{
  "best_metric": 2.477691650390625,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.02730561813093044,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0001365280906546522,
      "grad_norm": 0.27194273471832275,
      "learning_rate": 1e-05,
      "loss": 2.7279,
      "step": 1
    },
    {
      "epoch": 0.0001365280906546522,
      "eval_loss": 3.104750156402588,
      "eval_runtime": 365.268,
      "eval_samples_per_second": 33.772,
      "eval_steps_per_second": 8.443,
      "step": 1
    },
    {
      "epoch": 0.0002730561813093044,
      "grad_norm": 0.2629146873950958,
      "learning_rate": 2e-05,
      "loss": 2.8941,
      "step": 2
    },
    {
      "epoch": 0.00040958427196395656,
      "grad_norm": 0.35042041540145874,
      "learning_rate": 3e-05,
      "loss": 2.7248,
      "step": 3
    },
    {
      "epoch": 0.0005461123626186088,
      "grad_norm": 0.27044424414634705,
      "learning_rate": 4e-05,
      "loss": 2.9933,
      "step": 4
    },
    {
      "epoch": 0.0006826404532732609,
      "grad_norm": 0.2519526183605194,
      "learning_rate": 5e-05,
      "loss": 2.8809,
      "step": 5
    },
    {
      "epoch": 0.0008191685439279131,
      "grad_norm": 0.32828840613365173,
      "learning_rate": 6e-05,
      "loss": 2.9901,
      "step": 6
    },
    {
      "epoch": 0.0009556966345825653,
      "grad_norm": 0.3066588342189789,
      "learning_rate": 7e-05,
      "loss": 2.8243,
      "step": 7
    },
    {
      "epoch": 0.0010922247252372176,
      "grad_norm": 0.25900742411613464,
      "learning_rate": 8e-05,
      "loss": 2.9606,
      "step": 8
    },
    {
      "epoch": 0.0012287528158918697,
      "grad_norm": 0.26237908005714417,
      "learning_rate": 9e-05,
      "loss": 2.8853,
      "step": 9
    },
    {
      "epoch": 0.0013652809065465218,
      "grad_norm": 0.30526629090309143,
      "learning_rate": 0.0001,
      "loss": 3.0559,
      "step": 10
    },
    {
      "epoch": 0.0015018089972011742,
      "grad_norm": 0.2982544004917145,
      "learning_rate": 9.999316524962345e-05,
      "loss": 2.7064,
      "step": 11
    },
    {
      "epoch": 0.0016383370878558263,
      "grad_norm": 0.24506962299346924,
      "learning_rate": 9.997266286704631e-05,
      "loss": 2.6757,
      "step": 12
    },
    {
      "epoch": 0.0017748651785104786,
      "grad_norm": 0.2576853036880493,
      "learning_rate": 9.993849845741524e-05,
      "loss": 2.9024,
      "step": 13
    },
    {
      "epoch": 0.0019113932691651307,
      "grad_norm": 0.2723166048526764,
      "learning_rate": 9.989068136093873e-05,
      "loss": 2.7838,
      "step": 14
    },
    {
      "epoch": 0.0020479213598197828,
      "grad_norm": 0.29228994250297546,
      "learning_rate": 9.98292246503335e-05,
      "loss": 2.716,
      "step": 15
    },
    {
      "epoch": 0.0021844494504744353,
      "grad_norm": 0.35024645924568176,
      "learning_rate": 9.975414512725057e-05,
      "loss": 2.6975,
      "step": 16
    },
    {
      "epoch": 0.0023209775411290874,
      "grad_norm": 0.29096874594688416,
      "learning_rate": 9.966546331768191e-05,
      "loss": 2.6518,
      "step": 17
    },
    {
      "epoch": 0.0024575056317837395,
      "grad_norm": 0.45739424228668213,
      "learning_rate": 9.956320346634876e-05,
      "loss": 2.6521,
      "step": 18
    },
    {
      "epoch": 0.0025940337224383916,
      "grad_norm": 0.2922545075416565,
      "learning_rate": 9.944739353007344e-05,
      "loss": 2.7183,
      "step": 19
    },
    {
      "epoch": 0.0027305618130930437,
      "grad_norm": 0.3041123151779175,
      "learning_rate": 9.931806517013612e-05,
      "loss": 2.7687,
      "step": 20
    },
    {
      "epoch": 0.0028670899037476962,
      "grad_norm": 0.25775209069252014,
      "learning_rate": 9.917525374361912e-05,
      "loss": 2.6477,
      "step": 21
    },
    {
      "epoch": 0.0030036179944023483,
      "grad_norm": 0.2592810094356537,
      "learning_rate": 9.901899829374047e-05,
      "loss": 2.6581,
      "step": 22
    },
    {
      "epoch": 0.0031401460850570004,
      "grad_norm": 0.2812885046005249,
      "learning_rate": 9.884934153917997e-05,
      "loss": 2.7239,
      "step": 23
    },
    {
      "epoch": 0.0032766741757116525,
      "grad_norm": 0.3217126727104187,
      "learning_rate": 9.86663298624003e-05,
      "loss": 2.6797,
      "step": 24
    },
    {
      "epoch": 0.003413202266366305,
      "grad_norm": 0.2805332541465759,
      "learning_rate": 9.847001329696653e-05,
      "loss": 2.554,
      "step": 25
    },
    {
      "epoch": 0.003549730357020957,
      "grad_norm": 0.2860311269760132,
      "learning_rate": 9.826044551386744e-05,
      "loss": 2.7855,
      "step": 26
    },
    {
      "epoch": 0.0036862584476756092,
      "grad_norm": 0.28637662529945374,
      "learning_rate": 9.803768380684242e-05,
      "loss": 2.7207,
      "step": 27
    },
    {
      "epoch": 0.0038227865383302613,
      "grad_norm": 0.29103153944015503,
      "learning_rate": 9.780178907671789e-05,
      "loss": 2.6371,
      "step": 28
    },
    {
      "epoch": 0.003959314628984914,
      "grad_norm": 0.2585844099521637,
      "learning_rate": 9.755282581475769e-05,
      "loss": 2.6343,
      "step": 29
    },
    {
      "epoch": 0.0040958427196395655,
      "grad_norm": 0.27870482206344604,
      "learning_rate": 9.729086208503174e-05,
      "loss": 2.7855,
      "step": 30
    },
    {
      "epoch": 0.004232370810294218,
      "grad_norm": 0.39990758895874023,
      "learning_rate": 9.701596950580806e-05,
      "loss": 2.5737,
      "step": 31
    },
    {
      "epoch": 0.004368898900948871,
      "grad_norm": 0.29908424615859985,
      "learning_rate": 9.672822322997305e-05,
      "loss": 2.551,
      "step": 32
    },
    {
      "epoch": 0.004505426991603522,
      "grad_norm": 0.30026519298553467,
      "learning_rate": 9.642770192448536e-05,
      "loss": 2.5303,
      "step": 33
    },
    {
      "epoch": 0.004641955082258175,
      "grad_norm": 0.29965054988861084,
      "learning_rate": 9.611448774886924e-05,
      "loss": 2.6088,
      "step": 34
    },
    {
      "epoch": 0.0047784831729128265,
      "grad_norm": 0.46132540702819824,
      "learning_rate": 9.578866633275288e-05,
      "loss": 2.6285,
      "step": 35
    },
    {
      "epoch": 0.004915011263567479,
      "grad_norm": 0.34419891238212585,
      "learning_rate": 9.545032675245813e-05,
      "loss": 2.8227,
      "step": 36
    },
    {
      "epoch": 0.0050515393542221315,
      "grad_norm": 0.30205267667770386,
      "learning_rate": 9.509956150664796e-05,
      "loss": 2.481,
      "step": 37
    },
    {
      "epoch": 0.005188067444876783,
      "grad_norm": 0.32067328691482544,
      "learning_rate": 9.473646649103818e-05,
      "loss": 2.6081,
      "step": 38
    },
    {
      "epoch": 0.005324595535531436,
      "grad_norm": 0.36380669474601746,
      "learning_rate": 9.43611409721806e-05,
      "loss": 2.4902,
      "step": 39
    },
    {
      "epoch": 0.005461123626186087,
      "grad_norm": 0.3331817388534546,
      "learning_rate": 9.397368756032445e-05,
      "loss": 2.5083,
      "step": 40
    },
    {
      "epoch": 0.00559765171684074,
      "grad_norm": 0.33512428402900696,
      "learning_rate": 9.357421218136386e-05,
      "loss": 2.5393,
      "step": 41
    },
    {
      "epoch": 0.0057341798074953924,
      "grad_norm": 0.3623019754886627,
      "learning_rate": 9.316282404787871e-05,
      "loss": 2.6148,
      "step": 42
    },
    {
      "epoch": 0.005870707898150044,
      "grad_norm": 0.3712327182292938,
      "learning_rate": 9.273963562927695e-05,
      "loss": 2.6677,
      "step": 43
    },
    {
      "epoch": 0.006007235988804697,
      "grad_norm": 0.3600422441959381,
      "learning_rate": 9.230476262104677e-05,
      "loss": 2.5589,
      "step": 44
    },
    {
      "epoch": 0.006143764079459349,
      "grad_norm": 0.3411221504211426,
      "learning_rate": 9.185832391312644e-05,
      "loss": 2.5302,
      "step": 45
    },
    {
      "epoch": 0.006280292170114001,
      "grad_norm": 0.4312603175640106,
      "learning_rate": 9.140044155740101e-05,
      "loss": 2.8364,
      "step": 46
    },
    {
      "epoch": 0.006416820260768653,
      "grad_norm": 0.42747628688812256,
      "learning_rate": 9.093124073433463e-05,
      "loss": 2.9709,
      "step": 47
    },
    {
      "epoch": 0.006553348351423305,
      "grad_norm": 0.43163979053497314,
      "learning_rate": 9.045084971874738e-05,
      "loss": 2.746,
      "step": 48
    },
    {
      "epoch": 0.0066898764420779576,
      "grad_norm": 0.5412802696228027,
      "learning_rate": 8.995939984474624e-05,
      "loss": 2.9371,
      "step": 49
    },
    {
      "epoch": 0.00682640453273261,
      "grad_norm": 0.777123212814331,
      "learning_rate": 8.945702546981969e-05,
      "loss": 2.6875,
      "step": 50
    },
    {
      "epoch": 0.00682640453273261,
      "eval_loss": 2.62621808052063,
      "eval_runtime": 365.1041,
      "eval_samples_per_second": 33.788,
      "eval_steps_per_second": 8.447,
      "step": 50
    },
    {
      "epoch": 0.006962932623387262,
      "grad_norm": 0.45155230164527893,
      "learning_rate": 8.894386393810563e-05,
      "loss": 2.4352,
      "step": 51
    },
    {
      "epoch": 0.007099460714041914,
      "grad_norm": 0.46463319659233093,
      "learning_rate": 8.842005554284296e-05,
      "loss": 2.6966,
      "step": 52
    },
    {
      "epoch": 0.007235988804696566,
      "grad_norm": 0.5098180174827576,
      "learning_rate": 8.788574348801675e-05,
      "loss": 2.6095,
      "step": 53
    },
    {
      "epoch": 0.0073725168953512185,
      "grad_norm": 0.3428554832935333,
      "learning_rate": 8.73410738492077e-05,
      "loss": 2.6795,
      "step": 54
    },
    {
      "epoch": 0.007509044986005871,
      "grad_norm": 0.2610047459602356,
      "learning_rate": 8.678619553365659e-05,
      "loss": 2.6242,
      "step": 55
    },
    {
      "epoch": 0.007645573076660523,
      "grad_norm": 0.22234882414340973,
      "learning_rate": 8.622126023955446e-05,
      "loss": 2.5632,
      "step": 56
    },
    {
      "epoch": 0.007782101167315175,
      "grad_norm": 0.20495741069316864,
      "learning_rate": 8.564642241456986e-05,
      "loss": 2.6443,
      "step": 57
    },
    {
      "epoch": 0.007918629257969828,
      "grad_norm": 0.24717627465724945,
      "learning_rate": 8.506183921362443e-05,
      "loss": 2.6157,
      "step": 58
    },
    {
      "epoch": 0.00805515734862448,
      "grad_norm": 0.20852558314800262,
      "learning_rate": 8.44676704559283e-05,
      "loss": 2.6384,
      "step": 59
    },
    {
      "epoch": 0.008191685439279131,
      "grad_norm": 0.2658625841140747,
      "learning_rate": 8.386407858128706e-05,
      "loss": 2.6264,
      "step": 60
    },
    {
      "epoch": 0.008328213529933784,
      "grad_norm": 0.24918319284915924,
      "learning_rate": 8.32512286056924e-05,
      "loss": 2.5451,
      "step": 61
    },
    {
      "epoch": 0.008464741620588436,
      "grad_norm": 0.23944678902626038,
      "learning_rate": 8.262928807620843e-05,
      "loss": 2.4863,
      "step": 62
    },
    {
      "epoch": 0.008601269711243089,
      "grad_norm": 0.23544050753116608,
      "learning_rate": 8.199842702516583e-05,
      "loss": 2.5433,
      "step": 63
    },
    {
      "epoch": 0.008737797801897741,
      "grad_norm": 0.26784569025039673,
      "learning_rate": 8.135881792367686e-05,
      "loss": 2.6238,
      "step": 64
    },
    {
      "epoch": 0.008874325892552392,
      "grad_norm": 0.2575049102306366,
      "learning_rate": 8.07106356344834e-05,
      "loss": 2.4912,
      "step": 65
    },
    {
      "epoch": 0.009010853983207045,
      "grad_norm": 0.30827003717422485,
      "learning_rate": 8.005405736415126e-05,
      "loss": 2.4266,
      "step": 66
    },
    {
      "epoch": 0.009147382073861697,
      "grad_norm": 0.27325698733329773,
      "learning_rate": 7.938926261462366e-05,
      "loss": 2.5859,
      "step": 67
    },
    {
      "epoch": 0.00928391016451635,
      "grad_norm": 0.2419956475496292,
      "learning_rate": 7.871643313414718e-05,
      "loss": 2.5465,
      "step": 68
    },
    {
      "epoch": 0.009420438255171002,
      "grad_norm": 0.2588096261024475,
      "learning_rate": 7.803575286758364e-05,
      "loss": 2.5389,
      "step": 69
    },
    {
      "epoch": 0.009556966345825653,
      "grad_norm": 0.23883476853370667,
      "learning_rate": 7.734740790612136e-05,
      "loss": 2.4677,
      "step": 70
    },
    {
      "epoch": 0.009693494436480305,
      "grad_norm": 0.25533339381217957,
      "learning_rate": 7.66515864363997e-05,
      "loss": 2.683,
      "step": 71
    },
    {
      "epoch": 0.009830022527134958,
      "grad_norm": 0.24782323837280273,
      "learning_rate": 7.594847868906076e-05,
      "loss": 2.668,
      "step": 72
    },
    {
      "epoch": 0.00996655061778961,
      "grad_norm": 0.23901385068893433,
      "learning_rate": 7.52382768867422e-05,
      "loss": 2.4514,
      "step": 73
    },
    {
      "epoch": 0.010103078708444263,
      "grad_norm": 0.2513439357280731,
      "learning_rate": 7.452117519152542e-05,
      "loss": 2.5004,
      "step": 74
    },
    {
      "epoch": 0.010239606799098914,
      "grad_norm": 0.3062717616558075,
      "learning_rate": 7.379736965185368e-05,
      "loss": 2.4933,
      "step": 75
    },
    {
      "epoch": 0.010376134889753566,
      "grad_norm": 0.2644160985946655,
      "learning_rate": 7.30670581489344e-05,
      "loss": 2.5324,
      "step": 76
    },
    {
      "epoch": 0.010512662980408219,
      "grad_norm": 0.2861580550670624,
      "learning_rate": 7.233044034264034e-05,
      "loss": 2.7596,
      "step": 77
    },
    {
      "epoch": 0.010649191071062871,
      "grad_norm": 0.2912580966949463,
      "learning_rate": 7.158771761692464e-05,
      "loss": 2.5224,
      "step": 78
    },
    {
      "epoch": 0.010785719161717524,
      "grad_norm": 0.3743256628513336,
      "learning_rate": 7.083909302476453e-05,
      "loss": 2.4229,
      "step": 79
    },
    {
      "epoch": 0.010922247252372175,
      "grad_norm": 0.293729305267334,
      "learning_rate": 7.008477123264848e-05,
      "loss": 2.6703,
      "step": 80
    },
    {
      "epoch": 0.011058775343026827,
      "grad_norm": 0.2978295087814331,
      "learning_rate": 6.932495846462261e-05,
      "loss": 2.5538,
      "step": 81
    },
    {
      "epoch": 0.01119530343368148,
      "grad_norm": 0.29679563641548157,
      "learning_rate": 6.855986244591104e-05,
      "loss": 2.6379,
      "step": 82
    },
    {
      "epoch": 0.011331831524336132,
      "grad_norm": 0.2750690281391144,
      "learning_rate": 6.778969234612584e-05,
      "loss": 2.5469,
      "step": 83
    },
    {
      "epoch": 0.011468359614990785,
      "grad_norm": 0.2990764379501343,
      "learning_rate": 6.701465872208216e-05,
      "loss": 2.4238,
      "step": 84
    },
    {
      "epoch": 0.011604887705645437,
      "grad_norm": 0.3153710961341858,
      "learning_rate": 6.623497346023418e-05,
      "loss": 2.571,
      "step": 85
    },
    {
      "epoch": 0.011741415796300088,
      "grad_norm": 0.3125828802585602,
      "learning_rate": 6.545084971874738e-05,
      "loss": 2.4008,
      "step": 86
    },
    {
      "epoch": 0.01187794388695474,
      "grad_norm": 0.34010761976242065,
      "learning_rate": 6.466250186922325e-05,
      "loss": 2.3982,
      "step": 87
    },
    {
      "epoch": 0.012014471977609393,
      "grad_norm": 0.3255278766155243,
      "learning_rate": 6.387014543809223e-05,
      "loss": 2.4671,
      "step": 88
    },
    {
      "epoch": 0.012151000068264046,
      "grad_norm": 0.33447909355163574,
      "learning_rate": 6.307399704769099e-05,
      "loss": 2.5746,
      "step": 89
    },
    {
      "epoch": 0.012287528158918698,
      "grad_norm": 0.3527494966983795,
      "learning_rate": 6.227427435703997e-05,
      "loss": 2.6133,
      "step": 90
    },
    {
      "epoch": 0.012424056249573349,
      "grad_norm": 0.33412718772888184,
      "learning_rate": 6.147119600233758e-05,
      "loss": 2.4584,
      "step": 91
    },
    {
      "epoch": 0.012560584340228002,
      "grad_norm": 0.34142249822616577,
      "learning_rate": 6.066498153718735e-05,
      "loss": 2.6819,
      "step": 92
    },
    {
      "epoch": 0.012697112430882654,
      "grad_norm": 0.3621145486831665,
      "learning_rate": 5.985585137257401e-05,
      "loss": 2.4176,
      "step": 93
    },
    {
      "epoch": 0.012833640521537307,
      "grad_norm": 0.43017393350601196,
      "learning_rate": 5.90440267166055e-05,
      "loss": 2.6882,
      "step": 94
    },
    {
      "epoch": 0.01297016861219196,
      "grad_norm": 0.39249667525291443,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 2.4088,
      "step": 95
    },
    {
      "epoch": 0.01310669670284661,
      "grad_norm": 0.4208681881427765,
      "learning_rate": 5.74131823855921e-05,
      "loss": 2.6203,
      "step": 96
    },
    {
      "epoch": 0.013243224793501263,
      "grad_norm": 0.43923816084861755,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 2.9,
      "step": 97
    },
    {
      "epoch": 0.013379752884155915,
      "grad_norm": 0.47433754801750183,
      "learning_rate": 5.577423184847932e-05,
      "loss": 2.7912,
      "step": 98
    },
    {
      "epoch": 0.013516280974810568,
      "grad_norm": 0.5563204288482666,
      "learning_rate": 5.495227651252315e-05,
      "loss": 2.9368,
      "step": 99
    },
    {
      "epoch": 0.01365280906546522,
      "grad_norm": 0.7900686860084534,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 2.4241,
      "step": 100
    },
    {
      "epoch": 0.01365280906546522,
      "eval_loss": 2.5399112701416016,
      "eval_runtime": 365.0262,
      "eval_samples_per_second": 33.795,
      "eval_steps_per_second": 8.449,
      "step": 100
    },
    {
      "epoch": 0.013789337156119871,
      "grad_norm": 0.31755203008651733,
      "learning_rate": 5.330452921628497e-05,
      "loss": 2.6201,
      "step": 101
    },
    {
      "epoch": 0.013925865246774524,
      "grad_norm": 0.360796719789505,
      "learning_rate": 5.247918773366112e-05,
      "loss": 2.6936,
      "step": 102
    },
    {
      "epoch": 0.014062393337429176,
      "grad_norm": 0.3812713027000427,
      "learning_rate": 5.165316846586541e-05,
      "loss": 2.5351,
      "step": 103
    },
    {
      "epoch": 0.014198921428083829,
      "grad_norm": 0.3634807765483856,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 2.5735,
      "step": 104
    },
    {
      "epoch": 0.014335449518738481,
      "grad_norm": 0.3399415612220764,
      "learning_rate": 5e-05,
      "loss": 2.6437,
      "step": 105
    },
    {
      "epoch": 0.014471977609393132,
      "grad_norm": 0.3060910701751709,
      "learning_rate": 4.917330276168208e-05,
      "loss": 2.5458,
      "step": 106
    },
    {
      "epoch": 0.014608505700047784,
      "grad_norm": 0.26493722200393677,
      "learning_rate": 4.834683153413459e-05,
      "loss": 2.6479,
      "step": 107
    },
    {
      "epoch": 0.014745033790702437,
      "grad_norm": 0.25897419452667236,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 2.6475,
      "step": 108
    },
    {
      "epoch": 0.01488156188135709,
      "grad_norm": 0.25238776206970215,
      "learning_rate": 4.669547078371504e-05,
      "loss": 2.5239,
      "step": 109
    },
    {
      "epoch": 0.015018089972011742,
      "grad_norm": 0.35802146792411804,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 2.4212,
      "step": 110
    },
    {
      "epoch": 0.015154618062666393,
      "grad_norm": 0.25561901926994324,
      "learning_rate": 4.504772348747687e-05,
      "loss": 2.5157,
      "step": 111
    },
    {
      "epoch": 0.015291146153321045,
      "grad_norm": 0.2591532766819,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 2.4992,
      "step": 112
    },
    {
      "epoch": 0.015427674243975698,
      "grad_norm": 0.23964530229568481,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 2.4607,
      "step": 113
    },
    {
      "epoch": 0.01556420233463035,
      "grad_norm": 0.2713939845561981,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 2.5891,
      "step": 114
    },
    {
      "epoch": 0.015700730425285,
      "grad_norm": 0.2692546546459198,
      "learning_rate": 4.17702704859633e-05,
      "loss": 2.6841,
      "step": 115
    },
    {
      "epoch": 0.015837258515939655,
      "grad_norm": 0.2805223762989044,
      "learning_rate": 4.095597328339452e-05,
      "loss": 2.4114,
      "step": 116
    },
    {
      "epoch": 0.015973786606594306,
      "grad_norm": 0.2691623568534851,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 2.4756,
      "step": 117
    },
    {
      "epoch": 0.01611031469724896,
      "grad_norm": 0.33222973346710205,
      "learning_rate": 3.933501846281267e-05,
      "loss": 2.3796,
      "step": 118
    },
    {
      "epoch": 0.01624684278790361,
      "grad_norm": 0.2569933831691742,
      "learning_rate": 3.852880399766243e-05,
      "loss": 2.3322,
      "step": 119
    },
    {
      "epoch": 0.016383370878558262,
      "grad_norm": 0.26831957697868347,
      "learning_rate": 3.772572564296005e-05,
      "loss": 2.5066,
      "step": 120
    },
    {
      "epoch": 0.016519898969212916,
      "grad_norm": 0.2767285108566284,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 2.4728,
      "step": 121
    },
    {
      "epoch": 0.016656427059867567,
      "grad_norm": 0.2877205014228821,
      "learning_rate": 3.612985456190778e-05,
      "loss": 2.5232,
      "step": 122
    },
    {
      "epoch": 0.01679295515052222,
      "grad_norm": 0.28673556447029114,
      "learning_rate": 3.533749813077677e-05,
      "loss": 2.3521,
      "step": 123
    },
    {
      "epoch": 0.016929483241176872,
      "grad_norm": 0.28386664390563965,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 2.3625,
      "step": 124
    },
    {
      "epoch": 0.017066011331831523,
      "grad_norm": 0.29683011770248413,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 2.483,
      "step": 125
    },
    {
      "epoch": 0.017202539422486177,
      "grad_norm": 0.29346969723701477,
      "learning_rate": 3.298534127791785e-05,
      "loss": 2.5325,
      "step": 126
    },
    {
      "epoch": 0.017339067513140828,
      "grad_norm": 0.29306894540786743,
      "learning_rate": 3.221030765387417e-05,
      "loss": 2.5124,
      "step": 127
    },
    {
      "epoch": 0.017475595603795482,
      "grad_norm": 0.3129117786884308,
      "learning_rate": 3.144013755408895e-05,
      "loss": 2.3768,
      "step": 128
    },
    {
      "epoch": 0.017612123694450133,
      "grad_norm": 0.33181077241897583,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 2.5873,
      "step": 129
    },
    {
      "epoch": 0.017748651785104784,
      "grad_norm": 0.46079254150390625,
      "learning_rate": 2.991522876735154e-05,
      "loss": 2.2878,
      "step": 130
    },
    {
      "epoch": 0.017885179875759438,
      "grad_norm": 0.33316582441329956,
      "learning_rate": 2.916090697523549e-05,
      "loss": 2.5747,
      "step": 131
    },
    {
      "epoch": 0.01802170796641409,
      "grad_norm": 0.3477618396282196,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 2.7068,
      "step": 132
    },
    {
      "epoch": 0.018158236057068743,
      "grad_norm": 0.3639448583126068,
      "learning_rate": 2.766955965735968e-05,
      "loss": 2.4444,
      "step": 133
    },
    {
      "epoch": 0.018294764147723394,
      "grad_norm": 0.327484130859375,
      "learning_rate": 2.693294185106562e-05,
      "loss": 2.2098,
      "step": 134
    },
    {
      "epoch": 0.018431292238378045,
      "grad_norm": 0.3626045286655426,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 2.3977,
      "step": 135
    },
    {
      "epoch": 0.0185678203290327,
      "grad_norm": 0.32435932755470276,
      "learning_rate": 2.547882480847461e-05,
      "loss": 2.4311,
      "step": 136
    },
    {
      "epoch": 0.01870434841968735,
      "grad_norm": 0.356877863407135,
      "learning_rate": 2.476172311325783e-05,
      "loss": 2.6394,
      "step": 137
    },
    {
      "epoch": 0.018840876510342004,
      "grad_norm": 0.3856317400932312,
      "learning_rate": 2.405152131093926e-05,
      "loss": 2.2952,
      "step": 138
    },
    {
      "epoch": 0.018977404600996655,
      "grad_norm": 0.38227829337120056,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 2.3519,
      "step": 139
    },
    {
      "epoch": 0.019113932691651306,
      "grad_norm": 0.36895522475242615,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 2.3899,
      "step": 140
    },
    {
      "epoch": 0.01925046078230596,
      "grad_norm": 0.36597883701324463,
      "learning_rate": 2.196424713241637e-05,
      "loss": 2.4696,
      "step": 141
    },
    {
      "epoch": 0.01938698887296061,
      "grad_norm": 0.38308030366897583,
      "learning_rate": 2.128356686585282e-05,
      "loss": 2.406,
      "step": 142
    },
    {
      "epoch": 0.019523516963615265,
      "grad_norm": 0.3991979658603668,
      "learning_rate": 2.061073738537635e-05,
      "loss": 2.4898,
      "step": 143
    },
    {
      "epoch": 0.019660045054269916,
      "grad_norm": 0.42273375391960144,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 2.6218,
      "step": 144
    },
    {
      "epoch": 0.019796573144924567,
      "grad_norm": 0.46673694252967834,
      "learning_rate": 1.928936436551661e-05,
      "loss": 2.7304,
      "step": 145
    },
    {
      "epoch": 0.01993310123557922,
      "grad_norm": 0.46778738498687744,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 2.6397,
      "step": 146
    },
    {
      "epoch": 0.020069629326233872,
      "grad_norm": 0.4708595275878906,
      "learning_rate": 1.800157297483417e-05,
      "loss": 2.8079,
      "step": 147
    },
    {
      "epoch": 0.020206157416888526,
      "grad_norm": 0.5667031407356262,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 2.6724,
      "step": 148
    },
    {
      "epoch": 0.020342685507543177,
      "grad_norm": 0.5661789178848267,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 2.5638,
      "step": 149
    },
    {
      "epoch": 0.020479213598197828,
      "grad_norm": 0.7007172703742981,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 2.1129,
      "step": 150
    },
    {
      "epoch": 0.020479213598197828,
      "eval_loss": 2.4890899658203125,
      "eval_runtime": 365.1789,
      "eval_samples_per_second": 33.781,
      "eval_steps_per_second": 8.445,
      "step": 150
    },
    {
      "epoch": 0.020615741688852482,
      "grad_norm": 0.23158541321754456,
      "learning_rate": 1.553232954407171e-05,
      "loss": 2.4899,
      "step": 151
    },
    {
      "epoch": 0.020752269779507133,
      "grad_norm": 0.25515466928482056,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 2.5051,
      "step": 152
    },
    {
      "epoch": 0.020888797870161787,
      "grad_norm": 0.39401036500930786,
      "learning_rate": 1.435357758543015e-05,
      "loss": 2.2306,
      "step": 153
    },
    {
      "epoch": 0.021025325960816438,
      "grad_norm": 0.2779545187950134,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 2.5515,
      "step": 154
    },
    {
      "epoch": 0.02116185405147109,
      "grad_norm": 0.2996259033679962,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 2.4722,
      "step": 155
    },
    {
      "epoch": 0.021298382142125743,
      "grad_norm": 0.325000137090683,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 2.6084,
      "step": 156
    },
    {
      "epoch": 0.021434910232780394,
      "grad_norm": 0.29446229338645935,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 2.5902,
      "step": 157
    },
    {
      "epoch": 0.021571438323435048,
      "grad_norm": 0.2947021722793579,
      "learning_rate": 1.157994445715706e-05,
      "loss": 2.6818,
      "step": 158
    },
    {
      "epoch": 0.0217079664140897,
      "grad_norm": 0.2855950593948364,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 2.5575,
      "step": 159
    },
    {
      "epoch": 0.02184449450474435,
      "grad_norm": 0.30326858162879944,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 2.5987,
      "step": 160
    },
    {
      "epoch": 0.021981022595399004,
      "grad_norm": 0.2968631386756897,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 2.4874,
      "step": 161
    },
    {
      "epoch": 0.022117550686053655,
      "grad_norm": 0.2822592556476593,
      "learning_rate": 9.549150281252633e-06,
      "loss": 2.4668,
      "step": 162
    },
    {
      "epoch": 0.02225407877670831,
      "grad_norm": 0.30913037061691284,
      "learning_rate": 9.068759265665384e-06,
      "loss": 2.6763,
      "step": 163
    },
    {
      "epoch": 0.02239060686736296,
      "grad_norm": 0.3001090884208679,
      "learning_rate": 8.599558442598998e-06,
      "loss": 2.4296,
      "step": 164
    },
    {
      "epoch": 0.02252713495801761,
      "grad_norm": 0.29255884885787964,
      "learning_rate": 8.141676086873572e-06,
      "loss": 2.3285,
      "step": 165
    },
    {
      "epoch": 0.022663663048672265,
      "grad_norm": 0.3152393698692322,
      "learning_rate": 7.695237378953223e-06,
      "loss": 2.3928,
      "step": 166
    },
    {
      "epoch": 0.022800191139326915,
      "grad_norm": 0.30669286847114563,
      "learning_rate": 7.260364370723044e-06,
      "loss": 2.4607,
      "step": 167
    },
    {
      "epoch": 0.02293671922998157,
      "grad_norm": 0.5392299890518188,
      "learning_rate": 6.837175952121306e-06,
      "loss": 2.2386,
      "step": 168
    },
    {
      "epoch": 0.02307324732063622,
      "grad_norm": 0.3007963001728058,
      "learning_rate": 6.425787818636131e-06,
      "loss": 2.5599,
      "step": 169
    },
    {
      "epoch": 0.023209775411290875,
      "grad_norm": 0.28646934032440186,
      "learning_rate": 6.026312439675552e-06,
      "loss": 2.4544,
      "step": 170
    },
    {
      "epoch": 0.023346303501945526,
      "grad_norm": 0.28788676857948303,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 2.5325,
      "step": 171
    },
    {
      "epoch": 0.023482831592600176,
      "grad_norm": 0.28152015805244446,
      "learning_rate": 5.263533508961827e-06,
      "loss": 2.5105,
      "step": 172
    },
    {
      "epoch": 0.02361935968325483,
      "grad_norm": 0.297216534614563,
      "learning_rate": 4.900438493352055e-06,
      "loss": 2.5582,
      "step": 173
    },
    {
      "epoch": 0.02375588777390948,
      "grad_norm": 0.2965959906578064,
      "learning_rate": 4.549673247541875e-06,
      "loss": 2.572,
      "step": 174
    },
    {
      "epoch": 0.023892415864564136,
      "grad_norm": 0.2938275635242462,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 2.4671,
      "step": 175
    },
    {
      "epoch": 0.024028943955218787,
      "grad_norm": 0.3325875699520111,
      "learning_rate": 3.885512251130763e-06,
      "loss": 2.4831,
      "step": 176
    },
    {
      "epoch": 0.024165472045873437,
      "grad_norm": 0.285214364528656,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 2.4587,
      "step": 177
    },
    {
      "epoch": 0.02430200013652809,
      "grad_norm": 0.32118937373161316,
      "learning_rate": 3.271776770026963e-06,
      "loss": 2.3991,
      "step": 178
    },
    {
      "epoch": 0.024438528227182742,
      "grad_norm": 0.30175623297691345,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 2.3989,
      "step": 179
    },
    {
      "epoch": 0.024575056317837397,
      "grad_norm": 0.3708553612232208,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 2.2207,
      "step": 180
    },
    {
      "epoch": 0.024711584408492047,
      "grad_norm": 0.2987494170665741,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 2.4346,
      "step": 181
    },
    {
      "epoch": 0.024848112499146698,
      "grad_norm": 0.32223793864250183,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 2.2944,
      "step": 182
    },
    {
      "epoch": 0.024984640589801353,
      "grad_norm": 0.36161771416664124,
      "learning_rate": 1.962316193157593e-06,
      "loss": 2.3166,
      "step": 183
    },
    {
      "epoch": 0.025121168680456003,
      "grad_norm": 0.40299317240715027,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 2.2524,
      "step": 184
    },
    {
      "epoch": 0.025257696771110658,
      "grad_norm": 0.32681384682655334,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 2.5744,
      "step": 185
    },
    {
      "epoch": 0.02539422486176531,
      "grad_norm": 0.335816890001297,
      "learning_rate": 1.333670137599713e-06,
      "loss": 2.5213,
      "step": 186
    },
    {
      "epoch": 0.02553075295241996,
      "grad_norm": 0.376728892326355,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 2.5252,
      "step": 187
    },
    {
      "epoch": 0.025667281043074613,
      "grad_norm": 0.3863675892353058,
      "learning_rate": 9.810017062595322e-07,
      "loss": 2.4194,
      "step": 188
    },
    {
      "epoch": 0.025803809133729264,
      "grad_norm": 0.38680794835090637,
      "learning_rate": 8.247462563808817e-07,
      "loss": 2.4215,
      "step": 189
    },
    {
      "epoch": 0.02594033722438392,
      "grad_norm": 0.39536648988723755,
      "learning_rate": 6.819348298638839e-07,
      "loss": 2.3774,
      "step": 190
    },
    {
      "epoch": 0.02607686531503857,
      "grad_norm": 0.39887604117393494,
      "learning_rate": 5.526064699265753e-07,
      "loss": 2.5914,
      "step": 191
    },
    {
      "epoch": 0.02621339340569322,
      "grad_norm": 0.4228583872318268,
      "learning_rate": 4.367965336512403e-07,
      "loss": 2.6961,
      "step": 192
    },
    {
      "epoch": 0.026349921496347874,
      "grad_norm": 0.3940587341785431,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 2.3309,
      "step": 193
    },
    {
      "epoch": 0.026486449587002525,
      "grad_norm": 0.42821937799453735,
      "learning_rate": 2.458548727494292e-07,
      "loss": 2.5391,
      "step": 194
    },
    {
      "epoch": 0.02662297767765718,
      "grad_norm": 0.42225223779678345,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 2.4793,
      "step": 195
    },
    {
      "epoch": 0.02675950576831183,
      "grad_norm": 0.4540916085243225,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 2.3683,
      "step": 196
    },
    {
      "epoch": 0.02689603385896648,
      "grad_norm": 0.4423125982284546,
      "learning_rate": 6.150154258476315e-08,
      "loss": 2.543,
      "step": 197
    },
    {
      "epoch": 0.027032561949621135,
      "grad_norm": 0.6300118565559387,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 2.5739,
      "step": 198
    },
    {
      "epoch": 0.027169090040275786,
      "grad_norm": 0.665462851524353,
      "learning_rate": 6.834750376549792e-09,
      "loss": 2.9246,
      "step": 199
    },
    {
      "epoch": 0.02730561813093044,
      "grad_norm": 0.9178174734115601,
      "learning_rate": 0.0,
      "loss": 2.2389,
      "step": 200
    },
    {
      "epoch": 0.02730561813093044,
      "eval_loss": 2.477691650390625,
      "eval_runtime": 365.3403,
      "eval_samples_per_second": 33.766,
      "eval_steps_per_second": 8.441,
      "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": 6.651170610846106e+16,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}