{
  "best_metric": 1.3642938137054443,
  "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": 4.702356815338135,
      "learning_rate": 1e-05,
      "loss": 2.5608,
      "step": 1
    },
    {
      "epoch": 0.010752688172043012,
      "eval_loss": 2.5321671962738037,
      "eval_runtime": 4.7204,
      "eval_samples_per_second": 33.26,
      "eval_steps_per_second": 8.474,
      "step": 1
    },
    {
      "epoch": 0.021505376344086023,
      "grad_norm": 4.766476631164551,
      "learning_rate": 2e-05,
      "loss": 2.6608,
      "step": 2
    },
    {
      "epoch": 0.03225806451612903,
      "grad_norm": 4.727513313293457,
      "learning_rate": 3e-05,
      "loss": 2.1508,
      "step": 3
    },
    {
      "epoch": 0.043010752688172046,
      "grad_norm": 4.70632266998291,
      "learning_rate": 4e-05,
      "loss": 2.221,
      "step": 4
    },
    {
      "epoch": 0.053763440860215055,
      "grad_norm": 3.304126501083374,
      "learning_rate": 5e-05,
      "loss": 2.2863,
      "step": 5
    },
    {
      "epoch": 0.06451612903225806,
      "grad_norm": 3.3436529636383057,
      "learning_rate": 6e-05,
      "loss": 2.0891,
      "step": 6
    },
    {
      "epoch": 0.07526881720430108,
      "grad_norm": 2.757631540298462,
      "learning_rate": 7e-05,
      "loss": 2.0318,
      "step": 7
    },
    {
      "epoch": 0.08602150537634409,
      "grad_norm": 2.5397322177886963,
      "learning_rate": 8e-05,
      "loss": 1.8757,
      "step": 8
    },
    {
      "epoch": 0.0967741935483871,
      "grad_norm": 2.240288496017456,
      "learning_rate": 9e-05,
      "loss": 1.6397,
      "step": 9
    },
    {
      "epoch": 0.10752688172043011,
      "grad_norm": 2.0536859035491943,
      "learning_rate": 0.0001,
      "loss": 1.6601,
      "step": 10
    },
    {
      "epoch": 0.11827956989247312,
      "grad_norm": 1.8525017499923706,
      "learning_rate": 9.999316524962345e-05,
      "loss": 1.5248,
      "step": 11
    },
    {
      "epoch": 0.12903225806451613,
      "grad_norm": 1.9969803094863892,
      "learning_rate": 9.997266286704631e-05,
      "loss": 1.4807,
      "step": 12
    },
    {
      "epoch": 0.13978494623655913,
      "grad_norm": 2.01229190826416,
      "learning_rate": 9.993849845741524e-05,
      "loss": 1.4901,
      "step": 13
    },
    {
      "epoch": 0.15053763440860216,
      "grad_norm": 1.7899178266525269,
      "learning_rate": 9.989068136093873e-05,
      "loss": 1.5082,
      "step": 14
    },
    {
      "epoch": 0.16129032258064516,
      "grad_norm": 2.1389129161834717,
      "learning_rate": 9.98292246503335e-05,
      "loss": 1.5735,
      "step": 15
    },
    {
      "epoch": 0.17204301075268819,
      "grad_norm": 2.066209554672241,
      "learning_rate": 9.975414512725057e-05,
      "loss": 1.7002,
      "step": 16
    },
    {
      "epoch": 0.1827956989247312,
      "grad_norm": 2.373577117919922,
      "learning_rate": 9.966546331768191e-05,
      "loss": 1.4914,
      "step": 17
    },
    {
      "epoch": 0.1935483870967742,
      "grad_norm": 2.016629695892334,
      "learning_rate": 9.956320346634876e-05,
      "loss": 1.3667,
      "step": 18
    },
    {
      "epoch": 0.20430107526881722,
      "grad_norm": 2.0558924674987793,
      "learning_rate": 9.944739353007344e-05,
      "loss": 1.6329,
      "step": 19
    },
    {
      "epoch": 0.21505376344086022,
      "grad_norm": 2.0576963424682617,
      "learning_rate": 9.931806517013612e-05,
      "loss": 1.7884,
      "step": 20
    },
    {
      "epoch": 0.22580645161290322,
      "grad_norm": 2.1554651260375977,
      "learning_rate": 9.917525374361912e-05,
      "loss": 1.606,
      "step": 21
    },
    {
      "epoch": 0.23655913978494625,
      "grad_norm": 1.869352102279663,
      "learning_rate": 9.901899829374047e-05,
      "loss": 1.5538,
      "step": 22
    },
    {
      "epoch": 0.24731182795698925,
      "grad_norm": 1.7703973054885864,
      "learning_rate": 9.884934153917997e-05,
      "loss": 1.5749,
      "step": 23
    },
    {
      "epoch": 0.25806451612903225,
      "grad_norm": 1.570709228515625,
      "learning_rate": 9.86663298624003e-05,
      "loss": 1.4434,
      "step": 24
    },
    {
      "epoch": 0.26881720430107525,
      "grad_norm": 1.6146509647369385,
      "learning_rate": 9.847001329696653e-05,
      "loss": 1.4106,
      "step": 25
    },
    {
      "epoch": 0.27956989247311825,
      "grad_norm": 1.6668318510055542,
      "learning_rate": 9.826044551386744e-05,
      "loss": 1.5185,
      "step": 26
    },
    {
      "epoch": 0.2903225806451613,
      "grad_norm": 1.7098801136016846,
      "learning_rate": 9.803768380684242e-05,
      "loss": 1.4251,
      "step": 27
    },
    {
      "epoch": 0.3010752688172043,
      "grad_norm": 1.5546528100967407,
      "learning_rate": 9.780178907671789e-05,
      "loss": 1.4933,
      "step": 28
    },
    {
      "epoch": 0.3118279569892473,
      "grad_norm": 1.6766256093978882,
      "learning_rate": 9.755282581475769e-05,
      "loss": 1.5284,
      "step": 29
    },
    {
      "epoch": 0.3225806451612903,
      "grad_norm": 1.4459998607635498,
      "learning_rate": 9.729086208503174e-05,
      "loss": 1.3004,
      "step": 30
    },
    {
      "epoch": 0.3333333333333333,
      "grad_norm": 1.567784309387207,
      "learning_rate": 9.701596950580806e-05,
      "loss": 1.4096,
      "step": 31
    },
    {
      "epoch": 0.34408602150537637,
      "grad_norm": 1.7961149215698242,
      "learning_rate": 9.672822322997305e-05,
      "loss": 1.5655,
      "step": 32
    },
    {
      "epoch": 0.3548387096774194,
      "grad_norm": 1.873795509338379,
      "learning_rate": 9.642770192448536e-05,
      "loss": 1.4933,
      "step": 33
    },
    {
      "epoch": 0.3655913978494624,
      "grad_norm": 1.6129283905029297,
      "learning_rate": 9.611448774886924e-05,
      "loss": 1.5071,
      "step": 34
    },
    {
      "epoch": 0.3763440860215054,
      "grad_norm": 1.649135708808899,
      "learning_rate": 9.578866633275288e-05,
      "loss": 1.3613,
      "step": 35
    },
    {
      "epoch": 0.3870967741935484,
      "grad_norm": 1.5218567848205566,
      "learning_rate": 9.545032675245813e-05,
      "loss": 1.584,
      "step": 36
    },
    {
      "epoch": 0.3978494623655914,
      "grad_norm": 2.0004472732543945,
      "learning_rate": 9.509956150664796e-05,
      "loss": 1.481,
      "step": 37
    },
    {
      "epoch": 0.40860215053763443,
      "grad_norm": 1.8755414485931396,
      "learning_rate": 9.473646649103818e-05,
      "loss": 1.5232,
      "step": 38
    },
    {
      "epoch": 0.41935483870967744,
      "grad_norm": 1.5692195892333984,
      "learning_rate": 9.43611409721806e-05,
      "loss": 1.2805,
      "step": 39
    },
    {
      "epoch": 0.43010752688172044,
      "grad_norm": 1.687671184539795,
      "learning_rate": 9.397368756032445e-05,
      "loss": 1.5781,
      "step": 40
    },
    {
      "epoch": 0.44086021505376344,
      "grad_norm": 1.6162201166152954,
      "learning_rate": 9.357421218136386e-05,
      "loss": 1.6896,
      "step": 41
    },
    {
      "epoch": 0.45161290322580644,
      "grad_norm": 1.6264986991882324,
      "learning_rate": 9.316282404787871e-05,
      "loss": 1.5587,
      "step": 42
    },
    {
      "epoch": 0.46236559139784944,
      "grad_norm": 1.5933151245117188,
      "learning_rate": 9.273963562927695e-05,
      "loss": 1.3579,
      "step": 43
    },
    {
      "epoch": 0.4731182795698925,
      "grad_norm": 1.5279178619384766,
      "learning_rate": 9.230476262104677e-05,
      "loss": 1.426,
      "step": 44
    },
    {
      "epoch": 0.4838709677419355,
      "grad_norm": 1.5955722332000732,
      "learning_rate": 9.185832391312644e-05,
      "loss": 1.6286,
      "step": 45
    },
    {
      "epoch": 0.4946236559139785,
      "grad_norm": 1.3342571258544922,
      "learning_rate": 9.140044155740101e-05,
      "loss": 1.3073,
      "step": 46
    },
    {
      "epoch": 0.5053763440860215,
      "grad_norm": 1.3443530797958374,
      "learning_rate": 9.093124073433463e-05,
      "loss": 1.4065,
      "step": 47
    },
    {
      "epoch": 0.5161290322580645,
      "grad_norm": 1.672761082649231,
      "learning_rate": 9.045084971874738e-05,
      "loss": 1.6361,
      "step": 48
    },
    {
      "epoch": 0.5268817204301075,
      "grad_norm": 1.5579395294189453,
      "learning_rate": 8.995939984474624e-05,
      "loss": 1.4569,
      "step": 49
    },
    {
      "epoch": 0.5376344086021505,
      "grad_norm": 1.3008219003677368,
      "learning_rate": 8.945702546981969e-05,
      "loss": 1.3278,
      "step": 50
    },
    {
      "epoch": 0.5376344086021505,
      "eval_loss": 1.3941900730133057,
      "eval_runtime": 4.7378,
      "eval_samples_per_second": 33.138,
      "eval_steps_per_second": 8.443,
      "step": 50
    },
    {
      "epoch": 0.5483870967741935,
      "grad_norm": 1.3945226669311523,
      "learning_rate": 8.894386393810563e-05,
      "loss": 1.2764,
      "step": 51
    },
    {
      "epoch": 0.5591397849462365,
      "grad_norm": 1.3870302438735962,
      "learning_rate": 8.842005554284296e-05,
      "loss": 1.4109,
      "step": 52
    },
    {
      "epoch": 0.5698924731182796,
      "grad_norm": 1.3363792896270752,
      "learning_rate": 8.788574348801675e-05,
      "loss": 1.3709,
      "step": 53
    },
    {
      "epoch": 0.5806451612903226,
      "grad_norm": 1.4388351440429688,
      "learning_rate": 8.73410738492077e-05,
      "loss": 1.3211,
      "step": 54
    },
    {
      "epoch": 0.5913978494623656,
      "grad_norm": 1.3716356754302979,
      "learning_rate": 8.678619553365659e-05,
      "loss": 1.3606,
      "step": 55
    },
    {
      "epoch": 0.6021505376344086,
      "grad_norm": 1.4401875734329224,
      "learning_rate": 8.622126023955446e-05,
      "loss": 1.4787,
      "step": 56
    },
    {
      "epoch": 0.6129032258064516,
      "grad_norm": 1.4469499588012695,
      "learning_rate": 8.564642241456986e-05,
      "loss": 1.4999,
      "step": 57
    },
    {
      "epoch": 0.6236559139784946,
      "grad_norm": 1.427000641822815,
      "learning_rate": 8.506183921362443e-05,
      "loss": 1.3769,
      "step": 58
    },
    {
      "epoch": 0.6344086021505376,
      "grad_norm": 1.6219151020050049,
      "learning_rate": 8.44676704559283e-05,
      "loss": 1.4022,
      "step": 59
    },
    {
      "epoch": 0.6451612903225806,
      "grad_norm": 1.4375038146972656,
      "learning_rate": 8.386407858128706e-05,
      "loss": 1.3532,
      "step": 60
    },
    {
      "epoch": 0.6559139784946236,
      "grad_norm": 1.4883887767791748,
      "learning_rate": 8.32512286056924e-05,
      "loss": 1.518,
      "step": 61
    },
    {
      "epoch": 0.6666666666666666,
      "grad_norm": 1.5331352949142456,
      "learning_rate": 8.262928807620843e-05,
      "loss": 1.2898,
      "step": 62
    },
    {
      "epoch": 0.6774193548387096,
      "grad_norm": 1.4444113969802856,
      "learning_rate": 8.199842702516583e-05,
      "loss": 1.4499,
      "step": 63
    },
    {
      "epoch": 0.6881720430107527,
      "grad_norm": 1.4317717552185059,
      "learning_rate": 8.135881792367686e-05,
      "loss": 1.6092,
      "step": 64
    },
    {
      "epoch": 0.6989247311827957,
      "grad_norm": 1.9583007097244263,
      "learning_rate": 8.07106356344834e-05,
      "loss": 1.3679,
      "step": 65
    },
    {
      "epoch": 0.7096774193548387,
      "grad_norm": 1.5948585271835327,
      "learning_rate": 8.005405736415126e-05,
      "loss": 1.5299,
      "step": 66
    },
    {
      "epoch": 0.7204301075268817,
      "grad_norm": 1.474169135093689,
      "learning_rate": 7.938926261462366e-05,
      "loss": 1.4273,
      "step": 67
    },
    {
      "epoch": 0.7311827956989247,
      "grad_norm": 1.5605497360229492,
      "learning_rate": 7.871643313414718e-05,
      "loss": 1.4787,
      "step": 68
    },
    {
      "epoch": 0.7419354838709677,
      "grad_norm": 1.6736353635787964,
      "learning_rate": 7.803575286758364e-05,
      "loss": 1.5762,
      "step": 69
    },
    {
      "epoch": 0.7526881720430108,
      "grad_norm": 1.4765825271606445,
      "learning_rate": 7.734740790612136e-05,
      "loss": 1.4682,
      "step": 70
    },
    {
      "epoch": 0.7634408602150538,
      "grad_norm": 1.3116446733474731,
      "learning_rate": 7.66515864363997e-05,
      "loss": 1.3492,
      "step": 71
    },
    {
      "epoch": 0.7741935483870968,
      "grad_norm": 1.3846219778060913,
      "learning_rate": 7.594847868906076e-05,
      "loss": 1.3566,
      "step": 72
    },
    {
      "epoch": 0.7849462365591398,
      "grad_norm": 1.4503177404403687,
      "learning_rate": 7.52382768867422e-05,
      "loss": 1.4169,
      "step": 73
    },
    {
      "epoch": 0.7956989247311828,
      "grad_norm": 1.5589603185653687,
      "learning_rate": 7.452117519152542e-05,
      "loss": 1.3105,
      "step": 74
    },
    {
      "epoch": 0.8064516129032258,
      "grad_norm": 1.4112577438354492,
      "learning_rate": 7.379736965185368e-05,
      "loss": 1.4348,
      "step": 75
    },
    {
      "epoch": 0.8172043010752689,
      "grad_norm": 1.478387713432312,
      "learning_rate": 7.30670581489344e-05,
      "loss": 1.4845,
      "step": 76
    },
    {
      "epoch": 0.8279569892473119,
      "grad_norm": 1.4269083738327026,
      "learning_rate": 7.233044034264034e-05,
      "loss": 1.42,
      "step": 77
    },
    {
      "epoch": 0.8387096774193549,
      "grad_norm": 1.2202898263931274,
      "learning_rate": 7.158771761692464e-05,
      "loss": 1.4413,
      "step": 78
    },
    {
      "epoch": 0.8494623655913979,
      "grad_norm": 1.497951865196228,
      "learning_rate": 7.083909302476453e-05,
      "loss": 1.5208,
      "step": 79
    },
    {
      "epoch": 0.8602150537634409,
      "grad_norm": 1.2094298601150513,
      "learning_rate": 7.008477123264848e-05,
      "loss": 1.3276,
      "step": 80
    },
    {
      "epoch": 0.8709677419354839,
      "grad_norm": 1.1779359579086304,
      "learning_rate": 6.932495846462261e-05,
      "loss": 1.183,
      "step": 81
    },
    {
      "epoch": 0.8817204301075269,
      "grad_norm": 1.2660930156707764,
      "learning_rate": 6.855986244591104e-05,
      "loss": 1.3094,
      "step": 82
    },
    {
      "epoch": 0.8924731182795699,
      "grad_norm": 1.3192017078399658,
      "learning_rate": 6.778969234612584e-05,
      "loss": 1.1841,
      "step": 83
    },
    {
      "epoch": 0.9032258064516129,
      "grad_norm": 1.2833420038223267,
      "learning_rate": 6.701465872208216e-05,
      "loss": 1.4709,
      "step": 84
    },
    {
      "epoch": 0.9139784946236559,
      "grad_norm": 1.2637368440628052,
      "learning_rate": 6.623497346023418e-05,
      "loss": 1.2868,
      "step": 85
    },
    {
      "epoch": 0.9247311827956989,
      "grad_norm": 1.4691823720932007,
      "learning_rate": 6.545084971874738e-05,
      "loss": 1.5011,
      "step": 86
    },
    {
      "epoch": 0.9354838709677419,
      "grad_norm": 1.3752435445785522,
      "learning_rate": 6.466250186922325e-05,
      "loss": 1.5095,
      "step": 87
    },
    {
      "epoch": 0.946236559139785,
      "grad_norm": 1.3410090208053589,
      "learning_rate": 6.387014543809223e-05,
      "loss": 1.4567,
      "step": 88
    },
    {
      "epoch": 0.956989247311828,
      "grad_norm": 1.488720178604126,
      "learning_rate": 6.307399704769099e-05,
      "loss": 1.3362,
      "step": 89
    },
    {
      "epoch": 0.967741935483871,
      "grad_norm": 1.4510327577590942,
      "learning_rate": 6.227427435703997e-05,
      "loss": 1.5497,
      "step": 90
    },
    {
      "epoch": 0.978494623655914,
      "grad_norm": 1.362952470779419,
      "learning_rate": 6.147119600233758e-05,
      "loss": 1.3794,
      "step": 91
    },
    {
      "epoch": 0.989247311827957,
      "grad_norm": 1.3994568586349487,
      "learning_rate": 6.066498153718735e-05,
      "loss": 1.4286,
      "step": 92
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.45283043384552,
      "learning_rate": 5.985585137257401e-05,
      "loss": 1.3081,
      "step": 93
    },
    {
      "epoch": 1.010752688172043,
      "grad_norm": 1.0466877222061157,
      "learning_rate": 5.90440267166055e-05,
      "loss": 1.1259,
      "step": 94
    },
    {
      "epoch": 1.021505376344086,
      "grad_norm": 1.0868937969207764,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 1.1437,
      "step": 95
    },
    {
      "epoch": 1.032258064516129,
      "grad_norm": 1.0157090425491333,
      "learning_rate": 5.74131823855921e-05,
      "loss": 1.0368,
      "step": 96
    },
    {
      "epoch": 1.043010752688172,
      "grad_norm": 0.9955579042434692,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 1.0436,
      "step": 97
    },
    {
      "epoch": 1.053763440860215,
      "grad_norm": 1.1000343561172485,
      "learning_rate": 5.577423184847932e-05,
      "loss": 1.1144,
      "step": 98
    },
    {
      "epoch": 1.064516129032258,
      "grad_norm": 1.1350839138031006,
      "learning_rate": 5.495227651252315e-05,
      "loss": 1.1093,
      "step": 99
    },
    {
      "epoch": 1.075268817204301,
      "grad_norm": 1.1216412782669067,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 1.0636,
      "step": 100
    },
    {
      "epoch": 1.075268817204301,
      "eval_loss": 1.3642938137054443,
      "eval_runtime": 4.7381,
      "eval_samples_per_second": 33.135,
      "eval_steps_per_second": 8.442,
      "step": 100
    },
    {
      "epoch": 1.086021505376344,
      "grad_norm": 0.935893177986145,
      "learning_rate": 5.330452921628497e-05,
      "loss": 1.046,
      "step": 101
    },
    {
      "epoch": 1.096774193548387,
      "grad_norm": 1.0560184717178345,
      "learning_rate": 5.247918773366112e-05,
      "loss": 1.0836,
      "step": 102
    },
    {
      "epoch": 1.10752688172043,
      "grad_norm": 1.1073940992355347,
      "learning_rate": 5.165316846586541e-05,
      "loss": 1.1118,
      "step": 103
    },
    {
      "epoch": 1.118279569892473,
      "grad_norm": 1.1321074962615967,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 1.0707,
      "step": 104
    },
    {
      "epoch": 1.129032258064516,
      "grad_norm": 1.1407638788223267,
      "learning_rate": 5e-05,
      "loss": 0.9936,
      "step": 105
    },
    {
      "epoch": 1.139784946236559,
      "grad_norm": 1.505632758140564,
      "learning_rate": 4.917330276168208e-05,
      "loss": 1.1918,
      "step": 106
    },
    {
      "epoch": 1.1505376344086022,
      "grad_norm": 1.1546612977981567,
      "learning_rate": 4.834683153413459e-05,
      "loss": 1.0016,
      "step": 107
    },
    {
      "epoch": 1.1612903225806452,
      "grad_norm": 1.2602618932724,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 1.059,
      "step": 108
    },
    {
      "epoch": 1.1720430107526882,
      "grad_norm": 1.1792690753936768,
      "learning_rate": 4.669547078371504e-05,
      "loss": 0.9426,
      "step": 109
    },
    {
      "epoch": 1.1827956989247312,
      "grad_norm": 1.314300298690796,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 1.0949,
      "step": 110
    },
    {
      "epoch": 1.1935483870967742,
      "grad_norm": 1.3671072721481323,
      "learning_rate": 4.504772348747687e-05,
      "loss": 1.0288,
      "step": 111
    },
    {
      "epoch": 1.2043010752688172,
      "grad_norm": 1.364042043685913,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 1.0628,
      "step": 112
    },
    {
      "epoch": 1.2150537634408602,
      "grad_norm": 1.4929494857788086,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 1.0423,
      "step": 113
    },
    {
      "epoch": 1.2258064516129032,
      "grad_norm": 1.4592775106430054,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 1.0178,
      "step": 114
    },
    {
      "epoch": 1.2365591397849462,
      "grad_norm": 1.6401259899139404,
      "learning_rate": 4.17702704859633e-05,
      "loss": 1.2031,
      "step": 115
    },
    {
      "epoch": 1.2473118279569892,
      "grad_norm": 1.4892230033874512,
      "learning_rate": 4.095597328339452e-05,
      "loss": 1.1288,
      "step": 116
    },
    {
      "epoch": 1.2580645161290323,
      "grad_norm": 1.740278720855713,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 1.2741,
      "step": 117
    },
    {
      "epoch": 1.2688172043010753,
      "grad_norm": 1.8201625347137451,
      "learning_rate": 3.933501846281267e-05,
      "loss": 1.3326,
      "step": 118
    },
    {
      "epoch": 1.2795698924731183,
      "grad_norm": 1.5671497583389282,
      "learning_rate": 3.852880399766243e-05,
      "loss": 1.0878,
      "step": 119
    },
    {
      "epoch": 1.2903225806451613,
      "grad_norm": 1.2733285427093506,
      "learning_rate": 3.772572564296005e-05,
      "loss": 0.9754,
      "step": 120
    },
    {
      "epoch": 1.3010752688172043,
      "grad_norm": 1.3822678327560425,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 0.8714,
      "step": 121
    },
    {
      "epoch": 1.3118279569892473,
      "grad_norm": 1.6007723808288574,
      "learning_rate": 3.612985456190778e-05,
      "loss": 0.9819,
      "step": 122
    },
    {
      "epoch": 1.3225806451612903,
      "grad_norm": 1.5472182035446167,
      "learning_rate": 3.533749813077677e-05,
      "loss": 0.9527,
      "step": 123
    },
    {
      "epoch": 1.3333333333333333,
      "grad_norm": 1.585017204284668,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 1.1438,
      "step": 124
    },
    {
      "epoch": 1.3440860215053765,
      "grad_norm": 1.5065468549728394,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 1.1306,
      "step": 125
    },
    {
      "epoch": 1.3548387096774195,
      "grad_norm": 1.4847522974014282,
      "learning_rate": 3.298534127791785e-05,
      "loss": 1.0218,
      "step": 126
    },
    {
      "epoch": 1.3655913978494625,
      "grad_norm": 1.5103038549423218,
      "learning_rate": 3.221030765387417e-05,
      "loss": 1.0951,
      "step": 127
    },
    {
      "epoch": 1.3763440860215055,
      "grad_norm": 1.432877540588379,
      "learning_rate": 3.144013755408895e-05,
      "loss": 0.971,
      "step": 128
    },
    {
      "epoch": 1.3870967741935485,
      "grad_norm": 1.4108842611312866,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 1.0357,
      "step": 129
    },
    {
      "epoch": 1.3978494623655915,
      "grad_norm": 1.3557878732681274,
      "learning_rate": 2.991522876735154e-05,
      "loss": 1.0886,
      "step": 130
    },
    {
      "epoch": 1.4086021505376345,
      "grad_norm": 1.496570110321045,
      "learning_rate": 2.916090697523549e-05,
      "loss": 1.1439,
      "step": 131
    },
    {
      "epoch": 1.4193548387096775,
      "grad_norm": 1.5574095249176025,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 1.1395,
      "step": 132
    },
    {
      "epoch": 1.4301075268817205,
      "grad_norm": 1.6914658546447754,
      "learning_rate": 2.766955965735968e-05,
      "loss": 1.1831,
      "step": 133
    },
    {
      "epoch": 1.4408602150537635,
      "grad_norm": 1.6495730876922607,
      "learning_rate": 2.693294185106562e-05,
      "loss": 1.1531,
      "step": 134
    },
    {
      "epoch": 1.4516129032258065,
      "grad_norm": 1.6482576131820679,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 1.2152,
      "step": 135
    },
    {
      "epoch": 1.4623655913978495,
      "grad_norm": 1.3409982919692993,
      "learning_rate": 2.547882480847461e-05,
      "loss": 0.8851,
      "step": 136
    },
    {
      "epoch": 1.4731182795698925,
      "grad_norm": 1.4901268482208252,
      "learning_rate": 2.476172311325783e-05,
      "loss": 1.0615,
      "step": 137
    },
    {
      "epoch": 1.4838709677419355,
      "grad_norm": 1.6849151849746704,
      "learning_rate": 2.405152131093926e-05,
      "loss": 1.1009,
      "step": 138
    },
    {
      "epoch": 1.4946236559139785,
      "grad_norm": 1.4192452430725098,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 0.9832,
      "step": 139
    },
    {
      "epoch": 1.5053763440860215,
      "grad_norm": 1.598957896232605,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 1.1568,
      "step": 140
    },
    {
      "epoch": 1.5161290322580645,
      "grad_norm": 1.4840012788772583,
      "learning_rate": 2.196424713241637e-05,
      "loss": 1.0706,
      "step": 141
    },
    {
      "epoch": 1.5268817204301075,
      "grad_norm": 1.6134722232818604,
      "learning_rate": 2.128356686585282e-05,
      "loss": 1.123,
      "step": 142
    },
    {
      "epoch": 1.5376344086021505,
      "grad_norm": 1.54719877243042,
      "learning_rate": 2.061073738537635e-05,
      "loss": 1.0761,
      "step": 143
    },
    {
      "epoch": 1.5483870967741935,
      "grad_norm": 1.6617050170898438,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 1.2028,
      "step": 144
    },
    {
      "epoch": 1.5591397849462365,
      "grad_norm": 1.5799815654754639,
      "learning_rate": 1.928936436551661e-05,
      "loss": 1.1505,
      "step": 145
    },
    {
      "epoch": 1.5698924731182795,
      "grad_norm": 1.6557565927505493,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 1.1309,
      "step": 146
    },
    {
      "epoch": 1.5806451612903225,
      "grad_norm": 1.5026015043258667,
      "learning_rate": 1.800157297483417e-05,
      "loss": 1.0818,
      "step": 147
    },
    {
      "epoch": 1.5913978494623655,
      "grad_norm": 1.3791894912719727,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 0.9874,
      "step": 148
    },
    {
      "epoch": 1.6021505376344085,
      "grad_norm": 1.4911445379257202,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 1.1183,
      "step": 149
    },
    {
      "epoch": 1.6129032258064515,
      "grad_norm": 1.651066541671753,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 1.1194,
      "step": 150
    },
    {
      "epoch": 1.6129032258064515,
      "eval_loss": 1.3905954360961914,
      "eval_runtime": 4.7674,
      "eval_samples_per_second": 32.932,
      "eval_steps_per_second": 8.39,
      "step": 150
    },
    {
      "epoch": 1.6236559139784945,
      "grad_norm": 1.4255160093307495,
      "learning_rate": 1.553232954407171e-05,
      "loss": 1.0179,
      "step": 151
    },
    {
      "epoch": 1.6344086021505375,
      "grad_norm": 1.4603620767593384,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 0.8123,
      "step": 152
    },
    {
      "epoch": 1.6451612903225805,
      "grad_norm": 1.4084007740020752,
      "learning_rate": 1.435357758543015e-05,
      "loss": 1.129,
      "step": 153
    },
    {
      "epoch": 1.6559139784946235,
      "grad_norm": 1.5454281568527222,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 1.016,
      "step": 154
    },
    {
      "epoch": 1.6666666666666665,
      "grad_norm": 1.525089979171753,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 1.0878,
      "step": 155
    },
    {
      "epoch": 1.6774193548387095,
      "grad_norm": 1.5995702743530273,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 1.1535,
      "step": 156
    },
    {
      "epoch": 1.6881720430107527,
      "grad_norm": 1.5803050994873047,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 1.0537,
      "step": 157
    },
    {
      "epoch": 1.6989247311827957,
      "grad_norm": 1.6103012561798096,
      "learning_rate": 1.157994445715706e-05,
      "loss": 1.1765,
      "step": 158
    },
    {
      "epoch": 1.7096774193548387,
      "grad_norm": 1.8218728303909302,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 1.1703,
      "step": 159
    },
    {
      "epoch": 1.7204301075268817,
      "grad_norm": 1.4936676025390625,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 1.1043,
      "step": 160
    },
    {
      "epoch": 1.7311827956989247,
      "grad_norm": 1.5079119205474854,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 1.019,
      "step": 161
    },
    {
      "epoch": 1.7419354838709677,
      "grad_norm": 1.420454740524292,
      "learning_rate": 9.549150281252633e-06,
      "loss": 1.0531,
      "step": 162
    },
    {
      "epoch": 1.7526881720430108,
      "grad_norm": 1.4512486457824707,
      "learning_rate": 9.068759265665384e-06,
      "loss": 1.0896,
      "step": 163
    },
    {
      "epoch": 1.7634408602150538,
      "grad_norm": 1.605732798576355,
      "learning_rate": 8.599558442598998e-06,
      "loss": 1.0857,
      "step": 164
    },
    {
      "epoch": 1.7741935483870968,
      "grad_norm": 1.371429204940796,
      "learning_rate": 8.141676086873572e-06,
      "loss": 0.905,
      "step": 165
    },
    {
      "epoch": 1.7849462365591398,
      "grad_norm": 1.614424228668213,
      "learning_rate": 7.695237378953223e-06,
      "loss": 1.0574,
      "step": 166
    },
    {
      "epoch": 1.7956989247311828,
      "grad_norm": 1.5107542276382446,
      "learning_rate": 7.260364370723044e-06,
      "loss": 1.0767,
      "step": 167
    },
    {
      "epoch": 1.8064516129032258,
      "grad_norm": 1.4299854040145874,
      "learning_rate": 6.837175952121306e-06,
      "loss": 0.9596,
      "step": 168
    },
    {
      "epoch": 1.817204301075269,
      "grad_norm": 1.6855710744857788,
      "learning_rate": 6.425787818636131e-06,
      "loss": 1.1333,
      "step": 169
    },
    {
      "epoch": 1.827956989247312,
      "grad_norm": 1.5150099992752075,
      "learning_rate": 6.026312439675552e-06,
      "loss": 1.0099,
      "step": 170
    },
    {
      "epoch": 1.838709677419355,
      "grad_norm": 1.5539478063583374,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 1.04,
      "step": 171
    },
    {
      "epoch": 1.849462365591398,
      "grad_norm": 1.502110481262207,
      "learning_rate": 5.263533508961827e-06,
      "loss": 1.0692,
      "step": 172
    },
    {
      "epoch": 1.860215053763441,
      "grad_norm": 1.3206491470336914,
      "learning_rate": 4.900438493352055e-06,
      "loss": 1.0718,
      "step": 173
    },
    {
      "epoch": 1.870967741935484,
      "grad_norm": 1.462929606437683,
      "learning_rate": 4.549673247541875e-06,
      "loss": 0.9734,
      "step": 174
    },
    {
      "epoch": 1.881720430107527,
      "grad_norm": 1.5733832120895386,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 0.9554,
      "step": 175
    },
    {
      "epoch": 1.89247311827957,
      "grad_norm": 1.4133031368255615,
      "learning_rate": 3.885512251130763e-06,
      "loss": 0.9716,
      "step": 176
    },
    {
      "epoch": 1.903225806451613,
      "grad_norm": 1.4608386754989624,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 0.9521,
      "step": 177
    },
    {
      "epoch": 1.913978494623656,
      "grad_norm": 1.5045890808105469,
      "learning_rate": 3.271776770026963e-06,
      "loss": 0.9913,
      "step": 178
    },
    {
      "epoch": 1.924731182795699,
      "grad_norm": 1.577175259590149,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 1.0463,
      "step": 179
    },
    {
      "epoch": 1.935483870967742,
      "grad_norm": 1.683937430381775,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 1.2391,
      "step": 180
    },
    {
      "epoch": 1.946236559139785,
      "grad_norm": 1.6469801664352417,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 1.1313,
      "step": 181
    },
    {
      "epoch": 1.956989247311828,
      "grad_norm": 1.5587536096572876,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 1.0533,
      "step": 182
    },
    {
      "epoch": 1.967741935483871,
      "grad_norm": 1.9218688011169434,
      "learning_rate": 1.962316193157593e-06,
      "loss": 1.2219,
      "step": 183
    },
    {
      "epoch": 1.978494623655914,
      "grad_norm": 1.4584901332855225,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 1.0309,
      "step": 184
    },
    {
      "epoch": 1.989247311827957,
      "grad_norm": 1.6153608560562134,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 1.0525,
      "step": 185
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.873821496963501,
      "learning_rate": 1.333670137599713e-06,
      "loss": 1.1285,
      "step": 186
    },
    {
      "epoch": 2.010752688172043,
      "grad_norm": 1.4225022792816162,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 1.0531,
      "step": 187
    },
    {
      "epoch": 2.021505376344086,
      "grad_norm": 1.3832712173461914,
      "learning_rate": 9.810017062595322e-07,
      "loss": 1.0434,
      "step": 188
    },
    {
      "epoch": 2.032258064516129,
      "grad_norm": 1.2817425727844238,
      "learning_rate": 8.247462563808817e-07,
      "loss": 0.876,
      "step": 189
    },
    {
      "epoch": 2.043010752688172,
      "grad_norm": 1.2483127117156982,
      "learning_rate": 6.819348298638839e-07,
      "loss": 0.8236,
      "step": 190
    },
    {
      "epoch": 2.053763440860215,
      "grad_norm": 1.3870429992675781,
      "learning_rate": 5.526064699265753e-07,
      "loss": 0.909,
      "step": 191
    },
    {
      "epoch": 2.064516129032258,
      "grad_norm": 1.273902416229248,
      "learning_rate": 4.367965336512403e-07,
      "loss": 0.9588,
      "step": 192
    },
    {
      "epoch": 2.075268817204301,
      "grad_norm": 1.3933042287826538,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 0.932,
      "step": 193
    },
    {
      "epoch": 2.086021505376344,
      "grad_norm": 1.327836036682129,
      "learning_rate": 2.458548727494292e-07,
      "loss": 0.833,
      "step": 194
    },
    {
      "epoch": 2.096774193548387,
      "grad_norm": 1.3181606531143188,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 0.9239,
      "step": 195
    },
    {
      "epoch": 2.10752688172043,
      "grad_norm": 1.4119458198547363,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 0.9425,
      "step": 196
    },
    {
      "epoch": 2.118279569892473,
      "grad_norm": 1.3290451765060425,
      "learning_rate": 6.150154258476315e-08,
      "loss": 0.921,
      "step": 197
    },
    {
      "epoch": 2.129032258064516,
      "grad_norm": 1.3426759243011475,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 0.9362,
      "step": 198
    },
    {
      "epoch": 2.139784946236559,
      "grad_norm": 1.3566099405288696,
      "learning_rate": 6.834750376549792e-09,
      "loss": 0.9784,
      "step": 199
    },
    {
      "epoch": 2.150537634408602,
      "grad_norm": 1.3170688152313232,
      "learning_rate": 0.0,
      "loss": 0.9362,
      "step": 200
    },
    {
      "epoch": 2.150537634408602,
      "eval_loss": 1.383666753768921,
      "eval_runtime": 4.7303,
      "eval_samples_per_second": 33.191,
      "eval_steps_per_second": 8.456,
      "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
}