{
  "best_metric": 1.5611152648925781,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.5486968449931413,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0027434842249657062,
      "grad_norm": 0.3434993028640747,
      "learning_rate": 1e-05,
      "loss": 1.0919,
      "step": 1
    },
    {
      "epoch": 0.0027434842249657062,
      "eval_loss": 1.8083019256591797,
      "eval_runtime": 18.2512,
      "eval_samples_per_second": 33.642,
      "eval_steps_per_second": 8.438,
      "step": 1
    },
    {
      "epoch": 0.0054869684499314125,
      "grad_norm": 0.39688390493392944,
      "learning_rate": 2e-05,
      "loss": 1.5126,
      "step": 2
    },
    {
      "epoch": 0.00823045267489712,
      "grad_norm": 0.4337899684906006,
      "learning_rate": 3e-05,
      "loss": 1.5599,
      "step": 3
    },
    {
      "epoch": 0.010973936899862825,
      "grad_norm": 0.4767007827758789,
      "learning_rate": 4e-05,
      "loss": 1.6613,
      "step": 4
    },
    {
      "epoch": 0.013717421124828532,
      "grad_norm": 0.5760968327522278,
      "learning_rate": 5e-05,
      "loss": 1.5296,
      "step": 5
    },
    {
      "epoch": 0.01646090534979424,
      "grad_norm": 0.5192652344703674,
      "learning_rate": 6e-05,
      "loss": 1.5449,
      "step": 6
    },
    {
      "epoch": 0.019204389574759947,
      "grad_norm": 0.5282630920410156,
      "learning_rate": 7e-05,
      "loss": 1.5802,
      "step": 7
    },
    {
      "epoch": 0.02194787379972565,
      "grad_norm": 0.5215024352073669,
      "learning_rate": 8e-05,
      "loss": 1.6075,
      "step": 8
    },
    {
      "epoch": 0.024691358024691357,
      "grad_norm": 0.5709196925163269,
      "learning_rate": 9e-05,
      "loss": 1.6597,
      "step": 9
    },
    {
      "epoch": 0.027434842249657063,
      "grad_norm": 0.5125418901443481,
      "learning_rate": 0.0001,
      "loss": 1.5915,
      "step": 10
    },
    {
      "epoch": 0.03017832647462277,
      "grad_norm": 0.5333966016769409,
      "learning_rate": 9.999316524962345e-05,
      "loss": 1.6155,
      "step": 11
    },
    {
      "epoch": 0.03292181069958848,
      "grad_norm": 0.5041253566741943,
      "learning_rate": 9.997266286704631e-05,
      "loss": 1.5776,
      "step": 12
    },
    {
      "epoch": 0.03566529492455418,
      "grad_norm": 0.49540552496910095,
      "learning_rate": 9.993849845741524e-05,
      "loss": 1.581,
      "step": 13
    },
    {
      "epoch": 0.038408779149519894,
      "grad_norm": 0.5172403454780579,
      "learning_rate": 9.989068136093873e-05,
      "loss": 1.5653,
      "step": 14
    },
    {
      "epoch": 0.0411522633744856,
      "grad_norm": 0.5252892374992371,
      "learning_rate": 9.98292246503335e-05,
      "loss": 1.6227,
      "step": 15
    },
    {
      "epoch": 0.0438957475994513,
      "grad_norm": 0.49713313579559326,
      "learning_rate": 9.975414512725057e-05,
      "loss": 1.5489,
      "step": 16
    },
    {
      "epoch": 0.04663923182441701,
      "grad_norm": 0.5229028463363647,
      "learning_rate": 9.966546331768191e-05,
      "loss": 1.5732,
      "step": 17
    },
    {
      "epoch": 0.04938271604938271,
      "grad_norm": 0.49077802896499634,
      "learning_rate": 9.956320346634876e-05,
      "loss": 1.6113,
      "step": 18
    },
    {
      "epoch": 0.05212620027434842,
      "grad_norm": 0.4910929501056671,
      "learning_rate": 9.944739353007344e-05,
      "loss": 1.5832,
      "step": 19
    },
    {
      "epoch": 0.05486968449931413,
      "grad_norm": 0.4879602789878845,
      "learning_rate": 9.931806517013612e-05,
      "loss": 1.6103,
      "step": 20
    },
    {
      "epoch": 0.05761316872427984,
      "grad_norm": 0.49351829290390015,
      "learning_rate": 9.917525374361912e-05,
      "loss": 1.594,
      "step": 21
    },
    {
      "epoch": 0.06035665294924554,
      "grad_norm": 0.47767430543899536,
      "learning_rate": 9.901899829374047e-05,
      "loss": 1.5463,
      "step": 22
    },
    {
      "epoch": 0.06310013717421124,
      "grad_norm": 0.5049974918365479,
      "learning_rate": 9.884934153917997e-05,
      "loss": 1.6112,
      "step": 23
    },
    {
      "epoch": 0.06584362139917696,
      "grad_norm": 0.5194271802902222,
      "learning_rate": 9.86663298624003e-05,
      "loss": 1.5522,
      "step": 24
    },
    {
      "epoch": 0.06858710562414266,
      "grad_norm": 0.5424639582633972,
      "learning_rate": 9.847001329696653e-05,
      "loss": 1.6358,
      "step": 25
    },
    {
      "epoch": 0.07133058984910837,
      "grad_norm": 0.5647779703140259,
      "learning_rate": 9.826044551386744e-05,
      "loss": 1.6056,
      "step": 26
    },
    {
      "epoch": 0.07407407407407407,
      "grad_norm": 0.5801896452903748,
      "learning_rate": 9.803768380684242e-05,
      "loss": 1.6378,
      "step": 27
    },
    {
      "epoch": 0.07681755829903979,
      "grad_norm": 0.559781551361084,
      "learning_rate": 9.780178907671789e-05,
      "loss": 1.74,
      "step": 28
    },
    {
      "epoch": 0.07956104252400549,
      "grad_norm": 0.5181576013565063,
      "learning_rate": 9.755282581475769e-05,
      "loss": 1.5811,
      "step": 29
    },
    {
      "epoch": 0.0823045267489712,
      "grad_norm": 0.5499027967453003,
      "learning_rate": 9.729086208503174e-05,
      "loss": 1.6515,
      "step": 30
    },
    {
      "epoch": 0.0850480109739369,
      "grad_norm": 0.5976420640945435,
      "learning_rate": 9.701596950580806e-05,
      "loss": 1.6372,
      "step": 31
    },
    {
      "epoch": 0.0877914951989026,
      "grad_norm": 0.5729614496231079,
      "learning_rate": 9.672822322997305e-05,
      "loss": 1.6485,
      "step": 32
    },
    {
      "epoch": 0.09053497942386832,
      "grad_norm": 0.6087205410003662,
      "learning_rate": 9.642770192448536e-05,
      "loss": 1.6546,
      "step": 33
    },
    {
      "epoch": 0.09327846364883402,
      "grad_norm": 0.6267574429512024,
      "learning_rate": 9.611448774886924e-05,
      "loss": 1.6278,
      "step": 34
    },
    {
      "epoch": 0.09602194787379972,
      "grad_norm": 0.6141830682754517,
      "learning_rate": 9.578866633275288e-05,
      "loss": 1.5937,
      "step": 35
    },
    {
      "epoch": 0.09876543209876543,
      "grad_norm": 0.6453271508216858,
      "learning_rate": 9.545032675245813e-05,
      "loss": 1.6413,
      "step": 36
    },
    {
      "epoch": 0.10150891632373114,
      "grad_norm": 0.7297681570053101,
      "learning_rate": 9.509956150664796e-05,
      "loss": 1.708,
      "step": 37
    },
    {
      "epoch": 0.10425240054869685,
      "grad_norm": 0.7394252419471741,
      "learning_rate": 9.473646649103818e-05,
      "loss": 1.6882,
      "step": 38
    },
    {
      "epoch": 0.10699588477366255,
      "grad_norm": 0.8940777778625488,
      "learning_rate": 9.43611409721806e-05,
      "loss": 1.6182,
      "step": 39
    },
    {
      "epoch": 0.10973936899862825,
      "grad_norm": 0.8424510359764099,
      "learning_rate": 9.397368756032445e-05,
      "loss": 1.7334,
      "step": 40
    },
    {
      "epoch": 0.11248285322359397,
      "grad_norm": 0.8478288650512695,
      "learning_rate": 9.357421218136386e-05,
      "loss": 1.7342,
      "step": 41
    },
    {
      "epoch": 0.11522633744855967,
      "grad_norm": 0.8745148777961731,
      "learning_rate": 9.316282404787871e-05,
      "loss": 1.6544,
      "step": 42
    },
    {
      "epoch": 0.11796982167352538,
      "grad_norm": 0.960215151309967,
      "learning_rate": 9.273963562927695e-05,
      "loss": 1.7446,
      "step": 43
    },
    {
      "epoch": 0.12071330589849108,
      "grad_norm": 0.9448610544204712,
      "learning_rate": 9.230476262104677e-05,
      "loss": 1.6234,
      "step": 44
    },
    {
      "epoch": 0.12345679012345678,
      "grad_norm": 1.143613338470459,
      "learning_rate": 9.185832391312644e-05,
      "loss": 1.7175,
      "step": 45
    },
    {
      "epoch": 0.1262002743484225,
      "grad_norm": 1.270920753479004,
      "learning_rate": 9.140044155740101e-05,
      "loss": 1.5262,
      "step": 46
    },
    {
      "epoch": 0.1289437585733882,
      "grad_norm": 1.3764441013336182,
      "learning_rate": 9.093124073433463e-05,
      "loss": 1.604,
      "step": 47
    },
    {
      "epoch": 0.13168724279835392,
      "grad_norm": 1.5630559921264648,
      "learning_rate": 9.045084971874738e-05,
      "loss": 1.5084,
      "step": 48
    },
    {
      "epoch": 0.13443072702331962,
      "grad_norm": 1.8604649305343628,
      "learning_rate": 8.995939984474624e-05,
      "loss": 1.3938,
      "step": 49
    },
    {
      "epoch": 0.13717421124828533,
      "grad_norm": 3.0806515216827393,
      "learning_rate": 8.945702546981969e-05,
      "loss": 1.5744,
      "step": 50
    },
    {
      "epoch": 0.13717421124828533,
      "eval_loss": 1.772631287574768,
      "eval_runtime": 18.2965,
      "eval_samples_per_second": 33.558,
      "eval_steps_per_second": 8.417,
      "step": 50
    },
    {
      "epoch": 0.13991769547325103,
      "grad_norm": 1.3444758653640747,
      "learning_rate": 8.894386393810563e-05,
      "loss": 1.4937,
      "step": 51
    },
    {
      "epoch": 0.14266117969821673,
      "grad_norm": 1.166487693786621,
      "learning_rate": 8.842005554284296e-05,
      "loss": 1.6855,
      "step": 52
    },
    {
      "epoch": 0.14540466392318244,
      "grad_norm": 0.9816533327102661,
      "learning_rate": 8.788574348801675e-05,
      "loss": 1.7511,
      "step": 53
    },
    {
      "epoch": 0.14814814814814814,
      "grad_norm": 0.5922704339027405,
      "learning_rate": 8.73410738492077e-05,
      "loss": 1.6486,
      "step": 54
    },
    {
      "epoch": 0.15089163237311384,
      "grad_norm": 0.4463568925857544,
      "learning_rate": 8.678619553365659e-05,
      "loss": 1.5582,
      "step": 55
    },
    {
      "epoch": 0.15363511659807957,
      "grad_norm": 0.3850046396255493,
      "learning_rate": 8.622126023955446e-05,
      "loss": 1.4897,
      "step": 56
    },
    {
      "epoch": 0.15637860082304528,
      "grad_norm": 0.39507344365119934,
      "learning_rate": 8.564642241456986e-05,
      "loss": 1.5333,
      "step": 57
    },
    {
      "epoch": 0.15912208504801098,
      "grad_norm": 0.3807438015937805,
      "learning_rate": 8.506183921362443e-05,
      "loss": 1.4947,
      "step": 58
    },
    {
      "epoch": 0.16186556927297668,
      "grad_norm": 0.3942439556121826,
      "learning_rate": 8.44676704559283e-05,
      "loss": 1.5431,
      "step": 59
    },
    {
      "epoch": 0.1646090534979424,
      "grad_norm": 0.40099039673805237,
      "learning_rate": 8.386407858128706e-05,
      "loss": 1.5513,
      "step": 60
    },
    {
      "epoch": 0.1673525377229081,
      "grad_norm": 0.4164124131202698,
      "learning_rate": 8.32512286056924e-05,
      "loss": 1.5941,
      "step": 61
    },
    {
      "epoch": 0.1700960219478738,
      "grad_norm": 0.4109591245651245,
      "learning_rate": 8.262928807620843e-05,
      "loss": 1.579,
      "step": 62
    },
    {
      "epoch": 0.1728395061728395,
      "grad_norm": 0.4253193736076355,
      "learning_rate": 8.199842702516583e-05,
      "loss": 1.5839,
      "step": 63
    },
    {
      "epoch": 0.1755829903978052,
      "grad_norm": 0.4173336625099182,
      "learning_rate": 8.135881792367686e-05,
      "loss": 1.5536,
      "step": 64
    },
    {
      "epoch": 0.17832647462277093,
      "grad_norm": 0.4264042377471924,
      "learning_rate": 8.07106356344834e-05,
      "loss": 1.5566,
      "step": 65
    },
    {
      "epoch": 0.18106995884773663,
      "grad_norm": 0.4220297634601593,
      "learning_rate": 8.005405736415126e-05,
      "loss": 1.5098,
      "step": 66
    },
    {
      "epoch": 0.18381344307270234,
      "grad_norm": 0.4218011200428009,
      "learning_rate": 7.938926261462366e-05,
      "loss": 1.5441,
      "step": 67
    },
    {
      "epoch": 0.18655692729766804,
      "grad_norm": 0.45965126156806946,
      "learning_rate": 7.871643313414718e-05,
      "loss": 1.6386,
      "step": 68
    },
    {
      "epoch": 0.18930041152263374,
      "grad_norm": 0.4454924762248993,
      "learning_rate": 7.803575286758364e-05,
      "loss": 1.61,
      "step": 69
    },
    {
      "epoch": 0.19204389574759945,
      "grad_norm": 0.457846462726593,
      "learning_rate": 7.734740790612136e-05,
      "loss": 1.5239,
      "step": 70
    },
    {
      "epoch": 0.19478737997256515,
      "grad_norm": 0.43694180250167847,
      "learning_rate": 7.66515864363997e-05,
      "loss": 1.5791,
      "step": 71
    },
    {
      "epoch": 0.19753086419753085,
      "grad_norm": 0.4700647294521332,
      "learning_rate": 7.594847868906076e-05,
      "loss": 1.513,
      "step": 72
    },
    {
      "epoch": 0.20027434842249658,
      "grad_norm": 0.45643824338912964,
      "learning_rate": 7.52382768867422e-05,
      "loss": 1.5538,
      "step": 73
    },
    {
      "epoch": 0.2030178326474623,
      "grad_norm": 0.48366597294807434,
      "learning_rate": 7.452117519152542e-05,
      "loss": 1.5892,
      "step": 74
    },
    {
      "epoch": 0.205761316872428,
      "grad_norm": 0.48699089884757996,
      "learning_rate": 7.379736965185368e-05,
      "loss": 1.6062,
      "step": 75
    },
    {
      "epoch": 0.2085048010973937,
      "grad_norm": 0.48984402418136597,
      "learning_rate": 7.30670581489344e-05,
      "loss": 1.6072,
      "step": 76
    },
    {
      "epoch": 0.2112482853223594,
      "grad_norm": 0.5233974456787109,
      "learning_rate": 7.233044034264034e-05,
      "loss": 1.6272,
      "step": 77
    },
    {
      "epoch": 0.2139917695473251,
      "grad_norm": 0.537270724773407,
      "learning_rate": 7.158771761692464e-05,
      "loss": 1.614,
      "step": 78
    },
    {
      "epoch": 0.2167352537722908,
      "grad_norm": 0.5062639713287354,
      "learning_rate": 7.083909302476453e-05,
      "loss": 1.5397,
      "step": 79
    },
    {
      "epoch": 0.2194787379972565,
      "grad_norm": 0.5525569915771484,
      "learning_rate": 7.008477123264848e-05,
      "loss": 1.6221,
      "step": 80
    },
    {
      "epoch": 0.2222222222222222,
      "grad_norm": 0.5531335473060608,
      "learning_rate": 6.932495846462261e-05,
      "loss": 1.6083,
      "step": 81
    },
    {
      "epoch": 0.22496570644718794,
      "grad_norm": 0.6107179522514343,
      "learning_rate": 6.855986244591104e-05,
      "loss": 1.6478,
      "step": 82
    },
    {
      "epoch": 0.22770919067215364,
      "grad_norm": 0.5937504172325134,
      "learning_rate": 6.778969234612584e-05,
      "loss": 1.5921,
      "step": 83
    },
    {
      "epoch": 0.23045267489711935,
      "grad_norm": 0.6023827791213989,
      "learning_rate": 6.701465872208216e-05,
      "loss": 1.5433,
      "step": 84
    },
    {
      "epoch": 0.23319615912208505,
      "grad_norm": 0.6272398829460144,
      "learning_rate": 6.623497346023418e-05,
      "loss": 1.5674,
      "step": 85
    },
    {
      "epoch": 0.23593964334705075,
      "grad_norm": 0.6602590084075928,
      "learning_rate": 6.545084971874738e-05,
      "loss": 1.6381,
      "step": 86
    },
    {
      "epoch": 0.23868312757201646,
      "grad_norm": 0.6595104336738586,
      "learning_rate": 6.466250186922325e-05,
      "loss": 1.6106,
      "step": 87
    },
    {
      "epoch": 0.24142661179698216,
      "grad_norm": 0.7493829131126404,
      "learning_rate": 6.387014543809223e-05,
      "loss": 1.6783,
      "step": 88
    },
    {
      "epoch": 0.24417009602194786,
      "grad_norm": 0.7235624194145203,
      "learning_rate": 6.307399704769099e-05,
      "loss": 1.5739,
      "step": 89
    },
    {
      "epoch": 0.24691358024691357,
      "grad_norm": 0.7879060506820679,
      "learning_rate": 6.227427435703997e-05,
      "loss": 1.5853,
      "step": 90
    },
    {
      "epoch": 0.2496570644718793,
      "grad_norm": 0.842487633228302,
      "learning_rate": 6.147119600233758e-05,
      "loss": 1.6513,
      "step": 91
    },
    {
      "epoch": 0.252400548696845,
      "grad_norm": 0.8757365345954895,
      "learning_rate": 6.066498153718735e-05,
      "loss": 1.5479,
      "step": 92
    },
    {
      "epoch": 0.2551440329218107,
      "grad_norm": 0.9465636610984802,
      "learning_rate": 5.985585137257401e-05,
      "loss": 1.5751,
      "step": 93
    },
    {
      "epoch": 0.2578875171467764,
      "grad_norm": 1.0396233797073364,
      "learning_rate": 5.90440267166055e-05,
      "loss": 1.5078,
      "step": 94
    },
    {
      "epoch": 0.2606310013717421,
      "grad_norm": 1.1978845596313477,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 1.533,
      "step": 95
    },
    {
      "epoch": 0.26337448559670784,
      "grad_norm": 1.397278904914856,
      "learning_rate": 5.74131823855921e-05,
      "loss": 1.5851,
      "step": 96
    },
    {
      "epoch": 0.2661179698216735,
      "grad_norm": 1.9425033330917358,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 1.6791,
      "step": 97
    },
    {
      "epoch": 0.26886145404663925,
      "grad_norm": 1.922813057899475,
      "learning_rate": 5.577423184847932e-05,
      "loss": 1.6323,
      "step": 98
    },
    {
      "epoch": 0.2716049382716049,
      "grad_norm": 1.9005879163742065,
      "learning_rate": 5.495227651252315e-05,
      "loss": 1.4932,
      "step": 99
    },
    {
      "epoch": 0.27434842249657065,
      "grad_norm": 2.6369221210479736,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 1.6231,
      "step": 100
    },
    {
      "epoch": 0.27434842249657065,
      "eval_loss": 1.63627290725708,
      "eval_runtime": 18.3031,
      "eval_samples_per_second": 33.546,
      "eval_steps_per_second": 8.414,
      "step": 100
    },
    {
      "epoch": 0.27709190672153633,
      "grad_norm": 0.47432562708854675,
      "learning_rate": 5.330452921628497e-05,
      "loss": 1.2375,
      "step": 101
    },
    {
      "epoch": 0.27983539094650206,
      "grad_norm": 0.5977849364280701,
      "learning_rate": 5.247918773366112e-05,
      "loss": 1.5721,
      "step": 102
    },
    {
      "epoch": 0.2825788751714678,
      "grad_norm": 0.6470625996589661,
      "learning_rate": 5.165316846586541e-05,
      "loss": 1.6391,
      "step": 103
    },
    {
      "epoch": 0.28532235939643347,
      "grad_norm": 0.6095665693283081,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 1.6513,
      "step": 104
    },
    {
      "epoch": 0.2880658436213992,
      "grad_norm": 0.49119675159454346,
      "learning_rate": 5e-05,
      "loss": 1.4832,
      "step": 105
    },
    {
      "epoch": 0.2908093278463649,
      "grad_norm": 0.4163822531700134,
      "learning_rate": 4.917330276168208e-05,
      "loss": 1.4698,
      "step": 106
    },
    {
      "epoch": 0.2935528120713306,
      "grad_norm": 0.41564467549324036,
      "learning_rate": 4.834683153413459e-05,
      "loss": 1.5837,
      "step": 107
    },
    {
      "epoch": 0.2962962962962963,
      "grad_norm": 0.39721089601516724,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 1.5566,
      "step": 108
    },
    {
      "epoch": 0.299039780521262,
      "grad_norm": 0.4076736867427826,
      "learning_rate": 4.669547078371504e-05,
      "loss": 1.6122,
      "step": 109
    },
    {
      "epoch": 0.3017832647462277,
      "grad_norm": 0.4160033166408539,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 1.5648,
      "step": 110
    },
    {
      "epoch": 0.3045267489711934,
      "grad_norm": 0.39523667097091675,
      "learning_rate": 4.504772348747687e-05,
      "loss": 1.5649,
      "step": 111
    },
    {
      "epoch": 0.30727023319615915,
      "grad_norm": 0.3976440727710724,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 1.5404,
      "step": 112
    },
    {
      "epoch": 0.3100137174211248,
      "grad_norm": 0.4029238820075989,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 1.4939,
      "step": 113
    },
    {
      "epoch": 0.31275720164609055,
      "grad_norm": 0.4136357605457306,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 1.5426,
      "step": 114
    },
    {
      "epoch": 0.31550068587105623,
      "grad_norm": 0.4024861752986908,
      "learning_rate": 4.17702704859633e-05,
      "loss": 1.5082,
      "step": 115
    },
    {
      "epoch": 0.31824417009602196,
      "grad_norm": 0.4143868088722229,
      "learning_rate": 4.095597328339452e-05,
      "loss": 1.5333,
      "step": 116
    },
    {
      "epoch": 0.32098765432098764,
      "grad_norm": 0.3947871923446655,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 1.4405,
      "step": 117
    },
    {
      "epoch": 0.32373113854595337,
      "grad_norm": 0.4291187822818756,
      "learning_rate": 3.933501846281267e-05,
      "loss": 1.5684,
      "step": 118
    },
    {
      "epoch": 0.32647462277091904,
      "grad_norm": 0.42030757665634155,
      "learning_rate": 3.852880399766243e-05,
      "loss": 1.4987,
      "step": 119
    },
    {
      "epoch": 0.3292181069958848,
      "grad_norm": 0.4372823238372803,
      "learning_rate": 3.772572564296005e-05,
      "loss": 1.4744,
      "step": 120
    },
    {
      "epoch": 0.3319615912208505,
      "grad_norm": 0.4583459198474884,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 1.5896,
      "step": 121
    },
    {
      "epoch": 0.3347050754458162,
      "grad_norm": 0.49309661984443665,
      "learning_rate": 3.612985456190778e-05,
      "loss": 1.5024,
      "step": 122
    },
    {
      "epoch": 0.3374485596707819,
      "grad_norm": 0.4607110321521759,
      "learning_rate": 3.533749813077677e-05,
      "loss": 1.5847,
      "step": 123
    },
    {
      "epoch": 0.3401920438957476,
      "grad_norm": 0.4729166626930237,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 1.4996,
      "step": 124
    },
    {
      "epoch": 0.3429355281207133,
      "grad_norm": 0.4956277012825012,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 1.583,
      "step": 125
    },
    {
      "epoch": 0.345679012345679,
      "grad_norm": 0.48497462272644043,
      "learning_rate": 3.298534127791785e-05,
      "loss": 1.5984,
      "step": 126
    },
    {
      "epoch": 0.3484224965706447,
      "grad_norm": 0.48401322960853577,
      "learning_rate": 3.221030765387417e-05,
      "loss": 1.5429,
      "step": 127
    },
    {
      "epoch": 0.3511659807956104,
      "grad_norm": 0.5126416683197021,
      "learning_rate": 3.144013755408895e-05,
      "loss": 1.5965,
      "step": 128
    },
    {
      "epoch": 0.35390946502057613,
      "grad_norm": 0.515007495880127,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 1.4782,
      "step": 129
    },
    {
      "epoch": 0.35665294924554186,
      "grad_norm": 0.523879885673523,
      "learning_rate": 2.991522876735154e-05,
      "loss": 1.547,
      "step": 130
    },
    {
      "epoch": 0.35939643347050754,
      "grad_norm": 0.5102810859680176,
      "learning_rate": 2.916090697523549e-05,
      "loss": 1.6025,
      "step": 131
    },
    {
      "epoch": 0.36213991769547327,
      "grad_norm": 0.5761386156082153,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 1.5758,
      "step": 132
    },
    {
      "epoch": 0.36488340192043894,
      "grad_norm": 0.5534322261810303,
      "learning_rate": 2.766955965735968e-05,
      "loss": 1.5701,
      "step": 133
    },
    {
      "epoch": 0.3676268861454047,
      "grad_norm": 0.5841242074966431,
      "learning_rate": 2.693294185106562e-05,
      "loss": 1.7126,
      "step": 134
    },
    {
      "epoch": 0.37037037037037035,
      "grad_norm": 0.6193564534187317,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 1.5908,
      "step": 135
    },
    {
      "epoch": 0.3731138545953361,
      "grad_norm": 0.6218095421791077,
      "learning_rate": 2.547882480847461e-05,
      "loss": 1.6208,
      "step": 136
    },
    {
      "epoch": 0.37585733882030176,
      "grad_norm": 0.6127715110778809,
      "learning_rate": 2.476172311325783e-05,
      "loss": 1.5931,
      "step": 137
    },
    {
      "epoch": 0.3786008230452675,
      "grad_norm": 0.6064656376838684,
      "learning_rate": 2.405152131093926e-05,
      "loss": 1.4807,
      "step": 138
    },
    {
      "epoch": 0.3813443072702332,
      "grad_norm": 0.6779766082763672,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 1.5899,
      "step": 139
    },
    {
      "epoch": 0.3840877914951989,
      "grad_norm": 0.6981008052825928,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 1.5855,
      "step": 140
    },
    {
      "epoch": 0.3868312757201646,
      "grad_norm": 0.7448855042457581,
      "learning_rate": 2.196424713241637e-05,
      "loss": 1.6201,
      "step": 141
    },
    {
      "epoch": 0.3895747599451303,
      "grad_norm": 0.7847548127174377,
      "learning_rate": 2.128356686585282e-05,
      "loss": 1.5484,
      "step": 142
    },
    {
      "epoch": 0.39231824417009603,
      "grad_norm": 0.8773086667060852,
      "learning_rate": 2.061073738537635e-05,
      "loss": 1.5192,
      "step": 143
    },
    {
      "epoch": 0.3950617283950617,
      "grad_norm": 1.0284032821655273,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 1.6372,
      "step": 144
    },
    {
      "epoch": 0.39780521262002744,
      "grad_norm": 1.1878750324249268,
      "learning_rate": 1.928936436551661e-05,
      "loss": 1.5945,
      "step": 145
    },
    {
      "epoch": 0.40054869684499317,
      "grad_norm": 1.2274956703186035,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 1.6392,
      "step": 146
    },
    {
      "epoch": 0.40329218106995884,
      "grad_norm": 1.4721274375915527,
      "learning_rate": 1.800157297483417e-05,
      "loss": 1.6956,
      "step": 147
    },
    {
      "epoch": 0.4060356652949246,
      "grad_norm": 1.5409070253372192,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 1.3511,
      "step": 148
    },
    {
      "epoch": 0.40877914951989025,
      "grad_norm": 2.110781192779541,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 1.4294,
      "step": 149
    },
    {
      "epoch": 0.411522633744856,
      "grad_norm": 3.323831081390381,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 1.7199,
      "step": 150
    },
    {
      "epoch": 0.411522633744856,
      "eval_loss": 1.5677804946899414,
      "eval_runtime": 18.3897,
      "eval_samples_per_second": 33.388,
      "eval_steps_per_second": 8.374,
      "step": 150
    },
    {
      "epoch": 0.41426611796982166,
      "grad_norm": 0.32062894105911255,
      "learning_rate": 1.553232954407171e-05,
      "loss": 1.3738,
      "step": 151
    },
    {
      "epoch": 0.4170096021947874,
      "grad_norm": 0.37252163887023926,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 1.5285,
      "step": 152
    },
    {
      "epoch": 0.41975308641975306,
      "grad_norm": 0.3927411735057831,
      "learning_rate": 1.435357758543015e-05,
      "loss": 1.5806,
      "step": 153
    },
    {
      "epoch": 0.4224965706447188,
      "grad_norm": 0.38448622822761536,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 1.5292,
      "step": 154
    },
    {
      "epoch": 0.4252400548696845,
      "grad_norm": 0.38993939757347107,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 1.5491,
      "step": 155
    },
    {
      "epoch": 0.4279835390946502,
      "grad_norm": 0.3920053243637085,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 1.5022,
      "step": 156
    },
    {
      "epoch": 0.43072702331961593,
      "grad_norm": 0.40088677406311035,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 1.545,
      "step": 157
    },
    {
      "epoch": 0.4334705075445816,
      "grad_norm": 0.41268190741539,
      "learning_rate": 1.157994445715706e-05,
      "loss": 1.5467,
      "step": 158
    },
    {
      "epoch": 0.43621399176954734,
      "grad_norm": 0.41179367899894714,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 1.5797,
      "step": 159
    },
    {
      "epoch": 0.438957475994513,
      "grad_norm": 0.44291889667510986,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 1.6055,
      "step": 160
    },
    {
      "epoch": 0.44170096021947874,
      "grad_norm": 0.4080730378627777,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 1.5202,
      "step": 161
    },
    {
      "epoch": 0.4444444444444444,
      "grad_norm": 0.41179701685905457,
      "learning_rate": 9.549150281252633e-06,
      "loss": 1.4907,
      "step": 162
    },
    {
      "epoch": 0.44718792866941015,
      "grad_norm": 0.411336749792099,
      "learning_rate": 9.068759265665384e-06,
      "loss": 1.5045,
      "step": 163
    },
    {
      "epoch": 0.4499314128943759,
      "grad_norm": 0.43752506375312805,
      "learning_rate": 8.599558442598998e-06,
      "loss": 1.5745,
      "step": 164
    },
    {
      "epoch": 0.45267489711934156,
      "grad_norm": 0.4347054064273834,
      "learning_rate": 8.141676086873572e-06,
      "loss": 1.5239,
      "step": 165
    },
    {
      "epoch": 0.4554183813443073,
      "grad_norm": 0.41621142625808716,
      "learning_rate": 7.695237378953223e-06,
      "loss": 1.4387,
      "step": 166
    },
    {
      "epoch": 0.45816186556927296,
      "grad_norm": 0.4325125217437744,
      "learning_rate": 7.260364370723044e-06,
      "loss": 1.4959,
      "step": 167
    },
    {
      "epoch": 0.4609053497942387,
      "grad_norm": 0.4297638535499573,
      "learning_rate": 6.837175952121306e-06,
      "loss": 1.4761,
      "step": 168
    },
    {
      "epoch": 0.46364883401920437,
      "grad_norm": 0.45093488693237305,
      "learning_rate": 6.425787818636131e-06,
      "loss": 1.5184,
      "step": 169
    },
    {
      "epoch": 0.4663923182441701,
      "grad_norm": 0.44806966185569763,
      "learning_rate": 6.026312439675552e-06,
      "loss": 1.5341,
      "step": 170
    },
    {
      "epoch": 0.4691358024691358,
      "grad_norm": 0.4376004636287689,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 1.4844,
      "step": 171
    },
    {
      "epoch": 0.4718792866941015,
      "grad_norm": 0.46486717462539673,
      "learning_rate": 5.263533508961827e-06,
      "loss": 1.5246,
      "step": 172
    },
    {
      "epoch": 0.47462277091906724,
      "grad_norm": 0.48885199427604675,
      "learning_rate": 4.900438493352055e-06,
      "loss": 1.543,
      "step": 173
    },
    {
      "epoch": 0.4773662551440329,
      "grad_norm": 0.4764540493488312,
      "learning_rate": 4.549673247541875e-06,
      "loss": 1.5436,
      "step": 174
    },
    {
      "epoch": 0.48010973936899864,
      "grad_norm": 0.4586107134819031,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 1.5771,
      "step": 175
    },
    {
      "epoch": 0.4828532235939643,
      "grad_norm": 0.5206536054611206,
      "learning_rate": 3.885512251130763e-06,
      "loss": 1.566,
      "step": 176
    },
    {
      "epoch": 0.48559670781893005,
      "grad_norm": 0.492930144071579,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 1.5436,
      "step": 177
    },
    {
      "epoch": 0.4883401920438957,
      "grad_norm": 0.5178830027580261,
      "learning_rate": 3.271776770026963e-06,
      "loss": 1.5055,
      "step": 178
    },
    {
      "epoch": 0.49108367626886146,
      "grad_norm": 0.4968222677707672,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 1.5659,
      "step": 179
    },
    {
      "epoch": 0.49382716049382713,
      "grad_norm": 0.5313413143157959,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 1.5679,
      "step": 180
    },
    {
      "epoch": 0.49657064471879286,
      "grad_norm": 0.5328629016876221,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 1.5731,
      "step": 181
    },
    {
      "epoch": 0.4993141289437586,
      "grad_norm": 0.5572908520698547,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 1.5925,
      "step": 182
    },
    {
      "epoch": 0.5020576131687243,
      "grad_norm": 0.5933302640914917,
      "learning_rate": 1.962316193157593e-06,
      "loss": 1.6273,
      "step": 183
    },
    {
      "epoch": 0.50480109739369,
      "grad_norm": 0.6341251134872437,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 1.6401,
      "step": 184
    },
    {
      "epoch": 0.5075445816186557,
      "grad_norm": 0.609872043132782,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 1.5601,
      "step": 185
    },
    {
      "epoch": 0.5102880658436214,
      "grad_norm": 0.6693766713142395,
      "learning_rate": 1.333670137599713e-06,
      "loss": 1.5957,
      "step": 186
    },
    {
      "epoch": 0.5130315500685871,
      "grad_norm": 0.6638064384460449,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 1.6515,
      "step": 187
    },
    {
      "epoch": 0.5157750342935528,
      "grad_norm": 0.7410558462142944,
      "learning_rate": 9.810017062595322e-07,
      "loss": 1.634,
      "step": 188
    },
    {
      "epoch": 0.5185185185185185,
      "grad_norm": 0.8260947465896606,
      "learning_rate": 8.247462563808817e-07,
      "loss": 1.7651,
      "step": 189
    },
    {
      "epoch": 0.5212620027434842,
      "grad_norm": 0.7096025943756104,
      "learning_rate": 6.819348298638839e-07,
      "loss": 1.467,
      "step": 190
    },
    {
      "epoch": 0.52400548696845,
      "grad_norm": 0.8358923196792603,
      "learning_rate": 5.526064699265753e-07,
      "loss": 1.5747,
      "step": 191
    },
    {
      "epoch": 0.5267489711934157,
      "grad_norm": 0.8389124870300293,
      "learning_rate": 4.367965336512403e-07,
      "loss": 1.7134,
      "step": 192
    },
    {
      "epoch": 0.5294924554183813,
      "grad_norm": 0.978324830532074,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 1.6616,
      "step": 193
    },
    {
      "epoch": 0.532235939643347,
      "grad_norm": 0.9809422492980957,
      "learning_rate": 2.458548727494292e-07,
      "loss": 1.6386,
      "step": 194
    },
    {
      "epoch": 0.5349794238683128,
      "grad_norm": 0.993825376033783,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 1.5415,
      "step": 195
    },
    {
      "epoch": 0.5377229080932785,
      "grad_norm": 1.0987650156021118,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 1.4764,
      "step": 196
    },
    {
      "epoch": 0.5404663923182441,
      "grad_norm": 1.3133987188339233,
      "learning_rate": 6.150154258476315e-08,
      "loss": 1.5953,
      "step": 197
    },
    {
      "epoch": 0.5432098765432098,
      "grad_norm": 1.3211586475372314,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 1.4351,
      "step": 198
    },
    {
      "epoch": 0.5459533607681756,
      "grad_norm": 1.5410858392715454,
      "learning_rate": 6.834750376549792e-09,
      "loss": 1.2921,
      "step": 199
    },
    {
      "epoch": 0.5486968449931413,
      "grad_norm": 2.7132296562194824,
      "learning_rate": 0.0,
      "loss": 1.532,
      "step": 200
    },
    {
      "epoch": 0.5486968449931413,
      "eval_loss": 1.5611152648925781,
      "eval_runtime": 18.3186,
      "eval_samples_per_second": 33.518,
      "eval_steps_per_second": 8.407,
      "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": 5.463305468510208e+16,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}