{
  "best_metric": 1.973547101020813,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.05461123626186088,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0002730561813093044,
      "grad_norm": 43.446266174316406,
      "learning_rate": 1e-05,
      "loss": 10.0397,
      "step": 1
    },
    {
      "epoch": 0.0002730561813093044,
      "eval_loss": 4.238278865814209,
      "eval_runtime": 693.1619,
      "eval_samples_per_second": 8.898,
      "eval_steps_per_second": 2.225,
      "step": 1
    },
    {
      "epoch": 0.0005461123626186088,
      "grad_norm": 44.35715866088867,
      "learning_rate": 2e-05,
      "loss": 11.7295,
      "step": 2
    },
    {
      "epoch": 0.0008191685439279131,
      "grad_norm": 49.75259017944336,
      "learning_rate": 3e-05,
      "loss": 11.851,
      "step": 3
    },
    {
      "epoch": 0.0010922247252372176,
      "grad_norm": 46.56555938720703,
      "learning_rate": 4e-05,
      "loss": 11.3896,
      "step": 4
    },
    {
      "epoch": 0.0013652809065465218,
      "grad_norm": 40.56822967529297,
      "learning_rate": 5e-05,
      "loss": 10.777,
      "step": 5
    },
    {
      "epoch": 0.0016383370878558263,
      "grad_norm": 29.243772506713867,
      "learning_rate": 6e-05,
      "loss": 8.8778,
      "step": 6
    },
    {
      "epoch": 0.0019113932691651307,
      "grad_norm": 27.35118293762207,
      "learning_rate": 7e-05,
      "loss": 8.5317,
      "step": 7
    },
    {
      "epoch": 0.0021844494504744353,
      "grad_norm": 28.01230239868164,
      "learning_rate": 8e-05,
      "loss": 8.6976,
      "step": 8
    },
    {
      "epoch": 0.0024575056317837395,
      "grad_norm": 24.44605255126953,
      "learning_rate": 9e-05,
      "loss": 9.1925,
      "step": 9
    },
    {
      "epoch": 0.0027305618130930437,
      "grad_norm": 25.969579696655273,
      "learning_rate": 0.0001,
      "loss": 8.022,
      "step": 10
    },
    {
      "epoch": 0.0030036179944023483,
      "grad_norm": 27.08025550842285,
      "learning_rate": 9.999316524962345e-05,
      "loss": 9.1533,
      "step": 11
    },
    {
      "epoch": 0.0032766741757116525,
      "grad_norm": 18.790332794189453,
      "learning_rate": 9.997266286704631e-05,
      "loss": 7.5891,
      "step": 12
    },
    {
      "epoch": 0.003549730357020957,
      "grad_norm": 23.858049392700195,
      "learning_rate": 9.993849845741524e-05,
      "loss": 9.0651,
      "step": 13
    },
    {
      "epoch": 0.0038227865383302613,
      "grad_norm": 20.80962562561035,
      "learning_rate": 9.989068136093873e-05,
      "loss": 7.7469,
      "step": 14
    },
    {
      "epoch": 0.0040958427196395655,
      "grad_norm": 21.938926696777344,
      "learning_rate": 9.98292246503335e-05,
      "loss": 8.5726,
      "step": 15
    },
    {
      "epoch": 0.004368898900948871,
      "grad_norm": 19.670955657958984,
      "learning_rate": 9.975414512725057e-05,
      "loss": 7.9194,
      "step": 16
    },
    {
      "epoch": 0.004641955082258175,
      "grad_norm": 21.373003005981445,
      "learning_rate": 9.966546331768191e-05,
      "loss": 8.5548,
      "step": 17
    },
    {
      "epoch": 0.004915011263567479,
      "grad_norm": 21.677001953125,
      "learning_rate": 9.956320346634876e-05,
      "loss": 7.7842,
      "step": 18
    },
    {
      "epoch": 0.005188067444876783,
      "grad_norm": 21.132835388183594,
      "learning_rate": 9.944739353007344e-05,
      "loss": 9.1783,
      "step": 19
    },
    {
      "epoch": 0.005461123626186087,
      "grad_norm": 19.698659896850586,
      "learning_rate": 9.931806517013612e-05,
      "loss": 7.6789,
      "step": 20
    },
    {
      "epoch": 0.0057341798074953924,
      "grad_norm": 26.263071060180664,
      "learning_rate": 9.917525374361912e-05,
      "loss": 8.8006,
      "step": 21
    },
    {
      "epoch": 0.006007235988804697,
      "grad_norm": 19.799365997314453,
      "learning_rate": 9.901899829374047e-05,
      "loss": 8.5378,
      "step": 22
    },
    {
      "epoch": 0.006280292170114001,
      "grad_norm": 25.806068420410156,
      "learning_rate": 9.884934153917997e-05,
      "loss": 8.5017,
      "step": 23
    },
    {
      "epoch": 0.006553348351423305,
      "grad_norm": 24.34381866455078,
      "learning_rate": 9.86663298624003e-05,
      "loss": 8.9881,
      "step": 24
    },
    {
      "epoch": 0.00682640453273261,
      "grad_norm": 18.999231338500977,
      "learning_rate": 9.847001329696653e-05,
      "loss": 7.9945,
      "step": 25
    },
    {
      "epoch": 0.007099460714041914,
      "grad_norm": 20.11553382873535,
      "learning_rate": 9.826044551386744e-05,
      "loss": 8.4672,
      "step": 26
    },
    {
      "epoch": 0.0073725168953512185,
      "grad_norm": 21.312238693237305,
      "learning_rate": 9.803768380684242e-05,
      "loss": 8.6953,
      "step": 27
    },
    {
      "epoch": 0.007645573076660523,
      "grad_norm": 31.490800857543945,
      "learning_rate": 9.780178907671789e-05,
      "loss": 9.7065,
      "step": 28
    },
    {
      "epoch": 0.007918629257969828,
      "grad_norm": 20.707792282104492,
      "learning_rate": 9.755282581475769e-05,
      "loss": 8.5876,
      "step": 29
    },
    {
      "epoch": 0.008191685439279131,
      "grad_norm": 19.695018768310547,
      "learning_rate": 9.729086208503174e-05,
      "loss": 8.0395,
      "step": 30
    },
    {
      "epoch": 0.008464741620588436,
      "grad_norm": 21.563302993774414,
      "learning_rate": 9.701596950580806e-05,
      "loss": 7.869,
      "step": 31
    },
    {
      "epoch": 0.008737797801897741,
      "grad_norm": 25.480226516723633,
      "learning_rate": 9.672822322997305e-05,
      "loss": 8.0559,
      "step": 32
    },
    {
      "epoch": 0.009010853983207045,
      "grad_norm": 23.27593231201172,
      "learning_rate": 9.642770192448536e-05,
      "loss": 8.9705,
      "step": 33
    },
    {
      "epoch": 0.00928391016451635,
      "grad_norm": 22.600749969482422,
      "learning_rate": 9.611448774886924e-05,
      "loss": 8.3754,
      "step": 34
    },
    {
      "epoch": 0.009556966345825653,
      "grad_norm": 24.033321380615234,
      "learning_rate": 9.578866633275288e-05,
      "loss": 9.4395,
      "step": 35
    },
    {
      "epoch": 0.009830022527134958,
      "grad_norm": 25.064741134643555,
      "learning_rate": 9.545032675245813e-05,
      "loss": 8.8034,
      "step": 36
    },
    {
      "epoch": 0.010103078708444263,
      "grad_norm": 28.80820083618164,
      "learning_rate": 9.509956150664796e-05,
      "loss": 8.2912,
      "step": 37
    },
    {
      "epoch": 0.010376134889753566,
      "grad_norm": 29.365997314453125,
      "learning_rate": 9.473646649103818e-05,
      "loss": 7.6412,
      "step": 38
    },
    {
      "epoch": 0.010649191071062871,
      "grad_norm": 22.005088806152344,
      "learning_rate": 9.43611409721806e-05,
      "loss": 8.3727,
      "step": 39
    },
    {
      "epoch": 0.010922247252372175,
      "grad_norm": 22.131328582763672,
      "learning_rate": 9.397368756032445e-05,
      "loss": 7.4845,
      "step": 40
    },
    {
      "epoch": 0.01119530343368148,
      "grad_norm": 26.777101516723633,
      "learning_rate": 9.357421218136386e-05,
      "loss": 8.6799,
      "step": 41
    },
    {
      "epoch": 0.011468359614990785,
      "grad_norm": 27.420764923095703,
      "learning_rate": 9.316282404787871e-05,
      "loss": 8.6992,
      "step": 42
    },
    {
      "epoch": 0.011741415796300088,
      "grad_norm": 27.803956985473633,
      "learning_rate": 9.273963562927695e-05,
      "loss": 9.0848,
      "step": 43
    },
    {
      "epoch": 0.012014471977609393,
      "grad_norm": 32.42679214477539,
      "learning_rate": 9.230476262104677e-05,
      "loss": 10.5448,
      "step": 44
    },
    {
      "epoch": 0.012287528158918698,
      "grad_norm": 28.605180740356445,
      "learning_rate": 9.185832391312644e-05,
      "loss": 8.5635,
      "step": 45
    },
    {
      "epoch": 0.012560584340228002,
      "grad_norm": 28.457530975341797,
      "learning_rate": 9.140044155740101e-05,
      "loss": 9.0221,
      "step": 46
    },
    {
      "epoch": 0.012833640521537307,
      "grad_norm": 24.8094425201416,
      "learning_rate": 9.093124073433463e-05,
      "loss": 7.0862,
      "step": 47
    },
    {
      "epoch": 0.01310669670284661,
      "grad_norm": 30.082542419433594,
      "learning_rate": 9.045084971874738e-05,
      "loss": 8.431,
      "step": 48
    },
    {
      "epoch": 0.013379752884155915,
      "grad_norm": 26.035165786743164,
      "learning_rate": 8.995939984474624e-05,
      "loss": 8.7725,
      "step": 49
    },
    {
      "epoch": 0.01365280906546522,
      "grad_norm": 48.76018142700195,
      "learning_rate": 8.945702546981969e-05,
      "loss": 10.7992,
      "step": 50
    },
    {
      "epoch": 0.01365280906546522,
      "eval_loss": 2.271299362182617,
      "eval_runtime": 698.4085,
      "eval_samples_per_second": 8.832,
      "eval_steps_per_second": 2.208,
      "step": 50
    },
    {
      "epoch": 0.013925865246774524,
      "grad_norm": 24.909616470336914,
      "learning_rate": 8.894386393810563e-05,
      "loss": 7.0786,
      "step": 51
    },
    {
      "epoch": 0.014198921428083829,
      "grad_norm": 24.512235641479492,
      "learning_rate": 8.842005554284296e-05,
      "loss": 8.499,
      "step": 52
    },
    {
      "epoch": 0.014471977609393132,
      "grad_norm": 20.238571166992188,
      "learning_rate": 8.788574348801675e-05,
      "loss": 8.7019,
      "step": 53
    },
    {
      "epoch": 0.014745033790702437,
      "grad_norm": 20.472135543823242,
      "learning_rate": 8.73410738492077e-05,
      "loss": 7.191,
      "step": 54
    },
    {
      "epoch": 0.015018089972011742,
      "grad_norm": 18.48778533935547,
      "learning_rate": 8.678619553365659e-05,
      "loss": 8.5567,
      "step": 55
    },
    {
      "epoch": 0.015291146153321045,
      "grad_norm": 17.57303810119629,
      "learning_rate": 8.622126023955446e-05,
      "loss": 8.4159,
      "step": 56
    },
    {
      "epoch": 0.01556420233463035,
      "grad_norm": 17.54515266418457,
      "learning_rate": 8.564642241456986e-05,
      "loss": 8.3224,
      "step": 57
    },
    {
      "epoch": 0.015837258515939655,
      "grad_norm": 18.000179290771484,
      "learning_rate": 8.506183921362443e-05,
      "loss": 9.0583,
      "step": 58
    },
    {
      "epoch": 0.01611031469724896,
      "grad_norm": 22.772430419921875,
      "learning_rate": 8.44676704559283e-05,
      "loss": 8.5008,
      "step": 59
    },
    {
      "epoch": 0.016383370878558262,
      "grad_norm": 16.104167938232422,
      "learning_rate": 8.386407858128706e-05,
      "loss": 7.918,
      "step": 60
    },
    {
      "epoch": 0.016656427059867567,
      "grad_norm": 15.550981521606445,
      "learning_rate": 8.32512286056924e-05,
      "loss": 7.9553,
      "step": 61
    },
    {
      "epoch": 0.016929483241176872,
      "grad_norm": 17.316165924072266,
      "learning_rate": 8.262928807620843e-05,
      "loss": 7.9485,
      "step": 62
    },
    {
      "epoch": 0.017202539422486177,
      "grad_norm": 17.424301147460938,
      "learning_rate": 8.199842702516583e-05,
      "loss": 7.3209,
      "step": 63
    },
    {
      "epoch": 0.017475595603795482,
      "grad_norm": 16.4025936126709,
      "learning_rate": 8.135881792367686e-05,
      "loss": 8.1016,
      "step": 64
    },
    {
      "epoch": 0.017748651785104784,
      "grad_norm": 18.199373245239258,
      "learning_rate": 8.07106356344834e-05,
      "loss": 8.2282,
      "step": 65
    },
    {
      "epoch": 0.01802170796641409,
      "grad_norm": 18.115741729736328,
      "learning_rate": 8.005405736415126e-05,
      "loss": 8.0613,
      "step": 66
    },
    {
      "epoch": 0.018294764147723394,
      "grad_norm": 18.224075317382812,
      "learning_rate": 7.938926261462366e-05,
      "loss": 8.5954,
      "step": 67
    },
    {
      "epoch": 0.0185678203290327,
      "grad_norm": 17.102075576782227,
      "learning_rate": 7.871643313414718e-05,
      "loss": 8.2442,
      "step": 68
    },
    {
      "epoch": 0.018840876510342004,
      "grad_norm": 23.462993621826172,
      "learning_rate": 7.803575286758364e-05,
      "loss": 9.5302,
      "step": 69
    },
    {
      "epoch": 0.019113932691651306,
      "grad_norm": 19.042011260986328,
      "learning_rate": 7.734740790612136e-05,
      "loss": 8.4879,
      "step": 70
    },
    {
      "epoch": 0.01938698887296061,
      "grad_norm": 16.553205490112305,
      "learning_rate": 7.66515864363997e-05,
      "loss": 6.8034,
      "step": 71
    },
    {
      "epoch": 0.019660045054269916,
      "grad_norm": 21.598339080810547,
      "learning_rate": 7.594847868906076e-05,
      "loss": 7.9676,
      "step": 72
    },
    {
      "epoch": 0.01993310123557922,
      "grad_norm": 18.137998580932617,
      "learning_rate": 7.52382768867422e-05,
      "loss": 7.6405,
      "step": 73
    },
    {
      "epoch": 0.020206157416888526,
      "grad_norm": 18.159276962280273,
      "learning_rate": 7.452117519152542e-05,
      "loss": 8.6501,
      "step": 74
    },
    {
      "epoch": 0.020479213598197828,
      "grad_norm": 15.387510299682617,
      "learning_rate": 7.379736965185368e-05,
      "loss": 7.4218,
      "step": 75
    },
    {
      "epoch": 0.020752269779507133,
      "grad_norm": 23.088130950927734,
      "learning_rate": 7.30670581489344e-05,
      "loss": 8.0995,
      "step": 76
    },
    {
      "epoch": 0.021025325960816438,
      "grad_norm": 17.979206085205078,
      "learning_rate": 7.233044034264034e-05,
      "loss": 8.097,
      "step": 77
    },
    {
      "epoch": 0.021298382142125743,
      "grad_norm": 20.9732723236084,
      "learning_rate": 7.158771761692464e-05,
      "loss": 9.4713,
      "step": 78
    },
    {
      "epoch": 0.021571438323435048,
      "grad_norm": 20.32724952697754,
      "learning_rate": 7.083909302476453e-05,
      "loss": 7.8127,
      "step": 79
    },
    {
      "epoch": 0.02184449450474435,
      "grad_norm": 20.800214767456055,
      "learning_rate": 7.008477123264848e-05,
      "loss": 8.0962,
      "step": 80
    },
    {
      "epoch": 0.022117550686053655,
      "grad_norm": 19.788745880126953,
      "learning_rate": 6.932495846462261e-05,
      "loss": 9.1246,
      "step": 81
    },
    {
      "epoch": 0.02239060686736296,
      "grad_norm": 20.843629837036133,
      "learning_rate": 6.855986244591104e-05,
      "loss": 8.05,
      "step": 82
    },
    {
      "epoch": 0.022663663048672265,
      "grad_norm": 16.206398010253906,
      "learning_rate": 6.778969234612584e-05,
      "loss": 7.2646,
      "step": 83
    },
    {
      "epoch": 0.02293671922998157,
      "grad_norm": 18.066072463989258,
      "learning_rate": 6.701465872208216e-05,
      "loss": 7.4305,
      "step": 84
    },
    {
      "epoch": 0.023209775411290875,
      "grad_norm": 20.16105079650879,
      "learning_rate": 6.623497346023418e-05,
      "loss": 8.6175,
      "step": 85
    },
    {
      "epoch": 0.023482831592600176,
      "grad_norm": 23.420856475830078,
      "learning_rate": 6.545084971874738e-05,
      "loss": 8.8125,
      "step": 86
    },
    {
      "epoch": 0.02375588777390948,
      "grad_norm": 23.20645523071289,
      "learning_rate": 6.466250186922325e-05,
      "loss": 7.9372,
      "step": 87
    },
    {
      "epoch": 0.024028943955218787,
      "grad_norm": 18.95828628540039,
      "learning_rate": 6.387014543809223e-05,
      "loss": 8.2753,
      "step": 88
    },
    {
      "epoch": 0.02430200013652809,
      "grad_norm": 20.9344482421875,
      "learning_rate": 6.307399704769099e-05,
      "loss": 8.9119,
      "step": 89
    },
    {
      "epoch": 0.024575056317837397,
      "grad_norm": 22.957317352294922,
      "learning_rate": 6.227427435703997e-05,
      "loss": 8.6468,
      "step": 90
    },
    {
      "epoch": 0.024848112499146698,
      "grad_norm": 20.31450653076172,
      "learning_rate": 6.147119600233758e-05,
      "loss": 7.4481,
      "step": 91
    },
    {
      "epoch": 0.025121168680456003,
      "grad_norm": 25.291793823242188,
      "learning_rate": 6.066498153718735e-05,
      "loss": 7.6528,
      "step": 92
    },
    {
      "epoch": 0.02539422486176531,
      "grad_norm": 23.33091926574707,
      "learning_rate": 5.985585137257401e-05,
      "loss": 8.0942,
      "step": 93
    },
    {
      "epoch": 0.025667281043074613,
      "grad_norm": 26.418405532836914,
      "learning_rate": 5.90440267166055e-05,
      "loss": 7.4102,
      "step": 94
    },
    {
      "epoch": 0.02594033722438392,
      "grad_norm": 25.400789260864258,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 8.5998,
      "step": 95
    },
    {
      "epoch": 0.02621339340569322,
      "grad_norm": 32.99055480957031,
      "learning_rate": 5.74131823855921e-05,
      "loss": 7.8845,
      "step": 96
    },
    {
      "epoch": 0.026486449587002525,
      "grad_norm": 25.209062576293945,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 6.6911,
      "step": 97
    },
    {
      "epoch": 0.02675950576831183,
      "grad_norm": 26.858041763305664,
      "learning_rate": 5.577423184847932e-05,
      "loss": 7.1683,
      "step": 98
    },
    {
      "epoch": 0.027032561949621135,
      "grad_norm": 28.698999404907227,
      "learning_rate": 5.495227651252315e-05,
      "loss": 7.3205,
      "step": 99
    },
    {
      "epoch": 0.02730561813093044,
      "grad_norm": 51.56425094604492,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 11.4165,
      "step": 100
    },
    {
      "epoch": 0.02730561813093044,
      "eval_loss": 2.1340065002441406,
      "eval_runtime": 697.8133,
      "eval_samples_per_second": 8.839,
      "eval_steps_per_second": 2.21,
      "step": 100
    },
    {
      "epoch": 0.027578674312239742,
      "grad_norm": 18.07147789001465,
      "learning_rate": 5.330452921628497e-05,
      "loss": 6.5228,
      "step": 101
    },
    {
      "epoch": 0.027851730493549047,
      "grad_norm": 20.77764892578125,
      "learning_rate": 5.247918773366112e-05,
      "loss": 7.8123,
      "step": 102
    },
    {
      "epoch": 0.028124786674858352,
      "grad_norm": 20.727760314941406,
      "learning_rate": 5.165316846586541e-05,
      "loss": 8.3957,
      "step": 103
    },
    {
      "epoch": 0.028397842856167657,
      "grad_norm": 16.201143264770508,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 7.7034,
      "step": 104
    },
    {
      "epoch": 0.028670899037476962,
      "grad_norm": 14.143803596496582,
      "learning_rate": 5e-05,
      "loss": 7.5807,
      "step": 105
    },
    {
      "epoch": 0.028943955218786264,
      "grad_norm": 16.559654235839844,
      "learning_rate": 4.917330276168208e-05,
      "loss": 7.6732,
      "step": 106
    },
    {
      "epoch": 0.02921701140009557,
      "grad_norm": 15.340529441833496,
      "learning_rate": 4.834683153413459e-05,
      "loss": 7.8505,
      "step": 107
    },
    {
      "epoch": 0.029490067581404874,
      "grad_norm": 17.633358001708984,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 8.5092,
      "step": 108
    },
    {
      "epoch": 0.02976312376271418,
      "grad_norm": 16.75809097290039,
      "learning_rate": 4.669547078371504e-05,
      "loss": 8.4787,
      "step": 109
    },
    {
      "epoch": 0.030036179944023484,
      "grad_norm": 16.6685848236084,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 7.757,
      "step": 110
    },
    {
      "epoch": 0.030309236125332786,
      "grad_norm": 16.05257225036621,
      "learning_rate": 4.504772348747687e-05,
      "loss": 8.5722,
      "step": 111
    },
    {
      "epoch": 0.03058229230664209,
      "grad_norm": 15.737967491149902,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 8.13,
      "step": 112
    },
    {
      "epoch": 0.030855348487951396,
      "grad_norm": 18.064294815063477,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 8.5067,
      "step": 113
    },
    {
      "epoch": 0.0311284046692607,
      "grad_norm": 23.488189697265625,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 8.6982,
      "step": 114
    },
    {
      "epoch": 0.03140146085057,
      "grad_norm": 17.5887508392334,
      "learning_rate": 4.17702704859633e-05,
      "loss": 8.8168,
      "step": 115
    },
    {
      "epoch": 0.03167451703187931,
      "grad_norm": 14.242671966552734,
      "learning_rate": 4.095597328339452e-05,
      "loss": 6.7286,
      "step": 116
    },
    {
      "epoch": 0.03194757321318861,
      "grad_norm": 14.122462272644043,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 8.5156,
      "step": 117
    },
    {
      "epoch": 0.03222062939449792,
      "grad_norm": 15.14977741241455,
      "learning_rate": 3.933501846281267e-05,
      "loss": 7.9296,
      "step": 118
    },
    {
      "epoch": 0.03249368557580722,
      "grad_norm": 17.33123779296875,
      "learning_rate": 3.852880399766243e-05,
      "loss": 8.8162,
      "step": 119
    },
    {
      "epoch": 0.032766741757116524,
      "grad_norm": 14.457858085632324,
      "learning_rate": 3.772572564296005e-05,
      "loss": 8.0185,
      "step": 120
    },
    {
      "epoch": 0.03303979793842583,
      "grad_norm": 15.224176406860352,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 8.4122,
      "step": 121
    },
    {
      "epoch": 0.033312854119735134,
      "grad_norm": 17.36722755432129,
      "learning_rate": 3.612985456190778e-05,
      "loss": 7.9025,
      "step": 122
    },
    {
      "epoch": 0.03358591030104444,
      "grad_norm": 15.94758415222168,
      "learning_rate": 3.533749813077677e-05,
      "loss": 8.2387,
      "step": 123
    },
    {
      "epoch": 0.033858966482353745,
      "grad_norm": 18.829530715942383,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 8.4182,
      "step": 124
    },
    {
      "epoch": 0.034132022663663046,
      "grad_norm": 18.859539031982422,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 8.3133,
      "step": 125
    },
    {
      "epoch": 0.034405078844972355,
      "grad_norm": 15.820075035095215,
      "learning_rate": 3.298534127791785e-05,
      "loss": 7.9701,
      "step": 126
    },
    {
      "epoch": 0.034678135026281656,
      "grad_norm": 18.90699577331543,
      "learning_rate": 3.221030765387417e-05,
      "loss": 8.6803,
      "step": 127
    },
    {
      "epoch": 0.034951191207590965,
      "grad_norm": 16.5141658782959,
      "learning_rate": 3.144013755408895e-05,
      "loss": 7.4541,
      "step": 128
    },
    {
      "epoch": 0.035224247388900266,
      "grad_norm": 18.62548828125,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 7.9896,
      "step": 129
    },
    {
      "epoch": 0.03549730357020957,
      "grad_norm": 19.355186462402344,
      "learning_rate": 2.991522876735154e-05,
      "loss": 8.5288,
      "step": 130
    },
    {
      "epoch": 0.035770359751518876,
      "grad_norm": 21.866207122802734,
      "learning_rate": 2.916090697523549e-05,
      "loss": 8.3981,
      "step": 131
    },
    {
      "epoch": 0.03604341593282818,
      "grad_norm": 17.40390396118164,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 8.2206,
      "step": 132
    },
    {
      "epoch": 0.03631647211413749,
      "grad_norm": 16.635597229003906,
      "learning_rate": 2.766955965735968e-05,
      "loss": 6.883,
      "step": 133
    },
    {
      "epoch": 0.03658952829544679,
      "grad_norm": 16.17715072631836,
      "learning_rate": 2.693294185106562e-05,
      "loss": 7.0958,
      "step": 134
    },
    {
      "epoch": 0.03686258447675609,
      "grad_norm": 19.20148468017578,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 8.0087,
      "step": 135
    },
    {
      "epoch": 0.0371356406580654,
      "grad_norm": 16.3226318359375,
      "learning_rate": 2.547882480847461e-05,
      "loss": 6.7139,
      "step": 136
    },
    {
      "epoch": 0.0374086968393747,
      "grad_norm": 22.199859619140625,
      "learning_rate": 2.476172311325783e-05,
      "loss": 8.521,
      "step": 137
    },
    {
      "epoch": 0.03768175302068401,
      "grad_norm": 25.420183181762695,
      "learning_rate": 2.405152131093926e-05,
      "loss": 7.8769,
      "step": 138
    },
    {
      "epoch": 0.03795480920199331,
      "grad_norm": 17.60880470275879,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 7.6496,
      "step": 139
    },
    {
      "epoch": 0.03822786538330261,
      "grad_norm": 19.348995208740234,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 8.1288,
      "step": 140
    },
    {
      "epoch": 0.03850092156461192,
      "grad_norm": 19.610876083374023,
      "learning_rate": 2.196424713241637e-05,
      "loss": 8.0972,
      "step": 141
    },
    {
      "epoch": 0.03877397774592122,
      "grad_norm": 33.40858840942383,
      "learning_rate": 2.128356686585282e-05,
      "loss": 9.7113,
      "step": 142
    },
    {
      "epoch": 0.03904703392723053,
      "grad_norm": 22.39961051940918,
      "learning_rate": 2.061073738537635e-05,
      "loss": 8.1275,
      "step": 143
    },
    {
      "epoch": 0.03932009010853983,
      "grad_norm": 24.164146423339844,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 8.4301,
      "step": 144
    },
    {
      "epoch": 0.039593146289849133,
      "grad_norm": 28.657455444335938,
      "learning_rate": 1.928936436551661e-05,
      "loss": 7.9669,
      "step": 145
    },
    {
      "epoch": 0.03986620247115844,
      "grad_norm": 30.79471206665039,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 8.4107,
      "step": 146
    },
    {
      "epoch": 0.040139258652467744,
      "grad_norm": 35.513553619384766,
      "learning_rate": 1.800157297483417e-05,
      "loss": 8.0686,
      "step": 147
    },
    {
      "epoch": 0.04041231483377705,
      "grad_norm": 20.78765106201172,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 7.3615,
      "step": 148
    },
    {
      "epoch": 0.040685371015086354,
      "grad_norm": 27.29558563232422,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 7.879,
      "step": 149
    },
    {
      "epoch": 0.040958427196395655,
      "grad_norm": 46.89344024658203,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 9.8891,
      "step": 150
    },
    {
      "epoch": 0.040958427196395655,
      "eval_loss": 2.0223450660705566,
      "eval_runtime": 698.0638,
      "eval_samples_per_second": 8.836,
      "eval_steps_per_second": 2.209,
      "step": 150
    },
    {
      "epoch": 0.041231483377704964,
      "grad_norm": 16.523544311523438,
      "learning_rate": 1.553232954407171e-05,
      "loss": 7.1991,
      "step": 151
    },
    {
      "epoch": 0.041504539559014265,
      "grad_norm": 17.955978393554688,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 7.6936,
      "step": 152
    },
    {
      "epoch": 0.041777595740323574,
      "grad_norm": 17.638277053833008,
      "learning_rate": 1.435357758543015e-05,
      "loss": 8.7365,
      "step": 153
    },
    {
      "epoch": 0.042050651921632876,
      "grad_norm": 16.629852294921875,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 6.3438,
      "step": 154
    },
    {
      "epoch": 0.04232370810294218,
      "grad_norm": 19.368207931518555,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 8.4654,
      "step": 155
    },
    {
      "epoch": 0.042596764284251486,
      "grad_norm": 16.003515243530273,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 6.7485,
      "step": 156
    },
    {
      "epoch": 0.04286982046556079,
      "grad_norm": 16.423484802246094,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 8.965,
      "step": 157
    },
    {
      "epoch": 0.043142876646870096,
      "grad_norm": 15.11328411102295,
      "learning_rate": 1.157994445715706e-05,
      "loss": 7.471,
      "step": 158
    },
    {
      "epoch": 0.0434159328281794,
      "grad_norm": 14.444985389709473,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 8.2315,
      "step": 159
    },
    {
      "epoch": 0.0436889890094887,
      "grad_norm": 15.822388648986816,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 8.7615,
      "step": 160
    },
    {
      "epoch": 0.04396204519079801,
      "grad_norm": 17.411666870117188,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 8.797,
      "step": 161
    },
    {
      "epoch": 0.04423510137210731,
      "grad_norm": 15.159920692443848,
      "learning_rate": 9.549150281252633e-06,
      "loss": 8.2739,
      "step": 162
    },
    {
      "epoch": 0.04450815755341662,
      "grad_norm": 13.763167381286621,
      "learning_rate": 9.068759265665384e-06,
      "loss": 6.8688,
      "step": 163
    },
    {
      "epoch": 0.04478121373472592,
      "grad_norm": 16.808382034301758,
      "learning_rate": 8.599558442598998e-06,
      "loss": 7.8035,
      "step": 164
    },
    {
      "epoch": 0.04505426991603522,
      "grad_norm": 14.98381519317627,
      "learning_rate": 8.141676086873572e-06,
      "loss": 8.4142,
      "step": 165
    },
    {
      "epoch": 0.04532732609734453,
      "grad_norm": 16.07741355895996,
      "learning_rate": 7.695237378953223e-06,
      "loss": 7.7865,
      "step": 166
    },
    {
      "epoch": 0.04560038227865383,
      "grad_norm": 14.964221000671387,
      "learning_rate": 7.260364370723044e-06,
      "loss": 7.954,
      "step": 167
    },
    {
      "epoch": 0.04587343845996314,
      "grad_norm": 15.643426895141602,
      "learning_rate": 6.837175952121306e-06,
      "loss": 8.7938,
      "step": 168
    },
    {
      "epoch": 0.04614649464127244,
      "grad_norm": 15.192752838134766,
      "learning_rate": 6.425787818636131e-06,
      "loss": 7.7352,
      "step": 169
    },
    {
      "epoch": 0.04641955082258175,
      "grad_norm": 16.6499080657959,
      "learning_rate": 6.026312439675552e-06,
      "loss": 8.0943,
      "step": 170
    },
    {
      "epoch": 0.04669260700389105,
      "grad_norm": 16.1639461517334,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 8.5706,
      "step": 171
    },
    {
      "epoch": 0.04696566318520035,
      "grad_norm": 15.77147388458252,
      "learning_rate": 5.263533508961827e-06,
      "loss": 7.8765,
      "step": 172
    },
    {
      "epoch": 0.04723871936650966,
      "grad_norm": 16.122617721557617,
      "learning_rate": 4.900438493352055e-06,
      "loss": 7.8293,
      "step": 173
    },
    {
      "epoch": 0.04751177554781896,
      "grad_norm": 17.912139892578125,
      "learning_rate": 4.549673247541875e-06,
      "loss": 7.7461,
      "step": 174
    },
    {
      "epoch": 0.04778483172912827,
      "grad_norm": 16.062545776367188,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 7.9451,
      "step": 175
    },
    {
      "epoch": 0.04805788791043757,
      "grad_norm": 16.40523910522461,
      "learning_rate": 3.885512251130763e-06,
      "loss": 8.1723,
      "step": 176
    },
    {
      "epoch": 0.048330944091746875,
      "grad_norm": 15.758687973022461,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 7.5553,
      "step": 177
    },
    {
      "epoch": 0.04860400027305618,
      "grad_norm": 20.640363693237305,
      "learning_rate": 3.271776770026963e-06,
      "loss": 8.0343,
      "step": 178
    },
    {
      "epoch": 0.048877056454365485,
      "grad_norm": 23.85805320739746,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 8.2615,
      "step": 179
    },
    {
      "epoch": 0.04915011263567479,
      "grad_norm": 17.42931365966797,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 7.8427,
      "step": 180
    },
    {
      "epoch": 0.049423168816984095,
      "grad_norm": 16.42197036743164,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 7.7824,
      "step": 181
    },
    {
      "epoch": 0.049696224998293397,
      "grad_norm": 16.903270721435547,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 7.4511,
      "step": 182
    },
    {
      "epoch": 0.049969281179602705,
      "grad_norm": 26.65274429321289,
      "learning_rate": 1.962316193157593e-06,
      "loss": 8.9241,
      "step": 183
    },
    {
      "epoch": 0.05024233736091201,
      "grad_norm": 24.605058670043945,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 7.8053,
      "step": 184
    },
    {
      "epoch": 0.050515393542221315,
      "grad_norm": 22.182266235351562,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 8.1749,
      "step": 185
    },
    {
      "epoch": 0.05078844972353062,
      "grad_norm": 19.286502838134766,
      "learning_rate": 1.333670137599713e-06,
      "loss": 7.6814,
      "step": 186
    },
    {
      "epoch": 0.05106150590483992,
      "grad_norm": 20.933347702026367,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 8.4776,
      "step": 187
    },
    {
      "epoch": 0.05133456208614923,
      "grad_norm": 19.077287673950195,
      "learning_rate": 9.810017062595322e-07,
      "loss": 7.317,
      "step": 188
    },
    {
      "epoch": 0.05160761826745853,
      "grad_norm": 22.509185791015625,
      "learning_rate": 8.247462563808817e-07,
      "loss": 6.9666,
      "step": 189
    },
    {
      "epoch": 0.05188067444876784,
      "grad_norm": 20.019676208496094,
      "learning_rate": 6.819348298638839e-07,
      "loss": 7.0849,
      "step": 190
    },
    {
      "epoch": 0.05215373063007714,
      "grad_norm": 24.726022720336914,
      "learning_rate": 5.526064699265753e-07,
      "loss": 7.4983,
      "step": 191
    },
    {
      "epoch": 0.05242678681138644,
      "grad_norm": 21.559818267822266,
      "learning_rate": 4.367965336512403e-07,
      "loss": 8.1443,
      "step": 192
    },
    {
      "epoch": 0.05269984299269575,
      "grad_norm": 19.59793472290039,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 7.9091,
      "step": 193
    },
    {
      "epoch": 0.05297289917400505,
      "grad_norm": 21.68771743774414,
      "learning_rate": 2.458548727494292e-07,
      "loss": 8.5354,
      "step": 194
    },
    {
      "epoch": 0.05324595535531436,
      "grad_norm": 21.212215423583984,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 7.3903,
      "step": 195
    },
    {
      "epoch": 0.05351901153662366,
      "grad_norm": 24.273283004760742,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 8.1784,
      "step": 196
    },
    {
      "epoch": 0.05379206771793296,
      "grad_norm": 22.68973731994629,
      "learning_rate": 6.150154258476315e-08,
      "loss": 7.6137,
      "step": 197
    },
    {
      "epoch": 0.05406512389924227,
      "grad_norm": 23.224477767944336,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 6.6664,
      "step": 198
    },
    {
      "epoch": 0.05433818008055157,
      "grad_norm": 34.137840270996094,
      "learning_rate": 6.834750376549792e-09,
      "loss": 9.638,
      "step": 199
    },
    {
      "epoch": 0.05461123626186088,
      "grad_norm": 43.94276428222656,
      "learning_rate": 0.0,
      "loss": 9.4179,
      "step": 200
    },
    {
      "epoch": 0.05461123626186088,
      "eval_loss": 1.973547101020813,
      "eval_runtime": 698.3906,
      "eval_samples_per_second": 8.832,
      "eval_steps_per_second": 2.208,
      "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
}