{
  "best_metric": 0.9760225669957687,
  "best_model_checkpoint": "melanoma-v3\\checkpoint-12523",
  "epoch": 7.0,
  "eval_steps": 500,
  "global_step": 12523,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.027948574622694244,
      "grad_norm": 9.020837783813477,
      "learning_rate": 1.8355945730247407e-05,
      "loss": 0.6846,
      "step": 50
    },
    {
      "epoch": 0.05589714924538849,
      "grad_norm": 15.282539367675781,
      "learning_rate": 3.790901835594573e-05,
      "loss": 0.6027,
      "step": 100
    },
    {
      "epoch": 0.08384572386808273,
      "grad_norm": 17.592166900634766,
      "learning_rate": 5.78611332801277e-05,
      "loss": 0.4504,
      "step": 150
    },
    {
      "epoch": 0.11179429849077697,
      "grad_norm": 11.997602462768555,
      "learning_rate": 7.781324820430965e-05,
      "loss": 0.3523,
      "step": 200
    },
    {
      "epoch": 0.13974287311347122,
      "grad_norm": 16.85348129272461,
      "learning_rate": 9.776536312849161e-05,
      "loss": 0.2612,
      "step": 250
    },
    {
      "epoch": 0.16769144773616546,
      "grad_norm": 5.394933700561523,
      "learning_rate": 0.00011771747805267359,
      "loss": 0.2493,
      "step": 300
    },
    {
      "epoch": 0.1956400223588597,
      "grad_norm": 15.290493965148926,
      "learning_rate": 0.00013766959297685554,
      "loss": 0.2092,
      "step": 350
    },
    {
      "epoch": 0.22358859698155395,
      "grad_norm": 15.310522079467773,
      "learning_rate": 0.00015762170790103753,
      "loss": 0.1995,
      "step": 400
    },
    {
      "epoch": 0.2515371716042482,
      "grad_norm": 20.243497848510742,
      "learning_rate": 0.00017757382282521946,
      "loss": 0.2205,
      "step": 450
    },
    {
      "epoch": 0.27948574622694244,
      "grad_norm": 5.062504291534424,
      "learning_rate": 0.00019752593774940145,
      "loss": 0.2124,
      "step": 500
    },
    {
      "epoch": 0.3074343208496367,
      "grad_norm": 5.295910358428955,
      "learning_rate": 0.00021747805267358338,
      "loss": 0.2178,
      "step": 550
    },
    {
      "epoch": 0.3353828954723309,
      "grad_norm": 1.7261399030685425,
      "learning_rate": 0.00023743016759776537,
      "loss": 0.1946,
      "step": 600
    },
    {
      "epoch": 0.36333147009502514,
      "grad_norm": 5.8339338302612305,
      "learning_rate": 0.00025738228252194736,
      "loss": 0.2097,
      "step": 650
    },
    {
      "epoch": 0.3912800447177194,
      "grad_norm": 2.142676830291748,
      "learning_rate": 0.0002773343974461293,
      "loss": 0.2165,
      "step": 700
    },
    {
      "epoch": 0.41922861934041367,
      "grad_norm": 2.970517873764038,
      "learning_rate": 0.00029728651237031123,
      "loss": 0.2051,
      "step": 750
    },
    {
      "epoch": 0.4471771939631079,
      "grad_norm": 4.551942348480225,
      "learning_rate": 0.0003172386272944932,
      "loss": 0.2288,
      "step": 800
    },
    {
      "epoch": 0.47512576858580213,
      "grad_norm": 5.390867710113525,
      "learning_rate": 0.0003371907422186752,
      "loss": 0.2091,
      "step": 850
    },
    {
      "epoch": 0.5030743432084964,
      "grad_norm": 1.3094195127487183,
      "learning_rate": 0.00035714285714285714,
      "loss": 0.2302,
      "step": 900
    },
    {
      "epoch": 0.5310229178311906,
      "grad_norm": 2.8628153800964355,
      "learning_rate": 0.00037709497206703913,
      "loss": 0.2244,
      "step": 950
    },
    {
      "epoch": 0.5589714924538849,
      "grad_norm": 2.0258989334106445,
      "learning_rate": 0.00039704708699122107,
      "loss": 0.1956,
      "step": 1000
    },
    {
      "epoch": 0.5869200670765791,
      "grad_norm": 2.2154078483581543,
      "learning_rate": 0.000416999201915403,
      "loss": 0.2345,
      "step": 1050
    },
    {
      "epoch": 0.6148686416992734,
      "grad_norm": 1.6475298404693604,
      "learning_rate": 0.000436951316839585,
      "loss": 0.2137,
      "step": 1100
    },
    {
      "epoch": 0.6428172163219675,
      "grad_norm": 2.1503453254699707,
      "learning_rate": 0.000456903431763767,
      "loss": 0.2324,
      "step": 1150
    },
    {
      "epoch": 0.6707657909446618,
      "grad_norm": 1.5340570211410522,
      "learning_rate": 0.00047685554668794897,
      "loss": 0.2288,
      "step": 1200
    },
    {
      "epoch": 0.6987143655673561,
      "grad_norm": 2.333324432373047,
      "learning_rate": 0.0004968076616121309,
      "loss": 0.2061,
      "step": 1250
    },
    {
      "epoch": 0.7266629401900503,
      "grad_norm": 1.712575912475586,
      "learning_rate": 0.000498136645962733,
      "loss": 0.2189,
      "step": 1300
    },
    {
      "epoch": 0.7546115148127446,
      "grad_norm": 1.3168742656707764,
      "learning_rate": 0.0004959183673469388,
      "loss": 0.1978,
      "step": 1350
    },
    {
      "epoch": 0.7825600894354388,
      "grad_norm": 4.792356967926025,
      "learning_rate": 0.0004937000887311446,
      "loss": 0.2333,
      "step": 1400
    },
    {
      "epoch": 0.810508664058133,
      "grad_norm": 1.0282349586486816,
      "learning_rate": 0.0004914818101153504,
      "loss": 0.23,
      "step": 1450
    },
    {
      "epoch": 0.8384572386808273,
      "grad_norm": 0.6257219910621643,
      "learning_rate": 0.0004892635314995564,
      "loss": 0.2082,
      "step": 1500
    },
    {
      "epoch": 0.8664058133035215,
      "grad_norm": 1.1651033163070679,
      "learning_rate": 0.00048704525288376224,
      "loss": 0.1976,
      "step": 1550
    },
    {
      "epoch": 0.8943543879262158,
      "grad_norm": 1.0239338874816895,
      "learning_rate": 0.00048482697426796806,
      "loss": 0.1945,
      "step": 1600
    },
    {
      "epoch": 0.92230296254891,
      "grad_norm": 1.616534948348999,
      "learning_rate": 0.0004826086956521739,
      "loss": 0.1801,
      "step": 1650
    },
    {
      "epoch": 0.9502515371716043,
      "grad_norm": 1.4642572402954102,
      "learning_rate": 0.00048039041703637974,
      "loss": 0.1888,
      "step": 1700
    },
    {
      "epoch": 0.9782001117942984,
      "grad_norm": 5.763888359069824,
      "learning_rate": 0.00047817213842058566,
      "loss": 0.1798,
      "step": 1750
    },
    {
      "epoch": 1.0,
      "eval_f1": 0.8859060402684564,
      "eval_loss": 0.2657495439052582,
      "eval_runtime": 17.3445,
      "eval_samples_per_second": 211.017,
      "eval_steps_per_second": 6.63,
      "step": 1789
    },
    {
      "epoch": 1.0061486864169928,
      "grad_norm": 0.4531048536300659,
      "learning_rate": 0.0004759538598047915,
      "loss": 0.1844,
      "step": 1800
    },
    {
      "epoch": 1.034097261039687,
      "grad_norm": 1.737192153930664,
      "learning_rate": 0.00047373558118899735,
      "loss": 0.2026,
      "step": 1850
    },
    {
      "epoch": 1.0620458356623812,
      "grad_norm": 1.7833003997802734,
      "learning_rate": 0.00047151730257320317,
      "loss": 0.1724,
      "step": 1900
    },
    {
      "epoch": 1.0899944102850754,
      "grad_norm": 1.5082063674926758,
      "learning_rate": 0.0004692990239574091,
      "loss": 0.1891,
      "step": 1950
    },
    {
      "epoch": 1.1179429849077698,
      "grad_norm": 1.3961833715438843,
      "learning_rate": 0.00046708074534161496,
      "loss": 0.1737,
      "step": 2000
    },
    {
      "epoch": 1.145891559530464,
      "grad_norm": 0.9128915667533875,
      "learning_rate": 0.0004648624667258208,
      "loss": 0.1584,
      "step": 2050
    },
    {
      "epoch": 1.1738401341531581,
      "grad_norm": 1.8712306022644043,
      "learning_rate": 0.00046264418811002664,
      "loss": 0.1765,
      "step": 2100
    },
    {
      "epoch": 1.2017887087758523,
      "grad_norm": 1.0604172945022583,
      "learning_rate": 0.00046042590949423246,
      "loss": 0.2011,
      "step": 2150
    },
    {
      "epoch": 1.2297372833985467,
      "grad_norm": 2.5036442279815674,
      "learning_rate": 0.0004582076308784384,
      "loss": 0.1873,
      "step": 2200
    },
    {
      "epoch": 1.2576858580212409,
      "grad_norm": 2.0800774097442627,
      "learning_rate": 0.0004559893522626442,
      "loss": 0.1795,
      "step": 2250
    },
    {
      "epoch": 1.2856344326439353,
      "grad_norm": 1.3409812450408936,
      "learning_rate": 0.00045377107364685007,
      "loss": 0.1708,
      "step": 2300
    },
    {
      "epoch": 1.3135830072666295,
      "grad_norm": 1.0831228494644165,
      "learning_rate": 0.0004515527950310559,
      "loss": 0.1738,
      "step": 2350
    },
    {
      "epoch": 1.3415315818893236,
      "grad_norm": 0.5806594491004944,
      "learning_rate": 0.00044933451641526175,
      "loss": 0.1659,
      "step": 2400
    },
    {
      "epoch": 1.3694801565120178,
      "grad_norm": 0.9568463563919067,
      "learning_rate": 0.0004471162377994676,
      "loss": 0.1621,
      "step": 2450
    },
    {
      "epoch": 1.397428731134712,
      "grad_norm": 0.8637075424194336,
      "learning_rate": 0.0004448979591836735,
      "loss": 0.2007,
      "step": 2500
    },
    {
      "epoch": 1.4253773057574064,
      "grad_norm": 1.1704365015029907,
      "learning_rate": 0.0004426796805678793,
      "loss": 0.1556,
      "step": 2550
    },
    {
      "epoch": 1.4533258803801006,
      "grad_norm": 0.8730977773666382,
      "learning_rate": 0.0004404614019520852,
      "loss": 0.1741,
      "step": 2600
    },
    {
      "epoch": 1.481274455002795,
      "grad_norm": 0.7711090445518494,
      "learning_rate": 0.00043824312333629105,
      "loss": 0.1628,
      "step": 2650
    },
    {
      "epoch": 1.5092230296254892,
      "grad_norm": 0.6016321182250977,
      "learning_rate": 0.0004360248447204969,
      "loss": 0.1393,
      "step": 2700
    },
    {
      "epoch": 1.5371716042481833,
      "grad_norm": 0.9047902822494507,
      "learning_rate": 0.00043380656610470273,
      "loss": 0.1607,
      "step": 2750
    },
    {
      "epoch": 1.5651201788708775,
      "grad_norm": 1.0270776748657227,
      "learning_rate": 0.0004315882874889086,
      "loss": 0.1736,
      "step": 2800
    },
    {
      "epoch": 1.5930687534935717,
      "grad_norm": 0.7460580468177795,
      "learning_rate": 0.0004293700088731145,
      "loss": 0.1627,
      "step": 2850
    },
    {
      "epoch": 1.621017328116266,
      "grad_norm": 0.8114007115364075,
      "learning_rate": 0.00042715173025732034,
      "loss": 0.1756,
      "step": 2900
    },
    {
      "epoch": 1.6489659027389603,
      "grad_norm": 0.7761090397834778,
      "learning_rate": 0.0004249334516415262,
      "loss": 0.1491,
      "step": 2950
    },
    {
      "epoch": 1.6769144773616547,
      "grad_norm": 0.7432196140289307,
      "learning_rate": 0.00042271517302573203,
      "loss": 0.1663,
      "step": 3000
    },
    {
      "epoch": 1.7048630519843488,
      "grad_norm": 0.5656224489212036,
      "learning_rate": 0.0004204968944099379,
      "loss": 0.1808,
      "step": 3050
    },
    {
      "epoch": 1.732811626607043,
      "grad_norm": 1.192177653312683,
      "learning_rate": 0.00041827861579414377,
      "loss": 0.1605,
      "step": 3100
    },
    {
      "epoch": 1.7607602012297372,
      "grad_norm": 0.8815297484397888,
      "learning_rate": 0.00041606033717834964,
      "loss": 0.1595,
      "step": 3150
    },
    {
      "epoch": 1.7887087758524314,
      "grad_norm": 0.7288667559623718,
      "learning_rate": 0.00041384205856255545,
      "loss": 0.1529,
      "step": 3200
    },
    {
      "epoch": 1.8166573504751258,
      "grad_norm": 0.41470789909362793,
      "learning_rate": 0.0004116237799467613,
      "loss": 0.1661,
      "step": 3250
    },
    {
      "epoch": 1.8446059250978202,
      "grad_norm": 0.5929239392280579,
      "learning_rate": 0.00040940550133096714,
      "loss": 0.1658,
      "step": 3300
    },
    {
      "epoch": 1.8725544997205144,
      "grad_norm": 0.5527342557907104,
      "learning_rate": 0.00040718722271517306,
      "loss": 0.1686,
      "step": 3350
    },
    {
      "epoch": 1.9005030743432085,
      "grad_norm": 0.7223379611968994,
      "learning_rate": 0.0004049689440993789,
      "loss": 0.1616,
      "step": 3400
    },
    {
      "epoch": 1.9284516489659027,
      "grad_norm": 1.6212414503097534,
      "learning_rate": 0.00040275066548358475,
      "loss": 0.1455,
      "step": 3450
    },
    {
      "epoch": 1.9564002235885969,
      "grad_norm": 0.5442690849304199,
      "learning_rate": 0.00040053238686779056,
      "loss": 0.1666,
      "step": 3500
    },
    {
      "epoch": 1.984348798211291,
      "grad_norm": 0.4307806193828583,
      "learning_rate": 0.00039831410825199643,
      "loss": 0.1634,
      "step": 3550
    },
    {
      "epoch": 2.0,
      "eval_f1": 0.8839446782922429,
      "eval_loss": 0.26385268568992615,
      "eval_runtime": 15.9608,
      "eval_samples_per_second": 229.312,
      "eval_steps_per_second": 7.205,
      "step": 3578
    },
    {
      "epoch": 2.0122973728339857,
      "grad_norm": 1.0025759935379028,
      "learning_rate": 0.00039609582963620236,
      "loss": 0.16,
      "step": 3600
    },
    {
      "epoch": 2.04024594745668,
      "grad_norm": 0.7586847543716431,
      "learning_rate": 0.00039387755102040817,
      "loss": 0.18,
      "step": 3650
    },
    {
      "epoch": 2.068194522079374,
      "grad_norm": 1.3456194400787354,
      "learning_rate": 0.00039165927240461404,
      "loss": 0.1344,
      "step": 3700
    },
    {
      "epoch": 2.096143096702068,
      "grad_norm": 0.5425747632980347,
      "learning_rate": 0.00038944099378881986,
      "loss": 0.1386,
      "step": 3750
    },
    {
      "epoch": 2.1240916713247624,
      "grad_norm": 0.8643823266029358,
      "learning_rate": 0.0003872227151730258,
      "loss": 0.1609,
      "step": 3800
    },
    {
      "epoch": 2.1520402459474566,
      "grad_norm": 0.6121143698692322,
      "learning_rate": 0.0003850044365572316,
      "loss": 0.1433,
      "step": 3850
    },
    {
      "epoch": 2.1799888205701508,
      "grad_norm": 1.2089327573776245,
      "learning_rate": 0.00038278615794143747,
      "loss": 0.1635,
      "step": 3900
    },
    {
      "epoch": 2.207937395192845,
      "grad_norm": 0.845815122127533,
      "learning_rate": 0.0003805678793256433,
      "loss": 0.1633,
      "step": 3950
    },
    {
      "epoch": 2.2358859698155396,
      "grad_norm": 0.8113688230514526,
      "learning_rate": 0.00037834960070984915,
      "loss": 0.1497,
      "step": 4000
    },
    {
      "epoch": 2.2638345444382337,
      "grad_norm": 1.681497573852539,
      "learning_rate": 0.000376131322094055,
      "loss": 0.1402,
      "step": 4050
    },
    {
      "epoch": 2.291783119060928,
      "grad_norm": 0.9330260157585144,
      "learning_rate": 0.0003739130434782609,
      "loss": 0.1561,
      "step": 4100
    },
    {
      "epoch": 2.319731693683622,
      "grad_norm": 1.3728777170181274,
      "learning_rate": 0.0003716947648624667,
      "loss": 0.1228,
      "step": 4150
    },
    {
      "epoch": 2.3476802683063163,
      "grad_norm": 1.523483395576477,
      "learning_rate": 0.0003694764862466726,
      "loss": 0.1327,
      "step": 4200
    },
    {
      "epoch": 2.3756288429290104,
      "grad_norm": 0.6407974362373352,
      "learning_rate": 0.00036725820763087845,
      "loss": 0.1499,
      "step": 4250
    },
    {
      "epoch": 2.4035774175517046,
      "grad_norm": 0.5848202109336853,
      "learning_rate": 0.0003650399290150843,
      "loss": 0.1082,
      "step": 4300
    },
    {
      "epoch": 2.4315259921743992,
      "grad_norm": 0.21722891926765442,
      "learning_rate": 0.0003628216503992902,
      "loss": 0.1498,
      "step": 4350
    },
    {
      "epoch": 2.4594745667970934,
      "grad_norm": 0.7241097092628479,
      "learning_rate": 0.000360603371783496,
      "loss": 0.1354,
      "step": 4400
    },
    {
      "epoch": 2.4874231414197876,
      "grad_norm": 1.2298479080200195,
      "learning_rate": 0.00035838509316770187,
      "loss": 0.1409,
      "step": 4450
    },
    {
      "epoch": 2.5153717160424818,
      "grad_norm": 0.6272071003913879,
      "learning_rate": 0.00035616681455190774,
      "loss": 0.1358,
      "step": 4500
    },
    {
      "epoch": 2.543320290665176,
      "grad_norm": 0.9323145151138306,
      "learning_rate": 0.0003539485359361136,
      "loss": 0.1316,
      "step": 4550
    },
    {
      "epoch": 2.5712688652878706,
      "grad_norm": 0.44958174228668213,
      "learning_rate": 0.0003517302573203194,
      "loss": 0.1283,
      "step": 4600
    },
    {
      "epoch": 2.5992174399105643,
      "grad_norm": 0.30582261085510254,
      "learning_rate": 0.0003495119787045253,
      "loss": 0.1225,
      "step": 4650
    },
    {
      "epoch": 2.627166014533259,
      "grad_norm": 0.9854662418365479,
      "learning_rate": 0.00034729370008873117,
      "loss": 0.1527,
      "step": 4700
    },
    {
      "epoch": 2.655114589155953,
      "grad_norm": 0.3222338557243347,
      "learning_rate": 0.00034507542147293703,
      "loss": 0.1485,
      "step": 4750
    },
    {
      "epoch": 2.6830631637786473,
      "grad_norm": 0.7984679341316223,
      "learning_rate": 0.00034285714285714285,
      "loss": 0.15,
      "step": 4800
    },
    {
      "epoch": 2.7110117384013415,
      "grad_norm": 0.27581000328063965,
      "learning_rate": 0.0003406388642413487,
      "loss": 0.1423,
      "step": 4850
    },
    {
      "epoch": 2.7389603130240356,
      "grad_norm": 0.36392349004745483,
      "learning_rate": 0.00033842058562555454,
      "loss": 0.12,
      "step": 4900
    },
    {
      "epoch": 2.7669088876467303,
      "grad_norm": 0.8806673884391785,
      "learning_rate": 0.00033620230700976046,
      "loss": 0.1353,
      "step": 4950
    },
    {
      "epoch": 2.794857462269424,
      "grad_norm": 0.4529147446155548,
      "learning_rate": 0.0003339840283939663,
      "loss": 0.1524,
      "step": 5000
    },
    {
      "epoch": 2.8228060368921186,
      "grad_norm": 0.7661668658256531,
      "learning_rate": 0.00033176574977817214,
      "loss": 0.1431,
      "step": 5050
    },
    {
      "epoch": 2.850754611514813,
      "grad_norm": 0.7224089503288269,
      "learning_rate": 0.000329547471162378,
      "loss": 0.1328,
      "step": 5100
    },
    {
      "epoch": 2.878703186137507,
      "grad_norm": 0.7597375512123108,
      "learning_rate": 0.00032732919254658383,
      "loss": 0.1322,
      "step": 5150
    },
    {
      "epoch": 2.906651760760201,
      "grad_norm": 1.273445725440979,
      "learning_rate": 0.00032511091393078975,
      "loss": 0.1234,
      "step": 5200
    },
    {
      "epoch": 2.9346003353828953,
      "grad_norm": 1.428731918334961,
      "learning_rate": 0.00032289263531499557,
      "loss": 0.1476,
      "step": 5250
    },
    {
      "epoch": 2.96254891000559,
      "grad_norm": 0.4086410701274872,
      "learning_rate": 0.00032067435669920144,
      "loss": 0.1288,
      "step": 5300
    },
    {
      "epoch": 2.9904974846282837,
      "grad_norm": 1.4084666967391968,
      "learning_rate": 0.00031845607808340725,
      "loss": 0.1403,
      "step": 5350
    },
    {
      "epoch": 3.0,
      "eval_f1": 0.922771152754117,
      "eval_loss": 0.1881607472896576,
      "eval_runtime": 15.5665,
      "eval_samples_per_second": 235.12,
      "eval_steps_per_second": 7.388,
      "step": 5367
    },
    {
      "epoch": 3.0184460592509783,
      "grad_norm": 1.1583423614501953,
      "learning_rate": 0.0003162377994676132,
      "loss": 0.1348,
      "step": 5400
    },
    {
      "epoch": 3.0463946338736725,
      "grad_norm": 0.5038284063339233,
      "learning_rate": 0.000314019520851819,
      "loss": 0.1332,
      "step": 5450
    },
    {
      "epoch": 3.0743432084963667,
      "grad_norm": 1.139427900314331,
      "learning_rate": 0.00031180124223602486,
      "loss": 0.1071,
      "step": 5500
    },
    {
      "epoch": 3.102291783119061,
      "grad_norm": 0.632847011089325,
      "learning_rate": 0.0003095829636202307,
      "loss": 0.129,
      "step": 5550
    },
    {
      "epoch": 3.130240357741755,
      "grad_norm": 0.9477543830871582,
      "learning_rate": 0.00030736468500443655,
      "loss": 0.1296,
      "step": 5600
    },
    {
      "epoch": 3.1581889323644496,
      "grad_norm": 0.7813945412635803,
      "learning_rate": 0.0003051464063886424,
      "loss": 0.1356,
      "step": 5650
    },
    {
      "epoch": 3.186137506987144,
      "grad_norm": 1.0547813177108765,
      "learning_rate": 0.0003029281277728483,
      "loss": 0.1223,
      "step": 5700
    },
    {
      "epoch": 3.214086081609838,
      "grad_norm": 0.8783809542655945,
      "learning_rate": 0.0003007098491570541,
      "loss": 0.1086,
      "step": 5750
    },
    {
      "epoch": 3.242034656232532,
      "grad_norm": 1.7557278871536255,
      "learning_rate": 0.00029849157054126,
      "loss": 0.1176,
      "step": 5800
    },
    {
      "epoch": 3.2699832308552264,
      "grad_norm": 0.8250858783721924,
      "learning_rate": 0.0002962732919254659,
      "loss": 0.1244,
      "step": 5850
    },
    {
      "epoch": 3.2979318054779205,
      "grad_norm": 0.5830790996551514,
      "learning_rate": 0.0002940550133096717,
      "loss": 0.1255,
      "step": 5900
    },
    {
      "epoch": 3.3258803801006147,
      "grad_norm": 0.39319124817848206,
      "learning_rate": 0.0002918367346938776,
      "loss": 0.1145,
      "step": 5950
    },
    {
      "epoch": 3.3538289547233093,
      "grad_norm": 0.7558562755584717,
      "learning_rate": 0.0002896184560780834,
      "loss": 0.1449,
      "step": 6000
    },
    {
      "epoch": 3.3817775293460035,
      "grad_norm": 0.686841607093811,
      "learning_rate": 0.00028740017746228927,
      "loss": 0.1269,
      "step": 6050
    },
    {
      "epoch": 3.4097261039686977,
      "grad_norm": 0.3600720167160034,
      "learning_rate": 0.00028518189884649514,
      "loss": 0.1095,
      "step": 6100
    },
    {
      "epoch": 3.437674678591392,
      "grad_norm": 0.38374069333076477,
      "learning_rate": 0.000282963620230701,
      "loss": 0.11,
      "step": 6150
    },
    {
      "epoch": 3.465623253214086,
      "grad_norm": 0.736842691898346,
      "learning_rate": 0.0002807453416149068,
      "loss": 0.1099,
      "step": 6200
    },
    {
      "epoch": 3.49357182783678,
      "grad_norm": 0.21129830181598663,
      "learning_rate": 0.0002785270629991127,
      "loss": 0.1291,
      "step": 6250
    },
    {
      "epoch": 3.5215204024594744,
      "grad_norm": 0.6699479222297668,
      "learning_rate": 0.0002763087843833185,
      "loss": 0.1018,
      "step": 6300
    },
    {
      "epoch": 3.549468977082169,
      "grad_norm": 1.1297788619995117,
      "learning_rate": 0.00027409050576752443,
      "loss": 0.0986,
      "step": 6350
    },
    {
      "epoch": 3.577417551704863,
      "grad_norm": 0.7633982300758362,
      "learning_rate": 0.00027187222715173025,
      "loss": 0.1249,
      "step": 6400
    },
    {
      "epoch": 3.6053661263275574,
      "grad_norm": 1.3231158256530762,
      "learning_rate": 0.0002696539485359361,
      "loss": 0.1178,
      "step": 6450
    },
    {
      "epoch": 3.6333147009502516,
      "grad_norm": 0.5314428806304932,
      "learning_rate": 0.00026743566992014193,
      "loss": 0.1094,
      "step": 6500
    },
    {
      "epoch": 3.6612632755729457,
      "grad_norm": 0.3701280355453491,
      "learning_rate": 0.00026521739130434786,
      "loss": 0.12,
      "step": 6550
    },
    {
      "epoch": 3.68921185019564,
      "grad_norm": 0.6473478674888611,
      "learning_rate": 0.00026299911268855367,
      "loss": 0.106,
      "step": 6600
    },
    {
      "epoch": 3.717160424818334,
      "grad_norm": 0.4988769590854645,
      "learning_rate": 0.00026078083407275954,
      "loss": 0.1396,
      "step": 6650
    },
    {
      "epoch": 3.7451089994410287,
      "grad_norm": 1.0782082080841064,
      "learning_rate": 0.0002585625554569654,
      "loss": 0.1144,
      "step": 6700
    },
    {
      "epoch": 3.773057574063723,
      "grad_norm": 0.846603274345398,
      "learning_rate": 0.00025634427684117123,
      "loss": 0.1142,
      "step": 6750
    },
    {
      "epoch": 3.801006148686417,
      "grad_norm": 1.2522852420806885,
      "learning_rate": 0.00025412599822537715,
      "loss": 0.1086,
      "step": 6800
    },
    {
      "epoch": 3.8289547233091112,
      "grad_norm": 0.30715683102607727,
      "learning_rate": 0.00025190771960958297,
      "loss": 0.1023,
      "step": 6850
    },
    {
      "epoch": 3.8569032979318054,
      "grad_norm": 1.1746628284454346,
      "learning_rate": 0.00024968944099378884,
      "loss": 0.1182,
      "step": 6900
    },
    {
      "epoch": 3.8848518725544996,
      "grad_norm": 2.2346272468566895,
      "learning_rate": 0.0002474711623779947,
      "loss": 0.1207,
      "step": 6950
    },
    {
      "epoch": 3.9128004471771938,
      "grad_norm": 1.6879889965057373,
      "learning_rate": 0.0002452528837622005,
      "loss": 0.101,
      "step": 7000
    },
    {
      "epoch": 3.9407490217998884,
      "grad_norm": 0.777055025100708,
      "learning_rate": 0.00024303460514640642,
      "loss": 0.1051,
      "step": 7050
    },
    {
      "epoch": 3.9686975964225826,
      "grad_norm": 1.065346598625183,
      "learning_rate": 0.00024081632653061226,
      "loss": 0.1055,
      "step": 7100
    },
    {
      "epoch": 3.9966461710452768,
      "grad_norm": 2.3775737285614014,
      "learning_rate": 0.0002385980479148181,
      "loss": 0.097,
      "step": 7150
    },
    {
      "epoch": 4.0,
      "eval_f1": 0.9374824782730586,
      "eval_loss": 0.15188030898571014,
      "eval_runtime": 15.6619,
      "eval_samples_per_second": 233.688,
      "eval_steps_per_second": 7.343,
      "step": 7156
    },
    {
      "epoch": 4.024594745667971,
      "grad_norm": 0.8017207980155945,
      "learning_rate": 0.00023637976929902397,
      "loss": 0.1102,
      "step": 7200
    },
    {
      "epoch": 4.052543320290665,
      "grad_norm": 2.158339023590088,
      "learning_rate": 0.00023416149068322982,
      "loss": 0.1033,
      "step": 7250
    },
    {
      "epoch": 4.08049189491336,
      "grad_norm": 0.8033800721168518,
      "learning_rate": 0.00023194321206743569,
      "loss": 0.1054,
      "step": 7300
    },
    {
      "epoch": 4.1084404695360535,
      "grad_norm": 1.1149848699569702,
      "learning_rate": 0.00022972493345164153,
      "loss": 0.0884,
      "step": 7350
    },
    {
      "epoch": 4.136389044158748,
      "grad_norm": 0.18090881407260895,
      "learning_rate": 0.0002275066548358474,
      "loss": 0.091,
      "step": 7400
    },
    {
      "epoch": 4.164337618781442,
      "grad_norm": 0.6261000633239746,
      "learning_rate": 0.00022528837622005324,
      "loss": 0.0842,
      "step": 7450
    },
    {
      "epoch": 4.192286193404136,
      "grad_norm": 0.44622424244880676,
      "learning_rate": 0.00022307009760425908,
      "loss": 0.0933,
      "step": 7500
    },
    {
      "epoch": 4.220234768026831,
      "grad_norm": 1.2650983333587646,
      "learning_rate": 0.00022085181898846495,
      "loss": 0.0982,
      "step": 7550
    },
    {
      "epoch": 4.248183342649525,
      "grad_norm": 2.512014150619507,
      "learning_rate": 0.0002186335403726708,
      "loss": 0.0959,
      "step": 7600
    },
    {
      "epoch": 4.276131917272219,
      "grad_norm": 0.9572253227233887,
      "learning_rate": 0.00021641526175687667,
      "loss": 0.1197,
      "step": 7650
    },
    {
      "epoch": 4.304080491894913,
      "grad_norm": 0.8612358570098877,
      "learning_rate": 0.0002141969831410825,
      "loss": 0.0898,
      "step": 7700
    },
    {
      "epoch": 4.332029066517608,
      "grad_norm": 0.13563434779644012,
      "learning_rate": 0.00021197870452528838,
      "loss": 0.1268,
      "step": 7750
    },
    {
      "epoch": 4.3599776411403015,
      "grad_norm": 0.9207750558853149,
      "learning_rate": 0.00020976042590949425,
      "loss": 0.1138,
      "step": 7800
    },
    {
      "epoch": 4.387926215762996,
      "grad_norm": 0.15072673559188843,
      "learning_rate": 0.00020754214729370012,
      "loss": 0.0854,
      "step": 7850
    },
    {
      "epoch": 4.41587479038569,
      "grad_norm": 0.5902436375617981,
      "learning_rate": 0.00020532386867790596,
      "loss": 0.0967,
      "step": 7900
    },
    {
      "epoch": 4.4438233650083845,
      "grad_norm": 2.2286317348480225,
      "learning_rate": 0.0002031055900621118,
      "loss": 0.0939,
      "step": 7950
    },
    {
      "epoch": 4.471771939631079,
      "grad_norm": 0.6109320521354675,
      "learning_rate": 0.00020088731144631767,
      "loss": 0.0887,
      "step": 8000
    },
    {
      "epoch": 4.499720514253773,
      "grad_norm": 1.340773582458496,
      "learning_rate": 0.00019866903283052352,
      "loss": 0.0837,
      "step": 8050
    },
    {
      "epoch": 4.5276690888764675,
      "grad_norm": 1.1809900999069214,
      "learning_rate": 0.00019645075421472938,
      "loss": 0.0765,
      "step": 8100
    },
    {
      "epoch": 4.555617663499161,
      "grad_norm": 0.8299195170402527,
      "learning_rate": 0.00019423247559893523,
      "loss": 0.0972,
      "step": 8150
    },
    {
      "epoch": 4.583566238121856,
      "grad_norm": 0.18145732581615448,
      "learning_rate": 0.0001920141969831411,
      "loss": 0.075,
      "step": 8200
    },
    {
      "epoch": 4.61151481274455,
      "grad_norm": 0.2446570098400116,
      "learning_rate": 0.00018979591836734694,
      "loss": 0.0857,
      "step": 8250
    },
    {
      "epoch": 4.639463387367244,
      "grad_norm": 1.534759759902954,
      "learning_rate": 0.00018757763975155278,
      "loss": 0.0793,
      "step": 8300
    },
    {
      "epoch": 4.667411961989939,
      "grad_norm": 1.6062694787979126,
      "learning_rate": 0.00018535936113575865,
      "loss": 0.0931,
      "step": 8350
    },
    {
      "epoch": 4.6953605366126325,
      "grad_norm": 2.1251513957977295,
      "learning_rate": 0.0001831410825199645,
      "loss": 0.1003,
      "step": 8400
    },
    {
      "epoch": 4.723309111235327,
      "grad_norm": 0.9858797788619995,
      "learning_rate": 0.00018092280390417036,
      "loss": 0.082,
      "step": 8450
    },
    {
      "epoch": 4.751257685858021,
      "grad_norm": 1.1654304265975952,
      "learning_rate": 0.0001787045252883762,
      "loss": 0.0603,
      "step": 8500
    },
    {
      "epoch": 4.7792062604807155,
      "grad_norm": 1.731586217880249,
      "learning_rate": 0.00017648624667258208,
      "loss": 0.0593,
      "step": 8550
    },
    {
      "epoch": 4.807154835103409,
      "grad_norm": 0.524991512298584,
      "learning_rate": 0.00017426796805678795,
      "loss": 0.0928,
      "step": 8600
    },
    {
      "epoch": 4.835103409726104,
      "grad_norm": 0.76012122631073,
      "learning_rate": 0.00017204968944099382,
      "loss": 0.0901,
      "step": 8650
    },
    {
      "epoch": 4.8630519843487985,
      "grad_norm": 1.038218379020691,
      "learning_rate": 0.00016983141082519966,
      "loss": 0.0712,
      "step": 8700
    },
    {
      "epoch": 4.891000558971492,
      "grad_norm": 0.7694293260574341,
      "learning_rate": 0.0001676131322094055,
      "loss": 0.0801,
      "step": 8750
    },
    {
      "epoch": 4.918949133594187,
      "grad_norm": 1.4149705171585083,
      "learning_rate": 0.00016539485359361137,
      "loss": 0.1054,
      "step": 8800
    },
    {
      "epoch": 4.946897708216881,
      "grad_norm": 0.35135123133659363,
      "learning_rate": 0.00016317657497781721,
      "loss": 0.0917,
      "step": 8850
    },
    {
      "epoch": 4.974846282839575,
      "grad_norm": 1.8134843111038208,
      "learning_rate": 0.00016095829636202308,
      "loss": 0.0946,
      "step": 8900
    },
    {
      "epoch": 5.0,
      "eval_f1": 0.9494430162810625,
      "eval_loss": 0.12213503569364548,
      "eval_runtime": 16.0206,
      "eval_samples_per_second": 228.456,
      "eval_steps_per_second": 7.178,
      "step": 8945
    },
    {
      "epoch": 5.00279485746227,
      "grad_norm": 0.8835757374763489,
      "learning_rate": 0.00015874001774622893,
      "loss": 0.0792,
      "step": 8950
    },
    {
      "epoch": 5.0307434320849636,
      "grad_norm": 0.5846070647239685,
      "learning_rate": 0.0001565217391304348,
      "loss": 0.0834,
      "step": 9000
    },
    {
      "epoch": 5.058692006707658,
      "grad_norm": 0.4132053256034851,
      "learning_rate": 0.00015430346051464064,
      "loss": 0.0622,
      "step": 9050
    },
    {
      "epoch": 5.086640581330352,
      "grad_norm": 3.0904829502105713,
      "learning_rate": 0.00015208518189884648,
      "loss": 0.0626,
      "step": 9100
    },
    {
      "epoch": 5.1145891559530465,
      "grad_norm": 2.243579387664795,
      "learning_rate": 0.00014986690328305235,
      "loss": 0.0795,
      "step": 9150
    },
    {
      "epoch": 5.14253773057574,
      "grad_norm": 1.4317448139190674,
      "learning_rate": 0.0001476486246672582,
      "loss": 0.0642,
      "step": 9200
    },
    {
      "epoch": 5.170486305198435,
      "grad_norm": 0.3954031765460968,
      "learning_rate": 0.00014543034605146406,
      "loss": 0.0748,
      "step": 9250
    },
    {
      "epoch": 5.1984348798211295,
      "grad_norm": 0.7129592895507812,
      "learning_rate": 0.0001432120674356699,
      "loss": 0.0631,
      "step": 9300
    },
    {
      "epoch": 5.226383454443823,
      "grad_norm": 0.7119801044464111,
      "learning_rate": 0.0001409937888198758,
      "loss": 0.0672,
      "step": 9350
    },
    {
      "epoch": 5.254332029066518,
      "grad_norm": 1.8398321866989136,
      "learning_rate": 0.00013877551020408165,
      "loss": 0.0627,
      "step": 9400
    },
    {
      "epoch": 5.282280603689212,
      "grad_norm": 1.1668697595596313,
      "learning_rate": 0.0001365572315882875,
      "loss": 0.0661,
      "step": 9450
    },
    {
      "epoch": 5.310229178311906,
      "grad_norm": 0.6019765734672546,
      "learning_rate": 0.00013433895297249336,
      "loss": 0.0615,
      "step": 9500
    },
    {
      "epoch": 5.3381777529346,
      "grad_norm": 2.365870237350464,
      "learning_rate": 0.0001321206743566992,
      "loss": 0.0582,
      "step": 9550
    },
    {
      "epoch": 5.366126327557295,
      "grad_norm": 0.10710655897855759,
      "learning_rate": 0.00012990239574090507,
      "loss": 0.052,
      "step": 9600
    },
    {
      "epoch": 5.394074902179989,
      "grad_norm": 0.6959599852561951,
      "learning_rate": 0.0001276841171251109,
      "loss": 0.0558,
      "step": 9650
    },
    {
      "epoch": 5.422023476802683,
      "grad_norm": 0.11615774035453796,
      "learning_rate": 0.00012546583850931678,
      "loss": 0.0625,
      "step": 9700
    },
    {
      "epoch": 5.4499720514253776,
      "grad_norm": 0.057329438626766205,
      "learning_rate": 0.00012324755989352263,
      "loss": 0.0455,
      "step": 9750
    },
    {
      "epoch": 5.477920626048071,
      "grad_norm": 3.285327196121216,
      "learning_rate": 0.00012102928127772848,
      "loss": 0.0674,
      "step": 9800
    },
    {
      "epoch": 5.505869200670766,
      "grad_norm": 2.17206072807312,
      "learning_rate": 0.00011881100266193434,
      "loss": 0.0678,
      "step": 9850
    },
    {
      "epoch": 5.5338177752934605,
      "grad_norm": 0.2885390520095825,
      "learning_rate": 0.0001165927240461402,
      "loss": 0.0512,
      "step": 9900
    },
    {
      "epoch": 5.561766349916154,
      "grad_norm": 2.1970298290252686,
      "learning_rate": 0.00011437444543034606,
      "loss": 0.0598,
      "step": 9950
    },
    {
      "epoch": 5.589714924538849,
      "grad_norm": 0.1274396926164627,
      "learning_rate": 0.00011215616681455192,
      "loss": 0.0717,
      "step": 10000
    },
    {
      "epoch": 5.617663499161543,
      "grad_norm": 0.7299688458442688,
      "learning_rate": 0.00010993788819875776,
      "loss": 0.0797,
      "step": 10050
    },
    {
      "epoch": 5.645612073784237,
      "grad_norm": 0.5048921704292297,
      "learning_rate": 0.00010771960958296362,
      "loss": 0.0592,
      "step": 10100
    },
    {
      "epoch": 5.673560648406931,
      "grad_norm": 1.5433499813079834,
      "learning_rate": 0.00010550133096716947,
      "loss": 0.0702,
      "step": 10150
    },
    {
      "epoch": 5.701509223029626,
      "grad_norm": 0.23040899634361267,
      "learning_rate": 0.00010328305235137533,
      "loss": 0.0654,
      "step": 10200
    },
    {
      "epoch": 5.729457797652319,
      "grad_norm": 1.693513035774231,
      "learning_rate": 0.00010106477373558119,
      "loss": 0.0687,
      "step": 10250
    },
    {
      "epoch": 5.757406372275014,
      "grad_norm": 4.076319694519043,
      "learning_rate": 9.884649511978706e-05,
      "loss": 0.0639,
      "step": 10300
    },
    {
      "epoch": 5.785354946897709,
      "grad_norm": 0.48212429881095886,
      "learning_rate": 9.662821650399291e-05,
      "loss": 0.0604,
      "step": 10350
    },
    {
      "epoch": 5.813303521520402,
      "grad_norm": 0.04306049272418022,
      "learning_rate": 9.440993788819877e-05,
      "loss": 0.0495,
      "step": 10400
    },
    {
      "epoch": 5.841252096143097,
      "grad_norm": 0.23358555138111115,
      "learning_rate": 9.223602484472051e-05,
      "loss": 0.0615,
      "step": 10450
    },
    {
      "epoch": 5.869200670765791,
      "grad_norm": 1.8047716617584229,
      "learning_rate": 9.001774622892636e-05,
      "loss": 0.0531,
      "step": 10500
    },
    {
      "epoch": 5.897149245388485,
      "grad_norm": 0.1505836844444275,
      "learning_rate": 8.779946761313221e-05,
      "loss": 0.0529,
      "step": 10550
    },
    {
      "epoch": 5.92509782001118,
      "grad_norm": 0.42754456400871277,
      "learning_rate": 8.558118899733806e-05,
      "loss": 0.0567,
      "step": 10600
    },
    {
      "epoch": 5.953046394633874,
      "grad_norm": 1.9926128387451172,
      "learning_rate": 8.336291038154392e-05,
      "loss": 0.0473,
      "step": 10650
    },
    {
      "epoch": 5.980994969256568,
      "grad_norm": 0.3347943425178528,
      "learning_rate": 8.114463176574978e-05,
      "loss": 0.0688,
      "step": 10700
    },
    {
      "epoch": 6.0,
      "eval_f1": 0.9666100735710244,
      "eval_loss": 0.08690152317285538,
      "eval_runtime": 15.4749,
      "eval_samples_per_second": 236.512,
      "eval_steps_per_second": 7.431,
      "step": 10734
    },
    {
      "epoch": 6.008943543879262,
      "grad_norm": 0.8442544341087341,
      "learning_rate": 7.892635314995563e-05,
      "loss": 0.0456,
      "step": 10750
    },
    {
      "epoch": 6.036892118501957,
      "grad_norm": 1.1793649196624756,
      "learning_rate": 7.67080745341615e-05,
      "loss": 0.0434,
      "step": 10800
    },
    {
      "epoch": 6.06484069312465,
      "grad_norm": 0.9821094870567322,
      "learning_rate": 7.448979591836736e-05,
      "loss": 0.0455,
      "step": 10850
    },
    {
      "epoch": 6.092789267747345,
      "grad_norm": 0.5856317281723022,
      "learning_rate": 7.227151730257321e-05,
      "loss": 0.0365,
      "step": 10900
    },
    {
      "epoch": 6.120737842370039,
      "grad_norm": 0.5461094379425049,
      "learning_rate": 7.005323868677906e-05,
      "loss": 0.0342,
      "step": 10950
    },
    {
      "epoch": 6.148686416992733,
      "grad_norm": 2.0931904315948486,
      "learning_rate": 6.783496007098491e-05,
      "loss": 0.0423,
      "step": 11000
    },
    {
      "epoch": 6.176634991615428,
      "grad_norm": 0.046953145414590836,
      "learning_rate": 6.561668145519077e-05,
      "loss": 0.0523,
      "step": 11050
    },
    {
      "epoch": 6.204583566238122,
      "grad_norm": 1.0641871690750122,
      "learning_rate": 6.339840283939662e-05,
      "loss": 0.0421,
      "step": 11100
    },
    {
      "epoch": 6.232532140860816,
      "grad_norm": 0.4122409224510193,
      "learning_rate": 6.118012422360248e-05,
      "loss": 0.0441,
      "step": 11150
    },
    {
      "epoch": 6.26048071548351,
      "grad_norm": 0.2935519218444824,
      "learning_rate": 5.8961845607808344e-05,
      "loss": 0.0354,
      "step": 11200
    },
    {
      "epoch": 6.288429290106205,
      "grad_norm": 0.8135849833488464,
      "learning_rate": 5.6743566992014193e-05,
      "loss": 0.0333,
      "step": 11250
    },
    {
      "epoch": 6.316377864728899,
      "grad_norm": 1.1806564331054688,
      "learning_rate": 5.4525288376220056e-05,
      "loss": 0.0614,
      "step": 11300
    },
    {
      "epoch": 6.344326439351593,
      "grad_norm": 0.5362429022789001,
      "learning_rate": 5.230700976042591e-05,
      "loss": 0.0453,
      "step": 11350
    },
    {
      "epoch": 6.372275013974288,
      "grad_norm": 0.10853467881679535,
      "learning_rate": 5.008873114463176e-05,
      "loss": 0.0371,
      "step": 11400
    },
    {
      "epoch": 6.400223588596981,
      "grad_norm": 2.1878767013549805,
      "learning_rate": 4.787045252883762e-05,
      "loss": 0.0418,
      "step": 11450
    },
    {
      "epoch": 6.428172163219676,
      "grad_norm": 1.1209208965301514,
      "learning_rate": 4.565217391304348e-05,
      "loss": 0.0257,
      "step": 11500
    },
    {
      "epoch": 6.45612073784237,
      "grad_norm": 1.4330588579177856,
      "learning_rate": 4.343389529724934e-05,
      "loss": 0.0595,
      "step": 11550
    },
    {
      "epoch": 6.484069312465064,
      "grad_norm": 1.9309511184692383,
      "learning_rate": 4.1215616681455187e-05,
      "loss": 0.0357,
      "step": 11600
    },
    {
      "epoch": 6.512017887087758,
      "grad_norm": 0.004909028764814138,
      "learning_rate": 3.899733806566105e-05,
      "loss": 0.0332,
      "step": 11650
    },
    {
      "epoch": 6.539966461710453,
      "grad_norm": 0.5627462267875671,
      "learning_rate": 3.6779059449866906e-05,
      "loss": 0.0416,
      "step": 11700
    },
    {
      "epoch": 6.567915036333147,
      "grad_norm": 4.456967830657959,
      "learning_rate": 3.456078083407276e-05,
      "loss": 0.045,
      "step": 11750
    },
    {
      "epoch": 6.595863610955841,
      "grad_norm": 2.8317513465881348,
      "learning_rate": 3.234250221827861e-05,
      "loss": 0.0305,
      "step": 11800
    },
    {
      "epoch": 6.623812185578536,
      "grad_norm": 2.5559463500976562,
      "learning_rate": 3.012422360248447e-05,
      "loss": 0.0297,
      "step": 11850
    },
    {
      "epoch": 6.651760760201229,
      "grad_norm": 0.5523830056190491,
      "learning_rate": 2.790594498669033e-05,
      "loss": 0.0325,
      "step": 11900
    },
    {
      "epoch": 6.679709334823924,
      "grad_norm": 0.04988823086023331,
      "learning_rate": 2.5687666370896183e-05,
      "loss": 0.0313,
      "step": 11950
    },
    {
      "epoch": 6.707657909446619,
      "grad_norm": 1.5902944803237915,
      "learning_rate": 2.3469387755102043e-05,
      "loss": 0.0369,
      "step": 12000
    },
    {
      "epoch": 6.735606484069312,
      "grad_norm": 1.3617613315582275,
      "learning_rate": 2.1251109139307896e-05,
      "loss": 0.0338,
      "step": 12050
    },
    {
      "epoch": 6.763555058692007,
      "grad_norm": 0.13116395473480225,
      "learning_rate": 1.9032830523513755e-05,
      "loss": 0.0254,
      "step": 12100
    },
    {
      "epoch": 6.791503633314701,
      "grad_norm": 0.7528799176216125,
      "learning_rate": 1.6814551907719608e-05,
      "loss": 0.0322,
      "step": 12150
    },
    {
      "epoch": 6.819452207937395,
      "grad_norm": 1.894806146621704,
      "learning_rate": 1.4596273291925467e-05,
      "loss": 0.0254,
      "step": 12200
    },
    {
      "epoch": 6.847400782560089,
      "grad_norm": 0.5565615892410278,
      "learning_rate": 1.2377994676131324e-05,
      "loss": 0.0284,
      "step": 12250
    },
    {
      "epoch": 6.875349357182784,
      "grad_norm": 0.45591992139816284,
      "learning_rate": 1.015971606033718e-05,
      "loss": 0.0356,
      "step": 12300
    },
    {
      "epoch": 6.9032979318054775,
      "grad_norm": 1.1384029388427734,
      "learning_rate": 7.941437444543036e-06,
      "loss": 0.0308,
      "step": 12350
    },
    {
      "epoch": 6.931246506428172,
      "grad_norm": 0.24669720232486725,
      "learning_rate": 5.7231588287488905e-06,
      "loss": 0.0385,
      "step": 12400
    },
    {
      "epoch": 6.959195081050867,
      "grad_norm": 2.1776230335235596,
      "learning_rate": 3.5048802129547475e-06,
      "loss": 0.0271,
      "step": 12450
    },
    {
      "epoch": 6.98714365567356,
      "grad_norm": 0.8091051578521729,
      "learning_rate": 1.2866015971606034e-06,
      "loss": 0.0303,
      "step": 12500
    },
    {
      "epoch": 7.0,
      "eval_f1": 0.9760225669957687,
      "eval_loss": 0.06743501126766205,
      "eval_runtime": 16.4262,
      "eval_samples_per_second": 222.815,
      "eval_steps_per_second": 7.001,
      "step": 12523
    },
    {
      "epoch": 7.0,
      "step": 12523,
      "total_flos": 2.1045216318570566e+19,
      "train_loss": 0.12371139868869004,
      "train_runtime": 2341.9199,
      "train_samples_per_second": 171.111,
      "train_steps_per_second": 5.347
    }
  ],
  "logging_steps": 50,
  "max_steps": 12523,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 7,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 2.1045216318570566e+19,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": null
}