{
  "best_metric": 0.0965760126709938,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 2.727891156462585,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.013605442176870748,
      "grad_norm": 0.38160058856010437,
      "learning_rate": 1e-05,
      "loss": 0.248,
      "step": 1
    },
    {
      "epoch": 0.013605442176870748,
      "eval_loss": 0.34779831767082214,
      "eval_runtime": 17.1404,
      "eval_samples_per_second": 7.234,
      "eval_steps_per_second": 1.809,
      "step": 1
    },
    {
      "epoch": 0.027210884353741496,
      "grad_norm": 0.42016950249671936,
      "learning_rate": 2e-05,
      "loss": 0.2666,
      "step": 2
    },
    {
      "epoch": 0.04081632653061224,
      "grad_norm": 0.41273173689842224,
      "learning_rate": 3e-05,
      "loss": 0.2279,
      "step": 3
    },
    {
      "epoch": 0.05442176870748299,
      "grad_norm": 0.3237794041633606,
      "learning_rate": 4e-05,
      "loss": 0.3293,
      "step": 4
    },
    {
      "epoch": 0.06802721088435375,
      "grad_norm": 0.2477637529373169,
      "learning_rate": 5e-05,
      "loss": 0.245,
      "step": 5
    },
    {
      "epoch": 0.08163265306122448,
      "grad_norm": 0.33046668767929077,
      "learning_rate": 6e-05,
      "loss": 0.2308,
      "step": 6
    },
    {
      "epoch": 0.09523809523809523,
      "grad_norm": 0.46886196732521057,
      "learning_rate": 7e-05,
      "loss": 0.2475,
      "step": 7
    },
    {
      "epoch": 0.10884353741496598,
      "grad_norm": 0.54930579662323,
      "learning_rate": 8e-05,
      "loss": 0.302,
      "step": 8
    },
    {
      "epoch": 0.12244897959183673,
      "grad_norm": 0.3451319932937622,
      "learning_rate": 9e-05,
      "loss": 0.2401,
      "step": 9
    },
    {
      "epoch": 0.1360544217687075,
      "grad_norm": 0.14674012362957,
      "learning_rate": 0.0001,
      "loss": 0.1967,
      "step": 10
    },
    {
      "epoch": 0.14965986394557823,
      "grad_norm": 0.3813375234603882,
      "learning_rate": 9.999316524962345e-05,
      "loss": 0.2254,
      "step": 11
    },
    {
      "epoch": 0.16326530612244897,
      "grad_norm": 0.39614465832710266,
      "learning_rate": 9.997266286704631e-05,
      "loss": 0.302,
      "step": 12
    },
    {
      "epoch": 0.17687074829931973,
      "grad_norm": 0.2938721776008606,
      "learning_rate": 9.993849845741524e-05,
      "loss": 0.1603,
      "step": 13
    },
    {
      "epoch": 0.19047619047619047,
      "grad_norm": 0.15735051035881042,
      "learning_rate": 9.989068136093873e-05,
      "loss": 0.1359,
      "step": 14
    },
    {
      "epoch": 0.20408163265306123,
      "grad_norm": 0.2035001814365387,
      "learning_rate": 9.98292246503335e-05,
      "loss": 0.1623,
      "step": 15
    },
    {
      "epoch": 0.21768707482993196,
      "grad_norm": 0.1951669156551361,
      "learning_rate": 9.975414512725057e-05,
      "loss": 0.1964,
      "step": 16
    },
    {
      "epoch": 0.23129251700680273,
      "grad_norm": 0.17128324508666992,
      "learning_rate": 9.966546331768191e-05,
      "loss": 0.1539,
      "step": 17
    },
    {
      "epoch": 0.24489795918367346,
      "grad_norm": 0.26221445202827454,
      "learning_rate": 9.956320346634876e-05,
      "loss": 0.2932,
      "step": 18
    },
    {
      "epoch": 0.2585034013605442,
      "grad_norm": 0.12474635988473892,
      "learning_rate": 9.944739353007344e-05,
      "loss": 0.1369,
      "step": 19
    },
    {
      "epoch": 0.272108843537415,
      "grad_norm": 0.10079390555620193,
      "learning_rate": 9.931806517013612e-05,
      "loss": 0.159,
      "step": 20
    },
    {
      "epoch": 0.2857142857142857,
      "grad_norm": 0.10520423948764801,
      "learning_rate": 9.917525374361912e-05,
      "loss": 0.1462,
      "step": 21
    },
    {
      "epoch": 0.29931972789115646,
      "grad_norm": 0.10551194846630096,
      "learning_rate": 9.901899829374047e-05,
      "loss": 0.1223,
      "step": 22
    },
    {
      "epoch": 0.3129251700680272,
      "grad_norm": 0.13030225038528442,
      "learning_rate": 9.884934153917997e-05,
      "loss": 0.1256,
      "step": 23
    },
    {
      "epoch": 0.32653061224489793,
      "grad_norm": 0.12631310522556305,
      "learning_rate": 9.86663298624003e-05,
      "loss": 0.1243,
      "step": 24
    },
    {
      "epoch": 0.3401360544217687,
      "grad_norm": 0.12957267463207245,
      "learning_rate": 9.847001329696653e-05,
      "loss": 0.1343,
      "step": 25
    },
    {
      "epoch": 0.35374149659863946,
      "grad_norm": 0.21590958535671234,
      "learning_rate": 9.826044551386744e-05,
      "loss": 0.1651,
      "step": 26
    },
    {
      "epoch": 0.3673469387755102,
      "grad_norm": 0.09845131635665894,
      "learning_rate": 9.803768380684242e-05,
      "loss": 0.1025,
      "step": 27
    },
    {
      "epoch": 0.38095238095238093,
      "grad_norm": 0.29299595952033997,
      "learning_rate": 9.780178907671789e-05,
      "loss": 0.1128,
      "step": 28
    },
    {
      "epoch": 0.3945578231292517,
      "grad_norm": 0.13073989748954773,
      "learning_rate": 9.755282581475769e-05,
      "loss": 0.1336,
      "step": 29
    },
    {
      "epoch": 0.40816326530612246,
      "grad_norm": 0.29173561930656433,
      "learning_rate": 9.729086208503174e-05,
      "loss": 0.1152,
      "step": 30
    },
    {
      "epoch": 0.4217687074829932,
      "grad_norm": 0.1853582262992859,
      "learning_rate": 9.701596950580806e-05,
      "loss": 0.1797,
      "step": 31
    },
    {
      "epoch": 0.43537414965986393,
      "grad_norm": 0.4235008955001831,
      "learning_rate": 9.672822322997305e-05,
      "loss": 0.0756,
      "step": 32
    },
    {
      "epoch": 0.4489795918367347,
      "grad_norm": 0.12367819249629974,
      "learning_rate": 9.642770192448536e-05,
      "loss": 0.1862,
      "step": 33
    },
    {
      "epoch": 0.46258503401360546,
      "grad_norm": 0.4744042456150055,
      "learning_rate": 9.611448774886924e-05,
      "loss": 0.0931,
      "step": 34
    },
    {
      "epoch": 0.47619047619047616,
      "grad_norm": 0.12108553200960159,
      "learning_rate": 9.578866633275288e-05,
      "loss": 0.0951,
      "step": 35
    },
    {
      "epoch": 0.4897959183673469,
      "grad_norm": 0.40535086393356323,
      "learning_rate": 9.545032675245813e-05,
      "loss": 0.1674,
      "step": 36
    },
    {
      "epoch": 0.5034013605442177,
      "grad_norm": 0.12884244322776794,
      "learning_rate": 9.509956150664796e-05,
      "loss": 0.176,
      "step": 37
    },
    {
      "epoch": 0.5170068027210885,
      "grad_norm": 0.23265424370765686,
      "learning_rate": 9.473646649103818e-05,
      "loss": 0.1912,
      "step": 38
    },
    {
      "epoch": 0.5306122448979592,
      "grad_norm": 0.17942707240581512,
      "learning_rate": 9.43611409721806e-05,
      "loss": 0.0725,
      "step": 39
    },
    {
      "epoch": 0.54421768707483,
      "grad_norm": 0.12855759263038635,
      "learning_rate": 9.397368756032445e-05,
      "loss": 0.093,
      "step": 40
    },
    {
      "epoch": 0.5578231292517006,
      "grad_norm": 0.11313574016094208,
      "learning_rate": 9.357421218136386e-05,
      "loss": 0.1004,
      "step": 41
    },
    {
      "epoch": 0.5714285714285714,
      "grad_norm": 0.2591482102870941,
      "learning_rate": 9.316282404787871e-05,
      "loss": 0.0795,
      "step": 42
    },
    {
      "epoch": 0.5850340136054422,
      "grad_norm": 0.16203844547271729,
      "learning_rate": 9.273963562927695e-05,
      "loss": 0.0608,
      "step": 43
    },
    {
      "epoch": 0.5986394557823129,
      "grad_norm": 0.24320553243160248,
      "learning_rate": 9.230476262104677e-05,
      "loss": 0.0861,
      "step": 44
    },
    {
      "epoch": 0.6122448979591837,
      "grad_norm": 0.14639556407928467,
      "learning_rate": 9.185832391312644e-05,
      "loss": 0.0966,
      "step": 45
    },
    {
      "epoch": 0.6258503401360545,
      "grad_norm": 0.3061509430408478,
      "learning_rate": 9.140044155740101e-05,
      "loss": 0.1479,
      "step": 46
    },
    {
      "epoch": 0.6394557823129252,
      "grad_norm": 0.12761077284812927,
      "learning_rate": 9.093124073433463e-05,
      "loss": 0.0717,
      "step": 47
    },
    {
      "epoch": 0.6530612244897959,
      "grad_norm": 0.4294169247150421,
      "learning_rate": 9.045084971874738e-05,
      "loss": 0.0923,
      "step": 48
    },
    {
      "epoch": 0.6666666666666666,
      "grad_norm": 0.2987173795700073,
      "learning_rate": 8.995939984474624e-05,
      "loss": 0.0822,
      "step": 49
    },
    {
      "epoch": 0.6802721088435374,
      "grad_norm": 0.3335931599140167,
      "learning_rate": 8.945702546981969e-05,
      "loss": 0.101,
      "step": 50
    },
    {
      "epoch": 0.6802721088435374,
      "eval_loss": 0.11009099334478378,
      "eval_runtime": 17.4614,
      "eval_samples_per_second": 7.101,
      "eval_steps_per_second": 1.775,
      "step": 50
    },
    {
      "epoch": 0.6938775510204082,
      "grad_norm": 0.4437973201274872,
      "learning_rate": 8.894386393810563e-05,
      "loss": 0.0727,
      "step": 51
    },
    {
      "epoch": 0.7074829931972789,
      "grad_norm": 0.40452876687049866,
      "learning_rate": 8.842005554284296e-05,
      "loss": 0.1474,
      "step": 52
    },
    {
      "epoch": 0.7210884353741497,
      "grad_norm": 1.0304673910140991,
      "learning_rate": 8.788574348801675e-05,
      "loss": 0.0964,
      "step": 53
    },
    {
      "epoch": 0.7346938775510204,
      "grad_norm": 0.20635385811328888,
      "learning_rate": 8.73410738492077e-05,
      "loss": 0.0952,
      "step": 54
    },
    {
      "epoch": 0.7482993197278912,
      "grad_norm": 0.5117630362510681,
      "learning_rate": 8.678619553365659e-05,
      "loss": 0.1705,
      "step": 55
    },
    {
      "epoch": 0.7619047619047619,
      "grad_norm": 0.20666487514972687,
      "learning_rate": 8.622126023955446e-05,
      "loss": 0.0887,
      "step": 56
    },
    {
      "epoch": 0.7755102040816326,
      "grad_norm": 0.3259621262550354,
      "learning_rate": 8.564642241456986e-05,
      "loss": 0.1556,
      "step": 57
    },
    {
      "epoch": 0.7891156462585034,
      "grad_norm": 0.3114393949508667,
      "learning_rate": 8.506183921362443e-05,
      "loss": 0.105,
      "step": 58
    },
    {
      "epoch": 0.8027210884353742,
      "grad_norm": 0.09949564188718796,
      "learning_rate": 8.44676704559283e-05,
      "loss": 0.1817,
      "step": 59
    },
    {
      "epoch": 0.8163265306122449,
      "grad_norm": 0.30416229367256165,
      "learning_rate": 8.386407858128706e-05,
      "loss": 0.127,
      "step": 60
    },
    {
      "epoch": 0.8299319727891157,
      "grad_norm": 0.09363894909620285,
      "learning_rate": 8.32512286056924e-05,
      "loss": 0.0547,
      "step": 61
    },
    {
      "epoch": 0.8435374149659864,
      "grad_norm": 0.14224383234977722,
      "learning_rate": 8.262928807620843e-05,
      "loss": 0.1314,
      "step": 62
    },
    {
      "epoch": 0.8571428571428571,
      "grad_norm": 0.14530447125434875,
      "learning_rate": 8.199842702516583e-05,
      "loss": 0.1005,
      "step": 63
    },
    {
      "epoch": 0.8707482993197279,
      "grad_norm": 0.09581568092107773,
      "learning_rate": 8.135881792367686e-05,
      "loss": 0.0586,
      "step": 64
    },
    {
      "epoch": 0.8843537414965986,
      "grad_norm": 0.11573351174592972,
      "learning_rate": 8.07106356344834e-05,
      "loss": 0.0759,
      "step": 65
    },
    {
      "epoch": 0.8979591836734694,
      "grad_norm": 0.2774716913700104,
      "learning_rate": 8.005405736415126e-05,
      "loss": 0.1155,
      "step": 66
    },
    {
      "epoch": 0.9115646258503401,
      "grad_norm": 0.1492365598678589,
      "learning_rate": 7.938926261462366e-05,
      "loss": 0.0804,
      "step": 67
    },
    {
      "epoch": 0.9251700680272109,
      "grad_norm": 0.33371686935424805,
      "learning_rate": 7.871643313414718e-05,
      "loss": 0.0591,
      "step": 68
    },
    {
      "epoch": 0.9387755102040817,
      "grad_norm": 0.11553733050823212,
      "learning_rate": 7.803575286758364e-05,
      "loss": 0.1288,
      "step": 69
    },
    {
      "epoch": 0.9523809523809523,
      "grad_norm": 0.8253076076507568,
      "learning_rate": 7.734740790612136e-05,
      "loss": 0.0456,
      "step": 70
    },
    {
      "epoch": 0.9659863945578231,
      "grad_norm": 0.7074836492538452,
      "learning_rate": 7.66515864363997e-05,
      "loss": 0.0674,
      "step": 71
    },
    {
      "epoch": 0.9795918367346939,
      "grad_norm": 0.5858570337295532,
      "learning_rate": 7.594847868906076e-05,
      "loss": 0.0868,
      "step": 72
    },
    {
      "epoch": 0.9931972789115646,
      "grad_norm": 0.6751317381858826,
      "learning_rate": 7.52382768867422e-05,
      "loss": 0.1258,
      "step": 73
    },
    {
      "epoch": 1.010204081632653,
      "grad_norm": 1.2607001066207886,
      "learning_rate": 7.452117519152542e-05,
      "loss": 0.1321,
      "step": 74
    },
    {
      "epoch": 1.0238095238095237,
      "grad_norm": 0.8847551345825195,
      "learning_rate": 7.379736965185368e-05,
      "loss": 0.1178,
      "step": 75
    },
    {
      "epoch": 1.0374149659863945,
      "grad_norm": 1.0244680643081665,
      "learning_rate": 7.30670581489344e-05,
      "loss": 0.1095,
      "step": 76
    },
    {
      "epoch": 1.0510204081632653,
      "grad_norm": 0.09410584717988968,
      "learning_rate": 7.233044034264034e-05,
      "loss": 0.1332,
      "step": 77
    },
    {
      "epoch": 1.064625850340136,
      "grad_norm": 0.8557324409484863,
      "learning_rate": 7.158771761692464e-05,
      "loss": 0.0588,
      "step": 78
    },
    {
      "epoch": 1.0782312925170068,
      "grad_norm": 0.61385178565979,
      "learning_rate": 7.083909302476453e-05,
      "loss": 0.0968,
      "step": 79
    },
    {
      "epoch": 1.0918367346938775,
      "grad_norm": 0.25118139386177063,
      "learning_rate": 7.008477123264848e-05,
      "loss": 0.0515,
      "step": 80
    },
    {
      "epoch": 1.1054421768707483,
      "grad_norm": 0.42364707589149475,
      "learning_rate": 6.932495846462261e-05,
      "loss": 0.1369,
      "step": 81
    },
    {
      "epoch": 1.119047619047619,
      "grad_norm": 0.3666705787181854,
      "learning_rate": 6.855986244591104e-05,
      "loss": 0.0426,
      "step": 82
    },
    {
      "epoch": 1.1326530612244898,
      "grad_norm": 0.15041208267211914,
      "learning_rate": 6.778969234612584e-05,
      "loss": 0.1212,
      "step": 83
    },
    {
      "epoch": 1.1462585034013606,
      "grad_norm": 0.25074779987335205,
      "learning_rate": 6.701465872208216e-05,
      "loss": 0.1141,
      "step": 84
    },
    {
      "epoch": 1.1598639455782314,
      "grad_norm": 0.2678222954273224,
      "learning_rate": 6.623497346023418e-05,
      "loss": 0.0809,
      "step": 85
    },
    {
      "epoch": 1.1734693877551021,
      "grad_norm": 0.3604568541049957,
      "learning_rate": 6.545084971874738e-05,
      "loss": 0.0763,
      "step": 86
    },
    {
      "epoch": 1.1870748299319729,
      "grad_norm": 0.11493130773305893,
      "learning_rate": 6.466250186922325e-05,
      "loss": 0.0499,
      "step": 87
    },
    {
      "epoch": 1.2006802721088436,
      "grad_norm": 0.5648812651634216,
      "learning_rate": 6.387014543809223e-05,
      "loss": 0.1179,
      "step": 88
    },
    {
      "epoch": 1.2142857142857142,
      "grad_norm": 0.18333379924297333,
      "learning_rate": 6.307399704769099e-05,
      "loss": 0.0693,
      "step": 89
    },
    {
      "epoch": 1.227891156462585,
      "grad_norm": 0.17602325975894928,
      "learning_rate": 6.227427435703997e-05,
      "loss": 0.0393,
      "step": 90
    },
    {
      "epoch": 1.2414965986394557,
      "grad_norm": 0.44050687551498413,
      "learning_rate": 6.147119600233758e-05,
      "loss": 0.0715,
      "step": 91
    },
    {
      "epoch": 1.2551020408163265,
      "grad_norm": 0.11689498275518417,
      "learning_rate": 6.066498153718735e-05,
      "loss": 0.0801,
      "step": 92
    },
    {
      "epoch": 1.2687074829931972,
      "grad_norm": 0.10933611541986465,
      "learning_rate": 5.985585137257401e-05,
      "loss": 0.1041,
      "step": 93
    },
    {
      "epoch": 1.282312925170068,
      "grad_norm": 0.12335681170225143,
      "learning_rate": 5.90440267166055e-05,
      "loss": 0.0722,
      "step": 94
    },
    {
      "epoch": 1.2959183673469388,
      "grad_norm": 0.24435262382030487,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 0.06,
      "step": 95
    },
    {
      "epoch": 1.3095238095238095,
      "grad_norm": 0.12334996461868286,
      "learning_rate": 5.74131823855921e-05,
      "loss": 0.062,
      "step": 96
    },
    {
      "epoch": 1.3231292517006803,
      "grad_norm": 0.1946018934249878,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 0.1155,
      "step": 97
    },
    {
      "epoch": 1.336734693877551,
      "grad_norm": 0.16367290914058685,
      "learning_rate": 5.577423184847932e-05,
      "loss": 0.0444,
      "step": 98
    },
    {
      "epoch": 1.3503401360544218,
      "grad_norm": 0.22132612764835358,
      "learning_rate": 5.495227651252315e-05,
      "loss": 0.0869,
      "step": 99
    },
    {
      "epoch": 1.3639455782312926,
      "grad_norm": 0.24844536185264587,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 0.063,
      "step": 100
    },
    {
      "epoch": 1.3639455782312926,
      "eval_loss": 0.09808482229709625,
      "eval_runtime": 17.4676,
      "eval_samples_per_second": 7.099,
      "eval_steps_per_second": 1.775,
      "step": 100
    },
    {
      "epoch": 1.3775510204081631,
      "grad_norm": 0.16212226450443268,
      "learning_rate": 5.330452921628497e-05,
      "loss": 0.0869,
      "step": 101
    },
    {
      "epoch": 1.391156462585034,
      "grad_norm": 0.29015618562698364,
      "learning_rate": 5.247918773366112e-05,
      "loss": 0.1069,
      "step": 102
    },
    {
      "epoch": 1.4047619047619047,
      "grad_norm": 0.31713011860847473,
      "learning_rate": 5.165316846586541e-05,
      "loss": 0.0594,
      "step": 103
    },
    {
      "epoch": 1.4183673469387754,
      "grad_norm": 0.3650248944759369,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 0.072,
      "step": 104
    },
    {
      "epoch": 1.4319727891156462,
      "grad_norm": 0.22428473830223083,
      "learning_rate": 5e-05,
      "loss": 0.0748,
      "step": 105
    },
    {
      "epoch": 1.445578231292517,
      "grad_norm": 0.08111453801393509,
      "learning_rate": 4.917330276168208e-05,
      "loss": 0.0392,
      "step": 106
    },
    {
      "epoch": 1.4591836734693877,
      "grad_norm": 0.18333624303340912,
      "learning_rate": 4.834683153413459e-05,
      "loss": 0.0386,
      "step": 107
    },
    {
      "epoch": 1.4727891156462585,
      "grad_norm": 0.1260935366153717,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 0.0597,
      "step": 108
    },
    {
      "epoch": 1.4863945578231292,
      "grad_norm": 0.13858090341091156,
      "learning_rate": 4.669547078371504e-05,
      "loss": 0.0361,
      "step": 109
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.13214397430419922,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 0.1367,
      "step": 110
    },
    {
      "epoch": 1.5136054421768708,
      "grad_norm": 0.2434731423854828,
      "learning_rate": 4.504772348747687e-05,
      "loss": 0.0513,
      "step": 111
    },
    {
      "epoch": 1.5272108843537415,
      "grad_norm": 0.12276334315538406,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 0.0689,
      "step": 112
    },
    {
      "epoch": 1.5408163265306123,
      "grad_norm": 0.08630312234163284,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 0.0589,
      "step": 113
    },
    {
      "epoch": 1.554421768707483,
      "grad_norm": 0.4985141158103943,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 0.1188,
      "step": 114
    },
    {
      "epoch": 1.5680272108843538,
      "grad_norm": 0.1788254678249359,
      "learning_rate": 4.17702704859633e-05,
      "loss": 0.0762,
      "step": 115
    },
    {
      "epoch": 1.5816326530612246,
      "grad_norm": 0.24353311955928802,
      "learning_rate": 4.095597328339452e-05,
      "loss": 0.1022,
      "step": 116
    },
    {
      "epoch": 1.5952380952380953,
      "grad_norm": 0.23691025376319885,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 0.072,
      "step": 117
    },
    {
      "epoch": 1.608843537414966,
      "grad_norm": 0.14665265381336212,
      "learning_rate": 3.933501846281267e-05,
      "loss": 0.1178,
      "step": 118
    },
    {
      "epoch": 1.6224489795918369,
      "grad_norm": 0.07901093363761902,
      "learning_rate": 3.852880399766243e-05,
      "loss": 0.036,
      "step": 119
    },
    {
      "epoch": 1.6360544217687076,
      "grad_norm": 0.3640722334384918,
      "learning_rate": 3.772572564296005e-05,
      "loss": 0.0446,
      "step": 120
    },
    {
      "epoch": 1.6496598639455784,
      "grad_norm": 0.4317243695259094,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 0.1259,
      "step": 121
    },
    {
      "epoch": 1.663265306122449,
      "grad_norm": 0.10436530411243439,
      "learning_rate": 3.612985456190778e-05,
      "loss": 0.0822,
      "step": 122
    },
    {
      "epoch": 1.6768707482993197,
      "grad_norm": 0.22549955546855927,
      "learning_rate": 3.533749813077677e-05,
      "loss": 0.0455,
      "step": 123
    },
    {
      "epoch": 1.6904761904761905,
      "grad_norm": 0.3236393630504608,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 0.1039,
      "step": 124
    },
    {
      "epoch": 1.7040816326530612,
      "grad_norm": 0.11495677381753922,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 0.0433,
      "step": 125
    },
    {
      "epoch": 1.717687074829932,
      "grad_norm": 0.2573133111000061,
      "learning_rate": 3.298534127791785e-05,
      "loss": 0.0573,
      "step": 126
    },
    {
      "epoch": 1.7312925170068028,
      "grad_norm": 0.44331222772598267,
      "learning_rate": 3.221030765387417e-05,
      "loss": 0.175,
      "step": 127
    },
    {
      "epoch": 1.7448979591836735,
      "grad_norm": 0.38616031408309937,
      "learning_rate": 3.144013755408895e-05,
      "loss": 0.0968,
      "step": 128
    },
    {
      "epoch": 1.758503401360544,
      "grad_norm": 0.29330262541770935,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 0.1327,
      "step": 129
    },
    {
      "epoch": 1.7721088435374148,
      "grad_norm": 0.48555678129196167,
      "learning_rate": 2.991522876735154e-05,
      "loss": 0.0728,
      "step": 130
    },
    {
      "epoch": 1.7857142857142856,
      "grad_norm": 0.33002161979675293,
      "learning_rate": 2.916090697523549e-05,
      "loss": 0.0762,
      "step": 131
    },
    {
      "epoch": 1.7993197278911564,
      "grad_norm": 0.15878871083259583,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 0.0614,
      "step": 132
    },
    {
      "epoch": 1.8129251700680271,
      "grad_norm": 0.41857343912124634,
      "learning_rate": 2.766955965735968e-05,
      "loss": 0.0365,
      "step": 133
    },
    {
      "epoch": 1.8265306122448979,
      "grad_norm": 0.3290554881095886,
      "learning_rate": 2.693294185106562e-05,
      "loss": 0.0443,
      "step": 134
    },
    {
      "epoch": 1.8401360544217686,
      "grad_norm": 0.10365504026412964,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 0.0663,
      "step": 135
    },
    {
      "epoch": 1.8537414965986394,
      "grad_norm": 0.32278478145599365,
      "learning_rate": 2.547882480847461e-05,
      "loss": 0.0407,
      "step": 136
    },
    {
      "epoch": 1.8673469387755102,
      "grad_norm": 0.14270636439323425,
      "learning_rate": 2.476172311325783e-05,
      "loss": 0.0681,
      "step": 137
    },
    {
      "epoch": 1.880952380952381,
      "grad_norm": 0.21360638737678528,
      "learning_rate": 2.405152131093926e-05,
      "loss": 0.0436,
      "step": 138
    },
    {
      "epoch": 1.8945578231292517,
      "grad_norm": 0.10645715892314911,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 0.0703,
      "step": 139
    },
    {
      "epoch": 1.9081632653061225,
      "grad_norm": 0.10856879502534866,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 0.0862,
      "step": 140
    },
    {
      "epoch": 1.9217687074829932,
      "grad_norm": 0.3066421151161194,
      "learning_rate": 2.196424713241637e-05,
      "loss": 0.027,
      "step": 141
    },
    {
      "epoch": 1.935374149659864,
      "grad_norm": 0.1592700034379959,
      "learning_rate": 2.128356686585282e-05,
      "loss": 0.1025,
      "step": 142
    },
    {
      "epoch": 1.9489795918367347,
      "grad_norm": 0.12343791127204895,
      "learning_rate": 2.061073738537635e-05,
      "loss": 0.0291,
      "step": 143
    },
    {
      "epoch": 1.9625850340136055,
      "grad_norm": 0.1284944862127304,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 0.0366,
      "step": 144
    },
    {
      "epoch": 1.9761904761904763,
      "grad_norm": 0.1971082091331482,
      "learning_rate": 1.928936436551661e-05,
      "loss": 0.0329,
      "step": 145
    },
    {
      "epoch": 1.989795918367347,
      "grad_norm": 0.11205492913722992,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 0.1142,
      "step": 146
    },
    {
      "epoch": 2.006802721088435,
      "grad_norm": 0.5591760873794556,
      "learning_rate": 1.800157297483417e-05,
      "loss": 0.1608,
      "step": 147
    },
    {
      "epoch": 2.020408163265306,
      "grad_norm": 0.16187676787376404,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 0.0471,
      "step": 148
    },
    {
      "epoch": 2.0340136054421767,
      "grad_norm": 0.0873565748333931,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 0.0749,
      "step": 149
    },
    {
      "epoch": 2.0476190476190474,
      "grad_norm": 0.20911337435245514,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 0.1276,
      "step": 150
    },
    {
      "epoch": 2.0476190476190474,
      "eval_loss": 0.09702116250991821,
      "eval_runtime": 17.4701,
      "eval_samples_per_second": 7.098,
      "eval_steps_per_second": 1.774,
      "step": 150
    },
    {
      "epoch": 2.061224489795918,
      "grad_norm": 0.2917337119579315,
      "learning_rate": 1.553232954407171e-05,
      "loss": 0.0378,
      "step": 151
    },
    {
      "epoch": 2.074829931972789,
      "grad_norm": 0.09112036973237991,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 0.052,
      "step": 152
    },
    {
      "epoch": 2.0884353741496597,
      "grad_norm": 0.09712621569633484,
      "learning_rate": 1.435357758543015e-05,
      "loss": 0.0679,
      "step": 153
    },
    {
      "epoch": 2.1020408163265305,
      "grad_norm": 0.149315744638443,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 0.0764,
      "step": 154
    },
    {
      "epoch": 2.1156462585034013,
      "grad_norm": 0.3866598606109619,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 0.0889,
      "step": 155
    },
    {
      "epoch": 2.129251700680272,
      "grad_norm": 0.08537255972623825,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 0.0319,
      "step": 156
    },
    {
      "epoch": 2.142857142857143,
      "grad_norm": 0.09636199474334717,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 0.05,
      "step": 157
    },
    {
      "epoch": 2.1564625850340136,
      "grad_norm": 0.1146092414855957,
      "learning_rate": 1.157994445715706e-05,
      "loss": 0.0735,
      "step": 158
    },
    {
      "epoch": 2.1700680272108843,
      "grad_norm": 0.08625253289937973,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 0.0479,
      "step": 159
    },
    {
      "epoch": 2.183673469387755,
      "grad_norm": 0.1440863460302353,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 0.0519,
      "step": 160
    },
    {
      "epoch": 2.197278911564626,
      "grad_norm": 0.2233511507511139,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 0.0215,
      "step": 161
    },
    {
      "epoch": 2.2108843537414966,
      "grad_norm": 0.09955249726772308,
      "learning_rate": 9.549150281252633e-06,
      "loss": 0.0834,
      "step": 162
    },
    {
      "epoch": 2.2244897959183674,
      "grad_norm": 0.07915307581424713,
      "learning_rate": 9.068759265665384e-06,
      "loss": 0.0266,
      "step": 163
    },
    {
      "epoch": 2.238095238095238,
      "grad_norm": 0.1018153503537178,
      "learning_rate": 8.599558442598998e-06,
      "loss": 0.0355,
      "step": 164
    },
    {
      "epoch": 2.251700680272109,
      "grad_norm": 0.171899676322937,
      "learning_rate": 8.141676086873572e-06,
      "loss": 0.1047,
      "step": 165
    },
    {
      "epoch": 2.2653061224489797,
      "grad_norm": 0.22086167335510254,
      "learning_rate": 7.695237378953223e-06,
      "loss": 0.0733,
      "step": 166
    },
    {
      "epoch": 2.2789115646258504,
      "grad_norm": 0.3049245774745941,
      "learning_rate": 7.260364370723044e-06,
      "loss": 0.0533,
      "step": 167
    },
    {
      "epoch": 2.292517006802721,
      "grad_norm": 0.30612751841545105,
      "learning_rate": 6.837175952121306e-06,
      "loss": 0.0699,
      "step": 168
    },
    {
      "epoch": 2.306122448979592,
      "grad_norm": 0.09367984533309937,
      "learning_rate": 6.425787818636131e-06,
      "loss": 0.0768,
      "step": 169
    },
    {
      "epoch": 2.3197278911564627,
      "grad_norm": 0.23378770053386688,
      "learning_rate": 6.026312439675552e-06,
      "loss": 0.1211,
      "step": 170
    },
    {
      "epoch": 2.3333333333333335,
      "grad_norm": 0.25364062190055847,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 0.0345,
      "step": 171
    },
    {
      "epoch": 2.3469387755102042,
      "grad_norm": 0.2675899267196655,
      "learning_rate": 5.263533508961827e-06,
      "loss": 0.0816,
      "step": 172
    },
    {
      "epoch": 2.360544217687075,
      "grad_norm": 0.1582193374633789,
      "learning_rate": 4.900438493352055e-06,
      "loss": 0.0346,
      "step": 173
    },
    {
      "epoch": 2.3741496598639458,
      "grad_norm": 0.09332657605409622,
      "learning_rate": 4.549673247541875e-06,
      "loss": 0.0392,
      "step": 174
    },
    {
      "epoch": 2.387755102040816,
      "grad_norm": 0.10757287591695786,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 0.0953,
      "step": 175
    },
    {
      "epoch": 2.4013605442176873,
      "grad_norm": 0.09674852341413498,
      "learning_rate": 3.885512251130763e-06,
      "loss": 0.0426,
      "step": 176
    },
    {
      "epoch": 2.4149659863945576,
      "grad_norm": 0.20280176401138306,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 0.1943,
      "step": 177
    },
    {
      "epoch": 2.4285714285714284,
      "grad_norm": 0.10712963342666626,
      "learning_rate": 3.271776770026963e-06,
      "loss": 0.0289,
      "step": 178
    },
    {
      "epoch": 2.442176870748299,
      "grad_norm": 0.15586324036121368,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 0.0308,
      "step": 179
    },
    {
      "epoch": 2.45578231292517,
      "grad_norm": 0.11648932099342346,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 0.02,
      "step": 180
    },
    {
      "epoch": 2.4693877551020407,
      "grad_norm": 0.11573922634124756,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 0.0267,
      "step": 181
    },
    {
      "epoch": 2.4829931972789114,
      "grad_norm": 0.12493091821670532,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 0.0451,
      "step": 182
    },
    {
      "epoch": 2.496598639455782,
      "grad_norm": 0.11222916841506958,
      "learning_rate": 1.962316193157593e-06,
      "loss": 0.051,
      "step": 183
    },
    {
      "epoch": 2.510204081632653,
      "grad_norm": 0.1220880001783371,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 0.0439,
      "step": 184
    },
    {
      "epoch": 2.5238095238095237,
      "grad_norm": 0.08894781768321991,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 0.082,
      "step": 185
    },
    {
      "epoch": 2.5374149659863945,
      "grad_norm": 0.12918499112129211,
      "learning_rate": 1.333670137599713e-06,
      "loss": 0.0444,
      "step": 186
    },
    {
      "epoch": 2.5510204081632653,
      "grad_norm": 0.15321969985961914,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 0.0472,
      "step": 187
    },
    {
      "epoch": 2.564625850340136,
      "grad_norm": 0.13897962868213654,
      "learning_rate": 9.810017062595322e-07,
      "loss": 0.1051,
      "step": 188
    },
    {
      "epoch": 2.578231292517007,
      "grad_norm": 0.1774894893169403,
      "learning_rate": 8.247462563808817e-07,
      "loss": 0.036,
      "step": 189
    },
    {
      "epoch": 2.5918367346938775,
      "grad_norm": 0.10312793403863907,
      "learning_rate": 6.819348298638839e-07,
      "loss": 0.0522,
      "step": 190
    },
    {
      "epoch": 2.6054421768707483,
      "grad_norm": 0.1317165493965149,
      "learning_rate": 5.526064699265753e-07,
      "loss": 0.0398,
      "step": 191
    },
    {
      "epoch": 2.619047619047619,
      "grad_norm": 0.0993337333202362,
      "learning_rate": 4.367965336512403e-07,
      "loss": 0.0499,
      "step": 192
    },
    {
      "epoch": 2.63265306122449,
      "grad_norm": 0.22714680433273315,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 0.0382,
      "step": 193
    },
    {
      "epoch": 2.6462585034013606,
      "grad_norm": 0.10576555132865906,
      "learning_rate": 2.458548727494292e-07,
      "loss": 0.0446,
      "step": 194
    },
    {
      "epoch": 2.6598639455782314,
      "grad_norm": 0.09238826483488083,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 0.0409,
      "step": 195
    },
    {
      "epoch": 2.673469387755102,
      "grad_norm": 0.0816134363412857,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 0.0299,
      "step": 196
    },
    {
      "epoch": 2.687074829931973,
      "grad_norm": 0.14053530991077423,
      "learning_rate": 6.150154258476315e-08,
      "loss": 0.0596,
      "step": 197
    },
    {
      "epoch": 2.7006802721088436,
      "grad_norm": 0.17438529431819916,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 0.07,
      "step": 198
    },
    {
      "epoch": 2.7142857142857144,
      "grad_norm": 0.1229928582906723,
      "learning_rate": 6.834750376549792e-09,
      "loss": 0.0274,
      "step": 199
    },
    {
      "epoch": 2.727891156462585,
      "grad_norm": 0.1715364158153534,
      "learning_rate": 0.0,
      "loss": 0.068,
      "step": 200
    },
    {
      "epoch": 2.727891156462585,
      "eval_loss": 0.0965760126709938,
      "eval_runtime": 17.4515,
      "eval_samples_per_second": 7.105,
      "eval_steps_per_second": 1.776,
      "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": 0
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 5.322034636652544e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}