{
  "best_metric": 1.3869588375091553,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 1.0624169986719787,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.005312084993359893,
      "grad_norm": 1.9961893558502197,
      "learning_rate": 1e-05,
      "loss": 1.4674,
      "step": 1
    },
    {
      "epoch": 0.005312084993359893,
      "eval_loss": 2.1604528427124023,
      "eval_runtime": 7.2105,
      "eval_samples_per_second": 43.964,
      "eval_steps_per_second": 11.095,
      "step": 1
    },
    {
      "epoch": 0.010624169986719787,
      "grad_norm": 2.7986509799957275,
      "learning_rate": 2e-05,
      "loss": 1.7848,
      "step": 2
    },
    {
      "epoch": 0.01593625498007968,
      "grad_norm": 3.2918195724487305,
      "learning_rate": 3e-05,
      "loss": 1.8963,
      "step": 3
    },
    {
      "epoch": 0.021248339973439574,
      "grad_norm": 3.381787061691284,
      "learning_rate": 4e-05,
      "loss": 1.9838,
      "step": 4
    },
    {
      "epoch": 0.02656042496679947,
      "grad_norm": 3.0417654514312744,
      "learning_rate": 5e-05,
      "loss": 1.8711,
      "step": 5
    },
    {
      "epoch": 0.03187250996015936,
      "grad_norm": 3.1310834884643555,
      "learning_rate": 6e-05,
      "loss": 1.906,
      "step": 6
    },
    {
      "epoch": 0.03718459495351926,
      "grad_norm": 3.5859055519104004,
      "learning_rate": 7e-05,
      "loss": 1.9728,
      "step": 7
    },
    {
      "epoch": 0.04249667994687915,
      "grad_norm": 3.6353094577789307,
      "learning_rate": 8e-05,
      "loss": 2.2269,
      "step": 8
    },
    {
      "epoch": 0.04780876494023904,
      "grad_norm": 3.999406337738037,
      "learning_rate": 9e-05,
      "loss": 1.9924,
      "step": 9
    },
    {
      "epoch": 0.05312084993359894,
      "grad_norm": 4.358706474304199,
      "learning_rate": 0.0001,
      "loss": 2.1834,
      "step": 10
    },
    {
      "epoch": 0.05843293492695883,
      "grad_norm": 4.461287498474121,
      "learning_rate": 9.999316524962345e-05,
      "loss": 2.0307,
      "step": 11
    },
    {
      "epoch": 0.06374501992031872,
      "grad_norm": 4.280468463897705,
      "learning_rate": 9.997266286704631e-05,
      "loss": 1.9541,
      "step": 12
    },
    {
      "epoch": 0.06905710491367861,
      "grad_norm": 4.736527919769287,
      "learning_rate": 9.993849845741524e-05,
      "loss": 2.2975,
      "step": 13
    },
    {
      "epoch": 0.07436918990703852,
      "grad_norm": 4.972255229949951,
      "learning_rate": 9.989068136093873e-05,
      "loss": 2.4966,
      "step": 14
    },
    {
      "epoch": 0.0796812749003984,
      "grad_norm": 4.466799259185791,
      "learning_rate": 9.98292246503335e-05,
      "loss": 1.7539,
      "step": 15
    },
    {
      "epoch": 0.0849933598937583,
      "grad_norm": 3.188546657562256,
      "learning_rate": 9.975414512725057e-05,
      "loss": 1.7978,
      "step": 16
    },
    {
      "epoch": 0.0903054448871182,
      "grad_norm": 2.740100383758545,
      "learning_rate": 9.966546331768191e-05,
      "loss": 1.4716,
      "step": 17
    },
    {
      "epoch": 0.09561752988047809,
      "grad_norm": 3.1012604236602783,
      "learning_rate": 9.956320346634876e-05,
      "loss": 1.7421,
      "step": 18
    },
    {
      "epoch": 0.10092961487383798,
      "grad_norm": 3.222794771194458,
      "learning_rate": 9.944739353007344e-05,
      "loss": 1.6516,
      "step": 19
    },
    {
      "epoch": 0.10624169986719788,
      "grad_norm": 2.949683427810669,
      "learning_rate": 9.931806517013612e-05,
      "loss": 1.5846,
      "step": 20
    },
    {
      "epoch": 0.11155378486055777,
      "grad_norm": 3.35553240776062,
      "learning_rate": 9.917525374361912e-05,
      "loss": 1.7396,
      "step": 21
    },
    {
      "epoch": 0.11686586985391766,
      "grad_norm": 3.088747501373291,
      "learning_rate": 9.901899829374047e-05,
      "loss": 1.7153,
      "step": 22
    },
    {
      "epoch": 0.12217795484727756,
      "grad_norm": 2.7159929275512695,
      "learning_rate": 9.884934153917997e-05,
      "loss": 1.4142,
      "step": 23
    },
    {
      "epoch": 0.12749003984063745,
      "grad_norm": 2.6091272830963135,
      "learning_rate": 9.86663298624003e-05,
      "loss": 1.601,
      "step": 24
    },
    {
      "epoch": 0.13280212483399734,
      "grad_norm": 2.3251020908355713,
      "learning_rate": 9.847001329696653e-05,
      "loss": 1.5346,
      "step": 25
    },
    {
      "epoch": 0.13811420982735723,
      "grad_norm": 2.717583656311035,
      "learning_rate": 9.826044551386744e-05,
      "loss": 1.5922,
      "step": 26
    },
    {
      "epoch": 0.14342629482071714,
      "grad_norm": 2.178107500076294,
      "learning_rate": 9.803768380684242e-05,
      "loss": 1.4873,
      "step": 27
    },
    {
      "epoch": 0.14873837981407703,
      "grad_norm": 2.3208775520324707,
      "learning_rate": 9.780178907671789e-05,
      "loss": 1.5378,
      "step": 28
    },
    {
      "epoch": 0.15405046480743692,
      "grad_norm": 2.824413299560547,
      "learning_rate": 9.755282581475769e-05,
      "loss": 1.5492,
      "step": 29
    },
    {
      "epoch": 0.1593625498007968,
      "grad_norm": 2.6755924224853516,
      "learning_rate": 9.729086208503174e-05,
      "loss": 1.4338,
      "step": 30
    },
    {
      "epoch": 0.1646746347941567,
      "grad_norm": 2.3994293212890625,
      "learning_rate": 9.701596950580806e-05,
      "loss": 1.4735,
      "step": 31
    },
    {
      "epoch": 0.1699867197875166,
      "grad_norm": 3.284780740737915,
      "learning_rate": 9.672822322997305e-05,
      "loss": 1.7002,
      "step": 32
    },
    {
      "epoch": 0.1752988047808765,
      "grad_norm": 2.3807883262634277,
      "learning_rate": 9.642770192448536e-05,
      "loss": 1.4379,
      "step": 33
    },
    {
      "epoch": 0.1806108897742364,
      "grad_norm": 2.436025381088257,
      "learning_rate": 9.611448774886924e-05,
      "loss": 1.3348,
      "step": 34
    },
    {
      "epoch": 0.18592297476759628,
      "grad_norm": 2.633558511734009,
      "learning_rate": 9.578866633275288e-05,
      "loss": 1.5603,
      "step": 35
    },
    {
      "epoch": 0.19123505976095617,
      "grad_norm": 2.8258442878723145,
      "learning_rate": 9.545032675245813e-05,
      "loss": 1.8494,
      "step": 36
    },
    {
      "epoch": 0.19654714475431606,
      "grad_norm": 2.2452099323272705,
      "learning_rate": 9.509956150664796e-05,
      "loss": 1.6173,
      "step": 37
    },
    {
      "epoch": 0.20185922974767595,
      "grad_norm": 2.274542808532715,
      "learning_rate": 9.473646649103818e-05,
      "loss": 1.3304,
      "step": 38
    },
    {
      "epoch": 0.20717131474103587,
      "grad_norm": 2.4645605087280273,
      "learning_rate": 9.43611409721806e-05,
      "loss": 1.4591,
      "step": 39
    },
    {
      "epoch": 0.21248339973439576,
      "grad_norm": 2.094710350036621,
      "learning_rate": 9.397368756032445e-05,
      "loss": 1.4263,
      "step": 40
    },
    {
      "epoch": 0.21779548472775564,
      "grad_norm": 2.7573795318603516,
      "learning_rate": 9.357421218136386e-05,
      "loss": 1.4008,
      "step": 41
    },
    {
      "epoch": 0.22310756972111553,
      "grad_norm": 2.463797092437744,
      "learning_rate": 9.316282404787871e-05,
      "loss": 1.5969,
      "step": 42
    },
    {
      "epoch": 0.22841965471447542,
      "grad_norm": 2.386338233947754,
      "learning_rate": 9.273963562927695e-05,
      "loss": 1.3936,
      "step": 43
    },
    {
      "epoch": 0.2337317397078353,
      "grad_norm": 2.6396546363830566,
      "learning_rate": 9.230476262104677e-05,
      "loss": 1.5281,
      "step": 44
    },
    {
      "epoch": 0.23904382470119523,
      "grad_norm": 3.105961322784424,
      "learning_rate": 9.185832391312644e-05,
      "loss": 1.682,
      "step": 45
    },
    {
      "epoch": 0.24435590969455512,
      "grad_norm": 3.3824150562286377,
      "learning_rate": 9.140044155740101e-05,
      "loss": 1.6895,
      "step": 46
    },
    {
      "epoch": 0.249667994687915,
      "grad_norm": 4.8954386711120605,
      "learning_rate": 9.093124073433463e-05,
      "loss": 2.2261,
      "step": 47
    },
    {
      "epoch": 0.2549800796812749,
      "grad_norm": 1.651041865348816,
      "learning_rate": 9.045084971874738e-05,
      "loss": 1.5351,
      "step": 48
    },
    {
      "epoch": 0.2602921646746348,
      "grad_norm": 1.8729718923568726,
      "learning_rate": 8.995939984474624e-05,
      "loss": 1.543,
      "step": 49
    },
    {
      "epoch": 0.2656042496679947,
      "grad_norm": 1.6984103918075562,
      "learning_rate": 8.945702546981969e-05,
      "loss": 1.4056,
      "step": 50
    },
    {
      "epoch": 0.2656042496679947,
      "eval_loss": 1.5304877758026123,
      "eval_runtime": 7.256,
      "eval_samples_per_second": 43.688,
      "eval_steps_per_second": 11.025,
      "step": 50
    },
    {
      "epoch": 0.27091633466135456,
      "grad_norm": 1.9310680627822876,
      "learning_rate": 8.894386393810563e-05,
      "loss": 1.6063,
      "step": 51
    },
    {
      "epoch": 0.27622841965471445,
      "grad_norm": 2.0711851119995117,
      "learning_rate": 8.842005554284296e-05,
      "loss": 1.621,
      "step": 52
    },
    {
      "epoch": 0.2815405046480744,
      "grad_norm": 2.1960079669952393,
      "learning_rate": 8.788574348801675e-05,
      "loss": 1.7252,
      "step": 53
    },
    {
      "epoch": 0.2868525896414343,
      "grad_norm": 2.22192645072937,
      "learning_rate": 8.73410738492077e-05,
      "loss": 1.6359,
      "step": 54
    },
    {
      "epoch": 0.2921646746347942,
      "grad_norm": 2.208574056625366,
      "learning_rate": 8.678619553365659e-05,
      "loss": 1.8207,
      "step": 55
    },
    {
      "epoch": 0.29747675962815406,
      "grad_norm": 2.642925262451172,
      "learning_rate": 8.622126023955446e-05,
      "loss": 1.8832,
      "step": 56
    },
    {
      "epoch": 0.30278884462151395,
      "grad_norm": 2.668045997619629,
      "learning_rate": 8.564642241456986e-05,
      "loss": 2.0417,
      "step": 57
    },
    {
      "epoch": 0.30810092961487384,
      "grad_norm": 3.248166084289551,
      "learning_rate": 8.506183921362443e-05,
      "loss": 2.2328,
      "step": 58
    },
    {
      "epoch": 0.31341301460823373,
      "grad_norm": 4.173149585723877,
      "learning_rate": 8.44676704559283e-05,
      "loss": 2.5463,
      "step": 59
    },
    {
      "epoch": 0.3187250996015936,
      "grad_norm": 2.323002338409424,
      "learning_rate": 8.386407858128706e-05,
      "loss": 1.4174,
      "step": 60
    },
    {
      "epoch": 0.3240371845949535,
      "grad_norm": 2.4155280590057373,
      "learning_rate": 8.32512286056924e-05,
      "loss": 1.3838,
      "step": 61
    },
    {
      "epoch": 0.3293492695883134,
      "grad_norm": 2.469053268432617,
      "learning_rate": 8.262928807620843e-05,
      "loss": 1.6345,
      "step": 62
    },
    {
      "epoch": 0.3346613545816733,
      "grad_norm": 2.412416458129883,
      "learning_rate": 8.199842702516583e-05,
      "loss": 1.6597,
      "step": 63
    },
    {
      "epoch": 0.3399734395750332,
      "grad_norm": 2.0266129970550537,
      "learning_rate": 8.135881792367686e-05,
      "loss": 1.4379,
      "step": 64
    },
    {
      "epoch": 0.3452855245683931,
      "grad_norm": 2.5606510639190674,
      "learning_rate": 8.07106356344834e-05,
      "loss": 1.4501,
      "step": 65
    },
    {
      "epoch": 0.350597609561753,
      "grad_norm": 2.0793964862823486,
      "learning_rate": 8.005405736415126e-05,
      "loss": 1.4243,
      "step": 66
    },
    {
      "epoch": 0.3559096945551129,
      "grad_norm": 2.2079596519470215,
      "learning_rate": 7.938926261462366e-05,
      "loss": 1.3425,
      "step": 67
    },
    {
      "epoch": 0.3612217795484728,
      "grad_norm": 2.1000986099243164,
      "learning_rate": 7.871643313414718e-05,
      "loss": 1.3721,
      "step": 68
    },
    {
      "epoch": 0.3665338645418327,
      "grad_norm": 2.2770254611968994,
      "learning_rate": 7.803575286758364e-05,
      "loss": 1.4939,
      "step": 69
    },
    {
      "epoch": 0.37184594953519257,
      "grad_norm": 2.091768741607666,
      "learning_rate": 7.734740790612136e-05,
      "loss": 1.4325,
      "step": 70
    },
    {
      "epoch": 0.37715803452855245,
      "grad_norm": 2.333314895629883,
      "learning_rate": 7.66515864363997e-05,
      "loss": 1.4477,
      "step": 71
    },
    {
      "epoch": 0.38247011952191234,
      "grad_norm": 2.3316617012023926,
      "learning_rate": 7.594847868906076e-05,
      "loss": 1.6194,
      "step": 72
    },
    {
      "epoch": 0.38778220451527223,
      "grad_norm": 2.106593370437622,
      "learning_rate": 7.52382768867422e-05,
      "loss": 1.3938,
      "step": 73
    },
    {
      "epoch": 0.3930942895086321,
      "grad_norm": 2.1388003826141357,
      "learning_rate": 7.452117519152542e-05,
      "loss": 1.4695,
      "step": 74
    },
    {
      "epoch": 0.398406374501992,
      "grad_norm": 2.21095609664917,
      "learning_rate": 7.379736965185368e-05,
      "loss": 1.4777,
      "step": 75
    },
    {
      "epoch": 0.4037184594953519,
      "grad_norm": 2.0675554275512695,
      "learning_rate": 7.30670581489344e-05,
      "loss": 1.3245,
      "step": 76
    },
    {
      "epoch": 0.40903054448871184,
      "grad_norm": 1.7925864458084106,
      "learning_rate": 7.233044034264034e-05,
      "loss": 1.4433,
      "step": 77
    },
    {
      "epoch": 0.41434262948207173,
      "grad_norm": 2.0382237434387207,
      "learning_rate": 7.158771761692464e-05,
      "loss": 1.3347,
      "step": 78
    },
    {
      "epoch": 0.4196547144754316,
      "grad_norm": 2.0112431049346924,
      "learning_rate": 7.083909302476453e-05,
      "loss": 1.4446,
      "step": 79
    },
    {
      "epoch": 0.4249667994687915,
      "grad_norm": 1.8658033609390259,
      "learning_rate": 7.008477123264848e-05,
      "loss": 1.5634,
      "step": 80
    },
    {
      "epoch": 0.4302788844621514,
      "grad_norm": 2.0809333324432373,
      "learning_rate": 6.932495846462261e-05,
      "loss": 1.5047,
      "step": 81
    },
    {
      "epoch": 0.4355909694555113,
      "grad_norm": 1.7902474403381348,
      "learning_rate": 6.855986244591104e-05,
      "loss": 1.4596,
      "step": 82
    },
    {
      "epoch": 0.4409030544488712,
      "grad_norm": 2.0131309032440186,
      "learning_rate": 6.778969234612584e-05,
      "loss": 1.4884,
      "step": 83
    },
    {
      "epoch": 0.44621513944223107,
      "grad_norm": 2.0654282569885254,
      "learning_rate": 6.701465872208216e-05,
      "loss": 1.5301,
      "step": 84
    },
    {
      "epoch": 0.45152722443559096,
      "grad_norm": 2.1842639446258545,
      "learning_rate": 6.623497346023418e-05,
      "loss": 1.4964,
      "step": 85
    },
    {
      "epoch": 0.45683930942895085,
      "grad_norm": 2.017879009246826,
      "learning_rate": 6.545084971874738e-05,
      "loss": 1.4413,
      "step": 86
    },
    {
      "epoch": 0.46215139442231074,
      "grad_norm": 2.046215295791626,
      "learning_rate": 6.466250186922325e-05,
      "loss": 1.5414,
      "step": 87
    },
    {
      "epoch": 0.4674634794156706,
      "grad_norm": 2.1428616046905518,
      "learning_rate": 6.387014543809223e-05,
      "loss": 1.4064,
      "step": 88
    },
    {
      "epoch": 0.47277556440903057,
      "grad_norm": 2.3681371212005615,
      "learning_rate": 6.307399704769099e-05,
      "loss": 1.5363,
      "step": 89
    },
    {
      "epoch": 0.47808764940239046,
      "grad_norm": 2.3135945796966553,
      "learning_rate": 6.227427435703997e-05,
      "loss": 1.5381,
      "step": 90
    },
    {
      "epoch": 0.48339973439575035,
      "grad_norm": 2.8573532104492188,
      "learning_rate": 6.147119600233758e-05,
      "loss": 1.5686,
      "step": 91
    },
    {
      "epoch": 0.48871181938911024,
      "grad_norm": 2.3224003314971924,
      "learning_rate": 6.066498153718735e-05,
      "loss": 1.4144,
      "step": 92
    },
    {
      "epoch": 0.4940239043824701,
      "grad_norm": 2.596035957336426,
      "learning_rate": 5.985585137257401e-05,
      "loss": 1.5747,
      "step": 93
    },
    {
      "epoch": 0.49933598937583,
      "grad_norm": 2.7561702728271484,
      "learning_rate": 5.90440267166055e-05,
      "loss": 1.5332,
      "step": 94
    },
    {
      "epoch": 0.5046480743691899,
      "grad_norm": 1.6050623655319214,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 1.555,
      "step": 95
    },
    {
      "epoch": 0.5099601593625498,
      "grad_norm": 1.707343578338623,
      "learning_rate": 5.74131823855921e-05,
      "loss": 1.5333,
      "step": 96
    },
    {
      "epoch": 0.5152722443559097,
      "grad_norm": 1.707116961479187,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 1.5806,
      "step": 97
    },
    {
      "epoch": 0.5205843293492696,
      "grad_norm": 2.074115037918091,
      "learning_rate": 5.577423184847932e-05,
      "loss": 1.5777,
      "step": 98
    },
    {
      "epoch": 0.5258964143426295,
      "grad_norm": 1.7206108570098877,
      "learning_rate": 5.495227651252315e-05,
      "loss": 1.5205,
      "step": 99
    },
    {
      "epoch": 0.5312084993359893,
      "grad_norm": 1.8175201416015625,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 1.5327,
      "step": 100
    },
    {
      "epoch": 0.5312084993359893,
      "eval_loss": 1.45778489112854,
      "eval_runtime": 7.2112,
      "eval_samples_per_second": 43.96,
      "eval_steps_per_second": 11.094,
      "step": 100
    },
    {
      "epoch": 0.5365205843293492,
      "grad_norm": 1.9435057640075684,
      "learning_rate": 5.330452921628497e-05,
      "loss": 1.6681,
      "step": 101
    },
    {
      "epoch": 0.5418326693227091,
      "grad_norm": 2.2853028774261475,
      "learning_rate": 5.247918773366112e-05,
      "loss": 1.7598,
      "step": 102
    },
    {
      "epoch": 0.547144754316069,
      "grad_norm": 2.1462156772613525,
      "learning_rate": 5.165316846586541e-05,
      "loss": 1.7742,
      "step": 103
    },
    {
      "epoch": 0.5524568393094289,
      "grad_norm": 2.3823187351226807,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 1.7956,
      "step": 104
    },
    {
      "epoch": 0.5577689243027888,
      "grad_norm": 3.0888214111328125,
      "learning_rate": 5e-05,
      "loss": 2.0159,
      "step": 105
    },
    {
      "epoch": 0.5630810092961488,
      "grad_norm": 2.3878273963928223,
      "learning_rate": 4.917330276168208e-05,
      "loss": 1.7263,
      "step": 106
    },
    {
      "epoch": 0.5683930942895087,
      "grad_norm": 2.035449981689453,
      "learning_rate": 4.834683153413459e-05,
      "loss": 1.422,
      "step": 107
    },
    {
      "epoch": 0.5737051792828686,
      "grad_norm": 2.135831117630005,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 1.3636,
      "step": 108
    },
    {
      "epoch": 0.5790172642762285,
      "grad_norm": 2.0532004833221436,
      "learning_rate": 4.669547078371504e-05,
      "loss": 1.5042,
      "step": 109
    },
    {
      "epoch": 0.5843293492695883,
      "grad_norm": 2.3273632526397705,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 1.3496,
      "step": 110
    },
    {
      "epoch": 0.5896414342629482,
      "grad_norm": 2.2188820838928223,
      "learning_rate": 4.504772348747687e-05,
      "loss": 1.4727,
      "step": 111
    },
    {
      "epoch": 0.5949535192563081,
      "grad_norm": 1.8522779941558838,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 1.3787,
      "step": 112
    },
    {
      "epoch": 0.600265604249668,
      "grad_norm": 1.7978854179382324,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 1.4079,
      "step": 113
    },
    {
      "epoch": 0.6055776892430279,
      "grad_norm": 1.8319061994552612,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 1.3754,
      "step": 114
    },
    {
      "epoch": 0.6108897742363878,
      "grad_norm": 2.030869245529175,
      "learning_rate": 4.17702704859633e-05,
      "loss": 1.5467,
      "step": 115
    },
    {
      "epoch": 0.6162018592297477,
      "grad_norm": 1.717625617980957,
      "learning_rate": 4.095597328339452e-05,
      "loss": 1.3397,
      "step": 116
    },
    {
      "epoch": 0.6215139442231076,
      "grad_norm": 1.5522135496139526,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 1.0999,
      "step": 117
    },
    {
      "epoch": 0.6268260292164675,
      "grad_norm": 1.6645989418029785,
      "learning_rate": 3.933501846281267e-05,
      "loss": 1.2908,
      "step": 118
    },
    {
      "epoch": 0.6321381142098274,
      "grad_norm": 1.68788480758667,
      "learning_rate": 3.852880399766243e-05,
      "loss": 1.227,
      "step": 119
    },
    {
      "epoch": 0.6374501992031872,
      "grad_norm": 2.076498031616211,
      "learning_rate": 3.772572564296005e-05,
      "loss": 1.3627,
      "step": 120
    },
    {
      "epoch": 0.6427622841965471,
      "grad_norm": 1.9502153396606445,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 1.2332,
      "step": 121
    },
    {
      "epoch": 0.648074369189907,
      "grad_norm": 2.1534640789031982,
      "learning_rate": 3.612985456190778e-05,
      "loss": 1.3397,
      "step": 122
    },
    {
      "epoch": 0.6533864541832669,
      "grad_norm": 1.6815413236618042,
      "learning_rate": 3.533749813077677e-05,
      "loss": 1.2821,
      "step": 123
    },
    {
      "epoch": 0.6586985391766268,
      "grad_norm": 2.2450623512268066,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 1.4655,
      "step": 124
    },
    {
      "epoch": 0.6640106241699867,
      "grad_norm": 2.0245814323425293,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 1.4755,
      "step": 125
    },
    {
      "epoch": 0.6693227091633466,
      "grad_norm": 2.033900260925293,
      "learning_rate": 3.298534127791785e-05,
      "loss": 1.31,
      "step": 126
    },
    {
      "epoch": 0.6746347941567065,
      "grad_norm": 1.9057337045669556,
      "learning_rate": 3.221030765387417e-05,
      "loss": 1.3086,
      "step": 127
    },
    {
      "epoch": 0.6799468791500664,
      "grad_norm": 1.8459917306900024,
      "learning_rate": 3.144013755408895e-05,
      "loss": 1.1782,
      "step": 128
    },
    {
      "epoch": 0.6852589641434262,
      "grad_norm": 1.8653852939605713,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 1.2667,
      "step": 129
    },
    {
      "epoch": 0.6905710491367862,
      "grad_norm": 1.894116759300232,
      "learning_rate": 2.991522876735154e-05,
      "loss": 1.5062,
      "step": 130
    },
    {
      "epoch": 0.6958831341301461,
      "grad_norm": 1.9408761262893677,
      "learning_rate": 2.916090697523549e-05,
      "loss": 1.3065,
      "step": 131
    },
    {
      "epoch": 0.701195219123506,
      "grad_norm": 2.3585686683654785,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 1.3078,
      "step": 132
    },
    {
      "epoch": 0.7065073041168659,
      "grad_norm": 1.979691505432129,
      "learning_rate": 2.766955965735968e-05,
      "loss": 1.5929,
      "step": 133
    },
    {
      "epoch": 0.7118193891102258,
      "grad_norm": 1.8617613315582275,
      "learning_rate": 2.693294185106562e-05,
      "loss": 1.4103,
      "step": 134
    },
    {
      "epoch": 0.7171314741035857,
      "grad_norm": 1.8611036539077759,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 1.3035,
      "step": 135
    },
    {
      "epoch": 0.7224435590969456,
      "grad_norm": 2.0097100734710693,
      "learning_rate": 2.547882480847461e-05,
      "loss": 1.3619,
      "step": 136
    },
    {
      "epoch": 0.7277556440903055,
      "grad_norm": 2.175086259841919,
      "learning_rate": 2.476172311325783e-05,
      "loss": 1.6451,
      "step": 137
    },
    {
      "epoch": 0.7330677290836654,
      "grad_norm": 2.2641239166259766,
      "learning_rate": 2.405152131093926e-05,
      "loss": 1.2518,
      "step": 138
    },
    {
      "epoch": 0.7383798140770252,
      "grad_norm": 2.5203285217285156,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 1.6061,
      "step": 139
    },
    {
      "epoch": 0.7436918990703851,
      "grad_norm": 2.3696491718292236,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 1.5867,
      "step": 140
    },
    {
      "epoch": 0.749003984063745,
      "grad_norm": 3.6379926204681396,
      "learning_rate": 2.196424713241637e-05,
      "loss": 1.8054,
      "step": 141
    },
    {
      "epoch": 0.7543160690571049,
      "grad_norm": 1.260744571685791,
      "learning_rate": 2.128356686585282e-05,
      "loss": 1.5475,
      "step": 142
    },
    {
      "epoch": 0.7596281540504648,
      "grad_norm": 1.4957962036132812,
      "learning_rate": 2.061073738537635e-05,
      "loss": 1.4532,
      "step": 143
    },
    {
      "epoch": 0.7649402390438247,
      "grad_norm": 1.5174098014831543,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 1.4491,
      "step": 144
    },
    {
      "epoch": 0.7702523240371846,
      "grad_norm": 1.7261474132537842,
      "learning_rate": 1.928936436551661e-05,
      "loss": 1.5931,
      "step": 145
    },
    {
      "epoch": 0.7755644090305445,
      "grad_norm": 1.5767656564712524,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 1.4129,
      "step": 146
    },
    {
      "epoch": 0.7808764940239044,
      "grad_norm": 1.7982088327407837,
      "learning_rate": 1.800157297483417e-05,
      "loss": 1.4406,
      "step": 147
    },
    {
      "epoch": 0.7861885790172642,
      "grad_norm": 2.116525888442993,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 1.7292,
      "step": 148
    },
    {
      "epoch": 0.7915006640106241,
      "grad_norm": 2.237140417098999,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 1.5599,
      "step": 149
    },
    {
      "epoch": 0.796812749003984,
      "grad_norm": 2.6314172744750977,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 1.7428,
      "step": 150
    },
    {
      "epoch": 0.796812749003984,
      "eval_loss": 1.4133718013763428,
      "eval_runtime": 7.2262,
      "eval_samples_per_second": 43.868,
      "eval_steps_per_second": 11.071,
      "step": 150
    },
    {
      "epoch": 0.8021248339973439,
      "grad_norm": 2.130983591079712,
      "learning_rate": 1.553232954407171e-05,
      "loss": 1.5484,
      "step": 151
    },
    {
      "epoch": 0.8074369189907038,
      "grad_norm": 2.9864869117736816,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 1.6533,
      "step": 152
    },
    {
      "epoch": 0.8127490039840638,
      "grad_norm": 2.8473427295684814,
      "learning_rate": 1.435357758543015e-05,
      "loss": 1.7678,
      "step": 153
    },
    {
      "epoch": 0.8180610889774237,
      "grad_norm": 3.733130931854248,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 2.1122,
      "step": 154
    },
    {
      "epoch": 0.8233731739707836,
      "grad_norm": 1.6506917476654053,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 1.1746,
      "step": 155
    },
    {
      "epoch": 0.8286852589641435,
      "grad_norm": 1.7522071599960327,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 1.499,
      "step": 156
    },
    {
      "epoch": 0.8339973439575034,
      "grad_norm": 1.7502012252807617,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 1.3589,
      "step": 157
    },
    {
      "epoch": 0.8393094289508632,
      "grad_norm": 1.8208122253417969,
      "learning_rate": 1.157994445715706e-05,
      "loss": 1.6344,
      "step": 158
    },
    {
      "epoch": 0.8446215139442231,
      "grad_norm": 1.6915513277053833,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 1.3799,
      "step": 159
    },
    {
      "epoch": 0.849933598937583,
      "grad_norm": 1.798140287399292,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 1.2604,
      "step": 160
    },
    {
      "epoch": 0.8552456839309429,
      "grad_norm": 1.6726526021957397,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 1.2363,
      "step": 161
    },
    {
      "epoch": 0.8605577689243028,
      "grad_norm": 1.6930738687515259,
      "learning_rate": 9.549150281252633e-06,
      "loss": 1.2784,
      "step": 162
    },
    {
      "epoch": 0.8658698539176627,
      "grad_norm": 1.7110973596572876,
      "learning_rate": 9.068759265665384e-06,
      "loss": 1.3878,
      "step": 163
    },
    {
      "epoch": 0.8711819389110226,
      "grad_norm": 1.7826908826828003,
      "learning_rate": 8.599558442598998e-06,
      "loss": 1.5197,
      "step": 164
    },
    {
      "epoch": 0.8764940239043825,
      "grad_norm": 1.5502291917800903,
      "learning_rate": 8.141676086873572e-06,
      "loss": 1.2793,
      "step": 165
    },
    {
      "epoch": 0.8818061088977424,
      "grad_norm": 1.7470778226852417,
      "learning_rate": 7.695237378953223e-06,
      "loss": 1.3057,
      "step": 166
    },
    {
      "epoch": 0.8871181938911022,
      "grad_norm": 1.684496521949768,
      "learning_rate": 7.260364370723044e-06,
      "loss": 1.384,
      "step": 167
    },
    {
      "epoch": 0.8924302788844621,
      "grad_norm": 1.6192524433135986,
      "learning_rate": 6.837175952121306e-06,
      "loss": 1.3263,
      "step": 168
    },
    {
      "epoch": 0.897742363877822,
      "grad_norm": 1.8296061754226685,
      "learning_rate": 6.425787818636131e-06,
      "loss": 1.3344,
      "step": 169
    },
    {
      "epoch": 0.9030544488711819,
      "grad_norm": 1.6950888633728027,
      "learning_rate": 6.026312439675552e-06,
      "loss": 1.376,
      "step": 170
    },
    {
      "epoch": 0.9083665338645418,
      "grad_norm": 1.7980575561523438,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 1.369,
      "step": 171
    },
    {
      "epoch": 0.9136786188579017,
      "grad_norm": 1.9877363443374634,
      "learning_rate": 5.263533508961827e-06,
      "loss": 1.4185,
      "step": 172
    },
    {
      "epoch": 0.9189907038512616,
      "grad_norm": 1.9132676124572754,
      "learning_rate": 4.900438493352055e-06,
      "loss": 1.5245,
      "step": 173
    },
    {
      "epoch": 0.9243027888446215,
      "grad_norm": 1.9815455675125122,
      "learning_rate": 4.549673247541875e-06,
      "loss": 1.3448,
      "step": 174
    },
    {
      "epoch": 0.9296148738379814,
      "grad_norm": 1.7365641593933105,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 1.4778,
      "step": 175
    },
    {
      "epoch": 0.9349269588313412,
      "grad_norm": 1.7712080478668213,
      "learning_rate": 3.885512251130763e-06,
      "loss": 1.3158,
      "step": 176
    },
    {
      "epoch": 0.9402390438247012,
      "grad_norm": 2.0998706817626953,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 1.3785,
      "step": 177
    },
    {
      "epoch": 0.9455511288180611,
      "grad_norm": 2.02655291557312,
      "learning_rate": 3.271776770026963e-06,
      "loss": 1.4345,
      "step": 178
    },
    {
      "epoch": 0.950863213811421,
      "grad_norm": 2.129718065261841,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 1.5927,
      "step": 179
    },
    {
      "epoch": 0.9561752988047809,
      "grad_norm": 1.8841341733932495,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 1.2823,
      "step": 180
    },
    {
      "epoch": 0.9614873837981408,
      "grad_norm": 1.7093276977539062,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 1.3175,
      "step": 181
    },
    {
      "epoch": 0.9667994687915007,
      "grad_norm": 2.389735221862793,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 1.6821,
      "step": 182
    },
    {
      "epoch": 0.9721115537848606,
      "grad_norm": 2.1029300689697266,
      "learning_rate": 1.962316193157593e-06,
      "loss": 1.4846,
      "step": 183
    },
    {
      "epoch": 0.9774236387782205,
      "grad_norm": 1.910742998123169,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 1.3262,
      "step": 184
    },
    {
      "epoch": 0.9827357237715804,
      "grad_norm": 2.089308500289917,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 1.4581,
      "step": 185
    },
    {
      "epoch": 0.9880478087649402,
      "grad_norm": 2.3141303062438965,
      "learning_rate": 1.333670137599713e-06,
      "loss": 1.4081,
      "step": 186
    },
    {
      "epoch": 0.9933598937583001,
      "grad_norm": 2.4875192642211914,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 1.6058,
      "step": 187
    },
    {
      "epoch": 0.99867197875166,
      "grad_norm": 2.889101505279541,
      "learning_rate": 9.810017062595322e-07,
      "loss": 1.6184,
      "step": 188
    },
    {
      "epoch": 1.00398406374502,
      "grad_norm": 1.6815223693847656,
      "learning_rate": 8.247462563808817e-07,
      "loss": 1.18,
      "step": 189
    },
    {
      "epoch": 1.0092961487383798,
      "grad_norm": 1.2229230403900146,
      "learning_rate": 6.819348298638839e-07,
      "loss": 1.3061,
      "step": 190
    },
    {
      "epoch": 1.0146082337317397,
      "grad_norm": 1.3119114637374878,
      "learning_rate": 5.526064699265753e-07,
      "loss": 1.2225,
      "step": 191
    },
    {
      "epoch": 1.0199203187250996,
      "grad_norm": 1.5561548471450806,
      "learning_rate": 4.367965336512403e-07,
      "loss": 1.4897,
      "step": 192
    },
    {
      "epoch": 1.0252324037184595,
      "grad_norm": 1.4772052764892578,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 1.1711,
      "step": 193
    },
    {
      "epoch": 1.0305444887118194,
      "grad_norm": 1.6754026412963867,
      "learning_rate": 2.458548727494292e-07,
      "loss": 1.4283,
      "step": 194
    },
    {
      "epoch": 1.0358565737051793,
      "grad_norm": 1.7270011901855469,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 1.307,
      "step": 195
    },
    {
      "epoch": 1.0411686586985391,
      "grad_norm": 2.0306849479675293,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 1.5166,
      "step": 196
    },
    {
      "epoch": 1.046480743691899,
      "grad_norm": 2.1373815536499023,
      "learning_rate": 6.150154258476315e-08,
      "loss": 1.6676,
      "step": 197
    },
    {
      "epoch": 1.051792828685259,
      "grad_norm": 2.171025514602661,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 1.3999,
      "step": 198
    },
    {
      "epoch": 1.0571049136786188,
      "grad_norm": 2.575199842453003,
      "learning_rate": 6.834750376549792e-09,
      "loss": 1.8693,
      "step": 199
    },
    {
      "epoch": 1.0624169986719787,
      "grad_norm": 2.987534284591675,
      "learning_rate": 0.0,
      "loss": 1.8453,
      "step": 200
    },
    {
      "epoch": 1.0624169986719787,
      "eval_loss": 1.3869588375091553,
      "eval_runtime": 7.2053,
      "eval_samples_per_second": 43.995,
      "eval_steps_per_second": 11.103,
      "step": 200
    }
  ],
  "logging_steps": 1,
  "max_steps": 200,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "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": 2.032832394638131e+16,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}