{
  "best_metric": 0.03573817387223244,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.32948929159802304,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0016474464579901153,
      "grad_norm": 7.155981063842773,
      "learning_rate": 1e-05,
      "loss": 0.5664,
      "step": 1
    },
    {
      "epoch": 0.0016474464579901153,
      "eval_loss": 0.25151872634887695,
      "eval_runtime": 82.7381,
      "eval_samples_per_second": 12.364,
      "eval_steps_per_second": 3.094,
      "step": 1
    },
    {
      "epoch": 0.0032948929159802307,
      "grad_norm": 10.420136451721191,
      "learning_rate": 2e-05,
      "loss": 0.758,
      "step": 2
    },
    {
      "epoch": 0.004942339373970346,
      "grad_norm": 7.768281936645508,
      "learning_rate": 3e-05,
      "loss": 0.6635,
      "step": 3
    },
    {
      "epoch": 0.006589785831960461,
      "grad_norm": 4.629105091094971,
      "learning_rate": 4e-05,
      "loss": 0.4235,
      "step": 4
    },
    {
      "epoch": 0.008237232289950576,
      "grad_norm": 3.1388938426971436,
      "learning_rate": 5e-05,
      "loss": 0.3459,
      "step": 5
    },
    {
      "epoch": 0.009884678747940691,
      "grad_norm": 2.086496114730835,
      "learning_rate": 6e-05,
      "loss": 0.2722,
      "step": 6
    },
    {
      "epoch": 0.011532125205930808,
      "grad_norm": 2.1365911960601807,
      "learning_rate": 7e-05,
      "loss": 0.2953,
      "step": 7
    },
    {
      "epoch": 0.013179571663920923,
      "grad_norm": 1.511433482170105,
      "learning_rate": 8e-05,
      "loss": 0.217,
      "step": 8
    },
    {
      "epoch": 0.014827018121911038,
      "grad_norm": 3.4163413047790527,
      "learning_rate": 9e-05,
      "loss": 0.2602,
      "step": 9
    },
    {
      "epoch": 0.016474464579901153,
      "grad_norm": 1.4421342611312866,
      "learning_rate": 0.0001,
      "loss": 0.2017,
      "step": 10
    },
    {
      "epoch": 0.018121911037891267,
      "grad_norm": 2.9272472858428955,
      "learning_rate": 9.999316524962345e-05,
      "loss": 0.2411,
      "step": 11
    },
    {
      "epoch": 0.019769357495881382,
      "grad_norm": 1.5919667482376099,
      "learning_rate": 9.997266286704631e-05,
      "loss": 0.2338,
      "step": 12
    },
    {
      "epoch": 0.0214168039538715,
      "grad_norm": 1.3033331632614136,
      "learning_rate": 9.993849845741524e-05,
      "loss": 0.2178,
      "step": 13
    },
    {
      "epoch": 0.023064250411861616,
      "grad_norm": 1.3951294422149658,
      "learning_rate": 9.989068136093873e-05,
      "loss": 0.1715,
      "step": 14
    },
    {
      "epoch": 0.02471169686985173,
      "grad_norm": 1.15043044090271,
      "learning_rate": 9.98292246503335e-05,
      "loss": 0.232,
      "step": 15
    },
    {
      "epoch": 0.026359143327841845,
      "grad_norm": 1.1190717220306396,
      "learning_rate": 9.975414512725057e-05,
      "loss": 0.1977,
      "step": 16
    },
    {
      "epoch": 0.02800658978583196,
      "grad_norm": 1.0910860300064087,
      "learning_rate": 9.966546331768191e-05,
      "loss": 0.2124,
      "step": 17
    },
    {
      "epoch": 0.029654036243822075,
      "grad_norm": 1.1268714666366577,
      "learning_rate": 9.956320346634876e-05,
      "loss": 0.203,
      "step": 18
    },
    {
      "epoch": 0.03130148270181219,
      "grad_norm": 1.6683168411254883,
      "learning_rate": 9.944739353007344e-05,
      "loss": 0.2573,
      "step": 19
    },
    {
      "epoch": 0.032948929159802305,
      "grad_norm": 1.4160244464874268,
      "learning_rate": 9.931806517013612e-05,
      "loss": 0.1958,
      "step": 20
    },
    {
      "epoch": 0.03459637561779242,
      "grad_norm": 1.0933629274368286,
      "learning_rate": 9.917525374361912e-05,
      "loss": 0.2156,
      "step": 21
    },
    {
      "epoch": 0.036243822075782535,
      "grad_norm": 0.9035831689834595,
      "learning_rate": 9.901899829374047e-05,
      "loss": 0.18,
      "step": 22
    },
    {
      "epoch": 0.03789126853377265,
      "grad_norm": 0.9546683430671692,
      "learning_rate": 9.884934153917997e-05,
      "loss": 0.1628,
      "step": 23
    },
    {
      "epoch": 0.039538714991762765,
      "grad_norm": 0.8303844928741455,
      "learning_rate": 9.86663298624003e-05,
      "loss": 0.1539,
      "step": 24
    },
    {
      "epoch": 0.04118616144975288,
      "grad_norm": 0.9744294285774231,
      "learning_rate": 9.847001329696653e-05,
      "loss": 0.1714,
      "step": 25
    },
    {
      "epoch": 0.042833607907743,
      "grad_norm": 0.7936011552810669,
      "learning_rate": 9.826044551386744e-05,
      "loss": 0.1639,
      "step": 26
    },
    {
      "epoch": 0.044481054365733116,
      "grad_norm": 0.7597372531890869,
      "learning_rate": 9.803768380684242e-05,
      "loss": 0.1817,
      "step": 27
    },
    {
      "epoch": 0.04612850082372323,
      "grad_norm": 0.856248676776886,
      "learning_rate": 9.780178907671789e-05,
      "loss": 0.165,
      "step": 28
    },
    {
      "epoch": 0.047775947281713346,
      "grad_norm": 1.2012642621994019,
      "learning_rate": 9.755282581475769e-05,
      "loss": 0.1965,
      "step": 29
    },
    {
      "epoch": 0.04942339373970346,
      "grad_norm": 1.0729221105575562,
      "learning_rate": 9.729086208503174e-05,
      "loss": 0.2448,
      "step": 30
    },
    {
      "epoch": 0.051070840197693576,
      "grad_norm": 1.1273274421691895,
      "learning_rate": 9.701596950580806e-05,
      "loss": 0.1845,
      "step": 31
    },
    {
      "epoch": 0.05271828665568369,
      "grad_norm": 1.2683639526367188,
      "learning_rate": 9.672822322997305e-05,
      "loss": 0.208,
      "step": 32
    },
    {
      "epoch": 0.054365733113673806,
      "grad_norm": 0.8238076567649841,
      "learning_rate": 9.642770192448536e-05,
      "loss": 0.2034,
      "step": 33
    },
    {
      "epoch": 0.05601317957166392,
      "grad_norm": 0.7510798573493958,
      "learning_rate": 9.611448774886924e-05,
      "loss": 0.1608,
      "step": 34
    },
    {
      "epoch": 0.057660626029654036,
      "grad_norm": 0.8858838677406311,
      "learning_rate": 9.578866633275288e-05,
      "loss": 0.198,
      "step": 35
    },
    {
      "epoch": 0.05930807248764415,
      "grad_norm": 0.8198444247245789,
      "learning_rate": 9.545032675245813e-05,
      "loss": 0.1685,
      "step": 36
    },
    {
      "epoch": 0.060955518945634266,
      "grad_norm": 0.8573437929153442,
      "learning_rate": 9.509956150664796e-05,
      "loss": 0.2081,
      "step": 37
    },
    {
      "epoch": 0.06260296540362438,
      "grad_norm": 1.0294137001037598,
      "learning_rate": 9.473646649103818e-05,
      "loss": 0.1767,
      "step": 38
    },
    {
      "epoch": 0.0642504118616145,
      "grad_norm": 0.8891019225120544,
      "learning_rate": 9.43611409721806e-05,
      "loss": 0.1469,
      "step": 39
    },
    {
      "epoch": 0.06589785831960461,
      "grad_norm": 0.9600830674171448,
      "learning_rate": 9.397368756032445e-05,
      "loss": 0.1532,
      "step": 40
    },
    {
      "epoch": 0.06754530477759473,
      "grad_norm": 0.9224026203155518,
      "learning_rate": 9.357421218136386e-05,
      "loss": 0.142,
      "step": 41
    },
    {
      "epoch": 0.06919275123558484,
      "grad_norm": 0.9329553842544556,
      "learning_rate": 9.316282404787871e-05,
      "loss": 0.1637,
      "step": 42
    },
    {
      "epoch": 0.07084019769357495,
      "grad_norm": 1.4444706439971924,
      "learning_rate": 9.273963562927695e-05,
      "loss": 0.2154,
      "step": 43
    },
    {
      "epoch": 0.07248764415156507,
      "grad_norm": 1.2399431467056274,
      "learning_rate": 9.230476262104677e-05,
      "loss": 0.2448,
      "step": 44
    },
    {
      "epoch": 0.07413509060955518,
      "grad_norm": 1.3063863515853882,
      "learning_rate": 9.185832391312644e-05,
      "loss": 0.1442,
      "step": 45
    },
    {
      "epoch": 0.0757825370675453,
      "grad_norm": 0.7861459255218506,
      "learning_rate": 9.140044155740101e-05,
      "loss": 0.1271,
      "step": 46
    },
    {
      "epoch": 0.07742998352553541,
      "grad_norm": 1.4548641443252563,
      "learning_rate": 9.093124073433463e-05,
      "loss": 0.2289,
      "step": 47
    },
    {
      "epoch": 0.07907742998352553,
      "grad_norm": 1.101926326751709,
      "learning_rate": 9.045084971874738e-05,
      "loss": 0.1027,
      "step": 48
    },
    {
      "epoch": 0.08072487644151564,
      "grad_norm": 0.968700110912323,
      "learning_rate": 8.995939984474624e-05,
      "loss": 0.1505,
      "step": 49
    },
    {
      "epoch": 0.08237232289950576,
      "grad_norm": 1.304606556892395,
      "learning_rate": 8.945702546981969e-05,
      "loss": 0.1527,
      "step": 50
    },
    {
      "epoch": 0.08237232289950576,
      "eval_loss": 0.05358460545539856,
      "eval_runtime": 84.2038,
      "eval_samples_per_second": 12.149,
      "eval_steps_per_second": 3.04,
      "step": 50
    },
    {
      "epoch": 0.08401976935749589,
      "grad_norm": 0.8823680877685547,
      "learning_rate": 8.894386393810563e-05,
      "loss": 0.2333,
      "step": 51
    },
    {
      "epoch": 0.085667215815486,
      "grad_norm": 0.8000332117080688,
      "learning_rate": 8.842005554284296e-05,
      "loss": 0.2449,
      "step": 52
    },
    {
      "epoch": 0.08731466227347612,
      "grad_norm": 0.9639336466789246,
      "learning_rate": 8.788574348801675e-05,
      "loss": 0.2464,
      "step": 53
    },
    {
      "epoch": 0.08896210873146623,
      "grad_norm": 0.8865578174591064,
      "learning_rate": 8.73410738492077e-05,
      "loss": 0.2153,
      "step": 54
    },
    {
      "epoch": 0.09060955518945635,
      "grad_norm": 0.7511582374572754,
      "learning_rate": 8.678619553365659e-05,
      "loss": 0.2069,
      "step": 55
    },
    {
      "epoch": 0.09225700164744646,
      "grad_norm": 0.7353391647338867,
      "learning_rate": 8.622126023955446e-05,
      "loss": 0.1938,
      "step": 56
    },
    {
      "epoch": 0.09390444810543658,
      "grad_norm": 0.6765355467796326,
      "learning_rate": 8.564642241456986e-05,
      "loss": 0.2333,
      "step": 57
    },
    {
      "epoch": 0.09555189456342669,
      "grad_norm": 0.6801488399505615,
      "learning_rate": 8.506183921362443e-05,
      "loss": 0.1974,
      "step": 58
    },
    {
      "epoch": 0.09719934102141681,
      "grad_norm": 0.6033599972724915,
      "learning_rate": 8.44676704559283e-05,
      "loss": 0.1688,
      "step": 59
    },
    {
      "epoch": 0.09884678747940692,
      "grad_norm": 0.7072098851203918,
      "learning_rate": 8.386407858128706e-05,
      "loss": 0.2107,
      "step": 60
    },
    {
      "epoch": 0.10049423393739704,
      "grad_norm": 0.7184508442878723,
      "learning_rate": 8.32512286056924e-05,
      "loss": 0.1756,
      "step": 61
    },
    {
      "epoch": 0.10214168039538715,
      "grad_norm": 0.6584041118621826,
      "learning_rate": 8.262928807620843e-05,
      "loss": 0.1566,
      "step": 62
    },
    {
      "epoch": 0.10378912685337727,
      "grad_norm": 0.6446477770805359,
      "learning_rate": 8.199842702516583e-05,
      "loss": 0.1917,
      "step": 63
    },
    {
      "epoch": 0.10543657331136738,
      "grad_norm": 0.8058052659034729,
      "learning_rate": 8.135881792367686e-05,
      "loss": 0.1951,
      "step": 64
    },
    {
      "epoch": 0.1070840197693575,
      "grad_norm": 0.678106427192688,
      "learning_rate": 8.07106356344834e-05,
      "loss": 0.1991,
      "step": 65
    },
    {
      "epoch": 0.10873146622734761,
      "grad_norm": 0.7449752688407898,
      "learning_rate": 8.005405736415126e-05,
      "loss": 0.2028,
      "step": 66
    },
    {
      "epoch": 0.11037891268533773,
      "grad_norm": 0.7255269289016724,
      "learning_rate": 7.938926261462366e-05,
      "loss": 0.1769,
      "step": 67
    },
    {
      "epoch": 0.11202635914332784,
      "grad_norm": 0.8328843116760254,
      "learning_rate": 7.871643313414718e-05,
      "loss": 0.2876,
      "step": 68
    },
    {
      "epoch": 0.11367380560131796,
      "grad_norm": 0.5144248008728027,
      "learning_rate": 7.803575286758364e-05,
      "loss": 0.1326,
      "step": 69
    },
    {
      "epoch": 0.11532125205930807,
      "grad_norm": 0.7131458520889282,
      "learning_rate": 7.734740790612136e-05,
      "loss": 0.1873,
      "step": 70
    },
    {
      "epoch": 0.11696869851729819,
      "grad_norm": 0.5682728290557861,
      "learning_rate": 7.66515864363997e-05,
      "loss": 0.1717,
      "step": 71
    },
    {
      "epoch": 0.1186161449752883,
      "grad_norm": 0.4846824109554291,
      "learning_rate": 7.594847868906076e-05,
      "loss": 0.1093,
      "step": 72
    },
    {
      "epoch": 0.12026359143327842,
      "grad_norm": 0.8429383039474487,
      "learning_rate": 7.52382768867422e-05,
      "loss": 0.2256,
      "step": 73
    },
    {
      "epoch": 0.12191103789126853,
      "grad_norm": 0.7328072786331177,
      "learning_rate": 7.452117519152542e-05,
      "loss": 0.1978,
      "step": 74
    },
    {
      "epoch": 0.12355848434925865,
      "grad_norm": 0.6630023121833801,
      "learning_rate": 7.379736965185368e-05,
      "loss": 0.1364,
      "step": 75
    },
    {
      "epoch": 0.12520593080724876,
      "grad_norm": 0.9222315549850464,
      "learning_rate": 7.30670581489344e-05,
      "loss": 0.1996,
      "step": 76
    },
    {
      "epoch": 0.12685337726523888,
      "grad_norm": 0.8272705078125,
      "learning_rate": 7.233044034264034e-05,
      "loss": 0.1856,
      "step": 77
    },
    {
      "epoch": 0.128500823723229,
      "grad_norm": 0.8853323459625244,
      "learning_rate": 7.158771761692464e-05,
      "loss": 0.1633,
      "step": 78
    },
    {
      "epoch": 0.1301482701812191,
      "grad_norm": 0.729363203048706,
      "learning_rate": 7.083909302476453e-05,
      "loss": 0.1668,
      "step": 79
    },
    {
      "epoch": 0.13179571663920922,
      "grad_norm": 0.9551092386245728,
      "learning_rate": 7.008477123264848e-05,
      "loss": 0.1381,
      "step": 80
    },
    {
      "epoch": 0.13344316309719934,
      "grad_norm": 0.8506342768669128,
      "learning_rate": 6.932495846462261e-05,
      "loss": 0.1826,
      "step": 81
    },
    {
      "epoch": 0.13509060955518945,
      "grad_norm": 0.9394629597663879,
      "learning_rate": 6.855986244591104e-05,
      "loss": 0.1269,
      "step": 82
    },
    {
      "epoch": 0.13673805601317957,
      "grad_norm": 0.7383860349655151,
      "learning_rate": 6.778969234612584e-05,
      "loss": 0.1626,
      "step": 83
    },
    {
      "epoch": 0.13838550247116968,
      "grad_norm": 1.0318554639816284,
      "learning_rate": 6.701465872208216e-05,
      "loss": 0.2443,
      "step": 84
    },
    {
      "epoch": 0.1400329489291598,
      "grad_norm": 0.7354373335838318,
      "learning_rate": 6.623497346023418e-05,
      "loss": 0.175,
      "step": 85
    },
    {
      "epoch": 0.1416803953871499,
      "grad_norm": 0.8678761720657349,
      "learning_rate": 6.545084971874738e-05,
      "loss": 0.2204,
      "step": 86
    },
    {
      "epoch": 0.14332784184514002,
      "grad_norm": 0.564337968826294,
      "learning_rate": 6.466250186922325e-05,
      "loss": 0.0978,
      "step": 87
    },
    {
      "epoch": 0.14497528830313014,
      "grad_norm": 0.8833882808685303,
      "learning_rate": 6.387014543809223e-05,
      "loss": 0.1214,
      "step": 88
    },
    {
      "epoch": 0.14662273476112025,
      "grad_norm": 0.6987719535827637,
      "learning_rate": 6.307399704769099e-05,
      "loss": 0.1769,
      "step": 89
    },
    {
      "epoch": 0.14827018121911037,
      "grad_norm": 0.7246816158294678,
      "learning_rate": 6.227427435703997e-05,
      "loss": 0.158,
      "step": 90
    },
    {
      "epoch": 0.14991762767710048,
      "grad_norm": 0.7187811136245728,
      "learning_rate": 6.147119600233758e-05,
      "loss": 0.1393,
      "step": 91
    },
    {
      "epoch": 0.1515650741350906,
      "grad_norm": 0.6391703486442566,
      "learning_rate": 6.066498153718735e-05,
      "loss": 0.1127,
      "step": 92
    },
    {
      "epoch": 0.15321252059308071,
      "grad_norm": 0.7246570587158203,
      "learning_rate": 5.985585137257401e-05,
      "loss": 0.1757,
      "step": 93
    },
    {
      "epoch": 0.15485996705107083,
      "grad_norm": 0.6877361536026001,
      "learning_rate": 5.90440267166055e-05,
      "loss": 0.1327,
      "step": 94
    },
    {
      "epoch": 0.15650741350906094,
      "grad_norm": 0.9020498394966125,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 0.1763,
      "step": 95
    },
    {
      "epoch": 0.15815485996705106,
      "grad_norm": 0.6346703767776489,
      "learning_rate": 5.74131823855921e-05,
      "loss": 0.1508,
      "step": 96
    },
    {
      "epoch": 0.15980230642504117,
      "grad_norm": 0.8068584203720093,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 0.2025,
      "step": 97
    },
    {
      "epoch": 0.1614497528830313,
      "grad_norm": 0.5660220980644226,
      "learning_rate": 5.577423184847932e-05,
      "loss": 0.1149,
      "step": 98
    },
    {
      "epoch": 0.1630971993410214,
      "grad_norm": 0.5893715023994446,
      "learning_rate": 5.495227651252315e-05,
      "loss": 0.069,
      "step": 99
    },
    {
      "epoch": 0.16474464579901152,
      "grad_norm": 0.7794016599655151,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 0.1549,
      "step": 100
    },
    {
      "epoch": 0.16474464579901152,
      "eval_loss": 0.04567544162273407,
      "eval_runtime": 84.2517,
      "eval_samples_per_second": 12.142,
      "eval_steps_per_second": 3.039,
      "step": 100
    },
    {
      "epoch": 0.16639209225700163,
      "grad_norm": 0.9307386875152588,
      "learning_rate": 5.330452921628497e-05,
      "loss": 0.2645,
      "step": 101
    },
    {
      "epoch": 0.16803953871499178,
      "grad_norm": 0.7548027634620667,
      "learning_rate": 5.247918773366112e-05,
      "loss": 0.2228,
      "step": 102
    },
    {
      "epoch": 0.1696869851729819,
      "grad_norm": 0.5785323977470398,
      "learning_rate": 5.165316846586541e-05,
      "loss": 0.1922,
      "step": 103
    },
    {
      "epoch": 0.171334431630972,
      "grad_norm": 0.5022085905075073,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 0.1828,
      "step": 104
    },
    {
      "epoch": 0.17298187808896212,
      "grad_norm": 0.5046458840370178,
      "learning_rate": 5e-05,
      "loss": 0.165,
      "step": 105
    },
    {
      "epoch": 0.17462932454695224,
      "grad_norm": 0.6378635764122009,
      "learning_rate": 4.917330276168208e-05,
      "loss": 0.2051,
      "step": 106
    },
    {
      "epoch": 0.17627677100494235,
      "grad_norm": 0.5523620843887329,
      "learning_rate": 4.834683153413459e-05,
      "loss": 0.1925,
      "step": 107
    },
    {
      "epoch": 0.17792421746293247,
      "grad_norm": 0.4054763913154602,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 0.1083,
      "step": 108
    },
    {
      "epoch": 0.17957166392092258,
      "grad_norm": 0.5592818856239319,
      "learning_rate": 4.669547078371504e-05,
      "loss": 0.1956,
      "step": 109
    },
    {
      "epoch": 0.1812191103789127,
      "grad_norm": 0.6077075004577637,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 0.2045,
      "step": 110
    },
    {
      "epoch": 0.1828665568369028,
      "grad_norm": 0.6907646059989929,
      "learning_rate": 4.504772348747687e-05,
      "loss": 0.2208,
      "step": 111
    },
    {
      "epoch": 0.18451400329489293,
      "grad_norm": 0.6144804358482361,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 0.1776,
      "step": 112
    },
    {
      "epoch": 0.18616144975288304,
      "grad_norm": 0.5765202641487122,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 0.1738,
      "step": 113
    },
    {
      "epoch": 0.18780889621087316,
      "grad_norm": 0.4645479917526245,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 0.1554,
      "step": 114
    },
    {
      "epoch": 0.18945634266886327,
      "grad_norm": 0.5722184181213379,
      "learning_rate": 4.17702704859633e-05,
      "loss": 0.1935,
      "step": 115
    },
    {
      "epoch": 0.19110378912685339,
      "grad_norm": 0.6504333019256592,
      "learning_rate": 4.095597328339452e-05,
      "loss": 0.216,
      "step": 116
    },
    {
      "epoch": 0.1927512355848435,
      "grad_norm": 0.5393025279045105,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 0.1731,
      "step": 117
    },
    {
      "epoch": 0.19439868204283361,
      "grad_norm": 0.6009727120399475,
      "learning_rate": 3.933501846281267e-05,
      "loss": 0.1825,
      "step": 118
    },
    {
      "epoch": 0.19604612850082373,
      "grad_norm": 0.5557132363319397,
      "learning_rate": 3.852880399766243e-05,
      "loss": 0.1767,
      "step": 119
    },
    {
      "epoch": 0.19769357495881384,
      "grad_norm": 0.5873655080795288,
      "learning_rate": 3.772572564296005e-05,
      "loss": 0.1601,
      "step": 120
    },
    {
      "epoch": 0.19934102141680396,
      "grad_norm": 0.6219244599342346,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 0.1794,
      "step": 121
    },
    {
      "epoch": 0.20098846787479407,
      "grad_norm": 0.5566337704658508,
      "learning_rate": 3.612985456190778e-05,
      "loss": 0.1517,
      "step": 122
    },
    {
      "epoch": 0.2026359143327842,
      "grad_norm": 0.7192270755767822,
      "learning_rate": 3.533749813077677e-05,
      "loss": 0.1929,
      "step": 123
    },
    {
      "epoch": 0.2042833607907743,
      "grad_norm": 0.6337162852287292,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 0.1656,
      "step": 124
    },
    {
      "epoch": 0.20593080724876442,
      "grad_norm": 0.6705856919288635,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 0.1737,
      "step": 125
    },
    {
      "epoch": 0.20757825370675453,
      "grad_norm": 0.5509961843490601,
      "learning_rate": 3.298534127791785e-05,
      "loss": 0.0977,
      "step": 126
    },
    {
      "epoch": 0.20922570016474465,
      "grad_norm": 0.48886772990226746,
      "learning_rate": 3.221030765387417e-05,
      "loss": 0.1088,
      "step": 127
    },
    {
      "epoch": 0.21087314662273476,
      "grad_norm": 0.6949488520622253,
      "learning_rate": 3.144013755408895e-05,
      "loss": 0.1667,
      "step": 128
    },
    {
      "epoch": 0.21252059308072488,
      "grad_norm": 0.5871797204017639,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 0.1447,
      "step": 129
    },
    {
      "epoch": 0.214168039538715,
      "grad_norm": 0.8036453723907471,
      "learning_rate": 2.991522876735154e-05,
      "loss": 0.2199,
      "step": 130
    },
    {
      "epoch": 0.2158154859967051,
      "grad_norm": 0.7174505591392517,
      "learning_rate": 2.916090697523549e-05,
      "loss": 0.1673,
      "step": 131
    },
    {
      "epoch": 0.21746293245469522,
      "grad_norm": 0.5246807336807251,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 0.1216,
      "step": 132
    },
    {
      "epoch": 0.21911037891268534,
      "grad_norm": 0.8228816390037537,
      "learning_rate": 2.766955965735968e-05,
      "loss": 0.1701,
      "step": 133
    },
    {
      "epoch": 0.22075782537067545,
      "grad_norm": 0.573482871055603,
      "learning_rate": 2.693294185106562e-05,
      "loss": 0.1361,
      "step": 134
    },
    {
      "epoch": 0.22240527182866557,
      "grad_norm": 0.529029130935669,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 0.1267,
      "step": 135
    },
    {
      "epoch": 0.22405271828665568,
      "grad_norm": 0.7330461740493774,
      "learning_rate": 2.547882480847461e-05,
      "loss": 0.2012,
      "step": 136
    },
    {
      "epoch": 0.2257001647446458,
      "grad_norm": 0.7127494812011719,
      "learning_rate": 2.476172311325783e-05,
      "loss": 0.1901,
      "step": 137
    },
    {
      "epoch": 0.2273476112026359,
      "grad_norm": 0.748142659664154,
      "learning_rate": 2.405152131093926e-05,
      "loss": 0.1538,
      "step": 138
    },
    {
      "epoch": 0.22899505766062603,
      "grad_norm": 0.755768358707428,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 0.1968,
      "step": 139
    },
    {
      "epoch": 0.23064250411861614,
      "grad_norm": 0.6093936562538147,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 0.14,
      "step": 140
    },
    {
      "epoch": 0.23228995057660626,
      "grad_norm": 0.6513073444366455,
      "learning_rate": 2.196424713241637e-05,
      "loss": 0.1318,
      "step": 141
    },
    {
      "epoch": 0.23393739703459637,
      "grad_norm": 0.4579865038394928,
      "learning_rate": 2.128356686585282e-05,
      "loss": 0.0715,
      "step": 142
    },
    {
      "epoch": 0.2355848434925865,
      "grad_norm": 0.6177977919578552,
      "learning_rate": 2.061073738537635e-05,
      "loss": 0.0951,
      "step": 143
    },
    {
      "epoch": 0.2372322899505766,
      "grad_norm": 0.653860330581665,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 0.1404,
      "step": 144
    },
    {
      "epoch": 0.23887973640856672,
      "grad_norm": 0.6635512113571167,
      "learning_rate": 1.928936436551661e-05,
      "loss": 0.1278,
      "step": 145
    },
    {
      "epoch": 0.24052718286655683,
      "grad_norm": 0.6198115944862366,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 0.1124,
      "step": 146
    },
    {
      "epoch": 0.24217462932454695,
      "grad_norm": 0.5520274639129639,
      "learning_rate": 1.800157297483417e-05,
      "loss": 0.0924,
      "step": 147
    },
    {
      "epoch": 0.24382207578253706,
      "grad_norm": 0.8238588571548462,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 0.2046,
      "step": 148
    },
    {
      "epoch": 0.24546952224052718,
      "grad_norm": 0.5436566472053528,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 0.1069,
      "step": 149
    },
    {
      "epoch": 0.2471169686985173,
      "grad_norm": 1.7290219068527222,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 0.21,
      "step": 150
    },
    {
      "epoch": 0.2471169686985173,
      "eval_loss": 0.03943052142858505,
      "eval_runtime": 84.2421,
      "eval_samples_per_second": 12.144,
      "eval_steps_per_second": 3.039,
      "step": 150
    },
    {
      "epoch": 0.2487644151565074,
      "grad_norm": 0.5151445269584656,
      "learning_rate": 1.553232954407171e-05,
      "loss": 0.16,
      "step": 151
    },
    {
      "epoch": 0.2504118616144975,
      "grad_norm": 0.6329994797706604,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 0.2003,
      "step": 152
    },
    {
      "epoch": 0.25205930807248766,
      "grad_norm": 0.6546493768692017,
      "learning_rate": 1.435357758543015e-05,
      "loss": 0.1878,
      "step": 153
    },
    {
      "epoch": 0.25370675453047775,
      "grad_norm": 0.5207836627960205,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 0.1798,
      "step": 154
    },
    {
      "epoch": 0.2553542009884679,
      "grad_norm": 0.5409952998161316,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 0.1735,
      "step": 155
    },
    {
      "epoch": 0.257001647446458,
      "grad_norm": 0.4438985586166382,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 0.1577,
      "step": 156
    },
    {
      "epoch": 0.2586490939044481,
      "grad_norm": 0.5339425802230835,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 0.1995,
      "step": 157
    },
    {
      "epoch": 0.2602965403624382,
      "grad_norm": 0.5247253179550171,
      "learning_rate": 1.157994445715706e-05,
      "loss": 0.1777,
      "step": 158
    },
    {
      "epoch": 0.26194398682042835,
      "grad_norm": 0.43265119194984436,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 0.1535,
      "step": 159
    },
    {
      "epoch": 0.26359143327841844,
      "grad_norm": 0.4654902517795563,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 0.162,
      "step": 160
    },
    {
      "epoch": 0.2652388797364086,
      "grad_norm": 0.5447457432746887,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 0.166,
      "step": 161
    },
    {
      "epoch": 0.26688632619439867,
      "grad_norm": 0.47465264797210693,
      "learning_rate": 9.549150281252633e-06,
      "loss": 0.1604,
      "step": 162
    },
    {
      "epoch": 0.2685337726523888,
      "grad_norm": 0.5062435865402222,
      "learning_rate": 9.068759265665384e-06,
      "loss": 0.1685,
      "step": 163
    },
    {
      "epoch": 0.2701812191103789,
      "grad_norm": 0.5680501461029053,
      "learning_rate": 8.599558442598998e-06,
      "loss": 0.1819,
      "step": 164
    },
    {
      "epoch": 0.27182866556836904,
      "grad_norm": 0.48920100927352905,
      "learning_rate": 8.141676086873572e-06,
      "loss": 0.1563,
      "step": 165
    },
    {
      "epoch": 0.27347611202635913,
      "grad_norm": 0.5608918070793152,
      "learning_rate": 7.695237378953223e-06,
      "loss": 0.1693,
      "step": 166
    },
    {
      "epoch": 0.2751235584843493,
      "grad_norm": 0.5724075436592102,
      "learning_rate": 7.260364370723044e-06,
      "loss": 0.1826,
      "step": 167
    },
    {
      "epoch": 0.27677100494233936,
      "grad_norm": 1.1055313348770142,
      "learning_rate": 6.837175952121306e-06,
      "loss": 0.2617,
      "step": 168
    },
    {
      "epoch": 0.2784184514003295,
      "grad_norm": 0.5200842022895813,
      "learning_rate": 6.425787818636131e-06,
      "loss": 0.1497,
      "step": 169
    },
    {
      "epoch": 0.2800658978583196,
      "grad_norm": 0.4871165454387665,
      "learning_rate": 6.026312439675552e-06,
      "loss": 0.1362,
      "step": 170
    },
    {
      "epoch": 0.28171334431630973,
      "grad_norm": 0.7262673377990723,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 0.288,
      "step": 171
    },
    {
      "epoch": 0.2833607907742998,
      "grad_norm": 0.4437982439994812,
      "learning_rate": 5.263533508961827e-06,
      "loss": 0.1314,
      "step": 172
    },
    {
      "epoch": 0.28500823723228996,
      "grad_norm": 0.5510936975479126,
      "learning_rate": 4.900438493352055e-06,
      "loss": 0.178,
      "step": 173
    },
    {
      "epoch": 0.28665568369028005,
      "grad_norm": 0.46276822686195374,
      "learning_rate": 4.549673247541875e-06,
      "loss": 0.1342,
      "step": 174
    },
    {
      "epoch": 0.2883031301482702,
      "grad_norm": 0.559018611907959,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 0.1601,
      "step": 175
    },
    {
      "epoch": 0.2899505766062603,
      "grad_norm": 0.5030566453933716,
      "learning_rate": 3.885512251130763e-06,
      "loss": 0.1217,
      "step": 176
    },
    {
      "epoch": 0.2915980230642504,
      "grad_norm": 0.4926145672798157,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 0.1505,
      "step": 177
    },
    {
      "epoch": 0.2932454695222405,
      "grad_norm": 0.6175839304924011,
      "learning_rate": 3.271776770026963e-06,
      "loss": 0.1407,
      "step": 178
    },
    {
      "epoch": 0.29489291598023065,
      "grad_norm": 0.5800250768661499,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 0.1563,
      "step": 179
    },
    {
      "epoch": 0.29654036243822074,
      "grad_norm": 0.5808983445167542,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 0.1399,
      "step": 180
    },
    {
      "epoch": 0.2981878088962109,
      "grad_norm": 0.604492723941803,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 0.1442,
      "step": 181
    },
    {
      "epoch": 0.29983525535420097,
      "grad_norm": 0.6174325346946716,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 0.1343,
      "step": 182
    },
    {
      "epoch": 0.3014827018121911,
      "grad_norm": 0.6970261931419373,
      "learning_rate": 1.962316193157593e-06,
      "loss": 0.2081,
      "step": 183
    },
    {
      "epoch": 0.3031301482701812,
      "grad_norm": 0.6464529633522034,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 0.1841,
      "step": 184
    },
    {
      "epoch": 0.30477759472817134,
      "grad_norm": 0.5047242641448975,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 0.1433,
      "step": 185
    },
    {
      "epoch": 0.30642504118616143,
      "grad_norm": 0.659187376499176,
      "learning_rate": 1.333670137599713e-06,
      "loss": 0.1724,
      "step": 186
    },
    {
      "epoch": 0.30807248764415157,
      "grad_norm": 0.642983078956604,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 0.1273,
      "step": 187
    },
    {
      "epoch": 0.30971993410214166,
      "grad_norm": 0.48810455203056335,
      "learning_rate": 9.810017062595322e-07,
      "loss": 0.1064,
      "step": 188
    },
    {
      "epoch": 0.3113673805601318,
      "grad_norm": 0.44890350103378296,
      "learning_rate": 8.247462563808817e-07,
      "loss": 0.0808,
      "step": 189
    },
    {
      "epoch": 0.3130148270181219,
      "grad_norm": 0.5581908226013184,
      "learning_rate": 6.819348298638839e-07,
      "loss": 0.1335,
      "step": 190
    },
    {
      "epoch": 0.31466227347611203,
      "grad_norm": 0.5685779452323914,
      "learning_rate": 5.526064699265753e-07,
      "loss": 0.1121,
      "step": 191
    },
    {
      "epoch": 0.3163097199341021,
      "grad_norm": 0.4967321455478668,
      "learning_rate": 4.367965336512403e-07,
      "loss": 0.1154,
      "step": 192
    },
    {
      "epoch": 0.31795716639209226,
      "grad_norm": 0.4525260031223297,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 0.1,
      "step": 193
    },
    {
      "epoch": 0.31960461285008235,
      "grad_norm": 0.5718578100204468,
      "learning_rate": 2.458548727494292e-07,
      "loss": 0.0937,
      "step": 194
    },
    {
      "epoch": 0.3212520593080725,
      "grad_norm": 0.5804625749588013,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 0.1273,
      "step": 195
    },
    {
      "epoch": 0.3228995057660626,
      "grad_norm": 0.5510926246643066,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 0.122,
      "step": 196
    },
    {
      "epoch": 0.3245469522240527,
      "grad_norm": 0.6820921301841736,
      "learning_rate": 6.150154258476315e-08,
      "loss": 0.1458,
      "step": 197
    },
    {
      "epoch": 0.3261943986820428,
      "grad_norm": 0.6273231506347656,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 0.1376,
      "step": 198
    },
    {
      "epoch": 0.32784184514003295,
      "grad_norm": 0.51357102394104,
      "learning_rate": 6.834750376549792e-09,
      "loss": 0.099,
      "step": 199
    },
    {
      "epoch": 0.32948929159802304,
      "grad_norm": 0.7212042808532715,
      "learning_rate": 0.0,
      "loss": 0.1419,
      "step": 200
    },
    {
      "epoch": 0.32948929159802304,
      "eval_loss": 0.03573817387223244,
      "eval_runtime": 84.2598,
      "eval_samples_per_second": 12.141,
      "eval_steps_per_second": 3.038,
      "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": 3.1495562654121984e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}