{
  "best_metric": 0.0014145222958177328,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.3225806451612903,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0016129032258064516,
      "grad_norm": 0.603843629360199,
      "learning_rate": 1e-05,
      "loss": 0.9039,
      "step": 1
    },
    {
      "epoch": 0.0016129032258064516,
      "eval_loss": 0.7379059195518494,
      "eval_runtime": 107.1116,
      "eval_samples_per_second": 9.747,
      "eval_steps_per_second": 2.437,
      "step": 1
    },
    {
      "epoch": 0.0032258064516129032,
      "grad_norm": 0.738902747631073,
      "learning_rate": 2e-05,
      "loss": 1.1299,
      "step": 2
    },
    {
      "epoch": 0.004838709677419355,
      "grad_norm": 0.8352855443954468,
      "learning_rate": 3e-05,
      "loss": 1.2141,
      "step": 3
    },
    {
      "epoch": 0.0064516129032258064,
      "grad_norm": 0.8586352467536926,
      "learning_rate": 4e-05,
      "loss": 1.2945,
      "step": 4
    },
    {
      "epoch": 0.008064516129032258,
      "grad_norm": 0.9727372527122498,
      "learning_rate": 5e-05,
      "loss": 1.265,
      "step": 5
    },
    {
      "epoch": 0.00967741935483871,
      "grad_norm": 0.9425803422927856,
      "learning_rate": 6e-05,
      "loss": 1.2004,
      "step": 6
    },
    {
      "epoch": 0.01129032258064516,
      "grad_norm": 0.9808979034423828,
      "learning_rate": 7e-05,
      "loss": 1.0226,
      "step": 7
    },
    {
      "epoch": 0.012903225806451613,
      "grad_norm": 0.8568642735481262,
      "learning_rate": 8e-05,
      "loss": 0.791,
      "step": 8
    },
    {
      "epoch": 0.014516129032258065,
      "grad_norm": 0.9525200724601746,
      "learning_rate": 9e-05,
      "loss": 0.7293,
      "step": 9
    },
    {
      "epoch": 0.016129032258064516,
      "grad_norm": 1.1046411991119385,
      "learning_rate": 0.0001,
      "loss": 0.4723,
      "step": 10
    },
    {
      "epoch": 0.017741935483870968,
      "grad_norm": 0.8412634134292603,
      "learning_rate": 9.999316524962345e-05,
      "loss": 0.2159,
      "step": 11
    },
    {
      "epoch": 0.01935483870967742,
      "grad_norm": 0.9623184204101562,
      "learning_rate": 9.997266286704631e-05,
      "loss": 0.2193,
      "step": 12
    },
    {
      "epoch": 0.020967741935483872,
      "grad_norm": 1.2478351593017578,
      "learning_rate": 9.993849845741524e-05,
      "loss": 0.2806,
      "step": 13
    },
    {
      "epoch": 0.02258064516129032,
      "grad_norm": 1.0087507963180542,
      "learning_rate": 9.989068136093873e-05,
      "loss": 0.1489,
      "step": 14
    },
    {
      "epoch": 0.024193548387096774,
      "grad_norm": 1.6914600133895874,
      "learning_rate": 9.98292246503335e-05,
      "loss": 0.1748,
      "step": 15
    },
    {
      "epoch": 0.025806451612903226,
      "grad_norm": 1.0039892196655273,
      "learning_rate": 9.975414512725057e-05,
      "loss": 0.1199,
      "step": 16
    },
    {
      "epoch": 0.027419354838709678,
      "grad_norm": 0.7692573070526123,
      "learning_rate": 9.966546331768191e-05,
      "loss": 0.0675,
      "step": 17
    },
    {
      "epoch": 0.02903225806451613,
      "grad_norm": 0.8827784657478333,
      "learning_rate": 9.956320346634876e-05,
      "loss": 0.0686,
      "step": 18
    },
    {
      "epoch": 0.03064516129032258,
      "grad_norm": 0.7692415714263916,
      "learning_rate": 9.944739353007344e-05,
      "loss": 0.0545,
      "step": 19
    },
    {
      "epoch": 0.03225806451612903,
      "grad_norm": 0.9053983688354492,
      "learning_rate": 9.931806517013612e-05,
      "loss": 0.0553,
      "step": 20
    },
    {
      "epoch": 0.03387096774193549,
      "grad_norm": 0.6906158328056335,
      "learning_rate": 9.917525374361912e-05,
      "loss": 0.0291,
      "step": 21
    },
    {
      "epoch": 0.035483870967741936,
      "grad_norm": 1.2245749235153198,
      "learning_rate": 9.901899829374047e-05,
      "loss": 0.0855,
      "step": 22
    },
    {
      "epoch": 0.037096774193548385,
      "grad_norm": 0.5348228812217712,
      "learning_rate": 9.884934153917997e-05,
      "loss": 0.0254,
      "step": 23
    },
    {
      "epoch": 0.03870967741935484,
      "grad_norm": 1.3067644834518433,
      "learning_rate": 9.86663298624003e-05,
      "loss": 0.1753,
      "step": 24
    },
    {
      "epoch": 0.04032258064516129,
      "grad_norm": 0.7000195980072021,
      "learning_rate": 9.847001329696653e-05,
      "loss": 0.0257,
      "step": 25
    },
    {
      "epoch": 0.041935483870967745,
      "grad_norm": 0.23263350129127502,
      "learning_rate": 9.826044551386744e-05,
      "loss": 0.0083,
      "step": 26
    },
    {
      "epoch": 0.043548387096774194,
      "grad_norm": 1.1544290781021118,
      "learning_rate": 9.803768380684242e-05,
      "loss": 0.0753,
      "step": 27
    },
    {
      "epoch": 0.04516129032258064,
      "grad_norm": 1.6351916790008545,
      "learning_rate": 9.780178907671789e-05,
      "loss": 0.0563,
      "step": 28
    },
    {
      "epoch": 0.0467741935483871,
      "grad_norm": 0.8347451686859131,
      "learning_rate": 9.755282581475769e-05,
      "loss": 0.1168,
      "step": 29
    },
    {
      "epoch": 0.04838709677419355,
      "grad_norm": 0.9437413811683655,
      "learning_rate": 9.729086208503174e-05,
      "loss": 0.0662,
      "step": 30
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9113280177116394,
      "learning_rate": 9.701596950580806e-05,
      "loss": 0.0212,
      "step": 31
    },
    {
      "epoch": 0.05161290322580645,
      "grad_norm": 0.594906747341156,
      "learning_rate": 9.672822322997305e-05,
      "loss": 0.0483,
      "step": 32
    },
    {
      "epoch": 0.0532258064516129,
      "grad_norm": 0.32695919275283813,
      "learning_rate": 9.642770192448536e-05,
      "loss": 0.0147,
      "step": 33
    },
    {
      "epoch": 0.054838709677419356,
      "grad_norm": 0.33313068747520447,
      "learning_rate": 9.611448774886924e-05,
      "loss": 0.0206,
      "step": 34
    },
    {
      "epoch": 0.056451612903225805,
      "grad_norm": 1.2069950103759766,
      "learning_rate": 9.578866633275288e-05,
      "loss": 0.0513,
      "step": 35
    },
    {
      "epoch": 0.05806451612903226,
      "grad_norm": 0.43889063596725464,
      "learning_rate": 9.545032675245813e-05,
      "loss": 0.0133,
      "step": 36
    },
    {
      "epoch": 0.05967741935483871,
      "grad_norm": 0.7685791254043579,
      "learning_rate": 9.509956150664796e-05,
      "loss": 0.0634,
      "step": 37
    },
    {
      "epoch": 0.06129032258064516,
      "grad_norm": 1.0193281173706055,
      "learning_rate": 9.473646649103818e-05,
      "loss": 0.0348,
      "step": 38
    },
    {
      "epoch": 0.06290322580645161,
      "grad_norm": 0.7231410145759583,
      "learning_rate": 9.43611409721806e-05,
      "loss": 0.0469,
      "step": 39
    },
    {
      "epoch": 0.06451612903225806,
      "grad_norm": 0.6480370163917542,
      "learning_rate": 9.397368756032445e-05,
      "loss": 0.0197,
      "step": 40
    },
    {
      "epoch": 0.06612903225806452,
      "grad_norm": 1.3558857440948486,
      "learning_rate": 9.357421218136386e-05,
      "loss": 0.1396,
      "step": 41
    },
    {
      "epoch": 0.06774193548387097,
      "grad_norm": 1.1803086996078491,
      "learning_rate": 9.316282404787871e-05,
      "loss": 0.0097,
      "step": 42
    },
    {
      "epoch": 0.06935483870967742,
      "grad_norm": 0.35275307297706604,
      "learning_rate": 9.273963562927695e-05,
      "loss": 0.0084,
      "step": 43
    },
    {
      "epoch": 0.07096774193548387,
      "grad_norm": 0.7042427062988281,
      "learning_rate": 9.230476262104677e-05,
      "loss": 0.0235,
      "step": 44
    },
    {
      "epoch": 0.07258064516129033,
      "grad_norm": 1.0765535831451416,
      "learning_rate": 9.185832391312644e-05,
      "loss": 0.0177,
      "step": 45
    },
    {
      "epoch": 0.07419354838709677,
      "grad_norm": 0.8105307817459106,
      "learning_rate": 9.140044155740101e-05,
      "loss": 0.0174,
      "step": 46
    },
    {
      "epoch": 0.07580645161290323,
      "grad_norm": 0.12413424253463745,
      "learning_rate": 9.093124073433463e-05,
      "loss": 0.0031,
      "step": 47
    },
    {
      "epoch": 0.07741935483870968,
      "grad_norm": 1.49933660030365,
      "learning_rate": 9.045084971874738e-05,
      "loss": 0.1259,
      "step": 48
    },
    {
      "epoch": 0.07903225806451612,
      "grad_norm": 3.263685703277588,
      "learning_rate": 8.995939984474624e-05,
      "loss": 0.073,
      "step": 49
    },
    {
      "epoch": 0.08064516129032258,
      "grad_norm": 1.535475254058838,
      "learning_rate": 8.945702546981969e-05,
      "loss": 0.0535,
      "step": 50
    },
    {
      "epoch": 0.08064516129032258,
      "eval_loss": 0.005853595212101936,
      "eval_runtime": 108.8836,
      "eval_samples_per_second": 9.588,
      "eval_steps_per_second": 2.397,
      "step": 50
    },
    {
      "epoch": 0.08225806451612903,
      "grad_norm": 0.23463483154773712,
      "learning_rate": 8.894386393810563e-05,
      "loss": 0.0046,
      "step": 51
    },
    {
      "epoch": 0.08387096774193549,
      "grad_norm": 0.17040041089057922,
      "learning_rate": 8.842005554284296e-05,
      "loss": 0.0168,
      "step": 52
    },
    {
      "epoch": 0.08548387096774193,
      "grad_norm": 0.5608404278755188,
      "learning_rate": 8.788574348801675e-05,
      "loss": 0.0148,
      "step": 53
    },
    {
      "epoch": 0.08709677419354839,
      "grad_norm": 0.6543734669685364,
      "learning_rate": 8.73410738492077e-05,
      "loss": 0.0197,
      "step": 54
    },
    {
      "epoch": 0.08870967741935484,
      "grad_norm": 0.3093203604221344,
      "learning_rate": 8.678619553365659e-05,
      "loss": 0.0463,
      "step": 55
    },
    {
      "epoch": 0.09032258064516129,
      "grad_norm": 0.10087056457996368,
      "learning_rate": 8.622126023955446e-05,
      "loss": 0.0044,
      "step": 56
    },
    {
      "epoch": 0.09193548387096774,
      "grad_norm": 0.267266184091568,
      "learning_rate": 8.564642241456986e-05,
      "loss": 0.0129,
      "step": 57
    },
    {
      "epoch": 0.0935483870967742,
      "grad_norm": 0.31064850091934204,
      "learning_rate": 8.506183921362443e-05,
      "loss": 0.0484,
      "step": 58
    },
    {
      "epoch": 0.09516129032258064,
      "grad_norm": 0.2512148916721344,
      "learning_rate": 8.44676704559283e-05,
      "loss": 0.0344,
      "step": 59
    },
    {
      "epoch": 0.0967741935483871,
      "grad_norm": 1.4665775299072266,
      "learning_rate": 8.386407858128706e-05,
      "loss": 0.0336,
      "step": 60
    },
    {
      "epoch": 0.09838709677419355,
      "grad_norm": 0.33774101734161377,
      "learning_rate": 8.32512286056924e-05,
      "loss": 0.0152,
      "step": 61
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.22360168397426605,
      "learning_rate": 8.262928807620843e-05,
      "loss": 0.0147,
      "step": 62
    },
    {
      "epoch": 0.10161290322580645,
      "grad_norm": 0.2551698684692383,
      "learning_rate": 8.199842702516583e-05,
      "loss": 0.0167,
      "step": 63
    },
    {
      "epoch": 0.1032258064516129,
      "grad_norm": 0.4298221170902252,
      "learning_rate": 8.135881792367686e-05,
      "loss": 0.0314,
      "step": 64
    },
    {
      "epoch": 0.10483870967741936,
      "grad_norm": 0.292525053024292,
      "learning_rate": 8.07106356344834e-05,
      "loss": 0.0198,
      "step": 65
    },
    {
      "epoch": 0.1064516129032258,
      "grad_norm": 0.3462181091308594,
      "learning_rate": 8.005405736415126e-05,
      "loss": 0.0245,
      "step": 66
    },
    {
      "epoch": 0.10806451612903226,
      "grad_norm": 0.45068007707595825,
      "learning_rate": 7.938926261462366e-05,
      "loss": 0.0589,
      "step": 67
    },
    {
      "epoch": 0.10967741935483871,
      "grad_norm": 0.7323125004768372,
      "learning_rate": 7.871643313414718e-05,
      "loss": 0.0683,
      "step": 68
    },
    {
      "epoch": 0.11129032258064517,
      "grad_norm": 0.07402742654085159,
      "learning_rate": 7.803575286758364e-05,
      "loss": 0.0051,
      "step": 69
    },
    {
      "epoch": 0.11290322580645161,
      "grad_norm": 0.3788389265537262,
      "learning_rate": 7.734740790612136e-05,
      "loss": 0.0351,
      "step": 70
    },
    {
      "epoch": 0.11451612903225807,
      "grad_norm": 0.25281745195388794,
      "learning_rate": 7.66515864363997e-05,
      "loss": 0.0136,
      "step": 71
    },
    {
      "epoch": 0.11612903225806452,
      "grad_norm": 0.1284315288066864,
      "learning_rate": 7.594847868906076e-05,
      "loss": 0.0052,
      "step": 72
    },
    {
      "epoch": 0.11774193548387096,
      "grad_norm": 0.41634923219680786,
      "learning_rate": 7.52382768867422e-05,
      "loss": 0.0169,
      "step": 73
    },
    {
      "epoch": 0.11935483870967742,
      "grad_norm": 0.2020844668149948,
      "learning_rate": 7.452117519152542e-05,
      "loss": 0.0286,
      "step": 74
    },
    {
      "epoch": 0.12096774193548387,
      "grad_norm": 0.4097118377685547,
      "learning_rate": 7.379736965185368e-05,
      "loss": 0.008,
      "step": 75
    },
    {
      "epoch": 0.12258064516129032,
      "grad_norm": 0.11246518790721893,
      "learning_rate": 7.30670581489344e-05,
      "loss": 0.0045,
      "step": 76
    },
    {
      "epoch": 0.12419354838709677,
      "grad_norm": 0.45286452770233154,
      "learning_rate": 7.233044034264034e-05,
      "loss": 0.0197,
      "step": 77
    },
    {
      "epoch": 0.12580645161290321,
      "grad_norm": 0.6798192262649536,
      "learning_rate": 7.158771761692464e-05,
      "loss": 0.0305,
      "step": 78
    },
    {
      "epoch": 0.12741935483870967,
      "grad_norm": 0.9303900599479675,
      "learning_rate": 7.083909302476453e-05,
      "loss": 0.0721,
      "step": 79
    },
    {
      "epoch": 0.12903225806451613,
      "grad_norm": 0.7090916633605957,
      "learning_rate": 7.008477123264848e-05,
      "loss": 0.0278,
      "step": 80
    },
    {
      "epoch": 0.13064516129032258,
      "grad_norm": 0.7841101884841919,
      "learning_rate": 6.932495846462261e-05,
      "loss": 0.0387,
      "step": 81
    },
    {
      "epoch": 0.13225806451612904,
      "grad_norm": 0.03361155465245247,
      "learning_rate": 6.855986244591104e-05,
      "loss": 0.0017,
      "step": 82
    },
    {
      "epoch": 0.1338709677419355,
      "grad_norm": 0.1516152322292328,
      "learning_rate": 6.778969234612584e-05,
      "loss": 0.0035,
      "step": 83
    },
    {
      "epoch": 0.13548387096774195,
      "grad_norm": 0.5641607046127319,
      "learning_rate": 6.701465872208216e-05,
      "loss": 0.0269,
      "step": 84
    },
    {
      "epoch": 0.13709677419354838,
      "grad_norm": 1.0881990194320679,
      "learning_rate": 6.623497346023418e-05,
      "loss": 0.0965,
      "step": 85
    },
    {
      "epoch": 0.13870967741935483,
      "grad_norm": 0.9148401021957397,
      "learning_rate": 6.545084971874738e-05,
      "loss": 0.0939,
      "step": 86
    },
    {
      "epoch": 0.1403225806451613,
      "grad_norm": 0.5051007866859436,
      "learning_rate": 6.466250186922325e-05,
      "loss": 0.0244,
      "step": 87
    },
    {
      "epoch": 0.14193548387096774,
      "grad_norm": 0.33277037739753723,
      "learning_rate": 6.387014543809223e-05,
      "loss": 0.0206,
      "step": 88
    },
    {
      "epoch": 0.1435483870967742,
      "grad_norm": 0.05165182799100876,
      "learning_rate": 6.307399704769099e-05,
      "loss": 0.0026,
      "step": 89
    },
    {
      "epoch": 0.14516129032258066,
      "grad_norm": 0.8755428194999695,
      "learning_rate": 6.227427435703997e-05,
      "loss": 0.0526,
      "step": 90
    },
    {
      "epoch": 0.14677419354838708,
      "grad_norm": 0.4138418734073639,
      "learning_rate": 6.147119600233758e-05,
      "loss": 0.0187,
      "step": 91
    },
    {
      "epoch": 0.14838709677419354,
      "grad_norm": 1.0819470882415771,
      "learning_rate": 6.066498153718735e-05,
      "loss": 0.0331,
      "step": 92
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.10007546097040176,
      "learning_rate": 5.985585137257401e-05,
      "loss": 0.0043,
      "step": 93
    },
    {
      "epoch": 0.15161290322580645,
      "grad_norm": 0.0890575647354126,
      "learning_rate": 5.90440267166055e-05,
      "loss": 0.0046,
      "step": 94
    },
    {
      "epoch": 0.1532258064516129,
      "grad_norm": 0.9824152588844299,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 0.0265,
      "step": 95
    },
    {
      "epoch": 0.15483870967741936,
      "grad_norm": 0.061418917030096054,
      "learning_rate": 5.74131823855921e-05,
      "loss": 0.003,
      "step": 96
    },
    {
      "epoch": 0.15645161290322582,
      "grad_norm": 0.07597987353801727,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 0.0037,
      "step": 97
    },
    {
      "epoch": 0.15806451612903225,
      "grad_norm": 1.8876456022262573,
      "learning_rate": 5.577423184847932e-05,
      "loss": 0.0701,
      "step": 98
    },
    {
      "epoch": 0.1596774193548387,
      "grad_norm": 0.22971439361572266,
      "learning_rate": 5.495227651252315e-05,
      "loss": 0.0046,
      "step": 99
    },
    {
      "epoch": 0.16129032258064516,
      "grad_norm": 0.6317911148071289,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 0.0164,
      "step": 100
    },
    {
      "epoch": 0.16129032258064516,
      "eval_loss": 0.0026835536118596792,
      "eval_runtime": 109.2889,
      "eval_samples_per_second": 9.553,
      "eval_steps_per_second": 2.388,
      "step": 100
    },
    {
      "epoch": 0.1629032258064516,
      "grad_norm": 0.13364097476005554,
      "learning_rate": 5.330452921628497e-05,
      "loss": 0.0067,
      "step": 101
    },
    {
      "epoch": 0.16451612903225807,
      "grad_norm": 0.36112499237060547,
      "learning_rate": 5.247918773366112e-05,
      "loss": 0.0166,
      "step": 102
    },
    {
      "epoch": 0.16612903225806452,
      "grad_norm": 0.3356890082359314,
      "learning_rate": 5.165316846586541e-05,
      "loss": 0.0147,
      "step": 103
    },
    {
      "epoch": 0.16774193548387098,
      "grad_norm": 0.1406908631324768,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 0.0142,
      "step": 104
    },
    {
      "epoch": 0.1693548387096774,
      "grad_norm": 0.3340584635734558,
      "learning_rate": 5e-05,
      "loss": 0.0101,
      "step": 105
    },
    {
      "epoch": 0.17096774193548386,
      "grad_norm": 0.24642154574394226,
      "learning_rate": 4.917330276168208e-05,
      "loss": 0.0096,
      "step": 106
    },
    {
      "epoch": 0.17258064516129032,
      "grad_norm": 0.2866198718547821,
      "learning_rate": 4.834683153413459e-05,
      "loss": 0.0103,
      "step": 107
    },
    {
      "epoch": 0.17419354838709677,
      "grad_norm": 0.063316710293293,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 0.0019,
      "step": 108
    },
    {
      "epoch": 0.17580645161290323,
      "grad_norm": 0.3815213441848755,
      "learning_rate": 4.669547078371504e-05,
      "loss": 0.0111,
      "step": 109
    },
    {
      "epoch": 0.1774193548387097,
      "grad_norm": 0.49047574400901794,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 0.0176,
      "step": 110
    },
    {
      "epoch": 0.17903225806451614,
      "grad_norm": 0.329447865486145,
      "learning_rate": 4.504772348747687e-05,
      "loss": 0.0208,
      "step": 111
    },
    {
      "epoch": 0.18064516129032257,
      "grad_norm": 0.23379217088222504,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 0.0129,
      "step": 112
    },
    {
      "epoch": 0.18225806451612903,
      "grad_norm": 0.3418707251548767,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 0.023,
      "step": 113
    },
    {
      "epoch": 0.18387096774193548,
      "grad_norm": 0.3231382668018341,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 0.0211,
      "step": 114
    },
    {
      "epoch": 0.18548387096774194,
      "grad_norm": 0.2852570414543152,
      "learning_rate": 4.17702704859633e-05,
      "loss": 0.0215,
      "step": 115
    },
    {
      "epoch": 0.1870967741935484,
      "grad_norm": 0.37215906381607056,
      "learning_rate": 4.095597328339452e-05,
      "loss": 0.0114,
      "step": 116
    },
    {
      "epoch": 0.18870967741935485,
      "grad_norm": 0.1520654708147049,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 0.0055,
      "step": 117
    },
    {
      "epoch": 0.19032258064516128,
      "grad_norm": 0.2428535670042038,
      "learning_rate": 3.933501846281267e-05,
      "loss": 0.0073,
      "step": 118
    },
    {
      "epoch": 0.19193548387096773,
      "grad_norm": 0.33777958154678345,
      "learning_rate": 3.852880399766243e-05,
      "loss": 0.0269,
      "step": 119
    },
    {
      "epoch": 0.1935483870967742,
      "grad_norm": 0.16641807556152344,
      "learning_rate": 3.772572564296005e-05,
      "loss": 0.005,
      "step": 120
    },
    {
      "epoch": 0.19516129032258064,
      "grad_norm": 0.2343873828649521,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 0.0077,
      "step": 121
    },
    {
      "epoch": 0.1967741935483871,
      "grad_norm": 0.05236392095685005,
      "learning_rate": 3.612985456190778e-05,
      "loss": 0.0023,
      "step": 122
    },
    {
      "epoch": 0.19838709677419356,
      "grad_norm": 0.35883286595344543,
      "learning_rate": 3.533749813077677e-05,
      "loss": 0.0192,
      "step": 123
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.22238776087760925,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 0.0082,
      "step": 124
    },
    {
      "epoch": 0.20161290322580644,
      "grad_norm": 0.40833768248558044,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 0.0072,
      "step": 125
    },
    {
      "epoch": 0.2032258064516129,
      "grad_norm": 0.08354021608829498,
      "learning_rate": 3.298534127791785e-05,
      "loss": 0.0026,
      "step": 126
    },
    {
      "epoch": 0.20483870967741935,
      "grad_norm": 0.10409270972013474,
      "learning_rate": 3.221030765387417e-05,
      "loss": 0.0029,
      "step": 127
    },
    {
      "epoch": 0.2064516129032258,
      "grad_norm": 0.09194610267877579,
      "learning_rate": 3.144013755408895e-05,
      "loss": 0.0032,
      "step": 128
    },
    {
      "epoch": 0.20806451612903226,
      "grad_norm": 0.05515625327825546,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 0.0024,
      "step": 129
    },
    {
      "epoch": 0.20967741935483872,
      "grad_norm": 0.41631606221199036,
      "learning_rate": 2.991522876735154e-05,
      "loss": 0.0086,
      "step": 130
    },
    {
      "epoch": 0.21129032258064517,
      "grad_norm": 0.42728376388549805,
      "learning_rate": 2.916090697523549e-05,
      "loss": 0.0199,
      "step": 131
    },
    {
      "epoch": 0.2129032258064516,
      "grad_norm": 0.028320245444774628,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 0.0011,
      "step": 132
    },
    {
      "epoch": 0.21451612903225806,
      "grad_norm": 0.5399638414382935,
      "learning_rate": 2.766955965735968e-05,
      "loss": 0.0074,
      "step": 133
    },
    {
      "epoch": 0.2161290322580645,
      "grad_norm": 0.4056585133075714,
      "learning_rate": 2.693294185106562e-05,
      "loss": 0.0373,
      "step": 134
    },
    {
      "epoch": 0.21774193548387097,
      "grad_norm": 0.024112937971949577,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 0.001,
      "step": 135
    },
    {
      "epoch": 0.21935483870967742,
      "grad_norm": 0.5993265509605408,
      "learning_rate": 2.547882480847461e-05,
      "loss": 0.0085,
      "step": 136
    },
    {
      "epoch": 0.22096774193548388,
      "grad_norm": 0.01217877957969904,
      "learning_rate": 2.476172311325783e-05,
      "loss": 0.0007,
      "step": 137
    },
    {
      "epoch": 0.22258064516129034,
      "grad_norm": 0.027175255119800568,
      "learning_rate": 2.405152131093926e-05,
      "loss": 0.0011,
      "step": 138
    },
    {
      "epoch": 0.22419354838709676,
      "grad_norm": 0.2425202876329422,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 0.0095,
      "step": 139
    },
    {
      "epoch": 0.22580645161290322,
      "grad_norm": 0.023758340626955032,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 0.0011,
      "step": 140
    },
    {
      "epoch": 0.22741935483870968,
      "grad_norm": 0.09955063462257385,
      "learning_rate": 2.196424713241637e-05,
      "loss": 0.0026,
      "step": 141
    },
    {
      "epoch": 0.22903225806451613,
      "grad_norm": 0.053477123379707336,
      "learning_rate": 2.128356686585282e-05,
      "loss": 0.0016,
      "step": 142
    },
    {
      "epoch": 0.2306451612903226,
      "grad_norm": 0.24645189940929413,
      "learning_rate": 2.061073738537635e-05,
      "loss": 0.0056,
      "step": 143
    },
    {
      "epoch": 0.23225806451612904,
      "grad_norm": 0.9248583316802979,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 0.0523,
      "step": 144
    },
    {
      "epoch": 0.23387096774193547,
      "grad_norm": 0.018480895087122917,
      "learning_rate": 1.928936436551661e-05,
      "loss": 0.0008,
      "step": 145
    },
    {
      "epoch": 0.23548387096774193,
      "grad_norm": 0.11874135583639145,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 0.0032,
      "step": 146
    },
    {
      "epoch": 0.23709677419354838,
      "grad_norm": 1.3977549076080322,
      "learning_rate": 1.800157297483417e-05,
      "loss": 0.0507,
      "step": 147
    },
    {
      "epoch": 0.23870967741935484,
      "grad_norm": 0.04528358578681946,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 0.0012,
      "step": 148
    },
    {
      "epoch": 0.2403225806451613,
      "grad_norm": 2.37878155708313,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 0.1519,
      "step": 149
    },
    {
      "epoch": 0.24193548387096775,
      "grad_norm": 2.7409751415252686,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 0.0659,
      "step": 150
    },
    {
      "epoch": 0.24193548387096775,
      "eval_loss": 0.0014413581229746342,
      "eval_runtime": 109.2477,
      "eval_samples_per_second": 9.556,
      "eval_steps_per_second": 2.389,
      "step": 150
    },
    {
      "epoch": 0.2435483870967742,
      "grad_norm": 0.20558814704418182,
      "learning_rate": 1.553232954407171e-05,
      "loss": 0.0059,
      "step": 151
    },
    {
      "epoch": 0.24516129032258063,
      "grad_norm": 0.4324539303779602,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 0.019,
      "step": 152
    },
    {
      "epoch": 0.2467741935483871,
      "grad_norm": 0.1525496542453766,
      "learning_rate": 1.435357758543015e-05,
      "loss": 0.006,
      "step": 153
    },
    {
      "epoch": 0.24838709677419354,
      "grad_norm": 0.108416847884655,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 0.0055,
      "step": 154
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.06048719584941864,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 0.0018,
      "step": 155
    },
    {
      "epoch": 0.25161290322580643,
      "grad_norm": 0.08209064602851868,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 0.0019,
      "step": 156
    },
    {
      "epoch": 0.2532258064516129,
      "grad_norm": 0.3528808057308197,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 0.0289,
      "step": 157
    },
    {
      "epoch": 0.25483870967741934,
      "grad_norm": 0.7726386785507202,
      "learning_rate": 1.157994445715706e-05,
      "loss": 0.0571,
      "step": 158
    },
    {
      "epoch": 0.2564516129032258,
      "grad_norm": 0.17986443638801575,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 0.007,
      "step": 159
    },
    {
      "epoch": 0.25806451612903225,
      "grad_norm": 0.0820995420217514,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 0.0017,
      "step": 160
    },
    {
      "epoch": 0.25967741935483873,
      "grad_norm": 0.06043541431427002,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 0.0011,
      "step": 161
    },
    {
      "epoch": 0.26129032258064516,
      "grad_norm": 0.19384455680847168,
      "learning_rate": 9.549150281252633e-06,
      "loss": 0.0052,
      "step": 162
    },
    {
      "epoch": 0.2629032258064516,
      "grad_norm": 0.1950886845588684,
      "learning_rate": 9.068759265665384e-06,
      "loss": 0.0042,
      "step": 163
    },
    {
      "epoch": 0.2645161290322581,
      "grad_norm": 0.10976294428110123,
      "learning_rate": 8.599558442598998e-06,
      "loss": 0.0041,
      "step": 164
    },
    {
      "epoch": 0.2661290322580645,
      "grad_norm": 0.16872967779636383,
      "learning_rate": 8.141676086873572e-06,
      "loss": 0.008,
      "step": 165
    },
    {
      "epoch": 0.267741935483871,
      "grad_norm": 0.014217187650501728,
      "learning_rate": 7.695237378953223e-06,
      "loss": 0.0007,
      "step": 166
    },
    {
      "epoch": 0.2693548387096774,
      "grad_norm": 0.6291452050209045,
      "learning_rate": 7.260364370723044e-06,
      "loss": 0.0248,
      "step": 167
    },
    {
      "epoch": 0.2709677419354839,
      "grad_norm": 0.13673190772533417,
      "learning_rate": 6.837175952121306e-06,
      "loss": 0.0054,
      "step": 168
    },
    {
      "epoch": 0.2725806451612903,
      "grad_norm": 0.19489002227783203,
      "learning_rate": 6.425787818636131e-06,
      "loss": 0.0105,
      "step": 169
    },
    {
      "epoch": 0.27419354838709675,
      "grad_norm": 0.020141128450632095,
      "learning_rate": 6.026312439675552e-06,
      "loss": 0.0009,
      "step": 170
    },
    {
      "epoch": 0.27580645161290324,
      "grad_norm": 0.10351195186376572,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 0.0048,
      "step": 171
    },
    {
      "epoch": 0.27741935483870966,
      "grad_norm": 0.05539055913686752,
      "learning_rate": 5.263533508961827e-06,
      "loss": 0.002,
      "step": 172
    },
    {
      "epoch": 0.27903225806451615,
      "grad_norm": 0.22317983210086823,
      "learning_rate": 4.900438493352055e-06,
      "loss": 0.0061,
      "step": 173
    },
    {
      "epoch": 0.2806451612903226,
      "grad_norm": 0.050031088292598724,
      "learning_rate": 4.549673247541875e-06,
      "loss": 0.0012,
      "step": 174
    },
    {
      "epoch": 0.28225806451612906,
      "grad_norm": 0.7509785294532776,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 0.0258,
      "step": 175
    },
    {
      "epoch": 0.2838709677419355,
      "grad_norm": 0.07676703482866287,
      "learning_rate": 3.885512251130763e-06,
      "loss": 0.0024,
      "step": 176
    },
    {
      "epoch": 0.2854838709677419,
      "grad_norm": 1.8527617454528809,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 0.0128,
      "step": 177
    },
    {
      "epoch": 0.2870967741935484,
      "grad_norm": 0.32566869258880615,
      "learning_rate": 3.271776770026963e-06,
      "loss": 0.0298,
      "step": 178
    },
    {
      "epoch": 0.2887096774193548,
      "grad_norm": 0.09900537133216858,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 0.0031,
      "step": 179
    },
    {
      "epoch": 0.2903225806451613,
      "grad_norm": 0.5248034000396729,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 0.035,
      "step": 180
    },
    {
      "epoch": 0.29193548387096774,
      "grad_norm": 0.07472267746925354,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 0.0036,
      "step": 181
    },
    {
      "epoch": 0.29354838709677417,
      "grad_norm": 0.5043706297874451,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 0.0229,
      "step": 182
    },
    {
      "epoch": 0.29516129032258065,
      "grad_norm": 0.05585373193025589,
      "learning_rate": 1.962316193157593e-06,
      "loss": 0.0014,
      "step": 183
    },
    {
      "epoch": 0.2967741935483871,
      "grad_norm": 0.8581807613372803,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 0.0046,
      "step": 184
    },
    {
      "epoch": 0.29838709677419356,
      "grad_norm": 0.5122403502464294,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 0.0553,
      "step": 185
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.04005197435617447,
      "learning_rate": 1.333670137599713e-06,
      "loss": 0.0012,
      "step": 186
    },
    {
      "epoch": 0.3016129032258065,
      "grad_norm": 0.12912607192993164,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 0.0031,
      "step": 187
    },
    {
      "epoch": 0.3032258064516129,
      "grad_norm": 0.01144851092249155,
      "learning_rate": 9.810017062595322e-07,
      "loss": 0.0004,
      "step": 188
    },
    {
      "epoch": 0.30483870967741933,
      "grad_norm": 0.8459249138832092,
      "learning_rate": 8.247462563808817e-07,
      "loss": 0.016,
      "step": 189
    },
    {
      "epoch": 0.3064516129032258,
      "grad_norm": 0.14609886705875397,
      "learning_rate": 6.819348298638839e-07,
      "loss": 0.0032,
      "step": 190
    },
    {
      "epoch": 0.30806451612903224,
      "grad_norm": 0.4559611976146698,
      "learning_rate": 5.526064699265753e-07,
      "loss": 0.013,
      "step": 191
    },
    {
      "epoch": 0.3096774193548387,
      "grad_norm": 0.17169395089149475,
      "learning_rate": 4.367965336512403e-07,
      "loss": 0.0035,
      "step": 192
    },
    {
      "epoch": 0.31129032258064515,
      "grad_norm": 0.008081377483904362,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 0.0004,
      "step": 193
    },
    {
      "epoch": 0.31290322580645163,
      "grad_norm": 0.13021500408649445,
      "learning_rate": 2.458548727494292e-07,
      "loss": 0.003,
      "step": 194
    },
    {
      "epoch": 0.31451612903225806,
      "grad_norm": 1.0013407468795776,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 0.0558,
      "step": 195
    },
    {
      "epoch": 0.3161290322580645,
      "grad_norm": 0.12518633902072906,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 0.0026,
      "step": 196
    },
    {
      "epoch": 0.317741935483871,
      "grad_norm": 0.517772376537323,
      "learning_rate": 6.150154258476315e-08,
      "loss": 0.0114,
      "step": 197
    },
    {
      "epoch": 0.3193548387096774,
      "grad_norm": 4.395148277282715,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 0.0673,
      "step": 198
    },
    {
      "epoch": 0.3209677419354839,
      "grad_norm": 0.10897130519151688,
      "learning_rate": 6.834750376549792e-09,
      "loss": 0.0038,
      "step": 199
    },
    {
      "epoch": 0.3225806451612903,
      "grad_norm": 0.2314477562904358,
      "learning_rate": 0.0,
      "loss": 0.0058,
      "step": 200
    },
    {
      "epoch": 0.3225806451612903,
      "eval_loss": 0.0014145222958177328,
      "eval_runtime": 109.1299,
      "eval_samples_per_second": 9.567,
      "eval_steps_per_second": 2.392,
      "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": 5.09187566075904e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}