{
  "best_metric": 2.5915141105651855,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.008984826873617178,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 4.49241343680859e-05,
      "grad_norm": 7.704527854919434,
      "learning_rate": 1e-05,
      "loss": 11.6685,
      "step": 1
    },
    {
      "epoch": 4.49241343680859e-05,
      "eval_loss": 3.3177123069763184,
      "eval_runtime": 424.7102,
      "eval_samples_per_second": 88.272,
      "eval_steps_per_second": 22.069,
      "step": 1
    },
    {
      "epoch": 8.98482687361718e-05,
      "grad_norm": 9.044937133789062,
      "learning_rate": 2e-05,
      "loss": 12.2313,
      "step": 2
    },
    {
      "epoch": 0.00013477240310425767,
      "grad_norm": 8.883275032043457,
      "learning_rate": 3e-05,
      "loss": 12.1757,
      "step": 3
    },
    {
      "epoch": 0.0001796965374723436,
      "grad_norm": 8.799221992492676,
      "learning_rate": 4e-05,
      "loss": 11.9713,
      "step": 4
    },
    {
      "epoch": 0.00022462067184042948,
      "grad_norm": 7.929937839508057,
      "learning_rate": 5e-05,
      "loss": 12.6204,
      "step": 5
    },
    {
      "epoch": 0.00026954480620851535,
      "grad_norm": 8.440561294555664,
      "learning_rate": 6e-05,
      "loss": 11.012,
      "step": 6
    },
    {
      "epoch": 0.0003144689405766013,
      "grad_norm": 7.886547565460205,
      "learning_rate": 7e-05,
      "loss": 10.9689,
      "step": 7
    },
    {
      "epoch": 0.0003593930749446872,
      "grad_norm": 7.742301940917969,
      "learning_rate": 8e-05,
      "loss": 10.8692,
      "step": 8
    },
    {
      "epoch": 0.00040431720931277307,
      "grad_norm": 9.232806205749512,
      "learning_rate": 9e-05,
      "loss": 11.615,
      "step": 9
    },
    {
      "epoch": 0.00044924134368085896,
      "grad_norm": 9.356942176818848,
      "learning_rate": 0.0001,
      "loss": 11.7267,
      "step": 10
    },
    {
      "epoch": 0.0004941654780489448,
      "grad_norm": 8.897214889526367,
      "learning_rate": 9.999316524962345e-05,
      "loss": 11.7287,
      "step": 11
    },
    {
      "epoch": 0.0005390896124170307,
      "grad_norm": 8.982020378112793,
      "learning_rate": 9.997266286704631e-05,
      "loss": 11.2378,
      "step": 12
    },
    {
      "epoch": 0.0005840137467851166,
      "grad_norm": 8.378556251525879,
      "learning_rate": 9.993849845741524e-05,
      "loss": 11.2928,
      "step": 13
    },
    {
      "epoch": 0.0006289378811532026,
      "grad_norm": 8.130147933959961,
      "learning_rate": 9.989068136093873e-05,
      "loss": 10.8345,
      "step": 14
    },
    {
      "epoch": 0.0006738620155212885,
      "grad_norm": 8.747434616088867,
      "learning_rate": 9.98292246503335e-05,
      "loss": 11.3198,
      "step": 15
    },
    {
      "epoch": 0.0007187861498893744,
      "grad_norm": 7.390082359313965,
      "learning_rate": 9.975414512725057e-05,
      "loss": 10.2041,
      "step": 16
    },
    {
      "epoch": 0.0007637102842574603,
      "grad_norm": 7.369154930114746,
      "learning_rate": 9.966546331768191e-05,
      "loss": 10.7358,
      "step": 17
    },
    {
      "epoch": 0.0008086344186255461,
      "grad_norm": 7.815392017364502,
      "learning_rate": 9.956320346634876e-05,
      "loss": 10.8887,
      "step": 18
    },
    {
      "epoch": 0.000853558552993632,
      "grad_norm": 6.990930080413818,
      "learning_rate": 9.944739353007344e-05,
      "loss": 10.1232,
      "step": 19
    },
    {
      "epoch": 0.0008984826873617179,
      "grad_norm": 7.39527702331543,
      "learning_rate": 9.931806517013612e-05,
      "loss": 11.4211,
      "step": 20
    },
    {
      "epoch": 0.0009434068217298038,
      "grad_norm": 7.208984851837158,
      "learning_rate": 9.917525374361912e-05,
      "loss": 10.0958,
      "step": 21
    },
    {
      "epoch": 0.0009883309560978896,
      "grad_norm": 7.318514823913574,
      "learning_rate": 9.901899829374047e-05,
      "loss": 11.2377,
      "step": 22
    },
    {
      "epoch": 0.0010332550904659755,
      "grad_norm": 6.986603260040283,
      "learning_rate": 9.884934153917997e-05,
      "loss": 10.0796,
      "step": 23
    },
    {
      "epoch": 0.0010781792248340614,
      "grad_norm": 7.050904273986816,
      "learning_rate": 9.86663298624003e-05,
      "loss": 10.6751,
      "step": 24
    },
    {
      "epoch": 0.0011231033592021473,
      "grad_norm": 7.63284969329834,
      "learning_rate": 9.847001329696653e-05,
      "loss": 10.2685,
      "step": 25
    },
    {
      "epoch": 0.0011680274935702332,
      "grad_norm": 7.250881195068359,
      "learning_rate": 9.826044551386744e-05,
      "loss": 10.7424,
      "step": 26
    },
    {
      "epoch": 0.0012129516279383193,
      "grad_norm": 8.56254768371582,
      "learning_rate": 9.803768380684242e-05,
      "loss": 12.3304,
      "step": 27
    },
    {
      "epoch": 0.0012578757623064052,
      "grad_norm": 7.6560845375061035,
      "learning_rate": 9.780178907671789e-05,
      "loss": 9.874,
      "step": 28
    },
    {
      "epoch": 0.001302799896674491,
      "grad_norm": 7.455990314483643,
      "learning_rate": 9.755282581475769e-05,
      "loss": 11.0279,
      "step": 29
    },
    {
      "epoch": 0.001347724031042577,
      "grad_norm": 7.6860575675964355,
      "learning_rate": 9.729086208503174e-05,
      "loss": 10.4641,
      "step": 30
    },
    {
      "epoch": 0.0013926481654106628,
      "grad_norm": 7.540174961090088,
      "learning_rate": 9.701596950580806e-05,
      "loss": 9.6715,
      "step": 31
    },
    {
      "epoch": 0.0014375722997787487,
      "grad_norm": 7.6859588623046875,
      "learning_rate": 9.672822322997305e-05,
      "loss": 10.3341,
      "step": 32
    },
    {
      "epoch": 0.0014824964341468346,
      "grad_norm": 7.605548858642578,
      "learning_rate": 9.642770192448536e-05,
      "loss": 10.1505,
      "step": 33
    },
    {
      "epoch": 0.0015274205685149205,
      "grad_norm": 8.288911819458008,
      "learning_rate": 9.611448774886924e-05,
      "loss": 10.5214,
      "step": 34
    },
    {
      "epoch": 0.0015723447028830064,
      "grad_norm": 8.075105667114258,
      "learning_rate": 9.578866633275288e-05,
      "loss": 9.7556,
      "step": 35
    },
    {
      "epoch": 0.0016172688372510923,
      "grad_norm": 8.709887504577637,
      "learning_rate": 9.545032675245813e-05,
      "loss": 10.2588,
      "step": 36
    },
    {
      "epoch": 0.0016621929716191782,
      "grad_norm": 8.617378234863281,
      "learning_rate": 9.509956150664796e-05,
      "loss": 9.1325,
      "step": 37
    },
    {
      "epoch": 0.001707117105987264,
      "grad_norm": 8.220441818237305,
      "learning_rate": 9.473646649103818e-05,
      "loss": 10.2441,
      "step": 38
    },
    {
      "epoch": 0.00175204124035535,
      "grad_norm": 8.359474182128906,
      "learning_rate": 9.43611409721806e-05,
      "loss": 9.32,
      "step": 39
    },
    {
      "epoch": 0.0017969653747234359,
      "grad_norm": 8.244251251220703,
      "learning_rate": 9.397368756032445e-05,
      "loss": 9.466,
      "step": 40
    },
    {
      "epoch": 0.0018418895090915217,
      "grad_norm": 9.545090675354004,
      "learning_rate": 9.357421218136386e-05,
      "loss": 9.2706,
      "step": 41
    },
    {
      "epoch": 0.0018868136434596076,
      "grad_norm": 10.26917552947998,
      "learning_rate": 9.316282404787871e-05,
      "loss": 10.4483,
      "step": 42
    },
    {
      "epoch": 0.0019317377778276935,
      "grad_norm": 9.775924682617188,
      "learning_rate": 9.273963562927695e-05,
      "loss": 9.8433,
      "step": 43
    },
    {
      "epoch": 0.001976661912195779,
      "grad_norm": 10.140954971313477,
      "learning_rate": 9.230476262104677e-05,
      "loss": 8.8684,
      "step": 44
    },
    {
      "epoch": 0.0020215860465638653,
      "grad_norm": 11.226240158081055,
      "learning_rate": 9.185832391312644e-05,
      "loss": 9.7435,
      "step": 45
    },
    {
      "epoch": 0.002066510180931951,
      "grad_norm": 11.512527465820312,
      "learning_rate": 9.140044155740101e-05,
      "loss": 8.6754,
      "step": 46
    },
    {
      "epoch": 0.002111434315300037,
      "grad_norm": 14.677433013916016,
      "learning_rate": 9.093124073433463e-05,
      "loss": 10.7675,
      "step": 47
    },
    {
      "epoch": 0.0021563584496681228,
      "grad_norm": 15.949084281921387,
      "learning_rate": 9.045084971874738e-05,
      "loss": 10.85,
      "step": 48
    },
    {
      "epoch": 0.002201282584036209,
      "grad_norm": 17.321060180664062,
      "learning_rate": 8.995939984474624e-05,
      "loss": 11.9178,
      "step": 49
    },
    {
      "epoch": 0.0022462067184042945,
      "grad_norm": 34.57277297973633,
      "learning_rate": 8.945702546981969e-05,
      "loss": 14.001,
      "step": 50
    },
    {
      "epoch": 0.0022462067184042945,
      "eval_loss": 2.7662832736968994,
      "eval_runtime": 424.3836,
      "eval_samples_per_second": 88.34,
      "eval_steps_per_second": 22.086,
      "step": 50
    },
    {
      "epoch": 0.0022911308527723807,
      "grad_norm": 10.19485855102539,
      "learning_rate": 8.894386393810563e-05,
      "loss": 12.0763,
      "step": 51
    },
    {
      "epoch": 0.0023360549871404663,
      "grad_norm": 8.745186805725098,
      "learning_rate": 8.842005554284296e-05,
      "loss": 12.1067,
      "step": 52
    },
    {
      "epoch": 0.0023809791215085524,
      "grad_norm": 8.259309768676758,
      "learning_rate": 8.788574348801675e-05,
      "loss": 11.7137,
      "step": 53
    },
    {
      "epoch": 0.0024259032558766385,
      "grad_norm": 7.260303020477295,
      "learning_rate": 8.73410738492077e-05,
      "loss": 11.5238,
      "step": 54
    },
    {
      "epoch": 0.002470827390244724,
      "grad_norm": 6.761685848236084,
      "learning_rate": 8.678619553365659e-05,
      "loss": 11.58,
      "step": 55
    },
    {
      "epoch": 0.0025157515246128103,
      "grad_norm": 5.698697566986084,
      "learning_rate": 8.622126023955446e-05,
      "loss": 11.5472,
      "step": 56
    },
    {
      "epoch": 0.002560675658980896,
      "grad_norm": 5.167198181152344,
      "learning_rate": 8.564642241456986e-05,
      "loss": 11.3553,
      "step": 57
    },
    {
      "epoch": 0.002605599793348982,
      "grad_norm": 5.575362682342529,
      "learning_rate": 8.506183921362443e-05,
      "loss": 10.859,
      "step": 58
    },
    {
      "epoch": 0.0026505239277170678,
      "grad_norm": 6.121374130249023,
      "learning_rate": 8.44676704559283e-05,
      "loss": 11.1941,
      "step": 59
    },
    {
      "epoch": 0.002695448062085154,
      "grad_norm": 6.187102794647217,
      "learning_rate": 8.386407858128706e-05,
      "loss": 10.8339,
      "step": 60
    },
    {
      "epoch": 0.0027403721964532396,
      "grad_norm": 6.996326446533203,
      "learning_rate": 8.32512286056924e-05,
      "loss": 11.0405,
      "step": 61
    },
    {
      "epoch": 0.0027852963308213257,
      "grad_norm": 6.2773756980896,
      "learning_rate": 8.262928807620843e-05,
      "loss": 11.4055,
      "step": 62
    },
    {
      "epoch": 0.0028302204651894113,
      "grad_norm": 5.850597381591797,
      "learning_rate": 8.199842702516583e-05,
      "loss": 11.3701,
      "step": 63
    },
    {
      "epoch": 0.0028751445995574974,
      "grad_norm": 5.669310569763184,
      "learning_rate": 8.135881792367686e-05,
      "loss": 11.3048,
      "step": 64
    },
    {
      "epoch": 0.002920068733925583,
      "grad_norm": 5.014891147613525,
      "learning_rate": 8.07106356344834e-05,
      "loss": 11.5824,
      "step": 65
    },
    {
      "epoch": 0.0029649928682936692,
      "grad_norm": 4.967021942138672,
      "learning_rate": 8.005405736415126e-05,
      "loss": 10.4609,
      "step": 66
    },
    {
      "epoch": 0.003009917002661755,
      "grad_norm": 5.1634697914123535,
      "learning_rate": 7.938926261462366e-05,
      "loss": 10.3548,
      "step": 67
    },
    {
      "epoch": 0.003054841137029841,
      "grad_norm": 5.596578121185303,
      "learning_rate": 7.871643313414718e-05,
      "loss": 11.1188,
      "step": 68
    },
    {
      "epoch": 0.0030997652713979267,
      "grad_norm": 5.4398603439331055,
      "learning_rate": 7.803575286758364e-05,
      "loss": 11.1937,
      "step": 69
    },
    {
      "epoch": 0.003144689405766013,
      "grad_norm": 5.600924968719482,
      "learning_rate": 7.734740790612136e-05,
      "loss": 11.1241,
      "step": 70
    },
    {
      "epoch": 0.0031896135401340985,
      "grad_norm": 5.143842697143555,
      "learning_rate": 7.66515864363997e-05,
      "loss": 10.5599,
      "step": 71
    },
    {
      "epoch": 0.0032345376745021846,
      "grad_norm": 5.467100143432617,
      "learning_rate": 7.594847868906076e-05,
      "loss": 10.6663,
      "step": 72
    },
    {
      "epoch": 0.0032794618088702703,
      "grad_norm": 6.066835403442383,
      "learning_rate": 7.52382768867422e-05,
      "loss": 10.7799,
      "step": 73
    },
    {
      "epoch": 0.0033243859432383564,
      "grad_norm": 5.52231502532959,
      "learning_rate": 7.452117519152542e-05,
      "loss": 10.1481,
      "step": 74
    },
    {
      "epoch": 0.003369310077606442,
      "grad_norm": 6.379177093505859,
      "learning_rate": 7.379736965185368e-05,
      "loss": 10.2155,
      "step": 75
    },
    {
      "epoch": 0.003414234211974528,
      "grad_norm": 5.022770881652832,
      "learning_rate": 7.30670581489344e-05,
      "loss": 10.2953,
      "step": 76
    },
    {
      "epoch": 0.003459158346342614,
      "grad_norm": 5.834427833557129,
      "learning_rate": 7.233044034264034e-05,
      "loss": 11.272,
      "step": 77
    },
    {
      "epoch": 0.0035040824807107,
      "grad_norm": 5.810269355773926,
      "learning_rate": 7.158771761692464e-05,
      "loss": 10.079,
      "step": 78
    },
    {
      "epoch": 0.0035490066150787856,
      "grad_norm": 5.524205207824707,
      "learning_rate": 7.083909302476453e-05,
      "loss": 9.5685,
      "step": 79
    },
    {
      "epoch": 0.0035939307494468717,
      "grad_norm": 5.5137176513671875,
      "learning_rate": 7.008477123264848e-05,
      "loss": 10.5274,
      "step": 80
    },
    {
      "epoch": 0.0036388548838149574,
      "grad_norm": 6.164163589477539,
      "learning_rate": 6.932495846462261e-05,
      "loss": 9.7509,
      "step": 81
    },
    {
      "epoch": 0.0036837790181830435,
      "grad_norm": 6.019806861877441,
      "learning_rate": 6.855986244591104e-05,
      "loss": 10.082,
      "step": 82
    },
    {
      "epoch": 0.003728703152551129,
      "grad_norm": 6.2448201179504395,
      "learning_rate": 6.778969234612584e-05,
      "loss": 11.4841,
      "step": 83
    },
    {
      "epoch": 0.0037736272869192153,
      "grad_norm": 6.547962665557861,
      "learning_rate": 6.701465872208216e-05,
      "loss": 10.1931,
      "step": 84
    },
    {
      "epoch": 0.003818551421287301,
      "grad_norm": 6.18972110748291,
      "learning_rate": 6.623497346023418e-05,
      "loss": 9.4638,
      "step": 85
    },
    {
      "epoch": 0.003863475555655387,
      "grad_norm": 6.580770969390869,
      "learning_rate": 6.545084971874738e-05,
      "loss": 11.16,
      "step": 86
    },
    {
      "epoch": 0.003908399690023473,
      "grad_norm": 6.748318195343018,
      "learning_rate": 6.466250186922325e-05,
      "loss": 10.6037,
      "step": 87
    },
    {
      "epoch": 0.003953323824391558,
      "grad_norm": 6.5974040031433105,
      "learning_rate": 6.387014543809223e-05,
      "loss": 10.9843,
      "step": 88
    },
    {
      "epoch": 0.0039982479587596445,
      "grad_norm": 7.050424575805664,
      "learning_rate": 6.307399704769099e-05,
      "loss": 9.8075,
      "step": 89
    },
    {
      "epoch": 0.004043172093127731,
      "grad_norm": 7.172850131988525,
      "learning_rate": 6.227427435703997e-05,
      "loss": 10.382,
      "step": 90
    },
    {
      "epoch": 0.004088096227495817,
      "grad_norm": 7.529332637786865,
      "learning_rate": 6.147119600233758e-05,
      "loss": 10.2463,
      "step": 91
    },
    {
      "epoch": 0.004133020361863902,
      "grad_norm": 8.266640663146973,
      "learning_rate": 6.066498153718735e-05,
      "loss": 9.7778,
      "step": 92
    },
    {
      "epoch": 0.004177944496231988,
      "grad_norm": 8.065340995788574,
      "learning_rate": 5.985585137257401e-05,
      "loss": 10.3596,
      "step": 93
    },
    {
      "epoch": 0.004222868630600074,
      "grad_norm": 9.381607055664062,
      "learning_rate": 5.90440267166055e-05,
      "loss": 12.3793,
      "step": 94
    },
    {
      "epoch": 0.00426779276496816,
      "grad_norm": 8.951251983642578,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 10.5589,
      "step": 95
    },
    {
      "epoch": 0.0043127168993362455,
      "grad_norm": 9.510485649108887,
      "learning_rate": 5.74131823855921e-05,
      "loss": 8.9771,
      "step": 96
    },
    {
      "epoch": 0.004357641033704332,
      "grad_norm": 10.000716209411621,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 8.6483,
      "step": 97
    },
    {
      "epoch": 0.004402565168072418,
      "grad_norm": 12.402448654174805,
      "learning_rate": 5.577423184847932e-05,
      "loss": 11.0121,
      "step": 98
    },
    {
      "epoch": 0.004447489302440504,
      "grad_norm": 17.327621459960938,
      "learning_rate": 5.495227651252315e-05,
      "loss": 10.9437,
      "step": 99
    },
    {
      "epoch": 0.004492413436808589,
      "grad_norm": 28.959142684936523,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 16.3508,
      "step": 100
    },
    {
      "epoch": 0.004492413436808589,
      "eval_loss": 2.6626319885253906,
      "eval_runtime": 425.0812,
      "eval_samples_per_second": 88.195,
      "eval_steps_per_second": 22.05,
      "step": 100
    },
    {
      "epoch": 0.004537337571176675,
      "grad_norm": 5.03517484664917,
      "learning_rate": 5.330452921628497e-05,
      "loss": 11.8379,
      "step": 101
    },
    {
      "epoch": 0.004582261705544761,
      "grad_norm": 5.368809223175049,
      "learning_rate": 5.247918773366112e-05,
      "loss": 12.2468,
      "step": 102
    },
    {
      "epoch": 0.004627185839912847,
      "grad_norm": 5.362026691436768,
      "learning_rate": 5.165316846586541e-05,
      "loss": 10.919,
      "step": 103
    },
    {
      "epoch": 0.004672109974280933,
      "grad_norm": 5.336542129516602,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 11.2734,
      "step": 104
    },
    {
      "epoch": 0.004717034108649019,
      "grad_norm": 5.056530952453613,
      "learning_rate": 5e-05,
      "loss": 11.3126,
      "step": 105
    },
    {
      "epoch": 0.004761958243017105,
      "grad_norm": 5.093935489654541,
      "learning_rate": 4.917330276168208e-05,
      "loss": 11.3321,
      "step": 106
    },
    {
      "epoch": 0.004806882377385191,
      "grad_norm": 4.618561267852783,
      "learning_rate": 4.834683153413459e-05,
      "loss": 10.4597,
      "step": 107
    },
    {
      "epoch": 0.004851806511753277,
      "grad_norm": 4.551071643829346,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 11.2809,
      "step": 108
    },
    {
      "epoch": 0.004896730646121362,
      "grad_norm": 4.672351837158203,
      "learning_rate": 4.669547078371504e-05,
      "loss": 10.801,
      "step": 109
    },
    {
      "epoch": 0.004941654780489448,
      "grad_norm": 4.537178993225098,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 10.6807,
      "step": 110
    },
    {
      "epoch": 0.0049865789148575345,
      "grad_norm": 4.401553630828857,
      "learning_rate": 4.504772348747687e-05,
      "loss": 11.837,
      "step": 111
    },
    {
      "epoch": 0.005031503049225621,
      "grad_norm": 4.264166831970215,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 10.3927,
      "step": 112
    },
    {
      "epoch": 0.005076427183593706,
      "grad_norm": 4.5442047119140625,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 11.1291,
      "step": 113
    },
    {
      "epoch": 0.005121351317961792,
      "grad_norm": 4.373701095581055,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 10.8542,
      "step": 114
    },
    {
      "epoch": 0.005166275452329878,
      "grad_norm": 4.6788482666015625,
      "learning_rate": 4.17702704859633e-05,
      "loss": 11.7209,
      "step": 115
    },
    {
      "epoch": 0.005211199586697964,
      "grad_norm": 4.559150218963623,
      "learning_rate": 4.095597328339452e-05,
      "loss": 10.8493,
      "step": 116
    },
    {
      "epoch": 0.0052561237210660494,
      "grad_norm": 4.511398792266846,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 10.4847,
      "step": 117
    },
    {
      "epoch": 0.0053010478554341356,
      "grad_norm": 5.186944961547852,
      "learning_rate": 3.933501846281267e-05,
      "loss": 10.9482,
      "step": 118
    },
    {
      "epoch": 0.005345971989802222,
      "grad_norm": 5.235041618347168,
      "learning_rate": 3.852880399766243e-05,
      "loss": 11.0369,
      "step": 119
    },
    {
      "epoch": 0.005390896124170308,
      "grad_norm": 4.865777969360352,
      "learning_rate": 3.772572564296005e-05,
      "loss": 10.7916,
      "step": 120
    },
    {
      "epoch": 0.005435820258538393,
      "grad_norm": 5.442884922027588,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 11.3708,
      "step": 121
    },
    {
      "epoch": 0.005480744392906479,
      "grad_norm": 4.7792229652404785,
      "learning_rate": 3.612985456190778e-05,
      "loss": 11.045,
      "step": 122
    },
    {
      "epoch": 0.005525668527274565,
      "grad_norm": 4.93920373916626,
      "learning_rate": 3.533749813077677e-05,
      "loss": 10.2748,
      "step": 123
    },
    {
      "epoch": 0.005570592661642651,
      "grad_norm": 5.340206146240234,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 10.6015,
      "step": 124
    },
    {
      "epoch": 0.005615516796010737,
      "grad_norm": 4.95137882232666,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 10.1575,
      "step": 125
    },
    {
      "epoch": 0.005660440930378823,
      "grad_norm": 4.919768333435059,
      "learning_rate": 3.298534127791785e-05,
      "loss": 9.4554,
      "step": 126
    },
    {
      "epoch": 0.005705365064746909,
      "grad_norm": 5.368086338043213,
      "learning_rate": 3.221030765387417e-05,
      "loss": 9.7716,
      "step": 127
    },
    {
      "epoch": 0.005750289199114995,
      "grad_norm": 5.496703147888184,
      "learning_rate": 3.144013755408895e-05,
      "loss": 9.7555,
      "step": 128
    },
    {
      "epoch": 0.00579521333348308,
      "grad_norm": 5.635049819946289,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 9.5854,
      "step": 129
    },
    {
      "epoch": 0.005840137467851166,
      "grad_norm": 6.144560813903809,
      "learning_rate": 2.991522876735154e-05,
      "loss": 11.2076,
      "step": 130
    },
    {
      "epoch": 0.005885061602219252,
      "grad_norm": 5.222030162811279,
      "learning_rate": 2.916090697523549e-05,
      "loss": 10.2521,
      "step": 131
    },
    {
      "epoch": 0.0059299857365873385,
      "grad_norm": 5.709193229675293,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 9.2299,
      "step": 132
    },
    {
      "epoch": 0.005974909870955424,
      "grad_norm": 6.369167804718018,
      "learning_rate": 2.766955965735968e-05,
      "loss": 10.2951,
      "step": 133
    },
    {
      "epoch": 0.00601983400532351,
      "grad_norm": 5.345949649810791,
      "learning_rate": 2.693294185106562e-05,
      "loss": 9.528,
      "step": 134
    },
    {
      "epoch": 0.006064758139691596,
      "grad_norm": 5.7703752517700195,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 10.0742,
      "step": 135
    },
    {
      "epoch": 0.006109682274059682,
      "grad_norm": 6.811999797821045,
      "learning_rate": 2.547882480847461e-05,
      "loss": 10.3258,
      "step": 136
    },
    {
      "epoch": 0.006154606408427767,
      "grad_norm": 6.249923229217529,
      "learning_rate": 2.476172311325783e-05,
      "loss": 9.2141,
      "step": 137
    },
    {
      "epoch": 0.006199530542795853,
      "grad_norm": 6.040198802947998,
      "learning_rate": 2.405152131093926e-05,
      "loss": 10.075,
      "step": 138
    },
    {
      "epoch": 0.0062444546771639395,
      "grad_norm": 6.58219051361084,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 10.0423,
      "step": 139
    },
    {
      "epoch": 0.006289378811532026,
      "grad_norm": 7.655426502227783,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 10.7645,
      "step": 140
    },
    {
      "epoch": 0.006334302945900111,
      "grad_norm": 6.7975945472717285,
      "learning_rate": 2.196424713241637e-05,
      "loss": 8.926,
      "step": 141
    },
    {
      "epoch": 0.006379227080268197,
      "grad_norm": 6.947317600250244,
      "learning_rate": 2.128356686585282e-05,
      "loss": 9.5132,
      "step": 142
    },
    {
      "epoch": 0.006424151214636283,
      "grad_norm": 7.4242329597473145,
      "learning_rate": 2.061073738537635e-05,
      "loss": 9.6655,
      "step": 143
    },
    {
      "epoch": 0.006469075349004369,
      "grad_norm": 8.75672435760498,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 10.6343,
      "step": 144
    },
    {
      "epoch": 0.006513999483372454,
      "grad_norm": 8.709526062011719,
      "learning_rate": 1.928936436551661e-05,
      "loss": 8.8825,
      "step": 145
    },
    {
      "epoch": 0.0065589236177405405,
      "grad_norm": 11.966184616088867,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 11.2628,
      "step": 146
    },
    {
      "epoch": 0.006603847752108627,
      "grad_norm": 9.536019325256348,
      "learning_rate": 1.800157297483417e-05,
      "loss": 10.0251,
      "step": 147
    },
    {
      "epoch": 0.006648771886476713,
      "grad_norm": 14.3245210647583,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 9.713,
      "step": 148
    },
    {
      "epoch": 0.006693696020844798,
      "grad_norm": 16.303739547729492,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 10.6701,
      "step": 149
    },
    {
      "epoch": 0.006738620155212884,
      "grad_norm": 28.5744571685791,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 13.0938,
      "step": 150
    },
    {
      "epoch": 0.006738620155212884,
      "eval_loss": 2.60650372505188,
      "eval_runtime": 424.1197,
      "eval_samples_per_second": 88.395,
      "eval_steps_per_second": 22.1,
      "step": 150
    },
    {
      "epoch": 0.00678354428958097,
      "grad_norm": 3.7291252613067627,
      "learning_rate": 1.553232954407171e-05,
      "loss": 10.3543,
      "step": 151
    },
    {
      "epoch": 0.006828468423949056,
      "grad_norm": 4.018054962158203,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 10.737,
      "step": 152
    },
    {
      "epoch": 0.0068733925583171415,
      "grad_norm": 4.267710208892822,
      "learning_rate": 1.435357758543015e-05,
      "loss": 11.0661,
      "step": 153
    },
    {
      "epoch": 0.006918316692685228,
      "grad_norm": 4.368105888366699,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 11.3082,
      "step": 154
    },
    {
      "epoch": 0.006963240827053314,
      "grad_norm": 4.373310565948486,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 11.525,
      "step": 155
    },
    {
      "epoch": 0.0070081649614214,
      "grad_norm": 4.366888046264648,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 11.4385,
      "step": 156
    },
    {
      "epoch": 0.007053089095789485,
      "grad_norm": 4.487154006958008,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 11.7924,
      "step": 157
    },
    {
      "epoch": 0.007098013230157571,
      "grad_norm": 4.255626678466797,
      "learning_rate": 1.157994445715706e-05,
      "loss": 10.429,
      "step": 158
    },
    {
      "epoch": 0.007142937364525657,
      "grad_norm": 4.393832206726074,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 10.5053,
      "step": 159
    },
    {
      "epoch": 0.007187861498893743,
      "grad_norm": 4.706411838531494,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 10.9192,
      "step": 160
    },
    {
      "epoch": 0.0072327856332618295,
      "grad_norm": 4.519069671630859,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 11.0699,
      "step": 161
    },
    {
      "epoch": 0.007277709767629915,
      "grad_norm": 4.460075378417969,
      "learning_rate": 9.549150281252633e-06,
      "loss": 10.9185,
      "step": 162
    },
    {
      "epoch": 0.007322633901998001,
      "grad_norm": 4.843297481536865,
      "learning_rate": 9.068759265665384e-06,
      "loss": 10.3561,
      "step": 163
    },
    {
      "epoch": 0.007367558036366087,
      "grad_norm": 4.548454284667969,
      "learning_rate": 8.599558442598998e-06,
      "loss": 10.1688,
      "step": 164
    },
    {
      "epoch": 0.007412482170734173,
      "grad_norm": 4.506471633911133,
      "learning_rate": 8.141676086873572e-06,
      "loss": 10.3788,
      "step": 165
    },
    {
      "epoch": 0.007457406305102258,
      "grad_norm": 4.909854888916016,
      "learning_rate": 7.695237378953223e-06,
      "loss": 11.1137,
      "step": 166
    },
    {
      "epoch": 0.007502330439470344,
      "grad_norm": 4.584915637969971,
      "learning_rate": 7.260364370723044e-06,
      "loss": 10.4652,
      "step": 167
    },
    {
      "epoch": 0.0075472545738384305,
      "grad_norm": 4.439782619476318,
      "learning_rate": 6.837175952121306e-06,
      "loss": 9.3493,
      "step": 168
    },
    {
      "epoch": 0.007592178708206517,
      "grad_norm": 4.64847469329834,
      "learning_rate": 6.425787818636131e-06,
      "loss": 10.5949,
      "step": 169
    },
    {
      "epoch": 0.007637102842574602,
      "grad_norm": 4.924140453338623,
      "learning_rate": 6.026312439675552e-06,
      "loss": 10.2094,
      "step": 170
    },
    {
      "epoch": 0.007682026976942688,
      "grad_norm": 5.154047012329102,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 10.731,
      "step": 171
    },
    {
      "epoch": 0.007726951111310774,
      "grad_norm": 5.032504081726074,
      "learning_rate": 5.263533508961827e-06,
      "loss": 10.6782,
      "step": 172
    },
    {
      "epoch": 0.00777187524567886,
      "grad_norm": 4.745563507080078,
      "learning_rate": 4.900438493352055e-06,
      "loss": 10.2672,
      "step": 173
    },
    {
      "epoch": 0.007816799380046946,
      "grad_norm": 5.153327465057373,
      "learning_rate": 4.549673247541875e-06,
      "loss": 11.1209,
      "step": 174
    },
    {
      "epoch": 0.007861723514415032,
      "grad_norm": 5.027349948883057,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 10.276,
      "step": 175
    },
    {
      "epoch": 0.007906647648783117,
      "grad_norm": 5.011510848999023,
      "learning_rate": 3.885512251130763e-06,
      "loss": 10.4033,
      "step": 176
    },
    {
      "epoch": 0.007951571783151203,
      "grad_norm": 5.007534980773926,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 9.8234,
      "step": 177
    },
    {
      "epoch": 0.007996495917519289,
      "grad_norm": 5.256812572479248,
      "learning_rate": 3.271776770026963e-06,
      "loss": 10.3819,
      "step": 178
    },
    {
      "epoch": 0.008041420051887375,
      "grad_norm": 5.25607442855835,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 10.3275,
      "step": 179
    },
    {
      "epoch": 0.008086344186255461,
      "grad_norm": 5.792912006378174,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 11.3204,
      "step": 180
    },
    {
      "epoch": 0.008131268320623547,
      "grad_norm": 5.476194858551025,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 10.4698,
      "step": 181
    },
    {
      "epoch": 0.008176192454991633,
      "grad_norm": 4.973291873931885,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 9.3087,
      "step": 182
    },
    {
      "epoch": 0.00822111658935972,
      "grad_norm": 5.625941753387451,
      "learning_rate": 1.962316193157593e-06,
      "loss": 8.8482,
      "step": 183
    },
    {
      "epoch": 0.008266040723727804,
      "grad_norm": 5.804088592529297,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 10.804,
      "step": 184
    },
    {
      "epoch": 0.00831096485809589,
      "grad_norm": 5.885905742645264,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 9.9611,
      "step": 185
    },
    {
      "epoch": 0.008355888992463976,
      "grad_norm": 6.2364888191223145,
      "learning_rate": 1.333670137599713e-06,
      "loss": 11.005,
      "step": 186
    },
    {
      "epoch": 0.008400813126832062,
      "grad_norm": 5.6353020668029785,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 9.6686,
      "step": 187
    },
    {
      "epoch": 0.008445737261200148,
      "grad_norm": 6.8391218185424805,
      "learning_rate": 9.810017062595322e-07,
      "loss": 10.2593,
      "step": 188
    },
    {
      "epoch": 0.008490661395568234,
      "grad_norm": 6.948604583740234,
      "learning_rate": 8.247462563808817e-07,
      "loss": 10.0941,
      "step": 189
    },
    {
      "epoch": 0.00853558552993632,
      "grad_norm": 7.298179626464844,
      "learning_rate": 6.819348298638839e-07,
      "loss": 10.9747,
      "step": 190
    },
    {
      "epoch": 0.008580509664304407,
      "grad_norm": 7.452898025512695,
      "learning_rate": 5.526064699265753e-07,
      "loss": 10.5757,
      "step": 191
    },
    {
      "epoch": 0.008625433798672491,
      "grad_norm": 8.143848419189453,
      "learning_rate": 4.367965336512403e-07,
      "loss": 10.6709,
      "step": 192
    },
    {
      "epoch": 0.008670357933040577,
      "grad_norm": 7.728236198425293,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 9.5987,
      "step": 193
    },
    {
      "epoch": 0.008715282067408663,
      "grad_norm": 7.441531658172607,
      "learning_rate": 2.458548727494292e-07,
      "loss": 9.425,
      "step": 194
    },
    {
      "epoch": 0.00876020620177675,
      "grad_norm": 8.133403778076172,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 10.581,
      "step": 195
    },
    {
      "epoch": 0.008805130336144835,
      "grad_norm": 11.413058280944824,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 11.4768,
      "step": 196
    },
    {
      "epoch": 0.008850054470512922,
      "grad_norm": 10.818037986755371,
      "learning_rate": 6.150154258476315e-08,
      "loss": 10.5721,
      "step": 197
    },
    {
      "epoch": 0.008894978604881008,
      "grad_norm": 12.806747436523438,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 10.7664,
      "step": 198
    },
    {
      "epoch": 0.008939902739249094,
      "grad_norm": 15.246770858764648,
      "learning_rate": 6.834750376549792e-09,
      "loss": 11.3166,
      "step": 199
    },
    {
      "epoch": 0.008984826873617178,
      "grad_norm": 27.020919799804688,
      "learning_rate": 0.0,
      "loss": 13.6054,
      "step": 200
    },
    {
      "epoch": 0.008984826873617178,
      "eval_loss": 2.5915141105651855,
      "eval_runtime": 424.3026,
      "eval_samples_per_second": 88.357,
      "eval_steps_per_second": 22.09,
      "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": 1.30495065096192e+16,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}