{
  "best_metric": 1.4780741930007935,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.08300477277443453,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0004150238638721727,
      "grad_norm": 0.9040806293487549,
      "learning_rate": 1e-05,
      "loss": 1.5984,
      "step": 1
    },
    {
      "epoch": 0.0004150238638721727,
      "eval_loss": 2.097278356552124,
      "eval_runtime": 308.8969,
      "eval_samples_per_second": 13.137,
      "eval_steps_per_second": 3.286,
      "step": 1
    },
    {
      "epoch": 0.0008300477277443454,
      "grad_norm": 0.9268214702606201,
      "learning_rate": 2e-05,
      "loss": 1.5467,
      "step": 2
    },
    {
      "epoch": 0.0012450715916165179,
      "grad_norm": 1.0235897302627563,
      "learning_rate": 3e-05,
      "loss": 1.7765,
      "step": 3
    },
    {
      "epoch": 0.0016600954554886907,
      "grad_norm": 0.9821894764900208,
      "learning_rate": 4e-05,
      "loss": 1.7285,
      "step": 4
    },
    {
      "epoch": 0.002075119319360863,
      "grad_norm": 0.8883850574493408,
      "learning_rate": 5e-05,
      "loss": 1.7356,
      "step": 5
    },
    {
      "epoch": 0.0024901431832330357,
      "grad_norm": 0.7983221411705017,
      "learning_rate": 6e-05,
      "loss": 1.5957,
      "step": 6
    },
    {
      "epoch": 0.0029051670471052084,
      "grad_norm": 0.7555496692657471,
      "learning_rate": 7e-05,
      "loss": 1.6702,
      "step": 7
    },
    {
      "epoch": 0.0033201909109773814,
      "grad_norm": 0.9533426761627197,
      "learning_rate": 8e-05,
      "loss": 1.8611,
      "step": 8
    },
    {
      "epoch": 0.003735214774849554,
      "grad_norm": 0.9437768459320068,
      "learning_rate": 9e-05,
      "loss": 1.6254,
      "step": 9
    },
    {
      "epoch": 0.004150238638721726,
      "grad_norm": 0.7665088176727295,
      "learning_rate": 0.0001,
      "loss": 1.5651,
      "step": 10
    },
    {
      "epoch": 0.004565262502593899,
      "grad_norm": 1.0628373622894287,
      "learning_rate": 9.999316524962345e-05,
      "loss": 1.7285,
      "step": 11
    },
    {
      "epoch": 0.0049802863664660715,
      "grad_norm": 0.7029485702514648,
      "learning_rate": 9.997266286704631e-05,
      "loss": 1.5528,
      "step": 12
    },
    {
      "epoch": 0.0053953102303382445,
      "grad_norm": 0.9037666320800781,
      "learning_rate": 9.993849845741524e-05,
      "loss": 1.6942,
      "step": 13
    },
    {
      "epoch": 0.005810334094210417,
      "grad_norm": 0.7749373316764832,
      "learning_rate": 9.989068136093873e-05,
      "loss": 1.628,
      "step": 14
    },
    {
      "epoch": 0.00622535795808259,
      "grad_norm": 0.6726087927818298,
      "learning_rate": 9.98292246503335e-05,
      "loss": 1.6615,
      "step": 15
    },
    {
      "epoch": 0.006640381821954763,
      "grad_norm": 0.709281861782074,
      "learning_rate": 9.975414512725057e-05,
      "loss": 1.7138,
      "step": 16
    },
    {
      "epoch": 0.007055405685826935,
      "grad_norm": 0.6763333678245544,
      "learning_rate": 9.966546331768191e-05,
      "loss": 1.6198,
      "step": 17
    },
    {
      "epoch": 0.007470429549699108,
      "grad_norm": 0.7684789896011353,
      "learning_rate": 9.956320346634876e-05,
      "loss": 1.5476,
      "step": 18
    },
    {
      "epoch": 0.00788545341357128,
      "grad_norm": 0.7368987798690796,
      "learning_rate": 9.944739353007344e-05,
      "loss": 1.6289,
      "step": 19
    },
    {
      "epoch": 0.008300477277443452,
      "grad_norm": 0.6821125149726868,
      "learning_rate": 9.931806517013612e-05,
      "loss": 1.518,
      "step": 20
    },
    {
      "epoch": 0.008715501141315626,
      "grad_norm": 0.7184584736824036,
      "learning_rate": 9.917525374361912e-05,
      "loss": 1.6058,
      "step": 21
    },
    {
      "epoch": 0.009130525005187799,
      "grad_norm": 0.7115007638931274,
      "learning_rate": 9.901899829374047e-05,
      "loss": 1.5502,
      "step": 22
    },
    {
      "epoch": 0.00954554886905997,
      "grad_norm": 0.84979248046875,
      "learning_rate": 9.884934153917997e-05,
      "loss": 1.5956,
      "step": 23
    },
    {
      "epoch": 0.009960572732932143,
      "grad_norm": 0.7301867604255676,
      "learning_rate": 9.86663298624003e-05,
      "loss": 1.5356,
      "step": 24
    },
    {
      "epoch": 0.010375596596804317,
      "grad_norm": 0.7334578633308411,
      "learning_rate": 9.847001329696653e-05,
      "loss": 1.5731,
      "step": 25
    },
    {
      "epoch": 0.010790620460676489,
      "grad_norm": 0.8474494814872742,
      "learning_rate": 9.826044551386744e-05,
      "loss": 1.5658,
      "step": 26
    },
    {
      "epoch": 0.011205644324548661,
      "grad_norm": 0.7066594362258911,
      "learning_rate": 9.803768380684242e-05,
      "loss": 1.5892,
      "step": 27
    },
    {
      "epoch": 0.011620668188420833,
      "grad_norm": 0.6848791837692261,
      "learning_rate": 9.780178907671789e-05,
      "loss": 1.5779,
      "step": 28
    },
    {
      "epoch": 0.012035692052293007,
      "grad_norm": 0.69795823097229,
      "learning_rate": 9.755282581475769e-05,
      "loss": 1.5696,
      "step": 29
    },
    {
      "epoch": 0.01245071591616518,
      "grad_norm": 0.6559748649597168,
      "learning_rate": 9.729086208503174e-05,
      "loss": 1.5898,
      "step": 30
    },
    {
      "epoch": 0.012865739780037352,
      "grad_norm": 0.9328115582466125,
      "learning_rate": 9.701596950580806e-05,
      "loss": 1.6485,
      "step": 31
    },
    {
      "epoch": 0.013280763643909526,
      "grad_norm": 0.7966737151145935,
      "learning_rate": 9.672822322997305e-05,
      "loss": 1.6068,
      "step": 32
    },
    {
      "epoch": 0.013695787507781698,
      "grad_norm": 0.6859557032585144,
      "learning_rate": 9.642770192448536e-05,
      "loss": 1.5533,
      "step": 33
    },
    {
      "epoch": 0.01411081137165387,
      "grad_norm": 0.7181488275527954,
      "learning_rate": 9.611448774886924e-05,
      "loss": 1.5552,
      "step": 34
    },
    {
      "epoch": 0.014525835235526042,
      "grad_norm": 0.8105568289756775,
      "learning_rate": 9.578866633275288e-05,
      "loss": 1.5617,
      "step": 35
    },
    {
      "epoch": 0.014940859099398216,
      "grad_norm": 1.056746482849121,
      "learning_rate": 9.545032675245813e-05,
      "loss": 1.5397,
      "step": 36
    },
    {
      "epoch": 0.015355882963270388,
      "grad_norm": 0.8355746865272522,
      "learning_rate": 9.509956150664796e-05,
      "loss": 1.4344,
      "step": 37
    },
    {
      "epoch": 0.01577090682714256,
      "grad_norm": 1.0142078399658203,
      "learning_rate": 9.473646649103818e-05,
      "loss": 1.5112,
      "step": 38
    },
    {
      "epoch": 0.016185930691014733,
      "grad_norm": 0.8601869344711304,
      "learning_rate": 9.43611409721806e-05,
      "loss": 1.5725,
      "step": 39
    },
    {
      "epoch": 0.016600954554886905,
      "grad_norm": 1.0650964975357056,
      "learning_rate": 9.397368756032445e-05,
      "loss": 1.5563,
      "step": 40
    },
    {
      "epoch": 0.017015978418759077,
      "grad_norm": 0.8317297697067261,
      "learning_rate": 9.357421218136386e-05,
      "loss": 1.497,
      "step": 41
    },
    {
      "epoch": 0.017431002282631253,
      "grad_norm": 0.9435349702835083,
      "learning_rate": 9.316282404787871e-05,
      "loss": 1.6221,
      "step": 42
    },
    {
      "epoch": 0.017846026146503425,
      "grad_norm": 0.8460783362388611,
      "learning_rate": 9.273963562927695e-05,
      "loss": 1.6298,
      "step": 43
    },
    {
      "epoch": 0.018261050010375597,
      "grad_norm": 0.9758943319320679,
      "learning_rate": 9.230476262104677e-05,
      "loss": 1.5457,
      "step": 44
    },
    {
      "epoch": 0.01867607387424777,
      "grad_norm": 0.8983597159385681,
      "learning_rate": 9.185832391312644e-05,
      "loss": 1.5731,
      "step": 45
    },
    {
      "epoch": 0.01909109773811994,
      "grad_norm": 1.0872970819473267,
      "learning_rate": 9.140044155740101e-05,
      "loss": 1.4918,
      "step": 46
    },
    {
      "epoch": 0.019506121601992114,
      "grad_norm": 0.9300501942634583,
      "learning_rate": 9.093124073433463e-05,
      "loss": 1.4961,
      "step": 47
    },
    {
      "epoch": 0.019921145465864286,
      "grad_norm": 1.2505871057510376,
      "learning_rate": 9.045084971874738e-05,
      "loss": 1.4185,
      "step": 48
    },
    {
      "epoch": 0.02033616932973646,
      "grad_norm": 1.1213619709014893,
      "learning_rate": 8.995939984474624e-05,
      "loss": 1.3337,
      "step": 49
    },
    {
      "epoch": 0.020751193193608634,
      "grad_norm": 1.48298180103302,
      "learning_rate": 8.945702546981969e-05,
      "loss": 1.4725,
      "step": 50
    },
    {
      "epoch": 0.020751193193608634,
      "eval_loss": 1.6225911378860474,
      "eval_runtime": 311.2428,
      "eval_samples_per_second": 13.038,
      "eval_steps_per_second": 3.261,
      "step": 50
    },
    {
      "epoch": 0.021166217057480806,
      "grad_norm": 1.295230746269226,
      "learning_rate": 8.894386393810563e-05,
      "loss": 1.4678,
      "step": 51
    },
    {
      "epoch": 0.021581240921352978,
      "grad_norm": 1.1907988786697388,
      "learning_rate": 8.842005554284296e-05,
      "loss": 1.5818,
      "step": 52
    },
    {
      "epoch": 0.02199626478522515,
      "grad_norm": 0.927335798740387,
      "learning_rate": 8.788574348801675e-05,
      "loss": 1.5013,
      "step": 53
    },
    {
      "epoch": 0.022411288649097322,
      "grad_norm": 0.6489278078079224,
      "learning_rate": 8.73410738492077e-05,
      "loss": 1.475,
      "step": 54
    },
    {
      "epoch": 0.022826312512969495,
      "grad_norm": 0.8121264576911926,
      "learning_rate": 8.678619553365659e-05,
      "loss": 1.5256,
      "step": 55
    },
    {
      "epoch": 0.023241336376841667,
      "grad_norm": 0.6559225916862488,
      "learning_rate": 8.622126023955446e-05,
      "loss": 1.5329,
      "step": 56
    },
    {
      "epoch": 0.023656360240713842,
      "grad_norm": 0.6731387376785278,
      "learning_rate": 8.564642241456986e-05,
      "loss": 1.6484,
      "step": 57
    },
    {
      "epoch": 0.024071384104586015,
      "grad_norm": 0.7704599499702454,
      "learning_rate": 8.506183921362443e-05,
      "loss": 1.5695,
      "step": 58
    },
    {
      "epoch": 0.024486407968458187,
      "grad_norm": 0.5911169052124023,
      "learning_rate": 8.44676704559283e-05,
      "loss": 1.5327,
      "step": 59
    },
    {
      "epoch": 0.02490143183233036,
      "grad_norm": 0.6432330012321472,
      "learning_rate": 8.386407858128706e-05,
      "loss": 1.5431,
      "step": 60
    },
    {
      "epoch": 0.02531645569620253,
      "grad_norm": 0.6107537150382996,
      "learning_rate": 8.32512286056924e-05,
      "loss": 1.449,
      "step": 61
    },
    {
      "epoch": 0.025731479560074703,
      "grad_norm": 0.5994659066200256,
      "learning_rate": 8.262928807620843e-05,
      "loss": 1.4917,
      "step": 62
    },
    {
      "epoch": 0.026146503423946876,
      "grad_norm": 0.5583532452583313,
      "learning_rate": 8.199842702516583e-05,
      "loss": 1.5868,
      "step": 63
    },
    {
      "epoch": 0.02656152728781905,
      "grad_norm": 0.6556219458580017,
      "learning_rate": 8.135881792367686e-05,
      "loss": 1.5438,
      "step": 64
    },
    {
      "epoch": 0.026976551151691223,
      "grad_norm": 0.602249026298523,
      "learning_rate": 8.07106356344834e-05,
      "loss": 1.4724,
      "step": 65
    },
    {
      "epoch": 0.027391575015563396,
      "grad_norm": 0.5924007892608643,
      "learning_rate": 8.005405736415126e-05,
      "loss": 1.466,
      "step": 66
    },
    {
      "epoch": 0.027806598879435568,
      "grad_norm": 0.5764862298965454,
      "learning_rate": 7.938926261462366e-05,
      "loss": 1.5642,
      "step": 67
    },
    {
      "epoch": 0.02822162274330774,
      "grad_norm": 0.717204749584198,
      "learning_rate": 7.871643313414718e-05,
      "loss": 1.5692,
      "step": 68
    },
    {
      "epoch": 0.028636646607179912,
      "grad_norm": 0.586876630783081,
      "learning_rate": 7.803575286758364e-05,
      "loss": 1.5773,
      "step": 69
    },
    {
      "epoch": 0.029051670471052084,
      "grad_norm": 0.6259408593177795,
      "learning_rate": 7.734740790612136e-05,
      "loss": 1.5671,
      "step": 70
    },
    {
      "epoch": 0.029466694334924257,
      "grad_norm": 0.9274705052375793,
      "learning_rate": 7.66515864363997e-05,
      "loss": 1.515,
      "step": 71
    },
    {
      "epoch": 0.029881718198796432,
      "grad_norm": 0.79607754945755,
      "learning_rate": 7.594847868906076e-05,
      "loss": 1.6209,
      "step": 72
    },
    {
      "epoch": 0.030296742062668604,
      "grad_norm": 0.6282968521118164,
      "learning_rate": 7.52382768867422e-05,
      "loss": 1.5848,
      "step": 73
    },
    {
      "epoch": 0.030711765926540777,
      "grad_norm": 0.6528509259223938,
      "learning_rate": 7.452117519152542e-05,
      "loss": 1.5166,
      "step": 74
    },
    {
      "epoch": 0.03112678979041295,
      "grad_norm": 0.5978628396987915,
      "learning_rate": 7.379736965185368e-05,
      "loss": 1.4337,
      "step": 75
    },
    {
      "epoch": 0.03154181365428512,
      "grad_norm": 0.6321048140525818,
      "learning_rate": 7.30670581489344e-05,
      "loss": 1.549,
      "step": 76
    },
    {
      "epoch": 0.0319568375181573,
      "grad_norm": 0.6578341126441956,
      "learning_rate": 7.233044034264034e-05,
      "loss": 1.4733,
      "step": 77
    },
    {
      "epoch": 0.032371861382029465,
      "grad_norm": 0.6550039052963257,
      "learning_rate": 7.158771761692464e-05,
      "loss": 1.4077,
      "step": 78
    },
    {
      "epoch": 0.03278688524590164,
      "grad_norm": 0.6624751091003418,
      "learning_rate": 7.083909302476453e-05,
      "loss": 1.5121,
      "step": 79
    },
    {
      "epoch": 0.03320190910977381,
      "grad_norm": 0.6436450481414795,
      "learning_rate": 7.008477123264848e-05,
      "loss": 1.5863,
      "step": 80
    },
    {
      "epoch": 0.033616932973645985,
      "grad_norm": 0.7376863956451416,
      "learning_rate": 6.932495846462261e-05,
      "loss": 1.6854,
      "step": 81
    },
    {
      "epoch": 0.034031956837518154,
      "grad_norm": 0.6503894329071045,
      "learning_rate": 6.855986244591104e-05,
      "loss": 1.4428,
      "step": 82
    },
    {
      "epoch": 0.03444698070139033,
      "grad_norm": 0.7245578169822693,
      "learning_rate": 6.778969234612584e-05,
      "loss": 1.609,
      "step": 83
    },
    {
      "epoch": 0.034862004565262505,
      "grad_norm": 0.6532803177833557,
      "learning_rate": 6.701465872208216e-05,
      "loss": 1.4628,
      "step": 84
    },
    {
      "epoch": 0.035277028429134674,
      "grad_norm": 0.6491965055465698,
      "learning_rate": 6.623497346023418e-05,
      "loss": 1.5758,
      "step": 85
    },
    {
      "epoch": 0.03569205229300685,
      "grad_norm": 0.7369991540908813,
      "learning_rate": 6.545084971874738e-05,
      "loss": 1.572,
      "step": 86
    },
    {
      "epoch": 0.03610707615687902,
      "grad_norm": 0.6906344890594482,
      "learning_rate": 6.466250186922325e-05,
      "loss": 1.5037,
      "step": 87
    },
    {
      "epoch": 0.036522100020751194,
      "grad_norm": 0.7680234313011169,
      "learning_rate": 6.387014543809223e-05,
      "loss": 1.5832,
      "step": 88
    },
    {
      "epoch": 0.03693712388462336,
      "grad_norm": 0.7673617005348206,
      "learning_rate": 6.307399704769099e-05,
      "loss": 1.5151,
      "step": 89
    },
    {
      "epoch": 0.03735214774849554,
      "grad_norm": 0.7738460302352905,
      "learning_rate": 6.227427435703997e-05,
      "loss": 1.3993,
      "step": 90
    },
    {
      "epoch": 0.037767171612367714,
      "grad_norm": 0.7285815477371216,
      "learning_rate": 6.147119600233758e-05,
      "loss": 1.4983,
      "step": 91
    },
    {
      "epoch": 0.03818219547623988,
      "grad_norm": 0.7915642857551575,
      "learning_rate": 6.066498153718735e-05,
      "loss": 1.5477,
      "step": 92
    },
    {
      "epoch": 0.03859721934011206,
      "grad_norm": 0.7667887806892395,
      "learning_rate": 5.985585137257401e-05,
      "loss": 1.5767,
      "step": 93
    },
    {
      "epoch": 0.03901224320398423,
      "grad_norm": 0.7948265671730042,
      "learning_rate": 5.90440267166055e-05,
      "loss": 1.4944,
      "step": 94
    },
    {
      "epoch": 0.0394272670678564,
      "grad_norm": 0.9050247669219971,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 1.3873,
      "step": 95
    },
    {
      "epoch": 0.03984229093172857,
      "grad_norm": 0.8589797616004944,
      "learning_rate": 5.74131823855921e-05,
      "loss": 1.4529,
      "step": 96
    },
    {
      "epoch": 0.04025731479560075,
      "grad_norm": 0.9197651147842407,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 1.3978,
      "step": 97
    },
    {
      "epoch": 0.04067233865947292,
      "grad_norm": 0.9008936882019043,
      "learning_rate": 5.577423184847932e-05,
      "loss": 1.4307,
      "step": 98
    },
    {
      "epoch": 0.04108736252334509,
      "grad_norm": 1.0464752912521362,
      "learning_rate": 5.495227651252315e-05,
      "loss": 1.3833,
      "step": 99
    },
    {
      "epoch": 0.04150238638721727,
      "grad_norm": 1.2581040859222412,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 1.3548,
      "step": 100
    },
    {
      "epoch": 0.04150238638721727,
      "eval_loss": 1.5165632963180542,
      "eval_runtime": 311.0295,
      "eval_samples_per_second": 13.047,
      "eval_steps_per_second": 3.263,
      "step": 100
    },
    {
      "epoch": 0.041917410251089436,
      "grad_norm": 0.7076584100723267,
      "learning_rate": 5.330452921628497e-05,
      "loss": 1.4733,
      "step": 101
    },
    {
      "epoch": 0.04233243411496161,
      "grad_norm": 0.7008097171783447,
      "learning_rate": 5.247918773366112e-05,
      "loss": 1.4991,
      "step": 102
    },
    {
      "epoch": 0.04274745797883378,
      "grad_norm": 0.6747682094573975,
      "learning_rate": 5.165316846586541e-05,
      "loss": 1.5272,
      "step": 103
    },
    {
      "epoch": 0.043162481842705956,
      "grad_norm": 0.6513725519180298,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 1.5562,
      "step": 104
    },
    {
      "epoch": 0.043577505706578125,
      "grad_norm": 0.49187928438186646,
      "learning_rate": 5e-05,
      "loss": 1.5052,
      "step": 105
    },
    {
      "epoch": 0.0439925295704503,
      "grad_norm": 0.5867658257484436,
      "learning_rate": 4.917330276168208e-05,
      "loss": 1.5016,
      "step": 106
    },
    {
      "epoch": 0.044407553434322476,
      "grad_norm": 0.6284681558609009,
      "learning_rate": 4.834683153413459e-05,
      "loss": 1.5437,
      "step": 107
    },
    {
      "epoch": 0.044822577298194645,
      "grad_norm": 0.6012783050537109,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 1.5688,
      "step": 108
    },
    {
      "epoch": 0.04523760116206682,
      "grad_norm": 0.5611281991004944,
      "learning_rate": 4.669547078371504e-05,
      "loss": 1.5515,
      "step": 109
    },
    {
      "epoch": 0.04565262502593899,
      "grad_norm": 0.608805239200592,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 1.5294,
      "step": 110
    },
    {
      "epoch": 0.046067648889811165,
      "grad_norm": 0.6421564817428589,
      "learning_rate": 4.504772348747687e-05,
      "loss": 1.5386,
      "step": 111
    },
    {
      "epoch": 0.046482672753683334,
      "grad_norm": 0.647364616394043,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 1.5305,
      "step": 112
    },
    {
      "epoch": 0.04689769661755551,
      "grad_norm": 0.641161322593689,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 1.514,
      "step": 113
    },
    {
      "epoch": 0.047312720481427685,
      "grad_norm": 0.6211668252944946,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 1.5206,
      "step": 114
    },
    {
      "epoch": 0.047727744345299854,
      "grad_norm": 0.558771550655365,
      "learning_rate": 4.17702704859633e-05,
      "loss": 1.4915,
      "step": 115
    },
    {
      "epoch": 0.04814276820917203,
      "grad_norm": 0.5996015667915344,
      "learning_rate": 4.095597328339452e-05,
      "loss": 1.5172,
      "step": 116
    },
    {
      "epoch": 0.0485577920730442,
      "grad_norm": 0.529927670955658,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 1.6221,
      "step": 117
    },
    {
      "epoch": 0.048972815936916374,
      "grad_norm": 0.5847812294960022,
      "learning_rate": 3.933501846281267e-05,
      "loss": 1.4873,
      "step": 118
    },
    {
      "epoch": 0.04938783980078854,
      "grad_norm": 0.5653445720672607,
      "learning_rate": 3.852880399766243e-05,
      "loss": 1.5883,
      "step": 119
    },
    {
      "epoch": 0.04980286366466072,
      "grad_norm": 0.596710741519928,
      "learning_rate": 3.772572564296005e-05,
      "loss": 1.5829,
      "step": 120
    },
    {
      "epoch": 0.050217887528532894,
      "grad_norm": 0.7643442749977112,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 1.5568,
      "step": 121
    },
    {
      "epoch": 0.05063291139240506,
      "grad_norm": 0.5649930238723755,
      "learning_rate": 3.612985456190778e-05,
      "loss": 1.4868,
      "step": 122
    },
    {
      "epoch": 0.05104793525627724,
      "grad_norm": 0.6768708229064941,
      "learning_rate": 3.533749813077677e-05,
      "loss": 1.593,
      "step": 123
    },
    {
      "epoch": 0.05146295912014941,
      "grad_norm": 0.6045772433280945,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 1.5593,
      "step": 124
    },
    {
      "epoch": 0.05187798298402158,
      "grad_norm": 0.6199007034301758,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 1.4505,
      "step": 125
    },
    {
      "epoch": 0.05229300684789375,
      "grad_norm": 0.615681529045105,
      "learning_rate": 3.298534127791785e-05,
      "loss": 1.5054,
      "step": 126
    },
    {
      "epoch": 0.05270803071176593,
      "grad_norm": 0.6058111786842346,
      "learning_rate": 3.221030765387417e-05,
      "loss": 1.6281,
      "step": 127
    },
    {
      "epoch": 0.0531230545756381,
      "grad_norm": 0.5869006514549255,
      "learning_rate": 3.144013755408895e-05,
      "loss": 1.5096,
      "step": 128
    },
    {
      "epoch": 0.05353807843951027,
      "grad_norm": 0.6569715738296509,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 1.5622,
      "step": 129
    },
    {
      "epoch": 0.05395310230338245,
      "grad_norm": 0.692600429058075,
      "learning_rate": 2.991522876735154e-05,
      "loss": 1.5635,
      "step": 130
    },
    {
      "epoch": 0.054368126167254616,
      "grad_norm": 0.6849222779273987,
      "learning_rate": 2.916090697523549e-05,
      "loss": 1.4116,
      "step": 131
    },
    {
      "epoch": 0.05478315003112679,
      "grad_norm": 0.622251033782959,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 1.5347,
      "step": 132
    },
    {
      "epoch": 0.05519817389499896,
      "grad_norm": 0.6617666482925415,
      "learning_rate": 2.766955965735968e-05,
      "loss": 1.4188,
      "step": 133
    },
    {
      "epoch": 0.055613197758871136,
      "grad_norm": 0.7757525444030762,
      "learning_rate": 2.693294185106562e-05,
      "loss": 1.4783,
      "step": 134
    },
    {
      "epoch": 0.056028221622743304,
      "grad_norm": 0.6824130415916443,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 1.5326,
      "step": 135
    },
    {
      "epoch": 0.05644324548661548,
      "grad_norm": 0.7134312391281128,
      "learning_rate": 2.547882480847461e-05,
      "loss": 1.5048,
      "step": 136
    },
    {
      "epoch": 0.056858269350487656,
      "grad_norm": 0.6936268210411072,
      "learning_rate": 2.476172311325783e-05,
      "loss": 1.3719,
      "step": 137
    },
    {
      "epoch": 0.057273293214359824,
      "grad_norm": 0.6818733811378479,
      "learning_rate": 2.405152131093926e-05,
      "loss": 1.5015,
      "step": 138
    },
    {
      "epoch": 0.057688317078232,
      "grad_norm": 0.7464825510978699,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 1.5357,
      "step": 139
    },
    {
      "epoch": 0.05810334094210417,
      "grad_norm": 0.6866446733474731,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 1.4472,
      "step": 140
    },
    {
      "epoch": 0.058518364805976344,
      "grad_norm": 0.7462300062179565,
      "learning_rate": 2.196424713241637e-05,
      "loss": 1.3845,
      "step": 141
    },
    {
      "epoch": 0.05893338866984851,
      "grad_norm": 0.8067519664764404,
      "learning_rate": 2.128356686585282e-05,
      "loss": 1.4749,
      "step": 142
    },
    {
      "epoch": 0.05934841253372069,
      "grad_norm": 0.8725991249084473,
      "learning_rate": 2.061073738537635e-05,
      "loss": 1.4463,
      "step": 143
    },
    {
      "epoch": 0.059763436397592865,
      "grad_norm": 0.775001585483551,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 1.4209,
      "step": 144
    },
    {
      "epoch": 0.06017846026146503,
      "grad_norm": 0.8111575841903687,
      "learning_rate": 1.928936436551661e-05,
      "loss": 1.5426,
      "step": 145
    },
    {
      "epoch": 0.06059348412533721,
      "grad_norm": 0.9825055003166199,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 1.53,
      "step": 146
    },
    {
      "epoch": 0.06100850798920938,
      "grad_norm": 0.8568292260169983,
      "learning_rate": 1.800157297483417e-05,
      "loss": 1.4772,
      "step": 147
    },
    {
      "epoch": 0.06142353185308155,
      "grad_norm": 0.924245297908783,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 1.3636,
      "step": 148
    },
    {
      "epoch": 0.06183855571695372,
      "grad_norm": 1.1757372617721558,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 1.3777,
      "step": 149
    },
    {
      "epoch": 0.0622535795808259,
      "grad_norm": 1.3622688055038452,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 1.1867,
      "step": 150
    },
    {
      "epoch": 0.0622535795808259,
      "eval_loss": 1.482372760772705,
      "eval_runtime": 311.5867,
      "eval_samples_per_second": 13.024,
      "eval_steps_per_second": 3.258,
      "step": 150
    },
    {
      "epoch": 0.06266860344469807,
      "grad_norm": 0.47829145193099976,
      "learning_rate": 1.553232954407171e-05,
      "loss": 1.4022,
      "step": 151
    },
    {
      "epoch": 0.06308362730857024,
      "grad_norm": 0.5311530232429504,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 1.4899,
      "step": 152
    },
    {
      "epoch": 0.06349865117244241,
      "grad_norm": 0.528104841709137,
      "learning_rate": 1.435357758543015e-05,
      "loss": 1.3796,
      "step": 153
    },
    {
      "epoch": 0.0639136750363146,
      "grad_norm": 0.5498363971710205,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 1.4685,
      "step": 154
    },
    {
      "epoch": 0.06432869890018676,
      "grad_norm": 0.5535220503807068,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 1.4337,
      "step": 155
    },
    {
      "epoch": 0.06474372276405893,
      "grad_norm": 0.5592207312583923,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 1.3985,
      "step": 156
    },
    {
      "epoch": 0.0651587466279311,
      "grad_norm": 0.5554097294807434,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 1.5028,
      "step": 157
    },
    {
      "epoch": 0.06557377049180328,
      "grad_norm": 0.586708664894104,
      "learning_rate": 1.157994445715706e-05,
      "loss": 1.4503,
      "step": 158
    },
    {
      "epoch": 0.06598879435567545,
      "grad_norm": 0.5825499296188354,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 1.3706,
      "step": 159
    },
    {
      "epoch": 0.06640381821954762,
      "grad_norm": 0.5954872965812683,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 1.4461,
      "step": 160
    },
    {
      "epoch": 0.0668188420834198,
      "grad_norm": 0.5384337306022644,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 1.4507,
      "step": 161
    },
    {
      "epoch": 0.06723386594729197,
      "grad_norm": 0.5707142353057861,
      "learning_rate": 9.549150281252633e-06,
      "loss": 1.457,
      "step": 162
    },
    {
      "epoch": 0.06764888981116414,
      "grad_norm": 0.5536739826202393,
      "learning_rate": 9.068759265665384e-06,
      "loss": 1.5144,
      "step": 163
    },
    {
      "epoch": 0.06806391367503631,
      "grad_norm": 0.5960381627082825,
      "learning_rate": 8.599558442598998e-06,
      "loss": 1.4161,
      "step": 164
    },
    {
      "epoch": 0.06847893753890849,
      "grad_norm": 0.5277539491653442,
      "learning_rate": 8.141676086873572e-06,
      "loss": 1.5111,
      "step": 165
    },
    {
      "epoch": 0.06889396140278066,
      "grad_norm": 0.545833170413971,
      "learning_rate": 7.695237378953223e-06,
      "loss": 1.5037,
      "step": 166
    },
    {
      "epoch": 0.06930898526665283,
      "grad_norm": 0.6068612337112427,
      "learning_rate": 7.260364370723044e-06,
      "loss": 1.4978,
      "step": 167
    },
    {
      "epoch": 0.06972400913052501,
      "grad_norm": 0.6075426936149597,
      "learning_rate": 6.837175952121306e-06,
      "loss": 1.4755,
      "step": 168
    },
    {
      "epoch": 0.07013903299439718,
      "grad_norm": 0.5732592940330505,
      "learning_rate": 6.425787818636131e-06,
      "loss": 1.5248,
      "step": 169
    },
    {
      "epoch": 0.07055405685826935,
      "grad_norm": 0.5757129192352295,
      "learning_rate": 6.026312439675552e-06,
      "loss": 1.5244,
      "step": 170
    },
    {
      "epoch": 0.07096908072214152,
      "grad_norm": 0.5329376459121704,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 1.5506,
      "step": 171
    },
    {
      "epoch": 0.0713841045860137,
      "grad_norm": 0.6607252359390259,
      "learning_rate": 5.263533508961827e-06,
      "loss": 1.5311,
      "step": 172
    },
    {
      "epoch": 0.07179912844988587,
      "grad_norm": 0.610490620136261,
      "learning_rate": 4.900438493352055e-06,
      "loss": 1.5266,
      "step": 173
    },
    {
      "epoch": 0.07221415231375804,
      "grad_norm": 0.5758323073387146,
      "learning_rate": 4.549673247541875e-06,
      "loss": 1.5106,
      "step": 174
    },
    {
      "epoch": 0.07262917617763022,
      "grad_norm": 0.6813466548919678,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 1.6156,
      "step": 175
    },
    {
      "epoch": 0.07304420004150239,
      "grad_norm": 0.603428840637207,
      "learning_rate": 3.885512251130763e-06,
      "loss": 1.5227,
      "step": 176
    },
    {
      "epoch": 0.07345922390537456,
      "grad_norm": 0.6178284287452698,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 1.633,
      "step": 177
    },
    {
      "epoch": 0.07387424776924673,
      "grad_norm": 0.6623697876930237,
      "learning_rate": 3.271776770026963e-06,
      "loss": 1.5413,
      "step": 178
    },
    {
      "epoch": 0.07428927163311891,
      "grad_norm": 0.6070805191993713,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 1.6421,
      "step": 179
    },
    {
      "epoch": 0.07470429549699108,
      "grad_norm": 0.6213012337684631,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 1.3722,
      "step": 180
    },
    {
      "epoch": 0.07511931936086325,
      "grad_norm": 0.7087394595146179,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 1.5178,
      "step": 181
    },
    {
      "epoch": 0.07553434322473543,
      "grad_norm": 0.6422320008277893,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 1.4012,
      "step": 182
    },
    {
      "epoch": 0.0759493670886076,
      "grad_norm": 0.6422367095947266,
      "learning_rate": 1.962316193157593e-06,
      "loss": 1.5803,
      "step": 183
    },
    {
      "epoch": 0.07636439095247977,
      "grad_norm": 0.6316154599189758,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 1.386,
      "step": 184
    },
    {
      "epoch": 0.07677941481635193,
      "grad_norm": 0.6911137104034424,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 1.599,
      "step": 185
    },
    {
      "epoch": 0.07719443868022412,
      "grad_norm": 0.7753531336784363,
      "learning_rate": 1.333670137599713e-06,
      "loss": 1.535,
      "step": 186
    },
    {
      "epoch": 0.07760946254409629,
      "grad_norm": 0.6699145436286926,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 1.5651,
      "step": 187
    },
    {
      "epoch": 0.07802448640796845,
      "grad_norm": 0.7628930807113647,
      "learning_rate": 9.810017062595322e-07,
      "loss": 1.512,
      "step": 188
    },
    {
      "epoch": 0.07843951027184064,
      "grad_norm": 0.7047752141952515,
      "learning_rate": 8.247462563808817e-07,
      "loss": 1.4936,
      "step": 189
    },
    {
      "epoch": 0.0788545341357128,
      "grad_norm": 0.7719866037368774,
      "learning_rate": 6.819348298638839e-07,
      "loss": 1.5573,
      "step": 190
    },
    {
      "epoch": 0.07926955799958497,
      "grad_norm": 0.800010621547699,
      "learning_rate": 5.526064699265753e-07,
      "loss": 1.4549,
      "step": 191
    },
    {
      "epoch": 0.07968458186345714,
      "grad_norm": 0.7894272208213806,
      "learning_rate": 4.367965336512403e-07,
      "loss": 1.3724,
      "step": 192
    },
    {
      "epoch": 0.08009960572732933,
      "grad_norm": 0.8304757475852966,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 1.5447,
      "step": 193
    },
    {
      "epoch": 0.0805146295912015,
      "grad_norm": 1.0382047891616821,
      "learning_rate": 2.458548727494292e-07,
      "loss": 1.563,
      "step": 194
    },
    {
      "epoch": 0.08092965345507366,
      "grad_norm": 0.8752549290657043,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 1.6341,
      "step": 195
    },
    {
      "epoch": 0.08134467731894585,
      "grad_norm": 0.8815712928771973,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 1.408,
      "step": 196
    },
    {
      "epoch": 0.08175970118281801,
      "grad_norm": 1.002856969833374,
      "learning_rate": 6.150154258476315e-08,
      "loss": 1.2939,
      "step": 197
    },
    {
      "epoch": 0.08217472504669018,
      "grad_norm": 0.9047574400901794,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 1.34,
      "step": 198
    },
    {
      "epoch": 0.08258974891056235,
      "grad_norm": 1.1825281381607056,
      "learning_rate": 6.834750376549792e-09,
      "loss": 1.4277,
      "step": 199
    },
    {
      "epoch": 0.08300477277443453,
      "grad_norm": 1.1798052787780762,
      "learning_rate": 0.0,
      "loss": 1.3763,
      "step": 200
    },
    {
      "epoch": 0.08300477277443453,
      "eval_loss": 1.4780741930007935,
      "eval_runtime": 311.1553,
      "eval_samples_per_second": 13.042,
      "eval_steps_per_second": 3.262,
      "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": 3.0547399969603584e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}