{
  "best_metric": 1.664876103401184,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.0521614396557345,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0002608071982786725,
      "grad_norm": 15.92446517944336,
      "learning_rate": 1e-05,
      "loss": 8.4942,
      "step": 1
    },
    {
      "epoch": 0.0002608071982786725,
      "eval_loss": 2.6137824058532715,
      "eval_runtime": 465.4458,
      "eval_samples_per_second": 13.875,
      "eval_steps_per_second": 3.47,
      "step": 1
    },
    {
      "epoch": 0.000521614396557345,
      "grad_norm": 19.50524139404297,
      "learning_rate": 2e-05,
      "loss": 8.2499,
      "step": 2
    },
    {
      "epoch": 0.0007824215948360175,
      "grad_norm": 19.322851181030273,
      "learning_rate": 3e-05,
      "loss": 8.1554,
      "step": 3
    },
    {
      "epoch": 0.00104322879311469,
      "grad_norm": 18.32716941833496,
      "learning_rate": 4e-05,
      "loss": 7.7783,
      "step": 4
    },
    {
      "epoch": 0.0013040359913933624,
      "grad_norm": 17.61314582824707,
      "learning_rate": 5e-05,
      "loss": 7.5059,
      "step": 5
    },
    {
      "epoch": 0.001564843189672035,
      "grad_norm": 18.145336151123047,
      "learning_rate": 6e-05,
      "loss": 8.0427,
      "step": 6
    },
    {
      "epoch": 0.0018256503879507074,
      "grad_norm": 19.722463607788086,
      "learning_rate": 7e-05,
      "loss": 7.2592,
      "step": 7
    },
    {
      "epoch": 0.00208645758622938,
      "grad_norm": 19.03963851928711,
      "learning_rate": 8e-05,
      "loss": 7.5553,
      "step": 8
    },
    {
      "epoch": 0.0023472647845080522,
      "grad_norm": 21.876693725585938,
      "learning_rate": 9e-05,
      "loss": 7.6506,
      "step": 9
    },
    {
      "epoch": 0.002608071982786725,
      "grad_norm": 19.55409049987793,
      "learning_rate": 0.0001,
      "loss": 6.9203,
      "step": 10
    },
    {
      "epoch": 0.0028688791810653974,
      "grad_norm": 21.01988410949707,
      "learning_rate": 9.999316524962345e-05,
      "loss": 7.162,
      "step": 11
    },
    {
      "epoch": 0.00312968637934407,
      "grad_norm": 18.654972076416016,
      "learning_rate": 9.997266286704631e-05,
      "loss": 6.9283,
      "step": 12
    },
    {
      "epoch": 0.0033904935776227422,
      "grad_norm": 20.48809051513672,
      "learning_rate": 9.993849845741524e-05,
      "loss": 7.0406,
      "step": 13
    },
    {
      "epoch": 0.003651300775901415,
      "grad_norm": 16.33949851989746,
      "learning_rate": 9.989068136093873e-05,
      "loss": 6.7105,
      "step": 14
    },
    {
      "epoch": 0.003912107974180087,
      "grad_norm": 18.511028289794922,
      "learning_rate": 9.98292246503335e-05,
      "loss": 7.4781,
      "step": 15
    },
    {
      "epoch": 0.00417291517245876,
      "grad_norm": 18.401561737060547,
      "learning_rate": 9.975414512725057e-05,
      "loss": 6.6241,
      "step": 16
    },
    {
      "epoch": 0.004433722370737432,
      "grad_norm": 17.54397964477539,
      "learning_rate": 9.966546331768191e-05,
      "loss": 7.11,
      "step": 17
    },
    {
      "epoch": 0.0046945295690161044,
      "grad_norm": 18.686908721923828,
      "learning_rate": 9.956320346634876e-05,
      "loss": 6.9694,
      "step": 18
    },
    {
      "epoch": 0.0049553367672947775,
      "grad_norm": 17.026294708251953,
      "learning_rate": 9.944739353007344e-05,
      "loss": 6.8269,
      "step": 19
    },
    {
      "epoch": 0.00521614396557345,
      "grad_norm": 17.851730346679688,
      "learning_rate": 9.931806517013612e-05,
      "loss": 7.0332,
      "step": 20
    },
    {
      "epoch": 0.005476951163852123,
      "grad_norm": 18.39048194885254,
      "learning_rate": 9.917525374361912e-05,
      "loss": 6.944,
      "step": 21
    },
    {
      "epoch": 0.005737758362130795,
      "grad_norm": 18.222721099853516,
      "learning_rate": 9.901899829374047e-05,
      "loss": 6.9778,
      "step": 22
    },
    {
      "epoch": 0.005998565560409467,
      "grad_norm": 18.133663177490234,
      "learning_rate": 9.884934153917997e-05,
      "loss": 7.1548,
      "step": 23
    },
    {
      "epoch": 0.00625937275868814,
      "grad_norm": 18.435504913330078,
      "learning_rate": 9.86663298624003e-05,
      "loss": 6.8018,
      "step": 24
    },
    {
      "epoch": 0.006520179956966812,
      "grad_norm": 16.382251739501953,
      "learning_rate": 9.847001329696653e-05,
      "loss": 6.8027,
      "step": 25
    },
    {
      "epoch": 0.0067809871552454845,
      "grad_norm": 17.43813705444336,
      "learning_rate": 9.826044551386744e-05,
      "loss": 7.1021,
      "step": 26
    },
    {
      "epoch": 0.0070417943535241575,
      "grad_norm": 16.30862808227539,
      "learning_rate": 9.803768380684242e-05,
      "loss": 6.6197,
      "step": 27
    },
    {
      "epoch": 0.00730260155180283,
      "grad_norm": 16.058307647705078,
      "learning_rate": 9.780178907671789e-05,
      "loss": 6.5051,
      "step": 28
    },
    {
      "epoch": 0.007563408750081502,
      "grad_norm": 19.310583114624023,
      "learning_rate": 9.755282581475769e-05,
      "loss": 7.6339,
      "step": 29
    },
    {
      "epoch": 0.007824215948360174,
      "grad_norm": 17.598913192749023,
      "learning_rate": 9.729086208503174e-05,
      "loss": 7.3727,
      "step": 30
    },
    {
      "epoch": 0.008085023146638847,
      "grad_norm": 18.276451110839844,
      "learning_rate": 9.701596950580806e-05,
      "loss": 6.9771,
      "step": 31
    },
    {
      "epoch": 0.00834583034491752,
      "grad_norm": 16.577896118164062,
      "learning_rate": 9.672822322997305e-05,
      "loss": 7.3515,
      "step": 32
    },
    {
      "epoch": 0.008606637543196191,
      "grad_norm": 16.773691177368164,
      "learning_rate": 9.642770192448536e-05,
      "loss": 6.9659,
      "step": 33
    },
    {
      "epoch": 0.008867444741474865,
      "grad_norm": 17.366744995117188,
      "learning_rate": 9.611448774886924e-05,
      "loss": 7.7237,
      "step": 34
    },
    {
      "epoch": 0.009128251939753538,
      "grad_norm": 19.125431060791016,
      "learning_rate": 9.578866633275288e-05,
      "loss": 7.7703,
      "step": 35
    },
    {
      "epoch": 0.009389059138032209,
      "grad_norm": 18.292224884033203,
      "learning_rate": 9.545032675245813e-05,
      "loss": 7.593,
      "step": 36
    },
    {
      "epoch": 0.009649866336310882,
      "grad_norm": 18.408029556274414,
      "learning_rate": 9.509956150664796e-05,
      "loss": 7.2722,
      "step": 37
    },
    {
      "epoch": 0.009910673534589555,
      "grad_norm": 18.77663803100586,
      "learning_rate": 9.473646649103818e-05,
      "loss": 7.6034,
      "step": 38
    },
    {
      "epoch": 0.010171480732868228,
      "grad_norm": 18.99334716796875,
      "learning_rate": 9.43611409721806e-05,
      "loss": 6.8617,
      "step": 39
    },
    {
      "epoch": 0.0104322879311469,
      "grad_norm": 19.755197525024414,
      "learning_rate": 9.397368756032445e-05,
      "loss": 7.2095,
      "step": 40
    },
    {
      "epoch": 0.010693095129425572,
      "grad_norm": 20.920644760131836,
      "learning_rate": 9.357421218136386e-05,
      "loss": 7.0221,
      "step": 41
    },
    {
      "epoch": 0.010953902327704245,
      "grad_norm": 19.787565231323242,
      "learning_rate": 9.316282404787871e-05,
      "loss": 7.3358,
      "step": 42
    },
    {
      "epoch": 0.011214709525982917,
      "grad_norm": 21.234771728515625,
      "learning_rate": 9.273963562927695e-05,
      "loss": 7.7954,
      "step": 43
    },
    {
      "epoch": 0.01147551672426159,
      "grad_norm": 22.54673957824707,
      "learning_rate": 9.230476262104677e-05,
      "loss": 7.7269,
      "step": 44
    },
    {
      "epoch": 0.011736323922540263,
      "grad_norm": 18.309297561645508,
      "learning_rate": 9.185832391312644e-05,
      "loss": 8.0643,
      "step": 45
    },
    {
      "epoch": 0.011997131120818934,
      "grad_norm": 24.732643127441406,
      "learning_rate": 9.140044155740101e-05,
      "loss": 7.9585,
      "step": 46
    },
    {
      "epoch": 0.012257938319097607,
      "grad_norm": 17.507062911987305,
      "learning_rate": 9.093124073433463e-05,
      "loss": 7.3288,
      "step": 47
    },
    {
      "epoch": 0.01251874551737628,
      "grad_norm": 19.189712524414062,
      "learning_rate": 9.045084971874738e-05,
      "loss": 7.202,
      "step": 48
    },
    {
      "epoch": 0.012779552715654952,
      "grad_norm": 22.24582862854004,
      "learning_rate": 8.995939984474624e-05,
      "loss": 8.096,
      "step": 49
    },
    {
      "epoch": 0.013040359913933625,
      "grad_norm": 25.58473777770996,
      "learning_rate": 8.945702546981969e-05,
      "loss": 8.0074,
      "step": 50
    },
    {
      "epoch": 0.013040359913933625,
      "eval_loss": 2.1982150077819824,
      "eval_runtime": 468.8245,
      "eval_samples_per_second": 13.775,
      "eval_steps_per_second": 3.445,
      "step": 50
    },
    {
      "epoch": 0.013301167112212298,
      "grad_norm": 38.178443908691406,
      "learning_rate": 8.894386393810563e-05,
      "loss": 10.5693,
      "step": 51
    },
    {
      "epoch": 0.013561974310490969,
      "grad_norm": 18.324586868286133,
      "learning_rate": 8.842005554284296e-05,
      "loss": 8.6958,
      "step": 52
    },
    {
      "epoch": 0.013822781508769642,
      "grad_norm": 15.406447410583496,
      "learning_rate": 8.788574348801675e-05,
      "loss": 7.5827,
      "step": 53
    },
    {
      "epoch": 0.014083588707048315,
      "grad_norm": 12.764050483703613,
      "learning_rate": 8.73410738492077e-05,
      "loss": 8.0207,
      "step": 54
    },
    {
      "epoch": 0.014344395905326986,
      "grad_norm": 12.358431816101074,
      "learning_rate": 8.678619553365659e-05,
      "loss": 7.2749,
      "step": 55
    },
    {
      "epoch": 0.01460520310360566,
      "grad_norm": 11.963732719421387,
      "learning_rate": 8.622126023955446e-05,
      "loss": 7.7075,
      "step": 56
    },
    {
      "epoch": 0.014866010301884332,
      "grad_norm": 12.6146821975708,
      "learning_rate": 8.564642241456986e-05,
      "loss": 7.1581,
      "step": 57
    },
    {
      "epoch": 0.015126817500163004,
      "grad_norm": 14.263411521911621,
      "learning_rate": 8.506183921362443e-05,
      "loss": 7.1021,
      "step": 58
    },
    {
      "epoch": 0.015387624698441677,
      "grad_norm": 13.39104175567627,
      "learning_rate": 8.44676704559283e-05,
      "loss": 6.8996,
      "step": 59
    },
    {
      "epoch": 0.015648431896720348,
      "grad_norm": 11.874906539916992,
      "learning_rate": 8.386407858128706e-05,
      "loss": 6.2483,
      "step": 60
    },
    {
      "epoch": 0.01590923909499902,
      "grad_norm": 14.36909008026123,
      "learning_rate": 8.32512286056924e-05,
      "loss": 7.5651,
      "step": 61
    },
    {
      "epoch": 0.016170046293277694,
      "grad_norm": 13.23278522491455,
      "learning_rate": 8.262928807620843e-05,
      "loss": 6.8975,
      "step": 62
    },
    {
      "epoch": 0.016430853491556367,
      "grad_norm": 16.982460021972656,
      "learning_rate": 8.199842702516583e-05,
      "loss": 7.1759,
      "step": 63
    },
    {
      "epoch": 0.01669166068983504,
      "grad_norm": 12.389324188232422,
      "learning_rate": 8.135881792367686e-05,
      "loss": 6.6332,
      "step": 64
    },
    {
      "epoch": 0.016952467888113713,
      "grad_norm": 13.926976203918457,
      "learning_rate": 8.07106356344834e-05,
      "loss": 6.9457,
      "step": 65
    },
    {
      "epoch": 0.017213275086392383,
      "grad_norm": 12.722073554992676,
      "learning_rate": 8.005405736415126e-05,
      "loss": 6.6812,
      "step": 66
    },
    {
      "epoch": 0.017474082284671056,
      "grad_norm": 13.6449556350708,
      "learning_rate": 7.938926261462366e-05,
      "loss": 6.8593,
      "step": 67
    },
    {
      "epoch": 0.01773488948294973,
      "grad_norm": 13.449191093444824,
      "learning_rate": 7.871643313414718e-05,
      "loss": 6.5819,
      "step": 68
    },
    {
      "epoch": 0.017995696681228402,
      "grad_norm": 12.721368789672852,
      "learning_rate": 7.803575286758364e-05,
      "loss": 6.3668,
      "step": 69
    },
    {
      "epoch": 0.018256503879507075,
      "grad_norm": 14.585891723632812,
      "learning_rate": 7.734740790612136e-05,
      "loss": 6.6679,
      "step": 70
    },
    {
      "epoch": 0.018517311077785748,
      "grad_norm": 13.508546829223633,
      "learning_rate": 7.66515864363997e-05,
      "loss": 7.2189,
      "step": 71
    },
    {
      "epoch": 0.018778118276064418,
      "grad_norm": 14.999940872192383,
      "learning_rate": 7.594847868906076e-05,
      "loss": 7.0786,
      "step": 72
    },
    {
      "epoch": 0.01903892547434309,
      "grad_norm": 15.689528465270996,
      "learning_rate": 7.52382768867422e-05,
      "loss": 6.7797,
      "step": 73
    },
    {
      "epoch": 0.019299732672621764,
      "grad_norm": 13.87425422668457,
      "learning_rate": 7.452117519152542e-05,
      "loss": 6.5213,
      "step": 74
    },
    {
      "epoch": 0.019560539870900437,
      "grad_norm": 14.658243179321289,
      "learning_rate": 7.379736965185368e-05,
      "loss": 6.2372,
      "step": 75
    },
    {
      "epoch": 0.01982134706917911,
      "grad_norm": 14.30899429321289,
      "learning_rate": 7.30670581489344e-05,
      "loss": 6.5088,
      "step": 76
    },
    {
      "epoch": 0.020082154267457783,
      "grad_norm": 15.50216293334961,
      "learning_rate": 7.233044034264034e-05,
      "loss": 6.4836,
      "step": 77
    },
    {
      "epoch": 0.020342961465736456,
      "grad_norm": 15.188352584838867,
      "learning_rate": 7.158771761692464e-05,
      "loss": 6.6832,
      "step": 78
    },
    {
      "epoch": 0.020603768664015126,
      "grad_norm": 14.053617477416992,
      "learning_rate": 7.083909302476453e-05,
      "loss": 6.3503,
      "step": 79
    },
    {
      "epoch": 0.0208645758622938,
      "grad_norm": 15.538619995117188,
      "learning_rate": 7.008477123264848e-05,
      "loss": 7.2871,
      "step": 80
    },
    {
      "epoch": 0.02112538306057247,
      "grad_norm": 15.445680618286133,
      "learning_rate": 6.932495846462261e-05,
      "loss": 6.4864,
      "step": 81
    },
    {
      "epoch": 0.021386190258851145,
      "grad_norm": 13.660173416137695,
      "learning_rate": 6.855986244591104e-05,
      "loss": 6.7253,
      "step": 82
    },
    {
      "epoch": 0.021646997457129818,
      "grad_norm": 14.401267051696777,
      "learning_rate": 6.778969234612584e-05,
      "loss": 6.5632,
      "step": 83
    },
    {
      "epoch": 0.02190780465540849,
      "grad_norm": 15.632433891296387,
      "learning_rate": 6.701465872208216e-05,
      "loss": 6.9986,
      "step": 84
    },
    {
      "epoch": 0.02216861185368716,
      "grad_norm": 13.99183177947998,
      "learning_rate": 6.623497346023418e-05,
      "loss": 6.3404,
      "step": 85
    },
    {
      "epoch": 0.022429419051965833,
      "grad_norm": 16.06236457824707,
      "learning_rate": 6.545084971874738e-05,
      "loss": 6.7379,
      "step": 86
    },
    {
      "epoch": 0.022690226250244507,
      "grad_norm": 15.202803611755371,
      "learning_rate": 6.466250186922325e-05,
      "loss": 6.7154,
      "step": 87
    },
    {
      "epoch": 0.02295103344852318,
      "grad_norm": 16.386035919189453,
      "learning_rate": 6.387014543809223e-05,
      "loss": 6.9484,
      "step": 88
    },
    {
      "epoch": 0.023211840646801853,
      "grad_norm": 16.50779151916504,
      "learning_rate": 6.307399704769099e-05,
      "loss": 7.66,
      "step": 89
    },
    {
      "epoch": 0.023472647845080526,
      "grad_norm": 16.221988677978516,
      "learning_rate": 6.227427435703997e-05,
      "loss": 7.1399,
      "step": 90
    },
    {
      "epoch": 0.023733455043359195,
      "grad_norm": 15.444202423095703,
      "learning_rate": 6.147119600233758e-05,
      "loss": 7.2073,
      "step": 91
    },
    {
      "epoch": 0.02399426224163787,
      "grad_norm": 16.156230926513672,
      "learning_rate": 6.066498153718735e-05,
      "loss": 6.4703,
      "step": 92
    },
    {
      "epoch": 0.02425506943991654,
      "grad_norm": 17.191938400268555,
      "learning_rate": 5.985585137257401e-05,
      "loss": 6.9902,
      "step": 93
    },
    {
      "epoch": 0.024515876638195214,
      "grad_norm": 15.29601001739502,
      "learning_rate": 5.90440267166055e-05,
      "loss": 6.6085,
      "step": 94
    },
    {
      "epoch": 0.024776683836473887,
      "grad_norm": 15.107970237731934,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 7.0304,
      "step": 95
    },
    {
      "epoch": 0.02503749103475256,
      "grad_norm": 18.17102813720703,
      "learning_rate": 5.74131823855921e-05,
      "loss": 7.3946,
      "step": 96
    },
    {
      "epoch": 0.02529829823303123,
      "grad_norm": 18.876543045043945,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 7.4822,
      "step": 97
    },
    {
      "epoch": 0.025559105431309903,
      "grad_norm": 17.472850799560547,
      "learning_rate": 5.577423184847932e-05,
      "loss": 7.8264,
      "step": 98
    },
    {
      "epoch": 0.025819912629588576,
      "grad_norm": 17.297557830810547,
      "learning_rate": 5.495227651252315e-05,
      "loss": 7.2268,
      "step": 99
    },
    {
      "epoch": 0.02608071982786725,
      "grad_norm": 24.11049461364746,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 8.5908,
      "step": 100
    },
    {
      "epoch": 0.02608071982786725,
      "eval_loss": 2.0097198486328125,
      "eval_runtime": 468.9732,
      "eval_samples_per_second": 13.771,
      "eval_steps_per_second": 3.444,
      "step": 100
    },
    {
      "epoch": 0.026341527026145922,
      "grad_norm": 22.687538146972656,
      "learning_rate": 5.330452921628497e-05,
      "loss": 9.6004,
      "step": 101
    },
    {
      "epoch": 0.026602334224424595,
      "grad_norm": 19.59461212158203,
      "learning_rate": 5.247918773366112e-05,
      "loss": 8.6654,
      "step": 102
    },
    {
      "epoch": 0.02686314142270327,
      "grad_norm": 13.891752243041992,
      "learning_rate": 5.165316846586541e-05,
      "loss": 8.2593,
      "step": 103
    },
    {
      "epoch": 0.027123948620981938,
      "grad_norm": 10.809483528137207,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 7.4113,
      "step": 104
    },
    {
      "epoch": 0.02738475581926061,
      "grad_norm": 10.051604270935059,
      "learning_rate": 5e-05,
      "loss": 7.5772,
      "step": 105
    },
    {
      "epoch": 0.027645563017539284,
      "grad_norm": 10.854872703552246,
      "learning_rate": 4.917330276168208e-05,
      "loss": 7.7146,
      "step": 106
    },
    {
      "epoch": 0.027906370215817957,
      "grad_norm": 10.533751487731934,
      "learning_rate": 4.834683153413459e-05,
      "loss": 6.9586,
      "step": 107
    },
    {
      "epoch": 0.02816717741409663,
      "grad_norm": 11.13151741027832,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 7.2954,
      "step": 108
    },
    {
      "epoch": 0.028427984612375303,
      "grad_norm": 11.318047523498535,
      "learning_rate": 4.669547078371504e-05,
      "loss": 7.2726,
      "step": 109
    },
    {
      "epoch": 0.028688791810653973,
      "grad_norm": 11.068788528442383,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 6.8083,
      "step": 110
    },
    {
      "epoch": 0.028949599008932646,
      "grad_norm": 10.888313293457031,
      "learning_rate": 4.504772348747687e-05,
      "loss": 6.7556,
      "step": 111
    },
    {
      "epoch": 0.02921040620721132,
      "grad_norm": 10.240464210510254,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 6.4259,
      "step": 112
    },
    {
      "epoch": 0.029471213405489992,
      "grad_norm": 10.37086296081543,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 6.1604,
      "step": 113
    },
    {
      "epoch": 0.029732020603768665,
      "grad_norm": 11.893575668334961,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 6.8082,
      "step": 114
    },
    {
      "epoch": 0.029992827802047338,
      "grad_norm": 12.003168106079102,
      "learning_rate": 4.17702704859633e-05,
      "loss": 6.3759,
      "step": 115
    },
    {
      "epoch": 0.030253635000326008,
      "grad_norm": 12.625334739685059,
      "learning_rate": 4.095597328339452e-05,
      "loss": 6.5815,
      "step": 116
    },
    {
      "epoch": 0.03051444219860468,
      "grad_norm": 12.851235389709473,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 6.7071,
      "step": 117
    },
    {
      "epoch": 0.030775249396883354,
      "grad_norm": 12.898778915405273,
      "learning_rate": 3.933501846281267e-05,
      "loss": 6.8533,
      "step": 118
    },
    {
      "epoch": 0.031036056595162027,
      "grad_norm": 14.243744850158691,
      "learning_rate": 3.852880399766243e-05,
      "loss": 7.0537,
      "step": 119
    },
    {
      "epoch": 0.031296863793440696,
      "grad_norm": 11.919962882995605,
      "learning_rate": 3.772572564296005e-05,
      "loss": 6.4851,
      "step": 120
    },
    {
      "epoch": 0.03155767099171937,
      "grad_norm": 12.9480619430542,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 6.4566,
      "step": 121
    },
    {
      "epoch": 0.03181847818999804,
      "grad_norm": 12.840946197509766,
      "learning_rate": 3.612985456190778e-05,
      "loss": 6.2002,
      "step": 122
    },
    {
      "epoch": 0.032079285388276715,
      "grad_norm": 12.152085304260254,
      "learning_rate": 3.533749813077677e-05,
      "loss": 6.1678,
      "step": 123
    },
    {
      "epoch": 0.03234009258655539,
      "grad_norm": 13.635071754455566,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 6.5239,
      "step": 124
    },
    {
      "epoch": 0.03260089978483406,
      "grad_norm": 12.257086753845215,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 6.0963,
      "step": 125
    },
    {
      "epoch": 0.032861706983112735,
      "grad_norm": 13.802964210510254,
      "learning_rate": 3.298534127791785e-05,
      "loss": 6.4496,
      "step": 126
    },
    {
      "epoch": 0.03312251418139141,
      "grad_norm": 13.266039848327637,
      "learning_rate": 3.221030765387417e-05,
      "loss": 6.4984,
      "step": 127
    },
    {
      "epoch": 0.03338332137967008,
      "grad_norm": 11.946686744689941,
      "learning_rate": 3.144013755408895e-05,
      "loss": 6.2973,
      "step": 128
    },
    {
      "epoch": 0.033644128577948754,
      "grad_norm": 14.231830596923828,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 6.9088,
      "step": 129
    },
    {
      "epoch": 0.03390493577622743,
      "grad_norm": 12.960298538208008,
      "learning_rate": 2.991522876735154e-05,
      "loss": 6.6555,
      "step": 130
    },
    {
      "epoch": 0.0341657429745061,
      "grad_norm": 14.096085548400879,
      "learning_rate": 2.916090697523549e-05,
      "loss": 6.8805,
      "step": 131
    },
    {
      "epoch": 0.034426550172784766,
      "grad_norm": 14.523974418640137,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 6.7666,
      "step": 132
    },
    {
      "epoch": 0.03468735737106344,
      "grad_norm": 14.790218353271484,
      "learning_rate": 2.766955965735968e-05,
      "loss": 6.7825,
      "step": 133
    },
    {
      "epoch": 0.03494816456934211,
      "grad_norm": 13.652941703796387,
      "learning_rate": 2.693294185106562e-05,
      "loss": 6.6213,
      "step": 134
    },
    {
      "epoch": 0.035208971767620785,
      "grad_norm": 15.05577564239502,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 6.7036,
      "step": 135
    },
    {
      "epoch": 0.03546977896589946,
      "grad_norm": 13.610404968261719,
      "learning_rate": 2.547882480847461e-05,
      "loss": 6.4015,
      "step": 136
    },
    {
      "epoch": 0.03573058616417813,
      "grad_norm": 14.335271835327148,
      "learning_rate": 2.476172311325783e-05,
      "loss": 6.9272,
      "step": 137
    },
    {
      "epoch": 0.035991393362456804,
      "grad_norm": 16.21661376953125,
      "learning_rate": 2.405152131093926e-05,
      "loss": 6.8389,
      "step": 138
    },
    {
      "epoch": 0.03625220056073548,
      "grad_norm": 15.185527801513672,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 7.0456,
      "step": 139
    },
    {
      "epoch": 0.03651300775901415,
      "grad_norm": 14.693958282470703,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 6.7648,
      "step": 140
    },
    {
      "epoch": 0.03677381495729282,
      "grad_norm": 15.471109390258789,
      "learning_rate": 2.196424713241637e-05,
      "loss": 7.0808,
      "step": 141
    },
    {
      "epoch": 0.037034622155571496,
      "grad_norm": 14.067039489746094,
      "learning_rate": 2.128356686585282e-05,
      "loss": 6.6511,
      "step": 142
    },
    {
      "epoch": 0.03729542935385017,
      "grad_norm": 17.071073532104492,
      "learning_rate": 2.061073738537635e-05,
      "loss": 7.1006,
      "step": 143
    },
    {
      "epoch": 0.037556236552128835,
      "grad_norm": 15.53573989868164,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 7.0855,
      "step": 144
    },
    {
      "epoch": 0.03781704375040751,
      "grad_norm": 15.090848922729492,
      "learning_rate": 1.928936436551661e-05,
      "loss": 6.6323,
      "step": 145
    },
    {
      "epoch": 0.03807785094868618,
      "grad_norm": 16.284835815429688,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 7.2749,
      "step": 146
    },
    {
      "epoch": 0.038338658146964855,
      "grad_norm": 16.655628204345703,
      "learning_rate": 1.800157297483417e-05,
      "loss": 6.8157,
      "step": 147
    },
    {
      "epoch": 0.03859946534524353,
      "grad_norm": 15.952491760253906,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 6.8284,
      "step": 148
    },
    {
      "epoch": 0.0388602725435222,
      "grad_norm": 15.237507820129395,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 6.9314,
      "step": 149
    },
    {
      "epoch": 0.039121079741800874,
      "grad_norm": 21.894411087036133,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 7.7673,
      "step": 150
    },
    {
      "epoch": 0.039121079741800874,
      "eval_loss": 1.7088080644607544,
      "eval_runtime": 469.0356,
      "eval_samples_per_second": 13.769,
      "eval_steps_per_second": 3.443,
      "step": 150
    },
    {
      "epoch": 0.03938188694007955,
      "grad_norm": 8.972198486328125,
      "learning_rate": 1.553232954407171e-05,
      "loss": 8.0579,
      "step": 151
    },
    {
      "epoch": 0.03964269413835822,
      "grad_norm": 11.423454284667969,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 7.5822,
      "step": 152
    },
    {
      "epoch": 0.03990350133663689,
      "grad_norm": 9.619215965270996,
      "learning_rate": 1.435357758543015e-05,
      "loss": 7.148,
      "step": 153
    },
    {
      "epoch": 0.040164308534915566,
      "grad_norm": 9.98920726776123,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 7.303,
      "step": 154
    },
    {
      "epoch": 0.04042511573319424,
      "grad_norm": 10.115551948547363,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 7.073,
      "step": 155
    },
    {
      "epoch": 0.04068592293147291,
      "grad_norm": 10.321307182312012,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 7.4784,
      "step": 156
    },
    {
      "epoch": 0.04094673012975158,
      "grad_norm": 9.441727638244629,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 6.7862,
      "step": 157
    },
    {
      "epoch": 0.04120753732803025,
      "grad_norm": 10.315311431884766,
      "learning_rate": 1.157994445715706e-05,
      "loss": 6.7423,
      "step": 158
    },
    {
      "epoch": 0.041468344526308924,
      "grad_norm": 9.382549285888672,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 6.6966,
      "step": 159
    },
    {
      "epoch": 0.0417291517245876,
      "grad_norm": 11.195586204528809,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 6.725,
      "step": 160
    },
    {
      "epoch": 0.04198995892286627,
      "grad_norm": 9.782853126525879,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 6.5643,
      "step": 161
    },
    {
      "epoch": 0.04225076612114494,
      "grad_norm": 10.835838317871094,
      "learning_rate": 9.549150281252633e-06,
      "loss": 6.6496,
      "step": 162
    },
    {
      "epoch": 0.042511573319423616,
      "grad_norm": 9.901912689208984,
      "learning_rate": 9.068759265665384e-06,
      "loss": 5.9043,
      "step": 163
    },
    {
      "epoch": 0.04277238051770229,
      "grad_norm": 11.73845386505127,
      "learning_rate": 8.599558442598998e-06,
      "loss": 6.6078,
      "step": 164
    },
    {
      "epoch": 0.04303318771598096,
      "grad_norm": 9.563654899597168,
      "learning_rate": 8.141676086873572e-06,
      "loss": 6.155,
      "step": 165
    },
    {
      "epoch": 0.043293994914259636,
      "grad_norm": 10.07148551940918,
      "learning_rate": 7.695237378953223e-06,
      "loss": 6.0657,
      "step": 166
    },
    {
      "epoch": 0.04355480211253831,
      "grad_norm": 12.328645706176758,
      "learning_rate": 7.260364370723044e-06,
      "loss": 6.7078,
      "step": 167
    },
    {
      "epoch": 0.04381560931081698,
      "grad_norm": 10.790811538696289,
      "learning_rate": 6.837175952121306e-06,
      "loss": 6.3025,
      "step": 168
    },
    {
      "epoch": 0.04407641650909565,
      "grad_norm": 10.387198448181152,
      "learning_rate": 6.425787818636131e-06,
      "loss": 6.1726,
      "step": 169
    },
    {
      "epoch": 0.04433722370737432,
      "grad_norm": 10.43610954284668,
      "learning_rate": 6.026312439675552e-06,
      "loss": 6.0739,
      "step": 170
    },
    {
      "epoch": 0.044598030905652994,
      "grad_norm": 10.567517280578613,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 6.0931,
      "step": 171
    },
    {
      "epoch": 0.04485883810393167,
      "grad_norm": 12.047853469848633,
      "learning_rate": 5.263533508961827e-06,
      "loss": 6.7014,
      "step": 172
    },
    {
      "epoch": 0.04511964530221034,
      "grad_norm": 11.32982063293457,
      "learning_rate": 4.900438493352055e-06,
      "loss": 6.6474,
      "step": 173
    },
    {
      "epoch": 0.04538045250048901,
      "grad_norm": 10.81766128540039,
      "learning_rate": 4.549673247541875e-06,
      "loss": 6.4521,
      "step": 174
    },
    {
      "epoch": 0.045641259698767686,
      "grad_norm": 11.827414512634277,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 6.3422,
      "step": 175
    },
    {
      "epoch": 0.04590206689704636,
      "grad_norm": 14.590598106384277,
      "learning_rate": 3.885512251130763e-06,
      "loss": 6.3251,
      "step": 176
    },
    {
      "epoch": 0.04616287409532503,
      "grad_norm": 11.190947532653809,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 6.8135,
      "step": 177
    },
    {
      "epoch": 0.046423681293603705,
      "grad_norm": 12.350541114807129,
      "learning_rate": 3.271776770026963e-06,
      "loss": 6.5224,
      "step": 178
    },
    {
      "epoch": 0.04668448849188238,
      "grad_norm": 12.205690383911133,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 6.1394,
      "step": 179
    },
    {
      "epoch": 0.04694529569016105,
      "grad_norm": 11.804443359375,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 6.5934,
      "step": 180
    },
    {
      "epoch": 0.047206102888439724,
      "grad_norm": 13.251696586608887,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 6.3572,
      "step": 181
    },
    {
      "epoch": 0.04746691008671839,
      "grad_norm": 12.611340522766113,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 6.8046,
      "step": 182
    },
    {
      "epoch": 0.047727717284997064,
      "grad_norm": 12.854605674743652,
      "learning_rate": 1.962316193157593e-06,
      "loss": 6.1075,
      "step": 183
    },
    {
      "epoch": 0.04798852448327574,
      "grad_norm": 13.197980880737305,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 6.5498,
      "step": 184
    },
    {
      "epoch": 0.04824933168155441,
      "grad_norm": 11.321105003356934,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 6.173,
      "step": 185
    },
    {
      "epoch": 0.04851013887983308,
      "grad_norm": 13.08361530303955,
      "learning_rate": 1.333670137599713e-06,
      "loss": 6.8889,
      "step": 186
    },
    {
      "epoch": 0.048770946078111756,
      "grad_norm": 12.51752758026123,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 6.2587,
      "step": 187
    },
    {
      "epoch": 0.04903175327639043,
      "grad_norm": 14.71296215057373,
      "learning_rate": 9.810017062595322e-07,
      "loss": 6.7344,
      "step": 188
    },
    {
      "epoch": 0.0492925604746691,
      "grad_norm": 12.863415718078613,
      "learning_rate": 8.247462563808817e-07,
      "loss": 6.1479,
      "step": 189
    },
    {
      "epoch": 0.049553367672947775,
      "grad_norm": 13.14647388458252,
      "learning_rate": 6.819348298638839e-07,
      "loss": 6.7513,
      "step": 190
    },
    {
      "epoch": 0.04981417487122645,
      "grad_norm": 12.831649780273438,
      "learning_rate": 5.526064699265753e-07,
      "loss": 6.7909,
      "step": 191
    },
    {
      "epoch": 0.05007498206950512,
      "grad_norm": 15.013938903808594,
      "learning_rate": 4.367965336512403e-07,
      "loss": 7.2058,
      "step": 192
    },
    {
      "epoch": 0.050335789267783794,
      "grad_norm": 13.65225601196289,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 6.5087,
      "step": 193
    },
    {
      "epoch": 0.05059659646606246,
      "grad_norm": 13.648240089416504,
      "learning_rate": 2.458548727494292e-07,
      "loss": 7.2177,
      "step": 194
    },
    {
      "epoch": 0.05085740366434113,
      "grad_norm": 14.354628562927246,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 6.6091,
      "step": 195
    },
    {
      "epoch": 0.051118210862619806,
      "grad_norm": 15.314456939697266,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 7.0205,
      "step": 196
    },
    {
      "epoch": 0.05137901806089848,
      "grad_norm": 16.61842155456543,
      "learning_rate": 6.150154258476315e-08,
      "loss": 7.273,
      "step": 197
    },
    {
      "epoch": 0.05163982525917715,
      "grad_norm": 17.48015594482422,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 7.5824,
      "step": 198
    },
    {
      "epoch": 0.051900632457455825,
      "grad_norm": 16.759817123413086,
      "learning_rate": 6.834750376549792e-09,
      "loss": 7.3982,
      "step": 199
    },
    {
      "epoch": 0.0521614396557345,
      "grad_norm": 21.159927368164062,
      "learning_rate": 0.0,
      "loss": 8.399,
      "step": 200
    },
    {
      "epoch": 0.0521614396557345,
      "eval_loss": 1.664876103401184,
      "eval_runtime": 469.081,
      "eval_samples_per_second": 13.767,
      "eval_steps_per_second": 3.443,
      "step": 200
    }
  ],
  "logging_steps": 1,
  "max_steps": 200,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 50,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 5,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 0
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 2.861996017975296e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}