{
  "best_metric": 1.7642698287963867,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.15735641227380015,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0007867820613690008,
      "grad_norm": 1.484948754310608,
      "learning_rate": 1e-05,
      "loss": 3.1338,
      "step": 1
    },
    {
      "epoch": 0.0007867820613690008,
      "eval_loss": 3.409891128540039,
      "eval_runtime": 14.1845,
      "eval_samples_per_second": 150.939,
      "eval_steps_per_second": 37.788,
      "step": 1
    },
    {
      "epoch": 0.0015735641227380016,
      "grad_norm": 1.816478967666626,
      "learning_rate": 2e-05,
      "loss": 3.2501,
      "step": 2
    },
    {
      "epoch": 0.0023603461841070024,
      "grad_norm": 1.831554651260376,
      "learning_rate": 3e-05,
      "loss": 3.0871,
      "step": 3
    },
    {
      "epoch": 0.003147128245476003,
      "grad_norm": 1.860951542854309,
      "learning_rate": 4e-05,
      "loss": 3.043,
      "step": 4
    },
    {
      "epoch": 0.003933910306845004,
      "grad_norm": 1.626622200012207,
      "learning_rate": 5e-05,
      "loss": 3.1622,
      "step": 5
    },
    {
      "epoch": 0.004720692368214005,
      "grad_norm": 1.5488812923431396,
      "learning_rate": 6e-05,
      "loss": 3.1074,
      "step": 6
    },
    {
      "epoch": 0.0055074744295830055,
      "grad_norm": 1.3132474422454834,
      "learning_rate": 7e-05,
      "loss": 2.9825,
      "step": 7
    },
    {
      "epoch": 0.006294256490952006,
      "grad_norm": 1.2267699241638184,
      "learning_rate": 8e-05,
      "loss": 2.9639,
      "step": 8
    },
    {
      "epoch": 0.007081038552321007,
      "grad_norm": 1.1949219703674316,
      "learning_rate": 9e-05,
      "loss": 2.931,
      "step": 9
    },
    {
      "epoch": 0.007867820613690008,
      "grad_norm": 1.1969985961914062,
      "learning_rate": 0.0001,
      "loss": 2.9188,
      "step": 10
    },
    {
      "epoch": 0.00865460267505901,
      "grad_norm": 1.1242773532867432,
      "learning_rate": 9.999316524962345e-05,
      "loss": 2.8238,
      "step": 11
    },
    {
      "epoch": 0.00944138473642801,
      "grad_norm": 1.1578017473220825,
      "learning_rate": 9.997266286704631e-05,
      "loss": 2.7968,
      "step": 12
    },
    {
      "epoch": 0.010228166797797011,
      "grad_norm": 1.1513009071350098,
      "learning_rate": 9.993849845741524e-05,
      "loss": 2.699,
      "step": 13
    },
    {
      "epoch": 0.011014948859166011,
      "grad_norm": 1.1852365732192993,
      "learning_rate": 9.989068136093873e-05,
      "loss": 2.7504,
      "step": 14
    },
    {
      "epoch": 0.011801730920535013,
      "grad_norm": 1.2107021808624268,
      "learning_rate": 9.98292246503335e-05,
      "loss": 2.7013,
      "step": 15
    },
    {
      "epoch": 0.012588512981904013,
      "grad_norm": 1.2153959274291992,
      "learning_rate": 9.975414512725057e-05,
      "loss": 2.5866,
      "step": 16
    },
    {
      "epoch": 0.013375295043273014,
      "grad_norm": 1.2652902603149414,
      "learning_rate": 9.966546331768191e-05,
      "loss": 2.5578,
      "step": 17
    },
    {
      "epoch": 0.014162077104642014,
      "grad_norm": 1.3643229007720947,
      "learning_rate": 9.956320346634876e-05,
      "loss": 2.5252,
      "step": 18
    },
    {
      "epoch": 0.014948859166011016,
      "grad_norm": 1.2461212873458862,
      "learning_rate": 9.944739353007344e-05,
      "loss": 2.3801,
      "step": 19
    },
    {
      "epoch": 0.015735641227380016,
      "grad_norm": 1.3287397623062134,
      "learning_rate": 9.931806517013612e-05,
      "loss": 2.3786,
      "step": 20
    },
    {
      "epoch": 0.016522423288749016,
      "grad_norm": 1.4497843980789185,
      "learning_rate": 9.917525374361912e-05,
      "loss": 2.3213,
      "step": 21
    },
    {
      "epoch": 0.01730920535011802,
      "grad_norm": 1.351597547531128,
      "learning_rate": 9.901899829374047e-05,
      "loss": 2.3996,
      "step": 22
    },
    {
      "epoch": 0.01809598741148702,
      "grad_norm": 1.1487007141113281,
      "learning_rate": 9.884934153917997e-05,
      "loss": 2.2003,
      "step": 23
    },
    {
      "epoch": 0.01888276947285602,
      "grad_norm": 1.1297303438186646,
      "learning_rate": 9.86663298624003e-05,
      "loss": 2.2029,
      "step": 24
    },
    {
      "epoch": 0.01966955153422502,
      "grad_norm": 1.223523736000061,
      "learning_rate": 9.847001329696653e-05,
      "loss": 2.2771,
      "step": 25
    },
    {
      "epoch": 0.020456333595594022,
      "grad_norm": 1.0912593603134155,
      "learning_rate": 9.826044551386744e-05,
      "loss": 2.1437,
      "step": 26
    },
    {
      "epoch": 0.021243115656963022,
      "grad_norm": 1.0280300378799438,
      "learning_rate": 9.803768380684242e-05,
      "loss": 2.1182,
      "step": 27
    },
    {
      "epoch": 0.022029897718332022,
      "grad_norm": 1.1129682064056396,
      "learning_rate": 9.780178907671789e-05,
      "loss": 2.2856,
      "step": 28
    },
    {
      "epoch": 0.022816679779701022,
      "grad_norm": 1.0809565782546997,
      "learning_rate": 9.755282581475769e-05,
      "loss": 2.174,
      "step": 29
    },
    {
      "epoch": 0.023603461841070025,
      "grad_norm": 1.078463077545166,
      "learning_rate": 9.729086208503174e-05,
      "loss": 2.0672,
      "step": 30
    },
    {
      "epoch": 0.024390243902439025,
      "grad_norm": 1.1323771476745605,
      "learning_rate": 9.701596950580806e-05,
      "loss": 2.1847,
      "step": 31
    },
    {
      "epoch": 0.025177025963808025,
      "grad_norm": 1.0715173482894897,
      "learning_rate": 9.672822322997305e-05,
      "loss": 2.0494,
      "step": 32
    },
    {
      "epoch": 0.025963808025177025,
      "grad_norm": 1.095207691192627,
      "learning_rate": 9.642770192448536e-05,
      "loss": 2.0178,
      "step": 33
    },
    {
      "epoch": 0.02675059008654603,
      "grad_norm": 1.0888087749481201,
      "learning_rate": 9.611448774886924e-05,
      "loss": 2.1623,
      "step": 34
    },
    {
      "epoch": 0.02753737214791503,
      "grad_norm": 1.0621700286865234,
      "learning_rate": 9.578866633275288e-05,
      "loss": 2.0242,
      "step": 35
    },
    {
      "epoch": 0.02832415420928403,
      "grad_norm": 1.0141266584396362,
      "learning_rate": 9.545032675245813e-05,
      "loss": 1.9586,
      "step": 36
    },
    {
      "epoch": 0.029110936270653028,
      "grad_norm": 1.0606440305709839,
      "learning_rate": 9.509956150664796e-05,
      "loss": 1.9028,
      "step": 37
    },
    {
      "epoch": 0.02989771833202203,
      "grad_norm": 1.1259828805923462,
      "learning_rate": 9.473646649103818e-05,
      "loss": 2.0699,
      "step": 38
    },
    {
      "epoch": 0.03068450039339103,
      "grad_norm": 1.1965646743774414,
      "learning_rate": 9.43611409721806e-05,
      "loss": 2.0862,
      "step": 39
    },
    {
      "epoch": 0.03147128245476003,
      "grad_norm": 1.344648003578186,
      "learning_rate": 9.397368756032445e-05,
      "loss": 2.0498,
      "step": 40
    },
    {
      "epoch": 0.03225806451612903,
      "grad_norm": 1.100591778755188,
      "learning_rate": 9.357421218136386e-05,
      "loss": 1.9479,
      "step": 41
    },
    {
      "epoch": 0.03304484657749803,
      "grad_norm": 1.0336469411849976,
      "learning_rate": 9.316282404787871e-05,
      "loss": 1.9399,
      "step": 42
    },
    {
      "epoch": 0.03383162863886703,
      "grad_norm": 1.0810625553131104,
      "learning_rate": 9.273963562927695e-05,
      "loss": 1.949,
      "step": 43
    },
    {
      "epoch": 0.03461841070023604,
      "grad_norm": 1.1953938007354736,
      "learning_rate": 9.230476262104677e-05,
      "loss": 1.9253,
      "step": 44
    },
    {
      "epoch": 0.03540519276160504,
      "grad_norm": 1.1101166009902954,
      "learning_rate": 9.185832391312644e-05,
      "loss": 1.8107,
      "step": 45
    },
    {
      "epoch": 0.03619197482297404,
      "grad_norm": 1.1856448650360107,
      "learning_rate": 9.140044155740101e-05,
      "loss": 1.9452,
      "step": 46
    },
    {
      "epoch": 0.03697875688434304,
      "grad_norm": 1.1950342655181885,
      "learning_rate": 9.093124073433463e-05,
      "loss": 1.8977,
      "step": 47
    },
    {
      "epoch": 0.03776553894571204,
      "grad_norm": 1.3210840225219727,
      "learning_rate": 9.045084971874738e-05,
      "loss": 1.9713,
      "step": 48
    },
    {
      "epoch": 0.03855232100708104,
      "grad_norm": 1.3238219022750854,
      "learning_rate": 8.995939984474624e-05,
      "loss": 1.9458,
      "step": 49
    },
    {
      "epoch": 0.03933910306845004,
      "grad_norm": 1.7373554706573486,
      "learning_rate": 8.945702546981969e-05,
      "loss": 2.0043,
      "step": 50
    },
    {
      "epoch": 0.03933910306845004,
      "eval_loss": 1.9668817520141602,
      "eval_runtime": 14.1776,
      "eval_samples_per_second": 151.013,
      "eval_steps_per_second": 37.806,
      "step": 50
    },
    {
      "epoch": 0.04012588512981904,
      "grad_norm": 1.6901990175247192,
      "learning_rate": 8.894386393810563e-05,
      "loss": 2.4431,
      "step": 51
    },
    {
      "epoch": 0.040912667191188044,
      "grad_norm": 1.2382429838180542,
      "learning_rate": 8.842005554284296e-05,
      "loss": 2.138,
      "step": 52
    },
    {
      "epoch": 0.041699449252557044,
      "grad_norm": 1.6277223825454712,
      "learning_rate": 8.788574348801675e-05,
      "loss": 1.9867,
      "step": 53
    },
    {
      "epoch": 0.042486231313926044,
      "grad_norm": 1.155552625656128,
      "learning_rate": 8.73410738492077e-05,
      "loss": 2.0581,
      "step": 54
    },
    {
      "epoch": 0.043273013375295044,
      "grad_norm": 1.0387541055679321,
      "learning_rate": 8.678619553365659e-05,
      "loss": 2.1416,
      "step": 55
    },
    {
      "epoch": 0.044059795436664044,
      "grad_norm": 1.081500768661499,
      "learning_rate": 8.622126023955446e-05,
      "loss": 2.1029,
      "step": 56
    },
    {
      "epoch": 0.044846577498033044,
      "grad_norm": 0.9130778908729553,
      "learning_rate": 8.564642241456986e-05,
      "loss": 2.022,
      "step": 57
    },
    {
      "epoch": 0.045633359559402044,
      "grad_norm": 0.8940057754516602,
      "learning_rate": 8.506183921362443e-05,
      "loss": 2.0001,
      "step": 58
    },
    {
      "epoch": 0.046420141620771044,
      "grad_norm": 0.8492464423179626,
      "learning_rate": 8.44676704559283e-05,
      "loss": 2.0,
      "step": 59
    },
    {
      "epoch": 0.04720692368214005,
      "grad_norm": 1.0032528638839722,
      "learning_rate": 8.386407858128706e-05,
      "loss": 2.0364,
      "step": 60
    },
    {
      "epoch": 0.04799370574350905,
      "grad_norm": 1.0285426378250122,
      "learning_rate": 8.32512286056924e-05,
      "loss": 2.0403,
      "step": 61
    },
    {
      "epoch": 0.04878048780487805,
      "grad_norm": 0.94615238904953,
      "learning_rate": 8.262928807620843e-05,
      "loss": 1.9984,
      "step": 62
    },
    {
      "epoch": 0.04956726986624705,
      "grad_norm": 0.980772852897644,
      "learning_rate": 8.199842702516583e-05,
      "loss": 2.0517,
      "step": 63
    },
    {
      "epoch": 0.05035405192761605,
      "grad_norm": 0.945112407207489,
      "learning_rate": 8.135881792367686e-05,
      "loss": 1.956,
      "step": 64
    },
    {
      "epoch": 0.05114083398898505,
      "grad_norm": 0.9835402965545654,
      "learning_rate": 8.07106356344834e-05,
      "loss": 1.9872,
      "step": 65
    },
    {
      "epoch": 0.05192761605035405,
      "grad_norm": 0.9928127527236938,
      "learning_rate": 8.005405736415126e-05,
      "loss": 2.0058,
      "step": 66
    },
    {
      "epoch": 0.05271439811172305,
      "grad_norm": 0.9103164672851562,
      "learning_rate": 7.938926261462366e-05,
      "loss": 1.8624,
      "step": 67
    },
    {
      "epoch": 0.05350118017309206,
      "grad_norm": 0.9369115233421326,
      "learning_rate": 7.871643313414718e-05,
      "loss": 1.8945,
      "step": 68
    },
    {
      "epoch": 0.05428796223446106,
      "grad_norm": 0.9831304550170898,
      "learning_rate": 7.803575286758364e-05,
      "loss": 1.9843,
      "step": 69
    },
    {
      "epoch": 0.05507474429583006,
      "grad_norm": 0.9860329031944275,
      "learning_rate": 7.734740790612136e-05,
      "loss": 1.9,
      "step": 70
    },
    {
      "epoch": 0.05586152635719906,
      "grad_norm": 0.9668272733688354,
      "learning_rate": 7.66515864363997e-05,
      "loss": 1.8532,
      "step": 71
    },
    {
      "epoch": 0.05664830841856806,
      "grad_norm": 0.9727192521095276,
      "learning_rate": 7.594847868906076e-05,
      "loss": 1.8005,
      "step": 72
    },
    {
      "epoch": 0.057435090479937057,
      "grad_norm": 1.0073771476745605,
      "learning_rate": 7.52382768867422e-05,
      "loss": 1.893,
      "step": 73
    },
    {
      "epoch": 0.058221872541306056,
      "grad_norm": 1.1149795055389404,
      "learning_rate": 7.452117519152542e-05,
      "loss": 1.836,
      "step": 74
    },
    {
      "epoch": 0.059008654602675056,
      "grad_norm": 0.9467926621437073,
      "learning_rate": 7.379736965185368e-05,
      "loss": 1.8474,
      "step": 75
    },
    {
      "epoch": 0.05979543666404406,
      "grad_norm": 1.0009890794754028,
      "learning_rate": 7.30670581489344e-05,
      "loss": 1.8053,
      "step": 76
    },
    {
      "epoch": 0.06058221872541306,
      "grad_norm": 1.027714490890503,
      "learning_rate": 7.233044034264034e-05,
      "loss": 1.7908,
      "step": 77
    },
    {
      "epoch": 0.06136900078678206,
      "grad_norm": 1.0123565196990967,
      "learning_rate": 7.158771761692464e-05,
      "loss": 1.7715,
      "step": 78
    },
    {
      "epoch": 0.06215578284815106,
      "grad_norm": 1.0732414722442627,
      "learning_rate": 7.083909302476453e-05,
      "loss": 1.9212,
      "step": 79
    },
    {
      "epoch": 0.06294256490952006,
      "grad_norm": 1.0441950559616089,
      "learning_rate": 7.008477123264848e-05,
      "loss": 1.8414,
      "step": 80
    },
    {
      "epoch": 0.06372934697088907,
      "grad_norm": 1.068368673324585,
      "learning_rate": 6.932495846462261e-05,
      "loss": 1.884,
      "step": 81
    },
    {
      "epoch": 0.06451612903225806,
      "grad_norm": 1.0471543073654175,
      "learning_rate": 6.855986244591104e-05,
      "loss": 1.8253,
      "step": 82
    },
    {
      "epoch": 0.06530291109362707,
      "grad_norm": 1.0591351985931396,
      "learning_rate": 6.778969234612584e-05,
      "loss": 1.6889,
      "step": 83
    },
    {
      "epoch": 0.06608969315499606,
      "grad_norm": 1.068275809288025,
      "learning_rate": 6.701465872208216e-05,
      "loss": 1.6837,
      "step": 84
    },
    {
      "epoch": 0.06687647521636507,
      "grad_norm": 1.069745659828186,
      "learning_rate": 6.623497346023418e-05,
      "loss": 1.8113,
      "step": 85
    },
    {
      "epoch": 0.06766325727773406,
      "grad_norm": 1.1849879026412964,
      "learning_rate": 6.545084971874738e-05,
      "loss": 1.7003,
      "step": 86
    },
    {
      "epoch": 0.06845003933910307,
      "grad_norm": 1.1508926153182983,
      "learning_rate": 6.466250186922325e-05,
      "loss": 1.7825,
      "step": 87
    },
    {
      "epoch": 0.06923682140047208,
      "grad_norm": 1.1595426797866821,
      "learning_rate": 6.387014543809223e-05,
      "loss": 1.7222,
      "step": 88
    },
    {
      "epoch": 0.07002360346184107,
      "grad_norm": 1.1397861242294312,
      "learning_rate": 6.307399704769099e-05,
      "loss": 1.7975,
      "step": 89
    },
    {
      "epoch": 0.07081038552321008,
      "grad_norm": 1.1023906469345093,
      "learning_rate": 6.227427435703997e-05,
      "loss": 1.7034,
      "step": 90
    },
    {
      "epoch": 0.07159716758457907,
      "grad_norm": 1.11385178565979,
      "learning_rate": 6.147119600233758e-05,
      "loss": 1.6429,
      "step": 91
    },
    {
      "epoch": 0.07238394964594808,
      "grad_norm": 1.2370423078536987,
      "learning_rate": 6.066498153718735e-05,
      "loss": 1.7271,
      "step": 92
    },
    {
      "epoch": 0.07317073170731707,
      "grad_norm": 1.241693139076233,
      "learning_rate": 5.985585137257401e-05,
      "loss": 1.8362,
      "step": 93
    },
    {
      "epoch": 0.07395751376868608,
      "grad_norm": 1.1741639375686646,
      "learning_rate": 5.90440267166055e-05,
      "loss": 1.7965,
      "step": 94
    },
    {
      "epoch": 0.07474429583005507,
      "grad_norm": 1.2249914407730103,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 1.7735,
      "step": 95
    },
    {
      "epoch": 0.07553107789142408,
      "grad_norm": 1.2041676044464111,
      "learning_rate": 5.74131823855921e-05,
      "loss": 1.7219,
      "step": 96
    },
    {
      "epoch": 0.07631785995279308,
      "grad_norm": 1.1959495544433594,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 1.6604,
      "step": 97
    },
    {
      "epoch": 0.07710464201416208,
      "grad_norm": 1.332154393196106,
      "learning_rate": 5.577423184847932e-05,
      "loss": 1.8861,
      "step": 98
    },
    {
      "epoch": 0.07789142407553108,
      "grad_norm": 1.3559482097625732,
      "learning_rate": 5.495227651252315e-05,
      "loss": 1.6665,
      "step": 99
    },
    {
      "epoch": 0.07867820613690008,
      "grad_norm": 1.6197733879089355,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 1.6371,
      "step": 100
    },
    {
      "epoch": 0.07867820613690008,
      "eval_loss": 1.8378092050552368,
      "eval_runtime": 14.1432,
      "eval_samples_per_second": 151.38,
      "eval_steps_per_second": 37.898,
      "step": 100
    },
    {
      "epoch": 0.07946498819826908,
      "grad_norm": 1.5419577360153198,
      "learning_rate": 5.330452921628497e-05,
      "loss": 2.3035,
      "step": 101
    },
    {
      "epoch": 0.08025177025963807,
      "grad_norm": 1.2639943361282349,
      "learning_rate": 5.247918773366112e-05,
      "loss": 1.8866,
      "step": 102
    },
    {
      "epoch": 0.08103855232100708,
      "grad_norm": 1.1663724184036255,
      "learning_rate": 5.165316846586541e-05,
      "loss": 1.9968,
      "step": 103
    },
    {
      "epoch": 0.08182533438237609,
      "grad_norm": 1.1718730926513672,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 1.9829,
      "step": 104
    },
    {
      "epoch": 0.08261211644374508,
      "grad_norm": 1.1857966184616089,
      "learning_rate": 5e-05,
      "loss": 2.0207,
      "step": 105
    },
    {
      "epoch": 0.08339889850511409,
      "grad_norm": 1.079548954963684,
      "learning_rate": 4.917330276168208e-05,
      "loss": 2.0047,
      "step": 106
    },
    {
      "epoch": 0.08418568056648308,
      "grad_norm": 0.9886470437049866,
      "learning_rate": 4.834683153413459e-05,
      "loss": 1.9032,
      "step": 107
    },
    {
      "epoch": 0.08497246262785209,
      "grad_norm": 1.1596629619598389,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 2.0156,
      "step": 108
    },
    {
      "epoch": 0.08575924468922108,
      "grad_norm": 1.1242882013320923,
      "learning_rate": 4.669547078371504e-05,
      "loss": 1.8169,
      "step": 109
    },
    {
      "epoch": 0.08654602675059009,
      "grad_norm": 1.0429154634475708,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 1.9145,
      "step": 110
    },
    {
      "epoch": 0.08733280881195908,
      "grad_norm": 1.0206472873687744,
      "learning_rate": 4.504772348747687e-05,
      "loss": 1.7745,
      "step": 111
    },
    {
      "epoch": 0.08811959087332809,
      "grad_norm": 0.932558000087738,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 1.77,
      "step": 112
    },
    {
      "epoch": 0.0889063729346971,
      "grad_norm": 0.9783939719200134,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 1.8818,
      "step": 113
    },
    {
      "epoch": 0.08969315499606609,
      "grad_norm": 0.9316754341125488,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 1.6587,
      "step": 114
    },
    {
      "epoch": 0.0904799370574351,
      "grad_norm": 0.9783585071563721,
      "learning_rate": 4.17702704859633e-05,
      "loss": 1.7887,
      "step": 115
    },
    {
      "epoch": 0.09126671911880409,
      "grad_norm": 1.0170371532440186,
      "learning_rate": 4.095597328339452e-05,
      "loss": 1.6832,
      "step": 116
    },
    {
      "epoch": 0.0920535011801731,
      "grad_norm": 1.0139265060424805,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 1.8291,
      "step": 117
    },
    {
      "epoch": 0.09284028324154209,
      "grad_norm": 1.0537375211715698,
      "learning_rate": 3.933501846281267e-05,
      "loss": 1.6172,
      "step": 118
    },
    {
      "epoch": 0.0936270653029111,
      "grad_norm": 1.086940050125122,
      "learning_rate": 3.852880399766243e-05,
      "loss": 1.8984,
      "step": 119
    },
    {
      "epoch": 0.0944138473642801,
      "grad_norm": 1.1102848052978516,
      "learning_rate": 3.772572564296005e-05,
      "loss": 1.7912,
      "step": 120
    },
    {
      "epoch": 0.0952006294256491,
      "grad_norm": 1.1705005168914795,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 1.9006,
      "step": 121
    },
    {
      "epoch": 0.0959874114870181,
      "grad_norm": 1.005077838897705,
      "learning_rate": 3.612985456190778e-05,
      "loss": 1.6514,
      "step": 122
    },
    {
      "epoch": 0.0967741935483871,
      "grad_norm": 1.0830457210540771,
      "learning_rate": 3.533749813077677e-05,
      "loss": 1.6971,
      "step": 123
    },
    {
      "epoch": 0.0975609756097561,
      "grad_norm": 1.0748510360717773,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 1.7358,
      "step": 124
    },
    {
      "epoch": 0.0983477576711251,
      "grad_norm": 1.0849908590316772,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 1.7814,
      "step": 125
    },
    {
      "epoch": 0.0991345397324941,
      "grad_norm": 1.044701337814331,
      "learning_rate": 3.298534127791785e-05,
      "loss": 1.6957,
      "step": 126
    },
    {
      "epoch": 0.0999213217938631,
      "grad_norm": 1.065104603767395,
      "learning_rate": 3.221030765387417e-05,
      "loss": 1.7243,
      "step": 127
    },
    {
      "epoch": 0.1007081038552321,
      "grad_norm": 1.1763389110565186,
      "learning_rate": 3.144013755408895e-05,
      "loss": 1.7529,
      "step": 128
    },
    {
      "epoch": 0.10149488591660111,
      "grad_norm": 1.1484479904174805,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 1.8868,
      "step": 129
    },
    {
      "epoch": 0.1022816679779701,
      "grad_norm": 1.158713459968567,
      "learning_rate": 2.991522876735154e-05,
      "loss": 1.8405,
      "step": 130
    },
    {
      "epoch": 0.10306845003933911,
      "grad_norm": 1.2112756967544556,
      "learning_rate": 2.916090697523549e-05,
      "loss": 1.807,
      "step": 131
    },
    {
      "epoch": 0.1038552321007081,
      "grad_norm": 1.074357271194458,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 1.7989,
      "step": 132
    },
    {
      "epoch": 0.10464201416207711,
      "grad_norm": 1.1528265476226807,
      "learning_rate": 2.766955965735968e-05,
      "loss": 1.7523,
      "step": 133
    },
    {
      "epoch": 0.1054287962234461,
      "grad_norm": 1.1323232650756836,
      "learning_rate": 2.693294185106562e-05,
      "loss": 1.8373,
      "step": 134
    },
    {
      "epoch": 0.10621557828481511,
      "grad_norm": 1.2037923336029053,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 1.89,
      "step": 135
    },
    {
      "epoch": 0.10700236034618411,
      "grad_norm": 1.126426339149475,
      "learning_rate": 2.547882480847461e-05,
      "loss": 1.6399,
      "step": 136
    },
    {
      "epoch": 0.1077891424075531,
      "grad_norm": 1.1715989112854004,
      "learning_rate": 2.476172311325783e-05,
      "loss": 1.7243,
      "step": 137
    },
    {
      "epoch": 0.10857592446892211,
      "grad_norm": 1.0809516906738281,
      "learning_rate": 2.405152131093926e-05,
      "loss": 1.5418,
      "step": 138
    },
    {
      "epoch": 0.1093627065302911,
      "grad_norm": 1.146828532218933,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 1.5185,
      "step": 139
    },
    {
      "epoch": 0.11014948859166011,
      "grad_norm": 1.207365870475769,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 1.7831,
      "step": 140
    },
    {
      "epoch": 0.1109362706530291,
      "grad_norm": 1.133414387702942,
      "learning_rate": 2.196424713241637e-05,
      "loss": 1.5354,
      "step": 141
    },
    {
      "epoch": 0.11172305271439811,
      "grad_norm": 1.1466996669769287,
      "learning_rate": 2.128356686585282e-05,
      "loss": 1.7438,
      "step": 142
    },
    {
      "epoch": 0.1125098347757671,
      "grad_norm": 1.1880643367767334,
      "learning_rate": 2.061073738537635e-05,
      "loss": 1.6592,
      "step": 143
    },
    {
      "epoch": 0.11329661683713611,
      "grad_norm": 1.1415419578552246,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 1.6603,
      "step": 144
    },
    {
      "epoch": 0.11408339889850512,
      "grad_norm": 1.2165260314941406,
      "learning_rate": 1.928936436551661e-05,
      "loss": 1.6192,
      "step": 145
    },
    {
      "epoch": 0.11487018095987411,
      "grad_norm": 1.3562653064727783,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 1.9221,
      "step": 146
    },
    {
      "epoch": 0.11565696302124312,
      "grad_norm": 1.3774889707565308,
      "learning_rate": 1.800157297483417e-05,
      "loss": 1.7281,
      "step": 147
    },
    {
      "epoch": 0.11644374508261211,
      "grad_norm": 1.3174934387207031,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 1.5902,
      "step": 148
    },
    {
      "epoch": 0.11723052714398112,
      "grad_norm": 1.510408878326416,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 1.7194,
      "step": 149
    },
    {
      "epoch": 0.11801730920535011,
      "grad_norm": 1.7525125741958618,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 1.8281,
      "step": 150
    },
    {
      "epoch": 0.11801730920535011,
      "eval_loss": 1.77768874168396,
      "eval_runtime": 14.0978,
      "eval_samples_per_second": 151.868,
      "eval_steps_per_second": 38.02,
      "step": 150
    },
    {
      "epoch": 0.11880409126671912,
      "grad_norm": 1.6723418235778809,
      "learning_rate": 1.553232954407171e-05,
      "loss": 2.0954,
      "step": 151
    },
    {
      "epoch": 0.11959087332808813,
      "grad_norm": 1.332829475402832,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 1.9825,
      "step": 152
    },
    {
      "epoch": 0.12037765538945712,
      "grad_norm": 1.2287746667861938,
      "learning_rate": 1.435357758543015e-05,
      "loss": 1.9873,
      "step": 153
    },
    {
      "epoch": 0.12116443745082613,
      "grad_norm": 1.2369835376739502,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 1.9062,
      "step": 154
    },
    {
      "epoch": 0.12195121951219512,
      "grad_norm": 1.2607827186584473,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 2.0369,
      "step": 155
    },
    {
      "epoch": 0.12273800157356413,
      "grad_norm": 1.1202712059020996,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 1.965,
      "step": 156
    },
    {
      "epoch": 0.12352478363493312,
      "grad_norm": 1.0930640697479248,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 1.892,
      "step": 157
    },
    {
      "epoch": 0.12431156569630213,
      "grad_norm": 1.118685007095337,
      "learning_rate": 1.157994445715706e-05,
      "loss": 1.8864,
      "step": 158
    },
    {
      "epoch": 0.12509834775767112,
      "grad_norm": 1.0079420804977417,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 1.6432,
      "step": 159
    },
    {
      "epoch": 0.12588512981904013,
      "grad_norm": 1.169901728630066,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 1.9217,
      "step": 160
    },
    {
      "epoch": 0.12667191188040913,
      "grad_norm": 1.0619724988937378,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 1.7365,
      "step": 161
    },
    {
      "epoch": 0.12745869394177814,
      "grad_norm": 1.0279138088226318,
      "learning_rate": 9.549150281252633e-06,
      "loss": 1.7888,
      "step": 162
    },
    {
      "epoch": 0.12824547600314712,
      "grad_norm": 1.0158978700637817,
      "learning_rate": 9.068759265665384e-06,
      "loss": 1.7639,
      "step": 163
    },
    {
      "epoch": 0.12903225806451613,
      "grad_norm": 1.003627061843872,
      "learning_rate": 8.599558442598998e-06,
      "loss": 1.7006,
      "step": 164
    },
    {
      "epoch": 0.12981904012588513,
      "grad_norm": 1.041370153427124,
      "learning_rate": 8.141676086873572e-06,
      "loss": 1.7518,
      "step": 165
    },
    {
      "epoch": 0.13060582218725414,
      "grad_norm": 1.0391087532043457,
      "learning_rate": 7.695237378953223e-06,
      "loss": 1.7202,
      "step": 166
    },
    {
      "epoch": 0.13139260424862312,
      "grad_norm": 1.0339860916137695,
      "learning_rate": 7.260364370723044e-06,
      "loss": 1.8103,
      "step": 167
    },
    {
      "epoch": 0.13217938630999213,
      "grad_norm": 0.9981183409690857,
      "learning_rate": 6.837175952121306e-06,
      "loss": 1.6142,
      "step": 168
    },
    {
      "epoch": 0.13296616837136113,
      "grad_norm": 1.0310614109039307,
      "learning_rate": 6.425787818636131e-06,
      "loss": 1.5021,
      "step": 169
    },
    {
      "epoch": 0.13375295043273014,
      "grad_norm": 1.0159376859664917,
      "learning_rate": 6.026312439675552e-06,
      "loss": 1.6968,
      "step": 170
    },
    {
      "epoch": 0.13453973249409915,
      "grad_norm": 1.0290675163269043,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 1.6823,
      "step": 171
    },
    {
      "epoch": 0.13532651455546812,
      "grad_norm": 1.0565516948699951,
      "learning_rate": 5.263533508961827e-06,
      "loss": 1.7353,
      "step": 172
    },
    {
      "epoch": 0.13611329661683713,
      "grad_norm": 1.0285171270370483,
      "learning_rate": 4.900438493352055e-06,
      "loss": 1.704,
      "step": 173
    },
    {
      "epoch": 0.13690007867820614,
      "grad_norm": 1.1128296852111816,
      "learning_rate": 4.549673247541875e-06,
      "loss": 1.7787,
      "step": 174
    },
    {
      "epoch": 0.13768686073957515,
      "grad_norm": 1.0697580575942993,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 1.7385,
      "step": 175
    },
    {
      "epoch": 0.13847364280094415,
      "grad_norm": 1.0604289770126343,
      "learning_rate": 3.885512251130763e-06,
      "loss": 1.6894,
      "step": 176
    },
    {
      "epoch": 0.13926042486231313,
      "grad_norm": 1.1184186935424805,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 1.7189,
      "step": 177
    },
    {
      "epoch": 0.14004720692368214,
      "grad_norm": 1.0466569662094116,
      "learning_rate": 3.271776770026963e-06,
      "loss": 1.6031,
      "step": 178
    },
    {
      "epoch": 0.14083398898505115,
      "grad_norm": 1.0846339464187622,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 1.6342,
      "step": 179
    },
    {
      "epoch": 0.14162077104642015,
      "grad_norm": 1.1711846590042114,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 1.6639,
      "step": 180
    },
    {
      "epoch": 0.14240755310778913,
      "grad_norm": 1.1818076372146606,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 1.757,
      "step": 181
    },
    {
      "epoch": 0.14319433516915814,
      "grad_norm": 1.1868349313735962,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 1.7516,
      "step": 182
    },
    {
      "epoch": 0.14398111723052714,
      "grad_norm": 1.1716861724853516,
      "learning_rate": 1.962316193157593e-06,
      "loss": 1.8166,
      "step": 183
    },
    {
      "epoch": 0.14476789929189615,
      "grad_norm": 1.2448678016662598,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 1.752,
      "step": 184
    },
    {
      "epoch": 0.14555468135326516,
      "grad_norm": 1.1485395431518555,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 1.7708,
      "step": 185
    },
    {
      "epoch": 0.14634146341463414,
      "grad_norm": 1.1534854173660278,
      "learning_rate": 1.333670137599713e-06,
      "loss": 1.6237,
      "step": 186
    },
    {
      "epoch": 0.14712824547600314,
      "grad_norm": 1.2661713361740112,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 1.6317,
      "step": 187
    },
    {
      "epoch": 0.14791502753737215,
      "grad_norm": 1.305614948272705,
      "learning_rate": 9.810017062595322e-07,
      "loss": 1.7099,
      "step": 188
    },
    {
      "epoch": 0.14870180959874116,
      "grad_norm": 1.269962191581726,
      "learning_rate": 8.247462563808817e-07,
      "loss": 1.6072,
      "step": 189
    },
    {
      "epoch": 0.14948859166011014,
      "grad_norm": 1.2583258152008057,
      "learning_rate": 6.819348298638839e-07,
      "loss": 1.696,
      "step": 190
    },
    {
      "epoch": 0.15027537372147914,
      "grad_norm": 1.334059238433838,
      "learning_rate": 5.526064699265753e-07,
      "loss": 1.6755,
      "step": 191
    },
    {
      "epoch": 0.15106215578284815,
      "grad_norm": 1.6304258108139038,
      "learning_rate": 4.367965336512403e-07,
      "loss": 1.7412,
      "step": 192
    },
    {
      "epoch": 0.15184893784421716,
      "grad_norm": 1.7743124961853027,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 1.6613,
      "step": 193
    },
    {
      "epoch": 0.15263571990558616,
      "grad_norm": 1.8280112743377686,
      "learning_rate": 2.458548727494292e-07,
      "loss": 1.6361,
      "step": 194
    },
    {
      "epoch": 0.15342250196695514,
      "grad_norm": 1.9423319101333618,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 1.771,
      "step": 195
    },
    {
      "epoch": 0.15420928402832415,
      "grad_norm": 2.4467086791992188,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 1.7676,
      "step": 196
    },
    {
      "epoch": 0.15499606608969316,
      "grad_norm": 2.7606940269470215,
      "learning_rate": 6.150154258476315e-08,
      "loss": 1.8707,
      "step": 197
    },
    {
      "epoch": 0.15578284815106216,
      "grad_norm": 2.959803342819214,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 1.7694,
      "step": 198
    },
    {
      "epoch": 0.15656963021243114,
      "grad_norm": 2.8867545127868652,
      "learning_rate": 6.834750376549792e-09,
      "loss": 1.9107,
      "step": 199
    },
    {
      "epoch": 0.15735641227380015,
      "grad_norm": 4.06135892868042,
      "learning_rate": 0.0,
      "loss": 1.6157,
      "step": 200
    },
    {
      "epoch": 0.15735641227380015,
      "eval_loss": 1.7642698287963867,
      "eval_runtime": 14.1062,
      "eval_samples_per_second": 151.777,
      "eval_steps_per_second": 37.998,
      "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": 5953579646976000.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}