{
  "best_metric": 0.06332537531852722,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.5847953216374269,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0029239766081871343,
      "grad_norm": 2.0890398025512695,
      "learning_rate": 1e-05,
      "loss": 1.1411,
      "step": 1
    },
    {
      "epoch": 0.0029239766081871343,
      "eval_loss": 2.6981592178344727,
      "eval_runtime": 40.3893,
      "eval_samples_per_second": 14.261,
      "eval_steps_per_second": 3.565,
      "step": 1
    },
    {
      "epoch": 0.005847953216374269,
      "grad_norm": 2.310176372528076,
      "learning_rate": 2e-05,
      "loss": 1.4765,
      "step": 2
    },
    {
      "epoch": 0.008771929824561403,
      "grad_norm": 2.356250524520874,
      "learning_rate": 3e-05,
      "loss": 1.4442,
      "step": 3
    },
    {
      "epoch": 0.011695906432748537,
      "grad_norm": 3.2401864528656006,
      "learning_rate": 4e-05,
      "loss": 1.6593,
      "step": 4
    },
    {
      "epoch": 0.014619883040935672,
      "grad_norm": 2.55918025970459,
      "learning_rate": 5e-05,
      "loss": 1.3704,
      "step": 5
    },
    {
      "epoch": 0.017543859649122806,
      "grad_norm": 2.414872646331787,
      "learning_rate": 6e-05,
      "loss": 1.1148,
      "step": 6
    },
    {
      "epoch": 0.02046783625730994,
      "grad_norm": 3.3387632369995117,
      "learning_rate": 7e-05,
      "loss": 1.3199,
      "step": 7
    },
    {
      "epoch": 0.023391812865497075,
      "grad_norm": 2.7576286792755127,
      "learning_rate": 8e-05,
      "loss": 0.7685,
      "step": 8
    },
    {
      "epoch": 0.02631578947368421,
      "grad_norm": 2.428004741668701,
      "learning_rate": 9e-05,
      "loss": 0.475,
      "step": 9
    },
    {
      "epoch": 0.029239766081871343,
      "grad_norm": 2.4822936058044434,
      "learning_rate": 0.0001,
      "loss": 0.6621,
      "step": 10
    },
    {
      "epoch": 0.03216374269005848,
      "grad_norm": 2.04146409034729,
      "learning_rate": 9.999316524962345e-05,
      "loss": 0.4636,
      "step": 11
    },
    {
      "epoch": 0.03508771929824561,
      "grad_norm": 2.0360610485076904,
      "learning_rate": 9.997266286704631e-05,
      "loss": 0.3963,
      "step": 12
    },
    {
      "epoch": 0.038011695906432746,
      "grad_norm": 2.6192574501037598,
      "learning_rate": 9.993849845741524e-05,
      "loss": 0.5692,
      "step": 13
    },
    {
      "epoch": 0.04093567251461988,
      "grad_norm": 2.8487069606781006,
      "learning_rate": 9.989068136093873e-05,
      "loss": 0.601,
      "step": 14
    },
    {
      "epoch": 0.043859649122807015,
      "grad_norm": 2.339660882949829,
      "learning_rate": 9.98292246503335e-05,
      "loss": 0.339,
      "step": 15
    },
    {
      "epoch": 0.04678362573099415,
      "grad_norm": 2.0536000728607178,
      "learning_rate": 9.975414512725057e-05,
      "loss": 0.3482,
      "step": 16
    },
    {
      "epoch": 0.049707602339181284,
      "grad_norm": 1.7835501432418823,
      "learning_rate": 9.966546331768191e-05,
      "loss": 0.3778,
      "step": 17
    },
    {
      "epoch": 0.05263157894736842,
      "grad_norm": 2.382581949234009,
      "learning_rate": 9.956320346634876e-05,
      "loss": 0.3403,
      "step": 18
    },
    {
      "epoch": 0.05555555555555555,
      "grad_norm": 2.2955870628356934,
      "learning_rate": 9.944739353007344e-05,
      "loss": 0.4791,
      "step": 19
    },
    {
      "epoch": 0.05847953216374269,
      "grad_norm": 2.3345448970794678,
      "learning_rate": 9.931806517013612e-05,
      "loss": 0.4857,
      "step": 20
    },
    {
      "epoch": 0.06140350877192982,
      "grad_norm": 1.8483906984329224,
      "learning_rate": 9.917525374361912e-05,
      "loss": 0.3097,
      "step": 21
    },
    {
      "epoch": 0.06432748538011696,
      "grad_norm": 1.7428181171417236,
      "learning_rate": 9.901899829374047e-05,
      "loss": 0.2434,
      "step": 22
    },
    {
      "epoch": 0.06725146198830409,
      "grad_norm": 1.6632155179977417,
      "learning_rate": 9.884934153917997e-05,
      "loss": 0.3581,
      "step": 23
    },
    {
      "epoch": 0.07017543859649122,
      "grad_norm": 1.330684781074524,
      "learning_rate": 9.86663298624003e-05,
      "loss": 0.1676,
      "step": 24
    },
    {
      "epoch": 0.07309941520467836,
      "grad_norm": 1.7981983423233032,
      "learning_rate": 9.847001329696653e-05,
      "loss": 0.1969,
      "step": 25
    },
    {
      "epoch": 0.07602339181286549,
      "grad_norm": 2.4390034675598145,
      "learning_rate": 9.826044551386744e-05,
      "loss": 0.3363,
      "step": 26
    },
    {
      "epoch": 0.07894736842105263,
      "grad_norm": 1.953892469406128,
      "learning_rate": 9.803768380684242e-05,
      "loss": 0.2804,
      "step": 27
    },
    {
      "epoch": 0.08187134502923976,
      "grad_norm": 1.7866439819335938,
      "learning_rate": 9.780178907671789e-05,
      "loss": 0.3074,
      "step": 28
    },
    {
      "epoch": 0.0847953216374269,
      "grad_norm": 1.7630976438522339,
      "learning_rate": 9.755282581475769e-05,
      "loss": 0.2657,
      "step": 29
    },
    {
      "epoch": 0.08771929824561403,
      "grad_norm": 1.9363030195236206,
      "learning_rate": 9.729086208503174e-05,
      "loss": 0.245,
      "step": 30
    },
    {
      "epoch": 0.09064327485380116,
      "grad_norm": 2.9459056854248047,
      "learning_rate": 9.701596950580806e-05,
      "loss": 0.3752,
      "step": 31
    },
    {
      "epoch": 0.0935672514619883,
      "grad_norm": 2.1500279903411865,
      "learning_rate": 9.672822322997305e-05,
      "loss": 0.2837,
      "step": 32
    },
    {
      "epoch": 0.09649122807017543,
      "grad_norm": 4.954206943511963,
      "learning_rate": 9.642770192448536e-05,
      "loss": 0.1758,
      "step": 33
    },
    {
      "epoch": 0.09941520467836257,
      "grad_norm": 2.3973960876464844,
      "learning_rate": 9.611448774886924e-05,
      "loss": 0.3839,
      "step": 34
    },
    {
      "epoch": 0.1023391812865497,
      "grad_norm": 1.6900683641433716,
      "learning_rate": 9.578866633275288e-05,
      "loss": 0.3541,
      "step": 35
    },
    {
      "epoch": 0.10526315789473684,
      "grad_norm": 2.9898910522460938,
      "learning_rate": 9.545032675245813e-05,
      "loss": 0.2009,
      "step": 36
    },
    {
      "epoch": 0.10818713450292397,
      "grad_norm": 2.394251823425293,
      "learning_rate": 9.509956150664796e-05,
      "loss": 0.2474,
      "step": 37
    },
    {
      "epoch": 0.1111111111111111,
      "grad_norm": 2.4706475734710693,
      "learning_rate": 9.473646649103818e-05,
      "loss": 0.3232,
      "step": 38
    },
    {
      "epoch": 0.11403508771929824,
      "grad_norm": 1.4959150552749634,
      "learning_rate": 9.43611409721806e-05,
      "loss": 0.2161,
      "step": 39
    },
    {
      "epoch": 0.11695906432748537,
      "grad_norm": 1.0627198219299316,
      "learning_rate": 9.397368756032445e-05,
      "loss": 0.2196,
      "step": 40
    },
    {
      "epoch": 0.11988304093567251,
      "grad_norm": 1.5753737688064575,
      "learning_rate": 9.357421218136386e-05,
      "loss": 0.3423,
      "step": 41
    },
    {
      "epoch": 0.12280701754385964,
      "grad_norm": 2.046708822250366,
      "learning_rate": 9.316282404787871e-05,
      "loss": 0.1102,
      "step": 42
    },
    {
      "epoch": 0.12573099415204678,
      "grad_norm": 3.688920497894287,
      "learning_rate": 9.273963562927695e-05,
      "loss": 0.2954,
      "step": 43
    },
    {
      "epoch": 0.1286549707602339,
      "grad_norm": 3.269282102584839,
      "learning_rate": 9.230476262104677e-05,
      "loss": 0.2627,
      "step": 44
    },
    {
      "epoch": 0.13157894736842105,
      "grad_norm": 1.795954942703247,
      "learning_rate": 9.185832391312644e-05,
      "loss": 0.3369,
      "step": 45
    },
    {
      "epoch": 0.13450292397660818,
      "grad_norm": 2.5676660537719727,
      "learning_rate": 9.140044155740101e-05,
      "loss": 0.1511,
      "step": 46
    },
    {
      "epoch": 0.13742690058479531,
      "grad_norm": 2.3628110885620117,
      "learning_rate": 9.093124073433463e-05,
      "loss": 0.1901,
      "step": 47
    },
    {
      "epoch": 0.14035087719298245,
      "grad_norm": 5.900804042816162,
      "learning_rate": 9.045084971874738e-05,
      "loss": 0.3269,
      "step": 48
    },
    {
      "epoch": 0.14327485380116958,
      "grad_norm": 2.3997368812561035,
      "learning_rate": 8.995939984474624e-05,
      "loss": 0.1753,
      "step": 49
    },
    {
      "epoch": 0.14619883040935672,
      "grad_norm": 4.376775741577148,
      "learning_rate": 8.945702546981969e-05,
      "loss": 0.7191,
      "step": 50
    },
    {
      "epoch": 0.14619883040935672,
      "eval_loss": 0.23777985572814941,
      "eval_runtime": 40.709,
      "eval_samples_per_second": 14.149,
      "eval_steps_per_second": 3.537,
      "step": 50
    },
    {
      "epoch": 0.14912280701754385,
      "grad_norm": 1.410981297492981,
      "learning_rate": 8.894386393810563e-05,
      "loss": 0.37,
      "step": 51
    },
    {
      "epoch": 0.15204678362573099,
      "grad_norm": 1.1290719509124756,
      "learning_rate": 8.842005554284296e-05,
      "loss": 0.3229,
      "step": 52
    },
    {
      "epoch": 0.15497076023391812,
      "grad_norm": 1.164930820465088,
      "learning_rate": 8.788574348801675e-05,
      "loss": 0.1759,
      "step": 53
    },
    {
      "epoch": 0.15789473684210525,
      "grad_norm": 1.2031174898147583,
      "learning_rate": 8.73410738492077e-05,
      "loss": 0.1957,
      "step": 54
    },
    {
      "epoch": 0.1608187134502924,
      "grad_norm": 0.9945038557052612,
      "learning_rate": 8.678619553365659e-05,
      "loss": 0.1679,
      "step": 55
    },
    {
      "epoch": 0.16374269005847952,
      "grad_norm": 1.0364197492599487,
      "learning_rate": 8.622126023955446e-05,
      "loss": 0.1496,
      "step": 56
    },
    {
      "epoch": 0.16666666666666666,
      "grad_norm": 1.4977822303771973,
      "learning_rate": 8.564642241456986e-05,
      "loss": 0.1188,
      "step": 57
    },
    {
      "epoch": 0.1695906432748538,
      "grad_norm": 0.7542648315429688,
      "learning_rate": 8.506183921362443e-05,
      "loss": 0.1105,
      "step": 58
    },
    {
      "epoch": 0.17251461988304093,
      "grad_norm": 0.9274276494979858,
      "learning_rate": 8.44676704559283e-05,
      "loss": 0.1129,
      "step": 59
    },
    {
      "epoch": 0.17543859649122806,
      "grad_norm": 1.2470937967300415,
      "learning_rate": 8.386407858128706e-05,
      "loss": 0.1244,
      "step": 60
    },
    {
      "epoch": 0.1783625730994152,
      "grad_norm": 1.0916916131973267,
      "learning_rate": 8.32512286056924e-05,
      "loss": 0.0863,
      "step": 61
    },
    {
      "epoch": 0.18128654970760233,
      "grad_norm": 1.0528879165649414,
      "learning_rate": 8.262928807620843e-05,
      "loss": 0.149,
      "step": 62
    },
    {
      "epoch": 0.18421052631578946,
      "grad_norm": 1.1782444715499878,
      "learning_rate": 8.199842702516583e-05,
      "loss": 0.2064,
      "step": 63
    },
    {
      "epoch": 0.1871345029239766,
      "grad_norm": 1.1229543685913086,
      "learning_rate": 8.135881792367686e-05,
      "loss": 0.1819,
      "step": 64
    },
    {
      "epoch": 0.19005847953216373,
      "grad_norm": 1.4986158609390259,
      "learning_rate": 8.07106356344834e-05,
      "loss": 0.2466,
      "step": 65
    },
    {
      "epoch": 0.19298245614035087,
      "grad_norm": 1.2365769147872925,
      "learning_rate": 8.005405736415126e-05,
      "loss": 0.1109,
      "step": 66
    },
    {
      "epoch": 0.195906432748538,
      "grad_norm": 0.8835895657539368,
      "learning_rate": 7.938926261462366e-05,
      "loss": 0.1207,
      "step": 67
    },
    {
      "epoch": 0.19883040935672514,
      "grad_norm": 0.5641562938690186,
      "learning_rate": 7.871643313414718e-05,
      "loss": 0.0557,
      "step": 68
    },
    {
      "epoch": 0.20175438596491227,
      "grad_norm": 1.1681004762649536,
      "learning_rate": 7.803575286758364e-05,
      "loss": 0.0889,
      "step": 69
    },
    {
      "epoch": 0.2046783625730994,
      "grad_norm": 1.3729251623153687,
      "learning_rate": 7.734740790612136e-05,
      "loss": 0.1602,
      "step": 70
    },
    {
      "epoch": 0.20760233918128654,
      "grad_norm": 1.354580283164978,
      "learning_rate": 7.66515864363997e-05,
      "loss": 0.2044,
      "step": 71
    },
    {
      "epoch": 0.21052631578947367,
      "grad_norm": 1.806260108947754,
      "learning_rate": 7.594847868906076e-05,
      "loss": 0.2107,
      "step": 72
    },
    {
      "epoch": 0.2134502923976608,
      "grad_norm": 1.0779149532318115,
      "learning_rate": 7.52382768867422e-05,
      "loss": 0.0661,
      "step": 73
    },
    {
      "epoch": 0.21637426900584794,
      "grad_norm": 1.7838656902313232,
      "learning_rate": 7.452117519152542e-05,
      "loss": 0.1928,
      "step": 74
    },
    {
      "epoch": 0.21929824561403508,
      "grad_norm": 1.521530032157898,
      "learning_rate": 7.379736965185368e-05,
      "loss": 0.1036,
      "step": 75
    },
    {
      "epoch": 0.2222222222222222,
      "grad_norm": 1.5029271841049194,
      "learning_rate": 7.30670581489344e-05,
      "loss": 0.1351,
      "step": 76
    },
    {
      "epoch": 0.22514619883040934,
      "grad_norm": 1.5530059337615967,
      "learning_rate": 7.233044034264034e-05,
      "loss": 0.1452,
      "step": 77
    },
    {
      "epoch": 0.22807017543859648,
      "grad_norm": 1.1716041564941406,
      "learning_rate": 7.158771761692464e-05,
      "loss": 0.1679,
      "step": 78
    },
    {
      "epoch": 0.2309941520467836,
      "grad_norm": 1.0164728164672852,
      "learning_rate": 7.083909302476453e-05,
      "loss": 0.1187,
      "step": 79
    },
    {
      "epoch": 0.23391812865497075,
      "grad_norm": 1.504684329032898,
      "learning_rate": 7.008477123264848e-05,
      "loss": 0.1893,
      "step": 80
    },
    {
      "epoch": 0.23684210526315788,
      "grad_norm": 1.8343896865844727,
      "learning_rate": 6.932495846462261e-05,
      "loss": 0.0902,
      "step": 81
    },
    {
      "epoch": 0.23976608187134502,
      "grad_norm": 1.5082416534423828,
      "learning_rate": 6.855986244591104e-05,
      "loss": 0.1362,
      "step": 82
    },
    {
      "epoch": 0.24269005847953215,
      "grad_norm": 1.1749330759048462,
      "learning_rate": 6.778969234612584e-05,
      "loss": 0.1095,
      "step": 83
    },
    {
      "epoch": 0.24561403508771928,
      "grad_norm": 0.6425621509552002,
      "learning_rate": 6.701465872208216e-05,
      "loss": 0.0765,
      "step": 84
    },
    {
      "epoch": 0.24853801169590642,
      "grad_norm": 1.172538161277771,
      "learning_rate": 6.623497346023418e-05,
      "loss": 0.1592,
      "step": 85
    },
    {
      "epoch": 0.25146198830409355,
      "grad_norm": 2.3020501136779785,
      "learning_rate": 6.545084971874738e-05,
      "loss": 0.2583,
      "step": 86
    },
    {
      "epoch": 0.2543859649122807,
      "grad_norm": 2.4702718257904053,
      "learning_rate": 6.466250186922325e-05,
      "loss": 0.4123,
      "step": 87
    },
    {
      "epoch": 0.2573099415204678,
      "grad_norm": 1.3678455352783203,
      "learning_rate": 6.387014543809223e-05,
      "loss": 0.1621,
      "step": 88
    },
    {
      "epoch": 0.260233918128655,
      "grad_norm": 1.3469619750976562,
      "learning_rate": 6.307399704769099e-05,
      "loss": 0.1747,
      "step": 89
    },
    {
      "epoch": 0.2631578947368421,
      "grad_norm": 1.0145078897476196,
      "learning_rate": 6.227427435703997e-05,
      "loss": 0.0945,
      "step": 90
    },
    {
      "epoch": 0.26608187134502925,
      "grad_norm": 1.0948058366775513,
      "learning_rate": 6.147119600233758e-05,
      "loss": 0.1555,
      "step": 91
    },
    {
      "epoch": 0.26900584795321636,
      "grad_norm": 0.9066969752311707,
      "learning_rate": 6.066498153718735e-05,
      "loss": 0.0766,
      "step": 92
    },
    {
      "epoch": 0.2719298245614035,
      "grad_norm": 0.9737265706062317,
      "learning_rate": 5.985585137257401e-05,
      "loss": 0.1354,
      "step": 93
    },
    {
      "epoch": 0.27485380116959063,
      "grad_norm": 1.368283987045288,
      "learning_rate": 5.90440267166055e-05,
      "loss": 0.0831,
      "step": 94
    },
    {
      "epoch": 0.2777777777777778,
      "grad_norm": 1.1596975326538086,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 0.048,
      "step": 95
    },
    {
      "epoch": 0.2807017543859649,
      "grad_norm": 0.8607801795005798,
      "learning_rate": 5.74131823855921e-05,
      "loss": 0.033,
      "step": 96
    },
    {
      "epoch": 0.28362573099415206,
      "grad_norm": 0.877533495426178,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 0.0607,
      "step": 97
    },
    {
      "epoch": 0.28654970760233917,
      "grad_norm": 0.7952608466148376,
      "learning_rate": 5.577423184847932e-05,
      "loss": 0.0396,
      "step": 98
    },
    {
      "epoch": 0.2894736842105263,
      "grad_norm": 2.1482226848602295,
      "learning_rate": 5.495227651252315e-05,
      "loss": 0.1036,
      "step": 99
    },
    {
      "epoch": 0.29239766081871343,
      "grad_norm": 5.306232452392578,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 0.7334,
      "step": 100
    },
    {
      "epoch": 0.29239766081871343,
      "eval_loss": 0.13454046845436096,
      "eval_runtime": 40.7106,
      "eval_samples_per_second": 14.149,
      "eval_steps_per_second": 3.537,
      "step": 100
    },
    {
      "epoch": 0.2953216374269006,
      "grad_norm": 1.4431185722351074,
      "learning_rate": 5.330452921628497e-05,
      "loss": 0.2775,
      "step": 101
    },
    {
      "epoch": 0.2982456140350877,
      "grad_norm": 1.0740269422531128,
      "learning_rate": 5.247918773366112e-05,
      "loss": 0.0953,
      "step": 102
    },
    {
      "epoch": 0.30116959064327486,
      "grad_norm": 1.4454175233840942,
      "learning_rate": 5.165316846586541e-05,
      "loss": 0.1699,
      "step": 103
    },
    {
      "epoch": 0.30409356725146197,
      "grad_norm": 1.4981609582901,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 0.3032,
      "step": 104
    },
    {
      "epoch": 0.30701754385964913,
      "grad_norm": 0.5318459272384644,
      "learning_rate": 5e-05,
      "loss": 0.0412,
      "step": 105
    },
    {
      "epoch": 0.30994152046783624,
      "grad_norm": 1.0181541442871094,
      "learning_rate": 4.917330276168208e-05,
      "loss": 0.168,
      "step": 106
    },
    {
      "epoch": 0.3128654970760234,
      "grad_norm": 0.8908571600914001,
      "learning_rate": 4.834683153413459e-05,
      "loss": 0.1732,
      "step": 107
    },
    {
      "epoch": 0.3157894736842105,
      "grad_norm": 0.8806681632995605,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 0.0649,
      "step": 108
    },
    {
      "epoch": 0.31871345029239767,
      "grad_norm": 1.0315310955047607,
      "learning_rate": 4.669547078371504e-05,
      "loss": 0.0681,
      "step": 109
    },
    {
      "epoch": 0.3216374269005848,
      "grad_norm": 0.517672598361969,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 0.0379,
      "step": 110
    },
    {
      "epoch": 0.32456140350877194,
      "grad_norm": 0.5320312976837158,
      "learning_rate": 4.504772348747687e-05,
      "loss": 0.0536,
      "step": 111
    },
    {
      "epoch": 0.32748538011695905,
      "grad_norm": 0.761741042137146,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 0.071,
      "step": 112
    },
    {
      "epoch": 0.3304093567251462,
      "grad_norm": 0.6942750811576843,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 0.0842,
      "step": 113
    },
    {
      "epoch": 0.3333333333333333,
      "grad_norm": 1.1313939094543457,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 0.0752,
      "step": 114
    },
    {
      "epoch": 0.3362573099415205,
      "grad_norm": 0.5943699479103088,
      "learning_rate": 4.17702704859633e-05,
      "loss": 0.0218,
      "step": 115
    },
    {
      "epoch": 0.3391812865497076,
      "grad_norm": 0.5406585931777954,
      "learning_rate": 4.095597328339452e-05,
      "loss": 0.0455,
      "step": 116
    },
    {
      "epoch": 0.34210526315789475,
      "grad_norm": 1.1144094467163086,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 0.0653,
      "step": 117
    },
    {
      "epoch": 0.34502923976608185,
      "grad_norm": 0.7442683577537537,
      "learning_rate": 3.933501846281267e-05,
      "loss": 0.0631,
      "step": 118
    },
    {
      "epoch": 0.347953216374269,
      "grad_norm": 0.7475947737693787,
      "learning_rate": 3.852880399766243e-05,
      "loss": 0.0929,
      "step": 119
    },
    {
      "epoch": 0.3508771929824561,
      "grad_norm": 0.760625958442688,
      "learning_rate": 3.772572564296005e-05,
      "loss": 0.0581,
      "step": 120
    },
    {
      "epoch": 0.3538011695906433,
      "grad_norm": 1.111908197402954,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 0.0721,
      "step": 121
    },
    {
      "epoch": 0.3567251461988304,
      "grad_norm": 1.3549267053604126,
      "learning_rate": 3.612985456190778e-05,
      "loss": 0.0968,
      "step": 122
    },
    {
      "epoch": 0.35964912280701755,
      "grad_norm": 0.9203678965568542,
      "learning_rate": 3.533749813077677e-05,
      "loss": 0.1335,
      "step": 123
    },
    {
      "epoch": 0.36257309941520466,
      "grad_norm": 0.6739991307258606,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 0.127,
      "step": 124
    },
    {
      "epoch": 0.3654970760233918,
      "grad_norm": 0.75037682056427,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 0.0739,
      "step": 125
    },
    {
      "epoch": 0.3684210526315789,
      "grad_norm": 0.5134744644165039,
      "learning_rate": 3.298534127791785e-05,
      "loss": 0.0562,
      "step": 126
    },
    {
      "epoch": 0.3713450292397661,
      "grad_norm": 1.61324942111969,
      "learning_rate": 3.221030765387417e-05,
      "loss": 0.073,
      "step": 127
    },
    {
      "epoch": 0.3742690058479532,
      "grad_norm": 0.8383670449256897,
      "learning_rate": 3.144013755408895e-05,
      "loss": 0.0771,
      "step": 128
    },
    {
      "epoch": 0.37719298245614036,
      "grad_norm": 0.7300837635993958,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 0.0175,
      "step": 129
    },
    {
      "epoch": 0.38011695906432746,
      "grad_norm": 0.9605924487113953,
      "learning_rate": 2.991522876735154e-05,
      "loss": 0.0405,
      "step": 130
    },
    {
      "epoch": 0.3830409356725146,
      "grad_norm": 1.8118090629577637,
      "learning_rate": 2.916090697523549e-05,
      "loss": 0.2512,
      "step": 131
    },
    {
      "epoch": 0.38596491228070173,
      "grad_norm": 1.2128326892852783,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 0.1373,
      "step": 132
    },
    {
      "epoch": 0.3888888888888889,
      "grad_norm": 0.42887094616889954,
      "learning_rate": 2.766955965735968e-05,
      "loss": 0.0077,
      "step": 133
    },
    {
      "epoch": 0.391812865497076,
      "grad_norm": 0.89931720495224,
      "learning_rate": 2.693294185106562e-05,
      "loss": 0.0512,
      "step": 134
    },
    {
      "epoch": 0.39473684210526316,
      "grad_norm": 0.756951093673706,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 0.058,
      "step": 135
    },
    {
      "epoch": 0.39766081871345027,
      "grad_norm": 0.9209396839141846,
      "learning_rate": 2.547882480847461e-05,
      "loss": 0.0683,
      "step": 136
    },
    {
      "epoch": 0.40058479532163743,
      "grad_norm": 0.9183736443519592,
      "learning_rate": 2.476172311325783e-05,
      "loss": 0.1064,
      "step": 137
    },
    {
      "epoch": 0.40350877192982454,
      "grad_norm": 1.2212498188018799,
      "learning_rate": 2.405152131093926e-05,
      "loss": 0.0571,
      "step": 138
    },
    {
      "epoch": 0.4064327485380117,
      "grad_norm": 0.9035666584968567,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 0.0524,
      "step": 139
    },
    {
      "epoch": 0.4093567251461988,
      "grad_norm": 0.8271259665489197,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 0.066,
      "step": 140
    },
    {
      "epoch": 0.41228070175438597,
      "grad_norm": 1.1582491397857666,
      "learning_rate": 2.196424713241637e-05,
      "loss": 0.0217,
      "step": 141
    },
    {
      "epoch": 0.4152046783625731,
      "grad_norm": 0.21481887996196747,
      "learning_rate": 2.128356686585282e-05,
      "loss": 0.004,
      "step": 142
    },
    {
      "epoch": 0.41812865497076024,
      "grad_norm": 1.1510004997253418,
      "learning_rate": 2.061073738537635e-05,
      "loss": 0.0898,
      "step": 143
    },
    {
      "epoch": 0.42105263157894735,
      "grad_norm": 1.3552937507629395,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 0.0772,
      "step": 144
    },
    {
      "epoch": 0.4239766081871345,
      "grad_norm": 0.5917229652404785,
      "learning_rate": 1.928936436551661e-05,
      "loss": 0.0575,
      "step": 145
    },
    {
      "epoch": 0.4269005847953216,
      "grad_norm": 0.7931484580039978,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 0.0212,
      "step": 146
    },
    {
      "epoch": 0.4298245614035088,
      "grad_norm": 2.385178327560425,
      "learning_rate": 1.800157297483417e-05,
      "loss": 0.2112,
      "step": 147
    },
    {
      "epoch": 0.4327485380116959,
      "grad_norm": 3.091834306716919,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 0.2254,
      "step": 148
    },
    {
      "epoch": 0.43567251461988304,
      "grad_norm": 1.6704230308532715,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 0.1256,
      "step": 149
    },
    {
      "epoch": 0.43859649122807015,
      "grad_norm": 1.8730045557022095,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 0.1964,
      "step": 150
    },
    {
      "epoch": 0.43859649122807015,
      "eval_loss": 0.08593622595071793,
      "eval_runtime": 40.7161,
      "eval_samples_per_second": 14.147,
      "eval_steps_per_second": 3.537,
      "step": 150
    },
    {
      "epoch": 0.4415204678362573,
      "grad_norm": 1.6720114946365356,
      "learning_rate": 1.553232954407171e-05,
      "loss": 0.1055,
      "step": 151
    },
    {
      "epoch": 0.4444444444444444,
      "grad_norm": 0.5894430875778198,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 0.0597,
      "step": 152
    },
    {
      "epoch": 0.4473684210526316,
      "grad_norm": 0.60225909948349,
      "learning_rate": 1.435357758543015e-05,
      "loss": 0.0542,
      "step": 153
    },
    {
      "epoch": 0.4502923976608187,
      "grad_norm": 0.9882864952087402,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 0.0876,
      "step": 154
    },
    {
      "epoch": 0.45321637426900585,
      "grad_norm": 0.8175363540649414,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 0.0915,
      "step": 155
    },
    {
      "epoch": 0.45614035087719296,
      "grad_norm": 0.6284340023994446,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 0.0343,
      "step": 156
    },
    {
      "epoch": 0.4590643274853801,
      "grad_norm": 1.747030258178711,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 0.099,
      "step": 157
    },
    {
      "epoch": 0.4619883040935672,
      "grad_norm": 1.251965880393982,
      "learning_rate": 1.157994445715706e-05,
      "loss": 0.0637,
      "step": 158
    },
    {
      "epoch": 0.4649122807017544,
      "grad_norm": 1.2940280437469482,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 0.0607,
      "step": 159
    },
    {
      "epoch": 0.4678362573099415,
      "grad_norm": 0.6568557024002075,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 0.0664,
      "step": 160
    },
    {
      "epoch": 0.47076023391812866,
      "grad_norm": 0.9602758884429932,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 0.0812,
      "step": 161
    },
    {
      "epoch": 0.47368421052631576,
      "grad_norm": 0.9406545758247375,
      "learning_rate": 9.549150281252633e-06,
      "loss": 0.0148,
      "step": 162
    },
    {
      "epoch": 0.4766081871345029,
      "grad_norm": 0.6799633502960205,
      "learning_rate": 9.068759265665384e-06,
      "loss": 0.0619,
      "step": 163
    },
    {
      "epoch": 0.47953216374269003,
      "grad_norm": 0.8439313769340515,
      "learning_rate": 8.599558442598998e-06,
      "loss": 0.0992,
      "step": 164
    },
    {
      "epoch": 0.4824561403508772,
      "grad_norm": 0.39336708188056946,
      "learning_rate": 8.141676086873572e-06,
      "loss": 0.0087,
      "step": 165
    },
    {
      "epoch": 0.4853801169590643,
      "grad_norm": 0.5493365526199341,
      "learning_rate": 7.695237378953223e-06,
      "loss": 0.0568,
      "step": 166
    },
    {
      "epoch": 0.48830409356725146,
      "grad_norm": 0.6847161650657654,
      "learning_rate": 7.260364370723044e-06,
      "loss": 0.0821,
      "step": 167
    },
    {
      "epoch": 0.49122807017543857,
      "grad_norm": 0.6425371766090393,
      "learning_rate": 6.837175952121306e-06,
      "loss": 0.0357,
      "step": 168
    },
    {
      "epoch": 0.49415204678362573,
      "grad_norm": 1.0826213359832764,
      "learning_rate": 6.425787818636131e-06,
      "loss": 0.0486,
      "step": 169
    },
    {
      "epoch": 0.49707602339181284,
      "grad_norm": 0.775489091873169,
      "learning_rate": 6.026312439675552e-06,
      "loss": 0.0539,
      "step": 170
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4282867908477783,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 0.0943,
      "step": 171
    },
    {
      "epoch": 0.5029239766081871,
      "grad_norm": 1.7871519327163696,
      "learning_rate": 5.263533508961827e-06,
      "loss": 0.0416,
      "step": 172
    },
    {
      "epoch": 0.5058479532163743,
      "grad_norm": 1.3387969732284546,
      "learning_rate": 4.900438493352055e-06,
      "loss": 0.0825,
      "step": 173
    },
    {
      "epoch": 0.5087719298245614,
      "grad_norm": 0.6471429467201233,
      "learning_rate": 4.549673247541875e-06,
      "loss": 0.0205,
      "step": 174
    },
    {
      "epoch": 0.5116959064327485,
      "grad_norm": 0.5445294976234436,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 0.0188,
      "step": 175
    },
    {
      "epoch": 0.5146198830409356,
      "grad_norm": 0.6897689700126648,
      "learning_rate": 3.885512251130763e-06,
      "loss": 0.0399,
      "step": 176
    },
    {
      "epoch": 0.5175438596491229,
      "grad_norm": 0.9314203262329102,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 0.0104,
      "step": 177
    },
    {
      "epoch": 0.52046783625731,
      "grad_norm": 0.6671068668365479,
      "learning_rate": 3.271776770026963e-06,
      "loss": 0.0204,
      "step": 178
    },
    {
      "epoch": 0.5233918128654971,
      "grad_norm": 0.8153965473175049,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 0.0265,
      "step": 179
    },
    {
      "epoch": 0.5263157894736842,
      "grad_norm": 1.295872449874878,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 0.106,
      "step": 180
    },
    {
      "epoch": 0.5292397660818714,
      "grad_norm": 0.21367762982845306,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 0.0068,
      "step": 181
    },
    {
      "epoch": 0.5321637426900585,
      "grad_norm": 0.5155956745147705,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 0.0459,
      "step": 182
    },
    {
      "epoch": 0.5350877192982456,
      "grad_norm": 0.9211664199829102,
      "learning_rate": 1.962316193157593e-06,
      "loss": 0.0495,
      "step": 183
    },
    {
      "epoch": 0.5380116959064327,
      "grad_norm": 0.7172632217407227,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 0.0593,
      "step": 184
    },
    {
      "epoch": 0.5409356725146199,
      "grad_norm": 0.7884315848350525,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 0.0698,
      "step": 185
    },
    {
      "epoch": 0.543859649122807,
      "grad_norm": 0.6527345776557922,
      "learning_rate": 1.333670137599713e-06,
      "loss": 0.0308,
      "step": 186
    },
    {
      "epoch": 0.5467836257309941,
      "grad_norm": 0.5744469165802002,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 0.0198,
      "step": 187
    },
    {
      "epoch": 0.5497076023391813,
      "grad_norm": 1.0774519443511963,
      "learning_rate": 9.810017062595322e-07,
      "loss": 0.164,
      "step": 188
    },
    {
      "epoch": 0.5526315789473685,
      "grad_norm": 1.198975682258606,
      "learning_rate": 8.247462563808817e-07,
      "loss": 0.0843,
      "step": 189
    },
    {
      "epoch": 0.5555555555555556,
      "grad_norm": 0.34650421142578125,
      "learning_rate": 6.819348298638839e-07,
      "loss": 0.0173,
      "step": 190
    },
    {
      "epoch": 0.5584795321637427,
      "grad_norm": 0.7111528515815735,
      "learning_rate": 5.526064699265753e-07,
      "loss": 0.1108,
      "step": 191
    },
    {
      "epoch": 0.5614035087719298,
      "grad_norm": 1.4404373168945312,
      "learning_rate": 4.367965336512403e-07,
      "loss": 0.0561,
      "step": 192
    },
    {
      "epoch": 0.564327485380117,
      "grad_norm": 0.12150423973798752,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 0.0037,
      "step": 193
    },
    {
      "epoch": 0.5672514619883041,
      "grad_norm": 0.6229392886161804,
      "learning_rate": 2.458548727494292e-07,
      "loss": 0.0044,
      "step": 194
    },
    {
      "epoch": 0.5701754385964912,
      "grad_norm": 0.8171374797821045,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 0.068,
      "step": 195
    },
    {
      "epoch": 0.5730994152046783,
      "grad_norm": 0.08036511391401291,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 0.0022,
      "step": 196
    },
    {
      "epoch": 0.5760233918128655,
      "grad_norm": 0.930079460144043,
      "learning_rate": 6.150154258476315e-08,
      "loss": 0.0154,
      "step": 197
    },
    {
      "epoch": 0.5789473684210527,
      "grad_norm": 1.4315065145492554,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 0.0529,
      "step": 198
    },
    {
      "epoch": 0.5818713450292398,
      "grad_norm": 1.7034662961959839,
      "learning_rate": 6.834750376549792e-09,
      "loss": 0.0826,
      "step": 199
    },
    {
      "epoch": 0.5847953216374269,
      "grad_norm": 1.8194787502288818,
      "learning_rate": 0.0,
      "loss": 0.2167,
      "step": 200
    },
    {
      "epoch": 0.5847953216374269,
      "eval_loss": 0.06332537531852722,
      "eval_runtime": 40.6998,
      "eval_samples_per_second": 14.152,
      "eval_steps_per_second": 3.538,
      "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": 2.8544415358255104e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}