{
  "best_metric": 0.8085561394691467,
  "best_model_checkpoint": "miner_id_24/checkpoint-500",
  "epoch": 0.2516039753428104,
  "eval_steps": 100,
  "global_step": 500,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0005032079506856208,
      "grad_norm": 0.38311243057250977,
      "learning_rate": 5e-06,
      "loss": 0.5885,
      "step": 1
    },
    {
      "epoch": 0.0005032079506856208,
      "eval_loss": 1.0956110954284668,
      "eval_runtime": 315.5564,
      "eval_samples_per_second": 10.607,
      "eval_steps_per_second": 2.652,
      "step": 1
    },
    {
      "epoch": 0.0010064159013712416,
      "grad_norm": 0.5078917741775513,
      "learning_rate": 1e-05,
      "loss": 0.7928,
      "step": 2
    },
    {
      "epoch": 0.0015096238520568624,
      "grad_norm": 0.3885859251022339,
      "learning_rate": 1.5e-05,
      "loss": 0.727,
      "step": 3
    },
    {
      "epoch": 0.0020128318027424832,
      "grad_norm": 0.3169471323490143,
      "learning_rate": 2e-05,
      "loss": 0.6431,
      "step": 4
    },
    {
      "epoch": 0.002516039753428104,
      "grad_norm": 0.3564390540122986,
      "learning_rate": 2.5e-05,
      "loss": 0.8972,
      "step": 5
    },
    {
      "epoch": 0.003019247704113725,
      "grad_norm": 0.4155474007129669,
      "learning_rate": 3e-05,
      "loss": 0.7988,
      "step": 6
    },
    {
      "epoch": 0.0035224556547993457,
      "grad_norm": 0.3342589735984802,
      "learning_rate": 3.5e-05,
      "loss": 0.7426,
      "step": 7
    },
    {
      "epoch": 0.0040256636054849665,
      "grad_norm": 0.3079495429992676,
      "learning_rate": 4e-05,
      "loss": 0.6917,
      "step": 8
    },
    {
      "epoch": 0.004528871556170587,
      "grad_norm": 0.3665371537208557,
      "learning_rate": 4.5e-05,
      "loss": 0.7703,
      "step": 9
    },
    {
      "epoch": 0.005032079506856208,
      "grad_norm": 0.41909274458885193,
      "learning_rate": 5e-05,
      "loss": 0.8054,
      "step": 10
    },
    {
      "epoch": 0.005535287457541829,
      "grad_norm": 0.43584853410720825,
      "learning_rate": 5.500000000000001e-05,
      "loss": 0.8718,
      "step": 11
    },
    {
      "epoch": 0.00603849540822745,
      "grad_norm": 0.41220253705978394,
      "learning_rate": 6e-05,
      "loss": 0.8652,
      "step": 12
    },
    {
      "epoch": 0.0065417033589130705,
      "grad_norm": 0.40155696868896484,
      "learning_rate": 6.500000000000001e-05,
      "loss": 0.8386,
      "step": 13
    },
    {
      "epoch": 0.007044911309598691,
      "grad_norm": 0.4041687250137329,
      "learning_rate": 7e-05,
      "loss": 0.9532,
      "step": 14
    },
    {
      "epoch": 0.007548119260284312,
      "grad_norm": 0.398960679769516,
      "learning_rate": 7.500000000000001e-05,
      "loss": 0.9645,
      "step": 15
    },
    {
      "epoch": 0.008051327210969933,
      "grad_norm": 0.4464486241340637,
      "learning_rate": 8e-05,
      "loss": 0.9238,
      "step": 16
    },
    {
      "epoch": 0.008554535161655554,
      "grad_norm": 0.4189698100090027,
      "learning_rate": 8.5e-05,
      "loss": 0.8657,
      "step": 17
    },
    {
      "epoch": 0.009057743112341175,
      "grad_norm": 0.44621121883392334,
      "learning_rate": 9e-05,
      "loss": 0.8711,
      "step": 18
    },
    {
      "epoch": 0.009560951063026795,
      "grad_norm": 0.41292840242385864,
      "learning_rate": 9.5e-05,
      "loss": 0.8147,
      "step": 19
    },
    {
      "epoch": 0.010064159013712416,
      "grad_norm": 0.4491482079029083,
      "learning_rate": 0.0001,
      "loss": 0.8187,
      "step": 20
    },
    {
      "epoch": 0.010567366964398037,
      "grad_norm": 0.42417269945144653,
      "learning_rate": 9.999892908320647e-05,
      "loss": 0.9176,
      "step": 21
    },
    {
      "epoch": 0.011070574915083658,
      "grad_norm": 0.4301187992095947,
      "learning_rate": 9.999571637870036e-05,
      "loss": 0.9259,
      "step": 22
    },
    {
      "epoch": 0.011573782865769279,
      "grad_norm": 0.4333283305168152,
      "learning_rate": 9.999036202410325e-05,
      "loss": 0.9043,
      "step": 23
    },
    {
      "epoch": 0.0120769908164549,
      "grad_norm": 0.4324922263622284,
      "learning_rate": 9.998286624877786e-05,
      "loss": 0.8326,
      "step": 24
    },
    {
      "epoch": 0.01258019876714052,
      "grad_norm": 0.4286755919456482,
      "learning_rate": 9.997322937381829e-05,
      "loss": 0.9593,
      "step": 25
    },
    {
      "epoch": 0.013083406717826141,
      "grad_norm": 0.46805518865585327,
      "learning_rate": 9.996145181203615e-05,
      "loss": 0.8725,
      "step": 26
    },
    {
      "epoch": 0.013586614668511762,
      "grad_norm": 0.43730825185775757,
      "learning_rate": 9.994753406794301e-05,
      "loss": 0.8642,
      "step": 27
    },
    {
      "epoch": 0.014089822619197383,
      "grad_norm": 0.45640259981155396,
      "learning_rate": 9.99314767377287e-05,
      "loss": 0.8955,
      "step": 28
    },
    {
      "epoch": 0.014593030569883004,
      "grad_norm": 0.4871976375579834,
      "learning_rate": 9.991328050923581e-05,
      "loss": 0.9745,
      "step": 29
    },
    {
      "epoch": 0.015096238520568624,
      "grad_norm": 0.4469471275806427,
      "learning_rate": 9.989294616193017e-05,
      "loss": 0.8392,
      "step": 30
    },
    {
      "epoch": 0.015599446471254245,
      "grad_norm": 0.5271027088165283,
      "learning_rate": 9.98704745668676e-05,
      "loss": 0.9445,
      "step": 31
    },
    {
      "epoch": 0.016102654421939866,
      "grad_norm": 0.5062204599380493,
      "learning_rate": 9.98458666866564e-05,
      "loss": 0.8554,
      "step": 32
    },
    {
      "epoch": 0.01660586237262549,
      "grad_norm": 0.5133852958679199,
      "learning_rate": 9.981912357541627e-05,
      "loss": 0.8516,
      "step": 33
    },
    {
      "epoch": 0.017109070323311108,
      "grad_norm": 0.5246889591217041,
      "learning_rate": 9.97902463787331e-05,
      "loss": 0.7361,
      "step": 34
    },
    {
      "epoch": 0.01761227827399673,
      "grad_norm": 0.5814148187637329,
      "learning_rate": 9.975923633360985e-05,
      "loss": 0.855,
      "step": 35
    },
    {
      "epoch": 0.01811548622468235,
      "grad_norm": 0.6536831855773926,
      "learning_rate": 9.972609476841367e-05,
      "loss": 0.9576,
      "step": 36
    },
    {
      "epoch": 0.018618694175367972,
      "grad_norm": 0.602803647518158,
      "learning_rate": 9.969082310281891e-05,
      "loss": 0.8832,
      "step": 37
    },
    {
      "epoch": 0.01912190212605359,
      "grad_norm": 0.6678468585014343,
      "learning_rate": 9.965342284774632e-05,
      "loss": 1.0131,
      "step": 38
    },
    {
      "epoch": 0.019625110076739213,
      "grad_norm": 0.6718692183494568,
      "learning_rate": 9.961389560529836e-05,
      "loss": 1.0132,
      "step": 39
    },
    {
      "epoch": 0.020128318027424832,
      "grad_norm": 0.7489075660705566,
      "learning_rate": 9.957224306869053e-05,
      "loss": 1.0004,
      "step": 40
    },
    {
      "epoch": 0.020631525978110455,
      "grad_norm": 0.6968185901641846,
      "learning_rate": 9.952846702217886e-05,
      "loss": 0.9475,
      "step": 41
    },
    {
      "epoch": 0.021134733928796074,
      "grad_norm": 0.6813265681266785,
      "learning_rate": 9.948256934098352e-05,
      "loss": 0.8116,
      "step": 42
    },
    {
      "epoch": 0.021637941879481697,
      "grad_norm": 0.9425336122512817,
      "learning_rate": 9.943455199120837e-05,
      "loss": 1.1965,
      "step": 43
    },
    {
      "epoch": 0.022141149830167316,
      "grad_norm": 1.0524532794952393,
      "learning_rate": 9.938441702975689e-05,
      "loss": 1.1946,
      "step": 44
    },
    {
      "epoch": 0.022644357780852938,
      "grad_norm": 1.0975672006607056,
      "learning_rate": 9.933216660424395e-05,
      "loss": 1.0296,
      "step": 45
    },
    {
      "epoch": 0.023147565731538557,
      "grad_norm": 1.0887701511383057,
      "learning_rate": 9.927780295290389e-05,
      "loss": 1.2867,
      "step": 46
    },
    {
      "epoch": 0.02365077368222418,
      "grad_norm": 1.040777325630188,
      "learning_rate": 9.922132840449459e-05,
      "loss": 1.1358,
      "step": 47
    },
    {
      "epoch": 0.0241539816329098,
      "grad_norm": 1.1316983699798584,
      "learning_rate": 9.916274537819775e-05,
      "loss": 0.9627,
      "step": 48
    },
    {
      "epoch": 0.02465718958359542,
      "grad_norm": 1.6163930892944336,
      "learning_rate": 9.91020563835152e-05,
      "loss": 1.1563,
      "step": 49
    },
    {
      "epoch": 0.02516039753428104,
      "grad_norm": 1.776786208152771,
      "learning_rate": 9.903926402016153e-05,
      "loss": 1.0637,
      "step": 50
    },
    {
      "epoch": 0.025663605484966663,
      "grad_norm": 0.5479099750518799,
      "learning_rate": 9.897437097795257e-05,
      "loss": 0.642,
      "step": 51
    },
    {
      "epoch": 0.026166813435652282,
      "grad_norm": 0.6513462066650391,
      "learning_rate": 9.890738003669029e-05,
      "loss": 0.5686,
      "step": 52
    },
    {
      "epoch": 0.026670021386337905,
      "grad_norm": 0.445416122674942,
      "learning_rate": 9.883829406604363e-05,
      "loss": 0.674,
      "step": 53
    },
    {
      "epoch": 0.027173229337023524,
      "grad_norm": 0.43422967195510864,
      "learning_rate": 9.876711602542563e-05,
      "loss": 0.7273,
      "step": 54
    },
    {
      "epoch": 0.027676437287709146,
      "grad_norm": 0.36674436926841736,
      "learning_rate": 9.869384896386668e-05,
      "loss": 0.8264,
      "step": 55
    },
    {
      "epoch": 0.028179645238394765,
      "grad_norm": 0.2828488349914551,
      "learning_rate": 9.861849601988383e-05,
      "loss": 0.6785,
      "step": 56
    },
    {
      "epoch": 0.028682853189080388,
      "grad_norm": 0.3080351948738098,
      "learning_rate": 9.854106042134641e-05,
      "loss": 0.7158,
      "step": 57
    },
    {
      "epoch": 0.029186061139766007,
      "grad_norm": 0.3878094553947449,
      "learning_rate": 9.846154548533773e-05,
      "loss": 0.8812,
      "step": 58
    },
    {
      "epoch": 0.02968926909045163,
      "grad_norm": 0.4083269536495209,
      "learning_rate": 9.837995461801299e-05,
      "loss": 0.8511,
      "step": 59
    },
    {
      "epoch": 0.03019247704113725,
      "grad_norm": 0.3542426526546478,
      "learning_rate": 9.829629131445342e-05,
      "loss": 0.7856,
      "step": 60
    },
    {
      "epoch": 0.03069568499182287,
      "grad_norm": 0.36342018842697144,
      "learning_rate": 9.821055915851647e-05,
      "loss": 0.8066,
      "step": 61
    },
    {
      "epoch": 0.03119889294250849,
      "grad_norm": 0.3394208550453186,
      "learning_rate": 9.812276182268236e-05,
      "loss": 0.87,
      "step": 62
    },
    {
      "epoch": 0.03170210089319411,
      "grad_norm": 0.3435419797897339,
      "learning_rate": 9.803290306789676e-05,
      "loss": 0.7979,
      "step": 63
    },
    {
      "epoch": 0.03220530884387973,
      "grad_norm": 0.34052741527557373,
      "learning_rate": 9.794098674340965e-05,
      "loss": 0.7838,
      "step": 64
    },
    {
      "epoch": 0.03270851679456535,
      "grad_norm": 0.40174680948257446,
      "learning_rate": 9.784701678661045e-05,
      "loss": 0.9775,
      "step": 65
    },
    {
      "epoch": 0.03321172474525098,
      "grad_norm": 0.33692508935928345,
      "learning_rate": 9.775099722285935e-05,
      "loss": 0.8391,
      "step": 66
    },
    {
      "epoch": 0.033714932695936596,
      "grad_norm": 0.43034443259239197,
      "learning_rate": 9.765293216531486e-05,
      "loss": 0.8534,
      "step": 67
    },
    {
      "epoch": 0.034218140646622215,
      "grad_norm": 0.379113107919693,
      "learning_rate": 9.755282581475769e-05,
      "loss": 0.8039,
      "step": 68
    },
    {
      "epoch": 0.034721348597307834,
      "grad_norm": 0.38086801767349243,
      "learning_rate": 9.74506824594107e-05,
      "loss": 0.8454,
      "step": 69
    },
    {
      "epoch": 0.03522455654799346,
      "grad_norm": 0.40294283628463745,
      "learning_rate": 9.73465064747553e-05,
      "loss": 0.9549,
      "step": 70
    },
    {
      "epoch": 0.03572776449867908,
      "grad_norm": 0.382344514131546,
      "learning_rate": 9.724030232334391e-05,
      "loss": 0.8362,
      "step": 71
    },
    {
      "epoch": 0.0362309724493647,
      "grad_norm": 0.38100576400756836,
      "learning_rate": 9.713207455460894e-05,
      "loss": 0.8169,
      "step": 72
    },
    {
      "epoch": 0.03673418040005032,
      "grad_norm": 0.40086644887924194,
      "learning_rate": 9.702182780466775e-05,
      "loss": 0.7975,
      "step": 73
    },
    {
      "epoch": 0.037237388350735943,
      "grad_norm": 0.434026837348938,
      "learning_rate": 9.690956679612421e-05,
      "loss": 0.9394,
      "step": 74
    },
    {
      "epoch": 0.03774059630142156,
      "grad_norm": 0.47511646151542664,
      "learning_rate": 9.67952963378663e-05,
      "loss": 0.9199,
      "step": 75
    },
    {
      "epoch": 0.03824380425210718,
      "grad_norm": 0.4795306324958801,
      "learning_rate": 9.667902132486009e-05,
      "loss": 0.8249,
      "step": 76
    },
    {
      "epoch": 0.0387470122027928,
      "grad_norm": 0.4623278081417084,
      "learning_rate": 9.656074673794018e-05,
      "loss": 0.7846,
      "step": 77
    },
    {
      "epoch": 0.03925022015347843,
      "grad_norm": 0.41059255599975586,
      "learning_rate": 9.644047764359622e-05,
      "loss": 0.7837,
      "step": 78
    },
    {
      "epoch": 0.039753428104164046,
      "grad_norm": 0.43599745631217957,
      "learning_rate": 9.631821919375591e-05,
      "loss": 0.7853,
      "step": 79
    },
    {
      "epoch": 0.040256636054849665,
      "grad_norm": 0.4747604429721832,
      "learning_rate": 9.619397662556435e-05,
      "loss": 0.8343,
      "step": 80
    },
    {
      "epoch": 0.04075984400553529,
      "grad_norm": 0.5610798001289368,
      "learning_rate": 9.606775526115963e-05,
      "loss": 0.8586,
      "step": 81
    },
    {
      "epoch": 0.04126305195622091,
      "grad_norm": 0.5342729091644287,
      "learning_rate": 9.593956050744492e-05,
      "loss": 0.9634,
      "step": 82
    },
    {
      "epoch": 0.04176625990690653,
      "grad_norm": 0.5705016851425171,
      "learning_rate": 9.580939785585681e-05,
      "loss": 0.8676,
      "step": 83
    },
    {
      "epoch": 0.04226946785759215,
      "grad_norm": 0.5341368913650513,
      "learning_rate": 9.567727288213005e-05,
      "loss": 0.8717,
      "step": 84
    },
    {
      "epoch": 0.042772675808277774,
      "grad_norm": 0.5753642320632935,
      "learning_rate": 9.554319124605879e-05,
      "loss": 0.8475,
      "step": 85
    },
    {
      "epoch": 0.04327588375896339,
      "grad_norm": 0.5283560752868652,
      "learning_rate": 9.540715869125407e-05,
      "loss": 0.8437,
      "step": 86
    },
    {
      "epoch": 0.04377909170964901,
      "grad_norm": 0.6042847633361816,
      "learning_rate": 9.526918104489777e-05,
      "loss": 1.1062,
      "step": 87
    },
    {
      "epoch": 0.04428229966033463,
      "grad_norm": 0.6444901823997498,
      "learning_rate": 9.512926421749304e-05,
      "loss": 0.9332,
      "step": 88
    },
    {
      "epoch": 0.04478550761102026,
      "grad_norm": 0.6121519207954407,
      "learning_rate": 9.498741420261108e-05,
      "loss": 0.8726,
      "step": 89
    },
    {
      "epoch": 0.045288715561705876,
      "grad_norm": 0.6061900854110718,
      "learning_rate": 9.484363707663442e-05,
      "loss": 0.9538,
      "step": 90
    },
    {
      "epoch": 0.045791923512391496,
      "grad_norm": 0.6390529870986938,
      "learning_rate": 9.469793899849661e-05,
      "loss": 0.9553,
      "step": 91
    },
    {
      "epoch": 0.046295131463077115,
      "grad_norm": 0.6590822339057922,
      "learning_rate": 9.45503262094184e-05,
      "loss": 0.8401,
      "step": 92
    },
    {
      "epoch": 0.04679833941376274,
      "grad_norm": 0.8563952445983887,
      "learning_rate": 9.440080503264037e-05,
      "loss": 1.2022,
      "step": 93
    },
    {
      "epoch": 0.04730154736444836,
      "grad_norm": 0.869257926940918,
      "learning_rate": 9.42493818731521e-05,
      "loss": 0.9897,
      "step": 94
    },
    {
      "epoch": 0.04780475531513398,
      "grad_norm": 0.867311954498291,
      "learning_rate": 9.409606321741775e-05,
      "loss": 0.9971,
      "step": 95
    },
    {
      "epoch": 0.0483079632658196,
      "grad_norm": 0.9005740284919739,
      "learning_rate": 9.394085563309827e-05,
      "loss": 0.9732,
      "step": 96
    },
    {
      "epoch": 0.048811171216505224,
      "grad_norm": 1.0737452507019043,
      "learning_rate": 9.378376576876999e-05,
      "loss": 1.0972,
      "step": 97
    },
    {
      "epoch": 0.04931437916719084,
      "grad_norm": 0.9764959216117859,
      "learning_rate": 9.362480035363986e-05,
      "loss": 1.031,
      "step": 98
    },
    {
      "epoch": 0.04981758711787646,
      "grad_norm": 1.3426367044448853,
      "learning_rate": 9.34639661972572e-05,
      "loss": 1.1806,
      "step": 99
    },
    {
      "epoch": 0.05032079506856208,
      "grad_norm": 1.750638484954834,
      "learning_rate": 9.330127018922194e-05,
      "loss": 1.1607,
      "step": 100
    },
    {
      "epoch": 0.05032079506856208,
      "eval_loss": 0.8727328777313232,
      "eval_runtime": 313.2272,
      "eval_samples_per_second": 10.686,
      "eval_steps_per_second": 2.672,
      "step": 100
    },
    {
      "epoch": 0.05082400301924771,
      "grad_norm": 0.3322669267654419,
      "learning_rate": 9.31367192988896e-05,
      "loss": 0.6956,
      "step": 101
    },
    {
      "epoch": 0.051327210969933326,
      "grad_norm": 0.3668118715286255,
      "learning_rate": 9.297032057507264e-05,
      "loss": 0.6526,
      "step": 102
    },
    {
      "epoch": 0.051830418920618945,
      "grad_norm": 0.31805333495140076,
      "learning_rate": 9.280208114573859e-05,
      "loss": 0.5968,
      "step": 103
    },
    {
      "epoch": 0.052333626871304564,
      "grad_norm": 0.31701406836509705,
      "learning_rate": 9.263200821770461e-05,
      "loss": 0.7362,
      "step": 104
    },
    {
      "epoch": 0.05283683482199019,
      "grad_norm": 0.278466135263443,
      "learning_rate": 9.246010907632895e-05,
      "loss": 0.6471,
      "step": 105
    },
    {
      "epoch": 0.05334004277267581,
      "grad_norm": 0.2978034019470215,
      "learning_rate": 9.228639108519868e-05,
      "loss": 0.7367,
      "step": 106
    },
    {
      "epoch": 0.05384325072336143,
      "grad_norm": 0.2924036681652069,
      "learning_rate": 9.211086168581433e-05,
      "loss": 0.684,
      "step": 107
    },
    {
      "epoch": 0.05434645867404705,
      "grad_norm": 0.3505677878856659,
      "learning_rate": 9.193352839727121e-05,
      "loss": 0.7644,
      "step": 108
    },
    {
      "epoch": 0.054849666624732674,
      "grad_norm": 0.37053295969963074,
      "learning_rate": 9.175439881593716e-05,
      "loss": 0.8439,
      "step": 109
    },
    {
      "epoch": 0.05535287457541829,
      "grad_norm": 0.3657532334327698,
      "learning_rate": 9.157348061512727e-05,
      "loss": 0.7475,
      "step": 110
    },
    {
      "epoch": 0.05585608252610391,
      "grad_norm": 0.3392336368560791,
      "learning_rate": 9.139078154477512e-05,
      "loss": 0.8182,
      "step": 111
    },
    {
      "epoch": 0.05635929047678953,
      "grad_norm": 0.3058857321739197,
      "learning_rate": 9.120630943110077e-05,
      "loss": 0.6642,
      "step": 112
    },
    {
      "epoch": 0.05686249842747516,
      "grad_norm": 0.31133466958999634,
      "learning_rate": 9.102007217627568e-05,
      "loss": 0.7286,
      "step": 113
    },
    {
      "epoch": 0.057365706378160776,
      "grad_norm": 0.3490636348724365,
      "learning_rate": 9.083207775808396e-05,
      "loss": 0.8766,
      "step": 114
    },
    {
      "epoch": 0.057868914328846395,
      "grad_norm": 0.3371574878692627,
      "learning_rate": 9.064233422958077e-05,
      "loss": 0.7686,
      "step": 115
    },
    {
      "epoch": 0.058372122279532014,
      "grad_norm": 0.37452971935272217,
      "learning_rate": 9.045084971874738e-05,
      "loss": 0.7973,
      "step": 116
    },
    {
      "epoch": 0.05887533023021764,
      "grad_norm": 0.3957357704639435,
      "learning_rate": 9.025763242814291e-05,
      "loss": 0.8964,
      "step": 117
    },
    {
      "epoch": 0.05937853818090326,
      "grad_norm": 0.40561509132385254,
      "learning_rate": 9.006269063455304e-05,
      "loss": 0.9379,
      "step": 118
    },
    {
      "epoch": 0.05988174613158888,
      "grad_norm": 0.40388020873069763,
      "learning_rate": 8.986603268863536e-05,
      "loss": 0.9425,
      "step": 119
    },
    {
      "epoch": 0.0603849540822745,
      "grad_norm": 0.42222151160240173,
      "learning_rate": 8.966766701456177e-05,
      "loss": 0.945,
      "step": 120
    },
    {
      "epoch": 0.06088816203296012,
      "grad_norm": 0.3823082447052002,
      "learning_rate": 8.94676021096575e-05,
      "loss": 0.8699,
      "step": 121
    },
    {
      "epoch": 0.06139136998364574,
      "grad_norm": 0.4082789421081543,
      "learning_rate": 8.926584654403724e-05,
      "loss": 0.9377,
      "step": 122
    },
    {
      "epoch": 0.06189457793433136,
      "grad_norm": 0.4066939949989319,
      "learning_rate": 8.906240896023794e-05,
      "loss": 0.89,
      "step": 123
    },
    {
      "epoch": 0.06239778588501698,
      "grad_norm": 0.4092026352882385,
      "learning_rate": 8.885729807284856e-05,
      "loss": 0.9184,
      "step": 124
    },
    {
      "epoch": 0.0629009938357026,
      "grad_norm": 0.4130512773990631,
      "learning_rate": 8.865052266813685e-05,
      "loss": 0.847,
      "step": 125
    },
    {
      "epoch": 0.06340420178638823,
      "grad_norm": 0.40424537658691406,
      "learning_rate": 8.844209160367299e-05,
      "loss": 0.8562,
      "step": 126
    },
    {
      "epoch": 0.06390740973707384,
      "grad_norm": 0.4662040174007416,
      "learning_rate": 8.823201380795001e-05,
      "loss": 0.8301,
      "step": 127
    },
    {
      "epoch": 0.06441061768775946,
      "grad_norm": 0.47442004084587097,
      "learning_rate": 8.802029828000156e-05,
      "loss": 0.9385,
      "step": 128
    },
    {
      "epoch": 0.06491382563844508,
      "grad_norm": 0.5055505037307739,
      "learning_rate": 8.780695408901613e-05,
      "loss": 0.86,
      "step": 129
    },
    {
      "epoch": 0.0654170335891307,
      "grad_norm": 0.49873098731040955,
      "learning_rate": 8.759199037394887e-05,
      "loss": 0.9525,
      "step": 130
    },
    {
      "epoch": 0.06592024153981633,
      "grad_norm": 0.4767245054244995,
      "learning_rate": 8.737541634312985e-05,
      "loss": 0.8238,
      "step": 131
    },
    {
      "epoch": 0.06642344949050195,
      "grad_norm": 0.4893306791782379,
      "learning_rate": 8.715724127386972e-05,
      "loss": 0.8861,
      "step": 132
    },
    {
      "epoch": 0.06692665744118757,
      "grad_norm": 0.5542322993278503,
      "learning_rate": 8.693747451206232e-05,
      "loss": 0.9301,
      "step": 133
    },
    {
      "epoch": 0.06742986539187319,
      "grad_norm": 0.5872162580490112,
      "learning_rate": 8.671612547178428e-05,
      "loss": 0.8787,
      "step": 134
    },
    {
      "epoch": 0.06793307334255881,
      "grad_norm": 0.5158484578132629,
      "learning_rate": 8.649320363489179e-05,
      "loss": 0.7941,
      "step": 135
    },
    {
      "epoch": 0.06843628129324443,
      "grad_norm": 0.6727821826934814,
      "learning_rate": 8.626871855061438e-05,
      "loss": 1.0169,
      "step": 136
    },
    {
      "epoch": 0.06893948924393005,
      "grad_norm": 0.5748090744018555,
      "learning_rate": 8.604267983514594e-05,
      "loss": 0.8177,
      "step": 137
    },
    {
      "epoch": 0.06944269719461567,
      "grad_norm": 0.5611135363578796,
      "learning_rate": 8.581509717123273e-05,
      "loss": 0.8441,
      "step": 138
    },
    {
      "epoch": 0.0699459051453013,
      "grad_norm": 0.5927470922470093,
      "learning_rate": 8.558598030775857e-05,
      "loss": 0.8255,
      "step": 139
    },
    {
      "epoch": 0.07044911309598692,
      "grad_norm": 0.6794228553771973,
      "learning_rate": 8.535533905932738e-05,
      "loss": 0.9802,
      "step": 140
    },
    {
      "epoch": 0.07095232104667254,
      "grad_norm": 0.7753991484642029,
      "learning_rate": 8.51231833058426e-05,
      "loss": 0.9719,
      "step": 141
    },
    {
      "epoch": 0.07145552899735816,
      "grad_norm": 0.7393898963928223,
      "learning_rate": 8.488952299208401e-05,
      "loss": 0.9217,
      "step": 142
    },
    {
      "epoch": 0.07195873694804378,
      "grad_norm": 0.7829727530479431,
      "learning_rate": 8.46543681272818e-05,
      "loss": 1.0013,
      "step": 143
    },
    {
      "epoch": 0.0724619448987294,
      "grad_norm": 0.8072141408920288,
      "learning_rate": 8.44177287846877e-05,
      "loss": 0.9985,
      "step": 144
    },
    {
      "epoch": 0.07296515284941502,
      "grad_norm": 0.9565557837486267,
      "learning_rate": 8.417961510114356e-05,
      "loss": 1.1174,
      "step": 145
    },
    {
      "epoch": 0.07346836080010063,
      "grad_norm": 0.8486001491546631,
      "learning_rate": 8.39400372766471e-05,
      "loss": 1.1132,
      "step": 146
    },
    {
      "epoch": 0.07397156875078627,
      "grad_norm": 1.0413708686828613,
      "learning_rate": 8.36990055739149e-05,
      "loss": 1.1074,
      "step": 147
    },
    {
      "epoch": 0.07447477670147189,
      "grad_norm": 1.0939488410949707,
      "learning_rate": 8.345653031794292e-05,
      "loss": 1.0398,
      "step": 148
    },
    {
      "epoch": 0.0749779846521575,
      "grad_norm": 1.3568888902664185,
      "learning_rate": 8.321262189556409e-05,
      "loss": 1.2814,
      "step": 149
    },
    {
      "epoch": 0.07548119260284313,
      "grad_norm": 1.7241541147232056,
      "learning_rate": 8.296729075500344e-05,
      "loss": 0.9507,
      "step": 150
    },
    {
      "epoch": 0.07598440055352874,
      "grad_norm": 0.27183854579925537,
      "learning_rate": 8.272054740543052e-05,
      "loss": 0.546,
      "step": 151
    },
    {
      "epoch": 0.07648760850421436,
      "grad_norm": 0.3252801299095154,
      "learning_rate": 8.247240241650918e-05,
      "loss": 0.6398,
      "step": 152
    },
    {
      "epoch": 0.07699081645489998,
      "grad_norm": 0.3087828457355499,
      "learning_rate": 8.222286641794488e-05,
      "loss": 0.7522,
      "step": 153
    },
    {
      "epoch": 0.0774940244055856,
      "grad_norm": 0.27105748653411865,
      "learning_rate": 8.197195009902924e-05,
      "loss": 0.5907,
      "step": 154
    },
    {
      "epoch": 0.07799723235627123,
      "grad_norm": 0.29571643471717834,
      "learning_rate": 8.171966420818228e-05,
      "loss": 0.6671,
      "step": 155
    },
    {
      "epoch": 0.07850044030695685,
      "grad_norm": 0.30681684613227844,
      "learning_rate": 8.146601955249188e-05,
      "loss": 0.6634,
      "step": 156
    },
    {
      "epoch": 0.07900364825764247,
      "grad_norm": 0.2717362940311432,
      "learning_rate": 8.121102699725089e-05,
      "loss": 0.6372,
      "step": 157
    },
    {
      "epoch": 0.07950685620832809,
      "grad_norm": 0.30869171023368835,
      "learning_rate": 8.095469746549172e-05,
      "loss": 0.6234,
      "step": 158
    },
    {
      "epoch": 0.08001006415901371,
      "grad_norm": 0.31170839071273804,
      "learning_rate": 8.069704193751832e-05,
      "loss": 0.7328,
      "step": 159
    },
    {
      "epoch": 0.08051327210969933,
      "grad_norm": 0.320419579744339,
      "learning_rate": 8.043807145043604e-05,
      "loss": 0.7399,
      "step": 160
    },
    {
      "epoch": 0.08101648006038495,
      "grad_norm": 0.4064423143863678,
      "learning_rate": 8.017779709767858e-05,
      "loss": 0.9244,
      "step": 161
    },
    {
      "epoch": 0.08151968801107058,
      "grad_norm": 0.33918797969818115,
      "learning_rate": 7.991623002853296e-05,
      "loss": 0.7693,
      "step": 162
    },
    {
      "epoch": 0.0820228959617562,
      "grad_norm": 0.35222694277763367,
      "learning_rate": 7.965338144766186e-05,
      "loss": 0.7818,
      "step": 163
    },
    {
      "epoch": 0.08252610391244182,
      "grad_norm": 0.3331509530544281,
      "learning_rate": 7.938926261462366e-05,
      "loss": 0.7294,
      "step": 164
    },
    {
      "epoch": 0.08302931186312744,
      "grad_norm": 0.3317022919654846,
      "learning_rate": 7.912388484339012e-05,
      "loss": 0.7169,
      "step": 165
    },
    {
      "epoch": 0.08353251981381306,
      "grad_norm": 0.36114493012428284,
      "learning_rate": 7.88572595018617e-05,
      "loss": 0.7398,
      "step": 166
    },
    {
      "epoch": 0.08403572776449868,
      "grad_norm": 0.3843732178211212,
      "learning_rate": 7.858939801138061e-05,
      "loss": 0.8061,
      "step": 167
    },
    {
      "epoch": 0.0845389357151843,
      "grad_norm": 0.42287304997444153,
      "learning_rate": 7.832031184624164e-05,
      "loss": 0.903,
      "step": 168
    },
    {
      "epoch": 0.08504214366586992,
      "grad_norm": 0.4068758487701416,
      "learning_rate": 7.80500125332005e-05,
      "loss": 0.9174,
      "step": 169
    },
    {
      "epoch": 0.08554535161655555,
      "grad_norm": 0.42986854910850525,
      "learning_rate": 7.777851165098012e-05,
      "loss": 0.957,
      "step": 170
    },
    {
      "epoch": 0.08604855956724117,
      "grad_norm": 0.3900366723537445,
      "learning_rate": 7.750582082977467e-05,
      "loss": 0.7698,
      "step": 171
    },
    {
      "epoch": 0.08655176751792679,
      "grad_norm": 0.38784149289131165,
      "learning_rate": 7.723195175075136e-05,
      "loss": 0.8406,
      "step": 172
    },
    {
      "epoch": 0.0870549754686124,
      "grad_norm": 0.3956195116043091,
      "learning_rate": 7.695691614555003e-05,
      "loss": 0.8886,
      "step": 173
    },
    {
      "epoch": 0.08755818341929802,
      "grad_norm": 0.40850603580474854,
      "learning_rate": 7.668072579578058e-05,
      "loss": 0.8164,
      "step": 174
    },
    {
      "epoch": 0.08806139136998364,
      "grad_norm": 0.4286244511604309,
      "learning_rate": 7.64033925325184e-05,
      "loss": 0.8281,
      "step": 175
    },
    {
      "epoch": 0.08856459932066926,
      "grad_norm": 0.447592169046402,
      "learning_rate": 7.612492823579745e-05,
      "loss": 0.9133,
      "step": 176
    },
    {
      "epoch": 0.08906780727135488,
      "grad_norm": 0.412308931350708,
      "learning_rate": 7.584534483410137e-05,
      "loss": 0.9809,
      "step": 177
    },
    {
      "epoch": 0.08957101522204051,
      "grad_norm": 0.4190141260623932,
      "learning_rate": 7.55646543038526e-05,
      "loss": 0.7634,
      "step": 178
    },
    {
      "epoch": 0.09007422317272613,
      "grad_norm": 0.44869595766067505,
      "learning_rate": 7.528286866889924e-05,
      "loss": 0.8793,
      "step": 179
    },
    {
      "epoch": 0.09057743112341175,
      "grad_norm": 0.48549386858940125,
      "learning_rate": 7.500000000000001e-05,
      "loss": 0.8816,
      "step": 180
    },
    {
      "epoch": 0.09108063907409737,
      "grad_norm": 0.5181676149368286,
      "learning_rate": 7.471606041430723e-05,
      "loss": 0.9391,
      "step": 181
    },
    {
      "epoch": 0.09158384702478299,
      "grad_norm": 0.5079827904701233,
      "learning_rate": 7.443106207484776e-05,
      "loss": 0.8114,
      "step": 182
    },
    {
      "epoch": 0.09208705497546861,
      "grad_norm": 0.5033254623413086,
      "learning_rate": 7.414501719000187e-05,
      "loss": 0.8407,
      "step": 183
    },
    {
      "epoch": 0.09259026292615423,
      "grad_norm": 0.5162456035614014,
      "learning_rate": 7.385793801298042e-05,
      "loss": 0.7517,
      "step": 184
    },
    {
      "epoch": 0.09309347087683985,
      "grad_norm": 0.5596376061439514,
      "learning_rate": 7.35698368412999e-05,
      "loss": 0.8488,
      "step": 185
    },
    {
      "epoch": 0.09359667882752548,
      "grad_norm": 0.5955862402915955,
      "learning_rate": 7.328072601625557e-05,
      "loss": 0.8705,
      "step": 186
    },
    {
      "epoch": 0.0940998867782111,
      "grad_norm": 0.5844749808311462,
      "learning_rate": 7.2990617922393e-05,
      "loss": 0.9717,
      "step": 187
    },
    {
      "epoch": 0.09460309472889672,
      "grad_norm": 0.6185311079025269,
      "learning_rate": 7.269952498697734e-05,
      "loss": 0.8028,
      "step": 188
    },
    {
      "epoch": 0.09510630267958234,
      "grad_norm": 0.6681666374206543,
      "learning_rate": 7.240745967946113e-05,
      "loss": 0.9047,
      "step": 189
    },
    {
      "epoch": 0.09560951063026796,
      "grad_norm": 0.8303496241569519,
      "learning_rate": 7.211443451095007e-05,
      "loss": 0.9022,
      "step": 190
    },
    {
      "epoch": 0.09611271858095358,
      "grad_norm": 0.7427216172218323,
      "learning_rate": 7.18204620336671e-05,
      "loss": 0.9407,
      "step": 191
    },
    {
      "epoch": 0.0966159265316392,
      "grad_norm": 0.7446197867393494,
      "learning_rate": 7.152555484041476e-05,
      "loss": 1.1077,
      "step": 192
    },
    {
      "epoch": 0.09711913448232481,
      "grad_norm": 0.8290247917175293,
      "learning_rate": 7.122972556403567e-05,
      "loss": 0.969,
      "step": 193
    },
    {
      "epoch": 0.09762234243301045,
      "grad_norm": 0.8350086808204651,
      "learning_rate": 7.09329868768714e-05,
      "loss": 0.9224,
      "step": 194
    },
    {
      "epoch": 0.09812555038369607,
      "grad_norm": 0.9003720283508301,
      "learning_rate": 7.063535149021973e-05,
      "loss": 1.0217,
      "step": 195
    },
    {
      "epoch": 0.09862875833438169,
      "grad_norm": 0.9789468050003052,
      "learning_rate": 7.033683215379002e-05,
      "loss": 1.0498,
      "step": 196
    },
    {
      "epoch": 0.0991319662850673,
      "grad_norm": 0.9070441722869873,
      "learning_rate": 7.003744165515705e-05,
      "loss": 1.0021,
      "step": 197
    },
    {
      "epoch": 0.09963517423575292,
      "grad_norm": 1.0430539846420288,
      "learning_rate": 6.973719281921335e-05,
      "loss": 1.0564,
      "step": 198
    },
    {
      "epoch": 0.10013838218643854,
      "grad_norm": 1.1853272914886475,
      "learning_rate": 6.943609850761979e-05,
      "loss": 1.013,
      "step": 199
    },
    {
      "epoch": 0.10064159013712416,
      "grad_norm": 1.7435191869735718,
      "learning_rate": 6.91341716182545e-05,
      "loss": 0.9971,
      "step": 200
    },
    {
      "epoch": 0.10064159013712416,
      "eval_loss": 0.8363951444625854,
      "eval_runtime": 316.1265,
      "eval_samples_per_second": 10.588,
      "eval_steps_per_second": 2.648,
      "step": 200
    },
    {
      "epoch": 0.10114479808780978,
      "grad_norm": 0.2257886677980423,
      "learning_rate": 6.883142508466054e-05,
      "loss": 0.4313,
      "step": 201
    },
    {
      "epoch": 0.10164800603849541,
      "grad_norm": 0.2908216416835785,
      "learning_rate": 6.852787187549182e-05,
      "loss": 0.5445,
      "step": 202
    },
    {
      "epoch": 0.10215121398918103,
      "grad_norm": 0.2470732182264328,
      "learning_rate": 6.82235249939575e-05,
      "loss": 0.5322,
      "step": 203
    },
    {
      "epoch": 0.10265442193986665,
      "grad_norm": 0.3444390594959259,
      "learning_rate": 6.7918397477265e-05,
      "loss": 0.6779,
      "step": 204
    },
    {
      "epoch": 0.10315762989055227,
      "grad_norm": 0.3060758709907532,
      "learning_rate": 6.761250239606169e-05,
      "loss": 0.7115,
      "step": 205
    },
    {
      "epoch": 0.10366083784123789,
      "grad_norm": 0.3036004602909088,
      "learning_rate": 6.730585285387465e-05,
      "loss": 0.7163,
      "step": 206
    },
    {
      "epoch": 0.10416404579192351,
      "grad_norm": 0.2907506823539734,
      "learning_rate": 6.699846198654971e-05,
      "loss": 0.6395,
      "step": 207
    },
    {
      "epoch": 0.10466725374260913,
      "grad_norm": 0.3024415373802185,
      "learning_rate": 6.669034296168855e-05,
      "loss": 0.7051,
      "step": 208
    },
    {
      "epoch": 0.10517046169329475,
      "grad_norm": 0.302102655172348,
      "learning_rate": 6.638150897808468e-05,
      "loss": 0.75,
      "step": 209
    },
    {
      "epoch": 0.10567366964398038,
      "grad_norm": 0.3131133019924164,
      "learning_rate": 6.607197326515808e-05,
      "loss": 0.6947,
      "step": 210
    },
    {
      "epoch": 0.106176877594666,
      "grad_norm": 0.3494044542312622,
      "learning_rate": 6.57617490823885e-05,
      "loss": 0.8186,
      "step": 211
    },
    {
      "epoch": 0.10668008554535162,
      "grad_norm": 0.33517181873321533,
      "learning_rate": 6.545084971874738e-05,
      "loss": 0.6871,
      "step": 212
    },
    {
      "epoch": 0.10718329349603724,
      "grad_norm": 0.3964875638484955,
      "learning_rate": 6.513928849212873e-05,
      "loss": 0.8509,
      "step": 213
    },
    {
      "epoch": 0.10768650144672286,
      "grad_norm": 0.3203575313091278,
      "learning_rate": 6.482707874877854e-05,
      "loss": 0.7426,
      "step": 214
    },
    {
      "epoch": 0.10818970939740848,
      "grad_norm": 0.31022173166275024,
      "learning_rate": 6.451423386272312e-05,
      "loss": 0.7274,
      "step": 215
    },
    {
      "epoch": 0.1086929173480941,
      "grad_norm": 0.3365987241268158,
      "learning_rate": 6.420076723519614e-05,
      "loss": 0.7834,
      "step": 216
    },
    {
      "epoch": 0.10919612529877971,
      "grad_norm": 0.35897573828697205,
      "learning_rate": 6.388669229406462e-05,
      "loss": 0.8287,
      "step": 217
    },
    {
      "epoch": 0.10969933324946535,
      "grad_norm": 0.3722810745239258,
      "learning_rate": 6.357202249325371e-05,
      "loss": 0.9373,
      "step": 218
    },
    {
      "epoch": 0.11020254120015097,
      "grad_norm": 0.32477128505706787,
      "learning_rate": 6.32567713121704e-05,
      "loss": 0.8085,
      "step": 219
    },
    {
      "epoch": 0.11070574915083659,
      "grad_norm": 0.3690595328807831,
      "learning_rate": 6.294095225512603e-05,
      "loss": 0.8465,
      "step": 220
    },
    {
      "epoch": 0.1112089571015222,
      "grad_norm": 0.3620961904525757,
      "learning_rate": 6.26245788507579e-05,
      "loss": 0.9019,
      "step": 221
    },
    {
      "epoch": 0.11171216505220782,
      "grad_norm": 0.4240458309650421,
      "learning_rate": 6.230766465144967e-05,
      "loss": 0.8441,
      "step": 222
    },
    {
      "epoch": 0.11221537300289344,
      "grad_norm": 0.34328147768974304,
      "learning_rate": 6.199022323275083e-05,
      "loss": 0.7311,
      "step": 223
    },
    {
      "epoch": 0.11271858095357906,
      "grad_norm": 0.3980419337749481,
      "learning_rate": 6.167226819279528e-05,
      "loss": 0.782,
      "step": 224
    },
    {
      "epoch": 0.11322178890426468,
      "grad_norm": 0.39327988028526306,
      "learning_rate": 6.135381315171867e-05,
      "loss": 0.7983,
      "step": 225
    },
    {
      "epoch": 0.11372499685495031,
      "grad_norm": 0.41335606575012207,
      "learning_rate": 6.103487175107507e-05,
      "loss": 0.9564,
      "step": 226
    },
    {
      "epoch": 0.11422820480563593,
      "grad_norm": 0.4150274395942688,
      "learning_rate": 6.071545765325254e-05,
      "loss": 0.7931,
      "step": 227
    },
    {
      "epoch": 0.11473141275632155,
      "grad_norm": 0.46463119983673096,
      "learning_rate": 6.0395584540887963e-05,
      "loss": 0.878,
      "step": 228
    },
    {
      "epoch": 0.11523462070700717,
      "grad_norm": 0.5011468529701233,
      "learning_rate": 6.007526611628086e-05,
      "loss": 0.8926,
      "step": 229
    },
    {
      "epoch": 0.11573782865769279,
      "grad_norm": 0.47439324855804443,
      "learning_rate": 5.9754516100806423e-05,
      "loss": 0.9395,
      "step": 230
    },
    {
      "epoch": 0.11624103660837841,
      "grad_norm": 0.47574976086616516,
      "learning_rate": 5.9433348234327765e-05,
      "loss": 0.9035,
      "step": 231
    },
    {
      "epoch": 0.11674424455906403,
      "grad_norm": 0.513819694519043,
      "learning_rate": 5.911177627460739e-05,
      "loss": 0.7792,
      "step": 232
    },
    {
      "epoch": 0.11724745250974965,
      "grad_norm": 0.4985521733760834,
      "learning_rate": 5.8789813996717736e-05,
      "loss": 0.7791,
      "step": 233
    },
    {
      "epoch": 0.11775066046043528,
      "grad_norm": 0.5244281888008118,
      "learning_rate": 5.8467475192451226e-05,
      "loss": 0.7656,
      "step": 234
    },
    {
      "epoch": 0.1182538684111209,
      "grad_norm": 0.5643393397331238,
      "learning_rate": 5.814477366972945e-05,
      "loss": 0.8311,
      "step": 235
    },
    {
      "epoch": 0.11875707636180652,
      "grad_norm": 0.6539595723152161,
      "learning_rate": 5.782172325201155e-05,
      "loss": 1.0001,
      "step": 236
    },
    {
      "epoch": 0.11926028431249214,
      "grad_norm": 0.6218690872192383,
      "learning_rate": 5.749833777770225e-05,
      "loss": 0.907,
      "step": 237
    },
    {
      "epoch": 0.11976349226317776,
      "grad_norm": 0.6421785950660706,
      "learning_rate": 5.717463109955896e-05,
      "loss": 0.8911,
      "step": 238
    },
    {
      "epoch": 0.12026670021386338,
      "grad_norm": 0.6773377060890198,
      "learning_rate": 5.685061708409841e-05,
      "loss": 0.992,
      "step": 239
    },
    {
      "epoch": 0.120769908164549,
      "grad_norm": 0.6887732148170471,
      "learning_rate": 5.6526309611002594e-05,
      "loss": 0.8878,
      "step": 240
    },
    {
      "epoch": 0.12127311611523463,
      "grad_norm": 0.7765624523162842,
      "learning_rate": 5.6201722572524275e-05,
      "loss": 0.9145,
      "step": 241
    },
    {
      "epoch": 0.12177632406592025,
      "grad_norm": 0.714083731174469,
      "learning_rate": 5.587686987289189e-05,
      "loss": 0.8556,
      "step": 242
    },
    {
      "epoch": 0.12227953201660587,
      "grad_norm": 0.7760418057441711,
      "learning_rate": 5.5551765427713884e-05,
      "loss": 1.0028,
      "step": 243
    },
    {
      "epoch": 0.12278273996729148,
      "grad_norm": 0.8172394633293152,
      "learning_rate": 5.522642316338268e-05,
      "loss": 0.9939,
      "step": 244
    },
    {
      "epoch": 0.1232859479179771,
      "grad_norm": 0.9221703410148621,
      "learning_rate": 5.490085701647805e-05,
      "loss": 1.0789,
      "step": 245
    },
    {
      "epoch": 0.12378915586866272,
      "grad_norm": 1.002267599105835,
      "learning_rate": 5.457508093317013e-05,
      "loss": 1.1407,
      "step": 246
    },
    {
      "epoch": 0.12429236381934834,
      "grad_norm": 1.0404046773910522,
      "learning_rate": 5.4249108868622086e-05,
      "loss": 1.044,
      "step": 247
    },
    {
      "epoch": 0.12479557177003396,
      "grad_norm": 1.0855293273925781,
      "learning_rate": 5.392295478639225e-05,
      "loss": 1.0306,
      "step": 248
    },
    {
      "epoch": 0.1252987797207196,
      "grad_norm": 1.1987627744674683,
      "learning_rate": 5.359663265783598e-05,
      "loss": 1.0394,
      "step": 249
    },
    {
      "epoch": 0.1258019876714052,
      "grad_norm": 1.8623121976852417,
      "learning_rate": 5.327015646150716e-05,
      "loss": 1.2246,
      "step": 250
    },
    {
      "epoch": 0.12630519562209083,
      "grad_norm": 0.2524110674858093,
      "learning_rate": 5.294354018255945e-05,
      "loss": 0.5109,
      "step": 251
    },
    {
      "epoch": 0.12680840357277645,
      "grad_norm": 0.3118424713611603,
      "learning_rate": 5.26167978121472e-05,
      "loss": 0.5586,
      "step": 252
    },
    {
      "epoch": 0.12731161152346207,
      "grad_norm": 0.2597341239452362,
      "learning_rate": 5.228994334682604e-05,
      "loss": 0.6107,
      "step": 253
    },
    {
      "epoch": 0.1278148194741477,
      "grad_norm": 0.31350454688072205,
      "learning_rate": 5.196299078795344e-05,
      "loss": 0.6255,
      "step": 254
    },
    {
      "epoch": 0.1283180274248333,
      "grad_norm": 0.22063176333904266,
      "learning_rate": 5.1635954141088813e-05,
      "loss": 0.5468,
      "step": 255
    },
    {
      "epoch": 0.12882123537551893,
      "grad_norm": 0.2539903521537781,
      "learning_rate": 5.1308847415393666e-05,
      "loss": 0.6102,
      "step": 256
    },
    {
      "epoch": 0.12932444332620455,
      "grad_norm": 0.3015947937965393,
      "learning_rate": 5.0981684623031415e-05,
      "loss": 0.6774,
      "step": 257
    },
    {
      "epoch": 0.12982765127689017,
      "grad_norm": 0.3113625943660736,
      "learning_rate": 5.0654479778567223e-05,
      "loss": 0.7229,
      "step": 258
    },
    {
      "epoch": 0.13033085922757578,
      "grad_norm": 0.29673701524734497,
      "learning_rate": 5.0327246898367597e-05,
      "loss": 0.6658,
      "step": 259
    },
    {
      "epoch": 0.1308340671782614,
      "grad_norm": 0.3253553807735443,
      "learning_rate": 5e-05,
      "loss": 0.8489,
      "step": 260
    },
    {
      "epoch": 0.13133727512894705,
      "grad_norm": 0.3458925187587738,
      "learning_rate": 4.9672753101632415e-05,
      "loss": 0.8382,
      "step": 261
    },
    {
      "epoch": 0.13184048307963267,
      "grad_norm": 0.31306329369544983,
      "learning_rate": 4.934552022143279e-05,
      "loss": 0.7144,
      "step": 262
    },
    {
      "epoch": 0.1323436910303183,
      "grad_norm": 0.31457698345184326,
      "learning_rate": 4.901831537696859e-05,
      "loss": 0.7085,
      "step": 263
    },
    {
      "epoch": 0.1328468989810039,
      "grad_norm": 0.2948032021522522,
      "learning_rate": 4.869115258460635e-05,
      "loss": 0.6265,
      "step": 264
    },
    {
      "epoch": 0.13335010693168953,
      "grad_norm": 0.33766692876815796,
      "learning_rate": 4.83640458589112e-05,
      "loss": 0.861,
      "step": 265
    },
    {
      "epoch": 0.13385331488237515,
      "grad_norm": 0.3336288332939148,
      "learning_rate": 4.8037009212046586e-05,
      "loss": 0.7275,
      "step": 266
    },
    {
      "epoch": 0.13435652283306077,
      "grad_norm": 0.37023717164993286,
      "learning_rate": 4.7710056653173976e-05,
      "loss": 0.809,
      "step": 267
    },
    {
      "epoch": 0.13485973078374638,
      "grad_norm": 0.36774396896362305,
      "learning_rate": 4.738320218785281e-05,
      "loss": 0.7804,
      "step": 268
    },
    {
      "epoch": 0.135362938734432,
      "grad_norm": 0.4311372935771942,
      "learning_rate": 4.7056459817440544e-05,
      "loss": 0.87,
      "step": 269
    },
    {
      "epoch": 0.13586614668511762,
      "grad_norm": 0.358874648809433,
      "learning_rate": 4.6729843538492847e-05,
      "loss": 0.7542,
      "step": 270
    },
    {
      "epoch": 0.13636935463580324,
      "grad_norm": 0.3988489806652069,
      "learning_rate": 4.640336734216403e-05,
      "loss": 0.8881,
      "step": 271
    },
    {
      "epoch": 0.13687256258648886,
      "grad_norm": 0.39466479420661926,
      "learning_rate": 4.607704521360776e-05,
      "loss": 0.7696,
      "step": 272
    },
    {
      "epoch": 0.13737577053717448,
      "grad_norm": 0.3718567192554474,
      "learning_rate": 4.575089113137792e-05,
      "loss": 0.805,
      "step": 273
    },
    {
      "epoch": 0.1378789784878601,
      "grad_norm": 0.3889524042606354,
      "learning_rate": 4.542491906682989e-05,
      "loss": 0.8586,
      "step": 274
    },
    {
      "epoch": 0.13838218643854572,
      "grad_norm": 0.41434207558631897,
      "learning_rate": 4.509914298352197e-05,
      "loss": 0.8551,
      "step": 275
    },
    {
      "epoch": 0.13888539438923134,
      "grad_norm": 0.4007040560245514,
      "learning_rate": 4.477357683661734e-05,
      "loss": 0.7777,
      "step": 276
    },
    {
      "epoch": 0.13938860233991698,
      "grad_norm": 0.46113258600234985,
      "learning_rate": 4.444823457228612e-05,
      "loss": 0.979,
      "step": 277
    },
    {
      "epoch": 0.1398918102906026,
      "grad_norm": 0.41809511184692383,
      "learning_rate": 4.412313012710813e-05,
      "loss": 0.7977,
      "step": 278
    },
    {
      "epoch": 0.14039501824128822,
      "grad_norm": 0.41079187393188477,
      "learning_rate": 4.379827742747575e-05,
      "loss": 0.769,
      "step": 279
    },
    {
      "epoch": 0.14089822619197384,
      "grad_norm": 0.449725478887558,
      "learning_rate": 4.347369038899744e-05,
      "loss": 0.9074,
      "step": 280
    },
    {
      "epoch": 0.14140143414265946,
      "grad_norm": 0.4488917291164398,
      "learning_rate": 4.3149382915901606e-05,
      "loss": 0.7963,
      "step": 281
    },
    {
      "epoch": 0.14190464209334508,
      "grad_norm": 0.46613508462905884,
      "learning_rate": 4.282536890044104e-05,
      "loss": 0.8518,
      "step": 282
    },
    {
      "epoch": 0.1424078500440307,
      "grad_norm": 0.4330773651599884,
      "learning_rate": 4.250166222229774e-05,
      "loss": 0.7441,
      "step": 283
    },
    {
      "epoch": 0.14291105799471632,
      "grad_norm": 0.5291471481323242,
      "learning_rate": 4.2178276747988446e-05,
      "loss": 0.7925,
      "step": 284
    },
    {
      "epoch": 0.14341426594540194,
      "grad_norm": 0.5655127763748169,
      "learning_rate": 4.185522633027057e-05,
      "loss": 0.8825,
      "step": 285
    },
    {
      "epoch": 0.14391747389608756,
      "grad_norm": 0.6190987229347229,
      "learning_rate": 4.153252480754877e-05,
      "loss": 0.967,
      "step": 286
    },
    {
      "epoch": 0.14442068184677317,
      "grad_norm": 0.6031071543693542,
      "learning_rate": 4.1210186003282275e-05,
      "loss": 0.9445,
      "step": 287
    },
    {
      "epoch": 0.1449238897974588,
      "grad_norm": 0.6229695677757263,
      "learning_rate": 4.088822372539263e-05,
      "loss": 0.847,
      "step": 288
    },
    {
      "epoch": 0.1454270977481444,
      "grad_norm": 0.6028884053230286,
      "learning_rate": 4.0566651765672246e-05,
      "loss": 0.775,
      "step": 289
    },
    {
      "epoch": 0.14593030569883003,
      "grad_norm": 0.6752098798751831,
      "learning_rate": 4.0245483899193595e-05,
      "loss": 0.9906,
      "step": 290
    },
    {
      "epoch": 0.14643351364951565,
      "grad_norm": 0.7151780724525452,
      "learning_rate": 3.992473388371915e-05,
      "loss": 0.9065,
      "step": 291
    },
    {
      "epoch": 0.14693672160020127,
      "grad_norm": 0.6899400949478149,
      "learning_rate": 3.960441545911204e-05,
      "loss": 0.8839,
      "step": 292
    },
    {
      "epoch": 0.14743992955088692,
      "grad_norm": 0.8318858742713928,
      "learning_rate": 3.928454234674747e-05,
      "loss": 0.9673,
      "step": 293
    },
    {
      "epoch": 0.14794313750157254,
      "grad_norm": 0.7379890084266663,
      "learning_rate": 3.896512824892495e-05,
      "loss": 0.9573,
      "step": 294
    },
    {
      "epoch": 0.14844634545225815,
      "grad_norm": 0.8325116038322449,
      "learning_rate": 3.864618684828134e-05,
      "loss": 1.0124,
      "step": 295
    },
    {
      "epoch": 0.14894955340294377,
      "grad_norm": 0.974595308303833,
      "learning_rate": 3.832773180720475e-05,
      "loss": 1.0537,
      "step": 296
    },
    {
      "epoch": 0.1494527613536294,
      "grad_norm": 0.938064694404602,
      "learning_rate": 3.800977676724919e-05,
      "loss": 1.1143,
      "step": 297
    },
    {
      "epoch": 0.149955969304315,
      "grad_norm": 1.158764362335205,
      "learning_rate": 3.769233534855035e-05,
      "loss": 1.0022,
      "step": 298
    },
    {
      "epoch": 0.15045917725500063,
      "grad_norm": 1.1322569847106934,
      "learning_rate": 3.73754211492421e-05,
      "loss": 1.1288,
      "step": 299
    },
    {
      "epoch": 0.15096238520568625,
      "grad_norm": 1.96816086769104,
      "learning_rate": 3.705904774487396e-05,
      "loss": 0.9652,
      "step": 300
    },
    {
      "epoch": 0.15096238520568625,
      "eval_loss": 0.8184122443199158,
      "eval_runtime": 315.2685,
      "eval_samples_per_second": 10.616,
      "eval_steps_per_second": 2.655,
      "step": 300
    },
    {
      "epoch": 0.15146559315637187,
      "grad_norm": 0.27112269401550293,
      "learning_rate": 3.6743228687829595e-05,
      "loss": 0.5626,
      "step": 301
    },
    {
      "epoch": 0.1519688011070575,
      "grad_norm": 0.2685532867908478,
      "learning_rate": 3.642797750674629e-05,
      "loss": 0.4869,
      "step": 302
    },
    {
      "epoch": 0.1524720090577431,
      "grad_norm": 0.2729337215423584,
      "learning_rate": 3.6113307705935396e-05,
      "loss": 0.6047,
      "step": 303
    },
    {
      "epoch": 0.15297521700842873,
      "grad_norm": 0.22421544790267944,
      "learning_rate": 3.579923276480387e-05,
      "loss": 0.4895,
      "step": 304
    },
    {
      "epoch": 0.15347842495911435,
      "grad_norm": 0.27264493703842163,
      "learning_rate": 3.5485766137276894e-05,
      "loss": 0.7219,
      "step": 305
    },
    {
      "epoch": 0.15398163290979996,
      "grad_norm": 0.23473694920539856,
      "learning_rate": 3.5172921251221455e-05,
      "loss": 0.538,
      "step": 306
    },
    {
      "epoch": 0.15448484086048558,
      "grad_norm": 0.28241899609565735,
      "learning_rate": 3.486071150787128e-05,
      "loss": 0.603,
      "step": 307
    },
    {
      "epoch": 0.1549880488111712,
      "grad_norm": 0.2775544822216034,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 0.6043,
      "step": 308
    },
    {
      "epoch": 0.15549125676185685,
      "grad_norm": 0.3308035135269165,
      "learning_rate": 3.423825091761153e-05,
      "loss": 0.8242,
      "step": 309
    },
    {
      "epoch": 0.15599446471254247,
      "grad_norm": 0.3018944561481476,
      "learning_rate": 3.392802673484193e-05,
      "loss": 0.7775,
      "step": 310
    },
    {
      "epoch": 0.1564976726632281,
      "grad_norm": 0.3263493776321411,
      "learning_rate": 3.361849102191533e-05,
      "loss": 0.8695,
      "step": 311
    },
    {
      "epoch": 0.1570008806139137,
      "grad_norm": 0.3123197853565216,
      "learning_rate": 3.330965703831146e-05,
      "loss": 0.6601,
      "step": 312
    },
    {
      "epoch": 0.15750408856459933,
      "grad_norm": 0.3234642744064331,
      "learning_rate": 3.300153801345028e-05,
      "loss": 0.7108,
      "step": 313
    },
    {
      "epoch": 0.15800729651528495,
      "grad_norm": 0.3345732092857361,
      "learning_rate": 3.2694147146125345e-05,
      "loss": 0.8111,
      "step": 314
    },
    {
      "epoch": 0.15851050446597056,
      "grad_norm": 0.3351074159145355,
      "learning_rate": 3.2387497603938326e-05,
      "loss": 0.7729,
      "step": 315
    },
    {
      "epoch": 0.15901371241665618,
      "grad_norm": 0.3468862771987915,
      "learning_rate": 3.2081602522734986e-05,
      "loss": 0.7321,
      "step": 316
    },
    {
      "epoch": 0.1595169203673418,
      "grad_norm": 0.36144280433654785,
      "learning_rate": 3.177647500604252e-05,
      "loss": 0.808,
      "step": 317
    },
    {
      "epoch": 0.16002012831802742,
      "grad_norm": 0.3642672300338745,
      "learning_rate": 3.147212812450819e-05,
      "loss": 0.8032,
      "step": 318
    },
    {
      "epoch": 0.16052333626871304,
      "grad_norm": 0.34578797221183777,
      "learning_rate": 3.116857491533947e-05,
      "loss": 0.8504,
      "step": 319
    },
    {
      "epoch": 0.16102654421939866,
      "grad_norm": 0.3759293258190155,
      "learning_rate": 3.086582838174551e-05,
      "loss": 0.7563,
      "step": 320
    },
    {
      "epoch": 0.16152975217008428,
      "grad_norm": 0.3771149516105652,
      "learning_rate": 3.056390149238022e-05,
      "loss": 0.827,
      "step": 321
    },
    {
      "epoch": 0.1620329601207699,
      "grad_norm": 0.39360910654067993,
      "learning_rate": 3.0262807180786647e-05,
      "loss": 0.84,
      "step": 322
    },
    {
      "epoch": 0.16253616807145552,
      "grad_norm": 0.39732182025909424,
      "learning_rate": 2.996255834484296e-05,
      "loss": 0.8707,
      "step": 323
    },
    {
      "epoch": 0.16303937602214116,
      "grad_norm": 0.3919753134250641,
      "learning_rate": 2.9663167846209998e-05,
      "loss": 0.82,
      "step": 324
    },
    {
      "epoch": 0.16354258397282678,
      "grad_norm": 0.4043086767196655,
      "learning_rate": 2.936464850978027e-05,
      "loss": 0.8707,
      "step": 325
    },
    {
      "epoch": 0.1640457919235124,
      "grad_norm": 0.37297871708869934,
      "learning_rate": 2.9067013123128613e-05,
      "loss": 0.7362,
      "step": 326
    },
    {
      "epoch": 0.16454899987419802,
      "grad_norm": 0.4036833941936493,
      "learning_rate": 2.8770274435964355e-05,
      "loss": 0.7927,
      "step": 327
    },
    {
      "epoch": 0.16505220782488364,
      "grad_norm": 0.42356228828430176,
      "learning_rate": 2.8474445159585235e-05,
      "loss": 0.8197,
      "step": 328
    },
    {
      "epoch": 0.16555541577556926,
      "grad_norm": 0.4919668734073639,
      "learning_rate": 2.8179537966332887e-05,
      "loss": 0.9665,
      "step": 329
    },
    {
      "epoch": 0.16605862372625488,
      "grad_norm": 0.46561524271965027,
      "learning_rate": 2.7885565489049946e-05,
      "loss": 0.7955,
      "step": 330
    },
    {
      "epoch": 0.1665618316769405,
      "grad_norm": 0.5371547341346741,
      "learning_rate": 2.759254032053888e-05,
      "loss": 0.9057,
      "step": 331
    },
    {
      "epoch": 0.16706503962762612,
      "grad_norm": 0.5103256106376648,
      "learning_rate": 2.7300475013022663e-05,
      "loss": 0.808,
      "step": 332
    },
    {
      "epoch": 0.16756824757831174,
      "grad_norm": 0.5322546362876892,
      "learning_rate": 2.700938207760701e-05,
      "loss": 0.7863,
      "step": 333
    },
    {
      "epoch": 0.16807145552899735,
      "grad_norm": 0.57713383436203,
      "learning_rate": 2.671927398374443e-05,
      "loss": 0.7858,
      "step": 334
    },
    {
      "epoch": 0.16857466347968297,
      "grad_norm": 0.49595826864242554,
      "learning_rate": 2.6430163158700115e-05,
      "loss": 0.716,
      "step": 335
    },
    {
      "epoch": 0.1690778714303686,
      "grad_norm": 0.5743617415428162,
      "learning_rate": 2.6142061987019577e-05,
      "loss": 0.7695,
      "step": 336
    },
    {
      "epoch": 0.1695810793810542,
      "grad_norm": 0.6702653169631958,
      "learning_rate": 2.5854982809998153e-05,
      "loss": 0.8607,
      "step": 337
    },
    {
      "epoch": 0.17008428733173983,
      "grad_norm": 0.5764122605323792,
      "learning_rate": 2.556893792515227e-05,
      "loss": 0.8009,
      "step": 338
    },
    {
      "epoch": 0.17058749528242545,
      "grad_norm": 0.6570090651512146,
      "learning_rate": 2.5283939585692783e-05,
      "loss": 0.9141,
      "step": 339
    },
    {
      "epoch": 0.1710907032331111,
      "grad_norm": 0.6721312999725342,
      "learning_rate": 2.500000000000001e-05,
      "loss": 0.9218,
      "step": 340
    },
    {
      "epoch": 0.17159391118379672,
      "grad_norm": 0.7778568863868713,
      "learning_rate": 2.471713133110078e-05,
      "loss": 0.9495,
      "step": 341
    },
    {
      "epoch": 0.17209711913448233,
      "grad_norm": 0.9691055417060852,
      "learning_rate": 2.4435345696147403e-05,
      "loss": 1.2298,
      "step": 342
    },
    {
      "epoch": 0.17260032708516795,
      "grad_norm": 0.7954840660095215,
      "learning_rate": 2.4154655165898627e-05,
      "loss": 0.9362,
      "step": 343
    },
    {
      "epoch": 0.17310353503585357,
      "grad_norm": 0.7800099849700928,
      "learning_rate": 2.3875071764202563e-05,
      "loss": 0.8807,
      "step": 344
    },
    {
      "epoch": 0.1736067429865392,
      "grad_norm": 0.8425636291503906,
      "learning_rate": 2.3596607467481603e-05,
      "loss": 0.9545,
      "step": 345
    },
    {
      "epoch": 0.1741099509372248,
      "grad_norm": 0.8894269466400146,
      "learning_rate": 2.3319274204219428e-05,
      "loss": 0.8623,
      "step": 346
    },
    {
      "epoch": 0.17461315888791043,
      "grad_norm": 0.8730196952819824,
      "learning_rate": 2.3043083854449988e-05,
      "loss": 0.8751,
      "step": 347
    },
    {
      "epoch": 0.17511636683859605,
      "grad_norm": 1.0998096466064453,
      "learning_rate": 2.2768048249248648e-05,
      "loss": 0.8707,
      "step": 348
    },
    {
      "epoch": 0.17561957478928167,
      "grad_norm": 1.1269233226776123,
      "learning_rate": 2.2494179170225333e-05,
      "loss": 0.8427,
      "step": 349
    },
    {
      "epoch": 0.1761227827399673,
      "grad_norm": 1.5910190343856812,
      "learning_rate": 2.2221488349019903e-05,
      "loss": 0.9082,
      "step": 350
    },
    {
      "epoch": 0.1766259906906529,
      "grad_norm": 0.24401302635669708,
      "learning_rate": 2.194998746679952e-05,
      "loss": 0.4272,
      "step": 351
    },
    {
      "epoch": 0.17712919864133853,
      "grad_norm": 0.23979057371616364,
      "learning_rate": 2.167968815375837e-05,
      "loss": 0.3934,
      "step": 352
    },
    {
      "epoch": 0.17763240659202414,
      "grad_norm": 0.2592470943927765,
      "learning_rate": 2.1410601988619394e-05,
      "loss": 0.5218,
      "step": 353
    },
    {
      "epoch": 0.17813561454270976,
      "grad_norm": 0.2586510181427002,
      "learning_rate": 2.1142740498138324e-05,
      "loss": 0.6844,
      "step": 354
    },
    {
      "epoch": 0.17863882249339538,
      "grad_norm": 0.26831480860710144,
      "learning_rate": 2.08761151566099e-05,
      "loss": 0.6302,
      "step": 355
    },
    {
      "epoch": 0.17914203044408103,
      "grad_norm": 0.22921589016914368,
      "learning_rate": 2.061073738537635e-05,
      "loss": 0.5133,
      "step": 356
    },
    {
      "epoch": 0.17964523839476665,
      "grad_norm": 0.27421972155570984,
      "learning_rate": 2.034661855233815e-05,
      "loss": 0.6273,
      "step": 357
    },
    {
      "epoch": 0.18014844634545227,
      "grad_norm": 0.3044654428958893,
      "learning_rate": 2.008376997146705e-05,
      "loss": 0.7552,
      "step": 358
    },
    {
      "epoch": 0.1806516542961379,
      "grad_norm": 0.32453665137290955,
      "learning_rate": 1.982220290232143e-05,
      "loss": 0.7021,
      "step": 359
    },
    {
      "epoch": 0.1811548622468235,
      "grad_norm": 0.31701529026031494,
      "learning_rate": 1.9561928549563968e-05,
      "loss": 0.7761,
      "step": 360
    },
    {
      "epoch": 0.18165807019750912,
      "grad_norm": 0.3049759268760681,
      "learning_rate": 1.9302958062481673e-05,
      "loss": 0.6762,
      "step": 361
    },
    {
      "epoch": 0.18216127814819474,
      "grad_norm": 0.3287789821624756,
      "learning_rate": 1.9045302534508297e-05,
      "loss": 0.6225,
      "step": 362
    },
    {
      "epoch": 0.18266448609888036,
      "grad_norm": 0.31470227241516113,
      "learning_rate": 1.8788973002749112e-05,
      "loss": 0.7272,
      "step": 363
    },
    {
      "epoch": 0.18316769404956598,
      "grad_norm": 0.3436526954174042,
      "learning_rate": 1.8533980447508137e-05,
      "loss": 0.7964,
      "step": 364
    },
    {
      "epoch": 0.1836709020002516,
      "grad_norm": 0.36613336205482483,
      "learning_rate": 1.8280335791817733e-05,
      "loss": 0.7995,
      "step": 365
    },
    {
      "epoch": 0.18417410995093722,
      "grad_norm": 0.38511422276496887,
      "learning_rate": 1.8028049900970767e-05,
      "loss": 0.8626,
      "step": 366
    },
    {
      "epoch": 0.18467731790162284,
      "grad_norm": 0.353153258562088,
      "learning_rate": 1.777713358205514e-05,
      "loss": 0.7844,
      "step": 367
    },
    {
      "epoch": 0.18518052585230846,
      "grad_norm": 0.4062706530094147,
      "learning_rate": 1.7527597583490822e-05,
      "loss": 0.9175,
      "step": 368
    },
    {
      "epoch": 0.18568373380299408,
      "grad_norm": 0.3805887997150421,
      "learning_rate": 1.7279452594569483e-05,
      "loss": 0.8583,
      "step": 369
    },
    {
      "epoch": 0.1861869417536797,
      "grad_norm": 0.39910799264907837,
      "learning_rate": 1.703270924499656e-05,
      "loss": 0.7456,
      "step": 370
    },
    {
      "epoch": 0.18669014970436532,
      "grad_norm": 0.4065895974636078,
      "learning_rate": 1.678737810443593e-05,
      "loss": 0.8629,
      "step": 371
    },
    {
      "epoch": 0.18719335765505096,
      "grad_norm": 0.3785150349140167,
      "learning_rate": 1.6543469682057106e-05,
      "loss": 0.9208,
      "step": 372
    },
    {
      "epoch": 0.18769656560573658,
      "grad_norm": 0.4127970039844513,
      "learning_rate": 1.6300994426085103e-05,
      "loss": 0.854,
      "step": 373
    },
    {
      "epoch": 0.1881997735564222,
      "grad_norm": 0.39115816354751587,
      "learning_rate": 1.605996272335291e-05,
      "loss": 0.7288,
      "step": 374
    },
    {
      "epoch": 0.18870298150710782,
      "grad_norm": 0.8231402635574341,
      "learning_rate": 1.5820384898856434e-05,
      "loss": 0.9092,
      "step": 375
    },
    {
      "epoch": 0.18920618945779344,
      "grad_norm": 0.40067026019096375,
      "learning_rate": 1.5582271215312294e-05,
      "loss": 0.7805,
      "step": 376
    },
    {
      "epoch": 0.18970939740847906,
      "grad_norm": 0.42449265718460083,
      "learning_rate": 1.5345631872718214e-05,
      "loss": 0.824,
      "step": 377
    },
    {
      "epoch": 0.19021260535916468,
      "grad_norm": 0.4465072751045227,
      "learning_rate": 1.5110477007916001e-05,
      "loss": 0.9012,
      "step": 378
    },
    {
      "epoch": 0.1907158133098503,
      "grad_norm": 0.4620152413845062,
      "learning_rate": 1.4876816694157419e-05,
      "loss": 0.7812,
      "step": 379
    },
    {
      "epoch": 0.19121902126053592,
      "grad_norm": 0.4710691273212433,
      "learning_rate": 1.4644660940672627e-05,
      "loss": 0.79,
      "step": 380
    },
    {
      "epoch": 0.19172222921122153,
      "grad_norm": 0.5064547657966614,
      "learning_rate": 1.4414019692241437e-05,
      "loss": 0.8933,
      "step": 381
    },
    {
      "epoch": 0.19222543716190715,
      "grad_norm": 0.46711280941963196,
      "learning_rate": 1.4184902828767287e-05,
      "loss": 0.7944,
      "step": 382
    },
    {
      "epoch": 0.19272864511259277,
      "grad_norm": 0.5075110793113708,
      "learning_rate": 1.3957320164854059e-05,
      "loss": 0.897,
      "step": 383
    },
    {
      "epoch": 0.1932318530632784,
      "grad_norm": 0.496896892786026,
      "learning_rate": 1.373128144938563e-05,
      "loss": 0.7915,
      "step": 384
    },
    {
      "epoch": 0.193735061013964,
      "grad_norm": 0.5214306712150574,
      "learning_rate": 1.3506796365108232e-05,
      "loss": 0.778,
      "step": 385
    },
    {
      "epoch": 0.19423826896464963,
      "grad_norm": 0.5623003244400024,
      "learning_rate": 1.3283874528215733e-05,
      "loss": 0.7346,
      "step": 386
    },
    {
      "epoch": 0.19474147691533525,
      "grad_norm": 0.5645301938056946,
      "learning_rate": 1.3062525487937699e-05,
      "loss": 0.7334,
      "step": 387
    },
    {
      "epoch": 0.1952446848660209,
      "grad_norm": 0.656359076499939,
      "learning_rate": 1.2842758726130283e-05,
      "loss": 0.9588,
      "step": 388
    },
    {
      "epoch": 0.19574789281670651,
      "grad_norm": 0.6827101707458496,
      "learning_rate": 1.2624583656870154e-05,
      "loss": 0.9212,
      "step": 389
    },
    {
      "epoch": 0.19625110076739213,
      "grad_norm": 0.6078924536705017,
      "learning_rate": 1.2408009626051137e-05,
      "loss": 0.8709,
      "step": 390
    },
    {
      "epoch": 0.19675430871807775,
      "grad_norm": 0.716253399848938,
      "learning_rate": 1.2193045910983863e-05,
      "loss": 1.0122,
      "step": 391
    },
    {
      "epoch": 0.19725751666876337,
      "grad_norm": 0.701407253742218,
      "learning_rate": 1.1979701719998453e-05,
      "loss": 0.8495,
      "step": 392
    },
    {
      "epoch": 0.197760724619449,
      "grad_norm": 0.6870536208152771,
      "learning_rate": 1.1767986192049984e-05,
      "loss": 0.8584,
      "step": 393
    },
    {
      "epoch": 0.1982639325701346,
      "grad_norm": 0.8282470107078552,
      "learning_rate": 1.1557908396327028e-05,
      "loss": 0.9368,
      "step": 394
    },
    {
      "epoch": 0.19876714052082023,
      "grad_norm": 0.8186725974082947,
      "learning_rate": 1.134947733186315e-05,
      "loss": 1.044,
      "step": 395
    },
    {
      "epoch": 0.19927034847150585,
      "grad_norm": 0.8825912475585938,
      "learning_rate": 1.1142701927151456e-05,
      "loss": 0.9584,
      "step": 396
    },
    {
      "epoch": 0.19977355642219147,
      "grad_norm": 1.0319724082946777,
      "learning_rate": 1.0937591039762085e-05,
      "loss": 1.0113,
      "step": 397
    },
    {
      "epoch": 0.2002767643728771,
      "grad_norm": 0.8762199282646179,
      "learning_rate": 1.0734153455962765e-05,
      "loss": 1.0534,
      "step": 398
    },
    {
      "epoch": 0.2007799723235627,
      "grad_norm": 1.0712343454360962,
      "learning_rate": 1.0532397890342505e-05,
      "loss": 0.961,
      "step": 399
    },
    {
      "epoch": 0.20128318027424832,
      "grad_norm": 1.485298752784729,
      "learning_rate": 1.0332332985438248e-05,
      "loss": 0.9206,
      "step": 400
    },
    {
      "epoch": 0.20128318027424832,
      "eval_loss": 0.8115440011024475,
      "eval_runtime": 314.4876,
      "eval_samples_per_second": 10.643,
      "eval_steps_per_second": 2.661,
      "step": 400
    },
    {
      "epoch": 0.20178638822493394,
      "grad_norm": 0.2739810049533844,
      "learning_rate": 1.013396731136465e-05,
      "loss": 0.5771,
      "step": 401
    },
    {
      "epoch": 0.20228959617561956,
      "grad_norm": 0.32150766253471375,
      "learning_rate": 9.937309365446973e-06,
      "loss": 0.5294,
      "step": 402
    },
    {
      "epoch": 0.2027928041263052,
      "grad_norm": 0.23200996220111847,
      "learning_rate": 9.742367571857091e-06,
      "loss": 0.5972,
      "step": 403
    },
    {
      "epoch": 0.20329601207699083,
      "grad_norm": 0.23436395823955536,
      "learning_rate": 9.549150281252633e-06,
      "loss": 0.5296,
      "step": 404
    },
    {
      "epoch": 0.20379922002767645,
      "grad_norm": 0.22071292996406555,
      "learning_rate": 9.357665770419244e-06,
      "loss": 0.4927,
      "step": 405
    },
    {
      "epoch": 0.20430242797836207,
      "grad_norm": 0.24270054697990417,
      "learning_rate": 9.167922241916055e-06,
      "loss": 0.5776,
      "step": 406
    },
    {
      "epoch": 0.20480563592904769,
      "grad_norm": 0.2507602572441101,
      "learning_rate": 8.97992782372432e-06,
      "loss": 0.6028,
      "step": 407
    },
    {
      "epoch": 0.2053088438797333,
      "grad_norm": 0.3003790080547333,
      "learning_rate": 8.793690568899216e-06,
      "loss": 0.7235,
      "step": 408
    },
    {
      "epoch": 0.20581205183041892,
      "grad_norm": 0.30374565720558167,
      "learning_rate": 8.609218455224893e-06,
      "loss": 0.7252,
      "step": 409
    },
    {
      "epoch": 0.20631525978110454,
      "grad_norm": 0.30985498428344727,
      "learning_rate": 8.426519384872733e-06,
      "loss": 0.7916,
      "step": 410
    },
    {
      "epoch": 0.20681846773179016,
      "grad_norm": 0.3453262746334076,
      "learning_rate": 8.245601184062852e-06,
      "loss": 0.7118,
      "step": 411
    },
    {
      "epoch": 0.20732167568247578,
      "grad_norm": 0.37387192249298096,
      "learning_rate": 8.066471602728803e-06,
      "loss": 0.7777,
      "step": 412
    },
    {
      "epoch": 0.2078248836331614,
      "grad_norm": 0.32172510027885437,
      "learning_rate": 7.889138314185678e-06,
      "loss": 0.6585,
      "step": 413
    },
    {
      "epoch": 0.20832809158384702,
      "grad_norm": 0.3337942659854889,
      "learning_rate": 7.71360891480134e-06,
      "loss": 0.705,
      "step": 414
    },
    {
      "epoch": 0.20883129953453264,
      "grad_norm": 0.3572860062122345,
      "learning_rate": 7.539890923671062e-06,
      "loss": 0.7869,
      "step": 415
    },
    {
      "epoch": 0.20933450748521826,
      "grad_norm": 0.40350982546806335,
      "learning_rate": 7.367991782295391e-06,
      "loss": 0.7875,
      "step": 416
    },
    {
      "epoch": 0.20983771543590388,
      "grad_norm": 0.3487812578678131,
      "learning_rate": 7.197918854261432e-06,
      "loss": 0.7259,
      "step": 417
    },
    {
      "epoch": 0.2103409233865895,
      "grad_norm": 0.337096631526947,
      "learning_rate": 7.029679424927365e-06,
      "loss": 0.7244,
      "step": 418
    },
    {
      "epoch": 0.21084413133727514,
      "grad_norm": 0.42285338044166565,
      "learning_rate": 6.863280701110408e-06,
      "loss": 0.9008,
      "step": 419
    },
    {
      "epoch": 0.21134733928796076,
      "grad_norm": 0.40784189105033875,
      "learning_rate": 6.698729810778065e-06,
      "loss": 0.8874,
      "step": 420
    },
    {
      "epoch": 0.21185054723864638,
      "grad_norm": 0.3879254460334778,
      "learning_rate": 6.536033802742813e-06,
      "loss": 0.8339,
      "step": 421
    },
    {
      "epoch": 0.212353755189332,
      "grad_norm": 0.3816598951816559,
      "learning_rate": 6.375199646360142e-06,
      "loss": 0.7493,
      "step": 422
    },
    {
      "epoch": 0.21285696314001762,
      "grad_norm": 0.39769789576530457,
      "learning_rate": 6.216234231230012e-06,
      "loss": 0.8177,
      "step": 423
    },
    {
      "epoch": 0.21336017109070324,
      "grad_norm": 0.38587310910224915,
      "learning_rate": 6.059144366901736e-06,
      "loss": 0.7676,
      "step": 424
    },
    {
      "epoch": 0.21386337904138886,
      "grad_norm": 0.42682746052742004,
      "learning_rate": 5.903936782582253e-06,
      "loss": 0.8007,
      "step": 425
    },
    {
      "epoch": 0.21436658699207448,
      "grad_norm": 0.42841973900794983,
      "learning_rate": 5.750618126847912e-06,
      "loss": 0.9058,
      "step": 426
    },
    {
      "epoch": 0.2148697949427601,
      "grad_norm": 0.3857778310775757,
      "learning_rate": 5.599194967359639e-06,
      "loss": 0.6963,
      "step": 427
    },
    {
      "epoch": 0.21537300289344571,
      "grad_norm": 0.43843403458595276,
      "learning_rate": 5.449673790581611e-06,
      "loss": 0.8127,
      "step": 428
    },
    {
      "epoch": 0.21587621084413133,
      "grad_norm": 0.4387761950492859,
      "learning_rate": 5.302061001503394e-06,
      "loss": 0.8608,
      "step": 429
    },
    {
      "epoch": 0.21637941879481695,
      "grad_norm": 0.4464000463485718,
      "learning_rate": 5.156362923365588e-06,
      "loss": 0.7609,
      "step": 430
    },
    {
      "epoch": 0.21688262674550257,
      "grad_norm": 0.5585458278656006,
      "learning_rate": 5.012585797388936e-06,
      "loss": 0.939,
      "step": 431
    },
    {
      "epoch": 0.2173858346961882,
      "grad_norm": 0.5143706202507019,
      "learning_rate": 4.87073578250698e-06,
      "loss": 0.9182,
      "step": 432
    },
    {
      "epoch": 0.2178890426468738,
      "grad_norm": 0.48482105135917664,
      "learning_rate": 4.730818955102234e-06,
      "loss": 0.7851,
      "step": 433
    },
    {
      "epoch": 0.21839225059755943,
      "grad_norm": 0.5352709293365479,
      "learning_rate": 4.592841308745932e-06,
      "loss": 0.8665,
      "step": 434
    },
    {
      "epoch": 0.21889545854824508,
      "grad_norm": 0.5815681219100952,
      "learning_rate": 4.456808753941205e-06,
      "loss": 0.8981,
      "step": 435
    },
    {
      "epoch": 0.2193986664989307,
      "grad_norm": 0.5692911744117737,
      "learning_rate": 4.322727117869951e-06,
      "loss": 0.8123,
      "step": 436
    },
    {
      "epoch": 0.2199018744496163,
      "grad_norm": 0.60055011510849,
      "learning_rate": 4.190602144143207e-06,
      "loss": 0.8607,
      "step": 437
    },
    {
      "epoch": 0.22040508240030193,
      "grad_norm": 0.6022657155990601,
      "learning_rate": 4.06043949255509e-06,
      "loss": 0.7886,
      "step": 438
    },
    {
      "epoch": 0.22090829035098755,
      "grad_norm": 0.5964546799659729,
      "learning_rate": 3.932244738840379e-06,
      "loss": 0.7872,
      "step": 439
    },
    {
      "epoch": 0.22141149830167317,
      "grad_norm": 0.6574018597602844,
      "learning_rate": 3.8060233744356633e-06,
      "loss": 0.7414,
      "step": 440
    },
    {
      "epoch": 0.2219147062523588,
      "grad_norm": 0.6827805638313293,
      "learning_rate": 3.681780806244095e-06,
      "loss": 0.8722,
      "step": 441
    },
    {
      "epoch": 0.2224179142030444,
      "grad_norm": 0.9296197891235352,
      "learning_rate": 3.5595223564037884e-06,
      "loss": 1.0292,
      "step": 442
    },
    {
      "epoch": 0.22292112215373003,
      "grad_norm": 0.8508293628692627,
      "learning_rate": 3.4392532620598216e-06,
      "loss": 1.0509,
      "step": 443
    },
    {
      "epoch": 0.22342433010441565,
      "grad_norm": 0.8861533999443054,
      "learning_rate": 3.3209786751399187e-06,
      "loss": 1.1006,
      "step": 444
    },
    {
      "epoch": 0.22392753805510127,
      "grad_norm": 0.978121817111969,
      "learning_rate": 3.2047036621337236e-06,
      "loss": 1.0166,
      "step": 445
    },
    {
      "epoch": 0.22443074600578689,
      "grad_norm": 0.8798397183418274,
      "learning_rate": 3.0904332038757977e-06,
      "loss": 0.8607,
      "step": 446
    },
    {
      "epoch": 0.2249339539564725,
      "grad_norm": 0.9314451813697815,
      "learning_rate": 2.978172195332263e-06,
      "loss": 1.1433,
      "step": 447
    },
    {
      "epoch": 0.22543716190715812,
      "grad_norm": 1.0907480716705322,
      "learning_rate": 2.8679254453910785e-06,
      "loss": 0.9332,
      "step": 448
    },
    {
      "epoch": 0.22594036985784374,
      "grad_norm": 1.2249211072921753,
      "learning_rate": 2.759697676656098e-06,
      "loss": 0.8817,
      "step": 449
    },
    {
      "epoch": 0.22644357780852936,
      "grad_norm": 1.5159759521484375,
      "learning_rate": 2.653493525244721e-06,
      "loss": 0.8945,
      "step": 450
    },
    {
      "epoch": 0.226946785759215,
      "grad_norm": 0.23161232471466064,
      "learning_rate": 2.549317540589308e-06,
      "loss": 0.4277,
      "step": 451
    },
    {
      "epoch": 0.22744999370990063,
      "grad_norm": 0.2513332962989807,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 0.4781,
      "step": 452
    },
    {
      "epoch": 0.22795320166058625,
      "grad_norm": 0.2884768843650818,
      "learning_rate": 2.3470678346851518e-06,
      "loss": 0.6854,
      "step": 453
    },
    {
      "epoch": 0.22845640961127187,
      "grad_norm": 0.2403653860092163,
      "learning_rate": 2.2490027771406687e-06,
      "loss": 0.6633,
      "step": 454
    },
    {
      "epoch": 0.22895961756195748,
      "grad_norm": 0.24366974830627441,
      "learning_rate": 2.152983213389559e-06,
      "loss": 0.5886,
      "step": 455
    },
    {
      "epoch": 0.2294628255126431,
      "grad_norm": 0.26889801025390625,
      "learning_rate": 2.0590132565903476e-06,
      "loss": 0.597,
      "step": 456
    },
    {
      "epoch": 0.22996603346332872,
      "grad_norm": 0.2592047154903412,
      "learning_rate": 1.9670969321032407e-06,
      "loss": 0.592,
      "step": 457
    },
    {
      "epoch": 0.23046924141401434,
      "grad_norm": 0.2693350911140442,
      "learning_rate": 1.8772381773176417e-06,
      "loss": 0.6015,
      "step": 458
    },
    {
      "epoch": 0.23097244936469996,
      "grad_norm": 0.2780812978744507,
      "learning_rate": 1.7894408414835362e-06,
      "loss": 0.6346,
      "step": 459
    },
    {
      "epoch": 0.23147565731538558,
      "grad_norm": 0.3135295510292053,
      "learning_rate": 1.70370868554659e-06,
      "loss": 0.6892,
      "step": 460
    },
    {
      "epoch": 0.2319788652660712,
      "grad_norm": 0.2802668809890747,
      "learning_rate": 1.620045381987012e-06,
      "loss": 0.6123,
      "step": 461
    },
    {
      "epoch": 0.23248207321675682,
      "grad_norm": 0.3362366855144501,
      "learning_rate": 1.5384545146622852e-06,
      "loss": 0.8039,
      "step": 462
    },
    {
      "epoch": 0.23298528116744244,
      "grad_norm": 0.34370091557502747,
      "learning_rate": 1.4589395786535953e-06,
      "loss": 0.7433,
      "step": 463
    },
    {
      "epoch": 0.23348848911812806,
      "grad_norm": 0.30836600065231323,
      "learning_rate": 1.3815039801161721e-06,
      "loss": 0.7163,
      "step": 464
    },
    {
      "epoch": 0.23399169706881368,
      "grad_norm": 0.3486054241657257,
      "learning_rate": 1.3061510361333185e-06,
      "loss": 0.8756,
      "step": 465
    },
    {
      "epoch": 0.2344949050194993,
      "grad_norm": 0.35582247376441956,
      "learning_rate": 1.232883974574367e-06,
      "loss": 0.8586,
      "step": 466
    },
    {
      "epoch": 0.23499811297018494,
      "grad_norm": 0.3585059642791748,
      "learning_rate": 1.1617059339563807e-06,
      "loss": 0.8525,
      "step": 467
    },
    {
      "epoch": 0.23550132092087056,
      "grad_norm": 0.40273240208625793,
      "learning_rate": 1.0926199633097157e-06,
      "loss": 0.8213,
      "step": 468
    },
    {
      "epoch": 0.23600452887155618,
      "grad_norm": 0.3870213031768799,
      "learning_rate": 1.0256290220474307e-06,
      "loss": 0.821,
      "step": 469
    },
    {
      "epoch": 0.2365077368222418,
      "grad_norm": 0.3532085418701172,
      "learning_rate": 9.607359798384785e-07,
      "loss": 0.7606,
      "step": 470
    },
    {
      "epoch": 0.23701094477292742,
      "grad_norm": 0.3976037800312042,
      "learning_rate": 8.979436164848088e-07,
      "loss": 0.7803,
      "step": 471
    },
    {
      "epoch": 0.23751415272361304,
      "grad_norm": 0.4277723431587219,
      "learning_rate": 8.372546218022747e-07,
      "loss": 0.9184,
      "step": 472
    },
    {
      "epoch": 0.23801736067429866,
      "grad_norm": 0.38751596212387085,
      "learning_rate": 7.786715955054203e-07,
      "loss": 0.8418,
      "step": 473
    },
    {
      "epoch": 0.23852056862498427,
      "grad_norm": 0.392293244600296,
      "learning_rate": 7.221970470961125e-07,
      "loss": 0.8359,
      "step": 474
    },
    {
      "epoch": 0.2390237765756699,
      "grad_norm": 0.40017879009246826,
      "learning_rate": 6.678333957560512e-07,
      "loss": 0.8826,
      "step": 475
    },
    {
      "epoch": 0.2395269845263555,
      "grad_norm": 0.4077494144439697,
      "learning_rate": 6.15582970243117e-07,
      "loss": 0.8959,
      "step": 476
    },
    {
      "epoch": 0.24003019247704113,
      "grad_norm": 0.4908146858215332,
      "learning_rate": 5.654480087916303e-07,
      "loss": 0.9008,
      "step": 477
    },
    {
      "epoch": 0.24053340042772675,
      "grad_norm": 0.46262556314468384,
      "learning_rate": 5.174306590164879e-07,
      "loss": 0.8885,
      "step": 478
    },
    {
      "epoch": 0.24103660837841237,
      "grad_norm": 0.42933472990989685,
      "learning_rate": 4.715329778211375e-07,
      "loss": 0.7975,
      "step": 479
    },
    {
      "epoch": 0.241539816329098,
      "grad_norm": 0.46930062770843506,
      "learning_rate": 4.277569313094809e-07,
      "loss": 0.8802,
      "step": 480
    },
    {
      "epoch": 0.2420430242797836,
      "grad_norm": 0.4839300513267517,
      "learning_rate": 3.8610439470164737e-07,
      "loss": 0.8239,
      "step": 481
    },
    {
      "epoch": 0.24254623223046926,
      "grad_norm": 0.5210592746734619,
      "learning_rate": 3.465771522536854e-07,
      "loss": 0.8478,
      "step": 482
    },
    {
      "epoch": 0.24304944018115487,
      "grad_norm": 0.5569612979888916,
      "learning_rate": 3.09176897181096e-07,
      "loss": 0.8666,
      "step": 483
    },
    {
      "epoch": 0.2435526481318405,
      "grad_norm": 0.5378491878509521,
      "learning_rate": 2.7390523158633554e-07,
      "loss": 0.8229,
      "step": 484
    },
    {
      "epoch": 0.2440558560825261,
      "grad_norm": 0.5275207161903381,
      "learning_rate": 2.407636663901591e-07,
      "loss": 0.7684,
      "step": 485
    },
    {
      "epoch": 0.24455906403321173,
      "grad_norm": 0.5996338725090027,
      "learning_rate": 2.0975362126691712e-07,
      "loss": 0.8215,
      "step": 486
    },
    {
      "epoch": 0.24506227198389735,
      "grad_norm": 0.6922218203544617,
      "learning_rate": 1.8087642458373134e-07,
      "loss": 0.8545,
      "step": 487
    },
    {
      "epoch": 0.24556547993458297,
      "grad_norm": 0.6023167967796326,
      "learning_rate": 1.5413331334360182e-07,
      "loss": 0.8429,
      "step": 488
    },
    {
      "epoch": 0.2460686878852686,
      "grad_norm": 0.6781688928604126,
      "learning_rate": 1.2952543313240472e-07,
      "loss": 0.9136,
      "step": 489
    },
    {
      "epoch": 0.2465718958359542,
      "grad_norm": 0.597463846206665,
      "learning_rate": 1.0705383806982606e-07,
      "loss": 0.7757,
      "step": 490
    },
    {
      "epoch": 0.24707510378663983,
      "grad_norm": 0.6440747380256653,
      "learning_rate": 8.671949076420882e-08,
      "loss": 0.8941,
      "step": 491
    },
    {
      "epoch": 0.24757831173732545,
      "grad_norm": 0.7201170325279236,
      "learning_rate": 6.852326227130834e-08,
      "loss": 0.8291,
      "step": 492
    },
    {
      "epoch": 0.24808151968801107,
      "grad_norm": 0.8137110471725464,
      "learning_rate": 5.246593205699424e-08,
      "loss": 0.8731,
      "step": 493
    },
    {
      "epoch": 0.24858472763869668,
      "grad_norm": 0.769762396812439,
      "learning_rate": 3.8548187963854956e-08,
      "loss": 0.9042,
      "step": 494
    },
    {
      "epoch": 0.2490879355893823,
      "grad_norm": 0.7686090469360352,
      "learning_rate": 2.6770626181715773e-08,
      "loss": 1.01,
      "step": 495
    },
    {
      "epoch": 0.24959114354006792,
      "grad_norm": 0.8908253312110901,
      "learning_rate": 1.7133751222137007e-08,
      "loss": 0.9312,
      "step": 496
    },
    {
      "epoch": 0.25009435149075354,
      "grad_norm": 0.8430070877075195,
      "learning_rate": 9.637975896759077e-09,
      "loss": 0.9682,
      "step": 497
    },
    {
      "epoch": 0.2505975594414392,
      "grad_norm": 1.0221734046936035,
      "learning_rate": 4.2836212996499865e-09,
      "loss": 0.8625,
      "step": 498
    },
    {
      "epoch": 0.2511007673921248,
      "grad_norm": 1.10769784450531,
      "learning_rate": 1.0709167935385455e-09,
      "loss": 0.9659,
      "step": 499
    },
    {
      "epoch": 0.2516039753428104,
      "grad_norm": 1.3622961044311523,
      "learning_rate": 0.0,
      "loss": 0.8602,
      "step": 500
    },
    {
      "epoch": 0.2516039753428104,
      "eval_loss": 0.8085561394691467,
      "eval_runtime": 314.2655,
      "eval_samples_per_second": 10.65,
      "eval_steps_per_second": 2.663,
      "step": 500
    }
  ],
  "logging_steps": 1,
  "max_steps": 500,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 100,
  "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": 1.4280794742718464e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}