{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 5.0,
  "eval_steps": 500,
  "global_step": 2555,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0019588638589618022,
      "grad_norm": 1.1779376668292296,
      "learning_rate": 0.0,
      "loss": 0.6095,
      "num_tokens": 957228.0,
      "step": 1
    },
    {
      "epoch": 0.0039177277179236044,
      "grad_norm": 1.14453866876595,
      "learning_rate": 1.298701298701299e-07,
      "loss": 0.5863,
      "num_tokens": 1903020.0,
      "step": 2
    },
    {
      "epoch": 0.005876591576885406,
      "grad_norm": 1.1558548993371778,
      "learning_rate": 2.597402597402598e-07,
      "loss": 0.5987,
      "num_tokens": 2835325.0,
      "step": 3
    },
    {
      "epoch": 0.007835455435847209,
      "grad_norm": 1.1580675566497325,
      "learning_rate": 3.896103896103896e-07,
      "loss": 0.6001,
      "num_tokens": 3804624.0,
      "step": 4
    },
    {
      "epoch": 0.009794319294809012,
      "grad_norm": 1.131687899148094,
      "learning_rate": 5.194805194805196e-07,
      "loss": 0.591,
      "num_tokens": 4760897.0,
      "step": 5
    },
    {
      "epoch": 0.011753183153770812,
      "grad_norm": 1.1365950486885912,
      "learning_rate": 6.493506493506493e-07,
      "loss": 0.5938,
      "num_tokens": 5696321.0,
      "step": 6
    },
    {
      "epoch": 0.013712047012732615,
      "grad_norm": 1.1369691359607021,
      "learning_rate": 7.792207792207792e-07,
      "loss": 0.5961,
      "num_tokens": 6645763.0,
      "step": 7
    },
    {
      "epoch": 0.015670910871694418,
      "grad_norm": 1.1233025219879869,
      "learning_rate": 9.090909090909091e-07,
      "loss": 0.5764,
      "num_tokens": 7583888.0,
      "step": 8
    },
    {
      "epoch": 0.01762977473065622,
      "grad_norm": 1.1513047413325934,
      "learning_rate": 1.0389610389610392e-06,
      "loss": 0.6001,
      "num_tokens": 8544514.0,
      "step": 9
    },
    {
      "epoch": 0.019588638589618023,
      "grad_norm": 1.090111147138629,
      "learning_rate": 1.168831168831169e-06,
      "loss": 0.5682,
      "num_tokens": 9505325.0,
      "step": 10
    },
    {
      "epoch": 0.021547502448579822,
      "grad_norm": 1.128435799580429,
      "learning_rate": 1.2987012987012986e-06,
      "loss": 0.5969,
      "num_tokens": 10496025.0,
      "step": 11
    },
    {
      "epoch": 0.023506366307541625,
      "grad_norm": 1.1419469297779599,
      "learning_rate": 1.4285714285714286e-06,
      "loss": 0.5918,
      "num_tokens": 11483781.0,
      "step": 12
    },
    {
      "epoch": 0.025465230166503428,
      "grad_norm": 1.0837958411329958,
      "learning_rate": 1.5584415584415584e-06,
      "loss": 0.5797,
      "num_tokens": 12421527.0,
      "step": 13
    },
    {
      "epoch": 0.02742409402546523,
      "grad_norm": 1.1202969148222806,
      "learning_rate": 1.6883116883116885e-06,
      "loss": 0.5968,
      "num_tokens": 13402688.0,
      "step": 14
    },
    {
      "epoch": 0.029382957884427033,
      "grad_norm": 1.0672780406526166,
      "learning_rate": 1.8181818181818183e-06,
      "loss": 0.5771,
      "num_tokens": 14367162.0,
      "step": 15
    },
    {
      "epoch": 0.031341821743388835,
      "grad_norm": 1.0931389983274584,
      "learning_rate": 1.9480519480519483e-06,
      "loss": 0.6058,
      "num_tokens": 15278974.0,
      "step": 16
    },
    {
      "epoch": 0.03330068560235064,
      "grad_norm": 0.9912007092939249,
      "learning_rate": 2.0779220779220784e-06,
      "loss": 0.5904,
      "num_tokens": 16229440.0,
      "step": 17
    },
    {
      "epoch": 0.03525954946131244,
      "grad_norm": 0.9334645992333939,
      "learning_rate": 2.207792207792208e-06,
      "loss": 0.5673,
      "num_tokens": 17174371.0,
      "step": 18
    },
    {
      "epoch": 0.03721841332027424,
      "grad_norm": 0.9154496415543664,
      "learning_rate": 2.337662337662338e-06,
      "loss": 0.5739,
      "num_tokens": 18132871.0,
      "step": 19
    },
    {
      "epoch": 0.039177277179236046,
      "grad_norm": 0.9180404646017777,
      "learning_rate": 2.4675324675324676e-06,
      "loss": 0.5808,
      "num_tokens": 19079741.0,
      "step": 20
    },
    {
      "epoch": 0.04113614103819784,
      "grad_norm": 0.8626980715207839,
      "learning_rate": 2.597402597402597e-06,
      "loss": 0.5777,
      "num_tokens": 20037323.0,
      "step": 21
    },
    {
      "epoch": 0.043095004897159644,
      "grad_norm": 0.8510957187588554,
      "learning_rate": 2.7272727272727272e-06,
      "loss": 0.5781,
      "num_tokens": 20992105.0,
      "step": 22
    },
    {
      "epoch": 0.04505386875612145,
      "grad_norm": 0.822763278842823,
      "learning_rate": 2.8571428571428573e-06,
      "loss": 0.5522,
      "num_tokens": 21917740.0,
      "step": 23
    },
    {
      "epoch": 0.04701273261508325,
      "grad_norm": 0.5682119647768065,
      "learning_rate": 2.9870129870129873e-06,
      "loss": 0.5346,
      "num_tokens": 22868921.0,
      "step": 24
    },
    {
      "epoch": 0.04897159647404505,
      "grad_norm": 0.5212911432498203,
      "learning_rate": 3.116883116883117e-06,
      "loss": 0.5392,
      "num_tokens": 23827074.0,
      "step": 25
    },
    {
      "epoch": 0.050930460333006855,
      "grad_norm": 0.5034383592828012,
      "learning_rate": 3.246753246753247e-06,
      "loss": 0.5447,
      "num_tokens": 24746581.0,
      "step": 26
    },
    {
      "epoch": 0.05288932419196866,
      "grad_norm": 0.47309618987632196,
      "learning_rate": 3.376623376623377e-06,
      "loss": 0.546,
      "num_tokens": 25705115.0,
      "step": 27
    },
    {
      "epoch": 0.05484818805093046,
      "grad_norm": 0.44261582789260967,
      "learning_rate": 3.506493506493507e-06,
      "loss": 0.5519,
      "num_tokens": 26632617.0,
      "step": 28
    },
    {
      "epoch": 0.05680705190989226,
      "grad_norm": 0.4182589041396013,
      "learning_rate": 3.6363636363636366e-06,
      "loss": 0.5313,
      "num_tokens": 27575977.0,
      "step": 29
    },
    {
      "epoch": 0.058765915768854066,
      "grad_norm": 0.38193294303014225,
      "learning_rate": 3.7662337662337666e-06,
      "loss": 0.5208,
      "num_tokens": 28568378.0,
      "step": 30
    },
    {
      "epoch": 0.06072477962781587,
      "grad_norm": 0.35997287973769965,
      "learning_rate": 3.896103896103897e-06,
      "loss": 0.5199,
      "num_tokens": 29517430.0,
      "step": 31
    },
    {
      "epoch": 0.06268364348677767,
      "grad_norm": 0.24736828742042855,
      "learning_rate": 4.025974025974026e-06,
      "loss": 0.5232,
      "num_tokens": 30460107.0,
      "step": 32
    },
    {
      "epoch": 0.06464250734573947,
      "grad_norm": 0.29430500472969917,
      "learning_rate": 4.155844155844157e-06,
      "loss": 0.4902,
      "num_tokens": 31376862.0,
      "step": 33
    },
    {
      "epoch": 0.06660137120470128,
      "grad_norm": 0.3813996244800663,
      "learning_rate": 4.2857142857142855e-06,
      "loss": 0.5095,
      "num_tokens": 32333039.0,
      "step": 34
    },
    {
      "epoch": 0.06856023506366307,
      "grad_norm": 0.3734526706562682,
      "learning_rate": 4.415584415584416e-06,
      "loss": 0.4982,
      "num_tokens": 33301775.0,
      "step": 35
    },
    {
      "epoch": 0.07051909892262488,
      "grad_norm": 0.4071231857824654,
      "learning_rate": 4.5454545454545455e-06,
      "loss": 0.5053,
      "num_tokens": 34281803.0,
      "step": 36
    },
    {
      "epoch": 0.07247796278158668,
      "grad_norm": 0.38881600781955006,
      "learning_rate": 4.675324675324676e-06,
      "loss": 0.5156,
      "num_tokens": 35278446.0,
      "step": 37
    },
    {
      "epoch": 0.07443682664054849,
      "grad_norm": 0.34959618219486355,
      "learning_rate": 4.805194805194806e-06,
      "loss": 0.495,
      "num_tokens": 36174529.0,
      "step": 38
    },
    {
      "epoch": 0.07639569049951028,
      "grad_norm": 0.30582243304816603,
      "learning_rate": 4.935064935064935e-06,
      "loss": 0.4981,
      "num_tokens": 37096798.0,
      "step": 39
    },
    {
      "epoch": 0.07835455435847209,
      "grad_norm": 0.2683098108248299,
      "learning_rate": 5.064935064935065e-06,
      "loss": 0.5024,
      "num_tokens": 38025690.0,
      "step": 40
    },
    {
      "epoch": 0.08031341821743389,
      "grad_norm": 0.2227478708148392,
      "learning_rate": 5.194805194805194e-06,
      "loss": 0.49,
      "num_tokens": 38967788.0,
      "step": 41
    },
    {
      "epoch": 0.08227228207639568,
      "grad_norm": 0.18108191554443825,
      "learning_rate": 5.324675324675325e-06,
      "loss": 0.5014,
      "num_tokens": 39929380.0,
      "step": 42
    },
    {
      "epoch": 0.0842311459353575,
      "grad_norm": 0.15497597357842827,
      "learning_rate": 5.4545454545454545e-06,
      "loss": 0.4946,
      "num_tokens": 40891263.0,
      "step": 43
    },
    {
      "epoch": 0.08619000979431929,
      "grad_norm": 0.15671125686656093,
      "learning_rate": 5.584415584415585e-06,
      "loss": 0.4799,
      "num_tokens": 41833850.0,
      "step": 44
    },
    {
      "epoch": 0.0881488736532811,
      "grad_norm": 0.17507981763637595,
      "learning_rate": 5.7142857142857145e-06,
      "loss": 0.5,
      "num_tokens": 42760693.0,
      "step": 45
    },
    {
      "epoch": 0.0901077375122429,
      "grad_norm": 0.18455600553146348,
      "learning_rate": 5.844155844155844e-06,
      "loss": 0.4942,
      "num_tokens": 43665379.0,
      "step": 46
    },
    {
      "epoch": 0.0920666013712047,
      "grad_norm": 0.20963781093615366,
      "learning_rate": 5.9740259740259746e-06,
      "loss": 0.4791,
      "num_tokens": 44609358.0,
      "step": 47
    },
    {
      "epoch": 0.0940254652301665,
      "grad_norm": 0.20219312988770385,
      "learning_rate": 6.103896103896104e-06,
      "loss": 0.493,
      "num_tokens": 45561854.0,
      "step": 48
    },
    {
      "epoch": 0.09598432908912831,
      "grad_norm": 0.18454774054032433,
      "learning_rate": 6.233766233766234e-06,
      "loss": 0.4865,
      "num_tokens": 46546409.0,
      "step": 49
    },
    {
      "epoch": 0.0979431929480901,
      "grad_norm": 0.16319189525503663,
      "learning_rate": 6.363636363636364e-06,
      "loss": 0.4849,
      "num_tokens": 47486971.0,
      "step": 50
    },
    {
      "epoch": 0.09990205680705191,
      "grad_norm": 0.14530181600723818,
      "learning_rate": 6.493506493506494e-06,
      "loss": 0.4934,
      "num_tokens": 48457717.0,
      "step": 51
    },
    {
      "epoch": 0.10186092066601371,
      "grad_norm": 0.12857250564688383,
      "learning_rate": 6.623376623376624e-06,
      "loss": 0.4863,
      "num_tokens": 49385047.0,
      "step": 52
    },
    {
      "epoch": 0.10381978452497552,
      "grad_norm": 0.13135618685388173,
      "learning_rate": 6.753246753246754e-06,
      "loss": 0.489,
      "num_tokens": 50295374.0,
      "step": 53
    },
    {
      "epoch": 0.10577864838393732,
      "grad_norm": 0.134389045356148,
      "learning_rate": 6.8831168831168835e-06,
      "loss": 0.4723,
      "num_tokens": 51245492.0,
      "step": 54
    },
    {
      "epoch": 0.10773751224289912,
      "grad_norm": 0.144607400296704,
      "learning_rate": 7.012987012987014e-06,
      "loss": 0.4751,
      "num_tokens": 52187156.0,
      "step": 55
    },
    {
      "epoch": 0.10969637610186092,
      "grad_norm": 0.15328301289443136,
      "learning_rate": 7.1428571428571436e-06,
      "loss": 0.4743,
      "num_tokens": 53175990.0,
      "step": 56
    },
    {
      "epoch": 0.11165523996082272,
      "grad_norm": 0.13884142700520555,
      "learning_rate": 7.272727272727273e-06,
      "loss": 0.4647,
      "num_tokens": 54116901.0,
      "step": 57
    },
    {
      "epoch": 0.11361410381978453,
      "grad_norm": 0.13359747069670078,
      "learning_rate": 7.402597402597404e-06,
      "loss": 0.4692,
      "num_tokens": 55059892.0,
      "step": 58
    },
    {
      "epoch": 0.11557296767874632,
      "grad_norm": 0.1200875386005197,
      "learning_rate": 7.532467532467533e-06,
      "loss": 0.478,
      "num_tokens": 56011558.0,
      "step": 59
    },
    {
      "epoch": 0.11753183153770813,
      "grad_norm": 0.11530192743485264,
      "learning_rate": 7.662337662337663e-06,
      "loss": 0.4694,
      "num_tokens": 56978072.0,
      "step": 60
    },
    {
      "epoch": 0.11949069539666993,
      "grad_norm": 0.11869136818231486,
      "learning_rate": 7.792207792207793e-06,
      "loss": 0.483,
      "num_tokens": 57932775.0,
      "step": 61
    },
    {
      "epoch": 0.12144955925563174,
      "grad_norm": 0.10469749897829281,
      "learning_rate": 7.922077922077924e-06,
      "loss": 0.4744,
      "num_tokens": 58921803.0,
      "step": 62
    },
    {
      "epoch": 0.12340842311459353,
      "grad_norm": 0.1220032554685136,
      "learning_rate": 8.051948051948052e-06,
      "loss": 0.4739,
      "num_tokens": 59825350.0,
      "step": 63
    },
    {
      "epoch": 0.12536728697355534,
      "grad_norm": 0.12522096927717064,
      "learning_rate": 8.181818181818183e-06,
      "loss": 0.4808,
      "num_tokens": 60736848.0,
      "step": 64
    },
    {
      "epoch": 0.12732615083251714,
      "grad_norm": 1.7728242669152647,
      "learning_rate": 8.311688311688313e-06,
      "loss": 0.4775,
      "num_tokens": 61692068.0,
      "step": 65
    },
    {
      "epoch": 0.12928501469147893,
      "grad_norm": 0.1248389564577605,
      "learning_rate": 8.441558441558442e-06,
      "loss": 0.4897,
      "num_tokens": 62657029.0,
      "step": 66
    },
    {
      "epoch": 0.13124387855044076,
      "grad_norm": 0.11025857585858609,
      "learning_rate": 8.571428571428571e-06,
      "loss": 0.4659,
      "num_tokens": 63579287.0,
      "step": 67
    },
    {
      "epoch": 0.13320274240940255,
      "grad_norm": 0.11299800436889519,
      "learning_rate": 8.701298701298701e-06,
      "loss": 0.4816,
      "num_tokens": 64536200.0,
      "step": 68
    },
    {
      "epoch": 0.13516160626836435,
      "grad_norm": 0.10967372205625224,
      "learning_rate": 8.831168831168832e-06,
      "loss": 0.4781,
      "num_tokens": 65474334.0,
      "step": 69
    },
    {
      "epoch": 0.13712047012732614,
      "grad_norm": 0.10362839198604146,
      "learning_rate": 8.96103896103896e-06,
      "loss": 0.4756,
      "num_tokens": 66428712.0,
      "step": 70
    },
    {
      "epoch": 0.13907933398628794,
      "grad_norm": 0.10866195870559414,
      "learning_rate": 9.090909090909091e-06,
      "loss": 0.4682,
      "num_tokens": 67373340.0,
      "step": 71
    },
    {
      "epoch": 0.14103819784524976,
      "grad_norm": 0.10862245165189673,
      "learning_rate": 9.220779220779221e-06,
      "loss": 0.4941,
      "num_tokens": 68279915.0,
      "step": 72
    },
    {
      "epoch": 0.14299706170421156,
      "grad_norm": 0.10756775407424338,
      "learning_rate": 9.350649350649352e-06,
      "loss": 0.4757,
      "num_tokens": 69239390.0,
      "step": 73
    },
    {
      "epoch": 0.14495592556317335,
      "grad_norm": 0.10241016679452052,
      "learning_rate": 9.48051948051948e-06,
      "loss": 0.4838,
      "num_tokens": 70186275.0,
      "step": 74
    },
    {
      "epoch": 0.14691478942213515,
      "grad_norm": 0.10414405498619668,
      "learning_rate": 9.610389610389611e-06,
      "loss": 0.4751,
      "num_tokens": 71150805.0,
      "step": 75
    },
    {
      "epoch": 0.14887365328109697,
      "grad_norm": 0.10111059772940759,
      "learning_rate": 9.740259740259742e-06,
      "loss": 0.4675,
      "num_tokens": 72095683.0,
      "step": 76
    },
    {
      "epoch": 0.15083251714005877,
      "grad_norm": 0.1014881117391909,
      "learning_rate": 9.87012987012987e-06,
      "loss": 0.4621,
      "num_tokens": 73031911.0,
      "step": 77
    },
    {
      "epoch": 0.15279138099902057,
      "grad_norm": 0.10404653706199633,
      "learning_rate": 1e-05,
      "loss": 0.48,
      "num_tokens": 73977616.0,
      "step": 78
    },
    {
      "epoch": 0.15475024485798236,
      "grad_norm": 0.10386392762988193,
      "learning_rate": 9.99999598174872e-06,
      "loss": 0.4722,
      "num_tokens": 74938755.0,
      "step": 79
    },
    {
      "epoch": 0.15670910871694418,
      "grad_norm": 0.10369883566275276,
      "learning_rate": 9.999983927001336e-06,
      "loss": 0.4826,
      "num_tokens": 75890007.0,
      "step": 80
    },
    {
      "epoch": 0.15866797257590598,
      "grad_norm": 0.10335357131879185,
      "learning_rate": 9.999963835777224e-06,
      "loss": 0.4733,
      "num_tokens": 76850365.0,
      "step": 81
    },
    {
      "epoch": 0.16062683643486778,
      "grad_norm": 0.1014465030505948,
      "learning_rate": 9.99993570810868e-06,
      "loss": 0.4725,
      "num_tokens": 77789082.0,
      "step": 82
    },
    {
      "epoch": 0.16258570029382957,
      "grad_norm": 0.0969864906933695,
      "learning_rate": 9.999899544040908e-06,
      "loss": 0.4603,
      "num_tokens": 78746891.0,
      "step": 83
    },
    {
      "epoch": 0.16454456415279137,
      "grad_norm": 0.09908553798822536,
      "learning_rate": 9.999855343632037e-06,
      "loss": 0.4705,
      "num_tokens": 79652749.0,
      "step": 84
    },
    {
      "epoch": 0.1665034280117532,
      "grad_norm": 0.10178596303441723,
      "learning_rate": 9.99980310695311e-06,
      "loss": 0.4718,
      "num_tokens": 80592581.0,
      "step": 85
    },
    {
      "epoch": 0.168462291870715,
      "grad_norm": 0.10039756909225854,
      "learning_rate": 9.99974283408809e-06,
      "loss": 0.4702,
      "num_tokens": 81534344.0,
      "step": 86
    },
    {
      "epoch": 0.17042115572967678,
      "grad_norm": 0.10292980005118785,
      "learning_rate": 9.99967452513385e-06,
      "loss": 0.4774,
      "num_tokens": 82487549.0,
      "step": 87
    },
    {
      "epoch": 0.17238001958863858,
      "grad_norm": 0.10086921342177411,
      "learning_rate": 9.999598180200186e-06,
      "loss": 0.4843,
      "num_tokens": 83472867.0,
      "step": 88
    },
    {
      "epoch": 0.1743388834476004,
      "grad_norm": 0.10124601224604896,
      "learning_rate": 9.999513799409804e-06,
      "loss": 0.4555,
      "num_tokens": 84390727.0,
      "step": 89
    },
    {
      "epoch": 0.1762977473065622,
      "grad_norm": 0.10185638919754106,
      "learning_rate": 9.99942138289833e-06,
      "loss": 0.4529,
      "num_tokens": 85371181.0,
      "step": 90
    },
    {
      "epoch": 0.178256611165524,
      "grad_norm": 0.09902084863732517,
      "learning_rate": 9.999320930814307e-06,
      "loss": 0.465,
      "num_tokens": 86357901.0,
      "step": 91
    },
    {
      "epoch": 0.1802154750244858,
      "grad_norm": 0.09762296388519529,
      "learning_rate": 9.999212443319191e-06,
      "loss": 0.4628,
      "num_tokens": 87343535.0,
      "step": 92
    },
    {
      "epoch": 0.1821743388834476,
      "grad_norm": 0.09448978900098406,
      "learning_rate": 9.999095920587352e-06,
      "loss": 0.4452,
      "num_tokens": 88297905.0,
      "step": 93
    },
    {
      "epoch": 0.1841332027424094,
      "grad_norm": 0.09431248116101847,
      "learning_rate": 9.998971362806081e-06,
      "loss": 0.4453,
      "num_tokens": 89225471.0,
      "step": 94
    },
    {
      "epoch": 0.1860920666013712,
      "grad_norm": 0.10055354255483136,
      "learning_rate": 9.998838770175575e-06,
      "loss": 0.4678,
      "num_tokens": 90164738.0,
      "step": 95
    },
    {
      "epoch": 0.188050930460333,
      "grad_norm": 0.10230246968909389,
      "learning_rate": 9.998698142908954e-06,
      "loss": 0.4664,
      "num_tokens": 91125556.0,
      "step": 96
    },
    {
      "epoch": 0.19000979431929482,
      "grad_norm": 0.10625479777481586,
      "learning_rate": 9.998549481232247e-06,
      "loss": 0.4568,
      "num_tokens": 92093541.0,
      "step": 97
    },
    {
      "epoch": 0.19196865817825662,
      "grad_norm": 0.09861753488551372,
      "learning_rate": 9.998392785384397e-06,
      "loss": 0.4807,
      "num_tokens": 93025537.0,
      "step": 98
    },
    {
      "epoch": 0.1939275220372184,
      "grad_norm": 0.09927039259565387,
      "learning_rate": 9.998228055617264e-06,
      "loss": 0.4584,
      "num_tokens": 93946672.0,
      "step": 99
    },
    {
      "epoch": 0.1958863858961802,
      "grad_norm": 0.10248383003320141,
      "learning_rate": 9.998055292195615e-06,
      "loss": 0.4696,
      "num_tokens": 94908060.0,
      "step": 100
    },
    {
      "epoch": 0.197845249755142,
      "grad_norm": 0.10451441636833611,
      "learning_rate": 9.997874495397135e-06,
      "loss": 0.4729,
      "num_tokens": 95832620.0,
      "step": 101
    },
    {
      "epoch": 0.19980411361410383,
      "grad_norm": 0.09930874844991569,
      "learning_rate": 9.997685665512418e-06,
      "loss": 0.4687,
      "num_tokens": 96768246.0,
      "step": 102
    },
    {
      "epoch": 0.20176297747306562,
      "grad_norm": 0.09437867539319489,
      "learning_rate": 9.99748880284497e-06,
      "loss": 0.4647,
      "num_tokens": 97763421.0,
      "step": 103
    },
    {
      "epoch": 0.20372184133202742,
      "grad_norm": 0.09494153272122935,
      "learning_rate": 9.997283907711208e-06,
      "loss": 0.4711,
      "num_tokens": 98710496.0,
      "step": 104
    },
    {
      "epoch": 0.20568070519098922,
      "grad_norm": 0.09366365651865821,
      "learning_rate": 9.997070980440463e-06,
      "loss": 0.4488,
      "num_tokens": 99659612.0,
      "step": 105
    },
    {
      "epoch": 0.20763956904995104,
      "grad_norm": 0.10299014324035684,
      "learning_rate": 9.996850021374969e-06,
      "loss": 0.4773,
      "num_tokens": 100617208.0,
      "step": 106
    },
    {
      "epoch": 0.20959843290891284,
      "grad_norm": 0.09487793679625174,
      "learning_rate": 9.996621030869876e-06,
      "loss": 0.4669,
      "num_tokens": 101615021.0,
      "step": 107
    },
    {
      "epoch": 0.21155729676787463,
      "grad_norm": 0.09766041109201681,
      "learning_rate": 9.99638400929324e-06,
      "loss": 0.4668,
      "num_tokens": 102528013.0,
      "step": 108
    },
    {
      "epoch": 0.21351616062683643,
      "grad_norm": 0.09904813750168198,
      "learning_rate": 9.996138957026026e-06,
      "loss": 0.4649,
      "num_tokens": 103493513.0,
      "step": 109
    },
    {
      "epoch": 0.21547502448579825,
      "grad_norm": 0.09869728621227473,
      "learning_rate": 9.99588587446211e-06,
      "loss": 0.4734,
      "num_tokens": 104431043.0,
      "step": 110
    },
    {
      "epoch": 0.21743388834476005,
      "grad_norm": 0.09597599562526722,
      "learning_rate": 9.995624762008265e-06,
      "loss": 0.4577,
      "num_tokens": 105379859.0,
      "step": 111
    },
    {
      "epoch": 0.21939275220372184,
      "grad_norm": 0.09831372165074696,
      "learning_rate": 9.995355620084182e-06,
      "loss": 0.4609,
      "num_tokens": 106311570.0,
      "step": 112
    },
    {
      "epoch": 0.22135161606268364,
      "grad_norm": 0.09551136337814933,
      "learning_rate": 9.99507844912245e-06,
      "loss": 0.4609,
      "num_tokens": 107290051.0,
      "step": 113
    },
    {
      "epoch": 0.22331047992164543,
      "grad_norm": 0.09801363147795901,
      "learning_rate": 9.994793249568568e-06,
      "loss": 0.4541,
      "num_tokens": 108225334.0,
      "step": 114
    },
    {
      "epoch": 0.22526934378060726,
      "grad_norm": 0.09284290995644295,
      "learning_rate": 9.994500021880939e-06,
      "loss": 0.4575,
      "num_tokens": 109192825.0,
      "step": 115
    },
    {
      "epoch": 0.22722820763956905,
      "grad_norm": 0.09904086714200677,
      "learning_rate": 9.994198766530864e-06,
      "loss": 0.4591,
      "num_tokens": 110153495.0,
      "step": 116
    },
    {
      "epoch": 0.22918707149853085,
      "grad_norm": 0.09637123963622646,
      "learning_rate": 9.993889484002553e-06,
      "loss": 0.4464,
      "num_tokens": 111098418.0,
      "step": 117
    },
    {
      "epoch": 0.23114593535749264,
      "grad_norm": 0.10312546519736356,
      "learning_rate": 9.993572174793116e-06,
      "loss": 0.4524,
      "num_tokens": 112058709.0,
      "step": 118
    },
    {
      "epoch": 0.23310479921645447,
      "grad_norm": 0.09479021598118789,
      "learning_rate": 9.993246839412565e-06,
      "loss": 0.4572,
      "num_tokens": 113041947.0,
      "step": 119
    },
    {
      "epoch": 0.23506366307541626,
      "grad_norm": 0.0970162426335051,
      "learning_rate": 9.99291347838381e-06,
      "loss": 0.4597,
      "num_tokens": 113981192.0,
      "step": 120
    },
    {
      "epoch": 0.23702252693437806,
      "grad_norm": 0.09632502927371162,
      "learning_rate": 9.992572092242664e-06,
      "loss": 0.4809,
      "num_tokens": 114966318.0,
      "step": 121
    },
    {
      "epoch": 0.23898139079333985,
      "grad_norm": 0.10416236432460022,
      "learning_rate": 9.992222681537835e-06,
      "loss": 0.4567,
      "num_tokens": 115939878.0,
      "step": 122
    },
    {
      "epoch": 0.24094025465230168,
      "grad_norm": 0.09873768878554416,
      "learning_rate": 9.991865246830935e-06,
      "loss": 0.4664,
      "num_tokens": 116915166.0,
      "step": 123
    },
    {
      "epoch": 0.24289911851126347,
      "grad_norm": 0.1003905240372282,
      "learning_rate": 9.991499788696463e-06,
      "loss": 0.4707,
      "num_tokens": 117866964.0,
      "step": 124
    },
    {
      "epoch": 0.24485798237022527,
      "grad_norm": 0.0935735181673574,
      "learning_rate": 9.991126307721826e-06,
      "loss": 0.4684,
      "num_tokens": 118837592.0,
      "step": 125
    },
    {
      "epoch": 0.24681684622918706,
      "grad_norm": 0.0982612410275377,
      "learning_rate": 9.990744804507315e-06,
      "loss": 0.4701,
      "num_tokens": 119778663.0,
      "step": 126
    },
    {
      "epoch": 0.24877571008814886,
      "grad_norm": 0.09777596468391736,
      "learning_rate": 9.990355279666124e-06,
      "loss": 0.4645,
      "num_tokens": 120718675.0,
      "step": 127
    },
    {
      "epoch": 0.2507345739471107,
      "grad_norm": 0.10350691006912617,
      "learning_rate": 9.989957733824336e-06,
      "loss": 0.4781,
      "num_tokens": 121638180.0,
      "step": 128
    },
    {
      "epoch": 0.25269343780607245,
      "grad_norm": 0.10022440245488685,
      "learning_rate": 9.989552167620926e-06,
      "loss": 0.4612,
      "num_tokens": 122582958.0,
      "step": 129
    },
    {
      "epoch": 0.2546523016650343,
      "grad_norm": 0.10176036166639991,
      "learning_rate": 9.98913858170776e-06,
      "loss": 0.4776,
      "num_tokens": 123522480.0,
      "step": 130
    },
    {
      "epoch": 0.2566111655239961,
      "grad_norm": 0.09743072937780196,
      "learning_rate": 9.988716976749596e-06,
      "loss": 0.4612,
      "num_tokens": 124434998.0,
      "step": 131
    },
    {
      "epoch": 0.25857002938295787,
      "grad_norm": 0.10040714573425007,
      "learning_rate": 9.988287353424077e-06,
      "loss": 0.4632,
      "num_tokens": 125399319.0,
      "step": 132
    },
    {
      "epoch": 0.2605288932419197,
      "grad_norm": 0.09584888758063531,
      "learning_rate": 9.987849712421741e-06,
      "loss": 0.4506,
      "num_tokens": 126331517.0,
      "step": 133
    },
    {
      "epoch": 0.2624877571008815,
      "grad_norm": 0.09510636035826173,
      "learning_rate": 9.987404054446009e-06,
      "loss": 0.435,
      "num_tokens": 127286047.0,
      "step": 134
    },
    {
      "epoch": 0.2644466209598433,
      "grad_norm": 0.09737635875932876,
      "learning_rate": 9.986950380213182e-06,
      "loss": 0.4664,
      "num_tokens": 128243241.0,
      "step": 135
    },
    {
      "epoch": 0.2664054848188051,
      "grad_norm": 0.10324901586881302,
      "learning_rate": 9.986488690452456e-06,
      "loss": 0.4458,
      "num_tokens": 129183998.0,
      "step": 136
    },
    {
      "epoch": 0.2683643486777669,
      "grad_norm": 0.09511139653578601,
      "learning_rate": 9.986018985905901e-06,
      "loss": 0.4508,
      "num_tokens": 130112534.0,
      "step": 137
    },
    {
      "epoch": 0.2703232125367287,
      "grad_norm": 0.09701502066571129,
      "learning_rate": 9.985541267328479e-06,
      "loss": 0.4565,
      "num_tokens": 131075496.0,
      "step": 138
    },
    {
      "epoch": 0.2722820763956905,
      "grad_norm": 0.09540233096015516,
      "learning_rate": 9.98505553548802e-06,
      "loss": 0.4587,
      "num_tokens": 132049561.0,
      "step": 139
    },
    {
      "epoch": 0.2742409402546523,
      "grad_norm": 0.09394609726026784,
      "learning_rate": 9.984561791165245e-06,
      "loss": 0.4464,
      "num_tokens": 133008081.0,
      "step": 140
    },
    {
      "epoch": 0.2761998041136141,
      "grad_norm": 0.09773689866820262,
      "learning_rate": 9.984060035153752e-06,
      "loss": 0.4695,
      "num_tokens": 133956658.0,
      "step": 141
    },
    {
      "epoch": 0.2781586679725759,
      "grad_norm": 0.0961859719144124,
      "learning_rate": 9.983550268260009e-06,
      "loss": 0.4546,
      "num_tokens": 134926369.0,
      "step": 142
    },
    {
      "epoch": 0.2801175318315377,
      "grad_norm": 0.098413033893867,
      "learning_rate": 9.983032491303368e-06,
      "loss": 0.4702,
      "num_tokens": 135845886.0,
      "step": 143
    },
    {
      "epoch": 0.2820763956904995,
      "grad_norm": 0.09579879342736346,
      "learning_rate": 9.98250670511605e-06,
      "loss": 0.4672,
      "num_tokens": 136808351.0,
      "step": 144
    },
    {
      "epoch": 0.2840352595494613,
      "grad_norm": 0.09754525670821552,
      "learning_rate": 9.981972910543151e-06,
      "loss": 0.4784,
      "num_tokens": 137754233.0,
      "step": 145
    },
    {
      "epoch": 0.2859941234084231,
      "grad_norm": 0.1010148746273363,
      "learning_rate": 9.981431108442644e-06,
      "loss": 0.4505,
      "num_tokens": 138693576.0,
      "step": 146
    },
    {
      "epoch": 0.28795298726738494,
      "grad_norm": 0.1048396205973233,
      "learning_rate": 9.98088129968536e-06,
      "loss": 0.4667,
      "num_tokens": 139639294.0,
      "step": 147
    },
    {
      "epoch": 0.2899118511263467,
      "grad_norm": 0.09495042331450833,
      "learning_rate": 9.980323485155013e-06,
      "loss": 0.4508,
      "num_tokens": 140623819.0,
      "step": 148
    },
    {
      "epoch": 0.29187071498530853,
      "grad_norm": 0.09758568279279468,
      "learning_rate": 9.979757665748177e-06,
      "loss": 0.4433,
      "num_tokens": 141575936.0,
      "step": 149
    },
    {
      "epoch": 0.2938295788442703,
      "grad_norm": 0.09686099257479716,
      "learning_rate": 9.979183842374294e-06,
      "loss": 0.4663,
      "num_tokens": 142543786.0,
      "step": 150
    },
    {
      "epoch": 0.2957884427032321,
      "grad_norm": 0.09499912774299393,
      "learning_rate": 9.97860201595567e-06,
      "loss": 0.4562,
      "num_tokens": 143541287.0,
      "step": 151
    },
    {
      "epoch": 0.29774730656219395,
      "grad_norm": 0.09994523700180594,
      "learning_rate": 9.978012187427474e-06,
      "loss": 0.4721,
      "num_tokens": 144461732.0,
      "step": 152
    },
    {
      "epoch": 0.2997061704211557,
      "grad_norm": 0.10057207703671751,
      "learning_rate": 9.97741435773774e-06,
      "loss": 0.4609,
      "num_tokens": 145407103.0,
      "step": 153
    },
    {
      "epoch": 0.30166503428011754,
      "grad_norm": 0.10188797169880409,
      "learning_rate": 9.976808527847357e-06,
      "loss": 0.4613,
      "num_tokens": 146342159.0,
      "step": 154
    },
    {
      "epoch": 0.3036238981390793,
      "grad_norm": 0.10147521760797401,
      "learning_rate": 9.97619469873008e-06,
      "loss": 0.4612,
      "num_tokens": 147243557.0,
      "step": 155
    },
    {
      "epoch": 0.30558276199804113,
      "grad_norm": 0.09355508554161461,
      "learning_rate": 9.975572871372513e-06,
      "loss": 0.4498,
      "num_tokens": 148193129.0,
      "step": 156
    },
    {
      "epoch": 0.30754162585700295,
      "grad_norm": 0.09932265253341842,
      "learning_rate": 9.974943046774121e-06,
      "loss": 0.4669,
      "num_tokens": 149159142.0,
      "step": 157
    },
    {
      "epoch": 0.3095004897159647,
      "grad_norm": 0.10334059832200888,
      "learning_rate": 9.97430522594722e-06,
      "loss": 0.4565,
      "num_tokens": 150120735.0,
      "step": 158
    },
    {
      "epoch": 0.31145935357492655,
      "grad_norm": 0.10284601693636423,
      "learning_rate": 9.973659409916982e-06,
      "loss": 0.4619,
      "num_tokens": 151058295.0,
      "step": 159
    },
    {
      "epoch": 0.31341821743388837,
      "grad_norm": 0.1031684679282704,
      "learning_rate": 9.973005599721425e-06,
      "loss": 0.474,
      "num_tokens": 152000140.0,
      "step": 160
    },
    {
      "epoch": 0.31537708129285014,
      "grad_norm": 0.09461282557066732,
      "learning_rate": 9.972343796411423e-06,
      "loss": 0.4488,
      "num_tokens": 152947362.0,
      "step": 161
    },
    {
      "epoch": 0.31733594515181196,
      "grad_norm": 0.09383580011255986,
      "learning_rate": 9.971674001050687e-06,
      "loss": 0.4547,
      "num_tokens": 153891057.0,
      "step": 162
    },
    {
      "epoch": 0.31929480901077373,
      "grad_norm": 0.10351535825224005,
      "learning_rate": 9.970996214715782e-06,
      "loss": 0.4461,
      "num_tokens": 154832165.0,
      "step": 163
    },
    {
      "epoch": 0.32125367286973555,
      "grad_norm": 0.10053717067009102,
      "learning_rate": 9.970310438496115e-06,
      "loss": 0.466,
      "num_tokens": 155778733.0,
      "step": 164
    },
    {
      "epoch": 0.3232125367286974,
      "grad_norm": 0.0977421156293791,
      "learning_rate": 9.969616673493932e-06,
      "loss": 0.456,
      "num_tokens": 156722675.0,
      "step": 165
    },
    {
      "epoch": 0.32517140058765914,
      "grad_norm": 0.09462641000536073,
      "learning_rate": 9.968914920824327e-06,
      "loss": 0.4481,
      "num_tokens": 157706865.0,
      "step": 166
    },
    {
      "epoch": 0.32713026444662097,
      "grad_norm": 0.09685626465188592,
      "learning_rate": 9.96820518161522e-06,
      "loss": 0.4612,
      "num_tokens": 158677094.0,
      "step": 167
    },
    {
      "epoch": 0.32908912830558273,
      "grad_norm": 0.10525691473208647,
      "learning_rate": 9.967487457007382e-06,
      "loss": 0.4725,
      "num_tokens": 159628301.0,
      "step": 168
    },
    {
      "epoch": 0.33104799216454456,
      "grad_norm": 0.10515616858242463,
      "learning_rate": 9.96676174815441e-06,
      "loss": 0.4647,
      "num_tokens": 160509714.0,
      "step": 169
    },
    {
      "epoch": 0.3330068560235064,
      "grad_norm": 0.09361787506160345,
      "learning_rate": 9.966028056222734e-06,
      "loss": 0.4592,
      "num_tokens": 161487383.0,
      "step": 170
    },
    {
      "epoch": 0.33496571988246815,
      "grad_norm": 0.10545719861616981,
      "learning_rate": 9.965286382391619e-06,
      "loss": 0.4717,
      "num_tokens": 162412904.0,
      "step": 171
    },
    {
      "epoch": 0.33692458374143,
      "grad_norm": 0.09518391320721968,
      "learning_rate": 9.964536727853158e-06,
      "loss": 0.4557,
      "num_tokens": 163353241.0,
      "step": 172
    },
    {
      "epoch": 0.3388834476003918,
      "grad_norm": 0.09686861040030187,
      "learning_rate": 9.96377909381227e-06,
      "loss": 0.4688,
      "num_tokens": 164297299.0,
      "step": 173
    },
    {
      "epoch": 0.34084231145935356,
      "grad_norm": 0.09648744616787576,
      "learning_rate": 9.963013481486704e-06,
      "loss": 0.4438,
      "num_tokens": 165233075.0,
      "step": 174
    },
    {
      "epoch": 0.3428011753183154,
      "grad_norm": 0.09489418835775852,
      "learning_rate": 9.962239892107024e-06,
      "loss": 0.4666,
      "num_tokens": 166195433.0,
      "step": 175
    },
    {
      "epoch": 0.34476003917727716,
      "grad_norm": 0.09528846230202102,
      "learning_rate": 9.961458326916624e-06,
      "loss": 0.4514,
      "num_tokens": 167149954.0,
      "step": 176
    },
    {
      "epoch": 0.346718903036239,
      "grad_norm": 0.09420971492743258,
      "learning_rate": 9.960668787171713e-06,
      "loss": 0.4456,
      "num_tokens": 168131808.0,
      "step": 177
    },
    {
      "epoch": 0.3486777668952008,
      "grad_norm": 0.09878621581394058,
      "learning_rate": 9.959871274141319e-06,
      "loss": 0.4499,
      "num_tokens": 169108543.0,
      "step": 178
    },
    {
      "epoch": 0.35063663075416257,
      "grad_norm": 0.09722607708153193,
      "learning_rate": 9.959065789107285e-06,
      "loss": 0.4631,
      "num_tokens": 170024930.0,
      "step": 179
    },
    {
      "epoch": 0.3525954946131244,
      "grad_norm": 0.09715582363770695,
      "learning_rate": 9.958252333364266e-06,
      "loss": 0.458,
      "num_tokens": 170956490.0,
      "step": 180
    },
    {
      "epoch": 0.3545543584720862,
      "grad_norm": 0.10091101459947915,
      "learning_rate": 9.957430908219734e-06,
      "loss": 0.4649,
      "num_tokens": 171888577.0,
      "step": 181
    },
    {
      "epoch": 0.356513222331048,
      "grad_norm": 0.0956927902389695,
      "learning_rate": 9.956601514993962e-06,
      "loss": 0.454,
      "num_tokens": 172846773.0,
      "step": 182
    },
    {
      "epoch": 0.3584720861900098,
      "grad_norm": 0.0996617214313076,
      "learning_rate": 9.955764155020037e-06,
      "loss": 0.4674,
      "num_tokens": 173781976.0,
      "step": 183
    },
    {
      "epoch": 0.3604309500489716,
      "grad_norm": 0.09237992855339172,
      "learning_rate": 9.954918829643845e-06,
      "loss": 0.4489,
      "num_tokens": 174750604.0,
      "step": 184
    },
    {
      "epoch": 0.3623898139079334,
      "grad_norm": 0.09898987379939003,
      "learning_rate": 9.954065540224081e-06,
      "loss": 0.4517,
      "num_tokens": 175654902.0,
      "step": 185
    },
    {
      "epoch": 0.3643486777668952,
      "grad_norm": 0.10068051655230977,
      "learning_rate": 9.953204288132234e-06,
      "loss": 0.4414,
      "num_tokens": 176586402.0,
      "step": 186
    },
    {
      "epoch": 0.366307541625857,
      "grad_norm": 0.09827300657050554,
      "learning_rate": 9.9523350747526e-06,
      "loss": 0.4648,
      "num_tokens": 177515759.0,
      "step": 187
    },
    {
      "epoch": 0.3682664054848188,
      "grad_norm": 0.09367935695651718,
      "learning_rate": 9.95145790148226e-06,
      "loss": 0.4525,
      "num_tokens": 178476413.0,
      "step": 188
    },
    {
      "epoch": 0.3702252693437806,
      "grad_norm": 0.0968048834780351,
      "learning_rate": 9.9505727697311e-06,
      "loss": 0.4586,
      "num_tokens": 179446696.0,
      "step": 189
    },
    {
      "epoch": 0.3721841332027424,
      "grad_norm": 0.10315128405197668,
      "learning_rate": 9.94967968092179e-06,
      "loss": 0.4597,
      "num_tokens": 180353014.0,
      "step": 190
    },
    {
      "epoch": 0.37414299706170423,
      "grad_norm": 0.09643704412573983,
      "learning_rate": 9.948778636489793e-06,
      "loss": 0.4585,
      "num_tokens": 181352431.0,
      "step": 191
    },
    {
      "epoch": 0.376101860920666,
      "grad_norm": 0.09753806609190649,
      "learning_rate": 9.947869637883359e-06,
      "loss": 0.4436,
      "num_tokens": 182300791.0,
      "step": 192
    },
    {
      "epoch": 0.3780607247796278,
      "grad_norm": 0.09620530562443642,
      "learning_rate": 9.94695268656352e-06,
      "loss": 0.4474,
      "num_tokens": 183242162.0,
      "step": 193
    },
    {
      "epoch": 0.38001958863858964,
      "grad_norm": 0.09601966800599304,
      "learning_rate": 9.946027784004096e-06,
      "loss": 0.4384,
      "num_tokens": 184192887.0,
      "step": 194
    },
    {
      "epoch": 0.3819784524975514,
      "grad_norm": 0.09912862956296686,
      "learning_rate": 9.945094931691677e-06,
      "loss": 0.4514,
      "num_tokens": 185101798.0,
      "step": 195
    },
    {
      "epoch": 0.38393731635651324,
      "grad_norm": 0.09356569357224666,
      "learning_rate": 9.944154131125643e-06,
      "loss": 0.4453,
      "num_tokens": 186050228.0,
      "step": 196
    },
    {
      "epoch": 0.385896180215475,
      "grad_norm": 0.10936038967219135,
      "learning_rate": 9.943205383818142e-06,
      "loss": 0.4523,
      "num_tokens": 187011757.0,
      "step": 197
    },
    {
      "epoch": 0.3878550440744368,
      "grad_norm": 0.10152135007297276,
      "learning_rate": 9.942248691294092e-06,
      "loss": 0.4437,
      "num_tokens": 187933276.0,
      "step": 198
    },
    {
      "epoch": 0.38981390793339865,
      "grad_norm": 0.09454429685683488,
      "learning_rate": 9.941284055091191e-06,
      "loss": 0.4566,
      "num_tokens": 188880767.0,
      "step": 199
    },
    {
      "epoch": 0.3917727717923604,
      "grad_norm": 0.09458942177389343,
      "learning_rate": 9.940311476759896e-06,
      "loss": 0.4529,
      "num_tokens": 189882809.0,
      "step": 200
    },
    {
      "epoch": 0.39373163565132224,
      "grad_norm": 0.09888960327025009,
      "learning_rate": 9.939330957863433e-06,
      "loss": 0.4524,
      "num_tokens": 190814376.0,
      "step": 201
    },
    {
      "epoch": 0.395690499510284,
      "grad_norm": 0.09688253630738085,
      "learning_rate": 9.938342499977791e-06,
      "loss": 0.4648,
      "num_tokens": 191767854.0,
      "step": 202
    },
    {
      "epoch": 0.39764936336924583,
      "grad_norm": 0.09527070915427467,
      "learning_rate": 9.93734610469172e-06,
      "loss": 0.4523,
      "num_tokens": 192778592.0,
      "step": 203
    },
    {
      "epoch": 0.39960822722820766,
      "grad_norm": 0.09610244703120024,
      "learning_rate": 9.936341773606723e-06,
      "loss": 0.4563,
      "num_tokens": 193743846.0,
      "step": 204
    },
    {
      "epoch": 0.4015670910871694,
      "grad_norm": 0.0954323129485936,
      "learning_rate": 9.935329508337066e-06,
      "loss": 0.4221,
      "num_tokens": 194703539.0,
      "step": 205
    },
    {
      "epoch": 0.40352595494613125,
      "grad_norm": 0.09224066794714314,
      "learning_rate": 9.934309310509761e-06,
      "loss": 0.4633,
      "num_tokens": 195642924.0,
      "step": 206
    },
    {
      "epoch": 0.4054848188050931,
      "grad_norm": 0.09215081705810675,
      "learning_rate": 9.933281181764573e-06,
      "loss": 0.451,
      "num_tokens": 196600794.0,
      "step": 207
    },
    {
      "epoch": 0.40744368266405484,
      "grad_norm": 0.092979103110267,
      "learning_rate": 9.932245123754017e-06,
      "loss": 0.472,
      "num_tokens": 197554677.0,
      "step": 208
    },
    {
      "epoch": 0.40940254652301666,
      "grad_norm": 0.08910947696799769,
      "learning_rate": 9.931201138143342e-06,
      "loss": 0.4524,
      "num_tokens": 198536397.0,
      "step": 209
    },
    {
      "epoch": 0.41136141038197843,
      "grad_norm": 0.09604903925261911,
      "learning_rate": 9.930149226610555e-06,
      "loss": 0.4588,
      "num_tokens": 199467040.0,
      "step": 210
    },
    {
      "epoch": 0.41332027424094026,
      "grad_norm": 0.0996152785331448,
      "learning_rate": 9.929089390846389e-06,
      "loss": 0.4684,
      "num_tokens": 200421819.0,
      "step": 211
    },
    {
      "epoch": 0.4152791380999021,
      "grad_norm": 0.09463592045098039,
      "learning_rate": 9.928021632554318e-06,
      "loss": 0.4499,
      "num_tokens": 201345914.0,
      "step": 212
    },
    {
      "epoch": 0.41723800195886385,
      "grad_norm": 0.09975797702654883,
      "learning_rate": 9.926945953450555e-06,
      "loss": 0.4541,
      "num_tokens": 202269876.0,
      "step": 213
    },
    {
      "epoch": 0.41919686581782567,
      "grad_norm": 0.08908783685003974,
      "learning_rate": 9.925862355264035e-06,
      "loss": 0.4456,
      "num_tokens": 203204620.0,
      "step": 214
    },
    {
      "epoch": 0.42115572967678744,
      "grad_norm": 0.09551668100580148,
      "learning_rate": 9.924770839736429e-06,
      "loss": 0.4768,
      "num_tokens": 204123253.0,
      "step": 215
    },
    {
      "epoch": 0.42311459353574926,
      "grad_norm": 0.09703383213342595,
      "learning_rate": 9.923671408622128e-06,
      "loss": 0.4612,
      "num_tokens": 205070357.0,
      "step": 216
    },
    {
      "epoch": 0.4250734573947111,
      "grad_norm": 0.10500185746203437,
      "learning_rate": 9.922564063688248e-06,
      "loss": 0.4541,
      "num_tokens": 205981152.0,
      "step": 217
    },
    {
      "epoch": 0.42703232125367285,
      "grad_norm": 0.0944810782117902,
      "learning_rate": 9.92144880671463e-06,
      "loss": 0.444,
      "num_tokens": 206925820.0,
      "step": 218
    },
    {
      "epoch": 0.4289911851126347,
      "grad_norm": 0.09409864592076014,
      "learning_rate": 9.920325639493821e-06,
      "loss": 0.4492,
      "num_tokens": 207894358.0,
      "step": 219
    },
    {
      "epoch": 0.4309500489715965,
      "grad_norm": 0.09397986303935289,
      "learning_rate": 9.919194563831093e-06,
      "loss": 0.4509,
      "num_tokens": 208853063.0,
      "step": 220
    },
    {
      "epoch": 0.43290891283055827,
      "grad_norm": 0.09483722894944634,
      "learning_rate": 9.918055581544422e-06,
      "loss": 0.451,
      "num_tokens": 209803005.0,
      "step": 221
    },
    {
      "epoch": 0.4348677766895201,
      "grad_norm": 0.09240936963096365,
      "learning_rate": 9.916908694464494e-06,
      "loss": 0.4402,
      "num_tokens": 210743515.0,
      "step": 222
    },
    {
      "epoch": 0.43682664054848186,
      "grad_norm": 0.09969646231609247,
      "learning_rate": 9.915753904434702e-06,
      "loss": 0.4649,
      "num_tokens": 211678649.0,
      "step": 223
    },
    {
      "epoch": 0.4387855044074437,
      "grad_norm": 0.09630734303627361,
      "learning_rate": 9.914591213311142e-06,
      "loss": 0.4436,
      "num_tokens": 212640252.0,
      "step": 224
    },
    {
      "epoch": 0.4407443682664055,
      "grad_norm": 0.0901899193736916,
      "learning_rate": 9.913420622962606e-06,
      "loss": 0.458,
      "num_tokens": 213589770.0,
      "step": 225
    },
    {
      "epoch": 0.4427032321253673,
      "grad_norm": 0.09172175274308668,
      "learning_rate": 9.912242135270585e-06,
      "loss": 0.4562,
      "num_tokens": 214576120.0,
      "step": 226
    },
    {
      "epoch": 0.4446620959843291,
      "grad_norm": 0.09612569240774674,
      "learning_rate": 9.911055752129264e-06,
      "loss": 0.4434,
      "num_tokens": 215521108.0,
      "step": 227
    },
    {
      "epoch": 0.44662095984329087,
      "grad_norm": 0.09344083762321673,
      "learning_rate": 9.909861475445517e-06,
      "loss": 0.447,
      "num_tokens": 216458762.0,
      "step": 228
    },
    {
      "epoch": 0.4485798237022527,
      "grad_norm": 0.09106724915275911,
      "learning_rate": 9.908659307138905e-06,
      "loss": 0.4467,
      "num_tokens": 217433105.0,
      "step": 229
    },
    {
      "epoch": 0.4505386875612145,
      "grad_norm": 0.10357482160918961,
      "learning_rate": 9.907449249141673e-06,
      "loss": 0.4602,
      "num_tokens": 218362026.0,
      "step": 230
    },
    {
      "epoch": 0.4524975514201763,
      "grad_norm": 0.09185665370892589,
      "learning_rate": 9.90623130339875e-06,
      "loss": 0.4506,
      "num_tokens": 219311418.0,
      "step": 231
    },
    {
      "epoch": 0.4544564152791381,
      "grad_norm": 0.09647090136629549,
      "learning_rate": 9.90500547186774e-06,
      "loss": 0.4714,
      "num_tokens": 220312439.0,
      "step": 232
    },
    {
      "epoch": 0.4564152791380999,
      "grad_norm": 0.09242560530540733,
      "learning_rate": 9.903771756518919e-06,
      "loss": 0.4423,
      "num_tokens": 221246761.0,
      "step": 233
    },
    {
      "epoch": 0.4583741429970617,
      "grad_norm": 0.09337682539103548,
      "learning_rate": 9.902530159335245e-06,
      "loss": 0.453,
      "num_tokens": 222199221.0,
      "step": 234
    },
    {
      "epoch": 0.4603330068560235,
      "grad_norm": 0.10012730710127711,
      "learning_rate": 9.90128068231233e-06,
      "loss": 0.4518,
      "num_tokens": 223155514.0,
      "step": 235
    },
    {
      "epoch": 0.4622918707149853,
      "grad_norm": 0.10405044615190953,
      "learning_rate": 9.900023327458466e-06,
      "loss": 0.4471,
      "num_tokens": 224050665.0,
      "step": 236
    },
    {
      "epoch": 0.4642507345739471,
      "grad_norm": 0.09671175081752656,
      "learning_rate": 9.898758096794598e-06,
      "loss": 0.4449,
      "num_tokens": 224973216.0,
      "step": 237
    },
    {
      "epoch": 0.46620959843290893,
      "grad_norm": 0.09221217335060064,
      "learning_rate": 9.897484992354329e-06,
      "loss": 0.4615,
      "num_tokens": 225905305.0,
      "step": 238
    },
    {
      "epoch": 0.4681684622918707,
      "grad_norm": 0.09305484057306082,
      "learning_rate": 9.896204016183924e-06,
      "loss": 0.4534,
      "num_tokens": 226893498.0,
      "step": 239
    },
    {
      "epoch": 0.4701273261508325,
      "grad_norm": 0.08985035827138547,
      "learning_rate": 9.894915170342297e-06,
      "loss": 0.4474,
      "num_tokens": 227847424.0,
      "step": 240
    },
    {
      "epoch": 0.4720861900097943,
      "grad_norm": 0.09671387702973935,
      "learning_rate": 9.893618456901006e-06,
      "loss": 0.4598,
      "num_tokens": 228770094.0,
      "step": 241
    },
    {
      "epoch": 0.4740450538687561,
      "grad_norm": 0.09511629695889492,
      "learning_rate": 9.892313877944263e-06,
      "loss": 0.4375,
      "num_tokens": 229729772.0,
      "step": 242
    },
    {
      "epoch": 0.47600391772771794,
      "grad_norm": 0.09380621456050775,
      "learning_rate": 9.891001435568917e-06,
      "loss": 0.4508,
      "num_tokens": 230686754.0,
      "step": 243
    },
    {
      "epoch": 0.4779627815866797,
      "grad_norm": 0.09198683670913939,
      "learning_rate": 9.889681131884459e-06,
      "loss": 0.4535,
      "num_tokens": 231634183.0,
      "step": 244
    },
    {
      "epoch": 0.47992164544564153,
      "grad_norm": 0.09502283634944488,
      "learning_rate": 9.888352969013014e-06,
      "loss": 0.4624,
      "num_tokens": 232579231.0,
      "step": 245
    },
    {
      "epoch": 0.48188050930460335,
      "grad_norm": 0.09040936834978072,
      "learning_rate": 9.887016949089334e-06,
      "loss": 0.4523,
      "num_tokens": 233559115.0,
      "step": 246
    },
    {
      "epoch": 0.4838393731635651,
      "grad_norm": 0.09254363806857926,
      "learning_rate": 9.88567307426081e-06,
      "loss": 0.4495,
      "num_tokens": 234481493.0,
      "step": 247
    },
    {
      "epoch": 0.48579823702252695,
      "grad_norm": 0.1009075494272476,
      "learning_rate": 9.88432134668745e-06,
      "loss": 0.4732,
      "num_tokens": 235430540.0,
      "step": 248
    },
    {
      "epoch": 0.4877571008814887,
      "grad_norm": 0.09171640568723448,
      "learning_rate": 9.882961768541887e-06,
      "loss": 0.4429,
      "num_tokens": 236376157.0,
      "step": 249
    },
    {
      "epoch": 0.48971596474045054,
      "grad_norm": 0.08764846780429664,
      "learning_rate": 9.881594342009374e-06,
      "loss": 0.4515,
      "num_tokens": 237357177.0,
      "step": 250
    },
    {
      "epoch": 0.49167482859941236,
      "grad_norm": 0.09579368673659945,
      "learning_rate": 9.88021906928777e-06,
      "loss": 0.4603,
      "num_tokens": 238282985.0,
      "step": 251
    },
    {
      "epoch": 0.49363369245837413,
      "grad_norm": 0.09445820333911897,
      "learning_rate": 9.878835952587559e-06,
      "loss": 0.4509,
      "num_tokens": 239244676.0,
      "step": 252
    },
    {
      "epoch": 0.49559255631733595,
      "grad_norm": 0.08951040263052458,
      "learning_rate": 9.87744499413182e-06,
      "loss": 0.4417,
      "num_tokens": 240205678.0,
      "step": 253
    },
    {
      "epoch": 0.4975514201762977,
      "grad_norm": 0.09826525864645287,
      "learning_rate": 9.876046196156243e-06,
      "loss": 0.4571,
      "num_tokens": 241166522.0,
      "step": 254
    },
    {
      "epoch": 0.49951028403525954,
      "grad_norm": 0.09249167457583525,
      "learning_rate": 9.874639560909118e-06,
      "loss": 0.4318,
      "num_tokens": 242118412.0,
      "step": 255
    },
    {
      "epoch": 0.5014691478942214,
      "grad_norm": 0.1043974877643401,
      "learning_rate": 9.87322509065133e-06,
      "loss": 0.4604,
      "num_tokens": 243072617.0,
      "step": 256
    },
    {
      "epoch": 0.5034280117531832,
      "grad_norm": 0.08975347248528376,
      "learning_rate": 9.871802787656356e-06,
      "loss": 0.4518,
      "num_tokens": 244029342.0,
      "step": 257
    },
    {
      "epoch": 0.5053868756121449,
      "grad_norm": 0.09772314602053161,
      "learning_rate": 9.870372654210265e-06,
      "loss": 0.4596,
      "num_tokens": 244972993.0,
      "step": 258
    },
    {
      "epoch": 0.5073457394711067,
      "grad_norm": 0.09532417433774386,
      "learning_rate": 9.868934692611712e-06,
      "loss": 0.4616,
      "num_tokens": 245928400.0,
      "step": 259
    },
    {
      "epoch": 0.5093046033300686,
      "grad_norm": 0.09888763723078879,
      "learning_rate": 9.867488905171934e-06,
      "loss": 0.4406,
      "num_tokens": 246856281.0,
      "step": 260
    },
    {
      "epoch": 0.5112634671890304,
      "grad_norm": 0.09822568450389157,
      "learning_rate": 9.866035294214744e-06,
      "loss": 0.4516,
      "num_tokens": 247805102.0,
      "step": 261
    },
    {
      "epoch": 0.5132223310479922,
      "grad_norm": 0.09195695916021966,
      "learning_rate": 9.864573862076533e-06,
      "loss": 0.448,
      "num_tokens": 248747606.0,
      "step": 262
    },
    {
      "epoch": 0.515181194906954,
      "grad_norm": 0.08829654375672243,
      "learning_rate": 9.863104611106262e-06,
      "loss": 0.4433,
      "num_tokens": 249691221.0,
      "step": 263
    },
    {
      "epoch": 0.5171400587659157,
      "grad_norm": 0.10011745782468426,
      "learning_rate": 9.861627543665456e-06,
      "loss": 0.4511,
      "num_tokens": 250604365.0,
      "step": 264
    },
    {
      "epoch": 0.5190989226248776,
      "grad_norm": 0.09731081354203551,
      "learning_rate": 9.86014266212821e-06,
      "loss": 0.4464,
      "num_tokens": 251551288.0,
      "step": 265
    },
    {
      "epoch": 0.5210577864838394,
      "grad_norm": 0.09819161282037218,
      "learning_rate": 9.858649968881173e-06,
      "loss": 0.4551,
      "num_tokens": 252523236.0,
      "step": 266
    },
    {
      "epoch": 0.5230166503428012,
      "grad_norm": 0.09666669659878116,
      "learning_rate": 9.85714946632355e-06,
      "loss": 0.467,
      "num_tokens": 253453639.0,
      "step": 267
    },
    {
      "epoch": 0.524975514201763,
      "grad_norm": 0.08908227776173835,
      "learning_rate": 9.855641156867103e-06,
      "loss": 0.4469,
      "num_tokens": 254401672.0,
      "step": 268
    },
    {
      "epoch": 0.5269343780607247,
      "grad_norm": 0.09282226968659334,
      "learning_rate": 9.854125042936137e-06,
      "loss": 0.4521,
      "num_tokens": 255371297.0,
      "step": 269
    },
    {
      "epoch": 0.5288932419196866,
      "grad_norm": 0.09828684199638049,
      "learning_rate": 9.852601126967502e-06,
      "loss": 0.4532,
      "num_tokens": 256321317.0,
      "step": 270
    },
    {
      "epoch": 0.5308521057786484,
      "grad_norm": 0.09683561527174155,
      "learning_rate": 9.85106941141059e-06,
      "loss": 0.4416,
      "num_tokens": 257257547.0,
      "step": 271
    },
    {
      "epoch": 0.5328109696376102,
      "grad_norm": 0.09016846920338978,
      "learning_rate": 9.849529898727328e-06,
      "loss": 0.4521,
      "num_tokens": 258196782.0,
      "step": 272
    },
    {
      "epoch": 0.534769833496572,
      "grad_norm": 0.10246950371582385,
      "learning_rate": 9.847982591392173e-06,
      "loss": 0.4533,
      "num_tokens": 259113699.0,
      "step": 273
    },
    {
      "epoch": 0.5367286973555337,
      "grad_norm": 0.09970486600889314,
      "learning_rate": 9.846427491892117e-06,
      "loss": 0.4531,
      "num_tokens": 260062450.0,
      "step": 274
    },
    {
      "epoch": 0.5386875612144956,
      "grad_norm": 0.09086206195124324,
      "learning_rate": 9.844864602726671e-06,
      "loss": 0.4339,
      "num_tokens": 261015365.0,
      "step": 275
    },
    {
      "epoch": 0.5406464250734574,
      "grad_norm": 0.08951612961854831,
      "learning_rate": 9.843293926407866e-06,
      "loss": 0.4416,
      "num_tokens": 261951943.0,
      "step": 276
    },
    {
      "epoch": 0.5426052889324192,
      "grad_norm": 0.09759160821842901,
      "learning_rate": 9.841715465460251e-06,
      "loss": 0.4672,
      "num_tokens": 262870691.0,
      "step": 277
    },
    {
      "epoch": 0.544564152791381,
      "grad_norm": 0.09110634732133736,
      "learning_rate": 9.84012922242089e-06,
      "loss": 0.4566,
      "num_tokens": 263836081.0,
      "step": 278
    },
    {
      "epoch": 0.5465230166503428,
      "grad_norm": 0.09072382602508929,
      "learning_rate": 9.838535199839348e-06,
      "loss": 0.4506,
      "num_tokens": 264812493.0,
      "step": 279
    },
    {
      "epoch": 0.5484818805093046,
      "grad_norm": 0.08893797740730669,
      "learning_rate": 9.8369334002777e-06,
      "loss": 0.4456,
      "num_tokens": 265757155.0,
      "step": 280
    },
    {
      "epoch": 0.5504407443682664,
      "grad_norm": 0.09018712653195393,
      "learning_rate": 9.835323826310522e-06,
      "loss": 0.4578,
      "num_tokens": 266692706.0,
      "step": 281
    },
    {
      "epoch": 0.5523996082272282,
      "grad_norm": 0.09131692346053066,
      "learning_rate": 9.833706480524878e-06,
      "loss": 0.449,
      "num_tokens": 267648410.0,
      "step": 282
    },
    {
      "epoch": 0.55435847208619,
      "grad_norm": 0.08739332252346171,
      "learning_rate": 9.832081365520334e-06,
      "loss": 0.4508,
      "num_tokens": 268595990.0,
      "step": 283
    },
    {
      "epoch": 0.5563173359451518,
      "grad_norm": 0.09139986037545283,
      "learning_rate": 9.830448483908935e-06,
      "loss": 0.434,
      "num_tokens": 269537447.0,
      "step": 284
    },
    {
      "epoch": 0.5582761998041136,
      "grad_norm": 0.09065811066751005,
      "learning_rate": 9.82880783831521e-06,
      "loss": 0.4561,
      "num_tokens": 270486740.0,
      "step": 285
    },
    {
      "epoch": 0.5602350636630754,
      "grad_norm": 0.09054968942661933,
      "learning_rate": 9.827159431376176e-06,
      "loss": 0.4465,
      "num_tokens": 271440384.0,
      "step": 286
    },
    {
      "epoch": 0.5621939275220372,
      "grad_norm": 0.08513515561595217,
      "learning_rate": 9.825503265741314e-06,
      "loss": 0.4312,
      "num_tokens": 272416401.0,
      "step": 287
    },
    {
      "epoch": 0.564152791380999,
      "grad_norm": 0.091901688905363,
      "learning_rate": 9.823839344072582e-06,
      "loss": 0.4486,
      "num_tokens": 273346804.0,
      "step": 288
    },
    {
      "epoch": 0.5661116552399609,
      "grad_norm": 0.08792598502990735,
      "learning_rate": 9.822167669044398e-06,
      "loss": 0.4495,
      "num_tokens": 274284897.0,
      "step": 289
    },
    {
      "epoch": 0.5680705190989226,
      "grad_norm": 0.0933845119975612,
      "learning_rate": 9.82048824334365e-06,
      "loss": 0.4505,
      "num_tokens": 275232899.0,
      "step": 290
    },
    {
      "epoch": 0.5700293829578844,
      "grad_norm": 0.08847993498549075,
      "learning_rate": 9.818801069669679e-06,
      "loss": 0.4532,
      "num_tokens": 276190165.0,
      "step": 291
    },
    {
      "epoch": 0.5719882468168462,
      "grad_norm": 0.08838487689527708,
      "learning_rate": 9.817106150734281e-06,
      "loss": 0.4653,
      "num_tokens": 277157577.0,
      "step": 292
    },
    {
      "epoch": 0.5739471106758081,
      "grad_norm": 0.0891535266335383,
      "learning_rate": 9.815403489261698e-06,
      "loss": 0.453,
      "num_tokens": 278086368.0,
      "step": 293
    },
    {
      "epoch": 0.5759059745347699,
      "grad_norm": 0.09765021390877121,
      "learning_rate": 9.81369308798862e-06,
      "loss": 0.4517,
      "num_tokens": 279014464.0,
      "step": 294
    },
    {
      "epoch": 0.5778648383937316,
      "grad_norm": 0.08813812734907714,
      "learning_rate": 9.811974949664176e-06,
      "loss": 0.4426,
      "num_tokens": 279972567.0,
      "step": 295
    },
    {
      "epoch": 0.5798237022526934,
      "grad_norm": 0.09170718468035989,
      "learning_rate": 9.810249077049929e-06,
      "loss": 0.4549,
      "num_tokens": 280921291.0,
      "step": 296
    },
    {
      "epoch": 0.5817825661116552,
      "grad_norm": 0.09043515510180414,
      "learning_rate": 9.808515472919876e-06,
      "loss": 0.4519,
      "num_tokens": 281892629.0,
      "step": 297
    },
    {
      "epoch": 0.5837414299706171,
      "grad_norm": 0.08901240213313642,
      "learning_rate": 9.80677414006044e-06,
      "loss": 0.4554,
      "num_tokens": 282842117.0,
      "step": 298
    },
    {
      "epoch": 0.5857002938295789,
      "grad_norm": 0.09261848310830277,
      "learning_rate": 9.805025081270467e-06,
      "loss": 0.4364,
      "num_tokens": 283790343.0,
      "step": 299
    },
    {
      "epoch": 0.5876591576885406,
      "grad_norm": 0.09437822926353868,
      "learning_rate": 9.803268299361217e-06,
      "loss": 0.4555,
      "num_tokens": 284725518.0,
      "step": 300
    },
    {
      "epoch": 0.5896180215475024,
      "grad_norm": 0.10839960974848895,
      "learning_rate": 9.80150379715637e-06,
      "loss": 0.4791,
      "num_tokens": 285646307.0,
      "step": 301
    },
    {
      "epoch": 0.5915768854064642,
      "grad_norm": 0.09620758857257257,
      "learning_rate": 9.79973157749201e-06,
      "loss": 0.4481,
      "num_tokens": 286598500.0,
      "step": 302
    },
    {
      "epoch": 0.5935357492654261,
      "grad_norm": 0.08658250998764305,
      "learning_rate": 9.797951643216628e-06,
      "loss": 0.4597,
      "num_tokens": 287563888.0,
      "step": 303
    },
    {
      "epoch": 0.5954946131243879,
      "grad_norm": 0.09191132805686522,
      "learning_rate": 9.79616399719111e-06,
      "loss": 0.4766,
      "num_tokens": 288542116.0,
      "step": 304
    },
    {
      "epoch": 0.5974534769833496,
      "grad_norm": 0.1017523258754097,
      "learning_rate": 9.794368642288741e-06,
      "loss": 0.4523,
      "num_tokens": 289439264.0,
      "step": 305
    },
    {
      "epoch": 0.5994123408423114,
      "grad_norm": 0.09229643659596272,
      "learning_rate": 9.7925655813952e-06,
      "loss": 0.4422,
      "num_tokens": 290388706.0,
      "step": 306
    },
    {
      "epoch": 0.6013712047012733,
      "grad_norm": 0.09180609004222215,
      "learning_rate": 9.790754817408541e-06,
      "loss": 0.453,
      "num_tokens": 291356017.0,
      "step": 307
    },
    {
      "epoch": 0.6033300685602351,
      "grad_norm": 0.08948438390689484,
      "learning_rate": 9.788936353239211e-06,
      "loss": 0.4597,
      "num_tokens": 292342082.0,
      "step": 308
    },
    {
      "epoch": 0.6052889324191969,
      "grad_norm": 0.0884597685046138,
      "learning_rate": 9.787110191810027e-06,
      "loss": 0.4454,
      "num_tokens": 293275934.0,
      "step": 309
    },
    {
      "epoch": 0.6072477962781586,
      "grad_norm": 0.09238464655217382,
      "learning_rate": 9.78527633605618e-06,
      "loss": 0.4518,
      "num_tokens": 294228528.0,
      "step": 310
    },
    {
      "epoch": 0.6092066601371204,
      "grad_norm": 0.08783977724736827,
      "learning_rate": 9.783434788925227e-06,
      "loss": 0.4489,
      "num_tokens": 295190320.0,
      "step": 311
    },
    {
      "epoch": 0.6111655239960823,
      "grad_norm": 0.09515052060454643,
      "learning_rate": 9.781585553377086e-06,
      "loss": 0.4505,
      "num_tokens": 296154943.0,
      "step": 312
    },
    {
      "epoch": 0.6131243878550441,
      "grad_norm": 0.09104776922006945,
      "learning_rate": 9.779728632384035e-06,
      "loss": 0.4464,
      "num_tokens": 297092062.0,
      "step": 313
    },
    {
      "epoch": 0.6150832517140059,
      "grad_norm": 0.08777636231747861,
      "learning_rate": 9.777864028930705e-06,
      "loss": 0.4512,
      "num_tokens": 298062069.0,
      "step": 314
    },
    {
      "epoch": 0.6170421155729677,
      "grad_norm": 0.09365530562426334,
      "learning_rate": 9.775991746014073e-06,
      "loss": 0.4356,
      "num_tokens": 298979969.0,
      "step": 315
    },
    {
      "epoch": 0.6190009794319294,
      "grad_norm": 0.09484238379783078,
      "learning_rate": 9.77411178664346e-06,
      "loss": 0.4628,
      "num_tokens": 299928548.0,
      "step": 316
    },
    {
      "epoch": 0.6209598432908913,
      "grad_norm": 0.0923923755037889,
      "learning_rate": 9.772224153840528e-06,
      "loss": 0.4427,
      "num_tokens": 300875247.0,
      "step": 317
    },
    {
      "epoch": 0.6229187071498531,
      "grad_norm": 0.09215877415929054,
      "learning_rate": 9.770328850639268e-06,
      "loss": 0.441,
      "num_tokens": 301809712.0,
      "step": 318
    },
    {
      "epoch": 0.6248775710088149,
      "grad_norm": 0.08874374870559448,
      "learning_rate": 9.768425880086002e-06,
      "loss": 0.4455,
      "num_tokens": 302773650.0,
      "step": 319
    },
    {
      "epoch": 0.6268364348677767,
      "grad_norm": 0.090207249747187,
      "learning_rate": 9.766515245239377e-06,
      "loss": 0.4444,
      "num_tokens": 303722502.0,
      "step": 320
    },
    {
      "epoch": 0.6287952987267384,
      "grad_norm": 0.08503990355250717,
      "learning_rate": 9.764596949170356e-06,
      "loss": 0.4338,
      "num_tokens": 304706080.0,
      "step": 321
    },
    {
      "epoch": 0.6307541625857003,
      "grad_norm": 0.09025974958469772,
      "learning_rate": 9.762670994962215e-06,
      "loss": 0.4404,
      "num_tokens": 305650384.0,
      "step": 322
    },
    {
      "epoch": 0.6327130264446621,
      "grad_norm": 0.08638379902993888,
      "learning_rate": 9.760737385710546e-06,
      "loss": 0.438,
      "num_tokens": 306635856.0,
      "step": 323
    },
    {
      "epoch": 0.6346718903036239,
      "grad_norm": 0.08985329866314863,
      "learning_rate": 9.758796124523238e-06,
      "loss": 0.4458,
      "num_tokens": 307625254.0,
      "step": 324
    },
    {
      "epoch": 0.6366307541625857,
      "grad_norm": 0.09014705464588635,
      "learning_rate": 9.756847214520482e-06,
      "loss": 0.4522,
      "num_tokens": 308593837.0,
      "step": 325
    },
    {
      "epoch": 0.6385896180215475,
      "grad_norm": 0.09122915265370712,
      "learning_rate": 9.754890658834759e-06,
      "loss": 0.4434,
      "num_tokens": 309554650.0,
      "step": 326
    },
    {
      "epoch": 0.6405484818805093,
      "grad_norm": 0.08461284621938435,
      "learning_rate": 9.752926460610846e-06,
      "loss": 0.4386,
      "num_tokens": 310526204.0,
      "step": 327
    },
    {
      "epoch": 0.6425073457394711,
      "grad_norm": 0.08801795933423717,
      "learning_rate": 9.750954623005795e-06,
      "loss": 0.4608,
      "num_tokens": 311492787.0,
      "step": 328
    },
    {
      "epoch": 0.6444662095984329,
      "grad_norm": 0.0901115841850163,
      "learning_rate": 9.748975149188946e-06,
      "loss": 0.4603,
      "num_tokens": 312441688.0,
      "step": 329
    },
    {
      "epoch": 0.6464250734573947,
      "grad_norm": 0.08901845538892801,
      "learning_rate": 9.746988042341907e-06,
      "loss": 0.4529,
      "num_tokens": 313376394.0,
      "step": 330
    },
    {
      "epoch": 0.6483839373163565,
      "grad_norm": 0.0849474795941554,
      "learning_rate": 9.744993305658556e-06,
      "loss": 0.4346,
      "num_tokens": 314315578.0,
      "step": 331
    },
    {
      "epoch": 0.6503428011753183,
      "grad_norm": 0.0860787213869078,
      "learning_rate": 9.742990942345032e-06,
      "loss": 0.4547,
      "num_tokens": 315273796.0,
      "step": 332
    },
    {
      "epoch": 0.6523016650342801,
      "grad_norm": 0.08995415547258594,
      "learning_rate": 9.74098095561974e-06,
      "loss": 0.4513,
      "num_tokens": 316218943.0,
      "step": 333
    },
    {
      "epoch": 0.6542605288932419,
      "grad_norm": 0.09049418205944079,
      "learning_rate": 9.738963348713324e-06,
      "loss": 0.4356,
      "num_tokens": 317141320.0,
      "step": 334
    },
    {
      "epoch": 0.6562193927522038,
      "grad_norm": 0.09074946806627159,
      "learning_rate": 9.736938124868693e-06,
      "loss": 0.4474,
      "num_tokens": 318074977.0,
      "step": 335
    },
    {
      "epoch": 0.6581782566111655,
      "grad_norm": 0.09407292204888344,
      "learning_rate": 9.734905287340985e-06,
      "loss": 0.4617,
      "num_tokens": 319036215.0,
      "step": 336
    },
    {
      "epoch": 0.6601371204701273,
      "grad_norm": 0.09482236087266514,
      "learning_rate": 9.732864839397585e-06,
      "loss": 0.4313,
      "num_tokens": 319961419.0,
      "step": 337
    },
    {
      "epoch": 0.6620959843290891,
      "grad_norm": 0.08750525545608744,
      "learning_rate": 9.730816784318103e-06,
      "loss": 0.4385,
      "num_tokens": 320887567.0,
      "step": 338
    },
    {
      "epoch": 0.6640548481880509,
      "grad_norm": 0.09008099656070187,
      "learning_rate": 9.728761125394379e-06,
      "loss": 0.4453,
      "num_tokens": 321858337.0,
      "step": 339
    },
    {
      "epoch": 0.6660137120470128,
      "grad_norm": 0.0923712313256544,
      "learning_rate": 9.726697865930477e-06,
      "loss": 0.4468,
      "num_tokens": 322814978.0,
      "step": 340
    },
    {
      "epoch": 0.6679725759059746,
      "grad_norm": 0.08994049077738044,
      "learning_rate": 9.724627009242673e-06,
      "loss": 0.4477,
      "num_tokens": 323770176.0,
      "step": 341
    },
    {
      "epoch": 0.6699314397649363,
      "grad_norm": 0.08719580060077986,
      "learning_rate": 9.722548558659457e-06,
      "loss": 0.4355,
      "num_tokens": 324699181.0,
      "step": 342
    },
    {
      "epoch": 0.6718903036238981,
      "grad_norm": 0.08997896557435177,
      "learning_rate": 9.720462517521522e-06,
      "loss": 0.4407,
      "num_tokens": 325636045.0,
      "step": 343
    },
    {
      "epoch": 0.67384916748286,
      "grad_norm": 0.09408528451076777,
      "learning_rate": 9.718368889181763e-06,
      "loss": 0.4559,
      "num_tokens": 326598812.0,
      "step": 344
    },
    {
      "epoch": 0.6758080313418218,
      "grad_norm": 0.08601851635722685,
      "learning_rate": 9.716267677005273e-06,
      "loss": 0.435,
      "num_tokens": 327527276.0,
      "step": 345
    },
    {
      "epoch": 0.6777668952007836,
      "grad_norm": 0.08934376371860722,
      "learning_rate": 9.71415888436933e-06,
      "loss": 0.4614,
      "num_tokens": 328481063.0,
      "step": 346
    },
    {
      "epoch": 0.6797257590597453,
      "grad_norm": 0.09134788028869219,
      "learning_rate": 9.712042514663396e-06,
      "loss": 0.4563,
      "num_tokens": 329408909.0,
      "step": 347
    },
    {
      "epoch": 0.6816846229187071,
      "grad_norm": 0.09070727995108552,
      "learning_rate": 9.709918571289114e-06,
      "loss": 0.4499,
      "num_tokens": 330338376.0,
      "step": 348
    },
    {
      "epoch": 0.683643486777669,
      "grad_norm": 0.09732961017889742,
      "learning_rate": 9.7077870576603e-06,
      "loss": 0.4505,
      "num_tokens": 331305055.0,
      "step": 349
    },
    {
      "epoch": 0.6856023506366308,
      "grad_norm": 0.084475019256982,
      "learning_rate": 9.705647977202937e-06,
      "loss": 0.4342,
      "num_tokens": 332259264.0,
      "step": 350
    },
    {
      "epoch": 0.6875612144955926,
      "grad_norm": 0.08864252586488931,
      "learning_rate": 9.703501333355167e-06,
      "loss": 0.446,
      "num_tokens": 333210070.0,
      "step": 351
    },
    {
      "epoch": 0.6895200783545543,
      "grad_norm": 0.08872160192459715,
      "learning_rate": 9.701347129567296e-06,
      "loss": 0.4484,
      "num_tokens": 334175126.0,
      "step": 352
    },
    {
      "epoch": 0.6914789422135161,
      "grad_norm": 0.0930468611145152,
      "learning_rate": 9.699185369301773e-06,
      "loss": 0.4418,
      "num_tokens": 335155541.0,
      "step": 353
    },
    {
      "epoch": 0.693437806072478,
      "grad_norm": 0.09593356596707556,
      "learning_rate": 9.697016056033202e-06,
      "loss": 0.448,
      "num_tokens": 336093931.0,
      "step": 354
    },
    {
      "epoch": 0.6953966699314398,
      "grad_norm": 0.09139549920653682,
      "learning_rate": 9.694839193248315e-06,
      "loss": 0.4448,
      "num_tokens": 337044870.0,
      "step": 355
    },
    {
      "epoch": 0.6973555337904016,
      "grad_norm": 0.08923685382483237,
      "learning_rate": 9.692654784445988e-06,
      "loss": 0.448,
      "num_tokens": 337957570.0,
      "step": 356
    },
    {
      "epoch": 0.6993143976493633,
      "grad_norm": 0.09319457465664423,
      "learning_rate": 9.690462833137222e-06,
      "loss": 0.4372,
      "num_tokens": 338915262.0,
      "step": 357
    },
    {
      "epoch": 0.7012732615083251,
      "grad_norm": 0.09144500484833851,
      "learning_rate": 9.68826334284514e-06,
      "loss": 0.4353,
      "num_tokens": 339840823.0,
      "step": 358
    },
    {
      "epoch": 0.703232125367287,
      "grad_norm": 0.09068135189565031,
      "learning_rate": 9.686056317104986e-06,
      "loss": 0.4342,
      "num_tokens": 340787790.0,
      "step": 359
    },
    {
      "epoch": 0.7051909892262488,
      "grad_norm": 0.0897309059628541,
      "learning_rate": 9.683841759464114e-06,
      "loss": 0.4382,
      "num_tokens": 341705934.0,
      "step": 360
    },
    {
      "epoch": 0.7071498530852106,
      "grad_norm": 0.09186132785983991,
      "learning_rate": 9.68161967348198e-06,
      "loss": 0.4737,
      "num_tokens": 342661299.0,
      "step": 361
    },
    {
      "epoch": 0.7091087169441724,
      "grad_norm": 0.0926437461265092,
      "learning_rate": 9.679390062730147e-06,
      "loss": 0.4419,
      "num_tokens": 343595015.0,
      "step": 362
    },
    {
      "epoch": 0.7110675808031341,
      "grad_norm": 0.09275070419172855,
      "learning_rate": 9.677152930792271e-06,
      "loss": 0.4616,
      "num_tokens": 344524727.0,
      "step": 363
    },
    {
      "epoch": 0.713026444662096,
      "grad_norm": 0.08591248310796139,
      "learning_rate": 9.67490828126409e-06,
      "loss": 0.4498,
      "num_tokens": 345491807.0,
      "step": 364
    },
    {
      "epoch": 0.7149853085210578,
      "grad_norm": 0.09257358320354882,
      "learning_rate": 9.672656117753435e-06,
      "loss": 0.4483,
      "num_tokens": 346433272.0,
      "step": 365
    },
    {
      "epoch": 0.7169441723800196,
      "grad_norm": 0.0924765321124182,
      "learning_rate": 9.670396443880208e-06,
      "loss": 0.4459,
      "num_tokens": 347388710.0,
      "step": 366
    },
    {
      "epoch": 0.7189030362389814,
      "grad_norm": 0.08878238593226352,
      "learning_rate": 9.668129263276384e-06,
      "loss": 0.4534,
      "num_tokens": 348339683.0,
      "step": 367
    },
    {
      "epoch": 0.7208619000979432,
      "grad_norm": 0.09381300424385022,
      "learning_rate": 9.665854579586003e-06,
      "loss": 0.4446,
      "num_tokens": 349282115.0,
      "step": 368
    },
    {
      "epoch": 0.722820763956905,
      "grad_norm": 0.08907529279081705,
      "learning_rate": 9.663572396465165e-06,
      "loss": 0.4352,
      "num_tokens": 350225227.0,
      "step": 369
    },
    {
      "epoch": 0.7247796278158668,
      "grad_norm": 0.08936131212725164,
      "learning_rate": 9.661282717582028e-06,
      "loss": 0.461,
      "num_tokens": 351156796.0,
      "step": 370
    },
    {
      "epoch": 0.7267384916748286,
      "grad_norm": 0.09185757621293171,
      "learning_rate": 9.658985546616787e-06,
      "loss": 0.4448,
      "num_tokens": 352096120.0,
      "step": 371
    },
    {
      "epoch": 0.7286973555337904,
      "grad_norm": 0.09153254942213147,
      "learning_rate": 9.656680887261693e-06,
      "loss": 0.4416,
      "num_tokens": 353038771.0,
      "step": 372
    },
    {
      "epoch": 0.7306562193927522,
      "grad_norm": 0.09460985738793483,
      "learning_rate": 9.654368743221022e-06,
      "loss": 0.4386,
      "num_tokens": 353961679.0,
      "step": 373
    },
    {
      "epoch": 0.732615083251714,
      "grad_norm": 0.09009951630424058,
      "learning_rate": 9.652049118211085e-06,
      "loss": 0.4446,
      "num_tokens": 354922684.0,
      "step": 374
    },
    {
      "epoch": 0.7345739471106758,
      "grad_norm": 0.09185998933597071,
      "learning_rate": 9.649722015960219e-06,
      "loss": 0.4352,
      "num_tokens": 355865391.0,
      "step": 375
    },
    {
      "epoch": 0.7365328109696376,
      "grad_norm": 0.08574006743953584,
      "learning_rate": 9.647387440208772e-06,
      "loss": 0.4265,
      "num_tokens": 356807318.0,
      "step": 376
    },
    {
      "epoch": 0.7384916748285995,
      "grad_norm": 0.09213513041723538,
      "learning_rate": 9.645045394709113e-06,
      "loss": 0.4519,
      "num_tokens": 357756208.0,
      "step": 377
    },
    {
      "epoch": 0.7404505386875612,
      "grad_norm": 0.0914118115135455,
      "learning_rate": 9.64269588322561e-06,
      "loss": 0.453,
      "num_tokens": 358690434.0,
      "step": 378
    },
    {
      "epoch": 0.742409402546523,
      "grad_norm": 0.08515546604270077,
      "learning_rate": 9.640338909534636e-06,
      "loss": 0.4338,
      "num_tokens": 359600815.0,
      "step": 379
    },
    {
      "epoch": 0.7443682664054848,
      "grad_norm": 0.09099837438538474,
      "learning_rate": 9.637974477424556e-06,
      "loss": 0.4465,
      "num_tokens": 360546023.0,
      "step": 380
    },
    {
      "epoch": 0.7463271302644466,
      "grad_norm": 0.0916984793702679,
      "learning_rate": 9.63560259069572e-06,
      "loss": 0.4474,
      "num_tokens": 361484235.0,
      "step": 381
    },
    {
      "epoch": 0.7482859941234085,
      "grad_norm": 0.0907311847590698,
      "learning_rate": 9.633223253160468e-06,
      "loss": 0.4351,
      "num_tokens": 362430720.0,
      "step": 382
    },
    {
      "epoch": 0.7502448579823702,
      "grad_norm": 0.08487893107581852,
      "learning_rate": 9.630836468643105e-06,
      "loss": 0.4471,
      "num_tokens": 363419069.0,
      "step": 383
    },
    {
      "epoch": 0.752203721841332,
      "grad_norm": 0.08527321582104172,
      "learning_rate": 9.628442240979915e-06,
      "loss": 0.4448,
      "num_tokens": 364373467.0,
      "step": 384
    },
    {
      "epoch": 0.7541625857002938,
      "grad_norm": 0.09142315875170678,
      "learning_rate": 9.62604057401914e-06,
      "loss": 0.4359,
      "num_tokens": 365321364.0,
      "step": 385
    },
    {
      "epoch": 0.7561214495592556,
      "grad_norm": 0.09108466254026837,
      "learning_rate": 9.62363147162098e-06,
      "loss": 0.4489,
      "num_tokens": 366255055.0,
      "step": 386
    },
    {
      "epoch": 0.7580803134182175,
      "grad_norm": 0.09116160041104576,
      "learning_rate": 9.621214937657585e-06,
      "loss": 0.4473,
      "num_tokens": 367188147.0,
      "step": 387
    },
    {
      "epoch": 0.7600391772771793,
      "grad_norm": 0.09229959461153356,
      "learning_rate": 9.618790976013056e-06,
      "loss": 0.4511,
      "num_tokens": 368117102.0,
      "step": 388
    },
    {
      "epoch": 0.761998041136141,
      "grad_norm": 0.09153196002525947,
      "learning_rate": 9.616359590583424e-06,
      "loss": 0.4503,
      "num_tokens": 369092244.0,
      "step": 389
    },
    {
      "epoch": 0.7639569049951028,
      "grad_norm": 0.09012900386813807,
      "learning_rate": 9.613920785276655e-06,
      "loss": 0.4478,
      "num_tokens": 370033011.0,
      "step": 390
    },
    {
      "epoch": 0.7659157688540646,
      "grad_norm": 0.09032377835086211,
      "learning_rate": 9.611474564012646e-06,
      "loss": 0.4453,
      "num_tokens": 370970221.0,
      "step": 391
    },
    {
      "epoch": 0.7678746327130265,
      "grad_norm": 0.09174864200834358,
      "learning_rate": 9.609020930723208e-06,
      "loss": 0.4294,
      "num_tokens": 371927360.0,
      "step": 392
    },
    {
      "epoch": 0.7698334965719883,
      "grad_norm": 0.08799839349782142,
      "learning_rate": 9.606559889352065e-06,
      "loss": 0.4258,
      "num_tokens": 372880974.0,
      "step": 393
    },
    {
      "epoch": 0.77179236043095,
      "grad_norm": 0.09857544007537455,
      "learning_rate": 9.604091443854853e-06,
      "loss": 0.4495,
      "num_tokens": 373809205.0,
      "step": 394
    },
    {
      "epoch": 0.7737512242899118,
      "grad_norm": 0.0890729432210977,
      "learning_rate": 9.601615598199105e-06,
      "loss": 0.4502,
      "num_tokens": 374750745.0,
      "step": 395
    },
    {
      "epoch": 0.7757100881488737,
      "grad_norm": 0.09614982654170924,
      "learning_rate": 9.599132356364247e-06,
      "loss": 0.4502,
      "num_tokens": 375694515.0,
      "step": 396
    },
    {
      "epoch": 0.7776689520078355,
      "grad_norm": 0.09497610635385678,
      "learning_rate": 9.596641722341597e-06,
      "loss": 0.4625,
      "num_tokens": 376648630.0,
      "step": 397
    },
    {
      "epoch": 0.7796278158667973,
      "grad_norm": 0.08869489109692409,
      "learning_rate": 9.594143700134354e-06,
      "loss": 0.4422,
      "num_tokens": 377601707.0,
      "step": 398
    },
    {
      "epoch": 0.781586679725759,
      "grad_norm": 0.08748225330410787,
      "learning_rate": 9.591638293757584e-06,
      "loss": 0.4422,
      "num_tokens": 378528785.0,
      "step": 399
    },
    {
      "epoch": 0.7835455435847208,
      "grad_norm": 0.09171552222842465,
      "learning_rate": 9.589125507238234e-06,
      "loss": 0.4377,
      "num_tokens": 379465500.0,
      "step": 400
    },
    {
      "epoch": 0.7855044074436827,
      "grad_norm": 0.0924306305287081,
      "learning_rate": 9.586605344615105e-06,
      "loss": 0.4476,
      "num_tokens": 380436852.0,
      "step": 401
    },
    {
      "epoch": 0.7874632713026445,
      "grad_norm": 0.09194022766426356,
      "learning_rate": 9.584077809938856e-06,
      "loss": 0.4619,
      "num_tokens": 381419522.0,
      "step": 402
    },
    {
      "epoch": 0.7894221351616063,
      "grad_norm": 0.0953069976595774,
      "learning_rate": 9.581542907271994e-06,
      "loss": 0.4475,
      "num_tokens": 382373518.0,
      "step": 403
    },
    {
      "epoch": 0.791380999020568,
      "grad_norm": 0.09219210135779037,
      "learning_rate": 9.57900064068887e-06,
      "loss": 0.4469,
      "num_tokens": 383314714.0,
      "step": 404
    },
    {
      "epoch": 0.7933398628795298,
      "grad_norm": 0.08819815778989788,
      "learning_rate": 9.57645101427567e-06,
      "loss": 0.4405,
      "num_tokens": 384296808.0,
      "step": 405
    },
    {
      "epoch": 0.7952987267384917,
      "grad_norm": 0.09152627082774994,
      "learning_rate": 9.573894032130411e-06,
      "loss": 0.447,
      "num_tokens": 385238202.0,
      "step": 406
    },
    {
      "epoch": 0.7972575905974535,
      "grad_norm": 0.09332193664312469,
      "learning_rate": 9.571329698362931e-06,
      "loss": 0.4421,
      "num_tokens": 386206872.0,
      "step": 407
    },
    {
      "epoch": 0.7992164544564153,
      "grad_norm": 0.09435719314839941,
      "learning_rate": 9.568758017094884e-06,
      "loss": 0.4515,
      "num_tokens": 387163411.0,
      "step": 408
    },
    {
      "epoch": 0.801175318315377,
      "grad_norm": 0.09062165292179064,
      "learning_rate": 9.566178992459736e-06,
      "loss": 0.4581,
      "num_tokens": 388121654.0,
      "step": 409
    },
    {
      "epoch": 0.8031341821743389,
      "grad_norm": 0.09249823612388672,
      "learning_rate": 9.563592628602751e-06,
      "loss": 0.4378,
      "num_tokens": 389069363.0,
      "step": 410
    },
    {
      "epoch": 0.8050930460333007,
      "grad_norm": 0.09671609347878848,
      "learning_rate": 9.560998929681e-06,
      "loss": 0.4403,
      "num_tokens": 390021562.0,
      "step": 411
    },
    {
      "epoch": 0.8070519098922625,
      "grad_norm": 0.0903169040894827,
      "learning_rate": 9.55839789986333e-06,
      "loss": 0.4412,
      "num_tokens": 390968170.0,
      "step": 412
    },
    {
      "epoch": 0.8090107737512243,
      "grad_norm": 0.09307469451515846,
      "learning_rate": 9.555789543330383e-06,
      "loss": 0.4454,
      "num_tokens": 391927899.0,
      "step": 413
    },
    {
      "epoch": 0.8109696376101861,
      "grad_norm": 0.09144767479348721,
      "learning_rate": 9.553173864274567e-06,
      "loss": 0.443,
      "num_tokens": 392851142.0,
      "step": 414
    },
    {
      "epoch": 0.8129285014691479,
      "grad_norm": 0.09277036645308719,
      "learning_rate": 9.550550866900066e-06,
      "loss": 0.4479,
      "num_tokens": 393839937.0,
      "step": 415
    },
    {
      "epoch": 0.8148873653281097,
      "grad_norm": 0.09018958429415018,
      "learning_rate": 9.547920555422826e-06,
      "loss": 0.4363,
      "num_tokens": 394796096.0,
      "step": 416
    },
    {
      "epoch": 0.8168462291870715,
      "grad_norm": 0.09085236470619103,
      "learning_rate": 9.545282934070549e-06,
      "loss": 0.4483,
      "num_tokens": 395713544.0,
      "step": 417
    },
    {
      "epoch": 0.8188050930460333,
      "grad_norm": 0.08407649581832123,
      "learning_rate": 9.542638007082679e-06,
      "loss": 0.4391,
      "num_tokens": 396689825.0,
      "step": 418
    },
    {
      "epoch": 0.8207639569049952,
      "grad_norm": 0.09950920573968482,
      "learning_rate": 9.539985778710417e-06,
      "loss": 0.4534,
      "num_tokens": 397652133.0,
      "step": 419
    },
    {
      "epoch": 0.8227228207639569,
      "grad_norm": 0.09307640606019055,
      "learning_rate": 9.537326253216685e-06,
      "loss": 0.4423,
      "num_tokens": 398608999.0,
      "step": 420
    },
    {
      "epoch": 0.8246816846229187,
      "grad_norm": 0.09357630630578798,
      "learning_rate": 9.53465943487614e-06,
      "loss": 0.4524,
      "num_tokens": 399553537.0,
      "step": 421
    },
    {
      "epoch": 0.8266405484818805,
      "grad_norm": 0.09116527434719511,
      "learning_rate": 9.531985327975166e-06,
      "loss": 0.4318,
      "num_tokens": 400478774.0,
      "step": 422
    },
    {
      "epoch": 0.8285994123408423,
      "grad_norm": 0.0877207115367518,
      "learning_rate": 9.529303936811848e-06,
      "loss": 0.4458,
      "num_tokens": 401453462.0,
      "step": 423
    },
    {
      "epoch": 0.8305582761998042,
      "grad_norm": 0.10429597192469464,
      "learning_rate": 9.526615265695996e-06,
      "loss": 0.4588,
      "num_tokens": 402371978.0,
      "step": 424
    },
    {
      "epoch": 0.8325171400587659,
      "grad_norm": 0.0921247742066358,
      "learning_rate": 9.523919318949107e-06,
      "loss": 0.4431,
      "num_tokens": 403299490.0,
      "step": 425
    },
    {
      "epoch": 0.8344760039177277,
      "grad_norm": 0.08535437886957582,
      "learning_rate": 9.521216100904379e-06,
      "loss": 0.4504,
      "num_tokens": 404272460.0,
      "step": 426
    },
    {
      "epoch": 0.8364348677766895,
      "grad_norm": 0.09086008435179309,
      "learning_rate": 9.518505615906695e-06,
      "loss": 0.4364,
      "num_tokens": 405227719.0,
      "step": 427
    },
    {
      "epoch": 0.8383937316356513,
      "grad_norm": 0.08852926690081821,
      "learning_rate": 9.51578786831262e-06,
      "loss": 0.4333,
      "num_tokens": 406172073.0,
      "step": 428
    },
    {
      "epoch": 0.8403525954946132,
      "grad_norm": 0.09296742826394402,
      "learning_rate": 9.51306286249039e-06,
      "loss": 0.4435,
      "num_tokens": 407135296.0,
      "step": 429
    },
    {
      "epoch": 0.8423114593535749,
      "grad_norm": 0.09195865414003453,
      "learning_rate": 9.510330602819908e-06,
      "loss": 0.4338,
      "num_tokens": 408025872.0,
      "step": 430
    },
    {
      "epoch": 0.8442703232125367,
      "grad_norm": 0.08592651925000429,
      "learning_rate": 9.50759109369274e-06,
      "loss": 0.4405,
      "num_tokens": 408952437.0,
      "step": 431
    },
    {
      "epoch": 0.8462291870714985,
      "grad_norm": 0.08720910332233216,
      "learning_rate": 9.504844339512096e-06,
      "loss": 0.4618,
      "num_tokens": 409927896.0,
      "step": 432
    },
    {
      "epoch": 0.8481880509304603,
      "grad_norm": 0.09237479059899237,
      "learning_rate": 9.502090344692839e-06,
      "loss": 0.444,
      "num_tokens": 410872172.0,
      "step": 433
    },
    {
      "epoch": 0.8501469147894222,
      "grad_norm": 0.08611080491290468,
      "learning_rate": 9.499329113661462e-06,
      "loss": 0.4522,
      "num_tokens": 411842501.0,
      "step": 434
    },
    {
      "epoch": 0.8521057786483839,
      "grad_norm": 0.08974433119079651,
      "learning_rate": 9.496560650856097e-06,
      "loss": 0.4389,
      "num_tokens": 412802958.0,
      "step": 435
    },
    {
      "epoch": 0.8540646425073457,
      "grad_norm": 0.09282486852664872,
      "learning_rate": 9.493784960726494e-06,
      "loss": 0.443,
      "num_tokens": 413703549.0,
      "step": 436
    },
    {
      "epoch": 0.8560235063663075,
      "grad_norm": 0.09598607127715932,
      "learning_rate": 9.491002047734022e-06,
      "loss": 0.4448,
      "num_tokens": 414662771.0,
      "step": 437
    },
    {
      "epoch": 0.8579823702252694,
      "grad_norm": 0.0868526618332538,
      "learning_rate": 9.488211916351656e-06,
      "loss": 0.4522,
      "num_tokens": 415605121.0,
      "step": 438
    },
    {
      "epoch": 0.8599412340842312,
      "grad_norm": 0.08726492885006254,
      "learning_rate": 9.485414571063978e-06,
      "loss": 0.4413,
      "num_tokens": 416533609.0,
      "step": 439
    },
    {
      "epoch": 0.861900097943193,
      "grad_norm": 0.08893599449523941,
      "learning_rate": 9.482610016367162e-06,
      "loss": 0.458,
      "num_tokens": 417475629.0,
      "step": 440
    },
    {
      "epoch": 0.8638589618021547,
      "grad_norm": 0.08765700775405233,
      "learning_rate": 9.479798256768971e-06,
      "loss": 0.4465,
      "num_tokens": 418417455.0,
      "step": 441
    },
    {
      "epoch": 0.8658178256611165,
      "grad_norm": 0.09834057321360593,
      "learning_rate": 9.476979296788746e-06,
      "loss": 0.4373,
      "num_tokens": 419330611.0,
      "step": 442
    },
    {
      "epoch": 0.8677766895200784,
      "grad_norm": 0.09680606791405261,
      "learning_rate": 9.474153140957404e-06,
      "loss": 0.436,
      "num_tokens": 420290222.0,
      "step": 443
    },
    {
      "epoch": 0.8697355533790402,
      "grad_norm": 0.09250835079848582,
      "learning_rate": 9.471319793817427e-06,
      "loss": 0.4434,
      "num_tokens": 421235105.0,
      "step": 444
    },
    {
      "epoch": 0.871694417238002,
      "grad_norm": 0.08946786551563851,
      "learning_rate": 9.468479259922853e-06,
      "loss": 0.4477,
      "num_tokens": 422178339.0,
      "step": 445
    },
    {
      "epoch": 0.8736532810969637,
      "grad_norm": 0.09808169538863153,
      "learning_rate": 9.465631543839275e-06,
      "loss": 0.4365,
      "num_tokens": 423122510.0,
      "step": 446
    },
    {
      "epoch": 0.8756121449559255,
      "grad_norm": 0.10179959020277803,
      "learning_rate": 9.46277665014383e-06,
      "loss": 0.4658,
      "num_tokens": 424026386.0,
      "step": 447
    },
    {
      "epoch": 0.8775710088148874,
      "grad_norm": 0.09532360921540196,
      "learning_rate": 9.45991458342519e-06,
      "loss": 0.4615,
      "num_tokens": 425002226.0,
      "step": 448
    },
    {
      "epoch": 0.8795298726738492,
      "grad_norm": 0.10363014085944199,
      "learning_rate": 9.457045348283552e-06,
      "loss": 0.4442,
      "num_tokens": 425954442.0,
      "step": 449
    },
    {
      "epoch": 0.881488736532811,
      "grad_norm": 0.09131988082293717,
      "learning_rate": 9.454168949330644e-06,
      "loss": 0.4454,
      "num_tokens": 426882944.0,
      "step": 450
    },
    {
      "epoch": 0.8834476003917727,
      "grad_norm": 0.09361637101900908,
      "learning_rate": 9.451285391189701e-06,
      "loss": 0.4392,
      "num_tokens": 427859282.0,
      "step": 451
    },
    {
      "epoch": 0.8854064642507345,
      "grad_norm": 0.09877788574597322,
      "learning_rate": 9.448394678495469e-06,
      "loss": 0.4506,
      "num_tokens": 428809528.0,
      "step": 452
    },
    {
      "epoch": 0.8873653281096964,
      "grad_norm": 0.09549308711562945,
      "learning_rate": 9.445496815894192e-06,
      "loss": 0.4506,
      "num_tokens": 429768305.0,
      "step": 453
    },
    {
      "epoch": 0.8893241919686582,
      "grad_norm": 0.08871508739494441,
      "learning_rate": 9.442591808043604e-06,
      "loss": 0.4495,
      "num_tokens": 430729678.0,
      "step": 454
    },
    {
      "epoch": 0.89128305582762,
      "grad_norm": 0.0892212119857271,
      "learning_rate": 9.439679659612926e-06,
      "loss": 0.4267,
      "num_tokens": 431714983.0,
      "step": 455
    },
    {
      "epoch": 0.8932419196865817,
      "grad_norm": 0.09491973188699437,
      "learning_rate": 9.436760375282858e-06,
      "loss": 0.4422,
      "num_tokens": 432699169.0,
      "step": 456
    },
    {
      "epoch": 0.8952007835455436,
      "grad_norm": 0.09226562786893142,
      "learning_rate": 9.433833959745566e-06,
      "loss": 0.4502,
      "num_tokens": 433675703.0,
      "step": 457
    },
    {
      "epoch": 0.8971596474045054,
      "grad_norm": 0.10131891049086757,
      "learning_rate": 9.430900417704679e-06,
      "loss": 0.4572,
      "num_tokens": 434585297.0,
      "step": 458
    },
    {
      "epoch": 0.8991185112634672,
      "grad_norm": 0.09282059855847394,
      "learning_rate": 9.42795975387528e-06,
      "loss": 0.4597,
      "num_tokens": 435508190.0,
      "step": 459
    },
    {
      "epoch": 0.901077375122429,
      "grad_norm": 0.09559706065058528,
      "learning_rate": 9.4250119729839e-06,
      "loss": 0.4591,
      "num_tokens": 436464412.0,
      "step": 460
    },
    {
      "epoch": 0.9030362389813908,
      "grad_norm": 0.08935181573765572,
      "learning_rate": 9.422057079768508e-06,
      "loss": 0.4531,
      "num_tokens": 437430752.0,
      "step": 461
    },
    {
      "epoch": 0.9049951028403526,
      "grad_norm": 0.09115518617016825,
      "learning_rate": 9.419095078978506e-06,
      "loss": 0.4418,
      "num_tokens": 438400386.0,
      "step": 462
    },
    {
      "epoch": 0.9069539666993144,
      "grad_norm": 0.09088668038130367,
      "learning_rate": 9.416125975374722e-06,
      "loss": 0.4458,
      "num_tokens": 439320945.0,
      "step": 463
    },
    {
      "epoch": 0.9089128305582762,
      "grad_norm": 0.0877448959315513,
      "learning_rate": 9.413149773729393e-06,
      "loss": 0.4348,
      "num_tokens": 440254668.0,
      "step": 464
    },
    {
      "epoch": 0.910871694417238,
      "grad_norm": 0.09474539542626276,
      "learning_rate": 9.410166478826172e-06,
      "loss": 0.4618,
      "num_tokens": 441190447.0,
      "step": 465
    },
    {
      "epoch": 0.9128305582761999,
      "grad_norm": 0.08817808200419723,
      "learning_rate": 9.407176095460111e-06,
      "loss": 0.433,
      "num_tokens": 442141327.0,
      "step": 466
    },
    {
      "epoch": 0.9147894221351616,
      "grad_norm": 0.09427837087162373,
      "learning_rate": 9.404178628437652e-06,
      "loss": 0.4477,
      "num_tokens": 443083315.0,
      "step": 467
    },
    {
      "epoch": 0.9167482859941234,
      "grad_norm": 0.08978197754273248,
      "learning_rate": 9.40117408257663e-06,
      "loss": 0.4423,
      "num_tokens": 444004714.0,
      "step": 468
    },
    {
      "epoch": 0.9187071498530852,
      "grad_norm": 0.08991167756068928,
      "learning_rate": 9.398162462706249e-06,
      "loss": 0.4494,
      "num_tokens": 444898486.0,
      "step": 469
    },
    {
      "epoch": 0.920666013712047,
      "grad_norm": 0.08704060556036504,
      "learning_rate": 9.395143773667089e-06,
      "loss": 0.443,
      "num_tokens": 445863285.0,
      "step": 470
    },
    {
      "epoch": 0.9226248775710089,
      "grad_norm": 0.09328182815982616,
      "learning_rate": 9.392118020311088e-06,
      "loss": 0.4517,
      "num_tokens": 446830884.0,
      "step": 471
    },
    {
      "epoch": 0.9245837414299706,
      "grad_norm": 0.09946253242310275,
      "learning_rate": 9.389085207501542e-06,
      "loss": 0.433,
      "num_tokens": 447794765.0,
      "step": 472
    },
    {
      "epoch": 0.9265426052889324,
      "grad_norm": 0.08744402771416822,
      "learning_rate": 9.386045340113098e-06,
      "loss": 0.4447,
      "num_tokens": 448729864.0,
      "step": 473
    },
    {
      "epoch": 0.9285014691478942,
      "grad_norm": 0.09066553449079683,
      "learning_rate": 9.382998423031728e-06,
      "loss": 0.4412,
      "num_tokens": 449684785.0,
      "step": 474
    },
    {
      "epoch": 0.930460333006856,
      "grad_norm": 0.09078287148442435,
      "learning_rate": 9.379944461154747e-06,
      "loss": 0.4528,
      "num_tokens": 450636455.0,
      "step": 475
    },
    {
      "epoch": 0.9324191968658179,
      "grad_norm": 0.0943177774625156,
      "learning_rate": 9.37688345939079e-06,
      "loss": 0.4445,
      "num_tokens": 451543099.0,
      "step": 476
    },
    {
      "epoch": 0.9343780607247796,
      "grad_norm": 0.08755700718803858,
      "learning_rate": 9.373815422659806e-06,
      "loss": 0.4424,
      "num_tokens": 452484486.0,
      "step": 477
    },
    {
      "epoch": 0.9363369245837414,
      "grad_norm": 0.09366369233320135,
      "learning_rate": 9.370740355893054e-06,
      "loss": 0.4458,
      "num_tokens": 453400247.0,
      "step": 478
    },
    {
      "epoch": 0.9382957884427032,
      "grad_norm": 0.0894783976375864,
      "learning_rate": 9.367658264033089e-06,
      "loss": 0.4344,
      "num_tokens": 454365225.0,
      "step": 479
    },
    {
      "epoch": 0.940254652301665,
      "grad_norm": 0.09324735140123487,
      "learning_rate": 9.364569152033756e-06,
      "loss": 0.4532,
      "num_tokens": 455347718.0,
      "step": 480
    },
    {
      "epoch": 0.9422135161606269,
      "grad_norm": 0.09558477482548243,
      "learning_rate": 9.361473024860191e-06,
      "loss": 0.4503,
      "num_tokens": 456299070.0,
      "step": 481
    },
    {
      "epoch": 0.9441723800195886,
      "grad_norm": 0.08950461213093072,
      "learning_rate": 9.358369887488798e-06,
      "loss": 0.4422,
      "num_tokens": 457249785.0,
      "step": 482
    },
    {
      "epoch": 0.9461312438785504,
      "grad_norm": 0.08592069835625851,
      "learning_rate": 9.355259744907252e-06,
      "loss": 0.4533,
      "num_tokens": 458175313.0,
      "step": 483
    },
    {
      "epoch": 0.9480901077375122,
      "grad_norm": 0.09232436789338523,
      "learning_rate": 9.352142602114487e-06,
      "loss": 0.4516,
      "num_tokens": 459094336.0,
      "step": 484
    },
    {
      "epoch": 0.9500489715964741,
      "grad_norm": 0.0978701291776305,
      "learning_rate": 9.349018464120688e-06,
      "loss": 0.4434,
      "num_tokens": 460047468.0,
      "step": 485
    },
    {
      "epoch": 0.9520078354554359,
      "grad_norm": 0.08981980194266946,
      "learning_rate": 9.345887335947281e-06,
      "loss": 0.4315,
      "num_tokens": 461042755.0,
      "step": 486
    },
    {
      "epoch": 0.9539666993143977,
      "grad_norm": 0.08782893865713717,
      "learning_rate": 9.342749222626935e-06,
      "loss": 0.4355,
      "num_tokens": 461968138.0,
      "step": 487
    },
    {
      "epoch": 0.9559255631733594,
      "grad_norm": 0.09388681547459392,
      "learning_rate": 9.339604129203538e-06,
      "loss": 0.4296,
      "num_tokens": 462931977.0,
      "step": 488
    },
    {
      "epoch": 0.9578844270323212,
      "grad_norm": 0.09136708170681833,
      "learning_rate": 9.336452060732201e-06,
      "loss": 0.4553,
      "num_tokens": 463859433.0,
      "step": 489
    },
    {
      "epoch": 0.9598432908912831,
      "grad_norm": 0.08704641178999947,
      "learning_rate": 9.333293022279245e-06,
      "loss": 0.4473,
      "num_tokens": 464847912.0,
      "step": 490
    },
    {
      "epoch": 0.9618021547502449,
      "grad_norm": 0.08687383420459688,
      "learning_rate": 9.330127018922195e-06,
      "loss": 0.4588,
      "num_tokens": 465830652.0,
      "step": 491
    },
    {
      "epoch": 0.9637610186092067,
      "grad_norm": 0.09087441610941079,
      "learning_rate": 9.326954055749767e-06,
      "loss": 0.4488,
      "num_tokens": 466741968.0,
      "step": 492
    },
    {
      "epoch": 0.9657198824681684,
      "grad_norm": 0.0926523727188182,
      "learning_rate": 9.323774137861869e-06,
      "loss": 0.4372,
      "num_tokens": 467697264.0,
      "step": 493
    },
    {
      "epoch": 0.9676787463271302,
      "grad_norm": 0.09008100047290227,
      "learning_rate": 9.320587270369586e-06,
      "loss": 0.4415,
      "num_tokens": 468616073.0,
      "step": 494
    },
    {
      "epoch": 0.9696376101860921,
      "grad_norm": 0.0909613049269523,
      "learning_rate": 9.317393458395167e-06,
      "loss": 0.4486,
      "num_tokens": 469579890.0,
      "step": 495
    },
    {
      "epoch": 0.9715964740450539,
      "grad_norm": 0.08656271434525915,
      "learning_rate": 9.314192707072031e-06,
      "loss": 0.4296,
      "num_tokens": 470535384.0,
      "step": 496
    },
    {
      "epoch": 0.9735553379040157,
      "grad_norm": 0.09028048270821377,
      "learning_rate": 9.310985021544749e-06,
      "loss": 0.4505,
      "num_tokens": 471470747.0,
      "step": 497
    },
    {
      "epoch": 0.9755142017629774,
      "grad_norm": 0.08686211011620942,
      "learning_rate": 9.307770406969032e-06,
      "loss": 0.4325,
      "num_tokens": 472425879.0,
      "step": 498
    },
    {
      "epoch": 0.9774730656219393,
      "grad_norm": 0.10047507152636914,
      "learning_rate": 9.304548868511731e-06,
      "loss": 0.4386,
      "num_tokens": 473364667.0,
      "step": 499
    },
    {
      "epoch": 0.9794319294809011,
      "grad_norm": 0.09171293142795227,
      "learning_rate": 9.30132041135083e-06,
      "loss": 0.4326,
      "num_tokens": 474283129.0,
      "step": 500
    },
    {
      "epoch": 0.9813907933398629,
      "grad_norm": 0.08391550491357692,
      "learning_rate": 9.298085040675429e-06,
      "loss": 0.4317,
      "num_tokens": 475260218.0,
      "step": 501
    },
    {
      "epoch": 0.9833496571988247,
      "grad_norm": 0.09443388042449521,
      "learning_rate": 9.29484276168574e-06,
      "loss": 0.4311,
      "num_tokens": 476205061.0,
      "step": 502
    },
    {
      "epoch": 0.9853085210577864,
      "grad_norm": 0.09282268296918562,
      "learning_rate": 9.291593579593077e-06,
      "loss": 0.4311,
      "num_tokens": 477175331.0,
      "step": 503
    },
    {
      "epoch": 0.9872673849167483,
      "grad_norm": 0.09243941597551834,
      "learning_rate": 9.288337499619856e-06,
      "loss": 0.4432,
      "num_tokens": 478132730.0,
      "step": 504
    },
    {
      "epoch": 0.9892262487757101,
      "grad_norm": 0.08810022148509583,
      "learning_rate": 9.285074526999577e-06,
      "loss": 0.4351,
      "num_tokens": 479086824.0,
      "step": 505
    },
    {
      "epoch": 0.9911851126346719,
      "grad_norm": 0.08841532311346595,
      "learning_rate": 9.281804666976813e-06,
      "loss": 0.4279,
      "num_tokens": 479995722.0,
      "step": 506
    },
    {
      "epoch": 0.9931439764936337,
      "grad_norm": 0.08789829013435767,
      "learning_rate": 9.278527924807216e-06,
      "loss": 0.4458,
      "num_tokens": 480924330.0,
      "step": 507
    },
    {
      "epoch": 0.9951028403525954,
      "grad_norm": 0.09797516690458226,
      "learning_rate": 9.275244305757492e-06,
      "loss": 0.4475,
      "num_tokens": 481859349.0,
      "step": 508
    },
    {
      "epoch": 0.9970617042115573,
      "grad_norm": 0.08900989430801746,
      "learning_rate": 9.271953815105406e-06,
      "loss": 0.4461,
      "num_tokens": 482805292.0,
      "step": 509
    },
    {
      "epoch": 0.9990205680705191,
      "grad_norm": 0.09041197365212955,
      "learning_rate": 9.268656458139763e-06,
      "loss": 0.4436,
      "num_tokens": 483742048.0,
      "step": 510
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.09041197365212955,
      "learning_rate": 9.265352240160408e-06,
      "loss": 0.4253,
      "num_tokens": 484242036.0,
      "step": 511
    },
    {
      "epoch": 1.0019588638589618,
      "grad_norm": 0.1346384417662723,
      "learning_rate": 9.262041166478215e-06,
      "loss": 0.4222,
      "num_tokens": 485179674.0,
      "step": 512
    },
    {
      "epoch": 1.0039177277179236,
      "grad_norm": 0.088029458816831,
      "learning_rate": 9.25872324241507e-06,
      "loss": 0.4338,
      "num_tokens": 486117403.0,
      "step": 513
    },
    {
      "epoch": 1.0058765915768855,
      "grad_norm": 0.08878702794859034,
      "learning_rate": 9.255398473303873e-06,
      "loss": 0.4351,
      "num_tokens": 487071335.0,
      "step": 514
    },
    {
      "epoch": 1.0078354554358473,
      "grad_norm": 0.08822281799871601,
      "learning_rate": 9.252066864488532e-06,
      "loss": 0.4269,
      "num_tokens": 488069775.0,
      "step": 515
    },
    {
      "epoch": 1.0097943192948091,
      "grad_norm": 0.08939115776709602,
      "learning_rate": 9.24872842132394e-06,
      "loss": 0.428,
      "num_tokens": 489010259.0,
      "step": 516
    },
    {
      "epoch": 1.0117531831537707,
      "grad_norm": 0.08703369305219627,
      "learning_rate": 9.245383149175981e-06,
      "loss": 0.4301,
      "num_tokens": 489975699.0,
      "step": 517
    },
    {
      "epoch": 1.0137120470127325,
      "grad_norm": 0.08623685636983926,
      "learning_rate": 9.242031053421511e-06,
      "loss": 0.4405,
      "num_tokens": 490910628.0,
      "step": 518
    },
    {
      "epoch": 1.0156709108716944,
      "grad_norm": 0.08640005446608966,
      "learning_rate": 9.238672139448354e-06,
      "loss": 0.4328,
      "num_tokens": 491864581.0,
      "step": 519
    },
    {
      "epoch": 1.0176297747306562,
      "grad_norm": 0.08964058489696397,
      "learning_rate": 9.235306412655298e-06,
      "loss": 0.4445,
      "num_tokens": 492810592.0,
      "step": 520
    },
    {
      "epoch": 1.019588638589618,
      "grad_norm": 0.09360178318670839,
      "learning_rate": 9.231933878452075e-06,
      "loss": 0.4286,
      "num_tokens": 493725212.0,
      "step": 521
    },
    {
      "epoch": 1.0215475024485798,
      "grad_norm": 0.0849877652894003,
      "learning_rate": 9.22855454225936e-06,
      "loss": 0.4328,
      "num_tokens": 494682257.0,
      "step": 522
    },
    {
      "epoch": 1.0235063663075417,
      "grad_norm": 0.08922178043950774,
      "learning_rate": 9.225168409508763e-06,
      "loss": 0.4439,
      "num_tokens": 495656406.0,
      "step": 523
    },
    {
      "epoch": 1.0254652301665035,
      "grad_norm": 0.0872230033667074,
      "learning_rate": 9.221775485642817e-06,
      "loss": 0.4383,
      "num_tokens": 496573891.0,
      "step": 524
    },
    {
      "epoch": 1.0274240940254653,
      "grad_norm": 0.08890668382286442,
      "learning_rate": 9.21837577611497e-06,
      "loss": 0.4274,
      "num_tokens": 497525699.0,
      "step": 525
    },
    {
      "epoch": 1.0293829578844271,
      "grad_norm": 0.0878795163671064,
      "learning_rate": 9.214969286389577e-06,
      "loss": 0.4207,
      "num_tokens": 498456085.0,
      "step": 526
    },
    {
      "epoch": 1.0313418217433887,
      "grad_norm": 0.08547117859073326,
      "learning_rate": 9.21155602194189e-06,
      "loss": 0.4239,
      "num_tokens": 499434510.0,
      "step": 527
    },
    {
      "epoch": 1.0333006856023506,
      "grad_norm": 0.08449758677555291,
      "learning_rate": 9.20813598825805e-06,
      "loss": 0.4312,
      "num_tokens": 500402663.0,
      "step": 528
    },
    {
      "epoch": 1.0352595494613124,
      "grad_norm": 0.08679352085531529,
      "learning_rate": 9.204709190835081e-06,
      "loss": 0.4468,
      "num_tokens": 501376158.0,
      "step": 529
    },
    {
      "epoch": 1.0372184133202742,
      "grad_norm": 0.09127157452556493,
      "learning_rate": 9.201275635180877e-06,
      "loss": 0.4282,
      "num_tokens": 502317709.0,
      "step": 530
    },
    {
      "epoch": 1.039177277179236,
      "grad_norm": 0.08423552533543917,
      "learning_rate": 9.19783532681419e-06,
      "loss": 0.4399,
      "num_tokens": 503287116.0,
      "step": 531
    },
    {
      "epoch": 1.0411361410381978,
      "grad_norm": 0.09162636138190586,
      "learning_rate": 9.194388271264634e-06,
      "loss": 0.439,
      "num_tokens": 504232930.0,
      "step": 532
    },
    {
      "epoch": 1.0430950048971597,
      "grad_norm": 0.09074860458297218,
      "learning_rate": 9.190934474072658e-06,
      "loss": 0.4355,
      "num_tokens": 505193880.0,
      "step": 533
    },
    {
      "epoch": 1.0450538687561215,
      "grad_norm": 0.09258516439738017,
      "learning_rate": 9.187473940789558e-06,
      "loss": 0.4207,
      "num_tokens": 506177865.0,
      "step": 534
    },
    {
      "epoch": 1.0470127326150833,
      "grad_norm": 0.08837416243576877,
      "learning_rate": 9.184006676977444e-06,
      "loss": 0.4287,
      "num_tokens": 507129459.0,
      "step": 535
    },
    {
      "epoch": 1.0489715964740451,
      "grad_norm": 0.08628378811997846,
      "learning_rate": 9.180532688209256e-06,
      "loss": 0.4412,
      "num_tokens": 508068980.0,
      "step": 536
    },
    {
      "epoch": 1.050930460333007,
      "grad_norm": 0.08781731978793558,
      "learning_rate": 9.177051980068738e-06,
      "loss": 0.416,
      "num_tokens": 509021271.0,
      "step": 537
    },
    {
      "epoch": 1.0528893241919686,
      "grad_norm": 0.08690062866814949,
      "learning_rate": 9.173564558150429e-06,
      "loss": 0.4339,
      "num_tokens": 509978396.0,
      "step": 538
    },
    {
      "epoch": 1.0548481880509304,
      "grad_norm": 0.08606956839244426,
      "learning_rate": 9.17007042805967e-06,
      "loss": 0.4417,
      "num_tokens": 510933733.0,
      "step": 539
    },
    {
      "epoch": 1.0568070519098922,
      "grad_norm": 0.09521025231984558,
      "learning_rate": 9.166569595412576e-06,
      "loss": 0.4288,
      "num_tokens": 511841279.0,
      "step": 540
    },
    {
      "epoch": 1.058765915768854,
      "grad_norm": 0.0887698843811598,
      "learning_rate": 9.163062065836035e-06,
      "loss": 0.4382,
      "num_tokens": 512800311.0,
      "step": 541
    },
    {
      "epoch": 1.0607247796278159,
      "grad_norm": 0.0897644692815152,
      "learning_rate": 9.159547844967703e-06,
      "loss": 0.4309,
      "num_tokens": 513718965.0,
      "step": 542
    },
    {
      "epoch": 1.0626836434867777,
      "grad_norm": 0.10022140749613061,
      "learning_rate": 9.15602693845599e-06,
      "loss": 0.4341,
      "num_tokens": 514682927.0,
      "step": 543
    },
    {
      "epoch": 1.0646425073457395,
      "grad_norm": 0.09351376434364068,
      "learning_rate": 9.15249935196005e-06,
      "loss": 0.4414,
      "num_tokens": 515617431.0,
      "step": 544
    },
    {
      "epoch": 1.0666013712047013,
      "grad_norm": 0.08904044489307142,
      "learning_rate": 9.14896509114977e-06,
      "loss": 0.4471,
      "num_tokens": 516569839.0,
      "step": 545
    },
    {
      "epoch": 1.0685602350636632,
      "grad_norm": 0.09092820781392638,
      "learning_rate": 9.145424161705777e-06,
      "loss": 0.4432,
      "num_tokens": 517510193.0,
      "step": 546
    },
    {
      "epoch": 1.070519098922625,
      "grad_norm": 0.09330544530740836,
      "learning_rate": 9.141876569319405e-06,
      "loss": 0.4369,
      "num_tokens": 518445486.0,
      "step": 547
    },
    {
      "epoch": 1.0724779627815866,
      "grad_norm": 0.09016614273965744,
      "learning_rate": 9.138322319692701e-06,
      "loss": 0.4529,
      "num_tokens": 519407293.0,
      "step": 548
    },
    {
      "epoch": 1.0744368266405484,
      "grad_norm": 0.08597232757876415,
      "learning_rate": 9.134761418538411e-06,
      "loss": 0.4305,
      "num_tokens": 520351218.0,
      "step": 549
    },
    {
      "epoch": 1.0763956904995102,
      "grad_norm": 0.08545941413061858,
      "learning_rate": 9.131193871579975e-06,
      "loss": 0.4284,
      "num_tokens": 521304087.0,
      "step": 550
    },
    {
      "epoch": 1.078354554358472,
      "grad_norm": 0.09322913844311503,
      "learning_rate": 9.127619684551514e-06,
      "loss": 0.4369,
      "num_tokens": 522234338.0,
      "step": 551
    },
    {
      "epoch": 1.0803134182174339,
      "grad_norm": 0.08621966914343301,
      "learning_rate": 9.124038863197817e-06,
      "loss": 0.4223,
      "num_tokens": 523155859.0,
      "step": 552
    },
    {
      "epoch": 1.0822722820763957,
      "grad_norm": 0.09160924475456932,
      "learning_rate": 9.120451413274346e-06,
      "loss": 0.4364,
      "num_tokens": 524088614.0,
      "step": 553
    },
    {
      "epoch": 1.0842311459353575,
      "grad_norm": 0.08458715251971163,
      "learning_rate": 9.116857340547203e-06,
      "loss": 0.439,
      "num_tokens": 525041740.0,
      "step": 554
    },
    {
      "epoch": 1.0861900097943193,
      "grad_norm": 0.0898259350766786,
      "learning_rate": 9.11325665079315e-06,
      "loss": 0.451,
      "num_tokens": 525986172.0,
      "step": 555
    },
    {
      "epoch": 1.0881488736532812,
      "grad_norm": 0.08779432275187664,
      "learning_rate": 9.109649349799575e-06,
      "loss": 0.4395,
      "num_tokens": 526927161.0,
      "step": 556
    },
    {
      "epoch": 1.090107737512243,
      "grad_norm": 0.09289699034996378,
      "learning_rate": 9.106035443364493e-06,
      "loss": 0.4369,
      "num_tokens": 527843298.0,
      "step": 557
    },
    {
      "epoch": 1.0920666013712048,
      "grad_norm": 0.09181620819782729,
      "learning_rate": 9.102414937296542e-06,
      "loss": 0.4456,
      "num_tokens": 528785650.0,
      "step": 558
    },
    {
      "epoch": 1.0940254652301664,
      "grad_norm": 0.08490441522975409,
      "learning_rate": 9.098787837414957e-06,
      "loss": 0.4356,
      "num_tokens": 529760264.0,
      "step": 559
    },
    {
      "epoch": 1.0959843290891282,
      "grad_norm": 0.09289100848996348,
      "learning_rate": 9.095154149549584e-06,
      "loss": 0.4312,
      "num_tokens": 530731237.0,
      "step": 560
    },
    {
      "epoch": 1.09794319294809,
      "grad_norm": 0.08744910222088205,
      "learning_rate": 9.091513879540845e-06,
      "loss": 0.4249,
      "num_tokens": 531697682.0,
      "step": 561
    },
    {
      "epoch": 1.0999020568070519,
      "grad_norm": 0.08763731036004777,
      "learning_rate": 9.087867033239754e-06,
      "loss": 0.4455,
      "num_tokens": 532640988.0,
      "step": 562
    },
    {
      "epoch": 1.1018609206660137,
      "grad_norm": 0.09206640614015055,
      "learning_rate": 9.084213616507887e-06,
      "loss": 0.4235,
      "num_tokens": 533610700.0,
      "step": 563
    },
    {
      "epoch": 1.1038197845249755,
      "grad_norm": 0.0864258756510009,
      "learning_rate": 9.080553635217379e-06,
      "loss": 0.4462,
      "num_tokens": 534568004.0,
      "step": 564
    },
    {
      "epoch": 1.1057786483839374,
      "grad_norm": 0.09050850449279439,
      "learning_rate": 9.076887095250924e-06,
      "loss": 0.445,
      "num_tokens": 535487804.0,
      "step": 565
    },
    {
      "epoch": 1.1077375122428992,
      "grad_norm": 0.08886233313886589,
      "learning_rate": 9.07321400250175e-06,
      "loss": 0.4304,
      "num_tokens": 536431050.0,
      "step": 566
    },
    {
      "epoch": 1.109696376101861,
      "grad_norm": 0.09119087376797808,
      "learning_rate": 9.069534362873627e-06,
      "loss": 0.4403,
      "num_tokens": 537413743.0,
      "step": 567
    },
    {
      "epoch": 1.1116552399608226,
      "grad_norm": 0.08832284757786994,
      "learning_rate": 9.065848182280835e-06,
      "loss": 0.4237,
      "num_tokens": 538358346.0,
      "step": 568
    },
    {
      "epoch": 1.1136141038197844,
      "grad_norm": 0.09035342822708291,
      "learning_rate": 9.062155466648177e-06,
      "loss": 0.4336,
      "num_tokens": 539260955.0,
      "step": 569
    },
    {
      "epoch": 1.1155729676787463,
      "grad_norm": 0.08818202236921634,
      "learning_rate": 9.058456221910956e-06,
      "loss": 0.4718,
      "num_tokens": 540224054.0,
      "step": 570
    },
    {
      "epoch": 1.117531831537708,
      "grad_norm": 0.09349504095923021,
      "learning_rate": 9.05475045401497e-06,
      "loss": 0.4431,
      "num_tokens": 541167909.0,
      "step": 571
    },
    {
      "epoch": 1.11949069539667,
      "grad_norm": 0.09360804157578115,
      "learning_rate": 9.051038168916502e-06,
      "loss": 0.4372,
      "num_tokens": 542110697.0,
      "step": 572
    },
    {
      "epoch": 1.1214495592556317,
      "grad_norm": 0.08740201831701315,
      "learning_rate": 9.04731937258231e-06,
      "loss": 0.4276,
      "num_tokens": 543078503.0,
      "step": 573
    },
    {
      "epoch": 1.1234084231145935,
      "grad_norm": 0.08990834925551848,
      "learning_rate": 9.043594070989619e-06,
      "loss": 0.4283,
      "num_tokens": 544041670.0,
      "step": 574
    },
    {
      "epoch": 1.1253672869735554,
      "grad_norm": 0.09019501205224968,
      "learning_rate": 9.039862270126102e-06,
      "loss": 0.4402,
      "num_tokens": 544955716.0,
      "step": 575
    },
    {
      "epoch": 1.1273261508325172,
      "grad_norm": 0.09007146534175918,
      "learning_rate": 9.036123975989893e-06,
      "loss": 0.4464,
      "num_tokens": 545868897.0,
      "step": 576
    },
    {
      "epoch": 1.129285014691479,
      "grad_norm": 0.09113325715101009,
      "learning_rate": 9.032379194589546e-06,
      "loss": 0.4191,
      "num_tokens": 546840812.0,
      "step": 577
    },
    {
      "epoch": 1.1312438785504408,
      "grad_norm": 0.08730973899060335,
      "learning_rate": 9.028627931944054e-06,
      "loss": 0.4407,
      "num_tokens": 547780715.0,
      "step": 578
    },
    {
      "epoch": 1.1332027424094027,
      "grad_norm": 0.09309473090760453,
      "learning_rate": 9.024870194082824e-06,
      "loss": 0.4405,
      "num_tokens": 548711835.0,
      "step": 579
    },
    {
      "epoch": 1.1351616062683643,
      "grad_norm": 0.0909014800583608,
      "learning_rate": 9.021105987045668e-06,
      "loss": 0.4311,
      "num_tokens": 549694336.0,
      "step": 580
    },
    {
      "epoch": 1.137120470127326,
      "grad_norm": 0.08925871959051611,
      "learning_rate": 9.0173353168828e-06,
      "loss": 0.4416,
      "num_tokens": 550623086.0,
      "step": 581
    },
    {
      "epoch": 1.139079333986288,
      "grad_norm": 0.08698679272649354,
      "learning_rate": 9.013558189654819e-06,
      "loss": 0.4348,
      "num_tokens": 551575073.0,
      "step": 582
    },
    {
      "epoch": 1.1410381978452497,
      "grad_norm": 0.10080409823346292,
      "learning_rate": 9.009774611432703e-06,
      "loss": 0.416,
      "num_tokens": 552504245.0,
      "step": 583
    },
    {
      "epoch": 1.1429970617042116,
      "grad_norm": 0.09186094665712399,
      "learning_rate": 9.0059845882978e-06,
      "loss": 0.4396,
      "num_tokens": 553466835.0,
      "step": 584
    },
    {
      "epoch": 1.1449559255631734,
      "grad_norm": 0.08544410038312396,
      "learning_rate": 9.002188126341815e-06,
      "loss": 0.4168,
      "num_tokens": 554393292.0,
      "step": 585
    },
    {
      "epoch": 1.1469147894221352,
      "grad_norm": 0.09036750755990369,
      "learning_rate": 8.998385231666807e-06,
      "loss": 0.4428,
      "num_tokens": 555351017.0,
      "step": 586
    },
    {
      "epoch": 1.148873653281097,
      "grad_norm": 0.0941430314187494,
      "learning_rate": 8.994575910385164e-06,
      "loss": 0.4292,
      "num_tokens": 556298669.0,
      "step": 587
    },
    {
      "epoch": 1.1508325171400589,
      "grad_norm": 0.09190677924156279,
      "learning_rate": 8.990760168619616e-06,
      "loss": 0.4354,
      "num_tokens": 557209186.0,
      "step": 588
    },
    {
      "epoch": 1.1527913809990205,
      "grad_norm": 0.09438033334326829,
      "learning_rate": 8.986938012503203e-06,
      "loss": 0.4443,
      "num_tokens": 558171451.0,
      "step": 589
    },
    {
      "epoch": 1.1547502448579823,
      "grad_norm": 0.0910359513541571,
      "learning_rate": 8.983109448179281e-06,
      "loss": 0.4448,
      "num_tokens": 559135698.0,
      "step": 590
    },
    {
      "epoch": 1.156709108716944,
      "grad_norm": 0.08989170045297915,
      "learning_rate": 8.979274481801501e-06,
      "loss": 0.4302,
      "num_tokens": 560063946.0,
      "step": 591
    },
    {
      "epoch": 1.158667972575906,
      "grad_norm": 0.09086054010786371,
      "learning_rate": 8.975433119533809e-06,
      "loss": 0.4359,
      "num_tokens": 560999098.0,
      "step": 592
    },
    {
      "epoch": 1.1606268364348677,
      "grad_norm": 0.08845012178342421,
      "learning_rate": 8.971585367550426e-06,
      "loss": 0.4431,
      "num_tokens": 561936325.0,
      "step": 593
    },
    {
      "epoch": 1.1625857002938296,
      "grad_norm": 0.08962649423988109,
      "learning_rate": 8.967731232035848e-06,
      "loss": 0.4434,
      "num_tokens": 562877802.0,
      "step": 594
    },
    {
      "epoch": 1.1645445641527914,
      "grad_norm": 0.10172094046132377,
      "learning_rate": 8.963870719184829e-06,
      "loss": 0.4327,
      "num_tokens": 563804266.0,
      "step": 595
    },
    {
      "epoch": 1.1665034280117532,
      "grad_norm": 0.09120617978253029,
      "learning_rate": 8.960003835202369e-06,
      "loss": 0.4292,
      "num_tokens": 564776814.0,
      "step": 596
    },
    {
      "epoch": 1.168462291870715,
      "grad_norm": 0.09748825916269319,
      "learning_rate": 8.956130586303718e-06,
      "loss": 0.4155,
      "num_tokens": 565708746.0,
      "step": 597
    },
    {
      "epoch": 1.1704211557296769,
      "grad_norm": 0.09515446940350762,
      "learning_rate": 8.95225097871435e-06,
      "loss": 0.4335,
      "num_tokens": 566691749.0,
      "step": 598
    },
    {
      "epoch": 1.1723800195886387,
      "grad_norm": 0.09237283816575288,
      "learning_rate": 8.948365018669956e-06,
      "loss": 0.4434,
      "num_tokens": 567602865.0,
      "step": 599
    },
    {
      "epoch": 1.1743388834476005,
      "grad_norm": 0.0851188141049216,
      "learning_rate": 8.944472712416448e-06,
      "loss": 0.4235,
      "num_tokens": 568553534.0,
      "step": 600
    },
    {
      "epoch": 1.1762977473065621,
      "grad_norm": 0.09329542673487902,
      "learning_rate": 8.940574066209925e-06,
      "loss": 0.4197,
      "num_tokens": 569475834.0,
      "step": 601
    },
    {
      "epoch": 1.178256611165524,
      "grad_norm": 0.09616537034414288,
      "learning_rate": 8.93666908631669e-06,
      "loss": 0.4382,
      "num_tokens": 570398515.0,
      "step": 602
    },
    {
      "epoch": 1.1802154750244858,
      "grad_norm": 0.08888109211301855,
      "learning_rate": 8.932757779013214e-06,
      "loss": 0.4405,
      "num_tokens": 571317760.0,
      "step": 603
    },
    {
      "epoch": 1.1821743388834476,
      "grad_norm": 0.09427241505946338,
      "learning_rate": 8.928840150586145e-06,
      "loss": 0.4414,
      "num_tokens": 572262537.0,
      "step": 604
    },
    {
      "epoch": 1.1841332027424094,
      "grad_norm": 0.08439789988449452,
      "learning_rate": 8.924916207332288e-06,
      "loss": 0.4066,
      "num_tokens": 573190293.0,
      "step": 605
    },
    {
      "epoch": 1.1860920666013712,
      "grad_norm": 0.09258147163465,
      "learning_rate": 8.9209859555586e-06,
      "loss": 0.444,
      "num_tokens": 574116845.0,
      "step": 606
    },
    {
      "epoch": 1.188050930460333,
      "grad_norm": 0.09691647341818313,
      "learning_rate": 8.917049401582178e-06,
      "loss": 0.4416,
      "num_tokens": 575043499.0,
      "step": 607
    },
    {
      "epoch": 1.1900097943192949,
      "grad_norm": 0.09873606919009019,
      "learning_rate": 8.913106551730247e-06,
      "loss": 0.4327,
      "num_tokens": 576005125.0,
      "step": 608
    },
    {
      "epoch": 1.1919686581782567,
      "grad_norm": 0.08724247174727258,
      "learning_rate": 8.90915741234015e-06,
      "loss": 0.4271,
      "num_tokens": 576979176.0,
      "step": 609
    },
    {
      "epoch": 1.1939275220372183,
      "grad_norm": 0.08491734755637224,
      "learning_rate": 8.90520198975934e-06,
      "loss": 0.4386,
      "num_tokens": 577932131.0,
      "step": 610
    },
    {
      "epoch": 1.1958863858961801,
      "grad_norm": 0.08882017914507026,
      "learning_rate": 8.901240290345372e-06,
      "loss": 0.4183,
      "num_tokens": 578869316.0,
      "step": 611
    },
    {
      "epoch": 1.197845249755142,
      "grad_norm": 0.08848950314924899,
      "learning_rate": 8.897272320465887e-06,
      "loss": 0.4374,
      "num_tokens": 579828766.0,
      "step": 612
    },
    {
      "epoch": 1.1998041136141038,
      "grad_norm": 0.09418538155264382,
      "learning_rate": 8.893298086498604e-06,
      "loss": 0.4407,
      "num_tokens": 580760358.0,
      "step": 613
    },
    {
      "epoch": 1.2017629774730656,
      "grad_norm": 0.0951586839511284,
      "learning_rate": 8.889317594831313e-06,
      "loss": 0.4559,
      "num_tokens": 581739411.0,
      "step": 614
    },
    {
      "epoch": 1.2037218413320274,
      "grad_norm": 0.08649794305940979,
      "learning_rate": 8.885330851861858e-06,
      "loss": 0.448,
      "num_tokens": 582697954.0,
      "step": 615
    },
    {
      "epoch": 1.2056807051909892,
      "grad_norm": 0.090851455587811,
      "learning_rate": 8.881337863998137e-06,
      "loss": 0.4307,
      "num_tokens": 583654639.0,
      "step": 616
    },
    {
      "epoch": 1.207639569049951,
      "grad_norm": 0.08879120337346492,
      "learning_rate": 8.877338637658074e-06,
      "loss": 0.432,
      "num_tokens": 584588030.0,
      "step": 617
    },
    {
      "epoch": 1.209598432908913,
      "grad_norm": 0.0899944248152233,
      "learning_rate": 8.873333179269635e-06,
      "loss": 0.4434,
      "num_tokens": 585533383.0,
      "step": 618
    },
    {
      "epoch": 1.2115572967678747,
      "grad_norm": 0.09846557093533091,
      "learning_rate": 8.869321495270794e-06,
      "loss": 0.4223,
      "num_tokens": 586464702.0,
      "step": 619
    },
    {
      "epoch": 1.2135161606268365,
      "grad_norm": 0.09703993219197968,
      "learning_rate": 8.865303592109528e-06,
      "loss": 0.434,
      "num_tokens": 587434572.0,
      "step": 620
    },
    {
      "epoch": 1.2154750244857984,
      "grad_norm": 0.08482944334878705,
      "learning_rate": 8.86127947624382e-06,
      "loss": 0.4272,
      "num_tokens": 588407834.0,
      "step": 621
    },
    {
      "epoch": 1.21743388834476,
      "grad_norm": 0.08877645120102982,
      "learning_rate": 8.857249154141632e-06,
      "loss": 0.429,
      "num_tokens": 589366470.0,
      "step": 622
    },
    {
      "epoch": 1.2193927522037218,
      "grad_norm": 0.09451097486733544,
      "learning_rate": 8.853212632280901e-06,
      "loss": 0.4396,
      "num_tokens": 590324534.0,
      "step": 623
    },
    {
      "epoch": 1.2213516160626836,
      "grad_norm": 0.09283507786572938,
      "learning_rate": 8.849169917149532e-06,
      "loss": 0.4316,
      "num_tokens": 591288272.0,
      "step": 624
    },
    {
      "epoch": 1.2233104799216454,
      "grad_norm": 0.09120151965940637,
      "learning_rate": 8.845121015245384e-06,
      "loss": 0.435,
      "num_tokens": 592274577.0,
      "step": 625
    },
    {
      "epoch": 1.2252693437806073,
      "grad_norm": 0.08867685030216749,
      "learning_rate": 8.841065933076258e-06,
      "loss": 0.4526,
      "num_tokens": 593219134.0,
      "step": 626
    },
    {
      "epoch": 1.227228207639569,
      "grad_norm": 0.0873309302541802,
      "learning_rate": 8.83700467715989e-06,
      "loss": 0.4442,
      "num_tokens": 594165772.0,
      "step": 627
    },
    {
      "epoch": 1.229187071498531,
      "grad_norm": 0.0841848045908713,
      "learning_rate": 8.832937254023938e-06,
      "loss": 0.4159,
      "num_tokens": 595116617.0,
      "step": 628
    },
    {
      "epoch": 1.2311459353574927,
      "grad_norm": 0.09512852122433438,
      "learning_rate": 8.828863670205973e-06,
      "loss": 0.4298,
      "num_tokens": 596063491.0,
      "step": 629
    },
    {
      "epoch": 1.2331047992164546,
      "grad_norm": 0.09200472118564294,
      "learning_rate": 8.824783932253472e-06,
      "loss": 0.4128,
      "num_tokens": 597019290.0,
      "step": 630
    },
    {
      "epoch": 1.2350636630754162,
      "grad_norm": 0.0970630323465728,
      "learning_rate": 8.820698046723796e-06,
      "loss": 0.4428,
      "num_tokens": 597975720.0,
      "step": 631
    },
    {
      "epoch": 1.237022526934378,
      "grad_norm": 0.08991662641692448,
      "learning_rate": 8.816606020184191e-06,
      "loss": 0.4265,
      "num_tokens": 598874821.0,
      "step": 632
    },
    {
      "epoch": 1.2389813907933398,
      "grad_norm": 0.09894834121305396,
      "learning_rate": 8.812507859211775e-06,
      "loss": 0.4377,
      "num_tokens": 599815897.0,
      "step": 633
    },
    {
      "epoch": 1.2409402546523016,
      "grad_norm": 0.08975554446089534,
      "learning_rate": 8.808403570393525e-06,
      "loss": 0.4297,
      "num_tokens": 600767656.0,
      "step": 634
    },
    {
      "epoch": 1.2428991185112634,
      "grad_norm": 0.08981109968851751,
      "learning_rate": 8.804293160326263e-06,
      "loss": 0.438,
      "num_tokens": 601727806.0,
      "step": 635
    },
    {
      "epoch": 1.2448579823702253,
      "grad_norm": 0.08777442501895265,
      "learning_rate": 8.800176635616658e-06,
      "loss": 0.4264,
      "num_tokens": 602666137.0,
      "step": 636
    },
    {
      "epoch": 1.246816846229187,
      "grad_norm": 0.08464531266226324,
      "learning_rate": 8.796054002881196e-06,
      "loss": 0.4485,
      "num_tokens": 603593900.0,
      "step": 637
    },
    {
      "epoch": 1.248775710088149,
      "grad_norm": 0.09454628489660366,
      "learning_rate": 8.791925268746193e-06,
      "loss": 0.4236,
      "num_tokens": 604535548.0,
      "step": 638
    },
    {
      "epoch": 1.2507345739471107,
      "grad_norm": 0.08654271881186104,
      "learning_rate": 8.787790439847762e-06,
      "loss": 0.4161,
      "num_tokens": 605535752.0,
      "step": 639
    },
    {
      "epoch": 1.2526934378060726,
      "grad_norm": 0.08524410656224386,
      "learning_rate": 8.783649522831816e-06,
      "loss": 0.4247,
      "num_tokens": 606481476.0,
      "step": 640
    },
    {
      "epoch": 1.2546523016650344,
      "grad_norm": 0.08525806913461748,
      "learning_rate": 8.779502524354054e-06,
      "loss": 0.4362,
      "num_tokens": 607411055.0,
      "step": 641
    },
    {
      "epoch": 1.2566111655239962,
      "grad_norm": 0.0936767382051949,
      "learning_rate": 8.775349451079948e-06,
      "loss": 0.4396,
      "num_tokens": 608356108.0,
      "step": 642
    },
    {
      "epoch": 1.2585700293829578,
      "grad_norm": 0.08980676844521197,
      "learning_rate": 8.771190309684737e-06,
      "loss": 0.4133,
      "num_tokens": 609301945.0,
      "step": 643
    },
    {
      "epoch": 1.2605288932419196,
      "grad_norm": 0.08618636223485227,
      "learning_rate": 8.767025106853407e-06,
      "loss": 0.4156,
      "num_tokens": 610239096.0,
      "step": 644
    },
    {
      "epoch": 1.2624877571008815,
      "grad_norm": 0.08214414891863256,
      "learning_rate": 8.762853849280692e-06,
      "loss": 0.4292,
      "num_tokens": 611199816.0,
      "step": 645
    },
    {
      "epoch": 1.2644466209598433,
      "grad_norm": 0.08708778558013325,
      "learning_rate": 8.758676543671059e-06,
      "loss": 0.4354,
      "num_tokens": 612158169.0,
      "step": 646
    },
    {
      "epoch": 1.266405484818805,
      "grad_norm": 0.08721017054043191,
      "learning_rate": 8.754493196738692e-06,
      "loss": 0.4418,
      "num_tokens": 613075173.0,
      "step": 647
    },
    {
      "epoch": 1.268364348677767,
      "grad_norm": 0.09924361764756602,
      "learning_rate": 8.750303815207487e-06,
      "loss": 0.4139,
      "num_tokens": 614043095.0,
      "step": 648
    },
    {
      "epoch": 1.2703232125367288,
      "grad_norm": 0.0898894914120729,
      "learning_rate": 8.746108405811036e-06,
      "loss": 0.4384,
      "num_tokens": 614975110.0,
      "step": 649
    },
    {
      "epoch": 1.2722820763956906,
      "grad_norm": 0.0859163297853911,
      "learning_rate": 8.741906975292626e-06,
      "loss": 0.4349,
      "num_tokens": 615918684.0,
      "step": 650
    },
    {
      "epoch": 1.2742409402546522,
      "grad_norm": 0.09237780533611917,
      "learning_rate": 8.737699530405218e-06,
      "loss": 0.4398,
      "num_tokens": 616878905.0,
      "step": 651
    },
    {
      "epoch": 1.276199804113614,
      "grad_norm": 0.0995344708616368,
      "learning_rate": 8.73348607791144e-06,
      "loss": 0.4585,
      "num_tokens": 617842176.0,
      "step": 652
    },
    {
      "epoch": 1.2781586679725758,
      "grad_norm": 0.09267000447613578,
      "learning_rate": 8.729266624583574e-06,
      "loss": 0.4254,
      "num_tokens": 618781622.0,
      "step": 653
    },
    {
      "epoch": 1.2801175318315376,
      "grad_norm": 0.09010105944541431,
      "learning_rate": 8.725041177203556e-06,
      "loss": 0.4436,
      "num_tokens": 619724561.0,
      "step": 654
    },
    {
      "epoch": 1.2820763956904995,
      "grad_norm": 0.0884000275855435,
      "learning_rate": 8.720809742562942e-06,
      "loss": 0.4429,
      "num_tokens": 620692580.0,
      "step": 655
    },
    {
      "epoch": 1.2840352595494613,
      "grad_norm": 0.08443674869288785,
      "learning_rate": 8.716572327462923e-06,
      "loss": 0.4387,
      "num_tokens": 621602721.0,
      "step": 656
    },
    {
      "epoch": 1.2859941234084231,
      "grad_norm": 0.09753336457459878,
      "learning_rate": 8.712328938714299e-06,
      "loss": 0.4283,
      "num_tokens": 622548333.0,
      "step": 657
    },
    {
      "epoch": 1.287952987267385,
      "grad_norm": 0.08516333994023409,
      "learning_rate": 8.708079583137469e-06,
      "loss": 0.4375,
      "num_tokens": 623449645.0,
      "step": 658
    },
    {
      "epoch": 1.2899118511263468,
      "grad_norm": 0.09342874946755898,
      "learning_rate": 8.703824267562424e-06,
      "loss": 0.4435,
      "num_tokens": 624421631.0,
      "step": 659
    },
    {
      "epoch": 1.2918707149853086,
      "grad_norm": 0.0929486810483253,
      "learning_rate": 8.699562998828739e-06,
      "loss": 0.437,
      "num_tokens": 625372051.0,
      "step": 660
    },
    {
      "epoch": 1.2938295788442704,
      "grad_norm": 0.08918991633816341,
      "learning_rate": 8.695295783785549e-06,
      "loss": 0.431,
      "num_tokens": 626313867.0,
      "step": 661
    },
    {
      "epoch": 1.2957884427032322,
      "grad_norm": 0.09284217963926021,
      "learning_rate": 8.691022629291553e-06,
      "loss": 0.4426,
      "num_tokens": 627298613.0,
      "step": 662
    },
    {
      "epoch": 1.297747306562194,
      "grad_norm": 0.09180686540593182,
      "learning_rate": 8.686743542214994e-06,
      "loss": 0.4237,
      "num_tokens": 628224380.0,
      "step": 663
    },
    {
      "epoch": 1.2997061704211557,
      "grad_norm": 0.08633319535636999,
      "learning_rate": 8.682458529433651e-06,
      "loss": 0.4291,
      "num_tokens": 629166811.0,
      "step": 664
    },
    {
      "epoch": 1.3016650342801175,
      "grad_norm": 0.08866719171050685,
      "learning_rate": 8.678167597834825e-06,
      "loss": 0.4191,
      "num_tokens": 630106286.0,
      "step": 665
    },
    {
      "epoch": 1.3036238981390793,
      "grad_norm": 0.0865093360650587,
      "learning_rate": 8.673870754315336e-06,
      "loss": 0.4341,
      "num_tokens": 631049408.0,
      "step": 666
    },
    {
      "epoch": 1.3055827619980411,
      "grad_norm": 0.09236539788484789,
      "learning_rate": 8.669568005781503e-06,
      "loss": 0.4256,
      "num_tokens": 631990643.0,
      "step": 667
    },
    {
      "epoch": 1.307541625857003,
      "grad_norm": 0.08748296294076853,
      "learning_rate": 8.665259359149132e-06,
      "loss": 0.4424,
      "num_tokens": 632969463.0,
      "step": 668
    },
    {
      "epoch": 1.3095004897159648,
      "grad_norm": 0.0884592018171943,
      "learning_rate": 8.660944821343516e-06,
      "loss": 0.425,
      "num_tokens": 633925921.0,
      "step": 669
    },
    {
      "epoch": 1.3114593535749266,
      "grad_norm": 0.09219670228189061,
      "learning_rate": 8.656624399299414e-06,
      "loss": 0.4267,
      "num_tokens": 634873252.0,
      "step": 670
    },
    {
      "epoch": 1.3134182174338884,
      "grad_norm": 0.0934173112646401,
      "learning_rate": 8.652298099961041e-06,
      "loss": 0.4257,
      "num_tokens": 635791641.0,
      "step": 671
    },
    {
      "epoch": 1.31537708129285,
      "grad_norm": 0.08899742527365999,
      "learning_rate": 8.64796593028206e-06,
      "loss": 0.4316,
      "num_tokens": 636738438.0,
      "step": 672
    },
    {
      "epoch": 1.3173359451518118,
      "grad_norm": 0.09119691239377391,
      "learning_rate": 8.64362789722557e-06,
      "loss": 0.4269,
      "num_tokens": 637679723.0,
      "step": 673
    },
    {
      "epoch": 1.3192948090107737,
      "grad_norm": 0.09007407850006857,
      "learning_rate": 8.639284007764095e-06,
      "loss": 0.436,
      "num_tokens": 638618644.0,
      "step": 674
    },
    {
      "epoch": 1.3212536728697355,
      "grad_norm": 0.09146593267787054,
      "learning_rate": 8.63493426887957e-06,
      "loss": 0.4339,
      "num_tokens": 639568481.0,
      "step": 675
    },
    {
      "epoch": 1.3232125367286973,
      "grad_norm": 0.08399176778639582,
      "learning_rate": 8.630578687563331e-06,
      "loss": 0.4406,
      "num_tokens": 640515741.0,
      "step": 676
    },
    {
      "epoch": 1.3251714005876591,
      "grad_norm": 0.0880932161054987,
      "learning_rate": 8.62621727081611e-06,
      "loss": 0.4195,
      "num_tokens": 641452866.0,
      "step": 677
    },
    {
      "epoch": 1.327130264446621,
      "grad_norm": 0.08856200671088392,
      "learning_rate": 8.621850025648008e-06,
      "loss": 0.4537,
      "num_tokens": 642409372.0,
      "step": 678
    },
    {
      "epoch": 1.3290891283055828,
      "grad_norm": 0.09542380676899659,
      "learning_rate": 8.617476959078507e-06,
      "loss": 0.4415,
      "num_tokens": 643340797.0,
      "step": 679
    },
    {
      "epoch": 1.3310479921645446,
      "grad_norm": 0.09151225164745504,
      "learning_rate": 8.613098078136436e-06,
      "loss": 0.447,
      "num_tokens": 644281695.0,
      "step": 680
    },
    {
      "epoch": 1.3330068560235064,
      "grad_norm": 0.09136175561815078,
      "learning_rate": 8.608713389859974e-06,
      "loss": 0.4361,
      "num_tokens": 645255661.0,
      "step": 681
    },
    {
      "epoch": 1.3349657198824683,
      "grad_norm": 0.09106928179725267,
      "learning_rate": 8.60432290129663e-06,
      "loss": 0.4468,
      "num_tokens": 646193281.0,
      "step": 682
    },
    {
      "epoch": 1.33692458374143,
      "grad_norm": 0.08772595573487554,
      "learning_rate": 8.59992661950324e-06,
      "loss": 0.4227,
      "num_tokens": 647120974.0,
      "step": 683
    },
    {
      "epoch": 1.338883447600392,
      "grad_norm": 0.09534180082475596,
      "learning_rate": 8.59552455154595e-06,
      "loss": 0.4234,
      "num_tokens": 648053667.0,
      "step": 684
    },
    {
      "epoch": 1.3408423114593535,
      "grad_norm": 0.0934432959803238,
      "learning_rate": 8.591116704500208e-06,
      "loss": 0.4285,
      "num_tokens": 649004374.0,
      "step": 685
    },
    {
      "epoch": 1.3428011753183153,
      "grad_norm": 0.09172664210031523,
      "learning_rate": 8.586703085450746e-06,
      "loss": 0.4327,
      "num_tokens": 649928772.0,
      "step": 686
    },
    {
      "epoch": 1.3447600391772772,
      "grad_norm": 0.085466539788077,
      "learning_rate": 8.582283701491576e-06,
      "loss": 0.4237,
      "num_tokens": 650861928.0,
      "step": 687
    },
    {
      "epoch": 1.346718903036239,
      "grad_norm": 0.09167034135696252,
      "learning_rate": 8.577858559725978e-06,
      "loss": 0.4299,
      "num_tokens": 651825940.0,
      "step": 688
    },
    {
      "epoch": 1.3486777668952008,
      "grad_norm": 0.0970531851277602,
      "learning_rate": 8.573427667266485e-06,
      "loss": 0.4371,
      "num_tokens": 652783108.0,
      "step": 689
    },
    {
      "epoch": 1.3506366307541626,
      "grad_norm": 0.09072779638303974,
      "learning_rate": 8.56899103123487e-06,
      "loss": 0.4412,
      "num_tokens": 653718931.0,
      "step": 690
    },
    {
      "epoch": 1.3525954946131244,
      "grad_norm": 0.08960104365095135,
      "learning_rate": 8.564548658762143e-06,
      "loss": 0.4351,
      "num_tokens": 654637215.0,
      "step": 691
    },
    {
      "epoch": 1.3545543584720863,
      "grad_norm": 0.09304129721029504,
      "learning_rate": 8.56010055698853e-06,
      "loss": 0.4265,
      "num_tokens": 655560101.0,
      "step": 692
    },
    {
      "epoch": 1.3565132223310479,
      "grad_norm": 0.08696641031349327,
      "learning_rate": 8.555646733063469e-06,
      "loss": 0.4332,
      "num_tokens": 656510677.0,
      "step": 693
    },
    {
      "epoch": 1.3584720861900097,
      "grad_norm": 0.09476845731340805,
      "learning_rate": 8.551187194145591e-06,
      "loss": 0.4307,
      "num_tokens": 657450174.0,
      "step": 694
    },
    {
      "epoch": 1.3604309500489715,
      "grad_norm": 0.09233630418006109,
      "learning_rate": 8.54672194740272e-06,
      "loss": 0.4372,
      "num_tokens": 658393088.0,
      "step": 695
    },
    {
      "epoch": 1.3623898139079333,
      "grad_norm": 0.08536818889372678,
      "learning_rate": 8.54225100001184e-06,
      "loss": 0.4324,
      "num_tokens": 659330001.0,
      "step": 696
    },
    {
      "epoch": 1.3643486777668952,
      "grad_norm": 0.0899119840527955,
      "learning_rate": 8.537774359159117e-06,
      "loss": 0.4397,
      "num_tokens": 660298957.0,
      "step": 697
    },
    {
      "epoch": 1.366307541625857,
      "grad_norm": 0.09321774430912169,
      "learning_rate": 8.533292032039853e-06,
      "loss": 0.4277,
      "num_tokens": 661230044.0,
      "step": 698
    },
    {
      "epoch": 1.3682664054848188,
      "grad_norm": 0.09133714290085906,
      "learning_rate": 8.528804025858494e-06,
      "loss": 0.4353,
      "num_tokens": 662110838.0,
      "step": 699
    },
    {
      "epoch": 1.3702252693437806,
      "grad_norm": 0.08931284953678228,
      "learning_rate": 8.52431034782862e-06,
      "loss": 0.427,
      "num_tokens": 663060531.0,
      "step": 700
    },
    {
      "epoch": 1.3721841332027425,
      "grad_norm": 0.08622021222265192,
      "learning_rate": 8.519811005172916e-06,
      "loss": 0.441,
      "num_tokens": 664014829.0,
      "step": 701
    },
    {
      "epoch": 1.3741429970617043,
      "grad_norm": 0.0887106044030063,
      "learning_rate": 8.51530600512318e-06,
      "loss": 0.445,
      "num_tokens": 665012979.0,
      "step": 702
    },
    {
      "epoch": 1.376101860920666,
      "grad_norm": 0.08644656242360411,
      "learning_rate": 8.510795354920301e-06,
      "loss": 0.4048,
      "num_tokens": 665985315.0,
      "step": 703
    },
    {
      "epoch": 1.378060724779628,
      "grad_norm": 0.08195711926338874,
      "learning_rate": 8.506279061814248e-06,
      "loss": 0.4309,
      "num_tokens": 666939271.0,
      "step": 704
    },
    {
      "epoch": 1.3800195886385898,
      "grad_norm": 0.09408883419964502,
      "learning_rate": 8.501757133064066e-06,
      "loss": 0.429,
      "num_tokens": 667887572.0,
      "step": 705
    },
    {
      "epoch": 1.3819784524975514,
      "grad_norm": 0.08811305636721038,
      "learning_rate": 8.497229575937848e-06,
      "loss": 0.4357,
      "num_tokens": 668819611.0,
      "step": 706
    },
    {
      "epoch": 1.3839373163565132,
      "grad_norm": 0.08775982223125199,
      "learning_rate": 8.492696397712741e-06,
      "loss": 0.4343,
      "num_tokens": 669776929.0,
      "step": 707
    },
    {
      "epoch": 1.385896180215475,
      "grad_norm": 0.08536772157566545,
      "learning_rate": 8.488157605674924e-06,
      "loss": 0.4283,
      "num_tokens": 670736923.0,
      "step": 708
    },
    {
      "epoch": 1.3878550440744368,
      "grad_norm": 0.08791249109250661,
      "learning_rate": 8.483613207119602e-06,
      "loss": 0.4289,
      "num_tokens": 671662081.0,
      "step": 709
    },
    {
      "epoch": 1.3898139079333987,
      "grad_norm": 0.09847314582224101,
      "learning_rate": 8.479063209350986e-06,
      "loss": 0.4371,
      "num_tokens": 672630075.0,
      "step": 710
    },
    {
      "epoch": 1.3917727717923605,
      "grad_norm": 0.0924208886089575,
      "learning_rate": 8.47450761968229e-06,
      "loss": 0.427,
      "num_tokens": 673510735.0,
      "step": 711
    },
    {
      "epoch": 1.3937316356513223,
      "grad_norm": 0.09513808239437288,
      "learning_rate": 8.469946445435719e-06,
      "loss": 0.4292,
      "num_tokens": 674461452.0,
      "step": 712
    },
    {
      "epoch": 1.395690499510284,
      "grad_norm": 0.09173600111784203,
      "learning_rate": 8.465379693942448e-06,
      "loss": 0.4305,
      "num_tokens": 675426239.0,
      "step": 713
    },
    {
      "epoch": 1.3976493633692457,
      "grad_norm": 0.08839984025128103,
      "learning_rate": 8.460807372542618e-06,
      "loss": 0.4614,
      "num_tokens": 676375387.0,
      "step": 714
    },
    {
      "epoch": 1.3996082272282075,
      "grad_norm": 0.09374198120073221,
      "learning_rate": 8.456229488585328e-06,
      "loss": 0.4213,
      "num_tokens": 677327368.0,
      "step": 715
    },
    {
      "epoch": 1.4015670910871694,
      "grad_norm": 0.09072590715344986,
      "learning_rate": 8.451646049428607e-06,
      "loss": 0.4235,
      "num_tokens": 678283492.0,
      "step": 716
    },
    {
      "epoch": 1.4035259549461312,
      "grad_norm": 0.09335990897846304,
      "learning_rate": 8.447057062439425e-06,
      "loss": 0.433,
      "num_tokens": 679254020.0,
      "step": 717
    },
    {
      "epoch": 1.405484818805093,
      "grad_norm": 0.09508005476618756,
      "learning_rate": 8.44246253499366e-06,
      "loss": 0.4288,
      "num_tokens": 680237522.0,
      "step": 718
    },
    {
      "epoch": 1.4074436826640548,
      "grad_norm": 0.09470801427949821,
      "learning_rate": 8.437862474476099e-06,
      "loss": 0.4283,
      "num_tokens": 681198137.0,
      "step": 719
    },
    {
      "epoch": 1.4094025465230167,
      "grad_norm": 0.09040661798704991,
      "learning_rate": 8.433256888280422e-06,
      "loss": 0.4342,
      "num_tokens": 682109134.0,
      "step": 720
    },
    {
      "epoch": 1.4113614103819785,
      "grad_norm": 0.09718020598488185,
      "learning_rate": 8.428645783809187e-06,
      "loss": 0.4447,
      "num_tokens": 683041802.0,
      "step": 721
    },
    {
      "epoch": 1.4133202742409403,
      "grad_norm": 0.09386655251014273,
      "learning_rate": 8.424029168473829e-06,
      "loss": 0.4194,
      "num_tokens": 684005481.0,
      "step": 722
    },
    {
      "epoch": 1.4152791380999021,
      "grad_norm": 0.08941083494167625,
      "learning_rate": 8.419407049694634e-06,
      "loss": 0.4363,
      "num_tokens": 684974162.0,
      "step": 723
    },
    {
      "epoch": 1.417238001958864,
      "grad_norm": 0.08558228587510855,
      "learning_rate": 8.414779434900736e-06,
      "loss": 0.4388,
      "num_tokens": 685922624.0,
      "step": 724
    },
    {
      "epoch": 1.4191968658178258,
      "grad_norm": 0.08820784119716409,
      "learning_rate": 8.410146331530102e-06,
      "loss": 0.4401,
      "num_tokens": 686900397.0,
      "step": 725
    },
    {
      "epoch": 1.4211557296767874,
      "grad_norm": 0.08852337982491124,
      "learning_rate": 8.405507747029524e-06,
      "loss": 0.4253,
      "num_tokens": 687835960.0,
      "step": 726
    },
    {
      "epoch": 1.4231145935357492,
      "grad_norm": 0.09168139513037091,
      "learning_rate": 8.400863688854598e-06,
      "loss": 0.4409,
      "num_tokens": 688795372.0,
      "step": 727
    },
    {
      "epoch": 1.425073457394711,
      "grad_norm": 0.0878237833737238,
      "learning_rate": 8.396214164469723e-06,
      "loss": 0.4391,
      "num_tokens": 689754498.0,
      "step": 728
    },
    {
      "epoch": 1.4270323212536729,
      "grad_norm": 0.08916921173918818,
      "learning_rate": 8.391559181348081e-06,
      "loss": 0.4326,
      "num_tokens": 690700068.0,
      "step": 729
    },
    {
      "epoch": 1.4289911851126347,
      "grad_norm": 0.08885595175037213,
      "learning_rate": 8.38689874697163e-06,
      "loss": 0.425,
      "num_tokens": 691625569.0,
      "step": 730
    },
    {
      "epoch": 1.4309500489715965,
      "grad_norm": 0.09355201498428904,
      "learning_rate": 8.382232868831088e-06,
      "loss": 0.4247,
      "num_tokens": 692549153.0,
      "step": 731
    },
    {
      "epoch": 1.4329089128305583,
      "grad_norm": 0.08706942822188166,
      "learning_rate": 8.377561554425923e-06,
      "loss": 0.4386,
      "num_tokens": 693508761.0,
      "step": 732
    },
    {
      "epoch": 1.4348677766895201,
      "grad_norm": 0.09133725236247704,
      "learning_rate": 8.372884811264339e-06,
      "loss": 0.4408,
      "num_tokens": 694431253.0,
      "step": 733
    },
    {
      "epoch": 1.4368266405484817,
      "grad_norm": 0.08836235986857442,
      "learning_rate": 8.368202646863273e-06,
      "loss": 0.4312,
      "num_tokens": 695387922.0,
      "step": 734
    },
    {
      "epoch": 1.4387855044074436,
      "grad_norm": 0.09730075311717222,
      "learning_rate": 8.363515068748365e-06,
      "loss": 0.4301,
      "num_tokens": 696352116.0,
      "step": 735
    },
    {
      "epoch": 1.4407443682664054,
      "grad_norm": 0.0928184606574738,
      "learning_rate": 8.358822084453964e-06,
      "loss": 0.4446,
      "num_tokens": 697336443.0,
      "step": 736
    },
    {
      "epoch": 1.4427032321253672,
      "grad_norm": 0.09866917309104557,
      "learning_rate": 8.354123701523107e-06,
      "loss": 0.4301,
      "num_tokens": 698300399.0,
      "step": 737
    },
    {
      "epoch": 1.444662095984329,
      "grad_norm": 0.09600606306106796,
      "learning_rate": 8.349419927507505e-06,
      "loss": 0.437,
      "num_tokens": 699243353.0,
      "step": 738
    },
    {
      "epoch": 1.4466209598432909,
      "grad_norm": 0.09237112341447015,
      "learning_rate": 8.344710769967537e-06,
      "loss": 0.4272,
      "num_tokens": 700171912.0,
      "step": 739
    },
    {
      "epoch": 1.4485798237022527,
      "grad_norm": 0.08572563052517873,
      "learning_rate": 8.339996236472236e-06,
      "loss": 0.4351,
      "num_tokens": 701139037.0,
      "step": 740
    },
    {
      "epoch": 1.4505386875612145,
      "grad_norm": 0.08834229641252914,
      "learning_rate": 8.335276334599274e-06,
      "loss": 0.4296,
      "num_tokens": 702073841.0,
      "step": 741
    },
    {
      "epoch": 1.4524975514201763,
      "grad_norm": 0.0871311174805632,
      "learning_rate": 8.33055107193495e-06,
      "loss": 0.4299,
      "num_tokens": 703047084.0,
      "step": 742
    },
    {
      "epoch": 1.4544564152791382,
      "grad_norm": 0.0968390574225456,
      "learning_rate": 8.325820456074181e-06,
      "loss": 0.4363,
      "num_tokens": 704006650.0,
      "step": 743
    },
    {
      "epoch": 1.4564152791381,
      "grad_norm": 0.09770693695080601,
      "learning_rate": 8.321084494620489e-06,
      "loss": 0.4291,
      "num_tokens": 705005752.0,
      "step": 744
    },
    {
      "epoch": 1.4583741429970618,
      "grad_norm": 0.09042781818496945,
      "learning_rate": 8.316343195185988e-06,
      "loss": 0.433,
      "num_tokens": 705943750.0,
      "step": 745
    },
    {
      "epoch": 1.4603330068560236,
      "grad_norm": 0.09074059900347199,
      "learning_rate": 8.31159656539137e-06,
      "loss": 0.4443,
      "num_tokens": 706857464.0,
      "step": 746
    },
    {
      "epoch": 1.4622918707149852,
      "grad_norm": 0.09296990315373345,
      "learning_rate": 8.306844612865895e-06,
      "loss": 0.4278,
      "num_tokens": 707798061.0,
      "step": 747
    },
    {
      "epoch": 1.464250734573947,
      "grad_norm": 0.09192749101371916,
      "learning_rate": 8.30208734524738e-06,
      "loss": 0.4443,
      "num_tokens": 708757874.0,
      "step": 748
    },
    {
      "epoch": 1.4662095984329089,
      "grad_norm": 0.0920452893115045,
      "learning_rate": 8.297324770182183e-06,
      "loss": 0.4379,
      "num_tokens": 709711444.0,
      "step": 749
    },
    {
      "epoch": 1.4681684622918707,
      "grad_norm": 0.08734169812439341,
      "learning_rate": 8.292556895325195e-06,
      "loss": 0.4434,
      "num_tokens": 710675618.0,
      "step": 750
    },
    {
      "epoch": 1.4701273261508325,
      "grad_norm": 0.08918617592674521,
      "learning_rate": 8.28778372833982e-06,
      "loss": 0.4209,
      "num_tokens": 711617502.0,
      "step": 751
    },
    {
      "epoch": 1.4720861900097943,
      "grad_norm": 0.08764649345766538,
      "learning_rate": 8.283005276897975e-06,
      "loss": 0.4483,
      "num_tokens": 712567364.0,
      "step": 752
    },
    {
      "epoch": 1.4740450538687562,
      "grad_norm": 0.09092598413252254,
      "learning_rate": 8.278221548680068e-06,
      "loss": 0.4346,
      "num_tokens": 713498779.0,
      "step": 753
    },
    {
      "epoch": 1.476003917727718,
      "grad_norm": 0.09283053237607067,
      "learning_rate": 8.273432551374984e-06,
      "loss": 0.4308,
      "num_tokens": 714454721.0,
      "step": 754
    },
    {
      "epoch": 1.4779627815866796,
      "grad_norm": 0.09624790658111503,
      "learning_rate": 8.268638292680085e-06,
      "loss": 0.4348,
      "num_tokens": 715410148.0,
      "step": 755
    },
    {
      "epoch": 1.4799216454456414,
      "grad_norm": 0.0905438810326495,
      "learning_rate": 8.263838780301182e-06,
      "loss": 0.42,
      "num_tokens": 716371428.0,
      "step": 756
    },
    {
      "epoch": 1.4818805093046032,
      "grad_norm": 0.08900777996620862,
      "learning_rate": 8.259034021952537e-06,
      "loss": 0.4275,
      "num_tokens": 717334918.0,
      "step": 757
    },
    {
      "epoch": 1.483839373163565,
      "grad_norm": 0.09539894517349785,
      "learning_rate": 8.254224025356838e-06,
      "loss": 0.4309,
      "num_tokens": 718268899.0,
      "step": 758
    },
    {
      "epoch": 1.485798237022527,
      "grad_norm": 0.08907059530467444,
      "learning_rate": 8.249408798245196e-06,
      "loss": 0.4402,
      "num_tokens": 719225615.0,
      "step": 759
    },
    {
      "epoch": 1.4877571008814887,
      "grad_norm": 0.08749241883163766,
      "learning_rate": 8.244588348357127e-06,
      "loss": 0.4342,
      "num_tokens": 720155206.0,
      "step": 760
    },
    {
      "epoch": 1.4897159647404505,
      "grad_norm": 0.0881335687273988,
      "learning_rate": 8.239762683440542e-06,
      "loss": 0.4304,
      "num_tokens": 721082402.0,
      "step": 761
    },
    {
      "epoch": 1.4916748285994124,
      "grad_norm": 0.09397566961188934,
      "learning_rate": 8.234931811251739e-06,
      "loss": 0.4245,
      "num_tokens": 722036368.0,
      "step": 762
    },
    {
      "epoch": 1.4936336924583742,
      "grad_norm": 0.08868494862611866,
      "learning_rate": 8.230095739555377e-06,
      "loss": 0.426,
      "num_tokens": 722977150.0,
      "step": 763
    },
    {
      "epoch": 1.495592556317336,
      "grad_norm": 0.08822709518087088,
      "learning_rate": 8.225254476124479e-06,
      "loss": 0.4382,
      "num_tokens": 723921610.0,
      "step": 764
    },
    {
      "epoch": 1.4975514201762978,
      "grad_norm": 0.09222431290824455,
      "learning_rate": 8.220408028740408e-06,
      "loss": 0.4406,
      "num_tokens": 724882118.0,
      "step": 765
    },
    {
      "epoch": 1.4995102840352597,
      "grad_norm": 0.08633570371148419,
      "learning_rate": 8.215556405192863e-06,
      "loss": 0.4207,
      "num_tokens": 725829599.0,
      "step": 766
    },
    {
      "epoch": 1.5014691478942215,
      "grad_norm": 0.0868716952312777,
      "learning_rate": 8.21069961327986e-06,
      "loss": 0.4231,
      "num_tokens": 726772987.0,
      "step": 767
    },
    {
      "epoch": 1.5034280117531833,
      "grad_norm": 0.08768214349778954,
      "learning_rate": 8.205837660807726e-06,
      "loss": 0.4224,
      "num_tokens": 727706645.0,
      "step": 768
    },
    {
      "epoch": 1.505386875612145,
      "grad_norm": 0.09487800672671816,
      "learning_rate": 8.200970555591073e-06,
      "loss": 0.4434,
      "num_tokens": 728681109.0,
      "step": 769
    },
    {
      "epoch": 1.5073457394711067,
      "grad_norm": 0.08487833690630645,
      "learning_rate": 8.19609830545281e-06,
      "loss": 0.4337,
      "num_tokens": 729617569.0,
      "step": 770
    },
    {
      "epoch": 1.5093046033300686,
      "grad_norm": 0.09218774680180625,
      "learning_rate": 8.191220918224102e-06,
      "loss": 0.4281,
      "num_tokens": 730579210.0,
      "step": 771
    },
    {
      "epoch": 1.5112634671890304,
      "grad_norm": 0.08650153794869085,
      "learning_rate": 8.186338401744377e-06,
      "loss": 0.4446,
      "num_tokens": 731526742.0,
      "step": 772
    },
    {
      "epoch": 1.5132223310479922,
      "grad_norm": 0.08973200677638382,
      "learning_rate": 8.181450763861306e-06,
      "loss": 0.439,
      "num_tokens": 732472379.0,
      "step": 773
    },
    {
      "epoch": 1.515181194906954,
      "grad_norm": 0.08939353084355083,
      "learning_rate": 8.176558012430792e-06,
      "loss": 0.4327,
      "num_tokens": 733447256.0,
      "step": 774
    },
    {
      "epoch": 1.5171400587659156,
      "grad_norm": 0.08695642985328696,
      "learning_rate": 8.171660155316958e-06,
      "loss": 0.4347,
      "num_tokens": 734428538.0,
      "step": 775
    },
    {
      "epoch": 1.5190989226248774,
      "grad_norm": 0.08813042852208332,
      "learning_rate": 8.16675720039213e-06,
      "loss": 0.4313,
      "num_tokens": 735343982.0,
      "step": 776
    },
    {
      "epoch": 1.5210577864838393,
      "grad_norm": 0.09122332242998978,
      "learning_rate": 8.161849155536834e-06,
      "loss": 0.4309,
      "num_tokens": 736281010.0,
      "step": 777
    },
    {
      "epoch": 1.523016650342801,
      "grad_norm": 0.08858224521663992,
      "learning_rate": 8.156936028639768e-06,
      "loss": 0.433,
      "num_tokens": 737241458.0,
      "step": 778
    },
    {
      "epoch": 1.524975514201763,
      "grad_norm": 0.08752140528851679,
      "learning_rate": 8.152017827597807e-06,
      "loss": 0.4291,
      "num_tokens": 738191019.0,
      "step": 779
    },
    {
      "epoch": 1.5269343780607247,
      "grad_norm": 0.08985526676071166,
      "learning_rate": 8.147094560315978e-06,
      "loss": 0.4267,
      "num_tokens": 739125705.0,
      "step": 780
    },
    {
      "epoch": 1.5288932419196866,
      "grad_norm": 0.08674954335352558,
      "learning_rate": 8.142166234707448e-06,
      "loss": 0.4429,
      "num_tokens": 740067530.0,
      "step": 781
    },
    {
      "epoch": 1.5308521057786484,
      "grad_norm": 0.09647413877933882,
      "learning_rate": 8.13723285869352e-06,
      "loss": 0.433,
      "num_tokens": 741012286.0,
      "step": 782
    },
    {
      "epoch": 1.5328109696376102,
      "grad_norm": 0.08867424938439858,
      "learning_rate": 8.132294440203609e-06,
      "loss": 0.4392,
      "num_tokens": 742006599.0,
      "step": 783
    },
    {
      "epoch": 1.534769833496572,
      "grad_norm": 0.08614258269227894,
      "learning_rate": 8.127350987175242e-06,
      "loss": 0.4298,
      "num_tokens": 742928475.0,
      "step": 784
    },
    {
      "epoch": 1.5367286973555339,
      "grad_norm": 0.08941741651572559,
      "learning_rate": 8.12240250755403e-06,
      "loss": 0.4271,
      "num_tokens": 743897067.0,
      "step": 785
    },
    {
      "epoch": 1.5386875612144957,
      "grad_norm": 0.08307451835655041,
      "learning_rate": 8.117449009293668e-06,
      "loss": 0.415,
      "num_tokens": 744846396.0,
      "step": 786
    },
    {
      "epoch": 1.5406464250734575,
      "grad_norm": 0.08388009373690888,
      "learning_rate": 8.112490500355916e-06,
      "loss": 0.4321,
      "num_tokens": 745818252.0,
      "step": 787
    },
    {
      "epoch": 1.5426052889324193,
      "grad_norm": 0.08974967024873853,
      "learning_rate": 8.10752698871059e-06,
      "loss": 0.4361,
      "num_tokens": 746776422.0,
      "step": 788
    },
    {
      "epoch": 1.5445641527913812,
      "grad_norm": 0.08706306024197821,
      "learning_rate": 8.10255848233554e-06,
      "loss": 0.4301,
      "num_tokens": 747726198.0,
      "step": 789
    },
    {
      "epoch": 1.5465230166503428,
      "grad_norm": 0.08687534246069954,
      "learning_rate": 8.097584989216653e-06,
      "loss": 0.4191,
      "num_tokens": 748651636.0,
      "step": 790
    },
    {
      "epoch": 1.5484818805093046,
      "grad_norm": 0.08716398270684944,
      "learning_rate": 8.092606517347828e-06,
      "loss": 0.4342,
      "num_tokens": 749589769.0,
      "step": 791
    },
    {
      "epoch": 1.5504407443682664,
      "grad_norm": 0.086505218555598,
      "learning_rate": 8.08762307473096e-06,
      "loss": 0.4475,
      "num_tokens": 750588105.0,
      "step": 792
    },
    {
      "epoch": 1.5523996082272282,
      "grad_norm": 0.08746637865179134,
      "learning_rate": 8.082634669375944e-06,
      "loss": 0.4332,
      "num_tokens": 751558123.0,
      "step": 793
    },
    {
      "epoch": 1.55435847208619,
      "grad_norm": 0.0840126542938758,
      "learning_rate": 8.077641309300643e-06,
      "loss": 0.4138,
      "num_tokens": 752500192.0,
      "step": 794
    },
    {
      "epoch": 1.5563173359451516,
      "grad_norm": 0.08652533631916479,
      "learning_rate": 8.072643002530891e-06,
      "loss": 0.4258,
      "num_tokens": 753482524.0,
      "step": 795
    },
    {
      "epoch": 1.5582761998041135,
      "grad_norm": 0.08596044230527805,
      "learning_rate": 8.067639757100465e-06,
      "loss": 0.4314,
      "num_tokens": 754425718.0,
      "step": 796
    },
    {
      "epoch": 1.5602350636630753,
      "grad_norm": 0.09017531758130216,
      "learning_rate": 8.062631581051088e-06,
      "loss": 0.4343,
      "num_tokens": 755342924.0,
      "step": 797
    },
    {
      "epoch": 1.5621939275220371,
      "grad_norm": 0.08539720776155492,
      "learning_rate": 8.057618482432399e-06,
      "loss": 0.4099,
      "num_tokens": 756294294.0,
      "step": 798
    },
    {
      "epoch": 1.564152791380999,
      "grad_norm": 0.08714477235939926,
      "learning_rate": 8.052600469301958e-06,
      "loss": 0.4441,
      "num_tokens": 757290351.0,
      "step": 799
    },
    {
      "epoch": 1.5661116552399608,
      "grad_norm": 0.08212796379618922,
      "learning_rate": 8.047577549725219e-06,
      "loss": 0.427,
      "num_tokens": 758262720.0,
      "step": 800
    },
    {
      "epoch": 1.5680705190989226,
      "grad_norm": 0.08693270850999413,
      "learning_rate": 8.042549731775521e-06,
      "loss": 0.4102,
      "num_tokens": 759177946.0,
      "step": 801
    },
    {
      "epoch": 1.5700293829578844,
      "grad_norm": 0.08633453071203538,
      "learning_rate": 8.037517023534082e-06,
      "loss": 0.4446,
      "num_tokens": 760130872.0,
      "step": 802
    },
    {
      "epoch": 1.5719882468168462,
      "grad_norm": 0.09225219773572015,
      "learning_rate": 8.032479433089976e-06,
      "loss": 0.4411,
      "num_tokens": 761084377.0,
      "step": 803
    },
    {
      "epoch": 1.573947110675808,
      "grad_norm": 0.08702763450061118,
      "learning_rate": 8.027436968540124e-06,
      "loss": 0.4426,
      "num_tokens": 762015812.0,
      "step": 804
    },
    {
      "epoch": 1.5759059745347699,
      "grad_norm": 0.08846277303169,
      "learning_rate": 8.022389637989278e-06,
      "loss": 0.4322,
      "num_tokens": 762984337.0,
      "step": 805
    },
    {
      "epoch": 1.5778648383937317,
      "grad_norm": 0.09587811911424722,
      "learning_rate": 8.01733744955002e-06,
      "loss": 0.4328,
      "num_tokens": 763964568.0,
      "step": 806
    },
    {
      "epoch": 1.5798237022526935,
      "grad_norm": 0.08818955709529418,
      "learning_rate": 8.012280411342733e-06,
      "loss": 0.4465,
      "num_tokens": 764901824.0,
      "step": 807
    },
    {
      "epoch": 1.5817825661116554,
      "grad_norm": 0.09579378209022592,
      "learning_rate": 8.0072185314956e-06,
      "loss": 0.4245,
      "num_tokens": 765842584.0,
      "step": 808
    },
    {
      "epoch": 1.5837414299706172,
      "grad_norm": 0.09192508480952484,
      "learning_rate": 8.002151818144577e-06,
      "loss": 0.4381,
      "num_tokens": 766792416.0,
      "step": 809
    },
    {
      "epoch": 1.585700293829579,
      "grad_norm": 0.08778112997632395,
      "learning_rate": 7.997080279433402e-06,
      "loss": 0.4331,
      "num_tokens": 767750154.0,
      "step": 810
    },
    {
      "epoch": 1.5876591576885406,
      "grad_norm": 0.08591223462924581,
      "learning_rate": 7.992003923513557e-06,
      "loss": 0.4395,
      "num_tokens": 768681114.0,
      "step": 811
    },
    {
      "epoch": 1.5896180215475024,
      "grad_norm": 0.09197418425156342,
      "learning_rate": 7.986922758544272e-06,
      "loss": 0.4442,
      "num_tokens": 769613251.0,
      "step": 812
    },
    {
      "epoch": 1.5915768854064642,
      "grad_norm": 0.08981432669383796,
      "learning_rate": 7.981836792692508e-06,
      "loss": 0.4289,
      "num_tokens": 770564128.0,
      "step": 813
    },
    {
      "epoch": 1.593535749265426,
      "grad_norm": 0.08670656993160512,
      "learning_rate": 7.976746034132938e-06,
      "loss": 0.4487,
      "num_tokens": 771553074.0,
      "step": 814
    },
    {
      "epoch": 1.595494613124388,
      "grad_norm": 0.08789310833337932,
      "learning_rate": 7.971650491047944e-06,
      "loss": 0.4212,
      "num_tokens": 772481970.0,
      "step": 815
    },
    {
      "epoch": 1.5974534769833495,
      "grad_norm": 0.08862878479775865,
      "learning_rate": 7.966550171627592e-06,
      "loss": 0.4317,
      "num_tokens": 773369883.0,
      "step": 816
    },
    {
      "epoch": 1.5994123408423113,
      "grad_norm": 0.09019671038303287,
      "learning_rate": 7.96144508406963e-06,
      "loss": 0.4281,
      "num_tokens": 774334951.0,
      "step": 817
    },
    {
      "epoch": 1.6013712047012731,
      "grad_norm": 0.08631560650425865,
      "learning_rate": 7.956335236579468e-06,
      "loss": 0.4357,
      "num_tokens": 775272444.0,
      "step": 818
    },
    {
      "epoch": 1.603330068560235,
      "grad_norm": 0.09653679998604583,
      "learning_rate": 7.951220637370166e-06,
      "loss": 0.435,
      "num_tokens": 776199956.0,
      "step": 819
    },
    {
      "epoch": 1.6052889324191968,
      "grad_norm": 0.08792046484591014,
      "learning_rate": 7.946101294662418e-06,
      "loss": 0.4426,
      "num_tokens": 777169727.0,
      "step": 820
    },
    {
      "epoch": 1.6072477962781586,
      "grad_norm": 0.08557432522056063,
      "learning_rate": 7.940977216684553e-06,
      "loss": 0.4356,
      "num_tokens": 778113352.0,
      "step": 821
    },
    {
      "epoch": 1.6092066601371204,
      "grad_norm": 0.08792107016892484,
      "learning_rate": 7.935848411672499e-06,
      "loss": 0.4396,
      "num_tokens": 779071100.0,
      "step": 822
    },
    {
      "epoch": 1.6111655239960823,
      "grad_norm": 0.08278917692252104,
      "learning_rate": 7.93071488786979e-06,
      "loss": 0.4273,
      "num_tokens": 779994414.0,
      "step": 823
    },
    {
      "epoch": 1.613124387855044,
      "grad_norm": 0.0937239451472487,
      "learning_rate": 7.92557665352754e-06,
      "loss": 0.4343,
      "num_tokens": 780910389.0,
      "step": 824
    },
    {
      "epoch": 1.615083251714006,
      "grad_norm": 0.09212306734669672,
      "learning_rate": 7.920433716904436e-06,
      "loss": 0.4239,
      "num_tokens": 781843035.0,
      "step": 825
    },
    {
      "epoch": 1.6170421155729677,
      "grad_norm": 0.08931960362521048,
      "learning_rate": 7.915286086266723e-06,
      "loss": 0.427,
      "num_tokens": 782810958.0,
      "step": 826
    },
    {
      "epoch": 1.6190009794319296,
      "grad_norm": 0.0905389419758957,
      "learning_rate": 7.91013376988819e-06,
      "loss": 0.43,
      "num_tokens": 783786712.0,
      "step": 827
    },
    {
      "epoch": 1.6209598432908914,
      "grad_norm": 0.08644014656081704,
      "learning_rate": 7.904976776050156e-06,
      "loss": 0.4352,
      "num_tokens": 784736540.0,
      "step": 828
    },
    {
      "epoch": 1.6229187071498532,
      "grad_norm": 0.09199717032916593,
      "learning_rate": 7.899815113041463e-06,
      "loss": 0.4276,
      "num_tokens": 785728597.0,
      "step": 829
    },
    {
      "epoch": 1.624877571008815,
      "grad_norm": 0.08842558324087799,
      "learning_rate": 7.894648789158452e-06,
      "loss": 0.4468,
      "num_tokens": 786655785.0,
      "step": 830
    },
    {
      "epoch": 1.6268364348677768,
      "grad_norm": 0.10262169363913219,
      "learning_rate": 7.88947781270496e-06,
      "loss": 0.4191,
      "num_tokens": 787590282.0,
      "step": 831
    },
    {
      "epoch": 1.6287952987267384,
      "grad_norm": 0.09512038984159299,
      "learning_rate": 7.8843021919923e-06,
      "loss": 0.4421,
      "num_tokens": 788510805.0,
      "step": 832
    },
    {
      "epoch": 1.6307541625857003,
      "grad_norm": 0.0900043810946666,
      "learning_rate": 7.879121935339249e-06,
      "loss": 0.4198,
      "num_tokens": 789398786.0,
      "step": 833
    },
    {
      "epoch": 1.632713026444662,
      "grad_norm": 0.09477714069470886,
      "learning_rate": 7.873937051072037e-06,
      "loss": 0.4365,
      "num_tokens": 790336897.0,
      "step": 834
    },
    {
      "epoch": 1.634671890303624,
      "grad_norm": 0.08952014362266324,
      "learning_rate": 7.868747547524326e-06,
      "loss": 0.4265,
      "num_tokens": 791289556.0,
      "step": 835
    },
    {
      "epoch": 1.6366307541625857,
      "grad_norm": 0.09778589170144017,
      "learning_rate": 7.863553433037217e-06,
      "loss": 0.4462,
      "num_tokens": 792272860.0,
      "step": 836
    },
    {
      "epoch": 1.6385896180215473,
      "grad_norm": 0.09992143697321337,
      "learning_rate": 7.858354715959207e-06,
      "loss": 0.4414,
      "num_tokens": 793189011.0,
      "step": 837
    },
    {
      "epoch": 1.6405484818805092,
      "grad_norm": 0.09708392669394192,
      "learning_rate": 7.853151404646195e-06,
      "loss": 0.4259,
      "num_tokens": 794149245.0,
      "step": 838
    },
    {
      "epoch": 1.642507345739471,
      "grad_norm": 0.0875146819737148,
      "learning_rate": 7.847943507461471e-06,
      "loss": 0.4315,
      "num_tokens": 795082319.0,
      "step": 839
    },
    {
      "epoch": 1.6444662095984328,
      "grad_norm": 0.08844400418164805,
      "learning_rate": 7.842731032775688e-06,
      "loss": 0.4325,
      "num_tokens": 796044534.0,
      "step": 840
    },
    {
      "epoch": 1.6464250734573946,
      "grad_norm": 0.08993448791702277,
      "learning_rate": 7.83751398896686e-06,
      "loss": 0.4274,
      "num_tokens": 797002349.0,
      "step": 841
    },
    {
      "epoch": 1.6483839373163565,
      "grad_norm": 0.09779719083256716,
      "learning_rate": 7.832292384420342e-06,
      "loss": 0.427,
      "num_tokens": 797976708.0,
      "step": 842
    },
    {
      "epoch": 1.6503428011753183,
      "grad_norm": 0.09313945626158736,
      "learning_rate": 7.827066227528825e-06,
      "loss": 0.4324,
      "num_tokens": 798895955.0,
      "step": 843
    },
    {
      "epoch": 1.65230166503428,
      "grad_norm": 0.08946785732709807,
      "learning_rate": 7.821835526692311e-06,
      "loss": 0.4317,
      "num_tokens": 799870680.0,
      "step": 844
    },
    {
      "epoch": 1.654260528893242,
      "grad_norm": 0.0887193935127447,
      "learning_rate": 7.81660029031811e-06,
      "loss": 0.4385,
      "num_tokens": 800831985.0,
      "step": 845
    },
    {
      "epoch": 1.6562193927522038,
      "grad_norm": 0.08719254083646917,
      "learning_rate": 7.81136052682082e-06,
      "loss": 0.4296,
      "num_tokens": 801786190.0,
      "step": 846
    },
    {
      "epoch": 1.6581782566111656,
      "grad_norm": 0.08792814450101948,
      "learning_rate": 7.806116244622316e-06,
      "loss": 0.429,
      "num_tokens": 802786436.0,
      "step": 847
    },
    {
      "epoch": 1.6601371204701274,
      "grad_norm": 0.09345059438799186,
      "learning_rate": 7.80086745215173e-06,
      "loss": 0.4345,
      "num_tokens": 803764644.0,
      "step": 848
    },
    {
      "epoch": 1.6620959843290892,
      "grad_norm": 0.08817644282627775,
      "learning_rate": 7.795614157845457e-06,
      "loss": 0.4337,
      "num_tokens": 804691651.0,
      "step": 849
    },
    {
      "epoch": 1.664054848188051,
      "grad_norm": 0.09032795103795767,
      "learning_rate": 7.790356370147115e-06,
      "loss": 0.4229,
      "num_tokens": 805632028.0,
      "step": 850
    },
    {
      "epoch": 1.6660137120470129,
      "grad_norm": 0.08514604231993507,
      "learning_rate": 7.78509409750755e-06,
      "loss": 0.4238,
      "num_tokens": 806587305.0,
      "step": 851
    },
    {
      "epoch": 1.6679725759059747,
      "grad_norm": 0.08510216243372119,
      "learning_rate": 7.779827348384814e-06,
      "loss": 0.4276,
      "num_tokens": 807515943.0,
      "step": 852
    },
    {
      "epoch": 1.6699314397649363,
      "grad_norm": 0.09082030671758394,
      "learning_rate": 7.774556131244158e-06,
      "loss": 0.4262,
      "num_tokens": 808432184.0,
      "step": 853
    },
    {
      "epoch": 1.6718903036238981,
      "grad_norm": 0.09036312344258203,
      "learning_rate": 7.769280454558009e-06,
      "loss": 0.4348,
      "num_tokens": 809386146.0,
      "step": 854
    },
    {
      "epoch": 1.67384916748286,
      "grad_norm": 0.09592697261344595,
      "learning_rate": 7.764000326805967e-06,
      "loss": 0.4403,
      "num_tokens": 810355041.0,
      "step": 855
    },
    {
      "epoch": 1.6758080313418218,
      "grad_norm": 0.08847463534602111,
      "learning_rate": 7.758715756474785e-06,
      "loss": 0.4345,
      "num_tokens": 811304494.0,
      "step": 856
    },
    {
      "epoch": 1.6777668952007836,
      "grad_norm": 0.08942030009235737,
      "learning_rate": 7.753426752058354e-06,
      "loss": 0.4371,
      "num_tokens": 812264811.0,
      "step": 857
    },
    {
      "epoch": 1.6797257590597452,
      "grad_norm": 0.09016146484851315,
      "learning_rate": 7.748133322057693e-06,
      "loss": 0.4368,
      "num_tokens": 813213485.0,
      "step": 858
    },
    {
      "epoch": 1.681684622918707,
      "grad_norm": 0.09156888554192899,
      "learning_rate": 7.742835474980936e-06,
      "loss": 0.4525,
      "num_tokens": 814150318.0,
      "step": 859
    },
    {
      "epoch": 1.6836434867776688,
      "grad_norm": 0.08900818567746938,
      "learning_rate": 7.737533219343314e-06,
      "loss": 0.4324,
      "num_tokens": 815063347.0,
      "step": 860
    },
    {
      "epoch": 1.6856023506366307,
      "grad_norm": 0.09431275952391421,
      "learning_rate": 7.732226563667146e-06,
      "loss": 0.4361,
      "num_tokens": 816056695.0,
      "step": 861
    },
    {
      "epoch": 1.6875612144955925,
      "grad_norm": 0.08600326004812062,
      "learning_rate": 7.726915516481824e-06,
      "loss": 0.4276,
      "num_tokens": 816992130.0,
      "step": 862
    },
    {
      "epoch": 1.6895200783545543,
      "grad_norm": 0.09421732905820908,
      "learning_rate": 7.721600086323795e-06,
      "loss": 0.4333,
      "num_tokens": 817917134.0,
      "step": 863
    },
    {
      "epoch": 1.6914789422135161,
      "grad_norm": 0.08822550736347473,
      "learning_rate": 7.716280281736553e-06,
      "loss": 0.4417,
      "num_tokens": 818858756.0,
      "step": 864
    },
    {
      "epoch": 1.693437806072478,
      "grad_norm": 0.09570061746940511,
      "learning_rate": 7.710956111270621e-06,
      "loss": 0.4296,
      "num_tokens": 819800195.0,
      "step": 865
    },
    {
      "epoch": 1.6953966699314398,
      "grad_norm": 0.09744285532510016,
      "learning_rate": 7.705627583483543e-06,
      "loss": 0.4319,
      "num_tokens": 820767605.0,
      "step": 866
    },
    {
      "epoch": 1.6973555337904016,
      "grad_norm": 0.0897795504517481,
      "learning_rate": 7.700294706939864e-06,
      "loss": 0.4516,
      "num_tokens": 821738487.0,
      "step": 867
    },
    {
      "epoch": 1.6993143976493634,
      "grad_norm": 0.08888776807121071,
      "learning_rate": 7.69495749021112e-06,
      "loss": 0.4414,
      "num_tokens": 822682076.0,
      "step": 868
    },
    {
      "epoch": 1.7012732615083253,
      "grad_norm": 0.08984604263470013,
      "learning_rate": 7.68961594187582e-06,
      "loss": 0.4296,
      "num_tokens": 823672726.0,
      "step": 869
    },
    {
      "epoch": 1.703232125367287,
      "grad_norm": 0.08721830032120235,
      "learning_rate": 7.68427007051944e-06,
      "loss": 0.4438,
      "num_tokens": 824599914.0,
      "step": 870
    },
    {
      "epoch": 1.705190989226249,
      "grad_norm": 0.09343695241717352,
      "learning_rate": 7.6789198847344e-06,
      "loss": 0.4283,
      "num_tokens": 825498810.0,
      "step": 871
    },
    {
      "epoch": 1.7071498530852107,
      "grad_norm": 0.08978704569061438,
      "learning_rate": 7.673565393120056e-06,
      "loss": 0.4191,
      "num_tokens": 826460436.0,
      "step": 872
    },
    {
      "epoch": 1.7091087169441725,
      "grad_norm": 0.09005628935518344,
      "learning_rate": 7.668206604282687e-06,
      "loss": 0.4438,
      "num_tokens": 827426291.0,
      "step": 873
    },
    {
      "epoch": 1.7110675808031341,
      "grad_norm": 0.09040088145401211,
      "learning_rate": 7.662843526835475e-06,
      "loss": 0.4467,
      "num_tokens": 828379720.0,
      "step": 874
    },
    {
      "epoch": 1.713026444662096,
      "grad_norm": 0.08774775733491094,
      "learning_rate": 7.657476169398497e-06,
      "loss": 0.4198,
      "num_tokens": 829353691.0,
      "step": 875
    },
    {
      "epoch": 1.7149853085210578,
      "grad_norm": 0.0859301718564589,
      "learning_rate": 7.652104540598712e-06,
      "loss": 0.4294,
      "num_tokens": 830315563.0,
      "step": 876
    },
    {
      "epoch": 1.7169441723800196,
      "grad_norm": 0.08690655564830525,
      "learning_rate": 7.64672864906994e-06,
      "loss": 0.4319,
      "num_tokens": 831294006.0,
      "step": 877
    },
    {
      "epoch": 1.7189030362389814,
      "grad_norm": 0.0870783563945675,
      "learning_rate": 7.641348503452855e-06,
      "loss": 0.4221,
      "num_tokens": 832217236.0,
      "step": 878
    },
    {
      "epoch": 1.720861900097943,
      "grad_norm": 0.09367961873370947,
      "learning_rate": 7.635964112394966e-06,
      "loss": 0.4339,
      "num_tokens": 833159623.0,
      "step": 879
    },
    {
      "epoch": 1.7228207639569049,
      "grad_norm": 0.08801713160713069,
      "learning_rate": 7.630575484550611e-06,
      "loss": 0.4357,
      "num_tokens": 834099714.0,
      "step": 880
    },
    {
      "epoch": 1.7247796278158667,
      "grad_norm": 0.0867577050607559,
      "learning_rate": 7.6251826285809294e-06,
      "loss": 0.4378,
      "num_tokens": 835025020.0,
      "step": 881
    },
    {
      "epoch": 1.7267384916748285,
      "grad_norm": 0.08771599581109289,
      "learning_rate": 7.619785553153864e-06,
      "loss": 0.4341,
      "num_tokens": 835976099.0,
      "step": 882
    },
    {
      "epoch": 1.7286973555337903,
      "grad_norm": 0.09010045775355124,
      "learning_rate": 7.614384266944139e-06,
      "loss": 0.4338,
      "num_tokens": 836942966.0,
      "step": 883
    },
    {
      "epoch": 1.7306562193927522,
      "grad_norm": 0.08569022958178449,
      "learning_rate": 7.60897877863324e-06,
      "loss": 0.4356,
      "num_tokens": 837878994.0,
      "step": 884
    },
    {
      "epoch": 1.732615083251714,
      "grad_norm": 0.09054588136575797,
      "learning_rate": 7.603569096909414e-06,
      "loss": 0.4407,
      "num_tokens": 838812064.0,
      "step": 885
    },
    {
      "epoch": 1.7345739471106758,
      "grad_norm": 0.08455869017277894,
      "learning_rate": 7.598155230467645e-06,
      "loss": 0.4359,
      "num_tokens": 839761821.0,
      "step": 886
    },
    {
      "epoch": 1.7365328109696376,
      "grad_norm": 0.08910208308129197,
      "learning_rate": 7.592737188009643e-06,
      "loss": 0.4277,
      "num_tokens": 840698534.0,
      "step": 887
    },
    {
      "epoch": 1.7384916748285995,
      "grad_norm": 0.08419612378419906,
      "learning_rate": 7.5873149782438295e-06,
      "loss": 0.4428,
      "num_tokens": 841661118.0,
      "step": 888
    },
    {
      "epoch": 1.7404505386875613,
      "grad_norm": 0.09286261300836257,
      "learning_rate": 7.581888609885328e-06,
      "loss": 0.4477,
      "num_tokens": 842610772.0,
      "step": 889
    },
    {
      "epoch": 1.742409402546523,
      "grad_norm": 0.08567682262627445,
      "learning_rate": 7.5764580916559405e-06,
      "loss": 0.4219,
      "num_tokens": 843557703.0,
      "step": 890
    },
    {
      "epoch": 1.744368266405485,
      "grad_norm": 0.0874265045179033,
      "learning_rate": 7.571023432284142e-06,
      "loss": 0.4325,
      "num_tokens": 844529841.0,
      "step": 891
    },
    {
      "epoch": 1.7463271302644467,
      "grad_norm": 0.08527093511593209,
      "learning_rate": 7.565584640505061e-06,
      "loss": 0.4238,
      "num_tokens": 845454092.0,
      "step": 892
    },
    {
      "epoch": 1.7482859941234086,
      "grad_norm": 0.08786028916442253,
      "learning_rate": 7.560141725060477e-06,
      "loss": 0.4541,
      "num_tokens": 846388019.0,
      "step": 893
    },
    {
      "epoch": 1.7502448579823702,
      "grad_norm": 0.08810151204616025,
      "learning_rate": 7.554694694698783e-06,
      "loss": 0.4285,
      "num_tokens": 847341800.0,
      "step": 894
    },
    {
      "epoch": 1.752203721841332,
      "grad_norm": 0.090679696767153,
      "learning_rate": 7.549243558174999e-06,
      "loss": 0.4369,
      "num_tokens": 848297768.0,
      "step": 895
    },
    {
      "epoch": 1.7541625857002938,
      "grad_norm": 0.08945682012956972,
      "learning_rate": 7.543788324250738e-06,
      "loss": 0.4314,
      "num_tokens": 849233737.0,
      "step": 896
    },
    {
      "epoch": 1.7561214495592556,
      "grad_norm": 0.09006761166760972,
      "learning_rate": 7.5383290016942e-06,
      "loss": 0.4375,
      "num_tokens": 850207916.0,
      "step": 897
    },
    {
      "epoch": 1.7580803134182175,
      "grad_norm": 0.09358780555112266,
      "learning_rate": 7.532865599280157e-06,
      "loss": 0.4489,
      "num_tokens": 851122512.0,
      "step": 898
    },
    {
      "epoch": 1.7600391772771793,
      "grad_norm": 0.09016876350666427,
      "learning_rate": 7.5273981257899395e-06,
      "loss": 0.4327,
      "num_tokens": 852071033.0,
      "step": 899
    },
    {
      "epoch": 1.761998041136141,
      "grad_norm": 0.0915091856625939,
      "learning_rate": 7.521926590011419e-06,
      "loss": 0.4401,
      "num_tokens": 853014004.0,
      "step": 900
    },
    {
      "epoch": 1.7639569049951027,
      "grad_norm": 0.08645157239780024,
      "learning_rate": 7.516451000738998e-06,
      "loss": 0.4423,
      "num_tokens": 853945279.0,
      "step": 901
    },
    {
      "epoch": 1.7659157688540645,
      "grad_norm": 0.09315232618093792,
      "learning_rate": 7.5109713667735955e-06,
      "loss": 0.4327,
      "num_tokens": 854889767.0,
      "step": 902
    },
    {
      "epoch": 1.7678746327130264,
      "grad_norm": 0.08294235251717945,
      "learning_rate": 7.505487696922629e-06,
      "loss": 0.4306,
      "num_tokens": 855875946.0,
      "step": 903
    },
    {
      "epoch": 1.7698334965719882,
      "grad_norm": 0.08990498197963341,
      "learning_rate": 7.500000000000001e-06,
      "loss": 0.4272,
      "num_tokens": 856838276.0,
      "step": 904
    },
    {
      "epoch": 1.77179236043095,
      "grad_norm": 0.08574846062900512,
      "learning_rate": 7.494508284826093e-06,
      "loss": 0.4289,
      "num_tokens": 857779416.0,
      "step": 905
    },
    {
      "epoch": 1.7737512242899118,
      "grad_norm": 0.08718139719809676,
      "learning_rate": 7.489012560227742e-06,
      "loss": 0.4408,
      "num_tokens": 858742500.0,
      "step": 906
    },
    {
      "epoch": 1.7757100881488737,
      "grad_norm": 0.09264817251194245,
      "learning_rate": 7.483512835038227e-06,
      "loss": 0.4392,
      "num_tokens": 859679580.0,
      "step": 907
    },
    {
      "epoch": 1.7776689520078355,
      "grad_norm": 0.0928835377010871,
      "learning_rate": 7.47800911809726e-06,
      "loss": 0.4439,
      "num_tokens": 860614797.0,
      "step": 908
    },
    {
      "epoch": 1.7796278158667973,
      "grad_norm": 0.08645145576581885,
      "learning_rate": 7.472501418250968e-06,
      "loss": 0.4277,
      "num_tokens": 861590111.0,
      "step": 909
    },
    {
      "epoch": 1.7815866797257591,
      "grad_norm": 0.08750451884708292,
      "learning_rate": 7.46698974435188e-06,
      "loss": 0.4462,
      "num_tokens": 862532036.0,
      "step": 910
    },
    {
      "epoch": 1.783545543584721,
      "grad_norm": 0.09262248516729513,
      "learning_rate": 7.461474105258911e-06,
      "loss": 0.4351,
      "num_tokens": 863526420.0,
      "step": 911
    },
    {
      "epoch": 1.7855044074436828,
      "grad_norm": 0.08474017515659142,
      "learning_rate": 7.455954509837352e-06,
      "loss": 0.4121,
      "num_tokens": 864448770.0,
      "step": 912
    },
    {
      "epoch": 1.7874632713026446,
      "grad_norm": 0.093179644270419,
      "learning_rate": 7.450430966958851e-06,
      "loss": 0.446,
      "num_tokens": 865388472.0,
      "step": 913
    },
    {
      "epoch": 1.7894221351616064,
      "grad_norm": 0.09517537990548135,
      "learning_rate": 7.444903485501402e-06,
      "loss": 0.4477,
      "num_tokens": 866354253.0,
      "step": 914
    },
    {
      "epoch": 1.791380999020568,
      "grad_norm": 0.08913437808815497,
      "learning_rate": 7.439372074349328e-06,
      "loss": 0.4186,
      "num_tokens": 867334882.0,
      "step": 915
    },
    {
      "epoch": 1.7933398628795298,
      "grad_norm": 0.09609945805750214,
      "learning_rate": 7.43383674239327e-06,
      "loss": 0.4316,
      "num_tokens": 868257353.0,
      "step": 916
    },
    {
      "epoch": 1.7952987267384917,
      "grad_norm": 0.08434468534837705,
      "learning_rate": 7.428297498530168e-06,
      "loss": 0.4165,
      "num_tokens": 869172717.0,
      "step": 917
    },
    {
      "epoch": 1.7972575905974535,
      "grad_norm": 0.08753164647695391,
      "learning_rate": 7.422754351663252e-06,
      "loss": 0.4342,
      "num_tokens": 870135639.0,
      "step": 918
    },
    {
      "epoch": 1.7992164544564153,
      "grad_norm": 0.08935410902352021,
      "learning_rate": 7.417207310702025e-06,
      "loss": 0.4452,
      "num_tokens": 871098727.0,
      "step": 919
    },
    {
      "epoch": 1.801175318315377,
      "grad_norm": 0.09277493809398578,
      "learning_rate": 7.411656384562251e-06,
      "loss": 0.451,
      "num_tokens": 872054169.0,
      "step": 920
    },
    {
      "epoch": 1.8031341821743387,
      "grad_norm": 0.09091615700569183,
      "learning_rate": 7.406101582165933e-06,
      "loss": 0.4239,
      "num_tokens": 872998557.0,
      "step": 921
    },
    {
      "epoch": 1.8050930460333006,
      "grad_norm": 0.08686796941755942,
      "learning_rate": 7.400542912441309e-06,
      "loss": 0.4342,
      "num_tokens": 873932558.0,
      "step": 922
    },
    {
      "epoch": 1.8070519098922624,
      "grad_norm": 0.09195894264684597,
      "learning_rate": 7.394980384322834e-06,
      "loss": 0.4262,
      "num_tokens": 874903980.0,
      "step": 923
    },
    {
      "epoch": 1.8090107737512242,
      "grad_norm": 0.08911363486327942,
      "learning_rate": 7.389414006751159e-06,
      "loss": 0.4321,
      "num_tokens": 875848643.0,
      "step": 924
    },
    {
      "epoch": 1.810969637610186,
      "grad_norm": 0.09332062468254551,
      "learning_rate": 7.3838437886731264e-06,
      "loss": 0.4314,
      "num_tokens": 876794234.0,
      "step": 925
    },
    {
      "epoch": 1.8129285014691479,
      "grad_norm": 0.09478438211278445,
      "learning_rate": 7.378269739041751e-06,
      "loss": 0.4394,
      "num_tokens": 877747088.0,
      "step": 926
    },
    {
      "epoch": 1.8148873653281097,
      "grad_norm": 0.08971306555226388,
      "learning_rate": 7.3726918668162065e-06,
      "loss": 0.4485,
      "num_tokens": 878707539.0,
      "step": 927
    },
    {
      "epoch": 1.8168462291870715,
      "grad_norm": 0.08569446482743738,
      "learning_rate": 7.367110180961807e-06,
      "loss": 0.438,
      "num_tokens": 879651206.0,
      "step": 928
    },
    {
      "epoch": 1.8188050930460333,
      "grad_norm": 0.09119524659336786,
      "learning_rate": 7.3615246904500025e-06,
      "loss": 0.4192,
      "num_tokens": 880563698.0,
      "step": 929
    },
    {
      "epoch": 1.8207639569049952,
      "grad_norm": 0.0889603261333572,
      "learning_rate": 7.355935404258354e-06,
      "loss": 0.4325,
      "num_tokens": 881515407.0,
      "step": 930
    },
    {
      "epoch": 1.822722820763957,
      "grad_norm": 0.09026967037671438,
      "learning_rate": 7.350342331370521e-06,
      "loss": 0.4227,
      "num_tokens": 882453009.0,
      "step": 931
    },
    {
      "epoch": 1.8246816846229188,
      "grad_norm": 0.08652547878079349,
      "learning_rate": 7.3447454807762565e-06,
      "loss": 0.44,
      "num_tokens": 883449631.0,
      "step": 932
    },
    {
      "epoch": 1.8266405484818806,
      "grad_norm": 0.09162557590468148,
      "learning_rate": 7.339144861471379e-06,
      "loss": 0.4482,
      "num_tokens": 884419434.0,
      "step": 933
    },
    {
      "epoch": 1.8285994123408424,
      "grad_norm": 0.0938185836612543,
      "learning_rate": 7.333540482457766e-06,
      "loss": 0.4352,
      "num_tokens": 885398978.0,
      "step": 934
    },
    {
      "epoch": 1.8305582761998043,
      "grad_norm": 0.08630327514659179,
      "learning_rate": 7.327932352743341e-06,
      "loss": 0.4315,
      "num_tokens": 886348605.0,
      "step": 935
    },
    {
      "epoch": 1.8325171400587659,
      "grad_norm": 0.08570655142834177,
      "learning_rate": 7.322320481342053e-06,
      "loss": 0.41,
      "num_tokens": 887284913.0,
      "step": 936
    },
    {
      "epoch": 1.8344760039177277,
      "grad_norm": 0.08377633482101021,
      "learning_rate": 7.316704877273866e-06,
      "loss": 0.4266,
      "num_tokens": 888208855.0,
      "step": 937
    },
    {
      "epoch": 1.8364348677766895,
      "grad_norm": 0.08713332313204401,
      "learning_rate": 7.311085549564742e-06,
      "loss": 0.4228,
      "num_tokens": 889171102.0,
      "step": 938
    },
    {
      "epoch": 1.8383937316356513,
      "grad_norm": 0.10240927862922675,
      "learning_rate": 7.30546250724663e-06,
      "loss": 0.4324,
      "num_tokens": 890162092.0,
      "step": 939
    },
    {
      "epoch": 1.8403525954946132,
      "grad_norm": 0.09400164210750693,
      "learning_rate": 7.2998357593574485e-06,
      "loss": 0.4254,
      "num_tokens": 891131622.0,
      "step": 940
    },
    {
      "epoch": 1.8423114593535748,
      "grad_norm": 0.08731919697434702,
      "learning_rate": 7.294205314941074e-06,
      "loss": 0.4372,
      "num_tokens": 892096819.0,
      "step": 941
    },
    {
      "epoch": 1.8442703232125366,
      "grad_norm": 0.08865328013583497,
      "learning_rate": 7.288571183047321e-06,
      "loss": 0.4288,
      "num_tokens": 893011990.0,
      "step": 942
    },
    {
      "epoch": 1.8462291870714984,
      "grad_norm": 0.09348427289631799,
      "learning_rate": 7.282933372731933e-06,
      "loss": 0.4368,
      "num_tokens": 893946270.0,
      "step": 943
    },
    {
      "epoch": 1.8481880509304602,
      "grad_norm": 0.09244859490667034,
      "learning_rate": 7.277291893056565e-06,
      "loss": 0.4433,
      "num_tokens": 894870549.0,
      "step": 944
    },
    {
      "epoch": 1.850146914789422,
      "grad_norm": 0.0863169278489715,
      "learning_rate": 7.27164675308877e-06,
      "loss": 0.4366,
      "num_tokens": 895834311.0,
      "step": 945
    },
    {
      "epoch": 1.8521057786483839,
      "grad_norm": 0.08465771790873794,
      "learning_rate": 7.265997961901987e-06,
      "loss": 0.4441,
      "num_tokens": 896791695.0,
      "step": 946
    },
    {
      "epoch": 1.8540646425073457,
      "grad_norm": 0.08650374600013917,
      "learning_rate": 7.260345528575518e-06,
      "loss": 0.4229,
      "num_tokens": 897749765.0,
      "step": 947
    },
    {
      "epoch": 1.8560235063663075,
      "grad_norm": 0.0874449822383469,
      "learning_rate": 7.254689462194522e-06,
      "loss": 0.4355,
      "num_tokens": 898730878.0,
      "step": 948
    },
    {
      "epoch": 1.8579823702252694,
      "grad_norm": 0.09010161028563324,
      "learning_rate": 7.249029771849999e-06,
      "loss": 0.4204,
      "num_tokens": 899652213.0,
      "step": 949
    },
    {
      "epoch": 1.8599412340842312,
      "grad_norm": 0.0857329910328507,
      "learning_rate": 7.243366466638771e-06,
      "loss": 0.4296,
      "num_tokens": 900622061.0,
      "step": 950
    },
    {
      "epoch": 1.861900097943193,
      "grad_norm": 0.08716299164988617,
      "learning_rate": 7.237699555663472e-06,
      "loss": 0.4536,
      "num_tokens": 901545497.0,
      "step": 951
    },
    {
      "epoch": 1.8638589618021548,
      "grad_norm": 0.09347516841706188,
      "learning_rate": 7.232029048032532e-06,
      "loss": 0.4291,
      "num_tokens": 902490485.0,
      "step": 952
    },
    {
      "epoch": 1.8658178256611166,
      "grad_norm": 0.0854290686625699,
      "learning_rate": 7.226354952860157e-06,
      "loss": 0.4158,
      "num_tokens": 903429848.0,
      "step": 953
    },
    {
      "epoch": 1.8677766895200785,
      "grad_norm": 0.09123677673598038,
      "learning_rate": 7.220677279266327e-06,
      "loss": 0.4321,
      "num_tokens": 904397266.0,
      "step": 954
    },
    {
      "epoch": 1.8697355533790403,
      "grad_norm": 0.0875787526272707,
      "learning_rate": 7.214996036376769e-06,
      "loss": 0.4303,
      "num_tokens": 905341373.0,
      "step": 955
    },
    {
      "epoch": 1.8716944172380021,
      "grad_norm": 0.09385489528693176,
      "learning_rate": 7.2093112333229476e-06,
      "loss": 0.4367,
      "num_tokens": 906287476.0,
      "step": 956
    },
    {
      "epoch": 1.8736532810969637,
      "grad_norm": 0.08579907833807023,
      "learning_rate": 7.203622879242047e-06,
      "loss": 0.4205,
      "num_tokens": 907208011.0,
      "step": 957
    },
    {
      "epoch": 1.8756121449559255,
      "grad_norm": 0.08982081921348842,
      "learning_rate": 7.197930983276964e-06,
      "loss": 0.4565,
      "num_tokens": 908157808.0,
      "step": 958
    },
    {
      "epoch": 1.8775710088148874,
      "grad_norm": 0.09428717961917073,
      "learning_rate": 7.192235554576287e-06,
      "loss": 0.4298,
      "num_tokens": 909123454.0,
      "step": 959
    },
    {
      "epoch": 1.8795298726738492,
      "grad_norm": 0.08813977535803645,
      "learning_rate": 7.186536602294278e-06,
      "loss": 0.4298,
      "num_tokens": 910062164.0,
      "step": 960
    },
    {
      "epoch": 1.881488736532811,
      "grad_norm": 0.08709066569776673,
      "learning_rate": 7.180834135590869e-06,
      "loss": 0.4314,
      "num_tokens": 911009601.0,
      "step": 961
    },
    {
      "epoch": 1.8834476003917726,
      "grad_norm": 0.08719550875015987,
      "learning_rate": 7.175128163631636e-06,
      "loss": 0.4279,
      "num_tokens": 911954589.0,
      "step": 962
    },
    {
      "epoch": 1.8854064642507344,
      "grad_norm": 0.08836589818991197,
      "learning_rate": 7.169418695587791e-06,
      "loss": 0.4215,
      "num_tokens": 912889182.0,
      "step": 963
    },
    {
      "epoch": 1.8873653281096963,
      "grad_norm": 0.09352451611567275,
      "learning_rate": 7.1637057406361656e-06,
      "loss": 0.4474,
      "num_tokens": 913825514.0,
      "step": 964
    },
    {
      "epoch": 1.889324191968658,
      "grad_norm": 0.09249184003628412,
      "learning_rate": 7.1579893079591955e-06,
      "loss": 0.4419,
      "num_tokens": 914762452.0,
      "step": 965
    },
    {
      "epoch": 1.89128305582762,
      "grad_norm": 0.0883449220618962,
      "learning_rate": 7.152269406744904e-06,
      "loss": 0.4349,
      "num_tokens": 915690159.0,
      "step": 966
    },
    {
      "epoch": 1.8932419196865817,
      "grad_norm": 0.09220825335778882,
      "learning_rate": 7.146546046186893e-06,
      "loss": 0.4296,
      "num_tokens": 916610333.0,
      "step": 967
    },
    {
      "epoch": 1.8952007835455436,
      "grad_norm": 0.0872142532151571,
      "learning_rate": 7.140819235484322e-06,
      "loss": 0.4318,
      "num_tokens": 917540458.0,
      "step": 968
    },
    {
      "epoch": 1.8971596474045054,
      "grad_norm": 0.08882587423699995,
      "learning_rate": 7.135088983841898e-06,
      "loss": 0.4401,
      "num_tokens": 918487705.0,
      "step": 969
    },
    {
      "epoch": 1.8991185112634672,
      "grad_norm": 0.0936949575425075,
      "learning_rate": 7.129355300469854e-06,
      "loss": 0.43,
      "num_tokens": 919431101.0,
      "step": 970
    },
    {
      "epoch": 1.901077375122429,
      "grad_norm": 0.08974361619962315,
      "learning_rate": 7.123618194583947e-06,
      "loss": 0.4412,
      "num_tokens": 920379795.0,
      "step": 971
    },
    {
      "epoch": 1.9030362389813908,
      "grad_norm": 0.08792011295575772,
      "learning_rate": 7.117877675405427e-06,
      "loss": 0.4192,
      "num_tokens": 921314131.0,
      "step": 972
    },
    {
      "epoch": 1.9049951028403527,
      "grad_norm": 0.08528632260883838,
      "learning_rate": 7.112133752161034e-06,
      "loss": 0.4341,
      "num_tokens": 922226945.0,
      "step": 973
    },
    {
      "epoch": 1.9069539666993145,
      "grad_norm": 0.09026828044995538,
      "learning_rate": 7.106386434082979e-06,
      "loss": 0.4387,
      "num_tokens": 923195804.0,
      "step": 974
    },
    {
      "epoch": 1.9089128305582763,
      "grad_norm": 0.08532245075495483,
      "learning_rate": 7.10063573040893e-06,
      "loss": 0.432,
      "num_tokens": 924164462.0,
      "step": 975
    },
    {
      "epoch": 1.9108716944172381,
      "grad_norm": 0.08838154486612344,
      "learning_rate": 7.094881650381995e-06,
      "loss": 0.4166,
      "num_tokens": 925095845.0,
      "step": 976
    },
    {
      "epoch": 1.9128305582762,
      "grad_norm": 0.0894884729283402,
      "learning_rate": 7.089124203250709e-06,
      "loss": 0.4278,
      "num_tokens": 926021976.0,
      "step": 977
    },
    {
      "epoch": 1.9147894221351616,
      "grad_norm": 0.08564215618334733,
      "learning_rate": 7.083363398269022e-06,
      "loss": 0.4186,
      "num_tokens": 926979607.0,
      "step": 978
    },
    {
      "epoch": 1.9167482859941234,
      "grad_norm": 0.08583123541938284,
      "learning_rate": 7.077599244696278e-06,
      "loss": 0.4208,
      "num_tokens": 927957218.0,
      "step": 979
    },
    {
      "epoch": 1.9187071498530852,
      "grad_norm": 0.08357222255382876,
      "learning_rate": 7.071831751797202e-06,
      "loss": 0.4316,
      "num_tokens": 928925433.0,
      "step": 980
    },
    {
      "epoch": 1.920666013712047,
      "grad_norm": 0.08632462955622228,
      "learning_rate": 7.066060928841891e-06,
      "loss": 0.4249,
      "num_tokens": 929875395.0,
      "step": 981
    },
    {
      "epoch": 1.9226248775710089,
      "grad_norm": 0.08457636543108527,
      "learning_rate": 7.0602867851057925e-06,
      "loss": 0.4123,
      "num_tokens": 930807456.0,
      "step": 982
    },
    {
      "epoch": 1.9245837414299705,
      "grad_norm": 0.08805920979356217,
      "learning_rate": 7.054509329869685e-06,
      "loss": 0.4356,
      "num_tokens": 931763590.0,
      "step": 983
    },
    {
      "epoch": 1.9265426052889323,
      "grad_norm": 0.08623461210441301,
      "learning_rate": 7.048728572419681e-06,
      "loss": 0.4337,
      "num_tokens": 932714866.0,
      "step": 984
    },
    {
      "epoch": 1.928501469147894,
      "grad_norm": 0.0846789684294999,
      "learning_rate": 7.042944522047193e-06,
      "loss": 0.429,
      "num_tokens": 933676337.0,
      "step": 985
    },
    {
      "epoch": 1.930460333006856,
      "grad_norm": 0.08941298064535665,
      "learning_rate": 7.0371571880489274e-06,
      "loss": 0.43,
      "num_tokens": 934627135.0,
      "step": 986
    },
    {
      "epoch": 1.9324191968658178,
      "grad_norm": 0.08647750253606112,
      "learning_rate": 7.0313665797268704e-06,
      "loss": 0.4406,
      "num_tokens": 935574842.0,
      "step": 987
    },
    {
      "epoch": 1.9343780607247796,
      "grad_norm": 0.08733646879164099,
      "learning_rate": 7.025572706388268e-06,
      "loss": 0.4437,
      "num_tokens": 936533645.0,
      "step": 988
    },
    {
      "epoch": 1.9363369245837414,
      "grad_norm": 0.08713719779617433,
      "learning_rate": 7.019775577345618e-06,
      "loss": 0.4293,
      "num_tokens": 937489716.0,
      "step": 989
    },
    {
      "epoch": 1.9382957884427032,
      "grad_norm": 0.08620697776474966,
      "learning_rate": 7.0139752019166474e-06,
      "loss": 0.4166,
      "num_tokens": 938462701.0,
      "step": 990
    },
    {
      "epoch": 1.940254652301665,
      "grad_norm": 0.08687060493220213,
      "learning_rate": 7.008171589424302e-06,
      "loss": 0.4254,
      "num_tokens": 939428068.0,
      "step": 991
    },
    {
      "epoch": 1.9422135161606269,
      "grad_norm": 0.08367694885617118,
      "learning_rate": 7.0023647491967336e-06,
      "loss": 0.4191,
      "num_tokens": 940380046.0,
      "step": 992
    },
    {
      "epoch": 1.9441723800195887,
      "grad_norm": 0.08779227586016532,
      "learning_rate": 6.996554690567276e-06,
      "loss": 0.4353,
      "num_tokens": 941292759.0,
      "step": 993
    },
    {
      "epoch": 1.9461312438785505,
      "grad_norm": 0.09292159580303222,
      "learning_rate": 6.990741422874443e-06,
      "loss": 0.441,
      "num_tokens": 942242549.0,
      "step": 994
    },
    {
      "epoch": 1.9480901077375123,
      "grad_norm": 0.0834052261589026,
      "learning_rate": 6.984924955461901e-06,
      "loss": 0.4196,
      "num_tokens": 943203026.0,
      "step": 995
    },
    {
      "epoch": 1.9500489715964742,
      "grad_norm": 0.08523032764682975,
      "learning_rate": 6.979105297678462e-06,
      "loss": 0.4438,
      "num_tokens": 944179700.0,
      "step": 996
    },
    {
      "epoch": 1.952007835455436,
      "grad_norm": 0.09110533923101696,
      "learning_rate": 6.9732824588780635e-06,
      "loss": 0.44,
      "num_tokens": 945138832.0,
      "step": 997
    },
    {
      "epoch": 1.9539666993143978,
      "grad_norm": 0.08650261050569312,
      "learning_rate": 6.967456448419756e-06,
      "loss": 0.4327,
      "num_tokens": 946133903.0,
      "step": 998
    },
    {
      "epoch": 1.9559255631733594,
      "grad_norm": 0.08491087154817847,
      "learning_rate": 6.961627275667693e-06,
      "loss": 0.4222,
      "num_tokens": 947096770.0,
      "step": 999
    },
    {
      "epoch": 1.9578844270323212,
      "grad_norm": 0.08281487740287748,
      "learning_rate": 6.955794949991104e-06,
      "loss": 0.438,
      "num_tokens": 948045301.0,
      "step": 1000
    },
    {
      "epoch": 1.959843290891283,
      "grad_norm": 0.08839472335258977,
      "learning_rate": 6.9499594807642905e-06,
      "loss": 0.4351,
      "num_tokens": 949014144.0,
      "step": 1001
    },
    {
      "epoch": 1.9618021547502449,
      "grad_norm": 0.08677248282074429,
      "learning_rate": 6.944120877366605e-06,
      "loss": 0.4323,
      "num_tokens": 949983103.0,
      "step": 1002
    },
    {
      "epoch": 1.9637610186092067,
      "grad_norm": 0.08765327355089926,
      "learning_rate": 6.938279149182435e-06,
      "loss": 0.4335,
      "num_tokens": 950960048.0,
      "step": 1003
    },
    {
      "epoch": 1.9657198824681683,
      "grad_norm": 0.08593327105152829,
      "learning_rate": 6.932434305601197e-06,
      "loss": 0.4252,
      "num_tokens": 951887650.0,
      "step": 1004
    },
    {
      "epoch": 1.9676787463271301,
      "grad_norm": 0.08324228774379007,
      "learning_rate": 6.92658635601731e-06,
      "loss": 0.413,
      "num_tokens": 952856948.0,
      "step": 1005
    },
    {
      "epoch": 1.969637610186092,
      "grad_norm": 0.08417073884935165,
      "learning_rate": 6.920735309830184e-06,
      "loss": 0.4276,
      "num_tokens": 953790803.0,
      "step": 1006
    },
    {
      "epoch": 1.9715964740450538,
      "grad_norm": 0.08310230438234872,
      "learning_rate": 6.914881176444213e-06,
      "loss": 0.4334,
      "num_tokens": 954734145.0,
      "step": 1007
    },
    {
      "epoch": 1.9735553379040156,
      "grad_norm": 0.08890830673621009,
      "learning_rate": 6.909023965268746e-06,
      "loss": 0.4232,
      "num_tokens": 955704667.0,
      "step": 1008
    },
    {
      "epoch": 1.9755142017629774,
      "grad_norm": 0.0832389330871157,
      "learning_rate": 6.9031636857180795e-06,
      "loss": 0.4067,
      "num_tokens": 956644147.0,
      "step": 1009
    },
    {
      "epoch": 1.9774730656219393,
      "grad_norm": 0.08722851640571877,
      "learning_rate": 6.897300347211446e-06,
      "loss": 0.4313,
      "num_tokens": 957601288.0,
      "step": 1010
    },
    {
      "epoch": 1.979431929480901,
      "grad_norm": 0.09022898911624623,
      "learning_rate": 6.891433959172994e-06,
      "loss": 0.4287,
      "num_tokens": 958584024.0,
      "step": 1011
    },
    {
      "epoch": 1.981390793339863,
      "grad_norm": 0.08985029693519823,
      "learning_rate": 6.88556453103177e-06,
      "loss": 0.4312,
      "num_tokens": 959535098.0,
      "step": 1012
    },
    {
      "epoch": 1.9833496571988247,
      "grad_norm": 0.08472538395205044,
      "learning_rate": 6.879692072221709e-06,
      "loss": 0.416,
      "num_tokens": 960475278.0,
      "step": 1013
    },
    {
      "epoch": 1.9853085210577865,
      "grad_norm": 0.08796363623101505,
      "learning_rate": 6.873816592181617e-06,
      "loss": 0.4368,
      "num_tokens": 961396426.0,
      "step": 1014
    },
    {
      "epoch": 1.9872673849167484,
      "grad_norm": 0.09073188833185505,
      "learning_rate": 6.867938100355157e-06,
      "loss": 0.434,
      "num_tokens": 962325943.0,
      "step": 1015
    },
    {
      "epoch": 1.9892262487757102,
      "grad_norm": 0.08785016560910433,
      "learning_rate": 6.86205660619083e-06,
      "loss": 0.4367,
      "num_tokens": 963289686.0,
      "step": 1016
    },
    {
      "epoch": 1.991185112634672,
      "grad_norm": 0.08790596218123922,
      "learning_rate": 6.856172119141966e-06,
      "loss": 0.435,
      "num_tokens": 964244718.0,
      "step": 1017
    },
    {
      "epoch": 1.9931439764936338,
      "grad_norm": 0.08641718763995186,
      "learning_rate": 6.850284648666706e-06,
      "loss": 0.4283,
      "num_tokens": 965189309.0,
      "step": 1018
    },
    {
      "epoch": 1.9951028403525954,
      "grad_norm": 0.08831923186959566,
      "learning_rate": 6.844394204227981e-06,
      "loss": 0.4347,
      "num_tokens": 966118494.0,
      "step": 1019
    },
    {
      "epoch": 1.9970617042115573,
      "grad_norm": 0.08785677300124596,
      "learning_rate": 6.838500795293506e-06,
      "loss": 0.4374,
      "num_tokens": 967063563.0,
      "step": 1020
    },
    {
      "epoch": 1.999020568070519,
      "grad_norm": 0.09001442137711776,
      "learning_rate": 6.832604431335762e-06,
      "loss": 0.4204,
      "num_tokens": 968000355.0,
      "step": 1021
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.13990886438124683,
      "learning_rate": 6.8267051218319766e-06,
      "loss": 0.4285,
      "num_tokens": 968473493.0,
      "step": 1022
    },
    {
      "epoch": 2.001958863858962,
      "grad_norm": 0.09042271698661719,
      "learning_rate": 6.820802876264112e-06,
      "loss": 0.4339,
      "num_tokens": 969409433.0,
      "step": 1023
    },
    {
      "epoch": 2.0039177277179236,
      "grad_norm": 0.0894873000540885,
      "learning_rate": 6.814897704118851e-06,
      "loss": 0.445,
      "num_tokens": 970344157.0,
      "step": 1024
    },
    {
      "epoch": 2.0058765915768855,
      "grad_norm": 0.08828463646358492,
      "learning_rate": 6.808989614887582e-06,
      "loss": 0.4254,
      "num_tokens": 971264637.0,
      "step": 1025
    },
    {
      "epoch": 2.0078354554358473,
      "grad_norm": 0.08863652952689459,
      "learning_rate": 6.803078618066378e-06,
      "loss": 0.4221,
      "num_tokens": 972185363.0,
      "step": 1026
    },
    {
      "epoch": 2.009794319294809,
      "grad_norm": 0.0869582376314397,
      "learning_rate": 6.797164723155987e-06,
      "loss": 0.4186,
      "num_tokens": 973135478.0,
      "step": 1027
    },
    {
      "epoch": 2.011753183153771,
      "grad_norm": 0.0912341087882137,
      "learning_rate": 6.7912479396618155e-06,
      "loss": 0.4339,
      "num_tokens": 974087860.0,
      "step": 1028
    },
    {
      "epoch": 2.0137120470127328,
      "grad_norm": 0.08879814060006445,
      "learning_rate": 6.785328277093912e-06,
      "loss": 0.4243,
      "num_tokens": 975022047.0,
      "step": 1029
    },
    {
      "epoch": 2.0156709108716946,
      "grad_norm": 0.09211748606068026,
      "learning_rate": 6.7794057449669545e-06,
      "loss": 0.4426,
      "num_tokens": 975946956.0,
      "step": 1030
    },
    {
      "epoch": 2.0176297747306564,
      "grad_norm": 0.08710182517592917,
      "learning_rate": 6.773480352800234e-06,
      "loss": 0.4266,
      "num_tokens": 976870592.0,
      "step": 1031
    },
    {
      "epoch": 2.0195886385896182,
      "grad_norm": 0.08733356753832412,
      "learning_rate": 6.767552110117631e-06,
      "loss": 0.4175,
      "num_tokens": 977800895.0,
      "step": 1032
    },
    {
      "epoch": 2.0215475024485796,
      "grad_norm": 0.0871460092075375,
      "learning_rate": 6.761621026447616e-06,
      "loss": 0.4147,
      "num_tokens": 978770781.0,
      "step": 1033
    },
    {
      "epoch": 2.0235063663075414,
      "grad_norm": 0.08878069977801947,
      "learning_rate": 6.7556871113232234e-06,
      "loss": 0.4386,
      "num_tokens": 979720092.0,
      "step": 1034
    },
    {
      "epoch": 2.0254652301665033,
      "grad_norm": 0.08835686419539676,
      "learning_rate": 6.749750374282039e-06,
      "loss": 0.4288,
      "num_tokens": 980678968.0,
      "step": 1035
    },
    {
      "epoch": 2.027424094025465,
      "grad_norm": 0.0874628032269614,
      "learning_rate": 6.743810824866179e-06,
      "loss": 0.4125,
      "num_tokens": 981610602.0,
      "step": 1036
    },
    {
      "epoch": 2.029382957884427,
      "grad_norm": 0.08450736471281808,
      "learning_rate": 6.7378684726222875e-06,
      "loss": 0.4068,
      "num_tokens": 982563335.0,
      "step": 1037
    },
    {
      "epoch": 2.0313418217433887,
      "grad_norm": 0.08505868516958862,
      "learning_rate": 6.7319233271015104e-06,
      "loss": 0.4205,
      "num_tokens": 983513105.0,
      "step": 1038
    },
    {
      "epoch": 2.0333006856023506,
      "grad_norm": 0.08730489468212824,
      "learning_rate": 6.725975397859481e-06,
      "loss": 0.4407,
      "num_tokens": 984446753.0,
      "step": 1039
    },
    {
      "epoch": 2.0352595494613124,
      "grad_norm": 0.08587029827647688,
      "learning_rate": 6.720024694456311e-06,
      "loss": 0.4185,
      "num_tokens": 985383770.0,
      "step": 1040
    },
    {
      "epoch": 2.037218413320274,
      "grad_norm": 0.08595039343020731,
      "learning_rate": 6.714071226456568e-06,
      "loss": 0.4273,
      "num_tokens": 986353533.0,
      "step": 1041
    },
    {
      "epoch": 2.039177277179236,
      "grad_norm": 0.08648519850540613,
      "learning_rate": 6.708115003429265e-06,
      "loss": 0.4112,
      "num_tokens": 987291394.0,
      "step": 1042
    },
    {
      "epoch": 2.041136141038198,
      "grad_norm": 0.09070720713842331,
      "learning_rate": 6.702156034947842e-06,
      "loss": 0.4165,
      "num_tokens": 988209521.0,
      "step": 1043
    },
    {
      "epoch": 2.0430950048971597,
      "grad_norm": 0.08720773104648802,
      "learning_rate": 6.6961943305901515e-06,
      "loss": 0.4244,
      "num_tokens": 989143282.0,
      "step": 1044
    },
    {
      "epoch": 2.0450538687561215,
      "grad_norm": 0.08356870633033667,
      "learning_rate": 6.690229899938445e-06,
      "loss": 0.4244,
      "num_tokens": 990087629.0,
      "step": 1045
    },
    {
      "epoch": 2.0470127326150833,
      "grad_norm": 0.0885256763687902,
      "learning_rate": 6.684262752579354e-06,
      "loss": 0.4303,
      "num_tokens": 991029591.0,
      "step": 1046
    },
    {
      "epoch": 2.048971596474045,
      "grad_norm": 0.08845802673876547,
      "learning_rate": 6.678292898103879e-06,
      "loss": 0.422,
      "num_tokens": 991974727.0,
      "step": 1047
    },
    {
      "epoch": 2.050930460333007,
      "grad_norm": 0.08578130284905115,
      "learning_rate": 6.672320346107367e-06,
      "loss": 0.4242,
      "num_tokens": 992917717.0,
      "step": 1048
    },
    {
      "epoch": 2.052889324191969,
      "grad_norm": 0.09364007726543205,
      "learning_rate": 6.666345106189508e-06,
      "loss": 0.4237,
      "num_tokens": 993862862.0,
      "step": 1049
    },
    {
      "epoch": 2.0548481880509306,
      "grad_norm": 0.0860871126577842,
      "learning_rate": 6.660367187954304e-06,
      "loss": 0.4252,
      "num_tokens": 994830885.0,
      "step": 1050
    },
    {
      "epoch": 2.0568070519098924,
      "grad_norm": 0.09066905547094299,
      "learning_rate": 6.65438660101007e-06,
      "loss": 0.4233,
      "num_tokens": 995780263.0,
      "step": 1051
    },
    {
      "epoch": 2.0587659157688543,
      "grad_norm": 0.08451666053149698,
      "learning_rate": 6.648403354969404e-06,
      "loss": 0.4322,
      "num_tokens": 996762312.0,
      "step": 1052
    },
    {
      "epoch": 2.060724779627816,
      "grad_norm": 0.08147946498227737,
      "learning_rate": 6.642417459449181e-06,
      "loss": 0.4147,
      "num_tokens": 997760014.0,
      "step": 1053
    },
    {
      "epoch": 2.0626836434867775,
      "grad_norm": 0.0924097896999832,
      "learning_rate": 6.6364289240705324e-06,
      "loss": 0.4225,
      "num_tokens": 998707194.0,
      "step": 1054
    },
    {
      "epoch": 2.0646425073457393,
      "grad_norm": 0.08880772747352325,
      "learning_rate": 6.630437758458839e-06,
      "loss": 0.4222,
      "num_tokens": 999624872.0,
      "step": 1055
    },
    {
      "epoch": 2.066601371204701,
      "grad_norm": 0.08535151578135977,
      "learning_rate": 6.6244439722436985e-06,
      "loss": 0.4199,
      "num_tokens": 1000555666.0,
      "step": 1056
    },
    {
      "epoch": 2.068560235063663,
      "grad_norm": 0.08645940123807073,
      "learning_rate": 6.618447575058929e-06,
      "loss": 0.4278,
      "num_tokens": 1001482665.0,
      "step": 1057
    },
    {
      "epoch": 2.0705190989226248,
      "grad_norm": 0.08850702822202812,
      "learning_rate": 6.612448576542545e-06,
      "loss": 0.4124,
      "num_tokens": 1002454920.0,
      "step": 1058
    },
    {
      "epoch": 2.0724779627815866,
      "grad_norm": 0.08638920339828096,
      "learning_rate": 6.606446986336736e-06,
      "loss": 0.4184,
      "num_tokens": 1003425275.0,
      "step": 1059
    },
    {
      "epoch": 2.0744368266405484,
      "grad_norm": 0.0908983416828278,
      "learning_rate": 6.600442814087865e-06,
      "loss": 0.426,
      "num_tokens": 1004345342.0,
      "step": 1060
    },
    {
      "epoch": 2.0763956904995102,
      "grad_norm": 0.09052632215917387,
      "learning_rate": 6.594436069446437e-06,
      "loss": 0.4408,
      "num_tokens": 1005315478.0,
      "step": 1061
    },
    {
      "epoch": 2.078354554358472,
      "grad_norm": 0.08250486255633122,
      "learning_rate": 6.588426762067099e-06,
      "loss": 0.4113,
      "num_tokens": 1006245838.0,
      "step": 1062
    },
    {
      "epoch": 2.080313418217434,
      "grad_norm": 0.08852451483879102,
      "learning_rate": 6.582414901608611e-06,
      "loss": 0.4342,
      "num_tokens": 1007229627.0,
      "step": 1063
    },
    {
      "epoch": 2.0822722820763957,
      "grad_norm": 0.09031206410672629,
      "learning_rate": 6.576400497733841e-06,
      "loss": 0.4254,
      "num_tokens": 1008162960.0,
      "step": 1064
    },
    {
      "epoch": 2.0842311459353575,
      "grad_norm": 0.0864045142590658,
      "learning_rate": 6.570383560109745e-06,
      "loss": 0.4293,
      "num_tokens": 1009101606.0,
      "step": 1065
    },
    {
      "epoch": 2.0861900097943193,
      "grad_norm": 0.08922509587318524,
      "learning_rate": 6.564364098407348e-06,
      "loss": 0.434,
      "num_tokens": 1010024213.0,
      "step": 1066
    },
    {
      "epoch": 2.088148873653281,
      "grad_norm": 0.08691060584575491,
      "learning_rate": 6.558342122301733e-06,
      "loss": 0.4147,
      "num_tokens": 1010962414.0,
      "step": 1067
    },
    {
      "epoch": 2.090107737512243,
      "grad_norm": 0.08807462348673281,
      "learning_rate": 6.552317641472027e-06,
      "loss": 0.4292,
      "num_tokens": 1011926691.0,
      "step": 1068
    },
    {
      "epoch": 2.092066601371205,
      "grad_norm": 0.08543679010461865,
      "learning_rate": 6.5462906656013805e-06,
      "loss": 0.4237,
      "num_tokens": 1012871933.0,
      "step": 1069
    },
    {
      "epoch": 2.0940254652301666,
      "grad_norm": 0.08742364903188327,
      "learning_rate": 6.540261204376954e-06,
      "loss": 0.4274,
      "num_tokens": 1013811271.0,
      "step": 1070
    },
    {
      "epoch": 2.0959843290891285,
      "grad_norm": 0.08865922941911121,
      "learning_rate": 6.5342292674899065e-06,
      "loss": 0.4435,
      "num_tokens": 1014750776.0,
      "step": 1071
    },
    {
      "epoch": 2.0979431929480903,
      "grad_norm": 0.08850207818224096,
      "learning_rate": 6.52819486463537e-06,
      "loss": 0.4222,
      "num_tokens": 1015691930.0,
      "step": 1072
    },
    {
      "epoch": 2.099902056807052,
      "grad_norm": 0.08485792660808923,
      "learning_rate": 6.522158005512445e-06,
      "loss": 0.4059,
      "num_tokens": 1016642924.0,
      "step": 1073
    },
    {
      "epoch": 2.101860920666014,
      "grad_norm": 0.09106027897419366,
      "learning_rate": 6.516118699824179e-06,
      "loss": 0.4332,
      "num_tokens": 1017589589.0,
      "step": 1074
    },
    {
      "epoch": 2.1038197845249753,
      "grad_norm": 0.09628235122455635,
      "learning_rate": 6.510076957277549e-06,
      "loss": 0.4236,
      "num_tokens": 1018536454.0,
      "step": 1075
    },
    {
      "epoch": 2.105778648383937,
      "grad_norm": 0.09137519333502027,
      "learning_rate": 6.504032787583452e-06,
      "loss": 0.4332,
      "num_tokens": 1019489810.0,
      "step": 1076
    },
    {
      "epoch": 2.107737512242899,
      "grad_norm": 0.08896344922548269,
      "learning_rate": 6.497986200456684e-06,
      "loss": 0.4229,
      "num_tokens": 1020408403.0,
      "step": 1077
    },
    {
      "epoch": 2.109696376101861,
      "grad_norm": 0.09144395841876914,
      "learning_rate": 6.49193720561593e-06,
      "loss": 0.419,
      "num_tokens": 1021371009.0,
      "step": 1078
    },
    {
      "epoch": 2.1116552399608226,
      "grad_norm": 0.08419306375593151,
      "learning_rate": 6.48588581278374e-06,
      "loss": 0.4208,
      "num_tokens": 1022353024.0,
      "step": 1079
    },
    {
      "epoch": 2.1136141038197844,
      "grad_norm": 0.08908143142604028,
      "learning_rate": 6.479832031686522e-06,
      "loss": 0.4225,
      "num_tokens": 1023330905.0,
      "step": 1080
    },
    {
      "epoch": 2.1155729676787463,
      "grad_norm": 0.09110437348208388,
      "learning_rate": 6.473775872054522e-06,
      "loss": 0.4148,
      "num_tokens": 1024323661.0,
      "step": 1081
    },
    {
      "epoch": 2.117531831537708,
      "grad_norm": 0.08828334528110036,
      "learning_rate": 6.467717343621807e-06,
      "loss": 0.4261,
      "num_tokens": 1025321459.0,
      "step": 1082
    },
    {
      "epoch": 2.11949069539667,
      "grad_norm": 0.08723822032570723,
      "learning_rate": 6.461656456126254e-06,
      "loss": 0.4243,
      "num_tokens": 1026278632.0,
      "step": 1083
    },
    {
      "epoch": 2.1214495592556317,
      "grad_norm": 0.0913536062911726,
      "learning_rate": 6.45559321930953e-06,
      "loss": 0.4344,
      "num_tokens": 1027227510.0,
      "step": 1084
    },
    {
      "epoch": 2.1234084231145935,
      "grad_norm": 0.08782548059322276,
      "learning_rate": 6.449527642917079e-06,
      "loss": 0.4065,
      "num_tokens": 1028156348.0,
      "step": 1085
    },
    {
      "epoch": 2.1253672869735554,
      "grad_norm": 0.08968160127978006,
      "learning_rate": 6.443459736698106e-06,
      "loss": 0.4238,
      "num_tokens": 1029105257.0,
      "step": 1086
    },
    {
      "epoch": 2.127326150832517,
      "grad_norm": 0.08757246543778505,
      "learning_rate": 6.4373895104055575e-06,
      "loss": 0.4168,
      "num_tokens": 1030059584.0,
      "step": 1087
    },
    {
      "epoch": 2.129285014691479,
      "grad_norm": 0.0925208798180306,
      "learning_rate": 6.431316973796114e-06,
      "loss": 0.4254,
      "num_tokens": 1031018195.0,
      "step": 1088
    },
    {
      "epoch": 2.131243878550441,
      "grad_norm": 0.09027882568785589,
      "learning_rate": 6.425242136630163e-06,
      "loss": 0.4326,
      "num_tokens": 1031915323.0,
      "step": 1089
    },
    {
      "epoch": 2.1332027424094027,
      "grad_norm": 0.08698448537140553,
      "learning_rate": 6.419165008671796e-06,
      "loss": 0.4168,
      "num_tokens": 1032860249.0,
      "step": 1090
    },
    {
      "epoch": 2.1351616062683645,
      "grad_norm": 0.0904908892556419,
      "learning_rate": 6.413085599688786e-06,
      "loss": 0.4165,
      "num_tokens": 1033827505.0,
      "step": 1091
    },
    {
      "epoch": 2.1371204701273263,
      "grad_norm": 0.0903982867050371,
      "learning_rate": 6.407003919452565e-06,
      "loss": 0.4243,
      "num_tokens": 1034761620.0,
      "step": 1092
    },
    {
      "epoch": 2.139079333986288,
      "grad_norm": 0.08664524783286122,
      "learning_rate": 6.400919977738222e-06,
      "loss": 0.426,
      "num_tokens": 1035720831.0,
      "step": 1093
    },
    {
      "epoch": 2.14103819784525,
      "grad_norm": 0.09074898123458312,
      "learning_rate": 6.394833784324485e-06,
      "loss": 0.4153,
      "num_tokens": 1036661948.0,
      "step": 1094
    },
    {
      "epoch": 2.142997061704212,
      "grad_norm": 0.08912674580637855,
      "learning_rate": 6.388745348993689e-06,
      "loss": 0.419,
      "num_tokens": 1037608963.0,
      "step": 1095
    },
    {
      "epoch": 2.144955925563173,
      "grad_norm": 0.08584834284684544,
      "learning_rate": 6.3826546815317835e-06,
      "loss": 0.408,
      "num_tokens": 1038556073.0,
      "step": 1096
    },
    {
      "epoch": 2.146914789422135,
      "grad_norm": 0.08516378615177235,
      "learning_rate": 6.3765617917282985e-06,
      "loss": 0.4228,
      "num_tokens": 1039529123.0,
      "step": 1097
    },
    {
      "epoch": 2.148873653281097,
      "grad_norm": 0.08620322833951624,
      "learning_rate": 6.370466689376343e-06,
      "loss": 0.4344,
      "num_tokens": 1040489386.0,
      "step": 1098
    },
    {
      "epoch": 2.1508325171400586,
      "grad_norm": 0.08632778561522257,
      "learning_rate": 6.364369384272573e-06,
      "loss": 0.4088,
      "num_tokens": 1041423415.0,
      "step": 1099
    },
    {
      "epoch": 2.1527913809990205,
      "grad_norm": 0.0912855425604682,
      "learning_rate": 6.3582698862171945e-06,
      "loss": 0.4253,
      "num_tokens": 1042352125.0,
      "step": 1100
    },
    {
      "epoch": 2.1547502448579823,
      "grad_norm": 0.09338332366594856,
      "learning_rate": 6.3521682050139325e-06,
      "loss": 0.4271,
      "num_tokens": 1043332301.0,
      "step": 1101
    },
    {
      "epoch": 2.156709108716944,
      "grad_norm": 0.08712608719213073,
      "learning_rate": 6.346064350470019e-06,
      "loss": 0.4112,
      "num_tokens": 1044303032.0,
      "step": 1102
    },
    {
      "epoch": 2.158667972575906,
      "grad_norm": 0.08406921815221968,
      "learning_rate": 6.3399583323961864e-06,
      "loss": 0.4129,
      "num_tokens": 1045225033.0,
      "step": 1103
    },
    {
      "epoch": 2.1606268364348677,
      "grad_norm": 0.08523689137761957,
      "learning_rate": 6.333850160606641e-06,
      "loss": 0.4182,
      "num_tokens": 1046200892.0,
      "step": 1104
    },
    {
      "epoch": 2.1625857002938296,
      "grad_norm": 0.08849746271017883,
      "learning_rate": 6.327739844919047e-06,
      "loss": 0.419,
      "num_tokens": 1047132811.0,
      "step": 1105
    },
    {
      "epoch": 2.1645445641527914,
      "grad_norm": 0.09296922046548894,
      "learning_rate": 6.321627395154522e-06,
      "loss": 0.413,
      "num_tokens": 1048082197.0,
      "step": 1106
    },
    {
      "epoch": 2.166503428011753,
      "grad_norm": 0.0868108858528206,
      "learning_rate": 6.315512821137606e-06,
      "loss": 0.4343,
      "num_tokens": 1049080403.0,
      "step": 1107
    },
    {
      "epoch": 2.168462291870715,
      "grad_norm": 0.09214250216477304,
      "learning_rate": 6.30939613269626e-06,
      "loss": 0.4344,
      "num_tokens": 1050035591.0,
      "step": 1108
    },
    {
      "epoch": 2.170421155729677,
      "grad_norm": 0.08449882559402659,
      "learning_rate": 6.303277339661839e-06,
      "loss": 0.4193,
      "num_tokens": 1050975812.0,
      "step": 1109
    },
    {
      "epoch": 2.1723800195886387,
      "grad_norm": 0.09423781044233816,
      "learning_rate": 6.297156451869082e-06,
      "loss": 0.4347,
      "num_tokens": 1051895775.0,
      "step": 1110
    },
    {
      "epoch": 2.1743388834476005,
      "grad_norm": 0.09075475482659402,
      "learning_rate": 6.291033479156094e-06,
      "loss": 0.4351,
      "num_tokens": 1052848543.0,
      "step": 1111
    },
    {
      "epoch": 2.1762977473065623,
      "grad_norm": 0.08949276553966692,
      "learning_rate": 6.284908431364336e-06,
      "loss": 0.4243,
      "num_tokens": 1053810946.0,
      "step": 1112
    },
    {
      "epoch": 2.178256611165524,
      "grad_norm": 0.08706268467759441,
      "learning_rate": 6.278781318338597e-06,
      "loss": 0.4399,
      "num_tokens": 1054804296.0,
      "step": 1113
    },
    {
      "epoch": 2.180215475024486,
      "grad_norm": 0.08873194457519597,
      "learning_rate": 6.272652149926989e-06,
      "loss": 0.4205,
      "num_tokens": 1055765758.0,
      "step": 1114
    },
    {
      "epoch": 2.182174338883448,
      "grad_norm": 0.08708002276397027,
      "learning_rate": 6.26652093598093e-06,
      "loss": 0.4249,
      "num_tokens": 1056721494.0,
      "step": 1115
    },
    {
      "epoch": 2.1841332027424096,
      "grad_norm": 0.08684815034937152,
      "learning_rate": 6.260387686355121e-06,
      "loss": 0.4096,
      "num_tokens": 1057665717.0,
      "step": 1116
    },
    {
      "epoch": 2.186092066601371,
      "grad_norm": 0.08676643946482575,
      "learning_rate": 6.254252410907536e-06,
      "loss": 0.4168,
      "num_tokens": 1058645384.0,
      "step": 1117
    },
    {
      "epoch": 2.188050930460333,
      "grad_norm": 0.08720475768234767,
      "learning_rate": 6.2481151194994105e-06,
      "loss": 0.4305,
      "num_tokens": 1059591036.0,
      "step": 1118
    },
    {
      "epoch": 2.1900097943192947,
      "grad_norm": 0.08636773101059496,
      "learning_rate": 6.241975821995213e-06,
      "loss": 0.416,
      "num_tokens": 1060528437.0,
      "step": 1119
    },
    {
      "epoch": 2.1919686581782565,
      "grad_norm": 0.08644496055481303,
      "learning_rate": 6.235834528262641e-06,
      "loss": 0.4181,
      "num_tokens": 1061491146.0,
      "step": 1120
    },
    {
      "epoch": 2.1939275220372183,
      "grad_norm": 0.08715824158543502,
      "learning_rate": 6.229691248172599e-06,
      "loss": 0.4114,
      "num_tokens": 1062393644.0,
      "step": 1121
    },
    {
      "epoch": 2.19588638589618,
      "grad_norm": 0.08812619878603911,
      "learning_rate": 6.223545991599184e-06,
      "loss": 0.4261,
      "num_tokens": 1063329990.0,
      "step": 1122
    },
    {
      "epoch": 2.197845249755142,
      "grad_norm": 0.09132206578888642,
      "learning_rate": 6.217398768419668e-06,
      "loss": 0.4288,
      "num_tokens": 1064269256.0,
      "step": 1123
    },
    {
      "epoch": 2.1998041136141038,
      "grad_norm": 0.08502664270876768,
      "learning_rate": 6.211249588514489e-06,
      "loss": 0.4087,
      "num_tokens": 1065164894.0,
      "step": 1124
    },
    {
      "epoch": 2.2017629774730656,
      "grad_norm": 0.08789264797629244,
      "learning_rate": 6.2050984617672275e-06,
      "loss": 0.4136,
      "num_tokens": 1066078469.0,
      "step": 1125
    },
    {
      "epoch": 2.2037218413320274,
      "grad_norm": 0.08596189310712543,
      "learning_rate": 6.198945398064589e-06,
      "loss": 0.4292,
      "num_tokens": 1067041602.0,
      "step": 1126
    },
    {
      "epoch": 2.2056807051909892,
      "grad_norm": 0.08946989522624431,
      "learning_rate": 6.192790407296399e-06,
      "loss": 0.4299,
      "num_tokens": 1067977225.0,
      "step": 1127
    },
    {
      "epoch": 2.207639569049951,
      "grad_norm": 0.08530660571718063,
      "learning_rate": 6.186633499355576e-06,
      "loss": 0.4274,
      "num_tokens": 1068925447.0,
      "step": 1128
    },
    {
      "epoch": 2.209598432908913,
      "grad_norm": 0.08223275716237784,
      "learning_rate": 6.18047468413812e-06,
      "loss": 0.418,
      "num_tokens": 1069898052.0,
      "step": 1129
    },
    {
      "epoch": 2.2115572967678747,
      "grad_norm": 0.08439202910237378,
      "learning_rate": 6.1743139715431e-06,
      "loss": 0.4113,
      "num_tokens": 1070859566.0,
      "step": 1130
    },
    {
      "epoch": 2.2135161606268365,
      "grad_norm": 0.08836386425823387,
      "learning_rate": 6.168151371472634e-06,
      "loss": 0.4156,
      "num_tokens": 1071812516.0,
      "step": 1131
    },
    {
      "epoch": 2.2154750244857984,
      "grad_norm": 0.08253391333509251,
      "learning_rate": 6.161986893831867e-06,
      "loss": 0.4085,
      "num_tokens": 1072788298.0,
      "step": 1132
    },
    {
      "epoch": 2.21743388834476,
      "grad_norm": 0.08734497027399377,
      "learning_rate": 6.155820548528972e-06,
      "loss": 0.4098,
      "num_tokens": 1073729223.0,
      "step": 1133
    },
    {
      "epoch": 2.219392752203722,
      "grad_norm": 0.0872983973605972,
      "learning_rate": 6.149652345475118e-06,
      "loss": 0.4199,
      "num_tokens": 1074663715.0,
      "step": 1134
    },
    {
      "epoch": 2.221351616062684,
      "grad_norm": 0.08553499760591111,
      "learning_rate": 6.143482294584459e-06,
      "loss": 0.4327,
      "num_tokens": 1075616314.0,
      "step": 1135
    },
    {
      "epoch": 2.223310479921645,
      "grad_norm": 0.08813226772588481,
      "learning_rate": 6.137310405774122e-06,
      "loss": 0.4153,
      "num_tokens": 1076569428.0,
      "step": 1136
    },
    {
      "epoch": 2.2252693437806075,
      "grad_norm": 0.08466239192875728,
      "learning_rate": 6.1311366889641875e-06,
      "loss": 0.4267,
      "num_tokens": 1077546687.0,
      "step": 1137
    },
    {
      "epoch": 2.227228207639569,
      "grad_norm": 0.090136417478135,
      "learning_rate": 6.124961154077672e-06,
      "loss": 0.4172,
      "num_tokens": 1078449090.0,
      "step": 1138
    },
    {
      "epoch": 2.2291870714985307,
      "grad_norm": 0.08482683736174175,
      "learning_rate": 6.118783811040518e-06,
      "loss": 0.4105,
      "num_tokens": 1079416873.0,
      "step": 1139
    },
    {
      "epoch": 2.2311459353574925,
      "grad_norm": 0.08520286917446548,
      "learning_rate": 6.112604669781572e-06,
      "loss": 0.4122,
      "num_tokens": 1080347126.0,
      "step": 1140
    },
    {
      "epoch": 2.2331047992164543,
      "grad_norm": 0.08782176324254114,
      "learning_rate": 6.106423740232571e-06,
      "loss": 0.425,
      "num_tokens": 1081289664.0,
      "step": 1141
    },
    {
      "epoch": 2.235063663075416,
      "grad_norm": 0.08680345708766998,
      "learning_rate": 6.100241032328125e-06,
      "loss": 0.4326,
      "num_tokens": 1082237218.0,
      "step": 1142
    },
    {
      "epoch": 2.237022526934378,
      "grad_norm": 0.09063147118544396,
      "learning_rate": 6.094056556005703e-06,
      "loss": 0.4296,
      "num_tokens": 1083165502.0,
      "step": 1143
    },
    {
      "epoch": 2.23898139079334,
      "grad_norm": 0.0849138051543222,
      "learning_rate": 6.087870321205619e-06,
      "loss": 0.4224,
      "num_tokens": 1084127711.0,
      "step": 1144
    },
    {
      "epoch": 2.2409402546523016,
      "grad_norm": 0.08643945489123467,
      "learning_rate": 6.0816823378710086e-06,
      "loss": 0.4157,
      "num_tokens": 1085059824.0,
      "step": 1145
    },
    {
      "epoch": 2.2428991185112634,
      "grad_norm": 0.0895327436928162,
      "learning_rate": 6.075492615947824e-06,
      "loss": 0.4289,
      "num_tokens": 1085988313.0,
      "step": 1146
    },
    {
      "epoch": 2.2448579823702253,
      "grad_norm": 0.0879595063795222,
      "learning_rate": 6.069301165384807e-06,
      "loss": 0.4179,
      "num_tokens": 1086926661.0,
      "step": 1147
    },
    {
      "epoch": 2.246816846229187,
      "grad_norm": 0.09157089650253557,
      "learning_rate": 6.063107996133477e-06,
      "loss": 0.4356,
      "num_tokens": 1087829770.0,
      "step": 1148
    },
    {
      "epoch": 2.248775710088149,
      "grad_norm": 0.09318446896644245,
      "learning_rate": 6.056913118148122e-06,
      "loss": 0.4299,
      "num_tokens": 1088788350.0,
      "step": 1149
    },
    {
      "epoch": 2.2507345739471107,
      "grad_norm": 0.087298596899998,
      "learning_rate": 6.05071654138577e-06,
      "loss": 0.4161,
      "num_tokens": 1089765315.0,
      "step": 1150
    },
    {
      "epoch": 2.2526934378060726,
      "grad_norm": 0.08361724539249257,
      "learning_rate": 6.044518275806184e-06,
      "loss": 0.3956,
      "num_tokens": 1090656166.0,
      "step": 1151
    },
    {
      "epoch": 2.2546523016650344,
      "grad_norm": 0.08685786823092265,
      "learning_rate": 6.038318331371836e-06,
      "loss": 0.4331,
      "num_tokens": 1091577603.0,
      "step": 1152
    },
    {
      "epoch": 2.256611165523996,
      "grad_norm": 0.08956135119922624,
      "learning_rate": 6.032116718047905e-06,
      "loss": 0.4167,
      "num_tokens": 1092518192.0,
      "step": 1153
    },
    {
      "epoch": 2.258570029382958,
      "grad_norm": 0.09705063375610096,
      "learning_rate": 6.025913445802242e-06,
      "loss": 0.4174,
      "num_tokens": 1093426114.0,
      "step": 1154
    },
    {
      "epoch": 2.26052889324192,
      "grad_norm": 0.08764165278411953,
      "learning_rate": 6.019708524605372e-06,
      "loss": 0.4154,
      "num_tokens": 1094372037.0,
      "step": 1155
    },
    {
      "epoch": 2.2624877571008817,
      "grad_norm": 0.08653416837402352,
      "learning_rate": 6.013501964430468e-06,
      "loss": 0.4201,
      "num_tokens": 1095349695.0,
      "step": 1156
    },
    {
      "epoch": 2.264446620959843,
      "grad_norm": 0.086947342784056,
      "learning_rate": 6.007293775253339e-06,
      "loss": 0.4172,
      "num_tokens": 1096289633.0,
      "step": 1157
    },
    {
      "epoch": 2.2664054848188053,
      "grad_norm": 0.09222607274740083,
      "learning_rate": 6.001083967052408e-06,
      "loss": 0.4369,
      "num_tokens": 1097268758.0,
      "step": 1158
    },
    {
      "epoch": 2.2683643486777667,
      "grad_norm": 0.09039180454960108,
      "learning_rate": 5.994872549808702e-06,
      "loss": 0.4136,
      "num_tokens": 1098237143.0,
      "step": 1159
    },
    {
      "epoch": 2.2703232125367285,
      "grad_norm": 0.0910752643051445,
      "learning_rate": 5.988659533505839e-06,
      "loss": 0.4238,
      "num_tokens": 1099175653.0,
      "step": 1160
    },
    {
      "epoch": 2.2722820763956904,
      "grad_norm": 0.08708378242269066,
      "learning_rate": 5.982444928130001e-06,
      "loss": 0.4385,
      "num_tokens": 1100137220.0,
      "step": 1161
    },
    {
      "epoch": 2.274240940254652,
      "grad_norm": 0.08831809469667551,
      "learning_rate": 5.976228743669926e-06,
      "loss": 0.4136,
      "num_tokens": 1101100621.0,
      "step": 1162
    },
    {
      "epoch": 2.276199804113614,
      "grad_norm": 0.08991776132591153,
      "learning_rate": 5.970010990116892e-06,
      "loss": 0.425,
      "num_tokens": 1102049160.0,
      "step": 1163
    },
    {
      "epoch": 2.278158667972576,
      "grad_norm": 0.08300228647211032,
      "learning_rate": 5.963791677464696e-06,
      "loss": 0.4185,
      "num_tokens": 1103011905.0,
      "step": 1164
    },
    {
      "epoch": 2.2801175318315376,
      "grad_norm": 0.09026104617757048,
      "learning_rate": 5.957570815709643e-06,
      "loss": 0.4205,
      "num_tokens": 1103958699.0,
      "step": 1165
    },
    {
      "epoch": 2.2820763956904995,
      "grad_norm": 0.08812812922118217,
      "learning_rate": 5.951348414850529e-06,
      "loss": 0.4207,
      "num_tokens": 1104914696.0,
      "step": 1166
    },
    {
      "epoch": 2.2840352595494613,
      "grad_norm": 0.08719939930912697,
      "learning_rate": 5.94512448488862e-06,
      "loss": 0.4313,
      "num_tokens": 1105860497.0,
      "step": 1167
    },
    {
      "epoch": 2.285994123408423,
      "grad_norm": 0.09334636444021462,
      "learning_rate": 5.938899035827641e-06,
      "loss": 0.42,
      "num_tokens": 1106806217.0,
      "step": 1168
    },
    {
      "epoch": 2.287952987267385,
      "grad_norm": 0.08758036200203533,
      "learning_rate": 5.93267207767376e-06,
      "loss": 0.4235,
      "num_tokens": 1107719105.0,
      "step": 1169
    },
    {
      "epoch": 2.2899118511263468,
      "grad_norm": 0.0865580879757578,
      "learning_rate": 5.926443620435572e-06,
      "loss": 0.4266,
      "num_tokens": 1108682386.0,
      "step": 1170
    },
    {
      "epoch": 2.2918707149853086,
      "grad_norm": 0.08826947457120272,
      "learning_rate": 5.920213674124079e-06,
      "loss": 0.4237,
      "num_tokens": 1109596326.0,
      "step": 1171
    },
    {
      "epoch": 2.2938295788442704,
      "grad_norm": 0.08879768641173116,
      "learning_rate": 5.913982248752675e-06,
      "loss": 0.4264,
      "num_tokens": 1110537632.0,
      "step": 1172
    },
    {
      "epoch": 2.2957884427032322,
      "grad_norm": 0.09002457607624295,
      "learning_rate": 5.907749354337133e-06,
      "loss": 0.4157,
      "num_tokens": 1111464017.0,
      "step": 1173
    },
    {
      "epoch": 2.297747306562194,
      "grad_norm": 0.08699029703191959,
      "learning_rate": 5.901515000895591e-06,
      "loss": 0.4356,
      "num_tokens": 1112388838.0,
      "step": 1174
    },
    {
      "epoch": 2.299706170421156,
      "grad_norm": 0.09153113371090561,
      "learning_rate": 5.8952791984485245e-06,
      "loss": 0.4347,
      "num_tokens": 1113323477.0,
      "step": 1175
    },
    {
      "epoch": 2.3016650342801177,
      "grad_norm": 0.09107919287896965,
      "learning_rate": 5.889041957018745e-06,
      "loss": 0.419,
      "num_tokens": 1114287175.0,
      "step": 1176
    },
    {
      "epoch": 2.3036238981390795,
      "grad_norm": 0.09021894631319483,
      "learning_rate": 5.8828032866313725e-06,
      "loss": 0.4215,
      "num_tokens": 1115217358.0,
      "step": 1177
    },
    {
      "epoch": 2.305582761998041,
      "grad_norm": 0.08630194272004957,
      "learning_rate": 5.876563197313824e-06,
      "loss": 0.4117,
      "num_tokens": 1116132185.0,
      "step": 1178
    },
    {
      "epoch": 2.307541625857003,
      "grad_norm": 0.08634061646782061,
      "learning_rate": 5.8703216990958015e-06,
      "loss": 0.4332,
      "num_tokens": 1117096741.0,
      "step": 1179
    },
    {
      "epoch": 2.3095004897159646,
      "grad_norm": 0.08881950285497699,
      "learning_rate": 5.8640788020092634e-06,
      "loss": 0.4277,
      "num_tokens": 1118023443.0,
      "step": 1180
    },
    {
      "epoch": 2.3114593535749264,
      "grad_norm": 0.08708093470736415,
      "learning_rate": 5.857834516088426e-06,
      "loss": 0.4196,
      "num_tokens": 1118991817.0,
      "step": 1181
    },
    {
      "epoch": 2.313418217433888,
      "grad_norm": 0.09735277120936721,
      "learning_rate": 5.85158885136973e-06,
      "loss": 0.4278,
      "num_tokens": 1119946883.0,
      "step": 1182
    },
    {
      "epoch": 2.31537708129285,
      "grad_norm": 0.09013207888988778,
      "learning_rate": 5.845341817891838e-06,
      "loss": 0.4288,
      "num_tokens": 1120899043.0,
      "step": 1183
    },
    {
      "epoch": 2.317335945151812,
      "grad_norm": 0.08656646884508337,
      "learning_rate": 5.839093425695609e-06,
      "loss": 0.421,
      "num_tokens": 1121848313.0,
      "step": 1184
    },
    {
      "epoch": 2.3192948090107737,
      "grad_norm": 0.08680555823074611,
      "learning_rate": 5.832843684824087e-06,
      "loss": 0.4256,
      "num_tokens": 1122829981.0,
      "step": 1185
    },
    {
      "epoch": 2.3212536728697355,
      "grad_norm": 0.08741721014384901,
      "learning_rate": 5.826592605322482e-06,
      "loss": 0.425,
      "num_tokens": 1123749809.0,
      "step": 1186
    },
    {
      "epoch": 2.3232125367286973,
      "grad_norm": 0.08642901810838186,
      "learning_rate": 5.820340197238161e-06,
      "loss": 0.4226,
      "num_tokens": 1124698596.0,
      "step": 1187
    },
    {
      "epoch": 2.325171400587659,
      "grad_norm": 0.08588498258919515,
      "learning_rate": 5.81408647062062e-06,
      "loss": 0.4393,
      "num_tokens": 1125657206.0,
      "step": 1188
    },
    {
      "epoch": 2.327130264446621,
      "grad_norm": 0.08748373647641175,
      "learning_rate": 5.807831435521477e-06,
      "loss": 0.4232,
      "num_tokens": 1126610387.0,
      "step": 1189
    },
    {
      "epoch": 2.329089128305583,
      "grad_norm": 0.08927452797282809,
      "learning_rate": 5.8015751019944555e-06,
      "loss": 0.4213,
      "num_tokens": 1127539698.0,
      "step": 1190
    },
    {
      "epoch": 2.3310479921645446,
      "grad_norm": 0.0893519642205304,
      "learning_rate": 5.795317480095361e-06,
      "loss": 0.4125,
      "num_tokens": 1128510407.0,
      "step": 1191
    },
    {
      "epoch": 2.3330068560235064,
      "grad_norm": 0.08491636345490047,
      "learning_rate": 5.789058579882074e-06,
      "loss": 0.4121,
      "num_tokens": 1129470063.0,
      "step": 1192
    },
    {
      "epoch": 2.3349657198824683,
      "grad_norm": 0.08984252196487613,
      "learning_rate": 5.782798411414525e-06,
      "loss": 0.4331,
      "num_tokens": 1130367777.0,
      "step": 1193
    },
    {
      "epoch": 2.33692458374143,
      "grad_norm": 0.09053581250815942,
      "learning_rate": 5.776536984754691e-06,
      "loss": 0.4202,
      "num_tokens": 1131311686.0,
      "step": 1194
    },
    {
      "epoch": 2.338883447600392,
      "grad_norm": 0.09063978299387467,
      "learning_rate": 5.770274309966562e-06,
      "loss": 0.4223,
      "num_tokens": 1132235248.0,
      "step": 1195
    },
    {
      "epoch": 2.3408423114593537,
      "grad_norm": 0.08800259601948023,
      "learning_rate": 5.76401039711614e-06,
      "loss": 0.4338,
      "num_tokens": 1133183331.0,
      "step": 1196
    },
    {
      "epoch": 2.3428011753183156,
      "grad_norm": 0.08719622932715004,
      "learning_rate": 5.757745256271416e-06,
      "loss": 0.4308,
      "num_tokens": 1134132519.0,
      "step": 1197
    },
    {
      "epoch": 2.3447600391772774,
      "grad_norm": 0.08572206568995963,
      "learning_rate": 5.751478897502353e-06,
      "loss": 0.4096,
      "num_tokens": 1135054379.0,
      "step": 1198
    },
    {
      "epoch": 2.3467189030362388,
      "grad_norm": 0.08817276524972734,
      "learning_rate": 5.745211330880872e-06,
      "loss": 0.4294,
      "num_tokens": 1136022955.0,
      "step": 1199
    },
    {
      "epoch": 2.348677766895201,
      "grad_norm": 0.091228551530654,
      "learning_rate": 5.738942566480839e-06,
      "loss": 0.4303,
      "num_tokens": 1136958778.0,
      "step": 1200
    },
    {
      "epoch": 2.3506366307541624,
      "grad_norm": 0.08868420902004148,
      "learning_rate": 5.7326726143780394e-06,
      "loss": 0.4284,
      "num_tokens": 1137912240.0,
      "step": 1201
    },
    {
      "epoch": 2.3525954946131242,
      "grad_norm": 0.08948749533257055,
      "learning_rate": 5.726401484650171e-06,
      "loss": 0.4406,
      "num_tokens": 1138838672.0,
      "step": 1202
    },
    {
      "epoch": 2.354554358472086,
      "grad_norm": 0.09142155592338352,
      "learning_rate": 5.720129187376825e-06,
      "loss": 0.4306,
      "num_tokens": 1139765318.0,
      "step": 1203
    },
    {
      "epoch": 2.356513222331048,
      "grad_norm": 0.08728357874374511,
      "learning_rate": 5.713855732639466e-06,
      "loss": 0.4121,
      "num_tokens": 1140687203.0,
      "step": 1204
    },
    {
      "epoch": 2.3584720861900097,
      "grad_norm": 0.0923934935729649,
      "learning_rate": 5.707581130521424e-06,
      "loss": 0.423,
      "num_tokens": 1141640022.0,
      "step": 1205
    },
    {
      "epoch": 2.3604309500489715,
      "grad_norm": 0.08824418035661111,
      "learning_rate": 5.701305391107868e-06,
      "loss": 0.4153,
      "num_tokens": 1142588951.0,
      "step": 1206
    },
    {
      "epoch": 2.3623898139079333,
      "grad_norm": 0.08604299060885007,
      "learning_rate": 5.695028524485797e-06,
      "loss": 0.4159,
      "num_tokens": 1143525481.0,
      "step": 1207
    },
    {
      "epoch": 2.364348677766895,
      "grad_norm": 0.08486896431466706,
      "learning_rate": 5.688750540744021e-06,
      "loss": 0.4283,
      "num_tokens": 1144478548.0,
      "step": 1208
    },
    {
      "epoch": 2.366307541625857,
      "grad_norm": 0.08550242628203791,
      "learning_rate": 5.6824714499731505e-06,
      "loss": 0.4234,
      "num_tokens": 1145416931.0,
      "step": 1209
    },
    {
      "epoch": 2.368266405484819,
      "grad_norm": 0.08380409463016955,
      "learning_rate": 5.6761912622655675e-06,
      "loss": 0.4171,
      "num_tokens": 1146393240.0,
      "step": 1210
    },
    {
      "epoch": 2.3702252693437806,
      "grad_norm": 0.08748035599931478,
      "learning_rate": 5.6699099877154214e-06,
      "loss": 0.4148,
      "num_tokens": 1147358141.0,
      "step": 1211
    },
    {
      "epoch": 2.3721841332027425,
      "grad_norm": 0.08850046424705908,
      "learning_rate": 5.663627636418611e-06,
      "loss": 0.4325,
      "num_tokens": 1148312744.0,
      "step": 1212
    },
    {
      "epoch": 2.3741429970617043,
      "grad_norm": 0.08528350097972932,
      "learning_rate": 5.65734421847276e-06,
      "loss": 0.4274,
      "num_tokens": 1149299127.0,
      "step": 1213
    },
    {
      "epoch": 2.376101860920666,
      "grad_norm": 0.0858978532254676,
      "learning_rate": 5.6510597439772085e-06,
      "loss": 0.4281,
      "num_tokens": 1150243554.0,
      "step": 1214
    },
    {
      "epoch": 2.378060724779628,
      "grad_norm": 0.08659252218510509,
      "learning_rate": 5.644774223032997e-06,
      "loss": 0.4309,
      "num_tokens": 1151178767.0,
      "step": 1215
    },
    {
      "epoch": 2.3800195886385898,
      "grad_norm": 0.08581836773259258,
      "learning_rate": 5.638487665742849e-06,
      "loss": 0.4279,
      "num_tokens": 1152102608.0,
      "step": 1216
    },
    {
      "epoch": 2.3819784524975516,
      "grad_norm": 0.0867447544185725,
      "learning_rate": 5.632200082211146e-06,
      "loss": 0.4206,
      "num_tokens": 1153091123.0,
      "step": 1217
    },
    {
      "epoch": 2.3839373163565134,
      "grad_norm": 0.08594354254331536,
      "learning_rate": 5.625911482543928e-06,
      "loss": 0.427,
      "num_tokens": 1154072296.0,
      "step": 1218
    },
    {
      "epoch": 2.3858961802154752,
      "grad_norm": 0.08852540163719144,
      "learning_rate": 5.619621876848864e-06,
      "loss": 0.4211,
      "num_tokens": 1154992011.0,
      "step": 1219
    },
    {
      "epoch": 2.3878550440744366,
      "grad_norm": 0.08588867297587977,
      "learning_rate": 5.61333127523524e-06,
      "loss": 0.4348,
      "num_tokens": 1155926089.0,
      "step": 1220
    },
    {
      "epoch": 2.389813907933399,
      "grad_norm": 0.08344556567626972,
      "learning_rate": 5.607039687813942e-06,
      "loss": 0.4257,
      "num_tokens": 1156889859.0,
      "step": 1221
    },
    {
      "epoch": 2.3917727717923603,
      "grad_norm": 0.08750373287023493,
      "learning_rate": 5.600747124697442e-06,
      "loss": 0.4332,
      "num_tokens": 1157807428.0,
      "step": 1222
    },
    {
      "epoch": 2.393731635651322,
      "grad_norm": 0.0885322929769485,
      "learning_rate": 5.5944535959997825e-06,
      "loss": 0.4271,
      "num_tokens": 1158772647.0,
      "step": 1223
    },
    {
      "epoch": 2.395690499510284,
      "grad_norm": 0.08584501160678022,
      "learning_rate": 5.588159111836553e-06,
      "loss": 0.416,
      "num_tokens": 1159719729.0,
      "step": 1224
    },
    {
      "epoch": 2.3976493633692457,
      "grad_norm": 0.08383944177589096,
      "learning_rate": 5.581863682324882e-06,
      "loss": 0.4251,
      "num_tokens": 1160705818.0,
      "step": 1225
    },
    {
      "epoch": 2.3996082272282075,
      "grad_norm": 0.08580460156302355,
      "learning_rate": 5.575567317583415e-06,
      "loss": 0.4183,
      "num_tokens": 1161669076.0,
      "step": 1226
    },
    {
      "epoch": 2.4015670910871694,
      "grad_norm": 0.09267158512984244,
      "learning_rate": 5.569270027732306e-06,
      "loss": 0.4257,
      "num_tokens": 1162619969.0,
      "step": 1227
    },
    {
      "epoch": 2.403525954946131,
      "grad_norm": 0.08421035900643646,
      "learning_rate": 5.562971822893187e-06,
      "loss": 0.4094,
      "num_tokens": 1163571952.0,
      "step": 1228
    },
    {
      "epoch": 2.405484818805093,
      "grad_norm": 0.08596664522543533,
      "learning_rate": 5.55667271318917e-06,
      "loss": 0.4203,
      "num_tokens": 1164539426.0,
      "step": 1229
    },
    {
      "epoch": 2.407443682664055,
      "grad_norm": 0.09233976790408117,
      "learning_rate": 5.5503727087448155e-06,
      "loss": 0.4227,
      "num_tokens": 1165474032.0,
      "step": 1230
    },
    {
      "epoch": 2.4094025465230167,
      "grad_norm": 0.08688320768846537,
      "learning_rate": 5.544071819686125e-06,
      "loss": 0.4248,
      "num_tokens": 1166422493.0,
      "step": 1231
    },
    {
      "epoch": 2.4113614103819785,
      "grad_norm": 0.0945756226192389,
      "learning_rate": 5.53777005614052e-06,
      "loss": 0.4302,
      "num_tokens": 1167390753.0,
      "step": 1232
    },
    {
      "epoch": 2.4133202742409403,
      "grad_norm": 0.08552777818490351,
      "learning_rate": 5.531467428236827e-06,
      "loss": 0.429,
      "num_tokens": 1168369616.0,
      "step": 1233
    },
    {
      "epoch": 2.415279138099902,
      "grad_norm": 0.08703907151650343,
      "learning_rate": 5.525163946105266e-06,
      "loss": 0.4264,
      "num_tokens": 1169323260.0,
      "step": 1234
    },
    {
      "epoch": 2.417238001958864,
      "grad_norm": 0.0855132558577357,
      "learning_rate": 5.5188596198774244e-06,
      "loss": 0.4236,
      "num_tokens": 1170253457.0,
      "step": 1235
    },
    {
      "epoch": 2.419196865817826,
      "grad_norm": 0.09911091065156342,
      "learning_rate": 5.5125544596862505e-06,
      "loss": 0.4286,
      "num_tokens": 1171178861.0,
      "step": 1236
    },
    {
      "epoch": 2.4211557296767876,
      "grad_norm": 0.09515489730191913,
      "learning_rate": 5.506248475666031e-06,
      "loss": 0.433,
      "num_tokens": 1172118925.0,
      "step": 1237
    },
    {
      "epoch": 2.4231145935357494,
      "grad_norm": 0.0922862855799988,
      "learning_rate": 5.499941677952376e-06,
      "loss": 0.4112,
      "num_tokens": 1173043753.0,
      "step": 1238
    },
    {
      "epoch": 2.4250734573947113,
      "grad_norm": 0.0927490877834443,
      "learning_rate": 5.493634076682206e-06,
      "loss": 0.4255,
      "num_tokens": 1173983910.0,
      "step": 1239
    },
    {
      "epoch": 2.427032321253673,
      "grad_norm": 0.08991829452505166,
      "learning_rate": 5.4873256819937325e-06,
      "loss": 0.4304,
      "num_tokens": 1174924986.0,
      "step": 1240
    },
    {
      "epoch": 2.4289911851126345,
      "grad_norm": 0.09654364592883412,
      "learning_rate": 5.48101650402644e-06,
      "loss": 0.4223,
      "num_tokens": 1175880543.0,
      "step": 1241
    },
    {
      "epoch": 2.4309500489715967,
      "grad_norm": 0.09143355884515676,
      "learning_rate": 5.474706552921074e-06,
      "loss": 0.4293,
      "num_tokens": 1176851342.0,
      "step": 1242
    },
    {
      "epoch": 2.432908912830558,
      "grad_norm": 0.08487914794201284,
      "learning_rate": 5.468395838819624e-06,
      "loss": 0.4189,
      "num_tokens": 1177790517.0,
      "step": 1243
    },
    {
      "epoch": 2.43486777668952,
      "grad_norm": 0.08484709848706648,
      "learning_rate": 5.462084371865301e-06,
      "loss": 0.4103,
      "num_tokens": 1178737565.0,
      "step": 1244
    },
    {
      "epoch": 2.4368266405484817,
      "grad_norm": 0.08781895956074455,
      "learning_rate": 5.455772162202531e-06,
      "loss": 0.4263,
      "num_tokens": 1179660472.0,
      "step": 1245
    },
    {
      "epoch": 2.4387855044074436,
      "grad_norm": 0.0911733217803357,
      "learning_rate": 5.44945921997693e-06,
      "loss": 0.4263,
      "num_tokens": 1180588262.0,
      "step": 1246
    },
    {
      "epoch": 2.4407443682664054,
      "grad_norm": 0.08748336699199348,
      "learning_rate": 5.443145555335296e-06,
      "loss": 0.4336,
      "num_tokens": 1181543520.0,
      "step": 1247
    },
    {
      "epoch": 2.442703232125367,
      "grad_norm": 0.08665842688163865,
      "learning_rate": 5.436831178425582e-06,
      "loss": 0.4205,
      "num_tokens": 1182481681.0,
      "step": 1248
    },
    {
      "epoch": 2.444662095984329,
      "grad_norm": 0.08652704081826909,
      "learning_rate": 5.430516099396892e-06,
      "loss": 0.4119,
      "num_tokens": 1183408064.0,
      "step": 1249
    },
    {
      "epoch": 2.446620959843291,
      "grad_norm": 0.09089265625207198,
      "learning_rate": 5.424200328399456e-06,
      "loss": 0.4302,
      "num_tokens": 1184398506.0,
      "step": 1250
    },
    {
      "epoch": 2.4485798237022527,
      "grad_norm": 0.08665999856715143,
      "learning_rate": 5.417883875584613e-06,
      "loss": 0.4126,
      "num_tokens": 1185351585.0,
      "step": 1251
    },
    {
      "epoch": 2.4505386875612145,
      "grad_norm": 0.08772877346785295,
      "learning_rate": 5.411566751104804e-06,
      "loss": 0.4188,
      "num_tokens": 1186298168.0,
      "step": 1252
    },
    {
      "epoch": 2.4524975514201763,
      "grad_norm": 0.08846717146727087,
      "learning_rate": 5.405248965113546e-06,
      "loss": 0.4216,
      "num_tokens": 1187220175.0,
      "step": 1253
    },
    {
      "epoch": 2.454456415279138,
      "grad_norm": 0.09323937399381285,
      "learning_rate": 5.398930527765416e-06,
      "loss": 0.4173,
      "num_tokens": 1188169761.0,
      "step": 1254
    },
    {
      "epoch": 2.4564152791381,
      "grad_norm": 0.08635935317945805,
      "learning_rate": 5.392611449216045e-06,
      "loss": 0.4071,
      "num_tokens": 1189100120.0,
      "step": 1255
    },
    {
      "epoch": 2.458374142997062,
      "grad_norm": 0.09471159385477396,
      "learning_rate": 5.3862917396220925e-06,
      "loss": 0.4161,
      "num_tokens": 1190045776.0,
      "step": 1256
    },
    {
      "epoch": 2.4603330068560236,
      "grad_norm": 0.08940277625598954,
      "learning_rate": 5.379971409141227e-06,
      "loss": 0.4244,
      "num_tokens": 1190965780.0,
      "step": 1257
    },
    {
      "epoch": 2.4622918707149855,
      "grad_norm": 0.08333095986467484,
      "learning_rate": 5.373650467932122e-06,
      "loss": 0.4071,
      "num_tokens": 1191942307.0,
      "step": 1258
    },
    {
      "epoch": 2.4642507345739473,
      "grad_norm": 0.08622640845177042,
      "learning_rate": 5.367328926154428e-06,
      "loss": 0.4189,
      "num_tokens": 1192906717.0,
      "step": 1259
    },
    {
      "epoch": 2.466209598432909,
      "grad_norm": 0.08819087470337499,
      "learning_rate": 5.361006793968764e-06,
      "loss": 0.4399,
      "num_tokens": 1193847834.0,
      "step": 1260
    },
    {
      "epoch": 2.468168462291871,
      "grad_norm": 0.08392409479166801,
      "learning_rate": 5.354684081536693e-06,
      "loss": 0.4055,
      "num_tokens": 1194800527.0,
      "step": 1261
    },
    {
      "epoch": 2.4701273261508323,
      "grad_norm": 0.08873929760927023,
      "learning_rate": 5.348360799020717e-06,
      "loss": 0.4275,
      "num_tokens": 1195738245.0,
      "step": 1262
    },
    {
      "epoch": 2.472086190009794,
      "grad_norm": 0.0870619869669245,
      "learning_rate": 5.342036956584251e-06,
      "loss": 0.4162,
      "num_tokens": 1196709708.0,
      "step": 1263
    },
    {
      "epoch": 2.474045053868756,
      "grad_norm": 0.08833235986296435,
      "learning_rate": 5.335712564391609e-06,
      "loss": 0.4314,
      "num_tokens": 1197646062.0,
      "step": 1264
    },
    {
      "epoch": 2.4760039177277178,
      "grad_norm": 0.08809126754844328,
      "learning_rate": 5.329387632607989e-06,
      "loss": 0.4203,
      "num_tokens": 1198575151.0,
      "step": 1265
    },
    {
      "epoch": 2.4779627815866796,
      "grad_norm": 0.09220084623013948,
      "learning_rate": 5.32306217139946e-06,
      "loss": 0.4212,
      "num_tokens": 1199546127.0,
      "step": 1266
    },
    {
      "epoch": 2.4799216454456414,
      "grad_norm": 0.08788363086287648,
      "learning_rate": 5.316736190932936e-06,
      "loss": 0.4289,
      "num_tokens": 1200487335.0,
      "step": 1267
    },
    {
      "epoch": 2.4818805093046032,
      "grad_norm": 0.08637705985091716,
      "learning_rate": 5.31040970137617e-06,
      "loss": 0.4284,
      "num_tokens": 1201402960.0,
      "step": 1268
    },
    {
      "epoch": 2.483839373163565,
      "grad_norm": 0.08505942012147649,
      "learning_rate": 5.304082712897734e-06,
      "loss": 0.4073,
      "num_tokens": 1202357356.0,
      "step": 1269
    },
    {
      "epoch": 2.485798237022527,
      "grad_norm": 0.08793991445308141,
      "learning_rate": 5.297755235666996e-06,
      "loss": 0.4272,
      "num_tokens": 1203318073.0,
      "step": 1270
    },
    {
      "epoch": 2.4877571008814887,
      "grad_norm": 0.08715273201098046,
      "learning_rate": 5.291427279854117e-06,
      "loss": 0.4109,
      "num_tokens": 1204275795.0,
      "step": 1271
    },
    {
      "epoch": 2.4897159647404505,
      "grad_norm": 0.08511146390681669,
      "learning_rate": 5.28509885563002e-06,
      "loss": 0.4107,
      "num_tokens": 1205252135.0,
      "step": 1272
    },
    {
      "epoch": 2.4916748285994124,
      "grad_norm": 0.08417989921996578,
      "learning_rate": 5.2787699731663865e-06,
      "loss": 0.4074,
      "num_tokens": 1206216487.0,
      "step": 1273
    },
    {
      "epoch": 2.493633692458374,
      "grad_norm": 0.08565702386459073,
      "learning_rate": 5.2724406426356325e-06,
      "loss": 0.4189,
      "num_tokens": 1207143688.0,
      "step": 1274
    },
    {
      "epoch": 2.495592556317336,
      "grad_norm": 0.08517590404259694,
      "learning_rate": 5.266110874210893e-06,
      "loss": 0.4211,
      "num_tokens": 1208077199.0,
      "step": 1275
    },
    {
      "epoch": 2.497551420176298,
      "grad_norm": 0.08473219993111532,
      "learning_rate": 5.259780678066012e-06,
      "loss": 0.4222,
      "num_tokens": 1209022544.0,
      "step": 1276
    },
    {
      "epoch": 2.4995102840352597,
      "grad_norm": 0.08838379088708623,
      "learning_rate": 5.253450064375512e-06,
      "loss": 0.418,
      "num_tokens": 1209974535.0,
      "step": 1277
    },
    {
      "epoch": 2.5014691478942215,
      "grad_norm": 0.0935118444195652,
      "learning_rate": 5.247119043314592e-06,
      "loss": 0.4234,
      "num_tokens": 1210907077.0,
      "step": 1278
    },
    {
      "epoch": 2.5034280117531833,
      "grad_norm": 0.08726823635723774,
      "learning_rate": 5.24078762505911e-06,
      "loss": 0.4232,
      "num_tokens": 1211840227.0,
      "step": 1279
    },
    {
      "epoch": 2.505386875612145,
      "grad_norm": 0.08689683850520893,
      "learning_rate": 5.234455819785554e-06,
      "loss": 0.4266,
      "num_tokens": 1212811578.0,
      "step": 1280
    },
    {
      "epoch": 2.5073457394711065,
      "grad_norm": 0.08668393909851933,
      "learning_rate": 5.228123637671037e-06,
      "loss": 0.4243,
      "num_tokens": 1213755099.0,
      "step": 1281
    },
    {
      "epoch": 2.5093046033300688,
      "grad_norm": 0.08635148027331238,
      "learning_rate": 5.221791088893282e-06,
      "loss": 0.4186,
      "num_tokens": 1214715473.0,
      "step": 1282
    },
    {
      "epoch": 2.51126346718903,
      "grad_norm": 0.08660314040039935,
      "learning_rate": 5.215458183630597e-06,
      "loss": 0.4123,
      "num_tokens": 1215671606.0,
      "step": 1283
    },
    {
      "epoch": 2.5132223310479924,
      "grad_norm": 0.08970370400613235,
      "learning_rate": 5.209124932061862e-06,
      "loss": 0.4251,
      "num_tokens": 1216615570.0,
      "step": 1284
    },
    {
      "epoch": 2.515181194906954,
      "grad_norm": 0.09140977797954425,
      "learning_rate": 5.202791344366517e-06,
      "loss": 0.4104,
      "num_tokens": 1217500696.0,
      "step": 1285
    },
    {
      "epoch": 2.5171400587659156,
      "grad_norm": 0.09726820085104514,
      "learning_rate": 5.196457430724541e-06,
      "loss": 0.4518,
      "num_tokens": 1218426160.0,
      "step": 1286
    },
    {
      "epoch": 2.5190989226248774,
      "grad_norm": 0.09152544566287074,
      "learning_rate": 5.190123201316435e-06,
      "loss": 0.4338,
      "num_tokens": 1219386445.0,
      "step": 1287
    },
    {
      "epoch": 2.5210577864838393,
      "grad_norm": 0.08702100478160023,
      "learning_rate": 5.183788666323209e-06,
      "loss": 0.4083,
      "num_tokens": 1220323923.0,
      "step": 1288
    },
    {
      "epoch": 2.523016650342801,
      "grad_norm": 0.08763610530813823,
      "learning_rate": 5.177453835926366e-06,
      "loss": 0.4335,
      "num_tokens": 1221246258.0,
      "step": 1289
    },
    {
      "epoch": 2.524975514201763,
      "grad_norm": 0.09035880771317832,
      "learning_rate": 5.1711187203078826e-06,
      "loss": 0.4364,
      "num_tokens": 1222161108.0,
      "step": 1290
    },
    {
      "epoch": 2.5269343780607247,
      "grad_norm": 0.08886694883924055,
      "learning_rate": 5.164783329650192e-06,
      "loss": 0.422,
      "num_tokens": 1223089888.0,
      "step": 1291
    },
    {
      "epoch": 2.5288932419196866,
      "grad_norm": 0.0902986541107478,
      "learning_rate": 5.15844767413617e-06,
      "loss": 0.4333,
      "num_tokens": 1224025492.0,
      "step": 1292
    },
    {
      "epoch": 2.5308521057786484,
      "grad_norm": 0.08450434755008468,
      "learning_rate": 5.152111763949122e-06,
      "loss": 0.4113,
      "num_tokens": 1224962829.0,
      "step": 1293
    },
    {
      "epoch": 2.53281096963761,
      "grad_norm": 0.0842951630380676,
      "learning_rate": 5.145775609272757e-06,
      "loss": 0.4107,
      "num_tokens": 1225936176.0,
      "step": 1294
    },
    {
      "epoch": 2.534769833496572,
      "grad_norm": 0.09140174623138231,
      "learning_rate": 5.13943922029118e-06,
      "loss": 0.4277,
      "num_tokens": 1226884735.0,
      "step": 1295
    },
    {
      "epoch": 2.536728697355534,
      "grad_norm": 0.08431087858868833,
      "learning_rate": 5.133102607188875e-06,
      "loss": 0.427,
      "num_tokens": 1227836437.0,
      "step": 1296
    },
    {
      "epoch": 2.5386875612144957,
      "grad_norm": 0.09111438925586439,
      "learning_rate": 5.126765780150678e-06,
      "loss": 0.4398,
      "num_tokens": 1228767154.0,
      "step": 1297
    },
    {
      "epoch": 2.5406464250734575,
      "grad_norm": 0.09498884391890484,
      "learning_rate": 5.120428749361781e-06,
      "loss": 0.433,
      "num_tokens": 1229723280.0,
      "step": 1298
    },
    {
      "epoch": 2.5426052889324193,
      "grad_norm": 0.0889692529410911,
      "learning_rate": 5.1140915250076914e-06,
      "loss": 0.4161,
      "num_tokens": 1230657337.0,
      "step": 1299
    },
    {
      "epoch": 2.544564152791381,
      "grad_norm": 0.0868340073908007,
      "learning_rate": 5.107754117274236e-06,
      "loss": 0.4309,
      "num_tokens": 1231629360.0,
      "step": 1300
    },
    {
      "epoch": 2.546523016650343,
      "grad_norm": 0.08610435407697006,
      "learning_rate": 5.101416536347532e-06,
      "loss": 0.4287,
      "num_tokens": 1232581778.0,
      "step": 1301
    },
    {
      "epoch": 2.5484818805093044,
      "grad_norm": 0.08719213873117067,
      "learning_rate": 5.095078792413976e-06,
      "loss": 0.4213,
      "num_tokens": 1233538873.0,
      "step": 1302
    },
    {
      "epoch": 2.5504407443682666,
      "grad_norm": 0.08634673821346445,
      "learning_rate": 5.0887408956602316e-06,
      "loss": 0.4195,
      "num_tokens": 1234483567.0,
      "step": 1303
    },
    {
      "epoch": 2.552399608227228,
      "grad_norm": 0.09344386614951955,
      "learning_rate": 5.082402856273198e-06,
      "loss": 0.4174,
      "num_tokens": 1235412735.0,
      "step": 1304
    },
    {
      "epoch": 2.5543584720861903,
      "grad_norm": 0.08676201793077444,
      "learning_rate": 5.0760646844400105e-06,
      "loss": 0.4285,
      "num_tokens": 1236381151.0,
      "step": 1305
    },
    {
      "epoch": 2.5563173359451516,
      "grad_norm": 0.08656747436579534,
      "learning_rate": 5.069726390348019e-06,
      "loss": 0.4193,
      "num_tokens": 1237289451.0,
      "step": 1306
    },
    {
      "epoch": 2.5582761998041135,
      "grad_norm": 0.08486631831734234,
      "learning_rate": 5.063387984184763e-06,
      "loss": 0.427,
      "num_tokens": 1238216386.0,
      "step": 1307
    },
    {
      "epoch": 2.5602350636630753,
      "grad_norm": 0.09018745711808192,
      "learning_rate": 5.057049476137968e-06,
      "loss": 0.4369,
      "num_tokens": 1239152655.0,
      "step": 1308
    },
    {
      "epoch": 2.562193927522037,
      "grad_norm": 0.08541194231580124,
      "learning_rate": 5.05071087639552e-06,
      "loss": 0.4149,
      "num_tokens": 1240077458.0,
      "step": 1309
    },
    {
      "epoch": 2.564152791380999,
      "grad_norm": 0.09055586101454287,
      "learning_rate": 5.044372195145455e-06,
      "loss": 0.4256,
      "num_tokens": 1241039487.0,
      "step": 1310
    },
    {
      "epoch": 2.5661116552399608,
      "grad_norm": 0.08645832979757842,
      "learning_rate": 5.038033442575939e-06,
      "loss": 0.4135,
      "num_tokens": 1242000722.0,
      "step": 1311
    },
    {
      "epoch": 2.5680705190989226,
      "grad_norm": 0.08787443136369189,
      "learning_rate": 5.03169462887525e-06,
      "loss": 0.4254,
      "num_tokens": 1242927159.0,
      "step": 1312
    },
    {
      "epoch": 2.5700293829578844,
      "grad_norm": 0.08809899335198573,
      "learning_rate": 5.025355764231769e-06,
      "loss": 0.4267,
      "num_tokens": 1243890829.0,
      "step": 1313
    },
    {
      "epoch": 2.5719882468168462,
      "grad_norm": 0.09132544297416752,
      "learning_rate": 5.019016858833954e-06,
      "loss": 0.424,
      "num_tokens": 1244841292.0,
      "step": 1314
    },
    {
      "epoch": 2.573947110675808,
      "grad_norm": 0.08722016010171965,
      "learning_rate": 5.012677922870332e-06,
      "loss": 0.4221,
      "num_tokens": 1245805034.0,
      "step": 1315
    },
    {
      "epoch": 2.57590597453477,
      "grad_norm": 0.08219176566970758,
      "learning_rate": 5.006338966529479e-06,
      "loss": 0.422,
      "num_tokens": 1246760183.0,
      "step": 1316
    },
    {
      "epoch": 2.5778648383937317,
      "grad_norm": 0.08664017458337375,
      "learning_rate": 5e-06,
      "loss": 0.4288,
      "num_tokens": 1247697466.0,
      "step": 1317
    },
    {
      "epoch": 2.5798237022526935,
      "grad_norm": 0.08679633888158816,
      "learning_rate": 4.993661033470522e-06,
      "loss": 0.4233,
      "num_tokens": 1248618984.0,
      "step": 1318
    },
    {
      "epoch": 2.5817825661116554,
      "grad_norm": 0.08405379996897977,
      "learning_rate": 4.98732207712967e-06,
      "loss": 0.4293,
      "num_tokens": 1249596593.0,
      "step": 1319
    },
    {
      "epoch": 2.583741429970617,
      "grad_norm": 0.0876393179149807,
      "learning_rate": 4.980983141166047e-06,
      "loss": 0.4376,
      "num_tokens": 1250566838.0,
      "step": 1320
    },
    {
      "epoch": 2.585700293829579,
      "grad_norm": 0.08667325691006618,
      "learning_rate": 4.974644235768232e-06,
      "loss": 0.4462,
      "num_tokens": 1251527132.0,
      "step": 1321
    },
    {
      "epoch": 2.587659157688541,
      "grad_norm": 0.08681222762067749,
      "learning_rate": 4.9683053711247515e-06,
      "loss": 0.4271,
      "num_tokens": 1252486319.0,
      "step": 1322
    },
    {
      "epoch": 2.589618021547502,
      "grad_norm": 0.08560882662756536,
      "learning_rate": 4.961966557424063e-06,
      "loss": 0.4274,
      "num_tokens": 1253446832.0,
      "step": 1323
    },
    {
      "epoch": 2.5915768854064645,
      "grad_norm": 0.08701258458632462,
      "learning_rate": 4.9556278048545445e-06,
      "loss": 0.4228,
      "num_tokens": 1254392072.0,
      "step": 1324
    },
    {
      "epoch": 2.593535749265426,
      "grad_norm": 0.08551688483760193,
      "learning_rate": 4.949289123604481e-06,
      "loss": 0.4264,
      "num_tokens": 1255359716.0,
      "step": 1325
    },
    {
      "epoch": 2.595494613124388,
      "grad_norm": 0.08665994615475629,
      "learning_rate": 4.942950523862033e-06,
      "loss": 0.4441,
      "num_tokens": 1256344914.0,
      "step": 1326
    },
    {
      "epoch": 2.5974534769833495,
      "grad_norm": 0.08572831448333348,
      "learning_rate": 4.936612015815237e-06,
      "loss": 0.4255,
      "num_tokens": 1257311586.0,
      "step": 1327
    },
    {
      "epoch": 2.5994123408423113,
      "grad_norm": 0.08490205995471985,
      "learning_rate": 4.930273609651983e-06,
      "loss": 0.4107,
      "num_tokens": 1258277915.0,
      "step": 1328
    },
    {
      "epoch": 2.601371204701273,
      "grad_norm": 0.08611105268036397,
      "learning_rate": 4.9239353155599894e-06,
      "loss": 0.4151,
      "num_tokens": 1259241846.0,
      "step": 1329
    },
    {
      "epoch": 2.603330068560235,
      "grad_norm": 0.08371809003984203,
      "learning_rate": 4.917597143726802e-06,
      "loss": 0.4089,
      "num_tokens": 1260175108.0,
      "step": 1330
    },
    {
      "epoch": 2.605288932419197,
      "grad_norm": 0.08308040980042024,
      "learning_rate": 4.911259104339771e-06,
      "loss": 0.4026,
      "num_tokens": 1261131640.0,
      "step": 1331
    },
    {
      "epoch": 2.6072477962781586,
      "grad_norm": 0.08405177452374366,
      "learning_rate": 4.904921207586025e-06,
      "loss": 0.4115,
      "num_tokens": 1262089996.0,
      "step": 1332
    },
    {
      "epoch": 2.6092066601371204,
      "grad_norm": 0.08749694772730317,
      "learning_rate": 4.89858346365247e-06,
      "loss": 0.4085,
      "num_tokens": 1263023865.0,
      "step": 1333
    },
    {
      "epoch": 2.6111655239960823,
      "grad_norm": 0.08982798796986177,
      "learning_rate": 4.892245882725766e-06,
      "loss": 0.4291,
      "num_tokens": 1263981873.0,
      "step": 1334
    },
    {
      "epoch": 2.613124387855044,
      "grad_norm": 0.08715085872050161,
      "learning_rate": 4.88590847499231e-06,
      "loss": 0.4107,
      "num_tokens": 1264934131.0,
      "step": 1335
    },
    {
      "epoch": 2.615083251714006,
      "grad_norm": 0.08348569251643344,
      "learning_rate": 4.879571250638221e-06,
      "loss": 0.4313,
      "num_tokens": 1265881116.0,
      "step": 1336
    },
    {
      "epoch": 2.6170421155729677,
      "grad_norm": 0.08581338711154483,
      "learning_rate": 4.873234219849322e-06,
      "loss": 0.4221,
      "num_tokens": 1266830553.0,
      "step": 1337
    },
    {
      "epoch": 2.6190009794319296,
      "grad_norm": 0.0829385128259635,
      "learning_rate": 4.866897392811127e-06,
      "loss": 0.4105,
      "num_tokens": 1267818561.0,
      "step": 1338
    },
    {
      "epoch": 2.6209598432908914,
      "grad_norm": 0.08724719932512014,
      "learning_rate": 4.860560779708821e-06,
      "loss": 0.4315,
      "num_tokens": 1268764694.0,
      "step": 1339
    },
    {
      "epoch": 2.622918707149853,
      "grad_norm": 0.08767125652252683,
      "learning_rate": 4.8542243907272445e-06,
      "loss": 0.4325,
      "num_tokens": 1269711703.0,
      "step": 1340
    },
    {
      "epoch": 2.624877571008815,
      "grad_norm": 0.09372252963411086,
      "learning_rate": 4.847888236050879e-06,
      "loss": 0.4168,
      "num_tokens": 1270612157.0,
      "step": 1341
    },
    {
      "epoch": 2.626836434867777,
      "grad_norm": 0.08405713241309688,
      "learning_rate": 4.8415523258638306e-06,
      "loss": 0.4241,
      "num_tokens": 1271576623.0,
      "step": 1342
    },
    {
      "epoch": 2.6287952987267387,
      "grad_norm": 0.09110545895885466,
      "learning_rate": 4.83521667034981e-06,
      "loss": 0.4222,
      "num_tokens": 1272522556.0,
      "step": 1343
    },
    {
      "epoch": 2.6307541625857,
      "grad_norm": 0.08877118077984512,
      "learning_rate": 4.828881279692118e-06,
      "loss": 0.4246,
      "num_tokens": 1273476719.0,
      "step": 1344
    },
    {
      "epoch": 2.6327130264446623,
      "grad_norm": 0.08532625599007881,
      "learning_rate": 4.822546164073635e-06,
      "loss": 0.4214,
      "num_tokens": 1274440726.0,
      "step": 1345
    },
    {
      "epoch": 2.6346718903036237,
      "grad_norm": 0.08601418206646565,
      "learning_rate": 4.816211333676792e-06,
      "loss": 0.4246,
      "num_tokens": 1275390298.0,
      "step": 1346
    },
    {
      "epoch": 2.636630754162586,
      "grad_norm": 0.0844819289001155,
      "learning_rate": 4.809876798683567e-06,
      "loss": 0.4257,
      "num_tokens": 1276383425.0,
      "step": 1347
    },
    {
      "epoch": 2.6385896180215473,
      "grad_norm": 0.08513663433905257,
      "learning_rate": 4.803542569275461e-06,
      "loss": 0.4068,
      "num_tokens": 1277332941.0,
      "step": 1348
    },
    {
      "epoch": 2.640548481880509,
      "grad_norm": 0.0843116595718068,
      "learning_rate": 4.797208655633484e-06,
      "loss": 0.4133,
      "num_tokens": 1278246978.0,
      "step": 1349
    },
    {
      "epoch": 2.642507345739471,
      "grad_norm": 0.08388506139157995,
      "learning_rate": 4.7908750679381386e-06,
      "loss": 0.4188,
      "num_tokens": 1279192153.0,
      "step": 1350
    },
    {
      "epoch": 2.644466209598433,
      "grad_norm": 0.09031713181014736,
      "learning_rate": 4.784541816369403e-06,
      "loss": 0.4243,
      "num_tokens": 1280128374.0,
      "step": 1351
    },
    {
      "epoch": 2.6464250734573946,
      "grad_norm": 0.08951869887136361,
      "learning_rate": 4.778208911106718e-06,
      "loss": 0.412,
      "num_tokens": 1281048912.0,
      "step": 1352
    },
    {
      "epoch": 2.6483839373163565,
      "grad_norm": 0.08454202196508212,
      "learning_rate": 4.771876362328962e-06,
      "loss": 0.4201,
      "num_tokens": 1282010060.0,
      "step": 1353
    },
    {
      "epoch": 2.6503428011753183,
      "grad_norm": 0.0841667336847675,
      "learning_rate": 4.765544180214447e-06,
      "loss": 0.4237,
      "num_tokens": 1282970943.0,
      "step": 1354
    },
    {
      "epoch": 2.65230166503428,
      "grad_norm": 0.0909589185939779,
      "learning_rate": 4.7592123749408904e-06,
      "loss": 0.4118,
      "num_tokens": 1283937168.0,
      "step": 1355
    },
    {
      "epoch": 2.654260528893242,
      "grad_norm": 0.08799476261814788,
      "learning_rate": 4.752880956685407e-06,
      "loss": 0.4262,
      "num_tokens": 1284878060.0,
      "step": 1356
    },
    {
      "epoch": 2.6562193927522038,
      "grad_norm": 0.08905967083052413,
      "learning_rate": 4.746549935624491e-06,
      "loss": 0.4315,
      "num_tokens": 1285839805.0,
      "step": 1357
    },
    {
      "epoch": 2.6581782566111656,
      "grad_norm": 0.08352530100551155,
      "learning_rate": 4.740219321933991e-06,
      "loss": 0.4111,
      "num_tokens": 1286780353.0,
      "step": 1358
    },
    {
      "epoch": 2.6601371204701274,
      "grad_norm": 0.08869645410002136,
      "learning_rate": 4.7338891257891085e-06,
      "loss": 0.425,
      "num_tokens": 1287753264.0,
      "step": 1359
    },
    {
      "epoch": 2.6620959843290892,
      "grad_norm": 0.08426981837149752,
      "learning_rate": 4.72755935736437e-06,
      "loss": 0.4075,
      "num_tokens": 1288663758.0,
      "step": 1360
    },
    {
      "epoch": 2.664054848188051,
      "grad_norm": 0.08424764410311818,
      "learning_rate": 4.721230026833615e-06,
      "loss": 0.4235,
      "num_tokens": 1289635743.0,
      "step": 1361
    },
    {
      "epoch": 2.666013712047013,
      "grad_norm": 0.08468526648090785,
      "learning_rate": 4.714901144369982e-06,
      "loss": 0.4133,
      "num_tokens": 1290615669.0,
      "step": 1362
    },
    {
      "epoch": 2.6679725759059747,
      "grad_norm": 0.08674077530770553,
      "learning_rate": 4.708572720145886e-06,
      "loss": 0.4262,
      "num_tokens": 1291577461.0,
      "step": 1363
    },
    {
      "epoch": 2.6699314397649365,
      "grad_norm": 0.0879098956928888,
      "learning_rate": 4.702244764333006e-06,
      "loss": 0.4291,
      "num_tokens": 1292498086.0,
      "step": 1364
    },
    {
      "epoch": 2.671890303623898,
      "grad_norm": 0.08513834128773212,
      "learning_rate": 4.695917287102267e-06,
      "loss": 0.4232,
      "num_tokens": 1293435143.0,
      "step": 1365
    },
    {
      "epoch": 2.67384916748286,
      "grad_norm": 0.08643486884422578,
      "learning_rate": 4.689590298623831e-06,
      "loss": 0.4247,
      "num_tokens": 1294378264.0,
      "step": 1366
    },
    {
      "epoch": 2.6758080313418215,
      "grad_norm": 0.08745936633848232,
      "learning_rate": 4.683263809067065e-06,
      "loss": 0.4132,
      "num_tokens": 1295322682.0,
      "step": 1367
    },
    {
      "epoch": 2.677766895200784,
      "grad_norm": 0.08535549088178032,
      "learning_rate": 4.676937828600542e-06,
      "loss": 0.4225,
      "num_tokens": 1296243735.0,
      "step": 1368
    },
    {
      "epoch": 2.679725759059745,
      "grad_norm": 0.08626784633146645,
      "learning_rate": 4.670612367392012e-06,
      "loss": 0.4259,
      "num_tokens": 1297168892.0,
      "step": 1369
    },
    {
      "epoch": 2.681684622918707,
      "grad_norm": 0.0865592449270761,
      "learning_rate": 4.664287435608393e-06,
      "loss": 0.4221,
      "num_tokens": 1298117653.0,
      "step": 1370
    },
    {
      "epoch": 2.683643486777669,
      "grad_norm": 0.08852648756125193,
      "learning_rate": 4.657963043415751e-06,
      "loss": 0.4135,
      "num_tokens": 1299091473.0,
      "step": 1371
    },
    {
      "epoch": 2.6856023506366307,
      "grad_norm": 0.08803130086410157,
      "learning_rate": 4.651639200979284e-06,
      "loss": 0.4289,
      "num_tokens": 1300046508.0,
      "step": 1372
    },
    {
      "epoch": 2.6875612144955925,
      "grad_norm": 0.08728864770970363,
      "learning_rate": 4.645315918463308e-06,
      "loss": 0.4311,
      "num_tokens": 1300982987.0,
      "step": 1373
    },
    {
      "epoch": 2.6895200783545543,
      "grad_norm": 0.08542773274024708,
      "learning_rate": 4.638993206031238e-06,
      "loss": 0.4254,
      "num_tokens": 1301936022.0,
      "step": 1374
    },
    {
      "epoch": 2.691478942213516,
      "grad_norm": 0.08754845624593002,
      "learning_rate": 4.632671073845574e-06,
      "loss": 0.4166,
      "num_tokens": 1302874334.0,
      "step": 1375
    },
    {
      "epoch": 2.693437806072478,
      "grad_norm": 0.09093501073826295,
      "learning_rate": 4.626349532067879e-06,
      "loss": 0.4217,
      "num_tokens": 1303814830.0,
      "step": 1376
    },
    {
      "epoch": 2.69539666993144,
      "grad_norm": 0.08949043713831442,
      "learning_rate": 4.620028590858773e-06,
      "loss": 0.4151,
      "num_tokens": 1304721413.0,
      "step": 1377
    },
    {
      "epoch": 2.6973555337904016,
      "grad_norm": 0.08427046742392057,
      "learning_rate": 4.613708260377909e-06,
      "loss": 0.4013,
      "num_tokens": 1305673260.0,
      "step": 1378
    },
    {
      "epoch": 2.6993143976493634,
      "grad_norm": 0.08786768786837658,
      "learning_rate": 4.607388550783955e-06,
      "loss": 0.4373,
      "num_tokens": 1306626238.0,
      "step": 1379
    },
    {
      "epoch": 2.7012732615083253,
      "grad_norm": 0.09030499429070131,
      "learning_rate": 4.601069472234584e-06,
      "loss": 0.4281,
      "num_tokens": 1307559897.0,
      "step": 1380
    },
    {
      "epoch": 2.703232125367287,
      "grad_norm": 0.08626019707122737,
      "learning_rate": 4.594751034886457e-06,
      "loss": 0.4078,
      "num_tokens": 1308463462.0,
      "step": 1381
    },
    {
      "epoch": 2.705190989226249,
      "grad_norm": 0.09258734342286046,
      "learning_rate": 4.588433248895196e-06,
      "loss": 0.4219,
      "num_tokens": 1309366698.0,
      "step": 1382
    },
    {
      "epoch": 2.7071498530852107,
      "grad_norm": 0.08996091511584008,
      "learning_rate": 4.5821161244153885e-06,
      "loss": 0.4327,
      "num_tokens": 1310328296.0,
      "step": 1383
    },
    {
      "epoch": 2.7091087169441725,
      "grad_norm": 0.086387115622684,
      "learning_rate": 4.575799671600547e-06,
      "loss": 0.4131,
      "num_tokens": 1311296001.0,
      "step": 1384
    },
    {
      "epoch": 2.7110675808031344,
      "grad_norm": 0.08251139972896727,
      "learning_rate": 4.569483900603109e-06,
      "loss": 0.4102,
      "num_tokens": 1312243912.0,
      "step": 1385
    },
    {
      "epoch": 2.7130264446620957,
      "grad_norm": 0.0863231830166427,
      "learning_rate": 4.56316882157442e-06,
      "loss": 0.4275,
      "num_tokens": 1313180857.0,
      "step": 1386
    },
    {
      "epoch": 2.714985308521058,
      "grad_norm": 0.08047455742931099,
      "learning_rate": 4.556854444664706e-06,
      "loss": 0.4145,
      "num_tokens": 1314147053.0,
      "step": 1387
    },
    {
      "epoch": 2.7169441723800194,
      "grad_norm": 0.08545677192809085,
      "learning_rate": 4.5505407800230715e-06,
      "loss": 0.4177,
      "num_tokens": 1315061581.0,
      "step": 1388
    },
    {
      "epoch": 2.7189030362389817,
      "grad_norm": 0.08748335614431532,
      "learning_rate": 4.544227837797471e-06,
      "loss": 0.4288,
      "num_tokens": 1315977754.0,
      "step": 1389
    },
    {
      "epoch": 2.720861900097943,
      "grad_norm": 0.08688618754690962,
      "learning_rate": 4.5379156281347e-06,
      "loss": 0.425,
      "num_tokens": 1316917513.0,
      "step": 1390
    },
    {
      "epoch": 2.722820763956905,
      "grad_norm": 0.08300622578487153,
      "learning_rate": 4.5316041611803775e-06,
      "loss": 0.4073,
      "num_tokens": 1317872866.0,
      "step": 1391
    },
    {
      "epoch": 2.7247796278158667,
      "grad_norm": 0.08851146163457943,
      "learning_rate": 4.525293447078927e-06,
      "loss": 0.4242,
      "num_tokens": 1318779793.0,
      "step": 1392
    },
    {
      "epoch": 2.7267384916748285,
      "grad_norm": 0.08246445330676379,
      "learning_rate": 4.518983495973561e-06,
      "loss": 0.4181,
      "num_tokens": 1319728628.0,
      "step": 1393
    },
    {
      "epoch": 2.7286973555337903,
      "grad_norm": 0.08732434459091828,
      "learning_rate": 4.512674318006268e-06,
      "loss": 0.4294,
      "num_tokens": 1320692125.0,
      "step": 1394
    },
    {
      "epoch": 2.730656219392752,
      "grad_norm": 0.08878515380250739,
      "learning_rate": 4.506365923317796e-06,
      "loss": 0.4266,
      "num_tokens": 1321627423.0,
      "step": 1395
    },
    {
      "epoch": 2.732615083251714,
      "grad_norm": 0.08932684385252695,
      "learning_rate": 4.500058322047626e-06,
      "loss": 0.4331,
      "num_tokens": 1322577825.0,
      "step": 1396
    },
    {
      "epoch": 2.734573947110676,
      "grad_norm": 0.08577537821073135,
      "learning_rate": 4.493751524333971e-06,
      "loss": 0.4212,
      "num_tokens": 1323525074.0,
      "step": 1397
    },
    {
      "epoch": 2.7365328109696376,
      "grad_norm": 0.08887611706237718,
      "learning_rate": 4.487445540313752e-06,
      "loss": 0.4227,
      "num_tokens": 1324431844.0,
      "step": 1398
    },
    {
      "epoch": 2.7384916748285995,
      "grad_norm": 0.08666775303969908,
      "learning_rate": 4.481140380122576e-06,
      "loss": 0.4297,
      "num_tokens": 1325369039.0,
      "step": 1399
    },
    {
      "epoch": 2.7404505386875613,
      "grad_norm": 0.08570871349890383,
      "learning_rate": 4.474836053894735e-06,
      "loss": 0.4238,
      "num_tokens": 1326314871.0,
      "step": 1400
    },
    {
      "epoch": 2.742409402546523,
      "grad_norm": 0.0857332714167006,
      "learning_rate": 4.468532571763174e-06,
      "loss": 0.4227,
      "num_tokens": 1327305903.0,
      "step": 1401
    },
    {
      "epoch": 2.744368266405485,
      "grad_norm": 0.08624918511174663,
      "learning_rate": 4.462229943859481e-06,
      "loss": 0.4158,
      "num_tokens": 1328242905.0,
      "step": 1402
    },
    {
      "epoch": 2.7463271302644467,
      "grad_norm": 0.0906246795423964,
      "learning_rate": 4.455928180313876e-06,
      "loss": 0.4349,
      "num_tokens": 1329196365.0,
      "step": 1403
    },
    {
      "epoch": 2.7482859941234086,
      "grad_norm": 0.0868823878426707,
      "learning_rate": 4.4496272912551845e-06,
      "loss": 0.4238,
      "num_tokens": 1330152019.0,
      "step": 1404
    },
    {
      "epoch": 2.75024485798237,
      "grad_norm": 0.08839265708997726,
      "learning_rate": 4.44332728681083e-06,
      "loss": 0.4274,
      "num_tokens": 1331141273.0,
      "step": 1405
    },
    {
      "epoch": 2.752203721841332,
      "grad_norm": 0.08436293399608402,
      "learning_rate": 4.4370281771068135e-06,
      "loss": 0.4262,
      "num_tokens": 1332088742.0,
      "step": 1406
    },
    {
      "epoch": 2.7541625857002936,
      "grad_norm": 0.08799429873505175,
      "learning_rate": 4.430729972267695e-06,
      "loss": 0.4084,
      "num_tokens": 1333004796.0,
      "step": 1407
    },
    {
      "epoch": 2.756121449559256,
      "grad_norm": 0.08826510870107931,
      "learning_rate": 4.424432682416585e-06,
      "loss": 0.436,
      "num_tokens": 1333929542.0,
      "step": 1408
    },
    {
      "epoch": 2.7580803134182172,
      "grad_norm": 0.08445417563135003,
      "learning_rate": 4.418136317675119e-06,
      "loss": 0.4203,
      "num_tokens": 1334870165.0,
      "step": 1409
    },
    {
      "epoch": 2.7600391772771795,
      "grad_norm": 0.08253451675574697,
      "learning_rate": 4.411840888163449e-06,
      "loss": 0.4097,
      "num_tokens": 1335840737.0,
      "step": 1410
    },
    {
      "epoch": 2.761998041136141,
      "grad_norm": 0.08519991754388842,
      "learning_rate": 4.405546404000219e-06,
      "loss": 0.4294,
      "num_tokens": 1336789179.0,
      "step": 1411
    },
    {
      "epoch": 2.7639569049951027,
      "grad_norm": 0.08726131229535722,
      "learning_rate": 4.39925287530256e-06,
      "loss": 0.4093,
      "num_tokens": 1337739721.0,
      "step": 1412
    },
    {
      "epoch": 2.7659157688540645,
      "grad_norm": 0.08397532110608683,
      "learning_rate": 4.392960312186061e-06,
      "loss": 0.4333,
      "num_tokens": 1338727437.0,
      "step": 1413
    },
    {
      "epoch": 2.7678746327130264,
      "grad_norm": 0.08362502030987641,
      "learning_rate": 4.386668724764763e-06,
      "loss": 0.4031,
      "num_tokens": 1339705363.0,
      "step": 1414
    },
    {
      "epoch": 2.769833496571988,
      "grad_norm": 0.08847889470054021,
      "learning_rate": 4.380378123151139e-06,
      "loss": 0.4323,
      "num_tokens": 1340627521.0,
      "step": 1415
    },
    {
      "epoch": 2.77179236043095,
      "grad_norm": 0.08874335681524745,
      "learning_rate": 4.374088517456074e-06,
      "loss": 0.4297,
      "num_tokens": 1341568553.0,
      "step": 1416
    },
    {
      "epoch": 2.773751224289912,
      "grad_norm": 0.08843479663044887,
      "learning_rate": 4.367799917788855e-06,
      "loss": 0.4132,
      "num_tokens": 1342488029.0,
      "step": 1417
    },
    {
      "epoch": 2.7757100881488737,
      "grad_norm": 0.08588344536442485,
      "learning_rate": 4.361512334257153e-06,
      "loss": 0.424,
      "num_tokens": 1343431146.0,
      "step": 1418
    },
    {
      "epoch": 2.7776689520078355,
      "grad_norm": 0.08419682853981998,
      "learning_rate": 4.355225776967004e-06,
      "loss": 0.4122,
      "num_tokens": 1344386333.0,
      "step": 1419
    },
    {
      "epoch": 2.7796278158667973,
      "grad_norm": 0.08614637232405228,
      "learning_rate": 4.348940256022793e-06,
      "loss": 0.4191,
      "num_tokens": 1345337799.0,
      "step": 1420
    },
    {
      "epoch": 2.781586679725759,
      "grad_norm": 0.08609453928325388,
      "learning_rate": 4.342655781527242e-06,
      "loss": 0.4279,
      "num_tokens": 1346278180.0,
      "step": 1421
    },
    {
      "epoch": 2.783545543584721,
      "grad_norm": 0.08399908294756268,
      "learning_rate": 4.336372363581391e-06,
      "loss": 0.3971,
      "num_tokens": 1347235109.0,
      "step": 1422
    },
    {
      "epoch": 2.7855044074436828,
      "grad_norm": 0.08607710324760467,
      "learning_rate": 4.330090012284579e-06,
      "loss": 0.4106,
      "num_tokens": 1348195228.0,
      "step": 1423
    },
    {
      "epoch": 2.7874632713026446,
      "grad_norm": 0.08830464407820007,
      "learning_rate": 4.323808737734434e-06,
      "loss": 0.4243,
      "num_tokens": 1349140603.0,
      "step": 1424
    },
    {
      "epoch": 2.7894221351616064,
      "grad_norm": 0.08465188461532855,
      "learning_rate": 4.317528550026852e-06,
      "loss": 0.4166,
      "num_tokens": 1350080074.0,
      "step": 1425
    },
    {
      "epoch": 2.791380999020568,
      "grad_norm": 0.0853482332122759,
      "learning_rate": 4.3112494592559796e-06,
      "loss": 0.4375,
      "num_tokens": 1351042372.0,
      "step": 1426
    },
    {
      "epoch": 2.79333986287953,
      "grad_norm": 0.08623633168182623,
      "learning_rate": 4.304971475514204e-06,
      "loss": 0.4109,
      "num_tokens": 1351976139.0,
      "step": 1427
    },
    {
      "epoch": 2.7952987267384914,
      "grad_norm": 0.09794545108240034,
      "learning_rate": 4.298694608892134e-06,
      "loss": 0.4276,
      "num_tokens": 1352887962.0,
      "step": 1428
    },
    {
      "epoch": 2.7972575905974537,
      "grad_norm": 0.08865652806882626,
      "learning_rate": 4.292418869478577e-06,
      "loss": 0.4186,
      "num_tokens": 1353811342.0,
      "step": 1429
    },
    {
      "epoch": 2.799216454456415,
      "grad_norm": 0.08408678839635018,
      "learning_rate": 4.286144267360534e-06,
      "loss": 0.433,
      "num_tokens": 1354766417.0,
      "step": 1430
    },
    {
      "epoch": 2.801175318315377,
      "grad_norm": 0.08467383283729038,
      "learning_rate": 4.2798708126231756e-06,
      "loss": 0.4166,
      "num_tokens": 1355719962.0,
      "step": 1431
    },
    {
      "epoch": 2.8031341821743387,
      "grad_norm": 0.08517287543720582,
      "learning_rate": 4.2735985153498295e-06,
      "loss": 0.4185,
      "num_tokens": 1356676496.0,
      "step": 1432
    },
    {
      "epoch": 2.8050930460333006,
      "grad_norm": 0.08571886488062591,
      "learning_rate": 4.267327385621961e-06,
      "loss": 0.4272,
      "num_tokens": 1357628861.0,
      "step": 1433
    },
    {
      "epoch": 2.8070519098922624,
      "grad_norm": 0.08645861300094004,
      "learning_rate": 4.2610574335191615e-06,
      "loss": 0.4181,
      "num_tokens": 1358579701.0,
      "step": 1434
    },
    {
      "epoch": 2.809010773751224,
      "grad_norm": 0.08765285916897982,
      "learning_rate": 4.254788669119127e-06,
      "loss": 0.4426,
      "num_tokens": 1359550557.0,
      "step": 1435
    },
    {
      "epoch": 2.810969637610186,
      "grad_norm": 0.08613431350147782,
      "learning_rate": 4.248521102497649e-06,
      "loss": 0.423,
      "num_tokens": 1360494519.0,
      "step": 1436
    },
    {
      "epoch": 2.812928501469148,
      "grad_norm": 0.08199107687960296,
      "learning_rate": 4.242254743728586e-06,
      "loss": 0.4206,
      "num_tokens": 1361469588.0,
      "step": 1437
    },
    {
      "epoch": 2.8148873653281097,
      "grad_norm": 0.08731083624025529,
      "learning_rate": 4.235989602883862e-06,
      "loss": 0.4141,
      "num_tokens": 1362438118.0,
      "step": 1438
    },
    {
      "epoch": 2.8168462291870715,
      "grad_norm": 0.0844344702064925,
      "learning_rate": 4.22972569003344e-06,
      "loss": 0.4243,
      "num_tokens": 1363403750.0,
      "step": 1439
    },
    {
      "epoch": 2.8188050930460333,
      "grad_norm": 0.08283393914695329,
      "learning_rate": 4.223463015245311e-06,
      "loss": 0.417,
      "num_tokens": 1364397979.0,
      "step": 1440
    },
    {
      "epoch": 2.820763956904995,
      "grad_norm": 0.08904784088420996,
      "learning_rate": 4.217201588585475e-06,
      "loss": 0.442,
      "num_tokens": 1365365382.0,
      "step": 1441
    },
    {
      "epoch": 2.822722820763957,
      "grad_norm": 0.08607038877058544,
      "learning_rate": 4.210941420117929e-06,
      "loss": 0.4226,
      "num_tokens": 1366346611.0,
      "step": 1442
    },
    {
      "epoch": 2.824681684622919,
      "grad_norm": 0.08789243523209957,
      "learning_rate": 4.204682519904641e-06,
      "loss": 0.4252,
      "num_tokens": 1367319906.0,
      "step": 1443
    },
    {
      "epoch": 2.8266405484818806,
      "grad_norm": 0.0846557832077161,
      "learning_rate": 4.198424898005546e-06,
      "loss": 0.4267,
      "num_tokens": 1368275343.0,
      "step": 1444
    },
    {
      "epoch": 2.8285994123408424,
      "grad_norm": 0.08386220182738313,
      "learning_rate": 4.192168564478524e-06,
      "loss": 0.4133,
      "num_tokens": 1369274064.0,
      "step": 1445
    },
    {
      "epoch": 2.8305582761998043,
      "grad_norm": 0.08420007395172191,
      "learning_rate": 4.185913529379381e-06,
      "loss": 0.4157,
      "num_tokens": 1370227156.0,
      "step": 1446
    },
    {
      "epoch": 2.8325171400587656,
      "grad_norm": 0.08878866775347316,
      "learning_rate": 4.1796598027618406e-06,
      "loss": 0.4293,
      "num_tokens": 1371180761.0,
      "step": 1447
    },
    {
      "epoch": 2.834476003917728,
      "grad_norm": 0.08371984206402905,
      "learning_rate": 4.173407394677519e-06,
      "loss": 0.4233,
      "num_tokens": 1372171555.0,
      "step": 1448
    },
    {
      "epoch": 2.8364348677766893,
      "grad_norm": 0.08440853168229509,
      "learning_rate": 4.167156315175915e-06,
      "loss": 0.4038,
      "num_tokens": 1373114490.0,
      "step": 1449
    },
    {
      "epoch": 2.8383937316356516,
      "grad_norm": 0.08860207759262753,
      "learning_rate": 4.160906574304392e-06,
      "loss": 0.4165,
      "num_tokens": 1374077054.0,
      "step": 1450
    },
    {
      "epoch": 2.840352595494613,
      "grad_norm": 0.08813572880219034,
      "learning_rate": 4.154658182108163e-06,
      "loss": 0.4236,
      "num_tokens": 1375000445.0,
      "step": 1451
    },
    {
      "epoch": 2.8423114593535748,
      "grad_norm": 0.08708887346865538,
      "learning_rate": 4.148411148630271e-06,
      "loss": 0.4467,
      "num_tokens": 1375974231.0,
      "step": 1452
    },
    {
      "epoch": 2.8442703232125366,
      "grad_norm": 0.08739088490670505,
      "learning_rate": 4.142165483911575e-06,
      "loss": 0.4485,
      "num_tokens": 1376927049.0,
      "step": 1453
    },
    {
      "epoch": 2.8462291870714984,
      "grad_norm": 0.089933453616257,
      "learning_rate": 4.135921197990737e-06,
      "loss": 0.4181,
      "num_tokens": 1377873883.0,
      "step": 1454
    },
    {
      "epoch": 2.8481880509304602,
      "grad_norm": 0.08555850125055711,
      "learning_rate": 4.129678300904201e-06,
      "loss": 0.4252,
      "num_tokens": 1378793565.0,
      "step": 1455
    },
    {
      "epoch": 2.850146914789422,
      "grad_norm": 0.08400564416104261,
      "learning_rate": 4.123436802686176e-06,
      "loss": 0.4217,
      "num_tokens": 1379745839.0,
      "step": 1456
    },
    {
      "epoch": 2.852105778648384,
      "grad_norm": 0.08704734713245992,
      "learning_rate": 4.117196713368629e-06,
      "loss": 0.4292,
      "num_tokens": 1380716393.0,
      "step": 1457
    },
    {
      "epoch": 2.8540646425073457,
      "grad_norm": 0.08667779032238235,
      "learning_rate": 4.110958042981256e-06,
      "loss": 0.4191,
      "num_tokens": 1381659575.0,
      "step": 1458
    },
    {
      "epoch": 2.8560235063663075,
      "grad_norm": 0.08312185377817422,
      "learning_rate": 4.104720801551475e-06,
      "loss": 0.4211,
      "num_tokens": 1382617119.0,
      "step": 1459
    },
    {
      "epoch": 2.8579823702252694,
      "grad_norm": 0.08361729500249165,
      "learning_rate": 4.098484999104409e-06,
      "loss": 0.4199,
      "num_tokens": 1383604919.0,
      "step": 1460
    },
    {
      "epoch": 2.859941234084231,
      "grad_norm": 0.0883469687869528,
      "learning_rate": 4.092250645662867e-06,
      "loss": 0.4183,
      "num_tokens": 1384539101.0,
      "step": 1461
    },
    {
      "epoch": 2.861900097943193,
      "grad_norm": 0.08365888362340465,
      "learning_rate": 4.086017751247328e-06,
      "loss": 0.4154,
      "num_tokens": 1385494745.0,
      "step": 1462
    },
    {
      "epoch": 2.863858961802155,
      "grad_norm": 0.0857829252017472,
      "learning_rate": 4.079786325875924e-06,
      "loss": 0.4243,
      "num_tokens": 1386459640.0,
      "step": 1463
    },
    {
      "epoch": 2.8658178256611166,
      "grad_norm": 0.08579543156355285,
      "learning_rate": 4.073556379564429e-06,
      "loss": 0.4157,
      "num_tokens": 1387422495.0,
      "step": 1464
    },
    {
      "epoch": 2.8677766895200785,
      "grad_norm": 0.08403844361349817,
      "learning_rate": 4.067327922326242e-06,
      "loss": 0.4176,
      "num_tokens": 1388405073.0,
      "step": 1465
    },
    {
      "epoch": 2.8697355533790403,
      "grad_norm": 0.08300366668369293,
      "learning_rate": 4.0611009641723615e-06,
      "loss": 0.4197,
      "num_tokens": 1389346663.0,
      "step": 1466
    },
    {
      "epoch": 2.871694417238002,
      "grad_norm": 0.08600928440720594,
      "learning_rate": 4.0548755151113824e-06,
      "loss": 0.4198,
      "num_tokens": 1390300507.0,
      "step": 1467
    },
    {
      "epoch": 2.8736532810969635,
      "grad_norm": 0.08502182035386478,
      "learning_rate": 4.048651585149473e-06,
      "loss": 0.4164,
      "num_tokens": 1391204890.0,
      "step": 1468
    },
    {
      "epoch": 2.8756121449559258,
      "grad_norm": 0.08195440822126236,
      "learning_rate": 4.042429184290357e-06,
      "loss": 0.4112,
      "num_tokens": 1392185859.0,
      "step": 1469
    },
    {
      "epoch": 2.877571008814887,
      "grad_norm": 0.090892916068915,
      "learning_rate": 4.036208322535304e-06,
      "loss": 0.4295,
      "num_tokens": 1393131912.0,
      "step": 1470
    },
    {
      "epoch": 2.8795298726738494,
      "grad_norm": 0.0862667451758762,
      "learning_rate": 4.0299890098831096e-06,
      "loss": 0.4252,
      "num_tokens": 1394114081.0,
      "step": 1471
    },
    {
      "epoch": 2.881488736532811,
      "grad_norm": 0.08393258626423705,
      "learning_rate": 4.023771256330075e-06,
      "loss": 0.4228,
      "num_tokens": 1395089769.0,
      "step": 1472
    },
    {
      "epoch": 2.8834476003917726,
      "grad_norm": 0.08272992398126203,
      "learning_rate": 4.01755507187e-06,
      "loss": 0.4136,
      "num_tokens": 1396066601.0,
      "step": 1473
    },
    {
      "epoch": 2.8854064642507344,
      "grad_norm": 0.08551055948053857,
      "learning_rate": 4.011340466494162e-06,
      "loss": 0.4243,
      "num_tokens": 1397041867.0,
      "step": 1474
    },
    {
      "epoch": 2.8873653281096963,
      "grad_norm": 0.0873784242015668,
      "learning_rate": 4.005127450191299e-06,
      "loss": 0.4314,
      "num_tokens": 1398009067.0,
      "step": 1475
    },
    {
      "epoch": 2.889324191968658,
      "grad_norm": 0.08921150252123267,
      "learning_rate": 3.998916032947594e-06,
      "loss": 0.426,
      "num_tokens": 1398960877.0,
      "step": 1476
    },
    {
      "epoch": 2.89128305582762,
      "grad_norm": 0.08657560218415009,
      "learning_rate": 3.9927062247466625e-06,
      "loss": 0.4223,
      "num_tokens": 1399911107.0,
      "step": 1477
    },
    {
      "epoch": 2.8932419196865817,
      "grad_norm": 0.08431648518237164,
      "learning_rate": 3.986498035569533e-06,
      "loss": 0.422,
      "num_tokens": 1400868445.0,
      "step": 1478
    },
    {
      "epoch": 2.8952007835455436,
      "grad_norm": 0.08964906628252636,
      "learning_rate": 3.980291475394629e-06,
      "loss": 0.421,
      "num_tokens": 1401832152.0,
      "step": 1479
    },
    {
      "epoch": 2.8971596474045054,
      "grad_norm": 0.08505233137308539,
      "learning_rate": 3.974086554197759e-06,
      "loss": 0.4335,
      "num_tokens": 1402819847.0,
      "step": 1480
    },
    {
      "epoch": 2.899118511263467,
      "grad_norm": 0.08384794297954219,
      "learning_rate": 3.967883281952098e-06,
      "loss": 0.4249,
      "num_tokens": 1403772945.0,
      "step": 1481
    },
    {
      "epoch": 2.901077375122429,
      "grad_norm": 0.08397983241421397,
      "learning_rate": 3.9616816686281636e-06,
      "loss": 0.4198,
      "num_tokens": 1404751662.0,
      "step": 1482
    },
    {
      "epoch": 2.903036238981391,
      "grad_norm": 0.08733913341135212,
      "learning_rate": 3.955481724193817e-06,
      "loss": 0.4197,
      "num_tokens": 1405696873.0,
      "step": 1483
    },
    {
      "epoch": 2.9049951028403527,
      "grad_norm": 0.08910912116182969,
      "learning_rate": 3.9492834586142306e-06,
      "loss": 0.4219,
      "num_tokens": 1406591142.0,
      "step": 1484
    },
    {
      "epoch": 2.9069539666993145,
      "grad_norm": 0.09212989650364477,
      "learning_rate": 3.9430868818518786e-06,
      "loss": 0.413,
      "num_tokens": 1407541320.0,
      "step": 1485
    },
    {
      "epoch": 2.9089128305582763,
      "grad_norm": 0.08293600214295944,
      "learning_rate": 3.936892003866523e-06,
      "loss": 0.4002,
      "num_tokens": 1408502771.0,
      "step": 1486
    },
    {
      "epoch": 2.910871694417238,
      "grad_norm": 0.08590134879504807,
      "learning_rate": 3.930698834615195e-06,
      "loss": 0.4142,
      "num_tokens": 1409425939.0,
      "step": 1487
    },
    {
      "epoch": 2.9128305582762,
      "grad_norm": 0.08500131829727482,
      "learning_rate": 3.924507384052177e-06,
      "loss": 0.4266,
      "num_tokens": 1410401535.0,
      "step": 1488
    },
    {
      "epoch": 2.9147894221351613,
      "grad_norm": 0.08585076216249231,
      "learning_rate": 3.918317662128994e-06,
      "loss": 0.4295,
      "num_tokens": 1411329453.0,
      "step": 1489
    },
    {
      "epoch": 2.9167482859941236,
      "grad_norm": 0.08610441103438081,
      "learning_rate": 3.912129678794384e-06,
      "loss": 0.3994,
      "num_tokens": 1412308096.0,
      "step": 1490
    },
    {
      "epoch": 2.918707149853085,
      "grad_norm": 0.08848155507666867,
      "learning_rate": 3.905943443994299e-06,
      "loss": 0.4258,
      "num_tokens": 1413260722.0,
      "step": 1491
    },
    {
      "epoch": 2.9206660137120473,
      "grad_norm": 0.08558103907972713,
      "learning_rate": 3.899758967671879e-06,
      "loss": 0.4266,
      "num_tokens": 1414194902.0,
      "step": 1492
    },
    {
      "epoch": 2.9226248775710086,
      "grad_norm": 0.08695765787369268,
      "learning_rate": 3.893576259767431e-06,
      "loss": 0.4224,
      "num_tokens": 1415138861.0,
      "step": 1493
    },
    {
      "epoch": 2.9245837414299705,
      "grad_norm": 0.08563015648204321,
      "learning_rate": 3.887395330218429e-06,
      "loss": 0.4224,
      "num_tokens": 1416099459.0,
      "step": 1494
    },
    {
      "epoch": 2.9265426052889323,
      "grad_norm": 0.08342369327522459,
      "learning_rate": 3.8812161889594826e-06,
      "loss": 0.4152,
      "num_tokens": 1417054600.0,
      "step": 1495
    },
    {
      "epoch": 2.928501469147894,
      "grad_norm": 0.08796149298875809,
      "learning_rate": 3.875038845922329e-06,
      "loss": 0.4304,
      "num_tokens": 1418009232.0,
      "step": 1496
    },
    {
      "epoch": 2.930460333006856,
      "grad_norm": 0.08731540375781419,
      "learning_rate": 3.868863311035814e-06,
      "loss": 0.4346,
      "num_tokens": 1418974959.0,
      "step": 1497
    },
    {
      "epoch": 2.9324191968658178,
      "grad_norm": 0.08397040254879076,
      "learning_rate": 3.8626895942258804e-06,
      "loss": 0.4242,
      "num_tokens": 1419928020.0,
      "step": 1498
    },
    {
      "epoch": 2.9343780607247796,
      "grad_norm": 0.08829494871058116,
      "learning_rate": 3.856517705415543e-06,
      "loss": 0.4378,
      "num_tokens": 1420887581.0,
      "step": 1499
    },
    {
      "epoch": 2.9363369245837414,
      "grad_norm": 0.09057604720036475,
      "learning_rate": 3.850347654524884e-06,
      "loss": 0.4177,
      "num_tokens": 1421844392.0,
      "step": 1500
    },
    {
      "epoch": 2.9382957884427032,
      "grad_norm": 0.08520793043642402,
      "learning_rate": 3.84417945147103e-06,
      "loss": 0.4254,
      "num_tokens": 1422766353.0,
      "step": 1501
    },
    {
      "epoch": 2.940254652301665,
      "grad_norm": 0.08427816943750237,
      "learning_rate": 3.838013106168134e-06,
      "loss": 0.4078,
      "num_tokens": 1423742891.0,
      "step": 1502
    },
    {
      "epoch": 2.942213516160627,
      "grad_norm": 0.08386793424652861,
      "learning_rate": 3.831848628527368e-06,
      "loss": 0.4109,
      "num_tokens": 1424671790.0,
      "step": 1503
    },
    {
      "epoch": 2.9441723800195887,
      "grad_norm": 0.08416905709344,
      "learning_rate": 3.825686028456901e-06,
      "loss": 0.4178,
      "num_tokens": 1425628029.0,
      "step": 1504
    },
    {
      "epoch": 2.9461312438785505,
      "grad_norm": 0.08322343356214816,
      "learning_rate": 3.8195253158618815e-06,
      "loss": 0.4131,
      "num_tokens": 1426597089.0,
      "step": 1505
    },
    {
      "epoch": 2.9480901077375123,
      "grad_norm": 0.08331145269417914,
      "learning_rate": 3.813366500644426e-06,
      "loss": 0.4221,
      "num_tokens": 1427529099.0,
      "step": 1506
    },
    {
      "epoch": 2.950048971596474,
      "grad_norm": 0.08572693265651272,
      "learning_rate": 3.8072095927036034e-06,
      "loss": 0.434,
      "num_tokens": 1428501980.0,
      "step": 1507
    },
    {
      "epoch": 2.952007835455436,
      "grad_norm": 0.08641632406249106,
      "learning_rate": 3.8010546019354122e-06,
      "loss": 0.4251,
      "num_tokens": 1429428197.0,
      "step": 1508
    },
    {
      "epoch": 2.953966699314398,
      "grad_norm": 0.08316652705389987,
      "learning_rate": 3.7949015382327737e-06,
      "loss": 0.4088,
      "num_tokens": 1430395899.0,
      "step": 1509
    },
    {
      "epoch": 2.955925563173359,
      "grad_norm": 0.08947342170867507,
      "learning_rate": 3.788750411485511e-06,
      "loss": 0.4276,
      "num_tokens": 1431329285.0,
      "step": 1510
    },
    {
      "epoch": 2.9578844270323215,
      "grad_norm": 0.08295971476456522,
      "learning_rate": 3.782601231580332e-06,
      "loss": 0.4341,
      "num_tokens": 1432276277.0,
      "step": 1511
    },
    {
      "epoch": 2.959843290891283,
      "grad_norm": 0.08161559699458866,
      "learning_rate": 3.7764540084008166e-06,
      "loss": 0.4237,
      "num_tokens": 1433245547.0,
      "step": 1512
    },
    {
      "epoch": 2.961802154750245,
      "grad_norm": 0.09079736480700806,
      "learning_rate": 3.770308751827402e-06,
      "loss": 0.4183,
      "num_tokens": 1434187255.0,
      "step": 1513
    },
    {
      "epoch": 2.9637610186092065,
      "grad_norm": 0.0842775167541979,
      "learning_rate": 3.764165471737359e-06,
      "loss": 0.4121,
      "num_tokens": 1435139934.0,
      "step": 1514
    },
    {
      "epoch": 2.9657198824681683,
      "grad_norm": 0.0871417251165915,
      "learning_rate": 3.758024178004789e-06,
      "loss": 0.4232,
      "num_tokens": 1436085191.0,
      "step": 1515
    },
    {
      "epoch": 2.96767874632713,
      "grad_norm": 0.08558891087797421,
      "learning_rate": 3.751884880500591e-06,
      "loss": 0.4135,
      "num_tokens": 1437010923.0,
      "step": 1516
    },
    {
      "epoch": 2.969637610186092,
      "grad_norm": 0.08392466431363356,
      "learning_rate": 3.7457475890924656e-06,
      "loss": 0.4119,
      "num_tokens": 1437971337.0,
      "step": 1517
    },
    {
      "epoch": 2.971596474045054,
      "grad_norm": 0.0854219137398358,
      "learning_rate": 3.7396123136448824e-06,
      "loss": 0.4339,
      "num_tokens": 1438928174.0,
      "step": 1518
    },
    {
      "epoch": 2.9735553379040156,
      "grad_norm": 0.08432553105531877,
      "learning_rate": 3.7334790640190733e-06,
      "loss": 0.4185,
      "num_tokens": 1439879746.0,
      "step": 1519
    },
    {
      "epoch": 2.9755142017629774,
      "grad_norm": 0.0853445665439681,
      "learning_rate": 3.727347850073012e-06,
      "loss": 0.4119,
      "num_tokens": 1440799387.0,
      "step": 1520
    },
    {
      "epoch": 2.9774730656219393,
      "grad_norm": 0.0832484637801818,
      "learning_rate": 3.721218681661406e-06,
      "loss": 0.4194,
      "num_tokens": 1441766089.0,
      "step": 1521
    },
    {
      "epoch": 2.979431929480901,
      "grad_norm": 0.08445308558444874,
      "learning_rate": 3.7150915686356657e-06,
      "loss": 0.4244,
      "num_tokens": 1442732128.0,
      "step": 1522
    },
    {
      "epoch": 2.981390793339863,
      "grad_norm": 0.08667681178857112,
      "learning_rate": 3.708966520843906e-06,
      "loss": 0.4144,
      "num_tokens": 1443674826.0,
      "step": 1523
    },
    {
      "epoch": 2.9833496571988247,
      "grad_norm": 0.09038005129679207,
      "learning_rate": 3.70284354813092e-06,
      "loss": 0.4268,
      "num_tokens": 1444668576.0,
      "step": 1524
    },
    {
      "epoch": 2.9853085210577865,
      "grad_norm": 0.08268635260702975,
      "learning_rate": 3.6967226603381624e-06,
      "loss": 0.4137,
      "num_tokens": 1445635950.0,
      "step": 1525
    },
    {
      "epoch": 2.9872673849167484,
      "grad_norm": 0.0842139646863267,
      "learning_rate": 3.690603867303741e-06,
      "loss": 0.4133,
      "num_tokens": 1446543169.0,
      "step": 1526
    },
    {
      "epoch": 2.98922624877571,
      "grad_norm": 0.08709311275860508,
      "learning_rate": 3.6844871788623946e-06,
      "loss": 0.4116,
      "num_tokens": 1447487974.0,
      "step": 1527
    },
    {
      "epoch": 2.991185112634672,
      "grad_norm": 0.0857642104307539,
      "learning_rate": 3.6783726048454794e-06,
      "loss": 0.4199,
      "num_tokens": 1448428128.0,
      "step": 1528
    },
    {
      "epoch": 2.993143976493634,
      "grad_norm": 0.08473377933023256,
      "learning_rate": 3.6722601550809534e-06,
      "loss": 0.429,
      "num_tokens": 1449374978.0,
      "step": 1529
    },
    {
      "epoch": 2.9951028403525957,
      "grad_norm": 0.08998178379263423,
      "learning_rate": 3.6661498393933612e-06,
      "loss": 0.438,
      "num_tokens": 1450341064.0,
      "step": 1530
    },
    {
      "epoch": 2.997061704211557,
      "grad_norm": 0.08697424506888644,
      "learning_rate": 3.6600416676038144e-06,
      "loss": 0.4365,
      "num_tokens": 1451292835.0,
      "step": 1531
    },
    {
      "epoch": 2.9990205680705193,
      "grad_norm": 0.08657942646985065,
      "learning_rate": 3.6539356495299816e-06,
      "loss": 0.4194,
      "num_tokens": 1452234989.0,
      "step": 1532
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.08657942646985065,
      "learning_rate": 3.6478317949860696e-06,
      "loss": 0.4189,
      "num_tokens": 1452695717.0,
      "step": 1533
    },
    {
      "epoch": 3.001958863858962,
      "grad_norm": 0.13984079896830842,
      "learning_rate": 3.641730113782807e-06,
      "loss": 0.4149,
      "num_tokens": 1453658103.0,
      "step": 1534
    },
    {
      "epoch": 3.0039177277179236,
      "grad_norm": 0.08377869426784214,
      "learning_rate": 3.6356306157274274e-06,
      "loss": 0.412,
      "num_tokens": 1454618540.0,
      "step": 1535
    },
    {
      "epoch": 3.0058765915768855,
      "grad_norm": 0.08664720922269122,
      "learning_rate": 3.629533310623658e-06,
      "loss": 0.4126,
      "num_tokens": 1455545489.0,
      "step": 1536
    },
    {
      "epoch": 3.0078354554358473,
      "grad_norm": 0.08623098846156026,
      "learning_rate": 3.623438208271701e-06,
      "loss": 0.4137,
      "num_tokens": 1456513429.0,
      "step": 1537
    },
    {
      "epoch": 3.009794319294809,
      "grad_norm": 0.08688930262599263,
      "learning_rate": 3.6173453184682173e-06,
      "loss": 0.4174,
      "num_tokens": 1457483318.0,
      "step": 1538
    },
    {
      "epoch": 3.011753183153771,
      "grad_norm": 0.08597988636751401,
      "learning_rate": 3.611254651006311e-06,
      "loss": 0.424,
      "num_tokens": 1458412653.0,
      "step": 1539
    },
    {
      "epoch": 3.0137120470127328,
      "grad_norm": 0.08696752958556553,
      "learning_rate": 3.605166215675516e-06,
      "loss": 0.4008,
      "num_tokens": 1459381365.0,
      "step": 1540
    },
    {
      "epoch": 3.0156709108716946,
      "grad_norm": 0.08261014918185572,
      "learning_rate": 3.5990800222617774e-06,
      "loss": 0.425,
      "num_tokens": 1460343126.0,
      "step": 1541
    },
    {
      "epoch": 3.0176297747306564,
      "grad_norm": 0.08413938130327642,
      "learning_rate": 3.5929960805474386e-06,
      "loss": 0.4107,
      "num_tokens": 1461260689.0,
      "step": 1542
    },
    {
      "epoch": 3.0195886385896182,
      "grad_norm": 0.08383172821306109,
      "learning_rate": 3.5869144003112177e-06,
      "loss": 0.414,
      "num_tokens": 1462227916.0,
      "step": 1543
    },
    {
      "epoch": 3.0215475024485796,
      "grad_norm": 0.08419459918821942,
      "learning_rate": 3.580834991328206e-06,
      "loss": 0.4231,
      "num_tokens": 1463209714.0,
      "step": 1544
    },
    {
      "epoch": 3.0235063663075414,
      "grad_norm": 0.08149417035919176,
      "learning_rate": 3.5747578633698394e-06,
      "loss": 0.4307,
      "num_tokens": 1464181564.0,
      "step": 1545
    },
    {
      "epoch": 3.0254652301665033,
      "grad_norm": 0.0883918014012131,
      "learning_rate": 3.568683026203889e-06,
      "loss": 0.4338,
      "num_tokens": 1465174778.0,
      "step": 1546
    },
    {
      "epoch": 3.027424094025465,
      "grad_norm": 0.08517759734458666,
      "learning_rate": 3.5626104895944437e-06,
      "loss": 0.4053,
      "num_tokens": 1466129922.0,
      "step": 1547
    },
    {
      "epoch": 3.029382957884427,
      "grad_norm": 0.08446013585267594,
      "learning_rate": 3.5565402633018963e-06,
      "loss": 0.4182,
      "num_tokens": 1467089461.0,
      "step": 1548
    },
    {
      "epoch": 3.0313418217433887,
      "grad_norm": 0.08533340445757492,
      "learning_rate": 3.550472357082922e-06,
      "loss": 0.4122,
      "num_tokens": 1468046930.0,
      "step": 1549
    },
    {
      "epoch": 3.0333006856023506,
      "grad_norm": 0.09021042288751242,
      "learning_rate": 3.5444067806904703e-06,
      "loss": 0.4213,
      "num_tokens": 1468982281.0,
      "step": 1550
    },
    {
      "epoch": 3.0352595494613124,
      "grad_norm": 0.08618935150857403,
      "learning_rate": 3.538343543873748e-06,
      "loss": 0.4125,
      "num_tokens": 1469934810.0,
      "step": 1551
    },
    {
      "epoch": 3.037218413320274,
      "grad_norm": 0.0822883606080526,
      "learning_rate": 3.532282656378194e-06,
      "loss": 0.4208,
      "num_tokens": 1470853320.0,
      "step": 1552
    },
    {
      "epoch": 3.039177277179236,
      "grad_norm": 0.08475966762698743,
      "learning_rate": 3.526224127945479e-06,
      "loss": 0.3981,
      "num_tokens": 1471815045.0,
      "step": 1553
    },
    {
      "epoch": 3.041136141038198,
      "grad_norm": 0.0833098974921207,
      "learning_rate": 3.5201679683134793e-06,
      "loss": 0.4108,
      "num_tokens": 1472765711.0,
      "step": 1554
    },
    {
      "epoch": 3.0430950048971597,
      "grad_norm": 0.08204783615357081,
      "learning_rate": 3.5141141872162613e-06,
      "loss": 0.4309,
      "num_tokens": 1473756856.0,
      "step": 1555
    },
    {
      "epoch": 3.0450538687561215,
      "grad_norm": 0.08622732916217471,
      "learning_rate": 3.5080627943840717e-06,
      "loss": 0.4187,
      "num_tokens": 1474700106.0,
      "step": 1556
    },
    {
      "epoch": 3.0470127326150833,
      "grad_norm": 0.08823584351791991,
      "learning_rate": 3.5020137995433177e-06,
      "loss": 0.4175,
      "num_tokens": 1475676036.0,
      "step": 1557
    },
    {
      "epoch": 3.048971596474045,
      "grad_norm": 0.08673325416372975,
      "learning_rate": 3.49596721241655e-06,
      "loss": 0.423,
      "num_tokens": 1476615144.0,
      "step": 1558
    },
    {
      "epoch": 3.050930460333007,
      "grad_norm": 0.08522920122710141,
      "learning_rate": 3.489923042722453e-06,
      "loss": 0.4171,
      "num_tokens": 1477560825.0,
      "step": 1559
    },
    {
      "epoch": 3.052889324191969,
      "grad_norm": 0.08287045771927945,
      "learning_rate": 3.483881300175823e-06,
      "loss": 0.4105,
      "num_tokens": 1478493290.0,
      "step": 1560
    },
    {
      "epoch": 3.0548481880509306,
      "grad_norm": 0.08657532938138,
      "learning_rate": 3.4778419944875556e-06,
      "loss": 0.4117,
      "num_tokens": 1479436488.0,
      "step": 1561
    },
    {
      "epoch": 3.0568070519098924,
      "grad_norm": 0.08885440023003051,
      "learning_rate": 3.4718051353646304e-06,
      "loss": 0.4056,
      "num_tokens": 1480380043.0,
      "step": 1562
    },
    {
      "epoch": 3.0587659157688543,
      "grad_norm": 0.08556954522511925,
      "learning_rate": 3.465770732510095e-06,
      "loss": 0.4083,
      "num_tokens": 1481352426.0,
      "step": 1563
    },
    {
      "epoch": 3.060724779627816,
      "grad_norm": 0.0820891592346816,
      "learning_rate": 3.459738795623046e-06,
      "loss": 0.4106,
      "num_tokens": 1482296566.0,
      "step": 1564
    },
    {
      "epoch": 3.0626836434867775,
      "grad_norm": 0.08517211749613793,
      "learning_rate": 3.4537093343986207e-06,
      "loss": 0.4252,
      "num_tokens": 1483246603.0,
      "step": 1565
    },
    {
      "epoch": 3.0646425073457393,
      "grad_norm": 0.09074356856063315,
      "learning_rate": 3.4476823585279745e-06,
      "loss": 0.4251,
      "num_tokens": 1484209872.0,
      "step": 1566
    },
    {
      "epoch": 3.066601371204701,
      "grad_norm": 0.08704176577376241,
      "learning_rate": 3.4416578776982677e-06,
      "loss": 0.4231,
      "num_tokens": 1485204372.0,
      "step": 1567
    },
    {
      "epoch": 3.068560235063663,
      "grad_norm": 0.08499616024048211,
      "learning_rate": 3.4356359015926553e-06,
      "loss": 0.4249,
      "num_tokens": 1486151469.0,
      "step": 1568
    },
    {
      "epoch": 3.0705190989226248,
      "grad_norm": 0.08380466790436383,
      "learning_rate": 3.4296164398902576e-06,
      "loss": 0.4138,
      "num_tokens": 1487089176.0,
      "step": 1569
    },
    {
      "epoch": 3.0724779627815866,
      "grad_norm": 0.0866388761711634,
      "learning_rate": 3.42359950226616e-06,
      "loss": 0.4193,
      "num_tokens": 1488013711.0,
      "step": 1570
    },
    {
      "epoch": 3.0744368266405484,
      "grad_norm": 0.09210749037221054,
      "learning_rate": 3.4175850983913915e-06,
      "loss": 0.4262,
      "num_tokens": 1488958800.0,
      "step": 1571
    },
    {
      "epoch": 3.0763956904995102,
      "grad_norm": 0.09040379822716957,
      "learning_rate": 3.4115732379329038e-06,
      "loss": 0.4061,
      "num_tokens": 1489904078.0,
      "step": 1572
    },
    {
      "epoch": 3.078354554358472,
      "grad_norm": 0.08889792152673466,
      "learning_rate": 3.4055639305535647e-06,
      "loss": 0.417,
      "num_tokens": 1490873655.0,
      "step": 1573
    },
    {
      "epoch": 3.080313418217434,
      "grad_norm": 0.08498848149221166,
      "learning_rate": 3.399557185912138e-06,
      "loss": 0.4236,
      "num_tokens": 1491813490.0,
      "step": 1574
    },
    {
      "epoch": 3.0822722820763957,
      "grad_norm": 0.08783321336207509,
      "learning_rate": 3.393553013663265e-06,
      "loss": 0.4316,
      "num_tokens": 1492776079.0,
      "step": 1575
    },
    {
      "epoch": 3.0842311459353575,
      "grad_norm": 0.08725093162154747,
      "learning_rate": 3.387551423457456e-06,
      "loss": 0.433,
      "num_tokens": 1493699310.0,
      "step": 1576
    },
    {
      "epoch": 3.0861900097943193,
      "grad_norm": 0.0867956181889245,
      "learning_rate": 3.3815524249410716e-06,
      "loss": 0.4038,
      "num_tokens": 1494647913.0,
      "step": 1577
    },
    {
      "epoch": 3.088148873653281,
      "grad_norm": 0.08521664492400116,
      "learning_rate": 3.3755560277563028e-06,
      "loss": 0.4229,
      "num_tokens": 1495616948.0,
      "step": 1578
    },
    {
      "epoch": 3.090107737512243,
      "grad_norm": 0.0848634149347137,
      "learning_rate": 3.3695622415411626e-06,
      "loss": 0.4096,
      "num_tokens": 1496583967.0,
      "step": 1579
    },
    {
      "epoch": 3.092066601371205,
      "grad_norm": 0.08183021108472953,
      "learning_rate": 3.3635710759294684e-06,
      "loss": 0.4115,
      "num_tokens": 1497544251.0,
      "step": 1580
    },
    {
      "epoch": 3.0940254652301666,
      "grad_norm": 0.08651099819700418,
      "learning_rate": 3.3575825405508213e-06,
      "loss": 0.4123,
      "num_tokens": 1498477038.0,
      "step": 1581
    },
    {
      "epoch": 3.0959843290891285,
      "grad_norm": 0.08605916461647833,
      "learning_rate": 3.351596645030597e-06,
      "loss": 0.4135,
      "num_tokens": 1499429051.0,
      "step": 1582
    },
    {
      "epoch": 3.0979431929480903,
      "grad_norm": 0.08691192888813512,
      "learning_rate": 3.345613398989932e-06,
      "loss": 0.419,
      "num_tokens": 1500372240.0,
      "step": 1583
    },
    {
      "epoch": 3.099902056807052,
      "grad_norm": 0.0863879801250081,
      "learning_rate": 3.3396328120456968e-06,
      "loss": 0.4202,
      "num_tokens": 1501297993.0,
      "step": 1584
    },
    {
      "epoch": 3.101860920666014,
      "grad_norm": 0.08680344056610552,
      "learning_rate": 3.333654893810493e-06,
      "loss": 0.4152,
      "num_tokens": 1502237843.0,
      "step": 1585
    },
    {
      "epoch": 3.1038197845249753,
      "grad_norm": 0.09001976183098781,
      "learning_rate": 3.3276796538926335e-06,
      "loss": 0.4108,
      "num_tokens": 1503139935.0,
      "step": 1586
    },
    {
      "epoch": 3.105778648383937,
      "grad_norm": 0.0892083435236935,
      "learning_rate": 3.321707101896122e-06,
      "loss": 0.431,
      "num_tokens": 1504067449.0,
      "step": 1587
    },
    {
      "epoch": 3.107737512242899,
      "grad_norm": 0.08401671867988442,
      "learning_rate": 3.3157372474206463e-06,
      "loss": 0.4116,
      "num_tokens": 1505019240.0,
      "step": 1588
    },
    {
      "epoch": 3.109696376101861,
      "grad_norm": 0.08402281265217668,
      "learning_rate": 3.3097701000615543e-06,
      "loss": 0.396,
      "num_tokens": 1505980621.0,
      "step": 1589
    },
    {
      "epoch": 3.1116552399608226,
      "grad_norm": 0.08693116821436273,
      "learning_rate": 3.3038056694098485e-06,
      "loss": 0.4052,
      "num_tokens": 1506935827.0,
      "step": 1590
    },
    {
      "epoch": 3.1136141038197844,
      "grad_norm": 0.08520733780574463,
      "learning_rate": 3.2978439650521583e-06,
      "loss": 0.4238,
      "num_tokens": 1507907052.0,
      "step": 1591
    },
    {
      "epoch": 3.1155729676787463,
      "grad_norm": 0.08299873349532465,
      "learning_rate": 3.2918849965707346e-06,
      "loss": 0.4194,
      "num_tokens": 1508855217.0,
      "step": 1592
    },
    {
      "epoch": 3.117531831537708,
      "grad_norm": 0.08560173771938759,
      "learning_rate": 3.285928773543432e-06,
      "loss": 0.4118,
      "num_tokens": 1509765480.0,
      "step": 1593
    },
    {
      "epoch": 3.11949069539667,
      "grad_norm": 0.08741343363260318,
      "learning_rate": 3.279975305543691e-06,
      "loss": 0.4253,
      "num_tokens": 1510726718.0,
      "step": 1594
    },
    {
      "epoch": 3.1214495592556317,
      "grad_norm": 0.08673928707032301,
      "learning_rate": 3.274024602140521e-06,
      "loss": 0.403,
      "num_tokens": 1511680781.0,
      "step": 1595
    },
    {
      "epoch": 3.1234084231145935,
      "grad_norm": 0.08440194225050086,
      "learning_rate": 3.268076672898492e-06,
      "loss": 0.417,
      "num_tokens": 1512595531.0,
      "step": 1596
    },
    {
      "epoch": 3.1253672869735554,
      "grad_norm": 0.08530355178102447,
      "learning_rate": 3.262131527377715e-06,
      "loss": 0.414,
      "num_tokens": 1513537053.0,
      "step": 1597
    },
    {
      "epoch": 3.127326150832517,
      "grad_norm": 0.08542535879459515,
      "learning_rate": 3.2561891751338232e-06,
      "loss": 0.412,
      "num_tokens": 1514439201.0,
      "step": 1598
    },
    {
      "epoch": 3.129285014691479,
      "grad_norm": 0.08603236893444292,
      "learning_rate": 3.250249625717964e-06,
      "loss": 0.4107,
      "num_tokens": 1515384871.0,
      "step": 1599
    },
    {
      "epoch": 3.131243878550441,
      "grad_norm": 0.08403329310434217,
      "learning_rate": 3.2443128886767782e-06,
      "loss": 0.3974,
      "num_tokens": 1516352640.0,
      "step": 1600
    },
    {
      "epoch": 3.1332027424094027,
      "grad_norm": 0.08207199919198398,
      "learning_rate": 3.238378973552385e-06,
      "loss": 0.4253,
      "num_tokens": 1517325804.0,
      "step": 1601
    },
    {
      "epoch": 3.1351616062683645,
      "grad_norm": 0.0855836165023854,
      "learning_rate": 3.232447889882371e-06,
      "loss": 0.4163,
      "num_tokens": 1518280093.0,
      "step": 1602
    },
    {
      "epoch": 3.1371204701273263,
      "grad_norm": 0.08669293489065007,
      "learning_rate": 3.2265196471997682e-06,
      "loss": 0.4396,
      "num_tokens": 1519242000.0,
      "step": 1603
    },
    {
      "epoch": 3.139079333986288,
      "grad_norm": 0.08885090340405069,
      "learning_rate": 3.220594255033046e-06,
      "loss": 0.421,
      "num_tokens": 1520192835.0,
      "step": 1604
    },
    {
      "epoch": 3.14103819784525,
      "grad_norm": 0.08467497009528231,
      "learning_rate": 3.2146717229060885e-06,
      "loss": 0.4158,
      "num_tokens": 1521134188.0,
      "step": 1605
    },
    {
      "epoch": 3.142997061704212,
      "grad_norm": 0.0853642353840204,
      "learning_rate": 3.208752060338186e-06,
      "loss": 0.4115,
      "num_tokens": 1522125329.0,
      "step": 1606
    },
    {
      "epoch": 3.144955925563173,
      "grad_norm": 0.08515957490504761,
      "learning_rate": 3.202835276844015e-06,
      "loss": 0.4197,
      "num_tokens": 1523073841.0,
      "step": 1607
    },
    {
      "epoch": 3.146914789422135,
      "grad_norm": 0.08709305426130894,
      "learning_rate": 3.196921381933624e-06,
      "loss": 0.4091,
      "num_tokens": 1524011872.0,
      "step": 1608
    },
    {
      "epoch": 3.148873653281097,
      "grad_norm": 0.08384257861940272,
      "learning_rate": 3.1910103851124185e-06,
      "loss": 0.4121,
      "num_tokens": 1524958136.0,
      "step": 1609
    },
    {
      "epoch": 3.1508325171400586,
      "grad_norm": 0.09376876734480932,
      "learning_rate": 3.1851022958811495e-06,
      "loss": 0.416,
      "num_tokens": 1525942578.0,
      "step": 1610
    },
    {
      "epoch": 3.1527913809990205,
      "grad_norm": 0.08724936921017105,
      "learning_rate": 3.1791971237358893e-06,
      "loss": 0.4165,
      "num_tokens": 1526879385.0,
      "step": 1611
    },
    {
      "epoch": 3.1547502448579823,
      "grad_norm": 0.08507190978735882,
      "learning_rate": 3.173294878168025e-06,
      "loss": 0.4183,
      "num_tokens": 1527823497.0,
      "step": 1612
    },
    {
      "epoch": 3.156709108716944,
      "grad_norm": 0.08579655874793927,
      "learning_rate": 3.16739556866424e-06,
      "loss": 0.404,
      "num_tokens": 1528777994.0,
      "step": 1613
    },
    {
      "epoch": 3.158667972575906,
      "grad_norm": 0.08500964605985802,
      "learning_rate": 3.1614992047064947e-06,
      "loss": 0.413,
      "num_tokens": 1529718455.0,
      "step": 1614
    },
    {
      "epoch": 3.1606268364348677,
      "grad_norm": 0.0866587425423897,
      "learning_rate": 3.15560579577202e-06,
      "loss": 0.4246,
      "num_tokens": 1530618167.0,
      "step": 1615
    },
    {
      "epoch": 3.1625857002938296,
      "grad_norm": 0.0861475105612104,
      "learning_rate": 3.1497153513332956e-06,
      "loss": 0.4005,
      "num_tokens": 1531582422.0,
      "step": 1616
    },
    {
      "epoch": 3.1645445641527914,
      "grad_norm": 0.08225270518432634,
      "learning_rate": 3.1438278808580336e-06,
      "loss": 0.416,
      "num_tokens": 1532529952.0,
      "step": 1617
    },
    {
      "epoch": 3.166503428011753,
      "grad_norm": 0.09168438965808968,
      "learning_rate": 3.1379433938091695e-06,
      "loss": 0.428,
      "num_tokens": 1533478518.0,
      "step": 1618
    },
    {
      "epoch": 3.168462291870715,
      "grad_norm": 0.08600848156463216,
      "learning_rate": 3.132061899644845e-06,
      "loss": 0.4062,
      "num_tokens": 1534431461.0,
      "step": 1619
    },
    {
      "epoch": 3.170421155729677,
      "grad_norm": 0.0843262406890434,
      "learning_rate": 3.126183407818384e-06,
      "loss": 0.417,
      "num_tokens": 1535382785.0,
      "step": 1620
    },
    {
      "epoch": 3.1723800195886387,
      "grad_norm": 0.08220724066091184,
      "learning_rate": 3.1203079277782933e-06,
      "loss": 0.4228,
      "num_tokens": 1536337194.0,
      "step": 1621
    },
    {
      "epoch": 3.1743388834476005,
      "grad_norm": 0.08780076873904627,
      "learning_rate": 3.114435468968232e-06,
      "loss": 0.4177,
      "num_tokens": 1537278965.0,
      "step": 1622
    },
    {
      "epoch": 3.1762977473065623,
      "grad_norm": 0.09380512161834306,
      "learning_rate": 3.1085660408270072e-06,
      "loss": 0.423,
      "num_tokens": 1538242134.0,
      "step": 1623
    },
    {
      "epoch": 3.178256611165524,
      "grad_norm": 0.08714251180503718,
      "learning_rate": 3.1026996527885555e-06,
      "loss": 0.4224,
      "num_tokens": 1539190656.0,
      "step": 1624
    },
    {
      "epoch": 3.180215475024486,
      "grad_norm": 0.09076483906215926,
      "learning_rate": 3.0968363142819226e-06,
      "loss": 0.4038,
      "num_tokens": 1540168234.0,
      "step": 1625
    },
    {
      "epoch": 3.182174338883448,
      "grad_norm": 0.0853473473221615,
      "learning_rate": 3.090976034731257e-06,
      "loss": 0.415,
      "num_tokens": 1541115357.0,
      "step": 1626
    },
    {
      "epoch": 3.1841332027424096,
      "grad_norm": 0.08322543501652875,
      "learning_rate": 3.0851188235557892e-06,
      "loss": 0.4213,
      "num_tokens": 1542091792.0,
      "step": 1627
    },
    {
      "epoch": 3.186092066601371,
      "grad_norm": 0.08464197460586767,
      "learning_rate": 3.0792646901698165e-06,
      "loss": 0.4249,
      "num_tokens": 1543035468.0,
      "step": 1628
    },
    {
      "epoch": 3.188050930460333,
      "grad_norm": 0.08382929766448245,
      "learning_rate": 3.0734136439826913e-06,
      "loss": 0.3976,
      "num_tokens": 1543979369.0,
      "step": 1629
    },
    {
      "epoch": 3.1900097943192947,
      "grad_norm": 0.08628998152935578,
      "learning_rate": 3.0675656943988046e-06,
      "loss": 0.4146,
      "num_tokens": 1544914493.0,
      "step": 1630
    },
    {
      "epoch": 3.1919686581782565,
      "grad_norm": 0.0875463347687337,
      "learning_rate": 3.0617208508175662e-06,
      "loss": 0.4117,
      "num_tokens": 1545843104.0,
      "step": 1631
    },
    {
      "epoch": 3.1939275220372183,
      "grad_norm": 0.08789314380866718,
      "learning_rate": 3.0558791226333974e-06,
      "loss": 0.4157,
      "num_tokens": 1546744401.0,
      "step": 1632
    },
    {
      "epoch": 3.19588638589618,
      "grad_norm": 0.08655557771658778,
      "learning_rate": 3.050040519235711e-06,
      "loss": 0.4084,
      "num_tokens": 1547674396.0,
      "step": 1633
    },
    {
      "epoch": 3.197845249755142,
      "grad_norm": 0.08735170082717181,
      "learning_rate": 3.044205050008897e-06,
      "loss": 0.4233,
      "num_tokens": 1548641233.0,
      "step": 1634
    },
    {
      "epoch": 3.1998041136141038,
      "grad_norm": 0.08637476375081056,
      "learning_rate": 3.038372724332308e-06,
      "loss": 0.4184,
      "num_tokens": 1549614441.0,
      "step": 1635
    },
    {
      "epoch": 3.2017629774730656,
      "grad_norm": 0.0810595463890678,
      "learning_rate": 3.032543551580245e-06,
      "loss": 0.4113,
      "num_tokens": 1550594878.0,
      "step": 1636
    },
    {
      "epoch": 3.2037218413320274,
      "grad_norm": 0.08498515772322175,
      "learning_rate": 3.0267175411219386e-06,
      "loss": 0.4228,
      "num_tokens": 1551538367.0,
      "step": 1637
    },
    {
      "epoch": 3.2056807051909892,
      "grad_norm": 0.08391917564635133,
      "learning_rate": 3.020894702321539e-06,
      "loss": 0.4148,
      "num_tokens": 1552526924.0,
      "step": 1638
    },
    {
      "epoch": 3.207639569049951,
      "grad_norm": 0.08624378179678088,
      "learning_rate": 3.0150750445380995e-06,
      "loss": 0.419,
      "num_tokens": 1553490942.0,
      "step": 1639
    },
    {
      "epoch": 3.209598432908913,
      "grad_norm": 0.08507409916984543,
      "learning_rate": 3.0092585771255567e-06,
      "loss": 0.411,
      "num_tokens": 1554417219.0,
      "step": 1640
    },
    {
      "epoch": 3.2115572967678747,
      "grad_norm": 0.0921606250869294,
      "learning_rate": 3.003445309432723e-06,
      "loss": 0.4153,
      "num_tokens": 1555337275.0,
      "step": 1641
    },
    {
      "epoch": 3.2135161606268365,
      "grad_norm": 0.08743945172958979,
      "learning_rate": 2.997635250803268e-06,
      "loss": 0.4189,
      "num_tokens": 1556308265.0,
      "step": 1642
    },
    {
      "epoch": 3.2154750244857984,
      "grad_norm": 0.08537878053467987,
      "learning_rate": 2.991828410575698e-06,
      "loss": 0.4171,
      "num_tokens": 1557238231.0,
      "step": 1643
    },
    {
      "epoch": 3.21743388834476,
      "grad_norm": 0.08723290879714155,
      "learning_rate": 2.9860247980833534e-06,
      "loss": 0.4122,
      "num_tokens": 1558165499.0,
      "step": 1644
    },
    {
      "epoch": 3.219392752203722,
      "grad_norm": 0.08372142734691139,
      "learning_rate": 2.980224422654382e-06,
      "loss": 0.402,
      "num_tokens": 1559110342.0,
      "step": 1645
    },
    {
      "epoch": 3.221351616062684,
      "grad_norm": 0.08551788909129061,
      "learning_rate": 2.9744272936117323e-06,
      "loss": 0.413,
      "num_tokens": 1560057661.0,
      "step": 1646
    },
    {
      "epoch": 3.223310479921645,
      "grad_norm": 0.08613788402152842,
      "learning_rate": 2.9686334202731325e-06,
      "loss": 0.4307,
      "num_tokens": 1560992014.0,
      "step": 1647
    },
    {
      "epoch": 3.2252693437806075,
      "grad_norm": 0.08584891156901349,
      "learning_rate": 2.9628428119510742e-06,
      "loss": 0.4006,
      "num_tokens": 1561930651.0,
      "step": 1648
    },
    {
      "epoch": 3.227228207639569,
      "grad_norm": 0.084163946886269,
      "learning_rate": 2.9570554779528094e-06,
      "loss": 0.4209,
      "num_tokens": 1562892101.0,
      "step": 1649
    },
    {
      "epoch": 3.2291870714985307,
      "grad_norm": 0.08666807475053581,
      "learning_rate": 2.951271427580321e-06,
      "loss": 0.4087,
      "num_tokens": 1563818257.0,
      "step": 1650
    },
    {
      "epoch": 3.2311459353574925,
      "grad_norm": 0.08238976907237429,
      "learning_rate": 2.9454906701303167e-06,
      "loss": 0.4058,
      "num_tokens": 1564769030.0,
      "step": 1651
    },
    {
      "epoch": 3.2331047992164543,
      "grad_norm": 0.08506864633911439,
      "learning_rate": 2.9397132148942104e-06,
      "loss": 0.4039,
      "num_tokens": 1565715324.0,
      "step": 1652
    },
    {
      "epoch": 3.235063663075416,
      "grad_norm": 0.08222389882512811,
      "learning_rate": 2.9339390711581105e-06,
      "loss": 0.4202,
      "num_tokens": 1566677591.0,
      "step": 1653
    },
    {
      "epoch": 3.237022526934378,
      "grad_norm": 0.08610362353178959,
      "learning_rate": 2.928168248202799e-06,
      "loss": 0.4081,
      "num_tokens": 1567616036.0,
      "step": 1654
    },
    {
      "epoch": 3.23898139079334,
      "grad_norm": 0.0882402206039739,
      "learning_rate": 2.9224007553037236e-06,
      "loss": 0.3962,
      "num_tokens": 1568558614.0,
      "step": 1655
    },
    {
      "epoch": 3.2409402546523016,
      "grad_norm": 0.08387583936887746,
      "learning_rate": 2.91663660173098e-06,
      "loss": 0.4095,
      "num_tokens": 1569507348.0,
      "step": 1656
    },
    {
      "epoch": 3.2428991185112634,
      "grad_norm": 0.08395318958639249,
      "learning_rate": 2.9108757967492907e-06,
      "loss": 0.4125,
      "num_tokens": 1570467818.0,
      "step": 1657
    },
    {
      "epoch": 3.2448579823702253,
      "grad_norm": 0.08448289400753382,
      "learning_rate": 2.905118349618007e-06,
      "loss": 0.4128,
      "num_tokens": 1571390817.0,
      "step": 1658
    },
    {
      "epoch": 3.246816846229187,
      "grad_norm": 0.08247992022481007,
      "learning_rate": 2.899364269591072e-06,
      "loss": 0.4162,
      "num_tokens": 1572356564.0,
      "step": 1659
    },
    {
      "epoch": 3.248775710088149,
      "grad_norm": 0.09887974786405178,
      "learning_rate": 2.8936135659170217e-06,
      "loss": 0.4206,
      "num_tokens": 1573328765.0,
      "step": 1660
    },
    {
      "epoch": 3.2507345739471107,
      "grad_norm": 0.08430893870506403,
      "learning_rate": 2.8878662478389675e-06,
      "loss": 0.4043,
      "num_tokens": 1574270584.0,
      "step": 1661
    },
    {
      "epoch": 3.2526934378060726,
      "grad_norm": 0.08518091959222462,
      "learning_rate": 2.882122324594575e-06,
      "loss": 0.4209,
      "num_tokens": 1575234867.0,
      "step": 1662
    },
    {
      "epoch": 3.2546523016650344,
      "grad_norm": 0.08476071734482449,
      "learning_rate": 2.8763818054160538e-06,
      "loss": 0.4241,
      "num_tokens": 1576187296.0,
      "step": 1663
    },
    {
      "epoch": 3.256611165523996,
      "grad_norm": 0.08657136311691385,
      "learning_rate": 2.870644699530146e-06,
      "loss": 0.4269,
      "num_tokens": 1577180261.0,
      "step": 1664
    },
    {
      "epoch": 3.258570029382958,
      "grad_norm": 0.09207416692508284,
      "learning_rate": 2.8649110161581047e-06,
      "loss": 0.4084,
      "num_tokens": 1578094376.0,
      "step": 1665
    },
    {
      "epoch": 3.26052889324192,
      "grad_norm": 0.08785370437827834,
      "learning_rate": 2.859180764515678e-06,
      "loss": 0.4239,
      "num_tokens": 1579024168.0,
      "step": 1666
    },
    {
      "epoch": 3.2624877571008817,
      "grad_norm": 0.0852265339374843,
      "learning_rate": 2.853453953813108e-06,
      "loss": 0.4171,
      "num_tokens": 1579965477.0,
      "step": 1667
    },
    {
      "epoch": 3.264446620959843,
      "grad_norm": 0.0871308803527743,
      "learning_rate": 2.847730593255097e-06,
      "loss": 0.4339,
      "num_tokens": 1580915159.0,
      "step": 1668
    },
    {
      "epoch": 3.2664054848188053,
      "grad_norm": 0.08373682216483416,
      "learning_rate": 2.842010692040805e-06,
      "loss": 0.4171,
      "num_tokens": 1581839703.0,
      "step": 1669
    },
    {
      "epoch": 3.2683643486777667,
      "grad_norm": 0.08507930380153848,
      "learning_rate": 2.8362942593638344e-06,
      "loss": 0.4177,
      "num_tokens": 1582754801.0,
      "step": 1670
    },
    {
      "epoch": 3.2703232125367285,
      "grad_norm": 0.08766349076289924,
      "learning_rate": 2.83058130441221e-06,
      "loss": 0.4196,
      "num_tokens": 1583723619.0,
      "step": 1671
    },
    {
      "epoch": 3.2722820763956904,
      "grad_norm": 0.08387823426246532,
      "learning_rate": 2.824871836368364e-06,
      "loss": 0.4225,
      "num_tokens": 1584702117.0,
      "step": 1672
    },
    {
      "epoch": 3.274240940254652,
      "grad_norm": 0.08405211840878263,
      "learning_rate": 2.819165864409134e-06,
      "loss": 0.4155,
      "num_tokens": 1585632848.0,
      "step": 1673
    },
    {
      "epoch": 3.276199804113614,
      "grad_norm": 0.08611604714213371,
      "learning_rate": 2.8134633977057236e-06,
      "loss": 0.4352,
      "num_tokens": 1586592050.0,
      "step": 1674
    },
    {
      "epoch": 3.278158667972576,
      "grad_norm": 0.08485664401177756,
      "learning_rate": 2.8077644454237165e-06,
      "loss": 0.4179,
      "num_tokens": 1587529986.0,
      "step": 1675
    },
    {
      "epoch": 3.2801175318315376,
      "grad_norm": 0.08895927311901823,
      "learning_rate": 2.8020690167230384e-06,
      "loss": 0.4235,
      "num_tokens": 1588475204.0,
      "step": 1676
    },
    {
      "epoch": 3.2820763956904995,
      "grad_norm": 0.08252937131680212,
      "learning_rate": 2.7963771207579543e-06,
      "loss": 0.4155,
      "num_tokens": 1589437322.0,
      "step": 1677
    },
    {
      "epoch": 3.2840352595494613,
      "grad_norm": 0.08561585560174582,
      "learning_rate": 2.7906887666770554e-06,
      "loss": 0.4209,
      "num_tokens": 1590379994.0,
      "step": 1678
    },
    {
      "epoch": 3.285994123408423,
      "grad_norm": 0.08471052529676973,
      "learning_rate": 2.785003963623233e-06,
      "loss": 0.4077,
      "num_tokens": 1591316557.0,
      "step": 1679
    },
    {
      "epoch": 3.287952987267385,
      "grad_norm": 0.0837155669110085,
      "learning_rate": 2.7793227207336733e-06,
      "loss": 0.4112,
      "num_tokens": 1592285008.0,
      "step": 1680
    },
    {
      "epoch": 3.2899118511263468,
      "grad_norm": 0.08396722946953317,
      "learning_rate": 2.7736450471398435e-06,
      "loss": 0.4083,
      "num_tokens": 1593231866.0,
      "step": 1681
    },
    {
      "epoch": 3.2918707149853086,
      "grad_norm": 0.08267500220429273,
      "learning_rate": 2.7679709519674715e-06,
      "loss": 0.4092,
      "num_tokens": 1594187614.0,
      "step": 1682
    },
    {
      "epoch": 3.2938295788442704,
      "grad_norm": 0.08835444863536938,
      "learning_rate": 2.762300444336529e-06,
      "loss": 0.4347,
      "num_tokens": 1595121078.0,
      "step": 1683
    },
    {
      "epoch": 3.2957884427032322,
      "grad_norm": 0.08270484625014247,
      "learning_rate": 2.7566335333612303e-06,
      "loss": 0.4093,
      "num_tokens": 1596066499.0,
      "step": 1684
    },
    {
      "epoch": 3.297747306562194,
      "grad_norm": 0.08730471579599326,
      "learning_rate": 2.7509702281500028e-06,
      "loss": 0.4409,
      "num_tokens": 1597027924.0,
      "step": 1685
    },
    {
      "epoch": 3.299706170421156,
      "grad_norm": 0.08451779069624428,
      "learning_rate": 2.745310537805479e-06,
      "loss": 0.4092,
      "num_tokens": 1597997298.0,
      "step": 1686
    },
    {
      "epoch": 3.3016650342801177,
      "grad_norm": 0.08395958861025746,
      "learning_rate": 2.7396544714244837e-06,
      "loss": 0.4207,
      "num_tokens": 1598925723.0,
      "step": 1687
    },
    {
      "epoch": 3.3036238981390795,
      "grad_norm": 0.093120802790401,
      "learning_rate": 2.734002038098015e-06,
      "loss": 0.4162,
      "num_tokens": 1599886896.0,
      "step": 1688
    },
    {
      "epoch": 3.305582761998041,
      "grad_norm": 0.08465686828727743,
      "learning_rate": 2.7283532469112296e-06,
      "loss": 0.4085,
      "num_tokens": 1600897103.0,
      "step": 1689
    },
    {
      "epoch": 3.307541625857003,
      "grad_norm": 0.08666605127840357,
      "learning_rate": 2.7227081069434356e-06,
      "loss": 0.4164,
      "num_tokens": 1601870206.0,
      "step": 1690
    },
    {
      "epoch": 3.3095004897159646,
      "grad_norm": 0.08503362608005334,
      "learning_rate": 2.717066627268069e-06,
      "loss": 0.4161,
      "num_tokens": 1602797175.0,
      "step": 1691
    },
    {
      "epoch": 3.3114593535749264,
      "grad_norm": 0.08925765101820528,
      "learning_rate": 2.711428816952679e-06,
      "loss": 0.4113,
      "num_tokens": 1603727650.0,
      "step": 1692
    },
    {
      "epoch": 3.313418217433888,
      "grad_norm": 0.08134885457766461,
      "learning_rate": 2.705794685058927e-06,
      "loss": 0.4029,
      "num_tokens": 1604692227.0,
      "step": 1693
    },
    {
      "epoch": 3.31537708129285,
      "grad_norm": 0.08399004334986025,
      "learning_rate": 2.7001642406425523e-06,
      "loss": 0.4128,
      "num_tokens": 1605650385.0,
      "step": 1694
    },
    {
      "epoch": 3.317335945151812,
      "grad_norm": 0.08948639907925239,
      "learning_rate": 2.69453749275337e-06,
      "loss": 0.4103,
      "num_tokens": 1606585046.0,
      "step": 1695
    },
    {
      "epoch": 3.3192948090107737,
      "grad_norm": 0.08556440329567387,
      "learning_rate": 2.6889144504352592e-06,
      "loss": 0.4225,
      "num_tokens": 1607526201.0,
      "step": 1696
    },
    {
      "epoch": 3.3212536728697355,
      "grad_norm": 0.08999851006315714,
      "learning_rate": 2.683295122726136e-06,
      "loss": 0.4155,
      "num_tokens": 1608475502.0,
      "step": 1697
    },
    {
      "epoch": 3.3232125367286973,
      "grad_norm": 0.09137548845171062,
      "learning_rate": 2.6776795186579466e-06,
      "loss": 0.4229,
      "num_tokens": 1609438064.0,
      "step": 1698
    },
    {
      "epoch": 3.325171400587659,
      "grad_norm": 0.08318486747108624,
      "learning_rate": 2.6720676472566593e-06,
      "loss": 0.4069,
      "num_tokens": 1610394954.0,
      "step": 1699
    },
    {
      "epoch": 3.327130264446621,
      "grad_norm": 0.08257993944770921,
      "learning_rate": 2.6664595175422347e-06,
      "loss": 0.4225,
      "num_tokens": 1611363403.0,
      "step": 1700
    },
    {
      "epoch": 3.329089128305583,
      "grad_norm": 0.08548244142024326,
      "learning_rate": 2.6608551385286234e-06,
      "loss": 0.4245,
      "num_tokens": 1612312837.0,
      "step": 1701
    },
    {
      "epoch": 3.3310479921645446,
      "grad_norm": 0.08400158509048183,
      "learning_rate": 2.655254519223746e-06,
      "loss": 0.415,
      "num_tokens": 1613268041.0,
      "step": 1702
    },
    {
      "epoch": 3.3330068560235064,
      "grad_norm": 0.0869609313137376,
      "learning_rate": 2.6496576686294793e-06,
      "loss": 0.4091,
      "num_tokens": 1614232079.0,
      "step": 1703
    },
    {
      "epoch": 3.3349657198824683,
      "grad_norm": 0.08581379027992089,
      "learning_rate": 2.6440645957416483e-06,
      "loss": 0.4181,
      "num_tokens": 1615131466.0,
      "step": 1704
    },
    {
      "epoch": 3.33692458374143,
      "grad_norm": 0.08580102241383081,
      "learning_rate": 2.638475309549999e-06,
      "loss": 0.4153,
      "num_tokens": 1616107363.0,
      "step": 1705
    },
    {
      "epoch": 3.338883447600392,
      "grad_norm": 0.08371222806481414,
      "learning_rate": 2.6328898190381934e-06,
      "loss": 0.4162,
      "num_tokens": 1617034866.0,
      "step": 1706
    },
    {
      "epoch": 3.3408423114593537,
      "grad_norm": 0.08969662627288895,
      "learning_rate": 2.6273081331837956e-06,
      "loss": 0.4084,
      "num_tokens": 1617983797.0,
      "step": 1707
    },
    {
      "epoch": 3.3428011753183156,
      "grad_norm": 0.08618927633672457,
      "learning_rate": 2.6217302609582512e-06,
      "loss": 0.4187,
      "num_tokens": 1618963558.0,
      "step": 1708
    },
    {
      "epoch": 3.3447600391772774,
      "grad_norm": 0.08731475338949538,
      "learning_rate": 2.616156211326875e-06,
      "loss": 0.4184,
      "num_tokens": 1619928523.0,
      "step": 1709
    },
    {
      "epoch": 3.3467189030362388,
      "grad_norm": 0.08528902861499998,
      "learning_rate": 2.610585993248843e-06,
      "loss": 0.4256,
      "num_tokens": 1620886650.0,
      "step": 1710
    },
    {
      "epoch": 3.348677766895201,
      "grad_norm": 0.08460023583505598,
      "learning_rate": 2.605019615677169e-06,
      "loss": 0.4083,
      "num_tokens": 1621808402.0,
      "step": 1711
    },
    {
      "epoch": 3.3506366307541624,
      "grad_norm": 0.08335951649806546,
      "learning_rate": 2.599457087558691e-06,
      "loss": 0.4041,
      "num_tokens": 1622768658.0,
      "step": 1712
    },
    {
      "epoch": 3.3525954946131242,
      "grad_norm": 0.0847151441908554,
      "learning_rate": 2.5938984178340686e-06,
      "loss": 0.406,
      "num_tokens": 1623691927.0,
      "step": 1713
    },
    {
      "epoch": 3.354554358472086,
      "grad_norm": 0.08641642780825894,
      "learning_rate": 2.588343615437752e-06,
      "loss": 0.4106,
      "num_tokens": 1624642137.0,
      "step": 1714
    },
    {
      "epoch": 3.356513222331048,
      "grad_norm": 0.08299067286267417,
      "learning_rate": 2.582792689297975e-06,
      "loss": 0.4235,
      "num_tokens": 1625588285.0,
      "step": 1715
    },
    {
      "epoch": 3.3584720861900097,
      "grad_norm": 0.08682643569946621,
      "learning_rate": 2.57724564833675e-06,
      "loss": 0.4127,
      "num_tokens": 1626542713.0,
      "step": 1716
    },
    {
      "epoch": 3.3604309500489715,
      "grad_norm": 0.08258307177948966,
      "learning_rate": 2.5717025014698347e-06,
      "loss": 0.4092,
      "num_tokens": 1627518098.0,
      "step": 1717
    },
    {
      "epoch": 3.3623898139079333,
      "grad_norm": 0.08623905493619788,
      "learning_rate": 2.5661632576067315e-06,
      "loss": 0.4152,
      "num_tokens": 1628468117.0,
      "step": 1718
    },
    {
      "epoch": 3.364348677766895,
      "grad_norm": 0.08295413915630635,
      "learning_rate": 2.5606279256506726e-06,
      "loss": 0.4286,
      "num_tokens": 1629417735.0,
      "step": 1719
    },
    {
      "epoch": 3.366307541625857,
      "grad_norm": 0.08725054043735624,
      "learning_rate": 2.5550965144985993e-06,
      "loss": 0.4082,
      "num_tokens": 1630334417.0,
      "step": 1720
    },
    {
      "epoch": 3.368266405484819,
      "grad_norm": 0.08490228890645754,
      "learning_rate": 2.5495690330411488e-06,
      "loss": 0.4119,
      "num_tokens": 1631291605.0,
      "step": 1721
    },
    {
      "epoch": 3.3702252693437806,
      "grad_norm": 0.08891319542355251,
      "learning_rate": 2.5440454901626487e-06,
      "loss": 0.4205,
      "num_tokens": 1632218898.0,
      "step": 1722
    },
    {
      "epoch": 3.3721841332027425,
      "grad_norm": 0.08969720499224697,
      "learning_rate": 2.5385258947410908e-06,
      "loss": 0.4218,
      "num_tokens": 1633173091.0,
      "step": 1723
    },
    {
      "epoch": 3.3741429970617043,
      "grad_norm": 0.088167415198806,
      "learning_rate": 2.5330102556481206e-06,
      "loss": 0.4174,
      "num_tokens": 1634100421.0,
      "step": 1724
    },
    {
      "epoch": 3.376101860920666,
      "grad_norm": 0.08340958418579227,
      "learning_rate": 2.5274985817490327e-06,
      "loss": 0.4183,
      "num_tokens": 1635065644.0,
      "step": 1725
    },
    {
      "epoch": 3.378060724779628,
      "grad_norm": 0.09041616739164081,
      "learning_rate": 2.5219908819027415e-06,
      "loss": 0.4315,
      "num_tokens": 1636000356.0,
      "step": 1726
    },
    {
      "epoch": 3.3800195886385898,
      "grad_norm": 0.08300502331043028,
      "learning_rate": 2.516487164961775e-06,
      "loss": 0.4051,
      "num_tokens": 1636952582.0,
      "step": 1727
    },
    {
      "epoch": 3.3819784524975516,
      "grad_norm": 0.08426560137119644,
      "learning_rate": 2.510987439772261e-06,
      "loss": 0.4105,
      "num_tokens": 1637916269.0,
      "step": 1728
    },
    {
      "epoch": 3.3839373163565134,
      "grad_norm": 0.08522213583301294,
      "learning_rate": 2.505491715173908e-06,
      "loss": 0.4174,
      "num_tokens": 1638871893.0,
      "step": 1729
    },
    {
      "epoch": 3.3858961802154752,
      "grad_norm": 0.08801999823760176,
      "learning_rate": 2.5000000000000015e-06,
      "loss": 0.4127,
      "num_tokens": 1639846947.0,
      "step": 1730
    },
    {
      "epoch": 3.3878550440744366,
      "grad_norm": 0.08574862503642781,
      "learning_rate": 2.4945123030773755e-06,
      "loss": 0.404,
      "num_tokens": 1640807461.0,
      "step": 1731
    },
    {
      "epoch": 3.389813907933399,
      "grad_norm": 0.08277495860589412,
      "learning_rate": 2.489028633226406e-06,
      "loss": 0.4186,
      "num_tokens": 1641778118.0,
      "step": 1732
    },
    {
      "epoch": 3.3917727717923603,
      "grad_norm": 0.08805180798992134,
      "learning_rate": 2.4835489992610035e-06,
      "loss": 0.4125,
      "num_tokens": 1642692304.0,
      "step": 1733
    },
    {
      "epoch": 3.393731635651322,
      "grad_norm": 0.08482598218207796,
      "learning_rate": 2.478073409988583e-06,
      "loss": 0.406,
      "num_tokens": 1643641294.0,
      "step": 1734
    },
    {
      "epoch": 3.395690499510284,
      "grad_norm": 0.08397140001054886,
      "learning_rate": 2.4726018742100617e-06,
      "loss": 0.4282,
      "num_tokens": 1644637554.0,
      "step": 1735
    },
    {
      "epoch": 3.3976493633692457,
      "grad_norm": 0.08735993915989847,
      "learning_rate": 2.467134400719844e-06,
      "loss": 0.4122,
      "num_tokens": 1645572744.0,
      "step": 1736
    },
    {
      "epoch": 3.3996082272282075,
      "grad_norm": 0.08421742427245288,
      "learning_rate": 2.461670998305802e-06,
      "loss": 0.409,
      "num_tokens": 1646522141.0,
      "step": 1737
    },
    {
      "epoch": 3.4015670910871694,
      "grad_norm": 0.08333993516563225,
      "learning_rate": 2.4562116757492625e-06,
      "loss": 0.4205,
      "num_tokens": 1647474167.0,
      "step": 1738
    },
    {
      "epoch": 3.403525954946131,
      "grad_norm": 0.08824166242437516,
      "learning_rate": 2.4507564418250022e-06,
      "loss": 0.4139,
      "num_tokens": 1648409454.0,
      "step": 1739
    },
    {
      "epoch": 3.405484818805093,
      "grad_norm": 0.08423873827597493,
      "learning_rate": 2.4453053053012187e-06,
      "loss": 0.4078,
      "num_tokens": 1649366647.0,
      "step": 1740
    },
    {
      "epoch": 3.407443682664055,
      "grad_norm": 0.08439380435889429,
      "learning_rate": 2.439858274939525e-06,
      "loss": 0.4098,
      "num_tokens": 1650335154.0,
      "step": 1741
    },
    {
      "epoch": 3.4094025465230167,
      "grad_norm": 0.08659772005961898,
      "learning_rate": 2.43441535949494e-06,
      "loss": 0.4186,
      "num_tokens": 1651290121.0,
      "step": 1742
    },
    {
      "epoch": 3.4113614103819785,
      "grad_norm": 0.08613026733417209,
      "learning_rate": 2.4289765677158616e-06,
      "loss": 0.4192,
      "num_tokens": 1652203049.0,
      "step": 1743
    },
    {
      "epoch": 3.4133202742409403,
      "grad_norm": 0.08919702377634635,
      "learning_rate": 2.4235419083440615e-06,
      "loss": 0.4178,
      "num_tokens": 1653161862.0,
      "step": 1744
    },
    {
      "epoch": 3.415279138099902,
      "grad_norm": 0.08359942286714173,
      "learning_rate": 2.4181113901146737e-06,
      "loss": 0.4081,
      "num_tokens": 1654102435.0,
      "step": 1745
    },
    {
      "epoch": 3.417238001958864,
      "grad_norm": 0.08656263278449895,
      "learning_rate": 2.41268502175617e-06,
      "loss": 0.4124,
      "num_tokens": 1655048320.0,
      "step": 1746
    },
    {
      "epoch": 3.419196865817826,
      "grad_norm": 0.08764171546401789,
      "learning_rate": 2.4072628119903574e-06,
      "loss": 0.4097,
      "num_tokens": 1655975517.0,
      "step": 1747
    },
    {
      "epoch": 3.4211557296767876,
      "grad_norm": 0.08564032809521967,
      "learning_rate": 2.401844769532356e-06,
      "loss": 0.3934,
      "num_tokens": 1656923442.0,
      "step": 1748
    },
    {
      "epoch": 3.4231145935357494,
      "grad_norm": 0.08613560589077948,
      "learning_rate": 2.396430903090586e-06,
      "loss": 0.4071,
      "num_tokens": 1657886144.0,
      "step": 1749
    },
    {
      "epoch": 3.4250734573947113,
      "grad_norm": 0.08215056845240856,
      "learning_rate": 2.3910212213667605e-06,
      "loss": 0.4204,
      "num_tokens": 1658819601.0,
      "step": 1750
    },
    {
      "epoch": 3.427032321253673,
      "grad_norm": 0.08838797102360238,
      "learning_rate": 2.3856157330558625e-06,
      "loss": 0.4225,
      "num_tokens": 1659756549.0,
      "step": 1751
    },
    {
      "epoch": 3.4289911851126345,
      "grad_norm": 0.08737452904307148,
      "learning_rate": 2.380214446846137e-06,
      "loss": 0.4312,
      "num_tokens": 1660735494.0,
      "step": 1752
    },
    {
      "epoch": 3.4309500489715967,
      "grad_norm": 0.08352062693110998,
      "learning_rate": 2.3748173714190726e-06,
      "loss": 0.4055,
      "num_tokens": 1661703052.0,
      "step": 1753
    },
    {
      "epoch": 3.432908912830558,
      "grad_norm": 0.08615588980744507,
      "learning_rate": 2.3694245154493927e-06,
      "loss": 0.4248,
      "num_tokens": 1662666836.0,
      "step": 1754
    },
    {
      "epoch": 3.43486777668952,
      "grad_norm": 0.08368970100744642,
      "learning_rate": 2.3640358876050344e-06,
      "loss": 0.3958,
      "num_tokens": 1663611928.0,
      "step": 1755
    },
    {
      "epoch": 3.4368266405484817,
      "grad_norm": 0.08264730673412893,
      "learning_rate": 2.3586514965471462e-06,
      "loss": 0.4053,
      "num_tokens": 1664537902.0,
      "step": 1756
    },
    {
      "epoch": 3.4387855044074436,
      "grad_norm": 0.08571500674580414,
      "learning_rate": 2.3532713509300613e-06,
      "loss": 0.419,
      "num_tokens": 1665468709.0,
      "step": 1757
    },
    {
      "epoch": 3.4407443682664054,
      "grad_norm": 0.08610498491933342,
      "learning_rate": 2.3478954594012884e-06,
      "loss": 0.4087,
      "num_tokens": 1666395205.0,
      "step": 1758
    },
    {
      "epoch": 3.442703232125367,
      "grad_norm": 0.08877814131108148,
      "learning_rate": 2.3425238306015034e-06,
      "loss": 0.4179,
      "num_tokens": 1667309425.0,
      "step": 1759
    },
    {
      "epoch": 3.444662095984329,
      "grad_norm": 0.08596822518274606,
      "learning_rate": 2.337156473164526e-06,
      "loss": 0.4061,
      "num_tokens": 1668244302.0,
      "step": 1760
    },
    {
      "epoch": 3.446620959843291,
      "grad_norm": 0.08492643582989387,
      "learning_rate": 2.3317933957173146e-06,
      "loss": 0.4098,
      "num_tokens": 1669211039.0,
      "step": 1761
    },
    {
      "epoch": 3.4485798237022527,
      "grad_norm": 0.08455790693486658,
      "learning_rate": 2.3264346068799455e-06,
      "loss": 0.428,
      "num_tokens": 1670166151.0,
      "step": 1762
    },
    {
      "epoch": 3.4505386875612145,
      "grad_norm": 0.08526852184816038,
      "learning_rate": 2.321080115265601e-06,
      "loss": 0.4201,
      "num_tokens": 1671118037.0,
      "step": 1763
    },
    {
      "epoch": 3.4524975514201763,
      "grad_norm": 0.08569161667306968,
      "learning_rate": 2.3157299294805613e-06,
      "loss": 0.4136,
      "num_tokens": 1672063474.0,
      "step": 1764
    },
    {
      "epoch": 3.454456415279138,
      "grad_norm": 0.08450316742982002,
      "learning_rate": 2.310384058124181e-06,
      "loss": 0.4173,
      "num_tokens": 1673036663.0,
      "step": 1765
    },
    {
      "epoch": 3.4564152791381,
      "grad_norm": 0.0856099935556782,
      "learning_rate": 2.3050425097888802e-06,
      "loss": 0.4173,
      "num_tokens": 1673968432.0,
      "step": 1766
    },
    {
      "epoch": 3.458374142997062,
      "grad_norm": 0.08462805783740393,
      "learning_rate": 2.299705293060136e-06,
      "loss": 0.4144,
      "num_tokens": 1674934091.0,
      "step": 1767
    },
    {
      "epoch": 3.4603330068560236,
      "grad_norm": 0.0845664978809549,
      "learning_rate": 2.2943724165164583e-06,
      "loss": 0.4144,
      "num_tokens": 1675887098.0,
      "step": 1768
    },
    {
      "epoch": 3.4622918707149855,
      "grad_norm": 0.08952609770725463,
      "learning_rate": 2.289043888729379e-06,
      "loss": 0.4056,
      "num_tokens": 1676869567.0,
      "step": 1769
    },
    {
      "epoch": 3.4642507345739473,
      "grad_norm": 0.08254427905437188,
      "learning_rate": 2.2837197182634484e-06,
      "loss": 0.4281,
      "num_tokens": 1677838703.0,
      "step": 1770
    },
    {
      "epoch": 3.466209598432909,
      "grad_norm": 0.09066133696819692,
      "learning_rate": 2.2783999136762064e-06,
      "loss": 0.411,
      "num_tokens": 1678757313.0,
      "step": 1771
    },
    {
      "epoch": 3.468168462291871,
      "grad_norm": 0.08416745354032049,
      "learning_rate": 2.273084483518176e-06,
      "loss": 0.4241,
      "num_tokens": 1679689579.0,
      "step": 1772
    },
    {
      "epoch": 3.4701273261508323,
      "grad_norm": 0.09237198051598337,
      "learning_rate": 2.2677734363328534e-06,
      "loss": 0.4134,
      "num_tokens": 1680648641.0,
      "step": 1773
    },
    {
      "epoch": 3.472086190009794,
      "grad_norm": 0.08633902308442941,
      "learning_rate": 2.262466780656687e-06,
      "loss": 0.4256,
      "num_tokens": 1681548281.0,
      "step": 1774
    },
    {
      "epoch": 3.474045053868756,
      "grad_norm": 0.08747874225822623,
      "learning_rate": 2.2571645250190643e-06,
      "loss": 0.4254,
      "num_tokens": 1682526137.0,
      "step": 1775
    },
    {
      "epoch": 3.4760039177277178,
      "grad_norm": 0.0838620168697465,
      "learning_rate": 2.2518666779423078e-06,
      "loss": 0.411,
      "num_tokens": 1683479152.0,
      "step": 1776
    },
    {
      "epoch": 3.4779627815866796,
      "grad_norm": 0.08879126737134915,
      "learning_rate": 2.2465732479416476e-06,
      "loss": 0.4053,
      "num_tokens": 1684424051.0,
      "step": 1777
    },
    {
      "epoch": 3.4799216454456414,
      "grad_norm": 0.08211408450445824,
      "learning_rate": 2.241284243525215e-06,
      "loss": 0.4131,
      "num_tokens": 1685334537.0,
      "step": 1778
    },
    {
      "epoch": 3.4818805093046032,
      "grad_norm": 0.09247793438793202,
      "learning_rate": 2.2359996731940348e-06,
      "loss": 0.4113,
      "num_tokens": 1686285600.0,
      "step": 1779
    },
    {
      "epoch": 3.483839373163565,
      "grad_norm": 0.08267329357229923,
      "learning_rate": 2.230719545441992e-06,
      "loss": 0.4134,
      "num_tokens": 1687219733.0,
      "step": 1780
    },
    {
      "epoch": 3.485798237022527,
      "grad_norm": 0.08535859600841918,
      "learning_rate": 2.225443868755845e-06,
      "loss": 0.4141,
      "num_tokens": 1688156805.0,
      "step": 1781
    },
    {
      "epoch": 3.4877571008814887,
      "grad_norm": 0.08737140079566375,
      "learning_rate": 2.2201726516151885e-06,
      "loss": 0.4089,
      "num_tokens": 1689080953.0,
      "step": 1782
    },
    {
      "epoch": 3.4897159647404505,
      "grad_norm": 0.08292150516674358,
      "learning_rate": 2.214905902492452e-06,
      "loss": 0.4135,
      "num_tokens": 1690064045.0,
      "step": 1783
    },
    {
      "epoch": 3.4916748285994124,
      "grad_norm": 0.09017231440685045,
      "learning_rate": 2.2096436298528865e-06,
      "loss": 0.4166,
      "num_tokens": 1691013524.0,
      "step": 1784
    },
    {
      "epoch": 3.493633692458374,
      "grad_norm": 0.08898438261625373,
      "learning_rate": 2.204385842154545e-06,
      "loss": 0.4224,
      "num_tokens": 1691950030.0,
      "step": 1785
    },
    {
      "epoch": 3.495592556317336,
      "grad_norm": 0.092535999074826,
      "learning_rate": 2.1991325478482695e-06,
      "loss": 0.4252,
      "num_tokens": 1692853484.0,
      "step": 1786
    },
    {
      "epoch": 3.497551420176298,
      "grad_norm": 0.08892908936281196,
      "learning_rate": 2.1938837553776872e-06,
      "loss": 0.4145,
      "num_tokens": 1693797005.0,
      "step": 1787
    },
    {
      "epoch": 3.4995102840352597,
      "grad_norm": 0.08770780187605762,
      "learning_rate": 2.1886394731791814e-06,
      "loss": 0.4125,
      "num_tokens": 1694718544.0,
      "step": 1788
    },
    {
      "epoch": 3.5014691478942215,
      "grad_norm": 0.08925655869184221,
      "learning_rate": 2.1833997096818897e-06,
      "loss": 0.4255,
      "num_tokens": 1695664993.0,
      "step": 1789
    },
    {
      "epoch": 3.5034280117531833,
      "grad_norm": 0.0825690667279722,
      "learning_rate": 2.1781644733076897e-06,
      "loss": 0.4099,
      "num_tokens": 1696629356.0,
      "step": 1790
    },
    {
      "epoch": 3.505386875612145,
      "grad_norm": 0.08589096944740693,
      "learning_rate": 2.172933772471177e-06,
      "loss": 0.4028,
      "num_tokens": 1697588346.0,
      "step": 1791
    },
    {
      "epoch": 3.5073457394711065,
      "grad_norm": 0.087846075951044,
      "learning_rate": 2.1677076155796583e-06,
      "loss": 0.418,
      "num_tokens": 1698542320.0,
      "step": 1792
    },
    {
      "epoch": 3.5093046033300688,
      "grad_norm": 0.08342957076022849,
      "learning_rate": 2.162486011033142e-06,
      "loss": 0.4307,
      "num_tokens": 1699526083.0,
      "step": 1793
    },
    {
      "epoch": 3.51126346718903,
      "grad_norm": 0.08745366835845464,
      "learning_rate": 2.157268967224314e-06,
      "loss": 0.4268,
      "num_tokens": 1700468996.0,
      "step": 1794
    },
    {
      "epoch": 3.5132223310479924,
      "grad_norm": 0.09869232946554855,
      "learning_rate": 2.1520564925385294e-06,
      "loss": 0.4213,
      "num_tokens": 1701421419.0,
      "step": 1795
    },
    {
      "epoch": 3.515181194906954,
      "grad_norm": 0.08292798367719592,
      "learning_rate": 2.1468485953538054e-06,
      "loss": 0.4086,
      "num_tokens": 1702323652.0,
      "step": 1796
    },
    {
      "epoch": 3.5171400587659156,
      "grad_norm": 0.08837785187933994,
      "learning_rate": 2.141645284040796e-06,
      "loss": 0.4245,
      "num_tokens": 1703259414.0,
      "step": 1797
    },
    {
      "epoch": 3.5190989226248774,
      "grad_norm": 0.08503007192272861,
      "learning_rate": 2.136446566962784e-06,
      "loss": 0.4231,
      "num_tokens": 1704203270.0,
      "step": 1798
    },
    {
      "epoch": 3.5210577864838393,
      "grad_norm": 0.08523179691963496,
      "learning_rate": 2.1312524524756735e-06,
      "loss": 0.4128,
      "num_tokens": 1705165703.0,
      "step": 1799
    },
    {
      "epoch": 3.523016650342801,
      "grad_norm": 0.08678239735374162,
      "learning_rate": 2.1260629489279662e-06,
      "loss": 0.4208,
      "num_tokens": 1706076773.0,
      "step": 1800
    },
    {
      "epoch": 3.524975514201763,
      "grad_norm": 0.08584504515243888,
      "learning_rate": 2.1208780646607517e-06,
      "loss": 0.4174,
      "num_tokens": 1707027549.0,
      "step": 1801
    },
    {
      "epoch": 3.5269343780607247,
      "grad_norm": 0.08735485016531032,
      "learning_rate": 2.115697808007701e-06,
      "loss": 0.4231,
      "num_tokens": 1707936544.0,
      "step": 1802
    },
    {
      "epoch": 3.5288932419196866,
      "grad_norm": 0.09003271743520458,
      "learning_rate": 2.1105221872950403e-06,
      "loss": 0.4226,
      "num_tokens": 1708874517.0,
      "step": 1803
    },
    {
      "epoch": 3.5308521057786484,
      "grad_norm": 0.08535192791935681,
      "learning_rate": 2.1053512108415476e-06,
      "loss": 0.4063,
      "num_tokens": 1709823687.0,
      "step": 1804
    },
    {
      "epoch": 3.53281096963761,
      "grad_norm": 0.0829088964822582,
      "learning_rate": 2.1001848869585396e-06,
      "loss": 0.4083,
      "num_tokens": 1710757139.0,
      "step": 1805
    },
    {
      "epoch": 3.534769833496572,
      "grad_norm": 0.09425316059515856,
      "learning_rate": 2.095023223949845e-06,
      "loss": 0.4138,
      "num_tokens": 1711714145.0,
      "step": 1806
    },
    {
      "epoch": 3.536728697355534,
      "grad_norm": 0.08363603545769085,
      "learning_rate": 2.089866230111813e-06,
      "loss": 0.4084,
      "num_tokens": 1712644160.0,
      "step": 1807
    },
    {
      "epoch": 3.5386875612144957,
      "grad_norm": 0.08797244156608434,
      "learning_rate": 2.0847139137332795e-06,
      "loss": 0.4233,
      "num_tokens": 1713581770.0,
      "step": 1808
    },
    {
      "epoch": 3.5406464250734575,
      "grad_norm": 0.08429236686683463,
      "learning_rate": 2.079566283095565e-06,
      "loss": 0.4052,
      "num_tokens": 1714528173.0,
      "step": 1809
    },
    {
      "epoch": 3.5426052889324193,
      "grad_norm": 0.08511651079994069,
      "learning_rate": 2.074423346472461e-06,
      "loss": 0.4275,
      "num_tokens": 1715502773.0,
      "step": 1810
    },
    {
      "epoch": 3.544564152791381,
      "grad_norm": 0.081576130787359,
      "learning_rate": 2.0692851121302114e-06,
      "loss": 0.4255,
      "num_tokens": 1716463993.0,
      "step": 1811
    },
    {
      "epoch": 3.546523016650343,
      "grad_norm": 0.08356993766841732,
      "learning_rate": 2.064151588327501e-06,
      "loss": 0.4219,
      "num_tokens": 1717433634.0,
      "step": 1812
    },
    {
      "epoch": 3.5484818805093044,
      "grad_norm": 0.09094645398636986,
      "learning_rate": 2.0590227833154485e-06,
      "loss": 0.4293,
      "num_tokens": 1718382354.0,
      "step": 1813
    },
    {
      "epoch": 3.5504407443682666,
      "grad_norm": 0.08261497622365453,
      "learning_rate": 2.053898705337583e-06,
      "loss": 0.4125,
      "num_tokens": 1719353516.0,
      "step": 1814
    },
    {
      "epoch": 3.552399608227228,
      "grad_norm": 0.08463768612829826,
      "learning_rate": 2.0487793626298364e-06,
      "loss": 0.4137,
      "num_tokens": 1720312136.0,
      "step": 1815
    },
    {
      "epoch": 3.5543584720861903,
      "grad_norm": 0.08495475667173523,
      "learning_rate": 2.043664763420534e-06,
      "loss": 0.41,
      "num_tokens": 1721287673.0,
      "step": 1816
    },
    {
      "epoch": 3.5563173359451516,
      "grad_norm": 0.08244986328912285,
      "learning_rate": 2.0385549159303716e-06,
      "loss": 0.413,
      "num_tokens": 1722203009.0,
      "step": 1817
    },
    {
      "epoch": 3.5582761998041135,
      "grad_norm": 0.08699776816484583,
      "learning_rate": 2.033449828372408e-06,
      "loss": 0.4114,
      "num_tokens": 1723169179.0,
      "step": 1818
    },
    {
      "epoch": 3.5602350636630753,
      "grad_norm": 0.08655242975299911,
      "learning_rate": 2.028349508952057e-06,
      "loss": 0.4149,
      "num_tokens": 1724074552.0,
      "step": 1819
    },
    {
      "epoch": 3.562193927522037,
      "grad_norm": 0.08471053153555338,
      "learning_rate": 2.023253965867063e-06,
      "loss": 0.4231,
      "num_tokens": 1725034811.0,
      "step": 1820
    },
    {
      "epoch": 3.564152791380999,
      "grad_norm": 0.08479434667529531,
      "learning_rate": 2.0181632073074925e-06,
      "loss": 0.4239,
      "num_tokens": 1726000629.0,
      "step": 1821
    },
    {
      "epoch": 3.5661116552399608,
      "grad_norm": 0.08532687672569757,
      "learning_rate": 2.013077241455729e-06,
      "loss": 0.4114,
      "num_tokens": 1726961708.0,
      "step": 1822
    },
    {
      "epoch": 3.5680705190989226,
      "grad_norm": 0.0833140131608997,
      "learning_rate": 2.0079960764864453e-06,
      "loss": 0.4153,
      "num_tokens": 1727917436.0,
      "step": 1823
    },
    {
      "epoch": 3.5700293829578844,
      "grad_norm": 0.08153647040277169,
      "learning_rate": 2.002919720566599e-06,
      "loss": 0.4118,
      "num_tokens": 1728906407.0,
      "step": 1824
    },
    {
      "epoch": 3.5719882468168462,
      "grad_norm": 0.08754391286573099,
      "learning_rate": 1.997848181855423e-06,
      "loss": 0.4165,
      "num_tokens": 1729857227.0,
      "step": 1825
    },
    {
      "epoch": 3.573947110675808,
      "grad_norm": 0.08388968292243014,
      "learning_rate": 1.992781468504402e-06,
      "loss": 0.4054,
      "num_tokens": 1730804975.0,
      "step": 1826
    },
    {
      "epoch": 3.57590597453477,
      "grad_norm": 0.08525964255363742,
      "learning_rate": 1.9877195886572666e-06,
      "loss": 0.4117,
      "num_tokens": 1731781764.0,
      "step": 1827
    },
    {
      "epoch": 3.5778648383937317,
      "grad_norm": 0.08351103254975643,
      "learning_rate": 1.9826625504499807e-06,
      "loss": 0.4321,
      "num_tokens": 1732770357.0,
      "step": 1828
    },
    {
      "epoch": 3.5798237022526935,
      "grad_norm": 0.08216601091778715,
      "learning_rate": 1.9776103620107235e-06,
      "loss": 0.4188,
      "num_tokens": 1733702202.0,
      "step": 1829
    },
    {
      "epoch": 3.5817825661116554,
      "grad_norm": 0.08917446966353505,
      "learning_rate": 1.972563031459878e-06,
      "loss": 0.4361,
      "num_tokens": 1734625040.0,
      "step": 1830
    },
    {
      "epoch": 3.583741429970617,
      "grad_norm": 0.08677643703346584,
      "learning_rate": 1.967520566910025e-06,
      "loss": 0.4193,
      "num_tokens": 1735583871.0,
      "step": 1831
    },
    {
      "epoch": 3.585700293829579,
      "grad_norm": 0.08713642480637289,
      "learning_rate": 1.962482976465918e-06,
      "loss": 0.4135,
      "num_tokens": 1736537854.0,
      "step": 1832
    },
    {
      "epoch": 3.587659157688541,
      "grad_norm": 0.08973041504995834,
      "learning_rate": 1.9574502682244793e-06,
      "loss": 0.4196,
      "num_tokens": 1737479499.0,
      "step": 1833
    },
    {
      "epoch": 3.589618021547502,
      "grad_norm": 0.08471327244617989,
      "learning_rate": 1.952422450274784e-06,
      "loss": 0.4111,
      "num_tokens": 1738430189.0,
      "step": 1834
    },
    {
      "epoch": 3.5915768854064645,
      "grad_norm": 0.08432918776696068,
      "learning_rate": 1.947399530698043e-06,
      "loss": 0.4072,
      "num_tokens": 1739410135.0,
      "step": 1835
    },
    {
      "epoch": 3.593535749265426,
      "grad_norm": 0.08421464213829939,
      "learning_rate": 1.9423815175676027e-06,
      "loss": 0.4166,
      "num_tokens": 1740341982.0,
      "step": 1836
    },
    {
      "epoch": 3.595494613124388,
      "grad_norm": 0.08687752015803595,
      "learning_rate": 1.9373684189489156e-06,
      "loss": 0.4239,
      "num_tokens": 1741273342.0,
      "step": 1837
    },
    {
      "epoch": 3.5974534769833495,
      "grad_norm": 0.09046963770564236,
      "learning_rate": 1.9323602428995358e-06,
      "loss": 0.4207,
      "num_tokens": 1742235920.0,
      "step": 1838
    },
    {
      "epoch": 3.5994123408423113,
      "grad_norm": 0.08623590989277806,
      "learning_rate": 1.9273569974691107e-06,
      "loss": 0.427,
      "num_tokens": 1743203172.0,
      "step": 1839
    },
    {
      "epoch": 3.601371204701273,
      "grad_norm": 0.08598574944302184,
      "learning_rate": 1.9223586906993586e-06,
      "loss": 0.3957,
      "num_tokens": 1744130229.0,
      "step": 1840
    },
    {
      "epoch": 3.603330068560235,
      "grad_norm": 0.08426701231885612,
      "learning_rate": 1.917365330624057e-06,
      "loss": 0.4152,
      "num_tokens": 1745108939.0,
      "step": 1841
    },
    {
      "epoch": 3.605288932419197,
      "grad_norm": 0.08791378402919472,
      "learning_rate": 1.912376925269041e-06,
      "loss": 0.4241,
      "num_tokens": 1746050426.0,
      "step": 1842
    },
    {
      "epoch": 3.6072477962781586,
      "grad_norm": 0.08707336976847711,
      "learning_rate": 1.907393482652175e-06,
      "loss": 0.4153,
      "num_tokens": 1747061149.0,
      "step": 1843
    },
    {
      "epoch": 3.6092066601371204,
      "grad_norm": 0.08072513353149141,
      "learning_rate": 1.9024150107833473e-06,
      "loss": 0.4046,
      "num_tokens": 1747983306.0,
      "step": 1844
    },
    {
      "epoch": 3.6111655239960823,
      "grad_norm": 0.08520789486374623,
      "learning_rate": 1.8974415176644607e-06,
      "loss": 0.4198,
      "num_tokens": 1748924913.0,
      "step": 1845
    },
    {
      "epoch": 3.613124387855044,
      "grad_norm": 0.08932671544709427,
      "learning_rate": 1.8924730112894129e-06,
      "loss": 0.4189,
      "num_tokens": 1749899433.0,
      "step": 1846
    },
    {
      "epoch": 3.615083251714006,
      "grad_norm": 0.08653129735499375,
      "learning_rate": 1.887509499644084e-06,
      "loss": 0.4166,
      "num_tokens": 1750830578.0,
      "step": 1847
    },
    {
      "epoch": 3.6170421155729677,
      "grad_norm": 0.0872920067808658,
      "learning_rate": 1.8825509907063328e-06,
      "loss": 0.4284,
      "num_tokens": 1751755284.0,
      "step": 1848
    },
    {
      "epoch": 3.6190009794319296,
      "grad_norm": 0.08568505102634366,
      "learning_rate": 1.8775974924459716e-06,
      "loss": 0.4106,
      "num_tokens": 1752721987.0,
      "step": 1849
    },
    {
      "epoch": 3.6209598432908914,
      "grad_norm": 0.08651794932163524,
      "learning_rate": 1.872649012824758e-06,
      "loss": 0.4339,
      "num_tokens": 1753661497.0,
      "step": 1850
    },
    {
      "epoch": 3.622918707149853,
      "grad_norm": 0.08521313183514503,
      "learning_rate": 1.867705559796391e-06,
      "loss": 0.3973,
      "num_tokens": 1754612562.0,
      "step": 1851
    },
    {
      "epoch": 3.624877571008815,
      "grad_norm": 0.0826221539214511,
      "learning_rate": 1.8627671413064825e-06,
      "loss": 0.4198,
      "num_tokens": 1755555673.0,
      "step": 1852
    },
    {
      "epoch": 3.626836434867777,
      "grad_norm": 0.08246182928242993,
      "learning_rate": 1.8578337652925527e-06,
      "loss": 0.4093,
      "num_tokens": 1756497461.0,
      "step": 1853
    },
    {
      "epoch": 3.6287952987267387,
      "grad_norm": 0.08469517615317272,
      "learning_rate": 1.8529054396840234e-06,
      "loss": 0.4028,
      "num_tokens": 1757467410.0,
      "step": 1854
    },
    {
      "epoch": 3.6307541625857,
      "grad_norm": 0.08550398418039426,
      "learning_rate": 1.847982172402194e-06,
      "loss": 0.4088,
      "num_tokens": 1758414079.0,
      "step": 1855
    },
    {
      "epoch": 3.6327130264446623,
      "grad_norm": 0.08689088625548672,
      "learning_rate": 1.8430639713602317e-06,
      "loss": 0.4148,
      "num_tokens": 1759353247.0,
      "step": 1856
    },
    {
      "epoch": 3.6346718903036237,
      "grad_norm": 0.08363548236902174,
      "learning_rate": 1.8381508444631674e-06,
      "loss": 0.4135,
      "num_tokens": 1760292625.0,
      "step": 1857
    },
    {
      "epoch": 3.636630754162586,
      "grad_norm": 0.08520877105783718,
      "learning_rate": 1.8332427996078705e-06,
      "loss": 0.4119,
      "num_tokens": 1761257171.0,
      "step": 1858
    },
    {
      "epoch": 3.6385896180215473,
      "grad_norm": 0.08438989396775987,
      "learning_rate": 1.8283398446830442e-06,
      "loss": 0.4143,
      "num_tokens": 1762209289.0,
      "step": 1859
    },
    {
      "epoch": 3.640548481880509,
      "grad_norm": 0.08423480034920917,
      "learning_rate": 1.8234419875692105e-06,
      "loss": 0.4182,
      "num_tokens": 1763123626.0,
      "step": 1860
    },
    {
      "epoch": 3.642507345739471,
      "grad_norm": 0.084598886311569,
      "learning_rate": 1.8185492361386958e-06,
      "loss": 0.4093,
      "num_tokens": 1764058676.0,
      "step": 1861
    },
    {
      "epoch": 3.644466209598433,
      "grad_norm": 0.08399823936656207,
      "learning_rate": 1.8136615982556256e-06,
      "loss": 0.4009,
      "num_tokens": 1765020172.0,
      "step": 1862
    },
    {
      "epoch": 3.6464250734573946,
      "grad_norm": 0.08963877149509505,
      "learning_rate": 1.808779081775901e-06,
      "loss": 0.4216,
      "num_tokens": 1765959369.0,
      "step": 1863
    },
    {
      "epoch": 3.6483839373163565,
      "grad_norm": 0.08594458469416463,
      "learning_rate": 1.8039016945471915e-06,
      "loss": 0.4131,
      "num_tokens": 1766895719.0,
      "step": 1864
    },
    {
      "epoch": 3.6503428011753183,
      "grad_norm": 0.08596876907518203,
      "learning_rate": 1.799029444408928e-06,
      "loss": 0.4243,
      "num_tokens": 1767825589.0,
      "step": 1865
    },
    {
      "epoch": 3.65230166503428,
      "grad_norm": 0.08255312354050273,
      "learning_rate": 1.7941623391922775e-06,
      "loss": 0.4012,
      "num_tokens": 1768765820.0,
      "step": 1866
    },
    {
      "epoch": 3.654260528893242,
      "grad_norm": 0.08435650099899289,
      "learning_rate": 1.789300386720141e-06,
      "loss": 0.4177,
      "num_tokens": 1769725106.0,
      "step": 1867
    },
    {
      "epoch": 3.6562193927522038,
      "grad_norm": 0.08343230599318063,
      "learning_rate": 1.7844435948071387e-06,
      "loss": 0.4209,
      "num_tokens": 1770653757.0,
      "step": 1868
    },
    {
      "epoch": 3.6581782566111656,
      "grad_norm": 0.08452327308623443,
      "learning_rate": 1.7795919712595943e-06,
      "loss": 0.4277,
      "num_tokens": 1771637772.0,
      "step": 1869
    },
    {
      "epoch": 3.6601371204701274,
      "grad_norm": 0.0841760619426903,
      "learning_rate": 1.7747455238755223e-06,
      "loss": 0.4372,
      "num_tokens": 1772582387.0,
      "step": 1870
    },
    {
      "epoch": 3.6620959843290892,
      "grad_norm": 0.08574535958399185,
      "learning_rate": 1.7699042604446243e-06,
      "loss": 0.41,
      "num_tokens": 1773544097.0,
      "step": 1871
    },
    {
      "epoch": 3.664054848188051,
      "grad_norm": 0.08557968582975219,
      "learning_rate": 1.7650681887482628e-06,
      "loss": 0.421,
      "num_tokens": 1774479049.0,
      "step": 1872
    },
    {
      "epoch": 3.666013712047013,
      "grad_norm": 0.08368340503368406,
      "learning_rate": 1.7602373165594578e-06,
      "loss": 0.4137,
      "num_tokens": 1775401619.0,
      "step": 1873
    },
    {
      "epoch": 3.6679725759059747,
      "grad_norm": 0.08569041934847592,
      "learning_rate": 1.7554116516428753e-06,
      "loss": 0.4128,
      "num_tokens": 1776373602.0,
      "step": 1874
    },
    {
      "epoch": 3.6699314397649365,
      "grad_norm": 0.0847805857155678,
      "learning_rate": 1.7505912017548054e-06,
      "loss": 0.4331,
      "num_tokens": 1777351845.0,
      "step": 1875
    },
    {
      "epoch": 3.671890303623898,
      "grad_norm": 0.0849496557044052,
      "learning_rate": 1.7457759746431635e-06,
      "loss": 0.4131,
      "num_tokens": 1778286086.0,
      "step": 1876
    },
    {
      "epoch": 3.67384916748286,
      "grad_norm": 0.0859433881590437,
      "learning_rate": 1.7409659780474652e-06,
      "loss": 0.4101,
      "num_tokens": 1779223864.0,
      "step": 1877
    },
    {
      "epoch": 3.6758080313418215,
      "grad_norm": 0.08534200625136383,
      "learning_rate": 1.7361612196988177e-06,
      "loss": 0.4214,
      "num_tokens": 1780182672.0,
      "step": 1878
    },
    {
      "epoch": 3.677766895200784,
      "grad_norm": 0.08665887542010337,
      "learning_rate": 1.7313617073199163e-06,
      "loss": 0.417,
      "num_tokens": 1781173761.0,
      "step": 1879
    },
    {
      "epoch": 3.679725759059745,
      "grad_norm": 0.09091683884933376,
      "learning_rate": 1.7265674486250167e-06,
      "loss": 0.4146,
      "num_tokens": 1782087403.0,
      "step": 1880
    },
    {
      "epoch": 3.681684622918707,
      "grad_norm": 0.08603234926712564,
      "learning_rate": 1.7217784513199327e-06,
      "loss": 0.4231,
      "num_tokens": 1783062564.0,
      "step": 1881
    },
    {
      "epoch": 3.683643486777669,
      "grad_norm": 0.08683918881574754,
      "learning_rate": 1.7169947231020252e-06,
      "loss": 0.4184,
      "num_tokens": 1784022301.0,
      "step": 1882
    },
    {
      "epoch": 3.6856023506366307,
      "grad_norm": 0.08470340031448806,
      "learning_rate": 1.7122162716601808e-06,
      "loss": 0.4156,
      "num_tokens": 1784929460.0,
      "step": 1883
    },
    {
      "epoch": 3.6875612144955925,
      "grad_norm": 0.08464888603138751,
      "learning_rate": 1.7074431046748075e-06,
      "loss": 0.4234,
      "num_tokens": 1785858673.0,
      "step": 1884
    },
    {
      "epoch": 3.6895200783545543,
      "grad_norm": 0.09102148220171614,
      "learning_rate": 1.702675229817819e-06,
      "loss": 0.4066,
      "num_tokens": 1786793243.0,
      "step": 1885
    },
    {
      "epoch": 3.691478942213516,
      "grad_norm": 0.08619754730701117,
      "learning_rate": 1.6979126547526231e-06,
      "loss": 0.4183,
      "num_tokens": 1787739701.0,
      "step": 1886
    },
    {
      "epoch": 3.693437806072478,
      "grad_norm": 0.0834667381778445,
      "learning_rate": 1.6931553871341071e-06,
      "loss": 0.4213,
      "num_tokens": 1788684718.0,
      "step": 1887
    },
    {
      "epoch": 3.69539666993144,
      "grad_norm": 0.08408279900173775,
      "learning_rate": 1.6884034346086332e-06,
      "loss": 0.4225,
      "num_tokens": 1789637707.0,
      "step": 1888
    },
    {
      "epoch": 3.6973555337904016,
      "grad_norm": 0.08608790200130456,
      "learning_rate": 1.683656804814014e-06,
      "loss": 0.4242,
      "num_tokens": 1790633967.0,
      "step": 1889
    },
    {
      "epoch": 3.6993143976493634,
      "grad_norm": 0.0879383245546899,
      "learning_rate": 1.678915505379513e-06,
      "loss": 0.4284,
      "num_tokens": 1791628570.0,
      "step": 1890
    },
    {
      "epoch": 3.7012732615083253,
      "grad_norm": 0.0859372868559316,
      "learning_rate": 1.6741795439258218e-06,
      "loss": 0.4057,
      "num_tokens": 1792559909.0,
      "step": 1891
    },
    {
      "epoch": 3.703232125367287,
      "grad_norm": 0.08537582725539315,
      "learning_rate": 1.6694489280650516e-06,
      "loss": 0.4165,
      "num_tokens": 1793498999.0,
      "step": 1892
    },
    {
      "epoch": 3.705190989226249,
      "grad_norm": 0.08325153911763906,
      "learning_rate": 1.6647236654007276e-06,
      "loss": 0.4157,
      "num_tokens": 1794425533.0,
      "step": 1893
    },
    {
      "epoch": 3.7071498530852107,
      "grad_norm": 0.08411595431604199,
      "learning_rate": 1.660003763527765e-06,
      "loss": 0.4125,
      "num_tokens": 1795386101.0,
      "step": 1894
    },
    {
      "epoch": 3.7091087169441725,
      "grad_norm": 0.08360335887700575,
      "learning_rate": 1.6552892300324625e-06,
      "loss": 0.4143,
      "num_tokens": 1796324039.0,
      "step": 1895
    },
    {
      "epoch": 3.7110675808031344,
      "grad_norm": 0.08355029162210738,
      "learning_rate": 1.650580072492496e-06,
      "loss": 0.4267,
      "num_tokens": 1797278534.0,
      "step": 1896
    },
    {
      "epoch": 3.7130264446620957,
      "grad_norm": 0.08461827441161335,
      "learning_rate": 1.645876298476895e-06,
      "loss": 0.4347,
      "num_tokens": 1798197706.0,
      "step": 1897
    },
    {
      "epoch": 3.714985308521058,
      "grad_norm": 0.08811499346200773,
      "learning_rate": 1.641177915546036e-06,
      "loss": 0.3977,
      "num_tokens": 1799139831.0,
      "step": 1898
    },
    {
      "epoch": 3.7169441723800194,
      "grad_norm": 0.08492856828316475,
      "learning_rate": 1.6364849312516357e-06,
      "loss": 0.4243,
      "num_tokens": 1800099569.0,
      "step": 1899
    },
    {
      "epoch": 3.7189030362389817,
      "grad_norm": 0.08481567469199112,
      "learning_rate": 1.6317973531367293e-06,
      "loss": 0.41,
      "num_tokens": 1801061165.0,
      "step": 1900
    },
    {
      "epoch": 3.720861900097943,
      "grad_norm": 0.08403856147686903,
      "learning_rate": 1.6271151887356607e-06,
      "loss": 0.4062,
      "num_tokens": 1802007153.0,
      "step": 1901
    },
    {
      "epoch": 3.722820763956905,
      "grad_norm": 0.08631686762180507,
      "learning_rate": 1.6224384455740788e-06,
      "loss": 0.4212,
      "num_tokens": 1802930843.0,
      "step": 1902
    },
    {
      "epoch": 3.7247796278158667,
      "grad_norm": 0.09229068743186988,
      "learning_rate": 1.6177671311689141e-06,
      "loss": 0.4218,
      "num_tokens": 1803861097.0,
      "step": 1903
    },
    {
      "epoch": 3.7267384916748285,
      "grad_norm": 0.08531139766353613,
      "learning_rate": 1.6131012530283702e-06,
      "loss": 0.4215,
      "num_tokens": 1804762220.0,
      "step": 1904
    },
    {
      "epoch": 3.7286973555337903,
      "grad_norm": 0.0889313467619071,
      "learning_rate": 1.6084408186519195e-06,
      "loss": 0.4096,
      "num_tokens": 1805687334.0,
      "step": 1905
    },
    {
      "epoch": 3.730656219392752,
      "grad_norm": 0.08106353324046786,
      "learning_rate": 1.6037858355302788e-06,
      "loss": 0.4213,
      "num_tokens": 1806646527.0,
      "step": 1906
    },
    {
      "epoch": 3.732615083251714,
      "grad_norm": 0.08259892785556008,
      "learning_rate": 1.5991363111454023e-06,
      "loss": 0.4057,
      "num_tokens": 1807608616.0,
      "step": 1907
    },
    {
      "epoch": 3.734573947110676,
      "grad_norm": 0.0835525455556785,
      "learning_rate": 1.5944922529704776e-06,
      "loss": 0.4158,
      "num_tokens": 1808543058.0,
      "step": 1908
    },
    {
      "epoch": 3.7365328109696376,
      "grad_norm": 0.08231090323783154,
      "learning_rate": 1.589853668469899e-06,
      "loss": 0.4226,
      "num_tokens": 1809496623.0,
      "step": 1909
    },
    {
      "epoch": 3.7384916748285995,
      "grad_norm": 0.08512469678523701,
      "learning_rate": 1.5852205650992647e-06,
      "loss": 0.4038,
      "num_tokens": 1810466934.0,
      "step": 1910
    },
    {
      "epoch": 3.7404505386875613,
      "grad_norm": 0.08421443630211962,
      "learning_rate": 1.5805929503053685e-06,
      "loss": 0.4116,
      "num_tokens": 1811456339.0,
      "step": 1911
    },
    {
      "epoch": 3.742409402546523,
      "grad_norm": 0.0821152272671918,
      "learning_rate": 1.5759708315261724e-06,
      "loss": 0.4027,
      "num_tokens": 1812398210.0,
      "step": 1912
    },
    {
      "epoch": 3.744368266405485,
      "grad_norm": 0.09023731565749514,
      "learning_rate": 1.5713542161908141e-06,
      "loss": 0.4162,
      "num_tokens": 1813340179.0,
      "step": 1913
    },
    {
      "epoch": 3.7463271302644467,
      "grad_norm": 0.08504933339836979,
      "learning_rate": 1.5667431117195814e-06,
      "loss": 0.4253,
      "num_tokens": 1814310043.0,
      "step": 1914
    },
    {
      "epoch": 3.7482859941234086,
      "grad_norm": 0.08492851530814112,
      "learning_rate": 1.5621375255239018e-06,
      "loss": 0.4225,
      "num_tokens": 1815298802.0,
      "step": 1915
    },
    {
      "epoch": 3.75024485798237,
      "grad_norm": 0.08512399191992572,
      "learning_rate": 1.5575374650063406e-06,
      "loss": 0.4247,
      "num_tokens": 1816212697.0,
      "step": 1916
    },
    {
      "epoch": 3.752203721841332,
      "grad_norm": 0.08554790074590062,
      "learning_rate": 1.552942937560576e-06,
      "loss": 0.4228,
      "num_tokens": 1817152179.0,
      "step": 1917
    },
    {
      "epoch": 3.7541625857002936,
      "grad_norm": 0.08659548668907796,
      "learning_rate": 1.5483539505713918e-06,
      "loss": 0.4104,
      "num_tokens": 1818083399.0,
      "step": 1918
    },
    {
      "epoch": 3.756121449559256,
      "grad_norm": 0.08176585901436256,
      "learning_rate": 1.5437705114146735e-06,
      "loss": 0.4158,
      "num_tokens": 1819025814.0,
      "step": 1919
    },
    {
      "epoch": 3.7580803134182172,
      "grad_norm": 0.08507563429865035,
      "learning_rate": 1.539192627457382e-06,
      "loss": 0.4331,
      "num_tokens": 1819938914.0,
      "step": 1920
    },
    {
      "epoch": 3.7600391772771795,
      "grad_norm": 0.08605313690692393,
      "learning_rate": 1.5346203060575526e-06,
      "loss": 0.3975,
      "num_tokens": 1820911956.0,
      "step": 1921
    },
    {
      "epoch": 3.761998041136141,
      "grad_norm": 0.08198911838135789,
      "learning_rate": 1.530053554564282e-06,
      "loss": 0.4206,
      "num_tokens": 1821846063.0,
      "step": 1922
    },
    {
      "epoch": 3.7639569049951027,
      "grad_norm": 0.08800163657224465,
      "learning_rate": 1.5254923803177108e-06,
      "loss": 0.4188,
      "num_tokens": 1822752847.0,
      "step": 1923
    },
    {
      "epoch": 3.7659157688540645,
      "grad_norm": 0.0859162604766064,
      "learning_rate": 1.5209367906490152e-06,
      "loss": 0.4199,
      "num_tokens": 1823708565.0,
      "step": 1924
    },
    {
      "epoch": 3.7678746327130264,
      "grad_norm": 0.08619344356928228,
      "learning_rate": 1.5163867928803998e-06,
      "loss": 0.4169,
      "num_tokens": 1824668568.0,
      "step": 1925
    },
    {
      "epoch": 3.769833496571988,
      "grad_norm": 0.08501854812802791,
      "learning_rate": 1.511842394325077e-06,
      "loss": 0.4156,
      "num_tokens": 1825619257.0,
      "step": 1926
    },
    {
      "epoch": 3.77179236043095,
      "grad_norm": 0.08621020723153473,
      "learning_rate": 1.50730360228726e-06,
      "loss": 0.405,
      "num_tokens": 1826588277.0,
      "step": 1927
    },
    {
      "epoch": 3.773751224289912,
      "grad_norm": 0.0829383462089837,
      "learning_rate": 1.5027704240621533e-06,
      "loss": 0.4126,
      "num_tokens": 1827554644.0,
      "step": 1928
    },
    {
      "epoch": 3.7757100881488737,
      "grad_norm": 0.08857959937499182,
      "learning_rate": 1.4982428669359362e-06,
      "loss": 0.4133,
      "num_tokens": 1828492139.0,
      "step": 1929
    },
    {
      "epoch": 3.7776689520078355,
      "grad_norm": 0.08508540757727298,
      "learning_rate": 1.4937209381857509e-06,
      "loss": 0.4252,
      "num_tokens": 1829464363.0,
      "step": 1930
    },
    {
      "epoch": 3.7796278158667973,
      "grad_norm": 0.08805253930443299,
      "learning_rate": 1.4892046450797004e-06,
      "loss": 0.428,
      "num_tokens": 1830400969.0,
      "step": 1931
    },
    {
      "epoch": 3.781586679725759,
      "grad_norm": 0.0869356895237708,
      "learning_rate": 1.4846939948768218e-06,
      "loss": 0.4174,
      "num_tokens": 1831327031.0,
      "step": 1932
    },
    {
      "epoch": 3.783545543584721,
      "grad_norm": 0.08251678849294589,
      "learning_rate": 1.4801889948270852e-06,
      "loss": 0.4138,
      "num_tokens": 1832263846.0,
      "step": 1933
    },
    {
      "epoch": 3.7855044074436828,
      "grad_norm": 0.08346726500453072,
      "learning_rate": 1.4756896521713821e-06,
      "loss": 0.4145,
      "num_tokens": 1833211972.0,
      "step": 1934
    },
    {
      "epoch": 3.7874632713026446,
      "grad_norm": 0.08893905266831544,
      "learning_rate": 1.471195974141506e-06,
      "loss": 0.4193,
      "num_tokens": 1834160367.0,
      "step": 1935
    },
    {
      "epoch": 3.7894221351616064,
      "grad_norm": 0.08547630574089961,
      "learning_rate": 1.4667079679601475e-06,
      "loss": 0.4255,
      "num_tokens": 1835113199.0,
      "step": 1936
    },
    {
      "epoch": 3.791380999020568,
      "grad_norm": 0.08520224604626843,
      "learning_rate": 1.4622256408408852e-06,
      "loss": 0.4119,
      "num_tokens": 1836063274.0,
      "step": 1937
    },
    {
      "epoch": 3.79333986287953,
      "grad_norm": 0.09052160754255766,
      "learning_rate": 1.45774899998816e-06,
      "loss": 0.414,
      "num_tokens": 1837004423.0,
      "step": 1938
    },
    {
      "epoch": 3.7952987267384914,
      "grad_norm": 0.08285027997712183,
      "learning_rate": 1.4532780525972833e-06,
      "loss": 0.4152,
      "num_tokens": 1837949051.0,
      "step": 1939
    },
    {
      "epoch": 3.7972575905974537,
      "grad_norm": 0.08712884894974125,
      "learning_rate": 1.44881280585441e-06,
      "loss": 0.4155,
      "num_tokens": 1838888101.0,
      "step": 1940
    },
    {
      "epoch": 3.799216454456415,
      "grad_norm": 0.08826969607410523,
      "learning_rate": 1.4443532669365317e-06,
      "loss": 0.4223,
      "num_tokens": 1839861647.0,
      "step": 1941
    },
    {
      "epoch": 3.801175318315377,
      "grad_norm": 0.08673165545294591,
      "learning_rate": 1.4398994430114705e-06,
      "loss": 0.4163,
      "num_tokens": 1840829994.0,
      "step": 1942
    },
    {
      "epoch": 3.8031341821743387,
      "grad_norm": 0.08121303740640588,
      "learning_rate": 1.4354513412378586e-06,
      "loss": 0.4123,
      "num_tokens": 1841761720.0,
      "step": 1943
    },
    {
      "epoch": 3.8050930460333006,
      "grad_norm": 0.08931443056419079,
      "learning_rate": 1.4310089687651302e-06,
      "loss": 0.4283,
      "num_tokens": 1842697162.0,
      "step": 1944
    },
    {
      "epoch": 3.8070519098922624,
      "grad_norm": 0.08514268990515005,
      "learning_rate": 1.4265723327335162e-06,
      "loss": 0.4261,
      "num_tokens": 1843667675.0,
      "step": 1945
    },
    {
      "epoch": 3.809010773751224,
      "grad_norm": 0.08903059963344348,
      "learning_rate": 1.422141440274023e-06,
      "loss": 0.4224,
      "num_tokens": 1844619081.0,
      "step": 1946
    },
    {
      "epoch": 3.810969637610186,
      "grad_norm": 0.08401625713340918,
      "learning_rate": 1.4177162985084242e-06,
      "loss": 0.425,
      "num_tokens": 1845560968.0,
      "step": 1947
    },
    {
      "epoch": 3.812928501469148,
      "grad_norm": 0.0868968883943069,
      "learning_rate": 1.4132969145492559e-06,
      "loss": 0.4054,
      "num_tokens": 1846504685.0,
      "step": 1948
    },
    {
      "epoch": 3.8148873653281097,
      "grad_norm": 0.08180514482927043,
      "learning_rate": 1.4088832954997944e-06,
      "loss": 0.3966,
      "num_tokens": 1847433339.0,
      "step": 1949
    },
    {
      "epoch": 3.8168462291870715,
      "grad_norm": 0.0856552429613838,
      "learning_rate": 1.40447544845405e-06,
      "loss": 0.4307,
      "num_tokens": 1848386888.0,
      "step": 1950
    },
    {
      "epoch": 3.8188050930460333,
      "grad_norm": 0.08634696578800498,
      "learning_rate": 1.4000733804967608e-06,
      "loss": 0.4104,
      "num_tokens": 1849327866.0,
      "step": 1951
    },
    {
      "epoch": 3.820763956904995,
      "grad_norm": 0.08414651487379428,
      "learning_rate": 1.3956770987033719e-06,
      "loss": 0.4137,
      "num_tokens": 1850271679.0,
      "step": 1952
    },
    {
      "epoch": 3.822722820763957,
      "grad_norm": 0.08419625942296331,
      "learning_rate": 1.3912866101400275e-06,
      "loss": 0.417,
      "num_tokens": 1851203767.0,
      "step": 1953
    },
    {
      "epoch": 3.824681684622919,
      "grad_norm": 0.08390891064909364,
      "learning_rate": 1.3869019218635644e-06,
      "loss": 0.417,
      "num_tokens": 1852158239.0,
      "step": 1954
    },
    {
      "epoch": 3.8266405484818806,
      "grad_norm": 0.08567211789364376,
      "learning_rate": 1.3825230409214935e-06,
      "loss": 0.4113,
      "num_tokens": 1853119345.0,
      "step": 1955
    },
    {
      "epoch": 3.8285994123408424,
      "grad_norm": 0.08187069477773441,
      "learning_rate": 1.3781499743519911e-06,
      "loss": 0.4149,
      "num_tokens": 1854064557.0,
      "step": 1956
    },
    {
      "epoch": 3.8305582761998043,
      "grad_norm": 0.08442954503231062,
      "learning_rate": 1.3737827291838917e-06,
      "loss": 0.4119,
      "num_tokens": 1855003717.0,
      "step": 1957
    },
    {
      "epoch": 3.8325171400587656,
      "grad_norm": 0.08178202625417966,
      "learning_rate": 1.3694213124366696e-06,
      "loss": 0.4068,
      "num_tokens": 1855937892.0,
      "step": 1958
    },
    {
      "epoch": 3.834476003917728,
      "grad_norm": 0.08601072930203271,
      "learning_rate": 1.3650657311204307e-06,
      "loss": 0.428,
      "num_tokens": 1856862078.0,
      "step": 1959
    },
    {
      "epoch": 3.8364348677766893,
      "grad_norm": 0.08980409225787742,
      "learning_rate": 1.3607159922359058e-06,
      "loss": 0.4161,
      "num_tokens": 1857814343.0,
      "step": 1960
    },
    {
      "epoch": 3.8383937316356516,
      "grad_norm": 0.0881534624814249,
      "learning_rate": 1.3563721027744309e-06,
      "loss": 0.4172,
      "num_tokens": 1858760105.0,
      "step": 1961
    },
    {
      "epoch": 3.840352595494613,
      "grad_norm": 0.08242921726026739,
      "learning_rate": 1.3520340697179406e-06,
      "loss": 0.4187,
      "num_tokens": 1859705486.0,
      "step": 1962
    },
    {
      "epoch": 3.8423114593535748,
      "grad_norm": 0.08460184858489149,
      "learning_rate": 1.3477019000389624e-06,
      "loss": 0.4083,
      "num_tokens": 1860626769.0,
      "step": 1963
    },
    {
      "epoch": 3.8442703232125366,
      "grad_norm": 0.08800797682634066,
      "learning_rate": 1.3433756007005877e-06,
      "loss": 0.4187,
      "num_tokens": 1861580538.0,
      "step": 1964
    },
    {
      "epoch": 3.8462291870714984,
      "grad_norm": 0.08286833733684613,
      "learning_rate": 1.3390551786564853e-06,
      "loss": 0.414,
      "num_tokens": 1862536042.0,
      "step": 1965
    },
    {
      "epoch": 3.8481880509304602,
      "grad_norm": 0.08442314687884295,
      "learning_rate": 1.3347406408508695e-06,
      "loss": 0.4187,
      "num_tokens": 1863496597.0,
      "step": 1966
    },
    {
      "epoch": 3.850146914789422,
      "grad_norm": 0.08348039787032789,
      "learning_rate": 1.3304319942184985e-06,
      "loss": 0.4196,
      "num_tokens": 1864411280.0,
      "step": 1967
    },
    {
      "epoch": 3.852105778648384,
      "grad_norm": 0.08363557741924119,
      "learning_rate": 1.3261292456846648e-06,
      "loss": 0.4096,
      "num_tokens": 1865331381.0,
      "step": 1968
    },
    {
      "epoch": 3.8540646425073457,
      "grad_norm": 0.08820754563153728,
      "learning_rate": 1.3218324021651763e-06,
      "loss": 0.4234,
      "num_tokens": 1866285303.0,
      "step": 1969
    },
    {
      "epoch": 3.8560235063663075,
      "grad_norm": 0.08275129689993459,
      "learning_rate": 1.3175414705663508e-06,
      "loss": 0.4012,
      "num_tokens": 1867241446.0,
      "step": 1970
    },
    {
      "epoch": 3.8579823702252694,
      "grad_norm": 0.08405493005731622,
      "learning_rate": 1.3132564577850076e-06,
      "loss": 0.4108,
      "num_tokens": 1868209880.0,
      "step": 1971
    },
    {
      "epoch": 3.859941234084231,
      "grad_norm": 0.09041932226250768,
      "learning_rate": 1.3089773707084492e-06,
      "loss": 0.4313,
      "num_tokens": 1869153236.0,
      "step": 1972
    },
    {
      "epoch": 3.861900097943193,
      "grad_norm": 0.08449472039851497,
      "learning_rate": 1.3047042162144518e-06,
      "loss": 0.4224,
      "num_tokens": 1870126698.0,
      "step": 1973
    },
    {
      "epoch": 3.863858961802155,
      "grad_norm": 0.08411268105576536,
      "learning_rate": 1.3004370011712624e-06,
      "loss": 0.4139,
      "num_tokens": 1871077323.0,
      "step": 1974
    },
    {
      "epoch": 3.8658178256611166,
      "grad_norm": 0.0827968826755833,
      "learning_rate": 1.2961757324375768e-06,
      "loss": 0.4078,
      "num_tokens": 1872038384.0,
      "step": 1975
    },
    {
      "epoch": 3.8677766895200785,
      "grad_norm": 0.0845075442611009,
      "learning_rate": 1.2919204168625322e-06,
      "loss": 0.434,
      "num_tokens": 1873051905.0,
      "step": 1976
    },
    {
      "epoch": 3.8697355533790403,
      "grad_norm": 0.08462865608635815,
      "learning_rate": 1.2876710612857029e-06,
      "loss": 0.4223,
      "num_tokens": 1873979415.0,
      "step": 1977
    },
    {
      "epoch": 3.871694417238002,
      "grad_norm": 0.08460278437532708,
      "learning_rate": 1.2834276725370787e-06,
      "loss": 0.4137,
      "num_tokens": 1874950456.0,
      "step": 1978
    },
    {
      "epoch": 3.8736532810969635,
      "grad_norm": 0.08238758783440217,
      "learning_rate": 1.2791902574370591e-06,
      "loss": 0.4033,
      "num_tokens": 1875891822.0,
      "step": 1979
    },
    {
      "epoch": 3.8756121449559258,
      "grad_norm": 0.08342045939138873,
      "learning_rate": 1.2749588227964466e-06,
      "loss": 0.4222,
      "num_tokens": 1876851239.0,
      "step": 1980
    },
    {
      "epoch": 3.877571008814887,
      "grad_norm": 0.08437609123813213,
      "learning_rate": 1.2707333754164264e-06,
      "loss": 0.4162,
      "num_tokens": 1877805057.0,
      "step": 1981
    },
    {
      "epoch": 3.8795298726738494,
      "grad_norm": 0.08583678185088015,
      "learning_rate": 1.2665139220885615e-06,
      "loss": 0.4154,
      "num_tokens": 1878744460.0,
      "step": 1982
    },
    {
      "epoch": 3.881488736532811,
      "grad_norm": 0.08687076318955812,
      "learning_rate": 1.2623004695947832e-06,
      "loss": 0.4144,
      "num_tokens": 1879691675.0,
      "step": 1983
    },
    {
      "epoch": 3.8834476003917726,
      "grad_norm": 0.08710363190478325,
      "learning_rate": 1.2580930247073752e-06,
      "loss": 0.4163,
      "num_tokens": 1880638800.0,
      "step": 1984
    },
    {
      "epoch": 3.8854064642507344,
      "grad_norm": 0.08231975046148307,
      "learning_rate": 1.2538915941889646e-06,
      "loss": 0.4251,
      "num_tokens": 1881590715.0,
      "step": 1985
    },
    {
      "epoch": 3.8873653281096963,
      "grad_norm": 0.08920842056600325,
      "learning_rate": 1.2496961847925154e-06,
      "loss": 0.4257,
      "num_tokens": 1882517573.0,
      "step": 1986
    },
    {
      "epoch": 3.889324191968658,
      "grad_norm": 0.08646749713217497,
      "learning_rate": 1.2455068032613077e-06,
      "loss": 0.4185,
      "num_tokens": 1883463626.0,
      "step": 1987
    },
    {
      "epoch": 3.89128305582762,
      "grad_norm": 0.08299093958222617,
      "learning_rate": 1.2413234563289406e-06,
      "loss": 0.4042,
      "num_tokens": 1884403020.0,
      "step": 1988
    },
    {
      "epoch": 3.8932419196865817,
      "grad_norm": 0.07815162636046777,
      "learning_rate": 1.2371461507193077e-06,
      "loss": 0.4208,
      "num_tokens": 1885354725.0,
      "step": 1989
    },
    {
      "epoch": 3.8952007835455436,
      "grad_norm": 0.08675301159428482,
      "learning_rate": 1.2329748931465945e-06,
      "loss": 0.4108,
      "num_tokens": 1886282535.0,
      "step": 1990
    },
    {
      "epoch": 3.8971596474045054,
      "grad_norm": 0.084792093805616,
      "learning_rate": 1.2288096903152652e-06,
      "loss": 0.4041,
      "num_tokens": 1887247894.0,
      "step": 1991
    },
    {
      "epoch": 3.899118511263467,
      "grad_norm": 0.08283453301958113,
      "learning_rate": 1.2246505489200534e-06,
      "loss": 0.4058,
      "num_tokens": 1888196825.0,
      "step": 1992
    },
    {
      "epoch": 3.901077375122429,
      "grad_norm": 0.08651976409073864,
      "learning_rate": 1.2204974756459465e-06,
      "loss": 0.4234,
      "num_tokens": 1889162984.0,
      "step": 1993
    },
    {
      "epoch": 3.903036238981391,
      "grad_norm": 0.0844438315601082,
      "learning_rate": 1.216350477168185e-06,
      "loss": 0.4203,
      "num_tokens": 1890103421.0,
      "step": 1994
    },
    {
      "epoch": 3.9049951028403527,
      "grad_norm": 0.08637547521440192,
      "learning_rate": 1.2122095601522404e-06,
      "loss": 0.419,
      "num_tokens": 1891081047.0,
      "step": 1995
    },
    {
      "epoch": 3.9069539666993145,
      "grad_norm": 0.08730163040571967,
      "learning_rate": 1.2080747312538082e-06,
      "loss": 0.4268,
      "num_tokens": 1891995891.0,
      "step": 1996
    },
    {
      "epoch": 3.9089128305582763,
      "grad_norm": 0.08548199716073168,
      "learning_rate": 1.203945997118805e-06,
      "loss": 0.4124,
      "num_tokens": 1892946143.0,
      "step": 1997
    },
    {
      "epoch": 3.910871694417238,
      "grad_norm": 0.08191612584010766,
      "learning_rate": 1.1998233643833457e-06,
      "loss": 0.4058,
      "num_tokens": 1893888836.0,
      "step": 1998
    },
    {
      "epoch": 3.9128305582762,
      "grad_norm": 0.08330786533493431,
      "learning_rate": 1.195706839673738e-06,
      "loss": 0.4065,
      "num_tokens": 1894847818.0,
      "step": 1999
    },
    {
      "epoch": 3.9147894221351613,
      "grad_norm": 0.0824831852880592,
      "learning_rate": 1.1915964296064775e-06,
      "loss": 0.3953,
      "num_tokens": 1895811476.0,
      "step": 2000
    },
    {
      "epoch": 3.9167482859941236,
      "grad_norm": 0.08122688319719282,
      "learning_rate": 1.187492140788225e-06,
      "loss": 0.4158,
      "num_tokens": 1896787472.0,
      "step": 2001
    },
    {
      "epoch": 3.918707149853085,
      "grad_norm": 0.08136197637442369,
      "learning_rate": 1.1833939798158095e-06,
      "loss": 0.4195,
      "num_tokens": 1897705197.0,
      "step": 2002
    },
    {
      "epoch": 3.9206660137120473,
      "grad_norm": 0.08546440048957472,
      "learning_rate": 1.1793019532762057e-06,
      "loss": 0.4258,
      "num_tokens": 1898637706.0,
      "step": 2003
    },
    {
      "epoch": 3.9226248775710086,
      "grad_norm": 0.08585210376684785,
      "learning_rate": 1.1752160677465286e-06,
      "loss": 0.4117,
      "num_tokens": 1899588114.0,
      "step": 2004
    },
    {
      "epoch": 3.9245837414299705,
      "grad_norm": 0.0835230568809389,
      "learning_rate": 1.1711363297940265e-06,
      "loss": 0.414,
      "num_tokens": 1900483221.0,
      "step": 2005
    },
    {
      "epoch": 3.9265426052889323,
      "grad_norm": 0.08529726758501822,
      "learning_rate": 1.1670627459760637e-06,
      "loss": 0.4259,
      "num_tokens": 1901419892.0,
      "step": 2006
    },
    {
      "epoch": 3.928501469147894,
      "grad_norm": 0.09140885115656842,
      "learning_rate": 1.1629953228401109e-06,
      "loss": 0.4225,
      "num_tokens": 1902376792.0,
      "step": 2007
    },
    {
      "epoch": 3.930460333006856,
      "grad_norm": 0.08248469579214886,
      "learning_rate": 1.1589340669237432e-06,
      "loss": 0.4238,
      "num_tokens": 1903326764.0,
      "step": 2008
    },
    {
      "epoch": 3.9324191968658178,
      "grad_norm": 0.08597793447176653,
      "learning_rate": 1.1548789847546178e-06,
      "loss": 0.4076,
      "num_tokens": 1904290512.0,
      "step": 2009
    },
    {
      "epoch": 3.9343780607247796,
      "grad_norm": 0.08251527638436337,
      "learning_rate": 1.1508300828504682e-06,
      "loss": 0.4131,
      "num_tokens": 1905236568.0,
      "step": 2010
    },
    {
      "epoch": 3.9363369245837414,
      "grad_norm": 0.08440021634504866,
      "learning_rate": 1.1467873677191005e-06,
      "loss": 0.4179,
      "num_tokens": 1906158419.0,
      "step": 2011
    },
    {
      "epoch": 3.9382957884427032,
      "grad_norm": 0.08234066151314694,
      "learning_rate": 1.1427508458583703e-06,
      "loss": 0.4048,
      "num_tokens": 1907064319.0,
      "step": 2012
    },
    {
      "epoch": 3.940254652301665,
      "grad_norm": 0.0829201170387118,
      "learning_rate": 1.1387205237561804e-06,
      "loss": 0.4064,
      "num_tokens": 1908029968.0,
      "step": 2013
    },
    {
      "epoch": 3.942213516160627,
      "grad_norm": 0.08394438675727865,
      "learning_rate": 1.1346964078904727e-06,
      "loss": 0.4117,
      "num_tokens": 1908945871.0,
      "step": 2014
    },
    {
      "epoch": 3.9441723800195887,
      "grad_norm": 0.08613624973812307,
      "learning_rate": 1.1306785047292085e-06,
      "loss": 0.4318,
      "num_tokens": 1909891675.0,
      "step": 2015
    },
    {
      "epoch": 3.9461312438785505,
      "grad_norm": 0.08654788565549938,
      "learning_rate": 1.126666820730366e-06,
      "loss": 0.4198,
      "num_tokens": 1910827674.0,
      "step": 2016
    },
    {
      "epoch": 3.9480901077375123,
      "grad_norm": 0.08265006031279491,
      "learning_rate": 1.122661362341927e-06,
      "loss": 0.415,
      "num_tokens": 1911775363.0,
      "step": 2017
    },
    {
      "epoch": 3.950048971596474,
      "grad_norm": 0.08506299568498547,
      "learning_rate": 1.1186621360018656e-06,
      "loss": 0.416,
      "num_tokens": 1912745495.0,
      "step": 2018
    },
    {
      "epoch": 3.952007835455436,
      "grad_norm": 0.0862167601754671,
      "learning_rate": 1.1146691481381428e-06,
      "loss": 0.4165,
      "num_tokens": 1913679035.0,
      "step": 2019
    },
    {
      "epoch": 3.953966699314398,
      "grad_norm": 0.08483214721743973,
      "learning_rate": 1.1106824051686888e-06,
      "loss": 0.4114,
      "num_tokens": 1914583470.0,
      "step": 2020
    },
    {
      "epoch": 3.955925563173359,
      "grad_norm": 0.0844595520583096,
      "learning_rate": 1.1067019135013956e-06,
      "loss": 0.4318,
      "num_tokens": 1915529721.0,
      "step": 2021
    },
    {
      "epoch": 3.9578844270323215,
      "grad_norm": 0.08529514135489234,
      "learning_rate": 1.1027276795341135e-06,
      "loss": 0.4147,
      "num_tokens": 1916478178.0,
      "step": 2022
    },
    {
      "epoch": 3.959843290891283,
      "grad_norm": 0.08299401156766298,
      "learning_rate": 1.0987597096546288e-06,
      "loss": 0.4062,
      "num_tokens": 1917430861.0,
      "step": 2023
    },
    {
      "epoch": 3.961802154750245,
      "grad_norm": 0.08306066471753731,
      "learning_rate": 1.0947980102406597e-06,
      "loss": 0.4154,
      "num_tokens": 1918405975.0,
      "step": 2024
    },
    {
      "epoch": 3.9637610186092065,
      "grad_norm": 0.08160808801541311,
      "learning_rate": 1.0908425876598512e-06,
      "loss": 0.4112,
      "num_tokens": 1919405602.0,
      "step": 2025
    },
    {
      "epoch": 3.9657198824681683,
      "grad_norm": 0.08148928490862165,
      "learning_rate": 1.0868934482697547e-06,
      "loss": 0.4079,
      "num_tokens": 1920351831.0,
      "step": 2026
    },
    {
      "epoch": 3.96767874632713,
      "grad_norm": 0.0838535117640875,
      "learning_rate": 1.0829505984178212e-06,
      "loss": 0.4107,
      "num_tokens": 1921300601.0,
      "step": 2027
    },
    {
      "epoch": 3.969637610186092,
      "grad_norm": 0.08061820241323737,
      "learning_rate": 1.0790140444414e-06,
      "loss": 0.4098,
      "num_tokens": 1922253262.0,
      "step": 2028
    },
    {
      "epoch": 3.971596474045054,
      "grad_norm": 0.08164752940757682,
      "learning_rate": 1.0750837926677133e-06,
      "loss": 0.4127,
      "num_tokens": 1923196761.0,
      "step": 2029
    },
    {
      "epoch": 3.9735553379040156,
      "grad_norm": 0.08457874528495596,
      "learning_rate": 1.0711598494138564e-06,
      "loss": 0.4289,
      "num_tokens": 1924139440.0,
      "step": 2030
    },
    {
      "epoch": 3.9755142017629774,
      "grad_norm": 0.08340730751985082,
      "learning_rate": 1.0672422209867879e-06,
      "loss": 0.4147,
      "num_tokens": 1925118595.0,
      "step": 2031
    },
    {
      "epoch": 3.9774730656219393,
      "grad_norm": 0.08543135447738082,
      "learning_rate": 1.063330913683312e-06,
      "loss": 0.4201,
      "num_tokens": 1926054293.0,
      "step": 2032
    },
    {
      "epoch": 3.979431929480901,
      "grad_norm": 0.08344031544175283,
      "learning_rate": 1.0594259337900748e-06,
      "loss": 0.4042,
      "num_tokens": 1927014860.0,
      "step": 2033
    },
    {
      "epoch": 3.981390793339863,
      "grad_norm": 0.08259028738825627,
      "learning_rate": 1.0555272875835537e-06,
      "loss": 0.4175,
      "num_tokens": 1927959244.0,
      "step": 2034
    },
    {
      "epoch": 3.9833496571988247,
      "grad_norm": 0.08450279727167037,
      "learning_rate": 1.0516349813300447e-06,
      "loss": 0.4056,
      "num_tokens": 1928904971.0,
      "step": 2035
    },
    {
      "epoch": 3.9853085210577865,
      "grad_norm": 0.08607922460115165,
      "learning_rate": 1.047749021285651e-06,
      "loss": 0.4151,
      "num_tokens": 1929856382.0,
      "step": 2036
    },
    {
      "epoch": 3.9872673849167484,
      "grad_norm": 0.08731108771114565,
      "learning_rate": 1.0438694136962824e-06,
      "loss": 0.4262,
      "num_tokens": 1930766918.0,
      "step": 2037
    },
    {
      "epoch": 3.98922624877571,
      "grad_norm": 0.08561238047874237,
      "learning_rate": 1.0399961647976315e-06,
      "loss": 0.4198,
      "num_tokens": 1931709023.0,
      "step": 2038
    },
    {
      "epoch": 3.991185112634672,
      "grad_norm": 0.0840692465053406,
      "learning_rate": 1.0361292808151725e-06,
      "loss": 0.406,
      "num_tokens": 1932656941.0,
      "step": 2039
    },
    {
      "epoch": 3.993143976493634,
      "grad_norm": 0.08523835969256632,
      "learning_rate": 1.0322687679641523e-06,
      "loss": 0.4336,
      "num_tokens": 1933632222.0,
      "step": 2040
    },
    {
      "epoch": 3.9951028403525957,
      "grad_norm": 0.08560297472644865,
      "learning_rate": 1.0284146324495747e-06,
      "loss": 0.4214,
      "num_tokens": 1934570746.0,
      "step": 2041
    },
    {
      "epoch": 3.997061704211557,
      "grad_norm": 0.08318935314236181,
      "learning_rate": 1.024566880466193e-06,
      "loss": 0.4269,
      "num_tokens": 1935528915.0,
      "step": 2042
    },
    {
      "epoch": 3.9990205680705193,
      "grad_norm": 0.08849211155747533,
      "learning_rate": 1.020725518198501e-06,
      "loss": 0.4303,
      "num_tokens": 1936453215.0,
      "step": 2043
    },
    {
      "epoch": 4.0,
      "grad_norm": 0.1312278322783195,
      "learning_rate": 1.0168905518207206e-06,
      "loss": 0.4084,
      "num_tokens": 1936926192.0,
      "step": 2044
    },
    {
      "epoch": 4.001958863858961,
      "grad_norm": 0.08502193677350244,
      "learning_rate": 1.0130619874967983e-06,
      "loss": 0.4221,
      "num_tokens": 1937887261.0,
      "step": 2045
    },
    {
      "epoch": 4.003917727717924,
      "grad_norm": 0.08475491122092788,
      "learning_rate": 1.0092398313803864e-06,
      "loss": 0.4101,
      "num_tokens": 1938852537.0,
      "step": 2046
    },
    {
      "epoch": 4.005876591576885,
      "grad_norm": 0.08689762084378919,
      "learning_rate": 1.0054240896148364e-06,
      "loss": 0.4083,
      "num_tokens": 1939788905.0,
      "step": 2047
    },
    {
      "epoch": 4.007835455435847,
      "grad_norm": 0.08260209940359746,
      "learning_rate": 1.0016147683331955e-06,
      "loss": 0.4237,
      "num_tokens": 1940726418.0,
      "step": 2048
    },
    {
      "epoch": 4.009794319294809,
      "grad_norm": 0.08399095900727253,
      "learning_rate": 9.97811873658186e-07,
      "loss": 0.4279,
      "num_tokens": 1941663266.0,
      "step": 2049
    },
    {
      "epoch": 4.011753183153771,
      "grad_norm": 0.08421254211310467,
      "learning_rate": 9.940154117022005e-07,
      "loss": 0.4148,
      "num_tokens": 1942599604.0,
      "step": 2050
    },
    {
      "epoch": 4.013712047012732,
      "grad_norm": 0.08420527051018095,
      "learning_rate": 9.90225388567298e-07,
      "loss": 0.4098,
      "num_tokens": 1943527663.0,
      "step": 2051
    },
    {
      "epoch": 4.015670910871695,
      "grad_norm": 0.08317858749531441,
      "learning_rate": 9.86441810345183e-07,
      "loss": 0.4097,
      "num_tokens": 1944461056.0,
      "step": 2052
    },
    {
      "epoch": 4.017629774730656,
      "grad_norm": 0.08725770282762309,
      "learning_rate": 9.826646831172004e-07,
      "loss": 0.4267,
      "num_tokens": 1945386344.0,
      "step": 2053
    },
    {
      "epoch": 4.019588638589618,
      "grad_norm": 0.08222310778997291,
      "learning_rate": 9.788940129543323e-07,
      "loss": 0.4174,
      "num_tokens": 1946354473.0,
      "step": 2054
    },
    {
      "epoch": 4.02154750244858,
      "grad_norm": 0.08337402373949127,
      "learning_rate": 9.751298059171777e-07,
      "loss": 0.4104,
      "num_tokens": 1947324057.0,
      "step": 2055
    },
    {
      "epoch": 4.023506366307542,
      "grad_norm": 0.08333568851478758,
      "learning_rate": 9.713720680559463e-07,
      "loss": 0.3997,
      "num_tokens": 1948282104.0,
      "step": 2056
    },
    {
      "epoch": 4.025465230166503,
      "grad_norm": 0.08270716784118884,
      "learning_rate": 9.676208054104553e-07,
      "loss": 0.4138,
      "num_tokens": 1949279239.0,
      "step": 2057
    },
    {
      "epoch": 4.0274240940254655,
      "grad_norm": 0.08202636274056062,
      "learning_rate": 9.638760240101102e-07,
      "loss": 0.4014,
      "num_tokens": 1950240230.0,
      "step": 2058
    },
    {
      "epoch": 4.029382957884427,
      "grad_norm": 0.08458357494517589,
      "learning_rate": 9.60137729873898e-07,
      "loss": 0.4011,
      "num_tokens": 1951192910.0,
      "step": 2059
    },
    {
      "epoch": 4.031341821743389,
      "grad_norm": 0.08314252647160143,
      "learning_rate": 9.564059290103832e-07,
      "loss": 0.4162,
      "num_tokens": 1952150853.0,
      "step": 2060
    },
    {
      "epoch": 4.0333006856023506,
      "grad_norm": 0.08943419961461954,
      "learning_rate": 9.526806274176903e-07,
      "loss": 0.4166,
      "num_tokens": 1953085933.0,
      "step": 2061
    },
    {
      "epoch": 4.035259549461313,
      "grad_norm": 0.08656132443725874,
      "learning_rate": 9.489618310834975e-07,
      "loss": 0.4071,
      "num_tokens": 1954020911.0,
      "step": 2062
    },
    {
      "epoch": 4.037218413320274,
      "grad_norm": 0.08238655532832571,
      "learning_rate": 9.452495459850303e-07,
      "loss": 0.4095,
      "num_tokens": 1954980124.0,
      "step": 2063
    },
    {
      "epoch": 4.0391772771792365,
      "grad_norm": 0.08482853005158912,
      "learning_rate": 9.415437780890452e-07,
      "loss": 0.4153,
      "num_tokens": 1955937348.0,
      "step": 2064
    },
    {
      "epoch": 4.041136141038198,
      "grad_norm": 0.08503822379426298,
      "learning_rate": 9.378445333518238e-07,
      "loss": 0.408,
      "num_tokens": 1956877866.0,
      "step": 2065
    },
    {
      "epoch": 4.043095004897159,
      "grad_norm": 0.08649059908764865,
      "learning_rate": 9.34151817719166e-07,
      "loss": 0.401,
      "num_tokens": 1957796291.0,
      "step": 2066
    },
    {
      "epoch": 4.0450538687561215,
      "grad_norm": 0.08540282899796471,
      "learning_rate": 9.304656371263749e-07,
      "loss": 0.4129,
      "num_tokens": 1958743695.0,
      "step": 2067
    },
    {
      "epoch": 4.047012732615083,
      "grad_norm": 0.08589684990372319,
      "learning_rate": 9.26785997498249e-07,
      "loss": 0.4292,
      "num_tokens": 1959686627.0,
      "step": 2068
    },
    {
      "epoch": 4.048971596474045,
      "grad_norm": 0.08169380819766643,
      "learning_rate": 9.231129047490784e-07,
      "loss": 0.4122,
      "num_tokens": 1960669441.0,
      "step": 2069
    },
    {
      "epoch": 4.0509304603330065,
      "grad_norm": 0.08231001210924904,
      "learning_rate": 9.194463647826224e-07,
      "loss": 0.4121,
      "num_tokens": 1961638587.0,
      "step": 2070
    },
    {
      "epoch": 4.052889324191969,
      "grad_norm": 0.08227276084835788,
      "learning_rate": 9.157863834921155e-07,
      "loss": 0.4116,
      "num_tokens": 1962579088.0,
      "step": 2071
    },
    {
      "epoch": 4.05484818805093,
      "grad_norm": 0.08299856022370741,
      "learning_rate": 9.121329667602474e-07,
      "loss": 0.3987,
      "num_tokens": 1963493283.0,
      "step": 2072
    },
    {
      "epoch": 4.056807051909892,
      "grad_norm": 0.08386846753592783,
      "learning_rate": 9.08486120459155e-07,
      "loss": 0.417,
      "num_tokens": 1964424620.0,
      "step": 2073
    },
    {
      "epoch": 4.058765915768854,
      "grad_norm": 0.08497730471137754,
      "learning_rate": 9.048458504504181e-07,
      "loss": 0.4238,
      "num_tokens": 1965374010.0,
      "step": 2074
    },
    {
      "epoch": 4.060724779627816,
      "grad_norm": 0.08193153364959628,
      "learning_rate": 9.012121625850445e-07,
      "loss": 0.4058,
      "num_tokens": 1966341859.0,
      "step": 2075
    },
    {
      "epoch": 4.0626836434867775,
      "grad_norm": 0.08406692485203614,
      "learning_rate": 8.975850627034605e-07,
      "loss": 0.4098,
      "num_tokens": 1967288695.0,
      "step": 2076
    },
    {
      "epoch": 4.06464250734574,
      "grad_norm": 0.08458426120522519,
      "learning_rate": 8.939645566355082e-07,
      "loss": 0.4097,
      "num_tokens": 1968206492.0,
      "step": 2077
    },
    {
      "epoch": 4.066601371204701,
      "grad_norm": 0.08268780425945656,
      "learning_rate": 8.903506502004272e-07,
      "loss": 0.4073,
      "num_tokens": 1969156152.0,
      "step": 2078
    },
    {
      "epoch": 4.068560235063663,
      "grad_norm": 0.08509860396084373,
      "learning_rate": 8.867433492068506e-07,
      "loss": 0.4178,
      "num_tokens": 1970076813.0,
      "step": 2079
    },
    {
      "epoch": 4.070519098922625,
      "grad_norm": 0.08410925852511572,
      "learning_rate": 8.831426594527976e-07,
      "loss": 0.4091,
      "num_tokens": 1971000534.0,
      "step": 2080
    },
    {
      "epoch": 4.072477962781587,
      "grad_norm": 0.08214244051387433,
      "learning_rate": 8.79548586725657e-07,
      "loss": 0.4086,
      "num_tokens": 1971963735.0,
      "step": 2081
    },
    {
      "epoch": 4.074436826640548,
      "grad_norm": 0.08320219656143128,
      "learning_rate": 8.759611368021831e-07,
      "loss": 0.412,
      "num_tokens": 1972949208.0,
      "step": 2082
    },
    {
      "epoch": 4.076395690499511,
      "grad_norm": 0.08087905369433049,
      "learning_rate": 8.723803154484878e-07,
      "loss": 0.4048,
      "num_tokens": 1973900628.0,
      "step": 2083
    },
    {
      "epoch": 4.078354554358472,
      "grad_norm": 0.0839633275807735,
      "learning_rate": 8.688061284200266e-07,
      "loss": 0.4165,
      "num_tokens": 1974821386.0,
      "step": 2084
    },
    {
      "epoch": 4.080313418217434,
      "grad_norm": 0.08577869269252363,
      "learning_rate": 8.652385814615899e-07,
      "loss": 0.4085,
      "num_tokens": 1975746775.0,
      "step": 2085
    },
    {
      "epoch": 4.082272282076396,
      "grad_norm": 0.0858937256665813,
      "learning_rate": 8.616776803073013e-07,
      "loss": 0.4227,
      "num_tokens": 1976704407.0,
      "step": 2086
    },
    {
      "epoch": 4.084231145935357,
      "grad_norm": 0.08591592861928278,
      "learning_rate": 8.581234306805969e-07,
      "loss": 0.4153,
      "num_tokens": 1977661539.0,
      "step": 2087
    },
    {
      "epoch": 4.086190009794319,
      "grad_norm": 0.08264970198212282,
      "learning_rate": 8.545758382942232e-07,
      "loss": 0.4205,
      "num_tokens": 1978625050.0,
      "step": 2088
    },
    {
      "epoch": 4.088148873653281,
      "grad_norm": 0.08333784229963259,
      "learning_rate": 8.510349088502295e-07,
      "loss": 0.4277,
      "num_tokens": 1979620090.0,
      "step": 2089
    },
    {
      "epoch": 4.090107737512243,
      "grad_norm": 0.08789058389774676,
      "learning_rate": 8.475006480399528e-07,
      "loss": 0.4268,
      "num_tokens": 1980562137.0,
      "step": 2090
    },
    {
      "epoch": 4.092066601371204,
      "grad_norm": 0.0873505236627856,
      "learning_rate": 8.439730615440106e-07,
      "loss": 0.4108,
      "num_tokens": 1981474359.0,
      "step": 2091
    },
    {
      "epoch": 4.094025465230167,
      "grad_norm": 0.08388104843650347,
      "learning_rate": 8.404521550322975e-07,
      "loss": 0.4236,
      "num_tokens": 1982430028.0,
      "step": 2092
    },
    {
      "epoch": 4.095984329089128,
      "grad_norm": 0.08239524474820561,
      "learning_rate": 8.36937934163966e-07,
      "loss": 0.4132,
      "num_tokens": 1983395209.0,
      "step": 2093
    },
    {
      "epoch": 4.09794319294809,
      "grad_norm": 0.08362559051183868,
      "learning_rate": 8.334304045874248e-07,
      "loss": 0.4031,
      "num_tokens": 1984310413.0,
      "step": 2094
    },
    {
      "epoch": 4.099902056807052,
      "grad_norm": 0.08625867872649809,
      "learning_rate": 8.299295719403311e-07,
      "loss": 0.424,
      "num_tokens": 1985247194.0,
      "step": 2095
    },
    {
      "epoch": 4.101860920666014,
      "grad_norm": 0.08665866705409384,
      "learning_rate": 8.264354418495707e-07,
      "loss": 0.4172,
      "num_tokens": 1986214184.0,
      "step": 2096
    },
    {
      "epoch": 4.103819784524975,
      "grad_norm": 0.08319286746801058,
      "learning_rate": 8.229480199312645e-07,
      "loss": 0.4133,
      "num_tokens": 1987152093.0,
      "step": 2097
    },
    {
      "epoch": 4.105778648383938,
      "grad_norm": 0.08289013502815501,
      "learning_rate": 8.194673117907453e-07,
      "loss": 0.4112,
      "num_tokens": 1988115357.0,
      "step": 2098
    },
    {
      "epoch": 4.107737512242899,
      "grad_norm": 0.08516964773454462,
      "learning_rate": 8.159933230225564e-07,
      "loss": 0.425,
      "num_tokens": 1989081236.0,
      "step": 2099
    },
    {
      "epoch": 4.109696376101861,
      "grad_norm": 0.08260738635978855,
      "learning_rate": 8.125260592104445e-07,
      "loss": 0.4007,
      "num_tokens": 1990018687.0,
      "step": 2100
    },
    {
      "epoch": 4.111655239960823,
      "grad_norm": 0.08321095732955454,
      "learning_rate": 8.090655259273428e-07,
      "loss": 0.4162,
      "num_tokens": 1990965805.0,
      "step": 2101
    },
    {
      "epoch": 4.113614103819785,
      "grad_norm": 0.08618368115956755,
      "learning_rate": 8.05611728735367e-07,
      "loss": 0.4324,
      "num_tokens": 1991956498.0,
      "step": 2102
    },
    {
      "epoch": 4.115572967678746,
      "grad_norm": 0.08264899564453558,
      "learning_rate": 8.021646731858101e-07,
      "loss": 0.4065,
      "num_tokens": 1992912450.0,
      "step": 2103
    },
    {
      "epoch": 4.1175318315377085,
      "grad_norm": 0.08288342381673579,
      "learning_rate": 7.987243648191245e-07,
      "loss": 0.4015,
      "num_tokens": 1993820685.0,
      "step": 2104
    },
    {
      "epoch": 4.11949069539667,
      "grad_norm": 0.08393247390987402,
      "learning_rate": 7.952908091649186e-07,
      "loss": 0.4063,
      "num_tokens": 1994778014.0,
      "step": 2105
    },
    {
      "epoch": 4.121449559255632,
      "grad_norm": 0.08110493139268171,
      "learning_rate": 7.918640117419507e-07,
      "loss": 0.4023,
      "num_tokens": 1995745587.0,
      "step": 2106
    },
    {
      "epoch": 4.1234084231145935,
      "grad_norm": 0.08074313592907365,
      "learning_rate": 7.884439780581127e-07,
      "loss": 0.3961,
      "num_tokens": 1996736368.0,
      "step": 2107
    },
    {
      "epoch": 4.125367286973555,
      "grad_norm": 0.08225551126841155,
      "learning_rate": 7.850307136104246e-07,
      "loss": 0.4151,
      "num_tokens": 1997718180.0,
      "step": 2108
    },
    {
      "epoch": 4.127326150832517,
      "grad_norm": 0.08182874299364075,
      "learning_rate": 7.816242238850319e-07,
      "loss": 0.4091,
      "num_tokens": 1998709445.0,
      "step": 2109
    },
    {
      "epoch": 4.129285014691479,
      "grad_norm": 0.08526759626891284,
      "learning_rate": 7.782245143571848e-07,
      "loss": 0.4189,
      "num_tokens": 1999647519.0,
      "step": 2110
    },
    {
      "epoch": 4.131243878550441,
      "grad_norm": 0.08521423913269954,
      "learning_rate": 7.748315904912379e-07,
      "loss": 0.4089,
      "num_tokens": 2000565274.0,
      "step": 2111
    },
    {
      "epoch": 4.133202742409402,
      "grad_norm": 0.08322162936888877,
      "learning_rate": 7.71445457740641e-07,
      "loss": 0.4125,
      "num_tokens": 2001500032.0,
      "step": 2112
    },
    {
      "epoch": 4.1351616062683645,
      "grad_norm": 0.08482170959481423,
      "learning_rate": 7.680661215479268e-07,
      "loss": 0.4095,
      "num_tokens": 2002437137.0,
      "step": 2113
    },
    {
      "epoch": 4.137120470127326,
      "grad_norm": 0.08342381084957223,
      "learning_rate": 7.64693587344702e-07,
      "loss": 0.4075,
      "num_tokens": 2003378228.0,
      "step": 2114
    },
    {
      "epoch": 4.139079333986288,
      "grad_norm": 0.08461529304023764,
      "learning_rate": 7.613278605516455e-07,
      "loss": 0.4145,
      "num_tokens": 2004289444.0,
      "step": 2115
    },
    {
      "epoch": 4.1410381978452495,
      "grad_norm": 0.08102920812569286,
      "learning_rate": 7.57968946578489e-07,
      "loss": 0.4157,
      "num_tokens": 2005277377.0,
      "step": 2116
    },
    {
      "epoch": 4.142997061704212,
      "grad_norm": 0.08494344523553249,
      "learning_rate": 7.546168508240192e-07,
      "loss": 0.4159,
      "num_tokens": 2006212370.0,
      "step": 2117
    },
    {
      "epoch": 4.144955925563173,
      "grad_norm": 0.08396960848853971,
      "learning_rate": 7.512715786760604e-07,
      "loss": 0.425,
      "num_tokens": 2007166347.0,
      "step": 2118
    },
    {
      "epoch": 4.146914789422135,
      "grad_norm": 0.08594146064778084,
      "learning_rate": 7.479331355114682e-07,
      "loss": 0.4105,
      "num_tokens": 2008098904.0,
      "step": 2119
    },
    {
      "epoch": 4.148873653281097,
      "grad_norm": 0.08450597034766295,
      "learning_rate": 7.446015266961276e-07,
      "loss": 0.412,
      "num_tokens": 2009022591.0,
      "step": 2120
    },
    {
      "epoch": 4.150832517140059,
      "grad_norm": 0.08474797510859908,
      "learning_rate": 7.412767575849328e-07,
      "loss": 0.4148,
      "num_tokens": 2009954740.0,
      "step": 2121
    },
    {
      "epoch": 4.1527913809990205,
      "grad_norm": 0.08188769206637259,
      "learning_rate": 7.379588335217875e-07,
      "loss": 0.411,
      "num_tokens": 2010913111.0,
      "step": 2122
    },
    {
      "epoch": 4.154750244857983,
      "grad_norm": 0.08255063659786951,
      "learning_rate": 7.346477598395929e-07,
      "loss": 0.4195,
      "num_tokens": 2011914672.0,
      "step": 2123
    },
    {
      "epoch": 4.156709108716944,
      "grad_norm": 0.08454163666818436,
      "learning_rate": 7.313435418602388e-07,
      "loss": 0.4105,
      "num_tokens": 2012852391.0,
      "step": 2124
    },
    {
      "epoch": 4.158667972575906,
      "grad_norm": 0.08416398576314223,
      "learning_rate": 7.280461848945957e-07,
      "loss": 0.4219,
      "num_tokens": 2013781590.0,
      "step": 2125
    },
    {
      "epoch": 4.160626836434868,
      "grad_norm": 0.08449507440365876,
      "learning_rate": 7.247556942425094e-07,
      "loss": 0.4053,
      "num_tokens": 2014724385.0,
      "step": 2126
    },
    {
      "epoch": 4.16258570029383,
      "grad_norm": 0.08487978542152061,
      "learning_rate": 7.214720751927862e-07,
      "loss": 0.4163,
      "num_tokens": 2015687056.0,
      "step": 2127
    },
    {
      "epoch": 4.164544564152791,
      "grad_norm": 0.08553567224887644,
      "learning_rate": 7.181953330231878e-07,
      "loss": 0.4099,
      "num_tokens": 2016620091.0,
      "step": 2128
    },
    {
      "epoch": 4.166503428011753,
      "grad_norm": 0.0853119108817595,
      "learning_rate": 7.149254730004246e-07,
      "loss": 0.4157,
      "num_tokens": 2017554377.0,
      "step": 2129
    },
    {
      "epoch": 4.168462291870715,
      "grad_norm": 0.08093729368936112,
      "learning_rate": 7.116625003801436e-07,
      "loss": 0.3992,
      "num_tokens": 2018515914.0,
      "step": 2130
    },
    {
      "epoch": 4.170421155729676,
      "grad_norm": 0.08568561410916128,
      "learning_rate": 7.084064204069236e-07,
      "loss": 0.396,
      "num_tokens": 2019453253.0,
      "step": 2131
    },
    {
      "epoch": 4.172380019588639,
      "grad_norm": 0.0849945814766733,
      "learning_rate": 7.05157238314263e-07,
      "loss": 0.4296,
      "num_tokens": 2020398435.0,
      "step": 2132
    },
    {
      "epoch": 4.1743388834476,
      "grad_norm": 0.08783575006158219,
      "learning_rate": 7.019149593245722e-07,
      "loss": 0.4089,
      "num_tokens": 2021307637.0,
      "step": 2133
    },
    {
      "epoch": 4.176297747306562,
      "grad_norm": 0.0843997620643785,
      "learning_rate": 6.986795886491704e-07,
      "loss": 0.4212,
      "num_tokens": 2022262326.0,
      "step": 2134
    },
    {
      "epoch": 4.178256611165524,
      "grad_norm": 0.08199608429243559,
      "learning_rate": 6.954511314882695e-07,
      "loss": 0.4037,
      "num_tokens": 2023222076.0,
      "step": 2135
    },
    {
      "epoch": 4.180215475024486,
      "grad_norm": 0.08402598227606085,
      "learning_rate": 6.922295930309691e-07,
      "loss": 0.4227,
      "num_tokens": 2024171135.0,
      "step": 2136
    },
    {
      "epoch": 4.182174338883447,
      "grad_norm": 0.08157805519585351,
      "learning_rate": 6.890149784552519e-07,
      "loss": 0.4116,
      "num_tokens": 2025145287.0,
      "step": 2137
    },
    {
      "epoch": 4.18413320274241,
      "grad_norm": 0.08690398372361305,
      "learning_rate": 6.858072929279691e-07,
      "loss": 0.429,
      "num_tokens": 2026067710.0,
      "step": 2138
    },
    {
      "epoch": 4.186092066601371,
      "grad_norm": 0.08265160839080408,
      "learning_rate": 6.826065416048327e-07,
      "loss": 0.4135,
      "num_tokens": 2027026586.0,
      "step": 2139
    },
    {
      "epoch": 4.188050930460333,
      "grad_norm": 0.08346277029597797,
      "learning_rate": 6.794127296304154e-07,
      "loss": 0.4175,
      "num_tokens": 2028018740.0,
      "step": 2140
    },
    {
      "epoch": 4.190009794319295,
      "grad_norm": 0.0834534109184271,
      "learning_rate": 6.762258621381313e-07,
      "loss": 0.4191,
      "num_tokens": 2028994550.0,
      "step": 2141
    },
    {
      "epoch": 4.191968658178257,
      "grad_norm": 0.08465385376506786,
      "learning_rate": 6.73045944250233e-07,
      "loss": 0.4133,
      "num_tokens": 2029923918.0,
      "step": 2142
    },
    {
      "epoch": 4.193927522037218,
      "grad_norm": 0.08619047456406939,
      "learning_rate": 6.698729810778065e-07,
      "loss": 0.4306,
      "num_tokens": 2030894928.0,
      "step": 2143
    },
    {
      "epoch": 4.195886385896181,
      "grad_norm": 0.08419196548036316,
      "learning_rate": 6.667069777207563e-07,
      "loss": 0.41,
      "num_tokens": 2031844940.0,
      "step": 2144
    },
    {
      "epoch": 4.197845249755142,
      "grad_norm": 0.08361444455541482,
      "learning_rate": 6.63547939267799e-07,
      "loss": 0.4019,
      "num_tokens": 2032801382.0,
      "step": 2145
    },
    {
      "epoch": 4.199804113614104,
      "grad_norm": 0.08317461888285774,
      "learning_rate": 6.603958707964626e-07,
      "loss": 0.4127,
      "num_tokens": 2033772314.0,
      "step": 2146
    },
    {
      "epoch": 4.201762977473066,
      "grad_norm": 0.0848943280652552,
      "learning_rate": 6.572507773730657e-07,
      "loss": 0.412,
      "num_tokens": 2034723023.0,
      "step": 2147
    },
    {
      "epoch": 4.203721841332028,
      "grad_norm": 0.08437657318695536,
      "learning_rate": 6.541126640527195e-07,
      "loss": 0.4054,
      "num_tokens": 2035677290.0,
      "step": 2148
    },
    {
      "epoch": 4.205680705190989,
      "grad_norm": 0.08478795354565644,
      "learning_rate": 6.509815358793148e-07,
      "loss": 0.4207,
      "num_tokens": 2036642296.0,
      "step": 2149
    },
    {
      "epoch": 4.207639569049951,
      "grad_norm": 0.08779979141689355,
      "learning_rate": 6.478573978855146e-07,
      "loss": 0.4113,
      "num_tokens": 2037548272.0,
      "step": 2150
    },
    {
      "epoch": 4.209598432908913,
      "grad_norm": 0.08180382264066516,
      "learning_rate": 6.447402550927495e-07,
      "loss": 0.415,
      "num_tokens": 2038495772.0,
      "step": 2151
    },
    {
      "epoch": 4.211557296767874,
      "grad_norm": 0.08371757404507543,
      "learning_rate": 6.416301125112034e-07,
      "loss": 0.4111,
      "num_tokens": 2039428219.0,
      "step": 2152
    },
    {
      "epoch": 4.2135161606268365,
      "grad_norm": 0.08185201788840223,
      "learning_rate": 6.385269751398099e-07,
      "loss": 0.4062,
      "num_tokens": 2040362205.0,
      "step": 2153
    },
    {
      "epoch": 4.215475024485798,
      "grad_norm": 0.08487072656465926,
      "learning_rate": 6.354308479662447e-07,
      "loss": 0.4075,
      "num_tokens": 2041311062.0,
      "step": 2154
    },
    {
      "epoch": 4.21743388834476,
      "grad_norm": 0.08625274537260617,
      "learning_rate": 6.323417359669132e-07,
      "loss": 0.4073,
      "num_tokens": 2042225442.0,
      "step": 2155
    },
    {
      "epoch": 4.219392752203722,
      "grad_norm": 0.08221294186841123,
      "learning_rate": 6.292596441069465e-07,
      "loss": 0.4023,
      "num_tokens": 2043189469.0,
      "step": 2156
    },
    {
      "epoch": 4.221351616062684,
      "grad_norm": 0.08339468501248067,
      "learning_rate": 6.261845773401936e-07,
      "loss": 0.4097,
      "num_tokens": 2044130437.0,
      "step": 2157
    },
    {
      "epoch": 4.223310479921645,
      "grad_norm": 0.08487069018364485,
      "learning_rate": 6.23116540609211e-07,
      "loss": 0.4136,
      "num_tokens": 2045048907.0,
      "step": 2158
    },
    {
      "epoch": 4.2252693437806075,
      "grad_norm": 0.08453487202604214,
      "learning_rate": 6.200555388452539e-07,
      "loss": 0.427,
      "num_tokens": 2046006203.0,
      "step": 2159
    },
    {
      "epoch": 4.227228207639569,
      "grad_norm": 0.08761268157203629,
      "learning_rate": 6.170015769682742e-07,
      "loss": 0.4231,
      "num_tokens": 2046896363.0,
      "step": 2160
    },
    {
      "epoch": 4.229187071498531,
      "grad_norm": 0.08754788294493071,
      "learning_rate": 6.13954659886905e-07,
      "loss": 0.4117,
      "num_tokens": 2047831879.0,
      "step": 2161
    },
    {
      "epoch": 4.2311459353574925,
      "grad_norm": 0.08407342007603844,
      "learning_rate": 6.109147924984576e-07,
      "loss": 0.4114,
      "num_tokens": 2048792085.0,
      "step": 2162
    },
    {
      "epoch": 4.233104799216455,
      "grad_norm": 0.08328412407219687,
      "learning_rate": 6.078819796889135e-07,
      "loss": 0.4194,
      "num_tokens": 2049720176.0,
      "step": 2163
    },
    {
      "epoch": 4.235063663075416,
      "grad_norm": 0.08622209159468973,
      "learning_rate": 6.048562263329139e-07,
      "loss": 0.4161,
      "num_tokens": 2050659907.0,
      "step": 2164
    },
    {
      "epoch": 4.237022526934378,
      "grad_norm": 0.08357107207721784,
      "learning_rate": 6.018375372937524e-07,
      "loss": 0.4305,
      "num_tokens": 2051624536.0,
      "step": 2165
    },
    {
      "epoch": 4.23898139079334,
      "grad_norm": 0.08622754908099019,
      "learning_rate": 5.988259174233713e-07,
      "loss": 0.406,
      "num_tokens": 2052521740.0,
      "step": 2166
    },
    {
      "epoch": 4.240940254652302,
      "grad_norm": 0.08163031176710157,
      "learning_rate": 5.958213715623479e-07,
      "loss": 0.4083,
      "num_tokens": 2053504048.0,
      "step": 2167
    },
    {
      "epoch": 4.2428991185112634,
      "grad_norm": 0.08585538701462192,
      "learning_rate": 5.928239045398898e-07,
      "loss": 0.4179,
      "num_tokens": 2054424204.0,
      "step": 2168
    },
    {
      "epoch": 4.244857982370226,
      "grad_norm": 0.08322893752348214,
      "learning_rate": 5.898335211738287e-07,
      "loss": 0.4209,
      "num_tokens": 2055386498.0,
      "step": 2169
    },
    {
      "epoch": 4.246816846229187,
      "grad_norm": 0.0836286720621953,
      "learning_rate": 5.868502262706083e-07,
      "loss": 0.4166,
      "num_tokens": 2056326300.0,
      "step": 2170
    },
    {
      "epoch": 4.2487757100881485,
      "grad_norm": 0.08553714321910702,
      "learning_rate": 5.838740246252794e-07,
      "loss": 0.4254,
      "num_tokens": 2057227482.0,
      "step": 2171
    },
    {
      "epoch": 4.250734573947111,
      "grad_norm": 0.08630438171310291,
      "learning_rate": 5.80904921021494e-07,
      "loss": 0.4164,
      "num_tokens": 2058127330.0,
      "step": 2172
    },
    {
      "epoch": 4.252693437806072,
      "grad_norm": 0.0824815008477906,
      "learning_rate": 5.779429202314929e-07,
      "loss": 0.4168,
      "num_tokens": 2059096919.0,
      "step": 2173
    },
    {
      "epoch": 4.254652301665034,
      "grad_norm": 0.08562295305796984,
      "learning_rate": 5.749880270161024e-07,
      "loss": 0.4093,
      "num_tokens": 2060003289.0,
      "step": 2174
    },
    {
      "epoch": 4.256611165523996,
      "grad_norm": 0.08357240470701881,
      "learning_rate": 5.720402461247221e-07,
      "loss": 0.4029,
      "num_tokens": 2060957367.0,
      "step": 2175
    },
    {
      "epoch": 4.258570029382958,
      "grad_norm": 0.08382743661617692,
      "learning_rate": 5.690995822953226e-07,
      "loss": 0.4111,
      "num_tokens": 2061942571.0,
      "step": 2176
    },
    {
      "epoch": 4.260528893241919,
      "grad_norm": 0.08196817106127785,
      "learning_rate": 5.66166040254435e-07,
      "loss": 0.4078,
      "num_tokens": 2062894641.0,
      "step": 2177
    },
    {
      "epoch": 4.262487757100882,
      "grad_norm": 0.0853492542610442,
      "learning_rate": 5.632396247171429e-07,
      "loss": 0.415,
      "num_tokens": 2063836749.0,
      "step": 2178
    },
    {
      "epoch": 4.264446620959843,
      "grad_norm": 0.08622616065735546,
      "learning_rate": 5.603203403870739e-07,
      "loss": 0.4008,
      "num_tokens": 2064731613.0,
      "step": 2179
    },
    {
      "epoch": 4.266405484818805,
      "grad_norm": 0.08526980675529142,
      "learning_rate": 5.574081919563973e-07,
      "loss": 0.4228,
      "num_tokens": 2065718089.0,
      "step": 2180
    },
    {
      "epoch": 4.268364348677767,
      "grad_norm": 0.08631245593373647,
      "learning_rate": 5.545031841058101e-07,
      "loss": 0.4268,
      "num_tokens": 2066648903.0,
      "step": 2181
    },
    {
      "epoch": 4.270323212536729,
      "grad_norm": 0.08397979015459724,
      "learning_rate": 5.516053215045314e-07,
      "loss": 0.4056,
      "num_tokens": 2067601088.0,
      "step": 2182
    },
    {
      "epoch": 4.27228207639569,
      "grad_norm": 0.08476284187754984,
      "learning_rate": 5.487146088102995e-07,
      "loss": 0.4072,
      "num_tokens": 2068570800.0,
      "step": 2183
    },
    {
      "epoch": 4.274240940254653,
      "grad_norm": 0.082501835170805,
      "learning_rate": 5.45831050669357e-07,
      "loss": 0.4139,
      "num_tokens": 2069560541.0,
      "step": 2184
    },
    {
      "epoch": 4.276199804113614,
      "grad_norm": 0.08472218411987088,
      "learning_rate": 5.429546517164486e-07,
      "loss": 0.4236,
      "num_tokens": 2070500169.0,
      "step": 2185
    },
    {
      "epoch": 4.278158667972576,
      "grad_norm": 0.08415131804090892,
      "learning_rate": 5.400854165748126e-07,
      "loss": 0.4208,
      "num_tokens": 2071451080.0,
      "step": 2186
    },
    {
      "epoch": 4.280117531831538,
      "grad_norm": 0.08524541923262709,
      "learning_rate": 5.372233498561713e-07,
      "loss": 0.4186,
      "num_tokens": 2072377052.0,
      "step": 2187
    },
    {
      "epoch": 4.2820763956905,
      "grad_norm": 0.08498805380670119,
      "learning_rate": 5.343684561607249e-07,
      "loss": 0.4199,
      "num_tokens": 2073338734.0,
      "step": 2188
    },
    {
      "epoch": 4.284035259549461,
      "grad_norm": 0.08155953449800538,
      "learning_rate": 5.315207400771483e-07,
      "loss": 0.4129,
      "num_tokens": 2074298989.0,
      "step": 2189
    },
    {
      "epoch": 4.285994123408424,
      "grad_norm": 0.08224383241962627,
      "learning_rate": 5.286802061825752e-07,
      "loss": 0.4097,
      "num_tokens": 2075272934.0,
      "step": 2190
    },
    {
      "epoch": 4.287952987267385,
      "grad_norm": 0.08258556595270888,
      "learning_rate": 5.258468590425969e-07,
      "loss": 0.4188,
      "num_tokens": 2076247593.0,
      "step": 2191
    },
    {
      "epoch": 4.289911851126346,
      "grad_norm": 0.08684278617350298,
      "learning_rate": 5.230207032112549e-07,
      "loss": 0.4098,
      "num_tokens": 2077159368.0,
      "step": 2192
    },
    {
      "epoch": 4.291870714985309,
      "grad_norm": 0.08136552409319969,
      "learning_rate": 5.202017432310302e-07,
      "loss": 0.4165,
      "num_tokens": 2078142112.0,
      "step": 2193
    },
    {
      "epoch": 4.29382957884427,
      "grad_norm": 0.0829622828213875,
      "learning_rate": 5.173899836328383e-07,
      "loss": 0.4147,
      "num_tokens": 2079116049.0,
      "step": 2194
    },
    {
      "epoch": 4.295788442703232,
      "grad_norm": 0.08520556202686495,
      "learning_rate": 5.145854289360224e-07,
      "loss": 0.4275,
      "num_tokens": 2080079397.0,
      "step": 2195
    },
    {
      "epoch": 4.297747306562194,
      "grad_norm": 0.08559487717274157,
      "learning_rate": 5.117880836483452e-07,
      "loss": 0.4218,
      "num_tokens": 2081033883.0,
      "step": 2196
    },
    {
      "epoch": 4.299706170421156,
      "grad_norm": 0.08186257263284755,
      "learning_rate": 5.089979522659799e-07,
      "loss": 0.389,
      "num_tokens": 2081955163.0,
      "step": 2197
    },
    {
      "epoch": 4.301665034280117,
      "grad_norm": 0.08937897689299613,
      "learning_rate": 5.062150392735071e-07,
      "loss": 0.4127,
      "num_tokens": 2082866119.0,
      "step": 2198
    },
    {
      "epoch": 4.3036238981390795,
      "grad_norm": 0.08526523692463407,
      "learning_rate": 5.034393491439044e-07,
      "loss": 0.4102,
      "num_tokens": 2083846571.0,
      "step": 2199
    },
    {
      "epoch": 4.305582761998041,
      "grad_norm": 0.08259604272892286,
      "learning_rate": 5.006708863385379e-07,
      "loss": 0.4068,
      "num_tokens": 2084800432.0,
      "step": 2200
    },
    {
      "epoch": 4.307541625857003,
      "grad_norm": 0.08492654876361953,
      "learning_rate": 4.979096553071638e-07,
      "loss": 0.4294,
      "num_tokens": 2085782308.0,
      "step": 2201
    },
    {
      "epoch": 4.3095004897159646,
      "grad_norm": 0.08687307163056773,
      "learning_rate": 4.951556604879049e-07,
      "loss": 0.4122,
      "num_tokens": 2086736289.0,
      "step": 2202
    },
    {
      "epoch": 4.311459353574927,
      "grad_norm": 0.08451750896812836,
      "learning_rate": 4.924089063072613e-07,
      "loss": 0.4145,
      "num_tokens": 2087672786.0,
      "step": 2203
    },
    {
      "epoch": 4.313418217433888,
      "grad_norm": 0.08381739276592902,
      "learning_rate": 4.896693971800925e-07,
      "loss": 0.4216,
      "num_tokens": 2088630583.0,
      "step": 2204
    },
    {
      "epoch": 4.3153770812928505,
      "grad_norm": 0.08897984130235548,
      "learning_rate": 4.869371375096115e-07,
      "loss": 0.4133,
      "num_tokens": 2089555353.0,
      "step": 2205
    },
    {
      "epoch": 4.317335945151812,
      "grad_norm": 0.08334445269234497,
      "learning_rate": 4.842121316873821e-07,
      "loss": 0.3994,
      "num_tokens": 2090483402.0,
      "step": 2206
    },
    {
      "epoch": 4.319294809010774,
      "grad_norm": 0.08202820823655337,
      "learning_rate": 4.814943840933073e-07,
      "loss": 0.4025,
      "num_tokens": 2091426935.0,
      "step": 2207
    },
    {
      "epoch": 4.3212536728697355,
      "grad_norm": 0.08770868215761991,
      "learning_rate": 4.787838990956228e-07,
      "loss": 0.4196,
      "num_tokens": 2092364121.0,
      "step": 2208
    },
    {
      "epoch": 4.323212536728698,
      "grad_norm": 0.08734326654192257,
      "learning_rate": 4.76080681050895e-07,
      "loss": 0.4121,
      "num_tokens": 2093306474.0,
      "step": 2209
    },
    {
      "epoch": 4.325171400587659,
      "grad_norm": 0.08315994682821065,
      "learning_rate": 4.73384734304006e-07,
      "loss": 0.4115,
      "num_tokens": 2094259529.0,
      "step": 2210
    },
    {
      "epoch": 4.327130264446621,
      "grad_norm": 0.08058722757886908,
      "learning_rate": 4.7069606318815133e-07,
      "loss": 0.3994,
      "num_tokens": 2095200979.0,
      "step": 2211
    },
    {
      "epoch": 4.329089128305583,
      "grad_norm": 0.08591503600597408,
      "learning_rate": 4.680146720248363e-07,
      "loss": 0.4242,
      "num_tokens": 2096145206.0,
      "step": 2212
    },
    {
      "epoch": 4.331047992164544,
      "grad_norm": 0.08200645306346635,
      "learning_rate": 4.653405651238607e-07,
      "loss": 0.4078,
      "num_tokens": 2097099830.0,
      "step": 2213
    },
    {
      "epoch": 4.333006856023506,
      "grad_norm": 0.082613043339824,
      "learning_rate": 4.62673746783317e-07,
      "loss": 0.4056,
      "num_tokens": 2098086966.0,
      "step": 2214
    },
    {
      "epoch": 4.334965719882468,
      "grad_norm": 0.08304868702514229,
      "learning_rate": 4.600142212895853e-07,
      "loss": 0.4122,
      "num_tokens": 2099028865.0,
      "step": 2215
    },
    {
      "epoch": 4.33692458374143,
      "grad_norm": 0.08005491049253148,
      "learning_rate": 4.5736199291732174e-07,
      "loss": 0.4126,
      "num_tokens": 2100028776.0,
      "step": 2216
    },
    {
      "epoch": 4.3388834476003915,
      "grad_norm": 0.08318850806063391,
      "learning_rate": 4.547170659294536e-07,
      "loss": 0.4123,
      "num_tokens": 2100966982.0,
      "step": 2217
    },
    {
      "epoch": 4.340842311459354,
      "grad_norm": 0.08207739567563224,
      "learning_rate": 4.5207944457717477e-07,
      "loss": 0.4183,
      "num_tokens": 2101944250.0,
      "step": 2218
    },
    {
      "epoch": 4.342801175318315,
      "grad_norm": 0.08152074304049574,
      "learning_rate": 4.494491330999351e-07,
      "loss": 0.4055,
      "num_tokens": 2102886073.0,
      "step": 2219
    },
    {
      "epoch": 4.344760039177277,
      "grad_norm": 0.08474302020174879,
      "learning_rate": 4.468261357254339e-07,
      "loss": 0.4154,
      "num_tokens": 2103819573.0,
      "step": 2220
    },
    {
      "epoch": 4.346718903036239,
      "grad_norm": 0.08638245066452833,
      "learning_rate": 4.442104566696187e-07,
      "loss": 0.4304,
      "num_tokens": 2104761432.0,
      "step": 2221
    },
    {
      "epoch": 4.348677766895201,
      "grad_norm": 0.08386037076332054,
      "learning_rate": 4.416021001366699e-07,
      "loss": 0.4183,
      "num_tokens": 2105697909.0,
      "step": 2222
    },
    {
      "epoch": 4.350636630754162,
      "grad_norm": 0.08448003629016722,
      "learning_rate": 4.390010703190006e-07,
      "loss": 0.3955,
      "num_tokens": 2106637338.0,
      "step": 2223
    },
    {
      "epoch": 4.352595494613125,
      "grad_norm": 0.084315991238533,
      "learning_rate": 4.364073713972489e-07,
      "loss": 0.406,
      "num_tokens": 2107559331.0,
      "step": 2224
    },
    {
      "epoch": 4.354554358472086,
      "grad_norm": 0.08402430619713995,
      "learning_rate": 4.338210075402666e-07,
      "loss": 0.4289,
      "num_tokens": 2108527970.0,
      "step": 2225
    },
    {
      "epoch": 4.356513222331048,
      "grad_norm": 0.08512842873957444,
      "learning_rate": 4.312419829051173e-07,
      "loss": 0.4143,
      "num_tokens": 2109475714.0,
      "step": 2226
    },
    {
      "epoch": 4.35847208619001,
      "grad_norm": 0.08384900087958944,
      "learning_rate": 4.286703016370719e-07,
      "loss": 0.4208,
      "num_tokens": 2110460226.0,
      "step": 2227
    },
    {
      "epoch": 4.360430950048972,
      "grad_norm": 0.08390734204509748,
      "learning_rate": 4.261059678695906e-07,
      "loss": 0.4197,
      "num_tokens": 2111415213.0,
      "step": 2228
    },
    {
      "epoch": 4.362389813907933,
      "grad_norm": 0.08199699894461897,
      "learning_rate": 4.2354898572433136e-07,
      "loss": 0.4022,
      "num_tokens": 2112335020.0,
      "step": 2229
    },
    {
      "epoch": 4.364348677766896,
      "grad_norm": 0.08426215459972695,
      "learning_rate": 4.209993593111317e-07,
      "loss": 0.4065,
      "num_tokens": 2113274626.0,
      "step": 2230
    },
    {
      "epoch": 4.366307541625857,
      "grad_norm": 0.08147690133629366,
      "learning_rate": 4.184570927280074e-07,
      "loss": 0.4006,
      "num_tokens": 2114245201.0,
      "step": 2231
    },
    {
      "epoch": 4.368266405484819,
      "grad_norm": 0.08622121832133177,
      "learning_rate": 4.1592219006114597e-07,
      "loss": 0.4256,
      "num_tokens": 2115197965.0,
      "step": 2232
    },
    {
      "epoch": 4.370225269343781,
      "grad_norm": 0.08048327534010842,
      "learning_rate": 4.1339465538489666e-07,
      "loss": 0.4002,
      "num_tokens": 2116157469.0,
      "step": 2233
    },
    {
      "epoch": 4.372184133202742,
      "grad_norm": 0.08493546129868577,
      "learning_rate": 4.108744927617669e-07,
      "loss": 0.4058,
      "num_tokens": 2117128306.0,
      "step": 2234
    },
    {
      "epoch": 4.374142997061704,
      "grad_norm": 0.08182394768372599,
      "learning_rate": 4.083617062424172e-07,
      "loss": 0.4226,
      "num_tokens": 2118104858.0,
      "step": 2235
    },
    {
      "epoch": 4.376101860920666,
      "grad_norm": 0.08122698124721504,
      "learning_rate": 4.0585629986564954e-07,
      "loss": 0.4094,
      "num_tokens": 2119080292.0,
      "step": 2236
    },
    {
      "epoch": 4.378060724779628,
      "grad_norm": 0.08298022529596868,
      "learning_rate": 4.033582776584033e-07,
      "loss": 0.4056,
      "num_tokens": 2120028954.0,
      "step": 2237
    },
    {
      "epoch": 4.380019588638589,
      "grad_norm": 0.0846494002618381,
      "learning_rate": 4.008676436357539e-07,
      "loss": 0.405,
      "num_tokens": 2120966719.0,
      "step": 2238
    },
    {
      "epoch": 4.381978452497552,
      "grad_norm": 0.08372569010082824,
      "learning_rate": 3.98384401800897e-07,
      "loss": 0.4167,
      "num_tokens": 2121904322.0,
      "step": 2239
    },
    {
      "epoch": 4.383937316356513,
      "grad_norm": 0.08581288762711979,
      "learning_rate": 3.9590855614514757e-07,
      "loss": 0.4311,
      "num_tokens": 2122858061.0,
      "step": 2240
    },
    {
      "epoch": 4.385896180215475,
      "grad_norm": 0.08413316811852717,
      "learning_rate": 3.934401106479352e-07,
      "loss": 0.4107,
      "num_tokens": 2123804370.0,
      "step": 2241
    },
    {
      "epoch": 4.387855044074437,
      "grad_norm": 0.08331964490849027,
      "learning_rate": 3.909790692767934e-07,
      "loss": 0.403,
      "num_tokens": 2124702332.0,
      "step": 2242
    },
    {
      "epoch": 4.389813907933399,
      "grad_norm": 0.08498461246276752,
      "learning_rate": 3.8852543598735413e-07,
      "loss": 0.4057,
      "num_tokens": 2125605612.0,
      "step": 2243
    },
    {
      "epoch": 4.39177277179236,
      "grad_norm": 0.08226870123121025,
      "learning_rate": 3.86079214723345e-07,
      "loss": 0.4101,
      "num_tokens": 2126590824.0,
      "step": 2244
    },
    {
      "epoch": 4.3937316356513225,
      "grad_norm": 0.08362359404104432,
      "learning_rate": 3.8364040941657685e-07,
      "loss": 0.4187,
      "num_tokens": 2127552800.0,
      "step": 2245
    },
    {
      "epoch": 4.395690499510284,
      "grad_norm": 0.08441545341310425,
      "learning_rate": 3.812090239869448e-07,
      "loss": 0.4343,
      "num_tokens": 2128515212.0,
      "step": 2246
    },
    {
      "epoch": 4.397649363369246,
      "grad_norm": 0.08384512349331089,
      "learning_rate": 3.78785062342415e-07,
      "loss": 0.3941,
      "num_tokens": 2129442558.0,
      "step": 2247
    },
    {
      "epoch": 4.3996082272282075,
      "grad_norm": 0.08129428902959814,
      "learning_rate": 3.763685283790208e-07,
      "loss": 0.4083,
      "num_tokens": 2130423219.0,
      "step": 2248
    },
    {
      "epoch": 4.40156709108717,
      "grad_norm": 0.08373615458913583,
      "learning_rate": 3.73959425980861e-07,
      "loss": 0.422,
      "num_tokens": 2131383809.0,
      "step": 2249
    },
    {
      "epoch": 4.403525954946131,
      "grad_norm": 0.08493424193543722,
      "learning_rate": 3.7155775902008527e-07,
      "loss": 0.4104,
      "num_tokens": 2132363826.0,
      "step": 2250
    },
    {
      "epoch": 4.4054848188050935,
      "grad_norm": 0.08591869997551338,
      "learning_rate": 3.691635313568942e-07,
      "loss": 0.4136,
      "num_tokens": 2133338319.0,
      "step": 2251
    },
    {
      "epoch": 4.407443682664055,
      "grad_norm": 0.08332966368334885,
      "learning_rate": 3.667767468395328e-07,
      "loss": 0.4201,
      "num_tokens": 2134278623.0,
      "step": 2252
    },
    {
      "epoch": 4.409402546523017,
      "grad_norm": 0.08221686813731542,
      "learning_rate": 3.643974093042807e-07,
      "loss": 0.3929,
      "num_tokens": 2135231101.0,
      "step": 2253
    },
    {
      "epoch": 4.4113614103819785,
      "grad_norm": 0.0836099528841157,
      "learning_rate": 3.620255225754454e-07,
      "loss": 0.3987,
      "num_tokens": 2136147587.0,
      "step": 2254
    },
    {
      "epoch": 4.41332027424094,
      "grad_norm": 0.08010666325081656,
      "learning_rate": 3.596610904653652e-07,
      "loss": 0.3988,
      "num_tokens": 2137099188.0,
      "step": 2255
    },
    {
      "epoch": 4.415279138099902,
      "grad_norm": 0.08403197010040314,
      "learning_rate": 3.5730411677439125e-07,
      "loss": 0.4124,
      "num_tokens": 2138041070.0,
      "step": 2256
    },
    {
      "epoch": 4.4172380019588635,
      "grad_norm": 0.0852261359677142,
      "learning_rate": 3.5495460529088865e-07,
      "loss": 0.4135,
      "num_tokens": 2138959695.0,
      "step": 2257
    },
    {
      "epoch": 4.419196865817826,
      "grad_norm": 0.08243881441915217,
      "learning_rate": 3.5261255979122946e-07,
      "loss": 0.4095,
      "num_tokens": 2139883279.0,
      "step": 2258
    },
    {
      "epoch": 4.421155729676787,
      "grad_norm": 0.08835912267629452,
      "learning_rate": 3.5027798403978294e-07,
      "loss": 0.4239,
      "num_tokens": 2140831527.0,
      "step": 2259
    },
    {
      "epoch": 4.423114593535749,
      "grad_norm": 0.08537049208908011,
      "learning_rate": 3.4795088178891543e-07,
      "loss": 0.4229,
      "num_tokens": 2141779156.0,
      "step": 2260
    },
    {
      "epoch": 4.425073457394711,
      "grad_norm": 0.08880943437711142,
      "learning_rate": 3.4563125677897936e-07,
      "loss": 0.4103,
      "num_tokens": 2142725642.0,
      "step": 2261
    },
    {
      "epoch": 4.427032321253673,
      "grad_norm": 0.0833425437987292,
      "learning_rate": 3.433191127383079e-07,
      "loss": 0.4218,
      "num_tokens": 2143677235.0,
      "step": 2262
    },
    {
      "epoch": 4.4289911851126345,
      "grad_norm": 0.0846244693188937,
      "learning_rate": 3.410144533832127e-07,
      "loss": 0.4131,
      "num_tokens": 2144611510.0,
      "step": 2263
    },
    {
      "epoch": 4.430950048971597,
      "grad_norm": 0.08527690690387682,
      "learning_rate": 3.3871728241797406e-07,
      "loss": 0.4273,
      "num_tokens": 2145590702.0,
      "step": 2264
    },
    {
      "epoch": 4.432908912830558,
      "grad_norm": 0.08344372448433705,
      "learning_rate": 3.364276035348346e-07,
      "loss": 0.4171,
      "num_tokens": 2146540219.0,
      "step": 2265
    },
    {
      "epoch": 4.43486777668952,
      "grad_norm": 0.08558048600760117,
      "learning_rate": 3.3414542041399764e-07,
      "loss": 0.4152,
      "num_tokens": 2147496479.0,
      "step": 2266
    },
    {
      "epoch": 4.436826640548482,
      "grad_norm": 0.08380365344074671,
      "learning_rate": 3.318707367236174e-07,
      "loss": 0.4077,
      "num_tokens": 2148462954.0,
      "step": 2267
    },
    {
      "epoch": 4.438785504407444,
      "grad_norm": 0.08100272244534662,
      "learning_rate": 3.2960355611979245e-07,
      "loss": 0.4093,
      "num_tokens": 2149440780.0,
      "step": 2268
    },
    {
      "epoch": 4.440744368266405,
      "grad_norm": 0.08705743733780545,
      "learning_rate": 3.2734388224656575e-07,
      "loss": 0.418,
      "num_tokens": 2150369881.0,
      "step": 2269
    },
    {
      "epoch": 4.442703232125368,
      "grad_norm": 0.08330266108193204,
      "learning_rate": 3.2509171873591095e-07,
      "loss": 0.4,
      "num_tokens": 2151281705.0,
      "step": 2270
    },
    {
      "epoch": 4.444662095984329,
      "grad_norm": 0.08065185209022314,
      "learning_rate": 3.2284706920773067e-07,
      "loss": 0.4111,
      "num_tokens": 2152244910.0,
      "step": 2271
    },
    {
      "epoch": 4.44662095984329,
      "grad_norm": 0.0852183829272433,
      "learning_rate": 3.206099372698529e-07,
      "loss": 0.4036,
      "num_tokens": 2153198567.0,
      "step": 2272
    },
    {
      "epoch": 4.448579823702253,
      "grad_norm": 0.08618367096331993,
      "learning_rate": 3.183803265180208e-07,
      "loss": 0.4262,
      "num_tokens": 2154141710.0,
      "step": 2273
    },
    {
      "epoch": 4.450538687561215,
      "grad_norm": 0.08260727856664073,
      "learning_rate": 3.1615824053588764e-07,
      "loss": 0.4168,
      "num_tokens": 2155127523.0,
      "step": 2274
    },
    {
      "epoch": 4.452497551420176,
      "grad_norm": 0.08570730391592203,
      "learning_rate": 3.139436828950143e-07,
      "loss": 0.4113,
      "num_tokens": 2156076271.0,
      "step": 2275
    },
    {
      "epoch": 4.454456415279138,
      "grad_norm": 0.086741957208345,
      "learning_rate": 3.1173665715486076e-07,
      "loss": 0.4305,
      "num_tokens": 2157004084.0,
      "step": 2276
    },
    {
      "epoch": 4.4564152791381,
      "grad_norm": 0.08418931419252824,
      "learning_rate": 3.0953716686277944e-07,
      "loss": 0.3968,
      "num_tokens": 2157937532.0,
      "step": 2277
    },
    {
      "epoch": 4.458374142997061,
      "grad_norm": 0.08326658418956387,
      "learning_rate": 3.073452155540135e-07,
      "loss": 0.4109,
      "num_tokens": 2158874784.0,
      "step": 2278
    },
    {
      "epoch": 4.460333006856024,
      "grad_norm": 0.08141615131737735,
      "learning_rate": 3.0516080675168645e-07,
      "loss": 0.4004,
      "num_tokens": 2159841810.0,
      "step": 2279
    },
    {
      "epoch": 4.462291870714985,
      "grad_norm": 0.08284872071901574,
      "learning_rate": 3.029839439668003e-07,
      "loss": 0.4135,
      "num_tokens": 2160783954.0,
      "step": 2280
    },
    {
      "epoch": 4.464250734573947,
      "grad_norm": 0.08508399315997507,
      "learning_rate": 3.008146306982274e-07,
      "loss": 0.4084,
      "num_tokens": 2161701970.0,
      "step": 2281
    },
    {
      "epoch": 4.466209598432909,
      "grad_norm": 0.08563041485398272,
      "learning_rate": 2.9865287043270517e-07,
      "loss": 0.4097,
      "num_tokens": 2162627195.0,
      "step": 2282
    },
    {
      "epoch": 4.468168462291871,
      "grad_norm": 0.08574590565029692,
      "learning_rate": 2.9649866664483387e-07,
      "loss": 0.4021,
      "num_tokens": 2163552297.0,
      "step": 2283
    },
    {
      "epoch": 4.470127326150832,
      "grad_norm": 0.08449605980534906,
      "learning_rate": 2.943520227970653e-07,
      "loss": 0.4216,
      "num_tokens": 2164506134.0,
      "step": 2284
    },
    {
      "epoch": 4.472086190009795,
      "grad_norm": 0.08781311090859446,
      "learning_rate": 2.922129423397008e-07,
      "loss": 0.4009,
      "num_tokens": 2165440702.0,
      "step": 2285
    },
    {
      "epoch": 4.474045053868756,
      "grad_norm": 0.08384345960497813,
      "learning_rate": 2.9008142871088665e-07,
      "loss": 0.4124,
      "num_tokens": 2166396224.0,
      "step": 2286
    },
    {
      "epoch": 4.476003917727718,
      "grad_norm": 0.0826491722653405,
      "learning_rate": 2.8795748533660515e-07,
      "loss": 0.3996,
      "num_tokens": 2167356792.0,
      "step": 2287
    },
    {
      "epoch": 4.47796278158668,
      "grad_norm": 0.08388780111752804,
      "learning_rate": 2.8584111563067065e-07,
      "loss": 0.4069,
      "num_tokens": 2168322722.0,
      "step": 2288
    },
    {
      "epoch": 4.479921645445642,
      "grad_norm": 0.08150606305818417,
      "learning_rate": 2.83732322994727e-07,
      "loss": 0.3975,
      "num_tokens": 2169253064.0,
      "step": 2289
    },
    {
      "epoch": 4.481880509304603,
      "grad_norm": 0.08481783368686523,
      "learning_rate": 2.816311108182368e-07,
      "loss": 0.4199,
      "num_tokens": 2170195849.0,
      "step": 2290
    },
    {
      "epoch": 4.4838393731635655,
      "grad_norm": 0.08559890570711427,
      "learning_rate": 2.795374824784791e-07,
      "loss": 0.4134,
      "num_tokens": 2171151219.0,
      "step": 2291
    },
    {
      "epoch": 4.485798237022527,
      "grad_norm": 0.08538626033193797,
      "learning_rate": 2.774514413405444e-07,
      "loss": 0.4113,
      "num_tokens": 2172080606.0,
      "step": 2292
    },
    {
      "epoch": 4.487757100881488,
      "grad_norm": 0.08300910596689294,
      "learning_rate": 2.7537299075732824e-07,
      "loss": 0.4043,
      "num_tokens": 2173073159.0,
      "step": 2293
    },
    {
      "epoch": 4.4897159647404505,
      "grad_norm": 0.08618134206206447,
      "learning_rate": 2.7330213406952357e-07,
      "loss": 0.412,
      "num_tokens": 2174037855.0,
      "step": 2294
    },
    {
      "epoch": 4.491674828599413,
      "grad_norm": 0.09072514236070624,
      "learning_rate": 2.71238874605621e-07,
      "loss": 0.4361,
      "num_tokens": 2174993777.0,
      "step": 2295
    },
    {
      "epoch": 4.493633692458374,
      "grad_norm": 0.08157708889530797,
      "learning_rate": 2.69183215681898e-07,
      "loss": 0.4031,
      "num_tokens": 2175966634.0,
      "step": 2296
    },
    {
      "epoch": 4.495592556317336,
      "grad_norm": 0.08577225461564279,
      "learning_rate": 2.671351606024153e-07,
      "loss": 0.4098,
      "num_tokens": 2176900109.0,
      "step": 2297
    },
    {
      "epoch": 4.497551420176298,
      "grad_norm": 0.08731063558403147,
      "learning_rate": 2.6509471265901476e-07,
      "loss": 0.4135,
      "num_tokens": 2177793712.0,
      "step": 2298
    },
    {
      "epoch": 4.499510284035259,
      "grad_norm": 0.08477879481994724,
      "learning_rate": 2.6306187513130853e-07,
      "loss": 0.411,
      "num_tokens": 2178711336.0,
      "step": 2299
    },
    {
      "epoch": 4.5014691478942215,
      "grad_norm": 0.0847002053644622,
      "learning_rate": 2.6103665128667633e-07,
      "loss": 0.4204,
      "num_tokens": 2179653814.0,
      "step": 2300
    },
    {
      "epoch": 4.503428011753183,
      "grad_norm": 0.08441303701627194,
      "learning_rate": 2.590190443802626e-07,
      "loss": 0.4099,
      "num_tokens": 2180594507.0,
      "step": 2301
    },
    {
      "epoch": 4.505386875612145,
      "grad_norm": 0.08581206332258794,
      "learning_rate": 2.570090576549683e-07,
      "loss": 0.4142,
      "num_tokens": 2181528961.0,
      "step": 2302
    },
    {
      "epoch": 4.5073457394711065,
      "grad_norm": 0.08619673806619452,
      "learning_rate": 2.550066943414453e-07,
      "loss": 0.4108,
      "num_tokens": 2182439074.0,
      "step": 2303
    },
    {
      "epoch": 4.509304603330069,
      "grad_norm": 0.08664944180446546,
      "learning_rate": 2.530119576580936e-07,
      "loss": 0.4313,
      "num_tokens": 2183381993.0,
      "step": 2304
    },
    {
      "epoch": 4.51126346718903,
      "grad_norm": 0.0892073424588054,
      "learning_rate": 2.5102485081105464e-07,
      "loss": 0.4191,
      "num_tokens": 2184321837.0,
      "step": 2305
    },
    {
      "epoch": 4.513222331047992,
      "grad_norm": 0.08470427006361182,
      "learning_rate": 2.4904537699420585e-07,
      "loss": 0.4167,
      "num_tokens": 2185269117.0,
      "step": 2306
    },
    {
      "epoch": 4.515181194906954,
      "grad_norm": 0.08263284273956638,
      "learning_rate": 2.4707353938915656e-07,
      "loss": 0.4,
      "num_tokens": 2186215907.0,
      "step": 2307
    },
    {
      "epoch": 4.517140058765916,
      "grad_norm": 0.08322266205259499,
      "learning_rate": 2.4510934116524167e-07,
      "loss": 0.4045,
      "num_tokens": 2187156847.0,
      "step": 2308
    },
    {
      "epoch": 4.5190989226248774,
      "grad_norm": 0.08479976261062598,
      "learning_rate": 2.431527854795196e-07,
      "loss": 0.421,
      "num_tokens": 2188122917.0,
      "step": 2309
    },
    {
      "epoch": 4.52105778648384,
      "grad_norm": 0.08043269562909104,
      "learning_rate": 2.412038754767626e-07,
      "loss": 0.4051,
      "num_tokens": 2189076512.0,
      "step": 2310
    },
    {
      "epoch": 4.523016650342801,
      "grad_norm": 0.08492742322181865,
      "learning_rate": 2.3926261428945386e-07,
      "loss": 0.4235,
      "num_tokens": 2190025165.0,
      "step": 2311
    },
    {
      "epoch": 4.524975514201763,
      "grad_norm": 0.08560696506075864,
      "learning_rate": 2.3732900503778523e-07,
      "loss": 0.4143,
      "num_tokens": 2190951086.0,
      "step": 2312
    },
    {
      "epoch": 4.526934378060725,
      "grad_norm": 0.08009458542404425,
      "learning_rate": 2.354030508296462e-07,
      "loss": 0.3956,
      "num_tokens": 2191901846.0,
      "step": 2313
    },
    {
      "epoch": 4.528893241919686,
      "grad_norm": 0.0847903675690526,
      "learning_rate": 2.3348475476062448e-07,
      "loss": 0.4034,
      "num_tokens": 2192855624.0,
      "step": 2314
    },
    {
      "epoch": 4.530852105778648,
      "grad_norm": 0.08344768867181616,
      "learning_rate": 2.3157411991399925e-07,
      "loss": 0.4002,
      "num_tokens": 2193794408.0,
      "step": 2315
    },
    {
      "epoch": 4.532810969637611,
      "grad_norm": 0.08278655922588328,
      "learning_rate": 2.2967114936073342e-07,
      "loss": 0.4308,
      "num_tokens": 2194760213.0,
      "step": 2316
    },
    {
      "epoch": 4.534769833496572,
      "grad_norm": 0.08429760750186942,
      "learning_rate": 2.2777584615947313e-07,
      "loss": 0.4286,
      "num_tokens": 2195732735.0,
      "step": 2317
    },
    {
      "epoch": 4.536728697355533,
      "grad_norm": 0.0826348218518181,
      "learning_rate": 2.2588821335654044e-07,
      "loss": 0.4083,
      "num_tokens": 2196684662.0,
      "step": 2318
    },
    {
      "epoch": 4.538687561214496,
      "grad_norm": 0.08613819047594629,
      "learning_rate": 2.240082539859284e-07,
      "loss": 0.4262,
      "num_tokens": 2197653734.0,
      "step": 2319
    },
    {
      "epoch": 4.540646425073457,
      "grad_norm": 0.08738738991907348,
      "learning_rate": 2.2213597106929608e-07,
      "loss": 0.4253,
      "num_tokens": 2198584761.0,
      "step": 2320
    },
    {
      "epoch": 4.542605288932419,
      "grad_norm": 0.08283335755329545,
      "learning_rate": 2.2027136761596623e-07,
      "loss": 0.411,
      "num_tokens": 2199518797.0,
      "step": 2321
    },
    {
      "epoch": 4.544564152791381,
      "grad_norm": 0.08574179247280905,
      "learning_rate": 2.1841444662291543e-07,
      "loss": 0.4085,
      "num_tokens": 2200465168.0,
      "step": 2322
    },
    {
      "epoch": 4.546523016650343,
      "grad_norm": 0.08510442520869327,
      "learning_rate": 2.16565211074774e-07,
      "loss": 0.4178,
      "num_tokens": 2201420313.0,
      "step": 2323
    },
    {
      "epoch": 4.548481880509304,
      "grad_norm": 0.0827048706537416,
      "learning_rate": 2.147236639438205e-07,
      "loss": 0.406,
      "num_tokens": 2202345114.0,
      "step": 2324
    },
    {
      "epoch": 4.550440744368267,
      "grad_norm": 0.08470254714454753,
      "learning_rate": 2.1288980818997272e-07,
      "loss": 0.4162,
      "num_tokens": 2203283922.0,
      "step": 2325
    },
    {
      "epoch": 4.552399608227228,
      "grad_norm": 0.08173116438802071,
      "learning_rate": 2.1106364676078906e-07,
      "loss": 0.3965,
      "num_tokens": 2204239649.0,
      "step": 2326
    },
    {
      "epoch": 4.55435847208619,
      "grad_norm": 0.0815631082705351,
      "learning_rate": 2.0924518259145933e-07,
      "loss": 0.4143,
      "num_tokens": 2205198101.0,
      "step": 2327
    },
    {
      "epoch": 4.556317335945152,
      "grad_norm": 0.08418789645376833,
      "learning_rate": 2.0743441860480218e-07,
      "loss": 0.4087,
      "num_tokens": 2206132206.0,
      "step": 2328
    },
    {
      "epoch": 4.558276199804114,
      "grad_norm": 0.08308236583244617,
      "learning_rate": 2.0563135771125896e-07,
      "loss": 0.4166,
      "num_tokens": 2207092223.0,
      "step": 2329
    },
    {
      "epoch": 4.560235063663075,
      "grad_norm": 0.08685073165635805,
      "learning_rate": 2.0383600280889148e-07,
      "loss": 0.4254,
      "num_tokens": 2208040243.0,
      "step": 2330
    },
    {
      "epoch": 4.562193927522038,
      "grad_norm": 0.0816330532045113,
      "learning_rate": 2.0204835678337363e-07,
      "loss": 0.4104,
      "num_tokens": 2208988026.0,
      "step": 2331
    },
    {
      "epoch": 4.564152791380999,
      "grad_norm": 0.08342726520601144,
      "learning_rate": 2.0026842250799038e-07,
      "loss": 0.414,
      "num_tokens": 2209933323.0,
      "step": 2332
    },
    {
      "epoch": 4.566111655239961,
      "grad_norm": 0.08594110751750678,
      "learning_rate": 1.9849620284363047e-07,
      "loss": 0.4214,
      "num_tokens": 2210879277.0,
      "step": 2333
    },
    {
      "epoch": 4.568070519098923,
      "grad_norm": 0.08322421671137203,
      "learning_rate": 1.967317006387831e-07,
      "loss": 0.4018,
      "num_tokens": 2211792338.0,
      "step": 2334
    },
    {
      "epoch": 4.570029382957884,
      "grad_norm": 0.08366428902197062,
      "learning_rate": 1.9497491872953466e-07,
      "loss": 0.406,
      "num_tokens": 2212749293.0,
      "step": 2335
    },
    {
      "epoch": 4.571988246816846,
      "grad_norm": 0.08728341413641615,
      "learning_rate": 1.9322585993956145e-07,
      "loss": 0.4187,
      "num_tokens": 2213676954.0,
      "step": 2336
    },
    {
      "epoch": 4.5739471106758085,
      "grad_norm": 0.07969672172446281,
      "learning_rate": 1.9148452708012522e-07,
      "loss": 0.4113,
      "num_tokens": 2214654180.0,
      "step": 2337
    },
    {
      "epoch": 4.57590597453477,
      "grad_norm": 0.08440667566458711,
      "learning_rate": 1.897509229500727e-07,
      "loss": 0.4115,
      "num_tokens": 2215566546.0,
      "step": 2338
    },
    {
      "epoch": 4.577864838393731,
      "grad_norm": 0.08328162617540631,
      "learning_rate": 1.8802505033582608e-07,
      "loss": 0.388,
      "num_tokens": 2216506180.0,
      "step": 2339
    },
    {
      "epoch": 4.5798237022526935,
      "grad_norm": 0.08060675956140442,
      "learning_rate": 1.863069120113814e-07,
      "loss": 0.4117,
      "num_tokens": 2217488576.0,
      "step": 2340
    },
    {
      "epoch": 4.581782566111655,
      "grad_norm": 0.0814430156731195,
      "learning_rate": 1.8459651073830297e-07,
      "loss": 0.4115,
      "num_tokens": 2218451065.0,
      "step": 2341
    },
    {
      "epoch": 4.583741429970617,
      "grad_norm": 0.08540439417185391,
      "learning_rate": 1.8289384926572062e-07,
      "loss": 0.4227,
      "num_tokens": 2219388730.0,
      "step": 2342
    },
    {
      "epoch": 4.5857002938295786,
      "grad_norm": 0.08222524864127444,
      "learning_rate": 1.8119893033032132e-07,
      "loss": 0.4024,
      "num_tokens": 2220360189.0,
      "step": 2343
    },
    {
      "epoch": 4.587659157688541,
      "grad_norm": 0.0853117061101539,
      "learning_rate": 1.795117566563509e-07,
      "loss": 0.4223,
      "num_tokens": 2221316945.0,
      "step": 2344
    },
    {
      "epoch": 4.589618021547502,
      "grad_norm": 0.0867948644050997,
      "learning_rate": 1.778323309556035e-07,
      "loss": 0.438,
      "num_tokens": 2222244470.0,
      "step": 2345
    },
    {
      "epoch": 4.5915768854064645,
      "grad_norm": 0.0847031120690201,
      "learning_rate": 1.7616065592742038e-07,
      "loss": 0.4095,
      "num_tokens": 2223206358.0,
      "step": 2346
    },
    {
      "epoch": 4.593535749265426,
      "grad_norm": 0.08238375704959286,
      "learning_rate": 1.7449673425868674e-07,
      "loss": 0.4082,
      "num_tokens": 2224167450.0,
      "step": 2347
    },
    {
      "epoch": 4.595494613124388,
      "grad_norm": 0.08226070434458777,
      "learning_rate": 1.7284056862382492e-07,
      "loss": 0.3996,
      "num_tokens": 2225119088.0,
      "step": 2348
    },
    {
      "epoch": 4.5974534769833495,
      "grad_norm": 0.08492545584995367,
      "learning_rate": 1.7119216168478947e-07,
      "loss": 0.4193,
      "num_tokens": 2226073864.0,
      "step": 2349
    },
    {
      "epoch": 4.599412340842312,
      "grad_norm": 0.08158618882047468,
      "learning_rate": 1.6955151609106713e-07,
      "loss": 0.4164,
      "num_tokens": 2227067448.0,
      "step": 2350
    },
    {
      "epoch": 4.601371204701273,
      "grad_norm": 0.08423025376913923,
      "learning_rate": 1.6791863447966738e-07,
      "loss": 0.4263,
      "num_tokens": 2227998850.0,
      "step": 2351
    },
    {
      "epoch": 4.603330068560235,
      "grad_norm": 0.08110131092092367,
      "learning_rate": 1.6629351947512195e-07,
      "loss": 0.4043,
      "num_tokens": 2228982717.0,
      "step": 2352
    },
    {
      "epoch": 4.605288932419197,
      "grad_norm": 0.08481134528549475,
      "learning_rate": 1.6467617368947918e-07,
      "loss": 0.4204,
      "num_tokens": 2229924297.0,
      "step": 2353
    },
    {
      "epoch": 4.607247796278159,
      "grad_norm": 0.08910601063308517,
      "learning_rate": 1.6306659972230021e-07,
      "loss": 0.4131,
      "num_tokens": 2230849377.0,
      "step": 2354
    },
    {
      "epoch": 4.60920666013712,
      "grad_norm": 0.08470823640302579,
      "learning_rate": 1.614648001606528e-07,
      "loss": 0.4192,
      "num_tokens": 2231804698.0,
      "step": 2355
    },
    {
      "epoch": 4.611165523996082,
      "grad_norm": 0.07968500335160754,
      "learning_rate": 1.598707775791114e-07,
      "loss": 0.3972,
      "num_tokens": 2232778503.0,
      "step": 2356
    },
    {
      "epoch": 4.613124387855044,
      "grad_norm": 0.08441112155069348,
      "learning_rate": 1.582845345397488e-07,
      "loss": 0.406,
      "num_tokens": 2233701567.0,
      "step": 2357
    },
    {
      "epoch": 4.615083251714006,
      "grad_norm": 0.08406791439153873,
      "learning_rate": 1.5670607359213442e-07,
      "loss": 0.4166,
      "num_tokens": 2234655517.0,
      "step": 2358
    },
    {
      "epoch": 4.617042115572968,
      "grad_norm": 0.08634533203708583,
      "learning_rate": 1.551353972733305e-07,
      "loss": 0.4026,
      "num_tokens": 2235561182.0,
      "step": 2359
    },
    {
      "epoch": 4.619000979431929,
      "grad_norm": 0.08036240529465088,
      "learning_rate": 1.5357250810788316e-07,
      "loss": 0.4112,
      "num_tokens": 2236524351.0,
      "step": 2360
    },
    {
      "epoch": 4.620959843290891,
      "grad_norm": 0.08539634637474039,
      "learning_rate": 1.5201740860782687e-07,
      "loss": 0.4179,
      "num_tokens": 2237468329.0,
      "step": 2361
    },
    {
      "epoch": 4.622918707149853,
      "grad_norm": 0.08293143545477717,
      "learning_rate": 1.5047010127267393e-07,
      "loss": 0.4017,
      "num_tokens": 2238427138.0,
      "step": 2362
    },
    {
      "epoch": 4.624877571008815,
      "grad_norm": 0.07958122971564409,
      "learning_rate": 1.48930588589411e-07,
      "loss": 0.4081,
      "num_tokens": 2239422421.0,
      "step": 2363
    },
    {
      "epoch": 4.626836434867776,
      "grad_norm": 0.08192479917068426,
      "learning_rate": 1.4739887303249877e-07,
      "loss": 0.4127,
      "num_tokens": 2240381325.0,
      "step": 2364
    },
    {
      "epoch": 4.628795298726739,
      "grad_norm": 0.0851374889618941,
      "learning_rate": 1.45874957063864e-07,
      "loss": 0.4108,
      "num_tokens": 2241344982.0,
      "step": 2365
    },
    {
      "epoch": 4.6307541625857,
      "grad_norm": 0.08630612690419566,
      "learning_rate": 1.443588431328974e-07,
      "loss": 0.4118,
      "num_tokens": 2242308682.0,
      "step": 2366
    },
    {
      "epoch": 4.632713026444662,
      "grad_norm": 0.08219955862988715,
      "learning_rate": 1.4285053367645074e-07,
      "loss": 0.4091,
      "num_tokens": 2243249460.0,
      "step": 2367
    },
    {
      "epoch": 4.634671890303624,
      "grad_norm": 0.08206520215294191,
      "learning_rate": 1.4135003111882928e-07,
      "loss": 0.4262,
      "num_tokens": 2244236315.0,
      "step": 2368
    },
    {
      "epoch": 4.636630754162586,
      "grad_norm": 0.0843696440191487,
      "learning_rate": 1.3985733787179157e-07,
      "loss": 0.411,
      "num_tokens": 2245164535.0,
      "step": 2369
    },
    {
      "epoch": 4.638589618021547,
      "grad_norm": 0.0822541471251733,
      "learning_rate": 1.3837245633454512e-07,
      "loss": 0.4066,
      "num_tokens": 2246109231.0,
      "step": 2370
    },
    {
      "epoch": 4.64054848188051,
      "grad_norm": 0.0788509196853584,
      "learning_rate": 1.3689538889373965e-07,
      "loss": 0.3823,
      "num_tokens": 2247063965.0,
      "step": 2371
    },
    {
      "epoch": 4.642507345739471,
      "grad_norm": 0.08516640770529209,
      "learning_rate": 1.354261379234678e-07,
      "loss": 0.4042,
      "num_tokens": 2247990221.0,
      "step": 2372
    },
    {
      "epoch": 4.644466209598433,
      "grad_norm": 0.0846260946119953,
      "learning_rate": 1.3396470578525723e-07,
      "loss": 0.4203,
      "num_tokens": 2248905364.0,
      "step": 2373
    },
    {
      "epoch": 4.646425073457395,
      "grad_norm": 0.08562926935925291,
      "learning_rate": 1.3251109482806667e-07,
      "loss": 0.4071,
      "num_tokens": 2249826876.0,
      "step": 2374
    },
    {
      "epoch": 4.648383937316357,
      "grad_norm": 0.08280272498249136,
      "learning_rate": 1.3106530738828836e-07,
      "loss": 0.4088,
      "num_tokens": 2250778458.0,
      "step": 2375
    },
    {
      "epoch": 4.650342801175318,
      "grad_norm": 0.08534850115813652,
      "learning_rate": 1.2962734578973568e-07,
      "loss": 0.4254,
      "num_tokens": 2251720073.0,
      "step": 2376
    },
    {
      "epoch": 4.65230166503428,
      "grad_norm": 0.08661768596530278,
      "learning_rate": 1.2819721234364478e-07,
      "loss": 0.4202,
      "num_tokens": 2252622073.0,
      "step": 2377
    },
    {
      "epoch": 4.654260528893242,
      "grad_norm": 0.08497972478587038,
      "learning_rate": 1.2677490934867088e-07,
      "loss": 0.4124,
      "num_tokens": 2253561372.0,
      "step": 2378
    },
    {
      "epoch": 4.656219392752204,
      "grad_norm": 0.08358185604686996,
      "learning_rate": 1.253604390908819e-07,
      "loss": 0.4159,
      "num_tokens": 2254515196.0,
      "step": 2379
    },
    {
      "epoch": 4.658178256611166,
      "grad_norm": 0.08438690612513523,
      "learning_rate": 1.2395380384375654e-07,
      "loss": 0.4284,
      "num_tokens": 2255447919.0,
      "step": 2380
    },
    {
      "epoch": 4.660137120470127,
      "grad_norm": 0.08694154918463079,
      "learning_rate": 1.2255500586818015e-07,
      "loss": 0.4162,
      "num_tokens": 2256387955.0,
      "step": 2381
    },
    {
      "epoch": 4.662095984329089,
      "grad_norm": 0.0817943680700282,
      "learning_rate": 1.2116404741244204e-07,
      "loss": 0.4059,
      "num_tokens": 2257337818.0,
      "step": 2382
    },
    {
      "epoch": 4.664054848188051,
      "grad_norm": 0.08703181023937295,
      "learning_rate": 1.1978093071222995e-07,
      "loss": 0.4249,
      "num_tokens": 2258296586.0,
      "step": 2383
    },
    {
      "epoch": 4.666013712047013,
      "grad_norm": 0.08609012031456915,
      "learning_rate": 1.1840565799062843e-07,
      "loss": 0.4176,
      "num_tokens": 2259223903.0,
      "step": 2384
    },
    {
      "epoch": 4.667972575905974,
      "grad_norm": 0.08490192424131472,
      "learning_rate": 1.1703823145811422e-07,
      "loss": 0.4164,
      "num_tokens": 2260185068.0,
      "step": 2385
    },
    {
      "epoch": 4.6699314397649365,
      "grad_norm": 0.08470856216398916,
      "learning_rate": 1.1567865331255091e-07,
      "loss": 0.4233,
      "num_tokens": 2261120302.0,
      "step": 2386
    },
    {
      "epoch": 4.671890303623898,
      "grad_norm": 0.08384829499901271,
      "learning_rate": 1.1432692573919158e-07,
      "loss": 0.4372,
      "num_tokens": 2262131176.0,
      "step": 2387
    },
    {
      "epoch": 4.67384916748286,
      "grad_norm": 0.0868669309071135,
      "learning_rate": 1.1298305091066664e-07,
      "loss": 0.4049,
      "num_tokens": 2263079735.0,
      "step": 2388
    },
    {
      "epoch": 4.6758080313418215,
      "grad_norm": 0.083428730942915,
      "learning_rate": 1.1164703098698882e-07,
      "loss": 0.4097,
      "num_tokens": 2264024547.0,
      "step": 2389
    },
    {
      "epoch": 4.677766895200784,
      "grad_norm": 0.08185316461580272,
      "learning_rate": 1.1031886811554204e-07,
      "loss": 0.4084,
      "num_tokens": 2264994194.0,
      "step": 2390
    },
    {
      "epoch": 4.679725759059745,
      "grad_norm": 0.08459364324175946,
      "learning_rate": 1.0899856443108314e-07,
      "loss": 0.4031,
      "num_tokens": 2265923989.0,
      "step": 2391
    },
    {
      "epoch": 4.6816846229187075,
      "grad_norm": 0.08432846109620747,
      "learning_rate": 1.0768612205573791e-07,
      "loss": 0.4028,
      "num_tokens": 2266848216.0,
      "step": 2392
    },
    {
      "epoch": 4.683643486777669,
      "grad_norm": 0.0837381766860101,
      "learning_rate": 1.0638154309899562e-07,
      "loss": 0.4321,
      "num_tokens": 2267822961.0,
      "step": 2393
    },
    {
      "epoch": 4.685602350636631,
      "grad_norm": 0.08297454594420871,
      "learning_rate": 1.0508482965770506e-07,
      "loss": 0.4112,
      "num_tokens": 2268766814.0,
      "step": 2394
    },
    {
      "epoch": 4.6875612144955925,
      "grad_norm": 0.0880545235876429,
      "learning_rate": 1.0379598381607681e-07,
      "loss": 0.4264,
      "num_tokens": 2269698980.0,
      "step": 2395
    },
    {
      "epoch": 4.689520078354555,
      "grad_norm": 0.0828724427906624,
      "learning_rate": 1.0251500764567158e-07,
      "loss": 0.4116,
      "num_tokens": 2270630265.0,
      "step": 2396
    },
    {
      "epoch": 4.691478942213516,
      "grad_norm": 0.08548570246469113,
      "learning_rate": 1.012419032054035e-07,
      "loss": 0.4145,
      "num_tokens": 2271552753.0,
      "step": 2397
    },
    {
      "epoch": 4.6934378060724775,
      "grad_norm": 0.08550232441760104,
      "learning_rate": 9.997667254153464e-08,
      "loss": 0.4154,
      "num_tokens": 2272485419.0,
      "step": 2398
    },
    {
      "epoch": 4.69539666993144,
      "grad_norm": 0.0818702276118547,
      "learning_rate": 9.871931768767051e-08,
      "loss": 0.4239,
      "num_tokens": 2273461524.0,
      "step": 2399
    },
    {
      "epoch": 4.697355533790402,
      "grad_norm": 0.0831329454801517,
      "learning_rate": 9.746984066475728e-08,
      "loss": 0.4171,
      "num_tokens": 2274444871.0,
      "step": 2400
    },
    {
      "epoch": 4.699314397649363,
      "grad_norm": 0.08695654103675411,
      "learning_rate": 9.622824348108184e-08,
      "loss": 0.4365,
      "num_tokens": 2275392464.0,
      "step": 2401
    },
    {
      "epoch": 4.701273261508325,
      "grad_norm": 0.08276098441690038,
      "learning_rate": 9.499452813226284e-08,
      "loss": 0.4099,
      "num_tokens": 2276353262.0,
      "step": 2402
    },
    {
      "epoch": 4.703232125367287,
      "grad_norm": 0.08547505038028375,
      "learning_rate": 9.376869660125077e-08,
      "loss": 0.4178,
      "num_tokens": 2277284565.0,
      "step": 2403
    },
    {
      "epoch": 4.7051909892262485,
      "grad_norm": 0.08178310517058496,
      "learning_rate": 9.255075085832732e-08,
      "loss": 0.4089,
      "num_tokens": 2278245515.0,
      "step": 2404
    },
    {
      "epoch": 4.707149853085211,
      "grad_norm": 0.08547233860776729,
      "learning_rate": 9.134069286109604e-08,
      "loss": 0.421,
      "num_tokens": 2279198953.0,
      "step": 2405
    },
    {
      "epoch": 4.709108716944172,
      "grad_norm": 0.08344008488366243,
      "learning_rate": 9.013852455448335e-08,
      "loss": 0.4032,
      "num_tokens": 2280143860.0,
      "step": 2406
    },
    {
      "epoch": 4.711067580803134,
      "grad_norm": 0.08252842516593287,
      "learning_rate": 8.894424787073641e-08,
      "loss": 0.4172,
      "num_tokens": 2281116255.0,
      "step": 2407
    },
    {
      "epoch": 4.713026444662096,
      "grad_norm": 0.08361282068989813,
      "learning_rate": 8.775786472941528e-08,
      "loss": 0.4172,
      "num_tokens": 2282098453.0,
      "step": 2408
    },
    {
      "epoch": 4.714985308521058,
      "grad_norm": 0.0827717543188308,
      "learning_rate": 8.657937703739516e-08,
      "loss": 0.4299,
      "num_tokens": 2283055930.0,
      "step": 2409
    },
    {
      "epoch": 4.716944172380019,
      "grad_norm": 0.08230883377703196,
      "learning_rate": 8.540878668885977e-08,
      "loss": 0.4092,
      "num_tokens": 2284026427.0,
      "step": 2410
    },
    {
      "epoch": 4.718903036238982,
      "grad_norm": 0.09009461696717451,
      "learning_rate": 8.424609556529905e-08,
      "loss": 0.4155,
      "num_tokens": 2284956250.0,
      "step": 2411
    },
    {
      "epoch": 4.720861900097943,
      "grad_norm": 0.08096454719821111,
      "learning_rate": 8.309130553550815e-08,
      "loss": 0.404,
      "num_tokens": 2285916220.0,
      "step": 2412
    },
    {
      "epoch": 4.722820763956905,
      "grad_norm": 0.08291039517018342,
      "learning_rate": 8.19444184555801e-08,
      "loss": 0.4135,
      "num_tokens": 2286845304.0,
      "step": 2413
    },
    {
      "epoch": 4.724779627815867,
      "grad_norm": 0.0827014311360937,
      "learning_rate": 8.080543616890812e-08,
      "loss": 0.4009,
      "num_tokens": 2287812536.0,
      "step": 2414
    },
    {
      "epoch": 4.726738491674829,
      "grad_norm": 0.09025428355380949,
      "learning_rate": 7.967436050617893e-08,
      "loss": 0.4096,
      "num_tokens": 2288731826.0,
      "step": 2415
    },
    {
      "epoch": 4.72869735553379,
      "grad_norm": 0.0844809939820929,
      "learning_rate": 7.855119328537109e-08,
      "loss": 0.4264,
      "num_tokens": 2289698350.0,
      "step": 2416
    },
    {
      "epoch": 4.730656219392753,
      "grad_norm": 0.08579543628342719,
      "learning_rate": 7.743593631175106e-08,
      "loss": 0.4328,
      "num_tokens": 2290642899.0,
      "step": 2417
    },
    {
      "epoch": 4.732615083251714,
      "grad_norm": 0.08294522877285475,
      "learning_rate": 7.632859137787329e-08,
      "loss": 0.4104,
      "num_tokens": 2291575195.0,
      "step": 2418
    },
    {
      "epoch": 4.734573947110675,
      "grad_norm": 0.0838627781343406,
      "learning_rate": 7.522916026357297e-08,
      "loss": 0.4187,
      "num_tokens": 2292520282.0,
      "step": 2419
    },
    {
      "epoch": 4.736532810969638,
      "grad_norm": 0.08319567204494978,
      "learning_rate": 7.413764473596597e-08,
      "loss": 0.4165,
      "num_tokens": 2293445706.0,
      "step": 2420
    },
    {
      "epoch": 4.7384916748286,
      "grad_norm": 0.08012182889960814,
      "learning_rate": 7.30540465494467e-08,
      "loss": 0.3993,
      "num_tokens": 2294430713.0,
      "step": 2421
    },
    {
      "epoch": 4.740450538687561,
      "grad_norm": 0.08407500721762665,
      "learning_rate": 7.197836744568254e-08,
      "loss": 0.4075,
      "num_tokens": 2295408537.0,
      "step": 2422
    },
    {
      "epoch": 4.742409402546523,
      "grad_norm": 0.08365778938700633,
      "learning_rate": 7.09106091536127e-08,
      "loss": 0.4198,
      "num_tokens": 2296358794.0,
      "step": 2423
    },
    {
      "epoch": 4.744368266405485,
      "grad_norm": 0.08414292406344226,
      "learning_rate": 6.985077338944656e-08,
      "loss": 0.4173,
      "num_tokens": 2297315698.0,
      "step": 2424
    },
    {
      "epoch": 4.746327130264446,
      "grad_norm": 0.08603132400390481,
      "learning_rate": 6.879886185665818e-08,
      "loss": 0.4145,
      "num_tokens": 2298226505.0,
      "step": 2425
    },
    {
      "epoch": 4.748285994123409,
      "grad_norm": 0.08153954602319959,
      "learning_rate": 6.775487624598509e-08,
      "loss": 0.4012,
      "num_tokens": 2299201729.0,
      "step": 2426
    },
    {
      "epoch": 4.75024485798237,
      "grad_norm": 0.08304431063848801,
      "learning_rate": 6.67188182354267e-08,
      "loss": 0.4131,
      "num_tokens": 2300150837.0,
      "step": 2427
    },
    {
      "epoch": 4.752203721841332,
      "grad_norm": 0.08479816275786761,
      "learning_rate": 6.569068949023983e-08,
      "loss": 0.4105,
      "num_tokens": 2301073301.0,
      "step": 2428
    },
    {
      "epoch": 4.754162585700294,
      "grad_norm": 0.08213140338211498,
      "learning_rate": 6.467049166293483e-08,
      "loss": 0.4083,
      "num_tokens": 2302069644.0,
      "step": 2429
    },
    {
      "epoch": 4.756121449559256,
      "grad_norm": 0.08392926153014255,
      "learning_rate": 6.365822639327724e-08,
      "loss": 0.4025,
      "num_tokens": 2303024104.0,
      "step": 2430
    },
    {
      "epoch": 4.758080313418217,
      "grad_norm": 0.0844957988978168,
      "learning_rate": 6.26538953082817e-08,
      "loss": 0.4135,
      "num_tokens": 2303970182.0,
      "step": 2431
    },
    {
      "epoch": 4.7600391772771795,
      "grad_norm": 0.08504677682528261,
      "learning_rate": 6.165750002220916e-08,
      "loss": 0.4284,
      "num_tokens": 2304931055.0,
      "step": 2432
    },
    {
      "epoch": 4.761998041136141,
      "grad_norm": 0.08527646674148694,
      "learning_rate": 6.066904213656743e-08,
      "loss": 0.4101,
      "num_tokens": 2305861689.0,
      "step": 2433
    },
    {
      "epoch": 4.763956904995103,
      "grad_norm": 0.08725237255778938,
      "learning_rate": 5.968852324010454e-08,
      "loss": 0.4124,
      "num_tokens": 2306817684.0,
      "step": 2434
    },
    {
      "epoch": 4.7659157688540645,
      "grad_norm": 0.0824015101165882,
      "learning_rate": 5.8715944908809273e-08,
      "loss": 0.402,
      "num_tokens": 2307755058.0,
      "step": 2435
    },
    {
      "epoch": 4.767874632713027,
      "grad_norm": 0.08310793982906324,
      "learning_rate": 5.775130870590784e-08,
      "loss": 0.417,
      "num_tokens": 2308712133.0,
      "step": 2436
    },
    {
      "epoch": 4.769833496571988,
      "grad_norm": 0.08377938395008862,
      "learning_rate": 5.679461618185944e-08,
      "loss": 0.4167,
      "num_tokens": 2309670577.0,
      "step": 2437
    },
    {
      "epoch": 4.7717923604309505,
      "grad_norm": 0.08540182526090144,
      "learning_rate": 5.584586887435739e-08,
      "loss": 0.422,
      "num_tokens": 2310610225.0,
      "step": 2438
    },
    {
      "epoch": 4.773751224289912,
      "grad_norm": 0.08384391855385634,
      "learning_rate": 5.4905068308323536e-08,
      "loss": 0.4174,
      "num_tokens": 2311547266.0,
      "step": 2439
    },
    {
      "epoch": 4.775710088148873,
      "grad_norm": 0.08547930312731497,
      "learning_rate": 5.3972215995906076e-08,
      "loss": 0.4233,
      "num_tokens": 2312485067.0,
      "step": 2440
    },
    {
      "epoch": 4.7776689520078355,
      "grad_norm": 0.08399921760551363,
      "learning_rate": 5.3047313436480064e-08,
      "loss": 0.4145,
      "num_tokens": 2313432284.0,
      "step": 2441
    },
    {
      "epoch": 4.779627815866798,
      "grad_norm": 0.08206044911979275,
      "learning_rate": 5.213036211664191e-08,
      "loss": 0.4064,
      "num_tokens": 2314389898.0,
      "step": 2442
    },
    {
      "epoch": 4.781586679725759,
      "grad_norm": 0.08406160504043352,
      "learning_rate": 5.122136351020768e-08,
      "loss": 0.4017,
      "num_tokens": 2315315259.0,
      "step": 2443
    },
    {
      "epoch": 4.7835455435847205,
      "grad_norm": 0.08229692806506131,
      "learning_rate": 5.032031907821089e-08,
      "loss": 0.4073,
      "num_tokens": 2316294564.0,
      "step": 2444
    },
    {
      "epoch": 4.785504407443683,
      "grad_norm": 0.08599759770030403,
      "learning_rate": 4.942723026890139e-08,
      "loss": 0.4151,
      "num_tokens": 2317249683.0,
      "step": 2445
    },
    {
      "epoch": 4.787463271302644,
      "grad_norm": 0.08603985999117829,
      "learning_rate": 4.8542098517740945e-08,
      "loss": 0.4155,
      "num_tokens": 2318177699.0,
      "step": 2446
    },
    {
      "epoch": 4.789422135161606,
      "grad_norm": 0.08455414914704969,
      "learning_rate": 4.7664925247402074e-08,
      "loss": 0.4175,
      "num_tokens": 2319139001.0,
      "step": 2447
    },
    {
      "epoch": 4.791380999020568,
      "grad_norm": 0.08653806792097062,
      "learning_rate": 4.6795711867766436e-08,
      "loss": 0.4177,
      "num_tokens": 2320066572.0,
      "step": 2448
    },
    {
      "epoch": 4.79333986287953,
      "grad_norm": 0.0837413334352232,
      "learning_rate": 4.593445977592037e-08,
      "loss": 0.416,
      "num_tokens": 2321004533.0,
      "step": 2449
    },
    {
      "epoch": 4.7952987267384914,
      "grad_norm": 0.08591481470046484,
      "learning_rate": 4.5081170356156e-08,
      "loss": 0.4216,
      "num_tokens": 2321922365.0,
      "step": 2450
    },
    {
      "epoch": 4.797257590597454,
      "grad_norm": 0.08217979132395248,
      "learning_rate": 4.423584497996458e-08,
      "loss": 0.4087,
      "num_tokens": 2322892155.0,
      "step": 2451
    },
    {
      "epoch": 4.799216454456415,
      "grad_norm": 0.08410357083151844,
      "learning_rate": 4.339848500603816e-08,
      "loss": 0.3923,
      "num_tokens": 2323833595.0,
      "step": 2452
    },
    {
      "epoch": 4.801175318315377,
      "grad_norm": 0.08064405021012844,
      "learning_rate": 4.2569091780266245e-08,
      "loss": 0.4099,
      "num_tokens": 2324820336.0,
      "step": 2453
    },
    {
      "epoch": 4.803134182174339,
      "grad_norm": 0.08041802184099593,
      "learning_rate": 4.1747666635733594e-08,
      "loss": 0.4032,
      "num_tokens": 2325796312.0,
      "step": 2454
    },
    {
      "epoch": 4.805093046033301,
      "grad_norm": 0.08052566702220959,
      "learning_rate": 4.0934210892715765e-08,
      "loss": 0.4073,
      "num_tokens": 2326777747.0,
      "step": 2455
    },
    {
      "epoch": 4.807051909892262,
      "grad_norm": 0.08670948808326147,
      "learning_rate": 4.0128725858681884e-08,
      "loss": 0.4057,
      "num_tokens": 2327696411.0,
      "step": 2456
    },
    {
      "epoch": 4.809010773751225,
      "grad_norm": 0.08491674895829152,
      "learning_rate": 3.9331212828288e-08,
      "loss": 0.4243,
      "num_tokens": 2328622491.0,
      "step": 2457
    },
    {
      "epoch": 4.810969637610186,
      "grad_norm": 0.08322728782997706,
      "learning_rate": 3.8541673083377086e-08,
      "loss": 0.4024,
      "num_tokens": 2329553106.0,
      "step": 2458
    },
    {
      "epoch": 4.812928501469148,
      "grad_norm": 0.08174093923633531,
      "learning_rate": 3.776010789297735e-08,
      "loss": 0.4048,
      "num_tokens": 2330524979.0,
      "step": 2459
    },
    {
      "epoch": 4.81488736532811,
      "grad_norm": 0.08617383854563468,
      "learning_rate": 3.698651851329838e-08,
      "loss": 0.42,
      "num_tokens": 2331461473.0,
      "step": 2460
    },
    {
      "epoch": 4.816846229187071,
      "grad_norm": 0.08586259180754897,
      "learning_rate": 3.622090618773055e-08,
      "loss": 0.4211,
      "num_tokens": 2332400035.0,
      "step": 2461
    },
    {
      "epoch": 4.818805093046033,
      "grad_norm": 0.0838908968026586,
      "learning_rate": 3.546327214684342e-08,
      "loss": 0.4208,
      "num_tokens": 2333336979.0,
      "step": 2462
    },
    {
      "epoch": 4.820763956904996,
      "grad_norm": 0.08750160307910682,
      "learning_rate": 3.4713617608382324e-08,
      "loss": 0.4153,
      "num_tokens": 2334252618.0,
      "step": 2463
    },
    {
      "epoch": 4.822722820763957,
      "grad_norm": 0.08104017291582989,
      "learning_rate": 3.3971943777267866e-08,
      "loss": 0.4113,
      "num_tokens": 2335224929.0,
      "step": 2464
    },
    {
      "epoch": 4.824681684622918,
      "grad_norm": 0.08173541218523228,
      "learning_rate": 3.323825184559204e-08,
      "loss": 0.4079,
      "num_tokens": 2336185985.0,
      "step": 2465
    },
    {
      "epoch": 4.826640548481881,
      "grad_norm": 0.08300795903661226,
      "learning_rate": 3.251254299261875e-08,
      "loss": 0.4178,
      "num_tokens": 2337147503.0,
      "step": 2466
    },
    {
      "epoch": 4.828599412340842,
      "grad_norm": 0.08226662942561731,
      "learning_rate": 3.179481838477993e-08,
      "loss": 0.4053,
      "num_tokens": 2338153505.0,
      "step": 2467
    },
    {
      "epoch": 4.830558276199804,
      "grad_norm": 0.0848585446696675,
      "learning_rate": 3.108507917567505e-08,
      "loss": 0.4151,
      "num_tokens": 2339063128.0,
      "step": 2468
    },
    {
      "epoch": 4.832517140058766,
      "grad_norm": 0.08274741489835595,
      "learning_rate": 3.038332650606823e-08,
      "loss": 0.4069,
      "num_tokens": 2340015294.0,
      "step": 2469
    },
    {
      "epoch": 4.834476003917728,
      "grad_norm": 0.08290719552926558,
      "learning_rate": 2.9689561503886687e-08,
      "loss": 0.4069,
      "num_tokens": 2340986036.0,
      "step": 2470
    },
    {
      "epoch": 4.836434867776689,
      "grad_norm": 0.08749408764584028,
      "learning_rate": 2.9003785284218988e-08,
      "loss": 0.422,
      "num_tokens": 2341897246.0,
      "step": 2471
    },
    {
      "epoch": 4.838393731635652,
      "grad_norm": 0.08267423004795207,
      "learning_rate": 2.8325998949314536e-08,
      "loss": 0.4091,
      "num_tokens": 2342853731.0,
      "step": 2472
    },
    {
      "epoch": 4.840352595494613,
      "grad_norm": 0.08247784544045712,
      "learning_rate": 2.7656203588578566e-08,
      "loss": 0.4232,
      "num_tokens": 2343837974.0,
      "step": 2473
    },
    {
      "epoch": 4.842311459353575,
      "grad_norm": 0.0812281692364607,
      "learning_rate": 2.699440027857436e-08,
      "loss": 0.4074,
      "num_tokens": 2344803189.0,
      "step": 2474
    },
    {
      "epoch": 4.844270323212537,
      "grad_norm": 0.08427135044126924,
      "learning_rate": 2.6340590083018257e-08,
      "loss": 0.416,
      "num_tokens": 2345773157.0,
      "step": 2475
    },
    {
      "epoch": 4.846229187071499,
      "grad_norm": 0.08151715111615959,
      "learning_rate": 2.56947740527802e-08,
      "loss": 0.4202,
      "num_tokens": 2346741251.0,
      "step": 2476
    },
    {
      "epoch": 4.84818805093046,
      "grad_norm": 0.08897622249235938,
      "learning_rate": 2.505695322588042e-08,
      "loss": 0.4204,
      "num_tokens": 2347693106.0,
      "step": 2477
    },
    {
      "epoch": 4.8501469147894225,
      "grad_norm": 0.08461225830789412,
      "learning_rate": 2.442712862748775e-08,
      "loss": 0.4144,
      "num_tokens": 2348644307.0,
      "step": 2478
    },
    {
      "epoch": 4.852105778648384,
      "grad_norm": 0.08340120410325955,
      "learning_rate": 2.3805301269920754e-08,
      "loss": 0.4246,
      "num_tokens": 2349610681.0,
      "step": 2479
    },
    {
      "epoch": 4.854064642507346,
      "grad_norm": 0.0812242686859867,
      "learning_rate": 2.3191472152642726e-08,
      "loss": 0.4128,
      "num_tokens": 2350563195.0,
      "step": 2480
    },
    {
      "epoch": 4.8560235063663075,
      "grad_norm": 0.08351170707632818,
      "learning_rate": 2.2585642262260564e-08,
      "loss": 0.4143,
      "num_tokens": 2351510197.0,
      "step": 2481
    },
    {
      "epoch": 4.857982370225269,
      "grad_norm": 0.08392497822042437,
      "learning_rate": 2.1987812572526468e-08,
      "loss": 0.4245,
      "num_tokens": 2352457727.0,
      "step": 2482
    },
    {
      "epoch": 4.859941234084231,
      "grad_norm": 0.08431120254635958,
      "learning_rate": 2.1397984044331245e-08,
      "loss": 0.4261,
      "num_tokens": 2353403769.0,
      "step": 2483
    },
    {
      "epoch": 4.861900097943193,
      "grad_norm": 0.08552350387254533,
      "learning_rate": 2.0816157625706547e-08,
      "loss": 0.424,
      "num_tokens": 2354357230.0,
      "step": 2484
    },
    {
      "epoch": 4.863858961802155,
      "grad_norm": 0.0841283006789614,
      "learning_rate": 2.0242334251823204e-08,
      "loss": 0.4196,
      "num_tokens": 2355268335.0,
      "step": 2485
    },
    {
      "epoch": 4.865817825661116,
      "grad_norm": 0.08306796049433385,
      "learning_rate": 1.9676514844987338e-08,
      "loss": 0.4155,
      "num_tokens": 2356224375.0,
      "step": 2486
    },
    {
      "epoch": 4.8677766895200785,
      "grad_norm": 0.08321395206329488,
      "learning_rate": 1.911870031464036e-08,
      "loss": 0.4119,
      "num_tokens": 2357167198.0,
      "step": 2487
    },
    {
      "epoch": 4.86973555337904,
      "grad_norm": 0.08205135289045662,
      "learning_rate": 1.8568891557358413e-08,
      "loss": 0.4097,
      "num_tokens": 2358116578.0,
      "step": 2488
    },
    {
      "epoch": 4.871694417238002,
      "grad_norm": 0.08661824822951616,
      "learning_rate": 1.8027089456849055e-08,
      "loss": 0.4199,
      "num_tokens": 2359037560.0,
      "step": 2489
    },
    {
      "epoch": 4.8736532810969635,
      "grad_norm": 0.08594177887736316,
      "learning_rate": 1.749329488395124e-08,
      "loss": 0.4144,
      "num_tokens": 2359935151.0,
      "step": 2490
    },
    {
      "epoch": 4.875612144955926,
      "grad_norm": 0.08208582658409447,
      "learning_rate": 1.696750869663366e-08,
      "loss": 0.4114,
      "num_tokens": 2360896401.0,
      "step": 2491
    },
    {
      "epoch": 4.877571008814887,
      "grad_norm": 0.08144251104442016,
      "learning_rate": 1.6449731739991427e-08,
      "loss": 0.4018,
      "num_tokens": 2361843206.0,
      "step": 2492
    },
    {
      "epoch": 4.879529872673849,
      "grad_norm": 0.08323250894735935,
      "learning_rate": 1.593996484624938e-08,
      "loss": 0.418,
      "num_tokens": 2362816297.0,
      "step": 2493
    },
    {
      "epoch": 4.881488736532811,
      "grad_norm": 0.08389123523792696,
      "learning_rate": 1.5438208834754886e-08,
      "loss": 0.41,
      "num_tokens": 2363731796.0,
      "step": 2494
    },
    {
      "epoch": 4.883447600391773,
      "grad_norm": 0.08312659665297858,
      "learning_rate": 1.494446451198117e-08,
      "loss": 0.4108,
      "num_tokens": 2364729829.0,
      "step": 2495
    },
    {
      "epoch": 4.885406464250734,
      "grad_norm": 0.08208683931587697,
      "learning_rate": 1.4458732671523978e-08,
      "loss": 0.413,
      "num_tokens": 2365707631.0,
      "step": 2496
    },
    {
      "epoch": 4.887365328109697,
      "grad_norm": 0.08494231721901108,
      "learning_rate": 1.3981014094099354e-08,
      "loss": 0.4071,
      "num_tokens": 2366670586.0,
      "step": 2497
    },
    {
      "epoch": 4.889324191968658,
      "grad_norm": 0.0843058471808589,
      "learning_rate": 1.3511309547545315e-08,
      "loss": 0.4114,
      "num_tokens": 2367617080.0,
      "step": 2498
    },
    {
      "epoch": 4.89128305582762,
      "grad_norm": 0.08392439725095673,
      "learning_rate": 1.3049619786818512e-08,
      "loss": 0.4098,
      "num_tokens": 2368552710.0,
      "step": 2499
    },
    {
      "epoch": 4.893241919686582,
      "grad_norm": 0.08198489445557527,
      "learning_rate": 1.2595945553992572e-08,
      "loss": 0.4275,
      "num_tokens": 2369511071.0,
      "step": 2500
    },
    {
      "epoch": 4.895200783545544,
      "grad_norm": 0.08313640551389809,
      "learning_rate": 1.2150287578258646e-08,
      "loss": 0.4059,
      "num_tokens": 2370492381.0,
      "step": 2501
    },
    {
      "epoch": 4.897159647404505,
      "grad_norm": 0.08540445698858405,
      "learning_rate": 1.1712646575922637e-08,
      "loss": 0.4263,
      "num_tokens": 2371465884.0,
      "step": 2502
    },
    {
      "epoch": 4.899118511263467,
      "grad_norm": 0.08418418866658998,
      "learning_rate": 1.1283023250405756e-08,
      "loss": 0.4152,
      "num_tokens": 2372398758.0,
      "step": 2503
    },
    {
      "epoch": 4.901077375122429,
      "grad_norm": 0.0869585142660311,
      "learning_rate": 1.0861418292241188e-08,
      "loss": 0.4047,
      "num_tokens": 2373352246.0,
      "step": 2504
    },
    {
      "epoch": 4.903036238981391,
      "grad_norm": 0.08038143774441894,
      "learning_rate": 1.0447832379075761e-08,
      "loss": 0.4014,
      "num_tokens": 2374305208.0,
      "step": 2505
    },
    {
      "epoch": 4.904995102840353,
      "grad_norm": 0.08275289811279075,
      "learning_rate": 1.004226617566495e-08,
      "loss": 0.3972,
      "num_tokens": 2375248809.0,
      "step": 2506
    },
    {
      "epoch": 4.906953966699314,
      "grad_norm": 0.08178303547444857,
      "learning_rate": 9.6447203338762e-09,
      "loss": 0.3989,
      "num_tokens": 2376186884.0,
      "step": 2507
    },
    {
      "epoch": 4.908912830558276,
      "grad_norm": 0.08298956919319314,
      "learning_rate": 9.25519549268561e-09,
      "loss": 0.4182,
      "num_tokens": 2377153282.0,
      "step": 2508
    },
    {
      "epoch": 4.910871694417238,
      "grad_norm": 0.08168679032964059,
      "learning_rate": 8.873692278175693e-09,
      "loss": 0.3997,
      "num_tokens": 2378084196.0,
      "step": 2509
    },
    {
      "epoch": 4.9128305582762,
      "grad_norm": 0.08099350949175751,
      "learning_rate": 8.50021130353762e-09,
      "loss": 0.3998,
      "num_tokens": 2379064006.0,
      "step": 2510
    },
    {
      "epoch": 4.914789422135161,
      "grad_norm": 0.08448643905441347,
      "learning_rate": 8.134753169066756e-09,
      "loss": 0.4211,
      "num_tokens": 2380029423.0,
      "step": 2511
    },
    {
      "epoch": 4.916748285994124,
      "grad_norm": 0.08276948389547982,
      "learning_rate": 7.777318462164896e-09,
      "loss": 0.4025,
      "num_tokens": 2380955867.0,
      "step": 2512
    },
    {
      "epoch": 4.918707149853085,
      "grad_norm": 0.08353943599442933,
      "learning_rate": 7.427907757336927e-09,
      "loss": 0.412,
      "num_tokens": 2381873014.0,
      "step": 2513
    },
    {
      "epoch": 4.920666013712047,
      "grad_norm": 0.08724914833683893,
      "learning_rate": 7.0865216161902785e-09,
      "loss": 0.4237,
      "num_tokens": 2382796451.0,
      "step": 2514
    },
    {
      "epoch": 4.922624877571009,
      "grad_norm": 0.08241738671060188,
      "learning_rate": 6.75316058743547e-09,
      "loss": 0.4144,
      "num_tokens": 2383776478.0,
      "step": 2515
    },
    {
      "epoch": 4.924583741429971,
      "grad_norm": 0.08257681575010505,
      "learning_rate": 6.427825206883897e-09,
      "loss": 0.4087,
      "num_tokens": 2384725210.0,
      "step": 2516
    },
    {
      "epoch": 4.926542605288932,
      "grad_norm": 0.08247321302321747,
      "learning_rate": 6.110515997447275e-09,
      "loss": 0.4153,
      "num_tokens": 2385658811.0,
      "step": 2517
    },
    {
      "epoch": 4.9285014691478946,
      "grad_norm": 0.08527485644494695,
      "learning_rate": 5.80123346913708e-09,
      "loss": 0.4164,
      "num_tokens": 2386572645.0,
      "step": 2518
    },
    {
      "epoch": 4.930460333006856,
      "grad_norm": 0.0854930459702537,
      "learning_rate": 5.499978119062888e-09,
      "loss": 0.425,
      "num_tokens": 2387495071.0,
      "step": 2519
    },
    {
      "epoch": 4.932419196865818,
      "grad_norm": 0.08244657338826553,
      "learning_rate": 5.2067504314323725e-09,
      "loss": 0.4135,
      "num_tokens": 2388441374.0,
      "step": 2520
    },
    {
      "epoch": 4.93437806072478,
      "grad_norm": 0.08579930375525753,
      "learning_rate": 4.921550877550752e-09,
      "loss": 0.4094,
      "num_tokens": 2389388179.0,
      "step": 2521
    },
    {
      "epoch": 4.936336924583742,
      "grad_norm": 0.08239029943044571,
      "learning_rate": 4.644379915819674e-09,
      "loss": 0.4019,
      "num_tokens": 2390322786.0,
      "step": 2522
    },
    {
      "epoch": 4.938295788442703,
      "grad_norm": 0.08218263979258253,
      "learning_rate": 4.375237991736114e-09,
      "loss": 0.3947,
      "num_tokens": 2391252574.0,
      "step": 2523
    },
    {
      "epoch": 4.940254652301665,
      "grad_norm": 0.08460511024119019,
      "learning_rate": 4.114125537891811e-09,
      "loss": 0.4163,
      "num_tokens": 2392190407.0,
      "step": 2524
    },
    {
      "epoch": 4.942213516160627,
      "grad_norm": 0.08482161989770623,
      "learning_rate": 3.861042973973273e-09,
      "loss": 0.4197,
      "num_tokens": 2393133906.0,
      "step": 2525
    },
    {
      "epoch": 4.944172380019588,
      "grad_norm": 0.08367047310031385,
      "learning_rate": 3.6159907067601086e-09,
      "loss": 0.4013,
      "num_tokens": 2394067417.0,
      "step": 2526
    },
    {
      "epoch": 4.9461312438785505,
      "grad_norm": 0.08636739382428231,
      "learning_rate": 3.3789691301244766e-09,
      "loss": 0.4281,
      "num_tokens": 2395000779.0,
      "step": 2527
    },
    {
      "epoch": 4.948090107737512,
      "grad_norm": 0.08298122665582304,
      "learning_rate": 3.1499786250321904e-09,
      "loss": 0.417,
      "num_tokens": 2395987864.0,
      "step": 2528
    },
    {
      "epoch": 4.950048971596474,
      "grad_norm": 0.0872332731712304,
      "learning_rate": 2.9290195595388373e-09,
      "loss": 0.4263,
      "num_tokens": 2396948089.0,
      "step": 2529
    },
    {
      "epoch": 4.9520078354554355,
      "grad_norm": 0.08505589645696036,
      "learning_rate": 2.716092288792549e-09,
      "loss": 0.4138,
      "num_tokens": 2397876177.0,
      "step": 2530
    },
    {
      "epoch": 4.953966699314398,
      "grad_norm": 0.08201258823432063,
      "learning_rate": 2.511197155031231e-09,
      "loss": 0.4078,
      "num_tokens": 2398817796.0,
      "step": 2531
    },
    {
      "epoch": 4.955925563173359,
      "grad_norm": 0.0863739072094779,
      "learning_rate": 2.314334487583114e-09,
      "loss": 0.4171,
      "num_tokens": 2399760770.0,
      "step": 2532
    },
    {
      "epoch": 4.9578844270323215,
      "grad_norm": 0.08543826022418272,
      "learning_rate": 2.125504602866202e-09,
      "loss": 0.4112,
      "num_tokens": 2400685381.0,
      "step": 2533
    },
    {
      "epoch": 4.959843290891283,
      "grad_norm": 0.08517092811786608,
      "learning_rate": 1.944707804385493e-09,
      "loss": 0.4142,
      "num_tokens": 2401588105.0,
      "step": 2534
    },
    {
      "epoch": 4.961802154750245,
      "grad_norm": 0.08232903288303334,
      "learning_rate": 1.7719443827368677e-09,
      "loss": 0.4015,
      "num_tokens": 2402542987.0,
      "step": 2535
    },
    {
      "epoch": 4.9637610186092065,
      "grad_norm": 0.08489238637251854,
      "learning_rate": 1.6072146156032031e-09,
      "loss": 0.414,
      "num_tokens": 2403450808.0,
      "step": 2536
    },
    {
      "epoch": 4.965719882468169,
      "grad_norm": 0.08420075838921016,
      "learning_rate": 1.4505187677538169e-09,
      "loss": 0.4169,
      "num_tokens": 2404385191.0,
      "step": 2537
    },
    {
      "epoch": 4.96767874632713,
      "grad_norm": 0.08344347881043974,
      "learning_rate": 1.3018570910466876e-09,
      "loss": 0.4292,
      "num_tokens": 2405368652.0,
      "step": 2538
    },
    {
      "epoch": 4.969637610186092,
      "grad_norm": 0.084912657172186,
      "learning_rate": 1.1612298244256803e-09,
      "loss": 0.4253,
      "num_tokens": 2406298399.0,
      "step": 2539
    },
    {
      "epoch": 4.971596474045054,
      "grad_norm": 0.08283567224097792,
      "learning_rate": 1.0286371939205453e-09,
      "loss": 0.4155,
      "num_tokens": 2407276087.0,
      "step": 2540
    },
    {
      "epoch": 4.973555337904016,
      "grad_norm": 0.07901068391647909,
      "learning_rate": 9.040794126485841e-10,
      "loss": 0.4054,
      "num_tokens": 2408268769.0,
      "step": 2541
    },
    {
      "epoch": 4.975514201762977,
      "grad_norm": 0.08076771523672151,
      "learning_rate": 7.875566808107638e-10,
      "loss": 0.4196,
      "num_tokens": 2409263019.0,
      "step": 2542
    },
    {
      "epoch": 4.97747306562194,
      "grad_norm": 0.08504240350042797,
      "learning_rate": 6.790691856939369e-10,
      "loss": 0.418,
      "num_tokens": 2410207195.0,
      "step": 2543
    },
    {
      "epoch": 4.979431929480901,
      "grad_norm": 0.0836494206211676,
      "learning_rate": 5.78617101670842e-10,
      "loss": 0.4246,
      "num_tokens": 2411172709.0,
      "step": 2544
    },
    {
      "epoch": 4.9813907933398625,
      "grad_norm": 0.08370225143841217,
      "learning_rate": 4.862005901978828e-10,
      "loss": 0.4132,
      "num_tokens": 2412106660.0,
      "step": 2545
    },
    {
      "epoch": 4.983349657198825,
      "grad_norm": 0.08273121213895504,
      "learning_rate": 4.0181979981568364e-10,
      "loss": 0.4178,
      "num_tokens": 2413069391.0,
      "step": 2546
    },
    {
      "epoch": 4.985308521057786,
      "grad_norm": 0.08335935364884406,
      "learning_rate": 3.254748661507545e-10,
      "loss": 0.4155,
      "num_tokens": 2413994838.0,
      "step": 2547
    },
    {
      "epoch": 4.987267384916748,
      "grad_norm": 0.081987188110208,
      "learning_rate": 2.5716591191105034e-10,
      "loss": 0.4092,
      "num_tokens": 2414962822.0,
      "step": 2548
    },
    {
      "epoch": 4.98922624877571,
      "grad_norm": 0.08342080693098328,
      "learning_rate": 1.9689304688985667e-10,
      "loss": 0.4094,
      "num_tokens": 2415932912.0,
      "step": 2549
    },
    {
      "epoch": 4.991185112634672,
      "grad_norm": 0.08624338152957356,
      "learning_rate": 1.4465636796412442e-10,
      "loss": 0.4315,
      "num_tokens": 2416882851.0,
      "step": 2550
    },
    {
      "epoch": 4.993143976493633,
      "grad_norm": 0.0808935459746632,
      "learning_rate": 1.004559590939147e-10,
      "loss": 0.4055,
      "num_tokens": 2417834083.0,
      "step": 2551
    },
    {
      "epoch": 4.995102840352596,
      "grad_norm": 0.08319657490590793,
      "learning_rate": 6.429189132239888e-11,
      "loss": 0.4108,
      "num_tokens": 2418779500.0,
      "step": 2552
    },
    {
      "epoch": 4.997061704211557,
      "grad_norm": 0.08226287009167223,
      "learning_rate": 3.6164222776413626e-11,
      "loss": 0.401,
      "num_tokens": 2419700676.0,
      "step": 2553
    },
    {
      "epoch": 4.999020568070519,
      "grad_norm": 0.0838174095017364,
      "learning_rate": 1.6072998664795615e-11,
      "loss": 0.4207,
      "num_tokens": 2420659085.0,
      "step": 2554
    },
    {
      "epoch": 5.0,
      "grad_norm": 0.0838174095017364,
      "learning_rate": 4.018251280601959e-12,
      "loss": 0.4132,
      "num_tokens": 2421162947.0,
      "step": 2555
    },
    {
      "epoch": 5.0,
      "step": 2555,
      "total_flos": 2.4054064625943183e+18,
      "train_loss": 0.42931601253507656,
      "train_runtime": 13617.0429,
      "train_samples_per_second": 47.968,
      "train_steps_per_second": 0.188
    }
  ],
  "logging_steps": 1,
  "max_steps": 2555,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 5,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 2.4054064625943183e+18,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}