109211037699,
      "grad_norm": 0.2339731901884079,
      "learning_rate": 6.934263157894737e-05,
      "loss": 0.0118,
      "step": 68250
    },
    {
      "epoch": 53.06995724834823,
      "grad_norm": 0.0777270719408989,
      "learning_rate": 6.931631578947368e-05,
      "loss": 0.0159,
      "step": 68300
    },
    {
      "epoch": 53.108822386319474,
      "grad_norm": 0.6745654940605164,
      "learning_rate": 6.929e-05,
      "loss": 0.0032,
      "step": 68350
    },
    {
      "epoch": 53.14768752429071,
      "grad_norm": 0.10512412339448929,
      "learning_rate": 6.926368421052632e-05,
      "loss": 0.0162,
      "step": 68400
    },
    {
      "epoch": 53.18655266226195,
      "grad_norm": 0.20373064279556274,
      "learning_rate": 6.923736842105265e-05,
      "loss": 0.0112,
      "step": 68450
    },
    {
      "epoch": 53.22541780023319,
      "grad_norm": 0.12958596646785736,
      "learning_rate": 6.921105263157896e-05,
      "loss": 0.0071,
      "step": 68500
    },
    {
      "epoch": 53.26428293820443,
      "grad_norm": 0.0892186313867569,
      "learning_rate": 6.918473684210527e-05,
      "loss": 0.0152,
      "step": 68550
    },
    {
      "epoch": 53.303148076175674,
      "grad_norm": 4.232198238372803,
      "learning_rate": 6.915842105263158e-05,
      "loss": 0.0093,
      "step": 68600
    },
    {
      "epoch": 53.34201321414691,
      "grad_norm": 4.7500176429748535,
      "learning_rate": 6.913210526315789e-05,
      "loss": 0.0151,
      "step": 68650
    },
    {
      "epoch": 53.38087835211815,
      "grad_norm": 0.430583655834198,
      "learning_rate": 6.910578947368422e-05,
      "loss": 0.01,
      "step": 68700
    },
    {
      "epoch": 53.41974349008939,
      "grad_norm": 0.4367411434650421,
      "learning_rate": 6.907947368421053e-05,
      "loss": 0.0089,
      "step": 68750
    },
    {
      "epoch": 53.45860862806063,
      "grad_norm": 0.05042579025030136,
      "learning_rate": 6.905315789473684e-05,
      "loss": 0.006,
      "step": 68800
    },
    {
      "epoch": 53.49747376603187,
      "grad_norm": 0.14255771040916443,
      "learning_rate": 6.902684210526315e-05,
      "loss": 0.0101,
      "step": 68850
    },
    {
      "epoch": 53.53633890400311,
      "grad_norm": 0.374448299407959,
      "learning_rate": 6.900052631578948e-05,
      "loss": 0.0086,
      "step": 68900
    },
    {
      "epoch": 53.57520404197435,
      "grad_norm": 0.10216441005468369,
      "learning_rate": 6.89742105263158e-05,
      "loss": 0.0087,
      "step": 68950
    },
    {
      "epoch": 53.61406917994559,
      "grad_norm": 0.03489489480853081,
      "learning_rate": 6.894789473684211e-05,
      "loss": 0.0088,
      "step": 69000
    },
    {
      "epoch": 53.61406917994559,
      "eval_accuracy": 0.9985424865180003,
      "eval_runtime": 849.4139,
      "eval_samples_per_second": 24.232,
      "eval_steps_per_second": 3.029,
      "step": 69000
    },
    {
      "epoch": 53.65293431791683,
      "grad_norm": 0.0035805790685117245,
      "learning_rate": 6.892157894736843e-05,
      "loss": 0.0086,
      "step": 69050
    },
    {
      "epoch": 53.69179945588807,
      "grad_norm": 0.15195728838443756,
      "learning_rate": 6.889526315789474e-05,
      "loss": 0.0118,
      "step": 69100
    },
    {
      "epoch": 53.73066459385931,
      "grad_norm": 0.5311455726623535,
      "learning_rate": 6.886894736842105e-05,
      "loss": 0.011,
      "step": 69150
    },
    {
      "epoch": 53.769529731830545,
      "grad_norm": 0.30500563979148865,
      "learning_rate": 6.884263157894737e-05,
      "loss": 0.0097,
      "step": 69200
    },
    {
      "epoch": 53.80839486980179,
      "grad_norm": 0.05197153985500336,
      "learning_rate": 6.881631578947369e-05,
      "loss": 0.0103,
      "step": 69250
    },
    {
      "epoch": 53.84726000777303,
      "grad_norm": 0.369314581155777,
      "learning_rate": 6.879e-05,
      "loss": 0.0073,
      "step": 69300
    },
    {
      "epoch": 53.886125145744266,
      "grad_norm": 0.755934476852417,
      "learning_rate": 6.876368421052632e-05,
      "loss": 0.0071,
      "step": 69350
    },
    {
      "epoch": 53.92499028371551,
      "grad_norm": 2.4341537952423096,
      "learning_rate": 6.873736842105263e-05,
      "loss": 0.0191,
      "step": 69400
    },
    {
      "epoch": 53.963855421686745,
      "grad_norm": 6.1135993003845215,
      "learning_rate": 6.871105263157896e-05,
      "loss": 0.0125,
      "step": 69450
    },
    {
      "epoch": 54.00233190827827,
      "grad_norm": 0.027943432331085205,
      "learning_rate": 6.868473684210527e-05,
      "loss": 0.0124,
      "step": 69500
    },
    {
      "epoch": 54.041197046249515,
      "grad_norm": 3.6769490242004395,
      "learning_rate": 6.865842105263158e-05,
      "loss": 0.0121,
      "step": 69550
    },
    {
      "epoch": 54.08006218422075,
      "grad_norm": 2.0445716381073,
      "learning_rate": 6.86321052631579e-05,
      "loss": 0.0129,
      "step": 69600
    },
    {
      "epoch": 54.11892732219199,
      "grad_norm": 0.0530659481883049,
      "learning_rate": 6.860578947368422e-05,
      "loss": 0.0051,
      "step": 69650
    },
    {
      "epoch": 54.157792460163236,
      "grad_norm": 0.038605764508247375,
      "learning_rate": 6.857947368421053e-05,
      "loss": 0.0053,
      "step": 69700
    },
    {
      "epoch": 54.19665759813447,
      "grad_norm": 0.5865646004676819,
      "learning_rate": 6.855315789473684e-05,
      "loss": 0.01,
      "step": 69750
    },
    {
      "epoch": 54.235522736105715,
      "grad_norm": 1.5437185764312744,
      "learning_rate": 6.852684210526315e-05,
      "loss": 0.0087,
      "step": 69800
    },
    {
      "epoch": 54.27438787407695,
      "grad_norm": 0.030433477833867073,
      "learning_rate": 6.850052631578948e-05,
      "loss": 0.0044,
      "step": 69850
    },
    {
      "epoch": 54.31325301204819,
      "grad_norm": 0.1999322772026062,
      "learning_rate": 6.847421052631579e-05,
      "loss": 0.0098,
      "step": 69900
    },
    {
      "epoch": 54.352118150019436,
      "grad_norm": 2.545680046081543,
      "learning_rate": 6.844789473684212e-05,
      "loss": 0.0032,
      "step": 69950
    },
    {
      "epoch": 54.39098328799067,
      "grad_norm": 0.26193609833717346,
      "learning_rate": 6.842157894736843e-05,
      "loss": 0.012,
      "step": 70000
    },
    {
      "epoch": 54.39098328799067,
      "eval_accuracy": 0.9987854054316669,
      "eval_runtime": 852.7537,
      "eval_samples_per_second": 24.137,
      "eval_steps_per_second": 3.017,
      "step": 70000
    },
    {
      "epoch": 54.429848425961914,
      "grad_norm": 0.2691155970096588,
      "learning_rate": 6.839526315789474e-05,
      "loss": 0.0089,
      "step": 70050
    },
    {
      "epoch": 54.46871356393315,
      "grad_norm": 0.05729250609874725,
      "learning_rate": 6.836894736842105e-05,
      "loss": 0.0108,
      "step": 70100
    },
    {
      "epoch": 54.50757870190439,
      "grad_norm": 5.909592151641846,
      "learning_rate": 6.834263157894738e-05,
      "loss": 0.0159,
      "step": 70150
    },
    {
      "epoch": 54.54644383987563,
      "grad_norm": 0.48799797892570496,
      "learning_rate": 6.831631578947369e-05,
      "loss": 0.0125,
      "step": 70200
    },
    {
      "epoch": 54.58530897784687,
      "grad_norm": 1.2644726037979126,
      "learning_rate": 6.829e-05,
      "loss": 0.0113,
      "step": 70250
    },
    {
      "epoch": 54.624174115818114,
      "grad_norm": 0.2227022796869278,
      "learning_rate": 6.826368421052632e-05,
      "loss": 0.0047,
      "step": 70300
    },
    {
      "epoch": 54.66303925378935,
      "grad_norm": 0.06142238900065422,
      "learning_rate": 6.823736842105264e-05,
      "loss": 0.0089,
      "step": 70350
    },
    {
      "epoch": 54.70190439176059,
      "grad_norm": 2.738752603530884,
      "learning_rate": 6.821105263157895e-05,
      "loss": 0.006,
      "step": 70400
    },
    {
      "epoch": 54.74076952973183,
      "grad_norm": 0.5507923364639282,
      "learning_rate": 6.818473684210527e-05,
      "loss": 0.0095,
      "step": 70450
    },
    {
      "epoch": 54.77963466770307,
      "grad_norm": 0.024553149938583374,
      "learning_rate": 6.815842105263158e-05,
      "loss": 0.0066,
      "step": 70500
    },
    {
      "epoch": 54.81849980567431,
      "grad_norm": 0.1172686219215393,
      "learning_rate": 6.81321052631579e-05,
      "loss": 0.0113,
      "step": 70550
    },
    {
      "epoch": 54.85736494364555,
      "grad_norm": 0.08226852118968964,
      "learning_rate": 6.81057894736842e-05,
      "loss": 0.0083,
      "step": 70600
    },
    {
      "epoch": 54.89623008161679,
      "grad_norm": 5.019248008728027,
      "learning_rate": 6.807947368421053e-05,
      "loss": 0.0108,
      "step": 70650
    },
    {
      "epoch": 54.93509521958803,
      "grad_norm": 0.14996369183063507,
      "learning_rate": 6.805315789473684e-05,
      "loss": 0.0095,
      "step": 70700
    },
    {
      "epoch": 54.97396035755927,
      "grad_norm": 1.8933607339859009,
      "learning_rate": 6.802684210526317e-05,
      "loss": 0.0114,
      "step": 70750
    },
    {
      "epoch": 55.0124368441508,
      "grad_norm": 0.010230555199086666,
      "learning_rate": 6.800052631578948e-05,
      "loss": 0.0132,
      "step": 70800
    },
    {
      "epoch": 55.051301982122034,
      "grad_norm": 0.17247895896434784,
      "learning_rate": 6.797421052631579e-05,
      "loss": 0.0078,
      "step": 70850
    },
    {
      "epoch": 55.09016712009328,
      "grad_norm": 1.4789326190948486,
      "learning_rate": 6.794789473684212e-05,
      "loss": 0.0222,
      "step": 70900
    },
    {
      "epoch": 55.12903225806452,
      "grad_norm": 0.0069031622260808945,
      "learning_rate": 6.792157894736843e-05,
      "loss": 0.0061,
      "step": 70950
    },
    {
      "epoch": 55.167897396035755,
      "grad_norm": 0.054508525878190994,
      "learning_rate": 6.789526315789474e-05,
      "loss": 0.0049,
      "step": 71000
    },
    {
      "epoch": 55.167897396035755,
      "eval_accuracy": 0.9989797405626002,
      "eval_runtime": 862.2551,
      "eval_samples_per_second": 23.871,
      "eval_steps_per_second": 2.984,
      "step": 71000
    },
    {
      "epoch": 55.206762534007,
      "grad_norm": 0.043283432722091675,
      "learning_rate": 6.786894736842105e-05,
      "loss": 0.0066,
      "step": 71050
    },
    {
      "epoch": 55.245627671978234,
      "grad_norm": 0.06320410966873169,
      "learning_rate": 6.784263157894736e-05,
      "loss": 0.0055,
      "step": 71100
    },
    {
      "epoch": 55.28449280994948,
      "grad_norm": 0.04028366133570671,
      "learning_rate": 6.781631578947369e-05,
      "loss": 0.0049,
      "step": 71150
    },
    {
      "epoch": 55.32335794792071,
      "grad_norm": 2.822596311569214,
      "learning_rate": 6.779e-05,
      "loss": 0.0088,
      "step": 71200
    },
    {
      "epoch": 55.362223085891955,
      "grad_norm": 0.5370486378669739,
      "learning_rate": 6.776368421052632e-05,
      "loss": 0.008,
      "step": 71250
    },
    {
      "epoch": 55.4010882238632,
      "grad_norm": 1.14641535282135,
      "learning_rate": 6.773736842105264e-05,
      "loss": 0.0067,
      "step": 71300
    },
    {
      "epoch": 55.439953361834434,
      "grad_norm": 0.1291676163673401,
      "learning_rate": 6.771105263157895e-05,
      "loss": 0.0091,
      "step": 71350
    },
    {
      "epoch": 55.478818499805676,
      "grad_norm": 0.24983321130275726,
      "learning_rate": 6.768473684210527e-05,
      "loss": 0.014,
      "step": 71400
    },
    {
      "epoch": 55.51768363777691,
      "grad_norm": 0.03976181894540787,
      "learning_rate": 6.765842105263158e-05,
      "loss": 0.0126,
      "step": 71450
    },
    {
      "epoch": 55.556548775748155,
      "grad_norm": 0.37395182251930237,
      "learning_rate": 6.76321052631579e-05,
      "loss": 0.0071,
      "step": 71500
    },
    {
      "epoch": 55.59541391371939,
      "grad_norm": 0.05620402470231056,
      "learning_rate": 6.760578947368421e-05,
      "loss": 0.0068,
      "step": 71550
    },
    {
      "epoch": 55.63427905169063,
      "grad_norm": 0.27877360582351685,
      "learning_rate": 6.757947368421052e-05,
      "loss": 0.0107,
      "step": 71600
    },
    {
      "epoch": 55.673144189661876,
      "grad_norm": 0.04781031608581543,
      "learning_rate": 6.755315789473684e-05,
      "loss": 0.0144,
      "step": 71650
    },
    {
      "epoch": 55.71200932763311,
      "grad_norm": 0.015767989680171013,
      "learning_rate": 6.752684210526317e-05,
      "loss": 0.0173,
      "step": 71700
    },
    {
      "epoch": 55.750874465604355,
      "grad_norm": 0.03923904523253441,
      "learning_rate": 6.750052631578948e-05,
      "loss": 0.0067,
      "step": 71750
    },
    {
      "epoch": 55.78973960357559,
      "grad_norm": 3.509338140487671,
      "learning_rate": 6.747421052631579e-05,
      "loss": 0.0052,
      "step": 71800
    },
    {
      "epoch": 55.82860474154683,
      "grad_norm": 0.012454860843718052,
      "learning_rate": 6.74478947368421e-05,
      "loss": 0.0123,
      "step": 71850
    },
    {
      "epoch": 55.86746987951807,
      "grad_norm": 0.1035495400428772,
      "learning_rate": 6.742157894736843e-05,
      "loss": 0.0089,
      "step": 71900
    },
    {
      "epoch": 55.90633501748931,
      "grad_norm": 0.031217306852340698,
      "learning_rate": 6.739526315789474e-05,
      "loss": 0.0214,
      "step": 71950
    },
    {
      "epoch": 55.945200155460554,
      "grad_norm": 0.12328901886940002,
      "learning_rate": 6.736894736842105e-05,
      "loss": 0.008,
      "step": 72000
    },
    {
      "epoch": 55.945200155460554,
      "eval_accuracy": 0.9991254919108001,
      "eval_runtime": 855.5732,
      "eval_samples_per_second": 24.058,
      "eval_steps_per_second": 3.007,
      "step": 72000
    }
  ],
  "logging_steps": 50,
  "max_steps": 200000,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 156,
  "save_steps": 1000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 6.39666699518976e+17,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}