{
  "best_metric": 0.6898359656333923,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.3401360544217687,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0017006802721088435,
      "grad_norm": 0.2204342782497406,
      "learning_rate": 1e-05,
      "loss": 0.8125,
      "step": 1
    },
    {
      "epoch": 0.0017006802721088435,
      "eval_loss": 1.186415433883667,
      "eval_runtime": 125.7773,
      "eval_samples_per_second": 7.871,
      "eval_steps_per_second": 1.972,
      "step": 1
    },
    {
      "epoch": 0.003401360544217687,
      "grad_norm": 0.24905377626419067,
      "learning_rate": 2e-05,
      "loss": 0.8426,
      "step": 2
    },
    {
      "epoch": 0.00510204081632653,
      "grad_norm": 0.30998170375823975,
      "learning_rate": 3e-05,
      "loss": 0.9178,
      "step": 3
    },
    {
      "epoch": 0.006802721088435374,
      "grad_norm": 0.28902438282966614,
      "learning_rate": 4e-05,
      "loss": 0.9068,
      "step": 4
    },
    {
      "epoch": 0.008503401360544218,
      "grad_norm": 0.2688828706741333,
      "learning_rate": 5e-05,
      "loss": 0.8636,
      "step": 5
    },
    {
      "epoch": 0.01020408163265306,
      "grad_norm": 0.2665344178676605,
      "learning_rate": 6e-05,
      "loss": 0.8967,
      "step": 6
    },
    {
      "epoch": 0.011904761904761904,
      "grad_norm": 0.26651954650878906,
      "learning_rate": 7e-05,
      "loss": 0.8186,
      "step": 7
    },
    {
      "epoch": 0.013605442176870748,
      "grad_norm": 0.2417740821838379,
      "learning_rate": 8e-05,
      "loss": 0.8938,
      "step": 8
    },
    {
      "epoch": 0.015306122448979591,
      "grad_norm": 0.20842400193214417,
      "learning_rate": 9e-05,
      "loss": 0.8489,
      "step": 9
    },
    {
      "epoch": 0.017006802721088437,
      "grad_norm": 0.20249231159687042,
      "learning_rate": 0.0001,
      "loss": 0.9356,
      "step": 10
    },
    {
      "epoch": 0.01870748299319728,
      "grad_norm": 0.2657685875892639,
      "learning_rate": 9.999316524962345e-05,
      "loss": 0.7823,
      "step": 11
    },
    {
      "epoch": 0.02040816326530612,
      "grad_norm": 0.2690357267856598,
      "learning_rate": 9.997266286704631e-05,
      "loss": 0.7644,
      "step": 12
    },
    {
      "epoch": 0.022108843537414966,
      "grad_norm": 0.27657201886177063,
      "learning_rate": 9.993849845741524e-05,
      "loss": 0.8285,
      "step": 13
    },
    {
      "epoch": 0.023809523809523808,
      "grad_norm": 0.20251011848449707,
      "learning_rate": 9.989068136093873e-05,
      "loss": 0.7242,
      "step": 14
    },
    {
      "epoch": 0.025510204081632654,
      "grad_norm": 0.2040708065032959,
      "learning_rate": 9.98292246503335e-05,
      "loss": 0.8771,
      "step": 15
    },
    {
      "epoch": 0.027210884353741496,
      "grad_norm": 0.19819475710391998,
      "learning_rate": 9.975414512725057e-05,
      "loss": 0.9122,
      "step": 16
    },
    {
      "epoch": 0.02891156462585034,
      "grad_norm": 0.20104283094406128,
      "learning_rate": 9.966546331768191e-05,
      "loss": 0.8825,
      "step": 17
    },
    {
      "epoch": 0.030612244897959183,
      "grad_norm": 0.19711934030056,
      "learning_rate": 9.956320346634876e-05,
      "loss": 0.8565,
      "step": 18
    },
    {
      "epoch": 0.03231292517006803,
      "grad_norm": 0.21350829303264618,
      "learning_rate": 9.944739353007344e-05,
      "loss": 0.9454,
      "step": 19
    },
    {
      "epoch": 0.034013605442176874,
      "grad_norm": 0.2247208207845688,
      "learning_rate": 9.931806517013612e-05,
      "loss": 0.9034,
      "step": 20
    },
    {
      "epoch": 0.03571428571428571,
      "grad_norm": 0.20988470315933228,
      "learning_rate": 9.917525374361912e-05,
      "loss": 0.8926,
      "step": 21
    },
    {
      "epoch": 0.03741496598639456,
      "grad_norm": 0.21894076466560364,
      "learning_rate": 9.901899829374047e-05,
      "loss": 0.8922,
      "step": 22
    },
    {
      "epoch": 0.0391156462585034,
      "grad_norm": 0.19649578630924225,
      "learning_rate": 9.884934153917997e-05,
      "loss": 0.8124,
      "step": 23
    },
    {
      "epoch": 0.04081632653061224,
      "grad_norm": 0.2259817123413086,
      "learning_rate": 9.86663298624003e-05,
      "loss": 0.8659,
      "step": 24
    },
    {
      "epoch": 0.04251700680272109,
      "grad_norm": 0.20764009654521942,
      "learning_rate": 9.847001329696653e-05,
      "loss": 0.8369,
      "step": 25
    },
    {
      "epoch": 0.04421768707482993,
      "grad_norm": 0.21430037915706635,
      "learning_rate": 9.826044551386744e-05,
      "loss": 0.8563,
      "step": 26
    },
    {
      "epoch": 0.04591836734693878,
      "grad_norm": 0.23552869260311127,
      "learning_rate": 9.803768380684242e-05,
      "loss": 0.8486,
      "step": 27
    },
    {
      "epoch": 0.047619047619047616,
      "grad_norm": 0.22798550128936768,
      "learning_rate": 9.780178907671789e-05,
      "loss": 0.8939,
      "step": 28
    },
    {
      "epoch": 0.04931972789115646,
      "grad_norm": 0.25970762968063354,
      "learning_rate": 9.755282581475769e-05,
      "loss": 0.9529,
      "step": 29
    },
    {
      "epoch": 0.05102040816326531,
      "grad_norm": 0.2707545757293701,
      "learning_rate": 9.729086208503174e-05,
      "loss": 0.9145,
      "step": 30
    },
    {
      "epoch": 0.05272108843537415,
      "grad_norm": 0.2699640095233917,
      "learning_rate": 9.701596950580806e-05,
      "loss": 0.8509,
      "step": 31
    },
    {
      "epoch": 0.05442176870748299,
      "grad_norm": 0.2942357063293457,
      "learning_rate": 9.672822322997305e-05,
      "loss": 0.9295,
      "step": 32
    },
    {
      "epoch": 0.05612244897959184,
      "grad_norm": 0.28412437438964844,
      "learning_rate": 9.642770192448536e-05,
      "loss": 0.8857,
      "step": 33
    },
    {
      "epoch": 0.05782312925170068,
      "grad_norm": 0.3160867989063263,
      "learning_rate": 9.611448774886924e-05,
      "loss": 0.8412,
      "step": 34
    },
    {
      "epoch": 0.05952380952380952,
      "grad_norm": 0.34405621886253357,
      "learning_rate": 9.578866633275288e-05,
      "loss": 0.9591,
      "step": 35
    },
    {
      "epoch": 0.061224489795918366,
      "grad_norm": 0.30002671480178833,
      "learning_rate": 9.545032675245813e-05,
      "loss": 0.8489,
      "step": 36
    },
    {
      "epoch": 0.06292517006802721,
      "grad_norm": 0.3307010233402252,
      "learning_rate": 9.509956150664796e-05,
      "loss": 0.8392,
      "step": 37
    },
    {
      "epoch": 0.06462585034013606,
      "grad_norm": 0.34140732884407043,
      "learning_rate": 9.473646649103818e-05,
      "loss": 0.9821,
      "step": 38
    },
    {
      "epoch": 0.0663265306122449,
      "grad_norm": 0.3235897719860077,
      "learning_rate": 9.43611409721806e-05,
      "loss": 0.7999,
      "step": 39
    },
    {
      "epoch": 0.06802721088435375,
      "grad_norm": 0.29226845502853394,
      "learning_rate": 9.397368756032445e-05,
      "loss": 0.8352,
      "step": 40
    },
    {
      "epoch": 0.06972789115646258,
      "grad_norm": 0.2626186013221741,
      "learning_rate": 9.357421218136386e-05,
      "loss": 0.9251,
      "step": 41
    },
    {
      "epoch": 0.07142857142857142,
      "grad_norm": 0.25056135654449463,
      "learning_rate": 9.316282404787871e-05,
      "loss": 0.8486,
      "step": 42
    },
    {
      "epoch": 0.07312925170068027,
      "grad_norm": 0.25198081135749817,
      "learning_rate": 9.273963562927695e-05,
      "loss": 0.7745,
      "step": 43
    },
    {
      "epoch": 0.07482993197278912,
      "grad_norm": 0.268450528383255,
      "learning_rate": 9.230476262104677e-05,
      "loss": 0.8602,
      "step": 44
    },
    {
      "epoch": 0.07653061224489796,
      "grad_norm": 0.29419487714767456,
      "learning_rate": 9.185832391312644e-05,
      "loss": 0.8877,
      "step": 45
    },
    {
      "epoch": 0.0782312925170068,
      "grad_norm": 0.26198142766952515,
      "learning_rate": 9.140044155740101e-05,
      "loss": 0.7923,
      "step": 46
    },
    {
      "epoch": 0.07993197278911565,
      "grad_norm": 0.2817314863204956,
      "learning_rate": 9.093124073433463e-05,
      "loss": 0.77,
      "step": 47
    },
    {
      "epoch": 0.08163265306122448,
      "grad_norm": 0.29006093740463257,
      "learning_rate": 9.045084971874738e-05,
      "loss": 0.728,
      "step": 48
    },
    {
      "epoch": 0.08333333333333333,
      "grad_norm": 0.3132553994655609,
      "learning_rate": 8.995939984474624e-05,
      "loss": 0.7328,
      "step": 49
    },
    {
      "epoch": 0.08503401360544217,
      "grad_norm": 0.32709839940071106,
      "learning_rate": 8.945702546981969e-05,
      "loss": 0.693,
      "step": 50
    },
    {
      "epoch": 0.08503401360544217,
      "eval_loss": 0.7769989371299744,
      "eval_runtime": 127.5552,
      "eval_samples_per_second": 7.761,
      "eval_steps_per_second": 1.944,
      "step": 50
    },
    {
      "epoch": 0.08673469387755102,
      "grad_norm": 0.22816550731658936,
      "learning_rate": 8.894386393810563e-05,
      "loss": 0.6048,
      "step": 51
    },
    {
      "epoch": 0.08843537414965986,
      "grad_norm": 0.1930536925792694,
      "learning_rate": 8.842005554284296e-05,
      "loss": 0.6424,
      "step": 52
    },
    {
      "epoch": 0.09013605442176871,
      "grad_norm": 0.17307350039482117,
      "learning_rate": 8.788574348801675e-05,
      "loss": 0.6356,
      "step": 53
    },
    {
      "epoch": 0.09183673469387756,
      "grad_norm": 0.15257245302200317,
      "learning_rate": 8.73410738492077e-05,
      "loss": 0.6886,
      "step": 54
    },
    {
      "epoch": 0.0935374149659864,
      "grad_norm": 0.14338071644306183,
      "learning_rate": 8.678619553365659e-05,
      "loss": 0.6809,
      "step": 55
    },
    {
      "epoch": 0.09523809523809523,
      "grad_norm": 0.1425928920507431,
      "learning_rate": 8.622126023955446e-05,
      "loss": 0.6165,
      "step": 56
    },
    {
      "epoch": 0.09693877551020408,
      "grad_norm": 0.14232009649276733,
      "learning_rate": 8.564642241456986e-05,
      "loss": 0.6862,
      "step": 57
    },
    {
      "epoch": 0.09863945578231292,
      "grad_norm": 0.14961464703083038,
      "learning_rate": 8.506183921362443e-05,
      "loss": 0.7018,
      "step": 58
    },
    {
      "epoch": 0.10034013605442177,
      "grad_norm": 0.1577504426240921,
      "learning_rate": 8.44676704559283e-05,
      "loss": 0.6934,
      "step": 59
    },
    {
      "epoch": 0.10204081632653061,
      "grad_norm": 0.18428894877433777,
      "learning_rate": 8.386407858128706e-05,
      "loss": 0.7689,
      "step": 60
    },
    {
      "epoch": 0.10374149659863946,
      "grad_norm": 0.16556815803050995,
      "learning_rate": 8.32512286056924e-05,
      "loss": 0.6983,
      "step": 61
    },
    {
      "epoch": 0.1054421768707483,
      "grad_norm": 0.1856698989868164,
      "learning_rate": 8.262928807620843e-05,
      "loss": 0.7475,
      "step": 62
    },
    {
      "epoch": 0.10714285714285714,
      "grad_norm": 0.18615658581256866,
      "learning_rate": 8.199842702516583e-05,
      "loss": 0.709,
      "step": 63
    },
    {
      "epoch": 0.10884353741496598,
      "grad_norm": 0.32693877816200256,
      "learning_rate": 8.135881792367686e-05,
      "loss": 0.6919,
      "step": 64
    },
    {
      "epoch": 0.11054421768707483,
      "grad_norm": 0.1930025964975357,
      "learning_rate": 8.07106356344834e-05,
      "loss": 0.7233,
      "step": 65
    },
    {
      "epoch": 0.11224489795918367,
      "grad_norm": 0.19095459580421448,
      "learning_rate": 8.005405736415126e-05,
      "loss": 0.7997,
      "step": 66
    },
    {
      "epoch": 0.11394557823129252,
      "grad_norm": 0.18895427882671356,
      "learning_rate": 7.938926261462366e-05,
      "loss": 0.7756,
      "step": 67
    },
    {
      "epoch": 0.11564625850340136,
      "grad_norm": 0.19725216925144196,
      "learning_rate": 7.871643313414718e-05,
      "loss": 0.821,
      "step": 68
    },
    {
      "epoch": 0.11734693877551021,
      "grad_norm": 0.20601698756217957,
      "learning_rate": 7.803575286758364e-05,
      "loss": 0.77,
      "step": 69
    },
    {
      "epoch": 0.11904761904761904,
      "grad_norm": 0.22490014135837555,
      "learning_rate": 7.734740790612136e-05,
      "loss": 0.8944,
      "step": 70
    },
    {
      "epoch": 0.12074829931972789,
      "grad_norm": 0.21512475609779358,
      "learning_rate": 7.66515864363997e-05,
      "loss": 0.7938,
      "step": 71
    },
    {
      "epoch": 0.12244897959183673,
      "grad_norm": 0.19616465270519257,
      "learning_rate": 7.594847868906076e-05,
      "loss": 0.7361,
      "step": 72
    },
    {
      "epoch": 0.12414965986394558,
      "grad_norm": 0.20421065390110016,
      "learning_rate": 7.52382768867422e-05,
      "loss": 0.7421,
      "step": 73
    },
    {
      "epoch": 0.12585034013605442,
      "grad_norm": 0.21343572437763214,
      "learning_rate": 7.452117519152542e-05,
      "loss": 0.7467,
      "step": 74
    },
    {
      "epoch": 0.12755102040816327,
      "grad_norm": 0.22144010663032532,
      "learning_rate": 7.379736965185368e-05,
      "loss": 0.793,
      "step": 75
    },
    {
      "epoch": 0.1292517006802721,
      "grad_norm": 0.22638201713562012,
      "learning_rate": 7.30670581489344e-05,
      "loss": 0.7574,
      "step": 76
    },
    {
      "epoch": 0.13095238095238096,
      "grad_norm": 0.2620675563812256,
      "learning_rate": 7.233044034264034e-05,
      "loss": 0.6952,
      "step": 77
    },
    {
      "epoch": 0.1326530612244898,
      "grad_norm": 0.24176040291786194,
      "learning_rate": 7.158771761692464e-05,
      "loss": 0.7118,
      "step": 78
    },
    {
      "epoch": 0.13435374149659865,
      "grad_norm": 0.25384989380836487,
      "learning_rate": 7.083909302476453e-05,
      "loss": 0.7701,
      "step": 79
    },
    {
      "epoch": 0.1360544217687075,
      "grad_norm": 0.2719091773033142,
      "learning_rate": 7.008477123264848e-05,
      "loss": 0.7707,
      "step": 80
    },
    {
      "epoch": 0.1377551020408163,
      "grad_norm": 0.40553587675094604,
      "learning_rate": 6.932495846462261e-05,
      "loss": 0.8467,
      "step": 81
    },
    {
      "epoch": 0.13945578231292516,
      "grad_norm": 0.5253201723098755,
      "learning_rate": 6.855986244591104e-05,
      "loss": 0.7712,
      "step": 82
    },
    {
      "epoch": 0.141156462585034,
      "grad_norm": 0.3024201989173889,
      "learning_rate": 6.778969234612584e-05,
      "loss": 0.7769,
      "step": 83
    },
    {
      "epoch": 0.14285714285714285,
      "grad_norm": 0.32817140221595764,
      "learning_rate": 6.701465872208216e-05,
      "loss": 0.7806,
      "step": 84
    },
    {
      "epoch": 0.1445578231292517,
      "grad_norm": 0.30821433663368225,
      "learning_rate": 6.623497346023418e-05,
      "loss": 0.6983,
      "step": 85
    },
    {
      "epoch": 0.14625850340136054,
      "grad_norm": 0.2959168255329132,
      "learning_rate": 6.545084971874738e-05,
      "loss": 0.735,
      "step": 86
    },
    {
      "epoch": 0.14795918367346939,
      "grad_norm": 0.3219025731086731,
      "learning_rate": 6.466250186922325e-05,
      "loss": 0.7748,
      "step": 87
    },
    {
      "epoch": 0.14965986394557823,
      "grad_norm": 0.32275092601776123,
      "learning_rate": 6.387014543809223e-05,
      "loss": 0.7313,
      "step": 88
    },
    {
      "epoch": 0.15136054421768708,
      "grad_norm": 0.3146684169769287,
      "learning_rate": 6.307399704769099e-05,
      "loss": 0.7486,
      "step": 89
    },
    {
      "epoch": 0.15306122448979592,
      "grad_norm": 0.26414763927459717,
      "learning_rate": 6.227427435703997e-05,
      "loss": 0.7407,
      "step": 90
    },
    {
      "epoch": 0.15476190476190477,
      "grad_norm": 0.2503359317779541,
      "learning_rate": 6.147119600233758e-05,
      "loss": 0.771,
      "step": 91
    },
    {
      "epoch": 0.1564625850340136,
      "grad_norm": 0.2612239420413971,
      "learning_rate": 6.066498153718735e-05,
      "loss": 0.7619,
      "step": 92
    },
    {
      "epoch": 0.15816326530612246,
      "grad_norm": 0.2623046338558197,
      "learning_rate": 5.985585137257401e-05,
      "loss": 0.6933,
      "step": 93
    },
    {
      "epoch": 0.1598639455782313,
      "grad_norm": 0.25616154074668884,
      "learning_rate": 5.90440267166055e-05,
      "loss": 0.7269,
      "step": 94
    },
    {
      "epoch": 0.16156462585034015,
      "grad_norm": 0.29855290055274963,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 0.7612,
      "step": 95
    },
    {
      "epoch": 0.16326530612244897,
      "grad_norm": 0.2623254656791687,
      "learning_rate": 5.74131823855921e-05,
      "loss": 0.6954,
      "step": 96
    },
    {
      "epoch": 0.1649659863945578,
      "grad_norm": 0.3363308608531952,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 0.7446,
      "step": 97
    },
    {
      "epoch": 0.16666666666666666,
      "grad_norm": 0.30886736512184143,
      "learning_rate": 5.577423184847932e-05,
      "loss": 0.6852,
      "step": 98
    },
    {
      "epoch": 0.1683673469387755,
      "grad_norm": 0.3511720299720764,
      "learning_rate": 5.495227651252315e-05,
      "loss": 0.7374,
      "step": 99
    },
    {
      "epoch": 0.17006802721088435,
      "grad_norm": 0.391239732503891,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 0.7071,
      "step": 100
    },
    {
      "epoch": 0.17006802721088435,
      "eval_loss": 0.7250789403915405,
      "eval_runtime": 127.4104,
      "eval_samples_per_second": 7.77,
      "eval_steps_per_second": 1.946,
      "step": 100
    },
    {
      "epoch": 0.1717687074829932,
      "grad_norm": 0.15029719471931458,
      "learning_rate": 5.330452921628497e-05,
      "loss": 0.5888,
      "step": 101
    },
    {
      "epoch": 0.17346938775510204,
      "grad_norm": 0.15537625551223755,
      "learning_rate": 5.247918773366112e-05,
      "loss": 0.5555,
      "step": 102
    },
    {
      "epoch": 0.17517006802721088,
      "grad_norm": 0.1484169363975525,
      "learning_rate": 5.165316846586541e-05,
      "loss": 0.6206,
      "step": 103
    },
    {
      "epoch": 0.17687074829931973,
      "grad_norm": 0.16408419609069824,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 0.6376,
      "step": 104
    },
    {
      "epoch": 0.17857142857142858,
      "grad_norm": 0.14357221126556396,
      "learning_rate": 5e-05,
      "loss": 0.5652,
      "step": 105
    },
    {
      "epoch": 0.18027210884353742,
      "grad_norm": 0.1612963229417801,
      "learning_rate": 4.917330276168208e-05,
      "loss": 0.6299,
      "step": 106
    },
    {
      "epoch": 0.18197278911564627,
      "grad_norm": 0.14948496222496033,
      "learning_rate": 4.834683153413459e-05,
      "loss": 0.6943,
      "step": 107
    },
    {
      "epoch": 0.1836734693877551,
      "grad_norm": 0.16210488975048065,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 0.6526,
      "step": 108
    },
    {
      "epoch": 0.18537414965986396,
      "grad_norm": 0.15704397857189178,
      "learning_rate": 4.669547078371504e-05,
      "loss": 0.67,
      "step": 109
    },
    {
      "epoch": 0.1870748299319728,
      "grad_norm": 0.1561141163110733,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 0.5969,
      "step": 110
    },
    {
      "epoch": 0.18877551020408162,
      "grad_norm": 0.16394050419330597,
      "learning_rate": 4.504772348747687e-05,
      "loss": 0.6346,
      "step": 111
    },
    {
      "epoch": 0.19047619047619047,
      "grad_norm": 0.1791459321975708,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 0.7194,
      "step": 112
    },
    {
      "epoch": 0.1921768707482993,
      "grad_norm": 0.19085873663425446,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 0.759,
      "step": 113
    },
    {
      "epoch": 0.19387755102040816,
      "grad_norm": 0.1820947825908661,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 0.7621,
      "step": 114
    },
    {
      "epoch": 0.195578231292517,
      "grad_norm": 0.2045283019542694,
      "learning_rate": 4.17702704859633e-05,
      "loss": 0.7066,
      "step": 115
    },
    {
      "epoch": 0.19727891156462585,
      "grad_norm": 0.1924622654914856,
      "learning_rate": 4.095597328339452e-05,
      "loss": 0.6942,
      "step": 116
    },
    {
      "epoch": 0.1989795918367347,
      "grad_norm": 0.2077883780002594,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 0.7255,
      "step": 117
    },
    {
      "epoch": 0.20068027210884354,
      "grad_norm": 0.20034259557724,
      "learning_rate": 3.933501846281267e-05,
      "loss": 0.6539,
      "step": 118
    },
    {
      "epoch": 0.20238095238095238,
      "grad_norm": 0.20899388194084167,
      "learning_rate": 3.852880399766243e-05,
      "loss": 0.7123,
      "step": 119
    },
    {
      "epoch": 0.20408163265306123,
      "grad_norm": 0.23263196647167206,
      "learning_rate": 3.772572564296005e-05,
      "loss": 0.6713,
      "step": 120
    },
    {
      "epoch": 0.20578231292517007,
      "grad_norm": 0.20194005966186523,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 0.674,
      "step": 121
    },
    {
      "epoch": 0.20748299319727892,
      "grad_norm": 0.22397828102111816,
      "learning_rate": 3.612985456190778e-05,
      "loss": 0.7205,
      "step": 122
    },
    {
      "epoch": 0.20918367346938777,
      "grad_norm": 0.238456130027771,
      "learning_rate": 3.533749813077677e-05,
      "loss": 0.663,
      "step": 123
    },
    {
      "epoch": 0.2108843537414966,
      "grad_norm": 0.22754044830799103,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 0.7276,
      "step": 124
    },
    {
      "epoch": 0.21258503401360543,
      "grad_norm": 0.23544682562351227,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 0.806,
      "step": 125
    },
    {
      "epoch": 0.21428571428571427,
      "grad_norm": 0.27683186531066895,
      "learning_rate": 3.298534127791785e-05,
      "loss": 0.783,
      "step": 126
    },
    {
      "epoch": 0.21598639455782312,
      "grad_norm": 0.2276148796081543,
      "learning_rate": 3.221030765387417e-05,
      "loss": 0.7032,
      "step": 127
    },
    {
      "epoch": 0.21768707482993196,
      "grad_norm": 0.27521204948425293,
      "learning_rate": 3.144013755408895e-05,
      "loss": 0.8313,
      "step": 128
    },
    {
      "epoch": 0.2193877551020408,
      "grad_norm": 0.28434935212135315,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 0.7972,
      "step": 129
    },
    {
      "epoch": 0.22108843537414966,
      "grad_norm": 0.261618047952652,
      "learning_rate": 2.991522876735154e-05,
      "loss": 0.7797,
      "step": 130
    },
    {
      "epoch": 0.2227891156462585,
      "grad_norm": 0.277802050113678,
      "learning_rate": 2.916090697523549e-05,
      "loss": 0.7285,
      "step": 131
    },
    {
      "epoch": 0.22448979591836735,
      "grad_norm": 0.3588659465312958,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 0.7187,
      "step": 132
    },
    {
      "epoch": 0.2261904761904762,
      "grad_norm": 0.2697806656360626,
      "learning_rate": 2.766955965735968e-05,
      "loss": 0.7424,
      "step": 133
    },
    {
      "epoch": 0.22789115646258504,
      "grad_norm": 0.2989138662815094,
      "learning_rate": 2.693294185106562e-05,
      "loss": 0.7205,
      "step": 134
    },
    {
      "epoch": 0.22959183673469388,
      "grad_norm": 0.3319271504878998,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 0.7191,
      "step": 135
    },
    {
      "epoch": 0.23129251700680273,
      "grad_norm": 0.34353968501091003,
      "learning_rate": 2.547882480847461e-05,
      "loss": 0.7641,
      "step": 136
    },
    {
      "epoch": 0.23299319727891157,
      "grad_norm": 0.35848715901374817,
      "learning_rate": 2.476172311325783e-05,
      "loss": 0.6636,
      "step": 137
    },
    {
      "epoch": 0.23469387755102042,
      "grad_norm": 0.3979659378528595,
      "learning_rate": 2.405152131093926e-05,
      "loss": 0.6431,
      "step": 138
    },
    {
      "epoch": 0.23639455782312926,
      "grad_norm": 0.3438347280025482,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 0.7798,
      "step": 139
    },
    {
      "epoch": 0.23809523809523808,
      "grad_norm": 0.49516430497169495,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 0.817,
      "step": 140
    },
    {
      "epoch": 0.23979591836734693,
      "grad_norm": 0.28333780169487,
      "learning_rate": 2.196424713241637e-05,
      "loss": 0.7216,
      "step": 141
    },
    {
      "epoch": 0.24149659863945577,
      "grad_norm": 0.24931535124778748,
      "learning_rate": 2.128356686585282e-05,
      "loss": 0.6937,
      "step": 142
    },
    {
      "epoch": 0.24319727891156462,
      "grad_norm": 0.2970627248287201,
      "learning_rate": 2.061073738537635e-05,
      "loss": 0.735,
      "step": 143
    },
    {
      "epoch": 0.24489795918367346,
      "grad_norm": 0.2954237759113312,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 0.677,
      "step": 144
    },
    {
      "epoch": 0.2465986394557823,
      "grad_norm": 0.3293488919734955,
      "learning_rate": 1.928936436551661e-05,
      "loss": 0.8138,
      "step": 145
    },
    {
      "epoch": 0.24829931972789115,
      "grad_norm": 0.3099609613418579,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 0.6612,
      "step": 146
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.3167480528354645,
      "learning_rate": 1.800157297483417e-05,
      "loss": 0.6573,
      "step": 147
    },
    {
      "epoch": 0.25170068027210885,
      "grad_norm": 0.2991001009941101,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 0.5782,
      "step": 148
    },
    {
      "epoch": 0.2534013605442177,
      "grad_norm": 0.3861240744590759,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 0.6565,
      "step": 149
    },
    {
      "epoch": 0.25510204081632654,
      "grad_norm": 0.41108861565589905,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 0.6469,
      "step": 150
    },
    {
      "epoch": 0.25510204081632654,
      "eval_loss": 0.6988178491592407,
      "eval_runtime": 127.2393,
      "eval_samples_per_second": 7.781,
      "eval_steps_per_second": 1.949,
      "step": 150
    },
    {
      "epoch": 0.2568027210884354,
      "grad_norm": 0.13350877165794373,
      "learning_rate": 1.553232954407171e-05,
      "loss": 0.5609,
      "step": 151
    },
    {
      "epoch": 0.2585034013605442,
      "grad_norm": 0.13210421800613403,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 0.5359,
      "step": 152
    },
    {
      "epoch": 0.2602040816326531,
      "grad_norm": 0.15549764037132263,
      "learning_rate": 1.435357758543015e-05,
      "loss": 0.5749,
      "step": 153
    },
    {
      "epoch": 0.2619047619047619,
      "grad_norm": 0.15745224058628082,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 0.6344,
      "step": 154
    },
    {
      "epoch": 0.26360544217687076,
      "grad_norm": 0.16423070430755615,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 0.5869,
      "step": 155
    },
    {
      "epoch": 0.2653061224489796,
      "grad_norm": 0.16184845566749573,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 0.5716,
      "step": 156
    },
    {
      "epoch": 0.26700680272108845,
      "grad_norm": 0.1705847978591919,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 0.5796,
      "step": 157
    },
    {
      "epoch": 0.2687074829931973,
      "grad_norm": 0.17455095052719116,
      "learning_rate": 1.157994445715706e-05,
      "loss": 0.6581,
      "step": 158
    },
    {
      "epoch": 0.27040816326530615,
      "grad_norm": 0.1829204112291336,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 0.6812,
      "step": 159
    },
    {
      "epoch": 0.272108843537415,
      "grad_norm": 0.17729905247688293,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 0.5827,
      "step": 160
    },
    {
      "epoch": 0.27380952380952384,
      "grad_norm": 0.19451427459716797,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 0.6578,
      "step": 161
    },
    {
      "epoch": 0.2755102040816326,
      "grad_norm": 0.18990348279476166,
      "learning_rate": 9.549150281252633e-06,
      "loss": 0.6932,
      "step": 162
    },
    {
      "epoch": 0.27721088435374147,
      "grad_norm": 0.18601825833320618,
      "learning_rate": 9.068759265665384e-06,
      "loss": 0.5662,
      "step": 163
    },
    {
      "epoch": 0.2789115646258503,
      "grad_norm": 0.21824043989181519,
      "learning_rate": 8.599558442598998e-06,
      "loss": 0.663,
      "step": 164
    },
    {
      "epoch": 0.28061224489795916,
      "grad_norm": 0.21332722902297974,
      "learning_rate": 8.141676086873572e-06,
      "loss": 0.7489,
      "step": 165
    },
    {
      "epoch": 0.282312925170068,
      "grad_norm": 0.2076265513896942,
      "learning_rate": 7.695237378953223e-06,
      "loss": 0.7016,
      "step": 166
    },
    {
      "epoch": 0.28401360544217685,
      "grad_norm": 0.2165721207857132,
      "learning_rate": 7.260364370723044e-06,
      "loss": 0.6221,
      "step": 167
    },
    {
      "epoch": 0.2857142857142857,
      "grad_norm": 0.23035438358783722,
      "learning_rate": 6.837175952121306e-06,
      "loss": 0.7567,
      "step": 168
    },
    {
      "epoch": 0.28741496598639454,
      "grad_norm": 0.21310852468013763,
      "learning_rate": 6.425787818636131e-06,
      "loss": 0.7541,
      "step": 169
    },
    {
      "epoch": 0.2891156462585034,
      "grad_norm": 0.23578639328479767,
      "learning_rate": 6.026312439675552e-06,
      "loss": 0.7102,
      "step": 170
    },
    {
      "epoch": 0.29081632653061223,
      "grad_norm": 0.21661622822284698,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 0.6931,
      "step": 171
    },
    {
      "epoch": 0.2925170068027211,
      "grad_norm": 0.2488073706626892,
      "learning_rate": 5.263533508961827e-06,
      "loss": 0.6942,
      "step": 172
    },
    {
      "epoch": 0.2942176870748299,
      "grad_norm": 0.24162854254245758,
      "learning_rate": 4.900438493352055e-06,
      "loss": 0.6639,
      "step": 173
    },
    {
      "epoch": 0.29591836734693877,
      "grad_norm": 0.27713561058044434,
      "learning_rate": 4.549673247541875e-06,
      "loss": 0.7448,
      "step": 174
    },
    {
      "epoch": 0.2976190476190476,
      "grad_norm": 0.2654680013656616,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 0.7978,
      "step": 175
    },
    {
      "epoch": 0.29931972789115646,
      "grad_norm": 0.28977084159851074,
      "learning_rate": 3.885512251130763e-06,
      "loss": 0.7814,
      "step": 176
    },
    {
      "epoch": 0.3010204081632653,
      "grad_norm": 0.28827476501464844,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 0.7469,
      "step": 177
    },
    {
      "epoch": 0.30272108843537415,
      "grad_norm": 0.2795083224773407,
      "learning_rate": 3.271776770026963e-06,
      "loss": 0.7343,
      "step": 178
    },
    {
      "epoch": 0.304421768707483,
      "grad_norm": 0.31644758582115173,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 0.7288,
      "step": 179
    },
    {
      "epoch": 0.30612244897959184,
      "grad_norm": 0.34990450739860535,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 0.8129,
      "step": 180
    },
    {
      "epoch": 0.3078231292517007,
      "grad_norm": 0.3699513077735901,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 0.7217,
      "step": 181
    },
    {
      "epoch": 0.30952380952380953,
      "grad_norm": 0.2902442216873169,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 0.7548,
      "step": 182
    },
    {
      "epoch": 0.3112244897959184,
      "grad_norm": 0.34277424216270447,
      "learning_rate": 1.962316193157593e-06,
      "loss": 0.8659,
      "step": 183
    },
    {
      "epoch": 0.3129251700680272,
      "grad_norm": 0.3525453209877014,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 0.7295,
      "step": 184
    },
    {
      "epoch": 0.31462585034013607,
      "grad_norm": 0.35110706090927124,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 0.756,
      "step": 185
    },
    {
      "epoch": 0.3163265306122449,
      "grad_norm": 0.4118176996707916,
      "learning_rate": 1.333670137599713e-06,
      "loss": 0.7921,
      "step": 186
    },
    {
      "epoch": 0.31802721088435376,
      "grad_norm": 0.3601189851760864,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 0.7608,
      "step": 187
    },
    {
      "epoch": 0.3197278911564626,
      "grad_norm": 0.3018999993801117,
      "learning_rate": 9.810017062595322e-07,
      "loss": 0.6009,
      "step": 188
    },
    {
      "epoch": 0.32142857142857145,
      "grad_norm": 0.31316620111465454,
      "learning_rate": 8.247462563808817e-07,
      "loss": 0.8112,
      "step": 189
    },
    {
      "epoch": 0.3231292517006803,
      "grad_norm": 0.2805158197879791,
      "learning_rate": 6.819348298638839e-07,
      "loss": 0.7023,
      "step": 190
    },
    {
      "epoch": 0.32482993197278914,
      "grad_norm": 0.3001147210597992,
      "learning_rate": 5.526064699265753e-07,
      "loss": 0.7391,
      "step": 191
    },
    {
      "epoch": 0.32653061224489793,
      "grad_norm": 0.26582324504852295,
      "learning_rate": 4.367965336512403e-07,
      "loss": 0.6636,
      "step": 192
    },
    {
      "epoch": 0.3282312925170068,
      "grad_norm": 0.2941264510154724,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 0.7742,
      "step": 193
    },
    {
      "epoch": 0.3299319727891156,
      "grad_norm": 0.28889885544776917,
      "learning_rate": 2.458548727494292e-07,
      "loss": 0.6909,
      "step": 194
    },
    {
      "epoch": 0.33163265306122447,
      "grad_norm": 0.3120863437652588,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 0.6406,
      "step": 195
    },
    {
      "epoch": 0.3333333333333333,
      "grad_norm": 0.33334165811538696,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 0.731,
      "step": 196
    },
    {
      "epoch": 0.33503401360544216,
      "grad_norm": 0.3277623653411865,
      "learning_rate": 6.150154258476315e-08,
      "loss": 0.687,
      "step": 197
    },
    {
      "epoch": 0.336734693877551,
      "grad_norm": 0.351300984621048,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 0.5749,
      "step": 198
    },
    {
      "epoch": 0.33843537414965985,
      "grad_norm": 0.4301164150238037,
      "learning_rate": 6.834750376549792e-09,
      "loss": 0.6779,
      "step": 199
    },
    {
      "epoch": 0.3401360544217687,
      "grad_norm": 0.4669879078865051,
      "learning_rate": 0.0,
      "loss": 0.6887,
      "step": 200
    },
    {
      "epoch": 0.3401360544217687,
      "eval_loss": 0.6898359656333923,
      "eval_runtime": 127.4382,
      "eval_samples_per_second": 7.768,
      "eval_steps_per_second": 1.946,
      "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": 4.823093889466368e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}