{
  "best_metric": 0.28383296728134155,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.8592910848549946,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.004296455424274973,
      "grad_norm": 0.3637087345123291,
      "learning_rate": 1e-05,
      "loss": 0.8103,
      "step": 1
    },
    {
      "epoch": 0.004296455424274973,
      "eval_loss": 2.395321846008301,
      "eval_runtime": 27.149,
      "eval_samples_per_second": 14.439,
      "eval_steps_per_second": 3.61,
      "step": 1
    },
    {
      "epoch": 0.008592910848549946,
      "grad_norm": 0.5150983333587646,
      "learning_rate": 2e-05,
      "loss": 1.1294,
      "step": 2
    },
    {
      "epoch": 0.01288936627282492,
      "grad_norm": 0.6066375970840454,
      "learning_rate": 3e-05,
      "loss": 1.2703,
      "step": 3
    },
    {
      "epoch": 0.017185821697099892,
      "grad_norm": 0.5858229994773865,
      "learning_rate": 4e-05,
      "loss": 1.3463,
      "step": 4
    },
    {
      "epoch": 0.021482277121374866,
      "grad_norm": 0.6776552200317383,
      "learning_rate": 5e-05,
      "loss": 1.2584,
      "step": 5
    },
    {
      "epoch": 0.02577873254564984,
      "grad_norm": 0.606339693069458,
      "learning_rate": 6e-05,
      "loss": 1.2734,
      "step": 6
    },
    {
      "epoch": 0.03007518796992481,
      "grad_norm": 0.6469846963882446,
      "learning_rate": 7e-05,
      "loss": 1.1792,
      "step": 7
    },
    {
      "epoch": 0.034371643394199784,
      "grad_norm": 0.5836963057518005,
      "learning_rate": 8e-05,
      "loss": 1.1451,
      "step": 8
    },
    {
      "epoch": 0.03866809881847476,
      "grad_norm": 0.5917362570762634,
      "learning_rate": 9e-05,
      "loss": 0.9683,
      "step": 9
    },
    {
      "epoch": 0.04296455424274973,
      "grad_norm": 0.8102052211761475,
      "learning_rate": 0.0001,
      "loss": 0.8943,
      "step": 10
    },
    {
      "epoch": 0.047261009667024706,
      "grad_norm": 0.6655777096748352,
      "learning_rate": 9.999316524962345e-05,
      "loss": 0.739,
      "step": 11
    },
    {
      "epoch": 0.05155746509129968,
      "grad_norm": 0.7479202747344971,
      "learning_rate": 9.997266286704631e-05,
      "loss": 0.7188,
      "step": 12
    },
    {
      "epoch": 0.055853920515574654,
      "grad_norm": 0.7106651663780212,
      "learning_rate": 9.993849845741524e-05,
      "loss": 0.6258,
      "step": 13
    },
    {
      "epoch": 0.06015037593984962,
      "grad_norm": 0.5525503754615784,
      "learning_rate": 9.989068136093873e-05,
      "loss": 0.6035,
      "step": 14
    },
    {
      "epoch": 0.0644468313641246,
      "grad_norm": 0.45747628808021545,
      "learning_rate": 9.98292246503335e-05,
      "loss": 0.5847,
      "step": 15
    },
    {
      "epoch": 0.06874328678839957,
      "grad_norm": 0.40945225954055786,
      "learning_rate": 9.975414512725057e-05,
      "loss": 0.4871,
      "step": 16
    },
    {
      "epoch": 0.07303974221267455,
      "grad_norm": 0.5227960348129272,
      "learning_rate": 9.966546331768191e-05,
      "loss": 0.4554,
      "step": 17
    },
    {
      "epoch": 0.07733619763694952,
      "grad_norm": 0.3709550201892853,
      "learning_rate": 9.956320346634876e-05,
      "loss": 0.4575,
      "step": 18
    },
    {
      "epoch": 0.08163265306122448,
      "grad_norm": 0.35335588455200195,
      "learning_rate": 9.944739353007344e-05,
      "loss": 0.4892,
      "step": 19
    },
    {
      "epoch": 0.08592910848549946,
      "grad_norm": 0.41486284136772156,
      "learning_rate": 9.931806517013612e-05,
      "loss": 0.5214,
      "step": 20
    },
    {
      "epoch": 0.09022556390977443,
      "grad_norm": 0.4549630582332611,
      "learning_rate": 9.917525374361912e-05,
      "loss": 0.6713,
      "step": 21
    },
    {
      "epoch": 0.09452201933404941,
      "grad_norm": 0.4528137445449829,
      "learning_rate": 9.901899829374047e-05,
      "loss": 0.6117,
      "step": 22
    },
    {
      "epoch": 0.09881847475832438,
      "grad_norm": 0.47058239579200745,
      "learning_rate": 9.884934153917997e-05,
      "loss": 0.5362,
      "step": 23
    },
    {
      "epoch": 0.10311493018259936,
      "grad_norm": 0.5150260329246521,
      "learning_rate": 9.86663298624003e-05,
      "loss": 0.6023,
      "step": 24
    },
    {
      "epoch": 0.10741138560687433,
      "grad_norm": 0.4763900339603424,
      "learning_rate": 9.847001329696653e-05,
      "loss": 0.6096,
      "step": 25
    },
    {
      "epoch": 0.11170784103114931,
      "grad_norm": 0.4491954445838928,
      "learning_rate": 9.826044551386744e-05,
      "loss": 0.5383,
      "step": 26
    },
    {
      "epoch": 0.11600429645542427,
      "grad_norm": 0.37725311517715454,
      "learning_rate": 9.803768380684242e-05,
      "loss": 0.5781,
      "step": 27
    },
    {
      "epoch": 0.12030075187969924,
      "grad_norm": 0.48423704504966736,
      "learning_rate": 9.780178907671789e-05,
      "loss": 0.5574,
      "step": 28
    },
    {
      "epoch": 0.12459720730397422,
      "grad_norm": 0.4108438193798065,
      "learning_rate": 9.755282581475769e-05,
      "loss": 0.4861,
      "step": 29
    },
    {
      "epoch": 0.1288936627282492,
      "grad_norm": 0.40415096282958984,
      "learning_rate": 9.729086208503174e-05,
      "loss": 0.5138,
      "step": 30
    },
    {
      "epoch": 0.13319011815252416,
      "grad_norm": 0.4338022470474243,
      "learning_rate": 9.701596950580806e-05,
      "loss": 0.5121,
      "step": 31
    },
    {
      "epoch": 0.13748657357679914,
      "grad_norm": 0.639772355556488,
      "learning_rate": 9.672822322997305e-05,
      "loss": 0.5894,
      "step": 32
    },
    {
      "epoch": 0.14178302900107412,
      "grad_norm": 0.4205399453639984,
      "learning_rate": 9.642770192448536e-05,
      "loss": 0.4535,
      "step": 33
    },
    {
      "epoch": 0.1460794844253491,
      "grad_norm": 0.48501643538475037,
      "learning_rate": 9.611448774886924e-05,
      "loss": 0.4877,
      "step": 34
    },
    {
      "epoch": 0.15037593984962405,
      "grad_norm": 0.5719529390335083,
      "learning_rate": 9.578866633275288e-05,
      "loss": 0.5284,
      "step": 35
    },
    {
      "epoch": 0.15467239527389903,
      "grad_norm": 0.45450690388679504,
      "learning_rate": 9.545032675245813e-05,
      "loss": 0.4434,
      "step": 36
    },
    {
      "epoch": 0.158968850698174,
      "grad_norm": 0.5546385645866394,
      "learning_rate": 9.509956150664796e-05,
      "loss": 0.5373,
      "step": 37
    },
    {
      "epoch": 0.16326530612244897,
      "grad_norm": 0.5521821975708008,
      "learning_rate": 9.473646649103818e-05,
      "loss": 0.5153,
      "step": 38
    },
    {
      "epoch": 0.16756176154672395,
      "grad_norm": 0.4645980894565582,
      "learning_rate": 9.43611409721806e-05,
      "loss": 0.4344,
      "step": 39
    },
    {
      "epoch": 0.17185821697099893,
      "grad_norm": 0.5077691078186035,
      "learning_rate": 9.397368756032445e-05,
      "loss": 0.4193,
      "step": 40
    },
    {
      "epoch": 0.1761546723952739,
      "grad_norm": 0.6285257339477539,
      "learning_rate": 9.357421218136386e-05,
      "loss": 0.4512,
      "step": 41
    },
    {
      "epoch": 0.18045112781954886,
      "grad_norm": 0.7595347166061401,
      "learning_rate": 9.316282404787871e-05,
      "loss": 0.5183,
      "step": 42
    },
    {
      "epoch": 0.18474758324382384,
      "grad_norm": 0.6894388794898987,
      "learning_rate": 9.273963562927695e-05,
      "loss": 0.4657,
      "step": 43
    },
    {
      "epoch": 0.18904403866809882,
      "grad_norm": 0.6750317811965942,
      "learning_rate": 9.230476262104677e-05,
      "loss": 0.4279,
      "step": 44
    },
    {
      "epoch": 0.1933404940923738,
      "grad_norm": 0.9030443429946899,
      "learning_rate": 9.185832391312644e-05,
      "loss": 0.6653,
      "step": 45
    },
    {
      "epoch": 0.19763694951664876,
      "grad_norm": 0.7086827158927917,
      "learning_rate": 9.140044155740101e-05,
      "loss": 0.5327,
      "step": 46
    },
    {
      "epoch": 0.20193340494092374,
      "grad_norm": 0.887897789478302,
      "learning_rate": 9.093124073433463e-05,
      "loss": 0.5247,
      "step": 47
    },
    {
      "epoch": 0.20622986036519872,
      "grad_norm": 0.8612096309661865,
      "learning_rate": 9.045084971874738e-05,
      "loss": 0.5503,
      "step": 48
    },
    {
      "epoch": 0.21052631578947367,
      "grad_norm": 1.9472252130508423,
      "learning_rate": 8.995939984474624e-05,
      "loss": 0.6008,
      "step": 49
    },
    {
      "epoch": 0.21482277121374865,
      "grad_norm": 2.0724754333496094,
      "learning_rate": 8.945702546981969e-05,
      "loss": 0.4043,
      "step": 50
    },
    {
      "epoch": 0.21482277121374865,
      "eval_loss": 0.4881606698036194,
      "eval_runtime": 27.6633,
      "eval_samples_per_second": 14.17,
      "eval_steps_per_second": 3.543,
      "step": 50
    },
    {
      "epoch": 0.21911922663802363,
      "grad_norm": 0.44782543182373047,
      "learning_rate": 8.894386393810563e-05,
      "loss": 0.3396,
      "step": 51
    },
    {
      "epoch": 0.22341568206229862,
      "grad_norm": 0.5945596098899841,
      "learning_rate": 8.842005554284296e-05,
      "loss": 0.4482,
      "step": 52
    },
    {
      "epoch": 0.22771213748657357,
      "grad_norm": 0.5052401423454285,
      "learning_rate": 8.788574348801675e-05,
      "loss": 0.4347,
      "step": 53
    },
    {
      "epoch": 0.23200859291084855,
      "grad_norm": 0.4160156548023224,
      "learning_rate": 8.73410738492077e-05,
      "loss": 0.4182,
      "step": 54
    },
    {
      "epoch": 0.23630504833512353,
      "grad_norm": 0.3482251763343811,
      "learning_rate": 8.678619553365659e-05,
      "loss": 0.4322,
      "step": 55
    },
    {
      "epoch": 0.24060150375939848,
      "grad_norm": 0.3227390944957733,
      "learning_rate": 8.622126023955446e-05,
      "loss": 0.4172,
      "step": 56
    },
    {
      "epoch": 0.24489795918367346,
      "grad_norm": 0.29572582244873047,
      "learning_rate": 8.564642241456986e-05,
      "loss": 0.3785,
      "step": 57
    },
    {
      "epoch": 0.24919441460794844,
      "grad_norm": 0.35989752411842346,
      "learning_rate": 8.506183921362443e-05,
      "loss": 0.3799,
      "step": 58
    },
    {
      "epoch": 0.2534908700322234,
      "grad_norm": 0.385657399892807,
      "learning_rate": 8.44676704559283e-05,
      "loss": 0.3866,
      "step": 59
    },
    {
      "epoch": 0.2577873254564984,
      "grad_norm": 0.3310921788215637,
      "learning_rate": 8.386407858128706e-05,
      "loss": 0.3486,
      "step": 60
    },
    {
      "epoch": 0.2620837808807734,
      "grad_norm": 0.3422428071498871,
      "learning_rate": 8.32512286056924e-05,
      "loss": 0.3028,
      "step": 61
    },
    {
      "epoch": 0.2663802363050483,
      "grad_norm": 0.31822556257247925,
      "learning_rate": 8.262928807620843e-05,
      "loss": 0.297,
      "step": 62
    },
    {
      "epoch": 0.2706766917293233,
      "grad_norm": 0.3344513177871704,
      "learning_rate": 8.199842702516583e-05,
      "loss": 0.3756,
      "step": 63
    },
    {
      "epoch": 0.2749731471535983,
      "grad_norm": 0.2931362986564636,
      "learning_rate": 8.135881792367686e-05,
      "loss": 0.4118,
      "step": 64
    },
    {
      "epoch": 0.27926960257787325,
      "grad_norm": 0.3394784927368164,
      "learning_rate": 8.07106356344834e-05,
      "loss": 0.3515,
      "step": 65
    },
    {
      "epoch": 0.28356605800214824,
      "grad_norm": 0.317038357257843,
      "learning_rate": 8.005405736415126e-05,
      "loss": 0.3846,
      "step": 66
    },
    {
      "epoch": 0.2878625134264232,
      "grad_norm": 0.27550095319747925,
      "learning_rate": 7.938926261462366e-05,
      "loss": 0.3494,
      "step": 67
    },
    {
      "epoch": 0.2921589688506982,
      "grad_norm": 0.3647957146167755,
      "learning_rate": 7.871643313414718e-05,
      "loss": 0.3669,
      "step": 68
    },
    {
      "epoch": 0.2964554242749731,
      "grad_norm": 0.34974852204322815,
      "learning_rate": 7.803575286758364e-05,
      "loss": 0.3403,
      "step": 69
    },
    {
      "epoch": 0.3007518796992481,
      "grad_norm": 0.35612866282463074,
      "learning_rate": 7.734740790612136e-05,
      "loss": 0.3475,
      "step": 70
    },
    {
      "epoch": 0.3050483351235231,
      "grad_norm": 0.342788428068161,
      "learning_rate": 7.66515864363997e-05,
      "loss": 0.3453,
      "step": 71
    },
    {
      "epoch": 0.30934479054779807,
      "grad_norm": 0.3459738492965698,
      "learning_rate": 7.594847868906076e-05,
      "loss": 0.3246,
      "step": 72
    },
    {
      "epoch": 0.31364124597207305,
      "grad_norm": 0.4553300440311432,
      "learning_rate": 7.52382768867422e-05,
      "loss": 0.4952,
      "step": 73
    },
    {
      "epoch": 0.317937701396348,
      "grad_norm": 0.37579211592674255,
      "learning_rate": 7.452117519152542e-05,
      "loss": 0.3655,
      "step": 74
    },
    {
      "epoch": 0.322234156820623,
      "grad_norm": 0.42491230368614197,
      "learning_rate": 7.379736965185368e-05,
      "loss": 0.3746,
      "step": 75
    },
    {
      "epoch": 0.32653061224489793,
      "grad_norm": 0.4613445997238159,
      "learning_rate": 7.30670581489344e-05,
      "loss": 0.483,
      "step": 76
    },
    {
      "epoch": 0.3308270676691729,
      "grad_norm": 0.43322092294692993,
      "learning_rate": 7.233044034264034e-05,
      "loss": 0.3813,
      "step": 77
    },
    {
      "epoch": 0.3351235230934479,
      "grad_norm": 0.43508535623550415,
      "learning_rate": 7.158771761692464e-05,
      "loss": 0.4426,
      "step": 78
    },
    {
      "epoch": 0.3394199785177229,
      "grad_norm": 0.3977321982383728,
      "learning_rate": 7.083909302476453e-05,
      "loss": 0.3795,
      "step": 79
    },
    {
      "epoch": 0.34371643394199786,
      "grad_norm": 0.36455708742141724,
      "learning_rate": 7.008477123264848e-05,
      "loss": 0.3186,
      "step": 80
    },
    {
      "epoch": 0.34801288936627284,
      "grad_norm": 0.3600999116897583,
      "learning_rate": 6.932495846462261e-05,
      "loss": 0.3009,
      "step": 81
    },
    {
      "epoch": 0.3523093447905478,
      "grad_norm": 0.3982136845588684,
      "learning_rate": 6.855986244591104e-05,
      "loss": 0.3315,
      "step": 82
    },
    {
      "epoch": 0.35660580021482274,
      "grad_norm": 0.3989538550376892,
      "learning_rate": 6.778969234612584e-05,
      "loss": 0.3751,
      "step": 83
    },
    {
      "epoch": 0.3609022556390977,
      "grad_norm": 0.5305337905883789,
      "learning_rate": 6.701465872208216e-05,
      "loss": 0.4166,
      "step": 84
    },
    {
      "epoch": 0.3651987110633727,
      "grad_norm": 0.5150339007377625,
      "learning_rate": 6.623497346023418e-05,
      "loss": 0.3933,
      "step": 85
    },
    {
      "epoch": 0.3694951664876477,
      "grad_norm": 0.6410700678825378,
      "learning_rate": 6.545084971874738e-05,
      "loss": 0.5038,
      "step": 86
    },
    {
      "epoch": 0.37379162191192267,
      "grad_norm": 0.45256391167640686,
      "learning_rate": 6.466250186922325e-05,
      "loss": 0.3523,
      "step": 87
    },
    {
      "epoch": 0.37808807733619765,
      "grad_norm": 0.5341379046440125,
      "learning_rate": 6.387014543809223e-05,
      "loss": 0.3802,
      "step": 88
    },
    {
      "epoch": 0.38238453276047263,
      "grad_norm": 0.48598381876945496,
      "learning_rate": 6.307399704769099e-05,
      "loss": 0.4376,
      "step": 89
    },
    {
      "epoch": 0.3866809881847476,
      "grad_norm": 0.7272056341171265,
      "learning_rate": 6.227427435703997e-05,
      "loss": 0.4245,
      "step": 90
    },
    {
      "epoch": 0.39097744360902253,
      "grad_norm": 0.5760740041732788,
      "learning_rate": 6.147119600233758e-05,
      "loss": 0.3184,
      "step": 91
    },
    {
      "epoch": 0.3952738990332975,
      "grad_norm": 0.7006295323371887,
      "learning_rate": 6.066498153718735e-05,
      "loss": 0.4,
      "step": 92
    },
    {
      "epoch": 0.3995703544575725,
      "grad_norm": 0.635381281375885,
      "learning_rate": 5.985585137257401e-05,
      "loss": 0.4197,
      "step": 93
    },
    {
      "epoch": 0.4038668098818475,
      "grad_norm": 0.5333322882652283,
      "learning_rate": 5.90440267166055e-05,
      "loss": 0.3237,
      "step": 94
    },
    {
      "epoch": 0.40816326530612246,
      "grad_norm": 0.6799944043159485,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 0.352,
      "step": 95
    },
    {
      "epoch": 0.41245972073039744,
      "grad_norm": 0.6566717028617859,
      "learning_rate": 5.74131823855921e-05,
      "loss": 0.3699,
      "step": 96
    },
    {
      "epoch": 0.4167561761546724,
      "grad_norm": 0.7269343733787537,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 0.3615,
      "step": 97
    },
    {
      "epoch": 0.42105263157894735,
      "grad_norm": 0.982455313205719,
      "learning_rate": 5.577423184847932e-05,
      "loss": 0.5638,
      "step": 98
    },
    {
      "epoch": 0.4253490870032223,
      "grad_norm": 1.0673147439956665,
      "learning_rate": 5.495227651252315e-05,
      "loss": 0.3045,
      "step": 99
    },
    {
      "epoch": 0.4296455424274973,
      "grad_norm": 0.8240067362785339,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 0.2006,
      "step": 100
    },
    {
      "epoch": 0.4296455424274973,
      "eval_loss": 0.36948394775390625,
      "eval_runtime": 27.625,
      "eval_samples_per_second": 14.19,
      "eval_steps_per_second": 3.548,
      "step": 100
    },
    {
      "epoch": 0.4339419978517723,
      "grad_norm": 0.3018046021461487,
      "learning_rate": 5.330452921628497e-05,
      "loss": 0.2772,
      "step": 101
    },
    {
      "epoch": 0.43823845327604727,
      "grad_norm": 0.4544106721878052,
      "learning_rate": 5.247918773366112e-05,
      "loss": 0.3834,
      "step": 102
    },
    {
      "epoch": 0.44253490870032225,
      "grad_norm": 0.42616453766822815,
      "learning_rate": 5.165316846586541e-05,
      "loss": 0.4236,
      "step": 103
    },
    {
      "epoch": 0.44683136412459723,
      "grad_norm": 0.39616507291793823,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 0.3726,
      "step": 104
    },
    {
      "epoch": 0.45112781954887216,
      "grad_norm": 0.3530295789241791,
      "learning_rate": 5e-05,
      "loss": 0.3426,
      "step": 105
    },
    {
      "epoch": 0.45542427497314714,
      "grad_norm": 0.3598898947238922,
      "learning_rate": 4.917330276168208e-05,
      "loss": 0.3308,
      "step": 106
    },
    {
      "epoch": 0.4597207303974221,
      "grad_norm": 0.40101805329322815,
      "learning_rate": 4.834683153413459e-05,
      "loss": 0.3732,
      "step": 107
    },
    {
      "epoch": 0.4640171858216971,
      "grad_norm": 0.3908984065055847,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 0.397,
      "step": 108
    },
    {
      "epoch": 0.4683136412459721,
      "grad_norm": 0.3244476020336151,
      "learning_rate": 4.669547078371504e-05,
      "loss": 0.3096,
      "step": 109
    },
    {
      "epoch": 0.47261009667024706,
      "grad_norm": 0.3139345645904541,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 0.3013,
      "step": 110
    },
    {
      "epoch": 0.47690655209452204,
      "grad_norm": 0.330341637134552,
      "learning_rate": 4.504772348747687e-05,
      "loss": 0.2965,
      "step": 111
    },
    {
      "epoch": 0.48120300751879697,
      "grad_norm": 0.32211899757385254,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 0.2924,
      "step": 112
    },
    {
      "epoch": 0.48549946294307195,
      "grad_norm": 0.34253570437431335,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 0.2737,
      "step": 113
    },
    {
      "epoch": 0.4897959183673469,
      "grad_norm": 0.3796078562736511,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 0.3203,
      "step": 114
    },
    {
      "epoch": 0.4940923737916219,
      "grad_norm": 0.2806931436061859,
      "learning_rate": 4.17702704859633e-05,
      "loss": 0.2558,
      "step": 115
    },
    {
      "epoch": 0.4983888292158969,
      "grad_norm": 0.29928621649742126,
      "learning_rate": 4.095597328339452e-05,
      "loss": 0.2506,
      "step": 116
    },
    {
      "epoch": 0.5026852846401718,
      "grad_norm": 0.32902368903160095,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 0.2792,
      "step": 117
    },
    {
      "epoch": 0.5069817400644469,
      "grad_norm": 0.35551485419273376,
      "learning_rate": 3.933501846281267e-05,
      "loss": 0.2569,
      "step": 118
    },
    {
      "epoch": 0.5112781954887218,
      "grad_norm": 0.34765690565109253,
      "learning_rate": 3.852880399766243e-05,
      "loss": 0.2555,
      "step": 119
    },
    {
      "epoch": 0.5155746509129968,
      "grad_norm": 0.4137231707572937,
      "learning_rate": 3.772572564296005e-05,
      "loss": 0.3171,
      "step": 120
    },
    {
      "epoch": 0.5198711063372717,
      "grad_norm": 0.4357423186302185,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 0.3468,
      "step": 121
    },
    {
      "epoch": 0.5241675617615468,
      "grad_norm": 0.40948164463043213,
      "learning_rate": 3.612985456190778e-05,
      "loss": 0.2999,
      "step": 122
    },
    {
      "epoch": 0.5284640171858217,
      "grad_norm": 0.5223953723907471,
      "learning_rate": 3.533749813077677e-05,
      "loss": 0.3805,
      "step": 123
    },
    {
      "epoch": 0.5327604726100966,
      "grad_norm": 0.5923753976821899,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 0.4177,
      "step": 124
    },
    {
      "epoch": 0.5370569280343717,
      "grad_norm": 0.5544208884239197,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 0.3538,
      "step": 125
    },
    {
      "epoch": 0.5413533834586466,
      "grad_norm": 0.482532799243927,
      "learning_rate": 3.298534127791785e-05,
      "loss": 0.3155,
      "step": 126
    },
    {
      "epoch": 0.5456498388829216,
      "grad_norm": 0.5058260560035706,
      "learning_rate": 3.221030765387417e-05,
      "loss": 0.2917,
      "step": 127
    },
    {
      "epoch": 0.5499462943071965,
      "grad_norm": 0.4469728469848633,
      "learning_rate": 3.144013755408895e-05,
      "loss": 0.3291,
      "step": 128
    },
    {
      "epoch": 0.5542427497314716,
      "grad_norm": 0.4818762242794037,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 0.322,
      "step": 129
    },
    {
      "epoch": 0.5585392051557465,
      "grad_norm": 0.4870593249797821,
      "learning_rate": 2.991522876735154e-05,
      "loss": 0.3583,
      "step": 130
    },
    {
      "epoch": 0.5628356605800214,
      "grad_norm": 0.5139895677566528,
      "learning_rate": 2.916090697523549e-05,
      "loss": 0.3634,
      "step": 131
    },
    {
      "epoch": 0.5671321160042965,
      "grad_norm": 0.5461882948875427,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 0.3086,
      "step": 132
    },
    {
      "epoch": 0.5714285714285714,
      "grad_norm": 0.5530821084976196,
      "learning_rate": 2.766955965735968e-05,
      "loss": 0.3809,
      "step": 133
    },
    {
      "epoch": 0.5757250268528464,
      "grad_norm": 0.5471766591072083,
      "learning_rate": 2.693294185106562e-05,
      "loss": 0.3511,
      "step": 134
    },
    {
      "epoch": 0.5800214822771214,
      "grad_norm": 1.1091238260269165,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 0.3018,
      "step": 135
    },
    {
      "epoch": 0.5843179377013964,
      "grad_norm": 0.6000388264656067,
      "learning_rate": 2.547882480847461e-05,
      "loss": 0.4016,
      "step": 136
    },
    {
      "epoch": 0.5886143931256713,
      "grad_norm": 0.6467744708061218,
      "learning_rate": 2.476172311325783e-05,
      "loss": 0.3761,
      "step": 137
    },
    {
      "epoch": 0.5929108485499462,
      "grad_norm": 0.5213028192520142,
      "learning_rate": 2.405152131093926e-05,
      "loss": 0.3199,
      "step": 138
    },
    {
      "epoch": 0.5972073039742213,
      "grad_norm": 0.5980802774429321,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 0.2639,
      "step": 139
    },
    {
      "epoch": 0.6015037593984962,
      "grad_norm": 0.5194615721702576,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 0.3044,
      "step": 140
    },
    {
      "epoch": 0.6058002148227712,
      "grad_norm": 0.6527299880981445,
      "learning_rate": 2.196424713241637e-05,
      "loss": 0.3249,
      "step": 141
    },
    {
      "epoch": 0.6100966702470462,
      "grad_norm": 0.5338525772094727,
      "learning_rate": 2.128356686585282e-05,
      "loss": 0.2942,
      "step": 142
    },
    {
      "epoch": 0.6143931256713212,
      "grad_norm": 0.7685671448707581,
      "learning_rate": 2.061073738537635e-05,
      "loss": 0.4126,
      "step": 143
    },
    {
      "epoch": 0.6186895810955961,
      "grad_norm": 0.7643443942070007,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 0.3202,
      "step": 144
    },
    {
      "epoch": 0.6229860365198711,
      "grad_norm": 0.6941141486167908,
      "learning_rate": 1.928936436551661e-05,
      "loss": 0.3295,
      "step": 145
    },
    {
      "epoch": 0.6272824919441461,
      "grad_norm": 0.9832645058631897,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 0.383,
      "step": 146
    },
    {
      "epoch": 0.631578947368421,
      "grad_norm": 0.9147282242774963,
      "learning_rate": 1.800157297483417e-05,
      "loss": 0.3818,
      "step": 147
    },
    {
      "epoch": 0.635875402792696,
      "grad_norm": 0.786456823348999,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 0.2811,
      "step": 148
    },
    {
      "epoch": 0.640171858216971,
      "grad_norm": 0.8777568936347961,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 0.3394,
      "step": 149
    },
    {
      "epoch": 0.644468313641246,
      "grad_norm": 1.1083598136901855,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 0.2503,
      "step": 150
    },
    {
      "epoch": 0.644468313641246,
      "eval_loss": 0.29443585872650146,
      "eval_runtime": 27.5946,
      "eval_samples_per_second": 14.206,
      "eval_steps_per_second": 3.551,
      "step": 150
    },
    {
      "epoch": 0.6487647690655209,
      "grad_norm": 0.22200804948806763,
      "learning_rate": 1.553232954407171e-05,
      "loss": 0.1898,
      "step": 151
    },
    {
      "epoch": 0.6530612244897959,
      "grad_norm": 0.3020473122596741,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 0.2586,
      "step": 152
    },
    {
      "epoch": 0.6573576799140709,
      "grad_norm": 0.30981576442718506,
      "learning_rate": 1.435357758543015e-05,
      "loss": 0.2909,
      "step": 153
    },
    {
      "epoch": 0.6616541353383458,
      "grad_norm": 0.315384179353714,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 0.2658,
      "step": 154
    },
    {
      "epoch": 0.6659505907626209,
      "grad_norm": 0.33048534393310547,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 0.2657,
      "step": 155
    },
    {
      "epoch": 0.6702470461868958,
      "grad_norm": 0.32182732224464417,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 0.2593,
      "step": 156
    },
    {
      "epoch": 0.6745435016111708,
      "grad_norm": 0.3142335116863251,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 0.248,
      "step": 157
    },
    {
      "epoch": 0.6788399570354458,
      "grad_norm": 0.3597651422023773,
      "learning_rate": 1.157994445715706e-05,
      "loss": 0.2663,
      "step": 158
    },
    {
      "epoch": 0.6831364124597207,
      "grad_norm": 0.3391364812850952,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 0.3082,
      "step": 159
    },
    {
      "epoch": 0.6874328678839957,
      "grad_norm": 0.37021851539611816,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 0.3347,
      "step": 160
    },
    {
      "epoch": 0.6917293233082706,
      "grad_norm": 0.3648050129413605,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 0.25,
      "step": 161
    },
    {
      "epoch": 0.6960257787325457,
      "grad_norm": 0.38995301723480225,
      "learning_rate": 9.549150281252633e-06,
      "loss": 0.3182,
      "step": 162
    },
    {
      "epoch": 0.7003222341568206,
      "grad_norm": 0.3318188488483429,
      "learning_rate": 9.068759265665384e-06,
      "loss": 0.2502,
      "step": 163
    },
    {
      "epoch": 0.7046186895810956,
      "grad_norm": 0.4105289578437805,
      "learning_rate": 8.599558442598998e-06,
      "loss": 0.2879,
      "step": 164
    },
    {
      "epoch": 0.7089151450053706,
      "grad_norm": 0.41487324237823486,
      "learning_rate": 8.141676086873572e-06,
      "loss": 0.2896,
      "step": 165
    },
    {
      "epoch": 0.7132116004296455,
      "grad_norm": 0.4566102623939514,
      "learning_rate": 7.695237378953223e-06,
      "loss": 0.305,
      "step": 166
    },
    {
      "epoch": 0.7175080558539205,
      "grad_norm": 0.41210034489631653,
      "learning_rate": 7.260364370723044e-06,
      "loss": 0.2921,
      "step": 167
    },
    {
      "epoch": 0.7218045112781954,
      "grad_norm": 0.3816104531288147,
      "learning_rate": 6.837175952121306e-06,
      "loss": 0.2386,
      "step": 168
    },
    {
      "epoch": 0.7261009667024705,
      "grad_norm": 0.3759080171585083,
      "learning_rate": 6.425787818636131e-06,
      "loss": 0.2741,
      "step": 169
    },
    {
      "epoch": 0.7303974221267454,
      "grad_norm": 0.4245615303516388,
      "learning_rate": 6.026312439675552e-06,
      "loss": 0.3129,
      "step": 170
    },
    {
      "epoch": 0.7346938775510204,
      "grad_norm": 0.3779868185520172,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 0.2264,
      "step": 171
    },
    {
      "epoch": 0.7389903329752954,
      "grad_norm": 0.4320870339870453,
      "learning_rate": 5.263533508961827e-06,
      "loss": 0.2906,
      "step": 172
    },
    {
      "epoch": 0.7432867883995704,
      "grad_norm": 0.44205906987190247,
      "learning_rate": 4.900438493352055e-06,
      "loss": 0.2615,
      "step": 173
    },
    {
      "epoch": 0.7475832438238453,
      "grad_norm": 0.42238447070121765,
      "learning_rate": 4.549673247541875e-06,
      "loss": 0.2557,
      "step": 174
    },
    {
      "epoch": 0.7518796992481203,
      "grad_norm": 0.4558790326118469,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 0.2831,
      "step": 175
    },
    {
      "epoch": 0.7561761546723953,
      "grad_norm": 0.46227309107780457,
      "learning_rate": 3.885512251130763e-06,
      "loss": 0.2811,
      "step": 176
    },
    {
      "epoch": 0.7604726100966702,
      "grad_norm": 0.44037771224975586,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 0.2352,
      "step": 177
    },
    {
      "epoch": 0.7647690655209453,
      "grad_norm": 0.5309369564056396,
      "learning_rate": 3.271776770026963e-06,
      "loss": 0.2967,
      "step": 178
    },
    {
      "epoch": 0.7690655209452202,
      "grad_norm": 0.4705249071121216,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 0.2921,
      "step": 179
    },
    {
      "epoch": 0.7733619763694952,
      "grad_norm": 0.5692561268806458,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 0.295,
      "step": 180
    },
    {
      "epoch": 0.7776584317937701,
      "grad_norm": 0.4589831531047821,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 0.2454,
      "step": 181
    },
    {
      "epoch": 0.7819548872180451,
      "grad_norm": 0.5403589010238647,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 0.3231,
      "step": 182
    },
    {
      "epoch": 0.7862513426423201,
      "grad_norm": 0.5366402268409729,
      "learning_rate": 1.962316193157593e-06,
      "loss": 0.259,
      "step": 183
    },
    {
      "epoch": 0.790547798066595,
      "grad_norm": 0.6617404818534851,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 0.3531,
      "step": 184
    },
    {
      "epoch": 0.7948442534908701,
      "grad_norm": 0.6814762949943542,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 0.3581,
      "step": 185
    },
    {
      "epoch": 0.799140708915145,
      "grad_norm": 0.608530580997467,
      "learning_rate": 1.333670137599713e-06,
      "loss": 0.2712,
      "step": 186
    },
    {
      "epoch": 0.80343716433942,
      "grad_norm": 0.5977616906166077,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 0.2808,
      "step": 187
    },
    {
      "epoch": 0.807733619763695,
      "grad_norm": 0.5270123481750488,
      "learning_rate": 9.810017062595322e-07,
      "loss": 0.2795,
      "step": 188
    },
    {
      "epoch": 0.8120300751879699,
      "grad_norm": 0.4959504306316376,
      "learning_rate": 8.247462563808817e-07,
      "loss": 0.2393,
      "step": 189
    },
    {
      "epoch": 0.8163265306122449,
      "grad_norm": 0.6505959630012512,
      "learning_rate": 6.819348298638839e-07,
      "loss": 0.3502,
      "step": 190
    },
    {
      "epoch": 0.8206229860365198,
      "grad_norm": 0.6694881319999695,
      "learning_rate": 5.526064699265753e-07,
      "loss": 0.3087,
      "step": 191
    },
    {
      "epoch": 0.8249194414607949,
      "grad_norm": 0.535914421081543,
      "learning_rate": 4.367965336512403e-07,
      "loss": 0.2231,
      "step": 192
    },
    {
      "epoch": 0.8292158968850698,
      "grad_norm": 0.6710368394851685,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 0.2621,
      "step": 193
    },
    {
      "epoch": 0.8335123523093448,
      "grad_norm": 0.7072320580482483,
      "learning_rate": 2.458548727494292e-07,
      "loss": 0.281,
      "step": 194
    },
    {
      "epoch": 0.8378088077336198,
      "grad_norm": 0.7714508175849915,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 0.2793,
      "step": 195
    },
    {
      "epoch": 0.8421052631578947,
      "grad_norm": 0.6197285056114197,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 0.2679,
      "step": 196
    },
    {
      "epoch": 0.8464017185821697,
      "grad_norm": 0.7446801662445068,
      "learning_rate": 6.150154258476315e-08,
      "loss": 0.3773,
      "step": 197
    },
    {
      "epoch": 0.8506981740064447,
      "grad_norm": 0.7205499410629272,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 0.316,
      "step": 198
    },
    {
      "epoch": 0.8549946294307197,
      "grad_norm": 0.8960843086242676,
      "learning_rate": 6.834750376549792e-09,
      "loss": 0.3137,
      "step": 199
    },
    {
      "epoch": 0.8592910848549946,
      "grad_norm": 0.6217831373214722,
      "learning_rate": 0.0,
      "loss": 0.1739,
      "step": 200
    },
    {
      "epoch": 0.8592910848549946,
      "eval_loss": 0.28383296728134155,
      "eval_runtime": 27.6227,
      "eval_samples_per_second": 14.191,
      "eval_steps_per_second": 3.548,
      "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": 2.661403865382912e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}