{
  "best_metric": 0.4654270112514496,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.7835455435847208,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0039177277179236044,
      "grad_norm": 0.26972609758377075,
      "learning_rate": 1e-05,
      "loss": 0.5485,
      "step": 1
    },
    {
      "epoch": 0.0039177277179236044,
      "eval_loss": 1.0088223218917847,
      "eval_runtime": 30.0452,
      "eval_samples_per_second": 14.312,
      "eval_steps_per_second": 3.595,
      "step": 1
    },
    {
      "epoch": 0.007835455435847209,
      "grad_norm": 0.30441829562187195,
      "learning_rate": 2e-05,
      "loss": 0.5742,
      "step": 2
    },
    {
      "epoch": 0.011753183153770812,
      "grad_norm": 0.34726226329803467,
      "learning_rate": 3e-05,
      "loss": 0.6053,
      "step": 3
    },
    {
      "epoch": 0.015670910871694418,
      "grad_norm": 0.31042972207069397,
      "learning_rate": 4e-05,
      "loss": 0.8001,
      "step": 4
    },
    {
      "epoch": 0.019588638589618023,
      "grad_norm": 0.45918571949005127,
      "learning_rate": 5e-05,
      "loss": 0.6592,
      "step": 5
    },
    {
      "epoch": 0.023506366307541625,
      "grad_norm": 0.3573453724384308,
      "learning_rate": 6e-05,
      "loss": 0.8436,
      "step": 6
    },
    {
      "epoch": 0.02742409402546523,
      "grad_norm": 0.2946838438510895,
      "learning_rate": 7e-05,
      "loss": 0.6614,
      "step": 7
    },
    {
      "epoch": 0.031341821743388835,
      "grad_norm": 0.30500349402427673,
      "learning_rate": 8e-05,
      "loss": 0.7031,
      "step": 8
    },
    {
      "epoch": 0.03525954946131244,
      "grad_norm": 0.29546526074409485,
      "learning_rate": 9e-05,
      "loss": 0.6777,
      "step": 9
    },
    {
      "epoch": 0.039177277179236046,
      "grad_norm": 0.2998694181442261,
      "learning_rate": 0.0001,
      "loss": 0.6509,
      "step": 10
    },
    {
      "epoch": 0.043095004897159644,
      "grad_norm": 0.41219162940979004,
      "learning_rate": 9.999316524962345e-05,
      "loss": 0.7522,
      "step": 11
    },
    {
      "epoch": 0.04701273261508325,
      "grad_norm": 0.39883479475975037,
      "learning_rate": 9.997266286704631e-05,
      "loss": 0.7333,
      "step": 12
    },
    {
      "epoch": 0.050930460333006855,
      "grad_norm": 0.31704333424568176,
      "learning_rate": 9.993849845741524e-05,
      "loss": 0.6461,
      "step": 13
    },
    {
      "epoch": 0.05484818805093046,
      "grad_norm": 0.2685936987400055,
      "learning_rate": 9.989068136093873e-05,
      "loss": 0.6148,
      "step": 14
    },
    {
      "epoch": 0.058765915768854066,
      "grad_norm": 0.2569029927253723,
      "learning_rate": 9.98292246503335e-05,
      "loss": 0.7222,
      "step": 15
    },
    {
      "epoch": 0.06268364348677767,
      "grad_norm": 0.25974375009536743,
      "learning_rate": 9.975414512725057e-05,
      "loss": 0.5885,
      "step": 16
    },
    {
      "epoch": 0.06660137120470128,
      "grad_norm": 0.27935704588890076,
      "learning_rate": 9.966546331768191e-05,
      "loss": 0.6063,
      "step": 17
    },
    {
      "epoch": 0.07051909892262488,
      "grad_norm": 0.25238996744155884,
      "learning_rate": 9.956320346634876e-05,
      "loss": 0.6035,
      "step": 18
    },
    {
      "epoch": 0.07443682664054849,
      "grad_norm": 0.2595730423927307,
      "learning_rate": 9.944739353007344e-05,
      "loss": 0.5944,
      "step": 19
    },
    {
      "epoch": 0.07835455435847209,
      "grad_norm": 0.25267425179481506,
      "learning_rate": 9.931806517013612e-05,
      "loss": 0.6123,
      "step": 20
    },
    {
      "epoch": 0.08227228207639568,
      "grad_norm": 0.24565619230270386,
      "learning_rate": 9.917525374361912e-05,
      "loss": 0.5315,
      "step": 21
    },
    {
      "epoch": 0.08619000979431929,
      "grad_norm": 0.24319912493228912,
      "learning_rate": 9.901899829374047e-05,
      "loss": 0.5833,
      "step": 22
    },
    {
      "epoch": 0.0901077375122429,
      "grad_norm": 0.2320127934217453,
      "learning_rate": 9.884934153917997e-05,
      "loss": 0.6185,
      "step": 23
    },
    {
      "epoch": 0.0940254652301665,
      "grad_norm": 0.2828903794288635,
      "learning_rate": 9.86663298624003e-05,
      "loss": 0.6937,
      "step": 24
    },
    {
      "epoch": 0.0979431929480901,
      "grad_norm": 0.26696455478668213,
      "learning_rate": 9.847001329696653e-05,
      "loss": 0.6248,
      "step": 25
    },
    {
      "epoch": 0.10186092066601371,
      "grad_norm": 0.2754944860935211,
      "learning_rate": 9.826044551386744e-05,
      "loss": 0.6031,
      "step": 26
    },
    {
      "epoch": 0.10577864838393732,
      "grad_norm": 0.25244608521461487,
      "learning_rate": 9.803768380684242e-05,
      "loss": 0.5334,
      "step": 27
    },
    {
      "epoch": 0.10969637610186092,
      "grad_norm": 0.3373315632343292,
      "learning_rate": 9.780178907671789e-05,
      "loss": 0.6109,
      "step": 28
    },
    {
      "epoch": 0.11361410381978453,
      "grad_norm": 0.26254358887672424,
      "learning_rate": 9.755282581475769e-05,
      "loss": 0.5029,
      "step": 29
    },
    {
      "epoch": 0.11753183153770813,
      "grad_norm": 0.2713155746459961,
      "learning_rate": 9.729086208503174e-05,
      "loss": 0.5625,
      "step": 30
    },
    {
      "epoch": 0.12144955925563174,
      "grad_norm": 0.2678334414958954,
      "learning_rate": 9.701596950580806e-05,
      "loss": 0.5526,
      "step": 31
    },
    {
      "epoch": 0.12536728697355534,
      "grad_norm": 0.28181612491607666,
      "learning_rate": 9.672822322997305e-05,
      "loss": 0.5918,
      "step": 32
    },
    {
      "epoch": 0.12928501469147893,
      "grad_norm": 0.3311813473701477,
      "learning_rate": 9.642770192448536e-05,
      "loss": 0.6705,
      "step": 33
    },
    {
      "epoch": 0.13320274240940255,
      "grad_norm": 0.26722171902656555,
      "learning_rate": 9.611448774886924e-05,
      "loss": 0.5595,
      "step": 34
    },
    {
      "epoch": 0.13712047012732614,
      "grad_norm": 0.30288228392601013,
      "learning_rate": 9.578866633275288e-05,
      "loss": 0.6573,
      "step": 35
    },
    {
      "epoch": 0.14103819784524976,
      "grad_norm": 0.3098622262477875,
      "learning_rate": 9.545032675245813e-05,
      "loss": 0.5706,
      "step": 36
    },
    {
      "epoch": 0.14495592556317335,
      "grad_norm": 0.31600478291511536,
      "learning_rate": 9.509956150664796e-05,
      "loss": 0.4897,
      "step": 37
    },
    {
      "epoch": 0.14887365328109697,
      "grad_norm": 0.3423897922039032,
      "learning_rate": 9.473646649103818e-05,
      "loss": 0.642,
      "step": 38
    },
    {
      "epoch": 0.15279138099902057,
      "grad_norm": 0.333883136510849,
      "learning_rate": 9.43611409721806e-05,
      "loss": 0.5899,
      "step": 39
    },
    {
      "epoch": 0.15670910871694418,
      "grad_norm": 0.3087853193283081,
      "learning_rate": 9.397368756032445e-05,
      "loss": 0.6158,
      "step": 40
    },
    {
      "epoch": 0.16062683643486778,
      "grad_norm": 0.3516993820667267,
      "learning_rate": 9.357421218136386e-05,
      "loss": 0.5624,
      "step": 41
    },
    {
      "epoch": 0.16454456415279137,
      "grad_norm": 0.3407880365848541,
      "learning_rate": 9.316282404787871e-05,
      "loss": 0.5712,
      "step": 42
    },
    {
      "epoch": 0.168462291870715,
      "grad_norm": 0.3490033745765686,
      "learning_rate": 9.273963562927695e-05,
      "loss": 0.5637,
      "step": 43
    },
    {
      "epoch": 0.17238001958863858,
      "grad_norm": 0.30700936913490295,
      "learning_rate": 9.230476262104677e-05,
      "loss": 0.5535,
      "step": 44
    },
    {
      "epoch": 0.1762977473065622,
      "grad_norm": 0.3634184002876282,
      "learning_rate": 9.185832391312644e-05,
      "loss": 0.5552,
      "step": 45
    },
    {
      "epoch": 0.1802154750244858,
      "grad_norm": 0.3617565333843231,
      "learning_rate": 9.140044155740101e-05,
      "loss": 0.6254,
      "step": 46
    },
    {
      "epoch": 0.1841332027424094,
      "grad_norm": 0.4008702039718628,
      "learning_rate": 9.093124073433463e-05,
      "loss": 0.5005,
      "step": 47
    },
    {
      "epoch": 0.188050930460333,
      "grad_norm": 0.5387151837348938,
      "learning_rate": 9.045084971874738e-05,
      "loss": 0.561,
      "step": 48
    },
    {
      "epoch": 0.19196865817825662,
      "grad_norm": 0.35293614864349365,
      "learning_rate": 8.995939984474624e-05,
      "loss": 0.4946,
      "step": 49
    },
    {
      "epoch": 0.1958863858961802,
      "grad_norm": 0.39605939388275146,
      "learning_rate": 8.945702546981969e-05,
      "loss": 0.4718,
      "step": 50
    },
    {
      "epoch": 0.1958863858961802,
      "eval_loss": 0.5429767966270447,
      "eval_runtime": 30.5204,
      "eval_samples_per_second": 14.089,
      "eval_steps_per_second": 3.539,
      "step": 50
    },
    {
      "epoch": 0.19980411361410383,
      "grad_norm": 0.2771281898021698,
      "learning_rate": 8.894386393810563e-05,
      "loss": 0.3507,
      "step": 51
    },
    {
      "epoch": 0.20372184133202742,
      "grad_norm": 0.2725943922996521,
      "learning_rate": 8.842005554284296e-05,
      "loss": 0.4221,
      "step": 52
    },
    {
      "epoch": 0.20763956904995104,
      "grad_norm": 0.2671845853328705,
      "learning_rate": 8.788574348801675e-05,
      "loss": 0.3753,
      "step": 53
    },
    {
      "epoch": 0.21155729676787463,
      "grad_norm": 0.24789267778396606,
      "learning_rate": 8.73410738492077e-05,
      "loss": 0.4324,
      "step": 54
    },
    {
      "epoch": 0.21547502448579825,
      "grad_norm": 0.20853081345558167,
      "learning_rate": 8.678619553365659e-05,
      "loss": 0.3917,
      "step": 55
    },
    {
      "epoch": 0.21939275220372184,
      "grad_norm": 0.20116956532001495,
      "learning_rate": 8.622126023955446e-05,
      "loss": 0.4723,
      "step": 56
    },
    {
      "epoch": 0.22331047992164543,
      "grad_norm": 0.1907324492931366,
      "learning_rate": 8.564642241456986e-05,
      "loss": 0.459,
      "step": 57
    },
    {
      "epoch": 0.22722820763956905,
      "grad_norm": 0.1751759648323059,
      "learning_rate": 8.506183921362443e-05,
      "loss": 0.4553,
      "step": 58
    },
    {
      "epoch": 0.23114593535749264,
      "grad_norm": 0.17422771453857422,
      "learning_rate": 8.44676704559283e-05,
      "loss": 0.4141,
      "step": 59
    },
    {
      "epoch": 0.23506366307541626,
      "grad_norm": 0.18861931562423706,
      "learning_rate": 8.386407858128706e-05,
      "loss": 0.5011,
      "step": 60
    },
    {
      "epoch": 0.23898139079333985,
      "grad_norm": 0.2144758403301239,
      "learning_rate": 8.32512286056924e-05,
      "loss": 0.5125,
      "step": 61
    },
    {
      "epoch": 0.24289911851126347,
      "grad_norm": 0.1961142122745514,
      "learning_rate": 8.262928807620843e-05,
      "loss": 0.5073,
      "step": 62
    },
    {
      "epoch": 0.24681684622918706,
      "grad_norm": 0.18665722012519836,
      "learning_rate": 8.199842702516583e-05,
      "loss": 0.501,
      "step": 63
    },
    {
      "epoch": 0.2507345739471107,
      "grad_norm": 0.20960982143878937,
      "learning_rate": 8.135881792367686e-05,
      "loss": 0.5419,
      "step": 64
    },
    {
      "epoch": 0.2546523016650343,
      "grad_norm": 0.20306116342544556,
      "learning_rate": 8.07106356344834e-05,
      "loss": 0.5355,
      "step": 65
    },
    {
      "epoch": 0.25857002938295787,
      "grad_norm": 0.17925068736076355,
      "learning_rate": 8.005405736415126e-05,
      "loss": 0.4962,
      "step": 66
    },
    {
      "epoch": 0.2624877571008815,
      "grad_norm": 0.18048785626888275,
      "learning_rate": 7.938926261462366e-05,
      "loss": 0.4517,
      "step": 67
    },
    {
      "epoch": 0.2664054848188051,
      "grad_norm": 0.19975988566875458,
      "learning_rate": 7.871643313414718e-05,
      "loss": 0.5502,
      "step": 68
    },
    {
      "epoch": 0.2703232125367287,
      "grad_norm": 0.2168513536453247,
      "learning_rate": 7.803575286758364e-05,
      "loss": 0.5925,
      "step": 69
    },
    {
      "epoch": 0.2742409402546523,
      "grad_norm": 0.2047007828950882,
      "learning_rate": 7.734740790612136e-05,
      "loss": 0.4979,
      "step": 70
    },
    {
      "epoch": 0.2781586679725759,
      "grad_norm": 0.23418109118938446,
      "learning_rate": 7.66515864363997e-05,
      "loss": 0.6265,
      "step": 71
    },
    {
      "epoch": 0.2820763956904995,
      "grad_norm": 0.23406685888767242,
      "learning_rate": 7.594847868906076e-05,
      "loss": 0.5017,
      "step": 72
    },
    {
      "epoch": 0.2859941234084231,
      "grad_norm": 0.2153591364622116,
      "learning_rate": 7.52382768867422e-05,
      "loss": 0.5293,
      "step": 73
    },
    {
      "epoch": 0.2899118511263467,
      "grad_norm": 0.29084163904190063,
      "learning_rate": 7.452117519152542e-05,
      "loss": 0.4775,
      "step": 74
    },
    {
      "epoch": 0.2938295788442703,
      "grad_norm": 0.2232832908630371,
      "learning_rate": 7.379736965185368e-05,
      "loss": 0.5442,
      "step": 75
    },
    {
      "epoch": 0.29774730656219395,
      "grad_norm": 0.20817068219184875,
      "learning_rate": 7.30670581489344e-05,
      "loss": 0.5478,
      "step": 76
    },
    {
      "epoch": 0.30166503428011754,
      "grad_norm": 0.21714900434017181,
      "learning_rate": 7.233044034264034e-05,
      "loss": 0.5117,
      "step": 77
    },
    {
      "epoch": 0.30558276199804113,
      "grad_norm": 0.22392518818378448,
      "learning_rate": 7.158771761692464e-05,
      "loss": 0.5159,
      "step": 78
    },
    {
      "epoch": 0.3095004897159647,
      "grad_norm": 0.227036714553833,
      "learning_rate": 7.083909302476453e-05,
      "loss": 0.552,
      "step": 79
    },
    {
      "epoch": 0.31341821743388837,
      "grad_norm": 0.2379981279373169,
      "learning_rate": 7.008477123264848e-05,
      "loss": 0.502,
      "step": 80
    },
    {
      "epoch": 0.31733594515181196,
      "grad_norm": 0.23239147663116455,
      "learning_rate": 6.932495846462261e-05,
      "loss": 0.5056,
      "step": 81
    },
    {
      "epoch": 0.32125367286973555,
      "grad_norm": 0.47145965695381165,
      "learning_rate": 6.855986244591104e-05,
      "loss": 0.5404,
      "step": 82
    },
    {
      "epoch": 0.32517140058765914,
      "grad_norm": 0.26271262764930725,
      "learning_rate": 6.778969234612584e-05,
      "loss": 0.4829,
      "step": 83
    },
    {
      "epoch": 0.32908912830558273,
      "grad_norm": 0.2568938136100769,
      "learning_rate": 6.701465872208216e-05,
      "loss": 0.5703,
      "step": 84
    },
    {
      "epoch": 0.3330068560235064,
      "grad_norm": 0.2933399975299835,
      "learning_rate": 6.623497346023418e-05,
      "loss": 0.5882,
      "step": 85
    },
    {
      "epoch": 0.33692458374143,
      "grad_norm": 0.27287349104881287,
      "learning_rate": 6.545084971874738e-05,
      "loss": 0.5725,
      "step": 86
    },
    {
      "epoch": 0.34084231145935356,
      "grad_norm": 0.27714309096336365,
      "learning_rate": 6.466250186922325e-05,
      "loss": 0.5328,
      "step": 87
    },
    {
      "epoch": 0.34476003917727716,
      "grad_norm": 0.27044838666915894,
      "learning_rate": 6.387014543809223e-05,
      "loss": 0.506,
      "step": 88
    },
    {
      "epoch": 0.3486777668952008,
      "grad_norm": 0.3282981514930725,
      "learning_rate": 6.307399704769099e-05,
      "loss": 0.5982,
      "step": 89
    },
    {
      "epoch": 0.3525954946131244,
      "grad_norm": 0.31343716382980347,
      "learning_rate": 6.227427435703997e-05,
      "loss": 0.5665,
      "step": 90
    },
    {
      "epoch": 0.356513222331048,
      "grad_norm": 0.3406805992126465,
      "learning_rate": 6.147119600233758e-05,
      "loss": 0.5557,
      "step": 91
    },
    {
      "epoch": 0.3604309500489716,
      "grad_norm": 0.3100006878376007,
      "learning_rate": 6.066498153718735e-05,
      "loss": 0.4713,
      "step": 92
    },
    {
      "epoch": 0.3643486777668952,
      "grad_norm": 0.3171490430831909,
      "learning_rate": 5.985585137257401e-05,
      "loss": 0.5357,
      "step": 93
    },
    {
      "epoch": 0.3682664054848188,
      "grad_norm": 0.3275144696235657,
      "learning_rate": 5.90440267166055e-05,
      "loss": 0.5273,
      "step": 94
    },
    {
      "epoch": 0.3721841332027424,
      "grad_norm": 0.3621930778026581,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 0.5442,
      "step": 95
    },
    {
      "epoch": 0.376101860920666,
      "grad_norm": 0.33248838782310486,
      "learning_rate": 5.74131823855921e-05,
      "loss": 0.4757,
      "step": 96
    },
    {
      "epoch": 0.38001958863858964,
      "grad_norm": 0.34849095344543457,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 0.5506,
      "step": 97
    },
    {
      "epoch": 0.38393731635651324,
      "grad_norm": 0.34789741039276123,
      "learning_rate": 5.577423184847932e-05,
      "loss": 0.4696,
      "step": 98
    },
    {
      "epoch": 0.3878550440744368,
      "grad_norm": 0.3817988634109497,
      "learning_rate": 5.495227651252315e-05,
      "loss": 0.4175,
      "step": 99
    },
    {
      "epoch": 0.3917727717923604,
      "grad_norm": 0.5861048698425293,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 0.4571,
      "step": 100
    },
    {
      "epoch": 0.3917727717923604,
      "eval_loss": 0.4999079406261444,
      "eval_runtime": 30.5579,
      "eval_samples_per_second": 14.072,
      "eval_steps_per_second": 3.534,
      "step": 100
    },
    {
      "epoch": 0.395690499510284,
      "grad_norm": 0.17914611101150513,
      "learning_rate": 5.330452921628497e-05,
      "loss": 0.3187,
      "step": 101
    },
    {
      "epoch": 0.39960822722820766,
      "grad_norm": 0.2052203118801117,
      "learning_rate": 5.247918773366112e-05,
      "loss": 0.3622,
      "step": 102
    },
    {
      "epoch": 0.40352595494613125,
      "grad_norm": 0.226773202419281,
      "learning_rate": 5.165316846586541e-05,
      "loss": 0.376,
      "step": 103
    },
    {
      "epoch": 0.40744368266405484,
      "grad_norm": 0.23201745748519897,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 0.4014,
      "step": 104
    },
    {
      "epoch": 0.41136141038197843,
      "grad_norm": 0.22793354094028473,
      "learning_rate": 5e-05,
      "loss": 0.4412,
      "step": 105
    },
    {
      "epoch": 0.4152791380999021,
      "grad_norm": 0.1926497370004654,
      "learning_rate": 4.917330276168208e-05,
      "loss": 0.3756,
      "step": 106
    },
    {
      "epoch": 0.41919686581782567,
      "grad_norm": 0.2317059189081192,
      "learning_rate": 4.834683153413459e-05,
      "loss": 0.4315,
      "step": 107
    },
    {
      "epoch": 0.42311459353574926,
      "grad_norm": 0.20062124729156494,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 0.4152,
      "step": 108
    },
    {
      "epoch": 0.42703232125367285,
      "grad_norm": 0.20652806758880615,
      "learning_rate": 4.669547078371504e-05,
      "loss": 0.4956,
      "step": 109
    },
    {
      "epoch": 0.4309500489715965,
      "grad_norm": 0.2049548476934433,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 0.448,
      "step": 110
    },
    {
      "epoch": 0.4348677766895201,
      "grad_norm": 0.19221512973308563,
      "learning_rate": 4.504772348747687e-05,
      "loss": 0.4091,
      "step": 111
    },
    {
      "epoch": 0.4387855044074437,
      "grad_norm": 0.20272928476333618,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 0.5389,
      "step": 112
    },
    {
      "epoch": 0.4427032321253673,
      "grad_norm": 0.20781376957893372,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 0.4713,
      "step": 113
    },
    {
      "epoch": 0.44662095984329087,
      "grad_norm": 0.20585735142230988,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 0.4777,
      "step": 114
    },
    {
      "epoch": 0.4505386875612145,
      "grad_norm": 0.2058897614479065,
      "learning_rate": 4.17702704859633e-05,
      "loss": 0.4744,
      "step": 115
    },
    {
      "epoch": 0.4544564152791381,
      "grad_norm": 0.21956440806388855,
      "learning_rate": 4.095597328339452e-05,
      "loss": 0.4843,
      "step": 116
    },
    {
      "epoch": 0.4583741429970617,
      "grad_norm": 0.217080757021904,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 0.4818,
      "step": 117
    },
    {
      "epoch": 0.4622918707149853,
      "grad_norm": 0.21617946028709412,
      "learning_rate": 3.933501846281267e-05,
      "loss": 0.5324,
      "step": 118
    },
    {
      "epoch": 0.46620959843290893,
      "grad_norm": 0.2339426577091217,
      "learning_rate": 3.852880399766243e-05,
      "loss": 0.5273,
      "step": 119
    },
    {
      "epoch": 0.4701273261508325,
      "grad_norm": 0.21858668327331543,
      "learning_rate": 3.772572564296005e-05,
      "loss": 0.5648,
      "step": 120
    },
    {
      "epoch": 0.4740450538687561,
      "grad_norm": 0.1887381374835968,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 0.3954,
      "step": 121
    },
    {
      "epoch": 0.4779627815866797,
      "grad_norm": 0.23094674944877625,
      "learning_rate": 3.612985456190778e-05,
      "loss": 0.488,
      "step": 122
    },
    {
      "epoch": 0.48188050930460335,
      "grad_norm": 0.25744035840034485,
      "learning_rate": 3.533749813077677e-05,
      "loss": 0.6622,
      "step": 123
    },
    {
      "epoch": 0.48579823702252695,
      "grad_norm": 0.22198626399040222,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 0.4203,
      "step": 124
    },
    {
      "epoch": 0.48971596474045054,
      "grad_norm": 0.25090301036834717,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 0.5551,
      "step": 125
    },
    {
      "epoch": 0.49363369245837413,
      "grad_norm": 0.2793654203414917,
      "learning_rate": 3.298534127791785e-05,
      "loss": 0.5571,
      "step": 126
    },
    {
      "epoch": 0.4975514201762977,
      "grad_norm": 0.2615298330783844,
      "learning_rate": 3.221030765387417e-05,
      "loss": 0.5951,
      "step": 127
    },
    {
      "epoch": 0.5014691478942214,
      "grad_norm": 0.3096764087677002,
      "learning_rate": 3.144013755408895e-05,
      "loss": 0.5507,
      "step": 128
    },
    {
      "epoch": 0.5053868756121449,
      "grad_norm": 0.26822254061698914,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 0.5296,
      "step": 129
    },
    {
      "epoch": 0.5093046033300686,
      "grad_norm": 0.26826319098472595,
      "learning_rate": 2.991522876735154e-05,
      "loss": 0.5297,
      "step": 130
    },
    {
      "epoch": 0.5132223310479922,
      "grad_norm": 0.2675338685512543,
      "learning_rate": 2.916090697523549e-05,
      "loss": 0.5831,
      "step": 131
    },
    {
      "epoch": 0.5171400587659157,
      "grad_norm": 0.2670856714248657,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 0.54,
      "step": 132
    },
    {
      "epoch": 0.5210577864838394,
      "grad_norm": 0.2631172835826874,
      "learning_rate": 2.766955965735968e-05,
      "loss": 0.5452,
      "step": 133
    },
    {
      "epoch": 0.524975514201763,
      "grad_norm": 0.2749301493167877,
      "learning_rate": 2.693294185106562e-05,
      "loss": 0.5229,
      "step": 134
    },
    {
      "epoch": 0.5288932419196866,
      "grad_norm": 0.27897027134895325,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 0.5513,
      "step": 135
    },
    {
      "epoch": 0.5328109696376102,
      "grad_norm": 0.27677199244499207,
      "learning_rate": 2.547882480847461e-05,
      "loss": 0.5337,
      "step": 136
    },
    {
      "epoch": 0.5367286973555337,
      "grad_norm": 0.5336064696311951,
      "learning_rate": 2.476172311325783e-05,
      "loss": 0.4814,
      "step": 137
    },
    {
      "epoch": 0.5406464250734574,
      "grad_norm": 0.29900017380714417,
      "learning_rate": 2.405152131093926e-05,
      "loss": 0.5749,
      "step": 138
    },
    {
      "epoch": 0.544564152791381,
      "grad_norm": 0.27716895937919617,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 0.4772,
      "step": 139
    },
    {
      "epoch": 0.5484818805093046,
      "grad_norm": 0.30977749824523926,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 0.5454,
      "step": 140
    },
    {
      "epoch": 0.5523996082272282,
      "grad_norm": 0.28450608253479004,
      "learning_rate": 2.196424713241637e-05,
      "loss": 0.5248,
      "step": 141
    },
    {
      "epoch": 0.5563173359451518,
      "grad_norm": 0.2857397496700287,
      "learning_rate": 2.128356686585282e-05,
      "loss": 0.4744,
      "step": 142
    },
    {
      "epoch": 0.5602350636630754,
      "grad_norm": 0.2955208122730255,
      "learning_rate": 2.061073738537635e-05,
      "loss": 0.5052,
      "step": 143
    },
    {
      "epoch": 0.564152791380999,
      "grad_norm": 0.31489041447639465,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 0.5625,
      "step": 144
    },
    {
      "epoch": 0.5680705190989226,
      "grad_norm": 0.29186511039733887,
      "learning_rate": 1.928936436551661e-05,
      "loss": 0.4347,
      "step": 145
    },
    {
      "epoch": 0.5719882468168462,
      "grad_norm": 0.3489967882633209,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 0.4773,
      "step": 146
    },
    {
      "epoch": 0.5759059745347699,
      "grad_norm": 0.29148194193840027,
      "learning_rate": 1.800157297483417e-05,
      "loss": 0.4549,
      "step": 147
    },
    {
      "epoch": 0.5798237022526934,
      "grad_norm": 0.308432936668396,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 0.431,
      "step": 148
    },
    {
      "epoch": 0.5837414299706171,
      "grad_norm": 0.3864743709564209,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 0.4226,
      "step": 149
    },
    {
      "epoch": 0.5876591576885406,
      "grad_norm": 0.4305037260055542,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 0.411,
      "step": 150
    },
    {
      "epoch": 0.5876591576885406,
      "eval_loss": 0.4719695448875427,
      "eval_runtime": 30.5792,
      "eval_samples_per_second": 14.062,
      "eval_steps_per_second": 3.532,
      "step": 150
    },
    {
      "epoch": 0.5915768854064642,
      "grad_norm": 0.14875982701778412,
      "learning_rate": 1.553232954407171e-05,
      "loss": 0.2154,
      "step": 151
    },
    {
      "epoch": 0.5954946131243879,
      "grad_norm": 0.14534419775009155,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 0.267,
      "step": 152
    },
    {
      "epoch": 0.5994123408423114,
      "grad_norm": 0.18419203162193298,
      "learning_rate": 1.435357758543015e-05,
      "loss": 0.3598,
      "step": 153
    },
    {
      "epoch": 0.6033300685602351,
      "grad_norm": 0.1916617453098297,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 0.3982,
      "step": 154
    },
    {
      "epoch": 0.6072477962781586,
      "grad_norm": 0.18548016250133514,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 0.4342,
      "step": 155
    },
    {
      "epoch": 0.6111655239960823,
      "grad_norm": 0.31114447116851807,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 0.4788,
      "step": 156
    },
    {
      "epoch": 0.6150832517140059,
      "grad_norm": 0.18414777517318726,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 0.3486,
      "step": 157
    },
    {
      "epoch": 0.6190009794319294,
      "grad_norm": 0.20125900208950043,
      "learning_rate": 1.157994445715706e-05,
      "loss": 0.4401,
      "step": 158
    },
    {
      "epoch": 0.6229187071498531,
      "grad_norm": 0.1972198188304901,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 0.3918,
      "step": 159
    },
    {
      "epoch": 0.6268364348677767,
      "grad_norm": 0.18580587208271027,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 0.4195,
      "step": 160
    },
    {
      "epoch": 0.6307541625857003,
      "grad_norm": 0.23439495265483856,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 0.542,
      "step": 161
    },
    {
      "epoch": 0.6346718903036239,
      "grad_norm": 0.22220833599567413,
      "learning_rate": 9.549150281252633e-06,
      "loss": 0.5042,
      "step": 162
    },
    {
      "epoch": 0.6385896180215475,
      "grad_norm": 0.4385804831981659,
      "learning_rate": 9.068759265665384e-06,
      "loss": 0.4091,
      "step": 163
    },
    {
      "epoch": 0.6425073457394711,
      "grad_norm": 0.2172393947839737,
      "learning_rate": 8.599558442598998e-06,
      "loss": 0.4499,
      "step": 164
    },
    {
      "epoch": 0.6464250734573947,
      "grad_norm": 0.23506313562393188,
      "learning_rate": 8.141676086873572e-06,
      "loss": 0.5296,
      "step": 165
    },
    {
      "epoch": 0.6503428011753183,
      "grad_norm": 0.2332986444234848,
      "learning_rate": 7.695237378953223e-06,
      "loss": 0.4476,
      "step": 166
    },
    {
      "epoch": 0.6542605288932419,
      "grad_norm": 0.20083877444267273,
      "learning_rate": 7.260364370723044e-06,
      "loss": 0.3491,
      "step": 167
    },
    {
      "epoch": 0.6581782566111655,
      "grad_norm": 0.1972820907831192,
      "learning_rate": 6.837175952121306e-06,
      "loss": 0.3906,
      "step": 168
    },
    {
      "epoch": 0.6620959843290891,
      "grad_norm": 0.2078624665737152,
      "learning_rate": 6.425787818636131e-06,
      "loss": 0.4387,
      "step": 169
    },
    {
      "epoch": 0.6660137120470128,
      "grad_norm": 0.22298645973205566,
      "learning_rate": 6.026312439675552e-06,
      "loss": 0.4743,
      "step": 170
    },
    {
      "epoch": 0.6699314397649363,
      "grad_norm": 0.2238662987947464,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 0.5209,
      "step": 171
    },
    {
      "epoch": 0.67384916748286,
      "grad_norm": 0.22466623783111572,
      "learning_rate": 5.263533508961827e-06,
      "loss": 0.5031,
      "step": 172
    },
    {
      "epoch": 0.6777668952007836,
      "grad_norm": 0.23010987043380737,
      "learning_rate": 4.900438493352055e-06,
      "loss": 0.4522,
      "step": 173
    },
    {
      "epoch": 0.6816846229187071,
      "grad_norm": 0.2411428838968277,
      "learning_rate": 4.549673247541875e-06,
      "loss": 0.599,
      "step": 174
    },
    {
      "epoch": 0.6856023506366308,
      "grad_norm": 0.30372369289398193,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 0.5544,
      "step": 175
    },
    {
      "epoch": 0.6895200783545543,
      "grad_norm": 0.24095061421394348,
      "learning_rate": 3.885512251130763e-06,
      "loss": 0.577,
      "step": 176
    },
    {
      "epoch": 0.693437806072478,
      "grad_norm": 0.22726194560527802,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 0.5072,
      "step": 177
    },
    {
      "epoch": 0.6973555337904016,
      "grad_norm": 0.2847997844219208,
      "learning_rate": 3.271776770026963e-06,
      "loss": 0.5742,
      "step": 178
    },
    {
      "epoch": 0.7012732615083251,
      "grad_norm": 0.2630937695503235,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 0.5719,
      "step": 179
    },
    {
      "epoch": 0.7051909892262488,
      "grad_norm": 0.28436508774757385,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 0.5543,
      "step": 180
    },
    {
      "epoch": 0.7091087169441724,
      "grad_norm": 0.26273876428604126,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 0.4795,
      "step": 181
    },
    {
      "epoch": 0.713026444662096,
      "grad_norm": 0.2484874129295349,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 0.4465,
      "step": 182
    },
    {
      "epoch": 0.7169441723800196,
      "grad_norm": 0.2420506626367569,
      "learning_rate": 1.962316193157593e-06,
      "loss": 0.4831,
      "step": 183
    },
    {
      "epoch": 0.7208619000979432,
      "grad_norm": 0.2480272799730301,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 0.4454,
      "step": 184
    },
    {
      "epoch": 0.7247796278158668,
      "grad_norm": 0.2571113109588623,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 0.4716,
      "step": 185
    },
    {
      "epoch": 0.7286973555337904,
      "grad_norm": 0.30622658133506775,
      "learning_rate": 1.333670137599713e-06,
      "loss": 0.5097,
      "step": 186
    },
    {
      "epoch": 0.732615083251714,
      "grad_norm": 0.2853301465511322,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 0.5446,
      "step": 187
    },
    {
      "epoch": 0.7365328109696376,
      "grad_norm": 0.27071964740753174,
      "learning_rate": 9.810017062595322e-07,
      "loss": 0.5392,
      "step": 188
    },
    {
      "epoch": 0.7404505386875612,
      "grad_norm": 0.2724110782146454,
      "learning_rate": 8.247462563808817e-07,
      "loss": 0.4978,
      "step": 189
    },
    {
      "epoch": 0.7443682664054848,
      "grad_norm": 0.2743048071861267,
      "learning_rate": 6.819348298638839e-07,
      "loss": 0.4298,
      "step": 190
    },
    {
      "epoch": 0.7482859941234085,
      "grad_norm": 0.3096316158771515,
      "learning_rate": 5.526064699265753e-07,
      "loss": 0.5887,
      "step": 191
    },
    {
      "epoch": 0.752203721841332,
      "grad_norm": 0.2914092540740967,
      "learning_rate": 4.367965336512403e-07,
      "loss": 0.4921,
      "step": 192
    },
    {
      "epoch": 0.7561214495592556,
      "grad_norm": 0.31309422850608826,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 0.4696,
      "step": 193
    },
    {
      "epoch": 0.7600391772771793,
      "grad_norm": 0.2971559464931488,
      "learning_rate": 2.458548727494292e-07,
      "loss": 0.5126,
      "step": 194
    },
    {
      "epoch": 0.7639569049951028,
      "grad_norm": 0.3173765540122986,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 0.4157,
      "step": 195
    },
    {
      "epoch": 0.7678746327130265,
      "grad_norm": 0.3299669325351715,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 0.454,
      "step": 196
    },
    {
      "epoch": 0.77179236043095,
      "grad_norm": 0.3788973093032837,
      "learning_rate": 6.150154258476315e-08,
      "loss": 0.4909,
      "step": 197
    },
    {
      "epoch": 0.7757100881488737,
      "grad_norm": 0.36976227164268494,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 0.5051,
      "step": 198
    },
    {
      "epoch": 0.7796278158667973,
      "grad_norm": 0.3565472662448883,
      "learning_rate": 6.834750376549792e-09,
      "loss": 0.4093,
      "step": 199
    },
    {
      "epoch": 0.7835455435847208,
      "grad_norm": 0.38544148206710815,
      "learning_rate": 0.0,
      "loss": 0.3523,
      "step": 200
    },
    {
      "epoch": 0.7835455435847208,
      "eval_loss": 0.4654270112514496,
      "eval_runtime": 30.5891,
      "eval_samples_per_second": 14.057,
      "eval_steps_per_second": 3.531,
      "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": 2.661017318326272e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}