{
  "best_metric": 1.1696622371673584,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.08309961566427755,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00041549807832138777,
      "grad_norm": 2.299808979034424,
      "learning_rate": 1e-05,
      "loss": 1.5405,
      "step": 1
    },
    {
      "epoch": 0.00041549807832138777,
      "eval_loss": 1.9200119972229004,
      "eval_runtime": 181.6078,
      "eval_samples_per_second": 22.323,
      "eval_steps_per_second": 5.583,
      "step": 1
    },
    {
      "epoch": 0.0008309961566427755,
      "grad_norm": 2.7252023220062256,
      "learning_rate": 2e-05,
      "loss": 1.6341,
      "step": 2
    },
    {
      "epoch": 0.0012464942349641633,
      "grad_norm": 2.2323546409606934,
      "learning_rate": 3e-05,
      "loss": 1.3936,
      "step": 3
    },
    {
      "epoch": 0.001661992313285551,
      "grad_norm": 2.095740795135498,
      "learning_rate": 4e-05,
      "loss": 1.6867,
      "step": 4
    },
    {
      "epoch": 0.0020774903916069387,
      "grad_norm": 1.5457115173339844,
      "learning_rate": 5e-05,
      "loss": 1.56,
      "step": 5
    },
    {
      "epoch": 0.0024929884699283265,
      "grad_norm": 1.5662128925323486,
      "learning_rate": 6e-05,
      "loss": 1.5136,
      "step": 6
    },
    {
      "epoch": 0.0029084865482497143,
      "grad_norm": 1.6319241523742676,
      "learning_rate": 7e-05,
      "loss": 1.3458,
      "step": 7
    },
    {
      "epoch": 0.003323984626571102,
      "grad_norm": 1.3970435857772827,
      "learning_rate": 8e-05,
      "loss": 1.4031,
      "step": 8
    },
    {
      "epoch": 0.00373948270489249,
      "grad_norm": 1.4588398933410645,
      "learning_rate": 9e-05,
      "loss": 1.3809,
      "step": 9
    },
    {
      "epoch": 0.004154980783213877,
      "grad_norm": 1.294360876083374,
      "learning_rate": 0.0001,
      "loss": 1.2736,
      "step": 10
    },
    {
      "epoch": 0.004570478861535265,
      "grad_norm": 1.28848135471344,
      "learning_rate": 9.999316524962345e-05,
      "loss": 1.2026,
      "step": 11
    },
    {
      "epoch": 0.004985976939856653,
      "grad_norm": 1.3200008869171143,
      "learning_rate": 9.997266286704631e-05,
      "loss": 1.2535,
      "step": 12
    },
    {
      "epoch": 0.005401475018178041,
      "grad_norm": 1.4708698987960815,
      "learning_rate": 9.993849845741524e-05,
      "loss": 1.3189,
      "step": 13
    },
    {
      "epoch": 0.005816973096499429,
      "grad_norm": 1.0938494205474854,
      "learning_rate": 9.989068136093873e-05,
      "loss": 1.3409,
      "step": 14
    },
    {
      "epoch": 0.0062324711748208165,
      "grad_norm": 0.978922426700592,
      "learning_rate": 9.98292246503335e-05,
      "loss": 1.2778,
      "step": 15
    },
    {
      "epoch": 0.006647969253142204,
      "grad_norm": 1.1084084510803223,
      "learning_rate": 9.975414512725057e-05,
      "loss": 1.333,
      "step": 16
    },
    {
      "epoch": 0.007063467331463592,
      "grad_norm": 1.0981535911560059,
      "learning_rate": 9.966546331768191e-05,
      "loss": 1.3452,
      "step": 17
    },
    {
      "epoch": 0.00747896540978498,
      "grad_norm": 0.959189772605896,
      "learning_rate": 9.956320346634876e-05,
      "loss": 1.1853,
      "step": 18
    },
    {
      "epoch": 0.007894463488106367,
      "grad_norm": 1.0055214166641235,
      "learning_rate": 9.944739353007344e-05,
      "loss": 1.2398,
      "step": 19
    },
    {
      "epoch": 0.008309961566427755,
      "grad_norm": 1.0951157808303833,
      "learning_rate": 9.931806517013612e-05,
      "loss": 1.3896,
      "step": 20
    },
    {
      "epoch": 0.008725459644749143,
      "grad_norm": 1.1027330160140991,
      "learning_rate": 9.917525374361912e-05,
      "loss": 1.3814,
      "step": 21
    },
    {
      "epoch": 0.00914095772307053,
      "grad_norm": 1.0288090705871582,
      "learning_rate": 9.901899829374047e-05,
      "loss": 1.4062,
      "step": 22
    },
    {
      "epoch": 0.009556455801391918,
      "grad_norm": 1.0395174026489258,
      "learning_rate": 9.884934153917997e-05,
      "loss": 1.2857,
      "step": 23
    },
    {
      "epoch": 0.009971953879713306,
      "grad_norm": 1.0116403102874756,
      "learning_rate": 9.86663298624003e-05,
      "loss": 1.1877,
      "step": 24
    },
    {
      "epoch": 0.010387451958034694,
      "grad_norm": 1.0476988554000854,
      "learning_rate": 9.847001329696653e-05,
      "loss": 1.3134,
      "step": 25
    },
    {
      "epoch": 0.010802950036356082,
      "grad_norm": 1.0091525316238403,
      "learning_rate": 9.826044551386744e-05,
      "loss": 1.2218,
      "step": 26
    },
    {
      "epoch": 0.01121844811467747,
      "grad_norm": 1.1096187829971313,
      "learning_rate": 9.803768380684242e-05,
      "loss": 1.2584,
      "step": 27
    },
    {
      "epoch": 0.011633946192998857,
      "grad_norm": 0.9986774921417236,
      "learning_rate": 9.780178907671789e-05,
      "loss": 1.3315,
      "step": 28
    },
    {
      "epoch": 0.012049444271320245,
      "grad_norm": 0.998517632484436,
      "learning_rate": 9.755282581475769e-05,
      "loss": 1.2799,
      "step": 29
    },
    {
      "epoch": 0.012464942349641633,
      "grad_norm": 1.0310990810394287,
      "learning_rate": 9.729086208503174e-05,
      "loss": 1.2712,
      "step": 30
    },
    {
      "epoch": 0.012880440427963021,
      "grad_norm": 1.0967302322387695,
      "learning_rate": 9.701596950580806e-05,
      "loss": 1.3682,
      "step": 31
    },
    {
      "epoch": 0.013295938506284409,
      "grad_norm": 1.0782688856124878,
      "learning_rate": 9.672822322997305e-05,
      "loss": 1.2672,
      "step": 32
    },
    {
      "epoch": 0.013711436584605797,
      "grad_norm": 1.0700477361679077,
      "learning_rate": 9.642770192448536e-05,
      "loss": 1.2349,
      "step": 33
    },
    {
      "epoch": 0.014126934662927184,
      "grad_norm": 0.9679452180862427,
      "learning_rate": 9.611448774886924e-05,
      "loss": 1.1041,
      "step": 34
    },
    {
      "epoch": 0.014542432741248572,
      "grad_norm": 1.3080668449401855,
      "learning_rate": 9.578866633275288e-05,
      "loss": 1.4034,
      "step": 35
    },
    {
      "epoch": 0.01495793081956996,
      "grad_norm": 1.1441242694854736,
      "learning_rate": 9.545032675245813e-05,
      "loss": 1.1454,
      "step": 36
    },
    {
      "epoch": 0.015373428897891348,
      "grad_norm": 1.1910218000411987,
      "learning_rate": 9.509956150664796e-05,
      "loss": 1.2146,
      "step": 37
    },
    {
      "epoch": 0.015788926976212734,
      "grad_norm": 1.2756946086883545,
      "learning_rate": 9.473646649103818e-05,
      "loss": 1.333,
      "step": 38
    },
    {
      "epoch": 0.016204425054534122,
      "grad_norm": 1.2750654220581055,
      "learning_rate": 9.43611409721806e-05,
      "loss": 1.0768,
      "step": 39
    },
    {
      "epoch": 0.01661992313285551,
      "grad_norm": 1.2668075561523438,
      "learning_rate": 9.397368756032445e-05,
      "loss": 1.1589,
      "step": 40
    },
    {
      "epoch": 0.017035421211176897,
      "grad_norm": 1.3223789930343628,
      "learning_rate": 9.357421218136386e-05,
      "loss": 1.303,
      "step": 41
    },
    {
      "epoch": 0.017450919289498285,
      "grad_norm": 1.2531617879867554,
      "learning_rate": 9.316282404787871e-05,
      "loss": 1.1371,
      "step": 42
    },
    {
      "epoch": 0.017866417367819673,
      "grad_norm": 1.3075684309005737,
      "learning_rate": 9.273963562927695e-05,
      "loss": 1.2465,
      "step": 43
    },
    {
      "epoch": 0.01828191544614106,
      "grad_norm": 1.1558737754821777,
      "learning_rate": 9.230476262104677e-05,
      "loss": 1.0655,
      "step": 44
    },
    {
      "epoch": 0.01869741352446245,
      "grad_norm": 1.4140853881835938,
      "learning_rate": 9.185832391312644e-05,
      "loss": 1.052,
      "step": 45
    },
    {
      "epoch": 0.019112911602783837,
      "grad_norm": 1.154665470123291,
      "learning_rate": 9.140044155740101e-05,
      "loss": 0.9842,
      "step": 46
    },
    {
      "epoch": 0.019528409681105224,
      "grad_norm": 1.1338372230529785,
      "learning_rate": 9.093124073433463e-05,
      "loss": 0.9983,
      "step": 47
    },
    {
      "epoch": 0.019943907759426612,
      "grad_norm": 1.2601895332336426,
      "learning_rate": 9.045084971874738e-05,
      "loss": 0.867,
      "step": 48
    },
    {
      "epoch": 0.020359405837748,
      "grad_norm": 1.6295559406280518,
      "learning_rate": 8.995939984474624e-05,
      "loss": 0.956,
      "step": 49
    },
    {
      "epoch": 0.020774903916069388,
      "grad_norm": 2.3988633155822754,
      "learning_rate": 8.945702546981969e-05,
      "loss": 1.1806,
      "step": 50
    },
    {
      "epoch": 0.020774903916069388,
      "eval_loss": 1.439509391784668,
      "eval_runtime": 182.3291,
      "eval_samples_per_second": 22.235,
      "eval_steps_per_second": 5.561,
      "step": 50
    },
    {
      "epoch": 0.021190401994390776,
      "grad_norm": 3.1769156455993652,
      "learning_rate": 8.894386393810563e-05,
      "loss": 1.3909,
      "step": 51
    },
    {
      "epoch": 0.021605900072712163,
      "grad_norm": 3.3555750846862793,
      "learning_rate": 8.842005554284296e-05,
      "loss": 1.4634,
      "step": 52
    },
    {
      "epoch": 0.02202139815103355,
      "grad_norm": 2.0515425205230713,
      "learning_rate": 8.788574348801675e-05,
      "loss": 1.4994,
      "step": 53
    },
    {
      "epoch": 0.02243689622935494,
      "grad_norm": 1.1126914024353027,
      "learning_rate": 8.73410738492077e-05,
      "loss": 1.2088,
      "step": 54
    },
    {
      "epoch": 0.022852394307676327,
      "grad_norm": 0.7931724190711975,
      "learning_rate": 8.678619553365659e-05,
      "loss": 1.2538,
      "step": 55
    },
    {
      "epoch": 0.023267892385997715,
      "grad_norm": 0.9478723406791687,
      "learning_rate": 8.622126023955446e-05,
      "loss": 1.273,
      "step": 56
    },
    {
      "epoch": 0.023683390464319103,
      "grad_norm": 0.9318475127220154,
      "learning_rate": 8.564642241456986e-05,
      "loss": 1.2882,
      "step": 57
    },
    {
      "epoch": 0.02409888854264049,
      "grad_norm": 0.9630898237228394,
      "learning_rate": 8.506183921362443e-05,
      "loss": 1.2167,
      "step": 58
    },
    {
      "epoch": 0.024514386620961878,
      "grad_norm": 0.8153641819953918,
      "learning_rate": 8.44676704559283e-05,
      "loss": 1.1872,
      "step": 59
    },
    {
      "epoch": 0.024929884699283266,
      "grad_norm": 0.8451074361801147,
      "learning_rate": 8.386407858128706e-05,
      "loss": 1.2953,
      "step": 60
    },
    {
      "epoch": 0.025345382777604654,
      "grad_norm": 0.843819797039032,
      "learning_rate": 8.32512286056924e-05,
      "loss": 1.3942,
      "step": 61
    },
    {
      "epoch": 0.025760880855926042,
      "grad_norm": 0.8349138498306274,
      "learning_rate": 8.262928807620843e-05,
      "loss": 1.3803,
      "step": 62
    },
    {
      "epoch": 0.02617637893424743,
      "grad_norm": 0.8184905648231506,
      "learning_rate": 8.199842702516583e-05,
      "loss": 1.1478,
      "step": 63
    },
    {
      "epoch": 0.026591877012568817,
      "grad_norm": 0.8221478462219238,
      "learning_rate": 8.135881792367686e-05,
      "loss": 1.2182,
      "step": 64
    },
    {
      "epoch": 0.027007375090890205,
      "grad_norm": 0.845354437828064,
      "learning_rate": 8.07106356344834e-05,
      "loss": 1.2299,
      "step": 65
    },
    {
      "epoch": 0.027422873169211593,
      "grad_norm": 0.8170358538627625,
      "learning_rate": 8.005405736415126e-05,
      "loss": 1.1529,
      "step": 66
    },
    {
      "epoch": 0.02783837124753298,
      "grad_norm": 0.8231884837150574,
      "learning_rate": 7.938926261462366e-05,
      "loss": 1.3533,
      "step": 67
    },
    {
      "epoch": 0.02825386932585437,
      "grad_norm": 0.819172203540802,
      "learning_rate": 7.871643313414718e-05,
      "loss": 1.1787,
      "step": 68
    },
    {
      "epoch": 0.028669367404175757,
      "grad_norm": 0.7549390196800232,
      "learning_rate": 7.803575286758364e-05,
      "loss": 1.1401,
      "step": 69
    },
    {
      "epoch": 0.029084865482497144,
      "grad_norm": 0.8602890968322754,
      "learning_rate": 7.734740790612136e-05,
      "loss": 1.4203,
      "step": 70
    },
    {
      "epoch": 0.029500363560818532,
      "grad_norm": 0.7742926478385925,
      "learning_rate": 7.66515864363997e-05,
      "loss": 1.1202,
      "step": 71
    },
    {
      "epoch": 0.02991586163913992,
      "grad_norm": 0.800258457660675,
      "learning_rate": 7.594847868906076e-05,
      "loss": 1.1451,
      "step": 72
    },
    {
      "epoch": 0.030331359717461308,
      "grad_norm": 0.8634665012359619,
      "learning_rate": 7.52382768867422e-05,
      "loss": 1.2659,
      "step": 73
    },
    {
      "epoch": 0.030746857795782696,
      "grad_norm": 0.8620253801345825,
      "learning_rate": 7.452117519152542e-05,
      "loss": 1.288,
      "step": 74
    },
    {
      "epoch": 0.031162355874104083,
      "grad_norm": 0.8999612927436829,
      "learning_rate": 7.379736965185368e-05,
      "loss": 1.1273,
      "step": 75
    },
    {
      "epoch": 0.03157785395242547,
      "grad_norm": 1.0084432363510132,
      "learning_rate": 7.30670581489344e-05,
      "loss": 1.3074,
      "step": 76
    },
    {
      "epoch": 0.03199335203074686,
      "grad_norm": 0.8809545636177063,
      "learning_rate": 7.233044034264034e-05,
      "loss": 1.0526,
      "step": 77
    },
    {
      "epoch": 0.032408850109068243,
      "grad_norm": 0.9520617127418518,
      "learning_rate": 7.158771761692464e-05,
      "loss": 1.2366,
      "step": 78
    },
    {
      "epoch": 0.032824348187389635,
      "grad_norm": 0.9116857647895813,
      "learning_rate": 7.083909302476453e-05,
      "loss": 1.1937,
      "step": 79
    },
    {
      "epoch": 0.03323984626571102,
      "grad_norm": 0.9769972562789917,
      "learning_rate": 7.008477123264848e-05,
      "loss": 1.0908,
      "step": 80
    },
    {
      "epoch": 0.03365534434403241,
      "grad_norm": 1.044212818145752,
      "learning_rate": 6.932495846462261e-05,
      "loss": 1.3214,
      "step": 81
    },
    {
      "epoch": 0.034070842422353795,
      "grad_norm": 0.9704392552375793,
      "learning_rate": 6.855986244591104e-05,
      "loss": 1.1022,
      "step": 82
    },
    {
      "epoch": 0.034486340500675186,
      "grad_norm": 1.0969246625900269,
      "learning_rate": 6.778969234612584e-05,
      "loss": 1.1317,
      "step": 83
    },
    {
      "epoch": 0.03490183857899657,
      "grad_norm": 1.0266567468643188,
      "learning_rate": 6.701465872208216e-05,
      "loss": 1.3167,
      "step": 84
    },
    {
      "epoch": 0.03531733665731796,
      "grad_norm": 1.1352609395980835,
      "learning_rate": 6.623497346023418e-05,
      "loss": 1.4157,
      "step": 85
    },
    {
      "epoch": 0.035732834735639346,
      "grad_norm": 1.0722988843917847,
      "learning_rate": 6.545084971874738e-05,
      "loss": 1.1612,
      "step": 86
    },
    {
      "epoch": 0.03614833281396074,
      "grad_norm": 0.9580327272415161,
      "learning_rate": 6.466250186922325e-05,
      "loss": 1.1394,
      "step": 87
    },
    {
      "epoch": 0.03656383089228212,
      "grad_norm": 1.0610268115997314,
      "learning_rate": 6.387014543809223e-05,
      "loss": 1.1955,
      "step": 88
    },
    {
      "epoch": 0.03697932897060351,
      "grad_norm": 1.0625702142715454,
      "learning_rate": 6.307399704769099e-05,
      "loss": 1.1178,
      "step": 89
    },
    {
      "epoch": 0.0373948270489249,
      "grad_norm": 1.1978108882904053,
      "learning_rate": 6.227427435703997e-05,
      "loss": 1.2412,
      "step": 90
    },
    {
      "epoch": 0.03781032512724629,
      "grad_norm": 1.5245996713638306,
      "learning_rate": 6.147119600233758e-05,
      "loss": 1.1613,
      "step": 91
    },
    {
      "epoch": 0.03822582320556767,
      "grad_norm": 1.2561701536178589,
      "learning_rate": 6.066498153718735e-05,
      "loss": 1.2828,
      "step": 92
    },
    {
      "epoch": 0.038641321283889064,
      "grad_norm": 1.138969898223877,
      "learning_rate": 5.985585137257401e-05,
      "loss": 1.0369,
      "step": 93
    },
    {
      "epoch": 0.03905681936221045,
      "grad_norm": 1.075325608253479,
      "learning_rate": 5.90440267166055e-05,
      "loss": 0.9806,
      "step": 94
    },
    {
      "epoch": 0.03947231744053184,
      "grad_norm": 1.1118909120559692,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 0.9467,
      "step": 95
    },
    {
      "epoch": 0.039887815518853224,
      "grad_norm": 1.1962519884109497,
      "learning_rate": 5.74131823855921e-05,
      "loss": 1.0262,
      "step": 96
    },
    {
      "epoch": 0.040303313597174616,
      "grad_norm": 1.155336856842041,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 1.0079,
      "step": 97
    },
    {
      "epoch": 0.040718811675496,
      "grad_norm": 1.252110242843628,
      "learning_rate": 5.577423184847932e-05,
      "loss": 0.9519,
      "step": 98
    },
    {
      "epoch": 0.04113430975381739,
      "grad_norm": 1.4085887670516968,
      "learning_rate": 5.495227651252315e-05,
      "loss": 0.9516,
      "step": 99
    },
    {
      "epoch": 0.041549807832138776,
      "grad_norm": 2.212888240814209,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 1.0754,
      "step": 100
    },
    {
      "epoch": 0.041549807832138776,
      "eval_loss": 1.2469104528427124,
      "eval_runtime": 182.4041,
      "eval_samples_per_second": 22.225,
      "eval_steps_per_second": 5.559,
      "step": 100
    },
    {
      "epoch": 0.04196530591046017,
      "grad_norm": 1.7321544885635376,
      "learning_rate": 5.330452921628497e-05,
      "loss": 1.1575,
      "step": 101
    },
    {
      "epoch": 0.04238080398878155,
      "grad_norm": 1.7020206451416016,
      "learning_rate": 5.247918773366112e-05,
      "loss": 1.2863,
      "step": 102
    },
    {
      "epoch": 0.04279630206710294,
      "grad_norm": 1.254441499710083,
      "learning_rate": 5.165316846586541e-05,
      "loss": 1.28,
      "step": 103
    },
    {
      "epoch": 0.04321180014542433,
      "grad_norm": 0.9496528506278992,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 1.2196,
      "step": 104
    },
    {
      "epoch": 0.04362729822374572,
      "grad_norm": 0.723782479763031,
      "learning_rate": 5e-05,
      "loss": 1.1713,
      "step": 105
    },
    {
      "epoch": 0.0440427963020671,
      "grad_norm": 0.763882040977478,
      "learning_rate": 4.917330276168208e-05,
      "loss": 1.258,
      "step": 106
    },
    {
      "epoch": 0.044458294380388494,
      "grad_norm": 0.7494874000549316,
      "learning_rate": 4.834683153413459e-05,
      "loss": 1.2189,
      "step": 107
    },
    {
      "epoch": 0.04487379245870988,
      "grad_norm": 0.7877455949783325,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 1.1551,
      "step": 108
    },
    {
      "epoch": 0.04528929053703127,
      "grad_norm": 0.77644282579422,
      "learning_rate": 4.669547078371504e-05,
      "loss": 1.1094,
      "step": 109
    },
    {
      "epoch": 0.045704788615352654,
      "grad_norm": 0.7959705591201782,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 1.2283,
      "step": 110
    },
    {
      "epoch": 0.046120286693674045,
      "grad_norm": 0.8097056150436401,
      "learning_rate": 4.504772348747687e-05,
      "loss": 1.3555,
      "step": 111
    },
    {
      "epoch": 0.04653578477199543,
      "grad_norm": 0.7497302889823914,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 1.1416,
      "step": 112
    },
    {
      "epoch": 0.046951282850316814,
      "grad_norm": 0.7800290584564209,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 1.2519,
      "step": 113
    },
    {
      "epoch": 0.047366780928638205,
      "grad_norm": 0.7350729703903198,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 1.2185,
      "step": 114
    },
    {
      "epoch": 0.04778227900695959,
      "grad_norm": 0.7567837834358215,
      "learning_rate": 4.17702704859633e-05,
      "loss": 1.1682,
      "step": 115
    },
    {
      "epoch": 0.04819777708528098,
      "grad_norm": 0.7955023646354675,
      "learning_rate": 4.095597328339452e-05,
      "loss": 1.2352,
      "step": 116
    },
    {
      "epoch": 0.048613275163602365,
      "grad_norm": 0.71650230884552,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 1.1192,
      "step": 117
    },
    {
      "epoch": 0.049028773241923757,
      "grad_norm": 0.8122178316116333,
      "learning_rate": 3.933501846281267e-05,
      "loss": 1.258,
      "step": 118
    },
    {
      "epoch": 0.04944427132024514,
      "grad_norm": 0.8235985040664673,
      "learning_rate": 3.852880399766243e-05,
      "loss": 1.2185,
      "step": 119
    },
    {
      "epoch": 0.04985976939856653,
      "grad_norm": 0.8254658579826355,
      "learning_rate": 3.772572564296005e-05,
      "loss": 1.1837,
      "step": 120
    },
    {
      "epoch": 0.050275267476887917,
      "grad_norm": 0.9564327597618103,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 1.3123,
      "step": 121
    },
    {
      "epoch": 0.05069076555520931,
      "grad_norm": 0.8568813800811768,
      "learning_rate": 3.612985456190778e-05,
      "loss": 1.2052,
      "step": 122
    },
    {
      "epoch": 0.05110626363353069,
      "grad_norm": 0.876512348651886,
      "learning_rate": 3.533749813077677e-05,
      "loss": 1.241,
      "step": 123
    },
    {
      "epoch": 0.051521761711852083,
      "grad_norm": 0.8595829606056213,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 1.2149,
      "step": 124
    },
    {
      "epoch": 0.05193725979017347,
      "grad_norm": 0.8422783613204956,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 1.1904,
      "step": 125
    },
    {
      "epoch": 0.05235275786849486,
      "grad_norm": 0.7972773909568787,
      "learning_rate": 3.298534127791785e-05,
      "loss": 1.0429,
      "step": 126
    },
    {
      "epoch": 0.052768255946816243,
      "grad_norm": 0.8569056391716003,
      "learning_rate": 3.221030765387417e-05,
      "loss": 1.2627,
      "step": 127
    },
    {
      "epoch": 0.053183754025137635,
      "grad_norm": 0.8476487994194031,
      "learning_rate": 3.144013755408895e-05,
      "loss": 1.1083,
      "step": 128
    },
    {
      "epoch": 0.05359925210345902,
      "grad_norm": 0.906156063079834,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 1.2393,
      "step": 129
    },
    {
      "epoch": 0.05401475018178041,
      "grad_norm": 1.0482535362243652,
      "learning_rate": 2.991522876735154e-05,
      "loss": 1.2113,
      "step": 130
    },
    {
      "epoch": 0.054430248260101795,
      "grad_norm": 0.9307923316955566,
      "learning_rate": 2.916090697523549e-05,
      "loss": 1.23,
      "step": 131
    },
    {
      "epoch": 0.054845746338423186,
      "grad_norm": 0.9967564344406128,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 1.3719,
      "step": 132
    },
    {
      "epoch": 0.05526124441674457,
      "grad_norm": 0.9464885592460632,
      "learning_rate": 2.766955965735968e-05,
      "loss": 1.2785,
      "step": 133
    },
    {
      "epoch": 0.05567674249506596,
      "grad_norm": 1.049530267715454,
      "learning_rate": 2.693294185106562e-05,
      "loss": 1.2346,
      "step": 134
    },
    {
      "epoch": 0.056092240573387346,
      "grad_norm": 1.0776190757751465,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 1.4616,
      "step": 135
    },
    {
      "epoch": 0.05650773865170874,
      "grad_norm": 1.0015181303024292,
      "learning_rate": 2.547882480847461e-05,
      "loss": 1.1243,
      "step": 136
    },
    {
      "epoch": 0.05692323673003012,
      "grad_norm": 1.1369028091430664,
      "learning_rate": 2.476172311325783e-05,
      "loss": 1.2242,
      "step": 137
    },
    {
      "epoch": 0.05733873480835151,
      "grad_norm": 1.00320565700531,
      "learning_rate": 2.405152131093926e-05,
      "loss": 1.1998,
      "step": 138
    },
    {
      "epoch": 0.0577542328866729,
      "grad_norm": 1.065028429031372,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 1.1566,
      "step": 139
    },
    {
      "epoch": 0.05816973096499429,
      "grad_norm": 1.1584892272949219,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 1.1512,
      "step": 140
    },
    {
      "epoch": 0.05858522904331567,
      "grad_norm": 1.2334502935409546,
      "learning_rate": 2.196424713241637e-05,
      "loss": 1.3979,
      "step": 141
    },
    {
      "epoch": 0.059000727121637064,
      "grad_norm": 1.177543044090271,
      "learning_rate": 2.128356686585282e-05,
      "loss": 1.3537,
      "step": 142
    },
    {
      "epoch": 0.05941622519995845,
      "grad_norm": 1.0648972988128662,
      "learning_rate": 2.061073738537635e-05,
      "loss": 1.0492,
      "step": 143
    },
    {
      "epoch": 0.05983172327827984,
      "grad_norm": 0.9891376495361328,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 0.8041,
      "step": 144
    },
    {
      "epoch": 0.060247221356601224,
      "grad_norm": 0.9964181184768677,
      "learning_rate": 1.928936436551661e-05,
      "loss": 0.9136,
      "step": 145
    },
    {
      "epoch": 0.060662719434922616,
      "grad_norm": 1.0803135633468628,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 0.9841,
      "step": 146
    },
    {
      "epoch": 0.061078217513244,
      "grad_norm": 1.004699945449829,
      "learning_rate": 1.800157297483417e-05,
      "loss": 0.9483,
      "step": 147
    },
    {
      "epoch": 0.06149371559156539,
      "grad_norm": 0.9612074494361877,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 0.8652,
      "step": 148
    },
    {
      "epoch": 0.061909213669886776,
      "grad_norm": 1.1166809797286987,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 0.9987,
      "step": 149
    },
    {
      "epoch": 0.06232471174820817,
      "grad_norm": 1.6318724155426025,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 1.138,
      "step": 150
    },
    {
      "epoch": 0.06232471174820817,
      "eval_loss": 1.1831613779067993,
      "eval_runtime": 182.4426,
      "eval_samples_per_second": 22.221,
      "eval_steps_per_second": 5.558,
      "step": 150
    },
    {
      "epoch": 0.06274020982652956,
      "grad_norm": 0.6914190053939819,
      "learning_rate": 1.553232954407171e-05,
      "loss": 1.0126,
      "step": 151
    },
    {
      "epoch": 0.06315570790485094,
      "grad_norm": 0.8021866679191589,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 1.1986,
      "step": 152
    },
    {
      "epoch": 0.06357120598317233,
      "grad_norm": 0.7927551865577698,
      "learning_rate": 1.435357758543015e-05,
      "loss": 1.0682,
      "step": 153
    },
    {
      "epoch": 0.06398670406149372,
      "grad_norm": 0.8403190970420837,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 1.1827,
      "step": 154
    },
    {
      "epoch": 0.06440220213981511,
      "grad_norm": 0.8856049180030823,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 1.3011,
      "step": 155
    },
    {
      "epoch": 0.06481770021813649,
      "grad_norm": 0.8537068367004395,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 1.2328,
      "step": 156
    },
    {
      "epoch": 0.06523319829645788,
      "grad_norm": 0.8552418351173401,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 1.3571,
      "step": 157
    },
    {
      "epoch": 0.06564869637477927,
      "grad_norm": 0.7934669256210327,
      "learning_rate": 1.157994445715706e-05,
      "loss": 1.0858,
      "step": 158
    },
    {
      "epoch": 0.06606419445310066,
      "grad_norm": 0.7639363408088684,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 1.2377,
      "step": 159
    },
    {
      "epoch": 0.06647969253142204,
      "grad_norm": 0.7871673107147217,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 1.3006,
      "step": 160
    },
    {
      "epoch": 0.06689519060974343,
      "grad_norm": 0.7523965835571289,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 1.2872,
      "step": 161
    },
    {
      "epoch": 0.06731068868806482,
      "grad_norm": 0.8020960092544556,
      "learning_rate": 9.549150281252633e-06,
      "loss": 1.2768,
      "step": 162
    },
    {
      "epoch": 0.06772618676638621,
      "grad_norm": 0.7627438306808472,
      "learning_rate": 9.068759265665384e-06,
      "loss": 1.1554,
      "step": 163
    },
    {
      "epoch": 0.06814168484470759,
      "grad_norm": 0.8107988834381104,
      "learning_rate": 8.599558442598998e-06,
      "loss": 1.2288,
      "step": 164
    },
    {
      "epoch": 0.06855718292302898,
      "grad_norm": 0.7380449771881104,
      "learning_rate": 8.141676086873572e-06,
      "loss": 1.1236,
      "step": 165
    },
    {
      "epoch": 0.06897268100135037,
      "grad_norm": 0.8487104773521423,
      "learning_rate": 7.695237378953223e-06,
      "loss": 1.1555,
      "step": 166
    },
    {
      "epoch": 0.06938817907967175,
      "grad_norm": 0.8442949652671814,
      "learning_rate": 7.260364370723044e-06,
      "loss": 1.2705,
      "step": 167
    },
    {
      "epoch": 0.06980367715799314,
      "grad_norm": 0.825699508190155,
      "learning_rate": 6.837175952121306e-06,
      "loss": 1.3205,
      "step": 168
    },
    {
      "epoch": 0.07021917523631453,
      "grad_norm": 0.7826573252677917,
      "learning_rate": 6.425787818636131e-06,
      "loss": 1.109,
      "step": 169
    },
    {
      "epoch": 0.07063467331463592,
      "grad_norm": 0.8004805445671082,
      "learning_rate": 6.026312439675552e-06,
      "loss": 1.1775,
      "step": 170
    },
    {
      "epoch": 0.0710501713929573,
      "grad_norm": 0.8863964676856995,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 1.2052,
      "step": 171
    },
    {
      "epoch": 0.07146566947127869,
      "grad_norm": 0.7715016007423401,
      "learning_rate": 5.263533508961827e-06,
      "loss": 1.1256,
      "step": 172
    },
    {
      "epoch": 0.07188116754960008,
      "grad_norm": 0.9065338373184204,
      "learning_rate": 4.900438493352055e-06,
      "loss": 1.2882,
      "step": 173
    },
    {
      "epoch": 0.07229666562792147,
      "grad_norm": 0.9415071606636047,
      "learning_rate": 4.549673247541875e-06,
      "loss": 1.2399,
      "step": 174
    },
    {
      "epoch": 0.07271216370624285,
      "grad_norm": 0.9356192946434021,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 1.203,
      "step": 175
    },
    {
      "epoch": 0.07312766178456424,
      "grad_norm": 0.8843199610710144,
      "learning_rate": 3.885512251130763e-06,
      "loss": 1.2653,
      "step": 176
    },
    {
      "epoch": 0.07354315986288563,
      "grad_norm": 0.9166342616081238,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 1.1077,
      "step": 177
    },
    {
      "epoch": 0.07395865794120703,
      "grad_norm": 0.868156373500824,
      "learning_rate": 3.271776770026963e-06,
      "loss": 1.1893,
      "step": 178
    },
    {
      "epoch": 0.0743741560195284,
      "grad_norm": 0.9004489183425903,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 1.1978,
      "step": 179
    },
    {
      "epoch": 0.0747896540978498,
      "grad_norm": 0.8027791976928711,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 1.0252,
      "step": 180
    },
    {
      "epoch": 0.07520515217617119,
      "grad_norm": 0.9362911581993103,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 1.095,
      "step": 181
    },
    {
      "epoch": 0.07562065025449258,
      "grad_norm": 1.1003559827804565,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 1.1901,
      "step": 182
    },
    {
      "epoch": 0.07603614833281395,
      "grad_norm": 0.9305182099342346,
      "learning_rate": 1.962316193157593e-06,
      "loss": 1.12,
      "step": 183
    },
    {
      "epoch": 0.07645164641113535,
      "grad_norm": 1.020524501800537,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 1.3251,
      "step": 184
    },
    {
      "epoch": 0.07686714448945674,
      "grad_norm": 1.1219377517700195,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 1.077,
      "step": 185
    },
    {
      "epoch": 0.07728264256777813,
      "grad_norm": 1.0718369483947754,
      "learning_rate": 1.333670137599713e-06,
      "loss": 1.2123,
      "step": 186
    },
    {
      "epoch": 0.0776981406460995,
      "grad_norm": 1.1363760232925415,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 1.2269,
      "step": 187
    },
    {
      "epoch": 0.0781136387244209,
      "grad_norm": 1.120803713798523,
      "learning_rate": 9.810017062595322e-07,
      "loss": 1.1044,
      "step": 188
    },
    {
      "epoch": 0.07852913680274229,
      "grad_norm": 1.055897831916809,
      "learning_rate": 8.247462563808817e-07,
      "loss": 1.2073,
      "step": 189
    },
    {
      "epoch": 0.07894463488106368,
      "grad_norm": 1.136576533317566,
      "learning_rate": 6.819348298638839e-07,
      "loss": 1.0432,
      "step": 190
    },
    {
      "epoch": 0.07936013295938506,
      "grad_norm": 1.104294776916504,
      "learning_rate": 5.526064699265753e-07,
      "loss": 1.1118,
      "step": 191
    },
    {
      "epoch": 0.07977563103770645,
      "grad_norm": 0.9812235832214355,
      "learning_rate": 4.367965336512403e-07,
      "loss": 0.9029,
      "step": 192
    },
    {
      "epoch": 0.08019112911602784,
      "grad_norm": 1.1375207901000977,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 1.1935,
      "step": 193
    },
    {
      "epoch": 0.08060662719434923,
      "grad_norm": 1.0770201683044434,
      "learning_rate": 2.458548727494292e-07,
      "loss": 0.9794,
      "step": 194
    },
    {
      "epoch": 0.08102212527267061,
      "grad_norm": 0.9887418746948242,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 0.895,
      "step": 195
    },
    {
      "epoch": 0.081437623350992,
      "grad_norm": 1.0478863716125488,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 0.9464,
      "step": 196
    },
    {
      "epoch": 0.08185312142931339,
      "grad_norm": 1.0662038326263428,
      "learning_rate": 6.150154258476315e-08,
      "loss": 0.9947,
      "step": 197
    },
    {
      "epoch": 0.08226861950763478,
      "grad_norm": 1.2460801601409912,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 1.0595,
      "step": 198
    },
    {
      "epoch": 0.08268411758595616,
      "grad_norm": 1.1214287281036377,
      "learning_rate": 6.834750376549792e-09,
      "loss": 0.8416,
      "step": 199
    },
    {
      "epoch": 0.08309961566427755,
      "grad_norm": 1.6520289182662964,
      "learning_rate": 0.0,
      "loss": 0.95,
      "step": 200
    },
    {
      "epoch": 0.08309961566427755,
      "eval_loss": 1.1696622371673584,
      "eval_runtime": 182.5066,
      "eval_samples_per_second": 22.213,
      "eval_steps_per_second": 5.556,
      "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": 8.3184381232939e+16,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}