{
  "best_metric": 0.8241426348686218,
  "best_model_checkpoint": "miner_id_24/checkpoint-450",
  "epoch": 0.005982033958012768,
  "eval_steps": 150,
  "global_step": 450,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.329340879558393e-05,
      "grad_norm": 1.9388657808303833,
      "learning_rate": 5e-06,
      "loss": 1.3911,
      "step": 1
    },
    {
      "epoch": 1.329340879558393e-05,
      "eval_loss": 1.8459720611572266,
      "eval_runtime": 3461.4381,
      "eval_samples_per_second": 36.602,
      "eval_steps_per_second": 9.151,
      "step": 1
    },
    {
      "epoch": 2.658681759116786e-05,
      "grad_norm": 2.8200104236602783,
      "learning_rate": 1e-05,
      "loss": 1.3589,
      "step": 2
    },
    {
      "epoch": 3.988022638675179e-05,
      "grad_norm": 2.7405104637145996,
      "learning_rate": 1.5e-05,
      "loss": 1.3274,
      "step": 3
    },
    {
      "epoch": 5.317363518233572e-05,
      "grad_norm": 2.6861093044281006,
      "learning_rate": 2e-05,
      "loss": 1.2136,
      "step": 4
    },
    {
      "epoch": 6.646704397791965e-05,
      "grad_norm": 2.4807560443878174,
      "learning_rate": 2.5e-05,
      "loss": 1.2634,
      "step": 5
    },
    {
      "epoch": 7.976045277350358e-05,
      "grad_norm": 3.396902322769165,
      "learning_rate": 3e-05,
      "loss": 1.1798,
      "step": 6
    },
    {
      "epoch": 9.305386156908751e-05,
      "grad_norm": 2.9391775131225586,
      "learning_rate": 3.5e-05,
      "loss": 1.2212,
      "step": 7
    },
    {
      "epoch": 0.00010634727036467144,
      "grad_norm": 2.992940902709961,
      "learning_rate": 4e-05,
      "loss": 1.1953,
      "step": 8
    },
    {
      "epoch": 0.00011964067916025536,
      "grad_norm": 3.2214736938476562,
      "learning_rate": 4.5e-05,
      "loss": 1.1323,
      "step": 9
    },
    {
      "epoch": 0.0001329340879558393,
      "grad_norm": 3.6011428833007812,
      "learning_rate": 5e-05,
      "loss": 1.1466,
      "step": 10
    },
    {
      "epoch": 0.00014622749675142322,
      "grad_norm": 3.9849705696105957,
      "learning_rate": 5.500000000000001e-05,
      "loss": 1.0944,
      "step": 11
    },
    {
      "epoch": 0.00015952090554700716,
      "grad_norm": 3.5957984924316406,
      "learning_rate": 6e-05,
      "loss": 1.171,
      "step": 12
    },
    {
      "epoch": 0.00017281431434259108,
      "grad_norm": 2.5957956314086914,
      "learning_rate": 6.500000000000001e-05,
      "loss": 1.074,
      "step": 13
    },
    {
      "epoch": 0.00018610772313817502,
      "grad_norm": 3.4556539058685303,
      "learning_rate": 7e-05,
      "loss": 1.0507,
      "step": 14
    },
    {
      "epoch": 0.00019940113193375893,
      "grad_norm": 3.1671648025512695,
      "learning_rate": 7.500000000000001e-05,
      "loss": 1.0411,
      "step": 15
    },
    {
      "epoch": 0.00021269454072934287,
      "grad_norm": 2.902188539505005,
      "learning_rate": 8e-05,
      "loss": 1.0133,
      "step": 16
    },
    {
      "epoch": 0.00022598794952492681,
      "grad_norm": 2.787954568862915,
      "learning_rate": 8.5e-05,
      "loss": 1.0352,
      "step": 17
    },
    {
      "epoch": 0.00023928135832051073,
      "grad_norm": 2.525041103363037,
      "learning_rate": 9e-05,
      "loss": 0.9911,
      "step": 18
    },
    {
      "epoch": 0.00025257476711609464,
      "grad_norm": 2.6477138996124268,
      "learning_rate": 9.5e-05,
      "loss": 0.9524,
      "step": 19
    },
    {
      "epoch": 0.0002658681759116786,
      "grad_norm": 2.82763409614563,
      "learning_rate": 0.0001,
      "loss": 1.0342,
      "step": 20
    },
    {
      "epoch": 0.0002791615847072625,
      "grad_norm": 3.421452760696411,
      "learning_rate": 9.999866555428618e-05,
      "loss": 0.9806,
      "step": 21
    },
    {
      "epoch": 0.00029245499350284644,
      "grad_norm": 2.983229398727417,
      "learning_rate": 9.999466228837451e-05,
      "loss": 0.8953,
      "step": 22
    },
    {
      "epoch": 0.00030574840229843035,
      "grad_norm": 3.4214861392974854,
      "learning_rate": 9.998799041595064e-05,
      "loss": 0.9724,
      "step": 23
    },
    {
      "epoch": 0.0003190418110940143,
      "grad_norm": 3.5156090259552,
      "learning_rate": 9.997865029314463e-05,
      "loss": 0.9613,
      "step": 24
    },
    {
      "epoch": 0.00033233521988959824,
      "grad_norm": 3.8509676456451416,
      "learning_rate": 9.996664241851197e-05,
      "loss": 0.913,
      "step": 25
    },
    {
      "epoch": 0.00034562862868518215,
      "grad_norm": 3.1751909255981445,
      "learning_rate": 9.995196743300692e-05,
      "loss": 1.1327,
      "step": 26
    },
    {
      "epoch": 0.0003589220374807661,
      "grad_norm": 3.8788204193115234,
      "learning_rate": 9.993462611994832e-05,
      "loss": 0.7996,
      "step": 27
    },
    {
      "epoch": 0.00037221544627635004,
      "grad_norm": 3.27899432182312,
      "learning_rate": 9.991461940497786e-05,
      "loss": 0.9915,
      "step": 28
    },
    {
      "epoch": 0.00038550885507193395,
      "grad_norm": 3.78037691116333,
      "learning_rate": 9.989194835601048e-05,
      "loss": 1.0367,
      "step": 29
    },
    {
      "epoch": 0.00039880226386751786,
      "grad_norm": 4.236947536468506,
      "learning_rate": 9.986661418317759e-05,
      "loss": 0.9108,
      "step": 30
    },
    {
      "epoch": 0.00041209567266310183,
      "grad_norm": 3.913956642150879,
      "learning_rate": 9.983861823876231e-05,
      "loss": 1.0729,
      "step": 31
    },
    {
      "epoch": 0.00042538908145868575,
      "grad_norm": 4.501612186431885,
      "learning_rate": 9.980796201712734e-05,
      "loss": 1.0856,
      "step": 32
    },
    {
      "epoch": 0.00043868249025426966,
      "grad_norm": 4.474135875701904,
      "learning_rate": 9.977464715463524e-05,
      "loss": 0.897,
      "step": 33
    },
    {
      "epoch": 0.00045197589904985363,
      "grad_norm": 5.405597686767578,
      "learning_rate": 9.973867542956104e-05,
      "loss": 1.1942,
      "step": 34
    },
    {
      "epoch": 0.00046526930784543754,
      "grad_norm": 3.8748316764831543,
      "learning_rate": 9.97000487619973e-05,
      "loss": 1.1122,
      "step": 35
    },
    {
      "epoch": 0.00047856271664102146,
      "grad_norm": 3.826747417449951,
      "learning_rate": 9.965876921375165e-05,
      "loss": 1.097,
      "step": 36
    },
    {
      "epoch": 0.0004918561254366054,
      "grad_norm": 3.8462743759155273,
      "learning_rate": 9.961483898823678e-05,
      "loss": 0.8577,
      "step": 37
    },
    {
      "epoch": 0.0005051495342321893,
      "grad_norm": 4.05244255065918,
      "learning_rate": 9.956826043035268e-05,
      "loss": 0.9462,
      "step": 38
    },
    {
      "epoch": 0.0005184429430277732,
      "grad_norm": 4.053645610809326,
      "learning_rate": 9.951903602636166e-05,
      "loss": 0.9335,
      "step": 39
    },
    {
      "epoch": 0.0005317363518233572,
      "grad_norm": 4.824429988861084,
      "learning_rate": 9.946716840375551e-05,
      "loss": 1.0454,
      "step": 40
    },
    {
      "epoch": 0.0005450297606189411,
      "grad_norm": 5.300111770629883,
      "learning_rate": 9.94126603311153e-05,
      "loss": 0.9345,
      "step": 41
    },
    {
      "epoch": 0.000558323169414525,
      "grad_norm": 4.619444847106934,
      "learning_rate": 9.935551471796358e-05,
      "loss": 1.0301,
      "step": 42
    },
    {
      "epoch": 0.000571616578210109,
      "grad_norm": 4.485474586486816,
      "learning_rate": 9.92957346146091e-05,
      "loss": 0.9127,
      "step": 43
    },
    {
      "epoch": 0.0005849099870056929,
      "grad_norm": 5.240790843963623,
      "learning_rate": 9.923332321198395e-05,
      "loss": 1.1571,
      "step": 44
    },
    {
      "epoch": 0.0005982033958012768,
      "grad_norm": 5.170898914337158,
      "learning_rate": 9.916828384147331e-05,
      "loss": 1.0133,
      "step": 45
    },
    {
      "epoch": 0.0006114968045968607,
      "grad_norm": 6.1428446769714355,
      "learning_rate": 9.910061997473752e-05,
      "loss": 1.0003,
      "step": 46
    },
    {
      "epoch": 0.0006247902133924447,
      "grad_norm": 5.351384162902832,
      "learning_rate": 9.903033522352687e-05,
      "loss": 0.999,
      "step": 47
    },
    {
      "epoch": 0.0006380836221880286,
      "grad_norm": 5.867308616638184,
      "learning_rate": 9.895743333948874e-05,
      "loss": 1.0887,
      "step": 48
    },
    {
      "epoch": 0.0006513770309836126,
      "grad_norm": 7.773362159729004,
      "learning_rate": 9.888191821396744e-05,
      "loss": 1.1315,
      "step": 49
    },
    {
      "epoch": 0.0006646704397791965,
      "grad_norm": 9.756021499633789,
      "learning_rate": 9.880379387779637e-05,
      "loss": 1.2055,
      "step": 50
    },
    {
      "epoch": 0.0006779638485747804,
      "grad_norm": 3.411668062210083,
      "learning_rate": 9.872306450108292e-05,
      "loss": 1.4152,
      "step": 51
    },
    {
      "epoch": 0.0006912572573703643,
      "grad_norm": 2.6563198566436768,
      "learning_rate": 9.863973439298597e-05,
      "loss": 1.1962,
      "step": 52
    },
    {
      "epoch": 0.0007045506661659482,
      "grad_norm": 2.5267629623413086,
      "learning_rate": 9.855380800148572e-05,
      "loss": 1.1744,
      "step": 53
    },
    {
      "epoch": 0.0007178440749615322,
      "grad_norm": 3.2284536361694336,
      "learning_rate": 9.846528991314639e-05,
      "loss": 0.9512,
      "step": 54
    },
    {
      "epoch": 0.0007311374837571162,
      "grad_norm": 2.0135741233825684,
      "learning_rate": 9.837418485287127e-05,
      "loss": 1.0179,
      "step": 55
    },
    {
      "epoch": 0.0007444308925527001,
      "grad_norm": 2.1273906230926514,
      "learning_rate": 9.828049768365068e-05,
      "loss": 0.9856,
      "step": 56
    },
    {
      "epoch": 0.000757724301348284,
      "grad_norm": 2.432741165161133,
      "learning_rate": 9.818423340630228e-05,
      "loss": 1.0059,
      "step": 57
    },
    {
      "epoch": 0.0007710177101438679,
      "grad_norm": 2.2927942276000977,
      "learning_rate": 9.808539715920414e-05,
      "loss": 1.2061,
      "step": 58
    },
    {
      "epoch": 0.0007843111189394518,
      "grad_norm": 2.312389850616455,
      "learning_rate": 9.798399421802056e-05,
      "loss": 0.8977,
      "step": 59
    },
    {
      "epoch": 0.0007976045277350357,
      "grad_norm": 2.4386355876922607,
      "learning_rate": 9.78800299954203e-05,
      "loss": 0.8272,
      "step": 60
    },
    {
      "epoch": 0.0008108979365306198,
      "grad_norm": 1.9699569940567017,
      "learning_rate": 9.777351004078783e-05,
      "loss": 1.087,
      "step": 61
    },
    {
      "epoch": 0.0008241913453262037,
      "grad_norm": 2.6284430027008057,
      "learning_rate": 9.766444003992703e-05,
      "loss": 1.1052,
      "step": 62
    },
    {
      "epoch": 0.0008374847541217876,
      "grad_norm": 2.237579822540283,
      "learning_rate": 9.755282581475769e-05,
      "loss": 0.9417,
      "step": 63
    },
    {
      "epoch": 0.0008507781629173715,
      "grad_norm": 2.1193315982818604,
      "learning_rate": 9.743867332300478e-05,
      "loss": 0.9962,
      "step": 64
    },
    {
      "epoch": 0.0008640715717129554,
      "grad_norm": 2.354325294494629,
      "learning_rate": 9.732198865788047e-05,
      "loss": 1.0707,
      "step": 65
    },
    {
      "epoch": 0.0008773649805085393,
      "grad_norm": 2.586552858352661,
      "learning_rate": 9.72027780477588e-05,
      "loss": 0.9973,
      "step": 66
    },
    {
      "epoch": 0.0008906583893041232,
      "grad_norm": 2.397660255432129,
      "learning_rate": 9.708104785584323e-05,
      "loss": 0.9432,
      "step": 67
    },
    {
      "epoch": 0.0009039517980997073,
      "grad_norm": 2.3507635593414307,
      "learning_rate": 9.695680457982713e-05,
      "loss": 1.0085,
      "step": 68
    },
    {
      "epoch": 0.0009172452068952912,
      "grad_norm": 2.0967140197753906,
      "learning_rate": 9.683005485154677e-05,
      "loss": 0.9675,
      "step": 69
    },
    {
      "epoch": 0.0009305386156908751,
      "grad_norm": 2.5464537143707275,
      "learning_rate": 9.67008054366274e-05,
      "loss": 0.871,
      "step": 70
    },
    {
      "epoch": 0.000943832024486459,
      "grad_norm": 2.32565975189209,
      "learning_rate": 9.656906323412217e-05,
      "loss": 0.9763,
      "step": 71
    },
    {
      "epoch": 0.0009571254332820429,
      "grad_norm": 2.7051308155059814,
      "learning_rate": 9.643483527614372e-05,
      "loss": 1.1479,
      "step": 72
    },
    {
      "epoch": 0.0009704188420776268,
      "grad_norm": 2.812052011489868,
      "learning_rate": 9.629812872748901e-05,
      "loss": 0.8656,
      "step": 73
    },
    {
      "epoch": 0.0009837122508732107,
      "grad_norm": 2.9575998783111572,
      "learning_rate": 9.615895088525677e-05,
      "loss": 0.9026,
      "step": 74
    },
    {
      "epoch": 0.0009970056596687947,
      "grad_norm": 2.4681830406188965,
      "learning_rate": 9.601730917845797e-05,
      "loss": 1.0127,
      "step": 75
    },
    {
      "epoch": 0.0010102990684643786,
      "grad_norm": 2.636371374130249,
      "learning_rate": 9.587321116761938e-05,
      "loss": 0.9537,
      "step": 76
    },
    {
      "epoch": 0.0010235924772599625,
      "grad_norm": 4.0426554679870605,
      "learning_rate": 9.57266645443799e-05,
      "loss": 0.9712,
      "step": 77
    },
    {
      "epoch": 0.0010368858860555464,
      "grad_norm": 2.898866653442383,
      "learning_rate": 9.557767713108009e-05,
      "loss": 0.8165,
      "step": 78
    },
    {
      "epoch": 0.0010501792948511305,
      "grad_norm": 3.4302549362182617,
      "learning_rate": 9.542625688034449e-05,
      "loss": 0.9472,
      "step": 79
    },
    {
      "epoch": 0.0010634727036467144,
      "grad_norm": 2.756509304046631,
      "learning_rate": 9.527241187465734e-05,
      "loss": 0.7932,
      "step": 80
    },
    {
      "epoch": 0.0010767661124422984,
      "grad_norm": 2.6443912982940674,
      "learning_rate": 9.511615032593096e-05,
      "loss": 0.8982,
      "step": 81
    },
    {
      "epoch": 0.0010900595212378823,
      "grad_norm": 2.9568185806274414,
      "learning_rate": 9.49574805750675e-05,
      "loss": 0.8393,
      "step": 82
    },
    {
      "epoch": 0.0011033529300334662,
      "grad_norm": 3.2217254638671875,
      "learning_rate": 9.479641109151373e-05,
      "loss": 1.0166,
      "step": 83
    },
    {
      "epoch": 0.00111664633882905,
      "grad_norm": 2.9364144802093506,
      "learning_rate": 9.463295047280891e-05,
      "loss": 0.9075,
      "step": 84
    },
    {
      "epoch": 0.001129939747624634,
      "grad_norm": 3.6510021686553955,
      "learning_rate": 9.446710744412595e-05,
      "loss": 0.9151,
      "step": 85
    },
    {
      "epoch": 0.001143233156420218,
      "grad_norm": 3.259711742401123,
      "learning_rate": 9.429889085780557e-05,
      "loss": 0.7931,
      "step": 86
    },
    {
      "epoch": 0.0011565265652158018,
      "grad_norm": 3.346482276916504,
      "learning_rate": 9.41283096928839e-05,
      "loss": 0.8724,
      "step": 87
    },
    {
      "epoch": 0.0011698199740113858,
      "grad_norm": 3.270686626434326,
      "learning_rate": 9.395537305461311e-05,
      "loss": 0.7726,
      "step": 88
    },
    {
      "epoch": 0.0011831133828069697,
      "grad_norm": 3.747311592102051,
      "learning_rate": 9.378009017397542e-05,
      "loss": 0.8455,
      "step": 89
    },
    {
      "epoch": 0.0011964067916025536,
      "grad_norm": 3.821293592453003,
      "learning_rate": 9.360247040719039e-05,
      "loss": 0.757,
      "step": 90
    },
    {
      "epoch": 0.0012097002003981375,
      "grad_norm": 4.054718017578125,
      "learning_rate": 9.342252323521545e-05,
      "loss": 0.9279,
      "step": 91
    },
    {
      "epoch": 0.0012229936091937214,
      "grad_norm": 3.5439743995666504,
      "learning_rate": 9.324025826323994e-05,
      "loss": 0.7374,
      "step": 92
    },
    {
      "epoch": 0.0012362870179893056,
      "grad_norm": 4.841712951660156,
      "learning_rate": 9.305568522017227e-05,
      "loss": 0.7237,
      "step": 93
    },
    {
      "epoch": 0.0012495804267848895,
      "grad_norm": 5.552839279174805,
      "learning_rate": 9.286881395812066e-05,
      "loss": 0.8263,
      "step": 94
    },
    {
      "epoch": 0.0012628738355804734,
      "grad_norm": 5.980641841888428,
      "learning_rate": 9.267965445186733e-05,
      "loss": 1.141,
      "step": 95
    },
    {
      "epoch": 0.0012761672443760573,
      "grad_norm": 5.712211608886719,
      "learning_rate": 9.248821679833596e-05,
      "loss": 1.0009,
      "step": 96
    },
    {
      "epoch": 0.0012894606531716412,
      "grad_norm": 4.681507110595703,
      "learning_rate": 9.229451121605279e-05,
      "loss": 0.8064,
      "step": 97
    },
    {
      "epoch": 0.0013027540619672251,
      "grad_norm": 5.275393009185791,
      "learning_rate": 9.209854804460121e-05,
      "loss": 0.9145,
      "step": 98
    },
    {
      "epoch": 0.001316047470762809,
      "grad_norm": 5.1414361000061035,
      "learning_rate": 9.190033774406977e-05,
      "loss": 0.92,
      "step": 99
    },
    {
      "epoch": 0.001329340879558393,
      "grad_norm": 6.979901313781738,
      "learning_rate": 9.16998908944939e-05,
      "loss": 0.9631,
      "step": 100
    },
    {
      "epoch": 0.0013426342883539769,
      "grad_norm": 3.0721938610076904,
      "learning_rate": 9.149721819529119e-05,
      "loss": 1.3745,
      "step": 101
    },
    {
      "epoch": 0.0013559276971495608,
      "grad_norm": 2.620919704437256,
      "learning_rate": 9.129233046469022e-05,
      "loss": 1.2247,
      "step": 102
    },
    {
      "epoch": 0.0013692211059451447,
      "grad_norm": 2.184368371963501,
      "learning_rate": 9.108523863915314e-05,
      "loss": 1.0819,
      "step": 103
    },
    {
      "epoch": 0.0013825145147407286,
      "grad_norm": 1.9781068563461304,
      "learning_rate": 9.087595377279192e-05,
      "loss": 1.1311,
      "step": 104
    },
    {
      "epoch": 0.0013958079235363125,
      "grad_norm": 1.648494005203247,
      "learning_rate": 9.066448703677828e-05,
      "loss": 1.0057,
      "step": 105
    },
    {
      "epoch": 0.0014091013323318964,
      "grad_norm": 1.681408405303955,
      "learning_rate": 9.045084971874738e-05,
      "loss": 0.9557,
      "step": 106
    },
    {
      "epoch": 0.0014223947411274806,
      "grad_norm": 1.9565364122390747,
      "learning_rate": 9.023505322219536e-05,
      "loss": 0.9797,
      "step": 107
    },
    {
      "epoch": 0.0014356881499230645,
      "grad_norm": 1.8517160415649414,
      "learning_rate": 9.001710906587064e-05,
      "loss": 1.1434,
      "step": 108
    },
    {
      "epoch": 0.0014489815587186484,
      "grad_norm": 2.314161539077759,
      "learning_rate": 8.9797028883159e-05,
      "loss": 0.9643,
      "step": 109
    },
    {
      "epoch": 0.0014622749675142323,
      "grad_norm": 2.426609516143799,
      "learning_rate": 8.957482442146272e-05,
      "loss": 0.9958,
      "step": 110
    },
    {
      "epoch": 0.0014755683763098162,
      "grad_norm": 2.338745355606079,
      "learning_rate": 8.935050754157344e-05,
      "loss": 0.8958,
      "step": 111
    },
    {
      "epoch": 0.0014888617851054001,
      "grad_norm": 2.2527472972869873,
      "learning_rate": 8.912409021703913e-05,
      "loss": 0.8262,
      "step": 112
    },
    {
      "epoch": 0.001502155193900984,
      "grad_norm": 2.3767216205596924,
      "learning_rate": 8.889558453352492e-05,
      "loss": 0.9135,
      "step": 113
    },
    {
      "epoch": 0.001515448602696568,
      "grad_norm": 2.25586199760437,
      "learning_rate": 8.866500268816803e-05,
      "loss": 0.9426,
      "step": 114
    },
    {
      "epoch": 0.0015287420114921519,
      "grad_norm": 2.027083158493042,
      "learning_rate": 8.84323569889266e-05,
      "loss": 1.0912,
      "step": 115
    },
    {
      "epoch": 0.0015420354202877358,
      "grad_norm": 2.3341987133026123,
      "learning_rate": 8.819765985392296e-05,
      "loss": 0.964,
      "step": 116
    },
    {
      "epoch": 0.0015553288290833197,
      "grad_norm": 2.9502508640289307,
      "learning_rate": 8.79609238107805e-05,
      "loss": 0.9789,
      "step": 117
    },
    {
      "epoch": 0.0015686222378789036,
      "grad_norm": 2.1710007190704346,
      "learning_rate": 8.772216149595513e-05,
      "loss": 0.8787,
      "step": 118
    },
    {
      "epoch": 0.0015819156466744875,
      "grad_norm": 2.401892900466919,
      "learning_rate": 8.748138565406081e-05,
      "loss": 0.9684,
      "step": 119
    },
    {
      "epoch": 0.0015952090554700715,
      "grad_norm": 2.7512049674987793,
      "learning_rate": 8.72386091371891e-05,
      "loss": 0.8961,
      "step": 120
    },
    {
      "epoch": 0.0016085024642656556,
      "grad_norm": 2.9185006618499756,
      "learning_rate": 8.699384490422331e-05,
      "loss": 0.8282,
      "step": 121
    },
    {
      "epoch": 0.0016217958730612395,
      "grad_norm": 2.936816930770874,
      "learning_rate": 8.674710602014671e-05,
      "loss": 0.8772,
      "step": 122
    },
    {
      "epoch": 0.0016350892818568234,
      "grad_norm": 3.143145799636841,
      "learning_rate": 8.649840565534513e-05,
      "loss": 0.9349,
      "step": 123
    },
    {
      "epoch": 0.0016483826906524073,
      "grad_norm": 2.681953191757202,
      "learning_rate": 8.624775708490402e-05,
      "loss": 0.9328,
      "step": 124
    },
    {
      "epoch": 0.0016616760994479912,
      "grad_norm": 3.1768240928649902,
      "learning_rate": 8.59951736878998e-05,
      "loss": 0.8815,
      "step": 125
    },
    {
      "epoch": 0.0016749695082435752,
      "grad_norm": 3.0723233222961426,
      "learning_rate": 8.574066894668573e-05,
      "loss": 0.9168,
      "step": 126
    },
    {
      "epoch": 0.001688262917039159,
      "grad_norm": 3.1787002086639404,
      "learning_rate": 8.548425644617224e-05,
      "loss": 0.9053,
      "step": 127
    },
    {
      "epoch": 0.001701556325834743,
      "grad_norm": 3.2121903896331787,
      "learning_rate": 8.522594987310184e-05,
      "loss": 0.8593,
      "step": 128
    },
    {
      "epoch": 0.001714849734630327,
      "grad_norm": 2.9943299293518066,
      "learning_rate": 8.49657630153185e-05,
      "loss": 0.9997,
      "step": 129
    },
    {
      "epoch": 0.0017281431434259108,
      "grad_norm": 3.535433292388916,
      "learning_rate": 8.47037097610317e-05,
      "loss": 0.9114,
      "step": 130
    },
    {
      "epoch": 0.0017414365522214947,
      "grad_norm": 3.203751564025879,
      "learning_rate": 8.443980409807512e-05,
      "loss": 0.9689,
      "step": 131
    },
    {
      "epoch": 0.0017547299610170786,
      "grad_norm": 3.2908623218536377,
      "learning_rate": 8.417406011315998e-05,
      "loss": 1.0035,
      "step": 132
    },
    {
      "epoch": 0.0017680233698126626,
      "grad_norm": 2.752976179122925,
      "learning_rate": 8.390649199112315e-05,
      "loss": 0.9917,
      "step": 133
    },
    {
      "epoch": 0.0017813167786082465,
      "grad_norm": 3.3330466747283936,
      "learning_rate": 8.363711401417e-05,
      "loss": 1.0635,
      "step": 134
    },
    {
      "epoch": 0.0017946101874038306,
      "grad_norm": 3.2648167610168457,
      "learning_rate": 8.336594056111197e-05,
      "loss": 0.8795,
      "step": 135
    },
    {
      "epoch": 0.0018079035961994145,
      "grad_norm": 3.277704954147339,
      "learning_rate": 8.309298610659916e-05,
      "loss": 0.8441,
      "step": 136
    },
    {
      "epoch": 0.0018211970049949984,
      "grad_norm": 3.603541374206543,
      "learning_rate": 8.281826522034764e-05,
      "loss": 0.9423,
      "step": 137
    },
    {
      "epoch": 0.0018344904137905823,
      "grad_norm": 2.947814464569092,
      "learning_rate": 8.254179256636179e-05,
      "loss": 0.8583,
      "step": 138
    },
    {
      "epoch": 0.0018477838225861663,
      "grad_norm": 3.5735108852386475,
      "learning_rate": 8.226358290215151e-05,
      "loss": 1.0053,
      "step": 139
    },
    {
      "epoch": 0.0018610772313817502,
      "grad_norm": 4.395263195037842,
      "learning_rate": 8.198365107794457e-05,
      "loss": 0.908,
      "step": 140
    },
    {
      "epoch": 0.001874370640177334,
      "grad_norm": 4.384109973907471,
      "learning_rate": 8.17020120358939e-05,
      "loss": 0.9973,
      "step": 141
    },
    {
      "epoch": 0.001887664048972918,
      "grad_norm": 3.42045521736145,
      "learning_rate": 8.141868080927996e-05,
      "loss": 0.8164,
      "step": 142
    },
    {
      "epoch": 0.001900957457768502,
      "grad_norm": 3.931617259979248,
      "learning_rate": 8.113367252170844e-05,
      "loss": 0.9282,
      "step": 143
    },
    {
      "epoch": 0.0019142508665640858,
      "grad_norm": 4.341536045074463,
      "learning_rate": 8.084700238630283e-05,
      "loss": 0.9588,
      "step": 144
    },
    {
      "epoch": 0.0019275442753596697,
      "grad_norm": 4.990645885467529,
      "learning_rate": 8.055868570489247e-05,
      "loss": 0.9038,
      "step": 145
    },
    {
      "epoch": 0.0019408376841552537,
      "grad_norm": 5.111573219299316,
      "learning_rate": 8.026873786719573e-05,
      "loss": 0.9121,
      "step": 146
    },
    {
      "epoch": 0.0019541310929508376,
      "grad_norm": 4.322947978973389,
      "learning_rate": 7.997717434999861e-05,
      "loss": 0.8981,
      "step": 147
    },
    {
      "epoch": 0.0019674245017464215,
      "grad_norm": 4.9193010330200195,
      "learning_rate": 7.968401071632855e-05,
      "loss": 0.9212,
      "step": 148
    },
    {
      "epoch": 0.0019807179105420054,
      "grad_norm": 7.261848449707031,
      "learning_rate": 7.938926261462366e-05,
      "loss": 1.1311,
      "step": 149
    },
    {
      "epoch": 0.0019940113193375893,
      "grad_norm": 9.099236488342285,
      "learning_rate": 7.909294577789766e-05,
      "loss": 0.8936,
      "step": 150
    },
    {
      "epoch": 0.0019940113193375893,
      "eval_loss": 0.9891857504844666,
      "eval_runtime": 3478.2034,
      "eval_samples_per_second": 36.426,
      "eval_steps_per_second": 9.106,
      "step": 150
    },
    {
      "epoch": 0.0020073047281331732,
      "grad_norm": 2.0536725521087646,
      "learning_rate": 7.879507602289979e-05,
      "loss": 1.2124,
      "step": 151
    },
    {
      "epoch": 0.002020598136928757,
      "grad_norm": 2.2409486770629883,
      "learning_rate": 7.849566924927082e-05,
      "loss": 0.9982,
      "step": 152
    },
    {
      "epoch": 0.002033891545724341,
      "grad_norm": 1.8208754062652588,
      "learning_rate": 7.819474143869414e-05,
      "loss": 1.0108,
      "step": 153
    },
    {
      "epoch": 0.002047184954519925,
      "grad_norm": 1.8151131868362427,
      "learning_rate": 7.789230865404287e-05,
      "loss": 0.9152,
      "step": 154
    },
    {
      "epoch": 0.002060478363315509,
      "grad_norm": 1.7584021091461182,
      "learning_rate": 7.75883870385223e-05,
      "loss": 0.9911,
      "step": 155
    },
    {
      "epoch": 0.002073771772111093,
      "grad_norm": 1.859797477722168,
      "learning_rate": 7.728299281480833e-05,
      "loss": 0.8977,
      "step": 156
    },
    {
      "epoch": 0.002087065180906677,
      "grad_norm": 1.963524580001831,
      "learning_rate": 7.697614228418148e-05,
      "loss": 0.9215,
      "step": 157
    },
    {
      "epoch": 0.002100358589702261,
      "grad_norm": 1.8495042324066162,
      "learning_rate": 7.666785182565677e-05,
      "loss": 0.9008,
      "step": 158
    },
    {
      "epoch": 0.002113651998497845,
      "grad_norm": 1.8056210279464722,
      "learning_rate": 7.635813789510941e-05,
      "loss": 1.0834,
      "step": 159
    },
    {
      "epoch": 0.002126945407293429,
      "grad_norm": 2.264211893081665,
      "learning_rate": 7.604701702439651e-05,
      "loss": 1.0834,
      "step": 160
    },
    {
      "epoch": 0.002140238816089013,
      "grad_norm": 2.530268669128418,
      "learning_rate": 7.573450582047457e-05,
      "loss": 1.0092,
      "step": 161
    },
    {
      "epoch": 0.0021535322248845967,
      "grad_norm": 2.144922971725464,
      "learning_rate": 7.542062096451305e-05,
      "loss": 0.8963,
      "step": 162
    },
    {
      "epoch": 0.0021668256336801806,
      "grad_norm": 2.751955986022949,
      "learning_rate": 7.510537921100398e-05,
      "loss": 0.9765,
      "step": 163
    },
    {
      "epoch": 0.0021801190424757646,
      "grad_norm": 2.1858596801757812,
      "learning_rate": 7.47887973868676e-05,
      "loss": 1.0205,
      "step": 164
    },
    {
      "epoch": 0.0021934124512713485,
      "grad_norm": 2.055968999862671,
      "learning_rate": 7.447089239055428e-05,
      "loss": 0.8658,
      "step": 165
    },
    {
      "epoch": 0.0022067058600669324,
      "grad_norm": 2.7165462970733643,
      "learning_rate": 7.41516811911424e-05,
      "loss": 0.8954,
      "step": 166
    },
    {
      "epoch": 0.0022199992688625163,
      "grad_norm": 2.2613511085510254,
      "learning_rate": 7.383118082743262e-05,
      "loss": 1.0412,
      "step": 167
    },
    {
      "epoch": 0.0022332926776581,
      "grad_norm": 2.174492359161377,
      "learning_rate": 7.350940840703842e-05,
      "loss": 0.9415,
      "step": 168
    },
    {
      "epoch": 0.002246586086453684,
      "grad_norm": 2.3029043674468994,
      "learning_rate": 7.318638110547288e-05,
      "loss": 1.0003,
      "step": 169
    },
    {
      "epoch": 0.002259879495249268,
      "grad_norm": 2.20206356048584,
      "learning_rate": 7.286211616523193e-05,
      "loss": 1.0255,
      "step": 170
    },
    {
      "epoch": 0.002273172904044852,
      "grad_norm": 2.3058087825775146,
      "learning_rate": 7.253663089487395e-05,
      "loss": 0.9273,
      "step": 171
    },
    {
      "epoch": 0.002286466312840436,
      "grad_norm": 2.4932408332824707,
      "learning_rate": 7.220994266809591e-05,
      "loss": 0.9712,
      "step": 172
    },
    {
      "epoch": 0.0022997597216360198,
      "grad_norm": 2.348066568374634,
      "learning_rate": 7.188206892280594e-05,
      "loss": 0.8345,
      "step": 173
    },
    {
      "epoch": 0.0023130531304316037,
      "grad_norm": 2.508608341217041,
      "learning_rate": 7.155302716019263e-05,
      "loss": 1.069,
      "step": 174
    },
    {
      "epoch": 0.0023263465392271876,
      "grad_norm": 2.3190245628356934,
      "learning_rate": 7.122283494379076e-05,
      "loss": 0.7667,
      "step": 175
    },
    {
      "epoch": 0.0023396399480227715,
      "grad_norm": 2.9227256774902344,
      "learning_rate": 7.089150989854385e-05,
      "loss": 0.8513,
      "step": 176
    },
    {
      "epoch": 0.0023529333568183554,
      "grad_norm": 2.6290760040283203,
      "learning_rate": 7.055906970986336e-05,
      "loss": 0.853,
      "step": 177
    },
    {
      "epoch": 0.0023662267656139394,
      "grad_norm": 2.2298004627227783,
      "learning_rate": 7.022553212268469e-05,
      "loss": 0.8112,
      "step": 178
    },
    {
      "epoch": 0.0023795201744095233,
      "grad_norm": 2.7853963375091553,
      "learning_rate": 6.989091494051998e-05,
      "loss": 0.8567,
      "step": 179
    },
    {
      "epoch": 0.002392813583205107,
      "grad_norm": 2.7704803943634033,
      "learning_rate": 6.95552360245078e-05,
      "loss": 0.7658,
      "step": 180
    },
    {
      "epoch": 0.002406106992000691,
      "grad_norm": 3.0688202381134033,
      "learning_rate": 6.92185132924598e-05,
      "loss": 0.8616,
      "step": 181
    },
    {
      "epoch": 0.002419400400796275,
      "grad_norm": 3.337465763092041,
      "learning_rate": 6.888076471790424e-05,
      "loss": 0.9898,
      "step": 182
    },
    {
      "epoch": 0.002432693809591859,
      "grad_norm": 3.242480516433716,
      "learning_rate": 6.85420083291266e-05,
      "loss": 0.7605,
      "step": 183
    },
    {
      "epoch": 0.002445987218387443,
      "grad_norm": 3.2811238765716553,
      "learning_rate": 6.820226220820732e-05,
      "loss": 0.9702,
      "step": 184
    },
    {
      "epoch": 0.002459280627183027,
      "grad_norm": 3.416262149810791,
      "learning_rate": 6.786154449005665e-05,
      "loss": 0.8603,
      "step": 185
    },
    {
      "epoch": 0.002472574035978611,
      "grad_norm": 3.3136789798736572,
      "learning_rate": 6.751987336144648e-05,
      "loss": 0.9065,
      "step": 186
    },
    {
      "epoch": 0.002485867444774195,
      "grad_norm": 4.629530906677246,
      "learning_rate": 6.717726706003974e-05,
      "loss": 0.8809,
      "step": 187
    },
    {
      "epoch": 0.002499160853569779,
      "grad_norm": 3.5643258094787598,
      "learning_rate": 6.683374387341687e-05,
      "loss": 0.8574,
      "step": 188
    },
    {
      "epoch": 0.002512454262365363,
      "grad_norm": 3.7351303100585938,
      "learning_rate": 6.648932213809962e-05,
      "loss": 0.9614,
      "step": 189
    },
    {
      "epoch": 0.0025257476711609468,
      "grad_norm": 4.150784015655518,
      "learning_rate": 6.614402023857232e-05,
      "loss": 0.9416,
      "step": 190
    },
    {
      "epoch": 0.0025390410799565307,
      "grad_norm": 3.927623987197876,
      "learning_rate": 6.579785660630056e-05,
      "loss": 0.7178,
      "step": 191
    },
    {
      "epoch": 0.0025523344887521146,
      "grad_norm": 3.512359142303467,
      "learning_rate": 6.545084971874738e-05,
      "loss": 0.918,
      "step": 192
    },
    {
      "epoch": 0.0025656278975476985,
      "grad_norm": 3.812126874923706,
      "learning_rate": 6.510301809838689e-05,
      "loss": 0.885,
      "step": 193
    },
    {
      "epoch": 0.0025789213063432824,
      "grad_norm": 4.095211029052734,
      "learning_rate": 6.475438031171574e-05,
      "loss": 0.8032,
      "step": 194
    },
    {
      "epoch": 0.0025922147151388663,
      "grad_norm": 4.639618396759033,
      "learning_rate": 6.440495496826189e-05,
      "loss": 0.8512,
      "step": 195
    },
    {
      "epoch": 0.0026055081239344502,
      "grad_norm": 4.100682258605957,
      "learning_rate": 6.405476071959143e-05,
      "loss": 0.9202,
      "step": 196
    },
    {
      "epoch": 0.002618801532730034,
      "grad_norm": 4.494955539703369,
      "learning_rate": 6.370381625831292e-05,
      "loss": 0.7531,
      "step": 197
    },
    {
      "epoch": 0.002632094941525618,
      "grad_norm": 4.195039749145508,
      "learning_rate": 6.335214031707965e-05,
      "loss": 0.872,
      "step": 198
    },
    {
      "epoch": 0.002645388350321202,
      "grad_norm": 4.241918087005615,
      "learning_rate": 6.299975166758971e-05,
      "loss": 0.9042,
      "step": 199
    },
    {
      "epoch": 0.002658681759116786,
      "grad_norm": 7.482029438018799,
      "learning_rate": 6.264666911958404e-05,
      "loss": 0.968,
      "step": 200
    },
    {
      "epoch": 0.00267197516791237,
      "grad_norm": 1.535090684890747,
      "learning_rate": 6.229291151984233e-05,
      "loss": 1.1693,
      "step": 201
    },
    {
      "epoch": 0.0026852685767079537,
      "grad_norm": 2.0758211612701416,
      "learning_rate": 6.19384977511771e-05,
      "loss": 0.9393,
      "step": 202
    },
    {
      "epoch": 0.0026985619855035376,
      "grad_norm": 2.1649084091186523,
      "learning_rate": 6.158344673142573e-05,
      "loss": 1.0957,
      "step": 203
    },
    {
      "epoch": 0.0027118553942991216,
      "grad_norm": 1.799479365348816,
      "learning_rate": 6.122777741244067e-05,
      "loss": 1.1526,
      "step": 204
    },
    {
      "epoch": 0.0027251488030947055,
      "grad_norm": 1.7623484134674072,
      "learning_rate": 6.0871508779077856e-05,
      "loss": 0.8282,
      "step": 205
    },
    {
      "epoch": 0.0027384422118902894,
      "grad_norm": 1.7054765224456787,
      "learning_rate": 6.051465984818332e-05,
      "loss": 1.0077,
      "step": 206
    },
    {
      "epoch": 0.0027517356206858733,
      "grad_norm": 1.9788992404937744,
      "learning_rate": 6.015724966757812e-05,
      "loss": 0.869,
      "step": 207
    },
    {
      "epoch": 0.0027650290294814572,
      "grad_norm": 1.9093753099441528,
      "learning_rate": 5.979929731504158e-05,
      "loss": 0.9314,
      "step": 208
    },
    {
      "epoch": 0.002778322438277041,
      "grad_norm": 1.8340977430343628,
      "learning_rate": 5.944082189729301e-05,
      "loss": 0.9003,
      "step": 209
    },
    {
      "epoch": 0.002791615847072625,
      "grad_norm": 1.8456708192825317,
      "learning_rate": 5.908184254897182e-05,
      "loss": 1.0621,
      "step": 210
    },
    {
      "epoch": 0.002804909255868209,
      "grad_norm": 2.148777723312378,
      "learning_rate": 5.872237843161612e-05,
      "loss": 1.094,
      "step": 211
    },
    {
      "epoch": 0.002818202664663793,
      "grad_norm": 2.2333133220672607,
      "learning_rate": 5.8362448732639894e-05,
      "loss": 1.0306,
      "step": 212
    },
    {
      "epoch": 0.002831496073459377,
      "grad_norm": 2.136981248855591,
      "learning_rate": 5.800207266430895e-05,
      "loss": 0.9077,
      "step": 213
    },
    {
      "epoch": 0.002844789482254961,
      "grad_norm": 1.8241512775421143,
      "learning_rate": 5.764126946271526e-05,
      "loss": 1.0229,
      "step": 214
    },
    {
      "epoch": 0.002858082891050545,
      "grad_norm": 2.608855962753296,
      "learning_rate": 5.7280058386750255e-05,
      "loss": 0.9708,
      "step": 215
    },
    {
      "epoch": 0.002871376299846129,
      "grad_norm": 1.9406535625457764,
      "learning_rate": 5.6918458717076815e-05,
      "loss": 1.016,
      "step": 216
    },
    {
      "epoch": 0.002884669708641713,
      "grad_norm": 2.244218587875366,
      "learning_rate": 5.655648975510014e-05,
      "loss": 0.9253,
      "step": 217
    },
    {
      "epoch": 0.002897963117437297,
      "grad_norm": 2.27140474319458,
      "learning_rate": 5.61941708219374e-05,
      "loss": 0.81,
      "step": 218
    },
    {
      "epoch": 0.0029112565262328807,
      "grad_norm": 2.125356674194336,
      "learning_rate": 5.583152125738651e-05,
      "loss": 0.8303,
      "step": 219
    },
    {
      "epoch": 0.0029245499350284646,
      "grad_norm": 2.606123208999634,
      "learning_rate": 5.546856041889373e-05,
      "loss": 0.9575,
      "step": 220
    },
    {
      "epoch": 0.0029378433438240485,
      "grad_norm": 2.3250555992126465,
      "learning_rate": 5.510530768052047e-05,
      "loss": 0.9056,
      "step": 221
    },
    {
      "epoch": 0.0029511367526196325,
      "grad_norm": 2.9340219497680664,
      "learning_rate": 5.4741782431909136e-05,
      "loss": 0.881,
      "step": 222
    },
    {
      "epoch": 0.0029644301614152164,
      "grad_norm": 2.7451937198638916,
      "learning_rate": 5.437800407724812e-05,
      "loss": 0.8854,
      "step": 223
    },
    {
      "epoch": 0.0029777235702108003,
      "grad_norm": 2.617919921875,
      "learning_rate": 5.401399203423606e-05,
      "loss": 0.921,
      "step": 224
    },
    {
      "epoch": 0.002991016979006384,
      "grad_norm": 2.570213556289673,
      "learning_rate": 5.364976573304538e-05,
      "loss": 0.788,
      "step": 225
    },
    {
      "epoch": 0.003004310387801968,
      "grad_norm": 2.5218234062194824,
      "learning_rate": 5.328534461528515e-05,
      "loss": 0.8243,
      "step": 226
    },
    {
      "epoch": 0.003017603796597552,
      "grad_norm": 2.5379672050476074,
      "learning_rate": 5.29207481329633e-05,
      "loss": 0.9516,
      "step": 227
    },
    {
      "epoch": 0.003030897205393136,
      "grad_norm": 2.763002634048462,
      "learning_rate": 5.2555995747448364e-05,
      "loss": 0.916,
      "step": 228
    },
    {
      "epoch": 0.00304419061418872,
      "grad_norm": 2.5635182857513428,
      "learning_rate": 5.2191106928430644e-05,
      "loss": 0.8179,
      "step": 229
    },
    {
      "epoch": 0.0030574840229843038,
      "grad_norm": 2.8950819969177246,
      "learning_rate": 5.182610115288295e-05,
      "loss": 0.8488,
      "step": 230
    },
    {
      "epoch": 0.0030707774317798877,
      "grad_norm": 3.1561074256896973,
      "learning_rate": 5.1460997904021005e-05,
      "loss": 0.8642,
      "step": 231
    },
    {
      "epoch": 0.0030840708405754716,
      "grad_norm": 3.150465965270996,
      "learning_rate": 5.109581667026341e-05,
      "loss": 1.0022,
      "step": 232
    },
    {
      "epoch": 0.0030973642493710555,
      "grad_norm": 2.8613460063934326,
      "learning_rate": 5.073057694419147e-05,
      "loss": 0.9774,
      "step": 233
    },
    {
      "epoch": 0.0031106576581666394,
      "grad_norm": 2.627713203430176,
      "learning_rate": 5.036529822150865e-05,
      "loss": 0.8306,
      "step": 234
    },
    {
      "epoch": 0.0031239510669622233,
      "grad_norm": 3.2326531410217285,
      "learning_rate": 5e-05,
      "loss": 0.9001,
      "step": 235
    },
    {
      "epoch": 0.0031372444757578073,
      "grad_norm": 4.3131489753723145,
      "learning_rate": 4.963470177849135e-05,
      "loss": 1.0435,
      "step": 236
    },
    {
      "epoch": 0.003150537884553391,
      "grad_norm": 3.0774354934692383,
      "learning_rate": 4.9269423055808544e-05,
      "loss": 0.8572,
      "step": 237
    },
    {
      "epoch": 0.003163831293348975,
      "grad_norm": 2.716817855834961,
      "learning_rate": 4.8904183329736596e-05,
      "loss": 0.8809,
      "step": 238
    },
    {
      "epoch": 0.003177124702144559,
      "grad_norm": 2.6746835708618164,
      "learning_rate": 4.853900209597903e-05,
      "loss": 0.7999,
      "step": 239
    },
    {
      "epoch": 0.003190418110940143,
      "grad_norm": 3.784773826599121,
      "learning_rate": 4.817389884711705e-05,
      "loss": 0.9177,
      "step": 240
    },
    {
      "epoch": 0.003203711519735727,
      "grad_norm": 3.225341320037842,
      "learning_rate": 4.7808893071569374e-05,
      "loss": 0.7424,
      "step": 241
    },
    {
      "epoch": 0.003217004928531311,
      "grad_norm": 3.6971752643585205,
      "learning_rate": 4.744400425255165e-05,
      "loss": 0.7437,
      "step": 242
    },
    {
      "epoch": 0.003230298337326895,
      "grad_norm": 3.3899476528167725,
      "learning_rate": 4.707925186703671e-05,
      "loss": 0.5778,
      "step": 243
    },
    {
      "epoch": 0.003243591746122479,
      "grad_norm": 3.9439117908477783,
      "learning_rate": 4.671465538471486e-05,
      "loss": 0.8173,
      "step": 244
    },
    {
      "epoch": 0.003256885154918063,
      "grad_norm": 3.429020643234253,
      "learning_rate": 4.6350234266954626e-05,
      "loss": 0.7432,
      "step": 245
    },
    {
      "epoch": 0.003270178563713647,
      "grad_norm": 4.743805885314941,
      "learning_rate": 4.598600796576395e-05,
      "loss": 0.7756,
      "step": 246
    },
    {
      "epoch": 0.0032834719725092307,
      "grad_norm": 4.231058120727539,
      "learning_rate": 4.562199592275188e-05,
      "loss": 0.7174,
      "step": 247
    },
    {
      "epoch": 0.0032967653813048147,
      "grad_norm": 4.012794494628906,
      "learning_rate": 4.5258217568090876e-05,
      "loss": 0.673,
      "step": 248
    },
    {
      "epoch": 0.0033100587901003986,
      "grad_norm": 5.038784503936768,
      "learning_rate": 4.4894692319479544e-05,
      "loss": 0.786,
      "step": 249
    },
    {
      "epoch": 0.0033233521988959825,
      "grad_norm": 6.414555072784424,
      "learning_rate": 4.4531439581106295e-05,
      "loss": 0.7456,
      "step": 250
    },
    {
      "epoch": 0.0033366456076915664,
      "grad_norm": 1.2634704113006592,
      "learning_rate": 4.4168478742613506e-05,
      "loss": 0.9463,
      "step": 251
    },
    {
      "epoch": 0.0033499390164871503,
      "grad_norm": 1.8161299228668213,
      "learning_rate": 4.38058291780626e-05,
      "loss": 0.8844,
      "step": 252
    },
    {
      "epoch": 0.0033632324252827342,
      "grad_norm": 2.100287437438965,
      "learning_rate": 4.3443510244899864e-05,
      "loss": 0.973,
      "step": 253
    },
    {
      "epoch": 0.003376525834078318,
      "grad_norm": 1.925026297569275,
      "learning_rate": 4.308154128292318e-05,
      "loss": 0.8365,
      "step": 254
    },
    {
      "epoch": 0.003389819242873902,
      "grad_norm": 1.9287835359573364,
      "learning_rate": 4.271994161324977e-05,
      "loss": 1.0131,
      "step": 255
    },
    {
      "epoch": 0.003403112651669486,
      "grad_norm": 1.6935553550720215,
      "learning_rate": 4.235873053728475e-05,
      "loss": 0.8375,
      "step": 256
    },
    {
      "epoch": 0.00341640606046507,
      "grad_norm": 2.200186252593994,
      "learning_rate": 4.199792733569107e-05,
      "loss": 0.8687,
      "step": 257
    },
    {
      "epoch": 0.003429699469260654,
      "grad_norm": 1.9003078937530518,
      "learning_rate": 4.163755126736012e-05,
      "loss": 0.8753,
      "step": 258
    },
    {
      "epoch": 0.0034429928780562377,
      "grad_norm": 4.403824329376221,
      "learning_rate": 4.127762156838389e-05,
      "loss": 1.0423,
      "step": 259
    },
    {
      "epoch": 0.0034562862868518216,
      "grad_norm": 2.0057454109191895,
      "learning_rate": 4.0918157451028185e-05,
      "loss": 1.0278,
      "step": 260
    },
    {
      "epoch": 0.0034695796956474055,
      "grad_norm": 1.85740065574646,
      "learning_rate": 4.055917810270698e-05,
      "loss": 1.0295,
      "step": 261
    },
    {
      "epoch": 0.0034828731044429895,
      "grad_norm": 2.0854363441467285,
      "learning_rate": 4.020070268495843e-05,
      "loss": 0.883,
      "step": 262
    },
    {
      "epoch": 0.0034961665132385734,
      "grad_norm": 1.8586416244506836,
      "learning_rate": 3.9842750332421896e-05,
      "loss": 0.9127,
      "step": 263
    },
    {
      "epoch": 0.0035094599220341573,
      "grad_norm": 2.047469139099121,
      "learning_rate": 3.94853401518167e-05,
      "loss": 0.9311,
      "step": 264
    },
    {
      "epoch": 0.003522753330829741,
      "grad_norm": 2.0155272483825684,
      "learning_rate": 3.9128491220922156e-05,
      "loss": 0.806,
      "step": 265
    },
    {
      "epoch": 0.003536046739625325,
      "grad_norm": 1.9358389377593994,
      "learning_rate": 3.877222258755935e-05,
      "loss": 0.989,
      "step": 266
    },
    {
      "epoch": 0.003549340148420909,
      "grad_norm": 2.470038414001465,
      "learning_rate": 3.8416553268574285e-05,
      "loss": 0.8901,
      "step": 267
    },
    {
      "epoch": 0.003562633557216493,
      "grad_norm": 2.42508864402771,
      "learning_rate": 3.80615022488229e-05,
      "loss": 0.9417,
      "step": 268
    },
    {
      "epoch": 0.003575926966012077,
      "grad_norm": 2.4037420749664307,
      "learning_rate": 3.770708848015768e-05,
      "loss": 0.8843,
      "step": 269
    },
    {
      "epoch": 0.003589220374807661,
      "grad_norm": 2.5379691123962402,
      "learning_rate": 3.735333088041596e-05,
      "loss": 0.9576,
      "step": 270
    },
    {
      "epoch": 0.003602513783603245,
      "grad_norm": 2.117372989654541,
      "learning_rate": 3.7000248332410304e-05,
      "loss": 0.8887,
      "step": 271
    },
    {
      "epoch": 0.003615807192398829,
      "grad_norm": 2.1954541206359863,
      "learning_rate": 3.664785968292036e-05,
      "loss": 0.8735,
      "step": 272
    },
    {
      "epoch": 0.003629100601194413,
      "grad_norm": 2.301987648010254,
      "learning_rate": 3.629618374168711e-05,
      "loss": 0.9798,
      "step": 273
    },
    {
      "epoch": 0.003642394009989997,
      "grad_norm": 2.793152332305908,
      "learning_rate": 3.594523928040859e-05,
      "loss": 1.0083,
      "step": 274
    },
    {
      "epoch": 0.0036556874187855808,
      "grad_norm": 2.311340570449829,
      "learning_rate": 3.5595045031738125e-05,
      "loss": 0.8128,
      "step": 275
    },
    {
      "epoch": 0.0036689808275811647,
      "grad_norm": 2.1849329471588135,
      "learning_rate": 3.5245619688284274e-05,
      "loss": 0.7448,
      "step": 276
    },
    {
      "epoch": 0.0036822742363767486,
      "grad_norm": 2.5570812225341797,
      "learning_rate": 3.4896981901613104e-05,
      "loss": 0.8764,
      "step": 277
    },
    {
      "epoch": 0.0036955676451723325,
      "grad_norm": 2.5506277084350586,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 0.8251,
      "step": 278
    },
    {
      "epoch": 0.0037088610539679164,
      "grad_norm": 2.7928378582000732,
      "learning_rate": 3.420214339369944e-05,
      "loss": 0.8239,
      "step": 279
    },
    {
      "epoch": 0.0037221544627635004,
      "grad_norm": 3.0748653411865234,
      "learning_rate": 3.38559797614277e-05,
      "loss": 0.9272,
      "step": 280
    },
    {
      "epoch": 0.0037354478715590843,
      "grad_norm": 2.7534868717193604,
      "learning_rate": 3.351067786190038e-05,
      "loss": 0.8112,
      "step": 281
    },
    {
      "epoch": 0.003748741280354668,
      "grad_norm": 2.5092287063598633,
      "learning_rate": 3.316625612658315e-05,
      "loss": 0.8439,
      "step": 282
    },
    {
      "epoch": 0.003762034689150252,
      "grad_norm": 2.871694326400757,
      "learning_rate": 3.282273293996027e-05,
      "loss": 0.8854,
      "step": 283
    },
    {
      "epoch": 0.003775328097945836,
      "grad_norm": 3.3398563861846924,
      "learning_rate": 3.248012663855353e-05,
      "loss": 0.8582,
      "step": 284
    },
    {
      "epoch": 0.00378862150674142,
      "grad_norm": 3.9510514736175537,
      "learning_rate": 3.2138455509943366e-05,
      "loss": 0.8925,
      "step": 285
    },
    {
      "epoch": 0.003801914915537004,
      "grad_norm": 3.2462363243103027,
      "learning_rate": 3.179773779179267e-05,
      "loss": 0.8432,
      "step": 286
    },
    {
      "epoch": 0.0038152083243325878,
      "grad_norm": 3.1924219131469727,
      "learning_rate": 3.145799167087342e-05,
      "loss": 0.7827,
      "step": 287
    },
    {
      "epoch": 0.0038285017331281717,
      "grad_norm": 3.794776678085327,
      "learning_rate": 3.111923528209577e-05,
      "loss": 0.8214,
      "step": 288
    },
    {
      "epoch": 0.0038417951419237556,
      "grad_norm": 3.1330740451812744,
      "learning_rate": 3.078148670754022e-05,
      "loss": 0.8446,
      "step": 289
    },
    {
      "epoch": 0.0038550885507193395,
      "grad_norm": 3.1188437938690186,
      "learning_rate": 3.0444763975492208e-05,
      "loss": 0.8169,
      "step": 290
    },
    {
      "epoch": 0.0038683819595149234,
      "grad_norm": 3.640385866165161,
      "learning_rate": 3.0109085059480017e-05,
      "loss": 0.9897,
      "step": 291
    },
    {
      "epoch": 0.0038816753683105073,
      "grad_norm": 4.044271469116211,
      "learning_rate": 2.977446787731532e-05,
      "loss": 0.8255,
      "step": 292
    },
    {
      "epoch": 0.0038949687771060912,
      "grad_norm": 3.6685760021209717,
      "learning_rate": 2.944093029013664e-05,
      "loss": 0.8258,
      "step": 293
    },
    {
      "epoch": 0.003908262185901675,
      "grad_norm": 3.5113165378570557,
      "learning_rate": 2.910849010145617e-05,
      "loss": 0.7465,
      "step": 294
    },
    {
      "epoch": 0.0039215555946972595,
      "grad_norm": 3.854799747467041,
      "learning_rate": 2.8777165056209256e-05,
      "loss": 0.6862,
      "step": 295
    },
    {
      "epoch": 0.003934849003492843,
      "grad_norm": 4.6844587326049805,
      "learning_rate": 2.8446972839807384e-05,
      "loss": 0.7543,
      "step": 296
    },
    {
      "epoch": 0.003948142412288427,
      "grad_norm": 3.536104440689087,
      "learning_rate": 2.8117931077194065e-05,
      "loss": 0.6841,
      "step": 297
    },
    {
      "epoch": 0.003961435821084011,
      "grad_norm": 4.2677903175354,
      "learning_rate": 2.7790057331904117e-05,
      "loss": 0.6677,
      "step": 298
    },
    {
      "epoch": 0.003974729229879595,
      "grad_norm": 4.6824469566345215,
      "learning_rate": 2.746336910512606e-05,
      "loss": 0.8091,
      "step": 299
    },
    {
      "epoch": 0.003988022638675179,
      "grad_norm": 6.547423839569092,
      "learning_rate": 2.7137883834768073e-05,
      "loss": 0.9708,
      "step": 300
    },
    {
      "epoch": 0.003988022638675179,
      "eval_loss": 0.850064754486084,
      "eval_runtime": 3480.1436,
      "eval_samples_per_second": 36.405,
      "eval_steps_per_second": 9.101,
      "step": 300
    },
    {
      "epoch": 0.004001316047470763,
      "grad_norm": 1.0784815549850464,
      "learning_rate": 2.6813618894527138e-05,
      "loss": 0.9837,
      "step": 301
    },
    {
      "epoch": 0.0040146094562663465,
      "grad_norm": 1.8547368049621582,
      "learning_rate": 2.6490591592961578e-05,
      "loss": 0.9933,
      "step": 302
    },
    {
      "epoch": 0.004027902865061931,
      "grad_norm": 1.9853336811065674,
      "learning_rate": 2.6168819172567392e-05,
      "loss": 1.0929,
      "step": 303
    },
    {
      "epoch": 0.004041196273857514,
      "grad_norm": 1.8064887523651123,
      "learning_rate": 2.5848318808857606e-05,
      "loss": 0.9201,
      "step": 304
    },
    {
      "epoch": 0.004054489682653099,
      "grad_norm": 1.9184132814407349,
      "learning_rate": 2.5529107609445733e-05,
      "loss": 0.7944,
      "step": 305
    },
    {
      "epoch": 0.004067783091448682,
      "grad_norm": 1.625139832496643,
      "learning_rate": 2.521120261313241e-05,
      "loss": 1.0146,
      "step": 306
    },
    {
      "epoch": 0.0040810765002442665,
      "grad_norm": 2.2314794063568115,
      "learning_rate": 2.4894620788996037e-05,
      "loss": 0.9506,
      "step": 307
    },
    {
      "epoch": 0.00409436990903985,
      "grad_norm": 2.091251850128174,
      "learning_rate": 2.457937903548695e-05,
      "loss": 1.1053,
      "step": 308
    },
    {
      "epoch": 0.004107663317835434,
      "grad_norm": 2.8359029293060303,
      "learning_rate": 2.426549417952542e-05,
      "loss": 0.9342,
      "step": 309
    },
    {
      "epoch": 0.004120956726631018,
      "grad_norm": 2.370006799697876,
      "learning_rate": 2.3952982975603496e-05,
      "loss": 0.8265,
      "step": 310
    },
    {
      "epoch": 0.004134250135426602,
      "grad_norm": 1.7976568937301636,
      "learning_rate": 2.3641862104890595e-05,
      "loss": 1.0254,
      "step": 311
    },
    {
      "epoch": 0.004147543544222186,
      "grad_norm": 2.030775547027588,
      "learning_rate": 2.3332148174343254e-05,
      "loss": 0.915,
      "step": 312
    },
    {
      "epoch": 0.00416083695301777,
      "grad_norm": 2.1562678813934326,
      "learning_rate": 2.3023857715818532e-05,
      "loss": 0.9138,
      "step": 313
    },
    {
      "epoch": 0.004174130361813354,
      "grad_norm": 2.4217755794525146,
      "learning_rate": 2.2717007185191674e-05,
      "loss": 0.9814,
      "step": 314
    },
    {
      "epoch": 0.004187423770608938,
      "grad_norm": 2.0525031089782715,
      "learning_rate": 2.24116129614777e-05,
      "loss": 1.0232,
      "step": 315
    },
    {
      "epoch": 0.004200717179404522,
      "grad_norm": 1.8559058904647827,
      "learning_rate": 2.2107691345957133e-05,
      "loss": 0.8898,
      "step": 316
    },
    {
      "epoch": 0.004214010588200106,
      "grad_norm": 1.8364156484603882,
      "learning_rate": 2.1805258561305862e-05,
      "loss": 0.9021,
      "step": 317
    },
    {
      "epoch": 0.00422730399699569,
      "grad_norm": 2.247051954269409,
      "learning_rate": 2.1504330750729186e-05,
      "loss": 0.8632,
      "step": 318
    },
    {
      "epoch": 0.0042405974057912734,
      "grad_norm": 2.744231700897217,
      "learning_rate": 2.120492397710022e-05,
      "loss": 0.9776,
      "step": 319
    },
    {
      "epoch": 0.004253890814586858,
      "grad_norm": 2.21708345413208,
      "learning_rate": 2.090705422210237e-05,
      "loss": 0.8523,
      "step": 320
    },
    {
      "epoch": 0.004267184223382441,
      "grad_norm": 2.1800765991210938,
      "learning_rate": 2.061073738537635e-05,
      "loss": 0.9291,
      "step": 321
    },
    {
      "epoch": 0.004280477632178026,
      "grad_norm": 2.510124444961548,
      "learning_rate": 2.0315989283671473e-05,
      "loss": 0.8304,
      "step": 322
    },
    {
      "epoch": 0.004293771040973609,
      "grad_norm": 2.725311517715454,
      "learning_rate": 2.0022825650001387e-05,
      "loss": 0.77,
      "step": 323
    },
    {
      "epoch": 0.0043070644497691934,
      "grad_norm": 2.490511894226074,
      "learning_rate": 1.9731262132804274e-05,
      "loss": 0.8455,
      "step": 324
    },
    {
      "epoch": 0.004320357858564777,
      "grad_norm": 2.235755443572998,
      "learning_rate": 1.9441314295107537e-05,
      "loss": 0.7679,
      "step": 325
    },
    {
      "epoch": 0.004333651267360361,
      "grad_norm": 2.30128812789917,
      "learning_rate": 1.9152997613697183e-05,
      "loss": 0.7796,
      "step": 326
    },
    {
      "epoch": 0.004346944676155945,
      "grad_norm": 2.4752769470214844,
      "learning_rate": 1.8866327478291546e-05,
      "loss": 0.8884,
      "step": 327
    },
    {
      "epoch": 0.004360238084951529,
      "grad_norm": 2.7887279987335205,
      "learning_rate": 1.8581319190720035e-05,
      "loss": 0.7255,
      "step": 328
    },
    {
      "epoch": 0.004373531493747113,
      "grad_norm": 2.6239068508148193,
      "learning_rate": 1.8297987964106115e-05,
      "loss": 0.8943,
      "step": 329
    },
    {
      "epoch": 0.004386824902542697,
      "grad_norm": 2.581753969192505,
      "learning_rate": 1.801634892205545e-05,
      "loss": 0.7536,
      "step": 330
    },
    {
      "epoch": 0.00440011831133828,
      "grad_norm": 2.731560468673706,
      "learning_rate": 1.7736417097848506e-05,
      "loss": 0.8233,
      "step": 331
    },
    {
      "epoch": 0.004413411720133865,
      "grad_norm": 2.825575351715088,
      "learning_rate": 1.7458207433638223e-05,
      "loss": 0.8542,
      "step": 332
    },
    {
      "epoch": 0.004426705128929448,
      "grad_norm": 3.0620334148406982,
      "learning_rate": 1.718173477965236e-05,
      "loss": 0.8842,
      "step": 333
    },
    {
      "epoch": 0.004439998537725033,
      "grad_norm": 3.0803141593933105,
      "learning_rate": 1.6907013893400837e-05,
      "loss": 0.9173,
      "step": 334
    },
    {
      "epoch": 0.004453291946520616,
      "grad_norm": 2.6169607639312744,
      "learning_rate": 1.6634059438888033e-05,
      "loss": 0.9778,
      "step": 335
    },
    {
      "epoch": 0.0044665853553162,
      "grad_norm": 3.0732738971710205,
      "learning_rate": 1.636288598583e-05,
      "loss": 0.6455,
      "step": 336
    },
    {
      "epoch": 0.004479878764111784,
      "grad_norm": 2.649909496307373,
      "learning_rate": 1.6093508008876857e-05,
      "loss": 0.7877,
      "step": 337
    },
    {
      "epoch": 0.004493172172907368,
      "grad_norm": 3.0148892402648926,
      "learning_rate": 1.5825939886840037e-05,
      "loss": 0.8035,
      "step": 338
    },
    {
      "epoch": 0.004506465581702952,
      "grad_norm": 3.5838184356689453,
      "learning_rate": 1.5560195901924894e-05,
      "loss": 0.7526,
      "step": 339
    },
    {
      "epoch": 0.004519758990498536,
      "grad_norm": 3.0014920234680176,
      "learning_rate": 1.5296290238968303e-05,
      "loss": 0.8166,
      "step": 340
    },
    {
      "epoch": 0.00453305239929412,
      "grad_norm": 3.3715362548828125,
      "learning_rate": 1.50342369846815e-05,
      "loss": 0.8953,
      "step": 341
    },
    {
      "epoch": 0.004546345808089704,
      "grad_norm": 2.9526143074035645,
      "learning_rate": 1.4774050126898164e-05,
      "loss": 0.7074,
      "step": 342
    },
    {
      "epoch": 0.004559639216885288,
      "grad_norm": 3.7083990573883057,
      "learning_rate": 1.451574355382776e-05,
      "loss": 0.8644,
      "step": 343
    },
    {
      "epoch": 0.004572932625680872,
      "grad_norm": 4.097346782684326,
      "learning_rate": 1.425933105331429e-05,
      "loss": 0.787,
      "step": 344
    },
    {
      "epoch": 0.004586226034476456,
      "grad_norm": 3.9541990756988525,
      "learning_rate": 1.4004826312100216e-05,
      "loss": 0.9168,
      "step": 345
    },
    {
      "epoch": 0.0045995194432720396,
      "grad_norm": 4.238726615905762,
      "learning_rate": 1.3752242915095992e-05,
      "loss": 0.8646,
      "step": 346
    },
    {
      "epoch": 0.004612812852067624,
      "grad_norm": 4.998000621795654,
      "learning_rate": 1.3501594344654884e-05,
      "loss": 0.8752,
      "step": 347
    },
    {
      "epoch": 0.004626106260863207,
      "grad_norm": 4.411581516265869,
      "learning_rate": 1.3252893979853304e-05,
      "loss": 1.0685,
      "step": 348
    },
    {
      "epoch": 0.004639399669658792,
      "grad_norm": 5.595761299133301,
      "learning_rate": 1.3006155095776707e-05,
      "loss": 0.6555,
      "step": 349
    },
    {
      "epoch": 0.004652693078454375,
      "grad_norm": 6.310719013214111,
      "learning_rate": 1.2761390862810907e-05,
      "loss": 0.902,
      "step": 350
    },
    {
      "epoch": 0.00466598648724996,
      "grad_norm": 1.1895419359207153,
      "learning_rate": 1.2518614345939212e-05,
      "loss": 1.0112,
      "step": 351
    },
    {
      "epoch": 0.004679279896045543,
      "grad_norm": 1.4201432466506958,
      "learning_rate": 1.227783850404487e-05,
      "loss": 0.9783,
      "step": 352
    },
    {
      "epoch": 0.004692573304841127,
      "grad_norm": 1.685895323753357,
      "learning_rate": 1.2039076189219517e-05,
      "loss": 1.0757,
      "step": 353
    },
    {
      "epoch": 0.004705866713636711,
      "grad_norm": 1.7650642395019531,
      "learning_rate": 1.1802340146077045e-05,
      "loss": 0.8879,
      "step": 354
    },
    {
      "epoch": 0.004719160122432295,
      "grad_norm": 1.7274045944213867,
      "learning_rate": 1.1567643011073392e-05,
      "loss": 0.8959,
      "step": 355
    },
    {
      "epoch": 0.004732453531227879,
      "grad_norm": 2.2640771865844727,
      "learning_rate": 1.1334997311832002e-05,
      "loss": 0.8883,
      "step": 356
    },
    {
      "epoch": 0.004745746940023463,
      "grad_norm": 1.8365882635116577,
      "learning_rate": 1.1104415466475087e-05,
      "loss": 0.9858,
      "step": 357
    },
    {
      "epoch": 0.0047590403488190465,
      "grad_norm": 2.0218539237976074,
      "learning_rate": 1.0875909782960886e-05,
      "loss": 0.9811,
      "step": 358
    },
    {
      "epoch": 0.004772333757614631,
      "grad_norm": 1.7274595499038696,
      "learning_rate": 1.0649492458426564e-05,
      "loss": 1.0713,
      "step": 359
    },
    {
      "epoch": 0.004785627166410214,
      "grad_norm": 1.639711856842041,
      "learning_rate": 1.0425175578537299e-05,
      "loss": 0.9316,
      "step": 360
    },
    {
      "epoch": 0.004798920575205799,
      "grad_norm": 1.939886450767517,
      "learning_rate": 1.020297111684101e-05,
      "loss": 0.8692,
      "step": 361
    },
    {
      "epoch": 0.004812213984001382,
      "grad_norm": 2.1711244583129883,
      "learning_rate": 9.98289093412938e-06,
      "loss": 0.867,
      "step": 362
    },
    {
      "epoch": 0.0048255073927969665,
      "grad_norm": 1.8200219869613647,
      "learning_rate": 9.764946777804646e-06,
      "loss": 0.8237,
      "step": 363
    },
    {
      "epoch": 0.00483880080159255,
      "grad_norm": 2.027531623840332,
      "learning_rate": 9.549150281252633e-06,
      "loss": 0.8075,
      "step": 364
    },
    {
      "epoch": 0.004852094210388134,
      "grad_norm": 2.07029128074646,
      "learning_rate": 9.335512963221732e-06,
      "loss": 0.8874,
      "step": 365
    },
    {
      "epoch": 0.004865387619183718,
      "grad_norm": 2.0369820594787598,
      "learning_rate": 9.124046227208082e-06,
      "loss": 0.8996,
      "step": 366
    },
    {
      "epoch": 0.004878681027979302,
      "grad_norm": 2.115910530090332,
      "learning_rate": 8.914761360846869e-06,
      "loss": 0.8661,
      "step": 367
    },
    {
      "epoch": 0.004891974436774886,
      "grad_norm": 2.021585702896118,
      "learning_rate": 8.707669535309793e-06,
      "loss": 0.8805,
      "step": 368
    },
    {
      "epoch": 0.00490526784557047,
      "grad_norm": 2.294255018234253,
      "learning_rate": 8.502781804708826e-06,
      "loss": 0.8364,
      "step": 369
    },
    {
      "epoch": 0.004918561254366054,
      "grad_norm": 2.2873215675354004,
      "learning_rate": 8.30010910550611e-06,
      "loss": 0.8194,
      "step": 370
    },
    {
      "epoch": 0.004931854663161638,
      "grad_norm": 2.252678871154785,
      "learning_rate": 8.09966225593024e-06,
      "loss": 0.9445,
      "step": 371
    },
    {
      "epoch": 0.004945148071957222,
      "grad_norm": 2.10272216796875,
      "learning_rate": 7.901451955398792e-06,
      "loss": 0.8814,
      "step": 372
    },
    {
      "epoch": 0.004958441480752806,
      "grad_norm": 2.401695489883423,
      "learning_rate": 7.705488783947202e-06,
      "loss": 0.8314,
      "step": 373
    },
    {
      "epoch": 0.00497173488954839,
      "grad_norm": 2.3294260501861572,
      "learning_rate": 7.511783201664052e-06,
      "loss": 0.8886,
      "step": 374
    },
    {
      "epoch": 0.0049850282983439735,
      "grad_norm": 2.6172540187835693,
      "learning_rate": 7.320345548132679e-06,
      "loss": 0.9066,
      "step": 375
    },
    {
      "epoch": 0.004998321707139558,
      "grad_norm": 2.2096974849700928,
      "learning_rate": 7.131186041879357e-06,
      "loss": 0.7404,
      "step": 376
    },
    {
      "epoch": 0.005011615115935141,
      "grad_norm": 2.5697643756866455,
      "learning_rate": 6.944314779827749e-06,
      "loss": 0.8426,
      "step": 377
    },
    {
      "epoch": 0.005024908524730726,
      "grad_norm": 2.6028385162353516,
      "learning_rate": 6.759741736760061e-06,
      "loss": 0.9637,
      "step": 378
    },
    {
      "epoch": 0.005038201933526309,
      "grad_norm": 2.43548321723938,
      "learning_rate": 6.577476764784546e-06,
      "loss": 0.9475,
      "step": 379
    },
    {
      "epoch": 0.0050514953423218935,
      "grad_norm": 2.482872486114502,
      "learning_rate": 6.397529592809614e-06,
      "loss": 0.8607,
      "step": 380
    },
    {
      "epoch": 0.005064788751117477,
      "grad_norm": 2.3237485885620117,
      "learning_rate": 6.219909826024589e-06,
      "loss": 0.8284,
      "step": 381
    },
    {
      "epoch": 0.005078082159913061,
      "grad_norm": 2.8761892318725586,
      "learning_rate": 6.0446269453868945e-06,
      "loss": 0.7335,
      "step": 382
    },
    {
      "epoch": 0.005091375568708645,
      "grad_norm": 3.1307175159454346,
      "learning_rate": 5.871690307116107e-06,
      "loss": 0.9296,
      "step": 383
    },
    {
      "epoch": 0.005104668977504229,
      "grad_norm": 3.0153515338897705,
      "learning_rate": 5.701109142194422e-06,
      "loss": 0.7582,
      "step": 384
    },
    {
      "epoch": 0.005117962386299813,
      "grad_norm": 2.8852033615112305,
      "learning_rate": 5.532892555874059e-06,
      "loss": 0.8387,
      "step": 385
    },
    {
      "epoch": 0.005131255795095397,
      "grad_norm": 2.806442975997925,
      "learning_rate": 5.3670495271910925e-06,
      "loss": 0.6859,
      "step": 386
    },
    {
      "epoch": 0.0051445492038909805,
      "grad_norm": 2.6794891357421875,
      "learning_rate": 5.203588908486279e-06,
      "loss": 0.8166,
      "step": 387
    },
    {
      "epoch": 0.005157842612686565,
      "grad_norm": 2.7952828407287598,
      "learning_rate": 5.042519424932513e-06,
      "loss": 0.7706,
      "step": 388
    },
    {
      "epoch": 0.005171136021482148,
      "grad_norm": 2.773052930831909,
      "learning_rate": 4.883849674069058e-06,
      "loss": 0.8396,
      "step": 389
    },
    {
      "epoch": 0.005184429430277733,
      "grad_norm": 2.9820058345794678,
      "learning_rate": 4.727588125342669e-06,
      "loss": 0.7257,
      "step": 390
    },
    {
      "epoch": 0.005197722839073316,
      "grad_norm": 2.9437365531921387,
      "learning_rate": 4.573743119655516e-06,
      "loss": 0.7987,
      "step": 391
    },
    {
      "epoch": 0.0052110162478689005,
      "grad_norm": 3.896148920059204,
      "learning_rate": 4.422322868919937e-06,
      "loss": 0.7962,
      "step": 392
    },
    {
      "epoch": 0.005224309656664484,
      "grad_norm": 3.0379815101623535,
      "learning_rate": 4.273335455620097e-06,
      "loss": 0.6742,
      "step": 393
    },
    {
      "epoch": 0.005237603065460068,
      "grad_norm": 3.5610365867614746,
      "learning_rate": 4.126788832380629e-06,
      "loss": 0.8542,
      "step": 394
    },
    {
      "epoch": 0.005250896474255652,
      "grad_norm": 3.813161849975586,
      "learning_rate": 3.982690821542035e-06,
      "loss": 0.7634,
      "step": 395
    },
    {
      "epoch": 0.005264189883051236,
      "grad_norm": 4.287951946258545,
      "learning_rate": 3.8410491147432395e-06,
      "loss": 0.8481,
      "step": 396
    },
    {
      "epoch": 0.00527748329184682,
      "grad_norm": 3.525927782058716,
      "learning_rate": 3.7018712725109926e-06,
      "loss": 0.6304,
      "step": 397
    },
    {
      "epoch": 0.005290776700642404,
      "grad_norm": 4.333404541015625,
      "learning_rate": 3.5651647238562904e-06,
      "loss": 0.8299,
      "step": 398
    },
    {
      "epoch": 0.005304070109437988,
      "grad_norm": 5.827525615692139,
      "learning_rate": 3.430936765877857e-06,
      "loss": 0.9442,
      "step": 399
    },
    {
      "epoch": 0.005317363518233572,
      "grad_norm": 7.1007280349731445,
      "learning_rate": 3.299194563372604e-06,
      "loss": 0.6819,
      "step": 400
    },
    {
      "epoch": 0.005330656927029156,
      "grad_norm": 1.2109785079956055,
      "learning_rate": 3.1699451484532463e-06,
      "loss": 1.0173,
      "step": 401
    },
    {
      "epoch": 0.00534395033582474,
      "grad_norm": 1.6271941661834717,
      "learning_rate": 3.0431954201728784e-06,
      "loss": 0.9225,
      "step": 402
    },
    {
      "epoch": 0.005357243744620324,
      "grad_norm": 1.5182850360870361,
      "learning_rate": 2.9189521441567726e-06,
      "loss": 0.8122,
      "step": 403
    },
    {
      "epoch": 0.0053705371534159075,
      "grad_norm": 1.6241300106048584,
      "learning_rate": 2.797221952241219e-06,
      "loss": 0.6472,
      "step": 404
    },
    {
      "epoch": 0.005383830562211492,
      "grad_norm": 2.0764918327331543,
      "learning_rate": 2.6780113421195298e-06,
      "loss": 0.806,
      "step": 405
    },
    {
      "epoch": 0.005397123971007075,
      "grad_norm": 1.7580714225769043,
      "learning_rate": 2.561326676995218e-06,
      "loss": 0.7895,
      "step": 406
    },
    {
      "epoch": 0.00541041737980266,
      "grad_norm": 1.7050628662109375,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 0.9054,
      "step": 407
    },
    {
      "epoch": 0.005423710788598243,
      "grad_norm": 1.7846729755401611,
      "learning_rate": 2.3355599600729915e-06,
      "loss": 0.8994,
      "step": 408
    },
    {
      "epoch": 0.0054370041973938275,
      "grad_norm": 2.0640039443969727,
      "learning_rate": 2.2264899592121744e-06,
      "loss": 0.9436,
      "step": 409
    },
    {
      "epoch": 0.005450297606189411,
      "grad_norm": 1.7736494541168213,
      "learning_rate": 2.1199700045797077e-06,
      "loss": 0.8843,
      "step": 410
    },
    {
      "epoch": 0.005463591014984995,
      "grad_norm": 1.889944076538086,
      "learning_rate": 2.0160057819794466e-06,
      "loss": 0.8959,
      "step": 411
    },
    {
      "epoch": 0.005476884423780579,
      "grad_norm": 2.189314603805542,
      "learning_rate": 1.9146028407958484e-06,
      "loss": 1.0046,
      "step": 412
    },
    {
      "epoch": 0.005490177832576163,
      "grad_norm": 1.8977439403533936,
      "learning_rate": 1.8157665936977263e-06,
      "loss": 0.8848,
      "step": 413
    },
    {
      "epoch": 0.005503471241371747,
      "grad_norm": 2.0931496620178223,
      "learning_rate": 1.7195023163493252e-06,
      "loss": 1.0084,
      "step": 414
    },
    {
      "epoch": 0.005516764650167331,
      "grad_norm": 2.239429235458374,
      "learning_rate": 1.6258151471287396e-06,
      "loss": 0.8895,
      "step": 415
    },
    {
      "epoch": 0.0055300580589629144,
      "grad_norm": 2.0651838779449463,
      "learning_rate": 1.5347100868536246e-06,
      "loss": 0.8107,
      "step": 416
    },
    {
      "epoch": 0.005543351467758499,
      "grad_norm": 2.068511724472046,
      "learning_rate": 1.4461919985142735e-06,
      "loss": 0.7925,
      "step": 417
    },
    {
      "epoch": 0.005556644876554082,
      "grad_norm": 2.0597782135009766,
      "learning_rate": 1.3602656070140275e-06,
      "loss": 0.827,
      "step": 418
    },
    {
      "epoch": 0.005569938285349667,
      "grad_norm": 2.230896234512329,
      "learning_rate": 1.27693549891707e-06,
      "loss": 0.7999,
      "step": 419
    },
    {
      "epoch": 0.00558323169414525,
      "grad_norm": 2.3435306549072266,
      "learning_rate": 1.196206122203647e-06,
      "loss": 0.9503,
      "step": 420
    },
    {
      "epoch": 0.0055965251029408344,
      "grad_norm": 2.1370625495910645,
      "learning_rate": 1.1180817860325599e-06,
      "loss": 0.862,
      "step": 421
    },
    {
      "epoch": 0.005609818511736418,
      "grad_norm": 2.205033540725708,
      "learning_rate": 1.0425666605112517e-06,
      "loss": 0.8852,
      "step": 422
    },
    {
      "epoch": 0.005623111920532002,
      "grad_norm": 2.5713577270507812,
      "learning_rate": 9.696647764731337e-07,
      "loss": 0.8009,
      "step": 423
    },
    {
      "epoch": 0.005636405329327586,
      "grad_norm": 3.0529072284698486,
      "learning_rate": 8.993800252624862e-07,
      "loss": 0.9193,
      "step": 424
    },
    {
      "epoch": 0.00564969873812317,
      "grad_norm": 2.8869714736938477,
      "learning_rate": 8.317161585266964e-07,
      "loss": 0.9436,
      "step": 425
    },
    {
      "epoch": 0.005662992146918754,
      "grad_norm": 2.595672845840454,
      "learning_rate": 7.666767880160464e-07,
      "loss": 0.8529,
      "step": 426
    },
    {
      "epoch": 0.005676285555714338,
      "grad_norm": 2.3774523735046387,
      "learning_rate": 7.042653853909064e-07,
      "loss": 0.9693,
      "step": 427
    },
    {
      "epoch": 0.005689578964509922,
      "grad_norm": 2.5005154609680176,
      "learning_rate": 6.444852820364222e-07,
      "loss": 0.7297,
      "step": 428
    },
    {
      "epoch": 0.005702872373305506,
      "grad_norm": 2.437258243560791,
      "learning_rate": 5.87339668884701e-07,
      "loss": 0.7768,
      "step": 429
    },
    {
      "epoch": 0.00571616578210109,
      "grad_norm": 2.3614542484283447,
      "learning_rate": 5.328315962444874e-07,
      "loss": 0.7939,
      "step": 430
    },
    {
      "epoch": 0.005729459190896674,
      "grad_norm": 2.5789074897766113,
      "learning_rate": 4.809639736383431e-07,
      "loss": 0.7969,
      "step": 431
    },
    {
      "epoch": 0.005742752599692258,
      "grad_norm": 3.0158462524414062,
      "learning_rate": 4.317395696473214e-07,
      "loss": 0.9557,
      "step": 432
    },
    {
      "epoch": 0.005756046008487841,
      "grad_norm": 2.9326164722442627,
      "learning_rate": 3.851610117632354e-07,
      "loss": 0.877,
      "step": 433
    },
    {
      "epoch": 0.005769339417283426,
      "grad_norm": 3.167335033416748,
      "learning_rate": 3.4123078624834216e-07,
      "loss": 0.7724,
      "step": 434
    },
    {
      "epoch": 0.005782632826079009,
      "grad_norm": 2.828343152999878,
      "learning_rate": 2.9995123800270476e-07,
      "loss": 0.7906,
      "step": 435
    },
    {
      "epoch": 0.005795926234874594,
      "grad_norm": 2.650979518890381,
      "learning_rate": 2.613245704389644e-07,
      "loss": 0.8052,
      "step": 436
    },
    {
      "epoch": 0.005809219643670177,
      "grad_norm": 2.7367007732391357,
      "learning_rate": 2.2535284536476242e-07,
      "loss": 0.7699,
      "step": 437
    },
    {
      "epoch": 0.005822513052465761,
      "grad_norm": 2.9800891876220703,
      "learning_rate": 1.920379828726726e-07,
      "loss": 0.6601,
      "step": 438
    },
    {
      "epoch": 0.005835806461261345,
      "grad_norm": 3.0351688861846924,
      "learning_rate": 1.6138176123770554e-07,
      "loss": 0.692,
      "step": 439
    },
    {
      "epoch": 0.005849099870056929,
      "grad_norm": 3.4861412048339844,
      "learning_rate": 1.333858168224178e-07,
      "loss": 0.7502,
      "step": 440
    },
    {
      "epoch": 0.005862393278852513,
      "grad_norm": 3.223451614379883,
      "learning_rate": 1.0805164398952072e-07,
      "loss": 0.7937,
      "step": 441
    },
    {
      "epoch": 0.005875686687648097,
      "grad_norm": 3.0218589305877686,
      "learning_rate": 8.53805950221498e-08,
      "loss": 0.8031,
      "step": 442
    },
    {
      "epoch": 0.0058889800964436806,
      "grad_norm": 3.249140501022339,
      "learning_rate": 6.537388005167233e-08,
      "loss": 0.6787,
      "step": 443
    },
    {
      "epoch": 0.005902273505239265,
      "grad_norm": 3.6842355728149414,
      "learning_rate": 4.8032566993089225e-08,
      "loss": 0.839,
      "step": 444
    },
    {
      "epoch": 0.005915566914034848,
      "grad_norm": 3.5570058822631836,
      "learning_rate": 3.3357581488030475e-08,
      "loss": 0.6737,
      "step": 445
    },
    {
      "epoch": 0.005928860322830433,
      "grad_norm": 3.5006725788116455,
      "learning_rate": 2.134970685536697e-08,
      "loss": 0.7928,
      "step": 446
    },
    {
      "epoch": 0.005942153731626016,
      "grad_norm": 4.718527317047119,
      "learning_rate": 1.200958404936059e-08,
      "loss": 0.8572,
      "step": 447
    },
    {
      "epoch": 0.0059554471404216006,
      "grad_norm": 4.4138689041137695,
      "learning_rate": 5.337711625497121e-09,
      "loss": 0.9534,
      "step": 448
    },
    {
      "epoch": 0.005968740549217184,
      "grad_norm": 5.295405387878418,
      "learning_rate": 1.3344457138297906e-09,
      "loss": 0.7238,
      "step": 449
    },
    {
      "epoch": 0.005982033958012768,
      "grad_norm": 7.403181552886963,
      "learning_rate": 0.0,
      "loss": 0.9382,
      "step": 450
    },
    {
      "epoch": 0.005982033958012768,
      "eval_loss": 0.8241426348686218,
      "eval_runtime": 3479.6362,
      "eval_samples_per_second": 36.411,
      "eval_steps_per_second": 9.103,
      "step": 450
    }
  ],
  "logging_steps": 1,
  "max_steps": 450,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 150,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 2,
        "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": 1.4139709881266995e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}