{
  "best_metric": 0.6435211896896362,
  "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": 3.6940090656280518,
      "learning_rate": 1e-05,
      "loss": 1.3846,
      "step": 1
    },
    {
      "epoch": 0.004296455424274973,
      "eval_loss": 3.0925545692443848,
      "eval_runtime": 6.3974,
      "eval_samples_per_second": 61.275,
      "eval_steps_per_second": 15.319,
      "step": 1
    },
    {
      "epoch": 0.008592910848549946,
      "grad_norm": 4.802069664001465,
      "learning_rate": 2e-05,
      "loss": 1.8958,
      "step": 2
    },
    {
      "epoch": 0.01288936627282492,
      "grad_norm": 5.888405799865723,
      "learning_rate": 3e-05,
      "loss": 2.025,
      "step": 3
    },
    {
      "epoch": 0.017185821697099892,
      "grad_norm": 5.273263454437256,
      "learning_rate": 4e-05,
      "loss": 2.1644,
      "step": 4
    },
    {
      "epoch": 0.021482277121374866,
      "grad_norm": 3.2237682342529297,
      "learning_rate": 5e-05,
      "loss": 1.8597,
      "step": 5
    },
    {
      "epoch": 0.02577873254564984,
      "grad_norm": 3.056070327758789,
      "learning_rate": 6e-05,
      "loss": 1.7128,
      "step": 6
    },
    {
      "epoch": 0.03007518796992481,
      "grad_norm": 2.9855151176452637,
      "learning_rate": 7e-05,
      "loss": 1.5778,
      "step": 7
    },
    {
      "epoch": 0.034371643394199784,
      "grad_norm": 3.132558584213257,
      "learning_rate": 8e-05,
      "loss": 1.3197,
      "step": 8
    },
    {
      "epoch": 0.03866809881847476,
      "grad_norm": 2.819892644882202,
      "learning_rate": 9e-05,
      "loss": 1.2656,
      "step": 9
    },
    {
      "epoch": 0.04296455424274973,
      "grad_norm": 2.1864962577819824,
      "learning_rate": 0.0001,
      "loss": 1.2634,
      "step": 10
    },
    {
      "epoch": 0.047261009667024706,
      "grad_norm": 2.9201531410217285,
      "learning_rate": 9.999316524962345e-05,
      "loss": 1.4199,
      "step": 11
    },
    {
      "epoch": 0.05155746509129968,
      "grad_norm": 1.4994686841964722,
      "learning_rate": 9.997266286704631e-05,
      "loss": 0.9732,
      "step": 12
    },
    {
      "epoch": 0.055853920515574654,
      "grad_norm": 1.8198317289352417,
      "learning_rate": 9.993849845741524e-05,
      "loss": 1.1336,
      "step": 13
    },
    {
      "epoch": 0.06015037593984962,
      "grad_norm": 1.6827316284179688,
      "learning_rate": 9.989068136093873e-05,
      "loss": 1.0496,
      "step": 14
    },
    {
      "epoch": 0.0644468313641246,
      "grad_norm": 1.4762057065963745,
      "learning_rate": 9.98292246503335e-05,
      "loss": 1.0842,
      "step": 15
    },
    {
      "epoch": 0.06874328678839957,
      "grad_norm": 1.828031063079834,
      "learning_rate": 9.975414512725057e-05,
      "loss": 1.041,
      "step": 16
    },
    {
      "epoch": 0.07303974221267455,
      "grad_norm": 1.5681266784667969,
      "learning_rate": 9.966546331768191e-05,
      "loss": 0.9144,
      "step": 17
    },
    {
      "epoch": 0.07733619763694952,
      "grad_norm": 1.7685694694519043,
      "learning_rate": 9.956320346634876e-05,
      "loss": 0.9876,
      "step": 18
    },
    {
      "epoch": 0.08163265306122448,
      "grad_norm": 1.4676024913787842,
      "learning_rate": 9.944739353007344e-05,
      "loss": 0.9161,
      "step": 19
    },
    {
      "epoch": 0.08592910848549946,
      "grad_norm": 1.6293609142303467,
      "learning_rate": 9.931806517013612e-05,
      "loss": 0.9143,
      "step": 20
    },
    {
      "epoch": 0.09022556390977443,
      "grad_norm": 2.2325549125671387,
      "learning_rate": 9.917525374361912e-05,
      "loss": 1.1875,
      "step": 21
    },
    {
      "epoch": 0.09452201933404941,
      "grad_norm": 2.143954277038574,
      "learning_rate": 9.901899829374047e-05,
      "loss": 1.1853,
      "step": 22
    },
    {
      "epoch": 0.09881847475832438,
      "grad_norm": 2.0767698287963867,
      "learning_rate": 9.884934153917997e-05,
      "loss": 1.0234,
      "step": 23
    },
    {
      "epoch": 0.10311493018259936,
      "grad_norm": 2.171109437942505,
      "learning_rate": 9.86663298624003e-05,
      "loss": 1.111,
      "step": 24
    },
    {
      "epoch": 0.10741138560687433,
      "grad_norm": 2.1155309677124023,
      "learning_rate": 9.847001329696653e-05,
      "loss": 1.1015,
      "step": 25
    },
    {
      "epoch": 0.11170784103114931,
      "grad_norm": 2.353606939315796,
      "learning_rate": 9.826044551386744e-05,
      "loss": 1.1453,
      "step": 26
    },
    {
      "epoch": 0.11600429645542427,
      "grad_norm": 2.181079864501953,
      "learning_rate": 9.803768380684242e-05,
      "loss": 1.1244,
      "step": 27
    },
    {
      "epoch": 0.12030075187969924,
      "grad_norm": 1.8007853031158447,
      "learning_rate": 9.780178907671789e-05,
      "loss": 0.9651,
      "step": 28
    },
    {
      "epoch": 0.12459720730397422,
      "grad_norm": 1.9248244762420654,
      "learning_rate": 9.755282581475769e-05,
      "loss": 1.0424,
      "step": 29
    },
    {
      "epoch": 0.1288936627282492,
      "grad_norm": 1.8673564195632935,
      "learning_rate": 9.729086208503174e-05,
      "loss": 1.0949,
      "step": 30
    },
    {
      "epoch": 0.13319011815252416,
      "grad_norm": 2.051316499710083,
      "learning_rate": 9.701596950580806e-05,
      "loss": 0.9875,
      "step": 31
    },
    {
      "epoch": 0.13748657357679914,
      "grad_norm": 2.013021945953369,
      "learning_rate": 9.672822322997305e-05,
      "loss": 1.0598,
      "step": 32
    },
    {
      "epoch": 0.14178302900107412,
      "grad_norm": 2.5428507328033447,
      "learning_rate": 9.642770192448536e-05,
      "loss": 0.938,
      "step": 33
    },
    {
      "epoch": 0.1460794844253491,
      "grad_norm": 2.48836350440979,
      "learning_rate": 9.611448774886924e-05,
      "loss": 1.0257,
      "step": 34
    },
    {
      "epoch": 0.15037593984962405,
      "grad_norm": 2.3669607639312744,
      "learning_rate": 9.578866633275288e-05,
      "loss": 1.2544,
      "step": 35
    },
    {
      "epoch": 0.15467239527389903,
      "grad_norm": 2.0529329776763916,
      "learning_rate": 9.545032675245813e-05,
      "loss": 0.9965,
      "step": 36
    },
    {
      "epoch": 0.158968850698174,
      "grad_norm": 2.8352019786834717,
      "learning_rate": 9.509956150664796e-05,
      "loss": 0.9468,
      "step": 37
    },
    {
      "epoch": 0.16326530612244897,
      "grad_norm": 2.1075093746185303,
      "learning_rate": 9.473646649103818e-05,
      "loss": 0.9191,
      "step": 38
    },
    {
      "epoch": 0.16756176154672395,
      "grad_norm": 2.040109395980835,
      "learning_rate": 9.43611409721806e-05,
      "loss": 0.9363,
      "step": 39
    },
    {
      "epoch": 0.17185821697099893,
      "grad_norm": 2.551389694213867,
      "learning_rate": 9.397368756032445e-05,
      "loss": 1.0411,
      "step": 40
    },
    {
      "epoch": 0.1761546723952739,
      "grad_norm": 2.582689046859741,
      "learning_rate": 9.357421218136386e-05,
      "loss": 1.1275,
      "step": 41
    },
    {
      "epoch": 0.18045112781954886,
      "grad_norm": 2.425568103790283,
      "learning_rate": 9.316282404787871e-05,
      "loss": 1.0154,
      "step": 42
    },
    {
      "epoch": 0.18474758324382384,
      "grad_norm": 2.7150723934173584,
      "learning_rate": 9.273963562927695e-05,
      "loss": 0.9331,
      "step": 43
    },
    {
      "epoch": 0.18904403866809882,
      "grad_norm": 2.3676249980926514,
      "learning_rate": 9.230476262104677e-05,
      "loss": 0.7913,
      "step": 44
    },
    {
      "epoch": 0.1933404940923738,
      "grad_norm": 2.416627883911133,
      "learning_rate": 9.185832391312644e-05,
      "loss": 0.948,
      "step": 45
    },
    {
      "epoch": 0.19763694951664876,
      "grad_norm": 2.9219796657562256,
      "learning_rate": 9.140044155740101e-05,
      "loss": 1.1725,
      "step": 46
    },
    {
      "epoch": 0.20193340494092374,
      "grad_norm": 3.3446099758148193,
      "learning_rate": 9.093124073433463e-05,
      "loss": 1.1201,
      "step": 47
    },
    {
      "epoch": 0.20622986036519872,
      "grad_norm": 8.145545959472656,
      "learning_rate": 9.045084971874738e-05,
      "loss": 1.5186,
      "step": 48
    },
    {
      "epoch": 0.21052631578947367,
      "grad_norm": 9.392189025878906,
      "learning_rate": 8.995939984474624e-05,
      "loss": 0.6993,
      "step": 49
    },
    {
      "epoch": 0.21482277121374865,
      "grad_norm": 2.358957052230835,
      "learning_rate": 8.945702546981969e-05,
      "loss": 0.5121,
      "step": 50
    },
    {
      "epoch": 0.21482277121374865,
      "eval_loss": 1.173129916191101,
      "eval_runtime": 6.3582,
      "eval_samples_per_second": 61.653,
      "eval_steps_per_second": 15.413,
      "step": 50
    },
    {
      "epoch": 0.21911922663802363,
      "grad_norm": 4.924007892608643,
      "learning_rate": 8.894386393810563e-05,
      "loss": 1.1849,
      "step": 51
    },
    {
      "epoch": 0.22341568206229862,
      "grad_norm": 4.613038063049316,
      "learning_rate": 8.842005554284296e-05,
      "loss": 1.2046,
      "step": 52
    },
    {
      "epoch": 0.22771213748657357,
      "grad_norm": 3.8540916442871094,
      "learning_rate": 8.788574348801675e-05,
      "loss": 1.1985,
      "step": 53
    },
    {
      "epoch": 0.23200859291084855,
      "grad_norm": 2.9647669792175293,
      "learning_rate": 8.73410738492077e-05,
      "loss": 1.0424,
      "step": 54
    },
    {
      "epoch": 0.23630504833512353,
      "grad_norm": 2.0257620811462402,
      "learning_rate": 8.678619553365659e-05,
      "loss": 0.9737,
      "step": 55
    },
    {
      "epoch": 0.24060150375939848,
      "grad_norm": 1.637176752090454,
      "learning_rate": 8.622126023955446e-05,
      "loss": 0.9146,
      "step": 56
    },
    {
      "epoch": 0.24489795918367346,
      "grad_norm": 1.5860952138900757,
      "learning_rate": 8.564642241456986e-05,
      "loss": 0.8576,
      "step": 57
    },
    {
      "epoch": 0.24919441460794844,
      "grad_norm": 1.642493486404419,
      "learning_rate": 8.506183921362443e-05,
      "loss": 0.8621,
      "step": 58
    },
    {
      "epoch": 0.2534908700322234,
      "grad_norm": 1.9209811687469482,
      "learning_rate": 8.44676704559283e-05,
      "loss": 0.8713,
      "step": 59
    },
    {
      "epoch": 0.2577873254564984,
      "grad_norm": 1.705660104751587,
      "learning_rate": 8.386407858128706e-05,
      "loss": 1.0107,
      "step": 60
    },
    {
      "epoch": 0.2620837808807734,
      "grad_norm": 1.3240758180618286,
      "learning_rate": 8.32512286056924e-05,
      "loss": 0.7833,
      "step": 61
    },
    {
      "epoch": 0.2663802363050483,
      "grad_norm": 1.219561219215393,
      "learning_rate": 8.262928807620843e-05,
      "loss": 0.7791,
      "step": 62
    },
    {
      "epoch": 0.2706766917293233,
      "grad_norm": 1.3021117448806763,
      "learning_rate": 8.199842702516583e-05,
      "loss": 0.7396,
      "step": 63
    },
    {
      "epoch": 0.2749731471535983,
      "grad_norm": 1.2026478052139282,
      "learning_rate": 8.135881792367686e-05,
      "loss": 0.7569,
      "step": 64
    },
    {
      "epoch": 0.27926960257787325,
      "grad_norm": 1.1275808811187744,
      "learning_rate": 8.07106356344834e-05,
      "loss": 0.6994,
      "step": 65
    },
    {
      "epoch": 0.28356605800214824,
      "grad_norm": 1.2625210285186768,
      "learning_rate": 8.005405736415126e-05,
      "loss": 0.7437,
      "step": 66
    },
    {
      "epoch": 0.2878625134264232,
      "grad_norm": 1.2135088443756104,
      "learning_rate": 7.938926261462366e-05,
      "loss": 0.7016,
      "step": 67
    },
    {
      "epoch": 0.2921589688506982,
      "grad_norm": 1.2073503732681274,
      "learning_rate": 7.871643313414718e-05,
      "loss": 0.787,
      "step": 68
    },
    {
      "epoch": 0.2964554242749731,
      "grad_norm": 1.283071517944336,
      "learning_rate": 7.803575286758364e-05,
      "loss": 0.6606,
      "step": 69
    },
    {
      "epoch": 0.3007518796992481,
      "grad_norm": 1.2409520149230957,
      "learning_rate": 7.734740790612136e-05,
      "loss": 0.7781,
      "step": 70
    },
    {
      "epoch": 0.3050483351235231,
      "grad_norm": 1.3091660737991333,
      "learning_rate": 7.66515864363997e-05,
      "loss": 0.8053,
      "step": 71
    },
    {
      "epoch": 0.30934479054779807,
      "grad_norm": 1.3209784030914307,
      "learning_rate": 7.594847868906076e-05,
      "loss": 0.6767,
      "step": 72
    },
    {
      "epoch": 0.31364124597207305,
      "grad_norm": 1.303499460220337,
      "learning_rate": 7.52382768867422e-05,
      "loss": 0.7077,
      "step": 73
    },
    {
      "epoch": 0.317937701396348,
      "grad_norm": 1.4568078517913818,
      "learning_rate": 7.452117519152542e-05,
      "loss": 0.7705,
      "step": 74
    },
    {
      "epoch": 0.322234156820623,
      "grad_norm": 1.5832254886627197,
      "learning_rate": 7.379736965185368e-05,
      "loss": 0.8247,
      "step": 75
    },
    {
      "epoch": 0.32653061224489793,
      "grad_norm": 1.6621246337890625,
      "learning_rate": 7.30670581489344e-05,
      "loss": 0.8107,
      "step": 76
    },
    {
      "epoch": 0.3308270676691729,
      "grad_norm": 1.5627961158752441,
      "learning_rate": 7.233044034264034e-05,
      "loss": 0.7865,
      "step": 77
    },
    {
      "epoch": 0.3351235230934479,
      "grad_norm": 1.7858855724334717,
      "learning_rate": 7.158771761692464e-05,
      "loss": 0.8176,
      "step": 78
    },
    {
      "epoch": 0.3394199785177229,
      "grad_norm": 2.0798985958099365,
      "learning_rate": 7.083909302476453e-05,
      "loss": 1.049,
      "step": 79
    },
    {
      "epoch": 0.34371643394199786,
      "grad_norm": 1.6190752983093262,
      "learning_rate": 7.008477123264848e-05,
      "loss": 0.7368,
      "step": 80
    },
    {
      "epoch": 0.34801288936627284,
      "grad_norm": 1.7315847873687744,
      "learning_rate": 6.932495846462261e-05,
      "loss": 0.8219,
      "step": 81
    },
    {
      "epoch": 0.3523093447905478,
      "grad_norm": 1.643621802330017,
      "learning_rate": 6.855986244591104e-05,
      "loss": 0.8337,
      "step": 82
    },
    {
      "epoch": 0.35660580021482274,
      "grad_norm": 1.5921083688735962,
      "learning_rate": 6.778969234612584e-05,
      "loss": 0.7118,
      "step": 83
    },
    {
      "epoch": 0.3609022556390977,
      "grad_norm": 2.020669937133789,
      "learning_rate": 6.701465872208216e-05,
      "loss": 0.9622,
      "step": 84
    },
    {
      "epoch": 0.3651987110633727,
      "grad_norm": 1.422821283340454,
      "learning_rate": 6.623497346023418e-05,
      "loss": 0.7022,
      "step": 85
    },
    {
      "epoch": 0.3694951664876477,
      "grad_norm": 1.8762741088867188,
      "learning_rate": 6.545084971874738e-05,
      "loss": 0.8081,
      "step": 86
    },
    {
      "epoch": 0.37379162191192267,
      "grad_norm": 1.7387793064117432,
      "learning_rate": 6.466250186922325e-05,
      "loss": 0.8542,
      "step": 87
    },
    {
      "epoch": 0.37808807733619765,
      "grad_norm": 1.8297538757324219,
      "learning_rate": 6.387014543809223e-05,
      "loss": 0.8829,
      "step": 88
    },
    {
      "epoch": 0.38238453276047263,
      "grad_norm": 1.7551103830337524,
      "learning_rate": 6.307399704769099e-05,
      "loss": 0.7339,
      "step": 89
    },
    {
      "epoch": 0.3866809881847476,
      "grad_norm": 1.6879074573516846,
      "learning_rate": 6.227427435703997e-05,
      "loss": 0.6799,
      "step": 90
    },
    {
      "epoch": 0.39097744360902253,
      "grad_norm": 1.8937222957611084,
      "learning_rate": 6.147119600233758e-05,
      "loss": 0.6137,
      "step": 91
    },
    {
      "epoch": 0.3952738990332975,
      "grad_norm": 2.161207914352417,
      "learning_rate": 6.066498153718735e-05,
      "loss": 0.8405,
      "step": 92
    },
    {
      "epoch": 0.3995703544575725,
      "grad_norm": 1.8968161344528198,
      "learning_rate": 5.985585137257401e-05,
      "loss": 0.6836,
      "step": 93
    },
    {
      "epoch": 0.4038668098818475,
      "grad_norm": 1.91599702835083,
      "learning_rate": 5.90440267166055e-05,
      "loss": 0.7045,
      "step": 94
    },
    {
      "epoch": 0.40816326530612246,
      "grad_norm": 2.4522266387939453,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 0.9552,
      "step": 95
    },
    {
      "epoch": 0.41245972073039744,
      "grad_norm": 2.213533639907837,
      "learning_rate": 5.74131823855921e-05,
      "loss": 0.7822,
      "step": 96
    },
    {
      "epoch": 0.4167561761546724,
      "grad_norm": 2.3995201587677,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 0.8418,
      "step": 97
    },
    {
      "epoch": 0.42105263157894735,
      "grad_norm": 2.342036724090576,
      "learning_rate": 5.577423184847932e-05,
      "loss": 0.7122,
      "step": 98
    },
    {
      "epoch": 0.4253490870032223,
      "grad_norm": 3.3263285160064697,
      "learning_rate": 5.495227651252315e-05,
      "loss": 1.1119,
      "step": 99
    },
    {
      "epoch": 0.4296455424274973,
      "grad_norm": 1.8682502508163452,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 0.3896,
      "step": 100
    },
    {
      "epoch": 0.4296455424274973,
      "eval_loss": 0.8736681938171387,
      "eval_runtime": 6.3595,
      "eval_samples_per_second": 61.64,
      "eval_steps_per_second": 15.41,
      "step": 100
    },
    {
      "epoch": 0.4339419978517723,
      "grad_norm": 2.2557930946350098,
      "learning_rate": 5.330452921628497e-05,
      "loss": 0.8765,
      "step": 101
    },
    {
      "epoch": 0.43823845327604727,
      "grad_norm": 2.670783519744873,
      "learning_rate": 5.247918773366112e-05,
      "loss": 1.01,
      "step": 102
    },
    {
      "epoch": 0.44253490870032225,
      "grad_norm": 2.680856943130493,
      "learning_rate": 5.165316846586541e-05,
      "loss": 1.1138,
      "step": 103
    },
    {
      "epoch": 0.44683136412459723,
      "grad_norm": 2.409604072570801,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 0.9175,
      "step": 104
    },
    {
      "epoch": 0.45112781954887216,
      "grad_norm": 1.8556089401245117,
      "learning_rate": 5e-05,
      "loss": 0.8207,
      "step": 105
    },
    {
      "epoch": 0.45542427497314714,
      "grad_norm": 1.7843716144561768,
      "learning_rate": 4.917330276168208e-05,
      "loss": 0.8186,
      "step": 106
    },
    {
      "epoch": 0.4597207303974221,
      "grad_norm": 1.4535441398620605,
      "learning_rate": 4.834683153413459e-05,
      "loss": 0.6793,
      "step": 107
    },
    {
      "epoch": 0.4640171858216971,
      "grad_norm": 1.3760968446731567,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 0.7488,
      "step": 108
    },
    {
      "epoch": 0.4683136412459721,
      "grad_norm": 1.2591838836669922,
      "learning_rate": 4.669547078371504e-05,
      "loss": 0.7243,
      "step": 109
    },
    {
      "epoch": 0.47261009667024706,
      "grad_norm": 1.2640703916549683,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 0.7198,
      "step": 110
    },
    {
      "epoch": 0.47690655209452204,
      "grad_norm": 1.2857885360717773,
      "learning_rate": 4.504772348747687e-05,
      "loss": 0.7939,
      "step": 111
    },
    {
      "epoch": 0.48120300751879697,
      "grad_norm": 1.3929158449172974,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 0.7656,
      "step": 112
    },
    {
      "epoch": 0.48549946294307195,
      "grad_norm": 1.2389739751815796,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 0.6939,
      "step": 113
    },
    {
      "epoch": 0.4897959183673469,
      "grad_norm": 1.2491884231567383,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 0.6206,
      "step": 114
    },
    {
      "epoch": 0.4940923737916219,
      "grad_norm": 1.1597932577133179,
      "learning_rate": 4.17702704859633e-05,
      "loss": 0.5838,
      "step": 115
    },
    {
      "epoch": 0.4983888292158969,
      "grad_norm": 1.3803465366363525,
      "learning_rate": 4.095597328339452e-05,
      "loss": 0.7073,
      "step": 116
    },
    {
      "epoch": 0.5026852846401718,
      "grad_norm": 1.3922760486602783,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 0.6272,
      "step": 117
    },
    {
      "epoch": 0.5069817400644469,
      "grad_norm": 1.117069125175476,
      "learning_rate": 3.933501846281267e-05,
      "loss": 0.5342,
      "step": 118
    },
    {
      "epoch": 0.5112781954887218,
      "grad_norm": 1.5331922769546509,
      "learning_rate": 3.852880399766243e-05,
      "loss": 0.697,
      "step": 119
    },
    {
      "epoch": 0.5155746509129968,
      "grad_norm": 1.2803605794906616,
      "learning_rate": 3.772572564296005e-05,
      "loss": 0.5467,
      "step": 120
    },
    {
      "epoch": 0.5198711063372717,
      "grad_norm": 1.4712119102478027,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 0.595,
      "step": 121
    },
    {
      "epoch": 0.5241675617615468,
      "grad_norm": 1.7701973915100098,
      "learning_rate": 3.612985456190778e-05,
      "loss": 0.7672,
      "step": 122
    },
    {
      "epoch": 0.5284640171858217,
      "grad_norm": 1.5650845766067505,
      "learning_rate": 3.533749813077677e-05,
      "loss": 0.6085,
      "step": 123
    },
    {
      "epoch": 0.5327604726100966,
      "grad_norm": 1.6881626844406128,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 0.7314,
      "step": 124
    },
    {
      "epoch": 0.5370569280343717,
      "grad_norm": 1.5540963411331177,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 0.7373,
      "step": 125
    },
    {
      "epoch": 0.5413533834586466,
      "grad_norm": 1.708109974861145,
      "learning_rate": 3.298534127791785e-05,
      "loss": 0.7938,
      "step": 126
    },
    {
      "epoch": 0.5456498388829216,
      "grad_norm": 1.7486896514892578,
      "learning_rate": 3.221030765387417e-05,
      "loss": 0.7587,
      "step": 127
    },
    {
      "epoch": 0.5499462943071965,
      "grad_norm": 1.7386753559112549,
      "learning_rate": 3.144013755408895e-05,
      "loss": 0.7214,
      "step": 128
    },
    {
      "epoch": 0.5542427497314716,
      "grad_norm": 1.6774464845657349,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 0.7971,
      "step": 129
    },
    {
      "epoch": 0.5585392051557465,
      "grad_norm": 1.6203025579452515,
      "learning_rate": 2.991522876735154e-05,
      "loss": 0.6518,
      "step": 130
    },
    {
      "epoch": 0.5628356605800214,
      "grad_norm": 1.6531866788864136,
      "learning_rate": 2.916090697523549e-05,
      "loss": 0.6277,
      "step": 131
    },
    {
      "epoch": 0.5671321160042965,
      "grad_norm": 1.74118173122406,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 0.7141,
      "step": 132
    },
    {
      "epoch": 0.5714285714285714,
      "grad_norm": 1.673689365386963,
      "learning_rate": 2.766955965735968e-05,
      "loss": 0.5823,
      "step": 133
    },
    {
      "epoch": 0.5757250268528464,
      "grad_norm": 1.7893049716949463,
      "learning_rate": 2.693294185106562e-05,
      "loss": 0.9089,
      "step": 134
    },
    {
      "epoch": 0.5800214822771214,
      "grad_norm": 1.7646392583847046,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 0.7469,
      "step": 135
    },
    {
      "epoch": 0.5843179377013964,
      "grad_norm": 1.4430310726165771,
      "learning_rate": 2.547882480847461e-05,
      "loss": 0.5493,
      "step": 136
    },
    {
      "epoch": 0.5886143931256713,
      "grad_norm": 1.570407509803772,
      "learning_rate": 2.476172311325783e-05,
      "loss": 0.6485,
      "step": 137
    },
    {
      "epoch": 0.5929108485499462,
      "grad_norm": 1.5142794847488403,
      "learning_rate": 2.405152131093926e-05,
      "loss": 0.5923,
      "step": 138
    },
    {
      "epoch": 0.5972073039742213,
      "grad_norm": 1.714464545249939,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 0.6391,
      "step": 139
    },
    {
      "epoch": 0.6015037593984962,
      "grad_norm": 1.7961385250091553,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 0.609,
      "step": 140
    },
    {
      "epoch": 0.6058002148227712,
      "grad_norm": 1.7792097330093384,
      "learning_rate": 2.196424713241637e-05,
      "loss": 0.7087,
      "step": 141
    },
    {
      "epoch": 0.6100966702470462,
      "grad_norm": 1.6444392204284668,
      "learning_rate": 2.128356686585282e-05,
      "loss": 0.6374,
      "step": 142
    },
    {
      "epoch": 0.6143931256713212,
      "grad_norm": 1.7460147142410278,
      "learning_rate": 2.061073738537635e-05,
      "loss": 0.6115,
      "step": 143
    },
    {
      "epoch": 0.6186895810955961,
      "grad_norm": 1.7699378728866577,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 0.5404,
      "step": 144
    },
    {
      "epoch": 0.6229860365198711,
      "grad_norm": 1.999567985534668,
      "learning_rate": 1.928936436551661e-05,
      "loss": 0.694,
      "step": 145
    },
    {
      "epoch": 0.6272824919441461,
      "grad_norm": 2.3033158779144287,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 0.7751,
      "step": 146
    },
    {
      "epoch": 0.631578947368421,
      "grad_norm": 2.5667564868927,
      "learning_rate": 1.800157297483417e-05,
      "loss": 0.8225,
      "step": 147
    },
    {
      "epoch": 0.635875402792696,
      "grad_norm": 2.9949536323547363,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 1.1298,
      "step": 148
    },
    {
      "epoch": 0.640171858216971,
      "grad_norm": 2.703179359436035,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 0.7221,
      "step": 149
    },
    {
      "epoch": 0.644468313641246,
      "grad_norm": 2.3285953998565674,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 0.4887,
      "step": 150
    },
    {
      "epoch": 0.644468313641246,
      "eval_loss": 0.6901558041572571,
      "eval_runtime": 6.358,
      "eval_samples_per_second": 61.655,
      "eval_steps_per_second": 15.414,
      "step": 150
    },
    {
      "epoch": 0.6487647690655209,
      "grad_norm": 1.043041706085205,
      "learning_rate": 1.553232954407171e-05,
      "loss": 0.4883,
      "step": 151
    },
    {
      "epoch": 0.6530612244897959,
      "grad_norm": 1.3720712661743164,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 0.6869,
      "step": 152
    },
    {
      "epoch": 0.6573576799140709,
      "grad_norm": 1.5384008884429932,
      "learning_rate": 1.435357758543015e-05,
      "loss": 0.8041,
      "step": 153
    },
    {
      "epoch": 0.6616541353383458,
      "grad_norm": 1.4643335342407227,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 0.6682,
      "step": 154
    },
    {
      "epoch": 0.6659505907626209,
      "grad_norm": 1.5699212551116943,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 0.748,
      "step": 155
    },
    {
      "epoch": 0.6702470461868958,
      "grad_norm": 1.2308199405670166,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 0.618,
      "step": 156
    },
    {
      "epoch": 0.6745435016111708,
      "grad_norm": 1.519276738166809,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 0.7367,
      "step": 157
    },
    {
      "epoch": 0.6788399570354458,
      "grad_norm": 1.2963597774505615,
      "learning_rate": 1.157994445715706e-05,
      "loss": 0.6322,
      "step": 158
    },
    {
      "epoch": 0.6831364124597207,
      "grad_norm": 1.323813796043396,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 0.6544,
      "step": 159
    },
    {
      "epoch": 0.6874328678839957,
      "grad_norm": 1.4273834228515625,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 0.753,
      "step": 160
    },
    {
      "epoch": 0.6917293233082706,
      "grad_norm": 1.2868958711624146,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 0.6564,
      "step": 161
    },
    {
      "epoch": 0.6960257787325457,
      "grad_norm": 1.3547497987747192,
      "learning_rate": 9.549150281252633e-06,
      "loss": 0.6306,
      "step": 162
    },
    {
      "epoch": 0.7003222341568206,
      "grad_norm": 1.253730297088623,
      "learning_rate": 9.068759265665384e-06,
      "loss": 0.5483,
      "step": 163
    },
    {
      "epoch": 0.7046186895810956,
      "grad_norm": 1.283799409866333,
      "learning_rate": 8.599558442598998e-06,
      "loss": 0.5924,
      "step": 164
    },
    {
      "epoch": 0.7089151450053706,
      "grad_norm": 1.294709324836731,
      "learning_rate": 8.141676086873572e-06,
      "loss": 0.6982,
      "step": 165
    },
    {
      "epoch": 0.7132116004296455,
      "grad_norm": 1.219122290611267,
      "learning_rate": 7.695237378953223e-06,
      "loss": 0.5721,
      "step": 166
    },
    {
      "epoch": 0.7175080558539205,
      "grad_norm": 1.1943608522415161,
      "learning_rate": 7.260364370723044e-06,
      "loss": 0.5898,
      "step": 167
    },
    {
      "epoch": 0.7218045112781954,
      "grad_norm": 1.2280900478363037,
      "learning_rate": 6.837175952121306e-06,
      "loss": 0.5458,
      "step": 168
    },
    {
      "epoch": 0.7261009667024705,
      "grad_norm": 1.106630802154541,
      "learning_rate": 6.425787818636131e-06,
      "loss": 0.5234,
      "step": 169
    },
    {
      "epoch": 0.7303974221267454,
      "grad_norm": 1.2504066228866577,
      "learning_rate": 6.026312439675552e-06,
      "loss": 0.5506,
      "step": 170
    },
    {
      "epoch": 0.7346938775510204,
      "grad_norm": 1.121796727180481,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 0.5744,
      "step": 171
    },
    {
      "epoch": 0.7389903329752954,
      "grad_norm": 1.124097466468811,
      "learning_rate": 5.263533508961827e-06,
      "loss": 0.5643,
      "step": 172
    },
    {
      "epoch": 0.7432867883995704,
      "grad_norm": 1.2789398431777954,
      "learning_rate": 4.900438493352055e-06,
      "loss": 0.5907,
      "step": 173
    },
    {
      "epoch": 0.7475832438238453,
      "grad_norm": 1.4371399879455566,
      "learning_rate": 4.549673247541875e-06,
      "loss": 0.6561,
      "step": 174
    },
    {
      "epoch": 0.7518796992481203,
      "grad_norm": 1.566687822341919,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 0.7023,
      "step": 175
    },
    {
      "epoch": 0.7561761546723953,
      "grad_norm": 1.420259714126587,
      "learning_rate": 3.885512251130763e-06,
      "loss": 0.6446,
      "step": 176
    },
    {
      "epoch": 0.7604726100966702,
      "grad_norm": 1.5227923393249512,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 0.6625,
      "step": 177
    },
    {
      "epoch": 0.7647690655209453,
      "grad_norm": 1.4586358070373535,
      "learning_rate": 3.271776770026963e-06,
      "loss": 0.5485,
      "step": 178
    },
    {
      "epoch": 0.7690655209452202,
      "grad_norm": 1.4928873777389526,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 0.5758,
      "step": 179
    },
    {
      "epoch": 0.7733619763694952,
      "grad_norm": 1.6715686321258545,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 0.7357,
      "step": 180
    },
    {
      "epoch": 0.7776584317937701,
      "grad_norm": 1.6602483987808228,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 0.6134,
      "step": 181
    },
    {
      "epoch": 0.7819548872180451,
      "grad_norm": 1.560014009475708,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 0.5475,
      "step": 182
    },
    {
      "epoch": 0.7862513426423201,
      "grad_norm": 1.6969585418701172,
      "learning_rate": 1.962316193157593e-06,
      "loss": 0.6542,
      "step": 183
    },
    {
      "epoch": 0.790547798066595,
      "grad_norm": 1.8868225812911987,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 0.7808,
      "step": 184
    },
    {
      "epoch": 0.7948442534908701,
      "grad_norm": 1.5981675386428833,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 0.5874,
      "step": 185
    },
    {
      "epoch": 0.799140708915145,
      "grad_norm": 1.754606008529663,
      "learning_rate": 1.333670137599713e-06,
      "loss": 0.7141,
      "step": 186
    },
    {
      "epoch": 0.80343716433942,
      "grad_norm": 1.744737148284912,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 0.6619,
      "step": 187
    },
    {
      "epoch": 0.807733619763695,
      "grad_norm": 1.7298351526260376,
      "learning_rate": 9.810017062595322e-07,
      "loss": 0.6404,
      "step": 188
    },
    {
      "epoch": 0.8120300751879699,
      "grad_norm": 2.0197865962982178,
      "learning_rate": 8.247462563808817e-07,
      "loss": 0.7618,
      "step": 189
    },
    {
      "epoch": 0.8163265306122449,
      "grad_norm": 1.7302197217941284,
      "learning_rate": 6.819348298638839e-07,
      "loss": 0.5533,
      "step": 190
    },
    {
      "epoch": 0.8206229860365198,
      "grad_norm": 1.7119989395141602,
      "learning_rate": 5.526064699265753e-07,
      "loss": 0.5163,
      "step": 191
    },
    {
      "epoch": 0.8249194414607949,
      "grad_norm": 2.194359302520752,
      "learning_rate": 4.367965336512403e-07,
      "loss": 0.5844,
      "step": 192
    },
    {
      "epoch": 0.8292158968850698,
      "grad_norm": 1.8889172077178955,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 0.5934,
      "step": 193
    },
    {
      "epoch": 0.8335123523093448,
      "grad_norm": 2.432569742202759,
      "learning_rate": 2.458548727494292e-07,
      "loss": 0.6395,
      "step": 194
    },
    {
      "epoch": 0.8378088077336198,
      "grad_norm": 2.1962146759033203,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 0.6809,
      "step": 195
    },
    {
      "epoch": 0.8421052631578947,
      "grad_norm": 2.440392017364502,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 0.5871,
      "step": 196
    },
    {
      "epoch": 0.8464017185821697,
      "grad_norm": 2.8213558197021484,
      "learning_rate": 6.150154258476315e-08,
      "loss": 0.9241,
      "step": 197
    },
    {
      "epoch": 0.8506981740064447,
      "grad_norm": 3.18548583984375,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 0.7627,
      "step": 198
    },
    {
      "epoch": 0.8549946294307197,
      "grad_norm": 4.5898919105529785,
      "learning_rate": 6.834750376549792e-09,
      "loss": 0.7768,
      "step": 199
    },
    {
      "epoch": 0.8592910848549946,
      "grad_norm": 2.800640106201172,
      "learning_rate": 0.0,
      "loss": 0.37,
      "step": 200
    },
    {
      "epoch": 0.8592910848549946,
      "eval_loss": 0.6435211896896362,
      "eval_runtime": 6.3563,
      "eval_samples_per_second": 61.671,
      "eval_steps_per_second": 15.418,
      "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": 1.3317552734208e+16,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}