{
  "best_metric": 11.916446685791016,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.36413290851160673,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0018206645425580337,
      "grad_norm": 0.03413726016879082,
      "learning_rate": 1e-05,
      "loss": 11.9323,
      "step": 1
    },
    {
      "epoch": 0.0018206645425580337,
      "eval_loss": 11.93479061126709,
      "eval_runtime": 5.058,
      "eval_samples_per_second": 182.88,
      "eval_steps_per_second": 45.868,
      "step": 1
    },
    {
      "epoch": 0.0036413290851160674,
      "grad_norm": 0.04938232898712158,
      "learning_rate": 2e-05,
      "loss": 11.9373,
      "step": 2
    },
    {
      "epoch": 0.005461993627674101,
      "grad_norm": 0.03765185922384262,
      "learning_rate": 3e-05,
      "loss": 11.9313,
      "step": 3
    },
    {
      "epoch": 0.007282658170232135,
      "grad_norm": 0.03305572271347046,
      "learning_rate": 4e-05,
      "loss": 11.9374,
      "step": 4
    },
    {
      "epoch": 0.009103322712790168,
      "grad_norm": 0.04273687303066254,
      "learning_rate": 5e-05,
      "loss": 11.9327,
      "step": 5
    },
    {
      "epoch": 0.010923987255348202,
      "grad_norm": 0.04251832887530327,
      "learning_rate": 6e-05,
      "loss": 11.9344,
      "step": 6
    },
    {
      "epoch": 0.012744651797906235,
      "grad_norm": 0.04986019805073738,
      "learning_rate": 7e-05,
      "loss": 11.9337,
      "step": 7
    },
    {
      "epoch": 0.01456531634046427,
      "grad_norm": 0.0382511280477047,
      "learning_rate": 8e-05,
      "loss": 11.9358,
      "step": 8
    },
    {
      "epoch": 0.016385980883022302,
      "grad_norm": 0.04335345700383186,
      "learning_rate": 9e-05,
      "loss": 11.9329,
      "step": 9
    },
    {
      "epoch": 0.018206645425580335,
      "grad_norm": 0.043804410845041275,
      "learning_rate": 0.0001,
      "loss": 11.9339,
      "step": 10
    },
    {
      "epoch": 0.02002730996813837,
      "grad_norm": 0.050628215074539185,
      "learning_rate": 9.999316524962345e-05,
      "loss": 11.9351,
      "step": 11
    },
    {
      "epoch": 0.021847974510696404,
      "grad_norm": 0.043292414397001266,
      "learning_rate": 9.997266286704631e-05,
      "loss": 11.9353,
      "step": 12
    },
    {
      "epoch": 0.023668639053254437,
      "grad_norm": 0.05560588464140892,
      "learning_rate": 9.993849845741524e-05,
      "loss": 11.9345,
      "step": 13
    },
    {
      "epoch": 0.02548930359581247,
      "grad_norm": 0.06867741793394089,
      "learning_rate": 9.989068136093873e-05,
      "loss": 11.9322,
      "step": 14
    },
    {
      "epoch": 0.027309968138370506,
      "grad_norm": 0.07315599173307419,
      "learning_rate": 9.98292246503335e-05,
      "loss": 11.9318,
      "step": 15
    },
    {
      "epoch": 0.02913063268092854,
      "grad_norm": 0.07529256492853165,
      "learning_rate": 9.975414512725057e-05,
      "loss": 11.9328,
      "step": 16
    },
    {
      "epoch": 0.030951297223486572,
      "grad_norm": 0.07898218929767609,
      "learning_rate": 9.966546331768191e-05,
      "loss": 11.9336,
      "step": 17
    },
    {
      "epoch": 0.032771961766044605,
      "grad_norm": 0.08530876040458679,
      "learning_rate": 9.956320346634876e-05,
      "loss": 11.9336,
      "step": 18
    },
    {
      "epoch": 0.03459262630860264,
      "grad_norm": 0.08568832278251648,
      "learning_rate": 9.944739353007344e-05,
      "loss": 11.932,
      "step": 19
    },
    {
      "epoch": 0.03641329085116067,
      "grad_norm": 0.10579682141542435,
      "learning_rate": 9.931806517013612e-05,
      "loss": 11.9316,
      "step": 20
    },
    {
      "epoch": 0.03823395539371871,
      "grad_norm": 0.11739271134138107,
      "learning_rate": 9.917525374361912e-05,
      "loss": 11.9312,
      "step": 21
    },
    {
      "epoch": 0.04005461993627674,
      "grad_norm": 0.11046414822340012,
      "learning_rate": 9.901899829374047e-05,
      "loss": 11.9301,
      "step": 22
    },
    {
      "epoch": 0.041875284478834776,
      "grad_norm": 0.1442708969116211,
      "learning_rate": 9.884934153917997e-05,
      "loss": 11.9328,
      "step": 23
    },
    {
      "epoch": 0.04369594902139281,
      "grad_norm": 0.14422491192817688,
      "learning_rate": 9.86663298624003e-05,
      "loss": 11.9277,
      "step": 24
    },
    {
      "epoch": 0.04551661356395084,
      "grad_norm": 0.16111890971660614,
      "learning_rate": 9.847001329696653e-05,
      "loss": 11.9291,
      "step": 25
    },
    {
      "epoch": 0.047337278106508875,
      "grad_norm": 0.13617178797721863,
      "learning_rate": 9.826044551386744e-05,
      "loss": 11.9292,
      "step": 26
    },
    {
      "epoch": 0.04915794264906691,
      "grad_norm": 0.10961776971817017,
      "learning_rate": 9.803768380684242e-05,
      "loss": 11.9289,
      "step": 27
    },
    {
      "epoch": 0.05097860719162494,
      "grad_norm": 0.13653425872325897,
      "learning_rate": 9.780178907671789e-05,
      "loss": 11.9262,
      "step": 28
    },
    {
      "epoch": 0.05279927173418298,
      "grad_norm": 0.1738469898700714,
      "learning_rate": 9.755282581475769e-05,
      "loss": 11.923,
      "step": 29
    },
    {
      "epoch": 0.05461993627674101,
      "grad_norm": 0.13671448826789856,
      "learning_rate": 9.729086208503174e-05,
      "loss": 11.924,
      "step": 30
    },
    {
      "epoch": 0.056440600819299046,
      "grad_norm": 0.16584277153015137,
      "learning_rate": 9.701596950580806e-05,
      "loss": 11.9274,
      "step": 31
    },
    {
      "epoch": 0.05826126536185708,
      "grad_norm": 0.1861305832862854,
      "learning_rate": 9.672822322997305e-05,
      "loss": 11.9206,
      "step": 32
    },
    {
      "epoch": 0.06008192990441511,
      "grad_norm": 0.23096109926700592,
      "learning_rate": 9.642770192448536e-05,
      "loss": 11.9235,
      "step": 33
    },
    {
      "epoch": 0.061902594446973144,
      "grad_norm": 0.1698852926492691,
      "learning_rate": 9.611448774886924e-05,
      "loss": 11.919,
      "step": 34
    },
    {
      "epoch": 0.06372325898953118,
      "grad_norm": 0.13847710192203522,
      "learning_rate": 9.578866633275288e-05,
      "loss": 11.9193,
      "step": 35
    },
    {
      "epoch": 0.06554392353208921,
      "grad_norm": 0.20390431582927704,
      "learning_rate": 9.545032675245813e-05,
      "loss": 11.9244,
      "step": 36
    },
    {
      "epoch": 0.06736458807464725,
      "grad_norm": 0.13347230851650238,
      "learning_rate": 9.509956150664796e-05,
      "loss": 11.9165,
      "step": 37
    },
    {
      "epoch": 0.06918525261720528,
      "grad_norm": 0.13313481211662292,
      "learning_rate": 9.473646649103818e-05,
      "loss": 11.9185,
      "step": 38
    },
    {
      "epoch": 0.07100591715976332,
      "grad_norm": 0.14947770535945892,
      "learning_rate": 9.43611409721806e-05,
      "loss": 11.9168,
      "step": 39
    },
    {
      "epoch": 0.07282658170232134,
      "grad_norm": 0.10729013383388519,
      "learning_rate": 9.397368756032445e-05,
      "loss": 11.9199,
      "step": 40
    },
    {
      "epoch": 0.07464724624487938,
      "grad_norm": 0.09050336480140686,
      "learning_rate": 9.357421218136386e-05,
      "loss": 11.9181,
      "step": 41
    },
    {
      "epoch": 0.07646791078743742,
      "grad_norm": 0.14425118267536163,
      "learning_rate": 9.316282404787871e-05,
      "loss": 11.9193,
      "step": 42
    },
    {
      "epoch": 0.07828857532999545,
      "grad_norm": 0.09219750761985779,
      "learning_rate": 9.273963562927695e-05,
      "loss": 11.915,
      "step": 43
    },
    {
      "epoch": 0.08010923987255349,
      "grad_norm": 0.18210937082767487,
      "learning_rate": 9.230476262104677e-05,
      "loss": 11.9149,
      "step": 44
    },
    {
      "epoch": 0.08192990441511151,
      "grad_norm": 0.10966989398002625,
      "learning_rate": 9.185832391312644e-05,
      "loss": 11.9171,
      "step": 45
    },
    {
      "epoch": 0.08375056895766955,
      "grad_norm": 0.14103111624717712,
      "learning_rate": 9.140044155740101e-05,
      "loss": 11.9131,
      "step": 46
    },
    {
      "epoch": 0.08557123350022758,
      "grad_norm": 0.1283736675977707,
      "learning_rate": 9.093124073433463e-05,
      "loss": 11.9146,
      "step": 47
    },
    {
      "epoch": 0.08739189804278562,
      "grad_norm": 0.16424375772476196,
      "learning_rate": 9.045084971874738e-05,
      "loss": 11.9164,
      "step": 48
    },
    {
      "epoch": 0.08921256258534364,
      "grad_norm": 0.2161012589931488,
      "learning_rate": 8.995939984474624e-05,
      "loss": 11.9115,
      "step": 49
    },
    {
      "epoch": 0.09103322712790168,
      "grad_norm": 0.2343643307685852,
      "learning_rate": 8.945702546981969e-05,
      "loss": 11.9071,
      "step": 50
    },
    {
      "epoch": 0.09103322712790168,
      "eval_loss": 11.919342994689941,
      "eval_runtime": 5.0387,
      "eval_samples_per_second": 183.579,
      "eval_steps_per_second": 46.044,
      "step": 50
    },
    {
      "epoch": 0.09285389167045972,
      "grad_norm": 0.06608626991510391,
      "learning_rate": 8.894386393810563e-05,
      "loss": 11.9245,
      "step": 51
    },
    {
      "epoch": 0.09467455621301775,
      "grad_norm": 0.07884740829467773,
      "learning_rate": 8.842005554284296e-05,
      "loss": 11.9238,
      "step": 52
    },
    {
      "epoch": 0.09649522075557579,
      "grad_norm": 0.06321306526660919,
      "learning_rate": 8.788574348801675e-05,
      "loss": 11.9202,
      "step": 53
    },
    {
      "epoch": 0.09831588529813381,
      "grad_norm": 0.08037225157022476,
      "learning_rate": 8.73410738492077e-05,
      "loss": 11.9244,
      "step": 54
    },
    {
      "epoch": 0.10013654984069185,
      "grad_norm": 0.06967957317829132,
      "learning_rate": 8.678619553365659e-05,
      "loss": 11.9206,
      "step": 55
    },
    {
      "epoch": 0.10195721438324988,
      "grad_norm": 0.06570564955472946,
      "learning_rate": 8.622126023955446e-05,
      "loss": 11.9239,
      "step": 56
    },
    {
      "epoch": 0.10377787892580792,
      "grad_norm": 0.066615529358387,
      "learning_rate": 8.564642241456986e-05,
      "loss": 11.9257,
      "step": 57
    },
    {
      "epoch": 0.10559854346836596,
      "grad_norm": 0.06762253493070602,
      "learning_rate": 8.506183921362443e-05,
      "loss": 11.9227,
      "step": 58
    },
    {
      "epoch": 0.10741920801092399,
      "grad_norm": 0.06794265657663345,
      "learning_rate": 8.44676704559283e-05,
      "loss": 11.92,
      "step": 59
    },
    {
      "epoch": 0.10923987255348203,
      "grad_norm": 0.054183103144168854,
      "learning_rate": 8.386407858128706e-05,
      "loss": 11.9229,
      "step": 60
    },
    {
      "epoch": 0.11106053709604005,
      "grad_norm": 0.047181401401758194,
      "learning_rate": 8.32512286056924e-05,
      "loss": 11.9231,
      "step": 61
    },
    {
      "epoch": 0.11288120163859809,
      "grad_norm": 0.0388273261487484,
      "learning_rate": 8.262928807620843e-05,
      "loss": 11.9231,
      "step": 62
    },
    {
      "epoch": 0.11470186618115612,
      "grad_norm": 0.07356078177690506,
      "learning_rate": 8.199842702516583e-05,
      "loss": 11.9207,
      "step": 63
    },
    {
      "epoch": 0.11652253072371416,
      "grad_norm": 0.060196008533239365,
      "learning_rate": 8.135881792367686e-05,
      "loss": 11.9203,
      "step": 64
    },
    {
      "epoch": 0.11834319526627218,
      "grad_norm": 0.03406398370862007,
      "learning_rate": 8.07106356344834e-05,
      "loss": 11.9185,
      "step": 65
    },
    {
      "epoch": 0.12016385980883022,
      "grad_norm": 0.04454463720321655,
      "learning_rate": 8.005405736415126e-05,
      "loss": 11.9204,
      "step": 66
    },
    {
      "epoch": 0.12198452435138826,
      "grad_norm": 0.06781654059886932,
      "learning_rate": 7.938926261462366e-05,
      "loss": 11.9255,
      "step": 67
    },
    {
      "epoch": 0.12380518889394629,
      "grad_norm": 0.03592757135629654,
      "learning_rate": 7.871643313414718e-05,
      "loss": 11.9213,
      "step": 68
    },
    {
      "epoch": 0.12562585343650431,
      "grad_norm": 0.053486570715904236,
      "learning_rate": 7.803575286758364e-05,
      "loss": 11.9207,
      "step": 69
    },
    {
      "epoch": 0.12744651797906237,
      "grad_norm": 0.045916493982076645,
      "learning_rate": 7.734740790612136e-05,
      "loss": 11.9186,
      "step": 70
    },
    {
      "epoch": 0.1292671825216204,
      "grad_norm": 0.029214372858405113,
      "learning_rate": 7.66515864363997e-05,
      "loss": 11.9198,
      "step": 71
    },
    {
      "epoch": 0.13108784706417842,
      "grad_norm": 0.04013194516301155,
      "learning_rate": 7.594847868906076e-05,
      "loss": 11.9189,
      "step": 72
    },
    {
      "epoch": 0.13290851160673645,
      "grad_norm": 0.05094152316451073,
      "learning_rate": 7.52382768867422e-05,
      "loss": 11.9216,
      "step": 73
    },
    {
      "epoch": 0.1347291761492945,
      "grad_norm": 0.05009616166353226,
      "learning_rate": 7.452117519152542e-05,
      "loss": 11.9178,
      "step": 74
    },
    {
      "epoch": 0.13654984069185253,
      "grad_norm": 0.030104748904705048,
      "learning_rate": 7.379736965185368e-05,
      "loss": 11.9162,
      "step": 75
    },
    {
      "epoch": 0.13837050523441055,
      "grad_norm": 0.07186371088027954,
      "learning_rate": 7.30670581489344e-05,
      "loss": 11.919,
      "step": 76
    },
    {
      "epoch": 0.1401911697769686,
      "grad_norm": 0.04096841439604759,
      "learning_rate": 7.233044034264034e-05,
      "loss": 11.9182,
      "step": 77
    },
    {
      "epoch": 0.14201183431952663,
      "grad_norm": 0.04554932937026024,
      "learning_rate": 7.158771761692464e-05,
      "loss": 11.9147,
      "step": 78
    },
    {
      "epoch": 0.14383249886208466,
      "grad_norm": 0.036361921578645706,
      "learning_rate": 7.083909302476453e-05,
      "loss": 11.9184,
      "step": 79
    },
    {
      "epoch": 0.14565316340464268,
      "grad_norm": 0.03751020133495331,
      "learning_rate": 7.008477123264848e-05,
      "loss": 11.9168,
      "step": 80
    },
    {
      "epoch": 0.14747382794720074,
      "grad_norm": 0.033168449997901917,
      "learning_rate": 6.932495846462261e-05,
      "loss": 11.9182,
      "step": 81
    },
    {
      "epoch": 0.14929449248975876,
      "grad_norm": 0.02722890116274357,
      "learning_rate": 6.855986244591104e-05,
      "loss": 11.9191,
      "step": 82
    },
    {
      "epoch": 0.1511151570323168,
      "grad_norm": 0.048928141593933105,
      "learning_rate": 6.778969234612584e-05,
      "loss": 11.9201,
      "step": 83
    },
    {
      "epoch": 0.15293582157487484,
      "grad_norm": 0.06329765170812607,
      "learning_rate": 6.701465872208216e-05,
      "loss": 11.9189,
      "step": 84
    },
    {
      "epoch": 0.15475648611743287,
      "grad_norm": 0.04730033501982689,
      "learning_rate": 6.623497346023418e-05,
      "loss": 11.9143,
      "step": 85
    },
    {
      "epoch": 0.1565771506599909,
      "grad_norm": 0.06353859603404999,
      "learning_rate": 6.545084971874738e-05,
      "loss": 11.9123,
      "step": 86
    },
    {
      "epoch": 0.15839781520254892,
      "grad_norm": 0.06514666229486465,
      "learning_rate": 6.466250186922325e-05,
      "loss": 11.9186,
      "step": 87
    },
    {
      "epoch": 0.16021847974510697,
      "grad_norm": 0.06427396088838577,
      "learning_rate": 6.387014543809223e-05,
      "loss": 11.9153,
      "step": 88
    },
    {
      "epoch": 0.162039144287665,
      "grad_norm": 0.04870816320180893,
      "learning_rate": 6.307399704769099e-05,
      "loss": 11.9147,
      "step": 89
    },
    {
      "epoch": 0.16385980883022302,
      "grad_norm": 0.058200716972351074,
      "learning_rate": 6.227427435703997e-05,
      "loss": 11.9163,
      "step": 90
    },
    {
      "epoch": 0.16568047337278108,
      "grad_norm": 0.05362579971551895,
      "learning_rate": 6.147119600233758e-05,
      "loss": 11.9166,
      "step": 91
    },
    {
      "epoch": 0.1675011379153391,
      "grad_norm": 0.06777684390544891,
      "learning_rate": 6.066498153718735e-05,
      "loss": 11.9144,
      "step": 92
    },
    {
      "epoch": 0.16932180245789713,
      "grad_norm": 0.06118669733405113,
      "learning_rate": 5.985585137257401e-05,
      "loss": 11.9148,
      "step": 93
    },
    {
      "epoch": 0.17114246700045516,
      "grad_norm": 0.06644754856824875,
      "learning_rate": 5.90440267166055e-05,
      "loss": 11.9175,
      "step": 94
    },
    {
      "epoch": 0.1729631315430132,
      "grad_norm": 0.04351932927966118,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 11.9143,
      "step": 95
    },
    {
      "epoch": 0.17478379608557124,
      "grad_norm": 0.09682610630989075,
      "learning_rate": 5.74131823855921e-05,
      "loss": 11.9128,
      "step": 96
    },
    {
      "epoch": 0.17660446062812926,
      "grad_norm": 0.11314849555492401,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 11.9112,
      "step": 97
    },
    {
      "epoch": 0.1784251251706873,
      "grad_norm": 0.09338361769914627,
      "learning_rate": 5.577423184847932e-05,
      "loss": 11.9097,
      "step": 98
    },
    {
      "epoch": 0.18024578971324534,
      "grad_norm": 0.10834107547998428,
      "learning_rate": 5.495227651252315e-05,
      "loss": 11.9166,
      "step": 99
    },
    {
      "epoch": 0.18206645425580337,
      "grad_norm": 0.179707333445549,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 11.9038,
      "step": 100
    },
    {
      "epoch": 0.18206645425580337,
      "eval_loss": 11.917658805847168,
      "eval_runtime": 5.0402,
      "eval_samples_per_second": 183.523,
      "eval_steps_per_second": 46.03,
      "step": 100
    },
    {
      "epoch": 0.1838871187983614,
      "grad_norm": 0.043141648173332214,
      "learning_rate": 5.330452921628497e-05,
      "loss": 11.9247,
      "step": 101
    },
    {
      "epoch": 0.18570778334091945,
      "grad_norm": 0.05003609508275986,
      "learning_rate": 5.247918773366112e-05,
      "loss": 11.9214,
      "step": 102
    },
    {
      "epoch": 0.18752844788347747,
      "grad_norm": 0.04550706967711449,
      "learning_rate": 5.165316846586541e-05,
      "loss": 11.9198,
      "step": 103
    },
    {
      "epoch": 0.1893491124260355,
      "grad_norm": 0.047907501459121704,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 11.923,
      "step": 104
    },
    {
      "epoch": 0.19116977696859352,
      "grad_norm": 0.0394832044839859,
      "learning_rate": 5e-05,
      "loss": 11.9205,
      "step": 105
    },
    {
      "epoch": 0.19299044151115158,
      "grad_norm": 0.048778411000967026,
      "learning_rate": 4.917330276168208e-05,
      "loss": 11.9193,
      "step": 106
    },
    {
      "epoch": 0.1948111060537096,
      "grad_norm": 0.03681888058781624,
      "learning_rate": 4.834683153413459e-05,
      "loss": 11.9217,
      "step": 107
    },
    {
      "epoch": 0.19663177059626763,
      "grad_norm": 0.05845366790890694,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 11.9193,
      "step": 108
    },
    {
      "epoch": 0.19845243513882568,
      "grad_norm": 0.041342005133628845,
      "learning_rate": 4.669547078371504e-05,
      "loss": 11.9228,
      "step": 109
    },
    {
      "epoch": 0.2002730996813837,
      "grad_norm": 0.05567941814661026,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 11.9207,
      "step": 110
    },
    {
      "epoch": 0.20209376422394174,
      "grad_norm": 0.05213354527950287,
      "learning_rate": 4.504772348747687e-05,
      "loss": 11.9174,
      "step": 111
    },
    {
      "epoch": 0.20391442876649976,
      "grad_norm": 0.05773822218179703,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 11.9201,
      "step": 112
    },
    {
      "epoch": 0.20573509330905781,
      "grad_norm": 0.04012896120548248,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 11.9171,
      "step": 113
    },
    {
      "epoch": 0.20755575785161584,
      "grad_norm": 0.03765571862459183,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 11.921,
      "step": 114
    },
    {
      "epoch": 0.20937642239417387,
      "grad_norm": 0.04916438087821007,
      "learning_rate": 4.17702704859633e-05,
      "loss": 11.9183,
      "step": 115
    },
    {
      "epoch": 0.21119708693673192,
      "grad_norm": 0.06308427453041077,
      "learning_rate": 4.095597328339452e-05,
      "loss": 11.9179,
      "step": 116
    },
    {
      "epoch": 0.21301775147928995,
      "grad_norm": 0.06440264731645584,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 11.9203,
      "step": 117
    },
    {
      "epoch": 0.21483841602184797,
      "grad_norm": 0.04310610145330429,
      "learning_rate": 3.933501846281267e-05,
      "loss": 11.9183,
      "step": 118
    },
    {
      "epoch": 0.216659080564406,
      "grad_norm": 0.02499895729124546,
      "learning_rate": 3.852880399766243e-05,
      "loss": 11.9223,
      "step": 119
    },
    {
      "epoch": 0.21847974510696405,
      "grad_norm": 0.052417438477277756,
      "learning_rate": 3.772572564296005e-05,
      "loss": 11.9187,
      "step": 120
    },
    {
      "epoch": 0.22030040964952208,
      "grad_norm": 0.07833024114370346,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 11.9176,
      "step": 121
    },
    {
      "epoch": 0.2221210741920801,
      "grad_norm": 0.040134355425834656,
      "learning_rate": 3.612985456190778e-05,
      "loss": 11.9186,
      "step": 122
    },
    {
      "epoch": 0.22394173873463813,
      "grad_norm": 0.05635281652212143,
      "learning_rate": 3.533749813077677e-05,
      "loss": 11.9167,
      "step": 123
    },
    {
      "epoch": 0.22576240327719618,
      "grad_norm": 0.04688984528183937,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 11.9218,
      "step": 124
    },
    {
      "epoch": 0.2275830678197542,
      "grad_norm": 0.05027342960238457,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 11.9168,
      "step": 125
    },
    {
      "epoch": 0.22940373236231223,
      "grad_norm": 0.044746220111846924,
      "learning_rate": 3.298534127791785e-05,
      "loss": 11.918,
      "step": 126
    },
    {
      "epoch": 0.2312243969048703,
      "grad_norm": 0.05197661742568016,
      "learning_rate": 3.221030765387417e-05,
      "loss": 11.918,
      "step": 127
    },
    {
      "epoch": 0.23304506144742831,
      "grad_norm": 0.04914049431681633,
      "learning_rate": 3.144013755408895e-05,
      "loss": 11.9166,
      "step": 128
    },
    {
      "epoch": 0.23486572598998634,
      "grad_norm": 0.04531894996762276,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 11.9165,
      "step": 129
    },
    {
      "epoch": 0.23668639053254437,
      "grad_norm": 0.056826937943696976,
      "learning_rate": 2.991522876735154e-05,
      "loss": 11.9148,
      "step": 130
    },
    {
      "epoch": 0.23850705507510242,
      "grad_norm": 0.05140029639005661,
      "learning_rate": 2.916090697523549e-05,
      "loss": 11.9146,
      "step": 131
    },
    {
      "epoch": 0.24032771961766045,
      "grad_norm": 0.04223676770925522,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 11.9161,
      "step": 132
    },
    {
      "epoch": 0.24214838416021847,
      "grad_norm": 0.05060703679919243,
      "learning_rate": 2.766955965735968e-05,
      "loss": 11.9159,
      "step": 133
    },
    {
      "epoch": 0.24396904870277653,
      "grad_norm": 0.05956428498029709,
      "learning_rate": 2.693294185106562e-05,
      "loss": 11.9198,
      "step": 134
    },
    {
      "epoch": 0.24578971324533455,
      "grad_norm": 0.06842100620269775,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 11.9196,
      "step": 135
    },
    {
      "epoch": 0.24761037778789258,
      "grad_norm": 0.06286119669675827,
      "learning_rate": 2.547882480847461e-05,
      "loss": 11.9137,
      "step": 136
    },
    {
      "epoch": 0.2494310423304506,
      "grad_norm": 0.04629463702440262,
      "learning_rate": 2.476172311325783e-05,
      "loss": 11.9135,
      "step": 137
    },
    {
      "epoch": 0.25125170687300863,
      "grad_norm": 0.08288171142339706,
      "learning_rate": 2.405152131093926e-05,
      "loss": 11.915,
      "step": 138
    },
    {
      "epoch": 0.25307237141556665,
      "grad_norm": 0.05736103653907776,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 11.9104,
      "step": 139
    },
    {
      "epoch": 0.25489303595812474,
      "grad_norm": 0.06756237149238586,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 11.9124,
      "step": 140
    },
    {
      "epoch": 0.25671370050068276,
      "grad_norm": 0.06638480722904205,
      "learning_rate": 2.196424713241637e-05,
      "loss": 11.9117,
      "step": 141
    },
    {
      "epoch": 0.2585343650432408,
      "grad_norm": 0.09103255718946457,
      "learning_rate": 2.128356686585282e-05,
      "loss": 11.913,
      "step": 142
    },
    {
      "epoch": 0.2603550295857988,
      "grad_norm": 0.0980590358376503,
      "learning_rate": 2.061073738537635e-05,
      "loss": 11.9097,
      "step": 143
    },
    {
      "epoch": 0.26217569412835684,
      "grad_norm": 0.06975159049034119,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 11.9124,
      "step": 144
    },
    {
      "epoch": 0.26399635867091487,
      "grad_norm": 0.06156918779015541,
      "learning_rate": 1.928936436551661e-05,
      "loss": 11.9129,
      "step": 145
    },
    {
      "epoch": 0.2658170232134729,
      "grad_norm": 0.09512455016374588,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 11.9078,
      "step": 146
    },
    {
      "epoch": 0.267637687756031,
      "grad_norm": 0.0983736664056778,
      "learning_rate": 1.800157297483417e-05,
      "loss": 11.9101,
      "step": 147
    },
    {
      "epoch": 0.269458352298589,
      "grad_norm": 0.15988147258758545,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 11.8997,
      "step": 148
    },
    {
      "epoch": 0.271279016841147,
      "grad_norm": 0.1824086308479309,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 11.9061,
      "step": 149
    },
    {
      "epoch": 0.27309968138370505,
      "grad_norm": 0.17297427356243134,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 11.8977,
      "step": 150
    },
    {
      "epoch": 0.27309968138370505,
      "eval_loss": 11.91677474975586,
      "eval_runtime": 5.0403,
      "eval_samples_per_second": 183.522,
      "eval_steps_per_second": 46.029,
      "step": 150
    },
    {
      "epoch": 0.2749203459262631,
      "grad_norm": 0.054472945630550385,
      "learning_rate": 1.553232954407171e-05,
      "loss": 11.9221,
      "step": 151
    },
    {
      "epoch": 0.2767410104688211,
      "grad_norm": 0.06646860390901566,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 11.9219,
      "step": 152
    },
    {
      "epoch": 0.27856167501137913,
      "grad_norm": 0.036392826586961746,
      "learning_rate": 1.435357758543015e-05,
      "loss": 11.9211,
      "step": 153
    },
    {
      "epoch": 0.2803823395539372,
      "grad_norm": 0.04787255451083183,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 11.9228,
      "step": 154
    },
    {
      "epoch": 0.28220300409649524,
      "grad_norm": 0.058999769389629364,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 11.921,
      "step": 155
    },
    {
      "epoch": 0.28402366863905326,
      "grad_norm": 0.06873305886983871,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 11.9196,
      "step": 156
    },
    {
      "epoch": 0.2858443331816113,
      "grad_norm": 0.07329442352056503,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 11.9169,
      "step": 157
    },
    {
      "epoch": 0.2876649977241693,
      "grad_norm": 0.046734023839235306,
      "learning_rate": 1.157994445715706e-05,
      "loss": 11.9196,
      "step": 158
    },
    {
      "epoch": 0.28948566226672734,
      "grad_norm": 0.06411699950695038,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 11.9197,
      "step": 159
    },
    {
      "epoch": 0.29130632680928537,
      "grad_norm": 0.062398821115493774,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 11.9219,
      "step": 160
    },
    {
      "epoch": 0.29312699135184345,
      "grad_norm": 0.04219827428460121,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 11.9192,
      "step": 161
    },
    {
      "epoch": 0.2949476558944015,
      "grad_norm": 0.05368563532829285,
      "learning_rate": 9.549150281252633e-06,
      "loss": 11.9167,
      "step": 162
    },
    {
      "epoch": 0.2967683204369595,
      "grad_norm": 0.06402570754289627,
      "learning_rate": 9.068759265665384e-06,
      "loss": 11.9202,
      "step": 163
    },
    {
      "epoch": 0.2985889849795175,
      "grad_norm": 0.059248920530080795,
      "learning_rate": 8.599558442598998e-06,
      "loss": 11.9166,
      "step": 164
    },
    {
      "epoch": 0.30040964952207555,
      "grad_norm": 0.04016771912574768,
      "learning_rate": 8.141676086873572e-06,
      "loss": 11.9199,
      "step": 165
    },
    {
      "epoch": 0.3022303140646336,
      "grad_norm": 0.054599128663539886,
      "learning_rate": 7.695237378953223e-06,
      "loss": 11.9169,
      "step": 166
    },
    {
      "epoch": 0.3040509786071916,
      "grad_norm": 0.05375150218605995,
      "learning_rate": 7.260364370723044e-06,
      "loss": 11.9176,
      "step": 167
    },
    {
      "epoch": 0.3058716431497497,
      "grad_norm": 0.04632212594151497,
      "learning_rate": 6.837175952121306e-06,
      "loss": 11.9198,
      "step": 168
    },
    {
      "epoch": 0.3076923076923077,
      "grad_norm": 0.0518006756901741,
      "learning_rate": 6.425787818636131e-06,
      "loss": 11.9178,
      "step": 169
    },
    {
      "epoch": 0.30951297223486574,
      "grad_norm": 0.05097333341836929,
      "learning_rate": 6.026312439675552e-06,
      "loss": 11.9206,
      "step": 170
    },
    {
      "epoch": 0.31133363677742376,
      "grad_norm": 0.05397026240825653,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 11.9175,
      "step": 171
    },
    {
      "epoch": 0.3131543013199818,
      "grad_norm": 0.06216050311923027,
      "learning_rate": 5.263533508961827e-06,
      "loss": 11.9177,
      "step": 172
    },
    {
      "epoch": 0.3149749658625398,
      "grad_norm": 0.052663519978523254,
      "learning_rate": 4.900438493352055e-06,
      "loss": 11.9203,
      "step": 173
    },
    {
      "epoch": 0.31679563040509784,
      "grad_norm": 0.06635887920856476,
      "learning_rate": 4.549673247541875e-06,
      "loss": 11.9143,
      "step": 174
    },
    {
      "epoch": 0.3186162949476559,
      "grad_norm": 0.07404106110334396,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 11.9174,
      "step": 175
    },
    {
      "epoch": 0.32043695949021395,
      "grad_norm": 0.04695393890142441,
      "learning_rate": 3.885512251130763e-06,
      "loss": 11.915,
      "step": 176
    },
    {
      "epoch": 0.32225762403277197,
      "grad_norm": 0.03491644933819771,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 11.9182,
      "step": 177
    },
    {
      "epoch": 0.32407828857533,
      "grad_norm": 0.059455081820487976,
      "learning_rate": 3.271776770026963e-06,
      "loss": 11.9159,
      "step": 178
    },
    {
      "epoch": 0.325898953117888,
      "grad_norm": 0.0618712492287159,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 11.9125,
      "step": 179
    },
    {
      "epoch": 0.32771961766044605,
      "grad_norm": 0.053199317306280136,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 11.9191,
      "step": 180
    },
    {
      "epoch": 0.3295402822030041,
      "grad_norm": 0.06174110621213913,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 11.9122,
      "step": 181
    },
    {
      "epoch": 0.33136094674556216,
      "grad_norm": 0.060066673904657364,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 11.9147,
      "step": 182
    },
    {
      "epoch": 0.3331816112881202,
      "grad_norm": 0.09699099510908127,
      "learning_rate": 1.962316193157593e-06,
      "loss": 11.9165,
      "step": 183
    },
    {
      "epoch": 0.3350022758306782,
      "grad_norm": 0.0592498853802681,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 11.9158,
      "step": 184
    },
    {
      "epoch": 0.33682294037323623,
      "grad_norm": 0.058085668832063675,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 11.9133,
      "step": 185
    },
    {
      "epoch": 0.33864360491579426,
      "grad_norm": 0.06665123999118805,
      "learning_rate": 1.333670137599713e-06,
      "loss": 11.9161,
      "step": 186
    },
    {
      "epoch": 0.3404642694583523,
      "grad_norm": 0.08766186237335205,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 11.9147,
      "step": 187
    },
    {
      "epoch": 0.3422849340009103,
      "grad_norm": 0.09656717628240585,
      "learning_rate": 9.810017062595322e-07,
      "loss": 11.9125,
      "step": 188
    },
    {
      "epoch": 0.34410559854346834,
      "grad_norm": 0.05834021791815758,
      "learning_rate": 8.247462563808817e-07,
      "loss": 11.9139,
      "step": 189
    },
    {
      "epoch": 0.3459262630860264,
      "grad_norm": 0.0666501596570015,
      "learning_rate": 6.819348298638839e-07,
      "loss": 11.9167,
      "step": 190
    },
    {
      "epoch": 0.34774692762858445,
      "grad_norm": 0.06928762048482895,
      "learning_rate": 5.526064699265753e-07,
      "loss": 11.9133,
      "step": 191
    },
    {
      "epoch": 0.34956759217114247,
      "grad_norm": 0.09626271575689316,
      "learning_rate": 4.367965336512403e-07,
      "loss": 11.9158,
      "step": 192
    },
    {
      "epoch": 0.3513882567137005,
      "grad_norm": 0.07570219784975052,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 11.9133,
      "step": 193
    },
    {
      "epoch": 0.3532089212562585,
      "grad_norm": 0.06003417819738388,
      "learning_rate": 2.458548727494292e-07,
      "loss": 11.9131,
      "step": 194
    },
    {
      "epoch": 0.35502958579881655,
      "grad_norm": 0.09867273271083832,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 11.9096,
      "step": 195
    },
    {
      "epoch": 0.3568502503413746,
      "grad_norm": 0.10562506318092346,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 11.9041,
      "step": 196
    },
    {
      "epoch": 0.35867091488393266,
      "grad_norm": 0.11728859692811966,
      "learning_rate": 6.150154258476315e-08,
      "loss": 11.9091,
      "step": 197
    },
    {
      "epoch": 0.3604915794264907,
      "grad_norm": 0.10915789753198624,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 11.9103,
      "step": 198
    },
    {
      "epoch": 0.3623122439690487,
      "grad_norm": 0.16178463399410248,
      "learning_rate": 6.834750376549792e-09,
      "loss": 11.9047,
      "step": 199
    },
    {
      "epoch": 0.36413290851160673,
      "grad_norm": 0.2064269334077835,
      "learning_rate": 0.0,
      "loss": 11.9007,
      "step": 200
    },
    {
      "epoch": 0.36413290851160673,
      "eval_loss": 11.916446685791016,
      "eval_runtime": 5.0395,
      "eval_samples_per_second": 183.551,
      "eval_steps_per_second": 46.037,
      "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": 964165632000.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}