{
  "best_metric": 0.3340291678905487,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.016073294221650727,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 8.036647110825364e-05,
      "grad_norm": 0.18612001836299896,
      "learning_rate": 1e-05,
      "loss": 0.3914,
      "step": 1
    },
    {
      "epoch": 8.036647110825364e-05,
      "eval_loss": 1.0249946117401123,
      "eval_runtime": 1497.5594,
      "eval_samples_per_second": 13.994,
      "eval_steps_per_second": 3.499,
      "step": 1
    },
    {
      "epoch": 0.00016073294221650728,
      "grad_norm": 0.2979937493801117,
      "learning_rate": 2e-05,
      "loss": 0.5491,
      "step": 2
    },
    {
      "epoch": 0.0002410994133247609,
      "grad_norm": 0.38420426845550537,
      "learning_rate": 3e-05,
      "loss": 0.7028,
      "step": 3
    },
    {
      "epoch": 0.00032146588443301456,
      "grad_norm": 0.3543547987937927,
      "learning_rate": 4e-05,
      "loss": 0.6901,
      "step": 4
    },
    {
      "epoch": 0.00040183235554126816,
      "grad_norm": 0.348410964012146,
      "learning_rate": 5e-05,
      "loss": 0.5656,
      "step": 5
    },
    {
      "epoch": 0.0004821988266495218,
      "grad_norm": 0.36390647292137146,
      "learning_rate": 6e-05,
      "loss": 0.4986,
      "step": 6
    },
    {
      "epoch": 0.0005625652977577755,
      "grad_norm": 0.3201482892036438,
      "learning_rate": 7e-05,
      "loss": 0.5166,
      "step": 7
    },
    {
      "epoch": 0.0006429317688660291,
      "grad_norm": 0.3475126624107361,
      "learning_rate": 8e-05,
      "loss": 0.5452,
      "step": 8
    },
    {
      "epoch": 0.0007232982399742828,
      "grad_norm": 0.30419087409973145,
      "learning_rate": 9e-05,
      "loss": 0.6291,
      "step": 9
    },
    {
      "epoch": 0.0008036647110825363,
      "grad_norm": 0.3698471486568451,
      "learning_rate": 0.0001,
      "loss": 0.4441,
      "step": 10
    },
    {
      "epoch": 0.00088403118219079,
      "grad_norm": 0.41788730025291443,
      "learning_rate": 9.999316524962345e-05,
      "loss": 0.5673,
      "step": 11
    },
    {
      "epoch": 0.0009643976532990436,
      "grad_norm": 0.40829992294311523,
      "learning_rate": 9.997266286704631e-05,
      "loss": 0.639,
      "step": 12
    },
    {
      "epoch": 0.0010447641244072973,
      "grad_norm": 0.42904797196388245,
      "learning_rate": 9.993849845741524e-05,
      "loss": 0.4346,
      "step": 13
    },
    {
      "epoch": 0.001125130595515551,
      "grad_norm": 0.3262033462524414,
      "learning_rate": 9.989068136093873e-05,
      "loss": 0.379,
      "step": 14
    },
    {
      "epoch": 0.0012054970666238046,
      "grad_norm": 0.4094715416431427,
      "learning_rate": 9.98292246503335e-05,
      "loss": 0.5045,
      "step": 15
    },
    {
      "epoch": 0.0012858635377320583,
      "grad_norm": 0.489749938249588,
      "learning_rate": 9.975414512725057e-05,
      "loss": 0.4833,
      "step": 16
    },
    {
      "epoch": 0.001366230008840312,
      "grad_norm": 0.30741751194000244,
      "learning_rate": 9.966546331768191e-05,
      "loss": 0.3298,
      "step": 17
    },
    {
      "epoch": 0.0014465964799485656,
      "grad_norm": 0.38450294733047485,
      "learning_rate": 9.956320346634876e-05,
      "loss": 0.3483,
      "step": 18
    },
    {
      "epoch": 0.001526962951056819,
      "grad_norm": 0.33659470081329346,
      "learning_rate": 9.944739353007344e-05,
      "loss": 0.3724,
      "step": 19
    },
    {
      "epoch": 0.0016073294221650727,
      "grad_norm": 0.3183864653110504,
      "learning_rate": 9.931806517013612e-05,
      "loss": 0.381,
      "step": 20
    },
    {
      "epoch": 0.0016876958932733263,
      "grad_norm": 0.4514175355434418,
      "learning_rate": 9.917525374361912e-05,
      "loss": 0.4951,
      "step": 21
    },
    {
      "epoch": 0.00176806236438158,
      "grad_norm": 0.3846685290336609,
      "learning_rate": 9.901899829374047e-05,
      "loss": 0.41,
      "step": 22
    },
    {
      "epoch": 0.0018484288354898336,
      "grad_norm": 0.3974553942680359,
      "learning_rate": 9.884934153917997e-05,
      "loss": 0.3608,
      "step": 23
    },
    {
      "epoch": 0.0019287953065980873,
      "grad_norm": 0.33904963731765747,
      "learning_rate": 9.86663298624003e-05,
      "loss": 0.3576,
      "step": 24
    },
    {
      "epoch": 0.002009161777706341,
      "grad_norm": 0.38675615191459656,
      "learning_rate": 9.847001329696653e-05,
      "loss": 0.4196,
      "step": 25
    },
    {
      "epoch": 0.0020895282488145946,
      "grad_norm": 0.41639384627342224,
      "learning_rate": 9.826044551386744e-05,
      "loss": 0.3377,
      "step": 26
    },
    {
      "epoch": 0.0021698947199228482,
      "grad_norm": 0.29107943177223206,
      "learning_rate": 9.803768380684242e-05,
      "loss": 0.4185,
      "step": 27
    },
    {
      "epoch": 0.002250261191031102,
      "grad_norm": 0.37938907742500305,
      "learning_rate": 9.780178907671789e-05,
      "loss": 0.4009,
      "step": 28
    },
    {
      "epoch": 0.0023306276621393555,
      "grad_norm": 0.24359409511089325,
      "learning_rate": 9.755282581475769e-05,
      "loss": 0.2886,
      "step": 29
    },
    {
      "epoch": 0.002410994133247609,
      "grad_norm": 0.39020296931266785,
      "learning_rate": 9.729086208503174e-05,
      "loss": 0.3967,
      "step": 30
    },
    {
      "epoch": 0.002491360604355863,
      "grad_norm": 0.4349125027656555,
      "learning_rate": 9.701596950580806e-05,
      "loss": 0.2688,
      "step": 31
    },
    {
      "epoch": 0.0025717270754641165,
      "grad_norm": 0.3661207854747772,
      "learning_rate": 9.672822322997305e-05,
      "loss": 0.4181,
      "step": 32
    },
    {
      "epoch": 0.00265209354657237,
      "grad_norm": 0.5847356915473938,
      "learning_rate": 9.642770192448536e-05,
      "loss": 0.4327,
      "step": 33
    },
    {
      "epoch": 0.002732460017680624,
      "grad_norm": 0.3047882616519928,
      "learning_rate": 9.611448774886924e-05,
      "loss": 0.3267,
      "step": 34
    },
    {
      "epoch": 0.0028128264887888775,
      "grad_norm": 0.5811073780059814,
      "learning_rate": 9.578866633275288e-05,
      "loss": 0.4012,
      "step": 35
    },
    {
      "epoch": 0.002893192959897131,
      "grad_norm": 0.4881095588207245,
      "learning_rate": 9.545032675245813e-05,
      "loss": 0.3371,
      "step": 36
    },
    {
      "epoch": 0.0029735594310053843,
      "grad_norm": 0.5671091079711914,
      "learning_rate": 9.509956150664796e-05,
      "loss": 0.489,
      "step": 37
    },
    {
      "epoch": 0.003053925902113638,
      "grad_norm": 0.3962863087654114,
      "learning_rate": 9.473646649103818e-05,
      "loss": 0.3521,
      "step": 38
    },
    {
      "epoch": 0.0031342923732218917,
      "grad_norm": 0.3788892924785614,
      "learning_rate": 9.43611409721806e-05,
      "loss": 0.3665,
      "step": 39
    },
    {
      "epoch": 0.0032146588443301453,
      "grad_norm": 0.457553893327713,
      "learning_rate": 9.397368756032445e-05,
      "loss": 0.2819,
      "step": 40
    },
    {
      "epoch": 0.003295025315438399,
      "grad_norm": 0.6342529058456421,
      "learning_rate": 9.357421218136386e-05,
      "loss": 0.4221,
      "step": 41
    },
    {
      "epoch": 0.0033753917865466526,
      "grad_norm": 0.5456969738006592,
      "learning_rate": 9.316282404787871e-05,
      "loss": 0.4128,
      "step": 42
    },
    {
      "epoch": 0.0034557582576549063,
      "grad_norm": 0.5587855577468872,
      "learning_rate": 9.273963562927695e-05,
      "loss": 0.3647,
      "step": 43
    },
    {
      "epoch": 0.00353612472876316,
      "grad_norm": 0.9418520331382751,
      "learning_rate": 9.230476262104677e-05,
      "loss": 0.4885,
      "step": 44
    },
    {
      "epoch": 0.0036164911998714136,
      "grad_norm": 0.44635075330734253,
      "learning_rate": 9.185832391312644e-05,
      "loss": 0.4629,
      "step": 45
    },
    {
      "epoch": 0.0036968576709796672,
      "grad_norm": 0.4817911982536316,
      "learning_rate": 9.140044155740101e-05,
      "loss": 0.4139,
      "step": 46
    },
    {
      "epoch": 0.003777224142087921,
      "grad_norm": 0.514627993106842,
      "learning_rate": 9.093124073433463e-05,
      "loss": 0.4357,
      "step": 47
    },
    {
      "epoch": 0.0038575906131961745,
      "grad_norm": 0.3345630466938019,
      "learning_rate": 9.045084971874738e-05,
      "loss": 0.3273,
      "step": 48
    },
    {
      "epoch": 0.003937957084304428,
      "grad_norm": 0.9049527049064636,
      "learning_rate": 8.995939984474624e-05,
      "loss": 0.4914,
      "step": 49
    },
    {
      "epoch": 0.004018323555412682,
      "grad_norm": 0.5089755654335022,
      "learning_rate": 8.945702546981969e-05,
      "loss": 0.5391,
      "step": 50
    },
    {
      "epoch": 0.004018323555412682,
      "eval_loss": 0.3987460136413574,
      "eval_runtime": 1507.5251,
      "eval_samples_per_second": 13.902,
      "eval_steps_per_second": 3.476,
      "step": 50
    },
    {
      "epoch": 0.004098690026520935,
      "grad_norm": 0.2140122354030609,
      "learning_rate": 8.894386393810563e-05,
      "loss": 0.3368,
      "step": 51
    },
    {
      "epoch": 0.004179056497629189,
      "grad_norm": 0.1737431138753891,
      "learning_rate": 8.842005554284296e-05,
      "loss": 0.2711,
      "step": 52
    },
    {
      "epoch": 0.004259422968737442,
      "grad_norm": 0.19253864884376526,
      "learning_rate": 8.788574348801675e-05,
      "loss": 0.4067,
      "step": 53
    },
    {
      "epoch": 0.0043397894398456965,
      "grad_norm": 0.2398878037929535,
      "learning_rate": 8.73410738492077e-05,
      "loss": 0.3936,
      "step": 54
    },
    {
      "epoch": 0.00442015591095395,
      "grad_norm": 0.3259638547897339,
      "learning_rate": 8.678619553365659e-05,
      "loss": 0.3475,
      "step": 55
    },
    {
      "epoch": 0.004500522382062204,
      "grad_norm": 0.24697811901569366,
      "learning_rate": 8.622126023955446e-05,
      "loss": 0.3244,
      "step": 56
    },
    {
      "epoch": 0.004580888853170457,
      "grad_norm": 0.2508927285671234,
      "learning_rate": 8.564642241456986e-05,
      "loss": 0.4237,
      "step": 57
    },
    {
      "epoch": 0.004661255324278711,
      "grad_norm": 0.20806628465652466,
      "learning_rate": 8.506183921362443e-05,
      "loss": 0.3321,
      "step": 58
    },
    {
      "epoch": 0.004741621795386964,
      "grad_norm": 0.277308851480484,
      "learning_rate": 8.44676704559283e-05,
      "loss": 0.3384,
      "step": 59
    },
    {
      "epoch": 0.004821988266495218,
      "grad_norm": 0.3705958127975464,
      "learning_rate": 8.386407858128706e-05,
      "loss": 0.4142,
      "step": 60
    },
    {
      "epoch": 0.004902354737603472,
      "grad_norm": 0.23256395757198334,
      "learning_rate": 8.32512286056924e-05,
      "loss": 0.3226,
      "step": 61
    },
    {
      "epoch": 0.004982721208711726,
      "grad_norm": 0.29993152618408203,
      "learning_rate": 8.262928807620843e-05,
      "loss": 0.2895,
      "step": 62
    },
    {
      "epoch": 0.005063087679819979,
      "grad_norm": 0.2591308653354645,
      "learning_rate": 8.199842702516583e-05,
      "loss": 0.4285,
      "step": 63
    },
    {
      "epoch": 0.005143454150928233,
      "grad_norm": 0.296428918838501,
      "learning_rate": 8.135881792367686e-05,
      "loss": 0.2918,
      "step": 64
    },
    {
      "epoch": 0.005223820622036486,
      "grad_norm": 0.2771739065647125,
      "learning_rate": 8.07106356344834e-05,
      "loss": 0.289,
      "step": 65
    },
    {
      "epoch": 0.00530418709314474,
      "grad_norm": 0.4175921678543091,
      "learning_rate": 8.005405736415126e-05,
      "loss": 0.3932,
      "step": 66
    },
    {
      "epoch": 0.0053845535642529935,
      "grad_norm": 0.3970695734024048,
      "learning_rate": 7.938926261462366e-05,
      "loss": 0.3182,
      "step": 67
    },
    {
      "epoch": 0.005464920035361248,
      "grad_norm": 0.27005404233932495,
      "learning_rate": 7.871643313414718e-05,
      "loss": 0.3167,
      "step": 68
    },
    {
      "epoch": 0.005545286506469501,
      "grad_norm": 0.25059813261032104,
      "learning_rate": 7.803575286758364e-05,
      "loss": 0.3043,
      "step": 69
    },
    {
      "epoch": 0.005625652977577755,
      "grad_norm": 0.2868187725543976,
      "learning_rate": 7.734740790612136e-05,
      "loss": 0.4342,
      "step": 70
    },
    {
      "epoch": 0.005706019448686008,
      "grad_norm": 0.2638663351535797,
      "learning_rate": 7.66515864363997e-05,
      "loss": 0.2654,
      "step": 71
    },
    {
      "epoch": 0.005786385919794262,
      "grad_norm": 0.300005167722702,
      "learning_rate": 7.594847868906076e-05,
      "loss": 0.3505,
      "step": 72
    },
    {
      "epoch": 0.0058667523909025155,
      "grad_norm": 0.36400383710861206,
      "learning_rate": 7.52382768867422e-05,
      "loss": 0.354,
      "step": 73
    },
    {
      "epoch": 0.005947118862010769,
      "grad_norm": 0.3459414839744568,
      "learning_rate": 7.452117519152542e-05,
      "loss": 0.2994,
      "step": 74
    },
    {
      "epoch": 0.006027485333119023,
      "grad_norm": 0.3513219952583313,
      "learning_rate": 7.379736965185368e-05,
      "loss": 0.4194,
      "step": 75
    },
    {
      "epoch": 0.006107851804227276,
      "grad_norm": 0.3714442849159241,
      "learning_rate": 7.30670581489344e-05,
      "loss": 0.448,
      "step": 76
    },
    {
      "epoch": 0.00618821827533553,
      "grad_norm": 0.2990541458129883,
      "learning_rate": 7.233044034264034e-05,
      "loss": 0.4248,
      "step": 77
    },
    {
      "epoch": 0.006268584746443783,
      "grad_norm": 0.49857664108276367,
      "learning_rate": 7.158771761692464e-05,
      "loss": 0.4436,
      "step": 78
    },
    {
      "epoch": 0.006348951217552037,
      "grad_norm": 0.5330504775047302,
      "learning_rate": 7.083909302476453e-05,
      "loss": 0.3596,
      "step": 79
    },
    {
      "epoch": 0.006429317688660291,
      "grad_norm": 0.43941548466682434,
      "learning_rate": 7.008477123264848e-05,
      "loss": 0.3902,
      "step": 80
    },
    {
      "epoch": 0.006509684159768545,
      "grad_norm": 0.3333449363708496,
      "learning_rate": 6.932495846462261e-05,
      "loss": 0.3514,
      "step": 81
    },
    {
      "epoch": 0.006590050630876798,
      "grad_norm": 0.49027225375175476,
      "learning_rate": 6.855986244591104e-05,
      "loss": 0.4499,
      "step": 82
    },
    {
      "epoch": 0.006670417101985052,
      "grad_norm": 0.41234147548675537,
      "learning_rate": 6.778969234612584e-05,
      "loss": 0.427,
      "step": 83
    },
    {
      "epoch": 0.006750783573093305,
      "grad_norm": 0.4618304967880249,
      "learning_rate": 6.701465872208216e-05,
      "loss": 0.3542,
      "step": 84
    },
    {
      "epoch": 0.006831150044201559,
      "grad_norm": 0.4090670049190521,
      "learning_rate": 6.623497346023418e-05,
      "loss": 0.4029,
      "step": 85
    },
    {
      "epoch": 0.0069115165153098125,
      "grad_norm": 0.3413718640804291,
      "learning_rate": 6.545084971874738e-05,
      "loss": 0.4132,
      "step": 86
    },
    {
      "epoch": 0.006991882986418067,
      "grad_norm": 0.3925929665565491,
      "learning_rate": 6.466250186922325e-05,
      "loss": 0.3194,
      "step": 87
    },
    {
      "epoch": 0.00707224945752632,
      "grad_norm": 0.3401626944541931,
      "learning_rate": 6.387014543809223e-05,
      "loss": 0.237,
      "step": 88
    },
    {
      "epoch": 0.007152615928634574,
      "grad_norm": 0.3158605992794037,
      "learning_rate": 6.307399704769099e-05,
      "loss": 0.3277,
      "step": 89
    },
    {
      "epoch": 0.007232982399742827,
      "grad_norm": 0.29103630781173706,
      "learning_rate": 6.227427435703997e-05,
      "loss": 0.3033,
      "step": 90
    },
    {
      "epoch": 0.007313348870851081,
      "grad_norm": 0.46478354930877686,
      "learning_rate": 6.147119600233758e-05,
      "loss": 0.3307,
      "step": 91
    },
    {
      "epoch": 0.0073937153419593345,
      "grad_norm": 0.49609261751174927,
      "learning_rate": 6.066498153718735e-05,
      "loss": 0.3853,
      "step": 92
    },
    {
      "epoch": 0.0074740818130675886,
      "grad_norm": 0.522725522518158,
      "learning_rate": 5.985585137257401e-05,
      "loss": 0.4172,
      "step": 93
    },
    {
      "epoch": 0.007554448284175842,
      "grad_norm": 0.7370002269744873,
      "learning_rate": 5.90440267166055e-05,
      "loss": 0.4695,
      "step": 94
    },
    {
      "epoch": 0.007634814755284096,
      "grad_norm": 0.7781453728675842,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 0.401,
      "step": 95
    },
    {
      "epoch": 0.007715181226392349,
      "grad_norm": 0.5715779066085815,
      "learning_rate": 5.74131823855921e-05,
      "loss": 0.3563,
      "step": 96
    },
    {
      "epoch": 0.007795547697500603,
      "grad_norm": 0.5776063799858093,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 0.373,
      "step": 97
    },
    {
      "epoch": 0.007875914168608856,
      "grad_norm": 0.573598325252533,
      "learning_rate": 5.577423184847932e-05,
      "loss": 0.3351,
      "step": 98
    },
    {
      "epoch": 0.00795628063971711,
      "grad_norm": 0.6036813855171204,
      "learning_rate": 5.495227651252315e-05,
      "loss": 0.5931,
      "step": 99
    },
    {
      "epoch": 0.008036647110825364,
      "grad_norm": 0.5221275687217712,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 0.3439,
      "step": 100
    },
    {
      "epoch": 0.008036647110825364,
      "eval_loss": 0.3674781024456024,
      "eval_runtime": 1505.53,
      "eval_samples_per_second": 13.92,
      "eval_steps_per_second": 3.481,
      "step": 100
    },
    {
      "epoch": 0.008117013581933617,
      "grad_norm": 0.1746017336845398,
      "learning_rate": 5.330452921628497e-05,
      "loss": 0.4119,
      "step": 101
    },
    {
      "epoch": 0.00819738005304187,
      "grad_norm": 0.3522207736968994,
      "learning_rate": 5.247918773366112e-05,
      "loss": 0.5488,
      "step": 102
    },
    {
      "epoch": 0.008277746524150125,
      "grad_norm": 0.24777081608772278,
      "learning_rate": 5.165316846586541e-05,
      "loss": 0.3045,
      "step": 103
    },
    {
      "epoch": 0.008358112995258378,
      "grad_norm": 0.3415620028972626,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 0.4087,
      "step": 104
    },
    {
      "epoch": 0.008438479466366632,
      "grad_norm": 0.250539630651474,
      "learning_rate": 5e-05,
      "loss": 0.3469,
      "step": 105
    },
    {
      "epoch": 0.008518845937474885,
      "grad_norm": 0.31240588426589966,
      "learning_rate": 4.917330276168208e-05,
      "loss": 0.3877,
      "step": 106
    },
    {
      "epoch": 0.00859921240858314,
      "grad_norm": 0.29493534564971924,
      "learning_rate": 4.834683153413459e-05,
      "loss": 0.3546,
      "step": 107
    },
    {
      "epoch": 0.008679578879691393,
      "grad_norm": 0.24933046102523804,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 0.3164,
      "step": 108
    },
    {
      "epoch": 0.008759945350799646,
      "grad_norm": 0.3212600350379944,
      "learning_rate": 4.669547078371504e-05,
      "loss": 0.4184,
      "step": 109
    },
    {
      "epoch": 0.0088403118219079,
      "grad_norm": 0.4264766573905945,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 0.3748,
      "step": 110
    },
    {
      "epoch": 0.008920678293016154,
      "grad_norm": 0.30657777190208435,
      "learning_rate": 4.504772348747687e-05,
      "loss": 0.2886,
      "step": 111
    },
    {
      "epoch": 0.009001044764124408,
      "grad_norm": 0.3119131922721863,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 0.3675,
      "step": 112
    },
    {
      "epoch": 0.00908141123523266,
      "grad_norm": 0.3247963786125183,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 0.3164,
      "step": 113
    },
    {
      "epoch": 0.009161777706340914,
      "grad_norm": 0.34584736824035645,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 0.3701,
      "step": 114
    },
    {
      "epoch": 0.009242144177449169,
      "grad_norm": 0.3761036992073059,
      "learning_rate": 4.17702704859633e-05,
      "loss": 0.3521,
      "step": 115
    },
    {
      "epoch": 0.009322510648557422,
      "grad_norm": 0.2861977815628052,
      "learning_rate": 4.095597328339452e-05,
      "loss": 0.2289,
      "step": 116
    },
    {
      "epoch": 0.009402877119665675,
      "grad_norm": 0.32266029715538025,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 0.3258,
      "step": 117
    },
    {
      "epoch": 0.009483243590773929,
      "grad_norm": 0.2621302306652069,
      "learning_rate": 3.933501846281267e-05,
      "loss": 0.2928,
      "step": 118
    },
    {
      "epoch": 0.009563610061882184,
      "grad_norm": 0.30788081884384155,
      "learning_rate": 3.852880399766243e-05,
      "loss": 0.2959,
      "step": 119
    },
    {
      "epoch": 0.009643976532990437,
      "grad_norm": 0.3848606050014496,
      "learning_rate": 3.772572564296005e-05,
      "loss": 0.3143,
      "step": 120
    },
    {
      "epoch": 0.00972434300409869,
      "grad_norm": 0.5152484178543091,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 0.4244,
      "step": 121
    },
    {
      "epoch": 0.009804709475206943,
      "grad_norm": 0.28096693754196167,
      "learning_rate": 3.612985456190778e-05,
      "loss": 0.2511,
      "step": 122
    },
    {
      "epoch": 0.009885075946315196,
      "grad_norm": 0.34794381260871887,
      "learning_rate": 3.533749813077677e-05,
      "loss": 0.2989,
      "step": 123
    },
    {
      "epoch": 0.009965442417423451,
      "grad_norm": 0.3060310184955597,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 0.2905,
      "step": 124
    },
    {
      "epoch": 0.010045808888531705,
      "grad_norm": 0.4070822298526764,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 0.3016,
      "step": 125
    },
    {
      "epoch": 0.010126175359639958,
      "grad_norm": 0.3588668704032898,
      "learning_rate": 3.298534127791785e-05,
      "loss": 0.3573,
      "step": 126
    },
    {
      "epoch": 0.010206541830748211,
      "grad_norm": 0.45010945200920105,
      "learning_rate": 3.221030765387417e-05,
      "loss": 0.2986,
      "step": 127
    },
    {
      "epoch": 0.010286908301856466,
      "grad_norm": 0.41504713892936707,
      "learning_rate": 3.144013755408895e-05,
      "loss": 0.2613,
      "step": 128
    },
    {
      "epoch": 0.01036727477296472,
      "grad_norm": 0.4334265887737274,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 0.408,
      "step": 129
    },
    {
      "epoch": 0.010447641244072972,
      "grad_norm": 0.41168826818466187,
      "learning_rate": 2.991522876735154e-05,
      "loss": 0.2994,
      "step": 130
    },
    {
      "epoch": 0.010528007715181226,
      "grad_norm": 0.6202244162559509,
      "learning_rate": 2.916090697523549e-05,
      "loss": 0.4726,
      "step": 131
    },
    {
      "epoch": 0.01060837418628948,
      "grad_norm": 0.5010877847671509,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 0.3594,
      "step": 132
    },
    {
      "epoch": 0.010688740657397734,
      "grad_norm": 0.43364813923835754,
      "learning_rate": 2.766955965735968e-05,
      "loss": 0.2589,
      "step": 133
    },
    {
      "epoch": 0.010769107128505987,
      "grad_norm": 0.4264422655105591,
      "learning_rate": 2.693294185106562e-05,
      "loss": 0.2858,
      "step": 134
    },
    {
      "epoch": 0.01084947359961424,
      "grad_norm": 0.5041791796684265,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 0.298,
      "step": 135
    },
    {
      "epoch": 0.010929840070722495,
      "grad_norm": 0.5244119167327881,
      "learning_rate": 2.547882480847461e-05,
      "loss": 0.3159,
      "step": 136
    },
    {
      "epoch": 0.011010206541830748,
      "grad_norm": 0.5391702651977539,
      "learning_rate": 2.476172311325783e-05,
      "loss": 0.4065,
      "step": 137
    },
    {
      "epoch": 0.011090573012939002,
      "grad_norm": 0.5578972101211548,
      "learning_rate": 2.405152131093926e-05,
      "loss": 0.4387,
      "step": 138
    },
    {
      "epoch": 0.011170939484047255,
      "grad_norm": 0.46875637769699097,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 0.3113,
      "step": 139
    },
    {
      "epoch": 0.01125130595515551,
      "grad_norm": 0.9138067364692688,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 0.3789,
      "step": 140
    },
    {
      "epoch": 0.011331672426263763,
      "grad_norm": 0.4726915955543518,
      "learning_rate": 2.196424713241637e-05,
      "loss": 0.2958,
      "step": 141
    },
    {
      "epoch": 0.011412038897372016,
      "grad_norm": 0.48181626200675964,
      "learning_rate": 2.128356686585282e-05,
      "loss": 0.3791,
      "step": 142
    },
    {
      "epoch": 0.01149240536848027,
      "grad_norm": 0.7033829092979431,
      "learning_rate": 2.061073738537635e-05,
      "loss": 0.3671,
      "step": 143
    },
    {
      "epoch": 0.011572771839588524,
      "grad_norm": 0.5564422011375427,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 0.3811,
      "step": 144
    },
    {
      "epoch": 0.011653138310696778,
      "grad_norm": 0.7714976072311401,
      "learning_rate": 1.928936436551661e-05,
      "loss": 0.2818,
      "step": 145
    },
    {
      "epoch": 0.011733504781805031,
      "grad_norm": 0.43348315358161926,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 0.2385,
      "step": 146
    },
    {
      "epoch": 0.011813871252913284,
      "grad_norm": 0.5189164280891418,
      "learning_rate": 1.800157297483417e-05,
      "loss": 0.2692,
      "step": 147
    },
    {
      "epoch": 0.011894237724021537,
      "grad_norm": 0.7366738319396973,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 0.5457,
      "step": 148
    },
    {
      "epoch": 0.011974604195129792,
      "grad_norm": 0.6651752591133118,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 0.3807,
      "step": 149
    },
    {
      "epoch": 0.012054970666238046,
      "grad_norm": 0.49938416481018066,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 0.2675,
      "step": 150
    },
    {
      "epoch": 0.012054970666238046,
      "eval_loss": 0.3388899266719818,
      "eval_runtime": 1504.884,
      "eval_samples_per_second": 13.926,
      "eval_steps_per_second": 3.482,
      "step": 150
    },
    {
      "epoch": 0.012135337137346299,
      "grad_norm": 0.18534860014915466,
      "learning_rate": 1.553232954407171e-05,
      "loss": 0.2962,
      "step": 151
    },
    {
      "epoch": 0.012215703608454552,
      "grad_norm": 0.25620123744010925,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 0.333,
      "step": 152
    },
    {
      "epoch": 0.012296070079562807,
      "grad_norm": 0.3018810749053955,
      "learning_rate": 1.435357758543015e-05,
      "loss": 0.3106,
      "step": 153
    },
    {
      "epoch": 0.01237643655067106,
      "grad_norm": 0.2873912453651428,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 0.2809,
      "step": 154
    },
    {
      "epoch": 0.012456803021779313,
      "grad_norm": 0.2059762328863144,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 0.3118,
      "step": 155
    },
    {
      "epoch": 0.012537169492887567,
      "grad_norm": 0.3046092092990875,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 0.2337,
      "step": 156
    },
    {
      "epoch": 0.012617535963995822,
      "grad_norm": 0.2660495936870575,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 0.2903,
      "step": 157
    },
    {
      "epoch": 0.012697902435104075,
      "grad_norm": 0.32564952969551086,
      "learning_rate": 1.157994445715706e-05,
      "loss": 0.3222,
      "step": 158
    },
    {
      "epoch": 0.012778268906212328,
      "grad_norm": 0.25236859917640686,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 0.2677,
      "step": 159
    },
    {
      "epoch": 0.012858635377320581,
      "grad_norm": 0.397904634475708,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 0.4688,
      "step": 160
    },
    {
      "epoch": 0.012939001848428836,
      "grad_norm": 0.37300175428390503,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 0.295,
      "step": 161
    },
    {
      "epoch": 0.01301936831953709,
      "grad_norm": 0.29193902015686035,
      "learning_rate": 9.549150281252633e-06,
      "loss": 0.2537,
      "step": 162
    },
    {
      "epoch": 0.013099734790645343,
      "grad_norm": 0.36094874143600464,
      "learning_rate": 9.068759265665384e-06,
      "loss": 0.2717,
      "step": 163
    },
    {
      "epoch": 0.013180101261753596,
      "grad_norm": 0.3836531341075897,
      "learning_rate": 8.599558442598998e-06,
      "loss": 0.4108,
      "step": 164
    },
    {
      "epoch": 0.01326046773286185,
      "grad_norm": 0.2871370017528534,
      "learning_rate": 8.141676086873572e-06,
      "loss": 0.2845,
      "step": 165
    },
    {
      "epoch": 0.013340834203970104,
      "grad_norm": 0.4493572413921356,
      "learning_rate": 7.695237378953223e-06,
      "loss": 0.4717,
      "step": 166
    },
    {
      "epoch": 0.013421200675078357,
      "grad_norm": 0.31553685665130615,
      "learning_rate": 7.260364370723044e-06,
      "loss": 0.3308,
      "step": 167
    },
    {
      "epoch": 0.01350156714618661,
      "grad_norm": 0.4123665392398834,
      "learning_rate": 6.837175952121306e-06,
      "loss": 0.3841,
      "step": 168
    },
    {
      "epoch": 0.013581933617294865,
      "grad_norm": 0.39727526903152466,
      "learning_rate": 6.425787818636131e-06,
      "loss": 0.3622,
      "step": 169
    },
    {
      "epoch": 0.013662300088403119,
      "grad_norm": 0.42318227887153625,
      "learning_rate": 6.026312439675552e-06,
      "loss": 0.3221,
      "step": 170
    },
    {
      "epoch": 0.013742666559511372,
      "grad_norm": 0.34337565302848816,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 0.2302,
      "step": 171
    },
    {
      "epoch": 0.013823033030619625,
      "grad_norm": 0.4516986012458801,
      "learning_rate": 5.263533508961827e-06,
      "loss": 0.43,
      "step": 172
    },
    {
      "epoch": 0.013903399501727878,
      "grad_norm": 0.5286464095115662,
      "learning_rate": 4.900438493352055e-06,
      "loss": 0.3947,
      "step": 173
    },
    {
      "epoch": 0.013983765972836133,
      "grad_norm": 0.418790340423584,
      "learning_rate": 4.549673247541875e-06,
      "loss": 0.2461,
      "step": 174
    },
    {
      "epoch": 0.014064132443944386,
      "grad_norm": 0.5441297292709351,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 0.3717,
      "step": 175
    },
    {
      "epoch": 0.01414449891505264,
      "grad_norm": 0.3862937092781067,
      "learning_rate": 3.885512251130763e-06,
      "loss": 0.2792,
      "step": 176
    },
    {
      "epoch": 0.014224865386160893,
      "grad_norm": 0.49587640166282654,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 0.3326,
      "step": 177
    },
    {
      "epoch": 0.014305231857269148,
      "grad_norm": 0.6037700772285461,
      "learning_rate": 3.271776770026963e-06,
      "loss": 0.378,
      "step": 178
    },
    {
      "epoch": 0.014385598328377401,
      "grad_norm": 0.48618656396865845,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 0.3008,
      "step": 179
    },
    {
      "epoch": 0.014465964799485654,
      "grad_norm": 0.4510641396045685,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 0.3727,
      "step": 180
    },
    {
      "epoch": 0.014546331270593908,
      "grad_norm": 0.39510321617126465,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 0.2121,
      "step": 181
    },
    {
      "epoch": 0.014626697741702162,
      "grad_norm": 0.4688001573085785,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 0.3581,
      "step": 182
    },
    {
      "epoch": 0.014707064212810416,
      "grad_norm": 0.35157933831214905,
      "learning_rate": 1.962316193157593e-06,
      "loss": 0.2327,
      "step": 183
    },
    {
      "epoch": 0.014787430683918669,
      "grad_norm": 0.3844772279262543,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 0.3085,
      "step": 184
    },
    {
      "epoch": 0.014867797155026922,
      "grad_norm": 0.4743369221687317,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 0.2824,
      "step": 185
    },
    {
      "epoch": 0.014948163626135177,
      "grad_norm": 0.5135641098022461,
      "learning_rate": 1.333670137599713e-06,
      "loss": 0.4122,
      "step": 186
    },
    {
      "epoch": 0.01502853009724343,
      "grad_norm": 0.5278946757316589,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 0.3911,
      "step": 187
    },
    {
      "epoch": 0.015108896568351684,
      "grad_norm": 0.46897366642951965,
      "learning_rate": 9.810017062595322e-07,
      "loss": 0.2746,
      "step": 188
    },
    {
      "epoch": 0.015189263039459937,
      "grad_norm": 0.4527287781238556,
      "learning_rate": 8.247462563808817e-07,
      "loss": 0.2725,
      "step": 189
    },
    {
      "epoch": 0.015269629510568192,
      "grad_norm": 0.5292967557907104,
      "learning_rate": 6.819348298638839e-07,
      "loss": 0.3639,
      "step": 190
    },
    {
      "epoch": 0.015349995981676445,
      "grad_norm": 0.5562537908554077,
      "learning_rate": 5.526064699265753e-07,
      "loss": 0.2962,
      "step": 191
    },
    {
      "epoch": 0.015430362452784698,
      "grad_norm": 0.5636081099510193,
      "learning_rate": 4.367965336512403e-07,
      "loss": 0.3322,
      "step": 192
    },
    {
      "epoch": 0.015510728923892951,
      "grad_norm": 0.5668913722038269,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 0.3388,
      "step": 193
    },
    {
      "epoch": 0.015591095395001206,
      "grad_norm": 0.6294294595718384,
      "learning_rate": 2.458548727494292e-07,
      "loss": 0.234,
      "step": 194
    },
    {
      "epoch": 0.01567146186610946,
      "grad_norm": 0.6742649078369141,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 0.4989,
      "step": 195
    },
    {
      "epoch": 0.01575182833721771,
      "grad_norm": 0.5202977061271667,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 0.2358,
      "step": 196
    },
    {
      "epoch": 0.015832194808325966,
      "grad_norm": 0.6953601241111755,
      "learning_rate": 6.150154258476315e-08,
      "loss": 0.4657,
      "step": 197
    },
    {
      "epoch": 0.01591256127943422,
      "grad_norm": 0.7000541090965271,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 0.3249,
      "step": 198
    },
    {
      "epoch": 0.015992927750542472,
      "grad_norm": 0.8858475089073181,
      "learning_rate": 6.834750376549792e-09,
      "loss": 0.5271,
      "step": 199
    },
    {
      "epoch": 0.016073294221650727,
      "grad_norm": 0.7542904615402222,
      "learning_rate": 0.0,
      "loss": 0.2946,
      "step": 200
    },
    {
      "epoch": 0.016073294221650727,
      "eval_loss": 0.3340291678905487,
      "eval_runtime": 1505.6754,
      "eval_samples_per_second": 13.919,
      "eval_steps_per_second": 3.48,
      "step": 200
    }
  ],
  "logging_steps": 1,
  "max_steps": 200,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 50,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 5,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 0
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 2.8864340312457216e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}