{
  "best_metric": 0.5548620223999023,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.8016032064128257,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.004008016032064128,
      "grad_norm": 12.893512725830078,
      "learning_rate": 1e-05,
      "loss": 4.9938,
      "step": 1
    },
    {
      "epoch": 0.004008016032064128,
      "eval_loss": 1.1268008947372437,
      "eval_runtime": 88.7767,
      "eval_samples_per_second": 4.742,
      "eval_steps_per_second": 1.194,
      "step": 1
    },
    {
      "epoch": 0.008016032064128256,
      "grad_norm": 15.58875560760498,
      "learning_rate": 2e-05,
      "loss": 4.382,
      "step": 2
    },
    {
      "epoch": 0.012024048096192385,
      "grad_norm": 12.056736946105957,
      "learning_rate": 3e-05,
      "loss": 4.0357,
      "step": 3
    },
    {
      "epoch": 0.01603206412825651,
      "grad_norm": 9.07699966430664,
      "learning_rate": 4e-05,
      "loss": 3.6441,
      "step": 4
    },
    {
      "epoch": 0.02004008016032064,
      "grad_norm": 13.85738754272461,
      "learning_rate": 5e-05,
      "loss": 3.6025,
      "step": 5
    },
    {
      "epoch": 0.02404809619238477,
      "grad_norm": 8.968578338623047,
      "learning_rate": 6e-05,
      "loss": 3.2452,
      "step": 6
    },
    {
      "epoch": 0.028056112224448898,
      "grad_norm": 7.22355318069458,
      "learning_rate": 7e-05,
      "loss": 3.2993,
      "step": 7
    },
    {
      "epoch": 0.03206412825651302,
      "grad_norm": 6.018855571746826,
      "learning_rate": 8e-05,
      "loss": 3.12,
      "step": 8
    },
    {
      "epoch": 0.036072144288577156,
      "grad_norm": 6.162630081176758,
      "learning_rate": 9e-05,
      "loss": 3.0909,
      "step": 9
    },
    {
      "epoch": 0.04008016032064128,
      "grad_norm": 4.708962440490723,
      "learning_rate": 0.0001,
      "loss": 2.8983,
      "step": 10
    },
    {
      "epoch": 0.04408817635270541,
      "grad_norm": 5.808905124664307,
      "learning_rate": 9.999316524962345e-05,
      "loss": 2.771,
      "step": 11
    },
    {
      "epoch": 0.04809619238476954,
      "grad_norm": 5.182327747344971,
      "learning_rate": 9.997266286704631e-05,
      "loss": 2.7097,
      "step": 12
    },
    {
      "epoch": 0.052104208416833664,
      "grad_norm": 4.8496623039245605,
      "learning_rate": 9.993849845741524e-05,
      "loss": 2.6533,
      "step": 13
    },
    {
      "epoch": 0.056112224448897796,
      "grad_norm": 5.305124759674072,
      "learning_rate": 9.989068136093873e-05,
      "loss": 2.9562,
      "step": 14
    },
    {
      "epoch": 0.06012024048096192,
      "grad_norm": 4.643880844116211,
      "learning_rate": 9.98292246503335e-05,
      "loss": 2.7388,
      "step": 15
    },
    {
      "epoch": 0.06412825651302605,
      "grad_norm": 4.473149299621582,
      "learning_rate": 9.975414512725057e-05,
      "loss": 2.6393,
      "step": 16
    },
    {
      "epoch": 0.06813627254509018,
      "grad_norm": 4.139369010925293,
      "learning_rate": 9.966546331768191e-05,
      "loss": 2.5498,
      "step": 17
    },
    {
      "epoch": 0.07214428857715431,
      "grad_norm": 4.048318862915039,
      "learning_rate": 9.956320346634876e-05,
      "loss": 2.4195,
      "step": 18
    },
    {
      "epoch": 0.07615230460921844,
      "grad_norm": 4.301382064819336,
      "learning_rate": 9.944739353007344e-05,
      "loss": 2.7098,
      "step": 19
    },
    {
      "epoch": 0.08016032064128256,
      "grad_norm": 4.245019435882568,
      "learning_rate": 9.931806517013612e-05,
      "loss": 2.5701,
      "step": 20
    },
    {
      "epoch": 0.0841683366733467,
      "grad_norm": 4.314387321472168,
      "learning_rate": 9.917525374361912e-05,
      "loss": 2.6125,
      "step": 21
    },
    {
      "epoch": 0.08817635270541083,
      "grad_norm": 3.9001383781433105,
      "learning_rate": 9.901899829374047e-05,
      "loss": 2.5686,
      "step": 22
    },
    {
      "epoch": 0.09218436873747494,
      "grad_norm": 4.101955413818359,
      "learning_rate": 9.884934153917997e-05,
      "loss": 2.6433,
      "step": 23
    },
    {
      "epoch": 0.09619238476953908,
      "grad_norm": 4.369626045227051,
      "learning_rate": 9.86663298624003e-05,
      "loss": 2.5056,
      "step": 24
    },
    {
      "epoch": 0.10020040080160321,
      "grad_norm": 4.029824733734131,
      "learning_rate": 9.847001329696653e-05,
      "loss": 2.5307,
      "step": 25
    },
    {
      "epoch": 0.10420841683366733,
      "grad_norm": 4.034836292266846,
      "learning_rate": 9.826044551386744e-05,
      "loss": 2.5286,
      "step": 26
    },
    {
      "epoch": 0.10821643286573146,
      "grad_norm": 4.211627960205078,
      "learning_rate": 9.803768380684242e-05,
      "loss": 2.5892,
      "step": 27
    },
    {
      "epoch": 0.11222444889779559,
      "grad_norm": 3.898348093032837,
      "learning_rate": 9.780178907671789e-05,
      "loss": 2.567,
      "step": 28
    },
    {
      "epoch": 0.11623246492985972,
      "grad_norm": 3.6137986183166504,
      "learning_rate": 9.755282581475769e-05,
      "loss": 2.5256,
      "step": 29
    },
    {
      "epoch": 0.12024048096192384,
      "grad_norm": 3.7446653842926025,
      "learning_rate": 9.729086208503174e-05,
      "loss": 2.5732,
      "step": 30
    },
    {
      "epoch": 0.12424849699398798,
      "grad_norm": 6.043440341949463,
      "learning_rate": 9.701596950580806e-05,
      "loss": 2.6398,
      "step": 31
    },
    {
      "epoch": 0.1282565130260521,
      "grad_norm": 4.026174545288086,
      "learning_rate": 9.672822322997305e-05,
      "loss": 2.6369,
      "step": 32
    },
    {
      "epoch": 0.13226452905811623,
      "grad_norm": 3.9922075271606445,
      "learning_rate": 9.642770192448536e-05,
      "loss": 2.4741,
      "step": 33
    },
    {
      "epoch": 0.13627254509018036,
      "grad_norm": 3.7829277515411377,
      "learning_rate": 9.611448774886924e-05,
      "loss": 2.6832,
      "step": 34
    },
    {
      "epoch": 0.1402805611222445,
      "grad_norm": 3.7771103382110596,
      "learning_rate": 9.578866633275288e-05,
      "loss": 2.4489,
      "step": 35
    },
    {
      "epoch": 0.14428857715430862,
      "grad_norm": 3.7771096229553223,
      "learning_rate": 9.545032675245813e-05,
      "loss": 2.6347,
      "step": 36
    },
    {
      "epoch": 0.14829659318637275,
      "grad_norm": 3.7754852771759033,
      "learning_rate": 9.509956150664796e-05,
      "loss": 2.6407,
      "step": 37
    },
    {
      "epoch": 0.1523046092184369,
      "grad_norm": 3.922776222229004,
      "learning_rate": 9.473646649103818e-05,
      "loss": 2.6482,
      "step": 38
    },
    {
      "epoch": 0.156312625250501,
      "grad_norm": 3.9672491550445557,
      "learning_rate": 9.43611409721806e-05,
      "loss": 2.5384,
      "step": 39
    },
    {
      "epoch": 0.16032064128256512,
      "grad_norm": 3.7962536811828613,
      "learning_rate": 9.397368756032445e-05,
      "loss": 2.6941,
      "step": 40
    },
    {
      "epoch": 0.16432865731462926,
      "grad_norm": 3.718618154525757,
      "learning_rate": 9.357421218136386e-05,
      "loss": 2.7276,
      "step": 41
    },
    {
      "epoch": 0.1683366733466934,
      "grad_norm": 3.693268060684204,
      "learning_rate": 9.316282404787871e-05,
      "loss": 2.4831,
      "step": 42
    },
    {
      "epoch": 0.17234468937875752,
      "grad_norm": 3.918449640274048,
      "learning_rate": 9.273963562927695e-05,
      "loss": 2.5511,
      "step": 43
    },
    {
      "epoch": 0.17635270541082165,
      "grad_norm": 4.105891227722168,
      "learning_rate": 9.230476262104677e-05,
      "loss": 2.6428,
      "step": 44
    },
    {
      "epoch": 0.18036072144288579,
      "grad_norm": 4.407500267028809,
      "learning_rate": 9.185832391312644e-05,
      "loss": 2.763,
      "step": 45
    },
    {
      "epoch": 0.1843687374749499,
      "grad_norm": 4.594778537750244,
      "learning_rate": 9.140044155740101e-05,
      "loss": 2.456,
      "step": 46
    },
    {
      "epoch": 0.18837675350701402,
      "grad_norm": 4.5602898597717285,
      "learning_rate": 9.093124073433463e-05,
      "loss": 2.481,
      "step": 47
    },
    {
      "epoch": 0.19238476953907815,
      "grad_norm": 4.05642557144165,
      "learning_rate": 9.045084971874738e-05,
      "loss": 2.5056,
      "step": 48
    },
    {
      "epoch": 0.1963927855711423,
      "grad_norm": 4.247261047363281,
      "learning_rate": 8.995939984474624e-05,
      "loss": 2.775,
      "step": 49
    },
    {
      "epoch": 0.20040080160320642,
      "grad_norm": 4.612782001495361,
      "learning_rate": 8.945702546981969e-05,
      "loss": 2.7183,
      "step": 50
    },
    {
      "epoch": 0.20040080160320642,
      "eval_loss": 0.6524797677993774,
      "eval_runtime": 89.7961,
      "eval_samples_per_second": 4.688,
      "eval_steps_per_second": 1.18,
      "step": 50
    },
    {
      "epoch": 0.20440881763527055,
      "grad_norm": 4.573088645935059,
      "learning_rate": 8.894386393810563e-05,
      "loss": 4.6625,
      "step": 51
    },
    {
      "epoch": 0.20841683366733466,
      "grad_norm": 3.8097169399261475,
      "learning_rate": 8.842005554284296e-05,
      "loss": 2.5473,
      "step": 52
    },
    {
      "epoch": 0.2124248496993988,
      "grad_norm": 3.084719657897949,
      "learning_rate": 8.788574348801675e-05,
      "loss": 2.3826,
      "step": 53
    },
    {
      "epoch": 0.21643286573146292,
      "grad_norm": 3.2088465690612793,
      "learning_rate": 8.73410738492077e-05,
      "loss": 2.4925,
      "step": 54
    },
    {
      "epoch": 0.22044088176352705,
      "grad_norm": 3.3900623321533203,
      "learning_rate": 8.678619553365659e-05,
      "loss": 2.2819,
      "step": 55
    },
    {
      "epoch": 0.22444889779559118,
      "grad_norm": 3.358004331588745,
      "learning_rate": 8.622126023955446e-05,
      "loss": 2.3587,
      "step": 56
    },
    {
      "epoch": 0.22845691382765532,
      "grad_norm": 3.439387798309326,
      "learning_rate": 8.564642241456986e-05,
      "loss": 2.3835,
      "step": 57
    },
    {
      "epoch": 0.23246492985971945,
      "grad_norm": 3.3043360710144043,
      "learning_rate": 8.506183921362443e-05,
      "loss": 2.3327,
      "step": 58
    },
    {
      "epoch": 0.23647294589178355,
      "grad_norm": 3.154406785964966,
      "learning_rate": 8.44676704559283e-05,
      "loss": 2.3957,
      "step": 59
    },
    {
      "epoch": 0.24048096192384769,
      "grad_norm": 3.2822458744049072,
      "learning_rate": 8.386407858128706e-05,
      "loss": 2.2912,
      "step": 60
    },
    {
      "epoch": 0.24448897795591182,
      "grad_norm": 3.0163843631744385,
      "learning_rate": 8.32512286056924e-05,
      "loss": 2.3684,
      "step": 61
    },
    {
      "epoch": 0.24849699398797595,
      "grad_norm": 3.4419105052948,
      "learning_rate": 8.262928807620843e-05,
      "loss": 2.559,
      "step": 62
    },
    {
      "epoch": 0.25250501002004005,
      "grad_norm": 3.1099562644958496,
      "learning_rate": 8.199842702516583e-05,
      "loss": 2.3492,
      "step": 63
    },
    {
      "epoch": 0.2565130260521042,
      "grad_norm": 3.147329092025757,
      "learning_rate": 8.135881792367686e-05,
      "loss": 2.2366,
      "step": 64
    },
    {
      "epoch": 0.2605210420841683,
      "grad_norm": 3.2316110134124756,
      "learning_rate": 8.07106356344834e-05,
      "loss": 2.3532,
      "step": 65
    },
    {
      "epoch": 0.26452905811623245,
      "grad_norm": 3.0555813312530518,
      "learning_rate": 8.005405736415126e-05,
      "loss": 2.2404,
      "step": 66
    },
    {
      "epoch": 0.2685370741482966,
      "grad_norm": 3.0740108489990234,
      "learning_rate": 7.938926261462366e-05,
      "loss": 2.3033,
      "step": 67
    },
    {
      "epoch": 0.2725450901803607,
      "grad_norm": 3.300424337387085,
      "learning_rate": 7.871643313414718e-05,
      "loss": 2.3363,
      "step": 68
    },
    {
      "epoch": 0.27655310621242485,
      "grad_norm": 2.976027488708496,
      "learning_rate": 7.803575286758364e-05,
      "loss": 2.2247,
      "step": 69
    },
    {
      "epoch": 0.280561122244489,
      "grad_norm": 3.264178991317749,
      "learning_rate": 7.734740790612136e-05,
      "loss": 2.2702,
      "step": 70
    },
    {
      "epoch": 0.2845691382765531,
      "grad_norm": 3.1751465797424316,
      "learning_rate": 7.66515864363997e-05,
      "loss": 2.2457,
      "step": 71
    },
    {
      "epoch": 0.28857715430861725,
      "grad_norm": 3.1859261989593506,
      "learning_rate": 7.594847868906076e-05,
      "loss": 2.1528,
      "step": 72
    },
    {
      "epoch": 0.2925851703406814,
      "grad_norm": 3.2604966163635254,
      "learning_rate": 7.52382768867422e-05,
      "loss": 2.2474,
      "step": 73
    },
    {
      "epoch": 0.2965931863727455,
      "grad_norm": 3.3649673461914062,
      "learning_rate": 7.452117519152542e-05,
      "loss": 2.3483,
      "step": 74
    },
    {
      "epoch": 0.30060120240480964,
      "grad_norm": 3.1647846698760986,
      "learning_rate": 7.379736965185368e-05,
      "loss": 2.2055,
      "step": 75
    },
    {
      "epoch": 0.3046092184368738,
      "grad_norm": 3.2224955558776855,
      "learning_rate": 7.30670581489344e-05,
      "loss": 2.2656,
      "step": 76
    },
    {
      "epoch": 0.30861723446893785,
      "grad_norm": 3.2283167839050293,
      "learning_rate": 7.233044034264034e-05,
      "loss": 2.2712,
      "step": 77
    },
    {
      "epoch": 0.312625250501002,
      "grad_norm": 3.196347951889038,
      "learning_rate": 7.158771761692464e-05,
      "loss": 2.4321,
      "step": 78
    },
    {
      "epoch": 0.3166332665330661,
      "grad_norm": 3.3318557739257812,
      "learning_rate": 7.083909302476453e-05,
      "loss": 2.4075,
      "step": 79
    },
    {
      "epoch": 0.32064128256513025,
      "grad_norm": 3.2446036338806152,
      "learning_rate": 7.008477123264848e-05,
      "loss": 2.2987,
      "step": 80
    },
    {
      "epoch": 0.3246492985971944,
      "grad_norm": 3.335087299346924,
      "learning_rate": 6.932495846462261e-05,
      "loss": 2.6274,
      "step": 81
    },
    {
      "epoch": 0.3286573146292585,
      "grad_norm": 3.180757761001587,
      "learning_rate": 6.855986244591104e-05,
      "loss": 2.5807,
      "step": 82
    },
    {
      "epoch": 0.33266533066132264,
      "grad_norm": 3.3153293132781982,
      "learning_rate": 6.778969234612584e-05,
      "loss": 2.3855,
      "step": 83
    },
    {
      "epoch": 0.3366733466933868,
      "grad_norm": 3.253328561782837,
      "learning_rate": 6.701465872208216e-05,
      "loss": 2.3209,
      "step": 84
    },
    {
      "epoch": 0.3406813627254509,
      "grad_norm": 3.134979248046875,
      "learning_rate": 6.623497346023418e-05,
      "loss": 2.413,
      "step": 85
    },
    {
      "epoch": 0.34468937875751504,
      "grad_norm": 3.4511048793792725,
      "learning_rate": 6.545084971874738e-05,
      "loss": 2.4383,
      "step": 86
    },
    {
      "epoch": 0.3486973947895792,
      "grad_norm": 3.4455432891845703,
      "learning_rate": 6.466250186922325e-05,
      "loss": 2.3346,
      "step": 87
    },
    {
      "epoch": 0.3527054108216433,
      "grad_norm": 3.403756618499756,
      "learning_rate": 6.387014543809223e-05,
      "loss": 2.4421,
      "step": 88
    },
    {
      "epoch": 0.35671342685370744,
      "grad_norm": 3.2927913665771484,
      "learning_rate": 6.307399704769099e-05,
      "loss": 2.4442,
      "step": 89
    },
    {
      "epoch": 0.36072144288577157,
      "grad_norm": 3.3136186599731445,
      "learning_rate": 6.227427435703997e-05,
      "loss": 2.3426,
      "step": 90
    },
    {
      "epoch": 0.36472945891783565,
      "grad_norm": 3.2006802558898926,
      "learning_rate": 6.147119600233758e-05,
      "loss": 2.4803,
      "step": 91
    },
    {
      "epoch": 0.3687374749498998,
      "grad_norm": 3.3863649368286133,
      "learning_rate": 6.066498153718735e-05,
      "loss": 2.3128,
      "step": 92
    },
    {
      "epoch": 0.3727454909819639,
      "grad_norm": 3.251427173614502,
      "learning_rate": 5.985585137257401e-05,
      "loss": 2.4993,
      "step": 93
    },
    {
      "epoch": 0.37675350701402804,
      "grad_norm": 3.2351865768432617,
      "learning_rate": 5.90440267166055e-05,
      "loss": 2.307,
      "step": 94
    },
    {
      "epoch": 0.3807615230460922,
      "grad_norm": 3.3223328590393066,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 2.3093,
      "step": 95
    },
    {
      "epoch": 0.3847695390781563,
      "grad_norm": 3.440704822540283,
      "learning_rate": 5.74131823855921e-05,
      "loss": 2.394,
      "step": 96
    },
    {
      "epoch": 0.38877755511022044,
      "grad_norm": 3.421598434448242,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 2.4465,
      "step": 97
    },
    {
      "epoch": 0.3927855711422846,
      "grad_norm": 3.545168399810791,
      "learning_rate": 5.577423184847932e-05,
      "loss": 2.3908,
      "step": 98
    },
    {
      "epoch": 0.3967935871743487,
      "grad_norm": 3.563213348388672,
      "learning_rate": 5.495227651252315e-05,
      "loss": 2.3341,
      "step": 99
    },
    {
      "epoch": 0.40080160320641284,
      "grad_norm": 3.8546183109283447,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 2.5225,
      "step": 100
    },
    {
      "epoch": 0.40080160320641284,
      "eval_loss": 0.6069819927215576,
      "eval_runtime": 89.832,
      "eval_samples_per_second": 4.687,
      "eval_steps_per_second": 1.18,
      "step": 100
    },
    {
      "epoch": 0.40480961923847697,
      "grad_norm": 3.960508108139038,
      "learning_rate": 5.330452921628497e-05,
      "loss": 2.6546,
      "step": 101
    },
    {
      "epoch": 0.4088176352705411,
      "grad_norm": 3.3293697834014893,
      "learning_rate": 5.247918773366112e-05,
      "loss": 2.3414,
      "step": 102
    },
    {
      "epoch": 0.41282565130260523,
      "grad_norm": 2.85752534866333,
      "learning_rate": 5.165316846586541e-05,
      "loss": 2.2888,
      "step": 103
    },
    {
      "epoch": 0.4168336673346693,
      "grad_norm": 2.7095561027526855,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 2.3482,
      "step": 104
    },
    {
      "epoch": 0.42084168336673344,
      "grad_norm": 2.7856271266937256,
      "learning_rate": 5e-05,
      "loss": 2.1992,
      "step": 105
    },
    {
      "epoch": 0.4248496993987976,
      "grad_norm": 2.881302833557129,
      "learning_rate": 4.917330276168208e-05,
      "loss": 2.3239,
      "step": 106
    },
    {
      "epoch": 0.4288577154308617,
      "grad_norm": 3.0186617374420166,
      "learning_rate": 4.834683153413459e-05,
      "loss": 2.2834,
      "step": 107
    },
    {
      "epoch": 0.43286573146292584,
      "grad_norm": 2.964176893234253,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 2.3446,
      "step": 108
    },
    {
      "epoch": 0.43687374749499,
      "grad_norm": 2.7934885025024414,
      "learning_rate": 4.669547078371504e-05,
      "loss": 2.1322,
      "step": 109
    },
    {
      "epoch": 0.4408817635270541,
      "grad_norm": 2.769521713256836,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 2.2031,
      "step": 110
    },
    {
      "epoch": 0.44488977955911824,
      "grad_norm": 2.8491127490997314,
      "learning_rate": 4.504772348747687e-05,
      "loss": 2.1547,
      "step": 111
    },
    {
      "epoch": 0.44889779559118237,
      "grad_norm": 2.9389901161193848,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 2.168,
      "step": 112
    },
    {
      "epoch": 0.4529058116232465,
      "grad_norm": 3.47145676612854,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 2.2364,
      "step": 113
    },
    {
      "epoch": 0.45691382765531063,
      "grad_norm": 2.836869239807129,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 2.2266,
      "step": 114
    },
    {
      "epoch": 0.46092184368737477,
      "grad_norm": 2.892146587371826,
      "learning_rate": 4.17702704859633e-05,
      "loss": 2.2313,
      "step": 115
    },
    {
      "epoch": 0.4649298597194389,
      "grad_norm": 2.9002685546875,
      "learning_rate": 4.095597328339452e-05,
      "loss": 2.3847,
      "step": 116
    },
    {
      "epoch": 0.46893787575150303,
      "grad_norm": 2.843989133834839,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 2.1362,
      "step": 117
    },
    {
      "epoch": 0.4729458917835671,
      "grad_norm": 2.9691362380981445,
      "learning_rate": 3.933501846281267e-05,
      "loss": 2.1874,
      "step": 118
    },
    {
      "epoch": 0.47695390781563124,
      "grad_norm": 2.833814859390259,
      "learning_rate": 3.852880399766243e-05,
      "loss": 2.1156,
      "step": 119
    },
    {
      "epoch": 0.48096192384769537,
      "grad_norm": 2.9893736839294434,
      "learning_rate": 3.772572564296005e-05,
      "loss": 2.0828,
      "step": 120
    },
    {
      "epoch": 0.4849699398797595,
      "grad_norm": 2.917656660079956,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 2.1061,
      "step": 121
    },
    {
      "epoch": 0.48897795591182364,
      "grad_norm": 3.096787691116333,
      "learning_rate": 3.612985456190778e-05,
      "loss": 2.122,
      "step": 122
    },
    {
      "epoch": 0.49298597194388777,
      "grad_norm": 3.0050277709960938,
      "learning_rate": 3.533749813077677e-05,
      "loss": 2.2817,
      "step": 123
    },
    {
      "epoch": 0.4969939879759519,
      "grad_norm": 3.023724317550659,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 2.1993,
      "step": 124
    },
    {
      "epoch": 0.501002004008016,
      "grad_norm": 2.9125618934631348,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 2.1356,
      "step": 125
    },
    {
      "epoch": 0.5050100200400801,
      "grad_norm": 3.062142848968506,
      "learning_rate": 3.298534127791785e-05,
      "loss": 2.357,
      "step": 126
    },
    {
      "epoch": 0.5090180360721442,
      "grad_norm": 2.9896559715270996,
      "learning_rate": 3.221030765387417e-05,
      "loss": 2.249,
      "step": 127
    },
    {
      "epoch": 0.5130260521042084,
      "grad_norm": 2.967043161392212,
      "learning_rate": 3.144013755408895e-05,
      "loss": 2.3748,
      "step": 128
    },
    {
      "epoch": 0.5170340681362725,
      "grad_norm": 2.954469680786133,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 2.3238,
      "step": 129
    },
    {
      "epoch": 0.5210420841683366,
      "grad_norm": 2.980943202972412,
      "learning_rate": 2.991522876735154e-05,
      "loss": 2.3134,
      "step": 130
    },
    {
      "epoch": 0.5250501002004008,
      "grad_norm": 3.0086441040039062,
      "learning_rate": 2.916090697523549e-05,
      "loss": 2.4643,
      "step": 131
    },
    {
      "epoch": 0.5290581162324649,
      "grad_norm": 2.8430187702178955,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 2.2569,
      "step": 132
    },
    {
      "epoch": 0.533066132264529,
      "grad_norm": 2.837242603302002,
      "learning_rate": 2.766955965735968e-05,
      "loss": 2.1233,
      "step": 133
    },
    {
      "epoch": 0.5370741482965932,
      "grad_norm": 2.941136360168457,
      "learning_rate": 2.693294185106562e-05,
      "loss": 2.251,
      "step": 134
    },
    {
      "epoch": 0.5410821643286573,
      "grad_norm": 2.9835431575775146,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 2.2351,
      "step": 135
    },
    {
      "epoch": 0.5450901803607214,
      "grad_norm": 3.041640281677246,
      "learning_rate": 2.547882480847461e-05,
      "loss": 2.2436,
      "step": 136
    },
    {
      "epoch": 0.5490981963927856,
      "grad_norm": 3.0254502296447754,
      "learning_rate": 2.476172311325783e-05,
      "loss": 2.3884,
      "step": 137
    },
    {
      "epoch": 0.5531062124248497,
      "grad_norm": 3.1095502376556396,
      "learning_rate": 2.405152131093926e-05,
      "loss": 2.3294,
      "step": 138
    },
    {
      "epoch": 0.5571142284569138,
      "grad_norm": 3.104705810546875,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 2.2125,
      "step": 139
    },
    {
      "epoch": 0.561122244488978,
      "grad_norm": 3.1647305488586426,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 2.4404,
      "step": 140
    },
    {
      "epoch": 0.5651302605210421,
      "grad_norm": 3.0764007568359375,
      "learning_rate": 2.196424713241637e-05,
      "loss": 2.4113,
      "step": 141
    },
    {
      "epoch": 0.5691382765531062,
      "grad_norm": 3.1684370040893555,
      "learning_rate": 2.128356686585282e-05,
      "loss": 2.3454,
      "step": 142
    },
    {
      "epoch": 0.5731462925851704,
      "grad_norm": 3.1344246864318848,
      "learning_rate": 2.061073738537635e-05,
      "loss": 2.3477,
      "step": 143
    },
    {
      "epoch": 0.5771543086172345,
      "grad_norm": 3.2735211849212646,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 2.4967,
      "step": 144
    },
    {
      "epoch": 0.5811623246492986,
      "grad_norm": 3.2086429595947266,
      "learning_rate": 1.928936436551661e-05,
      "loss": 2.341,
      "step": 145
    },
    {
      "epoch": 0.5851703406813628,
      "grad_norm": 3.2624154090881348,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 2.187,
      "step": 146
    },
    {
      "epoch": 0.5891783567134269,
      "grad_norm": 3.386133909225464,
      "learning_rate": 1.800157297483417e-05,
      "loss": 2.4005,
      "step": 147
    },
    {
      "epoch": 0.593186372745491,
      "grad_norm": 3.231093168258667,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 2.1535,
      "step": 148
    },
    {
      "epoch": 0.5971943887775552,
      "grad_norm": 3.2708659172058105,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 2.1296,
      "step": 149
    },
    {
      "epoch": 0.6012024048096193,
      "grad_norm": 3.678691864013672,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 2.2936,
      "step": 150
    },
    {
      "epoch": 0.6012024048096193,
      "eval_loss": 0.5725266337394714,
      "eval_runtime": 89.8121,
      "eval_samples_per_second": 4.688,
      "eval_steps_per_second": 1.18,
      "step": 150
    },
    {
      "epoch": 0.6052104208416834,
      "grad_norm": 8.108134269714355,
      "learning_rate": 1.553232954407171e-05,
      "loss": 3.6303,
      "step": 151
    },
    {
      "epoch": 0.6092184368737475,
      "grad_norm": 3.3676862716674805,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 2.2755,
      "step": 152
    },
    {
      "epoch": 0.6132264529058116,
      "grad_norm": 3.3358988761901855,
      "learning_rate": 1.435357758543015e-05,
      "loss": 2.3166,
      "step": 153
    },
    {
      "epoch": 0.6172344689378757,
      "grad_norm": 3.06803822517395,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 2.153,
      "step": 154
    },
    {
      "epoch": 0.6212424849699398,
      "grad_norm": 2.962650775909424,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 2.2814,
      "step": 155
    },
    {
      "epoch": 0.625250501002004,
      "grad_norm": 2.8298065662384033,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 2.3628,
      "step": 156
    },
    {
      "epoch": 0.6292585170340681,
      "grad_norm": 2.636293888092041,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 2.1554,
      "step": 157
    },
    {
      "epoch": 0.6332665330661322,
      "grad_norm": 2.612872362136841,
      "learning_rate": 1.157994445715706e-05,
      "loss": 2.2008,
      "step": 158
    },
    {
      "epoch": 0.6372745490981964,
      "grad_norm": 2.6383156776428223,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 2.3371,
      "step": 159
    },
    {
      "epoch": 0.6412825651302605,
      "grad_norm": 2.748889207839966,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 2.1944,
      "step": 160
    },
    {
      "epoch": 0.6452905811623246,
      "grad_norm": 2.6395554542541504,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 2.053,
      "step": 161
    },
    {
      "epoch": 0.6492985971943888,
      "grad_norm": 2.716303825378418,
      "learning_rate": 9.549150281252633e-06,
      "loss": 2.1962,
      "step": 162
    },
    {
      "epoch": 0.6533066132264529,
      "grad_norm": 2.7696309089660645,
      "learning_rate": 9.068759265665384e-06,
      "loss": 2.0875,
      "step": 163
    },
    {
      "epoch": 0.657314629258517,
      "grad_norm": 2.7269880771636963,
      "learning_rate": 8.599558442598998e-06,
      "loss": 2.0274,
      "step": 164
    },
    {
      "epoch": 0.6613226452905812,
      "grad_norm": 2.7884132862091064,
      "learning_rate": 8.141676086873572e-06,
      "loss": 1.998,
      "step": 165
    },
    {
      "epoch": 0.6653306613226453,
      "grad_norm": 2.990201473236084,
      "learning_rate": 7.695237378953223e-06,
      "loss": 2.1909,
      "step": 166
    },
    {
      "epoch": 0.6693386773547094,
      "grad_norm": 2.7943427562713623,
      "learning_rate": 7.260364370723044e-06,
      "loss": 2.1746,
      "step": 167
    },
    {
      "epoch": 0.6733466933867736,
      "grad_norm": 2.7491326332092285,
      "learning_rate": 6.837175952121306e-06,
      "loss": 1.9397,
      "step": 168
    },
    {
      "epoch": 0.6773547094188377,
      "grad_norm": 2.8264925479888916,
      "learning_rate": 6.425787818636131e-06,
      "loss": 2.0807,
      "step": 169
    },
    {
      "epoch": 0.6813627254509018,
      "grad_norm": 2.836945056915283,
      "learning_rate": 6.026312439675552e-06,
      "loss": 1.9912,
      "step": 170
    },
    {
      "epoch": 0.685370741482966,
      "grad_norm": 3.041722297668457,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 2.1761,
      "step": 171
    },
    {
      "epoch": 0.6893787575150301,
      "grad_norm": 2.921405792236328,
      "learning_rate": 5.263533508961827e-06,
      "loss": 2.0174,
      "step": 172
    },
    {
      "epoch": 0.6933867735470942,
      "grad_norm": 3.0136797428131104,
      "learning_rate": 4.900438493352055e-06,
      "loss": 2.1859,
      "step": 173
    },
    {
      "epoch": 0.6973947895791583,
      "grad_norm": 2.9226717948913574,
      "learning_rate": 4.549673247541875e-06,
      "loss": 2.1546,
      "step": 174
    },
    {
      "epoch": 0.7014028056112225,
      "grad_norm": 2.92608904838562,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 2.0519,
      "step": 175
    },
    {
      "epoch": 0.7054108216432866,
      "grad_norm": 2.8183093070983887,
      "learning_rate": 3.885512251130763e-06,
      "loss": 2.1977,
      "step": 176
    },
    {
      "epoch": 0.7094188376753507,
      "grad_norm": 2.960336446762085,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 2.1262,
      "step": 177
    },
    {
      "epoch": 0.7134268537074149,
      "grad_norm": 2.8956258296966553,
      "learning_rate": 3.271776770026963e-06,
      "loss": 2.1811,
      "step": 178
    },
    {
      "epoch": 0.717434869739479,
      "grad_norm": 2.8759102821350098,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 2.1693,
      "step": 179
    },
    {
      "epoch": 0.7214428857715431,
      "grad_norm": 2.9955990314483643,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 2.2649,
      "step": 180
    },
    {
      "epoch": 0.7254509018036072,
      "grad_norm": 2.7741847038269043,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 2.1267,
      "step": 181
    },
    {
      "epoch": 0.7294589178356713,
      "grad_norm": 2.809300661087036,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 2.1727,
      "step": 182
    },
    {
      "epoch": 0.7334669338677354,
      "grad_norm": 2.9072818756103516,
      "learning_rate": 1.962316193157593e-06,
      "loss": 2.2581,
      "step": 183
    },
    {
      "epoch": 0.7374749498997996,
      "grad_norm": 2.9248111248016357,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 2.4392,
      "step": 184
    },
    {
      "epoch": 0.7414829659318637,
      "grad_norm": 2.9022226333618164,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 2.2164,
      "step": 185
    },
    {
      "epoch": 0.7454909819639278,
      "grad_norm": 2.9731595516204834,
      "learning_rate": 1.333670137599713e-06,
      "loss": 2.168,
      "step": 186
    },
    {
      "epoch": 0.749498997995992,
      "grad_norm": 3.0244717597961426,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 2.2642,
      "step": 187
    },
    {
      "epoch": 0.7535070140280561,
      "grad_norm": 3.100628614425659,
      "learning_rate": 9.810017062595322e-07,
      "loss": 2.362,
      "step": 188
    },
    {
      "epoch": 0.7575150300601202,
      "grad_norm": 2.922551155090332,
      "learning_rate": 8.247462563808817e-07,
      "loss": 2.2203,
      "step": 189
    },
    {
      "epoch": 0.7615230460921844,
      "grad_norm": 3.1363964080810547,
      "learning_rate": 6.819348298638839e-07,
      "loss": 2.5271,
      "step": 190
    },
    {
      "epoch": 0.7655310621242485,
      "grad_norm": 3.108311414718628,
      "learning_rate": 5.526064699265753e-07,
      "loss": 2.3528,
      "step": 191
    },
    {
      "epoch": 0.7695390781563126,
      "grad_norm": 3.085465908050537,
      "learning_rate": 4.367965336512403e-07,
      "loss": 2.3839,
      "step": 192
    },
    {
      "epoch": 0.7735470941883767,
      "grad_norm": 3.0662777423858643,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 2.2818,
      "step": 193
    },
    {
      "epoch": 0.7775551102204409,
      "grad_norm": 3.104541540145874,
      "learning_rate": 2.458548727494292e-07,
      "loss": 2.1621,
      "step": 194
    },
    {
      "epoch": 0.781563126252505,
      "grad_norm": 3.319154977798462,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 2.4567,
      "step": 195
    },
    {
      "epoch": 0.7855711422845691,
      "grad_norm": 3.316058874130249,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 2.2772,
      "step": 196
    },
    {
      "epoch": 0.7895791583166333,
      "grad_norm": 3.213953971862793,
      "learning_rate": 6.150154258476315e-08,
      "loss": 2.3096,
      "step": 197
    },
    {
      "epoch": 0.7935871743486974,
      "grad_norm": 3.3358776569366455,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 2.4532,
      "step": 198
    },
    {
      "epoch": 0.7975951903807615,
      "grad_norm": 3.6416778564453125,
      "learning_rate": 6.834750376549792e-09,
      "loss": 2.4639,
      "step": 199
    },
    {
      "epoch": 0.8016032064128257,
      "grad_norm": 4.0033955574035645,
      "learning_rate": 0.0,
      "loss": 2.4788,
      "step": 200
    },
    {
      "epoch": 0.8016032064128257,
      "eval_loss": 0.5548620223999023,
      "eval_runtime": 89.9725,
      "eval_samples_per_second": 4.679,
      "eval_steps_per_second": 1.178,
      "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": 5.094352911996027e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}