{
  "best_metric": 2.142829656600952,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.2946593001841621,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0014732965009208103,
      "grad_norm": 0.6813496947288513,
      "learning_rate": 1e-05,
      "loss": 2.5563,
      "step": 1
    },
    {
      "epoch": 0.0014732965009208103,
      "eval_loss": 2.918741226196289,
      "eval_runtime": 33.5541,
      "eval_samples_per_second": 34.094,
      "eval_steps_per_second": 8.524,
      "step": 1
    },
    {
      "epoch": 0.0029465930018416206,
      "grad_norm": 0.8075690269470215,
      "learning_rate": 2e-05,
      "loss": 2.5972,
      "step": 2
    },
    {
      "epoch": 0.004419889502762431,
      "grad_norm": 0.8233782649040222,
      "learning_rate": 3e-05,
      "loss": 2.4961,
      "step": 3
    },
    {
      "epoch": 0.005893186003683241,
      "grad_norm": 0.8552287817001343,
      "learning_rate": 4e-05,
      "loss": 2.6314,
      "step": 4
    },
    {
      "epoch": 0.007366482504604052,
      "grad_norm": 0.8532387018203735,
      "learning_rate": 5e-05,
      "loss": 2.6159,
      "step": 5
    },
    {
      "epoch": 0.008839779005524863,
      "grad_norm": 0.773468554019928,
      "learning_rate": 6e-05,
      "loss": 2.5155,
      "step": 6
    },
    {
      "epoch": 0.010313075506445672,
      "grad_norm": 0.7581596374511719,
      "learning_rate": 7e-05,
      "loss": 2.4491,
      "step": 7
    },
    {
      "epoch": 0.011786372007366482,
      "grad_norm": 0.7836837768554688,
      "learning_rate": 8e-05,
      "loss": 2.5402,
      "step": 8
    },
    {
      "epoch": 0.013259668508287293,
      "grad_norm": 0.7612143158912659,
      "learning_rate": 9e-05,
      "loss": 2.3494,
      "step": 9
    },
    {
      "epoch": 0.014732965009208104,
      "grad_norm": 0.7859139442443848,
      "learning_rate": 0.0001,
      "loss": 2.3823,
      "step": 10
    },
    {
      "epoch": 0.016206261510128914,
      "grad_norm": 0.9450372457504272,
      "learning_rate": 9.999316524962345e-05,
      "loss": 2.3155,
      "step": 11
    },
    {
      "epoch": 0.017679558011049725,
      "grad_norm": 1.120526909828186,
      "learning_rate": 9.997266286704631e-05,
      "loss": 2.323,
      "step": 12
    },
    {
      "epoch": 0.019152854511970532,
      "grad_norm": 0.7395462989807129,
      "learning_rate": 9.993849845741524e-05,
      "loss": 2.3224,
      "step": 13
    },
    {
      "epoch": 0.020626151012891343,
      "grad_norm": 0.9541445374488831,
      "learning_rate": 9.989068136093873e-05,
      "loss": 2.2665,
      "step": 14
    },
    {
      "epoch": 0.022099447513812154,
      "grad_norm": 0.8041669130325317,
      "learning_rate": 9.98292246503335e-05,
      "loss": 2.4497,
      "step": 15
    },
    {
      "epoch": 0.023572744014732964,
      "grad_norm": 0.6719281077384949,
      "learning_rate": 9.975414512725057e-05,
      "loss": 2.4481,
      "step": 16
    },
    {
      "epoch": 0.025046040515653775,
      "grad_norm": 0.6882573366165161,
      "learning_rate": 9.966546331768191e-05,
      "loss": 2.4399,
      "step": 17
    },
    {
      "epoch": 0.026519337016574586,
      "grad_norm": 0.6736729145050049,
      "learning_rate": 9.956320346634876e-05,
      "loss": 2.2692,
      "step": 18
    },
    {
      "epoch": 0.027992633517495397,
      "grad_norm": 0.6062433123588562,
      "learning_rate": 9.944739353007344e-05,
      "loss": 2.3338,
      "step": 19
    },
    {
      "epoch": 0.029465930018416207,
      "grad_norm": 0.6055423617362976,
      "learning_rate": 9.931806517013612e-05,
      "loss": 2.2801,
      "step": 20
    },
    {
      "epoch": 0.030939226519337018,
      "grad_norm": 0.6778382658958435,
      "learning_rate": 9.917525374361912e-05,
      "loss": 2.3912,
      "step": 21
    },
    {
      "epoch": 0.03241252302025783,
      "grad_norm": 0.628765344619751,
      "learning_rate": 9.901899829374047e-05,
      "loss": 2.3412,
      "step": 22
    },
    {
      "epoch": 0.03388581952117864,
      "grad_norm": 0.6677904725074768,
      "learning_rate": 9.884934153917997e-05,
      "loss": 2.3754,
      "step": 23
    },
    {
      "epoch": 0.03535911602209945,
      "grad_norm": 0.6703566312789917,
      "learning_rate": 9.86663298624003e-05,
      "loss": 2.3854,
      "step": 24
    },
    {
      "epoch": 0.03683241252302026,
      "grad_norm": 0.6443681716918945,
      "learning_rate": 9.847001329696653e-05,
      "loss": 2.5166,
      "step": 25
    },
    {
      "epoch": 0.038305709023941065,
      "grad_norm": 0.6533526182174683,
      "learning_rate": 9.826044551386744e-05,
      "loss": 2.4067,
      "step": 26
    },
    {
      "epoch": 0.039779005524861875,
      "grad_norm": 0.6256962418556213,
      "learning_rate": 9.803768380684242e-05,
      "loss": 2.2705,
      "step": 27
    },
    {
      "epoch": 0.041252302025782686,
      "grad_norm": 0.6230438351631165,
      "learning_rate": 9.780178907671789e-05,
      "loss": 2.3179,
      "step": 28
    },
    {
      "epoch": 0.0427255985267035,
      "grad_norm": 0.6809419393539429,
      "learning_rate": 9.755282581475769e-05,
      "loss": 2.4417,
      "step": 29
    },
    {
      "epoch": 0.04419889502762431,
      "grad_norm": 0.6697230935096741,
      "learning_rate": 9.729086208503174e-05,
      "loss": 2.4118,
      "step": 30
    },
    {
      "epoch": 0.04567219152854512,
      "grad_norm": 0.6355602145195007,
      "learning_rate": 9.701596950580806e-05,
      "loss": 2.2859,
      "step": 31
    },
    {
      "epoch": 0.04714548802946593,
      "grad_norm": 0.6855126023292542,
      "learning_rate": 9.672822322997305e-05,
      "loss": 2.3763,
      "step": 32
    },
    {
      "epoch": 0.04861878453038674,
      "grad_norm": 0.6802644729614258,
      "learning_rate": 9.642770192448536e-05,
      "loss": 2.4461,
      "step": 33
    },
    {
      "epoch": 0.05009208103130755,
      "grad_norm": 0.6710078120231628,
      "learning_rate": 9.611448774886924e-05,
      "loss": 2.4494,
      "step": 34
    },
    {
      "epoch": 0.05156537753222836,
      "grad_norm": 0.6675485372543335,
      "learning_rate": 9.578866633275288e-05,
      "loss": 2.3493,
      "step": 35
    },
    {
      "epoch": 0.05303867403314917,
      "grad_norm": 0.6746523380279541,
      "learning_rate": 9.545032675245813e-05,
      "loss": 2.3894,
      "step": 36
    },
    {
      "epoch": 0.05451197053406998,
      "grad_norm": 0.6553740501403809,
      "learning_rate": 9.509956150664796e-05,
      "loss": 2.1722,
      "step": 37
    },
    {
      "epoch": 0.05598526703499079,
      "grad_norm": 0.691763699054718,
      "learning_rate": 9.473646649103818e-05,
      "loss": 2.4571,
      "step": 38
    },
    {
      "epoch": 0.057458563535911604,
      "grad_norm": 0.6961422562599182,
      "learning_rate": 9.43611409721806e-05,
      "loss": 2.2901,
      "step": 39
    },
    {
      "epoch": 0.058931860036832415,
      "grad_norm": 0.6852315664291382,
      "learning_rate": 9.397368756032445e-05,
      "loss": 2.232,
      "step": 40
    },
    {
      "epoch": 0.060405156537753225,
      "grad_norm": 0.774763822555542,
      "learning_rate": 9.357421218136386e-05,
      "loss": 2.3699,
      "step": 41
    },
    {
      "epoch": 0.061878453038674036,
      "grad_norm": 0.701396107673645,
      "learning_rate": 9.316282404787871e-05,
      "loss": 2.3449,
      "step": 42
    },
    {
      "epoch": 0.06335174953959484,
      "grad_norm": 0.770219624042511,
      "learning_rate": 9.273963562927695e-05,
      "loss": 2.4661,
      "step": 43
    },
    {
      "epoch": 0.06482504604051566,
      "grad_norm": 0.7467716932296753,
      "learning_rate": 9.230476262104677e-05,
      "loss": 2.326,
      "step": 44
    },
    {
      "epoch": 0.06629834254143646,
      "grad_norm": 0.7156895995140076,
      "learning_rate": 9.185832391312644e-05,
      "loss": 2.2128,
      "step": 45
    },
    {
      "epoch": 0.06777163904235728,
      "grad_norm": 0.7966296076774597,
      "learning_rate": 9.140044155740101e-05,
      "loss": 2.2482,
      "step": 46
    },
    {
      "epoch": 0.06924493554327808,
      "grad_norm": 0.8367739915847778,
      "learning_rate": 9.093124073433463e-05,
      "loss": 2.4101,
      "step": 47
    },
    {
      "epoch": 0.0707182320441989,
      "grad_norm": 0.8258773684501648,
      "learning_rate": 9.045084971874738e-05,
      "loss": 2.1884,
      "step": 48
    },
    {
      "epoch": 0.0721915285451197,
      "grad_norm": 0.9507535099983215,
      "learning_rate": 8.995939984474624e-05,
      "loss": 2.3239,
      "step": 49
    },
    {
      "epoch": 0.07366482504604052,
      "grad_norm": 1.0835574865341187,
      "learning_rate": 8.945702546981969e-05,
      "loss": 2.2306,
      "step": 50
    },
    {
      "epoch": 0.07366482504604052,
      "eval_loss": 2.2473981380462646,
      "eval_runtime": 33.6255,
      "eval_samples_per_second": 34.022,
      "eval_steps_per_second": 8.505,
      "step": 50
    },
    {
      "epoch": 0.07513812154696133,
      "grad_norm": 1.010612964630127,
      "learning_rate": 8.894386393810563e-05,
      "loss": 2.1945,
      "step": 51
    },
    {
      "epoch": 0.07661141804788213,
      "grad_norm": 0.7759504914283752,
      "learning_rate": 8.842005554284296e-05,
      "loss": 2.2003,
      "step": 52
    },
    {
      "epoch": 0.07808471454880295,
      "grad_norm": 0.6330819129943848,
      "learning_rate": 8.788574348801675e-05,
      "loss": 2.1249,
      "step": 53
    },
    {
      "epoch": 0.07955801104972375,
      "grad_norm": 0.527708888053894,
      "learning_rate": 8.73410738492077e-05,
      "loss": 2.0861,
      "step": 54
    },
    {
      "epoch": 0.08103130755064457,
      "grad_norm": 0.48775210976600647,
      "learning_rate": 8.678619553365659e-05,
      "loss": 2.1054,
      "step": 55
    },
    {
      "epoch": 0.08250460405156537,
      "grad_norm": 0.4980153739452362,
      "learning_rate": 8.622126023955446e-05,
      "loss": 2.0805,
      "step": 56
    },
    {
      "epoch": 0.08397790055248619,
      "grad_norm": 0.5115144848823547,
      "learning_rate": 8.564642241456986e-05,
      "loss": 2.2559,
      "step": 57
    },
    {
      "epoch": 0.085451197053407,
      "grad_norm": 0.5000446438789368,
      "learning_rate": 8.506183921362443e-05,
      "loss": 2.0434,
      "step": 58
    },
    {
      "epoch": 0.08692449355432781,
      "grad_norm": 0.5144779682159424,
      "learning_rate": 8.44676704559283e-05,
      "loss": 2.0314,
      "step": 59
    },
    {
      "epoch": 0.08839779005524862,
      "grad_norm": 0.5057157874107361,
      "learning_rate": 8.386407858128706e-05,
      "loss": 2.1697,
      "step": 60
    },
    {
      "epoch": 0.08987108655616943,
      "grad_norm": 0.5526983141899109,
      "learning_rate": 8.32512286056924e-05,
      "loss": 2.1934,
      "step": 61
    },
    {
      "epoch": 0.09134438305709024,
      "grad_norm": 0.5022704601287842,
      "learning_rate": 8.262928807620843e-05,
      "loss": 2.1775,
      "step": 62
    },
    {
      "epoch": 0.09281767955801105,
      "grad_norm": 0.519891083240509,
      "learning_rate": 8.199842702516583e-05,
      "loss": 2.2059,
      "step": 63
    },
    {
      "epoch": 0.09429097605893186,
      "grad_norm": 0.552194356918335,
      "learning_rate": 8.135881792367686e-05,
      "loss": 2.1553,
      "step": 64
    },
    {
      "epoch": 0.09576427255985268,
      "grad_norm": 0.5596328973770142,
      "learning_rate": 8.07106356344834e-05,
      "loss": 2.2244,
      "step": 65
    },
    {
      "epoch": 0.09723756906077348,
      "grad_norm": 0.5474238991737366,
      "learning_rate": 8.005405736415126e-05,
      "loss": 2.3101,
      "step": 66
    },
    {
      "epoch": 0.0987108655616943,
      "grad_norm": 0.5688091516494751,
      "learning_rate": 7.938926261462366e-05,
      "loss": 2.1229,
      "step": 67
    },
    {
      "epoch": 0.1001841620626151,
      "grad_norm": 0.5471810102462769,
      "learning_rate": 7.871643313414718e-05,
      "loss": 2.1483,
      "step": 68
    },
    {
      "epoch": 0.10165745856353592,
      "grad_norm": 0.5143526792526245,
      "learning_rate": 7.803575286758364e-05,
      "loss": 2.1053,
      "step": 69
    },
    {
      "epoch": 0.10313075506445672,
      "grad_norm": 0.5542533993721008,
      "learning_rate": 7.734740790612136e-05,
      "loss": 2.2268,
      "step": 70
    },
    {
      "epoch": 0.10460405156537753,
      "grad_norm": 0.5456425547599792,
      "learning_rate": 7.66515864363997e-05,
      "loss": 2.1713,
      "step": 71
    },
    {
      "epoch": 0.10607734806629834,
      "grad_norm": 0.5739713907241821,
      "learning_rate": 7.594847868906076e-05,
      "loss": 2.245,
      "step": 72
    },
    {
      "epoch": 0.10755064456721915,
      "grad_norm": 0.550366222858429,
      "learning_rate": 7.52382768867422e-05,
      "loss": 2.2492,
      "step": 73
    },
    {
      "epoch": 0.10902394106813997,
      "grad_norm": 0.5844371914863586,
      "learning_rate": 7.452117519152542e-05,
      "loss": 2.3137,
      "step": 74
    },
    {
      "epoch": 0.11049723756906077,
      "grad_norm": 0.5785281658172607,
      "learning_rate": 7.379736965185368e-05,
      "loss": 2.2594,
      "step": 75
    },
    {
      "epoch": 0.11197053406998159,
      "grad_norm": 0.5865864157676697,
      "learning_rate": 7.30670581489344e-05,
      "loss": 2.2213,
      "step": 76
    },
    {
      "epoch": 0.11344383057090239,
      "grad_norm": 0.5580472350120544,
      "learning_rate": 7.233044034264034e-05,
      "loss": 2.2876,
      "step": 77
    },
    {
      "epoch": 0.11491712707182321,
      "grad_norm": 0.6060166358947754,
      "learning_rate": 7.158771761692464e-05,
      "loss": 2.3195,
      "step": 78
    },
    {
      "epoch": 0.11639042357274401,
      "grad_norm": 0.5804003477096558,
      "learning_rate": 7.083909302476453e-05,
      "loss": 2.2484,
      "step": 79
    },
    {
      "epoch": 0.11786372007366483,
      "grad_norm": 0.5879073739051819,
      "learning_rate": 7.008477123264848e-05,
      "loss": 2.1609,
      "step": 80
    },
    {
      "epoch": 0.11933701657458563,
      "grad_norm": 0.5739030838012695,
      "learning_rate": 6.932495846462261e-05,
      "loss": 2.1838,
      "step": 81
    },
    {
      "epoch": 0.12081031307550645,
      "grad_norm": 0.6036490797996521,
      "learning_rate": 6.855986244591104e-05,
      "loss": 2.1805,
      "step": 82
    },
    {
      "epoch": 0.12228360957642725,
      "grad_norm": 0.6051073670387268,
      "learning_rate": 6.778969234612584e-05,
      "loss": 2.2435,
      "step": 83
    },
    {
      "epoch": 0.12375690607734807,
      "grad_norm": 0.5920565724372864,
      "learning_rate": 6.701465872208216e-05,
      "loss": 2.14,
      "step": 84
    },
    {
      "epoch": 0.1252302025782689,
      "grad_norm": 0.6635783314704895,
      "learning_rate": 6.623497346023418e-05,
      "loss": 2.2595,
      "step": 85
    },
    {
      "epoch": 0.12670349907918968,
      "grad_norm": 0.6474425196647644,
      "learning_rate": 6.545084971874738e-05,
      "loss": 2.21,
      "step": 86
    },
    {
      "epoch": 0.1281767955801105,
      "grad_norm": 0.6599803566932678,
      "learning_rate": 6.466250186922325e-05,
      "loss": 2.1636,
      "step": 87
    },
    {
      "epoch": 0.12965009208103132,
      "grad_norm": 0.6417123079299927,
      "learning_rate": 6.387014543809223e-05,
      "loss": 2.2529,
      "step": 88
    },
    {
      "epoch": 0.1311233885819521,
      "grad_norm": 0.6307031512260437,
      "learning_rate": 6.307399704769099e-05,
      "loss": 2.2475,
      "step": 89
    },
    {
      "epoch": 0.13259668508287292,
      "grad_norm": 0.6963887810707092,
      "learning_rate": 6.227427435703997e-05,
      "loss": 2.3434,
      "step": 90
    },
    {
      "epoch": 0.13406998158379374,
      "grad_norm": 0.7633985877037048,
      "learning_rate": 6.147119600233758e-05,
      "loss": 2.3564,
      "step": 91
    },
    {
      "epoch": 0.13554327808471456,
      "grad_norm": 0.7439932823181152,
      "learning_rate": 6.066498153718735e-05,
      "loss": 2.3207,
      "step": 92
    },
    {
      "epoch": 0.13701657458563535,
      "grad_norm": 0.6978381872177124,
      "learning_rate": 5.985585137257401e-05,
      "loss": 2.1644,
      "step": 93
    },
    {
      "epoch": 0.13848987108655617,
      "grad_norm": 0.790519118309021,
      "learning_rate": 5.90440267166055e-05,
      "loss": 2.165,
      "step": 94
    },
    {
      "epoch": 0.13996316758747698,
      "grad_norm": 0.8105412721633911,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 2.4732,
      "step": 95
    },
    {
      "epoch": 0.1414364640883978,
      "grad_norm": 0.7491436004638672,
      "learning_rate": 5.74131823855921e-05,
      "loss": 2.2654,
      "step": 96
    },
    {
      "epoch": 0.1429097605893186,
      "grad_norm": 0.755374014377594,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 2.3541,
      "step": 97
    },
    {
      "epoch": 0.1443830570902394,
      "grad_norm": 0.8364269733428955,
      "learning_rate": 5.577423184847932e-05,
      "loss": 2.2961,
      "step": 98
    },
    {
      "epoch": 0.14585635359116023,
      "grad_norm": 0.8708286881446838,
      "learning_rate": 5.495227651252315e-05,
      "loss": 2.1471,
      "step": 99
    },
    {
      "epoch": 0.14732965009208104,
      "grad_norm": 1.0091307163238525,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 2.0601,
      "step": 100
    },
    {
      "epoch": 0.14732965009208104,
      "eval_loss": 2.177105188369751,
      "eval_runtime": 33.6215,
      "eval_samples_per_second": 34.026,
      "eval_steps_per_second": 8.506,
      "step": 100
    },
    {
      "epoch": 0.14880294659300183,
      "grad_norm": 0.5215460062026978,
      "learning_rate": 5.330452921628497e-05,
      "loss": 2.0475,
      "step": 101
    },
    {
      "epoch": 0.15027624309392265,
      "grad_norm": 0.6095430850982666,
      "learning_rate": 5.247918773366112e-05,
      "loss": 2.0592,
      "step": 102
    },
    {
      "epoch": 0.15174953959484347,
      "grad_norm": 0.5464235544204712,
      "learning_rate": 5.165316846586541e-05,
      "loss": 2.0183,
      "step": 103
    },
    {
      "epoch": 0.15322283609576426,
      "grad_norm": 0.5505194664001465,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 2.0362,
      "step": 104
    },
    {
      "epoch": 0.15469613259668508,
      "grad_norm": 0.5168222188949585,
      "learning_rate": 5e-05,
      "loss": 2.1467,
      "step": 105
    },
    {
      "epoch": 0.1561694290976059,
      "grad_norm": 0.5152841806411743,
      "learning_rate": 4.917330276168208e-05,
      "loss": 2.1926,
      "step": 106
    },
    {
      "epoch": 0.1576427255985267,
      "grad_norm": 0.49555912613868713,
      "learning_rate": 4.834683153413459e-05,
      "loss": 2.0815,
      "step": 107
    },
    {
      "epoch": 0.1591160220994475,
      "grad_norm": 0.4678058624267578,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 2.1123,
      "step": 108
    },
    {
      "epoch": 0.16058931860036832,
      "grad_norm": 0.4995936453342438,
      "learning_rate": 4.669547078371504e-05,
      "loss": 2.0178,
      "step": 109
    },
    {
      "epoch": 0.16206261510128914,
      "grad_norm": 0.4865174889564514,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 2.0839,
      "step": 110
    },
    {
      "epoch": 0.16353591160220995,
      "grad_norm": 0.46470630168914795,
      "learning_rate": 4.504772348747687e-05,
      "loss": 2.1032,
      "step": 111
    },
    {
      "epoch": 0.16500920810313074,
      "grad_norm": 0.4676339626312256,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 2.2317,
      "step": 112
    },
    {
      "epoch": 0.16648250460405156,
      "grad_norm": 0.48408418893814087,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 2.1775,
      "step": 113
    },
    {
      "epoch": 0.16795580110497238,
      "grad_norm": 0.5317076444625854,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 2.207,
      "step": 114
    },
    {
      "epoch": 0.1694290976058932,
      "grad_norm": 0.5113919973373413,
      "learning_rate": 4.17702704859633e-05,
      "loss": 2.1026,
      "step": 115
    },
    {
      "epoch": 0.170902394106814,
      "grad_norm": 0.5057461261749268,
      "learning_rate": 4.095597328339452e-05,
      "loss": 2.1784,
      "step": 116
    },
    {
      "epoch": 0.1723756906077348,
      "grad_norm": 0.5517849326133728,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 2.1082,
      "step": 117
    },
    {
      "epoch": 0.17384898710865562,
      "grad_norm": 0.5440712571144104,
      "learning_rate": 3.933501846281267e-05,
      "loss": 2.1231,
      "step": 118
    },
    {
      "epoch": 0.17532228360957644,
      "grad_norm": 0.5332390666007996,
      "learning_rate": 3.852880399766243e-05,
      "loss": 2.1617,
      "step": 119
    },
    {
      "epoch": 0.17679558011049723,
      "grad_norm": 0.5297031998634338,
      "learning_rate": 3.772572564296005e-05,
      "loss": 2.1625,
      "step": 120
    },
    {
      "epoch": 0.17826887661141805,
      "grad_norm": 0.5557818412780762,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 2.2473,
      "step": 121
    },
    {
      "epoch": 0.17974217311233887,
      "grad_norm": 0.5619638562202454,
      "learning_rate": 3.612985456190778e-05,
      "loss": 2.1909,
      "step": 122
    },
    {
      "epoch": 0.18121546961325966,
      "grad_norm": 0.5676457285881042,
      "learning_rate": 3.533749813077677e-05,
      "loss": 2.2397,
      "step": 123
    },
    {
      "epoch": 0.18268876611418047,
      "grad_norm": 0.5630333423614502,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 2.1458,
      "step": 124
    },
    {
      "epoch": 0.1841620626151013,
      "grad_norm": 0.5573937892913818,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 2.1902,
      "step": 125
    },
    {
      "epoch": 0.1856353591160221,
      "grad_norm": 0.5926021933555603,
      "learning_rate": 3.298534127791785e-05,
      "loss": 2.2485,
      "step": 126
    },
    {
      "epoch": 0.1871086556169429,
      "grad_norm": 0.5963653326034546,
      "learning_rate": 3.221030765387417e-05,
      "loss": 2.1565,
      "step": 127
    },
    {
      "epoch": 0.18858195211786372,
      "grad_norm": 0.6477805376052856,
      "learning_rate": 3.144013755408895e-05,
      "loss": 2.3577,
      "step": 128
    },
    {
      "epoch": 0.19005524861878453,
      "grad_norm": 0.5789662599563599,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 2.16,
      "step": 129
    },
    {
      "epoch": 0.19152854511970535,
      "grad_norm": 0.5775110721588135,
      "learning_rate": 2.991522876735154e-05,
      "loss": 2.1837,
      "step": 130
    },
    {
      "epoch": 0.19300184162062614,
      "grad_norm": 0.6224007606506348,
      "learning_rate": 2.916090697523549e-05,
      "loss": 2.3447,
      "step": 131
    },
    {
      "epoch": 0.19447513812154696,
      "grad_norm": 0.6071020364761353,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 2.3363,
      "step": 132
    },
    {
      "epoch": 0.19594843462246778,
      "grad_norm": 0.5970646142959595,
      "learning_rate": 2.766955965735968e-05,
      "loss": 2.2286,
      "step": 133
    },
    {
      "epoch": 0.1974217311233886,
      "grad_norm": 0.5931335091590881,
      "learning_rate": 2.693294185106562e-05,
      "loss": 2.2033,
      "step": 134
    },
    {
      "epoch": 0.19889502762430938,
      "grad_norm": 0.6456011533737183,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 2.3814,
      "step": 135
    },
    {
      "epoch": 0.2003683241252302,
      "grad_norm": 0.6193873882293701,
      "learning_rate": 2.547882480847461e-05,
      "loss": 2.1756,
      "step": 136
    },
    {
      "epoch": 0.20184162062615102,
      "grad_norm": 0.611831545829773,
      "learning_rate": 2.476172311325783e-05,
      "loss": 2.2781,
      "step": 137
    },
    {
      "epoch": 0.20331491712707184,
      "grad_norm": 0.6075538992881775,
      "learning_rate": 2.405152131093926e-05,
      "loss": 2.2085,
      "step": 138
    },
    {
      "epoch": 0.20478821362799263,
      "grad_norm": 0.6648210883140564,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 2.2603,
      "step": 139
    },
    {
      "epoch": 0.20626151012891344,
      "grad_norm": 0.6589313745498657,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 2.3493,
      "step": 140
    },
    {
      "epoch": 0.20773480662983426,
      "grad_norm": 0.6778343915939331,
      "learning_rate": 2.196424713241637e-05,
      "loss": 2.2593,
      "step": 141
    },
    {
      "epoch": 0.20920810313075505,
      "grad_norm": 0.6866799592971802,
      "learning_rate": 2.128356686585282e-05,
      "loss": 2.2085,
      "step": 142
    },
    {
      "epoch": 0.21068139963167587,
      "grad_norm": 0.6959302425384521,
      "learning_rate": 2.061073738537635e-05,
      "loss": 2.1706,
      "step": 143
    },
    {
      "epoch": 0.2121546961325967,
      "grad_norm": 0.6998744010925293,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 2.349,
      "step": 144
    },
    {
      "epoch": 0.2136279926335175,
      "grad_norm": 0.7252722382545471,
      "learning_rate": 1.928936436551661e-05,
      "loss": 2.2642,
      "step": 145
    },
    {
      "epoch": 0.2151012891344383,
      "grad_norm": 0.7717849016189575,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 2.2799,
      "step": 146
    },
    {
      "epoch": 0.2165745856353591,
      "grad_norm": 0.7438363432884216,
      "learning_rate": 1.800157297483417e-05,
      "loss": 2.1361,
      "step": 147
    },
    {
      "epoch": 0.21804788213627993,
      "grad_norm": 0.8306782245635986,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 2.2135,
      "step": 148
    },
    {
      "epoch": 0.21952117863720075,
      "grad_norm": 0.8231614828109741,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 2.228,
      "step": 149
    },
    {
      "epoch": 0.22099447513812154,
      "grad_norm": 0.8910000920295715,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 2.0673,
      "step": 150
    },
    {
      "epoch": 0.22099447513812154,
      "eval_loss": 2.1498870849609375,
      "eval_runtime": 33.6508,
      "eval_samples_per_second": 33.996,
      "eval_steps_per_second": 8.499,
      "step": 150
    },
    {
      "epoch": 0.22246777163904236,
      "grad_norm": 0.4497315287590027,
      "learning_rate": 1.553232954407171e-05,
      "loss": 2.1235,
      "step": 151
    },
    {
      "epoch": 0.22394106813996317,
      "grad_norm": 0.48219621181488037,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 2.0459,
      "step": 152
    },
    {
      "epoch": 0.225414364640884,
      "grad_norm": 0.4607875645160675,
      "learning_rate": 1.435357758543015e-05,
      "loss": 2.1236,
      "step": 153
    },
    {
      "epoch": 0.22688766114180478,
      "grad_norm": 0.4888078272342682,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 2.0777,
      "step": 154
    },
    {
      "epoch": 0.2283609576427256,
      "grad_norm": 0.5186671614646912,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 2.1288,
      "step": 155
    },
    {
      "epoch": 0.22983425414364642,
      "grad_norm": 0.5025704503059387,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 1.9971,
      "step": 156
    },
    {
      "epoch": 0.2313075506445672,
      "grad_norm": 0.4730200469493866,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 2.0245,
      "step": 157
    },
    {
      "epoch": 0.23278084714548802,
      "grad_norm": 0.5410052537918091,
      "learning_rate": 1.157994445715706e-05,
      "loss": 2.042,
      "step": 158
    },
    {
      "epoch": 0.23425414364640884,
      "grad_norm": 0.4994640648365021,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 1.9262,
      "step": 159
    },
    {
      "epoch": 0.23572744014732966,
      "grad_norm": 0.49574676156044006,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 2.1671,
      "step": 160
    },
    {
      "epoch": 0.23720073664825045,
      "grad_norm": 0.5467461943626404,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 2.248,
      "step": 161
    },
    {
      "epoch": 0.23867403314917127,
      "grad_norm": 0.5036174654960632,
      "learning_rate": 9.549150281252633e-06,
      "loss": 2.114,
      "step": 162
    },
    {
      "epoch": 0.24014732965009208,
      "grad_norm": 0.516216516494751,
      "learning_rate": 9.068759265665384e-06,
      "loss": 2.1475,
      "step": 163
    },
    {
      "epoch": 0.2416206261510129,
      "grad_norm": 0.5705137252807617,
      "learning_rate": 8.599558442598998e-06,
      "loss": 2.3152,
      "step": 164
    },
    {
      "epoch": 0.2430939226519337,
      "grad_norm": 0.5074183940887451,
      "learning_rate": 8.141676086873572e-06,
      "loss": 2.0175,
      "step": 165
    },
    {
      "epoch": 0.2445672191528545,
      "grad_norm": 0.5196369290351868,
      "learning_rate": 7.695237378953223e-06,
      "loss": 2.0298,
      "step": 166
    },
    {
      "epoch": 0.24604051565377533,
      "grad_norm": 0.5080201625823975,
      "learning_rate": 7.260364370723044e-06,
      "loss": 2.1187,
      "step": 167
    },
    {
      "epoch": 0.24751381215469614,
      "grad_norm": 0.5384064316749573,
      "learning_rate": 6.837175952121306e-06,
      "loss": 2.1803,
      "step": 168
    },
    {
      "epoch": 0.24898710865561693,
      "grad_norm": 0.5691060423851013,
      "learning_rate": 6.425787818636131e-06,
      "loss": 2.0255,
      "step": 169
    },
    {
      "epoch": 0.2504604051565378,
      "grad_norm": 0.550330638885498,
      "learning_rate": 6.026312439675552e-06,
      "loss": 2.1137,
      "step": 170
    },
    {
      "epoch": 0.25193370165745854,
      "grad_norm": 0.5306848883628845,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 2.1299,
      "step": 171
    },
    {
      "epoch": 0.25340699815837936,
      "grad_norm": 0.5351966619491577,
      "learning_rate": 5.263533508961827e-06,
      "loss": 2.1293,
      "step": 172
    },
    {
      "epoch": 0.2548802946593002,
      "grad_norm": 0.5528244972229004,
      "learning_rate": 4.900438493352055e-06,
      "loss": 2.1027,
      "step": 173
    },
    {
      "epoch": 0.256353591160221,
      "grad_norm": 0.5488362312316895,
      "learning_rate": 4.549673247541875e-06,
      "loss": 2.1509,
      "step": 174
    },
    {
      "epoch": 0.2578268876611418,
      "grad_norm": 0.5732300877571106,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 2.2063,
      "step": 175
    },
    {
      "epoch": 0.25930018416206263,
      "grad_norm": 0.5619857907295227,
      "learning_rate": 3.885512251130763e-06,
      "loss": 2.2253,
      "step": 176
    },
    {
      "epoch": 0.26077348066298345,
      "grad_norm": 0.5769329071044922,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 2.2133,
      "step": 177
    },
    {
      "epoch": 0.2622467771639042,
      "grad_norm": 0.5814343690872192,
      "learning_rate": 3.271776770026963e-06,
      "loss": 2.154,
      "step": 178
    },
    {
      "epoch": 0.263720073664825,
      "grad_norm": 0.6259899139404297,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 2.3186,
      "step": 179
    },
    {
      "epoch": 0.26519337016574585,
      "grad_norm": 0.5869902968406677,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 2.2518,
      "step": 180
    },
    {
      "epoch": 0.26666666666666666,
      "grad_norm": 0.6089134216308594,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 2.0913,
      "step": 181
    },
    {
      "epoch": 0.2681399631675875,
      "grad_norm": 0.5955935716629028,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 2.1964,
      "step": 182
    },
    {
      "epoch": 0.2696132596685083,
      "grad_norm": 0.5801297426223755,
      "learning_rate": 1.962316193157593e-06,
      "loss": 2.128,
      "step": 183
    },
    {
      "epoch": 0.2710865561694291,
      "grad_norm": 0.606168270111084,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 2.248,
      "step": 184
    },
    {
      "epoch": 0.27255985267034993,
      "grad_norm": 0.628900945186615,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 2.0947,
      "step": 185
    },
    {
      "epoch": 0.2740331491712707,
      "grad_norm": 0.6443435549736023,
      "learning_rate": 1.333670137599713e-06,
      "loss": 2.1588,
      "step": 186
    },
    {
      "epoch": 0.2755064456721915,
      "grad_norm": 0.6042181849479675,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 2.0953,
      "step": 187
    },
    {
      "epoch": 0.27697974217311233,
      "grad_norm": 0.6067960858345032,
      "learning_rate": 9.810017062595322e-07,
      "loss": 2.2178,
      "step": 188
    },
    {
      "epoch": 0.27845303867403315,
      "grad_norm": 0.6070239543914795,
      "learning_rate": 8.247462563808817e-07,
      "loss": 2.2549,
      "step": 189
    },
    {
      "epoch": 0.27992633517495397,
      "grad_norm": 0.6234919428825378,
      "learning_rate": 6.819348298638839e-07,
      "loss": 2.1084,
      "step": 190
    },
    {
      "epoch": 0.2813996316758748,
      "grad_norm": 0.6733989715576172,
      "learning_rate": 5.526064699265753e-07,
      "loss": 2.0763,
      "step": 191
    },
    {
      "epoch": 0.2828729281767956,
      "grad_norm": 0.7105913758277893,
      "learning_rate": 4.367965336512403e-07,
      "loss": 2.2615,
      "step": 192
    },
    {
      "epoch": 0.28434622467771636,
      "grad_norm": 0.718129575252533,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 2.3094,
      "step": 193
    },
    {
      "epoch": 0.2858195211786372,
      "grad_norm": 0.6917638182640076,
      "learning_rate": 2.458548727494292e-07,
      "loss": 2.1456,
      "step": 194
    },
    {
      "epoch": 0.287292817679558,
      "grad_norm": 0.7489725351333618,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 2.1397,
      "step": 195
    },
    {
      "epoch": 0.2887661141804788,
      "grad_norm": 0.7198150157928467,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 2.1637,
      "step": 196
    },
    {
      "epoch": 0.29023941068139963,
      "grad_norm": 0.758560299873352,
      "learning_rate": 6.150154258476315e-08,
      "loss": 2.221,
      "step": 197
    },
    {
      "epoch": 0.29171270718232045,
      "grad_norm": 0.764299750328064,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 2.2446,
      "step": 198
    },
    {
      "epoch": 0.29318600368324127,
      "grad_norm": 0.8316512703895569,
      "learning_rate": 6.834750376549792e-09,
      "loss": 2.1687,
      "step": 199
    },
    {
      "epoch": 0.2946593001841621,
      "grad_norm": 1.0795071125030518,
      "learning_rate": 0.0,
      "loss": 2.0839,
      "step": 200
    },
    {
      "epoch": 0.2946593001841621,
      "eval_loss": 2.142829656600952,
      "eval_runtime": 33.6479,
      "eval_samples_per_second": 33.999,
      "eval_steps_per_second": 8.5,
      "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": 5.456501850243072e+16,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}