{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.33297859761144616,
  "eval_steps": 500,
  "global_step": 2816,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00011824524062906468,
      "grad_norm": 1.7525548934936523,
      "learning_rate": 0.0,
      "loss": 0.9062,
      "num_tokens": 628048.0,
      "step": 1
    },
    {
      "epoch": 0.00023649048125812936,
      "grad_norm": 1.6681532859802246,
      "learning_rate": 7.058823529411765e-07,
      "loss": 0.8713,
      "num_tokens": 1266689.0,
      "step": 2
    },
    {
      "epoch": 0.000354735721887194,
      "grad_norm": 1.6785393953323364,
      "learning_rate": 1.411764705882353e-06,
      "loss": 0.8748,
      "num_tokens": 1900338.0,
      "step": 3
    },
    {
      "epoch": 0.0004729809625162587,
      "grad_norm": 1.6376346349716187,
      "learning_rate": 2.1176470588235296e-06,
      "loss": 0.8704,
      "num_tokens": 2539095.0,
      "step": 4
    },
    {
      "epoch": 0.0005912262031453234,
      "grad_norm": 1.6173648834228516,
      "learning_rate": 2.823529411764706e-06,
      "loss": 0.8767,
      "num_tokens": 3172740.0,
      "step": 5
    },
    {
      "epoch": 0.000709471443774388,
      "grad_norm": 1.3518872261047363,
      "learning_rate": 3.5294117647058825e-06,
      "loss": 0.8425,
      "num_tokens": 3810090.0,
      "step": 6
    },
    {
      "epoch": 0.0008277166844034528,
      "grad_norm": 1.2358812093734741,
      "learning_rate": 4.235294117647059e-06,
      "loss": 0.9167,
      "num_tokens": 4447913.0,
      "step": 7
    },
    {
      "epoch": 0.0009459619250325174,
      "grad_norm": 1.0764226913452148,
      "learning_rate": 4.941176470588235e-06,
      "loss": 0.8203,
      "num_tokens": 5085253.0,
      "step": 8
    },
    {
      "epoch": 0.001064207165661582,
      "grad_norm": 1.1310309171676636,
      "learning_rate": 5.647058823529412e-06,
      "loss": 0.8602,
      "num_tokens": 5721930.0,
      "step": 9
    },
    {
      "epoch": 0.0011824524062906468,
      "grad_norm": 1.1147594451904297,
      "learning_rate": 6.352941176470589e-06,
      "loss": 0.8557,
      "num_tokens": 6354428.0,
      "step": 10
    },
    {
      "epoch": 0.0013006976469197116,
      "grad_norm": 0.9504016637802124,
      "learning_rate": 7.058823529411765e-06,
      "loss": 0.8223,
      "num_tokens": 6988898.0,
      "step": 11
    },
    {
      "epoch": 0.001418942887548776,
      "grad_norm": 1.2047721147537231,
      "learning_rate": 7.764705882352943e-06,
      "loss": 0.8035,
      "num_tokens": 7627058.0,
      "step": 12
    },
    {
      "epoch": 0.0015371881281778408,
      "grad_norm": 1.0793852806091309,
      "learning_rate": 8.470588235294118e-06,
      "loss": 0.7834,
      "num_tokens": 8261730.0,
      "step": 13
    },
    {
      "epoch": 0.0016554333688069056,
      "grad_norm": 0.9650391340255737,
      "learning_rate": 9.176470588235295e-06,
      "loss": 0.7852,
      "num_tokens": 8889249.0,
      "step": 14
    },
    {
      "epoch": 0.0017736786094359701,
      "grad_norm": 0.6715760231018066,
      "learning_rate": 9.88235294117647e-06,
      "loss": 0.7063,
      "num_tokens": 9524710.0,
      "step": 15
    },
    {
      "epoch": 0.0018919238500650349,
      "grad_norm": 0.8713334202766418,
      "learning_rate": 1.0588235294117648e-05,
      "loss": 0.7234,
      "num_tokens": 10161723.0,
      "step": 16
    },
    {
      "epoch": 0.0020101690906940994,
      "grad_norm": 0.8030041456222534,
      "learning_rate": 1.1294117647058823e-05,
      "loss": 0.717,
      "num_tokens": 10793622.0,
      "step": 17
    },
    {
      "epoch": 0.002128414331323164,
      "grad_norm": 0.7192631363868713,
      "learning_rate": 1.2e-05,
      "loss": 0.7109,
      "num_tokens": 11420435.0,
      "step": 18
    },
    {
      "epoch": 0.002246659571952229,
      "grad_norm": 0.6215784549713135,
      "learning_rate": 1.2705882352941177e-05,
      "loss": 0.7001,
      "num_tokens": 12055174.0,
      "step": 19
    },
    {
      "epoch": 0.0023649048125812936,
      "grad_norm": 0.5151238441467285,
      "learning_rate": 1.3411764705882354e-05,
      "loss": 0.6322,
      "num_tokens": 12687424.0,
      "step": 20
    },
    {
      "epoch": 0.0024831500532103584,
      "grad_norm": 0.6060124635696411,
      "learning_rate": 1.411764705882353e-05,
      "loss": 0.683,
      "num_tokens": 13318953.0,
      "step": 21
    },
    {
      "epoch": 0.002601395293839423,
      "grad_norm": 0.5640995502471924,
      "learning_rate": 1.4823529411764707e-05,
      "loss": 0.6415,
      "num_tokens": 13937259.0,
      "step": 22
    },
    {
      "epoch": 0.0027196405344684874,
      "grad_norm": 0.5616605877876282,
      "learning_rate": 1.5529411764705886e-05,
      "loss": 0.6736,
      "num_tokens": 14569053.0,
      "step": 23
    },
    {
      "epoch": 0.002837885775097552,
      "grad_norm": 0.5228525400161743,
      "learning_rate": 1.623529411764706e-05,
      "loss": 0.6779,
      "num_tokens": 15199901.0,
      "step": 24
    },
    {
      "epoch": 0.002956131015726617,
      "grad_norm": 0.5150514841079712,
      "learning_rate": 1.6941176470588237e-05,
      "loss": 0.6548,
      "num_tokens": 15830276.0,
      "step": 25
    },
    {
      "epoch": 0.0030743762563556817,
      "grad_norm": 0.45277050137519836,
      "learning_rate": 1.7647058823529414e-05,
      "loss": 0.6527,
      "num_tokens": 16465223.0,
      "step": 26
    },
    {
      "epoch": 0.0031926214969847464,
      "grad_norm": 0.4658268094062805,
      "learning_rate": 1.835294117647059e-05,
      "loss": 0.6367,
      "num_tokens": 17102309.0,
      "step": 27
    },
    {
      "epoch": 0.003310866737613811,
      "grad_norm": 0.48463839292526245,
      "learning_rate": 1.9058823529411764e-05,
      "loss": 0.6367,
      "num_tokens": 17738787.0,
      "step": 28
    },
    {
      "epoch": 0.003429111978242876,
      "grad_norm": 0.4104596674442291,
      "learning_rate": 1.976470588235294e-05,
      "loss": 0.5729,
      "num_tokens": 18370503.0,
      "step": 29
    },
    {
      "epoch": 0.0035473572188719402,
      "grad_norm": 0.4189141094684601,
      "learning_rate": 2.047058823529412e-05,
      "loss": 0.6082,
      "num_tokens": 19008700.0,
      "step": 30
    },
    {
      "epoch": 0.003665602459501005,
      "grad_norm": 0.4681914448738098,
      "learning_rate": 2.1176470588235296e-05,
      "loss": 0.6145,
      "num_tokens": 19641076.0,
      "step": 31
    },
    {
      "epoch": 0.0037838477001300697,
      "grad_norm": 0.36613571643829346,
      "learning_rate": 2.188235294117647e-05,
      "loss": 0.5343,
      "num_tokens": 20268470.0,
      "step": 32
    },
    {
      "epoch": 0.0039020929407591345,
      "grad_norm": 0.37985900044441223,
      "learning_rate": 2.2588235294117646e-05,
      "loss": 0.6184,
      "num_tokens": 20903791.0,
      "step": 33
    },
    {
      "epoch": 0.004020338181388199,
      "grad_norm": 0.4060671031475067,
      "learning_rate": 2.3294117647058824e-05,
      "loss": 0.5974,
      "num_tokens": 21537889.0,
      "step": 34
    },
    {
      "epoch": 0.0041385834220172635,
      "grad_norm": 0.36260876059532166,
      "learning_rate": 2.4e-05,
      "loss": 0.5276,
      "num_tokens": 22169163.0,
      "step": 35
    },
    {
      "epoch": 0.004256828662646328,
      "grad_norm": 0.37660202383995056,
      "learning_rate": 2.4705882352941174e-05,
      "loss": 0.5824,
      "num_tokens": 22802993.0,
      "step": 36
    },
    {
      "epoch": 0.004375073903275393,
      "grad_norm": 0.3913825750350952,
      "learning_rate": 2.5411764705882355e-05,
      "loss": 0.5597,
      "num_tokens": 23436910.0,
      "step": 37
    },
    {
      "epoch": 0.004493319143904458,
      "grad_norm": 0.4073478579521179,
      "learning_rate": 2.6117647058823532e-05,
      "loss": 0.5951,
      "num_tokens": 24066237.0,
      "step": 38
    },
    {
      "epoch": 0.0046115643845335225,
      "grad_norm": 0.36139920353889465,
      "learning_rate": 2.682352941176471e-05,
      "loss": 0.5452,
      "num_tokens": 24702070.0,
      "step": 39
    },
    {
      "epoch": 0.004729809625162587,
      "grad_norm": 0.3636697232723236,
      "learning_rate": 2.7529411764705883e-05,
      "loss": 0.5321,
      "num_tokens": 25335554.0,
      "step": 40
    },
    {
      "epoch": 0.004848054865791652,
      "grad_norm": 0.3299705684185028,
      "learning_rate": 2.823529411764706e-05,
      "loss": 0.4893,
      "num_tokens": 25967347.0,
      "step": 41
    },
    {
      "epoch": 0.004966300106420717,
      "grad_norm": 0.36068227887153625,
      "learning_rate": 2.8941176470588237e-05,
      "loss": 0.5031,
      "num_tokens": 26599914.0,
      "step": 42
    },
    {
      "epoch": 0.0050845453470497815,
      "grad_norm": 0.429426908493042,
      "learning_rate": 2.9647058823529414e-05,
      "loss": 0.5608,
      "num_tokens": 27231412.0,
      "step": 43
    },
    {
      "epoch": 0.005202790587678846,
      "grad_norm": 0.40092170238494873,
      "learning_rate": 3.0352941176470588e-05,
      "loss": 0.5662,
      "num_tokens": 27868949.0,
      "step": 44
    },
    {
      "epoch": 0.005321035828307911,
      "grad_norm": 0.4001767635345459,
      "learning_rate": 3.105882352941177e-05,
      "loss": 0.5389,
      "num_tokens": 28505673.0,
      "step": 45
    },
    {
      "epoch": 0.005439281068936975,
      "grad_norm": 0.3943917155265808,
      "learning_rate": 3.1764705882352945e-05,
      "loss": 0.5483,
      "num_tokens": 29143956.0,
      "step": 46
    },
    {
      "epoch": 0.00555752630956604,
      "grad_norm": 0.3981671929359436,
      "learning_rate": 3.247058823529412e-05,
      "loss": 0.5405,
      "num_tokens": 29780377.0,
      "step": 47
    },
    {
      "epoch": 0.005675771550195104,
      "grad_norm": 0.4339058995246887,
      "learning_rate": 3.31764705882353e-05,
      "loss": 0.5444,
      "num_tokens": 30419988.0,
      "step": 48
    },
    {
      "epoch": 0.005794016790824169,
      "grad_norm": 0.46332406997680664,
      "learning_rate": 3.388235294117647e-05,
      "loss": 0.5356,
      "num_tokens": 31058641.0,
      "step": 49
    },
    {
      "epoch": 0.005912262031453234,
      "grad_norm": 0.3915022015571594,
      "learning_rate": 3.458823529411765e-05,
      "loss": 0.537,
      "num_tokens": 31696113.0,
      "step": 50
    },
    {
      "epoch": 0.006030507272082299,
      "grad_norm": 0.4514997601509094,
      "learning_rate": 3.529411764705883e-05,
      "loss": 0.5701,
      "num_tokens": 32335026.0,
      "step": 51
    },
    {
      "epoch": 0.006148752512711363,
      "grad_norm": 0.5017653107643127,
      "learning_rate": 3.6e-05,
      "loss": 0.4915,
      "num_tokens": 32968147.0,
      "step": 52
    },
    {
      "epoch": 0.006266997753340428,
      "grad_norm": 0.4031103253364563,
      "learning_rate": 3.670588235294118e-05,
      "loss": 0.5266,
      "num_tokens": 33599687.0,
      "step": 53
    },
    {
      "epoch": 0.006385242993969493,
      "grad_norm": 0.5270867943763733,
      "learning_rate": 3.7411764705882355e-05,
      "loss": 0.4989,
      "num_tokens": 34235952.0,
      "step": 54
    },
    {
      "epoch": 0.006503488234598558,
      "grad_norm": 0.3970652222633362,
      "learning_rate": 3.811764705882353e-05,
      "loss": 0.5222,
      "num_tokens": 34855614.0,
      "step": 55
    },
    {
      "epoch": 0.006621733475227622,
      "grad_norm": 0.5039134621620178,
      "learning_rate": 3.882352941176471e-05,
      "loss": 0.5615,
      "num_tokens": 35490773.0,
      "step": 56
    },
    {
      "epoch": 0.006739978715856687,
      "grad_norm": 0.42443519830703735,
      "learning_rate": 3.952941176470588e-05,
      "loss": 0.5441,
      "num_tokens": 36118336.0,
      "step": 57
    },
    {
      "epoch": 0.006858223956485752,
      "grad_norm": 0.4183923900127411,
      "learning_rate": 4.023529411764706e-05,
      "loss": 0.5106,
      "num_tokens": 36751911.0,
      "step": 58
    },
    {
      "epoch": 0.0069764691971148166,
      "grad_norm": 0.40575334429740906,
      "learning_rate": 4.094117647058824e-05,
      "loss": 0.504,
      "num_tokens": 37375413.0,
      "step": 59
    },
    {
      "epoch": 0.0070947144377438804,
      "grad_norm": 0.4154472351074219,
      "learning_rate": 4.164705882352941e-05,
      "loss": 0.4789,
      "num_tokens": 38006888.0,
      "step": 60
    },
    {
      "epoch": 0.007212959678372945,
      "grad_norm": 0.4931453764438629,
      "learning_rate": 4.235294117647059e-05,
      "loss": 0.4956,
      "num_tokens": 38640497.0,
      "step": 61
    },
    {
      "epoch": 0.00733120491900201,
      "grad_norm": 0.4350408911705017,
      "learning_rate": 4.3058823529411765e-05,
      "loss": 0.482,
      "num_tokens": 39273313.0,
      "step": 62
    },
    {
      "epoch": 0.007449450159631075,
      "grad_norm": 0.5501086115837097,
      "learning_rate": 4.376470588235294e-05,
      "loss": 0.6147,
      "num_tokens": 39910620.0,
      "step": 63
    },
    {
      "epoch": 0.007567695400260139,
      "grad_norm": 0.643365740776062,
      "learning_rate": 4.447058823529412e-05,
      "loss": 0.4939,
      "num_tokens": 40546868.0,
      "step": 64
    },
    {
      "epoch": 0.007685940640889204,
      "grad_norm": 0.4394073188304901,
      "learning_rate": 4.517647058823529e-05,
      "loss": 0.4614,
      "num_tokens": 41182518.0,
      "step": 65
    },
    {
      "epoch": 0.007804185881518269,
      "grad_norm": 0.5629273056983948,
      "learning_rate": 4.5882352941176467e-05,
      "loss": 0.5081,
      "num_tokens": 41817083.0,
      "step": 66
    },
    {
      "epoch": 0.007922431122147333,
      "grad_norm": 0.6570345163345337,
      "learning_rate": 4.658823529411765e-05,
      "loss": 0.5063,
      "num_tokens": 42454047.0,
      "step": 67
    },
    {
      "epoch": 0.008040676362776398,
      "grad_norm": 0.44039052724838257,
      "learning_rate": 4.729411764705882e-05,
      "loss": 0.4853,
      "num_tokens": 43066339.0,
      "step": 68
    },
    {
      "epoch": 0.008158921603405462,
      "grad_norm": 0.6671131253242493,
      "learning_rate": 4.8e-05,
      "loss": 0.532,
      "num_tokens": 43703254.0,
      "step": 69
    },
    {
      "epoch": 0.008277166844034527,
      "grad_norm": 0.4714736044406891,
      "learning_rate": 4.8705882352941175e-05,
      "loss": 0.5248,
      "num_tokens": 44339105.0,
      "step": 70
    },
    {
      "epoch": 0.008395412084663592,
      "grad_norm": 0.43229085206985474,
      "learning_rate": 4.941176470588235e-05,
      "loss": 0.4623,
      "num_tokens": 44969251.0,
      "step": 71
    },
    {
      "epoch": 0.008513657325292657,
      "grad_norm": 0.5161452889442444,
      "learning_rate": 5.0117647058823536e-05,
      "loss": 0.5103,
      "num_tokens": 45605957.0,
      "step": 72
    },
    {
      "epoch": 0.008631902565921721,
      "grad_norm": 0.39117467403411865,
      "learning_rate": 5.082352941176471e-05,
      "loss": 0.4618,
      "num_tokens": 46242216.0,
      "step": 73
    },
    {
      "epoch": 0.008750147806550786,
      "grad_norm": 0.5116973519325256,
      "learning_rate": 5.152941176470588e-05,
      "loss": 0.5111,
      "num_tokens": 46879736.0,
      "step": 74
    },
    {
      "epoch": 0.00886839304717985,
      "grad_norm": 0.42920881509780884,
      "learning_rate": 5.2235294117647064e-05,
      "loss": 0.4705,
      "num_tokens": 47508876.0,
      "step": 75
    },
    {
      "epoch": 0.008986638287808916,
      "grad_norm": 0.4628167748451233,
      "learning_rate": 5.294117647058824e-05,
      "loss": 0.4731,
      "num_tokens": 48148110.0,
      "step": 76
    },
    {
      "epoch": 0.00910488352843798,
      "grad_norm": 0.4489562511444092,
      "learning_rate": 5.364705882352942e-05,
      "loss": 0.4551,
      "num_tokens": 48779111.0,
      "step": 77
    },
    {
      "epoch": 0.009223128769067045,
      "grad_norm": 0.4363901615142822,
      "learning_rate": 5.435294117647059e-05,
      "loss": 0.4606,
      "num_tokens": 49414520.0,
      "step": 78
    },
    {
      "epoch": 0.00934137400969611,
      "grad_norm": 0.5536901354789734,
      "learning_rate": 5.5058823529411765e-05,
      "loss": 0.4838,
      "num_tokens": 50047364.0,
      "step": 79
    },
    {
      "epoch": 0.009459619250325174,
      "grad_norm": 0.39150169491767883,
      "learning_rate": 5.5764705882352946e-05,
      "loss": 0.4495,
      "num_tokens": 50679909.0,
      "step": 80
    },
    {
      "epoch": 0.00957786449095424,
      "grad_norm": 0.6313601136207581,
      "learning_rate": 5.647058823529412e-05,
      "loss": 0.4621,
      "num_tokens": 51313449.0,
      "step": 81
    },
    {
      "epoch": 0.009696109731583304,
      "grad_norm": 0.4643058776855469,
      "learning_rate": 5.717647058823529e-05,
      "loss": 0.4763,
      "num_tokens": 51949695.0,
      "step": 82
    },
    {
      "epoch": 0.009814354972212369,
      "grad_norm": 0.5276260375976562,
      "learning_rate": 5.7882352941176474e-05,
      "loss": 0.488,
      "num_tokens": 52582348.0,
      "step": 83
    },
    {
      "epoch": 0.009932600212841433,
      "grad_norm": 0.47664791345596313,
      "learning_rate": 5.858823529411765e-05,
      "loss": 0.4893,
      "num_tokens": 53211571.0,
      "step": 84
    },
    {
      "epoch": 0.010050845453470498,
      "grad_norm": 0.4026750922203064,
      "learning_rate": 5.929411764705883e-05,
      "loss": 0.4828,
      "num_tokens": 53846179.0,
      "step": 85
    },
    {
      "epoch": 0.010169090694099563,
      "grad_norm": 0.4486851096153259,
      "learning_rate": 6e-05,
      "loss": 0.4724,
      "num_tokens": 54478244.0,
      "step": 86
    },
    {
      "epoch": 0.010287335934728628,
      "grad_norm": 0.4775157570838928,
      "learning_rate": 5.99999821355329e-05,
      "loss": 0.4917,
      "num_tokens": 55113336.0,
      "step": 87
    },
    {
      "epoch": 0.010405581175357692,
      "grad_norm": 0.38246527314186096,
      "learning_rate": 5.9999928542155226e-05,
      "loss": 0.4554,
      "num_tokens": 55750901.0,
      "step": 88
    },
    {
      "epoch": 0.010523826415986757,
      "grad_norm": 0.516433596611023,
      "learning_rate": 5.999983921993793e-05,
      "loss": 0.5338,
      "num_tokens": 56387320.0,
      "step": 89
    },
    {
      "epoch": 0.010642071656615822,
      "grad_norm": 0.39624902606010437,
      "learning_rate": 5.999971416899916e-05,
      "loss": 0.4585,
      "num_tokens": 57023503.0,
      "step": 90
    },
    {
      "epoch": 0.010760316897244887,
      "grad_norm": 0.48605141043663025,
      "learning_rate": 5.999955338950444e-05,
      "loss": 0.5225,
      "num_tokens": 57651196.0,
      "step": 91
    },
    {
      "epoch": 0.01087856213787395,
      "grad_norm": 0.48312604427337646,
      "learning_rate": 5.9999356881666505e-05,
      "loss": 0.478,
      "num_tokens": 58290170.0,
      "step": 92
    },
    {
      "epoch": 0.010996807378503014,
      "grad_norm": 0.39719170331954956,
      "learning_rate": 5.9999124645745405e-05,
      "loss": 0.444,
      "num_tokens": 58929585.0,
      "step": 93
    },
    {
      "epoch": 0.01111505261913208,
      "grad_norm": 0.39644670486450195,
      "learning_rate": 5.999885668204845e-05,
      "loss": 0.4479,
      "num_tokens": 59569034.0,
      "step": 94
    },
    {
      "epoch": 0.011233297859761144,
      "grad_norm": 0.3955964148044586,
      "learning_rate": 5.999855299093024e-05,
      "loss": 0.4797,
      "num_tokens": 60197509.0,
      "step": 95
    },
    {
      "epoch": 0.011351543100390209,
      "grad_norm": 0.3650868535041809,
      "learning_rate": 5.9998213572792634e-05,
      "loss": 0.4547,
      "num_tokens": 60830564.0,
      "step": 96
    },
    {
      "epoch": 0.011469788341019273,
      "grad_norm": 0.47804784774780273,
      "learning_rate": 5.9997838428084794e-05,
      "loss": 0.4838,
      "num_tokens": 61466570.0,
      "step": 97
    },
    {
      "epoch": 0.011588033581648338,
      "grad_norm": 0.3899470269680023,
      "learning_rate": 5.999742755730315e-05,
      "loss": 0.4589,
      "num_tokens": 62100797.0,
      "step": 98
    },
    {
      "epoch": 0.011706278822277403,
      "grad_norm": 0.36873653531074524,
      "learning_rate": 5.99969809609914e-05,
      "loss": 0.4666,
      "num_tokens": 62734274.0,
      "step": 99
    },
    {
      "epoch": 0.011824524062906468,
      "grad_norm": 0.4157959222793579,
      "learning_rate": 5.9996498639740514e-05,
      "loss": 0.4896,
      "num_tokens": 63350637.0,
      "step": 100
    },
    {
      "epoch": 0.011942769303535532,
      "grad_norm": 0.3493950366973877,
      "learning_rate": 5.999598059418875e-05,
      "loss": 0.4667,
      "num_tokens": 63989248.0,
      "step": 101
    },
    {
      "epoch": 0.012061014544164597,
      "grad_norm": 0.38584789633750916,
      "learning_rate": 5.999542682502165e-05,
      "loss": 0.4768,
      "num_tokens": 64623402.0,
      "step": 102
    },
    {
      "epoch": 0.012179259784793662,
      "grad_norm": 0.36720627546310425,
      "learning_rate": 5.999483733297199e-05,
      "loss": 0.4301,
      "num_tokens": 65263084.0,
      "step": 103
    },
    {
      "epoch": 0.012297505025422727,
      "grad_norm": 0.3988032341003418,
      "learning_rate": 5.999421211881985e-05,
      "loss": 0.4946,
      "num_tokens": 65895544.0,
      "step": 104
    },
    {
      "epoch": 0.012415750266051791,
      "grad_norm": 0.4363256096839905,
      "learning_rate": 5.9993551183392574e-05,
      "loss": 0.4513,
      "num_tokens": 66526292.0,
      "step": 105
    },
    {
      "epoch": 0.012533995506680856,
      "grad_norm": 0.38990506529808044,
      "learning_rate": 5.9992854527564775e-05,
      "loss": 0.4651,
      "num_tokens": 67163008.0,
      "step": 106
    },
    {
      "epoch": 0.012652240747309921,
      "grad_norm": 0.37964877486228943,
      "learning_rate": 5.999212215225833e-05,
      "loss": 0.4597,
      "num_tokens": 67799173.0,
      "step": 107
    },
    {
      "epoch": 0.012770485987938986,
      "grad_norm": 0.368556410074234,
      "learning_rate": 5.999135405844239e-05,
      "loss": 0.4414,
      "num_tokens": 68433626.0,
      "step": 108
    },
    {
      "epoch": 0.01288873122856805,
      "grad_norm": 0.4006548225879669,
      "learning_rate": 5.9990550247133366e-05,
      "loss": 0.4724,
      "num_tokens": 69071597.0,
      "step": 109
    },
    {
      "epoch": 0.013006976469197115,
      "grad_norm": 0.4608422815799713,
      "learning_rate": 5.9989710719394935e-05,
      "loss": 0.4951,
      "num_tokens": 69708272.0,
      "step": 110
    },
    {
      "epoch": 0.01312522170982618,
      "grad_norm": 0.3924156427383423,
      "learning_rate": 5.9988835476338044e-05,
      "loss": 0.4799,
      "num_tokens": 70309426.0,
      "step": 111
    },
    {
      "epoch": 0.013243466950455245,
      "grad_norm": 0.5403560996055603,
      "learning_rate": 5.998792451912089e-05,
      "loss": 0.5011,
      "num_tokens": 70946347.0,
      "step": 112
    },
    {
      "epoch": 0.01336171219108431,
      "grad_norm": 0.3970711827278137,
      "learning_rate": 5.998697784894894e-05,
      "loss": 0.4399,
      "num_tokens": 71578069.0,
      "step": 113
    },
    {
      "epoch": 0.013479957431713374,
      "grad_norm": 0.37565523386001587,
      "learning_rate": 5.9985995467074924e-05,
      "loss": 0.4449,
      "num_tokens": 72216733.0,
      "step": 114
    },
    {
      "epoch": 0.013598202672342439,
      "grad_norm": 0.4285765290260315,
      "learning_rate": 5.998497737479881e-05,
      "loss": 0.4691,
      "num_tokens": 72844928.0,
      "step": 115
    },
    {
      "epoch": 0.013716447912971504,
      "grad_norm": 0.4407060444355011,
      "learning_rate": 5.998392357346784e-05,
      "loss": 0.4696,
      "num_tokens": 73482837.0,
      "step": 116
    },
    {
      "epoch": 0.013834693153600568,
      "grad_norm": 0.3529186546802521,
      "learning_rate": 5.998283406447651e-05,
      "loss": 0.4928,
      "num_tokens": 74118196.0,
      "step": 117
    },
    {
      "epoch": 0.013952938394229633,
      "grad_norm": 0.3553130328655243,
      "learning_rate": 5.998170884926654e-05,
      "loss": 0.436,
      "num_tokens": 74756174.0,
      "step": 118
    },
    {
      "epoch": 0.014071183634858696,
      "grad_norm": 0.34798574447631836,
      "learning_rate": 5.998054792932695e-05,
      "loss": 0.4969,
      "num_tokens": 75393367.0,
      "step": 119
    },
    {
      "epoch": 0.014189428875487761,
      "grad_norm": 0.3467991054058075,
      "learning_rate": 5.997935130619394e-05,
      "loss": 0.4438,
      "num_tokens": 76025875.0,
      "step": 120
    },
    {
      "epoch": 0.014307674116116826,
      "grad_norm": 0.3396257758140564,
      "learning_rate": 5.997811898145103e-05,
      "loss": 0.4239,
      "num_tokens": 76657710.0,
      "step": 121
    },
    {
      "epoch": 0.01442591935674589,
      "grad_norm": 0.3784518837928772,
      "learning_rate": 5.997685095672892e-05,
      "loss": 0.4644,
      "num_tokens": 77289318.0,
      "step": 122
    },
    {
      "epoch": 0.014544164597374955,
      "grad_norm": 0.4056454002857208,
      "learning_rate": 5.997554723370561e-05,
      "loss": 0.4493,
      "num_tokens": 77923482.0,
      "step": 123
    },
    {
      "epoch": 0.01466240983800402,
      "grad_norm": 0.39920276403427124,
      "learning_rate": 5.997420781410627e-05,
      "loss": 0.4695,
      "num_tokens": 78551959.0,
      "step": 124
    },
    {
      "epoch": 0.014780655078633085,
      "grad_norm": 0.3787338137626648,
      "learning_rate": 5.997283269970338e-05,
      "loss": 0.4583,
      "num_tokens": 79181459.0,
      "step": 125
    },
    {
      "epoch": 0.01489890031926215,
      "grad_norm": 0.37710845470428467,
      "learning_rate": 5.99714218923166e-05,
      "loss": 0.4281,
      "num_tokens": 79816398.0,
      "step": 126
    },
    {
      "epoch": 0.015017145559891214,
      "grad_norm": 0.3817121386528015,
      "learning_rate": 5.996997539381286e-05,
      "loss": 0.462,
      "num_tokens": 80451198.0,
      "step": 127
    },
    {
      "epoch": 0.015135390800520279,
      "grad_norm": 0.360310822725296,
      "learning_rate": 5.996849320610628e-05,
      "loss": 0.4697,
      "num_tokens": 81087693.0,
      "step": 128
    },
    {
      "epoch": 0.015253636041149344,
      "grad_norm": 0.3428126573562622,
      "learning_rate": 5.996697533115825e-05,
      "loss": 0.4215,
      "num_tokens": 81722590.0,
      "step": 129
    },
    {
      "epoch": 0.015371881281778408,
      "grad_norm": 0.4063381850719452,
      "learning_rate": 5.9965421770977355e-05,
      "loss": 0.5232,
      "num_tokens": 82358511.0,
      "step": 130
    },
    {
      "epoch": 0.015490126522407473,
      "grad_norm": 0.3583608567714691,
      "learning_rate": 5.996383252761942e-05,
      "loss": 0.4426,
      "num_tokens": 82997292.0,
      "step": 131
    },
    {
      "epoch": 0.015608371763036538,
      "grad_norm": 0.3704170882701874,
      "learning_rate": 5.996220760318747e-05,
      "loss": 0.4344,
      "num_tokens": 83636002.0,
      "step": 132
    },
    {
      "epoch": 0.0157266170036656,
      "grad_norm": 0.33110740780830383,
      "learning_rate": 5.9960546999831764e-05,
      "loss": 0.4562,
      "num_tokens": 84266256.0,
      "step": 133
    },
    {
      "epoch": 0.015844862244294666,
      "grad_norm": 0.374468594789505,
      "learning_rate": 5.9958850719749756e-05,
      "loss": 0.4743,
      "num_tokens": 84901903.0,
      "step": 134
    },
    {
      "epoch": 0.01596310748492373,
      "grad_norm": 0.33168625831604004,
      "learning_rate": 5.995711876518614e-05,
      "loss": 0.4431,
      "num_tokens": 85532297.0,
      "step": 135
    },
    {
      "epoch": 0.016081352725552795,
      "grad_norm": 0.3238625228404999,
      "learning_rate": 5.9955351138432795e-05,
      "loss": 0.397,
      "num_tokens": 86159004.0,
      "step": 136
    },
    {
      "epoch": 0.01619959796618186,
      "grad_norm": 0.3387395441532135,
      "learning_rate": 5.995354784182881e-05,
      "loss": 0.4474,
      "num_tokens": 86791668.0,
      "step": 137
    },
    {
      "epoch": 0.016317843206810925,
      "grad_norm": 0.36932268738746643,
      "learning_rate": 5.995170887776047e-05,
      "loss": 0.4581,
      "num_tokens": 87415500.0,
      "step": 138
    },
    {
      "epoch": 0.01643608844743999,
      "grad_norm": 0.347771555185318,
      "learning_rate": 5.994983424866128e-05,
      "loss": 0.4258,
      "num_tokens": 88054250.0,
      "step": 139
    },
    {
      "epoch": 0.016554333688069054,
      "grad_norm": 0.3125162720680237,
      "learning_rate": 5.994792395701191e-05,
      "loss": 0.4077,
      "num_tokens": 88683640.0,
      "step": 140
    },
    {
      "epoch": 0.01667257892869812,
      "grad_norm": 0.36889344453811646,
      "learning_rate": 5.994597800534024e-05,
      "loss": 0.4097,
      "num_tokens": 89322994.0,
      "step": 141
    },
    {
      "epoch": 0.016790824169327184,
      "grad_norm": 0.3739296495914459,
      "learning_rate": 5.9943996396221355e-05,
      "loss": 0.4557,
      "num_tokens": 89959870.0,
      "step": 142
    },
    {
      "epoch": 0.01690906940995625,
      "grad_norm": 0.4370843172073364,
      "learning_rate": 5.9941979132277475e-05,
      "loss": 0.4581,
      "num_tokens": 90595155.0,
      "step": 143
    },
    {
      "epoch": 0.017027314650585313,
      "grad_norm": 0.3677273690700531,
      "learning_rate": 5.993992621617805e-05,
      "loss": 0.4606,
      "num_tokens": 91229434.0,
      "step": 144
    },
    {
      "epoch": 0.017145559891214378,
      "grad_norm": 0.39628374576568604,
      "learning_rate": 5.993783765063969e-05,
      "loss": 0.4316,
      "num_tokens": 91864729.0,
      "step": 145
    },
    {
      "epoch": 0.017263805131843443,
      "grad_norm": 0.326407790184021,
      "learning_rate": 5.993571343842618e-05,
      "loss": 0.4114,
      "num_tokens": 92497696.0,
      "step": 146
    },
    {
      "epoch": 0.017382050372472507,
      "grad_norm": 0.4197523891925812,
      "learning_rate": 5.993355358234848e-05,
      "loss": 0.4856,
      "num_tokens": 93136961.0,
      "step": 147
    },
    {
      "epoch": 0.017500295613101572,
      "grad_norm": 0.36535730957984924,
      "learning_rate": 5.9931358085264714e-05,
      "loss": 0.4677,
      "num_tokens": 93763389.0,
      "step": 148
    },
    {
      "epoch": 0.017618540853730637,
      "grad_norm": 0.3775778114795685,
      "learning_rate": 5.992912695008017e-05,
      "loss": 0.4677,
      "num_tokens": 94396681.0,
      "step": 149
    },
    {
      "epoch": 0.0177367860943597,
      "grad_norm": 0.3371557295322418,
      "learning_rate": 5.992686017974728e-05,
      "loss": 0.4695,
      "num_tokens": 95024975.0,
      "step": 150
    },
    {
      "epoch": 0.017855031334988766,
      "grad_norm": 0.3853801190853119,
      "learning_rate": 5.992455777726568e-05,
      "loss": 0.4755,
      "num_tokens": 95647926.0,
      "step": 151
    },
    {
      "epoch": 0.01797327657561783,
      "grad_norm": 0.35996532440185547,
      "learning_rate": 5.9922219745682086e-05,
      "loss": 0.4435,
      "num_tokens": 96284873.0,
      "step": 152
    },
    {
      "epoch": 0.018091521816246896,
      "grad_norm": 0.3817720413208008,
      "learning_rate": 5.991984608809043e-05,
      "loss": 0.4561,
      "num_tokens": 96921347.0,
      "step": 153
    },
    {
      "epoch": 0.01820976705687596,
      "grad_norm": 0.35337361693382263,
      "learning_rate": 5.991743680763175e-05,
      "loss": 0.4072,
      "num_tokens": 97555490.0,
      "step": 154
    },
    {
      "epoch": 0.018328012297505025,
      "grad_norm": 0.32107430696487427,
      "learning_rate": 5.9914991907494234e-05,
      "loss": 0.422,
      "num_tokens": 98193055.0,
      "step": 155
    },
    {
      "epoch": 0.01844625753813409,
      "grad_norm": 0.40560588240623474,
      "learning_rate": 5.991251139091319e-05,
      "loss": 0.402,
      "num_tokens": 98826897.0,
      "step": 156
    },
    {
      "epoch": 0.018564502778763155,
      "grad_norm": 0.36023592948913574,
      "learning_rate": 5.990999526117108e-05,
      "loss": 0.4751,
      "num_tokens": 99462013.0,
      "step": 157
    },
    {
      "epoch": 0.01868274801939222,
      "grad_norm": 0.3359028100967407,
      "learning_rate": 5.9907443521597494e-05,
      "loss": 0.4851,
      "num_tokens": 100097628.0,
      "step": 158
    },
    {
      "epoch": 0.018800993260021284,
      "grad_norm": 0.34057164192199707,
      "learning_rate": 5.990485617556912e-05,
      "loss": 0.4222,
      "num_tokens": 100734556.0,
      "step": 159
    },
    {
      "epoch": 0.01891923850065035,
      "grad_norm": 0.3408139646053314,
      "learning_rate": 5.990223322650977e-05,
      "loss": 0.4679,
      "num_tokens": 101371427.0,
      "step": 160
    },
    {
      "epoch": 0.019037483741279414,
      "grad_norm": 0.3269839584827423,
      "learning_rate": 5.98995746778904e-05,
      "loss": 0.4716,
      "num_tokens": 102008014.0,
      "step": 161
    },
    {
      "epoch": 0.01915572898190848,
      "grad_norm": 0.3538208305835724,
      "learning_rate": 5.989688053322904e-05,
      "loss": 0.4351,
      "num_tokens": 102647195.0,
      "step": 162
    },
    {
      "epoch": 0.019273974222537543,
      "grad_norm": 0.30472224950790405,
      "learning_rate": 5.989415079609082e-05,
      "loss": 0.4485,
      "num_tokens": 103272117.0,
      "step": 163
    },
    {
      "epoch": 0.019392219463166608,
      "grad_norm": 0.30045244097709656,
      "learning_rate": 5.989138547008799e-05,
      "loss": 0.4532,
      "num_tokens": 103911617.0,
      "step": 164
    },
    {
      "epoch": 0.019510464703795673,
      "grad_norm": 0.33233657479286194,
      "learning_rate": 5.988858455887992e-05,
      "loss": 0.465,
      "num_tokens": 104546025.0,
      "step": 165
    },
    {
      "epoch": 0.019628709944424737,
      "grad_norm": 0.3113420307636261,
      "learning_rate": 5.9885748066172985e-05,
      "loss": 0.4252,
      "num_tokens": 105181492.0,
      "step": 166
    },
    {
      "epoch": 0.019746955185053802,
      "grad_norm": 0.31303688883781433,
      "learning_rate": 5.988287599572075e-05,
      "loss": 0.4236,
      "num_tokens": 105818707.0,
      "step": 167
    },
    {
      "epoch": 0.019865200425682867,
      "grad_norm": 0.27217867970466614,
      "learning_rate": 5.987996835132378e-05,
      "loss": 0.3983,
      "num_tokens": 106447672.0,
      "step": 168
    },
    {
      "epoch": 0.01998344566631193,
      "grad_norm": 0.3025364875793457,
      "learning_rate": 5.9877025136829754e-05,
      "loss": 0.4365,
      "num_tokens": 107079280.0,
      "step": 169
    },
    {
      "epoch": 0.020101690906940996,
      "grad_norm": 0.3207264840602875,
      "learning_rate": 5.987404635613339e-05,
      "loss": 0.4392,
      "num_tokens": 107716692.0,
      "step": 170
    },
    {
      "epoch": 0.02021993614757006,
      "grad_norm": 0.3323531746864319,
      "learning_rate": 5.9871032013176516e-05,
      "loss": 0.4369,
      "num_tokens": 108346190.0,
      "step": 171
    },
    {
      "epoch": 0.020338181388199126,
      "grad_norm": 0.35194694995880127,
      "learning_rate": 5.986798211194798e-05,
      "loss": 0.4255,
      "num_tokens": 108972655.0,
      "step": 172
    },
    {
      "epoch": 0.02045642662882819,
      "grad_norm": 0.375436395406723,
      "learning_rate": 5.98648966564837e-05,
      "loss": 0.4504,
      "num_tokens": 109606533.0,
      "step": 173
    },
    {
      "epoch": 0.020574671869457255,
      "grad_norm": 0.29956355690956116,
      "learning_rate": 5.986177565086665e-05,
      "loss": 0.4354,
      "num_tokens": 110242667.0,
      "step": 174
    },
    {
      "epoch": 0.02069291711008632,
      "grad_norm": 0.43081754446029663,
      "learning_rate": 5.985861909922682e-05,
      "loss": 0.4901,
      "num_tokens": 110879342.0,
      "step": 175
    },
    {
      "epoch": 0.020811162350715385,
      "grad_norm": 0.3560563921928406,
      "learning_rate": 5.985542700574125e-05,
      "loss": 0.4409,
      "num_tokens": 111515578.0,
      "step": 176
    },
    {
      "epoch": 0.02092940759134445,
      "grad_norm": 0.33471059799194336,
      "learning_rate": 5.985219937463405e-05,
      "loss": 0.4004,
      "num_tokens": 112148074.0,
      "step": 177
    },
    {
      "epoch": 0.021047652831973514,
      "grad_norm": 0.36091309785842896,
      "learning_rate": 5.98489362101763e-05,
      "loss": 0.4887,
      "num_tokens": 112784825.0,
      "step": 178
    },
    {
      "epoch": 0.02116589807260258,
      "grad_norm": 0.3456615209579468,
      "learning_rate": 5.984563751668612e-05,
      "loss": 0.4636,
      "num_tokens": 113415511.0,
      "step": 179
    },
    {
      "epoch": 0.021284143313231644,
      "grad_norm": 0.3597208559513092,
      "learning_rate": 5.984230329852868e-05,
      "loss": 0.4634,
      "num_tokens": 114014233.0,
      "step": 180
    },
    {
      "epoch": 0.02140238855386071,
      "grad_norm": 0.3545994162559509,
      "learning_rate": 5.98389335601161e-05,
      "loss": 0.4449,
      "num_tokens": 114646091.0,
      "step": 181
    },
    {
      "epoch": 0.021520633794489773,
      "grad_norm": 0.35723739862442017,
      "learning_rate": 5.9835528305907545e-05,
      "loss": 0.4196,
      "num_tokens": 115248849.0,
      "step": 182
    },
    {
      "epoch": 0.021638879035118838,
      "grad_norm": 0.3535923659801483,
      "learning_rate": 5.983208754040917e-05,
      "loss": 0.4149,
      "num_tokens": 115882858.0,
      "step": 183
    },
    {
      "epoch": 0.0217571242757479,
      "grad_norm": 0.3566373884677887,
      "learning_rate": 5.982861126817412e-05,
      "loss": 0.4492,
      "num_tokens": 116516321.0,
      "step": 184
    },
    {
      "epoch": 0.021875369516376964,
      "grad_norm": 0.32124945521354675,
      "learning_rate": 5.9825099493802524e-05,
      "loss": 0.425,
      "num_tokens": 117149980.0,
      "step": 185
    },
    {
      "epoch": 0.02199361475700603,
      "grad_norm": 0.34338271617889404,
      "learning_rate": 5.982155222194149e-05,
      "loss": 0.4097,
      "num_tokens": 117787055.0,
      "step": 186
    },
    {
      "epoch": 0.022111859997635094,
      "grad_norm": 0.37264284491539,
      "learning_rate": 5.98179694572851e-05,
      "loss": 0.4326,
      "num_tokens": 118414231.0,
      "step": 187
    },
    {
      "epoch": 0.02223010523826416,
      "grad_norm": 0.3190022110939026,
      "learning_rate": 5.981435120457439e-05,
      "loss": 0.4481,
      "num_tokens": 119042723.0,
      "step": 188
    },
    {
      "epoch": 0.022348350478893223,
      "grad_norm": 0.32043132185935974,
      "learning_rate": 5.98106974685974e-05,
      "loss": 0.476,
      "num_tokens": 119680244.0,
      "step": 189
    },
    {
      "epoch": 0.022466595719522288,
      "grad_norm": 0.42416223883628845,
      "learning_rate": 5.980700825418908e-05,
      "loss": 0.4332,
      "num_tokens": 120319137.0,
      "step": 190
    },
    {
      "epoch": 0.022584840960151353,
      "grad_norm": 0.33873337507247925,
      "learning_rate": 5.9803283566231345e-05,
      "loss": 0.4425,
      "num_tokens": 120954282.0,
      "step": 191
    },
    {
      "epoch": 0.022703086200780417,
      "grad_norm": 0.4517790675163269,
      "learning_rate": 5.979952340965306e-05,
      "loss": 0.4417,
      "num_tokens": 121589180.0,
      "step": 192
    },
    {
      "epoch": 0.022821331441409482,
      "grad_norm": 0.3599424958229065,
      "learning_rate": 5.9795727789430005e-05,
      "loss": 0.4758,
      "num_tokens": 122226045.0,
      "step": 193
    },
    {
      "epoch": 0.022939576682038547,
      "grad_norm": 0.4032820761203766,
      "learning_rate": 5.979189671058491e-05,
      "loss": 0.417,
      "num_tokens": 122863565.0,
      "step": 194
    },
    {
      "epoch": 0.02305782192266761,
      "grad_norm": 0.3690061867237091,
      "learning_rate": 5.9788030178187414e-05,
      "loss": 0.4114,
      "num_tokens": 123498631.0,
      "step": 195
    },
    {
      "epoch": 0.023176067163296676,
      "grad_norm": 0.3435264527797699,
      "learning_rate": 5.978412819735408e-05,
      "loss": 0.4407,
      "num_tokens": 124134670.0,
      "step": 196
    },
    {
      "epoch": 0.02329431240392574,
      "grad_norm": 0.33528485894203186,
      "learning_rate": 5.978019077324837e-05,
      "loss": 0.4188,
      "num_tokens": 124770499.0,
      "step": 197
    },
    {
      "epoch": 0.023412557644554806,
      "grad_norm": 0.3990873694419861,
      "learning_rate": 5.9776217911080664e-05,
      "loss": 0.4723,
      "num_tokens": 125409234.0,
      "step": 198
    },
    {
      "epoch": 0.02353080288518387,
      "grad_norm": 0.3296034038066864,
      "learning_rate": 5.9772209616108206e-05,
      "loss": 0.4415,
      "num_tokens": 126041885.0,
      "step": 199
    },
    {
      "epoch": 0.023649048125812935,
      "grad_norm": 0.36493048071861267,
      "learning_rate": 5.976816589363516e-05,
      "loss": 0.4607,
      "num_tokens": 126679673.0,
      "step": 200
    },
    {
      "epoch": 0.023767293366442,
      "grad_norm": 0.29360583424568176,
      "learning_rate": 5.976408674901256e-05,
      "loss": 0.4532,
      "num_tokens": 127318553.0,
      "step": 201
    },
    {
      "epoch": 0.023885538607071065,
      "grad_norm": 0.33180269598960876,
      "learning_rate": 5.9759972187638324e-05,
      "loss": 0.4232,
      "num_tokens": 127957565.0,
      "step": 202
    },
    {
      "epoch": 0.02400378384770013,
      "grad_norm": 0.3688262403011322,
      "learning_rate": 5.97558222149572e-05,
      "loss": 0.5016,
      "num_tokens": 128588826.0,
      "step": 203
    },
    {
      "epoch": 0.024122029088329194,
      "grad_norm": 0.31288033723831177,
      "learning_rate": 5.9751636836460835e-05,
      "loss": 0.4493,
      "num_tokens": 129222173.0,
      "step": 204
    },
    {
      "epoch": 0.02424027432895826,
      "grad_norm": 0.327677458524704,
      "learning_rate": 5.974741605768772e-05,
      "loss": 0.4646,
      "num_tokens": 129857385.0,
      "step": 205
    },
    {
      "epoch": 0.024358519569587324,
      "grad_norm": 0.2778308391571045,
      "learning_rate": 5.974315988422318e-05,
      "loss": 0.3875,
      "num_tokens": 130486516.0,
      "step": 206
    },
    {
      "epoch": 0.02447676481021639,
      "grad_norm": 0.2977498769760132,
      "learning_rate": 5.973886832169938e-05,
      "loss": 0.4597,
      "num_tokens": 131117306.0,
      "step": 207
    },
    {
      "epoch": 0.024595010050845453,
      "grad_norm": 0.2994617521762848,
      "learning_rate": 5.9734541375795316e-05,
      "loss": 0.4803,
      "num_tokens": 131749433.0,
      "step": 208
    },
    {
      "epoch": 0.024713255291474518,
      "grad_norm": 0.278374046087265,
      "learning_rate": 5.973017905223682e-05,
      "loss": 0.4307,
      "num_tokens": 132382579.0,
      "step": 209
    },
    {
      "epoch": 0.024831500532103583,
      "grad_norm": 0.31233716011047363,
      "learning_rate": 5.9725781356796527e-05,
      "loss": 0.4442,
      "num_tokens": 133017061.0,
      "step": 210
    },
    {
      "epoch": 0.024949745772732648,
      "grad_norm": 0.2855742871761322,
      "learning_rate": 5.972134829529387e-05,
      "loss": 0.435,
      "num_tokens": 133645920.0,
      "step": 211
    },
    {
      "epoch": 0.025067991013361712,
      "grad_norm": 0.278287410736084,
      "learning_rate": 5.971687987359511e-05,
      "loss": 0.4745,
      "num_tokens": 134277976.0,
      "step": 212
    },
    {
      "epoch": 0.025186236253990777,
      "grad_norm": 0.2984735667705536,
      "learning_rate": 5.971237609761326e-05,
      "loss": 0.4529,
      "num_tokens": 134913386.0,
      "step": 213
    },
    {
      "epoch": 0.025304481494619842,
      "grad_norm": 0.2837817370891571,
      "learning_rate": 5.970783697330815e-05,
      "loss": 0.438,
      "num_tokens": 135544399.0,
      "step": 214
    },
    {
      "epoch": 0.025422726735248907,
      "grad_norm": 0.30570748448371887,
      "learning_rate": 5.970326250668638e-05,
      "loss": 0.4475,
      "num_tokens": 136182707.0,
      "step": 215
    },
    {
      "epoch": 0.02554097197587797,
      "grad_norm": 0.2581757605075836,
      "learning_rate": 5.9698652703801296e-05,
      "loss": 0.4424,
      "num_tokens": 136819793.0,
      "step": 216
    },
    {
      "epoch": 0.025659217216507036,
      "grad_norm": 0.33199775218963623,
      "learning_rate": 5.969400757075305e-05,
      "loss": 0.4476,
      "num_tokens": 137459203.0,
      "step": 217
    },
    {
      "epoch": 0.0257774624571361,
      "grad_norm": 0.291856586933136,
      "learning_rate": 5.968932711368848e-05,
      "loss": 0.4408,
      "num_tokens": 138086108.0,
      "step": 218
    },
    {
      "epoch": 0.025895707697765166,
      "grad_norm": 0.26909223198890686,
      "learning_rate": 5.9684611338801244e-05,
      "loss": 0.4308,
      "num_tokens": 138717361.0,
      "step": 219
    },
    {
      "epoch": 0.02601395293839423,
      "grad_norm": 0.35612818598747253,
      "learning_rate": 5.967986025233169e-05,
      "loss": 0.4144,
      "num_tokens": 139353768.0,
      "step": 220
    },
    {
      "epoch": 0.026132198179023295,
      "grad_norm": 0.3378983736038208,
      "learning_rate": 5.967507386056688e-05,
      "loss": 0.4774,
      "num_tokens": 139986012.0,
      "step": 221
    },
    {
      "epoch": 0.02625044341965236,
      "grad_norm": 0.3200989365577698,
      "learning_rate": 5.9670252169840635e-05,
      "loss": 0.4127,
      "num_tokens": 140618557.0,
      "step": 222
    },
    {
      "epoch": 0.026368688660281425,
      "grad_norm": 0.31645864248275757,
      "learning_rate": 5.966539518653346e-05,
      "loss": 0.4037,
      "num_tokens": 141253328.0,
      "step": 223
    },
    {
      "epoch": 0.02648693390091049,
      "grad_norm": 0.30323508381843567,
      "learning_rate": 5.966050291707257e-05,
      "loss": 0.442,
      "num_tokens": 141885327.0,
      "step": 224
    },
    {
      "epoch": 0.026605179141539554,
      "grad_norm": 0.36218684911727905,
      "learning_rate": 5.965557536793189e-05,
      "loss": 0.4558,
      "num_tokens": 142518760.0,
      "step": 225
    },
    {
      "epoch": 0.02672342438216862,
      "grad_norm": 0.3109579384326935,
      "learning_rate": 5.965061254563201e-05,
      "loss": 0.462,
      "num_tokens": 143154500.0,
      "step": 226
    },
    {
      "epoch": 0.026841669622797684,
      "grad_norm": 0.29776236414909363,
      "learning_rate": 5.96456144567402e-05,
      "loss": 0.4467,
      "num_tokens": 143792062.0,
      "step": 227
    },
    {
      "epoch": 0.026959914863426748,
      "grad_norm": 0.298713743686676,
      "learning_rate": 5.964058110787039e-05,
      "loss": 0.3959,
      "num_tokens": 144424810.0,
      "step": 228
    },
    {
      "epoch": 0.027078160104055813,
      "grad_norm": 0.30076488852500916,
      "learning_rate": 5.963551250568319e-05,
      "loss": 0.4207,
      "num_tokens": 145021445.0,
      "step": 229
    },
    {
      "epoch": 0.027196405344684878,
      "grad_norm": 0.2877078652381897,
      "learning_rate": 5.963040865688585e-05,
      "loss": 0.4733,
      "num_tokens": 145656556.0,
      "step": 230
    },
    {
      "epoch": 0.027314650585313942,
      "grad_norm": 0.3390825092792511,
      "learning_rate": 5.962526956823225e-05,
      "loss": 0.4197,
      "num_tokens": 146293020.0,
      "step": 231
    },
    {
      "epoch": 0.027432895825943007,
      "grad_norm": 0.3341982364654541,
      "learning_rate": 5.962009524652293e-05,
      "loss": 0.4406,
      "num_tokens": 146922200.0,
      "step": 232
    },
    {
      "epoch": 0.027551141066572072,
      "grad_norm": 0.3136695921421051,
      "learning_rate": 5.961488569860504e-05,
      "loss": 0.4847,
      "num_tokens": 147560483.0,
      "step": 233
    },
    {
      "epoch": 0.027669386307201137,
      "grad_norm": 0.3606899380683899,
      "learning_rate": 5.9609640931372325e-05,
      "loss": 0.4268,
      "num_tokens": 148198475.0,
      "step": 234
    },
    {
      "epoch": 0.0277876315478302,
      "grad_norm": 0.28975990414619446,
      "learning_rate": 5.9604360951765175e-05,
      "loss": 0.4426,
      "num_tokens": 148811827.0,
      "step": 235
    },
    {
      "epoch": 0.027905876788459266,
      "grad_norm": 0.302625834941864,
      "learning_rate": 5.959904576677054e-05,
      "loss": 0.411,
      "num_tokens": 149446588.0,
      "step": 236
    },
    {
      "epoch": 0.028024122029088328,
      "grad_norm": 0.26446977257728577,
      "learning_rate": 5.959369538342198e-05,
      "loss": 0.3921,
      "num_tokens": 150080880.0,
      "step": 237
    },
    {
      "epoch": 0.028142367269717392,
      "grad_norm": 0.28426593542099,
      "learning_rate": 5.9588309808799606e-05,
      "loss": 0.462,
      "num_tokens": 150720517.0,
      "step": 238
    },
    {
      "epoch": 0.028260612510346457,
      "grad_norm": 0.31598517298698425,
      "learning_rate": 5.958288905003014e-05,
      "loss": 0.4303,
      "num_tokens": 151354865.0,
      "step": 239
    },
    {
      "epoch": 0.028378857750975522,
      "grad_norm": 0.2635610103607178,
      "learning_rate": 5.957743311428682e-05,
      "loss": 0.3943,
      "num_tokens": 151989302.0,
      "step": 240
    },
    {
      "epoch": 0.028497102991604586,
      "grad_norm": 0.3268015682697296,
      "learning_rate": 5.957194200878948e-05,
      "loss": 0.432,
      "num_tokens": 152622644.0,
      "step": 241
    },
    {
      "epoch": 0.02861534823223365,
      "grad_norm": 0.26707592606544495,
      "learning_rate": 5.9566415740804425e-05,
      "loss": 0.4668,
      "num_tokens": 153254978.0,
      "step": 242
    },
    {
      "epoch": 0.028733593472862716,
      "grad_norm": 0.2767883837223053,
      "learning_rate": 5.956085431764456e-05,
      "loss": 0.4062,
      "num_tokens": 153888439.0,
      "step": 243
    },
    {
      "epoch": 0.02885183871349178,
      "grad_norm": 0.252451628446579,
      "learning_rate": 5.955525774666927e-05,
      "loss": 0.3823,
      "num_tokens": 154521234.0,
      "step": 244
    },
    {
      "epoch": 0.028970083954120845,
      "grad_norm": 0.2643362283706665,
      "learning_rate": 5.954962603528448e-05,
      "loss": 0.4316,
      "num_tokens": 155156546.0,
      "step": 245
    },
    {
      "epoch": 0.02908832919474991,
      "grad_norm": 0.2746330499649048,
      "learning_rate": 5.954395919094257e-05,
      "loss": 0.4321,
      "num_tokens": 155791974.0,
      "step": 246
    },
    {
      "epoch": 0.029206574435378975,
      "grad_norm": 0.24371886253356934,
      "learning_rate": 5.953825722114247e-05,
      "loss": 0.3914,
      "num_tokens": 156417737.0,
      "step": 247
    },
    {
      "epoch": 0.02932481967600804,
      "grad_norm": 0.2923094928264618,
      "learning_rate": 5.953252013342955e-05,
      "loss": 0.4618,
      "num_tokens": 157054324.0,
      "step": 248
    },
    {
      "epoch": 0.029443064916637104,
      "grad_norm": 0.27547353506088257,
      "learning_rate": 5.9526747935395664e-05,
      "loss": 0.4488,
      "num_tokens": 157686875.0,
      "step": 249
    },
    {
      "epoch": 0.02956131015726617,
      "grad_norm": 0.31762856245040894,
      "learning_rate": 5.9520940634679115e-05,
      "loss": 0.4432,
      "num_tokens": 158318798.0,
      "step": 250
    },
    {
      "epoch": 0.029679555397895234,
      "grad_norm": 0.2715708911418915,
      "learning_rate": 5.951509823896469e-05,
      "loss": 0.457,
      "num_tokens": 158956587.0,
      "step": 251
    },
    {
      "epoch": 0.0297978006385243,
      "grad_norm": 0.2930499017238617,
      "learning_rate": 5.950922075598357e-05,
      "loss": 0.3783,
      "num_tokens": 159591859.0,
      "step": 252
    },
    {
      "epoch": 0.029916045879153363,
      "grad_norm": 0.29419347643852234,
      "learning_rate": 5.95033081935134e-05,
      "loss": 0.4169,
      "num_tokens": 160224798.0,
      "step": 253
    },
    {
      "epoch": 0.030034291119782428,
      "grad_norm": 0.2881893217563629,
      "learning_rate": 5.949736055937826e-05,
      "loss": 0.4212,
      "num_tokens": 160855622.0,
      "step": 254
    },
    {
      "epoch": 0.030152536360411493,
      "grad_norm": 0.29593420028686523,
      "learning_rate": 5.94913778614486e-05,
      "loss": 0.3898,
      "num_tokens": 161489536.0,
      "step": 255
    },
    {
      "epoch": 0.030270781601040558,
      "grad_norm": 0.31856146454811096,
      "learning_rate": 5.948536010764128e-05,
      "loss": 0.476,
      "num_tokens": 162123541.0,
      "step": 256
    },
    {
      "epoch": 0.030389026841669622,
      "grad_norm": 0.2588259279727936,
      "learning_rate": 5.947930730591957e-05,
      "loss": 0.4321,
      "num_tokens": 162754194.0,
      "step": 257
    },
    {
      "epoch": 0.030507272082298687,
      "grad_norm": 0.2927051782608032,
      "learning_rate": 5.94732194642931e-05,
      "loss": 0.4186,
      "num_tokens": 163389806.0,
      "step": 258
    },
    {
      "epoch": 0.030625517322927752,
      "grad_norm": 0.24726012349128723,
      "learning_rate": 5.946709659081787e-05,
      "loss": 0.4072,
      "num_tokens": 164022489.0,
      "step": 259
    },
    {
      "epoch": 0.030743762563556817,
      "grad_norm": 0.24236567318439484,
      "learning_rate": 5.946093869359624e-05,
      "loss": 0.4158,
      "num_tokens": 164655264.0,
      "step": 260
    },
    {
      "epoch": 0.03086200780418588,
      "grad_norm": 0.2743130922317505,
      "learning_rate": 5.945474578077693e-05,
      "loss": 0.408,
      "num_tokens": 165293663.0,
      "step": 261
    },
    {
      "epoch": 0.030980253044814946,
      "grad_norm": 0.26922914385795593,
      "learning_rate": 5.944851786055497e-05,
      "loss": 0.4308,
      "num_tokens": 165930344.0,
      "step": 262
    },
    {
      "epoch": 0.03109849828544401,
      "grad_norm": 0.27107909321784973,
      "learning_rate": 5.944225494117174e-05,
      "loss": 0.3898,
      "num_tokens": 166565332.0,
      "step": 263
    },
    {
      "epoch": 0.031216743526073076,
      "grad_norm": 0.2749185264110565,
      "learning_rate": 5.9435957030914914e-05,
      "loss": 0.4336,
      "num_tokens": 167200224.0,
      "step": 264
    },
    {
      "epoch": 0.03133498876670214,
      "grad_norm": 0.25396469235420227,
      "learning_rate": 5.942962413811848e-05,
      "loss": 0.4244,
      "num_tokens": 167832144.0,
      "step": 265
    },
    {
      "epoch": 0.0314532340073312,
      "grad_norm": 0.2632233202457428,
      "learning_rate": 5.942325627116272e-05,
      "loss": 0.4059,
      "num_tokens": 168469994.0,
      "step": 266
    },
    {
      "epoch": 0.03157147924796027,
      "grad_norm": 0.27231499552726746,
      "learning_rate": 5.941685343847419e-05,
      "loss": 0.4389,
      "num_tokens": 169106098.0,
      "step": 267
    },
    {
      "epoch": 0.03168972448858933,
      "grad_norm": 0.24510671198368073,
      "learning_rate": 5.941041564852572e-05,
      "loss": 0.3807,
      "num_tokens": 169737993.0,
      "step": 268
    },
    {
      "epoch": 0.0318079697292184,
      "grad_norm": 0.2536793351173401,
      "learning_rate": 5.940394290983639e-05,
      "loss": 0.3914,
      "num_tokens": 170368812.0,
      "step": 269
    },
    {
      "epoch": 0.03192621496984746,
      "grad_norm": 0.2706405818462372,
      "learning_rate": 5.939743523097154e-05,
      "loss": 0.4651,
      "num_tokens": 171006424.0,
      "step": 270
    },
    {
      "epoch": 0.03204446021047653,
      "grad_norm": 0.2482193559408188,
      "learning_rate": 5.939089262054276e-05,
      "loss": 0.4192,
      "num_tokens": 171645721.0,
      "step": 271
    },
    {
      "epoch": 0.03216270545110559,
      "grad_norm": 0.26313939690589905,
      "learning_rate": 5.9384315087207816e-05,
      "loss": 0.4055,
      "num_tokens": 172281075.0,
      "step": 272
    },
    {
      "epoch": 0.03228095069173466,
      "grad_norm": 0.23074069619178772,
      "learning_rate": 5.937770263967074e-05,
      "loss": 0.3695,
      "num_tokens": 172914063.0,
      "step": 273
    },
    {
      "epoch": 0.03239919593236372,
      "grad_norm": 0.2610953152179718,
      "learning_rate": 5.937105528668173e-05,
      "loss": 0.4333,
      "num_tokens": 173548325.0,
      "step": 274
    },
    {
      "epoch": 0.03251744117299279,
      "grad_norm": 0.2842814326286316,
      "learning_rate": 5.9364373037037196e-05,
      "loss": 0.431,
      "num_tokens": 174183975.0,
      "step": 275
    },
    {
      "epoch": 0.03263568641362185,
      "grad_norm": 0.25221580266952515,
      "learning_rate": 5.935765589957971e-05,
      "loss": 0.4047,
      "num_tokens": 174818549.0,
      "step": 276
    },
    {
      "epoch": 0.03275393165425092,
      "grad_norm": 0.2278755009174347,
      "learning_rate": 5.935090388319803e-05,
      "loss": 0.3917,
      "num_tokens": 175451113.0,
      "step": 277
    },
    {
      "epoch": 0.03287217689487998,
      "grad_norm": 0.32434311509132385,
      "learning_rate": 5.9344116996827055e-05,
      "loss": 0.4396,
      "num_tokens": 176087116.0,
      "step": 278
    },
    {
      "epoch": 0.03299042213550905,
      "grad_norm": 0.2737249433994293,
      "learning_rate": 5.933729524944781e-05,
      "loss": 0.4562,
      "num_tokens": 176724483.0,
      "step": 279
    },
    {
      "epoch": 0.03310866737613811,
      "grad_norm": 0.2810339033603668,
      "learning_rate": 5.9330438650087495e-05,
      "loss": 0.4315,
      "num_tokens": 177361164.0,
      "step": 280
    },
    {
      "epoch": 0.033226912616767176,
      "grad_norm": 0.2738715708255768,
      "learning_rate": 5.932354720781941e-05,
      "loss": 0.3845,
      "num_tokens": 177999098.0,
      "step": 281
    },
    {
      "epoch": 0.03334515785739624,
      "grad_norm": 0.28221046924591064,
      "learning_rate": 5.931662093176293e-05,
      "loss": 0.4291,
      "num_tokens": 178627820.0,
      "step": 282
    },
    {
      "epoch": 0.033463403098025306,
      "grad_norm": 0.25607389211654663,
      "learning_rate": 5.930965983108357e-05,
      "loss": 0.4149,
      "num_tokens": 179259176.0,
      "step": 283
    },
    {
      "epoch": 0.03358164833865437,
      "grad_norm": 0.28264743089675903,
      "learning_rate": 5.930266391499291e-05,
      "loss": 0.435,
      "num_tokens": 179890905.0,
      "step": 284
    },
    {
      "epoch": 0.033699893579283435,
      "grad_norm": 0.2758411467075348,
      "learning_rate": 5.929563319274862e-05,
      "loss": 0.4237,
      "num_tokens": 180518445.0,
      "step": 285
    },
    {
      "epoch": 0.0338181388199125,
      "grad_norm": 0.32677340507507324,
      "learning_rate": 5.9288567673654386e-05,
      "loss": 0.4559,
      "num_tokens": 181154667.0,
      "step": 286
    },
    {
      "epoch": 0.033936384060541565,
      "grad_norm": 0.2823239862918854,
      "learning_rate": 5.928146736705998e-05,
      "loss": 0.4384,
      "num_tokens": 181789184.0,
      "step": 287
    },
    {
      "epoch": 0.034054629301170626,
      "grad_norm": 0.2830294668674469,
      "learning_rate": 5.927433228236119e-05,
      "loss": 0.4218,
      "num_tokens": 182421462.0,
      "step": 288
    },
    {
      "epoch": 0.034172874541799694,
      "grad_norm": 0.30646511912345886,
      "learning_rate": 5.926716242899982e-05,
      "loss": 0.4081,
      "num_tokens": 183056216.0,
      "step": 289
    },
    {
      "epoch": 0.034291119782428756,
      "grad_norm": 0.27787357568740845,
      "learning_rate": 5.925995781646371e-05,
      "loss": 0.4386,
      "num_tokens": 183695586.0,
      "step": 290
    },
    {
      "epoch": 0.034409365023057824,
      "grad_norm": 0.3329799175262451,
      "learning_rate": 5.925271845428668e-05,
      "loss": 0.4078,
      "num_tokens": 184323170.0,
      "step": 291
    },
    {
      "epoch": 0.034527610263686885,
      "grad_norm": 0.32851293683052063,
      "learning_rate": 5.924544435204852e-05,
      "loss": 0.4838,
      "num_tokens": 184923402.0,
      "step": 292
    },
    {
      "epoch": 0.03464585550431595,
      "grad_norm": 0.2610059082508087,
      "learning_rate": 5.9238135519375016e-05,
      "loss": 0.4069,
      "num_tokens": 185562179.0,
      "step": 293
    },
    {
      "epoch": 0.034764100744945015,
      "grad_norm": 0.2528479993343353,
      "learning_rate": 5.9230791965937885e-05,
      "loss": 0.4184,
      "num_tokens": 186196488.0,
      "step": 294
    },
    {
      "epoch": 0.03488234598557408,
      "grad_norm": 0.27362382411956787,
      "learning_rate": 5.9223413701454826e-05,
      "loss": 0.4517,
      "num_tokens": 186795333.0,
      "step": 295
    },
    {
      "epoch": 0.035000591226203144,
      "grad_norm": 0.23410287499427795,
      "learning_rate": 5.921600073568944e-05,
      "loss": 0.3923,
      "num_tokens": 187431540.0,
      "step": 296
    },
    {
      "epoch": 0.03511883646683221,
      "grad_norm": 0.2733013927936554,
      "learning_rate": 5.920855307845126e-05,
      "loss": 0.4145,
      "num_tokens": 188067882.0,
      "step": 297
    },
    {
      "epoch": 0.035237081707461274,
      "grad_norm": 0.23181550204753876,
      "learning_rate": 5.920107073959573e-05,
      "loss": 0.452,
      "num_tokens": 188704706.0,
      "step": 298
    },
    {
      "epoch": 0.03535532694809034,
      "grad_norm": 0.26053255796432495,
      "learning_rate": 5.9193553729024175e-05,
      "loss": 0.3939,
      "num_tokens": 189339944.0,
      "step": 299
    },
    {
      "epoch": 0.0354735721887194,
      "grad_norm": 0.3104410767555237,
      "learning_rate": 5.918600205668382e-05,
      "loss": 0.4791,
      "num_tokens": 189977157.0,
      "step": 300
    },
    {
      "epoch": 0.03559181742934847,
      "grad_norm": 0.2637656629085541,
      "learning_rate": 5.917841573256773e-05,
      "loss": 0.4237,
      "num_tokens": 190586411.0,
      "step": 301
    },
    {
      "epoch": 0.03571006266997753,
      "grad_norm": 0.28651612997055054,
      "learning_rate": 5.917079476671486e-05,
      "loss": 0.4401,
      "num_tokens": 191219345.0,
      "step": 302
    },
    {
      "epoch": 0.0358283079106066,
      "grad_norm": 0.22322750091552734,
      "learning_rate": 5.916313916920998e-05,
      "loss": 0.4305,
      "num_tokens": 191855183.0,
      "step": 303
    },
    {
      "epoch": 0.03594655315123566,
      "grad_norm": 0.2441805899143219,
      "learning_rate": 5.915544895018368e-05,
      "loss": 0.4012,
      "num_tokens": 192485493.0,
      "step": 304
    },
    {
      "epoch": 0.03606479839186473,
      "grad_norm": 0.24022462964057922,
      "learning_rate": 5.914772411981241e-05,
      "loss": 0.3959,
      "num_tokens": 193122797.0,
      "step": 305
    },
    {
      "epoch": 0.03618304363249379,
      "grad_norm": 0.24951285123825073,
      "learning_rate": 5.9139964688318375e-05,
      "loss": 0.4222,
      "num_tokens": 193759752.0,
      "step": 306
    },
    {
      "epoch": 0.03630128887312286,
      "grad_norm": 0.24699069559574127,
      "learning_rate": 5.913217066596957e-05,
      "loss": 0.4144,
      "num_tokens": 194394288.0,
      "step": 307
    },
    {
      "epoch": 0.03641953411375192,
      "grad_norm": 0.2645781636238098,
      "learning_rate": 5.912434206307979e-05,
      "loss": 0.4124,
      "num_tokens": 195027778.0,
      "step": 308
    },
    {
      "epoch": 0.03653777935438099,
      "grad_norm": 0.24733687937259674,
      "learning_rate": 5.9116478890008596e-05,
      "loss": 0.3443,
      "num_tokens": 195660013.0,
      "step": 309
    },
    {
      "epoch": 0.03665602459501005,
      "grad_norm": 0.29403185844421387,
      "learning_rate": 5.910858115716124e-05,
      "loss": 0.4319,
      "num_tokens": 196290478.0,
      "step": 310
    },
    {
      "epoch": 0.03677426983563912,
      "grad_norm": 0.2918471395969391,
      "learning_rate": 5.910064887498876e-05,
      "loss": 0.4498,
      "num_tokens": 196927060.0,
      "step": 311
    },
    {
      "epoch": 0.03689251507626818,
      "grad_norm": 0.2860783040523529,
      "learning_rate": 5.9092682053987896e-05,
      "loss": 0.4462,
      "num_tokens": 197557085.0,
      "step": 312
    },
    {
      "epoch": 0.03701076031689725,
      "grad_norm": 0.30764201283454895,
      "learning_rate": 5.90846807047011e-05,
      "loss": 0.4329,
      "num_tokens": 198189535.0,
      "step": 313
    },
    {
      "epoch": 0.03712900555752631,
      "grad_norm": 0.2727659046649933,
      "learning_rate": 5.90766448377165e-05,
      "loss": 0.45,
      "num_tokens": 198820737.0,
      "step": 314
    },
    {
      "epoch": 0.03724725079815538,
      "grad_norm": 0.3000715374946594,
      "learning_rate": 5.906857446366791e-05,
      "loss": 0.407,
      "num_tokens": 199451792.0,
      "step": 315
    },
    {
      "epoch": 0.03736549603878444,
      "grad_norm": 0.2878008186817169,
      "learning_rate": 5.906046959323481e-05,
      "loss": 0.4122,
      "num_tokens": 200086012.0,
      "step": 316
    },
    {
      "epoch": 0.0374837412794135,
      "grad_norm": 0.26824718713760376,
      "learning_rate": 5.9052330237142316e-05,
      "loss": 0.4039,
      "num_tokens": 200723155.0,
      "step": 317
    },
    {
      "epoch": 0.03760198652004257,
      "grad_norm": 0.2783738076686859,
      "learning_rate": 5.90441564061612e-05,
      "loss": 0.4265,
      "num_tokens": 201362056.0,
      "step": 318
    },
    {
      "epoch": 0.03772023176067163,
      "grad_norm": 0.2819783091545105,
      "learning_rate": 5.9035948111107846e-05,
      "loss": 0.4094,
      "num_tokens": 201996006.0,
      "step": 319
    },
    {
      "epoch": 0.0378384770013007,
      "grad_norm": 0.24729867279529572,
      "learning_rate": 5.902770536284422e-05,
      "loss": 0.3785,
      "num_tokens": 202591057.0,
      "step": 320
    },
    {
      "epoch": 0.03795672224192976,
      "grad_norm": 0.26701050996780396,
      "learning_rate": 5.9019428172277925e-05,
      "loss": 0.3925,
      "num_tokens": 203228812.0,
      "step": 321
    },
    {
      "epoch": 0.03807496748255883,
      "grad_norm": 0.2603227198123932,
      "learning_rate": 5.901111655036209e-05,
      "loss": 0.423,
      "num_tokens": 203868294.0,
      "step": 322
    },
    {
      "epoch": 0.03819321272318789,
      "grad_norm": 0.23120039701461792,
      "learning_rate": 5.900277050809545e-05,
      "loss": 0.3883,
      "num_tokens": 204499481.0,
      "step": 323
    },
    {
      "epoch": 0.03831145796381696,
      "grad_norm": 0.25170889496803284,
      "learning_rate": 5.899439005652228e-05,
      "loss": 0.4032,
      "num_tokens": 205133494.0,
      "step": 324
    },
    {
      "epoch": 0.03842970320444602,
      "grad_norm": 0.2173241525888443,
      "learning_rate": 5.898597520673236e-05,
      "loss": 0.4166,
      "num_tokens": 205767024.0,
      "step": 325
    },
    {
      "epoch": 0.038547948445075086,
      "grad_norm": 0.2951737344264984,
      "learning_rate": 5.897752596986101e-05,
      "loss": 0.4365,
      "num_tokens": 206400825.0,
      "step": 326
    },
    {
      "epoch": 0.03866619368570415,
      "grad_norm": 0.2363547384738922,
      "learning_rate": 5.896904235708908e-05,
      "loss": 0.4203,
      "num_tokens": 207040239.0,
      "step": 327
    },
    {
      "epoch": 0.038784438926333216,
      "grad_norm": 0.26757577061653137,
      "learning_rate": 5.896052437964284e-05,
      "loss": 0.4025,
      "num_tokens": 207676732.0,
      "step": 328
    },
    {
      "epoch": 0.03890268416696228,
      "grad_norm": 0.2474527806043625,
      "learning_rate": 5.895197204879412e-05,
      "loss": 0.3947,
      "num_tokens": 208313021.0,
      "step": 329
    },
    {
      "epoch": 0.039020929407591345,
      "grad_norm": 0.27514564990997314,
      "learning_rate": 5.8943385375860145e-05,
      "loss": 0.455,
      "num_tokens": 208946344.0,
      "step": 330
    },
    {
      "epoch": 0.03913917464822041,
      "grad_norm": 0.2627350091934204,
      "learning_rate": 5.8934764372203596e-05,
      "loss": 0.4763,
      "num_tokens": 209577542.0,
      "step": 331
    },
    {
      "epoch": 0.039257419888849475,
      "grad_norm": 0.2800849676132202,
      "learning_rate": 5.892610904923262e-05,
      "loss": 0.4683,
      "num_tokens": 210213046.0,
      "step": 332
    },
    {
      "epoch": 0.039375665129478536,
      "grad_norm": 0.2469765841960907,
      "learning_rate": 5.891741941840074e-05,
      "loss": 0.4216,
      "num_tokens": 210848130.0,
      "step": 333
    },
    {
      "epoch": 0.039493910370107604,
      "grad_norm": 0.2550566494464874,
      "learning_rate": 5.890869549120689e-05,
      "loss": 0.4104,
      "num_tokens": 211482121.0,
      "step": 334
    },
    {
      "epoch": 0.039612155610736666,
      "grad_norm": 0.26573434472084045,
      "learning_rate": 5.8899937279195384e-05,
      "loss": 0.4571,
      "num_tokens": 212120668.0,
      "step": 335
    },
    {
      "epoch": 0.039730400851365734,
      "grad_norm": 0.23955504596233368,
      "learning_rate": 5.889114479395592e-05,
      "loss": 0.4236,
      "num_tokens": 212756102.0,
      "step": 336
    },
    {
      "epoch": 0.039848646091994795,
      "grad_norm": 0.27759063243865967,
      "learning_rate": 5.888231804712353e-05,
      "loss": 0.4502,
      "num_tokens": 213392130.0,
      "step": 337
    },
    {
      "epoch": 0.03996689133262386,
      "grad_norm": 0.24142037332057953,
      "learning_rate": 5.887345705037861e-05,
      "loss": 0.3996,
      "num_tokens": 214021365.0,
      "step": 338
    },
    {
      "epoch": 0.040085136573252925,
      "grad_norm": 0.2325737476348877,
      "learning_rate": 5.8864561815446864e-05,
      "loss": 0.4149,
      "num_tokens": 214651340.0,
      "step": 339
    },
    {
      "epoch": 0.04020338181388199,
      "grad_norm": 0.2732045352458954,
      "learning_rate": 5.885563235409927e-05,
      "loss": 0.4406,
      "num_tokens": 215284267.0,
      "step": 340
    },
    {
      "epoch": 0.040321627054511054,
      "grad_norm": 0.3444238305091858,
      "learning_rate": 5.884666867815216e-05,
      "loss": 0.4533,
      "num_tokens": 215920647.0,
      "step": 341
    },
    {
      "epoch": 0.04043987229514012,
      "grad_norm": 0.2464529573917389,
      "learning_rate": 5.88376707994671e-05,
      "loss": 0.3865,
      "num_tokens": 216555518.0,
      "step": 342
    },
    {
      "epoch": 0.040558117535769184,
      "grad_norm": 0.26775339245796204,
      "learning_rate": 5.882863872995093e-05,
      "loss": 0.3875,
      "num_tokens": 217191272.0,
      "step": 343
    },
    {
      "epoch": 0.04067636277639825,
      "grad_norm": 0.25909754633903503,
      "learning_rate": 5.8819572481555725e-05,
      "loss": 0.4182,
      "num_tokens": 217824700.0,
      "step": 344
    },
    {
      "epoch": 0.04079460801702731,
      "grad_norm": 0.25470298528671265,
      "learning_rate": 5.881047206627881e-05,
      "loss": 0.3802,
      "num_tokens": 218455560.0,
      "step": 345
    },
    {
      "epoch": 0.04091285325765638,
      "grad_norm": 0.26282477378845215,
      "learning_rate": 5.88013374961627e-05,
      "loss": 0.4288,
      "num_tokens": 219091298.0,
      "step": 346
    },
    {
      "epoch": 0.04103109849828544,
      "grad_norm": 0.2278517484664917,
      "learning_rate": 5.879216878329511e-05,
      "loss": 0.4011,
      "num_tokens": 219727733.0,
      "step": 347
    },
    {
      "epoch": 0.04114934373891451,
      "grad_norm": 0.2609797418117523,
      "learning_rate": 5.8782965939808954e-05,
      "loss": 0.4555,
      "num_tokens": 220360695.0,
      "step": 348
    },
    {
      "epoch": 0.04126758897954357,
      "grad_norm": 0.24882322549819946,
      "learning_rate": 5.877372897788229e-05,
      "loss": 0.4439,
      "num_tokens": 220993405.0,
      "step": 349
    },
    {
      "epoch": 0.04138583422017264,
      "grad_norm": 0.23803678154945374,
      "learning_rate": 5.8764457909738334e-05,
      "loss": 0.4379,
      "num_tokens": 221619384.0,
      "step": 350
    },
    {
      "epoch": 0.0415040794608017,
      "grad_norm": 0.2756311893463135,
      "learning_rate": 5.875515274764543e-05,
      "loss": 0.4223,
      "num_tokens": 222250253.0,
      "step": 351
    },
    {
      "epoch": 0.04162232470143077,
      "grad_norm": 0.25036007165908813,
      "learning_rate": 5.874581350391705e-05,
      "loss": 0.4087,
      "num_tokens": 222884935.0,
      "step": 352
    },
    {
      "epoch": 0.04174056994205983,
      "grad_norm": 0.26291555166244507,
      "learning_rate": 5.873644019091176e-05,
      "loss": 0.4097,
      "num_tokens": 223516629.0,
      "step": 353
    },
    {
      "epoch": 0.0418588151826889,
      "grad_norm": 0.263302743434906,
      "learning_rate": 5.872703282103319e-05,
      "loss": 0.4189,
      "num_tokens": 224150031.0,
      "step": 354
    },
    {
      "epoch": 0.04197706042331796,
      "grad_norm": 0.24322155117988586,
      "learning_rate": 5.871759140673007e-05,
      "loss": 0.4041,
      "num_tokens": 224786071.0,
      "step": 355
    },
    {
      "epoch": 0.04209530566394703,
      "grad_norm": 0.2757980525493622,
      "learning_rate": 5.870811596049618e-05,
      "loss": 0.4393,
      "num_tokens": 225421240.0,
      "step": 356
    },
    {
      "epoch": 0.04221355090457609,
      "grad_norm": 0.2762620151042938,
      "learning_rate": 5.8698606494870275e-05,
      "loss": 0.4039,
      "num_tokens": 226052174.0,
      "step": 357
    },
    {
      "epoch": 0.04233179614520516,
      "grad_norm": 0.2832268476486206,
      "learning_rate": 5.868906302243622e-05,
      "loss": 0.4447,
      "num_tokens": 226689398.0,
      "step": 358
    },
    {
      "epoch": 0.04245004138583422,
      "grad_norm": 0.23666320741176605,
      "learning_rate": 5.86794855558228e-05,
      "loss": 0.4085,
      "num_tokens": 227327616.0,
      "step": 359
    },
    {
      "epoch": 0.04256828662646329,
      "grad_norm": 0.27521422505378723,
      "learning_rate": 5.866987410770383e-05,
      "loss": 0.4136,
      "num_tokens": 227952686.0,
      "step": 360
    },
    {
      "epoch": 0.04268653186709235,
      "grad_norm": 0.26603540778160095,
      "learning_rate": 5.8660228690798094e-05,
      "loss": 0.4268,
      "num_tokens": 228587586.0,
      "step": 361
    },
    {
      "epoch": 0.04280477710772142,
      "grad_norm": 0.2510245740413666,
      "learning_rate": 5.86505493178693e-05,
      "loss": 0.4131,
      "num_tokens": 229185285.0,
      "step": 362
    },
    {
      "epoch": 0.04292302234835048,
      "grad_norm": 0.25701025128364563,
      "learning_rate": 5.8640836001726085e-05,
      "loss": 0.4067,
      "num_tokens": 229815886.0,
      "step": 363
    },
    {
      "epoch": 0.04304126758897955,
      "grad_norm": 0.2790713906288147,
      "learning_rate": 5.8631088755222046e-05,
      "loss": 0.4176,
      "num_tokens": 230449875.0,
      "step": 364
    },
    {
      "epoch": 0.04315951282960861,
      "grad_norm": 0.2744642198085785,
      "learning_rate": 5.8621307591255655e-05,
      "loss": 0.4202,
      "num_tokens": 231087769.0,
      "step": 365
    },
    {
      "epoch": 0.043277758070237676,
      "grad_norm": 0.26669076085090637,
      "learning_rate": 5.861149252277025e-05,
      "loss": 0.4053,
      "num_tokens": 231720719.0,
      "step": 366
    },
    {
      "epoch": 0.04339600331086674,
      "grad_norm": 0.24615958333015442,
      "learning_rate": 5.860164356275406e-05,
      "loss": 0.3992,
      "num_tokens": 232354694.0,
      "step": 367
    },
    {
      "epoch": 0.0435142485514958,
      "grad_norm": 0.26475226879119873,
      "learning_rate": 5.8591760724240154e-05,
      "loss": 0.4283,
      "num_tokens": 232986853.0,
      "step": 368
    },
    {
      "epoch": 0.04363249379212487,
      "grad_norm": 0.27385619282722473,
      "learning_rate": 5.858184402030643e-05,
      "loss": 0.4658,
      "num_tokens": 233626246.0,
      "step": 369
    },
    {
      "epoch": 0.04375073903275393,
      "grad_norm": 0.2431042194366455,
      "learning_rate": 5.85718934640756e-05,
      "loss": 0.3582,
      "num_tokens": 234256236.0,
      "step": 370
    },
    {
      "epoch": 0.043868984273383,
      "grad_norm": 0.28270891308784485,
      "learning_rate": 5.8561909068715165e-05,
      "loss": 0.4275,
      "num_tokens": 234894656.0,
      "step": 371
    },
    {
      "epoch": 0.04398722951401206,
      "grad_norm": 0.23586058616638184,
      "learning_rate": 5.855189084743743e-05,
      "loss": 0.4103,
      "num_tokens": 235530175.0,
      "step": 372
    },
    {
      "epoch": 0.044105474754641126,
      "grad_norm": 0.2272070050239563,
      "learning_rate": 5.854183881349944e-05,
      "loss": 0.3961,
      "num_tokens": 236165286.0,
      "step": 373
    },
    {
      "epoch": 0.04422371999527019,
      "grad_norm": 0.27485138177871704,
      "learning_rate": 5.8531752980203e-05,
      "loss": 0.3963,
      "num_tokens": 236799872.0,
      "step": 374
    },
    {
      "epoch": 0.044341965235899256,
      "grad_norm": 0.20811937749385834,
      "learning_rate": 5.8521633360894605e-05,
      "loss": 0.3703,
      "num_tokens": 237436361.0,
      "step": 375
    },
    {
      "epoch": 0.04446021047652832,
      "grad_norm": 0.2607218623161316,
      "learning_rate": 5.851147996896551e-05,
      "loss": 0.3905,
      "num_tokens": 238072053.0,
      "step": 376
    },
    {
      "epoch": 0.044578455717157385,
      "grad_norm": 0.2543987035751343,
      "learning_rate": 5.8501292817851625e-05,
      "loss": 0.4368,
      "num_tokens": 238707861.0,
      "step": 377
    },
    {
      "epoch": 0.044696700957786446,
      "grad_norm": 0.26138100028038025,
      "learning_rate": 5.849107192103355e-05,
      "loss": 0.4042,
      "num_tokens": 239340071.0,
      "step": 378
    },
    {
      "epoch": 0.044814946198415515,
      "grad_norm": 0.26055586338043213,
      "learning_rate": 5.848081729203653e-05,
      "loss": 0.4274,
      "num_tokens": 239974165.0,
      "step": 379
    },
    {
      "epoch": 0.044933191439044576,
      "grad_norm": 0.2314479798078537,
      "learning_rate": 5.847052894443045e-05,
      "loss": 0.3964,
      "num_tokens": 240606077.0,
      "step": 380
    },
    {
      "epoch": 0.045051436679673644,
      "grad_norm": 0.2874937951564789,
      "learning_rate": 5.8460206891829826e-05,
      "loss": 0.4107,
      "num_tokens": 241243195.0,
      "step": 381
    },
    {
      "epoch": 0.045169681920302705,
      "grad_norm": 0.2589111328125,
      "learning_rate": 5.8449851147893765e-05,
      "loss": 0.4553,
      "num_tokens": 241881692.0,
      "step": 382
    },
    {
      "epoch": 0.045287927160931774,
      "grad_norm": 0.28679126501083374,
      "learning_rate": 5.8439461726325956e-05,
      "loss": 0.4266,
      "num_tokens": 242520949.0,
      "step": 383
    },
    {
      "epoch": 0.045406172401560835,
      "grad_norm": 0.22613729536533356,
      "learning_rate": 5.8429038640874645e-05,
      "loss": 0.406,
      "num_tokens": 243160575.0,
      "step": 384
    },
    {
      "epoch": 0.0455244176421899,
      "grad_norm": 0.3079308271408081,
      "learning_rate": 5.841858190533266e-05,
      "loss": 0.4209,
      "num_tokens": 243790943.0,
      "step": 385
    },
    {
      "epoch": 0.045642662882818964,
      "grad_norm": 0.2861464023590088,
      "learning_rate": 5.840809153353732e-05,
      "loss": 0.468,
      "num_tokens": 244423844.0,
      "step": 386
    },
    {
      "epoch": 0.04576090812344803,
      "grad_norm": 0.24443581700325012,
      "learning_rate": 5.839756753937048e-05,
      "loss": 0.3987,
      "num_tokens": 245054470.0,
      "step": 387
    },
    {
      "epoch": 0.045879153364077094,
      "grad_norm": 0.2367839515209198,
      "learning_rate": 5.8387009936758456e-05,
      "loss": 0.3884,
      "num_tokens": 245688487.0,
      "step": 388
    },
    {
      "epoch": 0.04599739860470616,
      "grad_norm": 0.2450435906648636,
      "learning_rate": 5.837641873967209e-05,
      "loss": 0.4189,
      "num_tokens": 246325682.0,
      "step": 389
    },
    {
      "epoch": 0.04611564384533522,
      "grad_norm": 0.2244996875524521,
      "learning_rate": 5.836579396212662e-05,
      "loss": 0.439,
      "num_tokens": 246961506.0,
      "step": 390
    },
    {
      "epoch": 0.04623388908596429,
      "grad_norm": 0.2291857749223709,
      "learning_rate": 5.835513561818177e-05,
      "loss": 0.3873,
      "num_tokens": 247594032.0,
      "step": 391
    },
    {
      "epoch": 0.04635213432659335,
      "grad_norm": 0.25324153900146484,
      "learning_rate": 5.8344443721941645e-05,
      "loss": 0.4355,
      "num_tokens": 248227593.0,
      "step": 392
    },
    {
      "epoch": 0.04647037956722242,
      "grad_norm": 0.23453420400619507,
      "learning_rate": 5.833371828755477e-05,
      "loss": 0.4156,
      "num_tokens": 248859376.0,
      "step": 393
    },
    {
      "epoch": 0.04658862480785148,
      "grad_norm": 0.2575860619544983,
      "learning_rate": 5.832295932921405e-05,
      "loss": 0.4346,
      "num_tokens": 249493350.0,
      "step": 394
    },
    {
      "epoch": 0.04670687004848055,
      "grad_norm": 0.27037689089775085,
      "learning_rate": 5.8312166861156755e-05,
      "loss": 0.3934,
      "num_tokens": 250126507.0,
      "step": 395
    },
    {
      "epoch": 0.04682511528910961,
      "grad_norm": 0.23657937347888947,
      "learning_rate": 5.830134089766448e-05,
      "loss": 0.4322,
      "num_tokens": 250765825.0,
      "step": 396
    },
    {
      "epoch": 0.04694336052973868,
      "grad_norm": 0.2646414339542389,
      "learning_rate": 5.829048145306315e-05,
      "loss": 0.4261,
      "num_tokens": 251403372.0,
      "step": 397
    },
    {
      "epoch": 0.04706160577036774,
      "grad_norm": 0.26739707589149475,
      "learning_rate": 5.827958854172302e-05,
      "loss": 0.4417,
      "num_tokens": 252041762.0,
      "step": 398
    },
    {
      "epoch": 0.04717985101099681,
      "grad_norm": 0.25291574001312256,
      "learning_rate": 5.8268662178058614e-05,
      "loss": 0.412,
      "num_tokens": 252673716.0,
      "step": 399
    },
    {
      "epoch": 0.04729809625162587,
      "grad_norm": 0.24120593070983887,
      "learning_rate": 5.82577023765287e-05,
      "loss": 0.4209,
      "num_tokens": 253310805.0,
      "step": 400
    },
    {
      "epoch": 0.04741634149225494,
      "grad_norm": 0.23023198544979095,
      "learning_rate": 5.824670915163632e-05,
      "loss": 0.3477,
      "num_tokens": 253944650.0,
      "step": 401
    },
    {
      "epoch": 0.047534586732884,
      "grad_norm": 0.22422128915786743,
      "learning_rate": 5.823568251792876e-05,
      "loss": 0.3887,
      "num_tokens": 254579224.0,
      "step": 402
    },
    {
      "epoch": 0.04765283197351307,
      "grad_norm": 0.23010247945785522,
      "learning_rate": 5.8224622489997486e-05,
      "loss": 0.4055,
      "num_tokens": 255212393.0,
      "step": 403
    },
    {
      "epoch": 0.04777107721414213,
      "grad_norm": 0.22843380272388458,
      "learning_rate": 5.821352908247816e-05,
      "loss": 0.4294,
      "num_tokens": 255846216.0,
      "step": 404
    },
    {
      "epoch": 0.0478893224547712,
      "grad_norm": 0.24752621352672577,
      "learning_rate": 5.820240231005062e-05,
      "loss": 0.4379,
      "num_tokens": 256484541.0,
      "step": 405
    },
    {
      "epoch": 0.04800756769540026,
      "grad_norm": 0.26750537753105164,
      "learning_rate": 5.819124218743887e-05,
      "loss": 0.3972,
      "num_tokens": 257114691.0,
      "step": 406
    },
    {
      "epoch": 0.04812581293602933,
      "grad_norm": 0.22661860287189484,
      "learning_rate": 5.8180048729411025e-05,
      "loss": 0.4142,
      "num_tokens": 257748044.0,
      "step": 407
    },
    {
      "epoch": 0.04824405817665839,
      "grad_norm": 0.2039046287536621,
      "learning_rate": 5.816882195077929e-05,
      "loss": 0.369,
      "num_tokens": 258386984.0,
      "step": 408
    },
    {
      "epoch": 0.04836230341728746,
      "grad_norm": 0.2567331790924072,
      "learning_rate": 5.8157561866400044e-05,
      "loss": 0.4376,
      "num_tokens": 259023318.0,
      "step": 409
    },
    {
      "epoch": 0.04848054865791652,
      "grad_norm": 0.21170946955680847,
      "learning_rate": 5.814626849117365e-05,
      "loss": 0.3909,
      "num_tokens": 259657204.0,
      "step": 410
    },
    {
      "epoch": 0.048598793898545586,
      "grad_norm": 0.24077744781970978,
      "learning_rate": 5.813494184004459e-05,
      "loss": 0.3914,
      "num_tokens": 260290556.0,
      "step": 411
    },
    {
      "epoch": 0.04871703913917465,
      "grad_norm": 0.2552764415740967,
      "learning_rate": 5.812358192800132e-05,
      "loss": 0.4543,
      "num_tokens": 260928576.0,
      "step": 412
    },
    {
      "epoch": 0.048835284379803716,
      "grad_norm": 0.2142765074968338,
      "learning_rate": 5.811218877007637e-05,
      "loss": 0.4025,
      "num_tokens": 261563484.0,
      "step": 413
    },
    {
      "epoch": 0.04895352962043278,
      "grad_norm": 0.24795947968959808,
      "learning_rate": 5.8100762381346214e-05,
      "loss": 0.3724,
      "num_tokens": 262198058.0,
      "step": 414
    },
    {
      "epoch": 0.049071774861061845,
      "grad_norm": 0.2619639039039612,
      "learning_rate": 5.8089302776931333e-05,
      "loss": 0.4324,
      "num_tokens": 262825300.0,
      "step": 415
    },
    {
      "epoch": 0.04919002010169091,
      "grad_norm": 0.24405325949192047,
      "learning_rate": 5.8077809971996144e-05,
      "loss": 0.4465,
      "num_tokens": 263458362.0,
      "step": 416
    },
    {
      "epoch": 0.049308265342319975,
      "grad_norm": 0.2387343943119049,
      "learning_rate": 5.806628398174902e-05,
      "loss": 0.3886,
      "num_tokens": 264088367.0,
      "step": 417
    },
    {
      "epoch": 0.049426510582949036,
      "grad_norm": 0.254852831363678,
      "learning_rate": 5.805472482144222e-05,
      "loss": 0.4029,
      "num_tokens": 264724592.0,
      "step": 418
    },
    {
      "epoch": 0.049544755823578104,
      "grad_norm": 0.25719350576400757,
      "learning_rate": 5.8043132506371916e-05,
      "loss": 0.3904,
      "num_tokens": 265360262.0,
      "step": 419
    },
    {
      "epoch": 0.049663001064207166,
      "grad_norm": 0.2636720836162567,
      "learning_rate": 5.803150705187815e-05,
      "loss": 0.4441,
      "num_tokens": 265997689.0,
      "step": 420
    },
    {
      "epoch": 0.04978124630483623,
      "grad_norm": 0.29422527551651,
      "learning_rate": 5.80198484733448e-05,
      "loss": 0.4119,
      "num_tokens": 266631262.0,
      "step": 421
    },
    {
      "epoch": 0.049899491545465295,
      "grad_norm": 0.28710222244262695,
      "learning_rate": 5.800815678619961e-05,
      "loss": 0.4054,
      "num_tokens": 267268979.0,
      "step": 422
    },
    {
      "epoch": 0.050017736786094356,
      "grad_norm": 0.23449011147022247,
      "learning_rate": 5.799643200591411e-05,
      "loss": 0.41,
      "num_tokens": 267904627.0,
      "step": 423
    },
    {
      "epoch": 0.050135982026723425,
      "grad_norm": 0.2233870029449463,
      "learning_rate": 5.798467414800363e-05,
      "loss": 0.3647,
      "num_tokens": 268543181.0,
      "step": 424
    },
    {
      "epoch": 0.050254227267352486,
      "grad_norm": 0.2825268507003784,
      "learning_rate": 5.797288322802727e-05,
      "loss": 0.4117,
      "num_tokens": 269179996.0,
      "step": 425
    },
    {
      "epoch": 0.050372472507981554,
      "grad_norm": 0.21707065403461456,
      "learning_rate": 5.796105926158786e-05,
      "loss": 0.3921,
      "num_tokens": 269811253.0,
      "step": 426
    },
    {
      "epoch": 0.050490717748610615,
      "grad_norm": 0.24125707149505615,
      "learning_rate": 5.794920226433203e-05,
      "loss": 0.3987,
      "num_tokens": 270441984.0,
      "step": 427
    },
    {
      "epoch": 0.050608962989239684,
      "grad_norm": 0.2642814517021179,
      "learning_rate": 5.793731225195003e-05,
      "loss": 0.41,
      "num_tokens": 271074224.0,
      "step": 428
    },
    {
      "epoch": 0.050727208229868745,
      "grad_norm": 0.2821972370147705,
      "learning_rate": 5.792538924017587e-05,
      "loss": 0.4555,
      "num_tokens": 271705707.0,
      "step": 429
    },
    {
      "epoch": 0.05084545347049781,
      "grad_norm": 0.2493581920862198,
      "learning_rate": 5.791343324478718e-05,
      "loss": 0.3889,
      "num_tokens": 272318335.0,
      "step": 430
    },
    {
      "epoch": 0.050963698711126874,
      "grad_norm": 0.2921682596206665,
      "learning_rate": 5.790144428160525e-05,
      "loss": 0.4254,
      "num_tokens": 272955472.0,
      "step": 431
    },
    {
      "epoch": 0.05108194395175594,
      "grad_norm": 0.2746122181415558,
      "learning_rate": 5.7889422366495024e-05,
      "loss": 0.4138,
      "num_tokens": 273559901.0,
      "step": 432
    },
    {
      "epoch": 0.051200189192385004,
      "grad_norm": 0.24401576817035675,
      "learning_rate": 5.787736751536502e-05,
      "loss": 0.4117,
      "num_tokens": 274192648.0,
      "step": 433
    },
    {
      "epoch": 0.05131843443301407,
      "grad_norm": 0.2667708098888397,
      "learning_rate": 5.786527974416733e-05,
      "loss": 0.3815,
      "num_tokens": 274831825.0,
      "step": 434
    },
    {
      "epoch": 0.05143667967364313,
      "grad_norm": 0.23436440527439117,
      "learning_rate": 5.785315906889764e-05,
      "loss": 0.3677,
      "num_tokens": 275456510.0,
      "step": 435
    },
    {
      "epoch": 0.0515549249142722,
      "grad_norm": 0.27032753825187683,
      "learning_rate": 5.784100550559517e-05,
      "loss": 0.3713,
      "num_tokens": 276091876.0,
      "step": 436
    },
    {
      "epoch": 0.05167317015490126,
      "grad_norm": 0.2609861195087433,
      "learning_rate": 5.782881907034266e-05,
      "loss": 0.4188,
      "num_tokens": 276727335.0,
      "step": 437
    },
    {
      "epoch": 0.05179141539553033,
      "grad_norm": 0.2858883738517761,
      "learning_rate": 5.7816599779266336e-05,
      "loss": 0.4459,
      "num_tokens": 277359081.0,
      "step": 438
    },
    {
      "epoch": 0.05190966063615939,
      "grad_norm": 0.26549068093299866,
      "learning_rate": 5.78043476485359e-05,
      "loss": 0.3795,
      "num_tokens": 277990765.0,
      "step": 439
    },
    {
      "epoch": 0.05202790587678846,
      "grad_norm": 0.26636239886283875,
      "learning_rate": 5.779206269436455e-05,
      "loss": 0.4056,
      "num_tokens": 278612675.0,
      "step": 440
    },
    {
      "epoch": 0.05214615111741752,
      "grad_norm": 0.21832014620304108,
      "learning_rate": 5.7779744933008864e-05,
      "loss": 0.3829,
      "num_tokens": 279246447.0,
      "step": 441
    },
    {
      "epoch": 0.05226439635804659,
      "grad_norm": 0.2970152497291565,
      "learning_rate": 5.776739438076887e-05,
      "loss": 0.4533,
      "num_tokens": 279880451.0,
      "step": 442
    },
    {
      "epoch": 0.05238264159867565,
      "grad_norm": 0.27263355255126953,
      "learning_rate": 5.775501105398798e-05,
      "loss": 0.4431,
      "num_tokens": 280517149.0,
      "step": 443
    },
    {
      "epoch": 0.05250088683930472,
      "grad_norm": 0.23851481080055237,
      "learning_rate": 5.774259496905298e-05,
      "loss": 0.3809,
      "num_tokens": 281144009.0,
      "step": 444
    },
    {
      "epoch": 0.05261913207993378,
      "grad_norm": 0.26180770993232727,
      "learning_rate": 5.7730146142393974e-05,
      "loss": 0.3976,
      "num_tokens": 281781076.0,
      "step": 445
    },
    {
      "epoch": 0.05273737732056285,
      "grad_norm": 0.23845084011554718,
      "learning_rate": 5.771766459048445e-05,
      "loss": 0.4037,
      "num_tokens": 282418028.0,
      "step": 446
    },
    {
      "epoch": 0.05285562256119191,
      "grad_norm": 0.24678055942058563,
      "learning_rate": 5.770515032984114e-05,
      "loss": 0.3939,
      "num_tokens": 283052667.0,
      "step": 447
    },
    {
      "epoch": 0.05297386780182098,
      "grad_norm": 0.28906598687171936,
      "learning_rate": 5.769260337702411e-05,
      "loss": 0.4435,
      "num_tokens": 283689384.0,
      "step": 448
    },
    {
      "epoch": 0.05309211304245004,
      "grad_norm": 0.20284819602966309,
      "learning_rate": 5.768002374863666e-05,
      "loss": 0.369,
      "num_tokens": 284325980.0,
      "step": 449
    },
    {
      "epoch": 0.05321035828307911,
      "grad_norm": 0.2573983073234558,
      "learning_rate": 5.766741146132533e-05,
      "loss": 0.4255,
      "num_tokens": 284927410.0,
      "step": 450
    },
    {
      "epoch": 0.05332860352370817,
      "grad_norm": 0.26626330614089966,
      "learning_rate": 5.7654766531779884e-05,
      "loss": 0.4117,
      "num_tokens": 285561853.0,
      "step": 451
    },
    {
      "epoch": 0.05344684876433724,
      "grad_norm": 0.23338961601257324,
      "learning_rate": 5.7642088976733284e-05,
      "loss": 0.4227,
      "num_tokens": 286200667.0,
      "step": 452
    },
    {
      "epoch": 0.0535650940049663,
      "grad_norm": 0.2275472730398178,
      "learning_rate": 5.762937881296165e-05,
      "loss": 0.3856,
      "num_tokens": 286838297.0,
      "step": 453
    },
    {
      "epoch": 0.05368333924559537,
      "grad_norm": 0.2396043986082077,
      "learning_rate": 5.761663605728428e-05,
      "loss": 0.3895,
      "num_tokens": 287478013.0,
      "step": 454
    },
    {
      "epoch": 0.05380158448622443,
      "grad_norm": 0.27043527364730835,
      "learning_rate": 5.7603860726563574e-05,
      "loss": 0.4215,
      "num_tokens": 288113641.0,
      "step": 455
    },
    {
      "epoch": 0.053919829726853497,
      "grad_norm": 0.2147088646888733,
      "learning_rate": 5.759105283770505e-05,
      "loss": 0.4138,
      "num_tokens": 288748606.0,
      "step": 456
    },
    {
      "epoch": 0.05403807496748256,
      "grad_norm": 0.24205036461353302,
      "learning_rate": 5.757821240765731e-05,
      "loss": 0.3934,
      "num_tokens": 289387862.0,
      "step": 457
    },
    {
      "epoch": 0.054156320208111626,
      "grad_norm": 0.21859905123710632,
      "learning_rate": 5.756533945341201e-05,
      "loss": 0.3941,
      "num_tokens": 290025351.0,
      "step": 458
    },
    {
      "epoch": 0.05427456544874069,
      "grad_norm": 0.23405297100543976,
      "learning_rate": 5.755243399200386e-05,
      "loss": 0.3842,
      "num_tokens": 290660065.0,
      "step": 459
    },
    {
      "epoch": 0.054392810689369755,
      "grad_norm": 0.2372758984565735,
      "learning_rate": 5.7539496040510566e-05,
      "loss": 0.3901,
      "num_tokens": 291293769.0,
      "step": 460
    },
    {
      "epoch": 0.05451105592999882,
      "grad_norm": 0.22159790992736816,
      "learning_rate": 5.752652561605286e-05,
      "loss": 0.3958,
      "num_tokens": 291932970.0,
      "step": 461
    },
    {
      "epoch": 0.054629301170627885,
      "grad_norm": 0.22865094244480133,
      "learning_rate": 5.751352273579441e-05,
      "loss": 0.4059,
      "num_tokens": 292570196.0,
      "step": 462
    },
    {
      "epoch": 0.054747546411256946,
      "grad_norm": 0.2343677282333374,
      "learning_rate": 5.750048741694185e-05,
      "loss": 0.4382,
      "num_tokens": 293208953.0,
      "step": 463
    },
    {
      "epoch": 0.054865791651886014,
      "grad_norm": 0.22632277011871338,
      "learning_rate": 5.748741967674473e-05,
      "loss": 0.4056,
      "num_tokens": 293841194.0,
      "step": 464
    },
    {
      "epoch": 0.054984036892515076,
      "grad_norm": 0.24287892878055573,
      "learning_rate": 5.7474319532495534e-05,
      "loss": 0.4131,
      "num_tokens": 294478601.0,
      "step": 465
    },
    {
      "epoch": 0.055102282133144144,
      "grad_norm": 0.22494228184223175,
      "learning_rate": 5.7461187001529575e-05,
      "loss": 0.4069,
      "num_tokens": 295110727.0,
      "step": 466
    },
    {
      "epoch": 0.055220527373773205,
      "grad_norm": 0.22685138881206512,
      "learning_rate": 5.744802210122507e-05,
      "loss": 0.3812,
      "num_tokens": 295744614.0,
      "step": 467
    },
    {
      "epoch": 0.05533877261440227,
      "grad_norm": 0.22187015414237976,
      "learning_rate": 5.743482484900304e-05,
      "loss": 0.3966,
      "num_tokens": 296380861.0,
      "step": 468
    },
    {
      "epoch": 0.055457017855031335,
      "grad_norm": 0.22301408648490906,
      "learning_rate": 5.7421595262327336e-05,
      "loss": 0.405,
      "num_tokens": 297007279.0,
      "step": 469
    },
    {
      "epoch": 0.0555752630956604,
      "grad_norm": 0.22492894530296326,
      "learning_rate": 5.7408333358704576e-05,
      "loss": 0.3859,
      "num_tokens": 297623126.0,
      "step": 470
    },
    {
      "epoch": 0.055693508336289464,
      "grad_norm": 0.24347686767578125,
      "learning_rate": 5.739503915568416e-05,
      "loss": 0.4437,
      "num_tokens": 298255789.0,
      "step": 471
    },
    {
      "epoch": 0.05581175357691853,
      "grad_norm": 0.2275567501783371,
      "learning_rate": 5.738171267085824e-05,
      "loss": 0.4203,
      "num_tokens": 298885491.0,
      "step": 472
    },
    {
      "epoch": 0.055929998817547594,
      "grad_norm": 0.20554254949092865,
      "learning_rate": 5.736835392186165e-05,
      "loss": 0.3876,
      "num_tokens": 299521680.0,
      "step": 473
    },
    {
      "epoch": 0.056048244058176655,
      "grad_norm": 0.24628478288650513,
      "learning_rate": 5.735496292637196e-05,
      "loss": 0.3995,
      "num_tokens": 300161168.0,
      "step": 474
    },
    {
      "epoch": 0.05616648929880572,
      "grad_norm": 0.21095220744609833,
      "learning_rate": 5.734153970210938e-05,
      "loss": 0.3893,
      "num_tokens": 300793878.0,
      "step": 475
    },
    {
      "epoch": 0.056284734539434784,
      "grad_norm": 0.22608943283557892,
      "learning_rate": 5.732808426683678e-05,
      "loss": 0.4145,
      "num_tokens": 301427428.0,
      "step": 476
    },
    {
      "epoch": 0.05640297978006385,
      "grad_norm": 0.24192720651626587,
      "learning_rate": 5.731459663835966e-05,
      "loss": 0.4431,
      "num_tokens": 302063334.0,
      "step": 477
    },
    {
      "epoch": 0.056521225020692914,
      "grad_norm": 0.2017304003238678,
      "learning_rate": 5.730107683452611e-05,
      "loss": 0.3853,
      "num_tokens": 302696034.0,
      "step": 478
    },
    {
      "epoch": 0.05663947026132198,
      "grad_norm": 0.23585256934165955,
      "learning_rate": 5.7287524873226806e-05,
      "loss": 0.4106,
      "num_tokens": 303331655.0,
      "step": 479
    },
    {
      "epoch": 0.056757715501951043,
      "grad_norm": 0.26219794154167175,
      "learning_rate": 5.727394077239499e-05,
      "loss": 0.483,
      "num_tokens": 303962802.0,
      "step": 480
    },
    {
      "epoch": 0.05687596074258011,
      "grad_norm": 0.22113117575645447,
      "learning_rate": 5.7260324550006394e-05,
      "loss": 0.4007,
      "num_tokens": 304595703.0,
      "step": 481
    },
    {
      "epoch": 0.05699420598320917,
      "grad_norm": 0.23816938698291779,
      "learning_rate": 5.7246676224079296e-05,
      "loss": 0.4032,
      "num_tokens": 305227732.0,
      "step": 482
    },
    {
      "epoch": 0.05711245122383824,
      "grad_norm": 0.21741387248039246,
      "learning_rate": 5.723299581267444e-05,
      "loss": 0.4045,
      "num_tokens": 305861092.0,
      "step": 483
    },
    {
      "epoch": 0.0572306964644673,
      "grad_norm": 0.24627813696861267,
      "learning_rate": 5.721928333389502e-05,
      "loss": 0.4081,
      "num_tokens": 306497173.0,
      "step": 484
    },
    {
      "epoch": 0.05734894170509637,
      "grad_norm": 0.2061251550912857,
      "learning_rate": 5.72055388058867e-05,
      "loss": 0.3939,
      "num_tokens": 307133773.0,
      "step": 485
    },
    {
      "epoch": 0.05746718694572543,
      "grad_norm": 0.24562332034111023,
      "learning_rate": 5.719176224683751e-05,
      "loss": 0.3883,
      "num_tokens": 307760142.0,
      "step": 486
    },
    {
      "epoch": 0.0575854321863545,
      "grad_norm": 0.21945923566818237,
      "learning_rate": 5.717795367497789e-05,
      "loss": 0.4186,
      "num_tokens": 308392658.0,
      "step": 487
    },
    {
      "epoch": 0.05770367742698356,
      "grad_norm": 0.20955249667167664,
      "learning_rate": 5.7164113108580624e-05,
      "loss": 0.3745,
      "num_tokens": 309023200.0,
      "step": 488
    },
    {
      "epoch": 0.05782192266761263,
      "grad_norm": 0.2421126514673233,
      "learning_rate": 5.7150240565960876e-05,
      "loss": 0.4203,
      "num_tokens": 309657191.0,
      "step": 489
    },
    {
      "epoch": 0.05794016790824169,
      "grad_norm": 0.22103598713874817,
      "learning_rate": 5.713633606547608e-05,
      "loss": 0.4135,
      "num_tokens": 310290361.0,
      "step": 490
    },
    {
      "epoch": 0.05805841314887076,
      "grad_norm": 0.24057047069072723,
      "learning_rate": 5.712239962552599e-05,
      "loss": 0.4306,
      "num_tokens": 310922697.0,
      "step": 491
    },
    {
      "epoch": 0.05817665838949982,
      "grad_norm": 0.24529068171977997,
      "learning_rate": 5.710843126455258e-05,
      "loss": 0.422,
      "num_tokens": 311557453.0,
      "step": 492
    },
    {
      "epoch": 0.05829490363012889,
      "grad_norm": 0.22984932363033295,
      "learning_rate": 5.7094431001040134e-05,
      "loss": 0.423,
      "num_tokens": 312193193.0,
      "step": 493
    },
    {
      "epoch": 0.05841314887075795,
      "grad_norm": 0.21943072974681854,
      "learning_rate": 5.708039885351509e-05,
      "loss": 0.4211,
      "num_tokens": 312828473.0,
      "step": 494
    },
    {
      "epoch": 0.05853139411138702,
      "grad_norm": 0.21643073856830597,
      "learning_rate": 5.706633484054611e-05,
      "loss": 0.392,
      "num_tokens": 313462344.0,
      "step": 495
    },
    {
      "epoch": 0.05864963935201608,
      "grad_norm": 0.21632198989391327,
      "learning_rate": 5.705223898074402e-05,
      "loss": 0.3923,
      "num_tokens": 314080640.0,
      "step": 496
    },
    {
      "epoch": 0.05876788459264515,
      "grad_norm": 0.24517853558063507,
      "learning_rate": 5.703811129276178e-05,
      "loss": 0.4045,
      "num_tokens": 314717785.0,
      "step": 497
    },
    {
      "epoch": 0.05888612983327421,
      "grad_norm": 0.2291676551103592,
      "learning_rate": 5.7023951795294475e-05,
      "loss": 0.4608,
      "num_tokens": 315351322.0,
      "step": 498
    },
    {
      "epoch": 0.05900437507390328,
      "grad_norm": 0.23110775649547577,
      "learning_rate": 5.700976050707928e-05,
      "loss": 0.4141,
      "num_tokens": 315987690.0,
      "step": 499
    },
    {
      "epoch": 0.05912262031453234,
      "grad_norm": 197.8325958251953,
      "learning_rate": 5.699553744689544e-05,
      "loss": 5.1515,
      "num_tokens": 316588692.0,
      "step": 500
    },
    {
      "epoch": 0.05924086555516141,
      "grad_norm": 0.30840274691581726,
      "learning_rate": 5.6981282633564246e-05,
      "loss": 0.3961,
      "num_tokens": 317208507.0,
      "step": 501
    },
    {
      "epoch": 0.05935911079579047,
      "grad_norm": 0.24632979929447174,
      "learning_rate": 5.696699608594899e-05,
      "loss": 0.4435,
      "num_tokens": 317844277.0,
      "step": 502
    },
    {
      "epoch": 0.059477356036419536,
      "grad_norm": 0.24863989651203156,
      "learning_rate": 5.6952677822955e-05,
      "loss": 0.4299,
      "num_tokens": 318478032.0,
      "step": 503
    },
    {
      "epoch": 0.0595956012770486,
      "grad_norm": 0.25635260343551636,
      "learning_rate": 5.693832786352952e-05,
      "loss": 0.4575,
      "num_tokens": 319114663.0,
      "step": 504
    },
    {
      "epoch": 0.059713846517677666,
      "grad_norm": 0.2566048204898834,
      "learning_rate": 5.6923946226661766e-05,
      "loss": 0.3555,
      "num_tokens": 319744940.0,
      "step": 505
    },
    {
      "epoch": 0.05983209175830673,
      "grad_norm": 0.24760176241397858,
      "learning_rate": 5.690953293138289e-05,
      "loss": 0.4025,
      "num_tokens": 320377220.0,
      "step": 506
    },
    {
      "epoch": 0.059950336998935795,
      "grad_norm": 0.24116012454032898,
      "learning_rate": 5.68950879967659e-05,
      "loss": 0.4053,
      "num_tokens": 321010441.0,
      "step": 507
    },
    {
      "epoch": 0.060068582239564856,
      "grad_norm": 0.23760244250297546,
      "learning_rate": 5.688061144192569e-05,
      "loss": 0.387,
      "num_tokens": 321643644.0,
      "step": 508
    },
    {
      "epoch": 0.060186827480193925,
      "grad_norm": 0.22054804861545563,
      "learning_rate": 5.6866103286019014e-05,
      "loss": 0.3968,
      "num_tokens": 322275355.0,
      "step": 509
    },
    {
      "epoch": 0.060305072720822986,
      "grad_norm": 0.2905291020870209,
      "learning_rate": 5.6851563548244394e-05,
      "loss": 0.4227,
      "num_tokens": 322908591.0,
      "step": 510
    },
    {
      "epoch": 0.060423317961452054,
      "grad_norm": 0.2590107321739197,
      "learning_rate": 5.68369922478422e-05,
      "loss": 0.4215,
      "num_tokens": 323542459.0,
      "step": 511
    },
    {
      "epoch": 0.060541563202081115,
      "grad_norm": 0.25308477878570557,
      "learning_rate": 5.6822389404094526e-05,
      "loss": 0.4019,
      "num_tokens": 324176254.0,
      "step": 512
    },
    {
      "epoch": 0.060659808442710184,
      "grad_norm": 0.24590685963630676,
      "learning_rate": 5.680775503632525e-05,
      "loss": 0.3763,
      "num_tokens": 324803884.0,
      "step": 513
    },
    {
      "epoch": 0.060778053683339245,
      "grad_norm": 0.2458338737487793,
      "learning_rate": 5.679308916389993e-05,
      "loss": 0.4326,
      "num_tokens": 325442394.0,
      "step": 514
    },
    {
      "epoch": 0.06089629892396831,
      "grad_norm": 0.2458193451166153,
      "learning_rate": 5.677839180622581e-05,
      "loss": 0.4227,
      "num_tokens": 326081004.0,
      "step": 515
    },
    {
      "epoch": 0.061014544164597374,
      "grad_norm": 0.23497483134269714,
      "learning_rate": 5.676366298275183e-05,
      "loss": 0.4037,
      "num_tokens": 326711646.0,
      "step": 516
    },
    {
      "epoch": 0.06113278940522644,
      "grad_norm": 0.2159551978111267,
      "learning_rate": 5.674890271296856e-05,
      "loss": 0.3967,
      "num_tokens": 327346340.0,
      "step": 517
    },
    {
      "epoch": 0.061251034645855504,
      "grad_norm": 0.23301520943641663,
      "learning_rate": 5.673411101640814e-05,
      "loss": 0.3777,
      "num_tokens": 327982364.0,
      "step": 518
    },
    {
      "epoch": 0.06136927988648457,
      "grad_norm": 0.2253977358341217,
      "learning_rate": 5.6719287912644365e-05,
      "loss": 0.4402,
      "num_tokens": 328613513.0,
      "step": 519
    },
    {
      "epoch": 0.06148752512711363,
      "grad_norm": 0.23504820466041565,
      "learning_rate": 5.670443342129254e-05,
      "loss": 0.4156,
      "num_tokens": 329246797.0,
      "step": 520
    },
    {
      "epoch": 0.0616057703677427,
      "grad_norm": 0.23867231607437134,
      "learning_rate": 5.668954756200954e-05,
      "loss": 0.4268,
      "num_tokens": 329882051.0,
      "step": 521
    },
    {
      "epoch": 0.06172401560837176,
      "grad_norm": 0.23447009921073914,
      "learning_rate": 5.667463035449371e-05,
      "loss": 0.3978,
      "num_tokens": 330518781.0,
      "step": 522
    },
    {
      "epoch": 0.06184226084900083,
      "grad_norm": 0.2304748296737671,
      "learning_rate": 5.66596818184849e-05,
      "loss": 0.4029,
      "num_tokens": 331148822.0,
      "step": 523
    },
    {
      "epoch": 0.06196050608962989,
      "grad_norm": 0.22956591844558716,
      "learning_rate": 5.664470197376443e-05,
      "loss": 0.4224,
      "num_tokens": 331779530.0,
      "step": 524
    },
    {
      "epoch": 0.06207875133025896,
      "grad_norm": 0.21514792740345,
      "learning_rate": 5.6629690840155037e-05,
      "loss": 0.3633,
      "num_tokens": 332409781.0,
      "step": 525
    },
    {
      "epoch": 0.06219699657088802,
      "grad_norm": 0.2394063025712967,
      "learning_rate": 5.6614648437520845e-05,
      "loss": 0.4489,
      "num_tokens": 333047880.0,
      "step": 526
    },
    {
      "epoch": 0.06231524181151708,
      "grad_norm": 0.23102419078350067,
      "learning_rate": 5.6599574785767395e-05,
      "loss": 0.4125,
      "num_tokens": 333683586.0,
      "step": 527
    },
    {
      "epoch": 0.06243348705214615,
      "grad_norm": 0.22116991877555847,
      "learning_rate": 5.658446990484153e-05,
      "loss": 0.4137,
      "num_tokens": 334315506.0,
      "step": 528
    },
    {
      "epoch": 0.06255173229277522,
      "grad_norm": 0.2364656627178192,
      "learning_rate": 5.656933381473147e-05,
      "loss": 0.3834,
      "num_tokens": 334951291.0,
      "step": 529
    },
    {
      "epoch": 0.06266997753340428,
      "grad_norm": 0.23201730847358704,
      "learning_rate": 5.65541665354667e-05,
      "loss": 0.3932,
      "num_tokens": 335585867.0,
      "step": 530
    },
    {
      "epoch": 0.06278822277403334,
      "grad_norm": 0.2602282464504242,
      "learning_rate": 5.6538968087118e-05,
      "loss": 0.384,
      "num_tokens": 336220052.0,
      "step": 531
    },
    {
      "epoch": 0.0629064680146624,
      "grad_norm": 0.24242635071277618,
      "learning_rate": 5.652373848979735e-05,
      "loss": 0.4342,
      "num_tokens": 336856340.0,
      "step": 532
    },
    {
      "epoch": 0.06302471325529148,
      "grad_norm": 0.25628939270973206,
      "learning_rate": 5.6508477763658024e-05,
      "loss": 0.3873,
      "num_tokens": 337491980.0,
      "step": 533
    },
    {
      "epoch": 0.06314295849592054,
      "grad_norm": 0.2245175689458847,
      "learning_rate": 5.649318592889442e-05,
      "loss": 0.391,
      "num_tokens": 338130649.0,
      "step": 534
    },
    {
      "epoch": 0.0632612037365496,
      "grad_norm": 0.31959810853004456,
      "learning_rate": 5.6477863005742154e-05,
      "loss": 0.437,
      "num_tokens": 338765865.0,
      "step": 535
    },
    {
      "epoch": 0.06337944897717866,
      "grad_norm": 0.23283618688583374,
      "learning_rate": 5.6462509014477916e-05,
      "loss": 0.4077,
      "num_tokens": 339363366.0,
      "step": 536
    },
    {
      "epoch": 0.06349769421780774,
      "grad_norm": 0.23465241491794586,
      "learning_rate": 5.644712397541959e-05,
      "loss": 0.4525,
      "num_tokens": 339999650.0,
      "step": 537
    },
    {
      "epoch": 0.0636159394584368,
      "grad_norm": 0.26618894934654236,
      "learning_rate": 5.643170790892608e-05,
      "loss": 0.4114,
      "num_tokens": 340634297.0,
      "step": 538
    },
    {
      "epoch": 0.06373418469906586,
      "grad_norm": 0.23005127906799316,
      "learning_rate": 5.6416260835397386e-05,
      "loss": 0.4356,
      "num_tokens": 341269246.0,
      "step": 539
    },
    {
      "epoch": 0.06385242993969492,
      "grad_norm": 0.23913200199604034,
      "learning_rate": 5.6400782775274506e-05,
      "loss": 0.4225,
      "num_tokens": 341900107.0,
      "step": 540
    },
    {
      "epoch": 0.063970675180324,
      "grad_norm": 0.2653907835483551,
      "learning_rate": 5.638527374903947e-05,
      "loss": 0.4265,
      "num_tokens": 342539152.0,
      "step": 541
    },
    {
      "epoch": 0.06408892042095306,
      "grad_norm": 0.2343796193599701,
      "learning_rate": 5.6369733777215295e-05,
      "loss": 0.4254,
      "num_tokens": 343166271.0,
      "step": 542
    },
    {
      "epoch": 0.06420716566158212,
      "grad_norm": 0.21634334325790405,
      "learning_rate": 5.635416288036591e-05,
      "loss": 0.3838,
      "num_tokens": 343801657.0,
      "step": 543
    },
    {
      "epoch": 0.06432541090221118,
      "grad_norm": 0.22020314633846283,
      "learning_rate": 5.633856107909619e-05,
      "loss": 0.399,
      "num_tokens": 344436926.0,
      "step": 544
    },
    {
      "epoch": 0.06444365614284026,
      "grad_norm": 0.20335398614406586,
      "learning_rate": 5.632292839405191e-05,
      "loss": 0.3836,
      "num_tokens": 345071661.0,
      "step": 545
    },
    {
      "epoch": 0.06456190138346932,
      "grad_norm": 0.24160657823085785,
      "learning_rate": 5.63072648459197e-05,
      "loss": 0.4674,
      "num_tokens": 345710262.0,
      "step": 546
    },
    {
      "epoch": 0.06468014662409838,
      "grad_norm": 0.20083579421043396,
      "learning_rate": 5.629157045542704e-05,
      "loss": 0.4077,
      "num_tokens": 346345760.0,
      "step": 547
    },
    {
      "epoch": 0.06479839186472744,
      "grad_norm": 0.23045681416988373,
      "learning_rate": 5.627584524334222e-05,
      "loss": 0.3846,
      "num_tokens": 346978472.0,
      "step": 548
    },
    {
      "epoch": 0.06491663710535651,
      "grad_norm": 0.22518788278102875,
      "learning_rate": 5.626008923047432e-05,
      "loss": 0.3981,
      "num_tokens": 347615067.0,
      "step": 549
    },
    {
      "epoch": 0.06503488234598558,
      "grad_norm": 0.25393834710121155,
      "learning_rate": 5.624430243767317e-05,
      "loss": 0.421,
      "num_tokens": 348251545.0,
      "step": 550
    },
    {
      "epoch": 0.06515312758661464,
      "grad_norm": 0.20160511136054993,
      "learning_rate": 5.6228484885829335e-05,
      "loss": 0.4116,
      "num_tokens": 348888946.0,
      "step": 551
    },
    {
      "epoch": 0.0652713728272437,
      "grad_norm": 0.24297155439853668,
      "learning_rate": 5.621263659587408e-05,
      "loss": 0.4485,
      "num_tokens": 349519613.0,
      "step": 552
    },
    {
      "epoch": 0.06538961806787277,
      "grad_norm": 0.22340096533298492,
      "learning_rate": 5.619675758877937e-05,
      "loss": 0.4006,
      "num_tokens": 350152254.0,
      "step": 553
    },
    {
      "epoch": 0.06550786330850183,
      "grad_norm": 0.21662257611751556,
      "learning_rate": 5.618084788555778e-05,
      "loss": 0.3983,
      "num_tokens": 350786547.0,
      "step": 554
    },
    {
      "epoch": 0.0656261085491309,
      "grad_norm": 0.245235413312912,
      "learning_rate": 5.616490750726252e-05,
      "loss": 0.4061,
      "num_tokens": 351422383.0,
      "step": 555
    },
    {
      "epoch": 0.06574435378975996,
      "grad_norm": 0.23087024688720703,
      "learning_rate": 5.614893647498741e-05,
      "loss": 0.4178,
      "num_tokens": 352061087.0,
      "step": 556
    },
    {
      "epoch": 0.06586259903038903,
      "grad_norm": 0.2400379627943039,
      "learning_rate": 5.61329348098668e-05,
      "loss": 0.3789,
      "num_tokens": 352697788.0,
      "step": 557
    },
    {
      "epoch": 0.0659808442710181,
      "grad_norm": 0.21535944938659668,
      "learning_rate": 5.611690253307561e-05,
      "loss": 0.3992,
      "num_tokens": 353327836.0,
      "step": 558
    },
    {
      "epoch": 0.06609908951164715,
      "grad_norm": 0.21479098498821259,
      "learning_rate": 5.610083966582926e-05,
      "loss": 0.3911,
      "num_tokens": 353963351.0,
      "step": 559
    },
    {
      "epoch": 0.06621733475227622,
      "grad_norm": 0.23319794237613678,
      "learning_rate": 5.608474622938362e-05,
      "loss": 0.4251,
      "num_tokens": 354576403.0,
      "step": 560
    },
    {
      "epoch": 0.06633557999290529,
      "grad_norm": 0.22759908437728882,
      "learning_rate": 5.6068622245035045e-05,
      "loss": 0.4277,
      "num_tokens": 355211385.0,
      "step": 561
    },
    {
      "epoch": 0.06645382523353435,
      "grad_norm": 0.24206581711769104,
      "learning_rate": 5.605246773412032e-05,
      "loss": 0.4293,
      "num_tokens": 355842242.0,
      "step": 562
    },
    {
      "epoch": 0.06657207047416341,
      "grad_norm": 0.21902194619178772,
      "learning_rate": 5.6036282718016583e-05,
      "loss": 0.3806,
      "num_tokens": 356473678.0,
      "step": 563
    },
    {
      "epoch": 0.06669031571479248,
      "grad_norm": 0.21064871549606323,
      "learning_rate": 5.6020067218141395e-05,
      "loss": 0.3713,
      "num_tokens": 357112492.0,
      "step": 564
    },
    {
      "epoch": 0.06680856095542155,
      "grad_norm": 0.24632275104522705,
      "learning_rate": 5.6003821255952606e-05,
      "loss": 0.4005,
      "num_tokens": 357748442.0,
      "step": 565
    },
    {
      "epoch": 0.06692680619605061,
      "grad_norm": 0.2036793977022171,
      "learning_rate": 5.598754485294841e-05,
      "loss": 0.3638,
      "num_tokens": 358387012.0,
      "step": 566
    },
    {
      "epoch": 0.06704505143667967,
      "grad_norm": 0.1924084573984146,
      "learning_rate": 5.5971238030667256e-05,
      "loss": 0.3772,
      "num_tokens": 359025550.0,
      "step": 567
    },
    {
      "epoch": 0.06716329667730873,
      "grad_norm": 0.25849267840385437,
      "learning_rate": 5.5954900810687883e-05,
      "loss": 0.4201,
      "num_tokens": 359661915.0,
      "step": 568
    },
    {
      "epoch": 0.06728154191793781,
      "grad_norm": 0.20558328926563263,
      "learning_rate": 5.593853321462923e-05,
      "loss": 0.3806,
      "num_tokens": 360295739.0,
      "step": 569
    },
    {
      "epoch": 0.06739978715856687,
      "grad_norm": 0.2169775664806366,
      "learning_rate": 5.592213526415041e-05,
      "loss": 0.436,
      "num_tokens": 360929829.0,
      "step": 570
    },
    {
      "epoch": 0.06751803239919593,
      "grad_norm": 0.2151200920343399,
      "learning_rate": 5.5905706980950766e-05,
      "loss": 0.3729,
      "num_tokens": 361568418.0,
      "step": 571
    },
    {
      "epoch": 0.067636277639825,
      "grad_norm": 0.20888011157512665,
      "learning_rate": 5.588924838676973e-05,
      "loss": 0.3616,
      "num_tokens": 362197303.0,
      "step": 572
    },
    {
      "epoch": 0.06775452288045407,
      "grad_norm": 0.22360935807228088,
      "learning_rate": 5.587275950338685e-05,
      "loss": 0.4044,
      "num_tokens": 362825904.0,
      "step": 573
    },
    {
      "epoch": 0.06787276812108313,
      "grad_norm": 0.2185496687889099,
      "learning_rate": 5.5856240352621766e-05,
      "loss": 0.3965,
      "num_tokens": 363465140.0,
      "step": 574
    },
    {
      "epoch": 0.06799101336171219,
      "grad_norm": 0.19799013435840607,
      "learning_rate": 5.583969095633417e-05,
      "loss": 0.3984,
      "num_tokens": 364097168.0,
      "step": 575
    },
    {
      "epoch": 0.06810925860234125,
      "grad_norm": 0.20666642487049103,
      "learning_rate": 5.582311133642379e-05,
      "loss": 0.3763,
      "num_tokens": 364728470.0,
      "step": 576
    },
    {
      "epoch": 0.06822750384297033,
      "grad_norm": 0.20582424104213715,
      "learning_rate": 5.580650151483033e-05,
      "loss": 0.4387,
      "num_tokens": 365363497.0,
      "step": 577
    },
    {
      "epoch": 0.06834574908359939,
      "grad_norm": 0.2404540628194809,
      "learning_rate": 5.578986151353345e-05,
      "loss": 0.4566,
      "num_tokens": 366000699.0,
      "step": 578
    },
    {
      "epoch": 0.06846399432422845,
      "grad_norm": 0.19141684472560883,
      "learning_rate": 5.577319135455278e-05,
      "loss": 0.3639,
      "num_tokens": 366624851.0,
      "step": 579
    },
    {
      "epoch": 0.06858223956485751,
      "grad_norm": 0.22023002803325653,
      "learning_rate": 5.575649105994782e-05,
      "loss": 0.395,
      "num_tokens": 367261099.0,
      "step": 580
    },
    {
      "epoch": 0.06870048480548657,
      "grad_norm": 0.20336921513080597,
      "learning_rate": 5.5739760651817984e-05,
      "loss": 0.3665,
      "num_tokens": 367897458.0,
      "step": 581
    },
    {
      "epoch": 0.06881873004611565,
      "grad_norm": 0.23487062752246857,
      "learning_rate": 5.572300015230252e-05,
      "loss": 0.4354,
      "num_tokens": 368534927.0,
      "step": 582
    },
    {
      "epoch": 0.06893697528674471,
      "grad_norm": 0.233329176902771,
      "learning_rate": 5.570620958358048e-05,
      "loss": 0.426,
      "num_tokens": 369169358.0,
      "step": 583
    },
    {
      "epoch": 0.06905522052737377,
      "grad_norm": 0.21234264969825745,
      "learning_rate": 5.568938896787073e-05,
      "loss": 0.416,
      "num_tokens": 369806601.0,
      "step": 584
    },
    {
      "epoch": 0.06917346576800283,
      "grad_norm": 0.21879243850708008,
      "learning_rate": 5.567253832743187e-05,
      "loss": 0.4039,
      "num_tokens": 370442728.0,
      "step": 585
    },
    {
      "epoch": 0.0692917110086319,
      "grad_norm": 0.23026061058044434,
      "learning_rate": 5.565565768456228e-05,
      "loss": 0.415,
      "num_tokens": 371079275.0,
      "step": 586
    },
    {
      "epoch": 0.06940995624926097,
      "grad_norm": 0.2231062948703766,
      "learning_rate": 5.563874706159999e-05,
      "loss": 0.4094,
      "num_tokens": 371716282.0,
      "step": 587
    },
    {
      "epoch": 0.06952820148989003,
      "grad_norm": 0.21120983362197876,
      "learning_rate": 5.562180648092273e-05,
      "loss": 0.4116,
      "num_tokens": 372347661.0,
      "step": 588
    },
    {
      "epoch": 0.06964644673051909,
      "grad_norm": 0.21369966864585876,
      "learning_rate": 5.5604835964947864e-05,
      "loss": 0.4005,
      "num_tokens": 372977936.0,
      "step": 589
    },
    {
      "epoch": 0.06976469197114817,
      "grad_norm": 0.2117834985256195,
      "learning_rate": 5.558783553613237e-05,
      "loss": 0.42,
      "num_tokens": 373613192.0,
      "step": 590
    },
    {
      "epoch": 0.06988293721177723,
      "grad_norm": 0.2446117252111435,
      "learning_rate": 5.5570805216972816e-05,
      "loss": 0.4138,
      "num_tokens": 374251716.0,
      "step": 591
    },
    {
      "epoch": 0.07000118245240629,
      "grad_norm": 0.21218879520893097,
      "learning_rate": 5.555374503000531e-05,
      "loss": 0.3687,
      "num_tokens": 374880470.0,
      "step": 592
    },
    {
      "epoch": 0.07011942769303535,
      "grad_norm": 0.20297327637672424,
      "learning_rate": 5.5536654997805504e-05,
      "loss": 0.4106,
      "num_tokens": 375518014.0,
      "step": 593
    },
    {
      "epoch": 0.07023767293366442,
      "grad_norm": 0.23108522593975067,
      "learning_rate": 5.551953514298852e-05,
      "loss": 0.4121,
      "num_tokens": 376156899.0,
      "step": 594
    },
    {
      "epoch": 0.07035591817429349,
      "grad_norm": 0.20212985575199127,
      "learning_rate": 5.550238548820898e-05,
      "loss": 0.3945,
      "num_tokens": 376796416.0,
      "step": 595
    },
    {
      "epoch": 0.07047416341492255,
      "grad_norm": 0.22493824362754822,
      "learning_rate": 5.548520605616088e-05,
      "loss": 0.3948,
      "num_tokens": 377430299.0,
      "step": 596
    },
    {
      "epoch": 0.07059240865555161,
      "grad_norm": 0.1889137625694275,
      "learning_rate": 5.546799686957769e-05,
      "loss": 0.3441,
      "num_tokens": 378036010.0,
      "step": 597
    },
    {
      "epoch": 0.07071065389618068,
      "grad_norm": 0.24320292472839355,
      "learning_rate": 5.545075795123221e-05,
      "loss": 0.4358,
      "num_tokens": 378662576.0,
      "step": 598
    },
    {
      "epoch": 0.07082889913680974,
      "grad_norm": 0.20845460891723633,
      "learning_rate": 5.543348932393659e-05,
      "loss": 0.3632,
      "num_tokens": 379292706.0,
      "step": 599
    },
    {
      "epoch": 0.0709471443774388,
      "grad_norm": 0.2712063193321228,
      "learning_rate": 5.54161910105423e-05,
      "loss": 0.4241,
      "num_tokens": 379923752.0,
      "step": 600
    },
    {
      "epoch": 0.07106538961806787,
      "grad_norm": 0.2057293802499771,
      "learning_rate": 5.5398863033940096e-05,
      "loss": 0.3938,
      "num_tokens": 380556540.0,
      "step": 601
    },
    {
      "epoch": 0.07118363485869694,
      "grad_norm": 0.2481277585029602,
      "learning_rate": 5.538150541705998e-05,
      "loss": 0.3685,
      "num_tokens": 381191645.0,
      "step": 602
    },
    {
      "epoch": 0.071301880099326,
      "grad_norm": 0.2418372929096222,
      "learning_rate": 5.536411818287119e-05,
      "loss": 0.4251,
      "num_tokens": 381827763.0,
      "step": 603
    },
    {
      "epoch": 0.07142012533995507,
      "grad_norm": 0.19724474847316742,
      "learning_rate": 5.5346701354382125e-05,
      "loss": 0.3906,
      "num_tokens": 382466825.0,
      "step": 604
    },
    {
      "epoch": 0.07153837058058413,
      "grad_norm": 0.2242296040058136,
      "learning_rate": 5.532925495464039e-05,
      "loss": 0.3925,
      "num_tokens": 383103907.0,
      "step": 605
    },
    {
      "epoch": 0.0716566158212132,
      "grad_norm": 0.22040408849716187,
      "learning_rate": 5.5311779006732676e-05,
      "loss": 0.4127,
      "num_tokens": 383736705.0,
      "step": 606
    },
    {
      "epoch": 0.07177486106184226,
      "grad_norm": 0.23752287030220032,
      "learning_rate": 5.529427353378482e-05,
      "loss": 0.4197,
      "num_tokens": 384376310.0,
      "step": 607
    },
    {
      "epoch": 0.07189310630247132,
      "grad_norm": 0.21531735360622406,
      "learning_rate": 5.5276738558961686e-05,
      "loss": 0.3928,
      "num_tokens": 385007094.0,
      "step": 608
    },
    {
      "epoch": 0.07201135154310039,
      "grad_norm": 0.20882567763328552,
      "learning_rate": 5.525917410546721e-05,
      "loss": 0.3719,
      "num_tokens": 385646534.0,
      "step": 609
    },
    {
      "epoch": 0.07212959678372946,
      "grad_norm": 0.19994843006134033,
      "learning_rate": 5.5241580196544334e-05,
      "loss": 0.3651,
      "num_tokens": 386283106.0,
      "step": 610
    },
    {
      "epoch": 0.07224784202435852,
      "grad_norm": 0.22568222880363464,
      "learning_rate": 5.522395685547495e-05,
      "loss": 0.363,
      "num_tokens": 386916395.0,
      "step": 611
    },
    {
      "epoch": 0.07236608726498758,
      "grad_norm": 0.19781558215618134,
      "learning_rate": 5.520630410557994e-05,
      "loss": 0.4108,
      "num_tokens": 387553614.0,
      "step": 612
    },
    {
      "epoch": 0.07248433250561664,
      "grad_norm": 0.20043326914310455,
      "learning_rate": 5.5188621970219056e-05,
      "loss": 0.3767,
      "num_tokens": 388189908.0,
      "step": 613
    },
    {
      "epoch": 0.07260257774624572,
      "grad_norm": 0.22510674595832825,
      "learning_rate": 5.517091047279096e-05,
      "loss": 0.4186,
      "num_tokens": 388827274.0,
      "step": 614
    },
    {
      "epoch": 0.07272082298687478,
      "grad_norm": 0.19270777702331543,
      "learning_rate": 5.5153169636733194e-05,
      "loss": 0.3854,
      "num_tokens": 389460786.0,
      "step": 615
    },
    {
      "epoch": 0.07283906822750384,
      "grad_norm": 0.2161119431257248,
      "learning_rate": 5.5135399485522075e-05,
      "loss": 0.4093,
      "num_tokens": 390066154.0,
      "step": 616
    },
    {
      "epoch": 0.0729573134681329,
      "grad_norm": 0.20279474556446075,
      "learning_rate": 5.511760004267274e-05,
      "loss": 0.3766,
      "num_tokens": 390700370.0,
      "step": 617
    },
    {
      "epoch": 0.07307555870876198,
      "grad_norm": 0.23120693862438202,
      "learning_rate": 5.509977133173908e-05,
      "loss": 0.3815,
      "num_tokens": 391333815.0,
      "step": 618
    },
    {
      "epoch": 0.07319380394939104,
      "grad_norm": 0.21448132395744324,
      "learning_rate": 5.508191337631373e-05,
      "loss": 0.399,
      "num_tokens": 391972641.0,
      "step": 619
    },
    {
      "epoch": 0.0733120491900201,
      "grad_norm": 0.23450714349746704,
      "learning_rate": 5.5064026200028e-05,
      "loss": 0.4233,
      "num_tokens": 392607122.0,
      "step": 620
    },
    {
      "epoch": 0.07343029443064916,
      "grad_norm": 0.18135346472263336,
      "learning_rate": 5.5046109826551886e-05,
      "loss": 0.3721,
      "num_tokens": 393234821.0,
      "step": 621
    },
    {
      "epoch": 0.07354853967127824,
      "grad_norm": 0.26815736293792725,
      "learning_rate": 5.502816427959401e-05,
      "loss": 0.375,
      "num_tokens": 393867338.0,
      "step": 622
    },
    {
      "epoch": 0.0736667849119073,
      "grad_norm": 0.20579469203948975,
      "learning_rate": 5.5010189582901614e-05,
      "loss": 0.387,
      "num_tokens": 394505968.0,
      "step": 623
    },
    {
      "epoch": 0.07378503015253636,
      "grad_norm": 0.2314739227294922,
      "learning_rate": 5.499218576026049e-05,
      "loss": 0.4327,
      "num_tokens": 395143798.0,
      "step": 624
    },
    {
      "epoch": 0.07390327539316542,
      "grad_norm": 0.2270359843969345,
      "learning_rate": 5.497415283549501e-05,
      "loss": 0.3791,
      "num_tokens": 395781916.0,
      "step": 625
    },
    {
      "epoch": 0.0740215206337945,
      "grad_norm": 0.22188624739646912,
      "learning_rate": 5.4956090832468006e-05,
      "loss": 0.4233,
      "num_tokens": 396419440.0,
      "step": 626
    },
    {
      "epoch": 0.07413976587442356,
      "grad_norm": 0.18612177670001984,
      "learning_rate": 5.4937999775080824e-05,
      "loss": 0.3638,
      "num_tokens": 397053137.0,
      "step": 627
    },
    {
      "epoch": 0.07425801111505262,
      "grad_norm": 0.21651092171669006,
      "learning_rate": 5.4919879687273255e-05,
      "loss": 0.4044,
      "num_tokens": 397685003.0,
      "step": 628
    },
    {
      "epoch": 0.07437625635568168,
      "grad_norm": 0.19978873431682587,
      "learning_rate": 5.490173059302349e-05,
      "loss": 0.4171,
      "num_tokens": 398323686.0,
      "step": 629
    },
    {
      "epoch": 0.07449450159631076,
      "grad_norm": 0.21592815220355988,
      "learning_rate": 5.488355251634813e-05,
      "loss": 0.4154,
      "num_tokens": 398958685.0,
      "step": 630
    },
    {
      "epoch": 0.07461274683693982,
      "grad_norm": 0.2020513266324997,
      "learning_rate": 5.48653454813021e-05,
      "loss": 0.4077,
      "num_tokens": 399595823.0,
      "step": 631
    },
    {
      "epoch": 0.07473099207756888,
      "grad_norm": 0.2267449051141739,
      "learning_rate": 5.484710951197866e-05,
      "loss": 0.3978,
      "num_tokens": 400234363.0,
      "step": 632
    },
    {
      "epoch": 0.07484923731819794,
      "grad_norm": 0.22091686725616455,
      "learning_rate": 5.482884463250935e-05,
      "loss": 0.4122,
      "num_tokens": 400860312.0,
      "step": 633
    },
    {
      "epoch": 0.074967482558827,
      "grad_norm": 0.18706369400024414,
      "learning_rate": 5.481055086706399e-05,
      "loss": 0.405,
      "num_tokens": 401493314.0,
      "step": 634
    },
    {
      "epoch": 0.07508572779945608,
      "grad_norm": 0.22939816117286682,
      "learning_rate": 5.479222823985058e-05,
      "loss": 0.4171,
      "num_tokens": 402131410.0,
      "step": 635
    },
    {
      "epoch": 0.07520397304008514,
      "grad_norm": 0.20424328744411469,
      "learning_rate": 5.477387677511537e-05,
      "loss": 0.4301,
      "num_tokens": 402767694.0,
      "step": 636
    },
    {
      "epoch": 0.0753222182807142,
      "grad_norm": 0.19207920134067535,
      "learning_rate": 5.475549649714272e-05,
      "loss": 0.4156,
      "num_tokens": 403401170.0,
      "step": 637
    },
    {
      "epoch": 0.07544046352134326,
      "grad_norm": 0.2168489545583725,
      "learning_rate": 5.473708743025516e-05,
      "loss": 0.4025,
      "num_tokens": 404033223.0,
      "step": 638
    },
    {
      "epoch": 0.07555870876197233,
      "grad_norm": 0.20321166515350342,
      "learning_rate": 5.471864959881328e-05,
      "loss": 0.3945,
      "num_tokens": 404669142.0,
      "step": 639
    },
    {
      "epoch": 0.0756769540026014,
      "grad_norm": 0.20609459280967712,
      "learning_rate": 5.470018302721574e-05,
      "loss": 0.4091,
      "num_tokens": 405305698.0,
      "step": 640
    },
    {
      "epoch": 0.07579519924323046,
      "grad_norm": 0.2112593948841095,
      "learning_rate": 5.468168773989925e-05,
      "loss": 0.4354,
      "num_tokens": 405935366.0,
      "step": 641
    },
    {
      "epoch": 0.07591344448385952,
      "grad_norm": 0.21053996682167053,
      "learning_rate": 5.466316376133852e-05,
      "loss": 0.407,
      "num_tokens": 406563939.0,
      "step": 642
    },
    {
      "epoch": 0.0760316897244886,
      "grad_norm": 0.19004890322685242,
      "learning_rate": 5.46446111160462e-05,
      "loss": 0.3909,
      "num_tokens": 407201876.0,
      "step": 643
    },
    {
      "epoch": 0.07614993496511765,
      "grad_norm": 0.20329055190086365,
      "learning_rate": 5.4626029828572916e-05,
      "loss": 0.3846,
      "num_tokens": 407834876.0,
      "step": 644
    },
    {
      "epoch": 0.07626818020574672,
      "grad_norm": 0.19921205937862396,
      "learning_rate": 5.460741992350715e-05,
      "loss": 0.3864,
      "num_tokens": 408474126.0,
      "step": 645
    },
    {
      "epoch": 0.07638642544637578,
      "grad_norm": 0.19097398221492767,
      "learning_rate": 5.458878142547527e-05,
      "loss": 0.3468,
      "num_tokens": 409110752.0,
      "step": 646
    },
    {
      "epoch": 0.07650467068700485,
      "grad_norm": 0.22949428856372833,
      "learning_rate": 5.457011435914151e-05,
      "loss": 0.4106,
      "num_tokens": 409744230.0,
      "step": 647
    },
    {
      "epoch": 0.07662291592763391,
      "grad_norm": 0.2329481840133667,
      "learning_rate": 5.455141874920786e-05,
      "loss": 0.4005,
      "num_tokens": 410377730.0,
      "step": 648
    },
    {
      "epoch": 0.07674116116826298,
      "grad_norm": 0.21861277520656586,
      "learning_rate": 5.453269462041413e-05,
      "loss": 0.3794,
      "num_tokens": 411014041.0,
      "step": 649
    },
    {
      "epoch": 0.07685940640889204,
      "grad_norm": 0.25171056389808655,
      "learning_rate": 5.4513941997537826e-05,
      "loss": 0.389,
      "num_tokens": 411653738.0,
      "step": 650
    },
    {
      "epoch": 0.07697765164952111,
      "grad_norm": 0.20175261795520782,
      "learning_rate": 5.4495160905394185e-05,
      "loss": 0.3584,
      "num_tokens": 412288459.0,
      "step": 651
    },
    {
      "epoch": 0.07709589689015017,
      "grad_norm": 0.20550478994846344,
      "learning_rate": 5.447635136883611e-05,
      "loss": 0.3979,
      "num_tokens": 412924573.0,
      "step": 652
    },
    {
      "epoch": 0.07721414213077923,
      "grad_norm": 0.21349135041236877,
      "learning_rate": 5.445751341275414e-05,
      "loss": 0.353,
      "num_tokens": 413557882.0,
      "step": 653
    },
    {
      "epoch": 0.0773323873714083,
      "grad_norm": 0.20919151604175568,
      "learning_rate": 5.443864706207644e-05,
      "loss": 0.3545,
      "num_tokens": 414190765.0,
      "step": 654
    },
    {
      "epoch": 0.07745063261203737,
      "grad_norm": 0.17326898872852325,
      "learning_rate": 5.441975234176872e-05,
      "loss": 0.3625,
      "num_tokens": 414824481.0,
      "step": 655
    },
    {
      "epoch": 0.07756887785266643,
      "grad_norm": 0.22109591960906982,
      "learning_rate": 5.440082927683427e-05,
      "loss": 0.3867,
      "num_tokens": 415462470.0,
      "step": 656
    },
    {
      "epoch": 0.0776871230932955,
      "grad_norm": 0.21504008769989014,
      "learning_rate": 5.438187789231384e-05,
      "loss": 0.4169,
      "num_tokens": 416098823.0,
      "step": 657
    },
    {
      "epoch": 0.07780536833392455,
      "grad_norm": 0.21788012981414795,
      "learning_rate": 5.436289821328568e-05,
      "loss": 0.3969,
      "num_tokens": 416727909.0,
      "step": 658
    },
    {
      "epoch": 0.07792361357455363,
      "grad_norm": 0.2375272661447525,
      "learning_rate": 5.434389026486552e-05,
      "loss": 0.4216,
      "num_tokens": 417362676.0,
      "step": 659
    },
    {
      "epoch": 0.07804185881518269,
      "grad_norm": 0.20120349526405334,
      "learning_rate": 5.432485407220642e-05,
      "loss": 0.3754,
      "num_tokens": 417996258.0,
      "step": 660
    },
    {
      "epoch": 0.07816010405581175,
      "grad_norm": 0.20345494151115417,
      "learning_rate": 5.430578966049888e-05,
      "loss": 0.4174,
      "num_tokens": 418626034.0,
      "step": 661
    },
    {
      "epoch": 0.07827834929644081,
      "grad_norm": 0.22592638432979584,
      "learning_rate": 5.428669705497071e-05,
      "loss": 0.4421,
      "num_tokens": 419263225.0,
      "step": 662
    },
    {
      "epoch": 0.07839659453706989,
      "grad_norm": 0.18610456585884094,
      "learning_rate": 5.426757628088704e-05,
      "loss": 0.4022,
      "num_tokens": 419892898.0,
      "step": 663
    },
    {
      "epoch": 0.07851483977769895,
      "grad_norm": 0.17779679596424103,
      "learning_rate": 5.424842736355027e-05,
      "loss": 0.3886,
      "num_tokens": 420527528.0,
      "step": 664
    },
    {
      "epoch": 0.07863308501832801,
      "grad_norm": 0.198362335562706,
      "learning_rate": 5.422925032830005e-05,
      "loss": 0.3933,
      "num_tokens": 421156121.0,
      "step": 665
    },
    {
      "epoch": 0.07875133025895707,
      "grad_norm": 0.18584303557872772,
      "learning_rate": 5.421004520051323e-05,
      "loss": 0.3421,
      "num_tokens": 421790061.0,
      "step": 666
    },
    {
      "epoch": 0.07886957549958615,
      "grad_norm": 0.2071962207555771,
      "learning_rate": 5.419081200560384e-05,
      "loss": 0.4012,
      "num_tokens": 422426991.0,
      "step": 667
    },
    {
      "epoch": 0.07898782074021521,
      "grad_norm": 0.1821301132440567,
      "learning_rate": 5.417155076902304e-05,
      "loss": 0.3583,
      "num_tokens": 423063834.0,
      "step": 668
    },
    {
      "epoch": 0.07910606598084427,
      "grad_norm": 0.20297402143478394,
      "learning_rate": 5.415226151625912e-05,
      "loss": 0.3926,
      "num_tokens": 423703537.0,
      "step": 669
    },
    {
      "epoch": 0.07922431122147333,
      "grad_norm": 0.23203183710575104,
      "learning_rate": 5.413294427283741e-05,
      "loss": 0.4103,
      "num_tokens": 424330180.0,
      "step": 670
    },
    {
      "epoch": 0.0793425564621024,
      "grad_norm": 0.20411129295825958,
      "learning_rate": 5.411359906432033e-05,
      "loss": 0.4238,
      "num_tokens": 424966976.0,
      "step": 671
    },
    {
      "epoch": 0.07946080170273147,
      "grad_norm": 0.21121400594711304,
      "learning_rate": 5.409422591630725e-05,
      "loss": 0.3959,
      "num_tokens": 425604191.0,
      "step": 672
    },
    {
      "epoch": 0.07957904694336053,
      "grad_norm": 0.1865694373846054,
      "learning_rate": 5.4074824854434536e-05,
      "loss": 0.3455,
      "num_tokens": 426240218.0,
      "step": 673
    },
    {
      "epoch": 0.07969729218398959,
      "grad_norm": 0.2658561170101166,
      "learning_rate": 5.4055395904375516e-05,
      "loss": 0.4083,
      "num_tokens": 426878525.0,
      "step": 674
    },
    {
      "epoch": 0.07981553742461867,
      "grad_norm": 0.18216894567012787,
      "learning_rate": 5.403593909184038e-05,
      "loss": 0.3777,
      "num_tokens": 427512790.0,
      "step": 675
    },
    {
      "epoch": 0.07993378266524773,
      "grad_norm": 0.22485923767089844,
      "learning_rate": 5.401645444257622e-05,
      "loss": 0.4213,
      "num_tokens": 428150456.0,
      "step": 676
    },
    {
      "epoch": 0.08005202790587679,
      "grad_norm": 0.21150392293930054,
      "learning_rate": 5.399694198236695e-05,
      "loss": 0.3562,
      "num_tokens": 428783253.0,
      "step": 677
    },
    {
      "epoch": 0.08017027314650585,
      "grad_norm": 0.23367063701152802,
      "learning_rate": 5.397740173703328e-05,
      "loss": 0.3976,
      "num_tokens": 429419007.0,
      "step": 678
    },
    {
      "epoch": 0.08028851838713492,
      "grad_norm": 0.18307745456695557,
      "learning_rate": 5.395783373243272e-05,
      "loss": 0.3882,
      "num_tokens": 430051255.0,
      "step": 679
    },
    {
      "epoch": 0.08040676362776399,
      "grad_norm": 0.22259311378002167,
      "learning_rate": 5.3938237994459465e-05,
      "loss": 0.4348,
      "num_tokens": 430683994.0,
      "step": 680
    },
    {
      "epoch": 0.08052500886839305,
      "grad_norm": 0.18747752904891968,
      "learning_rate": 5.3918614549044446e-05,
      "loss": 0.3657,
      "num_tokens": 431314161.0,
      "step": 681
    },
    {
      "epoch": 0.08064325410902211,
      "grad_norm": 0.22098389267921448,
      "learning_rate": 5.389896342215524e-05,
      "loss": 0.389,
      "num_tokens": 431943550.0,
      "step": 682
    },
    {
      "epoch": 0.08076149934965117,
      "grad_norm": 0.18855495750904083,
      "learning_rate": 5.387928463979608e-05,
      "loss": 0.379,
      "num_tokens": 432580022.0,
      "step": 683
    },
    {
      "epoch": 0.08087974459028024,
      "grad_norm": 0.20843663811683655,
      "learning_rate": 5.385957822800776e-05,
      "loss": 0.4279,
      "num_tokens": 433215860.0,
      "step": 684
    },
    {
      "epoch": 0.0809979898309093,
      "grad_norm": 0.2124507576227188,
      "learning_rate": 5.3839844212867666e-05,
      "loss": 0.3809,
      "num_tokens": 433849766.0,
      "step": 685
    },
    {
      "epoch": 0.08111623507153837,
      "grad_norm": 0.17536836862564087,
      "learning_rate": 5.382008262048968e-05,
      "loss": 0.36,
      "num_tokens": 434484743.0,
      "step": 686
    },
    {
      "epoch": 0.08123448031216743,
      "grad_norm": 0.21414297819137573,
      "learning_rate": 5.3800293477024224e-05,
      "loss": 0.3713,
      "num_tokens": 435100372.0,
      "step": 687
    },
    {
      "epoch": 0.0813527255527965,
      "grad_norm": 0.21753741800785065,
      "learning_rate": 5.378047680865814e-05,
      "loss": 0.4386,
      "num_tokens": 435733891.0,
      "step": 688
    },
    {
      "epoch": 0.08147097079342557,
      "grad_norm": 0.21724680066108704,
      "learning_rate": 5.376063264161469e-05,
      "loss": 0.3765,
      "num_tokens": 436365690.0,
      "step": 689
    },
    {
      "epoch": 0.08158921603405463,
      "grad_norm": 0.21394267678260803,
      "learning_rate": 5.3740761002153536e-05,
      "loss": 0.3638,
      "num_tokens": 437003639.0,
      "step": 690
    },
    {
      "epoch": 0.08170746127468369,
      "grad_norm": 0.2070666402578354,
      "learning_rate": 5.372086191657073e-05,
      "loss": 0.3935,
      "num_tokens": 437642920.0,
      "step": 691
    },
    {
      "epoch": 0.08182570651531276,
      "grad_norm": 0.23084615170955658,
      "learning_rate": 5.370093541119857e-05,
      "loss": 0.3949,
      "num_tokens": 438277063.0,
      "step": 692
    },
    {
      "epoch": 0.08194395175594182,
      "grad_norm": 0.21036827564239502,
      "learning_rate": 5.368098151240571e-05,
      "loss": 0.4055,
      "num_tokens": 438915083.0,
      "step": 693
    },
    {
      "epoch": 0.08206219699657089,
      "grad_norm": 0.21798987686634064,
      "learning_rate": 5.366100024659699e-05,
      "loss": 0.4172,
      "num_tokens": 439554233.0,
      "step": 694
    },
    {
      "epoch": 0.08218044223719995,
      "grad_norm": 0.18871086835861206,
      "learning_rate": 5.3640991640213535e-05,
      "loss": 0.3707,
      "num_tokens": 440187716.0,
      "step": 695
    },
    {
      "epoch": 0.08229868747782902,
      "grad_norm": 0.20316803455352783,
      "learning_rate": 5.362095571973258e-05,
      "loss": 0.3978,
      "num_tokens": 440823462.0,
      "step": 696
    },
    {
      "epoch": 0.08241693271845808,
      "grad_norm": 0.19437344372272491,
      "learning_rate": 5.360089251166755e-05,
      "loss": 0.386,
      "num_tokens": 441461261.0,
      "step": 697
    },
    {
      "epoch": 0.08253517795908714,
      "grad_norm": 0.2129731923341751,
      "learning_rate": 5.358080204256795e-05,
      "loss": 0.4105,
      "num_tokens": 442092406.0,
      "step": 698
    },
    {
      "epoch": 0.0826534231997162,
      "grad_norm": 0.19358403980731964,
      "learning_rate": 5.356068433901938e-05,
      "loss": 0.415,
      "num_tokens": 442729228.0,
      "step": 699
    },
    {
      "epoch": 0.08277166844034528,
      "grad_norm": 0.20055876672267914,
      "learning_rate": 5.35405394276435e-05,
      "loss": 0.4277,
      "num_tokens": 443368453.0,
      "step": 700
    },
    {
      "epoch": 0.08288991368097434,
      "grad_norm": 0.19685010612010956,
      "learning_rate": 5.3520367335097916e-05,
      "loss": 0.3727,
      "num_tokens": 444004993.0,
      "step": 701
    },
    {
      "epoch": 0.0830081589216034,
      "grad_norm": 0.2335917055606842,
      "learning_rate": 5.350016808807624e-05,
      "loss": 0.4419,
      "num_tokens": 444641198.0,
      "step": 702
    },
    {
      "epoch": 0.08312640416223246,
      "grad_norm": 0.18580734729766846,
      "learning_rate": 5.3479941713308015e-05,
      "loss": 0.3951,
      "num_tokens": 445272456.0,
      "step": 703
    },
    {
      "epoch": 0.08324464940286154,
      "grad_norm": 0.23013712465763092,
      "learning_rate": 5.3459688237558684e-05,
      "loss": 0.4393,
      "num_tokens": 445909612.0,
      "step": 704
    },
    {
      "epoch": 0.0833628946434906,
      "grad_norm": 0.23703046143054962,
      "learning_rate": 5.343940768762954e-05,
      "loss": 0.4589,
      "num_tokens": 446544323.0,
      "step": 705
    },
    {
      "epoch": 0.08348113988411966,
      "grad_norm": 0.2044428586959839,
      "learning_rate": 5.341910009035771e-05,
      "loss": 0.4069,
      "num_tokens": 447177697.0,
      "step": 706
    },
    {
      "epoch": 0.08359938512474872,
      "grad_norm": 0.1987406462430954,
      "learning_rate": 5.3398765472616126e-05,
      "loss": 0.4164,
      "num_tokens": 447813578.0,
      "step": 707
    },
    {
      "epoch": 0.0837176303653778,
      "grad_norm": 0.19243036210536957,
      "learning_rate": 5.337840386131344e-05,
      "loss": 0.381,
      "num_tokens": 448448247.0,
      "step": 708
    },
    {
      "epoch": 0.08383587560600686,
      "grad_norm": 0.21122971177101135,
      "learning_rate": 5.335801528339407e-05,
      "loss": 0.4016,
      "num_tokens": 449081260.0,
      "step": 709
    },
    {
      "epoch": 0.08395412084663592,
      "grad_norm": 0.20589253306388855,
      "learning_rate": 5.3337599765838076e-05,
      "loss": 0.3636,
      "num_tokens": 449718960.0,
      "step": 710
    },
    {
      "epoch": 0.08407236608726498,
      "grad_norm": 0.22047452628612518,
      "learning_rate": 5.331715733566121e-05,
      "loss": 0.3761,
      "num_tokens": 450354221.0,
      "step": 711
    },
    {
      "epoch": 0.08419061132789406,
      "grad_norm": 0.22329314053058624,
      "learning_rate": 5.329668801991478e-05,
      "loss": 0.4277,
      "num_tokens": 450990864.0,
      "step": 712
    },
    {
      "epoch": 0.08430885656852312,
      "grad_norm": 0.21842564642429352,
      "learning_rate": 5.327619184568574e-05,
      "loss": 0.3584,
      "num_tokens": 451629963.0,
      "step": 713
    },
    {
      "epoch": 0.08442710180915218,
      "grad_norm": 0.201019287109375,
      "learning_rate": 5.325566884009654e-05,
      "loss": 0.3619,
      "num_tokens": 452263375.0,
      "step": 714
    },
    {
      "epoch": 0.08454534704978124,
      "grad_norm": 0.2327526956796646,
      "learning_rate": 5.323511903030513e-05,
      "loss": 0.4393,
      "num_tokens": 452899788.0,
      "step": 715
    },
    {
      "epoch": 0.08466359229041032,
      "grad_norm": 0.20655345916748047,
      "learning_rate": 5.3214542443504986e-05,
      "loss": 0.364,
      "num_tokens": 453532800.0,
      "step": 716
    },
    {
      "epoch": 0.08478183753103938,
      "grad_norm": 0.1914721131324768,
      "learning_rate": 5.319393910692494e-05,
      "loss": 0.4131,
      "num_tokens": 454167051.0,
      "step": 717
    },
    {
      "epoch": 0.08490008277166844,
      "grad_norm": 0.20321692526340485,
      "learning_rate": 5.317330904782927e-05,
      "loss": 0.4078,
      "num_tokens": 454801891.0,
      "step": 718
    },
    {
      "epoch": 0.0850183280122975,
      "grad_norm": 0.21024049818515778,
      "learning_rate": 5.315265229351762e-05,
      "loss": 0.3918,
      "num_tokens": 455434940.0,
      "step": 719
    },
    {
      "epoch": 0.08513657325292658,
      "grad_norm": 0.18730874359607697,
      "learning_rate": 5.313196887132494e-05,
      "loss": 0.4271,
      "num_tokens": 456070522.0,
      "step": 720
    },
    {
      "epoch": 0.08525481849355564,
      "grad_norm": 0.19962529838085175,
      "learning_rate": 5.311125880862147e-05,
      "loss": 0.4242,
      "num_tokens": 456700607.0,
      "step": 721
    },
    {
      "epoch": 0.0853730637341847,
      "grad_norm": 0.20009727776050568,
      "learning_rate": 5.309052213281273e-05,
      "loss": 0.3856,
      "num_tokens": 457333591.0,
      "step": 722
    },
    {
      "epoch": 0.08549130897481376,
      "grad_norm": 0.19860906898975372,
      "learning_rate": 5.30697588713394e-05,
      "loss": 0.4049,
      "num_tokens": 457972538.0,
      "step": 723
    },
    {
      "epoch": 0.08560955421544283,
      "grad_norm": 0.1968255639076233,
      "learning_rate": 5.304896905167741e-05,
      "loss": 0.3929,
      "num_tokens": 458608333.0,
      "step": 724
    },
    {
      "epoch": 0.0857277994560719,
      "grad_norm": 0.2130722552537918,
      "learning_rate": 5.302815270133777e-05,
      "loss": 0.3968,
      "num_tokens": 459238593.0,
      "step": 725
    },
    {
      "epoch": 0.08584604469670096,
      "grad_norm": 0.19868621230125427,
      "learning_rate": 5.300730984786665e-05,
      "loss": 0.3773,
      "num_tokens": 459848240.0,
      "step": 726
    },
    {
      "epoch": 0.08596428993733002,
      "grad_norm": 0.22184595465660095,
      "learning_rate": 5.298644051884526e-05,
      "loss": 0.4262,
      "num_tokens": 460483298.0,
      "step": 727
    },
    {
      "epoch": 0.0860825351779591,
      "grad_norm": 0.2032521814107895,
      "learning_rate": 5.2965544741889873e-05,
      "loss": 0.3885,
      "num_tokens": 461109500.0,
      "step": 728
    },
    {
      "epoch": 0.08620078041858815,
      "grad_norm": 0.22066350281238556,
      "learning_rate": 5.294462254465172e-05,
      "loss": 0.3901,
      "num_tokens": 461746594.0,
      "step": 729
    },
    {
      "epoch": 0.08631902565921722,
      "grad_norm": 0.19459012150764465,
      "learning_rate": 5.292367395481703e-05,
      "loss": 0.3939,
      "num_tokens": 462380799.0,
      "step": 730
    },
    {
      "epoch": 0.08643727089984628,
      "grad_norm": 0.19497635960578918,
      "learning_rate": 5.290269900010693e-05,
      "loss": 0.3987,
      "num_tokens": 463016862.0,
      "step": 731
    },
    {
      "epoch": 0.08655551614047535,
      "grad_norm": 0.2163160741329193,
      "learning_rate": 5.288169770827747e-05,
      "loss": 0.3683,
      "num_tokens": 463650306.0,
      "step": 732
    },
    {
      "epoch": 0.08667376138110441,
      "grad_norm": 0.2008008509874344,
      "learning_rate": 5.2860670107119505e-05,
      "loss": 0.389,
      "num_tokens": 464282371.0,
      "step": 733
    },
    {
      "epoch": 0.08679200662173348,
      "grad_norm": 0.20270873606204987,
      "learning_rate": 5.283961622445875e-05,
      "loss": 0.3805,
      "num_tokens": 464918493.0,
      "step": 734
    },
    {
      "epoch": 0.08691025186236254,
      "grad_norm": 0.21384552121162415,
      "learning_rate": 5.2818536088155667e-05,
      "loss": 0.3871,
      "num_tokens": 465551682.0,
      "step": 735
    },
    {
      "epoch": 0.0870284971029916,
      "grad_norm": 0.20108221471309662,
      "learning_rate": 5.279742972610548e-05,
      "loss": 0.3855,
      "num_tokens": 466189560.0,
      "step": 736
    },
    {
      "epoch": 0.08714674234362067,
      "grad_norm": 0.21781690418720245,
      "learning_rate": 5.27762971662381e-05,
      "loss": 0.3879,
      "num_tokens": 466826010.0,
      "step": 737
    },
    {
      "epoch": 0.08726498758424973,
      "grad_norm": 0.1880456656217575,
      "learning_rate": 5.2755138436518114e-05,
      "loss": 0.3571,
      "num_tokens": 467461700.0,
      "step": 738
    },
    {
      "epoch": 0.0873832328248788,
      "grad_norm": 0.1964695155620575,
      "learning_rate": 5.273395356494476e-05,
      "loss": 0.345,
      "num_tokens": 468097256.0,
      "step": 739
    },
    {
      "epoch": 0.08750147806550786,
      "grad_norm": 0.22812491655349731,
      "learning_rate": 5.271274257955181e-05,
      "loss": 0.4191,
      "num_tokens": 468731535.0,
      "step": 740
    },
    {
      "epoch": 0.08761972330613693,
      "grad_norm": 0.2017606943845749,
      "learning_rate": 5.269150550840766e-05,
      "loss": 0.3663,
      "num_tokens": 469363357.0,
      "step": 741
    },
    {
      "epoch": 0.087737968546766,
      "grad_norm": 0.19828036427497864,
      "learning_rate": 5.267024237961519e-05,
      "loss": 0.3901,
      "num_tokens": 469996228.0,
      "step": 742
    },
    {
      "epoch": 0.08785621378739505,
      "grad_norm": 0.19734492897987366,
      "learning_rate": 5.264895322131176e-05,
      "loss": 0.4177,
      "num_tokens": 470628575.0,
      "step": 743
    },
    {
      "epoch": 0.08797445902802412,
      "grad_norm": 0.19054393470287323,
      "learning_rate": 5.2627638061669204e-05,
      "loss": 0.3992,
      "num_tokens": 471265485.0,
      "step": 744
    },
    {
      "epoch": 0.08809270426865319,
      "grad_norm": 0.18877829611301422,
      "learning_rate": 5.260629692889371e-05,
      "loss": 0.3885,
      "num_tokens": 471898145.0,
      "step": 745
    },
    {
      "epoch": 0.08821094950928225,
      "grad_norm": 0.19310103356838226,
      "learning_rate": 5.258492985122588e-05,
      "loss": 0.3863,
      "num_tokens": 472533209.0,
      "step": 746
    },
    {
      "epoch": 0.08832919474991131,
      "grad_norm": 0.19495335221290588,
      "learning_rate": 5.2563536856940655e-05,
      "loss": 0.3962,
      "num_tokens": 473171790.0,
      "step": 747
    },
    {
      "epoch": 0.08844743999054037,
      "grad_norm": 0.1987488865852356,
      "learning_rate": 5.254211797434724e-05,
      "loss": 0.3608,
      "num_tokens": 473806600.0,
      "step": 748
    },
    {
      "epoch": 0.08856568523116945,
      "grad_norm": 0.20719455182552338,
      "learning_rate": 5.2520673231789095e-05,
      "loss": 0.3959,
      "num_tokens": 474446221.0,
      "step": 749
    },
    {
      "epoch": 0.08868393047179851,
      "grad_norm": 0.19631238281726837,
      "learning_rate": 5.2499202657643935e-05,
      "loss": 0.3641,
      "num_tokens": 475080946.0,
      "step": 750
    },
    {
      "epoch": 0.08880217571242757,
      "grad_norm": 0.17641642689704895,
      "learning_rate": 5.247770628032363e-05,
      "loss": 0.3358,
      "num_tokens": 475715589.0,
      "step": 751
    },
    {
      "epoch": 0.08892042095305663,
      "grad_norm": 0.21652136743068695,
      "learning_rate": 5.245618412827422e-05,
      "loss": 0.3982,
      "num_tokens": 476350099.0,
      "step": 752
    },
    {
      "epoch": 0.08903866619368571,
      "grad_norm": 0.20369255542755127,
      "learning_rate": 5.243463622997583e-05,
      "loss": 0.4119,
      "num_tokens": 476986608.0,
      "step": 753
    },
    {
      "epoch": 0.08915691143431477,
      "grad_norm": 0.18806983530521393,
      "learning_rate": 5.241306261394264e-05,
      "loss": 0.3947,
      "num_tokens": 477626286.0,
      "step": 754
    },
    {
      "epoch": 0.08927515667494383,
      "grad_norm": 0.20750251412391663,
      "learning_rate": 5.23914633087229e-05,
      "loss": 0.3962,
      "num_tokens": 478262327.0,
      "step": 755
    },
    {
      "epoch": 0.08939340191557289,
      "grad_norm": 0.20222771167755127,
      "learning_rate": 5.2369838342898844e-05,
      "loss": 0.3714,
      "num_tokens": 478899878.0,
      "step": 756
    },
    {
      "epoch": 0.08951164715620197,
      "grad_norm": 0.20789285004138947,
      "learning_rate": 5.2348187745086656e-05,
      "loss": 0.3862,
      "num_tokens": 479538362.0,
      "step": 757
    },
    {
      "epoch": 0.08962989239683103,
      "grad_norm": 0.19300925731658936,
      "learning_rate": 5.232651154393643e-05,
      "loss": 0.374,
      "num_tokens": 480177217.0,
      "step": 758
    },
    {
      "epoch": 0.08974813763746009,
      "grad_norm": 0.22329019010066986,
      "learning_rate": 5.2304809768132146e-05,
      "loss": 0.3966,
      "num_tokens": 480809891.0,
      "step": 759
    },
    {
      "epoch": 0.08986638287808915,
      "grad_norm": 0.20343434810638428,
      "learning_rate": 5.2283082446391654e-05,
      "loss": 0.3878,
      "num_tokens": 481446104.0,
      "step": 760
    },
    {
      "epoch": 0.08998462811871823,
      "grad_norm": 0.1930297464132309,
      "learning_rate": 5.226132960746656e-05,
      "loss": 0.3688,
      "num_tokens": 482084578.0,
      "step": 761
    },
    {
      "epoch": 0.09010287335934729,
      "grad_norm": 0.203902468085289,
      "learning_rate": 5.223955128014227e-05,
      "loss": 0.3799,
      "num_tokens": 482713767.0,
      "step": 762
    },
    {
      "epoch": 0.09022111859997635,
      "grad_norm": 0.23576878011226654,
      "learning_rate": 5.221774749323792e-05,
      "loss": 0.448,
      "num_tokens": 483353280.0,
      "step": 763
    },
    {
      "epoch": 0.09033936384060541,
      "grad_norm": 0.17504672706127167,
      "learning_rate": 5.219591827560631e-05,
      "loss": 0.3756,
      "num_tokens": 483985107.0,
      "step": 764
    },
    {
      "epoch": 0.09045760908123449,
      "grad_norm": 0.2097293883562088,
      "learning_rate": 5.2174063656133925e-05,
      "loss": 0.398,
      "num_tokens": 484615672.0,
      "step": 765
    },
    {
      "epoch": 0.09057585432186355,
      "grad_norm": 0.18698710203170776,
      "learning_rate": 5.215218366374085e-05,
      "loss": 0.3941,
      "num_tokens": 485253394.0,
      "step": 766
    },
    {
      "epoch": 0.09069409956249261,
      "grad_norm": 0.19252699613571167,
      "learning_rate": 5.213027832738073e-05,
      "loss": 0.3997,
      "num_tokens": 485886884.0,
      "step": 767
    },
    {
      "epoch": 0.09081234480312167,
      "grad_norm": 0.19506065547466278,
      "learning_rate": 5.2108347676040765e-05,
      "loss": 0.3886,
      "num_tokens": 486516495.0,
      "step": 768
    },
    {
      "epoch": 0.09093059004375074,
      "grad_norm": 0.2292434573173523,
      "learning_rate": 5.2086391738741674e-05,
      "loss": 0.4368,
      "num_tokens": 487151670.0,
      "step": 769
    },
    {
      "epoch": 0.0910488352843798,
      "grad_norm": 0.23250269889831543,
      "learning_rate": 5.206441054453758e-05,
      "loss": 0.3943,
      "num_tokens": 487786678.0,
      "step": 770
    },
    {
      "epoch": 0.09116708052500887,
      "grad_norm": 0.2288888543844223,
      "learning_rate": 5.204240412251609e-05,
      "loss": 0.4106,
      "num_tokens": 488423430.0,
      "step": 771
    },
    {
      "epoch": 0.09128532576563793,
      "grad_norm": 0.20469781756401062,
      "learning_rate": 5.2020372501798145e-05,
      "loss": 0.3816,
      "num_tokens": 489055057.0,
      "step": 772
    },
    {
      "epoch": 0.091403571006267,
      "grad_norm": 0.18626539409160614,
      "learning_rate": 5.199831571153806e-05,
      "loss": 0.3727,
      "num_tokens": 489690318.0,
      "step": 773
    },
    {
      "epoch": 0.09152181624689606,
      "grad_norm": 0.2176947295665741,
      "learning_rate": 5.197623378092347e-05,
      "loss": 0.4235,
      "num_tokens": 490315503.0,
      "step": 774
    },
    {
      "epoch": 0.09164006148752513,
      "grad_norm": 0.19569160044193268,
      "learning_rate": 5.1954126739175244e-05,
      "loss": 0.345,
      "num_tokens": 490951658.0,
      "step": 775
    },
    {
      "epoch": 0.09175830672815419,
      "grad_norm": 0.1919584423303604,
      "learning_rate": 5.193199461554748e-05,
      "loss": 0.3496,
      "num_tokens": 491582936.0,
      "step": 776
    },
    {
      "epoch": 0.09187655196878326,
      "grad_norm": 0.21266311407089233,
      "learning_rate": 5.1909837439327506e-05,
      "loss": 0.4171,
      "num_tokens": 492221808.0,
      "step": 777
    },
    {
      "epoch": 0.09199479720941232,
      "grad_norm": 0.21234160661697388,
      "learning_rate": 5.188765523983575e-05,
      "loss": 0.3959,
      "num_tokens": 492851928.0,
      "step": 778
    },
    {
      "epoch": 0.09211304245004139,
      "grad_norm": 0.19227167963981628,
      "learning_rate": 5.186544804642582e-05,
      "loss": 0.3497,
      "num_tokens": 493484521.0,
      "step": 779
    },
    {
      "epoch": 0.09223128769067045,
      "grad_norm": 0.22047032415866852,
      "learning_rate": 5.1843215888484313e-05,
      "loss": 0.4269,
      "num_tokens": 494120278.0,
      "step": 780
    },
    {
      "epoch": 0.09234953293129952,
      "grad_norm": 0.21378175914287567,
      "learning_rate": 5.182095879543093e-05,
      "loss": 0.421,
      "num_tokens": 494748375.0,
      "step": 781
    },
    {
      "epoch": 0.09246777817192858,
      "grad_norm": 0.24084855616092682,
      "learning_rate": 5.179867679671836e-05,
      "loss": 0.4098,
      "num_tokens": 495383913.0,
      "step": 782
    },
    {
      "epoch": 0.09258602341255764,
      "grad_norm": 0.20419980585575104,
      "learning_rate": 5.1776369921832214e-05,
      "loss": 0.3771,
      "num_tokens": 496008813.0,
      "step": 783
    },
    {
      "epoch": 0.0927042686531867,
      "grad_norm": 0.20603390038013458,
      "learning_rate": 5.1754038200291034e-05,
      "loss": 0.4053,
      "num_tokens": 496639648.0,
      "step": 784
    },
    {
      "epoch": 0.09282251389381578,
      "grad_norm": 0.19610314071178436,
      "learning_rate": 5.173168166164628e-05,
      "loss": 0.3518,
      "num_tokens": 497274171.0,
      "step": 785
    },
    {
      "epoch": 0.09294075913444484,
      "grad_norm": 0.19744454324245453,
      "learning_rate": 5.17093003354822e-05,
      "loss": 0.3817,
      "num_tokens": 497913138.0,
      "step": 786
    },
    {
      "epoch": 0.0930590043750739,
      "grad_norm": 0.2016357183456421,
      "learning_rate": 5.168689425141588e-05,
      "loss": 0.418,
      "num_tokens": 498551639.0,
      "step": 787
    },
    {
      "epoch": 0.09317724961570296,
      "grad_norm": 0.21235699951648712,
      "learning_rate": 5.166446343909713e-05,
      "loss": 0.4086,
      "num_tokens": 499182979.0,
      "step": 788
    },
    {
      "epoch": 0.09329549485633203,
      "grad_norm": 0.19116926193237305,
      "learning_rate": 5.164200792820852e-05,
      "loss": 0.3849,
      "num_tokens": 499817574.0,
      "step": 789
    },
    {
      "epoch": 0.0934137400969611,
      "grad_norm": 0.21546709537506104,
      "learning_rate": 5.1619527748465305e-05,
      "loss": 0.4332,
      "num_tokens": 500450987.0,
      "step": 790
    },
    {
      "epoch": 0.09353198533759016,
      "grad_norm": 0.2056332528591156,
      "learning_rate": 5.1597022929615354e-05,
      "loss": 0.4207,
      "num_tokens": 501086255.0,
      "step": 791
    },
    {
      "epoch": 0.09365023057821922,
      "grad_norm": 0.20669469237327576,
      "learning_rate": 5.1574493501439154e-05,
      "loss": 0.386,
      "num_tokens": 501724060.0,
      "step": 792
    },
    {
      "epoch": 0.09376847581884828,
      "grad_norm": 0.2072022706270218,
      "learning_rate": 5.155193949374976e-05,
      "loss": 0.3802,
      "num_tokens": 502336426.0,
      "step": 793
    },
    {
      "epoch": 0.09388672105947736,
      "grad_norm": 0.227503702044487,
      "learning_rate": 5.152936093639275e-05,
      "loss": 0.4131,
      "num_tokens": 502970977.0,
      "step": 794
    },
    {
      "epoch": 0.09400496630010642,
      "grad_norm": 0.18835189938545227,
      "learning_rate": 5.1506757859246195e-05,
      "loss": 0.3425,
      "num_tokens": 503604914.0,
      "step": 795
    },
    {
      "epoch": 0.09412321154073548,
      "grad_norm": 0.21025459468364716,
      "learning_rate": 5.148413029222061e-05,
      "loss": 0.3768,
      "num_tokens": 504241729.0,
      "step": 796
    },
    {
      "epoch": 0.09424145678136454,
      "grad_norm": 0.19752518832683563,
      "learning_rate": 5.146147826525892e-05,
      "loss": 0.3779,
      "num_tokens": 504874311.0,
      "step": 797
    },
    {
      "epoch": 0.09435970202199362,
      "grad_norm": 0.2358514666557312,
      "learning_rate": 5.14388018083364e-05,
      "loss": 0.3992,
      "num_tokens": 505507864.0,
      "step": 798
    },
    {
      "epoch": 0.09447794726262268,
      "grad_norm": 0.21056601405143738,
      "learning_rate": 5.141610095146066e-05,
      "loss": 0.4028,
      "num_tokens": 506141369.0,
      "step": 799
    },
    {
      "epoch": 0.09459619250325174,
      "grad_norm": 0.19571331143379211,
      "learning_rate": 5.139337572467163e-05,
      "loss": 0.3791,
      "num_tokens": 506777001.0,
      "step": 800
    },
    {
      "epoch": 0.0947144377438808,
      "grad_norm": 0.1915920525789261,
      "learning_rate": 5.137062615804145e-05,
      "loss": 0.3397,
      "num_tokens": 507406711.0,
      "step": 801
    },
    {
      "epoch": 0.09483268298450988,
      "grad_norm": 0.195199653506279,
      "learning_rate": 5.134785228167447e-05,
      "loss": 0.3947,
      "num_tokens": 508042267.0,
      "step": 802
    },
    {
      "epoch": 0.09495092822513894,
      "grad_norm": 0.20021317899227142,
      "learning_rate": 5.132505412570724e-05,
      "loss": 0.3644,
      "num_tokens": 508675609.0,
      "step": 803
    },
    {
      "epoch": 0.095069173465768,
      "grad_norm": 0.20839209854602814,
      "learning_rate": 5.1302231720308405e-05,
      "loss": 0.3671,
      "num_tokens": 509312065.0,
      "step": 804
    },
    {
      "epoch": 0.09518741870639706,
      "grad_norm": 0.20513001084327698,
      "learning_rate": 5.127938509567872e-05,
      "loss": 0.3783,
      "num_tokens": 509946717.0,
      "step": 805
    },
    {
      "epoch": 0.09530566394702614,
      "grad_norm": 0.19213712215423584,
      "learning_rate": 5.1256514282051e-05,
      "loss": 0.3806,
      "num_tokens": 510583518.0,
      "step": 806
    },
    {
      "epoch": 0.0954239091876552,
      "grad_norm": 0.2166701853275299,
      "learning_rate": 5.123361930969002e-05,
      "loss": 0.4097,
      "num_tokens": 511221107.0,
      "step": 807
    },
    {
      "epoch": 0.09554215442828426,
      "grad_norm": 0.19391781091690063,
      "learning_rate": 5.121070020889258e-05,
      "loss": 0.396,
      "num_tokens": 511860697.0,
      "step": 808
    },
    {
      "epoch": 0.09566039966891332,
      "grad_norm": 0.20301750302314758,
      "learning_rate": 5.1187757009987366e-05,
      "loss": 0.384,
      "num_tokens": 512498128.0,
      "step": 809
    },
    {
      "epoch": 0.0957786449095424,
      "grad_norm": 0.21539835631847382,
      "learning_rate": 5.1164789743335005e-05,
      "loss": 0.4199,
      "num_tokens": 513128610.0,
      "step": 810
    },
    {
      "epoch": 0.09589689015017146,
      "grad_norm": 0.18517307937145233,
      "learning_rate": 5.1141798439327895e-05,
      "loss": 0.3475,
      "num_tokens": 513765105.0,
      "step": 811
    },
    {
      "epoch": 0.09601513539080052,
      "grad_norm": 0.21439459919929504,
      "learning_rate": 5.111878312839033e-05,
      "loss": 0.4081,
      "num_tokens": 514394034.0,
      "step": 812
    },
    {
      "epoch": 0.09613338063142958,
      "grad_norm": 0.2012118101119995,
      "learning_rate": 5.10957438409783e-05,
      "loss": 0.3858,
      "num_tokens": 515033384.0,
      "step": 813
    },
    {
      "epoch": 0.09625162587205865,
      "grad_norm": 0.19617058336734772,
      "learning_rate": 5.107268060757957e-05,
      "loss": 0.3692,
      "num_tokens": 515667789.0,
      "step": 814
    },
    {
      "epoch": 0.09636987111268772,
      "grad_norm": 0.19214196503162384,
      "learning_rate": 5.1049593458713567e-05,
      "loss": 0.373,
      "num_tokens": 516301878.0,
      "step": 815
    },
    {
      "epoch": 0.09648811635331678,
      "grad_norm": 0.19815993309020996,
      "learning_rate": 5.1026482424931365e-05,
      "loss": 0.3633,
      "num_tokens": 516936923.0,
      "step": 816
    },
    {
      "epoch": 0.09660636159394584,
      "grad_norm": 0.2114209234714508,
      "learning_rate": 5.100334753681567e-05,
      "loss": 0.4106,
      "num_tokens": 517574893.0,
      "step": 817
    },
    {
      "epoch": 0.09672460683457491,
      "grad_norm": 0.18188177049160004,
      "learning_rate": 5.098018882498073e-05,
      "loss": 0.3795,
      "num_tokens": 518213624.0,
      "step": 818
    },
    {
      "epoch": 0.09684285207520398,
      "grad_norm": 0.20022772252559662,
      "learning_rate": 5.0957006320072334e-05,
      "loss": 0.3639,
      "num_tokens": 518852531.0,
      "step": 819
    },
    {
      "epoch": 0.09696109731583304,
      "grad_norm": 0.21430368721485138,
      "learning_rate": 5.093380005276774e-05,
      "loss": 0.4,
      "num_tokens": 519492204.0,
      "step": 820
    },
    {
      "epoch": 0.0970793425564621,
      "grad_norm": 0.19938379526138306,
      "learning_rate": 5.091057005377565e-05,
      "loss": 0.407,
      "num_tokens": 520130296.0,
      "step": 821
    },
    {
      "epoch": 0.09719758779709117,
      "grad_norm": 0.1988488733768463,
      "learning_rate": 5.088731635383619e-05,
      "loss": 0.3725,
      "num_tokens": 520765986.0,
      "step": 822
    },
    {
      "epoch": 0.09731583303772023,
      "grad_norm": 0.21430915594100952,
      "learning_rate": 5.0864038983720846e-05,
      "loss": 0.433,
      "num_tokens": 521400644.0,
      "step": 823
    },
    {
      "epoch": 0.0974340782783493,
      "grad_norm": 0.21381375193595886,
      "learning_rate": 5.0840737974232404e-05,
      "loss": 0.3746,
      "num_tokens": 522039463.0,
      "step": 824
    },
    {
      "epoch": 0.09755232351897836,
      "grad_norm": 0.2035173922777176,
      "learning_rate": 5.081741335620495e-05,
      "loss": 0.3772,
      "num_tokens": 522671508.0,
      "step": 825
    },
    {
      "epoch": 0.09767056875960743,
      "grad_norm": 0.1806216984987259,
      "learning_rate": 5.079406516050381e-05,
      "loss": 0.3692,
      "num_tokens": 523311225.0,
      "step": 826
    },
    {
      "epoch": 0.0977888140002365,
      "grad_norm": 0.1778479665517807,
      "learning_rate": 5.077069341802552e-05,
      "loss": 0.3509,
      "num_tokens": 523949133.0,
      "step": 827
    },
    {
      "epoch": 0.09790705924086555,
      "grad_norm": 0.18165086209774017,
      "learning_rate": 5.0747298159697744e-05,
      "loss": 0.3893,
      "num_tokens": 524577955.0,
      "step": 828
    },
    {
      "epoch": 0.09802530448149462,
      "grad_norm": 0.186898335814476,
      "learning_rate": 5.072387941647929e-05,
      "loss": 0.3885,
      "num_tokens": 525211514.0,
      "step": 829
    },
    {
      "epoch": 0.09814354972212369,
      "grad_norm": 0.1848706752061844,
      "learning_rate": 5.070043721936005e-05,
      "loss": 0.4114,
      "num_tokens": 525848086.0,
      "step": 830
    },
    {
      "epoch": 0.09826179496275275,
      "grad_norm": 0.1990211457014084,
      "learning_rate": 5.067697159936092e-05,
      "loss": 0.3728,
      "num_tokens": 526481722.0,
      "step": 831
    },
    {
      "epoch": 0.09838004020338181,
      "grad_norm": 0.20220965147018433,
      "learning_rate": 5.065348258753382e-05,
      "loss": 0.3912,
      "num_tokens": 527117956.0,
      "step": 832
    },
    {
      "epoch": 0.09849828544401087,
      "grad_norm": 0.20298202335834503,
      "learning_rate": 5.0629970214961606e-05,
      "loss": 0.4137,
      "num_tokens": 527750234.0,
      "step": 833
    },
    {
      "epoch": 0.09861653068463995,
      "grad_norm": 0.20618374645709991,
      "learning_rate": 5.060643451275808e-05,
      "loss": 0.3854,
      "num_tokens": 528386561.0,
      "step": 834
    },
    {
      "epoch": 0.09873477592526901,
      "grad_norm": 0.20952282845973969,
      "learning_rate": 5.0582875512067864e-05,
      "loss": 0.4212,
      "num_tokens": 529021750.0,
      "step": 835
    },
    {
      "epoch": 0.09885302116589807,
      "grad_norm": 0.19578179717063904,
      "learning_rate": 5.0559293244066455e-05,
      "loss": 0.3625,
      "num_tokens": 529661276.0,
      "step": 836
    },
    {
      "epoch": 0.09897126640652713,
      "grad_norm": 0.20233289897441864,
      "learning_rate": 5.0535687739960125e-05,
      "loss": 0.4032,
      "num_tokens": 530297609.0,
      "step": 837
    },
    {
      "epoch": 0.09908951164715621,
      "grad_norm": 0.18912136554718018,
      "learning_rate": 5.051205903098589e-05,
      "loss": 0.3768,
      "num_tokens": 530933315.0,
      "step": 838
    },
    {
      "epoch": 0.09920775688778527,
      "grad_norm": 0.19747322797775269,
      "learning_rate": 5.048840714841147e-05,
      "loss": 0.3904,
      "num_tokens": 531564788.0,
      "step": 839
    },
    {
      "epoch": 0.09932600212841433,
      "grad_norm": 0.2101273536682129,
      "learning_rate": 5.046473212353525e-05,
      "loss": 0.3947,
      "num_tokens": 532193430.0,
      "step": 840
    },
    {
      "epoch": 0.09944424736904339,
      "grad_norm": 0.21572358906269073,
      "learning_rate": 5.044103398768626e-05,
      "loss": 0.4025,
      "num_tokens": 532828601.0,
      "step": 841
    },
    {
      "epoch": 0.09956249260967245,
      "grad_norm": 0.2013438493013382,
      "learning_rate": 5.04173127722241e-05,
      "loss": 0.359,
      "num_tokens": 533466099.0,
      "step": 842
    },
    {
      "epoch": 0.09968073785030153,
      "grad_norm": 0.21333152055740356,
      "learning_rate": 5.03935685085389e-05,
      "loss": 0.3972,
      "num_tokens": 534102611.0,
      "step": 843
    },
    {
      "epoch": 0.09979898309093059,
      "grad_norm": 0.1875704973936081,
      "learning_rate": 5.036980122805129e-05,
      "loss": 0.3831,
      "num_tokens": 534731488.0,
      "step": 844
    },
    {
      "epoch": 0.09991722833155965,
      "grad_norm": 0.2032729685306549,
      "learning_rate": 5.0346010962212375e-05,
      "loss": 0.3874,
      "num_tokens": 535370116.0,
      "step": 845
    },
    {
      "epoch": 0.10003547357218871,
      "grad_norm": 0.18617123365402222,
      "learning_rate": 5.032219774250367e-05,
      "loss": 0.3922,
      "num_tokens": 536006561.0,
      "step": 846
    },
    {
      "epoch": 0.10015371881281779,
      "grad_norm": 0.2104693055152893,
      "learning_rate": 5.029836160043707e-05,
      "loss": 0.406,
      "num_tokens": 536639992.0,
      "step": 847
    },
    {
      "epoch": 0.10027196405344685,
      "grad_norm": 0.18108710646629333,
      "learning_rate": 5.027450256755477e-05,
      "loss": 0.3731,
      "num_tokens": 537253609.0,
      "step": 848
    },
    {
      "epoch": 0.10039020929407591,
      "grad_norm": 0.1897713840007782,
      "learning_rate": 5.025062067542931e-05,
      "loss": 0.3827,
      "num_tokens": 537890122.0,
      "step": 849
    },
    {
      "epoch": 0.10050845453470497,
      "grad_norm": 0.20141065120697021,
      "learning_rate": 5.022671595566343e-05,
      "loss": 0.4289,
      "num_tokens": 538526166.0,
      "step": 850
    },
    {
      "epoch": 0.10062669977533405,
      "grad_norm": 0.19830313324928284,
      "learning_rate": 5.0202788439890116e-05,
      "loss": 0.4037,
      "num_tokens": 539164864.0,
      "step": 851
    },
    {
      "epoch": 0.10074494501596311,
      "grad_norm": 0.18737861514091492,
      "learning_rate": 5.017883815977249e-05,
      "loss": 0.3712,
      "num_tokens": 539802408.0,
      "step": 852
    },
    {
      "epoch": 0.10086319025659217,
      "grad_norm": 0.18904809653759003,
      "learning_rate": 5.0154865147003816e-05,
      "loss": 0.3951,
      "num_tokens": 540432320.0,
      "step": 853
    },
    {
      "epoch": 0.10098143549722123,
      "grad_norm": 0.20891410112380981,
      "learning_rate": 5.013086943330743e-05,
      "loss": 0.3993,
      "num_tokens": 541067861.0,
      "step": 854
    },
    {
      "epoch": 0.1010996807378503,
      "grad_norm": 0.21142107248306274,
      "learning_rate": 5.010685105043673e-05,
      "loss": 0.4105,
      "num_tokens": 541704638.0,
      "step": 855
    },
    {
      "epoch": 0.10121792597847937,
      "grad_norm": 0.18683163821697235,
      "learning_rate": 5.008281003017507e-05,
      "loss": 0.3868,
      "num_tokens": 542339733.0,
      "step": 856
    },
    {
      "epoch": 0.10133617121910843,
      "grad_norm": 0.19870206713676453,
      "learning_rate": 5.0058746404335794e-05,
      "loss": 0.3736,
      "num_tokens": 542975991.0,
      "step": 857
    },
    {
      "epoch": 0.10145441645973749,
      "grad_norm": 0.17086005210876465,
      "learning_rate": 5.003466020476215e-05,
      "loss": 0.3705,
      "num_tokens": 543610157.0,
      "step": 858
    },
    {
      "epoch": 0.10157266170036656,
      "grad_norm": 0.20395129919052124,
      "learning_rate": 5.001055146332727e-05,
      "loss": 0.3879,
      "num_tokens": 544242202.0,
      "step": 859
    },
    {
      "epoch": 0.10169090694099563,
      "grad_norm": 0.19586624205112457,
      "learning_rate": 4.998642021193408e-05,
      "loss": 0.4023,
      "num_tokens": 544876956.0,
      "step": 860
    },
    {
      "epoch": 0.10180915218162469,
      "grad_norm": 0.20469500124454498,
      "learning_rate": 4.996226648251534e-05,
      "loss": 0.3793,
      "num_tokens": 545512522.0,
      "step": 861
    },
    {
      "epoch": 0.10192739742225375,
      "grad_norm": 0.21595309674739838,
      "learning_rate": 4.9938090307033514e-05,
      "loss": 0.4136,
      "num_tokens": 546149492.0,
      "step": 862
    },
    {
      "epoch": 0.10204564266288282,
      "grad_norm": 0.18492047488689423,
      "learning_rate": 4.9913891717480795e-05,
      "loss": 0.3663,
      "num_tokens": 546785991.0,
      "step": 863
    },
    {
      "epoch": 0.10216388790351189,
      "grad_norm": 0.18823403120040894,
      "learning_rate": 4.9889670745879035e-05,
      "loss": 0.4047,
      "num_tokens": 547421213.0,
      "step": 864
    },
    {
      "epoch": 0.10228213314414095,
      "grad_norm": 0.2192760556936264,
      "learning_rate": 4.986542742427968e-05,
      "loss": 0.4077,
      "num_tokens": 548057540.0,
      "step": 865
    },
    {
      "epoch": 0.10240037838477001,
      "grad_norm": 0.21480301022529602,
      "learning_rate": 4.984116178476379e-05,
      "loss": 0.4214,
      "num_tokens": 548694755.0,
      "step": 866
    },
    {
      "epoch": 0.10251862362539908,
      "grad_norm": 0.22339558601379395,
      "learning_rate": 4.9816873859441906e-05,
      "loss": 0.4529,
      "num_tokens": 549330967.0,
      "step": 867
    },
    {
      "epoch": 0.10263686886602814,
      "grad_norm": 0.1828283965587616,
      "learning_rate": 4.979256368045413e-05,
      "loss": 0.3812,
      "num_tokens": 549965342.0,
      "step": 868
    },
    {
      "epoch": 0.1027551141066572,
      "grad_norm": 0.1752055287361145,
      "learning_rate": 4.9768231279969936e-05,
      "loss": 0.3692,
      "num_tokens": 550595153.0,
      "step": 869
    },
    {
      "epoch": 0.10287335934728627,
      "grad_norm": 0.20586678385734558,
      "learning_rate": 4.9743876690188254e-05,
      "loss": 0.3934,
      "num_tokens": 551228741.0,
      "step": 870
    },
    {
      "epoch": 0.10299160458791534,
      "grad_norm": 0.18745943903923035,
      "learning_rate": 4.9719499943337376e-05,
      "loss": 0.4051,
      "num_tokens": 551838042.0,
      "step": 871
    },
    {
      "epoch": 0.1031098498285444,
      "grad_norm": 0.19367623329162598,
      "learning_rate": 4.969510107167488e-05,
      "loss": 0.3829,
      "num_tokens": 552467738.0,
      "step": 872
    },
    {
      "epoch": 0.10322809506917346,
      "grad_norm": 0.19626417756080627,
      "learning_rate": 4.9670680107487665e-05,
      "loss": 0.3905,
      "num_tokens": 553103599.0,
      "step": 873
    },
    {
      "epoch": 0.10334634030980253,
      "grad_norm": 0.1922215074300766,
      "learning_rate": 4.964623708309182e-05,
      "loss": 0.383,
      "num_tokens": 553736538.0,
      "step": 874
    },
    {
      "epoch": 0.1034645855504316,
      "grad_norm": 0.1745498776435852,
      "learning_rate": 4.962177203083267e-05,
      "loss": 0.3716,
      "num_tokens": 554369713.0,
      "step": 875
    },
    {
      "epoch": 0.10358283079106066,
      "grad_norm": 0.20037321746349335,
      "learning_rate": 4.959728498308465e-05,
      "loss": 0.3799,
      "num_tokens": 555000107.0,
      "step": 876
    },
    {
      "epoch": 0.10370107603168972,
      "grad_norm": 0.19089975953102112,
      "learning_rate": 4.957277597225133e-05,
      "loss": 0.368,
      "num_tokens": 555632229.0,
      "step": 877
    },
    {
      "epoch": 0.10381932127231878,
      "grad_norm": 0.18980231881141663,
      "learning_rate": 4.954824503076534e-05,
      "loss": 0.3878,
      "num_tokens": 556267726.0,
      "step": 878
    },
    {
      "epoch": 0.10393756651294786,
      "grad_norm": 0.19482596218585968,
      "learning_rate": 4.952369219108831e-05,
      "loss": 0.3518,
      "num_tokens": 556878328.0,
      "step": 879
    },
    {
      "epoch": 0.10405581175357692,
      "grad_norm": 0.1749802827835083,
      "learning_rate": 4.9499117485710866e-05,
      "loss": 0.3745,
      "num_tokens": 557513764.0,
      "step": 880
    },
    {
      "epoch": 0.10417405699420598,
      "grad_norm": 0.19543933868408203,
      "learning_rate": 4.947452094715258e-05,
      "loss": 0.3719,
      "num_tokens": 558143004.0,
      "step": 881
    },
    {
      "epoch": 0.10429230223483504,
      "grad_norm": 0.18101941049098969,
      "learning_rate": 4.944990260796188e-05,
      "loss": 0.3707,
      "num_tokens": 558778105.0,
      "step": 882
    },
    {
      "epoch": 0.10441054747546412,
      "grad_norm": 0.2059684544801712,
      "learning_rate": 4.9425262500716065e-05,
      "loss": 0.3884,
      "num_tokens": 559416233.0,
      "step": 883
    },
    {
      "epoch": 0.10452879271609318,
      "grad_norm": 0.17464634776115417,
      "learning_rate": 4.940060065802125e-05,
      "loss": 0.3874,
      "num_tokens": 560051126.0,
      "step": 884
    },
    {
      "epoch": 0.10464703795672224,
      "grad_norm": 0.19525666534900665,
      "learning_rate": 4.9375917112512283e-05,
      "loss": 0.3986,
      "num_tokens": 560687449.0,
      "step": 885
    },
    {
      "epoch": 0.1047652831973513,
      "grad_norm": 0.1954362988471985,
      "learning_rate": 4.9351211896852775e-05,
      "loss": 0.3728,
      "num_tokens": 561317929.0,
      "step": 886
    },
    {
      "epoch": 0.10488352843798038,
      "grad_norm": 0.19660866260528564,
      "learning_rate": 4.9326485043734956e-05,
      "loss": 0.3984,
      "num_tokens": 561956472.0,
      "step": 887
    },
    {
      "epoch": 0.10500177367860944,
      "grad_norm": 0.17625193297863007,
      "learning_rate": 4.9301736585879735e-05,
      "loss": 0.3933,
      "num_tokens": 562593057.0,
      "step": 888
    },
    {
      "epoch": 0.1051200189192385,
      "grad_norm": 0.20794154703617096,
      "learning_rate": 4.92769665560366e-05,
      "loss": 0.4091,
      "num_tokens": 563227832.0,
      "step": 889
    },
    {
      "epoch": 0.10523826415986756,
      "grad_norm": 0.166320338845253,
      "learning_rate": 4.9252174986983564e-05,
      "loss": 0.3735,
      "num_tokens": 563864035.0,
      "step": 890
    },
    {
      "epoch": 0.10535650940049664,
      "grad_norm": 0.20669224858283997,
      "learning_rate": 4.922736191152716e-05,
      "loss": 0.4333,
      "num_tokens": 564478244.0,
      "step": 891
    },
    {
      "epoch": 0.1054747546411257,
      "grad_norm": 0.1873048096895218,
      "learning_rate": 4.920252736250239e-05,
      "loss": 0.3448,
      "num_tokens": 565113319.0,
      "step": 892
    },
    {
      "epoch": 0.10559299988175476,
      "grad_norm": 0.20048575103282928,
      "learning_rate": 4.917767137277266e-05,
      "loss": 0.3979,
      "num_tokens": 565746893.0,
      "step": 893
    },
    {
      "epoch": 0.10571124512238382,
      "grad_norm": 0.19163161516189575,
      "learning_rate": 4.915279397522973e-05,
      "loss": 0.3874,
      "num_tokens": 566356947.0,
      "step": 894
    },
    {
      "epoch": 0.10582949036301288,
      "grad_norm": 0.211003378033638,
      "learning_rate": 4.912789520279373e-05,
      "loss": 0.4321,
      "num_tokens": 566994122.0,
      "step": 895
    },
    {
      "epoch": 0.10594773560364196,
      "grad_norm": 0.23276257514953613,
      "learning_rate": 4.9102975088413034e-05,
      "loss": 0.4238,
      "num_tokens": 567631151.0,
      "step": 896
    },
    {
      "epoch": 0.10606598084427102,
      "grad_norm": 0.2002890557050705,
      "learning_rate": 4.9078033665064294e-05,
      "loss": 0.3892,
      "num_tokens": 568268077.0,
      "step": 897
    },
    {
      "epoch": 0.10618422608490008,
      "grad_norm": 0.19507084786891937,
      "learning_rate": 4.905307096575234e-05,
      "loss": 0.3987,
      "num_tokens": 568906566.0,
      "step": 898
    },
    {
      "epoch": 0.10630247132552914,
      "grad_norm": 0.1791028529405594,
      "learning_rate": 4.902808702351013e-05,
      "loss": 0.3891,
      "num_tokens": 569542062.0,
      "step": 899
    },
    {
      "epoch": 0.10642071656615822,
      "grad_norm": 0.18552984297275543,
      "learning_rate": 4.900308187139879e-05,
      "loss": 0.386,
      "num_tokens": 570177721.0,
      "step": 900
    },
    {
      "epoch": 0.10653896180678728,
      "grad_norm": 0.21034976840019226,
      "learning_rate": 4.897805554250748e-05,
      "loss": 0.4465,
      "num_tokens": 570816714.0,
      "step": 901
    },
    {
      "epoch": 0.10665720704741634,
      "grad_norm": 0.178841233253479,
      "learning_rate": 4.8953008069953394e-05,
      "loss": 0.3741,
      "num_tokens": 571449734.0,
      "step": 902
    },
    {
      "epoch": 0.1067754522880454,
      "grad_norm": 0.2186090648174286,
      "learning_rate": 4.8927939486881676e-05,
      "loss": 0.4057,
      "num_tokens": 572088174.0,
      "step": 903
    },
    {
      "epoch": 0.10689369752867448,
      "grad_norm": 0.16908499598503113,
      "learning_rate": 4.890284982646545e-05,
      "loss": 0.3723,
      "num_tokens": 572726629.0,
      "step": 904
    },
    {
      "epoch": 0.10701194276930354,
      "grad_norm": 0.18440201878547668,
      "learning_rate": 4.887773912190571e-05,
      "loss": 0.3705,
      "num_tokens": 573356851.0,
      "step": 905
    },
    {
      "epoch": 0.1071301880099326,
      "grad_norm": 0.18719598650932312,
      "learning_rate": 4.885260740643127e-05,
      "loss": 0.3722,
      "num_tokens": 573993246.0,
      "step": 906
    },
    {
      "epoch": 0.10724843325056166,
      "grad_norm": 0.2074010819196701,
      "learning_rate": 4.8827454713298805e-05,
      "loss": 0.4022,
      "num_tokens": 574632107.0,
      "step": 907
    },
    {
      "epoch": 0.10736667849119073,
      "grad_norm": 0.17451098561286926,
      "learning_rate": 4.8802281075792705e-05,
      "loss": 0.3985,
      "num_tokens": 575267606.0,
      "step": 908
    },
    {
      "epoch": 0.1074849237318198,
      "grad_norm": 0.17612576484680176,
      "learning_rate": 4.877708652722509e-05,
      "loss": 0.3827,
      "num_tokens": 575898213.0,
      "step": 909
    },
    {
      "epoch": 0.10760316897244886,
      "grad_norm": 0.23687198758125305,
      "learning_rate": 4.875187110093576e-05,
      "loss": 0.4355,
      "num_tokens": 576535415.0,
      "step": 910
    },
    {
      "epoch": 0.10772141421307792,
      "grad_norm": 0.18335507810115814,
      "learning_rate": 4.8726634830292115e-05,
      "loss": 0.394,
      "num_tokens": 577170916.0,
      "step": 911
    },
    {
      "epoch": 0.10783965945370699,
      "grad_norm": 0.2192007303237915,
      "learning_rate": 4.8701377748689174e-05,
      "loss": 0.3917,
      "num_tokens": 577801174.0,
      "step": 912
    },
    {
      "epoch": 0.10795790469433605,
      "grad_norm": 0.18701738119125366,
      "learning_rate": 4.867609988954947e-05,
      "loss": 0.4236,
      "num_tokens": 578437720.0,
      "step": 913
    },
    {
      "epoch": 0.10807614993496512,
      "grad_norm": 0.19873884320259094,
      "learning_rate": 4.8650801286323057e-05,
      "loss": 0.3843,
      "num_tokens": 579073461.0,
      "step": 914
    },
    {
      "epoch": 0.10819439517559418,
      "grad_norm": 0.17789912223815918,
      "learning_rate": 4.8625481972487395e-05,
      "loss": 0.399,
      "num_tokens": 579712385.0,
      "step": 915
    },
    {
      "epoch": 0.10831264041622325,
      "grad_norm": 0.20318271219730377,
      "learning_rate": 4.86001419815474e-05,
      "loss": 0.3812,
      "num_tokens": 580346374.0,
      "step": 916
    },
    {
      "epoch": 0.10843088565685231,
      "grad_norm": 0.20375102758407593,
      "learning_rate": 4.8574781347035296e-05,
      "loss": 0.3876,
      "num_tokens": 580976923.0,
      "step": 917
    },
    {
      "epoch": 0.10854913089748137,
      "grad_norm": 0.20212046802043915,
      "learning_rate": 4.854940010251069e-05,
      "loss": 0.368,
      "num_tokens": 581611122.0,
      "step": 918
    },
    {
      "epoch": 0.10866737613811044,
      "grad_norm": 0.19325628876686096,
      "learning_rate": 4.8523998281560405e-05,
      "loss": 0.394,
      "num_tokens": 582241753.0,
      "step": 919
    },
    {
      "epoch": 0.10878562137873951,
      "grad_norm": 0.2155366986989975,
      "learning_rate": 4.849857591779851e-05,
      "loss": 0.4028,
      "num_tokens": 582880022.0,
      "step": 920
    },
    {
      "epoch": 0.10890386661936857,
      "grad_norm": 0.19220107793807983,
      "learning_rate": 4.847313304486629e-05,
      "loss": 0.3633,
      "num_tokens": 583516737.0,
      "step": 921
    },
    {
      "epoch": 0.10902211185999763,
      "grad_norm": 0.2031642198562622,
      "learning_rate": 4.844766969643212e-05,
      "loss": 0.3561,
      "num_tokens": 584155270.0,
      "step": 922
    },
    {
      "epoch": 0.1091403571006267,
      "grad_norm": 0.20130768418312073,
      "learning_rate": 4.8422185906191494e-05,
      "loss": 0.3893,
      "num_tokens": 584788413.0,
      "step": 923
    },
    {
      "epoch": 0.10925860234125577,
      "grad_norm": 0.19773003458976746,
      "learning_rate": 4.839668170786696e-05,
      "loss": 0.4072,
      "num_tokens": 585424357.0,
      "step": 924
    },
    {
      "epoch": 0.10937684758188483,
      "grad_norm": 0.21327199041843414,
      "learning_rate": 4.8371157135208066e-05,
      "loss": 0.3916,
      "num_tokens": 586058909.0,
      "step": 925
    },
    {
      "epoch": 0.10949509282251389,
      "grad_norm": 0.1890989989042282,
      "learning_rate": 4.834561222199132e-05,
      "loss": 0.3901,
      "num_tokens": 586688711.0,
      "step": 926
    },
    {
      "epoch": 0.10961333806314295,
      "grad_norm": 0.1838907152414322,
      "learning_rate": 4.832004700202017e-05,
      "loss": 0.3264,
      "num_tokens": 587318129.0,
      "step": 927
    },
    {
      "epoch": 0.10973158330377203,
      "grad_norm": 0.18171823024749756,
      "learning_rate": 4.829446150912488e-05,
      "loss": 0.3649,
      "num_tokens": 587954063.0,
      "step": 928
    },
    {
      "epoch": 0.10984982854440109,
      "grad_norm": 0.1974538266658783,
      "learning_rate": 4.826885577716258e-05,
      "loss": 0.3995,
      "num_tokens": 588586978.0,
      "step": 929
    },
    {
      "epoch": 0.10996807378503015,
      "grad_norm": 0.17991968989372253,
      "learning_rate": 4.824322984001721e-05,
      "loss": 0.3809,
      "num_tokens": 589223333.0,
      "step": 930
    },
    {
      "epoch": 0.11008631902565921,
      "grad_norm": 0.18628670275211334,
      "learning_rate": 4.821758373159938e-05,
      "loss": 0.3556,
      "num_tokens": 589857753.0,
      "step": 931
    },
    {
      "epoch": 0.11020456426628829,
      "grad_norm": 0.19802094995975494,
      "learning_rate": 4.819191748584646e-05,
      "loss": 0.3719,
      "num_tokens": 590492668.0,
      "step": 932
    },
    {
      "epoch": 0.11032280950691735,
      "grad_norm": 0.18508954346179962,
      "learning_rate": 4.81662311367224e-05,
      "loss": 0.3815,
      "num_tokens": 591130240.0,
      "step": 933
    },
    {
      "epoch": 0.11044105474754641,
      "grad_norm": 0.18970444798469543,
      "learning_rate": 4.814052471821782e-05,
      "loss": 0.4276,
      "num_tokens": 591744210.0,
      "step": 934
    },
    {
      "epoch": 0.11055929998817547,
      "grad_norm": 0.1893254518508911,
      "learning_rate": 4.8114798264349845e-05,
      "loss": 0.3506,
      "num_tokens": 592379832.0,
      "step": 935
    },
    {
      "epoch": 0.11067754522880455,
      "grad_norm": 0.19586031138896942,
      "learning_rate": 4.8089051809162144e-05,
      "loss": 0.3826,
      "num_tokens": 593014264.0,
      "step": 936
    },
    {
      "epoch": 0.11079579046943361,
      "grad_norm": 0.22042131423950195,
      "learning_rate": 4.8063285386724835e-05,
      "loss": 0.4103,
      "num_tokens": 593647737.0,
      "step": 937
    },
    {
      "epoch": 0.11091403571006267,
      "grad_norm": 0.1967095583677292,
      "learning_rate": 4.803749903113447e-05,
      "loss": 0.4202,
      "num_tokens": 594284342.0,
      "step": 938
    },
    {
      "epoch": 0.11103228095069173,
      "grad_norm": 0.1965053677558899,
      "learning_rate": 4.801169277651398e-05,
      "loss": 0.3929,
      "num_tokens": 594917472.0,
      "step": 939
    },
    {
      "epoch": 0.1111505261913208,
      "grad_norm": 0.19952984154224396,
      "learning_rate": 4.798586665701262e-05,
      "loss": 0.3875,
      "num_tokens": 595555577.0,
      "step": 940
    },
    {
      "epoch": 0.11126877143194987,
      "grad_norm": 0.20156317949295044,
      "learning_rate": 4.7960020706805934e-05,
      "loss": 0.4086,
      "num_tokens": 596194919.0,
      "step": 941
    },
    {
      "epoch": 0.11138701667257893,
      "grad_norm": 0.20338894426822662,
      "learning_rate": 4.7934154960095705e-05,
      "loss": 0.3963,
      "num_tokens": 596828877.0,
      "step": 942
    },
    {
      "epoch": 0.11150526191320799,
      "grad_norm": 0.21661074459552765,
      "learning_rate": 4.7908269451109926e-05,
      "loss": 0.4371,
      "num_tokens": 597463751.0,
      "step": 943
    },
    {
      "epoch": 0.11162350715383706,
      "grad_norm": 0.1983601450920105,
      "learning_rate": 4.7882364214102725e-05,
      "loss": 0.3729,
      "num_tokens": 598101711.0,
      "step": 944
    },
    {
      "epoch": 0.11174175239446613,
      "grad_norm": 0.19059373438358307,
      "learning_rate": 4.7856439283354345e-05,
      "loss": 0.3628,
      "num_tokens": 598735765.0,
      "step": 945
    },
    {
      "epoch": 0.11185999763509519,
      "grad_norm": 0.19377626478672028,
      "learning_rate": 4.7830494693171085e-05,
      "loss": 0.4069,
      "num_tokens": 599372011.0,
      "step": 946
    },
    {
      "epoch": 0.11197824287572425,
      "grad_norm": 0.2112198919057846,
      "learning_rate": 4.780453047788528e-05,
      "loss": 0.368,
      "num_tokens": 600001896.0,
      "step": 947
    },
    {
      "epoch": 0.11209648811635331,
      "grad_norm": 0.18281449377536774,
      "learning_rate": 4.7778546671855186e-05,
      "loss": 0.3621,
      "num_tokens": 600632801.0,
      "step": 948
    },
    {
      "epoch": 0.11221473335698239,
      "grad_norm": 0.17474791407585144,
      "learning_rate": 4.7752543309465044e-05,
      "loss": 0.3537,
      "num_tokens": 601266569.0,
      "step": 949
    },
    {
      "epoch": 0.11233297859761145,
      "grad_norm": 0.20017535984516144,
      "learning_rate": 4.772652042512491e-05,
      "loss": 0.3876,
      "num_tokens": 601904732.0,
      "step": 950
    },
    {
      "epoch": 0.11245122383824051,
      "grad_norm": 0.16637170314788818,
      "learning_rate": 4.770047805327074e-05,
      "loss": 0.3626,
      "num_tokens": 602535124.0,
      "step": 951
    },
    {
      "epoch": 0.11256946907886957,
      "grad_norm": 0.19100861251354218,
      "learning_rate": 4.7674416228364225e-05,
      "loss": 0.3927,
      "num_tokens": 603172120.0,
      "step": 952
    },
    {
      "epoch": 0.11268771431949864,
      "grad_norm": 0.20126068592071533,
      "learning_rate": 4.764833498489283e-05,
      "loss": 0.4082,
      "num_tokens": 603810653.0,
      "step": 953
    },
    {
      "epoch": 0.1128059595601277,
      "grad_norm": 0.1704559475183487,
      "learning_rate": 4.76222343573697e-05,
      "loss": 0.3481,
      "num_tokens": 604445259.0,
      "step": 954
    },
    {
      "epoch": 0.11292420480075677,
      "grad_norm": 0.1779528260231018,
      "learning_rate": 4.759611438033363e-05,
      "loss": 0.3596,
      "num_tokens": 605079841.0,
      "step": 955
    },
    {
      "epoch": 0.11304245004138583,
      "grad_norm": 0.20043320953845978,
      "learning_rate": 4.756997508834903e-05,
      "loss": 0.4095,
      "num_tokens": 605716847.0,
      "step": 956
    },
    {
      "epoch": 0.1131606952820149,
      "grad_norm": 0.19246995449066162,
      "learning_rate": 4.7543816516005865e-05,
      "loss": 0.3956,
      "num_tokens": 606349039.0,
      "step": 957
    },
    {
      "epoch": 0.11327894052264396,
      "grad_norm": 0.17408093810081482,
      "learning_rate": 4.7517638697919605e-05,
      "loss": 0.3872,
      "num_tokens": 606979557.0,
      "step": 958
    },
    {
      "epoch": 0.11339718576327303,
      "grad_norm": 0.18687652051448822,
      "learning_rate": 4.7491441668731214e-05,
      "loss": 0.3314,
      "num_tokens": 607606199.0,
      "step": 959
    },
    {
      "epoch": 0.11351543100390209,
      "grad_norm": 0.19232109189033508,
      "learning_rate": 4.746522546310704e-05,
      "loss": 0.3693,
      "num_tokens": 608215251.0,
      "step": 960
    },
    {
      "epoch": 0.11363367624453116,
      "grad_norm": 0.19829553365707397,
      "learning_rate": 4.7438990115738836e-05,
      "loss": 0.4049,
      "num_tokens": 608851164.0,
      "step": 961
    },
    {
      "epoch": 0.11375192148516022,
      "grad_norm": 0.19942665100097656,
      "learning_rate": 4.741273566134368e-05,
      "loss": 0.3855,
      "num_tokens": 609484264.0,
      "step": 962
    },
    {
      "epoch": 0.11387016672578928,
      "grad_norm": 0.17975790798664093,
      "learning_rate": 4.738646213466392e-05,
      "loss": 0.3708,
      "num_tokens": 610095969.0,
      "step": 963
    },
    {
      "epoch": 0.11398841196641835,
      "grad_norm": 0.18488220870494843,
      "learning_rate": 4.7360169570467165e-05,
      "loss": 0.3615,
      "num_tokens": 610726014.0,
      "step": 964
    },
    {
      "epoch": 0.11410665720704742,
      "grad_norm": 0.16256028413772583,
      "learning_rate": 4.733385800354619e-05,
      "loss": 0.3282,
      "num_tokens": 611354133.0,
      "step": 965
    },
    {
      "epoch": 0.11422490244767648,
      "grad_norm": 0.18399277329444885,
      "learning_rate": 4.7307527468718944e-05,
      "loss": 0.3705,
      "num_tokens": 611983546.0,
      "step": 966
    },
    {
      "epoch": 0.11434314768830554,
      "grad_norm": 0.1812833547592163,
      "learning_rate": 4.728117800082844e-05,
      "loss": 0.3815,
      "num_tokens": 612614615.0,
      "step": 967
    },
    {
      "epoch": 0.1144613929289346,
      "grad_norm": 0.1888638585805893,
      "learning_rate": 4.72548096347428e-05,
      "loss": 0.401,
      "num_tokens": 613251367.0,
      "step": 968
    },
    {
      "epoch": 0.11457963816956368,
      "grad_norm": 0.18793244659900665,
      "learning_rate": 4.7228422405355095e-05,
      "loss": 0.4288,
      "num_tokens": 613882406.0,
      "step": 969
    },
    {
      "epoch": 0.11469788341019274,
      "grad_norm": 0.1834799200296402,
      "learning_rate": 4.7202016347583395e-05,
      "loss": 0.3824,
      "num_tokens": 614516798.0,
      "step": 970
    },
    {
      "epoch": 0.1148161286508218,
      "grad_norm": 0.18810245394706726,
      "learning_rate": 4.717559149637066e-05,
      "loss": 0.4009,
      "num_tokens": 615155542.0,
      "step": 971
    },
    {
      "epoch": 0.11493437389145086,
      "grad_norm": 0.1651647984981537,
      "learning_rate": 4.7149147886684756e-05,
      "loss": 0.3113,
      "num_tokens": 615790325.0,
      "step": 972
    },
    {
      "epoch": 0.11505261913207994,
      "grad_norm": 0.17223705351352692,
      "learning_rate": 4.712268555351834e-05,
      "loss": 0.3466,
      "num_tokens": 616420924.0,
      "step": 973
    },
    {
      "epoch": 0.115170864372709,
      "grad_norm": 0.19105683267116547,
      "learning_rate": 4.709620453188883e-05,
      "loss": 0.3543,
      "num_tokens": 617058249.0,
      "step": 974
    },
    {
      "epoch": 0.11528910961333806,
      "grad_norm": 0.17421837151050568,
      "learning_rate": 4.706970485683845e-05,
      "loss": 0.3644,
      "num_tokens": 617692601.0,
      "step": 975
    },
    {
      "epoch": 0.11540735485396712,
      "grad_norm": 0.18170055747032166,
      "learning_rate": 4.7043186563434016e-05,
      "loss": 0.3664,
      "num_tokens": 618329266.0,
      "step": 976
    },
    {
      "epoch": 0.1155256000945962,
      "grad_norm": 0.19499434530735016,
      "learning_rate": 4.7016649686767036e-05,
      "loss": 0.3672,
      "num_tokens": 618961318.0,
      "step": 977
    },
    {
      "epoch": 0.11564384533522526,
      "grad_norm": 0.2040436565876007,
      "learning_rate": 4.69900942619536e-05,
      "loss": 0.4315,
      "num_tokens": 619595493.0,
      "step": 978
    },
    {
      "epoch": 0.11576209057585432,
      "grad_norm": 0.20933543145656586,
      "learning_rate": 4.696352032413432e-05,
      "loss": 0.3817,
      "num_tokens": 620231952.0,
      "step": 979
    },
    {
      "epoch": 0.11588033581648338,
      "grad_norm": 0.1690838783979416,
      "learning_rate": 4.693692790847436e-05,
      "loss": 0.3621,
      "num_tokens": 620865927.0,
      "step": 980
    },
    {
      "epoch": 0.11599858105711246,
      "grad_norm": 0.21045120060443878,
      "learning_rate": 4.691031705016327e-05,
      "loss": 0.4237,
      "num_tokens": 621502818.0,
      "step": 981
    },
    {
      "epoch": 0.11611682629774152,
      "grad_norm": 0.19847846031188965,
      "learning_rate": 4.688368778441504e-05,
      "loss": 0.3925,
      "num_tokens": 622139430.0,
      "step": 982
    },
    {
      "epoch": 0.11623507153837058,
      "grad_norm": 0.1916196495294571,
      "learning_rate": 4.685704014646803e-05,
      "loss": 0.3488,
      "num_tokens": 622775467.0,
      "step": 983
    },
    {
      "epoch": 0.11635331677899964,
      "grad_norm": 0.18403828144073486,
      "learning_rate": 4.683037417158486e-05,
      "loss": 0.3859,
      "num_tokens": 623412555.0,
      "step": 984
    },
    {
      "epoch": 0.11647156201962872,
      "grad_norm": 0.20068930089473724,
      "learning_rate": 4.6803689895052496e-05,
      "loss": 0.4022,
      "num_tokens": 624051117.0,
      "step": 985
    },
    {
      "epoch": 0.11658980726025778,
      "grad_norm": 0.19401557743549347,
      "learning_rate": 4.677698735218205e-05,
      "loss": 0.363,
      "num_tokens": 624689341.0,
      "step": 986
    },
    {
      "epoch": 0.11670805250088684,
      "grad_norm": 0.21485280990600586,
      "learning_rate": 4.6750266578308843e-05,
      "loss": 0.368,
      "num_tokens": 625321056.0,
      "step": 987
    },
    {
      "epoch": 0.1168262977415159,
      "grad_norm": 0.1814233511686325,
      "learning_rate": 4.6723527608792315e-05,
      "loss": 0.3721,
      "num_tokens": 625951744.0,
      "step": 988
    },
    {
      "epoch": 0.11694454298214498,
      "grad_norm": 0.19866612553596497,
      "learning_rate": 4.669677047901596e-05,
      "loss": 0.3925,
      "num_tokens": 626587647.0,
      "step": 989
    },
    {
      "epoch": 0.11706278822277404,
      "grad_norm": 0.19822439551353455,
      "learning_rate": 4.666999522438734e-05,
      "loss": 0.3769,
      "num_tokens": 627226191.0,
      "step": 990
    },
    {
      "epoch": 0.1171810334634031,
      "grad_norm": 0.19480082392692566,
      "learning_rate": 4.664320188033798e-05,
      "loss": 0.4012,
      "num_tokens": 627854887.0,
      "step": 991
    },
    {
      "epoch": 0.11729927870403216,
      "grad_norm": 0.17783285677433014,
      "learning_rate": 4.6616390482323344e-05,
      "loss": 0.3813,
      "num_tokens": 628470814.0,
      "step": 992
    },
    {
      "epoch": 0.11741752394466123,
      "grad_norm": 0.20395652949810028,
      "learning_rate": 4.65895610658228e-05,
      "loss": 0.3888,
      "num_tokens": 629095432.0,
      "step": 993
    },
    {
      "epoch": 0.1175357691852903,
      "grad_norm": 0.18363986909389496,
      "learning_rate": 4.6562713666339545e-05,
      "loss": 0.4066,
      "num_tokens": 629735034.0,
      "step": 994
    },
    {
      "epoch": 0.11765401442591936,
      "grad_norm": 0.18329794704914093,
      "learning_rate": 4.653584831940058e-05,
      "loss": 0.3867,
      "num_tokens": 630370989.0,
      "step": 995
    },
    {
      "epoch": 0.11777225966654842,
      "grad_norm": 0.17211481928825378,
      "learning_rate": 4.650896506055665e-05,
      "loss": 0.3991,
      "num_tokens": 631007330.0,
      "step": 996
    },
    {
      "epoch": 0.11789050490717749,
      "grad_norm": 0.19009293615818024,
      "learning_rate": 4.648206392538221e-05,
      "loss": 0.387,
      "num_tokens": 631644616.0,
      "step": 997
    },
    {
      "epoch": 0.11800875014780655,
      "grad_norm": 0.18162861466407776,
      "learning_rate": 4.6455144949475364e-05,
      "loss": 0.3846,
      "num_tokens": 632278554.0,
      "step": 998
    },
    {
      "epoch": 0.11812699538843562,
      "grad_norm": 0.16204237937927246,
      "learning_rate": 4.642820816845783e-05,
      "loss": 0.3872,
      "num_tokens": 632912104.0,
      "step": 999
    },
    {
      "epoch": 0.11824524062906468,
      "grad_norm": 0.17694039642810822,
      "learning_rate": 4.640125361797488e-05,
      "loss": 0.4142,
      "num_tokens": 633548167.0,
      "step": 1000
    },
    {
      "epoch": 0.11836348586969374,
      "grad_norm": 0.18365511298179626,
      "learning_rate": 4.6374281333695306e-05,
      "loss": 0.3618,
      "num_tokens": 634181746.0,
      "step": 1001
    },
    {
      "epoch": 0.11848173111032281,
      "grad_norm": 0.1922062337398529,
      "learning_rate": 4.634729135131137e-05,
      "loss": 0.4147,
      "num_tokens": 634816694.0,
      "step": 1002
    },
    {
      "epoch": 0.11859997635095187,
      "grad_norm": 0.1679506003856659,
      "learning_rate": 4.6320283706538755e-05,
      "loss": 0.3828,
      "num_tokens": 635455070.0,
      "step": 1003
    },
    {
      "epoch": 0.11871822159158094,
      "grad_norm": 0.18912957608699799,
      "learning_rate": 4.6293258435116506e-05,
      "loss": 0.4158,
      "num_tokens": 636090133.0,
      "step": 1004
    },
    {
      "epoch": 0.11883646683221,
      "grad_norm": 0.1797725409269333,
      "learning_rate": 4.6266215572806995e-05,
      "loss": 0.4062,
      "num_tokens": 636718996.0,
      "step": 1005
    },
    {
      "epoch": 0.11895471207283907,
      "grad_norm": 0.18465858697891235,
      "learning_rate": 4.623915515539587e-05,
      "loss": 0.4067,
      "num_tokens": 637355713.0,
      "step": 1006
    },
    {
      "epoch": 0.11907295731346813,
      "grad_norm": 0.1974141150712967,
      "learning_rate": 4.6212077218692036e-05,
      "loss": 0.3873,
      "num_tokens": 637990698.0,
      "step": 1007
    },
    {
      "epoch": 0.1191912025540972,
      "grad_norm": 0.18730655312538147,
      "learning_rate": 4.618498179852754e-05,
      "loss": 0.3847,
      "num_tokens": 638622647.0,
      "step": 1008
    },
    {
      "epoch": 0.11930944779472626,
      "grad_norm": 0.2009619027376175,
      "learning_rate": 4.61578689307576e-05,
      "loss": 0.4005,
      "num_tokens": 639251231.0,
      "step": 1009
    },
    {
      "epoch": 0.11942769303535533,
      "grad_norm": 0.18502581119537354,
      "learning_rate": 4.61307386512605e-05,
      "loss": 0.386,
      "num_tokens": 639886902.0,
      "step": 1010
    },
    {
      "epoch": 0.11954593827598439,
      "grad_norm": 0.18856503069400787,
      "learning_rate": 4.610359099593757e-05,
      "loss": 0.3709,
      "num_tokens": 640524463.0,
      "step": 1011
    },
    {
      "epoch": 0.11966418351661345,
      "grad_norm": 0.1779439002275467,
      "learning_rate": 4.607642600071314e-05,
      "loss": 0.4024,
      "num_tokens": 641163335.0,
      "step": 1012
    },
    {
      "epoch": 0.11978242875724252,
      "grad_norm": 0.19025777280330658,
      "learning_rate": 4.6049243701534476e-05,
      "loss": 0.3973,
      "num_tokens": 641802244.0,
      "step": 1013
    },
    {
      "epoch": 0.11990067399787159,
      "grad_norm": 0.19890962541103363,
      "learning_rate": 4.6022044134371756e-05,
      "loss": 0.4081,
      "num_tokens": 642437930.0,
      "step": 1014
    },
    {
      "epoch": 0.12001891923850065,
      "grad_norm": 0.18591611087322235,
      "learning_rate": 4.5994827335217994e-05,
      "loss": 0.3868,
      "num_tokens": 643076089.0,
      "step": 1015
    },
    {
      "epoch": 0.12013716447912971,
      "grad_norm": 0.17976440489292145,
      "learning_rate": 4.596759334008902e-05,
      "loss": 0.4177,
      "num_tokens": 643711689.0,
      "step": 1016
    },
    {
      "epoch": 0.12025540971975877,
      "grad_norm": 0.20330271124839783,
      "learning_rate": 4.59403421850234e-05,
      "loss": 0.3857,
      "num_tokens": 644349772.0,
      "step": 1017
    },
    {
      "epoch": 0.12037365496038785,
      "grad_norm": 0.18609577417373657,
      "learning_rate": 4.5913073906082425e-05,
      "loss": 0.3813,
      "num_tokens": 644988741.0,
      "step": 1018
    },
    {
      "epoch": 0.12049190020101691,
      "grad_norm": 0.18191839754581451,
      "learning_rate": 4.588578853935005e-05,
      "loss": 0.3577,
      "num_tokens": 645614498.0,
      "step": 1019
    },
    {
      "epoch": 0.12061014544164597,
      "grad_norm": 0.1779017597436905,
      "learning_rate": 4.585848612093283e-05,
      "loss": 0.3244,
      "num_tokens": 646252937.0,
      "step": 1020
    },
    {
      "epoch": 0.12072839068227503,
      "grad_norm": 0.19630023837089539,
      "learning_rate": 4.583116668695988e-05,
      "loss": 0.4196,
      "num_tokens": 646886273.0,
      "step": 1021
    },
    {
      "epoch": 0.12084663592290411,
      "grad_norm": 0.1875283569097519,
      "learning_rate": 4.5803830273582845e-05,
      "loss": 0.4122,
      "num_tokens": 647519729.0,
      "step": 1022
    },
    {
      "epoch": 0.12096488116353317,
      "grad_norm": 0.17070959508419037,
      "learning_rate": 4.5776476916975824e-05,
      "loss": 0.3611,
      "num_tokens": 648154792.0,
      "step": 1023
    },
    {
      "epoch": 0.12108312640416223,
      "grad_norm": 0.18787986040115356,
      "learning_rate": 4.5749106653335366e-05,
      "loss": 0.3684,
      "num_tokens": 648787406.0,
      "step": 1024
    },
    {
      "epoch": 0.12120137164479129,
      "grad_norm": 0.18751688301563263,
      "learning_rate": 4.5721719518880354e-05,
      "loss": 0.3911,
      "num_tokens": 649419550.0,
      "step": 1025
    },
    {
      "epoch": 0.12131961688542037,
      "grad_norm": 0.16276107728481293,
      "learning_rate": 4.569431554985202e-05,
      "loss": 0.3601,
      "num_tokens": 650055332.0,
      "step": 1026
    },
    {
      "epoch": 0.12143786212604943,
      "grad_norm": 0.19446961581707,
      "learning_rate": 4.566689478251388e-05,
      "loss": 0.4196,
      "num_tokens": 650692693.0,
      "step": 1027
    },
    {
      "epoch": 0.12155610736667849,
      "grad_norm": 0.18988844752311707,
      "learning_rate": 4.563945725315164e-05,
      "loss": 0.3915,
      "num_tokens": 651324330.0,
      "step": 1028
    },
    {
      "epoch": 0.12167435260730755,
      "grad_norm": 0.17802836000919342,
      "learning_rate": 4.561200299807324e-05,
      "loss": 0.3373,
      "num_tokens": 651953179.0,
      "step": 1029
    },
    {
      "epoch": 0.12179259784793663,
      "grad_norm": 0.19093678891658783,
      "learning_rate": 4.558453205360872e-05,
      "loss": 0.3633,
      "num_tokens": 652577781.0,
      "step": 1030
    },
    {
      "epoch": 0.12191084308856569,
      "grad_norm": 0.18175625801086426,
      "learning_rate": 4.55570444561102e-05,
      "loss": 0.3651,
      "num_tokens": 653215791.0,
      "step": 1031
    },
    {
      "epoch": 0.12202908832919475,
      "grad_norm": 0.1993628293275833,
      "learning_rate": 4.5529540241951866e-05,
      "loss": 0.398,
      "num_tokens": 653817011.0,
      "step": 1032
    },
    {
      "epoch": 0.12214733356982381,
      "grad_norm": 0.16179391741752625,
      "learning_rate": 4.550201944752987e-05,
      "loss": 0.3216,
      "num_tokens": 654453498.0,
      "step": 1033
    },
    {
      "epoch": 0.12226557881045289,
      "grad_norm": 0.19868984818458557,
      "learning_rate": 4.547448210926231e-05,
      "loss": 0.4069,
      "num_tokens": 655090159.0,
      "step": 1034
    },
    {
      "epoch": 0.12238382405108195,
      "grad_norm": 0.20796185731887817,
      "learning_rate": 4.544692826358916e-05,
      "loss": 0.4258,
      "num_tokens": 655725879.0,
      "step": 1035
    },
    {
      "epoch": 0.12250206929171101,
      "grad_norm": 0.19127629697322845,
      "learning_rate": 4.541935794697228e-05,
      "loss": 0.3639,
      "num_tokens": 656361031.0,
      "step": 1036
    },
    {
      "epoch": 0.12262031453234007,
      "grad_norm": 0.18504977226257324,
      "learning_rate": 4.53917711958953e-05,
      "loss": 0.4332,
      "num_tokens": 656995890.0,
      "step": 1037
    },
    {
      "epoch": 0.12273855977296914,
      "grad_norm": 0.1742541640996933,
      "learning_rate": 4.536416804686357e-05,
      "loss": 0.3658,
      "num_tokens": 657629671.0,
      "step": 1038
    },
    {
      "epoch": 0.1228568050135982,
      "grad_norm": 0.18453232944011688,
      "learning_rate": 4.53365485364042e-05,
      "loss": 0.3864,
      "num_tokens": 658262765.0,
      "step": 1039
    },
    {
      "epoch": 0.12297505025422727,
      "grad_norm": 0.18397222459316254,
      "learning_rate": 4.5308912701065896e-05,
      "loss": 0.3739,
      "num_tokens": 658892913.0,
      "step": 1040
    },
    {
      "epoch": 0.12309329549485633,
      "grad_norm": 0.19018225371837616,
      "learning_rate": 4.5281260577419006e-05,
      "loss": 0.3694,
      "num_tokens": 659528177.0,
      "step": 1041
    },
    {
      "epoch": 0.1232115407354854,
      "grad_norm": 0.1874883472919464,
      "learning_rate": 4.525359220205541e-05,
      "loss": 0.3817,
      "num_tokens": 660159039.0,
      "step": 1042
    },
    {
      "epoch": 0.12332978597611446,
      "grad_norm": 0.18878668546676636,
      "learning_rate": 4.522590761158849e-05,
      "loss": 0.3723,
      "num_tokens": 660793735.0,
      "step": 1043
    },
    {
      "epoch": 0.12344803121674353,
      "grad_norm": 0.18397924304008484,
      "learning_rate": 4.51982068426531e-05,
      "loss": 0.3895,
      "num_tokens": 661430126.0,
      "step": 1044
    },
    {
      "epoch": 0.12356627645737259,
      "grad_norm": 0.19383513927459717,
      "learning_rate": 4.517048993190551e-05,
      "loss": 0.3989,
      "num_tokens": 662069040.0,
      "step": 1045
    },
    {
      "epoch": 0.12368452169800166,
      "grad_norm": 0.1832335889339447,
      "learning_rate": 4.514275691602332e-05,
      "loss": 0.3551,
      "num_tokens": 662703135.0,
      "step": 1046
    },
    {
      "epoch": 0.12380276693863072,
      "grad_norm": 0.2143872082233429,
      "learning_rate": 4.511500783170546e-05,
      "loss": 0.4042,
      "num_tokens": 663337022.0,
      "step": 1047
    },
    {
      "epoch": 0.12392101217925978,
      "grad_norm": 0.20561742782592773,
      "learning_rate": 4.508724271567214e-05,
      "loss": 0.3758,
      "num_tokens": 663969631.0,
      "step": 1048
    },
    {
      "epoch": 0.12403925741988885,
      "grad_norm": 0.18738333880901337,
      "learning_rate": 4.505946160466476e-05,
      "loss": 0.4281,
      "num_tokens": 664607928.0,
      "step": 1049
    },
    {
      "epoch": 0.12415750266051792,
      "grad_norm": 0.1881737858057022,
      "learning_rate": 4.503166453544587e-05,
      "loss": 0.4124,
      "num_tokens": 665246971.0,
      "step": 1050
    },
    {
      "epoch": 0.12427574790114698,
      "grad_norm": 0.2155761420726776,
      "learning_rate": 4.500385154479921e-05,
      "loss": 0.3704,
      "num_tokens": 665880124.0,
      "step": 1051
    },
    {
      "epoch": 0.12439399314177604,
      "grad_norm": 0.17276139557361603,
      "learning_rate": 4.497602266952949e-05,
      "loss": 0.3857,
      "num_tokens": 666516590.0,
      "step": 1052
    },
    {
      "epoch": 0.1245122383824051,
      "grad_norm": 0.18994863331317902,
      "learning_rate": 4.494817794646252e-05,
      "loss": 0.3534,
      "num_tokens": 667141423.0,
      "step": 1053
    },
    {
      "epoch": 0.12463048362303417,
      "grad_norm": 0.2097732126712799,
      "learning_rate": 4.492031741244504e-05,
      "loss": 0.3947,
      "num_tokens": 667775999.0,
      "step": 1054
    },
    {
      "epoch": 0.12474872886366324,
      "grad_norm": 0.20371286571025848,
      "learning_rate": 4.4892441104344724e-05,
      "loss": 0.384,
      "num_tokens": 668413601.0,
      "step": 1055
    },
    {
      "epoch": 0.1248669741042923,
      "grad_norm": 0.17550261318683624,
      "learning_rate": 4.486454905905012e-05,
      "loss": 0.3732,
      "num_tokens": 669045590.0,
      "step": 1056
    },
    {
      "epoch": 0.12498521934492136,
      "grad_norm": 0.21981221437454224,
      "learning_rate": 4.48366413134706e-05,
      "loss": 0.399,
      "num_tokens": 669680260.0,
      "step": 1057
    },
    {
      "epoch": 0.12510346458555044,
      "grad_norm": 0.20199020206928253,
      "learning_rate": 4.480871790453632e-05,
      "loss": 0.3682,
      "num_tokens": 670314552.0,
      "step": 1058
    },
    {
      "epoch": 0.1252217098261795,
      "grad_norm": 0.20767715573310852,
      "learning_rate": 4.478077886919814e-05,
      "loss": 0.3715,
      "num_tokens": 670948491.0,
      "step": 1059
    },
    {
      "epoch": 0.12533995506680856,
      "grad_norm": 0.1847984343767166,
      "learning_rate": 4.4752824244427635e-05,
      "loss": 0.354,
      "num_tokens": 671580682.0,
      "step": 1060
    },
    {
      "epoch": 0.12545820030743762,
      "grad_norm": 0.1866157501935959,
      "learning_rate": 4.472485406721697e-05,
      "loss": 0.3585,
      "num_tokens": 672210589.0,
      "step": 1061
    },
    {
      "epoch": 0.12557644554806668,
      "grad_norm": 0.20393408834934235,
      "learning_rate": 4.4696868374578916e-05,
      "loss": 0.3824,
      "num_tokens": 672840526.0,
      "step": 1062
    },
    {
      "epoch": 0.12569469078869575,
      "grad_norm": 0.19358152151107788,
      "learning_rate": 4.4668867203546766e-05,
      "loss": 0.3702,
      "num_tokens": 673466186.0,
      "step": 1063
    },
    {
      "epoch": 0.1258129360293248,
      "grad_norm": 0.16034334897994995,
      "learning_rate": 4.46408505911743e-05,
      "loss": 0.3063,
      "num_tokens": 674089397.0,
      "step": 1064
    },
    {
      "epoch": 0.1259311812699539,
      "grad_norm": 0.2177608758211136,
      "learning_rate": 4.461281857453572e-05,
      "loss": 0.3839,
      "num_tokens": 674726487.0,
      "step": 1065
    },
    {
      "epoch": 0.12604942651058296,
      "grad_norm": 0.16433891654014587,
      "learning_rate": 4.4584771190725644e-05,
      "loss": 0.3526,
      "num_tokens": 675360889.0,
      "step": 1066
    },
    {
      "epoch": 0.12616767175121202,
      "grad_norm": 0.17344233393669128,
      "learning_rate": 4.4556708476858985e-05,
      "loss": 0.3615,
      "num_tokens": 675999411.0,
      "step": 1067
    },
    {
      "epoch": 0.12628591699184108,
      "grad_norm": 0.1769314557313919,
      "learning_rate": 4.4528630470070964e-05,
      "loss": 0.3477,
      "num_tokens": 676631267.0,
      "step": 1068
    },
    {
      "epoch": 0.12640416223247014,
      "grad_norm": 0.18346329033374786,
      "learning_rate": 4.450053720751703e-05,
      "loss": 0.3735,
      "num_tokens": 677241383.0,
      "step": 1069
    },
    {
      "epoch": 0.1265224074730992,
      "grad_norm": 0.1862960308790207,
      "learning_rate": 4.447242872637282e-05,
      "loss": 0.3876,
      "num_tokens": 677877358.0,
      "step": 1070
    },
    {
      "epoch": 0.12664065271372826,
      "grad_norm": 0.17572778463363647,
      "learning_rate": 4.444430506383414e-05,
      "loss": 0.3651,
      "num_tokens": 678505498.0,
      "step": 1071
    },
    {
      "epoch": 0.12675889795435732,
      "grad_norm": 0.19853392243385315,
      "learning_rate": 4.441616625711684e-05,
      "loss": 0.359,
      "num_tokens": 679124916.0,
      "step": 1072
    },
    {
      "epoch": 0.1268771431949864,
      "grad_norm": 0.1989092230796814,
      "learning_rate": 4.4388012343456826e-05,
      "loss": 0.403,
      "num_tokens": 679749193.0,
      "step": 1073
    },
    {
      "epoch": 0.12699538843561547,
      "grad_norm": 0.16579356789588928,
      "learning_rate": 4.435984336011001e-05,
      "loss": 0.3409,
      "num_tokens": 680380293.0,
      "step": 1074
    },
    {
      "epoch": 0.12711363367624454,
      "grad_norm": 0.17490504682064056,
      "learning_rate": 4.433165934435223e-05,
      "loss": 0.3768,
      "num_tokens": 681017380.0,
      "step": 1075
    },
    {
      "epoch": 0.1272318789168736,
      "grad_norm": 0.1843899041414261,
      "learning_rate": 4.430346033347923e-05,
      "loss": 0.3796,
      "num_tokens": 681647130.0,
      "step": 1076
    },
    {
      "epoch": 0.12735012415750266,
      "grad_norm": 0.17463913559913635,
      "learning_rate": 4.427524636480658e-05,
      "loss": 0.3586,
      "num_tokens": 682285081.0,
      "step": 1077
    },
    {
      "epoch": 0.12746836939813172,
      "grad_norm": 0.1665404886007309,
      "learning_rate": 4.4247017475669646e-05,
      "loss": 0.376,
      "num_tokens": 682919537.0,
      "step": 1078
    },
    {
      "epoch": 0.12758661463876078,
      "grad_norm": 0.18626192212104797,
      "learning_rate": 4.421877370342355e-05,
      "loss": 0.3868,
      "num_tokens": 683554203.0,
      "step": 1079
    },
    {
      "epoch": 0.12770485987938984,
      "grad_norm": 0.18904843926429749,
      "learning_rate": 4.4190515085443104e-05,
      "loss": 0.3929,
      "num_tokens": 684192561.0,
      "step": 1080
    },
    {
      "epoch": 0.12782310512001893,
      "grad_norm": 0.16643720865249634,
      "learning_rate": 4.416224165912276e-05,
      "loss": 0.3875,
      "num_tokens": 684826318.0,
      "step": 1081
    },
    {
      "epoch": 0.127941350360648,
      "grad_norm": 0.17822794616222382,
      "learning_rate": 4.4133953461876576e-05,
      "loss": 0.3997,
      "num_tokens": 685457932.0,
      "step": 1082
    },
    {
      "epoch": 0.12805959560127705,
      "grad_norm": 0.1729457676410675,
      "learning_rate": 4.410565053113814e-05,
      "loss": 0.3648,
      "num_tokens": 686094682.0,
      "step": 1083
    },
    {
      "epoch": 0.12817784084190612,
      "grad_norm": 0.17672215402126312,
      "learning_rate": 4.407733290436055e-05,
      "loss": 0.3787,
      "num_tokens": 686724923.0,
      "step": 1084
    },
    {
      "epoch": 0.12829608608253518,
      "grad_norm": 0.16756568849086761,
      "learning_rate": 4.4049000619016345e-05,
      "loss": 0.3775,
      "num_tokens": 687358762.0,
      "step": 1085
    },
    {
      "epoch": 0.12841433132316424,
      "grad_norm": 0.17241579294204712,
      "learning_rate": 4.402065371259747e-05,
      "loss": 0.3743,
      "num_tokens": 687997038.0,
      "step": 1086
    },
    {
      "epoch": 0.1285325765637933,
      "grad_norm": 0.16503721475601196,
      "learning_rate": 4.399229222261522e-05,
      "loss": 0.3895,
      "num_tokens": 688632728.0,
      "step": 1087
    },
    {
      "epoch": 0.12865082180442236,
      "grad_norm": 0.18014270067214966,
      "learning_rate": 4.3963916186600164e-05,
      "loss": 0.3913,
      "num_tokens": 689237615.0,
      "step": 1088
    },
    {
      "epoch": 0.12876906704505145,
      "grad_norm": 0.18278226256370544,
      "learning_rate": 4.393552564210214e-05,
      "loss": 0.3951,
      "num_tokens": 689872830.0,
      "step": 1089
    },
    {
      "epoch": 0.1288873122856805,
      "grad_norm": 0.17956025898456573,
      "learning_rate": 4.3907120626690194e-05,
      "loss": 0.3713,
      "num_tokens": 690505835.0,
      "step": 1090
    },
    {
      "epoch": 0.12900555752630957,
      "grad_norm": 0.18163177371025085,
      "learning_rate": 4.387870117795249e-05,
      "loss": 0.4,
      "num_tokens": 691138889.0,
      "step": 1091
    },
    {
      "epoch": 0.12912380276693863,
      "grad_norm": 0.17895618081092834,
      "learning_rate": 4.385026733349632e-05,
      "loss": 0.3824,
      "num_tokens": 691778212.0,
      "step": 1092
    },
    {
      "epoch": 0.1292420480075677,
      "grad_norm": 0.17014320194721222,
      "learning_rate": 4.382181913094801e-05,
      "loss": 0.356,
      "num_tokens": 692400014.0,
      "step": 1093
    },
    {
      "epoch": 0.12936029324819676,
      "grad_norm": 0.17010079324245453,
      "learning_rate": 4.37933566079529e-05,
      "loss": 0.3583,
      "num_tokens": 693035177.0,
      "step": 1094
    },
    {
      "epoch": 0.12947853848882582,
      "grad_norm": 0.1737738847732544,
      "learning_rate": 4.376487980217527e-05,
      "loss": 0.3793,
      "num_tokens": 693672098.0,
      "step": 1095
    },
    {
      "epoch": 0.12959678372945488,
      "grad_norm": 0.157965287566185,
      "learning_rate": 4.3736388751298306e-05,
      "loss": 0.3352,
      "num_tokens": 694306278.0,
      "step": 1096
    },
    {
      "epoch": 0.12971502897008397,
      "grad_norm": 0.19977279007434845,
      "learning_rate": 4.370788349302403e-05,
      "loss": 0.4292,
      "num_tokens": 694943202.0,
      "step": 1097
    },
    {
      "epoch": 0.12983327421071303,
      "grad_norm": 0.1841733455657959,
      "learning_rate": 4.3679364065073274e-05,
      "loss": 0.398,
      "num_tokens": 695581748.0,
      "step": 1098
    },
    {
      "epoch": 0.1299515194513421,
      "grad_norm": 0.18417219817638397,
      "learning_rate": 4.365083050518563e-05,
      "loss": 0.4095,
      "num_tokens": 696217907.0,
      "step": 1099
    },
    {
      "epoch": 0.13006976469197115,
      "grad_norm": 0.17778581380844116,
      "learning_rate": 4.362228285111939e-05,
      "loss": 0.3461,
      "num_tokens": 696832400.0,
      "step": 1100
    },
    {
      "epoch": 0.1301880099326002,
      "grad_norm": 0.20819556713104248,
      "learning_rate": 4.359372114065146e-05,
      "loss": 0.4048,
      "num_tokens": 697462714.0,
      "step": 1101
    },
    {
      "epoch": 0.13030625517322927,
      "grad_norm": 0.1679416298866272,
      "learning_rate": 4.356514541157742e-05,
      "loss": 0.3306,
      "num_tokens": 698096072.0,
      "step": 1102
    },
    {
      "epoch": 0.13042450041385834,
      "grad_norm": 0.17605753242969513,
      "learning_rate": 4.3536555701711316e-05,
      "loss": 0.3654,
      "num_tokens": 698732104.0,
      "step": 1103
    },
    {
      "epoch": 0.1305427456544874,
      "grad_norm": 0.18534035980701447,
      "learning_rate": 4.350795204888576e-05,
      "loss": 0.3637,
      "num_tokens": 699360778.0,
      "step": 1104
    },
    {
      "epoch": 0.13066099089511646,
      "grad_norm": 0.1880866438150406,
      "learning_rate": 4.3479334490951784e-05,
      "loss": 0.3639,
      "num_tokens": 699998919.0,
      "step": 1105
    },
    {
      "epoch": 0.13077923613574555,
      "grad_norm": 0.19816787540912628,
      "learning_rate": 4.345070306577882e-05,
      "loss": 0.3717,
      "num_tokens": 700636435.0,
      "step": 1106
    },
    {
      "epoch": 0.1308974813763746,
      "grad_norm": 0.1735011786222458,
      "learning_rate": 4.342205781125466e-05,
      "loss": 0.3669,
      "num_tokens": 701266798.0,
      "step": 1107
    },
    {
      "epoch": 0.13101572661700367,
      "grad_norm": 0.2126895934343338,
      "learning_rate": 4.33933987652854e-05,
      "loss": 0.398,
      "num_tokens": 701896284.0,
      "step": 1108
    },
    {
      "epoch": 0.13113397185763273,
      "grad_norm": 0.19128330051898956,
      "learning_rate": 4.336472596579538e-05,
      "loss": 0.4147,
      "num_tokens": 702532392.0,
      "step": 1109
    },
    {
      "epoch": 0.1312522170982618,
      "grad_norm": 0.17880013585090637,
      "learning_rate": 4.333603945072713e-05,
      "loss": 0.3671,
      "num_tokens": 703167803.0,
      "step": 1110
    },
    {
      "epoch": 0.13137046233889085,
      "grad_norm": 0.206880122423172,
      "learning_rate": 4.330733925804134e-05,
      "loss": 0.4211,
      "num_tokens": 703804312.0,
      "step": 1111
    },
    {
      "epoch": 0.13148870757951991,
      "grad_norm": 0.178485706448555,
      "learning_rate": 4.3278625425716805e-05,
      "loss": 0.3527,
      "num_tokens": 704434487.0,
      "step": 1112
    },
    {
      "epoch": 0.13160695282014898,
      "grad_norm": 0.2033427357673645,
      "learning_rate": 4.324989799175037e-05,
      "loss": 0.3989,
      "num_tokens": 705069074.0,
      "step": 1113
    },
    {
      "epoch": 0.13172519806077806,
      "grad_norm": 0.19211621582508087,
      "learning_rate": 4.322115699415685e-05,
      "loss": 0.3784,
      "num_tokens": 705694863.0,
      "step": 1114
    },
    {
      "epoch": 0.13184344330140713,
      "grad_norm": 0.19245471060276031,
      "learning_rate": 4.319240247096905e-05,
      "loss": 0.3802,
      "num_tokens": 706332062.0,
      "step": 1115
    },
    {
      "epoch": 0.1319616885420362,
      "grad_norm": 0.18468594551086426,
      "learning_rate": 4.316363446023764e-05,
      "loss": 0.3732,
      "num_tokens": 706961719.0,
      "step": 1116
    },
    {
      "epoch": 0.13207993378266525,
      "grad_norm": 0.18160372972488403,
      "learning_rate": 4.313485300003117e-05,
      "loss": 0.3549,
      "num_tokens": 707593241.0,
      "step": 1117
    },
    {
      "epoch": 0.1321981790232943,
      "grad_norm": 0.17797401547431946,
      "learning_rate": 4.310605812843595e-05,
      "loss": 0.4154,
      "num_tokens": 708223670.0,
      "step": 1118
    },
    {
      "epoch": 0.13231642426392337,
      "grad_norm": 0.19047312438488007,
      "learning_rate": 4.3077249883556075e-05,
      "loss": 0.3714,
      "num_tokens": 708853803.0,
      "step": 1119
    },
    {
      "epoch": 0.13243466950455243,
      "grad_norm": 0.19819514453411102,
      "learning_rate": 4.304842830351331e-05,
      "loss": 0.3865,
      "num_tokens": 709488191.0,
      "step": 1120
    },
    {
      "epoch": 0.1325529147451815,
      "grad_norm": 0.17723028361797333,
      "learning_rate": 4.301959342644706e-05,
      "loss": 0.3588,
      "num_tokens": 710121271.0,
      "step": 1121
    },
    {
      "epoch": 0.13267115998581058,
      "grad_norm": 0.1742127537727356,
      "learning_rate": 4.299074529051437e-05,
      "loss": 0.3779,
      "num_tokens": 710755348.0,
      "step": 1122
    },
    {
      "epoch": 0.13278940522643964,
      "grad_norm": 0.19025954604148865,
      "learning_rate": 4.296188393388978e-05,
      "loss": 0.4193,
      "num_tokens": 711388900.0,
      "step": 1123
    },
    {
      "epoch": 0.1329076504670687,
      "grad_norm": 0.18485389649868011,
      "learning_rate": 4.293300939476534e-05,
      "loss": 0.391,
      "num_tokens": 712027710.0,
      "step": 1124
    },
    {
      "epoch": 0.13302589570769777,
      "grad_norm": 0.20137916505336761,
      "learning_rate": 4.290412171135057e-05,
      "loss": 0.386,
      "num_tokens": 712662431.0,
      "step": 1125
    },
    {
      "epoch": 0.13314414094832683,
      "grad_norm": 0.18418936431407928,
      "learning_rate": 4.287522092187237e-05,
      "loss": 0.3778,
      "num_tokens": 713298321.0,
      "step": 1126
    },
    {
      "epoch": 0.1332623861889559,
      "grad_norm": 0.1847471296787262,
      "learning_rate": 4.284630706457495e-05,
      "loss": 0.3939,
      "num_tokens": 713929502.0,
      "step": 1127
    },
    {
      "epoch": 0.13338063142958495,
      "grad_norm": 0.18358609080314636,
      "learning_rate": 4.281738017771985e-05,
      "loss": 0.392,
      "num_tokens": 714563684.0,
      "step": 1128
    },
    {
      "epoch": 0.133498876670214,
      "grad_norm": 0.19019533693790436,
      "learning_rate": 4.278844029958586e-05,
      "loss": 0.4139,
      "num_tokens": 715194139.0,
      "step": 1129
    },
    {
      "epoch": 0.1336171219108431,
      "grad_norm": 0.17144839465618134,
      "learning_rate": 4.275948746846891e-05,
      "loss": 0.3872,
      "num_tokens": 715829633.0,
      "step": 1130
    },
    {
      "epoch": 0.13373536715147216,
      "grad_norm": 0.20015180110931396,
      "learning_rate": 4.273052172268213e-05,
      "loss": 0.3855,
      "num_tokens": 716437097.0,
      "step": 1131
    },
    {
      "epoch": 0.13385361239210122,
      "grad_norm": 0.18676939606666565,
      "learning_rate": 4.2701543100555715e-05,
      "loss": 0.3251,
      "num_tokens": 717069185.0,
      "step": 1132
    },
    {
      "epoch": 0.13397185763273028,
      "grad_norm": 0.16965261101722717,
      "learning_rate": 4.2672551640436876e-05,
      "loss": 0.3968,
      "num_tokens": 717707723.0,
      "step": 1133
    },
    {
      "epoch": 0.13409010287335935,
      "grad_norm": 0.1700984537601471,
      "learning_rate": 4.264354738068985e-05,
      "loss": 0.3727,
      "num_tokens": 718339982.0,
      "step": 1134
    },
    {
      "epoch": 0.1342083481139884,
      "grad_norm": 0.2030845433473587,
      "learning_rate": 4.2614530359695786e-05,
      "loss": 0.3893,
      "num_tokens": 718970674.0,
      "step": 1135
    },
    {
      "epoch": 0.13432659335461747,
      "grad_norm": 0.1986686736345291,
      "learning_rate": 4.258550061585275e-05,
      "loss": 0.4172,
      "num_tokens": 719606724.0,
      "step": 1136
    },
    {
      "epoch": 0.13444483859524653,
      "grad_norm": 0.17563633620738983,
      "learning_rate": 4.25564581875756e-05,
      "loss": 0.3858,
      "num_tokens": 720209640.0,
      "step": 1137
    },
    {
      "epoch": 0.13456308383587562,
      "grad_norm": 0.17847663164138794,
      "learning_rate": 4.2527403113296014e-05,
      "loss": 0.3804,
      "num_tokens": 720845332.0,
      "step": 1138
    },
    {
      "epoch": 0.13468132907650468,
      "grad_norm": 0.17714640498161316,
      "learning_rate": 4.249833543146239e-05,
      "loss": 0.3708,
      "num_tokens": 721471001.0,
      "step": 1139
    },
    {
      "epoch": 0.13479957431713374,
      "grad_norm": 0.17541660368442535,
      "learning_rate": 4.246925518053981e-05,
      "loss": 0.3568,
      "num_tokens": 722105712.0,
      "step": 1140
    },
    {
      "epoch": 0.1349178195577628,
      "grad_norm": 0.18325071036815643,
      "learning_rate": 4.2440162399010005e-05,
      "loss": 0.4037,
      "num_tokens": 722735260.0,
      "step": 1141
    },
    {
      "epoch": 0.13503606479839186,
      "grad_norm": 0.16364189982414246,
      "learning_rate": 4.241105712537126e-05,
      "loss": 0.3644,
      "num_tokens": 723369238.0,
      "step": 1142
    },
    {
      "epoch": 0.13515431003902093,
      "grad_norm": 0.18739944696426392,
      "learning_rate": 4.238193939813841e-05,
      "loss": 0.3838,
      "num_tokens": 724005091.0,
      "step": 1143
    },
    {
      "epoch": 0.13527255527965,
      "grad_norm": 0.1882290542125702,
      "learning_rate": 4.235280925584277e-05,
      "loss": 0.3571,
      "num_tokens": 724644703.0,
      "step": 1144
    },
    {
      "epoch": 0.13539080052027905,
      "grad_norm": 0.17208614945411682,
      "learning_rate": 4.2323666737032075e-05,
      "loss": 0.3854,
      "num_tokens": 725277008.0,
      "step": 1145
    },
    {
      "epoch": 0.13550904576090814,
      "grad_norm": 0.1727161407470703,
      "learning_rate": 4.2294511880270435e-05,
      "loss": 0.3558,
      "num_tokens": 725912854.0,
      "step": 1146
    },
    {
      "epoch": 0.1356272910015372,
      "grad_norm": 0.19190965592861176,
      "learning_rate": 4.226534472413829e-05,
      "loss": 0.4023,
      "num_tokens": 726552374.0,
      "step": 1147
    },
    {
      "epoch": 0.13574553624216626,
      "grad_norm": 0.18595655262470245,
      "learning_rate": 4.223616530723238e-05,
      "loss": 0.3585,
      "num_tokens": 727190596.0,
      "step": 1148
    },
    {
      "epoch": 0.13586378148279532,
      "grad_norm": 0.17069599032402039,
      "learning_rate": 4.220697366816563e-05,
      "loss": 0.3421,
      "num_tokens": 727820231.0,
      "step": 1149
    },
    {
      "epoch": 0.13598202672342438,
      "grad_norm": 0.1840461939573288,
      "learning_rate": 4.2177769845567157e-05,
      "loss": 0.3332,
      "num_tokens": 728440540.0,
      "step": 1150
    },
    {
      "epoch": 0.13610027196405344,
      "grad_norm": 0.2065514773130417,
      "learning_rate": 4.2148553878082203e-05,
      "loss": 0.4219,
      "num_tokens": 729072563.0,
      "step": 1151
    },
    {
      "epoch": 0.1362185172046825,
      "grad_norm": 0.19268664717674255,
      "learning_rate": 4.211932580437208e-05,
      "loss": 0.4187,
      "num_tokens": 729704300.0,
      "step": 1152
    },
    {
      "epoch": 0.13633676244531157,
      "grad_norm": 0.1874132603406906,
      "learning_rate": 4.2090085663114126e-05,
      "loss": 0.3949,
      "num_tokens": 730308427.0,
      "step": 1153
    },
    {
      "epoch": 0.13645500768594065,
      "grad_norm": 0.2015058696269989,
      "learning_rate": 4.206083349300163e-05,
      "loss": 0.4352,
      "num_tokens": 730946156.0,
      "step": 1154
    },
    {
      "epoch": 0.13657325292656972,
      "grad_norm": 0.1677466779947281,
      "learning_rate": 4.20315693327438e-05,
      "loss": 0.3534,
      "num_tokens": 731577262.0,
      "step": 1155
    },
    {
      "epoch": 0.13669149816719878,
      "grad_norm": 0.21406646072864532,
      "learning_rate": 4.2002293221065737e-05,
      "loss": 0.403,
      "num_tokens": 732204517.0,
      "step": 1156
    },
    {
      "epoch": 0.13680974340782784,
      "grad_norm": 0.17139025032520294,
      "learning_rate": 4.197300519670834e-05,
      "loss": 0.3519,
      "num_tokens": 732841654.0,
      "step": 1157
    },
    {
      "epoch": 0.1369279886484569,
      "grad_norm": 0.19764773547649384,
      "learning_rate": 4.1943705298428264e-05,
      "loss": 0.3785,
      "num_tokens": 733481344.0,
      "step": 1158
    },
    {
      "epoch": 0.13704623388908596,
      "grad_norm": 0.1817982941865921,
      "learning_rate": 4.191439356499788e-05,
      "loss": 0.3851,
      "num_tokens": 734116157.0,
      "step": 1159
    },
    {
      "epoch": 0.13716447912971502,
      "grad_norm": 0.1699836552143097,
      "learning_rate": 4.188507003520524e-05,
      "loss": 0.3549,
      "num_tokens": 734752960.0,
      "step": 1160
    },
    {
      "epoch": 0.13728272437034408,
      "grad_norm": 0.18850567936897278,
      "learning_rate": 4.1855734747853974e-05,
      "loss": 0.3689,
      "num_tokens": 735390605.0,
      "step": 1161
    },
    {
      "epoch": 0.13740096961097314,
      "grad_norm": 0.1694057732820511,
      "learning_rate": 4.1826387741763294e-05,
      "loss": 0.3493,
      "num_tokens": 736025653.0,
      "step": 1162
    },
    {
      "epoch": 0.13751921485160223,
      "grad_norm": 0.19740477204322815,
      "learning_rate": 4.1797029055767917e-05,
      "loss": 0.412,
      "num_tokens": 736662381.0,
      "step": 1163
    },
    {
      "epoch": 0.1376374600922313,
      "grad_norm": 0.1692422777414322,
      "learning_rate": 4.1767658728718005e-05,
      "loss": 0.3781,
      "num_tokens": 737298285.0,
      "step": 1164
    },
    {
      "epoch": 0.13775570533286036,
      "grad_norm": 0.18291229009628296,
      "learning_rate": 4.173827679947914e-05,
      "loss": 0.3646,
      "num_tokens": 737932027.0,
      "step": 1165
    },
    {
      "epoch": 0.13787395057348942,
      "grad_norm": 0.17919090390205383,
      "learning_rate": 4.170888330693226e-05,
      "loss": 0.402,
      "num_tokens": 738566204.0,
      "step": 1166
    },
    {
      "epoch": 0.13799219581411848,
      "grad_norm": 0.16608217358589172,
      "learning_rate": 4.167947828997356e-05,
      "loss": 0.3582,
      "num_tokens": 739201202.0,
      "step": 1167
    },
    {
      "epoch": 0.13811044105474754,
      "grad_norm": 0.1820799559354782,
      "learning_rate": 4.1650061787514534e-05,
      "loss": 0.3757,
      "num_tokens": 739833572.0,
      "step": 1168
    },
    {
      "epoch": 0.1382286862953766,
      "grad_norm": 0.1864081174135208,
      "learning_rate": 4.162063383848186e-05,
      "loss": 0.3813,
      "num_tokens": 740467809.0,
      "step": 1169
    },
    {
      "epoch": 0.13834693153600566,
      "grad_norm": 0.19109085202217102,
      "learning_rate": 4.159119448181737e-05,
      "loss": 0.3954,
      "num_tokens": 741101167.0,
      "step": 1170
    },
    {
      "epoch": 0.13846517677663475,
      "grad_norm": 0.18018218874931335,
      "learning_rate": 4.156174375647798e-05,
      "loss": 0.3666,
      "num_tokens": 741738147.0,
      "step": 1171
    },
    {
      "epoch": 0.1385834220172638,
      "grad_norm": 0.1801140159368515,
      "learning_rate": 4.1532281701435644e-05,
      "loss": 0.3747,
      "num_tokens": 742371844.0,
      "step": 1172
    },
    {
      "epoch": 0.13870166725789287,
      "grad_norm": 0.18970145285129547,
      "learning_rate": 4.150280835567733e-05,
      "loss": 0.3969,
      "num_tokens": 743011533.0,
      "step": 1173
    },
    {
      "epoch": 0.13881991249852194,
      "grad_norm": 0.20032352209091187,
      "learning_rate": 4.1473323758204925e-05,
      "loss": 0.4045,
      "num_tokens": 743646159.0,
      "step": 1174
    },
    {
      "epoch": 0.138938157739151,
      "grad_norm": 0.16128578782081604,
      "learning_rate": 4.144382794803523e-05,
      "loss": 0.3336,
      "num_tokens": 744282963.0,
      "step": 1175
    },
    {
      "epoch": 0.13905640297978006,
      "grad_norm": 0.2133799046278,
      "learning_rate": 4.1414320964199846e-05,
      "loss": 0.4102,
      "num_tokens": 744920861.0,
      "step": 1176
    },
    {
      "epoch": 0.13917464822040912,
      "grad_norm": 0.17610150575637817,
      "learning_rate": 4.1384802845745216e-05,
      "loss": 0.4052,
      "num_tokens": 745552170.0,
      "step": 1177
    },
    {
      "epoch": 0.13929289346103818,
      "grad_norm": 0.17871306836605072,
      "learning_rate": 4.1355273631732466e-05,
      "loss": 0.3641,
      "num_tokens": 746184040.0,
      "step": 1178
    },
    {
      "epoch": 0.13941113870166727,
      "grad_norm": 0.1855345219373703,
      "learning_rate": 4.132573336123741e-05,
      "loss": 0.3781,
      "num_tokens": 746821128.0,
      "step": 1179
    },
    {
      "epoch": 0.13952938394229633,
      "grad_norm": 0.1980353444814682,
      "learning_rate": 4.1296182073350536e-05,
      "loss": 0.393,
      "num_tokens": 747451607.0,
      "step": 1180
    },
    {
      "epoch": 0.1396476291829254,
      "grad_norm": 0.19999416172504425,
      "learning_rate": 4.126661980717686e-05,
      "loss": 0.378,
      "num_tokens": 748083163.0,
      "step": 1181
    },
    {
      "epoch": 0.13976587442355445,
      "grad_norm": 0.17589710652828217,
      "learning_rate": 4.123704660183597e-05,
      "loss": 0.3488,
      "num_tokens": 748721979.0,
      "step": 1182
    },
    {
      "epoch": 0.13988411966418351,
      "grad_norm": 0.20237183570861816,
      "learning_rate": 4.120746249646188e-05,
      "loss": 0.3976,
      "num_tokens": 749355743.0,
      "step": 1183
    },
    {
      "epoch": 0.14000236490481258,
      "grad_norm": 0.18042594194412231,
      "learning_rate": 4.1177867530203066e-05,
      "loss": 0.3601,
      "num_tokens": 749990459.0,
      "step": 1184
    },
    {
      "epoch": 0.14012061014544164,
      "grad_norm": 0.2047455608844757,
      "learning_rate": 4.114826174222237e-05,
      "loss": 0.3776,
      "num_tokens": 750625529.0,
      "step": 1185
    },
    {
      "epoch": 0.1402388553860707,
      "grad_norm": 0.17314770817756653,
      "learning_rate": 4.111864517169694e-05,
      "loss": 0.3613,
      "num_tokens": 751257308.0,
      "step": 1186
    },
    {
      "epoch": 0.1403571006266998,
      "grad_norm": 0.16046233475208282,
      "learning_rate": 4.108901785781821e-05,
      "loss": 0.3525,
      "num_tokens": 751890563.0,
      "step": 1187
    },
    {
      "epoch": 0.14047534586732885,
      "grad_norm": 0.18692271411418915,
      "learning_rate": 4.105937983979183e-05,
      "loss": 0.3497,
      "num_tokens": 752523163.0,
      "step": 1188
    },
    {
      "epoch": 0.1405935911079579,
      "grad_norm": 0.16772404313087463,
      "learning_rate": 4.102973115683758e-05,
      "loss": 0.3636,
      "num_tokens": 753157270.0,
      "step": 1189
    },
    {
      "epoch": 0.14071183634858697,
      "grad_norm": 0.1749587059020996,
      "learning_rate": 4.1000071848189406e-05,
      "loss": 0.3721,
      "num_tokens": 753790631.0,
      "step": 1190
    },
    {
      "epoch": 0.14083008158921603,
      "grad_norm": 0.17635898292064667,
      "learning_rate": 4.0970401953095275e-05,
      "loss": 0.4219,
      "num_tokens": 754422344.0,
      "step": 1191
    },
    {
      "epoch": 0.1409483268298451,
      "grad_norm": 0.1795620322227478,
      "learning_rate": 4.094072151081719e-05,
      "loss": 0.4066,
      "num_tokens": 755054602.0,
      "step": 1192
    },
    {
      "epoch": 0.14106657207047416,
      "grad_norm": 0.19329644739627838,
      "learning_rate": 4.091103056063108e-05,
      "loss": 0.3965,
      "num_tokens": 755686194.0,
      "step": 1193
    },
    {
      "epoch": 0.14118481731110322,
      "grad_norm": 0.18483629822731018,
      "learning_rate": 4.088132914182682e-05,
      "loss": 0.3922,
      "num_tokens": 756324764.0,
      "step": 1194
    },
    {
      "epoch": 0.1413030625517323,
      "grad_norm": 0.193592831492424,
      "learning_rate": 4.085161729370809e-05,
      "loss": 0.412,
      "num_tokens": 756963038.0,
      "step": 1195
    },
    {
      "epoch": 0.14142130779236137,
      "grad_norm": 0.1793985515832901,
      "learning_rate": 4.0821895055592415e-05,
      "loss": 0.3256,
      "num_tokens": 757590896.0,
      "step": 1196
    },
    {
      "epoch": 0.14153955303299043,
      "grad_norm": 0.1903679072856903,
      "learning_rate": 4.0792162466811046e-05,
      "loss": 0.3892,
      "num_tokens": 758226989.0,
      "step": 1197
    },
    {
      "epoch": 0.1416577982736195,
      "grad_norm": 0.18117378652095795,
      "learning_rate": 4.076241956670894e-05,
      "loss": 0.3966,
      "num_tokens": 758863618.0,
      "step": 1198
    },
    {
      "epoch": 0.14177604351424855,
      "grad_norm": 0.17214766144752502,
      "learning_rate": 4.0732666394644675e-05,
      "loss": 0.3483,
      "num_tokens": 759499176.0,
      "step": 1199
    },
    {
      "epoch": 0.1418942887548776,
      "grad_norm": 0.2005872279405594,
      "learning_rate": 4.0702902989990464e-05,
      "loss": 0.403,
      "num_tokens": 760138725.0,
      "step": 1200
    },
    {
      "epoch": 0.14201253399550667,
      "grad_norm": 0.16486693918704987,
      "learning_rate": 4.0673129392132014e-05,
      "loss": 0.3784,
      "num_tokens": 760778173.0,
      "step": 1201
    },
    {
      "epoch": 0.14213077923613573,
      "grad_norm": 0.20838482677936554,
      "learning_rate": 4.064334564046856e-05,
      "loss": 0.427,
      "num_tokens": 761411949.0,
      "step": 1202
    },
    {
      "epoch": 0.14224902447676482,
      "grad_norm": 0.1762368530035019,
      "learning_rate": 4.061355177441275e-05,
      "loss": 0.37,
      "num_tokens": 762049507.0,
      "step": 1203
    },
    {
      "epoch": 0.14236726971739389,
      "grad_norm": 0.19547240436077118,
      "learning_rate": 4.058374783339063e-05,
      "loss": 0.41,
      "num_tokens": 762685373.0,
      "step": 1204
    },
    {
      "epoch": 0.14248551495802295,
      "grad_norm": 0.18869923055171967,
      "learning_rate": 4.0553933856841584e-05,
      "loss": 0.3857,
      "num_tokens": 763323824.0,
      "step": 1205
    },
    {
      "epoch": 0.142603760198652,
      "grad_norm": 0.18279902637004852,
      "learning_rate": 4.052410988421824e-05,
      "loss": 0.421,
      "num_tokens": 763962195.0,
      "step": 1206
    },
    {
      "epoch": 0.14272200543928107,
      "grad_norm": 0.17345914244651794,
      "learning_rate": 4.049427595498649e-05,
      "loss": 0.3871,
      "num_tokens": 764599228.0,
      "step": 1207
    },
    {
      "epoch": 0.14284025067991013,
      "grad_norm": 0.1748848259449005,
      "learning_rate": 4.046443210862538e-05,
      "loss": 0.342,
      "num_tokens": 765233848.0,
      "step": 1208
    },
    {
      "epoch": 0.1429584959205392,
      "grad_norm": 0.17195607721805573,
      "learning_rate": 4.043457838462712e-05,
      "loss": 0.39,
      "num_tokens": 765873589.0,
      "step": 1209
    },
    {
      "epoch": 0.14307674116116825,
      "grad_norm": 0.16931648552417755,
      "learning_rate": 4.040471482249691e-05,
      "loss": 0.3322,
      "num_tokens": 766504261.0,
      "step": 1210
    },
    {
      "epoch": 0.14319498640179731,
      "grad_norm": 0.19360356032848358,
      "learning_rate": 4.037484146175306e-05,
      "loss": 0.4047,
      "num_tokens": 767140922.0,
      "step": 1211
    },
    {
      "epoch": 0.1433132316424264,
      "grad_norm": 0.2044503539800644,
      "learning_rate": 4.034495834192678e-05,
      "loss": 0.3987,
      "num_tokens": 767775931.0,
      "step": 1212
    },
    {
      "epoch": 0.14343147688305546,
      "grad_norm": 0.18666554987430573,
      "learning_rate": 4.031506550256223e-05,
      "loss": 0.4094,
      "num_tokens": 768411905.0,
      "step": 1213
    },
    {
      "epoch": 0.14354972212368453,
      "grad_norm": 0.1968887448310852,
      "learning_rate": 4.028516298321643e-05,
      "loss": 0.3707,
      "num_tokens": 769046890.0,
      "step": 1214
    },
    {
      "epoch": 0.1436679673643136,
      "grad_norm": 0.20507104694843292,
      "learning_rate": 4.025525082345917e-05,
      "loss": 0.3778,
      "num_tokens": 769671163.0,
      "step": 1215
    },
    {
      "epoch": 0.14378621260494265,
      "grad_norm": 0.18357984721660614,
      "learning_rate": 4.022532906287306e-05,
      "loss": 0.3759,
      "num_tokens": 770306885.0,
      "step": 1216
    },
    {
      "epoch": 0.1439044578455717,
      "grad_norm": 0.17793887853622437,
      "learning_rate": 4.0195397741053374e-05,
      "loss": 0.3404,
      "num_tokens": 770935952.0,
      "step": 1217
    },
    {
      "epoch": 0.14402270308620077,
      "grad_norm": 0.1842331439256668,
      "learning_rate": 4.016545689760804e-05,
      "loss": 0.3636,
      "num_tokens": 771566252.0,
      "step": 1218
    },
    {
      "epoch": 0.14414094832682983,
      "grad_norm": 0.17892083525657654,
      "learning_rate": 4.013550657215759e-05,
      "loss": 0.3593,
      "num_tokens": 772196019.0,
      "step": 1219
    },
    {
      "epoch": 0.14425919356745892,
      "grad_norm": 0.1608760952949524,
      "learning_rate": 4.01055468043351e-05,
      "loss": 0.3764,
      "num_tokens": 772829557.0,
      "step": 1220
    },
    {
      "epoch": 0.14437743880808798,
      "grad_norm": 0.17508582770824432,
      "learning_rate": 4.007557763378616e-05,
      "loss": 0.3865,
      "num_tokens": 773462306.0,
      "step": 1221
    },
    {
      "epoch": 0.14449568404871704,
      "grad_norm": 0.17223335802555084,
      "learning_rate": 4.0045599100168794e-05,
      "loss": 0.418,
      "num_tokens": 774094333.0,
      "step": 1222
    },
    {
      "epoch": 0.1446139292893461,
      "grad_norm": 0.18835633993148804,
      "learning_rate": 4.001561124315339e-05,
      "loss": 0.3892,
      "num_tokens": 774727506.0,
      "step": 1223
    },
    {
      "epoch": 0.14473217452997517,
      "grad_norm": 0.19669027626514435,
      "learning_rate": 3.998561410242271e-05,
      "loss": 0.4244,
      "num_tokens": 775365455.0,
      "step": 1224
    },
    {
      "epoch": 0.14485041977060423,
      "grad_norm": 0.18262512981891632,
      "learning_rate": 3.99556077176718e-05,
      "loss": 0.4127,
      "num_tokens": 775989745.0,
      "step": 1225
    },
    {
      "epoch": 0.1449686650112333,
      "grad_norm": 0.18306951224803925,
      "learning_rate": 3.992559212860789e-05,
      "loss": 0.3604,
      "num_tokens": 776627802.0,
      "step": 1226
    },
    {
      "epoch": 0.14508691025186235,
      "grad_norm": 0.17740747332572937,
      "learning_rate": 3.989556737495045e-05,
      "loss": 0.3525,
      "num_tokens": 777261077.0,
      "step": 1227
    },
    {
      "epoch": 0.14520515549249144,
      "grad_norm": 0.21090878546237946,
      "learning_rate": 3.9865533496431045e-05,
      "loss": 0.3955,
      "num_tokens": 777900544.0,
      "step": 1228
    },
    {
      "epoch": 0.1453234007331205,
      "grad_norm": 0.17277488112449646,
      "learning_rate": 3.983549053279333e-05,
      "loss": 0.3326,
      "num_tokens": 778526351.0,
      "step": 1229
    },
    {
      "epoch": 0.14544164597374956,
      "grad_norm": 0.18703323602676392,
      "learning_rate": 3.980543852379295e-05,
      "loss": 0.3794,
      "num_tokens": 779161010.0,
      "step": 1230
    },
    {
      "epoch": 0.14555989121437862,
      "grad_norm": 0.1652577966451645,
      "learning_rate": 3.977537750919757e-05,
      "loss": 0.3797,
      "num_tokens": 779795294.0,
      "step": 1231
    },
    {
      "epoch": 0.14567813645500768,
      "grad_norm": 0.17143456637859344,
      "learning_rate": 3.974530752878674e-05,
      "loss": 0.3601,
      "num_tokens": 780434525.0,
      "step": 1232
    },
    {
      "epoch": 0.14579638169563675,
      "grad_norm": 0.16284136474132538,
      "learning_rate": 3.971522862235187e-05,
      "loss": 0.3492,
      "num_tokens": 781047733.0,
      "step": 1233
    },
    {
      "epoch": 0.1459146269362658,
      "grad_norm": 0.1673222929239273,
      "learning_rate": 3.968514082969621e-05,
      "loss": 0.3657,
      "num_tokens": 781680882.0,
      "step": 1234
    },
    {
      "epoch": 0.14603287217689487,
      "grad_norm": 0.17836733162403107,
      "learning_rate": 3.9655044190634725e-05,
      "loss": 0.3779,
      "num_tokens": 782316078.0,
      "step": 1235
    },
    {
      "epoch": 0.14615111741752396,
      "grad_norm": 0.19530552625656128,
      "learning_rate": 3.9624938744994145e-05,
      "loss": 0.4156,
      "num_tokens": 782945690.0,
      "step": 1236
    },
    {
      "epoch": 0.14626936265815302,
      "grad_norm": 0.1764390617609024,
      "learning_rate": 3.9594824532612794e-05,
      "loss": 0.4112,
      "num_tokens": 783578548.0,
      "step": 1237
    },
    {
      "epoch": 0.14638760789878208,
      "grad_norm": 0.17299224436283112,
      "learning_rate": 3.956470159334064e-05,
      "loss": 0.3551,
      "num_tokens": 784211272.0,
      "step": 1238
    },
    {
      "epoch": 0.14650585313941114,
      "grad_norm": 0.17603452503681183,
      "learning_rate": 3.953456996703918e-05,
      "loss": 0.3959,
      "num_tokens": 784844206.0,
      "step": 1239
    },
    {
      "epoch": 0.1466240983800402,
      "grad_norm": 0.17413505911827087,
      "learning_rate": 3.95044296935814e-05,
      "loss": 0.3834,
      "num_tokens": 785479541.0,
      "step": 1240
    },
    {
      "epoch": 0.14674234362066926,
      "grad_norm": 0.17274941504001617,
      "learning_rate": 3.9474280812851754e-05,
      "loss": 0.3539,
      "num_tokens": 786105562.0,
      "step": 1241
    },
    {
      "epoch": 0.14686058886129832,
      "grad_norm": 0.18288247287273407,
      "learning_rate": 3.944412336474607e-05,
      "loss": 0.3926,
      "num_tokens": 786739257.0,
      "step": 1242
    },
    {
      "epoch": 0.14697883410192739,
      "grad_norm": 0.17496144771575928,
      "learning_rate": 3.941395738917149e-05,
      "loss": 0.3441,
      "num_tokens": 787367527.0,
      "step": 1243
    },
    {
      "epoch": 0.14709707934255647,
      "grad_norm": 0.1842147409915924,
      "learning_rate": 3.9383782926046496e-05,
      "loss": 0.3555,
      "num_tokens": 787999760.0,
      "step": 1244
    },
    {
      "epoch": 0.14721532458318554,
      "grad_norm": 0.1850278675556183,
      "learning_rate": 3.935360001530075e-05,
      "loss": 0.356,
      "num_tokens": 788637623.0,
      "step": 1245
    },
    {
      "epoch": 0.1473335698238146,
      "grad_norm": 0.16519823670387268,
      "learning_rate": 3.932340869687512e-05,
      "loss": 0.3481,
      "num_tokens": 789276023.0,
      "step": 1246
    },
    {
      "epoch": 0.14745181506444366,
      "grad_norm": 0.17701783776283264,
      "learning_rate": 3.929320901072159e-05,
      "loss": 0.3793,
      "num_tokens": 789907579.0,
      "step": 1247
    },
    {
      "epoch": 0.14757006030507272,
      "grad_norm": 0.18543802201747894,
      "learning_rate": 3.9263000996803226e-05,
      "loss": 0.4052,
      "num_tokens": 790542282.0,
      "step": 1248
    },
    {
      "epoch": 0.14768830554570178,
      "grad_norm": 0.1734195053577423,
      "learning_rate": 3.92327846950941e-05,
      "loss": 0.385,
      "num_tokens": 791177851.0,
      "step": 1249
    },
    {
      "epoch": 0.14780655078633084,
      "grad_norm": 0.16981084644794464,
      "learning_rate": 3.9202560145579255e-05,
      "loss": 0.3653,
      "num_tokens": 791809529.0,
      "step": 1250
    },
    {
      "epoch": 0.1479247960269599,
      "grad_norm": 0.180636465549469,
      "learning_rate": 3.917232738825466e-05,
      "loss": 0.4003,
      "num_tokens": 792448822.0,
      "step": 1251
    },
    {
      "epoch": 0.148043041267589,
      "grad_norm": 0.16907177865505219,
      "learning_rate": 3.914208646312713e-05,
      "loss": 0.3903,
      "num_tokens": 793083686.0,
      "step": 1252
    },
    {
      "epoch": 0.14816128650821805,
      "grad_norm": 0.1896291971206665,
      "learning_rate": 3.911183741021431e-05,
      "loss": 0.4406,
      "num_tokens": 793720893.0,
      "step": 1253
    },
    {
      "epoch": 0.14827953174884712,
      "grad_norm": 0.1729743331670761,
      "learning_rate": 3.908158026954457e-05,
      "loss": 0.3581,
      "num_tokens": 794350771.0,
      "step": 1254
    },
    {
      "epoch": 0.14839777698947618,
      "grad_norm": 0.17992742359638214,
      "learning_rate": 3.9051315081156996e-05,
      "loss": 0.3826,
      "num_tokens": 794988418.0,
      "step": 1255
    },
    {
      "epoch": 0.14851602223010524,
      "grad_norm": 0.18747080862522125,
      "learning_rate": 3.902104188510134e-05,
      "loss": 0.4175,
      "num_tokens": 795619860.0,
      "step": 1256
    },
    {
      "epoch": 0.1486342674707343,
      "grad_norm": 0.18046019971370697,
      "learning_rate": 3.899076072143793e-05,
      "loss": 0.3723,
      "num_tokens": 796258236.0,
      "step": 1257
    },
    {
      "epoch": 0.14875251271136336,
      "grad_norm": 0.17821986973285675,
      "learning_rate": 3.896047163023764e-05,
      "loss": 0.3618,
      "num_tokens": 796889178.0,
      "step": 1258
    },
    {
      "epoch": 0.14887075795199242,
      "grad_norm": 0.18142741918563843,
      "learning_rate": 3.893017465158184e-05,
      "loss": 0.3865,
      "num_tokens": 797526666.0,
      "step": 1259
    },
    {
      "epoch": 0.1489890031926215,
      "grad_norm": 0.16956116259098053,
      "learning_rate": 3.889986982556235e-05,
      "loss": 0.3713,
      "num_tokens": 798162731.0,
      "step": 1260
    },
    {
      "epoch": 0.14910724843325057,
      "grad_norm": 0.1800818145275116,
      "learning_rate": 3.886955719228133e-05,
      "loss": 0.3861,
      "num_tokens": 798797285.0,
      "step": 1261
    },
    {
      "epoch": 0.14922549367387963,
      "grad_norm": 0.1661878526210785,
      "learning_rate": 3.8839236791851344e-05,
      "loss": 0.3797,
      "num_tokens": 799430881.0,
      "step": 1262
    },
    {
      "epoch": 0.1493437389145087,
      "grad_norm": 0.15878503024578094,
      "learning_rate": 3.880890866439515e-05,
      "loss": 0.3368,
      "num_tokens": 800065809.0,
      "step": 1263
    },
    {
      "epoch": 0.14946198415513776,
      "grad_norm": 0.1672976016998291,
      "learning_rate": 3.8778572850045794e-05,
      "loss": 0.3263,
      "num_tokens": 800695505.0,
      "step": 1264
    },
    {
      "epoch": 0.14958022939576682,
      "grad_norm": 0.1818859577178955,
      "learning_rate": 3.874822938894647e-05,
      "loss": 0.3847,
      "num_tokens": 801324547.0,
      "step": 1265
    },
    {
      "epoch": 0.14969847463639588,
      "grad_norm": 0.17939628660678864,
      "learning_rate": 3.871787832125049e-05,
      "loss": 0.3743,
      "num_tokens": 801954766.0,
      "step": 1266
    },
    {
      "epoch": 0.14981671987702494,
      "grad_norm": 0.1639385223388672,
      "learning_rate": 3.868751968712125e-05,
      "loss": 0.3375,
      "num_tokens": 802585735.0,
      "step": 1267
    },
    {
      "epoch": 0.149934965117654,
      "grad_norm": 0.19370990991592407,
      "learning_rate": 3.8657153526732125e-05,
      "loss": 0.4178,
      "num_tokens": 803222978.0,
      "step": 1268
    },
    {
      "epoch": 0.1500532103582831,
      "grad_norm": 0.15046223998069763,
      "learning_rate": 3.862677988026647e-05,
      "loss": 0.2976,
      "num_tokens": 803861621.0,
      "step": 1269
    },
    {
      "epoch": 0.15017145559891215,
      "grad_norm": 0.15617400407791138,
      "learning_rate": 3.859639878791756e-05,
      "loss": 0.2841,
      "num_tokens": 804493467.0,
      "step": 1270
    },
    {
      "epoch": 0.1502897008395412,
      "grad_norm": 0.19728296995162964,
      "learning_rate": 3.856601028988849e-05,
      "loss": 0.366,
      "num_tokens": 805126588.0,
      "step": 1271
    },
    {
      "epoch": 0.15040794608017027,
      "grad_norm": 0.16635596752166748,
      "learning_rate": 3.85356144263922e-05,
      "loss": 0.3633,
      "num_tokens": 805765213.0,
      "step": 1272
    },
    {
      "epoch": 0.15052619132079934,
      "grad_norm": 0.1875937432050705,
      "learning_rate": 3.850521123765132e-05,
      "loss": 0.3738,
      "num_tokens": 806404016.0,
      "step": 1273
    },
    {
      "epoch": 0.1506444365614284,
      "grad_norm": 0.15865705907344818,
      "learning_rate": 3.847480076389821e-05,
      "loss": 0.3508,
      "num_tokens": 807036599.0,
      "step": 1274
    },
    {
      "epoch": 0.15076268180205746,
      "grad_norm": 0.16535243391990662,
      "learning_rate": 3.844438304537487e-05,
      "loss": 0.3677,
      "num_tokens": 807672182.0,
      "step": 1275
    },
    {
      "epoch": 0.15088092704268652,
      "grad_norm": 0.18247942626476288,
      "learning_rate": 3.8413958122332876e-05,
      "loss": 0.3947,
      "num_tokens": 808281458.0,
      "step": 1276
    },
    {
      "epoch": 0.1509991722833156,
      "grad_norm": 0.1740826964378357,
      "learning_rate": 3.838352603503335e-05,
      "loss": 0.3841,
      "num_tokens": 808881586.0,
      "step": 1277
    },
    {
      "epoch": 0.15111741752394467,
      "grad_norm": 0.18411533534526825,
      "learning_rate": 3.835308682374687e-05,
      "loss": 0.3988,
      "num_tokens": 809481267.0,
      "step": 1278
    },
    {
      "epoch": 0.15123566276457373,
      "grad_norm": 0.17475385963916779,
      "learning_rate": 3.8322640528753474e-05,
      "loss": 0.3473,
      "num_tokens": 810113774.0,
      "step": 1279
    },
    {
      "epoch": 0.1513539080052028,
      "grad_norm": 0.1791556179523468,
      "learning_rate": 3.829218719034255e-05,
      "loss": 0.3772,
      "num_tokens": 810745803.0,
      "step": 1280
    },
    {
      "epoch": 0.15147215324583185,
      "grad_norm": 0.17454758286476135,
      "learning_rate": 3.826172684881281e-05,
      "loss": 0.4019,
      "num_tokens": 811385263.0,
      "step": 1281
    },
    {
      "epoch": 0.15159039848646091,
      "grad_norm": 0.1617545187473297,
      "learning_rate": 3.8231259544472245e-05,
      "loss": 0.3424,
      "num_tokens": 812016051.0,
      "step": 1282
    },
    {
      "epoch": 0.15170864372708998,
      "grad_norm": 0.16357214748859406,
      "learning_rate": 3.820078531763805e-05,
      "loss": 0.3235,
      "num_tokens": 812649108.0,
      "step": 1283
    },
    {
      "epoch": 0.15182688896771904,
      "grad_norm": 0.1793568879365921,
      "learning_rate": 3.817030420863658e-05,
      "loss": 0.3588,
      "num_tokens": 813283409.0,
      "step": 1284
    },
    {
      "epoch": 0.15194513420834813,
      "grad_norm": 0.17174500226974487,
      "learning_rate": 3.81398162578033e-05,
      "loss": 0.3685,
      "num_tokens": 813917493.0,
      "step": 1285
    },
    {
      "epoch": 0.1520633794489772,
      "grad_norm": 0.185197114944458,
      "learning_rate": 3.810932150548272e-05,
      "loss": 0.3773,
      "num_tokens": 814556038.0,
      "step": 1286
    },
    {
      "epoch": 0.15218162468960625,
      "grad_norm": 0.16052372753620148,
      "learning_rate": 3.807881999202837e-05,
      "loss": 0.366,
      "num_tokens": 815190976.0,
      "step": 1287
    },
    {
      "epoch": 0.1522998699302353,
      "grad_norm": 0.1711684763431549,
      "learning_rate": 3.8048311757802705e-05,
      "loss": 0.3666,
      "num_tokens": 815830227.0,
      "step": 1288
    },
    {
      "epoch": 0.15241811517086437,
      "grad_norm": 0.17269568145275116,
      "learning_rate": 3.80177968431771e-05,
      "loss": 0.3716,
      "num_tokens": 816464763.0,
      "step": 1289
    },
    {
      "epoch": 0.15253636041149343,
      "grad_norm": 0.17598651349544525,
      "learning_rate": 3.7987275288531744e-05,
      "loss": 0.3783,
      "num_tokens": 817072689.0,
      "step": 1290
    },
    {
      "epoch": 0.1526546056521225,
      "grad_norm": 0.17497462034225464,
      "learning_rate": 3.795674713425562e-05,
      "loss": 0.3813,
      "num_tokens": 817703995.0,
      "step": 1291
    },
    {
      "epoch": 0.15277285089275155,
      "grad_norm": 0.15532535314559937,
      "learning_rate": 3.7926212420746455e-05,
      "loss": 0.3322,
      "num_tokens": 818306204.0,
      "step": 1292
    },
    {
      "epoch": 0.15289109613338064,
      "grad_norm": 0.18098512291908264,
      "learning_rate": 3.789567118841063e-05,
      "loss": 0.3948,
      "num_tokens": 818944794.0,
      "step": 1293
    },
    {
      "epoch": 0.1530093413740097,
      "grad_norm": 0.18815064430236816,
      "learning_rate": 3.78651234776632e-05,
      "loss": 0.4214,
      "num_tokens": 819580090.0,
      "step": 1294
    },
    {
      "epoch": 0.15312758661463877,
      "grad_norm": 0.1842847764492035,
      "learning_rate": 3.783456932892774e-05,
      "loss": 0.3878,
      "num_tokens": 820195511.0,
      "step": 1295
    },
    {
      "epoch": 0.15324583185526783,
      "grad_norm": 0.16440126299858093,
      "learning_rate": 3.7804008782636374e-05,
      "loss": 0.3707,
      "num_tokens": 820832912.0,
      "step": 1296
    },
    {
      "epoch": 0.1533640770958969,
      "grad_norm": 0.17478272318840027,
      "learning_rate": 3.7773441879229685e-05,
      "loss": 0.3795,
      "num_tokens": 821463804.0,
      "step": 1297
    },
    {
      "epoch": 0.15348232233652595,
      "grad_norm": 0.18590036034584045,
      "learning_rate": 3.774286865915667e-05,
      "loss": 0.376,
      "num_tokens": 822096771.0,
      "step": 1298
    },
    {
      "epoch": 0.153600567577155,
      "grad_norm": 0.2140692174434662,
      "learning_rate": 3.771228916287468e-05,
      "loss": 0.3896,
      "num_tokens": 822724976.0,
      "step": 1299
    },
    {
      "epoch": 0.15371881281778407,
      "grad_norm": 0.16716571152210236,
      "learning_rate": 3.768170343084937e-05,
      "loss": 0.3616,
      "num_tokens": 823360078.0,
      "step": 1300
    },
    {
      "epoch": 0.15383705805841316,
      "grad_norm": 0.16711951792240143,
      "learning_rate": 3.7651111503554676e-05,
      "loss": 0.3574,
      "num_tokens": 823995903.0,
      "step": 1301
    },
    {
      "epoch": 0.15395530329904222,
      "grad_norm": 0.2010648250579834,
      "learning_rate": 3.762051342147268e-05,
      "loss": 0.4097,
      "num_tokens": 824634564.0,
      "step": 1302
    },
    {
      "epoch": 0.15407354853967128,
      "grad_norm": 0.20303165912628174,
      "learning_rate": 3.758990922509364e-05,
      "loss": 0.4327,
      "num_tokens": 825267057.0,
      "step": 1303
    },
    {
      "epoch": 0.15419179378030035,
      "grad_norm": 0.17012476921081543,
      "learning_rate": 3.7559298954915914e-05,
      "loss": 0.3591,
      "num_tokens": 825899088.0,
      "step": 1304
    },
    {
      "epoch": 0.1543100390209294,
      "grad_norm": 0.18373215198516846,
      "learning_rate": 3.752868265144587e-05,
      "loss": 0.3915,
      "num_tokens": 826535290.0,
      "step": 1305
    },
    {
      "epoch": 0.15442828426155847,
      "grad_norm": 0.17862318456172943,
      "learning_rate": 3.7498060355197896e-05,
      "loss": 0.373,
      "num_tokens": 827173746.0,
      "step": 1306
    },
    {
      "epoch": 0.15454652950218753,
      "grad_norm": 0.1615932434797287,
      "learning_rate": 3.7467432106694276e-05,
      "loss": 0.3663,
      "num_tokens": 827804731.0,
      "step": 1307
    },
    {
      "epoch": 0.1546647747428166,
      "grad_norm": 0.16810201108455658,
      "learning_rate": 3.743679794646517e-05,
      "loss": 0.369,
      "num_tokens": 828436832.0,
      "step": 1308
    },
    {
      "epoch": 0.15478301998344568,
      "grad_norm": 0.16492970287799835,
      "learning_rate": 3.740615791504859e-05,
      "loss": 0.3439,
      "num_tokens": 829065711.0,
      "step": 1309
    },
    {
      "epoch": 0.15490126522407474,
      "grad_norm": 0.17141704261302948,
      "learning_rate": 3.737551205299029e-05,
      "loss": 0.3655,
      "num_tokens": 829699491.0,
      "step": 1310
    },
    {
      "epoch": 0.1550195104647038,
      "grad_norm": 0.1736629605293274,
      "learning_rate": 3.734486040084377e-05,
      "loss": 0.3821,
      "num_tokens": 830301332.0,
      "step": 1311
    },
    {
      "epoch": 0.15513775570533286,
      "grad_norm": 0.17586372792720795,
      "learning_rate": 3.731420299917015e-05,
      "loss": 0.3464,
      "num_tokens": 830932269.0,
      "step": 1312
    },
    {
      "epoch": 0.15525600094596193,
      "grad_norm": 0.18535244464874268,
      "learning_rate": 3.72835398885382e-05,
      "loss": 0.3894,
      "num_tokens": 831571426.0,
      "step": 1313
    },
    {
      "epoch": 0.155374246186591,
      "grad_norm": 0.1786143183708191,
      "learning_rate": 3.7252871109524224e-05,
      "loss": 0.3856,
      "num_tokens": 832211161.0,
      "step": 1314
    },
    {
      "epoch": 0.15549249142722005,
      "grad_norm": 0.19510836899280548,
      "learning_rate": 3.722219670271201e-05,
      "loss": 0.4069,
      "num_tokens": 832846569.0,
      "step": 1315
    },
    {
      "epoch": 0.1556107366678491,
      "grad_norm": 0.16250097751617432,
      "learning_rate": 3.719151670869284e-05,
      "loss": 0.3705,
      "num_tokens": 833477855.0,
      "step": 1316
    },
    {
      "epoch": 0.15572898190847817,
      "grad_norm": 0.18421269953250885,
      "learning_rate": 3.716083116806533e-05,
      "loss": 0.3765,
      "num_tokens": 834112533.0,
      "step": 1317
    },
    {
      "epoch": 0.15584722714910726,
      "grad_norm": 0.15677453577518463,
      "learning_rate": 3.713014012143551e-05,
      "loss": 0.3367,
      "num_tokens": 834750383.0,
      "step": 1318
    },
    {
      "epoch": 0.15596547238973632,
      "grad_norm": 0.1714835911989212,
      "learning_rate": 3.709944360941662e-05,
      "loss": 0.3865,
      "num_tokens": 835382487.0,
      "step": 1319
    },
    {
      "epoch": 0.15608371763036538,
      "grad_norm": 0.16007648408412933,
      "learning_rate": 3.706874167262916e-05,
      "loss": 0.3555,
      "num_tokens": 836011005.0,
      "step": 1320
    },
    {
      "epoch": 0.15620196287099444,
      "grad_norm": 0.16921104490756989,
      "learning_rate": 3.703803435170084e-05,
      "loss": 0.3547,
      "num_tokens": 836645534.0,
      "step": 1321
    },
    {
      "epoch": 0.1563202081116235,
      "grad_norm": 0.1705523282289505,
      "learning_rate": 3.700732168726645e-05,
      "loss": 0.3333,
      "num_tokens": 837281263.0,
      "step": 1322
    },
    {
      "epoch": 0.15643845335225257,
      "grad_norm": 0.1723070591688156,
      "learning_rate": 3.69766037199679e-05,
      "loss": 0.3559,
      "num_tokens": 837911899.0,
      "step": 1323
    },
    {
      "epoch": 0.15655669859288163,
      "grad_norm": 0.20030608773231506,
      "learning_rate": 3.6945880490454045e-05,
      "loss": 0.4059,
      "num_tokens": 838549137.0,
      "step": 1324
    },
    {
      "epoch": 0.1566749438335107,
      "grad_norm": 0.19786883890628815,
      "learning_rate": 3.691515203938077e-05,
      "loss": 0.3872,
      "num_tokens": 839185355.0,
      "step": 1325
    },
    {
      "epoch": 0.15679318907413978,
      "grad_norm": 0.18107512593269348,
      "learning_rate": 3.6884418407410845e-05,
      "loss": 0.3701,
      "num_tokens": 839817125.0,
      "step": 1326
    },
    {
      "epoch": 0.15691143431476884,
      "grad_norm": 0.1745939999818802,
      "learning_rate": 3.68536796352139e-05,
      "loss": 0.3598,
      "num_tokens": 840448393.0,
      "step": 1327
    },
    {
      "epoch": 0.1570296795553979,
      "grad_norm": 0.1749391406774521,
      "learning_rate": 3.6822935763466356e-05,
      "loss": 0.3511,
      "num_tokens": 841086731.0,
      "step": 1328
    },
    {
      "epoch": 0.15714792479602696,
      "grad_norm": 0.19824792444705963,
      "learning_rate": 3.6792186832851385e-05,
      "loss": 0.3976,
      "num_tokens": 841723258.0,
      "step": 1329
    },
    {
      "epoch": 0.15726617003665602,
      "grad_norm": 0.16669194400310516,
      "learning_rate": 3.676143288405887e-05,
      "loss": 0.3501,
      "num_tokens": 842354306.0,
      "step": 1330
    },
    {
      "epoch": 0.15738441527728508,
      "grad_norm": 0.16550509631633759,
      "learning_rate": 3.6730673957785314e-05,
      "loss": 0.3631,
      "num_tokens": 842990332.0,
      "step": 1331
    },
    {
      "epoch": 0.15750266051791414,
      "grad_norm": 0.17728877067565918,
      "learning_rate": 3.6699910094733816e-05,
      "loss": 0.3809,
      "num_tokens": 843624261.0,
      "step": 1332
    },
    {
      "epoch": 0.1576209057585432,
      "grad_norm": 0.1832035630941391,
      "learning_rate": 3.666914133561401e-05,
      "loss": 0.3751,
      "num_tokens": 844262144.0,
      "step": 1333
    },
    {
      "epoch": 0.1577391509991723,
      "grad_norm": 0.17735359072685242,
      "learning_rate": 3.6638367721142006e-05,
      "loss": 0.3715,
      "num_tokens": 844897584.0,
      "step": 1334
    },
    {
      "epoch": 0.15785739623980136,
      "grad_norm": 0.17729835212230682,
      "learning_rate": 3.660758929204034e-05,
      "loss": 0.3829,
      "num_tokens": 845532833.0,
      "step": 1335
    },
    {
      "epoch": 0.15797564148043042,
      "grad_norm": 0.18032696843147278,
      "learning_rate": 3.6576806089037916e-05,
      "loss": 0.3598,
      "num_tokens": 846132555.0,
      "step": 1336
    },
    {
      "epoch": 0.15809388672105948,
      "grad_norm": 0.19697603583335876,
      "learning_rate": 3.654601815286995e-05,
      "loss": 0.4304,
      "num_tokens": 846771073.0,
      "step": 1337
    },
    {
      "epoch": 0.15821213196168854,
      "grad_norm": 0.16080917418003082,
      "learning_rate": 3.6515225524277955e-05,
      "loss": 0.3532,
      "num_tokens": 847407047.0,
      "step": 1338
    },
    {
      "epoch": 0.1583303772023176,
      "grad_norm": 0.17054249346256256,
      "learning_rate": 3.648442824400959e-05,
      "loss": 0.3906,
      "num_tokens": 848038762.0,
      "step": 1339
    },
    {
      "epoch": 0.15844862244294666,
      "grad_norm": 0.17448587715625763,
      "learning_rate": 3.6453626352818744e-05,
      "loss": 0.3581,
      "num_tokens": 848671922.0,
      "step": 1340
    },
    {
      "epoch": 0.15856686768357572,
      "grad_norm": 0.18276441097259521,
      "learning_rate": 3.642281989146534e-05,
      "loss": 0.3741,
      "num_tokens": 849309141.0,
      "step": 1341
    },
    {
      "epoch": 0.1586851129242048,
      "grad_norm": 0.16970501840114594,
      "learning_rate": 3.63920089007154e-05,
      "loss": 0.3815,
      "num_tokens": 849947357.0,
      "step": 1342
    },
    {
      "epoch": 0.15880335816483387,
      "grad_norm": 0.18194201588630676,
      "learning_rate": 3.63611934213409e-05,
      "loss": 0.3632,
      "num_tokens": 850586919.0,
      "step": 1343
    },
    {
      "epoch": 0.15892160340546294,
      "grad_norm": 0.20223063230514526,
      "learning_rate": 3.633037349411978e-05,
      "loss": 0.4044,
      "num_tokens": 851164524.0,
      "step": 1344
    },
    {
      "epoch": 0.159039848646092,
      "grad_norm": 0.1658789962530136,
      "learning_rate": 3.629954915983588e-05,
      "loss": 0.3732,
      "num_tokens": 851795535.0,
      "step": 1345
    },
    {
      "epoch": 0.15915809388672106,
      "grad_norm": 0.16001170873641968,
      "learning_rate": 3.626872045927883e-05,
      "loss": 0.3657,
      "num_tokens": 852429267.0,
      "step": 1346
    },
    {
      "epoch": 0.15927633912735012,
      "grad_norm": 0.1767129898071289,
      "learning_rate": 3.6237887433244076e-05,
      "loss": 0.4073,
      "num_tokens": 853066581.0,
      "step": 1347
    },
    {
      "epoch": 0.15939458436797918,
      "grad_norm": 0.16027915477752686,
      "learning_rate": 3.6207050122532766e-05,
      "loss": 0.3671,
      "num_tokens": 853705006.0,
      "step": 1348
    },
    {
      "epoch": 0.15951282960860824,
      "grad_norm": 0.16998380422592163,
      "learning_rate": 3.617620856795171e-05,
      "loss": 0.3815,
      "num_tokens": 854335569.0,
      "step": 1349
    },
    {
      "epoch": 0.15963107484923733,
      "grad_norm": 0.1529499590396881,
      "learning_rate": 3.614536281031337e-05,
      "loss": 0.3546,
      "num_tokens": 854974579.0,
      "step": 1350
    },
    {
      "epoch": 0.1597493200898664,
      "grad_norm": 0.16457729041576385,
      "learning_rate": 3.611451289043573e-05,
      "loss": 0.3878,
      "num_tokens": 855612244.0,
      "step": 1351
    },
    {
      "epoch": 0.15986756533049545,
      "grad_norm": 0.17140960693359375,
      "learning_rate": 3.608365884914232e-05,
      "loss": 0.3636,
      "num_tokens": 856244164.0,
      "step": 1352
    },
    {
      "epoch": 0.15998581057112451,
      "grad_norm": 0.17074401676654816,
      "learning_rate": 3.605280072726208e-05,
      "loss": 0.376,
      "num_tokens": 856879791.0,
      "step": 1353
    },
    {
      "epoch": 0.16010405581175358,
      "grad_norm": 0.17245988547801971,
      "learning_rate": 3.602193856562937e-05,
      "loss": 0.3687,
      "num_tokens": 857513354.0,
      "step": 1354
    },
    {
      "epoch": 0.16022230105238264,
      "grad_norm": 0.15877239406108856,
      "learning_rate": 3.599107240508394e-05,
      "loss": 0.3487,
      "num_tokens": 858148669.0,
      "step": 1355
    },
    {
      "epoch": 0.1603405462930117,
      "grad_norm": 0.1687905192375183,
      "learning_rate": 3.596020228647074e-05,
      "loss": 0.3516,
      "num_tokens": 858784311.0,
      "step": 1356
    },
    {
      "epoch": 0.16045879153364076,
      "grad_norm": 0.15725862979888916,
      "learning_rate": 3.592932825064004e-05,
      "loss": 0.3544,
      "num_tokens": 859417782.0,
      "step": 1357
    },
    {
      "epoch": 0.16057703677426985,
      "grad_norm": 0.16414190828800201,
      "learning_rate": 3.5898450338447266e-05,
      "loss": 0.3568,
      "num_tokens": 860048747.0,
      "step": 1358
    },
    {
      "epoch": 0.1606952820148989,
      "grad_norm": 0.17501533031463623,
      "learning_rate": 3.586756859075294e-05,
      "loss": 0.3958,
      "num_tokens": 860684168.0,
      "step": 1359
    },
    {
      "epoch": 0.16081352725552797,
      "grad_norm": 0.1815037876367569,
      "learning_rate": 3.5836683048422714e-05,
      "loss": 0.3914,
      "num_tokens": 861317334.0,
      "step": 1360
    },
    {
      "epoch": 0.16093177249615703,
      "grad_norm": 0.1688801646232605,
      "learning_rate": 3.5805793752327226e-05,
      "loss": 0.3954,
      "num_tokens": 861948731.0,
      "step": 1361
    },
    {
      "epoch": 0.1610500177367861,
      "grad_norm": 0.17557959258556366,
      "learning_rate": 3.5774900743342096e-05,
      "loss": 0.4118,
      "num_tokens": 862586570.0,
      "step": 1362
    },
    {
      "epoch": 0.16116826297741516,
      "grad_norm": 0.16620570421218872,
      "learning_rate": 3.5744004062347845e-05,
      "loss": 0.3704,
      "num_tokens": 863222204.0,
      "step": 1363
    },
    {
      "epoch": 0.16128650821804422,
      "grad_norm": 0.17702950537204742,
      "learning_rate": 3.571310375022988e-05,
      "loss": 0.3715,
      "num_tokens": 863857243.0,
      "step": 1364
    },
    {
      "epoch": 0.16140475345867328,
      "grad_norm": 0.14711260795593262,
      "learning_rate": 3.568219984787838e-05,
      "loss": 0.3481,
      "num_tokens": 864495982.0,
      "step": 1365
    },
    {
      "epoch": 0.16152299869930234,
      "grad_norm": 0.1549108326435089,
      "learning_rate": 3.5651292396188284e-05,
      "loss": 0.3172,
      "num_tokens": 865131852.0,
      "step": 1366
    },
    {
      "epoch": 0.16164124393993143,
      "grad_norm": 0.15956881642341614,
      "learning_rate": 3.5620381436059246e-05,
      "loss": 0.3455,
      "num_tokens": 865767401.0,
      "step": 1367
    },
    {
      "epoch": 0.1617594891805605,
      "grad_norm": 0.17599697411060333,
      "learning_rate": 3.5589467008395545e-05,
      "loss": 0.3688,
      "num_tokens": 866405290.0,
      "step": 1368
    },
    {
      "epoch": 0.16187773442118955,
      "grad_norm": 0.17518922686576843,
      "learning_rate": 3.555854915410605e-05,
      "loss": 0.3865,
      "num_tokens": 867039158.0,
      "step": 1369
    },
    {
      "epoch": 0.1619959796618186,
      "grad_norm": 0.1700814664363861,
      "learning_rate": 3.5527627914104176e-05,
      "loss": 0.367,
      "num_tokens": 867673395.0,
      "step": 1370
    },
    {
      "epoch": 0.16211422490244767,
      "grad_norm": 0.17121228575706482,
      "learning_rate": 3.54967033293078e-05,
      "loss": 0.3901,
      "num_tokens": 868310685.0,
      "step": 1371
    },
    {
      "epoch": 0.16223247014307673,
      "grad_norm": 0.17767947912216187,
      "learning_rate": 3.546577544063924e-05,
      "loss": 0.3855,
      "num_tokens": 868923042.0,
      "step": 1372
    },
    {
      "epoch": 0.1623507153837058,
      "grad_norm": 0.1763809621334076,
      "learning_rate": 3.5434844289025174e-05,
      "loss": 0.4122,
      "num_tokens": 869556905.0,
      "step": 1373
    },
    {
      "epoch": 0.16246896062433486,
      "grad_norm": 0.15335842967033386,
      "learning_rate": 3.5403909915396624e-05,
      "loss": 0.321,
      "num_tokens": 870190559.0,
      "step": 1374
    },
    {
      "epoch": 0.16258720586496395,
      "grad_norm": 0.16744902729988098,
      "learning_rate": 3.537297236068884e-05,
      "loss": 0.353,
      "num_tokens": 870829637.0,
      "step": 1375
    },
    {
      "epoch": 0.162705451105593,
      "grad_norm": 0.1898547112941742,
      "learning_rate": 3.53420316658413e-05,
      "loss": 0.3477,
      "num_tokens": 871466630.0,
      "step": 1376
    },
    {
      "epoch": 0.16282369634622207,
      "grad_norm": 0.18806128203868866,
      "learning_rate": 3.531108787179764e-05,
      "loss": 0.3784,
      "num_tokens": 872077498.0,
      "step": 1377
    },
    {
      "epoch": 0.16294194158685113,
      "grad_norm": 0.18817409873008728,
      "learning_rate": 3.5280141019505576e-05,
      "loss": 0.3864,
      "num_tokens": 872706532.0,
      "step": 1378
    },
    {
      "epoch": 0.1630601868274802,
      "grad_norm": 0.17816320061683655,
      "learning_rate": 3.524919114991693e-05,
      "loss": 0.3553,
      "num_tokens": 873345301.0,
      "step": 1379
    },
    {
      "epoch": 0.16317843206810925,
      "grad_norm": 0.15472441911697388,
      "learning_rate": 3.521823830398743e-05,
      "loss": 0.3527,
      "num_tokens": 873977689.0,
      "step": 1380
    },
    {
      "epoch": 0.16329667730873831,
      "grad_norm": 0.17606398463249207,
      "learning_rate": 3.518728252267681e-05,
      "loss": 0.3672,
      "num_tokens": 874615260.0,
      "step": 1381
    },
    {
      "epoch": 0.16341492254936738,
      "grad_norm": 0.15786398947238922,
      "learning_rate": 3.515632384694865e-05,
      "loss": 0.3754,
      "num_tokens": 875248450.0,
      "step": 1382
    },
    {
      "epoch": 0.16353316778999646,
      "grad_norm": 0.1698189079761505,
      "learning_rate": 3.512536231777039e-05,
      "loss": 0.4117,
      "num_tokens": 875858896.0,
      "step": 1383
    },
    {
      "epoch": 0.16365141303062553,
      "grad_norm": 0.17061932384967804,
      "learning_rate": 3.5094397976113226e-05,
      "loss": 0.349,
      "num_tokens": 876495972.0,
      "step": 1384
    },
    {
      "epoch": 0.1637696582712546,
      "grad_norm": 0.1739518940448761,
      "learning_rate": 3.5063430862952073e-05,
      "loss": 0.3964,
      "num_tokens": 877126376.0,
      "step": 1385
    },
    {
      "epoch": 0.16388790351188365,
      "grad_norm": 0.16720589995384216,
      "learning_rate": 3.503246101926554e-05,
      "loss": 0.3339,
      "num_tokens": 877761789.0,
      "step": 1386
    },
    {
      "epoch": 0.1640061487525127,
      "grad_norm": 0.18069598078727722,
      "learning_rate": 3.500148848603582e-05,
      "loss": 0.3926,
      "num_tokens": 878356527.0,
      "step": 1387
    },
    {
      "epoch": 0.16412439399314177,
      "grad_norm": 0.1658611297607422,
      "learning_rate": 3.4970513304248674e-05,
      "loss": 0.3745,
      "num_tokens": 878990451.0,
      "step": 1388
    },
    {
      "epoch": 0.16424263923377083,
      "grad_norm": 0.1776137799024582,
      "learning_rate": 3.493953551489337e-05,
      "loss": 0.3313,
      "num_tokens": 879620939.0,
      "step": 1389
    },
    {
      "epoch": 0.1643608844743999,
      "grad_norm": 0.1723213642835617,
      "learning_rate": 3.490855515896263e-05,
      "loss": 0.3715,
      "num_tokens": 880257254.0,
      "step": 1390
    },
    {
      "epoch": 0.16447912971502898,
      "grad_norm": 0.19617415964603424,
      "learning_rate": 3.487757227745257e-05,
      "loss": 0.4065,
      "num_tokens": 880896268.0,
      "step": 1391
    },
    {
      "epoch": 0.16459737495565804,
      "grad_norm": 0.19237631559371948,
      "learning_rate": 3.484658691136265e-05,
      "loss": 0.3585,
      "num_tokens": 881531804.0,
      "step": 1392
    },
    {
      "epoch": 0.1647156201962871,
      "grad_norm": 0.208879753947258,
      "learning_rate": 3.48155991016956e-05,
      "loss": 0.3575,
      "num_tokens": 882135234.0,
      "step": 1393
    },
    {
      "epoch": 0.16483386543691617,
      "grad_norm": 0.17721176147460938,
      "learning_rate": 3.4784608889457406e-05,
      "loss": 0.3502,
      "num_tokens": 882730906.0,
      "step": 1394
    },
    {
      "epoch": 0.16495211067754523,
      "grad_norm": 0.18262936174869537,
      "learning_rate": 3.475361631565723e-05,
      "loss": 0.3943,
      "num_tokens": 883362306.0,
      "step": 1395
    },
    {
      "epoch": 0.1650703559181743,
      "grad_norm": 0.18020063638687134,
      "learning_rate": 3.472262142130736e-05,
      "loss": 0.3545,
      "num_tokens": 884001035.0,
      "step": 1396
    },
    {
      "epoch": 0.16518860115880335,
      "grad_norm": 0.18471196293830872,
      "learning_rate": 3.4691624247423126e-05,
      "loss": 0.3795,
      "num_tokens": 884635015.0,
      "step": 1397
    },
    {
      "epoch": 0.1653068463994324,
      "grad_norm": 0.17687906324863434,
      "learning_rate": 3.466062483502293e-05,
      "loss": 0.3879,
      "num_tokens": 885272613.0,
      "step": 1398
    },
    {
      "epoch": 0.1654250916400615,
      "grad_norm": 0.17874866724014282,
      "learning_rate": 3.4629623225128084e-05,
      "loss": 0.3235,
      "num_tokens": 885905643.0,
      "step": 1399
    },
    {
      "epoch": 0.16554333688069056,
      "grad_norm": 0.1700168401002884,
      "learning_rate": 3.459861945876282e-05,
      "loss": 0.3607,
      "num_tokens": 886536895.0,
      "step": 1400
    },
    {
      "epoch": 0.16566158212131962,
      "grad_norm": 0.16640955209732056,
      "learning_rate": 3.456761357695425e-05,
      "loss": 0.3412,
      "num_tokens": 887173320.0,
      "step": 1401
    },
    {
      "epoch": 0.16577982736194868,
      "grad_norm": 0.17980210483074188,
      "learning_rate": 3.453660562073225e-05,
      "loss": 0.3777,
      "num_tokens": 887807056.0,
      "step": 1402
    },
    {
      "epoch": 0.16589807260257775,
      "grad_norm": 0.19972193241119385,
      "learning_rate": 3.4505595631129484e-05,
      "loss": 0.386,
      "num_tokens": 888446558.0,
      "step": 1403
    },
    {
      "epoch": 0.1660163178432068,
      "grad_norm": 0.17348936200141907,
      "learning_rate": 3.4474583649181254e-05,
      "loss": 0.3606,
      "num_tokens": 889079741.0,
      "step": 1404
    },
    {
      "epoch": 0.16613456308383587,
      "grad_norm": 0.17330406606197357,
      "learning_rate": 3.4443569715925534e-05,
      "loss": 0.4002,
      "num_tokens": 889708375.0,
      "step": 1405
    },
    {
      "epoch": 0.16625280832446493,
      "grad_norm": 0.16449137032032013,
      "learning_rate": 3.441255387240288e-05,
      "loss": 0.3451,
      "num_tokens": 890343205.0,
      "step": 1406
    },
    {
      "epoch": 0.16637105356509402,
      "grad_norm": 0.1590339094400406,
      "learning_rate": 3.4381536159656354e-05,
      "loss": 0.3473,
      "num_tokens": 890980377.0,
      "step": 1407
    },
    {
      "epoch": 0.16648929880572308,
      "grad_norm": 0.16445298492908478,
      "learning_rate": 3.435051661873152e-05,
      "loss": 0.346,
      "num_tokens": 891610939.0,
      "step": 1408
    },
    {
      "epoch": 0.16660754404635214,
      "grad_norm": 0.18500487506389618,
      "learning_rate": 3.431949529067633e-05,
      "loss": 0.3941,
      "num_tokens": 892248590.0,
      "step": 1409
    },
    {
      "epoch": 0.1667257892869812,
      "grad_norm": 0.19658434391021729,
      "learning_rate": 3.4288472216541126e-05,
      "loss": 0.4552,
      "num_tokens": 892884534.0,
      "step": 1410
    },
    {
      "epoch": 0.16684403452761026,
      "grad_norm": 0.1724197417497635,
      "learning_rate": 3.4257447437378544e-05,
      "loss": 0.3398,
      "num_tokens": 893522439.0,
      "step": 1411
    },
    {
      "epoch": 0.16696227976823932,
      "grad_norm": 0.17510159313678741,
      "learning_rate": 3.42264209942435e-05,
      "loss": 0.3507,
      "num_tokens": 894159327.0,
      "step": 1412
    },
    {
      "epoch": 0.16708052500886839,
      "grad_norm": 0.20202413201332092,
      "learning_rate": 3.419539292819308e-05,
      "loss": 0.3959,
      "num_tokens": 894796406.0,
      "step": 1413
    },
    {
      "epoch": 0.16719877024949745,
      "grad_norm": 0.16532990336418152,
      "learning_rate": 3.416436328028654e-05,
      "loss": 0.3679,
      "num_tokens": 895433169.0,
      "step": 1414
    },
    {
      "epoch": 0.16731701549012654,
      "grad_norm": 0.16116783022880554,
      "learning_rate": 3.413333209158522e-05,
      "loss": 0.3715,
      "num_tokens": 896065522.0,
      "step": 1415
    },
    {
      "epoch": 0.1674352607307556,
      "grad_norm": 0.1760171353816986,
      "learning_rate": 3.410229940315251e-05,
      "loss": 0.3474,
      "num_tokens": 896697566.0,
      "step": 1416
    },
    {
      "epoch": 0.16755350597138466,
      "grad_norm": 0.17760524153709412,
      "learning_rate": 3.4071265256053755e-05,
      "loss": 0.3651,
      "num_tokens": 897331944.0,
      "step": 1417
    },
    {
      "epoch": 0.16767175121201372,
      "grad_norm": 0.1891304850578308,
      "learning_rate": 3.404022969135628e-05,
      "loss": 0.4184,
      "num_tokens": 897965210.0,
      "step": 1418
    },
    {
      "epoch": 0.16778999645264278,
      "grad_norm": 0.1605445295572281,
      "learning_rate": 3.400919275012923e-05,
      "loss": 0.3771,
      "num_tokens": 898598679.0,
      "step": 1419
    },
    {
      "epoch": 0.16790824169327184,
      "grad_norm": 0.18207436800003052,
      "learning_rate": 3.397815447344361e-05,
      "loss": 0.4081,
      "num_tokens": 899230065.0,
      "step": 1420
    },
    {
      "epoch": 0.1680264869339009,
      "grad_norm": 0.1534513235092163,
      "learning_rate": 3.3947114902372195e-05,
      "loss": 0.3266,
      "num_tokens": 899866736.0,
      "step": 1421
    },
    {
      "epoch": 0.16814473217452997,
      "grad_norm": 0.17347653210163116,
      "learning_rate": 3.391607407798943e-05,
      "loss": 0.347,
      "num_tokens": 900503362.0,
      "step": 1422
    },
    {
      "epoch": 0.16826297741515903,
      "grad_norm": 0.157567098736763,
      "learning_rate": 3.3885032041371474e-05,
      "loss": 0.3707,
      "num_tokens": 901142459.0,
      "step": 1423
    },
    {
      "epoch": 0.16838122265578812,
      "grad_norm": 0.1753988415002823,
      "learning_rate": 3.385398883359605e-05,
      "loss": 0.3704,
      "num_tokens": 901768662.0,
      "step": 1424
    },
    {
      "epoch": 0.16849946789641718,
      "grad_norm": 0.16486535966396332,
      "learning_rate": 3.3822944495742436e-05,
      "loss": 0.3642,
      "num_tokens": 902398246.0,
      "step": 1425
    },
    {
      "epoch": 0.16861771313704624,
      "grad_norm": 0.16971829533576965,
      "learning_rate": 3.379189906889144e-05,
      "loss": 0.3555,
      "num_tokens": 903032227.0,
      "step": 1426
    },
    {
      "epoch": 0.1687359583776753,
      "grad_norm": 0.1704730987548828,
      "learning_rate": 3.376085259412525e-05,
      "loss": 0.3856,
      "num_tokens": 903666346.0,
      "step": 1427
    },
    {
      "epoch": 0.16885420361830436,
      "grad_norm": 0.1611000895500183,
      "learning_rate": 3.372980511252752e-05,
      "loss": 0.3937,
      "num_tokens": 904303678.0,
      "step": 1428
    },
    {
      "epoch": 0.16897244885893342,
      "grad_norm": 0.18555743992328644,
      "learning_rate": 3.369875666518314e-05,
      "loss": 0.3608,
      "num_tokens": 904930363.0,
      "step": 1429
    },
    {
      "epoch": 0.16909069409956248,
      "grad_norm": 0.16819705069065094,
      "learning_rate": 3.366770729317837e-05,
      "loss": 0.4225,
      "num_tokens": 905564490.0,
      "step": 1430
    },
    {
      "epoch": 0.16920893934019154,
      "grad_norm": 0.17546449601650238,
      "learning_rate": 3.363665703760063e-05,
      "loss": 0.3882,
      "num_tokens": 906200069.0,
      "step": 1431
    },
    {
      "epoch": 0.16932718458082063,
      "grad_norm": 0.16625650227069855,
      "learning_rate": 3.3605605939538556e-05,
      "loss": 0.3445,
      "num_tokens": 906837954.0,
      "step": 1432
    },
    {
      "epoch": 0.1694454298214497,
      "grad_norm": 0.17123083770275116,
      "learning_rate": 3.357455404008186e-05,
      "loss": 0.3372,
      "num_tokens": 907469285.0,
      "step": 1433
    },
    {
      "epoch": 0.16956367506207876,
      "grad_norm": 0.1995430290699005,
      "learning_rate": 3.3543501380321315e-05,
      "loss": 0.4258,
      "num_tokens": 908105346.0,
      "step": 1434
    },
    {
      "epoch": 0.16968192030270782,
      "grad_norm": 0.16351737082004547,
      "learning_rate": 3.351244800134875e-05,
      "loss": 0.3553,
      "num_tokens": 908736680.0,
      "step": 1435
    },
    {
      "epoch": 0.16980016554333688,
      "grad_norm": 0.16494734585285187,
      "learning_rate": 3.348139394425687e-05,
      "loss": 0.3672,
      "num_tokens": 909376079.0,
      "step": 1436
    },
    {
      "epoch": 0.16991841078396594,
      "grad_norm": 0.17809386551380157,
      "learning_rate": 3.3450339250139353e-05,
      "loss": 0.3834,
      "num_tokens": 910014511.0,
      "step": 1437
    },
    {
      "epoch": 0.170036656024595,
      "grad_norm": 0.17304128408432007,
      "learning_rate": 3.341928396009067e-05,
      "loss": 0.3539,
      "num_tokens": 910650479.0,
      "step": 1438
    },
    {
      "epoch": 0.17015490126522406,
      "grad_norm": 0.1687421351671219,
      "learning_rate": 3.338822811520609e-05,
      "loss": 0.3586,
      "num_tokens": 911289548.0,
      "step": 1439
    },
    {
      "epoch": 0.17027314650585315,
      "grad_norm": 0.17560400068759918,
      "learning_rate": 3.335717175658164e-05,
      "loss": 0.3764,
      "num_tokens": 911920115.0,
      "step": 1440
    },
    {
      "epoch": 0.1703913917464822,
      "grad_norm": 0.18659047782421112,
      "learning_rate": 3.3326114925313995e-05,
      "loss": 0.3793,
      "num_tokens": 912552910.0,
      "step": 1441
    },
    {
      "epoch": 0.17050963698711127,
      "grad_norm": 0.15252403914928436,
      "learning_rate": 3.329505766250047e-05,
      "loss": 0.356,
      "num_tokens": 913186992.0,
      "step": 1442
    },
    {
      "epoch": 0.17062788222774034,
      "grad_norm": 0.1647413671016693,
      "learning_rate": 3.326400000923896e-05,
      "loss": 0.3644,
      "num_tokens": 913822413.0,
      "step": 1443
    },
    {
      "epoch": 0.1707461274683694,
      "grad_norm": 0.15846136212348938,
      "learning_rate": 3.323294200662785e-05,
      "loss": 0.3376,
      "num_tokens": 914458604.0,
      "step": 1444
    },
    {
      "epoch": 0.17086437270899846,
      "grad_norm": 0.16936224699020386,
      "learning_rate": 3.3201883695766014e-05,
      "loss": 0.3769,
      "num_tokens": 915097439.0,
      "step": 1445
    },
    {
      "epoch": 0.17098261794962752,
      "grad_norm": 0.17426526546478271,
      "learning_rate": 3.317082511775272e-05,
      "loss": 0.3862,
      "num_tokens": 915732434.0,
      "step": 1446
    },
    {
      "epoch": 0.17110086319025658,
      "grad_norm": 0.18202632665634155,
      "learning_rate": 3.313976631368759e-05,
      "loss": 0.3781,
      "num_tokens": 916361639.0,
      "step": 1447
    },
    {
      "epoch": 0.17121910843088567,
      "grad_norm": 0.18634334206581116,
      "learning_rate": 3.310870732467057e-05,
      "loss": 0.3626,
      "num_tokens": 916961319.0,
      "step": 1448
    },
    {
      "epoch": 0.17133735367151473,
      "grad_norm": 0.160066619515419,
      "learning_rate": 3.3077648191801817e-05,
      "loss": 0.3788,
      "num_tokens": 917600686.0,
      "step": 1449
    },
    {
      "epoch": 0.1714555989121438,
      "grad_norm": 0.18605881929397583,
      "learning_rate": 3.304658895618168e-05,
      "loss": 0.3704,
      "num_tokens": 918239869.0,
      "step": 1450
    },
    {
      "epoch": 0.17157384415277285,
      "grad_norm": 0.1764983981847763,
      "learning_rate": 3.3015529658910665e-05,
      "loss": 0.3624,
      "num_tokens": 918877611.0,
      "step": 1451
    },
    {
      "epoch": 0.17169208939340191,
      "grad_norm": 0.1722479909658432,
      "learning_rate": 3.2984470341089346e-05,
      "loss": 0.3914,
      "num_tokens": 919517243.0,
      "step": 1452
    },
    {
      "epoch": 0.17181033463403098,
      "grad_norm": 0.17516613006591797,
      "learning_rate": 3.295341104381833e-05,
      "loss": 0.2958,
      "num_tokens": 920155279.0,
      "step": 1453
    },
    {
      "epoch": 0.17192857987466004,
      "grad_norm": 0.18306639790534973,
      "learning_rate": 3.2922351808198195e-05,
      "loss": 0.3519,
      "num_tokens": 920791529.0,
      "step": 1454
    },
    {
      "epoch": 0.1720468251152891,
      "grad_norm": 0.18455427885055542,
      "learning_rate": 3.2891292675329434e-05,
      "loss": 0.4022,
      "num_tokens": 921421972.0,
      "step": 1455
    },
    {
      "epoch": 0.1721650703559182,
      "grad_norm": 0.17631007730960846,
      "learning_rate": 3.286023368631241e-05,
      "loss": 0.395,
      "num_tokens": 922059966.0,
      "step": 1456
    },
    {
      "epoch": 0.17228331559654725,
      "grad_norm": 0.17840716242790222,
      "learning_rate": 3.2829174882247295e-05,
      "loss": 0.3539,
      "num_tokens": 922694070.0,
      "step": 1457
    },
    {
      "epoch": 0.1724015608371763,
      "grad_norm": 0.1761758029460907,
      "learning_rate": 3.2798116304234e-05,
      "loss": 0.3708,
      "num_tokens": 923326220.0,
      "step": 1458
    },
    {
      "epoch": 0.17251980607780537,
      "grad_norm": 0.20283976197242737,
      "learning_rate": 3.276705799337217e-05,
      "loss": 0.3979,
      "num_tokens": 923963438.0,
      "step": 1459
    },
    {
      "epoch": 0.17263805131843443,
      "grad_norm": 0.18546903133392334,
      "learning_rate": 3.273599999076105e-05,
      "loss": 0.3961,
      "num_tokens": 924593547.0,
      "step": 1460
    },
    {
      "epoch": 0.1727562965590635,
      "grad_norm": 0.16381438076496124,
      "learning_rate": 3.270494233749953e-05,
      "loss": 0.3438,
      "num_tokens": 925228798.0,
      "step": 1461
    },
    {
      "epoch": 0.17287454179969255,
      "grad_norm": 0.18618075549602509,
      "learning_rate": 3.267388507468602e-05,
      "loss": 0.3847,
      "num_tokens": 925861981.0,
      "step": 1462
    },
    {
      "epoch": 0.17299278704032162,
      "grad_norm": 0.20563967525959015,
      "learning_rate": 3.2642828243418365e-05,
      "loss": 0.3768,
      "num_tokens": 926497587.0,
      "step": 1463
    },
    {
      "epoch": 0.1731110322809507,
      "grad_norm": 0.1740288883447647,
      "learning_rate": 3.261177188479391e-05,
      "loss": 0.3562,
      "num_tokens": 927132750.0,
      "step": 1464
    },
    {
      "epoch": 0.17322927752157977,
      "grad_norm": 0.1666458398103714,
      "learning_rate": 3.2580716039909335e-05,
      "loss": 0.3594,
      "num_tokens": 927764160.0,
      "step": 1465
    },
    {
      "epoch": 0.17334752276220883,
      "grad_norm": 0.166953906416893,
      "learning_rate": 3.254966074986065e-05,
      "loss": 0.3439,
      "num_tokens": 928403130.0,
      "step": 1466
    },
    {
      "epoch": 0.1734657680028379,
      "grad_norm": 0.18671217560768127,
      "learning_rate": 3.251860605574314e-05,
      "loss": 0.3665,
      "num_tokens": 929035983.0,
      "step": 1467
    },
    {
      "epoch": 0.17358401324346695,
      "grad_norm": 0.18403495848178864,
      "learning_rate": 3.248755199865126e-05,
      "loss": 0.3999,
      "num_tokens": 929673737.0,
      "step": 1468
    },
    {
      "epoch": 0.173702258484096,
      "grad_norm": 0.17461302876472473,
      "learning_rate": 3.245649861967869e-05,
      "loss": 0.3818,
      "num_tokens": 930303540.0,
      "step": 1469
    },
    {
      "epoch": 0.17382050372472507,
      "grad_norm": 0.2092389613389969,
      "learning_rate": 3.242544595991815e-05,
      "loss": 0.4015,
      "num_tokens": 930934690.0,
      "step": 1470
    },
    {
      "epoch": 0.17393874896535413,
      "grad_norm": 0.17646804451942444,
      "learning_rate": 3.239439406046145e-05,
      "loss": 0.3822,
      "num_tokens": 931570813.0,
      "step": 1471
    },
    {
      "epoch": 0.1740569942059832,
      "grad_norm": 0.16729430854320526,
      "learning_rate": 3.236334296239937e-05,
      "loss": 0.3666,
      "num_tokens": 932203713.0,
      "step": 1472
    },
    {
      "epoch": 0.17417523944661228,
      "grad_norm": 0.16469994187355042,
      "learning_rate": 3.233229270682163e-05,
      "loss": 0.3798,
      "num_tokens": 932842100.0,
      "step": 1473
    },
    {
      "epoch": 0.17429348468724135,
      "grad_norm": 0.16082432866096497,
      "learning_rate": 3.230124333481686e-05,
      "loss": 0.347,
      "num_tokens": 933475258.0,
      "step": 1474
    },
    {
      "epoch": 0.1744117299278704,
      "grad_norm": 0.16905388236045837,
      "learning_rate": 3.227019488747249e-05,
      "loss": 0.3569,
      "num_tokens": 934112965.0,
      "step": 1475
    },
    {
      "epoch": 0.17452997516849947,
      "grad_norm": 0.20290949940681458,
      "learning_rate": 3.2239147405874745e-05,
      "loss": 0.3906,
      "num_tokens": 934737655.0,
      "step": 1476
    },
    {
      "epoch": 0.17464822040912853,
      "grad_norm": 0.17375296354293823,
      "learning_rate": 3.220810093110857e-05,
      "loss": 0.3608,
      "num_tokens": 935377274.0,
      "step": 1477
    },
    {
      "epoch": 0.1747664656497576,
      "grad_norm": 0.17994755506515503,
      "learning_rate": 3.217705550425756e-05,
      "loss": 0.3519,
      "num_tokens": 936011871.0,
      "step": 1478
    },
    {
      "epoch": 0.17488471089038665,
      "grad_norm": 0.1969660371541977,
      "learning_rate": 3.214601116640396e-05,
      "loss": 0.4092,
      "num_tokens": 936647042.0,
      "step": 1479
    },
    {
      "epoch": 0.1750029561310157,
      "grad_norm": 0.1726323664188385,
      "learning_rate": 3.211496795862853e-05,
      "loss": 0.3596,
      "num_tokens": 937283368.0,
      "step": 1480
    },
    {
      "epoch": 0.1751212013716448,
      "grad_norm": 0.16877536475658417,
      "learning_rate": 3.208392592201058e-05,
      "loss": 0.3688,
      "num_tokens": 937919785.0,
      "step": 1481
    },
    {
      "epoch": 0.17523944661227386,
      "grad_norm": 0.17268769443035126,
      "learning_rate": 3.205288509762781e-05,
      "loss": 0.36,
      "num_tokens": 938554025.0,
      "step": 1482
    },
    {
      "epoch": 0.17535769185290292,
      "grad_norm": 0.17848555743694305,
      "learning_rate": 3.2021845526556394e-05,
      "loss": 0.3572,
      "num_tokens": 939191403.0,
      "step": 1483
    },
    {
      "epoch": 0.175475937093532,
      "grad_norm": 0.18568621575832367,
      "learning_rate": 3.199080724987078e-05,
      "loss": 0.3988,
      "num_tokens": 939827159.0,
      "step": 1484
    },
    {
      "epoch": 0.17559418233416105,
      "grad_norm": 0.1646295189857483,
      "learning_rate": 3.195977030864373e-05,
      "loss": 0.3475,
      "num_tokens": 940465371.0,
      "step": 1485
    },
    {
      "epoch": 0.1757124275747901,
      "grad_norm": 0.17712347209453583,
      "learning_rate": 3.1928734743946243e-05,
      "loss": 0.4087,
      "num_tokens": 941102114.0,
      "step": 1486
    },
    {
      "epoch": 0.17583067281541917,
      "grad_norm": 0.1864871382713318,
      "learning_rate": 3.189770059684749e-05,
      "loss": 0.3831,
      "num_tokens": 941738142.0,
      "step": 1487
    },
    {
      "epoch": 0.17594891805604823,
      "grad_norm": 0.16088663041591644,
      "learning_rate": 3.186666790841478e-05,
      "loss": 0.3303,
      "num_tokens": 942366943.0,
      "step": 1488
    },
    {
      "epoch": 0.17606716329667732,
      "grad_norm": 0.15922009944915771,
      "learning_rate": 3.183563671971347e-05,
      "loss": 0.3795,
      "num_tokens": 942995207.0,
      "step": 1489
    },
    {
      "epoch": 0.17618540853730638,
      "grad_norm": 0.1968304067850113,
      "learning_rate": 3.1804607071806916e-05,
      "loss": 0.3883,
      "num_tokens": 943631334.0,
      "step": 1490
    },
    {
      "epoch": 0.17630365377793544,
      "grad_norm": 0.17453576624393463,
      "learning_rate": 3.17735790057565e-05,
      "loss": 0.3661,
      "num_tokens": 944264751.0,
      "step": 1491
    },
    {
      "epoch": 0.1764218990185645,
      "grad_norm": 0.17541645467281342,
      "learning_rate": 3.1742552562621454e-05,
      "loss": 0.3984,
      "num_tokens": 944901410.0,
      "step": 1492
    },
    {
      "epoch": 0.17654014425919357,
      "grad_norm": 0.16635823249816895,
      "learning_rate": 3.171152778345888e-05,
      "loss": 0.3735,
      "num_tokens": 945538374.0,
      "step": 1493
    },
    {
      "epoch": 0.17665838949982263,
      "grad_norm": 0.1661323755979538,
      "learning_rate": 3.168050470932368e-05,
      "loss": 0.353,
      "num_tokens": 946170416.0,
      "step": 1494
    },
    {
      "epoch": 0.1767766347404517,
      "grad_norm": 0.17597319185733795,
      "learning_rate": 3.164948338126849e-05,
      "loss": 0.3577,
      "num_tokens": 946805320.0,
      "step": 1495
    },
    {
      "epoch": 0.17689487998108075,
      "grad_norm": 0.1651628017425537,
      "learning_rate": 3.161846384034366e-05,
      "loss": 0.4059,
      "num_tokens": 947437269.0,
      "step": 1496
    },
    {
      "epoch": 0.17701312522170984,
      "grad_norm": 0.18678541481494904,
      "learning_rate": 3.1587446127597126e-05,
      "loss": 0.3983,
      "num_tokens": 948074252.0,
      "step": 1497
    },
    {
      "epoch": 0.1771313704623389,
      "grad_norm": 0.15683230757713318,
      "learning_rate": 3.155643028407447e-05,
      "loss": 0.3214,
      "num_tokens": 948701457.0,
      "step": 1498
    },
    {
      "epoch": 0.17724961570296796,
      "grad_norm": 0.17081496119499207,
      "learning_rate": 3.152541635081875e-05,
      "loss": 0.3665,
      "num_tokens": 949332037.0,
      "step": 1499
    },
    {
      "epoch": 0.17736786094359702,
      "grad_norm": 0.17161160707473755,
      "learning_rate": 3.149440436887053e-05,
      "loss": 0.3598,
      "num_tokens": 949967678.0,
      "step": 1500
    },
    {
      "epoch": 0.17748610618422608,
      "grad_norm": 0.15939413011074066,
      "learning_rate": 3.1463394379267755e-05,
      "loss": 0.3541,
      "num_tokens": 950600709.0,
      "step": 1501
    },
    {
      "epoch": 0.17760435142485514,
      "grad_norm": 0.17879928648471832,
      "learning_rate": 3.1432386423045756e-05,
      "loss": 0.3888,
      "num_tokens": 951227404.0,
      "step": 1502
    },
    {
      "epoch": 0.1777225966654842,
      "grad_norm": 0.16773827373981476,
      "learning_rate": 3.1401380541237186e-05,
      "loss": 0.3721,
      "num_tokens": 951866625.0,
      "step": 1503
    },
    {
      "epoch": 0.17784084190611327,
      "grad_norm": 0.17299412190914154,
      "learning_rate": 3.137037677487193e-05,
      "loss": 0.3596,
      "num_tokens": 952502095.0,
      "step": 1504
    },
    {
      "epoch": 0.17795908714674236,
      "grad_norm": 0.16810157895088196,
      "learning_rate": 3.133937516497708e-05,
      "loss": 0.3773,
      "num_tokens": 953134327.0,
      "step": 1505
    },
    {
      "epoch": 0.17807733238737142,
      "grad_norm": 0.15980418026447296,
      "learning_rate": 3.130837575257688e-05,
      "loss": 0.356,
      "num_tokens": 953773701.0,
      "step": 1506
    },
    {
      "epoch": 0.17819557762800048,
      "grad_norm": 0.1758408099412918,
      "learning_rate": 3.1277378578692645e-05,
      "loss": 0.3844,
      "num_tokens": 954406325.0,
      "step": 1507
    },
    {
      "epoch": 0.17831382286862954,
      "grad_norm": 0.1675221174955368,
      "learning_rate": 3.124638368434277e-05,
      "loss": 0.3644,
      "num_tokens": 955041716.0,
      "step": 1508
    },
    {
      "epoch": 0.1784320681092586,
      "grad_norm": 0.18556548655033112,
      "learning_rate": 3.121539111054259e-05,
      "loss": 0.3729,
      "num_tokens": 955678383.0,
      "step": 1509
    },
    {
      "epoch": 0.17855031334988766,
      "grad_norm": 0.17029190063476562,
      "learning_rate": 3.118440089830441e-05,
      "loss": 0.3887,
      "num_tokens": 956315561.0,
      "step": 1510
    },
    {
      "epoch": 0.17866855859051672,
      "grad_norm": 0.17651323974132538,
      "learning_rate": 3.115341308863736e-05,
      "loss": 0.3898,
      "num_tokens": 956944551.0,
      "step": 1511
    },
    {
      "epoch": 0.17878680383114579,
      "grad_norm": 0.1759771704673767,
      "learning_rate": 3.1122427722547435e-05,
      "loss": 0.3802,
      "num_tokens": 957583170.0,
      "step": 1512
    },
    {
      "epoch": 0.17890504907177487,
      "grad_norm": 0.15872859954833984,
      "learning_rate": 3.109144484103738e-05,
      "loss": 0.3358,
      "num_tokens": 958218955.0,
      "step": 1513
    },
    {
      "epoch": 0.17902329431240394,
      "grad_norm": 0.17943841218948364,
      "learning_rate": 3.106046448510663e-05,
      "loss": 0.3519,
      "num_tokens": 958852938.0,
      "step": 1514
    },
    {
      "epoch": 0.179141539553033,
      "grad_norm": 0.17151592671871185,
      "learning_rate": 3.102948669575133e-05,
      "loss": 0.364,
      "num_tokens": 959490199.0,
      "step": 1515
    },
    {
      "epoch": 0.17925978479366206,
      "grad_norm": 0.18046090006828308,
      "learning_rate": 3.099851151396419e-05,
      "loss": 0.3769,
      "num_tokens": 960127283.0,
      "step": 1516
    },
    {
      "epoch": 0.17937803003429112,
      "grad_norm": 0.16845285892486572,
      "learning_rate": 3.096753898073447e-05,
      "loss": 0.3736,
      "num_tokens": 960756987.0,
      "step": 1517
    },
    {
      "epoch": 0.17949627527492018,
      "grad_norm": 0.16035690903663635,
      "learning_rate": 3.093656913704793e-05,
      "loss": 0.38,
      "num_tokens": 961395831.0,
      "step": 1518
    },
    {
      "epoch": 0.17961452051554924,
      "grad_norm": 0.16901935636997223,
      "learning_rate": 3.0905602023886786e-05,
      "loss": 0.363,
      "num_tokens": 962031388.0,
      "step": 1519
    },
    {
      "epoch": 0.1797327657561783,
      "grad_norm": 0.1605140119791031,
      "learning_rate": 3.087463768222962e-05,
      "loss": 0.3202,
      "num_tokens": 962666836.0,
      "step": 1520
    },
    {
      "epoch": 0.1798510109968074,
      "grad_norm": 0.17046019434928894,
      "learning_rate": 3.084367615305136e-05,
      "loss": 0.3469,
      "num_tokens": 963305624.0,
      "step": 1521
    },
    {
      "epoch": 0.17996925623743645,
      "grad_norm": 0.1570424735546112,
      "learning_rate": 3.08127174773232e-05,
      "loss": 0.3616,
      "num_tokens": 963941024.0,
      "step": 1522
    },
    {
      "epoch": 0.18008750147806551,
      "grad_norm": 0.17264924943447113,
      "learning_rate": 3.078176169601259e-05,
      "loss": 0.3659,
      "num_tokens": 964574609.0,
      "step": 1523
    },
    {
      "epoch": 0.18020574671869458,
      "grad_norm": 0.21252690255641937,
      "learning_rate": 3.075080885008308e-05,
      "loss": 0.4115,
      "num_tokens": 965214035.0,
      "step": 1524
    },
    {
      "epoch": 0.18032399195932364,
      "grad_norm": 0.15354739129543304,
      "learning_rate": 3.071985898049442e-05,
      "loss": 0.3541,
      "num_tokens": 965848776.0,
      "step": 1525
    },
    {
      "epoch": 0.1804422371999527,
      "grad_norm": 0.1739215850830078,
      "learning_rate": 3.068891212820237e-05,
      "loss": 0.3533,
      "num_tokens": 966478626.0,
      "step": 1526
    },
    {
      "epoch": 0.18056048244058176,
      "grad_norm": 0.19520846009254456,
      "learning_rate": 3.0657968334158715e-05,
      "loss": 0.3982,
      "num_tokens": 967079379.0,
      "step": 1527
    },
    {
      "epoch": 0.18067872768121082,
      "grad_norm": 0.16799034178256989,
      "learning_rate": 3.0627027639311165e-05,
      "loss": 0.3487,
      "num_tokens": 967715213.0,
      "step": 1528
    },
    {
      "epoch": 0.18079697292183988,
      "grad_norm": 0.19211645424365997,
      "learning_rate": 3.059609008460339e-05,
      "loss": 0.377,
      "num_tokens": 968326010.0,
      "step": 1529
    },
    {
      "epoch": 0.18091521816246897,
      "grad_norm": 0.17046518623828888,
      "learning_rate": 3.0565155710974824e-05,
      "loss": 0.3554,
      "num_tokens": 968964068.0,
      "step": 1530
    },
    {
      "epoch": 0.18103346340309803,
      "grad_norm": 0.167215034365654,
      "learning_rate": 3.053422455936077e-05,
      "loss": 0.3903,
      "num_tokens": 969598545.0,
      "step": 1531
    },
    {
      "epoch": 0.1811517086437271,
      "grad_norm": 0.18514011800289154,
      "learning_rate": 3.050329667069221e-05,
      "loss": 0.3952,
      "num_tokens": 970231579.0,
      "step": 1532
    },
    {
      "epoch": 0.18126995388435616,
      "grad_norm": 0.18290914595127106,
      "learning_rate": 3.047237208589583e-05,
      "loss": 0.3563,
      "num_tokens": 970868020.0,
      "step": 1533
    },
    {
      "epoch": 0.18138819912498522,
      "grad_norm": 0.18019051849842072,
      "learning_rate": 3.044145084589395e-05,
      "loss": 0.3801,
      "num_tokens": 971501577.0,
      "step": 1534
    },
    {
      "epoch": 0.18150644436561428,
      "grad_norm": 0.18275585770606995,
      "learning_rate": 3.041053299160446e-05,
      "loss": 0.4036,
      "num_tokens": 972136461.0,
      "step": 1535
    },
    {
      "epoch": 0.18162468960624334,
      "grad_norm": 0.16142059862613678,
      "learning_rate": 3.0379618563940756e-05,
      "loss": 0.3384,
      "num_tokens": 972774594.0,
      "step": 1536
    },
    {
      "epoch": 0.1817429348468724,
      "grad_norm": 0.20104844868183136,
      "learning_rate": 3.034870760381172e-05,
      "loss": 0.3699,
      "num_tokens": 973410075.0,
      "step": 1537
    },
    {
      "epoch": 0.1818611800875015,
      "grad_norm": 0.18426941335201263,
      "learning_rate": 3.0317800152121625e-05,
      "loss": 0.3835,
      "num_tokens": 974049163.0,
      "step": 1538
    },
    {
      "epoch": 0.18197942532813055,
      "grad_norm": 0.1625235378742218,
      "learning_rate": 3.0286896249770123e-05,
      "loss": 0.3486,
      "num_tokens": 974681502.0,
      "step": 1539
    },
    {
      "epoch": 0.1820976705687596,
      "grad_norm": 0.19007031619548798,
      "learning_rate": 3.0255995937652153e-05,
      "loss": 0.392,
      "num_tokens": 975312361.0,
      "step": 1540
    },
    {
      "epoch": 0.18221591580938867,
      "grad_norm": 0.1634375900030136,
      "learning_rate": 3.022509925665791e-05,
      "loss": 0.3726,
      "num_tokens": 975944571.0,
      "step": 1541
    },
    {
      "epoch": 0.18233416105001773,
      "grad_norm": 0.17391811311244965,
      "learning_rate": 3.0194206247672786e-05,
      "loss": 0.3431,
      "num_tokens": 976578990.0,
      "step": 1542
    },
    {
      "epoch": 0.1824524062906468,
      "grad_norm": 0.17183931171894073,
      "learning_rate": 3.0163316951577294e-05,
      "loss": 0.3434,
      "num_tokens": 977213113.0,
      "step": 1543
    },
    {
      "epoch": 0.18257065153127586,
      "grad_norm": 0.18079477548599243,
      "learning_rate": 3.0132431409247073e-05,
      "loss": 0.3712,
      "num_tokens": 977846686.0,
      "step": 1544
    },
    {
      "epoch": 0.18268889677190492,
      "grad_norm": 0.1796652376651764,
      "learning_rate": 3.0101549661552743e-05,
      "loss": 0.351,
      "num_tokens": 978484082.0,
      "step": 1545
    },
    {
      "epoch": 0.182807142012534,
      "grad_norm": 0.18004387617111206,
      "learning_rate": 3.0070671749359963e-05,
      "loss": 0.3715,
      "num_tokens": 979120007.0,
      "step": 1546
    },
    {
      "epoch": 0.18292538725316307,
      "grad_norm": 0.1583901345729828,
      "learning_rate": 3.0039797713529264e-05,
      "loss": 0.321,
      "num_tokens": 979754257.0,
      "step": 1547
    },
    {
      "epoch": 0.18304363249379213,
      "grad_norm": 0.1803177446126938,
      "learning_rate": 3.000892759491607e-05,
      "loss": 0.364,
      "num_tokens": 980379873.0,
      "step": 1548
    },
    {
      "epoch": 0.1831618777344212,
      "grad_norm": 0.16974784433841705,
      "learning_rate": 2.9978061434370635e-05,
      "loss": 0.3492,
      "num_tokens": 981012470.0,
      "step": 1549
    },
    {
      "epoch": 0.18328012297505025,
      "grad_norm": 0.17915767431259155,
      "learning_rate": 2.9947199272737932e-05,
      "loss": 0.41,
      "num_tokens": 981645741.0,
      "step": 1550
    },
    {
      "epoch": 0.1833983682156793,
      "grad_norm": 0.17839612066745758,
      "learning_rate": 2.9916341150857692e-05,
      "loss": 0.3728,
      "num_tokens": 982273942.0,
      "step": 1551
    },
    {
      "epoch": 0.18351661345630838,
      "grad_norm": 0.17455312609672546,
      "learning_rate": 2.988548710956428e-05,
      "loss": 0.3698,
      "num_tokens": 982905114.0,
      "step": 1552
    },
    {
      "epoch": 0.18363485869693744,
      "grad_norm": 32.93268966674805,
      "learning_rate": 2.9854637189686637e-05,
      "loss": 1.4744,
      "num_tokens": 983508182.0,
      "step": 1553
    },
    {
      "epoch": 0.18375310393756653,
      "grad_norm": 0.21559761464595795,
      "learning_rate": 2.9823791432048297e-05,
      "loss": 0.3675,
      "num_tokens": 984147363.0,
      "step": 1554
    },
    {
      "epoch": 0.1838713491781956,
      "grad_norm": 0.18873609602451324,
      "learning_rate": 2.9792949877467246e-05,
      "loss": 0.3667,
      "num_tokens": 984781250.0,
      "step": 1555
    },
    {
      "epoch": 0.18398959441882465,
      "grad_norm": 0.17174896597862244,
      "learning_rate": 2.9762112566755922e-05,
      "loss": 0.3658,
      "num_tokens": 985418934.0,
      "step": 1556
    },
    {
      "epoch": 0.1841078396594537,
      "grad_norm": 0.1909123659133911,
      "learning_rate": 2.9731279540721175e-05,
      "loss": 0.3605,
      "num_tokens": 986058183.0,
      "step": 1557
    },
    {
      "epoch": 0.18422608490008277,
      "grad_norm": 0.17101429402828217,
      "learning_rate": 2.970045084016412e-05,
      "loss": 0.3281,
      "num_tokens": 986689892.0,
      "step": 1558
    },
    {
      "epoch": 0.18434433014071183,
      "grad_norm": 0.17448195815086365,
      "learning_rate": 2.966962650588022e-05,
      "loss": 0.352,
      "num_tokens": 987323095.0,
      "step": 1559
    },
    {
      "epoch": 0.1844625753813409,
      "grad_norm": 0.19304078817367554,
      "learning_rate": 2.963880657865911e-05,
      "loss": 0.3399,
      "num_tokens": 987953264.0,
      "step": 1560
    },
    {
      "epoch": 0.18458082062196995,
      "grad_norm": 0.16238828003406525,
      "learning_rate": 2.960799109928461e-05,
      "loss": 0.3489,
      "num_tokens": 988590158.0,
      "step": 1561
    },
    {
      "epoch": 0.18469906586259904,
      "grad_norm": 0.1621323823928833,
      "learning_rate": 2.9577180108534666e-05,
      "loss": 0.3777,
      "num_tokens": 989215297.0,
      "step": 1562
    },
    {
      "epoch": 0.1848173111032281,
      "grad_norm": 0.17066222429275513,
      "learning_rate": 2.9546373647181265e-05,
      "loss": 0.3627,
      "num_tokens": 989854912.0,
      "step": 1563
    },
    {
      "epoch": 0.18493555634385717,
      "grad_norm": 0.1540173441171646,
      "learning_rate": 2.9515571755990414e-05,
      "loss": 0.3255,
      "num_tokens": 990494043.0,
      "step": 1564
    },
    {
      "epoch": 0.18505380158448623,
      "grad_norm": 0.18189725279808044,
      "learning_rate": 2.9484774475722057e-05,
      "loss": 0.3697,
      "num_tokens": 991126843.0,
      "step": 1565
    },
    {
      "epoch": 0.1851720468251153,
      "grad_norm": 0.17115595936775208,
      "learning_rate": 2.945398184713005e-05,
      "loss": 0.3548,
      "num_tokens": 991761760.0,
      "step": 1566
    },
    {
      "epoch": 0.18529029206574435,
      "grad_norm": 0.16794586181640625,
      "learning_rate": 2.9423193910962092e-05,
      "loss": 0.3356,
      "num_tokens": 992401240.0,
      "step": 1567
    },
    {
      "epoch": 0.1854085373063734,
      "grad_norm": 0.16514816880226135,
      "learning_rate": 2.939241070795967e-05,
      "loss": 0.3708,
      "num_tokens": 993035492.0,
      "step": 1568
    },
    {
      "epoch": 0.18552678254700247,
      "grad_norm": 0.17977920174598694,
      "learning_rate": 2.9361632278858003e-05,
      "loss": 0.3992,
      "num_tokens": 993670701.0,
      "step": 1569
    },
    {
      "epoch": 0.18564502778763156,
      "grad_norm": 0.17868345975875854,
      "learning_rate": 2.9330858664385995e-05,
      "loss": 0.3977,
      "num_tokens": 994304128.0,
      "step": 1570
    },
    {
      "epoch": 0.18576327302826062,
      "grad_norm": 0.16653095185756683,
      "learning_rate": 2.930008990526619e-05,
      "loss": 0.3277,
      "num_tokens": 994941330.0,
      "step": 1571
    },
    {
      "epoch": 0.18588151826888968,
      "grad_norm": 0.17928676307201385,
      "learning_rate": 2.9269326042214695e-05,
      "loss": 0.3898,
      "num_tokens": 995567115.0,
      "step": 1572
    },
    {
      "epoch": 0.18599976350951875,
      "grad_norm": 0.168831005692482,
      "learning_rate": 2.923856711594114e-05,
      "loss": 0.3929,
      "num_tokens": 996201713.0,
      "step": 1573
    },
    {
      "epoch": 0.1861180087501478,
      "grad_norm": 0.1701105684041977,
      "learning_rate": 2.9207813167148624e-05,
      "loss": 0.3397,
      "num_tokens": 996834913.0,
      "step": 1574
    },
    {
      "epoch": 0.18623625399077687,
      "grad_norm": 0.16095875203609467,
      "learning_rate": 2.917706423653365e-05,
      "loss": 0.3619,
      "num_tokens": 997471417.0,
      "step": 1575
    },
    {
      "epoch": 0.18635449923140593,
      "grad_norm": 0.18966802954673767,
      "learning_rate": 2.9146320364786106e-05,
      "loss": 0.4041,
      "num_tokens": 998105800.0,
      "step": 1576
    },
    {
      "epoch": 0.186472744472035,
      "grad_norm": 0.15751314163208008,
      "learning_rate": 2.9115581592589153e-05,
      "loss": 0.4016,
      "num_tokens": 998742897.0,
      "step": 1577
    },
    {
      "epoch": 0.18659098971266405,
      "grad_norm": 0.1764875054359436,
      "learning_rate": 2.908484796061923e-05,
      "loss": 0.3892,
      "num_tokens": 999375870.0,
      "step": 1578
    },
    {
      "epoch": 0.18670923495329314,
      "grad_norm": 0.16247054934501648,
      "learning_rate": 2.9054119509545957e-05,
      "loss": 0.3617,
      "num_tokens": 1000009533.0,
      "step": 1579
    },
    {
      "epoch": 0.1868274801939222,
      "grad_norm": 0.16236217319965363,
      "learning_rate": 2.902339628003211e-05,
      "loss": 0.3605,
      "num_tokens": 1000648319.0,
      "step": 1580
    },
    {
      "epoch": 0.18694572543455126,
      "grad_norm": 0.17573755979537964,
      "learning_rate": 2.8992678312733547e-05,
      "loss": 0.3755,
      "num_tokens": 1001283756.0,
      "step": 1581
    },
    {
      "epoch": 0.18706397067518032,
      "grad_norm": 0.1554783433675766,
      "learning_rate": 2.8961965648299166e-05,
      "loss": 0.3265,
      "num_tokens": 1001920534.0,
      "step": 1582
    },
    {
      "epoch": 0.18718221591580939,
      "grad_norm": 0.1593015044927597,
      "learning_rate": 2.8931258327370845e-05,
      "loss": 0.339,
      "num_tokens": 1002538598.0,
      "step": 1583
    },
    {
      "epoch": 0.18730046115643845,
      "grad_norm": 0.1607053279876709,
      "learning_rate": 2.8900556390583394e-05,
      "loss": 0.3348,
      "num_tokens": 1003174599.0,
      "step": 1584
    },
    {
      "epoch": 0.1874187063970675,
      "grad_norm": 0.16559194028377533,
      "learning_rate": 2.8869859878564498e-05,
      "loss": 0.3475,
      "num_tokens": 1003810442.0,
      "step": 1585
    },
    {
      "epoch": 0.18753695163769657,
      "grad_norm": 0.18042126297950745,
      "learning_rate": 2.883916883193467e-05,
      "loss": 0.4163,
      "num_tokens": 1004449210.0,
      "step": 1586
    },
    {
      "epoch": 0.18765519687832566,
      "grad_norm": 0.153243288397789,
      "learning_rate": 2.8808483291307167e-05,
      "loss": 0.3196,
      "num_tokens": 1005080267.0,
      "step": 1587
    },
    {
      "epoch": 0.18777344211895472,
      "grad_norm": 0.17950351536273956,
      "learning_rate": 2.877780329728799e-05,
      "loss": 0.4136,
      "num_tokens": 1005714238.0,
      "step": 1588
    },
    {
      "epoch": 0.18789168735958378,
      "grad_norm": 0.1732146143913269,
      "learning_rate": 2.8747128890475784e-05,
      "loss": 0.3666,
      "num_tokens": 1006353420.0,
      "step": 1589
    },
    {
      "epoch": 0.18800993260021284,
      "grad_norm": 0.19320537149906158,
      "learning_rate": 2.8716460111461807e-05,
      "loss": 0.3835,
      "num_tokens": 1006983318.0,
      "step": 1590
    },
    {
      "epoch": 0.1881281778408419,
      "grad_norm": 0.1723458468914032,
      "learning_rate": 2.8685797000829853e-05,
      "loss": 0.3519,
      "num_tokens": 1007615434.0,
      "step": 1591
    },
    {
      "epoch": 0.18824642308147096,
      "grad_norm": 0.1752302050590515,
      "learning_rate": 2.8655139599156236e-05,
      "loss": 0.3644,
      "num_tokens": 1008247783.0,
      "step": 1592
    },
    {
      "epoch": 0.18836466832210003,
      "grad_norm": 0.15836386382579803,
      "learning_rate": 2.8624487947009718e-05,
      "loss": 0.3466,
      "num_tokens": 1008884013.0,
      "step": 1593
    },
    {
      "epoch": 0.1884829135627291,
      "grad_norm": 0.17339007556438446,
      "learning_rate": 2.8593842084951417e-05,
      "loss": 0.3779,
      "num_tokens": 1009517526.0,
      "step": 1594
    },
    {
      "epoch": 0.18860115880335818,
      "grad_norm": 0.15724654495716095,
      "learning_rate": 2.8563202053534844e-05,
      "loss": 0.3636,
      "num_tokens": 1010147212.0,
      "step": 1595
    },
    {
      "epoch": 0.18871940404398724,
      "grad_norm": 0.15183372795581818,
      "learning_rate": 2.8532567893305735e-05,
      "loss": 0.369,
      "num_tokens": 1010778276.0,
      "step": 1596
    },
    {
      "epoch": 0.1888376492846163,
      "grad_norm": 0.17174997925758362,
      "learning_rate": 2.8501939644802106e-05,
      "loss": 0.3508,
      "num_tokens": 1011412350.0,
      "step": 1597
    },
    {
      "epoch": 0.18895589452524536,
      "grad_norm": 0.18961110711097717,
      "learning_rate": 2.8471317348554125e-05,
      "loss": 0.3696,
      "num_tokens": 1012042617.0,
      "step": 1598
    },
    {
      "epoch": 0.18907413976587442,
      "grad_norm": 0.16691336035728455,
      "learning_rate": 2.844070104508409e-05,
      "loss": 0.3571,
      "num_tokens": 1012678514.0,
      "step": 1599
    },
    {
      "epoch": 0.18919238500650348,
      "grad_norm": 0.1767212152481079,
      "learning_rate": 2.841009077490636e-05,
      "loss": 0.3939,
      "num_tokens": 1013317458.0,
      "step": 1600
    },
    {
      "epoch": 0.18931063024713254,
      "grad_norm": 0.1530785858631134,
      "learning_rate": 2.8379486578527328e-05,
      "loss": 0.3187,
      "num_tokens": 1013951330.0,
      "step": 1601
    },
    {
      "epoch": 0.1894288754877616,
      "grad_norm": 0.16864624619483948,
      "learning_rate": 2.8348888496445332e-05,
      "loss": 0.3559,
      "num_tokens": 1014583639.0,
      "step": 1602
    },
    {
      "epoch": 0.1895471207283907,
      "grad_norm": 0.1874200999736786,
      "learning_rate": 2.831829656915063e-05,
      "loss": 0.3987,
      "num_tokens": 1015222884.0,
      "step": 1603
    },
    {
      "epoch": 0.18966536596901976,
      "grad_norm": 0.16399650275707245,
      "learning_rate": 2.8287710837125326e-05,
      "loss": 0.3792,
      "num_tokens": 1015860526.0,
      "step": 1604
    },
    {
      "epoch": 0.18978361120964882,
      "grad_norm": 0.16257967054843903,
      "learning_rate": 2.825713134084334e-05,
      "loss": 0.3507,
      "num_tokens": 1016498048.0,
      "step": 1605
    },
    {
      "epoch": 0.18990185645027788,
      "grad_norm": 0.1760384738445282,
      "learning_rate": 2.8226558120770317e-05,
      "loss": 0.3818,
      "num_tokens": 1017129789.0,
      "step": 1606
    },
    {
      "epoch": 0.19002010169090694,
      "grad_norm": 0.16825507581233978,
      "learning_rate": 2.8195991217363634e-05,
      "loss": 0.4254,
      "num_tokens": 1017761831.0,
      "step": 1607
    },
    {
      "epoch": 0.190138346931536,
      "grad_norm": 0.18675942718982697,
      "learning_rate": 2.8165430671072266e-05,
      "loss": 0.3746,
      "num_tokens": 1018397544.0,
      "step": 1608
    },
    {
      "epoch": 0.19025659217216506,
      "grad_norm": 4.314301013946533,
      "learning_rate": 2.8134876522336804e-05,
      "loss": 0.4993,
      "num_tokens": 1018998279.0,
      "step": 1609
    },
    {
      "epoch": 0.19037483741279412,
      "grad_norm": 0.19050520658493042,
      "learning_rate": 2.8104328811589373e-05,
      "loss": 0.365,
      "num_tokens": 1019634188.0,
      "step": 1610
    },
    {
      "epoch": 0.1904930826534232,
      "grad_norm": 0.19670063257217407,
      "learning_rate": 2.8073787579253554e-05,
      "loss": 0.3534,
      "num_tokens": 1020259191.0,
      "step": 1611
    },
    {
      "epoch": 0.19061132789405227,
      "grad_norm": 0.15332163870334625,
      "learning_rate": 2.8043252865744388e-05,
      "loss": 0.3383,
      "num_tokens": 1020890735.0,
      "step": 1612
    },
    {
      "epoch": 0.19072957313468134,
      "grad_norm": 0.1597706526517868,
      "learning_rate": 2.8012724711468257e-05,
      "loss": 0.3426,
      "num_tokens": 1021516346.0,
      "step": 1613
    },
    {
      "epoch": 0.1908478183753104,
      "grad_norm": 0.1752755492925644,
      "learning_rate": 2.79822031568229e-05,
      "loss": 0.374,
      "num_tokens": 1022154101.0,
      "step": 1614
    },
    {
      "epoch": 0.19096606361593946,
      "grad_norm": 0.18779323995113373,
      "learning_rate": 2.7951688242197293e-05,
      "loss": 0.402,
      "num_tokens": 1022784679.0,
      "step": 1615
    },
    {
      "epoch": 0.19108430885656852,
      "grad_norm": 0.16711129248142242,
      "learning_rate": 2.7921180007971632e-05,
      "loss": 0.335,
      "num_tokens": 1023419003.0,
      "step": 1616
    },
    {
      "epoch": 0.19120255409719758,
      "grad_norm": 0.16607694327831268,
      "learning_rate": 2.7890678494517283e-05,
      "loss": 0.3741,
      "num_tokens": 1024053560.0,
      "step": 1617
    },
    {
      "epoch": 0.19132079933782664,
      "grad_norm": 0.15859146416187286,
      "learning_rate": 2.7860183742196708e-05,
      "loss": 0.3197,
      "num_tokens": 1024689470.0,
      "step": 1618
    },
    {
      "epoch": 0.19143904457845573,
      "grad_norm": 0.15850317478179932,
      "learning_rate": 2.7829695791363426e-05,
      "loss": 0.3432,
      "num_tokens": 1025315733.0,
      "step": 1619
    },
    {
      "epoch": 0.1915572898190848,
      "grad_norm": 0.17494311928749084,
      "learning_rate": 2.7799214682361958e-05,
      "loss": 0.3673,
      "num_tokens": 1025951911.0,
      "step": 1620
    },
    {
      "epoch": 0.19167553505971385,
      "grad_norm": 0.16905559599399567,
      "learning_rate": 2.7768740455527753e-05,
      "loss": 0.3741,
      "num_tokens": 1026586199.0,
      "step": 1621
    },
    {
      "epoch": 0.19179378030034291,
      "grad_norm": 0.1669166088104248,
      "learning_rate": 2.7738273151187193e-05,
      "loss": 0.3583,
      "num_tokens": 1027217632.0,
      "step": 1622
    },
    {
      "epoch": 0.19191202554097198,
      "grad_norm": 0.20083673298358917,
      "learning_rate": 2.7707812809657455e-05,
      "loss": 0.3712,
      "num_tokens": 1027847572.0,
      "step": 1623
    },
    {
      "epoch": 0.19203027078160104,
      "grad_norm": 0.17961816489696503,
      "learning_rate": 2.7677359471246534e-05,
      "loss": 0.3693,
      "num_tokens": 1028475842.0,
      "step": 1624
    },
    {
      "epoch": 0.1921485160222301,
      "grad_norm": 0.16991600394248962,
      "learning_rate": 2.7646913176253133e-05,
      "loss": 0.3568,
      "num_tokens": 1029111898.0,
      "step": 1625
    },
    {
      "epoch": 0.19226676126285916,
      "grad_norm": 0.16991819441318512,
      "learning_rate": 2.7616473964966657e-05,
      "loss": 0.3727,
      "num_tokens": 1029743386.0,
      "step": 1626
    },
    {
      "epoch": 0.19238500650348825,
      "grad_norm": 0.17013666033744812,
      "learning_rate": 2.758604187766713e-05,
      "loss": 0.3677,
      "num_tokens": 1030378115.0,
      "step": 1627
    },
    {
      "epoch": 0.1925032517441173,
      "grad_norm": 0.18004712462425232,
      "learning_rate": 2.755561695462513e-05,
      "loss": 0.4029,
      "num_tokens": 1031003252.0,
      "step": 1628
    },
    {
      "epoch": 0.19262149698474637,
      "grad_norm": 0.16978485882282257,
      "learning_rate": 2.7525199236101797e-05,
      "loss": 0.3712,
      "num_tokens": 1031616777.0,
      "step": 1629
    },
    {
      "epoch": 0.19273974222537543,
      "grad_norm": 0.18321801722049713,
      "learning_rate": 2.7494788762348694e-05,
      "loss": 0.3895,
      "num_tokens": 1032254017.0,
      "step": 1630
    },
    {
      "epoch": 0.1928579874660045,
      "grad_norm": 0.1728045642375946,
      "learning_rate": 2.7464385573607808e-05,
      "loss": 0.4044,
      "num_tokens": 1032888699.0,
      "step": 1631
    },
    {
      "epoch": 0.19297623270663355,
      "grad_norm": 0.17345483601093292,
      "learning_rate": 2.743398971011151e-05,
      "loss": 0.3602,
      "num_tokens": 1033518207.0,
      "step": 1632
    },
    {
      "epoch": 0.19309447794726262,
      "grad_norm": 0.15876762568950653,
      "learning_rate": 2.7403601212082446e-05,
      "loss": 0.3591,
      "num_tokens": 1034156394.0,
      "step": 1633
    },
    {
      "epoch": 0.19321272318789168,
      "grad_norm": 0.1655876785516739,
      "learning_rate": 2.737322011973353e-05,
      "loss": 0.3447,
      "num_tokens": 1034757861.0,
      "step": 1634
    },
    {
      "epoch": 0.19333096842852074,
      "grad_norm": 0.15525826811790466,
      "learning_rate": 2.734284647326788e-05,
      "loss": 0.3321,
      "num_tokens": 1035390904.0,
      "step": 1635
    },
    {
      "epoch": 0.19344921366914983,
      "grad_norm": 0.16308487951755524,
      "learning_rate": 2.7312480312878753e-05,
      "loss": 0.3468,
      "num_tokens": 1036022608.0,
      "step": 1636
    },
    {
      "epoch": 0.1935674589097789,
      "grad_norm": 0.15808793902397156,
      "learning_rate": 2.7282121678749508e-05,
      "loss": 0.3231,
      "num_tokens": 1036653565.0,
      "step": 1637
    },
    {
      "epoch": 0.19368570415040795,
      "grad_norm": 0.1726795732975006,
      "learning_rate": 2.7251770611053532e-05,
      "loss": 0.3604,
      "num_tokens": 1037289640.0,
      "step": 1638
    },
    {
      "epoch": 0.193803949391037,
      "grad_norm": 0.18068532645702362,
      "learning_rate": 2.7221427149954214e-05,
      "loss": 0.3619,
      "num_tokens": 1037923126.0,
      "step": 1639
    },
    {
      "epoch": 0.19392219463166607,
      "grad_norm": 0.15515606105327606,
      "learning_rate": 2.7191091335604855e-05,
      "loss": 0.3135,
      "num_tokens": 1038552223.0,
      "step": 1640
    },
    {
      "epoch": 0.19404043987229513,
      "grad_norm": 0.14960326254367828,
      "learning_rate": 2.7160763208148668e-05,
      "loss": 0.3557,
      "num_tokens": 1039187143.0,
      "step": 1641
    },
    {
      "epoch": 0.1941586851129242,
      "grad_norm": 0.19315950572490692,
      "learning_rate": 2.713044280771867e-05,
      "loss": 0.4029,
      "num_tokens": 1039825115.0,
      "step": 1642
    },
    {
      "epoch": 0.19427693035355326,
      "grad_norm": 0.16578054428100586,
      "learning_rate": 2.7100130174437664e-05,
      "loss": 0.3622,
      "num_tokens": 1040457103.0,
      "step": 1643
    },
    {
      "epoch": 0.19439517559418235,
      "grad_norm": 0.16419482231140137,
      "learning_rate": 2.7069825348418168e-05,
      "loss": 0.326,
      "num_tokens": 1041089601.0,
      "step": 1644
    },
    {
      "epoch": 0.1945134208348114,
      "grad_norm": 0.17337539792060852,
      "learning_rate": 2.7039528369762364e-05,
      "loss": 0.3629,
      "num_tokens": 1041725934.0,
      "step": 1645
    },
    {
      "epoch": 0.19463166607544047,
      "grad_norm": 0.18786442279815674,
      "learning_rate": 2.700923927856208e-05,
      "loss": 0.3456,
      "num_tokens": 1042365339.0,
      "step": 1646
    },
    {
      "epoch": 0.19474991131606953,
      "grad_norm": 0.19918936491012573,
      "learning_rate": 2.6978958114898666e-05,
      "loss": 0.3638,
      "num_tokens": 1042996166.0,
      "step": 1647
    },
    {
      "epoch": 0.1948681565566986,
      "grad_norm": 0.15421392023563385,
      "learning_rate": 2.6948684918843013e-05,
      "loss": 0.3328,
      "num_tokens": 1043623270.0,
      "step": 1648
    },
    {
      "epoch": 0.19498640179732765,
      "grad_norm": 0.18007415533065796,
      "learning_rate": 2.6918419730455445e-05,
      "loss": 0.3468,
      "num_tokens": 1044254863.0,
      "step": 1649
    },
    {
      "epoch": 0.1951046470379567,
      "grad_norm": 0.19268766045570374,
      "learning_rate": 2.6888162589785698e-05,
      "loss": 0.4225,
      "num_tokens": 1044894564.0,
      "step": 1650
    },
    {
      "epoch": 0.19522289227858577,
      "grad_norm": 0.16468672454357147,
      "learning_rate": 2.6857913536872868e-05,
      "loss": 0.3486,
      "num_tokens": 1045527684.0,
      "step": 1651
    },
    {
      "epoch": 0.19534113751921486,
      "grad_norm": 0.15274105966091156,
      "learning_rate": 2.682767261174534e-05,
      "loss": 0.3578,
      "num_tokens": 1046158648.0,
      "step": 1652
    },
    {
      "epoch": 0.19545938275984392,
      "grad_norm": 0.16642116010189056,
      "learning_rate": 2.6797439854420747e-05,
      "loss": 0.3553,
      "num_tokens": 1046792143.0,
      "step": 1653
    },
    {
      "epoch": 0.195577628000473,
      "grad_norm": 0.19157806038856506,
      "learning_rate": 2.676721530490591e-05,
      "loss": 0.3833,
      "num_tokens": 1047425498.0,
      "step": 1654
    },
    {
      "epoch": 0.19569587324110205,
      "grad_norm": 0.18173323571681976,
      "learning_rate": 2.6736999003196772e-05,
      "loss": 0.3635,
      "num_tokens": 1048059505.0,
      "step": 1655
    },
    {
      "epoch": 0.1958141184817311,
      "grad_norm": 0.18953822553157806,
      "learning_rate": 2.6706790989278408e-05,
      "loss": 0.3783,
      "num_tokens": 1048693153.0,
      "step": 1656
    },
    {
      "epoch": 0.19593236372236017,
      "grad_norm": 0.17194925248622894,
      "learning_rate": 2.667659130312488e-05,
      "loss": 0.3987,
      "num_tokens": 1049329857.0,
      "step": 1657
    },
    {
      "epoch": 0.19605060896298923,
      "grad_norm": 0.16898545622825623,
      "learning_rate": 2.6646399984699254e-05,
      "loss": 0.3372,
      "num_tokens": 1049961195.0,
      "step": 1658
    },
    {
      "epoch": 0.1961688542036183,
      "grad_norm": 0.21434806287288666,
      "learning_rate": 2.6616217073953512e-05,
      "loss": 0.3945,
      "num_tokens": 1050595023.0,
      "step": 1659
    },
    {
      "epoch": 0.19628709944424738,
      "grad_norm": 0.18597570061683655,
      "learning_rate": 2.658604261082851e-05,
      "loss": 0.3993,
      "num_tokens": 1051234499.0,
      "step": 1660
    },
    {
      "epoch": 0.19640534468487644,
      "grad_norm": 0.16699214279651642,
      "learning_rate": 2.655587663525395e-05,
      "loss": 0.3743,
      "num_tokens": 1051865728.0,
      "step": 1661
    },
    {
      "epoch": 0.1965235899255055,
      "grad_norm": 0.17103753983974457,
      "learning_rate": 2.6525719187148245e-05,
      "loss": 0.356,
      "num_tokens": 1052494691.0,
      "step": 1662
    },
    {
      "epoch": 0.19664183516613457,
      "grad_norm": 0.1663488894701004,
      "learning_rate": 2.6495570306418603e-05,
      "loss": 0.3462,
      "num_tokens": 1053129843.0,
      "step": 1663
    },
    {
      "epoch": 0.19676008040676363,
      "grad_norm": 0.18791471421718597,
      "learning_rate": 2.6465430032960825e-05,
      "loss": 0.3607,
      "num_tokens": 1053762168.0,
      "step": 1664
    },
    {
      "epoch": 0.1968783256473927,
      "grad_norm": 0.18649154901504517,
      "learning_rate": 2.643529840665936e-05,
      "loss": 0.3853,
      "num_tokens": 1054399760.0,
      "step": 1665
    },
    {
      "epoch": 0.19699657088802175,
      "grad_norm": 0.15763476490974426,
      "learning_rate": 2.640517546738721e-05,
      "loss": 0.3716,
      "num_tokens": 1055028654.0,
      "step": 1666
    },
    {
      "epoch": 0.1971148161286508,
      "grad_norm": 0.19678330421447754,
      "learning_rate": 2.637506125500586e-05,
      "loss": 0.3564,
      "num_tokens": 1055660000.0,
      "step": 1667
    },
    {
      "epoch": 0.1972330613692799,
      "grad_norm": 0.17163364589214325,
      "learning_rate": 2.634495580936528e-05,
      "loss": 0.3615,
      "num_tokens": 1056296438.0,
      "step": 1668
    },
    {
      "epoch": 0.19735130660990896,
      "grad_norm": 0.16406387090682983,
      "learning_rate": 2.6314859170303796e-05,
      "loss": 0.3503,
      "num_tokens": 1056927317.0,
      "step": 1669
    },
    {
      "epoch": 0.19746955185053802,
      "grad_norm": 0.17279218137264252,
      "learning_rate": 2.6284771377648132e-05,
      "loss": 0.3772,
      "num_tokens": 1057562315.0,
      "step": 1670
    },
    {
      "epoch": 0.19758779709116708,
      "grad_norm": 0.19749115407466888,
      "learning_rate": 2.625469247121327e-05,
      "loss": 0.3677,
      "num_tokens": 1058196627.0,
      "step": 1671
    },
    {
      "epoch": 0.19770604233179614,
      "grad_norm": 0.1557900458574295,
      "learning_rate": 2.6224622490802433e-05,
      "loss": 0.3297,
      "num_tokens": 1058830504.0,
      "step": 1672
    },
    {
      "epoch": 0.1978242875724252,
      "grad_norm": 0.1759173572063446,
      "learning_rate": 2.6194561476207054e-05,
      "loss": 0.3722,
      "num_tokens": 1059458402.0,
      "step": 1673
    },
    {
      "epoch": 0.19794253281305427,
      "grad_norm": 0.18761344254016876,
      "learning_rate": 2.6164509467206678e-05,
      "loss": 0.3305,
      "num_tokens": 1060092121.0,
      "step": 1674
    },
    {
      "epoch": 0.19806077805368333,
      "grad_norm": 0.1713910549879074,
      "learning_rate": 2.613446650356896e-05,
      "loss": 0.3497,
      "num_tokens": 1060729280.0,
      "step": 1675
    },
    {
      "epoch": 0.19817902329431242,
      "grad_norm": 0.15927839279174805,
      "learning_rate": 2.610443262504956e-05,
      "loss": 0.3563,
      "num_tokens": 1061357345.0,
      "step": 1676
    },
    {
      "epoch": 0.19829726853494148,
      "grad_norm": 0.1804131269454956,
      "learning_rate": 2.6074407871392115e-05,
      "loss": 0.3501,
      "num_tokens": 1061986290.0,
      "step": 1677
    },
    {
      "epoch": 0.19841551377557054,
      "grad_norm": 0.1799393892288208,
      "learning_rate": 2.604439228232821e-05,
      "loss": 0.3674,
      "num_tokens": 1062616696.0,
      "step": 1678
    },
    {
      "epoch": 0.1985337590161996,
      "grad_norm": 0.17438676953315735,
      "learning_rate": 2.6014385897577285e-05,
      "loss": 0.3641,
      "num_tokens": 1063253499.0,
      "step": 1679
    },
    {
      "epoch": 0.19865200425682866,
      "grad_norm": 0.15909676253795624,
      "learning_rate": 2.5984388756846617e-05,
      "loss": 0.3865,
      "num_tokens": 1063890393.0,
      "step": 1680
    },
    {
      "epoch": 0.19877024949745772,
      "grad_norm": 0.18874254822731018,
      "learning_rate": 2.595440089983121e-05,
      "loss": 0.3992,
      "num_tokens": 1064526042.0,
      "step": 1681
    },
    {
      "epoch": 0.19888849473808679,
      "grad_norm": 0.1858903169631958,
      "learning_rate": 2.5924422366213847e-05,
      "loss": 0.3896,
      "num_tokens": 1065162440.0,
      "step": 1682
    },
    {
      "epoch": 0.19900673997871585,
      "grad_norm": 0.15958355367183685,
      "learning_rate": 2.5894453195664908e-05,
      "loss": 0.3594,
      "num_tokens": 1065798256.0,
      "step": 1683
    },
    {
      "epoch": 0.1991249852193449,
      "grad_norm": 0.18213844299316406,
      "learning_rate": 2.586449342784242e-05,
      "loss": 0.3542,
      "num_tokens": 1066431917.0,
      "step": 1684
    },
    {
      "epoch": 0.199243230459974,
      "grad_norm": 0.162159264087677,
      "learning_rate": 2.583454310239197e-05,
      "loss": 0.3436,
      "num_tokens": 1067062494.0,
      "step": 1685
    },
    {
      "epoch": 0.19936147570060306,
      "grad_norm": 0.16742713749408722,
      "learning_rate": 2.580460225894663e-05,
      "loss": 0.3506,
      "num_tokens": 1067697069.0,
      "step": 1686
    },
    {
      "epoch": 0.19947972094123212,
      "grad_norm": 0.17315392196178436,
      "learning_rate": 2.5774670937126936e-05,
      "loss": 0.3955,
      "num_tokens": 1068295187.0,
      "step": 1687
    },
    {
      "epoch": 0.19959796618186118,
      "grad_norm": 0.1787833422422409,
      "learning_rate": 2.5744749176540828e-05,
      "loss": 0.3472,
      "num_tokens": 1068932515.0,
      "step": 1688
    },
    {
      "epoch": 0.19971621142249024,
      "grad_norm": 0.166206493973732,
      "learning_rate": 2.5714837016783576e-05,
      "loss": 0.3418,
      "num_tokens": 1069570945.0,
      "step": 1689
    },
    {
      "epoch": 0.1998344566631193,
      "grad_norm": 0.1507144272327423,
      "learning_rate": 2.568493449743777e-05,
      "loss": 0.3273,
      "num_tokens": 1070176282.0,
      "step": 1690
    },
    {
      "epoch": 0.19995270190374836,
      "grad_norm": 0.1695074588060379,
      "learning_rate": 2.565504165807322e-05,
      "loss": 0.3839,
      "num_tokens": 1070811105.0,
      "step": 1691
    },
    {
      "epoch": 0.20007094714437743,
      "grad_norm": 0.16034561395645142,
      "learning_rate": 2.5625158538246946e-05,
      "loss": 0.3275,
      "num_tokens": 1071447938.0,
      "step": 1692
    },
    {
      "epoch": 0.20018919238500651,
      "grad_norm": 0.17031002044677734,
      "learning_rate": 2.5595285177503097e-05,
      "loss": 0.3471,
      "num_tokens": 1072084659.0,
      "step": 1693
    },
    {
      "epoch": 0.20030743762563558,
      "grad_norm": 0.1551053524017334,
      "learning_rate": 2.556542161537289e-05,
      "loss": 0.3391,
      "num_tokens": 1072721591.0,
      "step": 1694
    },
    {
      "epoch": 0.20042568286626464,
      "grad_norm": 0.1724974513053894,
      "learning_rate": 2.5535567891374617e-05,
      "loss": 0.393,
      "num_tokens": 1073359151.0,
      "step": 1695
    },
    {
      "epoch": 0.2005439281068937,
      "grad_norm": 0.1494995504617691,
      "learning_rate": 2.5505724045013514e-05,
      "loss": 0.3014,
      "num_tokens": 1073993201.0,
      "step": 1696
    },
    {
      "epoch": 0.20066217334752276,
      "grad_norm": 0.17537805438041687,
      "learning_rate": 2.5475890115781768e-05,
      "loss": 0.3731,
      "num_tokens": 1074627609.0,
      "step": 1697
    },
    {
      "epoch": 0.20078041858815182,
      "grad_norm": 0.1633339673280716,
      "learning_rate": 2.544606614315842e-05,
      "loss": 0.3697,
      "num_tokens": 1075266812.0,
      "step": 1698
    },
    {
      "epoch": 0.20089866382878088,
      "grad_norm": 0.19342614710330963,
      "learning_rate": 2.5416252166609366e-05,
      "loss": 0.3752,
      "num_tokens": 1075904344.0,
      "step": 1699
    },
    {
      "epoch": 0.20101690906940994,
      "grad_norm": 0.17040970921516418,
      "learning_rate": 2.5386448225587255e-05,
      "loss": 0.3567,
      "num_tokens": 1076539387.0,
      "step": 1700
    },
    {
      "epoch": 0.20113515431003903,
      "grad_norm": 0.151948943734169,
      "learning_rate": 2.5356654359531447e-05,
      "loss": 0.3295,
      "num_tokens": 1077171298.0,
      "step": 1701
    },
    {
      "epoch": 0.2012533995506681,
      "grad_norm": 0.16101253032684326,
      "learning_rate": 2.5326870607868e-05,
      "loss": 0.3489,
      "num_tokens": 1077802493.0,
      "step": 1702
    },
    {
      "epoch": 0.20137164479129716,
      "grad_norm": 0.1759985238313675,
      "learning_rate": 2.529709701000955e-05,
      "loss": 0.3765,
      "num_tokens": 1078441368.0,
      "step": 1703
    },
    {
      "epoch": 0.20148989003192622,
      "grad_norm": 0.17887002229690552,
      "learning_rate": 2.526733360535533e-05,
      "loss": 0.33,
      "num_tokens": 1079072648.0,
      "step": 1704
    },
    {
      "epoch": 0.20160813527255528,
      "grad_norm": 0.15795493125915527,
      "learning_rate": 2.523758043329107e-05,
      "loss": 0.3403,
      "num_tokens": 1079706981.0,
      "step": 1705
    },
    {
      "epoch": 0.20172638051318434,
      "grad_norm": 0.19332772493362427,
      "learning_rate": 2.5207837533188952e-05,
      "loss": 0.4124,
      "num_tokens": 1080339530.0,
      "step": 1706
    },
    {
      "epoch": 0.2018446257538134,
      "grad_norm": 0.1674606204032898,
      "learning_rate": 2.5178104944407583e-05,
      "loss": 0.3376,
      "num_tokens": 1080970260.0,
      "step": 1707
    },
    {
      "epoch": 0.20196287099444246,
      "grad_norm": 0.17070183157920837,
      "learning_rate": 2.5148382706291906e-05,
      "loss": 0.3715,
      "num_tokens": 1081590177.0,
      "step": 1708
    },
    {
      "epoch": 0.20208111623507155,
      "grad_norm": 0.1682654172182083,
      "learning_rate": 2.5118670858173184e-05,
      "loss": 0.3623,
      "num_tokens": 1082225090.0,
      "step": 1709
    },
    {
      "epoch": 0.2021993614757006,
      "grad_norm": 0.17373022437095642,
      "learning_rate": 2.5088969439368924e-05,
      "loss": 0.3347,
      "num_tokens": 1082857496.0,
      "step": 1710
    },
    {
      "epoch": 0.20231760671632967,
      "grad_norm": 0.19974088668823242,
      "learning_rate": 2.5059278489182814e-05,
      "loss": 0.4124,
      "num_tokens": 1083488770.0,
      "step": 1711
    },
    {
      "epoch": 0.20243585195695873,
      "grad_norm": 0.1808067411184311,
      "learning_rate": 2.5029598046904727e-05,
      "loss": 0.3373,
      "num_tokens": 1084124695.0,
      "step": 1712
    },
    {
      "epoch": 0.2025540971975878,
      "grad_norm": 0.16108886897563934,
      "learning_rate": 2.49999281518106e-05,
      "loss": 0.3304,
      "num_tokens": 1084757117.0,
      "step": 1713
    },
    {
      "epoch": 0.20267234243821686,
      "grad_norm": 0.17718425393104553,
      "learning_rate": 2.4970268843162426e-05,
      "loss": 0.353,
      "num_tokens": 1085390682.0,
      "step": 1714
    },
    {
      "epoch": 0.20279058767884592,
      "grad_norm": 0.17533092200756073,
      "learning_rate": 2.494062016020818e-05,
      "loss": 0.33,
      "num_tokens": 1086017829.0,
      "step": 1715
    },
    {
      "epoch": 0.20290883291947498,
      "grad_norm": 0.1918751746416092,
      "learning_rate": 2.49109821421818e-05,
      "loss": 0.3822,
      "num_tokens": 1086653372.0,
      "step": 1716
    },
    {
      "epoch": 0.20302707816010407,
      "grad_norm": 0.1925252079963684,
      "learning_rate": 2.4881354828303063e-05,
      "loss": 0.3553,
      "num_tokens": 1087290946.0,
      "step": 1717
    },
    {
      "epoch": 0.20314532340073313,
      "grad_norm": 0.18237020075321198,
      "learning_rate": 2.4851738257777637e-05,
      "loss": 0.3843,
      "num_tokens": 1087920118.0,
      "step": 1718
    },
    {
      "epoch": 0.2032635686413622,
      "grad_norm": 0.1720273792743683,
      "learning_rate": 2.4822132469796943e-05,
      "loss": 0.3434,
      "num_tokens": 1088552138.0,
      "step": 1719
    },
    {
      "epoch": 0.20338181388199125,
      "grad_norm": 0.18255725502967834,
      "learning_rate": 2.4792537503538127e-05,
      "loss": 0.3755,
      "num_tokens": 1089182532.0,
      "step": 1720
    },
    {
      "epoch": 0.2035000591226203,
      "grad_norm": 0.1829424500465393,
      "learning_rate": 2.476295339816404e-05,
      "loss": 0.3441,
      "num_tokens": 1089818206.0,
      "step": 1721
    },
    {
      "epoch": 0.20361830436324937,
      "grad_norm": 0.17252077162265778,
      "learning_rate": 2.473338019282314e-05,
      "loss": 0.3621,
      "num_tokens": 1090453385.0,
      "step": 1722
    },
    {
      "epoch": 0.20373654960387844,
      "grad_norm": 0.17650392651557922,
      "learning_rate": 2.470381792664947e-05,
      "loss": 0.3314,
      "num_tokens": 1091090465.0,
      "step": 1723
    },
    {
      "epoch": 0.2038547948445075,
      "grad_norm": 0.16138392686843872,
      "learning_rate": 2.467426663876259e-05,
      "loss": 0.3463,
      "num_tokens": 1091724132.0,
      "step": 1724
    },
    {
      "epoch": 0.2039730400851366,
      "grad_norm": 0.18246984481811523,
      "learning_rate": 2.4644726368267546e-05,
      "loss": 0.3902,
      "num_tokens": 1092359583.0,
      "step": 1725
    },
    {
      "epoch": 0.20409128532576565,
      "grad_norm": 0.17352929711341858,
      "learning_rate": 2.461519715425479e-05,
      "loss": 0.3564,
      "num_tokens": 1092994941.0,
      "step": 1726
    },
    {
      "epoch": 0.2042095305663947,
      "grad_norm": 0.16574281454086304,
      "learning_rate": 2.4585679035800152e-05,
      "loss": 0.3718,
      "num_tokens": 1093633821.0,
      "step": 1727
    },
    {
      "epoch": 0.20432777580702377,
      "grad_norm": 0.15748518705368042,
      "learning_rate": 2.4556172051964777e-05,
      "loss": 0.3956,
      "num_tokens": 1094271599.0,
      "step": 1728
    },
    {
      "epoch": 0.20444602104765283,
      "grad_norm": 0.1616101861000061,
      "learning_rate": 2.452667624179508e-05,
      "loss": 0.3509,
      "num_tokens": 1094899179.0,
      "step": 1729
    },
    {
      "epoch": 0.2045642662882819,
      "grad_norm": 0.1773674488067627,
      "learning_rate": 2.4497191644322678e-05,
      "loss": 0.3386,
      "num_tokens": 1095533149.0,
      "step": 1730
    },
    {
      "epoch": 0.20468251152891095,
      "grad_norm": 0.21328280866146088,
      "learning_rate": 2.4467718298564364e-05,
      "loss": 0.4216,
      "num_tokens": 1096169826.0,
      "step": 1731
    },
    {
      "epoch": 0.20480075676954002,
      "grad_norm": 0.17868073284626007,
      "learning_rate": 2.4438256243522022e-05,
      "loss": 0.3998,
      "num_tokens": 1096805072.0,
      "step": 1732
    },
    {
      "epoch": 0.2049190020101691,
      "grad_norm": 0.18678660690784454,
      "learning_rate": 2.4408805518182634e-05,
      "loss": 0.3624,
      "num_tokens": 1097431523.0,
      "step": 1733
    },
    {
      "epoch": 0.20503724725079817,
      "grad_norm": 0.16271109879016876,
      "learning_rate": 2.437936616151814e-05,
      "loss": 0.345,
      "num_tokens": 1098064979.0,
      "step": 1734
    },
    {
      "epoch": 0.20515549249142723,
      "grad_norm": 0.1655609905719757,
      "learning_rate": 2.434993821248547e-05,
      "loss": 0.3303,
      "num_tokens": 1098696612.0,
      "step": 1735
    },
    {
      "epoch": 0.2052737377320563,
      "grad_norm": 0.18166378140449524,
      "learning_rate": 2.432052171002645e-05,
      "loss": 0.38,
      "num_tokens": 1099334882.0,
      "step": 1736
    },
    {
      "epoch": 0.20539198297268535,
      "grad_norm": 0.1831236183643341,
      "learning_rate": 2.429111669306775e-05,
      "loss": 0.3816,
      "num_tokens": 1099965725.0,
      "step": 1737
    },
    {
      "epoch": 0.2055102282133144,
      "grad_norm": 0.1645020991563797,
      "learning_rate": 2.4261723200520855e-05,
      "loss": 0.3243,
      "num_tokens": 1100599444.0,
      "step": 1738
    },
    {
      "epoch": 0.20562847345394347,
      "grad_norm": 0.17984479665756226,
      "learning_rate": 2.4232341271281997e-05,
      "loss": 0.3783,
      "num_tokens": 1101235842.0,
      "step": 1739
    },
    {
      "epoch": 0.20574671869457253,
      "grad_norm": 0.1687169224023819,
      "learning_rate": 2.420297094423209e-05,
      "loss": 0.361,
      "num_tokens": 1101872512.0,
      "step": 1740
    },
    {
      "epoch": 0.2058649639352016,
      "grad_norm": 0.18295352160930634,
      "learning_rate": 2.417361225823671e-05,
      "loss": 0.3829,
      "num_tokens": 1102511392.0,
      "step": 1741
    },
    {
      "epoch": 0.20598320917583068,
      "grad_norm": 0.1548542082309723,
      "learning_rate": 2.4144265252146034e-05,
      "loss": 0.3123,
      "num_tokens": 1103143502.0,
      "step": 1742
    },
    {
      "epoch": 0.20610145441645975,
      "grad_norm": 0.17851783335208893,
      "learning_rate": 2.4114929964794767e-05,
      "loss": 0.3264,
      "num_tokens": 1103778115.0,
      "step": 1743
    },
    {
      "epoch": 0.2062196996570888,
      "grad_norm": 0.1701647937297821,
      "learning_rate": 2.4085606435002127e-05,
      "loss": 0.339,
      "num_tokens": 1104417351.0,
      "step": 1744
    },
    {
      "epoch": 0.20633794489771787,
      "grad_norm": 0.17332208156585693,
      "learning_rate": 2.405629470157174e-05,
      "loss": 0.3602,
      "num_tokens": 1105051447.0,
      "step": 1745
    },
    {
      "epoch": 0.20645619013834693,
      "grad_norm": 0.1794467568397522,
      "learning_rate": 2.402699480329167e-05,
      "loss": 0.3944,
      "num_tokens": 1105684481.0,
      "step": 1746
    },
    {
      "epoch": 0.206574435378976,
      "grad_norm": 0.18091942369937897,
      "learning_rate": 2.3997706778934262e-05,
      "loss": 0.348,
      "num_tokens": 1106316069.0,
      "step": 1747
    },
    {
      "epoch": 0.20669268061960505,
      "grad_norm": 0.16843995451927185,
      "learning_rate": 2.3968430667256202e-05,
      "loss": 0.3588,
      "num_tokens": 1106946321.0,
      "step": 1748
    },
    {
      "epoch": 0.2068109258602341,
      "grad_norm": 0.1799001395702362,
      "learning_rate": 2.3939166506998377e-05,
      "loss": 0.3617,
      "num_tokens": 1107578837.0,
      "step": 1749
    },
    {
      "epoch": 0.2069291711008632,
      "grad_norm": 0.1762133687734604,
      "learning_rate": 2.3909914336885882e-05,
      "loss": 0.3737,
      "num_tokens": 1108209790.0,
      "step": 1750
    },
    {
      "epoch": 0.20704741634149226,
      "grad_norm": 0.1713460087776184,
      "learning_rate": 2.388067419562792e-05,
      "loss": 0.3642,
      "num_tokens": 1108848427.0,
      "step": 1751
    },
    {
      "epoch": 0.20716566158212132,
      "grad_norm": 0.1797306388616562,
      "learning_rate": 2.3851446121917798e-05,
      "loss": 0.3477,
      "num_tokens": 1109482368.0,
      "step": 1752
    },
    {
      "epoch": 0.20728390682275039,
      "grad_norm": 0.16762728989124298,
      "learning_rate": 2.3822230154432852e-05,
      "loss": 0.3364,
      "num_tokens": 1110116891.0,
      "step": 1753
    },
    {
      "epoch": 0.20740215206337945,
      "grad_norm": 0.17379309237003326,
      "learning_rate": 2.3793026331834378e-05,
      "loss": 0.3409,
      "num_tokens": 1110751086.0,
      "step": 1754
    },
    {
      "epoch": 0.2075203973040085,
      "grad_norm": 0.17598725855350494,
      "learning_rate": 2.3763834692767625e-05,
      "loss": 0.3564,
      "num_tokens": 1111379871.0,
      "step": 1755
    },
    {
      "epoch": 0.20763864254463757,
      "grad_norm": 0.16308774054050446,
      "learning_rate": 2.373465527586171e-05,
      "loss": 0.3693,
      "num_tokens": 1112019590.0,
      "step": 1756
    },
    {
      "epoch": 0.20775688778526663,
      "grad_norm": 0.17762301862239838,
      "learning_rate": 2.370548811972957e-05,
      "loss": 0.3517,
      "num_tokens": 1112657954.0,
      "step": 1757
    },
    {
      "epoch": 0.20787513302589572,
      "grad_norm": 0.16102401912212372,
      "learning_rate": 2.3676333262967934e-05,
      "loss": 0.3487,
      "num_tokens": 1113297065.0,
      "step": 1758
    },
    {
      "epoch": 0.20799337826652478,
      "grad_norm": 0.16705387830734253,
      "learning_rate": 2.364719074415723e-05,
      "loss": 0.3675,
      "num_tokens": 1113925418.0,
      "step": 1759
    },
    {
      "epoch": 0.20811162350715384,
      "grad_norm": 0.1803237944841385,
      "learning_rate": 2.3618060601861592e-05,
      "loss": 0.3594,
      "num_tokens": 1114557861.0,
      "step": 1760
    },
    {
      "epoch": 0.2082298687477829,
      "grad_norm": 0.1534406542778015,
      "learning_rate": 2.3588942874628747e-05,
      "loss": 0.3347,
      "num_tokens": 1115190865.0,
      "step": 1761
    },
    {
      "epoch": 0.20834811398841196,
      "grad_norm": 0.17831236124038696,
      "learning_rate": 2.3559837600989997e-05,
      "loss": 0.3416,
      "num_tokens": 1115828417.0,
      "step": 1762
    },
    {
      "epoch": 0.20846635922904103,
      "grad_norm": 0.16847272217273712,
      "learning_rate": 2.3530744819460194e-05,
      "loss": 0.3683,
      "num_tokens": 1116467782.0,
      "step": 1763
    },
    {
      "epoch": 0.2085846044696701,
      "grad_norm": 0.16194409132003784,
      "learning_rate": 2.350166456853762e-05,
      "loss": 0.3391,
      "num_tokens": 1117103530.0,
      "step": 1764
    },
    {
      "epoch": 0.20870284971029915,
      "grad_norm": 0.1709127426147461,
      "learning_rate": 2.3472596886703995e-05,
      "loss": 0.3738,
      "num_tokens": 1117737680.0,
      "step": 1765
    },
    {
      "epoch": 0.20882109495092824,
      "grad_norm": 0.17543098330497742,
      "learning_rate": 2.344354181242441e-05,
      "loss": 0.3768,
      "num_tokens": 1118373673.0,
      "step": 1766
    },
    {
      "epoch": 0.2089393401915573,
      "grad_norm": 0.18270130455493927,
      "learning_rate": 2.341449938414726e-05,
      "loss": 0.3374,
      "num_tokens": 1118998834.0,
      "step": 1767
    },
    {
      "epoch": 0.20905758543218636,
      "grad_norm": 0.20289045572280884,
      "learning_rate": 2.3385469640304216e-05,
      "loss": 0.3752,
      "num_tokens": 1119628929.0,
      "step": 1768
    },
    {
      "epoch": 0.20917583067281542,
      "grad_norm": 0.1586165726184845,
      "learning_rate": 2.335645261931015e-05,
      "loss": 0.3196,
      "num_tokens": 1120265763.0,
      "step": 1769
    },
    {
      "epoch": 0.20929407591344448,
      "grad_norm": 0.15601249039173126,
      "learning_rate": 2.3327448359563132e-05,
      "loss": 0.3641,
      "num_tokens": 1120900582.0,
      "step": 1770
    },
    {
      "epoch": 0.20941232115407354,
      "grad_norm": 0.1694789081811905,
      "learning_rate": 2.329845689944429e-05,
      "loss": 0.3457,
      "num_tokens": 1121525988.0,
      "step": 1771
    },
    {
      "epoch": 0.2095305663947026,
      "grad_norm": 0.17258980870246887,
      "learning_rate": 2.3269478277317866e-05,
      "loss": 0.3553,
      "num_tokens": 1122163965.0,
      "step": 1772
    },
    {
      "epoch": 0.20964881163533167,
      "grad_norm": 0.17151235044002533,
      "learning_rate": 2.3240512531531096e-05,
      "loss": 0.3624,
      "num_tokens": 1122801406.0,
      "step": 1773
    },
    {
      "epoch": 0.20976705687596076,
      "grad_norm": 0.16250571608543396,
      "learning_rate": 2.321155970041415e-05,
      "loss": 0.3499,
      "num_tokens": 1123437251.0,
      "step": 1774
    },
    {
      "epoch": 0.20988530211658982,
      "grad_norm": 0.1781967729330063,
      "learning_rate": 2.3182619822280153e-05,
      "loss": 0.3889,
      "num_tokens": 1124069900.0,
      "step": 1775
    },
    {
      "epoch": 0.21000354735721888,
      "grad_norm": 0.16175100207328796,
      "learning_rate": 2.3153692935425053e-05,
      "loss": 0.3405,
      "num_tokens": 1124707468.0,
      "step": 1776
    },
    {
      "epoch": 0.21012179259784794,
      "grad_norm": 0.16345171630382538,
      "learning_rate": 2.312477907812764e-05,
      "loss": 0.3356,
      "num_tokens": 1125310855.0,
      "step": 1777
    },
    {
      "epoch": 0.210240037838477,
      "grad_norm": 0.17669813334941864,
      "learning_rate": 2.309587828864943e-05,
      "loss": 0.3464,
      "num_tokens": 1125920063.0,
      "step": 1778
    },
    {
      "epoch": 0.21035828307910606,
      "grad_norm": 0.17977413535118103,
      "learning_rate": 2.306699060523466e-05,
      "loss": 0.3255,
      "num_tokens": 1126553148.0,
      "step": 1779
    },
    {
      "epoch": 0.21047652831973512,
      "grad_norm": 0.17399907112121582,
      "learning_rate": 2.3038116066110236e-05,
      "loss": 0.3508,
      "num_tokens": 1127180645.0,
      "step": 1780
    },
    {
      "epoch": 0.21059477356036418,
      "grad_norm": 0.16113130748271942,
      "learning_rate": 2.300925470948564e-05,
      "loss": 0.3891,
      "num_tokens": 1127814988.0,
      "step": 1781
    },
    {
      "epoch": 0.21071301880099327,
      "grad_norm": 0.16437307000160217,
      "learning_rate": 2.2980406573552945e-05,
      "loss": 0.3555,
      "num_tokens": 1128450513.0,
      "step": 1782
    },
    {
      "epoch": 0.21083126404162233,
      "grad_norm": 0.17271071672439575,
      "learning_rate": 2.29515716964867e-05,
      "loss": 0.3724,
      "num_tokens": 1129084385.0,
      "step": 1783
    },
    {
      "epoch": 0.2109495092822514,
      "grad_norm": 0.15812721848487854,
      "learning_rate": 2.2922750116443927e-05,
      "loss": 0.3517,
      "num_tokens": 1129723803.0,
      "step": 1784
    },
    {
      "epoch": 0.21106775452288046,
      "grad_norm": 0.17772532999515533,
      "learning_rate": 2.289394187156405e-05,
      "loss": 0.4023,
      "num_tokens": 1130361719.0,
      "step": 1785
    },
    {
      "epoch": 0.21118599976350952,
      "grad_norm": 0.17160187661647797,
      "learning_rate": 2.286514699996883e-05,
      "loss": 0.3672,
      "num_tokens": 1130996067.0,
      "step": 1786
    },
    {
      "epoch": 0.21130424500413858,
      "grad_norm": 0.1835414469242096,
      "learning_rate": 2.2836365539762362e-05,
      "loss": 0.3969,
      "num_tokens": 1131630388.0,
      "step": 1787
    },
    {
      "epoch": 0.21142249024476764,
      "grad_norm": 0.16501553356647491,
      "learning_rate": 2.2807597529030955e-05,
      "loss": 0.3428,
      "num_tokens": 1132265944.0,
      "step": 1788
    },
    {
      "epoch": 0.2115407354853967,
      "grad_norm": 0.16104234755039215,
      "learning_rate": 2.2778843005843156e-05,
      "loss": 0.3425,
      "num_tokens": 1132901366.0,
      "step": 1789
    },
    {
      "epoch": 0.21165898072602576,
      "grad_norm": 0.16584618389606476,
      "learning_rate": 2.275010200824964e-05,
      "loss": 0.3303,
      "num_tokens": 1133533829.0,
      "step": 1790
    },
    {
      "epoch": 0.21177722596665485,
      "grad_norm": 0.18711631000041962,
      "learning_rate": 2.27213745742832e-05,
      "loss": 0.3629,
      "num_tokens": 1134167426.0,
      "step": 1791
    },
    {
      "epoch": 0.21189547120728391,
      "grad_norm": 0.15587639808654785,
      "learning_rate": 2.2692660741958667e-05,
      "loss": 0.3351,
      "num_tokens": 1134798148.0,
      "step": 1792
    },
    {
      "epoch": 0.21201371644791298,
      "grad_norm": 0.1677485555410385,
      "learning_rate": 2.2663960549272876e-05,
      "loss": 0.3583,
      "num_tokens": 1135435451.0,
      "step": 1793
    },
    {
      "epoch": 0.21213196168854204,
      "grad_norm": 0.17804494500160217,
      "learning_rate": 2.2635274034204626e-05,
      "loss": 0.4105,
      "num_tokens": 1136074929.0,
      "step": 1794
    },
    {
      "epoch": 0.2122502069291711,
      "grad_norm": 0.16529813408851624,
      "learning_rate": 2.260660123471461e-05,
      "loss": 0.3778,
      "num_tokens": 1136709588.0,
      "step": 1795
    },
    {
      "epoch": 0.21236845216980016,
      "grad_norm": 0.15820159018039703,
      "learning_rate": 2.257794218874535e-05,
      "loss": 0.3406,
      "num_tokens": 1137345661.0,
      "step": 1796
    },
    {
      "epoch": 0.21248669741042922,
      "grad_norm": 0.16220960021018982,
      "learning_rate": 2.2549296934221193e-05,
      "loss": 0.3483,
      "num_tokens": 1137949799.0,
      "step": 1797
    },
    {
      "epoch": 0.21260494265105828,
      "grad_norm": 0.15993450582027435,
      "learning_rate": 2.2520665509048228e-05,
      "loss": 0.347,
      "num_tokens": 1138575265.0,
      "step": 1798
    },
    {
      "epoch": 0.21272318789168737,
      "grad_norm": 0.1684616208076477,
      "learning_rate": 2.2492047951114248e-05,
      "loss": 0.3428,
      "num_tokens": 1139211263.0,
      "step": 1799
    },
    {
      "epoch": 0.21284143313231643,
      "grad_norm": 0.17923472821712494,
      "learning_rate": 2.246344429828869e-05,
      "loss": 0.3602,
      "num_tokens": 1139850606.0,
      "step": 1800
    },
    {
      "epoch": 0.2129596783729455,
      "grad_norm": 0.1479499489068985,
      "learning_rate": 2.2434854588422596e-05,
      "loss": 0.3212,
      "num_tokens": 1140482025.0,
      "step": 1801
    },
    {
      "epoch": 0.21307792361357455,
      "grad_norm": 0.15312820672988892,
      "learning_rate": 2.2406278859348544e-05,
      "loss": 0.3394,
      "num_tokens": 1141111370.0,
      "step": 1802
    },
    {
      "epoch": 0.21319616885420362,
      "grad_norm": 0.1724095493555069,
      "learning_rate": 2.237771714888062e-05,
      "loss": 0.3533,
      "num_tokens": 1141741981.0,
      "step": 1803
    },
    {
      "epoch": 0.21331441409483268,
      "grad_norm": 0.17572695016860962,
      "learning_rate": 2.2349169494814382e-05,
      "loss": 0.369,
      "num_tokens": 1142381483.0,
      "step": 1804
    },
    {
      "epoch": 0.21343265933546174,
      "grad_norm": 0.14945538341999054,
      "learning_rate": 2.232063593492673e-05,
      "loss": 0.3266,
      "num_tokens": 1143015721.0,
      "step": 1805
    },
    {
      "epoch": 0.2135509045760908,
      "grad_norm": 0.15192392468452454,
      "learning_rate": 2.2292116506975985e-05,
      "loss": 0.3078,
      "num_tokens": 1143647685.0,
      "step": 1806
    },
    {
      "epoch": 0.2136691498167199,
      "grad_norm": 0.1759961098432541,
      "learning_rate": 2.2263611248701712e-05,
      "loss": 0.3452,
      "num_tokens": 1144279423.0,
      "step": 1807
    },
    {
      "epoch": 0.21378739505734895,
      "grad_norm": 0.1545567512512207,
      "learning_rate": 2.2235120197824736e-05,
      "loss": 0.3535,
      "num_tokens": 1144914752.0,
      "step": 1808
    },
    {
      "epoch": 0.213905640297978,
      "grad_norm": 0.16601650416851044,
      "learning_rate": 2.2206643392047105e-05,
      "loss": 0.373,
      "num_tokens": 1145548289.0,
      "step": 1809
    },
    {
      "epoch": 0.21402388553860707,
      "grad_norm": 0.17092914879322052,
      "learning_rate": 2.2178180869051998e-05,
      "loss": 0.3989,
      "num_tokens": 1146177907.0,
      "step": 1810
    },
    {
      "epoch": 0.21414213077923613,
      "grad_norm": 0.173860102891922,
      "learning_rate": 2.2149732666503694e-05,
      "loss": 0.3462,
      "num_tokens": 1146811669.0,
      "step": 1811
    },
    {
      "epoch": 0.2142603760198652,
      "grad_norm": 0.15742754936218262,
      "learning_rate": 2.2121298822047527e-05,
      "loss": 0.354,
      "num_tokens": 1147450564.0,
      "step": 1812
    },
    {
      "epoch": 0.21437862126049426,
      "grad_norm": 0.15500089526176453,
      "learning_rate": 2.209287937330982e-05,
      "loss": 0.3549,
      "num_tokens": 1148081304.0,
      "step": 1813
    },
    {
      "epoch": 0.21449686650112332,
      "grad_norm": 0.17439618706703186,
      "learning_rate": 2.206447435789787e-05,
      "loss": 0.3861,
      "num_tokens": 1148714735.0,
      "step": 1814
    },
    {
      "epoch": 0.2146151117417524,
      "grad_norm": 0.1759570837020874,
      "learning_rate": 2.2036083813399845e-05,
      "loss": 0.3825,
      "num_tokens": 1149346233.0,
      "step": 1815
    },
    {
      "epoch": 0.21473335698238147,
      "grad_norm": 0.15746824443340302,
      "learning_rate": 2.2007707777384788e-05,
      "loss": 0.3292,
      "num_tokens": 1149980039.0,
      "step": 1816
    },
    {
      "epoch": 0.21485160222301053,
      "grad_norm": 0.17177729308605194,
      "learning_rate": 2.1979346287402532e-05,
      "loss": 0.3406,
      "num_tokens": 1150610366.0,
      "step": 1817
    },
    {
      "epoch": 0.2149698474636396,
      "grad_norm": 0.1686304658651352,
      "learning_rate": 2.1950999380983664e-05,
      "loss": 0.3788,
      "num_tokens": 1151249431.0,
      "step": 1818
    },
    {
      "epoch": 0.21508809270426865,
      "grad_norm": 0.16297440230846405,
      "learning_rate": 2.1922667095639465e-05,
      "loss": 0.3473,
      "num_tokens": 1151882110.0,
      "step": 1819
    },
    {
      "epoch": 0.2152063379448977,
      "grad_norm": 0.15822143852710724,
      "learning_rate": 2.189434946886187e-05,
      "loss": 0.3684,
      "num_tokens": 1152521530.0,
      "step": 1820
    },
    {
      "epoch": 0.21532458318552677,
      "grad_norm": 0.17296434938907623,
      "learning_rate": 2.186604653812344e-05,
      "loss": 0.3491,
      "num_tokens": 1153156986.0,
      "step": 1821
    },
    {
      "epoch": 0.21544282842615584,
      "grad_norm": 0.15859577059745789,
      "learning_rate": 2.1837758340877247e-05,
      "loss": 0.3857,
      "num_tokens": 1153794759.0,
      "step": 1822
    },
    {
      "epoch": 0.21556107366678492,
      "grad_norm": 0.14899693429470062,
      "learning_rate": 2.1809484914556904e-05,
      "loss": 0.322,
      "num_tokens": 1154433046.0,
      "step": 1823
    },
    {
      "epoch": 0.21567931890741399,
      "grad_norm": 0.15135660767555237,
      "learning_rate": 2.1781226296576466e-05,
      "loss": 0.328,
      "num_tokens": 1155067168.0,
      "step": 1824
    },
    {
      "epoch": 0.21579756414804305,
      "grad_norm": 0.17024415731430054,
      "learning_rate": 2.1752982524330366e-05,
      "loss": 0.3625,
      "num_tokens": 1155703377.0,
      "step": 1825
    },
    {
      "epoch": 0.2159158093886721,
      "grad_norm": 0.16312475502490997,
      "learning_rate": 2.172475363519343e-05,
      "loss": 0.3427,
      "num_tokens": 1156333194.0,
      "step": 1826
    },
    {
      "epoch": 0.21603405462930117,
      "grad_norm": 0.16713650524616241,
      "learning_rate": 2.169653966652078e-05,
      "loss": 0.3487,
      "num_tokens": 1156965595.0,
      "step": 1827
    },
    {
      "epoch": 0.21615229986993023,
      "grad_norm": 0.1780661642551422,
      "learning_rate": 2.166834065564777e-05,
      "loss": 0.3925,
      "num_tokens": 1157595055.0,
      "step": 1828
    },
    {
      "epoch": 0.2162705451105593,
      "grad_norm": 0.17596584558486938,
      "learning_rate": 2.1640156639889993e-05,
      "loss": 0.397,
      "num_tokens": 1158227500.0,
      "step": 1829
    },
    {
      "epoch": 0.21638879035118835,
      "grad_norm": 0.16619384288787842,
      "learning_rate": 2.1611987656543183e-05,
      "loss": 0.3714,
      "num_tokens": 1158863732.0,
      "step": 1830
    },
    {
      "epoch": 0.21650703559181744,
      "grad_norm": 0.18381381034851074,
      "learning_rate": 2.1583833742883174e-05,
      "loss": 0.3617,
      "num_tokens": 1159489840.0,
      "step": 1831
    },
    {
      "epoch": 0.2166252808324465,
      "grad_norm": 0.17254766821861267,
      "learning_rate": 2.155569493616587e-05,
      "loss": 0.3697,
      "num_tokens": 1160128980.0,
      "step": 1832
    },
    {
      "epoch": 0.21674352607307557,
      "grad_norm": 0.1754225194454193,
      "learning_rate": 2.1527571273627178e-05,
      "loss": 0.3441,
      "num_tokens": 1160764145.0,
      "step": 1833
    },
    {
      "epoch": 0.21686177131370463,
      "grad_norm": 0.17318950593471527,
      "learning_rate": 2.149946279248298e-05,
      "loss": 0.3917,
      "num_tokens": 1161400464.0,
      "step": 1834
    },
    {
      "epoch": 0.2169800165543337,
      "grad_norm": 0.17255404591560364,
      "learning_rate": 2.147136952992905e-05,
      "loss": 0.3627,
      "num_tokens": 1162027629.0,
      "step": 1835
    },
    {
      "epoch": 0.21709826179496275,
      "grad_norm": 0.17934633791446686,
      "learning_rate": 2.1443291523141023e-05,
      "loss": 0.3925,
      "num_tokens": 1162661769.0,
      "step": 1836
    },
    {
      "epoch": 0.2172165070355918,
      "grad_norm": 0.1712150275707245,
      "learning_rate": 2.1415228809274368e-05,
      "loss": 0.3511,
      "num_tokens": 1163294488.0,
      "step": 1837
    },
    {
      "epoch": 0.21733475227622087,
      "grad_norm": 0.16180448234081268,
      "learning_rate": 2.1387181425464276e-05,
      "loss": 0.3528,
      "num_tokens": 1163907989.0,
      "step": 1838
    },
    {
      "epoch": 0.21745299751684993,
      "grad_norm": 0.16935303807258606,
      "learning_rate": 2.1359149408825705e-05,
      "loss": 0.3569,
      "num_tokens": 1164541061.0,
      "step": 1839
    },
    {
      "epoch": 0.21757124275747902,
      "grad_norm": 0.16335564851760864,
      "learning_rate": 2.1331132796453243e-05,
      "loss": 0.3667,
      "num_tokens": 1165172101.0,
      "step": 1840
    },
    {
      "epoch": 0.21768948799810808,
      "grad_norm": 0.15852032601833344,
      "learning_rate": 2.1303131625421096e-05,
      "loss": 0.313,
      "num_tokens": 1165801010.0,
      "step": 1841
    },
    {
      "epoch": 0.21780773323873714,
      "grad_norm": 0.15676049888134003,
      "learning_rate": 2.1275145932783048e-05,
      "loss": 0.3358,
      "num_tokens": 1166401559.0,
      "step": 1842
    },
    {
      "epoch": 0.2179259784793662,
      "grad_norm": 0.14861942827701569,
      "learning_rate": 2.1247175755572366e-05,
      "loss": 0.3039,
      "num_tokens": 1167038615.0,
      "step": 1843
    },
    {
      "epoch": 0.21804422371999527,
      "grad_norm": 0.17299704253673553,
      "learning_rate": 2.1219221130801866e-05,
      "loss": 0.371,
      "num_tokens": 1167676463.0,
      "step": 1844
    },
    {
      "epoch": 0.21816246896062433,
      "grad_norm": 0.17603974044322968,
      "learning_rate": 2.1191282095463685e-05,
      "loss": 0.3728,
      "num_tokens": 1168309031.0,
      "step": 1845
    },
    {
      "epoch": 0.2182807142012534,
      "grad_norm": 0.17879034578800201,
      "learning_rate": 2.1163358686529412e-05,
      "loss": 0.3727,
      "num_tokens": 1168935781.0,
      "step": 1846
    },
    {
      "epoch": 0.21839895944188245,
      "grad_norm": 0.17116892337799072,
      "learning_rate": 2.113545094094989e-05,
      "loss": 0.3907,
      "num_tokens": 1169567656.0,
      "step": 1847
    },
    {
      "epoch": 0.21851720468251154,
      "grad_norm": 0.17895719408988953,
      "learning_rate": 2.1107558895655284e-05,
      "loss": 0.3965,
      "num_tokens": 1170206328.0,
      "step": 1848
    },
    {
      "epoch": 0.2186354499231406,
      "grad_norm": 0.15772230923175812,
      "learning_rate": 2.107968258755497e-05,
      "loss": 0.3806,
      "num_tokens": 1170845768.0,
      "step": 1849
    },
    {
      "epoch": 0.21875369516376966,
      "grad_norm": 0.1658557802438736,
      "learning_rate": 2.1051822053537494e-05,
      "loss": 0.3329,
      "num_tokens": 1171477604.0,
      "step": 1850
    },
    {
      "epoch": 0.21887194040439872,
      "grad_norm": 0.1760043352842331,
      "learning_rate": 2.1023977330470522e-05,
      "loss": 0.4029,
      "num_tokens": 1172110395.0,
      "step": 1851
    },
    {
      "epoch": 0.21899018564502779,
      "grad_norm": 0.16340604424476624,
      "learning_rate": 2.0996148455200813e-05,
      "loss": 0.3348,
      "num_tokens": 1172747360.0,
      "step": 1852
    },
    {
      "epoch": 0.21910843088565685,
      "grad_norm": 0.15813681483268738,
      "learning_rate": 2.0968335464554125e-05,
      "loss": 0.3893,
      "num_tokens": 1173383816.0,
      "step": 1853
    },
    {
      "epoch": 0.2192266761262859,
      "grad_norm": 0.1610841453075409,
      "learning_rate": 2.0940538395335255e-05,
      "loss": 0.3427,
      "num_tokens": 1174017986.0,
      "step": 1854
    },
    {
      "epoch": 0.21934492136691497,
      "grad_norm": 0.16629956662654877,
      "learning_rate": 2.091275728432787e-05,
      "loss": 0.327,
      "num_tokens": 1174648098.0,
      "step": 1855
    },
    {
      "epoch": 0.21946316660754406,
      "grad_norm": 0.16128353774547577,
      "learning_rate": 2.088499216829455e-05,
      "loss": 0.3558,
      "num_tokens": 1175285162.0,
      "step": 1856
    },
    {
      "epoch": 0.21958141184817312,
      "grad_norm": 0.16071675717830658,
      "learning_rate": 2.085724308397668e-05,
      "loss": 0.3922,
      "num_tokens": 1175918416.0,
      "step": 1857
    },
    {
      "epoch": 0.21969965708880218,
      "grad_norm": 0.14863473176956177,
      "learning_rate": 2.0829510068094495e-05,
      "loss": 0.3345,
      "num_tokens": 1176553703.0,
      "step": 1858
    },
    {
      "epoch": 0.21981790232943124,
      "grad_norm": 0.15749160945415497,
      "learning_rate": 2.0801793157346902e-05,
      "loss": 0.3516,
      "num_tokens": 1177188385.0,
      "step": 1859
    },
    {
      "epoch": 0.2199361475700603,
      "grad_norm": 0.14628109335899353,
      "learning_rate": 2.0774092388411518e-05,
      "loss": 0.3319,
      "num_tokens": 1177825789.0,
      "step": 1860
    },
    {
      "epoch": 0.22005439281068936,
      "grad_norm": 0.1442510187625885,
      "learning_rate": 2.0746407797944604e-05,
      "loss": 0.3596,
      "num_tokens": 1178461722.0,
      "step": 1861
    },
    {
      "epoch": 0.22017263805131843,
      "grad_norm": 0.15561442077159882,
      "learning_rate": 2.0718739422580996e-05,
      "loss": 0.3587,
      "num_tokens": 1179095968.0,
      "step": 1862
    },
    {
      "epoch": 0.2202908832919475,
      "grad_norm": 0.14741671085357666,
      "learning_rate": 2.069108729893411e-05,
      "loss": 0.3306,
      "num_tokens": 1179730603.0,
      "step": 1863
    },
    {
      "epoch": 0.22040912853257658,
      "grad_norm": 0.15964746475219727,
      "learning_rate": 2.066345146359581e-05,
      "loss": 0.3892,
      "num_tokens": 1180365636.0,
      "step": 1864
    },
    {
      "epoch": 0.22052737377320564,
      "grad_norm": 0.1483248472213745,
      "learning_rate": 2.0635831953136443e-05,
      "loss": 0.3476,
      "num_tokens": 1181002112.0,
      "step": 1865
    },
    {
      "epoch": 0.2206456190138347,
      "grad_norm": 0.1502329409122467,
      "learning_rate": 2.060822880410472e-05,
      "loss": 0.3226,
      "num_tokens": 1181635764.0,
      "step": 1866
    },
    {
      "epoch": 0.22076386425446376,
      "grad_norm": 0.1646731048822403,
      "learning_rate": 2.0580642053027724e-05,
      "loss": 0.3723,
      "num_tokens": 1182274273.0,
      "step": 1867
    },
    {
      "epoch": 0.22088210949509282,
      "grad_norm": 0.16638028621673584,
      "learning_rate": 2.0553071736410846e-05,
      "loss": 0.351,
      "num_tokens": 1182907997.0,
      "step": 1868
    },
    {
      "epoch": 0.22100035473572188,
      "grad_norm": 0.164053276181221,
      "learning_rate": 2.0525517890737708e-05,
      "loss": 0.3696,
      "num_tokens": 1183539711.0,
      "step": 1869
    },
    {
      "epoch": 0.22111859997635094,
      "grad_norm": 0.1691984087228775,
      "learning_rate": 2.0497980552470146e-05,
      "loss": 0.3387,
      "num_tokens": 1184178482.0,
      "step": 1870
    },
    {
      "epoch": 0.22123684521698,
      "grad_norm": 0.15557564795017242,
      "learning_rate": 2.047045975804815e-05,
      "loss": 0.3557,
      "num_tokens": 1184815893.0,
      "step": 1871
    },
    {
      "epoch": 0.2213550904576091,
      "grad_norm": 0.15779171884059906,
      "learning_rate": 2.04429555438898e-05,
      "loss": 0.3745,
      "num_tokens": 1185451168.0,
      "step": 1872
    },
    {
      "epoch": 0.22147333569823816,
      "grad_norm": 0.17602765560150146,
      "learning_rate": 2.041546794639129e-05,
      "loss": 0.3826,
      "num_tokens": 1186086488.0,
      "step": 1873
    },
    {
      "epoch": 0.22159158093886722,
      "grad_norm": 0.155890092253685,
      "learning_rate": 2.038799700192677e-05,
      "loss": 0.3807,
      "num_tokens": 1186722050.0,
      "step": 1874
    },
    {
      "epoch": 0.22170982617949628,
      "grad_norm": 0.16406124830245972,
      "learning_rate": 2.0360542746848368e-05,
      "loss": 0.3492,
      "num_tokens": 1187352361.0,
      "step": 1875
    },
    {
      "epoch": 0.22182807142012534,
      "grad_norm": 0.15780766308307648,
      "learning_rate": 2.033310521748614e-05,
      "loss": 0.3253,
      "num_tokens": 1187987963.0,
      "step": 1876
    },
    {
      "epoch": 0.2219463166607544,
      "grad_norm": 0.15500418841838837,
      "learning_rate": 2.030568445014798e-05,
      "loss": 0.3191,
      "num_tokens": 1188622119.0,
      "step": 1877
    },
    {
      "epoch": 0.22206456190138346,
      "grad_norm": 0.15499471127986908,
      "learning_rate": 2.0278280481119655e-05,
      "loss": 0.3668,
      "num_tokens": 1189257899.0,
      "step": 1878
    },
    {
      "epoch": 0.22218280714201252,
      "grad_norm": 0.15043438971042633,
      "learning_rate": 2.0250893346664646e-05,
      "loss": 0.3691,
      "num_tokens": 1189890204.0,
      "step": 1879
    },
    {
      "epoch": 0.2223010523826416,
      "grad_norm": 0.15676124393939972,
      "learning_rate": 2.022352308302418e-05,
      "loss": 0.3604,
      "num_tokens": 1190523164.0,
      "step": 1880
    },
    {
      "epoch": 0.22241929762327067,
      "grad_norm": 0.15230387449264526,
      "learning_rate": 2.0196169726417173e-05,
      "loss": 0.3347,
      "num_tokens": 1191157689.0,
      "step": 1881
    },
    {
      "epoch": 0.22253754286389973,
      "grad_norm": 0.1457720249891281,
      "learning_rate": 2.016883331304013e-05,
      "loss": 0.2949,
      "num_tokens": 1191786401.0,
      "step": 1882
    },
    {
      "epoch": 0.2226557881045288,
      "grad_norm": 0.1585022658109665,
      "learning_rate": 2.014151387906718e-05,
      "loss": 0.3152,
      "num_tokens": 1192420292.0,
      "step": 1883
    },
    {
      "epoch": 0.22277403334515786,
      "grad_norm": 0.17930035293102264,
      "learning_rate": 2.011421146064996e-05,
      "loss": 0.3671,
      "num_tokens": 1193050347.0,
      "step": 1884
    },
    {
      "epoch": 0.22289227858578692,
      "grad_norm": 0.14422306418418884,
      "learning_rate": 2.0086926093917587e-05,
      "loss": 0.3369,
      "num_tokens": 1193683618.0,
      "step": 1885
    },
    {
      "epoch": 0.22301052382641598,
      "grad_norm": 0.1651952862739563,
      "learning_rate": 2.0059657814976615e-05,
      "loss": 0.3584,
      "num_tokens": 1194316482.0,
      "step": 1886
    },
    {
      "epoch": 0.22312876906704504,
      "grad_norm": 0.1569311022758484,
      "learning_rate": 2.003240665991099e-05,
      "loss": 0.3566,
      "num_tokens": 1194953767.0,
      "step": 1887
    },
    {
      "epoch": 0.22324701430767413,
      "grad_norm": 0.16590744256973267,
      "learning_rate": 2.0005172664782014e-05,
      "loss": 0.3481,
      "num_tokens": 1195589094.0,
      "step": 1888
    },
    {
      "epoch": 0.2233652595483032,
      "grad_norm": 0.168272465467453,
      "learning_rate": 1.9977955865628256e-05,
      "loss": 0.4074,
      "num_tokens": 1196219735.0,
      "step": 1889
    },
    {
      "epoch": 0.22348350478893225,
      "grad_norm": 0.15438778698444366,
      "learning_rate": 1.9950756298465535e-05,
      "loss": 0.3393,
      "num_tokens": 1196849912.0,
      "step": 1890
    },
    {
      "epoch": 0.2236017500295613,
      "grad_norm": 0.1607561856508255,
      "learning_rate": 1.9923573999286865e-05,
      "loss": 0.368,
      "num_tokens": 1197483986.0,
      "step": 1891
    },
    {
      "epoch": 0.22371999527019037,
      "grad_norm": 0.1599637120962143,
      "learning_rate": 1.989640900406244e-05,
      "loss": 0.3834,
      "num_tokens": 1198120332.0,
      "step": 1892
    },
    {
      "epoch": 0.22383824051081944,
      "grad_norm": 0.14852549135684967,
      "learning_rate": 1.986926134873951e-05,
      "loss": 0.3055,
      "num_tokens": 1198750956.0,
      "step": 1893
    },
    {
      "epoch": 0.2239564857514485,
      "grad_norm": 0.17405691742897034,
      "learning_rate": 1.9842131069242408e-05,
      "loss": 0.4235,
      "num_tokens": 1199388319.0,
      "step": 1894
    },
    {
      "epoch": 0.22407473099207756,
      "grad_norm": 0.15194599330425262,
      "learning_rate": 1.9815018201472474e-05,
      "loss": 0.3375,
      "num_tokens": 1200026396.0,
      "step": 1895
    },
    {
      "epoch": 0.22419297623270662,
      "grad_norm": 0.14832772314548492,
      "learning_rate": 1.978792278130797e-05,
      "loss": 0.3393,
      "num_tokens": 1200653214.0,
      "step": 1896
    },
    {
      "epoch": 0.2243112214733357,
      "grad_norm": 0.16637760400772095,
      "learning_rate": 1.976084484460413e-05,
      "loss": 0.3649,
      "num_tokens": 1201283854.0,
      "step": 1897
    },
    {
      "epoch": 0.22442946671396477,
      "grad_norm": 0.14979857206344604,
      "learning_rate": 1.9733784427193017e-05,
      "loss": 0.3534,
      "num_tokens": 1201916015.0,
      "step": 1898
    },
    {
      "epoch": 0.22454771195459383,
      "grad_norm": 0.16878588497638702,
      "learning_rate": 1.970674156488351e-05,
      "loss": 0.3841,
      "num_tokens": 1202554692.0,
      "step": 1899
    },
    {
      "epoch": 0.2246659571952229,
      "grad_norm": 0.14891672134399414,
      "learning_rate": 1.9679716293461257e-05,
      "loss": 0.3472,
      "num_tokens": 1203191938.0,
      "step": 1900
    },
    {
      "epoch": 0.22478420243585195,
      "grad_norm": 0.15611931681632996,
      "learning_rate": 1.965270864868863e-05,
      "loss": 0.3515,
      "num_tokens": 1203824362.0,
      "step": 1901
    },
    {
      "epoch": 0.22490244767648102,
      "grad_norm": 0.1861187368631363,
      "learning_rate": 1.9625718666304702e-05,
      "loss": 0.3427,
      "num_tokens": 1204462936.0,
      "step": 1902
    },
    {
      "epoch": 0.22502069291711008,
      "grad_norm": 0.16823303699493408,
      "learning_rate": 1.9598746382025132e-05,
      "loss": 0.3585,
      "num_tokens": 1205095441.0,
      "step": 1903
    },
    {
      "epoch": 0.22513893815773914,
      "grad_norm": 0.14871849119663239,
      "learning_rate": 1.957179183154219e-05,
      "loss": 0.3145,
      "num_tokens": 1205729677.0,
      "step": 1904
    },
    {
      "epoch": 0.22525718339836823,
      "grad_norm": 0.1572839319705963,
      "learning_rate": 1.954485505052465e-05,
      "loss": 0.3257,
      "num_tokens": 1206366321.0,
      "step": 1905
    },
    {
      "epoch": 0.2253754286389973,
      "grad_norm": 0.1735132485628128,
      "learning_rate": 1.95179360746178e-05,
      "loss": 0.3594,
      "num_tokens": 1207005752.0,
      "step": 1906
    },
    {
      "epoch": 0.22549367387962635,
      "grad_norm": 0.16872599720954895,
      "learning_rate": 1.9491034939443357e-05,
      "loss": 0.3981,
      "num_tokens": 1207633804.0,
      "step": 1907
    },
    {
      "epoch": 0.2256119191202554,
      "grad_norm": 0.16365472972393036,
      "learning_rate": 1.9464151680599428e-05,
      "loss": 0.3514,
      "num_tokens": 1208270549.0,
      "step": 1908
    },
    {
      "epoch": 0.22573016436088447,
      "grad_norm": 0.15993882715702057,
      "learning_rate": 1.9437286333660467e-05,
      "loss": 0.3293,
      "num_tokens": 1208909763.0,
      "step": 1909
    },
    {
      "epoch": 0.22584840960151353,
      "grad_norm": 0.14977790415287018,
      "learning_rate": 1.941043893417721e-05,
      "loss": 0.3612,
      "num_tokens": 1209548729.0,
      "step": 1910
    },
    {
      "epoch": 0.2259666548421426,
      "grad_norm": 0.17881079018115997,
      "learning_rate": 1.938360951767666e-05,
      "loss": 0.3851,
      "num_tokens": 1210184584.0,
      "step": 1911
    },
    {
      "epoch": 0.22608490008277166,
      "grad_norm": 0.16349364817142487,
      "learning_rate": 1.935679811966203e-05,
      "loss": 0.3471,
      "num_tokens": 1210820281.0,
      "step": 1912
    },
    {
      "epoch": 0.22620314532340074,
      "grad_norm": 0.16041794419288635,
      "learning_rate": 1.9330004775612675e-05,
      "loss": 0.3604,
      "num_tokens": 1211459450.0,
      "step": 1913
    },
    {
      "epoch": 0.2263213905640298,
      "grad_norm": 0.15849412977695465,
      "learning_rate": 1.9303229520984058e-05,
      "loss": 0.3428,
      "num_tokens": 1212089281.0,
      "step": 1914
    },
    {
      "epoch": 0.22643963580465887,
      "grad_norm": 0.16442883014678955,
      "learning_rate": 1.9276472391207707e-05,
      "loss": 0.3337,
      "num_tokens": 1212723355.0,
      "step": 1915
    },
    {
      "epoch": 0.22655788104528793,
      "grad_norm": 0.17862072587013245,
      "learning_rate": 1.9249733421691158e-05,
      "loss": 0.3599,
      "num_tokens": 1213354384.0,
      "step": 1916
    },
    {
      "epoch": 0.226676126285917,
      "grad_norm": 0.15677925944328308,
      "learning_rate": 1.9223012647817955e-05,
      "loss": 0.3329,
      "num_tokens": 1213990075.0,
      "step": 1917
    },
    {
      "epoch": 0.22679437152654605,
      "grad_norm": 0.16935351490974426,
      "learning_rate": 1.9196310104947513e-05,
      "loss": 0.3831,
      "num_tokens": 1214629125.0,
      "step": 1918
    },
    {
      "epoch": 0.2269126167671751,
      "grad_norm": 0.15546615421772003,
      "learning_rate": 1.9169625828415145e-05,
      "loss": 0.345,
      "num_tokens": 1215264589.0,
      "step": 1919
    },
    {
      "epoch": 0.22703086200780417,
      "grad_norm": 0.17183943092823029,
      "learning_rate": 1.9142959853531995e-05,
      "loss": 0.3633,
      "num_tokens": 1215902900.0,
      "step": 1920
    },
    {
      "epoch": 0.22714910724843326,
      "grad_norm": 0.15895579755306244,
      "learning_rate": 1.9116312215584967e-05,
      "loss": 0.3658,
      "num_tokens": 1216535762.0,
      "step": 1921
    },
    {
      "epoch": 0.22726735248906232,
      "grad_norm": 0.17026469111442566,
      "learning_rate": 1.9089682949836745e-05,
      "loss": 0.361,
      "num_tokens": 1217171423.0,
      "step": 1922
    },
    {
      "epoch": 0.22738559772969139,
      "grad_norm": 0.1577785164117813,
      "learning_rate": 1.9063072091525654e-05,
      "loss": 0.3583,
      "num_tokens": 1217808427.0,
      "step": 1923
    },
    {
      "epoch": 0.22750384297032045,
      "grad_norm": 0.16298867762088776,
      "learning_rate": 1.9036479675865684e-05,
      "loss": 0.3772,
      "num_tokens": 1218442317.0,
      "step": 1924
    },
    {
      "epoch": 0.2276220882109495,
      "grad_norm": 0.1716330200433731,
      "learning_rate": 1.900990573804641e-05,
      "loss": 0.3368,
      "num_tokens": 1219075224.0,
      "step": 1925
    },
    {
      "epoch": 0.22774033345157857,
      "grad_norm": 0.1755548119544983,
      "learning_rate": 1.898335031323297e-05,
      "loss": 0.3707,
      "num_tokens": 1219706682.0,
      "step": 1926
    },
    {
      "epoch": 0.22785857869220763,
      "grad_norm": 0.1405853033065796,
      "learning_rate": 1.8956813436565992e-05,
      "loss": 0.3538,
      "num_tokens": 1220346000.0,
      "step": 1927
    },
    {
      "epoch": 0.2279768239328367,
      "grad_norm": 0.17048880457878113,
      "learning_rate": 1.893029514316156e-05,
      "loss": 0.3885,
      "num_tokens": 1220976247.0,
      "step": 1928
    },
    {
      "epoch": 0.22809506917346578,
      "grad_norm": 0.15518951416015625,
      "learning_rate": 1.8903795468111172e-05,
      "loss": 0.3594,
      "num_tokens": 1221609494.0,
      "step": 1929
    },
    {
      "epoch": 0.22821331441409484,
      "grad_norm": 0.15728840231895447,
      "learning_rate": 1.8877314446481666e-05,
      "loss": 0.3474,
      "num_tokens": 1222243620.0,
      "step": 1930
    },
    {
      "epoch": 0.2283315596547239,
      "grad_norm": 0.16775105893611908,
      "learning_rate": 1.8850852113315253e-05,
      "loss": 0.3665,
      "num_tokens": 1222872003.0,
      "step": 1931
    },
    {
      "epoch": 0.22844980489535296,
      "grad_norm": 0.15760478377342224,
      "learning_rate": 1.8824408503629342e-05,
      "loss": 0.3556,
      "num_tokens": 1223509301.0,
      "step": 1932
    },
    {
      "epoch": 0.22856805013598203,
      "grad_norm": 0.15447109937667847,
      "learning_rate": 1.879798365241662e-05,
      "loss": 0.3195,
      "num_tokens": 1224141171.0,
      "step": 1933
    },
    {
      "epoch": 0.2286862953766111,
      "grad_norm": 0.1555739790201187,
      "learning_rate": 1.8771577594644917e-05,
      "loss": 0.3644,
      "num_tokens": 1224777797.0,
      "step": 1934
    },
    {
      "epoch": 0.22880454061724015,
      "grad_norm": 0.14799030125141144,
      "learning_rate": 1.8745190365257206e-05,
      "loss": 0.3647,
      "num_tokens": 1225410917.0,
      "step": 1935
    },
    {
      "epoch": 0.2289227858578692,
      "grad_norm": 0.1485302895307541,
      "learning_rate": 1.871882199917156e-05,
      "loss": 0.3351,
      "num_tokens": 1226044770.0,
      "step": 1936
    },
    {
      "epoch": 0.2290410310984983,
      "grad_norm": 0.17278270423412323,
      "learning_rate": 1.869247253128107e-05,
      "loss": 0.3745,
      "num_tokens": 1226684345.0,
      "step": 1937
    },
    {
      "epoch": 0.22915927633912736,
      "grad_norm": 0.15906833112239838,
      "learning_rate": 1.866614199645382e-05,
      "loss": 0.3528,
      "num_tokens": 1227323937.0,
      "step": 1938
    },
    {
      "epoch": 0.22927752157975642,
      "grad_norm": 0.15691624581813812,
      "learning_rate": 1.863983042953285e-05,
      "loss": 0.3271,
      "num_tokens": 1227957801.0,
      "step": 1939
    },
    {
      "epoch": 0.22939576682038548,
      "grad_norm": 0.16830727458000183,
      "learning_rate": 1.861353786533608e-05,
      "loss": 0.3724,
      "num_tokens": 1228594724.0,
      "step": 1940
    },
    {
      "epoch": 0.22951401206101454,
      "grad_norm": 0.16309845447540283,
      "learning_rate": 1.8587264338656328e-05,
      "loss": 0.3578,
      "num_tokens": 1229220164.0,
      "step": 1941
    },
    {
      "epoch": 0.2296322573016436,
      "grad_norm": 0.1481633484363556,
      "learning_rate": 1.8561009884261172e-05,
      "loss": 0.3372,
      "num_tokens": 1229851524.0,
      "step": 1942
    },
    {
      "epoch": 0.22975050254227267,
      "grad_norm": 0.16284671425819397,
      "learning_rate": 1.8534774536892972e-05,
      "loss": 0.356,
      "num_tokens": 1230484467.0,
      "step": 1943
    },
    {
      "epoch": 0.22986874778290173,
      "grad_norm": 0.15934152901172638,
      "learning_rate": 1.85085583312688e-05,
      "loss": 0.3263,
      "num_tokens": 1231116359.0,
      "step": 1944
    },
    {
      "epoch": 0.2299869930235308,
      "grad_norm": 0.1626678705215454,
      "learning_rate": 1.8482361302080397e-05,
      "loss": 0.3768,
      "num_tokens": 1231753859.0,
      "step": 1945
    },
    {
      "epoch": 0.23010523826415988,
      "grad_norm": 0.16223514080047607,
      "learning_rate": 1.8456183483994143e-05,
      "loss": 0.3259,
      "num_tokens": 1232388378.0,
      "step": 1946
    },
    {
      "epoch": 0.23022348350478894,
      "grad_norm": 0.15437988936901093,
      "learning_rate": 1.8430024911650977e-05,
      "loss": 0.3622,
      "num_tokens": 1233026448.0,
      "step": 1947
    },
    {
      "epoch": 0.230341728745418,
      "grad_norm": 0.16027230024337769,
      "learning_rate": 1.840388561966638e-05,
      "loss": 0.3378,
      "num_tokens": 1233660542.0,
      "step": 1948
    },
    {
      "epoch": 0.23045997398604706,
      "grad_norm": 0.15265357494354248,
      "learning_rate": 1.837776564263031e-05,
      "loss": 0.3416,
      "num_tokens": 1234287661.0,
      "step": 1949
    },
    {
      "epoch": 0.23057821922667612,
      "grad_norm": 0.15387766063213348,
      "learning_rate": 1.835166501510717e-05,
      "loss": 0.3404,
      "num_tokens": 1234922170.0,
      "step": 1950
    },
    {
      "epoch": 0.23069646446730518,
      "grad_norm": 0.15113648772239685,
      "learning_rate": 1.832558377163578e-05,
      "loss": 0.3178,
      "num_tokens": 1235555317.0,
      "step": 1951
    },
    {
      "epoch": 0.23081470970793425,
      "grad_norm": 0.15565477311611176,
      "learning_rate": 1.829952194672927e-05,
      "loss": 0.311,
      "num_tokens": 1236192199.0,
      "step": 1952
    },
    {
      "epoch": 0.2309329549485633,
      "grad_norm": 0.16680525243282318,
      "learning_rate": 1.8273479574875098e-05,
      "loss": 0.3504,
      "num_tokens": 1236799053.0,
      "step": 1953
    },
    {
      "epoch": 0.2310512001891924,
      "grad_norm": 0.14852072298526764,
      "learning_rate": 1.8247456690534978e-05,
      "loss": 0.3451,
      "num_tokens": 1237432544.0,
      "step": 1954
    },
    {
      "epoch": 0.23116944542982146,
      "grad_norm": 0.17207488417625427,
      "learning_rate": 1.8221453328144822e-05,
      "loss": 0.436,
      "num_tokens": 1238070500.0,
      "step": 1955
    },
    {
      "epoch": 0.23128769067045052,
      "grad_norm": 0.1915634572505951,
      "learning_rate": 1.8195469522114735e-05,
      "loss": 0.351,
      "num_tokens": 1238698864.0,
      "step": 1956
    },
    {
      "epoch": 0.23140593591107958,
      "grad_norm": 0.1548355221748352,
      "learning_rate": 1.816950530682892e-05,
      "loss": 0.3673,
      "num_tokens": 1239326142.0,
      "step": 1957
    },
    {
      "epoch": 0.23152418115170864,
      "grad_norm": 0.15716081857681274,
      "learning_rate": 1.814356071664567e-05,
      "loss": 0.3334,
      "num_tokens": 1239965759.0,
      "step": 1958
    },
    {
      "epoch": 0.2316424263923377,
      "grad_norm": 0.1694408804178238,
      "learning_rate": 1.811763578589728e-05,
      "loss": 0.3517,
      "num_tokens": 1240593937.0,
      "step": 1959
    },
    {
      "epoch": 0.23176067163296676,
      "grad_norm": 0.1577712446451187,
      "learning_rate": 1.8091730548890083e-05,
      "loss": 0.3392,
      "num_tokens": 1241232600.0,
      "step": 1960
    },
    {
      "epoch": 0.23187891687359583,
      "grad_norm": 0.17078039050102234,
      "learning_rate": 1.8065845039904307e-05,
      "loss": 0.3795,
      "num_tokens": 1241868072.0,
      "step": 1961
    },
    {
      "epoch": 0.23199716211422491,
      "grad_norm": 0.16106127202510834,
      "learning_rate": 1.803997929319408e-05,
      "loss": 0.3745,
      "num_tokens": 1242499329.0,
      "step": 1962
    },
    {
      "epoch": 0.23211540735485398,
      "grad_norm": 0.14302292466163635,
      "learning_rate": 1.8014133342987396e-05,
      "loss": 0.321,
      "num_tokens": 1243132947.0,
      "step": 1963
    },
    {
      "epoch": 0.23223365259548304,
      "grad_norm": 0.16162584722042084,
      "learning_rate": 1.7988307223486026e-05,
      "loss": 0.3751,
      "num_tokens": 1243768759.0,
      "step": 1964
    },
    {
      "epoch": 0.2323518978361121,
      "grad_norm": 0.16313178837299347,
      "learning_rate": 1.7962500968865538e-05,
      "loss": 0.3436,
      "num_tokens": 1244397968.0,
      "step": 1965
    },
    {
      "epoch": 0.23247014307674116,
      "grad_norm": 0.16243085265159607,
      "learning_rate": 1.7936714613275173e-05,
      "loss": 0.3467,
      "num_tokens": 1245030355.0,
      "step": 1966
    },
    {
      "epoch": 0.23258838831737022,
      "grad_norm": 0.1638815701007843,
      "learning_rate": 1.791094819083787e-05,
      "loss": 0.3549,
      "num_tokens": 1245652710.0,
      "step": 1967
    },
    {
      "epoch": 0.23270663355799928,
      "grad_norm": 0.15447312593460083,
      "learning_rate": 1.7885201735650167e-05,
      "loss": 0.3206,
      "num_tokens": 1246288548.0,
      "step": 1968
    },
    {
      "epoch": 0.23282487879862834,
      "grad_norm": 0.1560949981212616,
      "learning_rate": 1.7859475281782182e-05,
      "loss": 0.3352,
      "num_tokens": 1246925966.0,
      "step": 1969
    },
    {
      "epoch": 0.23294312403925743,
      "grad_norm": 0.1852051019668579,
      "learning_rate": 1.7833768863277604e-05,
      "loss": 0.3462,
      "num_tokens": 1247554529.0,
      "step": 1970
    },
    {
      "epoch": 0.2330613692798865,
      "grad_norm": 0.16054436564445496,
      "learning_rate": 1.7808082514153553e-05,
      "loss": 0.3926,
      "num_tokens": 1248186600.0,
      "step": 1971
    },
    {
      "epoch": 0.23317961452051555,
      "grad_norm": 0.159676194190979,
      "learning_rate": 1.7782416268400622e-05,
      "loss": 0.3597,
      "num_tokens": 1248802763.0,
      "step": 1972
    },
    {
      "epoch": 0.23329785976114462,
      "grad_norm": 0.17004135251045227,
      "learning_rate": 1.7756770159982806e-05,
      "loss": 0.3076,
      "num_tokens": 1249434773.0,
      "step": 1973
    },
    {
      "epoch": 0.23341610500177368,
      "grad_norm": 0.159078449010849,
      "learning_rate": 1.773114422283742e-05,
      "loss": 0.3524,
      "num_tokens": 1250071859.0,
      "step": 1974
    },
    {
      "epoch": 0.23353435024240274,
      "grad_norm": 0.1678972691297531,
      "learning_rate": 1.770553849087513e-05,
      "loss": 0.3434,
      "num_tokens": 1250709705.0,
      "step": 1975
    },
    {
      "epoch": 0.2336525954830318,
      "grad_norm": 0.16334383189678192,
      "learning_rate": 1.767995299797985e-05,
      "loss": 0.3898,
      "num_tokens": 1251344489.0,
      "step": 1976
    },
    {
      "epoch": 0.23377084072366086,
      "grad_norm": 0.164784237742424,
      "learning_rate": 1.7654387778008686e-05,
      "loss": 0.3489,
      "num_tokens": 1251982650.0,
      "step": 1977
    },
    {
      "epoch": 0.23388908596428995,
      "grad_norm": 0.1682564616203308,
      "learning_rate": 1.7628842864791946e-05,
      "loss": 0.3622,
      "num_tokens": 1252618531.0,
      "step": 1978
    },
    {
      "epoch": 0.234007331204919,
      "grad_norm": 0.1823110282421112,
      "learning_rate": 1.7603318292133045e-05,
      "loss": 0.3435,
      "num_tokens": 1253250959.0,
      "step": 1979
    },
    {
      "epoch": 0.23412557644554807,
      "grad_norm": 0.16742032766342163,
      "learning_rate": 1.7577814093808514e-05,
      "loss": 0.3631,
      "num_tokens": 1253889968.0,
      "step": 1980
    },
    {
      "epoch": 0.23424382168617713,
      "grad_norm": 0.16165918111801147,
      "learning_rate": 1.7552330303567894e-05,
      "loss": 0.3466,
      "num_tokens": 1254520645.0,
      "step": 1981
    },
    {
      "epoch": 0.2343620669268062,
      "grad_norm": 0.16288359463214874,
      "learning_rate": 1.7526866955133725e-05,
      "loss": 0.3784,
      "num_tokens": 1255148745.0,
      "step": 1982
    },
    {
      "epoch": 0.23448031216743526,
      "grad_norm": 0.1556723266839981,
      "learning_rate": 1.75014240822015e-05,
      "loss": 0.3634,
      "num_tokens": 1255784484.0,
      "step": 1983
    },
    {
      "epoch": 0.23459855740806432,
      "grad_norm": 0.16116824746131897,
      "learning_rate": 1.7476001718439606e-05,
      "loss": 0.3831,
      "num_tokens": 1256417769.0,
      "step": 1984
    },
    {
      "epoch": 0.23471680264869338,
      "grad_norm": 0.17379483580589294,
      "learning_rate": 1.7450599897489317e-05,
      "loss": 0.3431,
      "num_tokens": 1257052424.0,
      "step": 1985
    },
    {
      "epoch": 0.23483504788932247,
      "grad_norm": 0.14859539270401,
      "learning_rate": 1.742521865296471e-05,
      "loss": 0.3339,
      "num_tokens": 1257689558.0,
      "step": 1986
    },
    {
      "epoch": 0.23495329312995153,
      "grad_norm": 0.16267679631710052,
      "learning_rate": 1.7399858018452617e-05,
      "loss": 0.3677,
      "num_tokens": 1258323494.0,
      "step": 1987
    },
    {
      "epoch": 0.2350715383705806,
      "grad_norm": 0.16116073727607727,
      "learning_rate": 1.7374518027512616e-05,
      "loss": 0.3364,
      "num_tokens": 1258955356.0,
      "step": 1988
    },
    {
      "epoch": 0.23518978361120965,
      "grad_norm": 0.15993084013462067,
      "learning_rate": 1.734919871367695e-05,
      "loss": 0.3552,
      "num_tokens": 1259588215.0,
      "step": 1989
    },
    {
      "epoch": 0.2353080288518387,
      "grad_norm": 0.16909614205360413,
      "learning_rate": 1.732390011045053e-05,
      "loss": 0.3785,
      "num_tokens": 1260226408.0,
      "step": 1990
    },
    {
      "epoch": 0.23542627409246777,
      "grad_norm": 0.17005151510238647,
      "learning_rate": 1.729862225131083e-05,
      "loss": 0.353,
      "num_tokens": 1260860288.0,
      "step": 1991
    },
    {
      "epoch": 0.23554451933309684,
      "grad_norm": 0.17822960019111633,
      "learning_rate": 1.72733651697079e-05,
      "loss": 0.3646,
      "num_tokens": 1261494410.0,
      "step": 1992
    },
    {
      "epoch": 0.2356627645737259,
      "grad_norm": 0.15992970764636993,
      "learning_rate": 1.724812889906425e-05,
      "loss": 0.3578,
      "num_tokens": 1262129492.0,
      "step": 1993
    },
    {
      "epoch": 0.23578100981435499,
      "grad_norm": 0.1606341302394867,
      "learning_rate": 1.722291347277492e-05,
      "loss": 0.3344,
      "num_tokens": 1262767078.0,
      "step": 1994
    },
    {
      "epoch": 0.23589925505498405,
      "grad_norm": 0.17137299478054047,
      "learning_rate": 1.7197718924207304e-05,
      "loss": 0.3301,
      "num_tokens": 1263402389.0,
      "step": 1995
    },
    {
      "epoch": 0.2360175002956131,
      "grad_norm": 0.18807576596736908,
      "learning_rate": 1.7172545286701207e-05,
      "loss": 0.407,
      "num_tokens": 1264035180.0,
      "step": 1996
    },
    {
      "epoch": 0.23613574553624217,
      "grad_norm": 0.15306997299194336,
      "learning_rate": 1.714739259356874e-05,
      "loss": 0.3203,
      "num_tokens": 1264670288.0,
      "step": 1997
    },
    {
      "epoch": 0.23625399077687123,
      "grad_norm": 0.17509116232395172,
      "learning_rate": 1.71222608780943e-05,
      "loss": 0.3251,
      "num_tokens": 1265305830.0,
      "step": 1998
    },
    {
      "epoch": 0.2363722360175003,
      "grad_norm": 0.1641552448272705,
      "learning_rate": 1.709715017353455e-05,
      "loss": 0.3425,
      "num_tokens": 1265939517.0,
      "step": 1999
    },
    {
      "epoch": 0.23649048125812935,
      "grad_norm": 0.17969214916229248,
      "learning_rate": 1.7072060513118325e-05,
      "loss": 0.3889,
      "num_tokens": 1266577606.0,
      "step": 2000
    },
    {
      "epoch": 0.23660872649875841,
      "grad_norm": 0.17133449018001556,
      "learning_rate": 1.7046991930046615e-05,
      "loss": 0.3745,
      "num_tokens": 1267214198.0,
      "step": 2001
    },
    {
      "epoch": 0.23672697173938748,
      "grad_norm": 0.17307010293006897,
      "learning_rate": 1.7021944457492525e-05,
      "loss": 0.3641,
      "num_tokens": 1267844709.0,
      "step": 2002
    },
    {
      "epoch": 0.23684521698001657,
      "grad_norm": 0.14777572453022003,
      "learning_rate": 1.6996918128601207e-05,
      "loss": 0.3265,
      "num_tokens": 1268478703.0,
      "step": 2003
    },
    {
      "epoch": 0.23696346222064563,
      "grad_norm": 0.1670704334974289,
      "learning_rate": 1.6971912976489874e-05,
      "loss": 0.3539,
      "num_tokens": 1269115700.0,
      "step": 2004
    },
    {
      "epoch": 0.2370817074612747,
      "grad_norm": 0.1588928997516632,
      "learning_rate": 1.6946929034247678e-05,
      "loss": 0.3508,
      "num_tokens": 1269747201.0,
      "step": 2005
    },
    {
      "epoch": 0.23719995270190375,
      "grad_norm": 0.17382748425006866,
      "learning_rate": 1.6921966334935715e-05,
      "loss": 0.3284,
      "num_tokens": 1270382337.0,
      "step": 2006
    },
    {
      "epoch": 0.2373181979425328,
      "grad_norm": 0.1680542528629303,
      "learning_rate": 1.689702491158697e-05,
      "loss": 0.3535,
      "num_tokens": 1271020341.0,
      "step": 2007
    },
    {
      "epoch": 0.23743644318316187,
      "grad_norm": 0.16459088027477264,
      "learning_rate": 1.6872104797206273e-05,
      "loss": 0.3467,
      "num_tokens": 1271652135.0,
      "step": 2008
    },
    {
      "epoch": 0.23755468842379093,
      "grad_norm": 0.14711426198482513,
      "learning_rate": 1.6847206024770276e-05,
      "loss": 0.3425,
      "num_tokens": 1272287068.0,
      "step": 2009
    },
    {
      "epoch": 0.23767293366442,
      "grad_norm": 0.1500907987356186,
      "learning_rate": 1.6822328627227352e-05,
      "loss": 0.3461,
      "num_tokens": 1272924545.0,
      "step": 2010
    },
    {
      "epoch": 0.23779117890504908,
      "grad_norm": 0.17159508168697357,
      "learning_rate": 1.6797472637497623e-05,
      "loss": 0.3872,
      "num_tokens": 1273561738.0,
      "step": 2011
    },
    {
      "epoch": 0.23790942414567814,
      "grad_norm": 0.14653366804122925,
      "learning_rate": 1.6772638088472852e-05,
      "loss": 0.3359,
      "num_tokens": 1274195285.0,
      "step": 2012
    },
    {
      "epoch": 0.2380276693863072,
      "grad_norm": 0.1653970181941986,
      "learning_rate": 1.6747825013016444e-05,
      "loss": 0.3477,
      "num_tokens": 1274817746.0,
      "step": 2013
    },
    {
      "epoch": 0.23814591462693627,
      "grad_norm": 0.15158611536026,
      "learning_rate": 1.672303344396341e-05,
      "loss": 0.2962,
      "num_tokens": 1275435439.0,
      "step": 2014
    },
    {
      "epoch": 0.23826415986756533,
      "grad_norm": 0.1698819398880005,
      "learning_rate": 1.6698263414120277e-05,
      "loss": 0.3778,
      "num_tokens": 1276068813.0,
      "step": 2015
    },
    {
      "epoch": 0.2383824051081944,
      "grad_norm": 0.16204829514026642,
      "learning_rate": 1.667351495626506e-05,
      "loss": 0.3426,
      "num_tokens": 1276698654.0,
      "step": 2016
    },
    {
      "epoch": 0.23850065034882345,
      "grad_norm": 0.18355444073677063,
      "learning_rate": 1.6648788103147243e-05,
      "loss": 0.3838,
      "num_tokens": 1277333784.0,
      "step": 2017
    },
    {
      "epoch": 0.2386188955894525,
      "grad_norm": 0.17248386144638062,
      "learning_rate": 1.6624082887487718e-05,
      "loss": 0.3434,
      "num_tokens": 1277962469.0,
      "step": 2018
    },
    {
      "epoch": 0.2387371408300816,
      "grad_norm": 0.16043022274971008,
      "learning_rate": 1.6599399341978764e-05,
      "loss": 0.3751,
      "num_tokens": 1278598036.0,
      "step": 2019
    },
    {
      "epoch": 0.23885538607071066,
      "grad_norm": 0.16142123937606812,
      "learning_rate": 1.6574737499283947e-05,
      "loss": 0.3677,
      "num_tokens": 1279232150.0,
      "step": 2020
    },
    {
      "epoch": 0.23897363131133972,
      "grad_norm": 0.15865042805671692,
      "learning_rate": 1.655009739203814e-05,
      "loss": 0.3495,
      "num_tokens": 1279866936.0,
      "step": 2021
    },
    {
      "epoch": 0.23909187655196878,
      "grad_norm": 0.1501854509115219,
      "learning_rate": 1.6525479052847438e-05,
      "loss": 0.326,
      "num_tokens": 1280501697.0,
      "step": 2022
    },
    {
      "epoch": 0.23921012179259785,
      "grad_norm": 0.18602120876312256,
      "learning_rate": 1.6500882514289136e-05,
      "loss": 0.4071,
      "num_tokens": 1281129166.0,
      "step": 2023
    },
    {
      "epoch": 0.2393283670332269,
      "grad_norm": 0.16055260598659515,
      "learning_rate": 1.6476307808911697e-05,
      "loss": 0.3392,
      "num_tokens": 1281765313.0,
      "step": 2024
    },
    {
      "epoch": 0.23944661227385597,
      "grad_norm": 0.16048748791217804,
      "learning_rate": 1.645175496923467e-05,
      "loss": 0.3806,
      "num_tokens": 1282401699.0,
      "step": 2025
    },
    {
      "epoch": 0.23956485751448503,
      "grad_norm": 0.1903311014175415,
      "learning_rate": 1.6427224027748674e-05,
      "loss": 0.3922,
      "num_tokens": 1283036777.0,
      "step": 2026
    },
    {
      "epoch": 0.23968310275511412,
      "grad_norm": 0.1615394949913025,
      "learning_rate": 1.640271501691535e-05,
      "loss": 0.348,
      "num_tokens": 1283671025.0,
      "step": 2027
    },
    {
      "epoch": 0.23980134799574318,
      "grad_norm": 0.14248009026050568,
      "learning_rate": 1.637822796916734e-05,
      "loss": 0.3353,
      "num_tokens": 1284302915.0,
      "step": 2028
    },
    {
      "epoch": 0.23991959323637224,
      "grad_norm": 0.15840069949626923,
      "learning_rate": 1.6353762916908183e-05,
      "loss": 0.3626,
      "num_tokens": 1284941325.0,
      "step": 2029
    },
    {
      "epoch": 0.2400378384770013,
      "grad_norm": 0.15821804106235504,
      "learning_rate": 1.6329319892512347e-05,
      "loss": 0.3585,
      "num_tokens": 1285577381.0,
      "step": 2030
    },
    {
      "epoch": 0.24015608371763036,
      "grad_norm": 0.15751008689403534,
      "learning_rate": 1.6304898928325127e-05,
      "loss": 0.3771,
      "num_tokens": 1286210952.0,
      "step": 2031
    },
    {
      "epoch": 0.24027432895825943,
      "grad_norm": 0.14285607635974884,
      "learning_rate": 1.628050005666263e-05,
      "loss": 0.3395,
      "num_tokens": 1286839896.0,
      "step": 2032
    },
    {
      "epoch": 0.2403925741988885,
      "grad_norm": 0.15762567520141602,
      "learning_rate": 1.625612330981175e-05,
      "loss": 0.361,
      "num_tokens": 1287473506.0,
      "step": 2033
    },
    {
      "epoch": 0.24051081943951755,
      "grad_norm": 0.16265840828418732,
      "learning_rate": 1.6231768720030072e-05,
      "loss": 0.3703,
      "num_tokens": 1288110471.0,
      "step": 2034
    },
    {
      "epoch": 0.24062906468014664,
      "grad_norm": 0.1618996113538742,
      "learning_rate": 1.620743631954589e-05,
      "loss": 0.3638,
      "num_tokens": 1288746833.0,
      "step": 2035
    },
    {
      "epoch": 0.2407473099207757,
      "grad_norm": 0.16014817357063293,
      "learning_rate": 1.6183126140558103e-05,
      "loss": 0.359,
      "num_tokens": 1289384237.0,
      "step": 2036
    },
    {
      "epoch": 0.24086555516140476,
      "grad_norm": 0.16067685186862946,
      "learning_rate": 1.615883821523622e-05,
      "loss": 0.3702,
      "num_tokens": 1290019192.0,
      "step": 2037
    },
    {
      "epoch": 0.24098380040203382,
      "grad_norm": 0.16710048913955688,
      "learning_rate": 1.6134572575720325e-05,
      "loss": 0.3528,
      "num_tokens": 1290658276.0,
      "step": 2038
    },
    {
      "epoch": 0.24110204564266288,
      "grad_norm": 0.17060130834579468,
      "learning_rate": 1.6110329254120973e-05,
      "loss": 0.3591,
      "num_tokens": 1291293510.0,
      "step": 2039
    },
    {
      "epoch": 0.24122029088329194,
      "grad_norm": 0.14449091255664825,
      "learning_rate": 1.608610828251921e-05,
      "loss": 0.3289,
      "num_tokens": 1291917409.0,
      "step": 2040
    },
    {
      "epoch": 0.241338536123921,
      "grad_norm": 0.1513248085975647,
      "learning_rate": 1.6061909692966495e-05,
      "loss": 0.3397,
      "num_tokens": 1292552987.0,
      "step": 2041
    },
    {
      "epoch": 0.24145678136455007,
      "grad_norm": 0.16299116611480713,
      "learning_rate": 1.6037733517484666e-05,
      "loss": 0.3375,
      "num_tokens": 1293186477.0,
      "step": 2042
    },
    {
      "epoch": 0.24157502660517916,
      "grad_norm": 0.17591592669487,
      "learning_rate": 1.6013579788065923e-05,
      "loss": 0.3867,
      "num_tokens": 1293819286.0,
      "step": 2043
    },
    {
      "epoch": 0.24169327184580822,
      "grad_norm": 0.15682390332221985,
      "learning_rate": 1.598944853667274e-05,
      "loss": 0.3087,
      "num_tokens": 1294453861.0,
      "step": 2044
    },
    {
      "epoch": 0.24181151708643728,
      "grad_norm": 0.17341744899749756,
      "learning_rate": 1.5965339795237856e-05,
      "loss": 0.3969,
      "num_tokens": 1295089864.0,
      "step": 2045
    },
    {
      "epoch": 0.24192976232706634,
      "grad_norm": 0.16315826773643494,
      "learning_rate": 1.594125359566422e-05,
      "loss": 0.3458,
      "num_tokens": 1295726569.0,
      "step": 2046
    },
    {
      "epoch": 0.2420480075676954,
      "grad_norm": 0.1428024023771286,
      "learning_rate": 1.5917189969824936e-05,
      "loss": 0.325,
      "num_tokens": 1296360993.0,
      "step": 2047
    },
    {
      "epoch": 0.24216625280832446,
      "grad_norm": 0.17544488608837128,
      "learning_rate": 1.589314894956328e-05,
      "loss": 0.382,
      "num_tokens": 1297000395.0,
      "step": 2048
    },
    {
      "epoch": 0.24228449804895352,
      "grad_norm": 0.1569821685552597,
      "learning_rate": 1.586913056669257e-05,
      "loss": 0.3416,
      "num_tokens": 1297637794.0,
      "step": 2049
    },
    {
      "epoch": 0.24240274328958258,
      "grad_norm": 0.15297435224056244,
      "learning_rate": 1.584513485299619e-05,
      "loss": 0.3497,
      "num_tokens": 1298274450.0,
      "step": 2050
    },
    {
      "epoch": 0.24252098853021165,
      "grad_norm": 0.16815827786922455,
      "learning_rate": 1.5821161840227515e-05,
      "loss": 0.3923,
      "num_tokens": 1298912342.0,
      "step": 2051
    },
    {
      "epoch": 0.24263923377084073,
      "grad_norm": 0.17284280061721802,
      "learning_rate": 1.5797211560109886e-05,
      "loss": 0.3817,
      "num_tokens": 1299542171.0,
      "step": 2052
    },
    {
      "epoch": 0.2427574790114698,
      "grad_norm": 0.17015445232391357,
      "learning_rate": 1.5773284044336574e-05,
      "loss": 0.3735,
      "num_tokens": 1300176165.0,
      "step": 2053
    },
    {
      "epoch": 0.24287572425209886,
      "grad_norm": 0.1607237309217453,
      "learning_rate": 1.5749379324570697e-05,
      "loss": 0.3621,
      "num_tokens": 1300812545.0,
      "step": 2054
    },
    {
      "epoch": 0.24299396949272792,
      "grad_norm": 0.16473503410816193,
      "learning_rate": 1.5725497432445237e-05,
      "loss": 0.39,
      "num_tokens": 1301447175.0,
      "step": 2055
    },
    {
      "epoch": 0.24311221473335698,
      "grad_norm": 0.16471925377845764,
      "learning_rate": 1.5701638399562947e-05,
      "loss": 0.3688,
      "num_tokens": 1302081362.0,
      "step": 2056
    },
    {
      "epoch": 0.24323045997398604,
      "grad_norm": 0.15902522206306458,
      "learning_rate": 1.5677802257496335e-05,
      "loss": 0.3687,
      "num_tokens": 1302715490.0,
      "step": 2057
    },
    {
      "epoch": 0.2433487052146151,
      "grad_norm": 0.1593167930841446,
      "learning_rate": 1.565398903778763e-05,
      "loss": 0.3566,
      "num_tokens": 1303349090.0,
      "step": 2058
    },
    {
      "epoch": 0.24346695045524416,
      "grad_norm": 0.1661464273929596,
      "learning_rate": 1.5630198771948724e-05,
      "loss": 0.4114,
      "num_tokens": 1303981074.0,
      "step": 2059
    },
    {
      "epoch": 0.24358519569587325,
      "grad_norm": 0.15248556435108185,
      "learning_rate": 1.5606431491461118e-05,
      "loss": 0.3632,
      "num_tokens": 1304616565.0,
      "step": 2060
    },
    {
      "epoch": 0.2437034409365023,
      "grad_norm": 0.16896015405654907,
      "learning_rate": 1.55826872277759e-05,
      "loss": 0.3243,
      "num_tokens": 1305250305.0,
      "step": 2061
    },
    {
      "epoch": 0.24382168617713137,
      "grad_norm": 0.15472634136676788,
      "learning_rate": 1.555896601231374e-05,
      "loss": 0.3547,
      "num_tokens": 1305885171.0,
      "step": 2062
    },
    {
      "epoch": 0.24393993141776044,
      "grad_norm": 0.18401753902435303,
      "learning_rate": 1.553526787646475e-05,
      "loss": 0.3802,
      "num_tokens": 1306515269.0,
      "step": 2063
    },
    {
      "epoch": 0.2440581766583895,
      "grad_norm": 0.16186417639255524,
      "learning_rate": 1.5511592851588545e-05,
      "loss": 0.3458,
      "num_tokens": 1307140965.0,
      "step": 2064
    },
    {
      "epoch": 0.24417642189901856,
      "grad_norm": 0.14521846175193787,
      "learning_rate": 1.5487940969014125e-05,
      "loss": 0.3484,
      "num_tokens": 1307772843.0,
      "step": 2065
    },
    {
      "epoch": 0.24429466713964762,
      "grad_norm": 0.16249442100524902,
      "learning_rate": 1.5464312260039874e-05,
      "loss": 0.3702,
      "num_tokens": 1308403711.0,
      "step": 2066
    },
    {
      "epoch": 0.24441291238027668,
      "grad_norm": 0.14522194862365723,
      "learning_rate": 1.544070675593355e-05,
      "loss": 0.3237,
      "num_tokens": 1309036194.0,
      "step": 2067
    },
    {
      "epoch": 0.24453115762090577,
      "grad_norm": 0.1677798628807068,
      "learning_rate": 1.541712448793214e-05,
      "loss": 0.3669,
      "num_tokens": 1309668349.0,
      "step": 2068
    },
    {
      "epoch": 0.24464940286153483,
      "grad_norm": 0.15699172019958496,
      "learning_rate": 1.539356548724193e-05,
      "loss": 0.3702,
      "num_tokens": 1310305545.0,
      "step": 2069
    },
    {
      "epoch": 0.2447676481021639,
      "grad_norm": 0.1510690301656723,
      "learning_rate": 1.5370029785038402e-05,
      "loss": 0.3348,
      "num_tokens": 1310939242.0,
      "step": 2070
    },
    {
      "epoch": 0.24488589334279295,
      "grad_norm": 0.16037912666797638,
      "learning_rate": 1.5346517412466186e-05,
      "loss": 0.3513,
      "num_tokens": 1311572053.0,
      "step": 2071
    },
    {
      "epoch": 0.24500413858342202,
      "grad_norm": 0.14442725479602814,
      "learning_rate": 1.532302840063909e-05,
      "loss": 0.3478,
      "num_tokens": 1312205190.0,
      "step": 2072
    },
    {
      "epoch": 0.24512238382405108,
      "grad_norm": 0.15727399289608002,
      "learning_rate": 1.529956278063996e-05,
      "loss": 0.3902,
      "num_tokens": 1312841010.0,
      "step": 2073
    },
    {
      "epoch": 0.24524062906468014,
      "grad_norm": 0.16083134710788727,
      "learning_rate": 1.527612058352072e-05,
      "loss": 0.3668,
      "num_tokens": 1313472468.0,
      "step": 2074
    },
    {
      "epoch": 0.2453588743053092,
      "grad_norm": 0.1568143367767334,
      "learning_rate": 1.5252701840302268e-05,
      "loss": 0.3468,
      "num_tokens": 1314111009.0,
      "step": 2075
    },
    {
      "epoch": 0.2454771195459383,
      "grad_norm": 0.14847031235694885,
      "learning_rate": 1.5229306581974486e-05,
      "loss": 0.3506,
      "num_tokens": 1314744729.0,
      "step": 2076
    },
    {
      "epoch": 0.24559536478656735,
      "grad_norm": 0.15530890226364136,
      "learning_rate": 1.5205934839496191e-05,
      "loss": 0.3321,
      "num_tokens": 1315375503.0,
      "step": 2077
    },
    {
      "epoch": 0.2457136100271964,
      "grad_norm": 0.1590510606765747,
      "learning_rate": 1.5182586643795055e-05,
      "loss": 0.3644,
      "num_tokens": 1316012926.0,
      "step": 2078
    },
    {
      "epoch": 0.24583185526782547,
      "grad_norm": 0.15811114013195038,
      "learning_rate": 1.515926202576761e-05,
      "loss": 0.3715,
      "num_tokens": 1316650789.0,
      "step": 2079
    },
    {
      "epoch": 0.24595010050845453,
      "grad_norm": 0.1463797390460968,
      "learning_rate": 1.5135961016279167e-05,
      "loss": 0.3414,
      "num_tokens": 1317282748.0,
      "step": 2080
    },
    {
      "epoch": 0.2460683457490836,
      "grad_norm": 0.1546652913093567,
      "learning_rate": 1.5112683646163814e-05,
      "loss": 0.3419,
      "num_tokens": 1317922154.0,
      "step": 2081
    },
    {
      "epoch": 0.24618659098971266,
      "grad_norm": 0.15111781656742096,
      "learning_rate": 1.5089429946224357e-05,
      "loss": 0.3415,
      "num_tokens": 1318555926.0,
      "step": 2082
    },
    {
      "epoch": 0.24630483623034172,
      "grad_norm": 0.163945734500885,
      "learning_rate": 1.5066199947232275e-05,
      "loss": 0.3931,
      "num_tokens": 1319192513.0,
      "step": 2083
    },
    {
      "epoch": 0.2464230814709708,
      "grad_norm": 0.14155668020248413,
      "learning_rate": 1.5042993679927678e-05,
      "loss": 0.3145,
      "num_tokens": 1319824850.0,
      "step": 2084
    },
    {
      "epoch": 0.24654132671159987,
      "grad_norm": 0.16178345680236816,
      "learning_rate": 1.5019811175019274e-05,
      "loss": 0.3892,
      "num_tokens": 1320458462.0,
      "step": 2085
    },
    {
      "epoch": 0.24665957195222893,
      "grad_norm": 0.1564897894859314,
      "learning_rate": 1.499665246318433e-05,
      "loss": 0.344,
      "num_tokens": 1321092111.0,
      "step": 2086
    },
    {
      "epoch": 0.246777817192858,
      "grad_norm": 0.14574626088142395,
      "learning_rate": 1.4973517575068638e-05,
      "loss": 0.3251,
      "num_tokens": 1321724270.0,
      "step": 2087
    },
    {
      "epoch": 0.24689606243348705,
      "grad_norm": 0.16285115480422974,
      "learning_rate": 1.4950406541286447e-05,
      "loss": 0.3619,
      "num_tokens": 1322359786.0,
      "step": 2088
    },
    {
      "epoch": 0.2470143076741161,
      "grad_norm": 0.1563335806131363,
      "learning_rate": 1.4927319392420443e-05,
      "loss": 0.3497,
      "num_tokens": 1322984880.0,
      "step": 2089
    },
    {
      "epoch": 0.24713255291474517,
      "grad_norm": 0.15988491475582123,
      "learning_rate": 1.4904256159021713e-05,
      "loss": 0.3561,
      "num_tokens": 1323623217.0,
      "step": 2090
    },
    {
      "epoch": 0.24725079815537424,
      "grad_norm": 0.1461767852306366,
      "learning_rate": 1.4881216871609679e-05,
      "loss": 0.3369,
      "num_tokens": 1324257199.0,
      "step": 2091
    },
    {
      "epoch": 0.24736904339600332,
      "grad_norm": 0.16058190166950226,
      "learning_rate": 1.4858201560672108e-05,
      "loss": 0.3474,
      "num_tokens": 1324889167.0,
      "step": 2092
    },
    {
      "epoch": 0.24748728863663239,
      "grad_norm": 0.15663427114486694,
      "learning_rate": 1.483521025666501e-05,
      "loss": 0.351,
      "num_tokens": 1325522031.0,
      "step": 2093
    },
    {
      "epoch": 0.24760553387726145,
      "grad_norm": 0.17227518558502197,
      "learning_rate": 1.4812242990012637e-05,
      "loss": 0.3813,
      "num_tokens": 1326153928.0,
      "step": 2094
    },
    {
      "epoch": 0.2477237791178905,
      "grad_norm": 0.16479241847991943,
      "learning_rate": 1.4789299791107426e-05,
      "loss": 0.4066,
      "num_tokens": 1326785341.0,
      "step": 2095
    },
    {
      "epoch": 0.24784202435851957,
      "grad_norm": 0.1628458946943283,
      "learning_rate": 1.4766380690309985e-05,
      "loss": 0.3918,
      "num_tokens": 1327423021.0,
      "step": 2096
    },
    {
      "epoch": 0.24796026959914863,
      "grad_norm": 0.14676113426685333,
      "learning_rate": 1.474348571794901e-05,
      "loss": 0.3159,
      "num_tokens": 1328060086.0,
      "step": 2097
    },
    {
      "epoch": 0.2480785148397777,
      "grad_norm": 0.14470823109149933,
      "learning_rate": 1.4720614904321281e-05,
      "loss": 0.3668,
      "num_tokens": 1328698549.0,
      "step": 2098
    },
    {
      "epoch": 0.24819676008040675,
      "grad_norm": 0.14710788428783417,
      "learning_rate": 1.4697768279691603e-05,
      "loss": 0.3493,
      "num_tokens": 1329331694.0,
      "step": 2099
    },
    {
      "epoch": 0.24831500532103584,
      "grad_norm": 0.16010682284832,
      "learning_rate": 1.4674945874292765e-05,
      "loss": 0.3501,
      "num_tokens": 1329962005.0,
      "step": 2100
    },
    {
      "epoch": 0.2484332505616649,
      "grad_norm": 0.1616450697183609,
      "learning_rate": 1.4652147718325535e-05,
      "loss": 0.3719,
      "num_tokens": 1330597204.0,
      "step": 2101
    },
    {
      "epoch": 0.24855149580229396,
      "grad_norm": 0.16321998834609985,
      "learning_rate": 1.4629373841958563e-05,
      "loss": 0.3487,
      "num_tokens": 1331236517.0,
      "step": 2102
    },
    {
      "epoch": 0.24866974104292303,
      "grad_norm": 0.1300300657749176,
      "learning_rate": 1.4606624275328379e-05,
      "loss": 0.2921,
      "num_tokens": 1331871778.0,
      "step": 2103
    },
    {
      "epoch": 0.2487879862835521,
      "grad_norm": 0.16807685792446136,
      "learning_rate": 1.4583899048539347e-05,
      "loss": 0.3591,
      "num_tokens": 1332509099.0,
      "step": 2104
    },
    {
      "epoch": 0.24890623152418115,
      "grad_norm": 0.15710899233818054,
      "learning_rate": 1.4561198191663608e-05,
      "loss": 0.3425,
      "num_tokens": 1333140652.0,
      "step": 2105
    },
    {
      "epoch": 0.2490244767648102,
      "grad_norm": 0.17277449369430542,
      "learning_rate": 1.4538521734741089e-05,
      "loss": 0.4135,
      "num_tokens": 1333778201.0,
      "step": 2106
    },
    {
      "epoch": 0.24914272200543927,
      "grad_norm": 0.1700480431318283,
      "learning_rate": 1.451586970777939e-05,
      "loss": 0.3538,
      "num_tokens": 1334415211.0,
      "step": 2107
    },
    {
      "epoch": 0.24926096724606833,
      "grad_norm": 0.14761683344841003,
      "learning_rate": 1.4493242140753808e-05,
      "loss": 0.2949,
      "num_tokens": 1335048790.0,
      "step": 2108
    },
    {
      "epoch": 0.24937921248669742,
      "grad_norm": 0.15288735926151276,
      "learning_rate": 1.4470639063607256e-05,
      "loss": 0.3646,
      "num_tokens": 1335683687.0,
      "step": 2109
    },
    {
      "epoch": 0.24949745772732648,
      "grad_norm": 0.148744598031044,
      "learning_rate": 1.4448060506250244e-05,
      "loss": 0.3514,
      "num_tokens": 1336320525.0,
      "step": 2110
    },
    {
      "epoch": 0.24961570296795554,
      "grad_norm": 0.14903289079666138,
      "learning_rate": 1.4425506498560853e-05,
      "loss": 0.3495,
      "num_tokens": 1336952088.0,
      "step": 2111
    },
    {
      "epoch": 0.2497339482085846,
      "grad_norm": 0.15876129269599915,
      "learning_rate": 1.4402977070384653e-05,
      "loss": 0.3764,
      "num_tokens": 1337591066.0,
      "step": 2112
    },
    {
      "epoch": 0.24985219344921367,
      "grad_norm": 0.14318297803401947,
      "learning_rate": 1.43804722515347e-05,
      "loss": 0.308,
      "num_tokens": 1338227429.0,
      "step": 2113
    },
    {
      "epoch": 0.24997043868984273,
      "grad_norm": 0.1475268453359604,
      "learning_rate": 1.4357992071791484e-05,
      "loss": 0.3364,
      "num_tokens": 1338857099.0,
      "step": 2114
    },
    {
      "epoch": 0.2500886839304718,
      "grad_norm": 0.14468343555927277,
      "learning_rate": 1.4335536560902873e-05,
      "loss": 0.3265,
      "num_tokens": 1339491225.0,
      "step": 2115
    },
    {
      "epoch": 0.2502069291711009,
      "grad_norm": 0.152864471077919,
      "learning_rate": 1.4313105748584135e-05,
      "loss": 0.3405,
      "num_tokens": 1340126659.0,
      "step": 2116
    },
    {
      "epoch": 0.2503251744117299,
      "grad_norm": 0.15547658503055573,
      "learning_rate": 1.4290699664517809e-05,
      "loss": 0.3257,
      "num_tokens": 1340752701.0,
      "step": 2117
    },
    {
      "epoch": 0.250443419652359,
      "grad_norm": 0.16664119064807892,
      "learning_rate": 1.4268318338353732e-05,
      "loss": 0.3643,
      "num_tokens": 1341388806.0,
      "step": 2118
    },
    {
      "epoch": 0.25056166489298803,
      "grad_norm": 0.16034823656082153,
      "learning_rate": 1.4245961799708973e-05,
      "loss": 0.3646,
      "num_tokens": 1342025572.0,
      "step": 2119
    },
    {
      "epoch": 0.2506799101336171,
      "grad_norm": 0.16009068489074707,
      "learning_rate": 1.4223630078167796e-05,
      "loss": 0.3524,
      "num_tokens": 1342621450.0,
      "step": 2120
    },
    {
      "epoch": 0.2507981553742462,
      "grad_norm": 0.15944495797157288,
      "learning_rate": 1.420132320328165e-05,
      "loss": 0.3561,
      "num_tokens": 1343251996.0,
      "step": 2121
    },
    {
      "epoch": 0.25091640061487525,
      "grad_norm": 0.16853056848049164,
      "learning_rate": 1.4179041204569073e-05,
      "loss": 0.3782,
      "num_tokens": 1343888063.0,
      "step": 2122
    },
    {
      "epoch": 0.25103464585550433,
      "grad_norm": 0.153689906001091,
      "learning_rate": 1.4156784111515697e-05,
      "loss": 0.3152,
      "num_tokens": 1344524355.0,
      "step": 2123
    },
    {
      "epoch": 0.25115289109613337,
      "grad_norm": 0.16411302983760834,
      "learning_rate": 1.4134551953574201e-05,
      "loss": 0.3794,
      "num_tokens": 1345157130.0,
      "step": 2124
    },
    {
      "epoch": 0.25127113633676246,
      "grad_norm": 0.149836927652359,
      "learning_rate": 1.4112344760164249e-05,
      "loss": 0.3162,
      "num_tokens": 1345788742.0,
      "step": 2125
    },
    {
      "epoch": 0.2513893815773915,
      "grad_norm": 0.14406760036945343,
      "learning_rate": 1.4090162560672506e-05,
      "loss": 0.3558,
      "num_tokens": 1346424640.0,
      "step": 2126
    },
    {
      "epoch": 0.2515076268180206,
      "grad_norm": 0.14408567547798157,
      "learning_rate": 1.4068005384452529e-05,
      "loss": 0.3522,
      "num_tokens": 1347050079.0,
      "step": 2127
    },
    {
      "epoch": 0.2516258720586496,
      "grad_norm": 0.15319818258285522,
      "learning_rate": 1.4045873260824772e-05,
      "loss": 0.3417,
      "num_tokens": 1347683639.0,
      "step": 2128
    },
    {
      "epoch": 0.2517441172992787,
      "grad_norm": 0.16909605264663696,
      "learning_rate": 1.4023766219076532e-05,
      "loss": 0.4197,
      "num_tokens": 1348322094.0,
      "step": 2129
    },
    {
      "epoch": 0.2518623625399078,
      "grad_norm": 0.16063344478607178,
      "learning_rate": 1.4001684288461939e-05,
      "loss": 0.3496,
      "num_tokens": 1348957537.0,
      "step": 2130
    },
    {
      "epoch": 0.2519806077805368,
      "grad_norm": 0.15000493824481964,
      "learning_rate": 1.3979627498201864e-05,
      "loss": 0.3504,
      "num_tokens": 1349596973.0,
      "step": 2131
    },
    {
      "epoch": 0.2520988530211659,
      "grad_norm": 0.14872415363788605,
      "learning_rate": 1.3957595877483925e-05,
      "loss": 0.3262,
      "num_tokens": 1350233160.0,
      "step": 2132
    },
    {
      "epoch": 0.25221709826179495,
      "grad_norm": 0.16195665299892426,
      "learning_rate": 1.3935589455462432e-05,
      "loss": 0.3389,
      "num_tokens": 1350870883.0,
      "step": 2133
    },
    {
      "epoch": 0.25233534350242404,
      "grad_norm": 0.1620256006717682,
      "learning_rate": 1.3913608261258338e-05,
      "loss": 0.3235,
      "num_tokens": 1351499885.0,
      "step": 2134
    },
    {
      "epoch": 0.25245358874305307,
      "grad_norm": 0.13652187585830688,
      "learning_rate": 1.3891652323959228e-05,
      "loss": 0.3166,
      "num_tokens": 1352131223.0,
      "step": 2135
    },
    {
      "epoch": 0.25257183398368216,
      "grad_norm": 0.16615505516529083,
      "learning_rate": 1.3869721672619274e-05,
      "loss": 0.3438,
      "num_tokens": 1352768469.0,
      "step": 2136
    },
    {
      "epoch": 0.25269007922431125,
      "grad_norm": 0.14930325746536255,
      "learning_rate": 1.3847816336259156e-05,
      "loss": 0.3163,
      "num_tokens": 1353395367.0,
      "step": 2137
    },
    {
      "epoch": 0.2528083244649403,
      "grad_norm": 0.1527901440858841,
      "learning_rate": 1.3825936343866075e-05,
      "loss": 0.3608,
      "num_tokens": 1354024395.0,
      "step": 2138
    },
    {
      "epoch": 0.25292656970556937,
      "grad_norm": 0.1636216789484024,
      "learning_rate": 1.3804081724393691e-05,
      "loss": 0.3696,
      "num_tokens": 1354657702.0,
      "step": 2139
    },
    {
      "epoch": 0.2530448149461984,
      "grad_norm": 0.15648838877677917,
      "learning_rate": 1.3782252506762083e-05,
      "loss": 0.3834,
      "num_tokens": 1355295855.0,
      "step": 2140
    },
    {
      "epoch": 0.2531630601868275,
      "grad_norm": 0.1640644371509552,
      "learning_rate": 1.3760448719857729e-05,
      "loss": 0.3471,
      "num_tokens": 1355913131.0,
      "step": 2141
    },
    {
      "epoch": 0.2532813054274565,
      "grad_norm": 0.15252865850925446,
      "learning_rate": 1.3738670392533444e-05,
      "loss": 0.3549,
      "num_tokens": 1356552376.0,
      "step": 2142
    },
    {
      "epoch": 0.2533995506680856,
      "grad_norm": 0.16179129481315613,
      "learning_rate": 1.3716917553608351e-05,
      "loss": 0.3726,
      "num_tokens": 1357190876.0,
      "step": 2143
    },
    {
      "epoch": 0.25351779590871465,
      "grad_norm": 0.1642070710659027,
      "learning_rate": 1.3695190231867853e-05,
      "loss": 0.3587,
      "num_tokens": 1357827846.0,
      "step": 2144
    },
    {
      "epoch": 0.25363604114934374,
      "grad_norm": 0.14839212596416473,
      "learning_rate": 1.367348845606357e-05,
      "loss": 0.3342,
      "num_tokens": 1358460902.0,
      "step": 2145
    },
    {
      "epoch": 0.2537542863899728,
      "grad_norm": 0.15711148083209991,
      "learning_rate": 1.3651812254913344e-05,
      "loss": 0.3558,
      "num_tokens": 1359097940.0,
      "step": 2146
    },
    {
      "epoch": 0.25387253163060186,
      "grad_norm": 0.14963391423225403,
      "learning_rate": 1.3630161657101154e-05,
      "loss": 0.3304,
      "num_tokens": 1359727970.0,
      "step": 2147
    },
    {
      "epoch": 0.25399077687123095,
      "grad_norm": 0.16470035910606384,
      "learning_rate": 1.3608536691277099e-05,
      "loss": 0.3534,
      "num_tokens": 1360363518.0,
      "step": 2148
    },
    {
      "epoch": 0.25410902211186,
      "grad_norm": 0.18055987358093262,
      "learning_rate": 1.3586937386057365e-05,
      "loss": 0.3466,
      "num_tokens": 1360996423.0,
      "step": 2149
    },
    {
      "epoch": 0.2542272673524891,
      "grad_norm": 0.16418340802192688,
      "learning_rate": 1.3565363770024177e-05,
      "loss": 0.3435,
      "num_tokens": 1361629975.0,
      "step": 2150
    },
    {
      "epoch": 0.2543455125931181,
      "grad_norm": 0.16093459725379944,
      "learning_rate": 1.3543815871725777e-05,
      "loss": 0.3634,
      "num_tokens": 1362263332.0,
      "step": 2151
    },
    {
      "epoch": 0.2544637578337472,
      "grad_norm": 0.16369055211544037,
      "learning_rate": 1.3522293719676363e-05,
      "loss": 0.3681,
      "num_tokens": 1362893971.0,
      "step": 2152
    },
    {
      "epoch": 0.2545820030743763,
      "grad_norm": 0.14665311574935913,
      "learning_rate": 1.3500797342356067e-05,
      "loss": 0.3719,
      "num_tokens": 1363530574.0,
      "step": 2153
    },
    {
      "epoch": 0.2547002483150053,
      "grad_norm": 0.1517995148897171,
      "learning_rate": 1.3479326768210912e-05,
      "loss": 0.354,
      "num_tokens": 1364162896.0,
      "step": 2154
    },
    {
      "epoch": 0.2548184935556344,
      "grad_norm": 0.16105465590953827,
      "learning_rate": 1.3457882025652765e-05,
      "loss": 0.3433,
      "num_tokens": 1364795298.0,
      "step": 2155
    },
    {
      "epoch": 0.25493673879626344,
      "grad_norm": 0.14719729125499725,
      "learning_rate": 1.3436463143059343e-05,
      "loss": 0.3726,
      "num_tokens": 1365431336.0,
      "step": 2156
    },
    {
      "epoch": 0.25505498403689253,
      "grad_norm": 0.14745932817459106,
      "learning_rate": 1.3415070148774115e-05,
      "loss": 0.3386,
      "num_tokens": 1366063237.0,
      "step": 2157
    },
    {
      "epoch": 0.25517322927752156,
      "grad_norm": 0.15282244980335236,
      "learning_rate": 1.3393703071106295e-05,
      "loss": 0.3277,
      "num_tokens": 1366701239.0,
      "step": 2158
    },
    {
      "epoch": 0.25529147451815065,
      "grad_norm": 0.16405080258846283,
      "learning_rate": 1.3372361938330806e-05,
      "loss": 0.3404,
      "num_tokens": 1367334509.0,
      "step": 2159
    },
    {
      "epoch": 0.2554097197587797,
      "grad_norm": 0.15155361592769623,
      "learning_rate": 1.3351046778688231e-05,
      "loss": 0.3715,
      "num_tokens": 1367970895.0,
      "step": 2160
    },
    {
      "epoch": 0.2555279649994088,
      "grad_norm": 0.1698579043149948,
      "learning_rate": 1.3329757620384806e-05,
      "loss": 0.3459,
      "num_tokens": 1368606104.0,
      "step": 2161
    },
    {
      "epoch": 0.25564621024003786,
      "grad_norm": 0.15005841851234436,
      "learning_rate": 1.3308494491592337e-05,
      "loss": 0.35,
      "num_tokens": 1369242016.0,
      "step": 2162
    },
    {
      "epoch": 0.2557644554806669,
      "grad_norm": 0.16898460686206818,
      "learning_rate": 1.328725742044819e-05,
      "loss": 0.3533,
      "num_tokens": 1369878992.0,
      "step": 2163
    },
    {
      "epoch": 0.255882700721296,
      "grad_norm": 0.1578129231929779,
      "learning_rate": 1.3266046435055247e-05,
      "loss": 0.3561,
      "num_tokens": 1370511489.0,
      "step": 2164
    },
    {
      "epoch": 0.256000945961925,
      "grad_norm": 0.15820948779582977,
      "learning_rate": 1.3244861563481878e-05,
      "loss": 0.3651,
      "num_tokens": 1371142032.0,
      "step": 2165
    },
    {
      "epoch": 0.2561191912025541,
      "grad_norm": 0.14821572601795197,
      "learning_rate": 1.32237028337619e-05,
      "loss": 0.346,
      "num_tokens": 1371775322.0,
      "step": 2166
    },
    {
      "epoch": 0.25623743644318314,
      "grad_norm": 0.14930571615695953,
      "learning_rate": 1.3202570273894522e-05,
      "loss": 0.3376,
      "num_tokens": 1372409818.0,
      "step": 2167
    },
    {
      "epoch": 0.25635568168381223,
      "grad_norm": 0.16223351657390594,
      "learning_rate": 1.3181463911844337e-05,
      "loss": 0.3759,
      "num_tokens": 1373047668.0,
      "step": 2168
    },
    {
      "epoch": 0.25647392692444126,
      "grad_norm": 0.17227718234062195,
      "learning_rate": 1.3160383775541251e-05,
      "loss": 0.3755,
      "num_tokens": 1373685499.0,
      "step": 2169
    },
    {
      "epoch": 0.25659217216507035,
      "grad_norm": 0.15489031374454498,
      "learning_rate": 1.3139329892880495e-05,
      "loss": 0.3598,
      "num_tokens": 1374323849.0,
      "step": 2170
    },
    {
      "epoch": 0.25671041740569944,
      "grad_norm": 0.14238935708999634,
      "learning_rate": 1.3118302291722535e-05,
      "loss": 0.2983,
      "num_tokens": 1374959880.0,
      "step": 2171
    },
    {
      "epoch": 0.2568286626463285,
      "grad_norm": 0.14943967759609222,
      "learning_rate": 1.309730099989307e-05,
      "loss": 0.3409,
      "num_tokens": 1375595575.0,
      "step": 2172
    },
    {
      "epoch": 0.25694690788695757,
      "grad_norm": 0.15923093259334564,
      "learning_rate": 1.3076326045182975e-05,
      "loss": 0.363,
      "num_tokens": 1376233493.0,
      "step": 2173
    },
    {
      "epoch": 0.2570651531275866,
      "grad_norm": 0.17506137490272522,
      "learning_rate": 1.3055377455348276e-05,
      "loss": 0.35,
      "num_tokens": 1376867408.0,
      "step": 2174
    },
    {
      "epoch": 0.2571833983682157,
      "grad_norm": 0.14500072598457336,
      "learning_rate": 1.3034455258110127e-05,
      "loss": 0.3259,
      "num_tokens": 1377501464.0,
      "step": 2175
    },
    {
      "epoch": 0.2573016436088447,
      "grad_norm": 0.15769173204898834,
      "learning_rate": 1.3013559481154734e-05,
      "loss": 0.3356,
      "num_tokens": 1378107558.0,
      "step": 2176
    },
    {
      "epoch": 0.2574198888494738,
      "grad_norm": 0.16178518533706665,
      "learning_rate": 1.2992690152133351e-05,
      "loss": 0.3318,
      "num_tokens": 1378742114.0,
      "step": 2177
    },
    {
      "epoch": 0.2575381340901029,
      "grad_norm": 0.136235311627388,
      "learning_rate": 1.2971847298662235e-05,
      "loss": 0.3182,
      "num_tokens": 1379380501.0,
      "step": 2178
    },
    {
      "epoch": 0.25765637933073193,
      "grad_norm": 0.17029152810573578,
      "learning_rate": 1.2951030948322594e-05,
      "loss": 0.3309,
      "num_tokens": 1380013591.0,
      "step": 2179
    },
    {
      "epoch": 0.257774624571361,
      "grad_norm": 0.14048749208450317,
      "learning_rate": 1.2930241128660597e-05,
      "loss": 0.3193,
      "num_tokens": 1380648061.0,
      "step": 2180
    },
    {
      "epoch": 0.25789286981199006,
      "grad_norm": 0.15299919247627258,
      "learning_rate": 1.2909477867187276e-05,
      "loss": 0.3569,
      "num_tokens": 1381282667.0,
      "step": 2181
    },
    {
      "epoch": 0.25801111505261914,
      "grad_norm": 0.15771017968654633,
      "learning_rate": 1.2888741191378522e-05,
      "loss": 0.3695,
      "num_tokens": 1381921709.0,
      "step": 2182
    },
    {
      "epoch": 0.2581293602932482,
      "grad_norm": 0.15074072778224945,
      "learning_rate": 1.2868031128675058e-05,
      "loss": 0.3364,
      "num_tokens": 1382558484.0,
      "step": 2183
    },
    {
      "epoch": 0.25824760553387727,
      "grad_norm": 0.1526341289281845,
      "learning_rate": 1.2847347706482375e-05,
      "loss": 0.3643,
      "num_tokens": 1383194599.0,
      "step": 2184
    },
    {
      "epoch": 0.2583658507745063,
      "grad_norm": 0.15062035620212555,
      "learning_rate": 1.2826690952170725e-05,
      "loss": 0.3334,
      "num_tokens": 1383830599.0,
      "step": 2185
    },
    {
      "epoch": 0.2584840960151354,
      "grad_norm": 0.15836162865161896,
      "learning_rate": 1.2806060893075064e-05,
      "loss": 0.3451,
      "num_tokens": 1384469453.0,
      "step": 2186
    },
    {
      "epoch": 0.2586023412557645,
      "grad_norm": 0.1659291684627533,
      "learning_rate": 1.2785457556495023e-05,
      "loss": 0.3624,
      "num_tokens": 1385108151.0,
      "step": 2187
    },
    {
      "epoch": 0.2587205864963935,
      "grad_norm": 0.1547379344701767,
      "learning_rate": 1.2764880969694867e-05,
      "loss": 0.3677,
      "num_tokens": 1385737537.0,
      "step": 2188
    },
    {
      "epoch": 0.2588388317370226,
      "grad_norm": 0.16594819724559784,
      "learning_rate": 1.2744331159903462e-05,
      "loss": 0.3528,
      "num_tokens": 1386371978.0,
      "step": 2189
    },
    {
      "epoch": 0.25895707697765163,
      "grad_norm": 0.1468471735715866,
      "learning_rate": 1.2723808154314257e-05,
      "loss": 0.3031,
      "num_tokens": 1387006855.0,
      "step": 2190
    },
    {
      "epoch": 0.2590753222182807,
      "grad_norm": 0.14902792870998383,
      "learning_rate": 1.2703311980085216e-05,
      "loss": 0.3739,
      "num_tokens": 1387642922.0,
      "step": 2191
    },
    {
      "epoch": 0.25919356745890976,
      "grad_norm": 0.16405217349529266,
      "learning_rate": 1.2682842664338798e-05,
      "loss": 0.3412,
      "num_tokens": 1388273872.0,
      "step": 2192
    },
    {
      "epoch": 0.25931181269953885,
      "grad_norm": 0.1533263772726059,
      "learning_rate": 1.2662400234161925e-05,
      "loss": 0.3315,
      "num_tokens": 1388907732.0,
      "step": 2193
    },
    {
      "epoch": 0.25943005794016794,
      "grad_norm": 0.18562085926532745,
      "learning_rate": 1.2641984716605926e-05,
      "loss": 0.3749,
      "num_tokens": 1389539553.0,
      "step": 2194
    },
    {
      "epoch": 0.25954830318079697,
      "grad_norm": 0.1535794585943222,
      "learning_rate": 1.2621596138686556e-05,
      "loss": 0.3604,
      "num_tokens": 1390174842.0,
      "step": 2195
    },
    {
      "epoch": 0.25966654842142606,
      "grad_norm": 0.16793584823608398,
      "learning_rate": 1.2601234527383877e-05,
      "loss": 0.3613,
      "num_tokens": 1390810002.0,
      "step": 2196
    },
    {
      "epoch": 0.2597847936620551,
      "grad_norm": 0.15016791224479675,
      "learning_rate": 1.2580899909642288e-05,
      "loss": 0.3614,
      "num_tokens": 1391444325.0,
      "step": 2197
    },
    {
      "epoch": 0.2599030389026842,
      "grad_norm": 0.14536795020103455,
      "learning_rate": 1.2560592312370456e-05,
      "loss": 0.3375,
      "num_tokens": 1392075518.0,
      "step": 2198
    },
    {
      "epoch": 0.2600212841433132,
      "grad_norm": 0.16178761422634125,
      "learning_rate": 1.2540311762441314e-05,
      "loss": 0.3751,
      "num_tokens": 1392708819.0,
      "step": 2199
    },
    {
      "epoch": 0.2601395293839423,
      "grad_norm": 0.17659318447113037,
      "learning_rate": 1.2520058286691983e-05,
      "loss": 0.3666,
      "num_tokens": 1393340694.0,
      "step": 2200
    },
    {
      "epoch": 0.26025777462457134,
      "grad_norm": 0.14843176305294037,
      "learning_rate": 1.2499831911923764e-05,
      "loss": 0.3502,
      "num_tokens": 1393973875.0,
      "step": 2201
    },
    {
      "epoch": 0.2603760198652004,
      "grad_norm": 0.15998484194278717,
      "learning_rate": 1.2479632664902092e-05,
      "loss": 0.3457,
      "num_tokens": 1394610492.0,
      "step": 2202
    },
    {
      "epoch": 0.2604942651058295,
      "grad_norm": 0.16628916561603546,
      "learning_rate": 1.2459460572356502e-05,
      "loss": 0.3594,
      "num_tokens": 1395249777.0,
      "step": 2203
    },
    {
      "epoch": 0.26061251034645855,
      "grad_norm": 0.15287388861179352,
      "learning_rate": 1.2439315660980614e-05,
      "loss": 0.3491,
      "num_tokens": 1395881158.0,
      "step": 2204
    },
    {
      "epoch": 0.26073075558708764,
      "grad_norm": 0.1673702746629715,
      "learning_rate": 1.2419197957432053e-05,
      "loss": 0.3629,
      "num_tokens": 1396510147.0,
      "step": 2205
    },
    {
      "epoch": 0.26084900082771667,
      "grad_norm": 0.15775969624519348,
      "learning_rate": 1.2399107488332462e-05,
      "loss": 0.3576,
      "num_tokens": 1397143032.0,
      "step": 2206
    },
    {
      "epoch": 0.26096724606834576,
      "grad_norm": 0.1686294972896576,
      "learning_rate": 1.2379044280267427e-05,
      "loss": 0.3673,
      "num_tokens": 1397778901.0,
      "step": 2207
    },
    {
      "epoch": 0.2610854913089748,
      "grad_norm": 0.1551026701927185,
      "learning_rate": 1.2359008359786469e-05,
      "loss": 0.34,
      "num_tokens": 1398416648.0,
      "step": 2208
    },
    {
      "epoch": 0.2612037365496039,
      "grad_norm": 0.14781665802001953,
      "learning_rate": 1.2338999753403002e-05,
      "loss": 0.338,
      "num_tokens": 1399055774.0,
      "step": 2209
    },
    {
      "epoch": 0.2613219817902329,
      "grad_norm": 0.1639232337474823,
      "learning_rate": 1.2319018487594295e-05,
      "loss": 0.3909,
      "num_tokens": 1399679014.0,
      "step": 2210
    },
    {
      "epoch": 0.261440227030862,
      "grad_norm": 0.16171275079250336,
      "learning_rate": 1.2299064588801428e-05,
      "loss": 0.3538,
      "num_tokens": 1400310496.0,
      "step": 2211
    },
    {
      "epoch": 0.2615584722714911,
      "grad_norm": 0.1480337232351303,
      "learning_rate": 1.2279138083429277e-05,
      "loss": 0.3343,
      "num_tokens": 1400944987.0,
      "step": 2212
    },
    {
      "epoch": 0.2616767175121201,
      "grad_norm": 0.15325020253658295,
      "learning_rate": 1.2259238997846454e-05,
      "loss": 0.3251,
      "num_tokens": 1401580390.0,
      "step": 2213
    },
    {
      "epoch": 0.2617949627527492,
      "grad_norm": 0.16419720649719238,
      "learning_rate": 1.2239367358385317e-05,
      "loss": 0.3737,
      "num_tokens": 1402207325.0,
      "step": 2214
    },
    {
      "epoch": 0.26191320799337825,
      "grad_norm": 0.15797926485538483,
      "learning_rate": 1.2219523191341866e-05,
      "loss": 0.3559,
      "num_tokens": 1402846619.0,
      "step": 2215
    },
    {
      "epoch": 0.26203145323400734,
      "grad_norm": 0.15500512719154358,
      "learning_rate": 1.2199706522975776e-05,
      "loss": 0.3411,
      "num_tokens": 1403475238.0,
      "step": 2216
    },
    {
      "epoch": 0.2621496984746364,
      "grad_norm": 0.1467328667640686,
      "learning_rate": 1.2179917379510315e-05,
      "loss": 0.3186,
      "num_tokens": 1404109414.0,
      "step": 2217
    },
    {
      "epoch": 0.26226794371526546,
      "grad_norm": 0.15633271634578705,
      "learning_rate": 1.2160155787132334e-05,
      "loss": 0.3618,
      "num_tokens": 1404741649.0,
      "step": 2218
    },
    {
      "epoch": 0.26238618895589455,
      "grad_norm": 0.1830826699733734,
      "learning_rate": 1.2140421771992239e-05,
      "loss": 0.3871,
      "num_tokens": 1405372004.0,
      "step": 2219
    },
    {
      "epoch": 0.2625044341965236,
      "grad_norm": 0.15860310196876526,
      "learning_rate": 1.2120715360203924e-05,
      "loss": 0.3829,
      "num_tokens": 1406008107.0,
      "step": 2220
    },
    {
      "epoch": 0.2626226794371527,
      "grad_norm": 0.156002938747406,
      "learning_rate": 1.210103657784476e-05,
      "loss": 0.329,
      "num_tokens": 1406642830.0,
      "step": 2221
    },
    {
      "epoch": 0.2627409246777817,
      "grad_norm": 0.15903633832931519,
      "learning_rate": 1.2081385450955562e-05,
      "loss": 0.3507,
      "num_tokens": 1407279706.0,
      "step": 2222
    },
    {
      "epoch": 0.2628591699184108,
      "grad_norm": 0.14660197496414185,
      "learning_rate": 1.2061762005540534e-05,
      "loss": 0.3719,
      "num_tokens": 1407913980.0,
      "step": 2223
    },
    {
      "epoch": 0.26297741515903983,
      "grad_norm": 0.1643187403678894,
      "learning_rate": 1.2042166267567283e-05,
      "loss": 0.3839,
      "num_tokens": 1408544218.0,
      "step": 2224
    },
    {
      "epoch": 0.2630956603996689,
      "grad_norm": 0.1486949622631073,
      "learning_rate": 1.2022598262966715e-05,
      "loss": 0.3254,
      "num_tokens": 1409176076.0,
      "step": 2225
    },
    {
      "epoch": 0.26321390564029795,
      "grad_norm": 0.147776260972023,
      "learning_rate": 1.2003058017633055e-05,
      "loss": 0.3251,
      "num_tokens": 1409808144.0,
      "step": 2226
    },
    {
      "epoch": 0.26333215088092704,
      "grad_norm": 0.1610105186700821,
      "learning_rate": 1.1983545557423785e-05,
      "loss": 0.3559,
      "num_tokens": 1410438046.0,
      "step": 2227
    },
    {
      "epoch": 0.26345039612155613,
      "grad_norm": 0.15127472579479218,
      "learning_rate": 1.1964060908159617e-05,
      "loss": 0.3498,
      "num_tokens": 1411070928.0,
      "step": 2228
    },
    {
      "epoch": 0.26356864136218516,
      "grad_norm": 0.17238755524158478,
      "learning_rate": 1.1944604095624487e-05,
      "loss": 0.3671,
      "num_tokens": 1411705318.0,
      "step": 2229
    },
    {
      "epoch": 0.26368688660281425,
      "grad_norm": 0.14921915531158447,
      "learning_rate": 1.1925175145565464e-05,
      "loss": 0.3364,
      "num_tokens": 1412339837.0,
      "step": 2230
    },
    {
      "epoch": 0.2638051318434433,
      "grad_norm": 0.156773179769516,
      "learning_rate": 1.190577408369276e-05,
      "loss": 0.3049,
      "num_tokens": 1412944973.0,
      "step": 2231
    },
    {
      "epoch": 0.2639233770840724,
      "grad_norm": 0.15290521085262299,
      "learning_rate": 1.1886400935679674e-05,
      "loss": 0.3543,
      "num_tokens": 1413570089.0,
      "step": 2232
    },
    {
      "epoch": 0.2640416223247014,
      "grad_norm": 0.1462039202451706,
      "learning_rate": 1.1867055727162587e-05,
      "loss": 0.3502,
      "num_tokens": 1414206557.0,
      "step": 2233
    },
    {
      "epoch": 0.2641598675653305,
      "grad_norm": 0.15547192096710205,
      "learning_rate": 1.1847738483740886e-05,
      "loss": 0.3444,
      "num_tokens": 1414839263.0,
      "step": 2234
    },
    {
      "epoch": 0.2642781128059596,
      "grad_norm": 0.14497071504592896,
      "learning_rate": 1.1828449230976965e-05,
      "loss": 0.3448,
      "num_tokens": 1415473529.0,
      "step": 2235
    },
    {
      "epoch": 0.2643963580465886,
      "grad_norm": 0.14295479655265808,
      "learning_rate": 1.1809187994396168e-05,
      "loss": 0.3288,
      "num_tokens": 1416110661.0,
      "step": 2236
    },
    {
      "epoch": 0.2645146032872177,
      "grad_norm": 0.17718076705932617,
      "learning_rate": 1.178995479948677e-05,
      "loss": 0.3639,
      "num_tokens": 1416741736.0,
      "step": 2237
    },
    {
      "epoch": 0.26463284852784674,
      "grad_norm": 0.15077832341194153,
      "learning_rate": 1.177074967169995e-05,
      "loss": 0.3182,
      "num_tokens": 1417367685.0,
      "step": 2238
    },
    {
      "epoch": 0.26475109376847583,
      "grad_norm": 0.1541740894317627,
      "learning_rate": 1.175157263644973e-05,
      "loss": 0.3239,
      "num_tokens": 1418004275.0,
      "step": 2239
    },
    {
      "epoch": 0.26486933900910486,
      "grad_norm": 0.16033287346363068,
      "learning_rate": 1.1732423719112964e-05,
      "loss": 0.3284,
      "num_tokens": 1418632155.0,
      "step": 2240
    },
    {
      "epoch": 0.26498758424973395,
      "grad_norm": 0.15380583703517914,
      "learning_rate": 1.1713302945029295e-05,
      "loss": 0.3649,
      "num_tokens": 1419269496.0,
      "step": 2241
    },
    {
      "epoch": 0.265105829490363,
      "grad_norm": 0.14381130039691925,
      "learning_rate": 1.1694210339501118e-05,
      "loss": 0.3257,
      "num_tokens": 1419904473.0,
      "step": 2242
    },
    {
      "epoch": 0.2652240747309921,
      "grad_norm": 0.15737563371658325,
      "learning_rate": 1.1675145927793579e-05,
      "loss": 0.3254,
      "num_tokens": 1420538845.0,
      "step": 2243
    },
    {
      "epoch": 0.26534231997162117,
      "grad_norm": 0.1575727015733719,
      "learning_rate": 1.1656109735134482e-05,
      "loss": 0.3566,
      "num_tokens": 1421173978.0,
      "step": 2244
    },
    {
      "epoch": 0.2654605652122502,
      "grad_norm": 0.14479680359363556,
      "learning_rate": 1.1637101786714315e-05,
      "loss": 0.3718,
      "num_tokens": 1421805151.0,
      "step": 2245
    },
    {
      "epoch": 0.2655788104528793,
      "grad_norm": 0.1669468730688095,
      "learning_rate": 1.1618122107686168e-05,
      "loss": 0.3823,
      "num_tokens": 1422442066.0,
      "step": 2246
    },
    {
      "epoch": 0.2656970556935083,
      "grad_norm": 0.16510143876075745,
      "learning_rate": 1.1599170723165734e-05,
      "loss": 0.327,
      "num_tokens": 1423075090.0,
      "step": 2247
    },
    {
      "epoch": 0.2658153009341374,
      "grad_norm": 0.1484033763408661,
      "learning_rate": 1.1580247658231277e-05,
      "loss": 0.3169,
      "num_tokens": 1423675926.0,
      "step": 2248
    },
    {
      "epoch": 0.26593354617476644,
      "grad_norm": 0.15835952758789062,
      "learning_rate": 1.1561352937923563e-05,
      "loss": 0.3288,
      "num_tokens": 1424296031.0,
      "step": 2249
    },
    {
      "epoch": 0.26605179141539553,
      "grad_norm": 0.1563296765089035,
      "learning_rate": 1.154248658724586e-05,
      "loss": 0.3507,
      "num_tokens": 1424930538.0,
      "step": 2250
    },
    {
      "epoch": 0.2661700366560246,
      "grad_norm": 0.16382481157779694,
      "learning_rate": 1.1523648631163897e-05,
      "loss": 0.3681,
      "num_tokens": 1425559363.0,
      "step": 2251
    },
    {
      "epoch": 0.26628828189665366,
      "grad_norm": 0.15770530700683594,
      "learning_rate": 1.1504839094605815e-05,
      "loss": 0.3637,
      "num_tokens": 1426194887.0,
      "step": 2252
    },
    {
      "epoch": 0.26640652713728274,
      "grad_norm": 0.16853371262550354,
      "learning_rate": 1.1486058002462176e-05,
      "loss": 0.3491,
      "num_tokens": 1426827620.0,
      "step": 2253
    },
    {
      "epoch": 0.2665247723779118,
      "grad_norm": 0.1755160689353943,
      "learning_rate": 1.1467305379585872e-05,
      "loss": 0.3548,
      "num_tokens": 1427462285.0,
      "step": 2254
    },
    {
      "epoch": 0.26664301761854087,
      "grad_norm": 0.1687714010477066,
      "learning_rate": 1.1448581250792136e-05,
      "loss": 0.3571,
      "num_tokens": 1428098823.0,
      "step": 2255
    },
    {
      "epoch": 0.2667612628591699,
      "grad_norm": 0.14964935183525085,
      "learning_rate": 1.1429885640858493e-05,
      "loss": 0.3432,
      "num_tokens": 1428738293.0,
      "step": 2256
    },
    {
      "epoch": 0.266879508099799,
      "grad_norm": 0.15902990102767944,
      "learning_rate": 1.1411218574524726e-05,
      "loss": 0.3554,
      "num_tokens": 1429371807.0,
      "step": 2257
    },
    {
      "epoch": 0.266997753340428,
      "grad_norm": 0.15595631301403046,
      "learning_rate": 1.1392580076492852e-05,
      "loss": 0.3519,
      "num_tokens": 1430008824.0,
      "step": 2258
    },
    {
      "epoch": 0.2671159985810571,
      "grad_norm": 0.17033277451992035,
      "learning_rate": 1.1373970171427084e-05,
      "loss": 0.3639,
      "num_tokens": 1430648432.0,
      "step": 2259
    },
    {
      "epoch": 0.2672342438216862,
      "grad_norm": 0.16252721846103668,
      "learning_rate": 1.1355388883953794e-05,
      "loss": 0.3266,
      "num_tokens": 1431274507.0,
      "step": 2260
    },
    {
      "epoch": 0.26735248906231524,
      "grad_norm": 0.16513614356517792,
      "learning_rate": 1.133683623866148e-05,
      "loss": 0.3513,
      "num_tokens": 1431899378.0,
      "step": 2261
    },
    {
      "epoch": 0.2674707343029443,
      "grad_norm": 0.16827766597270966,
      "learning_rate": 1.1318312260100742e-05,
      "loss": 0.3575,
      "num_tokens": 1432532765.0,
      "step": 2262
    },
    {
      "epoch": 0.26758897954357336,
      "grad_norm": 0.156415656208992,
      "learning_rate": 1.1299816972784262e-05,
      "loss": 0.3473,
      "num_tokens": 1433167753.0,
      "step": 2263
    },
    {
      "epoch": 0.26770722478420245,
      "grad_norm": 0.1539047360420227,
      "learning_rate": 1.1281350401186727e-05,
      "loss": 0.3337,
      "num_tokens": 1433802888.0,
      "step": 2264
    },
    {
      "epoch": 0.2678254700248315,
      "grad_norm": 0.1548675000667572,
      "learning_rate": 1.1262912569744841e-05,
      "loss": 0.3582,
      "num_tokens": 1434439739.0,
      "step": 2265
    },
    {
      "epoch": 0.26794371526546057,
      "grad_norm": 0.1661369800567627,
      "learning_rate": 1.124450350285727e-05,
      "loss": 0.3617,
      "num_tokens": 1435075550.0,
      "step": 2266
    },
    {
      "epoch": 0.2680619605060896,
      "grad_norm": 0.16508784890174866,
      "learning_rate": 1.1226123224884625e-05,
      "loss": 0.358,
      "num_tokens": 1435702450.0,
      "step": 2267
    },
    {
      "epoch": 0.2681802057467187,
      "grad_norm": 0.1570100486278534,
      "learning_rate": 1.1207771760149412e-05,
      "loss": 0.354,
      "num_tokens": 1436341880.0,
      "step": 2268
    },
    {
      "epoch": 0.2682984509873478,
      "grad_norm": 0.15311993658542633,
      "learning_rate": 1.1189449132936016e-05,
      "loss": 0.3284,
      "num_tokens": 1436980764.0,
      "step": 2269
    },
    {
      "epoch": 0.2684166962279768,
      "grad_norm": 0.1685255914926529,
      "learning_rate": 1.1171155367490653e-05,
      "loss": 0.3593,
      "num_tokens": 1437607941.0,
      "step": 2270
    },
    {
      "epoch": 0.2685349414686059,
      "grad_norm": 0.15958444774150848,
      "learning_rate": 1.115289048802134e-05,
      "loss": 0.3655,
      "num_tokens": 1438242314.0,
      "step": 2271
    },
    {
      "epoch": 0.26865318670923494,
      "grad_norm": 0.165290966629982,
      "learning_rate": 1.1134654518697899e-05,
      "loss": 0.3474,
      "num_tokens": 1438871363.0,
      "step": 2272
    },
    {
      "epoch": 0.268771431949864,
      "grad_norm": 0.15361101925373077,
      "learning_rate": 1.1116447483651872e-05,
      "loss": 0.344,
      "num_tokens": 1439499872.0,
      "step": 2273
    },
    {
      "epoch": 0.26888967719049306,
      "grad_norm": 0.14768876135349274,
      "learning_rate": 1.1098269406976507e-05,
      "loss": 0.3208,
      "num_tokens": 1440133531.0,
      "step": 2274
    },
    {
      "epoch": 0.26900792243112215,
      "grad_norm": 0.14982637763023376,
      "learning_rate": 1.1080120312726752e-05,
      "loss": 0.3184,
      "num_tokens": 1440768688.0,
      "step": 2275
    },
    {
      "epoch": 0.26912616767175124,
      "grad_norm": 0.1513940989971161,
      "learning_rate": 1.1062000224919177e-05,
      "loss": 0.399,
      "num_tokens": 1441398656.0,
      "step": 2276
    },
    {
      "epoch": 0.26924441291238027,
      "grad_norm": 0.15859779715538025,
      "learning_rate": 1.1043909167532e-05,
      "loss": 0.3304,
      "num_tokens": 1442003989.0,
      "step": 2277
    },
    {
      "epoch": 0.26936265815300936,
      "grad_norm": 0.15063779056072235,
      "learning_rate": 1.1025847164504997e-05,
      "loss": 0.3779,
      "num_tokens": 1442643326.0,
      "step": 2278
    },
    {
      "epoch": 0.2694809033936384,
      "grad_norm": 0.1525985449552536,
      "learning_rate": 1.1007814239739506e-05,
      "loss": 0.3666,
      "num_tokens": 1443278975.0,
      "step": 2279
    },
    {
      "epoch": 0.2695991486342675,
      "grad_norm": 0.15241971611976624,
      "learning_rate": 1.098981041709839e-05,
      "loss": 0.3297,
      "num_tokens": 1443913604.0,
      "step": 2280
    },
    {
      "epoch": 0.2697173938748965,
      "grad_norm": 0.15431083738803864,
      "learning_rate": 1.0971835720405987e-05,
      "loss": 0.3506,
      "num_tokens": 1444549969.0,
      "step": 2281
    },
    {
      "epoch": 0.2698356391155256,
      "grad_norm": 0.14797011017799377,
      "learning_rate": 1.0953890173448117e-05,
      "loss": 0.3456,
      "num_tokens": 1445189491.0,
      "step": 2282
    },
    {
      "epoch": 0.26995388435615464,
      "grad_norm": 0.14169208705425262,
      "learning_rate": 1.0935973799972001e-05,
      "loss": 0.3292,
      "num_tokens": 1445820831.0,
      "step": 2283
    },
    {
      "epoch": 0.2700721295967837,
      "grad_norm": 0.15232716500759125,
      "learning_rate": 1.0918086623686275e-05,
      "loss": 0.3351,
      "num_tokens": 1446457219.0,
      "step": 2284
    },
    {
      "epoch": 0.2701903748374128,
      "grad_norm": 0.16128025949001312,
      "learning_rate": 1.0900228668260919e-05,
      "loss": 0.339,
      "num_tokens": 1447094737.0,
      "step": 2285
    },
    {
      "epoch": 0.27030862007804185,
      "grad_norm": 0.1798248142004013,
      "learning_rate": 1.0882399957327259e-05,
      "loss": 0.3763,
      "num_tokens": 1447727826.0,
      "step": 2286
    },
    {
      "epoch": 0.27042686531867094,
      "grad_norm": 0.15512481331825256,
      "learning_rate": 1.0864600514477925e-05,
      "loss": 0.3576,
      "num_tokens": 1448362348.0,
      "step": 2287
    },
    {
      "epoch": 0.2705451105593,
      "grad_norm": 0.1553766280412674,
      "learning_rate": 1.0846830363266808e-05,
      "loss": 0.3691,
      "num_tokens": 1448998191.0,
      "step": 2288
    },
    {
      "epoch": 0.27066335579992906,
      "grad_norm": 0.16342367231845856,
      "learning_rate": 1.0829089527209038e-05,
      "loss": 0.3467,
      "num_tokens": 1449624381.0,
      "step": 2289
    },
    {
      "epoch": 0.2707816010405581,
      "grad_norm": 0.1577490121126175,
      "learning_rate": 1.0811378029780954e-05,
      "loss": 0.3561,
      "num_tokens": 1450260226.0,
      "step": 2290
    },
    {
      "epoch": 0.2708998462811872,
      "grad_norm": 0.16117647290229797,
      "learning_rate": 1.0793695894420068e-05,
      "loss": 0.3552,
      "num_tokens": 1450890661.0,
      "step": 2291
    },
    {
      "epoch": 0.2710180915218163,
      "grad_norm": 0.1617818921804428,
      "learning_rate": 1.0776043144525048e-05,
      "loss": 0.3384,
      "num_tokens": 1451526642.0,
      "step": 2292
    },
    {
      "epoch": 0.2711363367624453,
      "grad_norm": 0.15790875256061554,
      "learning_rate": 1.0758419803455669e-05,
      "loss": 0.3605,
      "num_tokens": 1452159607.0,
      "step": 2293
    },
    {
      "epoch": 0.2712545820030744,
      "grad_norm": 0.15148423612117767,
      "learning_rate": 1.074082589453279e-05,
      "loss": 0.3745,
      "num_tokens": 1452797560.0,
      "step": 2294
    },
    {
      "epoch": 0.27137282724370343,
      "grad_norm": 0.15457701683044434,
      "learning_rate": 1.0723261441038317e-05,
      "loss": 0.3468,
      "num_tokens": 1453434624.0,
      "step": 2295
    },
    {
      "epoch": 0.2714910724843325,
      "grad_norm": 0.1459672898054123,
      "learning_rate": 1.0705726466215182e-05,
      "loss": 0.3503,
      "num_tokens": 1454067127.0,
      "step": 2296
    },
    {
      "epoch": 0.27160931772496155,
      "grad_norm": 0.1752711683511734,
      "learning_rate": 1.0688220993267324e-05,
      "loss": 0.3656,
      "num_tokens": 1454700667.0,
      "step": 2297
    },
    {
      "epoch": 0.27172756296559064,
      "grad_norm": 0.17267681658267975,
      "learning_rate": 1.0670745045359618e-05,
      "loss": 0.3873,
      "num_tokens": 1455337380.0,
      "step": 2298
    },
    {
      "epoch": 0.2718458082062197,
      "grad_norm": 0.14968270063400269,
      "learning_rate": 1.0653298645617877e-05,
      "loss": 0.3483,
      "num_tokens": 1455974456.0,
      "step": 2299
    },
    {
      "epoch": 0.27196405344684876,
      "grad_norm": 0.16421633958816528,
      "learning_rate": 1.063588181712881e-05,
      "loss": 0.3856,
      "num_tokens": 1456611249.0,
      "step": 2300
    },
    {
      "epoch": 0.27208229868747785,
      "grad_norm": 0.16939447820186615,
      "learning_rate": 1.0618494582940016e-05,
      "loss": 0.3393,
      "num_tokens": 1457243118.0,
      "step": 2301
    },
    {
      "epoch": 0.2722005439281069,
      "grad_norm": 0.15923595428466797,
      "learning_rate": 1.0601136966059902e-05,
      "loss": 0.3544,
      "num_tokens": 1457878878.0,
      "step": 2302
    },
    {
      "epoch": 0.272318789168736,
      "grad_norm": 0.17342090606689453,
      "learning_rate": 1.0583808989457697e-05,
      "loss": 0.3381,
      "num_tokens": 1458506707.0,
      "step": 2303
    },
    {
      "epoch": 0.272437034409365,
      "grad_norm": 0.17090991139411926,
      "learning_rate": 1.0566510676063415e-05,
      "loss": 0.3591,
      "num_tokens": 1459136085.0,
      "step": 2304
    },
    {
      "epoch": 0.2725552796499941,
      "grad_norm": 0.1702851802110672,
      "learning_rate": 1.0549242048767789e-05,
      "loss": 0.3738,
      "num_tokens": 1459775346.0,
      "step": 2305
    },
    {
      "epoch": 0.27267352489062313,
      "grad_norm": 0.1562279909849167,
      "learning_rate": 1.053200313042231e-05,
      "loss": 0.3837,
      "num_tokens": 1460412291.0,
      "step": 2306
    },
    {
      "epoch": 0.2727917701312522,
      "grad_norm": 0.14924395084381104,
      "learning_rate": 1.0514793943839119e-05,
      "loss": 0.322,
      "num_tokens": 1461048611.0,
      "step": 2307
    },
    {
      "epoch": 0.2729100153718813,
      "grad_norm": 0.15389004349708557,
      "learning_rate": 1.049761451179103e-05,
      "loss": 0.3586,
      "num_tokens": 1461677729.0,
      "step": 2308
    },
    {
      "epoch": 0.27302826061251034,
      "grad_norm": 0.16182515025138855,
      "learning_rate": 1.0480464857011482e-05,
      "loss": 0.37,
      "num_tokens": 1462315617.0,
      "step": 2309
    },
    {
      "epoch": 0.27314650585313943,
      "grad_norm": 0.14908406138420105,
      "learning_rate": 1.0463345002194496e-05,
      "loss": 0.3185,
      "num_tokens": 1462930982.0,
      "step": 2310
    },
    {
      "epoch": 0.27326475109376847,
      "grad_norm": 0.16309335827827454,
      "learning_rate": 1.044625496999469e-05,
      "loss": 0.3491,
      "num_tokens": 1463566971.0,
      "step": 2311
    },
    {
      "epoch": 0.27338299633439755,
      "grad_norm": 0.1465187668800354,
      "learning_rate": 1.0429194783027187e-05,
      "loss": 0.3295,
      "num_tokens": 1464202172.0,
      "step": 2312
    },
    {
      "epoch": 0.2735012415750266,
      "grad_norm": 0.15645480155944824,
      "learning_rate": 1.0412164463867636e-05,
      "loss": 0.3839,
      "num_tokens": 1464838092.0,
      "step": 2313
    },
    {
      "epoch": 0.2736194868156557,
      "grad_norm": 0.1486484557390213,
      "learning_rate": 1.0395164035052143e-05,
      "loss": 0.3429,
      "num_tokens": 1465477610.0,
      "step": 2314
    },
    {
      "epoch": 0.2737377320562847,
      "grad_norm": 0.16061237454414368,
      "learning_rate": 1.0378193519077273e-05,
      "loss": 0.3488,
      "num_tokens": 1466116332.0,
      "step": 2315
    },
    {
      "epoch": 0.2738559772969138,
      "grad_norm": 0.17339017987251282,
      "learning_rate": 1.0361252938400012e-05,
      "loss": 0.3766,
      "num_tokens": 1466749671.0,
      "step": 2316
    },
    {
      "epoch": 0.2739742225375429,
      "grad_norm": 0.16877548396587372,
      "learning_rate": 1.0344342315437721e-05,
      "loss": 0.3971,
      "num_tokens": 1467388729.0,
      "step": 2317
    },
    {
      "epoch": 0.2740924677781719,
      "grad_norm": 0.14575250446796417,
      "learning_rate": 1.0327461672568127e-05,
      "loss": 0.3433,
      "num_tokens": 1468022678.0,
      "step": 2318
    },
    {
      "epoch": 0.274210713018801,
      "grad_norm": 0.16383600234985352,
      "learning_rate": 1.0310611032129282e-05,
      "loss": 0.3782,
      "num_tokens": 1468653588.0,
      "step": 2319
    },
    {
      "epoch": 0.27432895825943004,
      "grad_norm": 0.15298399329185486,
      "learning_rate": 1.0293790416419522e-05,
      "loss": 0.3774,
      "num_tokens": 1469292071.0,
      "step": 2320
    },
    {
      "epoch": 0.27444720350005913,
      "grad_norm": 0.15629802644252777,
      "learning_rate": 1.0276999847697484e-05,
      "loss": 0.3509,
      "num_tokens": 1469926130.0,
      "step": 2321
    },
    {
      "epoch": 0.27456544874068817,
      "grad_norm": 0.1455887109041214,
      "learning_rate": 1.0260239348182013e-05,
      "loss": 0.3298,
      "num_tokens": 1470565682.0,
      "step": 2322
    },
    {
      "epoch": 0.27468369398131726,
      "grad_norm": 0.15099728107452393,
      "learning_rate": 1.0243508940052179e-05,
      "loss": 0.3377,
      "num_tokens": 1471199661.0,
      "step": 2323
    },
    {
      "epoch": 0.2748019392219463,
      "grad_norm": 0.15650267899036407,
      "learning_rate": 1.0226808645447229e-05,
      "loss": 0.3179,
      "num_tokens": 1471837079.0,
      "step": 2324
    },
    {
      "epoch": 0.2749201844625754,
      "grad_norm": 0.15078777074813843,
      "learning_rate": 1.0210138486466549e-05,
      "loss": 0.3531,
      "num_tokens": 1472470748.0,
      "step": 2325
    },
    {
      "epoch": 0.27503842970320447,
      "grad_norm": 0.14938890933990479,
      "learning_rate": 1.0193498485169674e-05,
      "loss": 0.3335,
      "num_tokens": 1473076369.0,
      "step": 2326
    },
    {
      "epoch": 0.2751566749438335,
      "grad_norm": 0.17257998883724213,
      "learning_rate": 1.0176888663576212e-05,
      "loss": 0.344,
      "num_tokens": 1473713052.0,
      "step": 2327
    },
    {
      "epoch": 0.2752749201844626,
      "grad_norm": 0.16291533410549164,
      "learning_rate": 1.0160309043665832e-05,
      "loss": 0.3318,
      "num_tokens": 1474350921.0,
      "step": 2328
    },
    {
      "epoch": 0.2753931654250916,
      "grad_norm": 0.15999779105186462,
      "learning_rate": 1.0143759647378238e-05,
      "loss": 0.4077,
      "num_tokens": 1474987408.0,
      "step": 2329
    },
    {
      "epoch": 0.2755114106657207,
      "grad_norm": 0.1630454659461975,
      "learning_rate": 1.0127240496613158e-05,
      "loss": 0.3422,
      "num_tokens": 1475621528.0,
      "step": 2330
    },
    {
      "epoch": 0.27562965590634975,
      "grad_norm": 0.15796898305416107,
      "learning_rate": 1.0110751613230277e-05,
      "loss": 0.3445,
      "num_tokens": 1476256968.0,
      "step": 2331
    },
    {
      "epoch": 0.27574790114697884,
      "grad_norm": 0.16335657238960266,
      "learning_rate": 1.0094293019049237e-05,
      "loss": 0.321,
      "num_tokens": 1476887629.0,
      "step": 2332
    },
    {
      "epoch": 0.2758661463876079,
      "grad_norm": 0.14884811639785767,
      "learning_rate": 1.0077864735849589e-05,
      "loss": 0.3735,
      "num_tokens": 1477522575.0,
      "step": 2333
    },
    {
      "epoch": 0.27598439162823696,
      "grad_norm": 0.15638722479343414,
      "learning_rate": 1.0061466785370775e-05,
      "loss": 0.3373,
      "num_tokens": 1478158656.0,
      "step": 2334
    },
    {
      "epoch": 0.27610263686886605,
      "grad_norm": 0.16069157421588898,
      "learning_rate": 1.0045099189312113e-05,
      "loss": 0.3649,
      "num_tokens": 1478794818.0,
      "step": 2335
    },
    {
      "epoch": 0.2762208821094951,
      "grad_norm": 0.14972923696041107,
      "learning_rate": 1.002876196933274e-05,
      "loss": 0.3375,
      "num_tokens": 1479428236.0,
      "step": 2336
    },
    {
      "epoch": 0.27633912735012417,
      "grad_norm": 0.153402641415596,
      "learning_rate": 1.0012455147051596e-05,
      "loss": 0.3564,
      "num_tokens": 1480061891.0,
      "step": 2337
    },
    {
      "epoch": 0.2764573725907532,
      "grad_norm": 0.16654513776302338,
      "learning_rate": 9.9961787440474e-06,
      "loss": 0.3946,
      "num_tokens": 1480701365.0,
      "step": 2338
    },
    {
      "epoch": 0.2765756178313823,
      "grad_norm": 0.1553163081407547,
      "learning_rate": 9.979932781858607e-06,
      "loss": 0.337,
      "num_tokens": 1481336637.0,
      "step": 2339
    },
    {
      "epoch": 0.2766938630720113,
      "grad_norm": 0.1627359241247177,
      "learning_rate": 9.963717281983415e-06,
      "loss": 0.3462,
      "num_tokens": 1481974800.0,
      "step": 2340
    },
    {
      "epoch": 0.2768121083126404,
      "grad_norm": 0.1517624706029892,
      "learning_rate": 9.947532265879685e-06,
      "loss": 0.3314,
      "num_tokens": 1482607289.0,
      "step": 2341
    },
    {
      "epoch": 0.2769303535532695,
      "grad_norm": 0.16129785776138306,
      "learning_rate": 9.931377754964957e-06,
      "loss": 0.3646,
      "num_tokens": 1483243948.0,
      "step": 2342
    },
    {
      "epoch": 0.27704859879389854,
      "grad_norm": 38.480674743652344,
      "learning_rate": 9.915253770616389e-06,
      "loss": 1.4308,
      "num_tokens": 1483845955.0,
      "step": 2343
    },
    {
      "epoch": 0.2771668440345276,
      "grad_norm": 0.16987748444080353,
      "learning_rate": 9.899160334170746e-06,
      "loss": 0.3543,
      "num_tokens": 1484479960.0,
      "step": 2344
    },
    {
      "epoch": 0.27728508927515666,
      "grad_norm": 0.1605212539434433,
      "learning_rate": 9.883097466924388e-06,
      "loss": 0.3566,
      "num_tokens": 1485115529.0,
      "step": 2345
    },
    {
      "epoch": 0.27740333451578575,
      "grad_norm": 0.15978476405143738,
      "learning_rate": 9.867065190133201e-06,
      "loss": 0.3562,
      "num_tokens": 1485751394.0,
      "step": 2346
    },
    {
      "epoch": 0.2775215797564148,
      "grad_norm": 0.15629255771636963,
      "learning_rate": 9.851063525012598e-06,
      "loss": 0.3737,
      "num_tokens": 1486390225.0,
      "step": 2347
    },
    {
      "epoch": 0.27763982499704387,
      "grad_norm": 0.1548636257648468,
      "learning_rate": 9.835092492737483e-06,
      "loss": 0.3469,
      "num_tokens": 1486988102.0,
      "step": 2348
    },
    {
      "epoch": 0.27775807023767296,
      "grad_norm": 0.15224124491214752,
      "learning_rate": 9.819152114442225e-06,
      "loss": 0.3431,
      "num_tokens": 1487623515.0,
      "step": 2349
    },
    {
      "epoch": 0.277876315478302,
      "grad_norm": 0.18440452218055725,
      "learning_rate": 9.803242411220631e-06,
      "loss": 0.3828,
      "num_tokens": 1488259828.0,
      "step": 2350
    },
    {
      "epoch": 0.2779945607189311,
      "grad_norm": 0.14934492111206055,
      "learning_rate": 9.78736340412592e-06,
      "loss": 0.3422,
      "num_tokens": 1488891872.0,
      "step": 2351
    },
    {
      "epoch": 0.2781128059595601,
      "grad_norm": 0.16854149103164673,
      "learning_rate": 9.771515114170672e-06,
      "loss": 0.4027,
      "num_tokens": 1489527101.0,
      "step": 2352
    },
    {
      "epoch": 0.2782310512001892,
      "grad_norm": 0.1552664041519165,
      "learning_rate": 9.755697562326836e-06,
      "loss": 0.3541,
      "num_tokens": 1490165911.0,
      "step": 2353
    },
    {
      "epoch": 0.27834929644081824,
      "grad_norm": 0.14984580874443054,
      "learning_rate": 9.73991076952568e-06,
      "loss": 0.3422,
      "num_tokens": 1490802923.0,
      "step": 2354
    },
    {
      "epoch": 0.27846754168144733,
      "grad_norm": 0.1623169183731079,
      "learning_rate": 9.724154756657781e-06,
      "loss": 0.3602,
      "num_tokens": 1491439721.0,
      "step": 2355
    },
    {
      "epoch": 0.27858578692207636,
      "grad_norm": 0.15729185938835144,
      "learning_rate": 9.70842954457296e-06,
      "loss": 0.3458,
      "num_tokens": 1492078820.0,
      "step": 2356
    },
    {
      "epoch": 0.27870403216270545,
      "grad_norm": 0.15174199640750885,
      "learning_rate": 9.692735154080303e-06,
      "loss": 0.3309,
      "num_tokens": 1492714650.0,
      "step": 2357
    },
    {
      "epoch": 0.27882227740333454,
      "grad_norm": 0.1500135362148285,
      "learning_rate": 9.677071605948097e-06,
      "loss": 0.3087,
      "num_tokens": 1493341798.0,
      "step": 2358
    },
    {
      "epoch": 0.2789405226439636,
      "grad_norm": 0.16156241297721863,
      "learning_rate": 9.661438920903808e-06,
      "loss": 0.3864,
      "num_tokens": 1493978267.0,
      "step": 2359
    },
    {
      "epoch": 0.27905876788459266,
      "grad_norm": 0.14532062411308289,
      "learning_rate": 9.645837119634094e-06,
      "loss": 0.3069,
      "num_tokens": 1494612569.0,
      "step": 2360
    },
    {
      "epoch": 0.2791770131252217,
      "grad_norm": 0.16516432166099548,
      "learning_rate": 9.630266222784709e-06,
      "loss": 0.3665,
      "num_tokens": 1495250024.0,
      "step": 2361
    },
    {
      "epoch": 0.2792952583658508,
      "grad_norm": 0.15118518471717834,
      "learning_rate": 9.61472625096053e-06,
      "loss": 0.3517,
      "num_tokens": 1495879619.0,
      "step": 2362
    },
    {
      "epoch": 0.2794135036064798,
      "grad_norm": 0.15301978588104248,
      "learning_rate": 9.599217224725496e-06,
      "loss": 0.3815,
      "num_tokens": 1496509723.0,
      "step": 2363
    },
    {
      "epoch": 0.2795317488471089,
      "grad_norm": 0.1620192974805832,
      "learning_rate": 9.583739164602621e-06,
      "loss": 0.3602,
      "num_tokens": 1497139086.0,
      "step": 2364
    },
    {
      "epoch": 0.27964999408773794,
      "grad_norm": 0.14597781002521515,
      "learning_rate": 9.568292091073922e-06,
      "loss": 0.3078,
      "num_tokens": 1497769353.0,
      "step": 2365
    },
    {
      "epoch": 0.27976823932836703,
      "grad_norm": 0.14736135303974152,
      "learning_rate": 9.552876024580412e-06,
      "loss": 0.3148,
      "num_tokens": 1498406713.0,
      "step": 2366
    },
    {
      "epoch": 0.2798864845689961,
      "grad_norm": 0.16254974901676178,
      "learning_rate": 9.537490985522081e-06,
      "loss": 0.3268,
      "num_tokens": 1499042860.0,
      "step": 2367
    },
    {
      "epoch": 0.28000472980962515,
      "grad_norm": 0.14526546001434326,
      "learning_rate": 9.52213699425785e-06,
      "loss": 0.3151,
      "num_tokens": 1499673955.0,
      "step": 2368
    },
    {
      "epoch": 0.28012297505025424,
      "grad_norm": 0.13753469288349152,
      "learning_rate": 9.506814071105575e-06,
      "loss": 0.3189,
      "num_tokens": 1500308786.0,
      "step": 2369
    },
    {
      "epoch": 0.2802412202908833,
      "grad_norm": 0.14920324087142944,
      "learning_rate": 9.491522236341976e-06,
      "loss": 0.316,
      "num_tokens": 1500943668.0,
      "step": 2370
    },
    {
      "epoch": 0.28035946553151236,
      "grad_norm": 0.14196032285690308,
      "learning_rate": 9.476261510202648e-06,
      "loss": 0.3167,
      "num_tokens": 1501575125.0,
      "step": 2371
    },
    {
      "epoch": 0.2804777107721414,
      "grad_norm": 0.14756308495998383,
      "learning_rate": 9.461031912882011e-06,
      "loss": 0.3828,
      "num_tokens": 1502205209.0,
      "step": 2372
    },
    {
      "epoch": 0.2805959560127705,
      "grad_norm": 0.15238507091999054,
      "learning_rate": 9.4458334645333e-06,
      "loss": 0.3794,
      "num_tokens": 1502841700.0,
      "step": 2373
    },
    {
      "epoch": 0.2807142012533996,
      "grad_norm": 0.14375653862953186,
      "learning_rate": 9.430666185268535e-06,
      "loss": 0.339,
      "num_tokens": 1503478757.0,
      "step": 2374
    },
    {
      "epoch": 0.2808324464940286,
      "grad_norm": 0.14228053390979767,
      "learning_rate": 9.415530095158473e-06,
      "loss": 0.3186,
      "num_tokens": 1504114367.0,
      "step": 2375
    },
    {
      "epoch": 0.2809506917346577,
      "grad_norm": 0.15334486961364746,
      "learning_rate": 9.400425214232616e-06,
      "loss": 0.3335,
      "num_tokens": 1504750346.0,
      "step": 2376
    },
    {
      "epoch": 0.28106893697528673,
      "grad_norm": 0.1638639122247696,
      "learning_rate": 9.385351562479158e-06,
      "loss": 0.3512,
      "num_tokens": 1505373996.0,
      "step": 2377
    },
    {
      "epoch": 0.2811871822159158,
      "grad_norm": 0.14417004585266113,
      "learning_rate": 9.370309159844967e-06,
      "loss": 0.3221,
      "num_tokens": 1506006742.0,
      "step": 2378
    },
    {
      "epoch": 0.28130542745654485,
      "grad_norm": 0.15065158903598785,
      "learning_rate": 9.35529802623557e-06,
      "loss": 0.3403,
      "num_tokens": 1506642591.0,
      "step": 2379
    },
    {
      "epoch": 0.28142367269717394,
      "grad_norm": 0.15715816617012024,
      "learning_rate": 9.3403181815151e-06,
      "loss": 0.3177,
      "num_tokens": 1507279859.0,
      "step": 2380
    },
    {
      "epoch": 0.281541917937803,
      "grad_norm": 0.16456620395183563,
      "learning_rate": 9.325369645506298e-06,
      "loss": 0.3342,
      "num_tokens": 1507913520.0,
      "step": 2381
    },
    {
      "epoch": 0.28166016317843207,
      "grad_norm": 0.15010853111743927,
      "learning_rate": 9.31045243799047e-06,
      "loss": 0.3382,
      "num_tokens": 1508547624.0,
      "step": 2382
    },
    {
      "epoch": 0.28177840841906115,
      "grad_norm": 0.16282407939434052,
      "learning_rate": 9.295566578707456e-06,
      "loss": 0.3657,
      "num_tokens": 1509186196.0,
      "step": 2383
    },
    {
      "epoch": 0.2818966536596902,
      "grad_norm": 0.14777715504169464,
      "learning_rate": 9.280712087355635e-06,
      "loss": 0.337,
      "num_tokens": 1509824371.0,
      "step": 2384
    },
    {
      "epoch": 0.2820148989003193,
      "grad_norm": 0.1608475148677826,
      "learning_rate": 9.265888983591861e-06,
      "loss": 0.3411,
      "num_tokens": 1510453337.0,
      "step": 2385
    },
    {
      "epoch": 0.2821331441409483,
      "grad_norm": 0.15267649292945862,
      "learning_rate": 9.251097287031453e-06,
      "loss": 0.323,
      "num_tokens": 1511089984.0,
      "step": 2386
    },
    {
      "epoch": 0.2822513893815774,
      "grad_norm": 0.14557266235351562,
      "learning_rate": 9.23633701724817e-06,
      "loss": 0.3359,
      "num_tokens": 1511723104.0,
      "step": 2387
    },
    {
      "epoch": 0.28236963462220643,
      "grad_norm": 0.15417338907718658,
      "learning_rate": 9.22160819377419e-06,
      "loss": 0.3459,
      "num_tokens": 1512361915.0,
      "step": 2388
    },
    {
      "epoch": 0.2824878798628355,
      "grad_norm": 0.15560688078403473,
      "learning_rate": 9.206910836100078e-06,
      "loss": 0.3907,
      "num_tokens": 1512989309.0,
      "step": 2389
    },
    {
      "epoch": 0.2826061251034646,
      "grad_norm": 0.1650230437517166,
      "learning_rate": 9.19224496367475e-06,
      "loss": 0.3583,
      "num_tokens": 1513623233.0,
      "step": 2390
    },
    {
      "epoch": 0.28272437034409365,
      "grad_norm": 0.16392244398593903,
      "learning_rate": 9.177610595905472e-06,
      "loss": 0.3566,
      "num_tokens": 1514257994.0,
      "step": 2391
    },
    {
      "epoch": 0.28284261558472273,
      "grad_norm": 0.16251103579998016,
      "learning_rate": 9.16300775215781e-06,
      "loss": 0.3546,
      "num_tokens": 1514895129.0,
      "step": 2392
    },
    {
      "epoch": 0.28296086082535177,
      "grad_norm": 0.15515641868114471,
      "learning_rate": 9.148436451755609e-06,
      "loss": 0.3172,
      "num_tokens": 1515533555.0,
      "step": 2393
    },
    {
      "epoch": 0.28307910606598086,
      "grad_norm": 0.14485909044742584,
      "learning_rate": 9.133896713980998e-06,
      "loss": 0.3138,
      "num_tokens": 1516170546.0,
      "step": 2394
    },
    {
      "epoch": 0.2831973513066099,
      "grad_norm": 0.14556768536567688,
      "learning_rate": 9.119388558074312e-06,
      "loss": 0.2966,
      "num_tokens": 1516799730.0,
      "step": 2395
    },
    {
      "epoch": 0.283315596547239,
      "grad_norm": 0.17382657527923584,
      "learning_rate": 9.104912003234105e-06,
      "loss": 0.3483,
      "num_tokens": 1517436780.0,
      "step": 2396
    },
    {
      "epoch": 0.283433841787868,
      "grad_norm": 0.15845876932144165,
      "learning_rate": 9.09046706861711e-06,
      "loss": 0.3901,
      "num_tokens": 1518072732.0,
      "step": 2397
    },
    {
      "epoch": 0.2835520870284971,
      "grad_norm": 0.15543769299983978,
      "learning_rate": 9.076053773338229e-06,
      "loss": 0.3586,
      "num_tokens": 1518710901.0,
      "step": 2398
    },
    {
      "epoch": 0.2836703322691262,
      "grad_norm": 0.15003569424152374,
      "learning_rate": 9.061672136470488e-06,
      "loss": 0.3336,
      "num_tokens": 1519345313.0,
      "step": 2399
    },
    {
      "epoch": 0.2837885775097552,
      "grad_norm": 0.1602921187877655,
      "learning_rate": 9.047322177045008e-06,
      "loss": 0.3591,
      "num_tokens": 1519980264.0,
      "step": 2400
    },
    {
      "epoch": 0.2839068227503843,
      "grad_norm": 0.17690017819404602,
      "learning_rate": 9.033003914051014e-06,
      "loss": 0.3711,
      "num_tokens": 1520615916.0,
      "step": 2401
    },
    {
      "epoch": 0.28402506799101335,
      "grad_norm": 0.1696273237466812,
      "learning_rate": 9.018717366435759e-06,
      "loss": 0.3715,
      "num_tokens": 1521252102.0,
      "step": 2402
    },
    {
      "epoch": 0.28414331323164244,
      "grad_norm": 0.14623281359672546,
      "learning_rate": 9.004462553104562e-06,
      "loss": 0.3344,
      "num_tokens": 1521879862.0,
      "step": 2403
    },
    {
      "epoch": 0.28426155847227147,
      "grad_norm": 0.1635226607322693,
      "learning_rate": 8.990239492920721e-06,
      "loss": 0.3781,
      "num_tokens": 1522505201.0,
      "step": 2404
    },
    {
      "epoch": 0.28437980371290056,
      "grad_norm": 0.143572598695755,
      "learning_rate": 8.976048204705527e-06,
      "loss": 0.3311,
      "num_tokens": 1523144696.0,
      "step": 2405
    },
    {
      "epoch": 0.28449804895352965,
      "grad_norm": 0.14635297656059265,
      "learning_rate": 8.961888707238222e-06,
      "loss": 0.3456,
      "num_tokens": 1523776301.0,
      "step": 2406
    },
    {
      "epoch": 0.2846162941941587,
      "grad_norm": 0.15382792055606842,
      "learning_rate": 8.94776101925598e-06,
      "loss": 0.3524,
      "num_tokens": 1524411665.0,
      "step": 2407
    },
    {
      "epoch": 0.28473453943478777,
      "grad_norm": 0.15080878138542175,
      "learning_rate": 8.93366515945389e-06,
      "loss": 0.3482,
      "num_tokens": 1525049675.0,
      "step": 2408
    },
    {
      "epoch": 0.2848527846754168,
      "grad_norm": 0.14847759902477264,
      "learning_rate": 8.919601146484912e-06,
      "loss": 0.3085,
      "num_tokens": 1525681212.0,
      "step": 2409
    },
    {
      "epoch": 0.2849710299160459,
      "grad_norm": 0.15085552632808685,
      "learning_rate": 8.90556899895987e-06,
      "loss": 0.3404,
      "num_tokens": 1526316616.0,
      "step": 2410
    },
    {
      "epoch": 0.2850892751566749,
      "grad_norm": 0.15188775956630707,
      "learning_rate": 8.891568735447421e-06,
      "loss": 0.3498,
      "num_tokens": 1526950609.0,
      "step": 2411
    },
    {
      "epoch": 0.285207520397304,
      "grad_norm": 0.1585957109928131,
      "learning_rate": 8.877600374474018e-06,
      "loss": 0.3697,
      "num_tokens": 1527587441.0,
      "step": 2412
    },
    {
      "epoch": 0.28532576563793305,
      "grad_norm": 0.15482857823371887,
      "learning_rate": 8.863663934523919e-06,
      "loss": 0.3782,
      "num_tokens": 1528223473.0,
      "step": 2413
    },
    {
      "epoch": 0.28544401087856214,
      "grad_norm": 0.14563055336475372,
      "learning_rate": 8.849759434039126e-06,
      "loss": 0.3582,
      "num_tokens": 1528862184.0,
      "step": 2414
    },
    {
      "epoch": 0.2855622561191912,
      "grad_norm": 0.1472221165895462,
      "learning_rate": 8.835886891419377e-06,
      "loss": 0.3469,
      "num_tokens": 1529498134.0,
      "step": 2415
    },
    {
      "epoch": 0.28568050135982026,
      "grad_norm": 0.15884767472743988,
      "learning_rate": 8.822046325022123e-06,
      "loss": 0.3557,
      "num_tokens": 1530132720.0,
      "step": 2416
    },
    {
      "epoch": 0.28579874660044935,
      "grad_norm": 0.14727573096752167,
      "learning_rate": 8.808237753162495e-06,
      "loss": 0.34,
      "num_tokens": 1530766312.0,
      "step": 2417
    },
    {
      "epoch": 0.2859169918410784,
      "grad_norm": 0.15706580877304077,
      "learning_rate": 8.794461194113302e-06,
      "loss": 0.3628,
      "num_tokens": 1531401703.0,
      "step": 2418
    },
    {
      "epoch": 0.28603523708170747,
      "grad_norm": 0.1499369889497757,
      "learning_rate": 8.780716666104976e-06,
      "loss": 0.3531,
      "num_tokens": 1532033358.0,
      "step": 2419
    },
    {
      "epoch": 0.2861534823223365,
      "grad_norm": 0.16059350967407227,
      "learning_rate": 8.767004187325567e-06,
      "loss": 0.3328,
      "num_tokens": 1532666946.0,
      "step": 2420
    },
    {
      "epoch": 0.2862717275629656,
      "grad_norm": 0.15611490607261658,
      "learning_rate": 8.75332377592071e-06,
      "loss": 0.3531,
      "num_tokens": 1533305486.0,
      "step": 2421
    },
    {
      "epoch": 0.28638997280359463,
      "grad_norm": 0.16063754260540009,
      "learning_rate": 8.739675449993611e-06,
      "loss": 0.3369,
      "num_tokens": 1533935802.0,
      "step": 2422
    },
    {
      "epoch": 0.2865082180442237,
      "grad_norm": 0.17527389526367188,
      "learning_rate": 8.726059227605018e-06,
      "loss": 0.3663,
      "num_tokens": 1534537936.0,
      "step": 2423
    },
    {
      "epoch": 0.2866264632848528,
      "grad_norm": 0.15854455530643463,
      "learning_rate": 8.712475126773192e-06,
      "loss": 0.3627,
      "num_tokens": 1535176503.0,
      "step": 2424
    },
    {
      "epoch": 0.28674470852548184,
      "grad_norm": 0.1538389027118683,
      "learning_rate": 8.698923165473894e-06,
      "loss": 0.306,
      "num_tokens": 1535809878.0,
      "step": 2425
    },
    {
      "epoch": 0.28686295376611093,
      "grad_norm": 0.14468614757061005,
      "learning_rate": 8.685403361640349e-06,
      "loss": 0.3063,
      "num_tokens": 1536444197.0,
      "step": 2426
    },
    {
      "epoch": 0.28698119900673996,
      "grad_norm": 0.15404705703258514,
      "learning_rate": 8.671915733163223e-06,
      "loss": 0.3336,
      "num_tokens": 1537077940.0,
      "step": 2427
    },
    {
      "epoch": 0.28709944424736905,
      "grad_norm": 0.14619845151901245,
      "learning_rate": 8.658460297890626e-06,
      "loss": 0.3092,
      "num_tokens": 1537709405.0,
      "step": 2428
    },
    {
      "epoch": 0.2872176894879981,
      "grad_norm": 0.1538306325674057,
      "learning_rate": 8.645037073628046e-06,
      "loss": 0.3657,
      "num_tokens": 1538346163.0,
      "step": 2429
    },
    {
      "epoch": 0.2873359347286272,
      "grad_norm": 0.1490619033575058,
      "learning_rate": 8.631646078138352e-06,
      "loss": 0.4068,
      "num_tokens": 1538982802.0,
      "step": 2430
    },
    {
      "epoch": 0.28745417996925626,
      "grad_norm": 0.1757480949163437,
      "learning_rate": 8.618287329141761e-06,
      "loss": 0.3608,
      "num_tokens": 1539617968.0,
      "step": 2431
    },
    {
      "epoch": 0.2875724252098853,
      "grad_norm": 0.14783507585525513,
      "learning_rate": 8.604960844315839e-06,
      "loss": 0.3197,
      "num_tokens": 1540248847.0,
      "step": 2432
    },
    {
      "epoch": 0.2876906704505144,
      "grad_norm": 0.14284883439540863,
      "learning_rate": 8.59166664129543e-06,
      "loss": 0.3541,
      "num_tokens": 1540884115.0,
      "step": 2433
    },
    {
      "epoch": 0.2878089156911434,
      "grad_norm": 0.16350941359996796,
      "learning_rate": 8.578404737672674e-06,
      "loss": 0.3229,
      "num_tokens": 1541519078.0,
      "step": 2434
    },
    {
      "epoch": 0.2879271609317725,
      "grad_norm": 0.1476208120584488,
      "learning_rate": 8.565175150996966e-06,
      "loss": 0.3008,
      "num_tokens": 1542154829.0,
      "step": 2435
    },
    {
      "epoch": 0.28804540617240154,
      "grad_norm": 0.14756110310554504,
      "learning_rate": 8.55197789877493e-06,
      "loss": 0.337,
      "num_tokens": 1542787615.0,
      "step": 2436
    },
    {
      "epoch": 0.28816365141303063,
      "grad_norm": 0.14862175285816193,
      "learning_rate": 8.53881299847042e-06,
      "loss": 0.3599,
      "num_tokens": 1543422823.0,
      "step": 2437
    },
    {
      "epoch": 0.28828189665365966,
      "grad_norm": 0.1573106050491333,
      "learning_rate": 8.525680467504469e-06,
      "loss": 0.3862,
      "num_tokens": 1544061139.0,
      "step": 2438
    },
    {
      "epoch": 0.28840014189428875,
      "grad_norm": 0.15238477289676666,
      "learning_rate": 8.512580323255266e-06,
      "loss": 0.3918,
      "num_tokens": 1544680188.0,
      "step": 2439
    },
    {
      "epoch": 0.28851838713491784,
      "grad_norm": 0.17096689343452454,
      "learning_rate": 8.499512583058153e-06,
      "loss": 0.4132,
      "num_tokens": 1545317687.0,
      "step": 2440
    },
    {
      "epoch": 0.2886366323755469,
      "grad_norm": 0.15198521316051483,
      "learning_rate": 8.486477264205592e-06,
      "loss": 0.3426,
      "num_tokens": 1545950953.0,
      "step": 2441
    },
    {
      "epoch": 0.28875487761617596,
      "grad_norm": 0.1503327190876007,
      "learning_rate": 8.473474383947138e-06,
      "loss": 0.3267,
      "num_tokens": 1546570745.0,
      "step": 2442
    },
    {
      "epoch": 0.288873122856805,
      "grad_norm": 0.1454770714044571,
      "learning_rate": 8.46050395948943e-06,
      "loss": 0.3253,
      "num_tokens": 1547208758.0,
      "step": 2443
    },
    {
      "epoch": 0.2889913680974341,
      "grad_norm": 0.16143880784511566,
      "learning_rate": 8.447566007996142e-06,
      "loss": 0.3414,
      "num_tokens": 1547841621.0,
      "step": 2444
    },
    {
      "epoch": 0.2891096133380631,
      "grad_norm": 0.16132719814777374,
      "learning_rate": 8.434660546587994e-06,
      "loss": 0.3751,
      "num_tokens": 1548474236.0,
      "step": 2445
    },
    {
      "epoch": 0.2892278585786922,
      "grad_norm": 0.15949557721614838,
      "learning_rate": 8.421787592342694e-06,
      "loss": 0.3318,
      "num_tokens": 1549102957.0,
      "step": 2446
    },
    {
      "epoch": 0.2893461038193213,
      "grad_norm": 0.1593041867017746,
      "learning_rate": 8.40894716229495e-06,
      "loss": 0.3544,
      "num_tokens": 1549738034.0,
      "step": 2447
    },
    {
      "epoch": 0.28946434905995033,
      "grad_norm": 0.15824440121650696,
      "learning_rate": 8.396139273436427e-06,
      "loss": 0.3356,
      "num_tokens": 1550367757.0,
      "step": 2448
    },
    {
      "epoch": 0.2895825943005794,
      "grad_norm": 0.14293761551380157,
      "learning_rate": 8.383363942715722e-06,
      "loss": 0.3418,
      "num_tokens": 1551002943.0,
      "step": 2449
    },
    {
      "epoch": 0.28970083954120845,
      "grad_norm": 0.155470073223114,
      "learning_rate": 8.37062118703835e-06,
      "loss": 0.332,
      "num_tokens": 1551640916.0,
      "step": 2450
    },
    {
      "epoch": 0.28981908478183754,
      "grad_norm": 0.16884839534759521,
      "learning_rate": 8.357911023266716e-06,
      "loss": 0.3752,
      "num_tokens": 1552279719.0,
      "step": 2451
    },
    {
      "epoch": 0.2899373300224666,
      "grad_norm": 0.15490181744098663,
      "learning_rate": 8.34523346822012e-06,
      "loss": 0.3553,
      "num_tokens": 1552909547.0,
      "step": 2452
    },
    {
      "epoch": 0.29005557526309567,
      "grad_norm": 0.1494501233100891,
      "learning_rate": 8.332588538674671e-06,
      "loss": 0.3218,
      "num_tokens": 1553547378.0,
      "step": 2453
    },
    {
      "epoch": 0.2901738205037247,
      "grad_norm": 0.16306337714195251,
      "learning_rate": 8.319976251363345e-06,
      "loss": 0.3499,
      "num_tokens": 1554186991.0,
      "step": 2454
    },
    {
      "epoch": 0.2902920657443538,
      "grad_norm": 0.16215814650058746,
      "learning_rate": 8.307396622975894e-06,
      "loss": 0.3755,
      "num_tokens": 1554823738.0,
      "step": 2455
    },
    {
      "epoch": 0.2904103109849829,
      "grad_norm": 0.14529335498809814,
      "learning_rate": 8.294849670158859e-06,
      "loss": 0.3713,
      "num_tokens": 1555463056.0,
      "step": 2456
    },
    {
      "epoch": 0.2905285562256119,
      "grad_norm": 0.18343862891197205,
      "learning_rate": 8.282335409515559e-06,
      "loss": 0.3582,
      "num_tokens": 1556098586.0,
      "step": 2457
    },
    {
      "epoch": 0.290646801466241,
      "grad_norm": 0.15133658051490784,
      "learning_rate": 8.26985385760603e-06,
      "loss": 0.3536,
      "num_tokens": 1556734027.0,
      "step": 2458
    },
    {
      "epoch": 0.29076504670687003,
      "grad_norm": 0.16418562829494476,
      "learning_rate": 8.257405030947032e-06,
      "loss": 0.3539,
      "num_tokens": 1557369690.0,
      "step": 2459
    },
    {
      "epoch": 0.2908832919474991,
      "grad_norm": 0.1541043072938919,
      "learning_rate": 8.244988946012026e-06,
      "loss": 0.3597,
      "num_tokens": 1557999303.0,
      "step": 2460
    },
    {
      "epoch": 0.29100153718812816,
      "grad_norm": 0.17861461639404297,
      "learning_rate": 8.232605619231132e-06,
      "loss": 0.3616,
      "num_tokens": 1558634807.0,
      "step": 2461
    },
    {
      "epoch": 0.29111978242875725,
      "grad_norm": 0.1503981500864029,
      "learning_rate": 8.220255066991141e-06,
      "loss": 0.304,
      "num_tokens": 1559269638.0,
      "step": 2462
    },
    {
      "epoch": 0.29123802766938633,
      "grad_norm": 0.150724858045578,
      "learning_rate": 8.207937305635455e-06,
      "loss": 0.3233,
      "num_tokens": 1559902736.0,
      "step": 2463
    },
    {
      "epoch": 0.29135627291001537,
      "grad_norm": 0.1694389432668686,
      "learning_rate": 8.1956523514641e-06,
      "loss": 0.3651,
      "num_tokens": 1560537476.0,
      "step": 2464
    },
    {
      "epoch": 0.29147451815064446,
      "grad_norm": 0.15489940345287323,
      "learning_rate": 8.18340022073367e-06,
      "loss": 0.3424,
      "num_tokens": 1561173697.0,
      "step": 2465
    },
    {
      "epoch": 0.2915927633912735,
      "grad_norm": 0.1514696478843689,
      "learning_rate": 8.171180929657341e-06,
      "loss": 0.3378,
      "num_tokens": 1561809371.0,
      "step": 2466
    },
    {
      "epoch": 0.2917110086319026,
      "grad_norm": 0.14901064336299896,
      "learning_rate": 8.158994494404828e-06,
      "loss": 0.3397,
      "num_tokens": 1562448843.0,
      "step": 2467
    },
    {
      "epoch": 0.2918292538725316,
      "grad_norm": 0.20931477844715118,
      "learning_rate": 8.146840931102362e-06,
      "loss": 0.3752,
      "num_tokens": 1563085995.0,
      "step": 2468
    },
    {
      "epoch": 0.2919474991131607,
      "grad_norm": 0.17268788814544678,
      "learning_rate": 8.13472025583268e-06,
      "loss": 0.4279,
      "num_tokens": 1563717068.0,
      "step": 2469
    },
    {
      "epoch": 0.29206574435378974,
      "grad_norm": 0.16520725190639496,
      "learning_rate": 8.122632484634991e-06,
      "loss": 0.3953,
      "num_tokens": 1564350110.0,
      "step": 2470
    },
    {
      "epoch": 0.2921839895944188,
      "grad_norm": 0.159401997923851,
      "learning_rate": 8.110577633504978e-06,
      "loss": 0.3674,
      "num_tokens": 1564985340.0,
      "step": 2471
    },
    {
      "epoch": 0.2923022348350479,
      "grad_norm": 0.14242856204509735,
      "learning_rate": 8.098555718394746e-06,
      "loss": 0.3257,
      "num_tokens": 1565619951.0,
      "step": 2472
    },
    {
      "epoch": 0.29242048007567695,
      "grad_norm": 0.1541231870651245,
      "learning_rate": 8.086566755212826e-06,
      "loss": 0.3472,
      "num_tokens": 1566247164.0,
      "step": 2473
    },
    {
      "epoch": 0.29253872531630604,
      "grad_norm": 0.1567128747701645,
      "learning_rate": 8.074610759824134e-06,
      "loss": 0.3766,
      "num_tokens": 1566885564.0,
      "step": 2474
    },
    {
      "epoch": 0.29265697055693507,
      "grad_norm": 0.14308445155620575,
      "learning_rate": 8.062687748049966e-06,
      "loss": 0.3116,
      "num_tokens": 1567517351.0,
      "step": 2475
    },
    {
      "epoch": 0.29277521579756416,
      "grad_norm": 0.1461796760559082,
      "learning_rate": 8.050797735667974e-06,
      "loss": 0.3298,
      "num_tokens": 1568152846.0,
      "step": 2476
    },
    {
      "epoch": 0.2928934610381932,
      "grad_norm": 0.1555767059326172,
      "learning_rate": 8.038940738412134e-06,
      "loss": 0.3536,
      "num_tokens": 1568780581.0,
      "step": 2477
    },
    {
      "epoch": 0.2930117062788223,
      "grad_norm": 0.14955316483974457,
      "learning_rate": 8.027116771972743e-06,
      "loss": 0.331,
      "num_tokens": 1569415183.0,
      "step": 2478
    },
    {
      "epoch": 0.2931299515194513,
      "grad_norm": 0.17126750946044922,
      "learning_rate": 8.015325851996378e-06,
      "loss": 0.3697,
      "num_tokens": 1570054656.0,
      "step": 2479
    },
    {
      "epoch": 0.2932481967600804,
      "grad_norm": 0.15955372154712677,
      "learning_rate": 8.00356799408589e-06,
      "loss": 0.3548,
      "num_tokens": 1570688875.0,
      "step": 2480
    },
    {
      "epoch": 0.2933664420007095,
      "grad_norm": 0.1520116925239563,
      "learning_rate": 7.991843213800385e-06,
      "loss": 0.3196,
      "num_tokens": 1571324827.0,
      "step": 2481
    },
    {
      "epoch": 0.2934846872413385,
      "grad_norm": 0.1544649600982666,
      "learning_rate": 7.980151526655194e-06,
      "loss": 0.3522,
      "num_tokens": 1571958663.0,
      "step": 2482
    },
    {
      "epoch": 0.2936029324819676,
      "grad_norm": 0.14922268688678741,
      "learning_rate": 7.968492948121852e-06,
      "loss": 0.3601,
      "num_tokens": 1572588508.0,
      "step": 2483
    },
    {
      "epoch": 0.29372117772259665,
      "grad_norm": 0.16230155527591705,
      "learning_rate": 7.956867493628082e-06,
      "loss": 0.3358,
      "num_tokens": 1573222363.0,
      "step": 2484
    },
    {
      "epoch": 0.29383942296322574,
      "grad_norm": 0.1554425209760666,
      "learning_rate": 7.945275178557774e-06,
      "loss": 0.3626,
      "num_tokens": 1573857267.0,
      "step": 2485
    },
    {
      "epoch": 0.29395766820385477,
      "grad_norm": 0.153177872300148,
      "learning_rate": 7.933716018250979e-06,
      "loss": 0.3245,
      "num_tokens": 1574489833.0,
      "step": 2486
    },
    {
      "epoch": 0.29407591344448386,
      "grad_norm": 0.1506984978914261,
      "learning_rate": 7.92219002800385e-06,
      "loss": 0.3347,
      "num_tokens": 1575127826.0,
      "step": 2487
    },
    {
      "epoch": 0.29419415868511295,
      "grad_norm": 0.16485412418842316,
      "learning_rate": 7.91069722306867e-06,
      "loss": 0.379,
      "num_tokens": 1575764942.0,
      "step": 2488
    },
    {
      "epoch": 0.294312403925742,
      "grad_norm": 0.131722092628479,
      "learning_rate": 7.899237618653793e-06,
      "loss": 0.3212,
      "num_tokens": 1576396971.0,
      "step": 2489
    },
    {
      "epoch": 0.2944306491663711,
      "grad_norm": 0.14967596530914307,
      "learning_rate": 7.887811229923635e-06,
      "loss": 0.3866,
      "num_tokens": 1577032933.0,
      "step": 2490
    },
    {
      "epoch": 0.2945488944070001,
      "grad_norm": 0.14169523119926453,
      "learning_rate": 7.876418071998677e-06,
      "loss": 0.3577,
      "num_tokens": 1577670215.0,
      "step": 2491
    },
    {
      "epoch": 0.2946671396476292,
      "grad_norm": 0.15427854657173157,
      "learning_rate": 7.865058159955415e-06,
      "loss": 0.3404,
      "num_tokens": 1578304041.0,
      "step": 2492
    },
    {
      "epoch": 0.29478538488825823,
      "grad_norm": 0.14424122869968414,
      "learning_rate": 7.853731508826346e-06,
      "loss": 0.3259,
      "num_tokens": 1578942969.0,
      "step": 2493
    },
    {
      "epoch": 0.2949036301288873,
      "grad_norm": 0.15108096599578857,
      "learning_rate": 7.842438133599956e-06,
      "loss": 0.3657,
      "num_tokens": 1579580392.0,
      "step": 2494
    },
    {
      "epoch": 0.29502187536951635,
      "grad_norm": 0.16395127773284912,
      "learning_rate": 7.831178049220699e-06,
      "loss": 0.3409,
      "num_tokens": 1580212819.0,
      "step": 2495
    },
    {
      "epoch": 0.29514012061014544,
      "grad_norm": 0.14246340095996857,
      "learning_rate": 7.819951270588985e-06,
      "loss": 0.3328,
      "num_tokens": 1580842095.0,
      "step": 2496
    },
    {
      "epoch": 0.29525836585077453,
      "grad_norm": 0.14513429999351501,
      "learning_rate": 7.808757812561132e-06,
      "loss": 0.3627,
      "num_tokens": 1581477926.0,
      "step": 2497
    },
    {
      "epoch": 0.29537661109140356,
      "grad_norm": 0.15380927920341492,
      "learning_rate": 7.79759768994938e-06,
      "loss": 0.3452,
      "num_tokens": 1582109535.0,
      "step": 2498
    },
    {
      "epoch": 0.29549485633203265,
      "grad_norm": 0.15641339123249054,
      "learning_rate": 7.786470917521844e-06,
      "loss": 0.355,
      "num_tokens": 1582745020.0,
      "step": 2499
    },
    {
      "epoch": 0.2956131015726617,
      "grad_norm": 0.1374625712633133,
      "learning_rate": 7.775377510002519e-06,
      "loss": 0.3431,
      "num_tokens": 1583382664.0,
      "step": 2500
    },
    {
      "epoch": 0.2957313468132908,
      "grad_norm": 0.1453300416469574,
      "learning_rate": 7.764317482071242e-06,
      "loss": 0.3383,
      "num_tokens": 1584021873.0,
      "step": 2501
    },
    {
      "epoch": 0.2958495920539198,
      "grad_norm": 0.1649884134531021,
      "learning_rate": 7.753290848363678e-06,
      "loss": 0.3865,
      "num_tokens": 1584661560.0,
      "step": 2502
    },
    {
      "epoch": 0.2959678372945489,
      "grad_norm": 5.652665615081787,
      "learning_rate": 7.742297623471309e-06,
      "loss": 0.5273,
      "num_tokens": 1585262729.0,
      "step": 2503
    },
    {
      "epoch": 0.296086082535178,
      "grad_norm": 0.16333463788032532,
      "learning_rate": 7.73133782194139e-06,
      "loss": 0.3621,
      "num_tokens": 1585884360.0,
      "step": 2504
    },
    {
      "epoch": 0.296204327775807,
      "grad_norm": 0.15930195152759552,
      "learning_rate": 7.720411458276974e-06,
      "loss": 0.392,
      "num_tokens": 1586513378.0,
      "step": 2505
    },
    {
      "epoch": 0.2963225730164361,
      "grad_norm": 0.15929850935935974,
      "learning_rate": 7.709518546936844e-06,
      "loss": 0.333,
      "num_tokens": 1587146261.0,
      "step": 2506
    },
    {
      "epoch": 0.29644081825706514,
      "grad_norm": 0.14589621126651764,
      "learning_rate": 7.698659102335525e-06,
      "loss": 0.3441,
      "num_tokens": 1587784709.0,
      "step": 2507
    },
    {
      "epoch": 0.29655906349769423,
      "grad_norm": 0.1511000245809555,
      "learning_rate": 7.687833138843248e-06,
      "loss": 0.3299,
      "num_tokens": 1588417863.0,
      "step": 2508
    },
    {
      "epoch": 0.29667730873832326,
      "grad_norm": 0.14837408065795898,
      "learning_rate": 7.677040670785942e-06,
      "loss": 0.3657,
      "num_tokens": 1589052471.0,
      "step": 2509
    },
    {
      "epoch": 0.29679555397895235,
      "grad_norm": 0.15598666667938232,
      "learning_rate": 7.666281712445225e-06,
      "loss": 0.3694,
      "num_tokens": 1589685366.0,
      "step": 2510
    },
    {
      "epoch": 0.2969137992195814,
      "grad_norm": 0.14032559096813202,
      "learning_rate": 7.655556278058356e-06,
      "loss": 0.3292,
      "num_tokens": 1590322671.0,
      "step": 2511
    },
    {
      "epoch": 0.2970320444602105,
      "grad_norm": 0.15163357555866241,
      "learning_rate": 7.644864381818234e-06,
      "loss": 0.3212,
      "num_tokens": 1590953678.0,
      "step": 2512
    },
    {
      "epoch": 0.29715028970083956,
      "grad_norm": 0.1445772647857666,
      "learning_rate": 7.63420603787338e-06,
      "loss": 0.3457,
      "num_tokens": 1591588343.0,
      "step": 2513
    },
    {
      "epoch": 0.2972685349414686,
      "grad_norm": 0.1554974913597107,
      "learning_rate": 7.623581260327911e-06,
      "loss": 0.3543,
      "num_tokens": 1592222393.0,
      "step": 2514
    },
    {
      "epoch": 0.2973867801820977,
      "grad_norm": 0.16030625998973846,
      "learning_rate": 7.6129900632415415e-06,
      "loss": 0.3763,
      "num_tokens": 1592836737.0,
      "step": 2515
    },
    {
      "epoch": 0.2975050254227267,
      "grad_norm": 0.15744347870349884,
      "learning_rate": 7.6024324606295274e-06,
      "loss": 0.3728,
      "num_tokens": 1593468479.0,
      "step": 2516
    },
    {
      "epoch": 0.2976232706633558,
      "grad_norm": 0.1460658609867096,
      "learning_rate": 7.591908466462682e-06,
      "loss": 0.3173,
      "num_tokens": 1594103240.0,
      "step": 2517
    },
    {
      "epoch": 0.29774151590398484,
      "grad_norm": 0.14362597465515137,
      "learning_rate": 7.5814180946673465e-06,
      "loss": 0.3054,
      "num_tokens": 1594736479.0,
      "step": 2518
    },
    {
      "epoch": 0.29785976114461393,
      "grad_norm": 0.14968115091323853,
      "learning_rate": 7.570961359125354e-06,
      "loss": 0.3409,
      "num_tokens": 1595369844.0,
      "step": 2519
    },
    {
      "epoch": 0.297978006385243,
      "grad_norm": 20.09839630126953,
      "learning_rate": 7.560538273674054e-06,
      "loss": 0.8587,
      "num_tokens": 1595969594.0,
      "step": 2520
    },
    {
      "epoch": 0.29809625162587206,
      "grad_norm": 0.1636546552181244,
      "learning_rate": 7.550148852106242e-06,
      "loss": 0.3541,
      "num_tokens": 1596607685.0,
      "step": 2521
    },
    {
      "epoch": 0.29821449686650114,
      "grad_norm": 0.1554463803768158,
      "learning_rate": 7.539793108170178e-06,
      "loss": 0.338,
      "num_tokens": 1597246441.0,
      "step": 2522
    },
    {
      "epoch": 0.2983327421071302,
      "grad_norm": 0.16188600659370422,
      "learning_rate": 7.529471055569549e-06,
      "loss": 0.3647,
      "num_tokens": 1597882201.0,
      "step": 2523
    },
    {
      "epoch": 0.29845098734775927,
      "grad_norm": 0.158077672123909,
      "learning_rate": 7.519182707963472e-06,
      "loss": 0.3499,
      "num_tokens": 1598512865.0,
      "step": 2524
    },
    {
      "epoch": 0.2985692325883883,
      "grad_norm": 0.14716975390911102,
      "learning_rate": 7.508928078966451e-06,
      "loss": 0.3353,
      "num_tokens": 1599149984.0,
      "step": 2525
    },
    {
      "epoch": 0.2986874778290174,
      "grad_norm": 0.15012039244174957,
      "learning_rate": 7.498707182148375e-06,
      "loss": 0.334,
      "num_tokens": 1599780457.0,
      "step": 2526
    },
    {
      "epoch": 0.2988057230696464,
      "grad_norm": 0.1462370604276657,
      "learning_rate": 7.488520031034491e-06,
      "loss": 0.3484,
      "num_tokens": 1600414611.0,
      "step": 2527
    },
    {
      "epoch": 0.2989239683102755,
      "grad_norm": 0.14537328481674194,
      "learning_rate": 7.478366639105394e-06,
      "loss": 0.3615,
      "num_tokens": 1601051962.0,
      "step": 2528
    },
    {
      "epoch": 0.2990422135509046,
      "grad_norm": 0.1465734839439392,
      "learning_rate": 7.468247019797004e-06,
      "loss": 0.3504,
      "num_tokens": 1601682372.0,
      "step": 2529
    },
    {
      "epoch": 0.29916045879153363,
      "grad_norm": 0.16139309108257294,
      "learning_rate": 7.458161186500556e-06,
      "loss": 0.3698,
      "num_tokens": 1602307670.0,
      "step": 2530
    },
    {
      "epoch": 0.2992787040321627,
      "grad_norm": 0.14881828427314758,
      "learning_rate": 7.448109152562571e-06,
      "loss": 0.3622,
      "num_tokens": 1602943700.0,
      "step": 2531
    },
    {
      "epoch": 0.29939694927279176,
      "grad_norm": 0.16552668809890747,
      "learning_rate": 7.438090931284837e-06,
      "loss": 0.353,
      "num_tokens": 1603576967.0,
      "step": 2532
    },
    {
      "epoch": 0.29951519451342085,
      "grad_norm": 0.14624199271202087,
      "learning_rate": 7.428106535924409e-06,
      "loss": 0.3338,
      "num_tokens": 1604215712.0,
      "step": 2533
    },
    {
      "epoch": 0.2996334397540499,
      "grad_norm": 0.1505880057811737,
      "learning_rate": 7.418155979693578e-06,
      "loss": 0.3329,
      "num_tokens": 1604816424.0,
      "step": 2534
    },
    {
      "epoch": 0.29975168499467897,
      "grad_norm": 0.1545705497264862,
      "learning_rate": 7.408239275759853e-06,
      "loss": 0.3326,
      "num_tokens": 1605422776.0,
      "step": 2535
    },
    {
      "epoch": 0.299869930235308,
      "grad_norm": 0.1565445214509964,
      "learning_rate": 7.398356437245944e-06,
      "loss": 0.3406,
      "num_tokens": 1606055073.0,
      "step": 2536
    },
    {
      "epoch": 0.2999881754759371,
      "grad_norm": 0.16159765422344208,
      "learning_rate": 7.388507477229758e-06,
      "loss": 0.3546,
      "num_tokens": 1606691741.0,
      "step": 2537
    },
    {
      "epoch": 0.3001064207165662,
      "grad_norm": 0.15116626024246216,
      "learning_rate": 7.3786924087443494e-06,
      "loss": 0.3564,
      "num_tokens": 1607324599.0,
      "step": 2538
    },
    {
      "epoch": 0.3002246659571952,
      "grad_norm": 0.1525968313217163,
      "learning_rate": 7.368911244777954e-06,
      "loss": 0.3124,
      "num_tokens": 1607960797.0,
      "step": 2539
    },
    {
      "epoch": 0.3003429111978243,
      "grad_norm": 0.1497633010149002,
      "learning_rate": 7.359163998273918e-06,
      "loss": 0.3612,
      "num_tokens": 1608569909.0,
      "step": 2540
    },
    {
      "epoch": 0.30046115643845334,
      "grad_norm": 0.15262489020824432,
      "learning_rate": 7.349450682130711e-06,
      "loss": 0.3599,
      "num_tokens": 1609206721.0,
      "step": 2541
    },
    {
      "epoch": 0.3005794016790824,
      "grad_norm": 0.15345512330532074,
      "learning_rate": 7.339771309201911e-06,
      "loss": 0.3594,
      "num_tokens": 1609836379.0,
      "step": 2542
    },
    {
      "epoch": 0.30069764691971146,
      "grad_norm": 0.15568511188030243,
      "learning_rate": 7.330125892296162e-06,
      "loss": 0.3374,
      "num_tokens": 1610467723.0,
      "step": 2543
    },
    {
      "epoch": 0.30081589216034055,
      "grad_norm": 0.1619885116815567,
      "learning_rate": 7.320514444177199e-06,
      "loss": 0.3578,
      "num_tokens": 1611102219.0,
      "step": 2544
    },
    {
      "epoch": 0.30093413740096964,
      "grad_norm": 0.15388430655002594,
      "learning_rate": 7.3109369775637855e-06,
      "loss": 0.3487,
      "num_tokens": 1611737217.0,
      "step": 2545
    },
    {
      "epoch": 0.30105238264159867,
      "grad_norm": 0.15204574167728424,
      "learning_rate": 7.301393505129721e-06,
      "loss": 0.3263,
      "num_tokens": 1612372064.0,
      "step": 2546
    },
    {
      "epoch": 0.30117062788222776,
      "grad_norm": 0.14779198169708252,
      "learning_rate": 7.291884039503831e-06,
      "loss": 0.3464,
      "num_tokens": 1613007358.0,
      "step": 2547
    },
    {
      "epoch": 0.3012888731228568,
      "grad_norm": 0.16607439517974854,
      "learning_rate": 7.282408593269922e-06,
      "loss": 0.3447,
      "num_tokens": 1613639636.0,
      "step": 2548
    },
    {
      "epoch": 0.3014071183634859,
      "grad_norm": 0.14590926468372345,
      "learning_rate": 7.272967178966806e-06,
      "loss": 0.3287,
      "num_tokens": 1614272364.0,
      "step": 2549
    },
    {
      "epoch": 0.3015253636041149,
      "grad_norm": 0.1410803347826004,
      "learning_rate": 7.263559809088245e-06,
      "loss": 0.3463,
      "num_tokens": 1614901916.0,
      "step": 2550
    },
    {
      "epoch": 0.301643608844744,
      "grad_norm": 0.14510096609592438,
      "learning_rate": 7.254186496082951e-06,
      "loss": 0.3264,
      "num_tokens": 1615538690.0,
      "step": 2551
    },
    {
      "epoch": 0.30176185408537304,
      "grad_norm": 0.15616920590400696,
      "learning_rate": 7.244847252354571e-06,
      "loss": 0.3539,
      "num_tokens": 1616173042.0,
      "step": 2552
    },
    {
      "epoch": 0.3018800993260021,
      "grad_norm": 0.15855540335178375,
      "learning_rate": 7.23554209026167e-06,
      "loss": 0.3964,
      "num_tokens": 1616809845.0,
      "step": 2553
    },
    {
      "epoch": 0.3019983445666312,
      "grad_norm": 0.1399592161178589,
      "learning_rate": 7.226271022117714e-06,
      "loss": 0.3345,
      "num_tokens": 1617446340.0,
      "step": 2554
    },
    {
      "epoch": 0.30211658980726025,
      "grad_norm": 0.14557190239429474,
      "learning_rate": 7.217034060191049e-06,
      "loss": 0.3508,
      "num_tokens": 1618075970.0,
      "step": 2555
    },
    {
      "epoch": 0.30223483504788934,
      "grad_norm": 0.15270313620567322,
      "learning_rate": 7.20783121670489e-06,
      "loss": 0.3556,
      "num_tokens": 1618702541.0,
      "step": 2556
    },
    {
      "epoch": 0.30235308028851837,
      "grad_norm": 0.15719246864318848,
      "learning_rate": 7.198662503837305e-06,
      "loss": 0.3453,
      "num_tokens": 1619339972.0,
      "step": 2557
    },
    {
      "epoch": 0.30247132552914746,
      "grad_norm": 0.14748413860797882,
      "learning_rate": 7.18952793372119e-06,
      "loss": 0.3571,
      "num_tokens": 1619975087.0,
      "step": 2558
    },
    {
      "epoch": 0.3025895707697765,
      "grad_norm": 0.16462109982967377,
      "learning_rate": 7.180427518444273e-06,
      "loss": 0.363,
      "num_tokens": 1620598247.0,
      "step": 2559
    },
    {
      "epoch": 0.3027078160104056,
      "grad_norm": 0.15592196583747864,
      "learning_rate": 7.1713612700490745e-06,
      "loss": 0.3595,
      "num_tokens": 1621233754.0,
      "step": 2560
    },
    {
      "epoch": 0.3028260612510347,
      "grad_norm": 0.16148506104946136,
      "learning_rate": 7.162329200532902e-06,
      "loss": 0.3835,
      "num_tokens": 1621865701.0,
      "step": 2561
    },
    {
      "epoch": 0.3029443064916637,
      "grad_norm": 0.14827576279640198,
      "learning_rate": 7.153331321847842e-06,
      "loss": 0.3607,
      "num_tokens": 1622502297.0,
      "step": 2562
    },
    {
      "epoch": 0.3030625517322928,
      "grad_norm": 0.15333227813243866,
      "learning_rate": 7.144367645900729e-06,
      "loss": 0.3723,
      "num_tokens": 1623133709.0,
      "step": 2563
    },
    {
      "epoch": 0.30318079697292183,
      "grad_norm": 0.14636582136154175,
      "learning_rate": 7.135438184553143e-06,
      "loss": 0.3669,
      "num_tokens": 1623765687.0,
      "step": 2564
    },
    {
      "epoch": 0.3032990422135509,
      "grad_norm": 0.15707650780677795,
      "learning_rate": 7.126542949621384e-06,
      "loss": 0.3607,
      "num_tokens": 1624390373.0,
      "step": 2565
    },
    {
      "epoch": 0.30341728745417995,
      "grad_norm": 0.15035566687583923,
      "learning_rate": 7.1176819528764625e-06,
      "loss": 0.3441,
      "num_tokens": 1625025113.0,
      "step": 2566
    },
    {
      "epoch": 0.30353553269480904,
      "grad_norm": 0.15434648096561432,
      "learning_rate": 7.108855206044077e-06,
      "loss": 0.3414,
      "num_tokens": 1625657956.0,
      "step": 2567
    },
    {
      "epoch": 0.3036537779354381,
      "grad_norm": 0.1429792046546936,
      "learning_rate": 7.100062720804617e-06,
      "loss": 0.3331,
      "num_tokens": 1626290658.0,
      "step": 2568
    },
    {
      "epoch": 0.30377202317606716,
      "grad_norm": 0.14561037719249725,
      "learning_rate": 7.091304508793116e-06,
      "loss": 0.3643,
      "num_tokens": 1626927675.0,
      "step": 2569
    },
    {
      "epoch": 0.30389026841669625,
      "grad_norm": 0.1610872894525528,
      "learning_rate": 7.082580581599263e-06,
      "loss": 0.3487,
      "num_tokens": 1627561965.0,
      "step": 2570
    },
    {
      "epoch": 0.3040085136573253,
      "grad_norm": 0.15274368226528168,
      "learning_rate": 7.073890950767382e-06,
      "loss": 0.302,
      "num_tokens": 1628198126.0,
      "step": 2571
    },
    {
      "epoch": 0.3041267588979544,
      "grad_norm": 0.1400538682937622,
      "learning_rate": 7.0652356277963985e-06,
      "loss": 0.3281,
      "num_tokens": 1628832165.0,
      "step": 2572
    },
    {
      "epoch": 0.3042450041385834,
      "grad_norm": 0.15818189084529877,
      "learning_rate": 7.056614624139859e-06,
      "loss": 0.3515,
      "num_tokens": 1629466351.0,
      "step": 2573
    },
    {
      "epoch": 0.3043632493792125,
      "grad_norm": 0.1607942134141922,
      "learning_rate": 7.048027951205882e-06,
      "loss": 0.3882,
      "num_tokens": 1630103503.0,
      "step": 2574
    },
    {
      "epoch": 0.30448149461984153,
      "grad_norm": 0.17156703770160675,
      "learning_rate": 7.039475620357156e-06,
      "loss": 0.3725,
      "num_tokens": 1630741258.0,
      "step": 2575
    },
    {
      "epoch": 0.3045997398604706,
      "grad_norm": 0.15256287157535553,
      "learning_rate": 7.030957642910928e-06,
      "loss": 0.335,
      "num_tokens": 1631378278.0,
      "step": 2576
    },
    {
      "epoch": 0.30471798510109965,
      "grad_norm": 0.1651080846786499,
      "learning_rate": 7.022474030138984e-06,
      "loss": 0.3239,
      "num_tokens": 1632012029.0,
      "step": 2577
    },
    {
      "epoch": 0.30483623034172874,
      "grad_norm": 0.14645303785800934,
      "learning_rate": 7.014024793267643e-06,
      "loss": 0.3164,
      "num_tokens": 1632649835.0,
      "step": 2578
    },
    {
      "epoch": 0.30495447558235783,
      "grad_norm": 0.1718732863664627,
      "learning_rate": 7.005609943477723e-06,
      "loss": 0.359,
      "num_tokens": 1633279626.0,
      "step": 2579
    },
    {
      "epoch": 0.30507272082298686,
      "grad_norm": 0.16827356815338135,
      "learning_rate": 6.9972294919045465e-06,
      "loss": 0.3314,
      "num_tokens": 1633913718.0,
      "step": 2580
    },
    {
      "epoch": 0.30519096606361595,
      "grad_norm": 0.1514851301908493,
      "learning_rate": 6.9888834496379095e-06,
      "loss": 0.3225,
      "num_tokens": 1634549519.0,
      "step": 2581
    },
    {
      "epoch": 0.305309211304245,
      "grad_norm": 0.13561248779296875,
      "learning_rate": 6.98057182772208e-06,
      "loss": 0.3311,
      "num_tokens": 1635181293.0,
      "step": 2582
    },
    {
      "epoch": 0.3054274565448741,
      "grad_norm": 0.1461554765701294,
      "learning_rate": 6.972294637155777e-06,
      "loss": 0.3516,
      "num_tokens": 1635817022.0,
      "step": 2583
    },
    {
      "epoch": 0.3055457017855031,
      "grad_norm": 0.1553107649087906,
      "learning_rate": 6.964051888892156e-06,
      "loss": 0.3586,
      "num_tokens": 1636452163.0,
      "step": 2584
    },
    {
      "epoch": 0.3056639470261322,
      "grad_norm": 0.17162953317165375,
      "learning_rate": 6.955843593838796e-06,
      "loss": 0.3717,
      "num_tokens": 1637091122.0,
      "step": 2585
    },
    {
      "epoch": 0.3057821922667613,
      "grad_norm": 0.15595267713069916,
      "learning_rate": 6.9476697628576825e-06,
      "loss": 0.3752,
      "num_tokens": 1637730453.0,
      "step": 2586
    },
    {
      "epoch": 0.3059004375073903,
      "grad_norm": 0.14371612668037415,
      "learning_rate": 6.939530406765194e-06,
      "loss": 0.3042,
      "num_tokens": 1638367703.0,
      "step": 2587
    },
    {
      "epoch": 0.3060186827480194,
      "grad_norm": 0.16365359723567963,
      "learning_rate": 6.931425536332091e-06,
      "loss": 0.3877,
      "num_tokens": 1638999767.0,
      "step": 2588
    },
    {
      "epoch": 0.30613692798864844,
      "grad_norm": 0.15286149084568024,
      "learning_rate": 6.923355162283503e-06,
      "loss": 0.3677,
      "num_tokens": 1639639175.0,
      "step": 2589
    },
    {
      "epoch": 0.30625517322927753,
      "grad_norm": 0.16202640533447266,
      "learning_rate": 6.915319295298903e-06,
      "loss": 0.3488,
      "num_tokens": 1640268937.0,
      "step": 2590
    },
    {
      "epoch": 0.30637341846990657,
      "grad_norm": 0.1711316555738449,
      "learning_rate": 6.907317946012105e-06,
      "loss": 0.3854,
      "num_tokens": 1640905144.0,
      "step": 2591
    },
    {
      "epoch": 0.30649166371053566,
      "grad_norm": 0.15269654989242554,
      "learning_rate": 6.89935112501124e-06,
      "loss": 0.3579,
      "num_tokens": 1641535980.0,
      "step": 2592
    },
    {
      "epoch": 0.3066099089511647,
      "grad_norm": 0.14511872828006744,
      "learning_rate": 6.891418842838766e-06,
      "loss": 0.3148,
      "num_tokens": 1642164823.0,
      "step": 2593
    },
    {
      "epoch": 0.3067281541917938,
      "grad_norm": 0.1409059464931488,
      "learning_rate": 6.883521109991411e-06,
      "loss": 0.3365,
      "num_tokens": 1642792813.0,
      "step": 2594
    },
    {
      "epoch": 0.30684639943242287,
      "grad_norm": 0.14985384047031403,
      "learning_rate": 6.875657936920204e-06,
      "loss": 0.3078,
      "num_tokens": 1643418555.0,
      "step": 2595
    },
    {
      "epoch": 0.3069646446730519,
      "grad_norm": 0.16204628348350525,
      "learning_rate": 6.86782933403043e-06,
      "loss": 0.3503,
      "num_tokens": 1644034211.0,
      "step": 2596
    },
    {
      "epoch": 0.307082889913681,
      "grad_norm": 0.14266403019428253,
      "learning_rate": 6.860035311681629e-06,
      "loss": 0.3332,
      "num_tokens": 1644670523.0,
      "step": 2597
    },
    {
      "epoch": 0.30720113515431,
      "grad_norm": 0.1625642627477646,
      "learning_rate": 6.852275880187587e-06,
      "loss": 0.3591,
      "num_tokens": 1645306242.0,
      "step": 2598
    },
    {
      "epoch": 0.3073193803949391,
      "grad_norm": 0.13715782761573792,
      "learning_rate": 6.844551049816312e-06,
      "loss": 0.3074,
      "num_tokens": 1645938779.0,
      "step": 2599
    },
    {
      "epoch": 0.30743762563556815,
      "grad_norm": 0.16159267723560333,
      "learning_rate": 6.8368608307900235e-06,
      "loss": 0.3655,
      "num_tokens": 1646575837.0,
      "step": 2600
    },
    {
      "epoch": 0.30755587087619723,
      "grad_norm": 0.1451670527458191,
      "learning_rate": 6.8292052332851405e-06,
      "loss": 0.3118,
      "num_tokens": 1647211588.0,
      "step": 2601
    },
    {
      "epoch": 0.3076741161168263,
      "grad_norm": 0.15132470428943634,
      "learning_rate": 6.821584267432268e-06,
      "loss": 0.3341,
      "num_tokens": 1647847594.0,
      "step": 2602
    },
    {
      "epoch": 0.30779236135745536,
      "grad_norm": 0.15491485595703125,
      "learning_rate": 6.813997943316185e-06,
      "loss": 0.3666,
      "num_tokens": 1648480650.0,
      "step": 2603
    },
    {
      "epoch": 0.30791060659808445,
      "grad_norm": 0.1458674818277359,
      "learning_rate": 6.8064462709758295e-06,
      "loss": 0.3328,
      "num_tokens": 1649117682.0,
      "step": 2604
    },
    {
      "epoch": 0.3080288518387135,
      "grad_norm": 0.14040695130825043,
      "learning_rate": 6.798929260404277e-06,
      "loss": 0.295,
      "num_tokens": 1649751752.0,
      "step": 2605
    },
    {
      "epoch": 0.30814709707934257,
      "grad_norm": 0.15380658209323883,
      "learning_rate": 6.791446921548744e-06,
      "loss": 0.3448,
      "num_tokens": 1650391077.0,
      "step": 2606
    },
    {
      "epoch": 0.3082653423199716,
      "grad_norm": 0.1441585123538971,
      "learning_rate": 6.783999264310565e-06,
      "loss": 0.3339,
      "num_tokens": 1651025851.0,
      "step": 2607
    },
    {
      "epoch": 0.3083835875606007,
      "grad_norm": 0.1481962352991104,
      "learning_rate": 6.776586298545181e-06,
      "loss": 0.3348,
      "num_tokens": 1651663895.0,
      "step": 2608
    },
    {
      "epoch": 0.3085018328012297,
      "grad_norm": 0.1441476196050644,
      "learning_rate": 6.769208034062119e-06,
      "loss": 0.3345,
      "num_tokens": 1652297334.0,
      "step": 2609
    },
    {
      "epoch": 0.3086200780418588,
      "grad_norm": 0.15080349147319794,
      "learning_rate": 6.761864480624994e-06,
      "loss": 0.3585,
      "num_tokens": 1652930319.0,
      "step": 2610
    },
    {
      "epoch": 0.3087383232824879,
      "grad_norm": 0.16040252149105072,
      "learning_rate": 6.754555647951481e-06,
      "loss": 0.3402,
      "num_tokens": 1653558106.0,
      "step": 2611
    },
    {
      "epoch": 0.30885656852311694,
      "grad_norm": 0.1390003263950348,
      "learning_rate": 6.747281545713321e-06,
      "loss": 0.2883,
      "num_tokens": 1654190622.0,
      "step": 2612
    },
    {
      "epoch": 0.308974813763746,
      "grad_norm": 0.1431875079870224,
      "learning_rate": 6.740042183536288e-06,
      "loss": 0.3487,
      "num_tokens": 1654804468.0,
      "step": 2613
    },
    {
      "epoch": 0.30909305900437506,
      "grad_norm": 0.15379014611244202,
      "learning_rate": 6.73283757100018e-06,
      "loss": 0.3333,
      "num_tokens": 1655436872.0,
      "step": 2614
    },
    {
      "epoch": 0.30921130424500415,
      "grad_norm": 0.17064590752124786,
      "learning_rate": 6.725667717638816e-06,
      "loss": 0.371,
      "num_tokens": 1656073073.0,
      "step": 2615
    },
    {
      "epoch": 0.3093295494856332,
      "grad_norm": 0.14026252925395966,
      "learning_rate": 6.718532632940025e-06,
      "loss": 0.3295,
      "num_tokens": 1656704728.0,
      "step": 2616
    },
    {
      "epoch": 0.30944779472626227,
      "grad_norm": 0.1622316539287567,
      "learning_rate": 6.711432326345615e-06,
      "loss": 0.3299,
      "num_tokens": 1657332271.0,
      "step": 2617
    },
    {
      "epoch": 0.30956603996689136,
      "grad_norm": 0.16228297352790833,
      "learning_rate": 6.704366807251383e-06,
      "loss": 0.3521,
      "num_tokens": 1657948359.0,
      "step": 2618
    },
    {
      "epoch": 0.3096842852075204,
      "grad_norm": 0.17386050522327423,
      "learning_rate": 6.697336085007086e-06,
      "loss": 0.4092,
      "num_tokens": 1658582924.0,
      "step": 2619
    },
    {
      "epoch": 0.3098025304481495,
      "grad_norm": 0.15338410437107086,
      "learning_rate": 6.690340168916435e-06,
      "loss": 0.3586,
      "num_tokens": 1659208653.0,
      "step": 2620
    },
    {
      "epoch": 0.3099207756887785,
      "grad_norm": 0.14601671695709229,
      "learning_rate": 6.683379068237077e-06,
      "loss": 0.3084,
      "num_tokens": 1659845271.0,
      "step": 2621
    },
    {
      "epoch": 0.3100390209294076,
      "grad_norm": 0.16629400849342346,
      "learning_rate": 6.676452792180601e-06,
      "loss": 0.3571,
      "num_tokens": 1660468585.0,
      "step": 2622
    },
    {
      "epoch": 0.31015726617003664,
      "grad_norm": 0.1492847353219986,
      "learning_rate": 6.669561349912505e-06,
      "loss": 0.3345,
      "num_tokens": 1661101485.0,
      "step": 2623
    },
    {
      "epoch": 0.3102755114106657,
      "grad_norm": 0.14829352498054504,
      "learning_rate": 6.66270475055219e-06,
      "loss": 0.319,
      "num_tokens": 1661735662.0,
      "step": 2624
    },
    {
      "epoch": 0.31039375665129476,
      "grad_norm": 0.15923385322093964,
      "learning_rate": 6.655883003172954e-06,
      "loss": 0.4076,
      "num_tokens": 1662373879.0,
      "step": 2625
    },
    {
      "epoch": 0.31051200189192385,
      "grad_norm": 0.15548096597194672,
      "learning_rate": 6.649096116801971e-06,
      "loss": 0.3547,
      "num_tokens": 1663008427.0,
      "step": 2626
    },
    {
      "epoch": 0.31063024713255294,
      "grad_norm": 0.15542833507061005,
      "learning_rate": 6.642344100420285e-06,
      "loss": 0.3544,
      "num_tokens": 1663638900.0,
      "step": 2627
    },
    {
      "epoch": 0.310748492373182,
      "grad_norm": 0.14559157192707062,
      "learning_rate": 6.635626962962804e-06,
      "loss": 0.3098,
      "num_tokens": 1664272324.0,
      "step": 2628
    },
    {
      "epoch": 0.31086673761381106,
      "grad_norm": 0.17403483390808105,
      "learning_rate": 6.62894471331827e-06,
      "loss": 0.3971,
      "num_tokens": 1664904228.0,
      "step": 2629
    },
    {
      "epoch": 0.3109849828544401,
      "grad_norm": 0.15496373176574707,
      "learning_rate": 6.6222973603292636e-06,
      "loss": 0.3408,
      "num_tokens": 1665537753.0,
      "step": 2630
    },
    {
      "epoch": 0.3111032280950692,
      "grad_norm": 0.16460204124450684,
      "learning_rate": 6.61568491279218e-06,
      "loss": 0.3652,
      "num_tokens": 1666174169.0,
      "step": 2631
    },
    {
      "epoch": 0.3112214733356982,
      "grad_norm": 0.16071675717830658,
      "learning_rate": 6.6091073794572425e-06,
      "loss": 0.3514,
      "num_tokens": 1666803754.0,
      "step": 2632
    },
    {
      "epoch": 0.3113397185763273,
      "grad_norm": 0.1600765883922577,
      "learning_rate": 6.602564769028456e-06,
      "loss": 0.3304,
      "num_tokens": 1667438241.0,
      "step": 2633
    },
    {
      "epoch": 0.31145796381695634,
      "grad_norm": 0.15328559279441833,
      "learning_rate": 6.596057090163614e-06,
      "loss": 0.3284,
      "num_tokens": 1668076450.0,
      "step": 2634
    },
    {
      "epoch": 0.31157620905758543,
      "grad_norm": 0.154999241232872,
      "learning_rate": 6.589584351474286e-06,
      "loss": 0.3584,
      "num_tokens": 1668712082.0,
      "step": 2635
    },
    {
      "epoch": 0.3116944542982145,
      "grad_norm": 0.14657165110111237,
      "learning_rate": 6.583146561525815e-06,
      "loss": 0.3807,
      "num_tokens": 1669348399.0,
      "step": 2636
    },
    {
      "epoch": 0.31181269953884355,
      "grad_norm": 0.15919843316078186,
      "learning_rate": 6.576743728837283e-06,
      "loss": 0.3351,
      "num_tokens": 1669984836.0,
      "step": 2637
    },
    {
      "epoch": 0.31193094477947264,
      "grad_norm": 0.15095235407352448,
      "learning_rate": 6.570375861881518e-06,
      "loss": 0.3035,
      "num_tokens": 1670619936.0,
      "step": 2638
    },
    {
      "epoch": 0.3120491900201017,
      "grad_norm": 0.16317443549633026,
      "learning_rate": 6.564042969085087e-06,
      "loss": 0.3509,
      "num_tokens": 1671255448.0,
      "step": 2639
    },
    {
      "epoch": 0.31216743526073076,
      "grad_norm": 0.1617879569530487,
      "learning_rate": 6.557745058828257e-06,
      "loss": 0.3665,
      "num_tokens": 1671888642.0,
      "step": 2640
    },
    {
      "epoch": 0.3122856805013598,
      "grad_norm": 0.1533808708190918,
      "learning_rate": 6.551482139445026e-06,
      "loss": 0.3438,
      "num_tokens": 1672528375.0,
      "step": 2641
    },
    {
      "epoch": 0.3124039257419889,
      "grad_norm": 0.15954667329788208,
      "learning_rate": 6.54525421922307e-06,
      "loss": 0.3904,
      "num_tokens": 1673167791.0,
      "step": 2642
    },
    {
      "epoch": 0.312522170982618,
      "grad_norm": 0.15365970134735107,
      "learning_rate": 6.5390613064037576e-06,
      "loss": 0.332,
      "num_tokens": 1673807365.0,
      "step": 2643
    },
    {
      "epoch": 0.312640416223247,
      "grad_norm": 0.15740928053855896,
      "learning_rate": 6.532903409182135e-06,
      "loss": 0.3659,
      "num_tokens": 1674442513.0,
      "step": 2644
    },
    {
      "epoch": 0.3127586614638761,
      "grad_norm": 0.15962831676006317,
      "learning_rate": 6.526780535706904e-06,
      "loss": 0.3486,
      "num_tokens": 1675075724.0,
      "step": 2645
    },
    {
      "epoch": 0.31287690670450513,
      "grad_norm": 0.16220061480998993,
      "learning_rate": 6.52069269408043e-06,
      "loss": 0.3534,
      "num_tokens": 1675710105.0,
      "step": 2646
    },
    {
      "epoch": 0.3129951519451342,
      "grad_norm": 0.15352794528007507,
      "learning_rate": 6.51463989235872e-06,
      "loss": 0.3395,
      "num_tokens": 1676349650.0,
      "step": 2647
    },
    {
      "epoch": 0.31311339718576325,
      "grad_norm": 0.1553868055343628,
      "learning_rate": 6.508622138551402e-06,
      "loss": 0.3394,
      "num_tokens": 1676989186.0,
      "step": 2648
    },
    {
      "epoch": 0.31323164242639234,
      "grad_norm": 0.15505124628543854,
      "learning_rate": 6.502639440621739e-06,
      "loss": 0.3812,
      "num_tokens": 1677628500.0,
      "step": 2649
    },
    {
      "epoch": 0.3133498876670214,
      "grad_norm": 0.14550581574440002,
      "learning_rate": 6.49669180648659e-06,
      "loss": 0.3103,
      "num_tokens": 1678243011.0,
      "step": 2650
    },
    {
      "epoch": 0.31346813290765047,
      "grad_norm": 0.16432583332061768,
      "learning_rate": 6.490779244016433e-06,
      "loss": 0.3595,
      "num_tokens": 1678877133.0,
      "step": 2651
    },
    {
      "epoch": 0.31358637814827955,
      "grad_norm": 0.14240172505378723,
      "learning_rate": 6.484901761035319e-06,
      "loss": 0.3482,
      "num_tokens": 1679511112.0,
      "step": 2652
    },
    {
      "epoch": 0.3137046233889086,
      "grad_norm": 0.1719961315393448,
      "learning_rate": 6.479059365320886e-06,
      "loss": 0.3884,
      "num_tokens": 1680144682.0,
      "step": 2653
    },
    {
      "epoch": 0.3138228686295377,
      "grad_norm": 0.15486003458499908,
      "learning_rate": 6.473252064604342e-06,
      "loss": 0.3532,
      "num_tokens": 1680769049.0,
      "step": 2654
    },
    {
      "epoch": 0.3139411138701667,
      "grad_norm": 0.14725635945796967,
      "learning_rate": 6.46747986657045e-06,
      "loss": 0.3594,
      "num_tokens": 1681408352.0,
      "step": 2655
    },
    {
      "epoch": 0.3140593591107958,
      "grad_norm": 0.15648604929447174,
      "learning_rate": 6.46174277885753e-06,
      "loss": 0.3512,
      "num_tokens": 1682038448.0,
      "step": 2656
    },
    {
      "epoch": 0.31417760435142483,
      "grad_norm": 0.15705391764640808,
      "learning_rate": 6.456040809057429e-06,
      "loss": 0.3592,
      "num_tokens": 1682676961.0,
      "step": 2657
    },
    {
      "epoch": 0.3142958495920539,
      "grad_norm": 0.15460099279880524,
      "learning_rate": 6.450373964715527e-06,
      "loss": 0.3493,
      "num_tokens": 1683310802.0,
      "step": 2658
    },
    {
      "epoch": 0.314414094832683,
      "grad_norm": 0.14110355079174042,
      "learning_rate": 6.444742253330732e-06,
      "loss": 0.356,
      "num_tokens": 1683948414.0,
      "step": 2659
    },
    {
      "epoch": 0.31453234007331204,
      "grad_norm": 0.15325553715229034,
      "learning_rate": 6.439145682355447e-06,
      "loss": 0.3738,
      "num_tokens": 1684582469.0,
      "step": 2660
    },
    {
      "epoch": 0.31465058531394113,
      "grad_norm": 0.15117822587490082,
      "learning_rate": 6.433584259195583e-06,
      "loss": 0.3557,
      "num_tokens": 1685219110.0,
      "step": 2661
    },
    {
      "epoch": 0.31476883055457017,
      "grad_norm": 0.15664592385292053,
      "learning_rate": 6.428057991210534e-06,
      "loss": 0.3209,
      "num_tokens": 1685852803.0,
      "step": 2662
    },
    {
      "epoch": 0.31488707579519926,
      "grad_norm": 0.1503203809261322,
      "learning_rate": 6.422566885713179e-06,
      "loss": 0.3275,
      "num_tokens": 1686488656.0,
      "step": 2663
    },
    {
      "epoch": 0.3150053210358283,
      "grad_norm": 0.16129250824451447,
      "learning_rate": 6.417110949969865e-06,
      "loss": 0.3434,
      "num_tokens": 1687123786.0,
      "step": 2664
    },
    {
      "epoch": 0.3151235662764574,
      "grad_norm": 0.15377874672412872,
      "learning_rate": 6.411690191200393e-06,
      "loss": 0.3625,
      "num_tokens": 1687756907.0,
      "step": 2665
    },
    {
      "epoch": 0.3152418115170864,
      "grad_norm": 0.1461174190044403,
      "learning_rate": 6.406304616578026e-06,
      "loss": 0.3234,
      "num_tokens": 1688393115.0,
      "step": 2666
    },
    {
      "epoch": 0.3153600567577155,
      "grad_norm": 0.14871680736541748,
      "learning_rate": 6.400954233229461e-06,
      "loss": 0.3529,
      "num_tokens": 1689028874.0,
      "step": 2667
    },
    {
      "epoch": 0.3154783019983446,
      "grad_norm": 0.16240054368972778,
      "learning_rate": 6.395639048234825e-06,
      "loss": 0.3768,
      "num_tokens": 1689668481.0,
      "step": 2668
    },
    {
      "epoch": 0.3155965472389736,
      "grad_norm": 0.14038841426372528,
      "learning_rate": 6.390359068627669e-06,
      "loss": 0.3306,
      "num_tokens": 1690307233.0,
      "step": 2669
    },
    {
      "epoch": 0.3157147924796027,
      "grad_norm": 0.15858112275600433,
      "learning_rate": 6.38511430139496e-06,
      "loss": 0.3381,
      "num_tokens": 1690936000.0,
      "step": 2670
    },
    {
      "epoch": 0.31583303772023175,
      "grad_norm": 0.14911122620105743,
      "learning_rate": 6.379904753477067e-06,
      "loss": 0.3699,
      "num_tokens": 1691574884.0,
      "step": 2671
    },
    {
      "epoch": 0.31595128296086084,
      "grad_norm": 0.14863522350788116,
      "learning_rate": 6.374730431767749e-06,
      "loss": 0.3198,
      "num_tokens": 1692206801.0,
      "step": 2672
    },
    {
      "epoch": 0.31606952820148987,
      "grad_norm": 0.14785444736480713,
      "learning_rate": 6.369591343114158e-06,
      "loss": 0.3431,
      "num_tokens": 1692843392.0,
      "step": 2673
    },
    {
      "epoch": 0.31618777344211896,
      "grad_norm": 0.156451016664505,
      "learning_rate": 6.364487494316815e-06,
      "loss": 0.3979,
      "num_tokens": 1693481747.0,
      "step": 2674
    },
    {
      "epoch": 0.31630601868274805,
      "grad_norm": 0.15187658369541168,
      "learning_rate": 6.359418892129616e-06,
      "loss": 0.3667,
      "num_tokens": 1694120880.0,
      "step": 2675
    },
    {
      "epoch": 0.3164242639233771,
      "grad_norm": 0.15651914477348328,
      "learning_rate": 6.354385543259807e-06,
      "loss": 0.3522,
      "num_tokens": 1694755717.0,
      "step": 2676
    },
    {
      "epoch": 0.31654250916400617,
      "grad_norm": 0.14550632238388062,
      "learning_rate": 6.349387454367989e-06,
      "loss": 0.3614,
      "num_tokens": 1695389815.0,
      "step": 2677
    },
    {
      "epoch": 0.3166607544046352,
      "grad_norm": 0.1500931680202484,
      "learning_rate": 6.344424632068104e-06,
      "loss": 0.3344,
      "num_tokens": 1696025843.0,
      "step": 2678
    },
    {
      "epoch": 0.3167789996452643,
      "grad_norm": 0.1742960661649704,
      "learning_rate": 6.339497082927423e-06,
      "loss": 0.355,
      "num_tokens": 1696659035.0,
      "step": 2679
    },
    {
      "epoch": 0.3168972448858933,
      "grad_norm": 0.15291573107242584,
      "learning_rate": 6.334604813466541e-06,
      "loss": 0.3375,
      "num_tokens": 1697289201.0,
      "step": 2680
    },
    {
      "epoch": 0.3170154901265224,
      "grad_norm": 0.15118835866451263,
      "learning_rate": 6.32974783015937e-06,
      "loss": 0.3301,
      "num_tokens": 1697925113.0,
      "step": 2681
    },
    {
      "epoch": 0.31713373536715145,
      "grad_norm": 0.1467960923910141,
      "learning_rate": 6.324926139433123e-06,
      "loss": 0.3395,
      "num_tokens": 1698558395.0,
      "step": 2682
    },
    {
      "epoch": 0.31725198060778054,
      "grad_norm": 0.15137924253940582,
      "learning_rate": 6.320139747668318e-06,
      "loss": 0.3641,
      "num_tokens": 1699196086.0,
      "step": 2683
    },
    {
      "epoch": 0.3173702258484096,
      "grad_norm": 0.13839225471019745,
      "learning_rate": 6.315388661198751e-06,
      "loss": 0.2986,
      "num_tokens": 1699829880.0,
      "step": 2684
    },
    {
      "epoch": 0.31748847108903866,
      "grad_norm": 0.16591571271419525,
      "learning_rate": 6.310672886311513e-06,
      "loss": 0.3798,
      "num_tokens": 1700459160.0,
      "step": 2685
    },
    {
      "epoch": 0.31760671632966775,
      "grad_norm": 0.14456845819950104,
      "learning_rate": 6.305992429246957e-06,
      "loss": 0.3326,
      "num_tokens": 1701090510.0,
      "step": 2686
    },
    {
      "epoch": 0.3177249615702968,
      "grad_norm": 0.15700989961624146,
      "learning_rate": 6.301347296198703e-06,
      "loss": 0.3672,
      "num_tokens": 1701726174.0,
      "step": 2687
    },
    {
      "epoch": 0.31784320681092587,
      "grad_norm": 0.175731360912323,
      "learning_rate": 6.2967374933136255e-06,
      "loss": 0.3796,
      "num_tokens": 1702361400.0,
      "step": 2688
    },
    {
      "epoch": 0.3179614520515549,
      "grad_norm": 0.1563163846731186,
      "learning_rate": 6.292163026691851e-06,
      "loss": 0.3667,
      "num_tokens": 1702993806.0,
      "step": 2689
    },
    {
      "epoch": 0.318079697292184,
      "grad_norm": 0.1482599675655365,
      "learning_rate": 6.28762390238674e-06,
      "loss": 0.3112,
      "num_tokens": 1703632945.0,
      "step": 2690
    },
    {
      "epoch": 0.318197942532813,
      "grad_norm": 0.15436552464962006,
      "learning_rate": 6.283120126404893e-06,
      "loss": 0.3446,
      "num_tokens": 1704265862.0,
      "step": 2691
    },
    {
      "epoch": 0.3183161877734421,
      "grad_norm": 0.1484438180923462,
      "learning_rate": 6.278651704706125e-06,
      "loss": 0.3433,
      "num_tokens": 1704866810.0,
      "step": 2692
    },
    {
      "epoch": 0.3184344330140712,
      "grad_norm": 0.15252937376499176,
      "learning_rate": 6.274218643203471e-06,
      "loss": 0.3451,
      "num_tokens": 1705501981.0,
      "step": 2693
    },
    {
      "epoch": 0.31855267825470024,
      "grad_norm": 0.16562378406524658,
      "learning_rate": 6.2698209477631745e-06,
      "loss": 0.355,
      "num_tokens": 1706139459.0,
      "step": 2694
    },
    {
      "epoch": 0.31867092349532933,
      "grad_norm": 0.15192921459674835,
      "learning_rate": 6.265458624204681e-06,
      "loss": 0.3204,
      "num_tokens": 1706776278.0,
      "step": 2695
    },
    {
      "epoch": 0.31878916873595836,
      "grad_norm": 0.1544736921787262,
      "learning_rate": 6.261131678300624e-06,
      "loss": 0.3347,
      "num_tokens": 1707408737.0,
      "step": 2696
    },
    {
      "epoch": 0.31890741397658745,
      "grad_norm": 0.17091700434684753,
      "learning_rate": 6.256840115776828e-06,
      "loss": 0.394,
      "num_tokens": 1708040485.0,
      "step": 2697
    },
    {
      "epoch": 0.3190256592172165,
      "grad_norm": 0.16079741716384888,
      "learning_rate": 6.252583942312284e-06,
      "loss": 0.3696,
      "num_tokens": 1708667258.0,
      "step": 2698
    },
    {
      "epoch": 0.3191439044578456,
      "grad_norm": 0.1671343594789505,
      "learning_rate": 6.248363163539164e-06,
      "loss": 0.3792,
      "num_tokens": 1709306576.0,
      "step": 2699
    },
    {
      "epoch": 0.31926214969847466,
      "grad_norm": 0.1549573540687561,
      "learning_rate": 6.2441777850428045e-06,
      "loss": 0.3605,
      "num_tokens": 1709931524.0,
      "step": 2700
    },
    {
      "epoch": 0.3193803949391037,
      "grad_norm": 0.15157970786094666,
      "learning_rate": 6.240027812361683e-06,
      "loss": 0.326,
      "num_tokens": 1710563229.0,
      "step": 2701
    },
    {
      "epoch": 0.3194986401797328,
      "grad_norm": 0.17513641715049744,
      "learning_rate": 6.235913250987438e-06,
      "loss": 0.3655,
      "num_tokens": 1711197349.0,
      "step": 2702
    },
    {
      "epoch": 0.3196168854203618,
      "grad_norm": 0.14705513417720795,
      "learning_rate": 6.231834106364838e-06,
      "loss": 0.3039,
      "num_tokens": 1711832054.0,
      "step": 2703
    },
    {
      "epoch": 0.3197351306609909,
      "grad_norm": 0.15651701390743256,
      "learning_rate": 6.227790383891798e-06,
      "loss": 0.3549,
      "num_tokens": 1712466588.0,
      "step": 2704
    },
    {
      "epoch": 0.31985337590161994,
      "grad_norm": 0.1544400155544281,
      "learning_rate": 6.223782088919343e-06,
      "loss": 0.333,
      "num_tokens": 1713104055.0,
      "step": 2705
    },
    {
      "epoch": 0.31997162114224903,
      "grad_norm": 0.15615926682949066,
      "learning_rate": 6.219809226751629e-06,
      "loss": 0.3661,
      "num_tokens": 1713741405.0,
      "step": 2706
    },
    {
      "epoch": 0.32008986638287806,
      "grad_norm": 0.14605312049388885,
      "learning_rate": 6.2158718026459194e-06,
      "loss": 0.3552,
      "num_tokens": 1714376139.0,
      "step": 2707
    },
    {
      "epoch": 0.32020811162350715,
      "grad_norm": 0.14911401271820068,
      "learning_rate": 6.211969821812582e-06,
      "loss": 0.3391,
      "num_tokens": 1715010719.0,
      "step": 2708
    },
    {
      "epoch": 0.32032635686413624,
      "grad_norm": 0.17062732577323914,
      "learning_rate": 6.20810328941509e-06,
      "loss": 0.3724,
      "num_tokens": 1715642575.0,
      "step": 2709
    },
    {
      "epoch": 0.3204446021047653,
      "grad_norm": 0.16017664968967438,
      "learning_rate": 6.204272210569996e-06,
      "loss": 0.3558,
      "num_tokens": 1716275068.0,
      "step": 2710
    },
    {
      "epoch": 0.32056284734539436,
      "grad_norm": 0.16315767168998718,
      "learning_rate": 6.200476590346943e-06,
      "loss": 0.3842,
      "num_tokens": 1716914533.0,
      "step": 2711
    },
    {
      "epoch": 0.3206810925860234,
      "grad_norm": 0.15205085277557373,
      "learning_rate": 6.196716433768656e-06,
      "loss": 0.3443,
      "num_tokens": 1717515924.0,
      "step": 2712
    },
    {
      "epoch": 0.3207993378266525,
      "grad_norm": 0.13858628273010254,
      "learning_rate": 6.19299174581092e-06,
      "loss": 0.3262,
      "num_tokens": 1718149602.0,
      "step": 2713
    },
    {
      "epoch": 0.3209175830672815,
      "grad_norm": 0.1503458470106125,
      "learning_rate": 6.189302531402601e-06,
      "loss": 0.35,
      "num_tokens": 1718783402.0,
      "step": 2714
    },
    {
      "epoch": 0.3210358283079106,
      "grad_norm": 0.14808592200279236,
      "learning_rate": 6.185648795425611e-06,
      "loss": 0.3547,
      "num_tokens": 1719417133.0,
      "step": 2715
    },
    {
      "epoch": 0.3211540735485397,
      "grad_norm": 0.14757515490055084,
      "learning_rate": 6.1820305427149105e-06,
      "loss": 0.3212,
      "num_tokens": 1720055957.0,
      "step": 2716
    },
    {
      "epoch": 0.32127231878916873,
      "grad_norm": 0.14903753995895386,
      "learning_rate": 6.178447778058516e-06,
      "loss": 0.3389,
      "num_tokens": 1720672344.0,
      "step": 2717
    },
    {
      "epoch": 0.3213905640297978,
      "grad_norm": 0.14562296867370605,
      "learning_rate": 6.174900506197477e-06,
      "loss": 0.3312,
      "num_tokens": 1721301318.0,
      "step": 2718
    },
    {
      "epoch": 0.32150880927042685,
      "grad_norm": 0.15241187810897827,
      "learning_rate": 6.171388731825877e-06,
      "loss": 0.3683,
      "num_tokens": 1721927703.0,
      "step": 2719
    },
    {
      "epoch": 0.32162705451105594,
      "grad_norm": 0.14671999216079712,
      "learning_rate": 6.167912459590827e-06,
      "loss": 0.3284,
      "num_tokens": 1722565193.0,
      "step": 2720
    },
    {
      "epoch": 0.321745299751685,
      "grad_norm": 0.15789209306240082,
      "learning_rate": 6.164471694092456e-06,
      "loss": 0.3568,
      "num_tokens": 1723200696.0,
      "step": 2721
    },
    {
      "epoch": 0.32186354499231407,
      "grad_norm": 0.16588889062404633,
      "learning_rate": 6.161066439883903e-06,
      "loss": 0.3705,
      "num_tokens": 1723839772.0,
      "step": 2722
    },
    {
      "epoch": 0.3219817902329431,
      "grad_norm": 0.16096794605255127,
      "learning_rate": 6.157696701471324e-06,
      "loss": 0.382,
      "num_tokens": 1724477404.0,
      "step": 2723
    },
    {
      "epoch": 0.3221000354735722,
      "grad_norm": 0.14348092675209045,
      "learning_rate": 6.154362483313876e-06,
      "loss": 0.3303,
      "num_tokens": 1725115337.0,
      "step": 2724
    },
    {
      "epoch": 0.3222182807142013,
      "grad_norm": 0.14821667969226837,
      "learning_rate": 6.151063789823705e-06,
      "loss": 0.3286,
      "num_tokens": 1725751875.0,
      "step": 2725
    },
    {
      "epoch": 0.3223365259548303,
      "grad_norm": 0.16842487454414368,
      "learning_rate": 6.1478006253659545e-06,
      "loss": 0.3153,
      "num_tokens": 1726386668.0,
      "step": 2726
    },
    {
      "epoch": 0.3224547711954594,
      "grad_norm": 0.14876829087734222,
      "learning_rate": 6.144572994258746e-06,
      "loss": 0.3477,
      "num_tokens": 1727024471.0,
      "step": 2727
    },
    {
      "epoch": 0.32257301643608843,
      "grad_norm": 0.16405583918094635,
      "learning_rate": 6.141380900773187e-06,
      "loss": 0.3453,
      "num_tokens": 1727657575.0,
      "step": 2728
    },
    {
      "epoch": 0.3226912616767175,
      "grad_norm": 0.16747678816318512,
      "learning_rate": 6.138224349133355e-06,
      "loss": 0.3412,
      "num_tokens": 1728294968.0,
      "step": 2729
    },
    {
      "epoch": 0.32280950691734656,
      "grad_norm": 0.14372828602790833,
      "learning_rate": 6.135103343516297e-06,
      "loss": 0.3232,
      "num_tokens": 1728933844.0,
      "step": 2730
    },
    {
      "epoch": 0.32292775215797564,
      "grad_norm": 0.1462886780500412,
      "learning_rate": 6.132017888052019e-06,
      "loss": 0.3362,
      "num_tokens": 1729563923.0,
      "step": 2731
    },
    {
      "epoch": 0.3230459973986047,
      "grad_norm": 0.15404145419597626,
      "learning_rate": 6.128967986823482e-06,
      "loss": 0.3458,
      "num_tokens": 1730193072.0,
      "step": 2732
    },
    {
      "epoch": 0.32316424263923377,
      "grad_norm": 0.15265800058841705,
      "learning_rate": 6.125953643866606e-06,
      "loss": 0.3348,
      "num_tokens": 1730826376.0,
      "step": 2733
    },
    {
      "epoch": 0.32328248787986286,
      "grad_norm": 0.17396466434001923,
      "learning_rate": 6.122974863170252e-06,
      "loss": 0.3679,
      "num_tokens": 1731457578.0,
      "step": 2734
    },
    {
      "epoch": 0.3234007331204919,
      "grad_norm": 0.1429223269224167,
      "learning_rate": 6.120031648676221e-06,
      "loss": 0.3076,
      "num_tokens": 1732094125.0,
      "step": 2735
    },
    {
      "epoch": 0.323518978361121,
      "grad_norm": 0.17876330018043518,
      "learning_rate": 6.1171240042792495e-06,
      "loss": 0.4053,
      "num_tokens": 1732695672.0,
      "step": 2736
    },
    {
      "epoch": 0.32363722360175,
      "grad_norm": 0.15872971713542938,
      "learning_rate": 6.114251933827009e-06,
      "loss": 0.3741,
      "num_tokens": 1733327345.0,
      "step": 2737
    },
    {
      "epoch": 0.3237554688423791,
      "grad_norm": 0.14188595116138458,
      "learning_rate": 6.111415441120089e-06,
      "loss": 0.3303,
      "num_tokens": 1733965576.0,
      "step": 2738
    },
    {
      "epoch": 0.32387371408300814,
      "grad_norm": 0.148431196808815,
      "learning_rate": 6.108614529912007e-06,
      "loss": 0.339,
      "num_tokens": 1734592394.0,
      "step": 2739
    },
    {
      "epoch": 0.3239919593236372,
      "grad_norm": 0.14060525596141815,
      "learning_rate": 6.10584920390919e-06,
      "loss": 0.3421,
      "num_tokens": 1735221890.0,
      "step": 2740
    },
    {
      "epoch": 0.3241102045642663,
      "grad_norm": 0.16482548415660858,
      "learning_rate": 6.103119466770971e-06,
      "loss": 0.3863,
      "num_tokens": 1735857891.0,
      "step": 2741
    },
    {
      "epoch": 0.32422844980489535,
      "grad_norm": 0.16050606966018677,
      "learning_rate": 6.100425322109603e-06,
      "loss": 0.3649,
      "num_tokens": 1736492956.0,
      "step": 2742
    },
    {
      "epoch": 0.32434669504552444,
      "grad_norm": 0.15739870071411133,
      "learning_rate": 6.097766773490227e-06,
      "loss": 0.3793,
      "num_tokens": 1737125568.0,
      "step": 2743
    },
    {
      "epoch": 0.32446494028615347,
      "grad_norm": 0.15872135758399963,
      "learning_rate": 6.095143824430883e-06,
      "loss": 0.3695,
      "num_tokens": 1737736959.0,
      "step": 2744
    },
    {
      "epoch": 0.32458318552678256,
      "grad_norm": 0.1529282182455063,
      "learning_rate": 6.0925564784025065e-06,
      "loss": 0.3506,
      "num_tokens": 1738364118.0,
      "step": 2745
    },
    {
      "epoch": 0.3247014307674116,
      "grad_norm": 0.1539255529642105,
      "learning_rate": 6.090004738828917e-06,
      "loss": 0.3245,
      "num_tokens": 1739000213.0,
      "step": 2746
    },
    {
      "epoch": 0.3248196760080407,
      "grad_norm": 0.16003718972206116,
      "learning_rate": 6.087488609086811e-06,
      "loss": 0.3712,
      "num_tokens": 1739635841.0,
      "step": 2747
    },
    {
      "epoch": 0.3249379212486697,
      "grad_norm": 0.14386966824531555,
      "learning_rate": 6.085008092505771e-06,
      "loss": 0.3046,
      "num_tokens": 1740270727.0,
      "step": 2748
    },
    {
      "epoch": 0.3250561664892988,
      "grad_norm": 0.14387185871601105,
      "learning_rate": 6.082563192368251e-06,
      "loss": 0.3092,
      "num_tokens": 1740910145.0,
      "step": 2749
    },
    {
      "epoch": 0.3251744117299279,
      "grad_norm": 0.1531536877155304,
      "learning_rate": 6.0801539119095686e-06,
      "loss": 0.3296,
      "num_tokens": 1741545955.0,
      "step": 2750
    },
    {
      "epoch": 0.3252926569705569,
      "grad_norm": 0.1645767241716385,
      "learning_rate": 6.077780254317911e-06,
      "loss": 0.3373,
      "num_tokens": 1742182116.0,
      "step": 2751
    },
    {
      "epoch": 0.325410902211186,
      "grad_norm": 0.15716831386089325,
      "learning_rate": 6.075442222734327e-06,
      "loss": 0.3461,
      "num_tokens": 1742817038.0,
      "step": 2752
    },
    {
      "epoch": 0.32552914745181505,
      "grad_norm": 0.15224866569042206,
      "learning_rate": 6.073139820252716e-06,
      "loss": 0.3191,
      "num_tokens": 1743447171.0,
      "step": 2753
    },
    {
      "epoch": 0.32564739269244414,
      "grad_norm": 0.137272447347641,
      "learning_rate": 6.070873049919833e-06,
      "loss": 0.3014,
      "num_tokens": 1744082013.0,
      "step": 2754
    },
    {
      "epoch": 0.32576563793307317,
      "grad_norm": 0.1579388976097107,
      "learning_rate": 6.068641914735285e-06,
      "loss": 0.3649,
      "num_tokens": 1744718914.0,
      "step": 2755
    },
    {
      "epoch": 0.32588388317370226,
      "grad_norm": 0.160564124584198,
      "learning_rate": 6.0664464176515185e-06,
      "loss": 0.3511,
      "num_tokens": 1745347427.0,
      "step": 2756
    },
    {
      "epoch": 0.32600212841433135,
      "grad_norm": 0.15549030900001526,
      "learning_rate": 6.064286561573819e-06,
      "loss": 0.35,
      "num_tokens": 1745987038.0,
      "step": 2757
    },
    {
      "epoch": 0.3261203736549604,
      "grad_norm": 0.1522926241159439,
      "learning_rate": 6.0621623493603116e-06,
      "loss": 0.3607,
      "num_tokens": 1746619830.0,
      "step": 2758
    },
    {
      "epoch": 0.32623861889558947,
      "grad_norm": 0.1506992131471634,
      "learning_rate": 6.060073783821954e-06,
      "loss": 0.3916,
      "num_tokens": 1747252651.0,
      "step": 2759
    },
    {
      "epoch": 0.3263568641362185,
      "grad_norm": 0.15200266242027283,
      "learning_rate": 6.05802086772253e-06,
      "loss": 0.3717,
      "num_tokens": 1747887273.0,
      "step": 2760
    },
    {
      "epoch": 0.3264751093768476,
      "grad_norm": 0.14301730692386627,
      "learning_rate": 6.056003603778654e-06,
      "loss": 0.3351,
      "num_tokens": 1748523868.0,
      "step": 2761
    },
    {
      "epoch": 0.32659335461747663,
      "grad_norm": 0.15410567820072174,
      "learning_rate": 6.054021994659758e-06,
      "loss": 0.34,
      "num_tokens": 1749155294.0,
      "step": 2762
    },
    {
      "epoch": 0.3267115998581057,
      "grad_norm": 0.15273074805736542,
      "learning_rate": 6.052076042988093e-06,
      "loss": 0.3257,
      "num_tokens": 1749791054.0,
      "step": 2763
    },
    {
      "epoch": 0.32682984509873475,
      "grad_norm": 0.1528049111366272,
      "learning_rate": 6.050165751338724e-06,
      "loss": 0.3165,
      "num_tokens": 1750428935.0,
      "step": 2764
    },
    {
      "epoch": 0.32694809033936384,
      "grad_norm": 0.1597285270690918,
      "learning_rate": 6.0482911222395316e-06,
      "loss": 0.3743,
      "num_tokens": 1751065445.0,
      "step": 2765
    },
    {
      "epoch": 0.32706633557999293,
      "grad_norm": 0.15127848088741302,
      "learning_rate": 6.046452158171194e-06,
      "loss": 0.3488,
      "num_tokens": 1751702074.0,
      "step": 2766
    },
    {
      "epoch": 0.32718458082062196,
      "grad_norm": 0.1535509079694748,
      "learning_rate": 6.044648861567207e-06,
      "loss": 0.3532,
      "num_tokens": 1752338344.0,
      "step": 2767
    },
    {
      "epoch": 0.32730282606125105,
      "grad_norm": 0.16607314348220825,
      "learning_rate": 6.042881234813861e-06,
      "loss": 0.3621,
      "num_tokens": 1752976568.0,
      "step": 2768
    },
    {
      "epoch": 0.3274210713018801,
      "grad_norm": 0.14180459082126617,
      "learning_rate": 6.041149280250246e-06,
      "loss": 0.332,
      "num_tokens": 1753603003.0,
      "step": 2769
    },
    {
      "epoch": 0.3275393165425092,
      "grad_norm": 0.1491679847240448,
      "learning_rate": 6.039453000168244e-06,
      "loss": 0.345,
      "num_tokens": 1754241935.0,
      "step": 2770
    },
    {
      "epoch": 0.3276575617831382,
      "grad_norm": 0.14957982301712036,
      "learning_rate": 6.037792396812535e-06,
      "loss": 0.366,
      "num_tokens": 1754872359.0,
      "step": 2771
    },
    {
      "epoch": 0.3277758070237673,
      "grad_norm": 0.14234739542007446,
      "learning_rate": 6.036167472380583e-06,
      "loss": 0.2863,
      "num_tokens": 1755511603.0,
      "step": 2772
    },
    {
      "epoch": 0.3278940522643964,
      "grad_norm": 0.16358140110969543,
      "learning_rate": 6.034578229022643e-06,
      "loss": 0.3651,
      "num_tokens": 1756143176.0,
      "step": 2773
    },
    {
      "epoch": 0.3280122975050254,
      "grad_norm": 0.16191521286964417,
      "learning_rate": 6.033024668841748e-06,
      "loss": 0.3497,
      "num_tokens": 1756756307.0,
      "step": 2774
    },
    {
      "epoch": 0.3281305427456545,
      "grad_norm": 0.14113369584083557,
      "learning_rate": 6.031506793893716e-06,
      "loss": 0.3439,
      "num_tokens": 1757390678.0,
      "step": 2775
    },
    {
      "epoch": 0.32824878798628354,
      "grad_norm": 0.14521457254886627,
      "learning_rate": 6.030024606187143e-06,
      "loss": 0.3371,
      "num_tokens": 1758029292.0,
      "step": 2776
    },
    {
      "epoch": 0.32836703322691263,
      "grad_norm": 0.1455880105495453,
      "learning_rate": 6.028578107683398e-06,
      "loss": 0.3538,
      "num_tokens": 1758663414.0,
      "step": 2777
    },
    {
      "epoch": 0.32848527846754166,
      "grad_norm": 0.143596813082695,
      "learning_rate": 6.0271673002966225e-06,
      "loss": 0.3284,
      "num_tokens": 1759297216.0,
      "step": 2778
    },
    {
      "epoch": 0.32860352370817075,
      "grad_norm": 0.14440219104290009,
      "learning_rate": 6.025792185893728e-06,
      "loss": 0.3409,
      "num_tokens": 1759927411.0,
      "step": 2779
    },
    {
      "epoch": 0.3287217689487998,
      "grad_norm": 0.16369512677192688,
      "learning_rate": 6.0244527662944e-06,
      "loss": 0.3611,
      "num_tokens": 1760561865.0,
      "step": 2780
    },
    {
      "epoch": 0.3288400141894289,
      "grad_norm": 0.160583034157753,
      "learning_rate": 6.023149043271077e-06,
      "loss": 0.3806,
      "num_tokens": 1761198485.0,
      "step": 2781
    },
    {
      "epoch": 0.32895825943005796,
      "grad_norm": 0.15042239427566528,
      "learning_rate": 6.021881018548975e-06,
      "loss": 0.301,
      "num_tokens": 1761832996.0,
      "step": 2782
    },
    {
      "epoch": 0.329076504670687,
      "grad_norm": 0.15653792023658752,
      "learning_rate": 6.020648693806062e-06,
      "loss": 0.3781,
      "num_tokens": 1762469073.0,
      "step": 2783
    },
    {
      "epoch": 0.3291947499113161,
      "grad_norm": 0.14766980707645416,
      "learning_rate": 6.0194520706730605e-06,
      "loss": 0.33,
      "num_tokens": 1763102453.0,
      "step": 2784
    },
    {
      "epoch": 0.3293129951519451,
      "grad_norm": 0.1543278843164444,
      "learning_rate": 6.018291150733459e-06,
      "loss": 0.3735,
      "num_tokens": 1763740288.0,
      "step": 2785
    },
    {
      "epoch": 0.3294312403925742,
      "grad_norm": 0.16179020702838898,
      "learning_rate": 6.017165935523497e-06,
      "loss": 0.3704,
      "num_tokens": 1764368292.0,
      "step": 2786
    },
    {
      "epoch": 0.32954948563320324,
      "grad_norm": 0.1551685631275177,
      "learning_rate": 6.01607642653216e-06,
      "loss": 0.3604,
      "num_tokens": 1764998177.0,
      "step": 2787
    },
    {
      "epoch": 0.32966773087383233,
      "grad_norm": 0.13794450461864471,
      "learning_rate": 6.015022625201192e-06,
      "loss": 0.3112,
      "num_tokens": 1765629891.0,
      "step": 2788
    },
    {
      "epoch": 0.32978597611446137,
      "grad_norm": 0.14764420688152313,
      "learning_rate": 6.0140045329250805e-06,
      "loss": 0.348,
      "num_tokens": 1766266964.0,
      "step": 2789
    },
    {
      "epoch": 0.32990422135509045,
      "grad_norm": 0.15834534168243408,
      "learning_rate": 6.013022151051061e-06,
      "loss": 0.3447,
      "num_tokens": 1766904529.0,
      "step": 2790
    },
    {
      "epoch": 0.33002246659571954,
      "grad_norm": 0.14628948271274567,
      "learning_rate": 6.012075480879112e-06,
      "loss": 0.3693,
      "num_tokens": 1767537573.0,
      "step": 2791
    },
    {
      "epoch": 0.3301407118363486,
      "grad_norm": 0.15087901055812836,
      "learning_rate": 6.011164523661962e-06,
      "loss": 0.3558,
      "num_tokens": 1768173295.0,
      "step": 2792
    },
    {
      "epoch": 0.33025895707697767,
      "grad_norm": 0.1705261468887329,
      "learning_rate": 6.010289280605068e-06,
      "loss": 0.3896,
      "num_tokens": 1768810003.0,
      "step": 2793
    },
    {
      "epoch": 0.3303772023176067,
      "grad_norm": 0.15986499190330505,
      "learning_rate": 6.00944975286664e-06,
      "loss": 0.3592,
      "num_tokens": 1769440243.0,
      "step": 2794
    },
    {
      "epoch": 0.3304954475582358,
      "grad_norm": 0.14154967665672302,
      "learning_rate": 6.0086459415576145e-06,
      "loss": 0.3463,
      "num_tokens": 1770079311.0,
      "step": 2795
    },
    {
      "epoch": 0.3306136927988648,
      "grad_norm": 0.15516899526119232,
      "learning_rate": 6.00787784774167e-06,
      "loss": 0.3453,
      "num_tokens": 1770715195.0,
      "step": 2796
    },
    {
      "epoch": 0.3307319380394939,
      "grad_norm": 0.15446634590625763,
      "learning_rate": 6.007145472435228e-06,
      "loss": 0.33,
      "num_tokens": 1771349599.0,
      "step": 2797
    },
    {
      "epoch": 0.330850183280123,
      "grad_norm": 0.15723447501659393,
      "learning_rate": 6.0064488166074275e-06,
      "loss": 0.3421,
      "num_tokens": 1771943463.0,
      "step": 2798
    },
    {
      "epoch": 0.33096842852075203,
      "grad_norm": 0.14613540470600128,
      "learning_rate": 6.005787881180151e-06,
      "loss": 0.3381,
      "num_tokens": 1772581089.0,
      "step": 2799
    },
    {
      "epoch": 0.3310866737613811,
      "grad_norm": 0.15577061474323273,
      "learning_rate": 6.005162667028012e-06,
      "loss": 0.3441,
      "num_tokens": 1773213267.0,
      "step": 2800
    },
    {
      "epoch": 0.33120491900201016,
      "grad_norm": 0.15103386342525482,
      "learning_rate": 6.004573174978353e-06,
      "loss": 0.3228,
      "num_tokens": 1773843997.0,
      "step": 2801
    },
    {
      "epoch": 0.33132316424263925,
      "grad_norm": 0.1580219715833664,
      "learning_rate": 6.004019405811245e-06,
      "loss": 0.3462,
      "num_tokens": 1774443566.0,
      "step": 2802
    },
    {
      "epoch": 0.3314414094832683,
      "grad_norm": 0.1480458378791809,
      "learning_rate": 6.003501360259485e-06,
      "loss": 0.3636,
      "num_tokens": 1775076095.0,
      "step": 2803
    },
    {
      "epoch": 0.33155965472389737,
      "grad_norm": 0.16500131785869598,
      "learning_rate": 6.003019039008604e-06,
      "loss": 0.3803,
      "num_tokens": 1775705995.0,
      "step": 2804
    },
    {
      "epoch": 0.3316778999645264,
      "grad_norm": 0.14276215434074402,
      "learning_rate": 6.0025724426968495e-06,
      "loss": 0.345,
      "num_tokens": 1776340157.0,
      "step": 2805
    },
    {
      "epoch": 0.3317961452051555,
      "grad_norm": 0.14830231666564941,
      "learning_rate": 6.002161571915202e-06,
      "loss": 0.3383,
      "num_tokens": 1776975868.0,
      "step": 2806
    },
    {
      "epoch": 0.3319143904457846,
      "grad_norm": 0.14599831402301788,
      "learning_rate": 6.001786427207366e-06,
      "loss": 0.3453,
      "num_tokens": 1777609915.0,
      "step": 2807
    },
    {
      "epoch": 0.3320326356864136,
      "grad_norm": 0.15465782582759857,
      "learning_rate": 6.0014470090697654e-06,
      "loss": 0.3297,
      "num_tokens": 1778243872.0,
      "step": 2808
    },
    {
      "epoch": 0.3321508809270427,
      "grad_norm": 0.1415361762046814,
      "learning_rate": 6.00114331795155e-06,
      "loss": 0.3511,
      "num_tokens": 1778876930.0,
      "step": 2809
    },
    {
      "epoch": 0.33226912616767174,
      "grad_norm": 0.1687016785144806,
      "learning_rate": 6.000875354254595e-06,
      "loss": 0.3536,
      "num_tokens": 1779509817.0,
      "step": 2810
    },
    {
      "epoch": 0.3323873714083008,
      "grad_norm": 0.1572532057762146,
      "learning_rate": 6.000643118333494e-06,
      "loss": 0.3789,
      "num_tokens": 1780144087.0,
      "step": 2811
    },
    {
      "epoch": 0.33250561664892986,
      "grad_norm": 0.15837253630161285,
      "learning_rate": 6.000446610495562e-06,
      "loss": 0.3558,
      "num_tokens": 1780778006.0,
      "step": 2812
    },
    {
      "epoch": 0.33262386188955895,
      "grad_norm": 0.1413705050945282,
      "learning_rate": 6.0002858310008385e-06,
      "loss": 0.3238,
      "num_tokens": 1781416088.0,
      "step": 2813
    },
    {
      "epoch": 0.33274210713018804,
      "grad_norm": 0.17787781357765198,
      "learning_rate": 6.00016078006208e-06,
      "loss": 0.3605,
      "num_tokens": 1782052246.0,
      "step": 2814
    },
    {
      "epoch": 0.33286035237081707,
      "grad_norm": 0.15967637300491333,
      "learning_rate": 6.000071457844771e-06,
      "loss": 0.3456,
      "num_tokens": 1782691473.0,
      "step": 2815
    },
    {
      "epoch": 0.33297859761144616,
      "grad_norm": 0.14508619904518127,
      "learning_rate": 6.000017864467102e-06,
      "loss": 0.3088,
      "num_tokens": 1783330384.0,
      "step": 2816
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 2816,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 2816,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.5080662592441549e+19,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}