{
  "best_metric": 0.958063608218407,
  "best_model_checkpoint": "melanoma-v4\\checkpoint-69000",
  "epoch": 5.0,
  "eval_steps": 1000,
  "global_step": 71615,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00349088878028346,
      "grad_norm": 44.0053596496582,
      "learning_rate": 3.1415805640882436e-07,
      "loss": 0.6936,
      "step": 50
    },
    {
      "epoch": 0.00698177756056692,
      "grad_norm": 52.222103118896484,
      "learning_rate": 6.632225635297403e-07,
      "loss": 0.6915,
      "step": 100
    },
    {
      "epoch": 0.010472666340850381,
      "grad_norm": 34.52608108520508,
      "learning_rate": 1.0053057805082381e-06,
      "loss": 0.6849,
      "step": 150
    },
    {
      "epoch": 0.01396355512113384,
      "grad_norm": 23.362762451171875,
      "learning_rate": 1.3543702876291538e-06,
      "loss": 0.6753,
      "step": 200
    },
    {
      "epoch": 0.017454443901417302,
      "grad_norm": 55.74956512451172,
      "learning_rate": 1.7034347947500698e-06,
      "loss": 0.6634,
      "step": 250
    },
    {
      "epoch": 0.020945332681700762,
      "grad_norm": 14.746822357177734,
      "learning_rate": 2.052499301870986e-06,
      "loss": 0.6475,
      "step": 300
    },
    {
      "epoch": 0.024436221461984223,
      "grad_norm": 22.278230667114258,
      "learning_rate": 2.401563808991902e-06,
      "loss": 0.6281,
      "step": 350
    },
    {
      "epoch": 0.02792711024226768,
      "grad_norm": 23.80645751953125,
      "learning_rate": 2.7506283161128178e-06,
      "loss": 0.6048,
      "step": 400
    },
    {
      "epoch": 0.031417999022551144,
      "grad_norm": 16.161684036254883,
      "learning_rate": 3.0996928232337337e-06,
      "loss": 0.5785,
      "step": 450
    },
    {
      "epoch": 0.034908887802834604,
      "grad_norm": 19.891162872314453,
      "learning_rate": 3.4487573303546497e-06,
      "loss": 0.5491,
      "step": 500
    },
    {
      "epoch": 0.038399776583118064,
      "grad_norm": 25.53693199157715,
      "learning_rate": 3.797821837475566e-06,
      "loss": 0.5138,
      "step": 550
    },
    {
      "epoch": 0.041890665363401525,
      "grad_norm": 36.38960647583008,
      "learning_rate": 4.146886344596482e-06,
      "loss": 0.482,
      "step": 600
    },
    {
      "epoch": 0.045381554143684985,
      "grad_norm": 29.26142120361328,
      "learning_rate": 4.4959508517173975e-06,
      "loss": 0.4375,
      "step": 650
    },
    {
      "epoch": 0.048872442923968445,
      "grad_norm": 55.49436950683594,
      "learning_rate": 4.845015358838314e-06,
      "loss": 0.4107,
      "step": 700
    },
    {
      "epoch": 0.052363331704251906,
      "grad_norm": 29.13718605041504,
      "learning_rate": 5.194079865959229e-06,
      "loss": 0.3595,
      "step": 750
    },
    {
      "epoch": 0.05585422048453536,
      "grad_norm": 27.340045928955078,
      "learning_rate": 5.543144373080146e-06,
      "loss": 0.313,
      "step": 800
    },
    {
      "epoch": 0.05934510926481882,
      "grad_norm": 18.835979461669922,
      "learning_rate": 5.892208880201061e-06,
      "loss": 0.2754,
      "step": 850
    },
    {
      "epoch": 0.06283599804510229,
      "grad_norm": 28.285400390625,
      "learning_rate": 6.2412733873219775e-06,
      "loss": 0.2423,
      "step": 900
    },
    {
      "epoch": 0.06632688682538575,
      "grad_norm": 14.74078369140625,
      "learning_rate": 6.590337894442894e-06,
      "loss": 0.1988,
      "step": 950
    },
    {
      "epoch": 0.06981777560566921,
      "grad_norm": 12.510848999023438,
      "learning_rate": 6.939402401563809e-06,
      "loss": 0.193,
      "step": 1000
    },
    {
      "epoch": 0.06981777560566921,
      "eval_f1": 0.0,
      "eval_loss": 1.0061070919036865,
      "eval_runtime": 16.8276,
      "eval_samples_per_second": 217.5,
      "eval_steps_per_second": 6.834,
      "step": 1000
    },
    {
      "epoch": 0.07330866438595267,
      "grad_norm": 24.07497215270996,
      "learning_rate": 7.288466908684726e-06,
      "loss": 0.1646,
      "step": 1050
    },
    {
      "epoch": 0.07679955316623613,
      "grad_norm": 34.75226593017578,
      "learning_rate": 7.637531415805642e-06,
      "loss": 0.1535,
      "step": 1100
    },
    {
      "epoch": 0.08029044194651959,
      "grad_norm": 8.6325101852417,
      "learning_rate": 7.986595922926558e-06,
      "loss": 0.1338,
      "step": 1150
    },
    {
      "epoch": 0.08378133072680305,
      "grad_norm": 7.621668338775635,
      "learning_rate": 8.335660430047473e-06,
      "loss": 0.1306,
      "step": 1200
    },
    {
      "epoch": 0.08727221950708651,
      "grad_norm": 12.712471008300781,
      "learning_rate": 8.68472493716839e-06,
      "loss": 0.1297,
      "step": 1250
    },
    {
      "epoch": 0.09076310828736997,
      "grad_norm": 24.750577926635742,
      "learning_rate": 9.033789444289304e-06,
      "loss": 0.1106,
      "step": 1300
    },
    {
      "epoch": 0.09425399706765343,
      "grad_norm": 6.89915657043457,
      "learning_rate": 9.382853951410221e-06,
      "loss": 0.087,
      "step": 1350
    },
    {
      "epoch": 0.09774488584793689,
      "grad_norm": 31.87893295288086,
      "learning_rate": 9.731918458531137e-06,
      "loss": 0.1054,
      "step": 1400
    },
    {
      "epoch": 0.10123577462822035,
      "grad_norm": 18.434345245361328,
      "learning_rate": 1.0080982965652052e-05,
      "loss": 0.1052,
      "step": 1450
    },
    {
      "epoch": 0.10472666340850381,
      "grad_norm": 2.470731258392334,
      "learning_rate": 1.043004747277297e-05,
      "loss": 0.0819,
      "step": 1500
    },
    {
      "epoch": 0.10821755218878727,
      "grad_norm": 4.6369853019714355,
      "learning_rate": 1.0779111979893885e-05,
      "loss": 0.0817,
      "step": 1550
    },
    {
      "epoch": 0.11170844096907072,
      "grad_norm": 59.104862213134766,
      "learning_rate": 1.11281764870148e-05,
      "loss": 0.0819,
      "step": 1600
    },
    {
      "epoch": 0.11519932974935418,
      "grad_norm": 46.61822509765625,
      "learning_rate": 1.1477240994135718e-05,
      "loss": 0.0556,
      "step": 1650
    },
    {
      "epoch": 0.11869021852963764,
      "grad_norm": 46.31196212768555,
      "learning_rate": 1.1826305501256633e-05,
      "loss": 0.0828,
      "step": 1700
    },
    {
      "epoch": 0.1221811073099211,
      "grad_norm": 3.6305160522460938,
      "learning_rate": 1.2175370008377549e-05,
      "loss": 0.0857,
      "step": 1750
    },
    {
      "epoch": 0.12567199609020457,
      "grad_norm": 6.165530204772949,
      "learning_rate": 1.2524434515498464e-05,
      "loss": 0.0811,
      "step": 1800
    },
    {
      "epoch": 0.12916288487048802,
      "grad_norm": 2.393606185913086,
      "learning_rate": 1.2873499022619382e-05,
      "loss": 0.0601,
      "step": 1850
    },
    {
      "epoch": 0.1326537736507715,
      "grad_norm": 9.886051177978516,
      "learning_rate": 1.3222563529740297e-05,
      "loss": 0.0757,
      "step": 1900
    },
    {
      "epoch": 0.13614466243105494,
      "grad_norm": 9.62782096862793,
      "learning_rate": 1.3571628036861211e-05,
      "loss": 0.0798,
      "step": 1950
    },
    {
      "epoch": 0.13963555121133842,
      "grad_norm": 4.66179895401001,
      "learning_rate": 1.3920692543982128e-05,
      "loss": 0.0585,
      "step": 2000
    },
    {
      "epoch": 0.13963555121133842,
      "eval_f1": 0.4475823705605477,
      "eval_loss": 1.2610578536987305,
      "eval_runtime": 16.5432,
      "eval_samples_per_second": 221.239,
      "eval_steps_per_second": 6.951,
      "step": 2000
    },
    {
      "epoch": 0.14312643999162186,
      "grad_norm": 7.12888240814209,
      "learning_rate": 1.4269757051103044e-05,
      "loss": 0.0682,
      "step": 2050
    },
    {
      "epoch": 0.14661732877190534,
      "grad_norm": 35.82215881347656,
      "learning_rate": 1.461882155822396e-05,
      "loss": 0.0577,
      "step": 2100
    },
    {
      "epoch": 0.15010821755218878,
      "grad_norm": 6.529672145843506,
      "learning_rate": 1.4960904775202458e-05,
      "loss": 0.0922,
      "step": 2150
    },
    {
      "epoch": 0.15359910633247226,
      "grad_norm": 8.664837837219238,
      "learning_rate": 1.5309969282323375e-05,
      "loss": 0.0574,
      "step": 2200
    },
    {
      "epoch": 0.1570899951127557,
      "grad_norm": 2.3119335174560547,
      "learning_rate": 1.5659033789444292e-05,
      "loss": 0.0593,
      "step": 2250
    },
    {
      "epoch": 0.16058088389303918,
      "grad_norm": 1.3561939001083374,
      "learning_rate": 1.6008098296565206e-05,
      "loss": 0.066,
      "step": 2300
    },
    {
      "epoch": 0.16407177267332262,
      "grad_norm": 38.09425735473633,
      "learning_rate": 1.635716280368612e-05,
      "loss": 0.0757,
      "step": 2350
    },
    {
      "epoch": 0.1675626614536061,
      "grad_norm": 61.25293731689453,
      "learning_rate": 1.6706227310807037e-05,
      "loss": 0.0778,
      "step": 2400
    },
    {
      "epoch": 0.17105355023388955,
      "grad_norm": 39.29595947265625,
      "learning_rate": 1.7055291817927955e-05,
      "loss": 0.0775,
      "step": 2450
    },
    {
      "epoch": 0.17454443901417302,
      "grad_norm": 35.1599235534668,
      "learning_rate": 1.7404356325048872e-05,
      "loss": 0.0775,
      "step": 2500
    },
    {
      "epoch": 0.17803532779445647,
      "grad_norm": 0.5723768472671509,
      "learning_rate": 1.7753420832169786e-05,
      "loss": 0.0658,
      "step": 2550
    },
    {
      "epoch": 0.18152621657473994,
      "grad_norm": 1.6450070142745972,
      "learning_rate": 1.81024853392907e-05,
      "loss": 0.0461,
      "step": 2600
    },
    {
      "epoch": 0.1850171053550234,
      "grad_norm": 0.417278528213501,
      "learning_rate": 1.8451549846411617e-05,
      "loss": 0.0648,
      "step": 2650
    },
    {
      "epoch": 0.18850799413530686,
      "grad_norm": 14.564908981323242,
      "learning_rate": 1.8800614353532534e-05,
      "loss": 0.0573,
      "step": 2700
    },
    {
      "epoch": 0.1919988829155903,
      "grad_norm": 0.4077770411968231,
      "learning_rate": 1.914967886065345e-05,
      "loss": 0.0695,
      "step": 2750
    },
    {
      "epoch": 0.19548977169587378,
      "grad_norm": 0.27182820439338684,
      "learning_rate": 1.9498743367774365e-05,
      "loss": 0.043,
      "step": 2800
    },
    {
      "epoch": 0.19898066047615723,
      "grad_norm": 3.3724520206451416,
      "learning_rate": 1.9847807874895282e-05,
      "loss": 0.0765,
      "step": 2850
    },
    {
      "epoch": 0.2024715492564407,
      "grad_norm": 27.066162109375,
      "learning_rate": 2.0196872382016196e-05,
      "loss": 0.047,
      "step": 2900
    },
    {
      "epoch": 0.20596243803672415,
      "grad_norm": 0.23048856854438782,
      "learning_rate": 2.0545936889137113e-05,
      "loss": 0.0589,
      "step": 2950
    },
    {
      "epoch": 0.20945332681700762,
      "grad_norm": 0.6853739619255066,
      "learning_rate": 2.089500139625803e-05,
      "loss": 0.0723,
      "step": 3000
    },
    {
      "epoch": 0.20945332681700762,
      "eval_f1": 0.8094940662086196,
      "eval_loss": 0.7442474961280823,
      "eval_runtime": 17.1452,
      "eval_samples_per_second": 213.471,
      "eval_steps_per_second": 6.707,
      "step": 3000
    },
    {
      "epoch": 0.21294421559729107,
      "grad_norm": 71.01643371582031,
      "learning_rate": 2.1244065903378947e-05,
      "loss": 0.0696,
      "step": 3050
    },
    {
      "epoch": 0.21643510437757454,
      "grad_norm": 54.42838668823242,
      "learning_rate": 2.159313041049986e-05,
      "loss": 0.0443,
      "step": 3100
    },
    {
      "epoch": 0.219925993157858,
      "grad_norm": 0.15738917887210846,
      "learning_rate": 2.194219491762078e-05,
      "loss": 0.0569,
      "step": 3150
    },
    {
      "epoch": 0.22341688193814144,
      "grad_norm": 0.6543312668800354,
      "learning_rate": 2.2291259424741692e-05,
      "loss": 0.0637,
      "step": 3200
    },
    {
      "epoch": 0.2269077707184249,
      "grad_norm": 0.44767776131629944,
      "learning_rate": 2.264032393186261e-05,
      "loss": 0.0498,
      "step": 3250
    },
    {
      "epoch": 0.23039865949870836,
      "grad_norm": 0.09834372252225876,
      "learning_rate": 2.2989388438983527e-05,
      "loss": 0.0344,
      "step": 3300
    },
    {
      "epoch": 0.23388954827899183,
      "grad_norm": 83.13866424560547,
      "learning_rate": 2.333845294610444e-05,
      "loss": 0.05,
      "step": 3350
    },
    {
      "epoch": 0.23738043705927528,
      "grad_norm": 14.352801322937012,
      "learning_rate": 2.3687517453225358e-05,
      "loss": 0.0536,
      "step": 3400
    },
    {
      "epoch": 0.24087132583955875,
      "grad_norm": 1.1875745058059692,
      "learning_rate": 2.4036581960346275e-05,
      "loss": 0.0617,
      "step": 3450
    },
    {
      "epoch": 0.2443622146198422,
      "grad_norm": 6.531946659088135,
      "learning_rate": 2.438564646746719e-05,
      "loss": 0.0636,
      "step": 3500
    },
    {
      "epoch": 0.24785310340012567,
      "grad_norm": 0.0965203046798706,
      "learning_rate": 2.4734710974588106e-05,
      "loss": 0.0423,
      "step": 3550
    },
    {
      "epoch": 0.25134399218040915,
      "grad_norm": 0.07983570545911789,
      "learning_rate": 2.508377548170902e-05,
      "loss": 0.036,
      "step": 3600
    },
    {
      "epoch": 0.25483488096069257,
      "grad_norm": 0.05020700767636299,
      "learning_rate": 2.5432839988829937e-05,
      "loss": 0.0448,
      "step": 3650
    },
    {
      "epoch": 0.25832576974097604,
      "grad_norm": 23.652816772460938,
      "learning_rate": 2.5781904495950854e-05,
      "loss": 0.0709,
      "step": 3700
    },
    {
      "epoch": 0.2618166585212595,
      "grad_norm": 0.33149588108062744,
      "learning_rate": 2.613096900307177e-05,
      "loss": 0.0426,
      "step": 3750
    },
    {
      "epoch": 0.265307547301543,
      "grad_norm": 57.82050704956055,
      "learning_rate": 2.6480033510192685e-05,
      "loss": 0.0535,
      "step": 3800
    },
    {
      "epoch": 0.2687984360818264,
      "grad_norm": 0.3976256847381592,
      "learning_rate": 2.6829098017313602e-05,
      "loss": 0.0591,
      "step": 3850
    },
    {
      "epoch": 0.2722893248621099,
      "grad_norm": 3.0774986743927,
      "learning_rate": 2.717816252443452e-05,
      "loss": 0.0312,
      "step": 3900
    },
    {
      "epoch": 0.27578021364239336,
      "grad_norm": 91.02516174316406,
      "learning_rate": 2.752722703155543e-05,
      "loss": 0.0541,
      "step": 3950
    },
    {
      "epoch": 0.27927110242267683,
      "grad_norm": 0.111998550593853,
      "learning_rate": 2.7876291538676347e-05,
      "loss": 0.0442,
      "step": 4000
    },
    {
      "epoch": 0.27927110242267683,
      "eval_f1": 0.8585769117212873,
      "eval_loss": 0.41572797298431396,
      "eval_runtime": 23.9384,
      "eval_samples_per_second": 152.893,
      "eval_steps_per_second": 4.804,
      "step": 4000
    },
    {
      "epoch": 0.28276199120296025,
      "grad_norm": 0.1660720556974411,
      "learning_rate": 2.8225356045797264e-05,
      "loss": 0.0418,
      "step": 4050
    },
    {
      "epoch": 0.2862528799832437,
      "grad_norm": 2.238628387451172,
      "learning_rate": 2.8574420552918178e-05,
      "loss": 0.0578,
      "step": 4100
    },
    {
      "epoch": 0.2897437687635272,
      "grad_norm": 34.74903106689453,
      "learning_rate": 2.8923485060039095e-05,
      "loss": 0.0474,
      "step": 4150
    },
    {
      "epoch": 0.2932346575438107,
      "grad_norm": 0.0780772715806961,
      "learning_rate": 2.9272549567160013e-05,
      "loss": 0.0311,
      "step": 4200
    },
    {
      "epoch": 0.2967255463240941,
      "grad_norm": 112.21017456054688,
      "learning_rate": 2.962161407428093e-05,
      "loss": 0.0683,
      "step": 4250
    },
    {
      "epoch": 0.30021643510437757,
      "grad_norm": 92.58731842041016,
      "learning_rate": 2.9970678581401847e-05,
      "loss": 0.0481,
      "step": 4300
    },
    {
      "epoch": 0.30370732388466104,
      "grad_norm": 0.12998327612876892,
      "learning_rate": 3.031974308852276e-05,
      "loss": 0.0473,
      "step": 4350
    },
    {
      "epoch": 0.3071982126649445,
      "grad_norm": 26.06481170654297,
      "learning_rate": 3.066880759564368e-05,
      "loss": 0.0474,
      "step": 4400
    },
    {
      "epoch": 0.31068910144522793,
      "grad_norm": 18.633922576904297,
      "learning_rate": 3.1017872102764595e-05,
      "loss": 0.064,
      "step": 4450
    },
    {
      "epoch": 0.3141799902255114,
      "grad_norm": 0.08274181932210922,
      "learning_rate": 3.136693660988551e-05,
      "loss": 0.0502,
      "step": 4500
    },
    {
      "epoch": 0.3176708790057949,
      "grad_norm": 15.958111763000488,
      "learning_rate": 3.171600111700642e-05,
      "loss": 0.0384,
      "step": 4550
    },
    {
      "epoch": 0.32116176778607836,
      "grad_norm": 0.12025760114192963,
      "learning_rate": 3.206506562412734e-05,
      "loss": 0.0612,
      "step": 4600
    },
    {
      "epoch": 0.3246526565663618,
      "grad_norm": 1.3214194774627686,
      "learning_rate": 3.241413013124826e-05,
      "loss": 0.0438,
      "step": 4650
    },
    {
      "epoch": 0.32814354534664525,
      "grad_norm": 19.262493133544922,
      "learning_rate": 3.2763194638369174e-05,
      "loss": 0.0352,
      "step": 4700
    },
    {
      "epoch": 0.3316344341269287,
      "grad_norm": 0.8211079239845276,
      "learning_rate": 3.3112259145490085e-05,
      "loss": 0.0594,
      "step": 4750
    },
    {
      "epoch": 0.3351253229072122,
      "grad_norm": 7.5860419273376465,
      "learning_rate": 3.3461323652611e-05,
      "loss": 0.0508,
      "step": 4800
    },
    {
      "epoch": 0.3386162116874956,
      "grad_norm": 0.6498265266418457,
      "learning_rate": 3.381038815973192e-05,
      "loss": 0.0394,
      "step": 4850
    },
    {
      "epoch": 0.3421071004677791,
      "grad_norm": 16.379024505615234,
      "learning_rate": 3.4159452666852836e-05,
      "loss": 0.0495,
      "step": 4900
    },
    {
      "epoch": 0.34559798924806256,
      "grad_norm": 16.531225204467773,
      "learning_rate": 3.4508517173973754e-05,
      "loss": 0.0447,
      "step": 4950
    },
    {
      "epoch": 0.34908887802834604,
      "grad_norm": 0.04507113993167877,
      "learning_rate": 3.485758168109467e-05,
      "loss": 0.0462,
      "step": 5000
    },
    {
      "epoch": 0.34908887802834604,
      "eval_f1": 0.7487179487179487,
      "eval_loss": 0.7435426712036133,
      "eval_runtime": 24.3491,
      "eval_samples_per_second": 150.313,
      "eval_steps_per_second": 4.723,
      "step": 5000
    },
    {
      "epoch": 0.35257976680862946,
      "grad_norm": 2.0328619480133057,
      "learning_rate": 3.520664618821559e-05,
      "loss": 0.0417,
      "step": 5050
    },
    {
      "epoch": 0.35607065558891293,
      "grad_norm": 2.6427454948425293,
      "learning_rate": 3.55557106953365e-05,
      "loss": 0.0445,
      "step": 5100
    },
    {
      "epoch": 0.3595615443691964,
      "grad_norm": 1.2316179275512695,
      "learning_rate": 3.5904775202457416e-05,
      "loss": 0.0536,
      "step": 5150
    },
    {
      "epoch": 0.3630524331494799,
      "grad_norm": 6.056459426879883,
      "learning_rate": 3.625383970957833e-05,
      "loss": 0.0532,
      "step": 5200
    },
    {
      "epoch": 0.3665433219297633,
      "grad_norm": 0.5685946345329285,
      "learning_rate": 3.660290421669924e-05,
      "loss": 0.0296,
      "step": 5250
    },
    {
      "epoch": 0.3700342107100468,
      "grad_norm": 0.04925481602549553,
      "learning_rate": 3.695196872382016e-05,
      "loss": 0.0366,
      "step": 5300
    },
    {
      "epoch": 0.37352509949033025,
      "grad_norm": 2.920793294906616,
      "learning_rate": 3.730103323094108e-05,
      "loss": 0.0384,
      "step": 5350
    },
    {
      "epoch": 0.3770159882706137,
      "grad_norm": 0.09377721697092056,
      "learning_rate": 3.7650097738061995e-05,
      "loss": 0.0322,
      "step": 5400
    },
    {
      "epoch": 0.38050687705089714,
      "grad_norm": 0.10469520837068558,
      "learning_rate": 3.799916224518291e-05,
      "loss": 0.0327,
      "step": 5450
    },
    {
      "epoch": 0.3839977658311806,
      "grad_norm": 8.783834457397461,
      "learning_rate": 3.834822675230383e-05,
      "loss": 0.0356,
      "step": 5500
    },
    {
      "epoch": 0.3874886546114641,
      "grad_norm": 0.16647844016551971,
      "learning_rate": 3.8697291259424747e-05,
      "loss": 0.0437,
      "step": 5550
    },
    {
      "epoch": 0.39097954339174756,
      "grad_norm": 1.108203411102295,
      "learning_rate": 3.9046355766545664e-05,
      "loss": 0.0422,
      "step": 5600
    },
    {
      "epoch": 0.394470432172031,
      "grad_norm": 0.06744211167097092,
      "learning_rate": 3.939542027366658e-05,
      "loss": 0.0345,
      "step": 5650
    },
    {
      "epoch": 0.39796132095231446,
      "grad_norm": 6.879347801208496,
      "learning_rate": 3.974448478078749e-05,
      "loss": 0.0625,
      "step": 5700
    },
    {
      "epoch": 0.40145220973259793,
      "grad_norm": 0.05510378256440163,
      "learning_rate": 4.00935492879084e-05,
      "loss": 0.0279,
      "step": 5750
    },
    {
      "epoch": 0.4049430985128814,
      "grad_norm": 0.2473091334104538,
      "learning_rate": 4.044261379502932e-05,
      "loss": 0.0381,
      "step": 5800
    },
    {
      "epoch": 0.4084339872931648,
      "grad_norm": 0.3688828945159912,
      "learning_rate": 4.0791678302150236e-05,
      "loss": 0.0428,
      "step": 5850
    },
    {
      "epoch": 0.4119248760734483,
      "grad_norm": 40.18030548095703,
      "learning_rate": 4.1140742809271153e-05,
      "loss": 0.0411,
      "step": 5900
    },
    {
      "epoch": 0.4154157648537318,
      "grad_norm": 0.1600964218378067,
      "learning_rate": 4.148980731639207e-05,
      "loss": 0.0373,
      "step": 5950
    },
    {
      "epoch": 0.41890665363401525,
      "grad_norm": 0.1175314262509346,
      "learning_rate": 4.183887182351299e-05,
      "loss": 0.041,
      "step": 6000
    },
    {
      "epoch": 0.41890665363401525,
      "eval_f1": 0.7534340659340659,
      "eval_loss": 0.6432412266731262,
      "eval_runtime": 23.6996,
      "eval_samples_per_second": 154.433,
      "eval_steps_per_second": 4.852,
      "step": 6000
    },
    {
      "epoch": 0.42239754241429867,
      "grad_norm": 1.2376866340637207,
      "learning_rate": 4.2187936330633905e-05,
      "loss": 0.0336,
      "step": 6050
    },
    {
      "epoch": 0.42588843119458214,
      "grad_norm": 0.07168062031269073,
      "learning_rate": 4.253700083775482e-05,
      "loss": 0.0167,
      "step": 6100
    },
    {
      "epoch": 0.4293793199748656,
      "grad_norm": 0.06772568076848984,
      "learning_rate": 4.288606534487574e-05,
      "loss": 0.0614,
      "step": 6150
    },
    {
      "epoch": 0.4328702087551491,
      "grad_norm": 0.06727208942174911,
      "learning_rate": 4.323512985199665e-05,
      "loss": 0.0464,
      "step": 6200
    },
    {
      "epoch": 0.4363610975354325,
      "grad_norm": 0.1292935311794281,
      "learning_rate": 4.358419435911757e-05,
      "loss": 0.0546,
      "step": 6250
    },
    {
      "epoch": 0.439851986315716,
      "grad_norm": 9.874855995178223,
      "learning_rate": 4.393325886623848e-05,
      "loss": 0.06,
      "step": 6300
    },
    {
      "epoch": 0.44334287509599946,
      "grad_norm": 7.21466588973999,
      "learning_rate": 4.4282323373359395e-05,
      "loss": 0.0611,
      "step": 6350
    },
    {
      "epoch": 0.4468337638762829,
      "grad_norm": 16.634078979492188,
      "learning_rate": 4.463138788048031e-05,
      "loss": 0.0524,
      "step": 6400
    },
    {
      "epoch": 0.45032465265656635,
      "grad_norm": 0.3858453929424286,
      "learning_rate": 4.498045238760123e-05,
      "loss": 0.0444,
      "step": 6450
    },
    {
      "epoch": 0.4538155414368498,
      "grad_norm": 0.7371519804000854,
      "learning_rate": 4.532253560457973e-05,
      "loss": 0.0587,
      "step": 6500
    },
    {
      "epoch": 0.4573064302171333,
      "grad_norm": 0.24755489826202393,
      "learning_rate": 4.5671600111700645e-05,
      "loss": 0.055,
      "step": 6550
    },
    {
      "epoch": 0.4607973189974167,
      "grad_norm": 0.055525269359350204,
      "learning_rate": 4.602066461882156e-05,
      "loss": 0.0381,
      "step": 6600
    },
    {
      "epoch": 0.4642882077777002,
      "grad_norm": 0.21897675096988678,
      "learning_rate": 4.636972912594247e-05,
      "loss": 0.042,
      "step": 6650
    },
    {
      "epoch": 0.46777909655798366,
      "grad_norm": 0.34035515785217285,
      "learning_rate": 4.671879363306339e-05,
      "loss": 0.0369,
      "step": 6700
    },
    {
      "epoch": 0.47126998533826714,
      "grad_norm": 0.18382132053375244,
      "learning_rate": 4.706785814018431e-05,
      "loss": 0.0569,
      "step": 6750
    },
    {
      "epoch": 0.47476087411855056,
      "grad_norm": 0.040306150913238525,
      "learning_rate": 4.7416922647305225e-05,
      "loss": 0.0353,
      "step": 6800
    },
    {
      "epoch": 0.47825176289883403,
      "grad_norm": 6.798317909240723,
      "learning_rate": 4.776598715442614e-05,
      "loss": 0.0672,
      "step": 6850
    },
    {
      "epoch": 0.4817426516791175,
      "grad_norm": 0.12166763097047806,
      "learning_rate": 4.811505166154706e-05,
      "loss": 0.0291,
      "step": 6900
    },
    {
      "epoch": 0.485233540459401,
      "grad_norm": 0.11069754511117935,
      "learning_rate": 4.846411616866797e-05,
      "loss": 0.0451,
      "step": 6950
    },
    {
      "epoch": 0.4887244292396844,
      "grad_norm": 1.7939095497131348,
      "learning_rate": 4.881318067578889e-05,
      "loss": 0.0449,
      "step": 7000
    },
    {
      "epoch": 0.4887244292396844,
      "eval_f1": 0.8512851897184822,
      "eval_loss": 0.42608362436294556,
      "eval_runtime": 24.2404,
      "eval_samples_per_second": 150.987,
      "eval_steps_per_second": 4.744,
      "step": 7000
    },
    {
      "epoch": 0.4922153180199679,
      "grad_norm": 11.614912033081055,
      "learning_rate": 4.9162245182909804e-05,
      "loss": 0.0316,
      "step": 7050
    },
    {
      "epoch": 0.49570620680025135,
      "grad_norm": 0.2757989764213562,
      "learning_rate": 4.951130969003072e-05,
      "loss": 0.0491,
      "step": 7100
    },
    {
      "epoch": 0.4991970955805348,
      "grad_norm": 0.044196244329214096,
      "learning_rate": 4.986037419715164e-05,
      "loss": 0.0377,
      "step": 7150
    },
    {
      "epoch": 0.5026879843608183,
      "grad_norm": 1.8137636184692383,
      "learning_rate": 4.99767272275922e-05,
      "loss": 0.0446,
      "step": 7200
    },
    {
      "epoch": 0.5061788731411018,
      "grad_norm": 0.11980023980140686,
      "learning_rate": 4.9937939273579196e-05,
      "loss": 0.0507,
      "step": 7250
    },
    {
      "epoch": 0.5096697619213851,
      "grad_norm": 13.373326301574707,
      "learning_rate": 4.9899151319566196e-05,
      "loss": 0.051,
      "step": 7300
    },
    {
      "epoch": 0.5131606507016686,
      "grad_norm": 0.118846096098423,
      "learning_rate": 4.9860363365553196e-05,
      "loss": 0.0312,
      "step": 7350
    },
    {
      "epoch": 0.5166515394819521,
      "grad_norm": 0.17438098788261414,
      "learning_rate": 4.9821575411540196e-05,
      "loss": 0.0352,
      "step": 7400
    },
    {
      "epoch": 0.5201424282622356,
      "grad_norm": 0.04345332458615303,
      "learning_rate": 4.978278745752719e-05,
      "loss": 0.0423,
      "step": 7450
    },
    {
      "epoch": 0.523633317042519,
      "grad_norm": 0.9469249844551086,
      "learning_rate": 4.9743999503514196e-05,
      "loss": 0.0336,
      "step": 7500
    },
    {
      "epoch": 0.5271242058228025,
      "grad_norm": 0.14183463156223297,
      "learning_rate": 4.970521154950119e-05,
      "loss": 0.043,
      "step": 7550
    },
    {
      "epoch": 0.530615094603086,
      "grad_norm": 15.920584678649902,
      "learning_rate": 4.966642359548819e-05,
      "loss": 0.0447,
      "step": 7600
    },
    {
      "epoch": 0.5341059833833695,
      "grad_norm": 0.06846634298563004,
      "learning_rate": 4.962763564147518e-05,
      "loss": 0.0323,
      "step": 7650
    },
    {
      "epoch": 0.5375968721636528,
      "grad_norm": 0.02816701866686344,
      "learning_rate": 4.958884768746219e-05,
      "loss": 0.0547,
      "step": 7700
    },
    {
      "epoch": 0.5410877609439363,
      "grad_norm": 6.20529842376709,
      "learning_rate": 4.955005973344918e-05,
      "loss": 0.0415,
      "step": 7750
    },
    {
      "epoch": 0.5445786497242198,
      "grad_norm": 0.5864895582199097,
      "learning_rate": 4.951127177943618e-05,
      "loss": 0.0339,
      "step": 7800
    },
    {
      "epoch": 0.5480695385045032,
      "grad_norm": 0.13454869389533997,
      "learning_rate": 4.9472483825423176e-05,
      "loss": 0.0349,
      "step": 7850
    },
    {
      "epoch": 0.5515604272847867,
      "grad_norm": 0.06532914191484451,
      "learning_rate": 4.9433695871410176e-05,
      "loss": 0.0412,
      "step": 7900
    },
    {
      "epoch": 0.5550513160650702,
      "grad_norm": 0.11592119932174683,
      "learning_rate": 4.9394907917397176e-05,
      "loss": 0.0285,
      "step": 7950
    },
    {
      "epoch": 0.5585422048453537,
      "grad_norm": 0.1591779738664627,
      "learning_rate": 4.935611996338417e-05,
      "loss": 0.0537,
      "step": 8000
    },
    {
      "epoch": 0.5585422048453537,
      "eval_f1": 0.7704807041299933,
      "eval_loss": 0.5664303302764893,
      "eval_runtime": 23.2978,
      "eval_samples_per_second": 157.096,
      "eval_steps_per_second": 4.936,
      "step": 8000
    },
    {
      "epoch": 0.5620330936256371,
      "grad_norm": 0.6088069081306458,
      "learning_rate": 4.931733200937117e-05,
      "loss": 0.0362,
      "step": 8050
    },
    {
      "epoch": 0.5655239824059205,
      "grad_norm": 0.14050616323947906,
      "learning_rate": 4.927854405535817e-05,
      "loss": 0.0383,
      "step": 8100
    },
    {
      "epoch": 0.569014871186204,
      "grad_norm": 0.5788790583610535,
      "learning_rate": 4.923975610134517e-05,
      "loss": 0.0283,
      "step": 8150
    },
    {
      "epoch": 0.5725057599664874,
      "grad_norm": 0.0380774587392807,
      "learning_rate": 4.920096814733216e-05,
      "loss": 0.0211,
      "step": 8200
    },
    {
      "epoch": 0.5759966487467709,
      "grad_norm": 2.891254186630249,
      "learning_rate": 4.916218019331916e-05,
      "loss": 0.0512,
      "step": 8250
    },
    {
      "epoch": 0.5794875375270544,
      "grad_norm": 0.1362515538930893,
      "learning_rate": 4.9123392239306163e-05,
      "loss": 0.0363,
      "step": 8300
    },
    {
      "epoch": 0.5829784263073379,
      "grad_norm": 0.1054207980632782,
      "learning_rate": 4.9084604285293163e-05,
      "loss": 0.0278,
      "step": 8350
    },
    {
      "epoch": 0.5864693150876213,
      "grad_norm": 0.07413887977600098,
      "learning_rate": 4.904581633128016e-05,
      "loss": 0.0308,
      "step": 8400
    },
    {
      "epoch": 0.5899602038679048,
      "grad_norm": 0.11982708424329758,
      "learning_rate": 4.900702837726716e-05,
      "loss": 0.0397,
      "step": 8450
    },
    {
      "epoch": 0.5934510926481882,
      "grad_norm": 8.392777442932129,
      "learning_rate": 4.896824042325416e-05,
      "loss": 0.0469,
      "step": 8500
    },
    {
      "epoch": 0.5969419814284717,
      "grad_norm": 1.516817569732666,
      "learning_rate": 4.892945246924116e-05,
      "loss": 0.0384,
      "step": 8550
    },
    {
      "epoch": 0.6004328702087551,
      "grad_norm": 1.3836275339126587,
      "learning_rate": 4.889066451522815e-05,
      "loss": 0.0451,
      "step": 8600
    },
    {
      "epoch": 0.6039237589890386,
      "grad_norm": 15.225459098815918,
      "learning_rate": 4.885187656121516e-05,
      "loss": 0.0364,
      "step": 8650
    },
    {
      "epoch": 0.6074146477693221,
      "grad_norm": 4.840157985687256,
      "learning_rate": 4.881308860720215e-05,
      "loss": 0.0309,
      "step": 8700
    },
    {
      "epoch": 0.6109055365496056,
      "grad_norm": 0.05251197889447212,
      "learning_rate": 4.877430065318915e-05,
      "loss": 0.036,
      "step": 8750
    },
    {
      "epoch": 0.614396425329889,
      "grad_norm": 0.1797259896993637,
      "learning_rate": 4.8735512699176144e-05,
      "loss": 0.029,
      "step": 8800
    },
    {
      "epoch": 0.6178873141101725,
      "grad_norm": 0.033766426146030426,
      "learning_rate": 4.8696724745163144e-05,
      "loss": 0.0344,
      "step": 8850
    },
    {
      "epoch": 0.6213782028904559,
      "grad_norm": 0.5324440598487854,
      "learning_rate": 4.8657936791150144e-05,
      "loss": 0.0452,
      "step": 8900
    },
    {
      "epoch": 0.6248690916707393,
      "grad_norm": 7.406557083129883,
      "learning_rate": 4.861914883713714e-05,
      "loss": 0.0359,
      "step": 8950
    },
    {
      "epoch": 0.6283599804510228,
      "grad_norm": 0.05340263247489929,
      "learning_rate": 4.858036088312414e-05,
      "loss": 0.0222,
      "step": 9000
    },
    {
      "epoch": 0.6283599804510228,
      "eval_f1": 0.8430634023854363,
      "eval_loss": 0.4622673988342285,
      "eval_runtime": 24.6998,
      "eval_samples_per_second": 148.179,
      "eval_steps_per_second": 4.656,
      "step": 9000
    },
    {
      "epoch": 0.6318508692313063,
      "grad_norm": 5.009503364562988,
      "learning_rate": 4.854157292911114e-05,
      "loss": 0.0504,
      "step": 9050
    },
    {
      "epoch": 0.6353417580115898,
      "grad_norm": 3.2266767024993896,
      "learning_rate": 4.850278497509814e-05,
      "loss": 0.0406,
      "step": 9100
    },
    {
      "epoch": 0.6388326467918732,
      "grad_norm": 3.8187813758850098,
      "learning_rate": 4.846399702108513e-05,
      "loss": 0.0347,
      "step": 9150
    },
    {
      "epoch": 0.6423235355721567,
      "grad_norm": 0.10258898138999939,
      "learning_rate": 4.842520906707213e-05,
      "loss": 0.0447,
      "step": 9200
    },
    {
      "epoch": 0.6458144243524402,
      "grad_norm": 2.6401562690734863,
      "learning_rate": 4.838642111305913e-05,
      "loss": 0.0313,
      "step": 9250
    },
    {
      "epoch": 0.6493053131327235,
      "grad_norm": 2.760939359664917,
      "learning_rate": 4.834763315904613e-05,
      "loss": 0.0412,
      "step": 9300
    },
    {
      "epoch": 0.652796201913007,
      "grad_norm": 0.3721374571323395,
      "learning_rate": 4.8308845205033124e-05,
      "loss": 0.0526,
      "step": 9350
    },
    {
      "epoch": 0.6562870906932905,
      "grad_norm": 11.350992202758789,
      "learning_rate": 4.8270057251020124e-05,
      "loss": 0.0435,
      "step": 9400
    },
    {
      "epoch": 0.659777979473574,
      "grad_norm": 0.2358410507440567,
      "learning_rate": 4.8231269297007124e-05,
      "loss": 0.0332,
      "step": 9450
    },
    {
      "epoch": 0.6632688682538574,
      "grad_norm": 0.27583521604537964,
      "learning_rate": 4.8192481342994125e-05,
      "loss": 0.0219,
      "step": 9500
    },
    {
      "epoch": 0.6667597570341409,
      "grad_norm": 0.19773100316524506,
      "learning_rate": 4.815369338898112e-05,
      "loss": 0.0365,
      "step": 9550
    },
    {
      "epoch": 0.6702506458144244,
      "grad_norm": 0.3891470730304718,
      "learning_rate": 4.811490543496812e-05,
      "loss": 0.04,
      "step": 9600
    },
    {
      "epoch": 0.6737415345947078,
      "grad_norm": 0.1269948035478592,
      "learning_rate": 4.807611748095512e-05,
      "loss": 0.0199,
      "step": 9650
    },
    {
      "epoch": 0.6772324233749912,
      "grad_norm": 1.7669986486434937,
      "learning_rate": 4.803732952694212e-05,
      "loss": 0.0342,
      "step": 9700
    },
    {
      "epoch": 0.6807233121552747,
      "grad_norm": 0.2495778650045395,
      "learning_rate": 4.799854157292911e-05,
      "loss": 0.0324,
      "step": 9750
    },
    {
      "epoch": 0.6842142009355582,
      "grad_norm": 0.7094061970710754,
      "learning_rate": 4.795975361891611e-05,
      "loss": 0.0268,
      "step": 9800
    },
    {
      "epoch": 0.6877050897158417,
      "grad_norm": 0.10349750518798828,
      "learning_rate": 4.792096566490311e-05,
      "loss": 0.0197,
      "step": 9850
    },
    {
      "epoch": 0.6911959784961251,
      "grad_norm": 0.08370551466941833,
      "learning_rate": 4.7882177710890105e-05,
      "loss": 0.026,
      "step": 9900
    },
    {
      "epoch": 0.6946868672764086,
      "grad_norm": 0.6155123114585876,
      "learning_rate": 4.7843389756877105e-05,
      "loss": 0.0354,
      "step": 9950
    },
    {
      "epoch": 0.6981777560566921,
      "grad_norm": 0.2413477748632431,
      "learning_rate": 4.7804601802864105e-05,
      "loss": 0.0372,
      "step": 10000
    },
    {
      "epoch": 0.6981777560566921,
      "eval_f1": 0.8886486486486487,
      "eval_loss": 0.30088678002357483,
      "eval_runtime": 26.0507,
      "eval_samples_per_second": 140.495,
      "eval_steps_per_second": 4.414,
      "step": 10000
    },
    {
      "epoch": 0.7016686448369754,
      "grad_norm": 0.028105752542614937,
      "learning_rate": 4.7765813848851105e-05,
      "loss": 0.021,
      "step": 10050
    },
    {
      "epoch": 0.7051595336172589,
      "grad_norm": 0.44947728514671326,
      "learning_rate": 4.77270258948381e-05,
      "loss": 0.0436,
      "step": 10100
    },
    {
      "epoch": 0.7086504223975424,
      "grad_norm": 0.13265953958034515,
      "learning_rate": 4.76882379408251e-05,
      "loss": 0.0224,
      "step": 10150
    },
    {
      "epoch": 0.7121413111778259,
      "grad_norm": 0.060906197875738144,
      "learning_rate": 4.76494499868121e-05,
      "loss": 0.0419,
      "step": 10200
    },
    {
      "epoch": 0.7156321999581093,
      "grad_norm": 3.1417033672332764,
      "learning_rate": 4.76106620327991e-05,
      "loss": 0.0324,
      "step": 10250
    },
    {
      "epoch": 0.7191230887383928,
      "grad_norm": 8.964985847473145,
      "learning_rate": 4.757187407878609e-05,
      "loss": 0.0326,
      "step": 10300
    },
    {
      "epoch": 0.7226139775186763,
      "grad_norm": 0.026810016483068466,
      "learning_rate": 4.753308612477309e-05,
      "loss": 0.034,
      "step": 10350
    },
    {
      "epoch": 0.7261048662989598,
      "grad_norm": 2.3609349727630615,
      "learning_rate": 4.749429817076009e-05,
      "loss": 0.036,
      "step": 10400
    },
    {
      "epoch": 0.7295957550792431,
      "grad_norm": 0.163139209151268,
      "learning_rate": 4.745551021674709e-05,
      "loss": 0.0251,
      "step": 10450
    },
    {
      "epoch": 0.7330866438595266,
      "grad_norm": 10.265725135803223,
      "learning_rate": 4.7416722262734085e-05,
      "loss": 0.0198,
      "step": 10500
    },
    {
      "epoch": 0.7365775326398101,
      "grad_norm": 0.017616290599107742,
      "learning_rate": 4.7377934308721085e-05,
      "loss": 0.0321,
      "step": 10550
    },
    {
      "epoch": 0.7400684214200935,
      "grad_norm": 4.75632381439209,
      "learning_rate": 4.7339146354708086e-05,
      "loss": 0.0406,
      "step": 10600
    },
    {
      "epoch": 0.743559310200377,
      "grad_norm": 4.056220531463623,
      "learning_rate": 4.7300358400695086e-05,
      "loss": 0.0268,
      "step": 10650
    },
    {
      "epoch": 0.7470501989806605,
      "grad_norm": 0.041627198457717896,
      "learning_rate": 4.726157044668208e-05,
      "loss": 0.0355,
      "step": 10700
    },
    {
      "epoch": 0.750541087760944,
      "grad_norm": 0.05310547724366188,
      "learning_rate": 4.722278249266908e-05,
      "loss": 0.0298,
      "step": 10750
    },
    {
      "epoch": 0.7540319765412274,
      "grad_norm": 3.478668451309204,
      "learning_rate": 4.718399453865608e-05,
      "loss": 0.0469,
      "step": 10800
    },
    {
      "epoch": 0.7575228653215108,
      "grad_norm": 0.046948205679655075,
      "learning_rate": 4.714520658464307e-05,
      "loss": 0.0217,
      "step": 10850
    },
    {
      "epoch": 0.7610137541017943,
      "grad_norm": 1.857437252998352,
      "learning_rate": 4.710641863063007e-05,
      "loss": 0.0452,
      "step": 10900
    },
    {
      "epoch": 0.7645046428820778,
      "grad_norm": 0.5096798539161682,
      "learning_rate": 4.706763067661707e-05,
      "loss": 0.0448,
      "step": 10950
    },
    {
      "epoch": 0.7679955316623612,
      "grad_norm": 0.05577833950519562,
      "learning_rate": 4.702884272260407e-05,
      "loss": 0.0236,
      "step": 11000
    },
    {
      "epoch": 0.7679955316623612,
      "eval_f1": 0.7422608695652174,
      "eval_loss": 0.6054657101631165,
      "eval_runtime": 18.0002,
      "eval_samples_per_second": 203.331,
      "eval_steps_per_second": 6.389,
      "step": 11000
    },
    {
      "epoch": 0.7714864204426447,
      "grad_norm": 0.32365405559539795,
      "learning_rate": 4.6990054768591066e-05,
      "loss": 0.0275,
      "step": 11050
    },
    {
      "epoch": 0.7749773092229282,
      "grad_norm": 0.054900918155908585,
      "learning_rate": 4.6951266814578066e-05,
      "loss": 0.0428,
      "step": 11100
    },
    {
      "epoch": 0.7784681980032117,
      "grad_norm": 0.1931043118238449,
      "learning_rate": 4.6912478860565066e-05,
      "loss": 0.0385,
      "step": 11150
    },
    {
      "epoch": 0.7819590867834951,
      "grad_norm": 11.70685863494873,
      "learning_rate": 4.6873690906552066e-05,
      "loss": 0.0386,
      "step": 11200
    },
    {
      "epoch": 0.7854499755637785,
      "grad_norm": 1.1998494863510132,
      "learning_rate": 4.683490295253906e-05,
      "loss": 0.0412,
      "step": 11250
    },
    {
      "epoch": 0.788940864344062,
      "grad_norm": 4.7984161376953125,
      "learning_rate": 4.679611499852606e-05,
      "loss": 0.0451,
      "step": 11300
    },
    {
      "epoch": 0.7924317531243454,
      "grad_norm": 0.10824266076087952,
      "learning_rate": 4.675732704451306e-05,
      "loss": 0.027,
      "step": 11350
    },
    {
      "epoch": 0.7959226419046289,
      "grad_norm": 11.501607894897461,
      "learning_rate": 4.671853909050006e-05,
      "loss": 0.0311,
      "step": 11400
    },
    {
      "epoch": 0.7994135306849124,
      "grad_norm": 4.8870978355407715,
      "learning_rate": 4.667975113648705e-05,
      "loss": 0.0382,
      "step": 11450
    },
    {
      "epoch": 0.8029044194651959,
      "grad_norm": 0.020871387794613838,
      "learning_rate": 4.664096318247405e-05,
      "loss": 0.0288,
      "step": 11500
    },
    {
      "epoch": 0.8063953082454793,
      "grad_norm": 1.9301238059997559,
      "learning_rate": 4.660217522846105e-05,
      "loss": 0.0388,
      "step": 11550
    },
    {
      "epoch": 0.8098861970257628,
      "grad_norm": 0.012374644167721272,
      "learning_rate": 4.656338727444805e-05,
      "loss": 0.0188,
      "step": 11600
    },
    {
      "epoch": 0.8133770858060462,
      "grad_norm": 0.08517163246870041,
      "learning_rate": 4.6524599320435046e-05,
      "loss": 0.0359,
      "step": 11650
    },
    {
      "epoch": 0.8168679745863296,
      "grad_norm": 0.027777288109064102,
      "learning_rate": 4.6485811366422046e-05,
      "loss": 0.0351,
      "step": 11700
    },
    {
      "epoch": 0.8203588633666131,
      "grad_norm": 4.146106719970703,
      "learning_rate": 4.6447023412409047e-05,
      "loss": 0.0286,
      "step": 11750
    },
    {
      "epoch": 0.8238497521468966,
      "grad_norm": 0.18585865199565887,
      "learning_rate": 4.640823545839604e-05,
      "loss": 0.0196,
      "step": 11800
    },
    {
      "epoch": 0.8273406409271801,
      "grad_norm": 0.03932729363441467,
      "learning_rate": 4.636944750438304e-05,
      "loss": 0.0205,
      "step": 11850
    },
    {
      "epoch": 0.8308315297074635,
      "grad_norm": 0.49756330251693726,
      "learning_rate": 4.633065955037004e-05,
      "loss": 0.0411,
      "step": 11900
    },
    {
      "epoch": 0.834322418487747,
      "grad_norm": 1.2074888944625854,
      "learning_rate": 4.629187159635704e-05,
      "loss": 0.0204,
      "step": 11950
    },
    {
      "epoch": 0.8378133072680305,
      "grad_norm": 0.08632475137710571,
      "learning_rate": 4.6253083642344033e-05,
      "loss": 0.0293,
      "step": 12000
    },
    {
      "epoch": 0.8378133072680305,
      "eval_f1": 0.8149117069980379,
      "eval_loss": 0.44084152579307556,
      "eval_runtime": 18.2839,
      "eval_samples_per_second": 200.176,
      "eval_steps_per_second": 6.29,
      "step": 12000
    },
    {
      "epoch": 0.8413041960483139,
      "grad_norm": 0.5310344696044922,
      "learning_rate": 4.6214295688331033e-05,
      "loss": 0.0364,
      "step": 12050
    },
    {
      "epoch": 0.8447950848285973,
      "grad_norm": 2.3562138080596924,
      "learning_rate": 4.617550773431803e-05,
      "loss": 0.0438,
      "step": 12100
    },
    {
      "epoch": 0.8482859736088808,
      "grad_norm": 3.931169271469116,
      "learning_rate": 4.6136719780305034e-05,
      "loss": 0.0181,
      "step": 12150
    },
    {
      "epoch": 0.8517768623891643,
      "grad_norm": 0.08786332607269287,
      "learning_rate": 4.609793182629203e-05,
      "loss": 0.0256,
      "step": 12200
    },
    {
      "epoch": 0.8552677511694478,
      "grad_norm": 0.5437902212142944,
      "learning_rate": 4.605914387227903e-05,
      "loss": 0.0353,
      "step": 12250
    },
    {
      "epoch": 0.8587586399497312,
      "grad_norm": 0.21293994784355164,
      "learning_rate": 4.602035591826603e-05,
      "loss": 0.0301,
      "step": 12300
    },
    {
      "epoch": 0.8622495287300147,
      "grad_norm": 0.2317512184381485,
      "learning_rate": 4.598156796425303e-05,
      "loss": 0.037,
      "step": 12350
    },
    {
      "epoch": 0.8657404175102982,
      "grad_norm": 4.126286029815674,
      "learning_rate": 4.594278001024002e-05,
      "loss": 0.0303,
      "step": 12400
    },
    {
      "epoch": 0.8692313062905815,
      "grad_norm": 0.08132177591323853,
      "learning_rate": 4.590399205622702e-05,
      "loss": 0.0292,
      "step": 12450
    },
    {
      "epoch": 0.872722195070865,
      "grad_norm": 4.297754764556885,
      "learning_rate": 4.586520410221402e-05,
      "loss": 0.0493,
      "step": 12500
    },
    {
      "epoch": 0.8762130838511485,
      "grad_norm": 6.928088188171387,
      "learning_rate": 4.582641614820102e-05,
      "loss": 0.0347,
      "step": 12550
    },
    {
      "epoch": 0.879703972631432,
      "grad_norm": 0.3550567924976349,
      "learning_rate": 4.5787628194188014e-05,
      "loss": 0.0265,
      "step": 12600
    },
    {
      "epoch": 0.8831948614117154,
      "grad_norm": 1.2661274671554565,
      "learning_rate": 4.5748840240175014e-05,
      "loss": 0.0343,
      "step": 12650
    },
    {
      "epoch": 0.8866857501919989,
      "grad_norm": 5.1844258308410645,
      "learning_rate": 4.5710052286162014e-05,
      "loss": 0.0209,
      "step": 12700
    },
    {
      "epoch": 0.8901766389722824,
      "grad_norm": 0.31868308782577515,
      "learning_rate": 4.5671264332149014e-05,
      "loss": 0.0238,
      "step": 12750
    },
    {
      "epoch": 0.8936675277525657,
      "grad_norm": 0.03575870022177696,
      "learning_rate": 4.563247637813601e-05,
      "loss": 0.0369,
      "step": 12800
    },
    {
      "epoch": 0.8971584165328492,
      "grad_norm": 1.8217113018035889,
      "learning_rate": 4.559368842412301e-05,
      "loss": 0.0298,
      "step": 12850
    },
    {
      "epoch": 0.9006493053131327,
      "grad_norm": 0.02744445390999317,
      "learning_rate": 4.555490047011001e-05,
      "loss": 0.0205,
      "step": 12900
    },
    {
      "epoch": 0.9041401940934162,
      "grad_norm": 0.18486034870147705,
      "learning_rate": 4.5516112516097e-05,
      "loss": 0.0399,
      "step": 12950
    },
    {
      "epoch": 0.9076310828736996,
      "grad_norm": 0.38794612884521484,
      "learning_rate": 4.5477324562084e-05,
      "loss": 0.0345,
      "step": 13000
    },
    {
      "epoch": 0.9076310828736996,
      "eval_f1": 0.6745126884884148,
      "eval_loss": 0.701119065284729,
      "eval_runtime": 19.4257,
      "eval_samples_per_second": 188.411,
      "eval_steps_per_second": 5.92,
      "step": 13000
    },
    {
      "epoch": 0.9111219716539831,
      "grad_norm": 0.06368507444858551,
      "learning_rate": 4.5438536608070994e-05,
      "loss": 0.0474,
      "step": 13050
    },
    {
      "epoch": 0.9146128604342666,
      "grad_norm": 1.3088841438293457,
      "learning_rate": 4.5399748654058e-05,
      "loss": 0.0306,
      "step": 13100
    },
    {
      "epoch": 0.9181037492145501,
      "grad_norm": 0.21806786954402924,
      "learning_rate": 4.5360960700044994e-05,
      "loss": 0.0299,
      "step": 13150
    },
    {
      "epoch": 0.9215946379948334,
      "grad_norm": 0.05096457153558731,
      "learning_rate": 4.5322172746031995e-05,
      "loss": 0.0261,
      "step": 13200
    },
    {
      "epoch": 0.9250855267751169,
      "grad_norm": 0.14264649152755737,
      "learning_rate": 4.528338479201899e-05,
      "loss": 0.0426,
      "step": 13250
    },
    {
      "epoch": 0.9285764155554004,
      "grad_norm": 0.4436454474925995,
      "learning_rate": 4.5244596838005995e-05,
      "loss": 0.0284,
      "step": 13300
    },
    {
      "epoch": 0.9320673043356839,
      "grad_norm": 2.399491786956787,
      "learning_rate": 4.520580888399299e-05,
      "loss": 0.0336,
      "step": 13350
    },
    {
      "epoch": 0.9355581931159673,
      "grad_norm": 0.05171108618378639,
      "learning_rate": 4.516702092997999e-05,
      "loss": 0.0151,
      "step": 13400
    },
    {
      "epoch": 0.9390490818962508,
      "grad_norm": 0.022188086062669754,
      "learning_rate": 4.512823297596699e-05,
      "loss": 0.0235,
      "step": 13450
    },
    {
      "epoch": 0.9425399706765343,
      "grad_norm": 8.833630561828613,
      "learning_rate": 4.508944502195399e-05,
      "loss": 0.0405,
      "step": 13500
    },
    {
      "epoch": 0.9460308594568178,
      "grad_norm": 0.18842175602912903,
      "learning_rate": 4.505065706794098e-05,
      "loss": 0.0237,
      "step": 13550
    },
    {
      "epoch": 0.9495217482371011,
      "grad_norm": 0.501763105392456,
      "learning_rate": 4.501186911392798e-05,
      "loss": 0.0247,
      "step": 13600
    },
    {
      "epoch": 0.9530126370173846,
      "grad_norm": 0.012047139927744865,
      "learning_rate": 4.497308115991498e-05,
      "loss": 0.0208,
      "step": 13650
    },
    {
      "epoch": 0.9565035257976681,
      "grad_norm": 0.04417157545685768,
      "learning_rate": 4.493429320590198e-05,
      "loss": 0.0314,
      "step": 13700
    },
    {
      "epoch": 0.9599944145779515,
      "grad_norm": 0.29428571462631226,
      "learning_rate": 4.4895505251888975e-05,
      "loss": 0.0279,
      "step": 13750
    },
    {
      "epoch": 0.963485303358235,
      "grad_norm": 6.524214744567871,
      "learning_rate": 4.4856717297875975e-05,
      "loss": 0.0225,
      "step": 13800
    },
    {
      "epoch": 0.9669761921385185,
      "grad_norm": 0.15136292576789856,
      "learning_rate": 4.4817929343862975e-05,
      "loss": 0.0158,
      "step": 13850
    },
    {
      "epoch": 0.970467080918802,
      "grad_norm": 0.1073521077632904,
      "learning_rate": 4.477914138984997e-05,
      "loss": 0.0219,
      "step": 13900
    },
    {
      "epoch": 0.9739579696990854,
      "grad_norm": 0.04597226157784462,
      "learning_rate": 4.474035343583697e-05,
      "loss": 0.0271,
      "step": 13950
    },
    {
      "epoch": 0.9774488584793688,
      "grad_norm": 0.5995834469795227,
      "learning_rate": 4.470156548182396e-05,
      "loss": 0.0655,
      "step": 14000
    },
    {
      "epoch": 0.9774488584793688,
      "eval_f1": 0.8513046211883055,
      "eval_loss": 0.3484158217906952,
      "eval_runtime": 16.0623,
      "eval_samples_per_second": 227.863,
      "eval_steps_per_second": 7.16,
      "step": 14000
    },
    {
      "epoch": 0.9809397472596523,
      "grad_norm": 0.03915255144238472,
      "learning_rate": 4.466277752781097e-05,
      "loss": 0.0194,
      "step": 14050
    },
    {
      "epoch": 0.9844306360399357,
      "grad_norm": 5.527812957763672,
      "learning_rate": 4.462398957379796e-05,
      "loss": 0.0187,
      "step": 14100
    },
    {
      "epoch": 0.9879215248202192,
      "grad_norm": 0.04913317784667015,
      "learning_rate": 4.458520161978496e-05,
      "loss": 0.0282,
      "step": 14150
    },
    {
      "epoch": 0.9914124136005027,
      "grad_norm": 0.11622585356235504,
      "learning_rate": 4.4546413665771955e-05,
      "loss": 0.0372,
      "step": 14200
    },
    {
      "epoch": 0.9949033023807862,
      "grad_norm": 16.156946182250977,
      "learning_rate": 4.450762571175896e-05,
      "loss": 0.0193,
      "step": 14250
    },
    {
      "epoch": 0.9983941911610696,
      "grad_norm": 0.05035267770290375,
      "learning_rate": 4.4468837757745955e-05,
      "loss": 0.0273,
      "step": 14300
    },
    {
      "epoch": 1.001885079941353,
      "grad_norm": 1.6708167791366577,
      "learning_rate": 4.4430049803732956e-05,
      "loss": 0.0203,
      "step": 14350
    },
    {
      "epoch": 1.0053759687216366,
      "grad_norm": 0.09074031561613083,
      "learning_rate": 4.439126184971995e-05,
      "loss": 0.0338,
      "step": 14400
    },
    {
      "epoch": 1.00886685750192,
      "grad_norm": 3.624009609222412,
      "learning_rate": 4.4352473895706956e-05,
      "loss": 0.0455,
      "step": 14450
    },
    {
      "epoch": 1.0123577462822035,
      "grad_norm": 0.1196989193558693,
      "learning_rate": 4.431368594169395e-05,
      "loss": 0.0419,
      "step": 14500
    },
    {
      "epoch": 1.015848635062487,
      "grad_norm": 0.21538716554641724,
      "learning_rate": 4.427489798768095e-05,
      "loss": 0.0332,
      "step": 14550
    },
    {
      "epoch": 1.0193395238427703,
      "grad_norm": 0.029338959604501724,
      "learning_rate": 4.423611003366794e-05,
      "loss": 0.0266,
      "step": 14600
    },
    {
      "epoch": 1.0228304126230539,
      "grad_norm": 0.02669925056397915,
      "learning_rate": 4.419732207965495e-05,
      "loss": 0.0321,
      "step": 14650
    },
    {
      "epoch": 1.0263213014033372,
      "grad_norm": 0.6535504460334778,
      "learning_rate": 4.415853412564194e-05,
      "loss": 0.0242,
      "step": 14700
    },
    {
      "epoch": 1.0298121901836208,
      "grad_norm": 0.07374177873134613,
      "learning_rate": 4.41205219307092e-05,
      "loss": 0.0381,
      "step": 14750
    },
    {
      "epoch": 1.0333030789639042,
      "grad_norm": 0.09146462380886078,
      "learning_rate": 4.40817339766962e-05,
      "loss": 0.0312,
      "step": 14800
    },
    {
      "epoch": 1.0367939677441877,
      "grad_norm": 0.1045895665884018,
      "learning_rate": 4.4042946022683194e-05,
      "loss": 0.0284,
      "step": 14850
    },
    {
      "epoch": 1.0402848565244711,
      "grad_norm": 0.008818292059004307,
      "learning_rate": 4.40041580686702e-05,
      "loss": 0.0116,
      "step": 14900
    },
    {
      "epoch": 1.0437757453047545,
      "grad_norm": 0.29010212421417236,
      "learning_rate": 4.3965370114657194e-05,
      "loss": 0.0395,
      "step": 14950
    },
    {
      "epoch": 1.047266634085038,
      "grad_norm": 0.12948091328144073,
      "learning_rate": 4.3926582160644194e-05,
      "loss": 0.0264,
      "step": 15000
    },
    {
      "epoch": 1.047266634085038,
      "eval_f1": 0.8383579217447081,
      "eval_loss": 0.4777594804763794,
      "eval_runtime": 16.0185,
      "eval_samples_per_second": 228.486,
      "eval_steps_per_second": 7.179,
      "step": 15000
    },
    {
      "epoch": 1.0507575228653214,
      "grad_norm": 0.032540276646614075,
      "learning_rate": 4.388779420663119e-05,
      "loss": 0.0378,
      "step": 15050
    },
    {
      "epoch": 1.054248411645605,
      "grad_norm": 0.2068353295326233,
      "learning_rate": 4.3849006252618194e-05,
      "loss": 0.0303,
      "step": 15100
    },
    {
      "epoch": 1.0577393004258884,
      "grad_norm": 0.1535184383392334,
      "learning_rate": 4.381021829860519e-05,
      "loss": 0.0143,
      "step": 15150
    },
    {
      "epoch": 1.061230189206172,
      "grad_norm": 0.21458381414413452,
      "learning_rate": 4.377143034459219e-05,
      "loss": 0.0464,
      "step": 15200
    },
    {
      "epoch": 1.0647210779864553,
      "grad_norm": 0.054905761033296585,
      "learning_rate": 4.373264239057918e-05,
      "loss": 0.0257,
      "step": 15250
    },
    {
      "epoch": 1.068211966766739,
      "grad_norm": 2.5384442806243896,
      "learning_rate": 4.369385443656618e-05,
      "loss": 0.0392,
      "step": 15300
    },
    {
      "epoch": 1.0717028555470223,
      "grad_norm": 1.0916022062301636,
      "learning_rate": 4.365506648255318e-05,
      "loss": 0.0273,
      "step": 15350
    },
    {
      "epoch": 1.0751937443273056,
      "grad_norm": 0.11858826130628586,
      "learning_rate": 4.3616278528540174e-05,
      "loss": 0.0271,
      "step": 15400
    },
    {
      "epoch": 1.0786846331075892,
      "grad_norm": 0.030543260276317596,
      "learning_rate": 4.3577490574527175e-05,
      "loss": 0.0331,
      "step": 15450
    },
    {
      "epoch": 1.0821755218878726,
      "grad_norm": 0.08327992260456085,
      "learning_rate": 4.3538702620514175e-05,
      "loss": 0.0375,
      "step": 15500
    },
    {
      "epoch": 1.0856664106681562,
      "grad_norm": 6.117712497711182,
      "learning_rate": 4.3499914666501175e-05,
      "loss": 0.037,
      "step": 15550
    },
    {
      "epoch": 1.0891572994484395,
      "grad_norm": 0.12089100480079651,
      "learning_rate": 4.346112671248817e-05,
      "loss": 0.0237,
      "step": 15600
    },
    {
      "epoch": 1.0926481882287231,
      "grad_norm": 1.5640662908554077,
      "learning_rate": 4.342233875847517e-05,
      "loss": 0.0227,
      "step": 15650
    },
    {
      "epoch": 1.0961390770090065,
      "grad_norm": 0.04615291208028793,
      "learning_rate": 4.338355080446217e-05,
      "loss": 0.0233,
      "step": 15700
    },
    {
      "epoch": 1.09962996578929,
      "grad_norm": 0.047694265842437744,
      "learning_rate": 4.334476285044917e-05,
      "loss": 0.0424,
      "step": 15750
    },
    {
      "epoch": 1.1031208545695734,
      "grad_norm": 0.022557301446795464,
      "learning_rate": 4.330597489643616e-05,
      "loss": 0.028,
      "step": 15800
    },
    {
      "epoch": 1.1066117433498568,
      "grad_norm": 0.41174036264419556,
      "learning_rate": 4.326718694242317e-05,
      "loss": 0.0423,
      "step": 15850
    },
    {
      "epoch": 1.1101026321301404,
      "grad_norm": 1.035745620727539,
      "learning_rate": 4.322839898841016e-05,
      "loss": 0.0158,
      "step": 15900
    },
    {
      "epoch": 1.1135935209104237,
      "grad_norm": 0.017003023996949196,
      "learning_rate": 4.318961103439716e-05,
      "loss": 0.0278,
      "step": 15950
    },
    {
      "epoch": 1.1170844096907073,
      "grad_norm": 0.030238742008805275,
      "learning_rate": 4.3150823080384155e-05,
      "loss": 0.0197,
      "step": 16000
    },
    {
      "epoch": 1.1170844096907073,
      "eval_f1": 0.8951467944877172,
      "eval_loss": 0.28839099407196045,
      "eval_runtime": 16.2773,
      "eval_samples_per_second": 224.853,
      "eval_steps_per_second": 7.065,
      "step": 16000
    },
    {
      "epoch": 1.1205752984709907,
      "grad_norm": 0.06707513332366943,
      "learning_rate": 4.311203512637116e-05,
      "loss": 0.0268,
      "step": 16050
    },
    {
      "epoch": 1.1240661872512743,
      "grad_norm": 0.29435157775878906,
      "learning_rate": 4.3073247172358155e-05,
      "loss": 0.0389,
      "step": 16100
    },
    {
      "epoch": 1.1275570760315576,
      "grad_norm": 0.4163244664669037,
      "learning_rate": 4.3034459218345155e-05,
      "loss": 0.0149,
      "step": 16150
    },
    {
      "epoch": 1.131047964811841,
      "grad_norm": 0.09609521925449371,
      "learning_rate": 4.299567126433215e-05,
      "loss": 0.0231,
      "step": 16200
    },
    {
      "epoch": 1.1345388535921246,
      "grad_norm": 1.1996419429779053,
      "learning_rate": 4.295688331031915e-05,
      "loss": 0.0279,
      "step": 16250
    },
    {
      "epoch": 1.138029742372408,
      "grad_norm": 0.03178354725241661,
      "learning_rate": 4.291809535630615e-05,
      "loss": 0.0396,
      "step": 16300
    },
    {
      "epoch": 1.1415206311526915,
      "grad_norm": 1.533177137374878,
      "learning_rate": 4.287930740229314e-05,
      "loss": 0.0274,
      "step": 16350
    },
    {
      "epoch": 1.145011519932975,
      "grad_norm": 0.05997175723314285,
      "learning_rate": 4.284051944828014e-05,
      "loss": 0.0196,
      "step": 16400
    },
    {
      "epoch": 1.1485024087132585,
      "grad_norm": 0.022537073120474815,
      "learning_rate": 4.280173149426714e-05,
      "loss": 0.0142,
      "step": 16450
    },
    {
      "epoch": 1.1519932974935418,
      "grad_norm": 0.022068103775382042,
      "learning_rate": 4.276294354025414e-05,
      "loss": 0.0226,
      "step": 16500
    },
    {
      "epoch": 1.1554841862738252,
      "grad_norm": 0.07768739014863968,
      "learning_rate": 4.2724155586241135e-05,
      "loss": 0.0323,
      "step": 16550
    },
    {
      "epoch": 1.1589750750541088,
      "grad_norm": 0.01296279113739729,
      "learning_rate": 4.2685367632228136e-05,
      "loss": 0.0253,
      "step": 16600
    },
    {
      "epoch": 1.1624659638343922,
      "grad_norm": 3.2529516220092773,
      "learning_rate": 4.2646579678215136e-05,
      "loss": 0.0266,
      "step": 16650
    },
    {
      "epoch": 1.1659568526146757,
      "grad_norm": 0.11319205164909363,
      "learning_rate": 4.2607791724202136e-05,
      "loss": 0.0225,
      "step": 16700
    },
    {
      "epoch": 1.169447741394959,
      "grad_norm": 0.36928442120552063,
      "learning_rate": 4.256900377018913e-05,
      "loss": 0.0373,
      "step": 16750
    },
    {
      "epoch": 1.1729386301752427,
      "grad_norm": 1.703551173210144,
      "learning_rate": 4.253021581617613e-05,
      "loss": 0.0229,
      "step": 16800
    },
    {
      "epoch": 1.176429518955526,
      "grad_norm": 4.041821479797363,
      "learning_rate": 4.249142786216313e-05,
      "loss": 0.0345,
      "step": 16850
    },
    {
      "epoch": 1.1799204077358096,
      "grad_norm": 0.21636897325515747,
      "learning_rate": 4.245263990815013e-05,
      "loss": 0.0255,
      "step": 16900
    },
    {
      "epoch": 1.183411296516093,
      "grad_norm": 5.224414348602295,
      "learning_rate": 4.241385195413712e-05,
      "loss": 0.0282,
      "step": 16950
    },
    {
      "epoch": 1.1869021852963764,
      "grad_norm": 2.6779046058654785,
      "learning_rate": 4.237506400012413e-05,
      "loss": 0.0327,
      "step": 17000
    },
    {
      "epoch": 1.1869021852963764,
      "eval_f1": 0.9023038786818315,
      "eval_loss": 0.2744286060333252,
      "eval_runtime": 15.7046,
      "eval_samples_per_second": 233.053,
      "eval_steps_per_second": 7.323,
      "step": 17000
    },
    {
      "epoch": 1.19039307407666,
      "grad_norm": 8.236310005187988,
      "learning_rate": 4.233627604611112e-05,
      "loss": 0.0313,
      "step": 17050
    },
    {
      "epoch": 1.1938839628569433,
      "grad_norm": 0.2507387101650238,
      "learning_rate": 4.229903961025864e-05,
      "loss": 0.0335,
      "step": 17100
    },
    {
      "epoch": 1.197374851637227,
      "grad_norm": 0.040820781141519547,
      "learning_rate": 4.226025165624564e-05,
      "loss": 0.0291,
      "step": 17150
    },
    {
      "epoch": 1.2008657404175103,
      "grad_norm": 0.06928090006113052,
      "learning_rate": 4.222146370223264e-05,
      "loss": 0.0247,
      "step": 17200
    },
    {
      "epoch": 1.2043566291977938,
      "grad_norm": 3.9058620929718018,
      "learning_rate": 4.218267574821964e-05,
      "loss": 0.0215,
      "step": 17250
    },
    {
      "epoch": 1.2078475179780772,
      "grad_norm": 0.13741998374462128,
      "learning_rate": 4.214388779420663e-05,
      "loss": 0.0164,
      "step": 17300
    },
    {
      "epoch": 1.2113384067583608,
      "grad_norm": 1.5019398927688599,
      "learning_rate": 4.210509984019363e-05,
      "loss": 0.0435,
      "step": 17350
    },
    {
      "epoch": 1.2148292955386442,
      "grad_norm": 6.695947647094727,
      "learning_rate": 4.206631188618063e-05,
      "loss": 0.0477,
      "step": 17400
    },
    {
      "epoch": 1.2183201843189275,
      "grad_norm": 0.024964775890111923,
      "learning_rate": 4.2027523932167626e-05,
      "loss": 0.0418,
      "step": 17450
    },
    {
      "epoch": 1.221811073099211,
      "grad_norm": 3.555797576904297,
      "learning_rate": 4.1988735978154626e-05,
      "loss": 0.0201,
      "step": 17500
    },
    {
      "epoch": 1.2253019618794945,
      "grad_norm": 0.15483196079730988,
      "learning_rate": 4.194994802414162e-05,
      "loss": 0.034,
      "step": 17550
    },
    {
      "epoch": 1.228792850659778,
      "grad_norm": 0.2912302017211914,
      "learning_rate": 4.1911160070128626e-05,
      "loss": 0.0295,
      "step": 17600
    },
    {
      "epoch": 1.2322837394400614,
      "grad_norm": 0.038849666714668274,
      "learning_rate": 4.187237211611562e-05,
      "loss": 0.0141,
      "step": 17650
    },
    {
      "epoch": 1.2357746282203448,
      "grad_norm": 0.03439799323678017,
      "learning_rate": 4.183358416210262e-05,
      "loss": 0.0345,
      "step": 17700
    },
    {
      "epoch": 1.2392655170006284,
      "grad_norm": 1.7664055824279785,
      "learning_rate": 4.179479620808961e-05,
      "loss": 0.0299,
      "step": 17750
    },
    {
      "epoch": 1.2427564057809117,
      "grad_norm": 5.20842981338501,
      "learning_rate": 4.175600825407662e-05,
      "loss": 0.0381,
      "step": 17800
    },
    {
      "epoch": 1.2462472945611953,
      "grad_norm": 1.018599510192871,
      "learning_rate": 4.171722030006361e-05,
      "loss": 0.0165,
      "step": 17850
    },
    {
      "epoch": 1.2497381833414787,
      "grad_norm": 3.1473236083984375,
      "learning_rate": 4.167843234605061e-05,
      "loss": 0.0236,
      "step": 17900
    },
    {
      "epoch": 1.2532290721217623,
      "grad_norm": 0.1991962492465973,
      "learning_rate": 4.163964439203761e-05,
      "loss": 0.0259,
      "step": 17950
    },
    {
      "epoch": 1.2567199609020456,
      "grad_norm": 3.431162118911743,
      "learning_rate": 4.160085643802461e-05,
      "loss": 0.0314,
      "step": 18000
    },
    {
      "epoch": 1.2567199609020456,
      "eval_f1": 0.8523869346733668,
      "eval_loss": 0.31631308794021606,
      "eval_runtime": 28.0145,
      "eval_samples_per_second": 130.646,
      "eval_steps_per_second": 4.105,
      "step": 18000
    },
    {
      "epoch": 1.2602108496823292,
      "grad_norm": 0.14350277185440063,
      "learning_rate": 4.1562068484011606e-05,
      "loss": 0.0316,
      "step": 18050
    },
    {
      "epoch": 1.2637017384626126,
      "grad_norm": 0.08211163431406021,
      "learning_rate": 4.1523280529998606e-05,
      "loss": 0.0288,
      "step": 18100
    },
    {
      "epoch": 1.267192627242896,
      "grad_norm": 3.4511756896972656,
      "learning_rate": 4.1484492575985606e-05,
      "loss": 0.0314,
      "step": 18150
    },
    {
      "epoch": 1.2706835160231795,
      "grad_norm": 0.03046209178864956,
      "learning_rate": 4.1445704621972606e-05,
      "loss": 0.0473,
      "step": 18200
    },
    {
      "epoch": 1.274174404803463,
      "grad_norm": 0.046366989612579346,
      "learning_rate": 4.14069166679596e-05,
      "loss": 0.0242,
      "step": 18250
    },
    {
      "epoch": 1.2776652935837465,
      "grad_norm": 0.03393428400158882,
      "learning_rate": 4.13681287139466e-05,
      "loss": 0.0106,
      "step": 18300
    },
    {
      "epoch": 1.2811561823640298,
      "grad_norm": 0.17253169417381287,
      "learning_rate": 4.13293407599336e-05,
      "loss": 0.0287,
      "step": 18350
    },
    {
      "epoch": 1.2846470711443134,
      "grad_norm": 2.6994237899780273,
      "learning_rate": 4.129055280592059e-05,
      "loss": 0.0269,
      "step": 18400
    },
    {
      "epoch": 1.2881379599245968,
      "grad_norm": 0.04223485663533211,
      "learning_rate": 4.125176485190759e-05,
      "loss": 0.0188,
      "step": 18450
    },
    {
      "epoch": 1.2916288487048804,
      "grad_norm": 0.0406014584004879,
      "learning_rate": 4.1212976897894587e-05,
      "loss": 0.0227,
      "step": 18500
    },
    {
      "epoch": 1.2951197374851637,
      "grad_norm": 4.774007797241211,
      "learning_rate": 4.1174188943881593e-05,
      "loss": 0.0263,
      "step": 18550
    },
    {
      "epoch": 1.298610626265447,
      "grad_norm": 7.6998677253723145,
      "learning_rate": 4.113540098986859e-05,
      "loss": 0.0469,
      "step": 18600
    },
    {
      "epoch": 1.3021015150457307,
      "grad_norm": 0.6437110304832458,
      "learning_rate": 4.109661303585559e-05,
      "loss": 0.0195,
      "step": 18650
    },
    {
      "epoch": 1.305592403826014,
      "grad_norm": 2.180750608444214,
      "learning_rate": 4.105782508184258e-05,
      "loss": 0.0281,
      "step": 18700
    },
    {
      "epoch": 1.3090832926062976,
      "grad_norm": 0.19855546951293945,
      "learning_rate": 4.101903712782959e-05,
      "loss": 0.0341,
      "step": 18750
    },
    {
      "epoch": 1.312574181386581,
      "grad_norm": 0.2433316558599472,
      "learning_rate": 4.098024917381658e-05,
      "loss": 0.0217,
      "step": 18800
    },
    {
      "epoch": 1.3160650701668644,
      "grad_norm": 5.1525163650512695,
      "learning_rate": 4.094146121980358e-05,
      "loss": 0.0245,
      "step": 18850
    },
    {
      "epoch": 1.319555958947148,
      "grad_norm": 4.688765525817871,
      "learning_rate": 4.0902673265790574e-05,
      "loss": 0.0361,
      "step": 18900
    },
    {
      "epoch": 1.3230468477274315,
      "grad_norm": 0.07672595232725143,
      "learning_rate": 4.086388531177758e-05,
      "loss": 0.0383,
      "step": 18950
    },
    {
      "epoch": 1.326537736507715,
      "grad_norm": 1.0029828548431396,
      "learning_rate": 4.0825097357764574e-05,
      "loss": 0.0172,
      "step": 19000
    },
    {
      "epoch": 1.326537736507715,
      "eval_f1": 0.8066115702479338,
      "eval_loss": 0.35703688859939575,
      "eval_runtime": 19.0041,
      "eval_samples_per_second": 192.59,
      "eval_steps_per_second": 6.051,
      "step": 19000
    },
    {
      "epoch": 1.3300286252879983,
      "grad_norm": 5.897487640380859,
      "learning_rate": 4.0786309403751574e-05,
      "loss": 0.031,
      "step": 19050
    },
    {
      "epoch": 1.3335195140682818,
      "grad_norm": 0.11918721348047256,
      "learning_rate": 4.0747521449738574e-05,
      "loss": 0.038,
      "step": 19100
    },
    {
      "epoch": 1.3370104028485652,
      "grad_norm": 1.9710204601287842,
      "learning_rate": 4.0708733495725574e-05,
      "loss": 0.0304,
      "step": 19150
    },
    {
      "epoch": 1.3405012916288488,
      "grad_norm": 2.3936004638671875,
      "learning_rate": 4.066994554171257e-05,
      "loss": 0.0344,
      "step": 19200
    },
    {
      "epoch": 1.3439921804091322,
      "grad_norm": 1.0621711015701294,
      "learning_rate": 4.063115758769957e-05,
      "loss": 0.03,
      "step": 19250
    },
    {
      "epoch": 1.3474830691894155,
      "grad_norm": 0.2894875407218933,
      "learning_rate": 4.059236963368657e-05,
      "loss": 0.02,
      "step": 19300
    },
    {
      "epoch": 1.350973957969699,
      "grad_norm": 4.131627559661865,
      "learning_rate": 4.055358167967356e-05,
      "loss": 0.0263,
      "step": 19350
    },
    {
      "epoch": 1.3544648467499825,
      "grad_norm": 0.20086219906806946,
      "learning_rate": 4.051479372566056e-05,
      "loss": 0.02,
      "step": 19400
    },
    {
      "epoch": 1.357955735530266,
      "grad_norm": 0.02272665873169899,
      "learning_rate": 4.0476005771647554e-05,
      "loss": 0.0281,
      "step": 19450
    },
    {
      "epoch": 1.3614466243105494,
      "grad_norm": 0.08684906363487244,
      "learning_rate": 4.043721781763456e-05,
      "loss": 0.0292,
      "step": 19500
    },
    {
      "epoch": 1.364937513090833,
      "grad_norm": 1.9985971450805664,
      "learning_rate": 4.0398429863621554e-05,
      "loss": 0.0269,
      "step": 19550
    },
    {
      "epoch": 1.3684284018711164,
      "grad_norm": 0.5104591250419617,
      "learning_rate": 4.0359641909608554e-05,
      "loss": 0.0305,
      "step": 19600
    },
    {
      "epoch": 1.3719192906514,
      "grad_norm": 0.08239644020795822,
      "learning_rate": 4.032085395559555e-05,
      "loss": 0.0235,
      "step": 19650
    },
    {
      "epoch": 1.3754101794316833,
      "grad_norm": 2.711775779724121,
      "learning_rate": 4.0282066001582554e-05,
      "loss": 0.0265,
      "step": 19700
    },
    {
      "epoch": 1.3789010682119667,
      "grad_norm": 0.027858775109052658,
      "learning_rate": 4.024327804756955e-05,
      "loss": 0.0155,
      "step": 19750
    },
    {
      "epoch": 1.3823919569922503,
      "grad_norm": 0.02158626727759838,
      "learning_rate": 4.020449009355655e-05,
      "loss": 0.0264,
      "step": 19800
    },
    {
      "epoch": 1.3858828457725336,
      "grad_norm": 0.09914042055606842,
      "learning_rate": 4.016570213954354e-05,
      "loss": 0.0249,
      "step": 19850
    },
    {
      "epoch": 1.3893737345528172,
      "grad_norm": 6.718291282653809,
      "learning_rate": 4.012691418553055e-05,
      "loss": 0.0213,
      "step": 19900
    },
    {
      "epoch": 1.3928646233331006,
      "grad_norm": 0.07551147788763046,
      "learning_rate": 4.008812623151754e-05,
      "loss": 0.0336,
      "step": 19950
    },
    {
      "epoch": 1.396355512113384,
      "grad_norm": 0.027326975017786026,
      "learning_rate": 4.004933827750454e-05,
      "loss": 0.0352,
      "step": 20000
    },
    {
      "epoch": 1.396355512113384,
      "eval_f1": 0.8510773130544994,
      "eval_loss": 0.35609981417655945,
      "eval_runtime": 17.6862,
      "eval_samples_per_second": 206.941,
      "eval_steps_per_second": 6.502,
      "step": 20000
    },
    {
      "epoch": 1.3998464008936675,
      "grad_norm": 0.031843628734350204,
      "learning_rate": 4.0010550323491535e-05,
      "loss": 0.028,
      "step": 20050
    },
    {
      "epoch": 1.403337289673951,
      "grad_norm": 3.878753900527954,
      "learning_rate": 3.997176236947854e-05,
      "loss": 0.0413,
      "step": 20100
    },
    {
      "epoch": 1.4068281784542345,
      "grad_norm": 0.048792481422424316,
      "learning_rate": 3.9932974415465535e-05,
      "loss": 0.0237,
      "step": 20150
    },
    {
      "epoch": 1.4103190672345178,
      "grad_norm": 0.07304860651493073,
      "learning_rate": 3.9894186461452535e-05,
      "loss": 0.031,
      "step": 20200
    },
    {
      "epoch": 1.4138099560148014,
      "grad_norm": 2.2251689434051514,
      "learning_rate": 3.9855398507439535e-05,
      "loss": 0.0415,
      "step": 20250
    },
    {
      "epoch": 1.4173008447950848,
      "grad_norm": 0.6833386421203613,
      "learning_rate": 3.981661055342653e-05,
      "loss": 0.0295,
      "step": 20300
    },
    {
      "epoch": 1.4207917335753684,
      "grad_norm": 0.09307420253753662,
      "learning_rate": 3.977782259941353e-05,
      "loss": 0.0194,
      "step": 20350
    },
    {
      "epoch": 1.4242826223556517,
      "grad_norm": 0.03538484126329422,
      "learning_rate": 3.973903464540052e-05,
      "loss": 0.024,
      "step": 20400
    },
    {
      "epoch": 1.427773511135935,
      "grad_norm": 0.026891808956861496,
      "learning_rate": 3.970024669138753e-05,
      "loss": 0.0182,
      "step": 20450
    },
    {
      "epoch": 1.4312643999162187,
      "grad_norm": 1.5537898540496826,
      "learning_rate": 3.966145873737452e-05,
      "loss": 0.0373,
      "step": 20500
    },
    {
      "epoch": 1.4347552886965023,
      "grad_norm": 0.010172019712626934,
      "learning_rate": 3.962267078336152e-05,
      "loss": 0.0301,
      "step": 20550
    },
    {
      "epoch": 1.4382461774767856,
      "grad_norm": 0.11504697799682617,
      "learning_rate": 3.9583882829348515e-05,
      "loss": 0.0182,
      "step": 20600
    },
    {
      "epoch": 1.441737066257069,
      "grad_norm": 0.08654824644327164,
      "learning_rate": 3.954509487533552e-05,
      "loss": 0.0281,
      "step": 20650
    },
    {
      "epoch": 1.4452279550373526,
      "grad_norm": 2.5057575702667236,
      "learning_rate": 3.9506306921322515e-05,
      "loss": 0.0352,
      "step": 20700
    },
    {
      "epoch": 1.448718843817636,
      "grad_norm": 0.024753259494900703,
      "learning_rate": 3.9467518967309515e-05,
      "loss": 0.0159,
      "step": 20750
    },
    {
      "epoch": 1.4522097325979195,
      "grad_norm": 0.1611536741256714,
      "learning_rate": 3.942873101329651e-05,
      "loss": 0.0307,
      "step": 20800
    },
    {
      "epoch": 1.4557006213782029,
      "grad_norm": 1.3975356817245483,
      "learning_rate": 3.9389943059283516e-05,
      "loss": 0.033,
      "step": 20850
    },
    {
      "epoch": 1.4591915101584862,
      "grad_norm": 3.3075716495513916,
      "learning_rate": 3.935115510527051e-05,
      "loss": 0.0307,
      "step": 20900
    },
    {
      "epoch": 1.4626823989387698,
      "grad_norm": 0.0137362414970994,
      "learning_rate": 3.931236715125751e-05,
      "loss": 0.0245,
      "step": 20950
    },
    {
      "epoch": 1.4661732877190532,
      "grad_norm": 6.11804723739624,
      "learning_rate": 3.92735791972445e-05,
      "loss": 0.0449,
      "step": 21000
    },
    {
      "epoch": 1.4661732877190532,
      "eval_f1": 0.8997429305912596,
      "eval_loss": 0.2522394359111786,
      "eval_runtime": 25.048,
      "eval_samples_per_second": 146.12,
      "eval_steps_per_second": 4.591,
      "step": 21000
    },
    {
      "epoch": 1.4696641764993368,
      "grad_norm": 0.04642185568809509,
      "learning_rate": 3.923479124323151e-05,
      "loss": 0.0183,
      "step": 21050
    },
    {
      "epoch": 1.4731550652796201,
      "grad_norm": 0.055214524269104004,
      "learning_rate": 3.91960032892185e-05,
      "loss": 0.0331,
      "step": 21100
    },
    {
      "epoch": 1.4766459540599037,
      "grad_norm": 0.2047278881072998,
      "learning_rate": 3.91572153352055e-05,
      "loss": 0.0197,
      "step": 21150
    },
    {
      "epoch": 1.480136842840187,
      "grad_norm": 4.8017578125,
      "learning_rate": 3.9118427381192496e-05,
      "loss": 0.0209,
      "step": 21200
    },
    {
      "epoch": 1.4836277316204707,
      "grad_norm": 0.14394736289978027,
      "learning_rate": 3.9079639427179496e-05,
      "loss": 0.0298,
      "step": 21250
    },
    {
      "epoch": 1.487118620400754,
      "grad_norm": 10.568550109863281,
      "learning_rate": 3.9040851473166496e-05,
      "loss": 0.0302,
      "step": 21300
    },
    {
      "epoch": 1.4906095091810374,
      "grad_norm": 9.700904846191406,
      "learning_rate": 3.900206351915349e-05,
      "loss": 0.0272,
      "step": 21350
    },
    {
      "epoch": 1.494100397961321,
      "grad_norm": 3.1005468368530273,
      "learning_rate": 3.8963275565140496e-05,
      "loss": 0.0375,
      "step": 21400
    },
    {
      "epoch": 1.4975912867416044,
      "grad_norm": 0.24566790461540222,
      "learning_rate": 3.892448761112749e-05,
      "loss": 0.0224,
      "step": 21450
    },
    {
      "epoch": 1.501082175521888,
      "grad_norm": 0.2529149651527405,
      "learning_rate": 3.888569965711449e-05,
      "loss": 0.0204,
      "step": 21500
    },
    {
      "epoch": 1.5045730643021713,
      "grad_norm": 0.12634265422821045,
      "learning_rate": 3.884691170310148e-05,
      "loss": 0.0412,
      "step": 21550
    },
    {
      "epoch": 1.5080639530824547,
      "grad_norm": 0.01092058327049017,
      "learning_rate": 3.880812374908849e-05,
      "loss": 0.0194,
      "step": 21600
    },
    {
      "epoch": 1.5115548418627383,
      "grad_norm": 0.050227969884872437,
      "learning_rate": 3.877011155415574e-05,
      "loss": 0.0212,
      "step": 21650
    },
    {
      "epoch": 1.5150457306430218,
      "grad_norm": 0.07757686078548431,
      "learning_rate": 3.873132360014274e-05,
      "loss": 0.0288,
      "step": 21700
    },
    {
      "epoch": 1.5185366194233052,
      "grad_norm": 3.599360466003418,
      "learning_rate": 3.869253564612974e-05,
      "loss": 0.0287,
      "step": 21750
    },
    {
      "epoch": 1.5220275082035886,
      "grad_norm": 0.23955953121185303,
      "learning_rate": 3.8653747692116734e-05,
      "loss": 0.024,
      "step": 21800
    },
    {
      "epoch": 1.5255183969838721,
      "grad_norm": 0.11281917244195938,
      "learning_rate": 3.8614959738103734e-05,
      "loss": 0.0255,
      "step": 21850
    },
    {
      "epoch": 1.5290092857641555,
      "grad_norm": 0.08529462665319443,
      "learning_rate": 3.8576171784090734e-05,
      "loss": 0.019,
      "step": 21900
    },
    {
      "epoch": 1.532500174544439,
      "grad_norm": 2.4979732036590576,
      "learning_rate": 3.8537383830077734e-05,
      "loss": 0.0259,
      "step": 21950
    },
    {
      "epoch": 1.5359910633247225,
      "grad_norm": 0.12030795216560364,
      "learning_rate": 3.849859587606473e-05,
      "loss": 0.0222,
      "step": 22000
    },
    {
      "epoch": 1.5359910633247225,
      "eval_f1": 0.8751926040061633,
      "eval_loss": 0.33271512389183044,
      "eval_runtime": 27.5904,
      "eval_samples_per_second": 132.655,
      "eval_steps_per_second": 4.168,
      "step": 22000
    },
    {
      "epoch": 1.5394819521050058,
      "grad_norm": 0.10101566463708878,
      "learning_rate": 3.845980792205173e-05,
      "loss": 0.04,
      "step": 22050
    },
    {
      "epoch": 1.5429728408852894,
      "grad_norm": 1.0485931634902954,
      "learning_rate": 3.842101996803873e-05,
      "loss": 0.0309,
      "step": 22100
    },
    {
      "epoch": 1.546463729665573,
      "grad_norm": 0.11740628629922867,
      "learning_rate": 3.838223201402573e-05,
      "loss": 0.0306,
      "step": 22150
    },
    {
      "epoch": 1.5499546184458564,
      "grad_norm": 0.20643308758735657,
      "learning_rate": 3.834344406001272e-05,
      "loss": 0.028,
      "step": 22200
    },
    {
      "epoch": 1.5534455072261397,
      "grad_norm": 0.04227438196539879,
      "learning_rate": 3.830465610599972e-05,
      "loss": 0.0268,
      "step": 22250
    },
    {
      "epoch": 1.556936396006423,
      "grad_norm": 0.17100311815738678,
      "learning_rate": 3.826586815198672e-05,
      "loss": 0.0267,
      "step": 22300
    },
    {
      "epoch": 1.5604272847867067,
      "grad_norm": 0.024188999086618423,
      "learning_rate": 3.822708019797372e-05,
      "loss": 0.0259,
      "step": 22350
    },
    {
      "epoch": 1.5639181735669903,
      "grad_norm": 0.13360099494457245,
      "learning_rate": 3.8188292243960715e-05,
      "loss": 0.0408,
      "step": 22400
    },
    {
      "epoch": 1.5674090623472736,
      "grad_norm": 0.06247904151678085,
      "learning_rate": 3.814950428994772e-05,
      "loss": 0.028,
      "step": 22450
    },
    {
      "epoch": 1.570899951127557,
      "grad_norm": 3.2922091484069824,
      "learning_rate": 3.8110716335934715e-05,
      "loss": 0.0246,
      "step": 22500
    },
    {
      "epoch": 1.5743908399078406,
      "grad_norm": 1.2226293087005615,
      "learning_rate": 3.8071928381921715e-05,
      "loss": 0.0358,
      "step": 22550
    },
    {
      "epoch": 1.5778817286881242,
      "grad_norm": 1.1220752000808716,
      "learning_rate": 3.803314042790871e-05,
      "loss": 0.0169,
      "step": 22600
    },
    {
      "epoch": 1.5813726174684075,
      "grad_norm": 0.08222747594118118,
      "learning_rate": 3.799435247389571e-05,
      "loss": 0.0168,
      "step": 22650
    },
    {
      "epoch": 1.5848635062486909,
      "grad_norm": 0.9396281242370605,
      "learning_rate": 3.795556451988271e-05,
      "loss": 0.0269,
      "step": 22700
    },
    {
      "epoch": 1.5883543950289742,
      "grad_norm": 0.038672447204589844,
      "learning_rate": 3.79167765658697e-05,
      "loss": 0.0172,
      "step": 22750
    },
    {
      "epoch": 1.5918452838092578,
      "grad_norm": 0.020772503688931465,
      "learning_rate": 3.78779886118567e-05,
      "loss": 0.0309,
      "step": 22800
    },
    {
      "epoch": 1.5953361725895414,
      "grad_norm": 0.19638319313526154,
      "learning_rate": 3.78392006578437e-05,
      "loss": 0.0246,
      "step": 22850
    },
    {
      "epoch": 1.5988270613698248,
      "grad_norm": 0.6562692523002625,
      "learning_rate": 3.78004127038307e-05,
      "loss": 0.0192,
      "step": 22900
    },
    {
      "epoch": 1.6023179501501081,
      "grad_norm": 3.3480851650238037,
      "learning_rate": 3.7761624749817695e-05,
      "loss": 0.0179,
      "step": 22950
    },
    {
      "epoch": 1.6058088389303917,
      "grad_norm": 0.07655413448810577,
      "learning_rate": 3.7722836795804695e-05,
      "loss": 0.0261,
      "step": 23000
    },
    {
      "epoch": 1.6058088389303917,
      "eval_f1": 0.9204448246364414,
      "eval_loss": 0.22322723269462585,
      "eval_runtime": 25.7811,
      "eval_samples_per_second": 141.964,
      "eval_steps_per_second": 4.461,
      "step": 23000
    },
    {
      "epoch": 1.609299727710675,
      "grad_norm": 1.4100903272628784,
      "learning_rate": 3.7684048841791695e-05,
      "loss": 0.0337,
      "step": 23050
    },
    {
      "epoch": 1.6127906164909587,
      "grad_norm": 0.710587739944458,
      "learning_rate": 3.7645260887778695e-05,
      "loss": 0.0236,
      "step": 23100
    },
    {
      "epoch": 1.616281505271242,
      "grad_norm": 0.6931687593460083,
      "learning_rate": 3.760647293376569e-05,
      "loss": 0.0225,
      "step": 23150
    },
    {
      "epoch": 1.6197723940515254,
      "grad_norm": 0.12429849058389664,
      "learning_rate": 3.756768497975269e-05,
      "loss": 0.0275,
      "step": 23200
    },
    {
      "epoch": 1.623263282831809,
      "grad_norm": 0.04798004776239395,
      "learning_rate": 3.752889702573969e-05,
      "loss": 0.033,
      "step": 23250
    },
    {
      "epoch": 1.6267541716120926,
      "grad_norm": 0.08050743490457535,
      "learning_rate": 3.749010907172669e-05,
      "loss": 0.0299,
      "step": 23300
    },
    {
      "epoch": 1.630245060392376,
      "grad_norm": 0.47208932042121887,
      "learning_rate": 3.745132111771368e-05,
      "loss": 0.0337,
      "step": 23350
    },
    {
      "epoch": 1.6337359491726593,
      "grad_norm": 9.010507583618164,
      "learning_rate": 3.741253316370068e-05,
      "loss": 0.0256,
      "step": 23400
    },
    {
      "epoch": 1.6372268379529427,
      "grad_norm": 0.23059742152690887,
      "learning_rate": 3.737374520968768e-05,
      "loss": 0.0214,
      "step": 23450
    },
    {
      "epoch": 1.6407177267332262,
      "grad_norm": 0.037250641733407974,
      "learning_rate": 3.733495725567468e-05,
      "loss": 0.0174,
      "step": 23500
    },
    {
      "epoch": 1.6442086155135098,
      "grad_norm": 0.04142063111066818,
      "learning_rate": 3.7296169301661676e-05,
      "loss": 0.0257,
      "step": 23550
    },
    {
      "epoch": 1.6476995042937932,
      "grad_norm": 0.061181291937828064,
      "learning_rate": 3.7257381347648676e-05,
      "loss": 0.0369,
      "step": 23600
    },
    {
      "epoch": 1.6511903930740766,
      "grad_norm": 0.12618523836135864,
      "learning_rate": 3.7218593393635676e-05,
      "loss": 0.028,
      "step": 23650
    },
    {
      "epoch": 1.6546812818543601,
      "grad_norm": 0.05392460525035858,
      "learning_rate": 3.717980543962267e-05,
      "loss": 0.0169,
      "step": 23700
    },
    {
      "epoch": 1.6581721706346437,
      "grad_norm": 0.589623749256134,
      "learning_rate": 3.714101748560967e-05,
      "loss": 0.0297,
      "step": 23750
    },
    {
      "epoch": 1.661663059414927,
      "grad_norm": 0.03963274881243706,
      "learning_rate": 3.710222953159667e-05,
      "loss": 0.0146,
      "step": 23800
    },
    {
      "epoch": 1.6651539481952105,
      "grad_norm": 0.0122640086337924,
      "learning_rate": 3.706344157758367e-05,
      "loss": 0.0118,
      "step": 23850
    },
    {
      "epoch": 1.6686448369754938,
      "grad_norm": 0.0617678239941597,
      "learning_rate": 3.702465362357066e-05,
      "loss": 0.0284,
      "step": 23900
    },
    {
      "epoch": 1.6721357257557774,
      "grad_norm": 0.14892399311065674,
      "learning_rate": 3.698586566955766e-05,
      "loss": 0.0248,
      "step": 23950
    },
    {
      "epoch": 1.675626614536061,
      "grad_norm": 0.011567726731300354,
      "learning_rate": 3.694707771554466e-05,
      "loss": 0.0173,
      "step": 24000
    },
    {
      "epoch": 1.675626614536061,
      "eval_f1": 0.9144698301113063,
      "eval_loss": 0.24581825733184814,
      "eval_runtime": 27.0574,
      "eval_samples_per_second": 135.268,
      "eval_steps_per_second": 4.25,
      "step": 24000
    },
    {
      "epoch": 1.6791175033163444,
      "grad_norm": 0.02023119479417801,
      "learning_rate": 3.690828976153166e-05,
      "loss": 0.0285,
      "step": 24050
    },
    {
      "epoch": 1.6826083920966277,
      "grad_norm": 0.08181091398000717,
      "learning_rate": 3.6869501807518656e-05,
      "loss": 0.0203,
      "step": 24100
    },
    {
      "epoch": 1.6860992808769113,
      "grad_norm": 0.04160789027810097,
      "learning_rate": 3.6830713853505656e-05,
      "loss": 0.0165,
      "step": 24150
    },
    {
      "epoch": 1.6895901696571949,
      "grad_norm": 0.5199412107467651,
      "learning_rate": 3.6791925899492656e-05,
      "loss": 0.0331,
      "step": 24200
    },
    {
      "epoch": 1.6930810584374782,
      "grad_norm": 0.016219686716794968,
      "learning_rate": 3.6753137945479657e-05,
      "loss": 0.0367,
      "step": 24250
    },
    {
      "epoch": 1.6965719472177616,
      "grad_norm": 0.06320388615131378,
      "learning_rate": 3.671434999146665e-05,
      "loss": 0.0172,
      "step": 24300
    },
    {
      "epoch": 1.700062835998045,
      "grad_norm": 0.04016570374369621,
      "learning_rate": 3.667556203745365e-05,
      "loss": 0.0104,
      "step": 24350
    },
    {
      "epoch": 1.7035537247783286,
      "grad_norm": 0.7291899919509888,
      "learning_rate": 3.663677408344065e-05,
      "loss": 0.0387,
      "step": 24400
    },
    {
      "epoch": 1.7070446135586121,
      "grad_norm": 0.44951528310775757,
      "learning_rate": 3.659798612942765e-05,
      "loss": 0.0234,
      "step": 24450
    },
    {
      "epoch": 1.7105355023388955,
      "grad_norm": 1.9208860397338867,
      "learning_rate": 3.655919817541464e-05,
      "loss": 0.0194,
      "step": 24500
    },
    {
      "epoch": 1.7140263911191789,
      "grad_norm": 0.12419503182172775,
      "learning_rate": 3.6520410221401643e-05,
      "loss": 0.0207,
      "step": 24550
    },
    {
      "epoch": 1.7175172798994625,
      "grad_norm": 0.03292817994952202,
      "learning_rate": 3.6481622267388644e-05,
      "loss": 0.0361,
      "step": 24600
    },
    {
      "epoch": 1.7210081686797458,
      "grad_norm": 3.2311935424804688,
      "learning_rate": 3.644283431337564e-05,
      "loss": 0.0169,
      "step": 24650
    },
    {
      "epoch": 1.7244990574600294,
      "grad_norm": 1.065648078918457,
      "learning_rate": 3.640404635936264e-05,
      "loss": 0.0237,
      "step": 24700
    },
    {
      "epoch": 1.7279899462403128,
      "grad_norm": 0.02137444168329239,
      "learning_rate": 3.636525840534964e-05,
      "loss": 0.0167,
      "step": 24750
    },
    {
      "epoch": 1.7314808350205961,
      "grad_norm": 1.4355806112289429,
      "learning_rate": 3.632647045133664e-05,
      "loss": 0.0215,
      "step": 24800
    },
    {
      "epoch": 1.7349717238008797,
      "grad_norm": 4.564198970794678,
      "learning_rate": 3.628768249732363e-05,
      "loss": 0.0226,
      "step": 24850
    },
    {
      "epoch": 1.7384626125811633,
      "grad_norm": 0.08031658828258514,
      "learning_rate": 3.624889454331063e-05,
      "loss": 0.0282,
      "step": 24900
    },
    {
      "epoch": 1.7419535013614467,
      "grad_norm": 1.2107548713684082,
      "learning_rate": 3.621010658929763e-05,
      "loss": 0.0411,
      "step": 24950
    },
    {
      "epoch": 1.74544439014173,
      "grad_norm": 0.18569612503051758,
      "learning_rate": 3.617131863528463e-05,
      "loss": 0.0165,
      "step": 25000
    },
    {
      "epoch": 1.74544439014173,
      "eval_f1": 0.8931159420289855,
      "eval_loss": 0.2845694422721863,
      "eval_runtime": 24.4848,
      "eval_samples_per_second": 149.48,
      "eval_steps_per_second": 4.697,
      "step": 25000
    },
    {
      "epoch": 1.7489352789220134,
      "grad_norm": 1.5806281566619873,
      "learning_rate": 3.6132530681271624e-05,
      "loss": 0.023,
      "step": 25050
    },
    {
      "epoch": 1.752426167702297,
      "grad_norm": 0.054229360073804855,
      "learning_rate": 3.6093742727258624e-05,
      "loss": 0.0202,
      "step": 25100
    },
    {
      "epoch": 1.7559170564825806,
      "grad_norm": 0.20487448573112488,
      "learning_rate": 3.6054954773245624e-05,
      "loss": 0.0389,
      "step": 25150
    },
    {
      "epoch": 1.759407945262864,
      "grad_norm": 0.7431159615516663,
      "learning_rate": 3.6016166819232624e-05,
      "loss": 0.0119,
      "step": 25200
    },
    {
      "epoch": 1.7628988340431473,
      "grad_norm": 0.22739042341709137,
      "learning_rate": 3.597737886521962e-05,
      "loss": 0.0383,
      "step": 25250
    },
    {
      "epoch": 1.7663897228234309,
      "grad_norm": 0.023175235837697983,
      "learning_rate": 3.593859091120662e-05,
      "loss": 0.0337,
      "step": 25300
    },
    {
      "epoch": 1.7698806116037145,
      "grad_norm": 0.03510722145438194,
      "learning_rate": 3.589980295719362e-05,
      "loss": 0.0225,
      "step": 25350
    },
    {
      "epoch": 1.7733715003839978,
      "grad_norm": 0.03466213867068291,
      "learning_rate": 3.586101500318062e-05,
      "loss": 0.0267,
      "step": 25400
    },
    {
      "epoch": 1.7768623891642812,
      "grad_norm": 0.23764640092849731,
      "learning_rate": 3.582222704916761e-05,
      "loss": 0.0231,
      "step": 25450
    },
    {
      "epoch": 1.7803532779445645,
      "grad_norm": 5.474936008453369,
      "learning_rate": 3.578343909515461e-05,
      "loss": 0.0276,
      "step": 25500
    },
    {
      "epoch": 1.7838441667248481,
      "grad_norm": 0.040409285575151443,
      "learning_rate": 3.574465114114161e-05,
      "loss": 0.0203,
      "step": 25550
    },
    {
      "epoch": 1.7873350555051317,
      "grad_norm": 0.04732607677578926,
      "learning_rate": 3.5705863187128604e-05,
      "loss": 0.0233,
      "step": 25600
    },
    {
      "epoch": 1.790825944285415,
      "grad_norm": 1.5996266603469849,
      "learning_rate": 3.566785099219587e-05,
      "loss": 0.0298,
      "step": 25650
    },
    {
      "epoch": 1.7943168330656984,
      "grad_norm": 0.021670788526535034,
      "learning_rate": 3.562906303818286e-05,
      "loss": 0.0258,
      "step": 25700
    },
    {
      "epoch": 1.797807721845982,
      "grad_norm": 0.11140725761651993,
      "learning_rate": 3.559027508416986e-05,
      "loss": 0.0253,
      "step": 25750
    },
    {
      "epoch": 1.8012986106262654,
      "grad_norm": 0.19464978575706482,
      "learning_rate": 3.5551487130156856e-05,
      "loss": 0.0394,
      "step": 25800
    },
    {
      "epoch": 1.804789499406549,
      "grad_norm": 0.0491335391998291,
      "learning_rate": 3.551269917614386e-05,
      "loss": 0.0254,
      "step": 25850
    },
    {
      "epoch": 1.8082803881868323,
      "grad_norm": 0.07059049606323242,
      "learning_rate": 3.5473911222130856e-05,
      "loss": 0.0111,
      "step": 25900
    },
    {
      "epoch": 1.8117712769671157,
      "grad_norm": 2.9037585258483887,
      "learning_rate": 3.5435123268117856e-05,
      "loss": 0.0287,
      "step": 25950
    },
    {
      "epoch": 1.8152621657473993,
      "grad_norm": 0.014848013408482075,
      "learning_rate": 3.5396335314104856e-05,
      "loss": 0.0068,
      "step": 26000
    },
    {
      "epoch": 1.8152621657473993,
      "eval_f1": 0.9202556265629341,
      "eval_loss": 0.22660647332668304,
      "eval_runtime": 18.1122,
      "eval_samples_per_second": 202.073,
      "eval_steps_per_second": 6.349,
      "step": 26000
    },
    {
      "epoch": 1.8187530545276829,
      "grad_norm": 1.870595932006836,
      "learning_rate": 3.535754736009185e-05,
      "loss": 0.034,
      "step": 26050
    },
    {
      "epoch": 1.8222439433079662,
      "grad_norm": 0.03695142641663551,
      "learning_rate": 3.531875940607885e-05,
      "loss": 0.0334,
      "step": 26100
    },
    {
      "epoch": 1.8257348320882496,
      "grad_norm": 0.5613416433334351,
      "learning_rate": 3.527997145206585e-05,
      "loss": 0.0274,
      "step": 26150
    },
    {
      "epoch": 1.829225720868533,
      "grad_norm": 2.8512015342712402,
      "learning_rate": 3.524118349805285e-05,
      "loss": 0.0262,
      "step": 26200
    },
    {
      "epoch": 1.8327166096488166,
      "grad_norm": 0.012033411301672459,
      "learning_rate": 3.520239554403984e-05,
      "loss": 0.02,
      "step": 26250
    },
    {
      "epoch": 1.8362074984291001,
      "grad_norm": 0.1745612472295761,
      "learning_rate": 3.516360759002684e-05,
      "loss": 0.0098,
      "step": 26300
    },
    {
      "epoch": 1.8396983872093835,
      "grad_norm": 0.3587130308151245,
      "learning_rate": 3.5124819636013836e-05,
      "loss": 0.0194,
      "step": 26350
    },
    {
      "epoch": 1.8431892759896669,
      "grad_norm": 1.0576293468475342,
      "learning_rate": 3.508603168200084e-05,
      "loss": 0.0266,
      "step": 26400
    },
    {
      "epoch": 1.8466801647699504,
      "grad_norm": 0.13543370366096497,
      "learning_rate": 3.5047243727987836e-05,
      "loss": 0.0206,
      "step": 26450
    },
    {
      "epoch": 1.850171053550234,
      "grad_norm": 0.039760392159223557,
      "learning_rate": 3.50092315330551e-05,
      "loss": 0.0271,
      "step": 26500
    },
    {
      "epoch": 1.8536619423305174,
      "grad_norm": 0.8882807493209839,
      "learning_rate": 3.4970443579042095e-05,
      "loss": 0.0298,
      "step": 26550
    },
    {
      "epoch": 1.8571528311108008,
      "grad_norm": 0.032096270471811295,
      "learning_rate": 3.4931655625029095e-05,
      "loss": 0.0096,
      "step": 26600
    },
    {
      "epoch": 1.8606437198910841,
      "grad_norm": 0.19736655056476593,
      "learning_rate": 3.489286767101609e-05,
      "loss": 0.0395,
      "step": 26650
    },
    {
      "epoch": 1.8641346086713677,
      "grad_norm": 0.04327971488237381,
      "learning_rate": 3.485407971700309e-05,
      "loss": 0.0226,
      "step": 26700
    },
    {
      "epoch": 1.8676254974516513,
      "grad_norm": 0.00801787618547678,
      "learning_rate": 3.481529176299009e-05,
      "loss": 0.0133,
      "step": 26750
    },
    {
      "epoch": 1.8711163862319347,
      "grad_norm": 6.338688850402832,
      "learning_rate": 3.477650380897708e-05,
      "loss": 0.0156,
      "step": 26800
    },
    {
      "epoch": 1.874607275012218,
      "grad_norm": 0.03296707198023796,
      "learning_rate": 3.473771585496408e-05,
      "loss": 0.0333,
      "step": 26850
    },
    {
      "epoch": 1.8780981637925016,
      "grad_norm": 0.0316268652677536,
      "learning_rate": 3.469892790095108e-05,
      "loss": 0.0156,
      "step": 26900
    },
    {
      "epoch": 1.8815890525727852,
      "grad_norm": 3.483025312423706,
      "learning_rate": 3.466013994693808e-05,
      "loss": 0.0242,
      "step": 26950
    },
    {
      "epoch": 1.8850799413530686,
      "grad_norm": 0.11344876885414124,
      "learning_rate": 3.4621351992925075e-05,
      "loss": 0.0252,
      "step": 27000
    },
    {
      "epoch": 1.8850799413530686,
      "eval_f1": 0.8898203592814371,
      "eval_loss": 0.2792615592479706,
      "eval_runtime": 383.384,
      "eval_samples_per_second": 9.547,
      "eval_steps_per_second": 0.3,
      "step": 27000
    },
    {
      "epoch": 1.888570830133352,
      "grad_norm": 0.025170810520648956,
      "learning_rate": 3.458256403891208e-05,
      "loss": 0.0292,
      "step": 27050
    },
    {
      "epoch": 1.8920617189136353,
      "grad_norm": 0.12406772375106812,
      "learning_rate": 3.4543776084899075e-05,
      "loss": 0.023,
      "step": 27100
    },
    {
      "epoch": 1.8955526076939189,
      "grad_norm": 0.05151906982064247,
      "learning_rate": 3.4504988130886075e-05,
      "loss": 0.0199,
      "step": 27150
    },
    {
      "epoch": 1.8990434964742025,
      "grad_norm": 0.07848247140645981,
      "learning_rate": 3.446620017687307e-05,
      "loss": 0.02,
      "step": 27200
    },
    {
      "epoch": 1.9025343852544858,
      "grad_norm": 0.01637980341911316,
      "learning_rate": 3.4427412222860075e-05,
      "loss": 0.0175,
      "step": 27250
    },
    {
      "epoch": 1.9060252740347692,
      "grad_norm": 0.6902241706848145,
      "learning_rate": 3.438862426884707e-05,
      "loss": 0.0387,
      "step": 27300
    },
    {
      "epoch": 1.9095161628150528,
      "grad_norm": 1.2976561784744263,
      "learning_rate": 3.434983631483407e-05,
      "loss": 0.0266,
      "step": 27350
    },
    {
      "epoch": 1.9130070515953361,
      "grad_norm": 0.7005116939544678,
      "learning_rate": 3.431104836082106e-05,
      "loss": 0.0173,
      "step": 27400
    },
    {
      "epoch": 1.9164979403756197,
      "grad_norm": 1.2808347940444946,
      "learning_rate": 3.427226040680807e-05,
      "loss": 0.036,
      "step": 27450
    },
    {
      "epoch": 1.919988829155903,
      "grad_norm": 0.03168126195669174,
      "learning_rate": 3.423347245279506e-05,
      "loss": 0.0343,
      "step": 27500
    },
    {
      "epoch": 1.9234797179361864,
      "grad_norm": 0.12667793035507202,
      "learning_rate": 3.419468449878206e-05,
      "loss": 0.0239,
      "step": 27550
    },
    {
      "epoch": 1.92697060671647,
      "grad_norm": 0.9971345663070679,
      "learning_rate": 3.4155896544769055e-05,
      "loss": 0.0198,
      "step": 27600
    },
    {
      "epoch": 1.9304614954967536,
      "grad_norm": 0.03455730155110359,
      "learning_rate": 3.4117108590756056e-05,
      "loss": 0.038,
      "step": 27650
    },
    {
      "epoch": 1.933952384277037,
      "grad_norm": 0.5763664245605469,
      "learning_rate": 3.4078320636743056e-05,
      "loss": 0.0328,
      "step": 27700
    },
    {
      "epoch": 1.9374432730573203,
      "grad_norm": 0.015370190143585205,
      "learning_rate": 3.403953268273005e-05,
      "loss": 0.0124,
      "step": 27750
    },
    {
      "epoch": 1.9409341618376037,
      "grad_norm": 0.5836357474327087,
      "learning_rate": 3.400074472871705e-05,
      "loss": 0.0227,
      "step": 27800
    },
    {
      "epoch": 1.9444250506178873,
      "grad_norm": 0.015789790078997612,
      "learning_rate": 3.396195677470405e-05,
      "loss": 0.0131,
      "step": 27850
    },
    {
      "epoch": 1.9479159393981709,
      "grad_norm": 0.02878633514046669,
      "learning_rate": 3.392316882069105e-05,
      "loss": 0.0317,
      "step": 27900
    },
    {
      "epoch": 1.9514068281784542,
      "grad_norm": 0.2344045788049698,
      "learning_rate": 3.388438086667804e-05,
      "loss": 0.0178,
      "step": 27950
    },
    {
      "epoch": 1.9548977169587376,
      "grad_norm": 0.2111712098121643,
      "learning_rate": 3.384559291266504e-05,
      "loss": 0.018,
      "step": 28000
    },
    {
      "epoch": 1.9548977169587376,
      "eval_f1": 0.9149679673849738,
      "eval_loss": 0.23677824437618256,
      "eval_runtime": 27.4697,
      "eval_samples_per_second": 133.237,
      "eval_steps_per_second": 4.186,
      "step": 28000
    },
    {
      "epoch": 1.9583886057390212,
      "grad_norm": 2.1588897705078125,
      "learning_rate": 3.380680495865204e-05,
      "loss": 0.0354,
      "step": 28050
    },
    {
      "epoch": 1.9618794945193048,
      "grad_norm": 0.13064005970954895,
      "learning_rate": 3.376801700463904e-05,
      "loss": 0.0231,
      "step": 28100
    },
    {
      "epoch": 1.9653703832995881,
      "grad_norm": 0.4351823329925537,
      "learning_rate": 3.3729229050626036e-05,
      "loss": 0.0381,
      "step": 28150
    },
    {
      "epoch": 1.9688612720798715,
      "grad_norm": 0.03735004737973213,
      "learning_rate": 3.369044109661304e-05,
      "loss": 0.0454,
      "step": 28200
    },
    {
      "epoch": 1.9723521608601549,
      "grad_norm": 2.105356454849243,
      "learning_rate": 3.3651653142600036e-05,
      "loss": 0.0196,
      "step": 28250
    },
    {
      "epoch": 1.9758430496404384,
      "grad_norm": 0.01047208160161972,
      "learning_rate": 3.3612865188587036e-05,
      "loss": 0.0262,
      "step": 28300
    },
    {
      "epoch": 1.979333938420722,
      "grad_norm": 0.2573576271533966,
      "learning_rate": 3.357407723457403e-05,
      "loss": 0.0283,
      "step": 28350
    },
    {
      "epoch": 1.9828248272010054,
      "grad_norm": 0.03263131529092789,
      "learning_rate": 3.3535289280561036e-05,
      "loss": 0.0222,
      "step": 28400
    },
    {
      "epoch": 1.9863157159812888,
      "grad_norm": 1.0335652828216553,
      "learning_rate": 3.349650132654803e-05,
      "loss": 0.0137,
      "step": 28450
    },
    {
      "epoch": 1.9898066047615723,
      "grad_norm": 0.029950108379125595,
      "learning_rate": 3.345771337253503e-05,
      "loss": 0.0257,
      "step": 28500
    },
    {
      "epoch": 1.9932974935418557,
      "grad_norm": 0.9945626854896545,
      "learning_rate": 3.341892541852202e-05,
      "loss": 0.0156,
      "step": 28550
    },
    {
      "epoch": 1.9967883823221393,
      "grad_norm": 0.18837442994117737,
      "learning_rate": 3.338013746450902e-05,
      "loss": 0.015,
      "step": 28600
    },
    {
      "epoch": 2.0002792711024227,
      "grad_norm": 0.023262226954102516,
      "learning_rate": 3.334134951049602e-05,
      "loss": 0.0184,
      "step": 28650
    },
    {
      "epoch": 2.003770159882706,
      "grad_norm": 0.7680675387382507,
      "learning_rate": 3.3302561556483016e-05,
      "loss": 0.0272,
      "step": 28700
    },
    {
      "epoch": 2.0072610486629894,
      "grad_norm": 0.5894890427589417,
      "learning_rate": 3.3263773602470017e-05,
      "loss": 0.036,
      "step": 28750
    },
    {
      "epoch": 2.010751937443273,
      "grad_norm": 0.16585896909236908,
      "learning_rate": 3.3224985648457017e-05,
      "loss": 0.0113,
      "step": 28800
    },
    {
      "epoch": 2.0142428262235565,
      "grad_norm": 0.09970059245824814,
      "learning_rate": 3.318619769444402e-05,
      "loss": 0.024,
      "step": 28850
    },
    {
      "epoch": 2.01773371500384,
      "grad_norm": 3.7176826000213623,
      "learning_rate": 3.3148185499511275e-05,
      "loss": 0.0298,
      "step": 28900
    },
    {
      "epoch": 2.0212246037841233,
      "grad_norm": 0.08712600916624069,
      "learning_rate": 3.3109397545498275e-05,
      "loss": 0.0169,
      "step": 28950
    },
    {
      "epoch": 2.024715492564407,
      "grad_norm": 0.592137336730957,
      "learning_rate": 3.307060959148527e-05,
      "loss": 0.0185,
      "step": 29000
    },
    {
      "epoch": 2.024715492564407,
      "eval_f1": 0.9016686531585221,
      "eval_loss": 0.27124351263046265,
      "eval_runtime": 24.7503,
      "eval_samples_per_second": 147.877,
      "eval_steps_per_second": 4.646,
      "step": 29000
    },
    {
      "epoch": 2.0282063813446904,
      "grad_norm": 41.94601058959961,
      "learning_rate": 3.303182163747227e-05,
      "loss": 0.0235,
      "step": 29050
    },
    {
      "epoch": 2.031697270124974,
      "grad_norm": 0.031210171058773994,
      "learning_rate": 3.299303368345927e-05,
      "loss": 0.0282,
      "step": 29100
    },
    {
      "epoch": 2.035188158905257,
      "grad_norm": 0.10925783962011337,
      "learning_rate": 3.295424572944626e-05,
      "loss": 0.0321,
      "step": 29150
    },
    {
      "epoch": 2.0386790476855405,
      "grad_norm": 3.2144949436187744,
      "learning_rate": 3.291545777543326e-05,
      "loss": 0.036,
      "step": 29200
    },
    {
      "epoch": 2.0421699364658243,
      "grad_norm": 0.024584254249930382,
      "learning_rate": 3.287666982142026e-05,
      "loss": 0.0294,
      "step": 29250
    },
    {
      "epoch": 2.0456608252461077,
      "grad_norm": 4.2165656089782715,
      "learning_rate": 3.283788186740726e-05,
      "loss": 0.025,
      "step": 29300
    },
    {
      "epoch": 2.049151714026391,
      "grad_norm": 11.206308364868164,
      "learning_rate": 3.2799093913394255e-05,
      "loss": 0.0176,
      "step": 29350
    },
    {
      "epoch": 2.0526426028066744,
      "grad_norm": 1.3034924268722534,
      "learning_rate": 3.2760305959381255e-05,
      "loss": 0.0274,
      "step": 29400
    },
    {
      "epoch": 2.0561334915869582,
      "grad_norm": 0.08774253726005554,
      "learning_rate": 3.2721518005368255e-05,
      "loss": 0.0205,
      "step": 29450
    },
    {
      "epoch": 2.0596243803672416,
      "grad_norm": 1.1402833461761475,
      "learning_rate": 3.2682730051355255e-05,
      "loss": 0.0234,
      "step": 29500
    },
    {
      "epoch": 2.063115269147525,
      "grad_norm": 0.5319716930389404,
      "learning_rate": 3.264394209734225e-05,
      "loss": 0.0266,
      "step": 29550
    },
    {
      "epoch": 2.0666061579278083,
      "grad_norm": 0.014304044656455517,
      "learning_rate": 3.260515414332925e-05,
      "loss": 0.0233,
      "step": 29600
    },
    {
      "epoch": 2.0700970467080917,
      "grad_norm": 0.031523484736680984,
      "learning_rate": 3.256636618931625e-05,
      "loss": 0.0285,
      "step": 29650
    },
    {
      "epoch": 2.0735879354883755,
      "grad_norm": 2.613784074783325,
      "learning_rate": 3.252757823530325e-05,
      "loss": 0.018,
      "step": 29700
    },
    {
      "epoch": 2.077078824268659,
      "grad_norm": 5.857016563415527,
      "learning_rate": 3.248879028129024e-05,
      "loss": 0.0272,
      "step": 29750
    },
    {
      "epoch": 2.0805697130489422,
      "grad_norm": 3.223562002182007,
      "learning_rate": 3.245000232727724e-05,
      "loss": 0.0249,
      "step": 29800
    },
    {
      "epoch": 2.0840606018292256,
      "grad_norm": 0.13457106053829193,
      "learning_rate": 3.241121437326424e-05,
      "loss": 0.0299,
      "step": 29850
    },
    {
      "epoch": 2.087551490609509,
      "grad_norm": 0.08261656016111374,
      "learning_rate": 3.237242641925124e-05,
      "loss": 0.0145,
      "step": 29900
    },
    {
      "epoch": 2.0910423793897928,
      "grad_norm": 0.022206531837582588,
      "learning_rate": 3.2333638465238236e-05,
      "loss": 0.0297,
      "step": 29950
    },
    {
      "epoch": 2.094533268170076,
      "grad_norm": 0.23456662893295288,
      "learning_rate": 3.2294850511225236e-05,
      "loss": 0.028,
      "step": 30000
    },
    {
      "epoch": 2.094533268170076,
      "eval_f1": 0.90849478390462,
      "eval_loss": 0.22531723976135254,
      "eval_runtime": 26.1579,
      "eval_samples_per_second": 139.92,
      "eval_steps_per_second": 4.396,
      "step": 30000
    },
    {
      "epoch": 2.0980241569503595,
      "grad_norm": 0.04046647995710373,
      "learning_rate": 3.2256062557212236e-05,
      "loss": 0.0219,
      "step": 30050
    },
    {
      "epoch": 2.101515045730643,
      "grad_norm": 0.1782691329717636,
      "learning_rate": 3.221727460319923e-05,
      "loss": 0.0185,
      "step": 30100
    },
    {
      "epoch": 2.1050059345109267,
      "grad_norm": 0.05876058712601662,
      "learning_rate": 3.217848664918623e-05,
      "loss": 0.025,
      "step": 30150
    },
    {
      "epoch": 2.10849682329121,
      "grad_norm": 0.05100909247994423,
      "learning_rate": 3.213969869517323e-05,
      "loss": 0.0353,
      "step": 30200
    },
    {
      "epoch": 2.1119877120714934,
      "grad_norm": 0.1494356095790863,
      "learning_rate": 3.210091074116023e-05,
      "loss": 0.0216,
      "step": 30250
    },
    {
      "epoch": 2.1154786008517767,
      "grad_norm": 0.019559986889362335,
      "learning_rate": 3.206212278714722e-05,
      "loss": 0.0179,
      "step": 30300
    },
    {
      "epoch": 2.11896948963206,
      "grad_norm": 0.009545271284878254,
      "learning_rate": 3.202333483313422e-05,
      "loss": 0.0209,
      "step": 30350
    },
    {
      "epoch": 2.122460378412344,
      "grad_norm": 0.016645213589072227,
      "learning_rate": 3.198454687912122e-05,
      "loss": 0.0328,
      "step": 30400
    },
    {
      "epoch": 2.1259512671926273,
      "grad_norm": 0.05545541271567345,
      "learning_rate": 3.194575892510822e-05,
      "loss": 0.0161,
      "step": 30450
    },
    {
      "epoch": 2.1294421559729106,
      "grad_norm": 4.6828789710998535,
      "learning_rate": 3.1906970971095216e-05,
      "loss": 0.0248,
      "step": 30500
    },
    {
      "epoch": 2.132933044753194,
      "grad_norm": 0.03040727972984314,
      "learning_rate": 3.1868183017082216e-05,
      "loss": 0.0233,
      "step": 30550
    },
    {
      "epoch": 2.136423933533478,
      "grad_norm": 0.04015226662158966,
      "learning_rate": 3.1829395063069216e-05,
      "loss": 0.0187,
      "step": 30600
    },
    {
      "epoch": 2.139914822313761,
      "grad_norm": 0.44110485911369324,
      "learning_rate": 3.1790607109056216e-05,
      "loss": 0.0358,
      "step": 30650
    },
    {
      "epoch": 2.1434057110940445,
      "grad_norm": 0.03207771107554436,
      "learning_rate": 3.175181915504321e-05,
      "loss": 0.0207,
      "step": 30700
    },
    {
      "epoch": 2.146896599874328,
      "grad_norm": 0.02635408379137516,
      "learning_rate": 3.171303120103021e-05,
      "loss": 0.0244,
      "step": 30750
    },
    {
      "epoch": 2.1503874886546113,
      "grad_norm": 0.026559866964817047,
      "learning_rate": 3.167424324701721e-05,
      "loss": 0.0218,
      "step": 30800
    },
    {
      "epoch": 2.153878377434895,
      "grad_norm": 0.03618308901786804,
      "learning_rate": 3.163545529300421e-05,
      "loss": 0.0307,
      "step": 30850
    },
    {
      "epoch": 2.1573692662151784,
      "grad_norm": 0.021506665274500847,
      "learning_rate": 3.15966673389912e-05,
      "loss": 0.0181,
      "step": 30900
    },
    {
      "epoch": 2.160860154995462,
      "grad_norm": 0.19482529163360596,
      "learning_rate": 3.15578793849782e-05,
      "loss": 0.0187,
      "step": 30950
    },
    {
      "epoch": 2.164351043775745,
      "grad_norm": 0.6267532706260681,
      "learning_rate": 3.15190914309652e-05,
      "loss": 0.0231,
      "step": 31000
    },
    {
      "epoch": 2.164351043775745,
      "eval_f1": 0.9244186046511628,
      "eval_loss": 0.22860726714134216,
      "eval_runtime": 24.3238,
      "eval_samples_per_second": 150.47,
      "eval_steps_per_second": 4.728,
      "step": 31000
    },
    {
      "epoch": 2.167841932556029,
      "grad_norm": 0.10762369632720947,
      "learning_rate": 3.1480303476952197e-05,
      "loss": 0.0166,
      "step": 31050
    },
    {
      "epoch": 2.1713328213363123,
      "grad_norm": 0.02598682790994644,
      "learning_rate": 3.14415155229392e-05,
      "loss": 0.0337,
      "step": 31100
    },
    {
      "epoch": 2.1748237101165957,
      "grad_norm": 2.0279221534729004,
      "learning_rate": 3.14027275689262e-05,
      "loss": 0.0139,
      "step": 31150
    },
    {
      "epoch": 2.178314598896879,
      "grad_norm": 0.023457834497094154,
      "learning_rate": 3.13639396149132e-05,
      "loss": 0.0213,
      "step": 31200
    },
    {
      "epoch": 2.1818054876771624,
      "grad_norm": 1.7613838911056519,
      "learning_rate": 3.132515166090019e-05,
      "loss": 0.0308,
      "step": 31250
    },
    {
      "epoch": 2.1852963764574462,
      "grad_norm": 0.7117096185684204,
      "learning_rate": 3.128636370688719e-05,
      "loss": 0.0155,
      "step": 31300
    },
    {
      "epoch": 2.1887872652377296,
      "grad_norm": 0.17505602538585663,
      "learning_rate": 3.1247575752874183e-05,
      "loss": 0.0203,
      "step": 31350
    },
    {
      "epoch": 2.192278154018013,
      "grad_norm": 0.0648331269621849,
      "learning_rate": 3.120878779886119e-05,
      "loss": 0.0202,
      "step": 31400
    },
    {
      "epoch": 2.1957690427982963,
      "grad_norm": 0.6441143155097961,
      "learning_rate": 3.1169999844848184e-05,
      "loss": 0.0442,
      "step": 31450
    },
    {
      "epoch": 2.19925993157858,
      "grad_norm": 0.027606986463069916,
      "learning_rate": 3.1131211890835184e-05,
      "loss": 0.0161,
      "step": 31500
    },
    {
      "epoch": 2.2027508203588635,
      "grad_norm": 0.007126821670681238,
      "learning_rate": 3.109242393682218e-05,
      "loss": 0.0097,
      "step": 31550
    },
    {
      "epoch": 2.206241709139147,
      "grad_norm": 0.008397717960178852,
      "learning_rate": 3.1053635982809184e-05,
      "loss": 0.0293,
      "step": 31600
    },
    {
      "epoch": 2.20973259791943,
      "grad_norm": 0.03310353308916092,
      "learning_rate": 3.101484802879618e-05,
      "loss": 0.0314,
      "step": 31650
    },
    {
      "epoch": 2.2132234866997136,
      "grad_norm": 0.02405470609664917,
      "learning_rate": 3.097606007478318e-05,
      "loss": 0.0274,
      "step": 31700
    },
    {
      "epoch": 2.2167143754799974,
      "grad_norm": 0.1666310578584671,
      "learning_rate": 3.093727212077018e-05,
      "loss": 0.0215,
      "step": 31750
    },
    {
      "epoch": 2.2202052642602808,
      "grad_norm": 0.08723346889019012,
      "learning_rate": 3.089848416675718e-05,
      "loss": 0.0177,
      "step": 31800
    },
    {
      "epoch": 2.223696153040564,
      "grad_norm": 0.019003497436642647,
      "learning_rate": 3.085969621274417e-05,
      "loss": 0.0149,
      "step": 31850
    },
    {
      "epoch": 2.2271870418208475,
      "grad_norm": 0.010868572629988194,
      "learning_rate": 3.082090825873117e-05,
      "loss": 0.0138,
      "step": 31900
    },
    {
      "epoch": 2.230677930601131,
      "grad_norm": 0.08453696966171265,
      "learning_rate": 3.078212030471817e-05,
      "loss": 0.0166,
      "step": 31950
    },
    {
      "epoch": 2.2341688193814147,
      "grad_norm": 0.06810794770717621,
      "learning_rate": 3.0743332350705164e-05,
      "loss": 0.0241,
      "step": 32000
    },
    {
      "epoch": 2.2341688193814147,
      "eval_f1": 0.9090370370370371,
      "eval_loss": 0.25670725107192993,
      "eval_runtime": 23.7993,
      "eval_samples_per_second": 153.786,
      "eval_steps_per_second": 4.832,
      "step": 32000
    },
    {
      "epoch": 2.237659708161698,
      "grad_norm": 0.08504435420036316,
      "learning_rate": 3.0704544396692164e-05,
      "loss": 0.0206,
      "step": 32050
    },
    {
      "epoch": 2.2411505969419814,
      "grad_norm": 0.009788773022592068,
      "learning_rate": 3.0665756442679164e-05,
      "loss": 0.01,
      "step": 32100
    },
    {
      "epoch": 2.2446414857222647,
      "grad_norm": 0.05060356482863426,
      "learning_rate": 3.0626968488666164e-05,
      "loss": 0.0194,
      "step": 32150
    },
    {
      "epoch": 2.2481323745025485,
      "grad_norm": 0.020789574831724167,
      "learning_rate": 3.058818053465316e-05,
      "loss": 0.0193,
      "step": 32200
    },
    {
      "epoch": 2.251623263282832,
      "grad_norm": 0.09519587457180023,
      "learning_rate": 3.054939258064016e-05,
      "loss": 0.0388,
      "step": 32250
    },
    {
      "epoch": 2.2551141520631153,
      "grad_norm": 0.09373360872268677,
      "learning_rate": 3.0510604626627154e-05,
      "loss": 0.0221,
      "step": 32300
    },
    {
      "epoch": 2.2586050408433986,
      "grad_norm": 0.022543421015143394,
      "learning_rate": 3.0471816672614158e-05,
      "loss": 0.0162,
      "step": 32350
    },
    {
      "epoch": 2.262095929623682,
      "grad_norm": 2.8079066276550293,
      "learning_rate": 3.0433028718601155e-05,
      "loss": 0.0262,
      "step": 32400
    },
    {
      "epoch": 2.265586818403966,
      "grad_norm": 0.032409828156232834,
      "learning_rate": 3.039424076458815e-05,
      "loss": 0.0325,
      "step": 32450
    },
    {
      "epoch": 2.269077707184249,
      "grad_norm": 0.03421691432595253,
      "learning_rate": 3.0355452810575148e-05,
      "loss": 0.0294,
      "step": 32500
    },
    {
      "epoch": 2.2725685959645325,
      "grad_norm": 0.3292216956615448,
      "learning_rate": 3.0316664856562148e-05,
      "loss": 0.0175,
      "step": 32550
    },
    {
      "epoch": 2.276059484744816,
      "grad_norm": 0.031362514942884445,
      "learning_rate": 3.0277876902549145e-05,
      "loss": 0.0233,
      "step": 32600
    },
    {
      "epoch": 2.2795503735250993,
      "grad_norm": 0.15582652390003204,
      "learning_rate": 3.023908894853614e-05,
      "loss": 0.0159,
      "step": 32650
    },
    {
      "epoch": 2.283041262305383,
      "grad_norm": 0.011120837181806564,
      "learning_rate": 3.0200300994523138e-05,
      "loss": 0.0194,
      "step": 32700
    },
    {
      "epoch": 2.2865321510856664,
      "grad_norm": 0.05831030383706093,
      "learning_rate": 3.016151304051014e-05,
      "loss": 0.0326,
      "step": 32750
    },
    {
      "epoch": 2.29002303986595,
      "grad_norm": 0.1667061150074005,
      "learning_rate": 3.0123500845577403e-05,
      "loss": 0.0261,
      "step": 32800
    },
    {
      "epoch": 2.293513928646233,
      "grad_norm": 0.03689422830939293,
      "learning_rate": 3.00847128915644e-05,
      "loss": 0.0211,
      "step": 32850
    },
    {
      "epoch": 2.297004817426517,
      "grad_norm": 0.08821934461593628,
      "learning_rate": 3.0045924937551396e-05,
      "loss": 0.0196,
      "step": 32900
    },
    {
      "epoch": 2.3004957062068003,
      "grad_norm": 0.026601577177643776,
      "learning_rate": 3.0007136983538393e-05,
      "loss": 0.0227,
      "step": 32950
    },
    {
      "epoch": 2.3039865949870837,
      "grad_norm": 6.168493747711182,
      "learning_rate": 2.9968349029525393e-05,
      "loss": 0.0292,
      "step": 33000
    },
    {
      "epoch": 2.3039865949870837,
      "eval_f1": 0.9086595492289442,
      "eval_loss": 0.22055290639400482,
      "eval_runtime": 17.4165,
      "eval_samples_per_second": 210.146,
      "eval_steps_per_second": 6.603,
      "step": 33000
    },
    {
      "epoch": 2.307477483767367,
      "grad_norm": 0.011802727356553078,
      "learning_rate": 2.992956107551239e-05,
      "loss": 0.0168,
      "step": 33050
    },
    {
      "epoch": 2.3109683725476504,
      "grad_norm": 0.6425135135650635,
      "learning_rate": 2.9890773121499386e-05,
      "loss": 0.0198,
      "step": 33100
    },
    {
      "epoch": 2.3144592613279342,
      "grad_norm": 0.06205461546778679,
      "learning_rate": 2.9851985167486387e-05,
      "loss": 0.0203,
      "step": 33150
    },
    {
      "epoch": 2.3179501501082176,
      "grad_norm": 0.03974612429738045,
      "learning_rate": 2.9813197213473387e-05,
      "loss": 0.0186,
      "step": 33200
    },
    {
      "epoch": 2.321441038888501,
      "grad_norm": 0.2282327115535736,
      "learning_rate": 2.9774409259460383e-05,
      "loss": 0.0312,
      "step": 33250
    },
    {
      "epoch": 2.3249319276687843,
      "grad_norm": 0.06931372731924057,
      "learning_rate": 2.973562130544738e-05,
      "loss": 0.0205,
      "step": 33300
    },
    {
      "epoch": 2.328422816449068,
      "grad_norm": 0.29048246145248413,
      "learning_rate": 2.9696833351434377e-05,
      "loss": 0.0207,
      "step": 33350
    },
    {
      "epoch": 2.3319137052293515,
      "grad_norm": 0.0800575315952301,
      "learning_rate": 2.965804539742138e-05,
      "loss": 0.0319,
      "step": 33400
    },
    {
      "epoch": 2.335404594009635,
      "grad_norm": 2.8189268112182617,
      "learning_rate": 2.9619257443408377e-05,
      "loss": 0.0298,
      "step": 33450
    },
    {
      "epoch": 2.338895482789918,
      "grad_norm": 0.019050445407629013,
      "learning_rate": 2.9580469489395374e-05,
      "loss": 0.0148,
      "step": 33500
    },
    {
      "epoch": 2.3423863715702016,
      "grad_norm": 2.150393009185791,
      "learning_rate": 2.954168153538237e-05,
      "loss": 0.0268,
      "step": 33550
    },
    {
      "epoch": 2.3458772603504854,
      "grad_norm": 0.049240246415138245,
      "learning_rate": 2.9502893581369374e-05,
      "loss": 0.0295,
      "step": 33600
    },
    {
      "epoch": 2.3493681491307687,
      "grad_norm": 0.012683813460171223,
      "learning_rate": 2.946410562735637e-05,
      "loss": 0.022,
      "step": 33650
    },
    {
      "epoch": 2.352859037911052,
      "grad_norm": 0.7272828221321106,
      "learning_rate": 2.9425317673343367e-05,
      "loss": 0.0178,
      "step": 33700
    },
    {
      "epoch": 2.3563499266913355,
      "grad_norm": 0.005036385729908943,
      "learning_rate": 2.9386529719330364e-05,
      "loss": 0.0133,
      "step": 33750
    },
    {
      "epoch": 2.3598408154716193,
      "grad_norm": 2.0128109455108643,
      "learning_rate": 2.9347741765317367e-05,
      "loss": 0.0333,
      "step": 33800
    },
    {
      "epoch": 2.3633317042519026,
      "grad_norm": 0.5187195539474487,
      "learning_rate": 2.9308953811304364e-05,
      "loss": 0.0217,
      "step": 33850
    },
    {
      "epoch": 2.366822593032186,
      "grad_norm": 0.034500740468502045,
      "learning_rate": 2.927016585729136e-05,
      "loss": 0.0232,
      "step": 33900
    },
    {
      "epoch": 2.3703134818124694,
      "grad_norm": 2.572962522506714,
      "learning_rate": 2.923137790327836e-05,
      "loss": 0.0157,
      "step": 33950
    },
    {
      "epoch": 2.3738043705927527,
      "grad_norm": 0.02022615261375904,
      "learning_rate": 2.9192589949265357e-05,
      "loss": 0.0269,
      "step": 34000
    },
    {
      "epoch": 2.3738043705927527,
      "eval_f1": 0.9133016627078385,
      "eval_loss": 0.217937171459198,
      "eval_runtime": 17.3487,
      "eval_samples_per_second": 210.967,
      "eval_steps_per_second": 6.629,
      "step": 34000
    },
    {
      "epoch": 2.3772952593730365,
      "grad_norm": 0.05787201225757599,
      "learning_rate": 2.9153801995252354e-05,
      "loss": 0.0187,
      "step": 34050
    },
    {
      "epoch": 2.38078614815332,
      "grad_norm": 0.08669839054346085,
      "learning_rate": 2.9115014041239354e-05,
      "loss": 0.0303,
      "step": 34100
    },
    {
      "epoch": 2.3842770369336033,
      "grad_norm": 3.055640697479248,
      "learning_rate": 2.9076226087226354e-05,
      "loss": 0.0314,
      "step": 34150
    },
    {
      "epoch": 2.3877679257138866,
      "grad_norm": 3.0248279571533203,
      "learning_rate": 2.903743813321335e-05,
      "loss": 0.0305,
      "step": 34200
    },
    {
      "epoch": 2.3912588144941704,
      "grad_norm": 0.35696566104888916,
      "learning_rate": 2.8998650179200348e-05,
      "loss": 0.0228,
      "step": 34250
    },
    {
      "epoch": 2.394749703274454,
      "grad_norm": 0.03579020872712135,
      "learning_rate": 2.8959862225187344e-05,
      "loss": 0.0229,
      "step": 34300
    },
    {
      "epoch": 2.398240592054737,
      "grad_norm": 0.5238425135612488,
      "learning_rate": 2.8921074271174348e-05,
      "loss": 0.0252,
      "step": 34350
    },
    {
      "epoch": 2.4017314808350205,
      "grad_norm": 0.07448386400938034,
      "learning_rate": 2.8882286317161344e-05,
      "loss": 0.0127,
      "step": 34400
    },
    {
      "epoch": 2.405222369615304,
      "grad_norm": 5.516473293304443,
      "learning_rate": 2.884349836314834e-05,
      "loss": 0.0286,
      "step": 34450
    },
    {
      "epoch": 2.4087132583955877,
      "grad_norm": 0.025009727105498314,
      "learning_rate": 2.8804710409135338e-05,
      "loss": 0.0161,
      "step": 34500
    },
    {
      "epoch": 2.412204147175871,
      "grad_norm": 1.4443304538726807,
      "learning_rate": 2.876592245512234e-05,
      "loss": 0.0327,
      "step": 34550
    },
    {
      "epoch": 2.4156950359561544,
      "grad_norm": 0.01991642825305462,
      "learning_rate": 2.8727134501109338e-05,
      "loss": 0.0251,
      "step": 34600
    },
    {
      "epoch": 2.419185924736438,
      "grad_norm": 0.244741290807724,
      "learning_rate": 2.8688346547096335e-05,
      "loss": 0.0218,
      "step": 34650
    },
    {
      "epoch": 2.4226768135167216,
      "grad_norm": 0.022914182394742966,
      "learning_rate": 2.864955859308333e-05,
      "loss": 0.0231,
      "step": 34700
    },
    {
      "epoch": 2.426167702297005,
      "grad_norm": 0.13576367497444153,
      "learning_rate": 2.8610770639070335e-05,
      "loss": 0.0207,
      "step": 34750
    },
    {
      "epoch": 2.4296585910772883,
      "grad_norm": 0.028347410261631012,
      "learning_rate": 2.857198268505733e-05,
      "loss": 0.0199,
      "step": 34800
    },
    {
      "epoch": 2.4331494798575717,
      "grad_norm": 0.04504719376564026,
      "learning_rate": 2.8533194731044328e-05,
      "loss": 0.0165,
      "step": 34850
    },
    {
      "epoch": 2.436640368637855,
      "grad_norm": 1.0004245042800903,
      "learning_rate": 2.8494406777031325e-05,
      "loss": 0.0178,
      "step": 34900
    },
    {
      "epoch": 2.4401312574181384,
      "grad_norm": 0.050408054143190384,
      "learning_rate": 2.8455618823018325e-05,
      "loss": 0.0224,
      "step": 34950
    },
    {
      "epoch": 2.443622146198422,
      "grad_norm": 0.011280260048806667,
      "learning_rate": 2.841683086900532e-05,
      "loss": 0.0157,
      "step": 35000
    },
    {
      "epoch": 2.443622146198422,
      "eval_f1": 0.9306651634723788,
      "eval_loss": 0.19316384196281433,
      "eval_runtime": 17.1785,
      "eval_samples_per_second": 213.056,
      "eval_steps_per_second": 6.694,
      "step": 35000
    },
    {
      "epoch": 2.4471130349787056,
      "grad_norm": 2.7738091945648193,
      "learning_rate": 2.837804291499232e-05,
      "loss": 0.0164,
      "step": 35050
    },
    {
      "epoch": 2.450603923758989,
      "grad_norm": 0.28194230794906616,
      "learning_rate": 2.834003072005958e-05,
      "loss": 0.0413,
      "step": 35100
    },
    {
      "epoch": 2.4540948125392723,
      "grad_norm": 0.0410354882478714,
      "learning_rate": 2.8301242766046576e-05,
      "loss": 0.0217,
      "step": 35150
    },
    {
      "epoch": 2.457585701319556,
      "grad_norm": 0.07318081706762314,
      "learning_rate": 2.826245481203358e-05,
      "loss": 0.0383,
      "step": 35200
    },
    {
      "epoch": 2.4610765900998395,
      "grad_norm": 0.08617233484983444,
      "learning_rate": 2.8223666858020576e-05,
      "loss": 0.0301,
      "step": 35250
    },
    {
      "epoch": 2.464567478880123,
      "grad_norm": 2.5360500812530518,
      "learning_rate": 2.8184878904007573e-05,
      "loss": 0.03,
      "step": 35300
    },
    {
      "epoch": 2.468058367660406,
      "grad_norm": 0.01551117654889822,
      "learning_rate": 2.814609094999457e-05,
      "loss": 0.0231,
      "step": 35350
    },
    {
      "epoch": 2.4715492564406896,
      "grad_norm": 0.09869514405727386,
      "learning_rate": 2.810730299598157e-05,
      "loss": 0.0238,
      "step": 35400
    },
    {
      "epoch": 2.4750401452209734,
      "grad_norm": 4.334355354309082,
      "learning_rate": 2.8068515041968567e-05,
      "loss": 0.0187,
      "step": 35450
    },
    {
      "epoch": 2.4785310340012567,
      "grad_norm": 0.08748903125524521,
      "learning_rate": 2.8029727087955567e-05,
      "loss": 0.0222,
      "step": 35500
    },
    {
      "epoch": 2.48202192278154,
      "grad_norm": 0.1268267035484314,
      "learning_rate": 2.7990939133942563e-05,
      "loss": 0.0226,
      "step": 35550
    },
    {
      "epoch": 2.4855128115618235,
      "grad_norm": 0.12871848046779633,
      "learning_rate": 2.7952151179929563e-05,
      "loss": 0.0266,
      "step": 35600
    },
    {
      "epoch": 2.4890037003421073,
      "grad_norm": 0.0586019866168499,
      "learning_rate": 2.791336322591656e-05,
      "loss": 0.0184,
      "step": 35650
    },
    {
      "epoch": 2.4924945891223906,
      "grad_norm": 4.743653297424316,
      "learning_rate": 2.7874575271903557e-05,
      "loss": 0.0301,
      "step": 35700
    },
    {
      "epoch": 2.495985477902674,
      "grad_norm": 0.12895400822162628,
      "learning_rate": 2.7835787317890553e-05,
      "loss": 0.018,
      "step": 35750
    },
    {
      "epoch": 2.4994763666829574,
      "grad_norm": 0.08446650207042694,
      "learning_rate": 2.7796999363877557e-05,
      "loss": 0.0146,
      "step": 35800
    },
    {
      "epoch": 2.5029672554632407,
      "grad_norm": 0.2824804186820984,
      "learning_rate": 2.7758211409864554e-05,
      "loss": 0.0262,
      "step": 35850
    },
    {
      "epoch": 2.5064581442435245,
      "grad_norm": 0.1400439441204071,
      "learning_rate": 2.771942345585155e-05,
      "loss": 0.0156,
      "step": 35900
    },
    {
      "epoch": 2.509949033023808,
      "grad_norm": 0.005719928536564112,
      "learning_rate": 2.7680635501838547e-05,
      "loss": 0.0163,
      "step": 35950
    },
    {
      "epoch": 2.5134399218040913,
      "grad_norm": 6.512287616729736,
      "learning_rate": 2.764184754782555e-05,
      "loss": 0.0175,
      "step": 36000
    },
    {
      "epoch": 2.5134399218040913,
      "eval_f1": 0.9277177006260672,
      "eval_loss": 0.20059193670749664,
      "eval_runtime": 17.1425,
      "eval_samples_per_second": 213.504,
      "eval_steps_per_second": 6.708,
      "step": 36000
    },
    {
      "epoch": 2.5169308105843746,
      "grad_norm": 1.4769384860992432,
      "learning_rate": 2.7603059593812547e-05,
      "loss": 0.0345,
      "step": 36050
    },
    {
      "epoch": 2.5204216993646584,
      "grad_norm": 0.19504880905151367,
      "learning_rate": 2.7564271639799544e-05,
      "loss": 0.0226,
      "step": 36100
    },
    {
      "epoch": 2.523912588144942,
      "grad_norm": 0.03159239888191223,
      "learning_rate": 2.7525483685786547e-05,
      "loss": 0.0238,
      "step": 36150
    },
    {
      "epoch": 2.527403476925225,
      "grad_norm": 0.06540489941835403,
      "learning_rate": 2.7486695731773544e-05,
      "loss": 0.0253,
      "step": 36200
    },
    {
      "epoch": 2.5308943657055085,
      "grad_norm": 0.23230262100696564,
      "learning_rate": 2.744790777776054e-05,
      "loss": 0.0158,
      "step": 36250
    },
    {
      "epoch": 2.534385254485792,
      "grad_norm": 0.030431071296334267,
      "learning_rate": 2.7409119823747537e-05,
      "loss": 0.0284,
      "step": 36300
    },
    {
      "epoch": 2.5378761432660757,
      "grad_norm": 7.270301342010498,
      "learning_rate": 2.7370331869734537e-05,
      "loss": 0.0228,
      "step": 36350
    },
    {
      "epoch": 2.541367032046359,
      "grad_norm": 0.03877014294266701,
      "learning_rate": 2.7331543915721538e-05,
      "loss": 0.0149,
      "step": 36400
    },
    {
      "epoch": 2.5448579208266424,
      "grad_norm": 0.07648701220750809,
      "learning_rate": 2.7292755961708534e-05,
      "loss": 0.0195,
      "step": 36450
    },
    {
      "epoch": 2.548348809606926,
      "grad_norm": 0.25087717175483704,
      "learning_rate": 2.725396800769553e-05,
      "loss": 0.0101,
      "step": 36500
    },
    {
      "epoch": 2.5518396983872096,
      "grad_norm": 0.12214798480272293,
      "learning_rate": 2.721518005368253e-05,
      "loss": 0.0195,
      "step": 36550
    },
    {
      "epoch": 2.555330587167493,
      "grad_norm": 0.00542970048263669,
      "learning_rate": 2.7176392099669528e-05,
      "loss": 0.0201,
      "step": 36600
    },
    {
      "epoch": 2.5588214759477763,
      "grad_norm": 4.020162105560303,
      "learning_rate": 2.7137604145656524e-05,
      "loss": 0.0274,
      "step": 36650
    },
    {
      "epoch": 2.5623123647280597,
      "grad_norm": 0.08753184229135513,
      "learning_rate": 2.709881619164352e-05,
      "loss": 0.0106,
      "step": 36700
    },
    {
      "epoch": 2.565803253508343,
      "grad_norm": 3.764026165008545,
      "learning_rate": 2.7060028237630524e-05,
      "loss": 0.028,
      "step": 36750
    },
    {
      "epoch": 2.569294142288627,
      "grad_norm": 0.01225785631686449,
      "learning_rate": 2.702124028361752e-05,
      "loss": 0.0167,
      "step": 36800
    },
    {
      "epoch": 2.57278503106891,
      "grad_norm": 0.028353124856948853,
      "learning_rate": 2.6982452329604518e-05,
      "loss": 0.0252,
      "step": 36850
    },
    {
      "epoch": 2.5762759198491936,
      "grad_norm": 0.026566436514258385,
      "learning_rate": 2.6943664375591515e-05,
      "loss": 0.0178,
      "step": 36900
    },
    {
      "epoch": 2.579766808629477,
      "grad_norm": 0.2791537344455719,
      "learning_rate": 2.6904876421578518e-05,
      "loss": 0.0289,
      "step": 36950
    },
    {
      "epoch": 2.5832576974097607,
      "grad_norm": 0.7140030860900879,
      "learning_rate": 2.6866088467565515e-05,
      "loss": 0.0155,
      "step": 37000
    },
    {
      "epoch": 2.5832576974097607,
      "eval_f1": 0.879188441438672,
      "eval_loss": 0.26918989419937134,
      "eval_runtime": 27.051,
      "eval_samples_per_second": 135.3,
      "eval_steps_per_second": 4.251,
      "step": 37000
    },
    {
      "epoch": 2.586748586190044,
      "grad_norm": 0.007281581871211529,
      "learning_rate": 2.682730051355251e-05,
      "loss": 0.0112,
      "step": 37050
    },
    {
      "epoch": 2.5902394749703275,
      "grad_norm": 0.3017968535423279,
      "learning_rate": 2.6788512559539508e-05,
      "loss": 0.0345,
      "step": 37100
    },
    {
      "epoch": 2.593730363750611,
      "grad_norm": 0.06797858327627182,
      "learning_rate": 2.674972460552651e-05,
      "loss": 0.01,
      "step": 37150
    },
    {
      "epoch": 2.597221252530894,
      "grad_norm": 0.012963666580617428,
      "learning_rate": 2.6710936651513508e-05,
      "loss": 0.0152,
      "step": 37200
    },
    {
      "epoch": 2.6007121413111776,
      "grad_norm": 0.07589650899171829,
      "learning_rate": 2.6672148697500505e-05,
      "loss": 0.0245,
      "step": 37250
    },
    {
      "epoch": 2.6042030300914614,
      "grad_norm": 0.028516370803117752,
      "learning_rate": 2.6633360743487505e-05,
      "loss": 0.0132,
      "step": 37300
    },
    {
      "epoch": 2.6076939188717447,
      "grad_norm": 0.24892716109752655,
      "learning_rate": 2.6594572789474505e-05,
      "loss": 0.0161,
      "step": 37350
    },
    {
      "epoch": 2.611184807652028,
      "grad_norm": 0.04786977171897888,
      "learning_rate": 2.6555784835461502e-05,
      "loss": 0.014,
      "step": 37400
    },
    {
      "epoch": 2.614675696432312,
      "grad_norm": 0.12335788458585739,
      "learning_rate": 2.65169968814485e-05,
      "loss": 0.029,
      "step": 37450
    },
    {
      "epoch": 2.6181665852125953,
      "grad_norm": 0.18296684324741364,
      "learning_rate": 2.64782089274355e-05,
      "loss": 0.0216,
      "step": 37500
    },
    {
      "epoch": 2.6216574739928786,
      "grad_norm": 0.020824022591114044,
      "learning_rate": 2.6439420973422495e-05,
      "loss": 0.0098,
      "step": 37550
    },
    {
      "epoch": 2.625148362773162,
      "grad_norm": 0.3721611797809601,
      "learning_rate": 2.6400633019409492e-05,
      "loss": 0.0104,
      "step": 37600
    },
    {
      "epoch": 2.6286392515534454,
      "grad_norm": 0.0648498460650444,
      "learning_rate": 2.636184506539649e-05,
      "loss": 0.029,
      "step": 37650
    },
    {
      "epoch": 2.6321301403337287,
      "grad_norm": 0.013140054419636726,
      "learning_rate": 2.6323057111383492e-05,
      "loss": 0.0175,
      "step": 37700
    },
    {
      "epoch": 2.6356210291140125,
      "grad_norm": 0.3959859609603882,
      "learning_rate": 2.628426915737049e-05,
      "loss": 0.0287,
      "step": 37750
    },
    {
      "epoch": 2.639111917894296,
      "grad_norm": 1.1734418869018555,
      "learning_rate": 2.6245481203357485e-05,
      "loss": 0.0185,
      "step": 37800
    },
    {
      "epoch": 2.6426028066745793,
      "grad_norm": 13.356185913085938,
      "learning_rate": 2.6206693249344482e-05,
      "loss": 0.009,
      "step": 37850
    },
    {
      "epoch": 2.646093695454863,
      "grad_norm": 6.345465183258057,
      "learning_rate": 2.6167905295331486e-05,
      "loss": 0.0185,
      "step": 37900
    },
    {
      "epoch": 2.6495845842351464,
      "grad_norm": 0.3218725025653839,
      "learning_rate": 2.6129117341318482e-05,
      "loss": 0.0325,
      "step": 37950
    },
    {
      "epoch": 2.65307547301543,
      "grad_norm": 0.013729876838624477,
      "learning_rate": 2.609032938730548e-05,
      "loss": 0.0147,
      "step": 38000
    },
    {
      "epoch": 2.65307547301543,
      "eval_f1": 0.9022556390977443,
      "eval_loss": 0.2909112870693207,
      "eval_runtime": 23.2176,
      "eval_samples_per_second": 157.639,
      "eval_steps_per_second": 4.953,
      "step": 38000
    },
    {
      "epoch": 2.656566361795713,
      "grad_norm": 0.33415481448173523,
      "learning_rate": 2.6051541433292476e-05,
      "loss": 0.0259,
      "step": 38050
    },
    {
      "epoch": 2.6600572505759965,
      "grad_norm": 0.009586818516254425,
      "learning_rate": 2.601275347927948e-05,
      "loss": 0.0227,
      "step": 38100
    },
    {
      "epoch": 2.66354813935628,
      "grad_norm": 0.024410735815763474,
      "learning_rate": 2.5973965525266476e-05,
      "loss": 0.0259,
      "step": 38150
    },
    {
      "epoch": 2.6670390281365637,
      "grad_norm": 0.03022683411836624,
      "learning_rate": 2.5935177571253472e-05,
      "loss": 0.0122,
      "step": 38200
    },
    {
      "epoch": 2.670529916916847,
      "grad_norm": 3.5583574771881104,
      "learning_rate": 2.589638961724047e-05,
      "loss": 0.0323,
      "step": 38250
    },
    {
      "epoch": 2.6740208056971304,
      "grad_norm": 0.07743709534406662,
      "learning_rate": 2.5857601663227473e-05,
      "loss": 0.0185,
      "step": 38300
    },
    {
      "epoch": 2.677511694477414,
      "grad_norm": 0.18533405661582947,
      "learning_rate": 2.581881370921447e-05,
      "loss": 0.0277,
      "step": 38350
    },
    {
      "epoch": 2.6810025832576976,
      "grad_norm": 0.05966994911432266,
      "learning_rate": 2.5780025755201466e-05,
      "loss": 0.0199,
      "step": 38400
    },
    {
      "epoch": 2.684493472037981,
      "grad_norm": 0.028219763189554214,
      "learning_rate": 2.5741237801188463e-05,
      "loss": 0.0184,
      "step": 38450
    },
    {
      "epoch": 2.6879843608182643,
      "grad_norm": 0.026486072689294815,
      "learning_rate": 2.5702449847175463e-05,
      "loss": 0.0152,
      "step": 38500
    },
    {
      "epoch": 2.6914752495985477,
      "grad_norm": 0.4646790623664856,
      "learning_rate": 2.566366189316246e-05,
      "loss": 0.0113,
      "step": 38550
    },
    {
      "epoch": 2.694966138378831,
      "grad_norm": 0.016069358214735985,
      "learning_rate": 2.5624873939149456e-05,
      "loss": 0.0272,
      "step": 38600
    },
    {
      "epoch": 2.698457027159115,
      "grad_norm": 0.1055779829621315,
      "learning_rate": 2.558608598513646e-05,
      "loss": 0.0372,
      "step": 38650
    },
    {
      "epoch": 2.701947915939398,
      "grad_norm": 0.08721558004617691,
      "learning_rate": 2.5547298031123456e-05,
      "loss": 0.0299,
      "step": 38700
    },
    {
      "epoch": 2.7054388047196816,
      "grad_norm": 4.360162258148193,
      "learning_rate": 2.5508510077110453e-05,
      "loss": 0.0159,
      "step": 38750
    },
    {
      "epoch": 2.708929693499965,
      "grad_norm": 0.04794733598828316,
      "learning_rate": 2.546972212309745e-05,
      "loss": 0.0152,
      "step": 38800
    },
    {
      "epoch": 2.7124205822802487,
      "grad_norm": 1.7673709392547607,
      "learning_rate": 2.5430934169084453e-05,
      "loss": 0.0284,
      "step": 38850
    },
    {
      "epoch": 2.715911471060532,
      "grad_norm": 0.17031490802764893,
      "learning_rate": 2.539214621507145e-05,
      "loss": 0.0154,
      "step": 38900
    },
    {
      "epoch": 2.7194023598408155,
      "grad_norm": 0.010257457382977009,
      "learning_rate": 2.5353358261058446e-05,
      "loss": 0.0092,
      "step": 38950
    },
    {
      "epoch": 2.722893248621099,
      "grad_norm": 0.011103369295597076,
      "learning_rate": 2.5314570307045443e-05,
      "loss": 0.017,
      "step": 39000
    },
    {
      "epoch": 2.722893248621099,
      "eval_f1": 0.9230318993268949,
      "eval_loss": 0.2264816164970398,
      "eval_runtime": 23.7732,
      "eval_samples_per_second": 153.955,
      "eval_steps_per_second": 4.837,
      "step": 39000
    },
    {
      "epoch": 2.726384137401382,
      "grad_norm": 0.024749957025051117,
      "learning_rate": 2.5275782353032447e-05,
      "loss": 0.025,
      "step": 39050
    },
    {
      "epoch": 2.729875026181666,
      "grad_norm": 0.016311727464199066,
      "learning_rate": 2.5236994399019443e-05,
      "loss": 0.019,
      "step": 39100
    },
    {
      "epoch": 2.7333659149619494,
      "grad_norm": 2.8876538276672363,
      "learning_rate": 2.519820644500644e-05,
      "loss": 0.0152,
      "step": 39150
    },
    {
      "epoch": 2.7368568037422327,
      "grad_norm": 0.019060427322983742,
      "learning_rate": 2.5159418490993437e-05,
      "loss": 0.0122,
      "step": 39200
    },
    {
      "epoch": 2.740347692522516,
      "grad_norm": 0.011108928360044956,
      "learning_rate": 2.5121406296060695e-05,
      "loss": 0.017,
      "step": 39250
    },
    {
      "epoch": 2.7438385813028,
      "grad_norm": 0.09083431214094162,
      "learning_rate": 2.508261834204769e-05,
      "loss": 0.0295,
      "step": 39300
    },
    {
      "epoch": 2.7473294700830833,
      "grad_norm": 0.005036790389567614,
      "learning_rate": 2.5043830388034695e-05,
      "loss": 0.0068,
      "step": 39350
    },
    {
      "epoch": 2.7508203588633666,
      "grad_norm": 0.047138676047325134,
      "learning_rate": 2.500504243402169e-05,
      "loss": 0.0169,
      "step": 39400
    },
    {
      "epoch": 2.75431124764365,
      "grad_norm": 0.696245551109314,
      "learning_rate": 2.4966254480008688e-05,
      "loss": 0.03,
      "step": 39450
    },
    {
      "epoch": 2.7578021364239333,
      "grad_norm": 2.528618097305298,
      "learning_rate": 2.4927466525995688e-05,
      "loss": 0.0273,
      "step": 39500
    },
    {
      "epoch": 2.761293025204217,
      "grad_norm": 0.01160145178437233,
      "learning_rate": 2.488867857198269e-05,
      "loss": 0.0171,
      "step": 39550
    },
    {
      "epoch": 2.7647839139845005,
      "grad_norm": 0.00931859016418457,
      "learning_rate": 2.4849890617969685e-05,
      "loss": 0.019,
      "step": 39600
    },
    {
      "epoch": 2.768274802764784,
      "grad_norm": 0.011263793334364891,
      "learning_rate": 2.4811102663956685e-05,
      "loss": 0.0227,
      "step": 39650
    },
    {
      "epoch": 2.7717656915450672,
      "grad_norm": 0.01920177787542343,
      "learning_rate": 2.4772314709943682e-05,
      "loss": 0.0195,
      "step": 39700
    },
    {
      "epoch": 2.775256580325351,
      "grad_norm": 0.0142729626968503,
      "learning_rate": 2.4733526755930682e-05,
      "loss": 0.0232,
      "step": 39750
    },
    {
      "epoch": 2.7787474691056344,
      "grad_norm": 0.10750947892665863,
      "learning_rate": 2.469473880191768e-05,
      "loss": 0.0152,
      "step": 39800
    },
    {
      "epoch": 2.782238357885918,
      "grad_norm": 0.018952248618006706,
      "learning_rate": 2.4655950847904675e-05,
      "loss": 0.0113,
      "step": 39850
    },
    {
      "epoch": 2.785729246666201,
      "grad_norm": 0.019632501527667046,
      "learning_rate": 2.4617162893891675e-05,
      "loss": 0.021,
      "step": 39900
    },
    {
      "epoch": 2.7892201354464845,
      "grad_norm": 0.0779823437333107,
      "learning_rate": 2.4578374939878672e-05,
      "loss": 0.025,
      "step": 39950
    },
    {
      "epoch": 2.792711024226768,
      "grad_norm": 0.02033323608338833,
      "learning_rate": 2.453958698586567e-05,
      "loss": 0.0235,
      "step": 40000
    },
    {
      "epoch": 2.792711024226768,
      "eval_f1": 0.9247813411078717,
      "eval_loss": 0.19202525913715363,
      "eval_runtime": 24.1493,
      "eval_samples_per_second": 151.557,
      "eval_steps_per_second": 4.762,
      "step": 40000
    },
    {
      "epoch": 2.7962019130070517,
      "grad_norm": 0.711127758026123,
      "learning_rate": 2.450079903185267e-05,
      "loss": 0.0281,
      "step": 40050
    },
    {
      "epoch": 2.799692801787335,
      "grad_norm": 0.013284056447446346,
      "learning_rate": 2.4462011077839665e-05,
      "loss": 0.0338,
      "step": 40100
    },
    {
      "epoch": 2.8031836905676184,
      "grad_norm": 0.11050266772508621,
      "learning_rate": 2.4423223123826666e-05,
      "loss": 0.0202,
      "step": 40150
    },
    {
      "epoch": 2.806674579347902,
      "grad_norm": 0.056580349802970886,
      "learning_rate": 2.4384435169813662e-05,
      "loss": 0.0235,
      "step": 40200
    },
    {
      "epoch": 2.8101654681281856,
      "grad_norm": 0.03578386455774307,
      "learning_rate": 2.4346422974880924e-05,
      "loss": 0.0262,
      "step": 40250
    },
    {
      "epoch": 2.813656356908469,
      "grad_norm": 0.02802601084113121,
      "learning_rate": 2.430763502086792e-05,
      "loss": 0.0245,
      "step": 40300
    },
    {
      "epoch": 2.8171472456887523,
      "grad_norm": 1.0409111976623535,
      "learning_rate": 2.426884706685492e-05,
      "loss": 0.0199,
      "step": 40350
    },
    {
      "epoch": 2.8206381344690357,
      "grad_norm": 0.031876832246780396,
      "learning_rate": 2.4230059112841917e-05,
      "loss": 0.0271,
      "step": 40400
    },
    {
      "epoch": 2.824129023249319,
      "grad_norm": 0.05730342119932175,
      "learning_rate": 2.4191271158828914e-05,
      "loss": 0.0203,
      "step": 40450
    },
    {
      "epoch": 2.827619912029603,
      "grad_norm": 4.108118534088135,
      "learning_rate": 2.4152483204815914e-05,
      "loss": 0.023,
      "step": 40500
    },
    {
      "epoch": 2.831110800809886,
      "grad_norm": 0.18447378277778625,
      "learning_rate": 2.411369525080291e-05,
      "loss": 0.0159,
      "step": 40550
    },
    {
      "epoch": 2.8346016895901696,
      "grad_norm": 0.10659290850162506,
      "learning_rate": 2.407490729678991e-05,
      "loss": 0.016,
      "step": 40600
    },
    {
      "epoch": 2.8380925783704534,
      "grad_norm": 0.01589849404990673,
      "learning_rate": 2.4036119342776907e-05,
      "loss": 0.0209,
      "step": 40650
    },
    {
      "epoch": 2.8415834671507367,
      "grad_norm": 0.14199930429458618,
      "learning_rate": 2.3997331388763907e-05,
      "loss": 0.0162,
      "step": 40700
    },
    {
      "epoch": 2.84507435593102,
      "grad_norm": 3.8175432682037354,
      "learning_rate": 2.3958543434750904e-05,
      "loss": 0.0339,
      "step": 40750
    },
    {
      "epoch": 2.8485652447113035,
      "grad_norm": 4.6632537841796875,
      "learning_rate": 2.3919755480737904e-05,
      "loss": 0.0371,
      "step": 40800
    },
    {
      "epoch": 2.852056133491587,
      "grad_norm": 3.703428268432617,
      "learning_rate": 2.38809675267249e-05,
      "loss": 0.013,
      "step": 40850
    },
    {
      "epoch": 2.85554702227187,
      "grad_norm": 0.14800503849983215,
      "learning_rate": 2.38421795727119e-05,
      "loss": 0.0126,
      "step": 40900
    },
    {
      "epoch": 2.859037911052154,
      "grad_norm": 6.090341567993164,
      "learning_rate": 2.3803391618698898e-05,
      "loss": 0.0239,
      "step": 40950
    },
    {
      "epoch": 2.8625287998324374,
      "grad_norm": 0.4346800744533539,
      "learning_rate": 2.3764603664685898e-05,
      "loss": 0.0238,
      "step": 41000
    },
    {
      "epoch": 2.8625287998324374,
      "eval_f1": 0.922173274596182,
      "eval_loss": 0.2260962426662445,
      "eval_runtime": 26.009,
      "eval_samples_per_second": 140.721,
      "eval_steps_per_second": 4.422,
      "step": 41000
    },
    {
      "epoch": 2.8660196886127207,
      "grad_norm": 3.542853593826294,
      "learning_rate": 2.3725815710672894e-05,
      "loss": 0.014,
      "step": 41050
    },
    {
      "epoch": 2.8695105773930045,
      "grad_norm": 4.3590779304504395,
      "learning_rate": 2.3687027756659894e-05,
      "loss": 0.0104,
      "step": 41100
    },
    {
      "epoch": 2.873001466173288,
      "grad_norm": 0.4096594452857971,
      "learning_rate": 2.364823980264689e-05,
      "loss": 0.0247,
      "step": 41150
    },
    {
      "epoch": 2.8764923549535713,
      "grad_norm": 0.01314621139317751,
      "learning_rate": 2.360945184863389e-05,
      "loss": 0.0285,
      "step": 41200
    },
    {
      "epoch": 2.8799832437338546,
      "grad_norm": 0.2991047501564026,
      "learning_rate": 2.3570663894620888e-05,
      "loss": 0.0205,
      "step": 41250
    },
    {
      "epoch": 2.883474132514138,
      "grad_norm": 5.858856678009033,
      "learning_rate": 2.3531875940607888e-05,
      "loss": 0.0201,
      "step": 41300
    },
    {
      "epoch": 2.8869650212944213,
      "grad_norm": 0.10207907110452652,
      "learning_rate": 2.3493087986594885e-05,
      "loss": 0.0238,
      "step": 41350
    },
    {
      "epoch": 2.890455910074705,
      "grad_norm": 0.19928042590618134,
      "learning_rate": 2.345430003258188e-05,
      "loss": 0.0167,
      "step": 41400
    },
    {
      "epoch": 2.8939467988549885,
      "grad_norm": 0.028837937861680984,
      "learning_rate": 2.341551207856888e-05,
      "loss": 0.0185,
      "step": 41450
    },
    {
      "epoch": 2.897437687635272,
      "grad_norm": 0.019361231476068497,
      "learning_rate": 2.3376724124555878e-05,
      "loss": 0.0158,
      "step": 41500
    },
    {
      "epoch": 2.9009285764155552,
      "grad_norm": 4.604653358459473,
      "learning_rate": 2.3337936170542875e-05,
      "loss": 0.0186,
      "step": 41550
    },
    {
      "epoch": 2.904419465195839,
      "grad_norm": 1.2114059925079346,
      "learning_rate": 2.3299148216529875e-05,
      "loss": 0.0311,
      "step": 41600
    },
    {
      "epoch": 2.9079103539761224,
      "grad_norm": 0.021827546879649162,
      "learning_rate": 2.326036026251687e-05,
      "loss": 0.0109,
      "step": 41650
    },
    {
      "epoch": 2.9114012427564058,
      "grad_norm": 2.5706710815429688,
      "learning_rate": 2.322157230850387e-05,
      "loss": 0.0106,
      "step": 41700
    },
    {
      "epoch": 2.914892131536689,
      "grad_norm": 0.039485763758420944,
      "learning_rate": 2.3182784354490868e-05,
      "loss": 0.0163,
      "step": 41750
    },
    {
      "epoch": 2.9183830203169725,
      "grad_norm": 0.16371645033359528,
      "learning_rate": 2.314399640047787e-05,
      "loss": 0.0101,
      "step": 41800
    },
    {
      "epoch": 2.9218739090972563,
      "grad_norm": 1.083182692527771,
      "learning_rate": 2.310520844646487e-05,
      "loss": 0.0193,
      "step": 41850
    },
    {
      "epoch": 2.9253647978775397,
      "grad_norm": 0.038459427654743195,
      "learning_rate": 2.3066420492451865e-05,
      "loss": 0.0191,
      "step": 41900
    },
    {
      "epoch": 2.928855686657823,
      "grad_norm": 0.054930657148361206,
      "learning_rate": 2.3027632538438865e-05,
      "loss": 0.0258,
      "step": 41950
    },
    {
      "epoch": 2.9323465754381064,
      "grad_norm": 0.01878916658461094,
      "learning_rate": 2.2988844584425862e-05,
      "loss": 0.0291,
      "step": 42000
    },
    {
      "epoch": 2.9323465754381064,
      "eval_f1": 0.9320612893899971,
      "eval_loss": 0.17577876150608063,
      "eval_runtime": 28.3368,
      "eval_samples_per_second": 129.161,
      "eval_steps_per_second": 4.058,
      "step": 42000
    },
    {
      "epoch": 2.93583746421839,
      "grad_norm": 1.7324750423431396,
      "learning_rate": 2.2950056630412862e-05,
      "loss": 0.0294,
      "step": 42050
    },
    {
      "epoch": 2.9393283529986736,
      "grad_norm": 0.13416598737239838,
      "learning_rate": 2.291126867639986e-05,
      "loss": 0.032,
      "step": 42100
    },
    {
      "epoch": 2.942819241778957,
      "grad_norm": 7.673727035522461,
      "learning_rate": 2.287248072238686e-05,
      "loss": 0.0158,
      "step": 42150
    },
    {
      "epoch": 2.9463101305592403,
      "grad_norm": 0.4483250677585602,
      "learning_rate": 2.2833692768373855e-05,
      "loss": 0.0229,
      "step": 42200
    },
    {
      "epoch": 2.9498010193395237,
      "grad_norm": 0.10611013323068619,
      "learning_rate": 2.2794904814360855e-05,
      "loss": 0.0146,
      "step": 42250
    },
    {
      "epoch": 2.9532919081198075,
      "grad_norm": 6.6183600425720215,
      "learning_rate": 2.2756116860347852e-05,
      "loss": 0.006,
      "step": 42300
    },
    {
      "epoch": 2.956782796900091,
      "grad_norm": 0.008068676106631756,
      "learning_rate": 2.271732890633485e-05,
      "loss": 0.0208,
      "step": 42350
    },
    {
      "epoch": 2.960273685680374,
      "grad_norm": 0.01248723454773426,
      "learning_rate": 2.267854095232185e-05,
      "loss": 0.0412,
      "step": 42400
    },
    {
      "epoch": 2.9637645744606576,
      "grad_norm": 1.5417126417160034,
      "learning_rate": 2.2639752998308846e-05,
      "loss": 0.0363,
      "step": 42450
    },
    {
      "epoch": 2.9672554632409414,
      "grad_norm": 0.12593162059783936,
      "learning_rate": 2.2600965044295842e-05,
      "loss": 0.0107,
      "step": 42500
    },
    {
      "epoch": 2.9707463520212247,
      "grad_norm": 2.7952985763549805,
      "learning_rate": 2.2562177090282842e-05,
      "loss": 0.0231,
      "step": 42550
    },
    {
      "epoch": 2.974237240801508,
      "grad_norm": 0.021743256598711014,
      "learning_rate": 2.252338913626984e-05,
      "loss": 0.0238,
      "step": 42600
    },
    {
      "epoch": 2.9777281295817914,
      "grad_norm": 0.01719530299305916,
      "learning_rate": 2.248460118225684e-05,
      "loss": 0.0104,
      "step": 42650
    },
    {
      "epoch": 2.981219018362075,
      "grad_norm": 0.009256424382328987,
      "learning_rate": 2.2445813228243836e-05,
      "loss": 0.01,
      "step": 42700
    },
    {
      "epoch": 2.984709907142358,
      "grad_norm": 0.010182956233620644,
      "learning_rate": 2.2407025274230836e-05,
      "loss": 0.0162,
      "step": 42750
    },
    {
      "epoch": 2.988200795922642,
      "grad_norm": 0.46795710921287537,
      "learning_rate": 2.2368237320217833e-05,
      "loss": 0.035,
      "step": 42800
    },
    {
      "epoch": 2.9916916847029253,
      "grad_norm": 0.037175968289375305,
      "learning_rate": 2.2329449366204833e-05,
      "loss": 0.0212,
      "step": 42850
    },
    {
      "epoch": 2.9951825734832087,
      "grad_norm": 0.2701772451400757,
      "learning_rate": 2.229066141219183e-05,
      "loss": 0.0215,
      "step": 42900
    },
    {
      "epoch": 2.9986734622634925,
      "grad_norm": 0.3241135776042938,
      "learning_rate": 2.225187345817883e-05,
      "loss": 0.0211,
      "step": 42950
    },
    {
      "epoch": 3.002164351043776,
      "grad_norm": 0.25792551040649414,
      "learning_rate": 2.221308550416583e-05,
      "loss": 0.018,
      "step": 43000
    },
    {
      "epoch": 3.002164351043776,
      "eval_f1": 0.9154761904761904,
      "eval_loss": 0.2519144117832184,
      "eval_runtime": 24.3041,
      "eval_samples_per_second": 150.592,
      "eval_steps_per_second": 4.732,
      "step": 43000
    },
    {
      "epoch": 3.0056552398240592,
      "grad_norm": 0.2332383692264557,
      "learning_rate": 2.2174297550152826e-05,
      "loss": 0.0236,
      "step": 43050
    },
    {
      "epoch": 3.0091461286043426,
      "grad_norm": 0.01284001674503088,
      "learning_rate": 2.2135509596139826e-05,
      "loss": 0.0154,
      "step": 43100
    },
    {
      "epoch": 3.012637017384626,
      "grad_norm": 0.05416431277990341,
      "learning_rate": 2.2096721642126823e-05,
      "loss": 0.0115,
      "step": 43150
    },
    {
      "epoch": 3.0161279061649098,
      "grad_norm": 0.12467977404594421,
      "learning_rate": 2.2057933688113823e-05,
      "loss": 0.0176,
      "step": 43200
    },
    {
      "epoch": 3.019618794945193,
      "grad_norm": 0.5407397150993347,
      "learning_rate": 2.201914573410082e-05,
      "loss": 0.0206,
      "step": 43250
    },
    {
      "epoch": 3.0231096837254765,
      "grad_norm": 0.13258038461208344,
      "learning_rate": 2.1980357780087816e-05,
      "loss": 0.0112,
      "step": 43300
    },
    {
      "epoch": 3.02660057250576,
      "grad_norm": 0.05076462775468826,
      "learning_rate": 2.1941569826074816e-05,
      "loss": 0.0146,
      "step": 43350
    },
    {
      "epoch": 3.0300914612860432,
      "grad_norm": 0.05165479704737663,
      "learning_rate": 2.1902781872061813e-05,
      "loss": 0.011,
      "step": 43400
    },
    {
      "epoch": 3.033582350066327,
      "grad_norm": 0.04711933061480522,
      "learning_rate": 2.186399391804881e-05,
      "loss": 0.0207,
      "step": 43450
    },
    {
      "epoch": 3.0370732388466104,
      "grad_norm": 0.08267899602651596,
      "learning_rate": 2.182520596403581e-05,
      "loss": 0.0277,
      "step": 43500
    },
    {
      "epoch": 3.0405641276268938,
      "grad_norm": 0.05017954856157303,
      "learning_rate": 2.1786418010022807e-05,
      "loss": 0.0162,
      "step": 43550
    },
    {
      "epoch": 3.044055016407177,
      "grad_norm": 0.008251439779996872,
      "learning_rate": 2.1747630056009807e-05,
      "loss": 0.0148,
      "step": 43600
    },
    {
      "epoch": 3.047545905187461,
      "grad_norm": 0.03382747620344162,
      "learning_rate": 2.1708842101996803e-05,
      "loss": 0.0128,
      "step": 43650
    },
    {
      "epoch": 3.0510367939677443,
      "grad_norm": 0.008427140302956104,
      "learning_rate": 2.1670054147983803e-05,
      "loss": 0.0173,
      "step": 43700
    },
    {
      "epoch": 3.0545276827480277,
      "grad_norm": 11.583622932434082,
      "learning_rate": 2.16312661939708e-05,
      "loss": 0.0258,
      "step": 43750
    },
    {
      "epoch": 3.058018571528311,
      "grad_norm": 0.026552023366093636,
      "learning_rate": 2.15924782399578e-05,
      "loss": 0.0326,
      "step": 43800
    },
    {
      "epoch": 3.0615094603085944,
      "grad_norm": 0.08250391483306885,
      "learning_rate": 2.1553690285944797e-05,
      "loss": 0.0064,
      "step": 43850
    },
    {
      "epoch": 3.065000349088878,
      "grad_norm": 0.06599225103855133,
      "learning_rate": 2.1514902331931797e-05,
      "loss": 0.0316,
      "step": 43900
    },
    {
      "epoch": 3.0684912378691616,
      "grad_norm": 0.062200937420129776,
      "learning_rate": 2.1476114377918794e-05,
      "loss": 0.0192,
      "step": 43950
    },
    {
      "epoch": 3.071982126649445,
      "grad_norm": 7.155509948730469,
      "learning_rate": 2.1437326423905794e-05,
      "loss": 0.0201,
      "step": 44000
    },
    {
      "epoch": 3.071982126649445,
      "eval_f1": 0.9021739130434783,
      "eval_loss": 0.27857547998428345,
      "eval_runtime": 25.994,
      "eval_samples_per_second": 140.802,
      "eval_steps_per_second": 4.424,
      "step": 44000
    },
    {
      "epoch": 3.0754730154297283,
      "grad_norm": 0.014235005713999271,
      "learning_rate": 2.139853846989279e-05,
      "loss": 0.022,
      "step": 44050
    },
    {
      "epoch": 3.0789639042100116,
      "grad_norm": 0.05011603608727455,
      "learning_rate": 2.135975051587979e-05,
      "loss": 0.0107,
      "step": 44100
    },
    {
      "epoch": 3.0824547929902955,
      "grad_norm": 1.0543460845947266,
      "learning_rate": 2.1320962561866787e-05,
      "loss": 0.0196,
      "step": 44150
    },
    {
      "epoch": 3.085945681770579,
      "grad_norm": 0.09093031287193298,
      "learning_rate": 2.1282174607853787e-05,
      "loss": 0.0145,
      "step": 44200
    },
    {
      "epoch": 3.089436570550862,
      "grad_norm": 0.04965496063232422,
      "learning_rate": 2.1243386653840784e-05,
      "loss": 0.0161,
      "step": 44250
    },
    {
      "epoch": 3.0929274593311455,
      "grad_norm": 2.1972270011901855,
      "learning_rate": 2.1204598699827784e-05,
      "loss": 0.0289,
      "step": 44300
    },
    {
      "epoch": 3.0964183481114294,
      "grad_norm": 0.02106853947043419,
      "learning_rate": 2.116581074581478e-05,
      "loss": 0.0303,
      "step": 44350
    },
    {
      "epoch": 3.0999092368917127,
      "grad_norm": 2.8582568168640137,
      "learning_rate": 2.1127022791801777e-05,
      "loss": 0.0177,
      "step": 44400
    },
    {
      "epoch": 3.103400125671996,
      "grad_norm": 0.010883578099310398,
      "learning_rate": 2.1088234837788777e-05,
      "loss": 0.0216,
      "step": 44450
    },
    {
      "epoch": 3.1068910144522794,
      "grad_norm": 0.20962642133235931,
      "learning_rate": 2.1049446883775774e-05,
      "loss": 0.0093,
      "step": 44500
    },
    {
      "epoch": 3.110381903232563,
      "grad_norm": 2.5480289459228516,
      "learning_rate": 2.1010658929762774e-05,
      "loss": 0.0104,
      "step": 44550
    },
    {
      "epoch": 3.1138727920128466,
      "grad_norm": 0.05912920832633972,
      "learning_rate": 2.097187097574977e-05,
      "loss": 0.0089,
      "step": 44600
    },
    {
      "epoch": 3.11736368079313,
      "grad_norm": 0.006363342050462961,
      "learning_rate": 2.0933858780817032e-05,
      "loss": 0.0145,
      "step": 44650
    },
    {
      "epoch": 3.1208545695734133,
      "grad_norm": 0.09209737926721573,
      "learning_rate": 2.0895846585884287e-05,
      "loss": 0.0143,
      "step": 44700
    },
    {
      "epoch": 3.1243454583536967,
      "grad_norm": 0.08498357236385345,
      "learning_rate": 2.0857058631871287e-05,
      "loss": 0.0236,
      "step": 44750
    },
    {
      "epoch": 3.1278363471339805,
      "grad_norm": 0.20750215649604797,
      "learning_rate": 2.0818270677858284e-05,
      "loss": 0.0222,
      "step": 44800
    },
    {
      "epoch": 3.131327235914264,
      "grad_norm": 1.05506432056427,
      "learning_rate": 2.0779482723845284e-05,
      "loss": 0.019,
      "step": 44850
    },
    {
      "epoch": 3.1348181246945472,
      "grad_norm": 5.7602763175964355,
      "learning_rate": 2.074069476983228e-05,
      "loss": 0.0324,
      "step": 44900
    },
    {
      "epoch": 3.1383090134748306,
      "grad_norm": 3.679661989212036,
      "learning_rate": 2.070190681581928e-05,
      "loss": 0.0235,
      "step": 44950
    },
    {
      "epoch": 3.141799902255114,
      "grad_norm": 0.003911898471415043,
      "learning_rate": 2.0663118861806277e-05,
      "loss": 0.0212,
      "step": 45000
    },
    {
      "epoch": 3.141799902255114,
      "eval_f1": 0.9351320321469575,
      "eval_loss": 0.1995823234319687,
      "eval_runtime": 24.8557,
      "eval_samples_per_second": 147.25,
      "eval_steps_per_second": 4.627,
      "step": 45000
    },
    {
      "epoch": 3.1452907910353978,
      "grad_norm": 0.02615140751004219,
      "learning_rate": 2.0624330907793277e-05,
      "loss": 0.0125,
      "step": 45050
    },
    {
      "epoch": 3.148781679815681,
      "grad_norm": 0.01764095388352871,
      "learning_rate": 2.0585542953780277e-05,
      "loss": 0.0036,
      "step": 45100
    },
    {
      "epoch": 3.1522725685959645,
      "grad_norm": 0.04237894341349602,
      "learning_rate": 2.0546754999767274e-05,
      "loss": 0.0299,
      "step": 45150
    },
    {
      "epoch": 3.155763457376248,
      "grad_norm": 0.008063646033406258,
      "learning_rate": 2.0507967045754274e-05,
      "loss": 0.022,
      "step": 45200
    },
    {
      "epoch": 3.1592543461565317,
      "grad_norm": 0.016073619946837425,
      "learning_rate": 2.046917909174127e-05,
      "loss": 0.0081,
      "step": 45250
    },
    {
      "epoch": 3.162745234936815,
      "grad_norm": 1.8636724948883057,
      "learning_rate": 2.0430391137728268e-05,
      "loss": 0.0213,
      "step": 45300
    },
    {
      "epoch": 3.1662361237170984,
      "grad_norm": 0.012820547446608543,
      "learning_rate": 2.0391603183715268e-05,
      "loss": 0.0296,
      "step": 45350
    },
    {
      "epoch": 3.1697270124973818,
      "grad_norm": 0.02494175173342228,
      "learning_rate": 2.0352815229702264e-05,
      "loss": 0.0114,
      "step": 45400
    },
    {
      "epoch": 3.173217901277665,
      "grad_norm": 0.0929480642080307,
      "learning_rate": 2.031402727568926e-05,
      "loss": 0.0297,
      "step": 45450
    },
    {
      "epoch": 3.176708790057949,
      "grad_norm": 0.06418604403734207,
      "learning_rate": 2.027523932167626e-05,
      "loss": 0.0181,
      "step": 45500
    },
    {
      "epoch": 3.1801996788382323,
      "grad_norm": 0.13698844611644745,
      "learning_rate": 2.0236451367663258e-05,
      "loss": 0.0223,
      "step": 45550
    },
    {
      "epoch": 3.1836905676185157,
      "grad_norm": 0.06596588343381882,
      "learning_rate": 2.0197663413650258e-05,
      "loss": 0.0213,
      "step": 45600
    },
    {
      "epoch": 3.187181456398799,
      "grad_norm": 1.1904255151748657,
      "learning_rate": 2.0158875459637254e-05,
      "loss": 0.0287,
      "step": 45650
    },
    {
      "epoch": 3.190672345179083,
      "grad_norm": 0.04259861633181572,
      "learning_rate": 2.0120087505624255e-05,
      "loss": 0.0409,
      "step": 45700
    },
    {
      "epoch": 3.194163233959366,
      "grad_norm": 0.02005760930478573,
      "learning_rate": 2.008129955161125e-05,
      "loss": 0.0159,
      "step": 45750
    },
    {
      "epoch": 3.1976541227396496,
      "grad_norm": 0.023527618497610092,
      "learning_rate": 2.004251159759825e-05,
      "loss": 0.011,
      "step": 45800
    },
    {
      "epoch": 3.201145011519933,
      "grad_norm": 0.054994259029626846,
      "learning_rate": 2.0003723643585248e-05,
      "loss": 0.0174,
      "step": 45850
    },
    {
      "epoch": 3.2046359003002163,
      "grad_norm": 3.069230556488037,
      "learning_rate": 1.9964935689572248e-05,
      "loss": 0.0333,
      "step": 45900
    },
    {
      "epoch": 3.2081267890805,
      "grad_norm": 0.053326524794101715,
      "learning_rate": 1.9926147735559245e-05,
      "loss": 0.0156,
      "step": 45950
    },
    {
      "epoch": 3.2116176778607834,
      "grad_norm": 0.0816662609577179,
      "learning_rate": 1.9887359781546245e-05,
      "loss": 0.0186,
      "step": 46000
    },
    {
      "epoch": 3.2116176778607834,
      "eval_f1": 0.9357903829542182,
      "eval_loss": 0.19323907792568207,
      "eval_runtime": 26.5504,
      "eval_samples_per_second": 137.851,
      "eval_steps_per_second": 4.331,
      "step": 46000
    },
    {
      "epoch": 3.215108566641067,
      "grad_norm": 4.011587619781494,
      "learning_rate": 1.984857182753324e-05,
      "loss": 0.0168,
      "step": 46050
    },
    {
      "epoch": 3.21859945542135,
      "grad_norm": 0.3233940899372101,
      "learning_rate": 1.980978387352024e-05,
      "loss": 0.0223,
      "step": 46100
    },
    {
      "epoch": 3.2220903442016335,
      "grad_norm": 0.6042135953903198,
      "learning_rate": 1.9770995919507238e-05,
      "loss": 0.0151,
      "step": 46150
    },
    {
      "epoch": 3.2255812329819173,
      "grad_norm": 0.35782158374786377,
      "learning_rate": 1.973220796549424e-05,
      "loss": 0.0112,
      "step": 46200
    },
    {
      "epoch": 3.2290721217622007,
      "grad_norm": 0.007998952642083168,
      "learning_rate": 1.9693420011481235e-05,
      "loss": 0.0123,
      "step": 46250
    },
    {
      "epoch": 3.232563010542484,
      "grad_norm": 0.012579442001879215,
      "learning_rate": 1.9654632057468235e-05,
      "loss": 0.0185,
      "step": 46300
    },
    {
      "epoch": 3.2360538993227674,
      "grad_norm": 0.025501690804958344,
      "learning_rate": 1.9615844103455232e-05,
      "loss": 0.0146,
      "step": 46350
    },
    {
      "epoch": 3.2395447881030512,
      "grad_norm": 0.026744967326521873,
      "learning_rate": 1.957705614944223e-05,
      "loss": 0.029,
      "step": 46400
    },
    {
      "epoch": 3.2430356768833346,
      "grad_norm": 0.45471131801605225,
      "learning_rate": 1.953826819542923e-05,
      "loss": 0.0241,
      "step": 46450
    },
    {
      "epoch": 3.246526565663618,
      "grad_norm": 0.0749790370464325,
      "learning_rate": 1.9499480241416225e-05,
      "loss": 0.0145,
      "step": 46500
    },
    {
      "epoch": 3.2500174544439013,
      "grad_norm": 0.06732578575611115,
      "learning_rate": 1.9460692287403225e-05,
      "loss": 0.0172,
      "step": 46550
    },
    {
      "epoch": 3.2535083432241847,
      "grad_norm": 0.022160114720463753,
      "learning_rate": 1.9421904333390222e-05,
      "loss": 0.0233,
      "step": 46600
    },
    {
      "epoch": 3.2569992320044685,
      "grad_norm": 0.010378474369645119,
      "learning_rate": 1.9383116379377222e-05,
      "loss": 0.0159,
      "step": 46650
    },
    {
      "epoch": 3.260490120784752,
      "grad_norm": 0.018741408362984657,
      "learning_rate": 1.934432842536422e-05,
      "loss": 0.0185,
      "step": 46700
    },
    {
      "epoch": 3.2639810095650352,
      "grad_norm": 7.819186687469482,
      "learning_rate": 1.930554047135122e-05,
      "loss": 0.0191,
      "step": 46750
    },
    {
      "epoch": 3.2674718983453186,
      "grad_norm": 0.05615635961294174,
      "learning_rate": 1.9266752517338216e-05,
      "loss": 0.0144,
      "step": 46800
    },
    {
      "epoch": 3.270962787125602,
      "grad_norm": 0.03543762490153313,
      "learning_rate": 1.9227964563325216e-05,
      "loss": 0.0117,
      "step": 46850
    },
    {
      "epoch": 3.2744536759058858,
      "grad_norm": 0.02200508303940296,
      "learning_rate": 1.9189176609312212e-05,
      "loss": 0.0123,
      "step": 46900
    },
    {
      "epoch": 3.277944564686169,
      "grad_norm": 0.06120089441537857,
      "learning_rate": 1.9150388655299212e-05,
      "loss": 0.0185,
      "step": 46950
    },
    {
      "epoch": 3.2814354534664525,
      "grad_norm": 0.012565530836582184,
      "learning_rate": 1.911160070128621e-05,
      "loss": 0.007,
      "step": 47000
    },
    {
      "epoch": 3.2814354534664525,
      "eval_f1": 0.9401757867876382,
      "eval_loss": 0.1720879226922989,
      "eval_runtime": 40.054,
      "eval_samples_per_second": 91.377,
      "eval_steps_per_second": 2.871,
      "step": 47000
    },
    {
      "epoch": 3.284926342246736,
      "grad_norm": 0.01494626235216856,
      "learning_rate": 1.907281274727321e-05,
      "loss": 0.0246,
      "step": 47050
    },
    {
      "epoch": 3.2884172310270197,
      "grad_norm": 0.023222725838422775,
      "learning_rate": 1.9034024793260206e-05,
      "loss": 0.0107,
      "step": 47100
    },
    {
      "epoch": 3.291908119807303,
      "grad_norm": 0.13517247140407562,
      "learning_rate": 1.8995236839247206e-05,
      "loss": 0.0241,
      "step": 47150
    },
    {
      "epoch": 3.2953990085875864,
      "grad_norm": 3.621380567550659,
      "learning_rate": 1.895722464431446e-05,
      "loss": 0.0354,
      "step": 47200
    },
    {
      "epoch": 3.2988898973678698,
      "grad_norm": 9.058627128601074,
      "learning_rate": 1.891843669030146e-05,
      "loss": 0.0428,
      "step": 47250
    },
    {
      "epoch": 3.302380786148153,
      "grad_norm": 0.03342050686478615,
      "learning_rate": 1.887964873628846e-05,
      "loss": 0.0189,
      "step": 47300
    },
    {
      "epoch": 3.305871674928437,
      "grad_norm": 0.00948148313909769,
      "learning_rate": 1.8840860782275457e-05,
      "loss": 0.0102,
      "step": 47350
    },
    {
      "epoch": 3.3093625637087203,
      "grad_norm": 0.10229407250881195,
      "learning_rate": 1.8802072828262457e-05,
      "loss": 0.0138,
      "step": 47400
    },
    {
      "epoch": 3.3128534524890036,
      "grad_norm": 5.57817268371582,
      "learning_rate": 1.8763284874249454e-05,
      "loss": 0.0327,
      "step": 47450
    },
    {
      "epoch": 3.316344341269287,
      "grad_norm": 0.01957140862941742,
      "learning_rate": 1.8724496920236454e-05,
      "loss": 0.0172,
      "step": 47500
    },
    {
      "epoch": 3.319835230049571,
      "grad_norm": 6.440301418304443,
      "learning_rate": 1.868570896622345e-05,
      "loss": 0.0186,
      "step": 47550
    },
    {
      "epoch": 3.323326118829854,
      "grad_norm": 0.012590606696903706,
      "learning_rate": 1.864692101221045e-05,
      "loss": 0.0159,
      "step": 47600
    },
    {
      "epoch": 3.3268170076101375,
      "grad_norm": 6.431277751922607,
      "learning_rate": 1.8608133058197448e-05,
      "loss": 0.0154,
      "step": 47650
    },
    {
      "epoch": 3.330307896390421,
      "grad_norm": 0.14014244079589844,
      "learning_rate": 1.8569345104184448e-05,
      "loss": 0.0149,
      "step": 47700
    },
    {
      "epoch": 3.3337987851707043,
      "grad_norm": 0.02405048906803131,
      "learning_rate": 1.8530557150171444e-05,
      "loss": 0.0219,
      "step": 47750
    },
    {
      "epoch": 3.337289673950988,
      "grad_norm": 4.332936763763428,
      "learning_rate": 1.849176919615844e-05,
      "loss": 0.026,
      "step": 47800
    },
    {
      "epoch": 3.3407805627312714,
      "grad_norm": 3.507913827896118,
      "learning_rate": 1.845298124214544e-05,
      "loss": 0.022,
      "step": 47850
    },
    {
      "epoch": 3.344271451511555,
      "grad_norm": 0.006527060177177191,
      "learning_rate": 1.8414193288132438e-05,
      "loss": 0.0186,
      "step": 47900
    },
    {
      "epoch": 3.347762340291838,
      "grad_norm": 0.1505638211965561,
      "learning_rate": 1.8375405334119435e-05,
      "loss": 0.0127,
      "step": 47950
    },
    {
      "epoch": 3.351253229072122,
      "grad_norm": 2.856977701187134,
      "learning_rate": 1.8336617380106435e-05,
      "loss": 0.0207,
      "step": 48000
    },
    {
      "epoch": 3.351253229072122,
      "eval_f1": 0.9406025824964132,
      "eval_loss": 0.18208463490009308,
      "eval_runtime": 20.8293,
      "eval_samples_per_second": 175.714,
      "eval_steps_per_second": 5.521,
      "step": 48000
    },
    {
      "epoch": 3.3547441178524053,
      "grad_norm": 0.006940176710486412,
      "learning_rate": 1.829782942609343e-05,
      "loss": 0.0085,
      "step": 48050
    },
    {
      "epoch": 3.3582350066326887,
      "grad_norm": 0.6002283096313477,
      "learning_rate": 1.825904147208043e-05,
      "loss": 0.0147,
      "step": 48100
    },
    {
      "epoch": 3.361725895412972,
      "grad_norm": 0.01274219062179327,
      "learning_rate": 1.8220253518067428e-05,
      "loss": 0.0151,
      "step": 48150
    },
    {
      "epoch": 3.3652167841932554,
      "grad_norm": 0.007482277695089579,
      "learning_rate": 1.8181465564054428e-05,
      "loss": 0.0181,
      "step": 48200
    },
    {
      "epoch": 3.3687076729735392,
      "grad_norm": 0.03832433000206947,
      "learning_rate": 1.8142677610041425e-05,
      "loss": 0.0308,
      "step": 48250
    },
    {
      "epoch": 3.3721985617538226,
      "grad_norm": 0.04535474628210068,
      "learning_rate": 1.8103889656028425e-05,
      "loss": 0.023,
      "step": 48300
    },
    {
      "epoch": 3.375689450534106,
      "grad_norm": 0.016268501058220863,
      "learning_rate": 1.806510170201542e-05,
      "loss": 0.0218,
      "step": 48350
    },
    {
      "epoch": 3.3791803393143893,
      "grad_norm": 5.86330509185791,
      "learning_rate": 1.802631374800242e-05,
      "loss": 0.029,
      "step": 48400
    },
    {
      "epoch": 3.382671228094673,
      "grad_norm": 0.035588301718235016,
      "learning_rate": 1.798752579398942e-05,
      "loss": 0.0349,
      "step": 48450
    },
    {
      "epoch": 3.3861621168749565,
      "grad_norm": 0.04482724890112877,
      "learning_rate": 1.794873783997642e-05,
      "loss": 0.0198,
      "step": 48500
    },
    {
      "epoch": 3.38965300565524,
      "grad_norm": 1.3957189321517944,
      "learning_rate": 1.790994988596342e-05,
      "loss": 0.0153,
      "step": 48550
    },
    {
      "epoch": 3.3931438944355232,
      "grad_norm": 0.030972935259342194,
      "learning_rate": 1.7871161931950415e-05,
      "loss": 0.014,
      "step": 48600
    },
    {
      "epoch": 3.3966347832158066,
      "grad_norm": 0.024676023051142693,
      "learning_rate": 1.7832373977937415e-05,
      "loss": 0.0158,
      "step": 48650
    },
    {
      "epoch": 3.4001256719960904,
      "grad_norm": 0.020046940073370934,
      "learning_rate": 1.7793586023924412e-05,
      "loss": 0.0226,
      "step": 48700
    },
    {
      "epoch": 3.4036165607763738,
      "grad_norm": 0.11487656086683273,
      "learning_rate": 1.775479806991141e-05,
      "loss": 0.019,
      "step": 48750
    },
    {
      "epoch": 3.407107449556657,
      "grad_norm": 0.07113959640264511,
      "learning_rate": 1.771601011589841e-05,
      "loss": 0.0116,
      "step": 48800
    },
    {
      "epoch": 3.4105983383369405,
      "grad_norm": 0.020373065024614334,
      "learning_rate": 1.7677222161885405e-05,
      "loss": 0.0234,
      "step": 48850
    },
    {
      "epoch": 3.4140892271172243,
      "grad_norm": 0.2534826099872589,
      "learning_rate": 1.7638434207872402e-05,
      "loss": 0.0127,
      "step": 48900
    },
    {
      "epoch": 3.4175801158975077,
      "grad_norm": 0.00617890153080225,
      "learning_rate": 1.7599646253859402e-05,
      "loss": 0.0093,
      "step": 48950
    },
    {
      "epoch": 3.421071004677791,
      "grad_norm": 0.12724170088768005,
      "learning_rate": 1.75608582998464e-05,
      "loss": 0.0136,
      "step": 49000
    },
    {
      "epoch": 3.421071004677791,
      "eval_f1": 0.9272142648348436,
      "eval_loss": 0.24522946774959564,
      "eval_runtime": 19.1642,
      "eval_samples_per_second": 190.981,
      "eval_steps_per_second": 6.001,
      "step": 49000
    },
    {
      "epoch": 3.4245618934580744,
      "grad_norm": 0.012064763344824314,
      "learning_rate": 1.75220703458334e-05,
      "loss": 0.0289,
      "step": 49050
    },
    {
      "epoch": 3.4280527822383577,
      "grad_norm": 0.21019980311393738,
      "learning_rate": 1.7483282391820396e-05,
      "loss": 0.0275,
      "step": 49100
    },
    {
      "epoch": 3.431543671018641,
      "grad_norm": 0.009219990111887455,
      "learning_rate": 1.7444494437807396e-05,
      "loss": 0.0265,
      "step": 49150
    },
    {
      "epoch": 3.435034559798925,
      "grad_norm": 0.44924938678741455,
      "learning_rate": 1.7405706483794392e-05,
      "loss": 0.0222,
      "step": 49200
    },
    {
      "epoch": 3.4385254485792083,
      "grad_norm": 0.12933841347694397,
      "learning_rate": 1.7366918529781392e-05,
      "loss": 0.0102,
      "step": 49250
    },
    {
      "epoch": 3.4420163373594916,
      "grad_norm": 0.07747853547334671,
      "learning_rate": 1.732813057576839e-05,
      "loss": 0.0122,
      "step": 49300
    },
    {
      "epoch": 3.445507226139775,
      "grad_norm": 2.5203640460968018,
      "learning_rate": 1.728934262175539e-05,
      "loss": 0.0183,
      "step": 49350
    },
    {
      "epoch": 3.448998114920059,
      "grad_norm": 0.3809491693973541,
      "learning_rate": 1.7250554667742386e-05,
      "loss": 0.0099,
      "step": 49400
    },
    {
      "epoch": 3.452489003700342,
      "grad_norm": 1.1953871250152588,
      "learning_rate": 1.7211766713729386e-05,
      "loss": 0.0173,
      "step": 49450
    },
    {
      "epoch": 3.4559798924806255,
      "grad_norm": 0.01619006134569645,
      "learning_rate": 1.7172978759716383e-05,
      "loss": 0.0372,
      "step": 49500
    },
    {
      "epoch": 3.459470781260909,
      "grad_norm": 0.032744113355875015,
      "learning_rate": 1.7134190805703383e-05,
      "loss": 0.0192,
      "step": 49550
    },
    {
      "epoch": 3.4629616700411923,
      "grad_norm": 0.015268692746758461,
      "learning_rate": 1.709540285169038e-05,
      "loss": 0.0201,
      "step": 49600
    },
    {
      "epoch": 3.466452558821476,
      "grad_norm": 0.4338730573654175,
      "learning_rate": 1.7057390656757637e-05,
      "loss": 0.0157,
      "step": 49650
    },
    {
      "epoch": 3.4699434476017594,
      "grad_norm": 0.1488252431154251,
      "learning_rate": 1.7018602702744638e-05,
      "loss": 0.0326,
      "step": 49700
    },
    {
      "epoch": 3.473434336382043,
      "grad_norm": 0.35929903388023376,
      "learning_rate": 1.6979814748731634e-05,
      "loss": 0.0084,
      "step": 49750
    },
    {
      "epoch": 3.476925225162326,
      "grad_norm": 0.09212347865104675,
      "learning_rate": 1.6941026794718634e-05,
      "loss": 0.0172,
      "step": 49800
    },
    {
      "epoch": 3.48041611394261,
      "grad_norm": 6.833462238311768,
      "learning_rate": 1.690223884070563e-05,
      "loss": 0.011,
      "step": 49850
    },
    {
      "epoch": 3.4839070027228933,
      "grad_norm": 0.06177880987524986,
      "learning_rate": 1.686345088669263e-05,
      "loss": 0.022,
      "step": 49900
    },
    {
      "epoch": 3.4873978915031767,
      "grad_norm": 0.015737321227788925,
      "learning_rate": 1.6824662932679628e-05,
      "loss": 0.0159,
      "step": 49950
    },
    {
      "epoch": 3.49088878028346,
      "grad_norm": 1.1112961769104004,
      "learning_rate": 1.6785874978666628e-05,
      "loss": 0.0113,
      "step": 50000
    },
    {
      "epoch": 3.49088878028346,
      "eval_f1": 0.9358381502890173,
      "eval_loss": 0.1911371946334839,
      "eval_runtime": 20.4886,
      "eval_samples_per_second": 178.636,
      "eval_steps_per_second": 5.613,
      "step": 50000
    },
    {
      "epoch": 3.4943796690637434,
      "grad_norm": 0.009405792690813541,
      "learning_rate": 1.6747087024653624e-05,
      "loss": 0.0201,
      "step": 50050
    },
    {
      "epoch": 3.4978705578440272,
      "grad_norm": 0.03674100339412689,
      "learning_rate": 1.6708299070640625e-05,
      "loss": 0.0269,
      "step": 50100
    },
    {
      "epoch": 3.5013614466243106,
      "grad_norm": 2.9473278522491455,
      "learning_rate": 1.666951111662762e-05,
      "loss": 0.0197,
      "step": 50150
    },
    {
      "epoch": 3.504852335404594,
      "grad_norm": 0.025787200778722763,
      "learning_rate": 1.663072316261462e-05,
      "loss": 0.0177,
      "step": 50200
    },
    {
      "epoch": 3.5083432241848773,
      "grad_norm": 0.025555545464158058,
      "learning_rate": 1.6591935208601618e-05,
      "loss": 0.0172,
      "step": 50250
    },
    {
      "epoch": 3.511834112965161,
      "grad_norm": 0.033668939024209976,
      "learning_rate": 1.6553147254588615e-05,
      "loss": 0.0139,
      "step": 50300
    },
    {
      "epoch": 3.5153250017454445,
      "grad_norm": 0.03965609520673752,
      "learning_rate": 1.6514359300575615e-05,
      "loss": 0.0163,
      "step": 50350
    },
    {
      "epoch": 3.518815890525728,
      "grad_norm": 0.02201533503830433,
      "learning_rate": 1.647557134656261e-05,
      "loss": 0.0183,
      "step": 50400
    },
    {
      "epoch": 3.522306779306011,
      "grad_norm": 0.02458702027797699,
      "learning_rate": 1.6436783392549608e-05,
      "loss": 0.0033,
      "step": 50450
    },
    {
      "epoch": 3.5257976680862946,
      "grad_norm": 0.03175359591841698,
      "learning_rate": 1.6397995438536608e-05,
      "loss": 0.023,
      "step": 50500
    },
    {
      "epoch": 3.5292885568665784,
      "grad_norm": 0.12282973527908325,
      "learning_rate": 1.6359207484523605e-05,
      "loss": 0.0231,
      "step": 50550
    },
    {
      "epoch": 3.5327794456468617,
      "grad_norm": 0.09555836021900177,
      "learning_rate": 1.6320419530510605e-05,
      "loss": 0.0225,
      "step": 50600
    },
    {
      "epoch": 3.536270334427145,
      "grad_norm": 0.03125094249844551,
      "learning_rate": 1.62816315764976e-05,
      "loss": 0.0057,
      "step": 50650
    },
    {
      "epoch": 3.5397612232074285,
      "grad_norm": 0.42508402466773987,
      "learning_rate": 1.6242843622484602e-05,
      "loss": 0.0121,
      "step": 50700
    },
    {
      "epoch": 3.5432521119877123,
      "grad_norm": 0.004737542010843754,
      "learning_rate": 1.6204055668471602e-05,
      "loss": 0.0266,
      "step": 50750
    },
    {
      "epoch": 3.5467430007679956,
      "grad_norm": 0.009997726418077946,
      "learning_rate": 1.61652677144586e-05,
      "loss": 0.0094,
      "step": 50800
    },
    {
      "epoch": 3.550233889548279,
      "grad_norm": 0.10814603418111801,
      "learning_rate": 1.61264797604456e-05,
      "loss": 0.009,
      "step": 50850
    },
    {
      "epoch": 3.5537247783285624,
      "grad_norm": 0.015030721202492714,
      "learning_rate": 1.6087691806432595e-05,
      "loss": 0.0059,
      "step": 50900
    },
    {
      "epoch": 3.5572156671088457,
      "grad_norm": 3.3664567470550537,
      "learning_rate": 1.6048903852419595e-05,
      "loss": 0.0265,
      "step": 50950
    },
    {
      "epoch": 3.560706555889129,
      "grad_norm": 0.037877023220062256,
      "learning_rate": 1.6010115898406592e-05,
      "loss": 0.0128,
      "step": 51000
    },
    {
      "epoch": 3.560706555889129,
      "eval_f1": 0.9240691879214307,
      "eval_loss": 0.2226642370223999,
      "eval_runtime": 19.3417,
      "eval_samples_per_second": 189.229,
      "eval_steps_per_second": 5.946,
      "step": 51000
    },
    {
      "epoch": 3.564197444669413,
      "grad_norm": 0.18292999267578125,
      "learning_rate": 1.5971327944393592e-05,
      "loss": 0.028,
      "step": 51050
    },
    {
      "epoch": 3.5676883334496963,
      "grad_norm": 0.03300943225622177,
      "learning_rate": 1.593253999038059e-05,
      "loss": 0.0178,
      "step": 51100
    },
    {
      "epoch": 3.5711792222299796,
      "grad_norm": 0.26565834879875183,
      "learning_rate": 1.589375203636759e-05,
      "loss": 0.0156,
      "step": 51150
    },
    {
      "epoch": 3.5746701110102634,
      "grad_norm": 0.006750601809471846,
      "learning_rate": 1.5854964082354586e-05,
      "loss": 0.0112,
      "step": 51200
    },
    {
      "epoch": 3.578160999790547,
      "grad_norm": 0.09060013294219971,
      "learning_rate": 1.5816176128341582e-05,
      "loss": 0.0323,
      "step": 51250
    },
    {
      "epoch": 3.58165188857083,
      "grad_norm": 2.0851974487304688,
      "learning_rate": 1.5777388174328582e-05,
      "loss": 0.0206,
      "step": 51300
    },
    {
      "epoch": 3.5851427773511135,
      "grad_norm": 0.012320557609200478,
      "learning_rate": 1.573860022031558e-05,
      "loss": 0.0157,
      "step": 51350
    },
    {
      "epoch": 3.588633666131397,
      "grad_norm": 19.57718276977539,
      "learning_rate": 1.5699812266302576e-05,
      "loss": 0.016,
      "step": 51400
    },
    {
      "epoch": 3.5921245549116803,
      "grad_norm": 0.014298966154456139,
      "learning_rate": 1.5661024312289576e-05,
      "loss": 0.0107,
      "step": 51450
    },
    {
      "epoch": 3.595615443691964,
      "grad_norm": 5.5826849937438965,
      "learning_rate": 1.5622236358276572e-05,
      "loss": 0.0175,
      "step": 51500
    },
    {
      "epoch": 3.5991063324722474,
      "grad_norm": 0.011979132890701294,
      "learning_rate": 1.5583448404263573e-05,
      "loss": 0.0246,
      "step": 51550
    },
    {
      "epoch": 3.602597221252531,
      "grad_norm": 0.021864114329218864,
      "learning_rate": 1.554466045025057e-05,
      "loss": 0.0199,
      "step": 51600
    },
    {
      "epoch": 3.6060881100328146,
      "grad_norm": 0.02277781441807747,
      "learning_rate": 1.550587249623757e-05,
      "loss": 0.0091,
      "step": 51650
    },
    {
      "epoch": 3.609578998813098,
      "grad_norm": 0.013841581530869007,
      "learning_rate": 1.5467084542224566e-05,
      "loss": 0.0059,
      "step": 51700
    },
    {
      "epoch": 3.6130698875933813,
      "grad_norm": 0.012927171774208546,
      "learning_rate": 1.5429072347291827e-05,
      "loss": 0.0201,
      "step": 51750
    },
    {
      "epoch": 3.6165607763736647,
      "grad_norm": 7.824684143066406,
      "learning_rate": 1.5390284393278824e-05,
      "loss": 0.0176,
      "step": 51800
    },
    {
      "epoch": 3.620051665153948,
      "grad_norm": 0.16372781991958618,
      "learning_rate": 1.535149643926582e-05,
      "loss": 0.0313,
      "step": 51850
    },
    {
      "epoch": 3.6235425539342314,
      "grad_norm": 0.11191274225711823,
      "learning_rate": 1.531270848525282e-05,
      "loss": 0.0192,
      "step": 51900
    },
    {
      "epoch": 3.627033442714515,
      "grad_norm": 0.09052783995866776,
      "learning_rate": 1.5273920531239818e-05,
      "loss": 0.015,
      "step": 51950
    },
    {
      "epoch": 3.6305243314947986,
      "grad_norm": 3.989870071411133,
      "learning_rate": 1.5235132577226818e-05,
      "loss": 0.0173,
      "step": 52000
    },
    {
      "epoch": 3.6305243314947986,
      "eval_f1": 0.9422750424448217,
      "eval_loss": 0.1699497252702713,
      "eval_runtime": 18.403,
      "eval_samples_per_second": 198.881,
      "eval_steps_per_second": 6.249,
      "step": 52000
    },
    {
      "epoch": 3.634015220275082,
      "grad_norm": 0.004953749943524599,
      "learning_rate": 1.5196344623213816e-05,
      "loss": 0.0073,
      "step": 52050
    },
    {
      "epoch": 3.6375061090553658,
      "grad_norm": 6.198831558227539,
      "learning_rate": 1.5157556669200814e-05,
      "loss": 0.026,
      "step": 52100
    },
    {
      "epoch": 3.640996997835649,
      "grad_norm": 0.018628481775522232,
      "learning_rate": 1.5118768715187811e-05,
      "loss": 0.0173,
      "step": 52150
    },
    {
      "epoch": 3.6444878866159325,
      "grad_norm": 0.025905922055244446,
      "learning_rate": 1.5079980761174811e-05,
      "loss": 0.0094,
      "step": 52200
    },
    {
      "epoch": 3.647978775396216,
      "grad_norm": 0.00478333467617631,
      "learning_rate": 1.5041192807161808e-05,
      "loss": 0.0182,
      "step": 52250
    },
    {
      "epoch": 3.651469664176499,
      "grad_norm": 3.3075673580169678,
      "learning_rate": 1.5002404853148808e-05,
      "loss": 0.0198,
      "step": 52300
    },
    {
      "epoch": 3.6549605529567826,
      "grad_norm": 0.05998103320598602,
      "learning_rate": 1.4963616899135805e-05,
      "loss": 0.0096,
      "step": 52350
    },
    {
      "epoch": 3.6584514417370664,
      "grad_norm": 4.651539325714111,
      "learning_rate": 1.4924828945122805e-05,
      "loss": 0.0248,
      "step": 52400
    },
    {
      "epoch": 3.6619423305173497,
      "grad_norm": 0.6373411417007446,
      "learning_rate": 1.4886040991109801e-05,
      "loss": 0.0338,
      "step": 52450
    },
    {
      "epoch": 3.665433219297633,
      "grad_norm": 1.2010835409164429,
      "learning_rate": 1.48472530370968e-05,
      "loss": 0.0103,
      "step": 52500
    },
    {
      "epoch": 3.668924108077917,
      "grad_norm": 0.018328318372368813,
      "learning_rate": 1.4808465083083798e-05,
      "loss": 0.0085,
      "step": 52550
    },
    {
      "epoch": 3.6724149968582003,
      "grad_norm": 0.03623636066913605,
      "learning_rate": 1.4769677129070796e-05,
      "loss": 0.0119,
      "step": 52600
    },
    {
      "epoch": 3.6759058856384836,
      "grad_norm": 0.05136825889348984,
      "learning_rate": 1.4730889175057793e-05,
      "loss": 0.0118,
      "step": 52650
    },
    {
      "epoch": 3.679396774418767,
      "grad_norm": 0.025166302919387817,
      "learning_rate": 1.4692101221044793e-05,
      "loss": 0.0148,
      "step": 52700
    },
    {
      "epoch": 3.6828876631990504,
      "grad_norm": 0.2804352045059204,
      "learning_rate": 1.465331326703179e-05,
      "loss": 0.0291,
      "step": 52750
    },
    {
      "epoch": 3.6863785519793337,
      "grad_norm": 0.01500253938138485,
      "learning_rate": 1.461452531301879e-05,
      "loss": 0.0157,
      "step": 52800
    },
    {
      "epoch": 3.6898694407596175,
      "grad_norm": 0.012079290114343166,
      "learning_rate": 1.4575737359005787e-05,
      "loss": 0.0143,
      "step": 52850
    },
    {
      "epoch": 3.693360329539901,
      "grad_norm": 0.09183007478713989,
      "learning_rate": 1.4536949404992787e-05,
      "loss": 0.0203,
      "step": 52900
    },
    {
      "epoch": 3.6968512183201843,
      "grad_norm": 0.020523011684417725,
      "learning_rate": 1.4498161450979783e-05,
      "loss": 0.0208,
      "step": 52950
    },
    {
      "epoch": 3.7003421071004676,
      "grad_norm": 2.2191498279571533,
      "learning_rate": 1.4459373496966783e-05,
      "loss": 0.0115,
      "step": 53000
    },
    {
      "epoch": 3.7003421071004676,
      "eval_f1": 0.9403368541250356,
      "eval_loss": 0.17281506955623627,
      "eval_runtime": 18.7663,
      "eval_samples_per_second": 195.031,
      "eval_steps_per_second": 6.128,
      "step": 53000
    },
    {
      "epoch": 3.7038329958807514,
      "grad_norm": 0.11821743845939636,
      "learning_rate": 1.4420585542953782e-05,
      "loss": 0.0144,
      "step": 53050
    },
    {
      "epoch": 3.707323884661035,
      "grad_norm": 0.015602991916239262,
      "learning_rate": 1.4381797588940779e-05,
      "loss": 0.0281,
      "step": 53100
    },
    {
      "epoch": 3.710814773441318,
      "grad_norm": 0.405454158782959,
      "learning_rate": 1.4343009634927779e-05,
      "loss": 0.0161,
      "step": 53150
    },
    {
      "epoch": 3.7143056622216015,
      "grad_norm": 0.7658965587615967,
      "learning_rate": 1.4304221680914775e-05,
      "loss": 0.0292,
      "step": 53200
    },
    {
      "epoch": 3.717796551001885,
      "grad_norm": 3.5675363540649414,
      "learning_rate": 1.4265433726901775e-05,
      "loss": 0.0109,
      "step": 53250
    },
    {
      "epoch": 3.7212874397821687,
      "grad_norm": 0.032014861702919006,
      "learning_rate": 1.4226645772888772e-05,
      "loss": 0.0223,
      "step": 53300
    },
    {
      "epoch": 3.724778328562452,
      "grad_norm": 0.12666158378124237,
      "learning_rate": 1.4187857818875772e-05,
      "loss": 0.0254,
      "step": 53350
    },
    {
      "epoch": 3.7282692173427354,
      "grad_norm": 0.029635407030582428,
      "learning_rate": 1.4149069864862769e-05,
      "loss": 0.0045,
      "step": 53400
    },
    {
      "epoch": 3.731760106123019,
      "grad_norm": 2.7315502166748047,
      "learning_rate": 1.4110281910849767e-05,
      "loss": 0.0421,
      "step": 53450
    },
    {
      "epoch": 3.7352509949033026,
      "grad_norm": 0.5258963704109192,
      "learning_rate": 1.4071493956836766e-05,
      "loss": 0.0206,
      "step": 53500
    },
    {
      "epoch": 3.738741883683586,
      "grad_norm": 0.05045127496123314,
      "learning_rate": 1.4032706002823764e-05,
      "loss": 0.0122,
      "step": 53550
    },
    {
      "epoch": 3.7422327724638693,
      "grad_norm": 0.008424119092524052,
      "learning_rate": 1.399391804881076e-05,
      "loss": 0.008,
      "step": 53600
    },
    {
      "epoch": 3.7457236612441527,
      "grad_norm": 0.026001902297139168,
      "learning_rate": 1.395513009479776e-05,
      "loss": 0.0204,
      "step": 53650
    },
    {
      "epoch": 3.749214550024436,
      "grad_norm": 0.009944004938006401,
      "learning_rate": 1.3916342140784757e-05,
      "loss": 0.0109,
      "step": 53700
    },
    {
      "epoch": 3.7527054388047194,
      "grad_norm": 0.48391270637512207,
      "learning_rate": 1.3877554186771758e-05,
      "loss": 0.0228,
      "step": 53750
    },
    {
      "epoch": 3.756196327585003,
      "grad_norm": 6.483001708984375,
      "learning_rate": 1.3838766232758754e-05,
      "loss": 0.0245,
      "step": 53800
    },
    {
      "epoch": 3.7596872163652866,
      "grad_norm": 0.23561638593673706,
      "learning_rate": 1.3799978278745754e-05,
      "loss": 0.0072,
      "step": 53850
    },
    {
      "epoch": 3.76317810514557,
      "grad_norm": 2.020230770111084,
      "learning_rate": 1.3761190324732751e-05,
      "loss": 0.0209,
      "step": 53900
    },
    {
      "epoch": 3.7666689939258537,
      "grad_norm": 2.9172303676605225,
      "learning_rate": 1.3722402370719751e-05,
      "loss": 0.0229,
      "step": 53950
    },
    {
      "epoch": 3.770159882706137,
      "grad_norm": 0.08300653100013733,
      "learning_rate": 1.3683614416706748e-05,
      "loss": 0.0226,
      "step": 54000
    },
    {
      "epoch": 3.770159882706137,
      "eval_f1": 0.9282156460591855,
      "eval_loss": 0.2161129117012024,
      "eval_runtime": 27.4237,
      "eval_samples_per_second": 133.461,
      "eval_steps_per_second": 4.193,
      "step": 54000
    },
    {
      "epoch": 3.7736507714864205,
      "grad_norm": 0.051579758524894714,
      "learning_rate": 1.3644826462693746e-05,
      "loss": 0.0128,
      "step": 54050
    },
    {
      "epoch": 3.777141660266704,
      "grad_norm": 0.029366178438067436,
      "learning_rate": 1.3606038508680743e-05,
      "loss": 0.0116,
      "step": 54100
    },
    {
      "epoch": 3.780632549046987,
      "grad_norm": 0.03355459123849869,
      "learning_rate": 1.3567250554667743e-05,
      "loss": 0.0171,
      "step": 54150
    },
    {
      "epoch": 3.7841234378272706,
      "grad_norm": 13.578819274902344,
      "learning_rate": 1.3528462600654743e-05,
      "loss": 0.0133,
      "step": 54200
    },
    {
      "epoch": 3.7876143266075544,
      "grad_norm": 0.4114447236061096,
      "learning_rate": 1.348967464664174e-05,
      "loss": 0.0133,
      "step": 54250
    },
    {
      "epoch": 3.7911052153878377,
      "grad_norm": 2.2274863719940186,
      "learning_rate": 1.345088669262874e-05,
      "loss": 0.0135,
      "step": 54300
    },
    {
      "epoch": 3.794596104168121,
      "grad_norm": 0.016799284145236015,
      "learning_rate": 1.3412098738615736e-05,
      "loss": 0.0103,
      "step": 54350
    },
    {
      "epoch": 3.798086992948405,
      "grad_norm": 0.021271638572216034,
      "learning_rate": 1.3373310784602735e-05,
      "loss": 0.0144,
      "step": 54400
    },
    {
      "epoch": 3.8015778817286883,
      "grad_norm": 0.40475618839263916,
      "learning_rate": 1.3334522830589733e-05,
      "loss": 0.0157,
      "step": 54450
    },
    {
      "epoch": 3.8050687705089716,
      "grad_norm": 0.014505240134894848,
      "learning_rate": 1.3295734876576732e-05,
      "loss": 0.0214,
      "step": 54500
    },
    {
      "epoch": 3.808559659289255,
      "grad_norm": 1.2870405912399292,
      "learning_rate": 1.3256946922563728e-05,
      "loss": 0.0199,
      "step": 54550
    },
    {
      "epoch": 3.8120505480695384,
      "grad_norm": 0.11923400312662125,
      "learning_rate": 1.3218158968550728e-05,
      "loss": 0.019,
      "step": 54600
    },
    {
      "epoch": 3.8155414368498217,
      "grad_norm": 6.796934127807617,
      "learning_rate": 1.3180146773617985e-05,
      "loss": 0.0189,
      "step": 54650
    },
    {
      "epoch": 3.8190323256301055,
      "grad_norm": 0.2147059142589569,
      "learning_rate": 1.3141358819604985e-05,
      "loss": 0.0148,
      "step": 54700
    },
    {
      "epoch": 3.822523214410389,
      "grad_norm": 0.009229235351085663,
      "learning_rate": 1.3102570865591981e-05,
      "loss": 0.0125,
      "step": 54750
    },
    {
      "epoch": 3.8260141031906723,
      "grad_norm": 0.012158773839473724,
      "learning_rate": 1.3063782911578981e-05,
      "loss": 0.0181,
      "step": 54800
    },
    {
      "epoch": 3.829504991970956,
      "grad_norm": 0.9341710209846497,
      "learning_rate": 1.3024994957565978e-05,
      "loss": 0.0118,
      "step": 54850
    },
    {
      "epoch": 3.8329958807512394,
      "grad_norm": 0.1444963812828064,
      "learning_rate": 1.2986207003552978e-05,
      "loss": 0.0219,
      "step": 54900
    },
    {
      "epoch": 3.836486769531523,
      "grad_norm": 0.2622317969799042,
      "learning_rate": 1.2947419049539975e-05,
      "loss": 0.0239,
      "step": 54950
    },
    {
      "epoch": 3.839977658311806,
      "grad_norm": 2.9175851345062256,
      "learning_rate": 1.2908631095526973e-05,
      "loss": 0.0229,
      "step": 55000
    },
    {
      "epoch": 3.839977658311806,
      "eval_f1": 0.9377182770663562,
      "eval_loss": 0.18816331028938293,
      "eval_runtime": 26.4818,
      "eval_samples_per_second": 138.208,
      "eval_steps_per_second": 4.343,
      "step": 55000
    },
    {
      "epoch": 3.8434685470920895,
      "grad_norm": 0.01847333274781704,
      "learning_rate": 1.2869843141513972e-05,
      "loss": 0.0123,
      "step": 55050
    },
    {
      "epoch": 3.846959435872373,
      "grad_norm": 0.14379319548606873,
      "learning_rate": 1.283105518750097e-05,
      "loss": 0.0135,
      "step": 55100
    },
    {
      "epoch": 3.8504503246526567,
      "grad_norm": 0.017605192959308624,
      "learning_rate": 1.2792267233487967e-05,
      "loss": 0.0051,
      "step": 55150
    },
    {
      "epoch": 3.85394121343294,
      "grad_norm": 0.00873847771435976,
      "learning_rate": 1.2753479279474967e-05,
      "loss": 0.0123,
      "step": 55200
    },
    {
      "epoch": 3.8574321022132234,
      "grad_norm": 0.006974588148295879,
      "learning_rate": 1.2714691325461967e-05,
      "loss": 0.0054,
      "step": 55250
    },
    {
      "epoch": 3.860922990993507,
      "grad_norm": 0.008401262573897839,
      "learning_rate": 1.2675903371448964e-05,
      "loss": 0.0115,
      "step": 55300
    },
    {
      "epoch": 3.8644138797737906,
      "grad_norm": 0.004291973542422056,
      "learning_rate": 1.2637115417435964e-05,
      "loss": 0.0233,
      "step": 55350
    },
    {
      "epoch": 3.867904768554074,
      "grad_norm": 0.039114680141210556,
      "learning_rate": 1.259832746342296e-05,
      "loss": 0.0301,
      "step": 55400
    },
    {
      "epoch": 3.8713956573343573,
      "grad_norm": 0.02999526634812355,
      "learning_rate": 1.2559539509409959e-05,
      "loss": 0.021,
      "step": 55450
    },
    {
      "epoch": 3.8748865461146407,
      "grad_norm": 0.14493048191070557,
      "learning_rate": 1.2520751555396955e-05,
      "loss": 0.0155,
      "step": 55500
    },
    {
      "epoch": 3.878377434894924,
      "grad_norm": 0.028276991099119186,
      "learning_rate": 1.2481963601383954e-05,
      "loss": 0.0093,
      "step": 55550
    },
    {
      "epoch": 3.881868323675208,
      "grad_norm": 0.984074056148529,
      "learning_rate": 1.2443175647370952e-05,
      "loss": 0.0121,
      "step": 55600
    },
    {
      "epoch": 3.885359212455491,
      "grad_norm": 0.006328861694782972,
      "learning_rate": 1.240438769335795e-05,
      "loss": 0.0144,
      "step": 55650
    },
    {
      "epoch": 3.8888501012357746,
      "grad_norm": 0.013225522823631763,
      "learning_rate": 1.2365599739344949e-05,
      "loss": 0.0112,
      "step": 55700
    },
    {
      "epoch": 3.892340990016058,
      "grad_norm": 0.16436105966567993,
      "learning_rate": 1.2326811785331947e-05,
      "loss": 0.0507,
      "step": 55750
    },
    {
      "epoch": 3.8958318787963417,
      "grad_norm": 0.026833226904273033,
      "learning_rate": 1.2288023831318946e-05,
      "loss": 0.0139,
      "step": 55800
    },
    {
      "epoch": 3.899322767576625,
      "grad_norm": 0.4603937268257141,
      "learning_rate": 1.2249235877305946e-05,
      "loss": 0.0146,
      "step": 55850
    },
    {
      "epoch": 3.9028136563569085,
      "grad_norm": 0.2762994170188904,
      "learning_rate": 1.2210447923292944e-05,
      "loss": 0.0118,
      "step": 55900
    },
    {
      "epoch": 3.906304545137192,
      "grad_norm": 0.040168821811676025,
      "learning_rate": 1.217165996927994e-05,
      "loss": 0.0176,
      "step": 55950
    },
    {
      "epoch": 3.909795433917475,
      "grad_norm": 0.010930743999779224,
      "learning_rate": 1.213287201526694e-05,
      "loss": 0.0073,
      "step": 56000
    },
    {
      "epoch": 3.909795433917475,
      "eval_f1": 0.9395429563205091,
      "eval_loss": 0.1781720668077469,
      "eval_runtime": 28.9085,
      "eval_samples_per_second": 126.607,
      "eval_steps_per_second": 3.978,
      "step": 56000
    },
    {
      "epoch": 3.913286322697759,
      "grad_norm": 0.014746199361979961,
      "learning_rate": 1.2094084061253938e-05,
      "loss": 0.0127,
      "step": 56050
    },
    {
      "epoch": 3.9167772114780424,
      "grad_norm": 0.014964771457016468,
      "learning_rate": 1.2055296107240936e-05,
      "loss": 0.0213,
      "step": 56100
    },
    {
      "epoch": 3.9202681002583257,
      "grad_norm": 1.6473639011383057,
      "learning_rate": 1.2016508153227934e-05,
      "loss": 0.0254,
      "step": 56150
    },
    {
      "epoch": 3.923758989038609,
      "grad_norm": 0.1399172842502594,
      "learning_rate": 1.1977720199214933e-05,
      "loss": 0.016,
      "step": 56200
    },
    {
      "epoch": 3.927249877818893,
      "grad_norm": 0.01769440248608589,
      "learning_rate": 1.1938932245201931e-05,
      "loss": 0.0232,
      "step": 56250
    },
    {
      "epoch": 3.9307407665991763,
      "grad_norm": 0.012674226425588131,
      "learning_rate": 1.190014429118893e-05,
      "loss": 0.0211,
      "step": 56300
    },
    {
      "epoch": 3.9342316553794596,
      "grad_norm": 6.944097518920898,
      "learning_rate": 1.1861356337175928e-05,
      "loss": 0.016,
      "step": 56350
    },
    {
      "epoch": 3.937722544159743,
      "grad_norm": 0.009510631673038006,
      "learning_rate": 1.1822568383162926e-05,
      "loss": 0.0236,
      "step": 56400
    },
    {
      "epoch": 3.9412134329400264,
      "grad_norm": 0.016936538740992546,
      "learning_rate": 1.1783780429149923e-05,
      "loss": 0.0301,
      "step": 56450
    },
    {
      "epoch": 3.9447043217203097,
      "grad_norm": 0.010942582041025162,
      "learning_rate": 1.1744992475136921e-05,
      "loss": 0.0082,
      "step": 56500
    },
    {
      "epoch": 3.9481952105005935,
      "grad_norm": 0.21423903107643127,
      "learning_rate": 1.170620452112392e-05,
      "loss": 0.0122,
      "step": 56550
    },
    {
      "epoch": 3.951686099280877,
      "grad_norm": 0.1346980482339859,
      "learning_rate": 1.1667416567110918e-05,
      "loss": 0.0156,
      "step": 56600
    },
    {
      "epoch": 3.9551769880611602,
      "grad_norm": 0.027757730334997177,
      "learning_rate": 1.1628628613097916e-05,
      "loss": 0.02,
      "step": 56650
    },
    {
      "epoch": 3.958667876841444,
      "grad_norm": 0.05131814628839493,
      "learning_rate": 1.1589840659084915e-05,
      "loss": 0.0097,
      "step": 56700
    },
    {
      "epoch": 3.9621587656217274,
      "grad_norm": 3.791733741760254,
      "learning_rate": 1.1551052705071913e-05,
      "loss": 0.0229,
      "step": 56750
    },
    {
      "epoch": 3.965649654402011,
      "grad_norm": 2.3265886306762695,
      "learning_rate": 1.1512264751058912e-05,
      "loss": 0.0137,
      "step": 56800
    },
    {
      "epoch": 3.969140543182294,
      "grad_norm": 2.308258533477783,
      "learning_rate": 1.147347679704591e-05,
      "loss": 0.0223,
      "step": 56850
    },
    {
      "epoch": 3.9726314319625775,
      "grad_norm": 0.004564803559333086,
      "learning_rate": 1.1434688843032908e-05,
      "loss": 0.0229,
      "step": 56900
    },
    {
      "epoch": 3.976122320742861,
      "grad_norm": 0.06186622008681297,
      "learning_rate": 1.1395900889019907e-05,
      "loss": 0.0118,
      "step": 56950
    },
    {
      "epoch": 3.9796132095231447,
      "grad_norm": 0.010163922794163227,
      "learning_rate": 1.1357888694087165e-05,
      "loss": 0.0116,
      "step": 57000
    },
    {
      "epoch": 3.9796132095231447,
      "eval_f1": 0.9382573571840739,
      "eval_loss": 0.17844714224338531,
      "eval_runtime": 26.9732,
      "eval_samples_per_second": 135.69,
      "eval_steps_per_second": 4.263,
      "step": 57000
    },
    {
      "epoch": 3.983104098303428,
      "grad_norm": 0.06274252384901047,
      "learning_rate": 1.1319100740074163e-05,
      "loss": 0.0119,
      "step": 57050
    },
    {
      "epoch": 3.9865949870837114,
      "grad_norm": 0.014810669235885143,
      "learning_rate": 1.1280312786061162e-05,
      "loss": 0.0028,
      "step": 57100
    },
    {
      "epoch": 3.990085875863995,
      "grad_norm": 0.0769757330417633,
      "learning_rate": 1.124152483204816e-05,
      "loss": 0.0133,
      "step": 57150
    },
    {
      "epoch": 3.9935767646442786,
      "grad_norm": 1.5741026401519775,
      "learning_rate": 1.1202736878035158e-05,
      "loss": 0.0324,
      "step": 57200
    },
    {
      "epoch": 3.997067653424562,
      "grad_norm": 1.7274805307388306,
      "learning_rate": 1.1163948924022157e-05,
      "loss": 0.0128,
      "step": 57250
    },
    {
      "epoch": 4.000558542204845,
      "grad_norm": 0.023459715768694878,
      "learning_rate": 1.1125160970009155e-05,
      "loss": 0.0336,
      "step": 57300
    },
    {
      "epoch": 4.004049430985129,
      "grad_norm": 0.21544323861598969,
      "learning_rate": 1.1086373015996153e-05,
      "loss": 0.0184,
      "step": 57350
    },
    {
      "epoch": 4.007540319765412,
      "grad_norm": 5.495270252227783,
      "learning_rate": 1.1047585061983152e-05,
      "loss": 0.0217,
      "step": 57400
    },
    {
      "epoch": 4.011031208545695,
      "grad_norm": 0.1595708429813385,
      "learning_rate": 1.100879710797015e-05,
      "loss": 0.02,
      "step": 57450
    },
    {
      "epoch": 4.014522097325979,
      "grad_norm": 0.32755452394485474,
      "learning_rate": 1.0970009153957147e-05,
      "loss": 0.0073,
      "step": 57500
    },
    {
      "epoch": 4.018012986106263,
      "grad_norm": 0.03040587343275547,
      "learning_rate": 1.0931221199944145e-05,
      "loss": 0.0189,
      "step": 57550
    },
    {
      "epoch": 4.021503874886546,
      "grad_norm": 0.006620501633733511,
      "learning_rate": 1.0892433245931144e-05,
      "loss": 0.0187,
      "step": 57600
    },
    {
      "epoch": 4.02499476366683,
      "grad_norm": 0.015734750777482986,
      "learning_rate": 1.0853645291918142e-05,
      "loss": 0.0181,
      "step": 57650
    },
    {
      "epoch": 4.028485652447113,
      "grad_norm": 0.08926963061094284,
      "learning_rate": 1.081485733790514e-05,
      "loss": 0.0199,
      "step": 57700
    },
    {
      "epoch": 4.0319765412273965,
      "grad_norm": 0.19889496266841888,
      "learning_rate": 1.0776069383892139e-05,
      "loss": 0.0104,
      "step": 57750
    },
    {
      "epoch": 4.03546743000768,
      "grad_norm": 0.008089990355074406,
      "learning_rate": 1.0737281429879137e-05,
      "loss": 0.0139,
      "step": 57800
    },
    {
      "epoch": 4.038958318787963,
      "grad_norm": 0.006437818054109812,
      "learning_rate": 1.0698493475866136e-05,
      "loss": 0.0173,
      "step": 57850
    },
    {
      "epoch": 4.0424492075682465,
      "grad_norm": 0.008485854603350163,
      "learning_rate": 1.0659705521853134e-05,
      "loss": 0.0193,
      "step": 57900
    },
    {
      "epoch": 4.04594009634853,
      "grad_norm": 0.01528970431536436,
      "learning_rate": 1.0620917567840132e-05,
      "loss": 0.0204,
      "step": 57950
    },
    {
      "epoch": 4.049430985128814,
      "grad_norm": 0.018422499299049377,
      "learning_rate": 1.0582129613827129e-05,
      "loss": 0.0101,
      "step": 58000
    },
    {
      "epoch": 4.049430985128814,
      "eval_f1": 0.9438202247191011,
      "eval_loss": 0.17257019877433777,
      "eval_runtime": 19.6793,
      "eval_samples_per_second": 185.983,
      "eval_steps_per_second": 5.844,
      "step": 58000
    },
    {
      "epoch": 4.0529218739090975,
      "grad_norm": 19.834749221801758,
      "learning_rate": 1.0543341659814127e-05,
      "loss": 0.0299,
      "step": 58050
    },
    {
      "epoch": 4.056412762689381,
      "grad_norm": 0.035838037729263306,
      "learning_rate": 1.0504553705801127e-05,
      "loss": 0.0132,
      "step": 58100
    },
    {
      "epoch": 4.059903651469664,
      "grad_norm": 0.014196299947798252,
      "learning_rate": 1.0465765751788126e-05,
      "loss": 0.0191,
      "step": 58150
    },
    {
      "epoch": 4.063394540249948,
      "grad_norm": 0.10412344336509705,
      "learning_rate": 1.0426977797775124e-05,
      "loss": 0.0193,
      "step": 58200
    },
    {
      "epoch": 4.066885429030231,
      "grad_norm": 0.012938785366714,
      "learning_rate": 1.0388189843762123e-05,
      "loss": 0.0074,
      "step": 58250
    },
    {
      "epoch": 4.070376317810514,
      "grad_norm": 0.07328232377767563,
      "learning_rate": 1.0349401889749121e-05,
      "loss": 0.0224,
      "step": 58300
    },
    {
      "epoch": 4.073867206590798,
      "grad_norm": 0.268594890832901,
      "learning_rate": 1.031061393573612e-05,
      "loss": 0.0156,
      "step": 58350
    },
    {
      "epoch": 4.077358095371081,
      "grad_norm": 0.05627800524234772,
      "learning_rate": 1.0271825981723118e-05,
      "loss": 0.0172,
      "step": 58400
    },
    {
      "epoch": 4.080848984151365,
      "grad_norm": 0.01628001220524311,
      "learning_rate": 1.0233038027710114e-05,
      "loss": 0.0092,
      "step": 58450
    },
    {
      "epoch": 4.084339872931649,
      "grad_norm": 0.15032631158828735,
      "learning_rate": 1.0194250073697113e-05,
      "loss": 0.0127,
      "step": 58500
    },
    {
      "epoch": 4.087830761711932,
      "grad_norm": 0.03321278095245361,
      "learning_rate": 1.0155462119684111e-05,
      "loss": 0.0071,
      "step": 58550
    },
    {
      "epoch": 4.091321650492215,
      "grad_norm": 1.3011609315872192,
      "learning_rate": 1.011667416567111e-05,
      "loss": 0.0193,
      "step": 58600
    },
    {
      "epoch": 4.094812539272499,
      "grad_norm": 0.009231241419911385,
      "learning_rate": 1.0077886211658108e-05,
      "loss": 0.0142,
      "step": 58650
    },
    {
      "epoch": 4.098303428052782,
      "grad_norm": 0.009372848086059093,
      "learning_rate": 1.0039098257645106e-05,
      "loss": 0.0047,
      "step": 58700
    },
    {
      "epoch": 4.1017943168330655,
      "grad_norm": 0.04598918929696083,
      "learning_rate": 1.0000310303632105e-05,
      "loss": 0.021,
      "step": 58750
    },
    {
      "epoch": 4.105285205613349,
      "grad_norm": 2.4926247596740723,
      "learning_rate": 9.961522349619103e-06,
      "loss": 0.0113,
      "step": 58800
    },
    {
      "epoch": 4.108776094393632,
      "grad_norm": 0.025690525770187378,
      "learning_rate": 9.922734395606101e-06,
      "loss": 0.0069,
      "step": 58850
    },
    {
      "epoch": 4.1122669831739165,
      "grad_norm": 0.009157177992165089,
      "learning_rate": 9.8839464415931e-06,
      "loss": 0.0059,
      "step": 58900
    },
    {
      "epoch": 4.1157578719542,
      "grad_norm": 0.015483526512980461,
      "learning_rate": 9.845158487580097e-06,
      "loss": 0.0243,
      "step": 58950
    },
    {
      "epoch": 4.119248760734483,
      "grad_norm": 0.03802795335650444,
      "learning_rate": 9.806370533567095e-06,
      "loss": 0.016,
      "step": 59000
    },
    {
      "epoch": 4.119248760734483,
      "eval_f1": 0.9442208165612421,
      "eval_loss": 0.170488640666008,
      "eval_runtime": 30.507,
      "eval_samples_per_second": 119.972,
      "eval_steps_per_second": 3.77,
      "step": 59000
    },
    {
      "epoch": 4.122739649514767,
      "grad_norm": 0.01446942612528801,
      "learning_rate": 9.767582579554093e-06,
      "loss": 0.0289,
      "step": 59050
    },
    {
      "epoch": 4.12623053829505,
      "grad_norm": 0.09928736090660095,
      "learning_rate": 9.728794625541092e-06,
      "loss": 0.0372,
      "step": 59100
    },
    {
      "epoch": 4.129721427075333,
      "grad_norm": 7.637205600738525,
      "learning_rate": 9.69000667152809e-06,
      "loss": 0.0091,
      "step": 59150
    },
    {
      "epoch": 4.133212315855617,
      "grad_norm": 0.10061842203140259,
      "learning_rate": 9.651218717515088e-06,
      "loss": 0.0267,
      "step": 59200
    },
    {
      "epoch": 4.1367032046359,
      "grad_norm": 0.00964583270251751,
      "learning_rate": 9.612430763502087e-06,
      "loss": 0.0095,
      "step": 59250
    },
    {
      "epoch": 4.140194093416183,
      "grad_norm": 2.4882123470306396,
      "learning_rate": 9.573642809489087e-06,
      "loss": 0.0095,
      "step": 59300
    },
    {
      "epoch": 4.143684982196468,
      "grad_norm": 0.07580289989709854,
      "learning_rate": 9.534854855476085e-06,
      "loss": 0.0095,
      "step": 59350
    },
    {
      "epoch": 4.147175870976751,
      "grad_norm": 1.8626923561096191,
      "learning_rate": 9.496066901463082e-06,
      "loss": 0.0148,
      "step": 59400
    },
    {
      "epoch": 4.150666759757034,
      "grad_norm": 0.015414468944072723,
      "learning_rate": 9.45727894745008e-06,
      "loss": 0.0292,
      "step": 59450
    },
    {
      "epoch": 4.154157648537318,
      "grad_norm": 0.10861571878194809,
      "learning_rate": 9.419266752517338e-06,
      "loss": 0.0146,
      "step": 59500
    },
    {
      "epoch": 4.157648537317601,
      "grad_norm": 0.01463202852755785,
      "learning_rate": 9.380478798504337e-06,
      "loss": 0.0114,
      "step": 59550
    },
    {
      "epoch": 4.1611394260978845,
      "grad_norm": 0.3748888075351715,
      "learning_rate": 9.341690844491335e-06,
      "loss": 0.0326,
      "step": 59600
    },
    {
      "epoch": 4.164630314878168,
      "grad_norm": 2.7507495880126953,
      "learning_rate": 9.302902890478333e-06,
      "loss": 0.0217,
      "step": 59650
    },
    {
      "epoch": 4.168121203658451,
      "grad_norm": 5.160195350646973,
      "learning_rate": 9.264114936465332e-06,
      "loss": 0.027,
      "step": 59700
    },
    {
      "epoch": 4.1716120924387345,
      "grad_norm": 0.021508311852812767,
      "learning_rate": 9.22532698245233e-06,
      "loss": 0.0141,
      "step": 59750
    },
    {
      "epoch": 4.175102981219018,
      "grad_norm": 3.139760732650757,
      "learning_rate": 9.186539028439329e-06,
      "loss": 0.0162,
      "step": 59800
    },
    {
      "epoch": 4.178593869999302,
      "grad_norm": 0.026574354618787766,
      "learning_rate": 9.147751074426327e-06,
      "loss": 0.0068,
      "step": 59850
    },
    {
      "epoch": 4.1820847587795855,
      "grad_norm": 0.01623447611927986,
      "learning_rate": 9.108963120413325e-06,
      "loss": 0.0115,
      "step": 59900
    },
    {
      "epoch": 4.185575647559869,
      "grad_norm": 0.012585906311869621,
      "learning_rate": 9.070175166400322e-06,
      "loss": 0.0123,
      "step": 59950
    },
    {
      "epoch": 4.189066536340152,
      "grad_norm": 0.12533676624298096,
      "learning_rate": 9.03138721238732e-06,
      "loss": 0.0073,
      "step": 60000
    },
    {
      "epoch": 4.189066536340152,
      "eval_f1": 0.9414816950129721,
      "eval_loss": 0.18086545169353485,
      "eval_runtime": 25.9786,
      "eval_samples_per_second": 140.885,
      "eval_steps_per_second": 4.427,
      "step": 60000
    },
    {
      "epoch": 4.192557425120436,
      "grad_norm": 0.9304799437522888,
      "learning_rate": 8.992599258374319e-06,
      "loss": 0.0074,
      "step": 60050
    },
    {
      "epoch": 4.196048313900719,
      "grad_norm": 0.00669199600815773,
      "learning_rate": 8.953811304361317e-06,
      "loss": 0.0107,
      "step": 60100
    },
    {
      "epoch": 4.199539202681002,
      "grad_norm": 4.687979698181152,
      "learning_rate": 8.915023350348316e-06,
      "loss": 0.0206,
      "step": 60150
    },
    {
      "epoch": 4.203030091461286,
      "grad_norm": 0.18070493638515472,
      "learning_rate": 8.876235396335314e-06,
      "loss": 0.0043,
      "step": 60200
    },
    {
      "epoch": 4.206520980241569,
      "grad_norm": 0.11669691652059555,
      "learning_rate": 8.837447442322312e-06,
      "loss": 0.0136,
      "step": 60250
    },
    {
      "epoch": 4.210011869021853,
      "grad_norm": 0.027609361335635185,
      "learning_rate": 8.79865948830931e-06,
      "loss": 0.0131,
      "step": 60300
    },
    {
      "epoch": 4.213502757802137,
      "grad_norm": 0.019433321431279182,
      "learning_rate": 8.75987153429631e-06,
      "loss": 0.0278,
      "step": 60350
    },
    {
      "epoch": 4.21699364658242,
      "grad_norm": 0.34241268038749695,
      "learning_rate": 8.721083580283308e-06,
      "loss": 0.0071,
      "step": 60400
    },
    {
      "epoch": 4.220484535362703,
      "grad_norm": 0.09645307064056396,
      "learning_rate": 8.682295626270306e-06,
      "loss": 0.0148,
      "step": 60450
    },
    {
      "epoch": 4.223975424142987,
      "grad_norm": 0.4123315215110779,
      "learning_rate": 8.643507672257304e-06,
      "loss": 0.0176,
      "step": 60500
    },
    {
      "epoch": 4.22746631292327,
      "grad_norm": 0.06280361115932465,
      "learning_rate": 8.604719718244303e-06,
      "loss": 0.0122,
      "step": 60550
    },
    {
      "epoch": 4.2309572017035535,
      "grad_norm": 0.013686880469322205,
      "learning_rate": 8.565931764231301e-06,
      "loss": 0.0036,
      "step": 60600
    },
    {
      "epoch": 4.234448090483837,
      "grad_norm": 0.013660120777785778,
      "learning_rate": 8.5271438102183e-06,
      "loss": 0.0157,
      "step": 60650
    },
    {
      "epoch": 4.23793897926412,
      "grad_norm": 0.07857243716716766,
      "learning_rate": 8.488355856205298e-06,
      "loss": 0.0147,
      "step": 60700
    },
    {
      "epoch": 4.2414298680444045,
      "grad_norm": 0.15520472824573517,
      "learning_rate": 8.449567902192296e-06,
      "loss": 0.0187,
      "step": 60750
    },
    {
      "epoch": 4.244920756824688,
      "grad_norm": 0.006114164367318153,
      "learning_rate": 8.410779948179295e-06,
      "loss": 0.0184,
      "step": 60800
    },
    {
      "epoch": 4.248411645604971,
      "grad_norm": 0.10630781948566437,
      "learning_rate": 8.371991994166293e-06,
      "loss": 0.0061,
      "step": 60850
    },
    {
      "epoch": 4.251902534385255,
      "grad_norm": 0.01164723839610815,
      "learning_rate": 8.33320404015329e-06,
      "loss": 0.0184,
      "step": 60900
    },
    {
      "epoch": 4.255393423165538,
      "grad_norm": 0.025141581892967224,
      "learning_rate": 8.294416086140288e-06,
      "loss": 0.0093,
      "step": 60950
    },
    {
      "epoch": 4.258884311945821,
      "grad_norm": 0.009035998955368996,
      "learning_rate": 8.255628132127286e-06,
      "loss": 0.0147,
      "step": 61000
    },
    {
      "epoch": 4.258884311945821,
      "eval_f1": 0.9440599769319492,
      "eval_loss": 0.1847628951072693,
      "eval_runtime": 25.1303,
      "eval_samples_per_second": 145.641,
      "eval_steps_per_second": 4.576,
      "step": 61000
    },
    {
      "epoch": 4.262375200726105,
      "grad_norm": 0.1899830847978592,
      "learning_rate": 8.216840178114285e-06,
      "loss": 0.0194,
      "step": 61050
    },
    {
      "epoch": 4.265866089506388,
      "grad_norm": 0.034703437238931656,
      "learning_rate": 8.178052224101283e-06,
      "loss": 0.0101,
      "step": 61100
    },
    {
      "epoch": 4.269356978286671,
      "grad_norm": 0.1625027358531952,
      "learning_rate": 8.139264270088282e-06,
      "loss": 0.0224,
      "step": 61150
    },
    {
      "epoch": 4.272847867066956,
      "grad_norm": 4.016665935516357,
      "learning_rate": 8.10047631607528e-06,
      "loss": 0.0082,
      "step": 61200
    },
    {
      "epoch": 4.276338755847239,
      "grad_norm": 0.006044914945960045,
      "learning_rate": 8.061688362062278e-06,
      "loss": 0.0081,
      "step": 61250
    },
    {
      "epoch": 4.279829644627522,
      "grad_norm": 6.473517894744873,
      "learning_rate": 8.022900408049277e-06,
      "loss": 0.0205,
      "step": 61300
    },
    {
      "epoch": 4.283320533407806,
      "grad_norm": 0.00561263645067811,
      "learning_rate": 7.984112454036275e-06,
      "loss": 0.029,
      "step": 61350
    },
    {
      "epoch": 4.286811422188089,
      "grad_norm": 0.012270646169781685,
      "learning_rate": 7.945324500023273e-06,
      "loss": 0.0207,
      "step": 61400
    },
    {
      "epoch": 4.2903023109683724,
      "grad_norm": 0.057795651257038116,
      "learning_rate": 7.90653654601027e-06,
      "loss": 0.0226,
      "step": 61450
    },
    {
      "epoch": 4.293793199748656,
      "grad_norm": 0.005627433769404888,
      "learning_rate": 7.867748591997268e-06,
      "loss": 0.0125,
      "step": 61500
    },
    {
      "epoch": 4.297284088528939,
      "grad_norm": 0.021934116259217262,
      "learning_rate": 7.828960637984269e-06,
      "loss": 0.0105,
      "step": 61550
    },
    {
      "epoch": 4.3007749773092225,
      "grad_norm": 6.289938926696777,
      "learning_rate": 7.790172683971267e-06,
      "loss": 0.0145,
      "step": 61600
    },
    {
      "epoch": 4.304265866089507,
      "grad_norm": 0.055504728108644485,
      "learning_rate": 7.751384729958265e-06,
      "loss": 0.0116,
      "step": 61650
    },
    {
      "epoch": 4.30775675486979,
      "grad_norm": 0.23014897108078003,
      "learning_rate": 7.712596775945264e-06,
      "loss": 0.0099,
      "step": 61700
    },
    {
      "epoch": 4.3112476436500735,
      "grad_norm": 0.048537444323301315,
      "learning_rate": 7.673808821932262e-06,
      "loss": 0.0141,
      "step": 61750
    },
    {
      "epoch": 4.314738532430357,
      "grad_norm": 0.0228910893201828,
      "learning_rate": 7.63502086791926e-06,
      "loss": 0.009,
      "step": 61800
    },
    {
      "epoch": 4.31822942121064,
      "grad_norm": 0.04024737700819969,
      "learning_rate": 7.596232913906258e-06,
      "loss": 0.0081,
      "step": 61850
    },
    {
      "epoch": 4.321720309990924,
      "grad_norm": 0.29908424615859985,
      "learning_rate": 7.557444959893256e-06,
      "loss": 0.021,
      "step": 61900
    },
    {
      "epoch": 4.325211198771207,
      "grad_norm": 0.006798740942031145,
      "learning_rate": 7.518657005880255e-06,
      "loss": 0.0186,
      "step": 61950
    },
    {
      "epoch": 4.32870208755149,
      "grad_norm": 0.05872134864330292,
      "learning_rate": 7.479869051867252e-06,
      "loss": 0.0273,
      "step": 62000
    },
    {
      "epoch": 4.32870208755149,
      "eval_f1": 0.9387518142235123,
      "eval_loss": 0.19446897506713867,
      "eval_runtime": 26.5965,
      "eval_samples_per_second": 137.612,
      "eval_steps_per_second": 4.324,
      "step": 62000
    },
    {
      "epoch": 4.332192976331774,
      "grad_norm": 0.012227371335029602,
      "learning_rate": 7.441081097854251e-06,
      "loss": 0.0096,
      "step": 62050
    },
    {
      "epoch": 4.335683865112058,
      "grad_norm": 0.06303926557302475,
      "learning_rate": 7.402293143841249e-06,
      "loss": 0.0146,
      "step": 62100
    },
    {
      "epoch": 4.339174753892341,
      "grad_norm": 0.23254437744617462,
      "learning_rate": 7.363505189828247e-06,
      "loss": 0.0144,
      "step": 62150
    },
    {
      "epoch": 4.342665642672625,
      "grad_norm": 0.0949549674987793,
      "learning_rate": 7.324717235815246e-06,
      "loss": 0.0051,
      "step": 62200
    },
    {
      "epoch": 4.346156531452908,
      "grad_norm": 0.016313236206769943,
      "learning_rate": 7.285929281802243e-06,
      "loss": 0.0346,
      "step": 62250
    },
    {
      "epoch": 4.349647420233191,
      "grad_norm": 0.381818950176239,
      "learning_rate": 7.247141327789242e-06,
      "loss": 0.0144,
      "step": 62300
    },
    {
      "epoch": 4.353138309013475,
      "grad_norm": 0.020029937848448753,
      "learning_rate": 7.20835337377624e-06,
      "loss": 0.0121,
      "step": 62350
    },
    {
      "epoch": 4.356629197793758,
      "grad_norm": 0.10468722879886627,
      "learning_rate": 7.170341178843498e-06,
      "loss": 0.0144,
      "step": 62400
    },
    {
      "epoch": 4.3601200865740415,
      "grad_norm": 0.0396365262567997,
      "learning_rate": 7.1315532248304965e-06,
      "loss": 0.0188,
      "step": 62450
    },
    {
      "epoch": 4.363610975354325,
      "grad_norm": 0.23057244718074799,
      "learning_rate": 7.092765270817495e-06,
      "loss": 0.0159,
      "step": 62500
    },
    {
      "epoch": 4.367101864134609,
      "grad_norm": 0.07276740670204163,
      "learning_rate": 7.053977316804493e-06,
      "loss": 0.0106,
      "step": 62550
    },
    {
      "epoch": 4.3705927529148925,
      "grad_norm": 0.016734370961785316,
      "learning_rate": 7.0151893627914925e-06,
      "loss": 0.0208,
      "step": 62600
    },
    {
      "epoch": 4.374083641695176,
      "grad_norm": 0.009831580333411694,
      "learning_rate": 6.976401408778491e-06,
      "loss": 0.0222,
      "step": 62650
    },
    {
      "epoch": 4.377574530475459,
      "grad_norm": 0.17393887042999268,
      "learning_rate": 6.937613454765489e-06,
      "loss": 0.0104,
      "step": 62700
    },
    {
      "epoch": 4.3810654192557426,
      "grad_norm": 0.10413938015699387,
      "learning_rate": 6.898825500752487e-06,
      "loss": 0.0107,
      "step": 62750
    },
    {
      "epoch": 4.384556308036026,
      "grad_norm": 6.596822261810303,
      "learning_rate": 6.860037546739485e-06,
      "loss": 0.0124,
      "step": 62800
    },
    {
      "epoch": 4.388047196816309,
      "grad_norm": 0.0255569014698267,
      "learning_rate": 6.8212495927264836e-06,
      "loss": 0.0131,
      "step": 62850
    },
    {
      "epoch": 4.391538085596593,
      "grad_norm": 0.005997411906719208,
      "learning_rate": 6.782461638713482e-06,
      "loss": 0.0194,
      "step": 62900
    },
    {
      "epoch": 4.395028974376876,
      "grad_norm": 0.031105129048228264,
      "learning_rate": 6.74367368470048e-06,
      "loss": 0.0153,
      "step": 62950
    },
    {
      "epoch": 4.39851986315716,
      "grad_norm": 0.02738456428050995,
      "learning_rate": 6.704885730687478e-06,
      "loss": 0.0191,
      "step": 63000
    },
    {
      "epoch": 4.39851986315716,
      "eval_f1": 0.9386806160999709,
      "eval_loss": 0.1916884183883667,
      "eval_runtime": 27.6313,
      "eval_samples_per_second": 132.459,
      "eval_steps_per_second": 4.162,
      "step": 63000
    },
    {
      "epoch": 4.402010751937444,
      "grad_norm": 0.1203995794057846,
      "learning_rate": 6.666097776674476e-06,
      "loss": 0.0173,
      "step": 63050
    },
    {
      "epoch": 4.405501640717727,
      "grad_norm": 12.34914493560791,
      "learning_rate": 6.627309822661475e-06,
      "loss": 0.0146,
      "step": 63100
    },
    {
      "epoch": 4.40899252949801,
      "grad_norm": 10.822372436523438,
      "learning_rate": 6.588521868648473e-06,
      "loss": 0.0147,
      "step": 63150
    },
    {
      "epoch": 4.412483418278294,
      "grad_norm": 0.018758252263069153,
      "learning_rate": 6.549733914635471e-06,
      "loss": 0.0136,
      "step": 63200
    },
    {
      "epoch": 4.415974307058577,
      "grad_norm": 0.007578504737466574,
      "learning_rate": 6.510945960622469e-06,
      "loss": 0.0065,
      "step": 63250
    },
    {
      "epoch": 4.41946519583886,
      "grad_norm": 0.05588122084736824,
      "learning_rate": 6.472158006609467e-06,
      "loss": 0.0052,
      "step": 63300
    },
    {
      "epoch": 4.422956084619144,
      "grad_norm": 0.007861234247684479,
      "learning_rate": 6.433370052596466e-06,
      "loss": 0.0196,
      "step": 63350
    },
    {
      "epoch": 4.426446973399427,
      "grad_norm": 1.294003963470459,
      "learning_rate": 6.394582098583464e-06,
      "loss": 0.0164,
      "step": 63400
    },
    {
      "epoch": 4.429937862179711,
      "grad_norm": 0.416669100522995,
      "learning_rate": 6.3557941445704624e-06,
      "loss": 0.0211,
      "step": 63450
    },
    {
      "epoch": 4.433428750959995,
      "grad_norm": 11.46147346496582,
      "learning_rate": 6.317006190557461e-06,
      "loss": 0.0134,
      "step": 63500
    },
    {
      "epoch": 4.436919639740278,
      "grad_norm": 0.03336512669920921,
      "learning_rate": 6.278218236544458e-06,
      "loss": 0.0092,
      "step": 63550
    },
    {
      "epoch": 4.4404105285205615,
      "grad_norm": 0.0666455328464508,
      "learning_rate": 6.2394302825314576e-06,
      "loss": 0.0207,
      "step": 63600
    },
    {
      "epoch": 4.443901417300845,
      "grad_norm": 0.02921397238969803,
      "learning_rate": 6.200642328518456e-06,
      "loss": 0.0189,
      "step": 63650
    },
    {
      "epoch": 4.447392306081128,
      "grad_norm": 0.02691628783941269,
      "learning_rate": 6.161854374505454e-06,
      "loss": 0.0055,
      "step": 63700
    },
    {
      "epoch": 4.450883194861412,
      "grad_norm": 0.008060934953391552,
      "learning_rate": 6.123066420492452e-06,
      "loss": 0.0221,
      "step": 63750
    },
    {
      "epoch": 4.454374083641695,
      "grad_norm": 0.03594188764691353,
      "learning_rate": 6.08427846647945e-06,
      "loss": 0.0089,
      "step": 63800
    },
    {
      "epoch": 4.457864972421978,
      "grad_norm": 0.6915578246116638,
      "learning_rate": 6.045490512466449e-06,
      "loss": 0.0213,
      "step": 63850
    },
    {
      "epoch": 4.461355861202262,
      "grad_norm": 0.01729670725762844,
      "learning_rate": 6.006702558453447e-06,
      "loss": 0.0102,
      "step": 63900
    },
    {
      "epoch": 4.464846749982546,
      "grad_norm": 0.0669393390417099,
      "learning_rate": 5.967914604440445e-06,
      "loss": 0.0253,
      "step": 63950
    },
    {
      "epoch": 4.468337638762829,
      "grad_norm": 0.006938809063285589,
      "learning_rate": 5.929126650427443e-06,
      "loss": 0.0126,
      "step": 64000
    },
    {
      "epoch": 4.468337638762829,
      "eval_f1": 0.9565217391304348,
      "eval_loss": 0.13514484465122223,
      "eval_runtime": 25.6105,
      "eval_samples_per_second": 142.91,
      "eval_steps_per_second": 4.49,
      "step": 64000
    },
    {
      "epoch": 4.471828527543113,
      "grad_norm": 0.051233816891908646,
      "learning_rate": 5.890338696414441e-06,
      "loss": 0.0111,
      "step": 64050
    },
    {
      "epoch": 4.475319416323396,
      "grad_norm": 0.02876574732363224,
      "learning_rate": 5.8515507424014405e-06,
      "loss": 0.0217,
      "step": 64100
    },
    {
      "epoch": 4.478810305103679,
      "grad_norm": 5.748332500457764,
      "learning_rate": 5.812762788388439e-06,
      "loss": 0.0104,
      "step": 64150
    },
    {
      "epoch": 4.482301193883963,
      "grad_norm": 0.004587912000715733,
      "learning_rate": 5.7739748343754364e-06,
      "loss": 0.0156,
      "step": 64200
    },
    {
      "epoch": 4.485792082664246,
      "grad_norm": 0.04000851511955261,
      "learning_rate": 5.735186880362435e-06,
      "loss": 0.0202,
      "step": 64250
    },
    {
      "epoch": 4.4892829714445295,
      "grad_norm": 3.99192476272583,
      "learning_rate": 5.696398926349433e-06,
      "loss": 0.0132,
      "step": 64300
    },
    {
      "epoch": 4.492773860224813,
      "grad_norm": 0.005991567857563496,
      "learning_rate": 5.6576109723364316e-06,
      "loss": 0.0149,
      "step": 64350
    },
    {
      "epoch": 4.496264749005097,
      "grad_norm": 0.05888019874691963,
      "learning_rate": 5.61882301832343e-06,
      "loss": 0.0162,
      "step": 64400
    },
    {
      "epoch": 4.4997556377853805,
      "grad_norm": 0.013234168291091919,
      "learning_rate": 5.580035064310428e-06,
      "loss": 0.0192,
      "step": 64450
    },
    {
      "epoch": 4.503246526565664,
      "grad_norm": 0.00775251816958189,
      "learning_rate": 5.542022869377686e-06,
      "loss": 0.0259,
      "step": 64500
    },
    {
      "epoch": 4.506737415345947,
      "grad_norm": 3.50691556930542,
      "learning_rate": 5.503234915364685e-06,
      "loss": 0.0176,
      "step": 64550
    },
    {
      "epoch": 4.5102283041262305,
      "grad_norm": 0.35396555066108704,
      "learning_rate": 5.464446961351683e-06,
      "loss": 0.0148,
      "step": 64600
    },
    {
      "epoch": 4.513719192906514,
      "grad_norm": 0.01665789633989334,
      "learning_rate": 5.4256590073386815e-06,
      "loss": 0.0074,
      "step": 64650
    },
    {
      "epoch": 4.517210081686797,
      "grad_norm": 0.05178999900817871,
      "learning_rate": 5.38687105332568e-06,
      "loss": 0.0095,
      "step": 64700
    },
    {
      "epoch": 4.520700970467081,
      "grad_norm": 0.1286119818687439,
      "learning_rate": 5.3480830993126774e-06,
      "loss": 0.0054,
      "step": 64750
    },
    {
      "epoch": 4.524191859247364,
      "grad_norm": 17.01955795288086,
      "learning_rate": 5.309295145299676e-06,
      "loss": 0.0178,
      "step": 64800
    },
    {
      "epoch": 4.527682748027647,
      "grad_norm": 0.08323328197002411,
      "learning_rate": 5.270507191286674e-06,
      "loss": 0.0115,
      "step": 64850
    },
    {
      "epoch": 4.531173636807932,
      "grad_norm": 4.3498640060424805,
      "learning_rate": 5.2317192372736726e-06,
      "loss": 0.0137,
      "step": 64900
    },
    {
      "epoch": 4.534664525588215,
      "grad_norm": 4.796928405761719,
      "learning_rate": 5.192931283260671e-06,
      "loss": 0.012,
      "step": 64950
    },
    {
      "epoch": 4.538155414368498,
      "grad_norm": 0.011186789721250534,
      "learning_rate": 5.1541433292476685e-06,
      "loss": 0.0096,
      "step": 65000
    },
    {
      "epoch": 4.538155414368498,
      "eval_f1": 0.9540295119182747,
      "eval_loss": 0.14705610275268555,
      "eval_runtime": 22.5602,
      "eval_samples_per_second": 162.232,
      "eval_steps_per_second": 5.097,
      "step": 65000
    },
    {
      "epoch": 4.541646303148782,
      "grad_norm": 0.11445566266775131,
      "learning_rate": 5.115355375234667e-06,
      "loss": 0.0255,
      "step": 65050
    },
    {
      "epoch": 4.545137191929065,
      "grad_norm": 0.2920547425746918,
      "learning_rate": 5.076567421221665e-06,
      "loss": 0.0126,
      "step": 65100
    },
    {
      "epoch": 4.548628080709348,
      "grad_norm": 0.010884485207498074,
      "learning_rate": 5.0377794672086645e-06,
      "loss": 0.0084,
      "step": 65150
    },
    {
      "epoch": 4.552118969489632,
      "grad_norm": 0.004941530991345644,
      "learning_rate": 4.998991513195663e-06,
      "loss": 0.0138,
      "step": 65200
    },
    {
      "epoch": 4.555609858269915,
      "grad_norm": 0.03210439160466194,
      "learning_rate": 4.96020355918266e-06,
      "loss": 0.0137,
      "step": 65250
    },
    {
      "epoch": 4.5591007470501985,
      "grad_norm": 0.16372257471084595,
      "learning_rate": 4.921415605169659e-06,
      "loss": 0.009,
      "step": 65300
    },
    {
      "epoch": 4.562591635830483,
      "grad_norm": 0.01041162759065628,
      "learning_rate": 4.882627651156657e-06,
      "loss": 0.0111,
      "step": 65350
    },
    {
      "epoch": 4.566082524610766,
      "grad_norm": 0.010886666364967823,
      "learning_rate": 4.8438396971436555e-06,
      "loss": 0.0065,
      "step": 65400
    },
    {
      "epoch": 4.5695734133910495,
      "grad_norm": 5.038321495056152,
      "learning_rate": 4.805051743130654e-06,
      "loss": 0.0123,
      "step": 65450
    },
    {
      "epoch": 4.573064302171333,
      "grad_norm": 0.14299964904785156,
      "learning_rate": 4.7662637891176514e-06,
      "loss": 0.0104,
      "step": 65500
    },
    {
      "epoch": 4.576555190951616,
      "grad_norm": 0.07817520946264267,
      "learning_rate": 4.72747583510465e-06,
      "loss": 0.0153,
      "step": 65550
    },
    {
      "epoch": 4.5800460797319,
      "grad_norm": 0.02830136939883232,
      "learning_rate": 4.688687881091648e-06,
      "loss": 0.0212,
      "step": 65600
    },
    {
      "epoch": 4.583536968512183,
      "grad_norm": 0.006670476868748665,
      "learning_rate": 4.649899927078647e-06,
      "loss": 0.0134,
      "step": 65650
    },
    {
      "epoch": 4.587027857292466,
      "grad_norm": 12.660615921020508,
      "learning_rate": 4.611111973065645e-06,
      "loss": 0.0125,
      "step": 65700
    },
    {
      "epoch": 4.59051874607275,
      "grad_norm": 0.030848925933241844,
      "learning_rate": 4.572324019052643e-06,
      "loss": 0.0122,
      "step": 65750
    },
    {
      "epoch": 4.594009634853034,
      "grad_norm": 0.028387153521180153,
      "learning_rate": 4.533536065039642e-06,
      "loss": 0.0137,
      "step": 65800
    },
    {
      "epoch": 4.597500523633317,
      "grad_norm": 0.05481830611824989,
      "learning_rate": 4.49474811102664e-06,
      "loss": 0.0243,
      "step": 65850
    },
    {
      "epoch": 4.600991412413601,
      "grad_norm": 0.021391669288277626,
      "learning_rate": 4.4559601570136385e-06,
      "loss": 0.0206,
      "step": 65900
    },
    {
      "epoch": 4.604482301193884,
      "grad_norm": 0.06175484508275986,
      "learning_rate": 4.417172203000636e-06,
      "loss": 0.0182,
      "step": 65950
    },
    {
      "epoch": 4.607973189974167,
      "grad_norm": 0.10749427229166031,
      "learning_rate": 4.378384248987634e-06,
      "loss": 0.0187,
      "step": 66000
    },
    {
      "epoch": 4.607973189974167,
      "eval_f1": 0.9402205455600696,
      "eval_loss": 0.2095240354537964,
      "eval_runtime": 23.1199,
      "eval_samples_per_second": 158.305,
      "eval_steps_per_second": 4.974,
      "step": 66000
    },
    {
      "epoch": 4.611464078754451,
      "grad_norm": 0.009061275981366634,
      "learning_rate": 4.339596294974633e-06,
      "loss": 0.0278,
      "step": 66050
    },
    {
      "epoch": 4.614954967534734,
      "grad_norm": 0.003557862713932991,
      "learning_rate": 4.300808340961631e-06,
      "loss": 0.0101,
      "step": 66100
    },
    {
      "epoch": 4.6184458563150175,
      "grad_norm": 0.05570525676012039,
      "learning_rate": 4.2620203869486295e-06,
      "loss": 0.0148,
      "step": 66150
    },
    {
      "epoch": 4.621936745095301,
      "grad_norm": 26.183717727661133,
      "learning_rate": 4.223232432935628e-06,
      "loss": 0.0293,
      "step": 66200
    },
    {
      "epoch": 4.625427633875585,
      "grad_norm": 0.01294758915901184,
      "learning_rate": 4.1844444789226254e-06,
      "loss": 0.014,
      "step": 66250
    },
    {
      "epoch": 4.6289185226558685,
      "grad_norm": 0.005879008211195469,
      "learning_rate": 4.146432283989884e-06,
      "loss": 0.0293,
      "step": 66300
    },
    {
      "epoch": 4.632409411436152,
      "grad_norm": 0.06964797526597977,
      "learning_rate": 4.107644329976883e-06,
      "loss": 0.0088,
      "step": 66350
    },
    {
      "epoch": 4.635900300216435,
      "grad_norm": 0.07942630350589752,
      "learning_rate": 4.068856375963881e-06,
      "loss": 0.0174,
      "step": 66400
    },
    {
      "epoch": 4.6393911889967185,
      "grad_norm": 0.12276989966630936,
      "learning_rate": 4.0300684219508795e-06,
      "loss": 0.0176,
      "step": 66450
    },
    {
      "epoch": 4.642882077777002,
      "grad_norm": 0.07290726155042648,
      "learning_rate": 3.991280467937877e-06,
      "loss": 0.0048,
      "step": 66500
    },
    {
      "epoch": 4.646372966557285,
      "grad_norm": 5.691318511962891,
      "learning_rate": 3.952492513924875e-06,
      "loss": 0.0185,
      "step": 66550
    },
    {
      "epoch": 4.649863855337569,
      "grad_norm": 0.01649297960102558,
      "learning_rate": 3.913704559911874e-06,
      "loss": 0.01,
      "step": 66600
    },
    {
      "epoch": 4.653354744117852,
      "grad_norm": 0.08620315790176392,
      "learning_rate": 3.874916605898872e-06,
      "loss": 0.0143,
      "step": 66650
    },
    {
      "epoch": 4.656845632898136,
      "grad_norm": 0.26478084921836853,
      "learning_rate": 3.8361286518858705e-06,
      "loss": 0.014,
      "step": 66700
    },
    {
      "epoch": 4.66033652167842,
      "grad_norm": 0.10022763907909393,
      "learning_rate": 3.7973406978728685e-06,
      "loss": 0.0032,
      "step": 66750
    },
    {
      "epoch": 4.663827410458703,
      "grad_norm": 6.073363780975342,
      "learning_rate": 3.7585527438598673e-06,
      "loss": 0.0149,
      "step": 66800
    },
    {
      "epoch": 4.667318299238986,
      "grad_norm": 0.014798264019191265,
      "learning_rate": 3.7197647898468657e-06,
      "loss": 0.0086,
      "step": 66850
    },
    {
      "epoch": 4.67080918801927,
      "grad_norm": 0.0035229383502155542,
      "learning_rate": 3.6809768358338636e-06,
      "loss": 0.0184,
      "step": 66900
    },
    {
      "epoch": 4.674300076799553,
      "grad_norm": 0.008204997517168522,
      "learning_rate": 3.642188881820862e-06,
      "loss": 0.0132,
      "step": 66950
    },
    {
      "epoch": 4.677790965579836,
      "grad_norm": 0.26443013548851013,
      "learning_rate": 3.6034009278078604e-06,
      "loss": 0.0175,
      "step": 67000
    },
    {
      "epoch": 4.677790965579836,
      "eval_f1": 0.9542743538767395,
      "eval_loss": 0.1465238481760025,
      "eval_runtime": 19.3418,
      "eval_samples_per_second": 189.227,
      "eval_steps_per_second": 5.946,
      "step": 67000
    },
    {
      "epoch": 4.68128185436012,
      "grad_norm": 0.03244328871369362,
      "learning_rate": 3.5646129737948583e-06,
      "loss": 0.0144,
      "step": 67050
    },
    {
      "epoch": 4.684772743140403,
      "grad_norm": 0.006314130965620279,
      "learning_rate": 3.5258250197818567e-06,
      "loss": 0.0162,
      "step": 67100
    },
    {
      "epoch": 4.688263631920687,
      "grad_norm": 3.670078992843628,
      "learning_rate": 3.4870370657688547e-06,
      "loss": 0.0102,
      "step": 67150
    },
    {
      "epoch": 4.691754520700971,
      "grad_norm": 0.016808710992336273,
      "learning_rate": 3.448249111755853e-06,
      "loss": 0.0114,
      "step": 67200
    },
    {
      "epoch": 4.695245409481254,
      "grad_norm": 7.48440408706665,
      "learning_rate": 3.4094611577428514e-06,
      "loss": 0.0159,
      "step": 67250
    },
    {
      "epoch": 4.6987362982615375,
      "grad_norm": 0.22316499054431915,
      "learning_rate": 3.3706732037298494e-06,
      "loss": 0.0077,
      "step": 67300
    },
    {
      "epoch": 4.702227187041821,
      "grad_norm": 6.541471004486084,
      "learning_rate": 3.3318852497168478e-06,
      "loss": 0.015,
      "step": 67350
    },
    {
      "epoch": 4.705718075822104,
      "grad_norm": 0.038210611790418625,
      "learning_rate": 3.2930972957038466e-06,
      "loss": 0.0139,
      "step": 67400
    },
    {
      "epoch": 4.709208964602388,
      "grad_norm": 0.008333321660757065,
      "learning_rate": 3.254309341690845e-06,
      "loss": 0.0192,
      "step": 67450
    },
    {
      "epoch": 4.712699853382671,
      "grad_norm": 0.031475260853767395,
      "learning_rate": 3.215521387677843e-06,
      "loss": 0.0059,
      "step": 67500
    },
    {
      "epoch": 4.716190742162954,
      "grad_norm": 0.0056376708671450615,
      "learning_rate": 3.1767334336648413e-06,
      "loss": 0.0081,
      "step": 67550
    },
    {
      "epoch": 4.719681630943239,
      "grad_norm": 0.013743482530117035,
      "learning_rate": 3.1379454796518397e-06,
      "loss": 0.0144,
      "step": 67600
    },
    {
      "epoch": 4.723172519723522,
      "grad_norm": 0.009420888498425484,
      "learning_rate": 3.0991575256388376e-06,
      "loss": 0.0115,
      "step": 67650
    },
    {
      "epoch": 4.726663408503805,
      "grad_norm": 0.007564714178442955,
      "learning_rate": 3.060369571625836e-06,
      "loss": 0.0119,
      "step": 67700
    },
    {
      "epoch": 4.730154297284089,
      "grad_norm": 0.13455595076084137,
      "learning_rate": 3.0215816176128344e-06,
      "loss": 0.0183,
      "step": 67750
    },
    {
      "epoch": 4.733645186064372,
      "grad_norm": 0.011826273053884506,
      "learning_rate": 2.9827936635998324e-06,
      "loss": 0.0192,
      "step": 67800
    },
    {
      "epoch": 4.737136074844655,
      "grad_norm": 0.02266635373234749,
      "learning_rate": 2.944005709586831e-06,
      "loss": 0.0083,
      "step": 67850
    },
    {
      "epoch": 4.740626963624939,
      "grad_norm": 2.2667205333709717,
      "learning_rate": 2.905217755573829e-06,
      "loss": 0.0208,
      "step": 67900
    },
    {
      "epoch": 4.744117852405222,
      "grad_norm": 0.08782637864351273,
      "learning_rate": 2.8664298015608275e-06,
      "loss": 0.0114,
      "step": 67950
    },
    {
      "epoch": 4.7476087411855055,
      "grad_norm": 0.013912571594119072,
      "learning_rate": 2.8276418475478255e-06,
      "loss": 0.0209,
      "step": 68000
    },
    {
      "epoch": 4.7476087411855055,
      "eval_f1": 0.9542223485925505,
      "eval_loss": 0.14947552978992462,
      "eval_runtime": 20.6326,
      "eval_samples_per_second": 177.389,
      "eval_steps_per_second": 5.574,
      "step": 68000
    },
    {
      "epoch": 4.75109962996579,
      "grad_norm": 0.07238892465829849,
      "learning_rate": 2.788853893534824e-06,
      "loss": 0.0115,
      "step": 68050
    },
    {
      "epoch": 4.754590518746073,
      "grad_norm": 0.007459067273885012,
      "learning_rate": 2.750065939521822e-06,
      "loss": 0.0148,
      "step": 68100
    },
    {
      "epoch": 4.758081407526356,
      "grad_norm": 9.393803596496582,
      "learning_rate": 2.7112779855088206e-06,
      "loss": 0.0145,
      "step": 68150
    },
    {
      "epoch": 4.76157229630664,
      "grad_norm": 0.1944788694381714,
      "learning_rate": 2.672490031495819e-06,
      "loss": 0.0166,
      "step": 68200
    },
    {
      "epoch": 4.765063185086923,
      "grad_norm": 0.086297906935215,
      "learning_rate": 2.633702077482817e-06,
      "loss": 0.0128,
      "step": 68250
    },
    {
      "epoch": 4.7685540738672065,
      "grad_norm": 0.007248733192682266,
      "learning_rate": 2.5949141234698153e-06,
      "loss": 0.0138,
      "step": 68300
    },
    {
      "epoch": 4.77204496264749,
      "grad_norm": 0.06873763352632523,
      "learning_rate": 2.5561261694568137e-06,
      "loss": 0.0196,
      "step": 68350
    },
    {
      "epoch": 4.775535851427773,
      "grad_norm": 0.011037365533411503,
      "learning_rate": 2.5173382154438116e-06,
      "loss": 0.0111,
      "step": 68400
    },
    {
      "epoch": 4.779026740208057,
      "grad_norm": 0.3673991858959198,
      "learning_rate": 2.4785502614308104e-06,
      "loss": 0.0147,
      "step": 68450
    },
    {
      "epoch": 4.782517628988341,
      "grad_norm": 0.014958728104829788,
      "learning_rate": 2.4397623074178084e-06,
      "loss": 0.0139,
      "step": 68500
    },
    {
      "epoch": 4.786008517768624,
      "grad_norm": 0.07658286392688751,
      "learning_rate": 2.4009743534048068e-06,
      "loss": 0.0091,
      "step": 68550
    },
    {
      "epoch": 4.789499406548908,
      "grad_norm": 0.09001337736845016,
      "learning_rate": 2.362186399391805e-06,
      "loss": 0.0095,
      "step": 68600
    },
    {
      "epoch": 4.792990295329191,
      "grad_norm": 0.0031005863565951586,
      "learning_rate": 2.323398445378803e-06,
      "loss": 0.0174,
      "step": 68650
    },
    {
      "epoch": 4.796481184109474,
      "grad_norm": 0.0415862575173378,
      "learning_rate": 2.2846104913658015e-06,
      "loss": 0.0038,
      "step": 68700
    },
    {
      "epoch": 4.799972072889758,
      "grad_norm": 4.582136631011963,
      "learning_rate": 2.2458225373528e-06,
      "loss": 0.0197,
      "step": 68750
    },
    {
      "epoch": 4.803462961670041,
      "grad_norm": 0.02210802398622036,
      "learning_rate": 2.2070345833397983e-06,
      "loss": 0.012,
      "step": 68800
    },
    {
      "epoch": 4.806953850450324,
      "grad_norm": 0.01928221434354782,
      "learning_rate": 2.1682466293267962e-06,
      "loss": 0.0144,
      "step": 68850
    },
    {
      "epoch": 4.810444739230608,
      "grad_norm": 0.016591308638453484,
      "learning_rate": 2.1294586753137946e-06,
      "loss": 0.0041,
      "step": 68900
    },
    {
      "epoch": 4.813935628010892,
      "grad_norm": 5.927182197570801,
      "learning_rate": 2.090670721300793e-06,
      "loss": 0.015,
      "step": 68950
    },
    {
      "epoch": 4.817426516791175,
      "grad_norm": 0.07412146776914597,
      "learning_rate": 2.051882767287791e-06,
      "loss": 0.0202,
      "step": 69000
    },
    {
      "epoch": 4.817426516791175,
      "eval_f1": 0.958063608218407,
      "eval_loss": 0.13213428854942322,
      "eval_runtime": 23.7215,
      "eval_samples_per_second": 154.29,
      "eval_steps_per_second": 4.848,
      "step": 69000
    },
    {
      "epoch": 4.820917405571459,
      "grad_norm": 0.011551358737051487,
      "learning_rate": 2.0130948132747897e-06,
      "loss": 0.0106,
      "step": 69050
    },
    {
      "epoch": 4.824408294351742,
      "grad_norm": 0.041600409895181656,
      "learning_rate": 1.9743068592617877e-06,
      "loss": 0.028,
      "step": 69100
    },
    {
      "epoch": 4.8278991831320255,
      "grad_norm": 0.011347083374857903,
      "learning_rate": 1.935518905248786e-06,
      "loss": 0.0197,
      "step": 69150
    },
    {
      "epoch": 4.831390071912309,
      "grad_norm": 0.1340709775686264,
      "learning_rate": 1.8967309512357842e-06,
      "loss": 0.0147,
      "step": 69200
    },
    {
      "epoch": 4.834880960692592,
      "grad_norm": 0.013690965250134468,
      "learning_rate": 1.8579429972227824e-06,
      "loss": 0.0118,
      "step": 69250
    },
    {
      "epoch": 4.838371849472876,
      "grad_norm": 7.221401214599609,
      "learning_rate": 1.8191550432097808e-06,
      "loss": 0.0314,
      "step": 69300
    },
    {
      "epoch": 4.841862738253159,
      "grad_norm": 0.0063622863963246346,
      "learning_rate": 1.7803670891967792e-06,
      "loss": 0.011,
      "step": 69350
    },
    {
      "epoch": 4.845353627033443,
      "grad_norm": 0.09204477816820145,
      "learning_rate": 1.7415791351837776e-06,
      "loss": 0.0089,
      "step": 69400
    },
    {
      "epoch": 4.8488445158137266,
      "grad_norm": 0.2040203958749771,
      "learning_rate": 1.7027911811707757e-06,
      "loss": 0.0019,
      "step": 69450
    },
    {
      "epoch": 4.85233540459401,
      "grad_norm": 0.009934721514582634,
      "learning_rate": 1.6640032271577739e-06,
      "loss": 0.0226,
      "step": 69500
    },
    {
      "epoch": 4.855826293374293,
      "grad_norm": 0.009200035594403744,
      "learning_rate": 1.625215273144772e-06,
      "loss": 0.0202,
      "step": 69550
    },
    {
      "epoch": 4.859317182154577,
      "grad_norm": 2.9772799015045166,
      "learning_rate": 1.5864273191317704e-06,
      "loss": 0.0267,
      "step": 69600
    },
    {
      "epoch": 4.86280807093486,
      "grad_norm": 0.046260032802820206,
      "learning_rate": 1.5476393651187688e-06,
      "loss": 0.0111,
      "step": 69650
    },
    {
      "epoch": 4.866298959715143,
      "grad_norm": 0.5659658312797546,
      "learning_rate": 1.509627170186027e-06,
      "loss": 0.024,
      "step": 69700
    },
    {
      "epoch": 4.869789848495427,
      "grad_norm": 0.029453273862600327,
      "learning_rate": 1.4708392161730255e-06,
      "loss": 0.0206,
      "step": 69750
    },
    {
      "epoch": 4.87328073727571,
      "grad_norm": 0.06688568741083145,
      "learning_rate": 1.4320512621600236e-06,
      "loss": 0.0081,
      "step": 69800
    },
    {
      "epoch": 4.876771626055994,
      "grad_norm": 0.09227997809648514,
      "learning_rate": 1.393263308147022e-06,
      "loss": 0.0158,
      "step": 69850
    },
    {
      "epoch": 4.880262514836277,
      "grad_norm": 0.01653113402426243,
      "learning_rate": 1.3544753541340202e-06,
      "loss": 0.012,
      "step": 69900
    },
    {
      "epoch": 4.883753403616561,
      "grad_norm": 0.1240030974149704,
      "learning_rate": 1.3156874001210183e-06,
      "loss": 0.0042,
      "step": 69950
    },
    {
      "epoch": 4.887244292396844,
      "grad_norm": 0.05169876664876938,
      "learning_rate": 1.2768994461080167e-06,
      "loss": 0.0181,
      "step": 70000
    },
    {
      "epoch": 4.887244292396844,
      "eval_f1": 0.9547652916073969,
      "eval_loss": 0.14620168507099152,
      "eval_runtime": 18.4218,
      "eval_samples_per_second": 198.678,
      "eval_steps_per_second": 6.243,
      "step": 70000
    },
    {
      "epoch": 4.890735181177128,
      "grad_norm": 0.005993202794343233,
      "learning_rate": 1.238111492095015e-06,
      "loss": 0.0088,
      "step": 70050
    },
    {
      "epoch": 4.894226069957411,
      "grad_norm": 0.014470288529992104,
      "learning_rate": 1.1993235380820133e-06,
      "loss": 0.0226,
      "step": 70100
    },
    {
      "epoch": 4.8977169587376945,
      "grad_norm": 4.7974162101745605,
      "learning_rate": 1.1605355840690116e-06,
      "loss": 0.0145,
      "step": 70150
    },
    {
      "epoch": 4.901207847517978,
      "grad_norm": 0.010994662530720234,
      "learning_rate": 1.1217476300560098e-06,
      "loss": 0.0174,
      "step": 70200
    },
    {
      "epoch": 4.904698736298261,
      "grad_norm": 0.006302454508841038,
      "learning_rate": 1.082959676043008e-06,
      "loss": 0.0128,
      "step": 70250
    },
    {
      "epoch": 4.908189625078545,
      "grad_norm": 0.1383901685476303,
      "learning_rate": 1.0441717220300066e-06,
      "loss": 0.016,
      "step": 70300
    },
    {
      "epoch": 4.911680513858828,
      "grad_norm": 0.06532856822013855,
      "learning_rate": 1.0053837680170047e-06,
      "loss": 0.0118,
      "step": 70350
    },
    {
      "epoch": 4.915171402639112,
      "grad_norm": 0.010423216037452221,
      "learning_rate": 9.66595814004003e-07,
      "loss": 0.0085,
      "step": 70400
    },
    {
      "epoch": 4.918662291419396,
      "grad_norm": 0.030837245285511017,
      "learning_rate": 9.278078599910013e-07,
      "loss": 0.015,
      "step": 70450
    },
    {
      "epoch": 4.922153180199679,
      "grad_norm": 0.025385385379195213,
      "learning_rate": 8.890199059779996e-07,
      "loss": 0.005,
      "step": 70500
    },
    {
      "epoch": 4.925644068979962,
      "grad_norm": 4.04787540435791,
      "learning_rate": 8.502319519649977e-07,
      "loss": 0.0135,
      "step": 70550
    },
    {
      "epoch": 4.929134957760246,
      "grad_norm": 0.018447715789079666,
      "learning_rate": 8.114439979519961e-07,
      "loss": 0.0159,
      "step": 70600
    },
    {
      "epoch": 4.932625846540529,
      "grad_norm": 0.5685107707977295,
      "learning_rate": 7.726560439389944e-07,
      "loss": 0.0022,
      "step": 70650
    },
    {
      "epoch": 4.936116735320812,
      "grad_norm": 0.05199820548295975,
      "learning_rate": 7.338680899259926e-07,
      "loss": 0.0176,
      "step": 70700
    },
    {
      "epoch": 4.939607624101096,
      "grad_norm": 0.18712732195854187,
      "learning_rate": 6.950801359129908e-07,
      "loss": 0.0236,
      "step": 70750
    },
    {
      "epoch": 4.943098512881379,
      "grad_norm": 0.009691745974123478,
      "learning_rate": 6.562921818999892e-07,
      "loss": 0.0238,
      "step": 70800
    },
    {
      "epoch": 4.946589401661663,
      "grad_norm": 0.0077186268754303455,
      "learning_rate": 6.175042278869874e-07,
      "loss": 0.0196,
      "step": 70850
    },
    {
      "epoch": 4.950080290441947,
      "grad_norm": 0.015092890709638596,
      "learning_rate": 5.787162738739857e-07,
      "loss": 0.0166,
      "step": 70900
    },
    {
      "epoch": 4.95357117922223,
      "grad_norm": 0.01764654368162155,
      "learning_rate": 5.39928319860984e-07,
      "loss": 0.0084,
      "step": 70950
    },
    {
      "epoch": 4.9570620680025135,
      "grad_norm": 0.030939076095819473,
      "learning_rate": 5.011403658479822e-07,
      "loss": 0.0067,
      "step": 71000
    },
    {
      "epoch": 4.9570620680025135,
      "eval_f1": 0.9526525955504849,
      "eval_loss": 0.150511234998703,
      "eval_runtime": 19.3329,
      "eval_samples_per_second": 189.315,
      "eval_steps_per_second": 5.948,
      "step": 71000
    },
    {
      "epoch": 4.960552956782797,
      "grad_norm": 0.016753925010561943,
      "learning_rate": 4.6235241183498053e-07,
      "loss": 0.0057,
      "step": 71050
    },
    {
      "epoch": 4.96404384556308,
      "grad_norm": 0.01155967079102993,
      "learning_rate": 4.2356445782197886e-07,
      "loss": 0.0089,
      "step": 71100
    },
    {
      "epoch": 4.967534734343364,
      "grad_norm": 0.04415968805551529,
      "learning_rate": 3.847765038089771e-07,
      "loss": 0.01,
      "step": 71150
    },
    {
      "epoch": 4.971025623123647,
      "grad_norm": 4.265117645263672,
      "learning_rate": 3.4598854979597536e-07,
      "loss": 0.017,
      "step": 71200
    },
    {
      "epoch": 4.97451651190393,
      "grad_norm": 0.02736731618642807,
      "learning_rate": 3.072005957829737e-07,
      "loss": 0.007,
      "step": 71250
    },
    {
      "epoch": 4.9780074006842145,
      "grad_norm": 0.029399313032627106,
      "learning_rate": 2.6841264176997196e-07,
      "loss": 0.0195,
      "step": 71300
    },
    {
      "epoch": 4.981498289464498,
      "grad_norm": 0.006107365246862173,
      "learning_rate": 2.296246877569702e-07,
      "loss": 0.0016,
      "step": 71350
    },
    {
      "epoch": 4.984989178244781,
      "grad_norm": 3.961512565612793,
      "learning_rate": 1.9083673374396848e-07,
      "loss": 0.0087,
      "step": 71400
    },
    {
      "epoch": 4.988480067025065,
      "grad_norm": 0.02385067753493786,
      "learning_rate": 1.5204877973096676e-07,
      "loss": 0.0261,
      "step": 71450
    },
    {
      "epoch": 4.991970955805348,
      "grad_norm": 0.15537860989570618,
      "learning_rate": 1.1326082571796503e-07,
      "loss": 0.0056,
      "step": 71500
    },
    {
      "epoch": 4.995461844585631,
      "grad_norm": 9.23886489868164,
      "learning_rate": 7.44728717049633e-08,
      "loss": 0.018,
      "step": 71550
    },
    {
      "epoch": 4.998952733365915,
      "grad_norm": 3.6488893032073975,
      "learning_rate": 3.5684917691961585e-08,
      "loss": 0.0126,
      "step": 71600
    },
    {
      "epoch": 5.0,
      "step": 71615,
      "total_flos": 1.2034503255440461e+20,
      "train_loss": 0.03158619465179086,
      "train_runtime": 20070.0469,
      "train_samples_per_second": 114.177,
      "train_steps_per_second": 3.568
    }
  ],
  "logging_steps": 50,
  "max_steps": 71615,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 5,
  "save_steps": 1000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.2034503255440461e+20,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": null
}