{
  "best_metric": 0.3507079780101776,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 2.2346368715083798,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0111731843575419,
      "grad_norm": 0.6019284725189209,
      "learning_rate": 1e-05,
      "loss": 0.9769,
      "step": 1
    },
    {
      "epoch": 0.0111731843575419,
      "eval_loss": 1.3166334629058838,
      "eval_runtime": 7.455,
      "eval_samples_per_second": 20.255,
      "eval_steps_per_second": 5.097,
      "step": 1
    },
    {
      "epoch": 0.0223463687150838,
      "grad_norm": 1.4022631645202637,
      "learning_rate": 2e-05,
      "loss": 1.2317,
      "step": 2
    },
    {
      "epoch": 0.0335195530726257,
      "grad_norm": 2.1945931911468506,
      "learning_rate": 3e-05,
      "loss": 1.1797,
      "step": 3
    },
    {
      "epoch": 0.0446927374301676,
      "grad_norm": 2.0695369243621826,
      "learning_rate": 4e-05,
      "loss": 1.2777,
      "step": 4
    },
    {
      "epoch": 0.055865921787709494,
      "grad_norm": 1.4945690631866455,
      "learning_rate": 5e-05,
      "loss": 1.1935,
      "step": 5
    },
    {
      "epoch": 0.0670391061452514,
      "grad_norm": 1.517856478691101,
      "learning_rate": 6e-05,
      "loss": 1.1827,
      "step": 6
    },
    {
      "epoch": 0.0782122905027933,
      "grad_norm": 1.3563342094421387,
      "learning_rate": 7e-05,
      "loss": 1.1087,
      "step": 7
    },
    {
      "epoch": 0.0893854748603352,
      "grad_norm": 1.3807897567749023,
      "learning_rate": 8e-05,
      "loss": 1.0623,
      "step": 8
    },
    {
      "epoch": 0.1005586592178771,
      "grad_norm": 1.1294442415237427,
      "learning_rate": 9e-05,
      "loss": 0.7778,
      "step": 9
    },
    {
      "epoch": 0.11173184357541899,
      "grad_norm": 1.1513280868530273,
      "learning_rate": 0.0001,
      "loss": 0.7984,
      "step": 10
    },
    {
      "epoch": 0.12290502793296089,
      "grad_norm": 1.1094616651535034,
      "learning_rate": 9.999316524962345e-05,
      "loss": 0.6654,
      "step": 11
    },
    {
      "epoch": 0.1340782122905028,
      "grad_norm": 1.8275506496429443,
      "learning_rate": 9.997266286704631e-05,
      "loss": 0.6868,
      "step": 12
    },
    {
      "epoch": 0.1452513966480447,
      "grad_norm": 2.272935628890991,
      "learning_rate": 9.993849845741524e-05,
      "loss": 0.699,
      "step": 13
    },
    {
      "epoch": 0.1564245810055866,
      "grad_norm": 1.5463398694992065,
      "learning_rate": 9.989068136093873e-05,
      "loss": 0.7178,
      "step": 14
    },
    {
      "epoch": 0.16759776536312848,
      "grad_norm": 1.9218133687973022,
      "learning_rate": 9.98292246503335e-05,
      "loss": 0.6616,
      "step": 15
    },
    {
      "epoch": 0.1787709497206704,
      "grad_norm": 1.970350980758667,
      "learning_rate": 9.975414512725057e-05,
      "loss": 0.5954,
      "step": 16
    },
    {
      "epoch": 0.18994413407821228,
      "grad_norm": 1.5509337186813354,
      "learning_rate": 9.966546331768191e-05,
      "loss": 0.375,
      "step": 17
    },
    {
      "epoch": 0.2011173184357542,
      "grad_norm": 4.704031467437744,
      "learning_rate": 9.956320346634876e-05,
      "loss": 0.587,
      "step": 18
    },
    {
      "epoch": 0.2122905027932961,
      "grad_norm": 1.4989049434661865,
      "learning_rate": 9.944739353007344e-05,
      "loss": 0.2974,
      "step": 19
    },
    {
      "epoch": 0.22346368715083798,
      "grad_norm": 1.6060658693313599,
      "learning_rate": 9.931806517013612e-05,
      "loss": 0.3291,
      "step": 20
    },
    {
      "epoch": 0.2346368715083799,
      "grad_norm": 0.9910858869552612,
      "learning_rate": 9.917525374361912e-05,
      "loss": 0.2913,
      "step": 21
    },
    {
      "epoch": 0.24581005586592178,
      "grad_norm": 0.7862236499786377,
      "learning_rate": 9.901899829374047e-05,
      "loss": 0.2667,
      "step": 22
    },
    {
      "epoch": 0.2569832402234637,
      "grad_norm": 1.6216517686843872,
      "learning_rate": 9.884934153917997e-05,
      "loss": 1.1384,
      "step": 23
    },
    {
      "epoch": 0.2681564245810056,
      "grad_norm": 0.8437603712081909,
      "learning_rate": 9.86663298624003e-05,
      "loss": 0.6747,
      "step": 24
    },
    {
      "epoch": 0.27932960893854747,
      "grad_norm": 0.997508704662323,
      "learning_rate": 9.847001329696653e-05,
      "loss": 0.714,
      "step": 25
    },
    {
      "epoch": 0.2905027932960894,
      "grad_norm": 0.7954604029655457,
      "learning_rate": 9.826044551386744e-05,
      "loss": 0.5472,
      "step": 26
    },
    {
      "epoch": 0.3016759776536313,
      "grad_norm": 0.6894015073776245,
      "learning_rate": 9.803768380684242e-05,
      "loss": 0.5464,
      "step": 27
    },
    {
      "epoch": 0.3128491620111732,
      "grad_norm": 0.749237596988678,
      "learning_rate": 9.780178907671789e-05,
      "loss": 0.597,
      "step": 28
    },
    {
      "epoch": 0.3240223463687151,
      "grad_norm": 0.8058210015296936,
      "learning_rate": 9.755282581475769e-05,
      "loss": 0.5662,
      "step": 29
    },
    {
      "epoch": 0.33519553072625696,
      "grad_norm": 0.8081727623939514,
      "learning_rate": 9.729086208503174e-05,
      "loss": 0.4091,
      "step": 30
    },
    {
      "epoch": 0.3463687150837989,
      "grad_norm": 0.8079054355621338,
      "learning_rate": 9.701596950580806e-05,
      "loss": 0.5468,
      "step": 31
    },
    {
      "epoch": 0.3575418994413408,
      "grad_norm": 0.9135511517524719,
      "learning_rate": 9.672822322997305e-05,
      "loss": 0.6132,
      "step": 32
    },
    {
      "epoch": 0.3687150837988827,
      "grad_norm": 0.9923297166824341,
      "learning_rate": 9.642770192448536e-05,
      "loss": 0.4688,
      "step": 33
    },
    {
      "epoch": 0.37988826815642457,
      "grad_norm": 0.922411322593689,
      "learning_rate": 9.611448774886924e-05,
      "loss": 0.4675,
      "step": 34
    },
    {
      "epoch": 0.39106145251396646,
      "grad_norm": 0.7654443383216858,
      "learning_rate": 9.578866633275288e-05,
      "loss": 0.3799,
      "step": 35
    },
    {
      "epoch": 0.4022346368715084,
      "grad_norm": 0.9367057085037231,
      "learning_rate": 9.545032675245813e-05,
      "loss": 0.4774,
      "step": 36
    },
    {
      "epoch": 0.4134078212290503,
      "grad_norm": 0.802885890007019,
      "learning_rate": 9.509956150664796e-05,
      "loss": 0.3753,
      "step": 37
    },
    {
      "epoch": 0.4245810055865922,
      "grad_norm": 0.7531136274337769,
      "learning_rate": 9.473646649103818e-05,
      "loss": 0.2679,
      "step": 38
    },
    {
      "epoch": 0.43575418994413406,
      "grad_norm": 1.2379066944122314,
      "learning_rate": 9.43611409721806e-05,
      "loss": 0.2816,
      "step": 39
    },
    {
      "epoch": 0.44692737430167595,
      "grad_norm": 0.7029897570610046,
      "learning_rate": 9.397368756032445e-05,
      "loss": 0.2044,
      "step": 40
    },
    {
      "epoch": 0.4581005586592179,
      "grad_norm": 1.0064433813095093,
      "learning_rate": 9.357421218136386e-05,
      "loss": 0.3445,
      "step": 41
    },
    {
      "epoch": 0.4692737430167598,
      "grad_norm": 0.5436912178993225,
      "learning_rate": 9.316282404787871e-05,
      "loss": 0.173,
      "step": 42
    },
    {
      "epoch": 0.48044692737430167,
      "grad_norm": 0.7081115245819092,
      "learning_rate": 9.273963562927695e-05,
      "loss": 0.2727,
      "step": 43
    },
    {
      "epoch": 0.49162011173184356,
      "grad_norm": 0.6203745603561401,
      "learning_rate": 9.230476262104677e-05,
      "loss": 0.1728,
      "step": 44
    },
    {
      "epoch": 0.5027932960893855,
      "grad_norm": 1.820703148841858,
      "learning_rate": 9.185832391312644e-05,
      "loss": 1.1963,
      "step": 45
    },
    {
      "epoch": 0.5139664804469274,
      "grad_norm": 0.7371804118156433,
      "learning_rate": 9.140044155740101e-05,
      "loss": 0.6524,
      "step": 46
    },
    {
      "epoch": 0.5251396648044693,
      "grad_norm": 0.787708044052124,
      "learning_rate": 9.093124073433463e-05,
      "loss": 0.623,
      "step": 47
    },
    {
      "epoch": 0.5363128491620112,
      "grad_norm": 0.6830984950065613,
      "learning_rate": 9.045084971874738e-05,
      "loss": 0.4794,
      "step": 48
    },
    {
      "epoch": 0.547486033519553,
      "grad_norm": 0.5908480286598206,
      "learning_rate": 8.995939984474624e-05,
      "loss": 0.3296,
      "step": 49
    },
    {
      "epoch": 0.5586592178770949,
      "grad_norm": 0.7622438073158264,
      "learning_rate": 8.945702546981969e-05,
      "loss": 0.4542,
      "step": 50
    },
    {
      "epoch": 0.5586592178770949,
      "eval_loss": 0.434339702129364,
      "eval_runtime": 7.5486,
      "eval_samples_per_second": 20.004,
      "eval_steps_per_second": 5.034,
      "step": 50
    },
    {
      "epoch": 0.5698324022346368,
      "grad_norm": 0.6984612345695496,
      "learning_rate": 8.894386393810563e-05,
      "loss": 0.5736,
      "step": 51
    },
    {
      "epoch": 0.5810055865921788,
      "grad_norm": 0.76580810546875,
      "learning_rate": 8.842005554284296e-05,
      "loss": 0.4749,
      "step": 52
    },
    {
      "epoch": 0.5921787709497207,
      "grad_norm": 0.9494268894195557,
      "learning_rate": 8.788574348801675e-05,
      "loss": 0.6063,
      "step": 53
    },
    {
      "epoch": 0.6033519553072626,
      "grad_norm": 0.6763572692871094,
      "learning_rate": 8.73410738492077e-05,
      "loss": 0.3317,
      "step": 54
    },
    {
      "epoch": 0.6145251396648045,
      "grad_norm": 0.6486566662788391,
      "learning_rate": 8.678619553365659e-05,
      "loss": 0.3562,
      "step": 55
    },
    {
      "epoch": 0.6256983240223464,
      "grad_norm": 0.7950789928436279,
      "learning_rate": 8.622126023955446e-05,
      "loss": 0.2263,
      "step": 56
    },
    {
      "epoch": 0.6368715083798883,
      "grad_norm": 0.6552952527999878,
      "learning_rate": 8.564642241456986e-05,
      "loss": 0.3347,
      "step": 57
    },
    {
      "epoch": 0.6480446927374302,
      "grad_norm": 0.6345714926719666,
      "learning_rate": 8.506183921362443e-05,
      "loss": 0.2841,
      "step": 58
    },
    {
      "epoch": 0.659217877094972,
      "grad_norm": 0.6420994997024536,
      "learning_rate": 8.44676704559283e-05,
      "loss": 0.3026,
      "step": 59
    },
    {
      "epoch": 0.6703910614525139,
      "grad_norm": 0.7764968276023865,
      "learning_rate": 8.386407858128706e-05,
      "loss": 0.1627,
      "step": 60
    },
    {
      "epoch": 0.6815642458100558,
      "grad_norm": 0.5726867318153381,
      "learning_rate": 8.32512286056924e-05,
      "loss": 0.2436,
      "step": 61
    },
    {
      "epoch": 0.6927374301675978,
      "grad_norm": 0.5297699570655823,
      "learning_rate": 8.262928807620843e-05,
      "loss": 0.2084,
      "step": 62
    },
    {
      "epoch": 0.7039106145251397,
      "grad_norm": 0.4522324800491333,
      "learning_rate": 8.199842702516583e-05,
      "loss": 0.2346,
      "step": 63
    },
    {
      "epoch": 0.7150837988826816,
      "grad_norm": 0.5540947318077087,
      "learning_rate": 8.135881792367686e-05,
      "loss": 0.1747,
      "step": 64
    },
    {
      "epoch": 0.7262569832402235,
      "grad_norm": 0.48012828826904297,
      "learning_rate": 8.07106356344834e-05,
      "loss": 0.1713,
      "step": 65
    },
    {
      "epoch": 0.7374301675977654,
      "grad_norm": 0.5377119779586792,
      "learning_rate": 8.005405736415126e-05,
      "loss": 0.2226,
      "step": 66
    },
    {
      "epoch": 0.7486033519553073,
      "grad_norm": 2.5457634925842285,
      "learning_rate": 7.938926261462366e-05,
      "loss": 0.9298,
      "step": 67
    },
    {
      "epoch": 0.7597765363128491,
      "grad_norm": 0.6084294319152832,
      "learning_rate": 7.871643313414718e-05,
      "loss": 0.5864,
      "step": 68
    },
    {
      "epoch": 0.770949720670391,
      "grad_norm": 0.60562664270401,
      "learning_rate": 7.803575286758364e-05,
      "loss": 0.6605,
      "step": 69
    },
    {
      "epoch": 0.7821229050279329,
      "grad_norm": 0.5156601071357727,
      "learning_rate": 7.734740790612136e-05,
      "loss": 0.458,
      "step": 70
    },
    {
      "epoch": 0.7932960893854749,
      "grad_norm": 0.6329552531242371,
      "learning_rate": 7.66515864363997e-05,
      "loss": 0.5411,
      "step": 71
    },
    {
      "epoch": 0.8044692737430168,
      "grad_norm": 0.6346179246902466,
      "learning_rate": 7.594847868906076e-05,
      "loss": 0.5311,
      "step": 72
    },
    {
      "epoch": 0.8156424581005587,
      "grad_norm": 0.6511710286140442,
      "learning_rate": 7.52382768867422e-05,
      "loss": 0.6218,
      "step": 73
    },
    {
      "epoch": 0.8268156424581006,
      "grad_norm": 0.4940270781517029,
      "learning_rate": 7.452117519152542e-05,
      "loss": 0.4921,
      "step": 74
    },
    {
      "epoch": 0.8379888268156425,
      "grad_norm": 0.6882925033569336,
      "learning_rate": 7.379736965185368e-05,
      "loss": 0.6295,
      "step": 75
    },
    {
      "epoch": 0.8491620111731844,
      "grad_norm": 0.6235405802726746,
      "learning_rate": 7.30670581489344e-05,
      "loss": 0.457,
      "step": 76
    },
    {
      "epoch": 0.8603351955307262,
      "grad_norm": 0.6313698291778564,
      "learning_rate": 7.233044034264034e-05,
      "loss": 0.2338,
      "step": 77
    },
    {
      "epoch": 0.8715083798882681,
      "grad_norm": 0.7784355878829956,
      "learning_rate": 7.158771761692464e-05,
      "loss": 0.4064,
      "step": 78
    },
    {
      "epoch": 0.88268156424581,
      "grad_norm": 0.7756523489952087,
      "learning_rate": 7.083909302476453e-05,
      "loss": 0.2674,
      "step": 79
    },
    {
      "epoch": 0.8938547486033519,
      "grad_norm": 0.7363424897193909,
      "learning_rate": 7.008477123264848e-05,
      "loss": 0.4045,
      "step": 80
    },
    {
      "epoch": 0.9050279329608939,
      "grad_norm": 0.650582492351532,
      "learning_rate": 6.932495846462261e-05,
      "loss": 0.2804,
      "step": 81
    },
    {
      "epoch": 0.9162011173184358,
      "grad_norm": 0.7414998412132263,
      "learning_rate": 6.855986244591104e-05,
      "loss": 0.2908,
      "step": 82
    },
    {
      "epoch": 0.9273743016759777,
      "grad_norm": 0.5577226877212524,
      "learning_rate": 6.778969234612584e-05,
      "loss": 0.1618,
      "step": 83
    },
    {
      "epoch": 0.9385474860335196,
      "grad_norm": 0.43591392040252686,
      "learning_rate": 6.701465872208216e-05,
      "loss": 0.1398,
      "step": 84
    },
    {
      "epoch": 0.9497206703910615,
      "grad_norm": 0.3598768413066864,
      "learning_rate": 6.623497346023418e-05,
      "loss": 0.1208,
      "step": 85
    },
    {
      "epoch": 0.9608938547486033,
      "grad_norm": 0.39774054288864136,
      "learning_rate": 6.545084971874738e-05,
      "loss": 0.1733,
      "step": 86
    },
    {
      "epoch": 0.9720670391061452,
      "grad_norm": 0.4294886887073517,
      "learning_rate": 6.466250186922325e-05,
      "loss": 0.1291,
      "step": 87
    },
    {
      "epoch": 0.9832402234636871,
      "grad_norm": 0.43344977498054504,
      "learning_rate": 6.387014543809223e-05,
      "loss": 0.159,
      "step": 88
    },
    {
      "epoch": 0.994413407821229,
      "grad_norm": 0.5479728579521179,
      "learning_rate": 6.307399704769099e-05,
      "loss": 0.586,
      "step": 89
    },
    {
      "epoch": 1.005586592178771,
      "grad_norm": 1.402073860168457,
      "learning_rate": 6.227427435703997e-05,
      "loss": 0.8381,
      "step": 90
    },
    {
      "epoch": 1.0167597765363128,
      "grad_norm": 0.38112547993659973,
      "learning_rate": 6.147119600233758e-05,
      "loss": 0.517,
      "step": 91
    },
    {
      "epoch": 1.0279329608938548,
      "grad_norm": 0.47864004969596863,
      "learning_rate": 6.066498153718735e-05,
      "loss": 0.516,
      "step": 92
    },
    {
      "epoch": 1.0391061452513966,
      "grad_norm": 0.4763987958431244,
      "learning_rate": 5.985585137257401e-05,
      "loss": 0.4618,
      "step": 93
    },
    {
      "epoch": 1.0502793296089385,
      "grad_norm": 0.38241147994995117,
      "learning_rate": 5.90440267166055e-05,
      "loss": 0.3306,
      "step": 94
    },
    {
      "epoch": 1.0614525139664805,
      "grad_norm": 0.43910539150238037,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 0.3816,
      "step": 95
    },
    {
      "epoch": 1.0726256983240223,
      "grad_norm": 0.5805469751358032,
      "learning_rate": 5.74131823855921e-05,
      "loss": 0.4254,
      "step": 96
    },
    {
      "epoch": 1.0837988826815643,
      "grad_norm": 0.4538100063800812,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 0.3432,
      "step": 97
    },
    {
      "epoch": 1.094972067039106,
      "grad_norm": 0.44317829608917236,
      "learning_rate": 5.577423184847932e-05,
      "loss": 0.3423,
      "step": 98
    },
    {
      "epoch": 1.106145251396648,
      "grad_norm": 0.5460543036460876,
      "learning_rate": 5.495227651252315e-05,
      "loss": 0.3158,
      "step": 99
    },
    {
      "epoch": 1.1173184357541899,
      "grad_norm": 0.4699908196926117,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 0.335,
      "step": 100
    },
    {
      "epoch": 1.1173184357541899,
      "eval_loss": 0.35996368527412415,
      "eval_runtime": 7.6399,
      "eval_samples_per_second": 19.765,
      "eval_steps_per_second": 4.974,
      "step": 100
    },
    {
      "epoch": 1.1284916201117319,
      "grad_norm": 0.49154403805732727,
      "learning_rate": 5.330452921628497e-05,
      "loss": 0.2768,
      "step": 101
    },
    {
      "epoch": 1.1396648044692737,
      "grad_norm": 0.4707217514514923,
      "learning_rate": 5.247918773366112e-05,
      "loss": 0.1589,
      "step": 102
    },
    {
      "epoch": 1.1508379888268156,
      "grad_norm": 0.46079739928245544,
      "learning_rate": 5.165316846586541e-05,
      "loss": 0.193,
      "step": 103
    },
    {
      "epoch": 1.1620111731843576,
      "grad_norm": 0.5139585733413696,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 0.2307,
      "step": 104
    },
    {
      "epoch": 1.1731843575418994,
      "grad_norm": 0.668438732624054,
      "learning_rate": 5e-05,
      "loss": 0.2282,
      "step": 105
    },
    {
      "epoch": 1.1843575418994414,
      "grad_norm": 0.5157988667488098,
      "learning_rate": 4.917330276168208e-05,
      "loss": 0.1328,
      "step": 106
    },
    {
      "epoch": 1.1955307262569832,
      "grad_norm": 0.39924243092536926,
      "learning_rate": 4.834683153413459e-05,
      "loss": 0.1255,
      "step": 107
    },
    {
      "epoch": 1.2067039106145252,
      "grad_norm": 0.42700207233428955,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 0.1351,
      "step": 108
    },
    {
      "epoch": 1.217877094972067,
      "grad_norm": 0.48354342579841614,
      "learning_rate": 4.669547078371504e-05,
      "loss": 0.1574,
      "step": 109
    },
    {
      "epoch": 1.229050279329609,
      "grad_norm": 0.3663063049316406,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 0.0886,
      "step": 110
    },
    {
      "epoch": 1.2402234636871508,
      "grad_norm": 0.42423301935195923,
      "learning_rate": 4.504772348747687e-05,
      "loss": 0.0983,
      "step": 111
    },
    {
      "epoch": 1.2513966480446927,
      "grad_norm": 0.5446560978889465,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 0.6438,
      "step": 112
    },
    {
      "epoch": 1.2625698324022347,
      "grad_norm": 0.3894984722137451,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 0.3759,
      "step": 113
    },
    {
      "epoch": 1.2737430167597765,
      "grad_norm": 0.43923401832580566,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 0.379,
      "step": 114
    },
    {
      "epoch": 1.2849162011173183,
      "grad_norm": 0.46830540895462036,
      "learning_rate": 4.17702704859633e-05,
      "loss": 0.3389,
      "step": 115
    },
    {
      "epoch": 1.2960893854748603,
      "grad_norm": 0.49750062823295593,
      "learning_rate": 4.095597328339452e-05,
      "loss": 0.3585,
      "step": 116
    },
    {
      "epoch": 1.3072625698324023,
      "grad_norm": 0.5573229789733887,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 0.417,
      "step": 117
    },
    {
      "epoch": 1.318435754189944,
      "grad_norm": 0.5657495260238647,
      "learning_rate": 3.933501846281267e-05,
      "loss": 0.4298,
      "step": 118
    },
    {
      "epoch": 1.329608938547486,
      "grad_norm": 0.4059884250164032,
      "learning_rate": 3.852880399766243e-05,
      "loss": 0.3385,
      "step": 119
    },
    {
      "epoch": 1.3407821229050279,
      "grad_norm": 0.5553444027900696,
      "learning_rate": 3.772572564296005e-05,
      "loss": 0.3361,
      "step": 120
    },
    {
      "epoch": 1.3519553072625698,
      "grad_norm": 0.49883168935775757,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 0.3448,
      "step": 121
    },
    {
      "epoch": 1.3631284916201118,
      "grad_norm": 0.5660603642463684,
      "learning_rate": 3.612985456190778e-05,
      "loss": 0.3264,
      "step": 122
    },
    {
      "epoch": 1.3743016759776536,
      "grad_norm": 0.5664513111114502,
      "learning_rate": 3.533749813077677e-05,
      "loss": 0.1938,
      "step": 123
    },
    {
      "epoch": 1.3854748603351954,
      "grad_norm": 0.5513072609901428,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 0.268,
      "step": 124
    },
    {
      "epoch": 1.3966480446927374,
      "grad_norm": 0.5621060132980347,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 0.1971,
      "step": 125
    },
    {
      "epoch": 1.4078212290502794,
      "grad_norm": 0.5587654113769531,
      "learning_rate": 3.298534127791785e-05,
      "loss": 0.2274,
      "step": 126
    },
    {
      "epoch": 1.4189944134078212,
      "grad_norm": 0.6601653695106506,
      "learning_rate": 3.221030765387417e-05,
      "loss": 0.3228,
      "step": 127
    },
    {
      "epoch": 1.4301675977653632,
      "grad_norm": 0.4269857704639435,
      "learning_rate": 3.144013755408895e-05,
      "loss": 0.1428,
      "step": 128
    },
    {
      "epoch": 1.441340782122905,
      "grad_norm": 0.45998716354370117,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 0.1673,
      "step": 129
    },
    {
      "epoch": 1.452513966480447,
      "grad_norm": 0.3854297995567322,
      "learning_rate": 2.991522876735154e-05,
      "loss": 0.133,
      "step": 130
    },
    {
      "epoch": 1.463687150837989,
      "grad_norm": 0.42208659648895264,
      "learning_rate": 2.916090697523549e-05,
      "loss": 0.1362,
      "step": 131
    },
    {
      "epoch": 1.4748603351955307,
      "grad_norm": 0.27260181307792664,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 0.0687,
      "step": 132
    },
    {
      "epoch": 1.4860335195530725,
      "grad_norm": 0.4132692217826843,
      "learning_rate": 2.766955965735968e-05,
      "loss": 0.1212,
      "step": 133
    },
    {
      "epoch": 1.4972067039106145,
      "grad_norm": 0.4734876751899719,
      "learning_rate": 2.693294185106562e-05,
      "loss": 0.7753,
      "step": 134
    },
    {
      "epoch": 1.5083798882681565,
      "grad_norm": 0.39801645278930664,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 0.4977,
      "step": 135
    },
    {
      "epoch": 1.5195530726256983,
      "grad_norm": 0.45366907119750977,
      "learning_rate": 2.547882480847461e-05,
      "loss": 0.3853,
      "step": 136
    },
    {
      "epoch": 1.5307262569832403,
      "grad_norm": 0.4831063151359558,
      "learning_rate": 2.476172311325783e-05,
      "loss": 0.3221,
      "step": 137
    },
    {
      "epoch": 1.541899441340782,
      "grad_norm": 0.44566017389297485,
      "learning_rate": 2.405152131093926e-05,
      "loss": 0.3098,
      "step": 138
    },
    {
      "epoch": 1.553072625698324,
      "grad_norm": 0.5208490490913391,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 0.3654,
      "step": 139
    },
    {
      "epoch": 1.564245810055866,
      "grad_norm": 0.6967378258705139,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 0.4679,
      "step": 140
    },
    {
      "epoch": 1.5754189944134078,
      "grad_norm": 0.44668933749198914,
      "learning_rate": 2.196424713241637e-05,
      "loss": 0.4452,
      "step": 141
    },
    {
      "epoch": 1.5865921787709496,
      "grad_norm": 0.5407074689865112,
      "learning_rate": 2.128356686585282e-05,
      "loss": 0.4012,
      "step": 142
    },
    {
      "epoch": 1.5977653631284916,
      "grad_norm": 0.4964157044887543,
      "learning_rate": 2.061073738537635e-05,
      "loss": 0.3284,
      "step": 143
    },
    {
      "epoch": 1.6089385474860336,
      "grad_norm": 0.504605233669281,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 0.2601,
      "step": 144
    },
    {
      "epoch": 1.6201117318435754,
      "grad_norm": 0.51454758644104,
      "learning_rate": 1.928936436551661e-05,
      "loss": 0.2471,
      "step": 145
    },
    {
      "epoch": 1.6312849162011172,
      "grad_norm": 0.6736803650856018,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 0.1578,
      "step": 146
    },
    {
      "epoch": 1.6424581005586592,
      "grad_norm": 0.5892282724380493,
      "learning_rate": 1.800157297483417e-05,
      "loss": 0.2802,
      "step": 147
    },
    {
      "epoch": 1.6536312849162011,
      "grad_norm": 0.5950078964233398,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 0.2356,
      "step": 148
    },
    {
      "epoch": 1.6648044692737431,
      "grad_norm": 0.5183621644973755,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 0.1884,
      "step": 149
    },
    {
      "epoch": 1.675977653631285,
      "grad_norm": 0.5445525646209717,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 0.1198,
      "step": 150
    },
    {
      "epoch": 1.675977653631285,
      "eval_loss": 0.3512671887874603,
      "eval_runtime": 7.5681,
      "eval_samples_per_second": 19.952,
      "eval_steps_per_second": 5.021,
      "step": 150
    },
    {
      "epoch": 1.6871508379888267,
      "grad_norm": 0.42454764246940613,
      "learning_rate": 1.553232954407171e-05,
      "loss": 0.1054,
      "step": 151
    },
    {
      "epoch": 1.6983240223463687,
      "grad_norm": 0.43263888359069824,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 0.1042,
      "step": 152
    },
    {
      "epoch": 1.7094972067039107,
      "grad_norm": 0.48480093479156494,
      "learning_rate": 1.435357758543015e-05,
      "loss": 0.1223,
      "step": 153
    },
    {
      "epoch": 1.7206703910614525,
      "grad_norm": 0.39971283078193665,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 0.1078,
      "step": 154
    },
    {
      "epoch": 1.7318435754189943,
      "grad_norm": 0.4790002703666687,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 0.1419,
      "step": 155
    },
    {
      "epoch": 1.7430167597765363,
      "grad_norm": 0.4774593710899353,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 0.7742,
      "step": 156
    },
    {
      "epoch": 1.7541899441340782,
      "grad_norm": 0.4144189953804016,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 0.4127,
      "step": 157
    },
    {
      "epoch": 1.7653631284916202,
      "grad_norm": 0.5643066167831421,
      "learning_rate": 1.157994445715706e-05,
      "loss": 0.4321,
      "step": 158
    },
    {
      "epoch": 1.776536312849162,
      "grad_norm": 0.5547996163368225,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 0.4448,
      "step": 159
    },
    {
      "epoch": 1.7877094972067038,
      "grad_norm": 0.5182667374610901,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 0.3853,
      "step": 160
    },
    {
      "epoch": 1.7988826815642458,
      "grad_norm": 0.43529149889945984,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 0.2178,
      "step": 161
    },
    {
      "epoch": 1.8100558659217878,
      "grad_norm": 0.4870721697807312,
      "learning_rate": 9.549150281252633e-06,
      "loss": 0.323,
      "step": 162
    },
    {
      "epoch": 1.8212290502793296,
      "grad_norm": 0.47201111912727356,
      "learning_rate": 9.068759265665384e-06,
      "loss": 0.3842,
      "step": 163
    },
    {
      "epoch": 1.8324022346368714,
      "grad_norm": 0.4745241701602936,
      "learning_rate": 8.599558442598998e-06,
      "loss": 0.2734,
      "step": 164
    },
    {
      "epoch": 1.8435754189944134,
      "grad_norm": 0.7225174903869629,
      "learning_rate": 8.141676086873572e-06,
      "loss": 0.3588,
      "step": 165
    },
    {
      "epoch": 1.8547486033519553,
      "grad_norm": 0.5225138068199158,
      "learning_rate": 7.695237378953223e-06,
      "loss": 0.2812,
      "step": 166
    },
    {
      "epoch": 1.8659217877094973,
      "grad_norm": 0.5475061535835266,
      "learning_rate": 7.260364370723044e-06,
      "loss": 0.2399,
      "step": 167
    },
    {
      "epoch": 1.8770949720670391,
      "grad_norm": 0.48459821939468384,
      "learning_rate": 6.837175952121306e-06,
      "loss": 0.1882,
      "step": 168
    },
    {
      "epoch": 1.888268156424581,
      "grad_norm": 0.5391102433204651,
      "learning_rate": 6.425787818636131e-06,
      "loss": 0.2384,
      "step": 169
    },
    {
      "epoch": 1.899441340782123,
      "grad_norm": 0.5096697211265564,
      "learning_rate": 6.026312439675552e-06,
      "loss": 0.1634,
      "step": 170
    },
    {
      "epoch": 1.910614525139665,
      "grad_norm": 0.5715564489364624,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 0.2364,
      "step": 171
    },
    {
      "epoch": 1.9217877094972067,
      "grad_norm": 0.4218137264251709,
      "learning_rate": 5.263533508961827e-06,
      "loss": 0.1267,
      "step": 172
    },
    {
      "epoch": 1.9329608938547485,
      "grad_norm": 0.5504609942436218,
      "learning_rate": 4.900438493352055e-06,
      "loss": 0.1345,
      "step": 173
    },
    {
      "epoch": 1.9441340782122905,
      "grad_norm": 0.3880223333835602,
      "learning_rate": 4.549673247541875e-06,
      "loss": 0.1191,
      "step": 174
    },
    {
      "epoch": 1.9553072625698324,
      "grad_norm": 0.4032748341560364,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 0.1058,
      "step": 175
    },
    {
      "epoch": 1.9664804469273744,
      "grad_norm": 0.4006319046020508,
      "learning_rate": 3.885512251130763e-06,
      "loss": 0.1106,
      "step": 176
    },
    {
      "epoch": 1.9776536312849162,
      "grad_norm": 0.3460472822189331,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 0.0651,
      "step": 177
    },
    {
      "epoch": 1.988826815642458,
      "grad_norm": 0.6127005815505981,
      "learning_rate": 3.271776770026963e-06,
      "loss": 0.4993,
      "step": 178
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.4792274236679077,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 0.1147,
      "step": 179
    },
    {
      "epoch": 2.011173184357542,
      "grad_norm": 0.3817814588546753,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 0.6282,
      "step": 180
    },
    {
      "epoch": 2.022346368715084,
      "grad_norm": 0.41862374544143677,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 0.4964,
      "step": 181
    },
    {
      "epoch": 2.0335195530726256,
      "grad_norm": 0.4326348304748535,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 0.3165,
      "step": 182
    },
    {
      "epoch": 2.0446927374301676,
      "grad_norm": 0.3941451609134674,
      "learning_rate": 1.962316193157593e-06,
      "loss": 0.2619,
      "step": 183
    },
    {
      "epoch": 2.0558659217877095,
      "grad_norm": 0.4871034622192383,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 0.3221,
      "step": 184
    },
    {
      "epoch": 2.0670391061452515,
      "grad_norm": 0.5044857859611511,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 0.3445,
      "step": 185
    },
    {
      "epoch": 2.078212290502793,
      "grad_norm": 0.3867410719394684,
      "learning_rate": 1.333670137599713e-06,
      "loss": 0.3141,
      "step": 186
    },
    {
      "epoch": 2.089385474860335,
      "grad_norm": 0.4248393476009369,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 0.2161,
      "step": 187
    },
    {
      "epoch": 2.100558659217877,
      "grad_norm": 0.48327651619911194,
      "learning_rate": 9.810017062595322e-07,
      "loss": 0.2747,
      "step": 188
    },
    {
      "epoch": 2.111731843575419,
      "grad_norm": 0.4485223889350891,
      "learning_rate": 8.247462563808817e-07,
      "loss": 0.1869,
      "step": 189
    },
    {
      "epoch": 2.122905027932961,
      "grad_norm": 0.4793951213359833,
      "learning_rate": 6.819348298638839e-07,
      "loss": 0.2088,
      "step": 190
    },
    {
      "epoch": 2.1340782122905027,
      "grad_norm": 0.4176865816116333,
      "learning_rate": 5.526064699265753e-07,
      "loss": 0.1131,
      "step": 191
    },
    {
      "epoch": 2.1452513966480447,
      "grad_norm": 0.5057032704353333,
      "learning_rate": 4.367965336512403e-07,
      "loss": 0.2373,
      "step": 192
    },
    {
      "epoch": 2.1564245810055866,
      "grad_norm": 0.5785103440284729,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 0.2305,
      "step": 193
    },
    {
      "epoch": 2.1675977653631286,
      "grad_norm": 0.4635258615016937,
      "learning_rate": 2.458548727494292e-07,
      "loss": 0.1885,
      "step": 194
    },
    {
      "epoch": 2.17877094972067,
      "grad_norm": 0.5159782767295837,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 0.1169,
      "step": 195
    },
    {
      "epoch": 2.189944134078212,
      "grad_norm": 0.40061166882514954,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 0.1152,
      "step": 196
    },
    {
      "epoch": 2.201117318435754,
      "grad_norm": 0.3203735053539276,
      "learning_rate": 6.150154258476315e-08,
      "loss": 0.0884,
      "step": 197
    },
    {
      "epoch": 2.212290502793296,
      "grad_norm": 0.3295019268989563,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 0.1021,
      "step": 198
    },
    {
      "epoch": 2.223463687150838,
      "grad_norm": 0.37430116534233093,
      "learning_rate": 6.834750376549792e-09,
      "loss": 0.0825,
      "step": 199
    },
    {
      "epoch": 2.2346368715083798,
      "grad_norm": 0.36784055829048157,
      "learning_rate": 0.0,
      "loss": 0.0834,
      "step": 200
    },
    {
      "epoch": 2.2346368715083798,
      "eval_loss": 0.3507079780101776,
      "eval_runtime": 7.5521,
      "eval_samples_per_second": 19.994,
      "eval_steps_per_second": 5.032,
      "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": 9.514860146589696e+16,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}