{
  "best_global_step": 2300,
  "best_metric": 0.7602248191833496,
  "best_model_checkpoint": "saves/qwen3-1.7B/Qwen3-1.7B-SFT-science-1e-5/checkpoint-2300",
  "epoch": 3.0,
  "eval_steps": 230,
  "global_step": 2313,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0012977269501388974,
      "grad_norm": 18.964550018310547,
      "learning_rate": 0.0,
      "loss": 1.341123104095459,
      "step": 1
    },
    {
      "epoch": 0.0025954539002777948,
      "grad_norm": 17.563854217529297,
      "learning_rate": 8.620689655172414e-08,
      "loss": 1.240975022315979,
      "step": 2
    },
    {
      "epoch": 0.003893180850416692,
      "grad_norm": 18.24137306213379,
      "learning_rate": 1.7241379310344828e-07,
      "loss": 1.3372515439987183,
      "step": 3
    },
    {
      "epoch": 0.0051909078005555895,
      "grad_norm": 19.47113800048828,
      "learning_rate": 2.5862068965517245e-07,
      "loss": 1.4065107107162476,
      "step": 4
    },
    {
      "epoch": 0.006488634750694487,
      "grad_norm": 17.434274673461914,
      "learning_rate": 3.4482758620689656e-07,
      "loss": 1.3200397491455078,
      "step": 5
    },
    {
      "epoch": 0.007786361700833384,
      "grad_norm": 18.338199615478516,
      "learning_rate": 4.3103448275862073e-07,
      "loss": 1.3047453165054321,
      "step": 6
    },
    {
      "epoch": 0.009084088650972282,
      "grad_norm": 19.371152877807617,
      "learning_rate": 5.172413793103449e-07,
      "loss": 1.282523512840271,
      "step": 7
    },
    {
      "epoch": 0.010381815601111179,
      "grad_norm": 17.54425621032715,
      "learning_rate": 6.034482758620691e-07,
      "loss": 1.2399928569793701,
      "step": 8
    },
    {
      "epoch": 0.011679542551250076,
      "grad_norm": 16.710424423217773,
      "learning_rate": 6.896551724137931e-07,
      "loss": 1.1754190921783447,
      "step": 9
    },
    {
      "epoch": 0.012977269501388973,
      "grad_norm": 17.185142517089844,
      "learning_rate": 7.758620689655173e-07,
      "loss": 1.2869014739990234,
      "step": 10
    },
    {
      "epoch": 0.01427499645152787,
      "grad_norm": 16.615570068359375,
      "learning_rate": 8.620689655172415e-07,
      "loss": 1.256578803062439,
      "step": 11
    },
    {
      "epoch": 0.015572723401666768,
      "grad_norm": 15.576726913452148,
      "learning_rate": 9.482758620689655e-07,
      "loss": 1.2133194208145142,
      "step": 12
    },
    {
      "epoch": 0.016870450351805667,
      "grad_norm": 15.714284896850586,
      "learning_rate": 1.0344827586206898e-06,
      "loss": 1.2498674392700195,
      "step": 13
    },
    {
      "epoch": 0.018168177301944564,
      "grad_norm": 15.78684139251709,
      "learning_rate": 1.120689655172414e-06,
      "loss": 1.235356092453003,
      "step": 14
    },
    {
      "epoch": 0.01946590425208346,
      "grad_norm": 14.27075481414795,
      "learning_rate": 1.2068965517241381e-06,
      "loss": 1.2290422916412354,
      "step": 15
    },
    {
      "epoch": 0.020763631202222358,
      "grad_norm": 12.652816772460938,
      "learning_rate": 1.2931034482758623e-06,
      "loss": 1.1969146728515625,
      "step": 16
    },
    {
      "epoch": 0.022061358152361255,
      "grad_norm": 12.271768569946289,
      "learning_rate": 1.3793103448275862e-06,
      "loss": 1.23379647731781,
      "step": 17
    },
    {
      "epoch": 0.023359085102500152,
      "grad_norm": 11.308884620666504,
      "learning_rate": 1.4655172413793104e-06,
      "loss": 1.1893874406814575,
      "step": 18
    },
    {
      "epoch": 0.02465681205263905,
      "grad_norm": 11.503107070922852,
      "learning_rate": 1.5517241379310346e-06,
      "loss": 1.2033772468566895,
      "step": 19
    },
    {
      "epoch": 0.025954539002777947,
      "grad_norm": 10.80382251739502,
      "learning_rate": 1.6379310344827587e-06,
      "loss": 1.1051026582717896,
      "step": 20
    },
    {
      "epoch": 0.027252265952916844,
      "grad_norm": 7.508317947387695,
      "learning_rate": 1.724137931034483e-06,
      "loss": 1.2109605073928833,
      "step": 21
    },
    {
      "epoch": 0.02854999290305574,
      "grad_norm": 6.523677349090576,
      "learning_rate": 1.810344827586207e-06,
      "loss": 1.0799193382263184,
      "step": 22
    },
    {
      "epoch": 0.029847719853194638,
      "grad_norm": 6.148171901702881,
      "learning_rate": 1.896551724137931e-06,
      "loss": 1.109641194343567,
      "step": 23
    },
    {
      "epoch": 0.031145446803333535,
      "grad_norm": 5.9545440673828125,
      "learning_rate": 1.982758620689655e-06,
      "loss": 1.0592708587646484,
      "step": 24
    },
    {
      "epoch": 0.032443173753472436,
      "grad_norm": 5.15648889541626,
      "learning_rate": 2.0689655172413796e-06,
      "loss": 1.044050693511963,
      "step": 25
    },
    {
      "epoch": 0.03374090070361133,
      "grad_norm": 4.722131252288818,
      "learning_rate": 2.1551724137931035e-06,
      "loss": 1.0839450359344482,
      "step": 26
    },
    {
      "epoch": 0.03503862765375023,
      "grad_norm": 4.583364963531494,
      "learning_rate": 2.241379310344828e-06,
      "loss": 1.0532729625701904,
      "step": 27
    },
    {
      "epoch": 0.03633635460388913,
      "grad_norm": 3.236476421356201,
      "learning_rate": 2.327586206896552e-06,
      "loss": 1.0504995584487915,
      "step": 28
    },
    {
      "epoch": 0.037634081554028025,
      "grad_norm": 2.3688511848449707,
      "learning_rate": 2.4137931034482762e-06,
      "loss": 1.0392580032348633,
      "step": 29
    },
    {
      "epoch": 0.03893180850416692,
      "grad_norm": 2.1980035305023193,
      "learning_rate": 2.5e-06,
      "loss": 1.0583772659301758,
      "step": 30
    },
    {
      "epoch": 0.04022953545430582,
      "grad_norm": 2.0409348011016846,
      "learning_rate": 2.5862068965517246e-06,
      "loss": 0.9697405695915222,
      "step": 31
    },
    {
      "epoch": 0.041527262404444716,
      "grad_norm": 2.0691511631011963,
      "learning_rate": 2.672413793103448e-06,
      "loss": 0.9957722425460815,
      "step": 32
    },
    {
      "epoch": 0.04282498935458361,
      "grad_norm": 1.81174898147583,
      "learning_rate": 2.7586206896551725e-06,
      "loss": 0.9902118444442749,
      "step": 33
    },
    {
      "epoch": 0.04412271630472251,
      "grad_norm": 1.8090497255325317,
      "learning_rate": 2.844827586206897e-06,
      "loss": 0.9399092793464661,
      "step": 34
    },
    {
      "epoch": 0.04542044325486141,
      "grad_norm": 1.7403111457824707,
      "learning_rate": 2.931034482758621e-06,
      "loss": 1.0072567462921143,
      "step": 35
    },
    {
      "epoch": 0.046718170205000305,
      "grad_norm": 1.588705062866211,
      "learning_rate": 3.017241379310345e-06,
      "loss": 1.0338897705078125,
      "step": 36
    },
    {
      "epoch": 0.0480158971551392,
      "grad_norm": 1.4160236120224,
      "learning_rate": 3.103448275862069e-06,
      "loss": 0.9362650513648987,
      "step": 37
    },
    {
      "epoch": 0.0493136241052781,
      "grad_norm": 1.4251086711883545,
      "learning_rate": 3.1896551724137935e-06,
      "loss": 0.9243776798248291,
      "step": 38
    },
    {
      "epoch": 0.050611351055416996,
      "grad_norm": 1.3788193464279175,
      "learning_rate": 3.2758620689655175e-06,
      "loss": 1.017249584197998,
      "step": 39
    },
    {
      "epoch": 0.05190907800555589,
      "grad_norm": 1.2869218587875366,
      "learning_rate": 3.362068965517242e-06,
      "loss": 0.9809561371803284,
      "step": 40
    },
    {
      "epoch": 0.05320680495569479,
      "grad_norm": 1.2191851139068604,
      "learning_rate": 3.448275862068966e-06,
      "loss": 0.9668518304824829,
      "step": 41
    },
    {
      "epoch": 0.05450453190583369,
      "grad_norm": 1.1019444465637207,
      "learning_rate": 3.5344827586206898e-06,
      "loss": 0.9050226807594299,
      "step": 42
    },
    {
      "epoch": 0.055802258855972585,
      "grad_norm": 1.1174124479293823,
      "learning_rate": 3.620689655172414e-06,
      "loss": 0.9214593172073364,
      "step": 43
    },
    {
      "epoch": 0.05709998580611148,
      "grad_norm": 1.0376768112182617,
      "learning_rate": 3.7068965517241385e-06,
      "loss": 0.9226070046424866,
      "step": 44
    },
    {
      "epoch": 0.05839771275625038,
      "grad_norm": 1.0166256427764893,
      "learning_rate": 3.793103448275862e-06,
      "loss": 0.9782117605209351,
      "step": 45
    },
    {
      "epoch": 0.059695439706389276,
      "grad_norm": 1.0689457654953003,
      "learning_rate": 3.8793103448275865e-06,
      "loss": 0.9090121388435364,
      "step": 46
    },
    {
      "epoch": 0.060993166656528174,
      "grad_norm": 0.9738860130310059,
      "learning_rate": 3.96551724137931e-06,
      "loss": 0.8602224588394165,
      "step": 47
    },
    {
      "epoch": 0.06229089360666707,
      "grad_norm": 0.9925969839096069,
      "learning_rate": 4.051724137931034e-06,
      "loss": 0.9660595059394836,
      "step": 48
    },
    {
      "epoch": 0.06358862055680597,
      "grad_norm": 0.9282276034355164,
      "learning_rate": 4.137931034482759e-06,
      "loss": 0.9271571040153503,
      "step": 49
    },
    {
      "epoch": 0.06488634750694487,
      "grad_norm": 0.8538461327552795,
      "learning_rate": 4.224137931034483e-06,
      "loss": 0.9294825196266174,
      "step": 50
    },
    {
      "epoch": 0.06618407445708377,
      "grad_norm": 0.8447784185409546,
      "learning_rate": 4.310344827586207e-06,
      "loss": 0.9223189353942871,
      "step": 51
    },
    {
      "epoch": 0.06748180140722267,
      "grad_norm": 0.9016849398612976,
      "learning_rate": 4.396551724137931e-06,
      "loss": 0.9693700075149536,
      "step": 52
    },
    {
      "epoch": 0.06877952835736156,
      "grad_norm": 0.8437271118164062,
      "learning_rate": 4.482758620689656e-06,
      "loss": 0.9623271822929382,
      "step": 53
    },
    {
      "epoch": 0.07007725530750046,
      "grad_norm": 0.8587411046028137,
      "learning_rate": 4.56896551724138e-06,
      "loss": 0.9010592103004456,
      "step": 54
    },
    {
      "epoch": 0.07137498225763936,
      "grad_norm": 0.8153098225593567,
      "learning_rate": 4.655172413793104e-06,
      "loss": 0.9027279615402222,
      "step": 55
    },
    {
      "epoch": 0.07267270920777825,
      "grad_norm": 0.7967116832733154,
      "learning_rate": 4.741379310344828e-06,
      "loss": 0.9301362037658691,
      "step": 56
    },
    {
      "epoch": 0.07397043615791715,
      "grad_norm": 0.7964755296707153,
      "learning_rate": 4.8275862068965525e-06,
      "loss": 0.8760148882865906,
      "step": 57
    },
    {
      "epoch": 0.07526816310805605,
      "grad_norm": 0.8177418112754822,
      "learning_rate": 4.9137931034482765e-06,
      "loss": 0.9806030988693237,
      "step": 58
    },
    {
      "epoch": 0.07656589005819495,
      "grad_norm": 0.8148566484451294,
      "learning_rate": 5e-06,
      "loss": 0.9486901760101318,
      "step": 59
    },
    {
      "epoch": 0.07786361700833384,
      "grad_norm": 0.7789106965065002,
      "learning_rate": 5.086206896551724e-06,
      "loss": 0.864636242389679,
      "step": 60
    },
    {
      "epoch": 0.07916134395847274,
      "grad_norm": 0.8153941035270691,
      "learning_rate": 5.172413793103449e-06,
      "loss": 0.8964623212814331,
      "step": 61
    },
    {
      "epoch": 0.08045907090861164,
      "grad_norm": 0.8125947713851929,
      "learning_rate": 5.258620689655173e-06,
      "loss": 0.951106607913971,
      "step": 62
    },
    {
      "epoch": 0.08175679785875054,
      "grad_norm": 0.8139016032218933,
      "learning_rate": 5.344827586206896e-06,
      "loss": 0.9181129932403564,
      "step": 63
    },
    {
      "epoch": 0.08305452480888943,
      "grad_norm": 0.7552460432052612,
      "learning_rate": 5.431034482758621e-06,
      "loss": 0.9456254839897156,
      "step": 64
    },
    {
      "epoch": 0.08435225175902833,
      "grad_norm": 0.7714299559593201,
      "learning_rate": 5.517241379310345e-06,
      "loss": 0.9329437613487244,
      "step": 65
    },
    {
      "epoch": 0.08564997870916723,
      "grad_norm": 0.7831487059593201,
      "learning_rate": 5.603448275862069e-06,
      "loss": 0.9335770010948181,
      "step": 66
    },
    {
      "epoch": 0.08694770565930612,
      "grad_norm": 0.8076956868171692,
      "learning_rate": 5.689655172413794e-06,
      "loss": 0.9263577461242676,
      "step": 67
    },
    {
      "epoch": 0.08824543260944502,
      "grad_norm": 0.7335968017578125,
      "learning_rate": 5.775862068965518e-06,
      "loss": 0.8535078167915344,
      "step": 68
    },
    {
      "epoch": 0.08954315955958392,
      "grad_norm": 0.7708919048309326,
      "learning_rate": 5.862068965517242e-06,
      "loss": 0.9572209119796753,
      "step": 69
    },
    {
      "epoch": 0.09084088650972282,
      "grad_norm": 0.7519110441207886,
      "learning_rate": 5.9482758620689665e-06,
      "loss": 0.9237591028213501,
      "step": 70
    },
    {
      "epoch": 0.09213861345986171,
      "grad_norm": 0.789342999458313,
      "learning_rate": 6.03448275862069e-06,
      "loss": 0.9293794631958008,
      "step": 71
    },
    {
      "epoch": 0.09343634041000061,
      "grad_norm": 0.8580161929130554,
      "learning_rate": 6.1206896551724135e-06,
      "loss": 1.0194649696350098,
      "step": 72
    },
    {
      "epoch": 0.0947340673601395,
      "grad_norm": 0.7663602232933044,
      "learning_rate": 6.206896551724138e-06,
      "loss": 0.8723046779632568,
      "step": 73
    },
    {
      "epoch": 0.0960317943102784,
      "grad_norm": 0.7687053680419922,
      "learning_rate": 6.293103448275862e-06,
      "loss": 0.8932123780250549,
      "step": 74
    },
    {
      "epoch": 0.0973295212604173,
      "grad_norm": 0.7912623882293701,
      "learning_rate": 6.379310344827587e-06,
      "loss": 0.8944480419158936,
      "step": 75
    },
    {
      "epoch": 0.0986272482105562,
      "grad_norm": 0.7491701245307922,
      "learning_rate": 6.465517241379311e-06,
      "loss": 0.9510178565979004,
      "step": 76
    },
    {
      "epoch": 0.0999249751606951,
      "grad_norm": 0.7807953953742981,
      "learning_rate": 6.551724137931035e-06,
      "loss": 0.9121882319450378,
      "step": 77
    },
    {
      "epoch": 0.10122270211083399,
      "grad_norm": 0.7805066108703613,
      "learning_rate": 6.63793103448276e-06,
      "loss": 0.8803927302360535,
      "step": 78
    },
    {
      "epoch": 0.10252042906097289,
      "grad_norm": 0.7737302780151367,
      "learning_rate": 6.724137931034484e-06,
      "loss": 0.959402859210968,
      "step": 79
    },
    {
      "epoch": 0.10381815601111179,
      "grad_norm": 0.7317953705787659,
      "learning_rate": 6.810344827586207e-06,
      "loss": 0.8773848414421082,
      "step": 80
    },
    {
      "epoch": 0.10511588296125068,
      "grad_norm": 0.7534870505332947,
      "learning_rate": 6.896551724137932e-06,
      "loss": 0.8883604407310486,
      "step": 81
    },
    {
      "epoch": 0.10641360991138958,
      "grad_norm": 0.7352930903434753,
      "learning_rate": 6.982758620689656e-06,
      "loss": 0.8512092232704163,
      "step": 82
    },
    {
      "epoch": 0.10771133686152848,
      "grad_norm": 0.7618373036384583,
      "learning_rate": 7.0689655172413796e-06,
      "loss": 0.867959201335907,
      "step": 83
    },
    {
      "epoch": 0.10900906381166738,
      "grad_norm": 0.7561816573143005,
      "learning_rate": 7.155172413793104e-06,
      "loss": 0.9354209899902344,
      "step": 84
    },
    {
      "epoch": 0.11030679076180627,
      "grad_norm": 0.7632707953453064,
      "learning_rate": 7.241379310344828e-06,
      "loss": 0.8821243047714233,
      "step": 85
    },
    {
      "epoch": 0.11160451771194517,
      "grad_norm": 0.7363986968994141,
      "learning_rate": 7.327586206896552e-06,
      "loss": 0.8989731073379517,
      "step": 86
    },
    {
      "epoch": 0.11290224466208407,
      "grad_norm": 0.7885032296180725,
      "learning_rate": 7.413793103448277e-06,
      "loss": 0.9085399508476257,
      "step": 87
    },
    {
      "epoch": 0.11419997161222296,
      "grad_norm": 0.7643342614173889,
      "learning_rate": 7.500000000000001e-06,
      "loss": 0.863010823726654,
      "step": 88
    },
    {
      "epoch": 0.11549769856236186,
      "grad_norm": 0.7284826040267944,
      "learning_rate": 7.586206896551724e-06,
      "loss": 0.9157944917678833,
      "step": 89
    },
    {
      "epoch": 0.11679542551250076,
      "grad_norm": 0.7478772401809692,
      "learning_rate": 7.672413793103449e-06,
      "loss": 0.9134526252746582,
      "step": 90
    },
    {
      "epoch": 0.11809315246263966,
      "grad_norm": 0.7836363315582275,
      "learning_rate": 7.758620689655173e-06,
      "loss": 0.908155620098114,
      "step": 91
    },
    {
      "epoch": 0.11939087941277855,
      "grad_norm": 0.7334380149841309,
      "learning_rate": 7.844827586206897e-06,
      "loss": 0.9568802118301392,
      "step": 92
    },
    {
      "epoch": 0.12068860636291745,
      "grad_norm": 0.7914179563522339,
      "learning_rate": 7.93103448275862e-06,
      "loss": 0.9015185832977295,
      "step": 93
    },
    {
      "epoch": 0.12198633331305635,
      "grad_norm": 0.6916558146476746,
      "learning_rate": 8.017241379310345e-06,
      "loss": 0.8272259831428528,
      "step": 94
    },
    {
      "epoch": 0.12328406026319524,
      "grad_norm": 0.7935004234313965,
      "learning_rate": 8.103448275862069e-06,
      "loss": 0.9952960014343262,
      "step": 95
    },
    {
      "epoch": 0.12458178721333414,
      "grad_norm": 0.7549696564674377,
      "learning_rate": 8.189655172413794e-06,
      "loss": 0.8569615483283997,
      "step": 96
    },
    {
      "epoch": 0.12587951416347304,
      "grad_norm": 0.8149930834770203,
      "learning_rate": 8.275862068965518e-06,
      "loss": 0.8843995928764343,
      "step": 97
    },
    {
      "epoch": 0.12717724111361195,
      "grad_norm": 0.7083914279937744,
      "learning_rate": 8.362068965517242e-06,
      "loss": 0.8912689089775085,
      "step": 98
    },
    {
      "epoch": 0.12847496806375083,
      "grad_norm": 0.7134708166122437,
      "learning_rate": 8.448275862068966e-06,
      "loss": 0.800317645072937,
      "step": 99
    },
    {
      "epoch": 0.12977269501388974,
      "grad_norm": 0.7577885985374451,
      "learning_rate": 8.53448275862069e-06,
      "loss": 0.9712153077125549,
      "step": 100
    },
    {
      "epoch": 0.13107042196402863,
      "grad_norm": 0.7569701671600342,
      "learning_rate": 8.620689655172414e-06,
      "loss": 0.9349263906478882,
      "step": 101
    },
    {
      "epoch": 0.13236814891416754,
      "grad_norm": 0.7419719099998474,
      "learning_rate": 8.706896551724138e-06,
      "loss": 0.8660463690757751,
      "step": 102
    },
    {
      "epoch": 0.13366587586430642,
      "grad_norm": 0.7070901989936829,
      "learning_rate": 8.793103448275862e-06,
      "loss": 0.8430792689323425,
      "step": 103
    },
    {
      "epoch": 0.13496360281444533,
      "grad_norm": 0.7656417489051819,
      "learning_rate": 8.879310344827588e-06,
      "loss": 0.9394369125366211,
      "step": 104
    },
    {
      "epoch": 0.13626132976458422,
      "grad_norm": 0.7935210466384888,
      "learning_rate": 8.965517241379312e-06,
      "loss": 0.9688374400138855,
      "step": 105
    },
    {
      "epoch": 0.13755905671472313,
      "grad_norm": 0.7562341690063477,
      "learning_rate": 9.051724137931036e-06,
      "loss": 0.9521839022636414,
      "step": 106
    },
    {
      "epoch": 0.138856783664862,
      "grad_norm": 0.7250262498855591,
      "learning_rate": 9.13793103448276e-06,
      "loss": 0.8643259406089783,
      "step": 107
    },
    {
      "epoch": 0.14015451061500092,
      "grad_norm": 0.6959007978439331,
      "learning_rate": 9.224137931034484e-06,
      "loss": 0.8793572187423706,
      "step": 108
    },
    {
      "epoch": 0.1414522375651398,
      "grad_norm": 0.7430084347724915,
      "learning_rate": 9.310344827586207e-06,
      "loss": 0.9341211318969727,
      "step": 109
    },
    {
      "epoch": 0.14274996451527872,
      "grad_norm": 0.7306565046310425,
      "learning_rate": 9.396551724137931e-06,
      "loss": 0.9853330850601196,
      "step": 110
    },
    {
      "epoch": 0.1440476914654176,
      "grad_norm": 0.7661026120185852,
      "learning_rate": 9.482758620689655e-06,
      "loss": 0.9588793516159058,
      "step": 111
    },
    {
      "epoch": 0.1453454184155565,
      "grad_norm": 0.7300426363945007,
      "learning_rate": 9.56896551724138e-06,
      "loss": 0.841079831123352,
      "step": 112
    },
    {
      "epoch": 0.1466431453656954,
      "grad_norm": 0.75968337059021,
      "learning_rate": 9.655172413793105e-06,
      "loss": 0.9170314073562622,
      "step": 113
    },
    {
      "epoch": 0.1479408723158343,
      "grad_norm": 0.7715488076210022,
      "learning_rate": 9.741379310344829e-06,
      "loss": 0.8987317681312561,
      "step": 114
    },
    {
      "epoch": 0.1492385992659732,
      "grad_norm": 0.7421614527702332,
      "learning_rate": 9.827586206896553e-06,
      "loss": 0.8786249160766602,
      "step": 115
    },
    {
      "epoch": 0.1505363262161121,
      "grad_norm": 0.7512428164482117,
      "learning_rate": 9.913793103448277e-06,
      "loss": 0.9067139625549316,
      "step": 116
    },
    {
      "epoch": 0.15183405316625098,
      "grad_norm": 0.7859955430030823,
      "learning_rate": 1e-05,
      "loss": 0.9494441747665405,
      "step": 117
    },
    {
      "epoch": 0.1531317801163899,
      "grad_norm": 0.7653732895851135,
      "learning_rate": 9.99999488813276e-06,
      "loss": 0.9384220242500305,
      "step": 118
    },
    {
      "epoch": 0.15442950706652878,
      "grad_norm": 0.7733427286148071,
      "learning_rate": 9.999979552541496e-06,
      "loss": 0.8465048670768738,
      "step": 119
    },
    {
      "epoch": 0.1557272340166677,
      "grad_norm": 0.7779337167739868,
      "learning_rate": 9.99995399325756e-06,
      "loss": 0.8923530578613281,
      "step": 120
    },
    {
      "epoch": 0.15702496096680657,
      "grad_norm": 0.7559251189231873,
      "learning_rate": 9.999918210333219e-06,
      "loss": 0.9556446075439453,
      "step": 121
    },
    {
      "epoch": 0.15832268791694548,
      "grad_norm": 0.7645485997200012,
      "learning_rate": 9.999872203841635e-06,
      "loss": 0.86847323179245,
      "step": 122
    },
    {
      "epoch": 0.15962041486708436,
      "grad_norm": 0.8056962490081787,
      "learning_rate": 9.999815973876888e-06,
      "loss": 0.8974254131317139,
      "step": 123
    },
    {
      "epoch": 0.16091814181722328,
      "grad_norm": 0.8095734119415283,
      "learning_rate": 9.999749520553945e-06,
      "loss": 0.9332894086837769,
      "step": 124
    },
    {
      "epoch": 0.16221586876736216,
      "grad_norm": 0.7671176791191101,
      "learning_rate": 9.99967284400869e-06,
      "loss": 0.8843300342559814,
      "step": 125
    },
    {
      "epoch": 0.16351359571750107,
      "grad_norm": 0.7800223231315613,
      "learning_rate": 9.99958594439791e-06,
      "loss": 0.857262134552002,
      "step": 126
    },
    {
      "epoch": 0.16481132266763995,
      "grad_norm": 0.7728240489959717,
      "learning_rate": 9.999488821899286e-06,
      "loss": 0.8976252675056458,
      "step": 127
    },
    {
      "epoch": 0.16610904961777886,
      "grad_norm": 0.7575803399085999,
      "learning_rate": 9.999381476711416e-06,
      "loss": 0.835990309715271,
      "step": 128
    },
    {
      "epoch": 0.16740677656791775,
      "grad_norm": 0.8353949785232544,
      "learning_rate": 9.999263909053789e-06,
      "loss": 0.8549575805664062,
      "step": 129
    },
    {
      "epoch": 0.16870450351805666,
      "grad_norm": 0.7656837105751038,
      "learning_rate": 9.999136119166803e-06,
      "loss": 0.822762668132782,
      "step": 130
    },
    {
      "epoch": 0.17000223046819554,
      "grad_norm": 0.7765395045280457,
      "learning_rate": 9.998998107311758e-06,
      "loss": 0.9564024209976196,
      "step": 131
    },
    {
      "epoch": 0.17129995741833445,
      "grad_norm": 0.7651109099388123,
      "learning_rate": 9.998849873770849e-06,
      "loss": 0.8914260268211365,
      "step": 132
    },
    {
      "epoch": 0.17259768436847334,
      "grad_norm": 0.7631571888923645,
      "learning_rate": 9.998691418847177e-06,
      "loss": 0.8151357173919678,
      "step": 133
    },
    {
      "epoch": 0.17389541131861225,
      "grad_norm": 0.8555284142494202,
      "learning_rate": 9.998522742864745e-06,
      "loss": 0.9253755211830139,
      "step": 134
    },
    {
      "epoch": 0.17519313826875113,
      "grad_norm": 0.7372084259986877,
      "learning_rate": 9.998343846168448e-06,
      "loss": 0.8820257186889648,
      "step": 135
    },
    {
      "epoch": 0.17649086521889004,
      "grad_norm": 0.7227676510810852,
      "learning_rate": 9.998154729124092e-06,
      "loss": 0.8299266695976257,
      "step": 136
    },
    {
      "epoch": 0.17778859216902893,
      "grad_norm": 0.7503435015678406,
      "learning_rate": 9.997955392118365e-06,
      "loss": 0.8857607245445251,
      "step": 137
    },
    {
      "epoch": 0.17908631911916784,
      "grad_norm": 0.7888970375061035,
      "learning_rate": 9.997745835558867e-06,
      "loss": 0.8129744529724121,
      "step": 138
    },
    {
      "epoch": 0.18038404606930672,
      "grad_norm": 0.7952077388763428,
      "learning_rate": 9.997526059874086e-06,
      "loss": 0.9076676368713379,
      "step": 139
    },
    {
      "epoch": 0.18168177301944563,
      "grad_norm": 0.7603996396064758,
      "learning_rate": 9.997296065513405e-06,
      "loss": 0.90814208984375,
      "step": 140
    },
    {
      "epoch": 0.1829794999695845,
      "grad_norm": 0.7346363663673401,
      "learning_rate": 9.997055852947109e-06,
      "loss": 0.9773215055465698,
      "step": 141
    },
    {
      "epoch": 0.18427722691972342,
      "grad_norm": 0.780322790145874,
      "learning_rate": 9.996805422666367e-06,
      "loss": 0.9135559797286987,
      "step": 142
    },
    {
      "epoch": 0.1855749538698623,
      "grad_norm": 0.7772685289382935,
      "learning_rate": 9.99654477518325e-06,
      "loss": 0.9268802404403687,
      "step": 143
    },
    {
      "epoch": 0.18687268082000122,
      "grad_norm": 0.725249707698822,
      "learning_rate": 9.996273911030714e-06,
      "loss": 0.8543541431427002,
      "step": 144
    },
    {
      "epoch": 0.1881704077701401,
      "grad_norm": 0.7833412289619446,
      "learning_rate": 9.995992830762608e-06,
      "loss": 1.0088237524032593,
      "step": 145
    },
    {
      "epoch": 0.189468134720279,
      "grad_norm": 0.7379457354545593,
      "learning_rate": 9.99570153495367e-06,
      "loss": 0.8407413363456726,
      "step": 146
    },
    {
      "epoch": 0.1907658616704179,
      "grad_norm": 0.7491745352745056,
      "learning_rate": 9.995400024199526e-06,
      "loss": 0.8960580825805664,
      "step": 147
    },
    {
      "epoch": 0.1920635886205568,
      "grad_norm": 0.7832504510879517,
      "learning_rate": 9.99508829911669e-06,
      "loss": 0.9582692980766296,
      "step": 148
    },
    {
      "epoch": 0.1933613155706957,
      "grad_norm": 0.8017767071723938,
      "learning_rate": 9.994766360342557e-06,
      "loss": 0.95718914270401,
      "step": 149
    },
    {
      "epoch": 0.1946590425208346,
      "grad_norm": 0.6941841840744019,
      "learning_rate": 9.994434208535415e-06,
      "loss": 0.8945623636245728,
      "step": 150
    },
    {
      "epoch": 0.19595676947097349,
      "grad_norm": 0.7576905488967896,
      "learning_rate": 9.994091844374431e-06,
      "loss": 0.9543700814247131,
      "step": 151
    },
    {
      "epoch": 0.1972544964211124,
      "grad_norm": 0.7699536681175232,
      "learning_rate": 9.993739268559648e-06,
      "loss": 0.9578084945678711,
      "step": 152
    },
    {
      "epoch": 0.19855222337125128,
      "grad_norm": 0.7555127143859863,
      "learning_rate": 9.993376481812001e-06,
      "loss": 0.9324721693992615,
      "step": 153
    },
    {
      "epoch": 0.1998499503213902,
      "grad_norm": 0.8148646354675293,
      "learning_rate": 9.99300348487329e-06,
      "loss": 0.9196079969406128,
      "step": 154
    },
    {
      "epoch": 0.20114767727152907,
      "grad_norm": 0.7479966878890991,
      "learning_rate": 9.992620278506203e-06,
      "loss": 0.915722131729126,
      "step": 155
    },
    {
      "epoch": 0.20244540422166798,
      "grad_norm": 0.7261396646499634,
      "learning_rate": 9.9922268634943e-06,
      "loss": 0.7793674468994141,
      "step": 156
    },
    {
      "epoch": 0.20374313117180687,
      "grad_norm": 0.8150226473808289,
      "learning_rate": 9.991823240642014e-06,
      "loss": 1.0025370121002197,
      "step": 157
    },
    {
      "epoch": 0.20504085812194578,
      "grad_norm": 0.7759719491004944,
      "learning_rate": 9.991409410774654e-06,
      "loss": 0.9169880747795105,
      "step": 158
    },
    {
      "epoch": 0.2063385850720847,
      "grad_norm": 0.7141395211219788,
      "learning_rate": 9.990985374738396e-06,
      "loss": 0.8525487184524536,
      "step": 159
    },
    {
      "epoch": 0.20763631202222357,
      "grad_norm": 0.7439554929733276,
      "learning_rate": 9.990551133400284e-06,
      "loss": 0.8688797950744629,
      "step": 160
    },
    {
      "epoch": 0.20893403897236248,
      "grad_norm": 0.7583606839179993,
      "learning_rate": 9.990106687648234e-06,
      "loss": 0.8457356095314026,
      "step": 161
    },
    {
      "epoch": 0.21023176592250137,
      "grad_norm": 0.7045748829841614,
      "learning_rate": 9.989652038391025e-06,
      "loss": 0.8181869983673096,
      "step": 162
    },
    {
      "epoch": 0.21152949287264028,
      "grad_norm": 0.7270849943161011,
      "learning_rate": 9.9891871865583e-06,
      "loss": 0.8501678705215454,
      "step": 163
    },
    {
      "epoch": 0.21282721982277916,
      "grad_norm": 0.7277626991271973,
      "learning_rate": 9.988712133100563e-06,
      "loss": 0.9715257287025452,
      "step": 164
    },
    {
      "epoch": 0.21412494677291807,
      "grad_norm": 0.7333799600601196,
      "learning_rate": 9.988226878989178e-06,
      "loss": 0.8683098554611206,
      "step": 165
    },
    {
      "epoch": 0.21542267372305696,
      "grad_norm": 0.7085330486297607,
      "learning_rate": 9.987731425216364e-06,
      "loss": 0.7768368721008301,
      "step": 166
    },
    {
      "epoch": 0.21672040067319587,
      "grad_norm": 0.751772403717041,
      "learning_rate": 9.987225772795204e-06,
      "loss": 0.9315556287765503,
      "step": 167
    },
    {
      "epoch": 0.21801812762333475,
      "grad_norm": 0.7408215999603271,
      "learning_rate": 9.986709922759626e-06,
      "loss": 0.8905249238014221,
      "step": 168
    },
    {
      "epoch": 0.21931585457347366,
      "grad_norm": 0.6908205151557922,
      "learning_rate": 9.986183876164412e-06,
      "loss": 0.7912262678146362,
      "step": 169
    },
    {
      "epoch": 0.22061358152361255,
      "grad_norm": 0.7142277956008911,
      "learning_rate": 9.985647634085197e-06,
      "loss": 0.8391699194908142,
      "step": 170
    },
    {
      "epoch": 0.22191130847375146,
      "grad_norm": 0.7526142597198486,
      "learning_rate": 9.985101197618456e-06,
      "loss": 0.8537833094596863,
      "step": 171
    },
    {
      "epoch": 0.22320903542389034,
      "grad_norm": 0.7814799547195435,
      "learning_rate": 9.98454456788152e-06,
      "loss": 0.8500316143035889,
      "step": 172
    },
    {
      "epoch": 0.22450676237402925,
      "grad_norm": 0.7822877168655396,
      "learning_rate": 9.983977746012547e-06,
      "loss": 0.9026069641113281,
      "step": 173
    },
    {
      "epoch": 0.22580448932416813,
      "grad_norm": 0.728121817111969,
      "learning_rate": 9.983400733170553e-06,
      "loss": 0.8231689929962158,
      "step": 174
    },
    {
      "epoch": 0.22710221627430704,
      "grad_norm": 0.7749063372612,
      "learning_rate": 9.982813530535377e-06,
      "loss": 0.8157176375389099,
      "step": 175
    },
    {
      "epoch": 0.22839994322444593,
      "grad_norm": 0.7811345458030701,
      "learning_rate": 9.982216139307705e-06,
      "loss": 0.9290957450866699,
      "step": 176
    },
    {
      "epoch": 0.22969767017458484,
      "grad_norm": 0.7119234204292297,
      "learning_rate": 9.981608560709044e-06,
      "loss": 0.7803032994270325,
      "step": 177
    },
    {
      "epoch": 0.23099539712472372,
      "grad_norm": 0.7667236328125,
      "learning_rate": 9.980990795981747e-06,
      "loss": 0.9145867824554443,
      "step": 178
    },
    {
      "epoch": 0.23229312407486263,
      "grad_norm": 0.754831075668335,
      "learning_rate": 9.980362846388978e-06,
      "loss": 0.9066612124443054,
      "step": 179
    },
    {
      "epoch": 0.23359085102500152,
      "grad_norm": 0.7749460935592651,
      "learning_rate": 9.97972471321474e-06,
      "loss": 0.889066755771637,
      "step": 180
    },
    {
      "epoch": 0.23488857797514043,
      "grad_norm": 0.7541179656982422,
      "learning_rate": 9.979076397763853e-06,
      "loss": 0.8429720997810364,
      "step": 181
    },
    {
      "epoch": 0.2361863049252793,
      "grad_norm": 0.7422946095466614,
      "learning_rate": 9.978417901361958e-06,
      "loss": 0.9534435272216797,
      "step": 182
    },
    {
      "epoch": 0.23748403187541822,
      "grad_norm": 0.7609230875968933,
      "learning_rate": 9.977749225355513e-06,
      "loss": 0.8388585448265076,
      "step": 183
    },
    {
      "epoch": 0.2387817588255571,
      "grad_norm": 0.7104296684265137,
      "learning_rate": 9.977070371111793e-06,
      "loss": 0.852005660533905,
      "step": 184
    },
    {
      "epoch": 0.24007948577569602,
      "grad_norm": 0.7545174360275269,
      "learning_rate": 9.976381340018879e-06,
      "loss": 0.8911629319190979,
      "step": 185
    },
    {
      "epoch": 0.2413772127258349,
      "grad_norm": 0.7062454223632812,
      "learning_rate": 9.97568213348567e-06,
      "loss": 0.8741620182991028,
      "step": 186
    },
    {
      "epoch": 0.2426749396759738,
      "grad_norm": 0.7272056937217712,
      "learning_rate": 9.974972752941861e-06,
      "loss": 0.8200283050537109,
      "step": 187
    },
    {
      "epoch": 0.2439726666261127,
      "grad_norm": 0.7598206996917725,
      "learning_rate": 9.97425319983796e-06,
      "loss": 0.9287367463111877,
      "step": 188
    },
    {
      "epoch": 0.2452703935762516,
      "grad_norm": 0.773642897605896,
      "learning_rate": 9.97352347564527e-06,
      "loss": 0.8710715174674988,
      "step": 189
    },
    {
      "epoch": 0.2465681205263905,
      "grad_norm": 0.7518259882926941,
      "learning_rate": 9.972783581855894e-06,
      "loss": 0.8873825669288635,
      "step": 190
    },
    {
      "epoch": 0.2478658474765294,
      "grad_norm": 0.7199155688285828,
      "learning_rate": 9.972033519982722e-06,
      "loss": 0.8325483798980713,
      "step": 191
    },
    {
      "epoch": 0.24916357442666828,
      "grad_norm": 0.7098877429962158,
      "learning_rate": 9.971273291559447e-06,
      "loss": 0.9047567248344421,
      "step": 192
    },
    {
      "epoch": 0.25046130137680717,
      "grad_norm": 0.7787808775901794,
      "learning_rate": 9.97050289814054e-06,
      "loss": 0.8593154549598694,
      "step": 193
    },
    {
      "epoch": 0.2517590283269461,
      "grad_norm": 0.6697104573249817,
      "learning_rate": 9.969722341301261e-06,
      "loss": 0.7885489463806152,
      "step": 194
    },
    {
      "epoch": 0.253056755277085,
      "grad_norm": 0.735886812210083,
      "learning_rate": 9.968931622637652e-06,
      "loss": 0.9099982976913452,
      "step": 195
    },
    {
      "epoch": 0.2543544822272239,
      "grad_norm": 0.7376393675804138,
      "learning_rate": 9.968130743766533e-06,
      "loss": 0.8479236364364624,
      "step": 196
    },
    {
      "epoch": 0.25565220917736275,
      "grad_norm": 0.7209894061088562,
      "learning_rate": 9.967319706325495e-06,
      "loss": 0.822541356086731,
      "step": 197
    },
    {
      "epoch": 0.25694993612750167,
      "grad_norm": 0.7586290240287781,
      "learning_rate": 9.96649851197291e-06,
      "loss": 0.8346601128578186,
      "step": 198
    },
    {
      "epoch": 0.2582476630776406,
      "grad_norm": 0.7636249661445618,
      "learning_rate": 9.965667162387908e-06,
      "loss": 0.9417949914932251,
      "step": 199
    },
    {
      "epoch": 0.2595453900277795,
      "grad_norm": 0.7661371827125549,
      "learning_rate": 9.964825659270391e-06,
      "loss": 0.8580296635627747,
      "step": 200
    },
    {
      "epoch": 0.26084311697791834,
      "grad_norm": 0.7609704732894897,
      "learning_rate": 9.963974004341019e-06,
      "loss": 0.8554142117500305,
      "step": 201
    },
    {
      "epoch": 0.26214084392805725,
      "grad_norm": 0.7663947343826294,
      "learning_rate": 9.963112199341212e-06,
      "loss": 0.8991423845291138,
      "step": 202
    },
    {
      "epoch": 0.26343857087819617,
      "grad_norm": 0.7254887223243713,
      "learning_rate": 9.96224024603314e-06,
      "loss": 0.8028452396392822,
      "step": 203
    },
    {
      "epoch": 0.2647362978283351,
      "grad_norm": 0.7555724382400513,
      "learning_rate": 9.961358146199729e-06,
      "loss": 0.8306048512458801,
      "step": 204
    },
    {
      "epoch": 0.26603402477847393,
      "grad_norm": 0.7537539601325989,
      "learning_rate": 9.960465901644651e-06,
      "loss": 0.8942010998725891,
      "step": 205
    },
    {
      "epoch": 0.26733175172861284,
      "grad_norm": 0.6972079873085022,
      "learning_rate": 9.959563514192317e-06,
      "loss": 0.8508002758026123,
      "step": 206
    },
    {
      "epoch": 0.26862947867875175,
      "grad_norm": 0.7849931716918945,
      "learning_rate": 9.958650985687884e-06,
      "loss": 0.9203708171844482,
      "step": 207
    },
    {
      "epoch": 0.26992720562889067,
      "grad_norm": 0.7529065012931824,
      "learning_rate": 9.95772831799724e-06,
      "loss": 0.9082123637199402,
      "step": 208
    },
    {
      "epoch": 0.2712249325790295,
      "grad_norm": 0.7322463393211365,
      "learning_rate": 9.956795513007008e-06,
      "loss": 0.8661291599273682,
      "step": 209
    },
    {
      "epoch": 0.27252265952916843,
      "grad_norm": 0.8052998185157776,
      "learning_rate": 9.955852572624538e-06,
      "loss": 0.9168900847434998,
      "step": 210
    },
    {
      "epoch": 0.27382038647930734,
      "grad_norm": 0.7243384718894958,
      "learning_rate": 9.954899498777903e-06,
      "loss": 0.9119362235069275,
      "step": 211
    },
    {
      "epoch": 0.27511811342944625,
      "grad_norm": 0.741720974445343,
      "learning_rate": 9.9539362934159e-06,
      "loss": 0.8204353451728821,
      "step": 212
    },
    {
      "epoch": 0.2764158403795851,
      "grad_norm": 0.7546416521072388,
      "learning_rate": 9.952962958508038e-06,
      "loss": 0.8421768546104431,
      "step": 213
    },
    {
      "epoch": 0.277713567329724,
      "grad_norm": 0.7767171859741211,
      "learning_rate": 9.951979496044544e-06,
      "loss": 0.8838454484939575,
      "step": 214
    },
    {
      "epoch": 0.27901129427986293,
      "grad_norm": 0.7499642968177795,
      "learning_rate": 9.950985908036346e-06,
      "loss": 0.8892335295677185,
      "step": 215
    },
    {
      "epoch": 0.28030902123000184,
      "grad_norm": 0.743560254573822,
      "learning_rate": 9.94998219651508e-06,
      "loss": 0.8514686226844788,
      "step": 216
    },
    {
      "epoch": 0.2816067481801407,
      "grad_norm": 0.7842270135879517,
      "learning_rate": 9.948968363533085e-06,
      "loss": 0.8559333086013794,
      "step": 217
    },
    {
      "epoch": 0.2829044751302796,
      "grad_norm": 0.8065765500068665,
      "learning_rate": 9.947944411163391e-06,
      "loss": 0.8402450084686279,
      "step": 218
    },
    {
      "epoch": 0.2842022020804185,
      "grad_norm": 0.7745425701141357,
      "learning_rate": 9.946910341499722e-06,
      "loss": 0.8811400532722473,
      "step": 219
    },
    {
      "epoch": 0.28549992903055743,
      "grad_norm": 0.7459537982940674,
      "learning_rate": 9.945866156656487e-06,
      "loss": 0.8989254236221313,
      "step": 220
    },
    {
      "epoch": 0.2867976559806963,
      "grad_norm": 0.7783303260803223,
      "learning_rate": 9.944811858768782e-06,
      "loss": 0.8823436498641968,
      "step": 221
    },
    {
      "epoch": 0.2880953829308352,
      "grad_norm": 0.8013091087341309,
      "learning_rate": 9.943747449992379e-06,
      "loss": 0.9165104627609253,
      "step": 222
    },
    {
      "epoch": 0.2893931098809741,
      "grad_norm": 0.7300430536270142,
      "learning_rate": 9.942672932503722e-06,
      "loss": 0.8838682174682617,
      "step": 223
    },
    {
      "epoch": 0.290690836831113,
      "grad_norm": 0.7317079305648804,
      "learning_rate": 9.941588308499932e-06,
      "loss": 0.9152242541313171,
      "step": 224
    },
    {
      "epoch": 0.2919885637812519,
      "grad_norm": 0.7530835866928101,
      "learning_rate": 9.940493580198787e-06,
      "loss": 0.8324201703071594,
      "step": 225
    },
    {
      "epoch": 0.2932862907313908,
      "grad_norm": 0.7751577496528625,
      "learning_rate": 9.93938874983873e-06,
      "loss": 0.8113334774971008,
      "step": 226
    },
    {
      "epoch": 0.2945840176815297,
      "grad_norm": 0.7621341347694397,
      "learning_rate": 9.93827381967886e-06,
      "loss": 0.8741900324821472,
      "step": 227
    },
    {
      "epoch": 0.2958817446316686,
      "grad_norm": 0.7917994260787964,
      "learning_rate": 9.937148791998926e-06,
      "loss": 0.8467355966567993,
      "step": 228
    },
    {
      "epoch": 0.29717947158180746,
      "grad_norm": 0.7442567348480225,
      "learning_rate": 9.936013669099326e-06,
      "loss": 0.8748873472213745,
      "step": 229
    },
    {
      "epoch": 0.2984771985319464,
      "grad_norm": 0.7540955543518066,
      "learning_rate": 9.9348684533011e-06,
      "loss": 0.9296368360519409,
      "step": 230
    },
    {
      "epoch": 0.2984771985319464,
      "eval_loss": 0.8298861980438232,
      "eval_runtime": 147.7609,
      "eval_samples_per_second": 35.138,
      "eval_steps_per_second": 8.784,
      "step": 230
    },
    {
      "epoch": 0.2997749254820853,
      "grad_norm": 0.7888861894607544,
      "learning_rate": 9.93371314694592e-06,
      "loss": 0.8800599575042725,
      "step": 231
    },
    {
      "epoch": 0.3010726524322242,
      "grad_norm": 0.7253603935241699,
      "learning_rate": 9.9325477523961e-06,
      "loss": 0.8651221990585327,
      "step": 232
    },
    {
      "epoch": 0.30237037938236305,
      "grad_norm": 0.722152054309845,
      "learning_rate": 9.931372272034573e-06,
      "loss": 0.8784995079040527,
      "step": 233
    },
    {
      "epoch": 0.30366810633250196,
      "grad_norm": 0.762865424156189,
      "learning_rate": 9.930186708264902e-06,
      "loss": 0.8278964161872864,
      "step": 234
    },
    {
      "epoch": 0.3049658332826409,
      "grad_norm": 0.7679449319839478,
      "learning_rate": 9.928991063511264e-06,
      "loss": 0.8578440546989441,
      "step": 235
    },
    {
      "epoch": 0.3062635602327798,
      "grad_norm": 0.747564971446991,
      "learning_rate": 9.927785340218448e-06,
      "loss": 0.8361713290214539,
      "step": 236
    },
    {
      "epoch": 0.3075612871829187,
      "grad_norm": 0.7898075580596924,
      "learning_rate": 9.926569540851856e-06,
      "loss": 0.9254142045974731,
      "step": 237
    },
    {
      "epoch": 0.30885901413305755,
      "grad_norm": 0.7446305751800537,
      "learning_rate": 9.925343667897487e-06,
      "loss": 0.8479958772659302,
      "step": 238
    },
    {
      "epoch": 0.31015674108319646,
      "grad_norm": 0.7183104157447815,
      "learning_rate": 9.924107723861944e-06,
      "loss": 0.8849471807479858,
      "step": 239
    },
    {
      "epoch": 0.3114544680333354,
      "grad_norm": 0.7643358707427979,
      "learning_rate": 9.922861711272417e-06,
      "loss": 0.9337932467460632,
      "step": 240
    },
    {
      "epoch": 0.3127521949834743,
      "grad_norm": 0.7303009629249573,
      "learning_rate": 9.921605632676688e-06,
      "loss": 0.8316428661346436,
      "step": 241
    },
    {
      "epoch": 0.31404992193361314,
      "grad_norm": 0.7191450595855713,
      "learning_rate": 9.920339490643119e-06,
      "loss": 0.7859864830970764,
      "step": 242
    },
    {
      "epoch": 0.31534764888375205,
      "grad_norm": 0.6739291548728943,
      "learning_rate": 9.91906328776065e-06,
      "loss": 0.7918820977210999,
      "step": 243
    },
    {
      "epoch": 0.31664537583389096,
      "grad_norm": 0.7382656335830688,
      "learning_rate": 9.917777026638794e-06,
      "loss": 0.8608056306838989,
      "step": 244
    },
    {
      "epoch": 0.3179431027840299,
      "grad_norm": 0.7110844254493713,
      "learning_rate": 9.916480709907626e-06,
      "loss": 0.7915103435516357,
      "step": 245
    },
    {
      "epoch": 0.31924082973416873,
      "grad_norm": 0.7153676748275757,
      "learning_rate": 9.91517434021779e-06,
      "loss": 0.8111603856086731,
      "step": 246
    },
    {
      "epoch": 0.32053855668430764,
      "grad_norm": 0.7526156902313232,
      "learning_rate": 9.913857920240481e-06,
      "loss": 0.9535894989967346,
      "step": 247
    },
    {
      "epoch": 0.32183628363444655,
      "grad_norm": 0.7110179662704468,
      "learning_rate": 9.912531452667441e-06,
      "loss": 0.8099456429481506,
      "step": 248
    },
    {
      "epoch": 0.32313401058458546,
      "grad_norm": 0.7142106890678406,
      "learning_rate": 9.911194940210964e-06,
      "loss": 0.8379517793655396,
      "step": 249
    },
    {
      "epoch": 0.3244317375347243,
      "grad_norm": 0.7095869183540344,
      "learning_rate": 9.909848385603878e-06,
      "loss": 0.9001181125640869,
      "step": 250
    },
    {
      "epoch": 0.32572946448486323,
      "grad_norm": 0.7466822266578674,
      "learning_rate": 9.908491791599546e-06,
      "loss": 0.8450853824615479,
      "step": 251
    },
    {
      "epoch": 0.32702719143500214,
      "grad_norm": 0.75234055519104,
      "learning_rate": 9.90712516097186e-06,
      "loss": 0.9213320016860962,
      "step": 252
    },
    {
      "epoch": 0.32832491838514105,
      "grad_norm": 0.7801260948181152,
      "learning_rate": 9.905748496515235e-06,
      "loss": 0.8934061527252197,
      "step": 253
    },
    {
      "epoch": 0.3296226453352799,
      "grad_norm": 0.7624495029449463,
      "learning_rate": 9.904361801044599e-06,
      "loss": 0.9207302927970886,
      "step": 254
    },
    {
      "epoch": 0.3309203722854188,
      "grad_norm": 0.7547022700309753,
      "learning_rate": 9.902965077395395e-06,
      "loss": 0.8502051830291748,
      "step": 255
    },
    {
      "epoch": 0.33221809923555773,
      "grad_norm": 0.7386295795440674,
      "learning_rate": 9.901558328423568e-06,
      "loss": 0.9277836680412292,
      "step": 256
    },
    {
      "epoch": 0.33351582618569664,
      "grad_norm": 0.7245653867721558,
      "learning_rate": 9.900141557005567e-06,
      "loss": 0.841809868812561,
      "step": 257
    },
    {
      "epoch": 0.3348135531358355,
      "grad_norm": 0.7325631380081177,
      "learning_rate": 9.898714766038326e-06,
      "loss": 0.869012713432312,
      "step": 258
    },
    {
      "epoch": 0.3361112800859744,
      "grad_norm": 0.759433925151825,
      "learning_rate": 9.897277958439274e-06,
      "loss": 0.9291174411773682,
      "step": 259
    },
    {
      "epoch": 0.3374090070361133,
      "grad_norm": 0.757870614528656,
      "learning_rate": 9.895831137146319e-06,
      "loss": 0.8791542053222656,
      "step": 260
    },
    {
      "epoch": 0.33870673398625223,
      "grad_norm": 0.7061610817909241,
      "learning_rate": 9.894374305117844e-06,
      "loss": 0.8448185324668884,
      "step": 261
    },
    {
      "epoch": 0.3400044609363911,
      "grad_norm": 0.7048200368881226,
      "learning_rate": 9.892907465332702e-06,
      "loss": 0.8903120756149292,
      "step": 262
    },
    {
      "epoch": 0.34130218788653,
      "grad_norm": 0.7492517232894897,
      "learning_rate": 9.891430620790208e-06,
      "loss": 0.7949280142784119,
      "step": 263
    },
    {
      "epoch": 0.3425999148366689,
      "grad_norm": 0.7763856053352356,
      "learning_rate": 9.889943774510136e-06,
      "loss": 0.88750821352005,
      "step": 264
    },
    {
      "epoch": 0.3438976417868078,
      "grad_norm": 0.7783463597297668,
      "learning_rate": 9.888446929532712e-06,
      "loss": 0.8366783261299133,
      "step": 265
    },
    {
      "epoch": 0.3451953687369467,
      "grad_norm": 0.7750303745269775,
      "learning_rate": 9.886940088918601e-06,
      "loss": 0.9069842100143433,
      "step": 266
    },
    {
      "epoch": 0.3464930956870856,
      "grad_norm": 0.7684900760650635,
      "learning_rate": 9.885423255748916e-06,
      "loss": 0.8415606021881104,
      "step": 267
    },
    {
      "epoch": 0.3477908226372245,
      "grad_norm": 0.764328122138977,
      "learning_rate": 9.883896433125193e-06,
      "loss": 0.8657584190368652,
      "step": 268
    },
    {
      "epoch": 0.3490885495873634,
      "grad_norm": 0.7822588682174683,
      "learning_rate": 9.8823596241694e-06,
      "loss": 0.8536877632141113,
      "step": 269
    },
    {
      "epoch": 0.35038627653750226,
      "grad_norm": 0.7310823202133179,
      "learning_rate": 9.88081283202392e-06,
      "loss": 0.8020272254943848,
      "step": 270
    },
    {
      "epoch": 0.3516840034876412,
      "grad_norm": 0.717753529548645,
      "learning_rate": 9.879256059851553e-06,
      "loss": 0.8834531307220459,
      "step": 271
    },
    {
      "epoch": 0.3529817304377801,
      "grad_norm": 0.7358869314193726,
      "learning_rate": 9.877689310835503e-06,
      "loss": 0.9175652861595154,
      "step": 272
    },
    {
      "epoch": 0.354279457387919,
      "grad_norm": 0.7447776198387146,
      "learning_rate": 9.876112588179378e-06,
      "loss": 0.8402864933013916,
      "step": 273
    },
    {
      "epoch": 0.35557718433805785,
      "grad_norm": 0.7582140564918518,
      "learning_rate": 9.874525895107175e-06,
      "loss": 0.8564626574516296,
      "step": 274
    },
    {
      "epoch": 0.35687491128819676,
      "grad_norm": 0.7297621965408325,
      "learning_rate": 9.872929234863277e-06,
      "loss": 0.8620203137397766,
      "step": 275
    },
    {
      "epoch": 0.35817263823833567,
      "grad_norm": 0.6891196966171265,
      "learning_rate": 9.871322610712452e-06,
      "loss": 0.795127809047699,
      "step": 276
    },
    {
      "epoch": 0.3594703651884746,
      "grad_norm": 0.7053247690200806,
      "learning_rate": 9.869706025939843e-06,
      "loss": 0.7824152112007141,
      "step": 277
    },
    {
      "epoch": 0.36076809213861344,
      "grad_norm": 0.7359561324119568,
      "learning_rate": 9.868079483850955e-06,
      "loss": 0.8618420958518982,
      "step": 278
    },
    {
      "epoch": 0.36206581908875235,
      "grad_norm": 0.7968899011611938,
      "learning_rate": 9.86644298777165e-06,
      "loss": 0.8965823650360107,
      "step": 279
    },
    {
      "epoch": 0.36336354603889126,
      "grad_norm": 0.762060284614563,
      "learning_rate": 9.864796541048155e-06,
      "loss": 0.9179255366325378,
      "step": 280
    },
    {
      "epoch": 0.36466127298903017,
      "grad_norm": 0.7475153207778931,
      "learning_rate": 9.863140147047034e-06,
      "loss": 0.8677244782447815,
      "step": 281
    },
    {
      "epoch": 0.365958999939169,
      "grad_norm": 0.7659401297569275,
      "learning_rate": 9.861473809155192e-06,
      "loss": 0.8148896098136902,
      "step": 282
    },
    {
      "epoch": 0.36725672688930794,
      "grad_norm": 0.7381513118743896,
      "learning_rate": 9.859797530779871e-06,
      "loss": 0.7781128883361816,
      "step": 283
    },
    {
      "epoch": 0.36855445383944685,
      "grad_norm": 0.7533836364746094,
      "learning_rate": 9.858111315348633e-06,
      "loss": 0.9026414752006531,
      "step": 284
    },
    {
      "epoch": 0.36985218078958576,
      "grad_norm": 0.7744868993759155,
      "learning_rate": 9.856415166309365e-06,
      "loss": 0.8830351233482361,
      "step": 285
    },
    {
      "epoch": 0.3711499077397246,
      "grad_norm": 0.7385957837104797,
      "learning_rate": 9.854709087130261e-06,
      "loss": 0.8755077123641968,
      "step": 286
    },
    {
      "epoch": 0.3724476346898635,
      "grad_norm": 0.7735070586204529,
      "learning_rate": 9.852993081299821e-06,
      "loss": 0.8712087869644165,
      "step": 287
    },
    {
      "epoch": 0.37374536164000244,
      "grad_norm": 0.7552995085716248,
      "learning_rate": 9.851267152326842e-06,
      "loss": 0.8405476808547974,
      "step": 288
    },
    {
      "epoch": 0.37504308859014135,
      "grad_norm": 0.7868830561637878,
      "learning_rate": 9.849531303740414e-06,
      "loss": 0.8763223886489868,
      "step": 289
    },
    {
      "epoch": 0.3763408155402802,
      "grad_norm": 0.7332586646080017,
      "learning_rate": 9.847785539089904e-06,
      "loss": 0.8512204885482788,
      "step": 290
    },
    {
      "epoch": 0.3776385424904191,
      "grad_norm": 0.75746089220047,
      "learning_rate": 9.846029861944964e-06,
      "loss": 0.8664734363555908,
      "step": 291
    },
    {
      "epoch": 0.378936269440558,
      "grad_norm": 0.7325623035430908,
      "learning_rate": 9.844264275895505e-06,
      "loss": 0.8830072283744812,
      "step": 292
    },
    {
      "epoch": 0.38023399639069694,
      "grad_norm": 0.7399287223815918,
      "learning_rate": 9.842488784551707e-06,
      "loss": 0.8711655735969543,
      "step": 293
    },
    {
      "epoch": 0.3815317233408358,
      "grad_norm": 0.7357679605484009,
      "learning_rate": 9.840703391543999e-06,
      "loss": 0.7781684994697571,
      "step": 294
    },
    {
      "epoch": 0.3828294502909747,
      "grad_norm": 0.7346934080123901,
      "learning_rate": 9.838908100523056e-06,
      "loss": 0.7901777029037476,
      "step": 295
    },
    {
      "epoch": 0.3841271772411136,
      "grad_norm": 0.7595230937004089,
      "learning_rate": 9.837102915159797e-06,
      "loss": 0.8042920827865601,
      "step": 296
    },
    {
      "epoch": 0.3854249041912525,
      "grad_norm": 0.7617968320846558,
      "learning_rate": 9.835287839145366e-06,
      "loss": 0.9179428815841675,
      "step": 297
    },
    {
      "epoch": 0.3867226311413914,
      "grad_norm": 0.7399303317070007,
      "learning_rate": 9.833462876191138e-06,
      "loss": 0.8573108911514282,
      "step": 298
    },
    {
      "epoch": 0.3880203580915303,
      "grad_norm": 0.7359921932220459,
      "learning_rate": 9.831628030028698e-06,
      "loss": 0.7790151238441467,
      "step": 299
    },
    {
      "epoch": 0.3893180850416692,
      "grad_norm": 0.7458881139755249,
      "learning_rate": 9.829783304409838e-06,
      "loss": 0.839961588382721,
      "step": 300
    },
    {
      "epoch": 0.3906158119918081,
      "grad_norm": 0.7518510222434998,
      "learning_rate": 9.827928703106562e-06,
      "loss": 0.9060590267181396,
      "step": 301
    },
    {
      "epoch": 0.39191353894194697,
      "grad_norm": 0.7119861841201782,
      "learning_rate": 9.826064229911056e-06,
      "loss": 0.8198513984680176,
      "step": 302
    },
    {
      "epoch": 0.3932112658920859,
      "grad_norm": 0.716594934463501,
      "learning_rate": 9.824189888635699e-06,
      "loss": 0.793634295463562,
      "step": 303
    },
    {
      "epoch": 0.3945089928422248,
      "grad_norm": 0.7235651612281799,
      "learning_rate": 9.82230568311304e-06,
      "loss": 0.8697519302368164,
      "step": 304
    },
    {
      "epoch": 0.3958067197923637,
      "grad_norm": 0.7011069655418396,
      "learning_rate": 9.820411617195807e-06,
      "loss": 0.7747907638549805,
      "step": 305
    },
    {
      "epoch": 0.39710444674250256,
      "grad_norm": 0.7803413271903992,
      "learning_rate": 9.818507694756883e-06,
      "loss": 0.9115470051765442,
      "step": 306
    },
    {
      "epoch": 0.39840217369264147,
      "grad_norm": 0.7196638584136963,
      "learning_rate": 9.816593919689305e-06,
      "loss": 0.8010602593421936,
      "step": 307
    },
    {
      "epoch": 0.3996999006427804,
      "grad_norm": 0.7336106896400452,
      "learning_rate": 9.814670295906265e-06,
      "loss": 0.8578990697860718,
      "step": 308
    },
    {
      "epoch": 0.4009976275929193,
      "grad_norm": 0.7282925844192505,
      "learning_rate": 9.81273682734108e-06,
      "loss": 0.8971216678619385,
      "step": 309
    },
    {
      "epoch": 0.40229535454305815,
      "grad_norm": 0.7463446259498596,
      "learning_rate": 9.81079351794721e-06,
      "loss": 0.8058433532714844,
      "step": 310
    },
    {
      "epoch": 0.40359308149319706,
      "grad_norm": 0.7901847958564758,
      "learning_rate": 9.808840371698226e-06,
      "loss": 0.896103024482727,
      "step": 311
    },
    {
      "epoch": 0.40489080844333597,
      "grad_norm": 0.72321617603302,
      "learning_rate": 9.80687739258782e-06,
      "loss": 0.8004294037818909,
      "step": 312
    },
    {
      "epoch": 0.4061885353934749,
      "grad_norm": 0.7592387199401855,
      "learning_rate": 9.804904584629786e-06,
      "loss": 0.8932079076766968,
      "step": 313
    },
    {
      "epoch": 0.40748626234361374,
      "grad_norm": 0.7692825198173523,
      "learning_rate": 9.80292195185802e-06,
      "loss": 0.8546319603919983,
      "step": 314
    },
    {
      "epoch": 0.40878398929375265,
      "grad_norm": 0.7320608496665955,
      "learning_rate": 9.800929498326502e-06,
      "loss": 0.8395287990570068,
      "step": 315
    },
    {
      "epoch": 0.41008171624389156,
      "grad_norm": 0.7121133208274841,
      "learning_rate": 9.798927228109294e-06,
      "loss": 0.8351479768753052,
      "step": 316
    },
    {
      "epoch": 0.41137944319403047,
      "grad_norm": 0.7649046778678894,
      "learning_rate": 9.796915145300534e-06,
      "loss": 0.90168696641922,
      "step": 317
    },
    {
      "epoch": 0.4126771701441694,
      "grad_norm": 0.759292721748352,
      "learning_rate": 9.794893254014421e-06,
      "loss": 0.8341972827911377,
      "step": 318
    },
    {
      "epoch": 0.41397489709430824,
      "grad_norm": 0.7619395852088928,
      "learning_rate": 9.792861558385212e-06,
      "loss": 0.8608151078224182,
      "step": 319
    },
    {
      "epoch": 0.41527262404444715,
      "grad_norm": 0.8039480447769165,
      "learning_rate": 9.790820062567208e-06,
      "loss": 0.8884308934211731,
      "step": 320
    },
    {
      "epoch": 0.41657035099458606,
      "grad_norm": 0.7317516207695007,
      "learning_rate": 9.788768770734753e-06,
      "loss": 0.8331541419029236,
      "step": 321
    },
    {
      "epoch": 0.41786807794472497,
      "grad_norm": 0.6900208592414856,
      "learning_rate": 9.78670768708222e-06,
      "loss": 0.7963373064994812,
      "step": 322
    },
    {
      "epoch": 0.4191658048948638,
      "grad_norm": 0.748304545879364,
      "learning_rate": 9.784636815824003e-06,
      "loss": 0.8610258102416992,
      "step": 323
    },
    {
      "epoch": 0.42046353184500274,
      "grad_norm": 0.7829216718673706,
      "learning_rate": 9.782556161194508e-06,
      "loss": 0.8490146994590759,
      "step": 324
    },
    {
      "epoch": 0.42176125879514165,
      "grad_norm": 0.7254793047904968,
      "learning_rate": 9.78046572744815e-06,
      "loss": 0.8324488401412964,
      "step": 325
    },
    {
      "epoch": 0.42305898574528056,
      "grad_norm": 0.7239816784858704,
      "learning_rate": 9.778365518859334e-06,
      "loss": 0.745983362197876,
      "step": 326
    },
    {
      "epoch": 0.4243567126954194,
      "grad_norm": 0.7268423438072205,
      "learning_rate": 9.776255539722457e-06,
      "loss": 0.8977281451225281,
      "step": 327
    },
    {
      "epoch": 0.4256544396455583,
      "grad_norm": 0.7335854172706604,
      "learning_rate": 9.774135794351892e-06,
      "loss": 0.8936282396316528,
      "step": 328
    },
    {
      "epoch": 0.42695216659569724,
      "grad_norm": 0.7278575897216797,
      "learning_rate": 9.77200628708198e-06,
      "loss": 0.7703112959861755,
      "step": 329
    },
    {
      "epoch": 0.42824989354583615,
      "grad_norm": 0.7280608415603638,
      "learning_rate": 9.769867022267028e-06,
      "loss": 0.9051704406738281,
      "step": 330
    },
    {
      "epoch": 0.429547620495975,
      "grad_norm": 0.7283042073249817,
      "learning_rate": 9.767718004281288e-06,
      "loss": 0.8391573429107666,
      "step": 331
    },
    {
      "epoch": 0.4308453474461139,
      "grad_norm": 0.7032507658004761,
      "learning_rate": 9.765559237518958e-06,
      "loss": 0.7801662087440491,
      "step": 332
    },
    {
      "epoch": 0.4321430743962528,
      "grad_norm": 0.7365294694900513,
      "learning_rate": 9.763390726394171e-06,
      "loss": 0.8532007932662964,
      "step": 333
    },
    {
      "epoch": 0.43344080134639174,
      "grad_norm": 0.7675997018814087,
      "learning_rate": 9.761212475340982e-06,
      "loss": 0.8515892624855042,
      "step": 334
    },
    {
      "epoch": 0.4347385282965306,
      "grad_norm": 0.7382540702819824,
      "learning_rate": 9.759024488813364e-06,
      "loss": 0.7981022596359253,
      "step": 335
    },
    {
      "epoch": 0.4360362552466695,
      "grad_norm": 0.7753154039382935,
      "learning_rate": 9.756826771285195e-06,
      "loss": 0.7946685552597046,
      "step": 336
    },
    {
      "epoch": 0.4373339821968084,
      "grad_norm": 0.6813466548919678,
      "learning_rate": 9.754619327250253e-06,
      "loss": 0.7935984134674072,
      "step": 337
    },
    {
      "epoch": 0.4386317091469473,
      "grad_norm": 0.7297438979148865,
      "learning_rate": 9.7524021612222e-06,
      "loss": 0.8265496492385864,
      "step": 338
    },
    {
      "epoch": 0.4399294360970862,
      "grad_norm": 0.7127614617347717,
      "learning_rate": 9.750175277734582e-06,
      "loss": 0.8402823209762573,
      "step": 339
    },
    {
      "epoch": 0.4412271630472251,
      "grad_norm": 0.7071698904037476,
      "learning_rate": 9.747938681340807e-06,
      "loss": 0.7340988516807556,
      "step": 340
    },
    {
      "epoch": 0.442524889997364,
      "grad_norm": 0.7566197514533997,
      "learning_rate": 9.745692376614154e-06,
      "loss": 0.8520500063896179,
      "step": 341
    },
    {
      "epoch": 0.4438226169475029,
      "grad_norm": 0.7088791728019714,
      "learning_rate": 9.743436368147745e-06,
      "loss": 0.7458124756813049,
      "step": 342
    },
    {
      "epoch": 0.44512034389764177,
      "grad_norm": 0.7586151361465454,
      "learning_rate": 9.741170660554548e-06,
      "loss": 0.83961021900177,
      "step": 343
    },
    {
      "epoch": 0.4464180708477807,
      "grad_norm": 0.7471207976341248,
      "learning_rate": 9.73889525846736e-06,
      "loss": 0.8236647844314575,
      "step": 344
    },
    {
      "epoch": 0.4477157977979196,
      "grad_norm": 0.7463803887367249,
      "learning_rate": 9.736610166538802e-06,
      "loss": 0.9211429357528687,
      "step": 345
    },
    {
      "epoch": 0.4490135247480585,
      "grad_norm": 0.7657600045204163,
      "learning_rate": 9.73431538944131e-06,
      "loss": 0.9619902968406677,
      "step": 346
    },
    {
      "epoch": 0.45031125169819736,
      "grad_norm": 0.7406160831451416,
      "learning_rate": 9.73201093186712e-06,
      "loss": 0.869321346282959,
      "step": 347
    },
    {
      "epoch": 0.45160897864833627,
      "grad_norm": 0.7429345846176147,
      "learning_rate": 9.729696798528268e-06,
      "loss": 0.8034260272979736,
      "step": 348
    },
    {
      "epoch": 0.4529067055984752,
      "grad_norm": 0.7394105195999146,
      "learning_rate": 9.727372994156568e-06,
      "loss": 0.8107494115829468,
      "step": 349
    },
    {
      "epoch": 0.4542044325486141,
      "grad_norm": 0.740679919719696,
      "learning_rate": 9.725039523503615e-06,
      "loss": 0.8520351052284241,
      "step": 350
    },
    {
      "epoch": 0.45550215949875295,
      "grad_norm": 0.7290498614311218,
      "learning_rate": 9.722696391340762e-06,
      "loss": 0.768467903137207,
      "step": 351
    },
    {
      "epoch": 0.45679988644889186,
      "grad_norm": 0.703825056552887,
      "learning_rate": 9.720343602459123e-06,
      "loss": 0.8336674571037292,
      "step": 352
    },
    {
      "epoch": 0.45809761339903077,
      "grad_norm": 0.7116036415100098,
      "learning_rate": 9.717981161669556e-06,
      "loss": 0.8118845820426941,
      "step": 353
    },
    {
      "epoch": 0.4593953403491697,
      "grad_norm": 0.7270032167434692,
      "learning_rate": 9.715609073802653e-06,
      "loss": 0.902847945690155,
      "step": 354
    },
    {
      "epoch": 0.46069306729930853,
      "grad_norm": 0.7389693260192871,
      "learning_rate": 9.713227343708737e-06,
      "loss": 0.8098748326301575,
      "step": 355
    },
    {
      "epoch": 0.46199079424944745,
      "grad_norm": 0.7012364268302917,
      "learning_rate": 9.71083597625784e-06,
      "loss": 0.7249214053153992,
      "step": 356
    },
    {
      "epoch": 0.46328852119958636,
      "grad_norm": 0.7171372175216675,
      "learning_rate": 9.708434976339704e-06,
      "loss": 0.8786948323249817,
      "step": 357
    },
    {
      "epoch": 0.46458624814972527,
      "grad_norm": 0.7394458055496216,
      "learning_rate": 9.706024348863766e-06,
      "loss": 0.8580265045166016,
      "step": 358
    },
    {
      "epoch": 0.4658839750998641,
      "grad_norm": 0.7083274722099304,
      "learning_rate": 9.703604098759148e-06,
      "loss": 0.8410412073135376,
      "step": 359
    },
    {
      "epoch": 0.46718170205000303,
      "grad_norm": 0.712253749370575,
      "learning_rate": 9.70117423097465e-06,
      "loss": 0.8000701665878296,
      "step": 360
    },
    {
      "epoch": 0.46847942900014194,
      "grad_norm": 0.7887440919876099,
      "learning_rate": 9.698734750478739e-06,
      "loss": 0.8630033731460571,
      "step": 361
    },
    {
      "epoch": 0.46977715595028086,
      "grad_norm": 0.7340478897094727,
      "learning_rate": 9.69628566225953e-06,
      "loss": 0.8977490663528442,
      "step": 362
    },
    {
      "epoch": 0.4710748829004197,
      "grad_norm": 0.7779148817062378,
      "learning_rate": 9.693826971324793e-06,
      "loss": 0.832167387008667,
      "step": 363
    },
    {
      "epoch": 0.4723726098505586,
      "grad_norm": 0.7122468948364258,
      "learning_rate": 9.691358682701927e-06,
      "loss": 0.857900857925415,
      "step": 364
    },
    {
      "epoch": 0.47367033680069753,
      "grad_norm": 0.7308494448661804,
      "learning_rate": 9.688880801437957e-06,
      "loss": 0.8315039873123169,
      "step": 365
    },
    {
      "epoch": 0.47496806375083644,
      "grad_norm": 0.7546078562736511,
      "learning_rate": 9.686393332599525e-06,
      "loss": 0.8919548988342285,
      "step": 366
    },
    {
      "epoch": 0.4762657907009753,
      "grad_norm": 0.7366290092468262,
      "learning_rate": 9.683896281272872e-06,
      "loss": 0.8298309445381165,
      "step": 367
    },
    {
      "epoch": 0.4775635176511142,
      "grad_norm": 0.7167381644248962,
      "learning_rate": 9.681389652563837e-06,
      "loss": 0.7846927642822266,
      "step": 368
    },
    {
      "epoch": 0.4788612446012531,
      "grad_norm": 0.7552225589752197,
      "learning_rate": 9.678873451597843e-06,
      "loss": 0.8357439637184143,
      "step": 369
    },
    {
      "epoch": 0.48015897155139203,
      "grad_norm": 0.7224664092063904,
      "learning_rate": 9.676347683519882e-06,
      "loss": 0.81117844581604,
      "step": 370
    },
    {
      "epoch": 0.4814566985015309,
      "grad_norm": 0.7118367552757263,
      "learning_rate": 9.673812353494513e-06,
      "loss": 0.8407114148139954,
      "step": 371
    },
    {
      "epoch": 0.4827544254516698,
      "grad_norm": 0.769687294960022,
      "learning_rate": 9.671267466705841e-06,
      "loss": 0.8985891342163086,
      "step": 372
    },
    {
      "epoch": 0.4840521524018087,
      "grad_norm": 0.7327465415000916,
      "learning_rate": 9.668713028357518e-06,
      "loss": 0.8324825167655945,
      "step": 373
    },
    {
      "epoch": 0.4853498793519476,
      "grad_norm": 0.7223183512687683,
      "learning_rate": 9.666149043672724e-06,
      "loss": 0.8175463080406189,
      "step": 374
    },
    {
      "epoch": 0.4866476063020865,
      "grad_norm": 0.7580711245536804,
      "learning_rate": 9.663575517894155e-06,
      "loss": 0.8757423162460327,
      "step": 375
    },
    {
      "epoch": 0.4879453332522254,
      "grad_norm": 0.7004430890083313,
      "learning_rate": 9.660992456284024e-06,
      "loss": 0.7426130771636963,
      "step": 376
    },
    {
      "epoch": 0.4892430602023643,
      "grad_norm": 0.7414857149124146,
      "learning_rate": 9.658399864124037e-06,
      "loss": 0.8230442404747009,
      "step": 377
    },
    {
      "epoch": 0.4905407871525032,
      "grad_norm": 0.7824345231056213,
      "learning_rate": 9.655797746715388e-06,
      "loss": 0.9073444604873657,
      "step": 378
    },
    {
      "epoch": 0.49183851410264207,
      "grad_norm": 0.7817385792732239,
      "learning_rate": 9.65318610937875e-06,
      "loss": 0.8738477826118469,
      "step": 379
    },
    {
      "epoch": 0.493136241052781,
      "grad_norm": 0.7348605394363403,
      "learning_rate": 9.650564957454258e-06,
      "loss": 0.8755439519882202,
      "step": 380
    },
    {
      "epoch": 0.4944339680029199,
      "grad_norm": 0.7649983763694763,
      "learning_rate": 9.647934296301506e-06,
      "loss": 1.0017733573913574,
      "step": 381
    },
    {
      "epoch": 0.4957316949530588,
      "grad_norm": 0.735651969909668,
      "learning_rate": 9.64529413129953e-06,
      "loss": 0.8582989573478699,
      "step": 382
    },
    {
      "epoch": 0.49702942190319765,
      "grad_norm": 0.7229171395301819,
      "learning_rate": 9.642644467846799e-06,
      "loss": 0.8139792680740356,
      "step": 383
    },
    {
      "epoch": 0.49832714885333657,
      "grad_norm": 0.7279280424118042,
      "learning_rate": 9.639985311361202e-06,
      "loss": 0.8503962159156799,
      "step": 384
    },
    {
      "epoch": 0.4996248758034755,
      "grad_norm": 0.7445419430732727,
      "learning_rate": 9.637316667280046e-06,
      "loss": 0.9100584387779236,
      "step": 385
    },
    {
      "epoch": 0.5009226027536143,
      "grad_norm": 0.7147620916366577,
      "learning_rate": 9.634638541060027e-06,
      "loss": 0.8684252500534058,
      "step": 386
    },
    {
      "epoch": 0.5022203297037533,
      "grad_norm": 0.7348777055740356,
      "learning_rate": 9.63195093817724e-06,
      "loss": 0.8720637559890747,
      "step": 387
    },
    {
      "epoch": 0.5035180566538922,
      "grad_norm": 0.743008553981781,
      "learning_rate": 9.62925386412715e-06,
      "loss": 0.8478811979293823,
      "step": 388
    },
    {
      "epoch": 0.504815783604031,
      "grad_norm": 0.7702372670173645,
      "learning_rate": 9.626547324424592e-06,
      "loss": 0.8293808102607727,
      "step": 389
    },
    {
      "epoch": 0.50611351055417,
      "grad_norm": 0.8113992810249329,
      "learning_rate": 9.623831324603755e-06,
      "loss": 0.9317741990089417,
      "step": 390
    },
    {
      "epoch": 0.5074112375043088,
      "grad_norm": 0.7273135781288147,
      "learning_rate": 9.621105870218167e-06,
      "loss": 0.840066134929657,
      "step": 391
    },
    {
      "epoch": 0.5087089644544478,
      "grad_norm": 0.7781491875648499,
      "learning_rate": 9.618370966840698e-06,
      "loss": 0.8440302610397339,
      "step": 392
    },
    {
      "epoch": 0.5100066914045867,
      "grad_norm": 0.7643720507621765,
      "learning_rate": 9.615626620063531e-06,
      "loss": 0.7832410335540771,
      "step": 393
    },
    {
      "epoch": 0.5113044183547255,
      "grad_norm": 0.7325314283370972,
      "learning_rate": 9.61287283549816e-06,
      "loss": 0.8389537334442139,
      "step": 394
    },
    {
      "epoch": 0.5126021453048645,
      "grad_norm": 0.7445440292358398,
      "learning_rate": 9.610109618775379e-06,
      "loss": 0.8239348530769348,
      "step": 395
    },
    {
      "epoch": 0.5138998722550033,
      "grad_norm": 0.7088785767555237,
      "learning_rate": 9.607336975545264e-06,
      "loss": 0.8060757517814636,
      "step": 396
    },
    {
      "epoch": 0.5151975992051423,
      "grad_norm": 0.8243458271026611,
      "learning_rate": 9.604554911477173e-06,
      "loss": 0.8679091334342957,
      "step": 397
    },
    {
      "epoch": 0.5164953261552812,
      "grad_norm": 0.8044202327728271,
      "learning_rate": 9.601763432259716e-06,
      "loss": 0.8719264268875122,
      "step": 398
    },
    {
      "epoch": 0.51779305310542,
      "grad_norm": 0.6826696991920471,
      "learning_rate": 9.59896254360077e-06,
      "loss": 0.8089145421981812,
      "step": 399
    },
    {
      "epoch": 0.519090780055559,
      "grad_norm": 0.7384952306747437,
      "learning_rate": 9.596152251227438e-06,
      "loss": 0.8086674213409424,
      "step": 400
    },
    {
      "epoch": 0.5203885070056978,
      "grad_norm": 0.7176761031150818,
      "learning_rate": 9.593332560886055e-06,
      "loss": 0.8248165845870972,
      "step": 401
    },
    {
      "epoch": 0.5216862339558367,
      "grad_norm": 0.7176565527915955,
      "learning_rate": 9.59050347834218e-06,
      "loss": 0.7899567484855652,
      "step": 402
    },
    {
      "epoch": 0.5229839609059757,
      "grad_norm": 0.7689147591590881,
      "learning_rate": 9.587665009380565e-06,
      "loss": 0.8428173065185547,
      "step": 403
    },
    {
      "epoch": 0.5242816878561145,
      "grad_norm": 0.7173596024513245,
      "learning_rate": 9.584817159805164e-06,
      "loss": 0.8865775465965271,
      "step": 404
    },
    {
      "epoch": 0.5255794148062535,
      "grad_norm": 0.742651641368866,
      "learning_rate": 9.58195993543911e-06,
      "loss": 0.8390700817108154,
      "step": 405
    },
    {
      "epoch": 0.5268771417563923,
      "grad_norm": 0.8015622496604919,
      "learning_rate": 9.579093342124699e-06,
      "loss": 0.8883824944496155,
      "step": 406
    },
    {
      "epoch": 0.5281748687065312,
      "grad_norm": 0.7367856502532959,
      "learning_rate": 9.576217385723391e-06,
      "loss": 0.7946388125419617,
      "step": 407
    },
    {
      "epoch": 0.5294725956566702,
      "grad_norm": 0.6854470372200012,
      "learning_rate": 9.57333207211579e-06,
      "loss": 0.7881499528884888,
      "step": 408
    },
    {
      "epoch": 0.530770322606809,
      "grad_norm": 0.7024526596069336,
      "learning_rate": 9.57043740720163e-06,
      "loss": 0.840889036655426,
      "step": 409
    },
    {
      "epoch": 0.5320680495569479,
      "grad_norm": 0.7624524235725403,
      "learning_rate": 9.567533396899769e-06,
      "loss": 0.8071510791778564,
      "step": 410
    },
    {
      "epoch": 0.5333657765070868,
      "grad_norm": 0.7562685012817383,
      "learning_rate": 9.564620047148174e-06,
      "loss": 0.8523662090301514,
      "step": 411
    },
    {
      "epoch": 0.5346635034572257,
      "grad_norm": 0.7493570446968079,
      "learning_rate": 9.561697363903908e-06,
      "loss": 0.8793864846229553,
      "step": 412
    },
    {
      "epoch": 0.5359612304073647,
      "grad_norm": 0.7925007343292236,
      "learning_rate": 9.558765353143116e-06,
      "loss": 0.8957589268684387,
      "step": 413
    },
    {
      "epoch": 0.5372589573575035,
      "grad_norm": 0.7443920969963074,
      "learning_rate": 9.555824020861022e-06,
      "loss": 0.8397520184516907,
      "step": 414
    },
    {
      "epoch": 0.5385566843076424,
      "grad_norm": 0.7451807260513306,
      "learning_rate": 9.5528733730719e-06,
      "loss": 0.8264951705932617,
      "step": 415
    },
    {
      "epoch": 0.5398544112577813,
      "grad_norm": 0.756137490272522,
      "learning_rate": 9.549913415809084e-06,
      "loss": 0.7959051728248596,
      "step": 416
    },
    {
      "epoch": 0.5411521382079202,
      "grad_norm": 0.6989119052886963,
      "learning_rate": 9.546944155124935e-06,
      "loss": 0.8344296813011169,
      "step": 417
    },
    {
      "epoch": 0.542449865158059,
      "grad_norm": 0.7302668690681458,
      "learning_rate": 9.54396559709084e-06,
      "loss": 0.8438407778739929,
      "step": 418
    },
    {
      "epoch": 0.543747592108198,
      "grad_norm": 0.7969427704811096,
      "learning_rate": 9.540977747797194e-06,
      "loss": 0.7917635440826416,
      "step": 419
    },
    {
      "epoch": 0.5450453190583369,
      "grad_norm": 0.7352253794670105,
      "learning_rate": 9.537980613353392e-06,
      "loss": 0.8681586384773254,
      "step": 420
    },
    {
      "epoch": 0.5463430460084758,
      "grad_norm": 0.7419816255569458,
      "learning_rate": 9.53497419988782e-06,
      "loss": 0.8770594596862793,
      "step": 421
    },
    {
      "epoch": 0.5476407729586147,
      "grad_norm": 0.7256708741188049,
      "learning_rate": 9.531958513547832e-06,
      "loss": 0.8533478379249573,
      "step": 422
    },
    {
      "epoch": 0.5489384999087535,
      "grad_norm": 0.8208673596382141,
      "learning_rate": 9.52893356049974e-06,
      "loss": 0.9058896899223328,
      "step": 423
    },
    {
      "epoch": 0.5502362268588925,
      "grad_norm": 0.7950146794319153,
      "learning_rate": 9.525899346928809e-06,
      "loss": 0.8501814603805542,
      "step": 424
    },
    {
      "epoch": 0.5515339538090314,
      "grad_norm": 0.7618380188941956,
      "learning_rate": 9.52285587903924e-06,
      "loss": 0.8879157900810242,
      "step": 425
    },
    {
      "epoch": 0.5528316807591702,
      "grad_norm": 0.6897137761116028,
      "learning_rate": 9.519803163054149e-06,
      "loss": 0.82085782289505,
      "step": 426
    },
    {
      "epoch": 0.5541294077093092,
      "grad_norm": 0.7675971984863281,
      "learning_rate": 9.51674120521557e-06,
      "loss": 0.8406592011451721,
      "step": 427
    },
    {
      "epoch": 0.555427134659448,
      "grad_norm": 0.7005506157875061,
      "learning_rate": 9.513670011784435e-06,
      "loss": 0.8337131142616272,
      "step": 428
    },
    {
      "epoch": 0.556724861609587,
      "grad_norm": 0.7315273880958557,
      "learning_rate": 9.510589589040554e-06,
      "loss": 0.7294406890869141,
      "step": 429
    },
    {
      "epoch": 0.5580225885597259,
      "grad_norm": 0.74063640832901,
      "learning_rate": 9.507499943282613e-06,
      "loss": 0.7584653496742249,
      "step": 430
    },
    {
      "epoch": 0.5593203155098647,
      "grad_norm": 0.7152555584907532,
      "learning_rate": 9.504401080828154e-06,
      "loss": 0.811851441860199,
      "step": 431
    },
    {
      "epoch": 0.5606180424600037,
      "grad_norm": 0.7302674055099487,
      "learning_rate": 9.501293008013568e-06,
      "loss": 0.8225210905075073,
      "step": 432
    },
    {
      "epoch": 0.5619157694101425,
      "grad_norm": 0.7426562905311584,
      "learning_rate": 9.498175731194077e-06,
      "loss": 0.8669119477272034,
      "step": 433
    },
    {
      "epoch": 0.5632134963602814,
      "grad_norm": 0.7812010049819946,
      "learning_rate": 9.495049256743723e-06,
      "loss": 0.9062970280647278,
      "step": 434
    },
    {
      "epoch": 0.5645112233104204,
      "grad_norm": 0.7181828022003174,
      "learning_rate": 9.491913591055356e-06,
      "loss": 0.7689467072486877,
      "step": 435
    },
    {
      "epoch": 0.5658089502605592,
      "grad_norm": 0.7515583038330078,
      "learning_rate": 9.488768740540615e-06,
      "loss": 0.8905704021453857,
      "step": 436
    },
    {
      "epoch": 0.5671066772106982,
      "grad_norm": 0.6940712332725525,
      "learning_rate": 9.485614711629927e-06,
      "loss": 0.7591986656188965,
      "step": 437
    },
    {
      "epoch": 0.568404404160837,
      "grad_norm": 0.7081801891326904,
      "learning_rate": 9.482451510772482e-06,
      "loss": 0.8202489018440247,
      "step": 438
    },
    {
      "epoch": 0.5697021311109759,
      "grad_norm": 0.7051956057548523,
      "learning_rate": 9.479279144436224e-06,
      "loss": 0.8634527921676636,
      "step": 439
    },
    {
      "epoch": 0.5709998580611149,
      "grad_norm": 0.7323376536369324,
      "learning_rate": 9.47609761910784e-06,
      "loss": 0.802325427532196,
      "step": 440
    },
    {
      "epoch": 0.5722975850112537,
      "grad_norm": 0.7712946534156799,
      "learning_rate": 9.472906941292746e-06,
      "loss": 0.8281590342521667,
      "step": 441
    },
    {
      "epoch": 0.5735953119613926,
      "grad_norm": 0.7743613719940186,
      "learning_rate": 9.469707117515068e-06,
      "loss": 0.8420519828796387,
      "step": 442
    },
    {
      "epoch": 0.5748930389115315,
      "grad_norm": 0.7498214840888977,
      "learning_rate": 9.466498154317635e-06,
      "loss": 0.7940037250518799,
      "step": 443
    },
    {
      "epoch": 0.5761907658616704,
      "grad_norm": 0.7323300838470459,
      "learning_rate": 9.463280058261965e-06,
      "loss": 0.8547796010971069,
      "step": 444
    },
    {
      "epoch": 0.5774884928118094,
      "grad_norm": 0.7174122929573059,
      "learning_rate": 9.460052835928254e-06,
      "loss": 0.8065235614776611,
      "step": 445
    },
    {
      "epoch": 0.5787862197619482,
      "grad_norm": 0.7354016304016113,
      "learning_rate": 9.45681649391535e-06,
      "loss": 0.7716302275657654,
      "step": 446
    },
    {
      "epoch": 0.5800839467120871,
      "grad_norm": 0.6930510997772217,
      "learning_rate": 9.453571038840755e-06,
      "loss": 0.7406535744667053,
      "step": 447
    },
    {
      "epoch": 0.581381673662226,
      "grad_norm": 0.7596374154090881,
      "learning_rate": 9.450316477340602e-06,
      "loss": 0.8508278727531433,
      "step": 448
    },
    {
      "epoch": 0.5826794006123649,
      "grad_norm": 0.7328705787658691,
      "learning_rate": 9.447052816069648e-06,
      "loss": 0.8054483532905579,
      "step": 449
    },
    {
      "epoch": 0.5839771275625038,
      "grad_norm": 0.7718313932418823,
      "learning_rate": 9.443780061701252e-06,
      "loss": 0.8918864727020264,
      "step": 450
    },
    {
      "epoch": 0.5852748545126427,
      "grad_norm": 0.7557240724563599,
      "learning_rate": 9.44049822092737e-06,
      "loss": 0.8130478858947754,
      "step": 451
    },
    {
      "epoch": 0.5865725814627816,
      "grad_norm": 0.7418437600135803,
      "learning_rate": 9.437207300458535e-06,
      "loss": 0.8093717098236084,
      "step": 452
    },
    {
      "epoch": 0.5878703084129205,
      "grad_norm": 0.7124131321907043,
      "learning_rate": 9.433907307023845e-06,
      "loss": 0.8172003626823425,
      "step": 453
    },
    {
      "epoch": 0.5891680353630594,
      "grad_norm": 0.7546778321266174,
      "learning_rate": 9.430598247370955e-06,
      "loss": 0.7902172207832336,
      "step": 454
    },
    {
      "epoch": 0.5904657623131983,
      "grad_norm": 0.7268347144126892,
      "learning_rate": 9.427280128266049e-06,
      "loss": 0.7638229727745056,
      "step": 455
    },
    {
      "epoch": 0.5917634892633372,
      "grad_norm": 0.7774848937988281,
      "learning_rate": 9.423952956493846e-06,
      "loss": 0.7811177372932434,
      "step": 456
    },
    {
      "epoch": 0.5930612162134761,
      "grad_norm": 0.7768798470497131,
      "learning_rate": 9.420616738857568e-06,
      "loss": 0.79379802942276,
      "step": 457
    },
    {
      "epoch": 0.5943589431636149,
      "grad_norm": 0.7454319596290588,
      "learning_rate": 9.417271482178938e-06,
      "loss": 0.8783116340637207,
      "step": 458
    },
    {
      "epoch": 0.5956566701137539,
      "grad_norm": 0.6949381232261658,
      "learning_rate": 9.413917193298153e-06,
      "loss": 0.8404789566993713,
      "step": 459
    },
    {
      "epoch": 0.5969543970638927,
      "grad_norm": 0.704766035079956,
      "learning_rate": 9.41055387907389e-06,
      "loss": 0.7789646983146667,
      "step": 460
    },
    {
      "epoch": 0.5969543970638927,
      "eval_loss": 0.8027161359786987,
      "eval_runtime": 139.819,
      "eval_samples_per_second": 37.134,
      "eval_steps_per_second": 9.283,
      "step": 460
    },
    {
      "epoch": 0.5982521240140317,
      "grad_norm": 0.7682297229766846,
      "learning_rate": 9.407181546383275e-06,
      "loss": 0.8239257335662842,
      "step": 461
    },
    {
      "epoch": 0.5995498509641706,
      "grad_norm": 0.7389854192733765,
      "learning_rate": 9.403800202121873e-06,
      "loss": 0.7810261845588684,
      "step": 462
    },
    {
      "epoch": 0.6008475779143094,
      "grad_norm": 0.6934480667114258,
      "learning_rate": 9.400409853203677e-06,
      "loss": 0.8098300099372864,
      "step": 463
    },
    {
      "epoch": 0.6021453048644484,
      "grad_norm": 0.7336373925209045,
      "learning_rate": 9.397010506561096e-06,
      "loss": 0.7502899765968323,
      "step": 464
    },
    {
      "epoch": 0.6034430318145872,
      "grad_norm": 0.7731090784072876,
      "learning_rate": 9.393602169144929e-06,
      "loss": 0.8462476134300232,
      "step": 465
    },
    {
      "epoch": 0.6047407587647261,
      "grad_norm": 0.7425335645675659,
      "learning_rate": 9.390184847924366e-06,
      "loss": 0.813231348991394,
      "step": 466
    },
    {
      "epoch": 0.6060384857148651,
      "grad_norm": 0.7691134214401245,
      "learning_rate": 9.386758549886964e-06,
      "loss": 0.872662365436554,
      "step": 467
    },
    {
      "epoch": 0.6073362126650039,
      "grad_norm": 0.7183594107627869,
      "learning_rate": 9.383323282038632e-06,
      "loss": 0.8333174586296082,
      "step": 468
    },
    {
      "epoch": 0.6086339396151429,
      "grad_norm": 0.7414330244064331,
      "learning_rate": 9.379879051403627e-06,
      "loss": 0.7815148234367371,
      "step": 469
    },
    {
      "epoch": 0.6099316665652817,
      "grad_norm": 0.753852903842926,
      "learning_rate": 9.376425865024527e-06,
      "loss": 0.8778871297836304,
      "step": 470
    },
    {
      "epoch": 0.6112293935154206,
      "grad_norm": 0.7895762920379639,
      "learning_rate": 9.372963729962227e-06,
      "loss": 0.930837869644165,
      "step": 471
    },
    {
      "epoch": 0.6125271204655596,
      "grad_norm": 0.75229412317276,
      "learning_rate": 9.369492653295913e-06,
      "loss": 0.7898579239845276,
      "step": 472
    },
    {
      "epoch": 0.6138248474156984,
      "grad_norm": 0.6872172951698303,
      "learning_rate": 9.366012642123061e-06,
      "loss": 0.8121792078018188,
      "step": 473
    },
    {
      "epoch": 0.6151225743658374,
      "grad_norm": 0.719164252281189,
      "learning_rate": 9.362523703559412e-06,
      "loss": 0.812069833278656,
      "step": 474
    },
    {
      "epoch": 0.6164203013159762,
      "grad_norm": 0.7363712787628174,
      "learning_rate": 9.359025844738962e-06,
      "loss": 0.8511337041854858,
      "step": 475
    },
    {
      "epoch": 0.6177180282661151,
      "grad_norm": 0.7480281591415405,
      "learning_rate": 9.355519072813946e-06,
      "loss": 0.8967547416687012,
      "step": 476
    },
    {
      "epoch": 0.6190157552162541,
      "grad_norm": 0.7672280073165894,
      "learning_rate": 9.352003394954827e-06,
      "loss": 0.8544362783432007,
      "step": 477
    },
    {
      "epoch": 0.6203134821663929,
      "grad_norm": 0.7103601694107056,
      "learning_rate": 9.348478818350277e-06,
      "loss": 0.8329444527626038,
      "step": 478
    },
    {
      "epoch": 0.6216112091165318,
      "grad_norm": 0.7819087505340576,
      "learning_rate": 9.34494535020716e-06,
      "loss": 0.868659496307373,
      "step": 479
    },
    {
      "epoch": 0.6229089360666707,
      "grad_norm": 0.8021608591079712,
      "learning_rate": 9.341402997750526e-06,
      "loss": 0.8001782298088074,
      "step": 480
    },
    {
      "epoch": 0.6242066630168096,
      "grad_norm": 0.7665423154830933,
      "learning_rate": 9.337851768223589e-06,
      "loss": 0.8383674025535583,
      "step": 481
    },
    {
      "epoch": 0.6255043899669486,
      "grad_norm": 0.7381864786148071,
      "learning_rate": 9.334291668887716e-06,
      "loss": 0.8331637382507324,
      "step": 482
    },
    {
      "epoch": 0.6268021169170874,
      "grad_norm": 0.6999797821044922,
      "learning_rate": 9.330722707022406e-06,
      "loss": 0.8383775353431702,
      "step": 483
    },
    {
      "epoch": 0.6280998438672263,
      "grad_norm": 0.7365379333496094,
      "learning_rate": 9.327144889925286e-06,
      "loss": 0.8450272679328918,
      "step": 484
    },
    {
      "epoch": 0.6293975708173652,
      "grad_norm": 0.7899283766746521,
      "learning_rate": 9.323558224912083e-06,
      "loss": 0.8851337432861328,
      "step": 485
    },
    {
      "epoch": 0.6306952977675041,
      "grad_norm": 0.7195940017700195,
      "learning_rate": 9.319962719316621e-06,
      "loss": 0.8370380401611328,
      "step": 486
    },
    {
      "epoch": 0.631993024717643,
      "grad_norm": 0.7461369037628174,
      "learning_rate": 9.3163583804908e-06,
      "loss": 0.7896135449409485,
      "step": 487
    },
    {
      "epoch": 0.6332907516677819,
      "grad_norm": 0.7383708357810974,
      "learning_rate": 9.312745215804577e-06,
      "loss": 0.9348940849304199,
      "step": 488
    },
    {
      "epoch": 0.6345884786179208,
      "grad_norm": 0.758083701133728,
      "learning_rate": 9.309123232645963e-06,
      "loss": 0.8204320669174194,
      "step": 489
    },
    {
      "epoch": 0.6358862055680597,
      "grad_norm": 0.7442320585250854,
      "learning_rate": 9.305492438420995e-06,
      "loss": 0.8042322397232056,
      "step": 490
    },
    {
      "epoch": 0.6371839325181986,
      "grad_norm": 0.7107361555099487,
      "learning_rate": 9.301852840553728e-06,
      "loss": 0.7795714735984802,
      "step": 491
    },
    {
      "epoch": 0.6384816594683375,
      "grad_norm": 0.760767936706543,
      "learning_rate": 9.298204446486221e-06,
      "loss": 0.8364303112030029,
      "step": 492
    },
    {
      "epoch": 0.6397793864184764,
      "grad_norm": 0.7145869135856628,
      "learning_rate": 9.294547263678515e-06,
      "loss": 0.7660549283027649,
      "step": 493
    },
    {
      "epoch": 0.6410771133686153,
      "grad_norm": 0.788299560546875,
      "learning_rate": 9.29088129960862e-06,
      "loss": 0.8712038993835449,
      "step": 494
    },
    {
      "epoch": 0.6423748403187541,
      "grad_norm": 0.7306452989578247,
      "learning_rate": 9.28720656177251e-06,
      "loss": 0.8043104410171509,
      "step": 495
    },
    {
      "epoch": 0.6436725672688931,
      "grad_norm": 0.7626951336860657,
      "learning_rate": 9.28352305768409e-06,
      "loss": 0.8437666296958923,
      "step": 496
    },
    {
      "epoch": 0.644970294219032,
      "grad_norm": 0.7394039630889893,
      "learning_rate": 9.279830794875194e-06,
      "loss": 0.8267931342124939,
      "step": 497
    },
    {
      "epoch": 0.6462680211691709,
      "grad_norm": 0.7661542296409607,
      "learning_rate": 9.276129780895566e-06,
      "loss": 0.8075953722000122,
      "step": 498
    },
    {
      "epoch": 0.6475657481193098,
      "grad_norm": 0.7846643328666687,
      "learning_rate": 9.272420023312843e-06,
      "loss": 0.9408265352249146,
      "step": 499
    },
    {
      "epoch": 0.6488634750694486,
      "grad_norm": 0.7860012054443359,
      "learning_rate": 9.268701529712541e-06,
      "loss": 0.8661449551582336,
      "step": 500
    },
    {
      "epoch": 0.6501612020195876,
      "grad_norm": 0.7225436568260193,
      "learning_rate": 9.264974307698034e-06,
      "loss": 0.7973988652229309,
      "step": 501
    },
    {
      "epoch": 0.6514589289697265,
      "grad_norm": 0.7316765189170837,
      "learning_rate": 9.261238364890553e-06,
      "loss": 0.8226317167282104,
      "step": 502
    },
    {
      "epoch": 0.6527566559198653,
      "grad_norm": 0.7312497496604919,
      "learning_rate": 9.257493708929153e-06,
      "loss": 0.8850467801094055,
      "step": 503
    },
    {
      "epoch": 0.6540543828700043,
      "grad_norm": 0.7464125752449036,
      "learning_rate": 9.253740347470708e-06,
      "loss": 0.8953540921211243,
      "step": 504
    },
    {
      "epoch": 0.6553521098201431,
      "grad_norm": 0.7750662565231323,
      "learning_rate": 9.24997828818989e-06,
      "loss": 0.860468327999115,
      "step": 505
    },
    {
      "epoch": 0.6566498367702821,
      "grad_norm": 0.72883540391922,
      "learning_rate": 9.246207538779162e-06,
      "loss": 0.8943373560905457,
      "step": 506
    },
    {
      "epoch": 0.657947563720421,
      "grad_norm": 0.7219229340553284,
      "learning_rate": 9.242428106948748e-06,
      "loss": 0.8636921048164368,
      "step": 507
    },
    {
      "epoch": 0.6592452906705598,
      "grad_norm": 0.7252084612846375,
      "learning_rate": 9.238640000426635e-06,
      "loss": 0.8572896718978882,
      "step": 508
    },
    {
      "epoch": 0.6605430176206988,
      "grad_norm": 0.6815743446350098,
      "learning_rate": 9.234843226958537e-06,
      "loss": 0.7989461421966553,
      "step": 509
    },
    {
      "epoch": 0.6618407445708376,
      "grad_norm": 0.7128654718399048,
      "learning_rate": 9.231037794307896e-06,
      "loss": 0.8661865592002869,
      "step": 510
    },
    {
      "epoch": 0.6631384715209765,
      "grad_norm": 0.6929757595062256,
      "learning_rate": 9.22722371025586e-06,
      "loss": 0.8198660612106323,
      "step": 511
    },
    {
      "epoch": 0.6644361984711155,
      "grad_norm": 0.7036609053611755,
      "learning_rate": 9.223400982601262e-06,
      "loss": 0.7628751993179321,
      "step": 512
    },
    {
      "epoch": 0.6657339254212543,
      "grad_norm": 0.7455263733863831,
      "learning_rate": 9.219569619160618e-06,
      "loss": 0.8409616351127625,
      "step": 513
    },
    {
      "epoch": 0.6670316523713933,
      "grad_norm": 0.7605084776878357,
      "learning_rate": 9.215729627768093e-06,
      "loss": 0.8993404507637024,
      "step": 514
    },
    {
      "epoch": 0.6683293793215321,
      "grad_norm": 0.7330998778343201,
      "learning_rate": 9.2118810162755e-06,
      "loss": 0.7985840439796448,
      "step": 515
    },
    {
      "epoch": 0.669627106271671,
      "grad_norm": 0.710595965385437,
      "learning_rate": 9.20802379255227e-06,
      "loss": 0.8102532625198364,
      "step": 516
    },
    {
      "epoch": 0.67092483322181,
      "grad_norm": 0.7394533753395081,
      "learning_rate": 9.204157964485454e-06,
      "loss": 0.8669203519821167,
      "step": 517
    },
    {
      "epoch": 0.6722225601719488,
      "grad_norm": 0.7361659407615662,
      "learning_rate": 9.200283539979691e-06,
      "loss": 0.8511008024215698,
      "step": 518
    },
    {
      "epoch": 0.6735202871220877,
      "grad_norm": 0.7361857891082764,
      "learning_rate": 9.196400526957198e-06,
      "loss": 0.8685998916625977,
      "step": 519
    },
    {
      "epoch": 0.6748180140722266,
      "grad_norm": 0.7673179507255554,
      "learning_rate": 9.192508933357753e-06,
      "loss": 0.8493822813034058,
      "step": 520
    },
    {
      "epoch": 0.6761157410223655,
      "grad_norm": 0.7651588320732117,
      "learning_rate": 9.188608767138683e-06,
      "loss": 0.895356297492981,
      "step": 521
    },
    {
      "epoch": 0.6774134679725045,
      "grad_norm": 0.7583249807357788,
      "learning_rate": 9.184700036274837e-06,
      "loss": 0.8651999831199646,
      "step": 522
    },
    {
      "epoch": 0.6787111949226433,
      "grad_norm": 0.7233076095581055,
      "learning_rate": 9.180782748758583e-06,
      "loss": 0.8715196251869202,
      "step": 523
    },
    {
      "epoch": 0.6800089218727822,
      "grad_norm": 0.7233548760414124,
      "learning_rate": 9.17685691259978e-06,
      "loss": 0.8916983008384705,
      "step": 524
    },
    {
      "epoch": 0.6813066488229211,
      "grad_norm": 0.7047330141067505,
      "learning_rate": 9.172922535825772e-06,
      "loss": 0.7544976472854614,
      "step": 525
    },
    {
      "epoch": 0.68260437577306,
      "grad_norm": 0.7306682467460632,
      "learning_rate": 9.168979626481364e-06,
      "loss": 0.8060271143913269,
      "step": 526
    },
    {
      "epoch": 0.6839021027231988,
      "grad_norm": 0.6905940175056458,
      "learning_rate": 9.165028192628803e-06,
      "loss": 0.8090964555740356,
      "step": 527
    },
    {
      "epoch": 0.6851998296733378,
      "grad_norm": 0.7123598456382751,
      "learning_rate": 9.161068242347777e-06,
      "loss": 0.7829641103744507,
      "step": 528
    },
    {
      "epoch": 0.6864975566234767,
      "grad_norm": 0.7222651243209839,
      "learning_rate": 9.157099783735378e-06,
      "loss": 0.7681039571762085,
      "step": 529
    },
    {
      "epoch": 0.6877952835736156,
      "grad_norm": 0.7189817428588867,
      "learning_rate": 9.1531228249061e-06,
      "loss": 0.8204732537269592,
      "step": 530
    },
    {
      "epoch": 0.6890930105237545,
      "grad_norm": 0.7915937900543213,
      "learning_rate": 9.149137373991819e-06,
      "loss": 0.9181404709815979,
      "step": 531
    },
    {
      "epoch": 0.6903907374738933,
      "grad_norm": 0.7157093286514282,
      "learning_rate": 9.145143439141771e-06,
      "loss": 0.8051410913467407,
      "step": 532
    },
    {
      "epoch": 0.6916884644240323,
      "grad_norm": 0.7645899057388306,
      "learning_rate": 9.141141028522544e-06,
      "loss": 0.8610852956771851,
      "step": 533
    },
    {
      "epoch": 0.6929861913741712,
      "grad_norm": 0.7532792687416077,
      "learning_rate": 9.137130150318055e-06,
      "loss": 0.777756929397583,
      "step": 534
    },
    {
      "epoch": 0.69428391832431,
      "grad_norm": 0.7266109585762024,
      "learning_rate": 9.133110812729532e-06,
      "loss": 0.8287126421928406,
      "step": 535
    },
    {
      "epoch": 0.695581645274449,
      "grad_norm": 0.7463825345039368,
      "learning_rate": 9.129083023975505e-06,
      "loss": 0.8141655921936035,
      "step": 536
    },
    {
      "epoch": 0.6968793722245878,
      "grad_norm": 0.7447496652603149,
      "learning_rate": 9.125046792291784e-06,
      "loss": 0.8310843110084534,
      "step": 537
    },
    {
      "epoch": 0.6981770991747268,
      "grad_norm": 0.72504723072052,
      "learning_rate": 9.121002125931436e-06,
      "loss": 0.7873820066452026,
      "step": 538
    },
    {
      "epoch": 0.6994748261248657,
      "grad_norm": 0.7359863519668579,
      "learning_rate": 9.116949033164785e-06,
      "loss": 0.8555896282196045,
      "step": 539
    },
    {
      "epoch": 0.7007725530750045,
      "grad_norm": 0.7197631001472473,
      "learning_rate": 9.112887522279378e-06,
      "loss": 0.8453904986381531,
      "step": 540
    },
    {
      "epoch": 0.7020702800251435,
      "grad_norm": 0.7554418444633484,
      "learning_rate": 9.108817601579978e-06,
      "loss": 0.8600095510482788,
      "step": 541
    },
    {
      "epoch": 0.7033680069752823,
      "grad_norm": 0.7778856754302979,
      "learning_rate": 9.104739279388542e-06,
      "loss": 0.8939447402954102,
      "step": 542
    },
    {
      "epoch": 0.7046657339254212,
      "grad_norm": 0.7399225831031799,
      "learning_rate": 9.100652564044206e-06,
      "loss": 0.831099808216095,
      "step": 543
    },
    {
      "epoch": 0.7059634608755602,
      "grad_norm": 0.7200485467910767,
      "learning_rate": 9.09655746390327e-06,
      "loss": 0.8666620254516602,
      "step": 544
    },
    {
      "epoch": 0.707261187825699,
      "grad_norm": 0.7057142853736877,
      "learning_rate": 9.092453987339174e-06,
      "loss": 0.8054381012916565,
      "step": 545
    },
    {
      "epoch": 0.708558914775838,
      "grad_norm": 0.7025026082992554,
      "learning_rate": 9.088342142742493e-06,
      "loss": 0.8074846863746643,
      "step": 546
    },
    {
      "epoch": 0.7098566417259768,
      "grad_norm": 0.7438026666641235,
      "learning_rate": 9.084221938520906e-06,
      "loss": 0.7710708379745483,
      "step": 547
    },
    {
      "epoch": 0.7111543686761157,
      "grad_norm": 0.7022576332092285,
      "learning_rate": 9.080093383099187e-06,
      "loss": 0.8335398435592651,
      "step": 548
    },
    {
      "epoch": 0.7124520956262547,
      "grad_norm": 0.7237393856048584,
      "learning_rate": 9.07595648491919e-06,
      "loss": 0.8371632695198059,
      "step": 549
    },
    {
      "epoch": 0.7137498225763935,
      "grad_norm": 0.7316430807113647,
      "learning_rate": 9.071811252439823e-06,
      "loss": 0.7962602972984314,
      "step": 550
    },
    {
      "epoch": 0.7150475495265324,
      "grad_norm": 0.679232656955719,
      "learning_rate": 9.067657694137038e-06,
      "loss": 0.8270066380500793,
      "step": 551
    },
    {
      "epoch": 0.7163452764766713,
      "grad_norm": 0.7692121267318726,
      "learning_rate": 9.063495818503809e-06,
      "loss": 0.8980318307876587,
      "step": 552
    },
    {
      "epoch": 0.7176430034268102,
      "grad_norm": 0.7516088485717773,
      "learning_rate": 9.059325634050118e-06,
      "loss": 0.8471689820289612,
      "step": 553
    },
    {
      "epoch": 0.7189407303769492,
      "grad_norm": 0.7334446907043457,
      "learning_rate": 9.05514714930294e-06,
      "loss": 0.8356298208236694,
      "step": 554
    },
    {
      "epoch": 0.720238457327088,
      "grad_norm": 0.6861634254455566,
      "learning_rate": 9.050960372806214e-06,
      "loss": 0.8037006258964539,
      "step": 555
    },
    {
      "epoch": 0.7215361842772269,
      "grad_norm": 0.7529132962226868,
      "learning_rate": 9.046765313120842e-06,
      "loss": 0.8884504437446594,
      "step": 556
    },
    {
      "epoch": 0.7228339112273658,
      "grad_norm": 0.7479069232940674,
      "learning_rate": 9.042561978824657e-06,
      "loss": 0.8784589767456055,
      "step": 557
    },
    {
      "epoch": 0.7241316381775047,
      "grad_norm": 0.7415169477462769,
      "learning_rate": 9.038350378512417e-06,
      "loss": 0.8879582285881042,
      "step": 558
    },
    {
      "epoch": 0.7254293651276437,
      "grad_norm": 0.7096434235572815,
      "learning_rate": 9.034130520795774e-06,
      "loss": 0.818169116973877,
      "step": 559
    },
    {
      "epoch": 0.7267270920777825,
      "grad_norm": 0.7252947688102722,
      "learning_rate": 9.029902414303273e-06,
      "loss": 0.8619629144668579,
      "step": 560
    },
    {
      "epoch": 0.7280248190279214,
      "grad_norm": 0.7012298703193665,
      "learning_rate": 9.025666067680319e-06,
      "loss": 0.7604071497917175,
      "step": 561
    },
    {
      "epoch": 0.7293225459780603,
      "grad_norm": 0.7263858914375305,
      "learning_rate": 9.021421489589169e-06,
      "loss": 0.8163655400276184,
      "step": 562
    },
    {
      "epoch": 0.7306202729281992,
      "grad_norm": 0.7177236676216125,
      "learning_rate": 9.017168688708913e-06,
      "loss": 0.7635133862495422,
      "step": 563
    },
    {
      "epoch": 0.731917999878338,
      "grad_norm": 0.7455753684043884,
      "learning_rate": 9.01290767373545e-06,
      "loss": 0.8407589197158813,
      "step": 564
    },
    {
      "epoch": 0.733215726828477,
      "grad_norm": 0.737720251083374,
      "learning_rate": 9.008638453381477e-06,
      "loss": 0.7209524512290955,
      "step": 565
    },
    {
      "epoch": 0.7345134537786159,
      "grad_norm": 0.7552280426025391,
      "learning_rate": 9.004361036376472e-06,
      "loss": 0.9184548854827881,
      "step": 566
    },
    {
      "epoch": 0.7358111807287548,
      "grad_norm": 0.7261860370635986,
      "learning_rate": 9.000075431466668e-06,
      "loss": 0.8386935591697693,
      "step": 567
    },
    {
      "epoch": 0.7371089076788937,
      "grad_norm": 0.7238007187843323,
      "learning_rate": 8.995781647415041e-06,
      "loss": 0.7940593957901001,
      "step": 568
    },
    {
      "epoch": 0.7384066346290326,
      "grad_norm": 0.7209489941596985,
      "learning_rate": 8.991479693001296e-06,
      "loss": 0.8524088859558105,
      "step": 569
    },
    {
      "epoch": 0.7397043615791715,
      "grad_norm": 0.7325084209442139,
      "learning_rate": 8.987169577021838e-06,
      "loss": 0.8363596200942993,
      "step": 570
    },
    {
      "epoch": 0.7410020885293104,
      "grad_norm": 0.7080730199813843,
      "learning_rate": 8.982851308289765e-06,
      "loss": 0.9011337757110596,
      "step": 571
    },
    {
      "epoch": 0.7422998154794492,
      "grad_norm": 0.7668154239654541,
      "learning_rate": 8.978524895634842e-06,
      "loss": 0.8450537323951721,
      "step": 572
    },
    {
      "epoch": 0.7435975424295882,
      "grad_norm": 0.7361171841621399,
      "learning_rate": 8.974190347903491e-06,
      "loss": 0.8203600645065308,
      "step": 573
    },
    {
      "epoch": 0.744895269379727,
      "grad_norm": 0.672437310218811,
      "learning_rate": 8.96984767395876e-06,
      "loss": 0.7632564306259155,
      "step": 574
    },
    {
      "epoch": 0.746192996329866,
      "grad_norm": 0.7395947575569153,
      "learning_rate": 8.965496882680322e-06,
      "loss": 0.8846946954727173,
      "step": 575
    },
    {
      "epoch": 0.7474907232800049,
      "grad_norm": 0.7199435234069824,
      "learning_rate": 8.961137982964445e-06,
      "loss": 0.7983006238937378,
      "step": 576
    },
    {
      "epoch": 0.7487884502301437,
      "grad_norm": 0.7480239272117615,
      "learning_rate": 8.95677098372397e-06,
      "loss": 0.8426603078842163,
      "step": 577
    },
    {
      "epoch": 0.7500861771802827,
      "grad_norm": 0.747464120388031,
      "learning_rate": 8.95239589388831e-06,
      "loss": 0.8379330635070801,
      "step": 578
    },
    {
      "epoch": 0.7513839041304216,
      "grad_norm": 0.7300353050231934,
      "learning_rate": 8.948012722403417e-06,
      "loss": 0.8795480132102966,
      "step": 579
    },
    {
      "epoch": 0.7526816310805604,
      "grad_norm": 0.790527880191803,
      "learning_rate": 8.943621478231764e-06,
      "loss": 0.8540128469467163,
      "step": 580
    },
    {
      "epoch": 0.7539793580306994,
      "grad_norm": 0.7547438144683838,
      "learning_rate": 8.939222170352333e-06,
      "loss": 0.8279592394828796,
      "step": 581
    },
    {
      "epoch": 0.7552770849808382,
      "grad_norm": 0.6995728015899658,
      "learning_rate": 8.9348148077606e-06,
      "loss": 0.7791587710380554,
      "step": 582
    },
    {
      "epoch": 0.7565748119309772,
      "grad_norm": 0.7065349817276001,
      "learning_rate": 8.9303993994685e-06,
      "loss": 0.8062193393707275,
      "step": 583
    },
    {
      "epoch": 0.757872538881116,
      "grad_norm": 0.7068513035774231,
      "learning_rate": 8.925975954504432e-06,
      "loss": 0.7748818397521973,
      "step": 584
    },
    {
      "epoch": 0.7591702658312549,
      "grad_norm": 0.7462431788444519,
      "learning_rate": 8.921544481913218e-06,
      "loss": 0.7763248085975647,
      "step": 585
    },
    {
      "epoch": 0.7604679927813939,
      "grad_norm": 0.7679768204689026,
      "learning_rate": 8.917104990756096e-06,
      "loss": 0.873698890209198,
      "step": 586
    },
    {
      "epoch": 0.7617657197315327,
      "grad_norm": 0.7324793338775635,
      "learning_rate": 8.912657490110705e-06,
      "loss": 0.8150361776351929,
      "step": 587
    },
    {
      "epoch": 0.7630634466816716,
      "grad_norm": 0.7165784239768982,
      "learning_rate": 8.908201989071055e-06,
      "loss": 0.8647459745407104,
      "step": 588
    },
    {
      "epoch": 0.7643611736318106,
      "grad_norm": 0.7230257391929626,
      "learning_rate": 8.903738496747523e-06,
      "loss": 0.8864815831184387,
      "step": 589
    },
    {
      "epoch": 0.7656589005819494,
      "grad_norm": 0.6890432238578796,
      "learning_rate": 8.899267022266815e-06,
      "loss": 0.8340771794319153,
      "step": 590
    },
    {
      "epoch": 0.7669566275320884,
      "grad_norm": 0.7173485159873962,
      "learning_rate": 8.894787574771968e-06,
      "loss": 0.8034731149673462,
      "step": 591
    },
    {
      "epoch": 0.7682543544822272,
      "grad_norm": 0.7507414817810059,
      "learning_rate": 8.890300163422319e-06,
      "loss": 0.8363530039787292,
      "step": 592
    },
    {
      "epoch": 0.7695520814323661,
      "grad_norm": 0.7558765411376953,
      "learning_rate": 8.885804797393484e-06,
      "loss": 0.8346325755119324,
      "step": 593
    },
    {
      "epoch": 0.770849808382505,
      "grad_norm": 0.767343282699585,
      "learning_rate": 8.881301485877355e-06,
      "loss": 0.8766120672225952,
      "step": 594
    },
    {
      "epoch": 0.7721475353326439,
      "grad_norm": 0.7072799801826477,
      "learning_rate": 8.87679023808206e-06,
      "loss": 0.7619128227233887,
      "step": 595
    },
    {
      "epoch": 0.7734452622827828,
      "grad_norm": 0.7522525191307068,
      "learning_rate": 8.87227106323196e-06,
      "loss": 0.8515980243682861,
      "step": 596
    },
    {
      "epoch": 0.7747429892329217,
      "grad_norm": 0.8023040890693665,
      "learning_rate": 8.867743970567625e-06,
      "loss": 0.7740094065666199,
      "step": 597
    },
    {
      "epoch": 0.7760407161830606,
      "grad_norm": 0.7332605719566345,
      "learning_rate": 8.86320896934581e-06,
      "loss": 0.892971396446228,
      "step": 598
    },
    {
      "epoch": 0.7773384431331996,
      "grad_norm": 0.725096583366394,
      "learning_rate": 8.858666068839447e-06,
      "loss": 0.7671830058097839,
      "step": 599
    },
    {
      "epoch": 0.7786361700833384,
      "grad_norm": 0.722201406955719,
      "learning_rate": 8.85411527833762e-06,
      "loss": 0.8197992444038391,
      "step": 600
    },
    {
      "epoch": 0.7799338970334773,
      "grad_norm": 0.682908833026886,
      "learning_rate": 8.849556607145541e-06,
      "loss": 0.7293855547904968,
      "step": 601
    },
    {
      "epoch": 0.7812316239836162,
      "grad_norm": 0.724932849407196,
      "learning_rate": 8.84499006458454e-06,
      "loss": 0.8873072266578674,
      "step": 602
    },
    {
      "epoch": 0.7825293509337551,
      "grad_norm": 0.6848104596138,
      "learning_rate": 8.840415659992038e-06,
      "loss": 0.8097598552703857,
      "step": 603
    },
    {
      "epoch": 0.7838270778838939,
      "grad_norm": 0.714169979095459,
      "learning_rate": 8.835833402721538e-06,
      "loss": 0.8175859451293945,
      "step": 604
    },
    {
      "epoch": 0.7851248048340329,
      "grad_norm": 0.7740371227264404,
      "learning_rate": 8.831243302142595e-06,
      "loss": 0.8492919206619263,
      "step": 605
    },
    {
      "epoch": 0.7864225317841718,
      "grad_norm": 0.7099517583847046,
      "learning_rate": 8.826645367640803e-06,
      "loss": 0.8585895895957947,
      "step": 606
    },
    {
      "epoch": 0.7877202587343107,
      "grad_norm": 0.7142400741577148,
      "learning_rate": 8.822039608617773e-06,
      "loss": 0.8533006906509399,
      "step": 607
    },
    {
      "epoch": 0.7890179856844496,
      "grad_norm": 0.7054215669631958,
      "learning_rate": 8.81742603449112e-06,
      "loss": 0.7479570508003235,
      "step": 608
    },
    {
      "epoch": 0.7903157126345884,
      "grad_norm": 0.7265973091125488,
      "learning_rate": 8.81280465469443e-06,
      "loss": 0.8953506946563721,
      "step": 609
    },
    {
      "epoch": 0.7916134395847274,
      "grad_norm": 0.7428868412971497,
      "learning_rate": 8.808175478677261e-06,
      "loss": 0.8039628267288208,
      "step": 610
    },
    {
      "epoch": 0.7929111665348663,
      "grad_norm": 0.7243048548698425,
      "learning_rate": 8.803538515905102e-06,
      "loss": 0.7982103228569031,
      "step": 611
    },
    {
      "epoch": 0.7942088934850051,
      "grad_norm": 0.7457985281944275,
      "learning_rate": 8.79889377585937e-06,
      "loss": 0.8665332198143005,
      "step": 612
    },
    {
      "epoch": 0.7955066204351441,
      "grad_norm": 0.776006817817688,
      "learning_rate": 8.79424126803738e-06,
      "loss": 0.8244978189468384,
      "step": 613
    },
    {
      "epoch": 0.7968043473852829,
      "grad_norm": 0.7452744245529175,
      "learning_rate": 8.789581001952339e-06,
      "loss": 0.8417704105377197,
      "step": 614
    },
    {
      "epoch": 0.7981020743354219,
      "grad_norm": 0.7379627823829651,
      "learning_rate": 8.784912987133305e-06,
      "loss": 0.8458194732666016,
      "step": 615
    },
    {
      "epoch": 0.7993998012855608,
      "grad_norm": 0.718325674533844,
      "learning_rate": 8.78023723312519e-06,
      "loss": 0.8211947679519653,
      "step": 616
    },
    {
      "epoch": 0.8006975282356996,
      "grad_norm": 0.6864851713180542,
      "learning_rate": 8.775553749488729e-06,
      "loss": 0.7400568127632141,
      "step": 617
    },
    {
      "epoch": 0.8019952551858386,
      "grad_norm": 0.7230079174041748,
      "learning_rate": 8.770862545800459e-06,
      "loss": 0.7876740097999573,
      "step": 618
    },
    {
      "epoch": 0.8032929821359774,
      "grad_norm": 0.7299041152000427,
      "learning_rate": 8.766163631652702e-06,
      "loss": 0.818249523639679,
      "step": 619
    },
    {
      "epoch": 0.8045907090861163,
      "grad_norm": 0.7503798604011536,
      "learning_rate": 8.76145701665355e-06,
      "loss": 0.8890657424926758,
      "step": 620
    },
    {
      "epoch": 0.8058884360362553,
      "grad_norm": 0.7524362802505493,
      "learning_rate": 8.756742710426842e-06,
      "loss": 0.7989673018455505,
      "step": 621
    },
    {
      "epoch": 0.8071861629863941,
      "grad_norm": 0.7557197213172913,
      "learning_rate": 8.752020722612135e-06,
      "loss": 0.819223940372467,
      "step": 622
    },
    {
      "epoch": 0.8084838899365331,
      "grad_norm": 0.7750368714332581,
      "learning_rate": 8.747291062864704e-06,
      "loss": 0.8548346757888794,
      "step": 623
    },
    {
      "epoch": 0.8097816168866719,
      "grad_norm": 0.7181532382965088,
      "learning_rate": 8.742553740855507e-06,
      "loss": 0.7740316390991211,
      "step": 624
    },
    {
      "epoch": 0.8110793438368108,
      "grad_norm": 0.744864821434021,
      "learning_rate": 8.737808766271163e-06,
      "loss": 0.795849621295929,
      "step": 625
    },
    {
      "epoch": 0.8123770707869498,
      "grad_norm": 0.7091443538665771,
      "learning_rate": 8.733056148813947e-06,
      "loss": 0.8226791024208069,
      "step": 626
    },
    {
      "epoch": 0.8136747977370886,
      "grad_norm": 0.7437657117843628,
      "learning_rate": 8.728295898201762e-06,
      "loss": 0.8405981063842773,
      "step": 627
    },
    {
      "epoch": 0.8149725246872275,
      "grad_norm": 0.7046335339546204,
      "learning_rate": 8.72352802416811e-06,
      "loss": 0.7682258486747742,
      "step": 628
    },
    {
      "epoch": 0.8162702516373664,
      "grad_norm": 0.7432284951210022,
      "learning_rate": 8.718752536462089e-06,
      "loss": 0.8703007102012634,
      "step": 629
    },
    {
      "epoch": 0.8175679785875053,
      "grad_norm": 0.7931978106498718,
      "learning_rate": 8.713969444848365e-06,
      "loss": 0.9198675155639648,
      "step": 630
    },
    {
      "epoch": 0.8188657055376443,
      "grad_norm": 0.7177122235298157,
      "learning_rate": 8.709178759107146e-06,
      "loss": 0.8506098389625549,
      "step": 631
    },
    {
      "epoch": 0.8201634324877831,
      "grad_norm": 0.7149267792701721,
      "learning_rate": 8.704380489034172e-06,
      "loss": 0.7816558480262756,
      "step": 632
    },
    {
      "epoch": 0.821461159437922,
      "grad_norm": 0.8047035932540894,
      "learning_rate": 8.699574644440696e-06,
      "loss": 0.8571796417236328,
      "step": 633
    },
    {
      "epoch": 0.8227588863880609,
      "grad_norm": 0.712013840675354,
      "learning_rate": 8.694761235153446e-06,
      "loss": 0.8653860092163086,
      "step": 634
    },
    {
      "epoch": 0.8240566133381998,
      "grad_norm": 0.6966065764427185,
      "learning_rate": 8.689940271014631e-06,
      "loss": 0.8233016133308411,
      "step": 635
    },
    {
      "epoch": 0.8253543402883388,
      "grad_norm": 0.731100857257843,
      "learning_rate": 8.685111761881902e-06,
      "loss": 0.8638073205947876,
      "step": 636
    },
    {
      "epoch": 0.8266520672384776,
      "grad_norm": 0.7355837821960449,
      "learning_rate": 8.680275717628336e-06,
      "loss": 0.8465107083320618,
      "step": 637
    },
    {
      "epoch": 0.8279497941886165,
      "grad_norm": 0.7684483528137207,
      "learning_rate": 8.675432148142423e-06,
      "loss": 0.8504188060760498,
      "step": 638
    },
    {
      "epoch": 0.8292475211387554,
      "grad_norm": 0.7174119353294373,
      "learning_rate": 8.670581063328031e-06,
      "loss": 0.7903969287872314,
      "step": 639
    },
    {
      "epoch": 0.8305452480888943,
      "grad_norm": 0.6824118494987488,
      "learning_rate": 8.665722473104407e-06,
      "loss": 0.79200679063797,
      "step": 640
    },
    {
      "epoch": 0.8318429750390332,
      "grad_norm": 0.7122912406921387,
      "learning_rate": 8.660856387406134e-06,
      "loss": 0.8188685774803162,
      "step": 641
    },
    {
      "epoch": 0.8331407019891721,
      "grad_norm": 0.7017625570297241,
      "learning_rate": 8.655982816183127e-06,
      "loss": 0.8316338062286377,
      "step": 642
    },
    {
      "epoch": 0.834438428939311,
      "grad_norm": 0.798739492893219,
      "learning_rate": 8.651101769400606e-06,
      "loss": 0.8281844854354858,
      "step": 643
    },
    {
      "epoch": 0.8357361558894499,
      "grad_norm": 0.7157971262931824,
      "learning_rate": 8.646213257039076e-06,
      "loss": 0.8818997740745544,
      "step": 644
    },
    {
      "epoch": 0.8370338828395888,
      "grad_norm": 0.7081462144851685,
      "learning_rate": 8.641317289094306e-06,
      "loss": 0.8395585417747498,
      "step": 645
    },
    {
      "epoch": 0.8383316097897276,
      "grad_norm": 0.752267062664032,
      "learning_rate": 8.636413875577314e-06,
      "loss": 0.852052628993988,
      "step": 646
    },
    {
      "epoch": 0.8396293367398666,
      "grad_norm": 0.7433504462242126,
      "learning_rate": 8.631503026514337e-06,
      "loss": 0.7837040424346924,
      "step": 647
    },
    {
      "epoch": 0.8409270636900055,
      "grad_norm": 0.7658749222755432,
      "learning_rate": 8.626584751946818e-06,
      "loss": 0.8357768654823303,
      "step": 648
    },
    {
      "epoch": 0.8422247906401443,
      "grad_norm": 0.7301930785179138,
      "learning_rate": 8.621659061931389e-06,
      "loss": 0.8085260391235352,
      "step": 649
    },
    {
      "epoch": 0.8435225175902833,
      "grad_norm": 0.694357693195343,
      "learning_rate": 8.616725966539831e-06,
      "loss": 0.7951889634132385,
      "step": 650
    },
    {
      "epoch": 0.8448202445404221,
      "grad_norm": 0.7343024015426636,
      "learning_rate": 8.611785475859083e-06,
      "loss": 0.8101716637611389,
      "step": 651
    },
    {
      "epoch": 0.8461179714905611,
      "grad_norm": 0.7503554224967957,
      "learning_rate": 8.606837599991194e-06,
      "loss": 0.8827271461486816,
      "step": 652
    },
    {
      "epoch": 0.8474156984407,
      "grad_norm": 0.7213442325592041,
      "learning_rate": 8.601882349053318e-06,
      "loss": 0.8409299254417419,
      "step": 653
    },
    {
      "epoch": 0.8487134253908388,
      "grad_norm": 0.7245302200317383,
      "learning_rate": 8.596919733177692e-06,
      "loss": 0.7459070682525635,
      "step": 654
    },
    {
      "epoch": 0.8500111523409778,
      "grad_norm": 0.7586632370948792,
      "learning_rate": 8.591949762511606e-06,
      "loss": 0.7964019775390625,
      "step": 655
    },
    {
      "epoch": 0.8513088792911166,
      "grad_norm": 0.7066137194633484,
      "learning_rate": 8.586972447217392e-06,
      "loss": 0.7685011625289917,
      "step": 656
    },
    {
      "epoch": 0.8526066062412555,
      "grad_norm": 0.7152141332626343,
      "learning_rate": 8.581987797472404e-06,
      "loss": 0.8681995868682861,
      "step": 657
    },
    {
      "epoch": 0.8539043331913945,
      "grad_norm": 0.732406735420227,
      "learning_rate": 8.576995823468984e-06,
      "loss": 0.8168764710426331,
      "step": 658
    },
    {
      "epoch": 0.8552020601415333,
      "grad_norm": 0.7218092083930969,
      "learning_rate": 8.571996535414457e-06,
      "loss": 0.7868984937667847,
      "step": 659
    },
    {
      "epoch": 0.8564997870916723,
      "grad_norm": 0.7385701537132263,
      "learning_rate": 8.566989943531106e-06,
      "loss": 0.8220299482345581,
      "step": 660
    },
    {
      "epoch": 0.8577975140418111,
      "grad_norm": 0.7344916462898254,
      "learning_rate": 8.561976058056138e-06,
      "loss": 0.8271669745445251,
      "step": 661
    },
    {
      "epoch": 0.85909524099195,
      "grad_norm": 0.7406896352767944,
      "learning_rate": 8.556954889241682e-06,
      "loss": 0.8336881399154663,
      "step": 662
    },
    {
      "epoch": 0.860392967942089,
      "grad_norm": 0.7561131715774536,
      "learning_rate": 8.551926447354759e-06,
      "loss": 0.7825358510017395,
      "step": 663
    },
    {
      "epoch": 0.8616906948922278,
      "grad_norm": 0.7565145492553711,
      "learning_rate": 8.546890742677259e-06,
      "loss": 0.7725045680999756,
      "step": 664
    },
    {
      "epoch": 0.8629884218423667,
      "grad_norm": 0.7259421348571777,
      "learning_rate": 8.541847785505921e-06,
      "loss": 0.8378758430480957,
      "step": 665
    },
    {
      "epoch": 0.8642861487925056,
      "grad_norm": 0.697087287902832,
      "learning_rate": 8.53679758615232e-06,
      "loss": 0.7955657839775085,
      "step": 666
    },
    {
      "epoch": 0.8655838757426445,
      "grad_norm": 0.7520134449005127,
      "learning_rate": 8.531740154942834e-06,
      "loss": 0.7764825224876404,
      "step": 667
    },
    {
      "epoch": 0.8668816026927835,
      "grad_norm": 0.7223501801490784,
      "learning_rate": 8.526675502218629e-06,
      "loss": 0.809272825717926,
      "step": 668
    },
    {
      "epoch": 0.8681793296429223,
      "grad_norm": 0.6946507692337036,
      "learning_rate": 8.521603638335638e-06,
      "loss": 0.8255141973495483,
      "step": 669
    },
    {
      "epoch": 0.8694770565930612,
      "grad_norm": 0.7150010466575623,
      "learning_rate": 8.516524573664539e-06,
      "loss": 0.845747709274292,
      "step": 670
    },
    {
      "epoch": 0.8707747835432001,
      "grad_norm": 0.7168200612068176,
      "learning_rate": 8.511438318590735e-06,
      "loss": 0.74006187915802,
      "step": 671
    },
    {
      "epoch": 0.872072510493339,
      "grad_norm": 0.7573256492614746,
      "learning_rate": 8.506344883514328e-06,
      "loss": 0.7870563268661499,
      "step": 672
    },
    {
      "epoch": 0.8733702374434779,
      "grad_norm": 0.7842072248458862,
      "learning_rate": 8.501244278850105e-06,
      "loss": 0.9044228196144104,
      "step": 673
    },
    {
      "epoch": 0.8746679643936168,
      "grad_norm": 0.7406667470932007,
      "learning_rate": 8.496136515027511e-06,
      "loss": 0.811069667339325,
      "step": 674
    },
    {
      "epoch": 0.8759656913437557,
      "grad_norm": 0.7430689334869385,
      "learning_rate": 8.491021602490632e-06,
      "loss": 0.8124207854270935,
      "step": 675
    },
    {
      "epoch": 0.8772634182938946,
      "grad_norm": 0.7479687929153442,
      "learning_rate": 8.485899551698166e-06,
      "loss": 0.7997281551361084,
      "step": 676
    },
    {
      "epoch": 0.8785611452440335,
      "grad_norm": 0.7981671690940857,
      "learning_rate": 8.480770373123415e-06,
      "loss": 0.8423473834991455,
      "step": 677
    },
    {
      "epoch": 0.8798588721941724,
      "grad_norm": 0.678821325302124,
      "learning_rate": 8.475634077254248e-06,
      "loss": 0.8249571323394775,
      "step": 678
    },
    {
      "epoch": 0.8811565991443113,
      "grad_norm": 0.725933313369751,
      "learning_rate": 8.470490674593091e-06,
      "loss": 0.860410213470459,
      "step": 679
    },
    {
      "epoch": 0.8824543260944502,
      "grad_norm": 0.7934810519218445,
      "learning_rate": 8.4653401756569e-06,
      "loss": 0.8338278532028198,
      "step": 680
    },
    {
      "epoch": 0.883752053044589,
      "grad_norm": 0.7282740473747253,
      "learning_rate": 8.460182590977142e-06,
      "loss": 0.8479672074317932,
      "step": 681
    },
    {
      "epoch": 0.885049779994728,
      "grad_norm": 0.789381206035614,
      "learning_rate": 8.455017931099772e-06,
      "loss": 0.8129116296768188,
      "step": 682
    },
    {
      "epoch": 0.8863475069448669,
      "grad_norm": 0.7255346179008484,
      "learning_rate": 8.449846206585211e-06,
      "loss": 0.821036696434021,
      "step": 683
    },
    {
      "epoch": 0.8876452338950058,
      "grad_norm": 0.7708171606063843,
      "learning_rate": 8.44466742800833e-06,
      "loss": 0.8153631687164307,
      "step": 684
    },
    {
      "epoch": 0.8889429608451447,
      "grad_norm": 0.7143636345863342,
      "learning_rate": 8.439481605958416e-06,
      "loss": 0.8277146816253662,
      "step": 685
    },
    {
      "epoch": 0.8902406877952835,
      "grad_norm": 0.8043286204338074,
      "learning_rate": 8.434288751039168e-06,
      "loss": 0.8356217741966248,
      "step": 686
    },
    {
      "epoch": 0.8915384147454225,
      "grad_norm": 0.7653588056564331,
      "learning_rate": 8.429088873868656e-06,
      "loss": 0.8771963715553284,
      "step": 687
    },
    {
      "epoch": 0.8928361416955614,
      "grad_norm": 0.6917808055877686,
      "learning_rate": 8.423881985079315e-06,
      "loss": 0.7479776740074158,
      "step": 688
    },
    {
      "epoch": 0.8941338686457002,
      "grad_norm": 0.7078230977058411,
      "learning_rate": 8.418668095317912e-06,
      "loss": 0.7685298323631287,
      "step": 689
    },
    {
      "epoch": 0.8954315955958392,
      "grad_norm": 0.7034929990768433,
      "learning_rate": 8.413447215245534e-06,
      "loss": 0.8008818626403809,
      "step": 690
    },
    {
      "epoch": 0.8954315955958392,
      "eval_loss": 0.7848750948905945,
      "eval_runtime": 143.9442,
      "eval_samples_per_second": 36.07,
      "eval_steps_per_second": 9.017,
      "step": 690
    },
    {
      "epoch": 0.896729322545978,
      "grad_norm": 0.7235103845596313,
      "learning_rate": 8.408219355537557e-06,
      "loss": 0.821499764919281,
      "step": 691
    },
    {
      "epoch": 0.898027049496117,
      "grad_norm": 0.7273632884025574,
      "learning_rate": 8.402984526883635e-06,
      "loss": 0.8295854330062866,
      "step": 692
    },
    {
      "epoch": 0.8993247764462559,
      "grad_norm": 0.7261776328086853,
      "learning_rate": 8.397742739987664e-06,
      "loss": 0.7892912030220032,
      "step": 693
    },
    {
      "epoch": 0.9006225033963947,
      "grad_norm": 0.7244385480880737,
      "learning_rate": 8.392494005567773e-06,
      "loss": 0.8177611231803894,
      "step": 694
    },
    {
      "epoch": 0.9019202303465337,
      "grad_norm": 0.7344761490821838,
      "learning_rate": 8.387238334356294e-06,
      "loss": 0.8023809194564819,
      "step": 695
    },
    {
      "epoch": 0.9032179572966725,
      "grad_norm": 0.736089825630188,
      "learning_rate": 8.381975737099745e-06,
      "loss": 0.8413804173469543,
      "step": 696
    },
    {
      "epoch": 0.9045156842468114,
      "grad_norm": 0.7272122502326965,
      "learning_rate": 8.376706224558807e-06,
      "loss": 0.8420959115028381,
      "step": 697
    },
    {
      "epoch": 0.9058134111969504,
      "grad_norm": 0.7177276015281677,
      "learning_rate": 8.3714298075083e-06,
      "loss": 0.8787964582443237,
      "step": 698
    },
    {
      "epoch": 0.9071111381470892,
      "grad_norm": 0.7416796088218689,
      "learning_rate": 8.366146496737158e-06,
      "loss": 0.8105108141899109,
      "step": 699
    },
    {
      "epoch": 0.9084088650972282,
      "grad_norm": 0.7222095727920532,
      "learning_rate": 8.360856303048417e-06,
      "loss": 0.7893627882003784,
      "step": 700
    },
    {
      "epoch": 0.909706592047367,
      "grad_norm": 0.7077576518058777,
      "learning_rate": 8.355559237259181e-06,
      "loss": 0.7560763955116272,
      "step": 701
    },
    {
      "epoch": 0.9110043189975059,
      "grad_norm": 0.6978065371513367,
      "learning_rate": 8.350255310200611e-06,
      "loss": 0.7838032841682434,
      "step": 702
    },
    {
      "epoch": 0.9123020459476449,
      "grad_norm": 0.7368733882904053,
      "learning_rate": 8.344944532717898e-06,
      "loss": 0.8452830910682678,
      "step": 703
    },
    {
      "epoch": 0.9135997728977837,
      "grad_norm": 0.7668716907501221,
      "learning_rate": 8.339626915670234e-06,
      "loss": 0.7398316264152527,
      "step": 704
    },
    {
      "epoch": 0.9148974998479226,
      "grad_norm": 0.7506059408187866,
      "learning_rate": 8.3343024699308e-06,
      "loss": 0.8178697824478149,
      "step": 705
    },
    {
      "epoch": 0.9161952267980615,
      "grad_norm": 0.771515965461731,
      "learning_rate": 8.328971206386742e-06,
      "loss": 0.8332667350769043,
      "step": 706
    },
    {
      "epoch": 0.9174929537482004,
      "grad_norm": 0.7568745017051697,
      "learning_rate": 8.323633135939145e-06,
      "loss": 0.8590246438980103,
      "step": 707
    },
    {
      "epoch": 0.9187906806983394,
      "grad_norm": 0.7116881608963013,
      "learning_rate": 8.318288269503007e-06,
      "loss": 0.8164006471633911,
      "step": 708
    },
    {
      "epoch": 0.9200884076484782,
      "grad_norm": 0.7115074396133423,
      "learning_rate": 8.312936618007232e-06,
      "loss": 0.827363133430481,
      "step": 709
    },
    {
      "epoch": 0.9213861345986171,
      "grad_norm": 0.7299380302429199,
      "learning_rate": 8.307578192394592e-06,
      "loss": 0.8012964129447937,
      "step": 710
    },
    {
      "epoch": 0.922683861548756,
      "grad_norm": 0.6973420977592468,
      "learning_rate": 8.30221300362171e-06,
      "loss": 0.7960250377655029,
      "step": 711
    },
    {
      "epoch": 0.9239815884988949,
      "grad_norm": 0.7379301190376282,
      "learning_rate": 8.29684106265904e-06,
      "loss": 0.8359223008155823,
      "step": 712
    },
    {
      "epoch": 0.9252793154490339,
      "grad_norm": 0.7026050090789795,
      "learning_rate": 8.291462380490842e-06,
      "loss": 0.8082996010780334,
      "step": 713
    },
    {
      "epoch": 0.9265770423991727,
      "grad_norm": 0.7560499906539917,
      "learning_rate": 8.286076968115158e-06,
      "loss": 0.7559288144111633,
      "step": 714
    },
    {
      "epoch": 0.9278747693493116,
      "grad_norm": 0.7483256459236145,
      "learning_rate": 8.280684836543794e-06,
      "loss": 0.8855387568473816,
      "step": 715
    },
    {
      "epoch": 0.9291724962994505,
      "grad_norm": 0.7110533714294434,
      "learning_rate": 8.275285996802293e-06,
      "loss": 0.8424345254898071,
      "step": 716
    },
    {
      "epoch": 0.9304702232495894,
      "grad_norm": 0.7900488376617432,
      "learning_rate": 8.269880459929919e-06,
      "loss": 0.8165364265441895,
      "step": 717
    },
    {
      "epoch": 0.9317679501997282,
      "grad_norm": 0.7415225505828857,
      "learning_rate": 8.264468236979626e-06,
      "loss": 0.8298836946487427,
      "step": 718
    },
    {
      "epoch": 0.9330656771498672,
      "grad_norm": 0.7365603446960449,
      "learning_rate": 8.259049339018036e-06,
      "loss": 0.855134904384613,
      "step": 719
    },
    {
      "epoch": 0.9343634041000061,
      "grad_norm": 0.755382239818573,
      "learning_rate": 8.25362377712543e-06,
      "loss": 0.8771618008613586,
      "step": 720
    },
    {
      "epoch": 0.935661131050145,
      "grad_norm": 0.7492240071296692,
      "learning_rate": 8.248191562395703e-06,
      "loss": 0.7913522720336914,
      "step": 721
    },
    {
      "epoch": 0.9369588580002839,
      "grad_norm": 0.726038932800293,
      "learning_rate": 8.242752705936363e-06,
      "loss": 0.8233412504196167,
      "step": 722
    },
    {
      "epoch": 0.9382565849504227,
      "grad_norm": 0.729804277420044,
      "learning_rate": 8.237307218868493e-06,
      "loss": 0.8285303115844727,
      "step": 723
    },
    {
      "epoch": 0.9395543119005617,
      "grad_norm": 0.7340808510780334,
      "learning_rate": 8.231855112326738e-06,
      "loss": 0.8087389469146729,
      "step": 724
    },
    {
      "epoch": 0.9408520388507006,
      "grad_norm": 0.6981987357139587,
      "learning_rate": 8.226396397459272e-06,
      "loss": 0.8206723928451538,
      "step": 725
    },
    {
      "epoch": 0.9421497658008394,
      "grad_norm": 0.7226195335388184,
      "learning_rate": 8.22093108542779e-06,
      "loss": 0.8090267181396484,
      "step": 726
    },
    {
      "epoch": 0.9434474927509784,
      "grad_norm": 0.7470179200172424,
      "learning_rate": 8.215459187407468e-06,
      "loss": 0.8644795417785645,
      "step": 727
    },
    {
      "epoch": 0.9447452197011172,
      "grad_norm": 0.7814143300056458,
      "learning_rate": 8.209980714586955e-06,
      "loss": 0.8789034485816956,
      "step": 728
    },
    {
      "epoch": 0.9460429466512562,
      "grad_norm": 0.7444256544113159,
      "learning_rate": 8.20449567816834e-06,
      "loss": 0.8365822434425354,
      "step": 729
    },
    {
      "epoch": 0.9473406736013951,
      "grad_norm": 0.7291041016578674,
      "learning_rate": 8.199004089367136e-06,
      "loss": 0.8873200416564941,
      "step": 730
    },
    {
      "epoch": 0.9486384005515339,
      "grad_norm": 0.6883243322372437,
      "learning_rate": 8.193505959412246e-06,
      "loss": 0.8070501685142517,
      "step": 731
    },
    {
      "epoch": 0.9499361275016729,
      "grad_norm": 0.6936081051826477,
      "learning_rate": 8.188001299545963e-06,
      "loss": 0.8187827467918396,
      "step": 732
    },
    {
      "epoch": 0.9512338544518117,
      "grad_norm": 0.7469680309295654,
      "learning_rate": 8.182490121023918e-06,
      "loss": 0.8218874335289001,
      "step": 733
    },
    {
      "epoch": 0.9525315814019506,
      "grad_norm": 0.7822017073631287,
      "learning_rate": 8.176972435115075e-06,
      "loss": 0.8489766716957092,
      "step": 734
    },
    {
      "epoch": 0.9538293083520896,
      "grad_norm": 0.7533932328224182,
      "learning_rate": 8.17144825310171e-06,
      "loss": 0.8591926693916321,
      "step": 735
    },
    {
      "epoch": 0.9551270353022284,
      "grad_norm": 0.6783921718597412,
      "learning_rate": 8.165917586279374e-06,
      "loss": 0.7739286422729492,
      "step": 736
    },
    {
      "epoch": 0.9564247622523674,
      "grad_norm": 0.7370641231536865,
      "learning_rate": 8.16038044595688e-06,
      "loss": 0.8205049633979797,
      "step": 737
    },
    {
      "epoch": 0.9577224892025062,
      "grad_norm": 0.723272442817688,
      "learning_rate": 8.15483684345628e-06,
      "loss": 0.8298395872116089,
      "step": 738
    },
    {
      "epoch": 0.9590202161526451,
      "grad_norm": 0.7543867230415344,
      "learning_rate": 8.149286790112838e-06,
      "loss": 0.784492552280426,
      "step": 739
    },
    {
      "epoch": 0.9603179431027841,
      "grad_norm": 0.750069797039032,
      "learning_rate": 8.143730297275008e-06,
      "loss": 0.7954182028770447,
      "step": 740
    },
    {
      "epoch": 0.9616156700529229,
      "grad_norm": 0.7066623568534851,
      "learning_rate": 8.138167376304411e-06,
      "loss": 0.8066163063049316,
      "step": 741
    },
    {
      "epoch": 0.9629133970030618,
      "grad_norm": 0.7153725028038025,
      "learning_rate": 8.132598038575814e-06,
      "loss": 0.7599750757217407,
      "step": 742
    },
    {
      "epoch": 0.9642111239532007,
      "grad_norm": 0.771787166595459,
      "learning_rate": 8.1270222954771e-06,
      "loss": 0.8843746781349182,
      "step": 743
    },
    {
      "epoch": 0.9655088509033396,
      "grad_norm": 0.7729872465133667,
      "learning_rate": 8.121440158409255e-06,
      "loss": 0.8271206021308899,
      "step": 744
    },
    {
      "epoch": 0.9668065778534786,
      "grad_norm": 0.7740857005119324,
      "learning_rate": 8.115851638786335e-06,
      "loss": 0.8692458868026733,
      "step": 745
    },
    {
      "epoch": 0.9681043048036174,
      "grad_norm": 0.7087443470954895,
      "learning_rate": 8.11025674803545e-06,
      "loss": 0.8108102083206177,
      "step": 746
    },
    {
      "epoch": 0.9694020317537563,
      "grad_norm": 0.7399817109107971,
      "learning_rate": 8.104655497596734e-06,
      "loss": 0.8514312505722046,
      "step": 747
    },
    {
      "epoch": 0.9706997587038952,
      "grad_norm": 0.749785304069519,
      "learning_rate": 8.099047898923326e-06,
      "loss": 0.7959103584289551,
      "step": 748
    },
    {
      "epoch": 0.9719974856540341,
      "grad_norm": 0.7259848117828369,
      "learning_rate": 8.093433963481348e-06,
      "loss": 0.8137398958206177,
      "step": 749
    },
    {
      "epoch": 0.973295212604173,
      "grad_norm": 0.7331299185752869,
      "learning_rate": 8.087813702749873e-06,
      "loss": 0.8132301568984985,
      "step": 750
    },
    {
      "epoch": 0.9745929395543119,
      "grad_norm": 0.7332584857940674,
      "learning_rate": 8.082187128220918e-06,
      "loss": 0.8191835880279541,
      "step": 751
    },
    {
      "epoch": 0.9758906665044508,
      "grad_norm": 0.6973693370819092,
      "learning_rate": 8.076554251399398e-06,
      "loss": 0.8266087174415588,
      "step": 752
    },
    {
      "epoch": 0.9771883934545897,
      "grad_norm": 0.7298718690872192,
      "learning_rate": 8.070915083803124e-06,
      "loss": 0.8183968663215637,
      "step": 753
    },
    {
      "epoch": 0.9784861204047286,
      "grad_norm": 0.7074362635612488,
      "learning_rate": 8.065269636962765e-06,
      "loss": 0.8740180730819702,
      "step": 754
    },
    {
      "epoch": 0.9797838473548675,
      "grad_norm": 0.7398948073387146,
      "learning_rate": 8.059617922421832e-06,
      "loss": 0.7935925722122192,
      "step": 755
    },
    {
      "epoch": 0.9810815743050064,
      "grad_norm": 0.709671676158905,
      "learning_rate": 8.053959951736647e-06,
      "loss": 0.8069880604743958,
      "step": 756
    },
    {
      "epoch": 0.9823793012551453,
      "grad_norm": 0.7370040416717529,
      "learning_rate": 8.048295736476332e-06,
      "loss": 0.8581700921058655,
      "step": 757
    },
    {
      "epoch": 0.9836770282052841,
      "grad_norm": 0.6889820098876953,
      "learning_rate": 8.042625288222774e-06,
      "loss": 0.7847036719322205,
      "step": 758
    },
    {
      "epoch": 0.9849747551554231,
      "grad_norm": 0.7171109914779663,
      "learning_rate": 8.036948618570601e-06,
      "loss": 0.7729103565216064,
      "step": 759
    },
    {
      "epoch": 0.986272482105562,
      "grad_norm": 0.7106866836547852,
      "learning_rate": 8.031265739127167e-06,
      "loss": 0.7267521023750305,
      "step": 760
    },
    {
      "epoch": 0.9875702090557009,
      "grad_norm": 0.7274038791656494,
      "learning_rate": 8.025576661512524e-06,
      "loss": 0.7723244428634644,
      "step": 761
    },
    {
      "epoch": 0.9888679360058398,
      "grad_norm": 0.6838043928146362,
      "learning_rate": 8.019881397359395e-06,
      "loss": 0.8244357109069824,
      "step": 762
    },
    {
      "epoch": 0.9901656629559786,
      "grad_norm": 0.7773102521896362,
      "learning_rate": 8.014179958313154e-06,
      "loss": 0.8146355152130127,
      "step": 763
    },
    {
      "epoch": 0.9914633899061176,
      "grad_norm": 0.6833189129829407,
      "learning_rate": 8.008472356031795e-06,
      "loss": 0.7225084900856018,
      "step": 764
    },
    {
      "epoch": 0.9927611168562565,
      "grad_norm": 0.7340080738067627,
      "learning_rate": 8.00275860218593e-06,
      "loss": 0.7879570126533508,
      "step": 765
    },
    {
      "epoch": 0.9940588438063953,
      "grad_norm": 0.744114100933075,
      "learning_rate": 7.99703870845873e-06,
      "loss": 0.8457169532775879,
      "step": 766
    },
    {
      "epoch": 0.9953565707565343,
      "grad_norm": 0.7166002988815308,
      "learning_rate": 7.991312686545939e-06,
      "loss": 0.786881685256958,
      "step": 767
    },
    {
      "epoch": 0.9966542977066731,
      "grad_norm": 0.7298569679260254,
      "learning_rate": 7.985580548155814e-06,
      "loss": 0.826994776725769,
      "step": 768
    },
    {
      "epoch": 0.9979520246568121,
      "grad_norm": 0.6964467167854309,
      "learning_rate": 7.979842305009133e-06,
      "loss": 0.7571007013320923,
      "step": 769
    },
    {
      "epoch": 0.999249751606951,
      "grad_norm": 0.7578926682472229,
      "learning_rate": 7.974097968839149e-06,
      "loss": 0.7799357771873474,
      "step": 770
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9857121706008911,
      "learning_rate": 7.968347551391574e-06,
      "loss": 0.7851473689079285,
      "step": 771
    },
    {
      "epoch": 1.0012977269501389,
      "grad_norm": 0.8775411248207092,
      "learning_rate": 7.962591064424558e-06,
      "loss": 0.7687423825263977,
      "step": 772
    },
    {
      "epoch": 1.0025954539002777,
      "grad_norm": 0.8086821436882019,
      "learning_rate": 7.95682851970866e-06,
      "loss": 0.7368807196617126,
      "step": 773
    },
    {
      "epoch": 1.0038931808504168,
      "grad_norm": 0.8149939775466919,
      "learning_rate": 7.951059929026826e-06,
      "loss": 0.7523797750473022,
      "step": 774
    },
    {
      "epoch": 1.0051909078005556,
      "grad_norm": 0.7600994110107422,
      "learning_rate": 7.94528530417436e-06,
      "loss": 0.7559916973114014,
      "step": 775
    },
    {
      "epoch": 1.0064886347506945,
      "grad_norm": 0.722993791103363,
      "learning_rate": 7.939504656958913e-06,
      "loss": 0.7532729506492615,
      "step": 776
    },
    {
      "epoch": 1.0077863617008334,
      "grad_norm": 0.7640414834022522,
      "learning_rate": 7.933717999200442e-06,
      "loss": 0.8589281439781189,
      "step": 777
    },
    {
      "epoch": 1.0090840886509722,
      "grad_norm": 0.7895022630691528,
      "learning_rate": 7.927925342731202e-06,
      "loss": 0.7896629571914673,
      "step": 778
    },
    {
      "epoch": 1.0103818156011113,
      "grad_norm": 0.8462852239608765,
      "learning_rate": 7.922126699395705e-06,
      "loss": 0.7882782816886902,
      "step": 779
    },
    {
      "epoch": 1.0116795425512501,
      "grad_norm": 0.7864810824394226,
      "learning_rate": 7.916322081050708e-06,
      "loss": 0.7486289143562317,
      "step": 780
    },
    {
      "epoch": 1.012977269501389,
      "grad_norm": 0.7887839674949646,
      "learning_rate": 7.910511499565192e-06,
      "loss": 0.6873072385787964,
      "step": 781
    },
    {
      "epoch": 1.0142749964515279,
      "grad_norm": 0.7757256627082825,
      "learning_rate": 7.90469496682032e-06,
      "loss": 0.7133891582489014,
      "step": 782
    },
    {
      "epoch": 1.0155727234016667,
      "grad_norm": 0.752653181552887,
      "learning_rate": 7.89887249470943e-06,
      "loss": 0.7424370050430298,
      "step": 783
    },
    {
      "epoch": 1.0168704503518056,
      "grad_norm": 0.7578316330909729,
      "learning_rate": 7.89304409513801e-06,
      "loss": 0.7230308055877686,
      "step": 784
    },
    {
      "epoch": 1.0181681773019446,
      "grad_norm": 0.763857901096344,
      "learning_rate": 7.887209780023652e-06,
      "loss": 0.7932077050209045,
      "step": 785
    },
    {
      "epoch": 1.0194659042520835,
      "grad_norm": 0.7570908069610596,
      "learning_rate": 7.881369561296061e-06,
      "loss": 0.8092459440231323,
      "step": 786
    },
    {
      "epoch": 1.0207636312022224,
      "grad_norm": 0.7562793493270874,
      "learning_rate": 7.875523450897004e-06,
      "loss": 0.7545458078384399,
      "step": 787
    },
    {
      "epoch": 1.0220613581523612,
      "grad_norm": 0.7494779825210571,
      "learning_rate": 7.869671460780297e-06,
      "loss": 0.7644586563110352,
      "step": 788
    },
    {
      "epoch": 1.0233590851025,
      "grad_norm": 0.7528582215309143,
      "learning_rate": 7.863813602911777e-06,
      "loss": 0.7303503751754761,
      "step": 789
    },
    {
      "epoch": 1.0246568120526391,
      "grad_norm": 0.7522894740104675,
      "learning_rate": 7.857949889269285e-06,
      "loss": 0.769363284111023,
      "step": 790
    },
    {
      "epoch": 1.025954539002778,
      "grad_norm": 0.7932534217834473,
      "learning_rate": 7.852080331842627e-06,
      "loss": 0.6980154514312744,
      "step": 791
    },
    {
      "epoch": 1.0272522659529169,
      "grad_norm": 0.7427641153335571,
      "learning_rate": 7.846204942633564e-06,
      "loss": 0.880423903465271,
      "step": 792
    },
    {
      "epoch": 1.0285499929030557,
      "grad_norm": 0.8040611743927002,
      "learning_rate": 7.84032373365578e-06,
      "loss": 0.7813893556594849,
      "step": 793
    },
    {
      "epoch": 1.0298477198531946,
      "grad_norm": 0.7074670791625977,
      "learning_rate": 7.834436716934859e-06,
      "loss": 0.6680439114570618,
      "step": 794
    },
    {
      "epoch": 1.0311454468033336,
      "grad_norm": 0.7425694465637207,
      "learning_rate": 7.828543904508258e-06,
      "loss": 0.7353805303573608,
      "step": 795
    },
    {
      "epoch": 1.0324431737534725,
      "grad_norm": 0.7460526823997498,
      "learning_rate": 7.82264530842529e-06,
      "loss": 0.7601878046989441,
      "step": 796
    },
    {
      "epoch": 1.0337409007036114,
      "grad_norm": 0.7570812702178955,
      "learning_rate": 7.816740940747089e-06,
      "loss": 0.7122874855995178,
      "step": 797
    },
    {
      "epoch": 1.0350386276537502,
      "grad_norm": 0.7967089414596558,
      "learning_rate": 7.810830813546594e-06,
      "loss": 0.7236475348472595,
      "step": 798
    },
    {
      "epoch": 1.036336354603889,
      "grad_norm": 0.7531136274337769,
      "learning_rate": 7.80491493890852e-06,
      "loss": 0.7801082730293274,
      "step": 799
    },
    {
      "epoch": 1.037634081554028,
      "grad_norm": 0.7200342416763306,
      "learning_rate": 7.798993328929328e-06,
      "loss": 0.7519978284835815,
      "step": 800
    },
    {
      "epoch": 1.038931808504167,
      "grad_norm": 0.7226322293281555,
      "learning_rate": 7.793065995717217e-06,
      "loss": 0.7664476633071899,
      "step": 801
    },
    {
      "epoch": 1.0402295354543059,
      "grad_norm": 0.7494815587997437,
      "learning_rate": 7.787132951392082e-06,
      "loss": 0.7169873714447021,
      "step": 802
    },
    {
      "epoch": 1.0415272624044447,
      "grad_norm": 0.7437771558761597,
      "learning_rate": 7.781194208085495e-06,
      "loss": 0.777241587638855,
      "step": 803
    },
    {
      "epoch": 1.0428249893545836,
      "grad_norm": 0.7736064195632935,
      "learning_rate": 7.775249777940685e-06,
      "loss": 0.7803325653076172,
      "step": 804
    },
    {
      "epoch": 1.0441227163047224,
      "grad_norm": 0.729405403137207,
      "learning_rate": 7.769299673112507e-06,
      "loss": 0.7208072543144226,
      "step": 805
    },
    {
      "epoch": 1.0454204432548615,
      "grad_norm": 0.7493116855621338,
      "learning_rate": 7.76334390576742e-06,
      "loss": 0.7370114326477051,
      "step": 806
    },
    {
      "epoch": 1.0467181702050004,
      "grad_norm": 0.7444543242454529,
      "learning_rate": 7.757382488083458e-06,
      "loss": 0.8371725678443909,
      "step": 807
    },
    {
      "epoch": 1.0480158971551392,
      "grad_norm": 0.7414549589157104,
      "learning_rate": 7.751415432250213e-06,
      "loss": 0.7497185468673706,
      "step": 808
    },
    {
      "epoch": 1.049313624105278,
      "grad_norm": 0.7469486594200134,
      "learning_rate": 7.745442750468803e-06,
      "loss": 0.7713848948478699,
      "step": 809
    },
    {
      "epoch": 1.050611351055417,
      "grad_norm": 0.7274091243743896,
      "learning_rate": 7.739464454951853e-06,
      "loss": 0.7730478048324585,
      "step": 810
    },
    {
      "epoch": 1.051909078005556,
      "grad_norm": 0.7138769626617432,
      "learning_rate": 7.733480557923464e-06,
      "loss": 0.6896777153015137,
      "step": 811
    },
    {
      "epoch": 1.0532068049556949,
      "grad_norm": 0.773325502872467,
      "learning_rate": 7.727491071619186e-06,
      "loss": 0.7608213424682617,
      "step": 812
    },
    {
      "epoch": 1.0545045319058337,
      "grad_norm": 0.7017092108726501,
      "learning_rate": 7.72149600828601e-06,
      "loss": 0.7796779870986938,
      "step": 813
    },
    {
      "epoch": 1.0558022588559726,
      "grad_norm": 0.7887934446334839,
      "learning_rate": 7.715495380182314e-06,
      "loss": 0.7536224722862244,
      "step": 814
    },
    {
      "epoch": 1.0570999858061114,
      "grad_norm": 0.7297992706298828,
      "learning_rate": 7.709489199577874e-06,
      "loss": 0.704574465751648,
      "step": 815
    },
    {
      "epoch": 1.0583977127562503,
      "grad_norm": 0.7437949180603027,
      "learning_rate": 7.7034774787538e-06,
      "loss": 0.741614818572998,
      "step": 816
    },
    {
      "epoch": 1.0596954397063894,
      "grad_norm": 0.7397854328155518,
      "learning_rate": 7.697460230002545e-06,
      "loss": 0.7390281558036804,
      "step": 817
    },
    {
      "epoch": 1.0609931666565282,
      "grad_norm": 0.7297989130020142,
      "learning_rate": 7.691437465627859e-06,
      "loss": 0.7850978970527649,
      "step": 818
    },
    {
      "epoch": 1.062290893606667,
      "grad_norm": 0.7123188376426697,
      "learning_rate": 7.685409197944768e-06,
      "loss": 0.7182658314704895,
      "step": 819
    },
    {
      "epoch": 1.063588620556806,
      "grad_norm": 0.755650520324707,
      "learning_rate": 7.679375439279557e-06,
      "loss": 0.7544305324554443,
      "step": 820
    },
    {
      "epoch": 1.0648863475069448,
      "grad_norm": 0.7906817197799683,
      "learning_rate": 7.673336201969733e-06,
      "loss": 0.8021898865699768,
      "step": 821
    },
    {
      "epoch": 1.0661840744570839,
      "grad_norm": 0.779934823513031,
      "learning_rate": 7.667291498364009e-06,
      "loss": 0.8222522139549255,
      "step": 822
    },
    {
      "epoch": 1.0674818014072227,
      "grad_norm": 0.7806147933006287,
      "learning_rate": 7.661241340822274e-06,
      "loss": 0.824374794960022,
      "step": 823
    },
    {
      "epoch": 1.0687795283573616,
      "grad_norm": 0.7627273797988892,
      "learning_rate": 7.655185741715569e-06,
      "loss": 0.7465693354606628,
      "step": 824
    },
    {
      "epoch": 1.0700772553075004,
      "grad_norm": 0.7400906682014465,
      "learning_rate": 7.64912471342606e-06,
      "loss": 0.7903148531913757,
      "step": 825
    },
    {
      "epoch": 1.0713749822576393,
      "grad_norm": 0.7551797032356262,
      "learning_rate": 7.643058268347015e-06,
      "loss": 0.7619248628616333,
      "step": 826
    },
    {
      "epoch": 1.0726727092077784,
      "grad_norm": 0.7665975689888,
      "learning_rate": 7.636986418882783e-06,
      "loss": 0.7794485092163086,
      "step": 827
    },
    {
      "epoch": 1.0739704361579172,
      "grad_norm": 0.712073028087616,
      "learning_rate": 7.630909177448755e-06,
      "loss": 0.7433626055717468,
      "step": 828
    },
    {
      "epoch": 1.075268163108056,
      "grad_norm": 0.7561034560203552,
      "learning_rate": 7.624826556471354e-06,
      "loss": 0.776486337184906,
      "step": 829
    },
    {
      "epoch": 1.076565890058195,
      "grad_norm": 0.7552739381790161,
      "learning_rate": 7.618738568388e-06,
      "loss": 0.7432718276977539,
      "step": 830
    },
    {
      "epoch": 1.0778636170083338,
      "grad_norm": 0.7300425171852112,
      "learning_rate": 7.612645225647086e-06,
      "loss": 0.6905062198638916,
      "step": 831
    },
    {
      "epoch": 1.0791613439584729,
      "grad_norm": 0.777538537979126,
      "learning_rate": 7.60654654070796e-06,
      "loss": 0.7303627133369446,
      "step": 832
    },
    {
      "epoch": 1.0804590709086117,
      "grad_norm": 0.7533742189407349,
      "learning_rate": 7.600442526040883e-06,
      "loss": 0.7780784964561462,
      "step": 833
    },
    {
      "epoch": 1.0817567978587506,
      "grad_norm": 0.7974081635475159,
      "learning_rate": 7.594333194127025e-06,
      "loss": 0.8020890355110168,
      "step": 834
    },
    {
      "epoch": 1.0830545248088894,
      "grad_norm": 0.7591211199760437,
      "learning_rate": 7.58821855745842e-06,
      "loss": 0.7216978073120117,
      "step": 835
    },
    {
      "epoch": 1.0843522517590283,
      "grad_norm": 0.7331092357635498,
      "learning_rate": 7.582098628537955e-06,
      "loss": 0.7495509386062622,
      "step": 836
    },
    {
      "epoch": 1.0856499787091671,
      "grad_norm": 0.7309387922286987,
      "learning_rate": 7.5759734198793365e-06,
      "loss": 0.7153578400611877,
      "step": 837
    },
    {
      "epoch": 1.0869477056593062,
      "grad_norm": 0.7279072403907776,
      "learning_rate": 7.5698429440070616e-06,
      "loss": 0.8049778938293457,
      "step": 838
    },
    {
      "epoch": 1.088245432609445,
      "grad_norm": 0.8327065706253052,
      "learning_rate": 7.563707213456405e-06,
      "loss": 0.8458483815193176,
      "step": 839
    },
    {
      "epoch": 1.089543159559584,
      "grad_norm": 0.7936592102050781,
      "learning_rate": 7.5575662407733815e-06,
      "loss": 0.7528290152549744,
      "step": 840
    },
    {
      "epoch": 1.0908408865097228,
      "grad_norm": 0.7855809330940247,
      "learning_rate": 7.551420038514726e-06,
      "loss": 0.685576319694519,
      "step": 841
    },
    {
      "epoch": 1.0921386134598616,
      "grad_norm": 0.7452473640441895,
      "learning_rate": 7.54526861924787e-06,
      "loss": 0.8216978311538696,
      "step": 842
    },
    {
      "epoch": 1.0934363404100007,
      "grad_norm": 0.7420814037322998,
      "learning_rate": 7.5391119955509026e-06,
      "loss": 0.725795328617096,
      "step": 843
    },
    {
      "epoch": 1.0947340673601396,
      "grad_norm": 0.780059278011322,
      "learning_rate": 7.532950180012564e-06,
      "loss": 0.7719560861587524,
      "step": 844
    },
    {
      "epoch": 1.0960317943102784,
      "grad_norm": 0.7614301443099976,
      "learning_rate": 7.526783185232208e-06,
      "loss": 0.7809950709342957,
      "step": 845
    },
    {
      "epoch": 1.0973295212604173,
      "grad_norm": 0.7960677742958069,
      "learning_rate": 7.520611023819779e-06,
      "loss": 0.7617411017417908,
      "step": 846
    },
    {
      "epoch": 1.0986272482105561,
      "grad_norm": 0.7018498778343201,
      "learning_rate": 7.514433708395783e-06,
      "loss": 0.7251595854759216,
      "step": 847
    },
    {
      "epoch": 1.099924975160695,
      "grad_norm": 0.7345300912857056,
      "learning_rate": 7.508251251591266e-06,
      "loss": 0.7545365691184998,
      "step": 848
    },
    {
      "epoch": 1.101222702110834,
      "grad_norm": 0.7398869395256042,
      "learning_rate": 7.5020636660477894e-06,
      "loss": 0.7734966278076172,
      "step": 849
    },
    {
      "epoch": 1.102520429060973,
      "grad_norm": 0.7710892558097839,
      "learning_rate": 7.4958709644174e-06,
      "loss": 0.805327296257019,
      "step": 850
    },
    {
      "epoch": 1.1038181560111118,
      "grad_norm": 0.7368630766868591,
      "learning_rate": 7.4896731593626015e-06,
      "loss": 0.7682947516441345,
      "step": 851
    },
    {
      "epoch": 1.1051158829612506,
      "grad_norm": 0.7361379265785217,
      "learning_rate": 7.4834702635563395e-06,
      "loss": 0.7895755767822266,
      "step": 852
    },
    {
      "epoch": 1.1064136099113895,
      "grad_norm": 0.7309215664863586,
      "learning_rate": 7.477262289681966e-06,
      "loss": 0.8211460113525391,
      "step": 853
    },
    {
      "epoch": 1.1077113368615286,
      "grad_norm": 0.7528488636016846,
      "learning_rate": 7.471049250433214e-06,
      "loss": 0.8033767938613892,
      "step": 854
    },
    {
      "epoch": 1.1090090638116674,
      "grad_norm": 0.772190511226654,
      "learning_rate": 7.464831158514179e-06,
      "loss": 0.7697012424468994,
      "step": 855
    },
    {
      "epoch": 1.1103067907618063,
      "grad_norm": 0.7335084080696106,
      "learning_rate": 7.458608026639285e-06,
      "loss": 0.7386154532432556,
      "step": 856
    },
    {
      "epoch": 1.1116045177119451,
      "grad_norm": 0.8107844591140747,
      "learning_rate": 7.45237986753326e-06,
      "loss": 0.8037740588188171,
      "step": 857
    },
    {
      "epoch": 1.112902244662084,
      "grad_norm": 0.7389320731163025,
      "learning_rate": 7.446146693931111e-06,
      "loss": 0.7920179963111877,
      "step": 858
    },
    {
      "epoch": 1.114199971612223,
      "grad_norm": 0.7466602325439453,
      "learning_rate": 7.439908518578105e-06,
      "loss": 0.7859473824501038,
      "step": 859
    },
    {
      "epoch": 1.115497698562362,
      "grad_norm": 0.752849817276001,
      "learning_rate": 7.433665354229731e-06,
      "loss": 0.7838953733444214,
      "step": 860
    },
    {
      "epoch": 1.1167954255125008,
      "grad_norm": 0.749245285987854,
      "learning_rate": 7.4274172136516766e-06,
      "loss": 0.772464394569397,
      "step": 861
    },
    {
      "epoch": 1.1180931524626396,
      "grad_norm": 0.7855739593505859,
      "learning_rate": 7.421164109619809e-06,
      "loss": 0.7547976970672607,
      "step": 862
    },
    {
      "epoch": 1.1193908794127785,
      "grad_norm": 0.7009910345077515,
      "learning_rate": 7.4149060549201455e-06,
      "loss": 0.7241532802581787,
      "step": 863
    },
    {
      "epoch": 1.1206886063629176,
      "grad_norm": 0.7256485223770142,
      "learning_rate": 7.408643062348824e-06,
      "loss": 0.716011106967926,
      "step": 864
    },
    {
      "epoch": 1.1219863333130564,
      "grad_norm": 0.7414820194244385,
      "learning_rate": 7.402375144712075e-06,
      "loss": 0.8162094950675964,
      "step": 865
    },
    {
      "epoch": 1.1232840602631953,
      "grad_norm": 0.7918742299079895,
      "learning_rate": 7.396102314826207e-06,
      "loss": 0.741178035736084,
      "step": 866
    },
    {
      "epoch": 1.1245817872133341,
      "grad_norm": 0.7321358323097229,
      "learning_rate": 7.389824585517569e-06,
      "loss": 0.7776750922203064,
      "step": 867
    },
    {
      "epoch": 1.125879514163473,
      "grad_norm": 0.7683652639389038,
      "learning_rate": 7.3835419696225275e-06,
      "loss": 0.8042243123054504,
      "step": 868
    },
    {
      "epoch": 1.1271772411136118,
      "grad_norm": 0.7723100185394287,
      "learning_rate": 7.377254479987445e-06,
      "loss": 0.779492199420929,
      "step": 869
    },
    {
      "epoch": 1.128474968063751,
      "grad_norm": 0.7506013512611389,
      "learning_rate": 7.370962129468642e-06,
      "loss": 0.7360951900482178,
      "step": 870
    },
    {
      "epoch": 1.1297726950138898,
      "grad_norm": 0.7726255655288696,
      "learning_rate": 7.364664930932385e-06,
      "loss": 0.8193867206573486,
      "step": 871
    },
    {
      "epoch": 1.1310704219640286,
      "grad_norm": 0.7944323420524597,
      "learning_rate": 7.35836289725485e-06,
      "loss": 0.7985553741455078,
      "step": 872
    },
    {
      "epoch": 1.1323681489141675,
      "grad_norm": 0.7620981931686401,
      "learning_rate": 7.352056041322103e-06,
      "loss": 0.7627980709075928,
      "step": 873
    },
    {
      "epoch": 1.1336658758643063,
      "grad_norm": 0.7579022645950317,
      "learning_rate": 7.345744376030066e-06,
      "loss": 0.7844129204750061,
      "step": 874
    },
    {
      "epoch": 1.1349636028144454,
      "grad_norm": 0.7362563014030457,
      "learning_rate": 7.339427914284498e-06,
      "loss": 0.7017800807952881,
      "step": 875
    },
    {
      "epoch": 1.1362613297645843,
      "grad_norm": 0.7596364617347717,
      "learning_rate": 7.3331066690009644e-06,
      "loss": 0.7282767295837402,
      "step": 876
    },
    {
      "epoch": 1.1375590567147231,
      "grad_norm": 0.7417566776275635,
      "learning_rate": 7.326780653104813e-06,
      "loss": 0.7459028363227844,
      "step": 877
    },
    {
      "epoch": 1.138856783664862,
      "grad_norm": 0.7634292840957642,
      "learning_rate": 7.320449879531143e-06,
      "loss": 0.8008609414100647,
      "step": 878
    },
    {
      "epoch": 1.1401545106150008,
      "grad_norm": 0.7351272106170654,
      "learning_rate": 7.314114361224785e-06,
      "loss": 0.7598878145217896,
      "step": 879
    },
    {
      "epoch": 1.1414522375651397,
      "grad_norm": 0.7726730108261108,
      "learning_rate": 7.30777411114027e-06,
      "loss": 0.7667620182037354,
      "step": 880
    },
    {
      "epoch": 1.1427499645152788,
      "grad_norm": 0.7361600995063782,
      "learning_rate": 7.301429142241805e-06,
      "loss": 0.7034899592399597,
      "step": 881
    },
    {
      "epoch": 1.1440476914654176,
      "grad_norm": 0.7282233834266663,
      "learning_rate": 7.295079467503247e-06,
      "loss": 0.6745363473892212,
      "step": 882
    },
    {
      "epoch": 1.1453454184155565,
      "grad_norm": 0.7537771463394165,
      "learning_rate": 7.288725099908071e-06,
      "loss": 0.7910201549530029,
      "step": 883
    },
    {
      "epoch": 1.1466431453656953,
      "grad_norm": 0.7625670433044434,
      "learning_rate": 7.282366052449351e-06,
      "loss": 0.8292834162712097,
      "step": 884
    },
    {
      "epoch": 1.1479408723158344,
      "grad_norm": 0.7329244613647461,
      "learning_rate": 7.276002338129731e-06,
      "loss": 0.7522168755531311,
      "step": 885
    },
    {
      "epoch": 1.1492385992659733,
      "grad_norm": 0.7891886234283447,
      "learning_rate": 7.269633969961395e-06,
      "loss": 0.8136672973632812,
      "step": 886
    },
    {
      "epoch": 1.1505363262161121,
      "grad_norm": 0.7690410017967224,
      "learning_rate": 7.2632609609660456e-06,
      "loss": 0.8092800974845886,
      "step": 887
    },
    {
      "epoch": 1.151834053166251,
      "grad_norm": 0.7545477151870728,
      "learning_rate": 7.256883324174871e-06,
      "loss": 0.7119185924530029,
      "step": 888
    },
    {
      "epoch": 1.1531317801163898,
      "grad_norm": 0.7471065521240234,
      "learning_rate": 7.250501072628524e-06,
      "loss": 0.7687282562255859,
      "step": 889
    },
    {
      "epoch": 1.1544295070665287,
      "grad_norm": 0.7187462449073792,
      "learning_rate": 7.2441142193770955e-06,
      "loss": 0.7572574019432068,
      "step": 890
    },
    {
      "epoch": 1.1557272340166678,
      "grad_norm": 0.745625376701355,
      "learning_rate": 7.237722777480083e-06,
      "loss": 0.8049079775810242,
      "step": 891
    },
    {
      "epoch": 1.1570249609668066,
      "grad_norm": 0.7539253234863281,
      "learning_rate": 7.231326760006368e-06,
      "loss": 0.8245639801025391,
      "step": 892
    },
    {
      "epoch": 1.1583226879169455,
      "grad_norm": 0.8297720551490784,
      "learning_rate": 7.224926180034186e-06,
      "loss": 0.8184964656829834,
      "step": 893
    },
    {
      "epoch": 1.1596204148670843,
      "grad_norm": 0.757693350315094,
      "learning_rate": 7.218521050651106e-06,
      "loss": 0.7818700671195984,
      "step": 894
    },
    {
      "epoch": 1.1609181418172232,
      "grad_norm": 0.723356306552887,
      "learning_rate": 7.212111384953993e-06,
      "loss": 0.7243102192878723,
      "step": 895
    },
    {
      "epoch": 1.1622158687673623,
      "grad_norm": 0.7468777298927307,
      "learning_rate": 7.205697196048992e-06,
      "loss": 0.7427289485931396,
      "step": 896
    },
    {
      "epoch": 1.1635135957175011,
      "grad_norm": 0.7640864253044128,
      "learning_rate": 7.199278497051498e-06,
      "loss": 0.7910446524620056,
      "step": 897
    },
    {
      "epoch": 1.16481132266764,
      "grad_norm": 0.7713980078697205,
      "learning_rate": 7.192855301086123e-06,
      "loss": 0.7965364456176758,
      "step": 898
    },
    {
      "epoch": 1.1661090496177788,
      "grad_norm": 0.776570200920105,
      "learning_rate": 7.186427621286678e-06,
      "loss": 0.753921627998352,
      "step": 899
    },
    {
      "epoch": 1.1674067765679177,
      "grad_norm": 0.7340149283409119,
      "learning_rate": 7.179995470796141e-06,
      "loss": 0.7870222330093384,
      "step": 900
    },
    {
      "epoch": 1.1687045035180565,
      "grad_norm": 0.7386748194694519,
      "learning_rate": 7.1735588627666346e-06,
      "loss": 0.7083443999290466,
      "step": 901
    },
    {
      "epoch": 1.1700022304681956,
      "grad_norm": 0.7222329378128052,
      "learning_rate": 7.167117810359387e-06,
      "loss": 0.6995740532875061,
      "step": 902
    },
    {
      "epoch": 1.1712999574183345,
      "grad_norm": 0.7578097581863403,
      "learning_rate": 7.160672326744726e-06,
      "loss": 0.7429664134979248,
      "step": 903
    },
    {
      "epoch": 1.1725976843684733,
      "grad_norm": 0.7582994699478149,
      "learning_rate": 7.154222425102033e-06,
      "loss": 0.7429041266441345,
      "step": 904
    },
    {
      "epoch": 1.1738954113186122,
      "grad_norm": 0.7732597589492798,
      "learning_rate": 7.1477681186197225e-06,
      "loss": 0.7760801911354065,
      "step": 905
    },
    {
      "epoch": 1.175193138268751,
      "grad_norm": 0.7655629515647888,
      "learning_rate": 7.141309420495219e-06,
      "loss": 0.7732428908348083,
      "step": 906
    },
    {
      "epoch": 1.1764908652188901,
      "grad_norm": 0.7582475543022156,
      "learning_rate": 7.134846343934924e-06,
      "loss": 0.8130234479904175,
      "step": 907
    },
    {
      "epoch": 1.177788592169029,
      "grad_norm": 0.7538527846336365,
      "learning_rate": 7.128378902154195e-06,
      "loss": 0.8220568895339966,
      "step": 908
    },
    {
      "epoch": 1.1790863191191678,
      "grad_norm": 0.7548943161964417,
      "learning_rate": 7.121907108377313e-06,
      "loss": 0.7736442685127258,
      "step": 909
    },
    {
      "epoch": 1.1803840460693067,
      "grad_norm": 0.7636051177978516,
      "learning_rate": 7.115430975837457e-06,
      "loss": 0.8169882297515869,
      "step": 910
    },
    {
      "epoch": 1.1816817730194455,
      "grad_norm": 0.767085075378418,
      "learning_rate": 7.10895051777668e-06,
      "loss": 0.7491641640663147,
      "step": 911
    },
    {
      "epoch": 1.1829794999695844,
      "grad_norm": 0.7884292602539062,
      "learning_rate": 7.1024657474458795e-06,
      "loss": 0.7490637302398682,
      "step": 912
    },
    {
      "epoch": 1.1842772269197235,
      "grad_norm": 0.7684943675994873,
      "learning_rate": 7.095976678104768e-06,
      "loss": 0.7587805390357971,
      "step": 913
    },
    {
      "epoch": 1.1855749538698623,
      "grad_norm": 0.7661841511726379,
      "learning_rate": 7.089483323021851e-06,
      "loss": 0.7323252558708191,
      "step": 914
    },
    {
      "epoch": 1.1868726808200012,
      "grad_norm": 0.7417365908622742,
      "learning_rate": 7.082985695474394e-06,
      "loss": 0.8243879675865173,
      "step": 915
    },
    {
      "epoch": 1.18817040777014,
      "grad_norm": 0.7636500000953674,
      "learning_rate": 7.076483808748402e-06,
      "loss": 0.7433345317840576,
      "step": 916
    },
    {
      "epoch": 1.1894681347202791,
      "grad_norm": 0.7614058256149292,
      "learning_rate": 7.069977676138588e-06,
      "loss": 0.7225663661956787,
      "step": 917
    },
    {
      "epoch": 1.190765861670418,
      "grad_norm": 0.70651775598526,
      "learning_rate": 7.063467310948346e-06,
      "loss": 0.7022136449813843,
      "step": 918
    },
    {
      "epoch": 1.1920635886205568,
      "grad_norm": 0.7883380055427551,
      "learning_rate": 7.0569527264897275e-06,
      "loss": 0.7680527567863464,
      "step": 919
    },
    {
      "epoch": 1.1933613155706957,
      "grad_norm": 0.7263570427894592,
      "learning_rate": 7.050433936083405e-06,
      "loss": 0.7278581857681274,
      "step": 920
    },
    {
      "epoch": 1.1933613155706957,
      "eval_loss": 0.776515781879425,
      "eval_runtime": 138.714,
      "eval_samples_per_second": 37.43,
      "eval_steps_per_second": 9.357,
      "step": 920
    },
    {
      "epoch": 1.1946590425208345,
      "grad_norm": 0.7014424800872803,
      "learning_rate": 7.043910953058657e-06,
      "loss": 0.7067049145698547,
      "step": 921
    },
    {
      "epoch": 1.1959567694709734,
      "grad_norm": 0.7746631503105164,
      "learning_rate": 7.037383790753333e-06,
      "loss": 0.7801349759101868,
      "step": 922
    },
    {
      "epoch": 1.1972544964211125,
      "grad_norm": 0.7254592180252075,
      "learning_rate": 7.030852462513827e-06,
      "loss": 0.7349493503570557,
      "step": 923
    },
    {
      "epoch": 1.1985522233712513,
      "grad_norm": 0.7621070742607117,
      "learning_rate": 7.024316981695053e-06,
      "loss": 0.719948410987854,
      "step": 924
    },
    {
      "epoch": 1.1998499503213902,
      "grad_norm": 0.7466797232627869,
      "learning_rate": 7.017777361660414e-06,
      "loss": 0.7515002489089966,
      "step": 925
    },
    {
      "epoch": 1.201147677271529,
      "grad_norm": 0.709510326385498,
      "learning_rate": 7.011233615781777e-06,
      "loss": 0.7294214367866516,
      "step": 926
    },
    {
      "epoch": 1.202445404221668,
      "grad_norm": 0.7353148460388184,
      "learning_rate": 7.004685757439449e-06,
      "loss": 0.8314340710639954,
      "step": 927
    },
    {
      "epoch": 1.203743131171807,
      "grad_norm": 0.7547840476036072,
      "learning_rate": 6.99813380002214e-06,
      "loss": 0.772587776184082,
      "step": 928
    },
    {
      "epoch": 1.2050408581219458,
      "grad_norm": 0.7102051973342896,
      "learning_rate": 6.991577756926948e-06,
      "loss": 0.8234133720397949,
      "step": 929
    },
    {
      "epoch": 1.2063385850720847,
      "grad_norm": 0.7403558492660522,
      "learning_rate": 6.9850176415593195e-06,
      "loss": 0.7458879947662354,
      "step": 930
    },
    {
      "epoch": 1.2076363120222235,
      "grad_norm": 0.7559409141540527,
      "learning_rate": 6.978453467333028e-06,
      "loss": 0.7612881660461426,
      "step": 931
    },
    {
      "epoch": 1.2089340389723624,
      "grad_norm": 0.7423329949378967,
      "learning_rate": 6.9718852476701535e-06,
      "loss": 0.8142660856246948,
      "step": 932
    },
    {
      "epoch": 1.2102317659225013,
      "grad_norm": 0.7359952330589294,
      "learning_rate": 6.965312996001038e-06,
      "loss": 0.7446624040603638,
      "step": 933
    },
    {
      "epoch": 1.2115294928726403,
      "grad_norm": 0.7642225027084351,
      "learning_rate": 6.958736725764275e-06,
      "loss": 0.7768579721450806,
      "step": 934
    },
    {
      "epoch": 1.2128272198227792,
      "grad_norm": 0.7549358606338501,
      "learning_rate": 6.952156450406673e-06,
      "loss": 0.6970754265785217,
      "step": 935
    },
    {
      "epoch": 1.214124946772918,
      "grad_norm": 0.7197482585906982,
      "learning_rate": 6.945572183383229e-06,
      "loss": 0.7270360589027405,
      "step": 936
    },
    {
      "epoch": 1.215422673723057,
      "grad_norm": 0.7698655724525452,
      "learning_rate": 6.9389839381571025e-06,
      "loss": 0.7863064408302307,
      "step": 937
    },
    {
      "epoch": 1.216720400673196,
      "grad_norm": 0.7359870076179504,
      "learning_rate": 6.932391728199587e-06,
      "loss": 0.7461609840393066,
      "step": 938
    },
    {
      "epoch": 1.2180181276233348,
      "grad_norm": 0.7642099857330322,
      "learning_rate": 6.925795566990083e-06,
      "loss": 0.7753203511238098,
      "step": 939
    },
    {
      "epoch": 1.2193158545734737,
      "grad_norm": 0.7594850659370422,
      "learning_rate": 6.919195468016073e-06,
      "loss": 0.720144510269165,
      "step": 940
    },
    {
      "epoch": 1.2206135815236125,
      "grad_norm": 0.7252835631370544,
      "learning_rate": 6.9125914447730865e-06,
      "loss": 0.7099956274032593,
      "step": 941
    },
    {
      "epoch": 1.2219113084737514,
      "grad_norm": 0.7514061331748962,
      "learning_rate": 6.905983510764681e-06,
      "loss": 0.7722299695014954,
      "step": 942
    },
    {
      "epoch": 1.2232090354238903,
      "grad_norm": 0.7274535894393921,
      "learning_rate": 6.899371679502408e-06,
      "loss": 0.7559299468994141,
      "step": 943
    },
    {
      "epoch": 1.2245067623740293,
      "grad_norm": 0.7720993757247925,
      "learning_rate": 6.89275596450579e-06,
      "loss": 0.6993024945259094,
      "step": 944
    },
    {
      "epoch": 1.2258044893241682,
      "grad_norm": 0.7331786155700684,
      "learning_rate": 6.886136379302288e-06,
      "loss": 0.7450038194656372,
      "step": 945
    },
    {
      "epoch": 1.227102216274307,
      "grad_norm": 0.750044047832489,
      "learning_rate": 6.87951293742728e-06,
      "loss": 0.7362930178642273,
      "step": 946
    },
    {
      "epoch": 1.228399943224446,
      "grad_norm": 0.7513757944107056,
      "learning_rate": 6.872885652424028e-06,
      "loss": 0.7076199054718018,
      "step": 947
    },
    {
      "epoch": 1.2296976701745848,
      "grad_norm": 0.7729170322418213,
      "learning_rate": 6.866254537843651e-06,
      "loss": 0.7380319237709045,
      "step": 948
    },
    {
      "epoch": 1.2309953971247238,
      "grad_norm": 0.8171294331550598,
      "learning_rate": 6.859619607245102e-06,
      "loss": 0.772607684135437,
      "step": 949
    },
    {
      "epoch": 1.2322931240748627,
      "grad_norm": 0.8113188147544861,
      "learning_rate": 6.852980874195132e-06,
      "loss": 0.7299817204475403,
      "step": 950
    },
    {
      "epoch": 1.2335908510250015,
      "grad_norm": 0.8075060844421387,
      "learning_rate": 6.846338352268273e-06,
      "loss": 0.8221969604492188,
      "step": 951
    },
    {
      "epoch": 1.2348885779751404,
      "grad_norm": 0.7619355916976929,
      "learning_rate": 6.839692055046801e-06,
      "loss": 0.7481426000595093,
      "step": 952
    },
    {
      "epoch": 1.2361863049252793,
      "grad_norm": 0.7421068549156189,
      "learning_rate": 6.833041996120707e-06,
      "loss": 0.7882356643676758,
      "step": 953
    },
    {
      "epoch": 1.2374840318754181,
      "grad_norm": 0.7260448932647705,
      "learning_rate": 6.826388189087683e-06,
      "loss": 0.8039664030075073,
      "step": 954
    },
    {
      "epoch": 1.2387817588255572,
      "grad_norm": 0.7982683777809143,
      "learning_rate": 6.819730647553079e-06,
      "loss": 0.7353008389472961,
      "step": 955
    },
    {
      "epoch": 1.240079485775696,
      "grad_norm": 0.7499701976776123,
      "learning_rate": 6.813069385129883e-06,
      "loss": 0.6967143416404724,
      "step": 956
    },
    {
      "epoch": 1.241377212725835,
      "grad_norm": 0.7564923167228699,
      "learning_rate": 6.806404415438689e-06,
      "loss": 0.730963945388794,
      "step": 957
    },
    {
      "epoch": 1.2426749396759738,
      "grad_norm": 0.7305687665939331,
      "learning_rate": 6.7997357521076735e-06,
      "loss": 0.7766451239585876,
      "step": 958
    },
    {
      "epoch": 1.2439726666261126,
      "grad_norm": 0.7542693018913269,
      "learning_rate": 6.793063408772565e-06,
      "loss": 0.7498062252998352,
      "step": 959
    },
    {
      "epoch": 1.2452703935762517,
      "grad_norm": 0.8000041842460632,
      "learning_rate": 6.78638739907662e-06,
      "loss": 0.7798739075660706,
      "step": 960
    },
    {
      "epoch": 1.2465681205263905,
      "grad_norm": 0.7799873948097229,
      "learning_rate": 6.779707736670585e-06,
      "loss": 0.723656415939331,
      "step": 961
    },
    {
      "epoch": 1.2478658474765294,
      "grad_norm": 0.7346177697181702,
      "learning_rate": 6.773024435212678e-06,
      "loss": 0.7384864687919617,
      "step": 962
    },
    {
      "epoch": 1.2491635744266683,
      "grad_norm": 0.7333899736404419,
      "learning_rate": 6.7663375083685635e-06,
      "loss": 0.7858769297599792,
      "step": 963
    },
    {
      "epoch": 1.250461301376807,
      "grad_norm": 0.7262735366821289,
      "learning_rate": 6.759646969811311e-06,
      "loss": 0.8483904004096985,
      "step": 964
    },
    {
      "epoch": 1.251759028326946,
      "grad_norm": 0.7812559008598328,
      "learning_rate": 6.752952833221379e-06,
      "loss": 0.8175226449966431,
      "step": 965
    },
    {
      "epoch": 1.253056755277085,
      "grad_norm": 0.795184314250946,
      "learning_rate": 6.7462551122865825e-06,
      "loss": 0.7282869815826416,
      "step": 966
    },
    {
      "epoch": 1.254354482227224,
      "grad_norm": 0.7277102470397949,
      "learning_rate": 6.739553820702067e-06,
      "loss": 0.7237769365310669,
      "step": 967
    },
    {
      "epoch": 1.2556522091773628,
      "grad_norm": 0.7736328840255737,
      "learning_rate": 6.732848972170276e-06,
      "loss": 0.6851608753204346,
      "step": 968
    },
    {
      "epoch": 1.2569499361275016,
      "grad_norm": 0.7597711682319641,
      "learning_rate": 6.726140580400928e-06,
      "loss": 0.7550280094146729,
      "step": 969
    },
    {
      "epoch": 1.2582476630776407,
      "grad_norm": 0.7755154371261597,
      "learning_rate": 6.719428659110987e-06,
      "loss": 0.7753418684005737,
      "step": 970
    },
    {
      "epoch": 1.2595453900277795,
      "grad_norm": 0.7772382497787476,
      "learning_rate": 6.712713222024633e-06,
      "loss": 0.7511424422264099,
      "step": 971
    },
    {
      "epoch": 1.2608431169779184,
      "grad_norm": 0.8128913044929504,
      "learning_rate": 6.705994282873233e-06,
      "loss": 0.8187769651412964,
      "step": 972
    },
    {
      "epoch": 1.2621408439280573,
      "grad_norm": 0.737055778503418,
      "learning_rate": 6.699271855395321e-06,
      "loss": 0.757542610168457,
      "step": 973
    },
    {
      "epoch": 1.263438570878196,
      "grad_norm": 0.785873293876648,
      "learning_rate": 6.6925459533365576e-06,
      "loss": 0.7776392102241516,
      "step": 974
    },
    {
      "epoch": 1.264736297828335,
      "grad_norm": 0.7526916265487671,
      "learning_rate": 6.685816590449708e-06,
      "loss": 0.8357568979263306,
      "step": 975
    },
    {
      "epoch": 1.2660340247784738,
      "grad_norm": 0.7440223693847656,
      "learning_rate": 6.67908378049462e-06,
      "loss": 0.7778916954994202,
      "step": 976
    },
    {
      "epoch": 1.267331751728613,
      "grad_norm": 0.8247221112251282,
      "learning_rate": 6.672347537238183e-06,
      "loss": 0.8135752081871033,
      "step": 977
    },
    {
      "epoch": 1.2686294786787518,
      "grad_norm": 0.8035563826560974,
      "learning_rate": 6.665607874454311e-06,
      "loss": 0.7972402572631836,
      "step": 978
    },
    {
      "epoch": 1.2699272056288906,
      "grad_norm": 0.7410714626312256,
      "learning_rate": 6.658864805923909e-06,
      "loss": 0.7736347913742065,
      "step": 979
    },
    {
      "epoch": 1.2712249325790295,
      "grad_norm": 0.8330723643302917,
      "learning_rate": 6.652118345434844e-06,
      "loss": 0.806992769241333,
      "step": 980
    },
    {
      "epoch": 1.2725226595291685,
      "grad_norm": 0.77059406042099,
      "learning_rate": 6.64536850678192e-06,
      "loss": 0.7687591314315796,
      "step": 981
    },
    {
      "epoch": 1.2738203864793074,
      "grad_norm": 0.7119758725166321,
      "learning_rate": 6.638615303766849e-06,
      "loss": 0.7087782621383667,
      "step": 982
    },
    {
      "epoch": 1.2751181134294463,
      "grad_norm": 0.7512334585189819,
      "learning_rate": 6.631858750198223e-06,
      "loss": 0.7549843192100525,
      "step": 983
    },
    {
      "epoch": 1.276415840379585,
      "grad_norm": 0.7844512462615967,
      "learning_rate": 6.625098859891483e-06,
      "loss": 0.7538207769393921,
      "step": 984
    },
    {
      "epoch": 1.277713567329724,
      "grad_norm": 0.8203350901603699,
      "learning_rate": 6.618335646668894e-06,
      "loss": 0.7637907862663269,
      "step": 985
    },
    {
      "epoch": 1.2790112942798628,
      "grad_norm": 0.7754982113838196,
      "learning_rate": 6.611569124359516e-06,
      "loss": 0.7811301350593567,
      "step": 986
    },
    {
      "epoch": 1.280309021230002,
      "grad_norm": 0.7798315286636353,
      "learning_rate": 6.604799306799172e-06,
      "loss": 0.7387815713882446,
      "step": 987
    },
    {
      "epoch": 1.2816067481801408,
      "grad_norm": 0.7446553111076355,
      "learning_rate": 6.598026207830428e-06,
      "loss": 0.7689957022666931,
      "step": 988
    },
    {
      "epoch": 1.2829044751302796,
      "grad_norm": 0.74884033203125,
      "learning_rate": 6.591249841302555e-06,
      "loss": 0.8296911120414734,
      "step": 989
    },
    {
      "epoch": 1.2842022020804185,
      "grad_norm": 0.7766373753547668,
      "learning_rate": 6.58447022107151e-06,
      "loss": 0.7640050053596497,
      "step": 990
    },
    {
      "epoch": 1.2854999290305575,
      "grad_norm": 0.7423699498176575,
      "learning_rate": 6.577687360999898e-06,
      "loss": 0.7463095188140869,
      "step": 991
    },
    {
      "epoch": 1.2867976559806964,
      "grad_norm": 0.7285832166671753,
      "learning_rate": 6.5709012749569535e-06,
      "loss": 0.7764305472373962,
      "step": 992
    },
    {
      "epoch": 1.2880953829308353,
      "grad_norm": 0.7299691438674927,
      "learning_rate": 6.564111976818501e-06,
      "loss": 0.7112918496131897,
      "step": 993
    },
    {
      "epoch": 1.289393109880974,
      "grad_norm": 0.7747628688812256,
      "learning_rate": 6.5573194804669416e-06,
      "loss": 0.8535696864128113,
      "step": 994
    },
    {
      "epoch": 1.290690836831113,
      "grad_norm": 0.7332835793495178,
      "learning_rate": 6.550523799791207e-06,
      "loss": 0.7880507707595825,
      "step": 995
    },
    {
      "epoch": 1.2919885637812518,
      "grad_norm": 0.7431394457817078,
      "learning_rate": 6.543724948686747e-06,
      "loss": 0.7584883570671082,
      "step": 996
    },
    {
      "epoch": 1.2932862907313907,
      "grad_norm": 0.7418090105056763,
      "learning_rate": 6.53692294105549e-06,
      "loss": 0.7503098249435425,
      "step": 997
    },
    {
      "epoch": 1.2945840176815298,
      "grad_norm": 0.7521533370018005,
      "learning_rate": 6.53011779080582e-06,
      "loss": 0.7553617358207703,
      "step": 998
    },
    {
      "epoch": 1.2958817446316686,
      "grad_norm": 0.7553898692131042,
      "learning_rate": 6.523309511852547e-06,
      "loss": 0.8658368587493896,
      "step": 999
    },
    {
      "epoch": 1.2971794715818075,
      "grad_norm": 0.7538703083992004,
      "learning_rate": 6.516498118116878e-06,
      "loss": 0.8231520056724548,
      "step": 1000
    },
    {
      "epoch": 1.2984771985319463,
      "grad_norm": 0.7610046863555908,
      "learning_rate": 6.5096836235263904e-06,
      "loss": 0.7878521084785461,
      "step": 1001
    },
    {
      "epoch": 1.2997749254820854,
      "grad_norm": 0.7674424648284912,
      "learning_rate": 6.502866042015e-06,
      "loss": 0.736729085445404,
      "step": 1002
    },
    {
      "epoch": 1.3010726524322243,
      "grad_norm": 0.7649716138839722,
      "learning_rate": 6.496045387522934e-06,
      "loss": 0.7443773746490479,
      "step": 1003
    },
    {
      "epoch": 1.302370379382363,
      "grad_norm": 0.7352249026298523,
      "learning_rate": 6.489221673996708e-06,
      "loss": 0.6777861714363098,
      "step": 1004
    },
    {
      "epoch": 1.303668106332502,
      "grad_norm": 0.7463338971138,
      "learning_rate": 6.482394915389085e-06,
      "loss": 0.7501753568649292,
      "step": 1005
    },
    {
      "epoch": 1.3049658332826408,
      "grad_norm": 0.7425395846366882,
      "learning_rate": 6.475565125659063e-06,
      "loss": 0.7762592434883118,
      "step": 1006
    },
    {
      "epoch": 1.3062635602327797,
      "grad_norm": 0.7330830097198486,
      "learning_rate": 6.4687323187718276e-06,
      "loss": 0.7636951208114624,
      "step": 1007
    },
    {
      "epoch": 1.3075612871829188,
      "grad_norm": 0.7125248908996582,
      "learning_rate": 6.461896508698744e-06,
      "loss": 0.6942975521087646,
      "step": 1008
    },
    {
      "epoch": 1.3088590141330576,
      "grad_norm": 0.7361205220222473,
      "learning_rate": 6.455057709417312e-06,
      "loss": 0.6826867461204529,
      "step": 1009
    },
    {
      "epoch": 1.3101567410831965,
      "grad_norm": 0.7715798020362854,
      "learning_rate": 6.448215934911145e-06,
      "loss": 0.8465424180030823,
      "step": 1010
    },
    {
      "epoch": 1.3114544680333353,
      "grad_norm": 0.754898190498352,
      "learning_rate": 6.441371199169942e-06,
      "loss": 0.7275772094726562,
      "step": 1011
    },
    {
      "epoch": 1.3127521949834744,
      "grad_norm": 0.7745576500892639,
      "learning_rate": 6.434523516189453e-06,
      "loss": 0.7738680243492126,
      "step": 1012
    },
    {
      "epoch": 1.3140499219336133,
      "grad_norm": 0.7302242517471313,
      "learning_rate": 6.427672899971457e-06,
      "loss": 0.7262543439865112,
      "step": 1013
    },
    {
      "epoch": 1.315347648883752,
      "grad_norm": 0.7484472990036011,
      "learning_rate": 6.4208193645237314e-06,
      "loss": 0.7359537482261658,
      "step": 1014
    },
    {
      "epoch": 1.316645375833891,
      "grad_norm": 0.7181548476219177,
      "learning_rate": 6.413962923860021e-06,
      "loss": 0.7458126544952393,
      "step": 1015
    },
    {
      "epoch": 1.3179431027840298,
      "grad_norm": 0.7399983406066895,
      "learning_rate": 6.407103592000009e-06,
      "loss": 0.713172197341919,
      "step": 1016
    },
    {
      "epoch": 1.3192408297341687,
      "grad_norm": 0.7715758681297302,
      "learning_rate": 6.400241382969297e-06,
      "loss": 0.8125825524330139,
      "step": 1017
    },
    {
      "epoch": 1.3205385566843075,
      "grad_norm": 0.7169333100318909,
      "learning_rate": 6.393376310799363e-06,
      "loss": 0.7719976305961609,
      "step": 1018
    },
    {
      "epoch": 1.3218362836344466,
      "grad_norm": 0.7504178285598755,
      "learning_rate": 6.386508389527544e-06,
      "loss": 0.7362254858016968,
      "step": 1019
    },
    {
      "epoch": 1.3231340105845855,
      "grad_norm": 0.7360901832580566,
      "learning_rate": 6.379637633196999e-06,
      "loss": 0.7390870451927185,
      "step": 1020
    },
    {
      "epoch": 1.3244317375347243,
      "grad_norm": 0.7370476722717285,
      "learning_rate": 6.3727640558566865e-06,
      "loss": 0.7379764318466187,
      "step": 1021
    },
    {
      "epoch": 1.3257294644848632,
      "grad_norm": 0.7515619397163391,
      "learning_rate": 6.3658876715613315e-06,
      "loss": 0.7954624891281128,
      "step": 1022
    },
    {
      "epoch": 1.3270271914350023,
      "grad_norm": 0.7417750954627991,
      "learning_rate": 6.3590084943713995e-06,
      "loss": 0.7774450182914734,
      "step": 1023
    },
    {
      "epoch": 1.328324918385141,
      "grad_norm": 0.7214071750640869,
      "learning_rate": 6.35212653835307e-06,
      "loss": 0.7376827001571655,
      "step": 1024
    },
    {
      "epoch": 1.32962264533528,
      "grad_norm": 0.7392934560775757,
      "learning_rate": 6.345241817578196e-06,
      "loss": 0.7806702852249146,
      "step": 1025
    },
    {
      "epoch": 1.3309203722854188,
      "grad_norm": 0.7314793467521667,
      "learning_rate": 6.3383543461242914e-06,
      "loss": 0.7208546996116638,
      "step": 1026
    },
    {
      "epoch": 1.3322180992355577,
      "grad_norm": 0.7446559071540833,
      "learning_rate": 6.331464138074493e-06,
      "loss": 0.7563031315803528,
      "step": 1027
    },
    {
      "epoch": 1.3335158261856965,
      "grad_norm": 0.7644572854042053,
      "learning_rate": 6.32457120751753e-06,
      "loss": 0.7995797991752625,
      "step": 1028
    },
    {
      "epoch": 1.3348135531358354,
      "grad_norm": 0.7222753167152405,
      "learning_rate": 6.317675568547704e-06,
      "loss": 0.7212015986442566,
      "step": 1029
    },
    {
      "epoch": 1.3361112800859745,
      "grad_norm": 0.7560867071151733,
      "learning_rate": 6.310777235264849e-06,
      "loss": 0.7757279872894287,
      "step": 1030
    },
    {
      "epoch": 1.3374090070361133,
      "grad_norm": 0.7711772322654724,
      "learning_rate": 6.303876221774311e-06,
      "loss": 0.7688950896263123,
      "step": 1031
    },
    {
      "epoch": 1.3387067339862522,
      "grad_norm": 0.7785727381706238,
      "learning_rate": 6.296972542186915e-06,
      "loss": 0.780545711517334,
      "step": 1032
    },
    {
      "epoch": 1.340004460936391,
      "grad_norm": 0.7750667929649353,
      "learning_rate": 6.2900662106189415e-06,
      "loss": 0.7841629385948181,
      "step": 1033
    },
    {
      "epoch": 1.34130218788653,
      "grad_norm": 0.7886035442352295,
      "learning_rate": 6.283157241192087e-06,
      "loss": 0.7848193645477295,
      "step": 1034
    },
    {
      "epoch": 1.342599914836669,
      "grad_norm": 0.7943287491798401,
      "learning_rate": 6.276245648033447e-06,
      "loss": 0.7743536233901978,
      "step": 1035
    },
    {
      "epoch": 1.3438976417868078,
      "grad_norm": 0.7815331220626831,
      "learning_rate": 6.2693314452754796e-06,
      "loss": 0.8415849208831787,
      "step": 1036
    },
    {
      "epoch": 1.3451953687369467,
      "grad_norm": 0.757429838180542,
      "learning_rate": 6.26241464705598e-06,
      "loss": 0.8025727272033691,
      "step": 1037
    },
    {
      "epoch": 1.3464930956870855,
      "grad_norm": 0.7403960227966309,
      "learning_rate": 6.25549526751805e-06,
      "loss": 0.7311757802963257,
      "step": 1038
    },
    {
      "epoch": 1.3477908226372244,
      "grad_norm": 0.7482463717460632,
      "learning_rate": 6.24857332081007e-06,
      "loss": 0.764176607131958,
      "step": 1039
    },
    {
      "epoch": 1.3490885495873635,
      "grad_norm": 0.749909520149231,
      "learning_rate": 6.241648821085666e-06,
      "loss": 0.7602924108505249,
      "step": 1040
    },
    {
      "epoch": 1.3503862765375023,
      "grad_norm": 0.7637724876403809,
      "learning_rate": 6.23472178250369e-06,
      "loss": 0.8251943588256836,
      "step": 1041
    },
    {
      "epoch": 1.3516840034876412,
      "grad_norm": 0.7882103323936462,
      "learning_rate": 6.227792219228183e-06,
      "loss": 0.8028998970985413,
      "step": 1042
    },
    {
      "epoch": 1.35298173043778,
      "grad_norm": 0.7483314275741577,
      "learning_rate": 6.220860145428347e-06,
      "loss": 0.7678120732307434,
      "step": 1043
    },
    {
      "epoch": 1.354279457387919,
      "grad_norm": 0.7539233565330505,
      "learning_rate": 6.213925575278518e-06,
      "loss": 0.7885809540748596,
      "step": 1044
    },
    {
      "epoch": 1.355577184338058,
      "grad_norm": 0.7156683206558228,
      "learning_rate": 6.206988522958135e-06,
      "loss": 0.707705557346344,
      "step": 1045
    },
    {
      "epoch": 1.3568749112881968,
      "grad_norm": 0.7801918387413025,
      "learning_rate": 6.200049002651718e-06,
      "loss": 0.811095654964447,
      "step": 1046
    },
    {
      "epoch": 1.3581726382383357,
      "grad_norm": 0.7180908918380737,
      "learning_rate": 6.19310702854883e-06,
      "loss": 0.671703577041626,
      "step": 1047
    },
    {
      "epoch": 1.3594703651884745,
      "grad_norm": 0.7936683893203735,
      "learning_rate": 6.186162614844047e-06,
      "loss": 0.7717031836509705,
      "step": 1048
    },
    {
      "epoch": 1.3607680921386134,
      "grad_norm": 0.759444534778595,
      "learning_rate": 6.17921577573694e-06,
      "loss": 0.7092019319534302,
      "step": 1049
    },
    {
      "epoch": 1.3620658190887522,
      "grad_norm": 0.7096357345581055,
      "learning_rate": 6.172266525432036e-06,
      "loss": 0.7081625461578369,
      "step": 1050
    },
    {
      "epoch": 1.3633635460388913,
      "grad_norm": 0.6968888640403748,
      "learning_rate": 6.165314878138794e-06,
      "loss": 0.6797256469726562,
      "step": 1051
    },
    {
      "epoch": 1.3646612729890302,
      "grad_norm": 0.7835313081741333,
      "learning_rate": 6.1583608480715705e-06,
      "loss": 0.8080659508705139,
      "step": 1052
    },
    {
      "epoch": 1.365958999939169,
      "grad_norm": 0.7313963174819946,
      "learning_rate": 6.1514044494496e-06,
      "loss": 0.666755199432373,
      "step": 1053
    },
    {
      "epoch": 1.3672567268893079,
      "grad_norm": 0.7819347977638245,
      "learning_rate": 6.144445696496955e-06,
      "loss": 0.8489786982536316,
      "step": 1054
    },
    {
      "epoch": 1.368554453839447,
      "grad_norm": 0.7251116037368774,
      "learning_rate": 6.137484603442524e-06,
      "loss": 0.7178645730018616,
      "step": 1055
    },
    {
      "epoch": 1.3698521807895858,
      "grad_norm": 0.8075553178787231,
      "learning_rate": 6.130521184519983e-06,
      "loss": 0.7850656509399414,
      "step": 1056
    },
    {
      "epoch": 1.3711499077397247,
      "grad_norm": 0.7090011239051819,
      "learning_rate": 6.123555453967759e-06,
      "loss": 0.7282134294509888,
      "step": 1057
    },
    {
      "epoch": 1.3724476346898635,
      "grad_norm": 0.7184298634529114,
      "learning_rate": 6.1165874260290074e-06,
      "loss": 0.7176535129547119,
      "step": 1058
    },
    {
      "epoch": 1.3737453616400024,
      "grad_norm": 0.7551043629646301,
      "learning_rate": 6.109617114951581e-06,
      "loss": 0.7516987323760986,
      "step": 1059
    },
    {
      "epoch": 1.3750430885901412,
      "grad_norm": 0.7810531258583069,
      "learning_rate": 6.102644534988006e-06,
      "loss": 0.7351160049438477,
      "step": 1060
    },
    {
      "epoch": 1.37634081554028,
      "grad_norm": 0.741783618927002,
      "learning_rate": 6.0956697003954404e-06,
      "loss": 0.7304431200027466,
      "step": 1061
    },
    {
      "epoch": 1.3776385424904192,
      "grad_norm": 0.7723316550254822,
      "learning_rate": 6.088692625435656e-06,
      "loss": 0.7627692818641663,
      "step": 1062
    },
    {
      "epoch": 1.378936269440558,
      "grad_norm": 0.7466050386428833,
      "learning_rate": 6.0817133243750046e-06,
      "loss": 0.7436075806617737,
      "step": 1063
    },
    {
      "epoch": 1.3802339963906969,
      "grad_norm": 0.784697413444519,
      "learning_rate": 6.074731811484391e-06,
      "loss": 0.7517378926277161,
      "step": 1064
    },
    {
      "epoch": 1.3815317233408357,
      "grad_norm": 0.7546585202217102,
      "learning_rate": 6.067748101039243e-06,
      "loss": 0.6963843703269958,
      "step": 1065
    },
    {
      "epoch": 1.3828294502909748,
      "grad_norm": 0.7581716179847717,
      "learning_rate": 6.060762207319479e-06,
      "loss": 0.7489246129989624,
      "step": 1066
    },
    {
      "epoch": 1.3841271772411137,
      "grad_norm": 0.7676309943199158,
      "learning_rate": 6.053774144609484e-06,
      "loss": 0.7299286127090454,
      "step": 1067
    },
    {
      "epoch": 1.3854249041912525,
      "grad_norm": 0.7663894891738892,
      "learning_rate": 6.046783927198079e-06,
      "loss": 0.7648952007293701,
      "step": 1068
    },
    {
      "epoch": 1.3867226311413914,
      "grad_norm": 0.7521557211875916,
      "learning_rate": 6.039791569378488e-06,
      "loss": 0.7617720365524292,
      "step": 1069
    },
    {
      "epoch": 1.3880203580915302,
      "grad_norm": 0.7601999044418335,
      "learning_rate": 6.032797085448315e-06,
      "loss": 0.7543430924415588,
      "step": 1070
    },
    {
      "epoch": 1.389318085041669,
      "grad_norm": 0.7353654503822327,
      "learning_rate": 6.025800489709505e-06,
      "loss": 0.799416720867157,
      "step": 1071
    },
    {
      "epoch": 1.3906158119918082,
      "grad_norm": 0.7517397999763489,
      "learning_rate": 6.018801796468328e-06,
      "loss": 0.82966148853302,
      "step": 1072
    },
    {
      "epoch": 1.391913538941947,
      "grad_norm": 0.8136231899261475,
      "learning_rate": 6.0118010200353396e-06,
      "loss": 0.8899416923522949,
      "step": 1073
    },
    {
      "epoch": 1.3932112658920859,
      "grad_norm": 0.7770370841026306,
      "learning_rate": 6.004798174725358e-06,
      "loss": 0.8107327818870544,
      "step": 1074
    },
    {
      "epoch": 1.3945089928422247,
      "grad_norm": 0.7465303540229797,
      "learning_rate": 5.997793274857427e-06,
      "loss": 0.769565999507904,
      "step": 1075
    },
    {
      "epoch": 1.3958067197923638,
      "grad_norm": 0.7476341128349304,
      "learning_rate": 5.990786334754795e-06,
      "loss": 0.8333680629730225,
      "step": 1076
    },
    {
      "epoch": 1.3971044467425027,
      "grad_norm": 0.7511208653450012,
      "learning_rate": 5.983777368744881e-06,
      "loss": 0.7648091316223145,
      "step": 1077
    },
    {
      "epoch": 1.3984021736926415,
      "grad_norm": 0.7745276093482971,
      "learning_rate": 5.9767663911592454e-06,
      "loss": 0.7386494278907776,
      "step": 1078
    },
    {
      "epoch": 1.3996999006427804,
      "grad_norm": 0.7422261238098145,
      "learning_rate": 5.9697534163335645e-06,
      "loss": 0.7428614497184753,
      "step": 1079
    },
    {
      "epoch": 1.4009976275929192,
      "grad_norm": 0.7582484483718872,
      "learning_rate": 5.9627384586075954e-06,
      "loss": 0.771366536617279,
      "step": 1080
    },
    {
      "epoch": 1.402295354543058,
      "grad_norm": 0.7568719387054443,
      "learning_rate": 5.955721532325151e-06,
      "loss": 0.7929791212081909,
      "step": 1081
    },
    {
      "epoch": 1.403593081493197,
      "grad_norm": 0.7572823762893677,
      "learning_rate": 5.94870265183407e-06,
      "loss": 0.7775267362594604,
      "step": 1082
    },
    {
      "epoch": 1.404890808443336,
      "grad_norm": 0.7899602055549622,
      "learning_rate": 5.941681831486188e-06,
      "loss": 0.7695810198783875,
      "step": 1083
    },
    {
      "epoch": 1.4061885353934749,
      "grad_norm": 0.7336994409561157,
      "learning_rate": 5.934659085637303e-06,
      "loss": 0.7132036685943604,
      "step": 1084
    },
    {
      "epoch": 1.4074862623436137,
      "grad_norm": 0.7522117495536804,
      "learning_rate": 5.927634428647154e-06,
      "loss": 0.7515596151351929,
      "step": 1085
    },
    {
      "epoch": 1.4087839892937526,
      "grad_norm": 0.7884212136268616,
      "learning_rate": 5.920607874879387e-06,
      "loss": 0.7542105317115784,
      "step": 1086
    },
    {
      "epoch": 1.4100817162438917,
      "grad_norm": 0.7597297430038452,
      "learning_rate": 5.913579438701525e-06,
      "loss": 0.8347772359848022,
      "step": 1087
    },
    {
      "epoch": 1.4113794431940305,
      "grad_norm": 0.7561913132667542,
      "learning_rate": 5.906549134484943e-06,
      "loss": 0.7754871249198914,
      "step": 1088
    },
    {
      "epoch": 1.4126771701441694,
      "grad_norm": 0.7146552801132202,
      "learning_rate": 5.899516976604832e-06,
      "loss": 0.7631995677947998,
      "step": 1089
    },
    {
      "epoch": 1.4139748970943082,
      "grad_norm": 0.7291439771652222,
      "learning_rate": 5.892482979440175e-06,
      "loss": 0.756955087184906,
      "step": 1090
    },
    {
      "epoch": 1.415272624044447,
      "grad_norm": 0.7642284631729126,
      "learning_rate": 5.885447157373716e-06,
      "loss": 0.7994847893714905,
      "step": 1091
    },
    {
      "epoch": 1.416570350994586,
      "grad_norm": 0.8073525428771973,
      "learning_rate": 5.878409524791931e-06,
      "loss": 0.7306899428367615,
      "step": 1092
    },
    {
      "epoch": 1.417868077944725,
      "grad_norm": 0.721283495426178,
      "learning_rate": 5.871370096084997e-06,
      "loss": 0.7839140892028809,
      "step": 1093
    },
    {
      "epoch": 1.4191658048948639,
      "grad_norm": 0.754751980304718,
      "learning_rate": 5.864328885646764e-06,
      "loss": 0.7405674457550049,
      "step": 1094
    },
    {
      "epoch": 1.4204635318450027,
      "grad_norm": 0.7623893618583679,
      "learning_rate": 5.857285907874725e-06,
      "loss": 0.7703812122344971,
      "step": 1095
    },
    {
      "epoch": 1.4217612587951416,
      "grad_norm": 0.7525649070739746,
      "learning_rate": 5.850241177169986e-06,
      "loss": 0.8114089965820312,
      "step": 1096
    },
    {
      "epoch": 1.4230589857452807,
      "grad_norm": 0.7692160606384277,
      "learning_rate": 5.84319470793724e-06,
      "loss": 0.7578598856925964,
      "step": 1097
    },
    {
      "epoch": 1.4243567126954195,
      "grad_norm": 0.7233715653419495,
      "learning_rate": 5.836146514584733e-06,
      "loss": 0.7252383828163147,
      "step": 1098
    },
    {
      "epoch": 1.4256544396455584,
      "grad_norm": 0.7710081338882446,
      "learning_rate": 5.829096611524235e-06,
      "loss": 0.7712035179138184,
      "step": 1099
    },
    {
      "epoch": 1.4269521665956972,
      "grad_norm": 0.7670133709907532,
      "learning_rate": 5.822045013171015e-06,
      "loss": 0.7823899388313293,
      "step": 1100
    },
    {
      "epoch": 1.428249893545836,
      "grad_norm": 0.7245627045631409,
      "learning_rate": 5.814991733943805e-06,
      "loss": 0.7533504962921143,
      "step": 1101
    },
    {
      "epoch": 1.429547620495975,
      "grad_norm": 0.7534570693969727,
      "learning_rate": 5.807936788264778e-06,
      "loss": 0.7617537379264832,
      "step": 1102
    },
    {
      "epoch": 1.4308453474461138,
      "grad_norm": 0.7101514935493469,
      "learning_rate": 5.800880190559511e-06,
      "loss": 0.7216631770133972,
      "step": 1103
    },
    {
      "epoch": 1.4321430743962529,
      "grad_norm": 0.7756243348121643,
      "learning_rate": 5.79382195525696e-06,
      "loss": 0.8646502494812012,
      "step": 1104
    },
    {
      "epoch": 1.4334408013463917,
      "grad_norm": 0.7338310480117798,
      "learning_rate": 5.786762096789431e-06,
      "loss": 0.7363216876983643,
      "step": 1105
    },
    {
      "epoch": 1.4347385282965306,
      "grad_norm": 0.7695474028587341,
      "learning_rate": 5.779700629592547e-06,
      "loss": 0.8422562479972839,
      "step": 1106
    },
    {
      "epoch": 1.4360362552466694,
      "grad_norm": 0.7274161577224731,
      "learning_rate": 5.7726375681052205e-06,
      "loss": 0.7430284023284912,
      "step": 1107
    },
    {
      "epoch": 1.4373339821968085,
      "grad_norm": 0.8229888677597046,
      "learning_rate": 5.765572926769625e-06,
      "loss": 0.8320131897926331,
      "step": 1108
    },
    {
      "epoch": 1.4386317091469474,
      "grad_norm": 0.7726709246635437,
      "learning_rate": 5.758506720031163e-06,
      "loss": 0.7141331434249878,
      "step": 1109
    },
    {
      "epoch": 1.4399294360970862,
      "grad_norm": 0.6873381733894348,
      "learning_rate": 5.751438962338441e-06,
      "loss": 0.6841472387313843,
      "step": 1110
    },
    {
      "epoch": 1.441227163047225,
      "grad_norm": 0.7722907066345215,
      "learning_rate": 5.744369668143233e-06,
      "loss": 0.7927265763282776,
      "step": 1111
    },
    {
      "epoch": 1.442524889997364,
      "grad_norm": 0.7216099500656128,
      "learning_rate": 5.737298851900457e-06,
      "loss": 0.7576711177825928,
      "step": 1112
    },
    {
      "epoch": 1.4438226169475028,
      "grad_norm": 0.7648237347602844,
      "learning_rate": 5.730226528068142e-06,
      "loss": 0.7920837998390198,
      "step": 1113
    },
    {
      "epoch": 1.4451203438976417,
      "grad_norm": 0.7815790772438049,
      "learning_rate": 5.7231527111074e-06,
      "loss": 0.8274810910224915,
      "step": 1114
    },
    {
      "epoch": 1.4464180708477807,
      "grad_norm": 0.7311232089996338,
      "learning_rate": 5.716077415482398e-06,
      "loss": 0.7946675419807434,
      "step": 1115
    },
    {
      "epoch": 1.4477157977979196,
      "grad_norm": 0.7482665181159973,
      "learning_rate": 5.709000655660324e-06,
      "loss": 0.758356511592865,
      "step": 1116
    },
    {
      "epoch": 1.4490135247480584,
      "grad_norm": 0.6977280378341675,
      "learning_rate": 5.7019224461113585e-06,
      "loss": 0.681710958480835,
      "step": 1117
    },
    {
      "epoch": 1.4503112516981973,
      "grad_norm": 0.8195312023162842,
      "learning_rate": 5.694842801308651e-06,
      "loss": 0.752457857131958,
      "step": 1118
    },
    {
      "epoch": 1.4516089786483364,
      "grad_norm": 0.7203164100646973,
      "learning_rate": 5.687761735728282e-06,
      "loss": 0.7394000291824341,
      "step": 1119
    },
    {
      "epoch": 1.4529067055984752,
      "grad_norm": 0.7854275107383728,
      "learning_rate": 5.680679263849241e-06,
      "loss": 0.7581281661987305,
      "step": 1120
    },
    {
      "epoch": 1.454204432548614,
      "grad_norm": 0.723127543926239,
      "learning_rate": 5.673595400153385e-06,
      "loss": 0.7252097725868225,
      "step": 1121
    },
    {
      "epoch": 1.455502159498753,
      "grad_norm": 0.7311640381813049,
      "learning_rate": 5.666510159125427e-06,
      "loss": 0.77386075258255,
      "step": 1122
    },
    {
      "epoch": 1.4567998864488918,
      "grad_norm": 0.735611617565155,
      "learning_rate": 5.65942355525289e-06,
      "loss": 0.730672299861908,
      "step": 1123
    },
    {
      "epoch": 1.4580976133990307,
      "grad_norm": 0.7109529972076416,
      "learning_rate": 5.652335603026084e-06,
      "loss": 0.6859084963798523,
      "step": 1124
    },
    {
      "epoch": 1.4593953403491697,
      "grad_norm": 0.7404365539550781,
      "learning_rate": 5.645246316938082e-06,
      "loss": 0.7842783331871033,
      "step": 1125
    },
    {
      "epoch": 1.4606930672993086,
      "grad_norm": 0.7049553394317627,
      "learning_rate": 5.638155711484674e-06,
      "loss": 0.7448842525482178,
      "step": 1126
    },
    {
      "epoch": 1.4619907942494474,
      "grad_norm": 0.7657853364944458,
      "learning_rate": 5.631063801164356e-06,
      "loss": 0.7713245749473572,
      "step": 1127
    },
    {
      "epoch": 1.4632885211995863,
      "grad_norm": 0.7475281953811646,
      "learning_rate": 5.62397060047829e-06,
      "loss": 0.7330006957054138,
      "step": 1128
    },
    {
      "epoch": 1.4645862481497254,
      "grad_norm": 0.7779927849769592,
      "learning_rate": 5.6168761239302745e-06,
      "loss": 0.7859174609184265,
      "step": 1129
    },
    {
      "epoch": 1.4658839750998642,
      "grad_norm": 0.7732834219932556,
      "learning_rate": 5.609780386026721e-06,
      "loss": 0.7622615694999695,
      "step": 1130
    },
    {
      "epoch": 1.467181702050003,
      "grad_norm": 0.7757659554481506,
      "learning_rate": 5.6026834012766155e-06,
      "loss": 0.7179020047187805,
      "step": 1131
    },
    {
      "epoch": 1.468479429000142,
      "grad_norm": 0.8133222460746765,
      "learning_rate": 5.595585184191496e-06,
      "loss": 0.8392655849456787,
      "step": 1132
    },
    {
      "epoch": 1.4697771559502808,
      "grad_norm": 0.741298496723175,
      "learning_rate": 5.58848574928542e-06,
      "loss": 0.770621657371521,
      "step": 1133
    },
    {
      "epoch": 1.4710748829004197,
      "grad_norm": 0.7434273362159729,
      "learning_rate": 5.5813851110749365e-06,
      "loss": 0.7763687372207642,
      "step": 1134
    },
    {
      "epoch": 1.4723726098505585,
      "grad_norm": 0.7332520484924316,
      "learning_rate": 5.574283284079049e-06,
      "loss": 0.7230371236801147,
      "step": 1135
    },
    {
      "epoch": 1.4736703368006976,
      "grad_norm": 0.7544539570808411,
      "learning_rate": 5.567180282819201e-06,
      "loss": 0.7774012684822083,
      "step": 1136
    },
    {
      "epoch": 1.4749680637508364,
      "grad_norm": 0.7133947610855103,
      "learning_rate": 5.560076121819229e-06,
      "loss": 0.7977353930473328,
      "step": 1137
    },
    {
      "epoch": 1.4762657907009753,
      "grad_norm": 0.7752287983894348,
      "learning_rate": 5.552970815605347e-06,
      "loss": 0.7770974040031433,
      "step": 1138
    },
    {
      "epoch": 1.4775635176511142,
      "grad_norm": 0.7379096150398254,
      "learning_rate": 5.545864378706106e-06,
      "loss": 0.7216660380363464,
      "step": 1139
    },
    {
      "epoch": 1.4788612446012532,
      "grad_norm": 0.7484530806541443,
      "learning_rate": 5.53875682565237e-06,
      "loss": 0.7281945943832397,
      "step": 1140
    },
    {
      "epoch": 1.480158971551392,
      "grad_norm": 0.7698119878768921,
      "learning_rate": 5.5316481709772886e-06,
      "loss": 0.7480529546737671,
      "step": 1141
    },
    {
      "epoch": 1.481456698501531,
      "grad_norm": 0.7418819665908813,
      "learning_rate": 5.524538429216258e-06,
      "loss": 0.8144744634628296,
      "step": 1142
    },
    {
      "epoch": 1.4827544254516698,
      "grad_norm": 0.7699658274650574,
      "learning_rate": 5.517427614906906e-06,
      "loss": 0.7931643128395081,
      "step": 1143
    },
    {
      "epoch": 1.4840521524018087,
      "grad_norm": 0.7453754544258118,
      "learning_rate": 5.510315742589042e-06,
      "loss": 0.8053094148635864,
      "step": 1144
    },
    {
      "epoch": 1.4853498793519475,
      "grad_norm": 0.7414461970329285,
      "learning_rate": 5.503202826804647e-06,
      "loss": 0.8256296515464783,
      "step": 1145
    },
    {
      "epoch": 1.4866476063020864,
      "grad_norm": 0.7538757920265198,
      "learning_rate": 5.496088882097836e-06,
      "loss": 0.776119589805603,
      "step": 1146
    },
    {
      "epoch": 1.4879453332522254,
      "grad_norm": 0.7799422740936279,
      "learning_rate": 5.488973923014821e-06,
      "loss": 0.7753342986106873,
      "step": 1147
    },
    {
      "epoch": 1.4892430602023643,
      "grad_norm": 0.7214382290840149,
      "learning_rate": 5.4818579641038974e-06,
      "loss": 0.7284688949584961,
      "step": 1148
    },
    {
      "epoch": 1.4905407871525032,
      "grad_norm": 0.7477608919143677,
      "learning_rate": 5.474741019915395e-06,
      "loss": 0.8078314065933228,
      "step": 1149
    },
    {
      "epoch": 1.491838514102642,
      "grad_norm": 0.7707608342170715,
      "learning_rate": 5.467623105001667e-06,
      "loss": 0.7474067211151123,
      "step": 1150
    },
    {
      "epoch": 1.491838514102642,
      "eval_loss": 0.7689123153686523,
      "eval_runtime": 143.0902,
      "eval_samples_per_second": 36.285,
      "eval_steps_per_second": 9.071,
      "step": 1150
    },
    {
      "epoch": 1.493136241052781,
      "grad_norm": 0.7697938084602356,
      "learning_rate": 5.460504233917047e-06,
      "loss": 0.7899725437164307,
      "step": 1151
    },
    {
      "epoch": 1.49443396800292,
      "grad_norm": 0.8120201826095581,
      "learning_rate": 5.453384421217823e-06,
      "loss": 0.7851537466049194,
      "step": 1152
    },
    {
      "epoch": 1.4957316949530588,
      "grad_norm": 0.7309612035751343,
      "learning_rate": 5.446263681462213e-06,
      "loss": 0.7107774019241333,
      "step": 1153
    },
    {
      "epoch": 1.4970294219031977,
      "grad_norm": 0.7438963651657104,
      "learning_rate": 5.439142029210323e-06,
      "loss": 0.788350522518158,
      "step": 1154
    },
    {
      "epoch": 1.4983271488533365,
      "grad_norm": 0.7466786503791809,
      "learning_rate": 5.4320194790241335e-06,
      "loss": 0.7024964094161987,
      "step": 1155
    },
    {
      "epoch": 1.4996248758034754,
      "grad_norm": 0.7446750998497009,
      "learning_rate": 5.424896045467455e-06,
      "loss": 0.7421405911445618,
      "step": 1156
    },
    {
      "epoch": 1.5009226027536142,
      "grad_norm": 0.7718220949172974,
      "learning_rate": 5.417771743105908e-06,
      "loss": 0.8068550825119019,
      "step": 1157
    },
    {
      "epoch": 1.5022203297037533,
      "grad_norm": 0.7445886731147766,
      "learning_rate": 5.4106465865068846e-06,
      "loss": 0.7749291658401489,
      "step": 1158
    },
    {
      "epoch": 1.5035180566538922,
      "grad_norm": 0.6862496733665466,
      "learning_rate": 5.403520590239527e-06,
      "loss": 0.6442810893058777,
      "step": 1159
    },
    {
      "epoch": 1.504815783604031,
      "grad_norm": 0.7648230791091919,
      "learning_rate": 5.396393768874696e-06,
      "loss": 0.8310628533363342,
      "step": 1160
    },
    {
      "epoch": 1.50611351055417,
      "grad_norm": 0.7615645527839661,
      "learning_rate": 5.389266136984939e-06,
      "loss": 0.738058865070343,
      "step": 1161
    },
    {
      "epoch": 1.507411237504309,
      "grad_norm": 0.7446813583374023,
      "learning_rate": 5.382137709144454e-06,
      "loss": 0.7968519330024719,
      "step": 1162
    },
    {
      "epoch": 1.5087089644544478,
      "grad_norm": 0.7255394458770752,
      "learning_rate": 5.3750084999290755e-06,
      "loss": 0.7655326724052429,
      "step": 1163
    },
    {
      "epoch": 1.5100066914045867,
      "grad_norm": 0.7433539628982544,
      "learning_rate": 5.3678785239162305e-06,
      "loss": 0.7728189826011658,
      "step": 1164
    },
    {
      "epoch": 1.5113044183547255,
      "grad_norm": 0.7390363216400146,
      "learning_rate": 5.360747795684916e-06,
      "loss": 0.6784142255783081,
      "step": 1165
    },
    {
      "epoch": 1.5126021453048644,
      "grad_norm": 0.7719975709915161,
      "learning_rate": 5.353616329815667e-06,
      "loss": 0.8225696086883545,
      "step": 1166
    },
    {
      "epoch": 1.5138998722550032,
      "grad_norm": 0.77885502576828,
      "learning_rate": 5.346484140890523e-06,
      "loss": 0.7254043817520142,
      "step": 1167
    },
    {
      "epoch": 1.5151975992051423,
      "grad_norm": 0.7126129269599915,
      "learning_rate": 5.339351243493008e-06,
      "loss": 0.7107834815979004,
      "step": 1168
    },
    {
      "epoch": 1.5164953261552812,
      "grad_norm": 0.7463387250900269,
      "learning_rate": 5.332217652208093e-06,
      "loss": 0.7344364523887634,
      "step": 1169
    },
    {
      "epoch": 1.51779305310542,
      "grad_norm": 0.7663910984992981,
      "learning_rate": 5.325083381622165e-06,
      "loss": 0.8212227821350098,
      "step": 1170
    },
    {
      "epoch": 1.519090780055559,
      "grad_norm": 0.7092857956886292,
      "learning_rate": 5.317948446322999e-06,
      "loss": 0.6895952820777893,
      "step": 1171
    },
    {
      "epoch": 1.520388507005698,
      "grad_norm": 0.7231706380844116,
      "learning_rate": 5.310812860899737e-06,
      "loss": 0.7563647627830505,
      "step": 1172
    },
    {
      "epoch": 1.5216862339558368,
      "grad_norm": 0.7281357049942017,
      "learning_rate": 5.303676639942841e-06,
      "loss": 0.7277164459228516,
      "step": 1173
    },
    {
      "epoch": 1.5229839609059757,
      "grad_norm": 0.7401008605957031,
      "learning_rate": 5.296539798044078e-06,
      "loss": 0.7237936854362488,
      "step": 1174
    },
    {
      "epoch": 1.5242816878561145,
      "grad_norm": 0.7612598538398743,
      "learning_rate": 5.289402349796484e-06,
      "loss": 0.7660357356071472,
      "step": 1175
    },
    {
      "epoch": 1.5255794148062534,
      "grad_norm": 0.7405344247817993,
      "learning_rate": 5.282264309794334e-06,
      "loss": 0.6943804621696472,
      "step": 1176
    },
    {
      "epoch": 1.5268771417563922,
      "grad_norm": 0.7092010974884033,
      "learning_rate": 5.2751256926331115e-06,
      "loss": 0.7395352721214294,
      "step": 1177
    },
    {
      "epoch": 1.528174868706531,
      "grad_norm": 0.7905942797660828,
      "learning_rate": 5.267986512909484e-06,
      "loss": 0.7439973950386047,
      "step": 1178
    },
    {
      "epoch": 1.5294725956566702,
      "grad_norm": 0.7583009600639343,
      "learning_rate": 5.2608467852212665e-06,
      "loss": 0.7969813346862793,
      "step": 1179
    },
    {
      "epoch": 1.530770322606809,
      "grad_norm": 0.7311235070228577,
      "learning_rate": 5.253706524167395e-06,
      "loss": 0.7828118801116943,
      "step": 1180
    },
    {
      "epoch": 1.5320680495569479,
      "grad_norm": 0.726702868938446,
      "learning_rate": 5.246565744347894e-06,
      "loss": 0.7190544605255127,
      "step": 1181
    },
    {
      "epoch": 1.533365776507087,
      "grad_norm": 0.7332654595375061,
      "learning_rate": 5.2394244603638536e-06,
      "loss": 0.7419191002845764,
      "step": 1182
    },
    {
      "epoch": 1.5346635034572258,
      "grad_norm": 0.6836192607879639,
      "learning_rate": 5.232282686817392e-06,
      "loss": 0.683068573474884,
      "step": 1183
    },
    {
      "epoch": 1.5359612304073647,
      "grad_norm": 0.722237765789032,
      "learning_rate": 5.2251404383116265e-06,
      "loss": 0.7589780688285828,
      "step": 1184
    },
    {
      "epoch": 1.5372589573575035,
      "grad_norm": 0.7364366054534912,
      "learning_rate": 5.217997729450649e-06,
      "loss": 0.7445809841156006,
      "step": 1185
    },
    {
      "epoch": 1.5385566843076424,
      "grad_norm": 0.7376484870910645,
      "learning_rate": 5.21085457483949e-06,
      "loss": 0.7111762166023254,
      "step": 1186
    },
    {
      "epoch": 1.5398544112577812,
      "grad_norm": 0.748837947845459,
      "learning_rate": 5.203710989084093e-06,
      "loss": 0.8105229139328003,
      "step": 1187
    },
    {
      "epoch": 1.54115213820792,
      "grad_norm": 0.7624726891517639,
      "learning_rate": 5.196566986791286e-06,
      "loss": 0.8003865480422974,
      "step": 1188
    },
    {
      "epoch": 1.542449865158059,
      "grad_norm": 0.7348607182502747,
      "learning_rate": 5.189422582568742e-06,
      "loss": 0.7221018671989441,
      "step": 1189
    },
    {
      "epoch": 1.543747592108198,
      "grad_norm": 0.748846709728241,
      "learning_rate": 5.182277791024959e-06,
      "loss": 0.7588199377059937,
      "step": 1190
    },
    {
      "epoch": 1.5450453190583369,
      "grad_norm": 0.7500258684158325,
      "learning_rate": 5.175132626769229e-06,
      "loss": 0.7608622312545776,
      "step": 1191
    },
    {
      "epoch": 1.546343046008476,
      "grad_norm": 0.7673987150192261,
      "learning_rate": 5.167987104411605e-06,
      "loss": 0.815061092376709,
      "step": 1192
    },
    {
      "epoch": 1.5476407729586148,
      "grad_norm": 0.7776307463645935,
      "learning_rate": 5.160841238562872e-06,
      "loss": 0.7571362257003784,
      "step": 1193
    },
    {
      "epoch": 1.5489384999087537,
      "grad_norm": 0.7552419304847717,
      "learning_rate": 5.153695043834513e-06,
      "loss": 0.7361883521080017,
      "step": 1194
    },
    {
      "epoch": 1.5502362268588925,
      "grad_norm": 0.7167699933052063,
      "learning_rate": 5.146548534838691e-06,
      "loss": 0.7665646076202393,
      "step": 1195
    },
    {
      "epoch": 1.5515339538090314,
      "grad_norm": 0.717226505279541,
      "learning_rate": 5.139401726188208e-06,
      "loss": 0.691736102104187,
      "step": 1196
    },
    {
      "epoch": 1.5528316807591702,
      "grad_norm": 0.7594795227050781,
      "learning_rate": 5.132254632496477e-06,
      "loss": 0.7824583053588867,
      "step": 1197
    },
    {
      "epoch": 1.554129407709309,
      "grad_norm": 0.7554059028625488,
      "learning_rate": 5.125107268377498e-06,
      "loss": 0.7830257415771484,
      "step": 1198
    },
    {
      "epoch": 1.555427134659448,
      "grad_norm": 0.7319567203521729,
      "learning_rate": 5.117959648445821e-06,
      "loss": 0.7027431130409241,
      "step": 1199
    },
    {
      "epoch": 1.556724861609587,
      "grad_norm": 0.783620297908783,
      "learning_rate": 5.1108117873165175e-06,
      "loss": 0.8177133798599243,
      "step": 1200
    },
    {
      "epoch": 1.5580225885597259,
      "grad_norm": 0.7518169283866882,
      "learning_rate": 5.1036636996051556e-06,
      "loss": 0.7404662370681763,
      "step": 1201
    },
    {
      "epoch": 1.5593203155098647,
      "grad_norm": 0.7498476505279541,
      "learning_rate": 5.096515399927767e-06,
      "loss": 0.71551513671875,
      "step": 1202
    },
    {
      "epoch": 1.5606180424600038,
      "grad_norm": 0.7556577324867249,
      "learning_rate": 5.089366902900813e-06,
      "loss": 0.7790558338165283,
      "step": 1203
    },
    {
      "epoch": 1.5619157694101427,
      "grad_norm": 0.7798899412155151,
      "learning_rate": 5.082218223141162e-06,
      "loss": 0.7935107350349426,
      "step": 1204
    },
    {
      "epoch": 1.5632134963602815,
      "grad_norm": 0.704472541809082,
      "learning_rate": 5.075069375266055e-06,
      "loss": 0.7023574709892273,
      "step": 1205
    },
    {
      "epoch": 1.5645112233104204,
      "grad_norm": 0.7147163152694702,
      "learning_rate": 5.067920373893075e-06,
      "loss": 0.6481233835220337,
      "step": 1206
    },
    {
      "epoch": 1.5658089502605592,
      "grad_norm": 0.824305534362793,
      "learning_rate": 5.060771233640122e-06,
      "loss": 0.7639428973197937,
      "step": 1207
    },
    {
      "epoch": 1.567106677210698,
      "grad_norm": 0.7788658738136292,
      "learning_rate": 5.0536219691253776e-06,
      "loss": 0.677037239074707,
      "step": 1208
    },
    {
      "epoch": 1.568404404160837,
      "grad_norm": 0.7424824833869934,
      "learning_rate": 5.046472594967279e-06,
      "loss": 0.7886664867401123,
      "step": 1209
    },
    {
      "epoch": 1.5697021311109758,
      "grad_norm": 0.7809754610061646,
      "learning_rate": 5.039323125784485e-06,
      "loss": 0.8234939575195312,
      "step": 1210
    },
    {
      "epoch": 1.5709998580611149,
      "grad_norm": 0.7964925169944763,
      "learning_rate": 5.0321735761958515e-06,
      "loss": 0.7490283846855164,
      "step": 1211
    },
    {
      "epoch": 1.5722975850112537,
      "grad_norm": 0.7560259699821472,
      "learning_rate": 5.025023960820399e-06,
      "loss": 0.7197484970092773,
      "step": 1212
    },
    {
      "epoch": 1.5735953119613926,
      "grad_norm": 0.7597184777259827,
      "learning_rate": 5.01787429427728e-06,
      "loss": 0.7696546316146851,
      "step": 1213
    },
    {
      "epoch": 1.5748930389115317,
      "grad_norm": 0.7353492975234985,
      "learning_rate": 5.010724591185752e-06,
      "loss": 0.7226786613464355,
      "step": 1214
    },
    {
      "epoch": 1.5761907658616705,
      "grad_norm": 0.7540714144706726,
      "learning_rate": 5.003574866165149e-06,
      "loss": 0.7219122648239136,
      "step": 1215
    },
    {
      "epoch": 1.5774884928118094,
      "grad_norm": 0.7779788374900818,
      "learning_rate": 4.9964251338348515e-06,
      "loss": 0.8082325458526611,
      "step": 1216
    },
    {
      "epoch": 1.5787862197619482,
      "grad_norm": 0.7652958631515503,
      "learning_rate": 4.989275408814251e-06,
      "loss": 0.7007216215133667,
      "step": 1217
    },
    {
      "epoch": 1.580083946712087,
      "grad_norm": 0.7894110679626465,
      "learning_rate": 4.982125705722722e-06,
      "loss": 0.7830923795700073,
      "step": 1218
    },
    {
      "epoch": 1.581381673662226,
      "grad_norm": 0.7912439703941345,
      "learning_rate": 4.974976039179604e-06,
      "loss": 0.7531068325042725,
      "step": 1219
    },
    {
      "epoch": 1.5826794006123648,
      "grad_norm": 0.7875705361366272,
      "learning_rate": 4.967826423804151e-06,
      "loss": 0.7552053332328796,
      "step": 1220
    },
    {
      "epoch": 1.5839771275625036,
      "grad_norm": 0.7627431154251099,
      "learning_rate": 4.960676874215518e-06,
      "loss": 0.7305505871772766,
      "step": 1221
    },
    {
      "epoch": 1.5852748545126427,
      "grad_norm": 0.7455613613128662,
      "learning_rate": 4.953527405032723e-06,
      "loss": 0.6995049715042114,
      "step": 1222
    },
    {
      "epoch": 1.5865725814627816,
      "grad_norm": 0.716137170791626,
      "learning_rate": 4.946378030874625e-06,
      "loss": 0.7280579209327698,
      "step": 1223
    },
    {
      "epoch": 1.5878703084129207,
      "grad_norm": 0.7123472094535828,
      "learning_rate": 4.9392287663598785e-06,
      "loss": 0.7188202142715454,
      "step": 1224
    },
    {
      "epoch": 1.5891680353630595,
      "grad_norm": 0.8411170244216919,
      "learning_rate": 4.932079626106926e-06,
      "loss": 0.8100149035453796,
      "step": 1225
    },
    {
      "epoch": 1.5904657623131984,
      "grad_norm": 0.755990207195282,
      "learning_rate": 4.924930624733947e-06,
      "loss": 0.8015238046646118,
      "step": 1226
    },
    {
      "epoch": 1.5917634892633372,
      "grad_norm": 0.780761182308197,
      "learning_rate": 4.91778177685884e-06,
      "loss": 0.775886595249176,
      "step": 1227
    },
    {
      "epoch": 1.593061216213476,
      "grad_norm": 0.8183882832527161,
      "learning_rate": 4.910633097099188e-06,
      "loss": 0.744438886642456,
      "step": 1228
    },
    {
      "epoch": 1.594358943163615,
      "grad_norm": 0.7258551120758057,
      "learning_rate": 4.903484600072236e-06,
      "loss": 0.6860997676849365,
      "step": 1229
    },
    {
      "epoch": 1.5956566701137538,
      "grad_norm": 0.695157527923584,
      "learning_rate": 4.896336300394845e-06,
      "loss": 0.7503056526184082,
      "step": 1230
    },
    {
      "epoch": 1.5969543970638926,
      "grad_norm": 0.7541502118110657,
      "learning_rate": 4.889188212683483e-06,
      "loss": 0.7562340497970581,
      "step": 1231
    },
    {
      "epoch": 1.5982521240140317,
      "grad_norm": 0.768254280090332,
      "learning_rate": 4.882040351554181e-06,
      "loss": 0.7829647064208984,
      "step": 1232
    },
    {
      "epoch": 1.5995498509641706,
      "grad_norm": 0.7578057646751404,
      "learning_rate": 4.874892731622503e-06,
      "loss": 0.7859159708023071,
      "step": 1233
    },
    {
      "epoch": 1.6008475779143094,
      "grad_norm": 0.7558201551437378,
      "learning_rate": 4.867745367503524e-06,
      "loss": 0.7639919519424438,
      "step": 1234
    },
    {
      "epoch": 1.6021453048644485,
      "grad_norm": 0.7190950512886047,
      "learning_rate": 4.860598273811793e-06,
      "loss": 0.7586647868156433,
      "step": 1235
    },
    {
      "epoch": 1.6034430318145874,
      "grad_norm": 0.7575900554656982,
      "learning_rate": 4.8534514651613104e-06,
      "loss": 0.7450016736984253,
      "step": 1236
    },
    {
      "epoch": 1.6047407587647262,
      "grad_norm": 0.7599023580551147,
      "learning_rate": 4.846304956165488e-06,
      "loss": 0.7748116254806519,
      "step": 1237
    },
    {
      "epoch": 1.606038485714865,
      "grad_norm": 0.7400065660476685,
      "learning_rate": 4.83915876143713e-06,
      "loss": 0.7867463827133179,
      "step": 1238
    },
    {
      "epoch": 1.607336212665004,
      "grad_norm": 0.7144732475280762,
      "learning_rate": 4.832012895588395e-06,
      "loss": 0.7374265789985657,
      "step": 1239
    },
    {
      "epoch": 1.6086339396151428,
      "grad_norm": 0.7419501543045044,
      "learning_rate": 4.824867373230772e-06,
      "loss": 0.785868227481842,
      "step": 1240
    },
    {
      "epoch": 1.6099316665652816,
      "grad_norm": 0.7401681542396545,
      "learning_rate": 4.817722208975041e-06,
      "loss": 0.7132147550582886,
      "step": 1241
    },
    {
      "epoch": 1.6112293935154205,
      "grad_norm": 0.7523838877677917,
      "learning_rate": 4.81057741743126e-06,
      "loss": 0.6855551600456238,
      "step": 1242
    },
    {
      "epoch": 1.6125271204655596,
      "grad_norm": 0.7317535281181335,
      "learning_rate": 4.8034330132087155e-06,
      "loss": 0.7517306208610535,
      "step": 1243
    },
    {
      "epoch": 1.6138248474156984,
      "grad_norm": 0.784957230091095,
      "learning_rate": 4.7962890109159085e-06,
      "loss": 0.8171036243438721,
      "step": 1244
    },
    {
      "epoch": 1.6151225743658375,
      "grad_norm": 0.794258177280426,
      "learning_rate": 4.789145425160511e-06,
      "loss": 0.7712936401367188,
      "step": 1245
    },
    {
      "epoch": 1.6164203013159764,
      "grad_norm": 0.7608366012573242,
      "learning_rate": 4.782002270549354e-06,
      "loss": 0.7149404883384705,
      "step": 1246
    },
    {
      "epoch": 1.6177180282661152,
      "grad_norm": 0.7840806245803833,
      "learning_rate": 4.774859561688374e-06,
      "loss": 0.8546009063720703,
      "step": 1247
    },
    {
      "epoch": 1.619015755216254,
      "grad_norm": 0.8032622337341309,
      "learning_rate": 4.767717313182611e-06,
      "loss": 0.82121741771698,
      "step": 1248
    },
    {
      "epoch": 1.620313482166393,
      "grad_norm": 0.7887139916419983,
      "learning_rate": 4.760575539636147e-06,
      "loss": 0.8578335642814636,
      "step": 1249
    },
    {
      "epoch": 1.6216112091165318,
      "grad_norm": 0.7383653521537781,
      "learning_rate": 4.753434255652108e-06,
      "loss": 0.7504441738128662,
      "step": 1250
    },
    {
      "epoch": 1.6229089360666706,
      "grad_norm": 0.7748722434043884,
      "learning_rate": 4.746293475832607e-06,
      "loss": 0.8303184509277344,
      "step": 1251
    },
    {
      "epoch": 1.6242066630168095,
      "grad_norm": 0.718803346157074,
      "learning_rate": 4.739153214778735e-06,
      "loss": 0.7208773493766785,
      "step": 1252
    },
    {
      "epoch": 1.6255043899669486,
      "grad_norm": 0.7380462884902954,
      "learning_rate": 4.732013487090517e-06,
      "loss": 0.7971898913383484,
      "step": 1253
    },
    {
      "epoch": 1.6268021169170874,
      "grad_norm": 0.7098410129547119,
      "learning_rate": 4.72487430736689e-06,
      "loss": 0.7401196360588074,
      "step": 1254
    },
    {
      "epoch": 1.6280998438672263,
      "grad_norm": 0.7748030424118042,
      "learning_rate": 4.7177356902056675e-06,
      "loss": 0.7918237447738647,
      "step": 1255
    },
    {
      "epoch": 1.6293975708173654,
      "grad_norm": 0.7241711616516113,
      "learning_rate": 4.7105976502035175e-06,
      "loss": 0.7051481008529663,
      "step": 1256
    },
    {
      "epoch": 1.6306952977675042,
      "grad_norm": 0.7323998808860779,
      "learning_rate": 4.703460201955924e-06,
      "loss": 0.7552366256713867,
      "step": 1257
    },
    {
      "epoch": 1.631993024717643,
      "grad_norm": 0.7378799915313721,
      "learning_rate": 4.696323360057162e-06,
      "loss": 0.7966490983963013,
      "step": 1258
    },
    {
      "epoch": 1.633290751667782,
      "grad_norm": 0.7449315190315247,
      "learning_rate": 4.689187139100265e-06,
      "loss": 0.8123936057090759,
      "step": 1259
    },
    {
      "epoch": 1.6345884786179208,
      "grad_norm": 0.735011100769043,
      "learning_rate": 4.682051553677001e-06,
      "loss": 0.7324624061584473,
      "step": 1260
    },
    {
      "epoch": 1.6358862055680596,
      "grad_norm": 0.7569977641105652,
      "learning_rate": 4.6749166183778375e-06,
      "loss": 0.7480865120887756,
      "step": 1261
    },
    {
      "epoch": 1.6371839325181985,
      "grad_norm": 0.7466624975204468,
      "learning_rate": 4.667782347791908e-06,
      "loss": 0.7476634383201599,
      "step": 1262
    },
    {
      "epoch": 1.6384816594683373,
      "grad_norm": 0.7527311444282532,
      "learning_rate": 4.660648756506993e-06,
      "loss": 0.7576185464859009,
      "step": 1263
    },
    {
      "epoch": 1.6397793864184764,
      "grad_norm": 0.7631829977035522,
      "learning_rate": 4.653515859109478e-06,
      "loss": 0.768151581287384,
      "step": 1264
    },
    {
      "epoch": 1.6410771133686153,
      "grad_norm": 0.7673115134239197,
      "learning_rate": 4.646383670184336e-06,
      "loss": 0.8062424659729004,
      "step": 1265
    },
    {
      "epoch": 1.6423748403187541,
      "grad_norm": 0.7701263427734375,
      "learning_rate": 4.639252204315086e-06,
      "loss": 0.7878407835960388,
      "step": 1266
    },
    {
      "epoch": 1.6436725672688932,
      "grad_norm": 0.7331056594848633,
      "learning_rate": 4.632121476083772e-06,
      "loss": 0.7622208595275879,
      "step": 1267
    },
    {
      "epoch": 1.644970294219032,
      "grad_norm": 0.7359024286270142,
      "learning_rate": 4.624991500070925e-06,
      "loss": 0.7808008193969727,
      "step": 1268
    },
    {
      "epoch": 1.646268021169171,
      "grad_norm": 0.7646325826644897,
      "learning_rate": 4.617862290855548e-06,
      "loss": 0.7223784923553467,
      "step": 1269
    },
    {
      "epoch": 1.6475657481193098,
      "grad_norm": 2.2569127082824707,
      "learning_rate": 4.610733863015063e-06,
      "loss": 0.8157532215118408,
      "step": 1270
    },
    {
      "epoch": 1.6488634750694486,
      "grad_norm": 0.7320986390113831,
      "learning_rate": 4.6036062311253055e-06,
      "loss": 0.7005835175514221,
      "step": 1271
    },
    {
      "epoch": 1.6501612020195875,
      "grad_norm": 0.7109540104866028,
      "learning_rate": 4.596479409760474e-06,
      "loss": 0.6603987812995911,
      "step": 1272
    },
    {
      "epoch": 1.6514589289697263,
      "grad_norm": 0.7034411430358887,
      "learning_rate": 4.589353413493118e-06,
      "loss": 0.7430425882339478,
      "step": 1273
    },
    {
      "epoch": 1.6527566559198652,
      "grad_norm": 0.740450382232666,
      "learning_rate": 4.582228256894093e-06,
      "loss": 0.7534090280532837,
      "step": 1274
    },
    {
      "epoch": 1.6540543828700043,
      "grad_norm": 0.758372962474823,
      "learning_rate": 4.575103954532547e-06,
      "loss": 0.7891790270805359,
      "step": 1275
    },
    {
      "epoch": 1.6553521098201431,
      "grad_norm": 0.7437251210212708,
      "learning_rate": 4.567980520975867e-06,
      "loss": 0.706454873085022,
      "step": 1276
    },
    {
      "epoch": 1.6566498367702822,
      "grad_norm": 0.7692578434944153,
      "learning_rate": 4.560857970789679e-06,
      "loss": 0.7044224739074707,
      "step": 1277
    },
    {
      "epoch": 1.657947563720421,
      "grad_norm": 0.7143827676773071,
      "learning_rate": 4.553736318537789e-06,
      "loss": 0.7750811576843262,
      "step": 1278
    },
    {
      "epoch": 1.65924529067056,
      "grad_norm": 0.7983530759811401,
      "learning_rate": 4.546615578782178e-06,
      "loss": 0.7537817358970642,
      "step": 1279
    },
    {
      "epoch": 1.6605430176206988,
      "grad_norm": 0.7134032249450684,
      "learning_rate": 4.5394957660829554e-06,
      "loss": 0.7322719097137451,
      "step": 1280
    },
    {
      "epoch": 1.6618407445708376,
      "grad_norm": 0.7665645480155945,
      "learning_rate": 4.532376894998335e-06,
      "loss": 0.7513355016708374,
      "step": 1281
    },
    {
      "epoch": 1.6631384715209765,
      "grad_norm": 0.7957409024238586,
      "learning_rate": 4.5252589800846054e-06,
      "loss": 0.7928126454353333,
      "step": 1282
    },
    {
      "epoch": 1.6644361984711153,
      "grad_norm": 0.7438947558403015,
      "learning_rate": 4.518142035896106e-06,
      "loss": 0.83896404504776,
      "step": 1283
    },
    {
      "epoch": 1.6657339254212542,
      "grad_norm": 0.7367180585861206,
      "learning_rate": 4.5110260769851804e-06,
      "loss": 0.7564518451690674,
      "step": 1284
    },
    {
      "epoch": 1.6670316523713933,
      "grad_norm": 0.7521037459373474,
      "learning_rate": 4.503911117902167e-06,
      "loss": 0.7650635838508606,
      "step": 1285
    },
    {
      "epoch": 1.6683293793215321,
      "grad_norm": 0.77150958776474,
      "learning_rate": 4.496797173195354e-06,
      "loss": 0.7417112588882446,
      "step": 1286
    },
    {
      "epoch": 1.669627106271671,
      "grad_norm": 0.7432476878166199,
      "learning_rate": 4.489684257410959e-06,
      "loss": 0.7239577770233154,
      "step": 1287
    },
    {
      "epoch": 1.67092483322181,
      "grad_norm": 0.7662708759307861,
      "learning_rate": 4.482572385093096e-06,
      "loss": 0.7873606085777283,
      "step": 1288
    },
    {
      "epoch": 1.672222560171949,
      "grad_norm": 0.7473525404930115,
      "learning_rate": 4.475461570783741e-06,
      "loss": 0.7298567891120911,
      "step": 1289
    },
    {
      "epoch": 1.6735202871220878,
      "grad_norm": 0.7975482940673828,
      "learning_rate": 4.468351829022713e-06,
      "loss": 0.8909924626350403,
      "step": 1290
    },
    {
      "epoch": 1.6748180140722266,
      "grad_norm": 0.7994987368583679,
      "learning_rate": 4.46124317434763e-06,
      "loss": 0.8102888464927673,
      "step": 1291
    },
    {
      "epoch": 1.6761157410223655,
      "grad_norm": 0.696455717086792,
      "learning_rate": 4.454135621293895e-06,
      "loss": 0.7557786703109741,
      "step": 1292
    },
    {
      "epoch": 1.6774134679725043,
      "grad_norm": 0.7524965405464172,
      "learning_rate": 4.447029184394654e-06,
      "loss": 0.6317112445831299,
      "step": 1293
    },
    {
      "epoch": 1.6787111949226432,
      "grad_norm": 0.7275172472000122,
      "learning_rate": 4.439923878180772e-06,
      "loss": 0.7290472388267517,
      "step": 1294
    },
    {
      "epoch": 1.680008921872782,
      "grad_norm": 0.7475486397743225,
      "learning_rate": 4.4328197171808e-06,
      "loss": 0.74071204662323,
      "step": 1295
    },
    {
      "epoch": 1.6813066488229211,
      "grad_norm": 0.711575984954834,
      "learning_rate": 4.425716715920952e-06,
      "loss": 0.770182728767395,
      "step": 1296
    },
    {
      "epoch": 1.68260437577306,
      "grad_norm": 0.7959085702896118,
      "learning_rate": 4.418614888925064e-06,
      "loss": 0.798844575881958,
      "step": 1297
    },
    {
      "epoch": 1.6839021027231988,
      "grad_norm": 0.7391006946563721,
      "learning_rate": 4.4115142507145806e-06,
      "loss": 0.7901135087013245,
      "step": 1298
    },
    {
      "epoch": 1.685199829673338,
      "grad_norm": 0.7250086665153503,
      "learning_rate": 4.4044148158085046e-06,
      "loss": 0.729693591594696,
      "step": 1299
    },
    {
      "epoch": 1.6864975566234768,
      "grad_norm": 0.7853794693946838,
      "learning_rate": 4.397316598723385e-06,
      "loss": 0.8253978490829468,
      "step": 1300
    },
    {
      "epoch": 1.6877952835736156,
      "grad_norm": 0.7512246370315552,
      "learning_rate": 4.39021961397328e-06,
      "loss": 0.7918137907981873,
      "step": 1301
    },
    {
      "epoch": 1.6890930105237545,
      "grad_norm": 0.7677380442619324,
      "learning_rate": 4.383123876069726e-06,
      "loss": 0.8250201940536499,
      "step": 1302
    },
    {
      "epoch": 1.6903907374738933,
      "grad_norm": 0.7484574317932129,
      "learning_rate": 4.376029399521711e-06,
      "loss": 0.790423572063446,
      "step": 1303
    },
    {
      "epoch": 1.6916884644240322,
      "grad_norm": 0.7543118000030518,
      "learning_rate": 4.368936198835646e-06,
      "loss": 0.7487786412239075,
      "step": 1304
    },
    {
      "epoch": 1.692986191374171,
      "grad_norm": 0.7802231907844543,
      "learning_rate": 4.361844288515327e-06,
      "loss": 0.7656086683273315,
      "step": 1305
    },
    {
      "epoch": 1.69428391832431,
      "grad_norm": 0.7723071575164795,
      "learning_rate": 4.354753683061921e-06,
      "loss": 0.7637503147125244,
      "step": 1306
    },
    {
      "epoch": 1.695581645274449,
      "grad_norm": 0.7176677584648132,
      "learning_rate": 4.347664396973917e-06,
      "loss": 0.6877071261405945,
      "step": 1307
    },
    {
      "epoch": 1.6968793722245878,
      "grad_norm": 0.7078595161437988,
      "learning_rate": 4.340576444747114e-06,
      "loss": 0.7922831773757935,
      "step": 1308
    },
    {
      "epoch": 1.698177099174727,
      "grad_norm": 0.8294334411621094,
      "learning_rate": 4.333489840874575e-06,
      "loss": 0.7542763352394104,
      "step": 1309
    },
    {
      "epoch": 1.6994748261248658,
      "grad_norm": 0.794155478477478,
      "learning_rate": 4.326404599846618e-06,
      "loss": 0.7189518213272095,
      "step": 1310
    },
    {
      "epoch": 1.7007725530750046,
      "grad_norm": 0.7483503818511963,
      "learning_rate": 4.319320736150762e-06,
      "loss": 0.7760790586471558,
      "step": 1311
    },
    {
      "epoch": 1.7020702800251435,
      "grad_norm": 0.7561513185501099,
      "learning_rate": 4.3122382642717196e-06,
      "loss": 0.7890743613243103,
      "step": 1312
    },
    {
      "epoch": 1.7033680069752823,
      "grad_norm": 0.7909126281738281,
      "learning_rate": 4.305157198691351e-06,
      "loss": 0.8085737228393555,
      "step": 1313
    },
    {
      "epoch": 1.7046657339254212,
      "grad_norm": 0.7337789535522461,
      "learning_rate": 4.298077553888644e-06,
      "loss": 0.7623671293258667,
      "step": 1314
    },
    {
      "epoch": 1.70596346087556,
      "grad_norm": 0.7426276803016663,
      "learning_rate": 4.290999344339678e-06,
      "loss": 0.760446310043335,
      "step": 1315
    },
    {
      "epoch": 1.707261187825699,
      "grad_norm": 0.7681460380554199,
      "learning_rate": 4.283922584517603e-06,
      "loss": 0.7784008383750916,
      "step": 1316
    },
    {
      "epoch": 1.708558914775838,
      "grad_norm": 0.75020432472229,
      "learning_rate": 4.276847288892601e-06,
      "loss": 0.7342908978462219,
      "step": 1317
    },
    {
      "epoch": 1.7098566417259768,
      "grad_norm": 0.7300113439559937,
      "learning_rate": 4.269773471931858e-06,
      "loss": 0.7642952799797058,
      "step": 1318
    },
    {
      "epoch": 1.7111543686761157,
      "grad_norm": 0.7630840539932251,
      "learning_rate": 4.262701148099544e-06,
      "loss": 0.8057624101638794,
      "step": 1319
    },
    {
      "epoch": 1.7124520956262548,
      "grad_norm": 0.7400659918785095,
      "learning_rate": 4.255630331856768e-06,
      "loss": 0.6892107129096985,
      "step": 1320
    },
    {
      "epoch": 1.7137498225763936,
      "grad_norm": 0.7181682586669922,
      "learning_rate": 4.248561037661561e-06,
      "loss": 0.6902667880058289,
      "step": 1321
    },
    {
      "epoch": 1.7150475495265325,
      "grad_norm": 0.7188737988471985,
      "learning_rate": 4.241493279968838e-06,
      "loss": 0.752339243888855,
      "step": 1322
    },
    {
      "epoch": 1.7163452764766713,
      "grad_norm": 0.7453453540802002,
      "learning_rate": 4.234427073230377e-06,
      "loss": 0.7290558815002441,
      "step": 1323
    },
    {
      "epoch": 1.7176430034268102,
      "grad_norm": 0.7392156720161438,
      "learning_rate": 4.22736243189478e-06,
      "loss": 0.7552244663238525,
      "step": 1324
    },
    {
      "epoch": 1.718940730376949,
      "grad_norm": 0.7706087827682495,
      "learning_rate": 4.220299370407454e-06,
      "loss": 0.8071123361587524,
      "step": 1325
    },
    {
      "epoch": 1.720238457327088,
      "grad_norm": 0.7364552617073059,
      "learning_rate": 4.2132379032105695e-06,
      "loss": 0.7699145078659058,
      "step": 1326
    },
    {
      "epoch": 1.7215361842772268,
      "grad_norm": 0.7225109934806824,
      "learning_rate": 4.206178044743041e-06,
      "loss": 0.7168894410133362,
      "step": 1327
    },
    {
      "epoch": 1.7228339112273658,
      "grad_norm": 0.7608538269996643,
      "learning_rate": 4.19911980944049e-06,
      "loss": 0.7722530961036682,
      "step": 1328
    },
    {
      "epoch": 1.7241316381775047,
      "grad_norm": 0.7884859442710876,
      "learning_rate": 4.1920632117352235e-06,
      "loss": 0.7422312498092651,
      "step": 1329
    },
    {
      "epoch": 1.7254293651276438,
      "grad_norm": 0.7449803352355957,
      "learning_rate": 4.185008266056195e-06,
      "loss": 0.6896304488182068,
      "step": 1330
    },
    {
      "epoch": 1.7267270920777826,
      "grad_norm": 0.7673011422157288,
      "learning_rate": 4.177954986828987e-06,
      "loss": 0.7713763117790222,
      "step": 1331
    },
    {
      "epoch": 1.7280248190279215,
      "grad_norm": 0.7378016710281372,
      "learning_rate": 4.170903388475766e-06,
      "loss": 0.7759542465209961,
      "step": 1332
    },
    {
      "epoch": 1.7293225459780603,
      "grad_norm": 0.7421383261680603,
      "learning_rate": 4.163853485415269e-06,
      "loss": 0.759284496307373,
      "step": 1333
    },
    {
      "epoch": 1.7306202729281992,
      "grad_norm": 0.7550718188285828,
      "learning_rate": 4.156805292062762e-06,
      "loss": 0.8264324069023132,
      "step": 1334
    },
    {
      "epoch": 1.731917999878338,
      "grad_norm": 0.730327308177948,
      "learning_rate": 4.1497588228300165e-06,
      "loss": 0.7197825312614441,
      "step": 1335
    },
    {
      "epoch": 1.733215726828477,
      "grad_norm": 0.7392425537109375,
      "learning_rate": 4.142714092125277e-06,
      "loss": 0.6834048628807068,
      "step": 1336
    },
    {
      "epoch": 1.7345134537786158,
      "grad_norm": 0.7402434945106506,
      "learning_rate": 4.135671114353239e-06,
      "loss": 0.8063099980354309,
      "step": 1337
    },
    {
      "epoch": 1.7358111807287548,
      "grad_norm": 0.680123507976532,
      "learning_rate": 4.128629903915004e-06,
      "loss": 0.671716034412384,
      "step": 1338
    },
    {
      "epoch": 1.7371089076788937,
      "grad_norm": 0.7551307678222656,
      "learning_rate": 4.121590475208071e-06,
      "loss": 0.7629652619361877,
      "step": 1339
    },
    {
      "epoch": 1.7384066346290326,
      "grad_norm": 0.7345308065414429,
      "learning_rate": 4.114552842626285e-06,
      "loss": 0.7416568994522095,
      "step": 1340
    },
    {
      "epoch": 1.7397043615791716,
      "grad_norm": 0.7788087725639343,
      "learning_rate": 4.107517020559827e-06,
      "loss": 0.7472527623176575,
      "step": 1341
    },
    {
      "epoch": 1.7410020885293105,
      "grad_norm": 0.7015308737754822,
      "learning_rate": 4.1004830233951696e-06,
      "loss": 0.7568359375,
      "step": 1342
    },
    {
      "epoch": 1.7422998154794493,
      "grad_norm": 0.7230126857757568,
      "learning_rate": 4.0934508655150585e-06,
      "loss": 0.7601989507675171,
      "step": 1343
    },
    {
      "epoch": 1.7435975424295882,
      "grad_norm": 0.7914838194847107,
      "learning_rate": 4.086420561298476e-06,
      "loss": 0.8481265306472778,
      "step": 1344
    },
    {
      "epoch": 1.744895269379727,
      "grad_norm": 0.7440057992935181,
      "learning_rate": 4.079392125120613e-06,
      "loss": 0.7708393335342407,
      "step": 1345
    },
    {
      "epoch": 1.746192996329866,
      "grad_norm": 0.6886211633682251,
      "learning_rate": 4.072365571352847e-06,
      "loss": 0.6713268160820007,
      "step": 1346
    },
    {
      "epoch": 1.7474907232800048,
      "grad_norm": 0.7026688456535339,
      "learning_rate": 4.065340914362697e-06,
      "loss": 0.7294650673866272,
      "step": 1347
    },
    {
      "epoch": 1.7487884502301436,
      "grad_norm": 0.7501193881034851,
      "learning_rate": 4.058318168513813e-06,
      "loss": 0.7362337708473206,
      "step": 1348
    },
    {
      "epoch": 1.7500861771802827,
      "grad_norm": 0.7246070504188538,
      "learning_rate": 4.05129734816593e-06,
      "loss": 0.7650129795074463,
      "step": 1349
    },
    {
      "epoch": 1.7513839041304216,
      "grad_norm": 0.7573635578155518,
      "learning_rate": 4.04427846767485e-06,
      "loss": 0.7343922853469849,
      "step": 1350
    },
    {
      "epoch": 1.7526816310805604,
      "grad_norm": 0.7699203491210938,
      "learning_rate": 4.037261541392405e-06,
      "loss": 0.7772881984710693,
      "step": 1351
    },
    {
      "epoch": 1.7539793580306995,
      "grad_norm": 0.7182446122169495,
      "learning_rate": 4.030246583666437e-06,
      "loss": 0.7065188884735107,
      "step": 1352
    },
    {
      "epoch": 1.7552770849808383,
      "grad_norm": 0.72418612241745,
      "learning_rate": 4.023233608840755e-06,
      "loss": 0.7692337036132812,
      "step": 1353
    },
    {
      "epoch": 1.7565748119309772,
      "grad_norm": 0.7533490657806396,
      "learning_rate": 4.016222631255121e-06,
      "loss": 0.7796276807785034,
      "step": 1354
    },
    {
      "epoch": 1.757872538881116,
      "grad_norm": 0.7471632957458496,
      "learning_rate": 4.0092136652452054e-06,
      "loss": 0.6910297274589539,
      "step": 1355
    },
    {
      "epoch": 1.759170265831255,
      "grad_norm": 0.7224332094192505,
      "learning_rate": 4.0022067251425736e-06,
      "loss": 0.8254155516624451,
      "step": 1356
    },
    {
      "epoch": 1.7604679927813938,
      "grad_norm": 0.7960043549537659,
      "learning_rate": 3.9952018252746424e-06,
      "loss": 0.7877900004386902,
      "step": 1357
    },
    {
      "epoch": 1.7617657197315326,
      "grad_norm": 0.7473553419113159,
      "learning_rate": 3.988198979964662e-06,
      "loss": 0.7509211897850037,
      "step": 1358
    },
    {
      "epoch": 1.7630634466816715,
      "grad_norm": 0.7365322709083557,
      "learning_rate": 3.981198203531673e-06,
      "loss": 0.7771757245063782,
      "step": 1359
    },
    {
      "epoch": 1.7643611736318106,
      "grad_norm": 0.7372840642929077,
      "learning_rate": 3.974199510290498e-06,
      "loss": 0.7108234763145447,
      "step": 1360
    },
    {
      "epoch": 1.7656589005819494,
      "grad_norm": 0.7740015387535095,
      "learning_rate": 3.967202914551688e-06,
      "loss": 0.7688103318214417,
      "step": 1361
    },
    {
      "epoch": 1.7669566275320885,
      "grad_norm": 0.7467756867408752,
      "learning_rate": 3.960208430621514e-06,
      "loss": 0.7408416867256165,
      "step": 1362
    },
    {
      "epoch": 1.7682543544822273,
      "grad_norm": 0.7169584035873413,
      "learning_rate": 3.953216072801922e-06,
      "loss": 0.7672545313835144,
      "step": 1363
    },
    {
      "epoch": 1.7695520814323662,
      "grad_norm": 0.7557554841041565,
      "learning_rate": 3.946225855390518e-06,
      "loss": 0.7705783843994141,
      "step": 1364
    },
    {
      "epoch": 1.770849808382505,
      "grad_norm": 0.773759663105011,
      "learning_rate": 3.9392377926805226e-06,
      "loss": 0.7769662737846375,
      "step": 1365
    },
    {
      "epoch": 1.772147535332644,
      "grad_norm": 0.8000966310501099,
      "learning_rate": 3.932251898960759e-06,
      "loss": 0.8040739297866821,
      "step": 1366
    },
    {
      "epoch": 1.7734452622827828,
      "grad_norm": 0.7309441566467285,
      "learning_rate": 3.925268188515611e-06,
      "loss": 0.782077968120575,
      "step": 1367
    },
    {
      "epoch": 1.7747429892329216,
      "grad_norm": 0.7728444337844849,
      "learning_rate": 3.918286675624998e-06,
      "loss": 0.784885585308075,
      "step": 1368
    },
    {
      "epoch": 1.7760407161830605,
      "grad_norm": 0.7408946752548218,
      "learning_rate": 3.911307374564346e-06,
      "loss": 0.656406044960022,
      "step": 1369
    },
    {
      "epoch": 1.7773384431331996,
      "grad_norm": 0.7466227412223816,
      "learning_rate": 3.904330299604562e-06,
      "loss": 0.7837309241294861,
      "step": 1370
    },
    {
      "epoch": 1.7786361700833384,
      "grad_norm": 0.7788135409355164,
      "learning_rate": 3.897355465011996e-06,
      "loss": 0.7761803269386292,
      "step": 1371
    },
    {
      "epoch": 1.7799338970334773,
      "grad_norm": 0.7268470525741577,
      "learning_rate": 3.89038288504842e-06,
      "loss": 0.6561414003372192,
      "step": 1372
    },
    {
      "epoch": 1.7812316239836163,
      "grad_norm": 0.704717218875885,
      "learning_rate": 3.883412573970995e-06,
      "loss": 0.7361816167831421,
      "step": 1373
    },
    {
      "epoch": 1.7825293509337552,
      "grad_norm": 0.724858820438385,
      "learning_rate": 3.876444546032242e-06,
      "loss": 0.7120435237884521,
      "step": 1374
    },
    {
      "epoch": 1.783827077883894,
      "grad_norm": 0.7575574517250061,
      "learning_rate": 3.8694788154800185e-06,
      "loss": 0.726084291934967,
      "step": 1375
    },
    {
      "epoch": 1.785124804834033,
      "grad_norm": 0.7479435801506042,
      "learning_rate": 3.862515396557476e-06,
      "loss": 0.7584422826766968,
      "step": 1376
    },
    {
      "epoch": 1.7864225317841718,
      "grad_norm": 0.7577545642852783,
      "learning_rate": 3.855554303503047e-06,
      "loss": 0.7728340029716492,
      "step": 1377
    },
    {
      "epoch": 1.7877202587343106,
      "grad_norm": 0.7183135747909546,
      "learning_rate": 3.848595550550401e-06,
      "loss": 0.7717944979667664,
      "step": 1378
    },
    {
      "epoch": 1.7890179856844495,
      "grad_norm": 0.7661336660385132,
      "learning_rate": 3.841639151928431e-06,
      "loss": 0.8036184310913086,
      "step": 1379
    },
    {
      "epoch": 1.7903157126345883,
      "grad_norm": 0.7206409573554993,
      "learning_rate": 3.834685121861208e-06,
      "loss": 0.7713849544525146,
      "step": 1380
    },
    {
      "epoch": 1.7903157126345883,
      "eval_loss": 0.7627881169319153,
      "eval_runtime": 141.991,
      "eval_samples_per_second": 36.566,
      "eval_steps_per_second": 9.141,
      "step": 1380
    },
    {
      "epoch": 1.7916134395847274,
      "grad_norm": 0.7314550280570984,
      "learning_rate": 3.827733474567966e-06,
      "loss": 0.7648483514785767,
      "step": 1381
    },
    {
      "epoch": 1.7929111665348663,
      "grad_norm": 0.7055811285972595,
      "learning_rate": 3.820784224263061e-06,
      "loss": 0.7016907334327698,
      "step": 1382
    },
    {
      "epoch": 1.7942088934850051,
      "grad_norm": 0.7724407911300659,
      "learning_rate": 3.8138373851559546e-06,
      "loss": 0.7384164333343506,
      "step": 1383
    },
    {
      "epoch": 1.7955066204351442,
      "grad_norm": 0.7461869120597839,
      "learning_rate": 3.8068929714511716e-06,
      "loss": 0.7625059485435486,
      "step": 1384
    },
    {
      "epoch": 1.796804347385283,
      "grad_norm": 0.7060610055923462,
      "learning_rate": 3.799950997348283e-06,
      "loss": 0.7166489362716675,
      "step": 1385
    },
    {
      "epoch": 1.798102074335422,
      "grad_norm": 0.7320153713226318,
      "learning_rate": 3.7930114770418654e-06,
      "loss": 0.674547553062439,
      "step": 1386
    },
    {
      "epoch": 1.7993998012855608,
      "grad_norm": 0.7460453510284424,
      "learning_rate": 3.7860744247214853e-06,
      "loss": 0.7099953293800354,
      "step": 1387
    },
    {
      "epoch": 1.8006975282356996,
      "grad_norm": 0.7489854693412781,
      "learning_rate": 3.7791398545716552e-06,
      "loss": 0.7331588268280029,
      "step": 1388
    },
    {
      "epoch": 1.8019952551858385,
      "grad_norm": 0.792487382888794,
      "learning_rate": 3.7722077807718193e-06,
      "loss": 0.8097270727157593,
      "step": 1389
    },
    {
      "epoch": 1.8032929821359773,
      "grad_norm": 0.7642346024513245,
      "learning_rate": 3.7652782174963107e-06,
      "loss": 0.736085832118988,
      "step": 1390
    },
    {
      "epoch": 1.8045907090861162,
      "grad_norm": 0.7119442224502563,
      "learning_rate": 3.758351178914336e-06,
      "loss": 0.7833069562911987,
      "step": 1391
    },
    {
      "epoch": 1.8058884360362553,
      "grad_norm": 0.7111310958862305,
      "learning_rate": 3.7514266791899324e-06,
      "loss": 0.6887351274490356,
      "step": 1392
    },
    {
      "epoch": 1.8071861629863941,
      "grad_norm": 0.7528010010719299,
      "learning_rate": 3.7445047324819517e-06,
      "loss": 0.8306547403335571,
      "step": 1393
    },
    {
      "epoch": 1.8084838899365332,
      "grad_norm": 0.7184954285621643,
      "learning_rate": 3.737585352944021e-06,
      "loss": 0.7377822995185852,
      "step": 1394
    },
    {
      "epoch": 1.809781616886672,
      "grad_norm": 0.7828357815742493,
      "learning_rate": 3.7306685547245225e-06,
      "loss": 0.7717545628547668,
      "step": 1395
    },
    {
      "epoch": 1.811079343836811,
      "grad_norm": 0.7384869456291199,
      "learning_rate": 3.7237543519665543e-06,
      "loss": 0.732141375541687,
      "step": 1396
    },
    {
      "epoch": 1.8123770707869498,
      "grad_norm": 0.7369826436042786,
      "learning_rate": 3.7168427588079153e-06,
      "loss": 0.7696800231933594,
      "step": 1397
    },
    {
      "epoch": 1.8136747977370886,
      "grad_norm": 0.6859968304634094,
      "learning_rate": 3.7099337893810593e-06,
      "loss": 0.7611477375030518,
      "step": 1398
    },
    {
      "epoch": 1.8149725246872275,
      "grad_norm": 0.7616201639175415,
      "learning_rate": 3.703027457813086e-06,
      "loss": 0.7165590524673462,
      "step": 1399
    },
    {
      "epoch": 1.8162702516373663,
      "grad_norm": 0.7798823118209839,
      "learning_rate": 3.696123778225691e-06,
      "loss": 0.8081139922142029,
      "step": 1400
    },
    {
      "epoch": 1.8175679785875052,
      "grad_norm": 0.736311674118042,
      "learning_rate": 3.6892227647351515e-06,
      "loss": 0.7922640442848206,
      "step": 1401
    },
    {
      "epoch": 1.8188657055376443,
      "grad_norm": 0.7466685771942139,
      "learning_rate": 3.6823244314522966e-06,
      "loss": 0.8202122449874878,
      "step": 1402
    },
    {
      "epoch": 1.8201634324877831,
      "grad_norm": 0.7271019220352173,
      "learning_rate": 3.67542879248247e-06,
      "loss": 0.7178265452384949,
      "step": 1403
    },
    {
      "epoch": 1.821461159437922,
      "grad_norm": 0.7438377737998962,
      "learning_rate": 3.668535861925509e-06,
      "loss": 0.7328163981437683,
      "step": 1404
    },
    {
      "epoch": 1.822758886388061,
      "grad_norm": 0.765792965888977,
      "learning_rate": 3.661645653875709e-06,
      "loss": 0.8025820255279541,
      "step": 1405
    },
    {
      "epoch": 1.8240566133382,
      "grad_norm": 0.7790916562080383,
      "learning_rate": 3.6547581824218057e-06,
      "loss": 0.6988804340362549,
      "step": 1406
    },
    {
      "epoch": 1.8253543402883388,
      "grad_norm": 0.7480283379554749,
      "learning_rate": 3.6478734616469324e-06,
      "loss": 0.7859005928039551,
      "step": 1407
    },
    {
      "epoch": 1.8266520672384776,
      "grad_norm": 0.7395159602165222,
      "learning_rate": 3.6409915056286017e-06,
      "loss": 0.7420051693916321,
      "step": 1408
    },
    {
      "epoch": 1.8279497941886165,
      "grad_norm": 0.7925701141357422,
      "learning_rate": 3.6341123284386694e-06,
      "loss": 0.7727212309837341,
      "step": 1409
    },
    {
      "epoch": 1.8292475211387553,
      "grad_norm": 0.804506242275238,
      "learning_rate": 3.627235944143315e-06,
      "loss": 0.8071032762527466,
      "step": 1410
    },
    {
      "epoch": 1.8305452480888942,
      "grad_norm": 0.7611510157585144,
      "learning_rate": 3.620362366803001e-06,
      "loss": 0.8068900108337402,
      "step": 1411
    },
    {
      "epoch": 1.831842975039033,
      "grad_norm": 0.7167320847511292,
      "learning_rate": 3.6134916104724573e-06,
      "loss": 0.72031170129776,
      "step": 1412
    },
    {
      "epoch": 1.8331407019891721,
      "grad_norm": 0.7381744384765625,
      "learning_rate": 3.606623689200637e-06,
      "loss": 0.8165542483329773,
      "step": 1413
    },
    {
      "epoch": 1.834438428939311,
      "grad_norm": 0.7563833594322205,
      "learning_rate": 3.599758617030704e-06,
      "loss": 0.7464354634284973,
      "step": 1414
    },
    {
      "epoch": 1.83573615588945,
      "grad_norm": 0.7599388360977173,
      "learning_rate": 3.5928964079999907e-06,
      "loss": 0.7536092400550842,
      "step": 1415
    },
    {
      "epoch": 1.837033882839589,
      "grad_norm": 0.748948335647583,
      "learning_rate": 3.5860370761399814e-06,
      "loss": 0.7360177636146545,
      "step": 1416
    },
    {
      "epoch": 1.8383316097897278,
      "grad_norm": 0.7518510818481445,
      "learning_rate": 3.5791806354762702e-06,
      "loss": 0.8265672922134399,
      "step": 1417
    },
    {
      "epoch": 1.8396293367398666,
      "grad_norm": 0.7552505731582642,
      "learning_rate": 3.572327100028545e-06,
      "loss": 0.7927607297897339,
      "step": 1418
    },
    {
      "epoch": 1.8409270636900055,
      "grad_norm": 0.7464638948440552,
      "learning_rate": 3.565476483810548e-06,
      "loss": 0.7352604866027832,
      "step": 1419
    },
    {
      "epoch": 1.8422247906401443,
      "grad_norm": 0.7622070908546448,
      "learning_rate": 3.55862880083006e-06,
      "loss": 0.7505785822868347,
      "step": 1420
    },
    {
      "epoch": 1.8435225175902832,
      "grad_norm": 0.7533313035964966,
      "learning_rate": 3.5517840650888564e-06,
      "loss": 0.7191623449325562,
      "step": 1421
    },
    {
      "epoch": 1.844820244540422,
      "grad_norm": 0.6952627897262573,
      "learning_rate": 3.544942290582691e-06,
      "loss": 0.6740431189537048,
      "step": 1422
    },
    {
      "epoch": 1.8461179714905611,
      "grad_norm": 0.7936224937438965,
      "learning_rate": 3.538103491301258e-06,
      "loss": 0.7648304104804993,
      "step": 1423
    },
    {
      "epoch": 1.8474156984407,
      "grad_norm": 0.7576262354850769,
      "learning_rate": 3.531267681228175e-06,
      "loss": 0.7879096865653992,
      "step": 1424
    },
    {
      "epoch": 1.8487134253908388,
      "grad_norm": 0.7862621545791626,
      "learning_rate": 3.5244348743409394e-06,
      "loss": 0.8113074898719788,
      "step": 1425
    },
    {
      "epoch": 1.850011152340978,
      "grad_norm": 0.744350254535675,
      "learning_rate": 3.517605084610917e-06,
      "loss": 0.8002179265022278,
      "step": 1426
    },
    {
      "epoch": 1.8513088792911168,
      "grad_norm": 0.7405024766921997,
      "learning_rate": 3.510778326003294e-06,
      "loss": 0.7966739535331726,
      "step": 1427
    },
    {
      "epoch": 1.8526066062412556,
      "grad_norm": 0.7228015065193176,
      "learning_rate": 3.5039546124770675e-06,
      "loss": 0.7743226289749146,
      "step": 1428
    },
    {
      "epoch": 1.8539043331913945,
      "grad_norm": 0.7095702886581421,
      "learning_rate": 3.4971339579850017e-06,
      "loss": 0.7427994608879089,
      "step": 1429
    },
    {
      "epoch": 1.8552020601415333,
      "grad_norm": 0.7417516708374023,
      "learning_rate": 3.4903163764736104e-06,
      "loss": 0.7906033396720886,
      "step": 1430
    },
    {
      "epoch": 1.8564997870916722,
      "grad_norm": 0.7653575539588928,
      "learning_rate": 3.4835018818831235e-06,
      "loss": 0.7567316293716431,
      "step": 1431
    },
    {
      "epoch": 1.857797514041811,
      "grad_norm": 0.7832647562026978,
      "learning_rate": 3.4766904881474535e-06,
      "loss": 0.8101804852485657,
      "step": 1432
    },
    {
      "epoch": 1.85909524099195,
      "grad_norm": 0.7221357226371765,
      "learning_rate": 3.4698822091941808e-06,
      "loss": 0.7595087289810181,
      "step": 1433
    },
    {
      "epoch": 1.860392967942089,
      "grad_norm": 0.7068979740142822,
      "learning_rate": 3.463077058944511e-06,
      "loss": 0.7103139162063599,
      "step": 1434
    },
    {
      "epoch": 1.8616906948922278,
      "grad_norm": 0.7623226046562195,
      "learning_rate": 3.456275051313255e-06,
      "loss": 0.7573720812797546,
      "step": 1435
    },
    {
      "epoch": 1.8629884218423667,
      "grad_norm": 0.7208322882652283,
      "learning_rate": 3.4494762002087934e-06,
      "loss": 0.7396218180656433,
      "step": 1436
    },
    {
      "epoch": 1.8642861487925058,
      "grad_norm": 0.7725145220756531,
      "learning_rate": 3.4426805195330605e-06,
      "loss": 0.8249402642250061,
      "step": 1437
    },
    {
      "epoch": 1.8655838757426446,
      "grad_norm": 0.7519071698188782,
      "learning_rate": 3.4358880231814983e-06,
      "loss": 0.7772566080093384,
      "step": 1438
    },
    {
      "epoch": 1.8668816026927835,
      "grad_norm": 0.7366478443145752,
      "learning_rate": 3.4290987250430486e-06,
      "loss": 0.7706719040870667,
      "step": 1439
    },
    {
      "epoch": 1.8681793296429223,
      "grad_norm": 0.7672808170318604,
      "learning_rate": 3.4223126390001025e-06,
      "loss": 0.8184731006622314,
      "step": 1440
    },
    {
      "epoch": 1.8694770565930612,
      "grad_norm": 0.7246538996696472,
      "learning_rate": 3.415529778928492e-06,
      "loss": 0.7725407481193542,
      "step": 1441
    },
    {
      "epoch": 1.8707747835432,
      "grad_norm": 0.7283217906951904,
      "learning_rate": 3.408750158697445e-06,
      "loss": 0.6893804669380188,
      "step": 1442
    },
    {
      "epoch": 1.872072510493339,
      "grad_norm": 0.7325669527053833,
      "learning_rate": 3.401973792169574e-06,
      "loss": 0.7318534255027771,
      "step": 1443
    },
    {
      "epoch": 1.8733702374434777,
      "grad_norm": 0.7319545745849609,
      "learning_rate": 3.39520069320083e-06,
      "loss": 0.7276860475540161,
      "step": 1444
    },
    {
      "epoch": 1.8746679643936168,
      "grad_norm": 0.7628985047340393,
      "learning_rate": 3.3884308756404873e-06,
      "loss": 0.760532021522522,
      "step": 1445
    },
    {
      "epoch": 1.8759656913437557,
      "grad_norm": 0.779719352722168,
      "learning_rate": 3.381664353331107e-06,
      "loss": 0.7530460953712463,
      "step": 1446
    },
    {
      "epoch": 1.8772634182938948,
      "grad_norm": 0.7683849930763245,
      "learning_rate": 3.3749011401085185e-06,
      "loss": 0.7530565857887268,
      "step": 1447
    },
    {
      "epoch": 1.8785611452440336,
      "grad_norm": 0.7684382796287537,
      "learning_rate": 3.3681412498017773e-06,
      "loss": 0.7397608757019043,
      "step": 1448
    },
    {
      "epoch": 1.8798588721941725,
      "grad_norm": 0.7616894841194153,
      "learning_rate": 3.361384696233152e-06,
      "loss": 0.7777236700057983,
      "step": 1449
    },
    {
      "epoch": 1.8811565991443113,
      "grad_norm": 0.771138608455658,
      "learning_rate": 3.354631493218081e-06,
      "loss": 0.7786779403686523,
      "step": 1450
    },
    {
      "epoch": 1.8824543260944502,
      "grad_norm": 0.7471014857292175,
      "learning_rate": 3.347881654565159e-06,
      "loss": 0.7542305588722229,
      "step": 1451
    },
    {
      "epoch": 1.883752053044589,
      "grad_norm": 0.7468750476837158,
      "learning_rate": 3.3411351940760924e-06,
      "loss": 0.7149202823638916,
      "step": 1452
    },
    {
      "epoch": 1.885049779994728,
      "grad_norm": 0.7417929768562317,
      "learning_rate": 3.3343921255456903e-06,
      "loss": 0.7159502506256104,
      "step": 1453
    },
    {
      "epoch": 1.8863475069448667,
      "grad_norm": 0.7415879368782043,
      "learning_rate": 3.3276524627618177e-06,
      "loss": 0.6990147829055786,
      "step": 1454
    },
    {
      "epoch": 1.8876452338950058,
      "grad_norm": 0.7450370788574219,
      "learning_rate": 3.3209162195053825e-06,
      "loss": 0.7316166162490845,
      "step": 1455
    },
    {
      "epoch": 1.8889429608451447,
      "grad_norm": 0.7386918067932129,
      "learning_rate": 3.314183409550293e-06,
      "loss": 0.7210360169410706,
      "step": 1456
    },
    {
      "epoch": 1.8902406877952835,
      "grad_norm": 0.7389383912086487,
      "learning_rate": 3.3074540466634454e-06,
      "loss": 0.7895267605781555,
      "step": 1457
    },
    {
      "epoch": 1.8915384147454226,
      "grad_norm": 0.7614936828613281,
      "learning_rate": 3.300728144604681e-06,
      "loss": 0.8177999258041382,
      "step": 1458
    },
    {
      "epoch": 1.8928361416955615,
      "grad_norm": 0.7167849540710449,
      "learning_rate": 3.294005717126767e-06,
      "loss": 0.6862820982933044,
      "step": 1459
    },
    {
      "epoch": 1.8941338686457003,
      "grad_norm": 0.7517713308334351,
      "learning_rate": 3.287286777975369e-06,
      "loss": 0.708810031414032,
      "step": 1460
    },
    {
      "epoch": 1.8954315955958392,
      "grad_norm": 0.7490662336349487,
      "learning_rate": 3.2805713408890134e-06,
      "loss": 0.733518123626709,
      "step": 1461
    },
    {
      "epoch": 1.896729322545978,
      "grad_norm": 0.7208186388015747,
      "learning_rate": 3.2738594195990725e-06,
      "loss": 0.7204444408416748,
      "step": 1462
    },
    {
      "epoch": 1.898027049496117,
      "grad_norm": 0.7559354901313782,
      "learning_rate": 3.267151027829725e-06,
      "loss": 0.7662622332572937,
      "step": 1463
    },
    {
      "epoch": 1.8993247764462557,
      "grad_norm": 0.7530090808868408,
      "learning_rate": 3.2604461792979346e-06,
      "loss": 0.7748895883560181,
      "step": 1464
    },
    {
      "epoch": 1.9006225033963946,
      "grad_norm": 0.7309477925300598,
      "learning_rate": 3.253744887713417e-06,
      "loss": 0.7579443454742432,
      "step": 1465
    },
    {
      "epoch": 1.9019202303465337,
      "grad_norm": 0.7468703985214233,
      "learning_rate": 3.2470471667786217e-06,
      "loss": 0.7491274476051331,
      "step": 1466
    },
    {
      "epoch": 1.9032179572966725,
      "grad_norm": 0.695270299911499,
      "learning_rate": 3.2403530301886897e-06,
      "loss": 0.7576136589050293,
      "step": 1467
    },
    {
      "epoch": 1.9045156842468114,
      "grad_norm": 0.7494797706604004,
      "learning_rate": 3.2336624916314385e-06,
      "loss": 0.7278751134872437,
      "step": 1468
    },
    {
      "epoch": 1.9058134111969505,
      "grad_norm": 0.7021095752716064,
      "learning_rate": 3.226975564787322e-06,
      "loss": 0.7184893488883972,
      "step": 1469
    },
    {
      "epoch": 1.9071111381470893,
      "grad_norm": 0.7631714940071106,
      "learning_rate": 3.2202922633294178e-06,
      "loss": 0.7584774494171143,
      "step": 1470
    },
    {
      "epoch": 1.9084088650972282,
      "grad_norm": 0.7526757121086121,
      "learning_rate": 3.2136126009233815e-06,
      "loss": 0.7316581010818481,
      "step": 1471
    },
    {
      "epoch": 1.909706592047367,
      "grad_norm": 0.8025736212730408,
      "learning_rate": 3.2069365912274364e-06,
      "loss": 0.7791706323623657,
      "step": 1472
    },
    {
      "epoch": 1.911004318997506,
      "grad_norm": 0.7570320963859558,
      "learning_rate": 3.2002642478923273e-06,
      "loss": 0.7199668288230896,
      "step": 1473
    },
    {
      "epoch": 1.9123020459476447,
      "grad_norm": 0.7405751347541809,
      "learning_rate": 3.1935955845613138e-06,
      "loss": 0.7453205585479736,
      "step": 1474
    },
    {
      "epoch": 1.9135997728977836,
      "grad_norm": 0.7665459513664246,
      "learning_rate": 3.1869306148701186e-06,
      "loss": 0.7860292792320251,
      "step": 1475
    },
    {
      "epoch": 1.9148974998479225,
      "grad_norm": 0.7148594856262207,
      "learning_rate": 3.1802693524469226e-06,
      "loss": 0.7073293328285217,
      "step": 1476
    },
    {
      "epoch": 1.9161952267980615,
      "grad_norm": 0.8274069428443909,
      "learning_rate": 3.1736118109123183e-06,
      "loss": 0.8189059495925903,
      "step": 1477
    },
    {
      "epoch": 1.9174929537482004,
      "grad_norm": 0.7543789148330688,
      "learning_rate": 3.1669580038792953e-06,
      "loss": 0.7467123866081238,
      "step": 1478
    },
    {
      "epoch": 1.9187906806983395,
      "grad_norm": 0.7482509613037109,
      "learning_rate": 3.1603079449532014e-06,
      "loss": 0.7409033179283142,
      "step": 1479
    },
    {
      "epoch": 1.9200884076484783,
      "grad_norm": 0.7407187819480896,
      "learning_rate": 3.1536616477317283e-06,
      "loss": 0.8021547198295593,
      "step": 1480
    },
    {
      "epoch": 1.9213861345986172,
      "grad_norm": 0.7338560223579407,
      "learning_rate": 3.147019125804869e-06,
      "loss": 0.7389906048774719,
      "step": 1481
    },
    {
      "epoch": 1.922683861548756,
      "grad_norm": 0.7023736238479614,
      "learning_rate": 3.140380392754901e-06,
      "loss": 0.6907408237457275,
      "step": 1482
    },
    {
      "epoch": 1.923981588498895,
      "grad_norm": 0.7529736757278442,
      "learning_rate": 3.13374546215635e-06,
      "loss": 0.7589248418807983,
      "step": 1483
    },
    {
      "epoch": 1.9252793154490337,
      "grad_norm": 0.7569683194160461,
      "learning_rate": 3.1271143475759745e-06,
      "loss": 0.8030762076377869,
      "step": 1484
    },
    {
      "epoch": 1.9265770423991726,
      "grad_norm": 0.7221974730491638,
      "learning_rate": 3.1204870625727216e-06,
      "loss": 0.7365057468414307,
      "step": 1485
    },
    {
      "epoch": 1.9278747693493115,
      "grad_norm": 0.7299054265022278,
      "learning_rate": 3.1138636206977147e-06,
      "loss": 0.7652139067649841,
      "step": 1486
    },
    {
      "epoch": 1.9291724962994505,
      "grad_norm": 0.7605259418487549,
      "learning_rate": 3.107244035494212e-06,
      "loss": 0.7461530566215515,
      "step": 1487
    },
    {
      "epoch": 1.9304702232495894,
      "grad_norm": 0.7828764319419861,
      "learning_rate": 3.100628320497592e-06,
      "loss": 0.7160741090774536,
      "step": 1488
    },
    {
      "epoch": 1.9317679501997282,
      "grad_norm": 0.7266847491264343,
      "learning_rate": 3.0940164892353197e-06,
      "loss": 0.7605788707733154,
      "step": 1489
    },
    {
      "epoch": 1.9330656771498673,
      "grad_norm": 0.7265951037406921,
      "learning_rate": 3.087408555226914e-06,
      "loss": 0.7256218791007996,
      "step": 1490
    },
    {
      "epoch": 1.9343634041000062,
      "grad_norm": 0.743874192237854,
      "learning_rate": 3.0808045319839285e-06,
      "loss": 0.7782658934593201,
      "step": 1491
    },
    {
      "epoch": 1.935661131050145,
      "grad_norm": 0.7700090408325195,
      "learning_rate": 3.0742044330099162e-06,
      "loss": 0.8411778211593628,
      "step": 1492
    },
    {
      "epoch": 1.936958858000284,
      "grad_norm": 0.7528886795043945,
      "learning_rate": 3.067608271800414e-06,
      "loss": 0.7456690669059753,
      "step": 1493
    },
    {
      "epoch": 1.9382565849504227,
      "grad_norm": 0.7644984722137451,
      "learning_rate": 3.0610160618428987e-06,
      "loss": 0.8341564536094666,
      "step": 1494
    },
    {
      "epoch": 1.9395543119005616,
      "grad_norm": 0.7603514790534973,
      "learning_rate": 3.054427816616773e-06,
      "loss": 0.7421169281005859,
      "step": 1495
    },
    {
      "epoch": 1.9408520388507005,
      "grad_norm": 0.750495195388794,
      "learning_rate": 3.0478435495933273e-06,
      "loss": 0.7318063974380493,
      "step": 1496
    },
    {
      "epoch": 1.9421497658008393,
      "grad_norm": 0.7440663576126099,
      "learning_rate": 3.0412632742357263e-06,
      "loss": 0.7301744222640991,
      "step": 1497
    },
    {
      "epoch": 1.9434474927509784,
      "grad_norm": 0.7638893723487854,
      "learning_rate": 3.0346870039989618e-06,
      "loss": 0.807081401348114,
      "step": 1498
    },
    {
      "epoch": 1.9447452197011172,
      "grad_norm": 0.7054970860481262,
      "learning_rate": 3.028114752329848e-06,
      "loss": 0.6948739886283875,
      "step": 1499
    },
    {
      "epoch": 1.9460429466512563,
      "grad_norm": 0.7606166005134583,
      "learning_rate": 3.0215465326669724e-06,
      "loss": 0.7260342240333557,
      "step": 1500
    },
    {
      "epoch": 1.9473406736013952,
      "grad_norm": 0.7610599398612976,
      "learning_rate": 3.0149823584406834e-06,
      "loss": 0.7050684690475464,
      "step": 1501
    },
    {
      "epoch": 1.948638400551534,
      "grad_norm": 0.7257749438285828,
      "learning_rate": 3.008422243073053e-06,
      "loss": 0.7687860131263733,
      "step": 1502
    },
    {
      "epoch": 1.9499361275016729,
      "grad_norm": 0.7967380285263062,
      "learning_rate": 3.001866199977861e-06,
      "loss": 0.7173135280609131,
      "step": 1503
    },
    {
      "epoch": 1.9512338544518117,
      "grad_norm": 0.7631365656852722,
      "learning_rate": 2.995314242560553e-06,
      "loss": 0.6869869232177734,
      "step": 1504
    },
    {
      "epoch": 1.9525315814019506,
      "grad_norm": 0.7799795269966125,
      "learning_rate": 2.988766384218225e-06,
      "loss": 0.8173317313194275,
      "step": 1505
    },
    {
      "epoch": 1.9538293083520895,
      "grad_norm": 0.7079163193702698,
      "learning_rate": 2.982222638339588e-06,
      "loss": 0.7003392577171326,
      "step": 1506
    },
    {
      "epoch": 1.9551270353022283,
      "grad_norm": 0.7494550347328186,
      "learning_rate": 2.9756830183049502e-06,
      "loss": 0.7491193413734436,
      "step": 1507
    },
    {
      "epoch": 1.9564247622523674,
      "grad_norm": 0.7550981640815735,
      "learning_rate": 2.969147537486175e-06,
      "loss": 0.6874207258224487,
      "step": 1508
    },
    {
      "epoch": 1.9577224892025062,
      "grad_norm": 0.7690152525901794,
      "learning_rate": 2.962616209246669e-06,
      "loss": 0.7523796558380127,
      "step": 1509
    },
    {
      "epoch": 1.959020216152645,
      "grad_norm": 0.7652460336685181,
      "learning_rate": 2.956089046941344e-06,
      "loss": 0.7440061569213867,
      "step": 1510
    },
    {
      "epoch": 1.9603179431027842,
      "grad_norm": 0.7546727657318115,
      "learning_rate": 2.9495660639165967e-06,
      "loss": 0.6781274676322937,
      "step": 1511
    },
    {
      "epoch": 1.961615670052923,
      "grad_norm": 0.7501972913742065,
      "learning_rate": 2.9430472735102733e-06,
      "loss": 0.8312241435050964,
      "step": 1512
    },
    {
      "epoch": 1.9629133970030619,
      "grad_norm": 0.7584221959114075,
      "learning_rate": 2.9365326890516543e-06,
      "loss": 0.7617495059967041,
      "step": 1513
    },
    {
      "epoch": 1.9642111239532007,
      "grad_norm": 0.7630670070648193,
      "learning_rate": 2.9300223238614135e-06,
      "loss": 0.7445218563079834,
      "step": 1514
    },
    {
      "epoch": 1.9655088509033396,
      "grad_norm": 0.7560063004493713,
      "learning_rate": 2.923516191251601e-06,
      "loss": 0.7189357280731201,
      "step": 1515
    },
    {
      "epoch": 1.9668065778534785,
      "grad_norm": 0.7539728283882141,
      "learning_rate": 2.917014304525609e-06,
      "loss": 0.7735335826873779,
      "step": 1516
    },
    {
      "epoch": 1.9681043048036173,
      "grad_norm": 0.7924235463142395,
      "learning_rate": 2.91051667697815e-06,
      "loss": 0.7291359305381775,
      "step": 1517
    },
    {
      "epoch": 1.9694020317537562,
      "grad_norm": 0.7921690344810486,
      "learning_rate": 2.904023321895234e-06,
      "loss": 0.7890439629554749,
      "step": 1518
    },
    {
      "epoch": 1.9706997587038952,
      "grad_norm": 0.7306675314903259,
      "learning_rate": 2.8975342525541217e-06,
      "loss": 0.7303510904312134,
      "step": 1519
    },
    {
      "epoch": 1.971997485654034,
      "grad_norm": 0.7390612959861755,
      "learning_rate": 2.8910494822233203e-06,
      "loss": 0.738096296787262,
      "step": 1520
    },
    {
      "epoch": 1.973295212604173,
      "grad_norm": 0.7136329412460327,
      "learning_rate": 2.8845690241625437e-06,
      "loss": 0.693221390247345,
      "step": 1521
    },
    {
      "epoch": 1.974592939554312,
      "grad_norm": 0.7404115796089172,
      "learning_rate": 2.878092891622688e-06,
      "loss": 0.7287771105766296,
      "step": 1522
    },
    {
      "epoch": 1.9758906665044509,
      "grad_norm": 0.7220072746276855,
      "learning_rate": 2.871621097845806e-06,
      "loss": 0.7384371161460876,
      "step": 1523
    },
    {
      "epoch": 1.9771883934545897,
      "grad_norm": 0.7408435344696045,
      "learning_rate": 2.865153656065076e-06,
      "loss": 0.8272619247436523,
      "step": 1524
    },
    {
      "epoch": 1.9784861204047286,
      "grad_norm": 0.7614798545837402,
      "learning_rate": 2.8586905795047813e-06,
      "loss": 0.7376547455787659,
      "step": 1525
    },
    {
      "epoch": 1.9797838473548675,
      "grad_norm": 0.7582762837409973,
      "learning_rate": 2.8522318813802796e-06,
      "loss": 0.7691409587860107,
      "step": 1526
    },
    {
      "epoch": 1.9810815743050063,
      "grad_norm": 0.7164722681045532,
      "learning_rate": 2.8457775748979664e-06,
      "loss": 0.8395357728004456,
      "step": 1527
    },
    {
      "epoch": 1.9823793012551452,
      "grad_norm": 0.7436051964759827,
      "learning_rate": 2.8393276732552745e-06,
      "loss": 0.7708083391189575,
      "step": 1528
    },
    {
      "epoch": 1.983677028205284,
      "grad_norm": 0.72344970703125,
      "learning_rate": 2.8328821896406132e-06,
      "loss": 0.7172407507896423,
      "step": 1529
    },
    {
      "epoch": 1.984974755155423,
      "grad_norm": 0.7652667760848999,
      "learning_rate": 2.826441137233368e-06,
      "loss": 0.8026193976402283,
      "step": 1530
    },
    {
      "epoch": 1.986272482105562,
      "grad_norm": 0.744799792766571,
      "learning_rate": 2.8200045292038596e-06,
      "loss": 0.7224990725517273,
      "step": 1531
    },
    {
      "epoch": 1.987570209055701,
      "grad_norm": 0.7798324227333069,
      "learning_rate": 2.8135723787133233e-06,
      "loss": 0.8296301960945129,
      "step": 1532
    },
    {
      "epoch": 1.9888679360058399,
      "grad_norm": 0.7641241550445557,
      "learning_rate": 2.8071446989138786e-06,
      "loss": 0.8053862452507019,
      "step": 1533
    },
    {
      "epoch": 1.9901656629559787,
      "grad_norm": 0.7807287573814392,
      "learning_rate": 2.800721502948506e-06,
      "loss": 0.8045886754989624,
      "step": 1534
    },
    {
      "epoch": 1.9914633899061176,
      "grad_norm": 0.7817100882530212,
      "learning_rate": 2.7943028039510085e-06,
      "loss": 0.7601321339607239,
      "step": 1535
    },
    {
      "epoch": 1.9927611168562565,
      "grad_norm": 0.7499108910560608,
      "learning_rate": 2.78788861504601e-06,
      "loss": 0.712813138961792,
      "step": 1536
    },
    {
      "epoch": 1.9940588438063953,
      "grad_norm": 0.7500565052032471,
      "learning_rate": 2.7814789493488947e-06,
      "loss": 0.784660816192627,
      "step": 1537
    },
    {
      "epoch": 1.9953565707565342,
      "grad_norm": 0.7742734551429749,
      "learning_rate": 2.7750738199658157e-06,
      "loss": 0.790881872177124,
      "step": 1538
    },
    {
      "epoch": 1.996654297706673,
      "grad_norm": 0.7405888438224792,
      "learning_rate": 2.7686732399936343e-06,
      "loss": 0.75714111328125,
      "step": 1539
    },
    {
      "epoch": 1.997952024656812,
      "grad_norm": 0.7482635974884033,
      "learning_rate": 2.762277222519919e-06,
      "loss": 0.8016012907028198,
      "step": 1540
    },
    {
      "epoch": 1.999249751606951,
      "grad_norm": 0.7615522146224976,
      "learning_rate": 2.7558857806229066e-06,
      "loss": 0.7230945825576782,
      "step": 1541
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.9185896515846252,
      "learning_rate": 2.749498927371478e-06,
      "loss": 0.6910619139671326,
      "step": 1542
    },
    {
      "epoch": 2.001297726950139,
      "grad_norm": 0.8090460300445557,
      "learning_rate": 2.7431166758251317e-06,
      "loss": 0.7160417437553406,
      "step": 1543
    },
    {
      "epoch": 2.0025954539002777,
      "grad_norm": 0.821662425994873,
      "learning_rate": 2.7367390390339565e-06,
      "loss": 0.8003557324409485,
      "step": 1544
    },
    {
      "epoch": 2.0038931808504166,
      "grad_norm": 0.8417342305183411,
      "learning_rate": 2.730366030038606e-06,
      "loss": 0.6696601510047913,
      "step": 1545
    },
    {
      "epoch": 2.0051909078005554,
      "grad_norm": 0.7647730112075806,
      "learning_rate": 2.72399766187027e-06,
      "loss": 0.7535327076911926,
      "step": 1546
    },
    {
      "epoch": 2.0064886347506943,
      "grad_norm": 0.7831124663352966,
      "learning_rate": 2.7176339475506515e-06,
      "loss": 0.6865320801734924,
      "step": 1547
    },
    {
      "epoch": 2.0077863617008336,
      "grad_norm": 0.8027868866920471,
      "learning_rate": 2.7112749000919304e-06,
      "loss": 0.7301362156867981,
      "step": 1548
    },
    {
      "epoch": 2.0090840886509724,
      "grad_norm": 0.7517098784446716,
      "learning_rate": 2.704920532496756e-06,
      "loss": 0.7181136012077332,
      "step": 1549
    },
    {
      "epoch": 2.0103818156011113,
      "grad_norm": 0.7687897086143494,
      "learning_rate": 2.698570857758195e-06,
      "loss": 0.6889755725860596,
      "step": 1550
    },
    {
      "epoch": 2.01167954255125,
      "grad_norm": 0.769771933555603,
      "learning_rate": 2.692225888859732e-06,
      "loss": 0.8039973378181458,
      "step": 1551
    },
    {
      "epoch": 2.012977269501389,
      "grad_norm": 0.7931247353553772,
      "learning_rate": 2.685885638775216e-06,
      "loss": 0.7709112167358398,
      "step": 1552
    },
    {
      "epoch": 2.014274996451528,
      "grad_norm": 0.7785104513168335,
      "learning_rate": 2.6795501204688586e-06,
      "loss": 0.7515279650688171,
      "step": 1553
    },
    {
      "epoch": 2.0155727234016667,
      "grad_norm": 0.7584104537963867,
      "learning_rate": 2.6732193468951882e-06,
      "loss": 0.751779317855835,
      "step": 1554
    },
    {
      "epoch": 2.0168704503518056,
      "grad_norm": 0.7819312214851379,
      "learning_rate": 2.666893330999035e-06,
      "loss": 0.6674883961677551,
      "step": 1555
    },
    {
      "epoch": 2.0181681773019444,
      "grad_norm": 0.7927038669586182,
      "learning_rate": 2.6605720857155017e-06,
      "loss": 0.6830568909645081,
      "step": 1556
    },
    {
      "epoch": 2.0194659042520833,
      "grad_norm": 0.8502561450004578,
      "learning_rate": 2.654255623969936e-06,
      "loss": 0.734710156917572,
      "step": 1557
    },
    {
      "epoch": 2.0207636312022226,
      "grad_norm": 0.7859963178634644,
      "learning_rate": 2.647943958677897e-06,
      "loss": 0.7186540365219116,
      "step": 1558
    },
    {
      "epoch": 2.0220613581523614,
      "grad_norm": 0.7992748618125916,
      "learning_rate": 2.6416371027451514e-06,
      "loss": 0.6743420362472534,
      "step": 1559
    },
    {
      "epoch": 2.0233590851025003,
      "grad_norm": 0.8139959573745728,
      "learning_rate": 2.635335069067617e-06,
      "loss": 0.7734099626541138,
      "step": 1560
    },
    {
      "epoch": 2.024656812052639,
      "grad_norm": 0.7677357196807861,
      "learning_rate": 2.62903787053136e-06,
      "loss": 0.7552749514579773,
      "step": 1561
    },
    {
      "epoch": 2.025954539002778,
      "grad_norm": 0.744732141494751,
      "learning_rate": 2.6227455200125575e-06,
      "loss": 0.6963829398155212,
      "step": 1562
    },
    {
      "epoch": 2.027252265952917,
      "grad_norm": 0.7503330111503601,
      "learning_rate": 2.6164580303774733e-06,
      "loss": 0.7289599776268005,
      "step": 1563
    },
    {
      "epoch": 2.0285499929030557,
      "grad_norm": 0.7900810241699219,
      "learning_rate": 2.6101754144824327e-06,
      "loss": 0.6874979138374329,
      "step": 1564
    },
    {
      "epoch": 2.0298477198531946,
      "grad_norm": 0.7571580410003662,
      "learning_rate": 2.603897685173794e-06,
      "loss": 0.7547987103462219,
      "step": 1565
    },
    {
      "epoch": 2.0311454468033334,
      "grad_norm": 0.7505063414573669,
      "learning_rate": 2.5976248552879264e-06,
      "loss": 0.7211095690727234,
      "step": 1566
    },
    {
      "epoch": 2.0324431737534723,
      "grad_norm": 0.7814156413078308,
      "learning_rate": 2.5913569376511806e-06,
      "loss": 0.6726822257041931,
      "step": 1567
    },
    {
      "epoch": 2.033740900703611,
      "grad_norm": 0.7896116971969604,
      "learning_rate": 2.5850939450798553e-06,
      "loss": 0.7849063277244568,
      "step": 1568
    },
    {
      "epoch": 2.0350386276537504,
      "grad_norm": 0.7852982878684998,
      "learning_rate": 2.5788358903801926e-06,
      "loss": 0.7045016884803772,
      "step": 1569
    },
    {
      "epoch": 2.0363363546038893,
      "grad_norm": 0.726755678653717,
      "learning_rate": 2.572582786348326e-06,
      "loss": 0.6949452757835388,
      "step": 1570
    },
    {
      "epoch": 2.037634081554028,
      "grad_norm": 0.7358665466308594,
      "learning_rate": 2.566334645770272e-06,
      "loss": 0.6689255237579346,
      "step": 1571
    },
    {
      "epoch": 2.038931808504167,
      "grad_norm": 0.7562324404716492,
      "learning_rate": 2.5600914814218963e-06,
      "loss": 0.6774943470954895,
      "step": 1572
    },
    {
      "epoch": 2.040229535454306,
      "grad_norm": 0.7547875642776489,
      "learning_rate": 2.553853306068888e-06,
      "loss": 0.7316610217094421,
      "step": 1573
    },
    {
      "epoch": 2.0415272624044447,
      "grad_norm": 0.7495048642158508,
      "learning_rate": 2.547620132466743e-06,
      "loss": 0.6789397597312927,
      "step": 1574
    },
    {
      "epoch": 2.0428249893545836,
      "grad_norm": 0.7583444714546204,
      "learning_rate": 2.541391973360717e-06,
      "loss": 0.6931465864181519,
      "step": 1575
    },
    {
      "epoch": 2.0441227163047224,
      "grad_norm": 0.7334174513816833,
      "learning_rate": 2.535168841485821e-06,
      "loss": 0.7701333165168762,
      "step": 1576
    },
    {
      "epoch": 2.0454204432548613,
      "grad_norm": 0.7112381458282471,
      "learning_rate": 2.5289507495667864e-06,
      "loss": 0.6884704232215881,
      "step": 1577
    },
    {
      "epoch": 2.046718170205,
      "grad_norm": 0.7634471654891968,
      "learning_rate": 2.5227377103180353e-06,
      "loss": 0.6711394786834717,
      "step": 1578
    },
    {
      "epoch": 2.048015897155139,
      "grad_norm": 0.7438578009605408,
      "learning_rate": 2.516529736443661e-06,
      "loss": 0.7194059491157532,
      "step": 1579
    },
    {
      "epoch": 2.0493136241052783,
      "grad_norm": 0.7477583289146423,
      "learning_rate": 2.5103268406374002e-06,
      "loss": 0.6969068050384521,
      "step": 1580
    },
    {
      "epoch": 2.050611351055417,
      "grad_norm": 0.7318229079246521,
      "learning_rate": 2.504129035582601e-06,
      "loss": 0.712550938129425,
      "step": 1581
    },
    {
      "epoch": 2.051909078005556,
      "grad_norm": 0.7574477195739746,
      "learning_rate": 2.497936333952212e-06,
      "loss": 0.7607018351554871,
      "step": 1582
    },
    {
      "epoch": 2.053206804955695,
      "grad_norm": 0.8140477538108826,
      "learning_rate": 2.491748748408735e-06,
      "loss": 0.6738405227661133,
      "step": 1583
    },
    {
      "epoch": 2.0545045319058337,
      "grad_norm": 0.7556366324424744,
      "learning_rate": 2.485566291604219e-06,
      "loss": 0.68988436460495,
      "step": 1584
    },
    {
      "epoch": 2.0558022588559726,
      "grad_norm": 0.7742814421653748,
      "learning_rate": 2.4793889761802225e-06,
      "loss": 0.708181619644165,
      "step": 1585
    },
    {
      "epoch": 2.0570999858061114,
      "grad_norm": 0.8035808205604553,
      "learning_rate": 2.4732168147677927e-06,
      "loss": 0.6929728984832764,
      "step": 1586
    },
    {
      "epoch": 2.0583977127562503,
      "grad_norm": 0.8000681400299072,
      "learning_rate": 2.467049819987437e-06,
      "loss": 0.7274913787841797,
      "step": 1587
    },
    {
      "epoch": 2.059695439706389,
      "grad_norm": 0.7306155562400818,
      "learning_rate": 2.460888004449099e-06,
      "loss": 0.6182838678359985,
      "step": 1588
    },
    {
      "epoch": 2.060993166656528,
      "grad_norm": 0.7635526061058044,
      "learning_rate": 2.454731380752132e-06,
      "loss": 0.7412324547767639,
      "step": 1589
    },
    {
      "epoch": 2.0622908936066673,
      "grad_norm": 0.7367591261863708,
      "learning_rate": 2.4485799614852755e-06,
      "loss": 0.6950958371162415,
      "step": 1590
    },
    {
      "epoch": 2.063588620556806,
      "grad_norm": 0.7613998055458069,
      "learning_rate": 2.442433759226619e-06,
      "loss": 0.7596891522407532,
      "step": 1591
    },
    {
      "epoch": 2.064886347506945,
      "grad_norm": 0.7433957457542419,
      "learning_rate": 2.4362927865435975e-06,
      "loss": 0.7979118227958679,
      "step": 1592
    },
    {
      "epoch": 2.066184074457084,
      "grad_norm": 0.8046156764030457,
      "learning_rate": 2.4301570559929405e-06,
      "loss": 0.773604691028595,
      "step": 1593
    },
    {
      "epoch": 2.0674818014072227,
      "grad_norm": 0.7190860509872437,
      "learning_rate": 2.4240265801206665e-06,
      "loss": 0.6870064735412598,
      "step": 1594
    },
    {
      "epoch": 2.0687795283573616,
      "grad_norm": 0.7501992583274841,
      "learning_rate": 2.4179013714620456e-06,
      "loss": 0.6922681927680969,
      "step": 1595
    },
    {
      "epoch": 2.0700772553075004,
      "grad_norm": 0.7564000487327576,
      "learning_rate": 2.4117814425415803e-06,
      "loss": 0.7309989929199219,
      "step": 1596
    },
    {
      "epoch": 2.0713749822576393,
      "grad_norm": 0.772293210029602,
      "learning_rate": 2.4056668058729766e-06,
      "loss": 0.7206372618675232,
      "step": 1597
    },
    {
      "epoch": 2.072672709207778,
      "grad_norm": 0.787385880947113,
      "learning_rate": 2.399557473959119e-06,
      "loss": 0.7057082653045654,
      "step": 1598
    },
    {
      "epoch": 2.073970436157917,
      "grad_norm": 0.836436927318573,
      "learning_rate": 2.3934534592920416e-06,
      "loss": 0.7267694473266602,
      "step": 1599
    },
    {
      "epoch": 2.075268163108056,
      "grad_norm": 0.7650021314620972,
      "learning_rate": 2.3873547743529157e-06,
      "loss": 0.7323647141456604,
      "step": 1600
    },
    {
      "epoch": 2.076565890058195,
      "grad_norm": 0.7755290269851685,
      "learning_rate": 2.3812614316120003e-06,
      "loss": 0.7333980798721313,
      "step": 1601
    },
    {
      "epoch": 2.077863617008334,
      "grad_norm": 0.7668667435646057,
      "learning_rate": 2.375173443528646e-06,
      "loss": 0.6830537915229797,
      "step": 1602
    },
    {
      "epoch": 2.079161343958473,
      "grad_norm": 0.8191787600517273,
      "learning_rate": 2.3690908225512464e-06,
      "loss": 0.6813746690750122,
      "step": 1603
    },
    {
      "epoch": 2.0804590709086117,
      "grad_norm": 0.7462801337242126,
      "learning_rate": 2.363013581117217e-06,
      "loss": 0.7685893177986145,
      "step": 1604
    },
    {
      "epoch": 2.0817567978587506,
      "grad_norm": 0.7931850552558899,
      "learning_rate": 2.356941731652986e-06,
      "loss": 0.7468531131744385,
      "step": 1605
    },
    {
      "epoch": 2.0830545248088894,
      "grad_norm": 0.7673047780990601,
      "learning_rate": 2.3508752865739425e-06,
      "loss": 0.7278560996055603,
      "step": 1606
    },
    {
      "epoch": 2.0843522517590283,
      "grad_norm": 0.7797254323959351,
      "learning_rate": 2.344814258284433e-06,
      "loss": 0.7439032793045044,
      "step": 1607
    },
    {
      "epoch": 2.085649978709167,
      "grad_norm": 0.7476287484169006,
      "learning_rate": 2.3387586591777274e-06,
      "loss": 0.7101130485534668,
      "step": 1608
    },
    {
      "epoch": 2.086947705659306,
      "grad_norm": 0.7510504722595215,
      "learning_rate": 2.3327085016359912e-06,
      "loss": 0.6794774532318115,
      "step": 1609
    },
    {
      "epoch": 2.088245432609445,
      "grad_norm": 0.7459760308265686,
      "learning_rate": 2.3266637980302677e-06,
      "loss": 0.7223816514015198,
      "step": 1610
    },
    {
      "epoch": 2.088245432609445,
      "eval_loss": 0.7627704739570618,
      "eval_runtime": 140.2916,
      "eval_samples_per_second": 37.009,
      "eval_steps_per_second": 9.252,
      "step": 1610
    },
    {
      "epoch": 2.089543159559584,
      "grad_norm": 0.7484630346298218,
      "learning_rate": 2.320624560720446e-06,
      "loss": 0.718350350856781,
      "step": 1611
    },
    {
      "epoch": 2.090840886509723,
      "grad_norm": 0.7986156940460205,
      "learning_rate": 2.314590802055232e-06,
      "loss": 0.7352898120880127,
      "step": 1612
    },
    {
      "epoch": 2.092138613459862,
      "grad_norm": 0.7424448728561401,
      "learning_rate": 2.308562534372144e-06,
      "loss": 0.6307387351989746,
      "step": 1613
    },
    {
      "epoch": 2.0934363404100007,
      "grad_norm": 0.8542348146438599,
      "learning_rate": 2.3025397699974555e-06,
      "loss": 0.7578550577163696,
      "step": 1614
    },
    {
      "epoch": 2.0947340673601396,
      "grad_norm": 0.7494445443153381,
      "learning_rate": 2.296522521246202e-06,
      "loss": 0.7531486749649048,
      "step": 1615
    },
    {
      "epoch": 2.0960317943102784,
      "grad_norm": 0.7711408138275146,
      "learning_rate": 2.290510800422129e-06,
      "loss": 0.7104783058166504,
      "step": 1616
    },
    {
      "epoch": 2.0973295212604173,
      "grad_norm": 0.7710621953010559,
      "learning_rate": 2.284504619817687e-06,
      "loss": 0.7426379919052124,
      "step": 1617
    },
    {
      "epoch": 2.098627248210556,
      "grad_norm": 0.755355715751648,
      "learning_rate": 2.2785039917139933e-06,
      "loss": 0.7569924592971802,
      "step": 1618
    },
    {
      "epoch": 2.099924975160695,
      "grad_norm": 0.8246397972106934,
      "learning_rate": 2.272508928380815e-06,
      "loss": 0.6701229810714722,
      "step": 1619
    },
    {
      "epoch": 2.101222702110834,
      "grad_norm": 0.7797556519508362,
      "learning_rate": 2.2665194420765386e-06,
      "loss": 0.8236826658248901,
      "step": 1620
    },
    {
      "epoch": 2.1025204290609727,
      "grad_norm": 0.7621787190437317,
      "learning_rate": 2.260535545048149e-06,
      "loss": 0.6628905534744263,
      "step": 1621
    },
    {
      "epoch": 2.103818156011112,
      "grad_norm": 0.7577031850814819,
      "learning_rate": 2.2545572495311966e-06,
      "loss": 0.6984103322029114,
      "step": 1622
    },
    {
      "epoch": 2.105115882961251,
      "grad_norm": 0.7571961283683777,
      "learning_rate": 2.2485845677497897e-06,
      "loss": 0.7148957848548889,
      "step": 1623
    },
    {
      "epoch": 2.1064136099113897,
      "grad_norm": 0.7815601825714111,
      "learning_rate": 2.2426175119165435e-06,
      "loss": 0.799562394618988,
      "step": 1624
    },
    {
      "epoch": 2.1077113368615286,
      "grad_norm": 0.8278639316558838,
      "learning_rate": 2.2366560942325833e-06,
      "loss": 0.7186278104782104,
      "step": 1625
    },
    {
      "epoch": 2.1090090638116674,
      "grad_norm": 0.7949169874191284,
      "learning_rate": 2.230700326887495e-06,
      "loss": 0.6756667494773865,
      "step": 1626
    },
    {
      "epoch": 2.1103067907618063,
      "grad_norm": 0.7379303574562073,
      "learning_rate": 2.2247502220593164e-06,
      "loss": 0.7049471139907837,
      "step": 1627
    },
    {
      "epoch": 2.111604517711945,
      "grad_norm": 0.7553898692131042,
      "learning_rate": 2.218805791914507e-06,
      "loss": 0.7118646502494812,
      "step": 1628
    },
    {
      "epoch": 2.112902244662084,
      "grad_norm": 0.7354833483695984,
      "learning_rate": 2.21286704860792e-06,
      "loss": 0.6721919775009155,
      "step": 1629
    },
    {
      "epoch": 2.114199971612223,
      "grad_norm": 0.7323917746543884,
      "learning_rate": 2.2069340042827846e-06,
      "loss": 0.6935778260231018,
      "step": 1630
    },
    {
      "epoch": 2.1154976985623617,
      "grad_norm": 0.7529373168945312,
      "learning_rate": 2.2010066710706734e-06,
      "loss": 0.6505716443061829,
      "step": 1631
    },
    {
      "epoch": 2.1167954255125006,
      "grad_norm": 0.7721433639526367,
      "learning_rate": 2.1950850610914824e-06,
      "loss": 0.6882133483886719,
      "step": 1632
    },
    {
      "epoch": 2.11809315246264,
      "grad_norm": 0.725283145904541,
      "learning_rate": 2.1891691864534065e-06,
      "loss": 0.6969849467277527,
      "step": 1633
    },
    {
      "epoch": 2.1193908794127787,
      "grad_norm": 0.8012808561325073,
      "learning_rate": 2.1832590592529128e-06,
      "loss": 0.7592130303382874,
      "step": 1634
    },
    {
      "epoch": 2.1206886063629176,
      "grad_norm": 0.7490873336791992,
      "learning_rate": 2.1773546915747103e-06,
      "loss": 0.6869245767593384,
      "step": 1635
    },
    {
      "epoch": 2.1219863333130564,
      "grad_norm": 0.7902940511703491,
      "learning_rate": 2.1714560954917437e-06,
      "loss": 0.7730790972709656,
      "step": 1636
    },
    {
      "epoch": 2.1232840602631953,
      "grad_norm": 0.7704285979270935,
      "learning_rate": 2.165563283065142e-06,
      "loss": 0.7789285182952881,
      "step": 1637
    },
    {
      "epoch": 2.124581787213334,
      "grad_norm": 0.783291757106781,
      "learning_rate": 2.159676266344222e-06,
      "loss": 0.7162942290306091,
      "step": 1638
    },
    {
      "epoch": 2.125879514163473,
      "grad_norm": 0.7617123126983643,
      "learning_rate": 2.1537950573664372e-06,
      "loss": 0.7395819425582886,
      "step": 1639
    },
    {
      "epoch": 2.127177241113612,
      "grad_norm": 0.7524591684341431,
      "learning_rate": 2.1479196681573745e-06,
      "loss": 0.7014380097389221,
      "step": 1640
    },
    {
      "epoch": 2.1284749680637507,
      "grad_norm": 0.7793992757797241,
      "learning_rate": 2.142050110730716e-06,
      "loss": 0.6641875505447388,
      "step": 1641
    },
    {
      "epoch": 2.1297726950138895,
      "grad_norm": 0.7960946559906006,
      "learning_rate": 2.136186397088223e-06,
      "loss": 0.827523410320282,
      "step": 1642
    },
    {
      "epoch": 2.1310704219640284,
      "grad_norm": 0.8080329895019531,
      "learning_rate": 2.1303285392197043e-06,
      "loss": 0.7528422474861145,
      "step": 1643
    },
    {
      "epoch": 2.1323681489141677,
      "grad_norm": 0.7847034931182861,
      "learning_rate": 2.1244765491029985e-06,
      "loss": 0.6985142827033997,
      "step": 1644
    },
    {
      "epoch": 2.1336658758643066,
      "grad_norm": 0.7830514907836914,
      "learning_rate": 2.118630438703939e-06,
      "loss": 0.7506939768791199,
      "step": 1645
    },
    {
      "epoch": 2.1349636028144454,
      "grad_norm": 0.771859884262085,
      "learning_rate": 2.1127902199763496e-06,
      "loss": 0.7569597363471985,
      "step": 1646
    },
    {
      "epoch": 2.1362613297645843,
      "grad_norm": 0.7885754108428955,
      "learning_rate": 2.1069559048619937e-06,
      "loss": 0.7057400345802307,
      "step": 1647
    },
    {
      "epoch": 2.137559056714723,
      "grad_norm": 0.7832199931144714,
      "learning_rate": 2.10112750529057e-06,
      "loss": 0.7383140921592712,
      "step": 1648
    },
    {
      "epoch": 2.138856783664862,
      "grad_norm": 0.7768014669418335,
      "learning_rate": 2.095305033179682e-06,
      "loss": 0.7292782664299011,
      "step": 1649
    },
    {
      "epoch": 2.140154510615001,
      "grad_norm": 0.762650191783905,
      "learning_rate": 2.0894885004348102e-06,
      "loss": 0.7502353191375732,
      "step": 1650
    },
    {
      "epoch": 2.1414522375651397,
      "grad_norm": 0.7974070310592651,
      "learning_rate": 2.0836779189492925e-06,
      "loss": 0.8040381669998169,
      "step": 1651
    },
    {
      "epoch": 2.1427499645152785,
      "grad_norm": 0.7788371443748474,
      "learning_rate": 2.077873300604297e-06,
      "loss": 0.7352241277694702,
      "step": 1652
    },
    {
      "epoch": 2.1440476914654174,
      "grad_norm": 0.7705795764923096,
      "learning_rate": 2.0720746572687995e-06,
      "loss": 0.696068525314331,
      "step": 1653
    },
    {
      "epoch": 2.1453454184155567,
      "grad_norm": 0.741032600402832,
      "learning_rate": 2.0662820007995592e-06,
      "loss": 0.710914671421051,
      "step": 1654
    },
    {
      "epoch": 2.1466431453656956,
      "grad_norm": 0.7707566022872925,
      "learning_rate": 2.060495343041087e-06,
      "loss": 0.7286970019340515,
      "step": 1655
    },
    {
      "epoch": 2.1479408723158344,
      "grad_norm": 0.7634013295173645,
      "learning_rate": 2.0547146958256416e-06,
      "loss": 0.6444496512413025,
      "step": 1656
    },
    {
      "epoch": 2.1492385992659733,
      "grad_norm": 0.8233134746551514,
      "learning_rate": 2.048940070973177e-06,
      "loss": 0.6969065070152283,
      "step": 1657
    },
    {
      "epoch": 2.150536326216112,
      "grad_norm": 0.79786217212677,
      "learning_rate": 2.04317148029134e-06,
      "loss": 0.6712714433670044,
      "step": 1658
    },
    {
      "epoch": 2.151834053166251,
      "grad_norm": 0.7395349740982056,
      "learning_rate": 2.0374089355754434e-06,
      "loss": 0.7371563911437988,
      "step": 1659
    },
    {
      "epoch": 2.15313178011639,
      "grad_norm": 0.7514559626579285,
      "learning_rate": 2.031652448608428e-06,
      "loss": 0.6969033479690552,
      "step": 1660
    },
    {
      "epoch": 2.1544295070665287,
      "grad_norm": 0.8158461451530457,
      "learning_rate": 2.025902031160853e-06,
      "loss": 0.7191423773765564,
      "step": 1661
    },
    {
      "epoch": 2.1557272340166675,
      "grad_norm": 0.7599233984947205,
      "learning_rate": 2.020157694990868e-06,
      "loss": 0.7718644738197327,
      "step": 1662
    },
    {
      "epoch": 2.1570249609668064,
      "grad_norm": 0.7463486790657043,
      "learning_rate": 2.014419451844186e-06,
      "loss": 0.750013530254364,
      "step": 1663
    },
    {
      "epoch": 2.1583226879169457,
      "grad_norm": 0.775913417339325,
      "learning_rate": 2.0086873134540626e-06,
      "loss": 0.7322590947151184,
      "step": 1664
    },
    {
      "epoch": 2.1596204148670846,
      "grad_norm": 0.7643508315086365,
      "learning_rate": 2.002961291541269e-06,
      "loss": 0.7270373702049255,
      "step": 1665
    },
    {
      "epoch": 2.1609181418172234,
      "grad_norm": 0.8092241883277893,
      "learning_rate": 1.997241397814071e-06,
      "loss": 0.7437032461166382,
      "step": 1666
    },
    {
      "epoch": 2.1622158687673623,
      "grad_norm": 0.7753113508224487,
      "learning_rate": 1.9915276439682056e-06,
      "loss": 0.7822949886322021,
      "step": 1667
    },
    {
      "epoch": 2.163513595717501,
      "grad_norm": 0.7699429392814636,
      "learning_rate": 1.985820041686848e-06,
      "loss": 0.7268073558807373,
      "step": 1668
    },
    {
      "epoch": 2.16481132266764,
      "grad_norm": 0.8058201670646667,
      "learning_rate": 1.9801186026406066e-06,
      "loss": 0.7283480763435364,
      "step": 1669
    },
    {
      "epoch": 2.166109049617779,
      "grad_norm": 0.7627425193786621,
      "learning_rate": 1.9744233384874766e-06,
      "loss": 0.7668908834457397,
      "step": 1670
    },
    {
      "epoch": 2.1674067765679177,
      "grad_norm": 0.7747920155525208,
      "learning_rate": 1.968734260872833e-06,
      "loss": 0.6658475995063782,
      "step": 1671
    },
    {
      "epoch": 2.1687045035180565,
      "grad_norm": 0.7950249910354614,
      "learning_rate": 1.9630513814294e-06,
      "loss": 0.778218150138855,
      "step": 1672
    },
    {
      "epoch": 2.1700022304681954,
      "grad_norm": 0.77446049451828,
      "learning_rate": 1.9573747117772272e-06,
      "loss": 0.7079744935035706,
      "step": 1673
    },
    {
      "epoch": 2.1712999574183343,
      "grad_norm": 0.7665644884109497,
      "learning_rate": 1.951704263523668e-06,
      "loss": 0.7032575607299805,
      "step": 1674
    },
    {
      "epoch": 2.1725976843684736,
      "grad_norm": 0.7874617576599121,
      "learning_rate": 1.9460400482633537e-06,
      "loss": 0.7241728901863098,
      "step": 1675
    },
    {
      "epoch": 2.1738954113186124,
      "grad_norm": 0.7628116011619568,
      "learning_rate": 1.9403820775781696e-06,
      "loss": 0.6686381101608276,
      "step": 1676
    },
    {
      "epoch": 2.1751931382687513,
      "grad_norm": 0.7952485084533691,
      "learning_rate": 1.9347303630372373e-06,
      "loss": 0.7865254282951355,
      "step": 1677
    },
    {
      "epoch": 2.17649086521889,
      "grad_norm": 0.7570112943649292,
      "learning_rate": 1.929084916196876e-06,
      "loss": 0.7014156579971313,
      "step": 1678
    },
    {
      "epoch": 2.177788592169029,
      "grad_norm": 0.7880426049232483,
      "learning_rate": 1.923445748600603e-06,
      "loss": 0.8080270886421204,
      "step": 1679
    },
    {
      "epoch": 2.179086319119168,
      "grad_norm": 0.7836724519729614,
      "learning_rate": 1.917812871779084e-06,
      "loss": 0.7680548429489136,
      "step": 1680
    },
    {
      "epoch": 2.1803840460693067,
      "grad_norm": 0.7716360092163086,
      "learning_rate": 1.912186297250128e-06,
      "loss": 0.7878873348236084,
      "step": 1681
    },
    {
      "epoch": 2.1816817730194455,
      "grad_norm": 0.7800530195236206,
      "learning_rate": 1.9065660365186545e-06,
      "loss": 0.7374768853187561,
      "step": 1682
    },
    {
      "epoch": 2.1829794999695844,
      "grad_norm": 0.7685821056365967,
      "learning_rate": 1.9009521010766756e-06,
      "loss": 0.7049664855003357,
      "step": 1683
    },
    {
      "epoch": 2.1842772269197233,
      "grad_norm": 0.7467005848884583,
      "learning_rate": 1.8953445024032679e-06,
      "loss": 0.7375454902648926,
      "step": 1684
    },
    {
      "epoch": 2.185574953869862,
      "grad_norm": 0.7665326595306396,
      "learning_rate": 1.889743251964553e-06,
      "loss": 0.721852719783783,
      "step": 1685
    },
    {
      "epoch": 2.1868726808200014,
      "grad_norm": 0.8213875889778137,
      "learning_rate": 1.8841483612136658e-06,
      "loss": 0.7483677864074707,
      "step": 1686
    },
    {
      "epoch": 2.1881704077701403,
      "grad_norm": 0.7615724802017212,
      "learning_rate": 1.8785598415907464e-06,
      "loss": 0.685491681098938,
      "step": 1687
    },
    {
      "epoch": 2.189468134720279,
      "grad_norm": 0.805755078792572,
      "learning_rate": 1.8729777045229009e-06,
      "loss": 0.6272298693656921,
      "step": 1688
    },
    {
      "epoch": 2.190765861670418,
      "grad_norm": 0.7846405506134033,
      "learning_rate": 1.8674019614241879e-06,
      "loss": 0.662231981754303,
      "step": 1689
    },
    {
      "epoch": 2.192063588620557,
      "grad_norm": 0.7738309502601624,
      "learning_rate": 1.8618326236955908e-06,
      "loss": 0.6917383074760437,
      "step": 1690
    },
    {
      "epoch": 2.1933613155706957,
      "grad_norm": 0.7525291442871094,
      "learning_rate": 1.8562697027249921e-06,
      "loss": 0.7249594926834106,
      "step": 1691
    },
    {
      "epoch": 2.1946590425208345,
      "grad_norm": 0.7994334697723389,
      "learning_rate": 1.8507132098871633e-06,
      "loss": 0.7203621864318848,
      "step": 1692
    },
    {
      "epoch": 2.1959567694709734,
      "grad_norm": 0.8307973146438599,
      "learning_rate": 1.8451631565437211e-06,
      "loss": 0.7953019738197327,
      "step": 1693
    },
    {
      "epoch": 2.1972544964211123,
      "grad_norm": 0.718991219997406,
      "learning_rate": 1.8396195540431205e-06,
      "loss": 0.7472409605979919,
      "step": 1694
    },
    {
      "epoch": 2.198552223371251,
      "grad_norm": 0.7688009738922119,
      "learning_rate": 1.834082413720627e-06,
      "loss": 0.754511833190918,
      "step": 1695
    },
    {
      "epoch": 2.19984995032139,
      "grad_norm": 0.8119645118713379,
      "learning_rate": 1.8285517468982905e-06,
      "loss": 0.7157659530639648,
      "step": 1696
    },
    {
      "epoch": 2.2011476772715293,
      "grad_norm": 0.7515578269958496,
      "learning_rate": 1.8230275648849243e-06,
      "loss": 0.7136957049369812,
      "step": 1697
    },
    {
      "epoch": 2.202445404221668,
      "grad_norm": 0.739663302898407,
      "learning_rate": 1.8175098789760848e-06,
      "loss": 0.7205515503883362,
      "step": 1698
    },
    {
      "epoch": 2.203743131171807,
      "grad_norm": 0.7999836802482605,
      "learning_rate": 1.8119987004540373e-06,
      "loss": 0.7069056034088135,
      "step": 1699
    },
    {
      "epoch": 2.205040858121946,
      "grad_norm": 0.7710497379302979,
      "learning_rate": 1.8064940405877546e-06,
      "loss": 0.7351633310317993,
      "step": 1700
    },
    {
      "epoch": 2.2063385850720847,
      "grad_norm": 0.7900813817977905,
      "learning_rate": 1.8009959106328655e-06,
      "loss": 0.6964262127876282,
      "step": 1701
    },
    {
      "epoch": 2.2076363120222235,
      "grad_norm": 0.7656311392784119,
      "learning_rate": 1.7955043218316615e-06,
      "loss": 0.7031038999557495,
      "step": 1702
    },
    {
      "epoch": 2.2089340389723624,
      "grad_norm": 0.779863715171814,
      "learning_rate": 1.7900192854130465e-06,
      "loss": 0.6933379769325256,
      "step": 1703
    },
    {
      "epoch": 2.2102317659225013,
      "grad_norm": 0.7669497132301331,
      "learning_rate": 1.7845408125925328e-06,
      "loss": 0.6850363612174988,
      "step": 1704
    },
    {
      "epoch": 2.21152949287264,
      "grad_norm": 0.7710373401641846,
      "learning_rate": 1.7790689145722111e-06,
      "loss": 0.7190161347389221,
      "step": 1705
    },
    {
      "epoch": 2.212827219822779,
      "grad_norm": 0.730596661567688,
      "learning_rate": 1.7736036025407282e-06,
      "loss": 0.6166294813156128,
      "step": 1706
    },
    {
      "epoch": 2.2141249467729183,
      "grad_norm": 0.7601070404052734,
      "learning_rate": 1.7681448876732632e-06,
      "loss": 0.6836649179458618,
      "step": 1707
    },
    {
      "epoch": 2.215422673723057,
      "grad_norm": 0.7664089202880859,
      "learning_rate": 1.7626927811315087e-06,
      "loss": 0.7623034119606018,
      "step": 1708
    },
    {
      "epoch": 2.216720400673196,
      "grad_norm": 0.8523539304733276,
      "learning_rate": 1.7572472940636375e-06,
      "loss": 0.7964061498641968,
      "step": 1709
    },
    {
      "epoch": 2.218018127623335,
      "grad_norm": 0.7584978342056274,
      "learning_rate": 1.7518084376042988e-06,
      "loss": 0.6912890672683716,
      "step": 1710
    },
    {
      "epoch": 2.2193158545734737,
      "grad_norm": 0.7614145874977112,
      "learning_rate": 1.7463762228745728e-06,
      "loss": 0.693087100982666,
      "step": 1711
    },
    {
      "epoch": 2.2206135815236125,
      "grad_norm": 0.8198820352554321,
      "learning_rate": 1.7409506609819648e-06,
      "loss": 0.6574248671531677,
      "step": 1712
    },
    {
      "epoch": 2.2219113084737514,
      "grad_norm": 0.728805422782898,
      "learning_rate": 1.735531763020376e-06,
      "loss": 0.6723518967628479,
      "step": 1713
    },
    {
      "epoch": 2.2232090354238903,
      "grad_norm": 0.7606318593025208,
      "learning_rate": 1.7301195400700815e-06,
      "loss": 0.626229465007782,
      "step": 1714
    },
    {
      "epoch": 2.224506762374029,
      "grad_norm": 0.7988364100456238,
      "learning_rate": 1.7247140031977073e-06,
      "loss": 0.7640826106071472,
      "step": 1715
    },
    {
      "epoch": 2.225804489324168,
      "grad_norm": 0.7772147059440613,
      "learning_rate": 1.7193151634562071e-06,
      "loss": 0.6884752511978149,
      "step": 1716
    },
    {
      "epoch": 2.2271022162743073,
      "grad_norm": 0.7542872428894043,
      "learning_rate": 1.7139230318848432e-06,
      "loss": 0.740616500377655,
      "step": 1717
    },
    {
      "epoch": 2.228399943224446,
      "grad_norm": 0.7642157077789307,
      "learning_rate": 1.7085376195091591e-06,
      "loss": 0.732641339302063,
      "step": 1718
    },
    {
      "epoch": 2.229697670174585,
      "grad_norm": 0.777519941329956,
      "learning_rate": 1.7031589373409596e-06,
      "loss": 0.6960573792457581,
      "step": 1719
    },
    {
      "epoch": 2.230995397124724,
      "grad_norm": 0.759345531463623,
      "learning_rate": 1.6977869963782895e-06,
      "loss": 0.7143913507461548,
      "step": 1720
    },
    {
      "epoch": 2.2322931240748627,
      "grad_norm": 0.782537043094635,
      "learning_rate": 1.6924218076054095e-06,
      "loss": 0.7428768277168274,
      "step": 1721
    },
    {
      "epoch": 2.2335908510250015,
      "grad_norm": 0.7905192971229553,
      "learning_rate": 1.6870633819927672e-06,
      "loss": 0.8565322160720825,
      "step": 1722
    },
    {
      "epoch": 2.2348885779751404,
      "grad_norm": 0.780165433883667,
      "learning_rate": 1.6817117304969944e-06,
      "loss": 0.705779492855072,
      "step": 1723
    },
    {
      "epoch": 2.2361863049252793,
      "grad_norm": 0.7498677372932434,
      "learning_rate": 1.676366864060856e-06,
      "loss": 0.7112017273902893,
      "step": 1724
    },
    {
      "epoch": 2.237484031875418,
      "grad_norm": 0.7308558821678162,
      "learning_rate": 1.6710287936132592e-06,
      "loss": 0.6326340436935425,
      "step": 1725
    },
    {
      "epoch": 2.238781758825557,
      "grad_norm": 0.7382625341415405,
      "learning_rate": 1.6656975300692008e-06,
      "loss": 0.6695979833602905,
      "step": 1726
    },
    {
      "epoch": 2.240079485775696,
      "grad_norm": 0.7628723382949829,
      "learning_rate": 1.660373084329767e-06,
      "loss": 0.770473062992096,
      "step": 1727
    },
    {
      "epoch": 2.241377212725835,
      "grad_norm": 0.7602196335792542,
      "learning_rate": 1.6550554672821028e-06,
      "loss": 0.6869367957115173,
      "step": 1728
    },
    {
      "epoch": 2.242674939675974,
      "grad_norm": 0.8013946413993835,
      "learning_rate": 1.6497446897993885e-06,
      "loss": 0.7376055717468262,
      "step": 1729
    },
    {
      "epoch": 2.243972666626113,
      "grad_norm": 0.8051080703735352,
      "learning_rate": 1.6444407627408194e-06,
      "loss": 0.7137443423271179,
      "step": 1730
    },
    {
      "epoch": 2.2452703935762517,
      "grad_norm": 0.7435747981071472,
      "learning_rate": 1.639143696951586e-06,
      "loss": 0.6921373605728149,
      "step": 1731
    },
    {
      "epoch": 2.2465681205263905,
      "grad_norm": 0.789109468460083,
      "learning_rate": 1.6338535032628427e-06,
      "loss": 0.7224962115287781,
      "step": 1732
    },
    {
      "epoch": 2.2478658474765294,
      "grad_norm": 0.7830778360366821,
      "learning_rate": 1.6285701924917025e-06,
      "loss": 0.7313556671142578,
      "step": 1733
    },
    {
      "epoch": 2.2491635744266683,
      "grad_norm": 0.7683809995651245,
      "learning_rate": 1.6232937754411938e-06,
      "loss": 0.6995819807052612,
      "step": 1734
    },
    {
      "epoch": 2.250461301376807,
      "grad_norm": 0.719525933265686,
      "learning_rate": 1.6180242629002558e-06,
      "loss": 0.6746504902839661,
      "step": 1735
    },
    {
      "epoch": 2.251759028326946,
      "grad_norm": 0.7692328095436096,
      "learning_rate": 1.6127616656437078e-06,
      "loss": 0.7401602268218994,
      "step": 1736
    },
    {
      "epoch": 2.253056755277085,
      "grad_norm": 0.815162181854248,
      "learning_rate": 1.6075059944322297e-06,
      "loss": 0.7262242436408997,
      "step": 1737
    },
    {
      "epoch": 2.2543544822272237,
      "grad_norm": 0.7180652022361755,
      "learning_rate": 1.6022572600123382e-06,
      "loss": 0.6500532031059265,
      "step": 1738
    },
    {
      "epoch": 2.255652209177363,
      "grad_norm": 0.7820153832435608,
      "learning_rate": 1.5970154731163667e-06,
      "loss": 0.7571191787719727,
      "step": 1739
    },
    {
      "epoch": 2.256949936127502,
      "grad_norm": 0.7669167518615723,
      "learning_rate": 1.5917806444624434e-06,
      "loss": 0.7316527962684631,
      "step": 1740
    },
    {
      "epoch": 2.2582476630776407,
      "grad_norm": 0.7556010484695435,
      "learning_rate": 1.5865527847544692e-06,
      "loss": 0.6542820930480957,
      "step": 1741
    },
    {
      "epoch": 2.2595453900277795,
      "grad_norm": 0.7612343430519104,
      "learning_rate": 1.581331904682089e-06,
      "loss": 0.7321107983589172,
      "step": 1742
    },
    {
      "epoch": 2.2608431169779184,
      "grad_norm": 0.7514542937278748,
      "learning_rate": 1.576118014920688e-06,
      "loss": 0.6985179781913757,
      "step": 1743
    },
    {
      "epoch": 2.2621408439280573,
      "grad_norm": 0.7584021091461182,
      "learning_rate": 1.5709111261313454e-06,
      "loss": 0.7909868359565735,
      "step": 1744
    },
    {
      "epoch": 2.263438570878196,
      "grad_norm": 0.7409632802009583,
      "learning_rate": 1.5657112489608316e-06,
      "loss": 0.6589058637619019,
      "step": 1745
    },
    {
      "epoch": 2.264736297828335,
      "grad_norm": 0.7241658568382263,
      "learning_rate": 1.5605183940415842e-06,
      "loss": 0.6412944197654724,
      "step": 1746
    },
    {
      "epoch": 2.266034024778474,
      "grad_norm": 0.815857470035553,
      "learning_rate": 1.5553325719916717e-06,
      "loss": 0.7096395492553711,
      "step": 1747
    },
    {
      "epoch": 2.2673317517286127,
      "grad_norm": 0.7670204043388367,
      "learning_rate": 1.5501537934147897e-06,
      "loss": 0.7624009847640991,
      "step": 1748
    },
    {
      "epoch": 2.2686294786787515,
      "grad_norm": 0.7562599182128906,
      "learning_rate": 1.5449820689002298e-06,
      "loss": 0.7044228911399841,
      "step": 1749
    },
    {
      "epoch": 2.269927205628891,
      "grad_norm": 0.7411571741104126,
      "learning_rate": 1.5398174090228595e-06,
      "loss": 0.6269949078559875,
      "step": 1750
    },
    {
      "epoch": 2.2712249325790297,
      "grad_norm": 0.7651223540306091,
      "learning_rate": 1.534659824343101e-06,
      "loss": 0.6997742056846619,
      "step": 1751
    },
    {
      "epoch": 2.2725226595291685,
      "grad_norm": 0.7359679937362671,
      "learning_rate": 1.5295093254069093e-06,
      "loss": 0.6871427297592163,
      "step": 1752
    },
    {
      "epoch": 2.2738203864793074,
      "grad_norm": 0.7561101913452148,
      "learning_rate": 1.524365922745752e-06,
      "loss": 0.6495131850242615,
      "step": 1753
    },
    {
      "epoch": 2.2751181134294463,
      "grad_norm": 0.7630066871643066,
      "learning_rate": 1.519229626876586e-06,
      "loss": 0.6955438852310181,
      "step": 1754
    },
    {
      "epoch": 2.276415840379585,
      "grad_norm": 0.747558057308197,
      "learning_rate": 1.5141004483018323e-06,
      "loss": 0.6892863512039185,
      "step": 1755
    },
    {
      "epoch": 2.277713567329724,
      "grad_norm": 0.7459293007850647,
      "learning_rate": 1.5089783975093698e-06,
      "loss": 0.6839847564697266,
      "step": 1756
    },
    {
      "epoch": 2.279011294279863,
      "grad_norm": 0.7482560873031616,
      "learning_rate": 1.5038634849724898e-06,
      "loss": 0.670592188835144,
      "step": 1757
    },
    {
      "epoch": 2.2803090212300017,
      "grad_norm": 0.7818513512611389,
      "learning_rate": 1.4987557211498966e-06,
      "loss": 0.6944141387939453,
      "step": 1758
    },
    {
      "epoch": 2.2816067481801405,
      "grad_norm": 0.8236380815505981,
      "learning_rate": 1.4936551164856739e-06,
      "loss": 0.7209656238555908,
      "step": 1759
    },
    {
      "epoch": 2.2829044751302794,
      "grad_norm": 0.7400614619255066,
      "learning_rate": 1.4885616814092673e-06,
      "loss": 0.6475507020950317,
      "step": 1760
    },
    {
      "epoch": 2.2842022020804187,
      "grad_norm": 0.7747074961662292,
      "learning_rate": 1.4834754263354628e-06,
      "loss": 0.7706501483917236,
      "step": 1761
    },
    {
      "epoch": 2.2854999290305575,
      "grad_norm": 0.7987594604492188,
      "learning_rate": 1.4783963616643654e-06,
      "loss": 0.7353149056434631,
      "step": 1762
    },
    {
      "epoch": 2.2867976559806964,
      "grad_norm": 0.7654763460159302,
      "learning_rate": 1.4733244977813726e-06,
      "loss": 0.781434953212738,
      "step": 1763
    },
    {
      "epoch": 2.2880953829308353,
      "grad_norm": 0.7649980783462524,
      "learning_rate": 1.468259845057169e-06,
      "loss": 0.686353325843811,
      "step": 1764
    },
    {
      "epoch": 2.289393109880974,
      "grad_norm": 0.743064820766449,
      "learning_rate": 1.4632024138476803e-06,
      "loss": 0.6991320252418518,
      "step": 1765
    },
    {
      "epoch": 2.290690836831113,
      "grad_norm": 0.756190836429596,
      "learning_rate": 1.4581522144940802e-06,
      "loss": 0.7311216592788696,
      "step": 1766
    },
    {
      "epoch": 2.291988563781252,
      "grad_norm": 0.7912296056747437,
      "learning_rate": 1.4531092573227434e-06,
      "loss": 0.6862029433250427,
      "step": 1767
    },
    {
      "epoch": 2.2932862907313907,
      "grad_norm": 0.7685181498527527,
      "learning_rate": 1.4480735526452427e-06,
      "loss": 0.7343103885650635,
      "step": 1768
    },
    {
      "epoch": 2.2945840176815295,
      "grad_norm": 0.7459105849266052,
      "learning_rate": 1.4430451107583187e-06,
      "loss": 0.6750067472457886,
      "step": 1769
    },
    {
      "epoch": 2.295881744631669,
      "grad_norm": 0.7575411200523376,
      "learning_rate": 1.4380239419438636e-06,
      "loss": 0.7810744643211365,
      "step": 1770
    },
    {
      "epoch": 2.2971794715818072,
      "grad_norm": 0.7612024545669556,
      "learning_rate": 1.433010056468896e-06,
      "loss": 0.7204791903495789,
      "step": 1771
    },
    {
      "epoch": 2.2984771985319465,
      "grad_norm": 0.7742648720741272,
      "learning_rate": 1.4280034645855429e-06,
      "loss": 0.7646610140800476,
      "step": 1772
    },
    {
      "epoch": 2.2997749254820854,
      "grad_norm": 0.7920587062835693,
      "learning_rate": 1.4230041765310171e-06,
      "loss": 0.7437126040458679,
      "step": 1773
    },
    {
      "epoch": 2.3010726524322243,
      "grad_norm": 0.7467794418334961,
      "learning_rate": 1.4180122025275972e-06,
      "loss": 0.6069324016571045,
      "step": 1774
    },
    {
      "epoch": 2.302370379382363,
      "grad_norm": 0.7249860167503357,
      "learning_rate": 1.4130275527826077e-06,
      "loss": 0.7492033243179321,
      "step": 1775
    },
    {
      "epoch": 2.303668106332502,
      "grad_norm": 0.79679274559021,
      "learning_rate": 1.4080502374883947e-06,
      "loss": 0.7478646636009216,
      "step": 1776
    },
    {
      "epoch": 2.304965833282641,
      "grad_norm": 0.7763088345527649,
      "learning_rate": 1.4030802668223097e-06,
      "loss": 0.721184253692627,
      "step": 1777
    },
    {
      "epoch": 2.3062635602327797,
      "grad_norm": 0.7542839646339417,
      "learning_rate": 1.398117650946681e-06,
      "loss": 0.7005340456962585,
      "step": 1778
    },
    {
      "epoch": 2.3075612871829185,
      "grad_norm": 0.7729418873786926,
      "learning_rate": 1.3931624000088073e-06,
      "loss": 0.6777938008308411,
      "step": 1779
    },
    {
      "epoch": 2.3088590141330574,
      "grad_norm": 0.7986354827880859,
      "learning_rate": 1.3882145241409184e-06,
      "loss": 0.7462781071662903,
      "step": 1780
    },
    {
      "epoch": 2.3101567410831967,
      "grad_norm": 0.7915672063827515,
      "learning_rate": 1.3832740334601692e-06,
      "loss": 0.7387913465499878,
      "step": 1781
    },
    {
      "epoch": 2.3114544680333355,
      "grad_norm": 0.7277045249938965,
      "learning_rate": 1.3783409380686135e-06,
      "loss": 0.6593598127365112,
      "step": 1782
    },
    {
      "epoch": 2.3127521949834744,
      "grad_norm": 0.7720676064491272,
      "learning_rate": 1.3734152480531821e-06,
      "loss": 0.7562534213066101,
      "step": 1783
    },
    {
      "epoch": 2.3140499219336133,
      "grad_norm": 0.7754230499267578,
      "learning_rate": 1.3684969734856646e-06,
      "loss": 0.6947664022445679,
      "step": 1784
    },
    {
      "epoch": 2.315347648883752,
      "grad_norm": 0.7693092226982117,
      "learning_rate": 1.363586124422689e-06,
      "loss": 0.7212303876876831,
      "step": 1785
    },
    {
      "epoch": 2.316645375833891,
      "grad_norm": 0.7735230326652527,
      "learning_rate": 1.3586827109056944e-06,
      "loss": 0.6982312798500061,
      "step": 1786
    },
    {
      "epoch": 2.31794310278403,
      "grad_norm": 0.7643104195594788,
      "learning_rate": 1.3537867429609263e-06,
      "loss": 0.678123950958252,
      "step": 1787
    },
    {
      "epoch": 2.3192408297341687,
      "grad_norm": 0.7314268946647644,
      "learning_rate": 1.3488982305993942e-06,
      "loss": 0.6732868552207947,
      "step": 1788
    },
    {
      "epoch": 2.3205385566843075,
      "grad_norm": 0.7716788649559021,
      "learning_rate": 1.3440171838168743e-06,
      "loss": 0.7167251110076904,
      "step": 1789
    },
    {
      "epoch": 2.3218362836344464,
      "grad_norm": 0.7513408660888672,
      "learning_rate": 1.3391436125938673e-06,
      "loss": 0.7298739552497864,
      "step": 1790
    },
    {
      "epoch": 2.3231340105845852,
      "grad_norm": 0.7510161399841309,
      "learning_rate": 1.3342775268955943e-06,
      "loss": 0.7235835194587708,
      "step": 1791
    },
    {
      "epoch": 2.3244317375347245,
      "grad_norm": 0.8174487948417664,
      "learning_rate": 1.329418936671969e-06,
      "loss": 0.7079183459281921,
      "step": 1792
    },
    {
      "epoch": 2.3257294644848634,
      "grad_norm": 0.7633885741233826,
      "learning_rate": 1.3245678518575782e-06,
      "loss": 0.7166120409965515,
      "step": 1793
    },
    {
      "epoch": 2.3270271914350023,
      "grad_norm": 0.7898662090301514,
      "learning_rate": 1.319724282371664e-06,
      "loss": 0.7210479378700256,
      "step": 1794
    },
    {
      "epoch": 2.328324918385141,
      "grad_norm": 0.7649275064468384,
      "learning_rate": 1.3148882381181e-06,
      "loss": 0.6867713928222656,
      "step": 1795
    },
    {
      "epoch": 2.32962264533528,
      "grad_norm": 0.7802595496177673,
      "learning_rate": 1.3100597289853689e-06,
      "loss": 0.7147336006164551,
      "step": 1796
    },
    {
      "epoch": 2.330920372285419,
      "grad_norm": 0.8007116913795471,
      "learning_rate": 1.3052387648465559e-06,
      "loss": 0.778854250907898,
      "step": 1797
    },
    {
      "epoch": 2.3322180992355577,
      "grad_norm": 0.811072051525116,
      "learning_rate": 1.3004253555593071e-06,
      "loss": 0.6877797245979309,
      "step": 1798
    },
    {
      "epoch": 2.3335158261856965,
      "grad_norm": 0.7577556371688843,
      "learning_rate": 1.2956195109658287e-06,
      "loss": 0.6524724364280701,
      "step": 1799
    },
    {
      "epoch": 2.3348135531358354,
      "grad_norm": 0.778110921382904,
      "learning_rate": 1.2908212408928561e-06,
      "loss": 0.8117507696151733,
      "step": 1800
    },
    {
      "epoch": 2.3361112800859742,
      "grad_norm": 0.7517038583755493,
      "learning_rate": 1.2860305551516355e-06,
      "loss": 0.7689738273620605,
      "step": 1801
    },
    {
      "epoch": 2.337409007036113,
      "grad_norm": 0.7671796679496765,
      "learning_rate": 1.281247463537912e-06,
      "loss": 0.6367142200469971,
      "step": 1802
    },
    {
      "epoch": 2.3387067339862524,
      "grad_norm": 0.8273651003837585,
      "learning_rate": 1.276471975831891e-06,
      "loss": 0.7301177978515625,
      "step": 1803
    },
    {
      "epoch": 2.3400044609363913,
      "grad_norm": 0.8008820414543152,
      "learning_rate": 1.2717041017982396e-06,
      "loss": 0.7587131857872009,
      "step": 1804
    },
    {
      "epoch": 2.34130218788653,
      "grad_norm": 0.7739439010620117,
      "learning_rate": 1.2669438511860527e-06,
      "loss": 0.6916797757148743,
      "step": 1805
    },
    {
      "epoch": 2.342599914836669,
      "grad_norm": 0.7357456088066101,
      "learning_rate": 1.2621912337288372e-06,
      "loss": 0.6751843690872192,
      "step": 1806
    },
    {
      "epoch": 2.343897641786808,
      "grad_norm": 0.7770068049430847,
      "learning_rate": 1.257446259144494e-06,
      "loss": 0.666739821434021,
      "step": 1807
    },
    {
      "epoch": 2.3451953687369467,
      "grad_norm": 0.7363953590393066,
      "learning_rate": 1.2527089371352968e-06,
      "loss": 0.6992002129554749,
      "step": 1808
    },
    {
      "epoch": 2.3464930956870855,
      "grad_norm": 0.759560227394104,
      "learning_rate": 1.2479792773878647e-06,
      "loss": 0.7391279339790344,
      "step": 1809
    },
    {
      "epoch": 2.3477908226372244,
      "grad_norm": 0.8206911087036133,
      "learning_rate": 1.243257289573161e-06,
      "loss": 0.7275311946868896,
      "step": 1810
    },
    {
      "epoch": 2.3490885495873632,
      "grad_norm": 0.7866367697715759,
      "learning_rate": 1.2385429833464513e-06,
      "loss": 0.6576771140098572,
      "step": 1811
    },
    {
      "epoch": 2.350386276537502,
      "grad_norm": 0.7733588218688965,
      "learning_rate": 1.2338363683472998e-06,
      "loss": 0.7235656380653381,
      "step": 1812
    },
    {
      "epoch": 2.351684003487641,
      "grad_norm": 0.7723690867424011,
      "learning_rate": 1.2291374541995437e-06,
      "loss": 0.7257228493690491,
      "step": 1813
    },
    {
      "epoch": 2.3529817304377803,
      "grad_norm": 0.8251050114631653,
      "learning_rate": 1.224446250511272e-06,
      "loss": 0.7229258418083191,
      "step": 1814
    },
    {
      "epoch": 2.354279457387919,
      "grad_norm": 0.7819148898124695,
      "learning_rate": 1.2197627668748101e-06,
      "loss": 0.7237058281898499,
      "step": 1815
    },
    {
      "epoch": 2.355577184338058,
      "grad_norm": 0.782592236995697,
      "learning_rate": 1.2150870128666959e-06,
      "loss": 0.6931707262992859,
      "step": 1816
    },
    {
      "epoch": 2.356874911288197,
      "grad_norm": 0.7515336275100708,
      "learning_rate": 1.2104189980476627e-06,
      "loss": 0.7242140173912048,
      "step": 1817
    },
    {
      "epoch": 2.3581726382383357,
      "grad_norm": 0.7545520067214966,
      "learning_rate": 1.2057587319626213e-06,
      "loss": 0.644026517868042,
      "step": 1818
    },
    {
      "epoch": 2.3594703651884745,
      "grad_norm": 0.7882485389709473,
      "learning_rate": 1.2011062241406313e-06,
      "loss": 0.7124985456466675,
      "step": 1819
    },
    {
      "epoch": 2.3607680921386134,
      "grad_norm": 0.7717565298080444,
      "learning_rate": 1.1964614840949002e-06,
      "loss": 0.7361290454864502,
      "step": 1820
    },
    {
      "epoch": 2.3620658190887522,
      "grad_norm": 0.7597373723983765,
      "learning_rate": 1.1918245213227408e-06,
      "loss": 0.702293336391449,
      "step": 1821
    },
    {
      "epoch": 2.363363546038891,
      "grad_norm": 0.7483997941017151,
      "learning_rate": 1.1871953453055707e-06,
      "loss": 0.7523679137229919,
      "step": 1822
    },
    {
      "epoch": 2.3646612729890304,
      "grad_norm": 0.7145439386367798,
      "learning_rate": 1.182573965508882e-06,
      "loss": 0.705971360206604,
      "step": 1823
    },
    {
      "epoch": 2.365958999939169,
      "grad_norm": 0.7735007405281067,
      "learning_rate": 1.1779603913822274e-06,
      "loss": 0.6811049580574036,
      "step": 1824
    },
    {
      "epoch": 2.367256726889308,
      "grad_norm": 0.7735781073570251,
      "learning_rate": 1.1733546323591981e-06,
      "loss": 0.7109265923500061,
      "step": 1825
    },
    {
      "epoch": 2.368554453839447,
      "grad_norm": 0.7741604447364807,
      "learning_rate": 1.168756697857406e-06,
      "loss": 0.7233576774597168,
      "step": 1826
    },
    {
      "epoch": 2.369852180789586,
      "grad_norm": 0.814807116985321,
      "learning_rate": 1.1641665972784628e-06,
      "loss": 0.7955071926116943,
      "step": 1827
    },
    {
      "epoch": 2.3711499077397247,
      "grad_norm": 0.7785446047782898,
      "learning_rate": 1.1595843400079636e-06,
      "loss": 0.695061445236206,
      "step": 1828
    },
    {
      "epoch": 2.3724476346898635,
      "grad_norm": 0.7668300271034241,
      "learning_rate": 1.1550099354154615e-06,
      "loss": 0.7401837110519409,
      "step": 1829
    },
    {
      "epoch": 2.3737453616400024,
      "grad_norm": 0.7792728543281555,
      "learning_rate": 1.1504433928544594e-06,
      "loss": 0.7388436794281006,
      "step": 1830
    },
    {
      "epoch": 2.3750430885901412,
      "grad_norm": 0.7757245302200317,
      "learning_rate": 1.1458847216623813e-06,
      "loss": 0.7079841494560242,
      "step": 1831
    },
    {
      "epoch": 2.37634081554028,
      "grad_norm": 0.7931995391845703,
      "learning_rate": 1.141333931160552e-06,
      "loss": 0.7631800174713135,
      "step": 1832
    },
    {
      "epoch": 2.377638542490419,
      "grad_norm": 0.8062769770622253,
      "learning_rate": 1.1367910306541918e-06,
      "loss": 0.80658358335495,
      "step": 1833
    },
    {
      "epoch": 2.3789362694405582,
      "grad_norm": 0.781043291091919,
      "learning_rate": 1.1322560294323775e-06,
      "loss": 0.7152395248413086,
      "step": 1834
    },
    {
      "epoch": 2.380233996390697,
      "grad_norm": 0.7616091966629028,
      "learning_rate": 1.1277289367680411e-06,
      "loss": 0.7314598560333252,
      "step": 1835
    },
    {
      "epoch": 2.381531723340836,
      "grad_norm": 0.7945448756217957,
      "learning_rate": 1.123209761917941e-06,
      "loss": 0.6685973405838013,
      "step": 1836
    },
    {
      "epoch": 2.382829450290975,
      "grad_norm": 0.7686976194381714,
      "learning_rate": 1.1186985141226458e-06,
      "loss": 0.6801435351371765,
      "step": 1837
    },
    {
      "epoch": 2.3841271772411137,
      "grad_norm": 0.7716367840766907,
      "learning_rate": 1.1141952026065156e-06,
      "loss": 0.746649444103241,
      "step": 1838
    },
    {
      "epoch": 2.3854249041912525,
      "grad_norm": 0.7480162978172302,
      "learning_rate": 1.1096998365776828e-06,
      "loss": 0.7064750790596008,
      "step": 1839
    },
    {
      "epoch": 2.3867226311413914,
      "grad_norm": 0.7676665782928467,
      "learning_rate": 1.1052124252280322e-06,
      "loss": 0.6626559495925903,
      "step": 1840
    },
    {
      "epoch": 2.3867226311413914,
      "eval_loss": 0.7613735198974609,
      "eval_runtime": 141.995,
      "eval_samples_per_second": 36.565,
      "eval_steps_per_second": 9.141,
      "step": 1840
    },
    {
      "epoch": 2.3880203580915302,
      "grad_norm": 0.7465634942054749,
      "learning_rate": 1.1007329777331866e-06,
      "loss": 0.717859148979187,
      "step": 1841
    },
    {
      "epoch": 2.389318085041669,
      "grad_norm": 0.7700461149215698,
      "learning_rate": 1.096261503252478e-06,
      "loss": 0.653079092502594,
      "step": 1842
    },
    {
      "epoch": 2.390615811991808,
      "grad_norm": 0.7783061265945435,
      "learning_rate": 1.0917980109289455e-06,
      "loss": 0.6487604975700378,
      "step": 1843
    },
    {
      "epoch": 2.391913538941947,
      "grad_norm": 0.7548145651817322,
      "learning_rate": 1.0873425098892964e-06,
      "loss": 0.6680834889411926,
      "step": 1844
    },
    {
      "epoch": 2.393211265892086,
      "grad_norm": 0.7790629267692566,
      "learning_rate": 1.082895009243905e-06,
      "loss": 0.707176923751831,
      "step": 1845
    },
    {
      "epoch": 2.394508992842225,
      "grad_norm": 0.7966057658195496,
      "learning_rate": 1.078455518086784e-06,
      "loss": 0.6602774858474731,
      "step": 1846
    },
    {
      "epoch": 2.395806719792364,
      "grad_norm": 0.7358562350273132,
      "learning_rate": 1.0740240454955692e-06,
      "loss": 0.6799494028091431,
      "step": 1847
    },
    {
      "epoch": 2.3971044467425027,
      "grad_norm": 0.7790771126747131,
      "learning_rate": 1.0696006005314996e-06,
      "loss": 0.7382240891456604,
      "step": 1848
    },
    {
      "epoch": 2.3984021736926415,
      "grad_norm": 0.7618228793144226,
      "learning_rate": 1.0651851922394035e-06,
      "loss": 0.6929839253425598,
      "step": 1849
    },
    {
      "epoch": 2.3996999006427804,
      "grad_norm": 0.7744076251983643,
      "learning_rate": 1.0607778296476679e-06,
      "loss": 0.7444416880607605,
      "step": 1850
    },
    {
      "epoch": 2.4009976275929192,
      "grad_norm": 0.72736656665802,
      "learning_rate": 1.05637852176824e-06,
      "loss": 0.6443988680839539,
      "step": 1851
    },
    {
      "epoch": 2.402295354543058,
      "grad_norm": 0.7641523480415344,
      "learning_rate": 1.051987277596585e-06,
      "loss": 0.7098280191421509,
      "step": 1852
    },
    {
      "epoch": 2.403593081493197,
      "grad_norm": 0.7642392516136169,
      "learning_rate": 1.0476041061116915e-06,
      "loss": 0.732379674911499,
      "step": 1853
    },
    {
      "epoch": 2.404890808443336,
      "grad_norm": 0.7198097109794617,
      "learning_rate": 1.0432290162760311e-06,
      "loss": 0.6814613342285156,
      "step": 1854
    },
    {
      "epoch": 2.4061885353934747,
      "grad_norm": 0.7901527285575867,
      "learning_rate": 1.038862017035558e-06,
      "loss": 0.7214125394821167,
      "step": 1855
    },
    {
      "epoch": 2.407486262343614,
      "grad_norm": 0.7606606483459473,
      "learning_rate": 1.0345031173196785e-06,
      "loss": 0.6583145260810852,
      "step": 1856
    },
    {
      "epoch": 2.408783989293753,
      "grad_norm": 0.7895342707633972,
      "learning_rate": 1.0301523260412405e-06,
      "loss": 0.7350203990936279,
      "step": 1857
    },
    {
      "epoch": 2.4100817162438917,
      "grad_norm": 0.8400049805641174,
      "learning_rate": 1.025809652096511e-06,
      "loss": 0.7966498136520386,
      "step": 1858
    },
    {
      "epoch": 2.4113794431940305,
      "grad_norm": 0.7552136778831482,
      "learning_rate": 1.0214751043651582e-06,
      "loss": 0.6972966194152832,
      "step": 1859
    },
    {
      "epoch": 2.4126771701441694,
      "grad_norm": 0.7996754050254822,
      "learning_rate": 1.0171486917102357e-06,
      "loss": 0.6762542724609375,
      "step": 1860
    },
    {
      "epoch": 2.4139748970943082,
      "grad_norm": 0.7461639046669006,
      "learning_rate": 1.0128304229781622e-06,
      "loss": 0.6995037198066711,
      "step": 1861
    },
    {
      "epoch": 2.415272624044447,
      "grad_norm": 0.769523024559021,
      "learning_rate": 1.008520306998706e-06,
      "loss": 0.6845777034759521,
      "step": 1862
    },
    {
      "epoch": 2.416570350994586,
      "grad_norm": 0.7502967119216919,
      "learning_rate": 1.0042183525849586e-06,
      "loss": 0.6307386159896851,
      "step": 1863
    },
    {
      "epoch": 2.417868077944725,
      "grad_norm": 0.7713011503219604,
      "learning_rate": 9.999245685333342e-07,
      "loss": 0.6450724601745605,
      "step": 1864
    },
    {
      "epoch": 2.4191658048948637,
      "grad_norm": 0.7563788294792175,
      "learning_rate": 9.95638963623528e-07,
      "loss": 0.7016159892082214,
      "step": 1865
    },
    {
      "epoch": 2.4204635318450025,
      "grad_norm": 0.7966086864471436,
      "learning_rate": 9.913615466185234e-07,
      "loss": 0.7306876182556152,
      "step": 1866
    },
    {
      "epoch": 2.421761258795142,
      "grad_norm": 0.7418493032455444,
      "learning_rate": 9.870923262645516e-07,
      "loss": 0.6521520018577576,
      "step": 1867
    },
    {
      "epoch": 2.4230589857452807,
      "grad_norm": 0.7862077355384827,
      "learning_rate": 9.828313112910887e-07,
      "loss": 0.7043705582618713,
      "step": 1868
    },
    {
      "epoch": 2.4243567126954195,
      "grad_norm": 0.7576665282249451,
      "learning_rate": 9.78578510410832e-07,
      "loss": 0.7256739139556885,
      "step": 1869
    },
    {
      "epoch": 2.4256544396455584,
      "grad_norm": 0.7524176239967346,
      "learning_rate": 9.743339323196827e-07,
      "loss": 0.708810031414032,
      "step": 1870
    },
    {
      "epoch": 2.4269521665956972,
      "grad_norm": 0.7555914521217346,
      "learning_rate": 9.700975856967287e-07,
      "loss": 0.6946004033088684,
      "step": 1871
    },
    {
      "epoch": 2.428249893545836,
      "grad_norm": 0.8047772645950317,
      "learning_rate": 9.658694792042284e-07,
      "loss": 0.7178459763526917,
      "step": 1872
    },
    {
      "epoch": 2.429547620495975,
      "grad_norm": 0.7859281897544861,
      "learning_rate": 9.616496214875847e-07,
      "loss": 0.6877878904342651,
      "step": 1873
    },
    {
      "epoch": 2.430845347446114,
      "grad_norm": 0.7489562034606934,
      "learning_rate": 9.574380211753442e-07,
      "loss": 0.6518415212631226,
      "step": 1874
    },
    {
      "epoch": 2.4321430743962527,
      "grad_norm": 0.7419933676719666,
      "learning_rate": 9.532346868791587e-07,
      "loss": 0.7486515641212463,
      "step": 1875
    },
    {
      "epoch": 2.433440801346392,
      "grad_norm": 0.7418408989906311,
      "learning_rate": 9.490396271937879e-07,
      "loss": 0.7523958086967468,
      "step": 1876
    },
    {
      "epoch": 2.4347385282965304,
      "grad_norm": 0.8075720071792603,
      "learning_rate": 9.448528506970628e-07,
      "loss": 0.6815677881240845,
      "step": 1877
    },
    {
      "epoch": 2.4360362552466697,
      "grad_norm": 0.7398018836975098,
      "learning_rate": 9.406743659498829e-07,
      "loss": 0.7089881896972656,
      "step": 1878
    },
    {
      "epoch": 2.4373339821968085,
      "grad_norm": 0.7852540016174316,
      "learning_rate": 9.365041814961928e-07,
      "loss": 0.6689083576202393,
      "step": 1879
    },
    {
      "epoch": 2.4386317091469474,
      "grad_norm": 0.7648181319236755,
      "learning_rate": 9.323423058629638e-07,
      "loss": 0.6946084499359131,
      "step": 1880
    },
    {
      "epoch": 2.4399294360970862,
      "grad_norm": 0.803335964679718,
      "learning_rate": 9.281887475601775e-07,
      "loss": 0.7401805520057678,
      "step": 1881
    },
    {
      "epoch": 2.441227163047225,
      "grad_norm": 0.7295672297477722,
      "learning_rate": 9.240435150808113e-07,
      "loss": 0.7000230550765991,
      "step": 1882
    },
    {
      "epoch": 2.442524889997364,
      "grad_norm": 0.7747576832771301,
      "learning_rate": 9.19906616900813e-07,
      "loss": 0.7612476348876953,
      "step": 1883
    },
    {
      "epoch": 2.443822616947503,
      "grad_norm": 0.7346387505531311,
      "learning_rate": 9.157780614790963e-07,
      "loss": 0.7002447843551636,
      "step": 1884
    },
    {
      "epoch": 2.4451203438976417,
      "grad_norm": 0.7575351595878601,
      "learning_rate": 9.116578572575091e-07,
      "loss": 0.7609678506851196,
      "step": 1885
    },
    {
      "epoch": 2.4464180708477805,
      "grad_norm": 0.7746056914329529,
      "learning_rate": 9.075460126608271e-07,
      "loss": 0.7523704767227173,
      "step": 1886
    },
    {
      "epoch": 2.44771579779792,
      "grad_norm": 0.745557427406311,
      "learning_rate": 9.034425360967319e-07,
      "loss": 0.7555814385414124,
      "step": 1887
    },
    {
      "epoch": 2.4490135247480587,
      "grad_norm": 0.7867469191551208,
      "learning_rate": 8.993474359557936e-07,
      "loss": 0.6881682276725769,
      "step": 1888
    },
    {
      "epoch": 2.4503112516981975,
      "grad_norm": 0.7527727484703064,
      "learning_rate": 8.952607206114588e-07,
      "loss": 0.7243494391441345,
      "step": 1889
    },
    {
      "epoch": 2.4516089786483364,
      "grad_norm": 0.7513570189476013,
      "learning_rate": 8.911823984200219e-07,
      "loss": 0.7017259001731873,
      "step": 1890
    },
    {
      "epoch": 2.4529067055984752,
      "grad_norm": 0.7603667974472046,
      "learning_rate": 8.871124777206213e-07,
      "loss": 0.7706648111343384,
      "step": 1891
    },
    {
      "epoch": 2.454204432548614,
      "grad_norm": 0.8234115839004517,
      "learning_rate": 8.83050966835215e-07,
      "loss": 0.7250049710273743,
      "step": 1892
    },
    {
      "epoch": 2.455502159498753,
      "grad_norm": 0.787039577960968,
      "learning_rate": 8.789978740685646e-07,
      "loss": 0.6726934909820557,
      "step": 1893
    },
    {
      "epoch": 2.456799886448892,
      "grad_norm": 0.7636567950248718,
      "learning_rate": 8.749532077082179e-07,
      "loss": 0.6961559057235718,
      "step": 1894
    },
    {
      "epoch": 2.4580976133990307,
      "grad_norm": 0.7643488049507141,
      "learning_rate": 8.709169760244968e-07,
      "loss": 0.7612512707710266,
      "step": 1895
    },
    {
      "epoch": 2.4593953403491695,
      "grad_norm": 0.7855246663093567,
      "learning_rate": 8.668891872704682e-07,
      "loss": 0.6644111275672913,
      "step": 1896
    },
    {
      "epoch": 2.4606930672993084,
      "grad_norm": 0.7694689631462097,
      "learning_rate": 8.628698496819471e-07,
      "loss": 0.6863073110580444,
      "step": 1897
    },
    {
      "epoch": 2.4619907942494477,
      "grad_norm": 0.734458327293396,
      "learning_rate": 8.58858971477457e-07,
      "loss": 0.777980387210846,
      "step": 1898
    },
    {
      "epoch": 2.4632885211995865,
      "grad_norm": 0.7516976594924927,
      "learning_rate": 8.548565608582299e-07,
      "loss": 0.7173190712928772,
      "step": 1899
    },
    {
      "epoch": 2.4645862481497254,
      "grad_norm": 0.7432575225830078,
      "learning_rate": 8.508626260081826e-07,
      "loss": 0.7015535235404968,
      "step": 1900
    },
    {
      "epoch": 2.4658839750998642,
      "grad_norm": 0.7617539763450623,
      "learning_rate": 8.468771750939009e-07,
      "loss": 0.7171428799629211,
      "step": 1901
    },
    {
      "epoch": 2.467181702050003,
      "grad_norm": 0.7451692819595337,
      "learning_rate": 8.429002162646233e-07,
      "loss": 0.708382785320282,
      "step": 1902
    },
    {
      "epoch": 2.468479429000142,
      "grad_norm": 0.7860012650489807,
      "learning_rate": 8.389317576522243e-07,
      "loss": 0.7629508376121521,
      "step": 1903
    },
    {
      "epoch": 2.469777155950281,
      "grad_norm": 0.7875171303749084,
      "learning_rate": 8.349718073711971e-07,
      "loss": 0.676824688911438,
      "step": 1904
    },
    {
      "epoch": 2.4710748829004197,
      "grad_norm": 0.7545873522758484,
      "learning_rate": 8.310203735186384e-07,
      "loss": 0.8134433627128601,
      "step": 1905
    },
    {
      "epoch": 2.4723726098505585,
      "grad_norm": 0.7484904527664185,
      "learning_rate": 8.270774641742275e-07,
      "loss": 0.7553561925888062,
      "step": 1906
    },
    {
      "epoch": 2.4736703368006974,
      "grad_norm": 0.7527977228164673,
      "learning_rate": 8.231430874002206e-07,
      "loss": 0.7250518798828125,
      "step": 1907
    },
    {
      "epoch": 2.4749680637508362,
      "grad_norm": 0.7821675539016724,
      "learning_rate": 8.192172512414187e-07,
      "loss": 0.699806809425354,
      "step": 1908
    },
    {
      "epoch": 2.4762657907009755,
      "grad_norm": 0.7531731128692627,
      "learning_rate": 8.152999637251641e-07,
      "loss": 0.6669670343399048,
      "step": 1909
    },
    {
      "epoch": 2.4775635176511144,
      "grad_norm": 0.7388619780540466,
      "learning_rate": 8.113912328613183e-07,
      "loss": 0.6381179094314575,
      "step": 1910
    },
    {
      "epoch": 2.4788612446012532,
      "grad_norm": 0.7556558847427368,
      "learning_rate": 8.074910666422475e-07,
      "loss": 0.6770901679992676,
      "step": 1911
    },
    {
      "epoch": 2.480158971551392,
      "grad_norm": 0.7798584699630737,
      "learning_rate": 8.035994730428031e-07,
      "loss": 0.7416208982467651,
      "step": 1912
    },
    {
      "epoch": 2.481456698501531,
      "grad_norm": 0.7625208497047424,
      "learning_rate": 7.997164600203111e-07,
      "loss": 0.6684930920600891,
      "step": 1913
    },
    {
      "epoch": 2.48275442545167,
      "grad_norm": 0.777184784412384,
      "learning_rate": 7.958420355145469e-07,
      "loss": 0.7176003456115723,
      "step": 1914
    },
    {
      "epoch": 2.4840521524018087,
      "grad_norm": 0.7437546849250793,
      "learning_rate": 7.919762074477311e-07,
      "loss": 0.6534523367881775,
      "step": 1915
    },
    {
      "epoch": 2.4853498793519475,
      "grad_norm": 0.7725997567176819,
      "learning_rate": 7.881189837245024e-07,
      "loss": 0.6518686413764954,
      "step": 1916
    },
    {
      "epoch": 2.4866476063020864,
      "grad_norm": 0.8138085007667542,
      "learning_rate": 7.842703722319073e-07,
      "loss": 0.7562498450279236,
      "step": 1917
    },
    {
      "epoch": 2.487945333252225,
      "grad_norm": 0.7612082958221436,
      "learning_rate": 7.804303808393831e-07,
      "loss": 0.6730899810791016,
      "step": 1918
    },
    {
      "epoch": 2.489243060202364,
      "grad_norm": 0.779849648475647,
      "learning_rate": 7.76599017398737e-07,
      "loss": 0.8186163902282715,
      "step": 1919
    },
    {
      "epoch": 2.4905407871525034,
      "grad_norm": 0.7707155346870422,
      "learning_rate": 7.727762897441421e-07,
      "loss": 0.6735822558403015,
      "step": 1920
    },
    {
      "epoch": 2.4918385141026422,
      "grad_norm": 0.7846003174781799,
      "learning_rate": 7.689622056921053e-07,
      "loss": 0.7295418381690979,
      "step": 1921
    },
    {
      "epoch": 2.493136241052781,
      "grad_norm": 0.7559599876403809,
      "learning_rate": 7.65156773041465e-07,
      "loss": 0.7741995453834534,
      "step": 1922
    },
    {
      "epoch": 2.49443396800292,
      "grad_norm": 0.791363000869751,
      "learning_rate": 7.613599995733667e-07,
      "loss": 0.7229835391044617,
      "step": 1923
    },
    {
      "epoch": 2.495731694953059,
      "grad_norm": 0.7598723769187927,
      "learning_rate": 7.575718930512516e-07,
      "loss": 0.6266255974769592,
      "step": 1924
    },
    {
      "epoch": 2.4970294219031977,
      "grad_norm": 0.7574074864387512,
      "learning_rate": 7.537924612208391e-07,
      "loss": 0.7148513793945312,
      "step": 1925
    },
    {
      "epoch": 2.4983271488533365,
      "grad_norm": 0.75996994972229,
      "learning_rate": 7.500217118101106e-07,
      "loss": 0.735736072063446,
      "step": 1926
    },
    {
      "epoch": 2.4996248758034754,
      "grad_norm": 0.7383106350898743,
      "learning_rate": 7.462596525292937e-07,
      "loss": 0.7258022427558899,
      "step": 1927
    },
    {
      "epoch": 2.500922602753614,
      "grad_norm": 0.7803204655647278,
      "learning_rate": 7.425062910708492e-07,
      "loss": 0.6895723342895508,
      "step": 1928
    },
    {
      "epoch": 2.5022203297037535,
      "grad_norm": 0.7795876264572144,
      "learning_rate": 7.387616351094473e-07,
      "loss": 0.6744094491004944,
      "step": 1929
    },
    {
      "epoch": 2.503518056653892,
      "grad_norm": 0.7904599905014038,
      "learning_rate": 7.350256923019666e-07,
      "loss": 0.6453460454940796,
      "step": 1930
    },
    {
      "epoch": 2.5048157836040312,
      "grad_norm": 0.7582366466522217,
      "learning_rate": 7.312984702874609e-07,
      "loss": 0.7346917390823364,
      "step": 1931
    },
    {
      "epoch": 2.50611351055417,
      "grad_norm": 0.7735522985458374,
      "learning_rate": 7.275799766871577e-07,
      "loss": 0.7309703826904297,
      "step": 1932
    },
    {
      "epoch": 2.507411237504309,
      "grad_norm": 0.8296225070953369,
      "learning_rate": 7.238702191044344e-07,
      "loss": 0.7618504166603088,
      "step": 1933
    },
    {
      "epoch": 2.508708964454448,
      "grad_norm": 0.7544898986816406,
      "learning_rate": 7.201692051248066e-07,
      "loss": 0.6686187386512756,
      "step": 1934
    },
    {
      "epoch": 2.5100066914045867,
      "grad_norm": 0.7645792365074158,
      "learning_rate": 7.164769423159113e-07,
      "loss": 0.7318588495254517,
      "step": 1935
    },
    {
      "epoch": 2.5113044183547255,
      "grad_norm": 0.8033185601234436,
      "learning_rate": 7.127934382274926e-07,
      "loss": 0.6869446635246277,
      "step": 1936
    },
    {
      "epoch": 2.5126021453048644,
      "grad_norm": 0.7405628561973572,
      "learning_rate": 7.091187003913802e-07,
      "loss": 0.7319486737251282,
      "step": 1937
    },
    {
      "epoch": 2.513899872255003,
      "grad_norm": 0.7678588032722473,
      "learning_rate": 7.054527363214875e-07,
      "loss": 0.7208606600761414,
      "step": 1938
    },
    {
      "epoch": 2.515197599205142,
      "grad_norm": 0.7541241645812988,
      "learning_rate": 7.017955535137788e-07,
      "loss": 0.710355818271637,
      "step": 1939
    },
    {
      "epoch": 2.5164953261552814,
      "grad_norm": 0.8112343549728394,
      "learning_rate": 6.981471594462719e-07,
      "loss": 0.7606580853462219,
      "step": 1940
    },
    {
      "epoch": 2.51779305310542,
      "grad_norm": 0.7551878690719604,
      "learning_rate": 6.945075615790059e-07,
      "loss": 0.7238883972167969,
      "step": 1941
    },
    {
      "epoch": 2.519090780055559,
      "grad_norm": 0.7674723863601685,
      "learning_rate": 6.908767673540384e-07,
      "loss": 0.7509013414382935,
      "step": 1942
    },
    {
      "epoch": 2.520388507005698,
      "grad_norm": 0.8111278414726257,
      "learning_rate": 6.872547841954241e-07,
      "loss": 0.7284641265869141,
      "step": 1943
    },
    {
      "epoch": 2.521686233955837,
      "grad_norm": 0.7864218354225159,
      "learning_rate": 6.836416195092021e-07,
      "loss": 0.7208187580108643,
      "step": 1944
    },
    {
      "epoch": 2.5229839609059757,
      "grad_norm": 0.7737844586372375,
      "learning_rate": 6.800372806833799e-07,
      "loss": 0.7160454988479614,
      "step": 1945
    },
    {
      "epoch": 2.5242816878561145,
      "grad_norm": 0.738142192363739,
      "learning_rate": 6.764417750879182e-07,
      "loss": 0.7124934196472168,
      "step": 1946
    },
    {
      "epoch": 2.5255794148062534,
      "grad_norm": 0.7452174425125122,
      "learning_rate": 6.728551100747155e-07,
      "loss": 0.7035548686981201,
      "step": 1947
    },
    {
      "epoch": 2.526877141756392,
      "grad_norm": 0.7617976069450378,
      "learning_rate": 6.692772929775943e-07,
      "loss": 0.7573691606521606,
      "step": 1948
    },
    {
      "epoch": 2.528174868706531,
      "grad_norm": 0.7748024463653564,
      "learning_rate": 6.657083311122858e-07,
      "loss": 0.7279133200645447,
      "step": 1949
    },
    {
      "epoch": 2.52947259565667,
      "grad_norm": 0.7774744033813477,
      "learning_rate": 6.621482317764105e-07,
      "loss": 0.6730536818504333,
      "step": 1950
    },
    {
      "epoch": 2.5307703226068092,
      "grad_norm": 0.7779364585876465,
      "learning_rate": 6.585970022494748e-07,
      "loss": 0.8089554309844971,
      "step": 1951
    },
    {
      "epoch": 2.5320680495569476,
      "grad_norm": 0.7630438804626465,
      "learning_rate": 6.550546497928401e-07,
      "loss": 0.6974608302116394,
      "step": 1952
    },
    {
      "epoch": 2.533365776507087,
      "grad_norm": 0.7758179306983948,
      "learning_rate": 6.515211816497247e-07,
      "loss": 0.7012693881988525,
      "step": 1953
    },
    {
      "epoch": 2.534663503457226,
      "grad_norm": 0.7640806436538696,
      "learning_rate": 6.479966050451736e-07,
      "loss": 0.6977714896202087,
      "step": 1954
    },
    {
      "epoch": 2.5359612304073647,
      "grad_norm": 0.7874693274497986,
      "learning_rate": 6.444809271860547e-07,
      "loss": 0.7300205826759338,
      "step": 1955
    },
    {
      "epoch": 2.5372589573575035,
      "grad_norm": 0.7789673805236816,
      "learning_rate": 6.409741552610399e-07,
      "loss": 0.7227955460548401,
      "step": 1956
    },
    {
      "epoch": 2.5385566843076424,
      "grad_norm": 0.7532787919044495,
      "learning_rate": 6.374762964405895e-07,
      "loss": 0.692006528377533,
      "step": 1957
    },
    {
      "epoch": 2.539854411257781,
      "grad_norm": 0.7724280953407288,
      "learning_rate": 6.339873578769401e-07,
      "loss": 0.715645432472229,
      "step": 1958
    },
    {
      "epoch": 2.54115213820792,
      "grad_norm": 0.7687965631484985,
      "learning_rate": 6.305073467040884e-07,
      "loss": 0.6982177495956421,
      "step": 1959
    },
    {
      "epoch": 2.542449865158059,
      "grad_norm": 0.7735276222229004,
      "learning_rate": 6.270362700377736e-07,
      "loss": 0.7986254692077637,
      "step": 1960
    },
    {
      "epoch": 2.543747592108198,
      "grad_norm": 0.7309845685958862,
      "learning_rate": 6.235741349754731e-07,
      "loss": 0.7566351890563965,
      "step": 1961
    },
    {
      "epoch": 2.545045319058337,
      "grad_norm": 0.7577828764915466,
      "learning_rate": 6.201209485963744e-07,
      "loss": 0.7308484315872192,
      "step": 1962
    },
    {
      "epoch": 2.546343046008476,
      "grad_norm": 0.7720513343811035,
      "learning_rate": 6.166767179613691e-07,
      "loss": 0.7158771753311157,
      "step": 1963
    },
    {
      "epoch": 2.547640772958615,
      "grad_norm": 0.7625830769538879,
      "learning_rate": 6.132414501130385e-07,
      "loss": 0.673353910446167,
      "step": 1964
    },
    {
      "epoch": 2.5489384999087537,
      "grad_norm": 0.754311740398407,
      "learning_rate": 6.098151520756357e-07,
      "loss": 0.7290247082710266,
      "step": 1965
    },
    {
      "epoch": 2.5502362268588925,
      "grad_norm": 0.7994941473007202,
      "learning_rate": 6.063978308550722e-07,
      "loss": 0.7406793832778931,
      "step": 1966
    },
    {
      "epoch": 2.5515339538090314,
      "grad_norm": 0.7528447508811951,
      "learning_rate": 6.029894934389058e-07,
      "loss": 0.7099549770355225,
      "step": 1967
    },
    {
      "epoch": 2.55283168075917,
      "grad_norm": 0.7489729523658752,
      "learning_rate": 5.995901467963228e-07,
      "loss": 0.630169689655304,
      "step": 1968
    },
    {
      "epoch": 2.554129407709309,
      "grad_norm": 0.8017000555992126,
      "learning_rate": 5.961997978781292e-07,
      "loss": 0.6770339012145996,
      "step": 1969
    },
    {
      "epoch": 2.555427134659448,
      "grad_norm": 0.7756592631340027,
      "learning_rate": 5.928184536167258e-07,
      "loss": 0.7437784671783447,
      "step": 1970
    },
    {
      "epoch": 2.5567248616095872,
      "grad_norm": 0.7770244479179382,
      "learning_rate": 5.89446120926111e-07,
      "loss": 0.6648685932159424,
      "step": 1971
    },
    {
      "epoch": 2.5580225885597256,
      "grad_norm": 0.782084584236145,
      "learning_rate": 5.860828067018481e-07,
      "loss": 0.7037726640701294,
      "step": 1972
    },
    {
      "epoch": 2.559320315509865,
      "grad_norm": 0.7928634881973267,
      "learning_rate": 5.82728517821064e-07,
      "loss": 0.7296996116638184,
      "step": 1973
    },
    {
      "epoch": 2.560618042460004,
      "grad_norm": 0.8191131353378296,
      "learning_rate": 5.793832611424322e-07,
      "loss": 0.7087576985359192,
      "step": 1974
    },
    {
      "epoch": 2.5619157694101427,
      "grad_norm": 0.7863832116127014,
      "learning_rate": 5.760470435061533e-07,
      "loss": 0.6903418898582458,
      "step": 1975
    },
    {
      "epoch": 2.5632134963602815,
      "grad_norm": 0.7777239084243774,
      "learning_rate": 5.727198717339511e-07,
      "loss": 0.7417172193527222,
      "step": 1976
    },
    {
      "epoch": 2.5645112233104204,
      "grad_norm": 0.7851623892784119,
      "learning_rate": 5.694017526290468e-07,
      "loss": 0.7211293578147888,
      "step": 1977
    },
    {
      "epoch": 2.565808950260559,
      "grad_norm": 0.7636955380439758,
      "learning_rate": 5.660926929761556e-07,
      "loss": 0.702522337436676,
      "step": 1978
    },
    {
      "epoch": 2.567106677210698,
      "grad_norm": 0.7507690787315369,
      "learning_rate": 5.627926995414662e-07,
      "loss": 0.6389472484588623,
      "step": 1979
    },
    {
      "epoch": 2.568404404160837,
      "grad_norm": 0.7766726016998291,
      "learning_rate": 5.59501779072631e-07,
      "loss": 0.7038560509681702,
      "step": 1980
    },
    {
      "epoch": 2.569702131110976,
      "grad_norm": 0.804925262928009,
      "learning_rate": 5.562199382987488e-07,
      "loss": 0.7217026352882385,
      "step": 1981
    },
    {
      "epoch": 2.570999858061115,
      "grad_norm": 0.731317937374115,
      "learning_rate": 5.529471839303541e-07,
      "loss": 0.6586759090423584,
      "step": 1982
    },
    {
      "epoch": 2.5722975850112535,
      "grad_norm": 0.7591803669929504,
      "learning_rate": 5.496835226593983e-07,
      "loss": 0.7452529072761536,
      "step": 1983
    },
    {
      "epoch": 2.573595311961393,
      "grad_norm": 0.7555832266807556,
      "learning_rate": 5.464289611592472e-07,
      "loss": 0.6921640634536743,
      "step": 1984
    },
    {
      "epoch": 2.5748930389115317,
      "grad_norm": 0.7568356990814209,
      "learning_rate": 5.431835060846519e-07,
      "loss": 0.7031154036521912,
      "step": 1985
    },
    {
      "epoch": 2.5761907658616705,
      "grad_norm": 0.7693030834197998,
      "learning_rate": 5.399471640717479e-07,
      "loss": 0.7538740634918213,
      "step": 1986
    },
    {
      "epoch": 2.5774884928118094,
      "grad_norm": 0.7669311165809631,
      "learning_rate": 5.367199417380347e-07,
      "loss": 0.6759027242660522,
      "step": 1987
    },
    {
      "epoch": 2.578786219761948,
      "grad_norm": 0.8306924104690552,
      "learning_rate": 5.335018456823665e-07,
      "loss": 0.7569330930709839,
      "step": 1988
    },
    {
      "epoch": 2.580083946712087,
      "grad_norm": 0.7967241406440735,
      "learning_rate": 5.302928824849335e-07,
      "loss": 0.7751659154891968,
      "step": 1989
    },
    {
      "epoch": 2.581381673662226,
      "grad_norm": 0.7510127425193787,
      "learning_rate": 5.270930587072548e-07,
      "loss": 0.6661146283149719,
      "step": 1990
    },
    {
      "epoch": 2.582679400612365,
      "grad_norm": 0.7510029673576355,
      "learning_rate": 5.239023808921595e-07,
      "loss": 0.6773824095726013,
      "step": 1991
    },
    {
      "epoch": 2.5839771275625036,
      "grad_norm": 0.7623904347419739,
      "learning_rate": 5.207208555637767e-07,
      "loss": 0.7609443664550781,
      "step": 1992
    },
    {
      "epoch": 2.585274854512643,
      "grad_norm": 0.7627770304679871,
      "learning_rate": 5.175484892275184e-07,
      "loss": 0.7122793197631836,
      "step": 1993
    },
    {
      "epoch": 2.5865725814627814,
      "grad_norm": 0.7585700154304504,
      "learning_rate": 5.14385288370074e-07,
      "loss": 0.743561863899231,
      "step": 1994
    },
    {
      "epoch": 2.5878703084129207,
      "grad_norm": 0.738996684551239,
      "learning_rate": 5.11231259459386e-07,
      "loss": 0.6380658745765686,
      "step": 1995
    },
    {
      "epoch": 2.5891680353630595,
      "grad_norm": 0.7691276669502258,
      "learning_rate": 5.080864089446464e-07,
      "loss": 0.6471886038780212,
      "step": 1996
    },
    {
      "epoch": 2.5904657623131984,
      "grad_norm": 0.7550168037414551,
      "learning_rate": 5.049507432562778e-07,
      "loss": 0.673279881477356,
      "step": 1997
    },
    {
      "epoch": 2.591763489263337,
      "grad_norm": 0.7882893085479736,
      "learning_rate": 5.018242688059238e-07,
      "loss": 0.786373496055603,
      "step": 1998
    },
    {
      "epoch": 2.593061216213476,
      "grad_norm": 0.7830791473388672,
      "learning_rate": 4.987069919864329e-07,
      "loss": 0.7720410823822021,
      "step": 1999
    },
    {
      "epoch": 2.594358943163615,
      "grad_norm": 0.7811652421951294,
      "learning_rate": 4.95598919171848e-07,
      "loss": 0.7637435793876648,
      "step": 2000
    },
    {
      "epoch": 2.595656670113754,
      "grad_norm": 0.771362841129303,
      "learning_rate": 4.925000567173882e-07,
      "loss": 0.7330818772315979,
      "step": 2001
    },
    {
      "epoch": 2.5969543970638926,
      "grad_norm": 0.7689074277877808,
      "learning_rate": 4.894104109594466e-07,
      "loss": 0.677016019821167,
      "step": 2002
    },
    {
      "epoch": 2.5982521240140315,
      "grad_norm": 0.7591285109519958,
      "learning_rate": 4.863299882155659e-07,
      "loss": 0.6757215261459351,
      "step": 2003
    },
    {
      "epoch": 2.599549850964171,
      "grad_norm": 0.768108069896698,
      "learning_rate": 4.832587947844297e-07,
      "loss": 0.6731127500534058,
      "step": 2004
    },
    {
      "epoch": 2.600847577914309,
      "grad_norm": 0.7346110939979553,
      "learning_rate": 4.801968369458531e-07,
      "loss": 0.7469443678855896,
      "step": 2005
    },
    {
      "epoch": 2.6021453048644485,
      "grad_norm": 0.7801742553710938,
      "learning_rate": 4.771441209607625e-07,
      "loss": 0.7497217059135437,
      "step": 2006
    },
    {
      "epoch": 2.6034430318145874,
      "grad_norm": 0.7419130206108093,
      "learning_rate": 4.7410065307119167e-07,
      "loss": 0.7378045916557312,
      "step": 2007
    },
    {
      "epoch": 2.604740758764726,
      "grad_norm": 0.7835217118263245,
      "learning_rate": 4.7106643950026067e-07,
      "loss": 0.7161265015602112,
      "step": 2008
    },
    {
      "epoch": 2.606038485714865,
      "grad_norm": 0.7818168997764587,
      "learning_rate": 4.6804148645216873e-07,
      "loss": 0.7108198404312134,
      "step": 2009
    },
    {
      "epoch": 2.607336212665004,
      "grad_norm": 0.7902902364730835,
      "learning_rate": 4.6502580011217934e-07,
      "loss": 0.7278609871864319,
      "step": 2010
    },
    {
      "epoch": 2.608633939615143,
      "grad_norm": 0.7904089093208313,
      "learning_rate": 4.6201938664660775e-07,
      "loss": 0.7432711720466614,
      "step": 2011
    },
    {
      "epoch": 2.6099316665652816,
      "grad_norm": 0.7738637328147888,
      "learning_rate": 4.590222522028082e-07,
      "loss": 0.7238620519638062,
      "step": 2012
    },
    {
      "epoch": 2.6112293935154205,
      "grad_norm": 0.7708808779716492,
      "learning_rate": 4.5603440290916347e-07,
      "loss": 0.7561792135238647,
      "step": 2013
    },
    {
      "epoch": 2.6125271204655593,
      "grad_norm": 0.7780309915542603,
      "learning_rate": 4.5305584487506605e-07,
      "loss": 0.7579943537712097,
      "step": 2014
    },
    {
      "epoch": 2.6138248474156986,
      "grad_norm": 0.7369161248207092,
      "learning_rate": 4.500865841909169e-07,
      "loss": 0.7089678049087524,
      "step": 2015
    },
    {
      "epoch": 2.6151225743658375,
      "grad_norm": 0.8075423240661621,
      "learning_rate": 4.471266269280994e-07,
      "loss": 0.7907909750938416,
      "step": 2016
    },
    {
      "epoch": 2.6164203013159764,
      "grad_norm": 0.797129213809967,
      "learning_rate": 4.441759791389799e-07,
      "loss": 0.7853899598121643,
      "step": 2017
    },
    {
      "epoch": 2.617718028266115,
      "grad_norm": 0.8074013590812683,
      "learning_rate": 4.41234646856884e-07,
      "loss": 0.6692730188369751,
      "step": 2018
    },
    {
      "epoch": 2.619015755216254,
      "grad_norm": 0.7978529930114746,
      "learning_rate": 4.383026360960929e-07,
      "loss": 0.7197666168212891,
      "step": 2019
    },
    {
      "epoch": 2.620313482166393,
      "grad_norm": 0.7521398067474365,
      "learning_rate": 4.35379952851826e-07,
      "loss": 0.6917500495910645,
      "step": 2020
    },
    {
      "epoch": 2.621611209116532,
      "grad_norm": 0.7865820527076721,
      "learning_rate": 4.324666031002311e-07,
      "loss": 0.6865369081497192,
      "step": 2021
    },
    {
      "epoch": 2.6229089360666706,
      "grad_norm": 0.758794903755188,
      "learning_rate": 4.29562592798371e-07,
      "loss": 0.7486015558242798,
      "step": 2022
    },
    {
      "epoch": 2.6242066630168095,
      "grad_norm": 0.7740626931190491,
      "learning_rate": 4.266679278842123e-07,
      "loss": 0.7256302237510681,
      "step": 2023
    },
    {
      "epoch": 2.625504389966949,
      "grad_norm": 0.7756375074386597,
      "learning_rate": 4.2378261427660994e-07,
      "loss": 0.7277344465255737,
      "step": 2024
    },
    {
      "epoch": 2.626802116917087,
      "grad_norm": 0.7711913585662842,
      "learning_rate": 4.209066578753035e-07,
      "loss": 0.683953583240509,
      "step": 2025
    },
    {
      "epoch": 2.6280998438672265,
      "grad_norm": 0.7805073857307434,
      "learning_rate": 4.1804006456089174e-07,
      "loss": 0.6939178109169006,
      "step": 2026
    },
    {
      "epoch": 2.6293975708173654,
      "grad_norm": 0.7380502223968506,
      "learning_rate": 4.1518284019483655e-07,
      "loss": 0.6786249279975891,
      "step": 2027
    },
    {
      "epoch": 2.630695297767504,
      "grad_norm": 0.7325124740600586,
      "learning_rate": 4.123349906194357e-07,
      "loss": 0.6845241785049438,
      "step": 2028
    },
    {
      "epoch": 2.631993024717643,
      "grad_norm": 0.7694509625434875,
      "learning_rate": 4.094965216578212e-07,
      "loss": 0.7099627256393433,
      "step": 2029
    },
    {
      "epoch": 2.633290751667782,
      "grad_norm": 0.7750458717346191,
      "learning_rate": 4.066674391139458e-07,
      "loss": 0.7063329219818115,
      "step": 2030
    },
    {
      "epoch": 2.634588478617921,
      "grad_norm": 0.772454559803009,
      "learning_rate": 4.038477487725645e-07,
      "loss": 0.6330828666687012,
      "step": 2031
    },
    {
      "epoch": 2.6358862055680596,
      "grad_norm": 0.792510449886322,
      "learning_rate": 4.0103745639923144e-07,
      "loss": 0.7432264685630798,
      "step": 2032
    },
    {
      "epoch": 2.6371839325181985,
      "grad_norm": 0.7578525543212891,
      "learning_rate": 3.9823656774028386e-07,
      "loss": 0.6670839786529541,
      "step": 2033
    },
    {
      "epoch": 2.6384816594683373,
      "grad_norm": 0.7778390049934387,
      "learning_rate": 3.9544508852282895e-07,
      "loss": 0.7466758489608765,
      "step": 2034
    },
    {
      "epoch": 2.6397793864184766,
      "grad_norm": 0.7715905904769897,
      "learning_rate": 3.9266302445473634e-07,
      "loss": 0.6792566776275635,
      "step": 2035
    },
    {
      "epoch": 2.641077113368615,
      "grad_norm": 0.8023421764373779,
      "learning_rate": 3.89890381224623e-07,
      "loss": 0.7303230166435242,
      "step": 2036
    },
    {
      "epoch": 2.6423748403187544,
      "grad_norm": 0.7726515531539917,
      "learning_rate": 3.8712716450183985e-07,
      "loss": 0.6872059106826782,
      "step": 2037
    },
    {
      "epoch": 2.643672567268893,
      "grad_norm": 0.7481380701065063,
      "learning_rate": 3.8437337993647017e-07,
      "loss": 0.7442598342895508,
      "step": 2038
    },
    {
      "epoch": 2.644970294219032,
      "grad_norm": 0.73259437084198,
      "learning_rate": 3.81629033159302e-07,
      "loss": 0.6850177049636841,
      "step": 2039
    },
    {
      "epoch": 2.646268021169171,
      "grad_norm": 0.7581216096878052,
      "learning_rate": 3.7889412978183324e-07,
      "loss": 0.7565430998802185,
      "step": 2040
    },
    {
      "epoch": 2.64756574811931,
      "grad_norm": 0.7624466419219971,
      "learning_rate": 3.7616867539624733e-07,
      "loss": 0.681561291217804,
      "step": 2041
    },
    {
      "epoch": 2.6488634750694486,
      "grad_norm": 0.7603456974029541,
      "learning_rate": 3.734526755754092e-07,
      "loss": 0.6617370247840881,
      "step": 2042
    },
    {
      "epoch": 2.6501612020195875,
      "grad_norm": 0.815494179725647,
      "learning_rate": 3.707461358728509e-07,
      "loss": 0.791693925857544,
      "step": 2043
    },
    {
      "epoch": 2.6514589289697263,
      "grad_norm": 0.784187376499176,
      "learning_rate": 3.680490618227611e-07,
      "loss": 0.7595443725585938,
      "step": 2044
    },
    {
      "epoch": 2.652756655919865,
      "grad_norm": 0.7431678771972656,
      "learning_rate": 3.6536145893997346e-07,
      "loss": 0.7058426737785339,
      "step": 2045
    },
    {
      "epoch": 2.6540543828700045,
      "grad_norm": 0.7702705264091492,
      "learning_rate": 3.626833327199564e-07,
      "loss": 0.7345145344734192,
      "step": 2046
    },
    {
      "epoch": 2.655352109820143,
      "grad_norm": 0.799102783203125,
      "learning_rate": 3.600146886387984e-07,
      "loss": 0.7478254437446594,
      "step": 2047
    },
    {
      "epoch": 2.656649836770282,
      "grad_norm": 0.7557673454284668,
      "learning_rate": 3.573555321532035e-07,
      "loss": 0.7010441422462463,
      "step": 2048
    },
    {
      "epoch": 2.657947563720421,
      "grad_norm": 0.8038543462753296,
      "learning_rate": 3.547058687004723e-07,
      "loss": 0.739691972732544,
      "step": 2049
    },
    {
      "epoch": 2.65924529067056,
      "grad_norm": 0.7559893727302551,
      "learning_rate": 3.520657036984959e-07,
      "loss": 0.6943920850753784,
      "step": 2050
    },
    {
      "epoch": 2.660543017620699,
      "grad_norm": 0.7952106595039368,
      "learning_rate": 3.494350425457438e-07,
      "loss": 0.6908611059188843,
      "step": 2051
    },
    {
      "epoch": 2.6618407445708376,
      "grad_norm": 0.7683407664299011,
      "learning_rate": 3.46813890621252e-07,
      "loss": 0.7314857840538025,
      "step": 2052
    },
    {
      "epoch": 2.6631384715209765,
      "grad_norm": 0.7509656548500061,
      "learning_rate": 3.4420225328461286e-07,
      "loss": 0.7266790866851807,
      "step": 2053
    },
    {
      "epoch": 2.6644361984711153,
      "grad_norm": 0.793124794960022,
      "learning_rate": 3.416001358759635e-07,
      "loss": 0.7625051736831665,
      "step": 2054
    },
    {
      "epoch": 2.665733925421254,
      "grad_norm": 0.7964074611663818,
      "learning_rate": 3.390075437159762e-07,
      "loss": 0.7503449320793152,
      "step": 2055
    },
    {
      "epoch": 2.667031652371393,
      "grad_norm": 0.7679505944252014,
      "learning_rate": 3.36424482105846e-07,
      "loss": 0.7547749876976013,
      "step": 2056
    },
    {
      "epoch": 2.6683293793215324,
      "grad_norm": 0.7538975477218628,
      "learning_rate": 3.338509563272774e-07,
      "loss": 0.6920310258865356,
      "step": 2057
    },
    {
      "epoch": 2.6696271062716708,
      "grad_norm": 0.7652698159217834,
      "learning_rate": 3.3128697164248213e-07,
      "loss": 0.7433255314826965,
      "step": 2058
    },
    {
      "epoch": 2.67092483322181,
      "grad_norm": 0.780931830406189,
      "learning_rate": 3.2873253329415986e-07,
      "loss": 0.7661755084991455,
      "step": 2059
    },
    {
      "epoch": 2.672222560171949,
      "grad_norm": 0.7813259959220886,
      "learning_rate": 3.2618764650548806e-07,
      "loss": 0.7323101162910461,
      "step": 2060
    },
    {
      "epoch": 2.673520287122088,
      "grad_norm": 0.7653396725654602,
      "learning_rate": 3.236523164801192e-07,
      "loss": 0.6384355425834656,
      "step": 2061
    },
    {
      "epoch": 2.6748180140722266,
      "grad_norm": 0.7511436343193054,
      "learning_rate": 3.2112654840215863e-07,
      "loss": 0.76376873254776,
      "step": 2062
    },
    {
      "epoch": 2.6761157410223655,
      "grad_norm": 0.8071112036705017,
      "learning_rate": 3.186103474361646e-07,
      "loss": 0.7349339127540588,
      "step": 2063
    },
    {
      "epoch": 2.6774134679725043,
      "grad_norm": 0.7712031006813049,
      "learning_rate": 3.161037187271304e-07,
      "loss": 0.7209377288818359,
      "step": 2064
    },
    {
      "epoch": 2.678711194922643,
      "grad_norm": 0.7776238918304443,
      "learning_rate": 3.136066674004773e-07,
      "loss": 0.7230247855186462,
      "step": 2065
    },
    {
      "epoch": 2.680008921872782,
      "grad_norm": 0.7257099747657776,
      "learning_rate": 3.1111919856204373e-07,
      "loss": 0.6789555549621582,
      "step": 2066
    },
    {
      "epoch": 2.681306648822921,
      "grad_norm": 0.7619670629501343,
      "learning_rate": 3.08641317298074e-07,
      "loss": 0.7058009505271912,
      "step": 2067
    },
    {
      "epoch": 2.68260437577306,
      "grad_norm": 0.7894755601882935,
      "learning_rate": 3.0617302867520736e-07,
      "loss": 0.691407322883606,
      "step": 2068
    },
    {
      "epoch": 2.6839021027231986,
      "grad_norm": 0.7572270631790161,
      "learning_rate": 3.0371433774047056e-07,
      "loss": 0.7379584312438965,
      "step": 2069
    },
    {
      "epoch": 2.685199829673338,
      "grad_norm": 0.7616877555847168,
      "learning_rate": 3.0126524952126203e-07,
      "loss": 0.7380081415176392,
      "step": 2070
    },
    {
      "epoch": 2.685199829673338,
      "eval_loss": 0.7602905035018921,
      "eval_runtime": 139.4785,
      "eval_samples_per_second": 37.224,
      "eval_steps_per_second": 9.306,
      "step": 2070
    },
    {
      "epoch": 2.686497556623477,
      "grad_norm": 0.7607947587966919,
      "learning_rate": 2.988257690253504e-07,
      "loss": 0.7469013333320618,
      "step": 2071
    },
    {
      "epoch": 2.6877952835736156,
      "grad_norm": 0.772251546382904,
      "learning_rate": 2.9639590124085296e-07,
      "loss": 0.7996762990951538,
      "step": 2072
    },
    {
      "epoch": 2.6890930105237545,
      "grad_norm": 0.7398597002029419,
      "learning_rate": 2.939756511362357e-07,
      "loss": 0.6718044281005859,
      "step": 2073
    },
    {
      "epoch": 2.6903907374738933,
      "grad_norm": 0.7646867632865906,
      "learning_rate": 2.915650236602974e-07,
      "loss": 0.7430347204208374,
      "step": 2074
    },
    {
      "epoch": 2.691688464424032,
      "grad_norm": 0.7866479158401489,
      "learning_rate": 2.891640237421611e-07,
      "loss": 0.7387747168540955,
      "step": 2075
    },
    {
      "epoch": 2.692986191374171,
      "grad_norm": 0.7951492071151733,
      "learning_rate": 2.8677265629126373e-07,
      "loss": 0.7158300280570984,
      "step": 2076
    },
    {
      "epoch": 2.69428391832431,
      "grad_norm": 0.7993662357330322,
      "learning_rate": 2.8439092619734655e-07,
      "loss": 0.7340144515037537,
      "step": 2077
    },
    {
      "epoch": 2.6955816452744488,
      "grad_norm": 0.7797279953956604,
      "learning_rate": 2.820188383304451e-07,
      "loss": 0.6969457864761353,
      "step": 2078
    },
    {
      "epoch": 2.696879372224588,
      "grad_norm": 0.7600167989730835,
      "learning_rate": 2.7965639754087893e-07,
      "loss": 0.7070989608764648,
      "step": 2079
    },
    {
      "epoch": 2.698177099174727,
      "grad_norm": 0.7879111170768738,
      "learning_rate": 2.7730360865923954e-07,
      "loss": 0.7690889239311218,
      "step": 2080
    },
    {
      "epoch": 2.699474826124866,
      "grad_norm": 0.7314602136611938,
      "learning_rate": 2.7496047649638757e-07,
      "loss": 0.6755863428115845,
      "step": 2081
    },
    {
      "epoch": 2.7007725530750046,
      "grad_norm": 0.780437707901001,
      "learning_rate": 2.726270058434327e-07,
      "loss": 0.7809361815452576,
      "step": 2082
    },
    {
      "epoch": 2.7020702800251435,
      "grad_norm": 0.7712891101837158,
      "learning_rate": 2.703032014717333e-07,
      "loss": 0.791184663772583,
      "step": 2083
    },
    {
      "epoch": 2.7033680069752823,
      "grad_norm": 0.7847064137458801,
      "learning_rate": 2.6798906813288117e-07,
      "loss": 0.7188366055488586,
      "step": 2084
    },
    {
      "epoch": 2.704665733925421,
      "grad_norm": 0.7422674298286438,
      "learning_rate": 2.656846105586919e-07,
      "loss": 0.7355191707611084,
      "step": 2085
    },
    {
      "epoch": 2.70596346087556,
      "grad_norm": 0.7793022394180298,
      "learning_rate": 2.633898334611995e-07,
      "loss": 0.7662262916564941,
      "step": 2086
    },
    {
      "epoch": 2.707261187825699,
      "grad_norm": 0.7930620312690735,
      "learning_rate": 2.6110474153264176e-07,
      "loss": 0.707592785358429,
      "step": 2087
    },
    {
      "epoch": 2.708558914775838,
      "grad_norm": 0.7076685428619385,
      "learning_rate": 2.588293394454533e-07,
      "loss": 0.6750243902206421,
      "step": 2088
    },
    {
      "epoch": 2.7098566417259766,
      "grad_norm": 0.7700866460800171,
      "learning_rate": 2.565636318522552e-07,
      "loss": 0.7775879502296448,
      "step": 2089
    },
    {
      "epoch": 2.711154368676116,
      "grad_norm": 0.7388302087783813,
      "learning_rate": 2.543076233858466e-07,
      "loss": 0.692900538444519,
      "step": 2090
    },
    {
      "epoch": 2.7124520956262548,
      "grad_norm": 0.7761010527610779,
      "learning_rate": 2.5206131865919303e-07,
      "loss": 0.7216837406158447,
      "step": 2091
    },
    {
      "epoch": 2.7137498225763936,
      "grad_norm": 0.7585817575454712,
      "learning_rate": 2.4982472226542045e-07,
      "loss": 0.6682706475257874,
      "step": 2092
    },
    {
      "epoch": 2.7150475495265325,
      "grad_norm": 0.76124107837677,
      "learning_rate": 2.475978387778e-07,
      "loss": 0.7049241065979004,
      "step": 2093
    },
    {
      "epoch": 2.7163452764766713,
      "grad_norm": 0.7612197995185852,
      "learning_rate": 2.453806727497482e-07,
      "loss": 0.7023073434829712,
      "step": 2094
    },
    {
      "epoch": 2.71764300342681,
      "grad_norm": 0.7852765321731567,
      "learning_rate": 2.431732287148053e-07,
      "loss": 0.7443613409996033,
      "step": 2095
    },
    {
      "epoch": 2.718940730376949,
      "grad_norm": 0.7986307144165039,
      "learning_rate": 2.409755111866369e-07,
      "loss": 0.7830970287322998,
      "step": 2096
    },
    {
      "epoch": 2.720238457327088,
      "grad_norm": 0.797486424446106,
      "learning_rate": 2.387875246590193e-07,
      "loss": 0.6837291717529297,
      "step": 2097
    },
    {
      "epoch": 2.7215361842772268,
      "grad_norm": 0.7682679295539856,
      "learning_rate": 2.3660927360583064e-07,
      "loss": 0.7355650663375854,
      "step": 2098
    },
    {
      "epoch": 2.722833911227366,
      "grad_norm": 0.7314669489860535,
      "learning_rate": 2.3444076248104297e-07,
      "loss": 0.6166611313819885,
      "step": 2099
    },
    {
      "epoch": 2.7241316381775045,
      "grad_norm": 0.7601985931396484,
      "learning_rate": 2.322819957187139e-07,
      "loss": 0.7629340291023254,
      "step": 2100
    },
    {
      "epoch": 2.7254293651276438,
      "grad_norm": 0.7485438585281372,
      "learning_rate": 2.3013297773297306e-07,
      "loss": 0.6586056351661682,
      "step": 2101
    },
    {
      "epoch": 2.7267270920777826,
      "grad_norm": 0.8016761541366577,
      "learning_rate": 2.279937129180204e-07,
      "loss": 0.7201554179191589,
      "step": 2102
    },
    {
      "epoch": 2.7280248190279215,
      "grad_norm": 0.7577768564224243,
      "learning_rate": 2.2586420564810863e-07,
      "loss": 0.7080650329589844,
      "step": 2103
    },
    {
      "epoch": 2.7293225459780603,
      "grad_norm": 0.7772164344787598,
      "learning_rate": 2.2374446027754405e-07,
      "loss": 0.7283114790916443,
      "step": 2104
    },
    {
      "epoch": 2.730620272928199,
      "grad_norm": 0.7661550641059875,
      "learning_rate": 2.2163448114066677e-07,
      "loss": 0.70591801404953,
      "step": 2105
    },
    {
      "epoch": 2.731917999878338,
      "grad_norm": 0.7516199946403503,
      "learning_rate": 2.1953427255185122e-07,
      "loss": 0.7168128490447998,
      "step": 2106
    },
    {
      "epoch": 2.733215726828477,
      "grad_norm": 0.7863711714744568,
      "learning_rate": 2.174438388054928e-07,
      "loss": 0.7204221487045288,
      "step": 2107
    },
    {
      "epoch": 2.7345134537786158,
      "grad_norm": 0.7704216837882996,
      "learning_rate": 2.1536318417599844e-07,
      "loss": 0.6871287822723389,
      "step": 2108
    },
    {
      "epoch": 2.7358111807287546,
      "grad_norm": 0.776828944683075,
      "learning_rate": 2.1329231291778108e-07,
      "loss": 0.7529214024543762,
      "step": 2109
    },
    {
      "epoch": 2.737108907678894,
      "grad_norm": 0.7328927516937256,
      "learning_rate": 2.1123122926524853e-07,
      "loss": 0.6814893484115601,
      "step": 2110
    },
    {
      "epoch": 2.7384066346290323,
      "grad_norm": 0.8207976222038269,
      "learning_rate": 2.0917993743279297e-07,
      "loss": 0.6837886571884155,
      "step": 2111
    },
    {
      "epoch": 2.7397043615791716,
      "grad_norm": 0.8150394558906555,
      "learning_rate": 2.0713844161479035e-07,
      "loss": 0.722987949848175,
      "step": 2112
    },
    {
      "epoch": 2.7410020885293105,
      "grad_norm": 0.7454245686531067,
      "learning_rate": 2.0510674598558045e-07,
      "loss": 0.6830301880836487,
      "step": 2113
    },
    {
      "epoch": 2.7422998154794493,
      "grad_norm": 0.757875919342041,
      "learning_rate": 2.0308485469946736e-07,
      "loss": 0.7439548373222351,
      "step": 2114
    },
    {
      "epoch": 2.743597542429588,
      "grad_norm": 0.7572675347328186,
      "learning_rate": 2.010727718907074e-07,
      "loss": 0.668762743473053,
      "step": 2115
    },
    {
      "epoch": 2.744895269379727,
      "grad_norm": 0.7444978356361389,
      "learning_rate": 1.9907050167349894e-07,
      "loss": 0.6855237483978271,
      "step": 2116
    },
    {
      "epoch": 2.746192996329866,
      "grad_norm": 0.7447678446769714,
      "learning_rate": 1.9707804814198096e-07,
      "loss": 0.6730313301086426,
      "step": 2117
    },
    {
      "epoch": 2.7474907232800048,
      "grad_norm": 0.7300037145614624,
      "learning_rate": 1.9509541537021392e-07,
      "loss": 0.6463099122047424,
      "step": 2118
    },
    {
      "epoch": 2.7487884502301436,
      "grad_norm": 0.7642115354537964,
      "learning_rate": 1.9312260741218114e-07,
      "loss": 0.6817460656166077,
      "step": 2119
    },
    {
      "epoch": 2.7500861771802825,
      "grad_norm": 0.7782988548278809,
      "learning_rate": 1.911596283017747e-07,
      "loss": 0.7150259613990784,
      "step": 2120
    },
    {
      "epoch": 2.7513839041304218,
      "grad_norm": 0.7755700945854187,
      "learning_rate": 1.8920648205279113e-07,
      "loss": 0.6812229156494141,
      "step": 2121
    },
    {
      "epoch": 2.75268163108056,
      "grad_norm": 0.7792496681213379,
      "learning_rate": 1.8726317265891968e-07,
      "loss": 0.7230182886123657,
      "step": 2122
    },
    {
      "epoch": 2.7539793580306995,
      "grad_norm": 0.7651738524436951,
      "learning_rate": 1.8532970409373684e-07,
      "loss": 0.6976583003997803,
      "step": 2123
    },
    {
      "epoch": 2.7552770849808383,
      "grad_norm": 0.7811776995658875,
      "learning_rate": 1.8340608031069462e-07,
      "loss": 0.7523854970932007,
      "step": 2124
    },
    {
      "epoch": 2.756574811930977,
      "grad_norm": 0.7669888138771057,
      "learning_rate": 1.8149230524311944e-07,
      "loss": 0.7343149781227112,
      "step": 2125
    },
    {
      "epoch": 2.757872538881116,
      "grad_norm": 0.7220097780227661,
      "learning_rate": 1.7958838280419387e-07,
      "loss": 0.6710058450698853,
      "step": 2126
    },
    {
      "epoch": 2.759170265831255,
      "grad_norm": 0.7744547724723816,
      "learning_rate": 1.7769431688696048e-07,
      "loss": 0.6996723413467407,
      "step": 2127
    },
    {
      "epoch": 2.7604679927813938,
      "grad_norm": 0.7946389317512512,
      "learning_rate": 1.7581011136430238e-07,
      "loss": 0.7537875771522522,
      "step": 2128
    },
    {
      "epoch": 2.7617657197315326,
      "grad_norm": 0.7883028388023376,
      "learning_rate": 1.739357700889438e-07,
      "loss": 0.732083261013031,
      "step": 2129
    },
    {
      "epoch": 2.7630634466816715,
      "grad_norm": 0.8040179014205933,
      "learning_rate": 1.720712968934385e-07,
      "loss": 0.8135465383529663,
      "step": 2130
    },
    {
      "epoch": 2.7643611736318103,
      "grad_norm": 0.7717416286468506,
      "learning_rate": 1.7021669559016184e-07,
      "loss": 0.696827232837677,
      "step": 2131
    },
    {
      "epoch": 2.7656589005819496,
      "grad_norm": 0.7489641308784485,
      "learning_rate": 1.6837196997130434e-07,
      "loss": 0.7431339025497437,
      "step": 2132
    },
    {
      "epoch": 2.7669566275320885,
      "grad_norm": 0.7535181045532227,
      "learning_rate": 1.6653712380886366e-07,
      "loss": 0.7000592350959778,
      "step": 2133
    },
    {
      "epoch": 2.7682543544822273,
      "grad_norm": 0.7728670239448547,
      "learning_rate": 1.6471216085463372e-07,
      "loss": 0.6311807036399841,
      "step": 2134
    },
    {
      "epoch": 2.769552081432366,
      "grad_norm": 0.7892481684684753,
      "learning_rate": 1.6289708484020395e-07,
      "loss": 0.7196534276008606,
      "step": 2135
    },
    {
      "epoch": 2.770849808382505,
      "grad_norm": 0.7621094584465027,
      "learning_rate": 1.6109189947694448e-07,
      "loss": 0.6594611406326294,
      "step": 2136
    },
    {
      "epoch": 2.772147535332644,
      "grad_norm": 0.8210483193397522,
      "learning_rate": 1.5929660845600215e-07,
      "loss": 0.736477255821228,
      "step": 2137
    },
    {
      "epoch": 2.7734452622827828,
      "grad_norm": 0.7322575449943542,
      "learning_rate": 1.575112154482933e-07,
      "loss": 0.7200109362602234,
      "step": 2138
    },
    {
      "epoch": 2.7747429892329216,
      "grad_norm": 0.7946739196777344,
      "learning_rate": 1.557357241044949e-07,
      "loss": 0.7807417511940002,
      "step": 2139
    },
    {
      "epoch": 2.7760407161830605,
      "grad_norm": 0.7464922666549683,
      "learning_rate": 1.539701380550368e-07,
      "loss": 0.6590375304222107,
      "step": 2140
    },
    {
      "epoch": 2.7773384431331998,
      "grad_norm": 0.774113118648529,
      "learning_rate": 1.5221446091009618e-07,
      "loss": 0.5982813239097595,
      "step": 2141
    },
    {
      "epoch": 2.778636170083338,
      "grad_norm": 0.8261796832084656,
      "learning_rate": 1.504686962595875e-07,
      "loss": 0.7688677906990051,
      "step": 2142
    },
    {
      "epoch": 2.7799338970334775,
      "grad_norm": 0.778244137763977,
      "learning_rate": 1.4873284767315864e-07,
      "loss": 0.6353880167007446,
      "step": 2143
    },
    {
      "epoch": 2.7812316239836163,
      "grad_norm": 0.7616711854934692,
      "learning_rate": 1.4700691870017991e-07,
      "loss": 0.7116013169288635,
      "step": 2144
    },
    {
      "epoch": 2.782529350933755,
      "grad_norm": 0.7771366834640503,
      "learning_rate": 1.4529091286973994e-07,
      "loss": 0.7575348615646362,
      "step": 2145
    },
    {
      "epoch": 2.783827077883894,
      "grad_norm": 0.7422629594802856,
      "learning_rate": 1.435848336906359e-07,
      "loss": 0.7005946040153503,
      "step": 2146
    },
    {
      "epoch": 2.785124804834033,
      "grad_norm": 0.748662531375885,
      "learning_rate": 1.418886846513673e-07,
      "loss": 0.7412527799606323,
      "step": 2147
    },
    {
      "epoch": 2.7864225317841718,
      "grad_norm": 0.8038177490234375,
      "learning_rate": 1.4020246922013093e-07,
      "loss": 0.6991332769393921,
      "step": 2148
    },
    {
      "epoch": 2.7877202587343106,
      "grad_norm": 0.7641218900680542,
      "learning_rate": 1.3852619084480933e-07,
      "loss": 0.6647689938545227,
      "step": 2149
    },
    {
      "epoch": 2.7890179856844495,
      "grad_norm": 0.7471143007278442,
      "learning_rate": 1.3685985295296798e-07,
      "loss": 0.735883355140686,
      "step": 2150
    },
    {
      "epoch": 2.7903157126345883,
      "grad_norm": 0.7782725095748901,
      "learning_rate": 1.3520345895184583e-07,
      "loss": 0.6804190278053284,
      "step": 2151
    },
    {
      "epoch": 2.7916134395847276,
      "grad_norm": 0.7638722658157349,
      "learning_rate": 1.3355701222835026e-07,
      "loss": 0.6945773363113403,
      "step": 2152
    },
    {
      "epoch": 2.792911166534866,
      "grad_norm": 0.7732612490653992,
      "learning_rate": 1.3192051614904722e-07,
      "loss": 0.6848861575126648,
      "step": 2153
    },
    {
      "epoch": 2.7942088934850053,
      "grad_norm": 0.7860000133514404,
      "learning_rate": 1.302939740601572e-07,
      "loss": 0.7503040432929993,
      "step": 2154
    },
    {
      "epoch": 2.795506620435144,
      "grad_norm": 0.7470289468765259,
      "learning_rate": 1.2867738928754703e-07,
      "loss": 0.687214195728302,
      "step": 2155
    },
    {
      "epoch": 2.796804347385283,
      "grad_norm": 0.7711262106895447,
      "learning_rate": 1.2707076513672423e-07,
      "loss": 0.7178612947463989,
      "step": 2156
    },
    {
      "epoch": 2.798102074335422,
      "grad_norm": 0.7954484224319458,
      "learning_rate": 1.2547410489282708e-07,
      "loss": 0.6872047781944275,
      "step": 2157
    },
    {
      "epoch": 2.7993998012855608,
      "grad_norm": 0.7886789441108704,
      "learning_rate": 1.2388741182062348e-07,
      "loss": 0.6837970614433289,
      "step": 2158
    },
    {
      "epoch": 2.8006975282356996,
      "grad_norm": 0.7388010621070862,
      "learning_rate": 1.2231068916449705e-07,
      "loss": 0.6859837174415588,
      "step": 2159
    },
    {
      "epoch": 2.8019952551858385,
      "grad_norm": 0.8267860412597656,
      "learning_rate": 1.2074394014844782e-07,
      "loss": 0.7664659023284912,
      "step": 2160
    },
    {
      "epoch": 2.8032929821359773,
      "grad_norm": 0.7906567454338074,
      "learning_rate": 1.1918716797608087e-07,
      "loss": 0.7450604438781738,
      "step": 2161
    },
    {
      "epoch": 2.804590709086116,
      "grad_norm": 0.7570729851722717,
      "learning_rate": 1.1764037583060162e-07,
      "loss": 0.7012235522270203,
      "step": 2162
    },
    {
      "epoch": 2.8058884360362555,
      "grad_norm": 0.7581570744514465,
      "learning_rate": 1.1610356687480728e-07,
      "loss": 0.7267595529556274,
      "step": 2163
    },
    {
      "epoch": 2.807186162986394,
      "grad_norm": 0.7544029951095581,
      "learning_rate": 1.1457674425108478e-07,
      "loss": 0.6990587115287781,
      "step": 2164
    },
    {
      "epoch": 2.808483889936533,
      "grad_norm": 0.7233387231826782,
      "learning_rate": 1.1305991108139847e-07,
      "loss": 0.7330483198165894,
      "step": 2165
    },
    {
      "epoch": 2.809781616886672,
      "grad_norm": 0.7781274914741516,
      "learning_rate": 1.1155307046728958e-07,
      "loss": 0.6811683773994446,
      "step": 2166
    },
    {
      "epoch": 2.811079343836811,
      "grad_norm": 0.776544988155365,
      "learning_rate": 1.1005622548986406e-07,
      "loss": 0.7087331414222717,
      "step": 2167
    },
    {
      "epoch": 2.8123770707869498,
      "grad_norm": 0.7924423217773438,
      "learning_rate": 1.0856937920979305e-07,
      "loss": 0.7304611802101135,
      "step": 2168
    },
    {
      "epoch": 2.8136747977370886,
      "grad_norm": 0.7693164944648743,
      "learning_rate": 1.0709253466729963e-07,
      "loss": 0.6941207051277161,
      "step": 2169
    },
    {
      "epoch": 2.8149725246872275,
      "grad_norm": 0.8000521063804626,
      "learning_rate": 1.0562569488215712e-07,
      "loss": 0.7666316628456116,
      "step": 2170
    },
    {
      "epoch": 2.8162702516373663,
      "grad_norm": 0.8068124055862427,
      "learning_rate": 1.0416886285368188e-07,
      "loss": 0.7321376800537109,
      "step": 2171
    },
    {
      "epoch": 2.817567978587505,
      "grad_norm": 0.7735725045204163,
      "learning_rate": 1.0272204156072663e-07,
      "loss": 0.7054159641265869,
      "step": 2172
    },
    {
      "epoch": 2.818865705537644,
      "grad_norm": 0.7532979249954224,
      "learning_rate": 1.012852339616749e-07,
      "loss": 0.6740676760673523,
      "step": 2173
    },
    {
      "epoch": 2.8201634324877833,
      "grad_norm": 0.7769542932510376,
      "learning_rate": 9.985844299443437e-08,
      "loss": 0.692727267742157,
      "step": 2174
    },
    {
      "epoch": 2.8214611594379218,
      "grad_norm": 0.7732757925987244,
      "learning_rate": 9.844167157643191e-08,
      "loss": 0.7220668196678162,
      "step": 2175
    },
    {
      "epoch": 2.822758886388061,
      "grad_norm": 0.748921275138855,
      "learning_rate": 9.703492260460578e-08,
      "loss": 0.7080860733985901,
      "step": 2176
    },
    {
      "epoch": 2.8240566133382,
      "grad_norm": 0.7941139936447144,
      "learning_rate": 9.563819895540172e-08,
      "loss": 0.8114986419677734,
      "step": 2177
    },
    {
      "epoch": 2.8253543402883388,
      "grad_norm": 0.7770174145698547,
      "learning_rate": 9.42515034847663e-08,
      "loss": 0.7355633974075317,
      "step": 2178
    },
    {
      "epoch": 2.8266520672384776,
      "grad_norm": 0.7767409086227417,
      "learning_rate": 9.287483902814087e-08,
      "loss": 0.7234621644020081,
      "step": 2179
    },
    {
      "epoch": 2.8279497941886165,
      "grad_norm": 0.791036069393158,
      "learning_rate": 9.150820840045483e-08,
      "loss": 0.7287334203720093,
      "step": 2180
    },
    {
      "epoch": 2.8292475211387553,
      "grad_norm": 0.7541413903236389,
      "learning_rate": 9.015161439612396e-08,
      "loss": 0.7405381798744202,
      "step": 2181
    },
    {
      "epoch": 2.830545248088894,
      "grad_norm": 0.7417975068092346,
      "learning_rate": 8.880505978903719e-08,
      "loss": 0.7482308149337769,
      "step": 2182
    },
    {
      "epoch": 2.831842975039033,
      "grad_norm": 0.7411255240440369,
      "learning_rate": 8.746854733255982e-08,
      "loss": 0.7017207145690918,
      "step": 2183
    },
    {
      "epoch": 2.833140701989172,
      "grad_norm": 0.7665935754776001,
      "learning_rate": 8.614207975952083e-08,
      "loss": 0.7161105871200562,
      "step": 2184
    },
    {
      "epoch": 2.834438428939311,
      "grad_norm": 0.7866116762161255,
      "learning_rate": 8.482565978221002e-08,
      "loss": 0.7364633679389954,
      "step": 2185
    },
    {
      "epoch": 2.83573615588945,
      "grad_norm": 0.7433639764785767,
      "learning_rate": 8.351929009237425e-08,
      "loss": 0.7004378437995911,
      "step": 2186
    },
    {
      "epoch": 2.837033882839589,
      "grad_norm": 0.7636539936065674,
      "learning_rate": 8.222297336120844e-08,
      "loss": 0.7470002770423889,
      "step": 2187
    },
    {
      "epoch": 2.8383316097897278,
      "grad_norm": 0.7847253680229187,
      "learning_rate": 8.093671223935118e-08,
      "loss": 0.6827197670936584,
      "step": 2188
    },
    {
      "epoch": 2.8396293367398666,
      "grad_norm": 0.7404457926750183,
      "learning_rate": 7.966050935688252e-08,
      "loss": 0.6898380517959595,
      "step": 2189
    },
    {
      "epoch": 2.8409270636900055,
      "grad_norm": 0.7589480876922607,
      "learning_rate": 7.839436732331285e-08,
      "loss": 0.6429364681243896,
      "step": 2190
    },
    {
      "epoch": 2.8422247906401443,
      "grad_norm": 0.768264889717102,
      "learning_rate": 7.7138288727584e-08,
      "loss": 0.7231125831604004,
      "step": 2191
    },
    {
      "epoch": 2.843522517590283,
      "grad_norm": 0.7590683698654175,
      "learning_rate": 7.589227613805705e-08,
      "loss": 0.7650172114372253,
      "step": 2192
    },
    {
      "epoch": 2.844820244540422,
      "grad_norm": 0.7730223536491394,
      "learning_rate": 7.465633210251344e-08,
      "loss": 0.7363346219062805,
      "step": 2193
    },
    {
      "epoch": 2.8461179714905613,
      "grad_norm": 0.7730546593666077,
      "learning_rate": 7.343045914814495e-08,
      "loss": 0.754916787147522,
      "step": 2194
    },
    {
      "epoch": 2.8474156984406997,
      "grad_norm": 0.7889825105667114,
      "learning_rate": 7.221465978155262e-08,
      "loss": 0.6751682758331299,
      "step": 2195
    },
    {
      "epoch": 2.848713425390839,
      "grad_norm": 0.807558000087738,
      "learning_rate": 7.10089364887373e-08,
      "loss": 0.7206065058708191,
      "step": 2196
    },
    {
      "epoch": 2.850011152340978,
      "grad_norm": 0.8067498207092285,
      "learning_rate": 6.981329173509909e-08,
      "loss": 0.7805840969085693,
      "step": 2197
    },
    {
      "epoch": 2.8513088792911168,
      "grad_norm": 0.7476110458374023,
      "learning_rate": 6.862772796542794e-08,
      "loss": 0.7597277760505676,
      "step": 2198
    },
    {
      "epoch": 2.8526066062412556,
      "grad_norm": 0.8052570819854736,
      "learning_rate": 6.745224760390246e-08,
      "loss": 0.7186450958251953,
      "step": 2199
    },
    {
      "epoch": 2.8539043331913945,
      "grad_norm": 0.7551132440567017,
      "learning_rate": 6.628685305408166e-08,
      "loss": 0.6708953976631165,
      "step": 2200
    },
    {
      "epoch": 2.8552020601415333,
      "grad_norm": 0.7911965250968933,
      "learning_rate": 6.513154669890221e-08,
      "loss": 0.6416589617729187,
      "step": 2201
    },
    {
      "epoch": 2.856499787091672,
      "grad_norm": 0.7725633382797241,
      "learning_rate": 6.398633090067497e-08,
      "loss": 0.6717783212661743,
      "step": 2202
    },
    {
      "epoch": 2.857797514041811,
      "grad_norm": 0.795801043510437,
      "learning_rate": 6.285120800107402e-08,
      "loss": 0.6958252787590027,
      "step": 2203
    },
    {
      "epoch": 2.85909524099195,
      "grad_norm": 0.7685935497283936,
      "learning_rate": 6.172618032114108e-08,
      "loss": 0.6823325753211975,
      "step": 2204
    },
    {
      "epoch": 2.860392967942089,
      "grad_norm": 0.8054656386375427,
      "learning_rate": 6.061125016127045e-08,
      "loss": 0.7556108236312866,
      "step": 2205
    },
    {
      "epoch": 2.8616906948922276,
      "grad_norm": 0.7606475949287415,
      "learning_rate": 5.950641980121352e-08,
      "loss": 0.7092191576957703,
      "step": 2206
    },
    {
      "epoch": 2.862988421842367,
      "grad_norm": 0.7591357231140137,
      "learning_rate": 5.84116915000682e-08,
      "loss": 0.7214552164077759,
      "step": 2207
    },
    {
      "epoch": 2.8642861487925058,
      "grad_norm": 0.74445641040802,
      "learning_rate": 5.732706749627726e-08,
      "loss": 0.7109599113464355,
      "step": 2208
    },
    {
      "epoch": 2.8655838757426446,
      "grad_norm": 0.7537177205085754,
      "learning_rate": 5.6252550007621645e-08,
      "loss": 0.7794804573059082,
      "step": 2209
    },
    {
      "epoch": 2.8668816026927835,
      "grad_norm": 0.7885766625404358,
      "learning_rate": 5.518814123121885e-08,
      "loss": 0.7224581837654114,
      "step": 2210
    },
    {
      "epoch": 2.8681793296429223,
      "grad_norm": 0.770478367805481,
      "learning_rate": 5.413384334351346e-08,
      "loss": 0.6955745816230774,
      "step": 2211
    },
    {
      "epoch": 2.869477056593061,
      "grad_norm": 0.7760496139526367,
      "learning_rate": 5.308965850027992e-08,
      "loss": 0.7298511862754822,
      "step": 2212
    },
    {
      "epoch": 2.8707747835432,
      "grad_norm": 0.8011616468429565,
      "learning_rate": 5.205558883661033e-08,
      "loss": 0.7658096551895142,
      "step": 2213
    },
    {
      "epoch": 2.872072510493339,
      "grad_norm": 0.760274350643158,
      "learning_rate": 5.103163646691611e-08,
      "loss": 0.7171127200126648,
      "step": 2214
    },
    {
      "epoch": 2.8733702374434777,
      "grad_norm": 0.7662195563316345,
      "learning_rate": 5.00178034849208e-08,
      "loss": 0.7367060780525208,
      "step": 2215
    },
    {
      "epoch": 2.874667964393617,
      "grad_norm": 0.7544783353805542,
      "learning_rate": 4.9014091963655584e-08,
      "loss": 0.6698785424232483,
      "step": 2216
    },
    {
      "epoch": 2.8759656913437555,
      "grad_norm": 0.8063926696777344,
      "learning_rate": 4.802050395545765e-08,
      "loss": 0.7871376276016235,
      "step": 2217
    },
    {
      "epoch": 2.8772634182938948,
      "grad_norm": 0.7655091881752014,
      "learning_rate": 4.703704149196187e-08,
      "loss": 0.7300557494163513,
      "step": 2218
    },
    {
      "epoch": 2.8785611452440336,
      "grad_norm": 0.7799139618873596,
      "learning_rate": 4.6063706584100196e-08,
      "loss": 0.6737034916877747,
      "step": 2219
    },
    {
      "epoch": 2.8798588721941725,
      "grad_norm": 0.7586493492126465,
      "learning_rate": 4.5100501222097304e-08,
      "loss": 0.825034499168396,
      "step": 2220
    },
    {
      "epoch": 2.8811565991443113,
      "grad_norm": 0.7479345798492432,
      "learning_rate": 4.414742737546274e-08,
      "loss": 0.7313747406005859,
      "step": 2221
    },
    {
      "epoch": 2.88245432609445,
      "grad_norm": 0.8680388331413269,
      "learning_rate": 4.320448699299262e-08,
      "loss": 0.8321331143379211,
      "step": 2222
    },
    {
      "epoch": 2.883752053044589,
      "grad_norm": 0.7766653895378113,
      "learning_rate": 4.227168200276077e-08,
      "loss": 0.6965093016624451,
      "step": 2223
    },
    {
      "epoch": 2.885049779994728,
      "grad_norm": 0.7875890135765076,
      "learning_rate": 4.134901431211702e-08,
      "loss": 0.7395253777503967,
      "step": 2224
    },
    {
      "epoch": 2.8863475069448667,
      "grad_norm": 0.7455689907073975,
      "learning_rate": 4.043648580768389e-08,
      "loss": 0.717605471611023,
      "step": 2225
    },
    {
      "epoch": 2.8876452338950056,
      "grad_norm": 0.7336239218711853,
      "learning_rate": 3.953409835535049e-08,
      "loss": 0.663152813911438,
      "step": 2226
    },
    {
      "epoch": 2.888942960845145,
      "grad_norm": 0.7704851627349854,
      "learning_rate": 3.8641853800271414e-08,
      "loss": 0.6977202892303467,
      "step": 2227
    },
    {
      "epoch": 2.8902406877952833,
      "grad_norm": 0.7432349324226379,
      "learning_rate": 3.77597539668606e-08,
      "loss": 0.6897647976875305,
      "step": 2228
    },
    {
      "epoch": 2.8915384147454226,
      "grad_norm": 0.7698100209236145,
      "learning_rate": 3.688780065878916e-08,
      "loss": 0.6869649291038513,
      "step": 2229
    },
    {
      "epoch": 2.8928361416955615,
      "grad_norm": 0.7458007335662842,
      "learning_rate": 3.602599565898091e-08,
      "loss": 0.6704892516136169,
      "step": 2230
    },
    {
      "epoch": 2.8941338686457003,
      "grad_norm": 0.7543532252311707,
      "learning_rate": 3.517434072960901e-08,
      "loss": 0.7490587830543518,
      "step": 2231
    },
    {
      "epoch": 2.895431595595839,
      "grad_norm": 0.7413858771324158,
      "learning_rate": 3.433283761209161e-08,
      "loss": 0.7437562942504883,
      "step": 2232
    },
    {
      "epoch": 2.896729322545978,
      "grad_norm": 0.8167586326599121,
      "learning_rate": 3.3501488027090635e-08,
      "loss": 0.6863840818405151,
      "step": 2233
    },
    {
      "epoch": 2.898027049496117,
      "grad_norm": 0.7584203481674194,
      "learning_rate": 3.268029367450465e-08,
      "loss": 0.7401635646820068,
      "step": 2234
    },
    {
      "epoch": 2.8993247764462557,
      "grad_norm": 0.7506893277168274,
      "learning_rate": 3.186925623346882e-08,
      "loss": 0.6926859617233276,
      "step": 2235
    },
    {
      "epoch": 2.9006225033963946,
      "grad_norm": 0.7473459839820862,
      "learning_rate": 3.10683773623488e-08,
      "loss": 0.7053976655006409,
      "step": 2236
    },
    {
      "epoch": 2.9019202303465335,
      "grad_norm": 0.7659155130386353,
      "learning_rate": 3.0277658698739665e-08,
      "loss": 0.7118337154388428,
      "step": 2237
    },
    {
      "epoch": 2.9032179572966728,
      "grad_norm": 0.7984611392021179,
      "learning_rate": 2.9497101859460865e-08,
      "loss": 0.6703416109085083,
      "step": 2238
    },
    {
      "epoch": 2.904515684246811,
      "grad_norm": 0.7309507727622986,
      "learning_rate": 2.872670844055403e-08,
      "loss": 0.7074156999588013,
      "step": 2239
    },
    {
      "epoch": 2.9058134111969505,
      "grad_norm": 0.770394504070282,
      "learning_rate": 2.7966480017277974e-08,
      "loss": 0.6736186146736145,
      "step": 2240
    },
    {
      "epoch": 2.9071111381470893,
      "grad_norm": 0.7793900370597839,
      "learning_rate": 2.7216418144107583e-08,
      "loss": 0.7187858819961548,
      "step": 2241
    },
    {
      "epoch": 2.908408865097228,
      "grad_norm": 0.7602792382240295,
      "learning_rate": 2.6476524354729917e-08,
      "loss": 0.6807897686958313,
      "step": 2242
    },
    {
      "epoch": 2.909706592047367,
      "grad_norm": 0.771725594997406,
      "learning_rate": 2.5746800162040342e-08,
      "loss": 0.7109719514846802,
      "step": 2243
    },
    {
      "epoch": 2.911004318997506,
      "grad_norm": 0.7642435431480408,
      "learning_rate": 2.5027247058139748e-08,
      "loss": 0.7177394032478333,
      "step": 2244
    },
    {
      "epoch": 2.9123020459476447,
      "grad_norm": 0.7589951753616333,
      "learning_rate": 2.4317866514332322e-08,
      "loss": 0.6757906675338745,
      "step": 2245
    },
    {
      "epoch": 2.9135997728977836,
      "grad_norm": 0.7996797561645508,
      "learning_rate": 2.361865998112223e-08,
      "loss": 0.7429489493370056,
      "step": 2246
    },
    {
      "epoch": 2.9148974998479225,
      "grad_norm": 0.760643482208252,
      "learning_rate": 2.2929628888209156e-08,
      "loss": 0.7641419172286987,
      "step": 2247
    },
    {
      "epoch": 2.9161952267980613,
      "grad_norm": 0.7626227736473083,
      "learning_rate": 2.2250774644487215e-08,
      "loss": 0.7242019176483154,
      "step": 2248
    },
    {
      "epoch": 2.9174929537482006,
      "grad_norm": 0.7831689715385437,
      "learning_rate": 2.158209863804217e-08,
      "loss": 0.7359440326690674,
      "step": 2249
    },
    {
      "epoch": 2.9187906806983395,
      "grad_norm": 0.7844473719596863,
      "learning_rate": 2.0923602236146977e-08,
      "loss": 0.7149016261100769,
      "step": 2250
    },
    {
      "epoch": 2.9200884076484783,
      "grad_norm": 0.7542704939842224,
      "learning_rate": 2.0275286785260694e-08,
      "loss": 0.7106453776359558,
      "step": 2251
    },
    {
      "epoch": 2.921386134598617,
      "grad_norm": 0.7972451448440552,
      "learning_rate": 1.9637153611022365e-08,
      "loss": 0.7038116455078125,
      "step": 2252
    },
    {
      "epoch": 2.922683861548756,
      "grad_norm": 0.7626628279685974,
      "learning_rate": 1.9009204018255456e-08,
      "loss": 0.6864380240440369,
      "step": 2253
    },
    {
      "epoch": 2.923981588498895,
      "grad_norm": 0.7927265763282776,
      "learning_rate": 1.839143929095566e-08,
      "loss": 0.6850247979164124,
      "step": 2254
    },
    {
      "epoch": 2.9252793154490337,
      "grad_norm": 0.736191987991333,
      "learning_rate": 1.7783860692296982e-08,
      "loss": 0.7299352884292603,
      "step": 2255
    },
    {
      "epoch": 2.9265770423991726,
      "grad_norm": 0.7745212316513062,
      "learning_rate": 1.718646946462288e-08,
      "loss": 0.7246094942092896,
      "step": 2256
    },
    {
      "epoch": 2.9278747693493115,
      "grad_norm": 0.8178421854972839,
      "learning_rate": 1.6599266829447902e-08,
      "loss": 0.7505685091018677,
      "step": 2257
    },
    {
      "epoch": 2.9291724962994508,
      "grad_norm": 0.7724591493606567,
      "learning_rate": 1.6022253987452717e-08,
      "loss": 0.6581344604492188,
      "step": 2258
    },
    {
      "epoch": 2.930470223249589,
      "grad_norm": 0.780160129070282,
      "learning_rate": 1.5455432118481884e-08,
      "loss": 0.7304115295410156,
      "step": 2259
    },
    {
      "epoch": 2.9317679501997285,
      "grad_norm": 0.752224862575531,
      "learning_rate": 1.4898802381543842e-08,
      "loss": 0.7192541360855103,
      "step": 2260
    },
    {
      "epoch": 2.9330656771498673,
      "grad_norm": 0.789572536945343,
      "learning_rate": 1.4352365914804822e-08,
      "loss": 0.7106218338012695,
      "step": 2261
    },
    {
      "epoch": 2.934363404100006,
      "grad_norm": 0.781013011932373,
      "learning_rate": 1.3816123835588835e-08,
      "loss": 0.7548207640647888,
      "step": 2262
    },
    {
      "epoch": 2.935661131050145,
      "grad_norm": 0.8188384175300598,
      "learning_rate": 1.3290077240375453e-08,
      "loss": 0.715192437171936,
      "step": 2263
    },
    {
      "epoch": 2.936958858000284,
      "grad_norm": 0.7365001440048218,
      "learning_rate": 1.277422720479704e-08,
      "loss": 0.7308194041252136,
      "step": 2264
    },
    {
      "epoch": 2.9382565849504227,
      "grad_norm": 0.7814714908599854,
      "learning_rate": 1.2268574783635968e-08,
      "loss": 0.8294280171394348,
      "step": 2265
    },
    {
      "epoch": 2.9395543119005616,
      "grad_norm": 0.7368665337562561,
      "learning_rate": 1.1773121010824063e-08,
      "loss": 0.7195689082145691,
      "step": 2266
    },
    {
      "epoch": 2.9408520388507005,
      "grad_norm": 0.7893995642662048,
      "learning_rate": 1.1287866899438171e-08,
      "loss": 0.7487694621086121,
      "step": 2267
    },
    {
      "epoch": 2.9421497658008393,
      "grad_norm": 0.737452507019043,
      "learning_rate": 1.081281344170071e-08,
      "loss": 0.6512093544006348,
      "step": 2268
    },
    {
      "epoch": 2.9434474927509786,
      "grad_norm": 0.8004632592201233,
      "learning_rate": 1.0347961608975221e-08,
      "loss": 0.7287921905517578,
      "step": 2269
    },
    {
      "epoch": 2.944745219701117,
      "grad_norm": 0.7655004262924194,
      "learning_rate": 9.893312351766382e-09,
      "loss": 0.7819275259971619,
      "step": 2270
    },
    {
      "epoch": 2.9460429466512563,
      "grad_norm": 0.7713659405708313,
      "learning_rate": 9.448866599717221e-09,
      "loss": 0.6854069232940674,
      "step": 2271
    },
    {
      "epoch": 2.947340673601395,
      "grad_norm": 0.746204137802124,
      "learning_rate": 9.014625261605791e-09,
      "loss": 0.7135393023490906,
      "step": 2272
    },
    {
      "epoch": 2.948638400551534,
      "grad_norm": 0.7569589614868164,
      "learning_rate": 8.590589225346834e-09,
      "loss": 0.7075472474098206,
      "step": 2273
    },
    {
      "epoch": 2.949936127501673,
      "grad_norm": 0.7612638473510742,
      "learning_rate": 8.17675935798623e-09,
      "loss": 0.7831802368164062,
      "step": 2274
    },
    {
      "epoch": 2.9512338544518117,
      "grad_norm": 0.7586123943328857,
      "learning_rate": 7.773136505700995e-09,
      "loss": 0.6527408361434937,
      "step": 2275
    },
    {
      "epoch": 2.9525315814019506,
      "grad_norm": 0.7621296048164368,
      "learning_rate": 7.379721493798176e-09,
      "loss": 0.7169144153594971,
      "step": 2276
    },
    {
      "epoch": 2.9538293083520895,
      "grad_norm": 0.7727404236793518,
      "learning_rate": 6.996515126711511e-09,
      "loss": 0.6721253991127014,
      "step": 2277
    },
    {
      "epoch": 2.9551270353022283,
      "grad_norm": 0.8339347243309021,
      "learning_rate": 6.623518188001443e-09,
      "loss": 0.7189649939537048,
      "step": 2278
    },
    {
      "epoch": 2.956424762252367,
      "grad_norm": 0.8069379925727844,
      "learning_rate": 6.260731440351775e-09,
      "loss": 0.7410060167312622,
      "step": 2279
    },
    {
      "epoch": 2.9577224892025065,
      "grad_norm": 0.7937390804290771,
      "learning_rate": 5.908155625570233e-09,
      "loss": 0.7048535943031311,
      "step": 2280
    },
    {
      "epoch": 2.959020216152645,
      "grad_norm": 0.72493577003479,
      "learning_rate": 5.56579146458458e-09,
      "loss": 0.723403811454773,
      "step": 2281
    },
    {
      "epoch": 2.960317943102784,
      "grad_norm": 0.7555978298187256,
      "learning_rate": 5.233639657443168e-09,
      "loss": 0.739270806312561,
      "step": 2282
    },
    {
      "epoch": 2.961615670052923,
      "grad_norm": 0.7471567392349243,
      "learning_rate": 4.911700883312165e-09,
      "loss": 0.6872268319129944,
      "step": 2283
    },
    {
      "epoch": 2.962913397003062,
      "grad_norm": 0.7928717732429504,
      "learning_rate": 4.599975800475553e-09,
      "loss": 0.7023618817329407,
      "step": 2284
    },
    {
      "epoch": 2.9642111239532007,
      "grad_norm": 0.733587384223938,
      "learning_rate": 4.298465046331246e-09,
      "loss": 0.715094268321991,
      "step": 2285
    },
    {
      "epoch": 2.9655088509033396,
      "grad_norm": 0.7716489434242249,
      "learning_rate": 4.007169237392749e-09,
      "loss": 0.651224672794342,
      "step": 2286
    },
    {
      "epoch": 2.9668065778534785,
      "grad_norm": 0.7328708171844482,
      "learning_rate": 3.726088969286945e-09,
      "loss": 0.7175962328910828,
      "step": 2287
    },
    {
      "epoch": 2.9681043048036173,
      "grad_norm": 0.7712464332580566,
      "learning_rate": 3.4552248167507576e-09,
      "loss": 0.7173080444335938,
      "step": 2288
    },
    {
      "epoch": 2.969402031753756,
      "grad_norm": 0.7157249450683594,
      "learning_rate": 3.1945773336333754e-09,
      "loss": 0.6980288028717041,
      "step": 2289
    },
    {
      "epoch": 2.970699758703895,
      "grad_norm": 0.7634878158569336,
      "learning_rate": 2.9441470528929206e-09,
      "loss": 0.7472588419914246,
      "step": 2290
    },
    {
      "epoch": 2.9719974856540343,
      "grad_norm": 0.8366067409515381,
      "learning_rate": 2.703934486595894e-09,
      "loss": 0.7743273377418518,
      "step": 2291
    },
    {
      "epoch": 2.9732952126041727,
      "grad_norm": 0.7398233413696289,
      "learning_rate": 2.4739401259160635e-09,
      "loss": 0.7106281518936157,
      "step": 2292
    },
    {
      "epoch": 2.974592939554312,
      "grad_norm": 0.8272362351417542,
      "learning_rate": 2.2541644411344653e-09,
      "loss": 0.7096326351165771,
      "step": 2293
    },
    {
      "epoch": 2.975890666504451,
      "grad_norm": 0.7860119938850403,
      "learning_rate": 2.0446078816355186e-09,
      "loss": 0.6308723092079163,
      "step": 2294
    },
    {
      "epoch": 2.9771883934545897,
      "grad_norm": 0.734978199005127,
      "learning_rate": 1.8452708759097993e-09,
      "loss": 0.7198217511177063,
      "step": 2295
    },
    {
      "epoch": 2.9784861204047286,
      "grad_norm": 0.7446017265319824,
      "learning_rate": 1.656153831551821e-09,
      "loss": 0.7086501717567444,
      "step": 2296
    },
    {
      "epoch": 2.9797838473548675,
      "grad_norm": 0.7879230380058289,
      "learning_rate": 1.4772571352567044e-09,
      "loss": 0.7350752353668213,
      "step": 2297
    },
    {
      "epoch": 2.9810815743050063,
      "grad_norm": 0.8034906983375549,
      "learning_rate": 1.3085811528240622e-09,
      "loss": 0.7501051425933838,
      "step": 2298
    },
    {
      "epoch": 2.982379301255145,
      "grad_norm": 0.8161899447441101,
      "learning_rate": 1.1501262291530034e-09,
      "loss": 0.7630376219749451,
      "step": 2299
    },
    {
      "epoch": 2.983677028205284,
      "grad_norm": 0.7429082989692688,
      "learning_rate": 1.0018926882443548e-09,
      "loss": 0.747724175453186,
      "step": 2300
    },
    {
      "epoch": 2.983677028205284,
      "eval_loss": 0.7602248191833496,
      "eval_runtime": 139.9566,
      "eval_samples_per_second": 37.097,
      "eval_steps_per_second": 9.274,
      "step": 2300
    },
    {
      "epoch": 2.984974755155423,
      "grad_norm": 0.7769709229469299,
      "learning_rate": 8.638808331973281e-10,
      "loss": 0.7226245403289795,
      "step": 2301
    },
    {
      "epoch": 2.986272482105562,
      "grad_norm": 0.7434839010238647,
      "learning_rate": 7.360909462111876e-10,
      "loss": 0.7368016242980957,
      "step": 2302
    },
    {
      "epoch": 2.987570209055701,
      "grad_norm": 0.7686272263526917,
      "learning_rate": 6.185232885846937e-10,
      "loss": 0.7323869466781616,
      "step": 2303
    },
    {
      "epoch": 2.98886793600584,
      "grad_norm": 0.7547258138656616,
      "learning_rate": 5.111781007138827e-10,
      "loss": 0.7003946900367737,
      "step": 2304
    },
    {
      "epoch": 2.9901656629559787,
      "grad_norm": 0.7800252437591553,
      "learning_rate": 4.1405560209206716e-10,
      "loss": 0.7231513261795044,
      "step": 2305
    },
    {
      "epoch": 2.9914633899061176,
      "grad_norm": 0.7863386869430542,
      "learning_rate": 3.2715599131039053e-10,
      "loss": 0.7120152115821838,
      "step": 2306
    },
    {
      "epoch": 2.9927611168562565,
      "grad_norm": 0.8008667230606079,
      "learning_rate": 2.5047944605616215e-10,
      "loss": 0.7078871726989746,
      "step": 2307
    },
    {
      "epoch": 2.9940588438063953,
      "grad_norm": 0.7796193361282349,
      "learning_rate": 1.840261231139673e-10,
      "loss": 0.6753217577934265,
      "step": 2308
    },
    {
      "epoch": 2.995356570756534,
      "grad_norm": 0.7583023905754089,
      "learning_rate": 1.2779615836455706e-10,
      "loss": 0.7384961247444153,
      "step": 2309
    },
    {
      "epoch": 2.996654297706673,
      "grad_norm": 0.7805288434028625,
      "learning_rate": 8.17896667826279e-11,
      "loss": 0.7757540941238403,
      "step": 2310
    },
    {
      "epoch": 2.9979520246568123,
      "grad_norm": 0.7969951033592224,
      "learning_rate": 4.600674244070735e-11,
      "loss": 0.6859312653541565,
      "step": 2311
    },
    {
      "epoch": 2.9992497516069507,
      "grad_norm": 0.8133782148361206,
      "learning_rate": 2.04474585052683e-11,
      "loss": 0.6985263824462891,
      "step": 2312
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.0103119611740112,
      "learning_rate": 5.11186723950452e-12,
      "loss": 0.8292257785797119,
      "step": 2313
    },
    {
      "epoch": 3.0,
      "step": 2313,
      "total_flos": 4.1917370482093916e+18,
      "train_loss": 0.07527349776537244,
      "train_runtime": 3319.5738,
      "train_samples_per_second": 89.138,
      "train_steps_per_second": 0.697
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 2313,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 230,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 4.1917370482093916e+18,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}