{
  "best_metric": 1.3634288311004639,
  "best_model_checkpoint": "miner_id_24/checkpoint-100",
  "epoch": 2.150537634408602,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.010752688172043012,
      "grad_norm": 4.756673812866211,
      "learning_rate": 1e-05,
      "loss": 2.5608,
      "step": 1
    },
    {
      "epoch": 0.010752688172043012,
      "eval_loss": 2.5321671962738037,
      "eval_runtime": 4.679,
      "eval_samples_per_second": 33.554,
      "eval_steps_per_second": 8.549,
      "step": 1
    },
    {
      "epoch": 0.021505376344086023,
      "grad_norm": 4.816965579986572,
      "learning_rate": 2e-05,
      "loss": 2.6608,
      "step": 2
    },
    {
      "epoch": 0.03225806451612903,
      "grad_norm": 4.787729740142822,
      "learning_rate": 3e-05,
      "loss": 2.149,
      "step": 3
    },
    {
      "epoch": 0.043010752688172046,
      "grad_norm": 4.805387496948242,
      "learning_rate": 4e-05,
      "loss": 2.223,
      "step": 4
    },
    {
      "epoch": 0.053763440860215055,
      "grad_norm": 3.398771047592163,
      "learning_rate": 5e-05,
      "loss": 2.2873,
      "step": 5
    },
    {
      "epoch": 0.06451612903225806,
      "grad_norm": 3.4376628398895264,
      "learning_rate": 6e-05,
      "loss": 2.0839,
      "step": 6
    },
    {
      "epoch": 0.07526881720430108,
      "grad_norm": 2.8605246543884277,
      "learning_rate": 7e-05,
      "loss": 2.0325,
      "step": 7
    },
    {
      "epoch": 0.08602150537634409,
      "grad_norm": 2.5606484413146973,
      "learning_rate": 8e-05,
      "loss": 1.8758,
      "step": 8
    },
    {
      "epoch": 0.0967741935483871,
      "grad_norm": 2.2748050689697266,
      "learning_rate": 9e-05,
      "loss": 1.6394,
      "step": 9
    },
    {
      "epoch": 0.10752688172043011,
      "grad_norm": 2.0811047554016113,
      "learning_rate": 0.0001,
      "loss": 1.6593,
      "step": 10
    },
    {
      "epoch": 0.11827956989247312,
      "grad_norm": 1.8930469751358032,
      "learning_rate": 9.999316524962345e-05,
      "loss": 1.5287,
      "step": 11
    },
    {
      "epoch": 0.12903225806451613,
      "grad_norm": 1.979614019393921,
      "learning_rate": 9.997266286704631e-05,
      "loss": 1.4822,
      "step": 12
    },
    {
      "epoch": 0.13978494623655913,
      "grad_norm": 2.106611490249634,
      "learning_rate": 9.993849845741524e-05,
      "loss": 1.4942,
      "step": 13
    },
    {
      "epoch": 0.15053763440860216,
      "grad_norm": 1.8150923252105713,
      "learning_rate": 9.989068136093873e-05,
      "loss": 1.5086,
      "step": 14
    },
    {
      "epoch": 0.16129032258064516,
      "grad_norm": 2.2123961448669434,
      "learning_rate": 9.98292246503335e-05,
      "loss": 1.572,
      "step": 15
    },
    {
      "epoch": 0.17204301075268819,
      "grad_norm": 2.053987741470337,
      "learning_rate": 9.975414512725057e-05,
      "loss": 1.6929,
      "step": 16
    },
    {
      "epoch": 0.1827956989247312,
      "grad_norm": 2.5067951679229736,
      "learning_rate": 9.966546331768191e-05,
      "loss": 1.4918,
      "step": 17
    },
    {
      "epoch": 0.1935483870967742,
      "grad_norm": 2.0683281421661377,
      "learning_rate": 9.956320346634876e-05,
      "loss": 1.3675,
      "step": 18
    },
    {
      "epoch": 0.20430107526881722,
      "grad_norm": 2.099067211151123,
      "learning_rate": 9.944739353007344e-05,
      "loss": 1.6336,
      "step": 19
    },
    {
      "epoch": 0.21505376344086022,
      "grad_norm": 2.0847325325012207,
      "learning_rate": 9.931806517013612e-05,
      "loss": 1.783,
      "step": 20
    },
    {
      "epoch": 0.22580645161290322,
      "grad_norm": 2.2268128395080566,
      "learning_rate": 9.917525374361912e-05,
      "loss": 1.6076,
      "step": 21
    },
    {
      "epoch": 0.23655913978494625,
      "grad_norm": 1.886684775352478,
      "learning_rate": 9.901899829374047e-05,
      "loss": 1.553,
      "step": 22
    },
    {
      "epoch": 0.24731182795698925,
      "grad_norm": 1.7694257497787476,
      "learning_rate": 9.884934153917997e-05,
      "loss": 1.5752,
      "step": 23
    },
    {
      "epoch": 0.25806451612903225,
      "grad_norm": 1.601288080215454,
      "learning_rate": 9.86663298624003e-05,
      "loss": 1.4431,
      "step": 24
    },
    {
      "epoch": 0.26881720430107525,
      "grad_norm": 1.6181790828704834,
      "learning_rate": 9.847001329696653e-05,
      "loss": 1.4098,
      "step": 25
    },
    {
      "epoch": 0.27956989247311825,
      "grad_norm": 1.6729943752288818,
      "learning_rate": 9.826044551386744e-05,
      "loss": 1.5202,
      "step": 26
    },
    {
      "epoch": 0.2903225806451613,
      "grad_norm": 1.6945127248764038,
      "learning_rate": 9.803768380684242e-05,
      "loss": 1.4196,
      "step": 27
    },
    {
      "epoch": 0.3010752688172043,
      "grad_norm": 1.5819880962371826,
      "learning_rate": 9.780178907671789e-05,
      "loss": 1.4923,
      "step": 28
    },
    {
      "epoch": 0.3118279569892473,
      "grad_norm": 1.6787797212600708,
      "learning_rate": 9.755282581475769e-05,
      "loss": 1.5269,
      "step": 29
    },
    {
      "epoch": 0.3225806451612903,
      "grad_norm": 1.463314414024353,
      "learning_rate": 9.729086208503174e-05,
      "loss": 1.3051,
      "step": 30
    },
    {
      "epoch": 0.3333333333333333,
      "grad_norm": 1.5788483619689941,
      "learning_rate": 9.701596950580806e-05,
      "loss": 1.4101,
      "step": 31
    },
    {
      "epoch": 0.34408602150537637,
      "grad_norm": 1.7896450757980347,
      "learning_rate": 9.672822322997305e-05,
      "loss": 1.5644,
      "step": 32
    },
    {
      "epoch": 0.3548387096774194,
      "grad_norm": 1.8968546390533447,
      "learning_rate": 9.642770192448536e-05,
      "loss": 1.494,
      "step": 33
    },
    {
      "epoch": 0.3655913978494624,
      "grad_norm": 1.6349527835845947,
      "learning_rate": 9.611448774886924e-05,
      "loss": 1.5081,
      "step": 34
    },
    {
      "epoch": 0.3763440860215054,
      "grad_norm": 1.6899402141571045,
      "learning_rate": 9.578866633275288e-05,
      "loss": 1.3583,
      "step": 35
    },
    {
      "epoch": 0.3870967741935484,
      "grad_norm": 1.5430152416229248,
      "learning_rate": 9.545032675245813e-05,
      "loss": 1.5876,
      "step": 36
    },
    {
      "epoch": 0.3978494623655914,
      "grad_norm": 1.892477035522461,
      "learning_rate": 9.509956150664796e-05,
      "loss": 1.4758,
      "step": 37
    },
    {
      "epoch": 0.40860215053763443,
      "grad_norm": 1.8843427896499634,
      "learning_rate": 9.473646649103818e-05,
      "loss": 1.5216,
      "step": 38
    },
    {
      "epoch": 0.41935483870967744,
      "grad_norm": 1.5942879915237427,
      "learning_rate": 9.43611409721806e-05,
      "loss": 1.2837,
      "step": 39
    },
    {
      "epoch": 0.43010752688172044,
      "grad_norm": 1.7020330429077148,
      "learning_rate": 9.397368756032445e-05,
      "loss": 1.5806,
      "step": 40
    },
    {
      "epoch": 0.44086021505376344,
      "grad_norm": 1.6075165271759033,
      "learning_rate": 9.357421218136386e-05,
      "loss": 1.6825,
      "step": 41
    },
    {
      "epoch": 0.45161290322580644,
      "grad_norm": 1.648186206817627,
      "learning_rate": 9.316282404787871e-05,
      "loss": 1.5647,
      "step": 42
    },
    {
      "epoch": 0.46236559139784944,
      "grad_norm": 1.6380871534347534,
      "learning_rate": 9.273963562927695e-05,
      "loss": 1.3593,
      "step": 43
    },
    {
      "epoch": 0.4731182795698925,
      "grad_norm": 1.568885087966919,
      "learning_rate": 9.230476262104677e-05,
      "loss": 1.4245,
      "step": 44
    },
    {
      "epoch": 0.4838709677419355,
      "grad_norm": 1.6013675928115845,
      "learning_rate": 9.185832391312644e-05,
      "loss": 1.6231,
      "step": 45
    },
    {
      "epoch": 0.4946236559139785,
      "grad_norm": 1.326374888420105,
      "learning_rate": 9.140044155740101e-05,
      "loss": 1.3063,
      "step": 46
    },
    {
      "epoch": 0.5053763440860215,
      "grad_norm": 1.3383835554122925,
      "learning_rate": 9.093124073433463e-05,
      "loss": 1.4064,
      "step": 47
    },
    {
      "epoch": 0.5161290322580645,
      "grad_norm": 1.6862118244171143,
      "learning_rate": 9.045084971874738e-05,
      "loss": 1.638,
      "step": 48
    },
    {
      "epoch": 0.5268817204301075,
      "grad_norm": 1.552420973777771,
      "learning_rate": 8.995939984474624e-05,
      "loss": 1.4579,
      "step": 49
    },
    {
      "epoch": 0.5376344086021505,
      "grad_norm": 1.317513108253479,
      "learning_rate": 8.945702546981969e-05,
      "loss": 1.3294,
      "step": 50
    },
    {
      "epoch": 0.5376344086021505,
      "eval_loss": 1.3945931196212769,
      "eval_runtime": 4.7012,
      "eval_samples_per_second": 33.395,
      "eval_steps_per_second": 8.508,
      "step": 50
    },
    {
      "epoch": 0.5483870967741935,
      "grad_norm": 1.4079197645187378,
      "learning_rate": 8.894386393810563e-05,
      "loss": 1.2827,
      "step": 51
    },
    {
      "epoch": 0.5591397849462365,
      "grad_norm": 1.3906307220458984,
      "learning_rate": 8.842005554284296e-05,
      "loss": 1.4083,
      "step": 52
    },
    {
      "epoch": 0.5698924731182796,
      "grad_norm": 1.3120077848434448,
      "learning_rate": 8.788574348801675e-05,
      "loss": 1.3644,
      "step": 53
    },
    {
      "epoch": 0.5806451612903226,
      "grad_norm": 1.432735562324524,
      "learning_rate": 8.73410738492077e-05,
      "loss": 1.3207,
      "step": 54
    },
    {
      "epoch": 0.5913978494623656,
      "grad_norm": 1.3783055543899536,
      "learning_rate": 8.678619553365659e-05,
      "loss": 1.3582,
      "step": 55
    },
    {
      "epoch": 0.6021505376344086,
      "grad_norm": 1.436294436454773,
      "learning_rate": 8.622126023955446e-05,
      "loss": 1.4784,
      "step": 56
    },
    {
      "epoch": 0.6129032258064516,
      "grad_norm": 1.423822045326233,
      "learning_rate": 8.564642241456986e-05,
      "loss": 1.5012,
      "step": 57
    },
    {
      "epoch": 0.6236559139784946,
      "grad_norm": 1.4416776895523071,
      "learning_rate": 8.506183921362443e-05,
      "loss": 1.3789,
      "step": 58
    },
    {
      "epoch": 0.6344086021505376,
      "grad_norm": 1.6241978406906128,
      "learning_rate": 8.44676704559283e-05,
      "loss": 1.4068,
      "step": 59
    },
    {
      "epoch": 0.6451612903225806,
      "grad_norm": 1.4760600328445435,
      "learning_rate": 8.386407858128706e-05,
      "loss": 1.3556,
      "step": 60
    },
    {
      "epoch": 0.6559139784946236,
      "grad_norm": 1.4955905675888062,
      "learning_rate": 8.32512286056924e-05,
      "loss": 1.5192,
      "step": 61
    },
    {
      "epoch": 0.6666666666666666,
      "grad_norm": 1.5181070566177368,
      "learning_rate": 8.262928807620843e-05,
      "loss": 1.2904,
      "step": 62
    },
    {
      "epoch": 0.6774193548387096,
      "grad_norm": 1.4489530324935913,
      "learning_rate": 8.199842702516583e-05,
      "loss": 1.4523,
      "step": 63
    },
    {
      "epoch": 0.6881720430107527,
      "grad_norm": 1.4414947032928467,
      "learning_rate": 8.135881792367686e-05,
      "loss": 1.612,
      "step": 64
    },
    {
      "epoch": 0.6989247311827957,
      "grad_norm": 2.020447015762329,
      "learning_rate": 8.07106356344834e-05,
      "loss": 1.3674,
      "step": 65
    },
    {
      "epoch": 0.7096774193548387,
      "grad_norm": 1.6104086637496948,
      "learning_rate": 8.005405736415126e-05,
      "loss": 1.5316,
      "step": 66
    },
    {
      "epoch": 0.7204301075268817,
      "grad_norm": 1.472418189048767,
      "learning_rate": 7.938926261462366e-05,
      "loss": 1.4308,
      "step": 67
    },
    {
      "epoch": 0.7311827956989247,
      "grad_norm": 1.569532871246338,
      "learning_rate": 7.871643313414718e-05,
      "loss": 1.484,
      "step": 68
    },
    {
      "epoch": 0.7419354838709677,
      "grad_norm": 1.660690188407898,
      "learning_rate": 7.803575286758364e-05,
      "loss": 1.5779,
      "step": 69
    },
    {
      "epoch": 0.7526881720430108,
      "grad_norm": 1.4673008918762207,
      "learning_rate": 7.734740790612136e-05,
      "loss": 1.4693,
      "step": 70
    },
    {
      "epoch": 0.7634408602150538,
      "grad_norm": 1.3039233684539795,
      "learning_rate": 7.66515864363997e-05,
      "loss": 1.3442,
      "step": 71
    },
    {
      "epoch": 0.7741935483870968,
      "grad_norm": 1.3886308670043945,
      "learning_rate": 7.594847868906076e-05,
      "loss": 1.361,
      "step": 72
    },
    {
      "epoch": 0.7849462365591398,
      "grad_norm": 1.4487396478652954,
      "learning_rate": 7.52382768867422e-05,
      "loss": 1.4151,
      "step": 73
    },
    {
      "epoch": 0.7956989247311828,
      "grad_norm": 1.542792558670044,
      "learning_rate": 7.452117519152542e-05,
      "loss": 1.3144,
      "step": 74
    },
    {
      "epoch": 0.8064516129032258,
      "grad_norm": 1.4200122356414795,
      "learning_rate": 7.379736965185368e-05,
      "loss": 1.4351,
      "step": 75
    },
    {
      "epoch": 0.8172043010752689,
      "grad_norm": 1.4879090785980225,
      "learning_rate": 7.30670581489344e-05,
      "loss": 1.4793,
      "step": 76
    },
    {
      "epoch": 0.8279569892473119,
      "grad_norm": 1.4579113721847534,
      "learning_rate": 7.233044034264034e-05,
      "loss": 1.4308,
      "step": 77
    },
    {
      "epoch": 0.8387096774193549,
      "grad_norm": 1.218188762664795,
      "learning_rate": 7.158771761692464e-05,
      "loss": 1.4405,
      "step": 78
    },
    {
      "epoch": 0.8494623655913979,
      "grad_norm": 1.4777249097824097,
      "learning_rate": 7.083909302476453e-05,
      "loss": 1.5185,
      "step": 79
    },
    {
      "epoch": 0.8602150537634409,
      "grad_norm": 1.2023836374282837,
      "learning_rate": 7.008477123264848e-05,
      "loss": 1.3254,
      "step": 80
    },
    {
      "epoch": 0.8709677419354839,
      "grad_norm": 1.1741511821746826,
      "learning_rate": 6.932495846462261e-05,
      "loss": 1.1883,
      "step": 81
    },
    {
      "epoch": 0.8817204301075269,
      "grad_norm": 1.263417363166809,
      "learning_rate": 6.855986244591104e-05,
      "loss": 1.314,
      "step": 82
    },
    {
      "epoch": 0.8924731182795699,
      "grad_norm": 1.3069579601287842,
      "learning_rate": 6.778969234612584e-05,
      "loss": 1.18,
      "step": 83
    },
    {
      "epoch": 0.9032258064516129,
      "grad_norm": 1.2708874940872192,
      "learning_rate": 6.701465872208216e-05,
      "loss": 1.4669,
      "step": 84
    },
    {
      "epoch": 0.9139784946236559,
      "grad_norm": 1.2953261137008667,
      "learning_rate": 6.623497346023418e-05,
      "loss": 1.2906,
      "step": 85
    },
    {
      "epoch": 0.9247311827956989,
      "grad_norm": 1.4633235931396484,
      "learning_rate": 6.545084971874738e-05,
      "loss": 1.5025,
      "step": 86
    },
    {
      "epoch": 0.9354838709677419,
      "grad_norm": 1.3885836601257324,
      "learning_rate": 6.466250186922325e-05,
      "loss": 1.5074,
      "step": 87
    },
    {
      "epoch": 0.946236559139785,
      "grad_norm": 1.3254637718200684,
      "learning_rate": 6.387014543809223e-05,
      "loss": 1.4577,
      "step": 88
    },
    {
      "epoch": 0.956989247311828,
      "grad_norm": 1.4725390672683716,
      "learning_rate": 6.307399704769099e-05,
      "loss": 1.3316,
      "step": 89
    },
    {
      "epoch": 0.967741935483871,
      "grad_norm": 1.4616190195083618,
      "learning_rate": 6.227427435703997e-05,
      "loss": 1.5571,
      "step": 90
    },
    {
      "epoch": 0.978494623655914,
      "grad_norm": 1.3754075765609741,
      "learning_rate": 6.147119600233758e-05,
      "loss": 1.3809,
      "step": 91
    },
    {
      "epoch": 0.989247311827957,
      "grad_norm": 1.3891860246658325,
      "learning_rate": 6.066498153718735e-05,
      "loss": 1.4299,
      "step": 92
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4610061645507812,
      "learning_rate": 5.985585137257401e-05,
      "loss": 1.3112,
      "step": 93
    },
    {
      "epoch": 1.010752688172043,
      "grad_norm": 1.038821816444397,
      "learning_rate": 5.90440267166055e-05,
      "loss": 1.1235,
      "step": 94
    },
    {
      "epoch": 1.021505376344086,
      "grad_norm": 1.0727566480636597,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 1.1396,
      "step": 95
    },
    {
      "epoch": 1.032258064516129,
      "grad_norm": 1.0206246376037598,
      "learning_rate": 5.74131823855921e-05,
      "loss": 1.038,
      "step": 96
    },
    {
      "epoch": 1.043010752688172,
      "grad_norm": 0.9992400407791138,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 1.0465,
      "step": 97
    },
    {
      "epoch": 1.053763440860215,
      "grad_norm": 1.1006141901016235,
      "learning_rate": 5.577423184847932e-05,
      "loss": 1.1107,
      "step": 98
    },
    {
      "epoch": 1.064516129032258,
      "grad_norm": 1.1604828834533691,
      "learning_rate": 5.495227651252315e-05,
      "loss": 1.1146,
      "step": 99
    },
    {
      "epoch": 1.075268817204301,
      "grad_norm": 1.125632405281067,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 1.0629,
      "step": 100
    },
    {
      "epoch": 1.075268817204301,
      "eval_loss": 1.3634288311004639,
      "eval_runtime": 4.7059,
      "eval_samples_per_second": 33.363,
      "eval_steps_per_second": 8.5,
      "step": 100
    },
    {
      "epoch": 1.086021505376344,
      "grad_norm": 0.9249867796897888,
      "learning_rate": 5.330452921628497e-05,
      "loss": 1.0458,
      "step": 101
    },
    {
      "epoch": 1.096774193548387,
      "grad_norm": 1.0629762411117554,
      "learning_rate": 5.247918773366112e-05,
      "loss": 1.0836,
      "step": 102
    },
    {
      "epoch": 1.10752688172043,
      "grad_norm": 1.1164374351501465,
      "learning_rate": 5.165316846586541e-05,
      "loss": 1.1097,
      "step": 103
    },
    {
      "epoch": 1.118279569892473,
      "grad_norm": 1.129228115081787,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 1.069,
      "step": 104
    },
    {
      "epoch": 1.129032258064516,
      "grad_norm": 1.1376335620880127,
      "learning_rate": 5e-05,
      "loss": 0.9951,
      "step": 105
    },
    {
      "epoch": 1.139784946236559,
      "grad_norm": 1.2229422330856323,
      "learning_rate": 4.917330276168208e-05,
      "loss": 1.1898,
      "step": 106
    },
    {
      "epoch": 1.1505376344086022,
      "grad_norm": 1.1418484449386597,
      "learning_rate": 4.834683153413459e-05,
      "loss": 0.9968,
      "step": 107
    },
    {
      "epoch": 1.1612903225806452,
      "grad_norm": 1.2595585584640503,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 1.0579,
      "step": 108
    },
    {
      "epoch": 1.1720430107526882,
      "grad_norm": 1.1968224048614502,
      "learning_rate": 4.669547078371504e-05,
      "loss": 0.9448,
      "step": 109
    },
    {
      "epoch": 1.1827956989247312,
      "grad_norm": 1.3096959590911865,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 1.0923,
      "step": 110
    },
    {
      "epoch": 1.1935483870967742,
      "grad_norm": 1.3598382472991943,
      "learning_rate": 4.504772348747687e-05,
      "loss": 1.0318,
      "step": 111
    },
    {
      "epoch": 1.2043010752688172,
      "grad_norm": 1.3797881603240967,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 1.0698,
      "step": 112
    },
    {
      "epoch": 1.2150537634408602,
      "grad_norm": 1.4478576183319092,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 1.0424,
      "step": 113
    },
    {
      "epoch": 1.2258064516129032,
      "grad_norm": 1.4821935892105103,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 1.0264,
      "step": 114
    },
    {
      "epoch": 1.2365591397849462,
      "grad_norm": 1.716293215751648,
      "learning_rate": 4.17702704859633e-05,
      "loss": 1.2165,
      "step": 115
    },
    {
      "epoch": 1.2473118279569892,
      "grad_norm": 1.488614559173584,
      "learning_rate": 4.095597328339452e-05,
      "loss": 1.1289,
      "step": 116
    },
    {
      "epoch": 1.2580645161290323,
      "grad_norm": 1.7371461391448975,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 1.2733,
      "step": 117
    },
    {
      "epoch": 1.2688172043010753,
      "grad_norm": 1.7669775485992432,
      "learning_rate": 3.933501846281267e-05,
      "loss": 1.3276,
      "step": 118
    },
    {
      "epoch": 1.2795698924731183,
      "grad_norm": 1.5520764589309692,
      "learning_rate": 3.852880399766243e-05,
      "loss": 1.0845,
      "step": 119
    },
    {
      "epoch": 1.2903225806451613,
      "grad_norm": 1.2955957651138306,
      "learning_rate": 3.772572564296005e-05,
      "loss": 0.9734,
      "step": 120
    },
    {
      "epoch": 1.3010752688172043,
      "grad_norm": 1.396894097328186,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 0.8713,
      "step": 121
    },
    {
      "epoch": 1.3118279569892473,
      "grad_norm": 1.6373274326324463,
      "learning_rate": 3.612985456190778e-05,
      "loss": 0.9747,
      "step": 122
    },
    {
      "epoch": 1.3225806451612903,
      "grad_norm": 1.5140986442565918,
      "learning_rate": 3.533749813077677e-05,
      "loss": 0.952,
      "step": 123
    },
    {
      "epoch": 1.3333333333333333,
      "grad_norm": 1.687639594078064,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 1.1532,
      "step": 124
    },
    {
      "epoch": 1.3440860215053765,
      "grad_norm": 1.5184952020645142,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 1.1347,
      "step": 125
    },
    {
      "epoch": 1.3548387096774195,
      "grad_norm": 1.493153691291809,
      "learning_rate": 3.298534127791785e-05,
      "loss": 1.0142,
      "step": 126
    },
    {
      "epoch": 1.3655913978494625,
      "grad_norm": 1.5351260900497437,
      "learning_rate": 3.221030765387417e-05,
      "loss": 1.096,
      "step": 127
    },
    {
      "epoch": 1.3763440860215055,
      "grad_norm": 1.4562873840332031,
      "learning_rate": 3.144013755408895e-05,
      "loss": 0.969,
      "step": 128
    },
    {
      "epoch": 1.3870967741935485,
      "grad_norm": 1.398634672164917,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 1.0367,
      "step": 129
    },
    {
      "epoch": 1.3978494623655915,
      "grad_norm": 1.3472834825515747,
      "learning_rate": 2.991522876735154e-05,
      "loss": 1.086,
      "step": 130
    },
    {
      "epoch": 1.4086021505376345,
      "grad_norm": 1.482953429222107,
      "learning_rate": 2.916090697523549e-05,
      "loss": 1.1435,
      "step": 131
    },
    {
      "epoch": 1.4193548387096775,
      "grad_norm": 1.5652285814285278,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 1.135,
      "step": 132
    },
    {
      "epoch": 1.4301075268817205,
      "grad_norm": 1.6983011960983276,
      "learning_rate": 2.766955965735968e-05,
      "loss": 1.188,
      "step": 133
    },
    {
      "epoch": 1.4408602150537635,
      "grad_norm": 1.6841248273849487,
      "learning_rate": 2.693294185106562e-05,
      "loss": 1.1523,
      "step": 134
    },
    {
      "epoch": 1.4516129032258065,
      "grad_norm": 1.623253345489502,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 1.2135,
      "step": 135
    },
    {
      "epoch": 1.4623655913978495,
      "grad_norm": 1.3781989812850952,
      "learning_rate": 2.547882480847461e-05,
      "loss": 0.886,
      "step": 136
    },
    {
      "epoch": 1.4731182795698925,
      "grad_norm": 1.4850106239318848,
      "learning_rate": 2.476172311325783e-05,
      "loss": 1.0595,
      "step": 137
    },
    {
      "epoch": 1.4838709677419355,
      "grad_norm": 1.6122136116027832,
      "learning_rate": 2.405152131093926e-05,
      "loss": 1.0925,
      "step": 138
    },
    {
      "epoch": 1.4946236559139785,
      "grad_norm": 1.4394659996032715,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 0.9858,
      "step": 139
    },
    {
      "epoch": 1.5053763440860215,
      "grad_norm": 1.6203980445861816,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 1.1627,
      "step": 140
    },
    {
      "epoch": 1.5161290322580645,
      "grad_norm": 1.4615089893341064,
      "learning_rate": 2.196424713241637e-05,
      "loss": 1.0633,
      "step": 141
    },
    {
      "epoch": 1.5268817204301075,
      "grad_norm": 1.6260236501693726,
      "learning_rate": 2.128356686585282e-05,
      "loss": 1.1186,
      "step": 142
    },
    {
      "epoch": 1.5376344086021505,
      "grad_norm": 1.5490525960922241,
      "learning_rate": 2.061073738537635e-05,
      "loss": 1.0727,
      "step": 143
    },
    {
      "epoch": 1.5483870967741935,
      "grad_norm": 1.6757961511611938,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 1.2063,
      "step": 144
    },
    {
      "epoch": 1.5591397849462365,
      "grad_norm": 1.5960594415664673,
      "learning_rate": 1.928936436551661e-05,
      "loss": 1.1443,
      "step": 145
    },
    {
      "epoch": 1.5698924731182795,
      "grad_norm": 1.5980165004730225,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 1.1234,
      "step": 146
    },
    {
      "epoch": 1.5806451612903225,
      "grad_norm": 1.5133894681930542,
      "learning_rate": 1.800157297483417e-05,
      "loss": 1.0842,
      "step": 147
    },
    {
      "epoch": 1.5913978494623655,
      "grad_norm": 1.3797662258148193,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 0.9913,
      "step": 148
    },
    {
      "epoch": 1.6021505376344085,
      "grad_norm": 1.5195506811141968,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 1.1191,
      "step": 149
    },
    {
      "epoch": 1.6129032258064515,
      "grad_norm": 1.61488676071167,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 1.1218,
      "step": 150
    },
    {
      "epoch": 1.6129032258064515,
      "eval_loss": 1.3883932828903198,
      "eval_runtime": 4.7039,
      "eval_samples_per_second": 33.377,
      "eval_steps_per_second": 8.504,
      "step": 150
    },
    {
      "epoch": 1.6236559139784945,
      "grad_norm": 1.46273672580719,
      "learning_rate": 1.553232954407171e-05,
      "loss": 1.0234,
      "step": 151
    },
    {
      "epoch": 1.6344086021505375,
      "grad_norm": 1.3821216821670532,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 0.8185,
      "step": 152
    },
    {
      "epoch": 1.6451612903225805,
      "grad_norm": 1.404694676399231,
      "learning_rate": 1.435357758543015e-05,
      "loss": 1.1315,
      "step": 153
    },
    {
      "epoch": 1.6559139784946235,
      "grad_norm": 1.514056921005249,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 1.0138,
      "step": 154
    },
    {
      "epoch": 1.6666666666666665,
      "grad_norm": 1.4897308349609375,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 1.0857,
      "step": 155
    },
    {
      "epoch": 1.6774193548387095,
      "grad_norm": 1.6354196071624756,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 1.1565,
      "step": 156
    },
    {
      "epoch": 1.6881720430107527,
      "grad_norm": 1.5802398920059204,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 1.0531,
      "step": 157
    },
    {
      "epoch": 1.6989247311827957,
      "grad_norm": 1.6117963790893555,
      "learning_rate": 1.157994445715706e-05,
      "loss": 1.1798,
      "step": 158
    },
    {
      "epoch": 1.7096774193548387,
      "grad_norm": 1.840330719947815,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 1.1744,
      "step": 159
    },
    {
      "epoch": 1.7204301075268817,
      "grad_norm": 1.4968127012252808,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 1.1037,
      "step": 160
    },
    {
      "epoch": 1.7311827956989247,
      "grad_norm": 1.4755455255508423,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 1.0174,
      "step": 161
    },
    {
      "epoch": 1.7419354838709677,
      "grad_norm": 1.4550976753234863,
      "learning_rate": 9.549150281252633e-06,
      "loss": 1.0547,
      "step": 162
    },
    {
      "epoch": 1.7526881720430108,
      "grad_norm": 1.4910399913787842,
      "learning_rate": 9.068759265665384e-06,
      "loss": 1.0814,
      "step": 163
    },
    {
      "epoch": 1.7634408602150538,
      "grad_norm": 1.5773422718048096,
      "learning_rate": 8.599558442598998e-06,
      "loss": 1.0778,
      "step": 164
    },
    {
      "epoch": 1.7741935483870968,
      "grad_norm": 1.363403081893921,
      "learning_rate": 8.141676086873572e-06,
      "loss": 0.9098,
      "step": 165
    },
    {
      "epoch": 1.7849462365591398,
      "grad_norm": 1.6571729183197021,
      "learning_rate": 7.695237378953223e-06,
      "loss": 1.0602,
      "step": 166
    },
    {
      "epoch": 1.7956989247311828,
      "grad_norm": 1.5179351568222046,
      "learning_rate": 7.260364370723044e-06,
      "loss": 1.0759,
      "step": 167
    },
    {
      "epoch": 1.8064516129032258,
      "grad_norm": 1.4653751850128174,
      "learning_rate": 6.837175952121306e-06,
      "loss": 0.9642,
      "step": 168
    },
    {
      "epoch": 1.817204301075269,
      "grad_norm": 1.6933156251907349,
      "learning_rate": 6.425787818636131e-06,
      "loss": 1.1237,
      "step": 169
    },
    {
      "epoch": 1.827956989247312,
      "grad_norm": 1.512691855430603,
      "learning_rate": 6.026312439675552e-06,
      "loss": 1.0108,
      "step": 170
    },
    {
      "epoch": 1.838709677419355,
      "grad_norm": 1.5427347421646118,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 1.0399,
      "step": 171
    },
    {
      "epoch": 1.849462365591398,
      "grad_norm": 1.5393648147583008,
      "learning_rate": 5.263533508961827e-06,
      "loss": 1.0656,
      "step": 172
    },
    {
      "epoch": 1.860215053763441,
      "grad_norm": 1.3335708379745483,
      "learning_rate": 4.900438493352055e-06,
      "loss": 1.0725,
      "step": 173
    },
    {
      "epoch": 1.870967741935484,
      "grad_norm": 1.4599882364273071,
      "learning_rate": 4.549673247541875e-06,
      "loss": 0.9811,
      "step": 174
    },
    {
      "epoch": 1.881720430107527,
      "grad_norm": 1.4409252405166626,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 0.9519,
      "step": 175
    },
    {
      "epoch": 1.89247311827957,
      "grad_norm": 1.3919806480407715,
      "learning_rate": 3.885512251130763e-06,
      "loss": 0.9694,
      "step": 176
    },
    {
      "epoch": 1.903225806451613,
      "grad_norm": 1.4734621047973633,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 0.9512,
      "step": 177
    },
    {
      "epoch": 1.913978494623656,
      "grad_norm": 1.4967057704925537,
      "learning_rate": 3.271776770026963e-06,
      "loss": 0.9916,
      "step": 178
    },
    {
      "epoch": 1.924731182795699,
      "grad_norm": 1.6163455247879028,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 1.055,
      "step": 179
    },
    {
      "epoch": 1.935483870967742,
      "grad_norm": 1.6834897994995117,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 1.2419,
      "step": 180
    },
    {
      "epoch": 1.946236559139785,
      "grad_norm": 1.642762541770935,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 1.136,
      "step": 181
    },
    {
      "epoch": 1.956989247311828,
      "grad_norm": 1.566754937171936,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 1.0523,
      "step": 182
    },
    {
      "epoch": 1.967741935483871,
      "grad_norm": 1.8782109022140503,
      "learning_rate": 1.962316193157593e-06,
      "loss": 1.2276,
      "step": 183
    },
    {
      "epoch": 1.978494623655914,
      "grad_norm": 1.4430046081542969,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 1.0263,
      "step": 184
    },
    {
      "epoch": 1.989247311827957,
      "grad_norm": 1.6009927988052368,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 1.0518,
      "step": 185
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.8476316928863525,
      "learning_rate": 1.333670137599713e-06,
      "loss": 1.1282,
      "step": 186
    },
    {
      "epoch": 2.010752688172043,
      "grad_norm": 1.425358772277832,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 1.0551,
      "step": 187
    },
    {
      "epoch": 2.021505376344086,
      "grad_norm": 1.3462239503860474,
      "learning_rate": 9.810017062595322e-07,
      "loss": 1.0377,
      "step": 188
    },
    {
      "epoch": 2.032258064516129,
      "grad_norm": 1.2908916473388672,
      "learning_rate": 8.247462563808817e-07,
      "loss": 0.8787,
      "step": 189
    },
    {
      "epoch": 2.043010752688172,
      "grad_norm": 1.2711652517318726,
      "learning_rate": 6.819348298638839e-07,
      "loss": 0.8211,
      "step": 190
    },
    {
      "epoch": 2.053763440860215,
      "grad_norm": 1.3738279342651367,
      "learning_rate": 5.526064699265753e-07,
      "loss": 0.9055,
      "step": 191
    },
    {
      "epoch": 2.064516129032258,
      "grad_norm": 1.2861653566360474,
      "learning_rate": 4.367965336512403e-07,
      "loss": 0.9609,
      "step": 192
    },
    {
      "epoch": 2.075268817204301,
      "grad_norm": 1.3780279159545898,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 0.9242,
      "step": 193
    },
    {
      "epoch": 2.086021505376344,
      "grad_norm": 1.331591248512268,
      "learning_rate": 2.458548727494292e-07,
      "loss": 0.8347,
      "step": 194
    },
    {
      "epoch": 2.096774193548387,
      "grad_norm": 1.3115837574005127,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 0.923,
      "step": 195
    },
    {
      "epoch": 2.10752688172043,
      "grad_norm": 1.381766438484192,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 0.9425,
      "step": 196
    },
    {
      "epoch": 2.118279569892473,
      "grad_norm": 1.3318949937820435,
      "learning_rate": 6.150154258476315e-08,
      "loss": 0.9235,
      "step": 197
    },
    {
      "epoch": 2.129032258064516,
      "grad_norm": 1.3440535068511963,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 0.9412,
      "step": 198
    },
    {
      "epoch": 2.139784946236559,
      "grad_norm": 1.4999085664749146,
      "learning_rate": 6.834750376549792e-09,
      "loss": 0.9707,
      "step": 199
    },
    {
      "epoch": 2.150537634408602,
      "grad_norm": 1.3091585636138916,
      "learning_rate": 0.0,
      "loss": 0.9311,
      "step": 200
    },
    {
      "epoch": 2.150537634408602,
      "eval_loss": 1.3825842142105103,
      "eval_runtime": 4.7143,
      "eval_samples_per_second": 33.303,
      "eval_steps_per_second": 8.485,
      "step": 200
    }
  ],
  "logging_steps": 1,
  "max_steps": 200,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 50,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 5,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 2
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 5.495622655279104e+16,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}