{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9999398785546805,
  "eval_steps": 500,
  "global_step": 8316,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00012024289063909097,
      "flos": 17735410237440.0,
      "grad_norm": 12.747928825148605,
      "learning_rate": 0.0,
      "loss": 1.7551,
      "num_input_tokens_seen": 20095,
      "step": 1
    },
    {
      "epoch": 0.00024048578127818193,
      "flos": 21856559554560.0,
      "grad_norm": 16.376437757432075,
      "learning_rate": 5.021476677069823e-07,
      "loss": 1.7109,
      "num_input_tokens_seen": 38475,
      "step": 2
    },
    {
      "epoch": 0.0003607286719172729,
      "flos": 13492340736000.0,
      "grad_norm": 12.649838026615573,
      "learning_rate": 7.958852231401551e-07,
      "loss": 1.5691,
      "num_input_tokens_seen": 56760,
      "step": 3
    },
    {
      "epoch": 0.00048097156255636386,
      "flos": 13700577239040.0,
      "grad_norm": 11.185202932823236,
      "learning_rate": 1.0042953354139647e-06,
      "loss": 1.6493,
      "num_input_tokens_seen": 75345,
      "step": 4
    },
    {
      "epoch": 0.0006012144531954548,
      "flos": 9873818050560.0,
      "grad_norm": 15.71763895101093,
      "learning_rate": 1.1659507774310057e-06,
      "loss": 1.7152,
      "num_input_tokens_seen": 92950,
      "step": 5
    },
    {
      "epoch": 0.0007214573438345458,
      "flos": 16455272632320.0,
      "grad_norm": 15.7781682093016,
      "learning_rate": 1.2980328908471373e-06,
      "loss": 1.5882,
      "num_input_tokens_seen": 112915,
      "step": 6
    },
    {
      "epoch": 0.0008417002344736367,
      "flos": 48451970273280.0,
      "grad_norm": 3.6980945526463356,
      "learning_rate": 1.4097067265369432e-06,
      "loss": 1.1283,
      "num_input_tokens_seen": 177630,
      "step": 7
    },
    {
      "epoch": 0.0009619431251127277,
      "flos": 15112136724480.0,
      "grad_norm": 23.717557251383546,
      "learning_rate": 1.506443003120947e-06,
      "loss": 1.5371,
      "num_input_tokens_seen": 192850,
      "step": 8
    },
    {
      "epoch": 0.0010821860157518186,
      "flos": 16586317209600.0,
      "grad_norm": 10.312682364036231,
      "learning_rate": 1.5917704462803102e-06,
      "loss": 1.6364,
      "num_input_tokens_seen": 209115,
      "step": 9
    },
    {
      "epoch": 0.0012024289063909096,
      "flos": 12042174013440.0,
      "grad_norm": 11.340662512424181,
      "learning_rate": 1.6680984451379884e-06,
      "loss": 1.6802,
      "num_input_tokens_seen": 221905,
      "step": 10
    },
    {
      "epoch": 0.0013226717970300007,
      "flos": 15117900533760.0,
      "grad_norm": 8.771848159276045,
      "learning_rate": 1.7371455188905097e-06,
      "loss": 1.4622,
      "num_input_tokens_seen": 241555,
      "step": 11
    },
    {
      "epoch": 0.0014429146876690916,
      "flos": 19394221424640.0,
      "grad_norm": 13.950534627097126,
      "learning_rate": 1.8001805585541196e-06,
      "loss": 1.4415,
      "num_input_tokens_seen": 262405,
      "step": 12
    },
    {
      "epoch": 0.0015631575783081825,
      "flos": 13518102036480.0,
      "grad_norm": 11.72601662724315,
      "learning_rate": 1.8581671739548328e-06,
      "loss": 1.5124,
      "num_input_tokens_seen": 279860,
      "step": 13
    },
    {
      "epoch": 0.0016834004689472734,
      "flos": 34415155261440.0,
      "grad_norm": 6.822045314757982,
      "learning_rate": 1.9118543942439254e-06,
      "loss": 1.3428,
      "num_input_tokens_seen": 301765,
      "step": 14
    },
    {
      "epoch": 0.0018036433595863645,
      "flos": 24950781296640.0,
      "grad_norm": 6.879192114570571,
      "learning_rate": 1.961836000571161e-06,
      "loss": 1.3525,
      "num_input_tokens_seen": 323140,
      "step": 15
    },
    {
      "epoch": 0.0019238862502254555,
      "flos": 42776116162560.0,
      "grad_norm": 3.0860240296412855,
      "learning_rate": 2.0085906708279293e-06,
      "loss": 0.9436,
      "num_input_tokens_seen": 378230,
      "step": 16
    },
    {
      "epoch": 0.0020441291408645466,
      "flos": 14777638379520.0,
      "grad_norm": 10.467127250045847,
      "learning_rate": 2.0525099325728135e-06,
      "loss": 1.4006,
      "num_input_tokens_seen": 396130,
      "step": 17
    },
    {
      "epoch": 0.0021643720315036373,
      "flos": 45714911662080.0,
      "grad_norm": 2.688858841981265,
      "learning_rate": 2.0939181139872922e-06,
      "loss": 1.0168,
      "num_input_tokens_seen": 457565,
      "step": 18
    },
    {
      "epoch": 0.0022846149221427284,
      "flos": 22302411571200.0,
      "grad_norm": 10.183023978954239,
      "learning_rate": 2.1330868934640175e-06,
      "loss": 1.2428,
      "num_input_tokens_seen": 477960,
      "step": 19
    },
    {
      "epoch": 0.002404857812781819,
      "flos": 36534494515200.0,
      "grad_norm": 2.415301255101124,
      "learning_rate": 2.170246112844971e-06,
      "loss": 1.0524,
      "num_input_tokens_seen": 532020,
      "step": 20
    },
    {
      "epoch": 0.0025251007034209102,
      "flos": 11184723701760.0,
      "grad_norm": 5.097995977664287,
      "learning_rate": 2.2055919496770983e-06,
      "loss": 1.2688,
      "num_input_tokens_seen": 549880,
      "step": 21
    },
    {
      "epoch": 0.0026453435940600014,
      "flos": 27021825208320.0,
      "grad_norm": 4.920639997275567,
      "learning_rate": 2.2392931865974923e-06,
      "loss": 1.1597,
      "num_input_tokens_seen": 572290,
      "step": 22
    },
    {
      "epoch": 0.002765586484699092,
      "flos": 15012831866880.0,
      "grad_norm": 6.579441256004261,
      "learning_rate": 2.271496085962064e-06,
      "loss": 1.2772,
      "num_input_tokens_seen": 589705,
      "step": 23
    },
    {
      "epoch": 0.002885829375338183,
      "flos": 14671711272960.0,
      "grad_norm": 8.553570846070663,
      "learning_rate": 2.3023282262611022e-06,
      "loss": 1.2857,
      "num_input_tokens_seen": 608200,
      "step": 24
    },
    {
      "epoch": 0.003006072265977274,
      "flos": 24845712629760.0,
      "grad_norm": 5.320106177453593,
      "learning_rate": 2.3319015548620114e-06,
      "loss": 1.1637,
      "num_input_tokens_seen": 629060,
      "step": 25
    },
    {
      "epoch": 0.003126315156616365,
      "flos": 17369641451520.0,
      "grad_norm": 3.222817822654203,
      "learning_rate": 2.3603148416618152e-06,
      "loss": 1.1784,
      "num_input_tokens_seen": 648295,
      "step": 26
    },
    {
      "epoch": 0.003246558047255456,
      "flos": 16795320176640.0,
      "grad_norm": 3.551802518750465,
      "learning_rate": 2.3876556694204647e-06,
      "loss": 1.2573,
      "num_input_tokens_seen": 668170,
      "step": 27
    },
    {
      "epoch": 0.003366800937894547,
      "flos": 12625386270720.0,
      "grad_norm": 4.850119064285344,
      "learning_rate": 2.414002061950908e-06,
      "loss": 1.1448,
      "num_input_tokens_seen": 686765,
      "step": 28
    },
    {
      "epoch": 0.003487043828533638,
      "flos": 17373136527360.0,
      "grad_norm": 4.344488067494663,
      "learning_rate": 2.4394238264681557e-06,
      "loss": 1.2579,
      "num_input_tokens_seen": 706220,
      "step": 29
    },
    {
      "epoch": 0.003607286719172729,
      "flos": 18605010739200.0,
      "grad_norm": 3.345718374846667,
      "learning_rate": 2.4639836682781433e-06,
      "loss": 1.243,
      "num_input_tokens_seen": 726070,
      "step": 30
    },
    {
      "epoch": 0.00372752960981182,
      "flos": 14618134548480.0,
      "grad_norm": 7.111203136775818,
      "learning_rate": 2.487738122623307e-06,
      "loss": 1.2276,
      "num_input_tokens_seen": 744700,
      "step": 31
    },
    {
      "epoch": 0.003847772500450911,
      "flos": 16114335989760.0,
      "grad_norm": 7.706784754357682,
      "learning_rate": 2.510738338534912e-06,
      "loss": 1.2158,
      "num_input_tokens_seen": 763145,
      "step": 32
    },
    {
      "epoch": 0.003968015391090002,
      "flos": 12731619962880.0,
      "grad_norm": 3.9603592490031194,
      "learning_rate": 2.5330307420306648e-06,
      "loss": 1.272,
      "num_input_tokens_seen": 779715,
      "step": 33
    },
    {
      "epoch": 0.004088258281729093,
      "flos": 19440899973120.0,
      "grad_norm": 3.4174331332559937,
      "learning_rate": 2.554657600279796e-06,
      "loss": 1.1137,
      "num_input_tokens_seen": 800710,
      "step": 34
    },
    {
      "epoch": 0.004208501172368184,
      "flos": 16534058803200.0,
      "grad_norm": 6.00992938820588,
      "learning_rate": 2.5756575039679493e-06,
      "loss": 1.2881,
      "num_input_tokens_seen": 819980,
      "step": 35
    },
    {
      "epoch": 0.0043287440630072746,
      "flos": 12260527841280.0,
      "grad_norm": 3.0672531610612985,
      "learning_rate": 2.5960657816942747e-06,
      "loss": 1.1935,
      "num_input_tokens_seen": 838615,
      "step": 36
    },
    {
      "epoch": 0.004448986953646365,
      "flos": 37978498867200.0,
      "grad_norm": 1.86347861214399,
      "learning_rate": 2.6159148575788668e-06,
      "loss": 0.909,
      "num_input_tokens_seen": 896370,
      "step": 37
    },
    {
      "epoch": 0.004569229844285457,
      "flos": 9480653660160.0,
      "grad_norm": 4.080059192078499,
      "learning_rate": 2.635234561171e-06,
      "loss": 1.2272,
      "num_input_tokens_seen": 914485,
      "step": 38
    },
    {
      "epoch": 0.0046894727349245475,
      "flos": 11467116933120.0,
      "grad_norm": 3.3771829962609927,
      "learning_rate": 2.6540523970949877e-06,
      "loss": 1.1859,
      "num_input_tokens_seen": 929050,
      "step": 39
    },
    {
      "epoch": 0.004809715625563638,
      "flos": 17005120266240.0,
      "grad_norm": 5.433983553651275,
      "learning_rate": 2.6723937805519533e-06,
      "loss": 1.1707,
      "num_input_tokens_seen": 946270,
      "step": 40
    },
    {
      "epoch": 0.00492995851620273,
      "flos": 14747125002240.0,
      "grad_norm": 5.506835158189562,
      "learning_rate": 2.690282243737839e-06,
      "loss": 1.1724,
      "num_input_tokens_seen": 964925,
      "step": 41
    },
    {
      "epoch": 0.0050502014068418205,
      "flos": 14435414077440.0,
      "grad_norm": 5.313202575899853,
      "learning_rate": 2.7077396173840807e-06,
      "loss": 1.2281,
      "num_input_tokens_seen": 982930,
      "step": 42
    },
    {
      "epoch": 0.005170444297480911,
      "flos": 18500555243520.0,
      "grad_norm": 7.999273982482796,
      "learning_rate": 2.7247861909342594e-06,
      "loss": 1.1574,
      "num_input_tokens_seen": 1003575,
      "step": 43
    },
    {
      "epoch": 0.005290687188120003,
      "flos": 14877035212800.0,
      "grad_norm": 6.960810633401276,
      "learning_rate": 2.7414408543044743e-06,
      "loss": 1.0657,
      "num_input_tokens_seen": 1018935,
      "step": 44
    },
    {
      "epoch": 0.005410930078759093,
      "flos": 11210576977920.0,
      "grad_norm": 7.751865777647854,
      "learning_rate": 2.7577212237113157e-06,
      "loss": 1.0305,
      "num_input_tokens_seen": 1035695,
      "step": 45
    },
    {
      "epoch": 0.005531172969398184,
      "flos": 14986181468160.0,
      "grad_norm": 5.02070659725479,
      "learning_rate": 2.7736437536690466e-06,
      "loss": 1.2776,
      "num_input_tokens_seen": 1055045,
      "step": 46
    },
    {
      "epoch": 0.005651415860037276,
      "flos": 14798616944640.0,
      "grad_norm": 3.417608937534333,
      "learning_rate": 2.789223836941131e-06,
      "loss": 1.3079,
      "num_input_tokens_seen": 1074900,
      "step": 47
    },
    {
      "epoch": 0.005771658750676366,
      "flos": 9348229447680.0,
      "grad_norm": 8.917056303929083,
      "learning_rate": 2.8044758939680847e-06,
      "loss": 1.2954,
      "num_input_tokens_seen": 1090690,
      "step": 48
    },
    {
      "epoch": 0.005891901641315457,
      "flos": 17372523356160.0,
      "grad_norm": 5.776102609774395,
      "learning_rate": 2.8194134530738863e-06,
      "loss": 1.2402,
      "num_input_tokens_seen": 1109180,
      "step": 49
    },
    {
      "epoch": 0.006012144531954548,
      "flos": 16402983567360.0,
      "grad_norm": 6.434233197673287,
      "learning_rate": 2.834049222568994e-06,
      "loss": 1.1323,
      "num_input_tokens_seen": 1126250,
      "step": 50
    },
    {
      "epoch": 0.006132387422593639,
      "flos": 16009543249920.0,
      "grad_norm": 10.561291507003224,
      "learning_rate": 2.848395155712969e-06,
      "loss": 1.1643,
      "num_input_tokens_seen": 1146190,
      "step": 51
    },
    {
      "epoch": 0.00625263031323273,
      "flos": 19673364848640.0,
      "grad_norm": 3.6847760759535215,
      "learning_rate": 2.8624625093687977e-06,
      "loss": 1.2198,
      "num_input_tokens_seen": 1163045,
      "step": 52
    },
    {
      "epoch": 0.006372873203871821,
      "flos": 16427487866880.0,
      "grad_norm": 4.5528908485417645,
      "learning_rate": 2.876261897070029e-06,
      "loss": 1.118,
      "num_input_tokens_seen": 1182895,
      "step": 53
    },
    {
      "epoch": 0.006493116094510912,
      "flos": 16244583444480.0,
      "grad_norm": 4.701409416711125,
      "learning_rate": 2.889803337127447e-06,
      "loss": 1.1586,
      "num_input_tokens_seen": 1201215,
      "step": 54
    },
    {
      "epoch": 0.006613358985150003,
      "flos": 16399151247360.0,
      "grad_norm": 5.97430800257395,
      "learning_rate": 2.903096296321516e-06,
      "loss": 1.0827,
      "num_input_tokens_seen": 1219080,
      "step": 55
    },
    {
      "epoch": 0.006733601875789094,
      "flos": 18890040606720.0,
      "grad_norm": 3.8432686356996104,
      "learning_rate": 2.9161497296578907e-06,
      "loss": 1.1604,
      "num_input_tokens_seen": 1238190,
      "step": 56
    },
    {
      "epoch": 0.006853844766428185,
      "flos": 10974709002240.0,
      "grad_norm": 15.644372268230905,
      "learning_rate": 2.928972116604173e-06,
      "loss": 1.0958,
      "num_input_tokens_seen": 1254185,
      "step": 57
    },
    {
      "epoch": 0.006974087657067276,
      "flos": 17242091950080.0,
      "grad_norm": 4.4140661729105295,
      "learning_rate": 2.9415714941751377e-06,
      "loss": 1.2399,
      "num_input_tokens_seen": 1275125,
      "step": 58
    },
    {
      "epoch": 0.007094330547706367,
      "flos": 18340284948480.0,
      "grad_norm": 4.361543943034003,
      "learning_rate": 2.9539554871897396e-06,
      "loss": 1.1662,
      "num_input_tokens_seen": 1295015,
      "step": 59
    },
    {
      "epoch": 0.007214573438345458,
      "flos": 15139338977280.0,
      "grad_norm": 3.657628117557893,
      "learning_rate": 2.9661313359851253e-06,
      "loss": 1.1988,
      "num_input_tokens_seen": 1312620,
      "step": 60
    },
    {
      "epoch": 0.007334816328984549,
      "flos": 17740386324480.0,
      "grad_norm": 3.9050882061363326,
      "learning_rate": 2.978105921839922e-06,
      "loss": 1.1751,
      "num_input_tokens_seen": 1332885,
      "step": 61
    },
    {
      "epoch": 0.00745505921962364,
      "flos": 13121105326080.0,
      "grad_norm": 10.188690677639075,
      "learning_rate": 2.9898857903302893e-06,
      "loss": 0.9684,
      "num_input_tokens_seen": 1351555,
      "step": 62
    },
    {
      "epoch": 0.007575302110262731,
      "flos": 13097919344640.0,
      "grad_norm": 7.20559405131348,
      "learning_rate": 3.001477172817253e-06,
      "loss": 1.1117,
      "num_input_tokens_seen": 1369165,
      "step": 63
    },
    {
      "epoch": 0.007695545000901822,
      "flos": 17766270259200.0,
      "grad_norm": 5.7607156171249,
      "learning_rate": 3.012886006241894e-06,
      "loss": 1.2003,
      "num_input_tokens_seen": 1388270,
      "step": 64
    },
    {
      "epoch": 0.007815787891540913,
      "flos": 15143784468480.0,
      "grad_norm": 3.4218433674782918,
      "learning_rate": 3.0241179513858383e-06,
      "loss": 1.1269,
      "num_input_tokens_seen": 1407300,
      "step": 65
    },
    {
      "epoch": 0.007936030782180003,
      "flos": 15324052254720.0,
      "grad_norm": 3.5213040069240953,
      "learning_rate": 3.035178409737647e-06,
      "loss": 1.1157,
      "num_input_tokens_seen": 1424470,
      "step": 66
    },
    {
      "epoch": 0.008056273672819095,
      "flos": 14278976102400.0,
      "grad_norm": 3.5679960185914,
      "learning_rate": 3.046072539090907e-06,
      "loss": 1.1119,
      "num_input_tokens_seen": 1442155,
      "step": 67
    },
    {
      "epoch": 0.008176516563458186,
      "flos": 12994230312960.0,
      "grad_norm": 4.557556400420907,
      "learning_rate": 3.056805267986779e-06,
      "loss": 1.2798,
      "num_input_tokens_seen": 1459345,
      "step": 68
    },
    {
      "epoch": 0.008296759454097276,
      "flos": 15537592688640.0,
      "grad_norm": 5.01826231832859,
      "learning_rate": 3.0673813091022194e-06,
      "loss": 1.1818,
      "num_input_tokens_seen": 1478285,
      "step": 69
    },
    {
      "epoch": 0.008417002344736368,
      "flos": 45395834572800.0,
      "grad_norm": 1.355521077402018,
      "learning_rate": 3.0778051716749317e-06,
      "loss": 0.8965,
      "num_input_tokens_seen": 1541550,
      "step": 70
    },
    {
      "epoch": 0.008537245235375458,
      "flos": 16324074762240.0,
      "grad_norm": 3.8689646327023413,
      "learning_rate": 3.0880811730470094e-06,
      "loss": 1.1373,
      "num_input_tokens_seen": 1560725,
      "step": 71
    },
    {
      "epoch": 0.008657488126014549,
      "flos": 44372135546880.0,
      "grad_norm": 1.1376504651836685,
      "learning_rate": 3.098213449401257e-06,
      "loss": 0.8437,
      "num_input_tokens_seen": 1627375,
      "step": 72
    },
    {
      "epoch": 0.00877773101665364,
      "flos": 21591803105280.0,
      "grad_norm": 5.670756752605141,
      "learning_rate": 3.1082059657570015e-06,
      "loss": 1.2086,
      "num_input_tokens_seen": 1646330,
      "step": 73
    },
    {
      "epoch": 0.00889797390729273,
      "flos": 16717668372480.0,
      "grad_norm": 5.119676942468075,
      "learning_rate": 3.1180625252858496e-06,
      "loss": 1.1813,
      "num_input_tokens_seen": 1664480,
      "step": 74
    },
    {
      "epoch": 0.009018216797931822,
      "flos": 16402676981760.0,
      "grad_norm": 3.817316108876163,
      "learning_rate": 3.1277867780021663e-06,
      "loss": 1.0269,
      "num_input_tokens_seen": 1680835,
      "step": 75
    },
    {
      "epoch": 0.009138459688570914,
      "flos": 11258328576000.0,
      "grad_norm": 3.910508055204195,
      "learning_rate": 3.1373822288779824e-06,
      "loss": 1.1853,
      "num_input_tokens_seen": 1697135,
      "step": 76
    },
    {
      "epoch": 0.009258702579210003,
      "flos": 19234809569280.0,
      "grad_norm": 5.005545020977361,
      "learning_rate": 3.1468522454274533e-06,
      "loss": 1.0135,
      "num_input_tokens_seen": 1717210,
      "step": 77
    },
    {
      "epoch": 0.009378945469849095,
      "flos": 19152834908160.0,
      "grad_norm": 9.491301318587185,
      "learning_rate": 3.15620006480197e-06,
      "loss": 1.142,
      "num_input_tokens_seen": 1736200,
      "step": 78
    },
    {
      "epoch": 0.009499188360488187,
      "flos": 25470514114560.0,
      "grad_norm": 5.04126836711433,
      "learning_rate": 3.1654288004333087e-06,
      "loss": 0.9802,
      "num_input_tokens_seen": 1754585,
      "step": 79
    },
    {
      "epoch": 0.009619431251127276,
      "flos": 15271947141120.0,
      "grad_norm": 6.710984679512563,
      "learning_rate": 3.1745414482589353e-06,
      "loss": 0.9779,
      "num_input_tokens_seen": 1773515,
      "step": 80
    },
    {
      "epoch": 0.009739674141766368,
      "flos": 12338762158080.0,
      "grad_norm": 3.101874424697789,
      "learning_rate": 3.1835408925606204e-06,
      "loss": 1.0828,
      "num_input_tokens_seen": 1791375,
      "step": 81
    },
    {
      "epoch": 0.00985991703240546,
      "flos": 19601722122240.0,
      "grad_norm": 3.443510315129791,
      "learning_rate": 3.1924299114448214e-06,
      "loss": 1.1114,
      "num_input_tokens_seen": 1811575,
      "step": 82
    },
    {
      "epoch": 0.00998015992304455,
      "flos": 9742068326400.0,
      "grad_norm": 6.395230077402996,
      "learning_rate": 3.2012111819909055e-06,
      "loss": 1.0677,
      "num_input_tokens_seen": 1828625,
      "step": 83
    },
    {
      "epoch": 0.010100402813683641,
      "flos": 14329394995200.0,
      "grad_norm": 6.439527173899391,
      "learning_rate": 3.2098872850910627e-06,
      "loss": 1.1749,
      "num_input_tokens_seen": 1845020,
      "step": 84
    },
    {
      "epoch": 0.010220645704322733,
      "flos": 12181005864960.0,
      "grad_norm": 3.7903315670349373,
      "learning_rate": 3.2184607100038194e-06,
      "loss": 1.1168,
      "num_input_tokens_seen": 1863380,
      "step": 85
    },
    {
      "epoch": 0.010340888594961822,
      "flos": 15248883793920.0,
      "grad_norm": 8.453329425814875,
      "learning_rate": 3.2269338586412414e-06,
      "loss": 1.1482,
      "num_input_tokens_seen": 1880685,
      "step": 86
    },
    {
      "epoch": 0.010461131485600914,
      "flos": 16350203965440.0,
      "grad_norm": 7.149000468781403,
      "learning_rate": 3.2353090496083106e-06,
      "loss": 1.1947,
      "num_input_tokens_seen": 1898240,
      "step": 87
    },
    {
      "epoch": 0.010581374376240005,
      "flos": 23927940710400.0,
      "grad_norm": 7.844729662504834,
      "learning_rate": 3.2435885220114572e-06,
      "loss": 1.0421,
      "num_input_tokens_seen": 1919310,
      "step": 88
    },
    {
      "epoch": 0.010701617266879095,
      "flos": 15459327713280.0,
      "grad_norm": 7.2507344893304175,
      "learning_rate": 3.2517744390519113e-06,
      "loss": 1.1598,
      "num_input_tokens_seen": 1937895,
      "step": 89
    },
    {
      "epoch": 0.010821860157518187,
      "flos": 13516415815680.0,
      "grad_norm": 3.213234481267124,
      "learning_rate": 3.259868891418298e-06,
      "loss": 0.9715,
      "num_input_tokens_seen": 1955380,
      "step": 90
    },
    {
      "epoch": 0.010942103048157278,
      "flos": 18107053608960.0,
      "grad_norm": 4.25652799216635,
      "learning_rate": 3.2678739004917757e-06,
      "loss": 1.0606,
      "num_input_tokens_seen": 1974835,
      "step": 91
    },
    {
      "epoch": 0.011062345938796368,
      "flos": 19578321530880.0,
      "grad_norm": 2.471888514890509,
      "learning_rate": 3.275791421376029e-06,
      "loss": 1.1463,
      "num_input_tokens_seen": 1998000,
      "step": 92
    },
    {
      "epoch": 0.01118258882943546,
      "flos": 11368670515200.0,
      "grad_norm": 19.222605050353476,
      "learning_rate": 3.2836233457634622e-06,
      "loss": 1.1741,
      "num_input_tokens_seen": 2015855,
      "step": 93
    },
    {
      "epoch": 0.011302831720074551,
      "flos": 14672753664000.0,
      "grad_norm": 3.306057837514007,
      "learning_rate": 3.2913715046481135e-06,
      "loss": 1.0657,
      "num_input_tokens_seen": 2035320,
      "step": 94
    },
    {
      "epoch": 0.011423074610713641,
      "flos": 9212402135040.0,
      "grad_norm": 4.0144787054016176,
      "learning_rate": 3.299037670895023e-06,
      "loss": 1.1023,
      "num_input_tokens_seen": 2051655,
      "step": 95
    },
    {
      "epoch": 0.011543317501352733,
      "flos": 21620047749120.0,
      "grad_norm": 11.326920733402403,
      "learning_rate": 3.3066235616750667e-06,
      "loss": 1.0331,
      "num_input_tokens_seen": 2072610,
      "step": 96
    },
    {
      "epoch": 0.011663560391991824,
      "flos": 10974525050880.0,
      "grad_norm": 3.6695561005934807,
      "learning_rate": 3.3141308407736276e-06,
      "loss": 1.1372,
      "num_input_tokens_seen": 2088965,
      "step": 97
    },
    {
      "epoch": 0.011783803282630914,
      "flos": 14121863639040.0,
      "grad_norm": 3.985380038491749,
      "learning_rate": 3.321561120780869e-06,
      "loss": 1.0821,
      "num_input_tokens_seen": 2107395,
      "step": 98
    },
    {
      "epoch": 0.011904046173270006,
      "flos": 15873593303040.0,
      "grad_norm": 3.392675245758302,
      "learning_rate": 3.3289159651708192e-06,
      "loss": 1.231,
      "num_input_tokens_seen": 2124690,
      "step": 99
    },
    {
      "epoch": 0.012024289063909096,
      "flos": 13545518899200.0,
      "grad_norm": 6.804161638334729,
      "learning_rate": 3.3361968902759768e-06,
      "loss": 1.198,
      "num_input_tokens_seen": 2144090,
      "step": 100
    },
    {
      "epoch": 0.012144531954548187,
      "flos": 10607459205120.0,
      "grad_norm": 3.9576173369025414,
      "learning_rate": 3.343405367163663e-06,
      "loss": 1.1696,
      "num_input_tokens_seen": 2160740,
      "step": 101
    },
    {
      "epoch": 0.012264774845187279,
      "flos": 10686429327360.0,
      "grad_norm": 4.387653037333899,
      "learning_rate": 3.350542823419951e-06,
      "loss": 1.0365,
      "num_input_tokens_seen": 2177060,
      "step": 102
    },
    {
      "epoch": 0.012385017735826368,
      "flos": 9842875453440.0,
      "grad_norm": 7.998329703485241,
      "learning_rate": 3.3576106448465615e-06,
      "loss": 1.0922,
      "num_input_tokens_seen": 2190160,
      "step": 103
    },
    {
      "epoch": 0.01250526062646546,
      "flos": 16796117299200.0,
      "grad_norm": 7.476088234778229,
      "learning_rate": 3.3646101770757797e-06,
      "loss": 1.1024,
      "num_input_tokens_seen": 2208670,
      "step": 104
    },
    {
      "epoch": 0.012625503517104552,
      "flos": 24713441710080.0,
      "grad_norm": 4.9308786540065155,
      "learning_rate": 3.371542727108104e-06,
      "loss": 1.0668,
      "num_input_tokens_seen": 2230565,
      "step": 105
    },
    {
      "epoch": 0.012745746407743641,
      "flos": 12626765905920.0,
      "grad_norm": 4.734334108283079,
      "learning_rate": 3.3784095647770114e-06,
      "loss": 1.1113,
      "num_input_tokens_seen": 2248930,
      "step": 106
    },
    {
      "epoch": 0.012865989298382733,
      "flos": 14619820769280.0,
      "grad_norm": 4.3976093820367375,
      "learning_rate": 3.3852119241449547e-06,
      "loss": 1.1014,
      "num_input_tokens_seen": 2267770,
      "step": 107
    },
    {
      "epoch": 0.012986232189021825,
      "flos": 17029839175680.0,
      "grad_norm": 3.2298068160637405,
      "learning_rate": 3.3919510048344295e-06,
      "loss": 1.175,
      "num_input_tokens_seen": 2285500,
      "step": 108
    },
    {
      "epoch": 0.013106475079660914,
      "flos": 16868557148160.0,
      "grad_norm": 19.342199901290556,
      "learning_rate": 3.3986279732976907e-06,
      "loss": 1.0713,
      "num_input_tokens_seen": 2303695,
      "step": 109
    },
    {
      "epoch": 0.013226717970300006,
      "flos": 14984004710400.0,
      "grad_norm": 3.0406913412919456,
      "learning_rate": 3.4052439640284983e-06,
      "loss": 1.1664,
      "num_input_tokens_seen": 2322330,
      "step": 110
    },
    {
      "epoch": 0.013346960860939098,
      "flos": 17687760015360.0,
      "grad_norm": 4.369454711483804,
      "learning_rate": 3.4118000807190217e-06,
      "loss": 1.0324,
      "num_input_tokens_seen": 2342930,
      "step": 111
    },
    {
      "epoch": 0.013467203751578187,
      "flos": 20072078438400.0,
      "grad_norm": 4.718880354741807,
      "learning_rate": 3.4182973973648723e-06,
      "loss": 0.9837,
      "num_input_tokens_seen": 2363860,
      "step": 112
    },
    {
      "epoch": 0.013587446642217279,
      "flos": 13413217320960.0,
      "grad_norm": 7.427785732303387,
      "learning_rate": 3.424736959321014e-06,
      "loss": 1.1507,
      "num_input_tokens_seen": 2381385,
      "step": 113
    },
    {
      "epoch": 0.01370768953285637,
      "flos": 17058758307840.0,
      "grad_norm": 6.584210623059426,
      "learning_rate": 3.431119784311155e-06,
      "loss": 1.1024,
      "num_input_tokens_seen": 2400780,
      "step": 114
    },
    {
      "epoch": 0.01382793242349546,
      "flos": 27992652656640.0,
      "grad_norm": 3.4547652409647496,
      "learning_rate": 3.43744686339307e-06,
      "loss": 0.9965,
      "num_input_tokens_seen": 2422820,
      "step": 115
    },
    {
      "epoch": 0.013948175314134552,
      "flos": 29538261258240.0,
      "grad_norm": 5.683284328991482,
      "learning_rate": 3.44371916188212e-06,
      "loss": 1.1198,
      "num_input_tokens_seen": 2443295,
      "step": 116
    },
    {
      "epoch": 0.014068418204773643,
      "flos": 15955169402880.0,
      "grad_norm": 6.94459899052207,
      "learning_rate": 3.449937620235143e-06,
      "loss": 1.0867,
      "num_input_tokens_seen": 2463610,
      "step": 117
    },
    {
      "epoch": 0.014188661095412733,
      "flos": 16926763315200.0,
      "grad_norm": 2.8496982280848444,
      "learning_rate": 3.456103154896722e-06,
      "loss": 1.0943,
      "num_input_tokens_seen": 2484605,
      "step": 118
    },
    {
      "epoch": 0.014308903986051825,
      "flos": 16822767697920.0,
      "grad_norm": 4.660840530230807,
      "learning_rate": 3.462216659109757e-06,
      "loss": 1.1425,
      "num_input_tokens_seen": 2504505,
      "step": 119
    },
    {
      "epoch": 0.014429146876690916,
      "flos": 14672232468480.0,
      "grad_norm": 8.930103148818336,
      "learning_rate": 3.4682790036921077e-06,
      "loss": 1.0606,
      "num_input_tokens_seen": 2522885,
      "step": 120
    },
    {
      "epoch": 0.014549389767330006,
      "flos": 14356750540800.0,
      "grad_norm": 9.79340696628728,
      "learning_rate": 3.4742910377810193e-06,
      "loss": 1.0518,
      "num_input_tokens_seen": 2540065,
      "step": 121
    },
    {
      "epoch": 0.014669632657969098,
      "flos": 12757749166080.0,
      "grad_norm": 4.533471135307687,
      "learning_rate": 3.4802535895469042e-06,
      "loss": 1.0993,
      "num_input_tokens_seen": 2558535,
      "step": 122
    },
    {
      "epoch": 0.01478987554860819,
      "flos": 16161842319360.0,
      "grad_norm": 5.70924797062947,
      "learning_rate": 3.4861674668779934e-06,
      "loss": 1.1147,
      "num_input_tokens_seen": 2576485,
      "step": 123
    },
    {
      "epoch": 0.01491011843924728,
      "flos": 12178399887360.0,
      "grad_norm": 25.516020884482433,
      "learning_rate": 3.492033458037272e-06,
      "loss": 1.0628,
      "num_input_tokens_seen": 2594775,
      "step": 124
    },
    {
      "epoch": 0.01503036132988637,
      "flos": 12520102993920.0,
      "grad_norm": 5.669061304395821,
      "learning_rate": 3.497852332293018e-06,
      "loss": 1.0851,
      "num_input_tokens_seen": 2610070,
      "step": 125
    },
    {
      "epoch": 0.015150604220525462,
      "flos": 13385708482560.0,
      "grad_norm": 4.281382092296745,
      "learning_rate": 3.5036248405242356e-06,
      "loss": 1.1922,
      "num_input_tokens_seen": 2628545,
      "step": 126
    },
    {
      "epoch": 0.015270847111164552,
      "flos": 28148876021760.0,
      "grad_norm": 4.378952856436854,
      "learning_rate": 3.509351715802146e-06,
      "loss": 1.0453,
      "num_input_tokens_seen": 2649150,
      "step": 127
    },
    {
      "epoch": 0.015391090001803644,
      "flos": 31269656186880.0,
      "grad_norm": 4.609453207096066,
      "learning_rate": 3.5150336739488763e-06,
      "loss": 1.0016,
      "num_input_tokens_seen": 2671155,
      "step": 128
    },
    {
      "epoch": 0.015511332892442733,
      "flos": 13413309296640.0,
      "grad_norm": 3.9876220636609,
      "learning_rate": 3.5206714140744143e-06,
      "loss": 1.0452,
      "num_input_tokens_seen": 2690930,
      "step": 129
    },
    {
      "epoch": 0.015631575783081827,
      "flos": 17451554795520.0,
      "grad_norm": 17.430499021616626,
      "learning_rate": 3.5262656190928208e-06,
      "loss": 1.0793,
      "num_input_tokens_seen": 2708950,
      "step": 130
    },
    {
      "epoch": 0.015751818673720917,
      "flos": 44616963932160.0,
      "grad_norm": 1.1675200860462362,
      "learning_rate": 3.5318169562186737e-06,
      "loss": 0.9354,
      "num_input_tokens_seen": 2777515,
      "step": 131
    },
    {
      "epoch": 0.015872061564360006,
      "flos": 16979573575680.0,
      "grad_norm": 5.8086932442512875,
      "learning_rate": 3.5373260774446292e-06,
      "loss": 1.0413,
      "num_input_tokens_seen": 2797685,
      "step": 132
    },
    {
      "epoch": 0.0159923044549991,
      "flos": 16612599705600.0,
      "grad_norm": 3.256676931088481,
      "learning_rate": 3.542793620000961e-06,
      "loss": 1.1215,
      "num_input_tokens_seen": 2816880,
      "step": 133
    },
    {
      "epoch": 0.01611254734563819,
      "flos": 12652925767680.0,
      "grad_norm": 12.889871545215199,
      "learning_rate": 3.5482202067978894e-06,
      "loss": 1.0793,
      "num_input_tokens_seen": 2833810,
      "step": 134
    },
    {
      "epoch": 0.01623279023627728,
      "flos": 14877648384000.0,
      "grad_norm": 5.299359394441471,
      "learning_rate": 3.553606446851471e-06,
      "loss": 0.9697,
      "num_input_tokens_seen": 2850270,
      "step": 135
    },
    {
      "epoch": 0.016353033126916373,
      "flos": 11132312002560.0,
      "grad_norm": 10.908013755909968,
      "learning_rate": 3.5589529356937613e-06,
      "loss": 1.0531,
      "num_input_tokens_seen": 2868385,
      "step": 136
    },
    {
      "epoch": 0.016473276017555463,
      "flos": 13334983004160.0,
      "grad_norm": 5.083931663612161,
      "learning_rate": 3.5642602557679627e-06,
      "loss": 0.9864,
      "num_input_tokens_seen": 2886555,
      "step": 137
    },
    {
      "epoch": 0.016593518908194552,
      "flos": 17320264949760.0,
      "grad_norm": 4.849305934991879,
      "learning_rate": 3.569528976809202e-06,
      "loss": 1.0639,
      "num_input_tokens_seen": 2903490,
      "step": 138
    },
    {
      "epoch": 0.016713761798833646,
      "flos": 15878253404160.0,
      "grad_norm": 3.67241960098625,
      "learning_rate": 3.5747596562115522e-06,
      "loss": 1.0993,
      "num_input_tokens_seen": 2923825,
      "step": 139
    },
    {
      "epoch": 0.016834004689472735,
      "flos": 12626704588800.0,
      "grad_norm": 5.305999908125179,
      "learning_rate": 3.5799528393819138e-06,
      "loss": 1.1261,
      "num_input_tokens_seen": 2942625,
      "step": 140
    },
    {
      "epoch": 0.016954247580111825,
      "flos": 14565170995200.0,
      "grad_norm": 3.777029409021885,
      "learning_rate": 3.585109060081286e-06,
      "loss": 1.0914,
      "num_input_tokens_seen": 2962145,
      "step": 141
    },
    {
      "epoch": 0.017074490470750915,
      "flos": 15693110906880.0,
      "grad_norm": 3.3658665590723356,
      "learning_rate": 3.590228840753992e-06,
      "loss": 0.9994,
      "num_input_tokens_seen": 2982295,
      "step": 142
    },
    {
      "epoch": 0.01719473336139001,
      "flos": 11315400376320.0,
      "grad_norm": 4.673212886859862,
      "learning_rate": 3.5953126928453423e-06,
      "loss": 1.0844,
      "num_input_tokens_seen": 2999565,
      "step": 143
    },
    {
      "epoch": 0.017314976252029098,
      "flos": 15983107461120.0,
      "grad_norm": 6.306333948489649,
      "learning_rate": 3.600361117108239e-06,
      "loss": 1.0206,
      "num_input_tokens_seen": 3019085,
      "step": 144
    },
    {
      "epoch": 0.017435219142668188,
      "flos": 15642508062720.0,
      "grad_norm": 6.535494506618831,
      "learning_rate": 3.6053746038991616e-06,
      "loss": 1.1783,
      "num_input_tokens_seen": 3037890,
      "step": 145
    },
    {
      "epoch": 0.01755546203330728,
      "flos": 51738884628480.0,
      "grad_norm": 1.1064904549691634,
      "learning_rate": 3.6103536334639843e-06,
      "loss": 0.8367,
      "num_input_tokens_seen": 3090875,
      "step": 146
    },
    {
      "epoch": 0.01767570492394637,
      "flos": 18027654266880.0,
      "grad_norm": 4.188813871683884,
      "learning_rate": 3.615298676214041e-06,
      "loss": 1.068,
      "num_input_tokens_seen": 3110875,
      "step": 147
    },
    {
      "epoch": 0.01779594781458546,
      "flos": 14514997370880.0,
      "grad_norm": 4.077367485695068,
      "learning_rate": 3.6202101929928317e-06,
      "loss": 1.1066,
      "num_input_tokens_seen": 3129185,
      "step": 148
    },
    {
      "epoch": 0.017916190705224554,
      "flos": 11499745751040.0,
      "grad_norm": 3.627436509631656,
      "learning_rate": 3.6250886353337413e-06,
      "loss": 1.1003,
      "num_input_tokens_seen": 3146435,
      "step": 149
    },
    {
      "epoch": 0.018036433595863644,
      "flos": 16585673379840.0,
      "grad_norm": 4.907891843118591,
      "learning_rate": 3.6299344457091488e-06,
      "loss": 1.0864,
      "num_input_tokens_seen": 3167015,
      "step": 150
    },
    {
      "epoch": 0.018156676486502734,
      "flos": 13177349345280.0,
      "grad_norm": 3.985133319572642,
      "learning_rate": 3.634748057771256e-06,
      "loss": 1.1362,
      "num_input_tokens_seen": 3182675,
      "step": 151
    },
    {
      "epoch": 0.018276919377141827,
      "flos": 18108187975680.0,
      "grad_norm": 5.104486862830827,
      "learning_rate": 3.639529896584965e-06,
      "loss": 1.0732,
      "num_input_tokens_seen": 3203770,
      "step": 152
    },
    {
      "epoch": 0.018397162267780917,
      "flos": 14226104524800.0,
      "grad_norm": 5.551058542840718,
      "learning_rate": 3.6442803788531233e-06,
      "loss": 1.0914,
      "num_input_tokens_seen": 3221450,
      "step": 153
    },
    {
      "epoch": 0.018517405158420007,
      "flos": 19629384253440.0,
      "grad_norm": 4.836633585933513,
      "learning_rate": 3.6489999131344357e-06,
      "loss": 1.1669,
      "num_input_tokens_seen": 3243945,
      "step": 154
    },
    {
      "epoch": 0.0186376480490591,
      "flos": 13727871467520.0,
      "grad_norm": 4.7174106743486925,
      "learning_rate": 3.653688900054313e-06,
      "loss": 1.1227,
      "num_input_tokens_seen": 3262195,
      "step": 155
    },
    {
      "epoch": 0.01875789093969819,
      "flos": 18998543032320.0,
      "grad_norm": 3.6291248924984054,
      "learning_rate": 3.6583477325089526e-06,
      "loss": 0.9771,
      "num_input_tokens_seen": 3282455,
      "step": 156
    },
    {
      "epoch": 0.01887813383033728,
      "flos": 17320571535360.0,
      "grad_norm": 5.6213485734239566,
      "learning_rate": 3.6629767958628916e-06,
      "loss": 1.2531,
      "num_input_tokens_seen": 3299550,
      "step": 157
    },
    {
      "epoch": 0.018998376720976373,
      "flos": 10345400709120.0,
      "grad_norm": 4.5274471768141895,
      "learning_rate": 3.667576468140291e-06,
      "loss": 1.0652,
      "num_input_tokens_seen": 3317085,
      "step": 158
    },
    {
      "epoch": 0.019118619611615463,
      "flos": 20878343393280.0,
      "grad_norm": 6.95257717074651,
      "learning_rate": 3.672147120210184e-06,
      "loss": 1.0978,
      "num_input_tokens_seen": 3333405,
      "step": 159
    },
    {
      "epoch": 0.019238862502254553,
      "flos": 14829406248960.0,
      "grad_norm": 27.305310034033443,
      "learning_rate": 3.6766891159659177e-06,
      "loss": 1.0842,
      "num_input_tokens_seen": 3351535,
      "step": 160
    },
    {
      "epoch": 0.019359105392893646,
      "flos": 15169974988800.0,
      "grad_norm": 5.494207452990743,
      "learning_rate": 3.6812028124990075e-06,
      "loss": 1.0874,
      "num_input_tokens_seen": 3368525,
      "step": 161
    },
    {
      "epoch": 0.019479348283532736,
      "flos": 11520540364800.0,
      "grad_norm": 4.334901428157935,
      "learning_rate": 3.6856885602676016e-06,
      "loss": 1.031,
      "num_input_tokens_seen": 3384280,
      "step": 162
    },
    {
      "epoch": 0.019599591174171826,
      "flos": 15695134371840.0,
      "grad_norm": 3.8704982572846554,
      "learning_rate": 3.6901467032597733e-06,
      "loss": 1.1589,
      "num_input_tokens_seen": 3402485,
      "step": 163
    },
    {
      "epoch": 0.01971983406481092,
      "flos": 13911603671040.0,
      "grad_norm": 10.265804480808292,
      "learning_rate": 3.694577579151804e-06,
      "loss": 1.0809,
      "num_input_tokens_seen": 3420615,
      "step": 164
    },
    {
      "epoch": 0.01984007695545001,
      "flos": 13544905728000.0,
      "grad_norm": 8.025733091911146,
      "learning_rate": 3.6989815194616703e-06,
      "loss": 0.959,
      "num_input_tokens_seen": 3437530,
      "step": 165
    },
    {
      "epoch": 0.0199603198460891,
      "flos": 14801621483520.0,
      "grad_norm": 6.234009696181892,
      "learning_rate": 3.703358849697888e-06,
      "loss": 1.0148,
      "num_input_tokens_seen": 3457160,
      "step": 166
    },
    {
      "epoch": 0.020080562736728192,
      "flos": 15536458321920.0,
      "grad_norm": 3.485175344470011,
      "learning_rate": 3.7077098895038803e-06,
      "loss": 1.0452,
      "num_input_tokens_seen": 3476250,
      "step": 167
    },
    {
      "epoch": 0.020200805627367282,
      "flos": 15405750988800.0,
      "grad_norm": 5.533086094828824,
      "learning_rate": 3.712034952798045e-06,
      "loss": 1.1818,
      "num_input_tokens_seen": 3494085,
      "step": 168
    },
    {
      "epoch": 0.02032104851800637,
      "flos": 23924690903040.0,
      "grad_norm": 6.093974730431421,
      "learning_rate": 3.7163343479096656e-06,
      "loss": 1.0406,
      "num_input_tokens_seen": 3515380,
      "step": 169
    },
    {
      "epoch": 0.020441291408645465,
      "flos": 22591396392960.0,
      "grad_norm": 7.357607230293925,
      "learning_rate": 3.720608377710802e-06,
      "loss": 1.0301,
      "num_input_tokens_seen": 3535190,
      "step": 170
    },
    {
      "epoch": 0.020561534299284555,
      "flos": 14829375590400.0,
      "grad_norm": 10.005805047249057,
      "learning_rate": 3.7248573397443277e-06,
      "loss": 1.0765,
      "num_input_tokens_seen": 3553835,
      "step": 171
    },
    {
      "epoch": 0.020681777189923645,
      "flos": 14908437688320.0,
      "grad_norm": 7.635876512059495,
      "learning_rate": 3.729081526348224e-06,
      "loss": 1.1931,
      "num_input_tokens_seen": 3572085,
      "step": 172
    },
    {
      "epoch": 0.020802020080562738,
      "flos": 20126758871040.0,
      "grad_norm": 6.845749375735345,
      "learning_rate": 3.7332812247762777e-06,
      "loss": 1.0509,
      "num_input_tokens_seen": 3593105,
      "step": 173
    },
    {
      "epoch": 0.020922262971201828,
      "flos": 13962881003520.0,
      "grad_norm": 4.4420546676521395,
      "learning_rate": 3.737456717315293e-06,
      "loss": 1.162,
      "num_input_tokens_seen": 3611790,
      "step": 174
    },
    {
      "epoch": 0.021042505861840918,
      "flos": 11077233008640.0,
      "grad_norm": 4.74888517447944,
      "learning_rate": 3.7416082813989552e-06,
      "loss": 1.1202,
      "num_input_tokens_seen": 3628505,
      "step": 175
    },
    {
      "epoch": 0.02116274875248001,
      "flos": 15012862525440.0,
      "grad_norm": 7.117192591844364,
      "learning_rate": 3.745736189718439e-06,
      "loss": 1.1173,
      "num_input_tokens_seen": 3647480,
      "step": 176
    },
    {
      "epoch": 0.0212829916431191,
      "flos": 17581005127680.0,
      "grad_norm": 7.356920484973098,
      "learning_rate": 3.749840710329894e-06,
      "loss": 0.9213,
      "num_input_tokens_seen": 3667905,
      "step": 177
    },
    {
      "epoch": 0.02140323453375819,
      "flos": 11780115517440.0,
      "grad_norm": 10.81496511593932,
      "learning_rate": 3.7539221067588938e-06,
      "loss": 1.1928,
      "num_input_tokens_seen": 3681600,
      "step": 178
    },
    {
      "epoch": 0.021523477424397284,
      "flos": 14406770872320.0,
      "grad_norm": 5.958689574504826,
      "learning_rate": 3.757980638101964e-06,
      "loss": 1.1336,
      "num_input_tokens_seen": 3694815,
      "step": 179
    },
    {
      "epoch": 0.021643720315036374,
      "flos": 18578942853120.0,
      "grad_norm": 11.441036824011993,
      "learning_rate": 3.7620165591252806e-06,
      "loss": 1.1095,
      "num_input_tokens_seen": 3712635,
      "step": 180
    },
    {
      "epoch": 0.021763963205675464,
      "flos": 17632527728640.0,
      "grad_norm": 7.0829315370781005,
      "learning_rate": 3.766030120360636e-06,
      "loss": 1.1458,
      "num_input_tokens_seen": 3731985,
      "step": 181
    },
    {
      "epoch": 0.021884206096314557,
      "flos": 18370399764480.0,
      "grad_norm": 6.960267245544134,
      "learning_rate": 3.7700215681987578e-06,
      "loss": 1.0917,
      "num_input_tokens_seen": 3751475,
      "step": 182
    },
    {
      "epoch": 0.022004448986953647,
      "flos": 14251405946880.0,
      "grad_norm": 6.673297382045757,
      "learning_rate": 3.7739911449800767e-06,
      "loss": 1.0327,
      "num_input_tokens_seen": 3771250,
      "step": 183
    },
    {
      "epoch": 0.022124691877592736,
      "flos": 14536895692800.0,
      "grad_norm": 11.753616101229362,
      "learning_rate": 3.7779390890830114e-06,
      "loss": 1.011,
      "num_input_tokens_seen": 3789620,
      "step": 184
    },
    {
      "epoch": 0.02224493476823183,
      "flos": 16769957437440.0,
      "grad_norm": 12.999032204623386,
      "learning_rate": 3.7818656350098723e-06,
      "loss": 1.067,
      "num_input_tokens_seen": 3810290,
      "step": 185
    },
    {
      "epoch": 0.02236517765887092,
      "flos": 11971389726720.0,
      "grad_norm": 17.016260785011365,
      "learning_rate": 3.7857710134704447e-06,
      "loss": 0.9792,
      "num_input_tokens_seen": 3828945,
      "step": 186
    },
    {
      "epoch": 0.02248542054951001,
      "flos": 31087058350080.0,
      "grad_norm": 4.756823808141228,
      "learning_rate": 3.7896554514633234e-06,
      "loss": 0.9971,
      "num_input_tokens_seen": 3852950,
      "step": 187
    },
    {
      "epoch": 0.022605663440149103,
      "flos": 16612385095680.0,
      "grad_norm": 5.178917644935326,
      "learning_rate": 3.7935191723550955e-06,
      "loss": 1.0597,
      "num_input_tokens_seen": 3871625,
      "step": 188
    },
    {
      "epoch": 0.022725906330788193,
      "flos": 20674153820160.0,
      "grad_norm": 17.15101062699641,
      "learning_rate": 3.797362395957408e-06,
      "loss": 1.1002,
      "num_input_tokens_seen": 3891910,
      "step": 189
    },
    {
      "epoch": 0.022846149221427282,
      "flos": 17423095541760.0,
      "grad_norm": 10.687642414415723,
      "learning_rate": 3.8011853386020055e-06,
      "loss": 0.9941,
      "num_input_tokens_seen": 3912535,
      "step": 190
    },
    {
      "epoch": 0.022966392112066376,
      "flos": 10975015587840.0,
      "grad_norm": 8.39280606981872,
      "learning_rate": 3.804988213213804e-06,
      "loss": 1.0935,
      "num_input_tokens_seen": 3930495,
      "step": 191
    },
    {
      "epoch": 0.023086635002705466,
      "flos": 39817353830400.0,
      "grad_norm": 1.1474144459497757,
      "learning_rate": 3.808771229382049e-06,
      "loss": 0.8658,
      "num_input_tokens_seen": 3989680,
      "step": 192
    },
    {
      "epoch": 0.023206877893344555,
      "flos": 13698124554240.0,
      "grad_norm": 7.21415848464681,
      "learning_rate": 3.8125345934296324e-06,
      "loss": 1.0703,
      "num_input_tokens_seen": 4007710,
      "step": 193
    },
    {
      "epoch": 0.02332712078398365,
      "flos": 16400224296960.0,
      "grad_norm": 12.372815488414574,
      "learning_rate": 3.81627850848061e-06,
      "loss": 1.0766,
      "num_input_tokens_seen": 4028025,
      "step": 194
    },
    {
      "epoch": 0.02344736367462274,
      "flos": 17373013893120.0,
      "grad_norm": 11.306181940028878,
      "learning_rate": 3.820003174525994e-06,
      "loss": 1.0518,
      "num_input_tokens_seen": 4047170,
      "step": 195
    },
    {
      "epoch": 0.02356760656526183,
      "flos": 15327486013440.0,
      "grad_norm": 9.985604032948233,
      "learning_rate": 3.823708788487851e-06,
      "loss": 1.0497,
      "num_input_tokens_seen": 4063890,
      "step": 196
    },
    {
      "epoch": 0.02368784945590092,
      "flos": 17924884992000.0,
      "grad_norm": 3.8539717066487933,
      "learning_rate": 3.827395544281781e-06,
      "loss": 1.0553,
      "num_input_tokens_seen": 4085950,
      "step": 197
    },
    {
      "epoch": 0.02380809234654001,
      "flos": 19626992885760.0,
      "grad_norm": 8.227902034371036,
      "learning_rate": 3.831063632877802e-06,
      "loss": 1.0079,
      "num_input_tokens_seen": 4105990,
      "step": 198
    },
    {
      "epoch": 0.0239283352371791,
      "flos": 12941481369600.0,
      "grad_norm": 31.455632041168318,
      "learning_rate": 3.834713242359712e-06,
      "loss": 0.9655,
      "num_input_tokens_seen": 4123540,
      "step": 199
    },
    {
      "epoch": 0.02404857812781819,
      "flos": 15194448629760.0,
      "grad_norm": 18.463290522651377,
      "learning_rate": 3.838344557982959e-06,
      "loss": 1.0867,
      "num_input_tokens_seen": 4144640,
      "step": 200
    },
    {
      "epoch": 0.024168821018457284,
      "flos": 11787596206080.0,
      "grad_norm": 33.563325612517524,
      "learning_rate": 3.841957762231063e-06,
      "loss": 1.0637,
      "num_input_tokens_seen": 4161890,
      "step": 201
    },
    {
      "epoch": 0.024289063909096374,
      "flos": 16219496632320.0,
      "grad_norm": 7.883544162030647,
      "learning_rate": 3.8455530348706454e-06,
      "loss": 1.0915,
      "num_input_tokens_seen": 4181210,
      "step": 202
    },
    {
      "epoch": 0.024409306799735464,
      "flos": 12574047621120.0,
      "grad_norm": 5.273376916608015,
      "learning_rate": 3.849130553005099e-06,
      "loss": 0.9926,
      "num_input_tokens_seen": 4199145,
      "step": 203
    },
    {
      "epoch": 0.024529549690374557,
      "flos": 15353645875200.0,
      "grad_norm": 8.803417097235062,
      "learning_rate": 3.852690491126933e-06,
      "loss": 1.0517,
      "num_input_tokens_seen": 4218915,
      "step": 204
    },
    {
      "epoch": 0.024649792581013647,
      "flos": 18181302312960.0,
      "grad_norm": 25.78157196768653,
      "learning_rate": 3.856233021168845e-06,
      "loss": 1.1174,
      "num_input_tokens_seen": 4238330,
      "step": 205
    },
    {
      "epoch": 0.024770035471652737,
      "flos": 24609783336960.0,
      "grad_norm": 6.169669614382505,
      "learning_rate": 3.859758312553544e-06,
      "loss": 1.1339,
      "num_input_tokens_seen": 4260270,
      "step": 206
    },
    {
      "epoch": 0.02489027836229183,
      "flos": 15273847971840.0,
      "grad_norm": 11.017639581961637,
      "learning_rate": 3.8632665322423735e-06,
      "loss": 1.1367,
      "num_input_tokens_seen": 4279645,
      "step": 207
    },
    {
      "epoch": 0.02501052125293092,
      "flos": 16505200988160.0,
      "grad_norm": 12.406909838600587,
      "learning_rate": 3.866757844782762e-06,
      "loss": 1.0643,
      "num_input_tokens_seen": 4299870,
      "step": 208
    },
    {
      "epoch": 0.02513076414357001,
      "flos": 18782764523520.0,
      "grad_norm": 6.131466935390352,
      "learning_rate": 3.870232412354527e-06,
      "loss": 1.1326,
      "num_input_tokens_seen": 4316435,
      "step": 209
    },
    {
      "epoch": 0.025251007034209103,
      "flos": 9585814302720.0,
      "grad_norm": 11.398246068046364,
      "learning_rate": 3.873690394815086e-06,
      "loss": 1.1401,
      "num_input_tokens_seen": 4332735,
      "step": 210
    },
    {
      "epoch": 0.025371249924848193,
      "flos": 10634140262400.0,
      "grad_norm": 16.835931155741264,
      "learning_rate": 3.877131949743587e-06,
      "loss": 1.1267,
      "num_input_tokens_seen": 4349320,
      "step": 211
    },
    {
      "epoch": 0.025491492815487283,
      "flos": 18183172485120.0,
      "grad_norm": 5.125286559490555,
      "learning_rate": 3.880557232483993e-06,
      "loss": 0.9971,
      "num_input_tokens_seen": 4368480,
      "step": 212
    },
    {
      "epoch": 0.025611735706126376,
      "flos": 14645367459840.0,
      "grad_norm": 10.709726166345499,
      "learning_rate": 3.883966396187164e-06,
      "loss": 1.0929,
      "num_input_tokens_seen": 4387470,
      "step": 213
    },
    {
      "epoch": 0.025731978596765466,
      "flos": 13518194012160.0,
      "grad_norm": 10.64411588925848,
      "learning_rate": 3.887359591851937e-06,
      "loss": 1.1181,
      "num_input_tokens_seen": 4404795,
      "step": 214
    },
    {
      "epoch": 0.025852221487404556,
      "flos": 15747239485440.0,
      "grad_norm": 10.457083415890615,
      "learning_rate": 3.890736968365265e-06,
      "loss": 1.1461,
      "num_input_tokens_seen": 4424830,
      "step": 215
    },
    {
      "epoch": 0.02597246437804365,
      "flos": 18893719633920.0,
      "grad_norm": 5.692756319905347,
      "learning_rate": 3.894098672541412e-06,
      "loss": 1.0648,
      "num_input_tokens_seen": 4445455,
      "step": 216
    },
    {
      "epoch": 0.02609270726868274,
      "flos": 23299030978560.0,
      "grad_norm": 5.931907509636412,
      "learning_rate": 3.89744484916025e-06,
      "loss": 0.9643,
      "num_input_tokens_seen": 4466285,
      "step": 217
    },
    {
      "epoch": 0.02621295015932183,
      "flos": 18678370344960.0,
      "grad_norm": 7.328224425464907,
      "learning_rate": 3.900775641004673e-06,
      "loss": 1.0874,
      "num_input_tokens_seen": 4485320,
      "step": 218
    },
    {
      "epoch": 0.026333193049960922,
      "flos": 30664944168960.0,
      "grad_norm": 28.53371108806268,
      "learning_rate": 3.904091188897156e-06,
      "loss": 0.9443,
      "num_input_tokens_seen": 4504175,
      "step": 219
    },
    {
      "epoch": 0.026453435940600012,
      "flos": 12728247521280.0,
      "grad_norm": 5.9981953768046194,
      "learning_rate": 3.90739163173548e-06,
      "loss": 1.0222,
      "num_input_tokens_seen": 4521730,
      "step": 220
    },
    {
      "epoch": 0.026573678831239102,
      "flos": 13461490114560.0,
      "grad_norm": 5.623743552708539,
      "learning_rate": 3.910677106527646e-06,
      "loss": 1.0886,
      "num_input_tokens_seen": 4538640,
      "step": 221
    },
    {
      "epoch": 0.026693921721878195,
      "flos": 20833718968320.0,
      "grad_norm": 7.658423041426042,
      "learning_rate": 3.913947748426004e-06,
      "loss": 1.0519,
      "num_input_tokens_seen": 4555180,
      "step": 222
    },
    {
      "epoch": 0.026814164612517285,
      "flos": 9970915491840.0,
      "grad_norm": 15.136383060793815,
      "learning_rate": 3.9172036907606136e-06,
      "loss": 0.9721,
      "num_input_tokens_seen": 4568380,
      "step": 223
    },
    {
      "epoch": 0.026934407503156375,
      "flos": 16715890176000.0,
      "grad_norm": 5.304905568108433,
      "learning_rate": 3.920445065071855e-06,
      "loss": 1.1521,
      "num_input_tokens_seen": 4589265,
      "step": 224
    },
    {
      "epoch": 0.027054650393795468,
      "flos": 20624624025600.0,
      "grad_norm": 6.7592436052409175,
      "learning_rate": 3.923672001142322e-06,
      "loss": 0.9896,
      "num_input_tokens_seen": 4609295,
      "step": 225
    },
    {
      "epoch": 0.027174893284434558,
      "flos": 22407848140800.0,
      "grad_norm": 8.166762786091983,
      "learning_rate": 3.926884627027996e-06,
      "loss": 1.0608,
      "num_input_tokens_seen": 4632785,
      "step": 226
    },
    {
      "epoch": 0.027295136175073648,
      "flos": 15668667924480.0,
      "grad_norm": 4.747745982984485,
      "learning_rate": 3.930083069088744e-06,
      "loss": 0.9943,
      "num_input_tokens_seen": 4652505,
      "step": 227
    },
    {
      "epoch": 0.02741537906571274,
      "flos": 43519598714880.0,
      "grad_norm": 1.1165536536254115,
      "learning_rate": 3.933267452018137e-06,
      "loss": 0.8447,
      "num_input_tokens_seen": 4712020,
      "step": 228
    },
    {
      "epoch": 0.02753562195635183,
      "flos": 17502310932480.0,
      "grad_norm": 9.948607564904183,
      "learning_rate": 3.936437898872622e-06,
      "loss": 1.0522,
      "num_input_tokens_seen": 4731715,
      "step": 229
    },
    {
      "epoch": 0.02765586484699092,
      "flos": 24321350369280.0,
      "grad_norm": 8.808751384398837,
      "learning_rate": 3.9395945311000525e-06,
      "loss": 1.0061,
      "num_input_tokens_seen": 4753060,
      "step": 230
    },
    {
      "epoch": 0.027776107737630014,
      "flos": 10476475944960.0,
      "grad_norm": 17.74725750435208,
      "learning_rate": 3.942737468567608e-06,
      "loss": 1.12,
      "num_input_tokens_seen": 4770795,
      "step": 231
    },
    {
      "epoch": 0.027896350628269104,
      "flos": 34257644236800.0,
      "grad_norm": 6.742998801491371,
      "learning_rate": 3.9458668295891026e-06,
      "loss": 1.0749,
      "num_input_tokens_seen": 4792800,
      "step": 232
    },
    {
      "epoch": 0.028016593518908194,
      "flos": 15402133278720.0,
      "grad_norm": 9.777127600737026,
      "learning_rate": 3.948982730951712e-06,
      "loss": 1.0813,
      "num_input_tokens_seen": 4810615,
      "step": 233
    },
    {
      "epoch": 0.028136836409547287,
      "flos": 13070257213440.0,
      "grad_norm": 6.269460669384643,
      "learning_rate": 3.9520852879421254e-06,
      "loss": 1.0335,
      "num_input_tokens_seen": 4827680,
      "step": 234
    },
    {
      "epoch": 0.028257079300186377,
      "flos": 22512242319360.0,
      "grad_norm": 8.827317354510718,
      "learning_rate": 3.955174614372137e-06,
      "loss": 1.0283,
      "num_input_tokens_seen": 4847165,
      "step": 235
    },
    {
      "epoch": 0.028377322190825467,
      "flos": 16717055201280.0,
      "grad_norm": 8.400508011488009,
      "learning_rate": 3.9582508226037045e-06,
      "loss": 1.0591,
      "num_input_tokens_seen": 4867025,
      "step": 236
    },
    {
      "epoch": 0.02849756508146456,
      "flos": 14536037253120.0,
      "grad_norm": 11.791749196836479,
      "learning_rate": 3.9613140235734636e-06,
      "loss": 1.137,
      "num_input_tokens_seen": 4883725,
      "step": 237
    },
    {
      "epoch": 0.02861780797210365,
      "flos": 10083771432960.0,
      "grad_norm": 5.926937304833728,
      "learning_rate": 3.96436432681674e-06,
      "loss": 1.0272,
      "num_input_tokens_seen": 4901435,
      "step": 238
    },
    {
      "epoch": 0.02873805086274274,
      "flos": 18366107566080.0,
      "grad_norm": 4.929444127188802,
      "learning_rate": 3.967401840491044e-06,
      "loss": 1.1062,
      "num_input_tokens_seen": 4920435,
      "step": 239
    },
    {
      "epoch": 0.028858293753381833,
      "flos": 12253813616640.0,
      "grad_norm": 6.373775888238175,
      "learning_rate": 3.97042667139909e-06,
      "loss": 1.0881,
      "num_input_tokens_seen": 4937480,
      "step": 240
    },
    {
      "epoch": 0.028978536644020923,
      "flos": 16375597363200.0,
      "grad_norm": 7.478350609876371,
      "learning_rate": 3.973438925011327e-06,
      "loss": 1.0838,
      "num_input_tokens_seen": 4955485,
      "step": 241
    },
    {
      "epoch": 0.029098779534660012,
      "flos": 20178526740480.0,
      "grad_norm": 6.194700424919044,
      "learning_rate": 3.976438705488002e-06,
      "loss": 1.1313,
      "num_input_tokens_seen": 4974865,
      "step": 242
    },
    {
      "epoch": 0.029219022425299106,
      "flos": 9795031879680.0,
      "grad_norm": 8.81185412914668,
      "learning_rate": 3.9794261157007744e-06,
      "loss": 1.1435,
      "num_input_tokens_seen": 4992340,
      "step": 243
    },
    {
      "epoch": 0.029339265315938196,
      "flos": 13776450846720.0,
      "grad_norm": 5.7761643594197745,
      "learning_rate": 3.982401257253887e-06,
      "loss": 1.0467,
      "num_input_tokens_seen": 5010400,
      "step": 244
    },
    {
      "epoch": 0.029459508206577285,
      "flos": 11079961620480.0,
      "grad_norm": 11.178200378249134,
      "learning_rate": 3.985364230504893e-06,
      "loss": 1.1053,
      "num_input_tokens_seen": 5028005,
      "step": 245
    },
    {
      "epoch": 0.02957975109721638,
      "flos": 20100476375040.0,
      "grad_norm": 28.68669895225562,
      "learning_rate": 3.988315134584976e-06,
      "loss": 1.0502,
      "num_input_tokens_seen": 5047405,
      "step": 246
    },
    {
      "epoch": 0.02969999398785547,
      "flos": 17293798502400.0,
      "grad_norm": 5.244315566798031,
      "learning_rate": 3.991254067418851e-06,
      "loss": 1.0179,
      "num_input_tokens_seen": 5067665,
      "step": 247
    },
    {
      "epoch": 0.02982023687849456,
      "flos": 25028678369280.0,
      "grad_norm": 5.380590503434489,
      "learning_rate": 3.994181125744254e-06,
      "loss": 1.038,
      "num_input_tokens_seen": 5089190,
      "step": 248
    },
    {
      "epoch": 0.02994047976913365,
      "flos": 18631538503680.0,
      "grad_norm": 12.336075417808825,
      "learning_rate": 3.99709640513106e-06,
      "loss": 0.9616,
      "num_input_tokens_seen": 5109790,
      "step": 249
    },
    {
      "epoch": 0.03006072265977274,
      "flos": 18234081914880.0,
      "grad_norm": 8.608405029155872,
      "learning_rate": 4e-06,
      "loss": 1.0598,
      "num_input_tokens_seen": 5129345,
      "step": 250
    },
    {
      "epoch": 0.03018096555041183,
      "flos": 15720803696640.0,
      "grad_norm": 14.48319138317248,
      "learning_rate": 3.999999848300794e-06,
      "loss": 1.1032,
      "num_input_tokens_seen": 5148050,
      "step": 251
    },
    {
      "epoch": 0.030301208441050925,
      "flos": 21513875374080.0,
      "grad_norm": 3.736380194357516,
      "learning_rate": 3.999999393203203e-06,
      "loss": 1.0975,
      "num_input_tokens_seen": 5170180,
      "step": 252
    },
    {
      "epoch": 0.030421451331690014,
      "flos": 16794523054080.0,
      "grad_norm": 19.08229552977175,
      "learning_rate": 3.999998634707293e-06,
      "loss": 1.0616,
      "num_input_tokens_seen": 5189450,
      "step": 253
    },
    {
      "epoch": 0.030541694222329104,
      "flos": 19890155089920.0,
      "grad_norm": 29.988087641438977,
      "learning_rate": 3.999997572813182e-06,
      "loss": 1.1822,
      "num_input_tokens_seen": 5206980,
      "step": 254
    },
    {
      "epoch": 0.030661937112968194,
      "flos": 13177717248000.0,
      "grad_norm": 6.42137422488638,
      "learning_rate": 3.999996207521028e-06,
      "loss": 1.0809,
      "num_input_tokens_seen": 5225410,
      "step": 255
    },
    {
      "epoch": 0.030782180003607287,
      "flos": 9139165163520.0,
      "grad_norm": 5.835104355585782,
      "learning_rate": 3.999994538831039e-06,
      "loss": 1.0234,
      "num_input_tokens_seen": 5241715,
      "step": 256
    },
    {
      "epoch": 0.030902422894246377,
      "flos": 16589750968320.0,
      "grad_norm": 15.301195105457397,
      "learning_rate": 3.99999256674347e-06,
      "loss": 1.0448,
      "num_input_tokens_seen": 5261585,
      "step": 257
    },
    {
      "epoch": 0.031022665784885467,
      "flos": 39416371507200.0,
      "grad_norm": 0.9364737297605722,
      "learning_rate": 3.999990291258618e-06,
      "loss": 0.7671,
      "num_input_tokens_seen": 5319995,
      "step": 258
    },
    {
      "epoch": 0.03114290867552456,
      "flos": 13701834240000.0,
      "grad_norm": 6.653087455675831,
      "learning_rate": 3.999987712376829e-06,
      "loss": 1.075,
      "num_input_tokens_seen": 5338035,
      "step": 259
    },
    {
      "epoch": 0.031263151566163654,
      "flos": 14881633996800.0,
      "grad_norm": 11.727509096329543,
      "learning_rate": 3.999984830098494e-06,
      "loss": 1.0235,
      "num_input_tokens_seen": 5357335,
      "step": 260
    },
    {
      "epoch": 0.03138339445680274,
      "flos": 10449886863360.0,
      "grad_norm": 5.00970280479619,
      "learning_rate": 3.999981644424051e-06,
      "loss": 1.1694,
      "num_input_tokens_seen": 5371855,
      "step": 261
    },
    {
      "epoch": 0.03150363734744183,
      "flos": 8195141406720.0,
      "grad_norm": 19.194850644372455,
      "learning_rate": 3.999978155353982e-06,
      "loss": 1.0766,
      "num_input_tokens_seen": 5388720,
      "step": 262
    },
    {
      "epoch": 0.03162388023808092,
      "flos": 24057728286720.0,
      "grad_norm": 6.076104384770632,
      "learning_rate": 3.9999743628888186e-06,
      "loss": 1.0084,
      "num_input_tokens_seen": 5410230,
      "step": 263
    },
    {
      "epoch": 0.03174412312872001,
      "flos": 14774572523520.0,
      "grad_norm": 6.301084867045435,
      "learning_rate": 3.999970267029133e-06,
      "loss": 1.0843,
      "num_input_tokens_seen": 5428910,
      "step": 264
    },
    {
      "epoch": 0.0318643660193591,
      "flos": 16870703247360.0,
      "grad_norm": 8.706330672942597,
      "learning_rate": 3.999965867775548e-06,
      "loss": 1.0142,
      "num_input_tokens_seen": 5449025,
      "step": 265
    },
    {
      "epoch": 0.0319846089099982,
      "flos": 9821007790080.0,
      "grad_norm": 8.455437024968202,
      "learning_rate": 3.9999611651287315e-06,
      "loss": 1.0721,
      "num_input_tokens_seen": 5466900,
      "step": 266
    },
    {
      "epoch": 0.03210485180063729,
      "flos": 10420875755520.0,
      "grad_norm": 19.010584242269406,
      "learning_rate": 3.999956159089396e-06,
      "loss": 1.0045,
      "num_input_tokens_seen": 5484070,
      "step": 267
    },
    {
      "epoch": 0.03222509469127638,
      "flos": 20597789675520.0,
      "grad_norm": 5.783531604779759,
      "learning_rate": 3.999950849658302e-06,
      "loss": 0.9931,
      "num_input_tokens_seen": 5502710,
      "step": 268
    },
    {
      "epoch": 0.03234533758191547,
      "flos": 11997580247040.0,
      "grad_norm": 7.4321616111734015,
      "learning_rate": 3.999945236836254e-06,
      "loss": 1.0533,
      "num_input_tokens_seen": 5521395,
      "step": 269
    },
    {
      "epoch": 0.03246558047255456,
      "flos": 13465567703040.0,
      "grad_norm": 6.607729876153454,
      "learning_rate": 3.999939320624103e-06,
      "loss": 1.1382,
      "num_input_tokens_seen": 5536265,
      "step": 270
    },
    {
      "epoch": 0.03258582336319365,
      "flos": 16871040491520.0,
      "grad_norm": 6.163042532850049,
      "learning_rate": 3.999933101022749e-06,
      "loss": 1.1053,
      "num_input_tokens_seen": 5556390,
      "step": 271
    },
    {
      "epoch": 0.032706066253832745,
      "flos": 19704706007040.0,
      "grad_norm": 35.55421970824982,
      "learning_rate": 3.999926578033132e-06,
      "loss": 1.0723,
      "num_input_tokens_seen": 5575925,
      "step": 272
    },
    {
      "epoch": 0.032826309144471835,
      "flos": 32608193310720.0,
      "grad_norm": 5.66422043152664,
      "learning_rate": 3.999919751656244e-06,
      "loss": 0.8571,
      "num_input_tokens_seen": 5602545,
      "step": 273
    },
    {
      "epoch": 0.032946552035110925,
      "flos": 18369388032000.0,
      "grad_norm": 4.778867240024389,
      "learning_rate": 3.9999126218931195e-06,
      "loss": 0.9798,
      "num_input_tokens_seen": 5620300,
      "step": 274
    },
    {
      "epoch": 0.033066794925750015,
      "flos": 10686398668800.0,
      "grad_norm": 7.623724908995412,
      "learning_rate": 3.99990518874484e-06,
      "loss": 1.11,
      "num_input_tokens_seen": 5636460,
      "step": 275
    },
    {
      "epoch": 0.033187037816389105,
      "flos": 16186622545920.0,
      "grad_norm": 5.4186314271759715,
      "learning_rate": 3.999897452212534e-06,
      "loss": 1.1276,
      "num_input_tokens_seen": 5653510,
      "step": 276
    },
    {
      "epoch": 0.033307280707028195,
      "flos": 16586378526720.0,
      "grad_norm": 10.77807676636877,
      "learning_rate": 3.999889412297374e-06,
      "loss": 1.2078,
      "num_input_tokens_seen": 5672655,
      "step": 277
    },
    {
      "epoch": 0.03342752359766729,
      "flos": 20545439293440.0,
      "grad_norm": 11.104354045687334,
      "learning_rate": 3.999881069000581e-06,
      "loss": 0.9995,
      "num_input_tokens_seen": 5692105,
      "step": 278
    },
    {
      "epoch": 0.03354776648830638,
      "flos": 13748972666880.0,
      "grad_norm": 6.395061566471807,
      "learning_rate": 3.99987242232342e-06,
      "loss": 1.0821,
      "num_input_tokens_seen": 5706830,
      "step": 279
    },
    {
      "epoch": 0.03366800937894547,
      "flos": 12653201694720.0,
      "grad_norm": 5.827292375175734,
      "learning_rate": 3.9998634722672026e-06,
      "loss": 1.0086,
      "num_input_tokens_seen": 5726605,
      "step": 280
    },
    {
      "epoch": 0.03378825226958456,
      "flos": 25655258050560.0,
      "grad_norm": 5.292783884571454,
      "learning_rate": 3.999854218833286e-06,
      "loss": 0.9937,
      "num_input_tokens_seen": 5747145,
      "step": 281
    },
    {
      "epoch": 0.03390849516022365,
      "flos": 18290356592640.0,
      "grad_norm": 5.169215506794863,
      "learning_rate": 3.999844662023075e-06,
      "loss": 1.0285,
      "num_input_tokens_seen": 5766740,
      "step": 282
    },
    {
      "epoch": 0.03402873805086274,
      "flos": 15114804019200.0,
      "grad_norm": 4.122766757197517,
      "learning_rate": 3.999834801838018e-06,
      "loss": 1.1332,
      "num_input_tokens_seen": 5785440,
      "step": 283
    },
    {
      "epoch": 0.03414898094150183,
      "flos": 16140403875840.0,
      "grad_norm": 3.7892945523336796,
      "learning_rate": 3.9998246382796115e-06,
      "loss": 0.9542,
      "num_input_tokens_seen": 5804740,
      "step": 284
    },
    {
      "epoch": 0.03426922383214093,
      "flos": 13387149434880.0,
      "grad_norm": 11.179742094555726,
      "learning_rate": 3.999814171349399e-06,
      "loss": 1.1111,
      "num_input_tokens_seen": 5822320,
      "step": 285
    },
    {
      "epoch": 0.03438946672278002,
      "flos": 24794220687360.0,
      "grad_norm": 4.382040843249215,
      "learning_rate": 3.9998034010489655e-06,
      "loss": 0.9466,
      "num_input_tokens_seen": 5845730,
      "step": 286
    },
    {
      "epoch": 0.03450970961341911,
      "flos": 15747484753920.0,
      "grad_norm": 7.279898980665952,
      "learning_rate": 3.999792327379946e-06,
      "loss": 0.9709,
      "num_input_tokens_seen": 5864825,
      "step": 287
    },
    {
      "epoch": 0.034629952504058197,
      "flos": 15432370728960.0,
      "grad_norm": 5.628970292075831,
      "learning_rate": 3.999780950344021e-06,
      "loss": 1.1763,
      "num_input_tokens_seen": 5882735,
      "step": 288
    },
    {
      "epoch": 0.034750195394697286,
      "flos": 14226687037440.0,
      "grad_norm": 3.8337570177430065,
      "learning_rate": 3.999769269942916e-06,
      "loss": 1.0264,
      "num_input_tokens_seen": 5902495,
      "step": 289
    },
    {
      "epoch": 0.034870438285336376,
      "flos": 19917173391360.0,
      "grad_norm": 4.369206543562858,
      "learning_rate": 3.999757286178402e-06,
      "loss": 1.0278,
      "num_input_tokens_seen": 5924650,
      "step": 290
    },
    {
      "epoch": 0.03499068117597547,
      "flos": 15642048184320.0,
      "grad_norm": 3.713650966211059,
      "learning_rate": 3.999744999052299e-06,
      "loss": 1.1122,
      "num_input_tokens_seen": 5945760,
      "step": 291
    },
    {
      "epoch": 0.03511092406661456,
      "flos": 40959466106880.0,
      "grad_norm": 1.0306913423208088,
      "learning_rate": 3.9997324085664675e-06,
      "loss": 0.8707,
      "num_input_tokens_seen": 6005710,
      "step": 292
    },
    {
      "epoch": 0.03523116695725365,
      "flos": 16296228679680.0,
      "grad_norm": 4.803270878558953,
      "learning_rate": 3.999719514722821e-06,
      "loss": 1.1291,
      "num_input_tokens_seen": 6025560,
      "step": 293
    },
    {
      "epoch": 0.03535140984789274,
      "flos": 26340074557440.0,
      "grad_norm": 4.522637963665744,
      "learning_rate": 3.999706317523314e-06,
      "loss": 0.9659,
      "num_input_tokens_seen": 6043840,
      "step": 294
    },
    {
      "epoch": 0.03547165273853183,
      "flos": 14514966712320.0,
      "grad_norm": 6.546938219880468,
      "learning_rate": 3.999692816969948e-06,
      "loss": 1.0665,
      "num_input_tokens_seen": 6063095,
      "step": 295
    },
    {
      "epoch": 0.03559189562917092,
      "flos": 50023708078080.0,
      "grad_norm": 1.0100463240924864,
      "learning_rate": 3.999679013064772e-06,
      "loss": 0.9147,
      "num_input_tokens_seen": 6129560,
      "step": 296
    },
    {
      "epoch": 0.03571213851981002,
      "flos": 15379223224320.0,
      "grad_norm": 5.251147190278029,
      "learning_rate": 3.99966490580988e-06,
      "loss": 1.069,
      "num_input_tokens_seen": 6146640,
      "step": 297
    },
    {
      "epoch": 0.03583238141044911,
      "flos": 31401375252480.0,
      "grad_norm": 4.685029052514017,
      "learning_rate": 3.999650495207411e-06,
      "loss": 0.8768,
      "num_input_tokens_seen": 6172385,
      "step": 298
    },
    {
      "epoch": 0.0359526243010882,
      "flos": 13408587878400.0,
      "grad_norm": 6.288231457126752,
      "learning_rate": 3.999635781259553e-06,
      "loss": 1.1161,
      "num_input_tokens_seen": 6187370,
      "step": 299
    },
    {
      "epoch": 0.03607286719172729,
      "flos": 44142499368960.0,
      "grad_norm": 0.991159311838692,
      "learning_rate": 3.999620763968535e-06,
      "loss": 0.7653,
      "num_input_tokens_seen": 6245965,
      "step": 300
    },
    {
      "epoch": 0.03619311008236638,
      "flos": 19811552870400.0,
      "grad_norm": 3.9692864243239936,
      "learning_rate": 3.999605443336638e-06,
      "loss": 1.0836,
      "num_input_tokens_seen": 6267815,
      "step": 301
    },
    {
      "epoch": 0.03631335297300547,
      "flos": 9607099453440.0,
      "grad_norm": 5.689652591015141,
      "learning_rate": 3.999589819366185e-06,
      "loss": 1.1111,
      "num_input_tokens_seen": 6281325,
      "step": 302
    },
    {
      "epoch": 0.036433595863644565,
      "flos": 19676308070400.0,
      "grad_norm": 3.429543042453543,
      "learning_rate": 3.999573892059547e-06,
      "loss": 1.0421,
      "num_input_tokens_seen": 6300175,
      "step": 303
    },
    {
      "epoch": 0.036553838754283655,
      "flos": 17477990584320.0,
      "grad_norm": 5.399605079621022,
      "learning_rate": 3.999557661419138e-06,
      "loss": 1.0165,
      "num_input_tokens_seen": 6320045,
      "step": 304
    },
    {
      "epoch": 0.036674081644922744,
      "flos": 16638759567360.0,
      "grad_norm": 3.4381220861553303,
      "learning_rate": 3.9995411274474225e-06,
      "loss": 1.0125,
      "num_input_tokens_seen": 6339045,
      "step": 305
    },
    {
      "epoch": 0.036794324535561834,
      "flos": 19574703820800.0,
      "grad_norm": 5.284744100014373,
      "learning_rate": 3.999524290146908e-06,
      "loss": 1.0401,
      "num_input_tokens_seen": 6358970,
      "step": 306
    },
    {
      "epoch": 0.036914567426200924,
      "flos": 13806504345600.0,
      "grad_norm": 3.7334716214953,
      "learning_rate": 3.9995071495201485e-06,
      "loss": 1.1123,
      "num_input_tokens_seen": 6375795,
      "step": 307
    },
    {
      "epoch": 0.037034810316840014,
      "flos": 15851756298240.0,
      "grad_norm": 5.781277713193711,
      "learning_rate": 3.999489705569744e-06,
      "loss": 1.1629,
      "num_input_tokens_seen": 6393215,
      "step": 308
    },
    {
      "epoch": 0.03715505320747911,
      "flos": 13177625272320.0,
      "grad_norm": 3.046445376110733,
      "learning_rate": 3.999471958298341e-06,
      "loss": 1.0771,
      "num_input_tokens_seen": 6411845,
      "step": 309
    },
    {
      "epoch": 0.0372752960981182,
      "flos": 25658937077760.0,
      "grad_norm": 3.7772933814587155,
      "learning_rate": 3.999453907708631e-06,
      "loss": 0.9818,
      "num_input_tokens_seen": 6433970,
      "step": 310
    },
    {
      "epoch": 0.03739553898875729,
      "flos": 14777209159680.0,
      "grad_norm": 5.426640683486057,
      "learning_rate": 3.999435553803353e-06,
      "loss": 1.0317,
      "num_input_tokens_seen": 6453090,
      "step": 311
    },
    {
      "epoch": 0.03751578187939638,
      "flos": 14381653401600.0,
      "grad_norm": 5.801027193860261,
      "learning_rate": 3.999416896585292e-06,
      "loss": 1.0555,
      "num_input_tokens_seen": 6469840,
      "step": 312
    },
    {
      "epoch": 0.03763602477003547,
      "flos": 14672293785600.0,
      "grad_norm": 5.375674538093053,
      "learning_rate": 3.9993979360572775e-06,
      "loss": 1.0648,
      "num_input_tokens_seen": 6489700,
      "step": 313
    },
    {
      "epoch": 0.03775626766067456,
      "flos": 11814338580480.0,
      "grad_norm": 5.861483344641671,
      "learning_rate": 3.999378672222185e-06,
      "loss": 1.0207,
      "num_input_tokens_seen": 6507205,
      "step": 314
    },
    {
      "epoch": 0.03787651055131366,
      "flos": 15012678574080.0,
      "grad_norm": 3.5788121792117993,
      "learning_rate": 3.9993591050829385e-06,
      "loss": 1.0417,
      "num_input_tokens_seen": 6524790,
      "step": 315
    },
    {
      "epoch": 0.037996753441952746,
      "flos": 15642538721280.0,
      "grad_norm": 9.720005947786246,
      "learning_rate": 3.999339234642506e-06,
      "loss": 1.0123,
      "num_input_tokens_seen": 6544260,
      "step": 316
    },
    {
      "epoch": 0.038116996332591836,
      "flos": 19732705382400.0,
      "grad_norm": 7.851103599384706,
      "learning_rate": 3.9993190609038994e-06,
      "loss": 1.0507,
      "num_input_tokens_seen": 6562745,
      "step": 317
    },
    {
      "epoch": 0.038237239223230926,
      "flos": 15038715801600.0,
      "grad_norm": 3.941299784860399,
      "learning_rate": 3.999298583870182e-06,
      "loss": 1.0569,
      "num_input_tokens_seen": 6582050,
      "step": 318
    },
    {
      "epoch": 0.038357482113870016,
      "flos": 18185533194240.0,
      "grad_norm": 34.804925118436906,
      "learning_rate": 3.999277803544458e-06,
      "loss": 0.9833,
      "num_input_tokens_seen": 6601925,
      "step": 319
    },
    {
      "epoch": 0.038477725004509106,
      "flos": 42388408995840.0,
      "grad_norm": 1.0366858930634535,
      "learning_rate": 3.999256719929882e-06,
      "loss": 0.8739,
      "num_input_tokens_seen": 6662920,
      "step": 320
    },
    {
      "epoch": 0.0385979678951482,
      "flos": 48201294213120.0,
      "grad_norm": 1.3361381692364938,
      "learning_rate": 3.999235333029651e-06,
      "loss": 0.9548,
      "num_input_tokens_seen": 6716580,
      "step": 321
    },
    {
      "epoch": 0.03871821078578729,
      "flos": 16166073200640.0,
      "grad_norm": 3.7560760864010057,
      "learning_rate": 3.999213642847009e-06,
      "loss": 1.0379,
      "num_input_tokens_seen": 6736885,
      "step": 322
    },
    {
      "epoch": 0.03883845367642638,
      "flos": 18705082060800.0,
      "grad_norm": 3.0931428255687896,
      "learning_rate": 3.999191649385247e-06,
      "loss": 1.1311,
      "num_input_tokens_seen": 6757780,
      "step": 323
    },
    {
      "epoch": 0.03895869656706547,
      "flos": 40760427171840.0,
      "grad_norm": 0.9143539355362025,
      "learning_rate": 3.999169352647702e-06,
      "loss": 0.8399,
      "num_input_tokens_seen": 6818680,
      "step": 324
    },
    {
      "epoch": 0.03907893945770456,
      "flos": 17687300136960.0,
      "grad_norm": 3.523527630585714,
      "learning_rate": 3.999146752637755e-06,
      "loss": 1.0309,
      "num_input_tokens_seen": 6839445,
      "step": 325
    },
    {
      "epoch": 0.03919918234834365,
      "flos": 13019593052160.0,
      "grad_norm": 7.753762101147507,
      "learning_rate": 3.999123849358836e-06,
      "loss": 1.1233,
      "num_input_tokens_seen": 6856830,
      "step": 326
    },
    {
      "epoch": 0.03931942523898275,
      "flos": 17947120558080.0,
      "grad_norm": 3.4966158173399986,
      "learning_rate": 3.999100642814418e-06,
      "loss": 0.9573,
      "num_input_tokens_seen": 6876990,
      "step": 327
    },
    {
      "epoch": 0.03943966812962184,
      "flos": 16533476290560.0,
      "grad_norm": 3.4836337408699185,
      "learning_rate": 3.999077133008022e-06,
      "loss": 1.1065,
      "num_input_tokens_seen": 6895295,
      "step": 328
    },
    {
      "epoch": 0.03955991102026093,
      "flos": 20781583196160.0,
      "grad_norm": 3.5399756190435623,
      "learning_rate": 3.9990533199432145e-06,
      "loss": 1.1173,
      "num_input_tokens_seen": 6916510,
      "step": 329
    },
    {
      "epoch": 0.03968015391090002,
      "flos": 12468978954240.0,
      "grad_norm": 7.497186768931543,
      "learning_rate": 3.999029203623608e-06,
      "loss": 0.9733,
      "num_input_tokens_seen": 6933950,
      "step": 330
    },
    {
      "epoch": 0.03980039680153911,
      "flos": 15485303623680.0,
      "grad_norm": 6.607408018940541,
      "learning_rate": 3.99900478405286e-06,
      "loss": 1.0706,
      "num_input_tokens_seen": 6952980,
      "step": 331
    },
    {
      "epoch": 0.0399206396921782,
      "flos": 10738994319360.0,
      "grad_norm": 19.125188290535533,
      "learning_rate": 3.998980061234676e-06,
      "loss": 1.0221,
      "num_input_tokens_seen": 6970615,
      "step": 332
    },
    {
      "epoch": 0.040040882582817294,
      "flos": 10183198924800.0,
      "grad_norm": 13.581232450625741,
      "learning_rate": 3.9989550351728055e-06,
      "loss": 0.9684,
      "num_input_tokens_seen": 6987265,
      "step": 333
    },
    {
      "epoch": 0.040161125473456384,
      "flos": 13675275816960.0,
      "grad_norm": 4.041636489807943,
      "learning_rate": 3.998929705871046e-06,
      "loss": 1.0499,
      "num_input_tokens_seen": 7004340,
      "step": 334
    },
    {
      "epoch": 0.040281368364095474,
      "flos": 12626091417600.0,
      "grad_norm": 24.450652580679595,
      "learning_rate": 3.99890407333324e-06,
      "loss": 1.104,
      "num_input_tokens_seen": 7022590,
      "step": 335
    },
    {
      "epoch": 0.040401611254734564,
      "flos": 13885873029120.0,
      "grad_norm": 4.765509532092436,
      "learning_rate": 3.998878137563275e-06,
      "loss": 1.0755,
      "num_input_tokens_seen": 7041860,
      "step": 336
    },
    {
      "epoch": 0.040521854145373654,
      "flos": 15668667924480.0,
      "grad_norm": 4.348917569942131,
      "learning_rate": 3.998851898565085e-06,
      "loss": 1.0524,
      "num_input_tokens_seen": 7061385,
      "step": 337
    },
    {
      "epoch": 0.04064209703601274,
      "flos": 16114366648320.0,
      "grad_norm": 5.092130438644989,
      "learning_rate": 3.998825356342653e-06,
      "loss": 1.0466,
      "num_input_tokens_seen": 7081280,
      "step": 338
    },
    {
      "epoch": 0.04076233992665183,
      "flos": 27546769981440.0,
      "grad_norm": 4.279125271119963,
      "learning_rate": 3.998798510900003e-06,
      "loss": 0.9352,
      "num_input_tokens_seen": 7103800,
      "step": 339
    },
    {
      "epoch": 0.04088258281729093,
      "flos": 18421799731200.0,
      "grad_norm": 108.58013403847178,
      "learning_rate": 3.998771362241207e-06,
      "loss": 1.0459,
      "num_input_tokens_seen": 7123925,
      "step": 340
    },
    {
      "epoch": 0.04100282570793002,
      "flos": 14040778076160.0,
      "grad_norm": 13.48804311710584,
      "learning_rate": 3.998743910370385e-06,
      "loss": 1.0839,
      "num_input_tokens_seen": 7142505,
      "step": 341
    },
    {
      "epoch": 0.04112306859856911,
      "flos": 16035672453120.0,
      "grad_norm": 7.529670771358388,
      "learning_rate": 3.998716155291702e-06,
      "loss": 0.9436,
      "num_input_tokens_seen": 7160065,
      "step": 342
    },
    {
      "epoch": 0.0412433114892082,
      "flos": 17814021857280.0,
      "grad_norm": 7.217337862419391,
      "learning_rate": 3.998688097009366e-06,
      "loss": 1.113,
      "num_input_tokens_seen": 7180550,
      "step": 343
    },
    {
      "epoch": 0.04136355437984729,
      "flos": 17922984161280.0,
      "grad_norm": 15.941954704483534,
      "learning_rate": 3.998659735527636e-06,
      "loss": 1.0188,
      "num_input_tokens_seen": 7199360,
      "step": 344
    },
    {
      "epoch": 0.04148379727048638,
      "flos": 16324350689280.0,
      "grad_norm": 5.4999628932741516,
      "learning_rate": 3.998631070850813e-06,
      "loss": 0.9869,
      "num_input_tokens_seen": 7219700,
      "step": 345
    },
    {
      "epoch": 0.041604040161125476,
      "flos": 9925616578560.0,
      "grad_norm": 3.600816761188529,
      "learning_rate": 3.9986021029832455e-06,
      "loss": 1.0475,
      "num_input_tokens_seen": 7236735,
      "step": 346
    },
    {
      "epoch": 0.041724283051764566,
      "flos": 8508477235200.0,
      "grad_norm": 11.955645286793716,
      "learning_rate": 3.9985728319293285e-06,
      "loss": 1.1284,
      "num_input_tokens_seen": 7250430,
      "step": 347
    },
    {
      "epoch": 0.041844525942403656,
      "flos": 8719564984320.0,
      "grad_norm": 37.33501957309568,
      "learning_rate": 3.998543257693501e-06,
      "loss": 1.0623,
      "num_input_tokens_seen": 7266905,
      "step": 348
    },
    {
      "epoch": 0.041964768833042745,
      "flos": 16901124648960.0,
      "grad_norm": 11.21655660030178,
      "learning_rate": 3.998513380280251e-06,
      "loss": 1.0955,
      "num_input_tokens_seen": 7286905,
      "step": 349
    },
    {
      "epoch": 0.042085011723681835,
      "flos": 8353173626880.0,
      "grad_norm": 6.376908259796233,
      "learning_rate": 3.99848319969411e-06,
      "loss": 1.1553,
      "num_input_tokens_seen": 7304225,
      "step": 350
    },
    {
      "epoch": 0.042205254614320925,
      "flos": 11945597767680.0,
      "grad_norm": 7.668419209180247,
      "learning_rate": 3.9984527159396564e-06,
      "loss": 0.9933,
      "num_input_tokens_seen": 7322585,
      "step": 351
    },
    {
      "epoch": 0.04232549750496002,
      "flos": 17870633779200.0,
      "grad_norm": 5.3073910550091385,
      "learning_rate": 3.9984219290215154e-06,
      "loss": 1.0426,
      "num_input_tokens_seen": 7342480,
      "step": 352
    },
    {
      "epoch": 0.04244574039559911,
      "flos": 19024948162560.0,
      "grad_norm": 13.495362226918632,
      "learning_rate": 3.998390838944356e-06,
      "loss": 1.1011,
      "num_input_tokens_seen": 7363705,
      "step": 353
    },
    {
      "epoch": 0.0425659832862382,
      "flos": 14855658086400.0,
      "grad_norm": 16.438293669943267,
      "learning_rate": 3.998359445712895e-06,
      "loss": 1.0943,
      "num_input_tokens_seen": 7382530,
      "step": 354
    },
    {
      "epoch": 0.04268622617687729,
      "flos": 16586133258240.0,
      "grad_norm": 18.614506370358292,
      "learning_rate": 3.9983277493318955e-06,
      "loss": 1.0146,
      "num_input_tokens_seen": 7401545,
      "step": 355
    },
    {
      "epoch": 0.04280646906751638,
      "flos": 18499819438080.0,
      "grad_norm": 3.9490000799144775,
      "learning_rate": 3.998295749806165e-06,
      "loss": 1.0177,
      "num_input_tokens_seen": 7422490,
      "step": 356
    },
    {
      "epoch": 0.04292671195815547,
      "flos": 19155931422720.0,
      "grad_norm": 4.378794523203895,
      "learning_rate": 3.998263447140558e-06,
      "loss": 1.0453,
      "num_input_tokens_seen": 7442410,
      "step": 357
    },
    {
      "epoch": 0.04304695484879457,
      "flos": 28175649054720.0,
      "grad_norm": 4.184380579311534,
      "learning_rate": 3.998230841339976e-06,
      "loss": 1.0247,
      "num_input_tokens_seen": 7464140,
      "step": 358
    },
    {
      "epoch": 0.04316719773943366,
      "flos": 13937947484160.0,
      "grad_norm": 11.718670572647495,
      "learning_rate": 3.998197932409363e-06,
      "loss": 1.0616,
      "num_input_tokens_seen": 7481870,
      "step": 359
    },
    {
      "epoch": 0.04328744063007275,
      "flos": 15954127011840.0,
      "grad_norm": 4.155657733220905,
      "learning_rate": 3.9981647203537125e-06,
      "loss": 1.0722,
      "num_input_tokens_seen": 7499090,
      "step": 360
    },
    {
      "epoch": 0.04340768352071184,
      "flos": 15113547018240.0,
      "grad_norm": 4.144545516439866,
      "learning_rate": 3.998131205178063e-06,
      "loss": 1.1662,
      "num_input_tokens_seen": 7517280,
      "step": 361
    },
    {
      "epoch": 0.04352792641135093,
      "flos": 8143618805760.0,
      "grad_norm": 3.9552150825052426,
      "learning_rate": 3.998097386887498e-06,
      "loss": 0.9656,
      "num_input_tokens_seen": 7534075,
      "step": 362
    },
    {
      "epoch": 0.04364816930199002,
      "flos": 16796362567680.0,
      "grad_norm": 2.9311176522315074,
      "learning_rate": 3.998063265487148e-06,
      "loss": 1.0503,
      "num_input_tokens_seen": 7554845,
      "step": 363
    },
    {
      "epoch": 0.043768412192629114,
      "flos": 10189024051200.0,
      "grad_norm": 5.45293751916255,
      "learning_rate": 3.99802884098219e-06,
      "loss": 1.0222,
      "num_input_tokens_seen": 7572675,
      "step": 364
    },
    {
      "epoch": 0.043888655083268203,
      "flos": 18841185300480.0,
      "grad_norm": 5.717671610563188,
      "learning_rate": 3.997994113377845e-06,
      "loss": 1.0242,
      "num_input_tokens_seen": 7591295,
      "step": 365
    },
    {
      "epoch": 0.04400889797390729,
      "flos": 19391983349760.0,
      "grad_norm": 3.870184089954471,
      "learning_rate": 3.9979590826793815e-06,
      "loss": 1.0426,
      "num_input_tokens_seen": 7612205,
      "step": 366
    },
    {
      "epoch": 0.04412914086454638,
      "flos": 14277903052800.0,
      "grad_norm": 3.5365405665384317,
      "learning_rate": 3.997923748892113e-06,
      "loss": 1.0124,
      "num_input_tokens_seen": 7631245,
      "step": 367
    },
    {
      "epoch": 0.04424938375518547,
      "flos": 16064438292480.0,
      "grad_norm": 2.641551650839702,
      "learning_rate": 3.9978881120214015e-06,
      "loss": 1.0968,
      "num_input_tokens_seen": 7652485,
      "step": 368
    },
    {
      "epoch": 0.04436962664582456,
      "flos": 17161772851200.0,
      "grad_norm": 51.0158212094236,
      "learning_rate": 3.997852172072652e-06,
      "loss": 1.006,
      "num_input_tokens_seen": 7673420,
      "step": 369
    },
    {
      "epoch": 0.04448986953646366,
      "flos": 12914922946560.0,
      "grad_norm": 9.574891681905067,
      "learning_rate": 3.9978159290513155e-06,
      "loss": 1.0973,
      "num_input_tokens_seen": 7691220,
      "step": 370
    },
    {
      "epoch": 0.04461011242710275,
      "flos": 21463732408320.0,
      "grad_norm": 8.024953073144532,
      "learning_rate": 3.997779382962892e-06,
      "loss": 1.0202,
      "num_input_tokens_seen": 7713825,
      "step": 371
    },
    {
      "epoch": 0.04473035531774184,
      "flos": 21201060741120.0,
      "grad_norm": 7.680380897613208,
      "learning_rate": 3.997742533812924e-06,
      "loss": 0.9503,
      "num_input_tokens_seen": 7736810,
      "step": 372
    },
    {
      "epoch": 0.04485059820838093,
      "flos": 9270271057920.0,
      "grad_norm": 6.598178334694038,
      "learning_rate": 3.997705381607001e-06,
      "loss": 1.1233,
      "num_input_tokens_seen": 7753345,
      "step": 373
    },
    {
      "epoch": 0.04497084109902002,
      "flos": 48760768634880.0,
      "grad_norm": 1.1023971425730874,
      "learning_rate": 3.997667926350761e-06,
      "loss": 0.8404,
      "num_input_tokens_seen": 7811395,
      "step": 374
    },
    {
      "epoch": 0.04509108398965911,
      "flos": 46065137848320.0,
      "grad_norm": 0.9127874365855545,
      "learning_rate": 3.997630168049886e-06,
      "loss": 0.8082,
      "num_input_tokens_seen": 7869480,
      "step": 375
    },
    {
      "epoch": 0.045211326880298205,
      "flos": 15824063508480.0,
      "grad_norm": 2.8375075016678784,
      "learning_rate": 3.997592106710101e-06,
      "loss": 0.9915,
      "num_input_tokens_seen": 7888660,
      "step": 376
    },
    {
      "epoch": 0.045331569770937295,
      "flos": 22930708131840.0,
      "grad_norm": 10.95032764798134,
      "learning_rate": 3.997553742337182e-06,
      "loss": 0.8772,
      "num_input_tokens_seen": 7907805,
      "step": 377
    },
    {
      "epoch": 0.045451812661576385,
      "flos": 15746932899840.0,
      "grad_norm": 2.7394251807993424,
      "learning_rate": 3.997515074936949e-06,
      "loss": 1.1244,
      "num_input_tokens_seen": 7928400,
      "step": 378
    },
    {
      "epoch": 0.045572055552215475,
      "flos": 11735736360960.0,
      "grad_norm": 7.484254202933054,
      "learning_rate": 3.997476104515268e-06,
      "loss": 1.0692,
      "num_input_tokens_seen": 7946310,
      "step": 379
    },
    {
      "epoch": 0.045692298442854565,
      "flos": 12469776076800.0,
      "grad_norm": 5.233580389488377,
      "learning_rate": 3.9974368310780485e-06,
      "loss": 0.9878,
      "num_input_tokens_seen": 7963205,
      "step": 380
    },
    {
      "epoch": 0.045812541333493655,
      "flos": 19051230658560.0,
      "grad_norm": 7.620224254442141,
      "learning_rate": 3.997397254631251e-06,
      "loss": 0.9501,
      "num_input_tokens_seen": 7983545,
      "step": 381
    },
    {
      "epoch": 0.04593278422413275,
      "flos": 43123613736960.0,
      "grad_norm": 1.1551509479456237,
      "learning_rate": 3.997357375180878e-06,
      "loss": 0.8667,
      "num_input_tokens_seen": 8047545,
      "step": 382
    },
    {
      "epoch": 0.04605302711477184,
      "flos": 15485272965120.0,
      "grad_norm": 4.471698348570202,
      "learning_rate": 3.997317192732979e-06,
      "loss": 0.9533,
      "num_input_tokens_seen": 8066045,
      "step": 383
    },
    {
      "epoch": 0.04617327000541093,
      "flos": 13803285196800.0,
      "grad_norm": 5.91009525486949,
      "learning_rate": 3.99727670729365e-06,
      "loss": 1.031,
      "num_input_tokens_seen": 8084325,
      "step": 384
    },
    {
      "epoch": 0.04629351289605002,
      "flos": 18081322967040.0,
      "grad_norm": 3.3138740137034075,
      "learning_rate": 3.997235918869033e-06,
      "loss": 0.9917,
      "num_input_tokens_seen": 8105080,
      "step": 385
    },
    {
      "epoch": 0.04641375578668911,
      "flos": 14593354321920.0,
      "grad_norm": 5.303849661144945,
      "learning_rate": 3.997194827465315e-06,
      "loss": 1.0478,
      "num_input_tokens_seen": 8123395,
      "step": 386
    },
    {
      "epoch": 0.0465339986773282,
      "flos": 9297013432320.0,
      "grad_norm": 14.327598903236762,
      "learning_rate": 3.997153433088728e-06,
      "loss": 1.1268,
      "num_input_tokens_seen": 8140240,
      "step": 387
    },
    {
      "epoch": 0.0466542415679673,
      "flos": 18185471877120.0,
      "grad_norm": 6.232680470485229,
      "learning_rate": 3.997111735745554e-06,
      "loss": 1.0186,
      "num_input_tokens_seen": 8162930,
      "step": 388
    },
    {
      "epoch": 0.04677448445860639,
      "flos": 15799651184640.0,
      "grad_norm": 13.371860028619803,
      "learning_rate": 3.997069735442118e-06,
      "loss": 1.0412,
      "num_input_tokens_seen": 8182345,
      "step": 389
    },
    {
      "epoch": 0.04689472734924548,
      "flos": 20047880724480.0,
      "grad_norm": 3.764450713233377,
      "learning_rate": 3.997027432184792e-06,
      "loss": 1.0124,
      "num_input_tokens_seen": 8206725,
      "step": 390
    },
    {
      "epoch": 0.04701497023988457,
      "flos": 16455241973760.0,
      "grad_norm": 3.908410291313979,
      "learning_rate": 3.99698482597999e-06,
      "loss": 1.099,
      "num_input_tokens_seen": 8224125,
      "step": 391
    },
    {
      "epoch": 0.04713521313052366,
      "flos": 46415486668800.0,
      "grad_norm": 1.090725886971724,
      "learning_rate": 3.99694191683418e-06,
      "loss": 0.8714,
      "num_input_tokens_seen": 8284645,
      "step": 392
    },
    {
      "epoch": 0.047255456021162746,
      "flos": 13309436313600.0,
      "grad_norm": 4.9566664685521005,
      "learning_rate": 3.996898704753867e-06,
      "loss": 1.0394,
      "num_input_tokens_seen": 8302315,
      "step": 393
    },
    {
      "epoch": 0.04737569891180184,
      "flos": 15668116070400.0,
      "grad_norm": 18.646856604545142,
      "learning_rate": 3.996855189745609e-06,
      "loss": 1.109,
      "num_input_tokens_seen": 8321300,
      "step": 394
    },
    {
      "epoch": 0.04749594180244093,
      "flos": 20701202780160.0,
      "grad_norm": 15.746397060462504,
      "learning_rate": 3.996811371816007e-06,
      "loss": 1.1391,
      "num_input_tokens_seen": 8343445,
      "step": 395
    },
    {
      "epoch": 0.04761618469308002,
      "flos": 25052508180480.0,
      "grad_norm": 6.1936322275122,
      "learning_rate": 3.996767250971707e-06,
      "loss": 0.9862,
      "num_input_tokens_seen": 8365905,
      "step": 396
    },
    {
      "epoch": 0.04773642758371911,
      "flos": 18238925967360.0,
      "grad_norm": 5.945396232265986,
      "learning_rate": 3.996722827219403e-06,
      "loss": 1.0725,
      "num_input_tokens_seen": 8387240,
      "step": 397
    },
    {
      "epoch": 0.0478566704743582,
      "flos": 14646011289600.0,
      "grad_norm": 9.242632019239238,
      "learning_rate": 3.996678100565833e-06,
      "loss": 1.0224,
      "num_input_tokens_seen": 8406015,
      "step": 398
    },
    {
      "epoch": 0.04797691336499729,
      "flos": 13354336665600.0,
      "grad_norm": 12.530466411948781,
      "learning_rate": 3.996633071017783e-06,
      "loss": 1.0847,
      "num_input_tokens_seen": 8422365,
      "step": 399
    },
    {
      "epoch": 0.04809715625563638,
      "flos": 14981644001280.0,
      "grad_norm": 5.011954892928994,
      "learning_rate": 3.996587738582084e-06,
      "loss": 1.0288,
      "num_input_tokens_seen": 8438885,
      "step": 400
    },
    {
      "epoch": 0.04821739914627548,
      "flos": 16927192535040.0,
      "grad_norm": 8.314152832916987,
      "learning_rate": 3.9965421032656115e-06,
      "loss": 1.0632,
      "num_input_tokens_seen": 8458535,
      "step": 401
    },
    {
      "epoch": 0.04833764203691457,
      "flos": 15773705932800.0,
      "grad_norm": 8.817785963845981,
      "learning_rate": 3.99649616507529e-06,
      "loss": 1.1532,
      "num_input_tokens_seen": 8477350,
      "step": 402
    },
    {
      "epoch": 0.04845788492755366,
      "flos": 47186884730880.0,
      "grad_norm": 1.1640707438132218,
      "learning_rate": 3.996449924018088e-06,
      "loss": 0.8905,
      "num_input_tokens_seen": 8537530,
      "step": 403
    },
    {
      "epoch": 0.04857812781819275,
      "flos": 13675613061120.0,
      "grad_norm": 9.12849965330613,
      "learning_rate": 3.99640338010102e-06,
      "loss": 1.0127,
      "num_input_tokens_seen": 8556355,
      "step": 404
    },
    {
      "epoch": 0.04869837070883184,
      "flos": 17111875153920.0,
      "grad_norm": 6.526393417129881,
      "learning_rate": 3.996356533331146e-06,
      "loss": 0.9881,
      "num_input_tokens_seen": 8577945,
      "step": 405
    },
    {
      "epoch": 0.04881861359947093,
      "flos": 17919397109760.0,
      "grad_norm": 3.6521836842431883,
      "learning_rate": 3.996309383715573e-06,
      "loss": 0.8093,
      "num_input_tokens_seen": 8596445,
      "step": 406
    },
    {
      "epoch": 0.048938856490110025,
      "flos": 11578286653440.0,
      "grad_norm": 5.479255395129761,
      "learning_rate": 3.996261931261454e-06,
      "loss": 0.9297,
      "num_input_tokens_seen": 8614745,
      "step": 407
    },
    {
      "epoch": 0.049059099380749115,
      "flos": 21303063552000.0,
      "grad_norm": 3.930417184599507,
      "learning_rate": 3.996214175975987e-06,
      "loss": 1.0776,
      "num_input_tokens_seen": 8634985,
      "step": 408
    },
    {
      "epoch": 0.049179342271388204,
      "flos": 25632715898880.0,
      "grad_norm": 6.210067131429053,
      "learning_rate": 3.996166117866417e-06,
      "loss": 1.0054,
      "num_input_tokens_seen": 8656640,
      "step": 409
    },
    {
      "epoch": 0.049299585162027294,
      "flos": 10319332823040.0,
      "grad_norm": 2.9591577661656694,
      "learning_rate": 3.996117756940035e-06,
      "loss": 1.0835,
      "num_input_tokens_seen": 8673045,
      "step": 410
    },
    {
      "epoch": 0.049419828052666384,
      "flos": 13880906342400.0,
      "grad_norm": 10.849710810450672,
      "learning_rate": 3.996069093204175e-06,
      "loss": 1.1856,
      "num_input_tokens_seen": 8688725,
      "step": 411
    },
    {
      "epoch": 0.049540070943305474,
      "flos": 9635282780160.0,
      "grad_norm": 78.84602590353714,
      "learning_rate": 3.996020126666221e-06,
      "loss": 1.0852,
      "num_input_tokens_seen": 8705425,
      "step": 412
    },
    {
      "epoch": 0.04966031383394457,
      "flos": 15508336312320.0,
      "grad_norm": 4.456144038969669,
      "learning_rate": 3.995970857333601e-06,
      "loss": 1.0373,
      "num_input_tokens_seen": 8725555,
      "step": 413
    },
    {
      "epoch": 0.04978055672458366,
      "flos": 20385904803840.0,
      "grad_norm": 3.1577120878703955,
      "learning_rate": 3.995921285213789e-06,
      "loss": 1.0069,
      "num_input_tokens_seen": 8745535,
      "step": 414
    },
    {
      "epoch": 0.04990079961522275,
      "flos": 14068624158720.0,
      "grad_norm": 11.014169410146648,
      "learning_rate": 3.995871410314305e-06,
      "loss": 1.007,
      "num_input_tokens_seen": 8763815,
      "step": 415
    },
    {
      "epoch": 0.05002104250586184,
      "flos": 44909811732480.0,
      "grad_norm": 1.106098871630114,
      "learning_rate": 3.995821232642714e-06,
      "loss": 0.867,
      "num_input_tokens_seen": 8821940,
      "step": 416
    },
    {
      "epoch": 0.05014128539650093,
      "flos": 19890216407040.0,
      "grad_norm": 3.7370312553080707,
      "learning_rate": 3.995770752206629e-06,
      "loss": 1.0277,
      "num_input_tokens_seen": 8842735,
      "step": 417
    },
    {
      "epoch": 0.05026152828714002,
      "flos": 12542246584320.0,
      "grad_norm": 9.024093168142947,
      "learning_rate": 3.995719969013709e-06,
      "loss": 1.1808,
      "num_input_tokens_seen": 8859635,
      "step": 418
    },
    {
      "epoch": 0.05038177117777912,
      "flos": 13568950149120.0,
      "grad_norm": 4.826451675165211,
      "learning_rate": 3.995668883071655e-06,
      "loss": 1.0554,
      "num_input_tokens_seen": 8875580,
      "step": 419
    },
    {
      "epoch": 0.050502014068418206,
      "flos": 14671987200000.0,
      "grad_norm": 3.5241544209830775,
      "learning_rate": 3.995617494388219e-06,
      "loss": 1.1108,
      "num_input_tokens_seen": 8893420,
      "step": 420
    },
    {
      "epoch": 0.050622256959057296,
      "flos": 15170342891520.0,
      "grad_norm": 7.321540395700152,
      "learning_rate": 3.995565802971196e-06,
      "loss": 1.0185,
      "num_input_tokens_seen": 8913115,
      "step": 421
    },
    {
      "epoch": 0.050742499849696386,
      "flos": 19707066716160.0,
      "grad_norm": 6.313893683654036,
      "learning_rate": 3.995513808828427e-06,
      "loss": 0.892,
      "num_input_tokens_seen": 8935630,
      "step": 422
    },
    {
      "epoch": 0.050862742740335476,
      "flos": 14095949045760.0,
      "grad_norm": 3.1169629314863156,
      "learning_rate": 3.9954615119678e-06,
      "loss": 0.9697,
      "num_input_tokens_seen": 8953905,
      "step": 423
    },
    {
      "epoch": 0.050982985630974566,
      "flos": 15689891758080.0,
      "grad_norm": 4.890692469707327,
      "learning_rate": 3.995408912397248e-06,
      "loss": 1.0067,
      "num_input_tokens_seen": 8971520,
      "step": 424
    },
    {
      "epoch": 0.05110322852161366,
      "flos": 14724674826240.0,
      "grad_norm": 10.804147658580845,
      "learning_rate": 3.99535601012475e-06,
      "loss": 1.1368,
      "num_input_tokens_seen": 8986570,
      "step": 425
    },
    {
      "epoch": 0.05122347141225275,
      "flos": 20334719447040.0,
      "grad_norm": 2.9018258496537572,
      "learning_rate": 3.995302805158333e-06,
      "loss": 0.9646,
      "num_input_tokens_seen": 9008945,
      "step": 426
    },
    {
      "epoch": 0.05134371430289184,
      "flos": 13993302405120.0,
      "grad_norm": 6.044760637490188,
      "learning_rate": 3.9952492975060665e-06,
      "loss": 1.0488,
      "num_input_tokens_seen": 9028735,
      "step": 427
    },
    {
      "epoch": 0.05146395719353093,
      "flos": 24583439523840.0,
      "grad_norm": 6.3283848085169385,
      "learning_rate": 3.995195487176067e-06,
      "loss": 1.0768,
      "num_input_tokens_seen": 9048685,
      "step": 428
    },
    {
      "epoch": 0.05158420008417002,
      "flos": 15456936345600.0,
      "grad_norm": 3.6536900614397325,
      "learning_rate": 3.995141374176499e-06,
      "loss": 1.0664,
      "num_input_tokens_seen": 9066800,
      "step": 429
    },
    {
      "epoch": 0.05170444297480911,
      "flos": 51967049195520.0,
      "grad_norm": 1.0291679634569983,
      "learning_rate": 3.995086958515572e-06,
      "loss": 0.881,
      "num_input_tokens_seen": 9124540,
      "step": 430
    },
    {
      "epoch": 0.05182468586544821,
      "flos": 44681914982400.0,
      "grad_norm": 0.9501969120839407,
      "learning_rate": 3.995032240201538e-06,
      "loss": 0.8477,
      "num_input_tokens_seen": 9186655,
      "step": 431
    },
    {
      "epoch": 0.0519449287560873,
      "flos": 29449104384000.0,
      "grad_norm": 1.020556697860971,
      "learning_rate": 3.9949772192427e-06,
      "loss": 0.8542,
      "num_input_tokens_seen": 9233000,
      "step": 432
    },
    {
      "epoch": 0.05206517164672639,
      "flos": 12391633735680.0,
      "grad_norm": 3.3776551151869305,
      "learning_rate": 3.994921895647405e-06,
      "loss": 1.0022,
      "num_input_tokens_seen": 9250890,
      "step": 433
    },
    {
      "epoch": 0.05218541453736548,
      "flos": 45819643084800.0,
      "grad_norm": 0.8895474993228204,
      "learning_rate": 3.994866269424043e-06,
      "loss": 0.7752,
      "num_input_tokens_seen": 9306980,
      "step": 434
    },
    {
      "epoch": 0.05230565742800457,
      "flos": 13701435678720.0,
      "grad_norm": 5.518311103139282,
      "learning_rate": 3.9948103405810545e-06,
      "loss": 0.9861,
      "num_input_tokens_seen": 9325650,
      "step": 435
    },
    {
      "epoch": 0.05242590031864366,
      "flos": 17999777525760.0,
      "grad_norm": 8.490277283434821,
      "learning_rate": 3.994754109126923e-06,
      "loss": 1.0594,
      "num_input_tokens_seen": 9346865,
      "step": 436
    },
    {
      "epoch": 0.052546143209282754,
      "flos": 18656042803200.0,
      "grad_norm": 4.753959611579109,
      "learning_rate": 3.994697575070181e-06,
      "loss": 1.134,
      "num_input_tokens_seen": 9366045,
      "step": 437
    },
    {
      "epoch": 0.052666386099921844,
      "flos": 15742395432960.0,
      "grad_norm": 4.417688983184357,
      "learning_rate": 3.994640738419402e-06,
      "loss": 1.1205,
      "num_input_tokens_seen": 9385140,
      "step": 438
    },
    {
      "epoch": 0.052786628990560934,
      "flos": 16982363504640.0,
      "grad_norm": 5.238121400082472,
      "learning_rate": 3.9945835991832075e-06,
      "loss": 1.014,
      "num_input_tokens_seen": 9406745,
      "step": 439
    },
    {
      "epoch": 0.052906871881200024,
      "flos": 17501237882880.0,
      "grad_norm": 3.546592936454532,
      "learning_rate": 3.994526157370268e-06,
      "loss": 1.135,
      "num_input_tokens_seen": 9425080,
      "step": 440
    },
    {
      "epoch": 0.053027114771839114,
      "flos": 40401516503040.0,
      "grad_norm": 1.1013433360145097,
      "learning_rate": 3.994468412989296e-06,
      "loss": 0.8221,
      "num_input_tokens_seen": 9486210,
      "step": 441
    },
    {
      "epoch": 0.053147357662478203,
      "flos": 12259669401600.0,
      "grad_norm": 9.931320860449688,
      "learning_rate": 3.994410366049052e-06,
      "loss": 1.1514,
      "num_input_tokens_seen": 9503790,
      "step": 442
    },
    {
      "epoch": 0.0532676005531173,
      "flos": 12154631393280.0,
      "grad_norm": 4.379630989311663,
      "learning_rate": 3.994352016558341e-06,
      "loss": 1.0385,
      "num_input_tokens_seen": 9520815,
      "step": 443
    },
    {
      "epoch": 0.05338784344375639,
      "flos": 19812319334400.0,
      "grad_norm": 4.137036650836708,
      "learning_rate": 3.994293364526014e-06,
      "loss": 0.9467,
      "num_input_tokens_seen": 9541420,
      "step": 444
    },
    {
      "epoch": 0.05350808633439548,
      "flos": 15275779461120.0,
      "grad_norm": 4.49844863429121,
      "learning_rate": 3.99423440996097e-06,
      "loss": 1.0358,
      "num_input_tokens_seen": 9560680,
      "step": 445
    },
    {
      "epoch": 0.05362832922503457,
      "flos": 14252754923520.0,
      "grad_norm": 8.553744329428058,
      "learning_rate": 3.994175152872152e-06,
      "loss": 1.0161,
      "num_input_tokens_seen": 9579485,
      "step": 446
    },
    {
      "epoch": 0.05374857211567366,
      "flos": 18602374103040.0,
      "grad_norm": 5.228662453870832,
      "learning_rate": 3.994115593268548e-06,
      "loss": 0.9993,
      "num_input_tokens_seen": 9598985,
      "step": 447
    },
    {
      "epoch": 0.05386881500631275,
      "flos": 19572251136000.0,
      "grad_norm": 10.822303869664873,
      "learning_rate": 3.994055731159195e-06,
      "loss": 1.0257,
      "num_input_tokens_seen": 9616175,
      "step": 448
    },
    {
      "epoch": 0.053989057896951846,
      "flos": 16768455168000.0,
      "grad_norm": 4.331485476567574,
      "learning_rate": 3.993995566553172e-06,
      "loss": 1.0864,
      "num_input_tokens_seen": 9634860,
      "step": 449
    },
    {
      "epoch": 0.054109300787590936,
      "flos": 17950063779840.0,
      "grad_norm": 3.1247657688096107,
      "learning_rate": 3.993935099459607e-06,
      "loss": 0.9762,
      "num_input_tokens_seen": 9656195,
      "step": 450
    },
    {
      "epoch": 0.054229543678230026,
      "flos": 16952892518400.0,
      "grad_norm": 9.001188411663518,
      "learning_rate": 3.993874329887673e-06,
      "loss": 0.9534,
      "num_input_tokens_seen": 9674570,
      "step": 451
    },
    {
      "epoch": 0.054349786568869116,
      "flos": 11547374714880.0,
      "grad_norm": 4.460689931971225,
      "learning_rate": 3.993813257846589e-06,
      "loss": 1.066,
      "num_input_tokens_seen": 9691045,
      "step": 452
    },
    {
      "epoch": 0.054470029459508205,
      "flos": 13229577093120.0,
      "grad_norm": 7.127969758694222,
      "learning_rate": 3.993751883345619e-06,
      "loss": 1.1361,
      "num_input_tokens_seen": 9709125,
      "step": 453
    },
    {
      "epoch": 0.054590272350147295,
      "flos": 12600452751360.0,
      "grad_norm": 6.123204805080501,
      "learning_rate": 3.993690206394073e-06,
      "loss": 1.0788,
      "num_input_tokens_seen": 9725145,
      "step": 454
    },
    {
      "epoch": 0.054710515240786385,
      "flos": 12601004605440.0,
      "grad_norm": 5.982409759618813,
      "learning_rate": 3.993628227001307e-06,
      "loss": 1.089,
      "num_input_tokens_seen": 9743065,
      "step": 455
    },
    {
      "epoch": 0.05483075813142548,
      "flos": 34466279301120.0,
      "grad_norm": 7.901525963307855,
      "learning_rate": 3.993565945176726e-06,
      "loss": 0.9225,
      "num_input_tokens_seen": 9763810,
      "step": 456
    },
    {
      "epoch": 0.05495100102206457,
      "flos": 13963708784640.0,
      "grad_norm": 5.489972076067063,
      "learning_rate": 3.993503360929776e-06,
      "loss": 1.0334,
      "num_input_tokens_seen": 9782415,
      "step": 457
    },
    {
      "epoch": 0.05507124391270366,
      "flos": 18763042959360.0,
      "grad_norm": 5.9834959016921205,
      "learning_rate": 3.99344047426995e-06,
      "loss": 1.0251,
      "num_input_tokens_seen": 9803395,
      "step": 458
    },
    {
      "epoch": 0.05519148680334275,
      "flos": 16061617704960.0,
      "grad_norm": 3.9903119304213535,
      "learning_rate": 3.993377285206789e-06,
      "loss": 1.1274,
      "num_input_tokens_seen": 9822900,
      "step": 459
    },
    {
      "epoch": 0.05531172969398184,
      "flos": 28962836275200.0,
      "grad_norm": 4.044245156302638,
      "learning_rate": 3.99331379374988e-06,
      "loss": 1.0788,
      "num_input_tokens_seen": 9846225,
      "step": 460
    },
    {
      "epoch": 0.05543197258462093,
      "flos": 16691723120640.0,
      "grad_norm": 4.539817970424568,
      "learning_rate": 3.993249999908852e-06,
      "loss": 1.006,
      "num_input_tokens_seen": 9866095,
      "step": 461
    },
    {
      "epoch": 0.05555221547526003,
      "flos": 13203662499840.0,
      "grad_norm": 5.327492058263379,
      "learning_rate": 3.993185903693384e-06,
      "loss": 1.0749,
      "num_input_tokens_seen": 9882615,
      "step": 462
    },
    {
      "epoch": 0.05567245836589912,
      "flos": 16770172047360.0,
      "grad_norm": 4.429959189612345,
      "learning_rate": 3.9931215051131995e-06,
      "loss": 1.0343,
      "num_input_tokens_seen": 9902980,
      "step": 463
    },
    {
      "epoch": 0.05579270125653821,
      "flos": 19759601049600.0,
      "grad_norm": 3.338667955950812,
      "learning_rate": 3.993056804178068e-06,
      "loss": 1.0114,
      "num_input_tokens_seen": 9924245,
      "step": 464
    },
    {
      "epoch": 0.0559129441471773,
      "flos": 19232786104320.0,
      "grad_norm": 3.978319660467446,
      "learning_rate": 3.992991800897803e-06,
      "loss": 1.0448,
      "num_input_tokens_seen": 9943770,
      "step": 465
    },
    {
      "epoch": 0.05603318703781639,
      "flos": 10763651911680.0,
      "grad_norm": 4.097882230943994,
      "learning_rate": 3.9929264952822665e-06,
      "loss": 1.0902,
      "num_input_tokens_seen": 9961025,
      "step": 466
    },
    {
      "epoch": 0.05615342992845548,
      "flos": 15820108554240.0,
      "grad_norm": 11.87056159345232,
      "learning_rate": 3.992860887341366e-06,
      "loss": 1.089,
      "num_input_tokens_seen": 9978915,
      "step": 467
    },
    {
      "epoch": 0.056273672819094574,
      "flos": 16768363192320.0,
      "grad_norm": 10.139497902983628,
      "learning_rate": 3.992794977085052e-06,
      "loss": 1.0153,
      "num_input_tokens_seen": 9996635,
      "step": 468
    },
    {
      "epoch": 0.056393915709733664,
      "flos": 14089725358080.0,
      "grad_norm": 5.464312862325239,
      "learning_rate": 3.992728764523326e-06,
      "loss": 1.0527,
      "num_input_tokens_seen": 10015300,
      "step": 469
    },
    {
      "epoch": 0.05651415860037275,
      "flos": 15747208826880.0,
      "grad_norm": 4.348616474291394,
      "learning_rate": 3.99266224966623e-06,
      "loss": 0.998,
      "num_input_tokens_seen": 10035935,
      "step": 470
    },
    {
      "epoch": 0.05663440149101184,
      "flos": 13806565662720.0,
      "grad_norm": 9.298885352494036,
      "learning_rate": 3.992595432523855e-06,
      "loss": 1.0791,
      "num_input_tokens_seen": 10052945,
      "step": 471
    },
    {
      "epoch": 0.05675464438165093,
      "flos": 16110595645440.0,
      "grad_norm": 3.9629915768853654,
      "learning_rate": 3.992528313106338e-06,
      "loss": 1.0705,
      "num_input_tokens_seen": 10070865,
      "step": 472
    },
    {
      "epoch": 0.05687488727229002,
      "flos": 11963909038080.0,
      "grad_norm": 19.173108596722617,
      "learning_rate": 3.9924608914238595e-06,
      "loss": 1.0227,
      "num_input_tokens_seen": 10085580,
      "step": 473
    },
    {
      "epoch": 0.05699513016292912,
      "flos": 20781705830400.0,
      "grad_norm": 5.977953460992073,
      "learning_rate": 3.992393167486648e-06,
      "loss": 1.0465,
      "num_input_tokens_seen": 10104450,
      "step": 474
    },
    {
      "epoch": 0.05711537305356821,
      "flos": 13413186662400.0,
      "grad_norm": 4.054924342417438,
      "learning_rate": 3.992325141304977e-06,
      "loss": 1.0094,
      "num_input_tokens_seen": 10122125,
      "step": 475
    },
    {
      "epoch": 0.0572356159442073,
      "flos": 19049421803520.0,
      "grad_norm": 5.704004841699458,
      "learning_rate": 3.992256812889166e-06,
      "loss": 1.0713,
      "num_input_tokens_seen": 10137950,
      "step": 476
    },
    {
      "epoch": 0.05735585883484639,
      "flos": 25055819304960.0,
      "grad_norm": 5.554545855426363,
      "learning_rate": 3.992188182249582e-06,
      "loss": 0.9688,
      "num_input_tokens_seen": 10159565,
      "step": 477
    },
    {
      "epoch": 0.05747610172548548,
      "flos": 13282601963520.0,
      "grad_norm": 3.9927739199771635,
      "learning_rate": 3.992119249396633e-06,
      "loss": 1.1155,
      "num_input_tokens_seen": 10177970,
      "step": 478
    },
    {
      "epoch": 0.05759634461612457,
      "flos": 19916222976000.0,
      "grad_norm": 4.355722955375979,
      "learning_rate": 3.992050014340778e-06,
      "loss": 1.031,
      "num_input_tokens_seen": 10198045,
      "step": 479
    },
    {
      "epoch": 0.057716587506763666,
      "flos": 49620763607040.0,
      "grad_norm": 1.0102854115813036,
      "learning_rate": 3.99198047709252e-06,
      "loss": 0.7737,
      "num_input_tokens_seen": 10259285,
      "step": 480
    },
    {
      "epoch": 0.057836830397402755,
      "flos": 17792092876800.0,
      "grad_norm": 4.031738724765705,
      "learning_rate": 3.991910637662408e-06,
      "loss": 0.9923,
      "num_input_tokens_seen": 10279295,
      "step": 481
    },
    {
      "epoch": 0.057957073288041845,
      "flos": 18212428861440.0,
      "grad_norm": 4.293566730702586,
      "learning_rate": 3.9918404960610355e-06,
      "loss": 1.013,
      "num_input_tokens_seen": 10298045,
      "step": 482
    },
    {
      "epoch": 0.058077316178680935,
      "flos": 14750865346560.0,
      "grad_norm": 4.512069199728076,
      "learning_rate": 3.991770052299043e-06,
      "loss": 0.9703,
      "num_input_tokens_seen": 10315995,
      "step": 483
    },
    {
      "epoch": 0.058197559069320025,
      "flos": 13413431930880.0,
      "grad_norm": 6.083311873446024,
      "learning_rate": 3.991699306387118e-06,
      "loss": 1.0788,
      "num_input_tokens_seen": 10334185,
      "step": 484
    },
    {
      "epoch": 0.058317801959959115,
      "flos": 17687269478400.0,
      "grad_norm": 3.2669622430268888,
      "learning_rate": 3.991628258335991e-06,
      "loss": 0.9977,
      "num_input_tokens_seen": 10356110,
      "step": 485
    },
    {
      "epoch": 0.05843804485059821,
      "flos": 16533537607680.0,
      "grad_norm": 6.223576038915103,
      "learning_rate": 3.991556908156442e-06,
      "loss": 1.0775,
      "num_input_tokens_seen": 10372355,
      "step": 486
    },
    {
      "epoch": 0.0585582877412373,
      "flos": 16456590950400.0,
      "grad_norm": 4.1889047117015945,
      "learning_rate": 3.9914852558592914e-06,
      "loss": 1.0728,
      "num_input_tokens_seen": 10393125,
      "step": 487
    },
    {
      "epoch": 0.05867853063187639,
      "flos": 16711138099200.0,
      "grad_norm": 6.035968733234458,
      "learning_rate": 3.991413301455413e-06,
      "loss": 0.9884,
      "num_input_tokens_seen": 10409295,
      "step": 488
    },
    {
      "epoch": 0.05879877352251548,
      "flos": 21016378122240.0,
      "grad_norm": 5.177264306981604,
      "learning_rate": 3.991341044955719e-06,
      "loss": 0.9781,
      "num_input_tokens_seen": 10428770,
      "step": 489
    },
    {
      "epoch": 0.05891901641315457,
      "flos": 14305381232640.0,
      "grad_norm": 4.28019044113721,
      "learning_rate": 3.991268486371172e-06,
      "loss": 1.0224,
      "num_input_tokens_seen": 10447045,
      "step": 490
    },
    {
      "epoch": 0.05903925930379366,
      "flos": 17529758453760.0,
      "grad_norm": 21.91046859456007,
      "learning_rate": 3.991195625712779e-06,
      "loss": 1.0631,
      "num_input_tokens_seen": 10463730,
      "step": 491
    },
    {
      "epoch": 0.05915950219443276,
      "flos": 15090606305280.0,
      "grad_norm": 4.429701924510796,
      "learning_rate": 3.991122462991592e-06,
      "loss": 1.013,
      "num_input_tokens_seen": 10482970,
      "step": 492
    },
    {
      "epoch": 0.05927974508507185,
      "flos": 6935175843840.0,
      "grad_norm": 443.5403377935315,
      "learning_rate": 3.991048998218712e-06,
      "loss": 1.0055,
      "num_input_tokens_seen": 10495995,
      "step": 493
    },
    {
      "epoch": 0.05939998797571094,
      "flos": 12941236101120.0,
      "grad_norm": 8.124661013060695,
      "learning_rate": 3.990975231405281e-06,
      "loss": 0.9551,
      "num_input_tokens_seen": 10514165,
      "step": 494
    },
    {
      "epoch": 0.05952023086635003,
      "flos": 20125961748480.0,
      "grad_norm": 9.64988049052657,
      "learning_rate": 3.990901162562491e-06,
      "loss": 0.9973,
      "num_input_tokens_seen": 10534575,
      "step": 495
    },
    {
      "epoch": 0.05964047375698912,
      "flos": 10528581058560.0,
      "grad_norm": 4.080774141058536,
      "learning_rate": 3.9908267917015765e-06,
      "loss": 1.0961,
      "num_input_tokens_seen": 10552355,
      "step": 496
    },
    {
      "epoch": 0.059760716647628206,
      "flos": 16481585786880.0,
      "grad_norm": 11.744530886216983,
      "learning_rate": 3.990752118833821e-06,
      "loss": 1.13,
      "num_input_tokens_seen": 10569515,
      "step": 497
    },
    {
      "epoch": 0.0598809595382673,
      "flos": 16166257152000.0,
      "grad_norm": 7.127018692747855,
      "learning_rate": 3.990677143970553e-06,
      "loss": 0.9853,
      "num_input_tokens_seen": 10590045,
      "step": 498
    },
    {
      "epoch": 0.06000120242890639,
      "flos": 15720926330880.0,
      "grad_norm": 6.39429441280253,
      "learning_rate": 3.990601867123144e-06,
      "loss": 1.012,
      "num_input_tokens_seen": 10609490,
      "step": 499
    },
    {
      "epoch": 0.06012144531954548,
      "flos": 13596888207360.0,
      "grad_norm": 7.827844550006871,
      "learning_rate": 3.990526288303014e-06,
      "loss": 1.0551,
      "num_input_tokens_seen": 10628000,
      "step": 500
    },
    {
      "epoch": 0.06024168821018457,
      "flos": 16192386355200.0,
      "grad_norm": 10.039202333656482,
      "learning_rate": 3.9904504075216295e-06,
      "loss": 1.1052,
      "num_input_tokens_seen": 10648480,
      "step": 501
    },
    {
      "epoch": 0.06036193110082366,
      "flos": 13308669849600.0,
      "grad_norm": 5.811871566435061,
      "learning_rate": 3.990374224790501e-06,
      "loss": 1.1492,
      "num_input_tokens_seen": 10666405,
      "step": 502
    },
    {
      "epoch": 0.06048217399146275,
      "flos": 12180944547840.0,
      "grad_norm": 14.524796428993797,
      "learning_rate": 3.990297740121185e-06,
      "loss": 0.9125,
      "num_input_tokens_seen": 10684060,
      "step": 503
    },
    {
      "epoch": 0.06060241688210185,
      "flos": 17084151705600.0,
      "grad_norm": 4.6819182668898565,
      "learning_rate": 3.990220953525284e-06,
      "loss": 0.9814,
      "num_input_tokens_seen": 10700890,
      "step": 504
    },
    {
      "epoch": 0.06072265977274094,
      "flos": 10319302164480.0,
      "grad_norm": 4.771526347415681,
      "learning_rate": 3.9901438650144465e-06,
      "loss": 0.9462,
      "num_input_tokens_seen": 10716860,
      "step": 505
    },
    {
      "epoch": 0.06084290266338003,
      "flos": 14593354321920.0,
      "grad_norm": 4.564201978140935,
      "learning_rate": 3.990066474600367e-06,
      "loss": 1.1105,
      "num_input_tokens_seen": 10734550,
      "step": 506
    },
    {
      "epoch": 0.06096314555401912,
      "flos": 15851664322560.0,
      "grad_norm": 5.2440818854368585,
      "learning_rate": 3.989988782294786e-06,
      "loss": 0.8848,
      "num_input_tokens_seen": 10754360,
      "step": 507
    },
    {
      "epoch": 0.06108338844465821,
      "flos": 13567325245440.0,
      "grad_norm": 4.276161901508192,
      "learning_rate": 3.989910788109489e-06,
      "loss": 1.1472,
      "num_input_tokens_seen": 10770730,
      "step": 508
    },
    {
      "epoch": 0.0612036313352973,
      "flos": 23954744401920.0,
      "grad_norm": 5.542643391668989,
      "learning_rate": 3.989832492056307e-06,
      "loss": 0.9419,
      "num_input_tokens_seen": 10791475,
      "step": 509
    },
    {
      "epoch": 0.06132387422593639,
      "flos": 19575439626240.0,
      "grad_norm": 4.847432760762875,
      "learning_rate": 3.989753894147119e-06,
      "loss": 1.0021,
      "num_input_tokens_seen": 10811320,
      "step": 510
    },
    {
      "epoch": 0.061444117116575485,
      "flos": 18421585121280.0,
      "grad_norm": 6.041437429705038,
      "learning_rate": 3.989674994393846e-06,
      "loss": 0.9955,
      "num_input_tokens_seen": 10830515,
      "step": 511
    },
    {
      "epoch": 0.061564360007214575,
      "flos": 20309540659200.0,
      "grad_norm": 2.9532262391190733,
      "learning_rate": 3.98959579280846e-06,
      "loss": 1.1441,
      "num_input_tokens_seen": 10848635,
      "step": 512
    },
    {
      "epoch": 0.061684602897853665,
      "flos": 8510592675840.0,
      "grad_norm": 4.312627154700933,
      "learning_rate": 3.989516289402973e-06,
      "loss": 1.0279,
      "num_input_tokens_seen": 10863985,
      "step": 513
    },
    {
      "epoch": 0.061804845788492754,
      "flos": 13855696896000.0,
      "grad_norm": 6.010720847469226,
      "learning_rate": 3.989436484189447e-06,
      "loss": 1.019,
      "num_input_tokens_seen": 10881650,
      "step": 514
    },
    {
      "epoch": 0.061925088679131844,
      "flos": 10843909693440.0,
      "grad_norm": 5.517197338906814,
      "learning_rate": 3.9893563771799885e-06,
      "loss": 1.0013,
      "num_input_tokens_seen": 10897845,
      "step": 515
    },
    {
      "epoch": 0.062045331569770934,
      "flos": 18446426664960.0,
      "grad_norm": 6.8555065882146655,
      "learning_rate": 3.989275968386749e-06,
      "loss": 1.0635,
      "num_input_tokens_seen": 10915475,
      "step": 516
    },
    {
      "epoch": 0.06216557446041003,
      "flos": 20021046374400.0,
      "grad_norm": 5.543726411616601,
      "learning_rate": 3.989195257821926e-06,
      "loss": 0.9715,
      "num_input_tokens_seen": 10933680,
      "step": 517
    },
    {
      "epoch": 0.06228581735104912,
      "flos": 16691968389120.0,
      "grad_norm": 20.230891501544505,
      "learning_rate": 3.989114245497765e-06,
      "loss": 1.0415,
      "num_input_tokens_seen": 10953200,
      "step": 518
    },
    {
      "epoch": 0.06240606024168821,
      "flos": 10739055636480.0,
      "grad_norm": 4.409957300096212,
      "learning_rate": 3.989032931426554e-06,
      "loss": 1.1511,
      "num_input_tokens_seen": 10970075,
      "step": 519
    },
    {
      "epoch": 0.06252630313232731,
      "flos": 14646287216640.0,
      "grad_norm": 4.708570603775343,
      "learning_rate": 3.9889513156206295e-06,
      "loss": 1.0683,
      "num_input_tokens_seen": 10989235,
      "step": 520
    },
    {
      "epoch": 0.06264654602296639,
      "flos": 14751417200640.0,
      "grad_norm": 5.414158851900117,
      "learning_rate": 3.988869398092371e-06,
      "loss": 0.8999,
      "num_input_tokens_seen": 11008865,
      "step": 521
    },
    {
      "epoch": 0.06276678891360549,
      "flos": 21095501537280.0,
      "grad_norm": 8.145594392542051,
      "learning_rate": 3.988787178854206e-06,
      "loss": 0.9734,
      "num_input_tokens_seen": 11028120,
      "step": 522
    },
    {
      "epoch": 0.06288703180424457,
      "flos": 15720129208320.0,
      "grad_norm": 6.880766236419033,
      "learning_rate": 3.988704657918608e-06,
      "loss": 1.0732,
      "num_input_tokens_seen": 11046900,
      "step": 523
    },
    {
      "epoch": 0.06300727469488367,
      "flos": 10582065807360.0,
      "grad_norm": 7.1557086589906325,
      "learning_rate": 3.988621835298094e-06,
      "loss": 1.0144,
      "num_input_tokens_seen": 11063835,
      "step": 524
    },
    {
      "epoch": 0.06312751758552275,
      "flos": 17451646771200.0,
      "grad_norm": 5.992608603908541,
      "learning_rate": 3.988538711005229e-06,
      "loss": 1.1295,
      "num_input_tokens_seen": 11083010,
      "step": 525
    },
    {
      "epoch": 0.06324776047616185,
      "flos": 15275196948480.0,
      "grad_norm": 8.231315956633525,
      "learning_rate": 3.988455285052622e-06,
      "loss": 1.096,
      "num_input_tokens_seen": 11098910,
      "step": 526
    },
    {
      "epoch": 0.06336800336680094,
      "flos": 15405107159040.0,
      "grad_norm": 5.220100091142683,
      "learning_rate": 3.98837155745293e-06,
      "loss": 1.0333,
      "num_input_tokens_seen": 11116670,
      "step": 527
    },
    {
      "epoch": 0.06348824625744003,
      "flos": 13833522647040.0,
      "grad_norm": 23.451031250229036,
      "learning_rate": 3.988287528218854e-06,
      "loss": 0.9641,
      "num_input_tokens_seen": 11135175,
      "step": 528
    },
    {
      "epoch": 0.06360848914807912,
      "flos": 10943827722240.0,
      "grad_norm": 16.063782399561717,
      "learning_rate": 3.98820319736314e-06,
      "loss": 1.0959,
      "num_input_tokens_seen": 11151510,
      "step": 529
    },
    {
      "epoch": 0.0637287320387182,
      "flos": 14618379816960.0,
      "grad_norm": 43.55029896612723,
      "learning_rate": 3.988118564898582e-06,
      "loss": 1.0557,
      "num_input_tokens_seen": 11170770,
      "step": 530
    },
    {
      "epoch": 0.0638489749293573,
      "flos": 12331220152320.0,
      "grad_norm": 16.148251352035476,
      "learning_rate": 3.988033630838019e-06,
      "loss": 1.0988,
      "num_input_tokens_seen": 11184530,
      "step": 531
    },
    {
      "epoch": 0.0639692178199964,
      "flos": 16928909414400.0,
      "grad_norm": 4.668833887860707,
      "learning_rate": 3.987948395194334e-06,
      "loss": 1.0794,
      "num_input_tokens_seen": 11206630,
      "step": 532
    },
    {
      "epoch": 0.06408946071063548,
      "flos": 13097551441920.0,
      "grad_norm": 3.981927576300251,
      "learning_rate": 3.987862857980458e-06,
      "loss": 0.9832,
      "num_input_tokens_seen": 11222295,
      "step": 533
    },
    {
      "epoch": 0.06420970360127458,
      "flos": 19339479674880.0,
      "grad_norm": 3.3981906845089918,
      "learning_rate": 3.987777019209368e-06,
      "loss": 0.9635,
      "num_input_tokens_seen": 11242530,
      "step": 534
    },
    {
      "epoch": 0.06432994649191366,
      "flos": 16822277160960.0,
      "grad_norm": 3.0728089566939203,
      "learning_rate": 3.987690878894084e-06,
      "loss": 1.0153,
      "num_input_tokens_seen": 11261965,
      "step": 535
    },
    {
      "epoch": 0.06445018938255276,
      "flos": 16638452981760.0,
      "grad_norm": 5.099256289241515,
      "learning_rate": 3.987604437047673e-06,
      "loss": 1.0455,
      "num_input_tokens_seen": 11281485,
      "step": 536
    },
    {
      "epoch": 0.06457043227319184,
      "flos": 13938284728320.0,
      "grad_norm": 8.686378461497473,
      "learning_rate": 3.987517693683251e-06,
      "loss": 0.9842,
      "num_input_tokens_seen": 11299780,
      "step": 537
    },
    {
      "epoch": 0.06469067516383094,
      "flos": 12020275691520.0,
      "grad_norm": 4.869423439941444,
      "learning_rate": 3.9874306488139745e-06,
      "loss": 1.1594,
      "num_input_tokens_seen": 11314760,
      "step": 538
    },
    {
      "epoch": 0.06481091805447003,
      "flos": 16561414348800.0,
      "grad_norm": 6.68157411222741,
      "learning_rate": 3.987343302453049e-06,
      "loss": 1.0702,
      "num_input_tokens_seen": 11335755,
      "step": 539
    },
    {
      "epoch": 0.06493116094510912,
      "flos": 20784587735040.0,
      "grad_norm": 8.816770127764466,
      "learning_rate": 3.987255654613724e-06,
      "loss": 1.0209,
      "num_input_tokens_seen": 11359240,
      "step": 540
    },
    {
      "epoch": 0.06505140383574821,
      "flos": 14095151923200.0,
      "grad_norm": 12.629500087530682,
      "learning_rate": 3.987167705309296e-06,
      "loss": 0.9055,
      "num_input_tokens_seen": 11378235,
      "step": 541
    },
    {
      "epoch": 0.0651716467263873,
      "flos": 12699757608960.0,
      "grad_norm": 3.596923223283246,
      "learning_rate": 3.987079454553108e-06,
      "loss": 1.1531,
      "num_input_tokens_seen": 11395905,
      "step": 542
    },
    {
      "epoch": 0.0652918896170264,
      "flos": 14796869406720.0,
      "grad_norm": 5.073596425764304,
      "learning_rate": 3.986990902358546e-06,
      "loss": 1.1223,
      "num_input_tokens_seen": 11412565,
      "step": 543
    },
    {
      "epoch": 0.06541213250766549,
      "flos": 15537960591360.0,
      "grad_norm": 6.8658119533888256,
      "learning_rate": 3.986902048739045e-06,
      "loss": 1.129,
      "num_input_tokens_seen": 11432230,
      "step": 544
    },
    {
      "epoch": 0.06553237539830457,
      "flos": 16427549184000.0,
      "grad_norm": 7.912340313835304,
      "learning_rate": 3.986812893708082e-06,
      "loss": 0.9931,
      "num_input_tokens_seen": 11448140,
      "step": 545
    },
    {
      "epoch": 0.06565261828894367,
      "flos": 12698960486400.0,
      "grad_norm": 4.632967709905325,
      "learning_rate": 3.9867234372791826e-06,
      "loss": 1.0187,
      "num_input_tokens_seen": 11465815,
      "step": 546
    },
    {
      "epoch": 0.06577286117958275,
      "flos": 16192141086720.0,
      "grad_norm": 3.3259942482756926,
      "learning_rate": 3.986633679465918e-06,
      "loss": 1.0731,
      "num_input_tokens_seen": 11485690,
      "step": 547
    },
    {
      "epoch": 0.06589310407022185,
      "flos": 16848528998400.0,
      "grad_norm": 9.571823860253527,
      "learning_rate": 3.986543620281904e-06,
      "loss": 1.0157,
      "num_input_tokens_seen": 11505060,
      "step": 548
    },
    {
      "epoch": 0.06601334696086093,
      "flos": 19124958167040.0,
      "grad_norm": 3.216748254034497,
      "learning_rate": 3.986453259740802e-06,
      "loss": 1.1165,
      "num_input_tokens_seen": 11522950,
      "step": 549
    },
    {
      "epoch": 0.06613358985150003,
      "flos": 8850640220160.0,
      "grad_norm": 6.900300790616056,
      "learning_rate": 3.986362597856319e-06,
      "loss": 0.9774,
      "num_input_tokens_seen": 11539170,
      "step": 550
    },
    {
      "epoch": 0.06625383274213913,
      "flos": 12993709117440.0,
      "grad_norm": 8.39452443360007,
      "learning_rate": 3.986271634642211e-06,
      "loss": 1.0087,
      "num_input_tokens_seen": 11555870,
      "step": 551
    },
    {
      "epoch": 0.06637407563277821,
      "flos": 10868413992960.0,
      "grad_norm": 6.933913945071445,
      "learning_rate": 3.986180370112274e-06,
      "loss": 1.0211,
      "num_input_tokens_seen": 11572110,
      "step": 552
    },
    {
      "epoch": 0.0664943185234173,
      "flos": 17085102120960.0,
      "grad_norm": 5.691885462623866,
      "learning_rate": 3.986088804280354e-06,
      "loss": 0.948,
      "num_input_tokens_seen": 11591560,
      "step": 553
    },
    {
      "epoch": 0.06661456141405639,
      "flos": 14879978434560.0,
      "grad_norm": 8.048148222997384,
      "learning_rate": 3.985996937160342e-06,
      "loss": 1.1349,
      "num_input_tokens_seen": 11610470,
      "step": 554
    },
    {
      "epoch": 0.06673480430469549,
      "flos": 37350333050880.0,
      "grad_norm": 5.488518261118998,
      "learning_rate": 3.985904768766173e-06,
      "loss": 0.8929,
      "num_input_tokens_seen": 11632965,
      "step": 555
    },
    {
      "epoch": 0.06685504719533458,
      "flos": 11473493913600.0,
      "grad_norm": 10.968921105770129,
      "learning_rate": 3.98581229911183e-06,
      "loss": 0.9478,
      "num_input_tokens_seen": 11651605,
      "step": 556
    },
    {
      "epoch": 0.06697529008597367,
      "flos": 15982678241280.0,
      "grad_norm": 16.011771640099276,
      "learning_rate": 3.985719528211341e-06,
      "loss": 1.1118,
      "num_input_tokens_seen": 11670695,
      "step": 557
    },
    {
      "epoch": 0.06709553297661276,
      "flos": 44874760888320.0,
      "grad_norm": 0.9654596501971237,
      "learning_rate": 3.985626456078777e-06,
      "loss": 0.8731,
      "num_input_tokens_seen": 11735070,
      "step": 558
    },
    {
      "epoch": 0.06721577586725185,
      "flos": 7879475527680.0,
      "grad_norm": 4.959771831863388,
      "learning_rate": 3.985533082728259e-06,
      "loss": 1.0636,
      "num_input_tokens_seen": 11750445,
      "step": 559
    },
    {
      "epoch": 0.06733601875789094,
      "flos": 18448358154240.0,
      "grad_norm": 3.8235306554248565,
      "learning_rate": 3.985439408173951e-06,
      "loss": 0.944,
      "num_input_tokens_seen": 11770390,
      "step": 560
    },
    {
      "epoch": 0.06745626164853002,
      "flos": 14776871915520.0,
      "grad_norm": 6.659881890426706,
      "learning_rate": 3.9853454324300634e-06,
      "loss": 0.9052,
      "num_input_tokens_seen": 11789320,
      "step": 561
    },
    {
      "epoch": 0.06757650453916912,
      "flos": 14069513256960.0,
      "grad_norm": 5.054565994240922,
      "learning_rate": 3.985251155510852e-06,
      "loss": 0.9765,
      "num_input_tokens_seen": 11808070,
      "step": 562
    },
    {
      "epoch": 0.06769674742980822,
      "flos": 18316792381440.0,
      "grad_norm": 6.445217506402246,
      "learning_rate": 3.98515657743062e-06,
      "loss": 1.0155,
      "num_input_tokens_seen": 11827255,
      "step": 563
    },
    {
      "epoch": 0.0678169903204473,
      "flos": 9214578892800.0,
      "grad_norm": 4.101662247671025,
      "learning_rate": 3.985061698203711e-06,
      "loss": 0.9802,
      "num_input_tokens_seen": 11844090,
      "step": 564
    },
    {
      "epoch": 0.0679372332110864,
      "flos": 50752505180160.0,
      "grad_norm": 0.9163050407853857,
      "learning_rate": 3.984966517844523e-06,
      "loss": 0.8676,
      "num_input_tokens_seen": 11899055,
      "step": 565
    },
    {
      "epoch": 0.06805747610172548,
      "flos": 20125808455680.0,
      "grad_norm": 6.051558541302281,
      "learning_rate": 3.984871036367492e-06,
      "loss": 1.0181,
      "num_input_tokens_seen": 11918800,
      "step": 566
    },
    {
      "epoch": 0.06817771899236458,
      "flos": 14278761492480.0,
      "grad_norm": 7.339325306414013,
      "learning_rate": 3.984775253787102e-06,
      "loss": 1.0406,
      "num_input_tokens_seen": 11936810,
      "step": 567
    },
    {
      "epoch": 0.06829796188300366,
      "flos": 12704816271360.0,
      "grad_norm": 6.704763785688129,
      "learning_rate": 3.984679170117885e-06,
      "loss": 1.067,
      "num_input_tokens_seen": 11952735,
      "step": 568
    },
    {
      "epoch": 0.06841820477364276,
      "flos": 10240546652160.0,
      "grad_norm": 7.41624838117424,
      "learning_rate": 3.984582785374415e-06,
      "loss": 0.9864,
      "num_input_tokens_seen": 11969895,
      "step": 569
    },
    {
      "epoch": 0.06853844766428185,
      "flos": 15584424529920.0,
      "grad_norm": 7.126732331209243,
      "learning_rate": 3.9844860995713155e-06,
      "loss": 1.0183,
      "num_input_tokens_seen": 11989155,
      "step": 570
    },
    {
      "epoch": 0.06865869055492094,
      "flos": 11892848824320.0,
      "grad_norm": 4.509650161606566,
      "learning_rate": 3.9843891127232524e-06,
      "loss": 1.0084,
      "num_input_tokens_seen": 12006410,
      "step": 571
    },
    {
      "epoch": 0.06877893344556003,
      "flos": 14146950451200.0,
      "grad_norm": 5.337735366098714,
      "learning_rate": 3.984291824844938e-06,
      "loss": 0.872,
      "num_input_tokens_seen": 12021225,
      "step": 572
    },
    {
      "epoch": 0.06889917633619912,
      "flos": 16376486461440.0,
      "grad_norm": 4.895138543791691,
      "learning_rate": 3.984194235951132e-06,
      "loss": 1.0374,
      "num_input_tokens_seen": 12037090,
      "step": 573
    },
    {
      "epoch": 0.06901941922683821,
      "flos": 14882584412160.0,
      "grad_norm": 6.243115715365944,
      "learning_rate": 3.9840963460566375e-06,
      "loss": 1.0277,
      "num_input_tokens_seen": 12055590,
      "step": 574
    },
    {
      "epoch": 0.06913966211747731,
      "flos": 17656786759680.0,
      "grad_norm": 3.6416540697055946,
      "learning_rate": 3.983998155176305e-06,
      "loss": 1.094,
      "num_input_tokens_seen": 12075670,
      "step": 575
    },
    {
      "epoch": 0.06925990500811639,
      "flos": 41772345200640.0,
      "grad_norm": 0.8838212992653176,
      "learning_rate": 3.9838996633250305e-06,
      "loss": 0.794,
      "num_input_tokens_seen": 12135905,
      "step": 576
    },
    {
      "epoch": 0.06938014789875549,
      "flos": 8981592821760.0,
      "grad_norm": 5.93223579342469,
      "learning_rate": 3.983800870517753e-06,
      "loss": 1.0883,
      "num_input_tokens_seen": 12152415,
      "step": 577
    },
    {
      "epoch": 0.06950039078939457,
      "flos": 16218944778240.0,
      "grad_norm": 6.810808164169734,
      "learning_rate": 3.983701776769463e-06,
      "loss": 0.9914,
      "num_input_tokens_seen": 12169545,
      "step": 578
    },
    {
      "epoch": 0.06962063368003367,
      "flos": 15586877214720.0,
      "grad_norm": 3.7945979319245167,
      "learning_rate": 3.9836023820951885e-06,
      "loss": 1.0582,
      "num_input_tokens_seen": 12188480,
      "step": 579
    },
    {
      "epoch": 0.06974087657067275,
      "flos": 14699771965440.0,
      "grad_norm": 4.023240760511483,
      "learning_rate": 3.983502686510011e-06,
      "loss": 0.8907,
      "num_input_tokens_seen": 12209030,
      "step": 580
    },
    {
      "epoch": 0.06986111946131185,
      "flos": 16088206786560.0,
      "grad_norm": 3.1681288973855244,
      "learning_rate": 3.9834026900290525e-06,
      "loss": 0.9326,
      "num_input_tokens_seen": 12228145,
      "step": 581
    },
    {
      "epoch": 0.06998136235195095,
      "flos": 19182275235840.0,
      "grad_norm": 4.981295312845286,
      "learning_rate": 3.983302392667482e-06,
      "loss": 1.1969,
      "num_input_tokens_seen": 12248710,
      "step": 582
    },
    {
      "epoch": 0.07010160524259003,
      "flos": 15983506022400.0,
      "grad_norm": 5.129914509319312,
      "learning_rate": 3.983201794440517e-06,
      "loss": 1.1431,
      "num_input_tokens_seen": 12268005,
      "step": 583
    },
    {
      "epoch": 0.07022184813322913,
      "flos": 12993739776000.0,
      "grad_norm": 3.715640523122536,
      "learning_rate": 3.9831008953634165e-06,
      "loss": 0.8741,
      "num_input_tokens_seen": 12287015,
      "step": 584
    },
    {
      "epoch": 0.07034209102386821,
      "flos": 17551595458560.0,
      "grad_norm": 4.89501483108409,
      "learning_rate": 3.9829996954514864e-06,
      "loss": 1.0178,
      "num_input_tokens_seen": 12305875,
      "step": 585
    },
    {
      "epoch": 0.0704623339145073,
      "flos": 18501413683200.0,
      "grad_norm": 3.855826163223775,
      "learning_rate": 3.982898194720079e-06,
      "loss": 1.0465,
      "num_input_tokens_seen": 12326325,
      "step": 586
    },
    {
      "epoch": 0.0705825768051464,
      "flos": 18028727316480.0,
      "grad_norm": 3.9907493912466303,
      "learning_rate": 3.982796393184592e-06,
      "loss": 1.0089,
      "num_input_tokens_seen": 12345125,
      "step": 587
    },
    {
      "epoch": 0.07070281969578548,
      "flos": 47289408737280.0,
      "grad_norm": 0.8899783078775484,
      "learning_rate": 3.98269429086047e-06,
      "loss": 0.852,
      "num_input_tokens_seen": 12402685,
      "step": 588
    },
    {
      "epoch": 0.07082306258642458,
      "flos": 16817065205760.0,
      "grad_norm": 5.4009438643494985,
      "learning_rate": 3.982591887763199e-06,
      "loss": 1.0534,
      "num_input_tokens_seen": 12419865,
      "step": 589
    },
    {
      "epoch": 0.07094330547706366,
      "flos": 9793652244480.0,
      "grad_norm": 4.804482655694283,
      "learning_rate": 3.982489183908316e-06,
      "loss": 1.0183,
      "num_input_tokens_seen": 12436005,
      "step": 590
    },
    {
      "epoch": 0.07106354836770276,
      "flos": 17530524917760.0,
      "grad_norm": 12.531885601866993,
      "learning_rate": 3.982386179311399e-06,
      "loss": 1.047,
      "num_input_tokens_seen": 12456245,
      "step": 591
    },
    {
      "epoch": 0.07118379125834184,
      "flos": 11473616547840.0,
      "grad_norm": 4.564728702033555,
      "learning_rate": 3.982282873988075e-06,
      "loss": 1.0676,
      "num_input_tokens_seen": 12473840,
      "step": 592
    },
    {
      "epoch": 0.07130403414898094,
      "flos": 13991003013120.0,
      "grad_norm": 4.055743893876294,
      "learning_rate": 3.982179267954016e-06,
      "loss": 1.0781,
      "num_input_tokens_seen": 12493990,
      "step": 593
    },
    {
      "epoch": 0.07142427703962004,
      "flos": 15537470054400.0,
      "grad_norm": 6.029061824763153,
      "learning_rate": 3.982075361224937e-06,
      "loss": 1.168,
      "num_input_tokens_seen": 12512075,
      "step": 594
    },
    {
      "epoch": 0.07154451993025912,
      "flos": 12967978475520.0,
      "grad_norm": 3.7933785622777023,
      "learning_rate": 3.981971153816602e-06,
      "loss": 1.0978,
      "num_input_tokens_seen": 12529400,
      "step": 595
    },
    {
      "epoch": 0.07166476282089822,
      "flos": 15744572190720.0,
      "grad_norm": 2.951047136055056,
      "learning_rate": 3.981866645744819e-06,
      "loss": 1.1691,
      "num_input_tokens_seen": 12549835,
      "step": 596
    },
    {
      "epoch": 0.0717850057115373,
      "flos": 9977966960640.0,
      "grad_norm": 4.742173623782725,
      "learning_rate": 3.9817618370254416e-06,
      "loss": 1.0133,
      "num_input_tokens_seen": 12566210,
      "step": 597
    },
    {
      "epoch": 0.0719052486021764,
      "flos": 22038206976000.0,
      "grad_norm": 5.04901663393245,
      "learning_rate": 3.9816567276743684e-06,
      "loss": 1.0768,
      "num_input_tokens_seen": 12585795,
      "step": 598
    },
    {
      "epoch": 0.0720254914928155,
      "flos": 15116950118400.0,
      "grad_norm": 2.900922917801232,
      "learning_rate": 3.9815513177075466e-06,
      "loss": 0.9591,
      "num_input_tokens_seen": 12604300,
      "step": 599
    },
    {
      "epoch": 0.07214573438345458,
      "flos": 19811644846080.0,
      "grad_norm": 2.549380047553046,
      "learning_rate": 3.9814456071409646e-06,
      "loss": 0.9113,
      "num_input_tokens_seen": 12624555,
      "step": 600
    },
    {
      "epoch": 0.07226597727409367,
      "flos": 18133029519360.0,
      "grad_norm": 9.435698684674534,
      "learning_rate": 3.981339595990659e-06,
      "loss": 1.0547,
      "num_input_tokens_seen": 12642805,
      "step": 601
    },
    {
      "epoch": 0.07238622016473276,
      "flos": 16717484421120.0,
      "grad_norm": 4.012192502227025,
      "learning_rate": 3.981233284272713e-06,
      "loss": 1.0198,
      "num_input_tokens_seen": 12662270,
      "step": 602
    },
    {
      "epoch": 0.07250646305537185,
      "flos": 18112296222720.0,
      "grad_norm": 6.497730739322936,
      "learning_rate": 3.981126672003253e-06,
      "loss": 1.1005,
      "num_input_tokens_seen": 12684665,
      "step": 603
    },
    {
      "epoch": 0.07262670594601094,
      "flos": 19333347962880.0,
      "grad_norm": 3.5608732290522993,
      "learning_rate": 3.981019759198451e-06,
      "loss": 0.9833,
      "num_input_tokens_seen": 12703335,
      "step": 604
    },
    {
      "epoch": 0.07274694883665003,
      "flos": 18972536463360.0,
      "grad_norm": 4.196936607334727,
      "learning_rate": 3.980912545874528e-06,
      "loss": 1.0257,
      "num_input_tokens_seen": 12723220,
      "step": 605
    },
    {
      "epoch": 0.07286719172728913,
      "flos": 21280030863360.0,
      "grad_norm": 13.34975096509587,
      "learning_rate": 3.980805032047746e-06,
      "loss": 1.0573,
      "num_input_tokens_seen": 12744410,
      "step": 606
    },
    {
      "epoch": 0.07298743461792821,
      "flos": 12310394880000.0,
      "grad_norm": 4.859306702946499,
      "learning_rate": 3.980697217734415e-06,
      "loss": 0.9996,
      "num_input_tokens_seen": 12761870,
      "step": 607
    },
    {
      "epoch": 0.07310767750856731,
      "flos": 13831223255040.0,
      "grad_norm": 3.997386078299715,
      "learning_rate": 3.980589102950891e-06,
      "loss": 1.1112,
      "num_input_tokens_seen": 12779755,
      "step": 608
    },
    {
      "epoch": 0.07322792039920639,
      "flos": 20781460561920.0,
      "grad_norm": 4.650091822548226,
      "learning_rate": 3.9804806877135755e-06,
      "loss": 0.9817,
      "num_input_tokens_seen": 12797520,
      "step": 609
    },
    {
      "epoch": 0.07334816328984549,
      "flos": 16691661803520.0,
      "grad_norm": 8.26854770957911,
      "learning_rate": 3.980371972038915e-06,
      "loss": 1.0693,
      "num_input_tokens_seen": 12817730,
      "step": 610
    },
    {
      "epoch": 0.07346840618048459,
      "flos": 16320671662080.0,
      "grad_norm": 4.650168114361167,
      "learning_rate": 3.980262955943399e-06,
      "loss": 1.0388,
      "num_input_tokens_seen": 12837115,
      "step": 611
    },
    {
      "epoch": 0.07358864907112367,
      "flos": 12520133652480.0,
      "grad_norm": 4.3473206259437385,
      "learning_rate": 3.980153639443569e-06,
      "loss": 1.087,
      "num_input_tokens_seen": 12852820,
      "step": 612
    },
    {
      "epoch": 0.07370889196176277,
      "flos": 17136532746240.0,
      "grad_norm": 7.224279886545814,
      "learning_rate": 3.980044022556005e-06,
      "loss": 1.008,
      "num_input_tokens_seen": 12872225,
      "step": 613
    },
    {
      "epoch": 0.07382913485240185,
      "flos": 18422320926720.0,
      "grad_norm": 7.918357534701453,
      "learning_rate": 3.9799341052973375e-06,
      "loss": 0.932,
      "num_input_tokens_seen": 12891780,
      "step": 614
    },
    {
      "epoch": 0.07394937774304094,
      "flos": 11945505792000.0,
      "grad_norm": 11.092930207386031,
      "learning_rate": 3.979823887684241e-06,
      "loss": 0.9352,
      "num_input_tokens_seen": 12910440,
      "step": 615
    },
    {
      "epoch": 0.07406962063368003,
      "flos": 14697595207680.0,
      "grad_norm": 6.5142853089143635,
      "learning_rate": 3.979713369733434e-06,
      "loss": 1.0624,
      "num_input_tokens_seen": 12928025,
      "step": 616
    },
    {
      "epoch": 0.07418986352431912,
      "flos": 15220148613120.0,
      "grad_norm": 29.71202933998282,
      "learning_rate": 3.979602551461683e-06,
      "loss": 1.0452,
      "num_input_tokens_seen": 12948525,
      "step": 617
    },
    {
      "epoch": 0.07431010641495822,
      "flos": 8457935708160.0,
      "grad_norm": 17.7475309132179,
      "learning_rate": 3.979491432885799e-06,
      "loss": 1.1289,
      "num_input_tokens_seen": 12964510,
      "step": 618
    },
    {
      "epoch": 0.0744303493055973,
      "flos": 14880039751680.0,
      "grad_norm": 4.22045163345778,
      "learning_rate": 3.97938001402264e-06,
      "loss": 1.0336,
      "num_input_tokens_seen": 12983355,
      "step": 619
    },
    {
      "epoch": 0.0745505921962364,
      "flos": 11499347189760.0,
      "grad_norm": 7.1895998333815605,
      "learning_rate": 3.979268294889105e-06,
      "loss": 1.0035,
      "num_input_tokens_seen": 12998625,
      "step": 620
    },
    {
      "epoch": 0.07467083508687548,
      "flos": 36430598983680.0,
      "grad_norm": 7.68746377579753,
      "learning_rate": 3.979156275502143e-06,
      "loss": 0.9558,
      "num_input_tokens_seen": 13022005,
      "step": 621
    },
    {
      "epoch": 0.07479107797751458,
      "flos": 12416720547840.0,
      "grad_norm": 5.8542194115196,
      "learning_rate": 3.979043955878749e-06,
      "loss": 1.1175,
      "num_input_tokens_seen": 13039570,
      "step": 622
    },
    {
      "epoch": 0.07491132086815366,
      "flos": 16688350679040.0,
      "grad_norm": 7.140566769285271,
      "learning_rate": 3.978931336035959e-06,
      "loss": 1.0344,
      "num_input_tokens_seen": 13058100,
      "step": 623
    },
    {
      "epoch": 0.07503156375879276,
      "flos": 14305442549760.0,
      "grad_norm": 12.511807167772194,
      "learning_rate": 3.9788184159908595e-06,
      "loss": 1.0237,
      "num_input_tokens_seen": 13074950,
      "step": 624
    },
    {
      "epoch": 0.07515180664943186,
      "flos": 10680880128000.0,
      "grad_norm": 4.415111099867128,
      "learning_rate": 3.97870519576058e-06,
      "loss": 1.0146,
      "num_input_tokens_seen": 13091095,
      "step": 625
    },
    {
      "epoch": 0.07527204954007094,
      "flos": 15117532631040.0,
      "grad_norm": 4.650743338944467,
      "learning_rate": 3.978591675362295e-06,
      "loss": 0.9987,
      "num_input_tokens_seen": 13109530,
      "step": 626
    },
    {
      "epoch": 0.07539229243071004,
      "flos": 15143753809920.0,
      "grad_norm": 4.297585612404628,
      "learning_rate": 3.978477854813226e-06,
      "loss": 1.0759,
      "num_input_tokens_seen": 13128590,
      "step": 627
    },
    {
      "epoch": 0.07551253532134912,
      "flos": 9191975424000.0,
      "grad_norm": 5.4373930120356375,
      "learning_rate": 3.97836373413064e-06,
      "loss": 1.0252,
      "num_input_tokens_seen": 13146365,
      "step": 628
    },
    {
      "epoch": 0.07563277821198822,
      "flos": 13623538606080.0,
      "grad_norm": 5.066087934836996,
      "learning_rate": 3.978249313331848e-06,
      "loss": 0.9565,
      "num_input_tokens_seen": 13164315,
      "step": 629
    },
    {
      "epoch": 0.07575302110262731,
      "flos": 13859590533120.0,
      "grad_norm": 18.89411655465424,
      "learning_rate": 3.978134592434208e-06,
      "loss": 0.8127,
      "num_input_tokens_seen": 13181785,
      "step": 630
    },
    {
      "epoch": 0.0758732639932664,
      "flos": 48667442196480.0,
      "grad_norm": 1.1502333632329154,
      "learning_rate": 3.978019571455123e-06,
      "loss": 0.8442,
      "num_input_tokens_seen": 13233450,
      "step": 631
    },
    {
      "epoch": 0.07599350688390549,
      "flos": 13465659678720.0,
      "grad_norm": 5.2219875138658365,
      "learning_rate": 3.977904250412042e-06,
      "loss": 1.0379,
      "num_input_tokens_seen": 13252125,
      "step": 632
    },
    {
      "epoch": 0.07611374977454458,
      "flos": 14960389509120.0,
      "grad_norm": 3.874329293564963,
      "learning_rate": 3.97778862932246e-06,
      "loss": 1.0609,
      "num_input_tokens_seen": 13269010,
      "step": 633
    },
    {
      "epoch": 0.07623399266518367,
      "flos": 13124477767680.0,
      "grad_norm": 11.075207324703058,
      "learning_rate": 3.9776727082039144e-06,
      "loss": 1.1387,
      "num_input_tokens_seen": 13285700,
      "step": 634
    },
    {
      "epoch": 0.07635423555582276,
      "flos": 31920241520640.0,
      "grad_norm": 0.8671633742816965,
      "learning_rate": 3.977556487073991e-06,
      "loss": 0.7768,
      "num_input_tokens_seen": 13339975,
      "step": 635
    },
    {
      "epoch": 0.07647447844646185,
      "flos": 15457672151040.0,
      "grad_norm": 2.5181421043286605,
      "learning_rate": 3.97743996595032e-06,
      "loss": 1.0063,
      "num_input_tokens_seen": 13359735,
      "step": 636
    },
    {
      "epoch": 0.07659472133710095,
      "flos": 16928173608960.0,
      "grad_norm": 3.692262063906024,
      "learning_rate": 3.9773231448505804e-06,
      "loss": 1.0166,
      "num_input_tokens_seen": 13381245,
      "step": 637
    },
    {
      "epoch": 0.07671496422774003,
      "flos": 15248515891200.0,
      "grad_norm": 6.035952617360696,
      "learning_rate": 3.977206023792491e-06,
      "loss": 0.9717,
      "num_input_tokens_seen": 13400855,
      "step": 638
    },
    {
      "epoch": 0.07683520711837913,
      "flos": 12022176522240.0,
      "grad_norm": 6.020262201625305,
      "learning_rate": 3.97708860279382e-06,
      "loss": 1.0166,
      "num_input_tokens_seen": 13418685,
      "step": 639
    },
    {
      "epoch": 0.07695545000901821,
      "flos": 16690742046720.0,
      "grad_norm": 5.387086984331868,
      "learning_rate": 3.97697088187238e-06,
      "loss": 0.9932,
      "num_input_tokens_seen": 13438920,
      "step": 640
    },
    {
      "epoch": 0.07707569289965731,
      "flos": 12491214520320.0,
      "grad_norm": 5.439269906867187,
      "learning_rate": 3.976852861046029e-06,
      "loss": 1.1107,
      "num_input_tokens_seen": 13455255,
      "step": 641
    },
    {
      "epoch": 0.0771959357902964,
      "flos": 18342982901760.0,
      "grad_norm": 4.587258455741349,
      "learning_rate": 3.97673454033267e-06,
      "loss": 1.0034,
      "num_input_tokens_seen": 13477075,
      "step": 642
    },
    {
      "epoch": 0.07731617868093549,
      "flos": 14068992061440.0,
      "grad_norm": 9.686135866657237,
      "learning_rate": 3.976615919750254e-06,
      "loss": 1.0228,
      "num_input_tokens_seen": 13494495,
      "step": 643
    },
    {
      "epoch": 0.07743642157157458,
      "flos": 15144152371200.0,
      "grad_norm": 6.577518099879927,
      "learning_rate": 3.976496999316775e-06,
      "loss": 1.0566,
      "num_input_tokens_seen": 13512970,
      "step": 644
    },
    {
      "epoch": 0.07755666446221367,
      "flos": 14169216675840.0,
      "grad_norm": 4.863551464004218,
      "learning_rate": 3.976377779050271e-06,
      "loss": 1.04,
      "num_input_tokens_seen": 13530820,
      "step": 645
    },
    {
      "epoch": 0.07767690735285276,
      "flos": 16796086640640.0,
      "grad_norm": 6.490397182354621,
      "learning_rate": 3.976258258968831e-06,
      "loss": 1.0331,
      "num_input_tokens_seen": 13549085,
      "step": 646
    },
    {
      "epoch": 0.07779715024349185,
      "flos": 15799221964800.0,
      "grad_norm": 6.8879790816216495,
      "learning_rate": 3.976138439090583e-06,
      "loss": 0.9321,
      "num_input_tokens_seen": 13566885,
      "step": 647
    },
    {
      "epoch": 0.07791739313413094,
      "flos": 14877924311040.0,
      "grad_norm": 4.753488163363645,
      "learning_rate": 3.976018319433706e-06,
      "loss": 1.0439,
      "num_input_tokens_seen": 13584150,
      "step": 648
    },
    {
      "epoch": 0.07803763602477004,
      "flos": 13697174138880.0,
      "grad_norm": 15.458771047367312,
      "learning_rate": 3.9758979000164205e-06,
      "loss": 1.1288,
      "num_input_tokens_seen": 13600690,
      "step": 649
    },
    {
      "epoch": 0.07815787891540912,
      "flos": 16140311900160.0,
      "grad_norm": 4.480452360407578,
      "learning_rate": 3.975777180856995e-06,
      "loss": 0.9117,
      "num_input_tokens_seen": 13619530,
      "step": 650
    },
    {
      "epoch": 0.07827812180604822,
      "flos": 16140587827200.0,
      "grad_norm": 9.324389759376515,
      "learning_rate": 3.975656161973742e-06,
      "loss": 1.0499,
      "num_input_tokens_seen": 13638335,
      "step": 651
    },
    {
      "epoch": 0.0783983646966873,
      "flos": 15432248094720.0,
      "grad_norm": 4.839820695503504,
      "learning_rate": 3.9755348433850194e-06,
      "loss": 1.0878,
      "num_input_tokens_seen": 13653395,
      "step": 652
    },
    {
      "epoch": 0.0785186075873264,
      "flos": 43402818478080.0,
      "grad_norm": 1.1002846197893448,
      "learning_rate": 3.975413225109232e-06,
      "loss": 0.9148,
      "num_input_tokens_seen": 13713665,
      "step": 653
    },
    {
      "epoch": 0.0786388504779655,
      "flos": 16979021721600.0,
      "grad_norm": 5.97423904233918,
      "learning_rate": 3.975291307164829e-06,
      "loss": 1.1281,
      "num_input_tokens_seen": 13732030,
      "step": 654
    },
    {
      "epoch": 0.07875909336860458,
      "flos": 10712803799040.0,
      "grad_norm": 5.415966145137755,
      "learning_rate": 3.975169089570306e-06,
      "loss": 1.0544,
      "num_input_tokens_seen": 13750125,
      "step": 655
    },
    {
      "epoch": 0.07887933625924368,
      "flos": 15799743160320.0,
      "grad_norm": 63.7869167772916,
      "learning_rate": 3.975046572344202e-06,
      "loss": 1.1193,
      "num_input_tokens_seen": 13766305,
      "step": 656
    },
    {
      "epoch": 0.07899957914988276,
      "flos": 14751141273600.0,
      "grad_norm": 4.923628818544088,
      "learning_rate": 3.974923755505103e-06,
      "loss": 0.9121,
      "num_input_tokens_seen": 13785255,
      "step": 657
    },
    {
      "epoch": 0.07911982204052186,
      "flos": 16350387916800.0,
      "grad_norm": 4.522474328047818,
      "learning_rate": 3.974800639071641e-06,
      "loss": 1.1121,
      "num_input_tokens_seen": 13805695,
      "step": 658
    },
    {
      "epoch": 0.07924006493116094,
      "flos": 16428101038080.0,
      "grad_norm": 6.460184664611291,
      "learning_rate": 3.974677223062492e-06,
      "loss": 1.2098,
      "num_input_tokens_seen": 13822630,
      "step": 659
    },
    {
      "epoch": 0.07936030782180004,
      "flos": 11656766238720.0,
      "grad_norm": 5.295928960617599,
      "learning_rate": 3.974553507496378e-06,
      "loss": 0.9313,
      "num_input_tokens_seen": 13840925,
      "step": 660
    },
    {
      "epoch": 0.07948055071243913,
      "flos": 16875271372800.0,
      "grad_norm": 5.708690362981723,
      "learning_rate": 3.974429492392068e-06,
      "loss": 1.0889,
      "num_input_tokens_seen": 13860670,
      "step": 661
    },
    {
      "epoch": 0.07960079360307822,
      "flos": 13885505126400.0,
      "grad_norm": 6.562923454064966,
      "learning_rate": 3.974305177768373e-06,
      "loss": 1.1041,
      "num_input_tokens_seen": 13878600,
      "step": 662
    },
    {
      "epoch": 0.07972103649371731,
      "flos": 16717116518400.0,
      "grad_norm": 11.064968027791798,
      "learning_rate": 3.974180563644152e-06,
      "loss": 1.0555,
      "num_input_tokens_seen": 13896885,
      "step": 663
    },
    {
      "epoch": 0.0798412793843564,
      "flos": 11839731978240.0,
      "grad_norm": 10.552093616316148,
      "learning_rate": 3.97405565003831e-06,
      "loss": 1.0889,
      "num_input_tokens_seen": 13912690,
      "step": 664
    },
    {
      "epoch": 0.07996152227499549,
      "flos": 12915290849280.0,
      "grad_norm": 18.286776341521865,
      "learning_rate": 3.973930436969794e-06,
      "loss": 0.9762,
      "num_input_tokens_seen": 13930865,
      "step": 665
    },
    {
      "epoch": 0.08008176516563459,
      "flos": 14619667476480.0,
      "grad_norm": 7.586725698866013,
      "learning_rate": 3.973804924457602e-06,
      "loss": 1.0595,
      "num_input_tokens_seen": 13948665,
      "step": 666
    },
    {
      "epoch": 0.08020200805627367,
      "flos": 22697476792320.0,
      "grad_norm": 15.27008385096263,
      "learning_rate": 3.973679112520771e-06,
      "loss": 1.0608,
      "num_input_tokens_seen": 13970100,
      "step": 667
    },
    {
      "epoch": 0.08032225094691277,
      "flos": 12598919823360.0,
      "grad_norm": 5.089650569830274,
      "learning_rate": 3.973553001178389e-06,
      "loss": 1.1907,
      "num_input_tokens_seen": 13987325,
      "step": 668
    },
    {
      "epoch": 0.08044249383755185,
      "flos": 17111292641280.0,
      "grad_norm": 5.441816874480962,
      "learning_rate": 3.973426590449585e-06,
      "loss": 0.9603,
      "num_input_tokens_seen": 14005000,
      "step": 669
    },
    {
      "epoch": 0.08056273672819095,
      "flos": 12915536117760.0,
      "grad_norm": 8.297803693904678,
      "learning_rate": 3.9732998803535364e-06,
      "loss": 0.9564,
      "num_input_tokens_seen": 14022780,
      "step": 670
    },
    {
      "epoch": 0.08068297961883003,
      "flos": 13959110000640.0,
      "grad_norm": 8.403731525474031,
      "learning_rate": 3.973172870909465e-06,
      "loss": 1.0611,
      "num_input_tokens_seen": 14037265,
      "step": 671
    },
    {
      "epoch": 0.08080322250946913,
      "flos": 16455058022400.0,
      "grad_norm": 24.52228762436391,
      "learning_rate": 3.973045562136638e-06,
      "loss": 1.0005,
      "num_input_tokens_seen": 14053800,
      "step": 672
    },
    {
      "epoch": 0.08092346540010822,
      "flos": 15459787591680.0,
      "grad_norm": 8.214187109693015,
      "learning_rate": 3.972917954054368e-06,
      "loss": 1.1153,
      "num_input_tokens_seen": 14072075,
      "step": 673
    },
    {
      "epoch": 0.08104370829074731,
      "flos": 14934137671680.0,
      "grad_norm": 8.667961268445397,
      "learning_rate": 3.972790046682013e-06,
      "loss": 1.0093,
      "num_input_tokens_seen": 14090470,
      "step": 674
    },
    {
      "epoch": 0.0811639511813864,
      "flos": 14251804508160.0,
      "grad_norm": 7.908224349347908,
      "learning_rate": 3.972661840038977e-06,
      "loss": 0.9855,
      "num_input_tokens_seen": 14110480,
      "step": 675
    },
    {
      "epoch": 0.08128419407202549,
      "flos": 11918732759040.0,
      "grad_norm": 4.290904721271071,
      "learning_rate": 3.972533334144707e-06,
      "loss": 1.021,
      "num_input_tokens_seen": 14127125,
      "step": 676
    },
    {
      "epoch": 0.08140443696266458,
      "flos": 16901063331840.0,
      "grad_norm": 9.645682567849049,
      "learning_rate": 3.972404529018699e-06,
      "loss": 0.9801,
      "num_input_tokens_seen": 14146705,
      "step": 677
    },
    {
      "epoch": 0.08152467985330367,
      "flos": 17347344568320.0,
      "grad_norm": 5.587074942923773,
      "learning_rate": 3.972275424680493e-06,
      "loss": 1.0539,
      "num_input_tokens_seen": 14166535,
      "step": 678
    },
    {
      "epoch": 0.08164492274394276,
      "flos": 13701803581440.0,
      "grad_norm": 4.772367516702537,
      "learning_rate": 3.972146021149673e-06,
      "loss": 1.1046,
      "num_input_tokens_seen": 14184530,
      "step": 679
    },
    {
      "epoch": 0.08176516563458186,
      "flos": 10476935823360.0,
      "grad_norm": 8.005454111555856,
      "learning_rate": 3.972016318445868e-06,
      "loss": 0.9958,
      "num_input_tokens_seen": 14202250,
      "step": 680
    },
    {
      "epoch": 0.08188540852522094,
      "flos": 16062476144640.0,
      "grad_norm": 5.620093855158995,
      "learning_rate": 3.971886316588757e-06,
      "loss": 1.1052,
      "num_input_tokens_seen": 14222475,
      "step": 681
    },
    {
      "epoch": 0.08200565141586004,
      "flos": 13806718955520.0,
      "grad_norm": 14.740642100702928,
      "learning_rate": 3.9717560155980595e-06,
      "loss": 0.9385,
      "num_input_tokens_seen": 14237845,
      "step": 682
    },
    {
      "epoch": 0.08212589430649912,
      "flos": 14619483525120.0,
      "grad_norm": 6.252836575545415,
      "learning_rate": 3.971625415493542e-06,
      "loss": 1.1134,
      "num_input_tokens_seen": 14255885,
      "step": 683
    },
    {
      "epoch": 0.08224613719713822,
      "flos": 18470256476160.0,
      "grad_norm": 8.812719896968352,
      "learning_rate": 3.971494516295017e-06,
      "loss": 1.0778,
      "num_input_tokens_seen": 14275055,
      "step": 684
    },
    {
      "epoch": 0.08236638008777732,
      "flos": 16900603453440.0,
      "grad_norm": 6.0953683122564914,
      "learning_rate": 3.971363318022341e-06,
      "loss": 1.0425,
      "num_input_tokens_seen": 14296115,
      "step": 685
    },
    {
      "epoch": 0.0824866229784164,
      "flos": 27702441492480.0,
      "grad_norm": 6.451767896513212,
      "learning_rate": 3.971231820695417e-06,
      "loss": 0.8884,
      "num_input_tokens_seen": 14319450,
      "step": 686
    },
    {
      "epoch": 0.0826068658690555,
      "flos": 16425004523520.0,
      "grad_norm": 8.622887836514707,
      "learning_rate": 3.971100024334193e-06,
      "loss": 1.0168,
      "num_input_tokens_seen": 14336690,
      "step": 687
    },
    {
      "epoch": 0.08272710875969458,
      "flos": 15008049131520.0,
      "grad_norm": 6.524715969875534,
      "learning_rate": 3.970967928958663e-06,
      "loss": 1.0653,
      "num_input_tokens_seen": 14353525,
      "step": 688
    },
    {
      "epoch": 0.08284735165033368,
      "flos": 13518807183360.0,
      "grad_norm": 3.8120270732786863,
      "learning_rate": 3.970835534588865e-06,
      "loss": 1.0341,
      "num_input_tokens_seen": 14370740,
      "step": 689
    },
    {
      "epoch": 0.08296759454097276,
      "flos": 11840222515200.0,
      "grad_norm": 7.931764948776326,
      "learning_rate": 3.970702841244883e-06,
      "loss": 1.0622,
      "num_input_tokens_seen": 14388780,
      "step": 690
    },
    {
      "epoch": 0.08308783743161186,
      "flos": 12758392995840.0,
      "grad_norm": 5.527113580123706,
      "learning_rate": 3.970569848946847e-06,
      "loss": 1.0193,
      "num_input_tokens_seen": 14408315,
      "step": 691
    },
    {
      "epoch": 0.08320808032225095,
      "flos": 10655302778880.0,
      "grad_norm": 8.229622846747414,
      "learning_rate": 3.970436557714932e-06,
      "loss": 1.0062,
      "num_input_tokens_seen": 14424555,
      "step": 692
    },
    {
      "epoch": 0.08332832321289003,
      "flos": 15904873144320.0,
      "grad_norm": 5.433385585853497,
      "learning_rate": 3.970302967569358e-06,
      "loss": 1.0509,
      "num_input_tokens_seen": 14442865,
      "step": 693
    },
    {
      "epoch": 0.08344856610352913,
      "flos": 17582384762880.0,
      "grad_norm": 10.815162639550174,
      "learning_rate": 3.9701690785303896e-06,
      "loss": 0.8765,
      "num_input_tokens_seen": 14461780,
      "step": 694
    },
    {
      "epoch": 0.08356880899416821,
      "flos": 18051024199680.0,
      "grad_norm": 8.35640097505256,
      "learning_rate": 3.970034890618339e-06,
      "loss": 1.0758,
      "num_input_tokens_seen": 14481190,
      "step": 695
    },
    {
      "epoch": 0.08368905188480731,
      "flos": 17320632852480.0,
      "grad_norm": 8.485886918550918,
      "learning_rate": 3.969900403853562e-06,
      "loss": 1.0714,
      "num_input_tokens_seen": 14499950,
      "step": 696
    },
    {
      "epoch": 0.08380929477544641,
      "flos": 12781119098880.0,
      "grad_norm": 5.5473575865665365,
      "learning_rate": 3.96976561825646e-06,
      "loss": 0.9823,
      "num_input_tokens_seen": 14516760,
      "step": 697
    },
    {
      "epoch": 0.08392953766608549,
      "flos": 18709926113280.0,
      "grad_norm": 5.172323247840507,
      "learning_rate": 3.969630533847479e-06,
      "loss": 1.0718,
      "num_input_tokens_seen": 14535440,
      "step": 698
    },
    {
      "epoch": 0.08404978055672459,
      "flos": 15983199436800.0,
      "grad_norm": 3.459485716111993,
      "learning_rate": 3.969495150647113e-06,
      "loss": 1.0412,
      "num_input_tokens_seen": 14553330,
      "step": 699
    },
    {
      "epoch": 0.08417002344736367,
      "flos": 17661078958080.0,
      "grad_norm": 4.0202035888437235,
      "learning_rate": 3.969359468675899e-06,
      "loss": 0.9668,
      "num_input_tokens_seen": 14573180,
      "step": 700
    },
    {
      "epoch": 0.08429026633800277,
      "flos": 11997181685760.0,
      "grad_norm": 6.245591096292635,
      "learning_rate": 3.969223487954418e-06,
      "loss": 1.0958,
      "num_input_tokens_seen": 14590360,
      "step": 701
    },
    {
      "epoch": 0.08441050922864185,
      "flos": 16953229762560.0,
      "grad_norm": 11.645377071134547,
      "learning_rate": 3.969087208503301e-06,
      "loss": 1.0226,
      "num_input_tokens_seen": 14610160,
      "step": 702
    },
    {
      "epoch": 0.08453075211928095,
      "flos": 18159373332480.0,
      "grad_norm": 7.836395828367619,
      "learning_rate": 3.968950630343219e-06,
      "loss": 1.0329,
      "num_input_tokens_seen": 14626865,
      "step": 703
    },
    {
      "epoch": 0.08465099500992004,
      "flos": 13855574261760.0,
      "grad_norm": 3.8782332682654683,
      "learning_rate": 3.968813753494892e-06,
      "loss": 1.1279,
      "num_input_tokens_seen": 14644745,
      "step": 704
    },
    {
      "epoch": 0.08477123790055913,
      "flos": 20912719749120.0,
      "grad_norm": 6.065526148026449,
      "learning_rate": 3.968676577979084e-06,
      "loss": 0.9563,
      "num_input_tokens_seen": 14664015,
      "step": 705
    },
    {
      "epoch": 0.08489148079119822,
      "flos": 13203570524160.0,
      "grad_norm": 8.145033801191902,
      "learning_rate": 3.968539103816605e-06,
      "loss": 0.9809,
      "num_input_tokens_seen": 14681535,
      "step": 706
    },
    {
      "epoch": 0.0850117236818373,
      "flos": 16686173921280.0,
      "grad_norm": 8.26682715516533,
      "learning_rate": 3.9684013310283085e-06,
      "loss": 1.0873,
      "num_input_tokens_seen": 14699940,
      "step": 707
    },
    {
      "epoch": 0.0851319665724764,
      "flos": 29014910730240.0,
      "grad_norm": 7.950582701243138,
      "learning_rate": 3.9682632596350956e-06,
      "loss": 0.8378,
      "num_input_tokens_seen": 14720825,
      "step": 708
    },
    {
      "epoch": 0.0852522094631155,
      "flos": 11229685370880.0,
      "grad_norm": 4.853103761448416,
      "learning_rate": 3.968124889657911e-06,
      "loss": 0.9829,
      "num_input_tokens_seen": 14735645,
      "step": 709
    },
    {
      "epoch": 0.08537245235375458,
      "flos": 10286581370880.0,
      "grad_norm": 5.376546947630805,
      "learning_rate": 3.967986221117746e-06,
      "loss": 1.1135,
      "num_input_tokens_seen": 14751305,
      "step": 710
    },
    {
      "epoch": 0.08549269524439368,
      "flos": 18841215959040.0,
      "grad_norm": 9.53246822388898,
      "learning_rate": 3.967847254035635e-06,
      "loss": 1.0686,
      "num_input_tokens_seen": 14770410,
      "step": 711
    },
    {
      "epoch": 0.08561293813503276,
      "flos": 9585630351360.0,
      "grad_norm": 7.873836128574093,
      "learning_rate": 3.967707988432661e-06,
      "loss": 1.0621,
      "num_input_tokens_seen": 14787835,
      "step": 712
    },
    {
      "epoch": 0.08573318102567186,
      "flos": 19182336552960.0,
      "grad_norm": 7.4358287824974285,
      "learning_rate": 3.967568424329949e-06,
      "loss": 1.0812,
      "num_input_tokens_seen": 14807980,
      "step": 713
    },
    {
      "epoch": 0.08585342391631094,
      "flos": 48193468170240.0,
      "grad_norm": 0.8577111870532358,
      "learning_rate": 3.967428561748671e-06,
      "loss": 0.7927,
      "num_input_tokens_seen": 14875670,
      "step": 714
    },
    {
      "epoch": 0.08597366680695004,
      "flos": 15957775380480.0,
      "grad_norm": 5.752994586848539,
      "learning_rate": 3.967288400710045e-06,
      "loss": 1.063,
      "num_input_tokens_seen": 14894855,
      "step": 715
    },
    {
      "epoch": 0.08609390969758914,
      "flos": 16743858892800.0,
      "grad_norm": 3.277256306333074,
      "learning_rate": 3.9671479412353335e-06,
      "loss": 1.0809,
      "num_input_tokens_seen": 14913040,
      "step": 716
    },
    {
      "epoch": 0.08621415258822822,
      "flos": 18422259609600.0,
      "grad_norm": 57.92782627235917,
      "learning_rate": 3.967007183345843e-06,
      "loss": 0.942,
      "num_input_tokens_seen": 14932615,
      "step": 717
    },
    {
      "epoch": 0.08633439547886732,
      "flos": 9585078497280.0,
      "grad_norm": 5.412484163256728,
      "learning_rate": 3.966866127062927e-06,
      "loss": 1.0894,
      "num_input_tokens_seen": 14949460,
      "step": 718
    },
    {
      "epoch": 0.0864546383695064,
      "flos": 41338856693760.0,
      "grad_norm": 0.9845007861469902,
      "learning_rate": 3.966724772407982e-06,
      "loss": 0.8815,
      "num_input_tokens_seen": 15006695,
      "step": 719
    },
    {
      "epoch": 0.0865748812601455,
      "flos": 14225797939200.0,
      "grad_norm": 8.000690907207689,
      "learning_rate": 3.966583119402454e-06,
      "loss": 1.095,
      "num_input_tokens_seen": 15023180,
      "step": 720
    },
    {
      "epoch": 0.08669512415078459,
      "flos": 25161010606080.0,
      "grad_norm": 5.122892923547,
      "learning_rate": 3.9664411680678305e-06,
      "loss": 1.021,
      "num_input_tokens_seen": 15044655,
      "step": 721
    },
    {
      "epoch": 0.08681536704142367,
      "flos": 47008364482560.0,
      "grad_norm": 0.8907357197253076,
      "learning_rate": 3.966298918425644e-06,
      "loss": 0.8382,
      "num_input_tokens_seen": 15101865,
      "step": 722
    },
    {
      "epoch": 0.08693560993206277,
      "flos": 24634195660800.0,
      "grad_norm": 6.119580732457449,
      "learning_rate": 3.966156370497476e-06,
      "loss": 1.0315,
      "num_input_tokens_seen": 15125195,
      "step": 723
    },
    {
      "epoch": 0.08705585282270185,
      "flos": 16455640535040.0,
      "grad_norm": 50.21598613660669,
      "learning_rate": 3.96601352430495e-06,
      "loss": 1.0822,
      "num_input_tokens_seen": 15144685,
      "step": 724
    },
    {
      "epoch": 0.08717609571334095,
      "flos": 21017911050240.0,
      "grad_norm": 4.793553298833733,
      "learning_rate": 3.965870379869735e-06,
      "loss": 1.0311,
      "num_input_tokens_seen": 15166450,
      "step": 725
    },
    {
      "epoch": 0.08729633860398003,
      "flos": 14671803248640.0,
      "grad_norm": 6.948023607001996,
      "learning_rate": 3.965726937213547e-06,
      "loss": 1.0608,
      "num_input_tokens_seen": 15184805,
      "step": 726
    },
    {
      "epoch": 0.08741658149461913,
      "flos": 13020359516160.0,
      "grad_norm": 13.581584669642636,
      "learning_rate": 3.965583196358144e-06,
      "loss": 1.0029,
      "num_input_tokens_seen": 15203560,
      "step": 727
    },
    {
      "epoch": 0.08753682438525823,
      "flos": 13278585692160.0,
      "grad_norm": 10.808468458457954,
      "learning_rate": 3.965439157325335e-06,
      "loss": 0.9339,
      "num_input_tokens_seen": 15220645,
      "step": 728
    },
    {
      "epoch": 0.08765706727589731,
      "flos": 19779813150720.0,
      "grad_norm": 8.68663009569497,
      "learning_rate": 3.965294820136968e-06,
      "loss": 0.95,
      "num_input_tokens_seen": 15242165,
      "step": 729
    },
    {
      "epoch": 0.08777731016653641,
      "flos": 17346486128640.0,
      "grad_norm": 9.623690825762395,
      "learning_rate": 3.965150184814938e-06,
      "loss": 1.0593,
      "num_input_tokens_seen": 15261370,
      "step": 730
    },
    {
      "epoch": 0.08789755305717549,
      "flos": 15616532152320.0,
      "grad_norm": 8.036164557542262,
      "learning_rate": 3.965005251381189e-06,
      "loss": 0.9489,
      "num_input_tokens_seen": 15279025,
      "step": 731
    },
    {
      "epoch": 0.08801779594781459,
      "flos": 41766090854400.0,
      "grad_norm": 0.9722417231526481,
      "learning_rate": 3.964860019857705e-06,
      "loss": 0.8661,
      "num_input_tokens_seen": 15343660,
      "step": 732
    },
    {
      "epoch": 0.08813803883845367,
      "flos": 16560034713600.0,
      "grad_norm": 10.551758128832134,
      "learning_rate": 3.964714490266518e-06,
      "loss": 1.0353,
      "num_input_tokens_seen": 15364025,
      "step": 733
    },
    {
      "epoch": 0.08825828172909277,
      "flos": 45405162885120.0,
      "grad_norm": 0.8935321408955329,
      "learning_rate": 3.964568662629706e-06,
      "loss": 0.857,
      "num_input_tokens_seen": 15425050,
      "step": 734
    },
    {
      "epoch": 0.08837852461973186,
      "flos": 19021912965120.0,
      "grad_norm": 5.934767244783561,
      "learning_rate": 3.9644225369693895e-06,
      "loss": 1.0365,
      "num_input_tokens_seen": 15445070,
      "step": 735
    },
    {
      "epoch": 0.08849876751037095,
      "flos": 19412808622080.0,
      "grad_norm": 5.9359935254784,
      "learning_rate": 3.964276113307735e-06,
      "loss": 1.0693,
      "num_input_tokens_seen": 15464755,
      "step": 736
    },
    {
      "epoch": 0.08861901040101004,
      "flos": 14069084037120.0,
      "grad_norm": 4.888761415168312,
      "learning_rate": 3.9641293916669574e-06,
      "loss": 1.0027,
      "num_input_tokens_seen": 15483435,
      "step": 737
    },
    {
      "epoch": 0.08873925329164913,
      "flos": 16821449379840.0,
      "grad_norm": 4.525321712965274,
      "learning_rate": 3.9639823720693115e-06,
      "loss": 1.0239,
      "num_input_tokens_seen": 15505010,
      "step": 738
    },
    {
      "epoch": 0.08885949618228822,
      "flos": 51448849305600.0,
      "grad_norm": 0.967757732059905,
      "learning_rate": 3.963835054537102e-06,
      "loss": 0.8546,
      "num_input_tokens_seen": 15573695,
      "step": 739
    },
    {
      "epoch": 0.08897973907292732,
      "flos": 15878284062720.0,
      "grad_norm": 11.712313987858902,
      "learning_rate": 3.963687439092676e-06,
      "loss": 0.8117,
      "num_input_tokens_seen": 15594100,
      "step": 740
    },
    {
      "epoch": 0.0890999819635664,
      "flos": 15091832647680.0,
      "grad_norm": 9.093821312274082,
      "learning_rate": 3.963539525758427e-06,
      "loss": 1.0045,
      "num_input_tokens_seen": 15613380,
      "step": 741
    },
    {
      "epoch": 0.0892202248542055,
      "flos": 18051606712320.0,
      "grad_norm": 8.756665551986982,
      "learning_rate": 3.9633913145567925e-06,
      "loss": 0.8696,
      "num_input_tokens_seen": 15633590,
      "step": 742
    },
    {
      "epoch": 0.08934046774484458,
      "flos": 17395341434880.0,
      "grad_norm": 6.242280999964639,
      "learning_rate": 3.9632428055102575e-06,
      "loss": 1.0186,
      "num_input_tokens_seen": 15653320,
      "step": 743
    },
    {
      "epoch": 0.08946071063548368,
      "flos": 25527708549120.0,
      "grad_norm": 8.001997221549843,
      "learning_rate": 3.9630939986413495e-06,
      "loss": 0.8666,
      "num_input_tokens_seen": 15674840,
      "step": 744
    },
    {
      "epoch": 0.08958095352612276,
      "flos": 10136428400640.0,
      "grad_norm": 6.644308778333959,
      "learning_rate": 3.962944893972643e-06,
      "loss": 0.9834,
      "num_input_tokens_seen": 15693010,
      "step": 745
    },
    {
      "epoch": 0.08970119641676186,
      "flos": 12648633569280.0,
      "grad_norm": 7.379320075495528,
      "learning_rate": 3.962795491526756e-06,
      "loss": 1.1063,
      "num_input_tokens_seen": 15709890,
      "step": 746
    },
    {
      "epoch": 0.08982143930740095,
      "flos": 14775185694720.0,
      "grad_norm": 4.636734381555369,
      "learning_rate": 3.962645791326354e-06,
      "loss": 1.0878,
      "num_input_tokens_seen": 15728865,
      "step": 747
    },
    {
      "epoch": 0.08994168219804004,
      "flos": 17137115258880.0,
      "grad_norm": 48.76084364576637,
      "learning_rate": 3.962495793394146e-06,
      "loss": 1.0305,
      "num_input_tokens_seen": 15747775,
      "step": 748
    },
    {
      "epoch": 0.09006192508867913,
      "flos": 40923855298560.0,
      "grad_norm": 0.7856339477053801,
      "learning_rate": 3.9623454977528864e-06,
      "loss": 0.8136,
      "num_input_tokens_seen": 15806150,
      "step": 749
    },
    {
      "epoch": 0.09018216797931822,
      "flos": 14542782136320.0,
      "grad_norm": 11.625576210098012,
      "learning_rate": 3.962194904425375e-06,
      "loss": 1.0505,
      "num_input_tokens_seen": 15826500,
      "step": 750
    },
    {
      "epoch": 0.09030241086995731,
      "flos": 16088114810880.0,
      "grad_norm": 5.9911118478205445,
      "learning_rate": 3.9620440134344566e-06,
      "loss": 0.8796,
      "num_input_tokens_seen": 15844375,
      "step": 751
    },
    {
      "epoch": 0.09042265376059641,
      "flos": 15537102151680.0,
      "grad_norm": 7.5191669960477086,
      "learning_rate": 3.9618928248030215e-06,
      "loss": 1.0252,
      "num_input_tokens_seen": 15863605,
      "step": 752
    },
    {
      "epoch": 0.0905428966512355,
      "flos": 17294074429440.0,
      "grad_norm": 5.231316153815584,
      "learning_rate": 3.961741338554005e-06,
      "loss": 1.0298,
      "num_input_tokens_seen": 15881665,
      "step": 753
    },
    {
      "epoch": 0.09066313954187459,
      "flos": 25579568394240.0,
      "grad_norm": 12.098083823570922,
      "learning_rate": 3.9615895547103865e-06,
      "loss": 0.9481,
      "num_input_tokens_seen": 15905030,
      "step": 754
    },
    {
      "epoch": 0.09078338243251367,
      "flos": 20781491220480.0,
      "grad_norm": 9.119976594459883,
      "learning_rate": 3.961437473295193e-06,
      "loss": 0.9763,
      "num_input_tokens_seen": 15924895,
      "step": 755
    },
    {
      "epoch": 0.09090362532315277,
      "flos": 15563078062080.0,
      "grad_norm": 5.710966556334257,
      "learning_rate": 3.961285094331495e-06,
      "loss": 0.9066,
      "num_input_tokens_seen": 15942530,
      "step": 756
    },
    {
      "epoch": 0.09102386821379185,
      "flos": 19470462935040.0,
      "grad_norm": 9.954193489056665,
      "learning_rate": 3.961132417842406e-06,
      "loss": 1.0553,
      "num_input_tokens_seen": 15962035,
      "step": 757
    },
    {
      "epoch": 0.09114411110443095,
      "flos": 14776749281280.0,
      "grad_norm": 15.535739671908779,
      "learning_rate": 3.960979443851089e-06,
      "loss": 0.9522,
      "num_input_tokens_seen": 15978780,
      "step": 758
    },
    {
      "epoch": 0.09126435399507005,
      "flos": 18609824133120.0,
      "grad_norm": 4.601454935937743,
      "learning_rate": 3.96082617238075e-06,
      "loss": 0.9972,
      "num_input_tokens_seen": 16001125,
      "step": 759
    },
    {
      "epoch": 0.09138459688570913,
      "flos": 17346302177280.0,
      "grad_norm": 5.423612904219332,
      "learning_rate": 3.960672603454639e-06,
      "loss": 0.9845,
      "num_input_tokens_seen": 16020825,
      "step": 760
    },
    {
      "epoch": 0.09150483977634823,
      "flos": 14934290964480.0,
      "grad_norm": 7.348953546260696,
      "learning_rate": 3.960518737096054e-06,
      "loss": 0.9558,
      "num_input_tokens_seen": 16040175,
      "step": 761
    },
    {
      "epoch": 0.09162508266698731,
      "flos": 16245411225600.0,
      "grad_norm": 8.110975633067508,
      "learning_rate": 3.960364573328334e-06,
      "loss": 0.9252,
      "num_input_tokens_seen": 16059220,
      "step": 762
    },
    {
      "epoch": 0.0917453255576264,
      "flos": 15431236362240.0,
      "grad_norm": 6.439536652097042,
      "learning_rate": 3.9602101121748675e-06,
      "loss": 1.0826,
      "num_input_tokens_seen": 16079435,
      "step": 763
    },
    {
      "epoch": 0.0918655684482655,
      "flos": 10317615943680.0,
      "grad_norm": 6.45656894660437,
      "learning_rate": 3.960055353659085e-06,
      "loss": 0.9196,
      "num_input_tokens_seen": 16096265,
      "step": 764
    },
    {
      "epoch": 0.09198581133890459,
      "flos": 16660044718080.0,
      "grad_norm": 5.148971506768575,
      "learning_rate": 3.959900297804465e-06,
      "loss": 1.0409,
      "num_input_tokens_seen": 16116155,
      "step": 765
    },
    {
      "epoch": 0.09210605422954368,
      "flos": 11888801894400.0,
      "grad_norm": 4.849042597392616,
      "learning_rate": 3.9597449446345276e-06,
      "loss": 0.9678,
      "num_input_tokens_seen": 16133120,
      "step": 766
    },
    {
      "epoch": 0.09222629712018277,
      "flos": 16114029404160.0,
      "grad_norm": 5.687738769885017,
      "learning_rate": 3.95958929417284e-06,
      "loss": 1.0265,
      "num_input_tokens_seen": 16150995,
      "step": 767
    },
    {
      "epoch": 0.09234654001082186,
      "flos": 50116658503680.0,
      "grad_norm": 0.78791699495381,
      "learning_rate": 3.9594333464430145e-06,
      "loss": 0.8086,
      "num_input_tokens_seen": 16220205,
      "step": 768
    },
    {
      "epoch": 0.09246678290146094,
      "flos": 14200404541440.0,
      "grad_norm": 13.548457456953933,
      "learning_rate": 3.959277101468709e-06,
      "loss": 1.0765,
      "num_input_tokens_seen": 16239475,
      "step": 769
    },
    {
      "epoch": 0.09258702579210004,
      "flos": 12573066547200.0,
      "grad_norm": 9.359351767891658,
      "learning_rate": 3.959120559273624e-06,
      "loss": 0.9783,
      "num_input_tokens_seen": 16256980,
      "step": 770
    },
    {
      "epoch": 0.09270726868273914,
      "flos": 14829252956160.0,
      "grad_norm": 4.916222029586281,
      "learning_rate": 3.958963719881509e-06,
      "loss": 1.0456,
      "num_input_tokens_seen": 16274790,
      "step": 771
    },
    {
      "epoch": 0.09282751157337822,
      "flos": 12046895431680.0,
      "grad_norm": 13.21320805813259,
      "learning_rate": 3.958806583316154e-06,
      "loss": 1.1249,
      "num_input_tokens_seen": 16292480,
      "step": 772
    },
    {
      "epoch": 0.09294775446401732,
      "flos": 23192368066560.0,
      "grad_norm": 5.451176248710625,
      "learning_rate": 3.9586491496013985e-06,
      "loss": 0.9884,
      "num_input_tokens_seen": 16314595,
      "step": 773
    },
    {
      "epoch": 0.0930679973546564,
      "flos": 12941450711040.0,
      "grad_norm": 24.9833110799896,
      "learning_rate": 3.958491418761124e-06,
      "loss": 1.0215,
      "num_input_tokens_seen": 16331885,
      "step": 774
    },
    {
      "epoch": 0.0931882402452955,
      "flos": 14981705318400.0,
      "grad_norm": 5.905599891366185,
      "learning_rate": 3.958333390819258e-06,
      "loss": 0.9261,
      "num_input_tokens_seen": 16348535,
      "step": 775
    },
    {
      "epoch": 0.0933084831359346,
      "flos": 17215870771200.0,
      "grad_norm": 7.616457296731183,
      "learning_rate": 3.9581750657997754e-06,
      "loss": 1.0049,
      "num_input_tokens_seen": 16367620,
      "step": 776
    },
    {
      "epoch": 0.09342872602657368,
      "flos": 18130208931840.0,
      "grad_norm": 3.44633361493588,
      "learning_rate": 3.95801644372669e-06,
      "loss": 1.0969,
      "num_input_tokens_seen": 16387245,
      "step": 777
    },
    {
      "epoch": 0.09354896891721277,
      "flos": 16455333949440.0,
      "grad_norm": 6.4696949439085865,
      "learning_rate": 3.957857524624068e-06,
      "loss": 1.0369,
      "num_input_tokens_seen": 16405845,
      "step": 778
    },
    {
      "epoch": 0.09366921180785186,
      "flos": 17267730616320.0,
      "grad_norm": 5.664457152199279,
      "learning_rate": 3.957698308516016e-06,
      "loss": 1.1029,
      "num_input_tokens_seen": 16426865,
      "step": 779
    },
    {
      "epoch": 0.09378945469849095,
      "flos": 13278892277760.0,
      "grad_norm": 7.232919237556955,
      "learning_rate": 3.957538795426688e-06,
      "loss": 1.0291,
      "num_input_tokens_seen": 16444010,
      "step": 780
    },
    {
      "epoch": 0.09390969758913004,
      "flos": 16505047695360.0,
      "grad_norm": 9.386757357260388,
      "learning_rate": 3.9573789853802804e-06,
      "loss": 0.9684,
      "num_input_tokens_seen": 16462205,
      "step": 781
    },
    {
      "epoch": 0.09402994047976913,
      "flos": 13938315386880.0,
      "grad_norm": 17.075892966561625,
      "learning_rate": 3.957218878401037e-06,
      "loss": 0.9527,
      "num_input_tokens_seen": 16480415,
      "step": 782
    },
    {
      "epoch": 0.09415018337040823,
      "flos": 20961912299520.0,
      "grad_norm": 13.539292973550458,
      "learning_rate": 3.957058474513246e-06,
      "loss": 1.0821,
      "num_input_tokens_seen": 16499990,
      "step": 783
    },
    {
      "epoch": 0.09427042626104731,
      "flos": 17478051901440.0,
      "grad_norm": 20.199024985171004,
      "learning_rate": 3.956897773741241e-06,
      "loss": 0.9781,
      "num_input_tokens_seen": 16518700,
      "step": 784
    },
    {
      "epoch": 0.09439066915168641,
      "flos": 18762644398080.0,
      "grad_norm": 11.597694404618949,
      "learning_rate": 3.956736776109398e-06,
      "loss": 0.9192,
      "num_input_tokens_seen": 16539595,
      "step": 785
    },
    {
      "epoch": 0.09451091204232549,
      "flos": 13780528435200.0,
      "grad_norm": 18.123407122894623,
      "learning_rate": 3.956575481642143e-06,
      "loss": 1.0325,
      "num_input_tokens_seen": 16558205,
      "step": 786
    },
    {
      "epoch": 0.09463115493296459,
      "flos": 18049184686080.0,
      "grad_norm": 8.764432938245571,
      "learning_rate": 3.956413890363943e-06,
      "loss": 0.9447,
      "num_input_tokens_seen": 16574905,
      "step": 787
    },
    {
      "epoch": 0.09475139782360369,
      "flos": 7093882552320.0,
      "grad_norm": 19.41443583016297,
      "learning_rate": 3.956252002299312e-06,
      "loss": 1.0219,
      "num_input_tokens_seen": 16590525,
      "step": 788
    },
    {
      "epoch": 0.09487164071424277,
      "flos": 12201217966080.0,
      "grad_norm": 6.647758978210104,
      "learning_rate": 3.956089817472807e-06,
      "loss": 1.1107,
      "num_input_tokens_seen": 16607550,
      "step": 789
    },
    {
      "epoch": 0.09499188360488187,
      "flos": 21989811548160.0,
      "grad_norm": 6.687262700229071,
      "learning_rate": 3.955927335909032e-06,
      "loss": 1.0562,
      "num_input_tokens_seen": 16630480,
      "step": 790
    },
    {
      "epoch": 0.09511212649552095,
      "flos": 20913056993280.0,
      "grad_norm": 10.603591713497956,
      "learning_rate": 3.955764557632634e-06,
      "loss": 0.9664,
      "num_input_tokens_seen": 16650010,
      "step": 791
    },
    {
      "epoch": 0.09523236938616005,
      "flos": 7277461463040.0,
      "grad_norm": 8.900359980128831,
      "learning_rate": 3.955601482668309e-06,
      "loss": 1.139,
      "num_input_tokens_seen": 16667590,
      "step": 792
    },
    {
      "epoch": 0.09535261227679913,
      "flos": 13517795450880.0,
      "grad_norm": 5.606472954202673,
      "learning_rate": 3.955438111040794e-06,
      "loss": 1.0839,
      "num_input_tokens_seen": 16685585,
      "step": 793
    },
    {
      "epoch": 0.09547285516743823,
      "flos": 14855443476480.0,
      "grad_norm": 12.035741983068922,
      "learning_rate": 3.955274442774873e-06,
      "loss": 1.0034,
      "num_input_tokens_seen": 16703885,
      "step": 794
    },
    {
      "epoch": 0.09559309805807732,
      "flos": 21489953587200.0,
      "grad_norm": 8.004857019390096,
      "learning_rate": 3.9551104778953725e-06,
      "loss": 0.9149,
      "num_input_tokens_seen": 16723900,
      "step": 795
    },
    {
      "epoch": 0.0957133409487164,
      "flos": 14958458019840.0,
      "grad_norm": 7.901011303260983,
      "learning_rate": 3.954946216427167e-06,
      "loss": 1.0501,
      "num_input_tokens_seen": 16744080,
      "step": 796
    },
    {
      "epoch": 0.0958335838393555,
      "flos": 51062315274240.0,
      "grad_norm": 0.8654387428603701,
      "learning_rate": 3.954781658395176e-06,
      "loss": 0.8461,
      "num_input_tokens_seen": 16800055,
      "step": 797
    },
    {
      "epoch": 0.09595382672999458,
      "flos": 15538359152640.0,
      "grad_norm": 5.3549897072174,
      "learning_rate": 3.95461680382436e-06,
      "loss": 1.1118,
      "num_input_tokens_seen": 16818700,
      "step": 798
    },
    {
      "epoch": 0.09607406962063368,
      "flos": 13254387978240.0,
      "grad_norm": 11.97645356787545,
      "learning_rate": 3.9544516527397295e-06,
      "loss": 1.0561,
      "num_input_tokens_seen": 16834770,
      "step": 799
    },
    {
      "epoch": 0.09619431251127276,
      "flos": 16038493040640.0,
      "grad_norm": 5.641039888546253,
      "learning_rate": 3.954286205166338e-06,
      "loss": 1.0026,
      "num_input_tokens_seen": 16855655,
      "step": 800
    },
    {
      "epoch": 0.09631455540191186,
      "flos": 10056998400000.0,
      "grad_norm": 6.947509418164074,
      "learning_rate": 3.954120461129282e-06,
      "loss": 1.0314,
      "num_input_tokens_seen": 16872785,
      "step": 801
    },
    {
      "epoch": 0.09643479829255096,
      "flos": 14724184289280.0,
      "grad_norm": 8.352091416541933,
      "learning_rate": 3.953954420653706e-06,
      "loss": 1.0521,
      "num_input_tokens_seen": 16889530,
      "step": 802
    },
    {
      "epoch": 0.09655504118319004,
      "flos": 17368323133440.0,
      "grad_norm": 7.615002941950069,
      "learning_rate": 3.953788083764798e-06,
      "loss": 1.0837,
      "num_input_tokens_seen": 16908485,
      "step": 803
    },
    {
      "epoch": 0.09667528407382914,
      "flos": 13072372654080.0,
      "grad_norm": 7.04160722635978,
      "learning_rate": 3.953621450487792e-06,
      "loss": 1.122,
      "num_input_tokens_seen": 16926825,
      "step": 804
    },
    {
      "epoch": 0.09679552696446822,
      "flos": 50720205496320.0,
      "grad_norm": 0.8812818846108662,
      "learning_rate": 3.953454520847964e-06,
      "loss": 0.8482,
      "num_input_tokens_seen": 16991390,
      "step": 805
    },
    {
      "epoch": 0.09691576985510732,
      "flos": 15590433607680.0,
      "grad_norm": 11.477977730346069,
      "learning_rate": 3.9532872948706395e-06,
      "loss": 0.9269,
      "num_input_tokens_seen": 17010605,
      "step": 806
    },
    {
      "epoch": 0.09703601274574641,
      "flos": 12729412546560.0,
      "grad_norm": 6.194556941829368,
      "learning_rate": 3.9531197725811845e-06,
      "loss": 1.0314,
      "num_input_tokens_seen": 17025710,
      "step": 807
    },
    {
      "epoch": 0.0971562556363855,
      "flos": 15745829191680.0,
      "grad_norm": 9.925861188678494,
      "learning_rate": 3.952951954005013e-06,
      "loss": 1.0776,
      "num_input_tokens_seen": 17045115,
      "step": 808
    },
    {
      "epoch": 0.0972764985270246,
      "flos": 18395486576640.0,
      "grad_norm": 7.64568886546785,
      "learning_rate": 3.952783839167584e-06,
      "loss": 1.0427,
      "num_input_tokens_seen": 17064880,
      "step": 809
    },
    {
      "epoch": 0.09739674141766368,
      "flos": 14436027248640.0,
      "grad_norm": 7.517461921029654,
      "learning_rate": 3.952615428094398e-06,
      "loss": 0.9279,
      "num_input_tokens_seen": 17084120,
      "step": 810
    },
    {
      "epoch": 0.09751698430830277,
      "flos": 11132863856640.0,
      "grad_norm": 4.6585183986600605,
      "learning_rate": 3.952446720811004e-06,
      "loss": 0.9426,
      "num_input_tokens_seen": 17102165,
      "step": 811
    },
    {
      "epoch": 0.09763722719894186,
      "flos": 45614932316160.0,
      "grad_norm": 0.902236073332304,
      "learning_rate": 3.952277717342995e-06,
      "loss": 0.8716,
      "num_input_tokens_seen": 17168320,
      "step": 812
    },
    {
      "epoch": 0.09775747008958095,
      "flos": 15694858444800.0,
      "grad_norm": 7.414482595201166,
      "learning_rate": 3.952108417716009e-06,
      "loss": 1.0573,
      "num_input_tokens_seen": 17187495,
      "step": 813
    },
    {
      "epoch": 0.09787771298022005,
      "flos": 15116122337280.0,
      "grad_norm": 6.032930604479617,
      "learning_rate": 3.951938821955727e-06,
      "loss": 1.0462,
      "num_input_tokens_seen": 17206615,
      "step": 814
    },
    {
      "epoch": 0.09799795587085913,
      "flos": 15668759900160.0,
      "grad_norm": 8.532141111456184,
      "learning_rate": 3.9517689300878786e-06,
      "loss": 0.9619,
      "num_input_tokens_seen": 17226070,
      "step": 815
    },
    {
      "epoch": 0.09811819876149823,
      "flos": 15747178168320.0,
      "grad_norm": 5.024165882324184,
      "learning_rate": 3.951598742138236e-06,
      "loss": 0.9827,
      "num_input_tokens_seen": 17244515,
      "step": 816
    },
    {
      "epoch": 0.09823844165213731,
      "flos": 15794837790720.0,
      "grad_norm": 5.1456539519186,
      "learning_rate": 3.951428258132615e-06,
      "loss": 0.9877,
      "num_input_tokens_seen": 17262355,
      "step": 817
    },
    {
      "epoch": 0.09835868454277641,
      "flos": 15979428433920.0,
      "grad_norm": 4.394766269757232,
      "learning_rate": 3.951257478096879e-06,
      "loss": 1.046,
      "num_input_tokens_seen": 17280440,
      "step": 818
    },
    {
      "epoch": 0.0984789274334155,
      "flos": 11577980067840.0,
      "grad_norm": 14.583213122644432,
      "learning_rate": 3.951086402056936e-06,
      "loss": 0.8846,
      "num_input_tokens_seen": 17294760,
      "step": 819
    },
    {
      "epoch": 0.09859917032405459,
      "flos": 17241907998720.0,
      "grad_norm": 7.483248249524646,
      "learning_rate": 3.950915030038735e-06,
      "loss": 1.0387,
      "num_input_tokens_seen": 17314275,
      "step": 820
    },
    {
      "epoch": 0.09871941321469369,
      "flos": 12338026352640.0,
      "grad_norm": 6.908316048607027,
      "learning_rate": 3.9507433620682765e-06,
      "loss": 1.0345,
      "num_input_tokens_seen": 17330930,
      "step": 821
    },
    {
      "epoch": 0.09883965610533277,
      "flos": 20284852408320.0,
      "grad_norm": 10.88541448614412,
      "learning_rate": 3.9505713981716e-06,
      "loss": 1.0828,
      "num_input_tokens_seen": 17353480,
      "step": 822
    },
    {
      "epoch": 0.09895989899597187,
      "flos": 16846597509120.0,
      "grad_norm": 6.379038111485399,
      "learning_rate": 3.950399138374795e-06,
      "loss": 1.0114,
      "num_input_tokens_seen": 17372280,
      "step": 823
    },
    {
      "epoch": 0.09908014188661095,
      "flos": 17555059875840.0,
      "grad_norm": 5.418265688679369,
      "learning_rate": 3.95022658270399e-06,
      "loss": 0.9376,
      "num_input_tokens_seen": 17392365,
      "step": 824
    },
    {
      "epoch": 0.09920038477725004,
      "flos": 9926321725440.0,
      "grad_norm": 5.358827790998616,
      "learning_rate": 3.9500537311853635e-06,
      "loss": 0.9795,
      "num_input_tokens_seen": 17410040,
      "step": 825
    },
    {
      "epoch": 0.09932062766788914,
      "flos": 9453911285760.0,
      "grad_norm": 5.786321611503118,
      "learning_rate": 3.949880583845136e-06,
      "loss": 1.0261,
      "num_input_tokens_seen": 17427835,
      "step": 826
    },
    {
      "epoch": 0.09944087055852822,
      "flos": 13833400012800.0,
      "grad_norm": 4.31029509417025,
      "learning_rate": 3.949707140709575e-06,
      "loss": 1.0143,
      "num_input_tokens_seen": 17447285,
      "step": 827
    },
    {
      "epoch": 0.09956111344916732,
      "flos": 12574476840960.0,
      "grad_norm": 8.723300416345214,
      "learning_rate": 3.949533401804991e-06,
      "loss": 1.0262,
      "num_input_tokens_seen": 17463910,
      "step": 828
    },
    {
      "epoch": 0.0996813563398064,
      "flos": 12731436011520.0,
      "grad_norm": 5.21248941031645,
      "learning_rate": 3.949359367157739e-06,
      "loss": 1.1125,
      "num_input_tokens_seen": 17482325,
      "step": 829
    },
    {
      "epoch": 0.0998015992304455,
      "flos": 12364554117120.0,
      "grad_norm": 9.29135619416013,
      "learning_rate": 3.949185036794222e-06,
      "loss": 0.9663,
      "num_input_tokens_seen": 17500055,
      "step": 830
    },
    {
      "epoch": 0.0999218421210846,
      "flos": 18424742952960.0,
      "grad_norm": 6.401570616165986,
      "learning_rate": 3.949010410740884e-06,
      "loss": 0.9726,
      "num_input_tokens_seen": 17522600,
      "step": 831
    },
    {
      "epoch": 0.10004208501172368,
      "flos": 15066102005760.0,
      "grad_norm": 6.244475707989886,
      "learning_rate": 3.948835489024216e-06,
      "loss": 1.0695,
      "num_input_tokens_seen": 17542055,
      "step": 832
    },
    {
      "epoch": 0.10016232790236278,
      "flos": 12286595727360.0,
      "grad_norm": 7.58524150385965,
      "learning_rate": 3.948660271670755e-06,
      "loss": 1.0896,
      "num_input_tokens_seen": 17558925,
      "step": 833
    },
    {
      "epoch": 0.10028257079300186,
      "flos": 18264595292160.0,
      "grad_norm": 7.41230417028288,
      "learning_rate": 3.948484758707079e-06,
      "loss": 1.0405,
      "num_input_tokens_seen": 17578245,
      "step": 834
    },
    {
      "epoch": 0.10040281368364096,
      "flos": 17897744056320.0,
      "grad_norm": 13.192470391265392,
      "learning_rate": 3.948308950159815e-06,
      "loss": 1.0413,
      "num_input_tokens_seen": 17596645,
      "step": 835
    },
    {
      "epoch": 0.10052305657428004,
      "flos": 12469714759680.0,
      "grad_norm": 12.374542536131008,
      "learning_rate": 3.9481328460556326e-06,
      "loss": 0.9501,
      "num_input_tokens_seen": 17613585,
      "step": 836
    },
    {
      "epoch": 0.10064329946491914,
      "flos": 13229086556160.0,
      "grad_norm": 20.840438648373997,
      "learning_rate": 3.9479564464212455e-06,
      "loss": 1.0762,
      "num_input_tokens_seen": 17632465,
      "step": 837
    },
    {
      "epoch": 0.10076354235555823,
      "flos": 12178798448640.0,
      "grad_norm": 5.010738476142408,
      "learning_rate": 3.947779751283414e-06,
      "loss": 0.967,
      "num_input_tokens_seen": 17649355,
      "step": 838
    },
    {
      "epoch": 0.10088378524619732,
      "flos": 16320487710720.0,
      "grad_norm": 5.930768182155559,
      "learning_rate": 3.947602760668944e-06,
      "loss": 0.9547,
      "num_input_tokens_seen": 17668865,
      "step": 839
    },
    {
      "epoch": 0.10100402813683641,
      "flos": 27045900288000.0,
      "grad_norm": 20.25276372440469,
      "learning_rate": 3.947425474604684e-06,
      "loss": 0.914,
      "num_input_tokens_seen": 17692520,
      "step": 840
    },
    {
      "epoch": 0.1011242710274755,
      "flos": 15301387468800.0,
      "grad_norm": 16.47515292986684,
      "learning_rate": 3.947247893117528e-06,
      "loss": 1.1221,
      "num_input_tokens_seen": 17710745,
      "step": 841
    },
    {
      "epoch": 0.10124451391811459,
      "flos": 9607375380480.0,
      "grad_norm": 16.47849976757119,
      "learning_rate": 3.947070016234413e-06,
      "loss": 0.8846,
      "num_input_tokens_seen": 17726255,
      "step": 842
    },
    {
      "epoch": 0.10136475680875369,
      "flos": 11783304007680.0,
      "grad_norm": 6.886018558420683,
      "learning_rate": 3.946891843982326e-06,
      "loss": 0.9437,
      "num_input_tokens_seen": 17743640,
      "step": 843
    },
    {
      "epoch": 0.10148499969939277,
      "flos": 13804419563520.0,
      "grad_norm": 16.511838422945786,
      "learning_rate": 3.9467133763882935e-06,
      "loss": 0.9504,
      "num_input_tokens_seen": 17761825,
      "step": 844
    },
    {
      "epoch": 0.10160524259003187,
      "flos": 14986181468160.0,
      "grad_norm": 12.428436504063852,
      "learning_rate": 3.9465346134793905e-06,
      "loss": 1.0581,
      "num_input_tokens_seen": 17781355,
      "step": 845
    },
    {
      "epoch": 0.10172548548067095,
      "flos": 12548316979200.0,
      "grad_norm": 5.506002964882238,
      "learning_rate": 3.9463555552827335e-06,
      "loss": 1.0028,
      "num_input_tokens_seen": 17798245,
      "step": 846
    },
    {
      "epoch": 0.10184572837131005,
      "flos": 14986058833920.0,
      "grad_norm": 10.259696281949655,
      "learning_rate": 3.946176201825487e-06,
      "loss": 1.0516,
      "num_input_tokens_seen": 17816000,
      "step": 847
    },
    {
      "epoch": 0.10196597126194913,
      "flos": 18552782991360.0,
      "grad_norm": 20.32645103839837,
      "learning_rate": 3.9459965531348575e-06,
      "loss": 1.0314,
      "num_input_tokens_seen": 17835375,
      "step": 848
    },
    {
      "epoch": 0.10208621415258823,
      "flos": 20886345277440.0,
      "grad_norm": 8.037995707886388,
      "learning_rate": 3.945816609238098e-06,
      "loss": 1.0466,
      "num_input_tokens_seen": 17854505,
      "step": 849
    },
    {
      "epoch": 0.10220645704322733,
      "flos": 16927315169280.0,
      "grad_norm": 5.990994370703806,
      "learning_rate": 3.945636370162507e-06,
      "loss": 1.0479,
      "num_input_tokens_seen": 17874335,
      "step": 850
    },
    {
      "epoch": 0.10232669993386641,
      "flos": 16504802426880.0,
      "grad_norm": 7.373249428095204,
      "learning_rate": 3.945455835935425e-06,
      "loss": 0.9945,
      "num_input_tokens_seen": 17893240,
      "step": 851
    },
    {
      "epoch": 0.1024469428245055,
      "flos": 16291844505600.0,
      "grad_norm": 9.162324003894632,
      "learning_rate": 3.94527500658424e-06,
      "loss": 0.9401,
      "num_input_tokens_seen": 17910625,
      "step": 852
    },
    {
      "epoch": 0.10256718571514459,
      "flos": 22356172247040.0,
      "grad_norm": 8.651257779619886,
      "learning_rate": 3.945093882136382e-06,
      "loss": 1.0057,
      "num_input_tokens_seen": 17934120,
      "step": 853
    },
    {
      "epoch": 0.10268742860578368,
      "flos": 16689147801600.0,
      "grad_norm": 4.28478345753366,
      "learning_rate": 3.944912462619329e-06,
      "loss": 1.0427,
      "num_input_tokens_seen": 17952805,
      "step": 854
    },
    {
      "epoch": 0.10280767149642277,
      "flos": 18159465308160.0,
      "grad_norm": 9.478954494463272,
      "learning_rate": 3.9447307480606025e-06,
      "loss": 1.0115,
      "num_input_tokens_seen": 17972610,
      "step": 855
    },
    {
      "epoch": 0.10292791438706186,
      "flos": 12285982556160.0,
      "grad_norm": 20.530917787694687,
      "learning_rate": 3.944548738487767e-06,
      "loss": 1.0974,
      "num_input_tokens_seen": 17989845,
      "step": 856
    },
    {
      "epoch": 0.10304815727770096,
      "flos": 19260754821120.0,
      "grad_norm": 7.078588205300871,
      "learning_rate": 3.944366433928434e-06,
      "loss": 1.1054,
      "num_input_tokens_seen": 18009545,
      "step": 857
    },
    {
      "epoch": 0.10316840016834004,
      "flos": 16191742525440.0,
      "grad_norm": 4.849883693598297,
      "learning_rate": 3.9441838344102594e-06,
      "loss": 1.0295,
      "num_input_tokens_seen": 18028990,
      "step": 858
    },
    {
      "epoch": 0.10328864305897914,
      "flos": 14698545623040.0,
      "grad_norm": 10.310847849426517,
      "learning_rate": 3.944000939960943e-06,
      "loss": 0.8807,
      "num_input_tokens_seen": 18047435,
      "step": 859
    },
    {
      "epoch": 0.10340888594961822,
      "flos": 20286109409280.0,
      "grad_norm": 6.282840053039714,
      "learning_rate": 3.943817750608229e-06,
      "loss": 1.0063,
      "num_input_tokens_seen": 18069705,
      "step": 860
    },
    {
      "epoch": 0.10352912884025732,
      "flos": 9427475496960.0,
      "grad_norm": 9.081496119987266,
      "learning_rate": 3.943634266379908e-06,
      "loss": 1.0081,
      "num_input_tokens_seen": 18086320,
      "step": 861
    },
    {
      "epoch": 0.10364937173089642,
      "flos": 18187188756480.0,
      "grad_norm": 5.498710828027209,
      "learning_rate": 3.943450487303815e-06,
      "loss": 1.0494,
      "num_input_tokens_seen": 18106535,
      "step": 862
    },
    {
      "epoch": 0.1037696146215355,
      "flos": 15065734103040.0,
      "grad_norm": 32.28331650270868,
      "learning_rate": 3.943266413407827e-06,
      "loss": 1.0463,
      "num_input_tokens_seen": 18125530,
      "step": 863
    },
    {
      "epoch": 0.1038898575121746,
      "flos": 18365126492160.0,
      "grad_norm": 7.23896818406277,
      "learning_rate": 3.94308204471987e-06,
      "loss": 1.0417,
      "num_input_tokens_seen": 18144265,
      "step": 864
    },
    {
      "epoch": 0.10401010040281368,
      "flos": 13518316646400.0,
      "grad_norm": 8.63850208437312,
      "learning_rate": 3.942897381267912e-06,
      "loss": 0.9416,
      "num_input_tokens_seen": 18160350,
      "step": 865
    },
    {
      "epoch": 0.10413034329345278,
      "flos": 11571633745920.0,
      "grad_norm": 14.70778965508026,
      "learning_rate": 3.942712423079965e-06,
      "loss": 0.8764,
      "num_input_tokens_seen": 18176460,
      "step": 866
    },
    {
      "epoch": 0.10425058618409186,
      "flos": 12205847408640.0,
      "grad_norm": 12.283349554153837,
      "learning_rate": 3.942527170184088e-06,
      "loss": 1.1067,
      "num_input_tokens_seen": 18192800,
      "step": 867
    },
    {
      "epoch": 0.10437082907473096,
      "flos": 12731252060160.0,
      "grad_norm": 32.24231338568834,
      "learning_rate": 3.942341622608385e-06,
      "loss": 0.9733,
      "num_input_tokens_seen": 18209550,
      "step": 868
    },
    {
      "epoch": 0.10449107196537005,
      "flos": 25894590443520.0,
      "grad_norm": 9.62615286127804,
      "learning_rate": 3.942155780381001e-06,
      "loss": 0.9701,
      "num_input_tokens_seen": 18233005,
      "step": 869
    },
    {
      "epoch": 0.10461131485600914,
      "flos": 16924157337600.0,
      "grad_norm": 3.7887960753223995,
      "learning_rate": 3.94196964353013e-06,
      "loss": 0.9574,
      "num_input_tokens_seen": 18252175,
      "step": 870
    },
    {
      "epoch": 0.10473155774664823,
      "flos": 13046151475200.0,
      "grad_norm": 7.810956390474006,
      "learning_rate": 3.941783212084008e-06,
      "loss": 0.9985,
      "num_input_tokens_seen": 18269650,
      "step": 871
    },
    {
      "epoch": 0.10485180063728732,
      "flos": 18211355811840.0,
      "grad_norm": 17.07649076553009,
      "learning_rate": 3.941596486070916e-06,
      "loss": 0.9659,
      "num_input_tokens_seen": 18287415,
      "step": 872
    },
    {
      "epoch": 0.10497204352792641,
      "flos": 19286608097280.0,
      "grad_norm": 8.920054662562638,
      "learning_rate": 3.941409465519182e-06,
      "loss": 0.7722,
      "num_input_tokens_seen": 18307660,
      "step": 873
    },
    {
      "epoch": 0.10509228641856551,
      "flos": 23433877217280.0,
      "grad_norm": 8.105296496663655,
      "learning_rate": 3.941222150457176e-06,
      "loss": 1.0532,
      "num_input_tokens_seen": 18330635,
      "step": 874
    },
    {
      "epoch": 0.10521252930920459,
      "flos": 10109410099200.0,
      "grad_norm": 21.19177527040451,
      "learning_rate": 3.941034540913311e-06,
      "loss": 0.908,
      "num_input_tokens_seen": 18347885,
      "step": 875
    },
    {
      "epoch": 0.10533277219984369,
      "flos": 15404248719360.0,
      "grad_norm": 10.28539621527415,
      "learning_rate": 3.940846636916051e-06,
      "loss": 1.0204,
      "num_input_tokens_seen": 18367640,
      "step": 876
    },
    {
      "epoch": 0.10545301509048277,
      "flos": 15822929141760.0,
      "grad_norm": 6.945986668893294,
      "learning_rate": 3.940658438493899e-06,
      "loss": 1.0557,
      "num_input_tokens_seen": 18385205,
      "step": 877
    },
    {
      "epoch": 0.10557325798112187,
      "flos": 15772970127360.0,
      "grad_norm": 8.311290681681767,
      "learning_rate": 3.940469945675405e-06,
      "loss": 0.9607,
      "num_input_tokens_seen": 18403310,
      "step": 878
    },
    {
      "epoch": 0.10569350087176095,
      "flos": 18343166853120.0,
      "grad_norm": 10.314936679455277,
      "learning_rate": 3.940281158489163e-06,
      "loss": 1.1102,
      "num_input_tokens_seen": 18422260,
      "step": 879
    },
    {
      "epoch": 0.10581374376240005,
      "flos": 12259914670080.0,
      "grad_norm": 13.794263028697213,
      "learning_rate": 3.940092076963812e-06,
      "loss": 1.023,
      "num_input_tokens_seen": 18439475,
      "step": 880
    },
    {
      "epoch": 0.10593398665303914,
      "flos": 24504745328640.0,
      "grad_norm": 10.877295942310367,
      "learning_rate": 3.9399027011280355e-06,
      "loss": 0.9891,
      "num_input_tokens_seen": 18461290,
      "step": 881
    },
    {
      "epoch": 0.10605422954367823,
      "flos": 16533537607680.0,
      "grad_norm": 10.752718204784495,
      "learning_rate": 3.939713031010561e-06,
      "loss": 0.9653,
      "num_input_tokens_seen": 18479375,
      "step": 882
    },
    {
      "epoch": 0.10617447243431732,
      "flos": 16218852802560.0,
      "grad_norm": 13.6992844666231,
      "learning_rate": 3.939523066640163e-06,
      "loss": 0.9765,
      "num_input_tokens_seen": 18497990,
      "step": 883
    },
    {
      "epoch": 0.10629471532495641,
      "flos": 17344309370880.0,
      "grad_norm": 7.583439110098548,
      "learning_rate": 3.939332808045657e-06,
      "loss": 1.0087,
      "num_input_tokens_seen": 18517360,
      "step": 884
    },
    {
      "epoch": 0.1064149582155955,
      "flos": 14986610688000.0,
      "grad_norm": 4.292204937311003,
      "learning_rate": 3.939142255255906e-06,
      "loss": 1.0347,
      "num_input_tokens_seen": 18537965,
      "step": 885
    },
    {
      "epoch": 0.1065352011062346,
      "flos": 14696368865280.0,
      "grad_norm": 6.763586992771591,
      "learning_rate": 3.938951408299817e-06,
      "loss": 1.0764,
      "num_input_tokens_seen": 18556525,
      "step": 886
    },
    {
      "epoch": 0.10665544399687368,
      "flos": 44855284592640.0,
      "grad_norm": 0.8570047810865149,
      "learning_rate": 3.938760267206342e-06,
      "loss": 0.7785,
      "num_input_tokens_seen": 18618065,
      "step": 887
    },
    {
      "epoch": 0.10677568688751278,
      "flos": 18605072056320.0,
      "grad_norm": 8.449857125964838,
      "learning_rate": 3.938568832004475e-06,
      "loss": 0.9729,
      "num_input_tokens_seen": 18636490,
      "step": 888
    },
    {
      "epoch": 0.10689592977815186,
      "flos": 9007905976320.0,
      "grad_norm": 6.012115961130334,
      "learning_rate": 3.938377102723257e-06,
      "loss": 0.9528,
      "num_input_tokens_seen": 18653345,
      "step": 889
    },
    {
      "epoch": 0.10701617266879096,
      "flos": 15720466452480.0,
      "grad_norm": 6.160522217846636,
      "learning_rate": 3.938185079391774e-06,
      "loss": 1.0374,
      "num_input_tokens_seen": 18670110,
      "step": 890
    },
    {
      "epoch": 0.10713641555943004,
      "flos": 14007965306880.0,
      "grad_norm": 14.71645199600046,
      "learning_rate": 3.937992762039157e-06,
      "loss": 1.2599,
      "num_input_tokens_seen": 18683155,
      "step": 891
    },
    {
      "epoch": 0.10725665845006914,
      "flos": 17033272934400.0,
      "grad_norm": 4.569691255520045,
      "learning_rate": 3.937800150694577e-06,
      "loss": 0.9969,
      "num_input_tokens_seen": 18704050,
      "step": 892
    },
    {
      "epoch": 0.10737690134070824,
      "flos": 13151281459200.0,
      "grad_norm": 9.241791665894702,
      "learning_rate": 3.937607245387255e-06,
      "loss": 0.9499,
      "num_input_tokens_seen": 18723135,
      "step": 893
    },
    {
      "epoch": 0.10749714423134732,
      "flos": 16137583288320.0,
      "grad_norm": 5.296765531685941,
      "learning_rate": 3.937414046146455e-06,
      "loss": 0.914,
      "num_input_tokens_seen": 18740810,
      "step": 894
    },
    {
      "epoch": 0.10761738712198642,
      "flos": 14987101224960.0,
      "grad_norm": 6.417033428593125,
      "learning_rate": 3.9372205530014845e-06,
      "loss": 0.9543,
      "num_input_tokens_seen": 18759010,
      "step": 895
    },
    {
      "epoch": 0.1077376300126255,
      "flos": 16898886574080.0,
      "grad_norm": 11.111598139791361,
      "learning_rate": 3.937026765981696e-06,
      "loss": 0.9322,
      "num_input_tokens_seen": 18778800,
      "step": 896
    },
    {
      "epoch": 0.1078578729032646,
      "flos": 14852530913280.0,
      "grad_norm": 6.657818963048377,
      "learning_rate": 3.936832685116488e-06,
      "loss": 0.9856,
      "num_input_tokens_seen": 18796615,
      "step": 897
    },
    {
      "epoch": 0.10797811579390369,
      "flos": 10476445286400.0,
      "grad_norm": 11.051331480425079,
      "learning_rate": 3.936638310435301e-06,
      "loss": 1.0911,
      "num_input_tokens_seen": 18814200,
      "step": 898
    },
    {
      "epoch": 0.10809835868454278,
      "flos": 13859406581760.0,
      "grad_norm": 5.896992851337087,
      "learning_rate": 3.936443641967623e-06,
      "loss": 1.0232,
      "num_input_tokens_seen": 18832750,
      "step": 899
    },
    {
      "epoch": 0.10821860157518187,
      "flos": 13072679239680.0,
      "grad_norm": 8.993993662276527,
      "learning_rate": 3.936248679742983e-06,
      "loss": 1.0273,
      "num_input_tokens_seen": 18850965,
      "step": 900
    },
    {
      "epoch": 0.10833884446582095,
      "flos": 35294097223680.0,
      "grad_norm": 1.041537556594223,
      "learning_rate": 3.936053423790959e-06,
      "loss": 0.9522,
      "num_input_tokens_seen": 18899005,
      "step": 901
    },
    {
      "epoch": 0.10845908735646005,
      "flos": 14488101703680.0,
      "grad_norm": 17.53439525874342,
      "learning_rate": 3.935857874141168e-06,
      "loss": 0.967,
      "num_input_tokens_seen": 18917560,
      "step": 902
    },
    {
      "epoch": 0.10857933024709913,
      "flos": 9900039229440.0,
      "grad_norm": 8.532201022284225,
      "learning_rate": 3.935662030823279e-06,
      "loss": 1.0354,
      "num_input_tokens_seen": 18933465,
      "step": 903
    },
    {
      "epoch": 0.10869957313773823,
      "flos": 9426341130240.0,
      "grad_norm": 22.171696754260612,
      "learning_rate": 3.935465893866998e-06,
      "loss": 0.9048,
      "num_input_tokens_seen": 18951410,
      "step": 904
    },
    {
      "epoch": 0.10881981602837733,
      "flos": 18364881223680.0,
      "grad_norm": 16.280618314881156,
      "learning_rate": 3.935269463302079e-06,
      "loss": 1.0063,
      "num_input_tokens_seen": 18969335,
      "step": 905
    },
    {
      "epoch": 0.10894005891901641,
      "flos": 14750987980800.0,
      "grad_norm": 6.925264858868469,
      "learning_rate": 3.935072739158322e-06,
      "loss": 0.9529,
      "num_input_tokens_seen": 18988765,
      "step": 906
    },
    {
      "epoch": 0.10906030180965551,
      "flos": 18971064852480.0,
      "grad_norm": 23.99440064279646,
      "learning_rate": 3.934875721465569e-06,
      "loss": 0.9941,
      "num_input_tokens_seen": 19008905,
      "step": 907
    },
    {
      "epoch": 0.10918054470029459,
      "flos": 26074459668480.0,
      "grad_norm": 9.230851432238659,
      "learning_rate": 3.9346784102537076e-06,
      "loss": 0.8911,
      "num_input_tokens_seen": 19030760,
      "step": 908
    },
    {
      "epoch": 0.10930078759093369,
      "flos": 15458867834880.0,
      "grad_norm": 5.67218605218444,
      "learning_rate": 3.934480805552669e-06,
      "loss": 0.9765,
      "num_input_tokens_seen": 19051490,
      "step": 909
    },
    {
      "epoch": 0.10942103048157277,
      "flos": 16062016266240.0,
      "grad_norm": 6.595347423411298,
      "learning_rate": 3.93428290739243e-06,
      "loss": 1.0718,
      "num_input_tokens_seen": 19070580,
      "step": 910
    },
    {
      "epoch": 0.10954127337221187,
      "flos": 10631197040640.0,
      "grad_norm": 8.232909562586162,
      "learning_rate": 3.9340847158030125e-06,
      "loss": 0.9944,
      "num_input_tokens_seen": 19083880,
      "step": 911
    },
    {
      "epoch": 0.10966151626285096,
      "flos": 15378395443200.0,
      "grad_norm": 7.493307815499258,
      "learning_rate": 3.9338862308144814e-06,
      "loss": 0.9611,
      "num_input_tokens_seen": 19102420,
      "step": 912
    },
    {
      "epoch": 0.10978175915349005,
      "flos": 14279466639360.0,
      "grad_norm": 8.489132385134758,
      "learning_rate": 3.933687452456946e-06,
      "loss": 1.0488,
      "num_input_tokens_seen": 19122040,
      "step": 913
    },
    {
      "epoch": 0.10990200204412914,
      "flos": 14488745533440.0,
      "grad_norm": 10.516469482249914,
      "learning_rate": 3.933488380760562e-06,
      "loss": 1.0527,
      "num_input_tokens_seen": 19141120,
      "step": 914
    },
    {
      "epoch": 0.11002224493476823,
      "flos": 12417088450560.0,
      "grad_norm": 6.542579279453149,
      "learning_rate": 3.9332890157555286e-06,
      "loss": 1.0772,
      "num_input_tokens_seen": 19157775,
      "step": 915
    },
    {
      "epoch": 0.11014248782540732,
      "flos": 8588643041280.0,
      "grad_norm": 12.158384317173955,
      "learning_rate": 3.933089357472088e-06,
      "loss": 0.9669,
      "num_input_tokens_seen": 19175525,
      "step": 916
    },
    {
      "epoch": 0.11026273071604642,
      "flos": 15904781168640.0,
      "grad_norm": 5.267908523715403,
      "learning_rate": 3.932889405940529e-06,
      "loss": 1.0597,
      "num_input_tokens_seen": 19193340,
      "step": 917
    },
    {
      "epoch": 0.1103829736066855,
      "flos": 14069329305600.0,
      "grad_norm": 11.156977952987853,
      "learning_rate": 3.932689161191184e-06,
      "loss": 1.0059,
      "num_input_tokens_seen": 19210765,
      "step": 918
    },
    {
      "epoch": 0.1105032164973246,
      "flos": 16110074449920.0,
      "grad_norm": 6.622927071498279,
      "learning_rate": 3.93248862325443e-06,
      "loss": 1.0736,
      "num_input_tokens_seen": 19229390,
      "step": 919
    },
    {
      "epoch": 0.11062345938796368,
      "flos": 47602468638720.0,
      "grad_norm": 1.067642333929412,
      "learning_rate": 3.932287792160688e-06,
      "loss": 0.852,
      "num_input_tokens_seen": 19287570,
      "step": 920
    },
    {
      "epoch": 0.11074370227860278,
      "flos": 15563139379200.0,
      "grad_norm": 80.33880588152122,
      "learning_rate": 3.932086667940424e-06,
      "loss": 1.0021,
      "num_input_tokens_seen": 19303995,
      "step": 921
    },
    {
      "epoch": 0.11086394516924186,
      "flos": 20414548008960.0,
      "grad_norm": 9.579190317871134,
      "learning_rate": 3.93188525062415e-06,
      "loss": 1.0188,
      "num_input_tokens_seen": 19324180,
      "step": 922
    },
    {
      "epoch": 0.11098418805988096,
      "flos": 17451524136960.0,
      "grad_norm": 7.220777350764695,
      "learning_rate": 3.931683540242418e-06,
      "loss": 1.067,
      "num_input_tokens_seen": 19344965,
      "step": 923
    },
    {
      "epoch": 0.11110443095052006,
      "flos": 16318893465600.0,
      "grad_norm": 45.45677999656652,
      "learning_rate": 3.9314815368258295e-06,
      "loss": 1.0998,
      "num_input_tokens_seen": 19361165,
      "step": 924
    },
    {
      "epoch": 0.11122467384115914,
      "flos": 13437139107840.0,
      "grad_norm": 10.785191266128416,
      "learning_rate": 3.9312792404050275e-06,
      "loss": 0.975,
      "num_input_tokens_seen": 19378940,
      "step": 925
    },
    {
      "epoch": 0.11134491673179824,
      "flos": 17845209722880.0,
      "grad_norm": 8.682247193025592,
      "learning_rate": 3.9310766510107e-06,
      "loss": 0.9766,
      "num_input_tokens_seen": 19397835,
      "step": 926
    },
    {
      "epoch": 0.11146515962243732,
      "flos": 17425609543680.0,
      "grad_norm": 8.567282573166159,
      "learning_rate": 3.9308737686735806e-06,
      "loss": 1.1313,
      "num_input_tokens_seen": 19417515,
      "step": 927
    },
    {
      "epoch": 0.11158540251307641,
      "flos": 15876229939200.0,
      "grad_norm": 9.583102104604,
      "learning_rate": 3.9306705934244455e-06,
      "loss": 1.0271,
      "num_input_tokens_seen": 19437315,
      "step": 928
    },
    {
      "epoch": 0.11170564540371551,
      "flos": 14122016931840.0,
      "grad_norm": 15.999369132002371,
      "learning_rate": 3.930467125294116e-06,
      "loss": 1.0825,
      "num_input_tokens_seen": 19456585,
      "step": 929
    },
    {
      "epoch": 0.1118258882943546,
      "flos": 46238936678400.0,
      "grad_norm": 1.0982172608196796,
      "learning_rate": 3.930263364313458e-06,
      "loss": 0.8381,
      "num_input_tokens_seen": 19506875,
      "step": 930
    },
    {
      "epoch": 0.11194613118499369,
      "flos": 12181281792000.0,
      "grad_norm": 6.214477298082189,
      "learning_rate": 3.930059310513384e-06,
      "loss": 1.0405,
      "num_input_tokens_seen": 19525635,
      "step": 931
    },
    {
      "epoch": 0.11206637407563277,
      "flos": 22717443624960.0,
      "grad_norm": 3.756929607130936,
      "learning_rate": 3.929854963924846e-06,
      "loss": 1.033,
      "num_input_tokens_seen": 19545620,
      "step": 932
    },
    {
      "epoch": 0.11218661696627187,
      "flos": 15590525583360.0,
      "grad_norm": 7.3288277165070665,
      "learning_rate": 3.929650324578845e-06,
      "loss": 0.9709,
      "num_input_tokens_seen": 19564805,
      "step": 933
    },
    {
      "epoch": 0.11230685985691095,
      "flos": 18415729336320.0,
      "grad_norm": 8.31059576035857,
      "learning_rate": 3.929445392506423e-06,
      "loss": 1.0208,
      "num_input_tokens_seen": 19582465,
      "step": 934
    },
    {
      "epoch": 0.11242710274755005,
      "flos": 15795175034880.0,
      "grad_norm": 6.319510014045943,
      "learning_rate": 3.92924016773867e-06,
      "loss": 0.9666,
      "num_input_tokens_seen": 19598680,
      "step": 935
    },
    {
      "epoch": 0.11254734563818915,
      "flos": 12547887759360.0,
      "grad_norm": 13.004536937431922,
      "learning_rate": 3.9290346503067175e-06,
      "loss": 0.9362,
      "num_input_tokens_seen": 19615065,
      "step": 936
    },
    {
      "epoch": 0.11266758852882823,
      "flos": 39296280145920.0,
      "grad_norm": 11.054641259841455,
      "learning_rate": 3.9288288402417415e-06,
      "loss": 0.9881,
      "num_input_tokens_seen": 19641045,
      "step": 937
    },
    {
      "epoch": 0.11278783141946733,
      "flos": 13385616506880.0,
      "grad_norm": 13.413374850421862,
      "learning_rate": 3.928622737574964e-06,
      "loss": 0.8961,
      "num_input_tokens_seen": 19656100,
      "step": 938
    },
    {
      "epoch": 0.11290807431010641,
      "flos": 18841491886080.0,
      "grad_norm": 6.281069017526371,
      "learning_rate": 3.928416342337652e-06,
      "loss": 1.1157,
      "num_input_tokens_seen": 19675555,
      "step": 939
    },
    {
      "epoch": 0.1130283172007455,
      "flos": 16140127948800.0,
      "grad_norm": 10.699613788800612,
      "learning_rate": 3.928209654561113e-06,
      "loss": 1.0318,
      "num_input_tokens_seen": 19696110,
      "step": 940
    },
    {
      "epoch": 0.1131485600913846,
      "flos": 16505967452160.0,
      "grad_norm": 5.72874968889782,
      "learning_rate": 3.928002674276703e-06,
      "loss": 1.0044,
      "num_input_tokens_seen": 19715220,
      "step": 941
    },
    {
      "epoch": 0.11326880298202369,
      "flos": 9926291066880.0,
      "grad_norm": 7.918750876748052,
      "learning_rate": 3.92779540151582e-06,
      "loss": 0.956,
      "num_input_tokens_seen": 19732025,
      "step": 942
    },
    {
      "epoch": 0.11338904587266278,
      "flos": 11551176376320.0,
      "grad_norm": 8.377322657753453,
      "learning_rate": 3.927587836309907e-06,
      "loss": 1.0572,
      "num_input_tokens_seen": 19749575,
      "step": 943
    },
    {
      "epoch": 0.11350928876330187,
      "flos": 17373259161600.0,
      "grad_norm": 5.1450100547759,
      "learning_rate": 3.927379978690452e-06,
      "loss": 0.9743,
      "num_input_tokens_seen": 19768560,
      "step": 944
    },
    {
      "epoch": 0.11362953165394096,
      "flos": 17423862005760.0,
      "grad_norm": 9.168698542267839,
      "learning_rate": 3.927171828688987e-06,
      "loss": 1.0634,
      "num_input_tokens_seen": 19787805,
      "step": 945
    },
    {
      "epoch": 0.11374977454458005,
      "flos": 17110556835840.0,
      "grad_norm": 7.052945315836061,
      "learning_rate": 3.926963386337088e-06,
      "loss": 1.0133,
      "num_input_tokens_seen": 19805755,
      "step": 946
    },
    {
      "epoch": 0.11387001743521914,
      "flos": 28175924981760.0,
      "grad_norm": 9.552614375460259,
      "learning_rate": 3.926754651666375e-06,
      "loss": 0.9037,
      "num_input_tokens_seen": 19826035,
      "step": 947
    },
    {
      "epoch": 0.11399026032585824,
      "flos": 17844780503040.0,
      "grad_norm": 8.155499303186614,
      "learning_rate": 3.926545624708513e-06,
      "loss": 0.9686,
      "num_input_tokens_seen": 19844995,
      "step": 948
    },
    {
      "epoch": 0.11411050321649732,
      "flos": 12726530641920.0,
      "grad_norm": 7.217155775920074,
      "learning_rate": 3.926336305495213e-06,
      "loss": 1.0634,
      "num_input_tokens_seen": 19863275,
      "step": 949
    },
    {
      "epoch": 0.11423074610713642,
      "flos": 15956763648000.0,
      "grad_norm": 6.126842922183923,
      "learning_rate": 3.926126694058226e-06,
      "loss": 1.0793,
      "num_input_tokens_seen": 19882145,
      "step": 950
    },
    {
      "epoch": 0.1143509889977755,
      "flos": 13989040865280.0,
      "grad_norm": 6.135316529506553,
      "learning_rate": 3.92591679042935e-06,
      "loss": 1.0167,
      "num_input_tokens_seen": 19901755,
      "step": 951
    },
    {
      "epoch": 0.1144712318884146,
      "flos": 14064056033280.0,
      "grad_norm": 9.772377743428358,
      "learning_rate": 3.92570659464043e-06,
      "loss": 1.0252,
      "num_input_tokens_seen": 19919535,
      "step": 952
    },
    {
      "epoch": 0.1145914747790537,
      "flos": 10554986188800.0,
      "grad_norm": 8.835189720236805,
      "learning_rate": 3.925496106723349e-06,
      "loss": 0.9991,
      "num_input_tokens_seen": 19936695,
      "step": 953
    },
    {
      "epoch": 0.11471171766969278,
      "flos": 14095121264640.0,
      "grad_norm": 8.111514593856542,
      "learning_rate": 3.9252853267100405e-06,
      "loss": 1.0359,
      "num_input_tokens_seen": 19955660,
      "step": 954
    },
    {
      "epoch": 0.11483196056033187,
      "flos": 16007581102080.0,
      "grad_norm": 6.626038379985216,
      "learning_rate": 3.9250742546324786e-06,
      "loss": 1.0363,
      "num_input_tokens_seen": 19975615,
      "step": 955
    },
    {
      "epoch": 0.11495220345097096,
      "flos": 20099893862400.0,
      "grad_norm": 3.44161948540393,
      "learning_rate": 3.924862890522683e-06,
      "loss": 1.0639,
      "num_input_tokens_seen": 19995345,
      "step": 956
    },
    {
      "epoch": 0.11507244634161005,
      "flos": 12389702246400.0,
      "grad_norm": 6.576960154715017,
      "learning_rate": 3.9246512344127174e-06,
      "loss": 1.0544,
      "num_input_tokens_seen": 20012725,
      "step": 957
    },
    {
      "epoch": 0.11519268923224914,
      "flos": 16035488501760.0,
      "grad_norm": 6.7285516259015115,
      "learning_rate": 3.9244392863346895e-06,
      "loss": 1.0239,
      "num_input_tokens_seen": 20031850,
      "step": 958
    },
    {
      "epoch": 0.11531293212288823,
      "flos": 12023770767360.0,
      "grad_norm": 7.927601085969922,
      "learning_rate": 3.9242270463207524e-06,
      "loss": 1.103,
      "num_input_tokens_seen": 20049960,
      "step": 959
    },
    {
      "epoch": 0.11543317501352733,
      "flos": 8745970114560.0,
      "grad_norm": 13.938144822877815,
      "learning_rate": 3.924014514403102e-06,
      "loss": 1.0378,
      "num_input_tokens_seen": 20065835,
      "step": 960
    },
    {
      "epoch": 0.11555341790416641,
      "flos": 14063013642240.0,
      "grad_norm": 6.481797774736429,
      "learning_rate": 3.92380169061398e-06,
      "loss": 1.1216,
      "num_input_tokens_seen": 20083335,
      "step": 961
    },
    {
      "epoch": 0.11567366079480551,
      "flos": 18316485795840.0,
      "grad_norm": 4.014316211966642,
      "learning_rate": 3.9235885749856705e-06,
      "loss": 1.036,
      "num_input_tokens_seen": 20101735,
      "step": 962
    },
    {
      "epoch": 0.1157939036854446,
      "flos": 12915260190720.0,
      "grad_norm": 16.99851893536886,
      "learning_rate": 3.9233751675505035e-06,
      "loss": 1.0298,
      "num_input_tokens_seen": 20120165,
      "step": 963
    },
    {
      "epoch": 0.11591414657608369,
      "flos": 16400898785280.0,
      "grad_norm": 14.033396533331436,
      "learning_rate": 3.923161468340853e-06,
      "loss": 1.0331,
      "num_input_tokens_seen": 20139720,
      "step": 964
    },
    {
      "epoch": 0.11603438946672277,
      "flos": 13805002076160.0,
      "grad_norm": 5.084192026289002,
      "learning_rate": 3.9229474773891374e-06,
      "loss": 1.0119,
      "num_input_tokens_seen": 20157980,
      "step": 965
    },
    {
      "epoch": 0.11615463235736187,
      "flos": 19101894819840.0,
      "grad_norm": 5.144206275031479,
      "learning_rate": 3.922733194727818e-06,
      "loss": 1.0444,
      "num_input_tokens_seen": 20177495,
      "step": 966
    },
    {
      "epoch": 0.11627487524800097,
      "flos": 13170788413440.0,
      "grad_norm": 4.016662226208862,
      "learning_rate": 3.922518620389402e-06,
      "loss": 1.0405,
      "num_input_tokens_seen": 20194080,
      "step": 967
    },
    {
      "epoch": 0.11639511813864005,
      "flos": 12862940467200.0,
      "grad_norm": 3.2621868231698037,
      "learning_rate": 3.922303754406439e-06,
      "loss": 1.1013,
      "num_input_tokens_seen": 20211640,
      "step": 968
    },
    {
      "epoch": 0.11651536102927915,
      "flos": 14854615695360.0,
      "grad_norm": 10.837609567055352,
      "learning_rate": 3.922088596811526e-06,
      "loss": 0.9882,
      "num_input_tokens_seen": 20230490,
      "step": 969
    },
    {
      "epoch": 0.11663560391991823,
      "flos": 11682956759040.0,
      "grad_norm": 26.91839247643932,
      "learning_rate": 3.9218731476373e-06,
      "loss": 1.0571,
      "num_input_tokens_seen": 20246395,
      "step": 970
    },
    {
      "epoch": 0.11675584681055733,
      "flos": 14095121264640.0,
      "grad_norm": 16.732123557404016,
      "learning_rate": 3.9216574069164455e-06,
      "loss": 1.0495,
      "num_input_tokens_seen": 20265090,
      "step": 971
    },
    {
      "epoch": 0.11687608970119642,
      "flos": 15589452533760.0,
      "grad_norm": 5.446821619364253,
      "learning_rate": 3.921441374681691e-06,
      "loss": 1.0073,
      "num_input_tokens_seen": 20284870,
      "step": 972
    },
    {
      "epoch": 0.1169963325918355,
      "flos": 17110832762880.0,
      "grad_norm": 3.9988071331916415,
      "learning_rate": 3.921225050965808e-06,
      "loss": 0.8483,
      "num_input_tokens_seen": 20304475,
      "step": 973
    },
    {
      "epoch": 0.1171165754824746,
      "flos": 16612814315520.0,
      "grad_norm": 7.981848389755409,
      "learning_rate": 3.921008435801612e-06,
      "loss": 0.9331,
      "num_input_tokens_seen": 20323280,
      "step": 974
    },
    {
      "epoch": 0.11723681837311369,
      "flos": 13151741337600.0,
      "grad_norm": 5.810073510130077,
      "learning_rate": 3.920791529221963e-06,
      "loss": 0.9462,
      "num_input_tokens_seen": 20341675,
      "step": 975
    },
    {
      "epoch": 0.11735706126375278,
      "flos": 16743889551360.0,
      "grad_norm": 4.043529644818421,
      "learning_rate": 3.920574331259768e-06,
      "loss": 0.9622,
      "num_input_tokens_seen": 20362595,
      "step": 976
    },
    {
      "epoch": 0.11747730415439187,
      "flos": 15903554826240.0,
      "grad_norm": 18.778676800897824,
      "learning_rate": 3.9203568419479716e-06,
      "loss": 0.986,
      "num_input_tokens_seen": 20382870,
      "step": 977
    },
    {
      "epoch": 0.11759754704503096,
      "flos": 15773828567040.0,
      "grad_norm": 7.209333611325305,
      "learning_rate": 3.92013906131957e-06,
      "loss": 0.9387,
      "num_input_tokens_seen": 20401520,
      "step": 978
    },
    {
      "epoch": 0.11771778993567006,
      "flos": 15852124200960.0,
      "grad_norm": 5.48164599737854,
      "learning_rate": 3.9199209894076e-06,
      "loss": 1.0117,
      "num_input_tokens_seen": 20421555,
      "step": 979
    },
    {
      "epoch": 0.11783803282630914,
      "flos": 15117931192320.0,
      "grad_norm": 29.748283785515994,
      "learning_rate": 3.919702626245142e-06,
      "loss": 1.095,
      "num_input_tokens_seen": 20440930,
      "step": 980
    },
    {
      "epoch": 0.11795827571694824,
      "flos": 18052158566400.0,
      "grad_norm": 5.873028038712727,
      "learning_rate": 3.919483971865322e-06,
      "loss": 0.846,
      "num_input_tokens_seen": 20460645,
      "step": 981
    },
    {
      "epoch": 0.11807851860758732,
      "flos": 16795228200960.0,
      "grad_norm": 5.039749075077657,
      "learning_rate": 3.91926502630131e-06,
      "loss": 1.06,
      "num_input_tokens_seen": 20480980,
      "step": 982
    },
    {
      "epoch": 0.11819876149822642,
      "flos": 17765687746560.0,
      "grad_norm": 6.354975647838404,
      "learning_rate": 3.91904578958632e-06,
      "loss": 0.9191,
      "num_input_tokens_seen": 20500115,
      "step": 983
    },
    {
      "epoch": 0.11831900438886551,
      "flos": 16350541209600.0,
      "grad_norm": 5.386362884252193,
      "learning_rate": 3.918826261753608e-06,
      "loss": 1.0476,
      "num_input_tokens_seen": 20519415,
      "step": 984
    },
    {
      "epoch": 0.1184392472795046,
      "flos": 19916560220160.0,
      "grad_norm": 10.945311454429822,
      "learning_rate": 3.918606442836478e-06,
      "loss": 0.8879,
      "num_input_tokens_seen": 20541355,
      "step": 985
    },
    {
      "epoch": 0.1185594901701437,
      "flos": 14118767124480.0,
      "grad_norm": 9.91574421500486,
      "learning_rate": 3.918386332868277e-06,
      "loss": 0.9756,
      "num_input_tokens_seen": 20559045,
      "step": 986
    },
    {
      "epoch": 0.11867973306078278,
      "flos": 13409906196480.0,
      "grad_norm": 9.5909165164996,
      "learning_rate": 3.918165931882394e-06,
      "loss": 1.1471,
      "num_input_tokens_seen": 20577165,
      "step": 987
    },
    {
      "epoch": 0.11879997595142187,
      "flos": 12023740108800.0,
      "grad_norm": 8.776969545177462,
      "learning_rate": 3.917945239912264e-06,
      "loss": 0.9471,
      "num_input_tokens_seen": 20594360,
      "step": 988
    },
    {
      "epoch": 0.11892021884206096,
      "flos": 12417303060480.0,
      "grad_norm": 5.357460325007849,
      "learning_rate": 3.917724256991367e-06,
      "loss": 0.9617,
      "num_input_tokens_seen": 20612825,
      "step": 989
    },
    {
      "epoch": 0.11904046173270005,
      "flos": 22065807790080.0,
      "grad_norm": 6.235505654052344,
      "learning_rate": 3.9175029831532245e-06,
      "loss": 1.0092,
      "num_input_tokens_seen": 20632060,
      "step": 990
    },
    {
      "epoch": 0.11916070462333915,
      "flos": 14305166622720.0,
      "grad_norm": 6.0522853097572265,
      "learning_rate": 3.917281418431404e-06,
      "loss": 1.0771,
      "num_input_tokens_seen": 20650825,
      "step": 991
    },
    {
      "epoch": 0.11928094751397823,
      "flos": 16744380088320.0,
      "grad_norm": 6.495585189428663,
      "learning_rate": 3.917059562859516e-06,
      "loss": 0.9593,
      "num_input_tokens_seen": 20669870,
      "step": 992
    },
    {
      "epoch": 0.11940119040461733,
      "flos": 17000153579520.0,
      "grad_norm": 7.559272014141908,
      "learning_rate": 3.916837416471218e-06,
      "loss": 1.0895,
      "num_input_tokens_seen": 20686210,
      "step": 993
    },
    {
      "epoch": 0.11952143329525641,
      "flos": 9768258846720.0,
      "grad_norm": 6.259252489693663,
      "learning_rate": 3.916614979300207e-06,
      "loss": 0.9119,
      "num_input_tokens_seen": 20700775,
      "step": 994
    },
    {
      "epoch": 0.11964167618589551,
      "flos": 19233951129600.0,
      "grad_norm": 4.4698777276368995,
      "learning_rate": 3.9163922513802274e-06,
      "loss": 0.9785,
      "num_input_tokens_seen": 20722830,
      "step": 995
    },
    {
      "epoch": 0.1197619190765346,
      "flos": 8850578903040.0,
      "grad_norm": 15.449565419625028,
      "learning_rate": 3.916169232745067e-06,
      "loss": 1.0293,
      "num_input_tokens_seen": 20740225,
      "step": 996
    },
    {
      "epoch": 0.11988216196717369,
      "flos": 11970960506880.0,
      "grad_norm": 19.528724836762898,
      "learning_rate": 3.915945923428559e-06,
      "loss": 1.1016,
      "num_input_tokens_seen": 20756470,
      "step": 997
    },
    {
      "epoch": 0.12000240485781279,
      "flos": 11473064693760.0,
      "grad_norm": 13.598603883777578,
      "learning_rate": 3.915722323464577e-06,
      "loss": 1.0284,
      "num_input_tokens_seen": 20774795,
      "step": 998
    },
    {
      "epoch": 0.12012264774845187,
      "flos": 35281649848320.0,
      "grad_norm": 11.707318643244196,
      "learning_rate": 3.91549843288704e-06,
      "loss": 0.9098,
      "num_input_tokens_seen": 20798195,
      "step": 999
    },
    {
      "epoch": 0.12024289063909097,
      "flos": 19208067194880.0,
      "grad_norm": 12.473641031433232,
      "learning_rate": 3.915274251729916e-06,
      "loss": 1.0074,
      "num_input_tokens_seen": 20819205,
      "step": 1000
    },
    {
      "epoch": 0.12036313352973005,
      "flos": 13859682508800.0,
      "grad_norm": 8.005983022724708,
      "learning_rate": 3.91504978002721e-06,
      "loss": 1.1016,
      "num_input_tokens_seen": 20837980,
      "step": 1001
    },
    {
      "epoch": 0.12048337642036915,
      "flos": 12228021657600.0,
      "grad_norm": 11.348717812867397,
      "learning_rate": 3.914825017812974e-06,
      "loss": 0.9722,
      "num_input_tokens_seen": 20854350,
      "step": 1002
    },
    {
      "epoch": 0.12060361931100824,
      "flos": 16245380567040.0,
      "grad_norm": 16.940136693697955,
      "learning_rate": 3.9145999651213065e-06,
      "loss": 0.9273,
      "num_input_tokens_seen": 20873310,
      "step": 1003
    },
    {
      "epoch": 0.12072386220164733,
      "flos": 11839670661120.0,
      "grad_norm": 23.773160222668256,
      "learning_rate": 3.9143746219863465e-06,
      "loss": 1.0743,
      "num_input_tokens_seen": 20890135,
      "step": 1004
    },
    {
      "epoch": 0.12084410509228642,
      "flos": 39454150963200.0,
      "grad_norm": 1.2417547083796707,
      "learning_rate": 3.914148988442278e-06,
      "loss": 0.9196,
      "num_input_tokens_seen": 20945645,
      "step": 1005
    },
    {
      "epoch": 0.1209643479829255,
      "flos": 19050494853120.0,
      "grad_norm": 9.338935760051628,
      "learning_rate": 3.91392306452333e-06,
      "loss": 1.1508,
      "num_input_tokens_seen": 20962440,
      "step": 1006
    },
    {
      "epoch": 0.1210845908735646,
      "flos": 7749105438720.0,
      "grad_norm": 11.187182544487749,
      "learning_rate": 3.913696850263774e-06,
      "loss": 0.8473,
      "num_input_tokens_seen": 20976525,
      "step": 1007
    },
    {
      "epoch": 0.1212048337642037,
      "flos": 14540329451520.0,
      "grad_norm": 8.13232138962068,
      "learning_rate": 3.913470345697929e-06,
      "loss": 0.9918,
      "num_input_tokens_seen": 20994875,
      "step": 1008
    },
    {
      "epoch": 0.12132507665484278,
      "flos": 15878192087040.0,
      "grad_norm": 25.592633904236,
      "learning_rate": 3.913243550860153e-06,
      "loss": 1.0545,
      "num_input_tokens_seen": 21012360,
      "step": 1009
    },
    {
      "epoch": 0.12144531954548188,
      "flos": 20886621204480.0,
      "grad_norm": 9.581677324155635,
      "learning_rate": 3.913016465784852e-06,
      "loss": 0.967,
      "num_input_tokens_seen": 21032755,
      "step": 1010
    },
    {
      "epoch": 0.12156556243612096,
      "flos": 14540850647040.0,
      "grad_norm": 9.858158477840512,
      "learning_rate": 3.912789090506474e-06,
      "loss": 0.9221,
      "num_input_tokens_seen": 21051735,
      "step": 1011
    },
    {
      "epoch": 0.12168580532676006,
      "flos": 11656643604480.0,
      "grad_norm": 6.7463097551550355,
      "learning_rate": 3.9125614250595114e-06,
      "loss": 0.9242,
      "num_input_tokens_seen": 21067665,
      "step": 1012
    },
    {
      "epoch": 0.12180604821739914,
      "flos": 10844062986240.0,
      "grad_norm": 13.119902283839616,
      "learning_rate": 3.912333469478502e-06,
      "loss": 1.0838,
      "num_input_tokens_seen": 21085350,
      "step": 1013
    },
    {
      "epoch": 0.12192629110803824,
      "flos": 13702048849920.0,
      "grad_norm": 13.113900277688712,
      "learning_rate": 3.912105223798025e-06,
      "loss": 0.9815,
      "num_input_tokens_seen": 21104490,
      "step": 1014
    },
    {
      "epoch": 0.12204653399867733,
      "flos": 34120735764480.0,
      "grad_norm": 1.0799864028145365,
      "learning_rate": 3.9118766880527065e-06,
      "loss": 0.916,
      "num_input_tokens_seen": 21158645,
      "step": 1015
    },
    {
      "epoch": 0.12216677688931642,
      "flos": 12914095165440.0,
      "grad_norm": 4.26970547305785,
      "learning_rate": 3.9116478622772145e-06,
      "loss": 0.9265,
      "num_input_tokens_seen": 21176940,
      "step": 1016
    },
    {
      "epoch": 0.12228701977995551,
      "flos": 19600465121280.0,
      "grad_norm": 5.405786042343491,
      "learning_rate": 3.911418746506261e-06,
      "loss": 1.0758,
      "num_input_tokens_seen": 21196790,
      "step": 1017
    },
    {
      "epoch": 0.1224072626705946,
      "flos": 15484322549760.0,
      "grad_norm": 8.913955086036664,
      "learning_rate": 3.911189340774604e-06,
      "loss": 0.982,
      "num_input_tokens_seen": 21216640,
      "step": 1018
    },
    {
      "epoch": 0.1225275055612337,
      "flos": 14697840476160.0,
      "grad_norm": 3.7600853405053587,
      "learning_rate": 3.910959645117043e-06,
      "loss": 0.9917,
      "num_input_tokens_seen": 21235695,
      "step": 1019
    },
    {
      "epoch": 0.12264774845187278,
      "flos": 41324960256000.0,
      "grad_norm": 0.8347766338188339,
      "learning_rate": 3.910729659568423e-06,
      "loss": 0.7933,
      "num_input_tokens_seen": 21292600,
      "step": 1020
    },
    {
      "epoch": 0.12276799134251187,
      "flos": 18789049528320.0,
      "grad_norm": 9.627492606676027,
      "learning_rate": 3.9104993841636344e-06,
      "loss": 1.025,
      "num_input_tokens_seen": 21312890,
      "step": 1021
    },
    {
      "epoch": 0.12288823423315097,
      "flos": 14956281262080.0,
      "grad_norm": 5.316001187154323,
      "learning_rate": 3.910268818937608e-06,
      "loss": 1.0092,
      "num_input_tokens_seen": 21330765,
      "step": 1022
    },
    {
      "epoch": 0.12300847712379005,
      "flos": 8667214602240.0,
      "grad_norm": 6.564644889902087,
      "learning_rate": 3.9100379639253196e-06,
      "loss": 1.0792,
      "num_input_tokens_seen": 21347205,
      "step": 1023
    },
    {
      "epoch": 0.12312872001442915,
      "flos": 11865002741760.0,
      "grad_norm": 4.412352103977825,
      "learning_rate": 3.909806819161791e-06,
      "loss": 1.0586,
      "num_input_tokens_seen": 21362400,
      "step": 1024
    },
    {
      "epoch": 0.12324896290506823,
      "flos": 13045691596800.0,
      "grad_norm": 6.588869720453975,
      "learning_rate": 3.909575384682086e-06,
      "loss": 1.0534,
      "num_input_tokens_seen": 21381000,
      "step": 1025
    },
    {
      "epoch": 0.12336920579570733,
      "flos": 13412512174080.0,
      "grad_norm": 3.3459568049988286,
      "learning_rate": 3.9093436605213144e-06,
      "loss": 0.8837,
      "num_input_tokens_seen": 21401220,
      "step": 1026
    },
    {
      "epoch": 0.12348944868634643,
      "flos": 16979420282880.0,
      "grad_norm": 3.9238359024935363,
      "learning_rate": 3.909111646714627e-06,
      "loss": 0.9841,
      "num_input_tokens_seen": 21421785,
      "step": 1027
    },
    {
      "epoch": 0.12360969157698551,
      "flos": 13492218101760.0,
      "grad_norm": 9.585558646751503,
      "learning_rate": 3.9088793432972206e-06,
      "loss": 0.9114,
      "num_input_tokens_seen": 21440325,
      "step": 1028
    },
    {
      "epoch": 0.1237299344676246,
      "flos": 9322774732800.0,
      "grad_norm": 8.235335292232387,
      "learning_rate": 3.908646750304336e-06,
      "loss": 1.0218,
      "num_input_tokens_seen": 21457730,
      "step": 1029
    },
    {
      "epoch": 0.12385017735826369,
      "flos": 14541218549760.0,
      "grad_norm": 9.082583148553644,
      "learning_rate": 3.908413867771257e-06,
      "loss": 1.0626,
      "num_input_tokens_seen": 21476360,
      "step": 1030
    },
    {
      "epoch": 0.12397042024890279,
      "flos": 12286105190400.0,
      "grad_norm": 4.601667781849502,
      "learning_rate": 3.908180695733311e-06,
      "loss": 0.9998,
      "num_input_tokens_seen": 21495570,
      "step": 1031
    },
    {
      "epoch": 0.12409066313954187,
      "flos": 14488469606400.0,
      "grad_norm": 12.43457978437802,
      "learning_rate": 3.907947234225871e-06,
      "loss": 1.0279,
      "num_input_tokens_seen": 21514300,
      "step": 1032
    },
    {
      "epoch": 0.12421090603018096,
      "flos": 14721026457600.0,
      "grad_norm": 4.4465514804575665,
      "learning_rate": 3.907713483284352e-06,
      "loss": 1.0693,
      "num_input_tokens_seen": 21533495,
      "step": 1033
    },
    {
      "epoch": 0.12433114892082006,
      "flos": 17425364275200.0,
      "grad_norm": 7.396856991125891,
      "learning_rate": 3.907479442944216e-06,
      "loss": 1.1756,
      "num_input_tokens_seen": 21551620,
      "step": 1034
    },
    {
      "epoch": 0.12445139181145914,
      "flos": 13963770101760.0,
      "grad_norm": 9.280621131630532,
      "learning_rate": 3.907245113240963e-06,
      "loss": 1.1212,
      "num_input_tokens_seen": 21569460,
      "step": 1035
    },
    {
      "epoch": 0.12457163470209824,
      "flos": 33181410877440.0,
      "grad_norm": 5.5735718339024665,
      "learning_rate": 3.907010494210144e-06,
      "loss": 0.9439,
      "num_input_tokens_seen": 21591840,
      "step": 1036
    },
    {
      "epoch": 0.12469187759273732,
      "flos": 14462616330240.0,
      "grad_norm": 7.294518950134216,
      "learning_rate": 3.9067755858873495e-06,
      "loss": 1.1254,
      "num_input_tokens_seen": 21608360,
      "step": 1037
    },
    {
      "epoch": 0.12481212048337642,
      "flos": 49573103984640.0,
      "grad_norm": 1.1269678812366617,
      "learning_rate": 3.906540388308214e-06,
      "loss": 0.8613,
      "num_input_tokens_seen": 21667665,
      "step": 1038
    },
    {
      "epoch": 0.12493236337401552,
      "flos": 12915597434880.0,
      "grad_norm": 7.624244527443871,
      "learning_rate": 3.906304901508417e-06,
      "loss": 1.0159,
      "num_input_tokens_seen": 21686285,
      "step": 1039
    },
    {
      "epoch": 0.12505260626465461,
      "flos": 21410799513600.0,
      "grad_norm": 6.222028778490286,
      "learning_rate": 3.9060691255236835e-06,
      "loss": 0.9616,
      "num_input_tokens_seen": 21706570,
      "step": 1040
    },
    {
      "epoch": 0.1251728491552937,
      "flos": 17379850752000.0,
      "grad_norm": 4.306722336597276,
      "learning_rate": 3.905833060389778e-06,
      "loss": 1.0222,
      "num_input_tokens_seen": 21730410,
      "step": 1041
    },
    {
      "epoch": 0.12529309204593278,
      "flos": 19308199833600.0,
      "grad_norm": 7.9491179738698055,
      "learning_rate": 3.905596706142513e-06,
      "loss": 0.9658,
      "num_input_tokens_seen": 21751540,
      "step": 1042
    },
    {
      "epoch": 0.12541333493657186,
      "flos": 21935284408320.0,
      "grad_norm": 4.4944056598973985,
      "learning_rate": 3.9053600628177435e-06,
      "loss": 1.0474,
      "num_input_tokens_seen": 21770870,
      "step": 1043
    },
    {
      "epoch": 0.12553357782721097,
      "flos": 16822185185280.0,
      "grad_norm": 3.8387548973178958,
      "learning_rate": 3.905123130451367e-06,
      "loss": 1.0462,
      "num_input_tokens_seen": 21791690,
      "step": 1044
    },
    {
      "epoch": 0.12565382071785006,
      "flos": 17687637381120.0,
      "grad_norm": 3.361358199131877,
      "learning_rate": 3.904885909079326e-06,
      "loss": 0.9906,
      "num_input_tokens_seen": 21810195,
      "step": 1045
    },
    {
      "epoch": 0.12577406360848914,
      "flos": 15170066964480.0,
      "grad_norm": 6.839079356112852,
      "learning_rate": 3.904648398737607e-06,
      "loss": 0.9762,
      "num_input_tokens_seen": 21828480,
      "step": 1046
    },
    {
      "epoch": 0.12589430649912825,
      "flos": 25790288240640.0,
      "grad_norm": 4.813358775218673,
      "learning_rate": 3.9044105994622406e-06,
      "loss": 0.9738,
      "num_input_tokens_seen": 21849345,
      "step": 1047
    },
    {
      "epoch": 0.12601454938976733,
      "flos": 18028175462400.0,
      "grad_norm": 6.19202604350446,
      "learning_rate": 3.9041725112893005e-06,
      "loss": 1.015,
      "num_input_tokens_seen": 21870290,
      "step": 1048
    },
    {
      "epoch": 0.12613479228040642,
      "flos": 11001298083840.0,
      "grad_norm": 17.768253880088096,
      "learning_rate": 3.903934134254904e-06,
      "loss": 0.9514,
      "num_input_tokens_seen": 21887800,
      "step": 1049
    },
    {
      "epoch": 0.1262550351710455,
      "flos": 15249037086720.0,
      "grad_norm": 5.351452724695271,
      "learning_rate": 3.903695468395213e-06,
      "loss": 1.0457,
      "num_input_tokens_seen": 21905390,
      "step": 1050
    },
    {
      "epoch": 0.1263752780616846,
      "flos": 22512058368000.0,
      "grad_norm": 6.213268221785505,
      "learning_rate": 3.903456513746434e-06,
      "loss": 0.7718,
      "num_input_tokens_seen": 21926085,
      "step": 1051
    },
    {
      "epoch": 0.1264955209523237,
      "flos": 20493763399680.0,
      "grad_norm": 5.021998268993084,
      "learning_rate": 3.903217270344815e-06,
      "loss": 1.0756,
      "num_input_tokens_seen": 21946055,
      "step": 1052
    },
    {
      "epoch": 0.12661576384296278,
      "flos": 20833443041280.0,
      "grad_norm": 4.171029565992174,
      "learning_rate": 3.902977738226648e-06,
      "loss": 1.0149,
      "num_input_tokens_seen": 21966510,
      "step": 1053
    },
    {
      "epoch": 0.12673600673360189,
      "flos": 14803277045760.0,
      "grad_norm": 5.188567270081608,
      "learning_rate": 3.902737917428273e-06,
      "loss": 1.1166,
      "num_input_tokens_seen": 21984395,
      "step": 1054
    },
    {
      "epoch": 0.12685624962424097,
      "flos": 17974874664960.0,
      "grad_norm": 104.97331320090359,
      "learning_rate": 3.902497807986068e-06,
      "loss": 1.0327,
      "num_input_tokens_seen": 22004135,
      "step": 1055
    },
    {
      "epoch": 0.12697649251488005,
      "flos": 19601722122240.0,
      "grad_norm": 6.437244595396157,
      "learning_rate": 3.902257409936458e-06,
      "loss": 1.0302,
      "num_input_tokens_seen": 22024620,
      "step": 1056
    },
    {
      "epoch": 0.12709673540551916,
      "flos": 15091863306240.0,
      "grad_norm": 7.332622002289958,
      "learning_rate": 3.902016723315912e-06,
      "loss": 1.0363,
      "num_input_tokens_seen": 22042280,
      "step": 1057
    },
    {
      "epoch": 0.12721697829615825,
      "flos": 18028206120960.0,
      "grad_norm": 4.61105905577501,
      "learning_rate": 3.901775748160941e-06,
      "loss": 0.8893,
      "num_input_tokens_seen": 22061180,
      "step": 1058
    },
    {
      "epoch": 0.12733722118679733,
      "flos": 44342265999360.0,
      "grad_norm": 1.016180678721228,
      "learning_rate": 3.901534484508101e-06,
      "loss": 0.8386,
      "num_input_tokens_seen": 22123575,
      "step": 1059
    },
    {
      "epoch": 0.1274574640774364,
      "flos": 19205767802880.0,
      "grad_norm": 3.938936122999064,
      "learning_rate": 3.901292932393991e-06,
      "loss": 0.948,
      "num_input_tokens_seen": 22142175,
      "step": 1060
    },
    {
      "epoch": 0.12757770696807552,
      "flos": 15799283281920.0,
      "grad_norm": 5.468510039378299,
      "learning_rate": 3.9010510918552555e-06,
      "loss": 1.0579,
      "num_input_tokens_seen": 22160970,
      "step": 1061
    },
    {
      "epoch": 0.1276979498587146,
      "flos": 20335240642560.0,
      "grad_norm": 10.791352334961784,
      "learning_rate": 3.900808962928581e-06,
      "loss": 0.9579,
      "num_input_tokens_seen": 22178305,
      "step": 1062
    },
    {
      "epoch": 0.1278181927493537,
      "flos": 12338148986880.0,
      "grad_norm": 5.1467082197524645,
      "learning_rate": 3.900566545650698e-06,
      "loss": 1.0898,
      "num_input_tokens_seen": 22195695,
      "step": 1063
    },
    {
      "epoch": 0.1279384356399928,
      "flos": 15009980620800.0,
      "grad_norm": 4.244608657105804,
      "learning_rate": 3.900323840058381e-06,
      "loss": 1.0136,
      "num_input_tokens_seen": 22213125,
      "step": 1064
    },
    {
      "epoch": 0.12805867853063188,
      "flos": 18918469201920.0,
      "grad_norm": 6.497323879131471,
      "learning_rate": 3.900080846188449e-06,
      "loss": 1.0125,
      "num_input_tokens_seen": 22231435,
      "step": 1065
    },
    {
      "epoch": 0.12817892142127096,
      "flos": 11631004938240.0,
      "grad_norm": 5.429748446993823,
      "learning_rate": 3.8998375640777625e-06,
      "loss": 1.0058,
      "num_input_tokens_seen": 22249025,
      "step": 1066
    },
    {
      "epoch": 0.12829916431191005,
      "flos": 43487215165440.0,
      "grad_norm": 0.9277991663744652,
      "learning_rate": 3.899593993763229e-06,
      "loss": 0.7529,
      "num_input_tokens_seen": 22309705,
      "step": 1067
    },
    {
      "epoch": 0.12841940720254916,
      "flos": 21225473064960.0,
      "grad_norm": 5.273510018309248,
      "learning_rate": 3.899350135281796e-06,
      "loss": 1.0002,
      "num_input_tokens_seen": 22330425,
      "step": 1068
    },
    {
      "epoch": 0.12853965009318824,
      "flos": 18468754206720.0,
      "grad_norm": 4.320938920313058,
      "learning_rate": 3.8991059886704585e-06,
      "loss": 0.9955,
      "num_input_tokens_seen": 22349650,
      "step": 1069
    },
    {
      "epoch": 0.12865989298382732,
      "flos": 21987512156160.0,
      "grad_norm": 6.478160555162769,
      "learning_rate": 3.898861553966252e-06,
      "loss": 1.0196,
      "num_input_tokens_seen": 22369020,
      "step": 1070
    },
    {
      "epoch": 0.12878013587446643,
      "flos": 18422198292480.0,
      "grad_norm": 5.417833212413362,
      "learning_rate": 3.898616831206257e-06,
      "loss": 1.075,
      "num_input_tokens_seen": 22389165,
      "step": 1071
    },
    {
      "epoch": 0.12890037876510552,
      "flos": 16587850137600.0,
      "grad_norm": 9.521326320803555,
      "learning_rate": 3.8983718204276e-06,
      "loss": 0.9704,
      "num_input_tokens_seen": 22411105,
      "step": 1072
    },
    {
      "epoch": 0.1290206216557446,
      "flos": 16770294681600.0,
      "grad_norm": 4.996937195162399,
      "learning_rate": 3.898126521667446e-06,
      "loss": 1.0417,
      "num_input_tokens_seen": 22430980,
      "step": 1073
    },
    {
      "epoch": 0.12914086454638368,
      "flos": 17189496299520.0,
      "grad_norm": 2.9931807898696543,
      "learning_rate": 3.897880934963007e-06,
      "loss": 1.0403,
      "num_input_tokens_seen": 22450250,
      "step": 1074
    },
    {
      "epoch": 0.1292611074370228,
      "flos": 14384228720640.0,
      "grad_norm": 4.203484842432126,
      "learning_rate": 3.89763506035154e-06,
      "loss": 0.9845,
      "num_input_tokens_seen": 22467820,
      "step": 1075
    },
    {
      "epoch": 0.12938135032766188,
      "flos": 19493863526400.0,
      "grad_norm": 4.16700809351458,
      "learning_rate": 3.897388897870343e-06,
      "loss": 1.023,
      "num_input_tokens_seen": 22488180,
      "step": 1076
    },
    {
      "epoch": 0.12950159321830096,
      "flos": 20859235000320.0,
      "grad_norm": 4.49734970036716,
      "learning_rate": 3.89714244755676e-06,
      "loss": 0.9359,
      "num_input_tokens_seen": 22509260,
      "step": 1077
    },
    {
      "epoch": 0.12962183610894007,
      "flos": 17451217551360.0,
      "grad_norm": 5.49869452502066,
      "learning_rate": 3.896895709448175e-06,
      "loss": 1.0613,
      "num_input_tokens_seen": 22528730,
      "step": 1078
    },
    {
      "epoch": 0.12974207899957915,
      "flos": 7878923673600.0,
      "grad_norm": 5.5140828254132686,
      "learning_rate": 3.896648683582019e-06,
      "loss": 0.9586,
      "num_input_tokens_seen": 22543785,
      "step": 1079
    },
    {
      "epoch": 0.12986232189021824,
      "flos": 17582813982720.0,
      "grad_norm": 8.316244328948587,
      "learning_rate": 3.896401369995766e-06,
      "loss": 1.0087,
      "num_input_tokens_seen": 22563310,
      "step": 1080
    },
    {
      "epoch": 0.12998256478085732,
      "flos": 17006070681600.0,
      "grad_norm": 3.688069315782329,
      "learning_rate": 3.896153768726932e-06,
      "loss": 0.9862,
      "num_input_tokens_seen": 22583340,
      "step": 1081
    },
    {
      "epoch": 0.13010280767149643,
      "flos": 13203325255680.0,
      "grad_norm": 9.262764037123146,
      "learning_rate": 3.8959058798130806e-06,
      "loss": 1.0706,
      "num_input_tokens_seen": 22601035,
      "step": 1082
    },
    {
      "epoch": 0.1302230505621355,
      "flos": 16192938209280.0,
      "grad_norm": 4.0511339083771984,
      "learning_rate": 3.895657703291814e-06,
      "loss": 0.9406,
      "num_input_tokens_seen": 22620860,
      "step": 1083
    },
    {
      "epoch": 0.1303432934527746,
      "flos": 15143263272960.0,
      "grad_norm": 7.098584111836829,
      "learning_rate": 3.895409239200781e-06,
      "loss": 0.9963,
      "num_input_tokens_seen": 22636465,
      "step": 1084
    },
    {
      "epoch": 0.1304635363434137,
      "flos": 14855014256640.0,
      "grad_norm": 5.297049460449675,
      "learning_rate": 3.895160487577673e-06,
      "loss": 1.1188,
      "num_input_tokens_seen": 22653755,
      "step": 1085
    },
    {
      "epoch": 0.1305837792340528,
      "flos": 45276754944000.0,
      "grad_norm": 0.8942042923471271,
      "learning_rate": 3.894911448460226e-06,
      "loss": 0.8313,
      "num_input_tokens_seen": 22712790,
      "step": 1086
    },
    {
      "epoch": 0.13070402212469187,
      "flos": 18811928924160.0,
      "grad_norm": 3.4224873870104484,
      "learning_rate": 3.8946621218862195e-06,
      "loss": 0.9269,
      "num_input_tokens_seen": 22733510,
      "step": 1087
    },
    {
      "epoch": 0.13082426501533098,
      "flos": 19706913423360.0,
      "grad_norm": 4.144183087623557,
      "learning_rate": 3.894412507893475e-06,
      "loss": 1.0823,
      "num_input_tokens_seen": 22753510,
      "step": 1088
    },
    {
      "epoch": 0.13094450790597006,
      "flos": 17660986982400.0,
      "grad_norm": 6.356431434464993,
      "learning_rate": 3.894162606519859e-06,
      "loss": 0.9198,
      "num_input_tokens_seen": 22772180,
      "step": 1089
    },
    {
      "epoch": 0.13106475079660915,
      "flos": 13518469939200.0,
      "grad_norm": 5.0022354850629,
      "learning_rate": 3.893912417803282e-06,
      "loss": 0.9729,
      "num_input_tokens_seen": 22791615,
      "step": 1090
    },
    {
      "epoch": 0.13118499368724823,
      "flos": 20597820334080.0,
      "grad_norm": 3.3564568198358895,
      "learning_rate": 3.8936619417816975e-06,
      "loss": 0.9622,
      "num_input_tokens_seen": 22811665,
      "step": 1091
    },
    {
      "epoch": 0.13130523657788734,
      "flos": 10083802091520.0,
      "grad_norm": 4.06040595444502,
      "learning_rate": 3.8934111784931015e-06,
      "loss": 0.9234,
      "num_input_tokens_seen": 22828835,
      "step": 1092
    },
    {
      "epoch": 0.13142547946852642,
      "flos": 50255621099520.0,
      "grad_norm": 1.026081891247666,
      "learning_rate": 3.893160127975535e-06,
      "loss": 0.8431,
      "num_input_tokens_seen": 22889245,
      "step": 1093
    },
    {
      "epoch": 0.1315457223591655,
      "flos": 32738348789760.0,
      "grad_norm": 4.9848997981829894,
      "learning_rate": 3.8929087902670826e-06,
      "loss": 1.012,
      "num_input_tokens_seen": 22910595,
      "step": 1094
    },
    {
      "epoch": 0.13166596524980462,
      "flos": 45014849740800.0,
      "grad_norm": 0.9178733289106769,
      "learning_rate": 3.8926571654058715e-06,
      "loss": 0.821,
      "num_input_tokens_seen": 22966960,
      "step": 1095
    },
    {
      "epoch": 0.1317862081404437,
      "flos": 16769681510400.0,
      "grad_norm": 4.974646757994108,
      "learning_rate": 3.892405253430074e-06,
      "loss": 0.9637,
      "num_input_tokens_seen": 22984200,
      "step": 1096
    },
    {
      "epoch": 0.13190645103108278,
      "flos": 14378771496960.0,
      "grad_norm": 3.8364098407637046,
      "learning_rate": 3.892153054377904e-06,
      "loss": 1.0239,
      "num_input_tokens_seen": 23001325,
      "step": 1097
    },
    {
      "epoch": 0.13202669392172187,
      "flos": 38239085752320.0,
      "grad_norm": 1.098335571769166,
      "learning_rate": 3.891900568287619e-06,
      "loss": 0.8353,
      "num_input_tokens_seen": 23053430,
      "step": 1098
    },
    {
      "epoch": 0.13214693681236098,
      "flos": 11210485002240.0,
      "grad_norm": 4.232715915100414,
      "learning_rate": 3.891647795197523e-06,
      "loss": 0.9265,
      "num_input_tokens_seen": 23069190,
      "step": 1099
    },
    {
      "epoch": 0.13226717970300006,
      "flos": 13727687516160.0,
      "grad_norm": 8.437026625175825,
      "learning_rate": 3.8913947351459605e-06,
      "loss": 0.8867,
      "num_input_tokens_seen": 23086450,
      "step": 1100
    },
    {
      "epoch": 0.13238742259363914,
      "flos": 14692996423680.0,
      "grad_norm": 6.40038406888587,
      "learning_rate": 3.89114138817132e-06,
      "loss": 0.8772,
      "num_input_tokens_seen": 23102835,
      "step": 1101
    },
    {
      "epoch": 0.13250766548427825,
      "flos": 14934107013120.0,
      "grad_norm": 3.2498694173548572,
      "learning_rate": 3.890887754312035e-06,
      "loss": 1.0507,
      "num_input_tokens_seen": 23120800,
      "step": 1102
    },
    {
      "epoch": 0.13262790837491734,
      "flos": 16087900200960.0,
      "grad_norm": 16.95181626572891,
      "learning_rate": 3.890633833606581e-06,
      "loss": 1.0719,
      "num_input_tokens_seen": 23140210,
      "step": 1103
    },
    {
      "epoch": 0.13274815126555642,
      "flos": 13964475248640.0,
      "grad_norm": 3.447659962174738,
      "learning_rate": 3.890379626093477e-06,
      "loss": 0.8943,
      "num_input_tokens_seen": 23159680,
      "step": 1104
    },
    {
      "epoch": 0.1328683941561955,
      "flos": 15138204610560.0,
      "grad_norm": 3.8215273932194265,
      "learning_rate": 3.890125131811287e-06,
      "loss": 1.1108,
      "num_input_tokens_seen": 23177450,
      "step": 1105
    },
    {
      "epoch": 0.1329886370468346,
      "flos": 9663834009600.0,
      "grad_norm": 4.032969736361678,
      "learning_rate": 3.889870350798618e-06,
      "loss": 0.9615,
      "num_input_tokens_seen": 23194515,
      "step": 1106
    },
    {
      "epoch": 0.1331088799374737,
      "flos": 14934076354560.0,
      "grad_norm": 19.540189360498548,
      "learning_rate": 3.889615283094119e-06,
      "loss": 0.9905,
      "num_input_tokens_seen": 23213425,
      "step": 1107
    },
    {
      "epoch": 0.13322912282811278,
      "flos": 12942033223680.0,
      "grad_norm": 6.11244097410071,
      "learning_rate": 3.889359928736485e-06,
      "loss": 1.0431,
      "num_input_tokens_seen": 23231090,
      "step": 1108
    },
    {
      "epoch": 0.1333493657187519,
      "flos": 17397886095360.0,
      "grad_norm": 5.894969036613081,
      "learning_rate": 3.889104287764451e-06,
      "loss": 1.1014,
      "num_input_tokens_seen": 23251185,
      "step": 1109
    },
    {
      "epoch": 0.13346960860939097,
      "flos": 15743069921280.0,
      "grad_norm": 5.649549699241455,
      "learning_rate": 3.888848360216798e-06,
      "loss": 1.1059,
      "num_input_tokens_seen": 23268550,
      "step": 1110
    },
    {
      "epoch": 0.13358985150003005,
      "flos": 48646249021440.0,
      "grad_norm": 0.8740204368838158,
      "learning_rate": 3.888592146132351e-06,
      "loss": 0.7808,
      "num_input_tokens_seen": 23329540,
      "step": 1111
    },
    {
      "epoch": 0.13371009439066917,
      "flos": 19103856967680.0,
      "grad_norm": 8.377089054839685,
      "learning_rate": 3.888335645549978e-06,
      "loss": 0.9712,
      "num_input_tokens_seen": 23349680,
      "step": 1112
    },
    {
      "epoch": 0.13383033728130825,
      "flos": 18736208609280.0,
      "grad_norm": 7.058652386741091,
      "learning_rate": 3.888078858508588e-06,
      "loss": 1.0259,
      "num_input_tokens_seen": 23369260,
      "step": 1113
    },
    {
      "epoch": 0.13395058017194733,
      "flos": 16034630062080.0,
      "grad_norm": 4.05702544854137,
      "learning_rate": 3.8878217850471365e-06,
      "loss": 1.0499,
      "num_input_tokens_seen": 23388895,
      "step": 1114
    },
    {
      "epoch": 0.13407082306258641,
      "flos": 18368161689600.0,
      "grad_norm": 3.8674677583500863,
      "learning_rate": 3.887564425204621e-06,
      "loss": 0.9364,
      "num_input_tokens_seen": 23410300,
      "step": 1115
    },
    {
      "epoch": 0.13419106595322552,
      "flos": 48934720757760.0,
      "grad_norm": 0.8551142406547898,
      "learning_rate": 3.887306779020083e-06,
      "loss": 0.7653,
      "num_input_tokens_seen": 23464675,
      "step": 1116
    },
    {
      "epoch": 0.1343113088438646,
      "flos": 14514905395200.0,
      "grad_norm": 7.9071396640008595,
      "learning_rate": 3.887048846532608e-06,
      "loss": 0.9032,
      "num_input_tokens_seen": 23481370,
      "step": 1117
    },
    {
      "epoch": 0.1344315517345037,
      "flos": 48253000765440.0,
      "grad_norm": 0.7715458712634156,
      "learning_rate": 3.8867906277813224e-06,
      "loss": 0.7921,
      "num_input_tokens_seen": 23539245,
      "step": 1118
    },
    {
      "epoch": 0.1345517946251428,
      "flos": 29093512949760.0,
      "grad_norm": 5.458356124403258,
      "learning_rate": 3.886532122805399e-06,
      "loss": 0.9396,
      "num_input_tokens_seen": 23561445,
      "step": 1119
    },
    {
      "epoch": 0.13467203751578188,
      "flos": 16216062873600.0,
      "grad_norm": 4.216264006396049,
      "learning_rate": 3.886273331644053e-06,
      "loss": 1.0985,
      "num_input_tokens_seen": 23580035,
      "step": 1120
    },
    {
      "epoch": 0.13479228040642097,
      "flos": 12101821132800.0,
      "grad_norm": 3.981681362000965,
      "learning_rate": 3.886014254336542e-06,
      "loss": 1.0276,
      "num_input_tokens_seen": 23596230,
      "step": 1121
    },
    {
      "epoch": 0.13491252329706005,
      "flos": 16872910663680.0,
      "grad_norm": 5.152777039744052,
      "learning_rate": 3.885754890922168e-06,
      "loss": 1.1201,
      "num_input_tokens_seen": 23616280,
      "step": 1122
    },
    {
      "epoch": 0.13503276618769916,
      "flos": 24344444375040.0,
      "grad_norm": 9.412361263554649,
      "learning_rate": 3.885495241440277e-06,
      "loss": 0.9913,
      "num_input_tokens_seen": 23640095,
      "step": 1123
    },
    {
      "epoch": 0.13515300907833824,
      "flos": 12548010393600.0,
      "grad_norm": 4.370452479281433,
      "learning_rate": 3.885235305930257e-06,
      "loss": 0.9494,
      "num_input_tokens_seen": 23658015,
      "step": 1124
    },
    {
      "epoch": 0.13527325196897733,
      "flos": 14378863472640.0,
      "grad_norm": 6.9636575883695375,
      "learning_rate": 3.884975084431539e-06,
      "loss": 1.0658,
      "num_input_tokens_seen": 23672685,
      "step": 1125
    },
    {
      "epoch": 0.13539349485961644,
      "flos": 12888763084800.0,
      "grad_norm": 4.675243783398263,
      "learning_rate": 3.8847145769836e-06,
      "loss": 1.1236,
      "num_input_tokens_seen": 23688825,
      "step": 1126
    },
    {
      "epoch": 0.13551373775025552,
      "flos": 13701558312960.0,
      "grad_norm": 8.300849723982282,
      "learning_rate": 3.884453783625959e-06,
      "loss": 0.8715,
      "num_input_tokens_seen": 23706155,
      "step": 1127
    },
    {
      "epoch": 0.1356339806408946,
      "flos": 14803675607040.0,
      "grad_norm": 7.19495516044,
      "learning_rate": 3.884192704398176e-06,
      "loss": 1.0546,
      "num_input_tokens_seen": 23723075,
      "step": 1128
    },
    {
      "epoch": 0.13575422353153369,
      "flos": 36094598369280.0,
      "grad_norm": 3.198096221621392,
      "learning_rate": 3.883931339339858e-06,
      "loss": 0.9415,
      "num_input_tokens_seen": 23747180,
      "step": 1129
    },
    {
      "epoch": 0.1358744664221728,
      "flos": 12863216394240.0,
      "grad_norm": 6.7926354526665325,
      "learning_rate": 3.883669688490654e-06,
      "loss": 0.987,
      "num_input_tokens_seen": 23764670,
      "step": 1130
    },
    {
      "epoch": 0.13599470931281188,
      "flos": 13073231093760.0,
      "grad_norm": 5.252733154932185,
      "learning_rate": 3.883407751890256e-06,
      "loss": 1.0549,
      "num_input_tokens_seen": 23782995,
      "step": 1131
    },
    {
      "epoch": 0.13611495220345096,
      "flos": 18992503296000.0,
      "grad_norm": 7.109842071281503,
      "learning_rate": 3.8831455295783994e-06,
      "loss": 1.0501,
      "num_input_tokens_seen": 23801965,
      "step": 1132
    },
    {
      "epoch": 0.13623519509409007,
      "flos": 15403574231040.0,
      "grad_norm": 5.606190066830564,
      "learning_rate": 3.882883021594864e-06,
      "loss": 0.9424,
      "num_input_tokens_seen": 23819825,
      "step": 1133
    },
    {
      "epoch": 0.13635543798472916,
      "flos": 10476874506240.0,
      "grad_norm": 5.530048132026456,
      "learning_rate": 3.8826202279794705e-06,
      "loss": 1.0637,
      "num_input_tokens_seen": 23836605,
      "step": 1134
    },
    {
      "epoch": 0.13647568087536824,
      "flos": 16268934451200.0,
      "grad_norm": 6.868436975262589,
      "learning_rate": 3.882357148772085e-06,
      "loss": 0.8993,
      "num_input_tokens_seen": 23853750,
      "step": 1135
    },
    {
      "epoch": 0.13659592376600732,
      "flos": 14147502305280.0,
      "grad_norm": 24.927769189356123,
      "learning_rate": 3.882093784012617e-06,
      "loss": 1.0552,
      "num_input_tokens_seen": 23872110,
      "step": 1136
    },
    {
      "epoch": 0.13671616665664643,
      "flos": 15218094489600.0,
      "grad_norm": 6.345482800980479,
      "learning_rate": 3.881830133741019e-06,
      "loss": 1.0323,
      "num_input_tokens_seen": 23890695,
      "step": 1137
    },
    {
      "epoch": 0.13683640954728551,
      "flos": 16187940864000.0,
      "grad_norm": 4.504039223156307,
      "learning_rate": 3.881566197997285e-06,
      "loss": 0.9639,
      "num_input_tokens_seen": 23906850,
      "step": 1138
    },
    {
      "epoch": 0.1369566524379246,
      "flos": 15432493363200.0,
      "grad_norm": 3.4495662803580114,
      "learning_rate": 3.881301976821456e-06,
      "loss": 0.9471,
      "num_input_tokens_seen": 23926600,
      "step": 1139
    },
    {
      "epoch": 0.1370768953285637,
      "flos": 13203539865600.0,
      "grad_norm": 5.730030711359669,
      "learning_rate": 3.881037470253612e-06,
      "loss": 1.0962,
      "num_input_tokens_seen": 23945835,
      "step": 1140
    },
    {
      "epoch": 0.1371971382192028,
      "flos": 10555262115840.0,
      "grad_norm": 6.799406294990429,
      "learning_rate": 3.88077267833388e-06,
      "loss": 0.997,
      "num_input_tokens_seen": 23962070,
      "step": 1141
    },
    {
      "epoch": 0.13731738110984187,
      "flos": 13490255953920.0,
      "grad_norm": 5.327784731562831,
      "learning_rate": 3.880507601102427e-06,
      "loss": 1.0337,
      "num_input_tokens_seen": 23979725,
      "step": 1142
    },
    {
      "epoch": 0.13743762400048098,
      "flos": 12889590865920.0,
      "grad_norm": 23.726141902347134,
      "learning_rate": 3.880242238599467e-06,
      "loss": 1.0191,
      "num_input_tokens_seen": 23995970,
      "step": 1143
    },
    {
      "epoch": 0.13755786689112007,
      "flos": 14933708451840.0,
      "grad_norm": 3.913186102723501,
      "learning_rate": 3.879976590865254e-06,
      "loss": 1.0374,
      "num_input_tokens_seen": 24015145,
      "step": 1144
    },
    {
      "epoch": 0.13767810978175915,
      "flos": 15169944330240.0,
      "grad_norm": 5.39268281884065,
      "learning_rate": 3.879710657940087e-06,
      "loss": 1.0752,
      "num_input_tokens_seen": 24033815,
      "step": 1145
    },
    {
      "epoch": 0.13779835267239823,
      "flos": 21804485099520.0,
      "grad_norm": 5.4457021653193864,
      "learning_rate": 3.879444439864308e-06,
      "loss": 0.8981,
      "num_input_tokens_seen": 24053110,
      "step": 1146
    },
    {
      "epoch": 0.13791859556303734,
      "flos": 16110289059840.0,
      "grad_norm": 7.04208854299371,
      "learning_rate": 3.879177936678301e-06,
      "loss": 1.0574,
      "num_input_tokens_seen": 24071835,
      "step": 1147
    },
    {
      "epoch": 0.13803883845367643,
      "flos": 25133256499200.0,
      "grad_norm": 12.175918475526036,
      "learning_rate": 3.878911148422496e-06,
      "loss": 0.9805,
      "num_input_tokens_seen": 24093030,
      "step": 1148
    },
    {
      "epoch": 0.1381590813443155,
      "flos": 22826712514560.0,
      "grad_norm": 7.373552083270791,
      "learning_rate": 3.878644075137364e-06,
      "loss": 0.9045,
      "num_input_tokens_seen": 24113400,
      "step": 1149
    },
    {
      "epoch": 0.13827932423495462,
      "flos": 12626735247360.0,
      "grad_norm": 4.315204159535973,
      "learning_rate": 3.878376716863418e-06,
      "loss": 0.9969,
      "num_input_tokens_seen": 24129420,
      "step": 1150
    },
    {
      "epoch": 0.1383995671255937,
      "flos": 13780375142400.0,
      "grad_norm": 6.3671877995590425,
      "learning_rate": 3.878109073641219e-06,
      "loss": 0.902,
      "num_input_tokens_seen": 24148170,
      "step": 1151
    },
    {
      "epoch": 0.13851981001623279,
      "flos": 20155187466240.0,
      "grad_norm": 3.555860522887187,
      "learning_rate": 3.877841145511366e-06,
      "loss": 1.0083,
      "num_input_tokens_seen": 24170630,
      "step": 1152
    },
    {
      "epoch": 0.13864005290687187,
      "flos": 15063679979520.0,
      "grad_norm": 5.250886388156586,
      "learning_rate": 3.8775729325145035e-06,
      "loss": 1.0312,
      "num_input_tokens_seen": 24189585,
      "step": 1153
    },
    {
      "epoch": 0.13876029579751098,
      "flos": 51321391779840.0,
      "grad_norm": 0.8308327150932417,
      "learning_rate": 3.877304434691321e-06,
      "loss": 0.841,
      "num_input_tokens_seen": 24256155,
      "step": 1154
    },
    {
      "epoch": 0.13888053868815006,
      "flos": 15586877214720.0,
      "grad_norm": 14.151991180399923,
      "learning_rate": 3.877035652082548e-06,
      "loss": 0.9966,
      "num_input_tokens_seen": 24275320,
      "step": 1155
    },
    {
      "epoch": 0.13900078157878915,
      "flos": 13910714572800.0,
      "grad_norm": 7.397589791060118,
      "learning_rate": 3.87676658472896e-06,
      "loss": 1.0455,
      "num_input_tokens_seen": 24293850,
      "step": 1156
    },
    {
      "epoch": 0.13912102446942826,
      "flos": 16088053493760.0,
      "grad_norm": 9.081434792241929,
      "learning_rate": 3.876497232671372e-06,
      "loss": 1.0514,
      "num_input_tokens_seen": 24313525,
      "step": 1157
    },
    {
      "epoch": 0.13924126736006734,
      "flos": 21122611814400.0,
      "grad_norm": 8.268089576003419,
      "learning_rate": 3.876227595950647e-06,
      "loss": 1.0197,
      "num_input_tokens_seen": 24332675,
      "step": 1158
    },
    {
      "epoch": 0.13936151025070642,
      "flos": 19523058585600.0,
      "grad_norm": 5.295172043437729,
      "learning_rate": 3.875957674607686e-06,
      "loss": 0.9855,
      "num_input_tokens_seen": 24354670,
      "step": 1159
    },
    {
      "epoch": 0.1394817531413455,
      "flos": 11604446515200.0,
      "grad_norm": 6.326618684051432,
      "learning_rate": 3.8756874686834386e-06,
      "loss": 1.0795,
      "num_input_tokens_seen": 24372605,
      "step": 1160
    },
    {
      "epoch": 0.13960199603198462,
      "flos": 22042069954560.0,
      "grad_norm": 5.173016839882508,
      "learning_rate": 3.875416978218893e-06,
      "loss": 1.0049,
      "num_input_tokens_seen": 24395520,
      "step": 1161
    },
    {
      "epoch": 0.1397222389226237,
      "flos": 12836535336960.0,
      "grad_norm": 7.289742032315346,
      "learning_rate": 3.8751462032550835e-06,
      "loss": 1.0249,
      "num_input_tokens_seen": 24412245,
      "step": 1162
    },
    {
      "epoch": 0.13984248181326278,
      "flos": 11944586035200.0,
      "grad_norm": 10.189260880578162,
      "learning_rate": 3.874875143833085e-06,
      "loss": 1.0166,
      "num_input_tokens_seen": 24430205,
      "step": 1163
    },
    {
      "epoch": 0.1399627247039019,
      "flos": 38714018304000.0,
      "grad_norm": 7.290232868007399,
      "learning_rate": 3.874603799994019e-06,
      "loss": 0.8854,
      "num_input_tokens_seen": 24453460,
      "step": 1164
    },
    {
      "epoch": 0.14008296759454097,
      "flos": 8274203504640.0,
      "grad_norm": 4.570512607464847,
      "learning_rate": 3.874332171779046e-06,
      "loss": 1.0842,
      "num_input_tokens_seen": 24468060,
      "step": 1165
    },
    {
      "epoch": 0.14020321048518006,
      "flos": 15642048184320.0,
      "grad_norm": 13.949709894041872,
      "learning_rate": 3.874060259229373e-06,
      "loss": 0.9591,
      "num_input_tokens_seen": 24489355,
      "step": 1166
    },
    {
      "epoch": 0.14032345337581917,
      "flos": 16638759567360.0,
      "grad_norm": 5.893771760319884,
      "learning_rate": 3.873788062386249e-06,
      "loss": 1.118,
      "num_input_tokens_seen": 24507335,
      "step": 1167
    },
    {
      "epoch": 0.14044369626645825,
      "flos": 21124696596480.0,
      "grad_norm": 5.602378285697485,
      "learning_rate": 3.873515581290965e-06,
      "loss": 1.0322,
      "num_input_tokens_seen": 24531860,
      "step": 1168
    },
    {
      "epoch": 0.14056393915709733,
      "flos": 12994107678720.0,
      "grad_norm": 4.475662114266892,
      "learning_rate": 3.8732428159848575e-06,
      "loss": 0.9553,
      "num_input_tokens_seen": 24550555,
      "step": 1169
    },
    {
      "epoch": 0.14068418204773642,
      "flos": 18998267105280.0,
      "grad_norm": 3.9243291342276017,
      "learning_rate": 3.872969766509304e-06,
      "loss": 0.991,
      "num_input_tokens_seen": 24570830,
      "step": 1170
    },
    {
      "epoch": 0.14080442493837553,
      "flos": 46724867543040.0,
      "grad_norm": 0.9371341627308303,
      "learning_rate": 3.872696432905726e-06,
      "loss": 0.7843,
      "num_input_tokens_seen": 24631370,
      "step": 1171
    },
    {
      "epoch": 0.1409246678290146,
      "flos": 18343872000000.0,
      "grad_norm": 11.242457595348231,
      "learning_rate": 3.872422815215589e-06,
      "loss": 0.9013,
      "num_input_tokens_seen": 24650170,
      "step": 1172
    },
    {
      "epoch": 0.1410449107196537,
      "flos": 15534557491200.0,
      "grad_norm": 4.0468695798859695,
      "learning_rate": 3.8721489134803994e-06,
      "loss": 0.9409,
      "num_input_tokens_seen": 24668680,
      "step": 1173
    },
    {
      "epoch": 0.1411651536102928,
      "flos": 11808176209920.0,
      "grad_norm": 15.14253517281571,
      "learning_rate": 3.871874727741707e-06,
      "loss": 0.9176,
      "num_input_tokens_seen": 24685630,
      "step": 1174
    },
    {
      "epoch": 0.1412853965009319,
      "flos": 14904728002560.0,
      "grad_norm": 10.879086833667275,
      "learning_rate": 3.871600258041108e-06,
      "loss": 1.1674,
      "num_input_tokens_seen": 24704875,
      "step": 1175
    },
    {
      "epoch": 0.14140563939157097,
      "flos": 14432532172800.0,
      "grad_norm": 3.9956660633790535,
      "learning_rate": 3.871325504420238e-06,
      "loss": 1.0449,
      "num_input_tokens_seen": 24723585,
      "step": 1176
    },
    {
      "epoch": 0.14152588228221005,
      "flos": 14960236216320.0,
      "grad_norm": 5.565802633116419,
      "learning_rate": 3.871050466920776e-06,
      "loss": 1.0222,
      "num_input_tokens_seen": 24743210,
      "step": 1177
    },
    {
      "epoch": 0.14164612517284916,
      "flos": 12915137556480.0,
      "grad_norm": 8.697688652902679,
      "learning_rate": 3.870775145584447e-06,
      "loss": 0.9969,
      "num_input_tokens_seen": 24760710,
      "step": 1178
    },
    {
      "epoch": 0.14176636806348825,
      "flos": 16164111052800.0,
      "grad_norm": 6.687754118182252,
      "learning_rate": 3.8704995404530145e-06,
      "loss": 0.8335,
      "num_input_tokens_seen": 24776055,
      "step": 1179
    },
    {
      "epoch": 0.14188661095412733,
      "flos": 15695011737600.0,
      "grad_norm": 4.40940624850506,
      "learning_rate": 3.87022365156829e-06,
      "loss": 1.0435,
      "num_input_tokens_seen": 24796490,
      "step": 1180
    },
    {
      "epoch": 0.14200685384476644,
      "flos": 17320050339840.0,
      "grad_norm": 10.793365305554177,
      "learning_rate": 3.869947478972123e-06,
      "loss": 1.0048,
      "num_input_tokens_seen": 24817235,
      "step": 1181
    },
    {
      "epoch": 0.14212709673540552,
      "flos": 17082373509120.0,
      "grad_norm": 4.72067868057128,
      "learning_rate": 3.869671022706412e-06,
      "loss": 1.0108,
      "num_input_tokens_seen": 24835685,
      "step": 1182
    },
    {
      "epoch": 0.1422473396260446,
      "flos": 18814442926080.0,
      "grad_norm": 4.331439680429614,
      "learning_rate": 3.869394282813092e-06,
      "loss": 0.8531,
      "num_input_tokens_seen": 24854605,
      "step": 1183
    },
    {
      "epoch": 0.1423675825166837,
      "flos": 12075783905280.0,
      "grad_norm": 5.352406233385375,
      "learning_rate": 3.869117259334147e-06,
      "loss": 1.0797,
      "num_input_tokens_seen": 24872250,
      "step": 1184
    },
    {
      "epoch": 0.1424878254073228,
      "flos": 12703835197440.0,
      "grad_norm": 3.306499011752715,
      "learning_rate": 3.868839952311599e-06,
      "loss": 1.0211,
      "num_input_tokens_seen": 24889925,
      "step": 1185
    },
    {
      "epoch": 0.14260806829796188,
      "flos": 14410419240960.0,
      "grad_norm": 19.907351476835526,
      "learning_rate": 3.868562361787516e-06,
      "loss": 0.999,
      "num_input_tokens_seen": 24908775,
      "step": 1186
    },
    {
      "epoch": 0.14272831118860096,
      "flos": 16481708421120.0,
      "grad_norm": 4.38649134330978,
      "learning_rate": 3.868284487804009e-06,
      "loss": 0.8926,
      "num_input_tokens_seen": 24927725,
      "step": 1187
    },
    {
      "epoch": 0.14284855407924008,
      "flos": 19389377372160.0,
      "grad_norm": 7.823138996187024,
      "learning_rate": 3.86800633040323e-06,
      "loss": 0.992,
      "num_input_tokens_seen": 24948035,
      "step": 1188
    },
    {
      "epoch": 0.14296879696987916,
      "flos": 20074009927680.0,
      "grad_norm": 23.50079733936569,
      "learning_rate": 3.867727889627376e-06,
      "loss": 0.9862,
      "num_input_tokens_seen": 24967370,
      "step": 1189
    },
    {
      "epoch": 0.14308903986051824,
      "flos": 13753878036480.0,
      "grad_norm": 9.322914393476212,
      "learning_rate": 3.867449165518687e-06,
      "loss": 0.9763,
      "num_input_tokens_seen": 24983560,
      "step": 1190
    },
    {
      "epoch": 0.14320928275115732,
      "flos": 12364952678400.0,
      "grad_norm": 5.703725955602296,
      "learning_rate": 3.867170158119444e-06,
      "loss": 0.9009,
      "num_input_tokens_seen": 25002280,
      "step": 1191
    },
    {
      "epoch": 0.14332952564179643,
      "flos": 15245174108160.0,
      "grad_norm": 5.976518139643376,
      "learning_rate": 3.866890867471972e-06,
      "loss": 0.9557,
      "num_input_tokens_seen": 25020470,
      "step": 1192
    },
    {
      "epoch": 0.14344976853243552,
      "flos": 11602453708800.0,
      "grad_norm": 8.24074381980251,
      "learning_rate": 3.86661129361864e-06,
      "loss": 1.0955,
      "num_input_tokens_seen": 25034680,
      "step": 1193
    },
    {
      "epoch": 0.1435700114230746,
      "flos": 13413646540800.0,
      "grad_norm": 6.420663686892172,
      "learning_rate": 3.866331436601859e-06,
      "loss": 1.0461,
      "num_input_tokens_seen": 25052395,
      "step": 1194
    },
    {
      "epoch": 0.1436902543137137,
      "flos": 14016304435200.0,
      "grad_norm": 14.328590377188736,
      "learning_rate": 3.866051296464083e-06,
      "loss": 0.9415,
      "num_input_tokens_seen": 25070950,
      "step": 1195
    },
    {
      "epoch": 0.1438104972043528,
      "flos": 10372296376320.0,
      "grad_norm": 5.497809580802847,
      "learning_rate": 3.86577087324781e-06,
      "loss": 1.055,
      "num_input_tokens_seen": 25087160,
      "step": 1196
    },
    {
      "epoch": 0.14393074009499188,
      "flos": 12102464962560.0,
      "grad_norm": 4.734750864953068,
      "learning_rate": 3.865490166995578e-06,
      "loss": 0.9687,
      "num_input_tokens_seen": 25105110,
      "step": 1197
    },
    {
      "epoch": 0.144050982985631,
      "flos": 21720333680640.0,
      "grad_norm": 5.346078108993633,
      "learning_rate": 3.86520917774997e-06,
      "loss": 1.0306,
      "num_input_tokens_seen": 25124265,
      "step": 1198
    },
    {
      "epoch": 0.14417122587627007,
      "flos": 12653079060480.0,
      "grad_norm": 5.358876196779248,
      "learning_rate": 3.864927905553614e-06,
      "loss": 0.9508,
      "num_input_tokens_seen": 25141895,
      "step": 1199
    },
    {
      "epoch": 0.14429146876690915,
      "flos": 15351683727360.0,
      "grad_norm": 7.3201691930009485,
      "learning_rate": 3.8646463504491765e-06,
      "loss": 1.0837,
      "num_input_tokens_seen": 25161750,
      "step": 1200
    },
    {
      "epoch": 0.14441171165754824,
      "flos": 16533660241920.0,
      "grad_norm": 3.615879555645832,
      "learning_rate": 3.8643645124793705e-06,
      "loss": 1.034,
      "num_input_tokens_seen": 25180370,
      "step": 1201
    },
    {
      "epoch": 0.14453195454818735,
      "flos": 30617437839360.0,
      "grad_norm": 3.4642874390453153,
      "learning_rate": 3.8640823916869515e-06,
      "loss": 0.9378,
      "num_input_tokens_seen": 25204400,
      "step": 1202
    },
    {
      "epoch": 0.14465219743882643,
      "flos": 19392504545280.0,
      "grad_norm": 2.950008301804063,
      "learning_rate": 3.863799988114714e-06,
      "loss": 0.985,
      "num_input_tokens_seen": 25226150,
      "step": 1203
    },
    {
      "epoch": 0.1447724403294655,
      "flos": 11813786726400.0,
      "grad_norm": 7.008969418768527,
      "learning_rate": 3.863517301805502e-06,
      "loss": 0.8996,
      "num_input_tokens_seen": 25244260,
      "step": 1204
    },
    {
      "epoch": 0.14489268322010462,
      "flos": 14243741306880.0,
      "grad_norm": 5.781820403131895,
      "learning_rate": 3.863234332802196e-06,
      "loss": 1.1647,
      "num_input_tokens_seen": 25256185,
      "step": 1205
    },
    {
      "epoch": 0.1450129261107437,
      "flos": 19313166520320.0,
      "grad_norm": 19.46451411082004,
      "learning_rate": 3.862951081147723e-06,
      "loss": 0.923,
      "num_input_tokens_seen": 25276070,
      "step": 1206
    },
    {
      "epoch": 0.1451331690013828,
      "flos": 18290233958400.0,
      "grad_norm": 8.826226980700403,
      "learning_rate": 3.862667546885053e-06,
      "loss": 0.9774,
      "num_input_tokens_seen": 25294340,
      "step": 1207
    },
    {
      "epoch": 0.14525341189202187,
      "flos": 18315228794880.0,
      "grad_norm": 7.6979403428749,
      "learning_rate": 3.8623837300571965e-06,
      "loss": 0.9289,
      "num_input_tokens_seen": 25313045,
      "step": 1208
    },
    {
      "epoch": 0.14537365478266098,
      "flos": 16401021419520.0,
      "grad_norm": 5.750907494623674,
      "learning_rate": 3.8620996307072085e-06,
      "loss": 1.0411,
      "num_input_tokens_seen": 25333470,
      "step": 1209
    },
    {
      "epoch": 0.14549389767330007,
      "flos": 14619790110720.0,
      "grad_norm": 11.588356593793698,
      "learning_rate": 3.861815248878188e-06,
      "loss": 0.8463,
      "num_input_tokens_seen": 25350675,
      "step": 1210
    },
    {
      "epoch": 0.14561414056393915,
      "flos": 10686245376000.0,
      "grad_norm": 9.156764513994345,
      "learning_rate": 3.861530584613274e-06,
      "loss": 0.9994,
      "num_input_tokens_seen": 25368395,
      "step": 1211
    },
    {
      "epoch": 0.14573438345457826,
      "flos": 13780712386560.0,
      "grad_norm": 6.492840859121299,
      "learning_rate": 3.86124563795565e-06,
      "loss": 1.0204,
      "num_input_tokens_seen": 25386930,
      "step": 1212
    },
    {
      "epoch": 0.14585462634521734,
      "flos": 17662397276160.0,
      "grad_norm": 11.086045203515749,
      "learning_rate": 3.860960408948543e-06,
      "loss": 0.9098,
      "num_input_tokens_seen": 25408400,
      "step": 1213
    },
    {
      "epoch": 0.14597486923585642,
      "flos": 10920917667840.0,
      "grad_norm": 8.59194538935918,
      "learning_rate": 3.860674897635222e-06,
      "loss": 1.1013,
      "num_input_tokens_seen": 25424605,
      "step": 1214
    },
    {
      "epoch": 0.1460951121264955,
      "flos": 11788362670080.0,
      "grad_norm": 7.179958201422201,
      "learning_rate": 3.860389104058998e-06,
      "loss": 1.0324,
      "num_input_tokens_seen": 25442555,
      "step": 1215
    },
    {
      "epoch": 0.14621535501713462,
      "flos": 17687238819840.0,
      "grad_norm": 20.810077514797367,
      "learning_rate": 3.860103028263227e-06,
      "loss": 0.9134,
      "num_input_tokens_seen": 25465380,
      "step": 1216
    },
    {
      "epoch": 0.1463355979077737,
      "flos": 17949327974400.0,
      "grad_norm": 4.929233439432042,
      "learning_rate": 3.859816670291304e-06,
      "loss": 0.8872,
      "num_input_tokens_seen": 25484195,
      "step": 1217
    },
    {
      "epoch": 0.14645584079841278,
      "flos": 15668391997440.0,
      "grad_norm": 4.838857234632801,
      "learning_rate": 3.859530030186672e-06,
      "loss": 1.0944,
      "num_input_tokens_seen": 25500925,
      "step": 1218
    },
    {
      "epoch": 0.1465760836890519,
      "flos": 16796638494720.0,
      "grad_norm": 4.122549311595688,
      "learning_rate": 3.859243107992813e-06,
      "loss": 1.0174,
      "num_input_tokens_seen": 25519450,
      "step": 1219
    },
    {
      "epoch": 0.14669632657969098,
      "flos": 26701928448000.0,
      "grad_norm": 6.108089536179248,
      "learning_rate": 3.858955903753252e-06,
      "loss": 0.9876,
      "num_input_tokens_seen": 25537810,
      "step": 1220
    },
    {
      "epoch": 0.14681656947033006,
      "flos": 20204502650880.0,
      "grad_norm": 2.964044154004324,
      "learning_rate": 3.858668417511559e-06,
      "loss": 1.0319,
      "num_input_tokens_seen": 25560280,
      "step": 1221
    },
    {
      "epoch": 0.14693681236096917,
      "flos": 13099482931200.0,
      "grad_norm": 8.895521910410526,
      "learning_rate": 3.8583806493113445e-06,
      "loss": 0.9586,
      "num_input_tokens_seen": 25578345,
      "step": 1222
    },
    {
      "epoch": 0.14705705525160825,
      "flos": 14751202590720.0,
      "grad_norm": 4.447224237771299,
      "learning_rate": 3.858092599196263e-06,
      "loss": 1.0267,
      "num_input_tokens_seen": 25596020,
      "step": 1223
    },
    {
      "epoch": 0.14717729814224734,
      "flos": 21330449756160.0,
      "grad_norm": 6.615755649118087,
      "learning_rate": 3.857804267210012e-06,
      "loss": 1.0242,
      "num_input_tokens_seen": 25615040,
      "step": 1224
    },
    {
      "epoch": 0.14729754103288642,
      "flos": 14226257817600.0,
      "grad_norm": 3.2767426549948944,
      "learning_rate": 3.857515653396331e-06,
      "loss": 1.0803,
      "num_input_tokens_seen": 25631970,
      "step": 1225
    },
    {
      "epoch": 0.14741778392352553,
      "flos": 13675797012480.0,
      "grad_norm": 5.355596606219042,
      "learning_rate": 3.857226757799002e-06,
      "loss": 1.0633,
      "num_input_tokens_seen": 25649245,
      "step": 1226
    },
    {
      "epoch": 0.1475380268141646,
      "flos": 18081016381440.0,
      "grad_norm": 5.2938989251276585,
      "learning_rate": 3.85693758046185e-06,
      "loss": 0.9243,
      "num_input_tokens_seen": 25667255,
      "step": 1227
    },
    {
      "epoch": 0.1476582697048037,
      "flos": 14800977653760.0,
      "grad_norm": 39.90735272801007,
      "learning_rate": 3.8566481214287435e-06,
      "loss": 1.0272,
      "num_input_tokens_seen": 25685095,
      "step": 1228
    },
    {
      "epoch": 0.1477785125954428,
      "flos": 9900345815040.0,
      "grad_norm": 5.494344578011007,
      "learning_rate": 3.8563583807435935e-06,
      "loss": 1.103,
      "num_input_tokens_seen": 25700960,
      "step": 1229
    },
    {
      "epoch": 0.1478987554860819,
      "flos": 14562779627520.0,
      "grad_norm": 8.143408477015946,
      "learning_rate": 3.856068358450353e-06,
      "loss": 0.9709,
      "num_input_tokens_seen": 25720630,
      "step": 1230
    },
    {
      "epoch": 0.14801899837672097,
      "flos": 12651822059520.0,
      "grad_norm": 4.381894809437086,
      "learning_rate": 3.8557780545930186e-06,
      "loss": 1.057,
      "num_input_tokens_seen": 25738765,
      "step": 1231
    },
    {
      "epoch": 0.14813924126736006,
      "flos": 14824868782080.0,
      "grad_norm": 3.550496676448958,
      "learning_rate": 3.855487469215628e-06,
      "loss": 0.9791,
      "num_input_tokens_seen": 25757415,
      "step": 1232
    },
    {
      "epoch": 0.14825948415799917,
      "flos": 26707201720320.0,
      "grad_norm": 5.138988849022817,
      "learning_rate": 3.855196602362264e-06,
      "loss": 0.92,
      "num_input_tokens_seen": 25780055,
      "step": 1233
    },
    {
      "epoch": 0.14837972704863825,
      "flos": 15639595499520.0,
      "grad_norm": 4.70492245617327,
      "learning_rate": 3.854905454077051e-06,
      "loss": 1.1427,
      "num_input_tokens_seen": 25797385,
      "step": 1234
    },
    {
      "epoch": 0.14849996993927733,
      "flos": 14908192419840.0,
      "grad_norm": 42.04903347070314,
      "learning_rate": 3.854614024404155e-06,
      "loss": 1.0716,
      "num_input_tokens_seen": 25815415,
      "step": 1235
    },
    {
      "epoch": 0.14862021282991644,
      "flos": 14226687037440.0,
      "grad_norm": 5.571882756902288,
      "learning_rate": 3.8543223133877865e-06,
      "loss": 1.0971,
      "num_input_tokens_seen": 25833730,
      "step": 1236
    },
    {
      "epoch": 0.14874045572055553,
      "flos": 16141354291200.0,
      "grad_norm": 6.162184280448059,
      "learning_rate": 3.854030321072198e-06,
      "loss": 1.0696,
      "num_input_tokens_seen": 25853355,
      "step": 1237
    },
    {
      "epoch": 0.1488606986111946,
      "flos": 18081169674240.0,
      "grad_norm": 6.09417597920166,
      "learning_rate": 3.853738047501682e-06,
      "loss": 0.9333,
      "num_input_tokens_seen": 25873635,
      "step": 1238
    },
    {
      "epoch": 0.1489809415018337,
      "flos": 12049501409280.0,
      "grad_norm": 8.344414730953181,
      "learning_rate": 3.85344549272058e-06,
      "loss": 0.9684,
      "num_input_tokens_seen": 25891335,
      "step": 1239
    },
    {
      "epoch": 0.1491011843924728,
      "flos": 24009240883200.0,
      "grad_norm": 15.951179408172887,
      "learning_rate": 3.853152656773269e-06,
      "loss": 1.0305,
      "num_input_tokens_seen": 25912490,
      "step": 1240
    },
    {
      "epoch": 0.14922142728311188,
      "flos": 15039512924160.0,
      "grad_norm": 4.421004462397994,
      "learning_rate": 3.852859539704174e-06,
      "loss": 1.0363,
      "num_input_tokens_seen": 25931510,
      "step": 1241
    },
    {
      "epoch": 0.14934167017375097,
      "flos": 21278497935360.0,
      "grad_norm": 4.841382438646665,
      "learning_rate": 3.85256614155776e-06,
      "loss": 0.9549,
      "num_input_tokens_seen": 25951360,
      "step": 1242
    },
    {
      "epoch": 0.14946191306439008,
      "flos": 12049899970560.0,
      "grad_norm": 3.4336992926753984,
      "learning_rate": 3.852272462378535e-06,
      "loss": 0.947,
      "num_input_tokens_seen": 25968955,
      "step": 1243
    },
    {
      "epoch": 0.14958215595502916,
      "flos": 11079532400640.0,
      "grad_norm": 9.48927303830015,
      "learning_rate": 3.85197850221105e-06,
      "loss": 0.9902,
      "num_input_tokens_seen": 25984975,
      "step": 1244
    },
    {
      "epoch": 0.14970239884566824,
      "flos": 23612888002560.0,
      "grad_norm": 3.5451241763900727,
      "learning_rate": 3.851684261099899e-06,
      "loss": 0.9595,
      "num_input_tokens_seen": 26006435,
      "step": 1245
    },
    {
      "epoch": 0.14982264173630733,
      "flos": 12626214051840.0,
      "grad_norm": 8.960270398691982,
      "learning_rate": 3.851389739089718e-06,
      "loss": 1.0661,
      "num_input_tokens_seen": 26022775,
      "step": 1246
    },
    {
      "epoch": 0.14994288462694644,
      "flos": 23109749575680.0,
      "grad_norm": 3.5922350800272276,
      "learning_rate": 3.851094936225186e-06,
      "loss": 1.0102,
      "num_input_tokens_seen": 26043380,
      "step": 1247
    },
    {
      "epoch": 0.15006312751758552,
      "flos": 22668557660160.0,
      "grad_norm": 2.49515703322514,
      "learning_rate": 3.850799852551024e-06,
      "loss": 0.9739,
      "num_input_tokens_seen": 26065520,
      "step": 1248
    },
    {
      "epoch": 0.1501833704082246,
      "flos": 11761589637120.0,
      "grad_norm": 6.79211343181233,
      "learning_rate": 3.850504488111995e-06,
      "loss": 1.0574,
      "num_input_tokens_seen": 26081915,
      "step": 1249
    },
    {
      "epoch": 0.15030361329886371,
      "flos": 16685989969920.0,
      "grad_norm": 5.083690422314927,
      "learning_rate": 3.850208842952907e-06,
      "loss": 1.0284,
      "num_input_tokens_seen": 26100440,
      "step": 1250
    },
    {
      "epoch": 0.1504238561895028,
      "flos": 18237822259200.0,
      "grad_norm": 3.972196269154332,
      "learning_rate": 3.849912917118608e-06,
      "loss": 0.9962,
      "num_input_tokens_seen": 26121200,
      "step": 1251
    },
    {
      "epoch": 0.15054409908014188,
      "flos": 37265000386560.0,
      "grad_norm": 1.2604548498417882,
      "learning_rate": 3.849616710653992e-06,
      "loss": 0.8247,
      "num_input_tokens_seen": 26182390,
      "step": 1252
    },
    {
      "epoch": 0.150664341970781,
      "flos": 13387425361920.0,
      "grad_norm": 4.323010007777422,
      "learning_rate": 3.84932022360399e-06,
      "loss": 0.955,
      "num_input_tokens_seen": 26200775,
      "step": 1253
    },
    {
      "epoch": 0.15078458486142007,
      "flos": 15746748948480.0,
      "grad_norm": 5.830239738128505,
      "learning_rate": 3.849023456013581e-06,
      "loss": 1.0247,
      "num_input_tokens_seen": 26218055,
      "step": 1254
    },
    {
      "epoch": 0.15090482775205916,
      "flos": 18972781731840.0,
      "grad_norm": 22.425310739599087,
      "learning_rate": 3.848726407927784e-06,
      "loss": 0.8203,
      "num_input_tokens_seen": 26238160,
      "step": 1255
    },
    {
      "epoch": 0.15102507064269824,
      "flos": 15485058355200.0,
      "grad_norm": 5.91602465231236,
      "learning_rate": 3.84842907939166e-06,
      "loss": 1.0703,
      "num_input_tokens_seen": 26257105,
      "step": 1256
    },
    {
      "epoch": 0.15114531353333735,
      "flos": 16219527290880.0,
      "grad_norm": 6.6991808831837965,
      "learning_rate": 3.8481314704503146e-06,
      "loss": 0.904,
      "num_input_tokens_seen": 26276655,
      "step": 1257
    },
    {
      "epoch": 0.15126555642397643,
      "flos": 13964076687360.0,
      "grad_norm": 5.300781969648072,
      "learning_rate": 3.847833581148895e-06,
      "loss": 1.0793,
      "num_input_tokens_seen": 26295285,
      "step": 1258
    },
    {
      "epoch": 0.15138579931461552,
      "flos": 20462667509760.0,
      "grad_norm": 5.065231528229515,
      "learning_rate": 3.84753541153259e-06,
      "loss": 0.9997,
      "num_input_tokens_seen": 26314575,
      "step": 1259
    },
    {
      "epoch": 0.15150604220525463,
      "flos": 15720742379520.0,
      "grad_norm": 3.4742657221849926,
      "learning_rate": 3.847236961646633e-06,
      "loss": 1.0365,
      "num_input_tokens_seen": 26333275,
      "step": 1260
    },
    {
      "epoch": 0.1516262850958937,
      "flos": 9138797260800.0,
      "grad_norm": 7.834706720390448,
      "learning_rate": 3.846938231536296e-06,
      "loss": 0.993,
      "num_input_tokens_seen": 26348615,
      "step": 1261
    },
    {
      "epoch": 0.1517465279865328,
      "flos": 15483464110080.0,
      "grad_norm": 4.367632739141256,
      "learning_rate": 3.8466392212468995e-06,
      "loss": 1.013,
      "num_input_tokens_seen": 26368525,
      "step": 1262
    },
    {
      "epoch": 0.15186677087717187,
      "flos": 41632317665280.0,
      "grad_norm": 0.839413838911739,
      "learning_rate": 3.8463399308238e-06,
      "loss": 0.8445,
      "num_input_tokens_seen": 26427350,
      "step": 1263
    },
    {
      "epoch": 0.15198701376781099,
      "flos": 23297344757760.0,
      "grad_norm": 3.7807362104932634,
      "learning_rate": 3.846040360312402e-06,
      "loss": 0.8467,
      "num_input_tokens_seen": 26450330,
      "step": 1264
    },
    {
      "epoch": 0.15210725665845007,
      "flos": 20230877122560.0,
      "grad_norm": 10.20943668137837,
      "learning_rate": 3.8457405097581485e-06,
      "loss": 1.0194,
      "num_input_tokens_seen": 26469040,
      "step": 1265
    },
    {
      "epoch": 0.15222749954908915,
      "flos": 14147716915200.0,
      "grad_norm": 3.8469589608494603,
      "learning_rate": 3.8454403792065275e-06,
      "loss": 0.9814,
      "num_input_tokens_seen": 26487580,
      "step": 1266
    },
    {
      "epoch": 0.15234774243972826,
      "flos": 15143784468480.0,
      "grad_norm": 5.804871834745558,
      "learning_rate": 3.845139968703068e-06,
      "loss": 1.0546,
      "num_input_tokens_seen": 26504820,
      "step": 1267
    },
    {
      "epoch": 0.15246798533036734,
      "flos": 18473843527680.0,
      "grad_norm": 20.24628932533842,
      "learning_rate": 3.844839278293342e-06,
      "loss": 1.0332,
      "num_input_tokens_seen": 26525390,
      "step": 1268
    },
    {
      "epoch": 0.15258822822100643,
      "flos": 18368989470720.0,
      "grad_norm": 3.9938992229158234,
      "learning_rate": 3.8445383080229654e-06,
      "loss": 0.9639,
      "num_input_tokens_seen": 26541125,
      "step": 1269
    },
    {
      "epoch": 0.1527084711116455,
      "flos": 17976376934400.0,
      "grad_norm": 18.116501914044072,
      "learning_rate": 3.844237057937593e-06,
      "loss": 0.9381,
      "num_input_tokens_seen": 26559850,
      "step": 1270
    },
    {
      "epoch": 0.15282871400228462,
      "flos": 20832676577280.0,
      "grad_norm": 4.853378275533941,
      "learning_rate": 3.843935528082926e-06,
      "loss": 0.9804,
      "num_input_tokens_seen": 26580595,
      "step": 1271
    },
    {
      "epoch": 0.1529489568929237,
      "flos": 14826187100160.0,
      "grad_norm": 3.2099891617256273,
      "learning_rate": 3.843633718504704e-06,
      "loss": 1.0551,
      "num_input_tokens_seen": 26598760,
      "step": 1272
    },
    {
      "epoch": 0.1530691997835628,
      "flos": 14252356362240.0,
      "grad_norm": 4.323499059704153,
      "learning_rate": 3.843331629248715e-06,
      "loss": 1.0947,
      "num_input_tokens_seen": 26616080,
      "step": 1273
    },
    {
      "epoch": 0.1531894426742019,
      "flos": 20486589296640.0,
      "grad_norm": 7.258757202292013,
      "learning_rate": 3.843029260360782e-06,
      "loss": 0.9744,
      "num_input_tokens_seen": 26634170,
      "step": 1274
    },
    {
      "epoch": 0.15330968556484098,
      "flos": 15799252623360.0,
      "grad_norm": 6.327385903983547,
      "learning_rate": 3.8427266118867755e-06,
      "loss": 0.988,
      "num_input_tokens_seen": 26653640,
      "step": 1275
    },
    {
      "epoch": 0.15342992845548006,
      "flos": 19602059366400.0,
      "grad_norm": 10.813748502589013,
      "learning_rate": 3.842423683872608e-06,
      "loss": 1.0276,
      "num_input_tokens_seen": 26673935,
      "step": 1276
    },
    {
      "epoch": 0.15355017134611917,
      "flos": 13911542353920.0,
      "grad_norm": 5.957442992942539,
      "learning_rate": 3.842120476364232e-06,
      "loss": 0.9779,
      "num_input_tokens_seen": 26692105,
      "step": 1277
    },
    {
      "epoch": 0.15367041423675826,
      "flos": 13098716467200.0,
      "grad_norm": 5.320995763073349,
      "learning_rate": 3.841816989407644e-06,
      "loss": 1.025,
      "num_input_tokens_seen": 26707315,
      "step": 1278
    },
    {
      "epoch": 0.15379065712739734,
      "flos": 29591715348480.0,
      "grad_norm": 3.7177357292216917,
      "learning_rate": 3.841513223048884e-06,
      "loss": 0.9592,
      "num_input_tokens_seen": 26727720,
      "step": 1279
    },
    {
      "epoch": 0.15391090001803642,
      "flos": 15668453314560.0,
      "grad_norm": 6.969342535855163,
      "learning_rate": 3.841209177334031e-06,
      "loss": 0.9656,
      "num_input_tokens_seen": 26745800,
      "step": 1280
    },
    {
      "epoch": 0.15403114290867553,
      "flos": 10922389278720.0,
      "grad_norm": 4.247164958742194,
      "learning_rate": 3.84090485230921e-06,
      "loss": 0.954,
      "num_input_tokens_seen": 26763760,
      "step": 1281
    },
    {
      "epoch": 0.15415138579931462,
      "flos": 12703559270400.0,
      "grad_norm": 5.465486610017203,
      "learning_rate": 3.840600248020588e-06,
      "loss": 0.9593,
      "num_input_tokens_seen": 26780420,
      "step": 1282
    },
    {
      "epoch": 0.1542716286899537,
      "flos": 8031529328640.0,
      "grad_norm": 3.8119565232430466,
      "learning_rate": 3.840295364514371e-06,
      "loss": 0.9954,
      "num_input_tokens_seen": 26797520,
      "step": 1283
    },
    {
      "epoch": 0.1543918715805928,
      "flos": 12338148986880.0,
      "grad_norm": 6.188474019769282,
      "learning_rate": 3.83999020183681e-06,
      "loss": 0.9795,
      "num_input_tokens_seen": 26815935,
      "step": 1284
    },
    {
      "epoch": 0.1545121144712319,
      "flos": 12601372508160.0,
      "grad_norm": 5.897077269691065,
      "learning_rate": 3.839684760034199e-06,
      "loss": 0.9822,
      "num_input_tokens_seen": 26833860,
      "step": 1285
    },
    {
      "epoch": 0.15463235736187098,
      "flos": 20099801886720.0,
      "grad_norm": 25.07560832387721,
      "learning_rate": 3.8393790391528716e-06,
      "loss": 0.8461,
      "num_input_tokens_seen": 26854275,
      "step": 1286
    },
    {
      "epoch": 0.15475260025251006,
      "flos": 16245319249920.0,
      "grad_norm": 12.317472896248024,
      "learning_rate": 3.8390730392392075e-06,
      "loss": 1.0852,
      "num_input_tokens_seen": 26873975,
      "step": 1287
    },
    {
      "epoch": 0.15487284314314917,
      "flos": 12469346856960.0,
      "grad_norm": 6.745618740514677,
      "learning_rate": 3.838766760339626e-06,
      "loss": 1.0052,
      "num_input_tokens_seen": 26892220,
      "step": 1288
    },
    {
      "epoch": 0.15499308603378825,
      "flos": 14250823434240.0,
      "grad_norm": 7.503228319275753,
      "learning_rate": 3.838460202500587e-06,
      "loss": 0.9723,
      "num_input_tokens_seen": 26907730,
      "step": 1289
    },
    {
      "epoch": 0.15511332892442733,
      "flos": 11257623429120.0,
      "grad_norm": 4.247438219686511,
      "learning_rate": 3.838153365768599e-06,
      "loss": 0.9397,
      "num_input_tokens_seen": 26923960,
      "step": 1290
    },
    {
      "epoch": 0.15523357181506645,
      "flos": 29487474462720.0,
      "grad_norm": 5.383445669287461,
      "learning_rate": 3.837846250190206e-06,
      "loss": 0.9475,
      "num_input_tokens_seen": 26946545,
      "step": 1291
    },
    {
      "epoch": 0.15535381470570553,
      "flos": 13098869760000.0,
      "grad_norm": 4.898144678249357,
      "learning_rate": 3.837538855811998e-06,
      "loss": 0.973,
      "num_input_tokens_seen": 26964440,
      "step": 1292
    },
    {
      "epoch": 0.1554740575963446,
      "flos": 9821253058560.0,
      "grad_norm": 4.4394924365147945,
      "learning_rate": 3.837231182680606e-06,
      "loss": 0.907,
      "num_input_tokens_seen": 26982125,
      "step": 1293
    },
    {
      "epoch": 0.1555943004869837,
      "flos": 14800487116800.0,
      "grad_norm": 5.163283111807426,
      "learning_rate": 3.836923230842706e-06,
      "loss": 0.9708,
      "num_input_tokens_seen": 27000960,
      "step": 1294
    },
    {
      "epoch": 0.1557145433776228,
      "flos": 15690566246400.0,
      "grad_norm": 12.882530010357959,
      "learning_rate": 3.836615000345011e-06,
      "loss": 1.0153,
      "num_input_tokens_seen": 27018860,
      "step": 1295
    },
    {
      "epoch": 0.1558347862682619,
      "flos": 14042617589760.0,
      "grad_norm": 4.1763884742521515,
      "learning_rate": 3.836306491234282e-06,
      "loss": 0.9826,
      "num_input_tokens_seen": 27036430,
      "step": 1296
    },
    {
      "epoch": 0.15595502915890097,
      "flos": 12206337945600.0,
      "grad_norm": 4.28928039775513,
      "learning_rate": 3.835997703557317e-06,
      "loss": 0.9381,
      "num_input_tokens_seen": 27052890,
      "step": 1297
    },
    {
      "epoch": 0.15607527204954008,
      "flos": 13990604451840.0,
      "grad_norm": 6.267755540979911,
      "learning_rate": 3.83568863736096e-06,
      "loss": 1.0014,
      "num_input_tokens_seen": 27071480,
      "step": 1298
    },
    {
      "epoch": 0.15619551494017916,
      "flos": 13125428183040.0,
      "grad_norm": 6.028143680715612,
      "learning_rate": 3.8353792926920975e-06,
      "loss": 1.0824,
      "num_input_tokens_seen": 27089850,
      "step": 1299
    },
    {
      "epoch": 0.15631575783081825,
      "flos": 14121863639040.0,
      "grad_norm": 4.352436628499321,
      "learning_rate": 3.835069669597655e-06,
      "loss": 1.0069,
      "num_input_tokens_seen": 27107960,
      "step": 1300
    },
    {
      "epoch": 0.15643600072145733,
      "flos": 14751079956480.0,
      "grad_norm": 6.119709143116027,
      "learning_rate": 3.834759768124603e-06,
      "loss": 1.0029,
      "num_input_tokens_seen": 27126555,
      "step": 1301
    },
    {
      "epoch": 0.15655624361209644,
      "flos": 13146743992320.0,
      "grad_norm": 9.062004128405713,
      "learning_rate": 3.834449588319953e-06,
      "loss": 0.9548,
      "num_input_tokens_seen": 27144310,
      "step": 1302
    },
    {
      "epoch": 0.15667648650273552,
      "flos": 17950646292480.0,
      "grad_norm": 5.068864753825863,
      "learning_rate": 3.834139130230758e-06,
      "loss": 1.0513,
      "num_input_tokens_seen": 27163335,
      "step": 1303
    },
    {
      "epoch": 0.1567967293933746,
      "flos": 17661262909440.0,
      "grad_norm": 2.76901941990681,
      "learning_rate": 3.833828393904117e-06,
      "loss": 1.0137,
      "num_input_tokens_seen": 27183335,
      "step": 1304
    },
    {
      "epoch": 0.15691697228401372,
      "flos": 13590725836800.0,
      "grad_norm": 4.197492832089109,
      "learning_rate": 3.833517379387165e-06,
      "loss": 0.9797,
      "num_input_tokens_seen": 27199510,
      "step": 1305
    },
    {
      "epoch": 0.1570372151746528,
      "flos": 17635317657600.0,
      "grad_norm": 3.4261994927575894,
      "learning_rate": 3.833206086727085e-06,
      "loss": 1.0899,
      "num_input_tokens_seen": 27218580,
      "step": 1306
    },
    {
      "epoch": 0.15715745806529188,
      "flos": 17687545405440.0,
      "grad_norm": 5.085957751255765,
      "learning_rate": 3.8328945159710994e-06,
      "loss": 0.9065,
      "num_input_tokens_seen": 27238480,
      "step": 1307
    },
    {
      "epoch": 0.157277700955931,
      "flos": 15537654005760.0,
      "grad_norm": 4.868627246489383,
      "learning_rate": 3.832582667166473e-06,
      "loss": 1.0726,
      "num_input_tokens_seen": 27258010,
      "step": 1308
    },
    {
      "epoch": 0.15739794384657008,
      "flos": 17449899233280.0,
      "grad_norm": 4.178449957343094,
      "learning_rate": 3.8322705403605125e-06,
      "loss": 1.0138,
      "num_input_tokens_seen": 27278075,
      "step": 1309
    },
    {
      "epoch": 0.15751818673720916,
      "flos": 12571257692160.0,
      "grad_norm": 3.8146776327619785,
      "learning_rate": 3.831958135600568e-06,
      "loss": 1.0096,
      "num_input_tokens_seen": 27295345,
      "step": 1310
    },
    {
      "epoch": 0.15763842962784824,
      "flos": 12653109719040.0,
      "grad_norm": 3.543337095180272,
      "learning_rate": 3.831645452934032e-06,
      "loss": 1.0038,
      "num_input_tokens_seen": 27313495,
      "step": 1311
    },
    {
      "epoch": 0.15775867251848735,
      "flos": 19209324195840.0,
      "grad_norm": 21.312618132233087,
      "learning_rate": 3.831332492408336e-06,
      "loss": 1.0039,
      "num_input_tokens_seen": 27334625,
      "step": 1312
    },
    {
      "epoch": 0.15787891540912644,
      "flos": 13645620879360.0,
      "grad_norm": 3.138875901261939,
      "learning_rate": 3.831019254070957e-06,
      "loss": 0.8894,
      "num_input_tokens_seen": 27352130,
      "step": 1313
    },
    {
      "epoch": 0.15799915829976552,
      "flos": 19417652674560.0,
      "grad_norm": 5.197902681972775,
      "learning_rate": 3.8307057379694135e-06,
      "loss": 1.1454,
      "num_input_tokens_seen": 27371185,
      "step": 1314
    },
    {
      "epoch": 0.15811940119040463,
      "flos": 14482644480000.0,
      "grad_norm": 7.04284634010373,
      "learning_rate": 3.830391944151264e-06,
      "loss": 1.0199,
      "num_input_tokens_seen": 27386785,
      "step": 1315
    },
    {
      "epoch": 0.1582396440810437,
      "flos": 23247140474880.0,
      "grad_norm": 3.9294650733444976,
      "learning_rate": 3.830077872664114e-06,
      "loss": 0.8753,
      "num_input_tokens_seen": 27407630,
      "step": 1316
    },
    {
      "epoch": 0.1583598869716828,
      "flos": 24110048010240.0,
      "grad_norm": 3.3317581905625833,
      "learning_rate": 3.829763523555604e-06,
      "loss": 0.9429,
      "num_input_tokens_seen": 27427750,
      "step": 1317
    },
    {
      "epoch": 0.15848012986232188,
      "flos": 17556531486720.0,
      "grad_norm": 3.7133250697219187,
      "learning_rate": 3.829448896873423e-06,
      "loss": 0.9846,
      "num_input_tokens_seen": 27446570,
      "step": 1318
    },
    {
      "epoch": 0.158600372752961,
      "flos": 16062261534720.0,
      "grad_norm": 3.2601062923491546,
      "learning_rate": 3.829133992665299e-06,
      "loss": 0.9931,
      "num_input_tokens_seen": 27465415,
      "step": 1319
    },
    {
      "epoch": 0.15872061564360007,
      "flos": 19889725870080.0,
      "grad_norm": 4.09729200113364,
      "learning_rate": 3.828818810979002e-06,
      "loss": 1.08,
      "num_input_tokens_seen": 27483465,
      "step": 1320
    },
    {
      "epoch": 0.15884085853423915,
      "flos": 16849050193920.0,
      "grad_norm": 3.2863625070633993,
      "learning_rate": 3.8285033518623454e-06,
      "loss": 0.999,
      "num_input_tokens_seen": 27503435,
      "step": 1321
    },
    {
      "epoch": 0.15896110142487826,
      "flos": 16428928819200.0,
      "grad_norm": 8.36446207290361,
      "learning_rate": 3.8281876153631845e-06,
      "loss": 1.0071,
      "num_input_tokens_seen": 27519910,
      "step": 1322
    },
    {
      "epoch": 0.15908134431551735,
      "flos": 10372664279040.0,
      "grad_norm": 7.624717081793434,
      "learning_rate": 3.827871601529416e-06,
      "loss": 0.8403,
      "num_input_tokens_seen": 27538150,
      "step": 1323
    },
    {
      "epoch": 0.15920158720615643,
      "flos": 14331295825920.0,
      "grad_norm": 5.172016061201773,
      "learning_rate": 3.827555310408979e-06,
      "loss": 1.0031,
      "num_input_tokens_seen": 27557265,
      "step": 1324
    },
    {
      "epoch": 0.1593218300967955,
      "flos": 17660864348160.0,
      "grad_norm": 3.712516625363921,
      "learning_rate": 3.827238742049854e-06,
      "loss": 1.0202,
      "num_input_tokens_seen": 27577280,
      "step": 1325
    },
    {
      "epoch": 0.15944207298743462,
      "flos": 20177668300800.0,
      "grad_norm": 3.4963273806800568,
      "learning_rate": 3.826921896500066e-06,
      "loss": 0.7341,
      "num_input_tokens_seen": 27598285,
      "step": 1326
    },
    {
      "epoch": 0.1595623158780737,
      "flos": 16322695127040.0,
      "grad_norm": 2.960886449046628,
      "learning_rate": 3.826604773807678e-06,
      "loss": 0.9772,
      "num_input_tokens_seen": 27615980,
      "step": 1327
    },
    {
      "epoch": 0.1596825587687128,
      "flos": 13982939811840.0,
      "grad_norm": 4.857626450130136,
      "learning_rate": 3.826287374020798e-06,
      "loss": 0.9364,
      "num_input_tokens_seen": 27630505,
      "step": 1328
    },
    {
      "epoch": 0.1598028016593519,
      "flos": 16087777566720.0,
      "grad_norm": 3.4859785145082696,
      "learning_rate": 3.825969697187575e-06,
      "loss": 1.0228,
      "num_input_tokens_seen": 27649555,
      "step": 1329
    },
    {
      "epoch": 0.15992304454999098,
      "flos": 14538857840640.0,
      "grad_norm": 8.947423587546332,
      "learning_rate": 3.8256517433562015e-06,
      "loss": 0.9066,
      "num_input_tokens_seen": 27667215,
      "step": 1330
    },
    {
      "epoch": 0.16004328744063007,
      "flos": 12522003824640.0,
      "grad_norm": 5.695454160082373,
      "learning_rate": 3.82533351257491e-06,
      "loss": 1.1226,
      "num_input_tokens_seen": 27684885,
      "step": 1331
    },
    {
      "epoch": 0.16016353033126918,
      "flos": 17137973698560.0,
      "grad_norm": 5.15604617554058,
      "learning_rate": 3.825015004891975e-06,
      "loss": 1.0911,
      "num_input_tokens_seen": 27703345,
      "step": 1332
    },
    {
      "epoch": 0.16028377322190826,
      "flos": 19678914048000.0,
      "grad_norm": 3.2151098315046247,
      "learning_rate": 3.824696220355716e-06,
      "loss": 0.9574,
      "num_input_tokens_seen": 27724655,
      "step": 1333
    },
    {
      "epoch": 0.16040401611254734,
      "flos": 14883136266240.0,
      "grad_norm": 4.121218479673475,
      "learning_rate": 3.824377159014491e-06,
      "loss": 0.9918,
      "num_input_tokens_seen": 27745270,
      "step": 1334
    },
    {
      "epoch": 0.16052425900318643,
      "flos": 15087755059200.0,
      "grad_norm": 5.447462754702155,
      "learning_rate": 3.824057820916702e-06,
      "loss": 1.0551,
      "num_input_tokens_seen": 27762195,
      "step": 1335
    },
    {
      "epoch": 0.16064450189382554,
      "flos": 10975414149120.0,
      "grad_norm": 3.920574808600645,
      "learning_rate": 3.8237382061107904e-06,
      "loss": 0.9186,
      "num_input_tokens_seen": 27778635,
      "step": 1336
    },
    {
      "epoch": 0.16076474478446462,
      "flos": 15039083704320.0,
      "grad_norm": 3.8104104741572637,
      "learning_rate": 3.823418314645243e-06,
      "loss": 0.9842,
      "num_input_tokens_seen": 27797230,
      "step": 1337
    },
    {
      "epoch": 0.1608849876751037,
      "flos": 13017784197120.0,
      "grad_norm": 4.015081346573752,
      "learning_rate": 3.823098146568588e-06,
      "loss": 0.9563,
      "num_input_tokens_seen": 27816655,
      "step": 1338
    },
    {
      "epoch": 0.1610052305657428,
      "flos": 21017420513280.0,
      "grad_norm": 2.933936534619037,
      "learning_rate": 3.822777701929394e-06,
      "loss": 0.9226,
      "num_input_tokens_seen": 27838200,
      "step": 1339
    },
    {
      "epoch": 0.1611254734563819,
      "flos": 19077697105920.0,
      "grad_norm": 4.173392144988248,
      "learning_rate": 3.8224569807762714e-06,
      "loss": 0.9433,
      "num_input_tokens_seen": 27857240,
      "step": 1340
    },
    {
      "epoch": 0.16124571634702098,
      "flos": 15931094323200.0,
      "grad_norm": 4.4152219076755985,
      "learning_rate": 3.822135983157873e-06,
      "loss": 0.9771,
      "num_input_tokens_seen": 27876235,
      "step": 1341
    },
    {
      "epoch": 0.16136595923766006,
      "flos": 7723374796800.0,
      "grad_norm": 7.949705576692983,
      "learning_rate": 3.821814709122896e-06,
      "loss": 1.0454,
      "num_input_tokens_seen": 27894005,
      "step": 1342
    },
    {
      "epoch": 0.16148620212829917,
      "flos": 15065304883200.0,
      "grad_norm": 3.6422304207372904,
      "learning_rate": 3.821493158720076e-06,
      "loss": 1.0558,
      "num_input_tokens_seen": 27912830,
      "step": 1343
    },
    {
      "epoch": 0.16160644501893826,
      "flos": 11861783592960.0,
      "grad_norm": 4.996664460537767,
      "learning_rate": 3.821171331998191e-06,
      "loss": 0.93,
      "num_input_tokens_seen": 27929080,
      "step": 1344
    },
    {
      "epoch": 0.16172668790957734,
      "flos": 46138037575680.0,
      "grad_norm": 0.9418715913707107,
      "learning_rate": 3.820849229006064e-06,
      "loss": 0.7933,
      "num_input_tokens_seen": 27996550,
      "step": 1345
    },
    {
      "epoch": 0.16184693080021645,
      "flos": 16533292339200.0,
      "grad_norm": 4.441451985871045,
      "learning_rate": 3.8205268497925564e-06,
      "loss": 0.9127,
      "num_input_tokens_seen": 28016740,
      "step": 1346
    },
    {
      "epoch": 0.16196717369085553,
      "flos": 12359495454720.0,
      "grad_norm": 3.7667329669845024,
      "learning_rate": 3.8202041944065725e-06,
      "loss": 0.9962,
      "num_input_tokens_seen": 28032280,
      "step": 1347
    },
    {
      "epoch": 0.16208741658149461,
      "flos": 16975158743040.0,
      "grad_norm": 4.739882686788934,
      "learning_rate": 3.819881262897061e-06,
      "loss": 0.9455,
      "num_input_tokens_seen": 28050135,
      "step": 1348
    },
    {
      "epoch": 0.1622076594721337,
      "flos": 18421370511360.0,
      "grad_norm": 3.3625738483246748,
      "learning_rate": 3.819558055313008e-06,
      "loss": 0.9435,
      "num_input_tokens_seen": 28070540,
      "step": 1349
    },
    {
      "epoch": 0.1623279023627728,
      "flos": 15298229637120.0,
      "grad_norm": 3.4747348052999794,
      "learning_rate": 3.819234571703444e-06,
      "loss": 0.9767,
      "num_input_tokens_seen": 28089085,
      "step": 1350
    },
    {
      "epoch": 0.1624481452534119,
      "flos": 15690167685120.0,
      "grad_norm": 3.0919521772330807,
      "learning_rate": 3.8189108121174435e-06,
      "loss": 1.0537,
      "num_input_tokens_seen": 28108570,
      "step": 1351
    },
    {
      "epoch": 0.16256838814405097,
      "flos": 19285535047680.0,
      "grad_norm": 2.800522626058383,
      "learning_rate": 3.818586776604118e-06,
      "loss": 1.0382,
      "num_input_tokens_seen": 28128930,
      "step": 1352
    },
    {
      "epoch": 0.16268863103469008,
      "flos": 14278608199680.0,
      "grad_norm": 4.4033966565446,
      "learning_rate": 3.818262465212625e-06,
      "loss": 0.838,
      "num_input_tokens_seen": 28148775,
      "step": 1353
    },
    {
      "epoch": 0.16280887392532917,
      "flos": 12993954385920.0,
      "grad_norm": 3.4035436395520846,
      "learning_rate": 3.817937877992161e-06,
      "loss": 0.9763,
      "num_input_tokens_seen": 28165790,
      "step": 1354
    },
    {
      "epoch": 0.16292911681596825,
      "flos": 8352315187200.0,
      "grad_norm": 17.188205773861604,
      "learning_rate": 3.817613014991967e-06,
      "loss": 1.0596,
      "num_input_tokens_seen": 28181650,
      "step": 1355
    },
    {
      "epoch": 0.16304935970660733,
      "flos": 18578482974720.0,
      "grad_norm": 2.927849846841018,
      "learning_rate": 3.817287876261323e-06,
      "loss": 0.9601,
      "num_input_tokens_seen": 28201705,
      "step": 1356
    },
    {
      "epoch": 0.16316960259724644,
      "flos": 20913271603200.0,
      "grad_norm": 4.785559214538005,
      "learning_rate": 3.816962461849553e-06,
      "loss": 1.0026,
      "num_input_tokens_seen": 28223295,
      "step": 1357
    },
    {
      "epoch": 0.16328984548788553,
      "flos": 14829436907520.0,
      "grad_norm": 3.880913436966175,
      "learning_rate": 3.8166367718060235e-06,
      "loss": 1.0514,
      "num_input_tokens_seen": 28242905,
      "step": 1358
    },
    {
      "epoch": 0.1634100883785246,
      "flos": 12784338247680.0,
      "grad_norm": 3.755281670745317,
      "learning_rate": 3.816310806180139e-06,
      "loss": 0.9467,
      "num_input_tokens_seen": 28261035,
      "step": 1359
    },
    {
      "epoch": 0.16353033126916372,
      "flos": 17478082560000.0,
      "grad_norm": 2.481239115117025,
      "learning_rate": 3.81598456502135e-06,
      "loss": 1.0063,
      "num_input_tokens_seen": 28280775,
      "step": 1360
    },
    {
      "epoch": 0.1636505741598028,
      "flos": 14113984389120.0,
      "grad_norm": 3.9408851483211613,
      "learning_rate": 3.8156580483791455e-06,
      "loss": 1.069,
      "num_input_tokens_seen": 28295685,
      "step": 1361
    },
    {
      "epoch": 0.16377081705044189,
      "flos": 20230907781120.0,
      "grad_norm": 4.085773647079096,
      "learning_rate": 3.815331256303059e-06,
      "loss": 0.9697,
      "num_input_tokens_seen": 28315435,
      "step": 1362
    },
    {
      "epoch": 0.163891059941081,
      "flos": 15563384647680.0,
      "grad_norm": 7.751533848733207,
      "learning_rate": 3.815004188842665e-06,
      "loss": 0.9548,
      "num_input_tokens_seen": 28333195,
      "step": 1363
    },
    {
      "epoch": 0.16401130283172008,
      "flos": 19077482496000.0,
      "grad_norm": 2.6415787278068046,
      "learning_rate": 3.814676846047578e-06,
      "loss": 0.9973,
      "num_input_tokens_seen": 28353790,
      "step": 1364
    },
    {
      "epoch": 0.16413154572235916,
      "flos": 23532783513600.0,
      "grad_norm": 3.5232317879250172,
      "learning_rate": 3.8143492279674565e-06,
      "loss": 0.9057,
      "num_input_tokens_seen": 28376205,
      "step": 1365
    },
    {
      "epoch": 0.16425178861299825,
      "flos": 28652190289920.0,
      "grad_norm": 0.9703193996018282,
      "learning_rate": 3.8140213346519997e-06,
      "loss": 0.8537,
      "num_input_tokens_seen": 28426520,
      "step": 1366
    },
    {
      "epoch": 0.16437203150363736,
      "flos": 18107145584640.0,
      "grad_norm": 2.797727316449377,
      "learning_rate": 3.813693166150948e-06,
      "loss": 0.9655,
      "num_input_tokens_seen": 28446450,
      "step": 1367
    },
    {
      "epoch": 0.16449227439427644,
      "flos": 16690864680960.0,
      "grad_norm": 4.072936517353253,
      "learning_rate": 3.813364722514086e-06,
      "loss": 1.0488,
      "num_input_tokens_seen": 28464505,
      "step": 1368
    },
    {
      "epoch": 0.16461251728491552,
      "flos": 9552143093760.0,
      "grad_norm": 3.750794135883796,
      "learning_rate": 3.8130360037912368e-06,
      "loss": 1.0104,
      "num_input_tokens_seen": 28480670,
      "step": 1369
    },
    {
      "epoch": 0.16473276017555463,
      "flos": 16350602526720.0,
      "grad_norm": 10.392138561637703,
      "learning_rate": 3.812707010032268e-06,
      "loss": 1.02,
      "num_input_tokens_seen": 28499445,
      "step": 1370
    },
    {
      "epoch": 0.16485300306619372,
      "flos": 17635103047680.0,
      "grad_norm": 4.120610068624121,
      "learning_rate": 3.8123777412870863e-06,
      "loss": 0.994,
      "num_input_tokens_seen": 28518665,
      "step": 1371
    },
    {
      "epoch": 0.1649732459568328,
      "flos": 14987469127680.0,
      "grad_norm": 3.794506700140627,
      "learning_rate": 3.812048197605643e-06,
      "loss": 0.9969,
      "num_input_tokens_seen": 28537280,
      "step": 1372
    },
    {
      "epoch": 0.16509348884747188,
      "flos": 14383860817920.0,
      "grad_norm": 3.251798425609044,
      "learning_rate": 3.8117183790379277e-06,
      "loss": 1.0017,
      "num_input_tokens_seen": 28555450,
      "step": 1373
    },
    {
      "epoch": 0.165213731738111,
      "flos": 7749504000000.0,
      "grad_norm": 3.960864184903032,
      "learning_rate": 3.811388285633976e-06,
      "loss": 1.1326,
      "num_input_tokens_seen": 28571155,
      "step": 1374
    },
    {
      "epoch": 0.16533397462875007,
      "flos": 21358909009920.0,
      "grad_norm": 4.297723415885234,
      "learning_rate": 3.811057917443861e-06,
      "loss": 0.8202,
      "num_input_tokens_seen": 28590140,
      "step": 1375
    },
    {
      "epoch": 0.16545421751938916,
      "flos": 46939389050880.0,
      "grad_norm": 0.8824791067746237,
      "learning_rate": 3.8107272745177e-06,
      "loss": 0.903,
      "num_input_tokens_seen": 28662190,
      "step": 1376
    },
    {
      "epoch": 0.16557446041002827,
      "flos": 15983383388160.0,
      "grad_norm": 3.7231013381203546,
      "learning_rate": 3.8103963569056513e-06,
      "loss": 0.9909,
      "num_input_tokens_seen": 28681045,
      "step": 1377
    },
    {
      "epoch": 0.16569470330066735,
      "flos": 17499275735040.0,
      "grad_norm": 2.4911780151440817,
      "learning_rate": 3.8100651646579146e-06,
      "loss": 1.0834,
      "num_input_tokens_seen": 28699975,
      "step": 1378
    },
    {
      "epoch": 0.16581494619130643,
      "flos": 10602339225600.0,
      "grad_norm": 6.374532944268808,
      "learning_rate": 3.8097336978247317e-06,
      "loss": 1.1257,
      "num_input_tokens_seen": 28716400,
      "step": 1379
    },
    {
      "epoch": 0.16593518908194552,
      "flos": 12338148986880.0,
      "grad_norm": 7.814993742624854,
      "learning_rate": 3.8094019564563854e-06,
      "loss": 1.0862,
      "num_input_tokens_seen": 28733050,
      "step": 1380
    },
    {
      "epoch": 0.16605543197258463,
      "flos": 14488714874880.0,
      "grad_norm": 4.539524284680188,
      "learning_rate": 3.809069940603201e-06,
      "loss": 0.9719,
      "num_input_tokens_seen": 28750725,
      "step": 1381
    },
    {
      "epoch": 0.1661756748632237,
      "flos": 10030777221120.0,
      "grad_norm": 3.6576107785911303,
      "learning_rate": 3.8087376503155452e-06,
      "loss": 0.9749,
      "num_input_tokens_seen": 28767930,
      "step": 1382
    },
    {
      "epoch": 0.1662959177538628,
      "flos": 47312931962880.0,
      "grad_norm": 1.032572447426458,
      "learning_rate": 3.808405085643826e-06,
      "loss": 0.8081,
      "num_input_tokens_seen": 28832530,
      "step": 1383
    },
    {
      "epoch": 0.1664161606445019,
      "flos": 14724337582080.0,
      "grad_norm": 5.457195347781017,
      "learning_rate": 3.8080722466384925e-06,
      "loss": 1.0993,
      "num_input_tokens_seen": 28850100,
      "step": 1384
    },
    {
      "epoch": 0.166536403535141,
      "flos": 17972513955840.0,
      "grad_norm": 3.510817296440995,
      "learning_rate": 3.8077391333500376e-06,
      "loss": 0.9034,
      "num_input_tokens_seen": 28868960,
      "step": 1385
    },
    {
      "epoch": 0.16665664642578007,
      "flos": 18107114926080.0,
      "grad_norm": 3.1561377477041317,
      "learning_rate": 3.8074057458289934e-06,
      "loss": 0.9629,
      "num_input_tokens_seen": 28889370,
      "step": 1386
    },
    {
      "epoch": 0.16677688931641918,
      "flos": 15773859225600.0,
      "grad_norm": 6.009330170891196,
      "learning_rate": 3.807072084125934e-06,
      "loss": 1.0238,
      "num_input_tokens_seen": 28910940,
      "step": 1387
    },
    {
      "epoch": 0.16689713220705826,
      "flos": 11997212344320.0,
      "grad_norm": 3.7169740427181788,
      "learning_rate": 3.806738148291477e-06,
      "loss": 0.9903,
      "num_input_tokens_seen": 28927485,
      "step": 1388
    },
    {
      "epoch": 0.16701737509769735,
      "flos": 25866315141120.0,
      "grad_norm": 3.830664216853933,
      "learning_rate": 3.8064039383762793e-06,
      "loss": 0.919,
      "num_input_tokens_seen": 28949570,
      "step": 1389
    },
    {
      "epoch": 0.16713761798833643,
      "flos": 16534120120320.0,
      "grad_norm": 4.101700844021023,
      "learning_rate": 3.8060694544310396e-06,
      "loss": 0.9722,
      "num_input_tokens_seen": 28967800,
      "step": 1390
    },
    {
      "epoch": 0.16725786087897554,
      "flos": 18003088650240.0,
      "grad_norm": 2.686276623153292,
      "learning_rate": 3.8057346965065006e-06,
      "loss": 0.9889,
      "num_input_tokens_seen": 28988750,
      "step": 1391
    },
    {
      "epoch": 0.16737810376961462,
      "flos": 22695116083200.0,
      "grad_norm": 2.3401637805777566,
      "learning_rate": 3.805399664653443e-06,
      "loss": 1.0555,
      "num_input_tokens_seen": 29010610,
      "step": 1392
    },
    {
      "epoch": 0.1674983466602537,
      "flos": 19913739632640.0,
      "grad_norm": 3.816576419968295,
      "learning_rate": 3.805064358922692e-06,
      "loss": 0.9503,
      "num_input_tokens_seen": 29028620,
      "step": 1393
    },
    {
      "epoch": 0.16761858955089282,
      "flos": 15458959810560.0,
      "grad_norm": 2.7889781265960303,
      "learning_rate": 3.8047287793651136e-06,
      "loss": 0.9951,
      "num_input_tokens_seen": 29049785,
      "step": 1394
    },
    {
      "epoch": 0.1677388324415319,
      "flos": 16927069900800.0,
      "grad_norm": 3.1081425510250775,
      "learning_rate": 3.8043929260316137e-06,
      "loss": 1.0813,
      "num_input_tokens_seen": 29067660,
      "step": 1395
    },
    {
      "epoch": 0.16785907533217098,
      "flos": 14593844858880.0,
      "grad_norm": 3.254784656673298,
      "learning_rate": 3.8040567989731417e-06,
      "loss": 1.0346,
      "num_input_tokens_seen": 29085325,
      "step": 1396
    },
    {
      "epoch": 0.16797931822281006,
      "flos": 11080053596160.0,
      "grad_norm": 3.0911310072909948,
      "learning_rate": 3.8037203982406876e-06,
      "loss": 1.0133,
      "num_input_tokens_seen": 29103210,
      "step": 1397
    },
    {
      "epoch": 0.16809956111344918,
      "flos": 11705774837760.0,
      "grad_norm": 3.0625625915359174,
      "learning_rate": 3.8033837238852835e-06,
      "loss": 0.9319,
      "num_input_tokens_seen": 29119630,
      "step": 1398
    },
    {
      "epoch": 0.16821980400408826,
      "flos": 16533629583360.0,
      "grad_norm": 2.764249556591174,
      "learning_rate": 3.8030467759580017e-06,
      "loss": 0.8904,
      "num_input_tokens_seen": 29140270,
      "step": 1399
    },
    {
      "epoch": 0.16834004689472734,
      "flos": 14748596613120.0,
      "grad_norm": 4.266765064853272,
      "learning_rate": 3.802709554509958e-06,
      "loss": 1.0754,
      "num_input_tokens_seen": 29157790,
      "step": 1400
    },
    {
      "epoch": 0.16846028978536645,
      "flos": 18998083153920.0,
      "grad_norm": 3.1750622600548497,
      "learning_rate": 3.8023720595923083e-06,
      "loss": 0.9869,
      "num_input_tokens_seen": 29176765,
      "step": 1401
    },
    {
      "epoch": 0.16858053267600553,
      "flos": 13360805621760.0,
      "grad_norm": 3.771112645941368,
      "learning_rate": 3.80203429125625e-06,
      "loss": 1.0755,
      "num_input_tokens_seen": 29194660,
      "step": 1402
    },
    {
      "epoch": 0.16870077556664462,
      "flos": 19757792194560.0,
      "grad_norm": 3.7391635726486294,
      "learning_rate": 3.8016962495530225e-06,
      "loss": 0.9008,
      "num_input_tokens_seen": 29213570,
      "step": 1403
    },
    {
      "epoch": 0.1688210184572837,
      "flos": 9685916282880.0,
      "grad_norm": 5.599458136909495,
      "learning_rate": 3.8013579345339063e-06,
      "loss": 0.9611,
      "num_input_tokens_seen": 29228155,
      "step": 1404
    },
    {
      "epoch": 0.1689412613479228,
      "flos": 18841093324800.0,
      "grad_norm": 5.550168888855925,
      "learning_rate": 3.801019346250224e-06,
      "loss": 0.8944,
      "num_input_tokens_seen": 29248020,
      "step": 1405
    },
    {
      "epoch": 0.1690615042385619,
      "flos": 15010164572160.0,
      "grad_norm": 3.2483560365679773,
      "learning_rate": 3.8006804847533395e-06,
      "loss": 1.0269,
      "num_input_tokens_seen": 29267255,
      "step": 1406
    },
    {
      "epoch": 0.16918174712920098,
      "flos": 14802418606080.0,
      "grad_norm": 3.4310273701024436,
      "learning_rate": 3.8003413500946556e-06,
      "loss": 1.0542,
      "num_input_tokens_seen": 29287085,
      "step": 1407
    },
    {
      "epoch": 0.1693019900198401,
      "flos": 12024138670080.0,
      "grad_norm": 6.55686927103562,
      "learning_rate": 3.8000019423256216e-06,
      "loss": 1.0206,
      "num_input_tokens_seen": 29304570,
      "step": 1408
    },
    {
      "epoch": 0.16942223291047917,
      "flos": 19076869324800.0,
      "grad_norm": 2.6968556285175693,
      "learning_rate": 3.7996622614977234e-06,
      "loss": 1.0784,
      "num_input_tokens_seen": 29325480,
      "step": 1409
    },
    {
      "epoch": 0.16954247580111825,
      "flos": 13173670318080.0,
      "grad_norm": 4.1587539816843915,
      "learning_rate": 3.799322307662492e-06,
      "loss": 0.998,
      "num_input_tokens_seen": 29343020,
      "step": 1410
    },
    {
      "epoch": 0.16966271869175734,
      "flos": 9867318435840.0,
      "grad_norm": 6.849375742290828,
      "learning_rate": 3.798982080871496e-06,
      "loss": 1.0461,
      "num_input_tokens_seen": 29357880,
      "step": 1411
    },
    {
      "epoch": 0.16978296158239645,
      "flos": 26759889346560.0,
      "grad_norm": 4.738538868718032,
      "learning_rate": 3.798641581176349e-06,
      "loss": 0.8807,
      "num_input_tokens_seen": 29379880,
      "step": 1412
    },
    {
      "epoch": 0.16990320447303553,
      "flos": 20177882910720.0,
      "grad_norm": 3.5063296711891656,
      "learning_rate": 3.7983008086287044e-06,
      "loss": 0.9416,
      "num_input_tokens_seen": 29400920,
      "step": 1413
    },
    {
      "epoch": 0.1700234473636746,
      "flos": 14326666383360.0,
      "grad_norm": 3.413258193748663,
      "learning_rate": 3.797959763280257e-06,
      "loss": 0.9882,
      "num_input_tokens_seen": 29419325,
      "step": 1414
    },
    {
      "epoch": 0.17014369025431372,
      "flos": 17683283865600.0,
      "grad_norm": 3.9805571571542684,
      "learning_rate": 3.797618445182743e-06,
      "loss": 0.9932,
      "num_input_tokens_seen": 29440440,
      "step": 1415
    },
    {
      "epoch": 0.1702639331449528,
      "flos": 11652566016000.0,
      "grad_norm": 6.2596346027629615,
      "learning_rate": 3.79727685438794e-06,
      "loss": 1.0422,
      "num_input_tokens_seen": 29454350,
      "step": 1416
    },
    {
      "epoch": 0.1703841760355919,
      "flos": 37559082639360.0,
      "grad_norm": 0.9092135151635208,
      "learning_rate": 3.796934990947667e-06,
      "loss": 0.8335,
      "num_input_tokens_seen": 29515755,
      "step": 1417
    },
    {
      "epoch": 0.170504418926231,
      "flos": 35303632035840.0,
      "grad_norm": 0.930024478651728,
      "learning_rate": 3.7965928549137854e-06,
      "loss": 0.8543,
      "num_input_tokens_seen": 29572290,
      "step": 1418
    },
    {
      "epoch": 0.17062466181687008,
      "flos": 18395915796480.0,
      "grad_norm": 4.864506309057413,
      "learning_rate": 3.7962504463381953e-06,
      "loss": 0.9479,
      "num_input_tokens_seen": 29593500,
      "step": 1419
    },
    {
      "epoch": 0.17074490470750917,
      "flos": 14882461777920.0,
      "grad_norm": 2.500766255251242,
      "learning_rate": 3.7959077652728412e-06,
      "loss": 0.9929,
      "num_input_tokens_seen": 29611675,
      "step": 1420
    },
    {
      "epoch": 0.17086514759814825,
      "flos": 14881909923840.0,
      "grad_norm": 9.140640858693509,
      "learning_rate": 3.795564811769707e-06,
      "loss": 0.9729,
      "num_input_tokens_seen": 29629750,
      "step": 1421
    },
    {
      "epoch": 0.17098539048878736,
      "flos": 20282062479360.0,
      "grad_norm": 3.229319874400226,
      "learning_rate": 3.795221585880818e-06,
      "loss": 0.9867,
      "num_input_tokens_seen": 29650150,
      "step": 1422
    },
    {
      "epoch": 0.17110563337942644,
      "flos": 11525905612800.0,
      "grad_norm": 11.138458683355625,
      "learning_rate": 3.794878087658242e-06,
      "loss": 1.1146,
      "num_input_tokens_seen": 29667640,
      "step": 1423
    },
    {
      "epoch": 0.17122587627006552,
      "flos": 21144050257920.0,
      "grad_norm": 2.688664915759463,
      "learning_rate": 3.7945343171540873e-06,
      "loss": 0.9839,
      "num_input_tokens_seen": 29688235,
      "step": 1424
    },
    {
      "epoch": 0.17134611916070464,
      "flos": 18028788633600.0,
      "grad_norm": 4.230998026390039,
      "learning_rate": 3.7941902744205033e-06,
      "loss": 0.9891,
      "num_input_tokens_seen": 29708990,
      "step": 1425
    },
    {
      "epoch": 0.17146636205134372,
      "flos": 9846799749120.0,
      "grad_norm": 6.33018027002206,
      "learning_rate": 3.7938459595096817e-06,
      "loss": 1.0392,
      "num_input_tokens_seen": 29727255,
      "step": 1426
    },
    {
      "epoch": 0.1715866049419828,
      "flos": 17005948047360.0,
      "grad_norm": 3.2762313973816704,
      "learning_rate": 3.7935013724738545e-06,
      "loss": 1.0569,
      "num_input_tokens_seen": 29747475,
      "step": 1427
    },
    {
      "epoch": 0.17170684783262188,
      "flos": 16139085557760.0,
      "grad_norm": 2.719616718916259,
      "learning_rate": 3.7931565133652945e-06,
      "loss": 0.9806,
      "num_input_tokens_seen": 29767270,
      "step": 1428
    },
    {
      "epoch": 0.171827090723261,
      "flos": 18944751697920.0,
      "grad_norm": 3.6589629712562566,
      "learning_rate": 3.792811382236317e-06,
      "loss": 0.8745,
      "num_input_tokens_seen": 29785500,
      "step": 1429
    },
    {
      "epoch": 0.17194733361390008,
      "flos": 20048463237120.0,
      "grad_norm": 3.8650479650050333,
      "learning_rate": 3.792465979139279e-06,
      "loss": 0.9755,
      "num_input_tokens_seen": 29807825,
      "step": 1430
    },
    {
      "epoch": 0.17206757650453916,
      "flos": 46917130936320.0,
      "grad_norm": 1.1087493744238774,
      "learning_rate": 3.792120304126576e-06,
      "loss": 0.9404,
      "num_input_tokens_seen": 29870920,
      "step": 1431
    },
    {
      "epoch": 0.17218781939517827,
      "flos": 15826086973440.0,
      "grad_norm": 3.1804573195684736,
      "learning_rate": 3.791774357250649e-06,
      "loss": 1.0389,
      "num_input_tokens_seen": 29889470,
      "step": 1432
    },
    {
      "epoch": 0.17230806228581735,
      "flos": 9978794741760.0,
      "grad_norm": 3.7341219125960365,
      "learning_rate": 3.7914281385639757e-06,
      "loss": 0.9971,
      "num_input_tokens_seen": 29907065,
      "step": 1433
    },
    {
      "epoch": 0.17242830517645644,
      "flos": 14698760232960.0,
      "grad_norm": 4.068975364432355,
      "learning_rate": 3.7910816481190784e-06,
      "loss": 0.9829,
      "num_input_tokens_seen": 29926600,
      "step": 1434
    },
    {
      "epoch": 0.17254854806709552,
      "flos": 21935652311040.0,
      "grad_norm": 3.929949954541742,
      "learning_rate": 3.7907348859685193e-06,
      "loss": 0.9536,
      "num_input_tokens_seen": 29948025,
      "step": 1435
    },
    {
      "epoch": 0.17266879095773463,
      "flos": 18945978040320.0,
      "grad_norm": 13.533593096064465,
      "learning_rate": 3.790387852164902e-06,
      "loss": 1.003,
      "num_input_tokens_seen": 29968475,
      "step": 1436
    },
    {
      "epoch": 0.1727890338483737,
      "flos": 14383492915200.0,
      "grad_norm": 3.9620846022190896,
      "learning_rate": 3.7900405467608707e-06,
      "loss": 0.9734,
      "num_input_tokens_seen": 29987740,
      "step": 1437
    },
    {
      "epoch": 0.1729092767390128,
      "flos": 12883060592640.0,
      "grad_norm": 12.537635603197067,
      "learning_rate": 3.7896929698091114e-06,
      "loss": 0.9799,
      "num_input_tokens_seen": 30000275,
      "step": 1438
    },
    {
      "epoch": 0.1730295196296519,
      "flos": 19049881681920.0,
      "grad_norm": 5.952622600191658,
      "learning_rate": 3.7893451213623518e-06,
      "loss": 0.9024,
      "num_input_tokens_seen": 30017225,
      "step": 1439
    },
    {
      "epoch": 0.173149762520291,
      "flos": 16953750958080.0,
      "grad_norm": 5.118913711705895,
      "learning_rate": 3.7889970014733606e-06,
      "loss": 1.0282,
      "num_input_tokens_seen": 30036050,
      "step": 1440
    },
    {
      "epoch": 0.17327000541093007,
      "flos": 16612844974080.0,
      "grad_norm": 3.3610178634137062,
      "learning_rate": 3.7886486101949463e-06,
      "loss": 0.9743,
      "num_input_tokens_seen": 30056950,
      "step": 1441
    },
    {
      "epoch": 0.17339024830156918,
      "flos": 12913543311360.0,
      "grad_norm": 4.453816463928476,
      "learning_rate": 3.7882999475799594e-06,
      "loss": 1.0758,
      "num_input_tokens_seen": 30074705,
      "step": 1442
    },
    {
      "epoch": 0.17351049119220827,
      "flos": 16586899722240.0,
      "grad_norm": 2.519995358326588,
      "learning_rate": 3.787951013681293e-06,
      "loss": 1.0187,
      "num_input_tokens_seen": 30092470,
      "step": 1443
    },
    {
      "epoch": 0.17363073408284735,
      "flos": 16925353021440.0,
      "grad_norm": 3.2329335709510225,
      "learning_rate": 3.787601808551879e-06,
      "loss": 0.9865,
      "num_input_tokens_seen": 30112005,
      "step": 1444
    },
    {
      "epoch": 0.17375097697348643,
      "flos": 13125121597440.0,
      "grad_norm": 4.488442732625927,
      "learning_rate": 3.7872523322446926e-06,
      "loss": 1.0281,
      "num_input_tokens_seen": 30130610,
      "step": 1445
    },
    {
      "epoch": 0.17387121986412554,
      "flos": 27759819878400.0,
      "grad_norm": 3.9260104439859327,
      "learning_rate": 3.7869025848127478e-06,
      "loss": 0.7998,
      "num_input_tokens_seen": 30154525,
      "step": 1446
    },
    {
      "epoch": 0.17399146275476463,
      "flos": 14462125793280.0,
      "grad_norm": 4.600447773908425,
      "learning_rate": 3.786552566309102e-06,
      "loss": 0.987,
      "num_input_tokens_seen": 30172455,
      "step": 1447
    },
    {
      "epoch": 0.1741117056454037,
      "flos": 13590541885440.0,
      "grad_norm": 3.274300181623903,
      "learning_rate": 3.7862022767868517e-06,
      "loss": 1.0658,
      "num_input_tokens_seen": 30189765,
      "step": 1448
    },
    {
      "epoch": 0.17423194853604282,
      "flos": 18054519275520.0,
      "grad_norm": 5.949891661611336,
      "learning_rate": 3.7858517162991367e-06,
      "loss": 1.0479,
      "num_input_tokens_seen": 30209560,
      "step": 1449
    },
    {
      "epoch": 0.1743521914266819,
      "flos": 17923413381120.0,
      "grad_norm": 4.621143279531694,
      "learning_rate": 3.7855008848991363e-06,
      "loss": 0.809,
      "num_input_tokens_seen": 30227485,
      "step": 1450
    },
    {
      "epoch": 0.17447243431732098,
      "flos": 18264625950720.0,
      "grad_norm": 3.3630395109026137,
      "learning_rate": 3.7851497826400714e-06,
      "loss": 0.9841,
      "num_input_tokens_seen": 30247345,
      "step": 1451
    },
    {
      "epoch": 0.17459267720796007,
      "flos": 25893425418240.0,
      "grad_norm": 9.552805255883527,
      "learning_rate": 3.7847984095752034e-06,
      "loss": 0.9474,
      "num_input_tokens_seen": 30270520,
      "step": 1452
    },
    {
      "epoch": 0.17471292009859918,
      "flos": 14200128614400.0,
      "grad_norm": 3.2868179309951486,
      "learning_rate": 3.784446765757836e-06,
      "loss": 0.9923,
      "num_input_tokens_seen": 30288885,
      "step": 1453
    },
    {
      "epoch": 0.17483316298923826,
      "flos": 19811706163200.0,
      "grad_norm": 7.412211533798623,
      "learning_rate": 3.7840948512413133e-06,
      "loss": 0.9855,
      "num_input_tokens_seen": 30306190,
      "step": 1454
    },
    {
      "epoch": 0.17495340587987734,
      "flos": 31453265756160.0,
      "grad_norm": 4.395615317676934,
      "learning_rate": 3.7837426660790196e-06,
      "loss": 0.9803,
      "num_input_tokens_seen": 30327325,
      "step": 1455
    },
    {
      "epoch": 0.17507364877051645,
      "flos": 14828639784960.0,
      "grad_norm": 2.7924277468548815,
      "learning_rate": 3.783390210324382e-06,
      "loss": 1.0121,
      "num_input_tokens_seen": 30346770,
      "step": 1456
    },
    {
      "epoch": 0.17519389166115554,
      "flos": 17582568714240.0,
      "grad_norm": 4.541618323579197,
      "learning_rate": 3.7830374840308676e-06,
      "loss": 0.9193,
      "num_input_tokens_seen": 30366645,
      "step": 1457
    },
    {
      "epoch": 0.17531413455179462,
      "flos": 16920784896000.0,
      "grad_norm": 2.890468283864124,
      "learning_rate": 3.7826844872519842e-06,
      "loss": 1.0113,
      "num_input_tokens_seen": 30384220,
      "step": 1458
    },
    {
      "epoch": 0.1754343774424337,
      "flos": 17478297169920.0,
      "grad_norm": 4.6508421760770755,
      "learning_rate": 3.782331220041282e-06,
      "loss": 0.9298,
      "num_input_tokens_seen": 30404005,
      "step": 1459
    },
    {
      "epoch": 0.17555462033307281,
      "flos": 12837240483840.0,
      "grad_norm": 5.319267898609061,
      "learning_rate": 3.7819776824523504e-06,
      "loss": 1.0336,
      "num_input_tokens_seen": 30421590,
      "step": 1460
    },
    {
      "epoch": 0.1756748632237119,
      "flos": 20204594626560.0,
      "grad_norm": 7.631417247492255,
      "learning_rate": 3.7816238745388213e-06,
      "loss": 1.0421,
      "num_input_tokens_seen": 30440855,
      "step": 1461
    },
    {
      "epoch": 0.17579510611435098,
      "flos": 18311427133440.0,
      "grad_norm": 2.8859072565415844,
      "learning_rate": 3.781269796354367e-06,
      "loss": 1.0761,
      "num_input_tokens_seen": 30460195,
      "step": 1462
    },
    {
      "epoch": 0.1759153490049901,
      "flos": 13177410662400.0,
      "grad_norm": 3.4289416499297634,
      "learning_rate": 3.7809154479527006e-06,
      "loss": 1.0697,
      "num_input_tokens_seen": 30479120,
      "step": 1463
    },
    {
      "epoch": 0.17603559189562917,
      "flos": 13199155691520.0,
      "grad_norm": 2.736308433358393,
      "learning_rate": 3.780560829387577e-06,
      "loss": 1.0402,
      "num_input_tokens_seen": 30497340,
      "step": 1464
    },
    {
      "epoch": 0.17615583478626826,
      "flos": 43323603087360.0,
      "grad_norm": 0.874859798643248,
      "learning_rate": 3.7802059407127915e-06,
      "loss": 0.8091,
      "num_input_tokens_seen": 30555610,
      "step": 1465
    },
    {
      "epoch": 0.17627607767690734,
      "flos": 16790108221440.0,
      "grad_norm": 3.159394010794142,
      "learning_rate": 3.7798507819821797e-06,
      "loss": 1.0556,
      "num_input_tokens_seen": 30572455,
      "step": 1466
    },
    {
      "epoch": 0.17639632056754645,
      "flos": 12495322767360.0,
      "grad_norm": 6.873261387939032,
      "learning_rate": 3.7794953532496197e-06,
      "loss": 0.9987,
      "num_input_tokens_seen": 30588080,
      "step": 1467
    },
    {
      "epoch": 0.17651656345818553,
      "flos": 41460235714560.0,
      "grad_norm": 1.970769663690249,
      "learning_rate": 3.7791396545690295e-06,
      "loss": 0.8335,
      "num_input_tokens_seen": 30649035,
      "step": 1468
    },
    {
      "epoch": 0.17663680634882462,
      "flos": 16297455022080.0,
      "grad_norm": 3.1939846721513083,
      "learning_rate": 3.7787836859943685e-06,
      "loss": 0.9974,
      "num_input_tokens_seen": 30667480,
      "step": 1469
    },
    {
      "epoch": 0.17675704923946373,
      "flos": 16087992176640.0,
      "grad_norm": 6.49879062846068,
      "learning_rate": 3.7784274475796363e-06,
      "loss": 0.9897,
      "num_input_tokens_seen": 30685830,
      "step": 1470
    },
    {
      "epoch": 0.1768772921301028,
      "flos": 19313932984320.0,
      "grad_norm": 3.446175049916996,
      "learning_rate": 3.7780709393788745e-06,
      "loss": 0.9709,
      "num_input_tokens_seen": 30706025,
      "step": 1471
    },
    {
      "epoch": 0.1769975350207419,
      "flos": 13597194792960.0,
      "grad_norm": 6.189670991057526,
      "learning_rate": 3.777714161446165e-06,
      "loss": 0.9605,
      "num_input_tokens_seen": 30725450,
      "step": 1472
    },
    {
      "epoch": 0.177117777911381,
      "flos": 25787620945920.0,
      "grad_norm": 4.163592972326784,
      "learning_rate": 3.7773571138356304e-06,
      "loss": 0.8753,
      "num_input_tokens_seen": 30745340,
      "step": 1473
    },
    {
      "epoch": 0.17723802080202009,
      "flos": 15694582517760.0,
      "grad_norm": 5.6669648409914455,
      "learning_rate": 3.776999796601435e-06,
      "loss": 1.0965,
      "num_input_tokens_seen": 30763820,
      "step": 1474
    },
    {
      "epoch": 0.17735826369265917,
      "flos": 21538379673600.0,
      "grad_norm": 2.815284622059437,
      "learning_rate": 3.776642209797783e-06,
      "loss": 0.9142,
      "num_input_tokens_seen": 30785370,
      "step": 1475
    },
    {
      "epoch": 0.17747850658329825,
      "flos": 15196564070400.0,
      "grad_norm": 3.3250903411294503,
      "learning_rate": 3.7762843534789205e-06,
      "loss": 0.9735,
      "num_input_tokens_seen": 30803840,
      "step": 1476
    },
    {
      "epoch": 0.17759874947393736,
      "flos": 12024199987200.0,
      "grad_norm": 4.274233226162426,
      "learning_rate": 3.7759262276991343e-06,
      "loss": 1.074,
      "num_input_tokens_seen": 30821170,
      "step": 1477
    },
    {
      "epoch": 0.17771899236457644,
      "flos": 8116447211520.0,
      "grad_norm": 3.5336290295746022,
      "learning_rate": 3.7755678325127506e-06,
      "loss": 0.999,
      "num_input_tokens_seen": 30838570,
      "step": 1478
    },
    {
      "epoch": 0.17783923525521553,
      "flos": 13335381565440.0,
      "grad_norm": 3.2940708185121106,
      "learning_rate": 3.7752091679741393e-06,
      "loss": 0.9451,
      "num_input_tokens_seen": 30856080,
      "step": 1479
    },
    {
      "epoch": 0.17795947814585464,
      "flos": 21672275496960.0,
      "grad_norm": 7.367796143585118,
      "learning_rate": 3.774850234137708e-06,
      "loss": 0.968,
      "num_input_tokens_seen": 30873095,
      "step": 1480
    },
    {
      "epoch": 0.17807972103649372,
      "flos": 17346424811520.0,
      "grad_norm": 11.135543798726642,
      "learning_rate": 3.7744910310579076e-06,
      "loss": 1.0256,
      "num_input_tokens_seen": 30891740,
      "step": 1481
    },
    {
      "epoch": 0.1781999639271328,
      "flos": 14408671703040.0,
      "grad_norm": 12.107603196990775,
      "learning_rate": 3.774131558789229e-06,
      "loss": 1.0596,
      "num_input_tokens_seen": 30910790,
      "step": 1482
    },
    {
      "epoch": 0.1783202068177719,
      "flos": 11263111311360.0,
      "grad_norm": 4.57644945540972,
      "learning_rate": 3.773771817386203e-06,
      "loss": 0.8969,
      "num_input_tokens_seen": 30927840,
      "step": 1483
    },
    {
      "epoch": 0.178440449708411,
      "flos": 14646225899520.0,
      "grad_norm": 3.355410485433464,
      "learning_rate": 3.773411806903403e-06,
      "loss": 0.9972,
      "num_input_tokens_seen": 30946640,
      "step": 1484
    },
    {
      "epoch": 0.17856069259905008,
      "flos": 15403420938240.0,
      "grad_norm": 3.793892902026991,
      "learning_rate": 3.7730515273954415e-06,
      "loss": 1.1325,
      "num_input_tokens_seen": 30964970,
      "step": 1485
    },
    {
      "epoch": 0.17868093548968916,
      "flos": 18915250053120.0,
      "grad_norm": 3.311183739903211,
      "learning_rate": 3.772690978916973e-06,
      "loss": 1.0375,
      "num_input_tokens_seen": 30984445,
      "step": 1486
    },
    {
      "epoch": 0.17880117838032827,
      "flos": 13171370926080.0,
      "grad_norm": 3.0234289834545747,
      "learning_rate": 3.772330161522693e-06,
      "loss": 1.0663,
      "num_input_tokens_seen": 31002075,
      "step": 1487
    },
    {
      "epoch": 0.17892142127096736,
      "flos": 18893075804160.0,
      "grad_norm": 9.758738196683332,
      "learning_rate": 3.7719690752673365e-06,
      "loss": 1.0047,
      "num_input_tokens_seen": 31022590,
      "step": 1488
    },
    {
      "epoch": 0.17904166416160644,
      "flos": 16974514913280.0,
      "grad_norm": 4.357995693532033,
      "learning_rate": 3.7716077202056796e-06,
      "loss": 0.9779,
      "num_input_tokens_seen": 31040785,
      "step": 1489
    },
    {
      "epoch": 0.17916190705224552,
      "flos": 13569931223040.0,
      "grad_norm": 4.995762555297352,
      "learning_rate": 3.7712460963925404e-06,
      "loss": 1.1332,
      "num_input_tokens_seen": 31056445,
      "step": 1490
    },
    {
      "epoch": 0.17928214994288463,
      "flos": 17894402273280.0,
      "grad_norm": 8.204691070747652,
      "learning_rate": 3.7708842038827775e-06,
      "loss": 0.951,
      "num_input_tokens_seen": 31075125,
      "step": 1491
    },
    {
      "epoch": 0.17940239283352372,
      "flos": 15906191462400.0,
      "grad_norm": 2.378075164794082,
      "learning_rate": 3.770522042731288e-06,
      "loss": 1.0512,
      "num_input_tokens_seen": 31096740,
      "step": 1492
    },
    {
      "epoch": 0.1795226357241628,
      "flos": 16475699343360.0,
      "grad_norm": 2.6684938866250607,
      "learning_rate": 3.7701596129930122e-06,
      "loss": 1.0852,
      "num_input_tokens_seen": 31115185,
      "step": 1493
    },
    {
      "epoch": 0.1796428786148019,
      "flos": 15694429224960.0,
      "grad_norm": 2.9563784162119284,
      "learning_rate": 3.7697969147229315e-06,
      "loss": 0.9264,
      "num_input_tokens_seen": 31133065,
      "step": 1494
    },
    {
      "epoch": 0.179763121505441,
      "flos": 15144121712640.0,
      "grad_norm": 3.563869815285448,
      "learning_rate": 3.7694339479760647e-06,
      "loss": 1.0562,
      "num_input_tokens_seen": 31151815,
      "step": 1495
    },
    {
      "epoch": 0.17988336439608008,
      "flos": 48808673525760.0,
      "grad_norm": 0.8166148453120075,
      "learning_rate": 3.769070712807476e-06,
      "loss": 0.8025,
      "num_input_tokens_seen": 31213565,
      "step": 1496
    },
    {
      "epoch": 0.18000360728671919,
      "flos": 15590035046400.0,
      "grad_norm": 2.6587579892230826,
      "learning_rate": 3.768707209272266e-06,
      "loss": 0.9795,
      "num_input_tokens_seen": 31233415,
      "step": 1497
    },
    {
      "epoch": 0.18012385017735827,
      "flos": 13463206993920.0,
      "grad_norm": 5.2672989445622616,
      "learning_rate": 3.768343437425579e-06,
      "loss": 0.9676,
      "num_input_tokens_seen": 31251705,
      "step": 1498
    },
    {
      "epoch": 0.18024409306799735,
      "flos": 14010080747520.0,
      "grad_norm": 10.868795795574524,
      "learning_rate": 3.7679793973225987e-06,
      "loss": 1.0575,
      "num_input_tokens_seen": 31267235,
      "step": 1499
    },
    {
      "epoch": 0.18036433595863643,
      "flos": 48138818396160.0,
      "grad_norm": 0.8816987498082101,
      "learning_rate": 3.767615089018549e-06,
      "loss": 0.834,
      "num_input_tokens_seen": 31329300,
      "step": 1500
    },
    {
      "epoch": 0.18048457884927555,
      "flos": 12884470886400.0,
      "grad_norm": 3.87345215431879,
      "learning_rate": 3.7672505125686966e-06,
      "loss": 1.0617,
      "num_input_tokens_seen": 31345385,
      "step": 1501
    },
    {
      "epoch": 0.18060482173991463,
      "flos": 11183374725120.0,
      "grad_norm": 5.36105916333547,
      "learning_rate": 3.7668856680283455e-06,
      "loss": 1.0467,
      "num_input_tokens_seen": 31362130,
      "step": 1502
    },
    {
      "epoch": 0.1807250646305537,
      "flos": 13176828149760.0,
      "grad_norm": 3.520205620482081,
      "learning_rate": 3.7665205554528437e-06,
      "loss": 1.0236,
      "num_input_tokens_seen": 31381205,
      "step": 1503
    },
    {
      "epoch": 0.18084530752119282,
      "flos": 16455395266560.0,
      "grad_norm": 3.5804893977489414,
      "learning_rate": 3.7661551748975782e-06,
      "loss": 0.9506,
      "num_input_tokens_seen": 31399100,
      "step": 1504
    },
    {
      "epoch": 0.1809655504118319,
      "flos": 42800957706240.0,
      "grad_norm": 0.8293783312272853,
      "learning_rate": 3.7657895264179772e-06,
      "loss": 0.8114,
      "num_input_tokens_seen": 31454795,
      "step": 1505
    },
    {
      "epoch": 0.181085793302471,
      "flos": 31585720627200.0,
      "grad_norm": 6.492234212659112,
      "learning_rate": 3.765423610069509e-06,
      "loss": 0.9521,
      "num_input_tokens_seen": 31479905,
      "step": 1506
    },
    {
      "epoch": 0.18120603619311007,
      "flos": 24899473305600.0,
      "grad_norm": 105.77809264121045,
      "learning_rate": 3.765057425907683e-06,
      "loss": 0.9298,
      "num_input_tokens_seen": 31501085,
      "step": 1507
    },
    {
      "epoch": 0.18132627908374918,
      "flos": 15274982338560.0,
      "grad_norm": 3.3551833860758222,
      "learning_rate": 3.764690973988048e-06,
      "loss": 0.9885,
      "num_input_tokens_seen": 31521145,
      "step": 1508
    },
    {
      "epoch": 0.18144652197438826,
      "flos": 20701478707200.0,
      "grad_norm": 2.79683229211348,
      "learning_rate": 3.7643242543661967e-06,
      "loss": 0.9453,
      "num_input_tokens_seen": 31543525,
      "step": 1509
    },
    {
      "epoch": 0.18156676486502735,
      "flos": 49177824153600.0,
      "grad_norm": 1.2402123637010827,
      "learning_rate": 3.7639572670977573e-06,
      "loss": 0.841,
      "num_input_tokens_seen": 31598740,
      "step": 1510
    },
    {
      "epoch": 0.18168700775566646,
      "flos": 18843423375360.0,
      "grad_norm": 2.329924125090862,
      "learning_rate": 3.7635900122384042e-06,
      "loss": 0.9739,
      "num_input_tokens_seen": 31621455,
      "step": 1511
    },
    {
      "epoch": 0.18180725064630554,
      "flos": 10601879347200.0,
      "grad_norm": 2.9753928301816694,
      "learning_rate": 3.7632224898438477e-06,
      "loss": 1.062,
      "num_input_tokens_seen": 31637650,
      "step": 1512
    },
    {
      "epoch": 0.18192749353694462,
      "flos": 13963831418880.0,
      "grad_norm": 4.9480734495788115,
      "learning_rate": 3.762854699969842e-06,
      "loss": 0.9959,
      "num_input_tokens_seen": 31657880,
      "step": 1513
    },
    {
      "epoch": 0.1820477364275837,
      "flos": 14696951377920.0,
      "grad_norm": 7.730624435884929,
      "learning_rate": 3.762486642672179e-06,
      "loss": 0.9201,
      "num_input_tokens_seen": 31674540,
      "step": 1514
    },
    {
      "epoch": 0.18216797931822282,
      "flos": 12128594165760.0,
      "grad_norm": 2.6882071340998253,
      "learning_rate": 3.7621183180066946e-06,
      "loss": 1.058,
      "num_input_tokens_seen": 31692220,
      "step": 1515
    },
    {
      "epoch": 0.1822882222088619,
      "flos": 20834669383680.0,
      "grad_norm": 2.2658215667692203,
      "learning_rate": 3.7617497260292625e-06,
      "loss": 0.9382,
      "num_input_tokens_seen": 31713995,
      "step": 1516
    },
    {
      "epoch": 0.18240846509950098,
      "flos": 12539824558080.0,
      "grad_norm": 4.369849524798854,
      "learning_rate": 3.7613808667957967e-06,
      "loss": 0.9964,
      "num_input_tokens_seen": 31726405,
      "step": 1517
    },
    {
      "epoch": 0.1825287079901401,
      "flos": 10447342202880.0,
      "grad_norm": 2.8378832318812854,
      "learning_rate": 3.7610117403622547e-06,
      "loss": 1.1173,
      "num_input_tokens_seen": 31742685,
      "step": 1518
    },
    {
      "epoch": 0.18264895088077918,
      "flos": 15591046778880.0,
      "grad_norm": 2.4606695120081974,
      "learning_rate": 3.7606423467846313e-06,
      "loss": 1.0992,
      "num_input_tokens_seen": 31762010,
      "step": 1519
    },
    {
      "epoch": 0.18276919377141826,
      "flos": 14829436907520.0,
      "grad_norm": 2.2935937812680955,
      "learning_rate": 3.760272686118964e-06,
      "loss": 0.9954,
      "num_input_tokens_seen": 31779950,
      "step": 1520
    },
    {
      "epoch": 0.18288943666205737,
      "flos": 15248423915520.0,
      "grad_norm": 4.242975151325497,
      "learning_rate": 3.7599027584213297e-06,
      "loss": 1.1224,
      "num_input_tokens_seen": 31798550,
      "step": 1521
    },
    {
      "epoch": 0.18300967955269645,
      "flos": 15297892392960.0,
      "grad_norm": 2.8460554900989137,
      "learning_rate": 3.7595325637478465e-06,
      "loss": 0.9904,
      "num_input_tokens_seen": 31816295,
      "step": 1522
    },
    {
      "epoch": 0.18312992244333554,
      "flos": 20571200593920.0,
      "grad_norm": 2.5688882609560566,
      "learning_rate": 3.7591621021546723e-06,
      "loss": 1.0252,
      "num_input_tokens_seen": 31838010,
      "step": 1523
    },
    {
      "epoch": 0.18325016533397462,
      "flos": 14278393589760.0,
      "grad_norm": 3.085711275331351,
      "learning_rate": 3.7587913736980062e-06,
      "loss": 1.018,
      "num_input_tokens_seen": 31857370,
      "step": 1524
    },
    {
      "epoch": 0.18337040822461373,
      "flos": 16584477696000.0,
      "grad_norm": 2.572805291569663,
      "learning_rate": 3.7584203784340865e-06,
      "loss": 1.0485,
      "num_input_tokens_seen": 31876260,
      "step": 1525
    },
    {
      "epoch": 0.1834906511152528,
      "flos": 17792399462400.0,
      "grad_norm": 3.3503781518017934,
      "learning_rate": 3.7580491164191938e-06,
      "loss": 1.0551,
      "num_input_tokens_seen": 31894290,
      "step": 1526
    },
    {
      "epoch": 0.1836108940058919,
      "flos": 48154646323200.0,
      "grad_norm": 0.8309338994000179,
      "learning_rate": 3.757677587709648e-06,
      "loss": 0.8295,
      "num_input_tokens_seen": 31957275,
      "step": 1527
    },
    {
      "epoch": 0.183731136896531,
      "flos": 18316363161600.0,
      "grad_norm": 3.0268166753870918,
      "learning_rate": 3.7573057923618095e-06,
      "loss": 0.9491,
      "num_input_tokens_seen": 31977090,
      "step": 1528
    },
    {
      "epoch": 0.1838513797871701,
      "flos": 14515181322240.0,
      "grad_norm": 2.8087053767531582,
      "learning_rate": 3.7569337304320793e-06,
      "loss": 0.937,
      "num_input_tokens_seen": 31996395,
      "step": 1529
    },
    {
      "epoch": 0.18397162267780917,
      "flos": 49098762055680.0,
      "grad_norm": 0.8685037744250111,
      "learning_rate": 3.756561401976899e-06,
      "loss": 0.857,
      "num_input_tokens_seen": 32055820,
      "step": 1530
    },
    {
      "epoch": 0.18409186556844825,
      "flos": 22773871595520.0,
      "grad_norm": 2.3237958763810234,
      "learning_rate": 3.7561888070527514e-06,
      "loss": 1.0322,
      "num_input_tokens_seen": 32077580,
      "step": 1531
    },
    {
      "epoch": 0.18421210845908736,
      "flos": 14278761492480.0,
      "grad_norm": 4.646822198338209,
      "learning_rate": 3.7558159457161577e-06,
      "loss": 0.9884,
      "num_input_tokens_seen": 32095265,
      "step": 1532
    },
    {
      "epoch": 0.18433235134972645,
      "flos": 16427334574080.0,
      "grad_norm": 4.9917372140797545,
      "learning_rate": 3.755442818023681e-06,
      "loss": 0.9745,
      "num_input_tokens_seen": 32114610,
      "step": 1533
    },
    {
      "epoch": 0.18445259424036553,
      "flos": 12963502325760.0,
      "grad_norm": 2.628982248100515,
      "learning_rate": 3.7550694240319246e-06,
      "loss": 0.9648,
      "num_input_tokens_seen": 32132205,
      "step": 1534
    },
    {
      "epoch": 0.18457283713100464,
      "flos": 15143661834240.0,
      "grad_norm": 3.741135355744296,
      "learning_rate": 3.7546957637975326e-06,
      "loss": 0.9594,
      "num_input_tokens_seen": 32149335,
      "step": 1535
    },
    {
      "epoch": 0.18469308002164372,
      "flos": 14462095134720.0,
      "grad_norm": 2.3788900139917004,
      "learning_rate": 3.7543218373771873e-06,
      "loss": 0.9454,
      "num_input_tokens_seen": 32168380,
      "step": 1536
    },
    {
      "epoch": 0.1848133229122828,
      "flos": 18817938001920.0,
      "grad_norm": 1.6389691670830187,
      "learning_rate": 3.753947644827615e-06,
      "loss": 0.9912,
      "num_input_tokens_seen": 32191560,
      "step": 1537
    },
    {
      "epoch": 0.1849335658029219,
      "flos": 50525037649920.0,
      "grad_norm": 1.003069517793837,
      "learning_rate": 3.753573186205579e-06,
      "loss": 0.7984,
      "num_input_tokens_seen": 32259400,
      "step": 1538
    },
    {
      "epoch": 0.185053808693561,
      "flos": 12312173076480.0,
      "grad_norm": 3.9513042614453413,
      "learning_rate": 3.753198461567885e-06,
      "loss": 0.9691,
      "num_input_tokens_seen": 32276365,
      "step": 1539
    },
    {
      "epoch": 0.18517405158420008,
      "flos": 20650262691840.0,
      "grad_norm": 3.7526402397813294,
      "learning_rate": 3.7528234709713783e-06,
      "loss": 1.118,
      "num_input_tokens_seen": 32298830,
      "step": 1540
    },
    {
      "epoch": 0.18529429447483917,
      "flos": 19075213762560.0,
      "grad_norm": 2.4740479034806584,
      "learning_rate": 3.7524482144729447e-06,
      "loss": 1.046,
      "num_input_tokens_seen": 32318005,
      "step": 1541
    },
    {
      "epoch": 0.18541453736547828,
      "flos": 9578272296960.0,
      "grad_norm": 4.040397969275065,
      "learning_rate": 3.7520726921295106e-06,
      "loss": 1.0414,
      "num_input_tokens_seen": 32334445,
      "step": 1542
    },
    {
      "epoch": 0.18553478025611736,
      "flos": 17085255413760.0,
      "grad_norm": 5.471103880972431,
      "learning_rate": 3.751696903998042e-06,
      "loss": 0.9397,
      "num_input_tokens_seen": 32352800,
      "step": 1543
    },
    {
      "epoch": 0.18565502314675644,
      "flos": 18421677096960.0,
      "grad_norm": 2.484492461108836,
      "learning_rate": 3.7513208501355456e-06,
      "loss": 0.8957,
      "num_input_tokens_seen": 32373625,
      "step": 1544
    },
    {
      "epoch": 0.18577526603739553,
      "flos": 13911940915200.0,
      "grad_norm": 2.7055527914434783,
      "learning_rate": 3.750944530599069e-06,
      "loss": 1.036,
      "num_input_tokens_seen": 32392915,
      "step": 1545
    },
    {
      "epoch": 0.18589550892803464,
      "flos": 13095558635520.0,
      "grad_norm": 2.767391351981055,
      "learning_rate": 3.7505679454456992e-06,
      "loss": 1.0066,
      "num_input_tokens_seen": 32409245,
      "step": 1546
    },
    {
      "epoch": 0.18601575181867372,
      "flos": 16743399014400.0,
      "grad_norm": 3.2566798594535227,
      "learning_rate": 3.750191094732564e-06,
      "loss": 0.8953,
      "num_input_tokens_seen": 32429830,
      "step": 1547
    },
    {
      "epoch": 0.1861359947093128,
      "flos": 18763165593600.0,
      "grad_norm": 3.168747887243887,
      "learning_rate": 3.7498139785168313e-06,
      "loss": 0.9559,
      "num_input_tokens_seen": 32450155,
      "step": 1548
    },
    {
      "epoch": 0.1862562375999519,
      "flos": 16586194575360.0,
      "grad_norm": 2.8359289247997936,
      "learning_rate": 3.749436596855709e-06,
      "loss": 0.983,
      "num_input_tokens_seen": 32469175,
      "step": 1549
    },
    {
      "epoch": 0.186376480490591,
      "flos": 11782445568000.0,
      "grad_norm": 3.65091740301064,
      "learning_rate": 3.749058949806446e-06,
      "loss": 1.1166,
      "num_input_tokens_seen": 32485620,
      "step": 1550
    },
    {
      "epoch": 0.18649672338123008,
      "flos": 15247381524480.0,
      "grad_norm": 2.6324590338175793,
      "learning_rate": 3.748681037426331e-06,
      "loss": 1.0404,
      "num_input_tokens_seen": 32504550,
      "step": 1551
    },
    {
      "epoch": 0.1866169662718692,
      "flos": 8667459870720.0,
      "grad_norm": 4.85533204503854,
      "learning_rate": 3.7483028597726936e-06,
      "loss": 1.1153,
      "num_input_tokens_seen": 32521040,
      "step": 1552
    },
    {
      "epoch": 0.18673720916250827,
      "flos": 16765205360640.0,
      "grad_norm": 5.782571507534178,
      "learning_rate": 3.7479244169029017e-06,
      "loss": 0.8255,
      "num_input_tokens_seen": 32540550,
      "step": 1553
    },
    {
      "epoch": 0.18685745205314735,
      "flos": 13989623377920.0,
      "grad_norm": 3.528708834865774,
      "learning_rate": 3.7475457088743658e-06,
      "loss": 0.9141,
      "num_input_tokens_seen": 32557520,
      "step": 1554
    },
    {
      "epoch": 0.18697769494378644,
      "flos": 24400381808640.0,
      "grad_norm": 6.292847815120558,
      "learning_rate": 3.7471667357445348e-06,
      "loss": 0.9479,
      "num_input_tokens_seen": 32577070,
      "step": 1555
    },
    {
      "epoch": 0.18709793783442555,
      "flos": 24427216158720.0,
      "grad_norm": 2.9586340164897504,
      "learning_rate": 3.7467874975709e-06,
      "loss": 0.9221,
      "num_input_tokens_seen": 32597595,
      "step": 1556
    },
    {
      "epoch": 0.18721818072506463,
      "flos": 29093604925440.0,
      "grad_norm": 3.2369526345149477,
      "learning_rate": 3.7464079944109904e-06,
      "loss": 0.9937,
      "num_input_tokens_seen": 32619175,
      "step": 1557
    },
    {
      "epoch": 0.18733842361570371,
      "flos": 15741751603200.0,
      "grad_norm": 3.2953225770075445,
      "learning_rate": 3.746028226322376e-06,
      "loss": 0.9769,
      "num_input_tokens_seen": 32634775,
      "step": 1558
    },
    {
      "epoch": 0.18745866650634282,
      "flos": 13412113612800.0,
      "grad_norm": 2.9802109993090813,
      "learning_rate": 3.745648193362669e-06,
      "loss": 0.9559,
      "num_input_tokens_seen": 32653850,
      "step": 1559
    },
    {
      "epoch": 0.1875789093969819,
      "flos": 13699136286720.0,
      "grad_norm": 3.724862212634555,
      "learning_rate": 3.745267895589518e-06,
      "loss": 0.9309,
      "num_input_tokens_seen": 32672110,
      "step": 1560
    },
    {
      "epoch": 0.187699152287621,
      "flos": 12049685360640.0,
      "grad_norm": 3.7510792768577916,
      "learning_rate": 3.7448873330606154e-06,
      "loss": 1.0135,
      "num_input_tokens_seen": 32689600,
      "step": 1561
    },
    {
      "epoch": 0.18781939517826007,
      "flos": 15878468014080.0,
      "grad_norm": 5.296484648688134,
      "learning_rate": 3.7445065058336914e-06,
      "loss": 1.0747,
      "num_input_tokens_seen": 32708190,
      "step": 1562
    },
    {
      "epoch": 0.18793963806889918,
      "flos": 10319547432960.0,
      "grad_norm": 3.95723525761285,
      "learning_rate": 3.7441254139665176e-06,
      "loss": 1.0615,
      "num_input_tokens_seen": 32724095,
      "step": 1563
    },
    {
      "epoch": 0.18805988095953827,
      "flos": 12364584775680.0,
      "grad_norm": 4.209213215211755,
      "learning_rate": 3.743744057516905e-06,
      "loss": 1.0249,
      "num_input_tokens_seen": 32741875,
      "step": 1564
    },
    {
      "epoch": 0.18818012385017735,
      "flos": 10629572136960.0,
      "grad_norm": 6.447010607270913,
      "learning_rate": 3.743362436542706e-06,
      "loss": 1.0773,
      "num_input_tokens_seen": 32756285,
      "step": 1565
    },
    {
      "epoch": 0.18830036674081646,
      "flos": 33992427909120.0,
      "grad_norm": 3.537973265784627,
      "learning_rate": 3.7429805511018115e-06,
      "loss": 0.9683,
      "num_input_tokens_seen": 32777665,
      "step": 1566
    },
    {
      "epoch": 0.18842060963145554,
      "flos": 21411014123520.0,
      "grad_norm": 3.2192031393240605,
      "learning_rate": 3.7425984012521524e-06,
      "loss": 0.9796,
      "num_input_tokens_seen": 32797585,
      "step": 1567
    },
    {
      "epoch": 0.18854085252209463,
      "flos": 50359034204160.0,
      "grad_norm": 0.7945478529494154,
      "learning_rate": 3.7422159870517025e-06,
      "loss": 0.8234,
      "num_input_tokens_seen": 32862560,
      "step": 1568
    },
    {
      "epoch": 0.1886610954127337,
      "flos": 15118912266240.0,
      "grad_norm": 2.5131692213348265,
      "learning_rate": 3.7418333085584717e-06,
      "loss": 0.9894,
      "num_input_tokens_seen": 32883465,
      "step": 1569
    },
    {
      "epoch": 0.18878133830337282,
      "flos": 12226795315200.0,
      "grad_norm": 3.1709924231047437,
      "learning_rate": 3.7414503658305128e-06,
      "loss": 1.1211,
      "num_input_tokens_seen": 32900420,
      "step": 1570
    },
    {
      "epoch": 0.1889015811940119,
      "flos": 18342584340480.0,
      "grad_norm": 3.122627578462474,
      "learning_rate": 3.7410671589259185e-06,
      "loss": 0.9752,
      "num_input_tokens_seen": 32918740,
      "step": 1571
    },
    {
      "epoch": 0.18902182408465099,
      "flos": 14933892403200.0,
      "grad_norm": 3.2265248146350225,
      "learning_rate": 3.7406836879028205e-06,
      "loss": 0.9863,
      "num_input_tokens_seen": 32938685,
      "step": 1572
    },
    {
      "epoch": 0.1891420669752901,
      "flos": 15824921948160.0,
      "grad_norm": 2.8221952541778283,
      "learning_rate": 3.7402999528193907e-06,
      "loss": 0.9725,
      "num_input_tokens_seen": 32957905,
      "step": 1573
    },
    {
      "epoch": 0.18926230986592918,
      "flos": 15642293452800.0,
      "grad_norm": 12.013634247675183,
      "learning_rate": 3.739915953733842e-06,
      "loss": 1.0405,
      "num_input_tokens_seen": 32975670,
      "step": 1574
    },
    {
      "epoch": 0.18938255275656826,
      "flos": 17399419023360.0,
      "grad_norm": 3.7364909669493045,
      "learning_rate": 3.7395316907044264e-06,
      "loss": 1.0188,
      "num_input_tokens_seen": 32996175,
      "step": 1575
    },
    {
      "epoch": 0.18950279564720737,
      "flos": 17373933649920.0,
      "grad_norm": 2.573873910686297,
      "learning_rate": 3.7391471637894364e-06,
      "loss": 0.9967,
      "num_input_tokens_seen": 33018160,
      "step": 1576
    },
    {
      "epoch": 0.18962303853784646,
      "flos": 14017070899200.0,
      "grad_norm": 5.659818587883706,
      "learning_rate": 3.738762373047205e-06,
      "loss": 1.0535,
      "num_input_tokens_seen": 33037800,
      "step": 1577
    },
    {
      "epoch": 0.18974328142848554,
      "flos": 14933769768960.0,
      "grad_norm": 4.240972018116515,
      "learning_rate": 3.738377318536103e-06,
      "loss": 1.033,
      "num_input_tokens_seen": 33057405,
      "step": 1578
    },
    {
      "epoch": 0.18986352431912462,
      "flos": 9136589844480.0,
      "grad_norm": 6.032871166814426,
      "learning_rate": 3.7379920003145447e-06,
      "loss": 0.9141,
      "num_input_tokens_seen": 33071400,
      "step": 1579
    },
    {
      "epoch": 0.18998376720976373,
      "flos": 16899898306560.0,
      "grad_norm": 4.879003397063745,
      "learning_rate": 3.7376064184409817e-06,
      "loss": 1.0468,
      "num_input_tokens_seen": 33090700,
      "step": 1580
    },
    {
      "epoch": 0.19010401010040281,
      "flos": 16323706859520.0,
      "grad_norm": 2.5020354567835272,
      "learning_rate": 3.7372205729739063e-06,
      "loss": 1.0754,
      "num_input_tokens_seen": 33112235,
      "step": 1581
    },
    {
      "epoch": 0.1902242529910419,
      "flos": 13570973614080.0,
      "grad_norm": 4.210088810652229,
      "learning_rate": 3.7368344639718514e-06,
      "loss": 0.9209,
      "num_input_tokens_seen": 33129890,
      "step": 1582
    },
    {
      "epoch": 0.190344495881681,
      "flos": 18133336104960.0,
      "grad_norm": 2.9713815369656276,
      "learning_rate": 3.7364480914933895e-06,
      "loss": 1.0112,
      "num_input_tokens_seen": 33149850,
      "step": 1583
    },
    {
      "epoch": 0.1904647387723201,
      "flos": 19073404907520.0,
      "grad_norm": 2.7191233762224964,
      "learning_rate": 3.7360614555971325e-06,
      "loss": 1.0169,
      "num_input_tokens_seen": 33169225,
      "step": 1584
    },
    {
      "epoch": 0.19058498166295917,
      "flos": 17058022502400.0,
      "grad_norm": 4.112543197451344,
      "learning_rate": 3.735674556341733e-06,
      "loss": 1.0451,
      "num_input_tokens_seen": 33188560,
      "step": 1585
    },
    {
      "epoch": 0.19070522455359826,
      "flos": 20152366878720.0,
      "grad_norm": 3.2188172564840722,
      "learning_rate": 3.7352873937858835e-06,
      "loss": 1.0319,
      "num_input_tokens_seen": 33209815,
      "step": 1586
    },
    {
      "epoch": 0.19082546744423737,
      "flos": 18259199385600.0,
      "grad_norm": 3.0727026568530253,
      "learning_rate": 3.734899967988316e-06,
      "loss": 0.9151,
      "num_input_tokens_seen": 33227715,
      "step": 1587
    },
    {
      "epoch": 0.19094571033487645,
      "flos": 13990573793280.0,
      "grad_norm": 3.124964904588086,
      "learning_rate": 3.7345122790078026e-06,
      "loss": 1.0475,
      "num_input_tokens_seen": 33245000,
      "step": 1588
    },
    {
      "epoch": 0.19106595322551553,
      "flos": 15353737850880.0,
      "grad_norm": 8.595572871300659,
      "learning_rate": 3.7341243269031556e-06,
      "loss": 1.1302,
      "num_input_tokens_seen": 33263710,
      "step": 1589
    },
    {
      "epoch": 0.19118619611615464,
      "flos": 21304596480000.0,
      "grad_norm": 4.6101141902657945,
      "learning_rate": 3.7337361117332275e-06,
      "loss": 0.9744,
      "num_input_tokens_seen": 33285170,
      "step": 1590
    },
    {
      "epoch": 0.19130643900679373,
      "flos": 12232436490240.0,
      "grad_norm": 4.688464459699802,
      "learning_rate": 3.7333476335569087e-06,
      "loss": 0.9629,
      "num_input_tokens_seen": 33302890,
      "step": 1591
    },
    {
      "epoch": 0.1914266818974328,
      "flos": 17660619079680.0,
      "grad_norm": 4.173938940752239,
      "learning_rate": 3.7329588924331325e-06,
      "loss": 0.8633,
      "num_input_tokens_seen": 33323815,
      "step": 1592
    },
    {
      "epoch": 0.1915469247880719,
      "flos": 13439223889920.0,
      "grad_norm": 2.8071244472508723,
      "learning_rate": 3.732569888420871e-06,
      "loss": 1.0182,
      "num_input_tokens_seen": 33343070,
      "step": 1593
    },
    {
      "epoch": 0.191667167678711,
      "flos": 14934536232960.0,
      "grad_norm": 7.425012957115299,
      "learning_rate": 3.732180621579134e-06,
      "loss": 1.0235,
      "num_input_tokens_seen": 33362005,
      "step": 1594
    },
    {
      "epoch": 0.1917874105693501,
      "flos": 24424088985600.0,
      "grad_norm": 2.9980039313903237,
      "learning_rate": 3.7317910919669745e-06,
      "loss": 1.0053,
      "num_input_tokens_seen": 33382920,
      "step": 1595
    },
    {
      "epoch": 0.19190765345998917,
      "flos": 16743950868480.0,
      "grad_norm": 3.6294858077555188,
      "learning_rate": 3.7314012996434826e-06,
      "loss": 0.9575,
      "num_input_tokens_seen": 33401370,
      "step": 1596
    },
    {
      "epoch": 0.19202789635062828,
      "flos": 14091963432960.0,
      "grad_norm": 2.375319710344036,
      "learning_rate": 3.7310112446677907e-06,
      "loss": 0.9999,
      "num_input_tokens_seen": 33419000,
      "step": 1597
    },
    {
      "epoch": 0.19214813924126736,
      "flos": 14855412817920.0,
      "grad_norm": 4.089804109018729,
      "learning_rate": 3.7306209270990695e-06,
      "loss": 0.8832,
      "num_input_tokens_seen": 33436725,
      "step": 1598
    },
    {
      "epoch": 0.19226838213190645,
      "flos": 18762767032320.0,
      "grad_norm": 2.2876323098862708,
      "learning_rate": 3.7302303469965292e-06,
      "loss": 1.065,
      "num_input_tokens_seen": 33455985,
      "step": 1599
    },
    {
      "epoch": 0.19238862502254553,
      "flos": 14803430338560.0,
      "grad_norm": 2.9406644942158064,
      "learning_rate": 3.7298395044194206e-06,
      "loss": 0.904,
      "num_input_tokens_seen": 33474515,
      "step": 1600
    },
    {
      "epoch": 0.19250886791318464,
      "flos": 15432646656000.0,
      "grad_norm": 3.9485588390211945,
      "learning_rate": 3.7294483994270356e-06,
      "loss": 1.1256,
      "num_input_tokens_seen": 33492560,
      "step": 1601
    },
    {
      "epoch": 0.19262911080382372,
      "flos": 16612538388480.0,
      "grad_norm": 3.620341629220655,
      "learning_rate": 3.7290570320787033e-06,
      "loss": 0.9647,
      "num_input_tokens_seen": 33511860,
      "step": 1602
    },
    {
      "epoch": 0.1927493536944628,
      "flos": 15588716728320.0,
      "grad_norm": 3.5329302275932357,
      "learning_rate": 3.728665402433793e-06,
      "loss": 0.913,
      "num_input_tokens_seen": 33530150,
      "step": 1603
    },
    {
      "epoch": 0.19286959658510192,
      "flos": 11709607157760.0,
      "grad_norm": 9.133469504944651,
      "learning_rate": 3.7282735105517164e-06,
      "loss": 1.0529,
      "num_input_tokens_seen": 33547995,
      "step": 1604
    },
    {
      "epoch": 0.192989839475741,
      "flos": 15354289704960.0,
      "grad_norm": 4.755658443522338,
      "learning_rate": 3.727881356491922e-06,
      "loss": 0.8803,
      "num_input_tokens_seen": 33566125,
      "step": 1605
    },
    {
      "epoch": 0.19311008236638008,
      "flos": 13675797012480.0,
      "grad_norm": 3.4897604067081778,
      "learning_rate": 3.7274889403139002e-06,
      "loss": 0.9634,
      "num_input_tokens_seen": 33583470,
      "step": 1606
    },
    {
      "epoch": 0.1932303252570192,
      "flos": 20409642639360.0,
      "grad_norm": 6.868995680159708,
      "learning_rate": 3.727096262077179e-06,
      "loss": 0.9932,
      "num_input_tokens_seen": 33602185,
      "step": 1607
    },
    {
      "epoch": 0.19335056814765827,
      "flos": 13019930296320.0,
      "grad_norm": 3.739511317389386,
      "learning_rate": 3.7267033218413285e-06,
      "loss": 1.0507,
      "num_input_tokens_seen": 33619700,
      "step": 1608
    },
    {
      "epoch": 0.19347081103829736,
      "flos": 9348811960320.0,
      "grad_norm": 3.6026423029222836,
      "learning_rate": 3.726310119665957e-06,
      "loss": 1.0125,
      "num_input_tokens_seen": 33635755,
      "step": 1609
    },
    {
      "epoch": 0.19359105392893644,
      "flos": 14407445360640.0,
      "grad_norm": 3.6153305263500184,
      "learning_rate": 3.725916655610713e-06,
      "loss": 1.0465,
      "num_input_tokens_seen": 33654805,
      "step": 1610
    },
    {
      "epoch": 0.19371129681957555,
      "flos": 14540298792960.0,
      "grad_norm": 5.859028635596038,
      "learning_rate": 3.725522929735284e-06,
      "loss": 0.9442,
      "num_input_tokens_seen": 33671460,
      "step": 1611
    },
    {
      "epoch": 0.19383153971021463,
      "flos": 21699048529920.0,
      "grad_norm": 5.980106434975285,
      "learning_rate": 3.725128942099399e-06,
      "loss": 0.9335,
      "num_input_tokens_seen": 33691580,
      "step": 1612
    },
    {
      "epoch": 0.19395178260085372,
      "flos": 17476212387840.0,
      "grad_norm": 4.248839724862349,
      "learning_rate": 3.7247346927628245e-06,
      "loss": 1.0033,
      "num_input_tokens_seen": 33711235,
      "step": 1613
    },
    {
      "epoch": 0.19407202549149283,
      "flos": 20624838635520.0,
      "grad_norm": 2.7539226991419206,
      "learning_rate": 3.7243401817853694e-06,
      "loss": 0.9844,
      "num_input_tokens_seen": 33731645,
      "step": 1614
    },
    {
      "epoch": 0.1941922683821319,
      "flos": 12758025093120.0,
      "grad_norm": 12.626772312365512,
      "learning_rate": 3.723945409226879e-06,
      "loss": 0.9184,
      "num_input_tokens_seen": 33749855,
      "step": 1615
    },
    {
      "epoch": 0.194312511272771,
      "flos": 6804100608000.0,
      "grad_norm": 3.6582659518930924,
      "learning_rate": 3.723550375147241e-06,
      "loss": 1.0082,
      "num_input_tokens_seen": 33764350,
      "step": 1616
    },
    {
      "epoch": 0.19443275416341008,
      "flos": 19234533642240.0,
      "grad_norm": 2.8514705277334054,
      "learning_rate": 3.7231550796063816e-06,
      "loss": 0.9951,
      "num_input_tokens_seen": 33784080,
      "step": 1617
    },
    {
      "epoch": 0.1945529970540492,
      "flos": 11206499389440.0,
      "grad_norm": 3.7491749173698414,
      "learning_rate": 3.722759522664266e-06,
      "loss": 0.8449,
      "num_input_tokens_seen": 33801100,
      "step": 1618
    },
    {
      "epoch": 0.19467323994468827,
      "flos": 13699442872320.0,
      "grad_norm": 4.596209017125044,
      "learning_rate": 3.7223637043809016e-06,
      "loss": 1.0119,
      "num_input_tokens_seen": 33819800,
      "step": 1619
    },
    {
      "epoch": 0.19479348283532735,
      "flos": 17162508656640.0,
      "grad_norm": 9.751869218107384,
      "learning_rate": 3.7219676248163322e-06,
      "loss": 1.0613,
      "num_input_tokens_seen": 33836685,
      "step": 1620
    },
    {
      "epoch": 0.19491372572596646,
      "flos": 18342952243200.0,
      "grad_norm": 4.776486885288395,
      "learning_rate": 3.721571284030643e-06,
      "loss": 1.1192,
      "num_input_tokens_seen": 33856215,
      "step": 1621
    },
    {
      "epoch": 0.19503396861660555,
      "flos": 13937027727360.0,
      "grad_norm": 4.349233774687201,
      "learning_rate": 3.7211746820839587e-06,
      "loss": 0.9842,
      "num_input_tokens_seen": 33873030,
      "step": 1622
    },
    {
      "epoch": 0.19515421150724463,
      "flos": 14934597550080.0,
      "grad_norm": 3.5496437592476684,
      "learning_rate": 3.7207778190364437e-06,
      "loss": 1.02,
      "num_input_tokens_seen": 33891175,
      "step": 1623
    },
    {
      "epoch": 0.1952744543978837,
      "flos": 23506838261760.0,
      "grad_norm": 4.374149348803121,
      "learning_rate": 3.720380694948302e-06,
      "loss": 0.9454,
      "num_input_tokens_seen": 33913780,
      "step": 1624
    },
    {
      "epoch": 0.19539469728852282,
      "flos": 45850187120640.0,
      "grad_norm": 0.9912334871780768,
      "learning_rate": 3.719983309879777e-06,
      "loss": 0.9422,
      "num_input_tokens_seen": 33973280,
      "step": 1625
    },
    {
      "epoch": 0.1955149401791619,
      "flos": 9427966033920.0,
      "grad_norm": 3.43035284379531,
      "learning_rate": 3.719585663891151e-06,
      "loss": 0.979,
      "num_input_tokens_seen": 33990535,
      "step": 1626
    },
    {
      "epoch": 0.195635183069801,
      "flos": 13277911203840.0,
      "grad_norm": 6.746626682137879,
      "learning_rate": 3.719187757042747e-06,
      "loss": 0.9772,
      "num_input_tokens_seen": 34008075,
      "step": 1627
    },
    {
      "epoch": 0.1957554259604401,
      "flos": 49639495987200.0,
      "grad_norm": 0.8027456241014382,
      "learning_rate": 3.7187895893949275e-06,
      "loss": 0.7942,
      "num_input_tokens_seen": 34074265,
      "step": 1628
    },
    {
      "epoch": 0.19587566885107918,
      "flos": 14960604119040.0,
      "grad_norm": 3.7217759075153514,
      "learning_rate": 3.7183911610080937e-06,
      "loss": 0.9349,
      "num_input_tokens_seen": 34090850,
      "step": 1629
    },
    {
      "epoch": 0.19599591174171827,
      "flos": 15768647270400.0,
      "grad_norm": 4.408919803906652,
      "learning_rate": 3.7179924719426872e-06,
      "loss": 0.9375,
      "num_input_tokens_seen": 34108465,
      "step": 1630
    },
    {
      "epoch": 0.19611615463235738,
      "flos": 16900296867840.0,
      "grad_norm": 7.170178846525525,
      "learning_rate": 3.7175935222591885e-06,
      "loss": 0.9504,
      "num_input_tokens_seen": 34127485,
      "step": 1631
    },
    {
      "epoch": 0.19623639752299646,
      "flos": 20385414266880.0,
      "grad_norm": 3.0104914637230604,
      "learning_rate": 3.717194312018118e-06,
      "loss": 0.9559,
      "num_input_tokens_seen": 34146190,
      "step": 1632
    },
    {
      "epoch": 0.19635664041363554,
      "flos": 14934321623040.0,
      "grad_norm": 5.183229274052756,
      "learning_rate": 3.716794841280036e-06,
      "loss": 0.9454,
      "num_input_tokens_seen": 34164615,
      "step": 1633
    },
    {
      "epoch": 0.19647688330427462,
      "flos": 13386996142080.0,
      "grad_norm": 3.2472996864473527,
      "learning_rate": 3.7163951101055407e-06,
      "loss": 0.97,
      "num_input_tokens_seen": 34182395,
      "step": 1634
    },
    {
      "epoch": 0.19659712619491373,
      "flos": 17241018900480.0,
      "grad_norm": 2.459401841362522,
      "learning_rate": 3.715995118555273e-06,
      "loss": 0.9794,
      "num_input_tokens_seen": 34202090,
      "step": 1635
    },
    {
      "epoch": 0.19671736908555282,
      "flos": 17582415421440.0,
      "grad_norm": 3.40796988927049,
      "learning_rate": 3.71559486668991e-06,
      "loss": 1.0406,
      "num_input_tokens_seen": 34220670,
      "step": 1636
    },
    {
      "epoch": 0.1968376119761919,
      "flos": 16953536348160.0,
      "grad_norm": 2.7820934773968102,
      "learning_rate": 3.715194354570169e-06,
      "loss": 0.9668,
      "num_input_tokens_seen": 34240395,
      "step": 1637
    },
    {
      "epoch": 0.196957854866831,
      "flos": 12836657971200.0,
      "grad_norm": 6.005349000163898,
      "learning_rate": 3.714793582256809e-06,
      "loss": 1.0263,
      "num_input_tokens_seen": 34257180,
      "step": 1638
    },
    {
      "epoch": 0.1970780977574701,
      "flos": 15380234956800.0,
      "grad_norm": 4.09242168745406,
      "learning_rate": 3.7143925498106253e-06,
      "loss": 1.0406,
      "num_input_tokens_seen": 34275440,
      "step": 1639
    },
    {
      "epoch": 0.19719834064810918,
      "flos": 14775308328960.0,
      "grad_norm": 3.1843172526715393,
      "learning_rate": 3.7139912572924558e-06,
      "loss": 0.9951,
      "num_input_tokens_seen": 34294190,
      "step": 1640
    },
    {
      "epoch": 0.19731858353874826,
      "flos": 16659860766720.0,
      "grad_norm": 5.385552626565436,
      "learning_rate": 3.7135897047631744e-06,
      "loss": 1.0137,
      "num_input_tokens_seen": 34311795,
      "step": 1641
    },
    {
      "epoch": 0.19743882642938737,
      "flos": 17058390405120.0,
      "grad_norm": 4.553498886770085,
      "learning_rate": 3.713187892283698e-06,
      "loss": 0.9426,
      "num_input_tokens_seen": 34331125,
      "step": 1642
    },
    {
      "epoch": 0.19755906932002645,
      "flos": 10600744980480.0,
      "grad_norm": 6.426712627220163,
      "learning_rate": 3.71278581991498e-06,
      "loss": 1.057,
      "num_input_tokens_seen": 34346705,
      "step": 1643
    },
    {
      "epoch": 0.19767931221066554,
      "flos": 13828034764800.0,
      "grad_norm": 3.1118129967838883,
      "learning_rate": 3.712383487718015e-06,
      "loss": 0.9895,
      "num_input_tokens_seen": 34364665,
      "step": 1644
    },
    {
      "epoch": 0.19779955510130465,
      "flos": 18315658014720.0,
      "grad_norm": 2.8263691040479646,
      "learning_rate": 3.7119808957538365e-06,
      "loss": 1.0693,
      "num_input_tokens_seen": 34383380,
      "step": 1645
    },
    {
      "epoch": 0.19791979799194373,
      "flos": 14751110615040.0,
      "grad_norm": 3.2677433134052447,
      "learning_rate": 3.711578044083517e-06,
      "loss": 0.9936,
      "num_input_tokens_seen": 34399900,
      "step": 1646
    },
    {
      "epoch": 0.1980400408825828,
      "flos": 18208259297280.0,
      "grad_norm": 9.471702593748077,
      "learning_rate": 3.7111749327681698e-06,
      "loss": 0.9513,
      "num_input_tokens_seen": 34419655,
      "step": 1647
    },
    {
      "epoch": 0.1981602837732219,
      "flos": 16717484421120.0,
      "grad_norm": 3.2651052837133614,
      "learning_rate": 3.7107715618689455e-06,
      "loss": 1.0554,
      "num_input_tokens_seen": 34438350,
      "step": 1648
    },
    {
      "epoch": 0.198280526663861,
      "flos": 16481401835520.0,
      "grad_norm": 1.805619432672492,
      "learning_rate": 3.710367931447035e-06,
      "loss": 1.0302,
      "num_input_tokens_seen": 34459850,
      "step": 1649
    },
    {
      "epoch": 0.1984007695545001,
      "flos": 15406333501440.0,
      "grad_norm": 4.36310027181746,
      "learning_rate": 3.70996404156367e-06,
      "loss": 1.0546,
      "num_input_tokens_seen": 34479205,
      "step": 1650
    },
    {
      "epoch": 0.19852101244513917,
      "flos": 25736926126080.0,
      "grad_norm": 2.1116598682048244,
      "learning_rate": 3.7095598922801187e-06,
      "loss": 0.924,
      "num_input_tokens_seen": 34501000,
      "step": 1651
    },
    {
      "epoch": 0.19864125533577828,
      "flos": 16423410278400.0,
      "grad_norm": 2.488229125238972,
      "learning_rate": 3.7091554836576914e-06,
      "loss": 0.957,
      "num_input_tokens_seen": 34517395,
      "step": 1652
    },
    {
      "epoch": 0.19876149822641737,
      "flos": 17504211763200.0,
      "grad_norm": 2.363293356115687,
      "learning_rate": 3.708750815757736e-06,
      "loss": 1.0267,
      "num_input_tokens_seen": 34537885,
      "step": 1653
    },
    {
      "epoch": 0.19888174111705645,
      "flos": 22957542481920.0,
      "grad_norm": 3.0099286450517764,
      "learning_rate": 3.7083458886416407e-06,
      "loss": 0.9128,
      "num_input_tokens_seen": 34556800,
      "step": 1654
    },
    {
      "epoch": 0.19900198400769553,
      "flos": 17504181104640.0,
      "grad_norm": 5.268774522608912,
      "learning_rate": 3.707940702370832e-06,
      "loss": 1.0781,
      "num_input_tokens_seen": 34577365,
      "step": 1655
    },
    {
      "epoch": 0.19912222689833464,
      "flos": 48633096499200.0,
      "grad_norm": 1.2730548026162,
      "learning_rate": 3.707535257006777e-06,
      "loss": 0.793,
      "num_input_tokens_seen": 34642710,
      "step": 1656
    },
    {
      "epoch": 0.19924246978897373,
      "flos": 11132250685440.0,
      "grad_norm": 4.189431954710497,
      "learning_rate": 3.707129552610981e-06,
      "loss": 1.0798,
      "num_input_tokens_seen": 34661080,
      "step": 1657
    },
    {
      "epoch": 0.1993627126796128,
      "flos": 12443278970880.0,
      "grad_norm": 3.360058808039146,
      "learning_rate": 3.70672358924499e-06,
      "loss": 0.938,
      "num_input_tokens_seen": 34680040,
      "step": 1658
    },
    {
      "epoch": 0.19948295557025192,
      "flos": 29172115169280.0,
      "grad_norm": 2.2995025760298615,
      "learning_rate": 3.706317366970386e-06,
      "loss": 0.9825,
      "num_input_tokens_seen": 34700760,
      "step": 1659
    },
    {
      "epoch": 0.199603198460891,
      "flos": 17844749844480.0,
      "grad_norm": 2.9805004273177365,
      "learning_rate": 3.705910885848795e-06,
      "loss": 1.0418,
      "num_input_tokens_seen": 34718855,
      "step": 1660
    },
    {
      "epoch": 0.19972344135153008,
      "flos": 14252601630720.0,
      "grad_norm": 3.09573793602273,
      "learning_rate": 3.705504145941879e-06,
      "loss": 1.0397,
      "num_input_tokens_seen": 34736745,
      "step": 1661
    },
    {
      "epoch": 0.1998436842421692,
      "flos": 16870304686080.0,
      "grad_norm": 3.5908440824359436,
      "learning_rate": 3.7050971473113403e-06,
      "loss": 0.9851,
      "num_input_tokens_seen": 34756240,
      "step": 1662
    },
    {
      "epoch": 0.19996392713280828,
      "flos": 25764741550080.0,
      "grad_norm": 1.9428629945718972,
      "learning_rate": 3.7046898900189196e-06,
      "loss": 1.0031,
      "num_input_tokens_seen": 34780295,
      "step": 1663
    },
    {
      "epoch": 0.20008417002344736,
      "flos": 16820560281600.0,
      "grad_norm": 1.841552265060468,
      "learning_rate": 3.704282374126398e-06,
      "loss": 1.0395,
      "num_input_tokens_seen": 34799695,
      "step": 1664
    },
    {
      "epoch": 0.20020441291408644,
      "flos": 15537592688640.0,
      "grad_norm": 11.055705424644934,
      "learning_rate": 3.7038745996955954e-06,
      "loss": 1.0834,
      "num_input_tokens_seen": 34818760,
      "step": 1665
    },
    {
      "epoch": 0.20032465580472555,
      "flos": 16476557783040.0,
      "grad_norm": 5.733978507239487,
      "learning_rate": 3.703466566788371e-06,
      "loss": 0.9059,
      "num_input_tokens_seen": 34837610,
      "step": 1666
    },
    {
      "epoch": 0.20044489869536464,
      "flos": 16975281377280.0,
      "grad_norm": 2.4981300035951013,
      "learning_rate": 3.703058275466622e-06,
      "loss": 0.9466,
      "num_input_tokens_seen": 34856565,
      "step": 1667
    },
    {
      "epoch": 0.20056514158600372,
      "flos": 15590188339200.0,
      "grad_norm": 3.272375257663053,
      "learning_rate": 3.7026497257922877e-06,
      "loss": 0.975,
      "num_input_tokens_seen": 34876595,
      "step": 1668
    },
    {
      "epoch": 0.20068538447664283,
      "flos": 16980094771200.0,
      "grad_norm": 3.0168688486817725,
      "learning_rate": 3.7022409178273436e-06,
      "loss": 1.0379,
      "num_input_tokens_seen": 34897295,
      "step": 1669
    },
    {
      "epoch": 0.2008056273672819,
      "flos": 13073108459520.0,
      "grad_norm": 2.9953897337888433,
      "learning_rate": 3.7018318516338054e-06,
      "loss": 0.9791,
      "num_input_tokens_seen": 34916175,
      "step": 1670
    },
    {
      "epoch": 0.200925870257921,
      "flos": 16822154526720.0,
      "grad_norm": 3.312205463486321,
      "learning_rate": 3.7014225272737284e-06,
      "loss": 1.0131,
      "num_input_tokens_seen": 34935120,
      "step": 1671
    },
    {
      "epoch": 0.20104611314856008,
      "flos": 11473524572160.0,
      "grad_norm": 2.925392685312734,
      "learning_rate": 3.701012944809207e-06,
      "loss": 0.939,
      "num_input_tokens_seen": 34951955,
      "step": 1672
    },
    {
      "epoch": 0.2011663560391992,
      "flos": 15196380119040.0,
      "grad_norm": 2.3709267269929044,
      "learning_rate": 3.700603104302374e-06,
      "loss": 0.9891,
      "num_input_tokens_seen": 34971485,
      "step": 1673
    },
    {
      "epoch": 0.20128659892983827,
      "flos": 44547160719360.0,
      "grad_norm": 0.915362485156994,
      "learning_rate": 3.7001930058154027e-06,
      "loss": 0.7814,
      "num_input_tokens_seen": 35036165,
      "step": 1674
    },
    {
      "epoch": 0.20140684182047736,
      "flos": 20256822374400.0,
      "grad_norm": 3.728075276224684,
      "learning_rate": 3.6997826494105037e-06,
      "loss": 0.998,
      "num_input_tokens_seen": 35056330,
      "step": 1675
    },
    {
      "epoch": 0.20152708471111647,
      "flos": 19994763878400.0,
      "grad_norm": 2.6701458248935186,
      "learning_rate": 3.6993720351499286e-06,
      "loss": 0.8889,
      "num_input_tokens_seen": 35077175,
      "step": 1676
    },
    {
      "epoch": 0.20164732760175555,
      "flos": 16743215063040.0,
      "grad_norm": 2.306669979570836,
      "learning_rate": 3.6989611630959666e-06,
      "loss": 0.9664,
      "num_input_tokens_seen": 35095450,
      "step": 1677
    },
    {
      "epoch": 0.20176757049239463,
      "flos": 50923383336960.0,
      "grad_norm": 0.7188403860563277,
      "learning_rate": 3.6985500333109474e-06,
      "loss": 0.8086,
      "num_input_tokens_seen": 35163500,
      "step": 1678
    },
    {
      "epoch": 0.20188781338303372,
      "flos": 15219443466240.0,
      "grad_norm": 2.633089679197712,
      "learning_rate": 3.6981386458572385e-06,
      "loss": 0.9666,
      "num_input_tokens_seen": 35181195,
      "step": 1679
    },
    {
      "epoch": 0.20200805627367283,
      "flos": 8116539187200.0,
      "grad_norm": 2.732725943170089,
      "learning_rate": 3.6977270007972468e-06,
      "loss": 0.9544,
      "num_input_tokens_seen": 35198450,
      "step": 1680
    },
    {
      "epoch": 0.2021282991643119,
      "flos": 20333094543360.0,
      "grad_norm": 3.0062933714468185,
      "learning_rate": 3.6973150981934196e-06,
      "loss": 0.9175,
      "num_input_tokens_seen": 35219400,
      "step": 1681
    },
    {
      "epoch": 0.202248542054951,
      "flos": 12698684559360.0,
      "grad_norm": 4.096240501823507,
      "learning_rate": 3.6969029381082415e-06,
      "loss": 1.0314,
      "num_input_tokens_seen": 35235115,
      "step": 1682
    },
    {
      "epoch": 0.2023687849455901,
      "flos": 14094784020480.0,
      "grad_norm": 2.3949996714103965,
      "learning_rate": 3.696490520604237e-06,
      "loss": 1.0065,
      "num_input_tokens_seen": 35253525,
      "step": 1683
    },
    {
      "epoch": 0.20248902783622919,
      "flos": 16035304550400.0,
      "grad_norm": 2.0357170869812777,
      "learning_rate": 3.696077845743968e-06,
      "loss": 1.0099,
      "num_input_tokens_seen": 35272835,
      "step": 1684
    },
    {
      "epoch": 0.20260927072686827,
      "flos": 16139392143360.0,
      "grad_norm": 2.778787615852221,
      "learning_rate": 3.69566491359004e-06,
      "loss": 0.921,
      "num_input_tokens_seen": 35289200,
      "step": 1685
    },
    {
      "epoch": 0.20272951361750738,
      "flos": 36488498565120.0,
      "grad_norm": 5.277744732580411,
      "learning_rate": 3.695251724205092e-06,
      "loss": 0.9094,
      "num_input_tokens_seen": 35313280,
      "step": 1686
    },
    {
      "epoch": 0.20284975650814646,
      "flos": 18919664885760.0,
      "grad_norm": 1.6668550941128317,
      "learning_rate": 3.6948382776518054e-06,
      "loss": 1.0581,
      "num_input_tokens_seen": 35333705,
      "step": 1687
    },
    {
      "epoch": 0.20296999939878554,
      "flos": 11342020116480.0,
      "grad_norm": 2.9565856678507276,
      "learning_rate": 3.6944245739929e-06,
      "loss": 1.0078,
      "num_input_tokens_seen": 35349585,
      "step": 1688
    },
    {
      "epoch": 0.20309024228942463,
      "flos": 13618939822080.0,
      "grad_norm": 5.467048947218617,
      "learning_rate": 3.6940106132911332e-06,
      "loss": 0.9276,
      "num_input_tokens_seen": 35366490,
      "step": 1689
    },
    {
      "epoch": 0.20321048518006374,
      "flos": 16219680583680.0,
      "grad_norm": 3.809522497937567,
      "learning_rate": 3.6935963956093037e-06,
      "loss": 1.0802,
      "num_input_tokens_seen": 35386295,
      "step": 1690
    },
    {
      "epoch": 0.20333072807070282,
      "flos": 13542391726080.0,
      "grad_norm": 2.205733790537294,
      "learning_rate": 3.6931819210102474e-06,
      "loss": 0.8983,
      "num_input_tokens_seen": 35405410,
      "step": 1691
    },
    {
      "epoch": 0.2034509709613419,
      "flos": 12884041666560.0,
      "grad_norm": 2.808045483636507,
      "learning_rate": 3.6927671895568402e-06,
      "loss": 1.058,
      "num_input_tokens_seen": 35424190,
      "step": 1692
    },
    {
      "epoch": 0.20357121385198101,
      "flos": 16292672286720.0,
      "grad_norm": 2.789629539314095,
      "learning_rate": 3.692352201311996e-06,
      "loss": 1.0759,
      "num_input_tokens_seen": 35442760,
      "step": 1693
    },
    {
      "epoch": 0.2036914567426201,
      "flos": 14854707671040.0,
      "grad_norm": 2.7110673065329376,
      "learning_rate": 3.6919369563386687e-06,
      "loss": 0.9643,
      "num_input_tokens_seen": 35462280,
      "step": 1694
    },
    {
      "epoch": 0.20381169963325918,
      "flos": 10971275243520.0,
      "grad_norm": 4.212626442299198,
      "learning_rate": 3.69152145469985e-06,
      "loss": 0.9991,
      "num_input_tokens_seen": 35479045,
      "step": 1695
    },
    {
      "epoch": 0.20393194252389826,
      "flos": 20539338240000.0,
      "grad_norm": 2.6810578604191906,
      "learning_rate": 3.691105696458572e-06,
      "loss": 1.024,
      "num_input_tokens_seen": 35496060,
      "step": 1696
    },
    {
      "epoch": 0.20405218541453737,
      "flos": 15980133580800.0,
      "grad_norm": 14.18599743385566,
      "learning_rate": 3.690689681677904e-06,
      "loss": 0.863,
      "num_input_tokens_seen": 35514250,
      "step": 1697
    },
    {
      "epoch": 0.20417242830517646,
      "flos": 18054611251200.0,
      "grad_norm": 2.046835629616,
      "learning_rate": 3.690273410420956e-06,
      "loss": 1.0849,
      "num_input_tokens_seen": 35533735,
      "step": 1698
    },
    {
      "epoch": 0.20429267119581554,
      "flos": 10449948180480.0,
      "grad_norm": 4.551567671118437,
      "learning_rate": 3.689856882750875e-06,
      "loss": 0.9752,
      "num_input_tokens_seen": 35548655,
      "step": 1699
    },
    {
      "epoch": 0.20441291408645465,
      "flos": 12597509529600.0,
      "grad_norm": 1.843757629378746,
      "learning_rate": 3.6894400987308486e-06,
      "loss": 0.9842,
      "num_input_tokens_seen": 35565895,
      "step": 1700
    },
    {
      "epoch": 0.20453315697709373,
      "flos": 11630545059840.0,
      "grad_norm": 2.337001698465183,
      "learning_rate": 3.6890230584241024e-06,
      "loss": 1.0409,
      "num_input_tokens_seen": 35582545,
      "step": 1701
    },
    {
      "epoch": 0.20465339986773282,
      "flos": 47768901304320.0,
      "grad_norm": 0.9978864507275025,
      "learning_rate": 3.6886057618939016e-06,
      "loss": 0.899,
      "num_input_tokens_seen": 35645085,
      "step": 1702
    },
    {
      "epoch": 0.2047736427583719,
      "flos": 29984327884800.0,
      "grad_norm": 3.886793954399631,
      "learning_rate": 3.6881882092035492e-06,
      "loss": 0.8834,
      "num_input_tokens_seen": 35666190,
      "step": 1703
    },
    {
      "epoch": 0.204893885649011,
      "flos": 50086030602240.0,
      "grad_norm": 1.060775198341152,
      "learning_rate": 3.6877704004163873e-06,
      "loss": 0.8818,
      "num_input_tokens_seen": 35726315,
      "step": 1704
    },
    {
      "epoch": 0.2050141285396501,
      "flos": 15773215395840.0,
      "grad_norm": 2.9996631751120884,
      "learning_rate": 3.6873523355957984e-06,
      "loss": 0.9889,
      "num_input_tokens_seen": 35745035,
      "step": 1705
    },
    {
      "epoch": 0.20513437143028918,
      "flos": 33087064596480.0,
      "grad_norm": 1.0811839728509294,
      "learning_rate": 3.686934014805201e-06,
      "loss": 0.9195,
      "num_input_tokens_seen": 35795385,
      "step": 1706
    },
    {
      "epoch": 0.20525461432092829,
      "flos": 15560196157440.0,
      "grad_norm": 2.7838218028105652,
      "learning_rate": 3.6865154381080552e-06,
      "loss": 1.0165,
      "num_input_tokens_seen": 35815790,
      "step": 1707
    },
    {
      "epoch": 0.20537485721156737,
      "flos": 15064691712000.0,
      "grad_norm": 3.571230446905832,
      "learning_rate": 3.6860966055678585e-06,
      "loss": 1.0259,
      "num_input_tokens_seen": 35831865,
      "step": 1708
    },
    {
      "epoch": 0.20549510010220645,
      "flos": 14329241702400.0,
      "grad_norm": 2.622652548252433,
      "learning_rate": 3.685677517248147e-06,
      "loss": 1.0674,
      "num_input_tokens_seen": 35850475,
      "step": 1709
    },
    {
      "epoch": 0.20561534299284553,
      "flos": 12047753871360.0,
      "grad_norm": 2.5175593740210362,
      "learning_rate": 3.6852581732124967e-06,
      "loss": 1.0043,
      "num_input_tokens_seen": 35867540,
      "step": 1710
    },
    {
      "epoch": 0.20573558588348465,
      "flos": 16269057085440.0,
      "grad_norm": 3.336326844412001,
      "learning_rate": 3.6848385735245213e-06,
      "loss": 0.9647,
      "num_input_tokens_seen": 35886350,
      "step": 1711
    },
    {
      "epoch": 0.20585582877412373,
      "flos": 17528838696960.0,
      "grad_norm": 3.4815602244797788,
      "learning_rate": 3.6844187182478734e-06,
      "loss": 1.0637,
      "num_input_tokens_seen": 35906925,
      "step": 1712
    },
    {
      "epoch": 0.2059760716647628,
      "flos": 17215594844160.0,
      "grad_norm": 2.6382757266469428,
      "learning_rate": 3.683998607446246e-06,
      "loss": 0.9486,
      "num_input_tokens_seen": 35925295,
      "step": 1713
    },
    {
      "epoch": 0.20609631455540192,
      "flos": 14357026467840.0,
      "grad_norm": 2.6669777052998414,
      "learning_rate": 3.6835782411833686e-06,
      "loss": 0.9488,
      "num_input_tokens_seen": 35944535,
      "step": 1714
    },
    {
      "epoch": 0.206216557446041,
      "flos": 14095059947520.0,
      "grad_norm": 2.101853928336999,
      "learning_rate": 3.68315761952301e-06,
      "loss": 0.9491,
      "num_input_tokens_seen": 35961485,
      "step": 1715
    },
    {
      "epoch": 0.2063368003366801,
      "flos": 17135919575040.0,
      "grad_norm": 3.592967072494988,
      "learning_rate": 3.6827367425289797e-06,
      "loss": 1.0331,
      "num_input_tokens_seen": 35980980,
      "step": 1716
    },
    {
      "epoch": 0.2064570432273192,
      "flos": 14436701736960.0,
      "grad_norm": 3.698374989363664,
      "learning_rate": 3.6823156102651225e-06,
      "loss": 0.9284,
      "num_input_tokens_seen": 35998855,
      "step": 1717
    },
    {
      "epoch": 0.20657728611795828,
      "flos": 14565906800640.0,
      "grad_norm": 2.908942011523057,
      "learning_rate": 3.6818942227953257e-06,
      "loss": 0.9128,
      "num_input_tokens_seen": 36019120,
      "step": 1718
    },
    {
      "epoch": 0.20669752900859736,
      "flos": 15485303623680.0,
      "grad_norm": 3.0771326314652767,
      "learning_rate": 3.681472580183512e-06,
      "loss": 0.8877,
      "num_input_tokens_seen": 36037490,
      "step": 1719
    },
    {
      "epoch": 0.20681777189923645,
      "flos": 10686429327360.0,
      "grad_norm": 2.362751950347228,
      "learning_rate": 3.6810506824936455e-06,
      "loss": 1.071,
      "num_input_tokens_seen": 36055290,
      "step": 1720
    },
    {
      "epoch": 0.20693801478987556,
      "flos": 40413144207360.0,
      "grad_norm": 1.0784052608936134,
      "learning_rate": 3.680628529789726e-06,
      "loss": 0.8608,
      "num_input_tokens_seen": 36107420,
      "step": 1721
    },
    {
      "epoch": 0.20705825768051464,
      "flos": 15351928995840.0,
      "grad_norm": 3.870514642567269,
      "learning_rate": 3.680206122135796e-06,
      "loss": 1.0565,
      "num_input_tokens_seen": 36127745,
      "step": 1722
    },
    {
      "epoch": 0.20717850057115372,
      "flos": 18395639869440.0,
      "grad_norm": 2.3852455599745226,
      "learning_rate": 3.6797834595959323e-06,
      "loss": 0.984,
      "num_input_tokens_seen": 36147365,
      "step": 1723
    },
    {
      "epoch": 0.20729874346179283,
      "flos": 20754043699200.0,
      "grad_norm": 3.2932193972947297,
      "learning_rate": 3.679360542234254e-06,
      "loss": 0.9803,
      "num_input_tokens_seen": 36166430,
      "step": 1724
    },
    {
      "epoch": 0.20741898635243192,
      "flos": 20677097041920.0,
      "grad_norm": 5.632338928547382,
      "learning_rate": 3.678937370114916e-06,
      "loss": 0.9285,
      "num_input_tokens_seen": 36185955,
      "step": 1725
    },
    {
      "epoch": 0.207539229243071,
      "flos": 11000746229760.0,
      "grad_norm": 2.8045036234391025,
      "learning_rate": 3.678513943302114e-06,
      "loss": 0.9967,
      "num_input_tokens_seen": 36202450,
      "step": 1726
    },
    {
      "epoch": 0.20765947213371008,
      "flos": 14566519971840.0,
      "grad_norm": 2.6502663028713274,
      "learning_rate": 3.678090261860082e-06,
      "loss": 1.048,
      "num_input_tokens_seen": 36221900,
      "step": 1727
    },
    {
      "epoch": 0.2077797150243492,
      "flos": 13728178053120.0,
      "grad_norm": 2.1297888929673707,
      "learning_rate": 3.6776663258530906e-06,
      "loss": 0.9811,
      "num_input_tokens_seen": 36240270,
      "step": 1728
    },
    {
      "epoch": 0.20789995791498828,
      "flos": 15506558115840.0,
      "grad_norm": 2.7540152453774183,
      "learning_rate": 3.6772421353454516e-06,
      "loss": 0.9127,
      "num_input_tokens_seen": 36258585,
      "step": 1729
    },
    {
      "epoch": 0.20802020080562736,
      "flos": 16455149998080.0,
      "grad_norm": 3.7898440469345767,
      "learning_rate": 3.6768176904015153e-06,
      "loss": 1.0779,
      "num_input_tokens_seen": 36278110,
      "step": 1730
    },
    {
      "epoch": 0.20814044369626647,
      "flos": 16399979028480.0,
      "grad_norm": 3.6349921793899465,
      "learning_rate": 3.6763929910856674e-06,
      "loss": 0.8105,
      "num_input_tokens_seen": 36296280,
      "step": 1731
    },
    {
      "epoch": 0.20826068658690555,
      "flos": 13910193377280.0,
      "grad_norm": 2.739330052801888,
      "learning_rate": 3.6759680374623365e-06,
      "loss": 0.9719,
      "num_input_tokens_seen": 36313915,
      "step": 1732
    },
    {
      "epoch": 0.20838092947754464,
      "flos": 18054304665600.0,
      "grad_norm": 7.0171277448190565,
      "learning_rate": 3.675542829595986e-06,
      "loss": 0.941,
      "num_input_tokens_seen": 36333300,
      "step": 1733
    },
    {
      "epoch": 0.20850117236818372,
      "flos": 17111323299840.0,
      "grad_norm": 1.8424115899664362,
      "learning_rate": 3.6751173675511213e-06,
      "loss": 0.9897,
      "num_input_tokens_seen": 36355065,
      "step": 1734
    },
    {
      "epoch": 0.20862141525882283,
      "flos": 14247205724160.0,
      "grad_norm": 2.8596139609955067,
      "learning_rate": 3.674691651392283e-06,
      "loss": 1.0851,
      "num_input_tokens_seen": 36372455,
      "step": 1735
    },
    {
      "epoch": 0.2087416581494619,
      "flos": 27857315880960.0,
      "grad_norm": 2.72463020125329,
      "learning_rate": 3.674265681184053e-06,
      "loss": 0.9626,
      "num_input_tokens_seen": 36395435,
      "step": 1736
    },
    {
      "epoch": 0.208861901040101,
      "flos": 18577225973760.0,
      "grad_norm": 18.80859740070058,
      "learning_rate": 3.6738394569910504e-06,
      "loss": 1.066,
      "num_input_tokens_seen": 36415695,
      "step": 1737
    },
    {
      "epoch": 0.2089821439307401,
      "flos": 20227596656640.0,
      "grad_norm": 4.254279058144332,
      "learning_rate": 3.6734129788779333e-06,
      "loss": 1.0218,
      "num_input_tokens_seen": 36434590,
      "step": 1738
    },
    {
      "epoch": 0.2091023868213792,
      "flos": 14960910704640.0,
      "grad_norm": 3.4701991447886944,
      "learning_rate": 3.6729862469093976e-06,
      "loss": 1.1016,
      "num_input_tokens_seen": 36453405,
      "step": 1739
    },
    {
      "epoch": 0.20922262971201827,
      "flos": 15956824965120.0,
      "grad_norm": 3.03793086382996,
      "learning_rate": 3.6725592611501782e-06,
      "loss": 1.0111,
      "num_input_tokens_seen": 36471800,
      "step": 1740
    },
    {
      "epoch": 0.20934287260265738,
      "flos": 19834953461760.0,
      "grad_norm": 2.7545227659305676,
      "learning_rate": 3.6721320216650496e-06,
      "loss": 0.9657,
      "num_input_tokens_seen": 36492135,
      "step": 1741
    },
    {
      "epoch": 0.20946311549329646,
      "flos": 11630238474240.0,
      "grad_norm": 2.8600789855507256,
      "learning_rate": 3.6717045285188215e-06,
      "loss": 1.0493,
      "num_input_tokens_seen": 36509550,
      "step": 1742
    },
    {
      "epoch": 0.20958335838393555,
      "flos": 15983689973760.0,
      "grad_norm": 6.6642047468882035,
      "learning_rate": 3.671276781776346e-06,
      "loss": 1.0623,
      "num_input_tokens_seen": 36527925,
      "step": 1743
    },
    {
      "epoch": 0.20970360127457463,
      "flos": 17946568704000.0,
      "grad_norm": 2.901652319221003,
      "learning_rate": 3.6708487815025128e-06,
      "loss": 0.8638,
      "num_input_tokens_seen": 36548225,
      "step": 1744
    },
    {
      "epoch": 0.20982384416521374,
      "flos": 13099207004160.0,
      "grad_norm": 3.0447253346708347,
      "learning_rate": 3.6704205277622463e-06,
      "loss": 0.9394,
      "num_input_tokens_seen": 36566385,
      "step": 1745
    },
    {
      "epoch": 0.20994408705585282,
      "flos": 18053875445760.0,
      "grad_norm": 3.693767329399652,
      "learning_rate": 3.6699920206205146e-06,
      "loss": 0.9985,
      "num_input_tokens_seen": 36586845,
      "step": 1746
    },
    {
      "epoch": 0.2100643299464919,
      "flos": 15140994539520.0,
      "grad_norm": 2.3574223806764683,
      "learning_rate": 3.669563260142321e-06,
      "loss": 1.0296,
      "num_input_tokens_seen": 36605455,
      "step": 1747
    },
    {
      "epoch": 0.21018457283713102,
      "flos": 13728147394560.0,
      "grad_norm": 8.075860360905683,
      "learning_rate": 3.6691342463927083e-06,
      "loss": 1.0425,
      "num_input_tokens_seen": 36624170,
      "step": 1748
    },
    {
      "epoch": 0.2103048157277701,
      "flos": 20178802667520.0,
      "grad_norm": 2.920149023094136,
      "learning_rate": 3.668704979436758e-06,
      "loss": 1.0194,
      "num_input_tokens_seen": 36643985,
      "step": 1749
    },
    {
      "epoch": 0.21042505861840918,
      "flos": 12364768727040.0,
      "grad_norm": 2.608112692311415,
      "learning_rate": 3.668275459339588e-06,
      "loss": 0.988,
      "num_input_tokens_seen": 36662185,
      "step": 1750
    },
    {
      "epoch": 0.21054530150904827,
      "flos": 10030838538240.0,
      "grad_norm": 5.908804850556742,
      "learning_rate": 3.667845686166358e-06,
      "loss": 1.0054,
      "num_input_tokens_seen": 36678830,
      "step": 1751
    },
    {
      "epoch": 0.21066554439968738,
      "flos": 13198205276160.0,
      "grad_norm": 2.4602605948393035,
      "learning_rate": 3.6674156599822634e-06,
      "loss": 1.0621,
      "num_input_tokens_seen": 36694345,
      "step": 1752
    },
    {
      "epoch": 0.21078578729032646,
      "flos": 16821602672640.0,
      "grad_norm": 3.0122481498069886,
      "learning_rate": 3.666985380852539e-06,
      "loss": 1.0097,
      "num_input_tokens_seen": 36713070,
      "step": 1753
    },
    {
      "epoch": 0.21090603018096554,
      "flos": 20908519526400.0,
      "grad_norm": 3.8088087128839025,
      "learning_rate": 3.6665548488424576e-06,
      "loss": 0.9486,
      "num_input_tokens_seen": 36731550,
      "step": 1754
    },
    {
      "epoch": 0.21102627307160465,
      "flos": 16536327536640.0,
      "grad_norm": 3.1862654445628555,
      "learning_rate": 3.6661240640173307e-06,
      "loss": 1.0728,
      "num_input_tokens_seen": 36752740,
      "step": 1755
    },
    {
      "epoch": 0.21114651596224374,
      "flos": 39087115776000.0,
      "grad_norm": 0.9799495705455347,
      "learning_rate": 3.6656930264425085e-06,
      "loss": 0.8341,
      "num_input_tokens_seen": 36816505,
      "step": 1756
    },
    {
      "epoch": 0.21126675885288282,
      "flos": 15301264834560.0,
      "grad_norm": 3.570618196918638,
      "learning_rate": 3.665261736183378e-06,
      "loss": 0.9632,
      "num_input_tokens_seen": 36836260,
      "step": 1757
    },
    {
      "epoch": 0.2113870017435219,
      "flos": 7696509788160.0,
      "grad_norm": 3.705154056203358,
      "learning_rate": 3.664830193305366e-06,
      "loss": 1.0836,
      "num_input_tokens_seen": 36853755,
      "step": 1758
    },
    {
      "epoch": 0.211507244634161,
      "flos": 11787228303360.0,
      "grad_norm": 2.7836823703342866,
      "learning_rate": 3.6643983978739373e-06,
      "loss": 0.9546,
      "num_input_tokens_seen": 36870090,
      "step": 1759
    },
    {
      "epoch": 0.2116274875248001,
      "flos": 14877065871360.0,
      "grad_norm": 2.1198223089929265,
      "learning_rate": 3.663966349954596e-06,
      "loss": 1.0119,
      "num_input_tokens_seen": 36889990,
      "step": 1760
    },
    {
      "epoch": 0.21174773041543918,
      "flos": 48833699020800.0,
      "grad_norm": 0.8153848418735832,
      "learning_rate": 3.6635340496128816e-06,
      "loss": 0.8131,
      "num_input_tokens_seen": 36946640,
      "step": 1761
    },
    {
      "epoch": 0.2118679733060783,
      "flos": 14672508395520.0,
      "grad_norm": 2.0922049899862323,
      "learning_rate": 3.6631014969143747e-06,
      "loss": 1.1189,
      "num_input_tokens_seen": 36966050,
      "step": 1762
    },
    {
      "epoch": 0.21198821619671737,
      "flos": 16508971991040.0,
      "grad_norm": 2.9774927803163416,
      "learning_rate": 3.662668691924693e-06,
      "loss": 1.0822,
      "num_input_tokens_seen": 36986820,
      "step": 1763
    },
    {
      "epoch": 0.21210845908735645,
      "flos": 17425119006720.0,
      "grad_norm": 2.309181925919261,
      "learning_rate": 3.6622356347094927e-06,
      "loss": 0.9103,
      "num_input_tokens_seen": 37008105,
      "step": 1764
    },
    {
      "epoch": 0.21222870197799554,
      "flos": 19287497195520.0,
      "grad_norm": 4.2264361406712245,
      "learning_rate": 3.6618023253344684e-06,
      "loss": 0.9862,
      "num_input_tokens_seen": 37026685,
      "step": 1765
    },
    {
      "epoch": 0.21234894486863465,
      "flos": 11945505792000.0,
      "grad_norm": 2.3880555550073095,
      "learning_rate": 3.6613687638653527e-06,
      "loss": 1.0437,
      "num_input_tokens_seen": 37044575,
      "step": 1766
    },
    {
      "epoch": 0.21246918775927373,
      "flos": 16689209118720.0,
      "grad_norm": 3.084874996329828,
      "learning_rate": 3.660934950367916e-06,
      "loss": 0.9812,
      "num_input_tokens_seen": 37063540,
      "step": 1767
    },
    {
      "epoch": 0.21258943064991281,
      "flos": 15904198656000.0,
      "grad_norm": 2.5951914631031703,
      "learning_rate": 3.660500884907968e-06,
      "loss": 1.0315,
      "num_input_tokens_seen": 37084000,
      "step": 1768
    },
    {
      "epoch": 0.21270967354055192,
      "flos": 42540340162560.0,
      "grad_norm": 0.9249051612486452,
      "learning_rate": 3.660066567551356e-06,
      "loss": 0.8197,
      "num_input_tokens_seen": 37143865,
      "step": 1769
    },
    {
      "epoch": 0.212829916431191,
      "flos": 15301878005760.0,
      "grad_norm": 3.151812313620614,
      "learning_rate": 3.6596319983639657e-06,
      "loss": 1.0275,
      "num_input_tokens_seen": 37162165,
      "step": 1770
    },
    {
      "epoch": 0.2129501593218301,
      "flos": 20651611668480.0,
      "grad_norm": 2.946938306304452,
      "learning_rate": 3.6591971774117214e-06,
      "loss": 1.065,
      "num_input_tokens_seen": 37184860,
      "step": 1771
    },
    {
      "epoch": 0.2130704022124692,
      "flos": 13334737735680.0,
      "grad_norm": 7.712822618955063,
      "learning_rate": 3.6587621047605833e-06,
      "loss": 1.0047,
      "num_input_tokens_seen": 37201750,
      "step": 1772
    },
    {
      "epoch": 0.21319064510310828,
      "flos": 9821253058560.0,
      "grad_norm": 3.559101826380762,
      "learning_rate": 3.6583267804765542e-06,
      "loss": 1.0713,
      "num_input_tokens_seen": 37215805,
      "step": 1773
    },
    {
      "epoch": 0.21331088799374737,
      "flos": 14881848606720.0,
      "grad_norm": 2.4356440790107987,
      "learning_rate": 3.6578912046256702e-06,
      "loss": 1.0532,
      "num_input_tokens_seen": 37234045,
      "step": 1774
    },
    {
      "epoch": 0.21343113088438645,
      "flos": 13203693158400.0,
      "grad_norm": 3.682681022177783,
      "learning_rate": 3.6574553772740083e-06,
      "loss": 0.9698,
      "num_input_tokens_seen": 37251695,
      "step": 1775
    },
    {
      "epoch": 0.21355137377502556,
      "flos": 48272928829440.0,
      "grad_norm": 0.9590143327248577,
      "learning_rate": 3.657019298487684e-06,
      "loss": 0.8615,
      "num_input_tokens_seen": 37316425,
      "step": 1776
    },
    {
      "epoch": 0.21367161666566464,
      "flos": 24636035174400.0,
      "grad_norm": 4.39043189876468,
      "learning_rate": 3.6565829683328495e-06,
      "loss": 1.014,
      "num_input_tokens_seen": 37338770,
      "step": 1777
    },
    {
      "epoch": 0.21379185955630373,
      "flos": 13465782312960.0,
      "grad_norm": 2.931552791613948,
      "learning_rate": 3.6561463868756965e-06,
      "loss": 1.0594,
      "num_input_tokens_seen": 37357190,
      "step": 1778
    },
    {
      "epoch": 0.21391210244694284,
      "flos": 20098360934400.0,
      "grad_norm": 2.3805418107852523,
      "learning_rate": 3.655709554182452e-06,
      "loss": 0.9891,
      "num_input_tokens_seen": 37377250,
      "step": 1779
    },
    {
      "epoch": 0.21403234533758192,
      "flos": 12362867896320.0,
      "grad_norm": 2.512476496800051,
      "learning_rate": 3.6552724703193855e-06,
      "loss": 1.0593,
      "num_input_tokens_seen": 37394160,
      "step": 1780
    },
    {
      "epoch": 0.214152588228221,
      "flos": 36933514690560.0,
      "grad_norm": 0.8384806727307066,
      "learning_rate": 3.654835135352801e-06,
      "loss": 0.7794,
      "num_input_tokens_seen": 37448690,
      "step": 1781
    },
    {
      "epoch": 0.21427283111886009,
      "flos": 13829475717120.0,
      "grad_norm": 3.0693294125765833,
      "learning_rate": 3.654397549349043e-06,
      "loss": 1.0774,
      "num_input_tokens_seen": 37465785,
      "step": 1782
    },
    {
      "epoch": 0.2143930740094992,
      "flos": 14253184143360.0,
      "grad_norm": 7.021788882044511,
      "learning_rate": 3.653959712374491e-06,
      "loss": 0.9402,
      "num_input_tokens_seen": 37483610,
      "step": 1783
    },
    {
      "epoch": 0.21451331690013828,
      "flos": 15484721111040.0,
      "grad_norm": 2.470297605987181,
      "learning_rate": 3.6535216244955663e-06,
      "loss": 1.0296,
      "num_input_tokens_seen": 37503225,
      "step": 1784
    },
    {
      "epoch": 0.21463355979077736,
      "flos": 23429952921600.0,
      "grad_norm": 2.536289050166817,
      "learning_rate": 3.653083285778726e-06,
      "loss": 0.9043,
      "num_input_tokens_seen": 37524315,
      "step": 1785
    },
    {
      "epoch": 0.21475380268141647,
      "flos": 15301694054400.0,
      "grad_norm": 3.521292902881801,
      "learning_rate": 3.6526446962904653e-06,
      "loss": 1.006,
      "num_input_tokens_seen": 37542750,
      "step": 1786
    },
    {
      "epoch": 0.21487404557205556,
      "flos": 22930401546240.0,
      "grad_norm": 3.229392521660995,
      "learning_rate": 3.652205856097318e-06,
      "loss": 0.9389,
      "num_input_tokens_seen": 37565655,
      "step": 1787
    },
    {
      "epoch": 0.21499428846269464,
      "flos": 8925839339520.0,
      "grad_norm": 2.9156829991761057,
      "learning_rate": 3.651766765265856e-06,
      "loss": 0.9885,
      "num_input_tokens_seen": 37582385,
      "step": 1788
    },
    {
      "epoch": 0.21511453135333372,
      "flos": 16686327214080.0,
      "grad_norm": 4.131869289594475,
      "learning_rate": 3.65132742386269e-06,
      "loss": 1.0014,
      "num_input_tokens_seen": 37597325,
      "step": 1789
    },
    {
      "epoch": 0.21523477424397283,
      "flos": 19182367211520.0,
      "grad_norm": 3.5774017002721212,
      "learning_rate": 3.6508878319544656e-06,
      "loss": 1.0413,
      "num_input_tokens_seen": 37617260,
      "step": 1790
    },
    {
      "epoch": 0.21535501713461191,
      "flos": 13413156003840.0,
      "grad_norm": 3.6953115967311922,
      "learning_rate": 3.65044798960787e-06,
      "loss": 1.0031,
      "num_input_tokens_seen": 37635320,
      "step": 1791
    },
    {
      "epoch": 0.215475260025251,
      "flos": 12679422873600.0,
      "grad_norm": 3.182451326368111,
      "learning_rate": 3.650007896889627e-06,
      "loss": 0.976,
      "num_input_tokens_seen": 37653620,
      "step": 1792
    },
    {
      "epoch": 0.2155955029158901,
      "flos": 11787626864640.0,
      "grad_norm": 2.585682836249507,
      "learning_rate": 3.6495675538664974e-06,
      "loss": 0.9997,
      "num_input_tokens_seen": 37672355,
      "step": 1793
    },
    {
      "epoch": 0.2157157458065292,
      "flos": 16794369761280.0,
      "grad_norm": 3.430064534728515,
      "learning_rate": 3.649126960605282e-06,
      "loss": 1.0259,
      "num_input_tokens_seen": 37693060,
      "step": 1794
    },
    {
      "epoch": 0.21583598869716827,
      "flos": 15720803696640.0,
      "grad_norm": 3.2009110405575143,
      "learning_rate": 3.6486861171728174e-06,
      "loss": 1.0231,
      "num_input_tokens_seen": 37711175,
      "step": 1795
    },
    {
      "epoch": 0.21595623158780738,
      "flos": 16665348648960.0,
      "grad_norm": 2.2158638893976925,
      "learning_rate": 3.6482450236359803e-06,
      "loss": 0.9876,
      "num_input_tokens_seen": 37732750,
      "step": 1796
    },
    {
      "epoch": 0.21607647447844647,
      "flos": 19155655495680.0,
      "grad_norm": 4.958494338725994,
      "learning_rate": 3.647803680061683e-06,
      "loss": 0.9699,
      "num_input_tokens_seen": 37752885,
      "step": 1797
    },
    {
      "epoch": 0.21619671736908555,
      "flos": 10235212062720.0,
      "grad_norm": 3.864966045393343,
      "learning_rate": 3.6473620865168776e-06,
      "loss": 0.941,
      "num_input_tokens_seen": 37769475,
      "step": 1798
    },
    {
      "epoch": 0.21631696025972463,
      "flos": 12705214832640.0,
      "grad_norm": 3.4037034901310763,
      "learning_rate": 3.646920243068554e-06,
      "loss": 1.0106,
      "num_input_tokens_seen": 37787090,
      "step": 1799
    },
    {
      "epoch": 0.21643720315036374,
      "flos": 17399173754880.0,
      "grad_norm": 2.8919275596694978,
      "learning_rate": 3.6464781497837384e-06,
      "loss": 0.9384,
      "num_input_tokens_seen": 37808785,
      "step": 1800
    },
    {
      "epoch": 0.21655744604100283,
      "flos": 20282369064960.0,
      "grad_norm": 3.8046082913341133,
      "learning_rate": 3.6460358067294965e-06,
      "loss": 0.9338,
      "num_input_tokens_seen": 37829735,
      "step": 1801
    },
    {
      "epoch": 0.2166776889316419,
      "flos": 14301518254080.0,
      "grad_norm": 5.760784484635575,
      "learning_rate": 3.645593213972932e-06,
      "loss": 0.9693,
      "num_input_tokens_seen": 37848360,
      "step": 1802
    },
    {
      "epoch": 0.21679793182228102,
      "flos": 10737584025600.0,
      "grad_norm": 3.8407496706016344,
      "learning_rate": 3.6451503715811852e-06,
      "loss": 0.9824,
      "num_input_tokens_seen": 37866390,
      "step": 1803
    },
    {
      "epoch": 0.2169181747129201,
      "flos": 12312449003520.0,
      "grad_norm": 4.923096714071694,
      "learning_rate": 3.6447072796214345e-06,
      "loss": 1.0055,
      "num_input_tokens_seen": 37884675,
      "step": 1804
    },
    {
      "epoch": 0.21703841760355919,
      "flos": 42054094602240.0,
      "grad_norm": 1.7245323017044918,
      "learning_rate": 3.644263938160898e-06,
      "loss": 0.8668,
      "num_input_tokens_seen": 37940360,
      "step": 1805
    },
    {
      "epoch": 0.21715866049419827,
      "flos": 15930143907840.0,
      "grad_norm": 4.187021939901883,
      "learning_rate": 3.6438203472668293e-06,
      "loss": 0.924,
      "num_input_tokens_seen": 37959725,
      "step": 1806
    },
    {
      "epoch": 0.21727890338483738,
      "flos": 12206889799680.0,
      "grad_norm": 5.778001357097209,
      "learning_rate": 3.6433765070065206e-06,
      "loss": 1.0266,
      "num_input_tokens_seen": 37977235,
      "step": 1807
    },
    {
      "epoch": 0.21739914627547646,
      "flos": 9472651776000.0,
      "grad_norm": 5.155675312730414,
      "learning_rate": 3.6429324174473025e-06,
      "loss": 1.0835,
      "num_input_tokens_seen": 37990495,
      "step": 1808
    },
    {
      "epoch": 0.21751938916611555,
      "flos": 14881695313920.0,
      "grad_norm": 3.892524715043767,
      "learning_rate": 3.6424880786565425e-06,
      "loss": 1.0445,
      "num_input_tokens_seen": 38006360,
      "step": 1809
    },
    {
      "epoch": 0.21763963205675466,
      "flos": 19654041845760.0,
      "grad_norm": 3.4135317182507485,
      "learning_rate": 3.6420434907016482e-06,
      "loss": 0.9851,
      "num_input_tokens_seen": 38025770,
      "step": 1810
    },
    {
      "epoch": 0.21775987494739374,
      "flos": 15219658076160.0,
      "grad_norm": 4.32006217134738,
      "learning_rate": 3.6415986536500606e-06,
      "loss": 1.0176,
      "num_input_tokens_seen": 38043820,
      "step": 1811
    },
    {
      "epoch": 0.21788011783803282,
      "flos": 12994107678720.0,
      "grad_norm": 3.6173559689271135,
      "learning_rate": 3.641153567569263e-06,
      "loss": 1.0165,
      "num_input_tokens_seen": 38061855,
      "step": 1812
    },
    {
      "epoch": 0.2180003607286719,
      "flos": 21567666708480.0,
      "grad_norm": 3.83350203322564,
      "learning_rate": 3.640708232526774e-06,
      "loss": 1.1499,
      "num_input_tokens_seen": 38080230,
      "step": 1813
    },
    {
      "epoch": 0.21812060361931102,
      "flos": 18130208931840.0,
      "grad_norm": 2.7225553693649305,
      "learning_rate": 3.6402626485901504e-06,
      "loss": 0.9787,
      "num_input_tokens_seen": 38099045,
      "step": 1814
    },
    {
      "epoch": 0.2182408465099501,
      "flos": 15563476623360.0,
      "grad_norm": 3.8486150950730855,
      "learning_rate": 3.639816815826988e-06,
      "loss": 0.9728,
      "num_input_tokens_seen": 38118090,
      "step": 1815
    },
    {
      "epoch": 0.21836108940058918,
      "flos": 16820284354560.0,
      "grad_norm": 3.160828696851869,
      "learning_rate": 3.6393707343049176e-06,
      "loss": 0.9745,
      "num_input_tokens_seen": 38138140,
      "step": 1816
    },
    {
      "epoch": 0.2184813322912283,
      "flos": 17556255559680.0,
      "grad_norm": 8.309320974873733,
      "learning_rate": 3.6389244040916104e-06,
      "loss": 0.9479,
      "num_input_tokens_seen": 38156935,
      "step": 1817
    },
    {
      "epoch": 0.21860157518186737,
      "flos": 18916691005440.0,
      "grad_norm": 3.0544474085434206,
      "learning_rate": 3.6384778252547747e-06,
      "loss": 0.988,
      "num_input_tokens_seen": 38172535,
      "step": 1818
    },
    {
      "epoch": 0.21872181807250646,
      "flos": 14829038346240.0,
      "grad_norm": 6.80596834691142,
      "learning_rate": 3.638030997862155e-06,
      "loss": 0.9806,
      "num_input_tokens_seen": 38191190,
      "step": 1819
    },
    {
      "epoch": 0.21884206096314554,
      "flos": 43814316687360.0,
      "grad_norm": 0.8738589339778589,
      "learning_rate": 3.6375839219815356e-06,
      "loss": 0.8214,
      "num_input_tokens_seen": 38248710,
      "step": 1820
    },
    {
      "epoch": 0.21896230385378465,
      "flos": 16688258703360.0,
      "grad_norm": 7.001872221028414,
      "learning_rate": 3.6371365976807375e-06,
      "loss": 1.0321,
      "num_input_tokens_seen": 38268825,
      "step": 1821
    },
    {
      "epoch": 0.21908254674442373,
      "flos": 17844320624640.0,
      "grad_norm": 3.5314331007217237,
      "learning_rate": 3.6366890250276185e-06,
      "loss": 1.0271,
      "num_input_tokens_seen": 38289500,
      "step": 1822
    },
    {
      "epoch": 0.21920278963506282,
      "flos": 16717423104000.0,
      "grad_norm": 3.4947688655801152,
      "learning_rate": 3.6362412040900764e-06,
      "loss": 1.0936,
      "num_input_tokens_seen": 38309010,
      "step": 1823
    },
    {
      "epoch": 0.21932303252570193,
      "flos": 20834393456640.0,
      "grad_norm": 2.9767699482234087,
      "learning_rate": 3.635793134936044e-06,
      "loss": 0.9993,
      "num_input_tokens_seen": 38329740,
      "step": 1824
    },
    {
      "epoch": 0.219443275416341,
      "flos": 14771077447680.0,
      "grad_norm": 2.7475610668279034,
      "learning_rate": 3.635344817633494e-06,
      "loss": 0.9306,
      "num_input_tokens_seen": 38348775,
      "step": 1825
    },
    {
      "epoch": 0.2195635183069801,
      "flos": 10240485335040.0,
      "grad_norm": 3.50762707872269,
      "learning_rate": 3.634896252250436e-06,
      "loss": 0.9558,
      "num_input_tokens_seen": 38365260,
      "step": 1826
    },
    {
      "epoch": 0.2196837611976192,
      "flos": 17241846681600.0,
      "grad_norm": 4.048035915396363,
      "learning_rate": 3.6344474388549157e-06,
      "loss": 1.0182,
      "num_input_tokens_seen": 38384635,
      "step": 1827
    },
    {
      "epoch": 0.2198040040882583,
      "flos": 12807953448960.0,
      "grad_norm": 5.251986718061493,
      "learning_rate": 3.6339983775150183e-06,
      "loss": 1.0158,
      "num_input_tokens_seen": 38400915,
      "step": 1828
    },
    {
      "epoch": 0.21992424697889737,
      "flos": 12599502336000.0,
      "grad_norm": 5.44660730716354,
      "learning_rate": 3.6335490682988664e-06,
      "loss": 1.0416,
      "num_input_tokens_seen": 38416245,
      "step": 1829
    },
    {
      "epoch": 0.22004448986953645,
      "flos": 12494832230400.0,
      "grad_norm": 3.563875702374124,
      "learning_rate": 3.63309951127462e-06,
      "loss": 1.0226,
      "num_input_tokens_seen": 38432875,
      "step": 1830
    },
    {
      "epoch": 0.22016473276017556,
      "flos": 15827834511360.0,
      "grad_norm": 4.226476756581749,
      "learning_rate": 3.6326497065104757e-06,
      "loss": 0.9447,
      "num_input_tokens_seen": 38453060,
      "step": 1831
    },
    {
      "epoch": 0.22028497565081465,
      "flos": 18185226608640.0,
      "grad_norm": 10.239892209200969,
      "learning_rate": 3.6321996540746697e-06,
      "loss": 0.9735,
      "num_input_tokens_seen": 38471855,
      "step": 1832
    },
    {
      "epoch": 0.22040521854145373,
      "flos": 25868798484480.0,
      "grad_norm": 3.117911271511985,
      "learning_rate": 3.6317493540354733e-06,
      "loss": 0.9938,
      "num_input_tokens_seen": 38494990,
      "step": 1833
    },
    {
      "epoch": 0.22052546143209284,
      "flos": 8326584545280.0,
      "grad_norm": 5.98748045342745,
      "learning_rate": 3.6312988064611976e-06,
      "loss": 0.9683,
      "num_input_tokens_seen": 38513020,
      "step": 1834
    },
    {
      "epoch": 0.22064570432273192,
      "flos": 17217556992000.0,
      "grad_norm": 2.647420502244467,
      "learning_rate": 3.6308480114201896e-06,
      "loss": 1.0075,
      "num_input_tokens_seen": 38534660,
      "step": 1835
    },
    {
      "epoch": 0.220765947213371,
      "flos": 12705429442560.0,
      "grad_norm": 2.5834140970054422,
      "learning_rate": 3.630396968980835e-06,
      "loss": 0.9691,
      "num_input_tokens_seen": 38552255,
      "step": 1836
    },
    {
      "epoch": 0.2208861901040101,
      "flos": 19047766241280.0,
      "grad_norm": 5.973209971675419,
      "learning_rate": 3.6299456792115575e-06,
      "loss": 1.0399,
      "num_input_tokens_seen": 38573230,
      "step": 1837
    },
    {
      "epoch": 0.2210064329946492,
      "flos": 12623424122880.0,
      "grad_norm": 2.800250014995871,
      "learning_rate": 3.629494142180815e-06,
      "loss": 1.008,
      "num_input_tokens_seen": 38591695,
      "step": 1838
    },
    {
      "epoch": 0.22112667588528828,
      "flos": 12731282718720.0,
      "grad_norm": 6.7304078208392735,
      "learning_rate": 3.6290423579571075e-06,
      "loss": 1.0459,
      "num_input_tokens_seen": 38607955,
      "step": 1839
    },
    {
      "epoch": 0.22124691877592736,
      "flos": 13020114247680.0,
      "grad_norm": 3.99096684914711,
      "learning_rate": 3.6285903266089694e-06,
      "loss": 0.9958,
      "num_input_tokens_seen": 38626950,
      "step": 1840
    },
    {
      "epoch": 0.22136716166656648,
      "flos": 14776565329920.0,
      "grad_norm": 2.9099863297183974,
      "learning_rate": 3.628138048204974e-06,
      "loss": 0.9686,
      "num_input_tokens_seen": 38647355,
      "step": 1841
    },
    {
      "epoch": 0.22148740455720556,
      "flos": 12521666580480.0,
      "grad_norm": 8.97114132359697,
      "learning_rate": 3.6276855228137304e-06,
      "loss": 0.9622,
      "num_input_tokens_seen": 38665280,
      "step": 1842
    },
    {
      "epoch": 0.22160764744784464,
      "flos": 15433198510080.0,
      "grad_norm": 2.694593427631407,
      "learning_rate": 3.6272327505038874e-06,
      "loss": 1.0132,
      "num_input_tokens_seen": 38681465,
      "step": 1843
    },
    {
      "epoch": 0.22172789033848372,
      "flos": 16897169694720.0,
      "grad_norm": 2.851581228534388,
      "learning_rate": 3.626779731344131e-06,
      "loss": 0.9866,
      "num_input_tokens_seen": 38700975,
      "step": 1844
    },
    {
      "epoch": 0.22184813322912283,
      "flos": 12023310888960.0,
      "grad_norm": 4.6383540187150105,
      "learning_rate": 3.6263264654031814e-06,
      "loss": 1.0545,
      "num_input_tokens_seen": 38717595,
      "step": 1845
    },
    {
      "epoch": 0.22196837611976192,
      "flos": 44254987407360.0,
      "grad_norm": 0.7688556552016861,
      "learning_rate": 3.6258729527498008e-06,
      "loss": 0.793,
      "num_input_tokens_seen": 38778160,
      "step": 1846
    },
    {
      "epoch": 0.222088619010401,
      "flos": 18186728878080.0,
      "grad_norm": 3.1276581712179365,
      "learning_rate": 3.6254191934527854e-06,
      "loss": 0.8193,
      "num_input_tokens_seen": 38797235,
      "step": 1847
    },
    {
      "epoch": 0.2222088619010401,
      "flos": 13702539386880.0,
      "grad_norm": 4.283574051283814,
      "learning_rate": 3.6249651875809715e-06,
      "loss": 0.8419,
      "num_input_tokens_seen": 38816835,
      "step": 1848
    },
    {
      "epoch": 0.2223291047916792,
      "flos": 13544905728000.0,
      "grad_norm": 6.648562007684032,
      "learning_rate": 3.62451093520323e-06,
      "loss": 1.0883,
      "num_input_tokens_seen": 38834460,
      "step": 1849
    },
    {
      "epoch": 0.22244934768231828,
      "flos": 14803338362880.0,
      "grad_norm": 5.805318354010075,
      "learning_rate": 3.6240564363884714e-06,
      "loss": 1.0907,
      "num_input_tokens_seen": 38854125,
      "step": 1850
    },
    {
      "epoch": 0.2225695905729574,
      "flos": 11053617807360.0,
      "grad_norm": 3.287977751041697,
      "learning_rate": 3.623601691205643e-06,
      "loss": 0.8965,
      "num_input_tokens_seen": 38872920,
      "step": 1851
    },
    {
      "epoch": 0.22268983346359647,
      "flos": 18053906104320.0,
      "grad_norm": 3.4493572686830514,
      "learning_rate": 3.623146699723729e-06,
      "loss": 1.0048,
      "num_input_tokens_seen": 38892100,
      "step": 1852
    },
    {
      "epoch": 0.22281007635423555,
      "flos": 9349302497280.0,
      "grad_norm": 4.3969268643940485,
      "learning_rate": 3.6226914620117507e-06,
      "loss": 0.9849,
      "num_input_tokens_seen": 38910440,
      "step": 1853
    },
    {
      "epoch": 0.22293031924487464,
      "flos": 10843419156480.0,
      "grad_norm": 3.6999647602774437,
      "learning_rate": 3.622235978138768e-06,
      "loss": 1.0092,
      "num_input_tokens_seen": 38927785,
      "step": 1854
    },
    {
      "epoch": 0.22305056213551375,
      "flos": 16035243233280.0,
      "grad_norm": 8.496595261778712,
      "learning_rate": 3.621780248173877e-06,
      "loss": 1.02,
      "num_input_tokens_seen": 38945705,
      "step": 1855
    },
    {
      "epoch": 0.22317080502615283,
      "flos": 46450668257280.0,
      "grad_norm": 0.9211976100357266,
      "learning_rate": 3.6213242721862125e-06,
      "loss": 0.8518,
      "num_input_tokens_seen": 39003880,
      "step": 1856
    },
    {
      "epoch": 0.2232910479167919,
      "flos": 18343258828800.0,
      "grad_norm": 2.2925256564203287,
      "learning_rate": 3.620868050244945e-06,
      "loss": 0.9516,
      "num_input_tokens_seen": 39024080,
      "step": 1857
    },
    {
      "epoch": 0.22341129080743102,
      "flos": 16528264335360.0,
      "grad_norm": 10.50122462805957,
      "learning_rate": 3.6204115824192817e-06,
      "loss": 0.9641,
      "num_input_tokens_seen": 39041275,
      "step": 1858
    },
    {
      "epoch": 0.2235315336980701,
      "flos": 15064936980480.0,
      "grad_norm": 3.17132383171146,
      "learning_rate": 3.619954868778471e-06,
      "loss": 0.9604,
      "num_input_tokens_seen": 39057690,
      "step": 1859
    },
    {
      "epoch": 0.2236517765887092,
      "flos": 14121710346240.0,
      "grad_norm": 2.377327230535873,
      "learning_rate": 3.6194979093917944e-06,
      "loss": 1.0167,
      "num_input_tokens_seen": 39076825,
      "step": 1860
    },
    {
      "epoch": 0.22377201947934827,
      "flos": 16501644595200.0,
      "grad_norm": 2.9799407451294098,
      "learning_rate": 3.6190407043285724e-06,
      "loss": 1.0793,
      "num_input_tokens_seen": 39094280,
      "step": 1861
    },
    {
      "epoch": 0.22389226236998738,
      "flos": 19074600591360.0,
      "grad_norm": 6.212281509947685,
      "learning_rate": 3.618583253658163e-06,
      "loss": 0.9483,
      "num_input_tokens_seen": 39114100,
      "step": 1862
    },
    {
      "epoch": 0.22401250526062647,
      "flos": 17189281689600.0,
      "grad_norm": 3.1935837973032664,
      "learning_rate": 3.618125557449961e-06,
      "loss": 1.0682,
      "num_input_tokens_seen": 39131875,
      "step": 1863
    },
    {
      "epoch": 0.22413274815126555,
      "flos": 11862887301120.0,
      "grad_norm": 4.672981054730076,
      "learning_rate": 3.6176676157733983e-06,
      "loss": 1.0318,
      "num_input_tokens_seen": 39146605,
      "step": 1864
    },
    {
      "epoch": 0.22425299104190466,
      "flos": 15169576427520.0,
      "grad_norm": 3.3064046956957265,
      "learning_rate": 3.6172094286979443e-06,
      "loss": 0.9555,
      "num_input_tokens_seen": 39163695,
      "step": 1865
    },
    {
      "epoch": 0.22437323393254374,
      "flos": 22935920087040.0,
      "grad_norm": 1.835839321676777,
      "learning_rate": 3.6167509962931064e-06,
      "loss": 1.0093,
      "num_input_tokens_seen": 39189115,
      "step": 1866
    },
    {
      "epoch": 0.22449347682318282,
      "flos": 12756584140800.0,
      "grad_norm": 3.7881286099442977,
      "learning_rate": 3.6162923186284276e-06,
      "loss": 0.9792,
      "num_input_tokens_seen": 39204795,
      "step": 1867
    },
    {
      "epoch": 0.2246137197138219,
      "flos": 13255859589120.0,
      "grad_norm": 2.962243428769253,
      "learning_rate": 3.6158333957734888e-06,
      "loss": 1.0589,
      "num_input_tokens_seen": 39223105,
      "step": 1868
    },
    {
      "epoch": 0.22473396260446102,
      "flos": 11022460600320.0,
      "grad_norm": 4.795969128888749,
      "learning_rate": 3.6153742277979088e-06,
      "loss": 1.0289,
      "num_input_tokens_seen": 39240255,
      "step": 1869
    },
    {
      "epoch": 0.2248542054951001,
      "flos": 10214632058880.0,
      "grad_norm": 3.2423617764248935,
      "learning_rate": 3.6149148147713434e-06,
      "loss": 0.9833,
      "num_input_tokens_seen": 39258210,
      "step": 1870
    },
    {
      "epoch": 0.22497444838573918,
      "flos": 13648594759680.0,
      "grad_norm": 3.6275097281748163,
      "learning_rate": 3.614455156763484e-06,
      "loss": 1.0737,
      "num_input_tokens_seen": 39276235,
      "step": 1871
    },
    {
      "epoch": 0.2250946912763783,
      "flos": 11971696312320.0,
      "grad_norm": 8.331238699559806,
      "learning_rate": 3.613995253844061e-06,
      "loss": 0.9159,
      "num_input_tokens_seen": 39293635,
      "step": 1872
    },
    {
      "epoch": 0.22521493416701738,
      "flos": 17556562145280.0,
      "grad_norm": 3.92724654906632,
      "learning_rate": 3.6135351060828414e-06,
      "loss": 0.9991,
      "num_input_tokens_seen": 39313830,
      "step": 1873
    },
    {
      "epoch": 0.22533517705765646,
      "flos": 12626827223040.0,
      "grad_norm": 3.476996302596401,
      "learning_rate": 3.6130747135496285e-06,
      "loss": 0.8828,
      "num_input_tokens_seen": 39332550,
      "step": 1874
    },
    {
      "epoch": 0.22545541994829554,
      "flos": 24031629742080.0,
      "grad_norm": 3.7073057372528333,
      "learning_rate": 3.6126140763142646e-06,
      "loss": 0.8602,
      "num_input_tokens_seen": 39357300,
      "step": 1875
    },
    {
      "epoch": 0.22557566283893465,
      "flos": 13596734914560.0,
      "grad_norm": 5.741272314930677,
      "learning_rate": 3.6121531944466275e-06,
      "loss": 1.0513,
      "num_input_tokens_seen": 39374345,
      "step": 1876
    },
    {
      "epoch": 0.22569590572957374,
      "flos": 14747339612160.0,
      "grad_norm": 5.240910973022575,
      "learning_rate": 3.611692068016633e-06,
      "loss": 0.9837,
      "num_input_tokens_seen": 39390395,
      "step": 1877
    },
    {
      "epoch": 0.22581614862021282,
      "flos": 13073016483840.0,
      "grad_norm": 3.0955669577702665,
      "learning_rate": 3.611230697094233e-06,
      "loss": 0.9458,
      "num_input_tokens_seen": 39406815,
      "step": 1878
    },
    {
      "epoch": 0.22593639151085193,
      "flos": 14227024281600.0,
      "grad_norm": 3.0311214773716175,
      "learning_rate": 3.6107690817494173e-06,
      "loss": 1.0755,
      "num_input_tokens_seen": 39426755,
      "step": 1879
    },
    {
      "epoch": 0.226056634401491,
      "flos": 9244080537600.0,
      "grad_norm": 3.6669382225972567,
      "learning_rate": 3.6103072220522117e-06,
      "loss": 0.9122,
      "num_input_tokens_seen": 39442005,
      "step": 1880
    },
    {
      "epoch": 0.2261768772921301,
      "flos": 13464954531840.0,
      "grad_norm": 2.099412009255671,
      "learning_rate": 3.609845118072682e-06,
      "loss": 1.1057,
      "num_input_tokens_seen": 39460395,
      "step": 1881
    },
    {
      "epoch": 0.2262971201827692,
      "flos": 14173416898560.0,
      "grad_norm": 2.32402178787287,
      "learning_rate": 3.6093827698809276e-06,
      "loss": 0.9881,
      "num_input_tokens_seen": 39479215,
      "step": 1882
    },
    {
      "epoch": 0.2264173630734083,
      "flos": 11787841474560.0,
      "grad_norm": 3.144501983316833,
      "learning_rate": 3.6089201775470864e-06,
      "loss": 1.0439,
      "num_input_tokens_seen": 39494390,
      "step": 1883
    },
    {
      "epoch": 0.22653760596404737,
      "flos": 17347068641280.0,
      "grad_norm": 3.7815043466319316,
      "learning_rate": 3.6084573411413334e-06,
      "loss": 0.9804,
      "num_input_tokens_seen": 39513505,
      "step": 1884
    },
    {
      "epoch": 0.22665784885468646,
      "flos": 12993739776000.0,
      "grad_norm": 2.8188655448965774,
      "learning_rate": 3.607994260733881e-06,
      "loss": 0.9999,
      "num_input_tokens_seen": 39532465,
      "step": 1885
    },
    {
      "epoch": 0.22677809174532557,
      "flos": 17108226785280.0,
      "grad_norm": 2.161784270809765,
      "learning_rate": 3.6075309363949776e-06,
      "loss": 0.9423,
      "num_input_tokens_seen": 39551355,
      "step": 1886
    },
    {
      "epoch": 0.22689833463596465,
      "flos": 14460868792320.0,
      "grad_norm": 15.338704757816448,
      "learning_rate": 3.6070673681949094e-06,
      "loss": 1.0044,
      "num_input_tokens_seen": 39569440,
      "step": 1887
    },
    {
      "epoch": 0.22701857752660373,
      "flos": 21465847848960.0,
      "grad_norm": 3.269018135232878,
      "learning_rate": 3.606603556203999e-06,
      "loss": 0.9961,
      "num_input_tokens_seen": 39594105,
      "step": 1888
    },
    {
      "epoch": 0.22713882041724284,
      "flos": 15983230095360.0,
      "grad_norm": 2.624756665531716,
      "learning_rate": 3.6061395004926066e-06,
      "loss": 1.0383,
      "num_input_tokens_seen": 39612760,
      "step": 1889
    },
    {
      "epoch": 0.22725906330788193,
      "flos": 14566918533120.0,
      "grad_norm": 5.3738609703941895,
      "learning_rate": 3.605675201131129e-06,
      "loss": 1.0348,
      "num_input_tokens_seen": 39630940,
      "step": 1890
    },
    {
      "epoch": 0.227379306198521,
      "flos": 13466272849920.0,
      "grad_norm": 4.193516522658002,
      "learning_rate": 3.60521065819e-06,
      "loss": 0.9895,
      "num_input_tokens_seen": 39647970,
      "step": 1891
    },
    {
      "epoch": 0.2274995490891601,
      "flos": 15484506501120.0,
      "grad_norm": 2.7891298601153216,
      "learning_rate": 3.60474587173969e-06,
      "loss": 1.0714,
      "num_input_tokens_seen": 39666175,
      "step": 1892
    },
    {
      "epoch": 0.2276197919797992,
      "flos": 13515097497600.0,
      "grad_norm": 3.1986494850385228,
      "learning_rate": 3.6042808418507084e-06,
      "loss": 1.0338,
      "num_input_tokens_seen": 39683580,
      "step": 1893
    },
    {
      "epoch": 0.22774003487043828,
      "flos": 13334553784320.0,
      "grad_norm": 10.471103323994914,
      "learning_rate": 3.6038155685935976e-06,
      "loss": 0.9726,
      "num_input_tokens_seen": 39699870,
      "step": 1894
    },
    {
      "epoch": 0.22786027776107737,
      "flos": 16349958696960.0,
      "grad_norm": 3.598562319102635,
      "learning_rate": 3.6033500520389404e-06,
      "loss": 0.8952,
      "num_input_tokens_seen": 39716260,
      "step": 1895
    },
    {
      "epoch": 0.22798052065171648,
      "flos": 47763597373440.0,
      "grad_norm": 0.8893036661070574,
      "learning_rate": 3.6028842922573553e-06,
      "loss": 0.8862,
      "num_input_tokens_seen": 39780125,
      "step": 1896
    },
    {
      "epoch": 0.22810076354235556,
      "flos": 44440037928960.0,
      "grad_norm": 0.8311227127367303,
      "learning_rate": 3.602418289319497e-06,
      "loss": 0.859,
      "num_input_tokens_seen": 39838400,
      "step": 1897
    },
    {
      "epoch": 0.22822100643299464,
      "flos": 16977703403520.0,
      "grad_norm": 13.601817716409439,
      "learning_rate": 3.601952043296059e-06,
      "loss": 0.9296,
      "num_input_tokens_seen": 39858115,
      "step": 1898
    },
    {
      "epoch": 0.22834124932363373,
      "flos": 14903930880000.0,
      "grad_norm": 3.347516086986994,
      "learning_rate": 3.6014855542577696e-06,
      "loss": 1.0119,
      "num_input_tokens_seen": 39875045,
      "step": 1899
    },
    {
      "epoch": 0.22846149221427284,
      "flos": 17714992926720.0,
      "grad_norm": 2.4561415670207842,
      "learning_rate": 3.6010188222753943e-06,
      "loss": 1.0378,
      "num_input_tokens_seen": 39895535,
      "step": 1900
    },
    {
      "epoch": 0.22858173510491192,
      "flos": 40164698112000.0,
      "grad_norm": 1.1191405717407001,
      "learning_rate": 3.6005518474197372e-06,
      "loss": 0.8709,
      "num_input_tokens_seen": 39947300,
      "step": 1901
    },
    {
      "epoch": 0.228701977995551,
      "flos": 17189680250880.0,
      "grad_norm": 3.134532275680101,
      "learning_rate": 3.6000846297616373e-06,
      "loss": 0.984,
      "num_input_tokens_seen": 39965320,
      "step": 1902
    },
    {
      "epoch": 0.22882222088619011,
      "flos": 15188286259200.0,
      "grad_norm": 3.4492457076646863,
      "learning_rate": 3.5996171693719717e-06,
      "loss": 0.9362,
      "num_input_tokens_seen": 39981135,
      "step": 1903
    },
    {
      "epoch": 0.2289424637768292,
      "flos": 46241818583040.0,
      "grad_norm": 1.46351355619574,
      "learning_rate": 3.5991494663216528e-06,
      "loss": 0.8647,
      "num_input_tokens_seen": 40043840,
      "step": 1904
    },
    {
      "epoch": 0.22906270666746828,
      "flos": 15746902241280.0,
      "grad_norm": 4.125188649805531,
      "learning_rate": 3.5986815206816314e-06,
      "loss": 1.0637,
      "num_input_tokens_seen": 40062380,
      "step": 1905
    },
    {
      "epoch": 0.2291829495581074,
      "flos": 18340131655680.0,
      "grad_norm": 2.3716696427899167,
      "learning_rate": 3.598213332522895e-06,
      "loss": 0.9438,
      "num_input_tokens_seen": 40082130,
      "step": 1906
    },
    {
      "epoch": 0.22930319244874647,
      "flos": 22221387325440.0,
      "grad_norm": 2.9199202419518846,
      "learning_rate": 3.597744901916466e-06,
      "loss": 0.9827,
      "num_input_tokens_seen": 40103135,
      "step": 1907
    },
    {
      "epoch": 0.22942343533938556,
      "flos": 16530226483200.0,
      "grad_norm": 3.010207612237803,
      "learning_rate": 3.5972762289334058e-06,
      "loss": 0.9595,
      "num_input_tokens_seen": 40122485,
      "step": 1908
    },
    {
      "epoch": 0.22954367823002464,
      "flos": 10319056896000.0,
      "grad_norm": 2.7271621175935263,
      "learning_rate": 3.5968073136448116e-06,
      "loss": 1.0486,
      "num_input_tokens_seen": 40140225,
      "step": 1909
    },
    {
      "epoch": 0.22966392112066375,
      "flos": 11866167767040.0,
      "grad_norm": 2.614730638303412,
      "learning_rate": 3.596338156121818e-06,
      "loss": 1.1061,
      "num_input_tokens_seen": 40158830,
      "step": 1910
    },
    {
      "epoch": 0.22978416401130283,
      "flos": 48315253862400.0,
      "grad_norm": 0.8279581801615744,
      "learning_rate": 3.595868756435595e-06,
      "loss": 0.8137,
      "num_input_tokens_seen": 40226230,
      "step": 1911
    },
    {
      "epoch": 0.22990440690194192,
      "flos": 14095213240320.0,
      "grad_norm": 3.124327167557675,
      "learning_rate": 3.5953991146573504e-06,
      "loss": 1.0031,
      "num_input_tokens_seen": 40244595,
      "step": 1912
    },
    {
      "epoch": 0.23002464979258103,
      "flos": 9368104304640.0,
      "grad_norm": 3.3055202435941227,
      "learning_rate": 3.5949292308583294e-06,
      "loss": 1.022,
      "num_input_tokens_seen": 40257560,
      "step": 1913
    },
    {
      "epoch": 0.2301448926832201,
      "flos": 15747208826880.0,
      "grad_norm": 6.840967178674172,
      "learning_rate": 3.594459105109811e-06,
      "loss": 1.0015,
      "num_input_tokens_seen": 40276460,
      "step": 1914
    },
    {
      "epoch": 0.2302651355738592,
      "flos": 14698453647360.0,
      "grad_norm": 2.6033201036918947,
      "learning_rate": 3.593988737483115e-06,
      "loss": 1.0147,
      "num_input_tokens_seen": 40296120,
      "step": 1915
    },
    {
      "epoch": 0.23038537846449827,
      "flos": 13178146467840.0,
      "grad_norm": 2.790362826993707,
      "learning_rate": 3.5935181280495947e-06,
      "loss": 0.9869,
      "num_input_tokens_seen": 40314420,
      "step": 1916
    },
    {
      "epoch": 0.23050562135513739,
      "flos": 45979177574400.0,
      "grad_norm": 0.9170382714599902,
      "learning_rate": 3.5930472768806412e-06,
      "loss": 0.7839,
      "num_input_tokens_seen": 40372810,
      "step": 1917
    },
    {
      "epoch": 0.23062586424577647,
      "flos": 12261631549440.0,
      "grad_norm": 2.57480476898515,
      "learning_rate": 3.5925761840476826e-06,
      "loss": 0.9629,
      "num_input_tokens_seen": 40391140,
      "step": 1918
    },
    {
      "epoch": 0.23074610713641555,
      "flos": 19838141952000.0,
      "grad_norm": 2.821721029041472,
      "learning_rate": 3.592104849622183e-06,
      "loss": 1.0221,
      "num_input_tokens_seen": 40413115,
      "step": 1919
    },
    {
      "epoch": 0.23086635002705466,
      "flos": 20546726952960.0,
      "grad_norm": 2.1653894425687685,
      "learning_rate": 3.591633273675644e-06,
      "loss": 0.9237,
      "num_input_tokens_seen": 40435070,
      "step": 1920
    },
    {
      "epoch": 0.23098659291769374,
      "flos": 45041293639680.0,
      "grad_norm": 0.990378284719004,
      "learning_rate": 3.591161456279602e-06,
      "loss": 0.8234,
      "num_input_tokens_seen": 40480335,
      "step": 1921
    },
    {
      "epoch": 0.23110683580833283,
      "flos": 16690803363840.0,
      "grad_norm": 2.0847108355922863,
      "learning_rate": 3.590689397505633e-06,
      "loss": 0.9938,
      "num_input_tokens_seen": 40500965,
      "step": 1922
    },
    {
      "epoch": 0.2312270786989719,
      "flos": 19417622016000.0,
      "grad_norm": 2.2427270751774793,
      "learning_rate": 3.590217097425347e-06,
      "loss": 1.065,
      "num_input_tokens_seen": 40520585,
      "step": 1923
    },
    {
      "epoch": 0.23134732158961102,
      "flos": 9349210521600.0,
      "grad_norm": 3.4108315549675425,
      "learning_rate": 3.589744556110391e-06,
      "loss": 0.8966,
      "num_input_tokens_seen": 40538295,
      "step": 1924
    },
    {
      "epoch": 0.2314675644802501,
      "flos": 26393743257600.0,
      "grad_norm": 2.0811480568786207,
      "learning_rate": 3.58927177363245e-06,
      "loss": 1.0436,
      "num_input_tokens_seen": 40560840,
      "step": 1925
    },
    {
      "epoch": 0.2315878073708892,
      "flos": 16953475031040.0,
      "grad_norm": 2.8927004399991523,
      "learning_rate": 3.5887987500632447e-06,
      "loss": 0.9296,
      "num_input_tokens_seen": 40578565,
      "step": 1926
    },
    {
      "epoch": 0.2317080502615283,
      "flos": 16376517120000.0,
      "grad_norm": 2.5842678061675692,
      "learning_rate": 3.5883254854745325e-06,
      "loss": 1.0352,
      "num_input_tokens_seen": 40596675,
      "step": 1927
    },
    {
      "epoch": 0.23182829315216738,
      "flos": 7907045683200.0,
      "grad_norm": 3.140589168211329,
      "learning_rate": 3.587851979938107e-06,
      "loss": 0.9543,
      "num_input_tokens_seen": 40613285,
      "step": 1928
    },
    {
      "epoch": 0.23194853604280646,
      "flos": 14069237329920.0,
      "grad_norm": 2.5554082477886406,
      "learning_rate": 3.5873782335257985e-06,
      "loss": 0.9718,
      "num_input_tokens_seen": 40631170,
      "step": 1929
    },
    {
      "epoch": 0.23206877893344555,
      "flos": 10818025758720.0,
      "grad_norm": 3.651299148745938,
      "learning_rate": 3.5869042463094744e-06,
      "loss": 0.9862,
      "num_input_tokens_seen": 40648605,
      "step": 1930
    },
    {
      "epoch": 0.23218902182408466,
      "flos": 16140894412800.0,
      "grad_norm": 2.7167825705331996,
      "learning_rate": 3.586430018361038e-06,
      "loss": 0.9752,
      "num_input_tokens_seen": 40668095,
      "step": 1931
    },
    {
      "epoch": 0.23230926471472374,
      "flos": 16140250583040.0,
      "grad_norm": 3.122446619192948,
      "learning_rate": 3.5859555497524283e-06,
      "loss": 0.9502,
      "num_input_tokens_seen": 40685050,
      "step": 1932
    },
    {
      "epoch": 0.23242950760536282,
      "flos": 14461941841920.0,
      "grad_norm": 4.440090014602019,
      "learning_rate": 3.5854808405556237e-06,
      "loss": 1.1232,
      "num_input_tokens_seen": 40704005,
      "step": 1933
    },
    {
      "epoch": 0.23254975049600193,
      "flos": 11970010091520.0,
      "grad_norm": 3.210557901261635,
      "learning_rate": 3.5850058908426355e-06,
      "loss": 0.9483,
      "num_input_tokens_seen": 40722275,
      "step": 1934
    },
    {
      "epoch": 0.23266999338664102,
      "flos": 16560065372160.0,
      "grad_norm": 2.9444350496378155,
      "learning_rate": 3.584530700685514e-06,
      "loss": 1.0446,
      "num_input_tokens_seen": 40742255,
      "step": 1935
    },
    {
      "epoch": 0.2327902362772801,
      "flos": 13882653880320.0,
      "grad_norm": 2.912968243463488,
      "learning_rate": 3.5840552701563448e-06,
      "loss": 1.0847,
      "num_input_tokens_seen": 40758175,
      "step": 1936
    },
    {
      "epoch": 0.2329104791679192,
      "flos": 11840283832320.0,
      "grad_norm": 3.925337277639349,
      "learning_rate": 3.5835795993272513e-06,
      "loss": 1.0086,
      "num_input_tokens_seen": 40776180,
      "step": 1937
    },
    {
      "epoch": 0.2330307220585583,
      "flos": 15747178168320.0,
      "grad_norm": 6.985036660755591,
      "learning_rate": 3.583103688270391e-06,
      "loss": 0.9086,
      "num_input_tokens_seen": 40795680,
      "step": 1938
    },
    {
      "epoch": 0.23315096494919738,
      "flos": 13701435678720.0,
      "grad_norm": 3.315042060136437,
      "learning_rate": 3.58262753705796e-06,
      "loss": 1.087,
      "num_input_tokens_seen": 40810290,
      "step": 1939
    },
    {
      "epoch": 0.23327120783983646,
      "flos": 37935897907200.0,
      "grad_norm": 0.8034133823830021,
      "learning_rate": 3.5821511457621902e-06,
      "loss": 0.7689,
      "num_input_tokens_seen": 40867310,
      "step": 1940
    },
    {
      "epoch": 0.23339145073047557,
      "flos": 12127122554880.0,
      "grad_norm": 4.785021920946188,
      "learning_rate": 3.5816745144553497e-06,
      "loss": 1.0107,
      "num_input_tokens_seen": 40882350,
      "step": 1941
    },
    {
      "epoch": 0.23351169362111465,
      "flos": 9218104627200.0,
      "grad_norm": 2.756425460109283,
      "learning_rate": 3.5811976432097424e-06,
      "loss": 0.9511,
      "num_input_tokens_seen": 40899740,
      "step": 1942
    },
    {
      "epoch": 0.23363193651175373,
      "flos": 11210454343680.0,
      "grad_norm": 8.529971072292135,
      "learning_rate": 3.58072053209771e-06,
      "loss": 1.0467,
      "num_input_tokens_seen": 40916015,
      "step": 1943
    },
    {
      "epoch": 0.23375217940239285,
      "flos": 14928404520960.0,
      "grad_norm": 3.1347732306177205,
      "learning_rate": 3.5802431811916296e-06,
      "loss": 0.9992,
      "num_input_tokens_seen": 40932345,
      "step": 1944
    },
    {
      "epoch": 0.23387242229303193,
      "flos": 14619514183680.0,
      "grad_norm": 2.8424454935261276,
      "learning_rate": 3.579765590563916e-06,
      "loss": 0.9985,
      "num_input_tokens_seen": 40951465,
      "step": 1945
    },
    {
      "epoch": 0.233992665183671,
      "flos": 17267669299200.0,
      "grad_norm": 2.7917733148349564,
      "learning_rate": 3.579287760287017e-06,
      "loss": 1.0098,
      "num_input_tokens_seen": 40971935,
      "step": 1946
    },
    {
      "epoch": 0.2341129080743101,
      "flos": 21490137538560.0,
      "grad_norm": 3.025677464815731,
      "learning_rate": 3.578809690433421e-06,
      "loss": 0.9315,
      "num_input_tokens_seen": 40993365,
      "step": 1947
    },
    {
      "epoch": 0.2342331509649492,
      "flos": 16193060843520.0,
      "grad_norm": 3.2181275886292604,
      "learning_rate": 3.578331381075651e-06,
      "loss": 0.9943,
      "num_input_tokens_seen": 41013585,
      "step": 1948
    },
    {
      "epoch": 0.2343533938555883,
      "flos": 16796025323520.0,
      "grad_norm": 4.592136644702882,
      "learning_rate": 3.5778528322862646e-06,
      "loss": 0.8844,
      "num_input_tokens_seen": 41032125,
      "step": 1949
    },
    {
      "epoch": 0.23447363674622737,
      "flos": 17476549632000.0,
      "grad_norm": 3.0560677378472074,
      "learning_rate": 3.5773740441378585e-06,
      "loss": 1.0667,
      "num_input_tokens_seen": 41052600,
      "step": 1950
    },
    {
      "epoch": 0.23459387963686648,
      "flos": 38009480232960.0,
      "grad_norm": 2.941179608736047,
      "learning_rate": 3.5768950167030633e-06,
      "loss": 0.9405,
      "num_input_tokens_seen": 41077020,
      "step": 1951
    },
    {
      "epoch": 0.23471412252750556,
      "flos": 17031923957760.0,
      "grad_norm": 6.030860374396737,
      "learning_rate": 3.576415750054548e-06,
      "loss": 0.9808,
      "num_input_tokens_seen": 41096860,
      "step": 1952
    },
    {
      "epoch": 0.23483436541814465,
      "flos": 11106642677760.0,
      "grad_norm": 4.077186681145488,
      "learning_rate": 3.5759362442650172e-06,
      "loss": 1.0577,
      "num_input_tokens_seen": 41113330,
      "step": 1953
    },
    {
      "epoch": 0.23495460830878373,
      "flos": 17739681177600.0,
      "grad_norm": 3.509855970813055,
      "learning_rate": 3.5754564994072113e-06,
      "loss": 1.0391,
      "num_input_tokens_seen": 41131890,
      "step": 1954
    },
    {
      "epoch": 0.23507485119942284,
      "flos": 21725852221440.0,
      "grad_norm": 4.617609666349216,
      "learning_rate": 3.5749765155539067e-06,
      "loss": 0.8095,
      "num_input_tokens_seen": 41152095,
      "step": 1955
    },
    {
      "epoch": 0.23519509409006192,
      "flos": 12990827212800.0,
      "grad_norm": 12.089396952567892,
      "learning_rate": 3.574496292777917e-06,
      "loss": 1.1152,
      "num_input_tokens_seen": 41170025,
      "step": 1956
    },
    {
      "epoch": 0.235315336980701,
      "flos": 21122519838720.0,
      "grad_norm": 2.7383071963486323,
      "learning_rate": 3.574015831152092e-06,
      "loss": 0.9091,
      "num_input_tokens_seen": 41190160,
      "step": 1957
    },
    {
      "epoch": 0.23543557987134012,
      "flos": 13151281459200.0,
      "grad_norm": 3.3833397857404046,
      "learning_rate": 3.573535130749316e-06,
      "loss": 1.0215,
      "num_input_tokens_seen": 41207830,
      "step": 1958
    },
    {
      "epoch": 0.2355558227619792,
      "flos": 17554569338880.0,
      "grad_norm": 3.0395316029138937,
      "learning_rate": 3.5730541916425127e-06,
      "loss": 0.9325,
      "num_input_tokens_seen": 41229030,
      "step": 1959
    },
    {
      "epoch": 0.23567606565261828,
      "flos": 15589820436480.0,
      "grad_norm": 2.357343392697328,
      "learning_rate": 3.572573013904639e-06,
      "loss": 1.0482,
      "num_input_tokens_seen": 41248660,
      "step": 1960
    },
    {
      "epoch": 0.2357963085432574,
      "flos": 9584894545920.0,
      "grad_norm": 3.154249255880882,
      "learning_rate": 3.572091597608689e-06,
      "loss": 1.1131,
      "num_input_tokens_seen": 41266505,
      "step": 1961
    },
    {
      "epoch": 0.23591655143389648,
      "flos": 15692957614080.0,
      "grad_norm": 3.6531907022668424,
      "learning_rate": 3.571609942827694e-06,
      "loss": 0.9204,
      "num_input_tokens_seen": 41285340,
      "step": 1962
    },
    {
      "epoch": 0.23603679432453556,
      "flos": 12048305725440.0,
      "grad_norm": 3.872964262742552,
      "learning_rate": 3.57112804963472e-06,
      "loss": 1.0746,
      "num_input_tokens_seen": 41303275,
      "step": 1963
    },
    {
      "epoch": 0.23615703721517464,
      "flos": 13596489646080.0,
      "grad_norm": 2.80872838065057,
      "learning_rate": 3.57064591810287e-06,
      "loss": 0.962,
      "num_input_tokens_seen": 41320495,
      "step": 1964
    },
    {
      "epoch": 0.23627728010581375,
      "flos": 13544445849600.0,
      "grad_norm": 6.674764038944443,
      "learning_rate": 3.570163548305284e-06,
      "loss": 0.9994,
      "num_input_tokens_seen": 41339145,
      "step": 1965
    },
    {
      "epoch": 0.23639752299645284,
      "flos": 10082790359040.0,
      "grad_norm": 3.312621348335594,
      "learning_rate": 3.569680940315135e-06,
      "loss": 0.8905,
      "num_input_tokens_seen": 41355265,
      "step": 1966
    },
    {
      "epoch": 0.23651776588709192,
      "flos": 16271571087360.0,
      "grad_norm": 2.950218263730652,
      "learning_rate": 3.5691980942056356e-06,
      "loss": 1.0111,
      "num_input_tokens_seen": 41374355,
      "step": 1967
    },
    {
      "epoch": 0.23663800877773103,
      "flos": 13203601182720.0,
      "grad_norm": 2.317387545098939,
      "learning_rate": 3.5687150100500332e-06,
      "loss": 0.9913,
      "num_input_tokens_seen": 41393775,
      "step": 1968
    },
    {
      "epoch": 0.2367582516683701,
      "flos": 18184858705920.0,
      "grad_norm": 2.5728291073210663,
      "learning_rate": 3.568231687921611e-06,
      "loss": 0.9439,
      "num_input_tokens_seen": 41413670,
      "step": 1969
    },
    {
      "epoch": 0.2368784945590092,
      "flos": 16560279982080.0,
      "grad_norm": 2.2402368227904566,
      "learning_rate": 3.5677481278936883e-06,
      "loss": 1.0134,
      "num_input_tokens_seen": 41432970,
      "step": 1970
    },
    {
      "epoch": 0.23699873744964828,
      "flos": 50031158108160.0,
      "grad_norm": 0.8848921662169136,
      "learning_rate": 3.5672643300396214e-06,
      "loss": 0.7917,
      "num_input_tokens_seen": 41501835,
      "step": 1971
    },
    {
      "epoch": 0.2371189803402874,
      "flos": 15510911631360.0,
      "grad_norm": 3.4403513786592463,
      "learning_rate": 3.566780294432802e-06,
      "loss": 0.8793,
      "num_input_tokens_seen": 41518730,
      "step": 1972
    },
    {
      "epoch": 0.23723922323092647,
      "flos": 15563752550400.0,
      "grad_norm": 3.508710784421164,
      "learning_rate": 3.566296021146657e-06,
      "loss": 0.9231,
      "num_input_tokens_seen": 41537830,
      "step": 1973
    },
    {
      "epoch": 0.23735946612156555,
      "flos": 23325804011520.0,
      "grad_norm": 2.3721560466250518,
      "learning_rate": 3.565811510254652e-06,
      "loss": 0.9202,
      "num_input_tokens_seen": 41558430,
      "step": 1974
    },
    {
      "epoch": 0.23747970901220466,
      "flos": 50524608430080.0,
      "grad_norm": 1.0331802221402682,
      "learning_rate": 3.5653267618302845e-06,
      "loss": 0.7891,
      "num_input_tokens_seen": 41625730,
      "step": 1975
    },
    {
      "epoch": 0.23759995190284375,
      "flos": 14802817167360.0,
      "grad_norm": 3.076041508894332,
      "learning_rate": 3.564841775947093e-06,
      "loss": 1.0538,
      "num_input_tokens_seen": 41646340,
      "step": 1976
    },
    {
      "epoch": 0.23772019479348283,
      "flos": 23477949788160.0,
      "grad_norm": 8.787114801551448,
      "learning_rate": 3.5643565526786475e-06,
      "loss": 0.9522,
      "num_input_tokens_seen": 41666000,
      "step": 1977
    },
    {
      "epoch": 0.2378404376841219,
      "flos": 23062580490240.0,
      "grad_norm": 2.4376964000903953,
      "learning_rate": 3.5638710920985574e-06,
      "loss": 0.9677,
      "num_input_tokens_seen": 41687180,
      "step": 1978
    },
    {
      "epoch": 0.23796068057476102,
      "flos": 16345758474240.0,
      "grad_norm": 5.411096594440919,
      "learning_rate": 3.5633853942804655e-06,
      "loss": 1.0054,
      "num_input_tokens_seen": 41705225,
      "step": 1979
    },
    {
      "epoch": 0.2380809234654001,
      "flos": 9507120107520.0,
      "grad_norm": 3.4917165280759783,
      "learning_rate": 3.5628994592980527e-06,
      "loss": 0.9609,
      "num_input_tokens_seen": 41722850,
      "step": 1980
    },
    {
      "epoch": 0.2382011663560392,
      "flos": 11943758254080.0,
      "grad_norm": 4.4116690569214105,
      "learning_rate": 3.562413287225034e-06,
      "loss": 0.9043,
      "num_input_tokens_seen": 41740680,
      "step": 1981
    },
    {
      "epoch": 0.2383214092466783,
      "flos": 13071667507200.0,
      "grad_norm": 4.091092477406421,
      "learning_rate": 3.5619268781351623e-06,
      "loss": 1.0825,
      "num_input_tokens_seen": 41758470,
      "step": 1982
    },
    {
      "epoch": 0.23844165213731738,
      "flos": 14016580362240.0,
      "grad_norm": 2.9941622838420803,
      "learning_rate": 3.5614402321022256e-06,
      "loss": 0.9688,
      "num_input_tokens_seen": 41776020,
      "step": 1983
    },
    {
      "epoch": 0.23856189502795647,
      "flos": 16612139827200.0,
      "grad_norm": 3.485335008235063,
      "learning_rate": 3.5609533492000463e-06,
      "loss": 1.0587,
      "num_input_tokens_seen": 41794630,
      "step": 1984
    },
    {
      "epoch": 0.23868213791859555,
      "flos": 16689301094400.0,
      "grad_norm": 2.9455718475214847,
      "learning_rate": 3.560466229502485e-06,
      "loss": 0.9783,
      "num_input_tokens_seen": 41813695,
      "step": 1985
    },
    {
      "epoch": 0.23880238080923466,
      "flos": 11761007124480.0,
      "grad_norm": 6.250051574494237,
      "learning_rate": 3.5599788730834384e-06,
      "loss": 1.0932,
      "num_input_tokens_seen": 41831375,
      "step": 1986
    },
    {
      "epoch": 0.23892262369987374,
      "flos": 12286595727360.0,
      "grad_norm": 3.7284420879419873,
      "learning_rate": 3.559491280016836e-06,
      "loss": 0.9891,
      "num_input_tokens_seen": 41849040,
      "step": 1987
    },
    {
      "epoch": 0.23904286659051283,
      "flos": 15852246835200.0,
      "grad_norm": 2.732153638716937,
      "learning_rate": 3.5590034503766465e-06,
      "loss": 0.9102,
      "num_input_tokens_seen": 41868425,
      "step": 1988
    },
    {
      "epoch": 0.23916310948115194,
      "flos": 15038869094400.0,
      "grad_norm": 3.846542460542563,
      "learning_rate": 3.558515384236874e-06,
      "loss": 1.0117,
      "num_input_tokens_seen": 41885575,
      "step": 1989
    },
    {
      "epoch": 0.23928335237179102,
      "flos": 9978825400320.0,
      "grad_norm": 2.343192377513184,
      "learning_rate": 3.558027081671556e-06,
      "loss": 1.0347,
      "num_input_tokens_seen": 41902280,
      "step": 1990
    },
    {
      "epoch": 0.2394035952624301,
      "flos": 16901339258880.0,
      "grad_norm": 3.3118728907791337,
      "learning_rate": 3.557538542754769e-06,
      "loss": 0.8873,
      "num_input_tokens_seen": 41921695,
      "step": 1991
    },
    {
      "epoch": 0.2395238381530692,
      "flos": 17214951014400.0,
      "grad_norm": 2.9938128566315276,
      "learning_rate": 3.557049767560623e-06,
      "loss": 0.8655,
      "num_input_tokens_seen": 41940330,
      "step": 1992
    },
    {
      "epoch": 0.2396440810437083,
      "flos": 17997324840960.0,
      "grad_norm": 2.9966481875768167,
      "learning_rate": 3.5565607561632655e-06,
      "loss": 1.0551,
      "num_input_tokens_seen": 41958890,
      "step": 1993
    },
    {
      "epoch": 0.23976432393434738,
      "flos": 20331132395520.0,
      "grad_norm": 5.296062395898024,
      "learning_rate": 3.5560715086368787e-06,
      "loss": 0.986,
      "num_input_tokens_seen": 41976480,
      "step": 1994
    },
    {
      "epoch": 0.23988456682498646,
      "flos": 13827942789120.0,
      "grad_norm": 2.958370656239219,
      "learning_rate": 3.5555820250556816e-06,
      "loss": 1.0202,
      "num_input_tokens_seen": 41993400,
      "step": 1995
    },
    {
      "epoch": 0.24000480971562557,
      "flos": 14383523573760.0,
      "grad_norm": 3.2284506383356586,
      "learning_rate": 3.5550923054939278e-06,
      "loss": 0.8869,
      "num_input_tokens_seen": 42012575,
      "step": 1996
    },
    {
      "epoch": 0.24012505260626466,
      "flos": 18103466557440.0,
      "grad_norm": 3.014782881882232,
      "learning_rate": 3.5546023500259083e-06,
      "loss": 0.9387,
      "num_input_tokens_seen": 42033390,
      "step": 1997
    },
    {
      "epoch": 0.24024529549690374,
      "flos": 10995534274560.0,
      "grad_norm": 5.388488523067709,
      "learning_rate": 3.5541121587259477e-06,
      "loss": 1.0097,
      "num_input_tokens_seen": 42050945,
      "step": 1998
    },
    {
      "epoch": 0.24036553838754285,
      "flos": 40876900823040.0,
      "grad_norm": 0.8189026843093253,
      "learning_rate": 3.553621731668408e-06,
      "loss": 0.7852,
      "num_input_tokens_seen": 42113875,
      "step": 1999
    },
    {
      "epoch": 0.24048578127818193,
      "flos": 17763510988800.0,
      "grad_norm": 2.439465048572241,
      "learning_rate": 3.553131068927688e-06,
      "loss": 1.0379,
      "num_input_tokens_seen": 42132000,
      "step": 2000
    },
    {
      "epoch": 0.24060602416882101,
      "flos": 16585979965440.0,
      "grad_norm": 3.071098557038816,
      "learning_rate": 3.552640170578219e-06,
      "loss": 1.0055,
      "num_input_tokens_seen": 42151970,
      "step": 2001
    },
    {
      "epoch": 0.2407262670594601,
      "flos": 10004556042240.0,
      "grad_norm": 4.052856502301441,
      "learning_rate": 3.5521490366944703e-06,
      "loss": 0.9831,
      "num_input_tokens_seen": 42169340,
      "step": 2002
    },
    {
      "epoch": 0.2408465099500992,
      "flos": 9638011392000.0,
      "grad_norm": 4.168370750368112,
      "learning_rate": 3.5516576673509474e-06,
      "loss": 1.0014,
      "num_input_tokens_seen": 42187060,
      "step": 2003
    },
    {
      "epoch": 0.2409667528407383,
      "flos": 22275791831040.0,
      "grad_norm": 2.339167782255839,
      "learning_rate": 3.5511660626221896e-06,
      "loss": 1.0595,
      "num_input_tokens_seen": 42207420,
      "step": 2004
    },
    {
      "epoch": 0.24108699573137737,
      "flos": 15773154078720.0,
      "grad_norm": 3.209290316317865,
      "learning_rate": 3.5506742225827744e-06,
      "loss": 1.0708,
      "num_input_tokens_seen": 42223995,
      "step": 2005
    },
    {
      "epoch": 0.24120723862201648,
      "flos": 18578115072000.0,
      "grad_norm": 2.9595871717372706,
      "learning_rate": 3.5501821473073116e-06,
      "loss": 1.0975,
      "num_input_tokens_seen": 42240300,
      "step": 2006
    },
    {
      "epoch": 0.24132748151265557,
      "flos": 13203938426880.0,
      "grad_norm": 2.7889004187411026,
      "learning_rate": 3.54968983687045e-06,
      "loss": 1.0667,
      "num_input_tokens_seen": 42256890,
      "step": 2007
    },
    {
      "epoch": 0.24144772440329465,
      "flos": 10791068774400.0,
      "grad_norm": 5.081258423295025,
      "learning_rate": 3.549197291346872e-06,
      "loss": 1.0863,
      "num_input_tokens_seen": 42273135,
      "step": 2008
    },
    {
      "epoch": 0.24156796729393373,
      "flos": 17084274339840.0,
      "grad_norm": 3.2923022915996736,
      "learning_rate": 3.548704510811297e-06,
      "loss": 0.994,
      "num_input_tokens_seen": 42292050,
      "step": 2009
    },
    {
      "epoch": 0.24168821018457284,
      "flos": 18710355333120.0,
      "grad_norm": 3.3028537721894846,
      "learning_rate": 3.5482114953384787e-06,
      "loss": 0.9402,
      "num_input_tokens_seen": 42311000,
      "step": 2010
    },
    {
      "epoch": 0.24180845307521193,
      "flos": 12915168215040.0,
      "grad_norm": 5.186343328372214,
      "learning_rate": 3.5477182450032077e-06,
      "loss": 1.0336,
      "num_input_tokens_seen": 42329320,
      "step": 2011
    },
    {
      "epoch": 0.241928695965851,
      "flos": 14515058688000.0,
      "grad_norm": 3.0166962455296966,
      "learning_rate": 3.5472247598803097e-06,
      "loss": 1.0271,
      "num_input_tokens_seen": 42348385,
      "step": 2012
    },
    {
      "epoch": 0.24204893885649012,
      "flos": 18185410560000.0,
      "grad_norm": 3.594462713804072,
      "learning_rate": 3.546731040044645e-06,
      "loss": 1.0515,
      "num_input_tokens_seen": 42363275,
      "step": 2013
    },
    {
      "epoch": 0.2421691817471292,
      "flos": 21852788551680.0,
      "grad_norm": 3.3050034825109003,
      "learning_rate": 3.546237085571112e-06,
      "loss": 0.9533,
      "num_input_tokens_seen": 42381430,
      "step": 2014
    },
    {
      "epoch": 0.24228942463776829,
      "flos": 15590004387840.0,
      "grad_norm": 5.124398917328355,
      "learning_rate": 3.5457428965346425e-06,
      "loss": 0.9154,
      "num_input_tokens_seen": 42400090,
      "step": 2015
    },
    {
      "epoch": 0.2424096675284074,
      "flos": 24242441564160.0,
      "grad_norm": 3.0179206925292923,
      "learning_rate": 3.545248473010205e-06,
      "loss": 0.9519,
      "num_input_tokens_seen": 42422615,
      "step": 2016
    },
    {
      "epoch": 0.24252991041904648,
      "flos": 15380480225280.0,
      "grad_norm": 5.126419987161179,
      "learning_rate": 3.544753815072802e-06,
      "loss": 1.0656,
      "num_input_tokens_seen": 42440990,
      "step": 2017
    },
    {
      "epoch": 0.24265015330968556,
      "flos": 15535875809280.0,
      "grad_norm": 12.133544428180292,
      "learning_rate": 3.544258922797474e-06,
      "loss": 1.0938,
      "num_input_tokens_seen": 42458830,
      "step": 2018
    },
    {
      "epoch": 0.24277039620032465,
      "flos": 18236841185280.0,
      "grad_norm": 8.336596905298096,
      "learning_rate": 3.543763796259295e-06,
      "loss": 0.9815,
      "num_input_tokens_seen": 42478505,
      "step": 2019
    },
    {
      "epoch": 0.24289063909096376,
      "flos": 18710171381760.0,
      "grad_norm": 3.47579639008805,
      "learning_rate": 3.5432684355333754e-06,
      "loss": 1.1081,
      "num_input_tokens_seen": 42496880,
      "step": 2020
    },
    {
      "epoch": 0.24301088198160284,
      "flos": 17838464839680.0,
      "grad_norm": 3.0066706880541156,
      "learning_rate": 3.5427728406948613e-06,
      "loss": 0.9589,
      "num_input_tokens_seen": 42515715,
      "step": 2021
    },
    {
      "epoch": 0.24313112487224192,
      "flos": 50061794119680.0,
      "grad_norm": 0.7779840870311002,
      "learning_rate": 3.542277011818934e-06,
      "loss": 0.7993,
      "num_input_tokens_seen": 42579270,
      "step": 2022
    },
    {
      "epoch": 0.24325136776288103,
      "flos": 29041806397440.0,
      "grad_norm": 3.2726535632281135,
      "learning_rate": 3.5417809489808104e-06,
      "loss": 0.9327,
      "num_input_tokens_seen": 42600600,
      "step": 2023
    },
    {
      "epoch": 0.24337161065352012,
      "flos": 17818927226880.0,
      "grad_norm": 2.8260017981823857,
      "learning_rate": 3.5412846522557422e-06,
      "loss": 0.9177,
      "num_input_tokens_seen": 42621210,
      "step": 2024
    },
    {
      "epoch": 0.2434918535441592,
      "flos": 13229944995840.0,
      "grad_norm": 6.110790976413927,
      "learning_rate": 3.540788121719018e-06,
      "loss": 0.9405,
      "num_input_tokens_seen": 42639350,
      "step": 2025
    },
    {
      "epoch": 0.24361209643479828,
      "flos": 17005886730240.0,
      "grad_norm": 2.513309242156684,
      "learning_rate": 3.5402913574459604e-06,
      "loss": 1.0258,
      "num_input_tokens_seen": 42658975,
      "step": 2026
    },
    {
      "epoch": 0.2437323393254374,
      "flos": 20414026813440.0,
      "grad_norm": 2.3426132497210364,
      "learning_rate": 3.5397943595119297e-06,
      "loss": 1.0479,
      "num_input_tokens_seen": 42680115,
      "step": 2027
    },
    {
      "epoch": 0.24385258221607647,
      "flos": 16743705600000.0,
      "grad_norm": 3.447754701496244,
      "learning_rate": 3.5392971279923177e-06,
      "loss": 0.9526,
      "num_input_tokens_seen": 42698055,
      "step": 2028
    },
    {
      "epoch": 0.24397282510671556,
      "flos": 18026826485760.0,
      "grad_norm": 4.083095984794494,
      "learning_rate": 3.5387996629625557e-06,
      "loss": 1.0247,
      "num_input_tokens_seen": 42715365,
      "step": 2029
    },
    {
      "epoch": 0.24409306799735467,
      "flos": 46673344942080.0,
      "grad_norm": 0.8522111015003182,
      "learning_rate": 3.5383019644981083e-06,
      "loss": 0.7936,
      "num_input_tokens_seen": 42778780,
      "step": 2030
    },
    {
      "epoch": 0.24421331088799375,
      "flos": 13859498557440.0,
      "grad_norm": 3.5021905855771296,
      "learning_rate": 3.5378040326744763e-06,
      "loss": 0.9195,
      "num_input_tokens_seen": 42797985,
      "step": 2031
    },
    {
      "epoch": 0.24433355377863283,
      "flos": 14960021606400.0,
      "grad_norm": 2.7788886000249953,
      "learning_rate": 3.5373058675671946e-06,
      "loss": 1.0593,
      "num_input_tokens_seen": 42815710,
      "step": 2032
    },
    {
      "epoch": 0.24445379666927192,
      "flos": 16087317688320.0,
      "grad_norm": 4.009437870950943,
      "learning_rate": 3.536807469251836e-06,
      "loss": 0.9196,
      "num_input_tokens_seen": 42834585,
      "step": 2033
    },
    {
      "epoch": 0.24457403955991103,
      "flos": 15091648696320.0,
      "grad_norm": 3.1916614369747123,
      "learning_rate": 3.5363088378040055e-06,
      "loss": 1.0187,
      "num_input_tokens_seen": 42853195,
      "step": 2034
    },
    {
      "epoch": 0.2446942824505501,
      "flos": 47972569681920.0,
      "grad_norm": 0.8416027013311075,
      "learning_rate": 3.5358099732993463e-06,
      "loss": 0.8772,
      "num_input_tokens_seen": 42912025,
      "step": 2035
    },
    {
      "epoch": 0.2448145253411892,
      "flos": 14487427215360.0,
      "grad_norm": 3.8580644984887837,
      "learning_rate": 3.535310875813535e-06,
      "loss": 1.0779,
      "num_input_tokens_seen": 42930140,
      "step": 2036
    },
    {
      "epoch": 0.2449347682318283,
      "flos": 20520015237120.0,
      "grad_norm": 3.2495370681885962,
      "learning_rate": 3.5348115454222843e-06,
      "loss": 1.0158,
      "num_input_tokens_seen": 42952445,
      "step": 2037
    },
    {
      "epoch": 0.2450550111224674,
      "flos": 16009727201280.0,
      "grad_norm": 3.040436813945043,
      "learning_rate": 3.5343119822013425e-06,
      "loss": 1.0458,
      "num_input_tokens_seen": 42971275,
      "step": 2038
    },
    {
      "epoch": 0.24517525401310647,
      "flos": 15455955271680.0,
      "grad_norm": 3.7797614075782313,
      "learning_rate": 3.533812186226493e-06,
      "loss": 0.9781,
      "num_input_tokens_seen": 42991705,
      "step": 2039
    },
    {
      "epoch": 0.24529549690374555,
      "flos": 17816290590720.0,
      "grad_norm": 11.89138851646894,
      "learning_rate": 3.5333121575735545e-06,
      "loss": 0.9624,
      "num_input_tokens_seen": 43011065,
      "step": 2040
    },
    {
      "epoch": 0.24541573979438466,
      "flos": 22905069465600.0,
      "grad_norm": 3.26195381825101,
      "learning_rate": 3.532811896318381e-06,
      "loss": 0.9507,
      "num_input_tokens_seen": 43032855,
      "step": 2041
    },
    {
      "epoch": 0.24553598268502375,
      "flos": 22354516684800.0,
      "grad_norm": 2.9969423924924645,
      "learning_rate": 3.5323114025368615e-06,
      "loss": 1.0004,
      "num_input_tokens_seen": 43047640,
      "step": 2042
    },
    {
      "epoch": 0.24565622557566283,
      "flos": 9899793960960.0,
      "grad_norm": 2.638582331156683,
      "learning_rate": 3.53181067630492e-06,
      "loss": 1.0125,
      "num_input_tokens_seen": 43064830,
      "step": 2043
    },
    {
      "epoch": 0.24577646846630194,
      "flos": 11735307141120.0,
      "grad_norm": 3.276488709789927,
      "learning_rate": 3.5313097176985175e-06,
      "loss": 0.9544,
      "num_input_tokens_seen": 43082860,
      "step": 2044
    },
    {
      "epoch": 0.24589671135694102,
      "flos": 13335105638400.0,
      "grad_norm": 2.3373095715314665,
      "learning_rate": 3.5308085267936482e-06,
      "loss": 1.0044,
      "num_input_tokens_seen": 43100295,
      "step": 2045
    },
    {
      "epoch": 0.2460169542475801,
      "flos": 14148023500800.0,
      "grad_norm": 2.2772930787405237,
      "learning_rate": 3.530307103666342e-06,
      "loss": 1.1029,
      "num_input_tokens_seen": 43119095,
      "step": 2046
    },
    {
      "epoch": 0.24613719713821922,
      "flos": 17190017495040.0,
      "grad_norm": 2.4471231892533214,
      "learning_rate": 3.5298054483926658e-06,
      "loss": 0.9977,
      "num_input_tokens_seen": 43139510,
      "step": 2047
    },
    {
      "epoch": 0.2462574400288583,
      "flos": 21537000038400.0,
      "grad_norm": 6.174163934733897,
      "learning_rate": 3.5293035610487187e-06,
      "loss": 1.0296,
      "num_input_tokens_seen": 43158595,
      "step": 2048
    },
    {
      "epoch": 0.24637768291949738,
      "flos": 49372102901760.0,
      "grad_norm": 0.7514041630745105,
      "learning_rate": 3.5288014417106374e-06,
      "loss": 0.839,
      "num_input_tokens_seen": 43224335,
      "step": 2049
    },
    {
      "epoch": 0.24649792581013646,
      "flos": 24529310945280.0,
      "grad_norm": 2.84810497793393,
      "learning_rate": 3.528299090454593e-06,
      "loss": 0.9572,
      "num_input_tokens_seen": 43244590,
      "step": 2050
    },
    {
      "epoch": 0.24661816870077558,
      "flos": 13962175856640.0,
      "grad_norm": 3.3081038471513553,
      "learning_rate": 3.527796507356792e-06,
      "loss": 1.0277,
      "num_input_tokens_seen": 43258200,
      "step": 2051
    },
    {
      "epoch": 0.24673841159141466,
      "flos": 14192586608640.0,
      "grad_norm": 3.7304022818031344,
      "learning_rate": 3.527293692493475e-06,
      "loss": 1.0797,
      "num_input_tokens_seen": 43273785,
      "step": 2052
    },
    {
      "epoch": 0.24685865448205374,
      "flos": 15374655098880.0,
      "grad_norm": 6.759282264302596,
      "learning_rate": 3.52679064594092e-06,
      "loss": 0.9301,
      "num_input_tokens_seen": 43290845,
      "step": 2053
    },
    {
      "epoch": 0.24697889737269285,
      "flos": 12725365616640.0,
      "grad_norm": 2.518604670040091,
      "learning_rate": 3.5262873677754375e-06,
      "loss": 0.9345,
      "num_input_tokens_seen": 43308570,
      "step": 2054
    },
    {
      "epoch": 0.24709914026333193,
      "flos": 19470554910720.0,
      "grad_norm": 3.161982702836483,
      "learning_rate": 3.5257838580733745e-06,
      "loss": 1.0042,
      "num_input_tokens_seen": 43327895,
      "step": 2055
    },
    {
      "epoch": 0.24721938315397102,
      "flos": 13670677032960.0,
      "grad_norm": 3.2168847762826998,
      "learning_rate": 3.5252801169111138e-06,
      "loss": 1.0581,
      "num_input_tokens_seen": 43345280,
      "step": 2056
    },
    {
      "epoch": 0.2473396260446101,
      "flos": 16481769738240.0,
      "grad_norm": 4.238899994437448,
      "learning_rate": 3.524776144365072e-06,
      "loss": 0.9906,
      "num_input_tokens_seen": 43363455,
      "step": 2057
    },
    {
      "epoch": 0.2474598689352492,
      "flos": 15013230428160.0,
      "grad_norm": 2.6388930560519683,
      "learning_rate": 3.5242719405117016e-06,
      "loss": 0.999,
      "num_input_tokens_seen": 43382980,
      "step": 2058
    },
    {
      "epoch": 0.2475801118258883,
      "flos": 15376157368320.0,
      "grad_norm": 3.122662700143451,
      "learning_rate": 3.5237675054274893e-06,
      "loss": 0.9293,
      "num_input_tokens_seen": 43401900,
      "step": 2059
    },
    {
      "epoch": 0.24770035471652738,
      "flos": 16114244014080.0,
      "grad_norm": 4.036497736912848,
      "learning_rate": 3.5232628391889584e-06,
      "loss": 1.0012,
      "num_input_tokens_seen": 43419910,
      "step": 2060
    },
    {
      "epoch": 0.2478205976071665,
      "flos": 15747239485440.0,
      "grad_norm": 4.037631875355736,
      "learning_rate": 3.522757941872666e-06,
      "loss": 0.8468,
      "num_input_tokens_seen": 43437785,
      "step": 2061
    },
    {
      "epoch": 0.24794084049780557,
      "flos": 17766208942080.0,
      "grad_norm": 2.2868969148355753,
      "learning_rate": 3.5222528135552042e-06,
      "loss": 1.0259,
      "num_input_tokens_seen": 43458965,
      "step": 2062
    },
    {
      "epoch": 0.24806108338844465,
      "flos": 12967794524160.0,
      "grad_norm": 3.383538218967155,
      "learning_rate": 3.521747454313201e-06,
      "loss": 1.0037,
      "num_input_tokens_seen": 43477365,
      "step": 2063
    },
    {
      "epoch": 0.24818132627908374,
      "flos": 13676348866560.0,
      "grad_norm": 4.030538333499943,
      "learning_rate": 3.521241864223319e-06,
      "loss": 0.8669,
      "num_input_tokens_seen": 43496045,
      "step": 2064
    },
    {
      "epoch": 0.24830156916972285,
      "flos": 50336185466880.0,
      "grad_norm": 0.8389098897063857,
      "learning_rate": 3.5207360433622552e-06,
      "loss": 0.8225,
      "num_input_tokens_seen": 43557765,
      "step": 2065
    },
    {
      "epoch": 0.24842181206036193,
      "flos": 28860128317440.0,
      "grad_norm": 4.382075011411791,
      "learning_rate": 3.5202299918067437e-06,
      "loss": 0.9457,
      "num_input_tokens_seen": 43581080,
      "step": 2066
    },
    {
      "epoch": 0.248542054951001,
      "flos": 14251160678400.0,
      "grad_norm": 6.483722552831039,
      "learning_rate": 3.519723709633551e-06,
      "loss": 0.8895,
      "num_input_tokens_seen": 43599560,
      "step": 2067
    },
    {
      "epoch": 0.24866229784164012,
      "flos": 16717576396800.0,
      "grad_norm": 3.1906665705887933,
      "learning_rate": 3.519217196919479e-06,
      "loss": 1.0273,
      "num_input_tokens_seen": 43618265,
      "step": 2068
    },
    {
      "epoch": 0.2487825407322792,
      "flos": 14095182581760.0,
      "grad_norm": 2.582473324603768,
      "learning_rate": 3.518710453741367e-06,
      "loss": 0.9271,
      "num_input_tokens_seen": 43637185,
      "step": 2069
    },
    {
      "epoch": 0.2489027836229183,
      "flos": 15642140160000.0,
      "grad_norm": 5.672151416947307,
      "learning_rate": 3.518203480176086e-06,
      "loss": 0.8735,
      "num_input_tokens_seen": 43655835,
      "step": 2070
    },
    {
      "epoch": 0.2490230265135574,
      "flos": 16559666810880.0,
      "grad_norm": 2.068371247715865,
      "learning_rate": 3.517696276300545e-06,
      "loss": 0.9915,
      "num_input_tokens_seen": 43677095,
      "step": 2071
    },
    {
      "epoch": 0.24914326940419648,
      "flos": 14067397816320.0,
      "grad_norm": 4.910856769030493,
      "learning_rate": 3.517188842191685e-06,
      "loss": 0.8977,
      "num_input_tokens_seen": 43694965,
      "step": 2072
    },
    {
      "epoch": 0.24926351229483557,
      "flos": 14356811857920.0,
      "grad_norm": 2.171802327429057,
      "learning_rate": 3.5166811779264837e-06,
      "loss": 0.9381,
      "num_input_tokens_seen": 43715005,
      "step": 2073
    },
    {
      "epoch": 0.24938375518547465,
      "flos": 16559850762240.0,
      "grad_norm": 6.471249010138669,
      "learning_rate": 3.5161732835819545e-06,
      "loss": 0.9726,
      "num_input_tokens_seen": 43734035,
      "step": 2074
    },
    {
      "epoch": 0.24950399807611376,
      "flos": 12260159938560.0,
      "grad_norm": 2.6189187771618365,
      "learning_rate": 3.515665159235143e-06,
      "loss": 1.0249,
      "num_input_tokens_seen": 43752640,
      "step": 2075
    },
    {
      "epoch": 0.24962424096675284,
      "flos": 13491175710720.0,
      "grad_norm": 2.7828915498569793,
      "learning_rate": 3.5151568049631318e-06,
      "loss": 0.9427,
      "num_input_tokens_seen": 43771075,
      "step": 2076
    },
    {
      "epoch": 0.24974448385739192,
      "flos": 23821247139840.0,
      "grad_norm": 5.05577978864193,
      "learning_rate": 3.5146482208430385e-06,
      "loss": 0.9957,
      "num_input_tokens_seen": 43792625,
      "step": 2077
    },
    {
      "epoch": 0.24986472674803104,
      "flos": 21384057139200.0,
      "grad_norm": 6.60064654795833,
      "learning_rate": 3.514139406952014e-06,
      "loss": 0.8744,
      "num_input_tokens_seen": 43814370,
      "step": 2078
    },
    {
      "epoch": 0.24998496963867012,
      "flos": 18944567746560.0,
      "grad_norm": 2.5940522167212,
      "learning_rate": 3.5136303633672454e-06,
      "loss": 1.0338,
      "num_input_tokens_seen": 43834220,
      "step": 2079
    },
    {
      "epoch": 0.25010521252930923,
      "flos": 16746832773120.0,
      "grad_norm": 3.15006473414193,
      "learning_rate": 3.5131210901659544e-06,
      "loss": 0.9494,
      "num_input_tokens_seen": 43855695,
      "step": 2080
    },
    {
      "epoch": 0.2502254554199483,
      "flos": 16665317990400.0,
      "grad_norm": 3.705927693599867,
      "learning_rate": 3.5126115874253967e-06,
      "loss": 1.0158,
      "num_input_tokens_seen": 43874970,
      "step": 2081
    },
    {
      "epoch": 0.2503456983105874,
      "flos": 20488183541760.0,
      "grad_norm": 2.2366425005900155,
      "learning_rate": 3.5121018552228644e-06,
      "loss": 0.9987,
      "num_input_tokens_seen": 43893195,
      "step": 2082
    },
    {
      "epoch": 0.2504659412012265,
      "flos": 13308301946880.0,
      "grad_norm": 3.188892135293558,
      "learning_rate": 3.5115918936356827e-06,
      "loss": 0.9748,
      "num_input_tokens_seen": 43909670,
      "step": 2083
    },
    {
      "epoch": 0.25058618409186556,
      "flos": 11945107230720.0,
      "grad_norm": 2.5374669152719713,
      "learning_rate": 3.5110817027412123e-06,
      "loss": 0.9848,
      "num_input_tokens_seen": 43928480,
      "step": 2084
    },
    {
      "epoch": 0.25070642698250467,
      "flos": 17372645990400.0,
      "grad_norm": 2.9032657847847916,
      "learning_rate": 3.5105712826168493e-06,
      "loss": 0.8787,
      "num_input_tokens_seen": 43947850,
      "step": 2085
    },
    {
      "epoch": 0.2508266698731437,
      "flos": 14378924789760.0,
      "grad_norm": 2.672527300531956,
      "learning_rate": 3.5100606333400235e-06,
      "loss": 0.9041,
      "num_input_tokens_seen": 43964705,
      "step": 2086
    },
    {
      "epoch": 0.25094691276378284,
      "flos": 13649821102080.0,
      "grad_norm": 5.061973261804398,
      "learning_rate": 3.5095497549882006e-06,
      "loss": 0.9761,
      "num_input_tokens_seen": 43982870,
      "step": 2087
    },
    {
      "epoch": 0.25106715565442195,
      "flos": 19182091284480.0,
      "grad_norm": 3.4952732593855775,
      "learning_rate": 3.50903864763888e-06,
      "loss": 0.9119,
      "num_input_tokens_seen": 44003380,
      "step": 2088
    },
    {
      "epoch": 0.251187398545061,
      "flos": 34573494067200.0,
      "grad_norm": 4.55256155660575,
      "learning_rate": 3.5085273113695965e-06,
      "loss": 0.9535,
      "num_input_tokens_seen": 44027670,
      "step": 2089
    },
    {
      "epoch": 0.2513076414357001,
      "flos": 19234441666560.0,
      "grad_norm": 11.521330580692899,
      "learning_rate": 3.508015746257919e-06,
      "loss": 0.9816,
      "num_input_tokens_seen": 44046430,
      "step": 2090
    },
    {
      "epoch": 0.2514278843263392,
      "flos": 13806626979840.0,
      "grad_norm": 4.483891158635026,
      "learning_rate": 3.5075039523814518e-06,
      "loss": 1.0272,
      "num_input_tokens_seen": 44065340,
      "step": 2091
    },
    {
      "epoch": 0.2515481272169783,
      "flos": 11938883543040.0,
      "grad_norm": 4.179652818197232,
      "learning_rate": 3.506991929817834e-06,
      "loss": 1.0115,
      "num_input_tokens_seen": 44081780,
      "step": 2092
    },
    {
      "epoch": 0.2516683701076174,
      "flos": 16874566225920.0,
      "grad_norm": 2.322454934067907,
      "learning_rate": 3.506479678644738e-06,
      "loss": 1.016,
      "num_input_tokens_seen": 44101895,
      "step": 2093
    },
    {
      "epoch": 0.2517886129982565,
      "flos": 19679864463360.0,
      "grad_norm": 3.4575269440699556,
      "learning_rate": 3.505967198939873e-06,
      "loss": 0.9331,
      "num_input_tokens_seen": 44118655,
      "step": 2094
    },
    {
      "epoch": 0.25190885588889556,
      "flos": 27203104727040.0,
      "grad_norm": 2.5528733464467144,
      "learning_rate": 3.5054544907809813e-06,
      "loss": 0.9706,
      "num_input_tokens_seen": 44138875,
      "step": 2095
    },
    {
      "epoch": 0.25202909877953467,
      "flos": 15823021117440.0,
      "grad_norm": 3.8724011966520244,
      "learning_rate": 3.50494155424584e-06,
      "loss": 0.9984,
      "num_input_tokens_seen": 44157500,
      "step": 2096
    },
    {
      "epoch": 0.2521493416701738,
      "flos": 15458193346560.0,
      "grad_norm": 3.079562708292234,
      "learning_rate": 3.504428389412262e-06,
      "loss": 1.0296,
      "num_input_tokens_seen": 44178030,
      "step": 2097
    },
    {
      "epoch": 0.25226958456081283,
      "flos": 19759662366720.0,
      "grad_norm": 4.046620482464542,
      "learning_rate": 3.5039149963580927e-06,
      "loss": 0.9321,
      "num_input_tokens_seen": 44197770,
      "step": 2098
    },
    {
      "epoch": 0.25238982745145194,
      "flos": 21904372469760.0,
      "grad_norm": 3.304463119645041,
      "learning_rate": 3.503401375161215e-06,
      "loss": 0.8913,
      "num_input_tokens_seen": 44217235,
      "step": 2099
    },
    {
      "epoch": 0.252510070342091,
      "flos": 14383523573760.0,
      "grad_norm": 6.760409770035324,
      "learning_rate": 3.502887525899544e-06,
      "loss": 1.0437,
      "num_input_tokens_seen": 44235935,
      "step": 2100
    },
    {
      "epoch": 0.2526303132327301,
      "flos": 16166625054720.0,
      "grad_norm": 2.501569731950284,
      "learning_rate": 3.50237344865103e-06,
      "loss": 1.0275,
      "num_input_tokens_seen": 44256655,
      "step": 2101
    },
    {
      "epoch": 0.2527505561233692,
      "flos": 21568279879680.0,
      "grad_norm": 5.374818310882704,
      "learning_rate": 3.501859143493658e-06,
      "loss": 0.9618,
      "num_input_tokens_seen": 44277005,
      "step": 2102
    },
    {
      "epoch": 0.2528707990140083,
      "flos": 41859754536960.0,
      "grad_norm": 0.8951219499075524,
      "learning_rate": 3.5013446105054488e-06,
      "loss": 0.8427,
      "num_input_tokens_seen": 44329645,
      "step": 2103
    },
    {
      "epoch": 0.2529910419046474,
      "flos": 17530555576320.0,
      "grad_norm": 2.4068290707255673,
      "learning_rate": 3.5008298497644555e-06,
      "loss": 0.9516,
      "num_input_tokens_seen": 44348410,
      "step": 2104
    },
    {
      "epoch": 0.2531112847952865,
      "flos": 16953413713920.0,
      "grad_norm": 2.836741435959101,
      "learning_rate": 3.500314861348767e-06,
      "loss": 1.082,
      "num_input_tokens_seen": 44368765,
      "step": 2105
    },
    {
      "epoch": 0.25323152768592555,
      "flos": 11420806287360.0,
      "grad_norm": 3.271405082994311,
      "learning_rate": 3.499799645336507e-06,
      "loss": 0.9693,
      "num_input_tokens_seen": 44385380,
      "step": 2106
    },
    {
      "epoch": 0.25335177057656466,
      "flos": 20233084538880.0,
      "grad_norm": 2.103364306774492,
      "learning_rate": 3.4992842018058336e-06,
      "loss": 1.0673,
      "num_input_tokens_seen": 44408000,
      "step": 2107
    },
    {
      "epoch": 0.25347201346720377,
      "flos": 13328483389440.0,
      "grad_norm": 3.1949258211267506,
      "learning_rate": 3.4987685308349384e-06,
      "loss": 1.0669,
      "num_input_tokens_seen": 44425450,
      "step": 2108
    },
    {
      "epoch": 0.2535922563578428,
      "flos": 11184846336000.0,
      "grad_norm": 6.626003261391493,
      "learning_rate": 3.4982526325020497e-06,
      "loss": 0.8182,
      "num_input_tokens_seen": 44442140,
      "step": 2109
    },
    {
      "epoch": 0.25371249924848194,
      "flos": 11545657835520.0,
      "grad_norm": 2.8962347909579895,
      "learning_rate": 3.4977365068854273e-06,
      "loss": 1.0167,
      "num_input_tokens_seen": 44457480,
      "step": 2110
    },
    {
      "epoch": 0.25383274213912105,
      "flos": 15458070712320.0,
      "grad_norm": 2.3888476378060637,
      "learning_rate": 3.4972201540633676e-06,
      "loss": 0.9232,
      "num_input_tokens_seen": 44476555,
      "step": 2111
    },
    {
      "epoch": 0.2539529850297601,
      "flos": 15195950899200.0,
      "grad_norm": 16.92456371471932,
      "learning_rate": 3.4967035741142008e-06,
      "loss": 1.0474,
      "num_input_tokens_seen": 44495095,
      "step": 2112
    },
    {
      "epoch": 0.2540732279203992,
      "flos": 17949726535680.0,
      "grad_norm": 4.090348089650419,
      "learning_rate": 3.4961867671162917e-06,
      "loss": 1.0158,
      "num_input_tokens_seen": 44514745,
      "step": 2113
    },
    {
      "epoch": 0.2541934708110383,
      "flos": 13780620410880.0,
      "grad_norm": 14.107697007951513,
      "learning_rate": 3.4956697331480402e-06,
      "loss": 0.9605,
      "num_input_tokens_seen": 44533035,
      "step": 2114
    },
    {
      "epoch": 0.2543137137016774,
      "flos": 17030237736960.0,
      "grad_norm": 9.804306205235312,
      "learning_rate": 3.495152472287879e-06,
      "loss": 0.9976,
      "num_input_tokens_seen": 44553465,
      "step": 2115
    },
    {
      "epoch": 0.2544339565923165,
      "flos": 18211907665920.0,
      "grad_norm": 5.081873804068105,
      "learning_rate": 3.4946349846142766e-06,
      "loss": 0.923,
      "num_input_tokens_seen": 44572325,
      "step": 2116
    },
    {
      "epoch": 0.25455419948295555,
      "flos": 15406333501440.0,
      "grad_norm": 4.5851551018858245,
      "learning_rate": 3.4941172702057353e-06,
      "loss": 0.9592,
      "num_input_tokens_seen": 44592105,
      "step": 2117
    },
    {
      "epoch": 0.25467444237359466,
      "flos": 18684164812800.0,
      "grad_norm": 8.48259748803649,
      "learning_rate": 3.4935993291407924e-06,
      "loss": 1.0033,
      "num_input_tokens_seen": 44610650,
      "step": 2118
    },
    {
      "epoch": 0.25479468526423377,
      "flos": 19208281804800.0,
      "grad_norm": 4.342742583549324,
      "learning_rate": 3.4930811614980183e-06,
      "loss": 0.9014,
      "num_input_tokens_seen": 44632065,
      "step": 2119
    },
    {
      "epoch": 0.2549149281548728,
      "flos": 16689944924160.0,
      "grad_norm": 2.804912759984789,
      "learning_rate": 3.4925627673560198e-06,
      "loss": 0.9902,
      "num_input_tokens_seen": 44652445,
      "step": 2120
    },
    {
      "epoch": 0.25503517104551193,
      "flos": 18369326714880.0,
      "grad_norm": 2.831830205687264,
      "learning_rate": 3.4920441467934357e-06,
      "loss": 1.0771,
      "num_input_tokens_seen": 44672680,
      "step": 2121
    },
    {
      "epoch": 0.25515541393615104,
      "flos": 18966987264000.0,
      "grad_norm": 3.127036079864668,
      "learning_rate": 3.491525299888941e-06,
      "loss": 1.0122,
      "num_input_tokens_seen": 44691245,
      "step": 2122
    },
    {
      "epoch": 0.2552756568267901,
      "flos": 47225982504960.0,
      "grad_norm": 1.0101569061395177,
      "learning_rate": 3.491006226721244e-06,
      "loss": 0.8818,
      "num_input_tokens_seen": 44755175,
      "step": 2123
    },
    {
      "epoch": 0.2553958997174292,
      "flos": 12705214832640.0,
      "grad_norm": 3.7621068704727674,
      "learning_rate": 3.4904869273690882e-06,
      "loss": 0.9657,
      "num_input_tokens_seen": 44772785,
      "step": 2124
    },
    {
      "epoch": 0.2555161426080683,
      "flos": 16612507729920.0,
      "grad_norm": 4.131629010011328,
      "learning_rate": 3.489967401911251e-06,
      "loss": 1.0891,
      "num_input_tokens_seen": 44791805,
      "step": 2125
    },
    {
      "epoch": 0.2556363854987074,
      "flos": 29014972047360.0,
      "grad_norm": 3.108787749536382,
      "learning_rate": 3.4894476504265428e-06,
      "loss": 0.8974,
      "num_input_tokens_seen": 44815765,
      "step": 2126
    },
    {
      "epoch": 0.2557566283893465,
      "flos": 48706793349120.0,
      "grad_norm": 0.7949725610568279,
      "learning_rate": 3.4889276729938104e-06,
      "loss": 0.7657,
      "num_input_tokens_seen": 44874015,
      "step": 2127
    },
    {
      "epoch": 0.2558768712799856,
      "flos": 16085999370240.0,
      "grad_norm": 3.789066401214164,
      "learning_rate": 3.488407469691934e-06,
      "loss": 0.9936,
      "num_input_tokens_seen": 44894430,
      "step": 2128
    },
    {
      "epoch": 0.25599711417062465,
      "flos": 18789110845440.0,
      "grad_norm": 3.918246836506035,
      "learning_rate": 3.487887040599828e-06,
      "loss": 1.0005,
      "num_input_tokens_seen": 44913950,
      "step": 2129
    },
    {
      "epoch": 0.25611735706126376,
      "flos": 16241180344320.0,
      "grad_norm": 4.6423044986676185,
      "learning_rate": 3.4873663857964407e-06,
      "loss": 0.9496,
      "num_input_tokens_seen": 44930885,
      "step": 2130
    },
    {
      "epoch": 0.2562375999519028,
      "flos": 16612569047040.0,
      "grad_norm": 3.9599398648869037,
      "learning_rate": 3.4868455053607556e-06,
      "loss": 0.8712,
      "num_input_tokens_seen": 44950220,
      "step": 2131
    },
    {
      "epoch": 0.2563578428425419,
      "flos": 16245380567040.0,
      "grad_norm": 33.656503887157534,
      "learning_rate": 3.486324399371789e-06,
      "loss": 0.9007,
      "num_input_tokens_seen": 44969240,
      "step": 2132
    },
    {
      "epoch": 0.25647808573318104,
      "flos": 15380878786560.0,
      "grad_norm": 4.604774584559962,
      "learning_rate": 3.485803067908593e-06,
      "loss": 0.976,
      "num_input_tokens_seen": 44988470,
      "step": 2133
    },
    {
      "epoch": 0.2565983286238201,
      "flos": 24060150312960.0,
      "grad_norm": 3.5899450974795535,
      "learning_rate": 3.485281511050253e-06,
      "loss": 0.9952,
      "num_input_tokens_seen": 45010325,
      "step": 2134
    },
    {
      "epoch": 0.2567185715144592,
      "flos": 11472574156800.0,
      "grad_norm": 5.908549189245065,
      "learning_rate": 3.484759728875889e-06,
      "loss": 1.0929,
      "num_input_tokens_seen": 45025410,
      "step": 2135
    },
    {
      "epoch": 0.2568388144050983,
      "flos": 12365780459520.0,
      "grad_norm": 3.101582793738709,
      "learning_rate": 3.4842377214646543e-06,
      "loss": 1.0057,
      "num_input_tokens_seen": 45043425,
      "step": 2136
    },
    {
      "epoch": 0.25695905729573737,
      "flos": 14830019420160.0,
      "grad_norm": 3.6066531532819472,
      "learning_rate": 3.483715488895737e-06,
      "loss": 0.8672,
      "num_input_tokens_seen": 45063475,
      "step": 2137
    },
    {
      "epoch": 0.2570793001863765,
      "flos": 17582292787200.0,
      "grad_norm": 3.3917369513450097,
      "learning_rate": 3.48319303124836e-06,
      "loss": 0.9772,
      "num_input_tokens_seen": 45083575,
      "step": 2138
    },
    {
      "epoch": 0.2571995430770156,
      "flos": 19155992739840.0,
      "grad_norm": 6.572565406888852,
      "learning_rate": 3.4826703486017798e-06,
      "loss": 0.8643,
      "num_input_tokens_seen": 45102920,
      "step": 2139
    },
    {
      "epoch": 0.25731978596765465,
      "flos": 14043108126720.0,
      "grad_norm": 8.214874862488633,
      "learning_rate": 3.4821474410352867e-06,
      "loss": 0.9627,
      "num_input_tokens_seen": 45121300,
      "step": 2140
    },
    {
      "epoch": 0.25744002885829376,
      "flos": 50535071109120.0,
      "grad_norm": 0.9794503200174699,
      "learning_rate": 3.481624308628205e-06,
      "loss": 0.8668,
      "num_input_tokens_seen": 45182390,
      "step": 2141
    },
    {
      "epoch": 0.25756027174893287,
      "flos": 12781854904320.0,
      "grad_norm": 6.716529088784495,
      "learning_rate": 3.481100951459893e-06,
      "loss": 1.1907,
      "num_input_tokens_seen": 45195130,
      "step": 2142
    },
    {
      "epoch": 0.2576805146395719,
      "flos": 16113876111360.0,
      "grad_norm": 3.6825199443124252,
      "learning_rate": 3.4805773696097453e-06,
      "loss": 0.9865,
      "num_input_tokens_seen": 45215740,
      "step": 2143
    },
    {
      "epoch": 0.25780075753021103,
      "flos": 11656827555840.0,
      "grad_norm": 3.0349685041373915,
      "learning_rate": 3.4800535631571874e-06,
      "loss": 1.073,
      "num_input_tokens_seen": 45230990,
      "step": 2144
    },
    {
      "epoch": 0.25792100042085014,
      "flos": 16219006095360.0,
      "grad_norm": 3.2031574526747786,
      "learning_rate": 3.4795295321816804e-06,
      "loss": 0.9669,
      "num_input_tokens_seen": 45249535,
      "step": 2145
    },
    {
      "epoch": 0.2580412433114892,
      "flos": 13255859589120.0,
      "grad_norm": 4.62355781958123,
      "learning_rate": 3.47900527676272e-06,
      "loss": 1.1179,
      "num_input_tokens_seen": 45267590,
      "step": 2146
    },
    {
      "epoch": 0.2581614862021283,
      "flos": 10083587481600.0,
      "grad_norm": 4.743155517319099,
      "learning_rate": 3.478480796979835e-06,
      "loss": 1.0842,
      "num_input_tokens_seen": 45285195,
      "step": 2147
    },
    {
      "epoch": 0.25828172909276736,
      "flos": 21018493562880.0,
      "grad_norm": 8.652545673433881,
      "learning_rate": 3.4779560929125894e-06,
      "loss": 0.9737,
      "num_input_tokens_seen": 45306460,
      "step": 2148
    },
    {
      "epoch": 0.2584019719834065,
      "flos": 48057303613440.0,
      "grad_norm": 0.8527377357014772,
      "learning_rate": 3.4774311646405783e-06,
      "loss": 0.7809,
      "num_input_tokens_seen": 45376085,
      "step": 2149
    },
    {
      "epoch": 0.2585222148740456,
      "flos": 16271693721600.0,
      "grad_norm": 3.2290604276511092,
      "learning_rate": 3.476906012243435e-06,
      "loss": 1.0317,
      "num_input_tokens_seen": 45394715,
      "step": 2150
    },
    {
      "epoch": 0.25864245776468464,
      "flos": 20594417233920.0,
      "grad_norm": 5.6449735580379174,
      "learning_rate": 3.476380635800824e-06,
      "loss": 1.0124,
      "num_input_tokens_seen": 45415635,
      "step": 2151
    },
    {
      "epoch": 0.25876270065532375,
      "flos": 10450040156160.0,
      "grad_norm": 8.708662206350434,
      "learning_rate": 3.475855035392444e-06,
      "loss": 1.0457,
      "num_input_tokens_seen": 45430675,
      "step": 2152
    },
    {
      "epoch": 0.25888294354596286,
      "flos": 43274786549760.0,
      "grad_norm": 3.416256965325824,
      "learning_rate": 3.475329211098029e-06,
      "loss": 0.8996,
      "num_input_tokens_seen": 45453550,
      "step": 2153
    },
    {
      "epoch": 0.2590031864366019,
      "flos": 19834033704960.0,
      "grad_norm": 5.700975823110756,
      "learning_rate": 3.4748031629973453e-06,
      "loss": 1.0181,
      "num_input_tokens_seen": 45474000,
      "step": 2154
    },
    {
      "epoch": 0.25912342932724103,
      "flos": 44687280230400.0,
      "grad_norm": 0.9200731869798519,
      "learning_rate": 3.4742768911701944e-06,
      "loss": 0.7671,
      "num_input_tokens_seen": 45536415,
      "step": 2155
    },
    {
      "epoch": 0.25924367221788014,
      "flos": 8714046443520.0,
      "grad_norm": 4.671185394730481,
      "learning_rate": 3.4737503956964113e-06,
      "loss": 0.8865,
      "num_input_tokens_seen": 45548440,
      "step": 2156
    },
    {
      "epoch": 0.2593639151085192,
      "flos": 10293510205440.0,
      "grad_norm": 10.310156723592472,
      "learning_rate": 3.473223676655865e-06,
      "loss": 0.8586,
      "num_input_tokens_seen": 45566160,
      "step": 2157
    },
    {
      "epoch": 0.2594841579991583,
      "flos": 10764632985600.0,
      "grad_norm": 3.3729429687329,
      "learning_rate": 3.472696734128459e-06,
      "loss": 0.9884,
      "num_input_tokens_seen": 45583745,
      "step": 2158
    },
    {
      "epoch": 0.2596044008897974,
      "flos": 16796393226240.0,
      "grad_norm": 3.804385696127871,
      "learning_rate": 3.4721695681941286e-06,
      "loss": 0.9445,
      "num_input_tokens_seen": 45602505,
      "step": 2159
    },
    {
      "epoch": 0.25972464378043647,
      "flos": 9716429660160.0,
      "grad_norm": 4.96538540146267,
      "learning_rate": 3.471642178932845e-06,
      "loss": 1.03,
      "num_input_tokens_seen": 45620870,
      "step": 2160
    },
    {
      "epoch": 0.2598448866710756,
      "flos": 13885719736320.0,
      "grad_norm": 8.679336334733536,
      "learning_rate": 3.471114566424613e-06,
      "loss": 1.1015,
      "num_input_tokens_seen": 45639050,
      "step": 2161
    },
    {
      "epoch": 0.25996512956171464,
      "flos": 15380326932480.0,
      "grad_norm": 6.6186354673164125,
      "learning_rate": 3.4705867307494715e-06,
      "loss": 0.9535,
      "num_input_tokens_seen": 45657840,
      "step": 2162
    },
    {
      "epoch": 0.26008537245235375,
      "flos": 12915413483520.0,
      "grad_norm": 4.665000731702218,
      "learning_rate": 3.470058671987492e-06,
      "loss": 1.0292,
      "num_input_tokens_seen": 45675825,
      "step": 2163
    },
    {
      "epoch": 0.26020561534299286,
      "flos": 17530770186240.0,
      "grad_norm": 11.469273526666464,
      "learning_rate": 3.4695303902187805e-06,
      "loss": 1.0314,
      "num_input_tokens_seen": 45695100,
      "step": 2164
    },
    {
      "epoch": 0.2603258582336319,
      "flos": 18341266022400.0,
      "grad_norm": 5.587890112995869,
      "learning_rate": 3.469001885523478e-06,
      "loss": 0.9908,
      "num_input_tokens_seen": 45715540,
      "step": 2165
    },
    {
      "epoch": 0.260446101124271,
      "flos": 20492138496000.0,
      "grad_norm": 3.388302759300603,
      "learning_rate": 3.4684731579817568e-06,
      "loss": 1.0103,
      "num_input_tokens_seen": 45736250,
      "step": 2166
    },
    {
      "epoch": 0.26056634401491013,
      "flos": 18264779243520.0,
      "grad_norm": 4.632795414241786,
      "learning_rate": 3.4679442076738247e-06,
      "loss": 0.9684,
      "num_input_tokens_seen": 45755685,
      "step": 2167
    },
    {
      "epoch": 0.2606865869055492,
      "flos": 19260816138240.0,
      "grad_norm": 12.112552506497392,
      "learning_rate": 3.4674150346799245e-06,
      "loss": 1.0358,
      "num_input_tokens_seen": 45775105,
      "step": 2168
    },
    {
      "epoch": 0.2608068297961883,
      "flos": 12548255662080.0,
      "grad_norm": 4.4466604324442,
      "learning_rate": 3.4668856390803295e-06,
      "loss": 0.9837,
      "num_input_tokens_seen": 45792705,
      "step": 2169
    },
    {
      "epoch": 0.2609270726868274,
      "flos": 13151220142080.0,
      "grad_norm": 4.008527725380789,
      "learning_rate": 3.4663560209553495e-06,
      "loss": 1.0941,
      "num_input_tokens_seen": 45810490,
      "step": 2170
    },
    {
      "epoch": 0.26104731557746647,
      "flos": 15511463485440.0,
      "grad_norm": 3.5462688986886826,
      "learning_rate": 3.4658261803853267e-06,
      "loss": 0.9713,
      "num_input_tokens_seen": 45828135,
      "step": 2171
    },
    {
      "epoch": 0.2611675584681056,
      "flos": 15406486794240.0,
      "grad_norm": 6.222104188768352,
      "learning_rate": 3.4652961174506383e-06,
      "loss": 1.0091,
      "num_input_tokens_seen": 45847725,
      "step": 2172
    },
    {
      "epoch": 0.2612878013587447,
      "flos": 51470011822080.0,
      "grad_norm": 1.0776303027108503,
      "learning_rate": 3.464765832231694e-06,
      "loss": 0.8038,
      "num_input_tokens_seen": 45901610,
      "step": 2173
    },
    {
      "epoch": 0.26140804424938374,
      "flos": 14355309588480.0,
      "grad_norm": 5.130028467526931,
      "learning_rate": 3.4642353248089373e-06,
      "loss": 0.9065,
      "num_input_tokens_seen": 45920090,
      "step": 2174
    },
    {
      "epoch": 0.26152828714002285,
      "flos": 18185931755520.0,
      "grad_norm": 3.962824325044519,
      "learning_rate": 3.463704595262846e-06,
      "loss": 0.9966,
      "num_input_tokens_seen": 45940690,
      "step": 2175
    },
    {
      "epoch": 0.26164853003066196,
      "flos": 18106317803520.0,
      "grad_norm": 4.519260513782168,
      "learning_rate": 3.463173643673931e-06,
      "loss": 0.8988,
      "num_input_tokens_seen": 45962935,
      "step": 2176
    },
    {
      "epoch": 0.261768772921301,
      "flos": 38592623063040.0,
      "grad_norm": 0.9553511839656608,
      "learning_rate": 3.4626424701227387e-06,
      "loss": 0.8581,
      "num_input_tokens_seen": 46017715,
      "step": 2177
    },
    {
      "epoch": 0.26188901581194013,
      "flos": 50624986337280.0,
      "grad_norm": 0.8781014042192202,
      "learning_rate": 3.4621110746898452e-06,
      "loss": 0.8131,
      "num_input_tokens_seen": 46085295,
      "step": 2178
    },
    {
      "epoch": 0.2620092587025792,
      "flos": 14960880046080.0,
      "grad_norm": 9.906877660271471,
      "learning_rate": 3.4615794574558654e-06,
      "loss": 0.9414,
      "num_input_tokens_seen": 46104025,
      "step": 2179
    },
    {
      "epoch": 0.2621295015932183,
      "flos": 13020451491840.0,
      "grad_norm": 17.13150680444149,
      "learning_rate": 3.4610476185014436e-06,
      "loss": 1.0231,
      "num_input_tokens_seen": 46121005,
      "step": 2180
    },
    {
      "epoch": 0.2622497444838574,
      "flos": 16822553088000.0,
      "grad_norm": 14.067208955505409,
      "learning_rate": 3.4605155579072597e-06,
      "loss": 0.9923,
      "num_input_tokens_seen": 46140580,
      "step": 2181
    },
    {
      "epoch": 0.26236998737449646,
      "flos": 15718013767680.0,
      "grad_norm": 4.0590430002803615,
      "learning_rate": 3.459983275754027e-06,
      "loss": 0.9146,
      "num_input_tokens_seen": 46159195,
      "step": 2182
    },
    {
      "epoch": 0.26249023026513557,
      "flos": 12679484190720.0,
      "grad_norm": 53.56498180145892,
      "learning_rate": 3.4594507721224918e-06,
      "loss": 0.9929,
      "num_input_tokens_seen": 46177565,
      "step": 2183
    },
    {
      "epoch": 0.2626104731557747,
      "flos": 12994077020160.0,
      "grad_norm": 5.192767123466983,
      "learning_rate": 3.4589180470934353e-06,
      "loss": 1.0089,
      "num_input_tokens_seen": 46197150,
      "step": 2184
    },
    {
      "epoch": 0.26273071604641374,
      "flos": 13701772922880.0,
      "grad_norm": 6.809831289030485,
      "learning_rate": 3.4583851007476713e-06,
      "loss": 0.9576,
      "num_input_tokens_seen": 46215340,
      "step": 2185
    },
    {
      "epoch": 0.26285095893705285,
      "flos": 12989416919040.0,
      "grad_norm": 5.32086151920143,
      "learning_rate": 3.4578519331660464e-06,
      "loss": 0.8808,
      "num_input_tokens_seen": 46232055,
      "step": 2186
    },
    {
      "epoch": 0.26297120182769196,
      "flos": 14331479777280.0,
      "grad_norm": 5.443112688391211,
      "learning_rate": 3.4573185444294426e-06,
      "loss": 1.0148,
      "num_input_tokens_seen": 46250140,
      "step": 2187
    },
    {
      "epoch": 0.263091444718331,
      "flos": 15930113249280.0,
      "grad_norm": 3.46149958401228,
      "learning_rate": 3.456784934618774e-06,
      "loss": 0.9789,
      "num_input_tokens_seen": 46271025,
      "step": 2188
    },
    {
      "epoch": 0.2632116876089701,
      "flos": 13491145052160.0,
      "grad_norm": 8.142215273232713,
      "learning_rate": 3.4562511038149897e-06,
      "loss": 0.9908,
      "num_input_tokens_seen": 46286240,
      "step": 2189
    },
    {
      "epoch": 0.26333193049960923,
      "flos": 48198005637120.0,
      "grad_norm": 0.9937799845076536,
      "learning_rate": 3.4557170520990705e-06,
      "loss": 0.7975,
      "num_input_tokens_seen": 46346635,
      "step": 2190
    },
    {
      "epoch": 0.2634521733902483,
      "flos": 17820981350400.0,
      "grad_norm": 2.266309963470333,
      "learning_rate": 3.4551827795520324e-06,
      "loss": 1.0616,
      "num_input_tokens_seen": 46369240,
      "step": 2191
    },
    {
      "epoch": 0.2635724162808874,
      "flos": 14619759452160.0,
      "grad_norm": 4.784568682038595,
      "learning_rate": 3.4546482862549226e-06,
      "loss": 1.0426,
      "num_input_tokens_seen": 46389275,
      "step": 2192
    },
    {
      "epoch": 0.2636926591715265,
      "flos": 13649238589440.0,
      "grad_norm": 10.906806122775112,
      "learning_rate": 3.4541135722888253e-06,
      "loss": 0.9835,
      "num_input_tokens_seen": 46405585,
      "step": 2193
    },
    {
      "epoch": 0.26381290206216557,
      "flos": 20520137871360.0,
      "grad_norm": 7.541942809161968,
      "learning_rate": 3.453578637734854e-06,
      "loss": 0.9978,
      "num_input_tokens_seen": 46426495,
      "step": 2194
    },
    {
      "epoch": 0.2639331449528047,
      "flos": 17791847608320.0,
      "grad_norm": 4.796310143032786,
      "learning_rate": 3.4530434826741605e-06,
      "loss": 0.9818,
      "num_input_tokens_seen": 46447155,
      "step": 2195
    },
    {
      "epoch": 0.26405338784344373,
      "flos": 33262833684480.0,
      "grad_norm": 4.574779967973443,
      "learning_rate": 3.452508107187926e-06,
      "loss": 0.8849,
      "num_input_tokens_seen": 46470250,
      "step": 2196
    },
    {
      "epoch": 0.26417363073408284,
      "flos": 15039911485440.0,
      "grad_norm": 4.770563287324584,
      "learning_rate": 3.451972511357366e-06,
      "loss": 0.9523,
      "num_input_tokens_seen": 46489515,
      "step": 2197
    },
    {
      "epoch": 0.26429387362472195,
      "flos": 16114458624000.0,
      "grad_norm": 5.594221212476415,
      "learning_rate": 3.45143669526373e-06,
      "loss": 1.0513,
      "num_input_tokens_seen": 46508995,
      "step": 2198
    },
    {
      "epoch": 0.264414116515361,
      "flos": 48103920844800.0,
      "grad_norm": 0.8280162020056395,
      "learning_rate": 3.450900658988302e-06,
      "loss": 0.8278,
      "num_input_tokens_seen": 46570265,
      "step": 2199
    },
    {
      "epoch": 0.2645343594060001,
      "flos": 18263031705600.0,
      "grad_norm": 7.622779647987182,
      "learning_rate": 3.450364402612397e-06,
      "loss": 0.9685,
      "num_input_tokens_seen": 46587140,
      "step": 2200
    },
    {
      "epoch": 0.26465460229663923,
      "flos": 15982923509760.0,
      "grad_norm": 7.349922605353576,
      "learning_rate": 3.449827926217366e-06,
      "loss": 1.0156,
      "num_input_tokens_seen": 46606295,
      "step": 2201
    },
    {
      "epoch": 0.2647748451872783,
      "flos": 20939155537920.0,
      "grad_norm": 8.792022071151173,
      "learning_rate": 3.449291229884591e-06,
      "loss": 0.9949,
      "num_input_tokens_seen": 46627255,
      "step": 2202
    },
    {
      "epoch": 0.2648950880779174,
      "flos": 19077635788800.0,
      "grad_norm": 5.184422510813512,
      "learning_rate": 3.4487543136954887e-06,
      "loss": 1.0491,
      "num_input_tokens_seen": 46646595,
      "step": 2203
    },
    {
      "epoch": 0.2650153309685565,
      "flos": 20546052464640.0,
      "grad_norm": 3.9930765315440144,
      "learning_rate": 3.448217177731509e-06,
      "loss": 1.1017,
      "num_input_tokens_seen": 46666800,
      "step": 2204
    },
    {
      "epoch": 0.26513557385919556,
      "flos": 14410112655360.0,
      "grad_norm": 14.62233872938909,
      "learning_rate": 3.4476798220741348e-06,
      "loss": 0.9729,
      "num_input_tokens_seen": 46685400,
      "step": 2205
    },
    {
      "epoch": 0.26525581674983467,
      "flos": 12522157117440.0,
      "grad_norm": 6.132517658119578,
      "learning_rate": 3.4471422468048826e-06,
      "loss": 0.9707,
      "num_input_tokens_seen": 46703845,
      "step": 2206
    },
    {
      "epoch": 0.2653760596404738,
      "flos": 19103397089280.0,
      "grad_norm": 4.4348289080005445,
      "learning_rate": 3.4466044520053022e-06,
      "loss": 0.9277,
      "num_input_tokens_seen": 46722570,
      "step": 2207
    },
    {
      "epoch": 0.26549630253111284,
      "flos": 16191374622720.0,
      "grad_norm": 3.8085317753371104,
      "learning_rate": 3.446066437756977e-06,
      "loss": 0.798,
      "num_input_tokens_seen": 46741495,
      "step": 2208
    },
    {
      "epoch": 0.26561654542175195,
      "flos": 16743429672960.0,
      "grad_norm": 5.081808497292023,
      "learning_rate": 3.4455282041415224e-06,
      "loss": 0.9464,
      "num_input_tokens_seen": 46760425,
      "step": 2209
    },
    {
      "epoch": 0.265736788312391,
      "flos": 19155808788480.0,
      "grad_norm": 5.43514560283338,
      "learning_rate": 3.4449897512405894e-06,
      "loss": 1.065,
      "num_input_tokens_seen": 46779295,
      "step": 2210
    },
    {
      "epoch": 0.2658570312030301,
      "flos": 16691079290880.0,
      "grad_norm": 4.253018597484822,
      "learning_rate": 3.444451079135859e-06,
      "loss": 0.9339,
      "num_input_tokens_seen": 46798525,
      "step": 2211
    },
    {
      "epoch": 0.2659772740936692,
      "flos": 15532748636160.0,
      "grad_norm": 7.669285976392803,
      "learning_rate": 3.4439121879090493e-06,
      "loss": 0.936,
      "num_input_tokens_seen": 46816025,
      "step": 2212
    },
    {
      "epoch": 0.2660975169843083,
      "flos": 14043568005120.0,
      "grad_norm": 5.329541686430677,
      "learning_rate": 3.4433730776419082e-06,
      "loss": 1.0261,
      "num_input_tokens_seen": 46834670,
      "step": 2213
    },
    {
      "epoch": 0.2662177598749474,
      "flos": 20673356697600.0,
      "grad_norm": 5.3794776540973395,
      "learning_rate": 3.4428337484162183e-06,
      "loss": 0.9955,
      "num_input_tokens_seen": 46855200,
      "step": 2214
    },
    {
      "epoch": 0.2663380027655865,
      "flos": 15458867834880.0,
      "grad_norm": 6.436806947011515,
      "learning_rate": 3.442294200313797e-06,
      "loss": 1.0433,
      "num_input_tokens_seen": 46872950,
      "step": 2215
    },
    {
      "epoch": 0.26645824565622556,
      "flos": 47962566881280.0,
      "grad_norm": 0.9239637666447187,
      "learning_rate": 3.4417544334164916e-06,
      "loss": 0.7793,
      "num_input_tokens_seen": 46936815,
      "step": 2216
    },
    {
      "epoch": 0.26657848854686467,
      "flos": 17975303884800.0,
      "grad_norm": 4.777656950396182,
      "learning_rate": 3.4412144478061854e-06,
      "loss": 0.962,
      "num_input_tokens_seen": 46958945,
      "step": 2217
    },
    {
      "epoch": 0.2666987314375038,
      "flos": 16849019535360.0,
      "grad_norm": 4.1927361844798305,
      "learning_rate": 3.4406742435647925e-06,
      "loss": 0.9508,
      "num_input_tokens_seen": 46978730,
      "step": 2218
    },
    {
      "epoch": 0.26681897432814283,
      "flos": 19257566330880.0,
      "grad_norm": 5.747715703592704,
      "learning_rate": 3.440133820774263e-06,
      "loss": 0.9791,
      "num_input_tokens_seen": 46998260,
      "step": 2219
    },
    {
      "epoch": 0.26693921721878194,
      "flos": 20650078740480.0,
      "grad_norm": 6.825303950545645,
      "learning_rate": 3.439593179516578e-06,
      "loss": 0.9983,
      "num_input_tokens_seen": 47017890,
      "step": 2220
    },
    {
      "epoch": 0.26705946010942105,
      "flos": 15275902095360.0,
      "grad_norm": 4.071496011658272,
      "learning_rate": 3.4390523198737524e-06,
      "loss": 1.0018,
      "num_input_tokens_seen": 47036770,
      "step": 2221
    },
    {
      "epoch": 0.2671797030000601,
      "flos": 15249619599360.0,
      "grad_norm": 3.1086797655835716,
      "learning_rate": 3.4385112419278333e-06,
      "loss": 0.9367,
      "num_input_tokens_seen": 47057715,
      "step": 2222
    },
    {
      "epoch": 0.2672999458906992,
      "flos": 45954795909120.0,
      "grad_norm": 0.8893175734680129,
      "learning_rate": 3.4379699457609033e-06,
      "loss": 0.887,
      "num_input_tokens_seen": 47115260,
      "step": 2223
    },
    {
      "epoch": 0.26742018878133833,
      "flos": 11970623262720.0,
      "grad_norm": 3.51024585764874,
      "learning_rate": 3.4374284314550755e-06,
      "loss": 1.086,
      "num_input_tokens_seen": 47134020,
      "step": 2224
    },
    {
      "epoch": 0.2675404316719774,
      "flos": 14671864565760.0,
      "grad_norm": 11.075409851068295,
      "learning_rate": 3.436886699092498e-06,
      "loss": 0.9911,
      "num_input_tokens_seen": 47152255,
      "step": 2225
    },
    {
      "epoch": 0.2676606745626165,
      "flos": 12383509217280.0,
      "grad_norm": 4.621409698030018,
      "learning_rate": 3.4363447487553502e-06,
      "loss": 0.901,
      "num_input_tokens_seen": 47165290,
      "step": 2226
    },
    {
      "epoch": 0.26778091745325555,
      "flos": 19835321364480.0,
      "grad_norm": 4.380653869734062,
      "learning_rate": 3.4358025805258455e-06,
      "loss": 0.9745,
      "num_input_tokens_seen": 47184715,
      "step": 2227
    },
    {
      "epoch": 0.26790116034389466,
      "flos": 14878690775040.0,
      "grad_norm": 4.202969799934346,
      "learning_rate": 3.435260194486232e-06,
      "loss": 1.0252,
      "num_input_tokens_seen": 47202405,
      "step": 2228
    },
    {
      "epoch": 0.2680214032345338,
      "flos": 12783970344960.0,
      "grad_norm": 9.151685182238552,
      "learning_rate": 3.4347175907187875e-06,
      "loss": 1.0037,
      "num_input_tokens_seen": 47219115,
      "step": 2229
    },
    {
      "epoch": 0.26814164612517283,
      "flos": 15930971688960.0,
      "grad_norm": 5.675670016223339,
      "learning_rate": 3.4341747693058254e-06,
      "loss": 1.0685,
      "num_input_tokens_seen": 47237310,
      "step": 2230
    },
    {
      "epoch": 0.26826188901581194,
      "flos": 25423283712000.0,
      "grad_norm": 4.23722512583032,
      "learning_rate": 3.4336317303296916e-06,
      "loss": 0.9597,
      "num_input_tokens_seen": 47258005,
      "step": 2231
    },
    {
      "epoch": 0.26838213190645105,
      "flos": 12495935938560.0,
      "grad_norm": 4.399025696474967,
      "learning_rate": 3.4330884738727635e-06,
      "loss": 0.9378,
      "num_input_tokens_seen": 47275900,
      "step": 2232
    },
    {
      "epoch": 0.2685023747970901,
      "flos": 16114519941120.0,
      "grad_norm": 4.723622213039844,
      "learning_rate": 3.4325450000174535e-06,
      "loss": 0.9139,
      "num_input_tokens_seen": 47292260,
      "step": 2233
    },
    {
      "epoch": 0.2686226176877292,
      "flos": 14278761492480.0,
      "grad_norm": 4.54582700417157,
      "learning_rate": 3.4320013088462067e-06,
      "loss": 0.9306,
      "num_input_tokens_seen": 47309340,
      "step": 2234
    },
    {
      "epoch": 0.2687428605783683,
      "flos": 15537562030080.0,
      "grad_norm": 3.226242825464299,
      "learning_rate": 3.431457400441499e-06,
      "loss": 1.0078,
      "num_input_tokens_seen": 47329455,
      "step": 2235
    },
    {
      "epoch": 0.2688631034690074,
      "flos": 50088115384320.0,
      "grad_norm": 1.0812680965077874,
      "learning_rate": 3.4309132748858424e-06,
      "loss": 0.8803,
      "num_input_tokens_seen": 47390165,
      "step": 2236
    },
    {
      "epoch": 0.2689833463596465,
      "flos": 16245104640000.0,
      "grad_norm": 2.874809258809379,
      "learning_rate": 3.430368932261779e-06,
      "loss": 1.0348,
      "num_input_tokens_seen": 47410240,
      "step": 2237
    },
    {
      "epoch": 0.2691035892502856,
      "flos": 12179810181120.0,
      "grad_norm": 3.768351853628987,
      "learning_rate": 3.429824372651886e-06,
      "loss": 0.9458,
      "num_input_tokens_seen": 47428110,
      "step": 2238
    },
    {
      "epoch": 0.26922383214092466,
      "flos": 12573097205760.0,
      "grad_norm": 11.332232552719091,
      "learning_rate": 3.4292795961387732e-06,
      "loss": 1.0307,
      "num_input_tokens_seen": 47445730,
      "step": 2239
    },
    {
      "epoch": 0.26934407503156377,
      "flos": 11441110364160.0,
      "grad_norm": 4.900441613051796,
      "learning_rate": 3.4287346028050818e-06,
      "loss": 1.0679,
      "num_input_tokens_seen": 47461520,
      "step": 2240
    },
    {
      "epoch": 0.2694643179222028,
      "flos": 16874934128640.0,
      "grad_norm": 2.8929180226319833,
      "learning_rate": 3.4281893927334866e-06,
      "loss": 0.9879,
      "num_input_tokens_seen": 47481150,
      "step": 2241
    },
    {
      "epoch": 0.26958456081284193,
      "flos": 17583028592640.0,
      "grad_norm": 3.6013656198305877,
      "learning_rate": 3.4276439660066963e-06,
      "loss": 0.9471,
      "num_input_tokens_seen": 47500570,
      "step": 2242
    },
    {
      "epoch": 0.26970480370348104,
      "flos": 12835523604480.0,
      "grad_norm": 3.9893837954403155,
      "learning_rate": 3.427098322707452e-06,
      "loss": 1.0421,
      "num_input_tokens_seen": 47516255,
      "step": 2243
    },
    {
      "epoch": 0.2698250465941201,
      "flos": 7592146268160.0,
      "grad_norm": 3.7854167584825307,
      "learning_rate": 3.426552462918526e-06,
      "loss": 1.0777,
      "num_input_tokens_seen": 47533910,
      "step": 2244
    },
    {
      "epoch": 0.2699452894847592,
      "flos": 12257124741120.0,
      "grad_norm": 5.828369135511679,
      "learning_rate": 3.426006386722726e-06,
      "loss": 0.9302,
      "num_input_tokens_seen": 47551690,
      "step": 2245
    },
    {
      "epoch": 0.2700655323753983,
      "flos": 12811049963520.0,
      "grad_norm": 20.755352754620915,
      "learning_rate": 3.4254600942028914e-06,
      "loss": 1.1151,
      "num_input_tokens_seen": 47569285,
      "step": 2246
    },
    {
      "epoch": 0.2701857752660374,
      "flos": 12888885719040.0,
      "grad_norm": 4.341841244009688,
      "learning_rate": 3.424913585441893e-06,
      "loss": 1.0064,
      "num_input_tokens_seen": 47586840,
      "step": 2247
    },
    {
      "epoch": 0.2703060181566765,
      "flos": 11546148372480.0,
      "grad_norm": 9.182143122350901,
      "learning_rate": 3.4243668605226374e-06,
      "loss": 1.0705,
      "num_input_tokens_seen": 47603585,
      "step": 2248
    },
    {
      "epoch": 0.2704262610473156,
      "flos": 13884677345280.0,
      "grad_norm": 6.171235256598565,
      "learning_rate": 3.423819919528061e-06,
      "loss": 1.0251,
      "num_input_tokens_seen": 47621390,
      "step": 2249
    },
    {
      "epoch": 0.27054650393795465,
      "flos": 14724490874880.0,
      "grad_norm": 3.964038364936874,
      "learning_rate": 3.4232727625411355e-06,
      "loss": 0.9776,
      "num_input_tokens_seen": 47640215,
      "step": 2250
    },
    {
      "epoch": 0.27066674682859376,
      "flos": 13227522969600.0,
      "grad_norm": 2.812813889683572,
      "learning_rate": 3.4227253896448626e-06,
      "loss": 1.0495,
      "num_input_tokens_seen": 47657795,
      "step": 2251
    },
    {
      "epoch": 0.2707869897192329,
      "flos": 16350142648320.0,
      "grad_norm": 5.979000536729604,
      "learning_rate": 3.42217780092228e-06,
      "loss": 0.9963,
      "num_input_tokens_seen": 47675855,
      "step": 2252
    },
    {
      "epoch": 0.27090723260987193,
      "flos": 41739532431360.0,
      "grad_norm": 0.8878606426179682,
      "learning_rate": 3.421629996456456e-06,
      "loss": 0.8334,
      "num_input_tokens_seen": 47734195,
      "step": 2253
    },
    {
      "epoch": 0.27102747550051104,
      "flos": 8431622553600.0,
      "grad_norm": 17.251465666051484,
      "learning_rate": 3.421081976330491e-06,
      "loss": 1.012,
      "num_input_tokens_seen": 47752430,
      "step": 2254
    },
    {
      "epoch": 0.27114771839115015,
      "flos": 14120392028160.0,
      "grad_norm": 4.674302507071221,
      "learning_rate": 3.4205337406275207e-06,
      "loss": 1.0672,
      "num_input_tokens_seen": 47772270,
      "step": 2255
    },
    {
      "epoch": 0.2712679612817892,
      "flos": 12993249239040.0,
      "grad_norm": 5.452813519432877,
      "learning_rate": 3.4199852894307114e-06,
      "loss": 0.9406,
      "num_input_tokens_seen": 47788740,
      "step": 2256
    },
    {
      "epoch": 0.2713882041724283,
      "flos": 17397395558400.0,
      "grad_norm": 3.767241071828816,
      "learning_rate": 3.419436622823262e-06,
      "loss": 0.9814,
      "num_input_tokens_seen": 47809180,
      "step": 2257
    },
    {
      "epoch": 0.27150844706306737,
      "flos": 16376425144320.0,
      "grad_norm": 4.294944523729505,
      "learning_rate": 3.4188877408884063e-06,
      "loss": 0.9443,
      "num_input_tokens_seen": 47829605,
      "step": 2258
    },
    {
      "epoch": 0.2716286899537065,
      "flos": 16034384793600.0,
      "grad_norm": 6.796833467904719,
      "learning_rate": 3.4183386437094088e-06,
      "loss": 0.8543,
      "num_input_tokens_seen": 47845990,
      "step": 2259
    },
    {
      "epoch": 0.2717489328443456,
      "flos": 9244019220480.0,
      "grad_norm": 6.830884851467188,
      "learning_rate": 3.417789331369565e-06,
      "loss": 1.0087,
      "num_input_tokens_seen": 47861500,
      "step": 2260
    },
    {
      "epoch": 0.27186917573498465,
      "flos": 20860583976960.0,
      "grad_norm": 4.526831636601663,
      "learning_rate": 3.4172398039522088e-06,
      "loss": 1.1019,
      "num_input_tokens_seen": 47882505,
      "step": 2261
    },
    {
      "epoch": 0.27198941862562376,
      "flos": 18527512227840.0,
      "grad_norm": 2.8718026667836702,
      "learning_rate": 3.4166900615407e-06,
      "loss": 0.9896,
      "num_input_tokens_seen": 47900140,
      "step": 2262
    },
    {
      "epoch": 0.27210966151626287,
      "flos": 23377571880960.0,
      "grad_norm": 7.335783317985049,
      "learning_rate": 3.416140104218436e-06,
      "loss": 0.9368,
      "num_input_tokens_seen": 47919225,
      "step": 2263
    },
    {
      "epoch": 0.2722299044069019,
      "flos": 46876062904320.0,
      "grad_norm": 0.9187365628252466,
      "learning_rate": 3.4155899320688437e-06,
      "loss": 0.9404,
      "num_input_tokens_seen": 47985020,
      "step": 2264
    },
    {
      "epoch": 0.27235014729754103,
      "flos": 10837961932800.0,
      "grad_norm": 3.49449583223689,
      "learning_rate": 3.415039545175384e-06,
      "loss": 0.939,
      "num_input_tokens_seen": 48000465,
      "step": 2265
    },
    {
      "epoch": 0.27247039018818014,
      "flos": 15537837957120.0,
      "grad_norm": 4.944711947021148,
      "learning_rate": 3.414488943621551e-06,
      "loss": 0.8527,
      "num_input_tokens_seen": 48018850,
      "step": 2266
    },
    {
      "epoch": 0.2725906330788192,
      "flos": 13254326661120.0,
      "grad_norm": 3.8292724135404694,
      "learning_rate": 3.41393812749087e-06,
      "loss": 0.9321,
      "num_input_tokens_seen": 48036615,
      "step": 2267
    },
    {
      "epoch": 0.2727108759694583,
      "flos": 12671727575040.0,
      "grad_norm": 18.97856939264407,
      "learning_rate": 3.4133870968668984e-06,
      "loss": 0.9181,
      "num_input_tokens_seen": 48051135,
      "step": 2268
    },
    {
      "epoch": 0.2728311188600974,
      "flos": 17398560583680.0,
      "grad_norm": 3.378963544944148,
      "learning_rate": 3.412835851833229e-06,
      "loss": 0.9709,
      "num_input_tokens_seen": 48073050,
      "step": 2269
    },
    {
      "epoch": 0.2729513617507365,
      "flos": 22092826091520.0,
      "grad_norm": 3.5826860035368666,
      "learning_rate": 3.4122843924734834e-06,
      "loss": 0.9704,
      "num_input_tokens_seen": 48095070,
      "step": 2270
    },
    {
      "epoch": 0.2730716046413756,
      "flos": 13540889456640.0,
      "grad_norm": 6.719443777770389,
      "learning_rate": 3.411732718871319e-06,
      "loss": 1.083,
      "num_input_tokens_seen": 48110630,
      "step": 2271
    },
    {
      "epoch": 0.27319184753201464,
      "flos": 19183225651200.0,
      "grad_norm": 3.086552580365328,
      "learning_rate": 3.4111808311104227e-06,
      "loss": 0.9784,
      "num_input_tokens_seen": 48132665,
      "step": 2272
    },
    {
      "epoch": 0.27331209042265375,
      "flos": 22643746775040.0,
      "grad_norm": 3.8662427576677416,
      "learning_rate": 3.410628729274517e-06,
      "loss": 0.8835,
      "num_input_tokens_seen": 48153905,
      "step": 2273
    },
    {
      "epoch": 0.27343233331329286,
      "flos": 18316669747200.0,
      "grad_norm": 3.544745255371121,
      "learning_rate": 3.4100764134473546e-06,
      "loss": 1.018,
      "num_input_tokens_seen": 48172910,
      "step": 2274
    },
    {
      "epoch": 0.2735525762039319,
      "flos": 17346731397120.0,
      "grad_norm": 3.190467649758843,
      "learning_rate": 3.4095238837127215e-06,
      "loss": 1.0351,
      "num_input_tokens_seen": 48191770,
      "step": 2275
    },
    {
      "epoch": 0.27367281909457103,
      "flos": 10214509424640.0,
      "grad_norm": 3.374600563430807,
      "learning_rate": 3.4089711401544355e-06,
      "loss": 0.9775,
      "num_input_tokens_seen": 48209085,
      "step": 2276
    },
    {
      "epoch": 0.27379306198521014,
      "flos": 16690956656640.0,
      "grad_norm": 6.018878483019853,
      "learning_rate": 3.4084181828563486e-06,
      "loss": 0.8659,
      "num_input_tokens_seen": 48225525,
      "step": 2277
    },
    {
      "epoch": 0.2739133048758492,
      "flos": 12365504532480.0,
      "grad_norm": 5.964985723998005,
      "learning_rate": 3.4078650119023428e-06,
      "loss": 0.8964,
      "num_input_tokens_seen": 48243560,
      "step": 2278
    },
    {
      "epoch": 0.2740335477664883,
      "flos": 13668622909440.0,
      "grad_norm": 5.511268862308955,
      "learning_rate": 3.4073116273763337e-06,
      "loss": 0.9373,
      "num_input_tokens_seen": 48257725,
      "step": 2279
    },
    {
      "epoch": 0.2741537906571274,
      "flos": 18579556024320.0,
      "grad_norm": 3.550460549151896,
      "learning_rate": 3.40675802936227e-06,
      "loss": 1.0091,
      "num_input_tokens_seen": 48278230,
      "step": 2280
    },
    {
      "epoch": 0.27427403354776647,
      "flos": 24371217408000.0,
      "grad_norm": 6.543675043502021,
      "learning_rate": 3.4062042179441318e-06,
      "loss": 0.907,
      "num_input_tokens_seen": 48298420,
      "step": 2281
    },
    {
      "epoch": 0.2743942764384056,
      "flos": 13305512017920.0,
      "grad_norm": 5.582789759959285,
      "learning_rate": 3.4056501932059314e-06,
      "loss": 1.0061,
      "num_input_tokens_seen": 48316215,
      "step": 2282
    },
    {
      "epoch": 0.2745145193290447,
      "flos": 46465913671680.0,
      "grad_norm": 0.8574799189982764,
      "learning_rate": 3.405095955231715e-06,
      "loss": 0.8115,
      "num_input_tokens_seen": 48367590,
      "step": 2283
    },
    {
      "epoch": 0.27463476221968375,
      "flos": 11414919843840.0,
      "grad_norm": 3.8682084402434174,
      "learning_rate": 3.4045415041055585e-06,
      "loss": 1.1317,
      "num_input_tokens_seen": 48382950,
      "step": 2284
    },
    {
      "epoch": 0.27475500511032286,
      "flos": 7275775242240.0,
      "grad_norm": 4.380907919857539,
      "learning_rate": 3.4039868399115728e-06,
      "loss": 0.9735,
      "num_input_tokens_seen": 48397310,
      "step": 2285
    },
    {
      "epoch": 0.27487524800096197,
      "flos": 12259638743040.0,
      "grad_norm": 8.387610663012282,
      "learning_rate": 3.4034319627339003e-06,
      "loss": 0.9893,
      "num_input_tokens_seen": 48413895,
      "step": 2286
    },
    {
      "epoch": 0.274995490891601,
      "flos": 19308537077760.0,
      "grad_norm": 63.54336148696022,
      "learning_rate": 3.402876872656715e-06,
      "loss": 0.8929,
      "num_input_tokens_seen": 48431935,
      "step": 2287
    },
    {
      "epoch": 0.27511573378224013,
      "flos": 16660964474880.0,
      "grad_norm": 3.943585008171474,
      "learning_rate": 3.402321569764223e-06,
      "loss": 1.0795,
      "num_input_tokens_seen": 48450960,
      "step": 2288
    },
    {
      "epoch": 0.2752359766728792,
      "flos": 11835593072640.0,
      "grad_norm": 3.2106784985261974,
      "learning_rate": 3.4017660541406635e-06,
      "loss": 1.0257,
      "num_input_tokens_seen": 48466745,
      "step": 2289
    },
    {
      "epoch": 0.2753562195635183,
      "flos": 17999011061760.0,
      "grad_norm": 2.4881799698223883,
      "learning_rate": 3.4012103258703092e-06,
      "loss": 0.9302,
      "num_input_tokens_seen": 48485220,
      "step": 2290
    },
    {
      "epoch": 0.2754764624541574,
      "flos": 19465741516800.0,
      "grad_norm": 6.902565642213651,
      "learning_rate": 3.4006543850374616e-06,
      "loss": 1.0364,
      "num_input_tokens_seen": 48499990,
      "step": 2291
    },
    {
      "epoch": 0.27559670534479647,
      "flos": 12207441653760.0,
      "grad_norm": 6.143330451245735,
      "learning_rate": 3.400098231726458e-06,
      "loss": 0.9479,
      "num_input_tokens_seen": 48516810,
      "step": 2292
    },
    {
      "epoch": 0.2757169482354356,
      "flos": 15584945725440.0,
      "grad_norm": 3.812160039034039,
      "learning_rate": 3.3995418660216657e-06,
      "loss": 1.0722,
      "num_input_tokens_seen": 48533985,
      "step": 2293
    },
    {
      "epoch": 0.2758371911260747,
      "flos": 14803552972800.0,
      "grad_norm": 3.646846710330707,
      "learning_rate": 3.3989852880074848e-06,
      "loss": 1.003,
      "num_input_tokens_seen": 48555135,
      "step": 2294
    },
    {
      "epoch": 0.27595743401671374,
      "flos": 49606100705280.0,
      "grad_norm": 0.9774346996164566,
      "learning_rate": 3.398428497768348e-06,
      "loss": 0.8391,
      "num_input_tokens_seen": 48620025,
      "step": 2295
    },
    {
      "epoch": 0.27607767690735285,
      "flos": 15065335541760.0,
      "grad_norm": 6.960724044935004,
      "learning_rate": 3.3978714953887205e-06,
      "loss": 0.9087,
      "num_input_tokens_seen": 48639500,
      "step": 2296
    },
    {
      "epoch": 0.27619791979799196,
      "flos": 17659760640000.0,
      "grad_norm": 3.656009372715862,
      "learning_rate": 3.397314280953098e-06,
      "loss": 1.0592,
      "num_input_tokens_seen": 48660045,
      "step": 2297
    },
    {
      "epoch": 0.276318162688631,
      "flos": 17608330014720.0,
      "grad_norm": 3.2685034810281546,
      "learning_rate": 3.3967568545460108e-06,
      "loss": 0.995,
      "num_input_tokens_seen": 48679305,
      "step": 2298
    },
    {
      "epoch": 0.27643840557927013,
      "flos": 12862541905920.0,
      "grad_norm": 5.942394608271982,
      "learning_rate": 3.3961992162520185e-06,
      "loss": 0.9921,
      "num_input_tokens_seen": 48697650,
      "step": 2299
    },
    {
      "epoch": 0.27655864846990924,
      "flos": 17658564956160.0,
      "grad_norm": 4.875828990387774,
      "learning_rate": 3.3956413661557156e-06,
      "loss": 0.9132,
      "num_input_tokens_seen": 48717545,
      "step": 2300
    },
    {
      "epoch": 0.2766788913605483,
      "flos": 14383738183680.0,
      "grad_norm": 3.41013667394868,
      "learning_rate": 3.3950833043417273e-06,
      "loss": 0.8549,
      "num_input_tokens_seen": 48735410,
      "step": 2301
    },
    {
      "epoch": 0.2767991342511874,
      "flos": 15248761159680.0,
      "grad_norm": 4.022899406523828,
      "learning_rate": 3.3945250308947105e-06,
      "loss": 0.9071,
      "num_input_tokens_seen": 48751435,
      "step": 2302
    },
    {
      "epoch": 0.2769193771418265,
      "flos": 47255982796800.0,
      "grad_norm": 1.2919389478767538,
      "learning_rate": 3.3939665458993556e-06,
      "loss": 0.8989,
      "num_input_tokens_seen": 48805575,
      "step": 2303
    },
    {
      "epoch": 0.27703962003246557,
      "flos": 14698422988800.0,
      "grad_norm": 3.659375281460146,
      "learning_rate": 3.3934078494403843e-06,
      "loss": 0.948,
      "num_input_tokens_seen": 48824870,
      "step": 2304
    },
    {
      "epoch": 0.2771598629231047,
      "flos": 16297792266240.0,
      "grad_norm": 16.666828510770184,
      "learning_rate": 3.3928489416025495e-06,
      "loss": 1.0021,
      "num_input_tokens_seen": 48845435,
      "step": 2305
    },
    {
      "epoch": 0.27728010581374374,
      "flos": 13020328857600.0,
      "grad_norm": 3.5623937222874136,
      "learning_rate": 3.392289822470638e-06,
      "loss": 0.9717,
      "num_input_tokens_seen": 48863135,
      "step": 2306
    },
    {
      "epoch": 0.27740034870438285,
      "flos": 13780835020800.0,
      "grad_norm": 3.7958790921168992,
      "learning_rate": 3.3917304921294674e-06,
      "loss": 0.9462,
      "num_input_tokens_seen": 48881020,
      "step": 2307
    },
    {
      "epoch": 0.27752059159502196,
      "flos": 15352419532800.0,
      "grad_norm": 3.1188950988408126,
      "learning_rate": 3.3911709506638876e-06,
      "loss": 0.9944,
      "num_input_tokens_seen": 48900050,
      "step": 2308
    },
    {
      "epoch": 0.277640834485661,
      "flos": 18940244889600.0,
      "grad_norm": 3.672858832833139,
      "learning_rate": 3.390611198158781e-06,
      "loss": 1.0045,
      "num_input_tokens_seen": 48917645,
      "step": 2309
    },
    {
      "epoch": 0.2777610773763001,
      "flos": 13826624471040.0,
      "grad_norm": 4.4342465361253485,
      "learning_rate": 3.3900512346990612e-06,
      "loss": 1.0916,
      "num_input_tokens_seen": 48933355,
      "step": 2310
    },
    {
      "epoch": 0.27788132026693924,
      "flos": 27337307136000.0,
      "grad_norm": 2.852517832801435,
      "learning_rate": 3.389491060369674e-06,
      "loss": 0.8399,
      "num_input_tokens_seen": 48958750,
      "step": 2311
    },
    {
      "epoch": 0.2780015631575783,
      "flos": 15904413265920.0,
      "grad_norm": 4.3602431878679235,
      "learning_rate": 3.388930675255598e-06,
      "loss": 1.0796,
      "num_input_tokens_seen": 48978320,
      "step": 2312
    },
    {
      "epoch": 0.2781218060482174,
      "flos": 8588735016960.0,
      "grad_norm": 5.7749903426977225,
      "learning_rate": 3.388370079441843e-06,
      "loss": 0.9646,
      "num_input_tokens_seen": 48993555,
      "step": 2313
    },
    {
      "epoch": 0.2782420489388565,
      "flos": 12831292723200.0,
      "grad_norm": 6.400134931572458,
      "learning_rate": 3.3878092730134505e-06,
      "loss": 1.1298,
      "num_input_tokens_seen": 49011260,
      "step": 2314
    },
    {
      "epoch": 0.27836229182949557,
      "flos": 13124661719040.0,
      "grad_norm": 3.445792997247921,
      "learning_rate": 3.3872482560554947e-06,
      "loss": 0.9969,
      "num_input_tokens_seen": 49029755,
      "step": 2315
    },
    {
      "epoch": 0.2784825347201347,
      "flos": 48032216801280.0,
      "grad_norm": 0.8568676523774228,
      "learning_rate": 3.386687028653082e-06,
      "loss": 0.7912,
      "num_input_tokens_seen": 49092320,
      "step": 2316
    },
    {
      "epoch": 0.2786027776107738,
      "flos": 16082136391680.0,
      "grad_norm": 3.3216987857391813,
      "learning_rate": 3.386125590891349e-06,
      "loss": 1.041,
      "num_input_tokens_seen": 49108915,
      "step": 2317
    },
    {
      "epoch": 0.27872302050141284,
      "flos": 11158471864320.0,
      "grad_norm": 4.412613406358482,
      "learning_rate": 3.3855639428554657e-06,
      "loss": 1.0179,
      "num_input_tokens_seen": 49126165,
      "step": 2318
    },
    {
      "epoch": 0.27884326339205195,
      "flos": 15720711720960.0,
      "grad_norm": 2.8322696559135383,
      "learning_rate": 3.385002084630635e-06,
      "loss": 1.0038,
      "num_input_tokens_seen": 49144855,
      "step": 2319
    },
    {
      "epoch": 0.278963506282691,
      "flos": 14593722224640.0,
      "grad_norm": 9.276909529311961,
      "learning_rate": 3.384440016302088e-06,
      "loss": 1.036,
      "num_input_tokens_seen": 49163250,
      "step": 2320
    },
    {
      "epoch": 0.2790837491733301,
      "flos": 15588226191360.0,
      "grad_norm": 5.695202317499867,
      "learning_rate": 3.3838777379550923e-06,
      "loss": 0.808,
      "num_input_tokens_seen": 49182415,
      "step": 2321
    },
    {
      "epoch": 0.27920399206396923,
      "flos": 18710355333120.0,
      "grad_norm": 4.412667437178198,
      "learning_rate": 3.383315249674944e-06,
      "loss": 0.9946,
      "num_input_tokens_seen": 49200700,
      "step": 2322
    },
    {
      "epoch": 0.2793242349546083,
      "flos": 18072370667520.0,
      "grad_norm": 5.757478908776807,
      "learning_rate": 3.3827525515469715e-06,
      "loss": 1.06,
      "num_input_tokens_seen": 49215325,
      "step": 2323
    },
    {
      "epoch": 0.2794444778452474,
      "flos": 14803277045760.0,
      "grad_norm": 4.771458591761592,
      "learning_rate": 3.3821896436565367e-06,
      "loss": 0.9027,
      "num_input_tokens_seen": 49234705,
      "step": 2324
    },
    {
      "epoch": 0.2795647207358865,
      "flos": 15324450816000.0,
      "grad_norm": 2.635356816955408,
      "learning_rate": 3.381626526089032e-06,
      "loss": 0.8993,
      "num_input_tokens_seen": 49253990,
      "step": 2325
    },
    {
      "epoch": 0.27968496362652556,
      "flos": 15250018160640.0,
      "grad_norm": 11.358349361580958,
      "learning_rate": 3.3810631989298815e-06,
      "loss": 0.9832,
      "num_input_tokens_seen": 49273320,
      "step": 2326
    },
    {
      "epoch": 0.2798052065171647,
      "flos": 16534150778880.0,
      "grad_norm": 4.7084146453227795,
      "learning_rate": 3.3804996622645423e-06,
      "loss": 1.0481,
      "num_input_tokens_seen": 49291040,
      "step": 2327
    },
    {
      "epoch": 0.2799254494078038,
      "flos": 15297861734400.0,
      "grad_norm": 4.014939305284959,
      "learning_rate": 3.3799359161785015e-06,
      "loss": 1.0825,
      "num_input_tokens_seen": 49310410,
      "step": 2328
    },
    {
      "epoch": 0.28004569229844284,
      "flos": 18787026063360.0,
      "grad_norm": 3.5678850933012773,
      "learning_rate": 3.3793719607572798e-06,
      "loss": 1.0483,
      "num_input_tokens_seen": 49331095,
      "step": 2329
    },
    {
      "epoch": 0.28016593518908195,
      "flos": 23928645857280.0,
      "grad_norm": 4.46922523799207,
      "learning_rate": 3.378807796086428e-06,
      "loss": 0.9631,
      "num_input_tokens_seen": 49353675,
      "step": 2330
    },
    {
      "epoch": 0.28028617807972106,
      "flos": 10843664424960.0,
      "grad_norm": 22.830829269580175,
      "learning_rate": 3.37824342225153e-06,
      "loss": 0.9514,
      "num_input_tokens_seen": 49369815,
      "step": 2331
    },
    {
      "epoch": 0.2804064209703601,
      "flos": 18159434649600.0,
      "grad_norm": 7.714267432983374,
      "learning_rate": 3.3776788393382006e-06,
      "loss": 0.9637,
      "num_input_tokens_seen": 49389015,
      "step": 2332
    },
    {
      "epoch": 0.2805266638609992,
      "flos": 20913118310400.0,
      "grad_norm": 7.026798992660311,
      "learning_rate": 3.3771140474320872e-06,
      "loss": 0.9609,
      "num_input_tokens_seen": 49408685,
      "step": 2333
    },
    {
      "epoch": 0.28064690675163834,
      "flos": 15243181301760.0,
      "grad_norm": 3.624616536872235,
      "learning_rate": 3.3765490466188664e-06,
      "loss": 0.9791,
      "num_input_tokens_seen": 49425805,
      "step": 2334
    },
    {
      "epoch": 0.2807671496422774,
      "flos": 14907609907200.0,
      "grad_norm": 13.950433999150293,
      "learning_rate": 3.3759838369842508e-06,
      "loss": 0.9164,
      "num_input_tokens_seen": 49443600,
      "step": 2335
    },
    {
      "epoch": 0.2808873925329165,
      "flos": 15274829045760.0,
      "grad_norm": 3.7448769224999707,
      "learning_rate": 3.375418418613981e-06,
      "loss": 0.9197,
      "num_input_tokens_seen": 49462345,
      "step": 2336
    },
    {
      "epoch": 0.28100763542355556,
      "flos": 11368333271040.0,
      "grad_norm": 5.630135459557678,
      "learning_rate": 3.374852791593831e-06,
      "loss": 1.0324,
      "num_input_tokens_seen": 49478265,
      "step": 2337
    },
    {
      "epoch": 0.28112787831419467,
      "flos": 13518469939200.0,
      "grad_norm": 4.4120938122235485,
      "learning_rate": 3.374286956009605e-06,
      "loss": 0.717,
      "num_input_tokens_seen": 49496550,
      "step": 2338
    },
    {
      "epoch": 0.2812481212048338,
      "flos": 9034648350720.0,
      "grad_norm": 3.3515902431380167,
      "learning_rate": 3.3737209119471405e-06,
      "loss": 0.9551,
      "num_input_tokens_seen": 49512780,
      "step": 2339
    },
    {
      "epoch": 0.28136836409547283,
      "flos": 11053740441600.0,
      "grad_norm": 5.704946399540566,
      "learning_rate": 3.373154659492306e-06,
      "loss": 0.8404,
      "num_input_tokens_seen": 49530640,
      "step": 2340
    },
    {
      "epoch": 0.28148860698611194,
      "flos": 14143915253760.0,
      "grad_norm": 3.411803871107955,
      "learning_rate": 3.3725881987310016e-06,
      "loss": 1.0396,
      "num_input_tokens_seen": 49547895,
      "step": 2341
    },
    {
      "epoch": 0.28160884987675106,
      "flos": 12364707409920.0,
      "grad_norm": 2.4064767904044135,
      "learning_rate": 3.372021529749159e-06,
      "loss": 1.0599,
      "num_input_tokens_seen": 49566675,
      "step": 2342
    },
    {
      "epoch": 0.2817290927673901,
      "flos": 11917046538240.0,
      "grad_norm": 3.061167524417576,
      "learning_rate": 3.3714546526327405e-06,
      "loss": 1.1146,
      "num_input_tokens_seen": 49584395,
      "step": 2343
    },
    {
      "epoch": 0.2818493356580292,
      "flos": 10896352051200.0,
      "grad_norm": 3.8931843256414678,
      "learning_rate": 3.3708875674677423e-06,
      "loss": 1.0638,
      "num_input_tokens_seen": 49602090,
      "step": 2344
    },
    {
      "epoch": 0.28196957854866833,
      "flos": 14488561582080.0,
      "grad_norm": 4.841463004384847,
      "learning_rate": 3.37032027434019e-06,
      "loss": 1.0198,
      "num_input_tokens_seen": 49621330,
      "step": 2345
    },
    {
      "epoch": 0.2820898214393074,
      "flos": 14172619776000.0,
      "grad_norm": 3.276026027785647,
      "learning_rate": 3.369752773336141e-06,
      "loss": 1.0284,
      "num_input_tokens_seen": 49640530,
      "step": 2346
    },
    {
      "epoch": 0.2822100643299465,
      "flos": 16009144688640.0,
      "grad_norm": 3.5158167807933904,
      "learning_rate": 3.3691850645416864e-06,
      "loss": 0.971,
      "num_input_tokens_seen": 49659960,
      "step": 2347
    },
    {
      "epoch": 0.2823303072205856,
      "flos": 8116937748480.0,
      "grad_norm": 3.8919560250662912,
      "learning_rate": 3.368617148042945e-06,
      "loss": 1.0292,
      "num_input_tokens_seen": 49677350,
      "step": 2348
    },
    {
      "epoch": 0.28245055011122466,
      "flos": 12941573345280.0,
      "grad_norm": 3.9326815717119055,
      "learning_rate": 3.368049023926071e-06,
      "loss": 1.0421,
      "num_input_tokens_seen": 49696065,
      "step": 2349
    },
    {
      "epoch": 0.2825707930018638,
      "flos": 17503874519040.0,
      "grad_norm": 3.634238342707314,
      "learning_rate": 3.3674806922772476e-06,
      "loss": 1.036,
      "num_input_tokens_seen": 49716670,
      "step": 2350
    },
    {
      "epoch": 0.28269103589250283,
      "flos": 17948469534720.0,
      "grad_norm": 3.7794190064245385,
      "learning_rate": 3.3669121531826904e-06,
      "loss": 0.9407,
      "num_input_tokens_seen": 49737370,
      "step": 2351
    },
    {
      "epoch": 0.28281127878314194,
      "flos": 13675827671040.0,
      "grad_norm": 3.5044275616846776,
      "learning_rate": 3.366343406728647e-06,
      "loss": 1.0224,
      "num_input_tokens_seen": 49756540,
      "step": 2352
    },
    {
      "epoch": 0.28293152167378105,
      "flos": 16979450941440.0,
      "grad_norm": 4.21008767804614,
      "learning_rate": 3.3657744530013946e-06,
      "loss": 0.8811,
      "num_input_tokens_seen": 49775495,
      "step": 2353
    },
    {
      "epoch": 0.2830517645644201,
      "flos": 31344088842240.0,
      "grad_norm": 3.1226645325681375,
      "learning_rate": 3.3652052920872437e-06,
      "loss": 0.9068,
      "num_input_tokens_seen": 49798080,
      "step": 2354
    },
    {
      "epoch": 0.2831720074550592,
      "flos": 18972659097600.0,
      "grad_norm": 3.0660942939447167,
      "learning_rate": 3.3646359240725355e-06,
      "loss": 1.0554,
      "num_input_tokens_seen": 49816990,
      "step": 2355
    },
    {
      "epoch": 0.2832922503456983,
      "flos": 22119200563200.0,
      "grad_norm": 5.126390036042492,
      "learning_rate": 3.364066349043643e-06,
      "loss": 0.8791,
      "num_input_tokens_seen": 49837915,
      "step": 2356
    },
    {
      "epoch": 0.2834124932363374,
      "flos": 14482797772800.0,
      "grad_norm": 2.805604067677817,
      "learning_rate": 3.363496567086969e-06,
      "loss": 1.02,
      "num_input_tokens_seen": 49854730,
      "step": 2357
    },
    {
      "epoch": 0.2835327361269765,
      "flos": 28123635916800.0,
      "grad_norm": 4.379676281536649,
      "learning_rate": 3.3629265782889506e-06,
      "loss": 0.9496,
      "num_input_tokens_seen": 49876275,
      "step": 2358
    },
    {
      "epoch": 0.2836529790176156,
      "flos": 21566685634560.0,
      "grad_norm": 5.565105042541559,
      "learning_rate": 3.362356382736054e-06,
      "loss": 0.8983,
      "num_input_tokens_seen": 49896600,
      "step": 2359
    },
    {
      "epoch": 0.28377322190825466,
      "flos": 8929855610880.0,
      "grad_norm": 7.329209757799822,
      "learning_rate": 3.361785980514777e-06,
      "loss": 1.1079,
      "num_input_tokens_seen": 49912520,
      "step": 2360
    },
    {
      "epoch": 0.28389346479889377,
      "flos": 12968039792640.0,
      "grad_norm": 3.2330270805699617,
      "learning_rate": 3.361215371711649e-06,
      "loss": 0.9634,
      "num_input_tokens_seen": 49931335,
      "step": 2361
    },
    {
      "epoch": 0.2840137076895329,
      "flos": 14483901480960.0,
      "grad_norm": 2.7864198105665583,
      "learning_rate": 3.3606445564132326e-06,
      "loss": 1.0243,
      "num_input_tokens_seen": 49948350,
      "step": 2362
    },
    {
      "epoch": 0.28413395058017193,
      "flos": 14226717696000.0,
      "grad_norm": 3.486713257788919,
      "learning_rate": 3.360073534706118e-06,
      "loss": 1.0133,
      "num_input_tokens_seen": 49965225,
      "step": 2363
    },
    {
      "epoch": 0.28425419347081105,
      "flos": 26885752627200.0,
      "grad_norm": 3.3397082563433282,
      "learning_rate": 3.35950230667693e-06,
      "loss": 0.9564,
      "num_input_tokens_seen": 49986640,
      "step": 2364
    },
    {
      "epoch": 0.28437443636145016,
      "flos": 9768718725120.0,
      "grad_norm": 4.799280414685028,
      "learning_rate": 3.358930872412323e-06,
      "loss": 1.0462,
      "num_input_tokens_seen": 50003525,
      "step": 2365
    },
    {
      "epoch": 0.2844946792520892,
      "flos": 16166962298880.0,
      "grad_norm": 4.645146591452104,
      "learning_rate": 3.3583592319989825e-06,
      "loss": 1.0014,
      "num_input_tokens_seen": 50022615,
      "step": 2366
    },
    {
      "epoch": 0.2846149221427283,
      "flos": 23115237457920.0,
      "grad_norm": 3.77876439643563,
      "learning_rate": 3.357787385523627e-06,
      "loss": 0.8813,
      "num_input_tokens_seen": 50043740,
      "step": 2367
    },
    {
      "epoch": 0.2847351650333674,
      "flos": 20283687383040.0,
      "grad_norm": 3.293375481118323,
      "learning_rate": 3.3572153330730048e-06,
      "loss": 1.0283,
      "num_input_tokens_seen": 50064555,
      "step": 2368
    },
    {
      "epoch": 0.2848554079240065,
      "flos": 39890498826240.0,
      "grad_norm": 0.8210145398523998,
      "learning_rate": 3.3566430747338956e-06,
      "loss": 0.8638,
      "num_input_tokens_seen": 50119480,
      "step": 2369
    },
    {
      "epoch": 0.2849756508146456,
      "flos": 8325266227200.0,
      "grad_norm": 6.891664872021271,
      "learning_rate": 3.35607061059311e-06,
      "loss": 1.063,
      "num_input_tokens_seen": 50134130,
      "step": 2370
    },
    {
      "epoch": 0.28509589370528465,
      "flos": 17897038909440.0,
      "grad_norm": 2.7119681298344105,
      "learning_rate": 3.3554979407374917e-06,
      "loss": 0.9406,
      "num_input_tokens_seen": 50155960,
      "step": 2371
    },
    {
      "epoch": 0.28521613659592376,
      "flos": 14174060728320.0,
      "grad_norm": 6.680095530757687,
      "learning_rate": 3.3549250652539134e-06,
      "loss": 0.9368,
      "num_input_tokens_seen": 50174775,
      "step": 2372
    },
    {
      "epoch": 0.2853363794865629,
      "flos": 16612661022720.0,
      "grad_norm": 3.0715179018482885,
      "learning_rate": 3.3543519842292794e-06,
      "loss": 1.008,
      "num_input_tokens_seen": 50194150,
      "step": 2373
    },
    {
      "epoch": 0.28545662237720193,
      "flos": 14092484628480.0,
      "grad_norm": 3.800454167781621,
      "learning_rate": 3.353778697750527e-06,
      "loss": 1.0363,
      "num_input_tokens_seen": 50212275,
      "step": 2374
    },
    {
      "epoch": 0.28557686526784104,
      "flos": 16979972136960.0,
      "grad_norm": 13.785599002878477,
      "learning_rate": 3.353205205904622e-06,
      "loss": 1.0845,
      "num_input_tokens_seen": 50231105,
      "step": 2375
    },
    {
      "epoch": 0.28569710815848015,
      "flos": 32079845437440.0,
      "grad_norm": 3.817863194064359,
      "learning_rate": 3.3526315087785637e-06,
      "loss": 0.921,
      "num_input_tokens_seen": 50251940,
      "step": 2376
    },
    {
      "epoch": 0.2858173510491192,
      "flos": 19099993989120.0,
      "grad_norm": 2.6505713404733555,
      "learning_rate": 3.3520576064593805e-06,
      "loss": 0.9998,
      "num_input_tokens_seen": 50271615,
      "step": 2377
    },
    {
      "epoch": 0.2859375939397583,
      "flos": 16455211315200.0,
      "grad_norm": 2.8723689770012246,
      "learning_rate": 3.3514834990341337e-06,
      "loss": 1.0191,
      "num_input_tokens_seen": 50291660,
      "step": 2378
    },
    {
      "epoch": 0.2860578368303974,
      "flos": 8535924756480.0,
      "grad_norm": 4.2125870511713215,
      "learning_rate": 3.3509091865899144e-06,
      "loss": 1.1308,
      "num_input_tokens_seen": 50306570,
      "step": 2379
    },
    {
      "epoch": 0.2861780797210365,
      "flos": 14147594280960.0,
      "grad_norm": 4.442647393163828,
      "learning_rate": 3.350334669213846e-06,
      "loss": 0.9006,
      "num_input_tokens_seen": 50323695,
      "step": 2380
    },
    {
      "epoch": 0.2862983226116756,
      "flos": 19627544739840.0,
      "grad_norm": 33.65176132149791,
      "learning_rate": 3.3497599469930816e-06,
      "loss": 0.9714,
      "num_input_tokens_seen": 50341625,
      "step": 2381
    },
    {
      "epoch": 0.28641856550231465,
      "flos": 15668238704640.0,
      "grad_norm": 10.121507583926634,
      "learning_rate": 3.349185020014807e-06,
      "loss": 1.0194,
      "num_input_tokens_seen": 50358610,
      "step": 2382
    },
    {
      "epoch": 0.28653880839295376,
      "flos": 15901194117120.0,
      "grad_norm": 3.342109348975987,
      "learning_rate": 3.348609888366237e-06,
      "loss": 0.9419,
      "num_input_tokens_seen": 50377345,
      "step": 2383
    },
    {
      "epoch": 0.28665905128359287,
      "flos": 16612661022720.0,
      "grad_norm": 8.077503527230544,
      "learning_rate": 3.348034552134619e-06,
      "loss": 0.8205,
      "num_input_tokens_seen": 50396470,
      "step": 2384
    },
    {
      "epoch": 0.2867792941742319,
      "flos": 14825114050560.0,
      "grad_norm": 3.0928966231486585,
      "learning_rate": 3.3474590114072316e-06,
      "loss": 1.0447,
      "num_input_tokens_seen": 50414190,
      "step": 2385
    },
    {
      "epoch": 0.28689953706487104,
      "flos": 14668829368320.0,
      "grad_norm": 6.295439980229955,
      "learning_rate": 3.3468832662713836e-06,
      "loss": 1.0202,
      "num_input_tokens_seen": 50432155,
      "step": 2386
    },
    {
      "epoch": 0.28701977995551015,
      "flos": 8927924121600.0,
      "grad_norm": 4.77400486347924,
      "learning_rate": 3.346307316814415e-06,
      "loss": 1.0315,
      "num_input_tokens_seen": 50447045,
      "step": 2387
    },
    {
      "epoch": 0.2871400228461492,
      "flos": 15092292526080.0,
      "grad_norm": 5.610524349357326,
      "learning_rate": 3.3457311631236965e-06,
      "loss": 0.9366,
      "num_input_tokens_seen": 50467750,
      "step": 2388
    },
    {
      "epoch": 0.2872602657367883,
      "flos": 17871277608960.0,
      "grad_norm": 4.097574791403151,
      "learning_rate": 3.345154805286631e-06,
      "loss": 1.0365,
      "num_input_tokens_seen": 50487730,
      "step": 2389
    },
    {
      "epoch": 0.2873805086274274,
      "flos": 11781065932800.0,
      "grad_norm": 3.9221276433658416,
      "learning_rate": 3.344578243390651e-06,
      "loss": 0.951,
      "num_input_tokens_seen": 50503010,
      "step": 2390
    },
    {
      "epoch": 0.2875007515180665,
      "flos": 12337965035520.0,
      "grad_norm": 4.588188281904425,
      "learning_rate": 3.3440014775232206e-06,
      "loss": 0.9731,
      "num_input_tokens_seen": 50520785,
      "step": 2391
    },
    {
      "epoch": 0.2876209944087056,
      "flos": 16659339571200.0,
      "grad_norm": 4.060891170318157,
      "learning_rate": 3.343424507771834e-06,
      "loss": 0.9153,
      "num_input_tokens_seen": 50538715,
      "step": 2392
    },
    {
      "epoch": 0.2877412372993447,
      "flos": 9689963212800.0,
      "grad_norm": 11.04010017847137,
      "learning_rate": 3.342847334224018e-06,
      "loss": 1.0655,
      "num_input_tokens_seen": 50555835,
      "step": 2393
    },
    {
      "epoch": 0.28786148018998375,
      "flos": 41566101504000.0,
      "grad_norm": 0.9040910391814219,
      "learning_rate": 3.342269956967329e-06,
      "loss": 0.8605,
      "num_input_tokens_seen": 50617460,
      "step": 2394
    },
    {
      "epoch": 0.28798172308062286,
      "flos": 16660075376640.0,
      "grad_norm": 17.87654672613149,
      "learning_rate": 3.341692376089355e-06,
      "loss": 0.9061,
      "num_input_tokens_seen": 50632735,
      "step": 2395
    },
    {
      "epoch": 0.288101965971262,
      "flos": 17863398359040.0,
      "grad_norm": 34.74800466393718,
      "learning_rate": 3.3411145916777146e-06,
      "loss": 1.0193,
      "num_input_tokens_seen": 50646615,
      "step": 2396
    },
    {
      "epoch": 0.28822220886190103,
      "flos": 11498856652800.0,
      "grad_norm": 15.471097517578949,
      "learning_rate": 3.3405366038200566e-06,
      "loss": 1.096,
      "num_input_tokens_seen": 50665270,
      "step": 2397
    },
    {
      "epoch": 0.28834245175254014,
      "flos": 17451861381120.0,
      "grad_norm": 4.982477574353011,
      "learning_rate": 3.3399584126040617e-06,
      "loss": 1.0378,
      "num_input_tokens_seen": 50684490,
      "step": 2398
    },
    {
      "epoch": 0.2884626946431792,
      "flos": 17477500047360.0,
      "grad_norm": 3.408432287693279,
      "learning_rate": 3.339380018117441e-06,
      "loss": 1.0929,
      "num_input_tokens_seen": 50705045,
      "step": 2399
    },
    {
      "epoch": 0.2885829375338183,
      "flos": 11708810035200.0,
      "grad_norm": 6.371778983402522,
      "learning_rate": 3.3388014204479366e-06,
      "loss": 0.972,
      "num_input_tokens_seen": 50722570,
      "step": 2400
    },
    {
      "epoch": 0.2887031804244574,
      "flos": 17110894080000.0,
      "grad_norm": 4.244649099128389,
      "learning_rate": 3.338222619683321e-06,
      "loss": 1.1104,
      "num_input_tokens_seen": 50742255,
      "step": 2401
    },
    {
      "epoch": 0.2888234233150965,
      "flos": 16585765355520.0,
      "grad_norm": 4.6420053811623525,
      "learning_rate": 3.337643615911398e-06,
      "loss": 0.9281,
      "num_input_tokens_seen": 50761600,
      "step": 2402
    },
    {
      "epoch": 0.2889436662057356,
      "flos": 15825473802240.0,
      "grad_norm": 3.27913285027549,
      "learning_rate": 3.3370644092200026e-06,
      "loss": 0.9813,
      "num_input_tokens_seen": 50778595,
      "step": 2403
    },
    {
      "epoch": 0.2890639090963747,
      "flos": 15354259046400.0,
      "grad_norm": 5.929397354767401,
      "learning_rate": 3.3364849996969985e-06,
      "loss": 0.9868,
      "num_input_tokens_seen": 50798335,
      "step": 2404
    },
    {
      "epoch": 0.28918415198701375,
      "flos": 20362442895360.0,
      "grad_norm": 3.574803705761899,
      "learning_rate": 3.335905387430283e-06,
      "loss": 1.0443,
      "num_input_tokens_seen": 50819490,
      "step": 2405
    },
    {
      "epoch": 0.28930439487765286,
      "flos": 15589728460800.0,
      "grad_norm": 3.0508672137976744,
      "learning_rate": 3.335325572507782e-06,
      "loss": 1.0165,
      "num_input_tokens_seen": 50839710,
      "step": 2406
    },
    {
      "epoch": 0.28942463776829197,
      "flos": 13676134256640.0,
      "grad_norm": 4.0789931969909725,
      "learning_rate": 3.3347455550174537e-06,
      "loss": 0.9338,
      "num_input_tokens_seen": 50858770,
      "step": 2407
    },
    {
      "epoch": 0.289544880658931,
      "flos": 10343837122560.0,
      "grad_norm": 4.426002328574718,
      "learning_rate": 3.3341653350472864e-06,
      "loss": 0.8793,
      "num_input_tokens_seen": 50875320,
      "step": 2408
    },
    {
      "epoch": 0.28966512354957014,
      "flos": 20388541440000.0,
      "grad_norm": 4.621934648171453,
      "learning_rate": 3.333584912685298e-06,
      "loss": 0.8886,
      "num_input_tokens_seen": 50893660,
      "step": 2409
    },
    {
      "epoch": 0.28978536644020925,
      "flos": 46331335249920.0,
      "grad_norm": 0.8481167715709375,
      "learning_rate": 3.3330042880195385e-06,
      "loss": 0.7826,
      "num_input_tokens_seen": 50947730,
      "step": 2410
    },
    {
      "epoch": 0.2899056093308483,
      "flos": 13203693158400.0,
      "grad_norm": 3.662331243279074,
      "learning_rate": 3.3324234611380888e-06,
      "loss": 0.9793,
      "num_input_tokens_seen": 50966180,
      "step": 2411
    },
    {
      "epoch": 0.2900258522214874,
      "flos": 16271847014400.0,
      "grad_norm": 2.3639405394949295,
      "learning_rate": 3.3318424321290596e-06,
      "loss": 1.0111,
      "num_input_tokens_seen": 50985615,
      "step": 2412
    },
    {
      "epoch": 0.2901460951121265,
      "flos": 50923084861440.0,
      "grad_norm": 0.852978849642963,
      "learning_rate": 3.3312612010805917e-06,
      "loss": 0.8395,
      "num_input_tokens_seen": 51044910,
      "step": 2413
    },
    {
      "epoch": 0.2902663380027656,
      "flos": 22932026449920.0,
      "grad_norm": 3.8205064595605576,
      "learning_rate": 3.330679768080858e-06,
      "loss": 0.9022,
      "num_input_tokens_seen": 51068515,
      "step": 2414
    },
    {
      "epoch": 0.2903865808934047,
      "flos": 20913026334720.0,
      "grad_norm": 3.806976694976121,
      "learning_rate": 3.3300981332180627e-06,
      "loss": 1.0389,
      "num_input_tokens_seen": 51087440,
      "step": 2415
    },
    {
      "epoch": 0.29050682378404374,
      "flos": 12099644375040.0,
      "grad_norm": 8.338099477282961,
      "learning_rate": 3.3295162965804373e-06,
      "loss": 1.0049,
      "num_input_tokens_seen": 51105655,
      "step": 2416
    },
    {
      "epoch": 0.29062706667468285,
      "flos": 12652864450560.0,
      "grad_norm": 3.7981847693891138,
      "learning_rate": 3.328934258256247e-06,
      "loss": 0.9642,
      "num_input_tokens_seen": 51123440,
      "step": 2417
    },
    {
      "epoch": 0.29074730956532197,
      "flos": 17267945226240.0,
      "grad_norm": 6.904163994920022,
      "learning_rate": 3.3283520183337856e-06,
      "loss": 0.8592,
      "num_input_tokens_seen": 51142865,
      "step": 2418
    },
    {
      "epoch": 0.290867552455961,
      "flos": 15873838571520.0,
      "grad_norm": 3.252048873637338,
      "learning_rate": 3.3277695769013797e-06,
      "loss": 0.8899,
      "num_input_tokens_seen": 51162030,
      "step": 2419
    },
    {
      "epoch": 0.29098779534660013,
      "flos": 16482076323840.0,
      "grad_norm": 3.478810884173884,
      "learning_rate": 3.327186934047385e-06,
      "loss": 0.9695,
      "num_input_tokens_seen": 51180445,
      "step": 2420
    },
    {
      "epoch": 0.29110803823723924,
      "flos": 10817320611840.0,
      "grad_norm": 5.184876295506676,
      "learning_rate": 3.3266040898601877e-06,
      "loss": 0.8395,
      "num_input_tokens_seen": 51198000,
      "step": 2421
    },
    {
      "epoch": 0.2912282811278783,
      "flos": 16056528384000.0,
      "grad_norm": 3.1359045888925268,
      "learning_rate": 3.3260210444282045e-06,
      "loss": 0.9653,
      "num_input_tokens_seen": 51215675,
      "step": 2422
    },
    {
      "epoch": 0.2913485240185174,
      "flos": 17424567152640.0,
      "grad_norm": 3.765235722471618,
      "learning_rate": 3.325437797839883e-06,
      "loss": 0.9186,
      "num_input_tokens_seen": 51233765,
      "step": 2423
    },
    {
      "epoch": 0.2914687669091565,
      "flos": 12705153515520.0,
      "grad_norm": 4.5497014545127366,
      "learning_rate": 3.3248543501837015e-06,
      "loss": 0.9371,
      "num_input_tokens_seen": 51250690,
      "step": 2424
    },
    {
      "epoch": 0.2915890097997956,
      "flos": 16010003128320.0,
      "grad_norm": 4.566918937416409,
      "learning_rate": 3.3242707015481684e-06,
      "loss": 0.9711,
      "num_input_tokens_seen": 51270345,
      "step": 2425
    },
    {
      "epoch": 0.2917092526904347,
      "flos": 9768841359360.0,
      "grad_norm": 2.6132994053869996,
      "learning_rate": 3.323686852021823e-06,
      "loss": 1.0052,
      "num_input_tokens_seen": 51287575,
      "step": 2426
    },
    {
      "epoch": 0.2918294955810738,
      "flos": 16114550599680.0,
      "grad_norm": 4.252611816299701,
      "learning_rate": 3.323102801693235e-06,
      "loss": 0.9926,
      "num_input_tokens_seen": 51306060,
      "step": 2427
    },
    {
      "epoch": 0.29194973847171285,
      "flos": 16662834647040.0,
      "grad_norm": 6.978167410148078,
      "learning_rate": 3.322518550651003e-06,
      "loss": 0.9869,
      "num_input_tokens_seen": 51325090,
      "step": 2428
    },
    {
      "epoch": 0.29206998136235196,
      "flos": 15564243087360.0,
      "grad_norm": 25.084762312865728,
      "learning_rate": 3.3219340989837586e-06,
      "loss": 1.0021,
      "num_input_tokens_seen": 51344800,
      "step": 2429
    },
    {
      "epoch": 0.292190224252991,
      "flos": 16502564352000.0,
      "grad_norm": 3.3165126464455406,
      "learning_rate": 3.3213494467801625e-06,
      "loss": 0.9983,
      "num_input_tokens_seen": 51363695,
      "step": 2430
    },
    {
      "epoch": 0.2923104671436301,
      "flos": 14724552192000.0,
      "grad_norm": 5.288328043713325,
      "learning_rate": 3.3207645941289063e-06,
      "loss": 0.9083,
      "num_input_tokens_seen": 51381760,
      "step": 2431
    },
    {
      "epoch": 0.29243071003426924,
      "flos": 25553807093760.0,
      "grad_norm": 6.810157303078673,
      "learning_rate": 3.320179541118711e-06,
      "loss": 0.9988,
      "num_input_tokens_seen": 51403980,
      "step": 2432
    },
    {
      "epoch": 0.2925509529249083,
      "flos": 41567726407680.0,
      "grad_norm": 1.0494518733991538,
      "learning_rate": 3.3195942878383293e-06,
      "loss": 0.8389,
      "num_input_tokens_seen": 51459800,
      "step": 2433
    },
    {
      "epoch": 0.2926711958155474,
      "flos": 15196165509120.0,
      "grad_norm": 4.4715893668195985,
      "learning_rate": 3.319008834376543e-06,
      "loss": 0.9732,
      "num_input_tokens_seen": 51479210,
      "step": 2434
    },
    {
      "epoch": 0.2927914387061865,
      "flos": 16481769738240.0,
      "grad_norm": 4.387386118397183,
      "learning_rate": 3.3184231808221654e-06,
      "loss": 1.0744,
      "num_input_tokens_seen": 51493255,
      "step": 2435
    },
    {
      "epoch": 0.29291168159682557,
      "flos": 15956886282240.0,
      "grad_norm": 4.186406462494279,
      "learning_rate": 3.3178373272640394e-06,
      "loss": 0.8181,
      "num_input_tokens_seen": 51512070,
      "step": 2436
    },
    {
      "epoch": 0.2930319244874647,
      "flos": 15032369479680.0,
      "grad_norm": 3.850622716307679,
      "learning_rate": 3.3172512737910387e-06,
      "loss": 1.0636,
      "num_input_tokens_seen": 51529300,
      "step": 2437
    },
    {
      "epoch": 0.2931521673781038,
      "flos": 22354761953280.0,
      "grad_norm": 9.497941446631097,
      "learning_rate": 3.3166650204920674e-06,
      "loss": 1.0778,
      "num_input_tokens_seen": 51550190,
      "step": 2438
    },
    {
      "epoch": 0.29327241026874284,
      "flos": 17210505523200.0,
      "grad_norm": 5.464339914062003,
      "learning_rate": 3.316078567456059e-06,
      "loss": 1.0217,
      "num_input_tokens_seen": 51567750,
      "step": 2439
    },
    {
      "epoch": 0.29339265315938196,
      "flos": 17241110876160.0,
      "grad_norm": 3.119243049253678,
      "learning_rate": 3.3154919147719786e-06,
      "loss": 0.9549,
      "num_input_tokens_seen": 51588485,
      "step": 2440
    },
    {
      "epoch": 0.29351289605002107,
      "flos": 11997794856960.0,
      "grad_norm": 3.9952182798945923,
      "learning_rate": 3.31490506252882e-06,
      "loss": 1.0666,
      "num_input_tokens_seen": 51607585,
      "step": 2441
    },
    {
      "epoch": 0.2936331389406601,
      "flos": 14069359964160.0,
      "grad_norm": 3.491521511535499,
      "learning_rate": 3.31431801081561e-06,
      "loss": 1.0444,
      "num_input_tokens_seen": 51626240,
      "step": 2442
    },
    {
      "epoch": 0.29375338183129923,
      "flos": 51149011353600.0,
      "grad_norm": 0.9552404035691708,
      "learning_rate": 3.313730759721402e-06,
      "loss": 0.8823,
      "num_input_tokens_seen": 51688890,
      "step": 2443
    },
    {
      "epoch": 0.29387362472193834,
      "flos": 15668698583040.0,
      "grad_norm": 4.150477021159595,
      "learning_rate": 3.313143309335282e-06,
      "loss": 1.0497,
      "num_input_tokens_seen": 51707100,
      "step": 2444
    },
    {
      "epoch": 0.2939938676125774,
      "flos": 16324044103680.0,
      "grad_norm": 3.652422735818718,
      "learning_rate": 3.3125556597463665e-06,
      "loss": 1.0327,
      "num_input_tokens_seen": 51726125,
      "step": 2445
    },
    {
      "epoch": 0.2941141105032165,
      "flos": 22355497758720.0,
      "grad_norm": 5.445357685093195,
      "learning_rate": 3.311967811043801e-06,
      "loss": 0.852,
      "num_input_tokens_seen": 51747765,
      "step": 2446
    },
    {
      "epoch": 0.29423435339385556,
      "flos": 16507868282880.0,
      "grad_norm": 4.681450564234976,
      "learning_rate": 3.3113797633167617e-06,
      "loss": 0.993,
      "num_input_tokens_seen": 51765780,
      "step": 2447
    },
    {
      "epoch": 0.2943545962844947,
      "flos": 19124344995840.0,
      "grad_norm": 4.88797227377762,
      "learning_rate": 3.310791516654455e-06,
      "loss": 0.8764,
      "num_input_tokens_seen": 51782560,
      "step": 2448
    },
    {
      "epoch": 0.2944748391751338,
      "flos": 14358865981440.0,
      "grad_norm": 7.600016955306565,
      "learning_rate": 3.3102030711461177e-06,
      "loss": 0.9926,
      "num_input_tokens_seen": 51801855,
      "step": 2449
    },
    {
      "epoch": 0.29459508206577284,
      "flos": 11289332490240.0,
      "grad_norm": 9.235348619491045,
      "learning_rate": 3.3096144268810156e-06,
      "loss": 0.8728,
      "num_input_tokens_seen": 51820335,
      "step": 2450
    },
    {
      "epoch": 0.29471532495641195,
      "flos": 14488929484800.0,
      "grad_norm": 4.543485971201827,
      "learning_rate": 3.3090255839484462e-06,
      "loss": 0.9147,
      "num_input_tokens_seen": 51838050,
      "step": 2451
    },
    {
      "epoch": 0.29483556784705106,
      "flos": 14462861598720.0,
      "grad_norm": 4.980606615738909,
      "learning_rate": 3.3084365424377366e-06,
      "loss": 1.0436,
      "num_input_tokens_seen": 51856535,
      "step": 2452
    },
    {
      "epoch": 0.2949558107376901,
      "flos": 49090668195840.0,
      "grad_norm": 0.8313464001278981,
      "learning_rate": 3.307847302438245e-06,
      "loss": 0.7795,
      "num_input_tokens_seen": 51910235,
      "step": 2453
    },
    {
      "epoch": 0.2950760536283292,
      "flos": 11394370498560.0,
      "grad_norm": 4.87733462258126,
      "learning_rate": 3.3072578640393562e-06,
      "loss": 0.9626,
      "num_input_tokens_seen": 51927290,
      "step": 2454
    },
    {
      "epoch": 0.29519629651896834,
      "flos": 14539471011840.0,
      "grad_norm": 4.233066708636379,
      "learning_rate": 3.3066682273304886e-06,
      "loss": 0.9861,
      "num_input_tokens_seen": 51944655,
      "step": 2455
    },
    {
      "epoch": 0.2953165394096074,
      "flos": 13413431930880.0,
      "grad_norm": 6.453938356732268,
      "learning_rate": 3.3060783924010904e-06,
      "loss": 0.9624,
      "num_input_tokens_seen": 51962300,
      "step": 2456
    },
    {
      "epoch": 0.2954367823002465,
      "flos": 14640002211840.0,
      "grad_norm": 13.29737818984001,
      "learning_rate": 3.3054883593406387e-06,
      "loss": 1.0478,
      "num_input_tokens_seen": 51976770,
      "step": 2457
    },
    {
      "epoch": 0.2955570251908856,
      "flos": 22223748034560.0,
      "grad_norm": 4.55780540259132,
      "learning_rate": 3.3048981282386404e-06,
      "loss": 0.8287,
      "num_input_tokens_seen": 51997800,
      "step": 2458
    },
    {
      "epoch": 0.29567726808152467,
      "flos": 15377843589120.0,
      "grad_norm": 4.372790405331026,
      "learning_rate": 3.304307699184634e-06,
      "loss": 1.0143,
      "num_input_tokens_seen": 52016110,
      "step": 2459
    },
    {
      "epoch": 0.2957975109721638,
      "flos": 17242183925760.0,
      "grad_norm": 6.544687274313012,
      "learning_rate": 3.3037170722681866e-06,
      "loss": 0.9878,
      "num_input_tokens_seen": 52036665,
      "step": 2460
    },
    {
      "epoch": 0.29591775386280283,
      "flos": 9506077716480.0,
      "grad_norm": 15.518193549981502,
      "learning_rate": 3.3031262475788956e-06,
      "loss": 0.8799,
      "num_input_tokens_seen": 52053325,
      "step": 2461
    },
    {
      "epoch": 0.29603799675344195,
      "flos": 12575059353600.0,
      "grad_norm": 4.3412305811813745,
      "learning_rate": 3.3025352252063897e-06,
      "loss": 0.9274,
      "num_input_tokens_seen": 52071740,
      "step": 2462
    },
    {
      "epoch": 0.29615823964408106,
      "flos": 16192876892160.0,
      "grad_norm": 5.101701953611787,
      "learning_rate": 3.3019440052403252e-06,
      "loss": 0.9389,
      "num_input_tokens_seen": 52091325,
      "step": 2463
    },
    {
      "epoch": 0.2962784825347201,
      "flos": 16717791006720.0,
      "grad_norm": 6.763246889171407,
      "learning_rate": 3.30135258777039e-06,
      "loss": 0.9014,
      "num_input_tokens_seen": 52110415,
      "step": 2464
    },
    {
      "epoch": 0.2963987254253592,
      "flos": 11578164019200.0,
      "grad_norm": 4.373469929849172,
      "learning_rate": 3.3007609728863024e-06,
      "loss": 0.8956,
      "num_input_tokens_seen": 52128225,
      "step": 2465
    },
    {
      "epoch": 0.29651896831599833,
      "flos": 23874118717440.0,
      "grad_norm": 7.667534433541905,
      "learning_rate": 3.300169160677809e-06,
      "loss": 0.9277,
      "num_input_tokens_seen": 52151860,
      "step": 2466
    },
    {
      "epoch": 0.2966392112066374,
      "flos": 16926947266560.0,
      "grad_norm": 7.049114404341728,
      "learning_rate": 3.2995771512346878e-06,
      "loss": 0.9468,
      "num_input_tokens_seen": 52169930,
      "step": 2467
    },
    {
      "epoch": 0.2967594540972765,
      "flos": 14147962183680.0,
      "grad_norm": 6.293352452804057,
      "learning_rate": 3.298984944646746e-06,
      "loss": 0.9142,
      "num_input_tokens_seen": 52188330,
      "step": 2468
    },
    {
      "epoch": 0.2968796969879156,
      "flos": 16874872811520.0,
      "grad_norm": 4.9245983521556065,
      "learning_rate": 3.298392541003822e-06,
      "loss": 1.0074,
      "num_input_tokens_seen": 52207455,
      "step": 2469
    },
    {
      "epoch": 0.29699993987855466,
      "flos": 16271816355840.0,
      "grad_norm": 3.754919639773888,
      "learning_rate": 3.2977999403957806e-06,
      "loss": 1.085,
      "num_input_tokens_seen": 52225935,
      "step": 2470
    },
    {
      "epoch": 0.2971201827691938,
      "flos": 24132436869120.0,
      "grad_norm": 4.583222867617071,
      "learning_rate": 3.2972071429125207e-06,
      "loss": 0.8596,
      "num_input_tokens_seen": 52246875,
      "step": 2471
    },
    {
      "epoch": 0.2972404256598329,
      "flos": 15668575948800.0,
      "grad_norm": 7.510945280803906,
      "learning_rate": 3.2966141486439682e-06,
      "loss": 1.0787,
      "num_input_tokens_seen": 52265785,
      "step": 2472
    },
    {
      "epoch": 0.29736066855047194,
      "flos": 22800951214080.0,
      "grad_norm": 5.845016587024189,
      "learning_rate": 3.29602095768008e-06,
      "loss": 0.846,
      "num_input_tokens_seen": 52286020,
      "step": 2473
    },
    {
      "epoch": 0.29748091144111105,
      "flos": 23901750190080.0,
      "grad_norm": 3.906957089323473,
      "learning_rate": 3.2954275701108437e-06,
      "loss": 0.8322,
      "num_input_tokens_seen": 52306920,
      "step": 2474
    },
    {
      "epoch": 0.29760115433175016,
      "flos": 29488240926720.0,
      "grad_norm": 3.0153196506373945,
      "learning_rate": 3.294833986026275e-06,
      "loss": 0.8818,
      "num_input_tokens_seen": 52329880,
      "step": 2475
    },
    {
      "epoch": 0.2977213972223892,
      "flos": 17420673515520.0,
      "grad_norm": 5.73177734078837,
      "learning_rate": 3.29424020551642e-06,
      "loss": 1.0431,
      "num_input_tokens_seen": 52348235,
      "step": 2476
    },
    {
      "epoch": 0.2978416401130283,
      "flos": 15115325214720.0,
      "grad_norm": 4.179621102349363,
      "learning_rate": 3.2936462286713546e-06,
      "loss": 0.9041,
      "num_input_tokens_seen": 52366305,
      "step": 2477
    },
    {
      "epoch": 0.2979618830036674,
      "flos": 18339978362880.0,
      "grad_norm": 4.593631665009647,
      "learning_rate": 3.2930520555811846e-06,
      "loss": 0.9636,
      "num_input_tokens_seen": 52385650,
      "step": 2478
    },
    {
      "epoch": 0.2980821258943065,
      "flos": 16690956656640.0,
      "grad_norm": 3.6152647110930083,
      "learning_rate": 3.292457686336046e-06,
      "loss": 0.9871,
      "num_input_tokens_seen": 52404690,
      "step": 2479
    },
    {
      "epoch": 0.2982023687849456,
      "flos": 49954610012160.0,
      "grad_norm": 0.8546912435557726,
      "learning_rate": 3.291863121026105e-06,
      "loss": 0.8446,
      "num_input_tokens_seen": 52468190,
      "step": 2480
    },
    {
      "epoch": 0.29832261167558466,
      "flos": 21253871001600.0,
      "grad_norm": 3.4933533909151473,
      "learning_rate": 3.2912683597415547e-06,
      "loss": 0.9513,
      "num_input_tokens_seen": 52491995,
      "step": 2481
    },
    {
      "epoch": 0.29844285456622377,
      "flos": 24188834181120.0,
      "grad_norm": 6.124122401055148,
      "learning_rate": 3.2906734025726213e-06,
      "loss": 0.9666,
      "num_input_tokens_seen": 52510980,
      "step": 2482
    },
    {
      "epoch": 0.2985630974568629,
      "flos": 16978193940480.0,
      "grad_norm": 4.028145474425516,
      "learning_rate": 3.290078249609559e-06,
      "loss": 1.0614,
      "num_input_tokens_seen": 52530120,
      "step": 2483
    },
    {
      "epoch": 0.29868334034750194,
      "flos": 15485150330880.0,
      "grad_norm": 5.15290455806446,
      "learning_rate": 3.2894829009426514e-06,
      "loss": 1.0585,
      "num_input_tokens_seen": 52547675,
      "step": 2484
    },
    {
      "epoch": 0.29880358323814105,
      "flos": 18264687267840.0,
      "grad_norm": 9.571703502228342,
      "learning_rate": 3.288887356662213e-06,
      "loss": 0.9645,
      "num_input_tokens_seen": 52568730,
      "step": 2485
    },
    {
      "epoch": 0.29892382612878016,
      "flos": 50852331233280.0,
      "grad_norm": 1.4803544053179423,
      "learning_rate": 3.288291616858588e-06,
      "loss": 0.806,
      "num_input_tokens_seen": 52623840,
      "step": 2486
    },
    {
      "epoch": 0.2990440690194192,
      "flos": 18130944737280.0,
      "grad_norm": 7.18869618952376,
      "learning_rate": 3.287695681622149e-06,
      "loss": 0.9694,
      "num_input_tokens_seen": 52642910,
      "step": 2487
    },
    {
      "epoch": 0.2991643119100583,
      "flos": 16874474250240.0,
      "grad_norm": 3.5072596056909484,
      "learning_rate": 3.2870995510432982e-06,
      "loss": 1.0057,
      "num_input_tokens_seen": 52661110,
      "step": 2488
    },
    {
      "epoch": 0.29928455480069743,
      "flos": 19522966609920.0,
      "grad_norm": 4.214912986717009,
      "learning_rate": 3.2865032252124697e-06,
      "loss": 0.9578,
      "num_input_tokens_seen": 52681345,
      "step": 2489
    },
    {
      "epoch": 0.2994047976913365,
      "flos": 24032948060160.0,
      "grad_norm": 3.2523718761920386,
      "learning_rate": 3.2859067042201243e-06,
      "loss": 0.9684,
      "num_input_tokens_seen": 52703105,
      "step": 2490
    },
    {
      "epoch": 0.2995250405819756,
      "flos": 11866474352640.0,
      "grad_norm": 5.640869980667339,
      "learning_rate": 3.2853099881567544e-06,
      "loss": 0.9643,
      "num_input_tokens_seen": 52721225,
      "step": 2491
    },
    {
      "epoch": 0.29964528347261465,
      "flos": 16321959321600.0,
      "grad_norm": 5.55864899447688,
      "learning_rate": 3.284713077112881e-06,
      "loss": 0.9813,
      "num_input_tokens_seen": 52740375,
      "step": 2492
    },
    {
      "epoch": 0.29976552636325376,
      "flos": 11990436802560.0,
      "grad_norm": 4.477757266186689,
      "learning_rate": 3.284115971179056e-06,
      "loss": 1.0375,
      "num_input_tokens_seen": 52754125,
      "step": 2493
    },
    {
      "epoch": 0.2998857692538929,
      "flos": 12076764979200.0,
      "grad_norm": 5.481570317337988,
      "learning_rate": 3.283518670445859e-06,
      "loss": 0.99,
      "num_input_tokens_seen": 52771755,
      "step": 2494
    },
    {
      "epoch": 0.30000601214453193,
      "flos": 49291691827200.0,
      "grad_norm": 2.8230346054458706,
      "learning_rate": 3.2829211750038995e-06,
      "loss": 0.7845,
      "num_input_tokens_seen": 52840105,
      "step": 2495
    },
    {
      "epoch": 0.30012625503517104,
      "flos": 12469684101120.0,
      "grad_norm": 3.7432069922776288,
      "learning_rate": 3.2823234849438183e-06,
      "loss": 1.0769,
      "num_input_tokens_seen": 52857860,
      "step": 2496
    },
    {
      "epoch": 0.30024649792581015,
      "flos": 15092476477440.0,
      "grad_norm": 8.82224724822361,
      "learning_rate": 3.2817256003562836e-06,
      "loss": 0.9397,
      "num_input_tokens_seen": 52877955,
      "step": 2497
    },
    {
      "epoch": 0.3003667408164492,
      "flos": 16350571868160.0,
      "grad_norm": 4.453249424317109,
      "learning_rate": 3.281127521331995e-06,
      "loss": 0.8537,
      "num_input_tokens_seen": 52898855,
      "step": 2498
    },
    {
      "epoch": 0.3004869837070883,
      "flos": 45986075750400.0,
      "grad_norm": 0.9107603761352583,
      "learning_rate": 3.2805292479616798e-06,
      "loss": 0.8388,
      "num_input_tokens_seen": 52957440,
      "step": 2499
    },
    {
      "epoch": 0.30060722659772743,
      "flos": 18682723860480.0,
      "grad_norm": 4.892989394515765,
      "learning_rate": 3.2799307803360955e-06,
      "loss": 1.1028,
      "num_input_tokens_seen": 52973090,
      "step": 2500
    },
    {
      "epoch": 0.3007274694883665,
      "flos": 17764308111360.0,
      "grad_norm": 2.9182883704218563,
      "learning_rate": 3.27933211854603e-06,
      "loss": 1.0118,
      "num_input_tokens_seen": 52991865,
      "step": 2501
    },
    {
      "epoch": 0.3008477123790056,
      "flos": 12076243783680.0,
      "grad_norm": 4.505971549183459,
      "learning_rate": 3.278733262682299e-06,
      "loss": 1.0613,
      "num_input_tokens_seen": 53009440,
      "step": 2502
    },
    {
      "epoch": 0.3009679552696447,
      "flos": 15274675752960.0,
      "grad_norm": 5.448100663743756,
      "learning_rate": 3.2781342128357484e-06,
      "loss": 1.0031,
      "num_input_tokens_seen": 53028515,
      "step": 2503
    },
    {
      "epoch": 0.30108819816028376,
      "flos": 15006025666560.0,
      "grad_norm": 7.625259395001881,
      "learning_rate": 3.2775349690972547e-06,
      "loss": 0.991,
      "num_input_tokens_seen": 53042385,
      "step": 2504
    },
    {
      "epoch": 0.30120844105092287,
      "flos": 50940498923520.0,
      "grad_norm": 0.7724877508358937,
      "learning_rate": 3.276935531557722e-06,
      "loss": 0.7559,
      "num_input_tokens_seen": 53107325,
      "step": 2505
    },
    {
      "epoch": 0.301328683941562,
      "flos": 14382021304320.0,
      "grad_norm": 7.095999071295332,
      "learning_rate": 3.2763359003080837e-06,
      "loss": 0.9963,
      "num_input_tokens_seen": 53124000,
      "step": 2506
    },
    {
      "epoch": 0.30144892683220104,
      "flos": 50596281815040.0,
      "grad_norm": 0.9287894904673174,
      "learning_rate": 3.2757360754393047e-06,
      "loss": 0.871,
      "num_input_tokens_seen": 53187790,
      "step": 2507
    },
    {
      "epoch": 0.30156916972284015,
      "flos": 16219220705280.0,
      "grad_norm": 7.817894621850526,
      "learning_rate": 3.2751360570423767e-06,
      "loss": 0.8142,
      "num_input_tokens_seen": 53205895,
      "step": 2508
    },
    {
      "epoch": 0.3016894126134792,
      "flos": 21306558627840.0,
      "grad_norm": 3.877567548132725,
      "learning_rate": 3.2745358452083236e-06,
      "loss": 0.9531,
      "num_input_tokens_seen": 53228515,
      "step": 2509
    },
    {
      "epoch": 0.3018096555041183,
      "flos": 15303840153600.0,
      "grad_norm": 2.3864959762670392,
      "learning_rate": 3.2739354400281955e-06,
      "loss": 1.0191,
      "num_input_tokens_seen": 53249455,
      "step": 2510
    },
    {
      "epoch": 0.3019298983947574,
      "flos": 42325258690560.0,
      "grad_norm": 0.9104178056920752,
      "learning_rate": 3.2733348415930744e-06,
      "loss": 0.8837,
      "num_input_tokens_seen": 53311045,
      "step": 2511
    },
    {
      "epoch": 0.3020501412853965,
      "flos": 24557739540480.0,
      "grad_norm": 5.3205677480988225,
      "learning_rate": 3.27273404999407e-06,
      "loss": 0.9968,
      "num_input_tokens_seen": 53332985,
      "step": 2512
    },
    {
      "epoch": 0.3021703841760356,
      "flos": 50854416015360.0,
      "grad_norm": 0.812027757294722,
      "learning_rate": 3.272133065322322e-06,
      "loss": 0.82,
      "num_input_tokens_seen": 53390975,
      "step": 2513
    },
    {
      "epoch": 0.3022906270666747,
      "flos": 15278201487360.0,
      "grad_norm": 4.937830274134796,
      "learning_rate": 3.271531887669e-06,
      "loss": 0.9838,
      "num_input_tokens_seen": 53410755,
      "step": 2514
    },
    {
      "epoch": 0.30241086995731375,
      "flos": 22191517777920.0,
      "grad_norm": 8.342619287119856,
      "learning_rate": 3.2709305171253015e-06,
      "loss": 0.8332,
      "num_input_tokens_seen": 53430595,
      "step": 2515
    },
    {
      "epoch": 0.30253111284795287,
      "flos": 16715522273280.0,
      "grad_norm": 8.659270382775428,
      "learning_rate": 3.2703289537824536e-06,
      "loss": 0.9717,
      "num_input_tokens_seen": 53450115,
      "step": 2516
    },
    {
      "epoch": 0.302651355738592,
      "flos": 13274906664960.0,
      "grad_norm": 7.510365253612888,
      "learning_rate": 3.269727197731714e-06,
      "loss": 0.9885,
      "num_input_tokens_seen": 53462600,
      "step": 2517
    },
    {
      "epoch": 0.30277159862923103,
      "flos": 15930511810560.0,
      "grad_norm": 11.863513410430029,
      "learning_rate": 3.269125249064367e-06,
      "loss": 0.9732,
      "num_input_tokens_seen": 53482015,
      "step": 2518
    },
    {
      "epoch": 0.30289184151987014,
      "flos": 15825780387840.0,
      "grad_norm": 4.572763890847682,
      "learning_rate": 3.2685231078717297e-06,
      "loss": 1.0217,
      "num_input_tokens_seen": 53501925,
      "step": 2519
    },
    {
      "epoch": 0.30301208441050925,
      "flos": 17947273850880.0,
      "grad_norm": 7.116714210559896,
      "learning_rate": 3.267920774245145e-06,
      "loss": 0.9257,
      "num_input_tokens_seen": 53521050,
      "step": 2520
    },
    {
      "epoch": 0.3031323273011483,
      "flos": 16376363827200.0,
      "grad_norm": 4.257553120683223,
      "learning_rate": 3.2673182482759876e-06,
      "loss": 1.0395,
      "num_input_tokens_seen": 53539885,
      "step": 2521
    },
    {
      "epoch": 0.3032525701917874,
      "flos": 13384390164480.0,
      "grad_norm": 5.702802257028002,
      "learning_rate": 3.266715530055659e-06,
      "loss": 0.8434,
      "num_input_tokens_seen": 53557755,
      "step": 2522
    },
    {
      "epoch": 0.30337281308242653,
      "flos": 12598184017920.0,
      "grad_norm": 6.348304954833262,
      "learning_rate": 3.2661126196755927e-06,
      "loss": 1.0015,
      "num_input_tokens_seen": 53576585,
      "step": 2523
    },
    {
      "epoch": 0.3034930559730656,
      "flos": 41382553251840.0,
      "grad_norm": 0.8933846066893658,
      "learning_rate": 3.265509517227248e-06,
      "loss": 0.8012,
      "num_input_tokens_seen": 53633120,
      "step": 2524
    },
    {
      "epoch": 0.3036132988637047,
      "flos": 10422868561920.0,
      "grad_norm": 3.815674257702504,
      "learning_rate": 3.264906222802115e-06,
      "loss": 0.9928,
      "num_input_tokens_seen": 53650690,
      "step": 2525
    },
    {
      "epoch": 0.30373354175434375,
      "flos": 14934750842880.0,
      "grad_norm": 4.1136860444514305,
      "learning_rate": 3.264302736491715e-06,
      "loss": 0.964,
      "num_input_tokens_seen": 53670530,
      "step": 2526
    },
    {
      "epoch": 0.30385378464498286,
      "flos": 15013904916480.0,
      "grad_norm": 8.80824252695021,
      "learning_rate": 3.263699058387594e-06,
      "loss": 1.0571,
      "num_input_tokens_seen": 53687685,
      "step": 2527
    },
    {
      "epoch": 0.30397402753562197,
      "flos": 14644079800320.0,
      "grad_norm": 9.474690503221513,
      "learning_rate": 3.2630951885813315e-06,
      "loss": 1.0972,
      "num_input_tokens_seen": 53704800,
      "step": 2528
    },
    {
      "epoch": 0.304094270426261,
      "flos": 10660330782720.0,
      "grad_norm": 5.829499356217532,
      "learning_rate": 3.262491127164533e-06,
      "loss": 0.9693,
      "num_input_tokens_seen": 53723335,
      "step": 2529
    },
    {
      "epoch": 0.30421451331690014,
      "flos": 9768841359360.0,
      "grad_norm": 7.684561811474679,
      "learning_rate": 3.2618868742288337e-06,
      "loss": 0.9947,
      "num_input_tokens_seen": 53739980,
      "step": 2530
    },
    {
      "epoch": 0.30433475620753925,
      "flos": 12312234393600.0,
      "grad_norm": 18.507671518300103,
      "learning_rate": 3.261282429865899e-06,
      "loss": 0.9215,
      "num_input_tokens_seen": 53757705,
      "step": 2531
    },
    {
      "epoch": 0.3044549990981783,
      "flos": 13413830492160.0,
      "grad_norm": 3.1553230057426527,
      "learning_rate": 3.2606777941674225e-06,
      "loss": 0.9215,
      "num_input_tokens_seen": 53776080,
      "step": 2532
    },
    {
      "epoch": 0.3045752419888174,
      "flos": 14959898972160.0,
      "grad_norm": 10.313156199529727,
      "learning_rate": 3.2600729672251276e-06,
      "loss": 1.04,
      "num_input_tokens_seen": 53793515,
      "step": 2533
    },
    {
      "epoch": 0.3046954848794565,
      "flos": 20729600716800.0,
      "grad_norm": 12.83307941400246,
      "learning_rate": 3.259467949130765e-06,
      "loss": 0.8533,
      "num_input_tokens_seen": 53814645,
      "step": 2534
    },
    {
      "epoch": 0.3048157277700956,
      "flos": 14403367772160.0,
      "grad_norm": 6.002957568795061,
      "learning_rate": 3.2588627399761164e-06,
      "loss": 1.0302,
      "num_input_tokens_seen": 53830360,
      "step": 2535
    },
    {
      "epoch": 0.3049359706607347,
      "flos": 16160248074240.0,
      "grad_norm": 5.550071117873369,
      "learning_rate": 3.2582573398529903e-06,
      "loss": 0.9074,
      "num_input_tokens_seen": 53847435,
      "step": 2536
    },
    {
      "epoch": 0.3050562135513738,
      "flos": 13066118307840.0,
      "grad_norm": 8.53196316487462,
      "learning_rate": 3.2576517488532265e-06,
      "loss": 0.9447,
      "num_input_tokens_seen": 53863505,
      "step": 2537
    },
    {
      "epoch": 0.30517645644201286,
      "flos": 14457527009280.0,
      "grad_norm": 4.955387597499824,
      "learning_rate": 3.257045967068692e-06,
      "loss": 1.0598,
      "num_input_tokens_seen": 53882480,
      "step": 2538
    },
    {
      "epoch": 0.30529669933265197,
      "flos": 15590096363520.0,
      "grad_norm": 8.0544649050838,
      "learning_rate": 3.2564399945912848e-06,
      "loss": 1.0088,
      "num_input_tokens_seen": 53901990,
      "step": 2539
    },
    {
      "epoch": 0.305416942223291,
      "flos": 15511218216960.0,
      "grad_norm": 7.513522609480445,
      "learning_rate": 3.2558338315129287e-06,
      "loss": 1.0167,
      "num_input_tokens_seen": 53919855,
      "step": 2540
    },
    {
      "epoch": 0.30553718511393013,
      "flos": 24190244474880.0,
      "grad_norm": 4.129412054372259,
      "learning_rate": 3.2552274779255785e-06,
      "loss": 0.952,
      "num_input_tokens_seen": 53940505,
      "step": 2541
    },
    {
      "epoch": 0.30565742800456924,
      "flos": 15821856092160.0,
      "grad_norm": 7.4342790569385775,
      "learning_rate": 3.2546209339212184e-06,
      "loss": 0.9527,
      "num_input_tokens_seen": 53959245,
      "step": 2542
    },
    {
      "epoch": 0.3057776708952083,
      "flos": 16035580477440.0,
      "grad_norm": 7.222597624322644,
      "learning_rate": 3.25401419959186e-06,
      "loss": 0.9609,
      "num_input_tokens_seen": 53979575,
      "step": 2543
    },
    {
      "epoch": 0.3058979137858474,
      "flos": 15485242306560.0,
      "grad_norm": 5.056142889333568,
      "learning_rate": 3.253407275029545e-06,
      "loss": 0.9571,
      "num_input_tokens_seen": 53998200,
      "step": 2544
    },
    {
      "epoch": 0.3060181566764865,
      "flos": 19208067194880.0,
      "grad_norm": 7.708217833057472,
      "learning_rate": 3.2528001603263425e-06,
      "loss": 0.9967,
      "num_input_tokens_seen": 54019990,
      "step": 2545
    },
    {
      "epoch": 0.3061383995671256,
      "flos": 14095182581760.0,
      "grad_norm": 5.274236231466825,
      "learning_rate": 3.2521928555743514e-06,
      "loss": 1.0034,
      "num_input_tokens_seen": 54037055,
      "step": 2546
    },
    {
      "epoch": 0.3062586424577647,
      "flos": 15721171599360.0,
      "grad_norm": 5.399467314831672,
      "learning_rate": 3.2515853608657e-06,
      "loss": 0.8582,
      "num_input_tokens_seen": 54054775,
      "step": 2547
    },
    {
      "epoch": 0.3063788853484038,
      "flos": 14799138140160.0,
      "grad_norm": 11.876673530307542,
      "learning_rate": 3.250977676292545e-06,
      "loss": 0.9248,
      "num_input_tokens_seen": 54072735,
      "step": 2548
    },
    {
      "epoch": 0.30649912823904285,
      "flos": 11467208908800.0,
      "grad_norm": 4.369291450444873,
      "learning_rate": 3.2503698019470712e-06,
      "loss": 0.9837,
      "num_input_tokens_seen": 54088225,
      "step": 2549
    },
    {
      "epoch": 0.30661937112968196,
      "flos": 13197868032000.0,
      "grad_norm": 6.021589854229412,
      "learning_rate": 3.249761737921492e-06,
      "loss": 0.9692,
      "num_input_tokens_seen": 54104475,
      "step": 2550
    },
    {
      "epoch": 0.30673961402032107,
      "flos": 22377978593280.0,
      "grad_norm": 18.05828322453655,
      "learning_rate": 3.249153484308051e-06,
      "loss": 0.9342,
      "num_input_tokens_seen": 54122810,
      "step": 2551
    },
    {
      "epoch": 0.3068598569109601,
      "flos": 14357271736320.0,
      "grad_norm": 7.474783658880725,
      "learning_rate": 3.2485450411990194e-06,
      "loss": 0.9689,
      "num_input_tokens_seen": 54141885,
      "step": 2552
    },
    {
      "epoch": 0.30698009980159924,
      "flos": 21092251729920.0,
      "grad_norm": 5.806411343438302,
      "learning_rate": 3.2479364086866983e-06,
      "loss": 1.0109,
      "num_input_tokens_seen": 54161860,
      "step": 2553
    },
    {
      "epoch": 0.30710034269223835,
      "flos": 16953658982400.0,
      "grad_norm": 7.353750618755153,
      "learning_rate": 3.247327586863416e-06,
      "loss": 1.0073,
      "num_input_tokens_seen": 54182460,
      "step": 2554
    },
    {
      "epoch": 0.3072205855828774,
      "flos": 18421523804160.0,
      "grad_norm": 7.704239399059298,
      "learning_rate": 3.2467185758215304e-06,
      "loss": 0.9768,
      "num_input_tokens_seen": 54201920,
      "step": 2555
    },
    {
      "epoch": 0.3073408284735165,
      "flos": 15799344599040.0,
      "grad_norm": 7.66277474288876,
      "learning_rate": 3.246109375653428e-06,
      "loss": 1.0335,
      "num_input_tokens_seen": 54218405,
      "step": 2556
    },
    {
      "epoch": 0.30746107136415557,
      "flos": 13833001451520.0,
      "grad_norm": 6.2280908716685985,
      "learning_rate": 3.2454999864515243e-06,
      "loss": 0.99,
      "num_input_tokens_seen": 54237500,
      "step": 2557
    },
    {
      "epoch": 0.3075813142547947,
      "flos": 15431573606400.0,
      "grad_norm": 11.077400130000573,
      "learning_rate": 3.244890408308263e-06,
      "loss": 0.8808,
      "num_input_tokens_seen": 54257925,
      "step": 2558
    },
    {
      "epoch": 0.3077015571454338,
      "flos": 17137329868800.0,
      "grad_norm": 4.966753672578033,
      "learning_rate": 3.2442806413161165e-06,
      "loss": 0.8,
      "num_input_tokens_seen": 54277290,
      "step": 2559
    },
    {
      "epoch": 0.30782180003607285,
      "flos": 13046151475200.0,
      "grad_norm": 12.395851524190084,
      "learning_rate": 3.243670685567586e-06,
      "loss": 0.9638,
      "num_input_tokens_seen": 54294410,
      "step": 2560
    },
    {
      "epoch": 0.30794204292671196,
      "flos": 16979573575680.0,
      "grad_norm": 5.335896571436645,
      "learning_rate": 3.2430605411552012e-06,
      "loss": 0.9862,
      "num_input_tokens_seen": 54314245,
      "step": 2561
    },
    {
      "epoch": 0.30806228581735107,
      "flos": 48643336458240.0,
      "grad_norm": 0.9287071829638717,
      "learning_rate": 3.2424502081715205e-06,
      "loss": 0.9133,
      "num_input_tokens_seen": 54377080,
      "step": 2562
    },
    {
      "epoch": 0.3081825287079901,
      "flos": 16954333470720.0,
      "grad_norm": 7.2533419568814645,
      "learning_rate": 3.241839686709132e-06,
      "loss": 0.9695,
      "num_input_tokens_seen": 54397735,
      "step": 2563
    },
    {
      "epoch": 0.30830277159862923,
      "flos": 11467423518720.0,
      "grad_norm": 5.822223365013522,
      "learning_rate": 3.2412289768606495e-06,
      "loss": 1.0051,
      "num_input_tokens_seen": 54414025,
      "step": 2564
    },
    {
      "epoch": 0.30842301448926834,
      "flos": 20911186821120.0,
      "grad_norm": 4.635790840505959,
      "learning_rate": 3.240618078718718e-06,
      "loss": 1.0095,
      "num_input_tokens_seen": 54435205,
      "step": 2565
    },
    {
      "epoch": 0.3085432573799074,
      "flos": 15590157680640.0,
      "grad_norm": 5.043165420254891,
      "learning_rate": 3.240006992376011e-06,
      "loss": 0.9271,
      "num_input_tokens_seen": 54454550,
      "step": 2566
    },
    {
      "epoch": 0.3086635002705465,
      "flos": 15668882534400.0,
      "grad_norm": 6.781536606036545,
      "learning_rate": 3.2393957179252284e-06,
      "loss": 0.9577,
      "num_input_tokens_seen": 54470805,
      "step": 2567
    },
    {
      "epoch": 0.3087837431611856,
      "flos": 23294125608960.0,
      "grad_norm": 6.133705647543182,
      "learning_rate": 3.2387842554591016e-06,
      "loss": 1.001,
      "num_input_tokens_seen": 54491340,
      "step": 2568
    },
    {
      "epoch": 0.3089039860518247,
      "flos": 12466250342400.0,
      "grad_norm": 9.489100578303312,
      "learning_rate": 3.238172605070388e-06,
      "loss": 1.0545,
      "num_input_tokens_seen": 54506475,
      "step": 2569
    },
    {
      "epoch": 0.3090242289424638,
      "flos": 10154770329600.0,
      "grad_norm": 7.576740067553429,
      "learning_rate": 3.2375607668518745e-06,
      "loss": 0.9725,
      "num_input_tokens_seen": 54519230,
      "step": 2570
    },
    {
      "epoch": 0.30914447183310284,
      "flos": 11366463098880.0,
      "grad_norm": 8.567620076166499,
      "learning_rate": 3.236948740896377e-06,
      "loss": 1.0992,
      "num_input_tokens_seen": 54533750,
      "step": 2571
    },
    {
      "epoch": 0.30926471472374195,
      "flos": 22981280317440.0,
      "grad_norm": 5.1049451377080235,
      "learning_rate": 3.2363365272967384e-06,
      "loss": 1.0319,
      "num_input_tokens_seen": 54556040,
      "step": 2572
    },
    {
      "epoch": 0.30938495761438106,
      "flos": 14457864253440.0,
      "grad_norm": 4.9307115461481175,
      "learning_rate": 3.235724126145832e-06,
      "loss": 1.01,
      "num_input_tokens_seen": 54571795,
      "step": 2573
    },
    {
      "epoch": 0.3095052005050201,
      "flos": 17110832762880.0,
      "grad_norm": 5.407313953290776,
      "learning_rate": 3.235111537536558e-06,
      "loss": 0.9601,
      "num_input_tokens_seen": 54592330,
      "step": 2574
    },
    {
      "epoch": 0.30962544339565923,
      "flos": 16636398858240.0,
      "grad_norm": 6.380330867498939,
      "learning_rate": 3.2344987615618456e-06,
      "loss": 1.0298,
      "num_input_tokens_seen": 54611885,
      "step": 2575
    },
    {
      "epoch": 0.30974568628629834,
      "flos": 24109557473280.0,
      "grad_norm": 11.061050233833614,
      "learning_rate": 3.2338857983146533e-06,
      "loss": 0.9657,
      "num_input_tokens_seen": 54633105,
      "step": 2576
    },
    {
      "epoch": 0.3098659291769374,
      "flos": 14357087784960.0,
      "grad_norm": 7.568831890491652,
      "learning_rate": 3.233272647887966e-06,
      "loss": 0.9535,
      "num_input_tokens_seen": 54651715,
      "step": 2577
    },
    {
      "epoch": 0.3099861720675765,
      "flos": 17604068474880.0,
      "grad_norm": 5.7829458687716055,
      "learning_rate": 3.2326593103747985e-06,
      "loss": 1.0877,
      "num_input_tokens_seen": 54670450,
      "step": 2578
    },
    {
      "epoch": 0.3101064149582156,
      "flos": 8274172846080.0,
      "grad_norm": 6.960674447891615,
      "learning_rate": 3.2320457858681936e-06,
      "loss": 1.0408,
      "num_input_tokens_seen": 54688560,
      "step": 2579
    },
    {
      "epoch": 0.31022665784885467,
      "flos": 16371243847680.0,
      "grad_norm": 7.549239710238159,
      "learning_rate": 3.2314320744612228e-06,
      "loss": 1.0601,
      "num_input_tokens_seen": 54703580,
      "step": 2580
    },
    {
      "epoch": 0.3103469007394938,
      "flos": 11525200465920.0,
      "grad_norm": 4.472473513938986,
      "learning_rate": 3.2308181762469854e-06,
      "loss": 0.9598,
      "num_input_tokens_seen": 54721245,
      "step": 2581
    },
    {
      "epoch": 0.3104671436301329,
      "flos": 21748762275840.0,
      "grad_norm": 4.8322846601396625,
      "learning_rate": 3.230204091318609e-06,
      "loss": 0.9751,
      "num_input_tokens_seen": 54741505,
      "step": 2582
    },
    {
      "epoch": 0.31058738652077195,
      "flos": 14226043207680.0,
      "grad_norm": 8.747443997491942,
      "learning_rate": 3.2295898197692503e-06,
      "loss": 1.04,
      "num_input_tokens_seen": 54760185,
      "step": 2583
    },
    {
      "epoch": 0.31070762941141106,
      "flos": 19995254415360.0,
      "grad_norm": 6.807784101852922,
      "learning_rate": 3.228975361692094e-06,
      "loss": 0.9829,
      "num_input_tokens_seen": 54780925,
      "step": 2584
    },
    {
      "epoch": 0.31082787230205017,
      "flos": 14567225118720.0,
      "grad_norm": 11.648150618952215,
      "learning_rate": 3.228360717180352e-06,
      "loss": 0.9935,
      "num_input_tokens_seen": 54798170,
      "step": 2585
    },
    {
      "epoch": 0.3109481151926892,
      "flos": 44702403010560.0,
      "grad_norm": 0.8646008171856857,
      "learning_rate": 3.227745886327266e-06,
      "loss": 0.8369,
      "num_input_tokens_seen": 54856585,
      "step": 2586
    },
    {
      "epoch": 0.31106835808332833,
      "flos": 31981850787840.0,
      "grad_norm": 1.0167290188997877,
      "learning_rate": 3.227130869226105e-06,
      "loss": 0.7825,
      "num_input_tokens_seen": 54913465,
      "step": 2587
    },
    {
      "epoch": 0.3111886009739674,
      "flos": 16637686517760.0,
      "grad_norm": 11.414847198813728,
      "learning_rate": 3.226515665970167e-06,
      "loss": 1.0076,
      "num_input_tokens_seen": 54930725,
      "step": 2588
    },
    {
      "epoch": 0.3113088438646065,
      "flos": 12417609646080.0,
      "grad_norm": 13.307681075617628,
      "learning_rate": 3.225900276652777e-06,
      "loss": 1.0456,
      "num_input_tokens_seen": 54947220,
      "step": 2589
    },
    {
      "epoch": 0.3114290867552456,
      "flos": 20204410675200.0,
      "grad_norm": 4.564910669862897,
      "learning_rate": 3.2252847013672906e-06,
      "loss": 0.9479,
      "num_input_tokens_seen": 54969600,
      "step": 2590
    },
    {
      "epoch": 0.31154932964588467,
      "flos": 19495120527360.0,
      "grad_norm": 9.599772100922419,
      "learning_rate": 3.224668940207089e-06,
      "loss": 0.9595,
      "num_input_tokens_seen": 54988305,
      "step": 2591
    },
    {
      "epoch": 0.3116695725365238,
      "flos": 18893627658240.0,
      "grad_norm": 4.779903446642884,
      "learning_rate": 3.2240529932655828e-06,
      "loss": 1.059,
      "num_input_tokens_seen": 55007290,
      "step": 2592
    },
    {
      "epoch": 0.3117898154271629,
      "flos": 15038317240320.0,
      "grad_norm": 10.850270182677912,
      "learning_rate": 3.223436860636211e-06,
      "loss": 1.0646,
      "num_input_tokens_seen": 55022645,
      "step": 2593
    },
    {
      "epoch": 0.31191005831780194,
      "flos": 19418787041280.0,
      "grad_norm": 6.255312378521224,
      "learning_rate": 3.2228205424124403e-06,
      "loss": 0.9326,
      "num_input_tokens_seen": 55045520,
      "step": 2594
    },
    {
      "epoch": 0.31203030120844105,
      "flos": 9134351769600.0,
      "grad_norm": 9.147076799560107,
      "learning_rate": 3.222204038687765e-06,
      "loss": 0.9356,
      "num_input_tokens_seen": 55058375,
      "step": 2595
    },
    {
      "epoch": 0.31215054409908016,
      "flos": 19626594324480.0,
      "grad_norm": 4.613512774672929,
      "learning_rate": 3.221587349555709e-06,
      "loss": 1.0784,
      "num_input_tokens_seen": 55078355,
      "step": 2596
    },
    {
      "epoch": 0.3122707869897192,
      "flos": 15274614435840.0,
      "grad_norm": 3.7362908592299964,
      "learning_rate": 3.2209704751098236e-06,
      "loss": 0.882,
      "num_input_tokens_seen": 55097105,
      "step": 2597
    },
    {
      "epoch": 0.31239102988035833,
      "flos": 10732065484800.0,
      "grad_norm": 9.719497308487075,
      "learning_rate": 3.2203534154436875e-06,
      "loss": 1.0088,
      "num_input_tokens_seen": 55111180,
      "step": 2598
    },
    {
      "epoch": 0.31251127277099744,
      "flos": 15668024094720.0,
      "grad_norm": 5.931052471766842,
      "learning_rate": 3.2197361706509084e-06,
      "loss": 0.9557,
      "num_input_tokens_seen": 55131655,
      "step": 2599
    },
    {
      "epoch": 0.3126315156616365,
      "flos": 10738013245440.0,
      "grad_norm": 10.381251278761773,
      "learning_rate": 3.2191187408251228e-06,
      "loss": 1.0133,
      "num_input_tokens_seen": 55148535,
      "step": 2600
    },
    {
      "epoch": 0.3127517585522756,
      "flos": 12858556293120.0,
      "grad_norm": 11.339141329864127,
      "learning_rate": 3.218501126059993e-06,
      "loss": 0.9647,
      "num_input_tokens_seen": 55163650,
      "step": 2601
    },
    {
      "epoch": 0.31287200144291466,
      "flos": 15563875184640.0,
      "grad_norm": 10.503321894190822,
      "learning_rate": 3.2178833264492116e-06,
      "loss": 1.0058,
      "num_input_tokens_seen": 55182075,
      "step": 2602
    },
    {
      "epoch": 0.31299224433355377,
      "flos": 21304688455680.0,
      "grad_norm": 6.3356575298424405,
      "learning_rate": 3.217265342086498e-06,
      "loss": 0.9548,
      "num_input_tokens_seen": 55202285,
      "step": 2603
    },
    {
      "epoch": 0.3131124872241929,
      "flos": 8195386675200.0,
      "grad_norm": 8.539892493114028,
      "learning_rate": 3.216647173065599e-06,
      "loss": 0.9253,
      "num_input_tokens_seen": 55217470,
      "step": 2604
    },
    {
      "epoch": 0.31323273011483194,
      "flos": 35642859909120.0,
      "grad_norm": 5.141384080306632,
      "learning_rate": 3.216028819480292e-06,
      "loss": 0.9327,
      "num_input_tokens_seen": 55238530,
      "step": 2605
    },
    {
      "epoch": 0.31335297300547105,
      "flos": 16061587046400.0,
      "grad_norm": 5.535705840636109,
      "learning_rate": 3.2154102814243793e-06,
      "loss": 0.9519,
      "num_input_tokens_seen": 55257390,
      "step": 2606
    },
    {
      "epoch": 0.31347321589611016,
      "flos": 24764013895680.0,
      "grad_norm": 7.025528270909708,
      "learning_rate": 3.2147915589916937e-06,
      "loss": 0.8686,
      "num_input_tokens_seen": 55278670,
      "step": 2607
    },
    {
      "epoch": 0.3135934587867492,
      "flos": 14016641679360.0,
      "grad_norm": 5.114991254006643,
      "learning_rate": 3.2141726522760938e-06,
      "loss": 1.0176,
      "num_input_tokens_seen": 55296450,
      "step": 2608
    },
    {
      "epoch": 0.3137137016773883,
      "flos": 47125819207680.0,
      "grad_norm": 0.7417651530750395,
      "learning_rate": 3.213553561371469e-06,
      "loss": 0.753,
      "num_input_tokens_seen": 55359905,
      "step": 2609
    },
    {
      "epoch": 0.31383394456802743,
      "flos": 11499009945600.0,
      "grad_norm": 5.2265690688019255,
      "learning_rate": 3.212934286371733e-06,
      "loss": 1.161,
      "num_input_tokens_seen": 55376335,
      "step": 2610
    },
    {
      "epoch": 0.3139541874586665,
      "flos": 27699344977920.0,
      "grad_norm": 7.190732809400002,
      "learning_rate": 3.2123148273708304e-06,
      "loss": 1.016,
      "num_input_tokens_seen": 55396245,
      "step": 2611
    },
    {
      "epoch": 0.3140744303493056,
      "flos": 17818865909760.0,
      "grad_norm": 6.5162654717439334,
      "learning_rate": 3.211695184462733e-06,
      "loss": 0.9608,
      "num_input_tokens_seen": 55417140,
      "step": 2612
    },
    {
      "epoch": 0.3141946732399447,
      "flos": 51929545666560.0,
      "grad_norm": 0.9876411988919449,
      "learning_rate": 3.2110753577414383e-06,
      "loss": 0.8392,
      "num_input_tokens_seen": 55478440,
      "step": 2613
    },
    {
      "epoch": 0.31431491613058377,
      "flos": 13645099683840.0,
      "grad_norm": 15.530466470584368,
      "learning_rate": 3.2104553473009757e-06,
      "loss": 0.9771,
      "num_input_tokens_seen": 55496280,
      "step": 2614
    },
    {
      "epoch": 0.3144351590212229,
      "flos": 25842362695680.0,
      "grad_norm": 5.298230383201682,
      "learning_rate": 3.209835153235399e-06,
      "loss": 0.8651,
      "num_input_tokens_seen": 55517555,
      "step": 2615
    },
    {
      "epoch": 0.314555401911862,
      "flos": 13151312117760.0,
      "grad_norm": 3.0036489230138343,
      "learning_rate": 3.2092147756387916e-06,
      "loss": 0.8729,
      "num_input_tokens_seen": 55537600,
      "step": 2616
    },
    {
      "epoch": 0.31467564480250104,
      "flos": 11578255994880.0,
      "grad_norm": 6.569885748068667,
      "learning_rate": 3.208594214605264e-06,
      "loss": 1.0329,
      "num_input_tokens_seen": 55555865,
      "step": 2617
    },
    {
      "epoch": 0.31479588769314015,
      "flos": 15379591127040.0,
      "grad_norm": 21.993148430383027,
      "learning_rate": 3.2079734702289553e-06,
      "loss": 0.9683,
      "num_input_tokens_seen": 55574480,
      "step": 2618
    },
    {
      "epoch": 0.3149161305837792,
      "flos": 47292413276160.0,
      "grad_norm": 0.8216031049036419,
      "learning_rate": 3.207352542604031e-06,
      "loss": 0.8368,
      "num_input_tokens_seen": 55636535,
      "step": 2619
    },
    {
      "epoch": 0.3150363734744183,
      "flos": 20651151790080.0,
      "grad_norm": 3.3313336732243437,
      "learning_rate": 3.2067314318246864e-06,
      "loss": 0.9711,
      "num_input_tokens_seen": 55656970,
      "step": 2620
    },
    {
      "epoch": 0.31515661636505743,
      "flos": 19680416317440.0,
      "grad_norm": 7.689434769472631,
      "learning_rate": 3.206110137985143e-06,
      "loss": 0.9608,
      "num_input_tokens_seen": 55676895,
      "step": 2621
    },
    {
      "epoch": 0.3152768592556965,
      "flos": 17501759078400.0,
      "grad_norm": 3.7141374464100267,
      "learning_rate": 3.2054886611796505e-06,
      "loss": 1.1128,
      "num_input_tokens_seen": 55695610,
      "step": 2622
    },
    {
      "epoch": 0.3153971021463356,
      "flos": 49033289809920.0,
      "grad_norm": 0.9680334332816454,
      "learning_rate": 3.204867001502487e-06,
      "loss": 0.8846,
      "num_input_tokens_seen": 55753985,
      "step": 2623
    },
    {
      "epoch": 0.3155173450369747,
      "flos": 18212275568640.0,
      "grad_norm": 4.312036904058555,
      "learning_rate": 3.2042451590479567e-06,
      "loss": 0.9922,
      "num_input_tokens_seen": 55774220,
      "step": 2624
    },
    {
      "epoch": 0.31563758792761376,
      "flos": 17288218644480.0,
      "grad_norm": 3.049121204783431,
      "learning_rate": 3.203623133910394e-06,
      "loss": 1.0457,
      "num_input_tokens_seen": 55792245,
      "step": 2625
    },
    {
      "epoch": 0.31575783081825287,
      "flos": 22747006586880.0,
      "grad_norm": 10.00409385806123,
      "learning_rate": 3.203000926184158e-06,
      "loss": 0.9586,
      "num_input_tokens_seen": 55810890,
      "step": 2626
    },
    {
      "epoch": 0.315878073708892,
      "flos": 21961781514240.0,
      "grad_norm": 35.47336148435894,
      "learning_rate": 3.202378535963639e-06,
      "loss": 0.9602,
      "num_input_tokens_seen": 55831525,
      "step": 2627
    },
    {
      "epoch": 0.31599831659953104,
      "flos": 15773430005760.0,
      "grad_norm": 5.456974802018804,
      "learning_rate": 3.2017559633432516e-06,
      "loss": 1.0284,
      "num_input_tokens_seen": 55850875,
      "step": 2628
    },
    {
      "epoch": 0.31611855949017015,
      "flos": 18212060958720.0,
      "grad_norm": 71.54051586448234,
      "learning_rate": 3.2011332084174398e-06,
      "loss": 0.85,
      "num_input_tokens_seen": 55871465,
      "step": 2629
    },
    {
      "epoch": 0.31623880238080926,
      "flos": 14618992988160.0,
      "grad_norm": 4.809380449139518,
      "learning_rate": 3.2005102712806756e-06,
      "loss": 1.0779,
      "num_input_tokens_seen": 55890015,
      "step": 2630
    },
    {
      "epoch": 0.3163590452714483,
      "flos": 9006526341120.0,
      "grad_norm": 46.8599168441497,
      "learning_rate": 3.1998871520274575e-06,
      "loss": 0.9177,
      "num_input_tokens_seen": 55905070,
      "step": 2631
    },
    {
      "epoch": 0.3164792881620874,
      "flos": 16378111365120.0,
      "grad_norm": 4.147722028109036,
      "learning_rate": 3.199263850752312e-06,
      "loss": 1.0347,
      "num_input_tokens_seen": 55925625,
      "step": 2632
    },
    {
      "epoch": 0.31659953105272653,
      "flos": 12967702548480.0,
      "grad_norm": 3.723143602715048,
      "learning_rate": 3.198640367549795e-06,
      "loss": 1.0464,
      "num_input_tokens_seen": 55944240,
      "step": 2633
    },
    {
      "epoch": 0.3167197739433656,
      "flos": 18291153715200.0,
      "grad_norm": 5.336453280228409,
      "learning_rate": 3.198016702514487e-06,
      "loss": 1.0527,
      "num_input_tokens_seen": 55964240,
      "step": 2634
    },
    {
      "epoch": 0.3168400168340047,
      "flos": 16740026572800.0,
      "grad_norm": 11.085296074643203,
      "learning_rate": 3.1973928557409972e-06,
      "loss": 1.0435,
      "num_input_tokens_seen": 55982000,
      "step": 2635
    },
    {
      "epoch": 0.31696025972464376,
      "flos": 20204962529280.0,
      "grad_norm": 5.289043206724874,
      "learning_rate": 3.1967688273239636e-06,
      "loss": 0.8942,
      "num_input_tokens_seen": 56001525,
      "step": 2636
    },
    {
      "epoch": 0.31708050261528287,
      "flos": 11604262563840.0,
      "grad_norm": 17.645077355534955,
      "learning_rate": 3.1961446173580503e-06,
      "loss": 1.0123,
      "num_input_tokens_seen": 56018185,
      "step": 2637
    },
    {
      "epoch": 0.317200745505922,
      "flos": 18656655974400.0,
      "grad_norm": 5.057733623183033,
      "learning_rate": 3.1955202259379502e-06,
      "loss": 0.965,
      "num_input_tokens_seen": 56039635,
      "step": 2638
    },
    {
      "epoch": 0.31732098839656103,
      "flos": 22348814192640.0,
      "grad_norm": 8.135382020089343,
      "learning_rate": 3.194895653158381e-06,
      "loss": 1.0133,
      "num_input_tokens_seen": 56058295,
      "step": 2639
    },
    {
      "epoch": 0.31744123128720014,
      "flos": 42219239608320.0,
      "grad_norm": 0.8162342602203889,
      "learning_rate": 3.194270899114093e-06,
      "loss": 0.7888,
      "num_input_tokens_seen": 56123810,
      "step": 2640
    },
    {
      "epoch": 0.31756147417783925,
      "flos": 12335788277760.0,
      "grad_norm": 11.26528062439577,
      "learning_rate": 3.193645963899858e-06,
      "loss": 1.0186,
      "num_input_tokens_seen": 56141145,
      "step": 2641
    },
    {
      "epoch": 0.3176817170684783,
      "flos": 18131404615680.0,
      "grad_norm": 8.56368364381062,
      "learning_rate": 3.193020847610479e-06,
      "loss": 1.028,
      "num_input_tokens_seen": 56161185,
      "step": 2642
    },
    {
      "epoch": 0.3178019599591174,
      "flos": 17765779722240.0,
      "grad_norm": 12.840848497537799,
      "learning_rate": 3.192395550340787e-06,
      "loss": 0.9058,
      "num_input_tokens_seen": 56178855,
      "step": 2643
    },
    {
      "epoch": 0.31792220284975653,
      "flos": 8746154065920.0,
      "grad_norm": 5.86701032470086,
      "learning_rate": 3.191770072185638e-06,
      "loss": 0.9594,
      "num_input_tokens_seen": 56195570,
      "step": 2644
    },
    {
      "epoch": 0.3180424457403956,
      "flos": 10947690700800.0,
      "grad_norm": 4.807006276408495,
      "learning_rate": 3.191144413239916e-06,
      "loss": 0.9185,
      "num_input_tokens_seen": 56211860,
      "step": 2645
    },
    {
      "epoch": 0.3181626886310347,
      "flos": 18629668331520.0,
      "grad_norm": 4.354628956664359,
      "learning_rate": 3.190518573598534e-06,
      "loss": 1.0719,
      "num_input_tokens_seen": 56228185,
      "step": 2646
    },
    {
      "epoch": 0.3182829315216738,
      "flos": 18132630958080.0,
      "grad_norm": 4.1403324815600655,
      "learning_rate": 3.1898925533564308e-06,
      "loss": 0.9786,
      "num_input_tokens_seen": 56249375,
      "step": 2647
    },
    {
      "epoch": 0.31840317441231286,
      "flos": 12836474019840.0,
      "grad_norm": 5.47688795311726,
      "learning_rate": 3.1892663526085733e-06,
      "loss": 0.835,
      "num_input_tokens_seen": 56267470,
      "step": 2648
    },
    {
      "epoch": 0.31852341730295197,
      "flos": 46352160522240.0,
      "grad_norm": 0.8116123045224073,
      "learning_rate": 3.188639971449956e-06,
      "loss": 0.7968,
      "num_input_tokens_seen": 56333240,
      "step": 2649
    },
    {
      "epoch": 0.318643660193591,
      "flos": 14672355102720.0,
      "grad_norm": 5.8021857516861965,
      "learning_rate": 3.1880134099756e-06,
      "loss": 0.9152,
      "num_input_tokens_seen": 56352595,
      "step": 2650
    },
    {
      "epoch": 0.31876390308423014,
      "flos": 19182183260160.0,
      "grad_norm": 6.23535056834859,
      "learning_rate": 3.1873866682805535e-06,
      "loss": 0.8986,
      "num_input_tokens_seen": 56373010,
      "step": 2651
    },
    {
      "epoch": 0.31888414597486925,
      "flos": 12785043394560.0,
      "grad_norm": 7.692121403712576,
      "learning_rate": 3.186759746459894e-06,
      "loss": 1.0758,
      "num_input_tokens_seen": 56391840,
      "step": 2652
    },
    {
      "epoch": 0.3190043888655083,
      "flos": 17892010905600.0,
      "grad_norm": 5.756466090881519,
      "learning_rate": 3.1861326446087246e-06,
      "loss": 0.9907,
      "num_input_tokens_seen": 56410300,
      "step": 2653
    },
    {
      "epoch": 0.3191246317561474,
      "flos": 15668177387520.0,
      "grad_norm": 4.683110941610359,
      "learning_rate": 3.1855053628221763e-06,
      "loss": 0.9074,
      "num_input_tokens_seen": 56429275,
      "step": 2654
    },
    {
      "epoch": 0.3192448746467865,
      "flos": 10527998545920.0,
      "grad_norm": 11.878143858098026,
      "learning_rate": 3.184877901195407e-06,
      "loss": 1.0844,
      "num_input_tokens_seen": 56445690,
      "step": 2655
    },
    {
      "epoch": 0.3193651175374256,
      "flos": 48145348669440.0,
      "grad_norm": 0.9037487766579438,
      "learning_rate": 3.184250259823602e-06,
      "loss": 0.8661,
      "num_input_tokens_seen": 56507940,
      "step": 2656
    },
    {
      "epoch": 0.3194853604280647,
      "flos": 8609315020800.0,
      "grad_norm": 6.492948081293809,
      "learning_rate": 3.183622438801974e-06,
      "loss": 0.9953,
      "num_input_tokens_seen": 56522950,
      "step": 2657
    },
    {
      "epoch": 0.3196056033187038,
      "flos": 10555599360000.0,
      "grad_norm": 3.869228864623536,
      "learning_rate": 3.1829944382257637e-06,
      "loss": 0.9576,
      "num_input_tokens_seen": 56540800,
      "step": 2658
    },
    {
      "epoch": 0.31972584620934286,
      "flos": 16900726087680.0,
      "grad_norm": 3.9411666993915113,
      "learning_rate": 3.1823662581902373e-06,
      "loss": 1.0067,
      "num_input_tokens_seen": 56558205,
      "step": 2659
    },
    {
      "epoch": 0.31984608909998197,
      "flos": 15091587379200.0,
      "grad_norm": 3.9998617148337727,
      "learning_rate": 3.1817378987906896e-06,
      "loss": 0.9307,
      "num_input_tokens_seen": 56577430,
      "step": 2660
    },
    {
      "epoch": 0.3199663319906211,
      "flos": 12967671889920.0,
      "grad_norm": 4.99598418138747,
      "learning_rate": 3.181109360122442e-06,
      "loss": 1.002,
      "num_input_tokens_seen": 56594740,
      "step": 2661
    },
    {
      "epoch": 0.32008657488126013,
      "flos": 13281958133760.0,
      "grad_norm": 6.820266636085769,
      "learning_rate": 3.1804806422808445e-06,
      "loss": 0.9732,
      "num_input_tokens_seen": 56611595,
      "step": 2662
    },
    {
      "epoch": 0.32020681777189924,
      "flos": 14620066037760.0,
      "grad_norm": 3.13708919556181,
      "learning_rate": 3.1798517453612714e-06,
      "loss": 0.9182,
      "num_input_tokens_seen": 56631120,
      "step": 2663
    },
    {
      "epoch": 0.32032706066253835,
      "flos": 25160366776320.0,
      "grad_norm": 6.381256922494868,
      "learning_rate": 3.1792226694591265e-06,
      "loss": 0.9502,
      "num_input_tokens_seen": 56652225,
      "step": 2664
    },
    {
      "epoch": 0.3204473035531774,
      "flos": 10817351270400.0,
      "grad_norm": 4.803036190445426,
      "learning_rate": 3.178593414669841e-06,
      "loss": 0.9944,
      "num_input_tokens_seen": 56670530,
      "step": 2665
    },
    {
      "epoch": 0.3205675464438165,
      "flos": 17399327047680.0,
      "grad_norm": 5.526105999609571,
      "learning_rate": 3.1779639810888707e-06,
      "loss": 0.8957,
      "num_input_tokens_seen": 56689845,
      "step": 2666
    },
    {
      "epoch": 0.3206877893344556,
      "flos": 15957468794880.0,
      "grad_norm": 10.450355881467214,
      "learning_rate": 3.1773343688117013e-06,
      "loss": 0.9514,
      "num_input_tokens_seen": 56710475,
      "step": 2667
    },
    {
      "epoch": 0.3208080322250947,
      "flos": 14488377630720.0,
      "grad_norm": 11.968112540139213,
      "learning_rate": 3.1767045779338445e-06,
      "loss": 1.0271,
      "num_input_tokens_seen": 56727855,
      "step": 2668
    },
    {
      "epoch": 0.3209282751157338,
      "flos": 15458499932160.0,
      "grad_norm": 13.228512597122421,
      "learning_rate": 3.176074608550839e-06,
      "loss": 1.1035,
      "num_input_tokens_seen": 56743395,
      "step": 2669
    },
    {
      "epoch": 0.32104851800637285,
      "flos": 15669158461440.0,
      "grad_norm": 9.229487176041518,
      "learning_rate": 3.17544446075825e-06,
      "loss": 1.0252,
      "num_input_tokens_seen": 56762280,
      "step": 2670
    },
    {
      "epoch": 0.32116876089701196,
      "flos": 26418400849920.0,
      "grad_norm": 4.2248305969586575,
      "learning_rate": 3.174814134651671e-06,
      "loss": 0.9112,
      "num_input_tokens_seen": 56784550,
      "step": 2671
    },
    {
      "epoch": 0.3212890037876511,
      "flos": 15614294077440.0,
      "grad_norm": 4.542040794898207,
      "learning_rate": 3.1741836303267215e-06,
      "loss": 1.0085,
      "num_input_tokens_seen": 56803805,
      "step": 2672
    },
    {
      "epoch": 0.32140924667829013,
      "flos": 7251362918400.0,
      "grad_norm": 3.25315630308659,
      "learning_rate": 3.1735529478790496e-06,
      "loss": 0.938,
      "num_input_tokens_seen": 56821515,
      "step": 2673
    },
    {
      "epoch": 0.32152948956892924,
      "flos": 36325898219520.0,
      "grad_norm": 4.969756978039909,
      "learning_rate": 3.172922087404328e-06,
      "loss": 0.9949,
      "num_input_tokens_seen": 56843495,
      "step": 2674
    },
    {
      "epoch": 0.32164973245956835,
      "flos": 45720246251520.0,
      "grad_norm": 0.8816228632390609,
      "learning_rate": 3.1722910489982586e-06,
      "loss": 0.7788,
      "num_input_tokens_seen": 56903575,
      "step": 2675
    },
    {
      "epoch": 0.3217699753502074,
      "flos": 16717576396800.0,
      "grad_norm": 3.1380379872888677,
      "learning_rate": 3.1716598327565694e-06,
      "loss": 1.001,
      "num_input_tokens_seen": 56922935,
      "step": 2676
    },
    {
      "epoch": 0.3218902182408465,
      "flos": 13518408622080.0,
      "grad_norm": 2.7735610289666583,
      "learning_rate": 3.171028438775015e-06,
      "loss": 1.036,
      "num_input_tokens_seen": 56941850,
      "step": 2677
    },
    {
      "epoch": 0.3220104611314856,
      "flos": 14462003159040.0,
      "grad_norm": 3.729144659279329,
      "learning_rate": 3.170396867149377e-06,
      "loss": 1.0369,
      "num_input_tokens_seen": 56959575,
      "step": 2678
    },
    {
      "epoch": 0.3221307040221247,
      "flos": 14275756953600.0,
      "grad_norm": 4.18451778062855,
      "learning_rate": 3.1697651179754653e-06,
      "loss": 1.0515,
      "num_input_tokens_seen": 56977955,
      "step": 2679
    },
    {
      "epoch": 0.3222509469127638,
      "flos": 17058053160960.0,
      "grad_norm": 6.936208243863941,
      "learning_rate": 3.1691331913491153e-06,
      "loss": 0.9348,
      "num_input_tokens_seen": 57000245,
      "step": 2680
    },
    {
      "epoch": 0.32237118980340285,
      "flos": 12521697239040.0,
      "grad_norm": 17.190764072162867,
      "learning_rate": 3.1685010873661898e-06,
      "loss": 1.0321,
      "num_input_tokens_seen": 57019140,
      "step": 2681
    },
    {
      "epoch": 0.32249143269404196,
      "flos": 16454414192640.0,
      "grad_norm": 3.043275712363637,
      "learning_rate": 3.167868806122578e-06,
      "loss": 0.9838,
      "num_input_tokens_seen": 57037910,
      "step": 2682
    },
    {
      "epoch": 0.32261167558468107,
      "flos": 17369886720000.0,
      "grad_norm": 3.071525508833683,
      "learning_rate": 3.1672363477141968e-06,
      "loss": 0.8661,
      "num_input_tokens_seen": 57056925,
      "step": 2683
    },
    {
      "epoch": 0.3227319184753201,
      "flos": 21642038046720.0,
      "grad_norm": 4.5473550759365065,
      "learning_rate": 3.1666037122369903e-06,
      "loss": 1.0362,
      "num_input_tokens_seen": 57077305,
      "step": 2684
    },
    {
      "epoch": 0.32285216136595923,
      "flos": 11997396295680.0,
      "grad_norm": 4.04798903283605,
      "learning_rate": 3.165970899786928e-06,
      "loss": 1.0493,
      "num_input_tokens_seen": 57094940,
      "step": 2685
    },
    {
      "epoch": 0.32297240425659834,
      "flos": 15616256225280.0,
      "grad_norm": 2.183430401146414,
      "learning_rate": 3.1653379104600067e-06,
      "loss": 0.9223,
      "num_input_tokens_seen": 57114805,
      "step": 2686
    },
    {
      "epoch": 0.3230926471472374,
      "flos": 16167176908800.0,
      "grad_norm": 2.5249488485461757,
      "learning_rate": 3.164704744352251e-06,
      "loss": 0.8857,
      "num_input_tokens_seen": 57135330,
      "step": 2687
    },
    {
      "epoch": 0.3232128900378765,
      "flos": 11994667683840.0,
      "grad_norm": 7.969929925665566,
      "learning_rate": 3.164071401559713e-06,
      "loss": 1.0034,
      "num_input_tokens_seen": 57152450,
      "step": 2688
    },
    {
      "epoch": 0.3233331329285156,
      "flos": 17083814461440.0,
      "grad_norm": 3.7673616955234683,
      "learning_rate": 3.1634378821784674e-06,
      "loss": 0.9043,
      "num_input_tokens_seen": 57172385,
      "step": 2689
    },
    {
      "epoch": 0.3234533758191547,
      "flos": 12836596654080.0,
      "grad_norm": 2.6454020071011524,
      "learning_rate": 3.1628041863046208e-06,
      "loss": 0.926,
      "num_input_tokens_seen": 57189520,
      "step": 2690
    },
    {
      "epoch": 0.3235736187097938,
      "flos": 11997365637120.0,
      "grad_norm": 6.695202251225459,
      "learning_rate": 3.162170314034304e-06,
      "loss": 1.1008,
      "num_input_tokens_seen": 57206655,
      "step": 2691
    },
    {
      "epoch": 0.3236938616004329,
      "flos": 15720865013760.0,
      "grad_norm": 11.408816197841597,
      "learning_rate": 3.1615362654636738e-06,
      "loss": 0.9982,
      "num_input_tokens_seen": 57227115,
      "step": 2692
    },
    {
      "epoch": 0.32381410449107195,
      "flos": 12154324807680.0,
      "grad_norm": 2.7716810225227566,
      "learning_rate": 3.1609020406889163e-06,
      "loss": 1.0706,
      "num_input_tokens_seen": 57244270,
      "step": 2693
    },
    {
      "epoch": 0.32393434738171106,
      "flos": 11732455895040.0,
      "grad_norm": 24.460651796878288,
      "learning_rate": 3.1602676398062416e-06,
      "loss": 1.0455,
      "num_input_tokens_seen": 57262900,
      "step": 2694
    },
    {
      "epoch": 0.3240545902723502,
      "flos": 18133060177920.0,
      "grad_norm": 4.492007256913883,
      "learning_rate": 3.1596330629118886e-06,
      "loss": 0.8118,
      "num_input_tokens_seen": 57282590,
      "step": 2695
    },
    {
      "epoch": 0.32417483316298923,
      "flos": 25580242882560.0,
      "grad_norm": 4.422049475149384,
      "learning_rate": 3.1589983101021223e-06,
      "loss": 0.9358,
      "num_input_tokens_seen": 57300940,
      "step": 2696
    },
    {
      "epoch": 0.32429507605362834,
      "flos": 21436714106880.0,
      "grad_norm": 4.545348864974095,
      "learning_rate": 3.1583633814732337e-06,
      "loss": 1.0194,
      "num_input_tokens_seen": 57320180,
      "step": 2697
    },
    {
      "epoch": 0.3244153189442674,
      "flos": 12915229532160.0,
      "grad_norm": 7.006025173585244,
      "learning_rate": 3.157728277121541e-06,
      "loss": 0.9046,
      "num_input_tokens_seen": 57338075,
      "step": 2698
    },
    {
      "epoch": 0.3245355618349065,
      "flos": 12546293514240.0,
      "grad_norm": 3.6933561473330028,
      "learning_rate": 3.1570929971433897e-06,
      "loss": 0.9814,
      "num_input_tokens_seen": 57353580,
      "step": 2699
    },
    {
      "epoch": 0.3246558047255456,
      "flos": 16664980746240.0,
      "grad_norm": 3.4835455895204785,
      "learning_rate": 3.1564575416351504e-06,
      "loss": 1.0184,
      "num_input_tokens_seen": 57372000,
      "step": 2700
    },
    {
      "epoch": 0.32477604761618467,
      "flos": 15457426882560.0,
      "grad_norm": 37.56990879513816,
      "learning_rate": 3.155821910693221e-06,
      "loss": 0.9354,
      "num_input_tokens_seen": 57391135,
      "step": 2701
    },
    {
      "epoch": 0.3248962905068238,
      "flos": 14068716134400.0,
      "grad_norm": 2.522387873839011,
      "learning_rate": 3.1551861044140275e-06,
      "loss": 1.0487,
      "num_input_tokens_seen": 57410490,
      "step": 2702
    },
    {
      "epoch": 0.3250165333974629,
      "flos": 17029195345920.0,
      "grad_norm": 3.1197827847990656,
      "learning_rate": 3.15455012289402e-06,
      "loss": 0.9664,
      "num_input_tokens_seen": 57429855,
      "step": 2703
    },
    {
      "epoch": 0.32513677628810195,
      "flos": 17059371479040.0,
      "grad_norm": 2.5080356201550553,
      "learning_rate": 3.153913966229677e-06,
      "loss": 1.0307,
      "num_input_tokens_seen": 57448695,
      "step": 2704
    },
    {
      "epoch": 0.32525701917874106,
      "flos": 50602750771200.0,
      "grad_norm": 0.6674811161599022,
      "learning_rate": 3.1532776345175027e-06,
      "loss": 0.7081,
      "num_input_tokens_seen": 57513560,
      "step": 2705
    },
    {
      "epoch": 0.32537726206938017,
      "flos": 13964046028800.0,
      "grad_norm": 2.645174430042439,
      "learning_rate": 3.1526411278540285e-06,
      "loss": 0.9708,
      "num_input_tokens_seen": 57531710,
      "step": 2706
    },
    {
      "epoch": 0.3254975049600192,
      "flos": 20487601029120.0,
      "grad_norm": 3.072437434354832,
      "learning_rate": 3.1520044463358116e-06,
      "loss": 1.0011,
      "num_input_tokens_seen": 57548160,
      "step": 2707
    },
    {
      "epoch": 0.32561774785065833,
      "flos": 13384972677120.0,
      "grad_norm": 3.0426760178917367,
      "learning_rate": 3.151367590059436e-06,
      "loss": 0.9928,
      "num_input_tokens_seen": 57566305,
      "step": 2708
    },
    {
      "epoch": 0.32573799074129745,
      "flos": 16428714209280.0,
      "grad_norm": 19.214609881810148,
      "learning_rate": 3.1507305591215117e-06,
      "loss": 1.0562,
      "num_input_tokens_seen": 57583935,
      "step": 2709
    },
    {
      "epoch": 0.3258582336319365,
      "flos": 51020664729600.0,
      "grad_norm": 0.7301426531919378,
      "learning_rate": 3.150093353618677e-06,
      "loss": 0.7764,
      "num_input_tokens_seen": 57648385,
      "step": 2710
    },
    {
      "epoch": 0.3259784765225756,
      "flos": 15957284843520.0,
      "grad_norm": 5.305001026264922,
      "learning_rate": 3.149455973647596e-06,
      "loss": 1.0683,
      "num_input_tokens_seen": 57666165,
      "step": 2711
    },
    {
      "epoch": 0.32609871941321467,
      "flos": 14540482744320.0,
      "grad_norm": 8.556463410841785,
      "learning_rate": 3.1488184193049563e-06,
      "loss": 0.9581,
      "num_input_tokens_seen": 57685420,
      "step": 2712
    },
    {
      "epoch": 0.3262189623038538,
      "flos": 15928733614080.0,
      "grad_norm": 2.7273708773782483,
      "learning_rate": 3.1481806906874767e-06,
      "loss": 0.9224,
      "num_input_tokens_seen": 57706450,
      "step": 2713
    },
    {
      "epoch": 0.3263392051944929,
      "flos": 14855872696320.0,
      "grad_norm": 19.23984963467344,
      "learning_rate": 3.147542787891899e-06,
      "loss": 1.0762,
      "num_input_tokens_seen": 57725515,
      "step": 2714
    },
    {
      "epoch": 0.32645944808513194,
      "flos": 17085040803840.0,
      "grad_norm": 3.346928876911562,
      "learning_rate": 3.1469047110149926e-06,
      "loss": 0.9358,
      "num_input_tokens_seen": 57743975,
      "step": 2715
    },
    {
      "epoch": 0.32657969097577105,
      "flos": 14934076354560.0,
      "grad_norm": 2.7786469869185213,
      "learning_rate": 3.146266460153554e-06,
      "loss": 1.0502,
      "num_input_tokens_seen": 57763405,
      "step": 2716
    },
    {
      "epoch": 0.32669993386641016,
      "flos": 16139852021760.0,
      "grad_norm": 2.762283893262298,
      "learning_rate": 3.145628035404404e-06,
      "loss": 0.9964,
      "num_input_tokens_seen": 57782325,
      "step": 2717
    },
    {
      "epoch": 0.3268201767570492,
      "flos": 51643504066560.0,
      "grad_norm": 0.8699700673204797,
      "learning_rate": 3.1449894368643922e-06,
      "loss": 0.7946,
      "num_input_tokens_seen": 57844360,
      "step": 2718
    },
    {
      "epoch": 0.32694041964768833,
      "flos": 17451462819840.0,
      "grad_norm": 2.182823959899398,
      "learning_rate": 3.1443506646303934e-06,
      "loss": 0.9158,
      "num_input_tokens_seen": 57865380,
      "step": 2719
    },
    {
      "epoch": 0.32706066253832744,
      "flos": 23666587361280.0,
      "grad_norm": 6.375262685113392,
      "learning_rate": 3.1437117187993086e-06,
      "loss": 0.8621,
      "num_input_tokens_seen": 57887420,
      "step": 2720
    },
    {
      "epoch": 0.3271809054289665,
      "flos": 17111231324160.0,
      "grad_norm": 13.380397384969084,
      "learning_rate": 3.143072599468065e-06,
      "loss": 0.9949,
      "num_input_tokens_seen": 57906965,
      "step": 2721
    },
    {
      "epoch": 0.3273011483196056,
      "flos": 27310748712960.0,
      "grad_norm": 2.088213875051539,
      "learning_rate": 3.1424333067336174e-06,
      "loss": 0.9565,
      "num_input_tokens_seen": 57929450,
      "step": 2722
    },
    {
      "epoch": 0.3274213912102447,
      "flos": 20698658119680.0,
      "grad_norm": 2.3209984789658034,
      "learning_rate": 3.141793840692945e-06,
      "loss": 0.9663,
      "num_input_tokens_seen": 57949920,
      "step": 2723
    },
    {
      "epoch": 0.32754163410088377,
      "flos": 20756128481280.0,
      "grad_norm": 5.940165032043315,
      "learning_rate": 3.1411542014430553e-06,
      "loss": 0.8061,
      "num_input_tokens_seen": 57970720,
      "step": 2724
    },
    {
      "epoch": 0.3276618769915229,
      "flos": 14645643386880.0,
      "grad_norm": 3.7103416205386095,
      "learning_rate": 3.1405143890809804e-06,
      "loss": 1.0214,
      "num_input_tokens_seen": 57989735,
      "step": 2725
    },
    {
      "epoch": 0.327782119882162,
      "flos": 13226756505600.0,
      "grad_norm": 7.46444258717334,
      "learning_rate": 3.1398744037037796e-06,
      "loss": 0.903,
      "num_input_tokens_seen": 58008790,
      "step": 2726
    },
    {
      "epoch": 0.32790236277280105,
      "flos": 15482329743360.0,
      "grad_norm": 3.813873793931839,
      "learning_rate": 3.139234245408538e-06,
      "loss": 1.0348,
      "num_input_tokens_seen": 58027390,
      "step": 2727
    },
    {
      "epoch": 0.32802260566344016,
      "flos": 16586378526720.0,
      "grad_norm": 2.4973365901301885,
      "learning_rate": 3.1385939142923666e-06,
      "loss": 0.9605,
      "num_input_tokens_seen": 58049500,
      "step": 2728
    },
    {
      "epoch": 0.3281428485540792,
      "flos": 17215625502720.0,
      "grad_norm": 3.2350525077320156,
      "learning_rate": 3.137953410452405e-06,
      "loss": 0.9739,
      "num_input_tokens_seen": 58069490,
      "step": 2729
    },
    {
      "epoch": 0.3282630914447183,
      "flos": 24346069278720.0,
      "grad_norm": 13.075854907706143,
      "learning_rate": 3.1373127339858146e-06,
      "loss": 0.9413,
      "num_input_tokens_seen": 58091810,
      "step": 2730
    },
    {
      "epoch": 0.32838333433535744,
      "flos": 19654716334080.0,
      "grad_norm": 3.9904941618044636,
      "learning_rate": 3.136671884989787e-06,
      "loss": 0.9283,
      "num_input_tokens_seen": 58111440,
      "step": 2731
    },
    {
      "epoch": 0.3285035772259965,
      "flos": 8693926318080.0,
      "grad_norm": 3.7520533979719572,
      "learning_rate": 3.1360308635615383e-06,
      "loss": 1.0564,
      "num_input_tokens_seen": 58129700,
      "step": 2732
    },
    {
      "epoch": 0.3286238201166356,
      "flos": 17293982453760.0,
      "grad_norm": 3.718148062578841,
      "learning_rate": 3.135389669798311e-06,
      "loss": 0.9873,
      "num_input_tokens_seen": 58147480,
      "step": 2733
    },
    {
      "epoch": 0.3287440630072747,
      "flos": 15192486481920.0,
      "grad_norm": 3.139329026083247,
      "learning_rate": 3.134748303797373e-06,
      "loss": 0.9993,
      "num_input_tokens_seen": 58164570,
      "step": 2734
    },
    {
      "epoch": 0.32886430589791377,
      "flos": 16874627543040.0,
      "grad_norm": 4.239850659889339,
      "learning_rate": 3.1341067656560203e-06,
      "loss": 1.0063,
      "num_input_tokens_seen": 58182135,
      "step": 2735
    },
    {
      "epoch": 0.3289845487885529,
      "flos": 15930358517760.0,
      "grad_norm": 3.156278704522165,
      "learning_rate": 3.133465055471572e-06,
      "loss": 1.0472,
      "num_input_tokens_seen": 58201640,
      "step": 2736
    },
    {
      "epoch": 0.329104791679192,
      "flos": 13964015370240.0,
      "grad_norm": 3.6241956673487556,
      "learning_rate": 3.1328231733413767e-06,
      "loss": 0.8455,
      "num_input_tokens_seen": 58218000,
      "step": 2737
    },
    {
      "epoch": 0.32922503456983104,
      "flos": 11315369717760.0,
      "grad_norm": 2.6751417746363084,
      "learning_rate": 3.1321811193628067e-06,
      "loss": 1.1019,
      "num_input_tokens_seen": 58235865,
      "step": 2738
    },
    {
      "epoch": 0.32934527746047015,
      "flos": 19103673016320.0,
      "grad_norm": 4.01826787877631,
      "learning_rate": 3.131538893633261e-06,
      "loss": 0.8996,
      "num_input_tokens_seen": 58255145,
      "step": 2739
    },
    {
      "epoch": 0.32946552035110926,
      "flos": 16638299688960.0,
      "grad_norm": 5.7419626204902645,
      "learning_rate": 3.130896496250165e-06,
      "loss": 0.9685,
      "num_input_tokens_seen": 58274690,
      "step": 2740
    },
    {
      "epoch": 0.3295857632417483,
      "flos": 10003881553920.0,
      "grad_norm": 3.944812221723696,
      "learning_rate": 3.1302539273109693e-06,
      "loss": 1.0549,
      "num_input_tokens_seen": 58291235,
      "step": 2741
    },
    {
      "epoch": 0.32970600613238743,
      "flos": 15770394808320.0,
      "grad_norm": 3.5915953438057944,
      "learning_rate": 3.1296111869131513e-06,
      "loss": 1.0046,
      "num_input_tokens_seen": 58308380,
      "step": 2742
    },
    {
      "epoch": 0.32982624902302654,
      "flos": 15668085411840.0,
      "grad_norm": 6.207828919403154,
      "learning_rate": 3.1289682751542153e-06,
      "loss": 1.042,
      "num_input_tokens_seen": 58327660,
      "step": 2743
    },
    {
      "epoch": 0.3299464919136656,
      "flos": 13230343557120.0,
      "grad_norm": 5.779621286623967,
      "learning_rate": 3.1283251921316883e-06,
      "loss": 0.9114,
      "num_input_tokens_seen": 58345125,
      "step": 2744
    },
    {
      "epoch": 0.3300667348043047,
      "flos": 9454156554240.0,
      "grad_norm": 11.087217830950966,
      "learning_rate": 3.1276819379431277e-06,
      "loss": 1.0173,
      "num_input_tokens_seen": 58362935,
      "step": 2745
    },
    {
      "epoch": 0.33018697769494376,
      "flos": 11132434636800.0,
      "grad_norm": 9.923224025539037,
      "learning_rate": 3.1270385126861134e-06,
      "loss": 0.9456,
      "num_input_tokens_seen": 58380640,
      "step": 2746
    },
    {
      "epoch": 0.3303072205855829,
      "flos": 12940469637120.0,
      "grad_norm": 4.130039258128383,
      "learning_rate": 3.1263949164582533e-06,
      "loss": 1.0127,
      "num_input_tokens_seen": 58400010,
      "step": 2747
    },
    {
      "epoch": 0.330427463476222,
      "flos": 12574752768000.0,
      "grad_norm": 2.616385903653813,
      "learning_rate": 3.1257511493571797e-06,
      "loss": 0.964,
      "num_input_tokens_seen": 58418235,
      "step": 2748
    },
    {
      "epoch": 0.33054770636686104,
      "flos": 19339786260480.0,
      "grad_norm": 6.532622708709397,
      "learning_rate": 3.125107211480552e-06,
      "loss": 0.9829,
      "num_input_tokens_seen": 58437890,
      "step": 2749
    },
    {
      "epoch": 0.33066794925750015,
      "flos": 14275910246400.0,
      "grad_norm": 8.944805014261858,
      "learning_rate": 3.124463102926054e-06,
      "loss": 0.9862,
      "num_input_tokens_seen": 58456945,
      "step": 2750
    },
    {
      "epoch": 0.33078819214813926,
      "flos": 50591499079680.0,
      "grad_norm": 0.7938982397109684,
      "learning_rate": 3.1238188237913984e-06,
      "loss": 0.8202,
      "num_input_tokens_seen": 58521205,
      "step": 2751
    },
    {
      "epoch": 0.3309084350387783,
      "flos": 15012801208320.0,
      "grad_norm": 6.985976204181533,
      "learning_rate": 3.1231743741743202e-06,
      "loss": 0.9519,
      "num_input_tokens_seen": 58540430,
      "step": 2752
    },
    {
      "epoch": 0.3310286779294174,
      "flos": 10030439976960.0,
      "grad_norm": 4.0978093877783595,
      "learning_rate": 3.122529754172582e-06,
      "loss": 1.026,
      "num_input_tokens_seen": 58557035,
      "step": 2753
    },
    {
      "epoch": 0.33114892082005654,
      "flos": 14751478517760.0,
      "grad_norm": 20.566518651758372,
      "learning_rate": 3.1218849638839736e-06,
      "loss": 0.9242,
      "num_input_tokens_seen": 58576015,
      "step": 2754
    },
    {
      "epoch": 0.3312691637106956,
      "flos": 12101115985920.0,
      "grad_norm": 4.1343765039650275,
      "learning_rate": 3.121240003406307e-06,
      "loss": 0.9785,
      "num_input_tokens_seen": 58594585,
      "step": 2755
    },
    {
      "epoch": 0.3313894066013347,
      "flos": 20987520307200.0,
      "grad_norm": 4.822080711382272,
      "learning_rate": 3.120594872837425e-06,
      "loss": 0.9116,
      "num_input_tokens_seen": 58612975,
      "step": 2756
    },
    {
      "epoch": 0.3315096494919738,
      "flos": 43963856486400.0,
      "grad_norm": 0.8508109529773028,
      "learning_rate": 3.1199495722751906e-06,
      "loss": 0.8431,
      "num_input_tokens_seen": 58672225,
      "step": 2757
    },
    {
      "epoch": 0.33162989238261287,
      "flos": 15380173639680.0,
      "grad_norm": 54.651144165458696,
      "learning_rate": 3.1193041018174972e-06,
      "loss": 1.0423,
      "num_input_tokens_seen": 58692660,
      "step": 2758
    },
    {
      "epoch": 0.331750135273252,
      "flos": 16114489282560.0,
      "grad_norm": 5.1401240791751714,
      "learning_rate": 3.118658461562261e-06,
      "loss": 1.1352,
      "num_input_tokens_seen": 58708480,
      "step": 2759
    },
    {
      "epoch": 0.33187037816389103,
      "flos": 16166195834880.0,
      "grad_norm": 2.3649909274550254,
      "learning_rate": 3.118012651607426e-06,
      "loss": 1.0491,
      "num_input_tokens_seen": 58729805,
      "step": 2760
    },
    {
      "epoch": 0.33199062105453014,
      "flos": 13619277066240.0,
      "grad_norm": 4.89180597543422,
      "learning_rate": 3.1173666720509603e-06,
      "loss": 1.0272,
      "num_input_tokens_seen": 58746460,
      "step": 2761
    },
    {
      "epoch": 0.33211086394516925,
      "flos": 22512303636480.0,
      "grad_norm": 34.48802441609261,
      "learning_rate": 3.116720522990859e-06,
      "loss": 0.877,
      "num_input_tokens_seen": 58767055,
      "step": 2762
    },
    {
      "epoch": 0.3322311068358083,
      "flos": 12706410516480.0,
      "grad_norm": 5.208597324038822,
      "learning_rate": 3.116074204525142e-06,
      "loss": 0.8107,
      "num_input_tokens_seen": 58784950,
      "step": 2763
    },
    {
      "epoch": 0.3323513497264474,
      "flos": 23010168791040.0,
      "grad_norm": 3.3300552143640036,
      "learning_rate": 3.1154277167518553e-06,
      "loss": 1.0226,
      "num_input_tokens_seen": 58806285,
      "step": 2764
    },
    {
      "epoch": 0.33247159261708653,
      "flos": 37675801559040.0,
      "grad_norm": 0.8767891031665366,
      "learning_rate": 3.114781059769072e-06,
      "loss": 0.8207,
      "num_input_tokens_seen": 58857330,
      "step": 2765
    },
    {
      "epoch": 0.3325918355077256,
      "flos": 19312829276160.0,
      "grad_norm": 4.493438175083839,
      "learning_rate": 3.1141342336748874e-06,
      "loss": 0.8764,
      "num_input_tokens_seen": 58876610,
      "step": 2766
    },
    {
      "epoch": 0.3327120783983647,
      "flos": 16826324090880.0,
      "grad_norm": 2.3441070284706056,
      "learning_rate": 3.1134872385674253e-06,
      "loss": 1.0059,
      "num_input_tokens_seen": 58900485,
      "step": 2767
    },
    {
      "epoch": 0.3328323212890038,
      "flos": 13596550963200.0,
      "grad_norm": 3.6404101918253793,
      "learning_rate": 3.1128400745448353e-06,
      "loss": 1.0669,
      "num_input_tokens_seen": 58919585,
      "step": 2768
    },
    {
      "epoch": 0.33295256417964286,
      "flos": 26917308395520.0,
      "grad_norm": 5.98525460653748,
      "learning_rate": 3.11219274170529e-06,
      "loss": 0.8248,
      "num_input_tokens_seen": 58941115,
      "step": 2769
    },
    {
      "epoch": 0.333072807070282,
      "flos": 18867927674880.0,
      "grad_norm": 3.4149593336328636,
      "learning_rate": 3.1115452401469903e-06,
      "loss": 1.0131,
      "num_input_tokens_seen": 58961235,
      "step": 2770
    },
    {
      "epoch": 0.3331930499609211,
      "flos": 15220087296000.0,
      "grad_norm": 3.644781008030055,
      "learning_rate": 3.1108975699681613e-06,
      "loss": 1.0622,
      "num_input_tokens_seen": 58978350,
      "step": 2771
    },
    {
      "epoch": 0.33331329285156014,
      "flos": 14665364951040.0,
      "grad_norm": 2.296936034345324,
      "learning_rate": 3.1102497312670542e-06,
      "loss": 0.9127,
      "num_input_tokens_seen": 58996075,
      "step": 2772
    },
    {
      "epoch": 0.33343353574219925,
      "flos": 19942750740480.0,
      "grad_norm": 3.816190715470502,
      "learning_rate": 3.109601724141946e-06,
      "loss": 0.9842,
      "num_input_tokens_seen": 59014790,
      "step": 2773
    },
    {
      "epoch": 0.33355377863283836,
      "flos": 16897384304640.0,
      "grad_norm": 3.2479771407596103,
      "learning_rate": 3.108953548691138e-06,
      "loss": 0.8896,
      "num_input_tokens_seen": 59034595,
      "step": 2774
    },
    {
      "epoch": 0.3336740215234774,
      "flos": 26970057338880.0,
      "grad_norm": 4.406366224988093,
      "learning_rate": 3.108305205012959e-06,
      "loss": 0.9125,
      "num_input_tokens_seen": 59055010,
      "step": 2775
    },
    {
      "epoch": 0.3337942644141165,
      "flos": 18158392258560.0,
      "grad_norm": 6.90107664471285,
      "learning_rate": 3.107656693205761e-06,
      "loss": 1.0827,
      "num_input_tokens_seen": 59074170,
      "step": 2776
    },
    {
      "epoch": 0.3339145073047556,
      "flos": 18496263045120.0,
      "grad_norm": 7.2282190352217865,
      "learning_rate": 3.107008013367924e-06,
      "loss": 0.8869,
      "num_input_tokens_seen": 59092685,
      "step": 2777
    },
    {
      "epoch": 0.3340347501953947,
      "flos": 13518439280640.0,
      "grad_norm": 3.434548237729387,
      "learning_rate": 3.1063591655978507e-06,
      "loss": 1.0468,
      "num_input_tokens_seen": 59108355,
      "step": 2778
    },
    {
      "epoch": 0.3341549930860338,
      "flos": 12833070919680.0,
      "grad_norm": 3.5188345267135825,
      "learning_rate": 3.105710149993972e-06,
      "loss": 0.9901,
      "num_input_tokens_seen": 59127405,
      "step": 2779
    },
    {
      "epoch": 0.33427523597667286,
      "flos": 16114642575360.0,
      "grad_norm": 6.477201533800894,
      "learning_rate": 3.1050609666547427e-06,
      "loss": 1.0543,
      "num_input_tokens_seen": 59146685,
      "step": 2780
    },
    {
      "epoch": 0.33439547886731197,
      "flos": 16088298762240.0,
      "grad_norm": 7.639352017630002,
      "learning_rate": 3.104411615678644e-06,
      "loss": 0.9659,
      "num_input_tokens_seen": 59165255,
      "step": 2781
    },
    {
      "epoch": 0.3345157217579511,
      "flos": 17135980892160.0,
      "grad_norm": 9.401863367973343,
      "learning_rate": 3.1037620971641803e-06,
      "loss": 0.9197,
      "num_input_tokens_seen": 59184765,
      "step": 2782
    },
    {
      "epoch": 0.33463596464859013,
      "flos": 13465904947200.0,
      "grad_norm": 10.658291349907438,
      "learning_rate": 3.1031124112098844e-06,
      "loss": 0.8445,
      "num_input_tokens_seen": 59202695,
      "step": 2783
    },
    {
      "epoch": 0.33475620753922924,
      "flos": 14462033817600.0,
      "grad_norm": 6.87222178418134,
      "learning_rate": 3.1024625579143127e-06,
      "loss": 0.9131,
      "num_input_tokens_seen": 59219935,
      "step": 2784
    },
    {
      "epoch": 0.33487645042986836,
      "flos": 12885421301760.0,
      "grad_norm": 19.766282721270265,
      "learning_rate": 3.101812537376048e-06,
      "loss": 0.9141,
      "num_input_tokens_seen": 59238675,
      "step": 2785
    },
    {
      "epoch": 0.3349966933205074,
      "flos": 18395578552320.0,
      "grad_norm": 3.3166427617832994,
      "learning_rate": 3.1011623496936973e-06,
      "loss": 1.0346,
      "num_input_tokens_seen": 59256690,
      "step": 2786
    },
    {
      "epoch": 0.3351169362111465,
      "flos": 20021751521280.0,
      "grad_norm": 2.833433953536799,
      "learning_rate": 3.100511994965893e-06,
      "loss": 0.8877,
      "num_input_tokens_seen": 59276365,
      "step": 2787
    },
    {
      "epoch": 0.33523717910178563,
      "flos": 16113232281600.0,
      "grad_norm": 2.7025381344231714,
      "learning_rate": 3.0998614732912947e-06,
      "loss": 1.0549,
      "num_input_tokens_seen": 59295460,
      "step": 2788
    },
    {
      "epoch": 0.3353574219924247,
      "flos": 11080176230400.0,
      "grad_norm": 3.673678989820894,
      "learning_rate": 3.0992107847685855e-06,
      "loss": 0.8719,
      "num_input_tokens_seen": 59312895,
      "step": 2789
    },
    {
      "epoch": 0.3354776648830638,
      "flos": 17635072389120.0,
      "grad_norm": 3.466900856715255,
      "learning_rate": 3.0985599294964736e-06,
      "loss": 0.9932,
      "num_input_tokens_seen": 59332170,
      "step": 2790
    },
    {
      "epoch": 0.33559790777370285,
      "flos": 20440983797760.0,
      "grad_norm": 3.235962998507548,
      "learning_rate": 3.097908907573695e-06,
      "loss": 0.9003,
      "num_input_tokens_seen": 59349870,
      "step": 2791
    },
    {
      "epoch": 0.33571815066434196,
      "flos": 15798823403520.0,
      "grad_norm": 5.375113730003639,
      "learning_rate": 3.0972577190990067e-06,
      "loss": 1.0885,
      "num_input_tokens_seen": 59368070,
      "step": 2792
    },
    {
      "epoch": 0.3358383935549811,
      "flos": 16953536348160.0,
      "grad_norm": 6.4079751273826195,
      "learning_rate": 3.096606364171196e-06,
      "loss": 0.9964,
      "num_input_tokens_seen": 59387580,
      "step": 2793
    },
    {
      "epoch": 0.33595863644562013,
      "flos": 15820813701120.0,
      "grad_norm": 10.451857508520577,
      "learning_rate": 3.0959548428890703e-06,
      "loss": 1.0387,
      "num_input_tokens_seen": 59406170,
      "step": 2794
    },
    {
      "epoch": 0.33607887933625924,
      "flos": 14278117662720.0,
      "grad_norm": 6.108141059850472,
      "learning_rate": 3.095303155351468e-06,
      "loss": 1.0303,
      "num_input_tokens_seen": 59426095,
      "step": 2795
    },
    {
      "epoch": 0.33619912222689835,
      "flos": 13702110167040.0,
      "grad_norm": 3.809511107469,
      "learning_rate": 3.0946513016572464e-06,
      "loss": 0.9871,
      "num_input_tokens_seen": 59444720,
      "step": 2796
    },
    {
      "epoch": 0.3363193651175374,
      "flos": 11892879482880.0,
      "grad_norm": 5.409166200024934,
      "learning_rate": 3.0939992819052938e-06,
      "loss": 0.9559,
      "num_input_tokens_seen": 59461950,
      "step": 2797
    },
    {
      "epoch": 0.3364396080081765,
      "flos": 16743674941440.0,
      "grad_norm": 9.002758874833782,
      "learning_rate": 3.0933470961945193e-06,
      "loss": 0.9992,
      "num_input_tokens_seen": 59479965,
      "step": 2798
    },
    {
      "epoch": 0.3365598508988156,
      "flos": 19968818626560.0,
      "grad_norm": 6.173312171971172,
      "learning_rate": 3.0926947446238597e-06,
      "loss": 0.8856,
      "num_input_tokens_seen": 59499255,
      "step": 2799
    },
    {
      "epoch": 0.3366800937894547,
      "flos": 12023709450240.0,
      "grad_norm": 4.454468855615059,
      "learning_rate": 3.092042227292276e-06,
      "loss": 1.0152,
      "num_input_tokens_seen": 59515810,
      "step": 2800
    },
    {
      "epoch": 0.3368003366800938,
      "flos": 17006040023040.0,
      "grad_norm": 2.762387811100171,
      "learning_rate": 3.0913895442987557e-06,
      "loss": 1.0786,
      "num_input_tokens_seen": 59536495,
      "step": 2801
    },
    {
      "epoch": 0.3369205795707329,
      "flos": 17635562926080.0,
      "grad_norm": 4.3899227137268735,
      "learning_rate": 3.090736695742308e-06,
      "loss": 1.0507,
      "num_input_tokens_seen": 59557345,
      "step": 2802
    },
    {
      "epoch": 0.33704082246137196,
      "flos": 12705214832640.0,
      "grad_norm": 6.66651232858661,
      "learning_rate": 3.0900836817219713e-06,
      "loss": 0.6979,
      "num_input_tokens_seen": 59573495,
      "step": 2803
    },
    {
      "epoch": 0.33716106535201107,
      "flos": 15116520898560.0,
      "grad_norm": 3.6075989948735296,
      "learning_rate": 3.089430502336807e-06,
      "loss": 1.0369,
      "num_input_tokens_seen": 59593185,
      "step": 2804
    },
    {
      "epoch": 0.3372813082426502,
      "flos": 13044005376000.0,
      "grad_norm": 8.466740331053982,
      "learning_rate": 3.088777157685902e-06,
      "loss": 1.0939,
      "num_input_tokens_seen": 59608495,
      "step": 2805
    },
    {
      "epoch": 0.33740155113328923,
      "flos": 12180791255040.0,
      "grad_norm": 15.061207316158704,
      "learning_rate": 3.088123647868367e-06,
      "loss": 1.0552,
      "num_input_tokens_seen": 59624765,
      "step": 2806
    },
    {
      "epoch": 0.33752179402392835,
      "flos": 20702551756800.0,
      "grad_norm": 4.456227289385393,
      "learning_rate": 3.0874699729833405e-06,
      "loss": 1.0088,
      "num_input_tokens_seen": 59645855,
      "step": 2807
    },
    {
      "epoch": 0.3376420369145674,
      "flos": 17843308892160.0,
      "grad_norm": 6.191401629040073,
      "learning_rate": 3.086816133129983e-06,
      "loss": 0.9901,
      "num_input_tokens_seen": 59665835,
      "step": 2808
    },
    {
      "epoch": 0.3377622798052065,
      "flos": 19575500943360.0,
      "grad_norm": 4.3823722849470546,
      "learning_rate": 3.0861621284074826e-06,
      "loss": 0.966,
      "num_input_tokens_seen": 59686080,
      "step": 2809
    },
    {
      "epoch": 0.3378825226958456,
      "flos": 15610860318720.0,
      "grad_norm": 2.5726060461016416,
      "learning_rate": 3.085507958915051e-06,
      "loss": 0.9224,
      "num_input_tokens_seen": 59704230,
      "step": 2810
    },
    {
      "epoch": 0.3380027655864847,
      "flos": 30379178496000.0,
      "grad_norm": 3.931069951869554,
      "learning_rate": 3.084853624751925e-06,
      "loss": 0.9022,
      "num_input_tokens_seen": 59725535,
      "step": 2811
    },
    {
      "epoch": 0.3381230084771238,
      "flos": 19025408040960.0,
      "grad_norm": 3.2354041779509015,
      "learning_rate": 3.0841991260173668e-06,
      "loss": 1.0439,
      "num_input_tokens_seen": 59745160,
      "step": 2812
    },
    {
      "epoch": 0.3382432513677629,
      "flos": 16139943997440.0,
      "grad_norm": 3.43596198614051,
      "learning_rate": 3.0835444628106634e-06,
      "loss": 0.9966,
      "num_input_tokens_seen": 59763860,
      "step": 2813
    },
    {
      "epoch": 0.33836349425840195,
      "flos": 15716818083840.0,
      "grad_norm": 4.579169561068748,
      "learning_rate": 3.082889635231126e-06,
      "loss": 1.034,
      "num_input_tokens_seen": 59782240,
      "step": 2814
    },
    {
      "epoch": 0.33848373714904106,
      "flos": 19444303073280.0,
      "grad_norm": 4.869814144089447,
      "learning_rate": 3.0822346433780925e-06,
      "loss": 0.944,
      "num_input_tokens_seen": 59802685,
      "step": 2815
    },
    {
      "epoch": 0.3386039800396802,
      "flos": 18395486576640.0,
      "grad_norm": 5.22253536521,
      "learning_rate": 3.0815794873509237e-06,
      "loss": 1.0411,
      "num_input_tokens_seen": 59820690,
      "step": 2816
    },
    {
      "epoch": 0.33872422293031923,
      "flos": 13387762606080.0,
      "grad_norm": 9.852880887323522,
      "learning_rate": 3.0809241672490066e-06,
      "loss": 0.9169,
      "num_input_tokens_seen": 59838580,
      "step": 2817
    },
    {
      "epoch": 0.33884446582095834,
      "flos": 16453555752960.0,
      "grad_norm": 3.7457319767145427,
      "learning_rate": 3.080268683171753e-06,
      "loss": 1.0392,
      "num_input_tokens_seen": 59858590,
      "step": 2818
    },
    {
      "epoch": 0.33896470871159745,
      "flos": 11315461693440.0,
      "grad_norm": 4.69249001974956,
      "learning_rate": 3.0796130352185985e-06,
      "loss": 1.0765,
      "num_input_tokens_seen": 59875165,
      "step": 2819
    },
    {
      "epoch": 0.3390849516022365,
      "flos": 24609936629760.0,
      "grad_norm": 3.415697487553172,
      "learning_rate": 3.0789572234890057e-06,
      "loss": 0.868,
      "num_input_tokens_seen": 59896525,
      "step": 2820
    },
    {
      "epoch": 0.3392051944928756,
      "flos": 11447027466240.0,
      "grad_norm": 3.92090646403288,
      "learning_rate": 3.0783012480824596e-06,
      "loss": 0.9671,
      "num_input_tokens_seen": 59915390,
      "step": 2821
    },
    {
      "epoch": 0.33932543738351467,
      "flos": 12097620910080.0,
      "grad_norm": 4.395965694706692,
      "learning_rate": 3.077645109098471e-06,
      "loss": 0.9428,
      "num_input_tokens_seen": 59931380,
      "step": 2822
    },
    {
      "epoch": 0.3394456802741538,
      "flos": 15720711720960.0,
      "grad_norm": 4.027881012978401,
      "learning_rate": 3.076988806636577e-06,
      "loss": 0.9105,
      "num_input_tokens_seen": 59948860,
      "step": 2823
    },
    {
      "epoch": 0.3395659231647929,
      "flos": 17948377559040.0,
      "grad_norm": 3.4462777167566143,
      "learning_rate": 3.0763323407963377e-06,
      "loss": 1.0753,
      "num_input_tokens_seen": 59968190,
      "step": 2824
    },
    {
      "epoch": 0.33968616605543195,
      "flos": 20729662033920.0,
      "grad_norm": 5.353401436156783,
      "learning_rate": 3.075675711677337e-06,
      "loss": 0.989,
      "num_input_tokens_seen": 59988005,
      "step": 2825
    },
    {
      "epoch": 0.33980640894607106,
      "flos": 15222693273600.0,
      "grad_norm": 5.095454580226419,
      "learning_rate": 3.0750189193791865e-06,
      "loss": 0.9732,
      "num_input_tokens_seen": 60007310,
      "step": 2826
    },
    {
      "epoch": 0.33992665183671017,
      "flos": 23168599572480.0,
      "grad_norm": 9.567349270205696,
      "learning_rate": 3.0743619640015203e-06,
      "loss": 0.8921,
      "num_input_tokens_seen": 60027280,
      "step": 2827
    },
    {
      "epoch": 0.3400468947273492,
      "flos": 12075875880960.0,
      "grad_norm": 5.52884807267682,
      "learning_rate": 3.073704845643999e-06,
      "loss": 1.116,
      "num_input_tokens_seen": 60044125,
      "step": 2828
    },
    {
      "epoch": 0.34016713761798834,
      "flos": 11944769986560.0,
      "grad_norm": 6.461854364546858,
      "learning_rate": 3.0730475644063063e-06,
      "loss": 0.9881,
      "num_input_tokens_seen": 60058945,
      "step": 2829
    },
    {
      "epoch": 0.34028738050862745,
      "flos": 15562618183680.0,
      "grad_norm": 5.0082299211756816,
      "learning_rate": 3.072390120388151e-06,
      "loss": 0.8551,
      "num_input_tokens_seen": 60076990,
      "step": 2830
    },
    {
      "epoch": 0.3404076233992665,
      "flos": 16165460029440.0,
      "grad_norm": 7.398962113049492,
      "learning_rate": 3.071732513689267e-06,
      "loss": 0.8921,
      "num_input_tokens_seen": 60095245,
      "step": 2831
    },
    {
      "epoch": 0.3405278662899056,
      "flos": 12073269903360.0,
      "grad_norm": 10.32442381500374,
      "learning_rate": 3.0710747444094134e-06,
      "loss": 0.8683,
      "num_input_tokens_seen": 60112995,
      "step": 2832
    },
    {
      "epoch": 0.3406481091805447,
      "flos": 30587292364800.0,
      "grad_norm": 4.335608887463078,
      "learning_rate": 3.070416812648372e-06,
      "loss": 0.849,
      "num_input_tokens_seen": 60136165,
      "step": 2833
    },
    {
      "epoch": 0.3407683520711838,
      "flos": 19051537244160.0,
      "grad_norm": 4.032050238144957,
      "learning_rate": 3.069758718505951e-06,
      "loss": 0.8369,
      "num_input_tokens_seen": 60157625,
      "step": 2834
    },
    {
      "epoch": 0.3408885949618229,
      "flos": 20493272862720.0,
      "grad_norm": 6.645627137177252,
      "learning_rate": 3.0691004620819836e-06,
      "loss": 0.9878,
      "num_input_tokens_seen": 60177475,
      "step": 2835
    },
    {
      "epoch": 0.341008837852462,
      "flos": 45514646384640.0,
      "grad_norm": 0.8318335069816569,
      "learning_rate": 3.0684420434763254e-06,
      "loss": 0.8279,
      "num_input_tokens_seen": 60243380,
      "step": 2836
    },
    {
      "epoch": 0.34112908074310105,
      "flos": 14775614914560.0,
      "grad_norm": 14.804051045851066,
      "learning_rate": 3.06778346278886e-06,
      "loss": 0.9545,
      "num_input_tokens_seen": 60261935,
      "step": 2837
    },
    {
      "epoch": 0.34124932363374016,
      "flos": 17769060188160.0,
      "grad_norm": 5.075876154694179,
      "learning_rate": 3.0671247201194906e-06,
      "loss": 0.9834,
      "num_input_tokens_seen": 60283790,
      "step": 2838
    },
    {
      "epoch": 0.3413695665243792,
      "flos": 20231214366720.0,
      "grad_norm": 4.568144559600454,
      "learning_rate": 3.066465815568151e-06,
      "loss": 0.9393,
      "num_input_tokens_seen": 60304340,
      "step": 2839
    },
    {
      "epoch": 0.34148980941501833,
      "flos": 18002659430400.0,
      "grad_norm": 4.778833514139133,
      "learning_rate": 3.0658067492347947e-06,
      "loss": 0.8842,
      "num_input_tokens_seen": 60326700,
      "step": 2840
    },
    {
      "epoch": 0.34161005230565744,
      "flos": 12417057792000.0,
      "grad_norm": 4.33593222128057,
      "learning_rate": 3.065147521219402e-06,
      "loss": 0.8646,
      "num_input_tokens_seen": 60345675,
      "step": 2841
    },
    {
      "epoch": 0.3417302951962965,
      "flos": 31188509306880.0,
      "grad_norm": 4.581864532013324,
      "learning_rate": 3.064488131621977e-06,
      "loss": 0.946,
      "num_input_tokens_seen": 60368720,
      "step": 2842
    },
    {
      "epoch": 0.3418505380869356,
      "flos": 22012752261120.0,
      "grad_norm": 4.879963622035781,
      "learning_rate": 3.063828580542549e-06,
      "loss": 0.9288,
      "num_input_tokens_seen": 60389635,
      "step": 2843
    },
    {
      "epoch": 0.3419707809775747,
      "flos": 13806841589760.0,
      "grad_norm": 3.833299650514205,
      "learning_rate": 3.0631688680811706e-06,
      "loss": 0.9114,
      "num_input_tokens_seen": 60408980,
      "step": 2844
    },
    {
      "epoch": 0.3420910238682138,
      "flos": 20464200437760.0,
      "grad_norm": 5.101387149431875,
      "learning_rate": 3.062508994337921e-06,
      "loss": 0.9527,
      "num_input_tokens_seen": 60428305,
      "step": 2845
    },
    {
      "epoch": 0.3422112667588529,
      "flos": 15196288143360.0,
      "grad_norm": 6.614088753860604,
      "learning_rate": 3.0618489594129013e-06,
      "loss": 0.9758,
      "num_input_tokens_seen": 60446165,
      "step": 2846
    },
    {
      "epoch": 0.342331509649492,
      "flos": 9794847928320.0,
      "grad_norm": 4.1866360812274745,
      "learning_rate": 3.061188763406239e-06,
      "loss": 0.9028,
      "num_input_tokens_seen": 60462030,
      "step": 2847
    },
    {
      "epoch": 0.34245175254013105,
      "flos": 20388173537280.0,
      "grad_norm": 5.174863199830783,
      "learning_rate": 3.060528406418085e-06,
      "loss": 1.0128,
      "num_input_tokens_seen": 60481600,
      "step": 2848
    },
    {
      "epoch": 0.34257199543077016,
      "flos": 24344444375040.0,
      "grad_norm": 6.3483527262137684,
      "learning_rate": 3.0598678885486145e-06,
      "loss": 0.8101,
      "num_input_tokens_seen": 60503860,
      "step": 2849
    },
    {
      "epoch": 0.34269223832140927,
      "flos": 14173386240000.0,
      "grad_norm": 5.835411936676204,
      "learning_rate": 3.0592072098980282e-06,
      "loss": 0.9407,
      "num_input_tokens_seen": 60523240,
      "step": 2850
    },
    {
      "epoch": 0.3428124812120483,
      "flos": 19392320593920.0,
      "grad_norm": 10.577045966014683,
      "learning_rate": 3.0585463705665514e-06,
      "loss": 0.9152,
      "num_input_tokens_seen": 60543335,
      "step": 2851
    },
    {
      "epoch": 0.34293272410268744,
      "flos": 17477009510400.0,
      "grad_norm": 7.416891058909813,
      "learning_rate": 3.0578853706544304e-06,
      "loss": 0.8904,
      "num_input_tokens_seen": 60560445,
      "step": 2852
    },
    {
      "epoch": 0.34305296699332655,
      "flos": 15275074314240.0,
      "grad_norm": 3.51304294286076,
      "learning_rate": 3.0572242102619404e-06,
      "loss": 0.8378,
      "num_input_tokens_seen": 60577320,
      "step": 2853
    },
    {
      "epoch": 0.3431732098839656,
      "flos": 17107000442880.0,
      "grad_norm": 5.265736926782431,
      "learning_rate": 3.0565628894893784e-06,
      "loss": 1.0086,
      "num_input_tokens_seen": 60597675,
      "step": 2854
    },
    {
      "epoch": 0.3432934527746047,
      "flos": 11893063434240.0,
      "grad_norm": 13.199074244547832,
      "learning_rate": 3.0559014084370655e-06,
      "loss": 0.9429,
      "num_input_tokens_seen": 60615920,
      "step": 2855
    },
    {
      "epoch": 0.34341369566524377,
      "flos": 16664060989440.0,
      "grad_norm": 3.7025127951061565,
      "learning_rate": 3.055239767205349e-06,
      "loss": 0.9784,
      "num_input_tokens_seen": 60637390,
      "step": 2856
    },
    {
      "epoch": 0.3435339385558829,
      "flos": 12227347169280.0,
      "grad_norm": 3.489617616095979,
      "learning_rate": 3.054577965894599e-06,
      "loss": 0.9637,
      "num_input_tokens_seen": 60653255,
      "step": 2857
    },
    {
      "epoch": 0.343654181446522,
      "flos": 15772663541760.0,
      "grad_norm": 4.187763945957966,
      "learning_rate": 3.0539160046052094e-06,
      "loss": 0.9195,
      "num_input_tokens_seen": 60672675,
      "step": 2858
    },
    {
      "epoch": 0.34377442433716104,
      "flos": 14121587712000.0,
      "grad_norm": 6.95451893067586,
      "learning_rate": 3.0532538834376003e-06,
      "loss": 0.891,
      "num_input_tokens_seen": 60691955,
      "step": 2859
    },
    {
      "epoch": 0.34389466722780015,
      "flos": 15771222589440.0,
      "grad_norm": 3.9856483967731267,
      "learning_rate": 3.0525916024922143e-06,
      "loss": 0.9738,
      "num_input_tokens_seen": 60710860,
      "step": 2860
    },
    {
      "epoch": 0.34401491011843927,
      "flos": 13203631841280.0,
      "grad_norm": 9.007059507501863,
      "learning_rate": 3.0519291618695193e-06,
      "loss": 1.0275,
      "num_input_tokens_seen": 60727980,
      "step": 2861
    },
    {
      "epoch": 0.3441351530090783,
      "flos": 12652833792000.0,
      "grad_norm": 8.073163218681309,
      "learning_rate": 3.0512665616700065e-06,
      "loss": 0.9453,
      "num_input_tokens_seen": 60746765,
      "step": 2862
    },
    {
      "epoch": 0.34425539589971743,
      "flos": 16429266063360.0,
      "grad_norm": 4.963564295948983,
      "learning_rate": 3.0506038019941933e-06,
      "loss": 1.0909,
      "num_input_tokens_seen": 60766495,
      "step": 2863
    },
    {
      "epoch": 0.34437563879035654,
      "flos": 15563078062080.0,
      "grad_norm": 6.959795617358343,
      "learning_rate": 3.049940882942617e-06,
      "loss": 0.8651,
      "num_input_tokens_seen": 60785000,
      "step": 2864
    },
    {
      "epoch": 0.3444958816809956,
      "flos": 16403596738560.0,
      "grad_norm": 3.566105139431326,
      "learning_rate": 3.0492778046158448e-06,
      "loss": 0.9842,
      "num_input_tokens_seen": 60806140,
      "step": 2865
    },
    {
      "epoch": 0.3446161245716347,
      "flos": 15563875184640.0,
      "grad_norm": 4.025323814492871,
      "learning_rate": 3.0486145671144633e-06,
      "loss": 0.963,
      "num_input_tokens_seen": 60825650,
      "step": 2866
    },
    {
      "epoch": 0.3447363674622738,
      "flos": 17865789726720.0,
      "grad_norm": 7.174090805665862,
      "learning_rate": 3.047951170539086e-06,
      "loss": 0.9368,
      "num_input_tokens_seen": 60844995,
      "step": 2867
    },
    {
      "epoch": 0.3448566103529129,
      "flos": 8431745187840.0,
      "grad_norm": 3.3992356338730456,
      "learning_rate": 3.047287614990349e-06,
      "loss": 1.0327,
      "num_input_tokens_seen": 60862635,
      "step": 2868
    },
    {
      "epoch": 0.344976853243552,
      "flos": 28700808437760.0,
      "grad_norm": 4.206740402042972,
      "learning_rate": 3.046623900568914e-06,
      "loss": 0.7947,
      "num_input_tokens_seen": 60884920,
      "step": 2869
    },
    {
      "epoch": 0.34509709613419104,
      "flos": 20460644044800.0,
      "grad_norm": 3.974054951157794,
      "learning_rate": 3.045960027375465e-06,
      "loss": 0.886,
      "num_input_tokens_seen": 60902475,
      "step": 2870
    },
    {
      "epoch": 0.34521733902483015,
      "flos": 21355107348480.0,
      "grad_norm": 5.729693172421858,
      "learning_rate": 3.045295995510711e-06,
      "loss": 1.0122,
      "num_input_tokens_seen": 60919165,
      "step": 2871
    },
    {
      "epoch": 0.34533758191546926,
      "flos": 19885556305920.0,
      "grad_norm": 4.890952199892252,
      "learning_rate": 3.0446318050753865e-06,
      "loss": 0.9277,
      "num_input_tokens_seen": 60939365,
      "step": 2872
    },
    {
      "epoch": 0.3454578248061083,
      "flos": 19313381130240.0,
      "grad_norm": 4.247736833154256,
      "learning_rate": 3.0439674561702474e-06,
      "loss": 0.9792,
      "num_input_tokens_seen": 60958585,
      "step": 2873
    },
    {
      "epoch": 0.3455780676967474,
      "flos": 13491206369280.0,
      "grad_norm": 3.1258542228481416,
      "learning_rate": 3.043302948896076e-06,
      "loss": 1.0663,
      "num_input_tokens_seen": 60976910,
      "step": 2874
    },
    {
      "epoch": 0.34569831058738654,
      "flos": 24610151239680.0,
      "grad_norm": 5.2889956043244615,
      "learning_rate": 3.0426382833536756e-06,
      "loss": 0.7923,
      "num_input_tokens_seen": 60999985,
      "step": 2875
    },
    {
      "epoch": 0.3458185534780256,
      "flos": 22197220270080.0,
      "grad_norm": 5.478828429405533,
      "learning_rate": 3.041973459643877e-06,
      "loss": 0.9822,
      "num_input_tokens_seen": 61019160,
      "step": 2876
    },
    {
      "epoch": 0.3459387963686647,
      "flos": 23141489295360.0,
      "grad_norm": 4.826930024740893,
      "learning_rate": 3.0413084778675334e-06,
      "loss": 0.8662,
      "num_input_tokens_seen": 61040130,
      "step": 2877
    },
    {
      "epoch": 0.3460590392593038,
      "flos": 17551687434240.0,
      "grad_norm": 3.3880440869756296,
      "learning_rate": 3.0406433381255214e-06,
      "loss": 1.0242,
      "num_input_tokens_seen": 61057885,
      "step": 2878
    },
    {
      "epoch": 0.34617928214994287,
      "flos": 13335013662720.0,
      "grad_norm": 4.807437760877101,
      "learning_rate": 3.0399780405187425e-06,
      "loss": 1.0134,
      "num_input_tokens_seen": 61076600,
      "step": 2879
    },
    {
      "epoch": 0.346299525040582,
      "flos": 17682425425920.0,
      "grad_norm": 3.9450505840118955,
      "learning_rate": 3.0393125851481216e-06,
      "loss": 0.9768,
      "num_input_tokens_seen": 61096195,
      "step": 2880
    },
    {
      "epoch": 0.3464197679312211,
      "flos": 11629901230080.0,
      "grad_norm": 3.4128150319436803,
      "learning_rate": 3.038646972114608e-06,
      "loss": 1.048,
      "num_input_tokens_seen": 61112240,
      "step": 2881
    },
    {
      "epoch": 0.34654001082186014,
      "flos": 15904351948800.0,
      "grad_norm": 5.541106499298878,
      "learning_rate": 3.037981201519174e-06,
      "loss": 0.8743,
      "num_input_tokens_seen": 61132560,
      "step": 2882
    },
    {
      "epoch": 0.34666025371249926,
      "flos": 13885995663360.0,
      "grad_norm": 4.330455827220593,
      "learning_rate": 3.0373152734628175e-06,
      "loss": 0.8933,
      "num_input_tokens_seen": 61150560,
      "step": 2883
    },
    {
      "epoch": 0.34678049660313837,
      "flos": 10791038115840.0,
      "grad_norm": 2.822959017465243,
      "learning_rate": 3.0366491880465584e-06,
      "loss": 0.9486,
      "num_input_tokens_seen": 61168300,
      "step": 2884
    },
    {
      "epoch": 0.3469007394937774,
      "flos": 15041689681920.0,
      "grad_norm": 3.5952808602543596,
      "learning_rate": 3.035982945371443e-06,
      "loss": 1.0147,
      "num_input_tokens_seen": 61189715,
      "step": 2885
    },
    {
      "epoch": 0.34702098238441653,
      "flos": 15898373529600.0,
      "grad_norm": 5.779077730088114,
      "learning_rate": 3.035316545538537e-06,
      "loss": 1.047,
      "num_input_tokens_seen": 61208230,
      "step": 2886
    },
    {
      "epoch": 0.3471412252750556,
      "flos": 16297393704960.0,
      "grad_norm": 8.68200993133237,
      "learning_rate": 3.034649988648935e-06,
      "loss": 0.9761,
      "num_input_tokens_seen": 61227715,
      "step": 2887
    },
    {
      "epoch": 0.3472614681656947,
      "flos": 15143815127040.0,
      "grad_norm": 5.024526902128195,
      "learning_rate": 3.033983274803752e-06,
      "loss": 1.0043,
      "num_input_tokens_seen": 61247225,
      "step": 2888
    },
    {
      "epoch": 0.3473817110563338,
      "flos": 16689883607040.0,
      "grad_norm": 8.645982953963903,
      "learning_rate": 3.0333164041041283e-06,
      "loss": 0.9187,
      "num_input_tokens_seen": 61263260,
      "step": 2889
    },
    {
      "epoch": 0.34750195394697286,
      "flos": 15878161428480.0,
      "grad_norm": 6.692915675492842,
      "learning_rate": 3.032649376651228e-06,
      "loss": 0.9125,
      "num_input_tokens_seen": 61282400,
      "step": 2890
    },
    {
      "epoch": 0.347622196837612,
      "flos": 20728405032960.0,
      "grad_norm": 4.616939221724671,
      "learning_rate": 3.031982192546238e-06,
      "loss": 0.9559,
      "num_input_tokens_seen": 61305215,
      "step": 2891
    },
    {
      "epoch": 0.3477424397282511,
      "flos": 15957039575040.0,
      "grad_norm": 4.842698899941855,
      "learning_rate": 3.0313148518903696e-06,
      "loss": 1.1393,
      "num_input_tokens_seen": 61324760,
      "step": 2892
    },
    {
      "epoch": 0.34786268261889014,
      "flos": 11158441205760.0,
      "grad_norm": 4.6301679863332215,
      "learning_rate": 3.030647354784859e-06,
      "loss": 0.9959,
      "num_input_tokens_seen": 61341520,
      "step": 2893
    },
    {
      "epoch": 0.34798292550952925,
      "flos": 14751049297920.0,
      "grad_norm": 3.3993322838752156,
      "learning_rate": 3.029979701330964e-06,
      "loss": 0.9643,
      "num_input_tokens_seen": 61360665,
      "step": 2894
    },
    {
      "epoch": 0.34810316840016836,
      "flos": 14147410329600.0,
      "grad_norm": 8.009248632653875,
      "learning_rate": 3.029311891629966e-06,
      "loss": 0.9917,
      "num_input_tokens_seen": 61378840,
      "step": 2895
    },
    {
      "epoch": 0.3482234112908074,
      "flos": 16796147957760.0,
      "grad_norm": 4.59624159317555,
      "learning_rate": 3.0286439257831744e-06,
      "loss": 0.9338,
      "num_input_tokens_seen": 61398030,
      "step": 2896
    },
    {
      "epoch": 0.3483436541814465,
      "flos": 17057133404160.0,
      "grad_norm": 3.5022494035990306,
      "learning_rate": 3.0279758038919156e-06,
      "loss": 0.9056,
      "num_input_tokens_seen": 61415975,
      "step": 2897
    },
    {
      "epoch": 0.34846389707208564,
      "flos": 16088298762240.0,
      "grad_norm": 4.69658279372934,
      "learning_rate": 3.0273075260575455e-06,
      "loss": 0.9718,
      "num_input_tokens_seen": 61434595,
      "step": 2898
    },
    {
      "epoch": 0.3485841399627247,
      "flos": 15479815741440.0,
      "grad_norm": 8.122725417794856,
      "learning_rate": 3.0266390923814396e-06,
      "loss": 0.9927,
      "num_input_tokens_seen": 61452375,
      "step": 2899
    },
    {
      "epoch": 0.3487043828533638,
      "flos": 12049746677760.0,
      "grad_norm": 5.239425909611156,
      "learning_rate": 3.0259705029650008e-06,
      "loss": 1.0215,
      "num_input_tokens_seen": 61470025,
      "step": 2900
    },
    {
      "epoch": 0.34882462574400286,
      "flos": 16061464412160.0,
      "grad_norm": 7.413055255028882,
      "learning_rate": 3.025301757909652e-06,
      "loss": 0.9249,
      "num_input_tokens_seen": 61489940,
      "step": 2901
    },
    {
      "epoch": 0.34894486863464197,
      "flos": 21279325716480.0,
      "grad_norm": 5.626972107092245,
      "learning_rate": 3.024632857316842e-06,
      "loss": 1.0091,
      "num_input_tokens_seen": 61510975,
      "step": 2902
    },
    {
      "epoch": 0.3490651115252811,
      "flos": 15716695449600.0,
      "grad_norm": 3.677533015058228,
      "learning_rate": 3.0239638012880412e-06,
      "loss": 0.9659,
      "num_input_tokens_seen": 61530590,
      "step": 2903
    },
    {
      "epoch": 0.34918535441592014,
      "flos": 8928905195520.0,
      "grad_norm": 5.744089820945852,
      "learning_rate": 3.0232945899247466e-06,
      "loss": 1.0098,
      "num_input_tokens_seen": 61547245,
      "step": 2904
    },
    {
      "epoch": 0.34930559730655925,
      "flos": 16481616445440.0,
      "grad_norm": 6.672000499679584,
      "learning_rate": 3.022625223328476e-06,
      "loss": 0.97,
      "num_input_tokens_seen": 61568705,
      "step": 2905
    },
    {
      "epoch": 0.34942584019719836,
      "flos": 16243725004800.0,
      "grad_norm": 4.074237241913332,
      "learning_rate": 3.0219557016007723e-06,
      "loss": 0.8879,
      "num_input_tokens_seen": 61588555,
      "step": 2906
    },
    {
      "epoch": 0.3495460830878374,
      "flos": 17373075210240.0,
      "grad_norm": 8.14622940180283,
      "learning_rate": 3.021286024843202e-06,
      "loss": 0.8939,
      "num_input_tokens_seen": 61606470,
      "step": 2907
    },
    {
      "epoch": 0.3496663259784765,
      "flos": 50134119444480.0,
      "grad_norm": 1.1191804089889552,
      "learning_rate": 3.0206161931573526e-06,
      "loss": 0.8994,
      "num_input_tokens_seen": 61658740,
      "step": 2908
    },
    {
      "epoch": 0.34978656886911563,
      "flos": 20439236259840.0,
      "grad_norm": 3.2581896666507353,
      "learning_rate": 3.0199462066448388e-06,
      "loss": 1.1334,
      "num_input_tokens_seen": 61680655,
      "step": 2909
    },
    {
      "epoch": 0.3499068117597547,
      "flos": 15012985159680.0,
      "grad_norm": 4.6644460753600105,
      "learning_rate": 3.019276065407296e-06,
      "loss": 0.8833,
      "num_input_tokens_seen": 61699495,
      "step": 2910
    },
    {
      "epoch": 0.3500270546503938,
      "flos": 16191834501120.0,
      "grad_norm": 6.7938321021768235,
      "learning_rate": 3.018605769546385e-06,
      "loss": 1.0061,
      "num_input_tokens_seen": 61719770,
      "step": 2911
    },
    {
      "epoch": 0.3501472975410329,
      "flos": 16272122941440.0,
      "grad_norm": 47.95980678032648,
      "learning_rate": 3.017935319163788e-06,
      "loss": 0.9924,
      "num_input_tokens_seen": 61738450,
      "step": 2912
    },
    {
      "epoch": 0.35026754043167196,
      "flos": 18106379120640.0,
      "grad_norm": 3.753978653129159,
      "learning_rate": 3.017264714361213e-06,
      "loss": 0.9088,
      "num_input_tokens_seen": 61757820,
      "step": 2913
    },
    {
      "epoch": 0.3503877833223111,
      "flos": 13885873029120.0,
      "grad_norm": 14.03995588768444,
      "learning_rate": 3.016593955240389e-06,
      "loss": 1.0143,
      "num_input_tokens_seen": 61776230,
      "step": 2914
    },
    {
      "epoch": 0.3505080262129502,
      "flos": 46589653401600.0,
      "grad_norm": 0.8296925445805586,
      "learning_rate": 3.015923041903071e-06,
      "loss": 0.8482,
      "num_input_tokens_seen": 61842075,
      "step": 2915
    },
    {
      "epoch": 0.35062826910358924,
      "flos": 20886283960320.0,
      "grad_norm": 6.302503167242021,
      "learning_rate": 3.0152519744510347e-06,
      "loss": 1.0296,
      "num_input_tokens_seen": 61861595,
      "step": 2916
    },
    {
      "epoch": 0.35074851199422835,
      "flos": 17057991843840.0,
      "grad_norm": 6.932044831392364,
      "learning_rate": 3.014580752986081e-06,
      "loss": 1.0204,
      "num_input_tokens_seen": 61880190,
      "step": 2917
    },
    {
      "epoch": 0.3508687548848674,
      "flos": 10759083786240.0,
      "grad_norm": 10.464825523733662,
      "learning_rate": 3.0139093776100345e-06,
      "loss": 0.9745,
      "num_input_tokens_seen": 61896500,
      "step": 2918
    },
    {
      "epoch": 0.3509889977755065,
      "flos": 15170894745600.0,
      "grad_norm": 3.7032081820002234,
      "learning_rate": 3.013237848424741e-06,
      "loss": 0.9433,
      "num_input_tokens_seen": 61915605,
      "step": 2919
    },
    {
      "epoch": 0.35110924066614563,
      "flos": 13570728345600.0,
      "grad_norm": 4.81198910069678,
      "learning_rate": 3.012566165532072e-06,
      "loss": 0.9556,
      "num_input_tokens_seen": 61934115,
      "step": 2920
    },
    {
      "epoch": 0.3512294835567847,
      "flos": 15615704371200.0,
      "grad_norm": 4.295423671800451,
      "learning_rate": 3.0118943290339207e-06,
      "loss": 0.9472,
      "num_input_tokens_seen": 61954045,
      "step": 2921
    },
    {
      "epoch": 0.3513497264474238,
      "flos": 12622841610240.0,
      "grad_norm": 5.28865660514405,
      "learning_rate": 3.011222339032204e-06,
      "loss": 0.8598,
      "num_input_tokens_seen": 61971915,
      "step": 2922
    },
    {
      "epoch": 0.3514699693380629,
      "flos": 19182244577280.0,
      "grad_norm": 3.5434684358991233,
      "learning_rate": 3.0105501956288626e-06,
      "loss": 0.8758,
      "num_input_tokens_seen": 61992105,
      "step": 2923
    },
    {
      "epoch": 0.35159021222870196,
      "flos": 10790976798720.0,
      "grad_norm": 27.190536154104358,
      "learning_rate": 3.0098778989258602e-06,
      "loss": 0.9167,
      "num_input_tokens_seen": 62010435,
      "step": 2924
    },
    {
      "epoch": 0.35171045511934107,
      "flos": 9867747655680.0,
      "grad_norm": 6.351126154964905,
      "learning_rate": 3.009205449025183e-06,
      "loss": 1.0756,
      "num_input_tokens_seen": 62026350,
      "step": 2925
    },
    {
      "epoch": 0.3518306980099802,
      "flos": 10083832750080.0,
      "grad_norm": 4.6453145283458195,
      "learning_rate": 3.008532846028842e-06,
      "loss": 0.8216,
      "num_input_tokens_seen": 62042830,
      "step": 2926
    },
    {
      "epoch": 0.35195094090061924,
      "flos": 19260908113920.0,
      "grad_norm": 5.634057615018734,
      "learning_rate": 3.0078600900388694e-06,
      "loss": 0.8951,
      "num_input_tokens_seen": 62062855,
      "step": 2927
    },
    {
      "epoch": 0.35207118379125835,
      "flos": 18237699624960.0,
      "grad_norm": 3.1217158035454586,
      "learning_rate": 3.007187181157323e-06,
      "loss": 0.9389,
      "num_input_tokens_seen": 62082585,
      "step": 2928
    },
    {
      "epoch": 0.35219142668189746,
      "flos": 12758209044480.0,
      "grad_norm": 4.46012804266676,
      "learning_rate": 3.006514119486282e-06,
      "loss": 0.8747,
      "num_input_tokens_seen": 62099135,
      "step": 2929
    },
    {
      "epoch": 0.3523116695725365,
      "flos": 9900345815040.0,
      "grad_norm": 5.5273354681916755,
      "learning_rate": 3.005840905127849e-06,
      "loss": 0.8874,
      "num_input_tokens_seen": 62115760,
      "step": 2930
    },
    {
      "epoch": 0.3524319124631756,
      "flos": 15222693273600.0,
      "grad_norm": 5.69336272512766,
      "learning_rate": 3.0051675381841516e-06,
      "loss": 1.059,
      "num_input_tokens_seen": 62132790,
      "step": 2931
    },
    {
      "epoch": 0.3525521553538147,
      "flos": 18736147292160.0,
      "grad_norm": 10.872746376584494,
      "learning_rate": 3.0044940187573363e-06,
      "loss": 0.9581,
      "num_input_tokens_seen": 62153520,
      "step": 2932
    },
    {
      "epoch": 0.3526723982444538,
      "flos": 15301264834560.0,
      "grad_norm": 6.245573994361364,
      "learning_rate": 3.003820346949578e-06,
      "loss": 0.8667,
      "num_input_tokens_seen": 62171320,
      "step": 2933
    },
    {
      "epoch": 0.3527926411350929,
      "flos": 16875577958400.0,
      "grad_norm": 3.707722059867754,
      "learning_rate": 3.003146522863071e-06,
      "loss": 0.98,
      "num_input_tokens_seen": 62191925,
      "step": 2934
    },
    {
      "epoch": 0.35291288402573195,
      "flos": 21699017871360.0,
      "grad_norm": 3.7491117061111825,
      "learning_rate": 3.0024725466000345e-06,
      "loss": 1.0377,
      "num_input_tokens_seen": 62211600,
      "step": 2935
    },
    {
      "epoch": 0.35303312691637107,
      "flos": 16428775526400.0,
      "grad_norm": 4.330227888422713,
      "learning_rate": 3.0017984182627087e-06,
      "loss": 0.9821,
      "num_input_tokens_seen": 62230645,
      "step": 2936
    },
    {
      "epoch": 0.3531533698070102,
      "flos": 15511463485440.0,
      "grad_norm": 4.583163148436943,
      "learning_rate": 3.00112413795336e-06,
      "loss": 1.0025,
      "num_input_tokens_seen": 62250200,
      "step": 2937
    },
    {
      "epoch": 0.35327361269764923,
      "flos": 11158594498560.0,
      "grad_norm": 4.00029276730105,
      "learning_rate": 3.000449705774275e-06,
      "loss": 1.0053,
      "num_input_tokens_seen": 62268160,
      "step": 2938
    },
    {
      "epoch": 0.35339385558828834,
      "flos": 15694429224960.0,
      "grad_norm": 3.7583421107920616,
      "learning_rate": 2.9997751218277654e-06,
      "loss": 0.9067,
      "num_input_tokens_seen": 62286035,
      "step": 2939
    },
    {
      "epoch": 0.35351409847892745,
      "flos": 17184713564160.0,
      "grad_norm": 3.6460128013398663,
      "learning_rate": 2.999100386216166e-06,
      "loss": 0.9642,
      "num_input_tokens_seen": 62304695,
      "step": 2940
    },
    {
      "epoch": 0.3536343413695665,
      "flos": 19260417576960.0,
      "grad_norm": 2.577498666224611,
      "learning_rate": 2.998425499041831e-06,
      "loss": 0.9384,
      "num_input_tokens_seen": 62324930,
      "step": 2941
    },
    {
      "epoch": 0.3537545842602056,
      "flos": 47251038658560.0,
      "grad_norm": 0.9035855290558231,
      "learning_rate": 2.997750460407142e-06,
      "loss": 0.8226,
      "num_input_tokens_seen": 62386005,
      "step": 2942
    },
    {
      "epoch": 0.35387482715084473,
      "flos": 13068049797120.0,
      "grad_norm": 5.265699618119298,
      "learning_rate": 2.997075270414501e-06,
      "loss": 0.8939,
      "num_input_tokens_seen": 62402940,
      "step": 2943
    },
    {
      "epoch": 0.3539950700414838,
      "flos": 46961624616960.0,
      "grad_norm": 0.7193931935890617,
      "learning_rate": 2.9963999291663347e-06,
      "loss": 0.7943,
      "num_input_tokens_seen": 62468440,
      "step": 2944
    },
    {
      "epoch": 0.3541153129321229,
      "flos": 14567010508800.0,
      "grad_norm": 9.145123592519727,
      "learning_rate": 2.9957244367650915e-06,
      "loss": 0.937,
      "num_input_tokens_seen": 62484405,
      "step": 2945
    },
    {
      "epoch": 0.354235555822762,
      "flos": 13885719736320.0,
      "grad_norm": 5.082118884657076,
      "learning_rate": 2.9950487933132425e-06,
      "loss": 1.0329,
      "num_input_tokens_seen": 62501540,
      "step": 2946
    },
    {
      "epoch": 0.35435579871340106,
      "flos": 14750742712320.0,
      "grad_norm": 3.8149556240635643,
      "learning_rate": 2.994372998913283e-06,
      "loss": 0.9028,
      "num_input_tokens_seen": 62519765,
      "step": 2947
    },
    {
      "epoch": 0.35447604160404017,
      "flos": 17032261201920.0,
      "grad_norm": 11.92896736734458,
      "learning_rate": 2.99369705366773e-06,
      "loss": 0.8212,
      "num_input_tokens_seen": 62539730,
      "step": 2948
    },
    {
      "epoch": 0.3545962844946792,
      "flos": 16660596572160.0,
      "grad_norm": 3.633657180865669,
      "learning_rate": 2.9930209576791244e-06,
      "loss": 1.0092,
      "num_input_tokens_seen": 62557925,
      "step": 2949
    },
    {
      "epoch": 0.35471652738531834,
      "flos": 15689922416640.0,
      "grad_norm": 3.6469017570448905,
      "learning_rate": 2.9923447110500285e-06,
      "loss": 0.8391,
      "num_input_tokens_seen": 62576390,
      "step": 2950
    },
    {
      "epoch": 0.35483677027595745,
      "flos": 19468224860160.0,
      "grad_norm": 3.8198857293338575,
      "learning_rate": 2.9916683138830295e-06,
      "loss": 0.9431,
      "num_input_tokens_seen": 62596775,
      "step": 2951
    },
    {
      "epoch": 0.3549570131665965,
      "flos": 9532268236800.0,
      "grad_norm": 5.182586181041834,
      "learning_rate": 2.9909917662807353e-06,
      "loss": 0.9954,
      "num_input_tokens_seen": 62614295,
      "step": 2952
    },
    {
      "epoch": 0.3550772560572356,
      "flos": 14830111395840.0,
      "grad_norm": 3.9551194779859737,
      "learning_rate": 2.9903150683457783e-06,
      "loss": 0.869,
      "num_input_tokens_seen": 62632560,
      "step": 2953
    },
    {
      "epoch": 0.3551974989478747,
      "flos": 14331663728640.0,
      "grad_norm": 5.72494554785274,
      "learning_rate": 2.9896382201808126e-06,
      "loss": 0.8433,
      "num_input_tokens_seen": 62649680,
      "step": 2954
    },
    {
      "epoch": 0.3553177418385138,
      "flos": 14069145354240.0,
      "grad_norm": 5.510189328134267,
      "learning_rate": 2.988961221888516e-06,
      "loss": 0.993,
      "num_input_tokens_seen": 62666075,
      "step": 2955
    },
    {
      "epoch": 0.3554379847291529,
      "flos": 10475801456640.0,
      "grad_norm": 6.16713406965153,
      "learning_rate": 2.988284073571589e-06,
      "loss": 0.9833,
      "num_input_tokens_seen": 62681880,
      "step": 2956
    },
    {
      "epoch": 0.355558227619792,
      "flos": 14540942622720.0,
      "grad_norm": 4.4019832649658115,
      "learning_rate": 2.9876067753327528e-06,
      "loss": 0.9133,
      "num_input_tokens_seen": 62699330,
      "step": 2957
    },
    {
      "epoch": 0.35567847051043106,
      "flos": 26885752627200.0,
      "grad_norm": 5.64638810288475,
      "learning_rate": 2.986929327274754e-06,
      "loss": 0.9851,
      "num_input_tokens_seen": 62719630,
      "step": 2958
    },
    {
      "epoch": 0.35579871340107017,
      "flos": 19181937991680.0,
      "grad_norm": 3.295762873369452,
      "learning_rate": 2.9862517295003617e-06,
      "loss": 0.9731,
      "num_input_tokens_seen": 62739765,
      "step": 2959
    },
    {
      "epoch": 0.3559189562917093,
      "flos": 20152397537280.0,
      "grad_norm": 4.019626152912692,
      "learning_rate": 2.9855739821123654e-06,
      "loss": 0.9256,
      "num_input_tokens_seen": 62761065,
      "step": 2960
    },
    {
      "epoch": 0.35603919918234833,
      "flos": 18262479851520.0,
      "grad_norm": 2.5755754290825097,
      "learning_rate": 2.98489608521358e-06,
      "loss": 1.0087,
      "num_input_tokens_seen": 62780725,
      "step": 2961
    },
    {
      "epoch": 0.35615944207298744,
      "flos": 16348517744640.0,
      "grad_norm": 5.284769183615939,
      "learning_rate": 2.9842180389068425e-06,
      "loss": 0.9887,
      "num_input_tokens_seen": 62797755,
      "step": 2962
    },
    {
      "epoch": 0.35627968496362655,
      "flos": 48875157504000.0,
      "grad_norm": 0.8013298946426903,
      "learning_rate": 2.98353984329501e-06,
      "loss": 0.8149,
      "num_input_tokens_seen": 62861820,
      "step": 2963
    },
    {
      "epoch": 0.3563999278542656,
      "flos": 16091180666880.0,
      "grad_norm": 12.916744371406056,
      "learning_rate": 2.982861498480965e-06,
      "loss": 0.9027,
      "num_input_tokens_seen": 62883920,
      "step": 2964
    },
    {
      "epoch": 0.3565201707449047,
      "flos": 18469336719360.0,
      "grad_norm": 5.426606238463777,
      "learning_rate": 2.9821830045676122e-06,
      "loss": 1.0166,
      "num_input_tokens_seen": 62903340,
      "step": 2965
    },
    {
      "epoch": 0.3566404136355438,
      "flos": 20283472773120.0,
      "grad_norm": 7.111391154709002,
      "learning_rate": 2.9815043616578793e-06,
      "loss": 0.9027,
      "num_input_tokens_seen": 62923855,
      "step": 2966
    },
    {
      "epoch": 0.3567606565261829,
      "flos": 27389320273920.0,
      "grad_norm": 3.7749138070681107,
      "learning_rate": 2.9808255698547145e-06,
      "loss": 0.971,
      "num_input_tokens_seen": 62946375,
      "step": 2967
    },
    {
      "epoch": 0.356880899416822,
      "flos": 15615827005440.0,
      "grad_norm": 3.9223491488467706,
      "learning_rate": 2.9801466292610913e-06,
      "loss": 0.9802,
      "num_input_tokens_seen": 62965980,
      "step": 2968
    },
    {
      "epoch": 0.35700114230746105,
      "flos": 13465598361600.0,
      "grad_norm": 27.795447721929214,
      "learning_rate": 2.979467539980003e-06,
      "loss": 0.9956,
      "num_input_tokens_seen": 62982490,
      "step": 2969
    },
    {
      "epoch": 0.35712138519810016,
      "flos": 14016917606400.0,
      "grad_norm": 10.100513754893509,
      "learning_rate": 2.978788302114468e-06,
      "loss": 0.9567,
      "num_input_tokens_seen": 62999325,
      "step": 2970
    },
    {
      "epoch": 0.35724162808873927,
      "flos": 25103049707520.0,
      "grad_norm": 5.783854459796631,
      "learning_rate": 2.9781089157675255e-06,
      "loss": 1.0099,
      "num_input_tokens_seen": 63017505,
      "step": 2971
    },
    {
      "epoch": 0.3573618709793783,
      "flos": 18185625169920.0,
      "grad_norm": 3.246247546136668,
      "learning_rate": 2.977429381042238e-06,
      "loss": 1.0737,
      "num_input_tokens_seen": 63037900,
      "step": 2972
    },
    {
      "epoch": 0.35748211387001744,
      "flos": 20755362017280.0,
      "grad_norm": 7.005437918567846,
      "learning_rate": 2.9767496980416913e-06,
      "loss": 1.0852,
      "num_input_tokens_seen": 63056915,
      "step": 2973
    },
    {
      "epoch": 0.35760235676065655,
      "flos": 9847075676160.0,
      "grad_norm": 6.6736354961957565,
      "learning_rate": 2.9760698668689914e-06,
      "loss": 1.0017,
      "num_input_tokens_seen": 63072860,
      "step": 2974
    },
    {
      "epoch": 0.3577225996512956,
      "flos": 31453909585920.0,
      "grad_norm": 3.6002460978420427,
      "learning_rate": 2.975389887627269e-06,
      "loss": 0.9098,
      "num_input_tokens_seen": 63095180,
      "step": 2975
    },
    {
      "epoch": 0.3578428425419347,
      "flos": 12075691929600.0,
      "grad_norm": 11.602734973371781,
      "learning_rate": 2.9747097604196764e-06,
      "loss": 1.0915,
      "num_input_tokens_seen": 63111545,
      "step": 2976
    },
    {
      "epoch": 0.3579630854325738,
      "flos": 51336238632960.0,
      "grad_norm": 0.7055929106383769,
      "learning_rate": 2.9740294853493875e-06,
      "loss": 0.7914,
      "num_input_tokens_seen": 63182825,
      "step": 2977
    },
    {
      "epoch": 0.3580833283232129,
      "flos": 17819141836800.0,
      "grad_norm": 3.863418129250115,
      "learning_rate": 2.9733490625196008e-06,
      "loss": 0.8677,
      "num_input_tokens_seen": 63202405,
      "step": 2978
    },
    {
      "epoch": 0.358203571213852,
      "flos": 9846186577920.0,
      "grad_norm": 6.746342833867849,
      "learning_rate": 2.9726684920335353e-06,
      "loss": 0.9265,
      "num_input_tokens_seen": 63219990,
      "step": 2979
    },
    {
      "epoch": 0.35832381410449105,
      "flos": 14409438167040.0,
      "grad_norm": 5.281796940578915,
      "learning_rate": 2.971987773994432e-06,
      "loss": 1.0138,
      "num_input_tokens_seen": 63235895,
      "step": 2980
    },
    {
      "epoch": 0.35844405699513016,
      "flos": 12024108011520.0,
      "grad_norm": 7.9146681287426555,
      "learning_rate": 2.9713069085055566e-06,
      "loss": 1.0195,
      "num_input_tokens_seen": 63253925,
      "step": 2981
    },
    {
      "epoch": 0.35856429988576927,
      "flos": 16503361474560.0,
      "grad_norm": 6.173823490883265,
      "learning_rate": 2.9706258956701958e-06,
      "loss": 0.9874,
      "num_input_tokens_seen": 63273635,
      "step": 2982
    },
    {
      "epoch": 0.3586845427764083,
      "flos": 16372378214400.0,
      "grad_norm": 17.981759896207148,
      "learning_rate": 2.9699447355916575e-06,
      "loss": 0.9511,
      "num_input_tokens_seen": 63292165,
      "step": 2983
    },
    {
      "epoch": 0.35880478566704743,
      "flos": 14199760711680.0,
      "grad_norm": 4.493477651012241,
      "learning_rate": 2.969263428373275e-06,
      "loss": 0.9333,
      "num_input_tokens_seen": 63310235,
      "step": 2984
    },
    {
      "epoch": 0.35892502855768654,
      "flos": 9663803351040.0,
      "grad_norm": 5.174883840243157,
      "learning_rate": 2.9685819741184007e-06,
      "loss": 0.9697,
      "num_input_tokens_seen": 63328395,
      "step": 2985
    },
    {
      "epoch": 0.3590452714483256,
      "flos": 12836811264000.0,
      "grad_norm": 5.326939604780452,
      "learning_rate": 2.967900372930411e-06,
      "loss": 0.8702,
      "num_input_tokens_seen": 63346625,
      "step": 2986
    },
    {
      "epoch": 0.3591655143389647,
      "flos": 12574691450880.0,
      "grad_norm": 7.932651751221173,
      "learning_rate": 2.9672186249127046e-06,
      "loss": 0.9822,
      "num_input_tokens_seen": 63365810,
      "step": 2987
    },
    {
      "epoch": 0.3592857572296038,
      "flos": 17946262118400.0,
      "grad_norm": 5.264018280577296,
      "learning_rate": 2.9665367301687014e-06,
      "loss": 0.9809,
      "num_input_tokens_seen": 63383775,
      "step": 2988
    },
    {
      "epoch": 0.3594060001202429,
      "flos": 20936059023360.0,
      "grad_norm": 16.599938339670402,
      "learning_rate": 2.965854688801845e-06,
      "loss": 0.9599,
      "num_input_tokens_seen": 63405555,
      "step": 2989
    },
    {
      "epoch": 0.359526243010882,
      "flos": 12073729781760.0,
      "grad_norm": 7.538614431714148,
      "learning_rate": 2.9651725009156005e-06,
      "loss": 0.9484,
      "num_input_tokens_seen": 63423020,
      "step": 2990
    },
    {
      "epoch": 0.3596464859015211,
      "flos": 16322909736960.0,
      "grad_norm": 4.141346533905561,
      "learning_rate": 2.964490166613454e-06,
      "loss": 0.9372,
      "num_input_tokens_seen": 63442665,
      "step": 2991
    },
    {
      "epoch": 0.35976672879216015,
      "flos": 39162836090880.0,
      "grad_norm": 0.8520775383625533,
      "learning_rate": 2.963807685998917e-06,
      "loss": 0.7938,
      "num_input_tokens_seen": 63498250,
      "step": 2992
    },
    {
      "epoch": 0.35988697168279926,
      "flos": 30821167534080.0,
      "grad_norm": 4.254099495581944,
      "learning_rate": 2.9631250591755196e-06,
      "loss": 0.9804,
      "num_input_tokens_seen": 63520685,
      "step": 2993
    },
    {
      "epoch": 0.36000721457343837,
      "flos": 25580028272640.0,
      "grad_norm": 6.775294065955793,
      "learning_rate": 2.962442286246817e-06,
      "loss": 0.7804,
      "num_input_tokens_seen": 63543235,
      "step": 2994
    },
    {
      "epoch": 0.3601274574640774,
      "flos": 12963287715840.0,
      "grad_norm": 3.3387118021937607,
      "learning_rate": 2.9617593673163853e-06,
      "loss": 0.9008,
      "num_input_tokens_seen": 63561775,
      "step": 2995
    },
    {
      "epoch": 0.36024770035471654,
      "flos": 9401346293760.0,
      "grad_norm": 6.691015727622014,
      "learning_rate": 2.9610763024878216e-06,
      "loss": 0.9611,
      "num_input_tokens_seen": 63577000,
      "step": 2996
    },
    {
      "epoch": 0.3603679432453556,
      "flos": 14384106086400.0,
      "grad_norm": 6.532111444198851,
      "learning_rate": 2.960393091864747e-06,
      "loss": 1.1093,
      "num_input_tokens_seen": 63595100,
      "step": 2997
    },
    {
      "epoch": 0.3604881861359947,
      "flos": 15953115279360.0,
      "grad_norm": 6.834019541207034,
      "learning_rate": 2.959709735550804e-06,
      "loss": 0.9386,
      "num_input_tokens_seen": 63614415,
      "step": 2998
    },
    {
      "epoch": 0.3606084290266338,
      "flos": 15668483973120.0,
      "grad_norm": 17.091801227180625,
      "learning_rate": 2.9590262336496575e-06,
      "loss": 0.9516,
      "num_input_tokens_seen": 63633865,
      "step": 2999
    },
    {
      "epoch": 0.36072867191727287,
      "flos": 11053617807360.0,
      "grad_norm": 10.654230815846995,
      "learning_rate": 2.9583425862649936e-06,
      "loss": 1.0461,
      "num_input_tokens_seen": 63651720,
      "step": 3000
    },
    {
      "epoch": 0.360848914807912,
      "flos": 13959232634880.0,
      "grad_norm": 10.529820195154576,
      "learning_rate": 2.9576587935005215e-06,
      "loss": 0.9401,
      "num_input_tokens_seen": 63669520,
      "step": 3001
    },
    {
      "epoch": 0.3609691576985511,
      "flos": 13385340579840.0,
      "grad_norm": 8.162972321799508,
      "learning_rate": 2.9569748554599713e-06,
      "loss": 0.9164,
      "num_input_tokens_seen": 63684850,
      "step": 3002
    },
    {
      "epoch": 0.36108940058919015,
      "flos": 30163246694400.0,
      "grad_norm": 3.950883433382312,
      "learning_rate": 2.956290772247097e-06,
      "loss": 0.9268,
      "num_input_tokens_seen": 63703245,
      "step": 3003
    },
    {
      "epoch": 0.36120964347982926,
      "flos": 16481217884160.0,
      "grad_norm": 9.148309903839053,
      "learning_rate": 2.9556065439656724e-06,
      "loss": 0.9365,
      "num_input_tokens_seen": 63722015,
      "step": 3004
    },
    {
      "epoch": 0.36132988637046837,
      "flos": 12835922165760.0,
      "grad_norm": 5.838235303161232,
      "learning_rate": 2.9549221707194952e-06,
      "loss": 1.0118,
      "num_input_tokens_seen": 63740585,
      "step": 3005
    },
    {
      "epoch": 0.3614501292611074,
      "flos": 19806463549440.0,
      "grad_norm": 6.126184154406491,
      "learning_rate": 2.954237652612384e-06,
      "loss": 0.9341,
      "num_input_tokens_seen": 63759355,
      "step": 3006
    },
    {
      "epoch": 0.36157037215174653,
      "flos": 16085386199040.0,
      "grad_norm": 6.638956304586511,
      "learning_rate": 2.9535529897481796e-06,
      "loss": 1.0416,
      "num_input_tokens_seen": 63776620,
      "step": 3007
    },
    {
      "epoch": 0.36169061504238564,
      "flos": 8874010152960.0,
      "grad_norm": 13.380340365655174,
      "learning_rate": 2.9528681822307446e-06,
      "loss": 0.9587,
      "num_input_tokens_seen": 63793190,
      "step": 3008
    },
    {
      "epoch": 0.3618108579330247,
      "flos": 18995814420480.0,
      "grad_norm": 5.901130839451551,
      "learning_rate": 2.952183230163964e-06,
      "loss": 1.0164,
      "num_input_tokens_seen": 63812485,
      "step": 3009
    },
    {
      "epoch": 0.3619311008236638,
      "flos": 16216216166400.0,
      "grad_norm": 5.327099832930192,
      "learning_rate": 2.9514981336517448e-06,
      "loss": 0.9358,
      "num_input_tokens_seen": 63831975,
      "step": 3010
    },
    {
      "epoch": 0.36205134371430286,
      "flos": 18446150737920.0,
      "grad_norm": 3.8942845303699887,
      "learning_rate": 2.950812892798015e-06,
      "loss": 1.0017,
      "num_input_tokens_seen": 63852590,
      "step": 3011
    },
    {
      "epoch": 0.362171586604942,
      "flos": 18604397568000.0,
      "grad_norm": 4.465871243798518,
      "learning_rate": 2.9501275077067256e-06,
      "loss": 1.0659,
      "num_input_tokens_seen": 63872930,
      "step": 3012
    },
    {
      "epoch": 0.3622918294955811,
      "flos": 19994886512640.0,
      "grad_norm": 5.937913097266041,
      "learning_rate": 2.949441978481848e-06,
      "loss": 1.0824,
      "num_input_tokens_seen": 63893550,
      "step": 3013
    },
    {
      "epoch": 0.36241207238622014,
      "flos": 14068440207360.0,
      "grad_norm": 4.724550635551041,
      "learning_rate": 2.9487563052273778e-06,
      "loss": 0.9954,
      "num_input_tokens_seen": 63910030,
      "step": 3014
    },
    {
      "epoch": 0.36253231527685925,
      "flos": 15196196167680.0,
      "grad_norm": 4.818746303057211,
      "learning_rate": 2.94807048804733e-06,
      "loss": 1.0536,
      "num_input_tokens_seen": 63929370,
      "step": 3015
    },
    {
      "epoch": 0.36265255816749836,
      "flos": 13014718341120.0,
      "grad_norm": 4.683510061348324,
      "learning_rate": 2.9473845270457434e-06,
      "loss": 1.0891,
      "num_input_tokens_seen": 63945905,
      "step": 3016
    },
    {
      "epoch": 0.3627728010581374,
      "flos": 13228994580480.0,
      "grad_norm": 5.818951707533701,
      "learning_rate": 2.946698422326677e-06,
      "loss": 0.8762,
      "num_input_tokens_seen": 63963085,
      "step": 3017
    },
    {
      "epoch": 0.36289304394877653,
      "flos": 19599300096000.0,
      "grad_norm": 7.370956088075939,
      "learning_rate": 2.946012173994213e-06,
      "loss": 0.9889,
      "num_input_tokens_seen": 63982590,
      "step": 3018
    },
    {
      "epoch": 0.36301328683941564,
      "flos": 24637016248320.0,
      "grad_norm": 3.0040864133939595,
      "learning_rate": 2.945325782152454e-06,
      "loss": 0.881,
      "num_input_tokens_seen": 64005345,
      "step": 3019
    },
    {
      "epoch": 0.3631335297300547,
      "flos": 13492126126080.0,
      "grad_norm": 3.9697108269598953,
      "learning_rate": 2.9446392469055257e-06,
      "loss": 0.9912,
      "num_input_tokens_seen": 64023100,
      "step": 3020
    },
    {
      "epoch": 0.3632537726206938,
      "flos": 13859222630400.0,
      "grad_norm": 6.40172435657369,
      "learning_rate": 2.9439525683575745e-06,
      "loss": 0.9939,
      "num_input_tokens_seen": 64041740,
      "step": 3021
    },
    {
      "epoch": 0.3633740155113329,
      "flos": 14960052264960.0,
      "grad_norm": 4.43107182369163,
      "learning_rate": 2.9432657466127694e-06,
      "loss": 0.9424,
      "num_input_tokens_seen": 64061030,
      "step": 3022
    },
    {
      "epoch": 0.36349425840197197,
      "flos": 14406249676800.0,
      "grad_norm": 5.021090088079223,
      "learning_rate": 2.9425787817753007e-06,
      "loss": 0.9621,
      "num_input_tokens_seen": 64079410,
      "step": 3023
    },
    {
      "epoch": 0.3636145012926111,
      "flos": 21175452733440.0,
      "grad_norm": 3.9986383981683864,
      "learning_rate": 2.94189167394938e-06,
      "loss": 0.9029,
      "num_input_tokens_seen": 64101565,
      "step": 3024
    },
    {
      "epoch": 0.3637347441832502,
      "flos": 15220884418560.0,
      "grad_norm": 3.673690931241845,
      "learning_rate": 2.941204423239241e-06,
      "loss": 1.0054,
      "num_input_tokens_seen": 64120160,
      "step": 3025
    },
    {
      "epoch": 0.36385498707388925,
      "flos": 21044009594880.0,
      "grad_norm": 4.511675206093985,
      "learning_rate": 2.9405170297491395e-06,
      "loss": 0.9589,
      "num_input_tokens_seen": 64139875,
      "step": 3026
    },
    {
      "epoch": 0.36397522996452836,
      "flos": 15799497891840.0,
      "grad_norm": 4.559297247492339,
      "learning_rate": 2.939829493583353e-06,
      "loss": 0.995,
      "num_input_tokens_seen": 64156240,
      "step": 3027
    },
    {
      "epoch": 0.3640954728551674,
      "flos": 15274767728640.0,
      "grad_norm": 6.191036870389126,
      "learning_rate": 2.939141814846179e-06,
      "loss": 1.0097,
      "num_input_tokens_seen": 64173375,
      "step": 3028
    },
    {
      "epoch": 0.3642157157458065,
      "flos": 12548470272000.0,
      "grad_norm": 6.998701247576727,
      "learning_rate": 2.938453993641938e-06,
      "loss": 1.0184,
      "num_input_tokens_seen": 64191470,
      "step": 3029
    },
    {
      "epoch": 0.36433595863644563,
      "flos": 12495721328640.0,
      "grad_norm": 6.095958759358595,
      "learning_rate": 2.937766030074973e-06,
      "loss": 0.8929,
      "num_input_tokens_seen": 64208445,
      "step": 3030
    },
    {
      "epoch": 0.3644562015270847,
      "flos": 19102998528000.0,
      "grad_norm": 4.0430281924978,
      "learning_rate": 2.937077924249646e-06,
      "loss": 1.0094,
      "num_input_tokens_seen": 64230755,
      "step": 3031
    },
    {
      "epoch": 0.3645764444177238,
      "flos": 10083219578880.0,
      "grad_norm": 5.498348161427436,
      "learning_rate": 2.9363896762703443e-06,
      "loss": 0.9372,
      "num_input_tokens_seen": 64247540,
      "step": 3032
    },
    {
      "epoch": 0.3646966873083629,
      "flos": 14671803248640.0,
      "grad_norm": 15.232053108315085,
      "learning_rate": 2.9357012862414725e-06,
      "loss": 1.0342,
      "num_input_tokens_seen": 64266620,
      "step": 3033
    },
    {
      "epoch": 0.36481693019900197,
      "flos": 19785362350080.0,
      "grad_norm": 12.833684016468938,
      "learning_rate": 2.9350127542674593e-06,
      "loss": 0.913,
      "num_input_tokens_seen": 64288550,
      "step": 3034
    },
    {
      "epoch": 0.3649371730896411,
      "flos": 13984687349760.0,
      "grad_norm": 4.37627871763542,
      "learning_rate": 2.934324080452755e-06,
      "loss": 0.9606,
      "num_input_tokens_seen": 64306060,
      "step": 3035
    },
    {
      "epoch": 0.3650574159802802,
      "flos": 17605724037120.0,
      "grad_norm": 3.009015365454862,
      "learning_rate": 2.9336352649018307e-06,
      "loss": 0.9772,
      "num_input_tokens_seen": 64325850,
      "step": 3036
    },
    {
      "epoch": 0.36517765887091924,
      "flos": 23429615677440.0,
      "grad_norm": 3.1900167066561136,
      "learning_rate": 2.9329463077191783e-06,
      "loss": 0.901,
      "num_input_tokens_seen": 64348945,
      "step": 3037
    },
    {
      "epoch": 0.36529790176155835,
      "flos": 14278577541120.0,
      "grad_norm": 4.437473780594855,
      "learning_rate": 2.9322572090093135e-06,
      "loss": 0.8369,
      "num_input_tokens_seen": 64367370,
      "step": 3038
    },
    {
      "epoch": 0.36541814465219746,
      "flos": 12495751987200.0,
      "grad_norm": 5.980063376712821,
      "learning_rate": 2.9315679688767713e-06,
      "loss": 0.9391,
      "num_input_tokens_seen": 64385100,
      "step": 3039
    },
    {
      "epoch": 0.3655383875428365,
      "flos": 16114121379840.0,
      "grad_norm": 4.843770673153968,
      "learning_rate": 2.9308785874261085e-06,
      "loss": 0.855,
      "num_input_tokens_seen": 64405010,
      "step": 3040
    },
    {
      "epoch": 0.36565863043347563,
      "flos": 15615949639680.0,
      "grad_norm": 8.433194340501585,
      "learning_rate": 2.9301890647619045e-06,
      "loss": 1.0015,
      "num_input_tokens_seen": 64424025,
      "step": 3041
    },
    {
      "epoch": 0.36577887332411474,
      "flos": 17661722787840.0,
      "grad_norm": 3.5932907474473756,
      "learning_rate": 2.929499400988759e-06,
      "loss": 0.9811,
      "num_input_tokens_seen": 64444905,
      "step": 3042
    },
    {
      "epoch": 0.3658991162147538,
      "flos": 20152274903040.0,
      "grad_norm": 4.156944274279232,
      "learning_rate": 2.9288095962112927e-06,
      "loss": 0.8422,
      "num_input_tokens_seen": 64465330,
      "step": 3043
    },
    {
      "epoch": 0.3660193591053929,
      "flos": 12600667361280.0,
      "grad_norm": 3.8660982081317377,
      "learning_rate": 2.9281196505341503e-06,
      "loss": 1.0413,
      "num_input_tokens_seen": 64482220,
      "step": 3044
    },
    {
      "epoch": 0.36613960199603196,
      "flos": 7251730821120.0,
      "grad_norm": 4.691571717913536,
      "learning_rate": 2.9274295640619946e-06,
      "loss": 0.9826,
      "num_input_tokens_seen": 64499070,
      "step": 3045
    },
    {
      "epoch": 0.36625984488667107,
      "flos": 14016488386560.0,
      "grad_norm": 4.523543967672451,
      "learning_rate": 2.9267393368995103e-06,
      "loss": 0.9808,
      "num_input_tokens_seen": 64518020,
      "step": 3046
    },
    {
      "epoch": 0.3663800877773102,
      "flos": 12520440238080.0,
      "grad_norm": 6.5108689621535145,
      "learning_rate": 2.926048969151407e-06,
      "loss": 0.9245,
      "num_input_tokens_seen": 64535025,
      "step": 3047
    },
    {
      "epoch": 0.36650033066794924,
      "flos": 14410419240960.0,
      "grad_norm": 7.61538765795335,
      "learning_rate": 2.92535846092241e-06,
      "loss": 0.8704,
      "num_input_tokens_seen": 64553760,
      "step": 3048
    },
    {
      "epoch": 0.36662057355858835,
      "flos": 17581403688960.0,
      "grad_norm": 4.454744385858179,
      "learning_rate": 2.9246678123172704e-06,
      "loss": 1.0153,
      "num_input_tokens_seen": 64573570,
      "step": 3049
    },
    {
      "epoch": 0.36674081644922746,
      "flos": 8719534325760.0,
      "grad_norm": 17.12316171845399,
      "learning_rate": 2.9239770234407596e-06,
      "loss": 0.9181,
      "num_input_tokens_seen": 64591595,
      "step": 3050
    },
    {
      "epoch": 0.3668610593398665,
      "flos": 14986978590720.0,
      "grad_norm": 4.334582857787516,
      "learning_rate": 2.9232860943976686e-06,
      "loss": 0.8729,
      "num_input_tokens_seen": 64612050,
      "step": 3051
    },
    {
      "epoch": 0.3669813022305056,
      "flos": 19048655339520.0,
      "grad_norm": 25.980134857743955,
      "learning_rate": 2.9225950252928115e-06,
      "loss": 1.0348,
      "num_input_tokens_seen": 64632620,
      "step": 3052
    },
    {
      "epoch": 0.36710154512114473,
      "flos": 13545059020800.0,
      "grad_norm": 4.313219977829608,
      "learning_rate": 2.9219038162310217e-06,
      "loss": 0.9965,
      "num_input_tokens_seen": 64650540,
      "step": 3053
    },
    {
      "epoch": 0.3672217880117838,
      "flos": 14775492280320.0,
      "grad_norm": 6.604292265683463,
      "learning_rate": 2.921212467317157e-06,
      "loss": 1.0152,
      "num_input_tokens_seen": 64669705,
      "step": 3054
    },
    {
      "epoch": 0.3673420309024229,
      "flos": 9585783644160.0,
      "grad_norm": 6.133823500963065,
      "learning_rate": 2.920520978656093e-06,
      "loss": 0.993,
      "num_input_tokens_seen": 64686390,
      "step": 3055
    },
    {
      "epoch": 0.367462273793062,
      "flos": 20650875863040.0,
      "grad_norm": 3.789050670487333,
      "learning_rate": 2.919829350352729e-06,
      "loss": 0.9576,
      "num_input_tokens_seen": 64707715,
      "step": 3056
    },
    {
      "epoch": 0.36758251668370107,
      "flos": 44841212313600.0,
      "grad_norm": 0.7632727006572044,
      "learning_rate": 2.919137582511983e-06,
      "loss": 0.8098,
      "num_input_tokens_seen": 64763875,
      "step": 3057
    },
    {
      "epoch": 0.3677027595743402,
      "flos": 8955739545600.0,
      "grad_norm": 90.17200360630233,
      "learning_rate": 2.918445675238797e-06,
      "loss": 0.8275,
      "num_input_tokens_seen": 64780520,
      "step": 3058
    },
    {
      "epoch": 0.36782300246497923,
      "flos": 17819080519680.0,
      "grad_norm": 19.028983245805346,
      "learning_rate": 2.917753628638132e-06,
      "loss": 0.877,
      "num_input_tokens_seen": 64800545,
      "step": 3059
    },
    {
      "epoch": 0.36794324535561834,
      "flos": 12337443840000.0,
      "grad_norm": 5.487288089909717,
      "learning_rate": 2.9170614428149716e-06,
      "loss": 0.8839,
      "num_input_tokens_seen": 64818600,
      "step": 3060
    },
    {
      "epoch": 0.36806348824625745,
      "flos": 17129787863040.0,
      "grad_norm": 7.477322690258714,
      "learning_rate": 2.9163691178743195e-06,
      "loss": 1.0525,
      "num_input_tokens_seen": 64836970,
      "step": 3061
    },
    {
      "epoch": 0.3681837311368965,
      "flos": 14567194460160.0,
      "grad_norm": 5.292713432206949,
      "learning_rate": 2.9156766539212006e-06,
      "loss": 0.9749,
      "num_input_tokens_seen": 64854335,
      "step": 3062
    },
    {
      "epoch": 0.3683039740275356,
      "flos": 15245664645120.0,
      "grad_norm": 7.910688880601602,
      "learning_rate": 2.9149840510606614e-06,
      "loss": 0.9066,
      "num_input_tokens_seen": 64872710,
      "step": 3063
    },
    {
      "epoch": 0.36842421691817473,
      "flos": 50402646896640.0,
      "grad_norm": 1.050714767224773,
      "learning_rate": 2.914291309397769e-06,
      "loss": 0.8766,
      "num_input_tokens_seen": 64929900,
      "step": 3064
    },
    {
      "epoch": 0.3685444598088138,
      "flos": 16586347868160.0,
      "grad_norm": 9.681571475380954,
      "learning_rate": 2.9135984290376117e-06,
      "loss": 0.9776,
      "num_input_tokens_seen": 64948485,
      "step": 3065
    },
    {
      "epoch": 0.3686647026994529,
      "flos": 16397924904960.0,
      "grad_norm": 5.533864066320802,
      "learning_rate": 2.9129054100853e-06,
      "loss": 1.0245,
      "num_input_tokens_seen": 64967045,
      "step": 3066
    },
    {
      "epoch": 0.368784945590092,
      "flos": 17871829463040.0,
      "grad_norm": 6.313384343511528,
      "learning_rate": 2.912212252645963e-06,
      "loss": 0.9422,
      "num_input_tokens_seen": 64989350,
      "step": 3067
    },
    {
      "epoch": 0.36890518848073106,
      "flos": 13072648581120.0,
      "grad_norm": 7.991392273181638,
      "learning_rate": 2.9115189568247523e-06,
      "loss": 0.9516,
      "num_input_tokens_seen": 65006630,
      "step": 3068
    },
    {
      "epoch": 0.36902543137137017,
      "flos": 11577888092160.0,
      "grad_norm": 11.529906817520388,
      "learning_rate": 2.910825522726841e-06,
      "loss": 1.1044,
      "num_input_tokens_seen": 65023875,
      "step": 3069
    },
    {
      "epoch": 0.3691456742620093,
      "flos": 8640686837760.0,
      "grad_norm": 17.811667332511647,
      "learning_rate": 2.9101319504574215e-06,
      "loss": 0.9562,
      "num_input_tokens_seen": 65040035,
      "step": 3070
    },
    {
      "epoch": 0.36926591715264834,
      "flos": 12601403166720.0,
      "grad_norm": 5.545055854041371,
      "learning_rate": 2.909438240121709e-06,
      "loss": 0.9577,
      "num_input_tokens_seen": 65060030,
      "step": 3071
    },
    {
      "epoch": 0.36938616004328745,
      "flos": 20622876487680.0,
      "grad_norm": 5.303858831342291,
      "learning_rate": 2.908744391824939e-06,
      "loss": 0.8907,
      "num_input_tokens_seen": 65080770,
      "step": 3072
    },
    {
      "epoch": 0.36950640293392656,
      "flos": 20807865692160.0,
      "grad_norm": 4.728684029619397,
      "learning_rate": 2.908050405672367e-06,
      "loss": 0.97,
      "num_input_tokens_seen": 65100035,
      "step": 3073
    },
    {
      "epoch": 0.3696266458245656,
      "flos": 17661845422080.0,
      "grad_norm": 5.727930589905349,
      "learning_rate": 2.9073562817692703e-06,
      "loss": 0.9828,
      "num_input_tokens_seen": 65118440,
      "step": 3074
    },
    {
      "epoch": 0.3697468887152047,
      "flos": 42862904217600.0,
      "grad_norm": 0.8340517325079675,
      "learning_rate": 2.9066620202209468e-06,
      "loss": 0.8046,
      "num_input_tokens_seen": 65180650,
      "step": 3075
    },
    {
      "epoch": 0.3698671316058438,
      "flos": 18602680688640.0,
      "grad_norm": 11.045922568781107,
      "learning_rate": 2.905967621132716e-06,
      "loss": 0.9813,
      "num_input_tokens_seen": 65197980,
      "step": 3076
    },
    {
      "epoch": 0.3699873744964829,
      "flos": 17503322664960.0,
      "grad_norm": 15.89278442953997,
      "learning_rate": 2.9052730846099172e-06,
      "loss": 0.9374,
      "num_input_tokens_seen": 65219045,
      "step": 3077
    },
    {
      "epoch": 0.370107617387122,
      "flos": 46454868480000.0,
      "grad_norm": 0.8881031270457804,
      "learning_rate": 2.9045784107579123e-06,
      "loss": 0.8432,
      "num_input_tokens_seen": 65278870,
      "step": 3078
    },
    {
      "epoch": 0.37022786027776106,
      "flos": 11289945661440.0,
      "grad_norm": 7.236949619060392,
      "learning_rate": 2.9038835996820807e-06,
      "loss": 0.8686,
      "num_input_tokens_seen": 65296200,
      "step": 3079
    },
    {
      "epoch": 0.37034810316840017,
      "flos": 13147203870720.0,
      "grad_norm": 8.539435962212865,
      "learning_rate": 2.903188651487826e-06,
      "loss": 0.9868,
      "num_input_tokens_seen": 65314475,
      "step": 3080
    },
    {
      "epoch": 0.3704683460590393,
      "flos": 12626796564480.0,
      "grad_norm": 7.009101400422614,
      "learning_rate": 2.902493566280571e-06,
      "loss": 1.0566,
      "num_input_tokens_seen": 65332300,
      "step": 3081
    },
    {
      "epoch": 0.37058858894967833,
      "flos": 9976035471360.0,
      "grad_norm": 6.338825889197814,
      "learning_rate": 2.9017983441657595e-06,
      "loss": 1.005,
      "num_input_tokens_seen": 65349350,
      "step": 3082
    },
    {
      "epoch": 0.37070883184031744,
      "flos": 9847351603200.0,
      "grad_norm": 17.12938684117699,
      "learning_rate": 2.9011029852488564e-06,
      "loss": 0.9508,
      "num_input_tokens_seen": 65366305,
      "step": 3083
    },
    {
      "epoch": 0.37082907473095655,
      "flos": 37418564567040.0,
      "grad_norm": 1.0755104206064527,
      "learning_rate": 2.9004074896353465e-06,
      "loss": 0.8726,
      "num_input_tokens_seen": 65420025,
      "step": 3084
    },
    {
      "epoch": 0.3709493176215956,
      "flos": 11316105523200.0,
      "grad_norm": 9.796948981728237,
      "learning_rate": 2.8997118574307362e-06,
      "loss": 1.0034,
      "num_input_tokens_seen": 65436700,
      "step": 3085
    },
    {
      "epoch": 0.3710695605122347,
      "flos": 14881971240960.0,
      "grad_norm": 17.352794481534072,
      "learning_rate": 2.899016088740553e-06,
      "loss": 0.9421,
      "num_input_tokens_seen": 65454530,
      "step": 3086
    },
    {
      "epoch": 0.37118980340287383,
      "flos": 10135569960960.0,
      "grad_norm": 7.818417680662557,
      "learning_rate": 2.898320183670344e-06,
      "loss": 0.9832,
      "num_input_tokens_seen": 65471665,
      "step": 3087
    },
    {
      "epoch": 0.3713100462935129,
      "flos": 18421861048320.0,
      "grad_norm": 4.821285403109052,
      "learning_rate": 2.8976241423256767e-06,
      "loss": 1.0784,
      "num_input_tokens_seen": 65491480,
      "step": 3088
    },
    {
      "epoch": 0.371430289184152,
      "flos": 21751889448960.0,
      "grad_norm": 6.794082804707826,
      "learning_rate": 2.896927964812142e-06,
      "loss": 0.8777,
      "num_input_tokens_seen": 65511765,
      "step": 3089
    },
    {
      "epoch": 0.37155053207479105,
      "flos": 11132158709760.0,
      "grad_norm": 8.771896865648959,
      "learning_rate": 2.8962316512353465e-06,
      "loss": 0.9268,
      "num_input_tokens_seen": 65529030,
      "step": 3090
    },
    {
      "epoch": 0.37167077496543016,
      "flos": 16638544957440.0,
      "grad_norm": 5.206126220990213,
      "learning_rate": 2.8955352017009233e-06,
      "loss": 0.9435,
      "num_input_tokens_seen": 65547995,
      "step": 3091
    },
    {
      "epoch": 0.3717910178560693,
      "flos": 15692926955520.0,
      "grad_norm": 4.5684111462970725,
      "learning_rate": 2.8948386163145212e-06,
      "loss": 0.9659,
      "num_input_tokens_seen": 65566925,
      "step": 3092
    },
    {
      "epoch": 0.3719112607467083,
      "flos": 19179301355520.0,
      "grad_norm": 14.273006672822019,
      "learning_rate": 2.8941418951818135e-06,
      "loss": 0.9865,
      "num_input_tokens_seen": 65586205,
      "step": 3093
    },
    {
      "epoch": 0.37203150363734744,
      "flos": 8562329886720.0,
      "grad_norm": 16.349588726132282,
      "learning_rate": 2.8934450384084903e-06,
      "loss": 0.9041,
      "num_input_tokens_seen": 65603440,
      "step": 3094
    },
    {
      "epoch": 0.37215174652798655,
      "flos": 16849724682240.0,
      "grad_norm": 41.74227888699523,
      "learning_rate": 2.8927480461002653e-06,
      "loss": 0.9048,
      "num_input_tokens_seen": 65623130,
      "step": 3095
    },
    {
      "epoch": 0.3722719894186256,
      "flos": 12673137868800.0,
      "grad_norm": 5.290185896378057,
      "learning_rate": 2.892050918362872e-06,
      "loss": 1.0295,
      "num_input_tokens_seen": 65637905,
      "step": 3096
    },
    {
      "epoch": 0.3723922323092647,
      "flos": 44680175554560.0,
      "grad_norm": 0.9627718281536092,
      "learning_rate": 2.8913536553020626e-06,
      "loss": 0.8031,
      "num_input_tokens_seen": 65691680,
      "step": 3097
    },
    {
      "epoch": 0.3725124751999038,
      "flos": 16376793047040.0,
      "grad_norm": 35.263755765702626,
      "learning_rate": 2.8906562570236137e-06,
      "loss": 1.0486,
      "num_input_tokens_seen": 65709310,
      "step": 3098
    },
    {
      "epoch": 0.3726327180905429,
      "flos": 14853358694400.0,
      "grad_norm": 10.440040507621665,
      "learning_rate": 2.889958723633318e-06,
      "loss": 0.9541,
      "num_input_tokens_seen": 65727970,
      "step": 3099
    },
    {
      "epoch": 0.372752960981182,
      "flos": 21804546416640.0,
      "grad_norm": 5.123220925786721,
      "learning_rate": 2.889261055236992e-06,
      "loss": 0.9302,
      "num_input_tokens_seen": 65749905,
      "step": 3100
    },
    {
      "epoch": 0.3728732038718211,
      "flos": 17869131509760.0,
      "grad_norm": 55.942501522210286,
      "learning_rate": 2.8885632519404704e-06,
      "loss": 1.024,
      "num_input_tokens_seen": 65769895,
      "step": 3101
    },
    {
      "epoch": 0.37299344676246016,
      "flos": 18002690088960.0,
      "grad_norm": 5.493219704579302,
      "learning_rate": 2.8878653138496107e-06,
      "loss": 0.952,
      "num_input_tokens_seen": 65790110,
      "step": 3102
    },
    {
      "epoch": 0.37311368965309927,
      "flos": 16953750958080.0,
      "grad_norm": 9.990227692751919,
      "learning_rate": 2.8871672410702878e-06,
      "loss": 0.9446,
      "num_input_tokens_seen": 65807190,
      "step": 3103
    },
    {
      "epoch": 0.3732339325437384,
      "flos": 18369020129280.0,
      "grad_norm": 13.396689823153086,
      "learning_rate": 2.8864690337084008e-06,
      "loss": 1.0127,
      "num_input_tokens_seen": 65826185,
      "step": 3104
    },
    {
      "epoch": 0.37335417543437743,
      "flos": 18653314191360.0,
      "grad_norm": 5.776210918876292,
      "learning_rate": 2.885770691869866e-06,
      "loss": 0.9783,
      "num_input_tokens_seen": 65846785,
      "step": 3105
    },
    {
      "epoch": 0.37347441832501654,
      "flos": 17083937095680.0,
      "grad_norm": 18.6126028699831,
      "learning_rate": 2.8850722156606207e-06,
      "loss": 0.9286,
      "num_input_tokens_seen": 65864895,
      "step": 3106
    },
    {
      "epoch": 0.3735946612156556,
      "flos": 13986680156160.0,
      "grad_norm": 7.807920777106557,
      "learning_rate": 2.8843736051866252e-06,
      "loss": 0.8575,
      "num_input_tokens_seen": 65883540,
      "step": 3107
    },
    {
      "epoch": 0.3737149041062947,
      "flos": 16376670412800.0,
      "grad_norm": 5.726422608710842,
      "learning_rate": 2.8836748605538557e-06,
      "loss": 0.8982,
      "num_input_tokens_seen": 65904900,
      "step": 3108
    },
    {
      "epoch": 0.3738351469969338,
      "flos": 24740981207040.0,
      "grad_norm": 7.164402744612541,
      "learning_rate": 2.882975981868313e-06,
      "loss": 0.8224,
      "num_input_tokens_seen": 65925005,
      "step": 3109
    },
    {
      "epoch": 0.3739553898875729,
      "flos": 31086874398720.0,
      "grad_norm": 6.189531397422404,
      "learning_rate": 2.882276969236016e-06,
      "loss": 0.8635,
      "num_input_tokens_seen": 65946085,
      "step": 3110
    },
    {
      "epoch": 0.374075632778212,
      "flos": 9057773015040.0,
      "grad_norm": 18.885331111769574,
      "learning_rate": 2.881577822763005e-06,
      "loss": 0.9465,
      "num_input_tokens_seen": 65963755,
      "step": 3111
    },
    {
      "epoch": 0.3741958756688511,
      "flos": 18520522076160.0,
      "grad_norm": 7.3465440994441815,
      "learning_rate": 2.880878542555338e-06,
      "loss": 1.0764,
      "num_input_tokens_seen": 65981240,
      "step": 3112
    },
    {
      "epoch": 0.37431611855949015,
      "flos": 15222631956480.0,
      "grad_norm": 5.023777500826208,
      "learning_rate": 2.8801791287190976e-06,
      "loss": 0.9999,
      "num_input_tokens_seen": 65998955,
      "step": 3113
    },
    {
      "epoch": 0.37443636145012926,
      "flos": 17215717478400.0,
      "grad_norm": 8.45928268241324,
      "learning_rate": 2.8794795813603817e-06,
      "loss": 1.0304,
      "num_input_tokens_seen": 66014140,
      "step": 3114
    },
    {
      "epoch": 0.3745566043407684,
      "flos": 10870989312000.0,
      "grad_norm": 5.747783067659312,
      "learning_rate": 2.878779900585314e-06,
      "loss": 1.0073,
      "num_input_tokens_seen": 66031700,
      "step": 3115
    },
    {
      "epoch": 0.37467684723140743,
      "flos": 17503966494720.0,
      "grad_norm": 5.5834060588294525,
      "learning_rate": 2.8780800865000336e-06,
      "loss": 0.961,
      "num_input_tokens_seen": 66052730,
      "step": 3116
    },
    {
      "epoch": 0.37479709012204654,
      "flos": 46090285977600.0,
      "grad_norm": 1.0827219717431609,
      "learning_rate": 2.877380139210702e-06,
      "loss": 0.8559,
      "num_input_tokens_seen": 66111120,
      "step": 3117
    },
    {
      "epoch": 0.37491733301268565,
      "flos": 16925751582720.0,
      "grad_norm": 6.595507636094678,
      "learning_rate": 2.876680058823501e-06,
      "loss": 0.9439,
      "num_input_tokens_seen": 66131240,
      "step": 3118
    },
    {
      "epoch": 0.3750375759033247,
      "flos": 22931995791360.0,
      "grad_norm": 5.121747915705744,
      "learning_rate": 2.8759798454446314e-06,
      "loss": 0.8514,
      "num_input_tokens_seen": 66154125,
      "step": 3119
    },
    {
      "epoch": 0.3751578187939638,
      "flos": 16612507729920.0,
      "grad_norm": 3.9978622446071688,
      "learning_rate": 2.8752794991803173e-06,
      "loss": 1.0049,
      "num_input_tokens_seen": 66171530,
      "step": 3120
    },
    {
      "epoch": 0.37527806168460287,
      "flos": 10314642063360.0,
      "grad_norm": 8.425189949407297,
      "learning_rate": 2.8745790201367976e-06,
      "loss": 0.949,
      "num_input_tokens_seen": 66187005,
      "step": 3121
    },
    {
      "epoch": 0.375398304575242,
      "flos": 18784389427200.0,
      "grad_norm": 10.200613369918429,
      "learning_rate": 2.8738784084203373e-06,
      "loss": 1.0539,
      "num_input_tokens_seen": 66206800,
      "step": 3122
    },
    {
      "epoch": 0.3755185474658811,
      "flos": 15799528550400.0,
      "grad_norm": 8.189164087210955,
      "learning_rate": 2.873177664137216e-06,
      "loss": 0.9784,
      "num_input_tokens_seen": 66227450,
      "step": 3123
    },
    {
      "epoch": 0.37563879035652015,
      "flos": 21963467735040.0,
      "grad_norm": 2.8971009990347936,
      "learning_rate": 2.8724767873937384e-06,
      "loss": 0.8819,
      "num_input_tokens_seen": 66251290,
      "step": 3124
    },
    {
      "epoch": 0.37575903324715926,
      "flos": 14747554222080.0,
      "grad_norm": 5.8415376279821105,
      "learning_rate": 2.871775778296225e-06,
      "loss": 1.0637,
      "num_input_tokens_seen": 66268100,
      "step": 3125
    },
    {
      "epoch": 0.37587927613779837,
      "flos": 13256196833280.0,
      "grad_norm": 8.318902554392174,
      "learning_rate": 2.8710746369510196e-06,
      "loss": 0.9826,
      "num_input_tokens_seen": 66285805,
      "step": 3126
    },
    {
      "epoch": 0.3759995190284374,
      "flos": 9610226626560.0,
      "grad_norm": 10.141544341512262,
      "learning_rate": 2.8703733634644846e-06,
      "loss": 1.0364,
      "num_input_tokens_seen": 66300280,
      "step": 3127
    },
    {
      "epoch": 0.37611976191907653,
      "flos": 14540513402880.0,
      "grad_norm": 15.270897726813985,
      "learning_rate": 2.869671957943002e-06,
      "loss": 0.9927,
      "num_input_tokens_seen": 66319155,
      "step": 3128
    },
    {
      "epoch": 0.37624000480971564,
      "flos": 15012678574080.0,
      "grad_norm": 11.213974980500057,
      "learning_rate": 2.8689704204929747e-06,
      "loss": 0.9371,
      "num_input_tokens_seen": 66338055,
      "step": 3129
    },
    {
      "epoch": 0.3763602477003547,
      "flos": 16035089940480.0,
      "grad_norm": 8.543153165162028,
      "learning_rate": 2.8682687512208253e-06,
      "loss": 1.0041,
      "num_input_tokens_seen": 66356785,
      "step": 3130
    },
    {
      "epoch": 0.3764804905909938,
      "flos": 19601507512320.0,
      "grad_norm": 5.794227015868966,
      "learning_rate": 2.8675669502329972e-06,
      "loss": 0.9992,
      "num_input_tokens_seen": 66378035,
      "step": 3131
    },
    {
      "epoch": 0.3766007334816329,
      "flos": 16008960737280.0,
      "grad_norm": 6.289197761550095,
      "learning_rate": 2.866865017635952e-06,
      "loss": 1.0239,
      "num_input_tokens_seen": 66395575,
      "step": 3132
    },
    {
      "epoch": 0.376720976372272,
      "flos": 18473291673600.0,
      "grad_norm": 5.1964737820425935,
      "learning_rate": 2.866162953536174e-06,
      "loss": 0.9885,
      "num_input_tokens_seen": 66416265,
      "step": 3133
    },
    {
      "epoch": 0.3768412192629111,
      "flos": 12784399564800.0,
      "grad_norm": 8.628757266566955,
      "learning_rate": 2.8654607580401634e-06,
      "loss": 0.9476,
      "num_input_tokens_seen": 66435720,
      "step": 3134
    },
    {
      "epoch": 0.3769614621535502,
      "flos": 45092532203520.0,
      "grad_norm": 0.9499289285275715,
      "learning_rate": 2.8647584312544446e-06,
      "loss": 0.8776,
      "num_input_tokens_seen": 66500645,
      "step": 3135
    },
    {
      "epoch": 0.37708170504418925,
      "flos": 16823994040320.0,
      "grad_norm": 3.597452130983371,
      "learning_rate": 2.864055973285559e-06,
      "loss": 1.0465,
      "num_input_tokens_seen": 66522365,
      "step": 3136
    },
    {
      "epoch": 0.37720194793482836,
      "flos": 17370622525440.0,
      "grad_norm": 4.165108325735635,
      "learning_rate": 2.8633533842400698e-06,
      "loss": 1.0569,
      "num_input_tokens_seen": 66542285,
      "step": 3137
    },
    {
      "epoch": 0.3773221908254674,
      "flos": 14908345712640.0,
      "grad_norm": 7.111387399763264,
      "learning_rate": 2.862650664224558e-06,
      "loss": 0.9746,
      "num_input_tokens_seen": 66560855,
      "step": 3138
    },
    {
      "epoch": 0.37744243371610653,
      "flos": 26863455744000.0,
      "grad_norm": 18.265247403068827,
      "learning_rate": 2.861947813345627e-06,
      "loss": 0.8952,
      "num_input_tokens_seen": 66583275,
      "step": 3139
    },
    {
      "epoch": 0.37756267660674564,
      "flos": 18605194690560.0,
      "grad_norm": 4.501834283613098,
      "learning_rate": 2.8612448317098974e-06,
      "loss": 0.9084,
      "num_input_tokens_seen": 66603330,
      "step": 3140
    },
    {
      "epoch": 0.3776829194973847,
      "flos": 13781049630720.0,
      "grad_norm": 9.15390286616072,
      "learning_rate": 2.8605417194240114e-06,
      "loss": 1.0304,
      "num_input_tokens_seen": 66621410,
      "step": 3141
    },
    {
      "epoch": 0.3778031623880238,
      "flos": 12310609489920.0,
      "grad_norm": 4.054962207698042,
      "learning_rate": 2.8598384765946315e-06,
      "loss": 0.9875,
      "num_input_tokens_seen": 66639785,
      "step": 3142
    },
    {
      "epoch": 0.3779234052786629,
      "flos": 19313350471680.0,
      "grad_norm": 5.818597364216331,
      "learning_rate": 2.8591351033284377e-06,
      "loss": 0.9097,
      "num_input_tokens_seen": 66659235,
      "step": 3143
    },
    {
      "epoch": 0.37804364816930197,
      "flos": 13964046028800.0,
      "grad_norm": 4.915784767498582,
      "learning_rate": 2.8584315997321325e-06,
      "loss": 1.0181,
      "num_input_tokens_seen": 66677960,
      "step": 3144
    },
    {
      "epoch": 0.3781638910599411,
      "flos": 16133567016960.0,
      "grad_norm": 7.380334508908041,
      "learning_rate": 2.8577279659124356e-06,
      "loss": 0.9709,
      "num_input_tokens_seen": 66695355,
      "step": 3145
    },
    {
      "epoch": 0.3782841339505802,
      "flos": 10345462026240.0,
      "grad_norm": 7.814547827312936,
      "learning_rate": 2.857024201976089e-06,
      "loss": 1.017,
      "num_input_tokens_seen": 66712635,
      "step": 3146
    },
    {
      "epoch": 0.37840437684121925,
      "flos": 23404436889600.0,
      "grad_norm": 17.09879274624107,
      "learning_rate": 2.8563203080298516e-06,
      "loss": 0.9195,
      "num_input_tokens_seen": 66733130,
      "step": 3147
    },
    {
      "epoch": 0.37852461973185836,
      "flos": 13020114247680.0,
      "grad_norm": 4.638292542055587,
      "learning_rate": 2.855616284180505e-06,
      "loss": 1.0745,
      "num_input_tokens_seen": 66749900,
      "step": 3148
    },
    {
      "epoch": 0.37864486262249747,
      "flos": 42587071918080.0,
      "grad_norm": 0.9814479543166403,
      "learning_rate": 2.8549121305348477e-06,
      "loss": 0.9195,
      "num_input_tokens_seen": 66809405,
      "step": 3149
    },
    {
      "epoch": 0.3787651055131365,
      "flos": 16608430141440.0,
      "grad_norm": 5.604983510138614,
      "learning_rate": 2.8542078471997006e-06,
      "loss": 1.0047,
      "num_input_tokens_seen": 66826740,
      "step": 3150
    },
    {
      "epoch": 0.37888534840377563,
      "flos": 17498141368320.0,
      "grad_norm": 6.048827494349338,
      "learning_rate": 2.8535034342819013e-06,
      "loss": 0.9505,
      "num_input_tokens_seen": 66843870,
      "step": 3151
    },
    {
      "epoch": 0.37900559129441475,
      "flos": 17057286696960.0,
      "grad_norm": 3.145735648619587,
      "learning_rate": 2.85279889188831e-06,
      "loss": 0.9167,
      "num_input_tokens_seen": 66863965,
      "step": 3152
    },
    {
      "epoch": 0.3791258341850538,
      "flos": 17530157015040.0,
      "grad_norm": 10.313973246446684,
      "learning_rate": 2.852094220125805e-06,
      "loss": 0.9965,
      "num_input_tokens_seen": 66883195,
      "step": 3153
    },
    {
      "epoch": 0.3792460770756929,
      "flos": 12338578206720.0,
      "grad_norm": 5.29187567991444,
      "learning_rate": 2.8513894191012846e-06,
      "loss": 0.8989,
      "num_input_tokens_seen": 66901895,
      "step": 3154
    },
    {
      "epoch": 0.37936631996633197,
      "flos": 17215257600000.0,
      "grad_norm": 6.749028736455198,
      "learning_rate": 2.8506844889216664e-06,
      "loss": 0.9734,
      "num_input_tokens_seen": 66921000,
      "step": 3155
    },
    {
      "epoch": 0.3794865628569711,
      "flos": 50345812254720.0,
      "grad_norm": 0.9442042009368056,
      "learning_rate": 2.849979429693887e-06,
      "loss": 0.866,
      "num_input_tokens_seen": 66981705,
      "step": 3156
    },
    {
      "epoch": 0.3796068057476102,
      "flos": 11158931742720.0,
      "grad_norm": 5.857377914869217,
      "learning_rate": 2.8492742415249042e-06,
      "loss": 0.9337,
      "num_input_tokens_seen": 66999070,
      "step": 3157
    },
    {
      "epoch": 0.37972704863824924,
      "flos": 17923260088320.0,
      "grad_norm": 4.957327134219995,
      "learning_rate": 2.848568924521694e-06,
      "loss": 0.9546,
      "num_input_tokens_seen": 67019570,
      "step": 3158
    },
    {
      "epoch": 0.37984729152888835,
      "flos": 18654969753600.0,
      "grad_norm": 6.929868607589687,
      "learning_rate": 2.8478634787912526e-06,
      "loss": 0.9211,
      "num_input_tokens_seen": 67037345,
      "step": 3159
    },
    {
      "epoch": 0.37996753441952746,
      "flos": 18237730283520.0,
      "grad_norm": 4.821964617605308,
      "learning_rate": 2.847157904440596e-06,
      "loss": 0.9666,
      "num_input_tokens_seen": 67056795,
      "step": 3160
    },
    {
      "epoch": 0.3800877773101665,
      "flos": 14276952637440.0,
      "grad_norm": 11.202258045909693,
      "learning_rate": 2.846452201576759e-06,
      "loss": 0.9424,
      "num_input_tokens_seen": 67075890,
      "step": 3161
    },
    {
      "epoch": 0.38020802020080563,
      "flos": 45137432555520.0,
      "grad_norm": 12.411398023724818,
      "learning_rate": 2.845746370306795e-06,
      "loss": 0.8512,
      "num_input_tokens_seen": 67140800,
      "step": 3162
    },
    {
      "epoch": 0.38032826309144474,
      "flos": 15118145802240.0,
      "grad_norm": 7.54026590338821,
      "learning_rate": 2.84504041073778e-06,
      "loss": 0.9724,
      "num_input_tokens_seen": 67158935,
      "step": 3163
    },
    {
      "epoch": 0.3804485059820838,
      "flos": 13440634183680.0,
      "grad_norm": 6.209870252743467,
      "learning_rate": 2.844334322976806e-06,
      "loss": 0.9983,
      "num_input_tokens_seen": 67178870,
      "step": 3164
    },
    {
      "epoch": 0.3805687488727229,
      "flos": 15509807923200.0,
      "grad_norm": 4.878780561894351,
      "learning_rate": 2.8436281071309866e-06,
      "loss": 1.0266,
      "num_input_tokens_seen": 67197130,
      "step": 3165
    },
    {
      "epoch": 0.380688991763362,
      "flos": 41900323921920.0,
      "grad_norm": 0.8612566029443635,
      "learning_rate": 2.842921763307455e-06,
      "loss": 0.7508,
      "num_input_tokens_seen": 67259660,
      "step": 3166
    },
    {
      "epoch": 0.38080923465400107,
      "flos": 16921949921280.0,
      "grad_norm": 6.38189623453248,
      "learning_rate": 2.842215291613361e-06,
      "loss": 1.0251,
      "num_input_tokens_seen": 67277760,
      "step": 3167
    },
    {
      "epoch": 0.3809294775446402,
      "flos": 39329973903360.0,
      "grad_norm": 0.888112355248933,
      "learning_rate": 2.8415086921558774e-06,
      "loss": 0.8323,
      "num_input_tokens_seen": 67340905,
      "step": 3168
    },
    {
      "epoch": 0.38104972043527924,
      "flos": 17529329233920.0,
      "grad_norm": 4.4051492817726565,
      "learning_rate": 2.840801965042194e-06,
      "loss": 0.9756,
      "num_input_tokens_seen": 67360085,
      "step": 3169
    },
    {
      "epoch": 0.38116996332591835,
      "flos": 16245227274240.0,
      "grad_norm": 7.3407841314002935,
      "learning_rate": 2.840095110379521e-06,
      "loss": 1.0244,
      "num_input_tokens_seen": 67379325,
      "step": 3170
    },
    {
      "epoch": 0.38129020621655746,
      "flos": 43544900136960.0,
      "grad_norm": 0.8093537891215581,
      "learning_rate": 2.8393881282750884e-06,
      "loss": 0.7481,
      "num_input_tokens_seen": 67441875,
      "step": 3171
    },
    {
      "epoch": 0.3814104491071965,
      "flos": 15375942758400.0,
      "grad_norm": 4.087595032686716,
      "learning_rate": 2.838681018836144e-06,
      "loss": 0.9728,
      "num_input_tokens_seen": 67458915,
      "step": 3172
    },
    {
      "epoch": 0.3815306919978356,
      "flos": 13544691118080.0,
      "grad_norm": 15.231021766377053,
      "learning_rate": 2.837973782169955e-06,
      "loss": 0.9722,
      "num_input_tokens_seen": 67477010,
      "step": 3173
    },
    {
      "epoch": 0.38165093488847474,
      "flos": 48022559354880.0,
      "grad_norm": 0.9347298385331234,
      "learning_rate": 2.8372664183838096e-06,
      "loss": 0.8198,
      "num_input_tokens_seen": 67539750,
      "step": 3174
    },
    {
      "epoch": 0.3817711777791138,
      "flos": 15873961205760.0,
      "grad_norm": 10.719652855219977,
      "learning_rate": 2.836558927585015e-06,
      "loss": 0.8738,
      "num_input_tokens_seen": 67556440,
      "step": 3175
    },
    {
      "epoch": 0.3818914206697529,
      "flos": 16219282022400.0,
      "grad_norm": 8.186856543029952,
      "learning_rate": 2.8358513098808957e-06,
      "loss": 1.0133,
      "num_input_tokens_seen": 67576475,
      "step": 3176
    },
    {
      "epoch": 0.382011663560392,
      "flos": 17343880151040.0,
      "grad_norm": 4.7618919885554885,
      "learning_rate": 2.835143565378798e-06,
      "loss": 0.9601,
      "num_input_tokens_seen": 67596660,
      "step": 3177
    },
    {
      "epoch": 0.38213190645103107,
      "flos": 15616286883840.0,
      "grad_norm": 47.935959374689425,
      "learning_rate": 2.8344356941860847e-06,
      "loss": 0.9739,
      "num_input_tokens_seen": 67616010,
      "step": 3178
    },
    {
      "epoch": 0.3822521493416702,
      "flos": 25343792394240.0,
      "grad_norm": 6.030373306991839,
      "learning_rate": 2.8337276964101403e-06,
      "loss": 0.8557,
      "num_input_tokens_seen": 67636170,
      "step": 3179
    },
    {
      "epoch": 0.3823723922323093,
      "flos": 14961155973120.0,
      "grad_norm": 3.9501519765469,
      "learning_rate": 2.833019572158367e-06,
      "loss": 0.9446,
      "num_input_tokens_seen": 67654325,
      "step": 3180
    },
    {
      "epoch": 0.38249263512294834,
      "flos": 14040195563520.0,
      "grad_norm": 4.233300554844469,
      "learning_rate": 2.8323113215381872e-06,
      "loss": 1.0058,
      "num_input_tokens_seen": 67672390,
      "step": 3181
    },
    {
      "epoch": 0.38261287801358745,
      "flos": 15222662615040.0,
      "grad_norm": 8.897518763617517,
      "learning_rate": 2.831602944657042e-06,
      "loss": 0.9484,
      "num_input_tokens_seen": 67690190,
      "step": 3182
    },
    {
      "epoch": 0.38273312090422656,
      "flos": 15616286883840.0,
      "grad_norm": 8.458338795426833,
      "learning_rate": 2.830894441622391e-06,
      "loss": 0.9397,
      "num_input_tokens_seen": 67706560,
      "step": 3183
    },
    {
      "epoch": 0.3828533637948656,
      "flos": 17635286999040.0,
      "grad_norm": 5.67242008669074,
      "learning_rate": 2.8301858125417134e-06,
      "loss": 0.9845,
      "num_input_tokens_seen": 67726120,
      "step": 3184
    },
    {
      "epoch": 0.38297360668550473,
      "flos": 15956794306560.0,
      "grad_norm": 6.642548836674625,
      "learning_rate": 2.8294770575225082e-06,
      "loss": 0.938,
      "num_input_tokens_seen": 67745970,
      "step": 3185
    },
    {
      "epoch": 0.3830938495761438,
      "flos": 17711007313920.0,
      "grad_norm": 4.642976415590277,
      "learning_rate": 2.828768176672293e-06,
      "loss": 1.0338,
      "num_input_tokens_seen": 67764805,
      "step": 3186
    },
    {
      "epoch": 0.3832140924667829,
      "flos": 23560997498880.0,
      "grad_norm": 9.264360158188206,
      "learning_rate": 2.8280591700986044e-06,
      "loss": 0.9119,
      "num_input_tokens_seen": 67786390,
      "step": 3187
    },
    {
      "epoch": 0.383334335357422,
      "flos": 22746883952640.0,
      "grad_norm": 4.250905781881158,
      "learning_rate": 2.827350037908999e-06,
      "loss": 0.9407,
      "num_input_tokens_seen": 67805550,
      "step": 3188
    },
    {
      "epoch": 0.38345457824806106,
      "flos": 14042556272640.0,
      "grad_norm": 6.605418506794429,
      "learning_rate": 2.8266407802110496e-06,
      "loss": 0.9863,
      "num_input_tokens_seen": 67823525,
      "step": 3189
    },
    {
      "epoch": 0.3835748211387002,
      "flos": 15930941030400.0,
      "grad_norm": 4.7092085618424955,
      "learning_rate": 2.8259313971123515e-06,
      "loss": 0.9486,
      "num_input_tokens_seen": 67844365,
      "step": 3190
    },
    {
      "epoch": 0.3836950640293393,
      "flos": 17870572462080.0,
      "grad_norm": 5.574824192713442,
      "learning_rate": 2.8252218887205166e-06,
      "loss": 0.9685,
      "num_input_tokens_seen": 67864775,
      "step": 3191
    },
    {
      "epoch": 0.38381530691997834,
      "flos": 15485242306560.0,
      "grad_norm": 4.515816971549248,
      "learning_rate": 2.824512255143178e-06,
      "loss": 0.9995,
      "num_input_tokens_seen": 67883730,
      "step": 3192
    },
    {
      "epoch": 0.38393554981061745,
      "flos": 15092568453120.0,
      "grad_norm": 5.9401993889529985,
      "learning_rate": 2.8238024964879855e-06,
      "loss": 0.9862,
      "num_input_tokens_seen": 67904345,
      "step": 3193
    },
    {
      "epoch": 0.38405579270125656,
      "flos": 12050022604800.0,
      "grad_norm": 5.581298096589463,
      "learning_rate": 2.8230926128626095e-06,
      "loss": 0.9524,
      "num_input_tokens_seen": 67922560,
      "step": 3194
    },
    {
      "epoch": 0.3841760355918956,
      "flos": 15511187558400.0,
      "grad_norm": 6.933245204707786,
      "learning_rate": 2.822382604374738e-06,
      "loss": 0.9722,
      "num_input_tokens_seen": 67941205,
      "step": 3195
    },
    {
      "epoch": 0.3842962784825347,
      "flos": 18442318417920.0,
      "grad_norm": 5.036017392359509,
      "learning_rate": 2.8216724711320793e-06,
      "loss": 0.8594,
      "num_input_tokens_seen": 67960050,
      "step": 3196
    },
    {
      "epoch": 0.38441652137317384,
      "flos": 18028022169600.0,
      "grad_norm": 4.464297337108196,
      "learning_rate": 2.820962213242361e-06,
      "loss": 0.9935,
      "num_input_tokens_seen": 67979100,
      "step": 3197
    },
    {
      "epoch": 0.3845367642638129,
      "flos": 12836290068480.0,
      "grad_norm": 5.446322069757206,
      "learning_rate": 2.8202518308133264e-06,
      "loss": 1.0541,
      "num_input_tokens_seen": 67996095,
      "step": 3198
    },
    {
      "epoch": 0.384657007154452,
      "flos": 17949665218560.0,
      "grad_norm": 5.222432373543983,
      "learning_rate": 2.8195413239527426e-06,
      "loss": 0.9384,
      "num_input_tokens_seen": 68015555,
      "step": 3199
    },
    {
      "epoch": 0.38477725004509106,
      "flos": 14095826411520.0,
      "grad_norm": 4.777001182270124,
      "learning_rate": 2.8188306927683906e-06,
      "loss": 1.0051,
      "num_input_tokens_seen": 68034745,
      "step": 3200
    },
    {
      "epoch": 0.38489749293573017,
      "flos": 12941144125440.0,
      "grad_norm": 5.92266566652315,
      "learning_rate": 2.818119937368074e-06,
      "loss": 0.9309,
      "num_input_tokens_seen": 68053100,
      "step": 3201
    },
    {
      "epoch": 0.3850177358263693,
      "flos": 17346516787200.0,
      "grad_norm": 3.4825296347864043,
      "learning_rate": 2.817409057859613e-06,
      "loss": 0.8464,
      "num_input_tokens_seen": 68071810,
      "step": 3202
    },
    {
      "epoch": 0.38513797871700833,
      "flos": 12518110187520.0,
      "grad_norm": 5.18244117508372,
      "learning_rate": 2.8166980543508482e-06,
      "loss": 0.978,
      "num_input_tokens_seen": 68087420,
      "step": 3203
    },
    {
      "epoch": 0.38525822160764744,
      "flos": 18317098967040.0,
      "grad_norm": 4.629148942081959,
      "learning_rate": 2.815986926949638e-06,
      "loss": 0.9748,
      "num_input_tokens_seen": 68105640,
      "step": 3204
    },
    {
      "epoch": 0.38537846449828655,
      "flos": 14331326484480.0,
      "grad_norm": 8.294703197538409,
      "learning_rate": 2.8152756757638597e-06,
      "loss": 0.9943,
      "num_input_tokens_seen": 68123860,
      "step": 3205
    },
    {
      "epoch": 0.3854987073889256,
      "flos": 16376455802880.0,
      "grad_norm": 5.763655328678384,
      "learning_rate": 2.8145643009014093e-06,
      "loss": 1.0286,
      "num_input_tokens_seen": 68142075,
      "step": 3206
    },
    {
      "epoch": 0.3856189502795647,
      "flos": 14328751165440.0,
      "grad_norm": 5.3940288025197,
      "learning_rate": 2.813852802470202e-06,
      "loss": 0.9796,
      "num_input_tokens_seen": 68159690,
      "step": 3207
    },
    {
      "epoch": 0.38573919317020383,
      "flos": 18290479226880.0,
      "grad_norm": 4.691138724418441,
      "learning_rate": 2.8131411805781717e-06,
      "loss": 0.9138,
      "num_input_tokens_seen": 68179535,
      "step": 3208
    },
    {
      "epoch": 0.3858594360608429,
      "flos": 21250008023040.0,
      "grad_norm": 4.825738310781121,
      "learning_rate": 2.8124294353332707e-06,
      "loss": 0.831,
      "num_input_tokens_seen": 68197930,
      "step": 3209
    },
    {
      "epoch": 0.385979678951482,
      "flos": 17634888437760.0,
      "grad_norm": 6.503253709288378,
      "learning_rate": 2.8117175668434713e-06,
      "loss": 0.9685,
      "num_input_tokens_seen": 68217310,
      "step": 3210
    },
    {
      "epoch": 0.3860999218421211,
      "flos": 15117685923840.0,
      "grad_norm": 7.28294909385354,
      "learning_rate": 2.811005575216762e-06,
      "loss": 0.8734,
      "num_input_tokens_seen": 68235745,
      "step": 3211
    },
    {
      "epoch": 0.38622016473276016,
      "flos": 17452781137920.0,
      "grad_norm": 6.373250683896339,
      "learning_rate": 2.8102934605611513e-06,
      "loss": 0.9752,
      "num_input_tokens_seen": 68257100,
      "step": 3212
    },
    {
      "epoch": 0.3863404076233993,
      "flos": 14593293004800.0,
      "grad_norm": 5.728857894437442,
      "learning_rate": 2.8095812229846665e-06,
      "loss": 0.8646,
      "num_input_tokens_seen": 68276780,
      "step": 3213
    },
    {
      "epoch": 0.3864606505140384,
      "flos": 15878376038400.0,
      "grad_norm": 8.757004032710402,
      "learning_rate": 2.808868862595355e-06,
      "loss": 0.8913,
      "num_input_tokens_seen": 68296745,
      "step": 3214
    },
    {
      "epoch": 0.38658089340467744,
      "flos": 18291307008000.0,
      "grad_norm": 5.487583015526501,
      "learning_rate": 2.8081563795012795e-06,
      "loss": 0.9832,
      "num_input_tokens_seen": 68316090,
      "step": 3215
    },
    {
      "epoch": 0.38670113629531655,
      "flos": 24111734231040.0,
      "grad_norm": 5.931642232774725,
      "learning_rate": 2.807443773810524e-06,
      "loss": 0.925,
      "num_input_tokens_seen": 68337070,
      "step": 3216
    },
    {
      "epoch": 0.3868213791859556,
      "flos": 16586470502400.0,
      "grad_norm": 11.66632021434994,
      "learning_rate": 2.80673104563119e-06,
      "loss": 1.0851,
      "num_input_tokens_seen": 68357415,
      "step": 3217
    },
    {
      "epoch": 0.3869416220765947,
      "flos": 13072004751360.0,
      "grad_norm": 4.35079450696041,
      "learning_rate": 2.8060181950713976e-06,
      "loss": 0.9677,
      "num_input_tokens_seen": 68373925,
      "step": 3218
    },
    {
      "epoch": 0.3870618649672338,
      "flos": 11053679124480.0,
      "grad_norm": 4.10294456811548,
      "learning_rate": 2.805305222239286e-06,
      "loss": 0.9988,
      "num_input_tokens_seen": 68390900,
      "step": 3219
    },
    {
      "epoch": 0.3871821078578729,
      "flos": 16717453762560.0,
      "grad_norm": 6.357232560065261,
      "learning_rate": 2.8045921272430118e-06,
      "loss": 0.9244,
      "num_input_tokens_seen": 68410300,
      "step": 3220
    },
    {
      "epoch": 0.387302350748512,
      "flos": 12594443673600.0,
      "grad_norm": 10.738957611020219,
      "learning_rate": 2.803878910190753e-06,
      "loss": 0.9665,
      "num_input_tokens_seen": 68426940,
      "step": 3221
    },
    {
      "epoch": 0.3874225936391511,
      "flos": 8084922101760.0,
      "grad_norm": 7.773629865710259,
      "learning_rate": 2.8031655711907017e-06,
      "loss": 1.0157,
      "num_input_tokens_seen": 68440365,
      "step": 3222
    },
    {
      "epoch": 0.38754283652979016,
      "flos": 15590127022080.0,
      "grad_norm": 10.665804073105736,
      "learning_rate": 2.8024521103510723e-06,
      "loss": 0.9973,
      "num_input_tokens_seen": 68456855,
      "step": 3223
    },
    {
      "epoch": 0.38766307942042927,
      "flos": 15036937605120.0,
      "grad_norm": 3.967519013540166,
      "learning_rate": 2.8017385277800952e-06,
      "loss": 0.9485,
      "num_input_tokens_seen": 68474930,
      "step": 3224
    },
    {
      "epoch": 0.3877833223110684,
      "flos": 19522752000000.0,
      "grad_norm": 27.644238431489324,
      "learning_rate": 2.8010248235860213e-06,
      "loss": 0.928,
      "num_input_tokens_seen": 68494765,
      "step": 3225
    },
    {
      "epoch": 0.38790356520170743,
      "flos": 46180109230080.0,
      "grad_norm": 0.8890663361424714,
      "learning_rate": 2.8003109978771192e-06,
      "loss": 0.8621,
      "num_input_tokens_seen": 68555650,
      "step": 3226
    },
    {
      "epoch": 0.38802380809234654,
      "flos": 15877640232960.0,
      "grad_norm": 4.585836432153975,
      "learning_rate": 2.799597050761674e-06,
      "loss": 0.9825,
      "num_input_tokens_seen": 68571575,
      "step": 3227
    },
    {
      "epoch": 0.38814405098298566,
      "flos": 17972881858560.0,
      "grad_norm": 4.631609343926374,
      "learning_rate": 2.7988829823479924e-06,
      "loss": 0.9859,
      "num_input_tokens_seen": 68589685,
      "step": 3228
    },
    {
      "epoch": 0.3882642938736247,
      "flos": 13358996766720.0,
      "grad_norm": 8.532246625766872,
      "learning_rate": 2.7981687927443976e-06,
      "loss": 0.8364,
      "num_input_tokens_seen": 68606205,
      "step": 3229
    },
    {
      "epoch": 0.3883845367642638,
      "flos": 15379959029760.0,
      "grad_norm": 6.262053806148088,
      "learning_rate": 2.797454482059231e-06,
      "loss": 1.049,
      "num_input_tokens_seen": 68626080,
      "step": 3230
    },
    {
      "epoch": 0.3885047796549029,
      "flos": 14592710492160.0,
      "grad_norm": 3.5921008762901323,
      "learning_rate": 2.7967400504008537e-06,
      "loss": 1.0396,
      "num_input_tokens_seen": 68645100,
      "step": 3231
    },
    {
      "epoch": 0.388625022545542,
      "flos": 46050820300800.0,
      "grad_norm": 0.8522110665297307,
      "learning_rate": 2.7960254978776456e-06,
      "loss": 0.8202,
      "num_input_tokens_seen": 68706910,
      "step": 3232
    },
    {
      "epoch": 0.3887452654361811,
      "flos": 12836627312640.0,
      "grad_norm": 5.663102356659462,
      "learning_rate": 2.7953108245980006e-06,
      "loss": 0.9941,
      "num_input_tokens_seen": 68725145,
      "step": 3233
    },
    {
      "epoch": 0.38886550832682015,
      "flos": 17768048455680.0,
      "grad_norm": 3.545005944466281,
      "learning_rate": 2.7945960306703365e-06,
      "loss": 0.9341,
      "num_input_tokens_seen": 68747850,
      "step": 3234
    },
    {
      "epoch": 0.38898575121745926,
      "flos": 19366375342080.0,
      "grad_norm": 3.9181039621406284,
      "learning_rate": 2.7938811162030865e-06,
      "loss": 0.8452,
      "num_input_tokens_seen": 68767835,
      "step": 3235
    },
    {
      "epoch": 0.3891059941080984,
      "flos": 20489931079680.0,
      "grad_norm": 5.025712087099884,
      "learning_rate": 2.793166081304702e-06,
      "loss": 1.0181,
      "num_input_tokens_seen": 68788050,
      "step": 3236
    },
    {
      "epoch": 0.38922623699873743,
      "flos": 16271601745920.0,
      "grad_norm": 4.004700997724644,
      "learning_rate": 2.7924509260836543e-06,
      "loss": 1.0001,
      "num_input_tokens_seen": 68806895,
      "step": 3237
    },
    {
      "epoch": 0.38934647988937654,
      "flos": 14042985492480.0,
      "grad_norm": 4.48727333962799,
      "learning_rate": 2.791735650648431e-06,
      "loss": 0.8725,
      "num_input_tokens_seen": 68825735,
      "step": 3238
    },
    {
      "epoch": 0.38946672278001565,
      "flos": 13617836113920.0,
      "grad_norm": 7.083755510502064,
      "learning_rate": 2.791020255107538e-06,
      "loss": 0.9321,
      "num_input_tokens_seen": 68842825,
      "step": 3239
    },
    {
      "epoch": 0.3895869656706547,
      "flos": 17739865128960.0,
      "grad_norm": 5.240746472019456,
      "learning_rate": 2.7903047395695023e-06,
      "loss": 0.9955,
      "num_input_tokens_seen": 68862445,
      "step": 3240
    },
    {
      "epoch": 0.3897072085612938,
      "flos": 17162845900800.0,
      "grad_norm": 5.107993402833793,
      "learning_rate": 2.789589104142865e-06,
      "loss": 1.0981,
      "num_input_tokens_seen": 68879790,
      "step": 3241
    },
    {
      "epoch": 0.3898274514519329,
      "flos": 12155796418560.0,
      "grad_norm": 5.168525279463488,
      "learning_rate": 2.7888733489361895e-06,
      "loss": 0.9578,
      "num_input_tokens_seen": 68897925,
      "step": 3242
    },
    {
      "epoch": 0.389947694342572,
      "flos": 47307750666240.0,
      "grad_norm": 0.8218846106601895,
      "learning_rate": 2.788157474058054e-06,
      "loss": 0.8436,
      "num_input_tokens_seen": 68959920,
      "step": 3243
    },
    {
      "epoch": 0.3900679372332111,
      "flos": 18318233333760.0,
      "grad_norm": 4.438621916452438,
      "learning_rate": 2.7874414796170555e-06,
      "loss": 0.8945,
      "num_input_tokens_seen": 68981130,
      "step": 3244
    },
    {
      "epoch": 0.3901881801238502,
      "flos": 8300087439360.0,
      "grad_norm": 7.019007353849862,
      "learning_rate": 2.7867253657218113e-06,
      "loss": 1.0224,
      "num_input_tokens_seen": 68994740,
      "step": 3245
    },
    {
      "epoch": 0.39030842301448926,
      "flos": 19445038878720.0,
      "grad_norm": 5.86682696701451,
      "learning_rate": 2.7860091324809544e-06,
      "loss": 0.9206,
      "num_input_tokens_seen": 69015520,
      "step": 3246
    },
    {
      "epoch": 0.39042866590512837,
      "flos": 19340430090240.0,
      "grad_norm": 4.070313905799901,
      "learning_rate": 2.7852927800031377e-06,
      "loss": 1.006,
      "num_input_tokens_seen": 69035405,
      "step": 3247
    },
    {
      "epoch": 0.3905489087957674,
      "flos": 21174900879360.0,
      "grad_norm": 6.47700688548191,
      "learning_rate": 2.7845763083970298e-06,
      "loss": 1.0288,
      "num_input_tokens_seen": 69055525,
      "step": 3248
    },
    {
      "epoch": 0.39066915168640653,
      "flos": 17425180323840.0,
      "grad_norm": 7.639625905250071,
      "learning_rate": 2.7838597177713205e-06,
      "loss": 1.0033,
      "num_input_tokens_seen": 69076335,
      "step": 3249
    },
    {
      "epoch": 0.39078939457704565,
      "flos": 14593599590400.0,
      "grad_norm": 6.452962139004973,
      "learning_rate": 2.7831430082347143e-06,
      "loss": 0.9307,
      "num_input_tokens_seen": 69095260,
      "step": 3250
    },
    {
      "epoch": 0.3909096374676847,
      "flos": 16192447672320.0,
      "grad_norm": 5.765659073726382,
      "learning_rate": 2.7824261798959373e-06,
      "loss": 1.0209,
      "num_input_tokens_seen": 69113160,
      "step": 3251
    },
    {
      "epoch": 0.3910298803583238,
      "flos": 16350449233920.0,
      "grad_norm": 5.371389906607841,
      "learning_rate": 2.78170923286373e-06,
      "loss": 0.9913,
      "num_input_tokens_seen": 69132480,
      "step": 3252
    },
    {
      "epoch": 0.3911501232489629,
      "flos": 17294135746560.0,
      "grad_norm": 6.992345166693059,
      "learning_rate": 2.780992167246854e-06,
      "loss": 1.0412,
      "num_input_tokens_seen": 69149725,
      "step": 3253
    },
    {
      "epoch": 0.391270366139602,
      "flos": 43568055459840.0,
      "grad_norm": 1.1382329235408815,
      "learning_rate": 2.7802749831540883e-06,
      "loss": 0.9754,
      "num_input_tokens_seen": 69208345,
      "step": 3254
    },
    {
      "epoch": 0.3913906090302411,
      "flos": 15301418127360.0,
      "grad_norm": 5.089884970813514,
      "learning_rate": 2.7795576806942268e-06,
      "loss": 1.0106,
      "num_input_tokens_seen": 69226870,
      "step": 3255
    },
    {
      "epoch": 0.3915108519208802,
      "flos": 35640245821440.0,
      "grad_norm": 0.7923196475470996,
      "learning_rate": 2.778840259976085e-06,
      "loss": 0.7677,
      "num_input_tokens_seen": 69281820,
      "step": 3256
    },
    {
      "epoch": 0.39163109481151925,
      "flos": 11681117245440.0,
      "grad_norm": 6.0595821949168345,
      "learning_rate": 2.778122721108495e-06,
      "loss": 0.9624,
      "num_input_tokens_seen": 69299770,
      "step": 3257
    },
    {
      "epoch": 0.39175133770215836,
      "flos": 18552691015680.0,
      "grad_norm": 3.794721330103771,
      "learning_rate": 2.7774050642003076e-06,
      "loss": 1.0753,
      "num_input_tokens_seen": 69320300,
      "step": 3258
    },
    {
      "epoch": 0.3918715805927975,
      "flos": 15537592688640.0,
      "grad_norm": 4.224847075273641,
      "learning_rate": 2.7766872893603896e-06,
      "loss": 1.132,
      "num_input_tokens_seen": 69339995,
      "step": 3259
    },
    {
      "epoch": 0.39199182348343653,
      "flos": 14462524354560.0,
      "grad_norm": 8.999120457375135,
      "learning_rate": 2.7759693966976275e-06,
      "loss": 0.9199,
      "num_input_tokens_seen": 69358220,
      "step": 3260
    },
    {
      "epoch": 0.39211206637407564,
      "flos": 15401274839040.0,
      "grad_norm": 4.038479259298382,
      "learning_rate": 2.7752513863209242e-06,
      "loss": 1.0318,
      "num_input_tokens_seen": 69376520,
      "step": 3261
    },
    {
      "epoch": 0.39223230926471475,
      "flos": 14958059458560.0,
      "grad_norm": 8.87857671311588,
      "learning_rate": 2.774533258339203e-06,
      "loss": 1.0311,
      "num_input_tokens_seen": 69393700,
      "step": 3262
    },
    {
      "epoch": 0.3923525521553538,
      "flos": 12469469491200.0,
      "grad_norm": 6.378779170025497,
      "learning_rate": 2.7738150128614014e-06,
      "loss": 0.9881,
      "num_input_tokens_seen": 69410825,
      "step": 3263
    },
    {
      "epoch": 0.3924727950459929,
      "flos": 14593354321920.0,
      "grad_norm": 5.0994029125670455,
      "learning_rate": 2.7730966499964777e-06,
      "loss": 1.0916,
      "num_input_tokens_seen": 69427495,
      "step": 3264
    },
    {
      "epoch": 0.39259303793663197,
      "flos": 11472666132480.0,
      "grad_norm": 7.670079784550301,
      "learning_rate": 2.772378169853408e-06,
      "loss": 0.9929,
      "num_input_tokens_seen": 69444785,
      "step": 3265
    },
    {
      "epoch": 0.3927132808272711,
      "flos": 11889752309760.0,
      "grad_norm": 5.054149380890465,
      "learning_rate": 2.771659572541183e-06,
      "loss": 0.9356,
      "num_input_tokens_seen": 69462435,
      "step": 3266
    },
    {
      "epoch": 0.3928335237179102,
      "flos": 14384290037760.0,
      "grad_norm": 4.585600390545655,
      "learning_rate": 2.7709408581688143e-06,
      "loss": 1.0698,
      "num_input_tokens_seen": 69482140,
      "step": 3267
    },
    {
      "epoch": 0.39295376660854925,
      "flos": 17766331576320.0,
      "grad_norm": 7.402674588583043,
      "learning_rate": 2.7702220268453307e-06,
      "loss": 1.0748,
      "num_input_tokens_seen": 69502220,
      "step": 3268
    },
    {
      "epoch": 0.39307400949918836,
      "flos": 13256288808960.0,
      "grad_norm": 5.664880396227661,
      "learning_rate": 2.7695030786797785e-06,
      "loss": 1.0353,
      "num_input_tokens_seen": 69517835,
      "step": 3269
    },
    {
      "epoch": 0.39319425238982747,
      "flos": 15927599247360.0,
      "grad_norm": 5.422417957042272,
      "learning_rate": 2.7687840137812206e-06,
      "loss": 0.9384,
      "num_input_tokens_seen": 69535640,
      "step": 3270
    },
    {
      "epoch": 0.3933144952804665,
      "flos": 47393833574400.0,
      "grad_norm": 0.8521742067324894,
      "learning_rate": 2.7680648322587395e-06,
      "loss": 0.8274,
      "num_input_tokens_seen": 69600235,
      "step": 3271
    },
    {
      "epoch": 0.39343473817110564,
      "flos": 10948886384640.0,
      "grad_norm": 4.585115167300829,
      "learning_rate": 2.7673455342214334e-06,
      "loss": 1.0043,
      "num_input_tokens_seen": 69616945,
      "step": 3272
    },
    {
      "epoch": 0.39355498106174475,
      "flos": 15144366981120.0,
      "grad_norm": 3.4107830437268376,
      "learning_rate": 2.7666261197784198e-06,
      "loss": 0.9414,
      "num_input_tokens_seen": 69635480,
      "step": 3273
    },
    {
      "epoch": 0.3936752239523838,
      "flos": 9374174699520.0,
      "grad_norm": 6.438315626899032,
      "learning_rate": 2.7659065890388336e-06,
      "loss": 0.9585,
      "num_input_tokens_seen": 69651200,
      "step": 3274
    },
    {
      "epoch": 0.3937954668430229,
      "flos": 11892450263040.0,
      "grad_norm": 5.707449665958792,
      "learning_rate": 2.7651869421118266e-06,
      "loss": 1.0254,
      "num_input_tokens_seen": 69667530,
      "step": 3275
    },
    {
      "epoch": 0.393915709733662,
      "flos": 14956311920640.0,
      "grad_norm": 3.6027240990024083,
      "learning_rate": 2.76446717910657e-06,
      "loss": 1.0178,
      "num_input_tokens_seen": 69687955,
      "step": 3276
    },
    {
      "epoch": 0.3940359526243011,
      "flos": 12154876661760.0,
      "grad_norm": 6.405711081562235,
      "learning_rate": 2.763747300132249e-06,
      "loss": 0.9455,
      "num_input_tokens_seen": 69705115,
      "step": 3277
    },
    {
      "epoch": 0.3941561955149402,
      "flos": 14907793858560.0,
      "grad_norm": 9.063153263092957,
      "learning_rate": 2.7630273052980704e-06,
      "loss": 1.0621,
      "num_input_tokens_seen": 69725425,
      "step": 3278
    },
    {
      "epoch": 0.39427643840557924,
      "flos": 13360897597440.0,
      "grad_norm": 8.317906965934718,
      "learning_rate": 2.7623071947132554e-06,
      "loss": 0.8703,
      "num_input_tokens_seen": 69742175,
      "step": 3279
    },
    {
      "epoch": 0.39439668129621835,
      "flos": 16534120120320.0,
      "grad_norm": 5.422768466999092,
      "learning_rate": 2.7615869684870458e-06,
      "loss": 0.9843,
      "num_input_tokens_seen": 69761205,
      "step": 3280
    },
    {
      "epoch": 0.39451692418685746,
      "flos": 18973118976000.0,
      "grad_norm": 11.287352879305788,
      "learning_rate": 2.7608666267286986e-06,
      "loss": 1.0401,
      "num_input_tokens_seen": 69781155,
      "step": 3281
    },
    {
      "epoch": 0.3946371670774965,
      "flos": 12940561612800.0,
      "grad_norm": 4.743693130011547,
      "learning_rate": 2.760146169547489e-06,
      "loss": 1.0487,
      "num_input_tokens_seen": 69797640,
      "step": 3282
    },
    {
      "epoch": 0.39475740996813563,
      "flos": 17214889697280.0,
      "grad_norm": 3.244312738840062,
      "learning_rate": 2.75942559705271e-06,
      "loss": 0.9569,
      "num_input_tokens_seen": 69817095,
      "step": 3283
    },
    {
      "epoch": 0.39487765285877474,
      "flos": 13701895557120.0,
      "grad_norm": 28.637462318558796,
      "learning_rate": 2.7587049093536713e-06,
      "loss": 1.0769,
      "num_input_tokens_seen": 69833145,
      "step": 3284
    },
    {
      "epoch": 0.3949978957494138,
      "flos": 12260221255680.0,
      "grad_norm": 7.134075341299796,
      "learning_rate": 2.757984106559701e-06,
      "loss": 0.9972,
      "num_input_tokens_seen": 69851850,
      "step": 3285
    },
    {
      "epoch": 0.3951181386400529,
      "flos": 25919156060160.0,
      "grad_norm": 4.059324461260409,
      "learning_rate": 2.7572631887801446e-06,
      "loss": 0.9078,
      "num_input_tokens_seen": 69873195,
      "step": 3286
    },
    {
      "epoch": 0.395238381530692,
      "flos": 16427211939840.0,
      "grad_norm": 3.8078375902002812,
      "learning_rate": 2.7565421561243654e-06,
      "loss": 0.9495,
      "num_input_tokens_seen": 69891080,
      "step": 3287
    },
    {
      "epoch": 0.3953586244213311,
      "flos": 17315972751360.0,
      "grad_norm": 6.168252929944075,
      "learning_rate": 2.7558210087017413e-06,
      "loss": 1.0161,
      "num_input_tokens_seen": 69910735,
      "step": 3288
    },
    {
      "epoch": 0.3954788673119702,
      "flos": 16664367575040.0,
      "grad_norm": 4.569493199133119,
      "learning_rate": 2.7550997466216724e-06,
      "loss": 0.9416,
      "num_input_tokens_seen": 69928250,
      "step": 3289
    },
    {
      "epoch": 0.3955991102026093,
      "flos": 12391235174400.0,
      "grad_norm": 3.8785099007599513,
      "learning_rate": 2.7543783699935714e-06,
      "loss": 1.0054,
      "num_input_tokens_seen": 69946000,
      "step": 3290
    },
    {
      "epoch": 0.39571935309324835,
      "flos": 12913573969920.0,
      "grad_norm": 5.703222989847392,
      "learning_rate": 2.753656878926872e-06,
      "loss": 1.0498,
      "num_input_tokens_seen": 69961600,
      "step": 3291
    },
    {
      "epoch": 0.39583959598388746,
      "flos": 12574231572480.0,
      "grad_norm": 11.650688419549038,
      "learning_rate": 2.752935273531023e-06,
      "loss": 0.9312,
      "num_input_tokens_seen": 69979470,
      "step": 3292
    },
    {
      "epoch": 0.39595983887452657,
      "flos": 13726430515200.0,
      "grad_norm": 4.738653987554855,
      "learning_rate": 2.752213553915492e-06,
      "loss": 0.9791,
      "num_input_tokens_seen": 69997545,
      "step": 3293
    },
    {
      "epoch": 0.3960800817651656,
      "flos": 43433239879680.0,
      "grad_norm": 0.821252302692036,
      "learning_rate": 2.751491720189762e-06,
      "loss": 0.8759,
      "num_input_tokens_seen": 70055375,
      "step": 3294
    },
    {
      "epoch": 0.39620032465580474,
      "flos": 11918732759040.0,
      "grad_norm": 4.457750989487563,
      "learning_rate": 2.7507697724633364e-06,
      "loss": 1.1031,
      "num_input_tokens_seen": 70071855,
      "step": 3295
    },
    {
      "epoch": 0.3963205675464438,
      "flos": 49462026240000.0,
      "grad_norm": 0.7820714827791194,
      "learning_rate": 2.7500477108457327e-06,
      "loss": 0.7588,
      "num_input_tokens_seen": 70123585,
      "step": 3296
    },
    {
      "epoch": 0.3964408104370829,
      "flos": 18265361756160.0,
      "grad_norm": 38.58028453099053,
      "learning_rate": 2.7493255354464877e-06,
      "loss": 1.0046,
      "num_input_tokens_seen": 70141115,
      "step": 3297
    },
    {
      "epoch": 0.396561053327722,
      "flos": 17265952419840.0,
      "grad_norm": 4.7298102704116785,
      "learning_rate": 2.748603246375156e-06,
      "loss": 0.9513,
      "num_input_tokens_seen": 70158850,
      "step": 3298
    },
    {
      "epoch": 0.39668129621836107,
      "flos": 14567562362880.0,
      "grad_norm": 4.5731557495368,
      "learning_rate": 2.7478808437413055e-06,
      "loss": 0.8821,
      "num_input_tokens_seen": 70177980,
      "step": 3299
    },
    {
      "epoch": 0.3968015391090002,
      "flos": 19260662845440.0,
      "grad_norm": 7.428885463742084,
      "learning_rate": 2.7471583276545263e-06,
      "loss": 0.8486,
      "num_input_tokens_seen": 70198360,
      "step": 3300
    },
    {
      "epoch": 0.3969217819996393,
      "flos": 8824940236800.0,
      "grad_norm": 5.561228574684527,
      "learning_rate": 2.7464356982244224e-06,
      "loss": 0.8881,
      "num_input_tokens_seen": 70216080,
      "step": 3301
    },
    {
      "epoch": 0.39704202489027834,
      "flos": 47430264053760.0,
      "grad_norm": 0.8938631786364397,
      "learning_rate": 2.745712955560617e-06,
      "loss": 0.8392,
      "num_input_tokens_seen": 70272005,
      "step": 3302
    },
    {
      "epoch": 0.39716226778091746,
      "flos": 12023893401600.0,
      "grad_norm": 20.778210685875734,
      "learning_rate": 2.7449900997727496e-06,
      "loss": 0.973,
      "num_input_tokens_seen": 70289835,
      "step": 3303
    },
    {
      "epoch": 0.39728251067155657,
      "flos": 16691079290880.0,
      "grad_norm": 6.632029420886337,
      "learning_rate": 2.744267130970476e-06,
      "loss": 1.0318,
      "num_input_tokens_seen": 70309280,
      "step": 3304
    },
    {
      "epoch": 0.3974027535621956,
      "flos": 14698422988800.0,
      "grad_norm": 3.5795475826284626,
      "learning_rate": 2.7435440492634697e-06,
      "loss": 0.9474,
      "num_input_tokens_seen": 70328325,
      "step": 3305
    },
    {
      "epoch": 0.39752299645283473,
      "flos": 15301234176000.0,
      "grad_norm": 11.155123955667575,
      "learning_rate": 2.7428208547614228e-06,
      "loss": 0.8546,
      "num_input_tokens_seen": 70347540,
      "step": 3306
    },
    {
      "epoch": 0.39764323934347384,
      "flos": 13623140044800.0,
      "grad_norm": 5.4879159826181185,
      "learning_rate": 2.742097547574043e-06,
      "loss": 0.9512,
      "num_input_tokens_seen": 70365485,
      "step": 3307
    },
    {
      "epoch": 0.3977634822341129,
      "flos": 14803154411520.0,
      "grad_norm": 14.880310136577872,
      "learning_rate": 2.7413741278110544e-06,
      "loss": 0.9658,
      "num_input_tokens_seen": 70383895,
      "step": 3308
    },
    {
      "epoch": 0.397883725124752,
      "flos": 28411946250240.0,
      "grad_norm": 11.759306772780748,
      "learning_rate": 2.7406505955822016e-06,
      "loss": 0.8689,
      "num_input_tokens_seen": 70404640,
      "step": 3309
    },
    {
      "epoch": 0.39800396801539106,
      "flos": 12308218122240.0,
      "grad_norm": 7.677267816846811,
      "learning_rate": 2.7399269509972415e-06,
      "loss": 0.8559,
      "num_input_tokens_seen": 70418515,
      "step": 3310
    },
    {
      "epoch": 0.3981242109060302,
      "flos": 13618418626560.0,
      "grad_norm": 4.706860652380494,
      "learning_rate": 2.7392031941659514e-06,
      "loss": 1.0346,
      "num_input_tokens_seen": 70436080,
      "step": 3311
    },
    {
      "epoch": 0.3982444537966693,
      "flos": 17472104140800.0,
      "grad_norm": 3.659353520357207,
      "learning_rate": 2.7384793251981244e-06,
      "loss": 1.0539,
      "num_input_tokens_seen": 70454785,
      "step": 3312
    },
    {
      "epoch": 0.39836469668730834,
      "flos": 18657943633920.0,
      "grad_norm": 7.871755022237859,
      "learning_rate": 2.737755344203571e-06,
      "loss": 0.9964,
      "num_input_tokens_seen": 70474455,
      "step": 3313
    },
    {
      "epoch": 0.39848493957794745,
      "flos": 19680017756160.0,
      "grad_norm": 4.070229088336794,
      "learning_rate": 2.7370312512921186e-06,
      "loss": 0.9926,
      "num_input_tokens_seen": 70495955,
      "step": 3314
    },
    {
      "epoch": 0.39860518246858656,
      "flos": 8614649610240.0,
      "grad_norm": 5.551195981493644,
      "learning_rate": 2.736307046573611e-06,
      "loss": 0.9545,
      "num_input_tokens_seen": 70511545,
      "step": 3315
    },
    {
      "epoch": 0.3987254253592256,
      "flos": 15901500702720.0,
      "grad_norm": 16.44407377312141,
      "learning_rate": 2.73558273015791e-06,
      "loss": 1.0167,
      "num_input_tokens_seen": 70531095,
      "step": 3316
    },
    {
      "epoch": 0.3988456682498647,
      "flos": 16717637713920.0,
      "grad_norm": 10.55790547319671,
      "learning_rate": 2.734858302154894e-06,
      "loss": 0.8891,
      "num_input_tokens_seen": 70552315,
      "step": 3317
    },
    {
      "epoch": 0.39896591114050384,
      "flos": 13623538606080.0,
      "grad_norm": 5.530551327475394,
      "learning_rate": 2.734133762674457e-06,
      "loss": 0.942,
      "num_input_tokens_seen": 70571625,
      "step": 3318
    },
    {
      "epoch": 0.3990861540311429,
      "flos": 20229558804480.0,
      "grad_norm": 6.463166559253772,
      "learning_rate": 2.7334091118265124e-06,
      "loss": 0.8933,
      "num_input_tokens_seen": 70593240,
      "step": 3319
    },
    {
      "epoch": 0.399206396921782,
      "flos": 44209197957120.0,
      "grad_norm": 0.7310642120531738,
      "learning_rate": 2.732684349720989e-06,
      "loss": 0.7807,
      "num_input_tokens_seen": 70660920,
      "step": 3320
    },
    {
      "epoch": 0.3993266398124211,
      "flos": 19995193098240.0,
      "grad_norm": 9.638840273668253,
      "learning_rate": 2.7319594764678318e-06,
      "loss": 0.9374,
      "num_input_tokens_seen": 70682740,
      "step": 3321
    },
    {
      "epoch": 0.39944688270306017,
      "flos": 16510382284800.0,
      "grad_norm": 4.137830258929067,
      "learning_rate": 2.7312344921770044e-06,
      "loss": 1.0178,
      "num_input_tokens_seen": 70704160,
      "step": 3322
    },
    {
      "epoch": 0.3995671255936993,
      "flos": 13754000670720.0,
      "grad_norm": 7.068199895876347,
      "learning_rate": 2.7305093969584857e-06,
      "loss": 0.9773,
      "num_input_tokens_seen": 70722705,
      "step": 3323
    },
    {
      "epoch": 0.3996873684843384,
      "flos": 16953842933760.0,
      "grad_norm": 4.431812994182547,
      "learning_rate": 2.729784190922272e-06,
      "loss": 0.982,
      "num_input_tokens_seen": 70743860,
      "step": 3324
    },
    {
      "epoch": 0.39980761137497745,
      "flos": 47672325058560.0,
      "grad_norm": 0.8130114852755599,
      "learning_rate": 2.729058874178378e-06,
      "loss": 0.7844,
      "num_input_tokens_seen": 70814260,
      "step": 3325
    },
    {
      "epoch": 0.39992785426561656,
      "flos": 20336313692160.0,
      "grad_norm": 4.512025659204124,
      "learning_rate": 2.7283334468368315e-06,
      "loss": 0.8841,
      "num_input_tokens_seen": 70835260,
      "step": 3326
    },
    {
      "epoch": 0.4000480971562556,
      "flos": 10634293555200.0,
      "grad_norm": 4.990550075630245,
      "learning_rate": 2.72760790900768e-06,
      "loss": 0.9232,
      "num_input_tokens_seen": 70851565,
      "step": 3327
    },
    {
      "epoch": 0.4001683400468947,
      "flos": 17006285291520.0,
      "grad_norm": 5.78157199272233,
      "learning_rate": 2.7268822608009875e-06,
      "loss": 0.9913,
      "num_input_tokens_seen": 70870660,
      "step": 3328
    },
    {
      "epoch": 0.40028858293753383,
      "flos": 17320295608320.0,
      "grad_norm": 5.101349573525958,
      "learning_rate": 2.726156502326834e-06,
      "loss": 0.9685,
      "num_input_tokens_seen": 70891680,
      "step": 3329
    },
    {
      "epoch": 0.4004088258281729,
      "flos": 47822171443200.0,
      "grad_norm": 0.7544702979498837,
      "learning_rate": 2.725430633695316e-06,
      "loss": 0.8235,
      "num_input_tokens_seen": 70954480,
      "step": 3330
    },
    {
      "epoch": 0.400529068718812,
      "flos": 41935014973440.0,
      "grad_norm": 0.9459122260768601,
      "learning_rate": 2.7247046550165485e-06,
      "loss": 0.8164,
      "num_input_tokens_seen": 71006325,
      "step": 3331
    },
    {
      "epoch": 0.4006493116094511,
      "flos": 18057217228800.0,
      "grad_norm": 2.943812790381211,
      "learning_rate": 2.7239785664006606e-06,
      "loss": 0.952,
      "num_input_tokens_seen": 71029585,
      "step": 3332
    },
    {
      "epoch": 0.40076955450009016,
      "flos": 43144622960640.0,
      "grad_norm": 0.8215869268164719,
      "learning_rate": 2.7232523679578002e-06,
      "loss": 0.8611,
      "num_input_tokens_seen": 71092385,
      "step": 3333
    },
    {
      "epoch": 0.4008897973907293,
      "flos": 11760945807360.0,
      "grad_norm": 6.985009354494312,
      "learning_rate": 2.7225260597981295e-06,
      "loss": 0.9912,
      "num_input_tokens_seen": 71109810,
      "step": 3334
    },
    {
      "epoch": 0.4010100402813684,
      "flos": 10869916262400.0,
      "grad_norm": 8.378686349824205,
      "learning_rate": 2.721799642031831e-06,
      "loss": 0.9763,
      "num_input_tokens_seen": 71125700,
      "step": 3335
    },
    {
      "epoch": 0.40113028317200744,
      "flos": 9376290140160.0,
      "grad_norm": 12.056910034907109,
      "learning_rate": 2.721073114769101e-06,
      "loss": 0.9693,
      "num_input_tokens_seen": 71143095,
      "step": 3336
    },
    {
      "epoch": 0.40125052606264655,
      "flos": 14672263127040.0,
      "grad_norm": 8.367643399111095,
      "learning_rate": 2.7203464781201523e-06,
      "loss": 0.948,
      "num_input_tokens_seen": 71162130,
      "step": 3337
    },
    {
      "epoch": 0.40137076895328566,
      "flos": 17503506616320.0,
      "grad_norm": 5.495349131384253,
      "learning_rate": 2.719619732195215e-06,
      "loss": 0.9701,
      "num_input_tokens_seen": 71183490,
      "step": 3338
    },
    {
      "epoch": 0.4014910118439247,
      "flos": 17215288258560.0,
      "grad_norm": 4.4368822984196115,
      "learning_rate": 2.7188928771045377e-06,
      "loss": 0.9298,
      "num_input_tokens_seen": 71204530,
      "step": 3339
    },
    {
      "epoch": 0.4016112547345638,
      "flos": 19025408040960.0,
      "grad_norm": 5.9664318116636865,
      "learning_rate": 2.7181659129583815e-06,
      "loss": 0.998,
      "num_input_tokens_seen": 71223840,
      "step": 3340
    },
    {
      "epoch": 0.4017314976252029,
      "flos": 15117808558080.0,
      "grad_norm": 9.539300839057697,
      "learning_rate": 2.7174388398670276e-06,
      "loss": 0.9611,
      "num_input_tokens_seen": 71242740,
      "step": 3341
    },
    {
      "epoch": 0.401851740515842,
      "flos": 18133979934720.0,
      "grad_norm": 6.358982083416105,
      "learning_rate": 2.716711657940773e-06,
      "loss": 1.1121,
      "num_input_tokens_seen": 71263470,
      "step": 3342
    },
    {
      "epoch": 0.4019719834064811,
      "flos": 39634081505280.0,
      "grad_norm": 0.9398973330779218,
      "learning_rate": 2.7159843672899284e-06,
      "loss": 0.7972,
      "num_input_tokens_seen": 71327390,
      "step": 3343
    },
    {
      "epoch": 0.40209222629712016,
      "flos": 12883305861120.0,
      "grad_norm": 5.226216531430595,
      "learning_rate": 2.715256968024825e-06,
      "loss": 1.0096,
      "num_input_tokens_seen": 71344185,
      "step": 3344
    },
    {
      "epoch": 0.40221246918775927,
      "flos": 18476572139520.0,
      "grad_norm": 4.83212183973035,
      "learning_rate": 2.7145294602558083e-06,
      "loss": 1.0241,
      "num_input_tokens_seen": 71364615,
      "step": 3345
    },
    {
      "epoch": 0.4023327120783984,
      "flos": 24137863434240.0,
      "grad_norm": 6.819606329331668,
      "learning_rate": 2.713801844093241e-06,
      "loss": 0.8899,
      "num_input_tokens_seen": 71385485,
      "step": 3346
    },
    {
      "epoch": 0.40245295496903744,
      "flos": 19150320906240.0,
      "grad_norm": 5.314536568218957,
      "learning_rate": 2.7130741196475014e-06,
      "loss": 1.062,
      "num_input_tokens_seen": 71403335,
      "step": 3347
    },
    {
      "epoch": 0.40257319785967655,
      "flos": 26339645337600.0,
      "grad_norm": 5.5341525463369665,
      "learning_rate": 2.7123462870289848e-06,
      "loss": 0.9857,
      "num_input_tokens_seen": 71423105,
      "step": 3348
    },
    {
      "epoch": 0.40269344075031566,
      "flos": 17320724828160.0,
      "grad_norm": 10.974422194579668,
      "learning_rate": 2.711618346348102e-06,
      "loss": 0.9935,
      "num_input_tokens_seen": 71443350,
      "step": 3349
    },
    {
      "epoch": 0.4028136836409547,
      "flos": 10159737016320.0,
      "grad_norm": 18.440412905491026,
      "learning_rate": 2.7108902977152825e-06,
      "loss": 0.829,
      "num_input_tokens_seen": 71460970,
      "step": 3350
    },
    {
      "epoch": 0.4029339265315938,
      "flos": 18601362370560.0,
      "grad_norm": 7.343002735834081,
      "learning_rate": 2.7101621412409704e-06,
      "loss": 0.9502,
      "num_input_tokens_seen": 71480175,
      "step": 3351
    },
    {
      "epoch": 0.40305416942223293,
      "flos": 16532127313920.0,
      "grad_norm": 6.645612499749453,
      "learning_rate": 2.7094338770356256e-06,
      "loss": 1.0511,
      "num_input_tokens_seen": 71498980,
      "step": 3352
    },
    {
      "epoch": 0.403174412312872,
      "flos": 19287374561280.0,
      "grad_norm": 7.8105774001497785,
      "learning_rate": 2.708705505209726e-06,
      "loss": 0.8377,
      "num_input_tokens_seen": 71519475,
      "step": 3353
    },
    {
      "epoch": 0.4032946552035111,
      "flos": 15453563904000.0,
      "grad_norm": 5.442470514921154,
      "learning_rate": 2.7079770258737646e-06,
      "loss": 1.1019,
      "num_input_tokens_seen": 71537105,
      "step": 3354
    },
    {
      "epoch": 0.4034148980941502,
      "flos": 12282794065920.0,
      "grad_norm": 9.410164357613775,
      "learning_rate": 2.707248439138251e-06,
      "loss": 0.9386,
      "num_input_tokens_seen": 71553060,
      "step": 3355
    },
    {
      "epoch": 0.40353514098478926,
      "flos": 15642324111360.0,
      "grad_norm": 3.91985788944103,
      "learning_rate": 2.7065197451137114e-06,
      "loss": 0.8435,
      "num_input_tokens_seen": 71574160,
      "step": 3356
    },
    {
      "epoch": 0.4036553838754284,
      "flos": 10057182351360.0,
      "grad_norm": 6.782349668721644,
      "learning_rate": 2.7057909439106894e-06,
      "loss": 0.873,
      "num_input_tokens_seen": 71591735,
      "step": 3357
    },
    {
      "epoch": 0.40377562676606743,
      "flos": 17635072389120.0,
      "grad_norm": 11.941900750310792,
      "learning_rate": 2.7050620356397417e-06,
      "loss": 0.9829,
      "num_input_tokens_seen": 71610405,
      "step": 3358
    },
    {
      "epoch": 0.40389586965670654,
      "flos": 17110894080000.0,
      "grad_norm": 5.9559530914693415,
      "learning_rate": 2.7043330204114437e-06,
      "loss": 0.9096,
      "num_input_tokens_seen": 71628835,
      "step": 3359
    },
    {
      "epoch": 0.40401611254734565,
      "flos": 11604691783680.0,
      "grad_norm": 7.220823576389185,
      "learning_rate": 2.7036038983363862e-06,
      "loss": 1.0435,
      "num_input_tokens_seen": 71645160,
      "step": 3360
    },
    {
      "epoch": 0.4041363554379847,
      "flos": 17058421063680.0,
      "grad_norm": 32.818267722246745,
      "learning_rate": 2.702874669525177e-06,
      "loss": 1.0304,
      "num_input_tokens_seen": 71663360,
      "step": 3361
    },
    {
      "epoch": 0.4042565983286238,
      "flos": 20230355927040.0,
      "grad_norm": 5.338080538551247,
      "learning_rate": 2.7021453340884394e-06,
      "loss": 0.8877,
      "num_input_tokens_seen": 71680805,
      "step": 3362
    },
    {
      "epoch": 0.40437684121926293,
      "flos": 12546845368320.0,
      "grad_norm": 7.210041027613716,
      "learning_rate": 2.7014158921368125e-06,
      "loss": 0.9178,
      "num_input_tokens_seen": 71698850,
      "step": 3363
    },
    {
      "epoch": 0.404497084109902,
      "flos": 17079736872960.0,
      "grad_norm": 5.247953470163393,
      "learning_rate": 2.700686343780953e-06,
      "loss": 1.0509,
      "num_input_tokens_seen": 71718440,
      "step": 3364
    },
    {
      "epoch": 0.4046173270005411,
      "flos": 16297761607680.0,
      "grad_norm": 4.8144408436709485,
      "learning_rate": 2.699956689131532e-06,
      "loss": 1.0646,
      "num_input_tokens_seen": 71738145,
      "step": 3365
    },
    {
      "epoch": 0.4047375698911802,
      "flos": 14672355102720.0,
      "grad_norm": 4.484844116768049,
      "learning_rate": 2.699226928299238e-06,
      "loss": 1.0333,
      "num_input_tokens_seen": 71755885,
      "step": 3366
    },
    {
      "epoch": 0.40485781278181926,
      "flos": 20597544407040.0,
      "grad_norm": 17.610501433581824,
      "learning_rate": 2.698497061394774e-06,
      "loss": 0.9674,
      "num_input_tokens_seen": 71774090,
      "step": 3367
    },
    {
      "epoch": 0.40497805567245837,
      "flos": 16454751436800.0,
      "grad_norm": 6.962661889212781,
      "learning_rate": 2.6977670885288627e-06,
      "loss": 0.9975,
      "num_input_tokens_seen": 71795210,
      "step": 3368
    },
    {
      "epoch": 0.4050982985630975,
      "flos": 11525384417280.0,
      "grad_norm": 11.973957412982964,
      "learning_rate": 2.6970370098122378e-06,
      "loss": 0.9483,
      "num_input_tokens_seen": 71811915,
      "step": 3369
    },
    {
      "epoch": 0.40521854145373654,
      "flos": 24583899402240.0,
      "grad_norm": 3.434487566088101,
      "learning_rate": 2.6963068253556535e-06,
      "loss": 1.0492,
      "num_input_tokens_seen": 71833020,
      "step": 3370
    },
    {
      "epoch": 0.40533878434437565,
      "flos": 18022564945920.0,
      "grad_norm": 7.131165363367846,
      "learning_rate": 2.6955765352698763e-06,
      "loss": 1.0466,
      "num_input_tokens_seen": 71852885,
      "step": 3371
    },
    {
      "epoch": 0.40545902723501476,
      "flos": 10607336570880.0,
      "grad_norm": 7.855200688489462,
      "learning_rate": 2.6948461396656923e-06,
      "loss": 0.9216,
      "num_input_tokens_seen": 71870015,
      "step": 3372
    },
    {
      "epoch": 0.4055792701256538,
      "flos": 18160139796480.0,
      "grad_norm": 22.938447272674104,
      "learning_rate": 2.6941156386539013e-06,
      "loss": 0.9372,
      "num_input_tokens_seen": 71889685,
      "step": 3373
    },
    {
      "epoch": 0.4056995130162929,
      "flos": 13886578176000.0,
      "grad_norm": 10.622861041017968,
      "learning_rate": 2.6933850323453203e-06,
      "loss": 0.9819,
      "num_input_tokens_seen": 71907850,
      "step": 3374
    },
    {
      "epoch": 0.405819755906932,
      "flos": 10896106782720.0,
      "grad_norm": 7.509629997149779,
      "learning_rate": 2.6926543208507806e-06,
      "loss": 0.9376,
      "num_input_tokens_seen": 71926250,
      "step": 3375
    },
    {
      "epoch": 0.4059399987975711,
      "flos": 15222662615040.0,
      "grad_norm": 7.924998373037466,
      "learning_rate": 2.6919235042811316e-06,
      "loss": 0.9766,
      "num_input_tokens_seen": 71944755,
      "step": 3376
    },
    {
      "epoch": 0.4060602416882102,
      "flos": 18185809121280.0,
      "grad_norm": 65.35922745321619,
      "learning_rate": 2.691192582747237e-06,
      "loss": 0.9426,
      "num_input_tokens_seen": 71964105,
      "step": 3377
    },
    {
      "epoch": 0.40618048457884925,
      "flos": 16898917232640.0,
      "grad_norm": 7.293182942003931,
      "learning_rate": 2.6904615563599765e-06,
      "loss": 0.9293,
      "num_input_tokens_seen": 71983625,
      "step": 3378
    },
    {
      "epoch": 0.40630072746948837,
      "flos": 12495843962880.0,
      "grad_norm": 4.8519200486579885,
      "learning_rate": 2.6897304252302477e-06,
      "loss": 1.0161,
      "num_input_tokens_seen": 72000665,
      "step": 3379
    },
    {
      "epoch": 0.4064209703601275,
      "flos": 43544808161280.0,
      "grad_norm": 0.8438105738053949,
      "learning_rate": 2.688999189468962e-06,
      "loss": 0.764,
      "num_input_tokens_seen": 72056815,
      "step": 3380
    },
    {
      "epoch": 0.40654121325076653,
      "flos": 17084335656960.0,
      "grad_norm": 6.479345223111952,
      "learning_rate": 2.6882678491870464e-06,
      "loss": 0.94,
      "num_input_tokens_seen": 72076970,
      "step": 3381
    },
    {
      "epoch": 0.40666145614140564,
      "flos": 19470248325120.0,
      "grad_norm": 7.751149984377948,
      "learning_rate": 2.6875364044954453e-06,
      "loss": 0.9019,
      "num_input_tokens_seen": 72096920,
      "step": 3382
    },
    {
      "epoch": 0.40678169903204475,
      "flos": 18631170600960.0,
      "grad_norm": 4.474511120906748,
      "learning_rate": 2.6868048555051185e-06,
      "loss": 1.0117,
      "num_input_tokens_seen": 72118170,
      "step": 3383
    },
    {
      "epoch": 0.4069019419226838,
      "flos": 20389001318400.0,
      "grad_norm": 8.614728921000603,
      "learning_rate": 2.686073202327041e-06,
      "loss": 1.0434,
      "num_input_tokens_seen": 72136890,
      "step": 3384
    },
    {
      "epoch": 0.4070221848133229,
      "flos": 17950186414080.0,
      "grad_norm": 15.535092406205248,
      "learning_rate": 2.6853414450722043e-06,
      "loss": 0.9285,
      "num_input_tokens_seen": 72156275,
      "step": 3385
    },
    {
      "epoch": 0.40714242770396203,
      "flos": 13045936865280.0,
      "grad_norm": 16.145056264902735,
      "learning_rate": 2.684609583851616e-06,
      "loss": 1.0391,
      "num_input_tokens_seen": 72174170,
      "step": 3386
    },
    {
      "epoch": 0.4072626705946011,
      "flos": 21543714263040.0,
      "grad_norm": 5.333211043068446,
      "learning_rate": 2.683877618776297e-06,
      "loss": 0.9948,
      "num_input_tokens_seen": 72196145,
      "step": 3387
    },
    {
      "epoch": 0.4073829134852402,
      "flos": 15510482411520.0,
      "grad_norm": 15.712696767954242,
      "learning_rate": 2.6831455499572876e-06,
      "loss": 0.9353,
      "num_input_tokens_seen": 72213800,
      "step": 3388
    },
    {
      "epoch": 0.40750315637587925,
      "flos": 17971900784640.0,
      "grad_norm": 4.323499059704153,
      "learning_rate": 2.682413377505641e-06,
      "loss": 0.9605,
      "num_input_tokens_seen": 72232325,
      "step": 3389
    },
    {
      "epoch": 0.40762339926651836,
      "flos": 13984840642560.0,
      "grad_norm": 8.687138172546023,
      "learning_rate": 2.6816811015324284e-06,
      "loss": 0.9557,
      "num_input_tokens_seen": 72250095,
      "step": 3390
    },
    {
      "epoch": 0.40774364215715747,
      "flos": 51171798773760.0,
      "grad_norm": 0.8084433595425092,
      "learning_rate": 2.6809487221487343e-06,
      "loss": 0.8122,
      "num_input_tokens_seen": 72309300,
      "step": 3391
    },
    {
      "epoch": 0.4078638850477965,
      "flos": 10659012464640.0,
      "grad_norm": 6.049096460891003,
      "learning_rate": 2.6802162394656605e-06,
      "loss": 1.0077,
      "num_input_tokens_seen": 72325730,
      "step": 3392
    },
    {
      "epoch": 0.40798412793843564,
      "flos": 16954057543680.0,
      "grad_norm": 12.174316788132815,
      "learning_rate": 2.679483653594324e-06,
      "loss": 0.9047,
      "num_input_tokens_seen": 72347220,
      "step": 3393
    },
    {
      "epoch": 0.40810437082907475,
      "flos": 14957630238720.0,
      "grad_norm": 11.298856991374102,
      "learning_rate": 2.678750964645857e-06,
      "loss": 0.9425,
      "num_input_tokens_seen": 72366020,
      "step": 3394
    },
    {
      "epoch": 0.4082246137197138,
      "flos": 7954153451520.0,
      "grad_norm": 27.108446264710906,
      "learning_rate": 2.6780181727314094e-06,
      "loss": 1.0325,
      "num_input_tokens_seen": 72380645,
      "step": 3395
    },
    {
      "epoch": 0.4083448566103529,
      "flos": 13518439280640.0,
      "grad_norm": 7.272065357951103,
      "learning_rate": 2.6772852779621435e-06,
      "loss": 0.969,
      "num_input_tokens_seen": 72398225,
      "step": 3396
    },
    {
      "epoch": 0.408465099500992,
      "flos": 16743644282880.0,
      "grad_norm": 4.244850180130956,
      "learning_rate": 2.676552280449239e-06,
      "loss": 1.0408,
      "num_input_tokens_seen": 72417830,
      "step": 3397
    },
    {
      "epoch": 0.4085853423916311,
      "flos": 8981654138880.0,
      "grad_norm": 9.132104903351708,
      "learning_rate": 2.6758191803038917e-06,
      "loss": 0.9294,
      "num_input_tokens_seen": 72436045,
      "step": 3398
    },
    {
      "epoch": 0.4087055852822702,
      "flos": 17320632852480.0,
      "grad_norm": 3.8787936437267754,
      "learning_rate": 2.6750859776373125e-06,
      "loss": 1.0204,
      "num_input_tokens_seen": 72455220,
      "step": 3399
    },
    {
      "epoch": 0.4088258281729093,
      "flos": 47535669964800.0,
      "grad_norm": 0.8371063688066683,
      "learning_rate": 2.674352672560727e-06,
      "loss": 0.8067,
      "num_input_tokens_seen": 72516385,
      "step": 3400
    },
    {
      "epoch": 0.40894607106354836,
      "flos": 14515058688000.0,
      "grad_norm": 4.7057297296177145,
      "learning_rate": 2.673619265185377e-06,
      "loss": 0.9672,
      "num_input_tokens_seen": 72535945,
      "step": 3401
    },
    {
      "epoch": 0.40906631395418747,
      "flos": 19261551943680.0,
      "grad_norm": 5.353268895662041,
      "learning_rate": 2.672885755622521e-06,
      "loss": 0.9669,
      "num_input_tokens_seen": 72558080,
      "step": 3402
    },
    {
      "epoch": 0.4091865568448266,
      "flos": 18133520056320.0,
      "grad_norm": 8.66206907931721,
      "learning_rate": 2.67215214398343e-06,
      "loss": 0.895,
      "num_input_tokens_seen": 72577815,
      "step": 3403
    },
    {
      "epoch": 0.40930679973546563,
      "flos": 20414333399040.0,
      "grad_norm": 27.93538512688701,
      "learning_rate": 2.671418430379393e-06,
      "loss": 0.9661,
      "num_input_tokens_seen": 72596220,
      "step": 3404
    },
    {
      "epoch": 0.40942704262610474,
      "flos": 14829038346240.0,
      "grad_norm": 12.37846988620293,
      "learning_rate": 2.670684614921715e-06,
      "loss": 1.0073,
      "num_input_tokens_seen": 72614915,
      "step": 3405
    },
    {
      "epoch": 0.4095472855167438,
      "flos": 15355240120320.0,
      "grad_norm": 6.175320129745015,
      "learning_rate": 2.6699506977217128e-06,
      "loss": 0.8764,
      "num_input_tokens_seen": 72634810,
      "step": 3406
    },
    {
      "epoch": 0.4096675284073829,
      "flos": 19888162283520.0,
      "grad_norm": 6.314155588122265,
      "learning_rate": 2.6692166788907233e-06,
      "loss": 0.8751,
      "num_input_tokens_seen": 72654725,
      "step": 3407
    },
    {
      "epoch": 0.409787771298022,
      "flos": 13621699092480.0,
      "grad_norm": 6.181974776217472,
      "learning_rate": 2.6684825585400957e-06,
      "loss": 0.9611,
      "num_input_tokens_seen": 72673390,
      "step": 3408
    },
    {
      "epoch": 0.4099080141886611,
      "flos": 42419045007360.0,
      "grad_norm": 0.9469089111701646,
      "learning_rate": 2.6677483367811947e-06,
      "loss": 0.9155,
      "num_input_tokens_seen": 72733150,
      "step": 3409
    },
    {
      "epoch": 0.4100282570793002,
      "flos": 15562005012480.0,
      "grad_norm": 16.439706400456277,
      "learning_rate": 2.6670140137254028e-06,
      "loss": 0.9539,
      "num_input_tokens_seen": 72752345,
      "step": 3410
    },
    {
      "epoch": 0.4101484999699393,
      "flos": 13151189483520.0,
      "grad_norm": 38.113857554639004,
      "learning_rate": 2.666279589484115e-06,
      "loss": 1.0662,
      "num_input_tokens_seen": 72769965,
      "step": 3411
    },
    {
      "epoch": 0.41026874286057835,
      "flos": 13540797480960.0,
      "grad_norm": 4.7272297033606225,
      "learning_rate": 2.6655450641687435e-06,
      "loss": 0.9875,
      "num_input_tokens_seen": 72787250,
      "step": 3412
    },
    {
      "epoch": 0.41038898575121746,
      "flos": 22247884431360.0,
      "grad_norm": 5.8152023104426,
      "learning_rate": 2.664810437890715e-06,
      "loss": 0.8815,
      "num_input_tokens_seen": 72808640,
      "step": 3413
    },
    {
      "epoch": 0.41050922864185657,
      "flos": 10135815229440.0,
      "grad_norm": 7.589860335963607,
      "learning_rate": 2.6640757107614714e-06,
      "loss": 0.9899,
      "num_input_tokens_seen": 72826455,
      "step": 3414
    },
    {
      "epoch": 0.4106294715324956,
      "flos": 22066145034240.0,
      "grad_norm": 13.155086076223583,
      "learning_rate": 2.6633408828924697e-06,
      "loss": 0.9005,
      "num_input_tokens_seen": 72845040,
      "step": 3415
    },
    {
      "epoch": 0.41074971442313474,
      "flos": 17394850897920.0,
      "grad_norm": 11.567151592837352,
      "learning_rate": 2.662605954395185e-06,
      "loss": 0.8957,
      "num_input_tokens_seen": 72864720,
      "step": 3416
    },
    {
      "epoch": 0.41086995731377385,
      "flos": 15300988907520.0,
      "grad_norm": 3.5416706907960056,
      "learning_rate": 2.6618709253811027e-06,
      "loss": 1.0336,
      "num_input_tokens_seen": 72883895,
      "step": 3417
    },
    {
      "epoch": 0.4109902002044129,
      "flos": 14696982036480.0,
      "grad_norm": 3.118795262315984,
      "learning_rate": 2.6611357959617277e-06,
      "loss": 1.0722,
      "num_input_tokens_seen": 72903235,
      "step": 3418
    },
    {
      "epoch": 0.411110443095052,
      "flos": 12883213885440.0,
      "grad_norm": 5.922317528608878,
      "learning_rate": 2.660400566248578e-06,
      "loss": 1.0901,
      "num_input_tokens_seen": 72921080,
      "step": 3419
    },
    {
      "epoch": 0.41123068598569107,
      "flos": 10293540864000.0,
      "grad_norm": 7.0134396282210325,
      "learning_rate": 2.6596652363531876e-06,
      "loss": 0.8658,
      "num_input_tokens_seen": 72936675,
      "step": 3420
    },
    {
      "epoch": 0.4113509288763302,
      "flos": 15038654484480.0,
      "grad_norm": 5.283314859897475,
      "learning_rate": 2.6589298063871055e-06,
      "loss": 0.9787,
      "num_input_tokens_seen": 72956570,
      "step": 3421
    },
    {
      "epoch": 0.4114711717669693,
      "flos": 13072587264000.0,
      "grad_norm": 4.101259754451835,
      "learning_rate": 2.658194276461895e-06,
      "loss": 0.8898,
      "num_input_tokens_seen": 72974215,
      "step": 3422
    },
    {
      "epoch": 0.41159141465760835,
      "flos": 19390389104640.0,
      "grad_norm": 20.806966143286978,
      "learning_rate": 2.6574586466891368e-06,
      "loss": 0.8698,
      "num_input_tokens_seen": 72994410,
      "step": 3423
    },
    {
      "epoch": 0.41171165754824746,
      "flos": 14196510904320.0,
      "grad_norm": 6.413677691905365,
      "learning_rate": 2.6567229171804247e-06,
      "loss": 0.825,
      "num_input_tokens_seen": 73012015,
      "step": 3424
    },
    {
      "epoch": 0.41183190043888657,
      "flos": 12757871800320.0,
      "grad_norm": 73.63416817385968,
      "learning_rate": 2.655987088047368e-06,
      "loss": 1.0648,
      "num_input_tokens_seen": 73030080,
      "step": 3425
    },
    {
      "epoch": 0.4119521433295256,
      "flos": 19340460748800.0,
      "grad_norm": 11.30133584827769,
      "learning_rate": 2.6552511594015912e-06,
      "loss": 0.9674,
      "num_input_tokens_seen": 73050190,
      "step": 3426
    },
    {
      "epoch": 0.41207238622016473,
      "flos": 10686429327360.0,
      "grad_norm": 12.045387898648439,
      "learning_rate": 2.654515131354735e-06,
      "loss": 1.0554,
      "num_input_tokens_seen": 73068175,
      "step": 3427
    },
    {
      "epoch": 0.41219262911080384,
      "flos": 19260908113920.0,
      "grad_norm": 4.83932470166192,
      "learning_rate": 2.653779004018453e-06,
      "loss": 1.0501,
      "num_input_tokens_seen": 73088460,
      "step": 3428
    },
    {
      "epoch": 0.4123128720014429,
      "flos": 17560823685120.0,
      "grad_norm": 4.227425703933233,
      "learning_rate": 2.653042777504417e-06,
      "loss": 1.0139,
      "num_input_tokens_seen": 73110770,
      "step": 3429
    },
    {
      "epoch": 0.412433114892082,
      "flos": 18679412736000.0,
      "grad_norm": 10.217845071743662,
      "learning_rate": 2.6523064519243105e-06,
      "loss": 0.994,
      "num_input_tokens_seen": 73130060,
      "step": 3430
    },
    {
      "epoch": 0.4125533577827211,
      "flos": 15170281574400.0,
      "grad_norm": 11.810245036297207,
      "learning_rate": 2.6515700273898333e-06,
      "loss": 0.9757,
      "num_input_tokens_seen": 73147655,
      "step": 3431
    },
    {
      "epoch": 0.4126736006733602,
      "flos": 18552752332800.0,
      "grad_norm": 5.998700954954552,
      "learning_rate": 2.6508335040127018e-06,
      "loss": 0.8844,
      "num_input_tokens_seen": 73167070,
      "step": 3432
    },
    {
      "epoch": 0.4127938435639993,
      "flos": 18264258048000.0,
      "grad_norm": 3.901479875562783,
      "learning_rate": 2.6500968819046446e-06,
      "loss": 0.9614,
      "num_input_tokens_seen": 73187090,
      "step": 3433
    },
    {
      "epoch": 0.4129140864546384,
      "flos": 12750697697280.0,
      "grad_norm": 7.310507217532029,
      "learning_rate": 2.649360161177408e-06,
      "loss": 0.771,
      "num_input_tokens_seen": 73201870,
      "step": 3434
    },
    {
      "epoch": 0.41303432934527745,
      "flos": 16874934128640.0,
      "grad_norm": 11.837280116712767,
      "learning_rate": 2.6486233419427504e-06,
      "loss": 0.9244,
      "num_input_tokens_seen": 73221405,
      "step": 3435
    },
    {
      "epoch": 0.41315457223591656,
      "flos": 14016212459520.0,
      "grad_norm": 7.058314340589147,
      "learning_rate": 2.6478864243124484e-06,
      "loss": 0.9425,
      "num_input_tokens_seen": 73240790,
      "step": 3436
    },
    {
      "epoch": 0.4132748151265556,
      "flos": 14855719403520.0,
      "grad_norm": 11.622070599700127,
      "learning_rate": 2.6471494083982903e-06,
      "loss": 1.0459,
      "num_input_tokens_seen": 73259895,
      "step": 3437
    },
    {
      "epoch": 0.4133950580171947,
      "flos": 22983242465280.0,
      "grad_norm": 5.013976970167797,
      "learning_rate": 2.6464122943120818e-06,
      "loss": 0.9448,
      "num_input_tokens_seen": 73279840,
      "step": 3438
    },
    {
      "epoch": 0.41351530090783384,
      "flos": 16558072565760.0,
      "grad_norm": 41.10612337482397,
      "learning_rate": 2.645675082165642e-06,
      "loss": 1.0214,
      "num_input_tokens_seen": 73295770,
      "step": 3439
    },
    {
      "epoch": 0.4136355437984729,
      "flos": 18212152934400.0,
      "grad_norm": 9.138984311219856,
      "learning_rate": 2.644937772070806e-06,
      "loss": 0.9561,
      "num_input_tokens_seen": 73313935,
      "step": 3440
    },
    {
      "epoch": 0.413755786689112,
      "flos": 14068961402880.0,
      "grad_norm": 7.6879976933343865,
      "learning_rate": 2.6442003641394225e-06,
      "loss": 1.0279,
      "num_input_tokens_seen": 73331250,
      "step": 3441
    },
    {
      "epoch": 0.4138760295797511,
      "flos": 19129802219520.0,
      "grad_norm": 2.873161557396295,
      "learning_rate": 2.643462858483356e-06,
      "loss": 1.0307,
      "num_input_tokens_seen": 73351255,
      "step": 3442
    },
    {
      "epoch": 0.41399627247039017,
      "flos": 11604385198080.0,
      "grad_norm": 6.781147336402451,
      "learning_rate": 2.6427252552144856e-06,
      "loss": 0.9342,
      "num_input_tokens_seen": 73369625,
      "step": 3443
    },
    {
      "epoch": 0.4141165153610293,
      "flos": 16298068193280.0,
      "grad_norm": 5.663196660653445,
      "learning_rate": 2.6419875544447044e-06,
      "loss": 0.9468,
      "num_input_tokens_seen": 73390745,
      "step": 3444
    },
    {
      "epoch": 0.4142367582516684,
      "flos": 17923995893760.0,
      "grad_norm": 4.680875445916666,
      "learning_rate": 2.6412497562859218e-06,
      "loss": 0.9046,
      "num_input_tokens_seen": 73411745,
      "step": 3445
    },
    {
      "epoch": 0.41435700114230745,
      "flos": 15407191941120.0,
      "grad_norm": 5.989884432535967,
      "learning_rate": 2.6405118608500617e-06,
      "loss": 0.9581,
      "num_input_tokens_seen": 73430290,
      "step": 3446
    },
    {
      "epoch": 0.41447724403294656,
      "flos": 18500800512000.0,
      "grad_norm": 5.9359218708665,
      "learning_rate": 2.6397738682490613e-06,
      "loss": 0.9976,
      "num_input_tokens_seen": 73450910,
      "step": 3447
    },
    {
      "epoch": 0.41459748692358567,
      "flos": 12941389393920.0,
      "grad_norm": 5.051601971638146,
      "learning_rate": 2.6390357785948734e-06,
      "loss": 0.9448,
      "num_input_tokens_seen": 73467745,
      "step": 3448
    },
    {
      "epoch": 0.4147177298142247,
      "flos": 17187656785920.0,
      "grad_norm": 4.405202971165243,
      "learning_rate": 2.6382975919994667e-06,
      "loss": 0.994,
      "num_input_tokens_seen": 73488040,
      "step": 3449
    },
    {
      "epoch": 0.41483797270486383,
      "flos": 14253061509120.0,
      "grad_norm": 2.958534574156106,
      "learning_rate": 2.637559308574822e-06,
      "loss": 0.9117,
      "num_input_tokens_seen": 73507505,
      "step": 3450
    },
    {
      "epoch": 0.4149582155955029,
      "flos": 21437173985280.0,
      "grad_norm": 5.514179850902633,
      "learning_rate": 2.6368209284329376e-06,
      "loss": 0.91,
      "num_input_tokens_seen": 73527855,
      "step": 3451
    },
    {
      "epoch": 0.415078458486142,
      "flos": 11866842255360.0,
      "grad_norm": 6.003671158783807,
      "learning_rate": 2.636082451685825e-06,
      "loss": 0.9391,
      "num_input_tokens_seen": 73545775,
      "step": 3452
    },
    {
      "epoch": 0.4151987013767811,
      "flos": 18528769228800.0,
      "grad_norm": 4.351240985162959,
      "learning_rate": 2.6353438784455094e-06,
      "loss": 1.0533,
      "num_input_tokens_seen": 73568780,
      "step": 3453
    },
    {
      "epoch": 0.41531894426742016,
      "flos": 17503782543360.0,
      "grad_norm": 6.242076887122273,
      "learning_rate": 2.6346052088240326e-06,
      "loss": 0.8979,
      "num_input_tokens_seen": 73588020,
      "step": 3454
    },
    {
      "epoch": 0.4154391871580593,
      "flos": 10581299343360.0,
      "grad_norm": 6.891989229057694,
      "learning_rate": 2.63386644293345e-06,
      "loss": 0.9657,
      "num_input_tokens_seen": 73604085,
      "step": 3455
    },
    {
      "epoch": 0.4155594300486984,
      "flos": 10345094123520.0,
      "grad_norm": 15.020652286674753,
      "learning_rate": 2.633127580885833e-06,
      "loss": 1.0237,
      "num_input_tokens_seen": 73618305,
      "step": 3456
    },
    {
      "epoch": 0.41567967293933744,
      "flos": 21017635123200.0,
      "grad_norm": 4.931816403117376,
      "learning_rate": 2.632388622793265e-06,
      "loss": 0.8373,
      "num_input_tokens_seen": 73637180,
      "step": 3457
    },
    {
      "epoch": 0.41579991582997655,
      "flos": 13643904000000.0,
      "grad_norm": 5.705946634507688,
      "learning_rate": 2.6316495687678457e-06,
      "loss": 0.8754,
      "num_input_tokens_seen": 73655550,
      "step": 3458
    },
    {
      "epoch": 0.41592015872061566,
      "flos": 17398897827840.0,
      "grad_norm": 6.471288505441115,
      "learning_rate": 2.6309104189216887e-06,
      "loss": 0.9445,
      "num_input_tokens_seen": 73672835,
      "step": 3459
    },
    {
      "epoch": 0.4160404016112547,
      "flos": 14749363077120.0,
      "grad_norm": 3.6184148498189956,
      "learning_rate": 2.630171173366923e-06,
      "loss": 0.9385,
      "num_input_tokens_seen": 73688355,
      "step": 3460
    },
    {
      "epoch": 0.41616064450189383,
      "flos": 9819812106240.0,
      "grad_norm": 8.318799836821274,
      "learning_rate": 2.629431832215691e-06,
      "loss": 0.9351,
      "num_input_tokens_seen": 73702880,
      "step": 3461
    },
    {
      "epoch": 0.41628088739253294,
      "flos": 14199883345920.0,
      "grad_norm": 5.651541926541901,
      "learning_rate": 2.628692395580151e-06,
      "loss": 1.0524,
      "num_input_tokens_seen": 73722690,
      "step": 3462
    },
    {
      "epoch": 0.416401130283172,
      "flos": 20781552537600.0,
      "grad_norm": 5.208385111233514,
      "learning_rate": 2.6279528635724747e-06,
      "loss": 0.9805,
      "num_input_tokens_seen": 73742565,
      "step": 3463
    },
    {
      "epoch": 0.4165213731738111,
      "flos": 11493951283200.0,
      "grad_norm": 10.277676317785264,
      "learning_rate": 2.627213236304848e-06,
      "loss": 0.9753,
      "num_input_tokens_seen": 73759085,
      "step": 3464
    },
    {
      "epoch": 0.4166416160644502,
      "flos": 24084991856640.0,
      "grad_norm": 16.37714773755119,
      "learning_rate": 2.626473513889472e-06,
      "loss": 0.8931,
      "num_input_tokens_seen": 73781185,
      "step": 3465
    },
    {
      "epoch": 0.41676185895508927,
      "flos": 14850752716800.0,
      "grad_norm": 9.396913679199992,
      "learning_rate": 2.625733696438562e-06,
      "loss": 1.0201,
      "num_input_tokens_seen": 73798410,
      "step": 3466
    },
    {
      "epoch": 0.4168821018457284,
      "flos": 13046703329280.0,
      "grad_norm": 5.29442978855029,
      "learning_rate": 2.6249937840643476e-06,
      "loss": 0.949,
      "num_input_tokens_seen": 73816435,
      "step": 3467
    },
    {
      "epoch": 0.41700234473636744,
      "flos": 13256810004480.0,
      "grad_norm": 5.000448016122003,
      "learning_rate": 2.6242537768790733e-06,
      "loss": 0.8583,
      "num_input_tokens_seen": 73835310,
      "step": 3468
    },
    {
      "epoch": 0.41712258762700655,
      "flos": 22122021150720.0,
      "grad_norm": 9.654760943416036,
      "learning_rate": 2.6235136749949975e-06,
      "loss": 0.8829,
      "num_input_tokens_seen": 73858480,
      "step": 3469
    },
    {
      "epoch": 0.41724283051764566,
      "flos": 25629220823040.0,
      "grad_norm": 6.667886940220162,
      "learning_rate": 2.6227734785243924e-06,
      "loss": 0.8048,
      "num_input_tokens_seen": 73878160,
      "step": 3470
    },
    {
      "epoch": 0.4173630734082847,
      "flos": 18024741703680.0,
      "grad_norm": 7.7426447266603455,
      "learning_rate": 2.6220331875795466e-06,
      "loss": 0.9861,
      "num_input_tokens_seen": 73897230,
      "step": 3471
    },
    {
      "epoch": 0.4174833162989238,
      "flos": 18996458250240.0,
      "grad_norm": 11.673117534473505,
      "learning_rate": 2.62129280227276e-06,
      "loss": 0.9478,
      "num_input_tokens_seen": 73916950,
      "step": 3472
    },
    {
      "epoch": 0.41760355918956293,
      "flos": 53715966382080.0,
      "grad_norm": 9.122442017588153,
      "learning_rate": 2.62055232271635e-06,
      "loss": 0.8723,
      "num_input_tokens_seen": 73943855,
      "step": 3473
    },
    {
      "epoch": 0.417723802080202,
      "flos": 10107999805440.0,
      "grad_norm": 8.117670127465136,
      "learning_rate": 2.619811749022646e-06,
      "loss": 1.0794,
      "num_input_tokens_seen": 73958885,
      "step": 3474
    },
    {
      "epoch": 0.4178440449708411,
      "flos": 10342212218880.0,
      "grad_norm": 14.140717800109856,
      "learning_rate": 2.6190710813039917e-06,
      "loss": 0.9038,
      "num_input_tokens_seen": 73971730,
      "step": 3475
    },
    {
      "epoch": 0.4179642878614802,
      "flos": 15275442216960.0,
      "grad_norm": 7.240057869780025,
      "learning_rate": 2.618330319672747e-06,
      "loss": 1.0128,
      "num_input_tokens_seen": 73990870,
      "step": 3476
    },
    {
      "epoch": 0.41808453075211927,
      "flos": 13072403312640.0,
      "grad_norm": 6.362647861643712,
      "learning_rate": 2.617589464241284e-06,
      "loss": 1.1179,
      "num_input_tokens_seen": 74004990,
      "step": 3477
    },
    {
      "epoch": 0.4182047736427584,
      "flos": 14408211824640.0,
      "grad_norm": 6.1163755148352825,
      "learning_rate": 2.6168485151219914e-06,
      "loss": 0.9204,
      "num_input_tokens_seen": 74024330,
      "step": 3478
    },
    {
      "epoch": 0.4183250165333975,
      "flos": 13384328847360.0,
      "grad_norm": 7.368773854923219,
      "learning_rate": 2.616107472427269e-06,
      "loss": 0.9039,
      "num_input_tokens_seen": 74038745,
      "step": 3479
    },
    {
      "epoch": 0.41844525942403654,
      "flos": 12566750883840.0,
      "grad_norm": 7.486982365261998,
      "learning_rate": 2.615366336269533e-06,
      "loss": 0.9598,
      "num_input_tokens_seen": 74052130,
      "step": 3480
    },
    {
      "epoch": 0.41856550231467565,
      "flos": 13013430681600.0,
      "grad_norm": 8.334110935806653,
      "learning_rate": 2.6146251067612126e-06,
      "loss": 0.978,
      "num_input_tokens_seen": 74067325,
      "step": 3481
    },
    {
      "epoch": 0.41868574520531476,
      "flos": 16191068037120.0,
      "grad_norm": 5.38197477380007,
      "learning_rate": 2.6138837840147525e-06,
      "loss": 1.0111,
      "num_input_tokens_seen": 74086080,
      "step": 3482
    },
    {
      "epoch": 0.4188059880959538,
      "flos": 9663864668160.0,
      "grad_norm": 10.458714654260259,
      "learning_rate": 2.6131423681426103e-06,
      "loss": 0.9644,
      "num_input_tokens_seen": 74101715,
      "step": 3483
    },
    {
      "epoch": 0.41892623098659293,
      "flos": 26998087372800.0,
      "grad_norm": 4.27791983848824,
      "learning_rate": 2.6124008592572587e-06,
      "loss": 0.9175,
      "num_input_tokens_seen": 74125420,
      "step": 3484
    },
    {
      "epoch": 0.419046473877232,
      "flos": 16534304071680.0,
      "grad_norm": 5.409617351037276,
      "learning_rate": 2.6116592574711835e-06,
      "loss": 0.9928,
      "num_input_tokens_seen": 74143440,
      "step": 3485
    },
    {
      "epoch": 0.4191667167678711,
      "flos": 14724736143360.0,
      "grad_norm": 6.988972288010457,
      "learning_rate": 2.6109175628968853e-06,
      "loss": 1.0214,
      "num_input_tokens_seen": 74162925,
      "step": 3486
    },
    {
      "epoch": 0.4192869596585102,
      "flos": 16769589534720.0,
      "grad_norm": 6.144605128092104,
      "learning_rate": 2.610175775646878e-06,
      "loss": 1.0256,
      "num_input_tokens_seen": 74181225,
      "step": 3487
    },
    {
      "epoch": 0.41940720254914926,
      "flos": 17843768770560.0,
      "grad_norm": 6.138118164257566,
      "learning_rate": 2.6094338958336907e-06,
      "loss": 0.9406,
      "num_input_tokens_seen": 74199615,
      "step": 3488
    },
    {
      "epoch": 0.41952744543978837,
      "flos": 10996208762880.0,
      "grad_norm": 4.3429623994405135,
      "learning_rate": 2.608691923569867e-06,
      "loss": 1.0049,
      "num_input_tokens_seen": 74216210,
      "step": 3489
    },
    {
      "epoch": 0.4196476883304275,
      "flos": 17530248990720.0,
      "grad_norm": 4.2327195200630126,
      "learning_rate": 2.6079498589679616e-06,
      "loss": 0.9527,
      "num_input_tokens_seen": 74237020,
      "step": 3490
    },
    {
      "epoch": 0.41976793122106654,
      "flos": 17448182353920.0,
      "grad_norm": 8.769672187063476,
      "learning_rate": 2.6072077021405465e-06,
      "loss": 0.9463,
      "num_input_tokens_seen": 74255575,
      "step": 3491
    },
    {
      "epoch": 0.41988817411170565,
      "flos": 15036355092480.0,
      "grad_norm": 4.5899044211415605,
      "learning_rate": 2.6064654532002054e-06,
      "loss": 0.8885,
      "num_input_tokens_seen": 74274305,
      "step": 3492
    },
    {
      "epoch": 0.42000841700234476,
      "flos": 22564255457280.0,
      "grad_norm": 4.4597995996997915,
      "learning_rate": 2.6057231122595375e-06,
      "loss": 0.9489,
      "num_input_tokens_seen": 74295335,
      "step": 3493
    },
    {
      "epoch": 0.4201286598929838,
      "flos": 15112259358720.0,
      "grad_norm": 4.541908094422827,
      "learning_rate": 2.604980679431154e-06,
      "loss": 0.9274,
      "num_input_tokens_seen": 74313295,
      "step": 3494
    },
    {
      "epoch": 0.4202489027836229,
      "flos": 13146866626560.0,
      "grad_norm": 4.798182405618091,
      "learning_rate": 2.604238154827684e-06,
      "loss": 0.9546,
      "num_input_tokens_seen": 74329640,
      "step": 3495
    },
    {
      "epoch": 0.42036914567426203,
      "flos": 13701711605760.0,
      "grad_norm": 6.438593799293809,
      "learning_rate": 2.6034955385617656e-06,
      "loss": 0.9138,
      "num_input_tokens_seen": 74347690,
      "step": 3496
    },
    {
      "epoch": 0.4204893885649011,
      "flos": 48580746117120.0,
      "grad_norm": 0.7689424382457156,
      "learning_rate": 2.6027528307460544e-06,
      "loss": 0.8548,
      "num_input_tokens_seen": 74411415,
      "step": 3497
    },
    {
      "epoch": 0.4206096314555402,
      "flos": 15563997818880.0,
      "grad_norm": 17.14048879703178,
      "learning_rate": 2.602010031493217e-06,
      "loss": 1.0489,
      "num_input_tokens_seen": 74429365,
      "step": 3498
    },
    {
      "epoch": 0.42072987434617926,
      "flos": 20860154757120.0,
      "grad_norm": 4.815308581136559,
      "learning_rate": 2.6012671409159367e-06,
      "loss": 1.0526,
      "num_input_tokens_seen": 74450420,
      "step": 3499
    },
    {
      "epoch": 0.42085011723681837,
      "flos": 19654716334080.0,
      "grad_norm": 8.304418780195004,
      "learning_rate": 2.6005241591269097e-06,
      "loss": 1.0237,
      "num_input_tokens_seen": 74469510,
      "step": 3500
    },
    {
      "epoch": 0.4209703601274575,
      "flos": 19811184967680.0,
      "grad_norm": 4.92757896634894,
      "learning_rate": 2.5997810862388454e-06,
      "loss": 0.9915,
      "num_input_tokens_seen": 74489070,
      "step": 3501
    },
    {
      "epoch": 0.42109060301809653,
      "flos": 19600373145600.0,
      "grad_norm": 8.095436656878203,
      "learning_rate": 2.599037922364467e-06,
      "loss": 0.954,
      "num_input_tokens_seen": 74507690,
      "step": 3502
    },
    {
      "epoch": 0.42121084590873564,
      "flos": 20886038691840.0,
      "grad_norm": 5.549447164408939,
      "learning_rate": 2.5982946676165112e-06,
      "loss": 0.948,
      "num_input_tokens_seen": 74527180,
      "step": 3503
    },
    {
      "epoch": 0.42133108879937475,
      "flos": 48260726722560.0,
      "grad_norm": 0.7719521951105196,
      "learning_rate": 2.5975513221077313e-06,
      "loss": 0.7933,
      "num_input_tokens_seen": 74590870,
      "step": 3504
    },
    {
      "epoch": 0.4214513316900138,
      "flos": 16423532912640.0,
      "grad_norm": 11.04549398247514,
      "learning_rate": 2.5968078859508897e-06,
      "loss": 1.0731,
      "num_input_tokens_seen": 74607790,
      "step": 3505
    },
    {
      "epoch": 0.4215715745806529,
      "flos": 10839709470720.0,
      "grad_norm": 4.587079453486462,
      "learning_rate": 2.5960643592587673e-06,
      "loss": 0.9899,
      "num_input_tokens_seen": 74624920,
      "step": 3506
    },
    {
      "epoch": 0.42169181747129203,
      "flos": 15720956989440.0,
      "grad_norm": 4.093854597080377,
      "learning_rate": 2.5953207421441553e-06,
      "loss": 1.0161,
      "num_input_tokens_seen": 74643240,
      "step": 3507
    },
    {
      "epoch": 0.4218120603619311,
      "flos": 16081799147520.0,
      "grad_norm": 5.698240764915203,
      "learning_rate": 2.5945770347198603e-06,
      "loss": 0.9357,
      "num_input_tokens_seen": 74661115,
      "step": 3508
    },
    {
      "epoch": 0.4219323032525702,
      "flos": 13964199321600.0,
      "grad_norm": 6.007656616252199,
      "learning_rate": 2.593833237098701e-06,
      "loss": 1.0206,
      "num_input_tokens_seen": 74678435,
      "step": 3509
    },
    {
      "epoch": 0.4220525461432093,
      "flos": 21516052131840.0,
      "grad_norm": 7.630732648279045,
      "learning_rate": 2.593089349393512e-06,
      "loss": 0.8276,
      "num_input_tokens_seen": 74698645,
      "step": 3510
    },
    {
      "epoch": 0.42217278903384836,
      "flos": 17293859819520.0,
      "grad_norm": 5.090175006008819,
      "learning_rate": 2.592345371717141e-06,
      "loss": 1.0256,
      "num_input_tokens_seen": 74717895,
      "step": 3511
    },
    {
      "epoch": 0.42229303192448747,
      "flos": 12102587596800.0,
      "grad_norm": 16.7218236076579,
      "learning_rate": 2.591601304182448e-06,
      "loss": 0.8891,
      "num_input_tokens_seen": 74735585,
      "step": 3512
    },
    {
      "epoch": 0.4224132748151266,
      "flos": 16193398087680.0,
      "grad_norm": 4.014131141987422,
      "learning_rate": 2.5908571469023067e-06,
      "loss": 0.9865,
      "num_input_tokens_seen": 74754790,
      "step": 3513
    },
    {
      "epoch": 0.42253351770576564,
      "flos": 12624374538240.0,
      "grad_norm": 5.195177578428973,
      "learning_rate": 2.5901128999896067e-06,
      "loss": 0.9512,
      "num_input_tokens_seen": 74769940,
      "step": 3514
    },
    {
      "epoch": 0.42265376059640475,
      "flos": 20309847244800.0,
      "grad_norm": 7.578053519314941,
      "learning_rate": 2.5893685635572487e-06,
      "loss": 0.8834,
      "num_input_tokens_seen": 74790510,
      "step": 3515
    },
    {
      "epoch": 0.4227740034870438,
      "flos": 11499531141120.0,
      "grad_norm": 4.269465802834958,
      "learning_rate": 2.5886241377181483e-06,
      "loss": 0.8781,
      "num_input_tokens_seen": 74809100,
      "step": 3516
    },
    {
      "epoch": 0.4228942463776829,
      "flos": 17997263523840.0,
      "grad_norm": 29.681155777711577,
      "learning_rate": 2.587879622585234e-06,
      "loss": 0.9953,
      "num_input_tokens_seen": 74827420,
      "step": 3517
    },
    {
      "epoch": 0.423014489268322,
      "flos": 18788620308480.0,
      "grad_norm": 7.050713620693664,
      "learning_rate": 2.5871350182714486e-06,
      "loss": 0.9441,
      "num_input_tokens_seen": 74848020,
      "step": 3518
    },
    {
      "epoch": 0.4231347321589611,
      "flos": 12232927027200.0,
      "grad_norm": 9.523094214870131,
      "learning_rate": 2.586390324889748e-06,
      "loss": 0.9992,
      "num_input_tokens_seen": 74863640,
      "step": 3519
    },
    {
      "epoch": 0.4232549750496002,
      "flos": 16347475353600.0,
      "grad_norm": 8.631437691008928,
      "learning_rate": 2.5856455425531003e-06,
      "loss": 0.8587,
      "num_input_tokens_seen": 74884835,
      "step": 3520
    },
    {
      "epoch": 0.4233752179402393,
      "flos": 15088858767360.0,
      "grad_norm": 5.284984825495045,
      "learning_rate": 2.5849006713744902e-06,
      "loss": 1.0019,
      "num_input_tokens_seen": 74903350,
      "step": 3521
    },
    {
      "epoch": 0.42349546083087836,
      "flos": 14698300354560.0,
      "grad_norm": 12.694345516057718,
      "learning_rate": 2.5841557114669135e-06,
      "loss": 0.9229,
      "num_input_tokens_seen": 74919930,
      "step": 3522
    },
    {
      "epoch": 0.42361570372151747,
      "flos": 13174835343360.0,
      "grad_norm": 8.427831917526484,
      "learning_rate": 2.58341066294338e-06,
      "loss": 0.8511,
      "num_input_tokens_seen": 74936315,
      "step": 3523
    },
    {
      "epoch": 0.4237359466121566,
      "flos": 14881664655360.0,
      "grad_norm": 8.40039403082063,
      "learning_rate": 2.5826655259169124e-06,
      "loss": 1.0485,
      "num_input_tokens_seen": 74954690,
      "step": 3524
    },
    {
      "epoch": 0.42385618950279563,
      "flos": 12781854904320.0,
      "grad_norm": 3.3656908855878465,
      "learning_rate": 2.5819203005005475e-06,
      "loss": 1.0897,
      "num_input_tokens_seen": 74971745,
      "step": 3525
    },
    {
      "epoch": 0.42397643239343474,
      "flos": 16900940697600.0,
      "grad_norm": 5.556542281368314,
      "learning_rate": 2.581174986807336e-06,
      "loss": 0.9815,
      "num_input_tokens_seen": 74991700,
      "step": 3526
    },
    {
      "epoch": 0.42409667528407385,
      "flos": 11709177937920.0,
      "grad_norm": 4.362667686059672,
      "learning_rate": 2.580429584950341e-06,
      "loss": 1.1072,
      "num_input_tokens_seen": 75007170,
      "step": 3527
    },
    {
      "epoch": 0.4242169181747129,
      "flos": 11341989457920.0,
      "grad_norm": 3.648700190037848,
      "learning_rate": 2.5796840950426397e-06,
      "loss": 0.8396,
      "num_input_tokens_seen": 75023975,
      "step": 3528
    },
    {
      "epoch": 0.424337161065352,
      "flos": 14252509655040.0,
      "grad_norm": 33.943711304295746,
      "learning_rate": 2.578938517197322e-06,
      "loss": 0.8536,
      "num_input_tokens_seen": 75041790,
      "step": 3529
    },
    {
      "epoch": 0.4244574039559911,
      "flos": 16979420282880.0,
      "grad_norm": 4.418815497920578,
      "learning_rate": 2.5781928515274916e-06,
      "loss": 0.8134,
      "num_input_tokens_seen": 75060230,
      "step": 3530
    },
    {
      "epoch": 0.4245776468466302,
      "flos": 12442727116800.0,
      "grad_norm": 3.9819656567244532,
      "learning_rate": 2.577447098146265e-06,
      "loss": 0.8657,
      "num_input_tokens_seen": 75077125,
      "step": 3531
    },
    {
      "epoch": 0.4246978897372693,
      "flos": 19779935784960.0,
      "grad_norm": 9.128936336599972,
      "learning_rate": 2.5767012571667724e-06,
      "loss": 0.9771,
      "num_input_tokens_seen": 75096325,
      "step": 3532
    },
    {
      "epoch": 0.42481813262790835,
      "flos": 11027549921280.0,
      "grad_norm": 4.575877187881491,
      "learning_rate": 2.5759553287021587e-06,
      "loss": 0.8702,
      "num_input_tokens_seen": 75114375,
      "step": 3533
    },
    {
      "epoch": 0.42493837551854746,
      "flos": 17031648030720.0,
      "grad_norm": 3.742953737691012,
      "learning_rate": 2.5752093128655786e-06,
      "loss": 0.9628,
      "num_input_tokens_seen": 75132340,
      "step": 3534
    },
    {
      "epoch": 0.4250586184091866,
      "flos": 14776718622720.0,
      "grad_norm": 3.782635395579208,
      "learning_rate": 2.574463209770204e-06,
      "loss": 0.9259,
      "num_input_tokens_seen": 75151375,
      "step": 3535
    },
    {
      "epoch": 0.42517886129982563,
      "flos": 21645318512640.0,
      "grad_norm": 7.571047657641401,
      "learning_rate": 2.5737170195292165e-06,
      "loss": 0.987,
      "num_input_tokens_seen": 75174430,
      "step": 3536
    },
    {
      "epoch": 0.42529910419046474,
      "flos": 14249627750400.0,
      "grad_norm": 5.443933824135574,
      "learning_rate": 2.572970742255814e-06,
      "loss": 0.9706,
      "num_input_tokens_seen": 75192640,
      "step": 3537
    },
    {
      "epoch": 0.42541934708110385,
      "flos": 16083270758400.0,
      "grad_norm": 4.476008569022286,
      "learning_rate": 2.5722243780632046e-06,
      "loss": 1.0151,
      "num_input_tokens_seen": 75210625,
      "step": 3538
    },
    {
      "epoch": 0.4255395899717429,
      "flos": 47400425164800.0,
      "grad_norm": 0.8283148853760143,
      "learning_rate": 2.5714779270646125e-06,
      "loss": 0.8521,
      "num_input_tokens_seen": 75271115,
      "step": 3539
    },
    {
      "epoch": 0.425659832862382,
      "flos": 12705828003840.0,
      "grad_norm": 5.801575926732544,
      "learning_rate": 2.5707313893732735e-06,
      "loss": 0.9578,
      "num_input_tokens_seen": 75289375,
      "step": 3540
    },
    {
      "epoch": 0.4257800757530211,
      "flos": 17082802728960.0,
      "grad_norm": 7.92334018979266,
      "learning_rate": 2.5699847651024364e-06,
      "loss": 0.9605,
      "num_input_tokens_seen": 75309735,
      "step": 3541
    },
    {
      "epoch": 0.4259003186436602,
      "flos": 16848375705600.0,
      "grad_norm": 4.859883468863488,
      "learning_rate": 2.5692380543653627e-06,
      "loss": 0.9743,
      "num_input_tokens_seen": 75327610,
      "step": 3542
    },
    {
      "epoch": 0.4260205615342993,
      "flos": 10784323891200.0,
      "grad_norm": 4.8384638336239645,
      "learning_rate": 2.5684912572753293e-06,
      "loss": 0.8832,
      "num_input_tokens_seen": 75343005,
      "step": 3543
    },
    {
      "epoch": 0.4261408044249384,
      "flos": 21857142067200.0,
      "grad_norm": 9.295002929949696,
      "learning_rate": 2.5677443739456245e-06,
      "loss": 1.0418,
      "num_input_tokens_seen": 75364385,
      "step": 3544
    },
    {
      "epoch": 0.42626104731557746,
      "flos": 16533476290560.0,
      "grad_norm": 5.199594027003868,
      "learning_rate": 2.5669974044895495e-06,
      "loss": 0.9844,
      "num_input_tokens_seen": 75380500,
      "step": 3545
    },
    {
      "epoch": 0.42638129020621657,
      "flos": 18421339852800.0,
      "grad_norm": 8.990934680727332,
      "learning_rate": 2.5662503490204187e-06,
      "loss": 0.9849,
      "num_input_tokens_seen": 75400385,
      "step": 3546
    },
    {
      "epoch": 0.4265015330968556,
      "flos": 18864831160320.0,
      "grad_norm": 4.379895985731305,
      "learning_rate": 2.5655032076515603e-06,
      "loss": 0.9552,
      "num_input_tokens_seen": 75419430,
      "step": 3547
    },
    {
      "epoch": 0.42662177598749473,
      "flos": 17346823372800.0,
      "grad_norm": 6.423294244689202,
      "learning_rate": 2.5647559804963155e-06,
      "loss": 1.0094,
      "num_input_tokens_seen": 75439080,
      "step": 3548
    },
    {
      "epoch": 0.42674201887813384,
      "flos": 16454966046720.0,
      "grad_norm": 4.113926676570125,
      "learning_rate": 2.5640086676680364e-06,
      "loss": 0.9814,
      "num_input_tokens_seen": 75460295,
      "step": 3549
    },
    {
      "epoch": 0.4268622617687729,
      "flos": 15406456135680.0,
      "grad_norm": 5.905281431515385,
      "learning_rate": 2.5632612692800923e-06,
      "loss": 0.9845,
      "num_input_tokens_seen": 75479080,
      "step": 3550
    },
    {
      "epoch": 0.426982504659412,
      "flos": 16664428892160.0,
      "grad_norm": 4.957700908089089,
      "learning_rate": 2.5625137854458603e-06,
      "loss": 0.9403,
      "num_input_tokens_seen": 75497815,
      "step": 3551
    },
    {
      "epoch": 0.4271027475500511,
      "flos": 13413493248000.0,
      "grad_norm": 5.3883409166897245,
      "learning_rate": 2.561766216278735e-06,
      "loss": 0.9911,
      "num_input_tokens_seen": 75515130,
      "step": 3552
    },
    {
      "epoch": 0.4272229904406902,
      "flos": 19129894195200.0,
      "grad_norm": 5.201156472940019,
      "learning_rate": 2.561018561892121e-06,
      "loss": 1.0046,
      "num_input_tokens_seen": 75533990,
      "step": 3553
    },
    {
      "epoch": 0.4273432333313293,
      "flos": 17031893299200.0,
      "grad_norm": 6.253653107189317,
      "learning_rate": 2.5602708223994363e-06,
      "loss": 0.9571,
      "num_input_tokens_seen": 75555575,
      "step": 3554
    },
    {
      "epoch": 0.4274634762219684,
      "flos": 21070169456640.0,
      "grad_norm": 7.6197656894813806,
      "learning_rate": 2.559522997914115e-06,
      "loss": 0.8673,
      "num_input_tokens_seen": 75574875,
      "step": 3555
    },
    {
      "epoch": 0.42758371911260745,
      "flos": 15222846566400.0,
      "grad_norm": 3.837545582255178,
      "learning_rate": 2.558775088549599e-06,
      "loss": 1.0529,
      "num_input_tokens_seen": 75594175,
      "step": 3556
    },
    {
      "epoch": 0.42770396200324656,
      "flos": 10420507852800.0,
      "grad_norm": 13.16941656698746,
      "learning_rate": 2.5580270944193467e-06,
      "loss": 0.8469,
      "num_input_tokens_seen": 75610715,
      "step": 3557
    },
    {
      "epoch": 0.4278242048938857,
      "flos": 50601831014400.0,
      "grad_norm": 0.7471382858169865,
      "learning_rate": 2.557279015636827e-06,
      "loss": 0.7673,
      "num_input_tokens_seen": 75670845,
      "step": 3558
    },
    {
      "epoch": 0.42794444778452473,
      "flos": 49674217697280.0,
      "grad_norm": 0.8351387414622494,
      "learning_rate": 2.5565308523155245e-06,
      "loss": 0.8514,
      "num_input_tokens_seen": 75730165,
      "step": 3559
    },
    {
      "epoch": 0.42806469067516384,
      "flos": 12908453990400.0,
      "grad_norm": 6.516980805431253,
      "learning_rate": 2.5557826045689336e-06,
      "loss": 1.0173,
      "num_input_tokens_seen": 75746125,
      "step": 3560
    },
    {
      "epoch": 0.4281849335658029,
      "flos": 39018538905600.0,
      "grad_norm": 0.8390202232806147,
      "learning_rate": 2.5550342725105643e-06,
      "loss": 0.8005,
      "num_input_tokens_seen": 75804010,
      "step": 3561
    },
    {
      "epoch": 0.428305176456442,
      "flos": 12233632174080.0,
      "grad_norm": 10.256109138372267,
      "learning_rate": 2.554285856253937e-06,
      "loss": 1.0108,
      "num_input_tokens_seen": 75822565,
      "step": 3562
    },
    {
      "epoch": 0.4284254193470811,
      "flos": 18762583080960.0,
      "grad_norm": 4.300641105232524,
      "learning_rate": 2.5535373559125855e-06,
      "loss": 0.9556,
      "num_input_tokens_seen": 75842650,
      "step": 3563
    },
    {
      "epoch": 0.42854566223772017,
      "flos": 21173306634240.0,
      "grad_norm": 4.67115538295338,
      "learning_rate": 2.552788771600057e-06,
      "loss": 1.0156,
      "num_input_tokens_seen": 75862680,
      "step": 3564
    },
    {
      "epoch": 0.4286659051283593,
      "flos": 15642600038400.0,
      "grad_norm": 5.493351819707882,
      "learning_rate": 2.5520401034299118e-06,
      "loss": 1.0126,
      "num_input_tokens_seen": 75880160,
      "step": 3565
    },
    {
      "epoch": 0.4287861480189984,
      "flos": 9401744855040.0,
      "grad_norm": 5.7672097113963705,
      "learning_rate": 2.551291351515722e-06,
      "loss": 1.0674,
      "num_input_tokens_seen": 75896895,
      "step": 3566
    },
    {
      "epoch": 0.42890639090963745,
      "flos": 18972475146240.0,
      "grad_norm": 4.72949107845718,
      "learning_rate": 2.5505425159710726e-06,
      "loss": 1.0513,
      "num_input_tokens_seen": 75916425,
      "step": 3567
    },
    {
      "epoch": 0.42902663380027656,
      "flos": 17106233978880.0,
      "grad_norm": 5.805123519437613,
      "learning_rate": 2.549793596909561e-06,
      "loss": 1.0272,
      "num_input_tokens_seen": 75934765,
      "step": 3568
    },
    {
      "epoch": 0.42914687669091567,
      "flos": 11053311221760.0,
      "grad_norm": 5.512112371608657,
      "learning_rate": 2.5490445944447976e-06,
      "loss": 0.8504,
      "num_input_tokens_seen": 75952980,
      "step": 3569
    },
    {
      "epoch": 0.4292671195815547,
      "flos": 22433854709760.0,
      "grad_norm": 8.78412100670019,
      "learning_rate": 2.548295508690406e-06,
      "loss": 0.8465,
      "num_input_tokens_seen": 75973995,
      "step": 3570
    },
    {
      "epoch": 0.42938736247219383,
      "flos": 21563405168640.0,
      "grad_norm": 7.024761817668319,
      "learning_rate": 2.5475463397600217e-06,
      "loss": 0.9583,
      "num_input_tokens_seen": 75993795,
      "step": 3571
    },
    {
      "epoch": 0.42950760536283294,
      "flos": 20911340113920.0,
      "grad_norm": 8.303343354592643,
      "learning_rate": 2.546797087767293e-06,
      "loss": 0.9751,
      "num_input_tokens_seen": 76013640,
      "step": 3572
    },
    {
      "epoch": 0.429627848253472,
      "flos": 19129372999680.0,
      "grad_norm": 5.252570113558844,
      "learning_rate": 2.546047752825881e-06,
      "loss": 1.0584,
      "num_input_tokens_seen": 76033965,
      "step": 3573
    },
    {
      "epoch": 0.4297480911441111,
      "flos": 9794694635520.0,
      "grad_norm": 4.368658210543241,
      "learning_rate": 2.5452983350494595e-06,
      "loss": 1.1249,
      "num_input_tokens_seen": 76049240,
      "step": 3574
    },
    {
      "epoch": 0.4298683340347502,
      "flos": 14724828119040.0,
      "grad_norm": 9.344490965995645,
      "learning_rate": 2.544548834551713e-06,
      "loss": 0.8624,
      "num_input_tokens_seen": 76067965,
      "step": 3575
    },
    {
      "epoch": 0.4299885769253893,
      "flos": 14825481953280.0,
      "grad_norm": 4.650554681325029,
      "learning_rate": 2.5437992514463424e-06,
      "loss": 1.1163,
      "num_input_tokens_seen": 76081010,
      "step": 3576
    },
    {
      "epoch": 0.4301088198160284,
      "flos": 18133918617600.0,
      "grad_norm": 6.799972915595475,
      "learning_rate": 2.5430495858470565e-06,
      "loss": 1.0719,
      "num_input_tokens_seen": 76100200,
      "step": 3577
    },
    {
      "epoch": 0.43022906270666744,
      "flos": 12941174784000.0,
      "grad_norm": 5.649982499627211,
      "learning_rate": 2.54229983786758e-06,
      "loss": 0.9643,
      "num_input_tokens_seen": 76117865,
      "step": 3578
    },
    {
      "epoch": 0.43034930559730655,
      "flos": 16634681978880.0,
      "grad_norm": 6.433267035002693,
      "learning_rate": 2.541550007621651e-06,
      "loss": 1.0353,
      "num_input_tokens_seen": 76136075,
      "step": 3579
    },
    {
      "epoch": 0.43046954848794566,
      "flos": 20074163220480.0,
      "grad_norm": 4.702370415158966,
      "learning_rate": 2.5408000952230156e-06,
      "loss": 0.9963,
      "num_input_tokens_seen": 76154585,
      "step": 3580
    },
    {
      "epoch": 0.4305897913785847,
      "flos": 20358120038400.0,
      "grad_norm": 5.71128917190567,
      "learning_rate": 2.5400501007854357e-06,
      "loss": 1.0877,
      "num_input_tokens_seen": 76173750,
      "step": 3581
    },
    {
      "epoch": 0.43071003426922383,
      "flos": 14514598809600.0,
      "grad_norm": 5.108642712300252,
      "learning_rate": 2.539300024422685e-06,
      "loss": 0.9485,
      "num_input_tokens_seen": 76191415,
      "step": 3582
    },
    {
      "epoch": 0.43083027715986294,
      "flos": 37191894159360.0,
      "grad_norm": 0.8097792702025272,
      "learning_rate": 2.538549866248549e-06,
      "loss": 0.8327,
      "num_input_tokens_seen": 76246115,
      "step": 3583
    },
    {
      "epoch": 0.430950520050502,
      "flos": 11813940019200.0,
      "grad_norm": 6.91670864352928,
      "learning_rate": 2.5377996263768274e-06,
      "loss": 0.9998,
      "num_input_tokens_seen": 76263915,
      "step": 3584
    },
    {
      "epoch": 0.4310707629411411,
      "flos": 17504150446080.0,
      "grad_norm": 4.705972310216188,
      "learning_rate": 2.5370493049213293e-06,
      "loss": 0.8674,
      "num_input_tokens_seen": 76283280,
      "step": 3585
    },
    {
      "epoch": 0.4311910058317802,
      "flos": 18815454658560.0,
      "grad_norm": 3.8107562220251787,
      "learning_rate": 2.536298901995878e-06,
      "loss": 0.9934,
      "num_input_tokens_seen": 76302210,
      "step": 3586
    },
    {
      "epoch": 0.43131124872241927,
      "flos": 17897621422080.0,
      "grad_norm": 6.072462881767121,
      "learning_rate": 2.535548417714311e-06,
      "loss": 0.9931,
      "num_input_tokens_seen": 76321230,
      "step": 3587
    },
    {
      "epoch": 0.4314314916130584,
      "flos": 15352450191360.0,
      "grad_norm": 2.9131185112262803,
      "learning_rate": 2.534797852190474e-06,
      "loss": 1.0618,
      "num_input_tokens_seen": 76341130,
      "step": 3588
    },
    {
      "epoch": 0.4315517345036975,
      "flos": 13670247813120.0,
      "grad_norm": 6.173767572267113,
      "learning_rate": 2.5340472055382283e-06,
      "loss": 0.9994,
      "num_input_tokens_seen": 76356880,
      "step": 3589
    },
    {
      "epoch": 0.43167197739433655,
      "flos": 17262855905280.0,
      "grad_norm": 5.707275886697173,
      "learning_rate": 2.5332964778714468e-06,
      "loss": 1.005,
      "num_input_tokens_seen": 76373785,
      "step": 3590
    },
    {
      "epoch": 0.43179222028497566,
      "flos": 11940355153920.0,
      "grad_norm": 4.447491210791581,
      "learning_rate": 2.5325456693040123e-06,
      "loss": 0.8591,
      "num_input_tokens_seen": 76390700,
      "step": 3591
    },
    {
      "epoch": 0.43191246317561477,
      "flos": 12495445401600.0,
      "grad_norm": 10.047495394190985,
      "learning_rate": 2.531794779949824e-06,
      "loss": 0.9446,
      "num_input_tokens_seen": 76408320,
      "step": 3592
    },
    {
      "epoch": 0.4320327060662538,
      "flos": 16979788185600.0,
      "grad_norm": 5.470094700748448,
      "learning_rate": 2.5310438099227903e-06,
      "loss": 1.0721,
      "num_input_tokens_seen": 76425305,
      "step": 3593
    },
    {
      "epoch": 0.43215294895689293,
      "flos": 47541617725440.0,
      "grad_norm": 1.3531860566845315,
      "learning_rate": 2.530292759336833e-06,
      "loss": 0.7545,
      "num_input_tokens_seen": 76485760,
      "step": 3594
    },
    {
      "epoch": 0.432273191847532,
      "flos": 14619146280960.0,
      "grad_norm": 4.6886621687576575,
      "learning_rate": 2.5295416283058855e-06,
      "loss": 0.897,
      "num_input_tokens_seen": 76504345,
      "step": 3595
    },
    {
      "epoch": 0.4323934347381711,
      "flos": 13676318208000.0,
      "grad_norm": 3.730130516816449,
      "learning_rate": 2.5287904169438943e-06,
      "loss": 0.8577,
      "num_input_tokens_seen": 76523270,
      "step": 3596
    },
    {
      "epoch": 0.4325136776288102,
      "flos": 15432861265920.0,
      "grad_norm": 8.713496331011271,
      "learning_rate": 2.528039125364817e-06,
      "loss": 0.8276,
      "num_input_tokens_seen": 76541795,
      "step": 3597
    },
    {
      "epoch": 0.43263392051944927,
      "flos": 15873654620160.0,
      "grad_norm": 8.350063721190947,
      "learning_rate": 2.5272877536826246e-06,
      "loss": 0.9435,
      "num_input_tokens_seen": 76560310,
      "step": 3598
    },
    {
      "epoch": 0.4327541634100884,
      "flos": 20781215293440.0,
      "grad_norm": 15.829158888524812,
      "learning_rate": 2.5265363020112986e-06,
      "loss": 0.8751,
      "num_input_tokens_seen": 76580350,
      "step": 3599
    },
    {
      "epoch": 0.4328744063007275,
      "flos": 18552660357120.0,
      "grad_norm": 4.2865385262028255,
      "learning_rate": 2.5257847704648344e-06,
      "loss": 1.0441,
      "num_input_tokens_seen": 76601300,
      "step": 3600
    },
    {
      "epoch": 0.43299464919136654,
      "flos": 11735215165440.0,
      "grad_norm": 4.039873938151688,
      "learning_rate": 2.525033159157239e-06,
      "loss": 0.9608,
      "num_input_tokens_seen": 76617335,
      "step": 3601
    },
    {
      "epoch": 0.43311489208200565,
      "flos": 11394523791360.0,
      "grad_norm": 7.402147404223159,
      "learning_rate": 2.52428146820253e-06,
      "loss": 0.9663,
      "num_input_tokens_seen": 76635310,
      "step": 3602
    },
    {
      "epoch": 0.43323513497264476,
      "flos": 16298068193280.0,
      "grad_norm": 3.6164333801504496,
      "learning_rate": 2.52352969771474e-06,
      "loss": 1.0134,
      "num_input_tokens_seen": 76654255,
      "step": 3603
    },
    {
      "epoch": 0.4333553778632838,
      "flos": 18000114769920.0,
      "grad_norm": 4.925592090625989,
      "learning_rate": 2.5227778478079106e-06,
      "loss": 1.0819,
      "num_input_tokens_seen": 76673385,
      "step": 3604
    },
    {
      "epoch": 0.43347562075392293,
      "flos": 13751210741760.0,
      "grad_norm": 4.042139058510353,
      "learning_rate": 2.522025918596098e-06,
      "loss": 0.9637,
      "num_input_tokens_seen": 76691405,
      "step": 3605
    },
    {
      "epoch": 0.43359586364456204,
      "flos": 18738875904000.0,
      "grad_norm": 3.436133511908074,
      "learning_rate": 2.521273910193368e-06,
      "loss": 0.8521,
      "num_input_tokens_seen": 76714305,
      "step": 3606
    },
    {
      "epoch": 0.4337161065352011,
      "flos": 11308563517440.0,
      "grad_norm": 6.023873200843768,
      "learning_rate": 2.5205218227138006e-06,
      "loss": 1.0583,
      "num_input_tokens_seen": 76726980,
      "step": 3607
    },
    {
      "epoch": 0.4338363494258402,
      "flos": 14352918220800.0,
      "grad_norm": 8.929883579294083,
      "learning_rate": 2.519769656271486e-06,
      "loss": 0.9868,
      "num_input_tokens_seen": 76744120,
      "step": 3608
    },
    {
      "epoch": 0.43395659231647926,
      "flos": 14252417679360.0,
      "grad_norm": 6.144789198634425,
      "learning_rate": 2.5190174109805285e-06,
      "loss": 0.8734,
      "num_input_tokens_seen": 76763665,
      "step": 3609
    },
    {
      "epoch": 0.43407683520711837,
      "flos": 14121587712000.0,
      "grad_norm": 6.310774765693238,
      "learning_rate": 2.518265086955042e-06,
      "loss": 0.8279,
      "num_input_tokens_seen": 76781105,
      "step": 3610
    },
    {
      "epoch": 0.4341970780977575,
      "flos": 16425832304640.0,
      "grad_norm": 3.967970523732437,
      "learning_rate": 2.5175126843091534e-06,
      "loss": 1.0164,
      "num_input_tokens_seen": 76800195,
      "step": 3611
    },
    {
      "epoch": 0.43431732098839654,
      "flos": 26702265692160.0,
      "grad_norm": 3.2528719317148322,
      "learning_rate": 2.5167602031570034e-06,
      "loss": 0.9358,
      "num_input_tokens_seen": 76820100,
      "step": 3612
    },
    {
      "epoch": 0.43443756387903565,
      "flos": 22721490554880.0,
      "grad_norm": 7.28468758965054,
      "learning_rate": 2.51600764361274e-06,
      "loss": 0.9263,
      "num_input_tokens_seen": 76841345,
      "step": 3613
    },
    {
      "epoch": 0.43455780676967476,
      "flos": 16691079290880.0,
      "grad_norm": 8.160865851446923,
      "learning_rate": 2.5152550057905283e-06,
      "loss": 0.9839,
      "num_input_tokens_seen": 76860955,
      "step": 3614
    },
    {
      "epoch": 0.4346780496603138,
      "flos": 17215686819840.0,
      "grad_norm": 4.742943390835167,
      "learning_rate": 2.5145022898045415e-06,
      "loss": 0.9413,
      "num_input_tokens_seen": 76879860,
      "step": 3615
    },
    {
      "epoch": 0.4347982925509529,
      "flos": 12102495621120.0,
      "grad_norm": 6.26082985054541,
      "learning_rate": 2.5137494957689664e-06,
      "loss": 1.0936,
      "num_input_tokens_seen": 76895190,
      "step": 3616
    },
    {
      "epoch": 0.43491853544159204,
      "flos": 43623502356480.0,
      "grad_norm": 0.758586645436829,
      "learning_rate": 2.5129966237980016e-06,
      "loss": 0.7927,
      "num_input_tokens_seen": 76957905,
      "step": 3617
    },
    {
      "epoch": 0.4350387783322311,
      "flos": 15589697802240.0,
      "grad_norm": 7.724546163616209,
      "learning_rate": 2.512243674005857e-06,
      "loss": 0.972,
      "num_input_tokens_seen": 76976990,
      "step": 3618
    },
    {
      "epoch": 0.4351590212228702,
      "flos": 17844995112960.0,
      "grad_norm": 3.7884549941870707,
      "learning_rate": 2.5114906465067537e-06,
      "loss": 1.0598,
      "num_input_tokens_seen": 76997695,
      "step": 3619
    },
    {
      "epoch": 0.4352792641135093,
      "flos": 15274798387200.0,
      "grad_norm": 10.65724155142181,
      "learning_rate": 2.5107375414149264e-06,
      "loss": 0.9355,
      "num_input_tokens_seen": 77016660,
      "step": 3620
    },
    {
      "epoch": 0.43539950700414837,
      "flos": 11499623116800.0,
      "grad_norm": 7.932115545757639,
      "learning_rate": 2.5099843588446197e-06,
      "loss": 0.8952,
      "num_input_tokens_seen": 77034700,
      "step": 3621
    },
    {
      "epoch": 0.4355197498947875,
      "flos": 11814737141760.0,
      "grad_norm": 3.9913644558115977,
      "learning_rate": 2.509231098910091e-06,
      "loss": 0.8097,
      "num_input_tokens_seen": 77054290,
      "step": 3622
    },
    {
      "epoch": 0.4356399927854266,
      "flos": 11578133360640.0,
      "grad_norm": 4.881948165687143,
      "learning_rate": 2.508477761725611e-06,
      "loss": 0.9259,
      "num_input_tokens_seen": 77072285,
      "step": 3623
    },
    {
      "epoch": 0.43576023567606564,
      "flos": 12181343109120.0,
      "grad_norm": 3.507459321115817,
      "learning_rate": 2.507724347405458e-06,
      "loss": 0.9985,
      "num_input_tokens_seen": 77089955,
      "step": 3624
    },
    {
      "epoch": 0.43588047856670475,
      "flos": 11257286184960.0,
      "grad_norm": 10.875699426367728,
      "learning_rate": 2.5069708560639243e-06,
      "loss": 1.0043,
      "num_input_tokens_seen": 77107585,
      "step": 3625
    },
    {
      "epoch": 0.4360007214573438,
      "flos": 16822185185280.0,
      "grad_norm": 14.669259680209827,
      "learning_rate": 2.5062172878153158e-06,
      "loss": 0.803,
      "num_input_tokens_seen": 77126580,
      "step": 3626
    },
    {
      "epoch": 0.4361209643479829,
      "flos": 15613895516160.0,
      "grad_norm": 4.710573906729595,
      "learning_rate": 2.505463642773947e-06,
      "loss": 1.0607,
      "num_input_tokens_seen": 77146265,
      "step": 3627
    },
    {
      "epoch": 0.43624120723862203,
      "flos": 12337995694080.0,
      "grad_norm": 5.627721001222657,
      "learning_rate": 2.504709921054146e-06,
      "loss": 0.944,
      "num_input_tokens_seen": 77162800,
      "step": 3628
    },
    {
      "epoch": 0.4363614501292611,
      "flos": 12679576166400.0,
      "grad_norm": 6.744506578545025,
      "learning_rate": 2.50395612277025e-06,
      "loss": 1.0384,
      "num_input_tokens_seen": 77178375,
      "step": 3629
    },
    {
      "epoch": 0.4364816930199002,
      "flos": 14409806069760.0,
      "grad_norm": 4.503199181882154,
      "learning_rate": 2.503202248036612e-06,
      "loss": 0.9279,
      "num_input_tokens_seen": 77196950,
      "step": 3630
    },
    {
      "epoch": 0.4366019359105393,
      "flos": 17110986055680.0,
      "grad_norm": 4.634386589668583,
      "learning_rate": 2.5024482969675927e-06,
      "loss": 0.9187,
      "num_input_tokens_seen": 77216625,
      "step": 3631
    },
    {
      "epoch": 0.43672217880117836,
      "flos": 15451663073280.0,
      "grad_norm": 6.079792960807882,
      "learning_rate": 2.501694269677566e-06,
      "loss": 1.0351,
      "num_input_tokens_seen": 77234115,
      "step": 3632
    },
    {
      "epoch": 0.4368424216918175,
      "flos": 12779310243840.0,
      "grad_norm": 5.291509290542672,
      "learning_rate": 2.500940166280918e-06,
      "loss": 0.993,
      "num_input_tokens_seen": 77252265,
      "step": 3633
    },
    {
      "epoch": 0.4369626645824566,
      "flos": 18106838999040.0,
      "grad_norm": 4.608563471331073,
      "learning_rate": 2.500185986892045e-06,
      "loss": 0.9777,
      "num_input_tokens_seen": 77271470,
      "step": 3634
    },
    {
      "epoch": 0.43708290747309564,
      "flos": 18003241943040.0,
      "grad_norm": 4.43998176055461,
      "learning_rate": 2.499431731625355e-06,
      "loss": 0.9669,
      "num_input_tokens_seen": 77290215,
      "step": 3635
    },
    {
      "epoch": 0.43720315036373475,
      "flos": 22510678732800.0,
      "grad_norm": 7.253946742785473,
      "learning_rate": 2.4986774005952686e-06,
      "loss": 0.9978,
      "num_input_tokens_seen": 77312310,
      "step": 3636
    },
    {
      "epoch": 0.43732339325437386,
      "flos": 16428622233600.0,
      "grad_norm": 5.337570990418656,
      "learning_rate": 2.4979229939162166e-06,
      "loss": 1.0305,
      "num_input_tokens_seen": 77330810,
      "step": 3637
    },
    {
      "epoch": 0.4374436361450129,
      "flos": 19759263805440.0,
      "grad_norm": 3.3952314315365673,
      "learning_rate": 2.4971685117026433e-06,
      "loss": 0.9981,
      "num_input_tokens_seen": 77350295,
      "step": 3638
    },
    {
      "epoch": 0.437563879035652,
      "flos": 17190998568960.0,
      "grad_norm": 3.6756757828503215,
      "learning_rate": 2.4964139540690018e-06,
      "loss": 0.9601,
      "num_input_tokens_seen": 77373350,
      "step": 3639
    },
    {
      "epoch": 0.4376841219262911,
      "flos": 16691140608000.0,
      "grad_norm": 5.087817522867353,
      "learning_rate": 2.495659321129758e-06,
      "loss": 0.9171,
      "num_input_tokens_seen": 77390815,
      "step": 3640
    },
    {
      "epoch": 0.4378043648169302,
      "flos": 18107421511680.0,
      "grad_norm": 5.554460180167503,
      "learning_rate": 2.494904612999389e-06,
      "loss": 0.9487,
      "num_input_tokens_seen": 77409245,
      "step": 3641
    },
    {
      "epoch": 0.4379246077075693,
      "flos": 38571491205120.0,
      "grad_norm": 1.4559598772892484,
      "learning_rate": 2.4941498297923843e-06,
      "loss": 0.7996,
      "num_input_tokens_seen": 77469535,
      "step": 3642
    },
    {
      "epoch": 0.43804485059820836,
      "flos": 14614056960000.0,
      "grad_norm": 23.424756952447765,
      "learning_rate": 2.4933949716232424e-06,
      "loss": 0.8983,
      "num_input_tokens_seen": 77486780,
      "step": 3643
    },
    {
      "epoch": 0.43816509348884747,
      "flos": 16977580769280.0,
      "grad_norm": 4.0180158214418995,
      "learning_rate": 2.492640038606476e-06,
      "loss": 0.9208,
      "num_input_tokens_seen": 77504865,
      "step": 3644
    },
    {
      "epoch": 0.4382853363794866,
      "flos": 10371989790720.0,
      "grad_norm": 6.4235730675839475,
      "learning_rate": 2.491885030856608e-06,
      "loss": 0.9833,
      "num_input_tokens_seen": 77522680,
      "step": 3645
    },
    {
      "epoch": 0.43840557927012563,
      "flos": 12154907320320.0,
      "grad_norm": 5.5439543435885135,
      "learning_rate": 2.4911299484881713e-06,
      "loss": 1.0162,
      "num_input_tokens_seen": 77539930,
      "step": 3646
    },
    {
      "epoch": 0.43852582216076474,
      "flos": 13754215280640.0,
      "grad_norm": 3.4620723198576897,
      "learning_rate": 2.490374791615712e-06,
      "loss": 1.0041,
      "num_input_tokens_seen": 77559675,
      "step": 3647
    },
    {
      "epoch": 0.43864606505140386,
      "flos": 12808474644480.0,
      "grad_norm": 21.458416846646752,
      "learning_rate": 2.4896195603537867e-06,
      "loss": 0.9622,
      "num_input_tokens_seen": 77574005,
      "step": 3648
    },
    {
      "epoch": 0.4387663079420429,
      "flos": 13936751800320.0,
      "grad_norm": 5.046109073495757,
      "learning_rate": 2.488864254816964e-06,
      "loss": 0.9323,
      "num_input_tokens_seen": 77592415,
      "step": 3649
    },
    {
      "epoch": 0.438886550832682,
      "flos": 13990451159040.0,
      "grad_norm": 4.706965199873721,
      "learning_rate": 2.4881088751198218e-06,
      "loss": 0.8646,
      "num_input_tokens_seen": 77610295,
      "step": 3650
    },
    {
      "epoch": 0.43900679372332113,
      "flos": 10265541488640.0,
      "grad_norm": 4.807784507454259,
      "learning_rate": 2.4873534213769517e-06,
      "loss": 0.8209,
      "num_input_tokens_seen": 77625245,
      "step": 3651
    },
    {
      "epoch": 0.4391270366139602,
      "flos": 17106632540160.0,
      "grad_norm": 8.193331672703467,
      "learning_rate": 2.4865978937029547e-06,
      "loss": 0.9046,
      "num_input_tokens_seen": 77643945,
      "step": 3652
    },
    {
      "epoch": 0.4392472795045993,
      "flos": 22484365578240.0,
      "grad_norm": 4.290945615965647,
      "learning_rate": 2.485842292212445e-06,
      "loss": 0.8612,
      "num_input_tokens_seen": 77664880,
      "step": 3653
    },
    {
      "epoch": 0.4393675223952384,
      "flos": 10502605148160.0,
      "grad_norm": 7.880504561405569,
      "learning_rate": 2.485086617020045e-06,
      "loss": 0.987,
      "num_input_tokens_seen": 77683095,
      "step": 3654
    },
    {
      "epoch": 0.43948776528587746,
      "flos": 10473042186240.0,
      "grad_norm": 4.545450623250483,
      "learning_rate": 2.4843308682403903e-06,
      "loss": 1.0034,
      "num_input_tokens_seen": 77699730,
      "step": 3655
    },
    {
      "epoch": 0.4396080081765166,
      "flos": 9818156544000.0,
      "grad_norm": 3.070315295502067,
      "learning_rate": 2.4835750459881294e-06,
      "loss": 1.0156,
      "num_input_tokens_seen": 77716075,
      "step": 3656
    },
    {
      "epoch": 0.43972825106715563,
      "flos": 12914984263680.0,
      "grad_norm": 5.165970211875061,
      "learning_rate": 2.4828191503779177e-06,
      "loss": 1.0056,
      "num_input_tokens_seen": 77733895,
      "step": 3657
    },
    {
      "epoch": 0.43984849395779474,
      "flos": 11944984596480.0,
      "grad_norm": 3.737359932919022,
      "learning_rate": 2.482063181524425e-06,
      "loss": 1.0856,
      "num_input_tokens_seen": 77749515,
      "step": 3658
    },
    {
      "epoch": 0.43996873684843385,
      "flos": 13250923560960.0,
      "grad_norm": 4.600379778899501,
      "learning_rate": 2.4813071395423307e-06,
      "loss": 0.9963,
      "num_input_tokens_seen": 77766800,
      "step": 3659
    },
    {
      "epoch": 0.4400889797390729,
      "flos": 16816973230080.0,
      "grad_norm": 4.887635118416805,
      "learning_rate": 2.4805510245463263e-06,
      "loss": 0.8483,
      "num_input_tokens_seen": 77786675,
      "step": 3660
    },
    {
      "epoch": 0.440209222629712,
      "flos": 16455456583680.0,
      "grad_norm": 3.8161970879703526,
      "learning_rate": 2.4797948366511137e-06,
      "loss": 0.7719,
      "num_input_tokens_seen": 77806105,
      "step": 3661
    },
    {
      "epoch": 0.4403294655203511,
      "flos": 17658105077760.0,
      "grad_norm": 4.943881582638222,
      "learning_rate": 2.4790385759714055e-06,
      "loss": 0.9431,
      "num_input_tokens_seen": 77824890,
      "step": 3662
    },
    {
      "epoch": 0.4404497084109902,
      "flos": 16035611136000.0,
      "grad_norm": 5.318182744438843,
      "learning_rate": 2.478282242621926e-06,
      "loss": 0.9024,
      "num_input_tokens_seen": 77845070,
      "step": 3663
    },
    {
      "epoch": 0.4405699513016293,
      "flos": 46514791526400.0,
      "grad_norm": 0.9299504845778791,
      "learning_rate": 2.477525836717411e-06,
      "loss": 0.827,
      "num_input_tokens_seen": 77912555,
      "step": 3664
    },
    {
      "epoch": 0.4406901941922684,
      "flos": 25447879987200.0,
      "grad_norm": 9.162012988128149,
      "learning_rate": 2.476769358372606e-06,
      "loss": 0.9829,
      "num_input_tokens_seen": 77933925,
      "step": 3665
    },
    {
      "epoch": 0.44081043708290746,
      "flos": 12784062320640.0,
      "grad_norm": 4.2898828941314155,
      "learning_rate": 2.4760128077022683e-06,
      "loss": 0.9372,
      "num_input_tokens_seen": 77951780,
      "step": 3666
    },
    {
      "epoch": 0.44093067997354657,
      "flos": 21489187123200.0,
      "grad_norm": 4.255236373055693,
      "learning_rate": 2.4752561848211672e-06,
      "loss": 0.8738,
      "num_input_tokens_seen": 77973900,
      "step": 3667
    },
    {
      "epoch": 0.4410509228641857,
      "flos": 16531054264320.0,
      "grad_norm": 7.07103814042187,
      "learning_rate": 2.4744994898440797e-06,
      "loss": 0.9041,
      "num_input_tokens_seen": 77992410,
      "step": 3668
    },
    {
      "epoch": 0.44117116575482473,
      "flos": 13833246720000.0,
      "grad_norm": 6.264489125629236,
      "learning_rate": 2.473742722885797e-06,
      "loss": 1.0281,
      "num_input_tokens_seen": 78011150,
      "step": 3669
    },
    {
      "epoch": 0.44129140864546385,
      "flos": 19261122723840.0,
      "grad_norm": 9.959962038634242,
      "learning_rate": 2.4729858840611197e-06,
      "loss": 0.8484,
      "num_input_tokens_seen": 78029780,
      "step": 3670
    },
    {
      "epoch": 0.4414116515361029,
      "flos": 18577317949440.0,
      "grad_norm": 4.678961543101863,
      "learning_rate": 2.4722289734848605e-06,
      "loss": 0.9159,
      "num_input_tokens_seen": 78049965,
      "step": 3671
    },
    {
      "epoch": 0.441531894426742,
      "flos": 15561698426880.0,
      "grad_norm": 4.893261476105192,
      "learning_rate": 2.471471991271841e-06,
      "loss": 0.972,
      "num_input_tokens_seen": 78066810,
      "step": 3672
    },
    {
      "epoch": 0.4416521373173811,
      "flos": 16661822914560.0,
      "grad_norm": 15.829952213603203,
      "learning_rate": 2.470714937536896e-06,
      "loss": 0.9925,
      "num_input_tokens_seen": 78085255,
      "step": 3673
    },
    {
      "epoch": 0.4417723802080202,
      "flos": 14431735050240.0,
      "grad_norm": 6.312296949085768,
      "learning_rate": 2.469957812394868e-06,
      "loss": 0.9034,
      "num_input_tokens_seen": 78103785,
      "step": 3674
    },
    {
      "epoch": 0.4418926230986593,
      "flos": 13387793264640.0,
      "grad_norm": 7.672964853851313,
      "learning_rate": 2.4692006159606148e-06,
      "loss": 0.9538,
      "num_input_tokens_seen": 78121035,
      "step": 3675
    },
    {
      "epoch": 0.4420128659892984,
      "flos": 13807301468160.0,
      "grad_norm": 4.209805876182027,
      "learning_rate": 2.468443348349e-06,
      "loss": 0.9756,
      "num_input_tokens_seen": 78138630,
      "step": 3676
    },
    {
      "epoch": 0.44213310887993745,
      "flos": 12678779043840.0,
      "grad_norm": 9.495493974657343,
      "learning_rate": 2.467686009674902e-06,
      "loss": 1.0175,
      "num_input_tokens_seen": 78152800,
      "step": 3677
    },
    {
      "epoch": 0.44225335177057656,
      "flos": 13620043530240.0,
      "grad_norm": 5.982015677525604,
      "learning_rate": 2.466928600053209e-06,
      "loss": 1.0453,
      "num_input_tokens_seen": 78167825,
      "step": 3678
    },
    {
      "epoch": 0.4423735946612157,
      "flos": 16686112604160.0,
      "grad_norm": 3.326387766866864,
      "learning_rate": 2.466171119598818e-06,
      "loss": 0.9056,
      "num_input_tokens_seen": 78187515,
      "step": 3679
    },
    {
      "epoch": 0.44249383755185473,
      "flos": 18996458250240.0,
      "grad_norm": 13.842782221392715,
      "learning_rate": 2.465413568426639e-06,
      "loss": 0.9534,
      "num_input_tokens_seen": 78208185,
      "step": 3680
    },
    {
      "epoch": 0.44261408044249384,
      "flos": 16454138265600.0,
      "grad_norm": 4.759207785581324,
      "learning_rate": 2.464655946651591e-06,
      "loss": 1.0185,
      "num_input_tokens_seen": 78226910,
      "step": 3681
    },
    {
      "epoch": 0.44273432333313295,
      "flos": 17399327047680.0,
      "grad_norm": 6.968673089804713,
      "learning_rate": 2.4638982543886065e-06,
      "loss": 0.9946,
      "num_input_tokens_seen": 78246670,
      "step": 3682
    },
    {
      "epoch": 0.442854566223772,
      "flos": 12416076718080.0,
      "grad_norm": 9.055013576689072,
      "learning_rate": 2.4631404917526254e-06,
      "loss": 1.0608,
      "num_input_tokens_seen": 78263345,
      "step": 3683
    },
    {
      "epoch": 0.4429748091144111,
      "flos": 17710486118400.0,
      "grad_norm": 6.808771075304867,
      "learning_rate": 2.4623826588586e-06,
      "loss": 0.9801,
      "num_input_tokens_seen": 78283335,
      "step": 3684
    },
    {
      "epoch": 0.4430950520050502,
      "flos": 15352358215680.0,
      "grad_norm": 5.4616351248224495,
      "learning_rate": 2.461624755821492e-06,
      "loss": 1.0195,
      "num_input_tokens_seen": 78302535,
      "step": 3685
    },
    {
      "epoch": 0.4432152948956893,
      "flos": 17478205194240.0,
      "grad_norm": 11.117703563215349,
      "learning_rate": 2.4608667827562763e-06,
      "loss": 0.9695,
      "num_input_tokens_seen": 78321585,
      "step": 3686
    },
    {
      "epoch": 0.4433355377863284,
      "flos": 15458622566400.0,
      "grad_norm": 6.247360893004454,
      "learning_rate": 2.460108739777936e-06,
      "loss": 1.0883,
      "num_input_tokens_seen": 78340440,
      "step": 3687
    },
    {
      "epoch": 0.44345578067696745,
      "flos": 14252908216320.0,
      "grad_norm": 7.545690597568806,
      "learning_rate": 2.4593506270014656e-06,
      "loss": 0.9571,
      "num_input_tokens_seen": 78359130,
      "step": 3688
    },
    {
      "epoch": 0.44357602356760656,
      "flos": 17188545884160.0,
      "grad_norm": 4.542261673800977,
      "learning_rate": 2.45859244454187e-06,
      "loss": 1.0085,
      "num_input_tokens_seen": 78378640,
      "step": 3689
    },
    {
      "epoch": 0.44369626645824567,
      "flos": 16137675264000.0,
      "grad_norm": 12.470638122628163,
      "learning_rate": 2.4578341925141655e-06,
      "loss": 0.8551,
      "num_input_tokens_seen": 78397575,
      "step": 3690
    },
    {
      "epoch": 0.4438165093488847,
      "flos": 27149466685440.0,
      "grad_norm": 4.560274391284334,
      "learning_rate": 2.457075871033378e-06,
      "loss": 0.903,
      "num_input_tokens_seen": 78419170,
      "step": 3691
    },
    {
      "epoch": 0.44393675223952384,
      "flos": 10974739660800.0,
      "grad_norm": 5.138164267557555,
      "learning_rate": 2.4563174802145445e-06,
      "loss": 1.0786,
      "num_input_tokens_seen": 78436140,
      "step": 3692
    },
    {
      "epoch": 0.44405699513016295,
      "flos": 46234176491520.0,
      "grad_norm": 0.6140003726193916,
      "learning_rate": 2.455559020172712e-06,
      "loss": 0.691,
      "num_input_tokens_seen": 78503215,
      "step": 3693
    },
    {
      "epoch": 0.444177238020802,
      "flos": 17057715916800.0,
      "grad_norm": 6.474102662110095,
      "learning_rate": 2.4548004910229385e-06,
      "loss": 1.0807,
      "num_input_tokens_seen": 78520510,
      "step": 3694
    },
    {
      "epoch": 0.4442974809114411,
      "flos": 16033894256640.0,
      "grad_norm": 7.745130670478272,
      "learning_rate": 2.4540418928802913e-06,
      "loss": 1.0595,
      "num_input_tokens_seen": 78538965,
      "step": 3695
    },
    {
      "epoch": 0.4444177238020802,
      "flos": 12521666580480.0,
      "grad_norm": 6.402965764476231,
      "learning_rate": 2.4532832258598506e-06,
      "loss": 0.8468,
      "num_input_tokens_seen": 78556515,
      "step": 3696
    },
    {
      "epoch": 0.4445379666927193,
      "flos": 20388296171520.0,
      "grad_norm": 3.6273464632568464,
      "learning_rate": 2.4525244900767047e-06,
      "loss": 1.0064,
      "num_input_tokens_seen": 78577050,
      "step": 3697
    },
    {
      "epoch": 0.4446582095833584,
      "flos": 50483601715200.0,
      "grad_norm": 0.8292503268402532,
      "learning_rate": 2.4517656856459536e-06,
      "loss": 0.8269,
      "num_input_tokens_seen": 78642615,
      "step": 3698
    },
    {
      "epoch": 0.4447784524739975,
      "flos": 18867253186560.0,
      "grad_norm": 10.302506106772183,
      "learning_rate": 2.4510068126827073e-06,
      "loss": 0.8817,
      "num_input_tokens_seen": 78663335,
      "step": 3699
    },
    {
      "epoch": 0.44489869536463655,
      "flos": 8195662602240.0,
      "grad_norm": 6.000531490945458,
      "learning_rate": 2.450247871302086e-06,
      "loss": 1.0167,
      "num_input_tokens_seen": 78680830,
      "step": 3700
    },
    {
      "epoch": 0.44501893825527566,
      "flos": 14514384199680.0,
      "grad_norm": 7.415780803910039,
      "learning_rate": 2.44948886161922e-06,
      "loss": 1.0307,
      "num_input_tokens_seen": 78699565,
      "step": 3701
    },
    {
      "epoch": 0.4451391811459148,
      "flos": 12942585077760.0,
      "grad_norm": 4.0892962040948095,
      "learning_rate": 2.4487297837492524e-06,
      "loss": 1.0409,
      "num_input_tokens_seen": 78718450,
      "step": 3702
    },
    {
      "epoch": 0.44525942403655383,
      "flos": 11971573678080.0,
      "grad_norm": 12.806627965574421,
      "learning_rate": 2.4479706378073323e-06,
      "loss": 0.8048,
      "num_input_tokens_seen": 78736710,
      "step": 3703
    },
    {
      "epoch": 0.44537966692719294,
      "flos": 16534273413120.0,
      "grad_norm": 3.7089058223399007,
      "learning_rate": 2.447211423908623e-06,
      "loss": 1.0304,
      "num_input_tokens_seen": 78756475,
      "step": 3704
    },
    {
      "epoch": 0.445499909817832,
      "flos": 15431021752320.0,
      "grad_norm": 13.199662082855095,
      "learning_rate": 2.4464521421682966e-06,
      "loss": 0.9389,
      "num_input_tokens_seen": 78773785,
      "step": 3705
    },
    {
      "epoch": 0.4456201527084711,
      "flos": 17058206453760.0,
      "grad_norm": 2.678140176843574,
      "learning_rate": 2.4456927927015345e-06,
      "loss": 1.069,
      "num_input_tokens_seen": 78794545,
      "step": 3706
    },
    {
      "epoch": 0.4457403955991102,
      "flos": 13334799052800.0,
      "grad_norm": 5.425603370689306,
      "learning_rate": 2.4449333756235307e-06,
      "loss": 0.9483,
      "num_input_tokens_seen": 78810980,
      "step": 3707
    },
    {
      "epoch": 0.4458606384897493,
      "flos": 13622680166400.0,
      "grad_norm": 9.306553523005308,
      "learning_rate": 2.4441738910494876e-06,
      "loss": 0.9865,
      "num_input_tokens_seen": 78825435,
      "step": 3708
    },
    {
      "epoch": 0.4459808813803884,
      "flos": 15170496184320.0,
      "grad_norm": 8.31273725357494,
      "learning_rate": 2.4434143390946176e-06,
      "loss": 1.0133,
      "num_input_tokens_seen": 78843965,
      "step": 3709
    },
    {
      "epoch": 0.4461011242710275,
      "flos": 16555068026880.0,
      "grad_norm": 4.669403340944623,
      "learning_rate": 2.4426547198741457e-06,
      "loss": 1.044,
      "num_input_tokens_seen": 78861890,
      "step": 3710
    },
    {
      "epoch": 0.44622136716166655,
      "flos": 14330774630400.0,
      "grad_norm": 5.527530432470251,
      "learning_rate": 2.441895033503305e-06,
      "loss": 0.9333,
      "num_input_tokens_seen": 78879530,
      "step": 3711
    },
    {
      "epoch": 0.44634161005230566,
      "flos": 15113547018240.0,
      "grad_norm": 4.339056978919336,
      "learning_rate": 2.4411352800973375e-06,
      "loss": 1.0252,
      "num_input_tokens_seen": 78897685,
      "step": 3712
    },
    {
      "epoch": 0.44646185294294477,
      "flos": 16297332387840.0,
      "grad_norm": 7.40797217582602,
      "learning_rate": 2.4403754597715005e-06,
      "loss": 0.9251,
      "num_input_tokens_seen": 78916850,
      "step": 3713
    },
    {
      "epoch": 0.4465820958335838,
      "flos": 16087746908160.0,
      "grad_norm": 4.542737828664611,
      "learning_rate": 2.4396155726410553e-06,
      "loss": 1.1265,
      "num_input_tokens_seen": 78935180,
      "step": 3714
    },
    {
      "epoch": 0.44670233872422294,
      "flos": 16112557793280.0,
      "grad_norm": 5.85740852418035,
      "learning_rate": 2.438855618821278e-06,
      "loss": 1.1176,
      "num_input_tokens_seen": 78950700,
      "step": 3715
    },
    {
      "epoch": 0.44682258161486205,
      "flos": 16765971824640.0,
      "grad_norm": 3.7470594957248746,
      "learning_rate": 2.4380955984274517e-06,
      "loss": 0.8573,
      "num_input_tokens_seen": 78969075,
      "step": 3716
    },
    {
      "epoch": 0.4469428245055011,
      "flos": 18863972720640.0,
      "grad_norm": 3.652813462025469,
      "learning_rate": 2.4373355115748716e-06,
      "loss": 0.9544,
      "num_input_tokens_seen": 78989625,
      "step": 3717
    },
    {
      "epoch": 0.4470630673961402,
      "flos": 15273418752000.0,
      "grad_norm": 5.613221446271544,
      "learning_rate": 2.436575358378842e-06,
      "loss": 0.919,
      "num_input_tokens_seen": 79008835,
      "step": 3718
    },
    {
      "epoch": 0.44718331028677927,
      "flos": 11441631559680.0,
      "grad_norm": 5.231572098198814,
      "learning_rate": 2.4358151389546782e-06,
      "loss": 1.0218,
      "num_input_tokens_seen": 79025240,
      "step": 3719
    },
    {
      "epoch": 0.4473035531774184,
      "flos": 13962850344960.0,
      "grad_norm": 8.136414476878054,
      "learning_rate": 2.4350548534177035e-06,
      "loss": 0.9455,
      "num_input_tokens_seen": 79041790,
      "step": 3720
    },
    {
      "epoch": 0.4474237960680575,
      "flos": 29591654031360.0,
      "grad_norm": 3.7657680880004465,
      "learning_rate": 2.434294501883254e-06,
      "loss": 0.8684,
      "num_input_tokens_seen": 79064605,
      "step": 3721
    },
    {
      "epoch": 0.44754403895869654,
      "flos": 16269792890880.0,
      "grad_norm": 4.301097667671659,
      "learning_rate": 2.433534084466674e-06,
      "loss": 0.8534,
      "num_input_tokens_seen": 79083545,
      "step": 3722
    },
    {
      "epoch": 0.44766428184933565,
      "flos": 18238558064640.0,
      "grad_norm": 3.249508013533293,
      "learning_rate": 2.4327736012833178e-06,
      "loss": 0.9121,
      "num_input_tokens_seen": 79104985,
      "step": 3723
    },
    {
      "epoch": 0.44778452473997477,
      "flos": 14514813419520.0,
      "grad_norm": 5.494041856948609,
      "learning_rate": 2.4320130524485506e-06,
      "loss": 0.9517,
      "num_input_tokens_seen": 79123500,
      "step": 3724
    },
    {
      "epoch": 0.4479047676306138,
      "flos": 15611074928640.0,
      "grad_norm": 4.2395836582543005,
      "learning_rate": 2.431252438077746e-06,
      "loss": 0.9936,
      "num_input_tokens_seen": 79142720,
      "step": 3725
    },
    {
      "epoch": 0.44802501052125293,
      "flos": 15246155182080.0,
      "grad_norm": 5.660137379195596,
      "learning_rate": 2.4304917582862906e-06,
      "loss": 0.9598,
      "num_input_tokens_seen": 79161620,
      "step": 3726
    },
    {
      "epoch": 0.44814525341189204,
      "flos": 15720711720960.0,
      "grad_norm": 6.136293548286472,
      "learning_rate": 2.4297310131895774e-06,
      "loss": 1.0769,
      "num_input_tokens_seen": 79179885,
      "step": 3727
    },
    {
      "epoch": 0.4482654963025311,
      "flos": 11787320279040.0,
      "grad_norm": 9.829924644577211,
      "learning_rate": 2.4289702029030113e-06,
      "loss": 0.9391,
      "num_input_tokens_seen": 79197075,
      "step": 3728
    },
    {
      "epoch": 0.4483857391931702,
      "flos": 13359058083840.0,
      "grad_norm": 3.528107857134573,
      "learning_rate": 2.4282093275420057e-06,
      "loss": 1.0192,
      "num_input_tokens_seen": 79215825,
      "step": 3729
    },
    {
      "epoch": 0.4485059820838093,
      "flos": 14459458498560.0,
      "grad_norm": 4.519877929051876,
      "learning_rate": 2.4274483872219863e-06,
      "loss": 0.8906,
      "num_input_tokens_seen": 79232905,
      "step": 3730
    },
    {
      "epoch": 0.4486262249744484,
      "flos": 14226472427520.0,
      "grad_norm": 4.580494134225245,
      "learning_rate": 2.426687382058386e-06,
      "loss": 1.1228,
      "num_input_tokens_seen": 79250905,
      "step": 3731
    },
    {
      "epoch": 0.4487464678650875,
      "flos": 46246999879680.0,
      "grad_norm": 0.9324070878568314,
      "learning_rate": 2.425926312166649e-06,
      "loss": 0.8108,
      "num_input_tokens_seen": 79303500,
      "step": 3732
    },
    {
      "epoch": 0.4488667107557266,
      "flos": 14744365731840.0,
      "grad_norm": 11.0596347697487,
      "learning_rate": 2.42516517766223e-06,
      "loss": 0.9149,
      "num_input_tokens_seen": 79321300,
      "step": 3733
    },
    {
      "epoch": 0.44898695364636565,
      "flos": 17032199884800.0,
      "grad_norm": 4.055010891110697,
      "learning_rate": 2.4244039786605907e-06,
      "loss": 0.8768,
      "num_input_tokens_seen": 79342025,
      "step": 3734
    },
    {
      "epoch": 0.44910719653700476,
      "flos": 13203447889920.0,
      "grad_norm": 4.085504979422941,
      "learning_rate": 2.4236427152772055e-06,
      "loss": 1.018,
      "num_input_tokens_seen": 79360150,
      "step": 3735
    },
    {
      "epoch": 0.4492274394276438,
      "flos": 40810846064640.0,
      "grad_norm": 0.940804855286224,
      "learning_rate": 2.422881387627557e-06,
      "loss": 0.8043,
      "num_input_tokens_seen": 79412320,
      "step": 3736
    },
    {
      "epoch": 0.4493476823182829,
      "flos": 16530900971520.0,
      "grad_norm": 3.288376689095951,
      "learning_rate": 2.422119995827139e-06,
      "loss": 0.9674,
      "num_input_tokens_seen": 79432165,
      "step": 3737
    },
    {
      "epoch": 0.44946792520892204,
      "flos": 11184539750400.0,
      "grad_norm": 13.516661358388163,
      "learning_rate": 2.4213585399914528e-06,
      "loss": 0.9299,
      "num_input_tokens_seen": 79449090,
      "step": 3738
    },
    {
      "epoch": 0.4495881680995611,
      "flos": 13911940915200.0,
      "grad_norm": 7.158413888865108,
      "learning_rate": 2.4205970202360113e-06,
      "loss": 1.0412,
      "num_input_tokens_seen": 79468375,
      "step": 3739
    },
    {
      "epoch": 0.4497084109902002,
      "flos": 18526868398080.0,
      "grad_norm": 5.503144319100725,
      "learning_rate": 2.4198354366763354e-06,
      "loss": 0.9768,
      "num_input_tokens_seen": 79486735,
      "step": 3740
    },
    {
      "epoch": 0.4498286538808393,
      "flos": 10450469376000.0,
      "grad_norm": 13.554994255354567,
      "learning_rate": 2.4190737894279587e-06,
      "loss": 0.9748,
      "num_input_tokens_seen": 79503825,
      "step": 3741
    },
    {
      "epoch": 0.44994889677147837,
      "flos": 10922235985920.0,
      "grad_norm": 6.882126322563875,
      "learning_rate": 2.4183120786064203e-06,
      "loss": 0.9949,
      "num_input_tokens_seen": 79520420,
      "step": 3742
    },
    {
      "epoch": 0.4500691396621175,
      "flos": 15484475842560.0,
      "grad_norm": 9.52202061791242,
      "learning_rate": 2.417550304327273e-06,
      "loss": 1.049,
      "num_input_tokens_seen": 79538180,
      "step": 3743
    },
    {
      "epoch": 0.4501893825527566,
      "flos": 23091162378240.0,
      "grad_norm": 18.547677129611053,
      "learning_rate": 2.4167884667060763e-06,
      "loss": 0.9502,
      "num_input_tokens_seen": 79560610,
      "step": 3744
    },
    {
      "epoch": 0.45030962544339564,
      "flos": 11709055303680.0,
      "grad_norm": 7.437695156069462,
      "learning_rate": 2.4160265658584e-06,
      "loss": 1.0742,
      "num_input_tokens_seen": 79575220,
      "step": 3745
    },
    {
      "epoch": 0.45042986833403476,
      "flos": 13885259857920.0,
      "grad_norm": 4.873686980441521,
      "learning_rate": 2.4152646018998253e-06,
      "loss": 0.862,
      "num_input_tokens_seen": 79593890,
      "step": 3746
    },
    {
      "epoch": 0.45055011122467387,
      "flos": 16429020794880.0,
      "grad_norm": 5.329376163296882,
      "learning_rate": 2.4145025749459403e-06,
      "loss": 0.9042,
      "num_input_tokens_seen": 79614635,
      "step": 3747
    },
    {
      "epoch": 0.4506703541153129,
      "flos": 14144681717760.0,
      "grad_norm": 4.477440766198849,
      "learning_rate": 2.413740485112344e-06,
      "loss": 0.8995,
      "num_input_tokens_seen": 79632695,
      "step": 3748
    },
    {
      "epoch": 0.45079059700595203,
      "flos": 13832357621760.0,
      "grad_norm": 5.253511072208322,
      "learning_rate": 2.412978332514646e-06,
      "loss": 1.0153,
      "num_input_tokens_seen": 79651195,
      "step": 3749
    },
    {
      "epoch": 0.4509108398965911,
      "flos": 19680171048960.0,
      "grad_norm": 5.521474702635674,
      "learning_rate": 2.4122161172684623e-06,
      "loss": 0.9021,
      "num_input_tokens_seen": 79671710,
      "step": 3750
    },
    {
      "epoch": 0.4510310827872302,
      "flos": 14907671224320.0,
      "grad_norm": 6.770005708632887,
      "learning_rate": 2.4114538394894216e-06,
      "loss": 1.0189,
      "num_input_tokens_seen": 79689070,
      "step": 3751
    },
    {
      "epoch": 0.4511513256778693,
      "flos": 11473156669440.0,
      "grad_norm": 6.4232639564894365,
      "learning_rate": 2.410691499293161e-06,
      "loss": 1.0254,
      "num_input_tokens_seen": 79706945,
      "step": 3752
    },
    {
      "epoch": 0.45127156856850836,
      "flos": 17918998548480.0,
      "grad_norm": 4.664643166848182,
      "learning_rate": 2.409929096795326e-06,
      "loss": 0.9266,
      "num_input_tokens_seen": 79727035,
      "step": 3753
    },
    {
      "epoch": 0.4513918114591475,
      "flos": 14488622899200.0,
      "grad_norm": 6.85344771892121,
      "learning_rate": 2.409166632111573e-06,
      "loss": 0.9803,
      "num_input_tokens_seen": 79744890,
      "step": 3754
    },
    {
      "epoch": 0.4515120543497866,
      "flos": 18969869168640.0,
      "grad_norm": 15.584549201953998,
      "learning_rate": 2.4084041053575674e-06,
      "loss": 0.9799,
      "num_input_tokens_seen": 79764030,
      "step": 3755
    },
    {
      "epoch": 0.45163229724042564,
      "flos": 14619851427840.0,
      "grad_norm": 4.8480949357056975,
      "learning_rate": 2.4076415166489834e-06,
      "loss": 0.9163,
      "num_input_tokens_seen": 79783160,
      "step": 3756
    },
    {
      "epoch": 0.45175254013106475,
      "flos": 15405934940160.0,
      "grad_norm": 3.939917654735773,
      "learning_rate": 2.406878866101506e-06,
      "loss": 0.9921,
      "num_input_tokens_seen": 79801845,
      "step": 3757
    },
    {
      "epoch": 0.45187278302170386,
      "flos": 13386229678080.0,
      "grad_norm": 5.447822506773474,
      "learning_rate": 2.4061161538308273e-06,
      "loss": 0.9756,
      "num_input_tokens_seen": 79818410,
      "step": 3758
    },
    {
      "epoch": 0.4519930259123429,
      "flos": 13172627927040.0,
      "grad_norm": 5.876896978067619,
      "learning_rate": 2.4053533799526523e-06,
      "loss": 1.0899,
      "num_input_tokens_seen": 79833850,
      "step": 3759
    },
    {
      "epoch": 0.452113268802982,
      "flos": 17923842600960.0,
      "grad_norm": 5.1515529624380925,
      "learning_rate": 2.404590544582691e-06,
      "loss": 1.0444,
      "num_input_tokens_seen": 79851805,
      "step": 3760
    },
    {
      "epoch": 0.45223351169362114,
      "flos": 28118577254400.0,
      "grad_norm": 5.242900771438031,
      "learning_rate": 2.403827647836666e-06,
      "loss": 1.0024,
      "num_input_tokens_seen": 79872080,
      "step": 3761
    },
    {
      "epoch": 0.4523537545842602,
      "flos": 15329509478400.0,
      "grad_norm": 6.964382017259264,
      "learning_rate": 2.4030646898303075e-06,
      "loss": 0.8801,
      "num_input_tokens_seen": 79893290,
      "step": 3762
    },
    {
      "epoch": 0.4524739974748993,
      "flos": 20257496862720.0,
      "grad_norm": 7.9949362941587125,
      "learning_rate": 2.4023016706793566e-06,
      "loss": 1.0125,
      "num_input_tokens_seen": 79912805,
      "step": 3763
    },
    {
      "epoch": 0.4525942403655384,
      "flos": 44061321830400.0,
      "grad_norm": 0.981345677418877,
      "learning_rate": 2.401538590499561e-06,
      "loss": 0.7958,
      "num_input_tokens_seen": 79972980,
      "step": 3764
    },
    {
      "epoch": 0.45271448325617747,
      "flos": 19604052172800.0,
      "grad_norm": 9.913417978581677,
      "learning_rate": 2.400775449406682e-06,
      "loss": 0.8989,
      "num_input_tokens_seen": 79995895,
      "step": 3765
    },
    {
      "epoch": 0.4528347261468166,
      "flos": 15954372280320.0,
      "grad_norm": 6.370506348419559,
      "learning_rate": 2.400012247516485e-06,
      "loss": 0.9254,
      "num_input_tokens_seen": 80016180,
      "step": 3766
    },
    {
      "epoch": 0.45295496903745563,
      "flos": 14985231052800.0,
      "grad_norm": 5.308321397768952,
      "learning_rate": 2.3992489849447484e-06,
      "loss": 1.0848,
      "num_input_tokens_seen": 80034355,
      "step": 3767
    },
    {
      "epoch": 0.45307521192809475,
      "flos": 16507040501760.0,
      "grad_norm": 5.334158336885386,
      "learning_rate": 2.3984856618072584e-06,
      "loss": 0.9884,
      "num_input_tokens_seen": 80054110,
      "step": 3768
    },
    {
      "epoch": 0.45319545481873386,
      "flos": 11001298083840.0,
      "grad_norm": 5.3401398538284806,
      "learning_rate": 2.3977222782198098e-06,
      "loss": 0.9259,
      "num_input_tokens_seen": 80072465,
      "step": 3769
    },
    {
      "epoch": 0.4533156977093729,
      "flos": 15589789777920.0,
      "grad_norm": 5.388290828714572,
      "learning_rate": 2.3969588342982077e-06,
      "loss": 0.9452,
      "num_input_tokens_seen": 80091560,
      "step": 3770
    },
    {
      "epoch": 0.453435940600012,
      "flos": 17240742973440.0,
      "grad_norm": 4.46106448552295,
      "learning_rate": 2.396195330158267e-06,
      "loss": 0.9208,
      "num_input_tokens_seen": 80111170,
      "step": 3771
    },
    {
      "epoch": 0.45355618349065113,
      "flos": 16664612843520.0,
      "grad_norm": 17.87803954844331,
      "learning_rate": 2.3954317659158094e-06,
      "loss": 0.9848,
      "num_input_tokens_seen": 80131225,
      "step": 3772
    },
    {
      "epoch": 0.4536764263812902,
      "flos": 47906537472000.0,
      "grad_norm": 0.9539710182989969,
      "learning_rate": 2.394668141686667e-06,
      "loss": 0.8164,
      "num_input_tokens_seen": 80192910,
      "step": 3773
    },
    {
      "epoch": 0.4537966692719293,
      "flos": 30534512762880.0,
      "grad_norm": 5.533424941227648,
      "learning_rate": 2.3939044575866813e-06,
      "loss": 0.8777,
      "num_input_tokens_seen": 80215380,
      "step": 3774
    },
    {
      "epoch": 0.4539169121625684,
      "flos": 25366641131520.0,
      "grad_norm": 4.792338758133139,
      "learning_rate": 2.3931407137317024e-06,
      "loss": 0.9459,
      "num_input_tokens_seen": 80235255,
      "step": 3775
    },
    {
      "epoch": 0.45403715505320746,
      "flos": 13124631060480.0,
      "grad_norm": 4.773509622831266,
      "learning_rate": 2.3923769102375907e-06,
      "loss": 1.0448,
      "num_input_tokens_seen": 80253840,
      "step": 3776
    },
    {
      "epoch": 0.4541573979438466,
      "flos": 17818436689920.0,
      "grad_norm": 5.578478035320365,
      "learning_rate": 2.391613047220213e-06,
      "loss": 0.983,
      "num_input_tokens_seen": 80273460,
      "step": 3777
    },
    {
      "epoch": 0.4542776408344857,
      "flos": 12993617141760.0,
      "grad_norm": 4.653663275169585,
      "learning_rate": 2.390849124795447e-06,
      "loss": 0.991,
      "num_input_tokens_seen": 80289180,
      "step": 3778
    },
    {
      "epoch": 0.45439788372512474,
      "flos": 14695633059840.0,
      "grad_norm": 4.432239261710369,
      "learning_rate": 2.3900851430791804e-06,
      "loss": 1.0267,
      "num_input_tokens_seen": 80306920,
      "step": 3779
    },
    {
      "epoch": 0.45451812661576385,
      "flos": 15851848273920.0,
      "grad_norm": 8.42464615166487,
      "learning_rate": 2.389321102187307e-06,
      "loss": 1.0412,
      "num_input_tokens_seen": 80325420,
      "step": 3780
    },
    {
      "epoch": 0.4546383695064029,
      "flos": 15459511664640.0,
      "grad_norm": 5.9276774219875215,
      "learning_rate": 2.3885570022357326e-06,
      "loss": 1.0136,
      "num_input_tokens_seen": 80344270,
      "step": 3781
    },
    {
      "epoch": 0.454758612397042,
      "flos": 45994966732800.0,
      "grad_norm": 0.836279576267506,
      "learning_rate": 2.38779284334037e-06,
      "loss": 0.8212,
      "num_input_tokens_seen": 80408965,
      "step": 3782
    },
    {
      "epoch": 0.4548788552876811,
      "flos": 19440961290240.0,
      "grad_norm": 6.3340494186068295,
      "learning_rate": 2.387028625617141e-06,
      "loss": 0.9749,
      "num_input_tokens_seen": 80427900,
      "step": 3783
    },
    {
      "epoch": 0.4549990981783202,
      "flos": 16245533859840.0,
      "grad_norm": 3.8784186756911887,
      "learning_rate": 2.3862643491819766e-06,
      "loss": 1.0424,
      "num_input_tokens_seen": 80446185,
      "step": 3784
    },
    {
      "epoch": 0.4551193410689593,
      "flos": 16534058803200.0,
      "grad_norm": 8.163665791049338,
      "learning_rate": 2.3855000141508186e-06,
      "loss": 1.0248,
      "num_input_tokens_seen": 80466060,
      "step": 3785
    },
    {
      "epoch": 0.4552395839595984,
      "flos": 14750957322240.0,
      "grad_norm": 19.160671103556677,
      "learning_rate": 2.3847356206396143e-06,
      "loss": 1.0264,
      "num_input_tokens_seen": 80483090,
      "step": 3786
    },
    {
      "epoch": 0.45535982685023746,
      "flos": 16533139046400.0,
      "grad_norm": 2.787830925016704,
      "learning_rate": 2.3839711687643227e-06,
      "loss": 0.9779,
      "num_input_tokens_seen": 80504035,
      "step": 3787
    },
    {
      "epoch": 0.45548006974087657,
      "flos": 13937794191360.0,
      "grad_norm": 6.081128317388873,
      "learning_rate": 2.38320665864091e-06,
      "loss": 0.9176,
      "num_input_tokens_seen": 80523105,
      "step": 3788
    },
    {
      "epoch": 0.4556003126315157,
      "flos": 14226503086080.0,
      "grad_norm": 6.108578212907085,
      "learning_rate": 2.3824420903853516e-06,
      "loss": 1.0101,
      "num_input_tokens_seen": 80541290,
      "step": 3789
    },
    {
      "epoch": 0.45572055552215474,
      "flos": 16318341611520.0,
      "grad_norm": 16.43615670752908,
      "learning_rate": 2.3816774641136324e-06,
      "loss": 0.9987,
      "num_input_tokens_seen": 80558265,
      "step": 3790
    },
    {
      "epoch": 0.45584079841279385,
      "flos": 23614053027840.0,
      "grad_norm": 4.8581435430936715,
      "learning_rate": 2.380912779941745e-06,
      "loss": 0.9158,
      "num_input_tokens_seen": 80581105,
      "step": 3791
    },
    {
      "epoch": 0.45596104130343296,
      "flos": 19418572431360.0,
      "grad_norm": 4.209637103172348,
      "learning_rate": 2.3801480379856918e-06,
      "loss": 1.0042,
      "num_input_tokens_seen": 80602535,
      "step": 3792
    },
    {
      "epoch": 0.456081284194072,
      "flos": 15327486013440.0,
      "grad_norm": 5.176659528961141,
      "learning_rate": 2.379383238361484e-06,
      "loss": 1.0293,
      "num_input_tokens_seen": 80621615,
      "step": 3793
    },
    {
      "epoch": 0.4562015270847111,
      "flos": 25632685240320.0,
      "grad_norm": 8.862871075651123,
      "learning_rate": 2.3786183811851407e-06,
      "loss": 0.9792,
      "num_input_tokens_seen": 80642040,
      "step": 3794
    },
    {
      "epoch": 0.45632176997535023,
      "flos": 9585262448640.0,
      "grad_norm": 4.18465566114099,
      "learning_rate": 2.3778534665726892e-06,
      "loss": 0.9871,
      "num_input_tokens_seen": 80658590,
      "step": 3795
    },
    {
      "epoch": 0.4564420128659893,
      "flos": 23272993751040.0,
      "grad_norm": 2.990200249335808,
      "learning_rate": 2.377088494640168e-06,
      "loss": 0.9211,
      "num_input_tokens_seen": 80680060,
      "step": 3796
    },
    {
      "epoch": 0.4565622557566284,
      "flos": 14463812014080.0,
      "grad_norm": 3.276259923961386,
      "learning_rate": 2.3763234655036216e-06,
      "loss": 0.9733,
      "num_input_tokens_seen": 80698980,
      "step": 3797
    },
    {
      "epoch": 0.45668249864726745,
      "flos": 18054580592640.0,
      "grad_norm": 7.366556797963156,
      "learning_rate": 2.3755583792791046e-06,
      "loss": 1.0616,
      "num_input_tokens_seen": 80718570,
      "step": 3798
    },
    {
      "epoch": 0.45680274153790656,
      "flos": 11001052815360.0,
      "grad_norm": 4.6234959399161255,
      "learning_rate": 2.3747932360826803e-06,
      "loss": 0.9387,
      "num_input_tokens_seen": 80735220,
      "step": 3799
    },
    {
      "epoch": 0.4569229844285457,
      "flos": 14043384053760.0,
      "grad_norm": 4.6691000361638295,
      "learning_rate": 2.3740280360304205e-06,
      "loss": 1.0063,
      "num_input_tokens_seen": 80752665,
      "step": 3800
    },
    {
      "epoch": 0.45704322731918473,
      "flos": 17133896110080.0,
      "grad_norm": 6.421053126107197,
      "learning_rate": 2.3732627792384038e-06,
      "loss": 0.8754,
      "num_input_tokens_seen": 80773455,
      "step": 3801
    },
    {
      "epoch": 0.45716347020982384,
      "flos": 22325505576960.0,
      "grad_norm": 4.778667874679527,
      "learning_rate": 2.3724974658227207e-06,
      "loss": 0.9459,
      "num_input_tokens_seen": 80793965,
      "step": 3802
    },
    {
      "epoch": 0.45728371310046295,
      "flos": 18864003379200.0,
      "grad_norm": 4.828017742854451,
      "learning_rate": 2.3717320958994687e-06,
      "loss": 0.8995,
      "num_input_tokens_seen": 80811245,
      "step": 3803
    },
    {
      "epoch": 0.457403955991102,
      "flos": 12703773880320.0,
      "grad_norm": 9.427077038729482,
      "learning_rate": 2.3709666695847534e-06,
      "loss": 0.9025,
      "num_input_tokens_seen": 80829145,
      "step": 3804
    },
    {
      "epoch": 0.4575241988817411,
      "flos": 30168857210880.0,
      "grad_norm": 3.857212464648081,
      "learning_rate": 2.370201186994689e-06,
      "loss": 0.902,
      "num_input_tokens_seen": 80852550,
      "step": 3805
    },
    {
      "epoch": 0.45764444177238023,
      "flos": 21463211212800.0,
      "grad_norm": 4.5649645693888345,
      "learning_rate": 2.369435648245399e-06,
      "loss": 0.8889,
      "num_input_tokens_seen": 80872485,
      "step": 3806
    },
    {
      "epoch": 0.4577646846630193,
      "flos": 17110096957440.0,
      "grad_norm": 7.793162836320579,
      "learning_rate": 2.368670053453015e-06,
      "loss": 1.0498,
      "num_input_tokens_seen": 80893895,
      "step": 3807
    },
    {
      "epoch": 0.4578849275536584,
      "flos": 12336646717440.0,
      "grad_norm": 5.32507433772248,
      "learning_rate": 2.3679044027336757e-06,
      "loss": 0.9175,
      "num_input_tokens_seen": 80909505,
      "step": 3808
    },
    {
      "epoch": 0.4580051704442975,
      "flos": 9527148257280.0,
      "grad_norm": 6.830314372522705,
      "learning_rate": 2.3671386962035326e-06,
      "loss": 0.8697,
      "num_input_tokens_seen": 80926695,
      "step": 3809
    },
    {
      "epoch": 0.45812541333493656,
      "flos": 12781364367360.0,
      "grad_norm": 6.91603961844539,
      "learning_rate": 2.3663729339787405e-06,
      "loss": 0.872,
      "num_input_tokens_seen": 80943350,
      "step": 3810
    },
    {
      "epoch": 0.45824565622557567,
      "flos": 14350772121600.0,
      "grad_norm": 5.552638366706488,
      "learning_rate": 2.365607116175466e-06,
      "loss": 0.9066,
      "num_input_tokens_seen": 80958495,
      "step": 3811
    },
    {
      "epoch": 0.4583658991162148,
      "flos": 14094998630400.0,
      "grad_norm": 13.76271495737839,
      "learning_rate": 2.3648412429098825e-06,
      "loss": 0.8371,
      "num_input_tokens_seen": 80976565,
      "step": 3812
    },
    {
      "epoch": 0.45848614200685384,
      "flos": 14931960913920.0,
      "grad_norm": 3.6682760579133826,
      "learning_rate": 2.364075314298172e-06,
      "loss": 1.0065,
      "num_input_tokens_seen": 80993740,
      "step": 3813
    },
    {
      "epoch": 0.45860638489749295,
      "flos": 14960634777600.0,
      "grad_norm": 6.241615164136422,
      "learning_rate": 2.3633093304565267e-06,
      "loss": 0.8857,
      "num_input_tokens_seen": 81012515,
      "step": 3814
    },
    {
      "epoch": 0.458726627788132,
      "flos": 19103519723520.0,
      "grad_norm": 22.210877731409205,
      "learning_rate": 2.3625432915011443e-06,
      "loss": 0.8115,
      "num_input_tokens_seen": 81034145,
      "step": 3815
    },
    {
      "epoch": 0.4588468706787711,
      "flos": 17137299210240.0,
      "grad_norm": 5.982655888338513,
      "learning_rate": 2.3617771975482334e-06,
      "loss": 0.8377,
      "num_input_tokens_seen": 81052695,
      "step": 3816
    },
    {
      "epoch": 0.4589671135694102,
      "flos": 12674486845440.0,
      "grad_norm": 3.31345195406976,
      "learning_rate": 2.3610110487140083e-06,
      "loss": 0.9373,
      "num_input_tokens_seen": 81070850,
      "step": 3817
    },
    {
      "epoch": 0.4590873564600493,
      "flos": 18235737477120.0,
      "grad_norm": 4.910997559935395,
      "learning_rate": 2.360244845114695e-06,
      "loss": 1.006,
      "num_input_tokens_seen": 81090190,
      "step": 3818
    },
    {
      "epoch": 0.4592075993506884,
      "flos": 13124692377600.0,
      "grad_norm": 6.733745076397402,
      "learning_rate": 2.3594785868665245e-06,
      "loss": 0.8822,
      "num_input_tokens_seen": 81106215,
      "step": 3819
    },
    {
      "epoch": 0.4593278422413275,
      "flos": 14645674045440.0,
      "grad_norm": 14.82924867317753,
      "learning_rate": 2.3587122740857386e-06,
      "loss": 1.0016,
      "num_input_tokens_seen": 81123035,
      "step": 3820
    },
    {
      "epoch": 0.45944808513196655,
      "flos": 15168135475200.0,
      "grad_norm": 3.3992165560442387,
      "learning_rate": 2.357945906888586e-06,
      "loss": 0.9743,
      "num_input_tokens_seen": 81142195,
      "step": 3821
    },
    {
      "epoch": 0.45956832802260567,
      "flos": 15217420001280.0,
      "grad_norm": 4.024395697111395,
      "learning_rate": 2.357179485391324e-06,
      "loss": 0.9826,
      "num_input_tokens_seen": 81159770,
      "step": 3822
    },
    {
      "epoch": 0.4596885709132448,
      "flos": 15904505241600.0,
      "grad_norm": 4.663216518950445,
      "learning_rate": 2.3564130097102173e-06,
      "loss": 1.0469,
      "num_input_tokens_seen": 81179145,
      "step": 3823
    },
    {
      "epoch": 0.45980881380388383,
      "flos": 20646675640320.0,
      "grad_norm": 3.580670113741273,
      "learning_rate": 2.355646479961541e-06,
      "loss": 0.934,
      "num_input_tokens_seen": 81198175,
      "step": 3824
    },
    {
      "epoch": 0.45992905669452294,
      "flos": 23819376967680.0,
      "grad_norm": 4.545600004469681,
      "learning_rate": 2.354879896261576e-06,
      "loss": 0.9096,
      "num_input_tokens_seen": 81218105,
      "step": 3825
    },
    {
      "epoch": 0.46004929958516205,
      "flos": 25919431987200.0,
      "grad_norm": 4.751683589316885,
      "learning_rate": 2.3541132587266133e-06,
      "loss": 0.7721,
      "num_input_tokens_seen": 81240545,
      "step": 3826
    },
    {
      "epoch": 0.4601695424758011,
      "flos": 12207502970880.0,
      "grad_norm": 5.555600291177985,
      "learning_rate": 2.3533465674729515e-06,
      "loss": 0.892,
      "num_input_tokens_seen": 81257495,
      "step": 3827
    },
    {
      "epoch": 0.4602897853664402,
      "flos": 11237258035200.0,
      "grad_norm": 4.875259588128179,
      "learning_rate": 2.352579822616895e-06,
      "loss": 0.9228,
      "num_input_tokens_seen": 81274650,
      "step": 3828
    },
    {
      "epoch": 0.4604100282570793,
      "flos": 18082304040960.0,
      "grad_norm": 7.2884663852236535,
      "learning_rate": 2.351813024274761e-06,
      "loss": 0.973,
      "num_input_tokens_seen": 81295725,
      "step": 3829
    },
    {
      "epoch": 0.4605302711477184,
      "flos": 19675388313600.0,
      "grad_norm": 5.65621206365688,
      "learning_rate": 2.3510461725628693e-06,
      "loss": 0.9215,
      "num_input_tokens_seen": 81315910,
      "step": 3830
    },
    {
      "epoch": 0.4606505140383575,
      "flos": 16951298273280.0,
      "grad_norm": 5.094694459567967,
      "learning_rate": 2.350279267597554e-06,
      "loss": 0.8954,
      "num_input_tokens_seen": 81336270,
      "step": 3831
    },
    {
      "epoch": 0.46077075692899655,
      "flos": 11394309181440.0,
      "grad_norm": 4.794836947122113,
      "learning_rate": 2.3495123094951515e-06,
      "loss": 1.0212,
      "num_input_tokens_seen": 81354335,
      "step": 3832
    },
    {
      "epoch": 0.46089099981963566,
      "flos": 34888792043520.0,
      "grad_norm": 3.7146298961677098,
      "learning_rate": 2.34874529837201e-06,
      "loss": 0.9401,
      "num_input_tokens_seen": 81377600,
      "step": 3833
    },
    {
      "epoch": 0.46101124271027477,
      "flos": 13544813752320.0,
      "grad_norm": 4.947410006097394,
      "learning_rate": 2.347978234344483e-06,
      "loss": 0.9971,
      "num_input_tokens_seen": 81393525,
      "step": 3834
    },
    {
      "epoch": 0.4611314856009138,
      "flos": 28097046835200.0,
      "grad_norm": 4.163845607698606,
      "learning_rate": 2.347211117528935e-06,
      "loss": 0.892,
      "num_input_tokens_seen": 81415545,
      "step": 3835
    },
    {
      "epoch": 0.46125172849155294,
      "flos": 14774419230720.0,
      "grad_norm": 3.0061150851678047,
      "learning_rate": 2.3464439480417374e-06,
      "loss": 0.9097,
      "num_input_tokens_seen": 81434785,
      "step": 3836
    },
    {
      "epoch": 0.46137197138219205,
      "flos": 12704969564160.0,
      "grad_norm": 6.238892718153824,
      "learning_rate": 2.3456767259992676e-06,
      "loss": 0.9399,
      "num_input_tokens_seen": 81452150,
      "step": 3837
    },
    {
      "epoch": 0.4614922142728311,
      "flos": 11918640783360.0,
      "grad_norm": 4.387914071826543,
      "learning_rate": 2.3449094515179135e-06,
      "loss": 1.0626,
      "num_input_tokens_seen": 81469330,
      "step": 3838
    },
    {
      "epoch": 0.4616124571634702,
      "flos": 18946131333120.0,
      "grad_norm": 3.1992090618652913,
      "learning_rate": 2.34414212471407e-06,
      "loss": 1.0084,
      "num_input_tokens_seen": 81488845,
      "step": 3839
    },
    {
      "epoch": 0.4617327000541093,
      "flos": 14436671078400.0,
      "grad_norm": 7.014573322134537,
      "learning_rate": 2.3433747457041394e-06,
      "loss": 0.9339,
      "num_input_tokens_seen": 81507270,
      "step": 3840
    },
    {
      "epoch": 0.4618529429447484,
      "flos": 21070506700800.0,
      "grad_norm": 6.402880866636048,
      "learning_rate": 2.342607314604533e-06,
      "loss": 1.0418,
      "num_input_tokens_seen": 81526740,
      "step": 3841
    },
    {
      "epoch": 0.4619731858353875,
      "flos": 14038386708480.0,
      "grad_norm": 6.680745073442641,
      "learning_rate": 2.3418398315316694e-06,
      "loss": 1.033,
      "num_input_tokens_seen": 81544280,
      "step": 3842
    },
    {
      "epoch": 0.4620934287260266,
      "flos": 13438089523200.0,
      "grad_norm": 4.480108803722124,
      "learning_rate": 2.3410722966019755e-06,
      "loss": 0.964,
      "num_input_tokens_seen": 81559115,
      "step": 3843
    },
    {
      "epoch": 0.46221367161666566,
      "flos": 26655249899520.0,
      "grad_norm": 4.9351206007795065,
      "learning_rate": 2.3403047099318848e-06,
      "loss": 0.852,
      "num_input_tokens_seen": 81582905,
      "step": 3844
    },
    {
      "epoch": 0.46233391450730477,
      "flos": 10188012318720.0,
      "grad_norm": 3.4556890435993,
      "learning_rate": 2.3395370716378405e-06,
      "loss": 0.9291,
      "num_input_tokens_seen": 81600070,
      "step": 3845
    },
    {
      "epoch": 0.4624541573979438,
      "flos": 15983965900800.0,
      "grad_norm": 4.916335897580322,
      "learning_rate": 2.338769381836292e-06,
      "loss": 0.9171,
      "num_input_tokens_seen": 81619400,
      "step": 3846
    },
    {
      "epoch": 0.46257440028858293,
      "flos": 10214264156160.0,
      "grad_norm": 5.760154713566186,
      "learning_rate": 2.3380016406436984e-06,
      "loss": 0.9129,
      "num_input_tokens_seen": 81636600,
      "step": 3847
    },
    {
      "epoch": 0.46269464317922204,
      "flos": 16587053015040.0,
      "grad_norm": 33.89961952738648,
      "learning_rate": 2.337233848176524e-06,
      "loss": 1.0169,
      "num_input_tokens_seen": 81654090,
      "step": 3848
    },
    {
      "epoch": 0.4628148860698611,
      "flos": 13151557386240.0,
      "grad_norm": 9.085483214943684,
      "learning_rate": 2.3364660045512435e-06,
      "loss": 1.0101,
      "num_input_tokens_seen": 81672570,
      "step": 3849
    },
    {
      "epoch": 0.4629351289605002,
      "flos": 50612806778880.0,
      "grad_norm": 0.8654568216163327,
      "learning_rate": 2.335698109884337e-06,
      "loss": 0.82,
      "num_input_tokens_seen": 81737495,
      "step": 3850
    },
    {
      "epoch": 0.4630553718511393,
      "flos": 42721672888320.0,
      "grad_norm": 0.8555023326117387,
      "learning_rate": 2.334930164292294e-06,
      "loss": 0.8486,
      "num_input_tokens_seen": 81799765,
      "step": 3851
    },
    {
      "epoch": 0.4631756147417784,
      "flos": 11289025904640.0,
      "grad_norm": 5.5809172151977515,
      "learning_rate": 2.334162167891612e-06,
      "loss": 0.9894,
      "num_input_tokens_seen": 81816750,
      "step": 3852
    },
    {
      "epoch": 0.4632958576324175,
      "flos": 11657410068480.0,
      "grad_norm": 4.246746613145721,
      "learning_rate": 2.333394120798795e-06,
      "loss": 0.9296,
      "num_input_tokens_seen": 81835205,
      "step": 3853
    },
    {
      "epoch": 0.4634161005230566,
      "flos": 15878559989760.0,
      "grad_norm": 3.717528839922991,
      "learning_rate": 2.3326260231303545e-06,
      "loss": 0.9279,
      "num_input_tokens_seen": 81853525,
      "step": 3854
    },
    {
      "epoch": 0.46353634341369565,
      "flos": 11132158709760.0,
      "grad_norm": 3.7244268091563693,
      "learning_rate": 2.331857875002811e-06,
      "loss": 1.069,
      "num_input_tokens_seen": 81871845,
      "step": 3855
    },
    {
      "epoch": 0.46365658630433476,
      "flos": 20178618716160.0,
      "grad_norm": 3.409301790593027,
      "learning_rate": 2.3310896765326916e-06,
      "loss": 0.9544,
      "num_input_tokens_seen": 81892565,
      "step": 3856
    },
    {
      "epoch": 0.46377682919497387,
      "flos": 17503843860480.0,
      "grad_norm": 4.91459131352045,
      "learning_rate": 2.330321427836531e-06,
      "loss": 1.0352,
      "num_input_tokens_seen": 81914155,
      "step": 3857
    },
    {
      "epoch": 0.4638970720856129,
      "flos": 13517059645440.0,
      "grad_norm": 6.589560095036219,
      "learning_rate": 2.3295531290308733e-06,
      "loss": 1.0291,
      "num_input_tokens_seen": 81932025,
      "step": 3858
    },
    {
      "epoch": 0.46401731497625204,
      "flos": 13092247511040.0,
      "grad_norm": 5.563261858597366,
      "learning_rate": 2.3287847802322678e-06,
      "loss": 0.924,
      "num_input_tokens_seen": 81947315,
      "step": 3859
    },
    {
      "epoch": 0.4641375578668911,
      "flos": 18552323112960.0,
      "grad_norm": 8.840221334595018,
      "learning_rate": 2.3280163815572723e-06,
      "loss": 1.0312,
      "num_input_tokens_seen": 81967630,
      "step": 3860
    },
    {
      "epoch": 0.4642578007575302,
      "flos": 13882991124480.0,
      "grad_norm": 4.882115965944648,
      "learning_rate": 2.3272479331224522e-06,
      "loss": 0.9609,
      "num_input_tokens_seen": 81984780,
      "step": 3861
    },
    {
      "epoch": 0.4643780436481693,
      "flos": 20075757465600.0,
      "grad_norm": 3.82036117803535,
      "learning_rate": 2.3264794350443817e-06,
      "loss": 0.9663,
      "num_input_tokens_seen": 82006595,
      "step": 3862
    },
    {
      "epoch": 0.46449828653880837,
      "flos": 18055347056640.0,
      "grad_norm": 4.004048444502495,
      "learning_rate": 2.3257108874396396e-06,
      "loss": 0.9758,
      "num_input_tokens_seen": 82027410,
      "step": 3863
    },
    {
      "epoch": 0.4646185294294475,
      "flos": 11342112092160.0,
      "grad_norm": 19.550332084625587,
      "learning_rate": 2.3249422904248152e-06,
      "loss": 0.9304,
      "num_input_tokens_seen": 82045565,
      "step": 3864
    },
    {
      "epoch": 0.4647387723200866,
      "flos": 18765648936960.0,
      "grad_norm": 3.381051607358079,
      "learning_rate": 2.324173644116504e-06,
      "loss": 1.0721,
      "num_input_tokens_seen": 82068135,
      "step": 3865
    },
    {
      "epoch": 0.46485901521072565,
      "flos": 19337762795520.0,
      "grad_norm": 4.391255723714424,
      "learning_rate": 2.3234049486313087e-06,
      "loss": 1.0079,
      "num_input_tokens_seen": 82089305,
      "step": 3866
    },
    {
      "epoch": 0.46497925810136476,
      "flos": 17267975884800.0,
      "grad_norm": 4.802219060716045,
      "learning_rate": 2.322636204085839e-06,
      "loss": 0.9587,
      "num_input_tokens_seen": 82109095,
      "step": 3867
    },
    {
      "epoch": 0.46509950099200387,
      "flos": 11499347189760.0,
      "grad_norm": 5.045571740591261,
      "learning_rate": 2.3218674105967143e-06,
      "loss": 0.963,
      "num_input_tokens_seen": 82127080,
      "step": 3868
    },
    {
      "epoch": 0.4652197438826429,
      "flos": 16666237747200.0,
      "grad_norm": 5.934852732733013,
      "learning_rate": 2.3210985682805593e-06,
      "loss": 1.0232,
      "num_input_tokens_seen": 82148580,
      "step": 3869
    },
    {
      "epoch": 0.46533998677328203,
      "flos": 11472911400960.0,
      "grad_norm": 7.387806780491684,
      "learning_rate": 2.320329677254007e-06,
      "loss": 0.8605,
      "num_input_tokens_seen": 82165630,
      "step": 3870
    },
    {
      "epoch": 0.46546022966392114,
      "flos": 15012739891200.0,
      "grad_norm": 4.859951954042907,
      "learning_rate": 2.319560737633697e-06,
      "loss": 0.9146,
      "num_input_tokens_seen": 82184070,
      "step": 3871
    },
    {
      "epoch": 0.4655804725545602,
      "flos": 29407891169280.0,
      "grad_norm": 3.186444425640867,
      "learning_rate": 2.3187917495362775e-06,
      "loss": 0.8836,
      "num_input_tokens_seen": 82208200,
      "step": 3872
    },
    {
      "epoch": 0.4657007154451993,
      "flos": 13883481661440.0,
      "grad_norm": 11.324993505749724,
      "learning_rate": 2.318022713078403e-06,
      "loss": 0.9476,
      "num_input_tokens_seen": 82222500,
      "step": 3873
    },
    {
      "epoch": 0.4658209583358384,
      "flos": 10970140876800.0,
      "grad_norm": 8.99952696510784,
      "learning_rate": 2.3172536283767354e-06,
      "loss": 1.0425,
      "num_input_tokens_seen": 82235980,
      "step": 3874
    },
    {
      "epoch": 0.4659412012264775,
      "flos": 10529163571200.0,
      "grad_norm": 6.353560955590671,
      "learning_rate": 2.3164844955479447e-06,
      "loss": 0.9974,
      "num_input_tokens_seen": 82251510,
      "step": 3875
    },
    {
      "epoch": 0.4660614441171166,
      "flos": 17372707307520.0,
      "grad_norm": 4.452710369951304,
      "learning_rate": 2.3157153147087082e-06,
      "loss": 0.8907,
      "num_input_tokens_seen": 82273120,
      "step": 3876
    },
    {
      "epoch": 0.46618168700775564,
      "flos": 15695011737600.0,
      "grad_norm": 5.641200493278988,
      "learning_rate": 2.314946085975709e-06,
      "loss": 1.0372,
      "num_input_tokens_seen": 82291820,
      "step": 3877
    },
    {
      "epoch": 0.46630192989839475,
      "flos": 18631385210880.0,
      "grad_norm": 4.136133601054429,
      "learning_rate": 2.3141768094656393e-06,
      "loss": 1.0184,
      "num_input_tokens_seen": 82310115,
      "step": 3878
    },
    {
      "epoch": 0.46642217278903386,
      "flos": 8090226032640.0,
      "grad_norm": 4.345992133999345,
      "learning_rate": 2.3134074852951966e-06,
      "loss": 1.0229,
      "num_input_tokens_seen": 82326425,
      "step": 3879
    },
    {
      "epoch": 0.4665424156796729,
      "flos": 23036819189760.0,
      "grad_norm": 5.593920657149281,
      "learning_rate": 2.312638113581088e-06,
      "loss": 0.9783,
      "num_input_tokens_seen": 82346630,
      "step": 3880
    },
    {
      "epoch": 0.46666265857031203,
      "flos": 13067436625920.0,
      "grad_norm": 5.4097492161301615,
      "learning_rate": 2.311868694440027e-06,
      "loss": 0.9746,
      "num_input_tokens_seen": 82360770,
      "step": 3881
    },
    {
      "epoch": 0.46678290146095114,
      "flos": 49010770206720.0,
      "grad_norm": 0.7359486216617918,
      "learning_rate": 2.3110992279887323e-06,
      "loss": 0.832,
      "num_input_tokens_seen": 82432415,
      "step": 3882
    },
    {
      "epoch": 0.4669031443515902,
      "flos": 12548163686400.0,
      "grad_norm": 3.792072414165492,
      "learning_rate": 2.310329714343932e-06,
      "loss": 1.0425,
      "num_input_tokens_seen": 82448285,
      "step": 3883
    },
    {
      "epoch": 0.4670233872422293,
      "flos": 17028796784640.0,
      "grad_norm": 2.982732350714857,
      "learning_rate": 2.309560153622361e-06,
      "loss": 1.0071,
      "num_input_tokens_seen": 82464915,
      "step": 3884
    },
    {
      "epoch": 0.4671436301328684,
      "flos": 20021598228480.0,
      "grad_norm": 19.786016531765874,
      "learning_rate": 2.3087905459407602e-06,
      "loss": 0.9285,
      "num_input_tokens_seen": 82483575,
      "step": 3885
    },
    {
      "epoch": 0.46726387302350747,
      "flos": 49677222236160.0,
      "grad_norm": 0.8520847084264378,
      "learning_rate": 2.3080208914158795e-06,
      "loss": 0.875,
      "num_input_tokens_seen": 82546295,
      "step": 3886
    },
    {
      "epoch": 0.4673841159141466,
      "flos": 18158913454080.0,
      "grad_norm": 5.735271331833905,
      "learning_rate": 2.3072511901644753e-06,
      "loss": 0.9005,
      "num_input_tokens_seen": 82565085,
      "step": 3887
    },
    {
      "epoch": 0.4675043588047857,
      "flos": 17425732177920.0,
      "grad_norm": 5.082884075553238,
      "learning_rate": 2.306481442303309e-06,
      "loss": 0.9949,
      "num_input_tokens_seen": 82584380,
      "step": 3888
    },
    {
      "epoch": 0.46762460169542475,
      "flos": 14882768363520.0,
      "grad_norm": 6.980528589471412,
      "learning_rate": 2.3057116479491515e-06,
      "loss": 0.9173,
      "num_input_tokens_seen": 82603510,
      "step": 3889
    },
    {
      "epoch": 0.46774484458606386,
      "flos": 13596796231680.0,
      "grad_norm": 4.422275484638664,
      "learning_rate": 2.30494180721878e-06,
      "loss": 0.9596,
      "num_input_tokens_seen": 82620570,
      "step": 3890
    },
    {
      "epoch": 0.4678650874767029,
      "flos": 12731527987200.0,
      "grad_norm": 5.7070843891904905,
      "learning_rate": 2.3041719202289794e-06,
      "loss": 1.0931,
      "num_input_tokens_seen": 82636465,
      "step": 3891
    },
    {
      "epoch": 0.467985330367342,
      "flos": 15170097623040.0,
      "grad_norm": 3.4359898804769835,
      "learning_rate": 2.30340198709654e-06,
      "loss": 0.9992,
      "num_input_tokens_seen": 82656020,
      "step": 3892
    },
    {
      "epoch": 0.46810557325798113,
      "flos": 14567163801600.0,
      "grad_norm": 10.550820297965506,
      "learning_rate": 2.3026320079382605e-06,
      "loss": 0.9512,
      "num_input_tokens_seen": 82672675,
      "step": 3893
    },
    {
      "epoch": 0.4682258161486202,
      "flos": 21463671091200.0,
      "grad_norm": 7.8860142295374285,
      "learning_rate": 2.3018619828709454e-06,
      "loss": 0.9475,
      "num_input_tokens_seen": 82693935,
      "step": 3894
    },
    {
      "epoch": 0.4683460590392593,
      "flos": 17995270717440.0,
      "grad_norm": 5.580554935110761,
      "learning_rate": 2.3010919120114084e-06,
      "loss": 1.0115,
      "num_input_tokens_seen": 82710185,
      "step": 3895
    },
    {
      "epoch": 0.4684663019298984,
      "flos": 10862220963840.0,
      "grad_norm": 4.938231981318654,
      "learning_rate": 2.3003217954764672e-06,
      "loss": 0.8406,
      "num_input_tokens_seen": 82724610,
      "step": 3896
    },
    {
      "epoch": 0.46858654482053747,
      "flos": 19782143201280.0,
      "grad_norm": 6.006734247762439,
      "learning_rate": 2.299551633382949e-06,
      "loss": 0.9925,
      "num_input_tokens_seen": 82744640,
      "step": 3897
    },
    {
      "epoch": 0.4687067877111766,
      "flos": 12783939686400.0,
      "grad_norm": 7.90550523951765,
      "learning_rate": 2.2987814258476854e-06,
      "loss": 1.0312,
      "num_input_tokens_seen": 82762160,
      "step": 3898
    },
    {
      "epoch": 0.4688270306018157,
      "flos": 12019110666240.0,
      "grad_norm": 6.158841744558914,
      "learning_rate": 2.2980111729875177e-06,
      "loss": 0.8593,
      "num_input_tokens_seen": 82778130,
      "step": 3899
    },
    {
      "epoch": 0.46894727349245474,
      "flos": 12626551296000.0,
      "grad_norm": 9.523475152458381,
      "learning_rate": 2.2972408749192917e-06,
      "loss": 1.0119,
      "num_input_tokens_seen": 82795580,
      "step": 3900
    },
    {
      "epoch": 0.46906751638309385,
      "flos": 15249680916480.0,
      "grad_norm": 9.462668300729394,
      "learning_rate": 2.296470531759861e-06,
      "loss": 0.8617,
      "num_input_tokens_seen": 82813400,
      "step": 3901
    },
    {
      "epoch": 0.46918775927373296,
      "flos": 14434708930560.0,
      "grad_norm": 4.557781758431958,
      "learning_rate": 2.2957001436260866e-06,
      "loss": 0.9818,
      "num_input_tokens_seen": 82830180,
      "step": 3902
    },
    {
      "epoch": 0.469308002164372,
      "flos": 13044250644480.0,
      "grad_norm": 4.088673712993401,
      "learning_rate": 2.294929710634836e-06,
      "loss": 0.9157,
      "num_input_tokens_seen": 82847990,
      "step": 3903
    },
    {
      "epoch": 0.46942824505501113,
      "flos": 27101132574720.0,
      "grad_norm": 3.5686747260764897,
      "learning_rate": 2.2941592329029823e-06,
      "loss": 0.8064,
      "num_input_tokens_seen": 82868815,
      "step": 3904
    },
    {
      "epoch": 0.46954848794565024,
      "flos": 15537960591360.0,
      "grad_norm": 3.379014629764728,
      "learning_rate": 2.2933887105474067e-06,
      "loss": 0.9853,
      "num_input_tokens_seen": 82886710,
      "step": 3905
    },
    {
      "epoch": 0.4696687308362893,
      "flos": 15641220403200.0,
      "grad_norm": 8.901833814651443,
      "learning_rate": 2.2926181436849974e-06,
      "loss": 1.0011,
      "num_input_tokens_seen": 82905785,
      "step": 3906
    },
    {
      "epoch": 0.4697889737269284,
      "flos": 15351469117440.0,
      "grad_norm": 5.707266863398699,
      "learning_rate": 2.2918475324326478e-06,
      "loss": 0.9213,
      "num_input_tokens_seen": 82925225,
      "step": 3907
    },
    {
      "epoch": 0.46990921661756746,
      "flos": 17949603901440.0,
      "grad_norm": 5.6922299346054155,
      "learning_rate": 2.2910768769072603e-06,
      "loss": 1.0979,
      "num_input_tokens_seen": 82943725,
      "step": 3908
    },
    {
      "epoch": 0.47002945950820657,
      "flos": 9767829626880.0,
      "grad_norm": 4.586909176566218,
      "learning_rate": 2.2903061772257417e-06,
      "loss": 0.9446,
      "num_input_tokens_seen": 82961430,
      "step": 3909
    },
    {
      "epoch": 0.4701497023988457,
      "flos": 18681313566720.0,
      "grad_norm": 4.555344705948678,
      "learning_rate": 2.289535433505007e-06,
      "loss": 0.983,
      "num_input_tokens_seen": 82982505,
      "step": 3910
    },
    {
      "epoch": 0.47026994528948474,
      "flos": 18237914234880.0,
      "grad_norm": 4.067366985826051,
      "learning_rate": 2.2887646458619767e-06,
      "loss": 0.8305,
      "num_input_tokens_seen": 83003590,
      "step": 3911
    },
    {
      "epoch": 0.47039018818012385,
      "flos": 14590043197440.0,
      "grad_norm": 6.037497333660287,
      "learning_rate": 2.2879938144135797e-06,
      "loss": 0.963,
      "num_input_tokens_seen": 83019415,
      "step": 3912
    },
    {
      "epoch": 0.47051043107076296,
      "flos": 15325309255680.0,
      "grad_norm": 4.562367685568035,
      "learning_rate": 2.2872229392767496e-06,
      "loss": 0.9591,
      "num_input_tokens_seen": 83039240,
      "step": 3913
    },
    {
      "epoch": 0.470630673961402,
      "flos": 13439837061120.0,
      "grad_norm": 3.368068358581609,
      "learning_rate": 2.286452020568428e-06,
      "loss": 0.9451,
      "num_input_tokens_seen": 83057035,
      "step": 3914
    },
    {
      "epoch": 0.4707509168520411,
      "flos": 14148391403520.0,
      "grad_norm": 5.11091277986226,
      "learning_rate": 2.2856810584055637e-06,
      "loss": 0.9364,
      "num_input_tokens_seen": 83074290,
      "step": 3915
    },
    {
      "epoch": 0.47087115974268023,
      "flos": 28651646545920.0,
      "grad_norm": 2.4503665649848005,
      "learning_rate": 2.2849100529051085e-06,
      "loss": 0.8689,
      "num_input_tokens_seen": 83100945,
      "step": 3916
    },
    {
      "epoch": 0.4709914026333193,
      "flos": 9558550732800.0,
      "grad_norm": 8.554540345480927,
      "learning_rate": 2.284139004184026e-06,
      "loss": 0.9939,
      "num_input_tokens_seen": 83117895,
      "step": 3917
    },
    {
      "epoch": 0.4711116455239584,
      "flos": 14173907435520.0,
      "grad_norm": 5.379699426926403,
      "learning_rate": 2.2833679123592814e-06,
      "loss": 0.928,
      "num_input_tokens_seen": 83134875,
      "step": 3918
    },
    {
      "epoch": 0.4712318884145975,
      "flos": 22906081198080.0,
      "grad_norm": 9.87668991115314,
      "learning_rate": 2.2825967775478508e-06,
      "loss": 0.8234,
      "num_input_tokens_seen": 83155695,
      "step": 3919
    },
    {
      "epoch": 0.47135213130523657,
      "flos": 14226564403200.0,
      "grad_norm": 4.491023329109139,
      "learning_rate": 2.2818255998667135e-06,
      "loss": 1.0278,
      "num_input_tokens_seen": 83173925,
      "step": 3920
    },
    {
      "epoch": 0.4714723741958757,
      "flos": 13780283166720.0,
      "grad_norm": 3.162354712484936,
      "learning_rate": 2.2810543794328566e-06,
      "loss": 0.9859,
      "num_input_tokens_seen": 83192680,
      "step": 3921
    },
    {
      "epoch": 0.4715926170865148,
      "flos": 14460562206720.0,
      "grad_norm": 3.2426408933265694,
      "learning_rate": 2.2802831163632735e-06,
      "loss": 1.0081,
      "num_input_tokens_seen": 83211120,
      "step": 3922
    },
    {
      "epoch": 0.47171285997715384,
      "flos": 16112864378880.0,
      "grad_norm": 16.292338924082923,
      "learning_rate": 2.279511810774965e-06,
      "loss": 0.9325,
      "num_input_tokens_seen": 83232370,
      "step": 3923
    },
    {
      "epoch": 0.47183310286779295,
      "flos": 14986549370880.0,
      "grad_norm": 5.674261762688929,
      "learning_rate": 2.2787404627849364e-06,
      "loss": 0.9068,
      "num_input_tokens_seen": 83251300,
      "step": 3924
    },
    {
      "epoch": 0.471953345758432,
      "flos": 15433075875840.0,
      "grad_norm": 6.743651901818509,
      "learning_rate": 2.277969072510202e-06,
      "loss": 0.9847,
      "num_input_tokens_seen": 83270000,
      "step": 3925
    },
    {
      "epoch": 0.4720735886490711,
      "flos": 14091748823040.0,
      "grad_norm": 6.59998247548869,
      "learning_rate": 2.2771976400677803e-06,
      "loss": 1.0165,
      "num_input_tokens_seen": 83288550,
      "step": 3926
    },
    {
      "epoch": 0.47219383153971023,
      "flos": 13596918865920.0,
      "grad_norm": 6.29321637131146,
      "learning_rate": 2.2764261655746965e-06,
      "loss": 0.9722,
      "num_input_tokens_seen": 83305765,
      "step": 3927
    },
    {
      "epoch": 0.4723140744303493,
      "flos": 16509462528000.0,
      "grad_norm": 3.192725293029911,
      "learning_rate": 2.2756546491479832e-06,
      "loss": 0.9558,
      "num_input_tokens_seen": 83326400,
      "step": 3928
    },
    {
      "epoch": 0.4724343173209884,
      "flos": 12915229532160.0,
      "grad_norm": 8.287388552553578,
      "learning_rate": 2.274883090904679e-06,
      "loss": 0.9995,
      "num_input_tokens_seen": 83343885,
      "step": 3929
    },
    {
      "epoch": 0.4725545602116275,
      "flos": 15091158159360.0,
      "grad_norm": 5.0235348895230025,
      "learning_rate": 2.2741114909618283e-06,
      "loss": 0.8699,
      "num_input_tokens_seen": 83359500,
      "step": 3930
    },
    {
      "epoch": 0.47267480310226656,
      "flos": 15222723932160.0,
      "grad_norm": 4.422612536493472,
      "learning_rate": 2.2733398494364828e-06,
      "loss": 0.9112,
      "num_input_tokens_seen": 83378465,
      "step": 3931
    },
    {
      "epoch": 0.47279504599290567,
      "flos": 13308516556800.0,
      "grad_norm": 5.738184523627422,
      "learning_rate": 2.272568166445699e-06,
      "loss": 1.0393,
      "num_input_tokens_seen": 83396750,
      "step": 3932
    },
    {
      "epoch": 0.4729152888835448,
      "flos": 14986825297920.0,
      "grad_norm": 4.992132482091195,
      "learning_rate": 2.271796442106541e-06,
      "loss": 0.8336,
      "num_input_tokens_seen": 83415825,
      "step": 3933
    },
    {
      "epoch": 0.47303553177418384,
      "flos": 50274361589760.0,
      "grad_norm": 0.8185857855665489,
      "learning_rate": 2.271024676536079e-06,
      "loss": 0.8043,
      "num_input_tokens_seen": 83475805,
      "step": 3934
    },
    {
      "epoch": 0.47315577466482295,
      "flos": 15956518379520.0,
      "grad_norm": 4.943669581090746,
      "learning_rate": 2.2702528698513894e-06,
      "loss": 0.9308,
      "num_input_tokens_seen": 83496650,
      "step": 3935
    },
    {
      "epoch": 0.47327601755546206,
      "flos": 17320448901120.0,
      "grad_norm": 6.443120484261258,
      "learning_rate": 2.269481022169554e-06,
      "loss": 0.9786,
      "num_input_tokens_seen": 83514965,
      "step": 3936
    },
    {
      "epoch": 0.4733962604461011,
      "flos": 16295124971520.0,
      "grad_norm": 3.2872365925786933,
      "learning_rate": 2.2687091336076614e-06,
      "loss": 0.9964,
      "num_input_tokens_seen": 83534025,
      "step": 3937
    },
    {
      "epoch": 0.4735165033367402,
      "flos": 12989324943360.0,
      "grad_norm": 6.670491202367287,
      "learning_rate": 2.267937204282807e-06,
      "loss": 0.9913,
      "num_input_tokens_seen": 83550885,
      "step": 3938
    },
    {
      "epoch": 0.4736367462273793,
      "flos": 16374462996480.0,
      "grad_norm": 5.542978384857729,
      "learning_rate": 2.2671652343120926e-06,
      "loss": 0.9656,
      "num_input_tokens_seen": 83571080,
      "step": 3939
    },
    {
      "epoch": 0.4737569891180184,
      "flos": 18054549934080.0,
      "grad_norm": 6.254800707526111,
      "learning_rate": 2.2663932238126236e-06,
      "loss": 0.9957,
      "num_input_tokens_seen": 83589360,
      "step": 3940
    },
    {
      "epoch": 0.4738772320086575,
      "flos": 18396007772160.0,
      "grad_norm": 7.847906584328566,
      "learning_rate": 2.265621172901515e-06,
      "loss": 0.9919,
      "num_input_tokens_seen": 83612195,
      "step": 3941
    },
    {
      "epoch": 0.47399747489929656,
      "flos": 19628648448000.0,
      "grad_norm": 4.583111959948912,
      "learning_rate": 2.2648490816958854e-06,
      "loss": 0.9128,
      "num_input_tokens_seen": 83632910,
      "step": 3942
    },
    {
      "epoch": 0.47411771778993567,
      "flos": 17687545405440.0,
      "grad_norm": 6.546209843242697,
      "learning_rate": 2.264076950312861e-06,
      "loss": 0.9074,
      "num_input_tokens_seen": 83651440,
      "step": 3943
    },
    {
      "epoch": 0.4742379606805748,
      "flos": 16167146250240.0,
      "grad_norm": 3.967004587434771,
      "learning_rate": 2.2633047788695727e-06,
      "loss": 1.0105,
      "num_input_tokens_seen": 83671465,
      "step": 3944
    },
    {
      "epoch": 0.47435820357121383,
      "flos": 13962758369280.0,
      "grad_norm": 4.8022055565539326,
      "learning_rate": 2.262532567483159e-06,
      "loss": 0.8382,
      "num_input_tokens_seen": 83689745,
      "step": 3945
    },
    {
      "epoch": 0.47447844646185294,
      "flos": 17949941145600.0,
      "grad_norm": 7.274358672618527,
      "learning_rate": 2.2617603162707635e-06,
      "loss": 0.9989,
      "num_input_tokens_seen": 83709875,
      "step": 3946
    },
    {
      "epoch": 0.47459868935249205,
      "flos": 17476794900480.0,
      "grad_norm": 4.31355714970991,
      "learning_rate": 2.2609880253495363e-06,
      "loss": 1.0171,
      "num_input_tokens_seen": 83729230,
      "step": 3947
    },
    {
      "epoch": 0.4747189322431311,
      "flos": 14593538273280.0,
      "grad_norm": 5.5452273218530355,
      "learning_rate": 2.260215694836633e-06,
      "loss": 1.052,
      "num_input_tokens_seen": 83748125,
      "step": 3948
    },
    {
      "epoch": 0.4748391751337702,
      "flos": 18500861829120.0,
      "grad_norm": 3.9002733819462523,
      "learning_rate": 2.2594433248492157e-06,
      "loss": 0.836,
      "num_input_tokens_seen": 83766820,
      "step": 3949
    },
    {
      "epoch": 0.47495941802440933,
      "flos": 16219649925120.0,
      "grad_norm": 3.507375439303441,
      "learning_rate": 2.2586709155044527e-06,
      "loss": 0.9873,
      "num_input_tokens_seen": 83787140,
      "step": 3950
    },
    {
      "epoch": 0.4750796609150484,
      "flos": 19863780618240.0,
      "grad_norm": 5.694691900236838,
      "learning_rate": 2.2578984669195167e-06,
      "loss": 0.9553,
      "num_input_tokens_seen": 83807825,
      "step": 3951
    },
    {
      "epoch": 0.4751999038056875,
      "flos": 25446285742080.0,
      "grad_norm": 7.521943606549918,
      "learning_rate": 2.2571259792115887e-06,
      "loss": 0.8699,
      "num_input_tokens_seen": 83828765,
      "step": 3952
    },
    {
      "epoch": 0.4753201466963266,
      "flos": 15694827786240.0,
      "grad_norm": 3.195647056618367,
      "learning_rate": 2.2563534524978544e-06,
      "loss": 0.9907,
      "num_input_tokens_seen": 83845955,
      "step": 3953
    },
    {
      "epoch": 0.47544038958696566,
      "flos": 21516144107520.0,
      "grad_norm": 5.1053148288771695,
      "learning_rate": 2.2555808868955052e-06,
      "loss": 0.9183,
      "num_input_tokens_seen": 83867805,
      "step": 3954
    },
    {
      "epoch": 0.47556063247760477,
      "flos": 16686879068160.0,
      "grad_norm": 6.941498025385139,
      "learning_rate": 2.254808282521738e-06,
      "loss": 0.9181,
      "num_input_tokens_seen": 83886275,
      "step": 3955
    },
    {
      "epoch": 0.4756808753682438,
      "flos": 17897038909440.0,
      "grad_norm": 3.965612417791801,
      "learning_rate": 2.2540356394937573e-06,
      "loss": 1.0067,
      "num_input_tokens_seen": 83904695,
      "step": 3956
    },
    {
      "epoch": 0.47580111825888294,
      "flos": 11080114913280.0,
      "grad_norm": 5.852237188865989,
      "learning_rate": 2.253262957928772e-06,
      "loss": 1.008,
      "num_input_tokens_seen": 83921300,
      "step": 3957
    },
    {
      "epoch": 0.47592136114952205,
      "flos": 12493176668160.0,
      "grad_norm": 3.72334275215162,
      "learning_rate": 2.2524902379439976e-06,
      "loss": 0.9107,
      "num_input_tokens_seen": 83939690,
      "step": 3958
    },
    {
      "epoch": 0.4760416040401611,
      "flos": 43962078289920.0,
      "grad_norm": 0.7730697759519822,
      "learning_rate": 2.251717479656655e-06,
      "loss": 0.849,
      "num_input_tokens_seen": 84004205,
      "step": 3959
    },
    {
      "epoch": 0.4761618469308002,
      "flos": 13046488719360.0,
      "grad_norm": 4.520100312934355,
      "learning_rate": 2.2509446831839704e-06,
      "loss": 0.9497,
      "num_input_tokens_seen": 84023365,
      "step": 3960
    },
    {
      "epoch": 0.4762820898214393,
      "flos": 12784184954880.0,
      "grad_norm": 6.667649514546196,
      "learning_rate": 2.250171848643177e-06,
      "loss": 0.9986,
      "num_input_tokens_seen": 84040375,
      "step": 3961
    },
    {
      "epoch": 0.4764023327120784,
      "flos": 13702232801280.0,
      "grad_norm": 3.12102515153611,
      "learning_rate": 2.249398976151513e-06,
      "loss": 1.0622,
      "num_input_tokens_seen": 84057645,
      "step": 3962
    },
    {
      "epoch": 0.4765225756027175,
      "flos": 16166502420480.0,
      "grad_norm": 6.420491387406552,
      "learning_rate": 2.248626065826223e-06,
      "loss": 0.9862,
      "num_input_tokens_seen": 84075570,
      "step": 3963
    },
    {
      "epoch": 0.4766428184933566,
      "flos": 47206728929280.0,
      "grad_norm": 0.7646485544255778,
      "learning_rate": 2.2478531177845564e-06,
      "loss": 0.8347,
      "num_input_tokens_seen": 84136285,
      "step": 3964
    },
    {
      "epoch": 0.47676306138399566,
      "flos": 17165298585600.0,
      "grad_norm": 3.320599424643152,
      "learning_rate": 2.247080132143769e-06,
      "loss": 1.0463,
      "num_input_tokens_seen": 84158360,
      "step": 3965
    },
    {
      "epoch": 0.47688330427463477,
      "flos": 8876984033280.0,
      "grad_norm": 4.518061736098378,
      "learning_rate": 2.246307109021121e-06,
      "loss": 0.8675,
      "num_input_tokens_seen": 84175485,
      "step": 3966
    },
    {
      "epoch": 0.4770035471652739,
      "flos": 15190585651200.0,
      "grad_norm": 5.441213633132925,
      "learning_rate": 2.2455340485338817e-06,
      "loss": 1.0174,
      "num_input_tokens_seen": 84192840,
      "step": 3967
    },
    {
      "epoch": 0.47712379005591293,
      "flos": 17897928007680.0,
      "grad_norm": 5.750115434897672,
      "learning_rate": 2.244760950799322e-06,
      "loss": 0.8742,
      "num_input_tokens_seen": 84210830,
      "step": 3968
    },
    {
      "epoch": 0.47724403294655204,
      "flos": 15668545290240.0,
      "grad_norm": 6.300065588231255,
      "learning_rate": 2.2439878159347203e-06,
      "loss": 0.9197,
      "num_input_tokens_seen": 84229975,
      "step": 3969
    },
    {
      "epoch": 0.4773642758371911,
      "flos": 50297846046720.0,
      "grad_norm": 0.826425283540628,
      "learning_rate": 2.2432146440573616e-06,
      "loss": 0.8152,
      "num_input_tokens_seen": 84295655,
      "step": 3970
    },
    {
      "epoch": 0.4774845187278302,
      "flos": 16742233989120.0,
      "grad_norm": 6.319312621291594,
      "learning_rate": 2.242441435284534e-06,
      "loss": 0.8539,
      "num_input_tokens_seen": 84314250,
      "step": 3971
    },
    {
      "epoch": 0.4776047616184693,
      "flos": 16402278420480.0,
      "grad_norm": 4.555869104609991,
      "learning_rate": 2.2416681897335337e-06,
      "loss": 1.04,
      "num_input_tokens_seen": 84332120,
      "step": 3972
    },
    {
      "epoch": 0.4777250045091084,
      "flos": 22742101217280.0,
      "grad_norm": 5.887148693966634,
      "learning_rate": 2.240894907521661e-06,
      "loss": 0.8552,
      "num_input_tokens_seen": 84350920,
      "step": 3973
    },
    {
      "epoch": 0.4778452473997475,
      "flos": 17266534932480.0,
      "grad_norm": 7.249977637947059,
      "learning_rate": 2.240121588766223e-06,
      "loss": 0.8288,
      "num_input_tokens_seen": 84370690,
      "step": 3974
    },
    {
      "epoch": 0.4779654902903866,
      "flos": 22510617415680.0,
      "grad_norm": 7.261079937609251,
      "learning_rate": 2.239348233584531e-06,
      "loss": 0.8941,
      "num_input_tokens_seen": 84391265,
      "step": 3975
    },
    {
      "epoch": 0.47808573318102565,
      "flos": 13833124085760.0,
      "grad_norm": 3.269739798363017,
      "learning_rate": 2.2385748420939013e-06,
      "loss": 0.9959,
      "num_input_tokens_seen": 84410180,
      "step": 3976
    },
    {
      "epoch": 0.47820597607166476,
      "flos": 16061464412160.0,
      "grad_norm": 3.873583996377224,
      "learning_rate": 2.2378014144116583e-06,
      "loss": 0.921,
      "num_input_tokens_seen": 84428710,
      "step": 3977
    },
    {
      "epoch": 0.4783262189623039,
      "flos": 16350357258240.0,
      "grad_norm": 4.694210488071795,
      "learning_rate": 2.23702795065513e-06,
      "loss": 0.9922,
      "num_input_tokens_seen": 84448010,
      "step": 3978
    },
    {
      "epoch": 0.47844646185294293,
      "flos": 35592770150400.0,
      "grad_norm": 1.0020287319754282,
      "learning_rate": 2.2362544509416493e-06,
      "loss": 0.8972,
      "num_input_tokens_seen": 84499845,
      "step": 3979
    },
    {
      "epoch": 0.47856670474358204,
      "flos": 14357057126400.0,
      "grad_norm": 6.8341995558613835,
      "learning_rate": 2.2354809153885572e-06,
      "loss": 1.0144,
      "num_input_tokens_seen": 84516635,
      "step": 3980
    },
    {
      "epoch": 0.47868694763422115,
      "flos": 14903593635840.0,
      "grad_norm": 3.8081908340560964,
      "learning_rate": 2.234707344113197e-06,
      "loss": 1.026,
      "num_input_tokens_seen": 84534450,
      "step": 3981
    },
    {
      "epoch": 0.4788071905248602,
      "flos": 13492616663040.0,
      "grad_norm": 3.3568302896036735,
      "learning_rate": 2.233933737232919e-06,
      "loss": 0.9673,
      "num_input_tokens_seen": 84551950,
      "step": 3982
    },
    {
      "epoch": 0.4789274334154993,
      "flos": 16349682769920.0,
      "grad_norm": 3.6162330900786506,
      "learning_rate": 2.2331600948650793e-06,
      "loss": 0.9671,
      "num_input_tokens_seen": 84571815,
      "step": 3983
    },
    {
      "epoch": 0.4790476763061384,
      "flos": 16402462371840.0,
      "grad_norm": 8.67001493283298,
      "learning_rate": 2.2323864171270386e-06,
      "loss": 0.9928,
      "num_input_tokens_seen": 84592805,
      "step": 3984
    },
    {
      "epoch": 0.4791679191967775,
      "flos": 15039758192640.0,
      "grad_norm": 4.933096745119044,
      "learning_rate": 2.231612704136164e-06,
      "loss": 0.9175,
      "num_input_tokens_seen": 84612895,
      "step": 3985
    },
    {
      "epoch": 0.4792881620874166,
      "flos": 15845072732160.0,
      "grad_norm": 3.555267487883351,
      "learning_rate": 2.2308389560098253e-06,
      "loss": 0.9251,
      "num_input_tokens_seen": 84628990,
      "step": 3986
    },
    {
      "epoch": 0.47940840497805565,
      "flos": 12338608865280.0,
      "grad_norm": 7.801805424207737,
      "learning_rate": 2.2300651728654008e-06,
      "loss": 0.958,
      "num_input_tokens_seen": 84643970,
      "step": 3987
    },
    {
      "epoch": 0.47952864786869476,
      "flos": 46794486804480.0,
      "grad_norm": 0.7770263177104182,
      "learning_rate": 2.229291354820272e-06,
      "loss": 0.8287,
      "num_input_tokens_seen": 84704700,
      "step": 3988
    },
    {
      "epoch": 0.47964889075933387,
      "flos": 11892143677440.0,
      "grad_norm": 11.785433969838442,
      "learning_rate": 2.228517501991828e-06,
      "loss": 0.9608,
      "num_input_tokens_seen": 84723220,
      "step": 3989
    },
    {
      "epoch": 0.4797691336499729,
      "flos": 50189374279680.0,
      "grad_norm": 0.8382013985037144,
      "learning_rate": 2.22774361449746e-06,
      "loss": 0.8341,
      "num_input_tokens_seen": 84779420,
      "step": 3990
    },
    {
      "epoch": 0.47988937654061203,
      "flos": 13439959695360.0,
      "grad_norm": 13.530030933466001,
      "learning_rate": 2.2269696924545668e-06,
      "loss": 0.8906,
      "num_input_tokens_seen": 84796970,
      "step": 3991
    },
    {
      "epoch": 0.48000961943125114,
      "flos": 10211627520000.0,
      "grad_norm": 4.816995081985557,
      "learning_rate": 2.2261957359805523e-06,
      "loss": 0.9699,
      "num_input_tokens_seen": 84813925,
      "step": 3992
    },
    {
      "epoch": 0.4801298623218902,
      "flos": 19259436503040.0,
      "grad_norm": 4.625009278984041,
      "learning_rate": 2.225421745192823e-06,
      "loss": 0.9466,
      "num_input_tokens_seen": 84833800,
      "step": 3993
    },
    {
      "epoch": 0.4802501052125293,
      "flos": 18758934712320.0,
      "grad_norm": 5.275746564780752,
      "learning_rate": 2.2246477202087955e-06,
      "loss": 0.9662,
      "num_input_tokens_seen": 84854200,
      "step": 3994
    },
    {
      "epoch": 0.4803703481031684,
      "flos": 14906138296320.0,
      "grad_norm": 5.634392250952862,
      "learning_rate": 2.223873661145887e-06,
      "loss": 1.0255,
      "num_input_tokens_seen": 84873975,
      "step": 3995
    },
    {
      "epoch": 0.4804905909938075,
      "flos": 14697809817600.0,
      "grad_norm": 8.71265660587148,
      "learning_rate": 2.2230995681215226e-06,
      "loss": 0.9086,
      "num_input_tokens_seen": 84893220,
      "step": 3996
    },
    {
      "epoch": 0.4806108338844466,
      "flos": 11788086743040.0,
      "grad_norm": 10.020264882376527,
      "learning_rate": 2.2223254412531305e-06,
      "loss": 0.9744,
      "num_input_tokens_seen": 84910310,
      "step": 3997
    },
    {
      "epoch": 0.4807310767750857,
      "flos": 14200312565760.0,
      "grad_norm": 3.2671251273712167,
      "learning_rate": 2.221551280658146e-06,
      "loss": 1.0104,
      "num_input_tokens_seen": 84929090,
      "step": 3998
    },
    {
      "epoch": 0.48085131966572475,
      "flos": 16481309859840.0,
      "grad_norm": 3.156714830068054,
      "learning_rate": 2.2207770864540085e-06,
      "loss": 0.9428,
      "num_input_tokens_seen": 84947785,
      "step": 3999
    },
    {
      "epoch": 0.48097156255636386,
      "flos": 14593660907520.0,
      "grad_norm": 4.641615636648924,
      "learning_rate": 2.220002858758162e-06,
      "loss": 0.917,
      "num_input_tokens_seen": 84965495,
      "step": 4000
    },
    {
      "epoch": 0.481091805447003,
      "flos": 50499368325120.0,
      "grad_norm": 0.8811439342112282,
      "learning_rate": 2.2192285976880573e-06,
      "loss": 0.8205,
      "num_input_tokens_seen": 85029470,
      "step": 4001
    },
    {
      "epoch": 0.48121204833764203,
      "flos": 25998586060800.0,
      "grad_norm": 6.2811977706939235,
      "learning_rate": 2.2184543033611485e-06,
      "loss": 0.996,
      "num_input_tokens_seen": 85050270,
      "step": 4002
    },
    {
      "epoch": 0.48133229122828114,
      "flos": 19575194357760.0,
      "grad_norm": 5.63767594361063,
      "learning_rate": 2.2176799758948957e-06,
      "loss": 1.0121,
      "num_input_tokens_seen": 85070150,
      "step": 4003
    },
    {
      "epoch": 0.4814525341189202,
      "flos": 30774488985600.0,
      "grad_norm": 4.076040385546906,
      "learning_rate": 2.2169056154067635e-06,
      "loss": 0.9196,
      "num_input_tokens_seen": 85093790,
      "step": 4004
    },
    {
      "epoch": 0.4815727770095593,
      "flos": 17236236165120.0,
      "grad_norm": 4.516408040617618,
      "learning_rate": 2.216131222014222e-06,
      "loss": 1.0218,
      "num_input_tokens_seen": 85111585,
      "step": 4005
    },
    {
      "epoch": 0.4816930199001984,
      "flos": 12836412702720.0,
      "grad_norm": 4.962563940448234,
      "learning_rate": 2.2153567958347455e-06,
      "loss": 1.0022,
      "num_input_tokens_seen": 85127515,
      "step": 4006
    },
    {
      "epoch": 0.48181326279083747,
      "flos": 12233540198400.0,
      "grad_norm": 4.386677661136041,
      "learning_rate": 2.214582336985815e-06,
      "loss": 0.9883,
      "num_input_tokens_seen": 85145135,
      "step": 4007
    },
    {
      "epoch": 0.4819335056814766,
      "flos": 10529224888320.0,
      "grad_norm": 8.653155383066693,
      "learning_rate": 2.2138078455849142e-06,
      "loss": 0.8485,
      "num_input_tokens_seen": 85162850,
      "step": 4008
    },
    {
      "epoch": 0.4820537485721157,
      "flos": 13649453199360.0,
      "grad_norm": 5.15214421346471,
      "learning_rate": 2.2130333217495334e-06,
      "loss": 0.976,
      "num_input_tokens_seen": 85181740,
      "step": 4009
    },
    {
      "epoch": 0.48217399146275475,
      "flos": 11341652213760.0,
      "grad_norm": 5.798355777116266,
      "learning_rate": 2.2122587655971665e-06,
      "loss": 0.8602,
      "num_input_tokens_seen": 85196405,
      "step": 4010
    },
    {
      "epoch": 0.48229423435339386,
      "flos": 17163336437760.0,
      "grad_norm": 6.6224412115244675,
      "learning_rate": 2.211484177245314e-06,
      "loss": 0.8347,
      "num_input_tokens_seen": 85215715,
      "step": 4011
    },
    {
      "epoch": 0.48241447724403297,
      "flos": 16927008583680.0,
      "grad_norm": 4.053287330500239,
      "learning_rate": 2.21070955681148e-06,
      "loss": 0.9171,
      "num_input_tokens_seen": 85234540,
      "step": 4012
    },
    {
      "epoch": 0.482534720134672,
      "flos": 16427181281280.0,
      "grad_norm": 3.0098796924361366,
      "learning_rate": 2.209934904413174e-06,
      "loss": 0.9683,
      "num_input_tokens_seen": 85255865,
      "step": 4013
    },
    {
      "epoch": 0.48265496302531113,
      "flos": 14856148623360.0,
      "grad_norm": 4.105328442238495,
      "learning_rate": 2.2091602201679095e-06,
      "loss": 0.9095,
      "num_input_tokens_seen": 85275195,
      "step": 4014
    },
    {
      "epoch": 0.48277520591595025,
      "flos": 10764816936960.0,
      "grad_norm": 5.567635897625256,
      "learning_rate": 2.208385504193206e-06,
      "loss": 1.0106,
      "num_input_tokens_seen": 85292415,
      "step": 4015
    },
    {
      "epoch": 0.4828954488065893,
      "flos": 12653232353280.0,
      "grad_norm": 6.547395598987331,
      "learning_rate": 2.2076107566065873e-06,
      "loss": 1.0055,
      "num_input_tokens_seen": 85309920,
      "step": 4016
    },
    {
      "epoch": 0.4830156916972284,
      "flos": 22879062896640.0,
      "grad_norm": 4.794027968483044,
      "learning_rate": 2.2068359775255816e-06,
      "loss": 0.9437,
      "num_input_tokens_seen": 85327950,
      "step": 4017
    },
    {
      "epoch": 0.48313593458786747,
      "flos": 15537286103040.0,
      "grad_norm": 18.021202315887702,
      "learning_rate": 2.206061167067723e-06,
      "loss": 0.9633,
      "num_input_tokens_seen": 85345780,
      "step": 4018
    },
    {
      "epoch": 0.4832561774785066,
      "flos": 16061525729280.0,
      "grad_norm": 3.2683679477765692,
      "learning_rate": 2.205286325350549e-06,
      "loss": 0.9803,
      "num_input_tokens_seen": 85364565,
      "step": 4019
    },
    {
      "epoch": 0.4833764203691457,
      "flos": 9474828533760.0,
      "grad_norm": 5.098142450173469,
      "learning_rate": 2.204511452491603e-06,
      "loss": 0.9291,
      "num_input_tokens_seen": 85380910,
      "step": 4020
    },
    {
      "epoch": 0.48349666325978474,
      "flos": 31532481146880.0,
      "grad_norm": 5.790484337454397,
      "learning_rate": 2.2037365486084316e-06,
      "loss": 0.9421,
      "num_input_tokens_seen": 85403870,
      "step": 4021
    },
    {
      "epoch": 0.48361690615042385,
      "flos": 18524231761920.0,
      "grad_norm": 3.7759115823722706,
      "learning_rate": 2.2029616138185886e-06,
      "loss": 0.9748,
      "num_input_tokens_seen": 85422590,
      "step": 4022
    },
    {
      "epoch": 0.48373714904106296,
      "flos": 15825933680640.0,
      "grad_norm": 4.005643201748852,
      "learning_rate": 2.202186648239629e-06,
      "loss": 1.0204,
      "num_input_tokens_seen": 85442245,
      "step": 4023
    },
    {
      "epoch": 0.483857391931702,
      "flos": 20151508439040.0,
      "grad_norm": 4.900758427682791,
      "learning_rate": 2.201411651989117e-06,
      "loss": 0.9113,
      "num_input_tokens_seen": 85463945,
      "step": 4024
    },
    {
      "epoch": 0.48397763482234113,
      "flos": 19523395829760.0,
      "grad_norm": 4.560769785103699,
      "learning_rate": 2.2006366251846167e-06,
      "loss": 0.9724,
      "num_input_tokens_seen": 85484305,
      "step": 4025
    },
    {
      "epoch": 0.48409787771298024,
      "flos": 11889966919680.0,
      "grad_norm": 9.559106561259359,
      "learning_rate": 2.1998615679436997e-06,
      "loss": 0.9461,
      "num_input_tokens_seen": 85501565,
      "step": 4026
    },
    {
      "epoch": 0.4842181206036193,
      "flos": 17845669601280.0,
      "grad_norm": 4.7587717276323405,
      "learning_rate": 2.199086480383942e-06,
      "loss": 0.9566,
      "num_input_tokens_seen": 85520660,
      "step": 4027
    },
    {
      "epoch": 0.4843383634942584,
      "flos": 21646422220800.0,
      "grad_norm": 8.988508305340535,
      "learning_rate": 2.1983113626229234e-06,
      "loss": 0.8565,
      "num_input_tokens_seen": 85539630,
      "step": 4028
    },
    {
      "epoch": 0.4844586063848975,
      "flos": 14489511997440.0,
      "grad_norm": 4.5139266783607805,
      "learning_rate": 2.1975362147782293e-06,
      "loss": 0.9803,
      "num_input_tokens_seen": 85558545,
      "step": 4029
    },
    {
      "epoch": 0.48457884927553657,
      "flos": 50351453429760.0,
      "grad_norm": 0.8257047807263793,
      "learning_rate": 2.196761036967448e-06,
      "loss": 0.7615,
      "num_input_tokens_seen": 85626230,
      "step": 4030
    },
    {
      "epoch": 0.4846990921661757,
      "flos": 14144589742080.0,
      "grad_norm": 3.1757073763317507,
      "learning_rate": 2.1959858293081743e-06,
      "loss": 0.9749,
      "num_input_tokens_seen": 85645085,
      "step": 4031
    },
    {
      "epoch": 0.4848193350568148,
      "flos": 16402768957440.0,
      "grad_norm": 4.008758968560916,
      "learning_rate": 2.1952105919180056e-06,
      "loss": 0.962,
      "num_input_tokens_seen": 85664060,
      "step": 4032
    },
    {
      "epoch": 0.48493957794745385,
      "flos": 15956824965120.0,
      "grad_norm": 9.332277329059846,
      "learning_rate": 2.1944353249145456e-06,
      "loss": 0.8704,
      "num_input_tokens_seen": 85682890,
      "step": 4033
    },
    {
      "epoch": 0.48505982083809296,
      "flos": 17818681958400.0,
      "grad_norm": 3.4076544989574775,
      "learning_rate": 2.193660028415401e-06,
      "loss": 0.9399,
      "num_input_tokens_seen": 85703390,
      "step": 4034
    },
    {
      "epoch": 0.485180063728732,
      "flos": 19051475927040.0,
      "grad_norm": 4.507799383240232,
      "learning_rate": 2.1928847025381852e-06,
      "loss": 1.0132,
      "num_input_tokens_seen": 85723715,
      "step": 4035
    },
    {
      "epoch": 0.4853003066193711,
      "flos": 17109115883520.0,
      "grad_norm": 4.5405281576395256,
      "learning_rate": 2.192109347400512e-06,
      "loss": 1.038,
      "num_input_tokens_seen": 85743650,
      "step": 4036
    },
    {
      "epoch": 0.48542054951001024,
      "flos": 16403136860160.0,
      "grad_norm": 3.7556164961039493,
      "learning_rate": 2.191333963120004e-06,
      "loss": 0.9815,
      "num_input_tokens_seen": 85762350,
      "step": 4037
    },
    {
      "epoch": 0.4855407924006493,
      "flos": 18263184998400.0,
      "grad_norm": 50.35966587722386,
      "learning_rate": 2.190558549814286e-06,
      "loss": 0.8901,
      "num_input_tokens_seen": 85782230,
      "step": 4038
    },
    {
      "epoch": 0.4856610352912884,
      "flos": 17058175795200.0,
      "grad_norm": 4.173168881825339,
      "learning_rate": 2.1897831076009872e-06,
      "loss": 0.9864,
      "num_input_tokens_seen": 85801590,
      "step": 4039
    },
    {
      "epoch": 0.4857812781819275,
      "flos": 17136348794880.0,
      "grad_norm": 4.268627633034631,
      "learning_rate": 2.1890076365977426e-06,
      "loss": 0.9876,
      "num_input_tokens_seen": 85821135,
      "step": 4040
    },
    {
      "epoch": 0.48590152107256657,
      "flos": 40259680112640.0,
      "grad_norm": 1.1080583630452918,
      "learning_rate": 2.188232136922189e-06,
      "loss": 0.7598,
      "num_input_tokens_seen": 85878975,
      "step": 4041
    },
    {
      "epoch": 0.4860217639632057,
      "flos": 14225828597760.0,
      "grad_norm": 3.6395787385162945,
      "learning_rate": 2.187456608691971e-06,
      "loss": 0.9494,
      "num_input_tokens_seen": 85897570,
      "step": 4042
    },
    {
      "epoch": 0.4861420068538448,
      "flos": 12626980515840.0,
      "grad_norm": 3.5877508095716504,
      "learning_rate": 2.1866810520247334e-06,
      "loss": 1.0691,
      "num_input_tokens_seen": 85916160,
      "step": 4043
    },
    {
      "epoch": 0.48626224974448384,
      "flos": 18683980861440.0,
      "grad_norm": 4.543079588071806,
      "learning_rate": 2.185905467038129e-06,
      "loss": 0.841,
      "num_input_tokens_seen": 85934785,
      "step": 4044
    },
    {
      "epoch": 0.48638249263512295,
      "flos": 15668698583040.0,
      "grad_norm": 2.8644960332341296,
      "learning_rate": 2.1851298538498127e-06,
      "loss": 0.9668,
      "num_input_tokens_seen": 85954220,
      "step": 4045
    },
    {
      "epoch": 0.48650273552576206,
      "flos": 17871860121600.0,
      "grad_norm": 3.301644152648461,
      "learning_rate": 2.184354212577446e-06,
      "loss": 0.9941,
      "num_input_tokens_seen": 85974245,
      "step": 4046
    },
    {
      "epoch": 0.4866229784164011,
      "flos": 12364462141440.0,
      "grad_norm": 6.303490885610187,
      "learning_rate": 2.1835785433386907e-06,
      "loss": 0.8111,
      "num_input_tokens_seen": 85992780,
      "step": 4047
    },
    {
      "epoch": 0.48674322130704023,
      "flos": 16586133258240.0,
      "grad_norm": 5.739722313095409,
      "learning_rate": 2.182802846251216e-06,
      "loss": 0.8442,
      "num_input_tokens_seen": 86012770,
      "step": 4048
    },
    {
      "epoch": 0.4868634641976793,
      "flos": 20519677992960.0,
      "grad_norm": 7.461965887486076,
      "learning_rate": 2.182027121432696e-06,
      "loss": 0.9124,
      "num_input_tokens_seen": 86033115,
      "step": 4049
    },
    {
      "epoch": 0.4869837070883184,
      "flos": 13492126126080.0,
      "grad_norm": 3.635795566229496,
      "learning_rate": 2.1812513690008054e-06,
      "loss": 1.0155,
      "num_input_tokens_seen": 86051955,
      "step": 4050
    },
    {
      "epoch": 0.4871039499789575,
      "flos": 10685908131840.0,
      "grad_norm": 10.30382825205997,
      "learning_rate": 2.180475589073227e-06,
      "loss": 0.9865,
      "num_input_tokens_seen": 86069375,
      "step": 4051
    },
    {
      "epoch": 0.48722419286959656,
      "flos": 18628901867520.0,
      "grad_norm": 4.925066201011981,
      "learning_rate": 2.1796997817676456e-06,
      "loss": 0.9294,
      "num_input_tokens_seen": 86090105,
      "step": 4052
    },
    {
      "epoch": 0.4873444357602357,
      "flos": 17085868584960.0,
      "grad_norm": 4.724901042890983,
      "learning_rate": 2.1789239472017494e-06,
      "loss": 0.8707,
      "num_input_tokens_seen": 86111475,
      "step": 4053
    },
    {
      "epoch": 0.4874646786508748,
      "flos": 16219128729600.0,
      "grad_norm": 4.581368504926196,
      "learning_rate": 2.1781480854932326e-06,
      "loss": 0.9223,
      "num_input_tokens_seen": 86130960,
      "step": 4054
    },
    {
      "epoch": 0.48758492154151384,
      "flos": 15117440655360.0,
      "grad_norm": 4.839822271882831,
      "learning_rate": 2.1773721967597933e-06,
      "loss": 0.993,
      "num_input_tokens_seen": 86149130,
      "step": 4055
    },
    {
      "epoch": 0.48770516443215295,
      "flos": 44554741493760.0,
      "grad_norm": 0.9878062437318563,
      "learning_rate": 2.1765962811191322e-06,
      "loss": 0.8238,
      "num_input_tokens_seen": 86203315,
      "step": 4056
    },
    {
      "epoch": 0.48782540732279206,
      "flos": 47352888176640.0,
      "grad_norm": 0.9357464284846528,
      "learning_rate": 2.1758203386889566e-06,
      "loss": 0.8755,
      "num_input_tokens_seen": 86265805,
      "step": 4057
    },
    {
      "epoch": 0.4879456502134311,
      "flos": 10316512235520.0,
      "grad_norm": 5.577149548978833,
      "learning_rate": 2.1750443695869746e-06,
      "loss": 1.0259,
      "num_input_tokens_seen": 86281095,
      "step": 4058
    },
    {
      "epoch": 0.4880658931040702,
      "flos": 13833154744320.0,
      "grad_norm": 4.214122579709955,
      "learning_rate": 2.174268373930901e-06,
      "loss": 1.043,
      "num_input_tokens_seen": 86298330,
      "step": 4059
    },
    {
      "epoch": 0.48818613599470934,
      "flos": 11836482170880.0,
      "grad_norm": 5.244114983352116,
      "learning_rate": 2.1734923518384537e-06,
      "loss": 1.0006,
      "num_input_tokens_seen": 86314655,
      "step": 4060
    },
    {
      "epoch": 0.4883063788853484,
      "flos": 19047674265600.0,
      "grad_norm": 4.409448206974455,
      "learning_rate": 2.1727163034273547e-06,
      "loss": 1.02,
      "num_input_tokens_seen": 86332540,
      "step": 4061
    },
    {
      "epoch": 0.4884266217759875,
      "flos": 11865983815680.0,
      "grad_norm": 5.6600298818062305,
      "learning_rate": 2.17194022881533e-06,
      "loss": 0.9693,
      "num_input_tokens_seen": 86348350,
      "step": 4062
    },
    {
      "epoch": 0.4885468646666266,
      "flos": 17215748136960.0,
      "grad_norm": 5.949008750089596,
      "learning_rate": 2.1711641281201092e-06,
      "loss": 0.8652,
      "num_input_tokens_seen": 86368000,
      "step": 4063
    },
    {
      "epoch": 0.48866710755726567,
      "flos": 10318964920320.0,
      "grad_norm": 6.61114694637271,
      "learning_rate": 2.1703880014594264e-06,
      "loss": 0.9755,
      "num_input_tokens_seen": 86385310,
      "step": 4064
    },
    {
      "epoch": 0.4887873504479048,
      "flos": 20519769968640.0,
      "grad_norm": 5.950095219210847,
      "learning_rate": 2.1696118489510182e-06,
      "loss": 0.9193,
      "num_input_tokens_seen": 86405960,
      "step": 4065
    },
    {
      "epoch": 0.48890759333854383,
      "flos": 16193275453440.0,
      "grad_norm": 3.288214422422843,
      "learning_rate": 2.1688356707126286e-06,
      "loss": 0.9095,
      "num_input_tokens_seen": 86425300,
      "step": 4066
    },
    {
      "epoch": 0.48902783622918294,
      "flos": 12601127239680.0,
      "grad_norm": 3.7384098234159,
      "learning_rate": 2.168059466862001e-06,
      "loss": 0.8951,
      "num_input_tokens_seen": 86443170,
      "step": 4067
    },
    {
      "epoch": 0.48914807911982205,
      "flos": 15852584079360.0,
      "grad_norm": 5.406117873148348,
      "learning_rate": 2.167283237516887e-06,
      "loss": 1.008,
      "num_input_tokens_seen": 86461165,
      "step": 4068
    },
    {
      "epoch": 0.4892683220104611,
      "flos": 11578378629120.0,
      "grad_norm": 5.630878852019298,
      "learning_rate": 2.1665069827950383e-06,
      "loss": 0.9492,
      "num_input_tokens_seen": 86478170,
      "step": 4069
    },
    {
      "epoch": 0.4893885649011002,
      "flos": 11129522073600.0,
      "grad_norm": 4.591035001660803,
      "learning_rate": 2.1657307028142126e-06,
      "loss": 1.0602,
      "num_input_tokens_seen": 86495430,
      "step": 4070
    },
    {
      "epoch": 0.48950880779173933,
      "flos": 20359407697920.0,
      "grad_norm": 4.155870463315877,
      "learning_rate": 2.164954397692171e-06,
      "loss": 0.8657,
      "num_input_tokens_seen": 86514575,
      "step": 4071
    },
    {
      "epoch": 0.4896290506823784,
      "flos": 47388161740800.0,
      "grad_norm": 1.1370987928678475,
      "learning_rate": 2.164178067546678e-06,
      "loss": 1.0219,
      "num_input_tokens_seen": 86573460,
      "step": 4072
    },
    {
      "epoch": 0.4897492935730175,
      "flos": 8824296407040.0,
      "grad_norm": 4.244100627543927,
      "learning_rate": 2.163401712495504e-06,
      "loss": 1.1018,
      "num_input_tokens_seen": 86590875,
      "step": 4073
    },
    {
      "epoch": 0.4898695364636566,
      "flos": 16690803363840.0,
      "grad_norm": 3.9358755726956964,
      "learning_rate": 2.1626253326564194e-06,
      "loss": 0.9876,
      "num_input_tokens_seen": 86609545,
      "step": 4074
    },
    {
      "epoch": 0.48998977935429566,
      "flos": 19337762795520.0,
      "grad_norm": 2.5261674882418035,
      "learning_rate": 2.161848928147201e-06,
      "loss": 0.9597,
      "num_input_tokens_seen": 86629535,
      "step": 4075
    },
    {
      "epoch": 0.4901100222449348,
      "flos": 14436211200000.0,
      "grad_norm": 5.271463931914325,
      "learning_rate": 2.161072499085629e-06,
      "loss": 0.984,
      "num_input_tokens_seen": 86648250,
      "step": 4076
    },
    {
      "epoch": 0.4902302651355739,
      "flos": 21699784335360.0,
      "grad_norm": 3.2980755251812846,
      "learning_rate": 2.160296045589487e-06,
      "loss": 1.0265,
      "num_input_tokens_seen": 86671430,
      "step": 4077
    },
    {
      "epoch": 0.49035050802621294,
      "flos": 13598942330880.0,
      "grad_norm": 4.448280242187661,
      "learning_rate": 2.159519567776562e-06,
      "loss": 0.884,
      "num_input_tokens_seen": 86690800,
      "step": 4078
    },
    {
      "epoch": 0.49047075091685205,
      "flos": 15792599715840.0,
      "grad_norm": 8.822712493179447,
      "learning_rate": 2.1587430657646463e-06,
      "loss": 0.8926,
      "num_input_tokens_seen": 86703955,
      "step": 4079
    },
    {
      "epoch": 0.4905909938074911,
      "flos": 14304553451520.0,
      "grad_norm": 3.4361702861400163,
      "learning_rate": 2.157966539671533e-06,
      "loss": 0.9711,
      "num_input_tokens_seen": 86723315,
      "step": 4080
    },
    {
      "epoch": 0.4907112366981302,
      "flos": 12181496401920.0,
      "grad_norm": 5.004024411905022,
      "learning_rate": 2.157189989615021e-06,
      "loss": 0.8496,
      "num_input_tokens_seen": 86741625,
      "step": 4081
    },
    {
      "epoch": 0.4908314795887693,
      "flos": 15405904281600.0,
      "grad_norm": 3.0877988971788537,
      "learning_rate": 2.156413415712913e-06,
      "loss": 0.9395,
      "num_input_tokens_seen": 86763395,
      "step": 4082
    },
    {
      "epoch": 0.4909517224794084,
      "flos": 18657790341120.0,
      "grad_norm": 2.954447455214765,
      "learning_rate": 2.155636818083014e-06,
      "loss": 0.9744,
      "num_input_tokens_seen": 86784485,
      "step": 4083
    },
    {
      "epoch": 0.4910719653700475,
      "flos": 16454751436800.0,
      "grad_norm": 4.406243506893487,
      "learning_rate": 2.154860196843134e-06,
      "loss": 1.0282,
      "num_input_tokens_seen": 86803400,
      "step": 4084
    },
    {
      "epoch": 0.4911922082606866,
      "flos": 16586746429440.0,
      "grad_norm": 9.320807316731987,
      "learning_rate": 2.154083552111085e-06,
      "loss": 0.9588,
      "num_input_tokens_seen": 86822290,
      "step": 4085
    },
    {
      "epoch": 0.49131245115132566,
      "flos": 20806026178560.0,
      "grad_norm": 4.4077001038737285,
      "learning_rate": 2.1533068840046834e-06,
      "loss": 1.0111,
      "num_input_tokens_seen": 86842275,
      "step": 4086
    },
    {
      "epoch": 0.49143269404196477,
      "flos": 14296827494400.0,
      "grad_norm": 4.499952104101663,
      "learning_rate": 2.152530192641749e-06,
      "loss": 0.8048,
      "num_input_tokens_seen": 86856905,
      "step": 4087
    },
    {
      "epoch": 0.4915529369326039,
      "flos": 17347068641280.0,
      "grad_norm": 3.418451556302304,
      "learning_rate": 2.1517534781401068e-06,
      "loss": 0.9169,
      "num_input_tokens_seen": 86874505,
      "step": 4088
    },
    {
      "epoch": 0.49167317982324293,
      "flos": 7382070251520.0,
      "grad_norm": 5.767719332952232,
      "learning_rate": 2.150976740617581e-06,
      "loss": 0.8897,
      "num_input_tokens_seen": 86890785,
      "step": 4089
    },
    {
      "epoch": 0.49179342271388204,
      "flos": 18212030300160.0,
      "grad_norm": 4.498345176753739,
      "learning_rate": 2.150199980192006e-06,
      "loss": 0.9219,
      "num_input_tokens_seen": 86909625,
      "step": 4090
    },
    {
      "epoch": 0.49191366560452116,
      "flos": 14983667466240.0,
      "grad_norm": 2.81300095229365,
      "learning_rate": 2.1494231969812114e-06,
      "loss": 1.0066,
      "num_input_tokens_seen": 86928335,
      "step": 4091
    },
    {
      "epoch": 0.4920339084951602,
      "flos": 18552721674240.0,
      "grad_norm": 4.5154446364552605,
      "learning_rate": 2.1486463911030372e-06,
      "loss": 0.9875,
      "num_input_tokens_seen": 86948705,
      "step": 4092
    },
    {
      "epoch": 0.4921541513857993,
      "flos": 17843768770560.0,
      "grad_norm": 5.5894892144949235,
      "learning_rate": 2.147869562675324e-06,
      "loss": 0.9436,
      "num_input_tokens_seen": 86967395,
      "step": 4093
    },
    {
      "epoch": 0.49227439427643843,
      "flos": 17347068641280.0,
      "grad_norm": 7.978449644005019,
      "learning_rate": 2.147092711815915e-06,
      "loss": 0.9155,
      "num_input_tokens_seen": 86986465,
      "step": 4094
    },
    {
      "epoch": 0.4923946371670775,
      "flos": 7985402634240.0,
      "grad_norm": 5.94104444216004,
      "learning_rate": 2.1463158386426593e-06,
      "loss": 1.0563,
      "num_input_tokens_seen": 87003995,
      "step": 4095
    },
    {
      "epoch": 0.4925148800577166,
      "flos": 21699447091200.0,
      "grad_norm": 6.26974268286396,
      "learning_rate": 2.145538943273407e-06,
      "loss": 0.965,
      "num_input_tokens_seen": 87023990,
      "step": 4096
    },
    {
      "epoch": 0.49263512294835565,
      "flos": 14803369021440.0,
      "grad_norm": 2.7357985796208784,
      "learning_rate": 2.144762025826013e-06,
      "loss": 0.9095,
      "num_input_tokens_seen": 87042800,
      "step": 4097
    },
    {
      "epoch": 0.49275536583899476,
      "flos": 16899530403840.0,
      "grad_norm": 4.357200162095319,
      "learning_rate": 2.143985086418334e-06,
      "loss": 1.0603,
      "num_input_tokens_seen": 87057700,
      "step": 4098
    },
    {
      "epoch": 0.4928756087296339,
      "flos": 15826117632000.0,
      "grad_norm": 2.77429847785345,
      "learning_rate": 2.1432081251682324e-06,
      "loss": 0.9664,
      "num_input_tokens_seen": 87077790,
      "step": 4099
    },
    {
      "epoch": 0.49299585162027293,
      "flos": 13937242337280.0,
      "grad_norm": 3.2283134235588062,
      "learning_rate": 2.142431142193572e-06,
      "loss": 1.0591,
      "num_input_tokens_seen": 87095290,
      "step": 4100
    },
    {
      "epoch": 0.49311609451091204,
      "flos": 27729950330880.0,
      "grad_norm": 5.635335455101881,
      "learning_rate": 2.1416541376122207e-06,
      "loss": 0.9231,
      "num_input_tokens_seen": 87115190,
      "step": 4101
    },
    {
      "epoch": 0.49323633740155115,
      "flos": 20178005544960.0,
      "grad_norm": 14.513180234279988,
      "learning_rate": 2.1408771115420496e-06,
      "loss": 0.9207,
      "num_input_tokens_seen": 87134770,
      "step": 4102
    },
    {
      "epoch": 0.4933565802921902,
      "flos": 15007374643200.0,
      "grad_norm": 4.026086857169096,
      "learning_rate": 2.140100064100932e-06,
      "loss": 0.8503,
      "num_input_tokens_seen": 87150465,
      "step": 4103
    },
    {
      "epoch": 0.4934768231828293,
      "flos": 12782560051200.0,
      "grad_norm": 4.802451604373907,
      "learning_rate": 2.139322995406746e-06,
      "loss": 0.9545,
      "num_input_tokens_seen": 87167820,
      "step": 4104
    },
    {
      "epoch": 0.4935970660734684,
      "flos": 16684947578880.0,
      "grad_norm": 4.628774185550148,
      "learning_rate": 2.1385459055773727e-06,
      "loss": 1.0004,
      "num_input_tokens_seen": 87185730,
      "step": 4105
    },
    {
      "epoch": 0.4937173089641075,
      "flos": 46157460664320.0,
      "grad_norm": 3.982756162404362,
      "learning_rate": 2.137768794730696e-06,
      "loss": 0.9303,
      "num_input_tokens_seen": 87208900,
      "step": 4106
    },
    {
      "epoch": 0.4938375518547466,
      "flos": 15878161428480.0,
      "grad_norm": 3.8578263237335864,
      "learning_rate": 2.1369916629846026e-06,
      "loss": 0.995,
      "num_input_tokens_seen": 87228370,
      "step": 4107
    },
    {
      "epoch": 0.4939577947453857,
      "flos": 12653263011840.0,
      "grad_norm": 7.225018046531447,
      "learning_rate": 2.136214510456983e-06,
      "loss": 0.94,
      "num_input_tokens_seen": 87246545,
      "step": 4108
    },
    {
      "epoch": 0.49407803763602476,
      "flos": 50178705100800.0,
      "grad_norm": 0.9471428545804097,
      "learning_rate": 2.1354373372657296e-06,
      "loss": 0.8989,
      "num_input_tokens_seen": 87304705,
      "step": 4109
    },
    {
      "epoch": 0.49419828052666387,
      "flos": 17295178137600.0,
      "grad_norm": 3.339163227909362,
      "learning_rate": 2.1346601435287404e-06,
      "loss": 0.9046,
      "num_input_tokens_seen": 87326695,
      "step": 4110
    },
    {
      "epoch": 0.494318523417303,
      "flos": 20937469317120.0,
      "grad_norm": 3.1937877496029947,
      "learning_rate": 2.1338829293639144e-06,
      "loss": 0.9988,
      "num_input_tokens_seen": 87346775,
      "step": 4111
    },
    {
      "epoch": 0.49443876630794203,
      "flos": 10791589969920.0,
      "grad_norm": 3.877294783827797,
      "learning_rate": 2.1331056948891547e-06,
      "loss": 1.0309,
      "num_input_tokens_seen": 87363595,
      "step": 4112
    },
    {
      "epoch": 0.49455900919858115,
      "flos": 8692178780160.0,
      "grad_norm": 4.908982020432415,
      "learning_rate": 2.1323284402223666e-06,
      "loss": 0.9564,
      "num_input_tokens_seen": 87379305,
      "step": 4113
    },
    {
      "epoch": 0.4946792520892202,
      "flos": 16189197864960.0,
      "grad_norm": 6.212199839780207,
      "learning_rate": 2.1315511654814597e-06,
      "loss": 1.0728,
      "num_input_tokens_seen": 87397435,
      "step": 4114
    },
    {
      "epoch": 0.4947994949798593,
      "flos": 16454536826880.0,
      "grad_norm": 3.23255050725931,
      "learning_rate": 2.1307738707843456e-06,
      "loss": 0.9657,
      "num_input_tokens_seen": 87416820,
      "step": 4115
    },
    {
      "epoch": 0.4949197378704984,
      "flos": 16822553088000.0,
      "grad_norm": 3.887197677779304,
      "learning_rate": 2.1299965562489385e-06,
      "loss": 0.8822,
      "num_input_tokens_seen": 87436345,
      "step": 4116
    },
    {
      "epoch": 0.4950399807611375,
      "flos": 19159917035520.0,
      "grad_norm": 2.5583002053103,
      "learning_rate": 2.129219221993158e-06,
      "loss": 0.9865,
      "num_input_tokens_seen": 87460850,
      "step": 4117
    },
    {
      "epoch": 0.4951602236517766,
      "flos": 48199240089600.0,
      "grad_norm": 0.8258525330409229,
      "learning_rate": 2.128441868134924e-06,
      "loss": 0.8356,
      "num_input_tokens_seen": 87522505,
      "step": 4118
    },
    {
      "epoch": 0.4952804665424157,
      "flos": 14121281126400.0,
      "grad_norm": 5.223057048885555,
      "learning_rate": 2.1276644947921606e-06,
      "loss": 1.0251,
      "num_input_tokens_seen": 87541140,
      "step": 4119
    },
    {
      "epoch": 0.49540070943305475,
      "flos": 13334584442880.0,
      "grad_norm": 4.15538578083297,
      "learning_rate": 2.126887102082795e-06,
      "loss": 1.0198,
      "num_input_tokens_seen": 87560885,
      "step": 4120
    },
    {
      "epoch": 0.49552095232369386,
      "flos": 17737902981120.0,
      "grad_norm": 2.895456431352718,
      "learning_rate": 2.126109690124757e-06,
      "loss": 0.8905,
      "num_input_tokens_seen": 87581420,
      "step": 4121
    },
    {
      "epoch": 0.495641195214333,
      "flos": 16245625835520.0,
      "grad_norm": 3.982026009858358,
      "learning_rate": 2.1253322590359786e-06,
      "loss": 0.8991,
      "num_input_tokens_seen": 87600475,
      "step": 4122
    },
    {
      "epoch": 0.49576143810497203,
      "flos": 18337924239360.0,
      "grad_norm": 5.421434016254605,
      "learning_rate": 2.124554808934397e-06,
      "loss": 0.9192,
      "num_input_tokens_seen": 87620775,
      "step": 4123
    },
    {
      "epoch": 0.49588168099561114,
      "flos": 15722305966080.0,
      "grad_norm": 10.921695473086398,
      "learning_rate": 2.1237773399379496e-06,
      "loss": 0.9271,
      "num_input_tokens_seen": 87641460,
      "step": 4124
    },
    {
      "epoch": 0.49600192388625025,
      "flos": 17344953200640.0,
      "grad_norm": 4.31234851515813,
      "learning_rate": 2.122999852164578e-06,
      "loss": 1.0635,
      "num_input_tokens_seen": 87661800,
      "step": 4125
    },
    {
      "epoch": 0.4961221667768893,
      "flos": 16245779128320.0,
      "grad_norm": 5.819308651371883,
      "learning_rate": 2.122222345732227e-06,
      "loss": 0.7724,
      "num_input_tokens_seen": 87681435,
      "step": 4126
    },
    {
      "epoch": 0.4962424096675284,
      "flos": 12652741816320.0,
      "grad_norm": 3.975132653284171,
      "learning_rate": 2.121444820758843e-06,
      "loss": 1.0281,
      "num_input_tokens_seen": 87699795,
      "step": 4127
    },
    {
      "epoch": 0.49636265255816747,
      "flos": 15480459571200.0,
      "grad_norm": 3.6054313219634375,
      "learning_rate": 2.120667277362376e-06,
      "loss": 0.9661,
      "num_input_tokens_seen": 87718230,
      "step": 4128
    },
    {
      "epoch": 0.4964828954488066,
      "flos": 11574607626240.0,
      "grad_norm": 3.867158400300869,
      "learning_rate": 2.1198897156607796e-06,
      "loss": 1.0416,
      "num_input_tokens_seen": 87735305,
      "step": 4129
    },
    {
      "epoch": 0.4966031383394457,
      "flos": 17577019514880.0,
      "grad_norm": 4.942453724206021,
      "learning_rate": 2.1191121357720085e-06,
      "loss": 0.937,
      "num_input_tokens_seen": 87753085,
      "step": 4130
    },
    {
      "epoch": 0.49672338123008475,
      "flos": 16298405437440.0,
      "grad_norm": 12.883311727554213,
      "learning_rate": 2.1183345378140206e-06,
      "loss": 0.9412,
      "num_input_tokens_seen": 87772550,
      "step": 4131
    },
    {
      "epoch": 0.49684362412072386,
      "flos": 47238805893120.0,
      "grad_norm": 1.012943542722297,
      "learning_rate": 2.1175569219047783e-06,
      "loss": 0.8549,
      "num_input_tokens_seen": 87833710,
      "step": 4132
    },
    {
      "epoch": 0.49696386701136297,
      "flos": 14171056189440.0,
      "grad_norm": 4.962488031252786,
      "learning_rate": 2.1167792881622437e-06,
      "loss": 0.9201,
      "num_input_tokens_seen": 87852450,
      "step": 4133
    },
    {
      "epoch": 0.497084109902002,
      "flos": 17605999964160.0,
      "grad_norm": 3.6069079137796582,
      "learning_rate": 2.116001636704384e-06,
      "loss": 1.0072,
      "num_input_tokens_seen": 87872555,
      "step": 4134
    },
    {
      "epoch": 0.49720435279264114,
      "flos": 15537378078720.0,
      "grad_norm": 6.7815017301887295,
      "learning_rate": 2.1152239676491685e-06,
      "loss": 0.9937,
      "num_input_tokens_seen": 87890380,
      "step": 4135
    },
    {
      "epoch": 0.49732459568328025,
      "flos": 16927315169280.0,
      "grad_norm": 3.2777086311705026,
      "learning_rate": 2.114446281114569e-06,
      "loss": 0.93,
      "num_input_tokens_seen": 87909120,
      "step": 4136
    },
    {
      "epoch": 0.4974448385739193,
      "flos": 14226472427520.0,
      "grad_norm": 5.351168652820403,
      "learning_rate": 2.1136685772185587e-06,
      "loss": 0.9468,
      "num_input_tokens_seen": 87927865,
      "step": 4137
    },
    {
      "epoch": 0.4975650814645584,
      "flos": 17656357539840.0,
      "grad_norm": 2.9729460927090168,
      "learning_rate": 2.1128908560791163e-06,
      "loss": 0.9768,
      "num_input_tokens_seen": 87947415,
      "step": 4138
    },
    {
      "epoch": 0.4976853243551975,
      "flos": 14069697208320.0,
      "grad_norm": 4.035623942447859,
      "learning_rate": 2.1121131178142203e-06,
      "loss": 0.9675,
      "num_input_tokens_seen": 87966500,
      "step": 4139
    },
    {
      "epoch": 0.4978055672458366,
      "flos": 16450152652800.0,
      "grad_norm": 3.876845627839046,
      "learning_rate": 2.1113353625418544e-06,
      "loss": 1.0155,
      "num_input_tokens_seen": 87984770,
      "step": 4140
    },
    {
      "epoch": 0.4979258101364757,
      "flos": 11001144791040.0,
      "grad_norm": 2.773950491729513,
      "learning_rate": 2.1105575903800017e-06,
      "loss": 0.9851,
      "num_input_tokens_seen": 88003210,
      "step": 4141
    },
    {
      "epoch": 0.4980460530271148,
      "flos": 18759731834880.0,
      "grad_norm": 3.7274964639709607,
      "learning_rate": 2.1097798014466502e-06,
      "loss": 1.0369,
      "num_input_tokens_seen": 88022530,
      "step": 4142
    },
    {
      "epoch": 0.49816629591775385,
      "flos": 12233386905600.0,
      "grad_norm": 10.513671693583966,
      "learning_rate": 2.109001995859791e-06,
      "loss": 0.7546,
      "num_input_tokens_seen": 88041150,
      "step": 4143
    },
    {
      "epoch": 0.49828653880839296,
      "flos": 46488876933120.0,
      "grad_norm": 0.7762793149342376,
      "learning_rate": 2.108224173737415e-06,
      "loss": 0.8259,
      "num_input_tokens_seen": 88104170,
      "step": 4144
    },
    {
      "epoch": 0.498406781699032,
      "flos": 19570534256640.0,
      "grad_norm": 3.3945639798569545,
      "learning_rate": 2.1074463351975183e-06,
      "loss": 0.9492,
      "num_input_tokens_seen": 88122775,
      "step": 4145
    },
    {
      "epoch": 0.49852702458967113,
      "flos": 22456304885760.0,
      "grad_norm": 4.013501032328137,
      "learning_rate": 2.106668480358098e-06,
      "loss": 0.9042,
      "num_input_tokens_seen": 88142720,
      "step": 4146
    },
    {
      "epoch": 0.49864726748031024,
      "flos": 16241885491200.0,
      "grad_norm": 3.2317140109558578,
      "learning_rate": 2.105890609337154e-06,
      "loss": 0.9051,
      "num_input_tokens_seen": 88160955,
      "step": 4147
    },
    {
      "epoch": 0.4987675103709493,
      "flos": 50423739985920.0,
      "grad_norm": 0.7016402356552642,
      "learning_rate": 2.1051127222526883e-06,
      "loss": 0.8386,
      "num_input_tokens_seen": 88232790,
      "step": 4148
    },
    {
      "epoch": 0.4988877532615884,
      "flos": 20493211545600.0,
      "grad_norm": 4.241065113952066,
      "learning_rate": 2.1043348192227067e-06,
      "loss": 0.9999,
      "num_input_tokens_seen": 88252880,
      "step": 4149
    },
    {
      "epoch": 0.4990079961522275,
      "flos": 11944524718080.0,
      "grad_norm": 4.305675912357532,
      "learning_rate": 2.1035569003652156e-06,
      "loss": 0.819,
      "num_input_tokens_seen": 88271230,
      "step": 4150
    },
    {
      "epoch": 0.4991282390428666,
      "flos": 9369637232640.0,
      "grad_norm": 9.627173240344716,
      "learning_rate": 2.1027789657982255e-06,
      "loss": 1.0045,
      "num_input_tokens_seen": 88285165,
      "step": 4151
    },
    {
      "epoch": 0.4992484819335057,
      "flos": 15296298147840.0,
      "grad_norm": 5.972407473665385,
      "learning_rate": 2.1020010156397482e-06,
      "loss": 0.9732,
      "num_input_tokens_seen": 88302105,
      "step": 4152
    },
    {
      "epoch": 0.4993687248241448,
      "flos": 17684510208000.0,
      "grad_norm": 11.112996687371735,
      "learning_rate": 2.101223050007797e-06,
      "loss": 0.9765,
      "num_input_tokens_seen": 88320375,
      "step": 4153
    },
    {
      "epoch": 0.49948896771478385,
      "flos": 45056592261120.0,
      "grad_norm": 0.8773043466553149,
      "learning_rate": 2.1004450690203904e-06,
      "loss": 0.775,
      "num_input_tokens_seen": 88376175,
      "step": 4154
    },
    {
      "epoch": 0.49960921060542296,
      "flos": 48898343485440.0,
      "grad_norm": 0.9828288636414565,
      "learning_rate": 2.099667072795546e-06,
      "loss": 0.9052,
      "num_input_tokens_seen": 88438015,
      "step": 4155
    },
    {
      "epoch": 0.49972945349606207,
      "flos": 16822001233920.0,
      "grad_norm": 4.205206486806526,
      "learning_rate": 2.0988890614512864e-06,
      "loss": 0.9833,
      "num_input_tokens_seen": 88457625,
      "step": 4156
    },
    {
      "epoch": 0.4998496963867011,
      "flos": 14016672337920.0,
      "grad_norm": 10.593227857477139,
      "learning_rate": 2.098111035105635e-06,
      "loss": 1.0378,
      "num_input_tokens_seen": 88475770,
      "step": 4157
    },
    {
      "epoch": 0.49996993927734024,
      "flos": 15819464724480.0,
      "grad_norm": 3.5502308837599466,
      "learning_rate": 2.0973329938766176e-06,
      "loss": 0.9372,
      "num_input_tokens_seen": 88492920,
      "step": 4158
    },
    {
      "epoch": 0.5000901821679793,
      "flos": 16582791475200.0,
      "grad_norm": 4.141894566938866,
      "learning_rate": 2.0965549378822618e-06,
      "loss": 0.9836,
      "num_input_tokens_seen": 88513930,
      "step": 4159
    },
    {
      "epoch": 0.5002104250586185,
      "flos": 14435720663040.0,
      "grad_norm": 4.014778493279256,
      "learning_rate": 2.095776867240599e-06,
      "loss": 1.0358,
      "num_input_tokens_seen": 88530640,
      "step": 4160
    },
    {
      "epoch": 0.5003306679492575,
      "flos": 9873910026240.0,
      "grad_norm": 6.4901168220245,
      "learning_rate": 2.094998782069661e-06,
      "loss": 1.0066,
      "num_input_tokens_seen": 88548065,
      "step": 4161
    },
    {
      "epoch": 0.5004509108398966,
      "flos": 19706361569280.0,
      "grad_norm": 2.8332129808183737,
      "learning_rate": 2.0942206824874845e-06,
      "loss": 0.9497,
      "num_input_tokens_seen": 88570560,
      "step": 4162
    },
    {
      "epoch": 0.5005711537305357,
      "flos": 10581636587520.0,
      "grad_norm": 6.387198949837455,
      "learning_rate": 2.093442568612105e-06,
      "loss": 0.9791,
      "num_input_tokens_seen": 88588085,
      "step": 4163
    },
    {
      "epoch": 0.5006913966211748,
      "flos": 18865597624320.0,
      "grad_norm": 4.4803897814123035,
      "learning_rate": 2.0926644405615613e-06,
      "loss": 1.0354,
      "num_input_tokens_seen": 88608705,
      "step": 4164
    },
    {
      "epoch": 0.5008116395118138,
      "flos": 14515089346560.0,
      "grad_norm": 3.5428746725123363,
      "learning_rate": 2.091886298453897e-06,
      "loss": 1.0134,
      "num_input_tokens_seen": 88626610,
      "step": 4165
    },
    {
      "epoch": 0.500931882402453,
      "flos": 15327547330560.0,
      "grad_norm": 5.3664694226587235,
      "learning_rate": 2.091108142407153e-06,
      "loss": 0.9118,
      "num_input_tokens_seen": 88645070,
      "step": 4166
    },
    {
      "epoch": 0.5010521252930921,
      "flos": 41353940705280.0,
      "grad_norm": 0.9141640891980882,
      "learning_rate": 2.090329972539377e-06,
      "loss": 0.8805,
      "num_input_tokens_seen": 88703355,
      "step": 4167
    },
    {
      "epoch": 0.5011723681837311,
      "flos": 13204428963840.0,
      "grad_norm": 5.205359176863423,
      "learning_rate": 2.089551788968616e-06,
      "loss": 0.8722,
      "num_input_tokens_seen": 88721040,
      "step": 4168
    },
    {
      "epoch": 0.5012926110743702,
      "flos": 39986484449280.0,
      "grad_norm": 0.8790397373131338,
      "learning_rate": 2.08877359181292e-06,
      "loss": 0.8307,
      "num_input_tokens_seen": 88777325,
      "step": 4169
    },
    {
      "epoch": 0.5014128539650093,
      "flos": 17237401190400.0,
      "grad_norm": 6.763728556559348,
      "learning_rate": 2.0879953811903396e-06,
      "loss": 1.0349,
      "num_input_tokens_seen": 88791930,
      "step": 4170
    },
    {
      "epoch": 0.5015330968556484,
      "flos": 19602028707840.0,
      "grad_norm": 3.202717080542402,
      "learning_rate": 2.08721715721893e-06,
      "loss": 0.9785,
      "num_input_tokens_seen": 88810975,
      "step": 4171
    },
    {
      "epoch": 0.5016533397462875,
      "flos": 16923053629440.0,
      "grad_norm": 4.796236374681341,
      "learning_rate": 2.0864389200167477e-06,
      "loss": 0.9677,
      "num_input_tokens_seen": 88828950,
      "step": 4172
    },
    {
      "epoch": 0.5017735826369266,
      "flos": 17996681011200.0,
      "grad_norm": 12.420004035592767,
      "learning_rate": 2.0856606697018504e-06,
      "loss": 0.9928,
      "num_input_tokens_seen": 88846680,
      "step": 4173
    },
    {
      "epoch": 0.5018938255275657,
      "flos": 11944953937920.0,
      "grad_norm": 4.1520557321619025,
      "learning_rate": 2.084882406392297e-06,
      "loss": 0.92,
      "num_input_tokens_seen": 88864360,
      "step": 4174
    },
    {
      "epoch": 0.5020140684182047,
      "flos": 18155295744000.0,
      "grad_norm": 14.552607372826914,
      "learning_rate": 2.0841041302061496e-06,
      "loss": 0.8901,
      "num_input_tokens_seen": 88883540,
      "step": 4175
    },
    {
      "epoch": 0.5021343113088439,
      "flos": 16821939916800.0,
      "grad_norm": 7.552798014181074,
      "learning_rate": 2.083325841261473e-06,
      "loss": 0.9429,
      "num_input_tokens_seen": 88902320,
      "step": 4176
    },
    {
      "epoch": 0.502254554199483,
      "flos": 17451033600000.0,
      "grad_norm": 3.995628710225467,
      "learning_rate": 2.0825475396763322e-06,
      "loss": 0.8623,
      "num_input_tokens_seen": 88922690,
      "step": 4177
    },
    {
      "epoch": 0.502374797090122,
      "flos": 24426327060480.0,
      "grad_norm": 3.309064415035834,
      "learning_rate": 2.081769225568796e-06,
      "loss": 0.8494,
      "num_input_tokens_seen": 88944860,
      "step": 4178
    },
    {
      "epoch": 0.5024950399807612,
      "flos": 18526561812480.0,
      "grad_norm": 3.1235256531847972,
      "learning_rate": 2.0809908990569327e-06,
      "loss": 0.9523,
      "num_input_tokens_seen": 88966360,
      "step": 4179
    },
    {
      "epoch": 0.5026152828714002,
      "flos": 15092353843200.0,
      "grad_norm": 3.918986316716397,
      "learning_rate": 2.0802125602588146e-06,
      "loss": 0.9898,
      "num_input_tokens_seen": 88985345,
      "step": 4180
    },
    {
      "epoch": 0.5027355257620393,
      "flos": 22065991741440.0,
      "grad_norm": 3.3732271129736704,
      "learning_rate": 2.0794342092925146e-06,
      "loss": 0.8571,
      "num_input_tokens_seen": 89006200,
      "step": 4181
    },
    {
      "epoch": 0.5028557686526784,
      "flos": 17635501608960.0,
      "grad_norm": 11.965935358708448,
      "learning_rate": 2.078655846276108e-06,
      "loss": 0.8753,
      "num_input_tokens_seen": 89026250,
      "step": 4182
    },
    {
      "epoch": 0.5029760115433175,
      "flos": 16324534640640.0,
      "grad_norm": 9.674495114107538,
      "learning_rate": 2.0778774713276727e-06,
      "loss": 0.8764,
      "num_input_tokens_seen": 89045445,
      "step": 4183
    },
    {
      "epoch": 0.5030962544339566,
      "flos": 10818148392960.0,
      "grad_norm": 5.8285687561456445,
      "learning_rate": 2.077099084565287e-06,
      "loss": 0.855,
      "num_input_tokens_seen": 89062570,
      "step": 4184
    },
    {
      "epoch": 0.5032164973245957,
      "flos": 17421961175040.0,
      "grad_norm": 7.16031341463492,
      "learning_rate": 2.0763206861070313e-06,
      "loss": 0.8335,
      "num_input_tokens_seen": 89081350,
      "step": 4185
    },
    {
      "epoch": 0.5033367402152348,
      "flos": 11470397399040.0,
      "grad_norm": 5.552660866115613,
      "learning_rate": 2.0755422760709876e-06,
      "loss": 0.9494,
      "num_input_tokens_seen": 89098470,
      "step": 4186
    },
    {
      "epoch": 0.5034569831058738,
      "flos": 15191260139520.0,
      "grad_norm": 5.251914992075212,
      "learning_rate": 2.0747638545752417e-06,
      "loss": 0.9654,
      "num_input_tokens_seen": 89116750,
      "step": 4187
    },
    {
      "epoch": 0.503577225996513,
      "flos": 14593752883200.0,
      "grad_norm": 4.520339141883229,
      "learning_rate": 2.073985421737878e-06,
      "loss": 1.0284,
      "num_input_tokens_seen": 89133780,
      "step": 4188
    },
    {
      "epoch": 0.5036974688871521,
      "flos": 19386771394560.0,
      "grad_norm": 6.724031865086979,
      "learning_rate": 2.0732069776769844e-06,
      "loss": 0.9499,
      "num_input_tokens_seen": 89150910,
      "step": 4189
    },
    {
      "epoch": 0.5038177117777911,
      "flos": 14488316313600.0,
      "grad_norm": 5.456785356228536,
      "learning_rate": 2.072428522510651e-06,
      "loss": 0.9221,
      "num_input_tokens_seen": 89167195,
      "step": 4190
    },
    {
      "epoch": 0.5039379546684303,
      "flos": 15563323330560.0,
      "grad_norm": 4.9951350386105755,
      "learning_rate": 2.071650056356968e-06,
      "loss": 0.9431,
      "num_input_tokens_seen": 89184455,
      "step": 4191
    },
    {
      "epoch": 0.5040581975590693,
      "flos": 14199791370240.0,
      "grad_norm": 3.8955237253686015,
      "learning_rate": 2.070871579334028e-06,
      "loss": 0.9865,
      "num_input_tokens_seen": 89203285,
      "step": 4192
    },
    {
      "epoch": 0.5041784404497084,
      "flos": 14881971240960.0,
      "grad_norm": 4.883249589811736,
      "learning_rate": 2.0700930915599264e-06,
      "loss": 0.9163,
      "num_input_tokens_seen": 89222735,
      "step": 4193
    },
    {
      "epoch": 0.5042986833403476,
      "flos": 8798504448000.0,
      "grad_norm": 9.575382904284918,
      "learning_rate": 2.0693145931527583e-06,
      "loss": 0.9775,
      "num_input_tokens_seen": 89238935,
      "step": 4194
    },
    {
      "epoch": 0.5044189262309866,
      "flos": 20805229056000.0,
      "grad_norm": 3.645717088571847,
      "learning_rate": 2.068536084230622e-06,
      "loss": 0.9794,
      "num_input_tokens_seen": 89260520,
      "step": 4195
    },
    {
      "epoch": 0.5045391691216257,
      "flos": 16975005450240.0,
      "grad_norm": 5.48052965430788,
      "learning_rate": 2.067757564911616e-06,
      "loss": 1.0695,
      "num_input_tokens_seen": 89278815,
      "step": 4196
    },
    {
      "epoch": 0.5046594120122648,
      "flos": 17530432942080.0,
      "grad_norm": 4.46054241080979,
      "learning_rate": 2.0669790353138407e-06,
      "loss": 1.1114,
      "num_input_tokens_seen": 89297500,
      "step": 4197
    },
    {
      "epoch": 0.5047796549029039,
      "flos": 16608276848640.0,
      "grad_norm": 6.482665790360607,
      "learning_rate": 2.0662004955553995e-06,
      "loss": 0.9259,
      "num_input_tokens_seen": 89316920,
      "step": 4198
    },
    {
      "epoch": 0.5048998977935429,
      "flos": 12253690982400.0,
      "grad_norm": 3.2050520891786056,
      "learning_rate": 2.065421945754395e-06,
      "loss": 0.9569,
      "num_input_tokens_seen": 89334370,
      "step": 4199
    },
    {
      "epoch": 0.505020140684182,
      "flos": 24869051904000.0,
      "grad_norm": 47.15980147328867,
      "learning_rate": 2.0646433860289344e-06,
      "loss": 0.9793,
      "num_input_tokens_seen": 89353015,
      "step": 4200
    },
    {
      "epoch": 0.5051403835748212,
      "flos": 17687821332480.0,
      "grad_norm": 7.247332641311448,
      "learning_rate": 2.0638648164971233e-06,
      "loss": 1.0164,
      "num_input_tokens_seen": 89371200,
      "step": 4201
    },
    {
      "epoch": 0.5052606264654602,
      "flos": 14881848606720.0,
      "grad_norm": 3.294575255228771,
      "learning_rate": 2.06308623727707e-06,
      "loss": 1.0865,
      "num_input_tokens_seen": 89391020,
      "step": 4202
    },
    {
      "epoch": 0.5053808693560993,
      "flos": 13934605701120.0,
      "grad_norm": 4.998367996901218,
      "learning_rate": 2.0623076484868846e-06,
      "loss": 0.9396,
      "num_input_tokens_seen": 89408195,
      "step": 4203
    },
    {
      "epoch": 0.5055011122467384,
      "flos": 48336508354560.0,
      "grad_norm": 0.9463448995172509,
      "learning_rate": 2.061529050244679e-06,
      "loss": 0.8808,
      "num_input_tokens_seen": 89467660,
      "step": 4204
    },
    {
      "epoch": 0.5056213551373775,
      "flos": 11944739328000.0,
      "grad_norm": 4.200487635324227,
      "learning_rate": 2.060750442668565e-06,
      "loss": 0.9423,
      "num_input_tokens_seen": 89485135,
      "step": 4205
    },
    {
      "epoch": 0.5057415980280165,
      "flos": 10837992591360.0,
      "grad_norm": 4.144592227181615,
      "learning_rate": 2.059971825876657e-06,
      "loss": 0.8375,
      "num_input_tokens_seen": 89499365,
      "step": 4206
    },
    {
      "epoch": 0.5058618409186557,
      "flos": 13492126126080.0,
      "grad_norm": 10.555573669774981,
      "learning_rate": 2.0591931999870713e-06,
      "loss": 0.9671,
      "num_input_tokens_seen": 89518010,
      "step": 4207
    },
    {
      "epoch": 0.5059820838092948,
      "flos": 45424301936640.0,
      "grad_norm": 0.912474891891981,
      "learning_rate": 2.0584145651179234e-06,
      "loss": 0.8278,
      "num_input_tokens_seen": 89573440,
      "step": 4208
    },
    {
      "epoch": 0.5061023266999338,
      "flos": 10915092541440.0,
      "grad_norm": 4.980872188010439,
      "learning_rate": 2.0576359213873327e-06,
      "loss": 0.9786,
      "num_input_tokens_seen": 89588310,
      "step": 4209
    },
    {
      "epoch": 0.506222569590573,
      "flos": 15953053962240.0,
      "grad_norm": 3.876186435329589,
      "learning_rate": 2.056857268913419e-06,
      "loss": 0.881,
      "num_input_tokens_seen": 89608080,
      "step": 4210
    },
    {
      "epoch": 0.506342812481212,
      "flos": 12436534087680.0,
      "grad_norm": 3.917136076265162,
      "learning_rate": 2.056078607814303e-06,
      "loss": 1.0381,
      "num_input_tokens_seen": 89623585,
      "step": 4211
    },
    {
      "epoch": 0.5064630553718511,
      "flos": 16636828078080.0,
      "grad_norm": 4.7555978816714735,
      "learning_rate": 2.055299938208106e-06,
      "loss": 0.98,
      "num_input_tokens_seen": 89644295,
      "step": 4212
    },
    {
      "epoch": 0.5065832982624903,
      "flos": 17057409331200.0,
      "grad_norm": 8.396107053120568,
      "learning_rate": 2.0545212602129526e-06,
      "loss": 1.0568,
      "num_input_tokens_seen": 89663870,
      "step": 4213
    },
    {
      "epoch": 0.5067035411531293,
      "flos": 15271824506880.0,
      "grad_norm": 6.8176239062842905,
      "learning_rate": 2.0537425739469673e-06,
      "loss": 0.8287,
      "num_input_tokens_seen": 89682525,
      "step": 4214
    },
    {
      "epoch": 0.5068237840437684,
      "flos": 47207587368960.0,
      "grad_norm": 0.9255126370289071,
      "learning_rate": 2.052963879528276e-06,
      "loss": 0.8307,
      "num_input_tokens_seen": 89742115,
      "step": 4215
    },
    {
      "epoch": 0.5069440269344075,
      "flos": 19411766231040.0,
      "grad_norm": 4.2264629923446835,
      "learning_rate": 2.052185177075007e-06,
      "loss": 0.9561,
      "num_input_tokens_seen": 89761405,
      "step": 4216
    },
    {
      "epoch": 0.5070642698250466,
      "flos": 16611250728960.0,
      "grad_norm": 3.76511933663192,
      "learning_rate": 2.051406466705288e-06,
      "loss": 1.0186,
      "num_input_tokens_seen": 89780665,
      "step": 4217
    },
    {
      "epoch": 0.5071845127156857,
      "flos": 14436671078400.0,
      "grad_norm": 3.525818014259632,
      "learning_rate": 2.0506277485372486e-06,
      "loss": 1.0016,
      "num_input_tokens_seen": 89799210,
      "step": 4218
    },
    {
      "epoch": 0.5073047556063248,
      "flos": 8509396992000.0,
      "grad_norm": 4.4254022016768655,
      "learning_rate": 2.04984902268902e-06,
      "loss": 0.8597,
      "num_input_tokens_seen": 89816130,
      "step": 4219
    },
    {
      "epoch": 0.5074249984969639,
      "flos": 13964229980160.0,
      "grad_norm": 4.235239088567386,
      "learning_rate": 2.0490702892787345e-06,
      "loss": 0.9307,
      "num_input_tokens_seen": 89834910,
      "step": 4220
    },
    {
      "epoch": 0.5075452413876029,
      "flos": 20491617300480.0,
      "grad_norm": 3.7719712340713336,
      "learning_rate": 2.0482915484245246e-06,
      "loss": 0.8228,
      "num_input_tokens_seen": 89856250,
      "step": 4221
    },
    {
      "epoch": 0.5076654842782421,
      "flos": 14436088565760.0,
      "grad_norm": 9.21419171932771,
      "learning_rate": 2.047512800244526e-06,
      "loss": 1.0268,
      "num_input_tokens_seen": 89871235,
      "step": 4222
    },
    {
      "epoch": 0.5077857271688812,
      "flos": 18762705715200.0,
      "grad_norm": 4.856658274255613,
      "learning_rate": 2.046734044856873e-06,
      "loss": 0.9747,
      "num_input_tokens_seen": 89890365,
      "step": 4223
    },
    {
      "epoch": 0.5079059700595202,
      "flos": 15485027696640.0,
      "grad_norm": 4.840983135847265,
      "learning_rate": 2.045955282379702e-06,
      "loss": 0.9978,
      "num_input_tokens_seen": 89908745,
      "step": 4224
    },
    {
      "epoch": 0.5080262129501594,
      "flos": 9296400261120.0,
      "grad_norm": 11.156297186743696,
      "learning_rate": 2.045176512931152e-06,
      "loss": 0.9464,
      "num_input_tokens_seen": 89923095,
      "step": 4225
    },
    {
      "epoch": 0.5081464558407984,
      "flos": 18002199552000.0,
      "grad_norm": 4.03836260104991,
      "learning_rate": 2.0443977366293604e-06,
      "loss": 0.944,
      "num_input_tokens_seen": 89940855,
      "step": 4226
    },
    {
      "epoch": 0.5082666987314375,
      "flos": 22061822177280.0,
      "grad_norm": 3.002266980990612,
      "learning_rate": 2.043618953592468e-06,
      "loss": 0.9698,
      "num_input_tokens_seen": 89963030,
      "step": 4227
    },
    {
      "epoch": 0.5083869416220766,
      "flos": 13963923394560.0,
      "grad_norm": 3.102246571778952,
      "learning_rate": 2.0428401639386144e-06,
      "loss": 1.0045,
      "num_input_tokens_seen": 89983315,
      "step": 4228
    },
    {
      "epoch": 0.5085071845127157,
      "flos": 51436892467200.0,
      "grad_norm": 0.9170248170030022,
      "learning_rate": 2.042061367785943e-06,
      "loss": 0.8409,
      "num_input_tokens_seen": 90036175,
      "step": 4229
    },
    {
      "epoch": 0.5086274274033548,
      "flos": 25655748587520.0,
      "grad_norm": 6.3168975318181495,
      "learning_rate": 2.041282565252594e-06,
      "loss": 0.9474,
      "num_input_tokens_seen": 90056060,
      "step": 4230
    },
    {
      "epoch": 0.5087476702939938,
      "flos": 16717392445440.0,
      "grad_norm": 2.9318657267010537,
      "learning_rate": 2.040503756456714e-06,
      "loss": 0.9562,
      "num_input_tokens_seen": 90074990,
      "step": 4231
    },
    {
      "epoch": 0.508867913184633,
      "flos": 10686490644480.0,
      "grad_norm": 9.719869371329182,
      "learning_rate": 2.0397249415164456e-06,
      "loss": 0.9946,
      "num_input_tokens_seen": 90092065,
      "step": 4232
    },
    {
      "epoch": 0.508988156075272,
      "flos": 18421799731200.0,
      "grad_norm": 4.036224607858562,
      "learning_rate": 2.0389461205499354e-06,
      "loss": 0.994,
      "num_input_tokens_seen": 90110920,
      "step": 4233
    },
    {
      "epoch": 0.5091083989659111,
      "flos": 9768534773760.0,
      "grad_norm": 3.8737889520115267,
      "learning_rate": 2.03816729367533e-06,
      "loss": 0.9262,
      "num_input_tokens_seen": 90128795,
      "step": 4234
    },
    {
      "epoch": 0.5092286418565503,
      "flos": 14986089492480.0,
      "grad_norm": 3.11516947424558,
      "learning_rate": 2.0373884610107765e-06,
      "loss": 0.8929,
      "num_input_tokens_seen": 90148830,
      "step": 4235
    },
    {
      "epoch": 0.5093488847471893,
      "flos": 13200841912320.0,
      "grad_norm": 6.3427237446330915,
      "learning_rate": 2.0366096226744225e-06,
      "loss": 0.8795,
      "num_input_tokens_seen": 90163745,
      "step": 4236
    },
    {
      "epoch": 0.5094691276378284,
      "flos": 16925598289920.0,
      "grad_norm": 3.83016875848909,
      "learning_rate": 2.035830778784418e-06,
      "loss": 0.9607,
      "num_input_tokens_seen": 90184140,
      "step": 4237
    },
    {
      "epoch": 0.5095893705284675,
      "flos": 12338210304000.0,
      "grad_norm": 4.60578142161155,
      "learning_rate": 2.0350519294589134e-06,
      "loss": 0.9985,
      "num_input_tokens_seen": 90201900,
      "step": 4238
    },
    {
      "epoch": 0.5097096134191066,
      "flos": 18396038430720.0,
      "grad_norm": 3.4202033710801962,
      "learning_rate": 2.0342730748160588e-06,
      "loss": 1.0193,
      "num_input_tokens_seen": 90222085,
      "step": 4239
    },
    {
      "epoch": 0.5098298563097456,
      "flos": 19758282731520.0,
      "grad_norm": 4.085521085974873,
      "learning_rate": 2.033494214974006e-06,
      "loss": 0.8829,
      "num_input_tokens_seen": 90242950,
      "step": 4240
    },
    {
      "epoch": 0.5099500992003848,
      "flos": 15168472719360.0,
      "grad_norm": 2.942616167003613,
      "learning_rate": 2.0327153500509067e-06,
      "loss": 1.0268,
      "num_input_tokens_seen": 90261695,
      "step": 4241
    },
    {
      "epoch": 0.5100703420910239,
      "flos": 14095949045760.0,
      "grad_norm": 6.167581267020544,
      "learning_rate": 2.031936480164916e-06,
      "loss": 1.0516,
      "num_input_tokens_seen": 90279155,
      "step": 4242
    },
    {
      "epoch": 0.5101905849816629,
      "flos": 17533682749440.0,
      "grad_norm": 6.336296024362309,
      "learning_rate": 2.0311576054341857e-06,
      "loss": 0.999,
      "num_input_tokens_seen": 90299490,
      "step": 4243
    },
    {
      "epoch": 0.5103108278723021,
      "flos": 16297945559040.0,
      "grad_norm": 3.2914575115986504,
      "learning_rate": 2.0303787259768715e-06,
      "loss": 0.8241,
      "num_input_tokens_seen": 90320110,
      "step": 4244
    },
    {
      "epoch": 0.5104310707629411,
      "flos": 15274859704320.0,
      "grad_norm": 4.823240506839402,
      "learning_rate": 2.0295998419111294e-06,
      "loss": 0.8679,
      "num_input_tokens_seen": 90337120,
      "step": 4245
    },
    {
      "epoch": 0.5105513136535802,
      "flos": 10529163571200.0,
      "grad_norm": 8.72874397462268,
      "learning_rate": 2.028820953355115e-06,
      "loss": 0.9142,
      "num_input_tokens_seen": 90354940,
      "step": 4246
    },
    {
      "epoch": 0.5106715565442194,
      "flos": 16062384168960.0,
      "grad_norm": 4.666697002493941,
      "learning_rate": 2.0280420604269834e-06,
      "loss": 0.974,
      "num_input_tokens_seen": 90374200,
      "step": 4247
    },
    {
      "epoch": 0.5107917994348584,
      "flos": 50870940979200.0,
      "grad_norm": 0.8365931212699413,
      "learning_rate": 2.027263163244895e-06,
      "loss": 0.8372,
      "num_input_tokens_seen": 90443045,
      "step": 4248
    },
    {
      "epoch": 0.5109120423254975,
      "flos": 17661538836480.0,
      "grad_norm": 2.701191116672699,
      "learning_rate": 2.026484261927005e-06,
      "loss": 0.9344,
      "num_input_tokens_seen": 90462260,
      "step": 4249
    },
    {
      "epoch": 0.5110322852161366,
      "flos": 15088276254720.0,
      "grad_norm": 4.223741982873791,
      "learning_rate": 2.025705356591475e-06,
      "loss": 0.9309,
      "num_input_tokens_seen": 90479670,
      "step": 4250
    },
    {
      "epoch": 0.5111525281067757,
      "flos": 47584862515200.0,
      "grad_norm": 0.8566199095343895,
      "learning_rate": 2.024926447356462e-06,
      "loss": 0.8149,
      "num_input_tokens_seen": 90541675,
      "step": 4251
    },
    {
      "epoch": 0.5112727709974147,
      "flos": 10502973050880.0,
      "grad_norm": 4.587589414800974,
      "learning_rate": 2.024147534340127e-06,
      "loss": 0.9831,
      "num_input_tokens_seen": 90559255,
      "step": 4252
    },
    {
      "epoch": 0.5113930138880539,
      "flos": 15038317240320.0,
      "grad_norm": 3.8405383579050527,
      "learning_rate": 2.02336861766063e-06,
      "loss": 0.9941,
      "num_input_tokens_seen": 90578035,
      "step": 4253
    },
    {
      "epoch": 0.511513256778693,
      "flos": 14485710336000.0,
      "grad_norm": 3.6852714140484464,
      "learning_rate": 2.0225896974361327e-06,
      "loss": 0.9786,
      "num_input_tokens_seen": 90597860,
      "step": 4254
    },
    {
      "epoch": 0.511633499669332,
      "flos": 50043889520640.0,
      "grad_norm": 0.945701960263959,
      "learning_rate": 2.0218107737847962e-06,
      "loss": 0.8475,
      "num_input_tokens_seen": 90659625,
      "step": 4255
    },
    {
      "epoch": 0.5117537425599712,
      "flos": 17137237893120.0,
      "grad_norm": 4.702603231997441,
      "learning_rate": 2.0210318468247826e-06,
      "loss": 0.951,
      "num_input_tokens_seen": 90678315,
      "step": 4256
    },
    {
      "epoch": 0.5118739854506102,
      "flos": 14881725972480.0,
      "grad_norm": 2.658348869908133,
      "learning_rate": 2.020252916674255e-06,
      "loss": 1.0113,
      "num_input_tokens_seen": 90697150,
      "step": 4257
    },
    {
      "epoch": 0.5119942283412493,
      "flos": 12364799385600.0,
      "grad_norm": 4.587943007363797,
      "learning_rate": 2.019473983451375e-06,
      "loss": 0.9986,
      "num_input_tokens_seen": 90715290,
      "step": 4258
    },
    {
      "epoch": 0.5121144712318885,
      "flos": 14958458019840.0,
      "grad_norm": 4.025405076772335,
      "learning_rate": 2.0186950472743076e-06,
      "loss": 0.9077,
      "num_input_tokens_seen": 90734915,
      "step": 4259
    },
    {
      "epoch": 0.5122347141225275,
      "flos": 14091503554560.0,
      "grad_norm": 3.5139085436553166,
      "learning_rate": 2.0179161082612162e-06,
      "loss": 0.9425,
      "num_input_tokens_seen": 90754025,
      "step": 4260
    },
    {
      "epoch": 0.5123549570131666,
      "flos": 16009297981440.0,
      "grad_norm": 3.389224558675754,
      "learning_rate": 2.017137166530266e-06,
      "loss": 0.9096,
      "num_input_tokens_seen": 90773280,
      "step": 4261
    },
    {
      "epoch": 0.5124751999038056,
      "flos": 14430907269120.0,
      "grad_norm": 4.2221787824125,
      "learning_rate": 2.0163582221996213e-06,
      "loss": 0.9989,
      "num_input_tokens_seen": 90791375,
      "step": 4262
    },
    {
      "epoch": 0.5125954427944448,
      "flos": 28411884933120.0,
      "grad_norm": 3.7235717288213657,
      "learning_rate": 2.015579275387446e-06,
      "loss": 0.8765,
      "num_input_tokens_seen": 90816415,
      "step": 4263
    },
    {
      "epoch": 0.5127156856850839,
      "flos": 20807865692160.0,
      "grad_norm": 4.35771842299683,
      "learning_rate": 2.0148003262119085e-06,
      "loss": 0.8734,
      "num_input_tokens_seen": 90837105,
      "step": 4264
    },
    {
      "epoch": 0.5128359285757229,
      "flos": 9559378513920.0,
      "grad_norm": 3.985477549275164,
      "learning_rate": 2.0140213747911728e-06,
      "loss": 0.9507,
      "num_input_tokens_seen": 90855360,
      "step": 4265
    },
    {
      "epoch": 0.5129561714663621,
      "flos": 17923597332480.0,
      "grad_norm": 3.80236018318182,
      "learning_rate": 2.013242421243406e-06,
      "loss": 0.9918,
      "num_input_tokens_seen": 90873985,
      "step": 4266
    },
    {
      "epoch": 0.5130764143570011,
      "flos": 12862725857280.0,
      "grad_norm": 3.0081999293905892,
      "learning_rate": 2.012463465686774e-06,
      "loss": 0.9863,
      "num_input_tokens_seen": 90893455,
      "step": 4267
    },
    {
      "epoch": 0.5131966572476402,
      "flos": 42798834155520.0,
      "grad_norm": 0.8815596476133137,
      "learning_rate": 2.0116845082394446e-06,
      "loss": 0.7917,
      "num_input_tokens_seen": 90958010,
      "step": 4268
    },
    {
      "epoch": 0.5133169001382794,
      "flos": 13125060280320.0,
      "grad_norm": 5.723890318624098,
      "learning_rate": 2.0109055490195836e-06,
      "loss": 0.9856,
      "num_input_tokens_seen": 90976185,
      "step": 4269
    },
    {
      "epoch": 0.5134371430289184,
      "flos": 10975015587840.0,
      "grad_norm": 9.366943152567838,
      "learning_rate": 2.0101265881453605e-06,
      "loss": 0.8385,
      "num_input_tokens_seen": 90994380,
      "step": 4270
    },
    {
      "epoch": 0.5135573859195575,
      "flos": 15222539980800.0,
      "grad_norm": 5.8317689523651,
      "learning_rate": 2.009347625734941e-06,
      "loss": 0.9794,
      "num_input_tokens_seen": 91014010,
      "step": 4271
    },
    {
      "epoch": 0.5136776288101966,
      "flos": 12548286320640.0,
      "grad_norm": 4.131267525640067,
      "learning_rate": 2.0085686619064954e-06,
      "loss": 0.9584,
      "num_input_tokens_seen": 91030170,
      "step": 4272
    },
    {
      "epoch": 0.5137978717008357,
      "flos": 11735491092480.0,
      "grad_norm": 4.989286770963377,
      "learning_rate": 2.00778969677819e-06,
      "loss": 1.0268,
      "num_input_tokens_seen": 91046925,
      "step": 4273
    },
    {
      "epoch": 0.5139181145914747,
      "flos": 14672447078400.0,
      "grad_norm": 2.656230253258402,
      "learning_rate": 2.0070107304681934e-06,
      "loss": 0.8383,
      "num_input_tokens_seen": 91065600,
      "step": 4274
    },
    {
      "epoch": 0.5140383574821139,
      "flos": 23482517913600.0,
      "grad_norm": 3.2822098554227424,
      "learning_rate": 2.006231763094675e-06,
      "loss": 0.9874,
      "num_input_tokens_seen": 91086340,
      "step": 4275
    },
    {
      "epoch": 0.514158600372753,
      "flos": 13859498557440.0,
      "grad_norm": 4.23541584806095,
      "learning_rate": 2.0054527947758027e-06,
      "loss": 1.0788,
      "num_input_tokens_seen": 91104860,
      "step": 4276
    },
    {
      "epoch": 0.514278843263392,
      "flos": 44759076249600.0,
      "grad_norm": 0.780028036053852,
      "learning_rate": 2.004673825629746e-06,
      "loss": 0.7831,
      "num_input_tokens_seen": 91165360,
      "step": 4277
    },
    {
      "epoch": 0.5143990861540312,
      "flos": 18419255070720.0,
      "grad_norm": 3.092931619928607,
      "learning_rate": 2.0038948557746744e-06,
      "loss": 0.9222,
      "num_input_tokens_seen": 91186935,
      "step": 4278
    },
    {
      "epoch": 0.5145193290446702,
      "flos": 16585734696960.0,
      "grad_norm": 2.986698544309432,
      "learning_rate": 2.0031158853287558e-06,
      "loss": 0.9399,
      "num_input_tokens_seen": 91207090,
      "step": 4279
    },
    {
      "epoch": 0.5146395719353093,
      "flos": 16245073981440.0,
      "grad_norm": 4.076747383905181,
      "learning_rate": 2.0023369144101593e-06,
      "loss": 0.9123,
      "num_input_tokens_seen": 91224980,
      "step": 4280
    },
    {
      "epoch": 0.5147598148259485,
      "flos": 18785033256960.0,
      "grad_norm": 2.799210580169228,
      "learning_rate": 2.0015579431370555e-06,
      "loss": 0.961,
      "num_input_tokens_seen": 91246380,
      "step": 4281
    },
    {
      "epoch": 0.5148800577165875,
      "flos": 21353482444800.0,
      "grad_norm": 3.628782863592965,
      "learning_rate": 2.000778971627612e-06,
      "loss": 0.9004,
      "num_input_tokens_seen": 91265565,
      "step": 4282
    },
    {
      "epoch": 0.5150003006072266,
      "flos": 12705460101120.0,
      "grad_norm": 4.910460396681053,
      "learning_rate": 2e-06,
      "loss": 1.0973,
      "num_input_tokens_seen": 91282880,
      "step": 4283
    },
    {
      "epoch": 0.5151205434978657,
      "flos": 13098839101440.0,
      "grad_norm": 4.954525144979229,
      "learning_rate": 1.9992210283723878e-06,
      "loss": 1.0498,
      "num_input_tokens_seen": 91299840,
      "step": 4284
    },
    {
      "epoch": 0.5152407863885048,
      "flos": 18031241318400.0,
      "grad_norm": 2.608277141302953,
      "learning_rate": 1.9984420568629448e-06,
      "loss": 0.9828,
      "num_input_tokens_seen": 91322325,
      "step": 4285
    },
    {
      "epoch": 0.5153610292791438,
      "flos": 12991348408320.0,
      "grad_norm": 3.961523971929967,
      "learning_rate": 1.9976630855898405e-06,
      "loss": 0.9813,
      "num_input_tokens_seen": 91339800,
      "step": 4286
    },
    {
      "epoch": 0.515481272169783,
      "flos": 21673624473600.0,
      "grad_norm": 4.113231401422184,
      "learning_rate": 1.9968841146712445e-06,
      "loss": 0.9444,
      "num_input_tokens_seen": 91359135,
      "step": 4287
    },
    {
      "epoch": 0.5156015150604221,
      "flos": 16374708264960.0,
      "grad_norm": 3.5462038186541815,
      "learning_rate": 1.996105144225326e-06,
      "loss": 0.9082,
      "num_input_tokens_seen": 91379090,
      "step": 4288
    },
    {
      "epoch": 0.5157217579510611,
      "flos": 12653140377600.0,
      "grad_norm": 6.381850507467109,
      "learning_rate": 1.995326174370254e-06,
      "loss": 0.9809,
      "num_input_tokens_seen": 91397910,
      "step": 4289
    },
    {
      "epoch": 0.5158420008417003,
      "flos": 13596642938880.0,
      "grad_norm": 2.765617930947445,
      "learning_rate": 1.994547205224197e-06,
      "loss": 0.9228,
      "num_input_tokens_seen": 91416535,
      "step": 4290
    },
    {
      "epoch": 0.5159622437323393,
      "flos": 15931063664640.0,
      "grad_norm": 5.095226612086613,
      "learning_rate": 1.993768236905325e-06,
      "loss": 0.8721,
      "num_input_tokens_seen": 91436925,
      "step": 4291
    },
    {
      "epoch": 0.5160824866229784,
      "flos": 17499521003520.0,
      "grad_norm": 3.035448607671837,
      "learning_rate": 1.992989269531807e-06,
      "loss": 0.8527,
      "num_input_tokens_seen": 91455455,
      "step": 4292
    },
    {
      "epoch": 0.5162027295136175,
      "flos": 12756522823680.0,
      "grad_norm": 6.289018524205724,
      "learning_rate": 1.99221030322181e-06,
      "loss": 0.8564,
      "num_input_tokens_seen": 91471980,
      "step": 4293
    },
    {
      "epoch": 0.5163229724042566,
      "flos": 19470554910720.0,
      "grad_norm": 2.513355534587856,
      "learning_rate": 1.991431338093505e-06,
      "loss": 1.0051,
      "num_input_tokens_seen": 91494055,
      "step": 4294
    },
    {
      "epoch": 0.5164432152948957,
      "flos": 15459051786240.0,
      "grad_norm": 2.9101688666358223,
      "learning_rate": 1.9906523742650587e-06,
      "loss": 0.9797,
      "num_input_tokens_seen": 91512635,
      "step": 4295
    },
    {
      "epoch": 0.5165634581855347,
      "flos": 18180167946240.0,
      "grad_norm": 4.235857376809895,
      "learning_rate": 1.9898734118546397e-06,
      "loss": 0.9515,
      "num_input_tokens_seen": 91532305,
      "step": 4296
    },
    {
      "epoch": 0.5166837010761739,
      "flos": 14121066516480.0,
      "grad_norm": 3.0258642112888494,
      "learning_rate": 1.989094450980416e-06,
      "loss": 1.0024,
      "num_input_tokens_seen": 91552125,
      "step": 4297
    },
    {
      "epoch": 0.516803943966813,
      "flos": 18967815045120.0,
      "grad_norm": 3.6573263035047363,
      "learning_rate": 1.9883154917605556e-06,
      "loss": 0.9495,
      "num_input_tokens_seen": 91571055,
      "step": 4298
    },
    {
      "epoch": 0.516924186857452,
      "flos": 13964229980160.0,
      "grad_norm": 6.40592129492072,
      "learning_rate": 1.9875365343132262e-06,
      "loss": 1.0171,
      "num_input_tokens_seen": 91587895,
      "step": 4299
    },
    {
      "epoch": 0.5170444297480912,
      "flos": 11285592145920.0,
      "grad_norm": 4.390649978729547,
      "learning_rate": 1.9867575787565946e-06,
      "loss": 1.0404,
      "num_input_tokens_seen": 91602275,
      "step": 4300
    },
    {
      "epoch": 0.5171646726387302,
      "flos": 10005322506240.0,
      "grad_norm": 16.717713168784478,
      "learning_rate": 1.9859786252088275e-06,
      "loss": 1.0579,
      "num_input_tokens_seen": 91619175,
      "step": 4301
    },
    {
      "epoch": 0.5172849155293693,
      "flos": 16762875310080.0,
      "grad_norm": 5.049762100810321,
      "learning_rate": 1.9851996737880914e-06,
      "loss": 0.8649,
      "num_input_tokens_seen": 91634080,
      "step": 4302
    },
    {
      "epoch": 0.5174051584200084,
      "flos": 10083924725760.0,
      "grad_norm": 3.6522176302589657,
      "learning_rate": 1.9844207246125537e-06,
      "loss": 0.9415,
      "num_input_tokens_seen": 91650380,
      "step": 4303
    },
    {
      "epoch": 0.5175254013106475,
      "flos": 27048935485440.0,
      "grad_norm": 4.228586216573007,
      "learning_rate": 1.983641777800379e-06,
      "loss": 0.8771,
      "num_input_tokens_seen": 91672745,
      "step": 4304
    },
    {
      "epoch": 0.5176456442012866,
      "flos": 49090537451520.0,
      "grad_norm": 0.8427187657663401,
      "learning_rate": 1.9828628334697343e-06,
      "loss": 0.819,
      "num_input_tokens_seen": 91739675,
      "step": 4305
    },
    {
      "epoch": 0.5177658870919257,
      "flos": 45879259545600.0,
      "grad_norm": 0.7937422864644276,
      "learning_rate": 1.982083891738784e-06,
      "loss": 0.7835,
      "num_input_tokens_seen": 91800265,
      "step": 4306
    },
    {
      "epoch": 0.5178861299825648,
      "flos": 18972229877760.0,
      "grad_norm": 2.647627625095382,
      "learning_rate": 1.9813049527256923e-06,
      "loss": 1.0247,
      "num_input_tokens_seen": 91820380,
      "step": 4307
    },
    {
      "epoch": 0.5180063728732038,
      "flos": 12705889320960.0,
      "grad_norm": 4.240607710418651,
      "learning_rate": 1.9805260165486252e-06,
      "loss": 1.0107,
      "num_input_tokens_seen": 91839470,
      "step": 4308
    },
    {
      "epoch": 0.518126615763843,
      "flos": 13833522647040.0,
      "grad_norm": 5.620066365294754,
      "learning_rate": 1.9797470833257457e-06,
      "loss": 1.0607,
      "num_input_tokens_seen": 91858890,
      "step": 4309
    },
    {
      "epoch": 0.5182468586544821,
      "flos": 14698116403200.0,
      "grad_norm": 3.045931782590401,
      "learning_rate": 1.9789681531752177e-06,
      "loss": 0.97,
      "num_input_tokens_seen": 91878830,
      "step": 4310
    },
    {
      "epoch": 0.5183671015451211,
      "flos": 16428714209280.0,
      "grad_norm": 3.503983818905192,
      "learning_rate": 1.978189226215204e-06,
      "loss": 0.923,
      "num_input_tokens_seen": 91899095,
      "step": 4311
    },
    {
      "epoch": 0.5184873444357603,
      "flos": 12464410828800.0,
      "grad_norm": 3.780509939361531,
      "learning_rate": 1.9774103025638675e-06,
      "loss": 0.962,
      "num_input_tokens_seen": 91916940,
      "step": 4312
    },
    {
      "epoch": 0.5186075873263993,
      "flos": 17740141056000.0,
      "grad_norm": 2.614307059678093,
      "learning_rate": 1.9766313823393696e-06,
      "loss": 0.9595,
      "num_input_tokens_seen": 91937525,
      "step": 4313
    },
    {
      "epoch": 0.5187278302170384,
      "flos": 10734702120960.0,
      "grad_norm": 3.7384470680635418,
      "learning_rate": 1.975852465659873e-06,
      "loss": 0.882,
      "num_input_tokens_seen": 91953225,
      "step": 4314
    },
    {
      "epoch": 0.5188480731076776,
      "flos": 17792276828160.0,
      "grad_norm": 3.7564302626531307,
      "learning_rate": 1.9750735526435377e-06,
      "loss": 0.894,
      "num_input_tokens_seen": 91969890,
      "step": 4315
    },
    {
      "epoch": 0.5189683159983166,
      "flos": 17634643169280.0,
      "grad_norm": 13.518489467951557,
      "learning_rate": 1.974294643408525e-06,
      "loss": 0.9811,
      "num_input_tokens_seen": 91987405,
      "step": 4316
    },
    {
      "epoch": 0.5190885588889557,
      "flos": 17608943185920.0,
      "grad_norm": 3.3472452879586267,
      "learning_rate": 1.9735157380729947e-06,
      "loss": 0.8584,
      "num_input_tokens_seen": 92007535,
      "step": 4317
    },
    {
      "epoch": 0.5192088017795948,
      "flos": 17578337832960.0,
      "grad_norm": 3.1482068301530575,
      "learning_rate": 1.9727368367551053e-06,
      "loss": 1.0294,
      "num_input_tokens_seen": 92025805,
      "step": 4318
    },
    {
      "epoch": 0.5193290446702339,
      "flos": 19387139297280.0,
      "grad_norm": 7.358088872988215,
      "learning_rate": 1.9719579395730164e-06,
      "loss": 0.8915,
      "num_input_tokens_seen": 92044900,
      "step": 4319
    },
    {
      "epoch": 0.5194492875608729,
      "flos": 8064188805120.0,
      "grad_norm": 3.467276595330316,
      "learning_rate": 1.9711790466448854e-06,
      "loss": 1.1298,
      "num_input_tokens_seen": 92058640,
      "step": 4320
    },
    {
      "epoch": 0.5195695304515121,
      "flos": 14435628687360.0,
      "grad_norm": 8.256254657336877,
      "learning_rate": 1.9704001580888704e-06,
      "loss": 0.9016,
      "num_input_tokens_seen": 92077100,
      "step": 4321
    },
    {
      "epoch": 0.5196897733421512,
      "flos": 14226962964480.0,
      "grad_norm": 4.85055359502631,
      "learning_rate": 1.9696212740231283e-06,
      "loss": 1.0663,
      "num_input_tokens_seen": 92095470,
      "step": 4322
    },
    {
      "epoch": 0.5198100162327902,
      "flos": 16927069900800.0,
      "grad_norm": 5.581350211581976,
      "learning_rate": 1.9688423945658146e-06,
      "loss": 1.0118,
      "num_input_tokens_seen": 92116055,
      "step": 4323
    },
    {
      "epoch": 0.5199302591234293,
      "flos": 17083998412800.0,
      "grad_norm": 4.7842868310938185,
      "learning_rate": 1.9680635198350845e-06,
      "loss": 0.9139,
      "num_input_tokens_seen": 92135485,
      "step": 4324
    },
    {
      "epoch": 0.5200505020140684,
      "flos": 18762460446720.0,
      "grad_norm": 4.25884045432332,
      "learning_rate": 1.967284649949093e-06,
      "loss": 0.9203,
      "num_input_tokens_seen": 92154415,
      "step": 4325
    },
    {
      "epoch": 0.5201707449047075,
      "flos": 28281392209920.0,
      "grad_norm": 3.1112740243158146,
      "learning_rate": 1.966505785025994e-06,
      "loss": 0.9221,
      "num_input_tokens_seen": 92176040,
      "step": 4326
    },
    {
      "epoch": 0.5202909877953465,
      "flos": 38398689669120.0,
      "grad_norm": 4.221023286170957,
      "learning_rate": 1.965726925183941e-06,
      "loss": 0.9646,
      "num_input_tokens_seen": 92198865,
      "step": 4327
    },
    {
      "epoch": 0.5204112306859857,
      "flos": 13859437240320.0,
      "grad_norm": 3.2412138798382246,
      "learning_rate": 1.964948070541087e-06,
      "loss": 1.035,
      "num_input_tokens_seen": 92217245,
      "step": 4328
    },
    {
      "epoch": 0.5205314735766248,
      "flos": 10817749831680.0,
      "grad_norm": 4.325517109733998,
      "learning_rate": 1.9641692212155816e-06,
      "loss": 0.8823,
      "num_input_tokens_seen": 92234730,
      "step": 4329
    },
    {
      "epoch": 0.5206517164672638,
      "flos": 42409395671040.0,
      "grad_norm": 3.248654013460024,
      "learning_rate": 1.9633903773255777e-06,
      "loss": 0.9259,
      "num_input_tokens_seen": 92256765,
      "step": 4330
    },
    {
      "epoch": 0.520771959357903,
      "flos": 19130384732160.0,
      "grad_norm": 2.9109713033862397,
      "learning_rate": 1.9626115389892237e-06,
      "loss": 0.9423,
      "num_input_tokens_seen": 92277275,
      "step": 4331
    },
    {
      "epoch": 0.520892202248542,
      "flos": 19156146032640.0,
      "grad_norm": 3.8055123251681677,
      "learning_rate": 1.96183270632467e-06,
      "loss": 1.0372,
      "num_input_tokens_seen": 92296845,
      "step": 4332
    },
    {
      "epoch": 0.5210124451391811,
      "flos": 18396099747840.0,
      "grad_norm": 3.0416438654911557,
      "learning_rate": 1.9610538794500644e-06,
      "loss": 0.9783,
      "num_input_tokens_seen": 92316115,
      "step": 4333
    },
    {
      "epoch": 0.5211326880298203,
      "flos": 50530464215040.0,
      "grad_norm": 0.8189589459489659,
      "learning_rate": 1.9602750584835542e-06,
      "loss": 0.8257,
      "num_input_tokens_seen": 92381770,
      "step": 4334
    },
    {
      "epoch": 0.5212529309204593,
      "flos": 11050184048640.0,
      "grad_norm": 4.895075031831559,
      "learning_rate": 1.959496243543286e-06,
      "loss": 1.0258,
      "num_input_tokens_seen": 92399370,
      "step": 4335
    },
    {
      "epoch": 0.5213731738110984,
      "flos": 18677603880960.0,
      "grad_norm": 3.202981340861927,
      "learning_rate": 1.9587174347474057e-06,
      "loss": 0.9878,
      "num_input_tokens_seen": 92415600,
      "step": 4336
    },
    {
      "epoch": 0.5214934167017375,
      "flos": 13772955770880.0,
      "grad_norm": 4.829837828573273,
      "learning_rate": 1.9579386322140574e-06,
      "loss": 0.991,
      "num_input_tokens_seen": 92431000,
      "step": 4337
    },
    {
      "epoch": 0.5216136595923766,
      "flos": 22065715814400.0,
      "grad_norm": 4.018183624082008,
      "learning_rate": 1.9571598360613854e-06,
      "loss": 1.0014,
      "num_input_tokens_seen": 92453595,
      "step": 4338
    },
    {
      "epoch": 0.5217339024830157,
      "flos": 15590035046400.0,
      "grad_norm": 3.1720352508202505,
      "learning_rate": 1.956381046407532e-06,
      "loss": 0.895,
      "num_input_tokens_seen": 92473610,
      "step": 4339
    },
    {
      "epoch": 0.5218541453736548,
      "flos": 14855627427840.0,
      "grad_norm": 3.6461591447843085,
      "learning_rate": 1.9556022633706394e-06,
      "loss": 1.0506,
      "num_input_tokens_seen": 92492120,
      "step": 4340
    },
    {
      "epoch": 0.5219743882642939,
      "flos": 17032138567680.0,
      "grad_norm": 3.1215211959918463,
      "learning_rate": 1.954823487068848e-06,
      "loss": 0.9912,
      "num_input_tokens_seen": 92512050,
      "step": 4341
    },
    {
      "epoch": 0.5220946311549329,
      "flos": 20515355136000.0,
      "grad_norm": 2.7120719536770013,
      "learning_rate": 1.9540447176202976e-06,
      "loss": 1.0027,
      "num_input_tokens_seen": 92533015,
      "step": 4342
    },
    {
      "epoch": 0.5222148740455721,
      "flos": 48111462850560.0,
      "grad_norm": 0.8457930166911923,
      "learning_rate": 1.9532659551431272e-06,
      "loss": 0.8448,
      "num_input_tokens_seen": 92599765,
      "step": 4343
    },
    {
      "epoch": 0.5223351169362112,
      "flos": 44272816250880.0,
      "grad_norm": 3.5631294363677077,
      "learning_rate": 1.9524871997554744e-06,
      "loss": 0.8637,
      "num_input_tokens_seen": 92627245,
      "step": 4344
    },
    {
      "epoch": 0.5224553598268502,
      "flos": 10345278074880.0,
      "grad_norm": 5.90379904295941,
      "learning_rate": 1.951708451575475e-06,
      "loss": 0.9776,
      "num_input_tokens_seen": 92644030,
      "step": 4345
    },
    {
      "epoch": 0.5225756027174894,
      "flos": 10473164820480.0,
      "grad_norm": 3.1445981225905637,
      "learning_rate": 1.9509297107212657e-06,
      "loss": 1.0073,
      "num_input_tokens_seen": 92660520,
      "step": 4346
    },
    {
      "epoch": 0.5226958456081284,
      "flos": 16715920834560.0,
      "grad_norm": 2.4765171564325446,
      "learning_rate": 1.95015097731098e-06,
      "loss": 0.9848,
      "num_input_tokens_seen": 92679730,
      "step": 4347
    },
    {
      "epoch": 0.5228160884987675,
      "flos": 13518745866240.0,
      "grad_norm": 3.7335371410279365,
      "learning_rate": 1.949372251462751e-06,
      "loss": 1.0158,
      "num_input_tokens_seen": 92696865,
      "step": 4348
    },
    {
      "epoch": 0.5229363313894067,
      "flos": 14955453480960.0,
      "grad_norm": 2.960354307208268,
      "learning_rate": 1.9485935332947124e-06,
      "loss": 1.0227,
      "num_input_tokens_seen": 92714495,
      "step": 4349
    },
    {
      "epoch": 0.5230565742800457,
      "flos": 10476874506240.0,
      "grad_norm": 5.084481073268326,
      "learning_rate": 1.947814822924993e-06,
      "loss": 1.0209,
      "num_input_tokens_seen": 92731725,
      "step": 4350
    },
    {
      "epoch": 0.5231768171706848,
      "flos": 18369939886080.0,
      "grad_norm": 4.551801077721594,
      "learning_rate": 1.9470361204717236e-06,
      "loss": 1.014,
      "num_input_tokens_seen": 92750585,
      "step": 4351
    },
    {
      "epoch": 0.5232970600613239,
      "flos": 16163559198720.0,
      "grad_norm": 2.553321403130265,
      "learning_rate": 1.9462574260530326e-06,
      "loss": 1.0046,
      "num_input_tokens_seen": 92770585,
      "step": 4352
    },
    {
      "epoch": 0.523417302951963,
      "flos": 12259332157440.0,
      "grad_norm": 2.9542037366111407,
      "learning_rate": 1.9454787397870472e-06,
      "loss": 1.0038,
      "num_input_tokens_seen": 92787625,
      "step": 4353
    },
    {
      "epoch": 0.523537545842602,
      "flos": 13151312117760.0,
      "grad_norm": 5.134875575325293,
      "learning_rate": 1.944700061791894e-06,
      "loss": 0.9159,
      "num_input_tokens_seen": 92805740,
      "step": 4354
    },
    {
      "epoch": 0.5236577887332411,
      "flos": 13990389841920.0,
      "grad_norm": 6.590047800356711,
      "learning_rate": 1.943921392185698e-06,
      "loss": 0.8557,
      "num_input_tokens_seen": 92824085,
      "step": 4355
    },
    {
      "epoch": 0.5237780316238803,
      "flos": 16743613624320.0,
      "grad_norm": 8.005612788575617,
      "learning_rate": 1.9431427310865814e-06,
      "loss": 0.9749,
      "num_input_tokens_seen": 92843410,
      "step": 4356
    },
    {
      "epoch": 0.5238982745145193,
      "flos": 15982984826880.0,
      "grad_norm": 3.5863499622367785,
      "learning_rate": 1.942364078612667e-06,
      "loss": 0.9901,
      "num_input_tokens_seen": 92861860,
      "step": 4357
    },
    {
      "epoch": 0.5240185174051584,
      "flos": 19286761390080.0,
      "grad_norm": 2.9084326638763245,
      "learning_rate": 1.9415854348820765e-06,
      "loss": 0.9417,
      "num_input_tokens_seen": 92881430,
      "step": 4358
    },
    {
      "epoch": 0.5241387602957975,
      "flos": 16114397306880.0,
      "grad_norm": 9.16550183842627,
      "learning_rate": 1.940806800012929e-06,
      "loss": 0.8677,
      "num_input_tokens_seen": 92901220,
      "step": 4359
    },
    {
      "epoch": 0.5242590031864366,
      "flos": 28963633397760.0,
      "grad_norm": 3.3896063755583166,
      "learning_rate": 1.9400281741233432e-06,
      "loss": 0.8346,
      "num_input_tokens_seen": 92925830,
      "step": 4360
    },
    {
      "epoch": 0.5243792460770756,
      "flos": 47742434856960.0,
      "grad_norm": 0.7434967940337704,
      "learning_rate": 1.939249557331435e-06,
      "loss": 0.7503,
      "num_input_tokens_seen": 92991365,
      "step": 4361
    },
    {
      "epoch": 0.5244994889677148,
      "flos": 20073550049280.0,
      "grad_norm": 6.147217606507664,
      "learning_rate": 1.938470949755321e-06,
      "loss": 0.9215,
      "num_input_tokens_seen": 93010965,
      "step": 4362
    },
    {
      "epoch": 0.5246197318583539,
      "flos": 47221230428160.0,
      "grad_norm": 0.8828307968539149,
      "learning_rate": 1.937692351513115e-06,
      "loss": 0.815,
      "num_input_tokens_seen": 93069680,
      "step": 4363
    },
    {
      "epoch": 0.5247399747489929,
      "flos": 14934996111360.0,
      "grad_norm": 3.3785532860713814,
      "learning_rate": 1.9369137627229297e-06,
      "loss": 0.993,
      "num_input_tokens_seen": 93087800,
      "step": 4364
    },
    {
      "epoch": 0.5248602176396321,
      "flos": 13492095467520.0,
      "grad_norm": 3.497623317854024,
      "learning_rate": 1.936135183502877e-06,
      "loss": 1.0777,
      "num_input_tokens_seen": 93104820,
      "step": 4365
    },
    {
      "epoch": 0.5249804605302711,
      "flos": 15773276712960.0,
      "grad_norm": 4.947415981726242,
      "learning_rate": 1.935356613971066e-06,
      "loss": 0.9964,
      "num_input_tokens_seen": 93125200,
      "step": 4366
    },
    {
      "epoch": 0.5251007034209102,
      "flos": 16927223193600.0,
      "grad_norm": 3.0099890030061194,
      "learning_rate": 1.9345780542456047e-06,
      "loss": 0.9644,
      "num_input_tokens_seen": 93144295,
      "step": 4367
    },
    {
      "epoch": 0.5252209463115494,
      "flos": 16559881420800.0,
      "grad_norm": 3.342697690996533,
      "learning_rate": 1.9337995044446007e-06,
      "loss": 0.9093,
      "num_input_tokens_seen": 93162855,
      "step": 4368
    },
    {
      "epoch": 0.5253411892021884,
      "flos": 14069022720000.0,
      "grad_norm": 3.473751826441979,
      "learning_rate": 1.9330209646861596e-06,
      "loss": 0.9947,
      "num_input_tokens_seen": 93181725,
      "step": 4369
    },
    {
      "epoch": 0.5254614320928275,
      "flos": 17162937876480.0,
      "grad_norm": 4.399432596555021,
      "learning_rate": 1.9322424350883843e-06,
      "loss": 0.9713,
      "num_input_tokens_seen": 93203280,
      "step": 4370
    },
    {
      "epoch": 0.5255816749834666,
      "flos": 17530402283520.0,
      "grad_norm": 3.2470626762128254,
      "learning_rate": 1.931463915769379e-06,
      "loss": 0.9845,
      "num_input_tokens_seen": 93223115,
      "step": 4371
    },
    {
      "epoch": 0.5257019178741057,
      "flos": 9978181570560.0,
      "grad_norm": 5.792160708352861,
      "learning_rate": 1.930685406847242e-06,
      "loss": 0.9359,
      "num_input_tokens_seen": 93237410,
      "step": 4372
    },
    {
      "epoch": 0.5258221607647448,
      "flos": 16742540574720.0,
      "grad_norm": 2.8951179846668773,
      "learning_rate": 1.9299069084400734e-06,
      "loss": 1.0228,
      "num_input_tokens_seen": 93257990,
      "step": 4373
    },
    {
      "epoch": 0.5259424036553839,
      "flos": 17760230522880.0,
      "grad_norm": 4.120086054968002,
      "learning_rate": 1.9291284206659717e-06,
      "loss": 0.8847,
      "num_input_tokens_seen": 93275895,
      "step": 4374
    },
    {
      "epoch": 0.526062646546023,
      "flos": 20490084372480.0,
      "grad_norm": 3.8260821438559316,
      "learning_rate": 1.928349943643032e-06,
      "loss": 0.9065,
      "num_input_tokens_seen": 93294715,
      "step": 4375
    },
    {
      "epoch": 0.526182889436662,
      "flos": 16219925852160.0,
      "grad_norm": 3.820361552479372,
      "learning_rate": 1.9275714774893493e-06,
      "loss": 1.0249,
      "num_input_tokens_seen": 93313890,
      "step": 4376
    },
    {
      "epoch": 0.5263031323273012,
      "flos": 16297424363520.0,
      "grad_norm": 14.591512831235203,
      "learning_rate": 1.9267930223230154e-06,
      "loss": 0.9115,
      "num_input_tokens_seen": 93332085,
      "step": 4377
    },
    {
      "epoch": 0.5264233752179402,
      "flos": 12574078279680.0,
      "grad_norm": 5.4601688744555545,
      "learning_rate": 1.9260145782621224e-06,
      "loss": 0.9721,
      "num_input_tokens_seen": 93349585,
      "step": 4378
    },
    {
      "epoch": 0.5265436181085793,
      "flos": 17369304207360.0,
      "grad_norm": 4.276065778335199,
      "learning_rate": 1.925236145424758e-06,
      "loss": 1.0713,
      "num_input_tokens_seen": 93368125,
      "step": 4379
    },
    {
      "epoch": 0.5266638609992185,
      "flos": 49563223818240.0,
      "grad_norm": 0.7111937511708091,
      "learning_rate": 1.924457723929012e-06,
      "loss": 0.7932,
      "num_input_tokens_seen": 93438655,
      "step": 4380
    },
    {
      "epoch": 0.5267841038898575,
      "flos": 14722743336960.0,
      "grad_norm": 4.6851866036202,
      "learning_rate": 1.9236793138929685e-06,
      "loss": 1.0167,
      "num_input_tokens_seen": 93457645,
      "step": 4381
    },
    {
      "epoch": 0.5269043467804966,
      "flos": 12204038553600.0,
      "grad_norm": 6.338029057617603,
      "learning_rate": 1.9229009154347133e-06,
      "loss": 0.9979,
      "num_input_tokens_seen": 93474955,
      "step": 4382
    },
    {
      "epoch": 0.5270245896711357,
      "flos": 12915137556480.0,
      "grad_norm": 3.978027433650806,
      "learning_rate": 1.922122528672327e-06,
      "loss": 0.9963,
      "num_input_tokens_seen": 93493340,
      "step": 4383
    },
    {
      "epoch": 0.5271448325617748,
      "flos": 15117409996800.0,
      "grad_norm": 4.363204750476429,
      "learning_rate": 1.9213441537238914e-06,
      "loss": 0.9747,
      "num_input_tokens_seen": 93509935,
      "step": 4384
    },
    {
      "epoch": 0.5272650754524139,
      "flos": 46892013465600.0,
      "grad_norm": 1.026446803060263,
      "learning_rate": 1.920565790707485e-06,
      "loss": 0.8479,
      "num_input_tokens_seen": 93575045,
      "step": 4385
    },
    {
      "epoch": 0.527385318343053,
      "flos": 13958742097920.0,
      "grad_norm": 5.997758446642991,
      "learning_rate": 1.9197874397411853e-06,
      "loss": 0.8508,
      "num_input_tokens_seen": 93591395,
      "step": 4386
    },
    {
      "epoch": 0.5275055612336921,
      "flos": 8954145300480.0,
      "grad_norm": 12.741849800092803,
      "learning_rate": 1.919009100943067e-06,
      "loss": 0.8211,
      "num_input_tokens_seen": 93606805,
      "step": 4387
    },
    {
      "epoch": 0.5276258041243311,
      "flos": 12574415523840.0,
      "grad_norm": 5.606337891296921,
      "learning_rate": 1.9182307744312043e-06,
      "loss": 0.8504,
      "num_input_tokens_seen": 93623630,
      "step": 4388
    },
    {
      "epoch": 0.5277460470149702,
      "flos": 16140005314560.0,
      "grad_norm": 2.9340850784354497,
      "learning_rate": 1.9174524603236676e-06,
      "loss": 0.9504,
      "num_input_tokens_seen": 93642300,
      "step": 4389
    },
    {
      "epoch": 0.5278662899056094,
      "flos": 14122354176000.0,
      "grad_norm": 3.50018718763879,
      "learning_rate": 1.916674158738527e-06,
      "loss": 0.9587,
      "num_input_tokens_seen": 93660925,
      "step": 4390
    },
    {
      "epoch": 0.5279865327962484,
      "flos": 12758638264320.0,
      "grad_norm": 6.02693203541582,
      "learning_rate": 1.9158958697938506e-06,
      "loss": 0.7866,
      "num_input_tokens_seen": 93679025,
      "step": 4391
    },
    {
      "epoch": 0.5281067756868875,
      "flos": 11263080652800.0,
      "grad_norm": 7.401623532327664,
      "learning_rate": 1.9151175936077032e-06,
      "loss": 1.0338,
      "num_input_tokens_seen": 93693715,
      "step": 4392
    },
    {
      "epoch": 0.5282270185775266,
      "flos": 13779853946880.0,
      "grad_norm": 3.571112278147693,
      "learning_rate": 1.9143393302981507e-06,
      "loss": 0.9909,
      "num_input_tokens_seen": 93711120,
      "step": 4393
    },
    {
      "epoch": 0.5283472614681657,
      "flos": 11604691783680.0,
      "grad_norm": 2.986954777337265,
      "learning_rate": 1.913561079983252e-06,
      "loss": 1.0225,
      "num_input_tokens_seen": 93729665,
      "step": 4394
    },
    {
      "epoch": 0.5284675043588047,
      "flos": 19050740121600.0,
      "grad_norm": 6.173673961516389,
      "learning_rate": 1.9127828427810693e-06,
      "loss": 0.9397,
      "num_input_tokens_seen": 93749950,
      "step": 4395
    },
    {
      "epoch": 0.5285877472494439,
      "flos": 14119962808320.0,
      "grad_norm": 4.516043990077211,
      "learning_rate": 1.9120046188096607e-06,
      "loss": 1.0018,
      "num_input_tokens_seen": 93767715,
      "step": 4396
    },
    {
      "epoch": 0.528707990140083,
      "flos": 14200251248640.0,
      "grad_norm": 4.025967234145126,
      "learning_rate": 1.9112264081870804e-06,
      "loss": 0.9445,
      "num_input_tokens_seen": 93785825,
      "step": 4397
    },
    {
      "epoch": 0.528828233030722,
      "flos": 14672201809920.0,
      "grad_norm": 5.245037094581425,
      "learning_rate": 1.9104482110313843e-06,
      "loss": 0.9519,
      "num_input_tokens_seen": 93805135,
      "step": 4398
    },
    {
      "epoch": 0.5289484759213612,
      "flos": 17923903918080.0,
      "grad_norm": 4.138839855088541,
      "learning_rate": 1.909670027460623e-06,
      "loss": 0.9433,
      "num_input_tokens_seen": 93822155,
      "step": 4399
    },
    {
      "epoch": 0.5290687188120002,
      "flos": 22197710807040.0,
      "grad_norm": 6.7644490195676745,
      "learning_rate": 1.908891857592847e-06,
      "loss": 0.9093,
      "num_input_tokens_seen": 93842945,
      "step": 4400
    },
    {
      "epoch": 0.5291889617026393,
      "flos": 14278178979840.0,
      "grad_norm": 3.6168498794160553,
      "learning_rate": 1.9081137015461034e-06,
      "loss": 1.0871,
      "num_input_tokens_seen": 93858740,
      "step": 4401
    },
    {
      "epoch": 0.5293092045932785,
      "flos": 13935770726400.0,
      "grad_norm": 3.900611286674052,
      "learning_rate": 1.9073355594384383e-06,
      "loss": 1.0906,
      "num_input_tokens_seen": 93876700,
      "step": 4402
    },
    {
      "epoch": 0.5294294474839175,
      "flos": 17295392747520.0,
      "grad_norm": 4.030264563951644,
      "learning_rate": 1.906557431387895e-06,
      "loss": 1.0027,
      "num_input_tokens_seen": 93895410,
      "step": 4403
    },
    {
      "epoch": 0.5295496903745566,
      "flos": 13383746334720.0,
      "grad_norm": 4.802346355436905,
      "learning_rate": 1.905779317512516e-06,
      "loss": 0.9918,
      "num_input_tokens_seen": 93912675,
      "step": 4404
    },
    {
      "epoch": 0.5296699332651957,
      "flos": 14855535452160.0,
      "grad_norm": 3.5768928905188284,
      "learning_rate": 1.9050012179303385e-06,
      "loss": 0.9879,
      "num_input_tokens_seen": 93930905,
      "step": 4405
    },
    {
      "epoch": 0.5297901761558348,
      "flos": 15662076334080.0,
      "grad_norm": 4.028973550254747,
      "learning_rate": 1.904223132759401e-06,
      "loss": 0.8729,
      "num_input_tokens_seen": 93949225,
      "step": 4406
    },
    {
      "epoch": 0.5299104190464738,
      "flos": 15484629135360.0,
      "grad_norm": 5.738719323603277,
      "learning_rate": 1.9034450621177383e-06,
      "loss": 0.875,
      "num_input_tokens_seen": 93967265,
      "step": 4407
    },
    {
      "epoch": 0.530030661937113,
      "flos": 10398088335360.0,
      "grad_norm": 3.5226976924784186,
      "learning_rate": 1.9026670061233824e-06,
      "loss": 0.8982,
      "num_input_tokens_seen": 93984420,
      "step": 4408
    },
    {
      "epoch": 0.5301509048277521,
      "flos": 15091556720640.0,
      "grad_norm": 10.06734070531802,
      "learning_rate": 1.901888964894365e-06,
      "loss": 0.9997,
      "num_input_tokens_seen": 94003180,
      "step": 4409
    },
    {
      "epoch": 0.5302711477183911,
      "flos": 18473199697920.0,
      "grad_norm": 4.591772575103184,
      "learning_rate": 1.9011109385487134e-06,
      "loss": 0.868,
      "num_input_tokens_seen": 94024150,
      "step": 4410
    },
    {
      "epoch": 0.5303913906090303,
      "flos": 15825780387840.0,
      "grad_norm": 5.781492815846743,
      "learning_rate": 1.900332927204454e-06,
      "loss": 0.8414,
      "num_input_tokens_seen": 94042320,
      "step": 4411
    },
    {
      "epoch": 0.5305116334996693,
      "flos": 17739466567680.0,
      "grad_norm": 3.4745281310765894,
      "learning_rate": 1.8995549309796097e-06,
      "loss": 0.956,
      "num_input_tokens_seen": 94061345,
      "step": 4412
    },
    {
      "epoch": 0.5306318763903084,
      "flos": 14327647457280.0,
      "grad_norm": 4.209160877607211,
      "learning_rate": 1.8987769499922028e-06,
      "loss": 0.9602,
      "num_input_tokens_seen": 94080035,
      "step": 4413
    },
    {
      "epoch": 0.5307521192809476,
      "flos": 14383554232320.0,
      "grad_norm": 5.69754984881968,
      "learning_rate": 1.897998984360252e-06,
      "loss": 0.8986,
      "num_input_tokens_seen": 94098725,
      "step": 4414
    },
    {
      "epoch": 0.5308723621715866,
      "flos": 20548842393600.0,
      "grad_norm": 3.5069248631164647,
      "learning_rate": 1.897221034201775e-06,
      "loss": 0.9793,
      "num_input_tokens_seen": 94122185,
      "step": 4415
    },
    {
      "epoch": 0.5309926050622257,
      "flos": 19552376279040.0,
      "grad_norm": 4.460747656315061,
      "learning_rate": 1.8964430996347842e-06,
      "loss": 0.8634,
      "num_input_tokens_seen": 94143455,
      "step": 4416
    },
    {
      "epoch": 0.5311128479528648,
      "flos": 14560940113920.0,
      "grad_norm": 3.598617839393884,
      "learning_rate": 1.8956651807772931e-06,
      "loss": 1.0254,
      "num_input_tokens_seen": 94161210,
      "step": 4417
    },
    {
      "epoch": 0.5312330908435039,
      "flos": 15196625387520.0,
      "grad_norm": 2.985174740991376,
      "learning_rate": 1.8948872777473115e-06,
      "loss": 1.0244,
      "num_input_tokens_seen": 94178885,
      "step": 4418
    },
    {
      "epoch": 0.531353333734143,
      "flos": 17582354104320.0,
      "grad_norm": 4.915364152179893,
      "learning_rate": 1.8941093906628458e-06,
      "loss": 0.8376,
      "num_input_tokens_seen": 94196390,
      "step": 4419
    },
    {
      "epoch": 0.531473576624782,
      "flos": 21724166000640.0,
      "grad_norm": 2.909744294982399,
      "learning_rate": 1.893331519641902e-06,
      "loss": 0.9126,
      "num_input_tokens_seen": 94218255,
      "step": 4420
    },
    {
      "epoch": 0.5315938195154212,
      "flos": 16350326599680.0,
      "grad_norm": 6.192731706999186,
      "learning_rate": 1.8925536648024815e-06,
      "loss": 0.9446,
      "num_input_tokens_seen": 94235395,
      "step": 4421
    },
    {
      "epoch": 0.5317140624060602,
      "flos": 16167636787200.0,
      "grad_norm": 6.711962869134529,
      "learning_rate": 1.8917758262625849e-06,
      "loss": 0.9582,
      "num_input_tokens_seen": 94255355,
      "step": 4422
    },
    {
      "epoch": 0.5318343052966993,
      "flos": 16219680583680.0,
      "grad_norm": 3.0032413773707822,
      "learning_rate": 1.8909980041402089e-06,
      "loss": 1.004,
      "num_input_tokens_seen": 94273670,
      "step": 4423
    },
    {
      "epoch": 0.5319545481873384,
      "flos": 9611330334720.0,
      "grad_norm": 5.534133762850963,
      "learning_rate": 1.8902201985533494e-06,
      "loss": 0.8522,
      "num_input_tokens_seen": 94290655,
      "step": 4424
    },
    {
      "epoch": 0.5320747910779775,
      "flos": 15746135777280.0,
      "grad_norm": 3.5973214357130807,
      "learning_rate": 1.8894424096199983e-06,
      "loss": 0.9471,
      "num_input_tokens_seen": 94309580,
      "step": 4425
    },
    {
      "epoch": 0.5321950339686166,
      "flos": 13177410662400.0,
      "grad_norm": 4.092231061018268,
      "learning_rate": 1.8886646374581463e-06,
      "loss": 1.0578,
      "num_input_tokens_seen": 94328525,
      "step": 4426
    },
    {
      "epoch": 0.5323152768592557,
      "flos": 16245257932800.0,
      "grad_norm": 3.814883253278535,
      "learning_rate": 1.8878868821857795e-06,
      "loss": 0.9005,
      "num_input_tokens_seen": 94347895,
      "step": 4427
    },
    {
      "epoch": 0.5324355197498948,
      "flos": 24216802897920.0,
      "grad_norm": 30.419400138483276,
      "learning_rate": 1.8871091439208838e-06,
      "loss": 0.9348,
      "num_input_tokens_seen": 94369225,
      "step": 4428
    },
    {
      "epoch": 0.5325557626405338,
      "flos": 16532464558080.0,
      "grad_norm": 5.024694116496592,
      "learning_rate": 1.8863314227814414e-06,
      "loss": 0.9591,
      "num_input_tokens_seen": 94387255,
      "step": 4429
    },
    {
      "epoch": 0.532676005531173,
      "flos": 19019950817280.0,
      "grad_norm": 4.223331930214986,
      "learning_rate": 1.8855537188854313e-06,
      "loss": 0.6611,
      "num_input_tokens_seen": 94405950,
      "step": 4430
    },
    {
      "epoch": 0.5327962484218121,
      "flos": 12678993653760.0,
      "grad_norm": 4.111421604860664,
      "learning_rate": 1.8847760323508315e-06,
      "loss": 0.9739,
      "num_input_tokens_seen": 94424575,
      "step": 4431
    },
    {
      "epoch": 0.5329164913124511,
      "flos": 12700340121600.0,
      "grad_norm": 2.9102445653021336,
      "learning_rate": 1.883998363295616e-06,
      "loss": 0.9484,
      "num_input_tokens_seen": 94441775,
      "step": 4432
    },
    {
      "epoch": 0.5330367342030903,
      "flos": 40969736724480.0,
      "grad_norm": 1.02039180990661,
      "learning_rate": 1.8832207118377565e-06,
      "loss": 0.8806,
      "num_input_tokens_seen": 94496865,
      "step": 4433
    },
    {
      "epoch": 0.5331569770937293,
      "flos": 12338302279680.0,
      "grad_norm": 4.365108315783299,
      "learning_rate": 1.882443078095222e-06,
      "loss": 0.8846,
      "num_input_tokens_seen": 94515465,
      "step": 4434
    },
    {
      "epoch": 0.5332772199843684,
      "flos": 40608955883520.0,
      "grad_norm": 0.8718871878042143,
      "learning_rate": 1.8816654621859794e-06,
      "loss": 0.8818,
      "num_input_tokens_seen": 94574850,
      "step": 4435
    },
    {
      "epoch": 0.5333974628750076,
      "flos": 13256258150400.0,
      "grad_norm": 6.418214520457791,
      "learning_rate": 1.8808878642279915e-06,
      "loss": 0.9028,
      "num_input_tokens_seen": 94589975,
      "step": 4436
    },
    {
      "epoch": 0.5335177057656466,
      "flos": 16927069900800.0,
      "grad_norm": 9.255243825477693,
      "learning_rate": 1.8801102843392209e-06,
      "loss": 0.8451,
      "num_input_tokens_seen": 94609100,
      "step": 4437
    },
    {
      "epoch": 0.5336379486562857,
      "flos": 17841561354240.0,
      "grad_norm": 3.6345104556372143,
      "learning_rate": 1.8793327226376238e-06,
      "loss": 1.0548,
      "num_input_tokens_seen": 94628140,
      "step": 4438
    },
    {
      "epoch": 0.5337581915469248,
      "flos": 15195797606400.0,
      "grad_norm": 3.616437863145607,
      "learning_rate": 1.8785551792411569e-06,
      "loss": 0.9958,
      "num_input_tokens_seen": 94646870,
      "step": 4439
    },
    {
      "epoch": 0.5338784344375639,
      "flos": 10502267904000.0,
      "grad_norm": 4.522081451870215,
      "learning_rate": 1.8777776542677733e-06,
      "loss": 1.0165,
      "num_input_tokens_seen": 94664640,
      "step": 4440
    },
    {
      "epoch": 0.5339986773282029,
      "flos": 14776749281280.0,
      "grad_norm": 3.645934330585829,
      "learning_rate": 1.8770001478354216e-06,
      "loss": 0.9028,
      "num_input_tokens_seen": 94684035,
      "step": 4441
    },
    {
      "epoch": 0.5341189202188421,
      "flos": 12732723671040.0,
      "grad_norm": 5.999582594021874,
      "learning_rate": 1.8762226600620504e-06,
      "loss": 1.021,
      "num_input_tokens_seen": 94702370,
      "step": 4442
    },
    {
      "epoch": 0.5342391631094812,
      "flos": 7746407485440.0,
      "grad_norm": 4.994458561020306,
      "learning_rate": 1.8754451910656031e-06,
      "loss": 0.7845,
      "num_input_tokens_seen": 94715990,
      "step": 4443
    },
    {
      "epoch": 0.5343594060001202,
      "flos": 10842223472640.0,
      "grad_norm": 7.58712140164701,
      "learning_rate": 1.8746677409640212e-06,
      "loss": 1.0115,
      "num_input_tokens_seen": 94732810,
      "step": 4444
    },
    {
      "epoch": 0.5344796488907594,
      "flos": 19150872760320.0,
      "grad_norm": 3.6263704011555347,
      "learning_rate": 1.8738903098752432e-06,
      "loss": 1.0353,
      "num_input_tokens_seen": 94751660,
      "step": 4445
    },
    {
      "epoch": 0.5345998917813984,
      "flos": 18081108357120.0,
      "grad_norm": 4.96082561697163,
      "learning_rate": 1.8731128979172052e-06,
      "loss": 0.9207,
      "num_input_tokens_seen": 94770580,
      "step": 4446
    },
    {
      "epoch": 0.5347201346720375,
      "flos": 23429523701760.0,
      "grad_norm": 3.9945411388051735,
      "learning_rate": 1.8723355052078394e-06,
      "loss": 0.8489,
      "num_input_tokens_seen": 94790335,
      "step": 4447
    },
    {
      "epoch": 0.5348403775626767,
      "flos": 12731742597120.0,
      "grad_norm": 4.281799281170069,
      "learning_rate": 1.8715581318650765e-06,
      "loss": 0.9588,
      "num_input_tokens_seen": 94809110,
      "step": 4448
    },
    {
      "epoch": 0.5349606204533157,
      "flos": 12469745418240.0,
      "grad_norm": 6.90340755190049,
      "learning_rate": 1.8707807780068422e-06,
      "loss": 1.0008,
      "num_input_tokens_seen": 94826645,
      "step": 4449
    },
    {
      "epoch": 0.5350808633439548,
      "flos": 20780724756480.0,
      "grad_norm": 4.44302818992407,
      "learning_rate": 1.8700034437510611e-06,
      "loss": 0.8654,
      "num_input_tokens_seen": 94846460,
      "step": 4450
    },
    {
      "epoch": 0.5352011062345938,
      "flos": 13832756183040.0,
      "grad_norm": 4.704716105961689,
      "learning_rate": 1.8692261292156549e-06,
      "loss": 1.0023,
      "num_input_tokens_seen": 94865415,
      "step": 4451
    },
    {
      "epoch": 0.535321349125233,
      "flos": 16691508510720.0,
      "grad_norm": 4.057304934021686,
      "learning_rate": 1.8684488345185401e-06,
      "loss": 1.0095,
      "num_input_tokens_seen": 94885310,
      "step": 4452
    },
    {
      "epoch": 0.535441592015872,
      "flos": 14534718935040.0,
      "grad_norm": 5.005472430970142,
      "learning_rate": 1.8676715597776332e-06,
      "loss": 0.9711,
      "num_input_tokens_seen": 94903375,
      "step": 4453
    },
    {
      "epoch": 0.5355618349065111,
      "flos": 13885351833600.0,
      "grad_norm": 3.567134586431362,
      "learning_rate": 1.8668943051108455e-06,
      "loss": 0.9427,
      "num_input_tokens_seen": 94920400,
      "step": 4454
    },
    {
      "epoch": 0.5356820777971503,
      "flos": 17084580925440.0,
      "grad_norm": 3.8221548431825303,
      "learning_rate": 1.8661170706360856e-06,
      "loss": 0.9536,
      "num_input_tokens_seen": 94939285,
      "step": 4455
    },
    {
      "epoch": 0.5358023206877893,
      "flos": 14827658711040.0,
      "grad_norm": 3.426735826553864,
      "learning_rate": 1.8653398564712594e-06,
      "loss": 1.0144,
      "num_input_tokens_seen": 94957950,
      "step": 4456
    },
    {
      "epoch": 0.5359225635784284,
      "flos": 15930603786240.0,
      "grad_norm": 2.4458553211408733,
      "learning_rate": 1.8645626627342704e-06,
      "loss": 1.0157,
      "num_input_tokens_seen": 94978435,
      "step": 4457
    },
    {
      "epoch": 0.5360428064690675,
      "flos": 17137176576000.0,
      "grad_norm": 4.475928669474485,
      "learning_rate": 1.8637854895430172e-06,
      "loss": 0.9835,
      "num_input_tokens_seen": 94997420,
      "step": 4458
    },
    {
      "epoch": 0.5361630493597066,
      "flos": 15223153152000.0,
      "grad_norm": 11.25683216747844,
      "learning_rate": 1.8630083370153978e-06,
      "loss": 0.8761,
      "num_input_tokens_seen": 95016780,
      "step": 4459
    },
    {
      "epoch": 0.5362832922503457,
      "flos": 49330644418560.0,
      "grad_norm": 0.7850557708974197,
      "learning_rate": 1.8622312052693041e-06,
      "loss": 0.7808,
      "num_input_tokens_seen": 95077680,
      "step": 4460
    },
    {
      "epoch": 0.5364035351409848,
      "flos": 6856880209920.0,
      "grad_norm": 4.849127559256149,
      "learning_rate": 1.8614540944226267e-06,
      "loss": 0.9264,
      "num_input_tokens_seen": 95094070,
      "step": 4461
    },
    {
      "epoch": 0.5365237780316239,
      "flos": 16556018442240.0,
      "grad_norm": 4.791937557800741,
      "learning_rate": 1.8606770045932537e-06,
      "loss": 0.8744,
      "num_input_tokens_seen": 95112905,
      "step": 4462
    },
    {
      "epoch": 0.5366440209222629,
      "flos": 18919664885760.0,
      "grad_norm": 6.1734502785950385,
      "learning_rate": 1.859899935899068e-06,
      "loss": 1.0073,
      "num_input_tokens_seen": 95132480,
      "step": 4463
    },
    {
      "epoch": 0.5367642638129021,
      "flos": 13910224035840.0,
      "grad_norm": 2.8927020884139543,
      "learning_rate": 1.8591228884579506e-06,
      "loss": 0.9917,
      "num_input_tokens_seen": 95150695,
      "step": 4464
    },
    {
      "epoch": 0.5368845067035412,
      "flos": 17005948047360.0,
      "grad_norm": 3.57120240721536,
      "learning_rate": 1.8583458623877795e-06,
      "loss": 1.0164,
      "num_input_tokens_seen": 95169515,
      "step": 4465
    },
    {
      "epoch": 0.5370047495941802,
      "flos": 11944892620800.0,
      "grad_norm": 7.151912524121471,
      "learning_rate": 1.8575688578064281e-06,
      "loss": 0.9308,
      "num_input_tokens_seen": 95187360,
      "step": 4466
    },
    {
      "epoch": 0.5371249924848194,
      "flos": 14724644167680.0,
      "grad_norm": 3.579230658270369,
      "learning_rate": 1.8567918748317674e-06,
      "loss": 0.9556,
      "num_input_tokens_seen": 95206430,
      "step": 4467
    },
    {
      "epoch": 0.5372452353754584,
      "flos": 12731895889920.0,
      "grad_norm": 3.759819145317121,
      "learning_rate": 1.8560149135816659e-06,
      "loss": 1.0177,
      "num_input_tokens_seen": 95222985,
      "step": 4468
    },
    {
      "epoch": 0.5373654782660975,
      "flos": 10869640335360.0,
      "grad_norm": 5.677230450418426,
      "learning_rate": 1.8552379741739873e-06,
      "loss": 1.0284,
      "num_input_tokens_seen": 95240050,
      "step": 4469
    },
    {
      "epoch": 0.5374857211567367,
      "flos": 49410963517440.0,
      "grad_norm": 0.9062924868391273,
      "learning_rate": 1.8544610567265935e-06,
      "loss": 0.7741,
      "num_input_tokens_seen": 95293710,
      "step": 4470
    },
    {
      "epoch": 0.5376059640473757,
      "flos": 10660054855680.0,
      "grad_norm": 8.269410389704902,
      "learning_rate": 1.853684161357341e-06,
      "loss": 1.032,
      "num_input_tokens_seen": 95311090,
      "step": 4471
    },
    {
      "epoch": 0.5377262069380148,
      "flos": 14043138785280.0,
      "grad_norm": 5.0787131570450965,
      "learning_rate": 1.852907288184085e-06,
      "loss": 0.9731,
      "num_input_tokens_seen": 95329695,
      "step": 4472
    },
    {
      "epoch": 0.5378464498286539,
      "flos": 21380470087680.0,
      "grad_norm": 3.363543386135014,
      "learning_rate": 1.8521304373246762e-06,
      "loss": 0.8962,
      "num_input_tokens_seen": 95350460,
      "step": 4473
    },
    {
      "epoch": 0.537966692719293,
      "flos": 15091311452160.0,
      "grad_norm": 4.261336745668588,
      "learning_rate": 1.8513536088969626e-06,
      "loss": 1.081,
      "num_input_tokens_seen": 95367845,
      "step": 4474
    },
    {
      "epoch": 0.538086935609932,
      "flos": 15301142200320.0,
      "grad_norm": 3.4263358803541784,
      "learning_rate": 1.8505768030187884e-06,
      "loss": 0.9867,
      "num_input_tokens_seen": 95387695,
      "step": 4475
    },
    {
      "epoch": 0.5382071785005712,
      "flos": 16166471761920.0,
      "grad_norm": 4.543853280049349,
      "learning_rate": 1.849800019807995e-06,
      "loss": 1.002,
      "num_input_tokens_seen": 95408640,
      "step": 4476
    },
    {
      "epoch": 0.5383274213912103,
      "flos": 17738240225280.0,
      "grad_norm": 93.41275801182084,
      "learning_rate": 1.8490232593824186e-06,
      "loss": 0.9125,
      "num_input_tokens_seen": 95424815,
      "step": 4477
    },
    {
      "epoch": 0.5384476642818493,
      "flos": 15852492103680.0,
      "grad_norm": 3.5706003018831587,
      "learning_rate": 1.8482465218598935e-06,
      "loss": 1.0338,
      "num_input_tokens_seen": 95444480,
      "step": 4478
    },
    {
      "epoch": 0.5385679071724885,
      "flos": 16140495851520.0,
      "grad_norm": 3.106676934297591,
      "learning_rate": 1.8474698073582508e-06,
      "loss": 1.0389,
      "num_input_tokens_seen": 95465570,
      "step": 4479
    },
    {
      "epoch": 0.5386881500631275,
      "flos": 11283170119680.0,
      "grad_norm": 4.5334292728893475,
      "learning_rate": 1.8466931159953166e-06,
      "loss": 1.0578,
      "num_input_tokens_seen": 95481925,
      "step": 4480
    },
    {
      "epoch": 0.5388083929537666,
      "flos": 17109882347520.0,
      "grad_norm": 3.5244604685310743,
      "learning_rate": 1.8459164478889158e-06,
      "loss": 1.0342,
      "num_input_tokens_seen": 95503040,
      "step": 4481
    },
    {
      "epoch": 0.5389286358444056,
      "flos": 15799773818880.0,
      "grad_norm": 4.619295881375999,
      "learning_rate": 1.8451398031568663e-06,
      "loss": 0.9515,
      "num_input_tokens_seen": 95522385,
      "step": 4482
    },
    {
      "epoch": 0.5390488787350448,
      "flos": 17269570129920.0,
      "grad_norm": 3.3151906439965835,
      "learning_rate": 1.844363181916986e-06,
      "loss": 0.9395,
      "num_input_tokens_seen": 95542830,
      "step": 4483
    },
    {
      "epoch": 0.5391691216256839,
      "flos": 11735245824000.0,
      "grad_norm": 6.170969449984072,
      "learning_rate": 1.8435865842870868e-06,
      "loss": 1.0203,
      "num_input_tokens_seen": 95560490,
      "step": 4484
    },
    {
      "epoch": 0.5392893645163229,
      "flos": 16585213501440.0,
      "grad_norm": 3.1830748714422397,
      "learning_rate": 1.8428100103849787e-06,
      "loss": 0.904,
      "num_input_tokens_seen": 95580005,
      "step": 4485
    },
    {
      "epoch": 0.5394096074069621,
      "flos": 11080176230400.0,
      "grad_norm": 4.875460871894974,
      "learning_rate": 1.842033460328467e-06,
      "loss": 0.9021,
      "num_input_tokens_seen": 95598445,
      "step": 4486
    },
    {
      "epoch": 0.5395298502976011,
      "flos": 16271601745920.0,
      "grad_norm": 5.226673438658187,
      "learning_rate": 1.8412569342353541e-06,
      "loss": 0.9462,
      "num_input_tokens_seen": 95618320,
      "step": 4487
    },
    {
      "epoch": 0.5396500931882402,
      "flos": 16953413713920.0,
      "grad_norm": 5.496754468991935,
      "learning_rate": 1.840480432223438e-06,
      "loss": 1.0433,
      "num_input_tokens_seen": 95637045,
      "step": 4488
    },
    {
      "epoch": 0.5397703360788794,
      "flos": 18736331243520.0,
      "grad_norm": 3.1737701316817013,
      "learning_rate": 1.8397039544105131e-06,
      "loss": 0.9533,
      "num_input_tokens_seen": 95655850,
      "step": 4489
    },
    {
      "epoch": 0.5398905789695184,
      "flos": 15064967639040.0,
      "grad_norm": 4.915721328286232,
      "learning_rate": 1.8389275009143711e-06,
      "loss": 0.8894,
      "num_input_tokens_seen": 95675310,
      "step": 4490
    },
    {
      "epoch": 0.5400108218601575,
      "flos": 17842971648000.0,
      "grad_norm": 9.404176603916808,
      "learning_rate": 1.8381510718527988e-06,
      "loss": 0.9217,
      "num_input_tokens_seen": 95694640,
      "step": 4491
    },
    {
      "epoch": 0.5401310647507966,
      "flos": 18760866201600.0,
      "grad_norm": 4.172307024242226,
      "learning_rate": 1.8373746673435812e-06,
      "loss": 0.8346,
      "num_input_tokens_seen": 95715385,
      "step": 4492
    },
    {
      "epoch": 0.5402513076414357,
      "flos": 19837590097920.0,
      "grad_norm": 5.036517116936014,
      "learning_rate": 1.8365982875044964e-06,
      "loss": 0.9837,
      "num_input_tokens_seen": 95735415,
      "step": 4493
    },
    {
      "epoch": 0.5403715505320748,
      "flos": 16271877672960.0,
      "grad_norm": 5.793377666998801,
      "learning_rate": 1.8358219324533217e-06,
      "loss": 0.9486,
      "num_input_tokens_seen": 95755400,
      "step": 4494
    },
    {
      "epoch": 0.5404917934227139,
      "flos": 21539636674560.0,
      "grad_norm": 4.288074483836616,
      "learning_rate": 1.8350456023078292e-06,
      "loss": 0.8951,
      "num_input_tokens_seen": 95777495,
      "step": 4495
    },
    {
      "epoch": 0.540612036313353,
      "flos": 14147992842240.0,
      "grad_norm": 5.645936953261185,
      "learning_rate": 1.8342692971857874e-06,
      "loss": 0.9513,
      "num_input_tokens_seen": 95796415,
      "step": 4496
    },
    {
      "epoch": 0.540732279203992,
      "flos": 17267761274880.0,
      "grad_norm": 11.680334168970857,
      "learning_rate": 1.833493017204962e-06,
      "loss": 0.9059,
      "num_input_tokens_seen": 95816240,
      "step": 4497
    },
    {
      "epoch": 0.5408525220946312,
      "flos": 14330958581760.0,
      "grad_norm": 15.514822486798712,
      "learning_rate": 1.8327167624831134e-06,
      "loss": 0.9628,
      "num_input_tokens_seen": 95833690,
      "step": 4498
    },
    {
      "epoch": 0.5409727649852702,
      "flos": 17163551047680.0,
      "grad_norm": 37.36483942941462,
      "learning_rate": 1.831940533137999e-06,
      "loss": 0.8995,
      "num_input_tokens_seen": 95852315,
      "step": 4499
    },
    {
      "epoch": 0.5410930078759093,
      "flos": 17003801948160.0,
      "grad_norm": 4.718236427906514,
      "learning_rate": 1.8311643292873718e-06,
      "loss": 0.9086,
      "num_input_tokens_seen": 95870855,
      "step": 4500
    },
    {
      "epoch": 0.5412132507665485,
      "flos": 14986304102400.0,
      "grad_norm": 3.229689147683043,
      "learning_rate": 1.8303881510489818e-06,
      "loss": 1.0795,
      "num_input_tokens_seen": 95888965,
      "step": 4501
    },
    {
      "epoch": 0.5413334936571875,
      "flos": 21542365286400.0,
      "grad_norm": 3.412609777676015,
      "learning_rate": 1.829611998540574e-06,
      "loss": 0.8798,
      "num_input_tokens_seen": 95909890,
      "step": 4502
    },
    {
      "epoch": 0.5414537365478266,
      "flos": 17267791933440.0,
      "grad_norm": 4.819760876814115,
      "learning_rate": 1.8288358718798914e-06,
      "loss": 0.9947,
      "num_input_tokens_seen": 95928800,
      "step": 4503
    },
    {
      "epoch": 0.5415739794384657,
      "flos": 11787841474560.0,
      "grad_norm": 5.2571937684443135,
      "learning_rate": 1.8280597711846703e-06,
      "loss": 0.9225,
      "num_input_tokens_seen": 95946760,
      "step": 4504
    },
    {
      "epoch": 0.5416942223291048,
      "flos": 16481708421120.0,
      "grad_norm": 6.933926048375187,
      "learning_rate": 1.8272836965726455e-06,
      "loss": 1.0334,
      "num_input_tokens_seen": 95965415,
      "step": 4505
    },
    {
      "epoch": 0.5418144652197439,
      "flos": 14410112655360.0,
      "grad_norm": 7.7057073570700005,
      "learning_rate": 1.8265076481615461e-06,
      "loss": 0.9744,
      "num_input_tokens_seen": 95985050,
      "step": 4506
    },
    {
      "epoch": 0.541934708110383,
      "flos": 8772313927680.0,
      "grad_norm": 5.095731010667758,
      "learning_rate": 1.8257316260690987e-06,
      "loss": 1.0639,
      "num_input_tokens_seen": 96002555,
      "step": 4507
    },
    {
      "epoch": 0.5420549510010221,
      "flos": 15093518868480.0,
      "grad_norm": 4.857460159868727,
      "learning_rate": 1.8249556304130254e-06,
      "loss": 0.9588,
      "num_input_tokens_seen": 96023555,
      "step": 4508
    },
    {
      "epoch": 0.5421751938916611,
      "flos": 21011871313920.0,
      "grad_norm": 4.016231030996092,
      "learning_rate": 1.824179661311044e-06,
      "loss": 0.8811,
      "num_input_tokens_seen": 96042025,
      "step": 4509
    },
    {
      "epoch": 0.5422954367823003,
      "flos": 13282479329280.0,
      "grad_norm": 11.000592302501687,
      "learning_rate": 1.823403718880868e-06,
      "loss": 0.9832,
      "num_input_tokens_seen": 96060505,
      "step": 4510
    },
    {
      "epoch": 0.5424156796729394,
      "flos": 28543512023040.0,
      "grad_norm": 5.675357474608944,
      "learning_rate": 1.822627803240207e-06,
      "loss": 0.8719,
      "num_input_tokens_seen": 96082555,
      "step": 4511
    },
    {
      "epoch": 0.5425359225635784,
      "flos": 8116815114240.0,
      "grad_norm": 4.289081177384197,
      "learning_rate": 1.8218519145067675e-06,
      "loss": 1.0463,
      "num_input_tokens_seen": 96097895,
      "step": 4512
    },
    {
      "epoch": 0.5426561654542175,
      "flos": 14357179760640.0,
      "grad_norm": 4.370086526603335,
      "learning_rate": 1.8210760527982508e-06,
      "loss": 1.0832,
      "num_input_tokens_seen": 96117900,
      "step": 4513
    },
    {
      "epoch": 0.5427764083448566,
      "flos": 15537255444480.0,
      "grad_norm": 4.959480131653875,
      "learning_rate": 1.8203002182323552e-06,
      "loss": 0.9375,
      "num_input_tokens_seen": 96135175,
      "step": 4514
    },
    {
      "epoch": 0.5428966512354957,
      "flos": 13933379358720.0,
      "grad_norm": 6.253521041872537,
      "learning_rate": 1.819524410926773e-06,
      "loss": 0.9507,
      "num_input_tokens_seen": 96152575,
      "step": 4515
    },
    {
      "epoch": 0.5430168941261347,
      "flos": 15904106680320.0,
      "grad_norm": 4.268445992989528,
      "learning_rate": 1.8187486309991944e-06,
      "loss": 0.9654,
      "num_input_tokens_seen": 96173175,
      "step": 4516
    },
    {
      "epoch": 0.5431371370167739,
      "flos": 13303120650240.0,
      "grad_norm": 6.435220851946736,
      "learning_rate": 1.817972878567304e-06,
      "loss": 0.9637,
      "num_input_tokens_seen": 96191550,
      "step": 4517
    },
    {
      "epoch": 0.543257379907413,
      "flos": 13333909954560.0,
      "grad_norm": 4.732037562101102,
      "learning_rate": 1.8171971537487834e-06,
      "loss": 0.9617,
      "num_input_tokens_seen": 96209920,
      "step": 4518
    },
    {
      "epoch": 0.543377622798052,
      "flos": 12390652661760.0,
      "grad_norm": 4.508936802965822,
      "learning_rate": 1.8164214566613093e-06,
      "loss": 0.9933,
      "num_input_tokens_seen": 96228265,
      "step": 4519
    },
    {
      "epoch": 0.5434978656886912,
      "flos": 13465751654400.0,
      "grad_norm": 10.008443700820822,
      "learning_rate": 1.8156457874225547e-06,
      "loss": 0.8372,
      "num_input_tokens_seen": 96246445,
      "step": 4520
    },
    {
      "epoch": 0.5436181085793302,
      "flos": 12233724149760.0,
      "grad_norm": 6.8410750783439775,
      "learning_rate": 1.814870146150187e-06,
      "loss": 0.9925,
      "num_input_tokens_seen": 96264275,
      "step": 4521
    },
    {
      "epoch": 0.5437383514699693,
      "flos": 13544261898240.0,
      "grad_norm": 6.35552727145383,
      "learning_rate": 1.814094532961871e-06,
      "loss": 0.9917,
      "num_input_tokens_seen": 96282570,
      "step": 4522
    },
    {
      "epoch": 0.5438585943606085,
      "flos": 16062813388800.0,
      "grad_norm": 7.248483696584977,
      "learning_rate": 1.8133189479752666e-06,
      "loss": 1.0174,
      "num_input_tokens_seen": 96301220,
      "step": 4523
    },
    {
      "epoch": 0.5439788372512475,
      "flos": 15484629135360.0,
      "grad_norm": 17.451307292598543,
      "learning_rate": 1.8125433913080292e-06,
      "loss": 1.0144,
      "num_input_tokens_seen": 96318640,
      "step": 4524
    },
    {
      "epoch": 0.5440990801418866,
      "flos": 11604477173760.0,
      "grad_norm": 14.356742609297186,
      "learning_rate": 1.811767863077811e-06,
      "loss": 1.0084,
      "num_input_tokens_seen": 96337310,
      "step": 4525
    },
    {
      "epoch": 0.5442193230325257,
      "flos": 15353155338240.0,
      "grad_norm": 3.1618508980792517,
      "learning_rate": 1.8109923634022577e-06,
      "loss": 0.9792,
      "num_input_tokens_seen": 96357055,
      "step": 4526
    },
    {
      "epoch": 0.5443395659231648,
      "flos": 10943797063680.0,
      "grad_norm": 5.975464244986764,
      "learning_rate": 1.8102168923990128e-06,
      "loss": 1.0646,
      "num_input_tokens_seen": 96370320,
      "step": 4527
    },
    {
      "epoch": 0.5444598088138038,
      "flos": 13308455239680.0,
      "grad_norm": 8.468255398091433,
      "learning_rate": 1.809441450185714e-06,
      "loss": 0.9886,
      "num_input_tokens_seen": 96388525,
      "step": 4528
    },
    {
      "epoch": 0.544580051704443,
      "flos": 15013138452480.0,
      "grad_norm": 6.160983826484621,
      "learning_rate": 1.8086660368799958e-06,
      "loss": 0.9228,
      "num_input_tokens_seen": 96406295,
      "step": 4529
    },
    {
      "epoch": 0.5447002945950821,
      "flos": 23169856573440.0,
      "grad_norm": 4.76826738987729,
      "learning_rate": 1.807890652599488e-06,
      "loss": 0.9611,
      "num_input_tokens_seen": 96430400,
      "step": 4530
    },
    {
      "epoch": 0.5448205374857211,
      "flos": 8297266851840.0,
      "grad_norm": 11.607360582929628,
      "learning_rate": 1.8071152974618156e-06,
      "loss": 1.015,
      "num_input_tokens_seen": 96447920,
      "step": 4531
    },
    {
      "epoch": 0.5449407803763603,
      "flos": 17162968535040.0,
      "grad_norm": 7.374507822973918,
      "learning_rate": 1.806339971584599e-06,
      "loss": 0.9766,
      "num_input_tokens_seen": 96464300,
      "step": 4532
    },
    {
      "epoch": 0.5450610232669993,
      "flos": 16534028144640.0,
      "grad_norm": 5.77472545417052,
      "learning_rate": 1.8055646750854546e-06,
      "loss": 1.0492,
      "num_input_tokens_seen": 96483530,
      "step": 4533
    },
    {
      "epoch": 0.5451812661576384,
      "flos": 12600268800000.0,
      "grad_norm": 19.006879665252132,
      "learning_rate": 1.8047894080819945e-06,
      "loss": 1.0101,
      "num_input_tokens_seen": 96500500,
      "step": 4534
    },
    {
      "epoch": 0.5453015090482776,
      "flos": 50895843840000.0,
      "grad_norm": 0.7565034394371979,
      "learning_rate": 1.8040141706918258e-06,
      "loss": 0.8575,
      "num_input_tokens_seen": 96561460,
      "step": 4535
    },
    {
      "epoch": 0.5454217519389166,
      "flos": 18181762191360.0,
      "grad_norm": 9.288805009178262,
      "learning_rate": 1.8032389630325525e-06,
      "loss": 0.964,
      "num_input_tokens_seen": 96579930,
      "step": 4536
    },
    {
      "epoch": 0.5455419948295557,
      "flos": 16821142794240.0,
      "grad_norm": 5.6730784250826884,
      "learning_rate": 1.8024637852217707e-06,
      "loss": 0.9594,
      "num_input_tokens_seen": 96599375,
      "step": 4537
    },
    {
      "epoch": 0.5456622377201948,
      "flos": 16638483640320.0,
      "grad_norm": 4.389451273902231,
      "learning_rate": 1.8016886373770766e-06,
      "loss": 1.0391,
      "num_input_tokens_seen": 96617610,
      "step": 4538
    },
    {
      "epoch": 0.5457824806108339,
      "flos": 17058267770880.0,
      "grad_norm": 3.909086494564282,
      "learning_rate": 1.8009135196160579e-06,
      "loss": 0.9681,
      "num_input_tokens_seen": 96636205,
      "step": 4539
    },
    {
      "epoch": 0.545902723501473,
      "flos": 16036285624320.0,
      "grad_norm": 4.018091535240491,
      "learning_rate": 1.8001384320563e-06,
      "loss": 1.0313,
      "num_input_tokens_seen": 96656180,
      "step": 4540
    },
    {
      "epoch": 0.5460229663921121,
      "flos": 39490589552640.0,
      "grad_norm": 0.8098866174956513,
      "learning_rate": 1.7993633748153833e-06,
      "loss": 0.7983,
      "num_input_tokens_seen": 96710505,
      "step": 4541
    },
    {
      "epoch": 0.5461432092827512,
      "flos": 10896168099840.0,
      "grad_norm": 5.115676214093441,
      "learning_rate": 1.7985883480108834e-06,
      "loss": 0.9195,
      "num_input_tokens_seen": 96727860,
      "step": 4542
    },
    {
      "epoch": 0.5462634521733902,
      "flos": 17084826193920.0,
      "grad_norm": 5.495049936694178,
      "learning_rate": 1.797813351760371e-06,
      "loss": 0.9199,
      "num_input_tokens_seen": 96749285,
      "step": 4543
    },
    {
      "epoch": 0.5463836950640293,
      "flos": 16219557949440.0,
      "grad_norm": 9.750255483556773,
      "learning_rate": 1.7970383861814116e-06,
      "loss": 0.969,
      "num_input_tokens_seen": 96768775,
      "step": 4544
    },
    {
      "epoch": 0.5465039379546685,
      "flos": 14514230906880.0,
      "grad_norm": 15.199865762218202,
      "learning_rate": 1.7962634513915684e-06,
      "loss": 0.9448,
      "num_input_tokens_seen": 96785845,
      "step": 4545
    },
    {
      "epoch": 0.5466241808453075,
      "flos": 12282242211840.0,
      "grad_norm": 11.64160741974362,
      "learning_rate": 1.7954885475083969e-06,
      "loss": 0.9933,
      "num_input_tokens_seen": 96803235,
      "step": 4546
    },
    {
      "epoch": 0.5467444237359466,
      "flos": 15354351022080.0,
      "grad_norm": 4.131747882060701,
      "learning_rate": 1.7947136746494513e-06,
      "loss": 0.9279,
      "num_input_tokens_seen": 96823870,
      "step": 4547
    },
    {
      "epoch": 0.5468646666265857,
      "flos": 17189373665280.0,
      "grad_norm": 24.58137171618416,
      "learning_rate": 1.793938832932277e-06,
      "loss": 1.0712,
      "num_input_tokens_seen": 96841700,
      "step": 4548
    },
    {
      "epoch": 0.5469849095172248,
      "flos": 19601630146560.0,
      "grad_norm": 3.9488781217081113,
      "learning_rate": 1.7931640224744185e-06,
      "loss": 0.8939,
      "num_input_tokens_seen": 96861970,
      "step": 4549
    },
    {
      "epoch": 0.5471051524078638,
      "flos": 19916161658880.0,
      "grad_norm": 10.48247082128269,
      "learning_rate": 1.7923892433934127e-06,
      "loss": 0.9308,
      "num_input_tokens_seen": 96882765,
      "step": 4550
    },
    {
      "epoch": 0.547225395298503,
      "flos": 12862971125760.0,
      "grad_norm": 15.25917152645674,
      "learning_rate": 1.7916144958067939e-06,
      "loss": 0.9822,
      "num_input_tokens_seen": 96900345,
      "step": 4551
    },
    {
      "epoch": 0.5473456381891421,
      "flos": 15170680135680.0,
      "grad_norm": 5.1716468968179,
      "learning_rate": 1.7908397798320905e-06,
      "loss": 0.9804,
      "num_input_tokens_seen": 96919800,
      "step": 4552
    },
    {
      "epoch": 0.5474658810797811,
      "flos": 14142566277120.0,
      "grad_norm": 3.7681130701339054,
      "learning_rate": 1.7900650955868265e-06,
      "loss": 0.9488,
      "num_input_tokens_seen": 96939165,
      "step": 4553
    },
    {
      "epoch": 0.5475861239704203,
      "flos": 36094751662080.0,
      "grad_norm": 4.275354711975137,
      "learning_rate": 1.7892904431885202e-06,
      "loss": 0.9629,
      "num_input_tokens_seen": 96966060,
      "step": 4554
    },
    {
      "epoch": 0.5477063668610593,
      "flos": 14699005501440.0,
      "grad_norm": 6.509987420745275,
      "learning_rate": 1.788515822754686e-06,
      "loss": 0.9453,
      "num_input_tokens_seen": 96986200,
      "step": 4555
    },
    {
      "epoch": 0.5478266097516984,
      "flos": 13911664988160.0,
      "grad_norm": 6.1039787324307415,
      "learning_rate": 1.7877412344028335e-06,
      "loss": 0.9662,
      "num_input_tokens_seen": 97005725,
      "step": 4556
    },
    {
      "epoch": 0.5479468526423376,
      "flos": 9087090708480.0,
      "grad_norm": 12.39943863613232,
      "learning_rate": 1.7869666782504668e-06,
      "loss": 0.9688,
      "num_input_tokens_seen": 97022025,
      "step": 4557
    },
    {
      "epoch": 0.5480670955329766,
      "flos": 13178023833600.0,
      "grad_norm": 7.043393647488853,
      "learning_rate": 1.7861921544150867e-06,
      "loss": 0.8824,
      "num_input_tokens_seen": 97040595,
      "step": 4558
    },
    {
      "epoch": 0.5481873384236157,
      "flos": 11283905925120.0,
      "grad_norm": 3.702337382452585,
      "learning_rate": 1.7854176630141856e-06,
      "loss": 0.952,
      "num_input_tokens_seen": 97057450,
      "step": 4559
    },
    {
      "epoch": 0.5483075813142548,
      "flos": 16192784916480.0,
      "grad_norm": 10.005997575843848,
      "learning_rate": 1.784643204165255e-06,
      "loss": 1.0268,
      "num_input_tokens_seen": 97076490,
      "step": 4560
    },
    {
      "epoch": 0.5484278242048939,
      "flos": 13701343703040.0,
      "grad_norm": 3.572821138548853,
      "learning_rate": 1.7838687779857783e-06,
      "loss": 0.9619,
      "num_input_tokens_seen": 97094085,
      "step": 4561
    },
    {
      "epoch": 0.5485480670955329,
      "flos": 16215357726720.0,
      "grad_norm": 4.914685232613473,
      "learning_rate": 1.7830943845932366e-06,
      "loss": 0.8281,
      "num_input_tokens_seen": 97113130,
      "step": 4562
    },
    {
      "epoch": 0.5486683099861721,
      "flos": 16111515402240.0,
      "grad_norm": 2.5406846718625196,
      "learning_rate": 1.7823200241051044e-06,
      "loss": 0.9473,
      "num_input_tokens_seen": 97131765,
      "step": 4563
    },
    {
      "epoch": 0.5487885528768112,
      "flos": 16455609876480.0,
      "grad_norm": 6.446180102590918,
      "learning_rate": 1.7815456966388513e-06,
      "loss": 0.9925,
      "num_input_tokens_seen": 97150580,
      "step": 4564
    },
    {
      "epoch": 0.5489087957674502,
      "flos": 15668208046080.0,
      "grad_norm": 10.556715605943243,
      "learning_rate": 1.780771402311943e-06,
      "loss": 1.0244,
      "num_input_tokens_seen": 97169135,
      "step": 4565
    },
    {
      "epoch": 0.5490290386580894,
      "flos": 17293890478080.0,
      "grad_norm": 4.9333757003477094,
      "learning_rate": 1.7799971412418374e-06,
      "loss": 0.9725,
      "num_input_tokens_seen": 97190250,
      "step": 4566
    },
    {
      "epoch": 0.5491492815487284,
      "flos": 12966568181760.0,
      "grad_norm": 23.487192113136363,
      "learning_rate": 1.7792229135459918e-06,
      "loss": 0.9326,
      "num_input_tokens_seen": 97206620,
      "step": 4567
    },
    {
      "epoch": 0.5492695244393675,
      "flos": 46213144719360.0,
      "grad_norm": 0.7848974532068688,
      "learning_rate": 1.7784487193418538e-06,
      "loss": 0.824,
      "num_input_tokens_seen": 97264190,
      "step": 4568
    },
    {
      "epoch": 0.5493897673300067,
      "flos": 12308279439360.0,
      "grad_norm": 9.319679309091939,
      "learning_rate": 1.7776745587468698e-06,
      "loss": 0.8028,
      "num_input_tokens_seen": 97281335,
      "step": 4569
    },
    {
      "epoch": 0.5495100102206457,
      "flos": 14121557053440.0,
      "grad_norm": 35.29528064681594,
      "learning_rate": 1.7769004318784776e-06,
      "loss": 1.0127,
      "num_input_tokens_seen": 97298700,
      "step": 4570
    },
    {
      "epoch": 0.5496302531112848,
      "flos": 11840007905280.0,
      "grad_norm": 4.387114835607856,
      "learning_rate": 1.776126338854113e-06,
      "loss": 1.0017,
      "num_input_tokens_seen": 97316210,
      "step": 4571
    },
    {
      "epoch": 0.5497504960019239,
      "flos": 17478266511360.0,
      "grad_norm": 4.009156000068561,
      "learning_rate": 1.7753522797912044e-06,
      "loss": 1.0422,
      "num_input_tokens_seen": 97336015,
      "step": 4572
    },
    {
      "epoch": 0.549870738892563,
      "flos": 10922511912960.0,
      "grad_norm": 16.26550023217037,
      "learning_rate": 1.7745782548071765e-06,
      "loss": 0.885,
      "num_input_tokens_seen": 97352630,
      "step": 4573
    },
    {
      "epoch": 0.549990981783202,
      "flos": 14960481484800.0,
      "grad_norm": 5.521986796483851,
      "learning_rate": 1.7738042640194482e-06,
      "loss": 0.9309,
      "num_input_tokens_seen": 97372015,
      "step": 4574
    },
    {
      "epoch": 0.5501112246738411,
      "flos": 15195123118080.0,
      "grad_norm": 3.860148487496306,
      "learning_rate": 1.7730303075454335e-06,
      "loss": 0.8981,
      "num_input_tokens_seen": 97390625,
      "step": 4575
    },
    {
      "epoch": 0.5502314675644803,
      "flos": 12364462141440.0,
      "grad_norm": 7.81202244634182,
      "learning_rate": 1.7722563855025402e-06,
      "loss": 1.0485,
      "num_input_tokens_seen": 97408375,
      "step": 4576
    },
    {
      "epoch": 0.5503517104551193,
      "flos": 17289138401280.0,
      "grad_norm": 8.01567639786153,
      "learning_rate": 1.7714824980081721e-06,
      "loss": 0.9052,
      "num_input_tokens_seen": 97427390,
      "step": 4577
    },
    {
      "epoch": 0.5504719533457584,
      "flos": 15930971688960.0,
      "grad_norm": 4.855212621462563,
      "learning_rate": 1.7707086451797276e-06,
      "loss": 0.9394,
      "num_input_tokens_seen": 97447985,
      "step": 4578
    },
    {
      "epoch": 0.5505921962363975,
      "flos": 48186140774400.0,
      "grad_norm": 0.7166062187236938,
      "learning_rate": 1.7699348271345993e-06,
      "loss": 0.7335,
      "num_input_tokens_seen": 97510330,
      "step": 4579
    },
    {
      "epoch": 0.5507124391270366,
      "flos": 32654863749120.0,
      "grad_norm": 0.7524582392488682,
      "learning_rate": 1.7691610439901753e-06,
      "loss": 0.7569,
      "num_input_tokens_seen": 97572985,
      "step": 4580
    },
    {
      "epoch": 0.5508326820176757,
      "flos": 15826393559040.0,
      "grad_norm": 11.151305569562773,
      "learning_rate": 1.7683872958638367e-06,
      "loss": 0.9477,
      "num_input_tokens_seen": 97591585,
      "step": 4581
    },
    {
      "epoch": 0.5509529249083148,
      "flos": 14201262981120.0,
      "grad_norm": 18.571898267905667,
      "learning_rate": 1.7676135828729614e-06,
      "loss": 1.0305,
      "num_input_tokens_seen": 97611015,
      "step": 4582
    },
    {
      "epoch": 0.5510731677989539,
      "flos": 15510697021440.0,
      "grad_norm": 7.892279181498266,
      "learning_rate": 1.7668399051349205e-06,
      "loss": 1.0252,
      "num_input_tokens_seen": 97630415,
      "step": 4583
    },
    {
      "epoch": 0.5511934106895929,
      "flos": 15246737694720.0,
      "grad_norm": 9.079428810605625,
      "learning_rate": 1.766066262767081e-06,
      "loss": 1.0155,
      "num_input_tokens_seen": 97647975,
      "step": 4584
    },
    {
      "epoch": 0.5513136535802321,
      "flos": 14960328192000.0,
      "grad_norm": 14.757571071509838,
      "learning_rate": 1.765292655886803e-06,
      "loss": 0.9703,
      "num_input_tokens_seen": 97666340,
      "step": 4585
    },
    {
      "epoch": 0.5514338964708712,
      "flos": 19808088453120.0,
      "grad_norm": 5.054706933103583,
      "learning_rate": 1.764519084611443e-06,
      "loss": 0.9098,
      "num_input_tokens_seen": 97686515,
      "step": 4586
    },
    {
      "epoch": 0.5515541393615102,
      "flos": 15563783208960.0,
      "grad_norm": 8.949751559597374,
      "learning_rate": 1.7637455490583505e-06,
      "loss": 0.9659,
      "num_input_tokens_seen": 97705560,
      "step": 4587
    },
    {
      "epoch": 0.5516743822521494,
      "flos": 14541003939840.0,
      "grad_norm": 6.934818608800297,
      "learning_rate": 1.7629720493448701e-06,
      "loss": 0.9723,
      "num_input_tokens_seen": 97722575,
      "step": 4588
    },
    {
      "epoch": 0.5517946251427884,
      "flos": 10555997921280.0,
      "grad_norm": 5.883534502403024,
      "learning_rate": 1.7621985855883418e-06,
      "loss": 1.0418,
      "num_input_tokens_seen": 97738995,
      "step": 4589
    },
    {
      "epoch": 0.5519148680334275,
      "flos": 13045630279680.0,
      "grad_norm": 7.384916201602256,
      "learning_rate": 1.7614251579060983e-06,
      "loss": 0.9115,
      "num_input_tokens_seen": 97757310,
      "step": 4590
    },
    {
      "epoch": 0.5520351109240667,
      "flos": 17866218946560.0,
      "grad_norm": 7.560458829973652,
      "learning_rate": 1.76065176641547e-06,
      "loss": 1.0426,
      "num_input_tokens_seen": 97779740,
      "step": 4591
    },
    {
      "epoch": 0.5521553538147057,
      "flos": 14960849387520.0,
      "grad_norm": 3.5577666308596188,
      "learning_rate": 1.759878411233777e-06,
      "loss": 0.9755,
      "num_input_tokens_seen": 97797920,
      "step": 4592
    },
    {
      "epoch": 0.5522755967053448,
      "flos": 13387210752000.0,
      "grad_norm": 4.26292675041105,
      "learning_rate": 1.7591050924783388e-06,
      "loss": 0.9525,
      "num_input_tokens_seen": 97814830,
      "step": 4593
    },
    {
      "epoch": 0.5523958395959839,
      "flos": 46305389998080.0,
      "grad_norm": 0.9104048098934876,
      "learning_rate": 1.7583318102664661e-06,
      "loss": 0.8283,
      "num_input_tokens_seen": 97882115,
      "step": 4594
    },
    {
      "epoch": 0.552516082486623,
      "flos": 7644772577280.0,
      "grad_norm": 7.014033555489259,
      "learning_rate": 1.757558564715466e-06,
      "loss": 0.9841,
      "num_input_tokens_seen": 97899910,
      "step": 4595
    },
    {
      "epoch": 0.552636325377262,
      "flos": 15773031444480.0,
      "grad_norm": 6.463374903943433,
      "learning_rate": 1.7567853559426386e-06,
      "loss": 0.9194,
      "num_input_tokens_seen": 97916680,
      "step": 4596
    },
    {
      "epoch": 0.5527565682679012,
      "flos": 17058390405120.0,
      "grad_norm": 25.89205002301925,
      "learning_rate": 1.7560121840652797e-06,
      "loss": 0.9465,
      "num_input_tokens_seen": 97935785,
      "step": 4597
    },
    {
      "epoch": 0.5528768111585403,
      "flos": 13990389841920.0,
      "grad_norm": 7.097754604264878,
      "learning_rate": 1.7552390492006782e-06,
      "loss": 0.8846,
      "num_input_tokens_seen": 97953825,
      "step": 4598
    },
    {
      "epoch": 0.5529970540491793,
      "flos": 18659752488960.0,
      "grad_norm": 13.536855615723228,
      "learning_rate": 1.7544659514661184e-06,
      "loss": 0.8528,
      "num_input_tokens_seen": 97976635,
      "step": 4599
    },
    {
      "epoch": 0.5531172969398185,
      "flos": 17373013893120.0,
      "grad_norm": 15.080208526707592,
      "learning_rate": 1.7536928909788786e-06,
      "loss": 0.9967,
      "num_input_tokens_seen": 97995660,
      "step": 4600
    },
    {
      "epoch": 0.5532375398304575,
      "flos": 46468703600640.0,
      "grad_norm": 0.9420049512656095,
      "learning_rate": 1.752919867856231e-06,
      "loss": 0.8344,
      "num_input_tokens_seen": 98047025,
      "step": 4601
    },
    {
      "epoch": 0.5533577827210966,
      "flos": 13964444590080.0,
      "grad_norm": 7.419156574526767,
      "learning_rate": 1.7521468822154436e-06,
      "loss": 0.9792,
      "num_input_tokens_seen": 98065660,
      "step": 4602
    },
    {
      "epoch": 0.5534780256117358,
      "flos": 23036543262720.0,
      "grad_norm": 3.6861228310894507,
      "learning_rate": 1.751373934173777e-06,
      "loss": 0.9291,
      "num_input_tokens_seen": 98088125,
      "step": 4603
    },
    {
      "epoch": 0.5535982685023748,
      "flos": 16508113551360.0,
      "grad_norm": 6.4738342649725,
      "learning_rate": 1.750601023848487e-06,
      "loss": 0.9249,
      "num_input_tokens_seen": 98108570,
      "step": 4604
    },
    {
      "epoch": 0.5537185113930139,
      "flos": 17320172974080.0,
      "grad_norm": 7.6744691149478905,
      "learning_rate": 1.749828151356823e-06,
      "loss": 0.9303,
      "num_input_tokens_seen": 98128485,
      "step": 4605
    },
    {
      "epoch": 0.553838754283653,
      "flos": 16742632550400.0,
      "grad_norm": 5.380980424852111,
      "learning_rate": 1.7490553168160297e-06,
      "loss": 0.9585,
      "num_input_tokens_seen": 98149275,
      "step": 4606
    },
    {
      "epoch": 0.5539589971742921,
      "flos": 12233816125440.0,
      "grad_norm": 7.197455369785132,
      "learning_rate": 1.748282520343345e-06,
      "loss": 0.9482,
      "num_input_tokens_seen": 98168025,
      "step": 4607
    },
    {
      "epoch": 0.5540792400649311,
      "flos": 19628433838080.0,
      "grad_norm": 4.764006192783843,
      "learning_rate": 1.7475097620560023e-06,
      "loss": 0.9744,
      "num_input_tokens_seen": 98187810,
      "step": 4608
    },
    {
      "epoch": 0.5541994829555702,
      "flos": 16979665551360.0,
      "grad_norm": 4.697265828027542,
      "learning_rate": 1.746737042071228e-06,
      "loss": 0.9034,
      "num_input_tokens_seen": 98206035,
      "step": 4609
    },
    {
      "epoch": 0.5543197258462094,
      "flos": 14274653245440.0,
      "grad_norm": 7.992695811822944,
      "learning_rate": 1.7459643605062424e-06,
      "loss": 0.9908,
      "num_input_tokens_seen": 98223015,
      "step": 4610
    },
    {
      "epoch": 0.5544399687368484,
      "flos": 14849955594240.0,
      "grad_norm": 8.770917032768189,
      "learning_rate": 1.745191717478262e-06,
      "loss": 1.0072,
      "num_input_tokens_seen": 98241315,
      "step": 4611
    },
    {
      "epoch": 0.5545602116274875,
      "flos": 18158055014400.0,
      "grad_norm": 5.866924498412125,
      "learning_rate": 1.7444191131044948e-06,
      "loss": 1.0004,
      "num_input_tokens_seen": 98261310,
      "step": 4612
    },
    {
      "epoch": 0.5546804545181266,
      "flos": 14907548590080.0,
      "grad_norm": 5.030297230608892,
      "learning_rate": 1.7436465475021456e-06,
      "loss": 0.9183,
      "num_input_tokens_seen": 98281080,
      "step": 4613
    },
    {
      "epoch": 0.5548006974087657,
      "flos": 19103305113600.0,
      "grad_norm": 9.463068198475794,
      "learning_rate": 1.7428740207884111e-06,
      "loss": 0.8995,
      "num_input_tokens_seen": 98301680,
      "step": 4614
    },
    {
      "epoch": 0.5549209402994048,
      "flos": 24006696222720.0,
      "grad_norm": 76.98015229691515,
      "learning_rate": 1.7421015330804833e-06,
      "loss": 0.8117,
      "num_input_tokens_seen": 98321320,
      "step": 4615
    },
    {
      "epoch": 0.5550411831900439,
      "flos": 16901277941760.0,
      "grad_norm": 6.207643338291085,
      "learning_rate": 1.7413290844955475e-06,
      "loss": 0.9188,
      "num_input_tokens_seen": 98341070,
      "step": 4616
    },
    {
      "epoch": 0.555161426080683,
      "flos": 15378579394560.0,
      "grad_norm": 3.702633338191207,
      "learning_rate": 1.7405566751507843e-06,
      "loss": 0.9801,
      "num_input_tokens_seen": 98358835,
      "step": 4617
    },
    {
      "epoch": 0.555281668971322,
      "flos": 35438394408960.0,
      "grad_norm": 7.707689028094516,
      "learning_rate": 1.7397843051633668e-06,
      "loss": 0.8651,
      "num_input_tokens_seen": 98381250,
      "step": 4618
    },
    {
      "epoch": 0.5554019118619612,
      "flos": 14725226680320.0,
      "grad_norm": 5.1090476021586495,
      "learning_rate": 1.739011974650464e-06,
      "loss": 0.9044,
      "num_input_tokens_seen": 98400300,
      "step": 4619
    },
    {
      "epoch": 0.5555221547526003,
      "flos": 18132722933760.0,
      "grad_norm": 18.314328880250688,
      "learning_rate": 1.7382396837292365e-06,
      "loss": 0.9554,
      "num_input_tokens_seen": 98420480,
      "step": 4620
    },
    {
      "epoch": 0.5556423976432393,
      "flos": 15458837176320.0,
      "grad_norm": 18.181135289632543,
      "learning_rate": 1.737467432516841e-06,
      "loss": 0.9294,
      "num_input_tokens_seen": 98440300,
      "step": 4621
    },
    {
      "epoch": 0.5557626405338785,
      "flos": 17713797242880.0,
      "grad_norm": 5.750021893003498,
      "learning_rate": 1.7366952211304274e-06,
      "loss": 0.9203,
      "num_input_tokens_seen": 98457865,
      "step": 4622
    },
    {
      "epoch": 0.5558828834245175,
      "flos": 13256043540480.0,
      "grad_norm": 6.72145203661149,
      "learning_rate": 1.735923049687139e-06,
      "loss": 1.0278,
      "num_input_tokens_seen": 98474160,
      "step": 4623
    },
    {
      "epoch": 0.5560031263151566,
      "flos": 19418572431360.0,
      "grad_norm": 4.4945465527915,
      "learning_rate": 1.7351509183041144e-06,
      "loss": 0.9382,
      "num_input_tokens_seen": 98494210,
      "step": 4624
    },
    {
      "epoch": 0.5561233692057957,
      "flos": 16638269030400.0,
      "grad_norm": 6.128008979193182,
      "learning_rate": 1.7343788270984852e-06,
      "loss": 0.9196,
      "num_input_tokens_seen": 98513070,
      "step": 4625
    },
    {
      "epoch": 0.5562436120964348,
      "flos": 26891945656320.0,
      "grad_norm": 12.605928122440053,
      "learning_rate": 1.7336067761873764e-06,
      "loss": 0.9405,
      "num_input_tokens_seen": 98535215,
      "step": 4626
    },
    {
      "epoch": 0.5563638549870739,
      "flos": 17897069568000.0,
      "grad_norm": 9.154785872434001,
      "learning_rate": 1.7328347656879076e-06,
      "loss": 0.95,
      "num_input_tokens_seen": 98554795,
      "step": 4627
    },
    {
      "epoch": 0.556484097877713,
      "flos": 9578456248320.0,
      "grad_norm": 6.464335681508399,
      "learning_rate": 1.7320627957171927e-06,
      "loss": 0.883,
      "num_input_tokens_seen": 98569175,
      "step": 4628
    },
    {
      "epoch": 0.5566043407683521,
      "flos": 17556562145280.0,
      "grad_norm": 8.472659851889507,
      "learning_rate": 1.7312908663923382e-06,
      "loss": 0.9967,
      "num_input_tokens_seen": 98585070,
      "step": 4629
    },
    {
      "epoch": 0.5567245836589911,
      "flos": 14613934325760.0,
      "grad_norm": 5.0742788920608275,
      "learning_rate": 1.7305189778304463e-06,
      "loss": 0.8601,
      "num_input_tokens_seen": 98602965,
      "step": 4630
    },
    {
      "epoch": 0.5568448265496303,
      "flos": 14698208378880.0,
      "grad_norm": 4.991534314234518,
      "learning_rate": 1.729747130148611e-06,
      "loss": 0.9984,
      "num_input_tokens_seen": 98621880,
      "step": 4631
    },
    {
      "epoch": 0.5569650694402694,
      "flos": 18003425894400.0,
      "grad_norm": 11.696605379022502,
      "learning_rate": 1.7289753234639208e-06,
      "loss": 0.9595,
      "num_input_tokens_seen": 98640575,
      "step": 4632
    },
    {
      "epoch": 0.5570853123309084,
      "flos": 13984718008320.0,
      "grad_norm": 5.858064632123041,
      "learning_rate": 1.7282035578934592e-06,
      "loss": 0.9535,
      "num_input_tokens_seen": 98658460,
      "step": 4633
    },
    {
      "epoch": 0.5572055552215476,
      "flos": 11395228938240.0,
      "grad_norm": 7.731106030689507,
      "learning_rate": 1.727431833554301e-06,
      "loss": 0.9776,
      "num_input_tokens_seen": 98676655,
      "step": 4634
    },
    {
      "epoch": 0.5573257981121866,
      "flos": 12128686141440.0,
      "grad_norm": 4.534943016224172,
      "learning_rate": 1.7266601505635175e-06,
      "loss": 0.9624,
      "num_input_tokens_seen": 98693715,
      "step": 4635
    },
    {
      "epoch": 0.5574460410028257,
      "flos": 13334768394240.0,
      "grad_norm": 10.402386655586506,
      "learning_rate": 1.7258885090381717e-06,
      "loss": 0.9518,
      "num_input_tokens_seen": 98711475,
      "step": 4636
    },
    {
      "epoch": 0.5575662838934649,
      "flos": 21122059960320.0,
      "grad_norm": 14.034015204625934,
      "learning_rate": 1.7251169090953213e-06,
      "loss": 0.9826,
      "num_input_tokens_seen": 98731670,
      "step": 4637
    },
    {
      "epoch": 0.5576865267841039,
      "flos": 15668667924480.0,
      "grad_norm": 4.742650821682559,
      "learning_rate": 1.7243453508520168e-06,
      "loss": 0.9494,
      "num_input_tokens_seen": 98748375,
      "step": 4638
    },
    {
      "epoch": 0.557806769674743,
      "flos": 12176223129600.0,
      "grad_norm": 15.867505805214492,
      "learning_rate": 1.7235738344253038e-06,
      "loss": 1.0405,
      "num_input_tokens_seen": 98761725,
      "step": 4639
    },
    {
      "epoch": 0.557927012565382,
      "flos": 17578399150080.0,
      "grad_norm": 3.9228390424479693,
      "learning_rate": 1.72280235993222e-06,
      "loss": 1.0087,
      "num_input_tokens_seen": 98779750,
      "step": 4640
    },
    {
      "epoch": 0.5580472554560212,
      "flos": 12023985377280.0,
      "grad_norm": 5.839965900687384,
      "learning_rate": 1.722030927489798e-06,
      "loss": 0.8757,
      "num_input_tokens_seen": 98796750,
      "step": 4641
    },
    {
      "epoch": 0.5581674983466602,
      "flos": 16713437491200.0,
      "grad_norm": 4.133387505781331,
      "learning_rate": 1.7212595372150634e-06,
      "loss": 0.9376,
      "num_input_tokens_seen": 98816450,
      "step": 4642
    },
    {
      "epoch": 0.5582877412372993,
      "flos": 9506568253440.0,
      "grad_norm": 7.08808249042809,
      "learning_rate": 1.720488189225035e-06,
      "loss": 0.9254,
      "num_input_tokens_seen": 98833870,
      "step": 4643
    },
    {
      "epoch": 0.5584079841279385,
      "flos": 15559582986240.0,
      "grad_norm": 7.434982258327227,
      "learning_rate": 1.7197168836367265e-06,
      "loss": 0.9854,
      "num_input_tokens_seen": 98850400,
      "step": 4644
    },
    {
      "epoch": 0.5585282270185775,
      "flos": 13361020231680.0,
      "grad_norm": 5.164332488096348,
      "learning_rate": 1.7189456205671433e-06,
      "loss": 1.0111,
      "num_input_tokens_seen": 98868965,
      "step": 4645
    },
    {
      "epoch": 0.5586484699092166,
      "flos": 15534496174080.0,
      "grad_norm": 4.5921582429050565,
      "learning_rate": 1.7181744001332866e-06,
      "loss": 1.017,
      "num_input_tokens_seen": 98887295,
      "step": 4646
    },
    {
      "epoch": 0.5587687127998557,
      "flos": 16271417794560.0,
      "grad_norm": 3.7894700922447044,
      "learning_rate": 1.7174032224521493e-06,
      "loss": 0.8328,
      "num_input_tokens_seen": 98905725,
      "step": 4647
    },
    {
      "epoch": 0.5588889556904948,
      "flos": 14409959362560.0,
      "grad_norm": 4.732104471467408,
      "learning_rate": 1.7166320876407184e-06,
      "loss": 0.8982,
      "num_input_tokens_seen": 98924865,
      "step": 4648
    },
    {
      "epoch": 0.5590091985811338,
      "flos": 11656766238720.0,
      "grad_norm": 4.985900262868724,
      "learning_rate": 1.7158609958159742e-06,
      "loss": 0.8703,
      "num_input_tokens_seen": 98941990,
      "step": 4649
    },
    {
      "epoch": 0.559129441471773,
      "flos": 10261709168640.0,
      "grad_norm": 8.174942753675948,
      "learning_rate": 1.7150899470948911e-06,
      "loss": 0.9828,
      "num_input_tokens_seen": 98956975,
      "step": 4650
    },
    {
      "epoch": 0.5592496843624121,
      "flos": 43320191877120.0,
      "grad_norm": 0.8297466019954219,
      "learning_rate": 1.7143189415944365e-06,
      "loss": 0.7954,
      "num_input_tokens_seen": 99021155,
      "step": 4651
    },
    {
      "epoch": 0.5593699272530511,
      "flos": 14829007687680.0,
      "grad_norm": 4.78574274945284,
      "learning_rate": 1.7135479794315714e-06,
      "loss": 0.956,
      "num_input_tokens_seen": 99037830,
      "step": 4652
    },
    {
      "epoch": 0.5594901701436903,
      "flos": 9087213342720.0,
      "grad_norm": 4.296241519354673,
      "learning_rate": 1.7127770607232502e-06,
      "loss": 0.981,
      "num_input_tokens_seen": 99056095,
      "step": 4653
    },
    {
      "epoch": 0.5596104130343293,
      "flos": 16428714209280.0,
      "grad_norm": 5.302851633141855,
      "learning_rate": 1.7120061855864204e-06,
      "loss": 0.9967,
      "num_input_tokens_seen": 99075825,
      "step": 4654
    },
    {
      "epoch": 0.5597306559249684,
      "flos": 18473843527680.0,
      "grad_norm": 3.3807289102770004,
      "learning_rate": 1.7112353541380233e-06,
      "loss": 0.9045,
      "num_input_tokens_seen": 99095405,
      "step": 4655
    },
    {
      "epoch": 0.5598508988156076,
      "flos": 15983322071040.0,
      "grad_norm": 7.210269322543596,
      "learning_rate": 1.7104645664949931e-06,
      "loss": 0.9197,
      "num_input_tokens_seen": 99117595,
      "step": 4656
    },
    {
      "epoch": 0.5599711417062466,
      "flos": 16429082112000.0,
      "grad_norm": 3.6957020463529635,
      "learning_rate": 1.7096938227742584e-06,
      "loss": 0.9169,
      "num_input_tokens_seen": 99138445,
      "step": 4657
    },
    {
      "epoch": 0.5600913845968857,
      "flos": 15872458936320.0,
      "grad_norm": 3.4202484027264073,
      "learning_rate": 1.70892312309274e-06,
      "loss": 1.0327,
      "num_input_tokens_seen": 99156055,
      "step": 4658
    },
    {
      "epoch": 0.5602116274875248,
      "flos": 12490815959040.0,
      "grad_norm": 5.222833007012634,
      "learning_rate": 1.7081524675673523e-06,
      "loss": 0.8664,
      "num_input_tokens_seen": 99171265,
      "step": 4659
    },
    {
      "epoch": 0.5603318703781639,
      "flos": 50215779409920.0,
      "grad_norm": 0.8409007318652976,
      "learning_rate": 1.7073818563150026e-06,
      "loss": 0.8369,
      "num_input_tokens_seen": 99233065,
      "step": 4660
    },
    {
      "epoch": 0.560452113268803,
      "flos": 13146284113920.0,
      "grad_norm": 3.7172483088877986,
      "learning_rate": 1.7066112894525935e-06,
      "loss": 1.0502,
      "num_input_tokens_seen": 99250865,
      "step": 4661
    },
    {
      "epoch": 0.5605723561594421,
      "flos": 17975702446080.0,
      "grad_norm": 3.8825054699723522,
      "learning_rate": 1.7058407670970177e-06,
      "loss": 0.9234,
      "num_input_tokens_seen": 99272060,
      "step": 4662
    },
    {
      "epoch": 0.5606925990500812,
      "flos": 14619728793600.0,
      "grad_norm": 7.898927697912892,
      "learning_rate": 1.7050702893651643e-06,
      "loss": 0.8047,
      "num_input_tokens_seen": 99291360,
      "step": 4663
    },
    {
      "epoch": 0.5608128419407202,
      "flos": 24976603914240.0,
      "grad_norm": 3.7564263275448995,
      "learning_rate": 1.7042998563739134e-06,
      "loss": 0.9494,
      "num_input_tokens_seen": 99309430,
      "step": 4664
    },
    {
      "epoch": 0.5609330848313594,
      "flos": 17526018109440.0,
      "grad_norm": 5.153723820396175,
      "learning_rate": 1.703529468240139e-06,
      "loss": 0.9026,
      "num_input_tokens_seen": 99328020,
      "step": 4665
    },
    {
      "epoch": 0.5610533277219985,
      "flos": 13302016942080.0,
      "grad_norm": 4.637341092001369,
      "learning_rate": 1.7027591250807088e-06,
      "loss": 0.9167,
      "num_input_tokens_seen": 99344915,
      "step": 4666
    },
    {
      "epoch": 0.5611735706126375,
      "flos": 10607612497920.0,
      "grad_norm": 5.5195212286489,
      "learning_rate": 1.7019888270124825e-06,
      "loss": 1.0154,
      "num_input_tokens_seen": 99361800,
      "step": 4667
    },
    {
      "epoch": 0.5612938135032767,
      "flos": 11653363138560.0,
      "grad_norm": 11.919649688898357,
      "learning_rate": 1.7012185741523147e-06,
      "loss": 1.0198,
      "num_input_tokens_seen": 99377845,
      "step": 4668
    },
    {
      "epoch": 0.5614140563939157,
      "flos": 18264503316480.0,
      "grad_norm": 3.672031740130374,
      "learning_rate": 1.7004483666170514e-06,
      "loss": 0.8027,
      "num_input_tokens_seen": 99398060,
      "step": 4669
    },
    {
      "epoch": 0.5615342992845548,
      "flos": 17582538055680.0,
      "grad_norm": 3.3379146246942937,
      "learning_rate": 1.699678204523533e-06,
      "loss": 0.9825,
      "num_input_tokens_seen": 99417645,
      "step": 4670
    },
    {
      "epoch": 0.5616545421751938,
      "flos": 15640699207680.0,
      "grad_norm": 6.248716909309888,
      "learning_rate": 1.6989080879885918e-06,
      "loss": 0.8647,
      "num_input_tokens_seen": 99435225,
      "step": 4671
    },
    {
      "epoch": 0.561774785065833,
      "flos": 38459984240640.0,
      "grad_norm": 0.902585579309659,
      "learning_rate": 1.6981380171290544e-06,
      "loss": 0.8305,
      "num_input_tokens_seen": 99495970,
      "step": 4672
    },
    {
      "epoch": 0.5618950279564721,
      "flos": 14011705651200.0,
      "grad_norm": 3.0944002026953332,
      "learning_rate": 1.6973679920617396e-06,
      "loss": 0.932,
      "num_input_tokens_seen": 99513225,
      "step": 4673
    },
    {
      "epoch": 0.5620152708471111,
      "flos": 11892818165760.0,
      "grad_norm": 6.56335414595797,
      "learning_rate": 1.6965980129034603e-06,
      "loss": 1.0513,
      "num_input_tokens_seen": 99530330,
      "step": 4674
    },
    {
      "epoch": 0.5621355137377503,
      "flos": 19077635788800.0,
      "grad_norm": 2.483412454185277,
      "learning_rate": 1.6958280797710209e-06,
      "loss": 0.9596,
      "num_input_tokens_seen": 99551975,
      "step": 4675
    },
    {
      "epoch": 0.5622557566283893,
      "flos": 43812262563840.0,
      "grad_norm": 0.7392470486259178,
      "learning_rate": 1.6950581927812198e-06,
      "loss": 0.7497,
      "num_input_tokens_seen": 99611265,
      "step": 4676
    },
    {
      "epoch": 0.5623759995190284,
      "flos": 18840878714880.0,
      "grad_norm": 4.447819704998599,
      "learning_rate": 1.6942883520508486e-06,
      "loss": 0.9636,
      "num_input_tokens_seen": 99629720,
      "step": 4677
    },
    {
      "epoch": 0.5624962424096676,
      "flos": 13754399232000.0,
      "grad_norm": 3.5910927941728077,
      "learning_rate": 1.693518557696691e-06,
      "loss": 0.9704,
      "num_input_tokens_seen": 99648580,
      "step": 4678
    },
    {
      "epoch": 0.5626164853003066,
      "flos": 14671680614400.0,
      "grad_norm": 6.5058552705631,
      "learning_rate": 1.6927488098355252e-06,
      "loss": 1.0794,
      "num_input_tokens_seen": 99665930,
      "step": 4679
    },
    {
      "epoch": 0.5627367281909457,
      "flos": 44931372810240.0,
      "grad_norm": 0.8931871335711724,
      "learning_rate": 1.6919791085841201e-06,
      "loss": 0.8669,
      "num_input_tokens_seen": 99723060,
      "step": 4680
    },
    {
      "epoch": 0.5628569710815848,
      "flos": 9008427171840.0,
      "grad_norm": 5.223916243449842,
      "learning_rate": 1.6912094540592396e-06,
      "loss": 0.9809,
      "num_input_tokens_seen": 99738300,
      "step": 4681
    },
    {
      "epoch": 0.5629772139722239,
      "flos": 9707722629120.0,
      "grad_norm": 5.246661259908124,
      "learning_rate": 1.6904398463776393e-06,
      "loss": 1.0012,
      "num_input_tokens_seen": 99751820,
      "step": 4682
    },
    {
      "epoch": 0.5630974568628629,
      "flos": 15246676377600.0,
      "grad_norm": 4.024856109890423,
      "learning_rate": 1.6896702856560683e-06,
      "loss": 0.9293,
      "num_input_tokens_seen": 99770635,
      "step": 4683
    },
    {
      "epoch": 0.5632176997535021,
      "flos": 10056722472960.0,
      "grad_norm": 12.921028893663943,
      "learning_rate": 1.6889007720112677e-06,
      "loss": 0.8655,
      "num_input_tokens_seen": 99788100,
      "step": 4684
    },
    {
      "epoch": 0.5633379426441412,
      "flos": 14775676231680.0,
      "grad_norm": 4.032439300401902,
      "learning_rate": 1.6881313055599734e-06,
      "loss": 0.9741,
      "num_input_tokens_seen": 99807750,
      "step": 4685
    },
    {
      "epoch": 0.5634581855347802,
      "flos": 16061249802240.0,
      "grad_norm": 3.9668129827190195,
      "learning_rate": 1.6873618864189117e-06,
      "loss": 1.0131,
      "num_input_tokens_seen": 99823240,
      "step": 4686
    },
    {
      "epoch": 0.5635784284254194,
      "flos": 15275166289920.0,
      "grad_norm": 4.093365600243382,
      "learning_rate": 1.686592514704803e-06,
      "loss": 0.971,
      "num_input_tokens_seen": 99840355,
      "step": 4687
    },
    {
      "epoch": 0.5636986713160584,
      "flos": 14067980328960.0,
      "grad_norm": 4.078920484107753,
      "learning_rate": 1.685823190534361e-06,
      "loss": 0.8953,
      "num_input_tokens_seen": 99858315,
      "step": 4688
    },
    {
      "epoch": 0.5638189142066975,
      "flos": 14043016151040.0,
      "grad_norm": 4.0719590618271555,
      "learning_rate": 1.6850539140242907e-06,
      "loss": 1.0159,
      "num_input_tokens_seen": 99877295,
      "step": 4689
    },
    {
      "epoch": 0.5639391570973367,
      "flos": 16271877672960.0,
      "grad_norm": 4.379609431790056,
      "learning_rate": 1.684284685291292e-06,
      "loss": 1.0071,
      "num_input_tokens_seen": 99898660,
      "step": 4690
    },
    {
      "epoch": 0.5640593999879757,
      "flos": 16869967441920.0,
      "grad_norm": 2.8617343861545597,
      "learning_rate": 1.683515504452055e-06,
      "loss": 0.9942,
      "num_input_tokens_seen": 99915755,
      "step": 4691
    },
    {
      "epoch": 0.5641796428786148,
      "flos": 16139729387520.0,
      "grad_norm": 3.409678562262023,
      "learning_rate": 1.6827463716232648e-06,
      "loss": 0.8581,
      "num_input_tokens_seen": 99936135,
      "step": 4692
    },
    {
      "epoch": 0.5642998857692539,
      "flos": 14042617589760.0,
      "grad_norm": 4.497988569218828,
      "learning_rate": 1.6819772869215972e-06,
      "loss": 0.955,
      "num_input_tokens_seen": 99954935,
      "step": 4693
    },
    {
      "epoch": 0.564420128659893,
      "flos": 16481861713920.0,
      "grad_norm": 5.393379016552288,
      "learning_rate": 1.6812082504637228e-06,
      "loss": 1.0099,
      "num_input_tokens_seen": 99975975,
      "step": 4694
    },
    {
      "epoch": 0.564540371550532,
      "flos": 16533782876160.0,
      "grad_norm": 2.55922694633872,
      "learning_rate": 1.6804392623663025e-06,
      "loss": 0.9438,
      "num_input_tokens_seen": 99996900,
      "step": 4695
    },
    {
      "epoch": 0.5646606144411712,
      "flos": 17792828682240.0,
      "grad_norm": 3.6295786913139696,
      "learning_rate": 1.6796703227459935e-06,
      "loss": 0.9656,
      "num_input_tokens_seen": 100014575,
      "step": 4696
    },
    {
      "epoch": 0.5647808573318103,
      "flos": 26078935818240.0,
      "grad_norm": 2.8226844344855504,
      "learning_rate": 1.6789014317194407e-06,
      "loss": 0.9527,
      "num_input_tokens_seen": 100035775,
      "step": 4697
    },
    {
      "epoch": 0.5649011002224493,
      "flos": 16009267322880.0,
      "grad_norm": 5.7324641315520894,
      "learning_rate": 1.6781325894032853e-06,
      "loss": 0.9151,
      "num_input_tokens_seen": 100054455,
      "step": 4698
    },
    {
      "epoch": 0.5650213431130885,
      "flos": 13124998963200.0,
      "grad_norm": 3.114315689137645,
      "learning_rate": 1.6773637959141608e-06,
      "loss": 1.1095,
      "num_input_tokens_seen": 100071150,
      "step": 4699
    },
    {
      "epoch": 0.5651415860037275,
      "flos": 12413838643200.0,
      "grad_norm": 5.169146137671341,
      "learning_rate": 1.6765950513686915e-06,
      "loss": 0.8371,
      "num_input_tokens_seen": 100088980,
      "step": 4700
    },
    {
      "epoch": 0.5652618288943666,
      "flos": 18159220039680.0,
      "grad_norm": 4.345351328808335,
      "learning_rate": 1.675826355883496e-06,
      "loss": 0.9549,
      "num_input_tokens_seen": 100107915,
      "step": 4701
    },
    {
      "epoch": 0.5653820717850057,
      "flos": 13964444590080.0,
      "grad_norm": 3.0900346429358128,
      "learning_rate": 1.6750577095751848e-06,
      "loss": 0.9935,
      "num_input_tokens_seen": 100126745,
      "step": 4702
    },
    {
      "epoch": 0.5655023146756448,
      "flos": 19208312463360.0,
      "grad_norm": 2.3846507501701093,
      "learning_rate": 1.6742891125603605e-06,
      "loss": 0.9298,
      "num_input_tokens_seen": 100147370,
      "step": 4703
    },
    {
      "epoch": 0.5656225575662839,
      "flos": 19703449006080.0,
      "grad_norm": 2.796381731063847,
      "learning_rate": 1.6735205649556185e-06,
      "loss": 0.9125,
      "num_input_tokens_seen": 100166960,
      "step": 4704
    },
    {
      "epoch": 0.5657428004569229,
      "flos": 17317781606400.0,
      "grad_norm": 2.136926058627557,
      "learning_rate": 1.6727520668775476e-06,
      "loss": 1.0373,
      "num_input_tokens_seen": 100186965,
      "step": 4705
    },
    {
      "epoch": 0.5658630433475621,
      "flos": 15589176606720.0,
      "grad_norm": 3.3170212893143525,
      "learning_rate": 1.6719836184427275e-06,
      "loss": 0.9489,
      "num_input_tokens_seen": 100206990,
      "step": 4706
    },
    {
      "epoch": 0.5659832862382012,
      "flos": 21672827351040.0,
      "grad_norm": 3.119165084828954,
      "learning_rate": 1.671215219767733e-06,
      "loss": 0.8404,
      "num_input_tokens_seen": 100226170,
      "step": 4707
    },
    {
      "epoch": 0.5661035291288402,
      "flos": 9296614871040.0,
      "grad_norm": 3.5415682311496797,
      "learning_rate": 1.670446870969127e-06,
      "loss": 0.9549,
      "num_input_tokens_seen": 100243555,
      "step": 4708
    },
    {
      "epoch": 0.5662237720194794,
      "flos": 11420131799040.0,
      "grad_norm": 4.869279096694786,
      "learning_rate": 1.6696785721634685e-06,
      "loss": 1.0026,
      "num_input_tokens_seen": 100257760,
      "step": 4709
    },
    {
      "epoch": 0.5663440149101184,
      "flos": 12522647654400.0,
      "grad_norm": 3.541752144773104,
      "learning_rate": 1.6689103234673086e-06,
      "loss": 0.9216,
      "num_input_tokens_seen": 100275800,
      "step": 4710
    },
    {
      "epoch": 0.5664642578007575,
      "flos": 16612967608320.0,
      "grad_norm": 3.057084257941912,
      "learning_rate": 1.668142124997189e-06,
      "loss": 0.9492,
      "num_input_tokens_seen": 100295180,
      "step": 4711
    },
    {
      "epoch": 0.5665845006913967,
      "flos": 46909918064640.0,
      "grad_norm": 0.7481016134050154,
      "learning_rate": 1.6673739768696453e-06,
      "loss": 0.8174,
      "num_input_tokens_seen": 100361470,
      "step": 4712
    },
    {
      "epoch": 0.5667047435820357,
      "flos": 18605501276160.0,
      "grad_norm": 4.00118786340223,
      "learning_rate": 1.6666058792012052e-06,
      "loss": 0.9568,
      "num_input_tokens_seen": 100382075,
      "step": 4713
    },
    {
      "epoch": 0.5668249864726748,
      "flos": 51474150727680.0,
      "grad_norm": 0.8584496805044136,
      "learning_rate": 1.6658378321083878e-06,
      "loss": 0.8908,
      "num_input_tokens_seen": 100446125,
      "step": 4714
    },
    {
      "epoch": 0.5669452293633139,
      "flos": 15769505710080.0,
      "grad_norm": 2.6451367540106907,
      "learning_rate": 1.6650698357077055e-06,
      "loss": 1.0193,
      "num_input_tokens_seen": 100462890,
      "step": 4715
    },
    {
      "epoch": 0.567065472253953,
      "flos": 12915168215040.0,
      "grad_norm": 6.677678962917214,
      "learning_rate": 1.6643018901156632e-06,
      "loss": 1.0056,
      "num_input_tokens_seen": 100481705,
      "step": 4716
    },
    {
      "epoch": 0.567185715144592,
      "flos": 14458569400320.0,
      "grad_norm": 3.5941555955307853,
      "learning_rate": 1.6635339954487566e-06,
      "loss": 0.9796,
      "num_input_tokens_seen": 100497300,
      "step": 4717
    },
    {
      "epoch": 0.5673059580352312,
      "flos": 16507193794560.0,
      "grad_norm": 2.6460789443830333,
      "learning_rate": 1.6627661518234765e-06,
      "loss": 1.0159,
      "num_input_tokens_seen": 100516275,
      "step": 4718
    },
    {
      "epoch": 0.5674262009258703,
      "flos": 15427128115200.0,
      "grad_norm": 2.5883055386726346,
      "learning_rate": 1.661998359356302e-06,
      "loss": 1.0435,
      "num_input_tokens_seen": 100535025,
      "step": 4719
    },
    {
      "epoch": 0.5675464438165093,
      "flos": 48311942737920.0,
      "grad_norm": 0.8127576346178159,
      "learning_rate": 1.6612306181637077e-06,
      "loss": 0.7741,
      "num_input_tokens_seen": 100594070,
      "step": 4720
    },
    {
      "epoch": 0.5676666867071485,
      "flos": 13386505605120.0,
      "grad_norm": 6.448239451408817,
      "learning_rate": 1.6604629283621598e-06,
      "loss": 0.8434,
      "num_input_tokens_seen": 100611720,
      "step": 4721
    },
    {
      "epoch": 0.5677869295977875,
      "flos": 23927879393280.0,
      "grad_norm": 3.0489097647821883,
      "learning_rate": 1.6596952900681152e-06,
      "loss": 0.934,
      "num_input_tokens_seen": 100632200,
      "step": 4722
    },
    {
      "epoch": 0.5679071724884266,
      "flos": 19968266772480.0,
      "grad_norm": 6.39453244311229,
      "learning_rate": 1.658927703398025e-06,
      "loss": 1.0151,
      "num_input_tokens_seen": 100651985,
      "step": 4723
    },
    {
      "epoch": 0.5680274153790658,
      "flos": 16743521648640.0,
      "grad_norm": 4.871508986724674,
      "learning_rate": 1.6581601684683309e-06,
      "loss": 0.9546,
      "num_input_tokens_seen": 100672130,
      "step": 4724
    },
    {
      "epoch": 0.5681476582697048,
      "flos": 15956549038080.0,
      "grad_norm": 3.858253346898695,
      "learning_rate": 1.6573926853954674e-06,
      "loss": 0.8785,
      "num_input_tokens_seen": 100689435,
      "step": 4725
    },
    {
      "epoch": 0.5682679011603439,
      "flos": 13859345264640.0,
      "grad_norm": 3.1736569965884196,
      "learning_rate": 1.6566252542958608e-06,
      "loss": 1.0184,
      "num_input_tokens_seen": 100708655,
      "step": 4726
    },
    {
      "epoch": 0.568388144050983,
      "flos": 20491433349120.0,
      "grad_norm": 3.5966249327643856,
      "learning_rate": 1.6558578752859305e-06,
      "loss": 0.9635,
      "num_input_tokens_seen": 100727335,
      "step": 4727
    },
    {
      "epoch": 0.5685083869416221,
      "flos": 15060583464960.0,
      "grad_norm": 3.8963094954098545,
      "learning_rate": 1.6550905484820865e-06,
      "loss": 0.9805,
      "num_input_tokens_seen": 100745515,
      "step": 4728
    },
    {
      "epoch": 0.5686286298322611,
      "flos": 17661201592320.0,
      "grad_norm": 5.694759556598567,
      "learning_rate": 1.6543232740007328e-06,
      "loss": 0.991,
      "num_input_tokens_seen": 100762350,
      "step": 4729
    },
    {
      "epoch": 0.5687488727229003,
      "flos": 18946315284480.0,
      "grad_norm": 4.154763551082693,
      "learning_rate": 1.653556051958263e-06,
      "loss": 0.8457,
      "num_input_tokens_seen": 100781750,
      "step": 4730
    },
    {
      "epoch": 0.5688691156135394,
      "flos": 14772671692800.0,
      "grad_norm": 3.0766100568374757,
      "learning_rate": 1.6527888824710642e-06,
      "loss": 0.9352,
      "num_input_tokens_seen": 100801070,
      "step": 4731
    },
    {
      "epoch": 0.5689893585041784,
      "flos": 18418090045440.0,
      "grad_norm": 4.421201620961586,
      "learning_rate": 1.6520217656555166e-06,
      "loss": 0.9403,
      "num_input_tokens_seen": 100820080,
      "step": 4732
    },
    {
      "epoch": 0.5691096013948175,
      "flos": 16691355217920.0,
      "grad_norm": 2.5125120817434854,
      "learning_rate": 1.65125470162799e-06,
      "loss": 0.8991,
      "num_input_tokens_seen": 100840155,
      "step": 4733
    },
    {
      "epoch": 0.5692298442854566,
      "flos": 12808873205760.0,
      "grad_norm": 4.979065940614887,
      "learning_rate": 1.6504876905048485e-06,
      "loss": 0.8931,
      "num_input_tokens_seen": 100856835,
      "step": 4734
    },
    {
      "epoch": 0.5693500871760957,
      "flos": 16376425144320.0,
      "grad_norm": 2.8972440252055667,
      "learning_rate": 1.6497207324024464e-06,
      "loss": 0.9278,
      "num_input_tokens_seen": 100875455,
      "step": 4735
    },
    {
      "epoch": 0.5694703300667348,
      "flos": 13466058240000.0,
      "grad_norm": 2.8014198756254944,
      "learning_rate": 1.6489538274371305e-06,
      "loss": 1.0175,
      "num_input_tokens_seen": 100893780,
      "step": 4736
    },
    {
      "epoch": 0.5695905729573739,
      "flos": 15563936501760.0,
      "grad_norm": 3.433888289281432,
      "learning_rate": 1.6481869757252396e-06,
      "loss": 1.024,
      "num_input_tokens_seen": 100911835,
      "step": 4737
    },
    {
      "epoch": 0.569710815848013,
      "flos": 20284269895680.0,
      "grad_norm": 3.3810026688293013,
      "learning_rate": 1.647420177383105e-06,
      "loss": 0.9116,
      "num_input_tokens_seen": 100934425,
      "step": 4738
    },
    {
      "epoch": 0.569831058738652,
      "flos": 20204717260800.0,
      "grad_norm": 3.3957145217674416,
      "learning_rate": 1.646653432527049e-06,
      "loss": 0.9242,
      "num_input_tokens_seen": 100954785,
      "step": 4739
    },
    {
      "epoch": 0.5699513016292912,
      "flos": 18396559626240.0,
      "grad_norm": 6.3601703568201255,
      "learning_rate": 1.645886741273387e-06,
      "loss": 0.9444,
      "num_input_tokens_seen": 100976320,
      "step": 4740
    },
    {
      "epoch": 0.5700715445199303,
      "flos": 12781578977280.0,
      "grad_norm": 6.345450638444179,
      "learning_rate": 1.645120103738424e-06,
      "loss": 0.9239,
      "num_input_tokens_seen": 100993550,
      "step": 4741
    },
    {
      "epoch": 0.5701917874105693,
      "flos": 8064188805120.0,
      "grad_norm": 6.374593385182777,
      "learning_rate": 1.6443535200384591e-06,
      "loss": 1.0245,
      "num_input_tokens_seen": 101011445,
      "step": 4742
    },
    {
      "epoch": 0.5703120303012085,
      "flos": 15458193346560.0,
      "grad_norm": 4.0642837129818306,
      "learning_rate": 1.6435869902897827e-06,
      "loss": 0.8988,
      "num_input_tokens_seen": 101029745,
      "step": 4743
    },
    {
      "epoch": 0.5704322731918475,
      "flos": 40605092904960.0,
      "grad_norm": 0.8043344334211334,
      "learning_rate": 1.6428205146086764e-06,
      "loss": 0.8295,
      "num_input_tokens_seen": 101091445,
      "step": 4744
    },
    {
      "epoch": 0.5705525160824866,
      "flos": 14724828119040.0,
      "grad_norm": 3.1224855606808104,
      "learning_rate": 1.6420540931114142e-06,
      "loss": 0.9052,
      "num_input_tokens_seen": 101111755,
      "step": 4745
    },
    {
      "epoch": 0.5706727589731257,
      "flos": 13309160386560.0,
      "grad_norm": 2.5104238634019533,
      "learning_rate": 1.6412877259142616e-06,
      "loss": 0.9799,
      "num_input_tokens_seen": 101131395,
      "step": 4746
    },
    {
      "epoch": 0.5707930018637648,
      "flos": 19678822072320.0,
      "grad_norm": 8.632682585278117,
      "learning_rate": 1.6405214131334757e-06,
      "loss": 0.9365,
      "num_input_tokens_seen": 101149640,
      "step": 4747
    },
    {
      "epoch": 0.5709132447544039,
      "flos": 19652110356480.0,
      "grad_norm": 3.8745930365698595,
      "learning_rate": 1.6397551548853052e-06,
      "loss": 0.9957,
      "num_input_tokens_seen": 101167525,
      "step": 4748
    },
    {
      "epoch": 0.571033487645043,
      "flos": 15403451596800.0,
      "grad_norm": 9.305262678354845,
      "learning_rate": 1.6389889512859917e-06,
      "loss": 0.8999,
      "num_input_tokens_seen": 101186905,
      "step": 4749
    },
    {
      "epoch": 0.5711537305356821,
      "flos": 50261844787200.0,
      "grad_norm": 0.8825670762006578,
      "learning_rate": 1.638222802451767e-06,
      "loss": 0.8508,
      "num_input_tokens_seen": 101248105,
      "step": 4750
    },
    {
      "epoch": 0.5712739734263211,
      "flos": 17582476738560.0,
      "grad_norm": 2.888905936785876,
      "learning_rate": 1.6374567084988561e-06,
      "loss": 0.9373,
      "num_input_tokens_seen": 101269010,
      "step": 4751
    },
    {
      "epoch": 0.5713942163169603,
      "flos": 18919695544320.0,
      "grad_norm": 3.0509827140686014,
      "learning_rate": 1.6366906695434738e-06,
      "loss": 0.9596,
      "num_input_tokens_seen": 101291250,
      "step": 4752
    },
    {
      "epoch": 0.5715144592075994,
      "flos": 15013230428160.0,
      "grad_norm": 3.7078390899360656,
      "learning_rate": 1.6359246857018275e-06,
      "loss": 1.0464,
      "num_input_tokens_seen": 101308500,
      "step": 4753
    },
    {
      "epoch": 0.5716347020982384,
      "flos": 16585397452800.0,
      "grad_norm": 17.219920594567302,
      "learning_rate": 1.6351587570901178e-06,
      "loss": 0.9738,
      "num_input_tokens_seen": 101328345,
      "step": 4754
    },
    {
      "epoch": 0.5717549449888776,
      "flos": 12041652817920.0,
      "grad_norm": 4.535018931962339,
      "learning_rate": 1.634392883824534e-06,
      "loss": 0.9478,
      "num_input_tokens_seen": 101340065,
      "step": 4755
    },
    {
      "epoch": 0.5718751878795166,
      "flos": 25344804126720.0,
      "grad_norm": 3.674460551834087,
      "learning_rate": 1.6336270660212595e-06,
      "loss": 0.8638,
      "num_input_tokens_seen": 101361380,
      "step": 4756
    },
    {
      "epoch": 0.5719954307701557,
      "flos": 27569005547520.0,
      "grad_norm": 3.6773826037871324,
      "learning_rate": 1.6328613037964676e-06,
      "loss": 0.8462,
      "num_input_tokens_seen": 101384165,
      "step": 4757
    },
    {
      "epoch": 0.5721156736607949,
      "flos": 14645674045440.0,
      "grad_norm": 3.7300952344912086,
      "learning_rate": 1.6320955972663241e-06,
      "loss": 0.8784,
      "num_input_tokens_seen": 101403480,
      "step": 4758
    },
    {
      "epoch": 0.5722359165514339,
      "flos": 26708060160000.0,
      "grad_norm": 10.698400459877277,
      "learning_rate": 1.6313299465469857e-06,
      "loss": 0.8376,
      "num_input_tokens_seen": 101425930,
      "step": 4759
    },
    {
      "epoch": 0.572356159442073,
      "flos": 15608867512320.0,
      "grad_norm": 4.92541880109813,
      "learning_rate": 1.6305643517546014e-06,
      "loss": 0.9726,
      "num_input_tokens_seen": 101441030,
      "step": 4760
    },
    {
      "epoch": 0.5724764023327121,
      "flos": 13570881638400.0,
      "grad_norm": 3.413178143872523,
      "learning_rate": 1.629798813005311e-06,
      "loss": 1.0436,
      "num_input_tokens_seen": 101460470,
      "step": 4761
    },
    {
      "epoch": 0.5725966452233512,
      "flos": 16218546216960.0,
      "grad_norm": 3.6061275785350273,
      "learning_rate": 1.6290333304152473e-06,
      "loss": 0.9012,
      "num_input_tokens_seen": 101480065,
      "step": 4762
    },
    {
      "epoch": 0.5727168881139902,
      "flos": 29640723947520.0,
      "grad_norm": 4.62793061481184,
      "learning_rate": 1.6282679041005314e-06,
      "loss": 0.7674,
      "num_input_tokens_seen": 101505375,
      "step": 4763
    },
    {
      "epoch": 0.5728371310046293,
      "flos": 10345523343360.0,
      "grad_norm": 8.41356150942924,
      "learning_rate": 1.6275025341772789e-06,
      "loss": 1.0706,
      "num_input_tokens_seen": 101521400,
      "step": 4764
    },
    {
      "epoch": 0.5729573738952685,
      "flos": 15274859704320.0,
      "grad_norm": 6.491640143611658,
      "learning_rate": 1.626737220761596e-06,
      "loss": 1.0208,
      "num_input_tokens_seen": 101538585,
      "step": 4765
    },
    {
      "epoch": 0.5730776167859075,
      "flos": 16794768322560.0,
      "grad_norm": 5.996746452667323,
      "learning_rate": 1.62597196396958e-06,
      "loss": 0.9827,
      "num_input_tokens_seen": 101556475,
      "step": 4766
    },
    {
      "epoch": 0.5731978596765466,
      "flos": 18316792381440.0,
      "grad_norm": 7.194972126517738,
      "learning_rate": 1.6252067639173197e-06,
      "loss": 1.0458,
      "num_input_tokens_seen": 101578105,
      "step": 4767
    },
    {
      "epoch": 0.5733181025671857,
      "flos": 18762583080960.0,
      "grad_norm": 3.18259561255728,
      "learning_rate": 1.6244416207208956e-06,
      "loss": 0.892,
      "num_input_tokens_seen": 101598760,
      "step": 4768
    },
    {
      "epoch": 0.5734383454578248,
      "flos": 20964824862720.0,
      "grad_norm": 5.106772970548964,
      "learning_rate": 1.6236765344963787e-06,
      "loss": 0.9348,
      "num_input_tokens_seen": 101619740,
      "step": 4769
    },
    {
      "epoch": 0.5735585883484638,
      "flos": 24950719979520.0,
      "grad_norm": 3.468588060602653,
      "learning_rate": 1.6229115053598322e-06,
      "loss": 0.8883,
      "num_input_tokens_seen": 101641215,
      "step": 4770
    },
    {
      "epoch": 0.573678831239103,
      "flos": 13308731166720.0,
      "grad_norm": 3.042854355900983,
      "learning_rate": 1.6221465334273108e-06,
      "loss": 0.9005,
      "num_input_tokens_seen": 101660145,
      "step": 4771
    },
    {
      "epoch": 0.5737990741297421,
      "flos": 18290693836800.0,
      "grad_norm": 4.685552370450612,
      "learning_rate": 1.6213816188148593e-06,
      "loss": 0.8012,
      "num_input_tokens_seen": 101678570,
      "step": 4772
    },
    {
      "epoch": 0.5739193170203811,
      "flos": 19416058429440.0,
      "grad_norm": 2.2198197915419593,
      "learning_rate": 1.6206167616385162e-06,
      "loss": 0.9655,
      "num_input_tokens_seen": 101699355,
      "step": 4773
    },
    {
      "epoch": 0.5740395599110203,
      "flos": 8614986854400.0,
      "grad_norm": 3.724938636632782,
      "learning_rate": 1.6198519620143078e-06,
      "loss": 0.9411,
      "num_input_tokens_seen": 101716230,
      "step": 4774
    },
    {
      "epoch": 0.5741598028016593,
      "flos": 18447683665920.0,
      "grad_norm": 2.5964732753378104,
      "learning_rate": 1.6190872200582546e-06,
      "loss": 0.9706,
      "num_input_tokens_seen": 101737690,
      "step": 4775
    },
    {
      "epoch": 0.5742800456922984,
      "flos": 13648809369600.0,
      "grad_norm": 8.551680139910053,
      "learning_rate": 1.6183225358863676e-06,
      "loss": 0.9751,
      "num_input_tokens_seen": 101754305,
      "step": 4776
    },
    {
      "epoch": 0.5744002885829376,
      "flos": 22040506368000.0,
      "grad_norm": 5.5778235773449625,
      "learning_rate": 1.617557909614648e-06,
      "loss": 0.9151,
      "num_input_tokens_seen": 101773460,
      "step": 4777
    },
    {
      "epoch": 0.5745205314735766,
      "flos": 16951604858880.0,
      "grad_norm": 2.8552937382243124,
      "learning_rate": 1.6167933413590899e-06,
      "loss": 1.0459,
      "num_input_tokens_seen": 101792085,
      "step": 4778
    },
    {
      "epoch": 0.5746407743642157,
      "flos": 8666754723840.0,
      "grad_norm": 3.717092449164369,
      "learning_rate": 1.6160288312356773e-06,
      "loss": 1.1008,
      "num_input_tokens_seen": 101808935,
      "step": 4779
    },
    {
      "epoch": 0.5747610172548548,
      "flos": 17162876559360.0,
      "grad_norm": 2.8827349231071966,
      "learning_rate": 1.6152643793603857e-06,
      "loss": 1.0211,
      "num_input_tokens_seen": 101829005,
      "step": 4780
    },
    {
      "epoch": 0.5748812601454939,
      "flos": 18078747648000.0,
      "grad_norm": 6.004612739012218,
      "learning_rate": 1.6144999858491815e-06,
      "loss": 1.079,
      "num_input_tokens_seen": 101847355,
      "step": 4781
    },
    {
      "epoch": 0.575001503036133,
      "flos": 21827609763840.0,
      "grad_norm": 3.5493866350292578,
      "learning_rate": 1.6137356508180232e-06,
      "loss": 1.0476,
      "num_input_tokens_seen": 101868785,
      "step": 4782
    },
    {
      "epoch": 0.5751217459267721,
      "flos": 15433075875840.0,
      "grad_norm": 2.935661714684436,
      "learning_rate": 1.6129713743828593e-06,
      "loss": 1.0136,
      "num_input_tokens_seen": 101887515,
      "step": 4783
    },
    {
      "epoch": 0.5752419888174112,
      "flos": 15378947297280.0,
      "grad_norm": 3.2873984723677,
      "learning_rate": 1.6122071566596306e-06,
      "loss": 0.9528,
      "num_input_tokens_seen": 101907510,
      "step": 4784
    },
    {
      "epoch": 0.5753622317080502,
      "flos": 12592512184320.0,
      "grad_norm": 4.679447632821146,
      "learning_rate": 1.6114429977642674e-06,
      "loss": 1.0282,
      "num_input_tokens_seen": 101921735,
      "step": 4785
    },
    {
      "epoch": 0.5754824745986894,
      "flos": 14040594124800.0,
      "grad_norm": 2.7072760841436856,
      "learning_rate": 1.6106788978126926e-06,
      "loss": 0.9232,
      "num_input_tokens_seen": 101940430,
      "step": 4786
    },
    {
      "epoch": 0.5756027174893285,
      "flos": 22086295818240.0,
      "grad_norm": 26.01137352648597,
      "learning_rate": 1.6099148569208196e-06,
      "loss": 0.9705,
      "num_input_tokens_seen": 101957370,
      "step": 4787
    },
    {
      "epoch": 0.5757229603799675,
      "flos": 20334351544320.0,
      "grad_norm": 3.5588398920516644,
      "learning_rate": 1.6091508752045523e-06,
      "loss": 0.8216,
      "num_input_tokens_seen": 101977970,
      "step": 4788
    },
    {
      "epoch": 0.5758432032706067,
      "flos": 16346954158080.0,
      "grad_norm": 2.3650882086726166,
      "learning_rate": 1.608386952779787e-06,
      "loss": 1.0626,
      "num_input_tokens_seen": 101997060,
      "step": 4789
    },
    {
      "epoch": 0.5759634461612457,
      "flos": 18317221601280.0,
      "grad_norm": 2.4311599440186735,
      "learning_rate": 1.6076230897624098e-06,
      "loss": 0.9425,
      "num_input_tokens_seen": 102018985,
      "step": 4790
    },
    {
      "epoch": 0.5760836890518848,
      "flos": 21804209172480.0,
      "grad_norm": 4.600886400484804,
      "learning_rate": 1.6068592862682974e-06,
      "loss": 0.9741,
      "num_input_tokens_seen": 102036860,
      "step": 4791
    },
    {
      "epoch": 0.576203931942524,
      "flos": 26078353305600.0,
      "grad_norm": 3.901850794298993,
      "learning_rate": 1.6060955424133187e-06,
      "loss": 0.9261,
      "num_input_tokens_seen": 102057505,
      "step": 4792
    },
    {
      "epoch": 0.576324174833163,
      "flos": 18156583403520.0,
      "grad_norm": 5.8762575792086436,
      "learning_rate": 1.6053318583133332e-06,
      "loss": 1.085,
      "num_input_tokens_seen": 102078095,
      "step": 4793
    },
    {
      "epoch": 0.5764444177238021,
      "flos": 17871277608960.0,
      "grad_norm": 3.537712462821773,
      "learning_rate": 1.6045682340841907e-06,
      "loss": 0.9475,
      "num_input_tokens_seen": 102096740,
      "step": 4794
    },
    {
      "epoch": 0.5765646606144411,
      "flos": 44532843171840.0,
      "grad_norm": 0.8067698894681554,
      "learning_rate": 1.6038046698417336e-06,
      "loss": 0.798,
      "num_input_tokens_seen": 102157355,
      "step": 4795
    },
    {
      "epoch": 0.5766849035050803,
      "flos": 17870848389120.0,
      "grad_norm": 4.324253816126451,
      "learning_rate": 1.6030411657017919e-06,
      "loss": 0.874,
      "num_input_tokens_seen": 102176730,
      "step": 4796
    },
    {
      "epoch": 0.5768051463957193,
      "flos": 11310127104000.0,
      "grad_norm": 2.869028108700433,
      "learning_rate": 1.6022777217801903e-06,
      "loss": 1.0466,
      "num_input_tokens_seen": 102193405,
      "step": 4797
    },
    {
      "epoch": 0.5769253892863584,
      "flos": 15747086192640.0,
      "grad_norm": 2.450000400932435,
      "learning_rate": 1.601514338192742e-06,
      "loss": 0.9241,
      "num_input_tokens_seen": 102213055,
      "step": 4798
    },
    {
      "epoch": 0.5770456321769976,
      "flos": 16244890030080.0,
      "grad_norm": 3.9176692220609803,
      "learning_rate": 1.6007510150552514e-06,
      "loss": 0.9072,
      "num_input_tokens_seen": 102230835,
      "step": 4799
    },
    {
      "epoch": 0.5771658750676366,
      "flos": 33132003717120.0,
      "grad_norm": 2.8290182989559534,
      "learning_rate": 1.599987752483515e-06,
      "loss": 0.8172,
      "num_input_tokens_seen": 102255000,
      "step": 4800
    },
    {
      "epoch": 0.5772861179582757,
      "flos": 15743959019520.0,
      "grad_norm": 3.1367681554036464,
      "learning_rate": 1.5992245505933184e-06,
      "loss": 0.8782,
      "num_input_tokens_seen": 102274420,
      "step": 4801
    },
    {
      "epoch": 0.5774063608489148,
      "flos": 22276558295040.0,
      "grad_norm": 3.1888644720007426,
      "learning_rate": 1.5984614095004388e-06,
      "loss": 0.9061,
      "num_input_tokens_seen": 102295275,
      "step": 4802
    },
    {
      "epoch": 0.5775266037395539,
      "flos": 16008132956160.0,
      "grad_norm": 3.5601923646281723,
      "learning_rate": 1.5976983293206438e-06,
      "loss": 0.9986,
      "num_input_tokens_seen": 102310800,
      "step": 4803
    },
    {
      "epoch": 0.577646846630193,
      "flos": 14960205557760.0,
      "grad_norm": 3.9186260246366293,
      "learning_rate": 1.5969353101696928e-06,
      "loss": 0.9095,
      "num_input_tokens_seen": 102328960,
      "step": 4804
    },
    {
      "epoch": 0.5777670895208321,
      "flos": 21173337292800.0,
      "grad_norm": 2.7230079894353163,
      "learning_rate": 1.5961723521633341e-06,
      "loss": 0.9959,
      "num_input_tokens_seen": 102349920,
      "step": 4805
    },
    {
      "epoch": 0.5778873324114712,
      "flos": 13833124085760.0,
      "grad_norm": 4.074634914777114,
      "learning_rate": 1.5954094554173097e-06,
      "loss": 1.1089,
      "num_input_tokens_seen": 102367630,
      "step": 4806
    },
    {
      "epoch": 0.5780075753021102,
      "flos": 9978396180480.0,
      "grad_norm": 4.665231120608218,
      "learning_rate": 1.5946466200473482e-06,
      "loss": 0.9896,
      "num_input_tokens_seen": 102385260,
      "step": 4807
    },
    {
      "epoch": 0.5781278181927494,
      "flos": 10786378014720.0,
      "grad_norm": 3.158095320806174,
      "learning_rate": 1.5938838461691723e-06,
      "loss": 1.0345,
      "num_input_tokens_seen": 102401890,
      "step": 4808
    },
    {
      "epoch": 0.5782480610833884,
      "flos": 11839701319680.0,
      "grad_norm": 9.23647040590084,
      "learning_rate": 1.593121133898494e-06,
      "loss": 1.0123,
      "num_input_tokens_seen": 102418815,
      "step": 4809
    },
    {
      "epoch": 0.5783683039740275,
      "flos": 18131803176960.0,
      "grad_norm": 6.70973999565752,
      "learning_rate": 1.592358483351016e-06,
      "loss": 0.9795,
      "num_input_tokens_seen": 102438710,
      "step": 4810
    },
    {
      "epoch": 0.5784885468646667,
      "flos": 13046458060800.0,
      "grad_norm": 3.604791678918614,
      "learning_rate": 1.5915958946424326e-06,
      "loss": 0.9207,
      "num_input_tokens_seen": 102457115,
      "step": 4811
    },
    {
      "epoch": 0.5786087897553057,
      "flos": 32950540247040.0,
      "grad_norm": 3.1312533563226617,
      "learning_rate": 1.5908333678884271e-06,
      "loss": 0.9403,
      "num_input_tokens_seen": 102483255,
      "step": 4812
    },
    {
      "epoch": 0.5787290326459448,
      "flos": 8720116838400.0,
      "grad_norm": 3.133590183146655,
      "learning_rate": 1.5900709032046743e-06,
      "loss": 0.9267,
      "num_input_tokens_seen": 102501050,
      "step": 4813
    },
    {
      "epoch": 0.5788492755365839,
      "flos": 16556233052160.0,
      "grad_norm": 3.192591023532842,
      "learning_rate": 1.5893085007068391e-06,
      "loss": 0.9828,
      "num_input_tokens_seen": 102518330,
      "step": 4814
    },
    {
      "epoch": 0.578969518427223,
      "flos": 17110740787200.0,
      "grad_norm": 3.1375877155337757,
      "learning_rate": 1.5885461605105786e-06,
      "loss": 0.9059,
      "num_input_tokens_seen": 102539650,
      "step": 4815
    },
    {
      "epoch": 0.579089761317862,
      "flos": 15537378078720.0,
      "grad_norm": 3.1312437624688374,
      "learning_rate": 1.5877838827315375e-06,
      "loss": 0.9628,
      "num_input_tokens_seen": 102557915,
      "step": 4816
    },
    {
      "epoch": 0.5792100042085012,
      "flos": 16297608314880.0,
      "grad_norm": 3.211390553832655,
      "learning_rate": 1.587021667485355e-06,
      "loss": 0.8863,
      "num_input_tokens_seen": 102577005,
      "step": 4817
    },
    {
      "epoch": 0.5793302470991403,
      "flos": 15248577208320.0,
      "grad_norm": 3.683835423168059,
      "learning_rate": 1.5862595148876559e-06,
      "loss": 0.9743,
      "num_input_tokens_seen": 102596830,
      "step": 4818
    },
    {
      "epoch": 0.5794504899897793,
      "flos": 8953286860800.0,
      "grad_norm": 4.604032879895112,
      "learning_rate": 1.58549742505406e-06,
      "loss": 0.9646,
      "num_input_tokens_seen": 102611295,
      "step": 4819
    },
    {
      "epoch": 0.5795707328804185,
      "flos": 10503248977920.0,
      "grad_norm": 51.01785265044482,
      "learning_rate": 1.5847353981001747e-06,
      "loss": 0.9434,
      "num_input_tokens_seen": 102628195,
      "step": 4820
    },
    {
      "epoch": 0.5796909757710575,
      "flos": 26000149647360.0,
      "grad_norm": 2.9003697455582635,
      "learning_rate": 1.5839734341415993e-06,
      "loss": 0.8878,
      "num_input_tokens_seen": 102650115,
      "step": 4821
    },
    {
      "epoch": 0.5798112186616966,
      "flos": 16376915681280.0,
      "grad_norm": 3.2353119161187656,
      "learning_rate": 1.5832115332939238e-06,
      "loss": 0.9644,
      "num_input_tokens_seen": 102668275,
      "step": 4822
    },
    {
      "epoch": 0.5799314615523358,
      "flos": 11786185912320.0,
      "grad_norm": 3.0629458978640463,
      "learning_rate": 1.5824496956727272e-06,
      "loss": 0.9448,
      "num_input_tokens_seen": 102685200,
      "step": 4823
    },
    {
      "epoch": 0.5800517044429748,
      "flos": 14541218549760.0,
      "grad_norm": 3.934001276447533,
      "learning_rate": 1.5816879213935797e-06,
      "loss": 0.9161,
      "num_input_tokens_seen": 102703730,
      "step": 4824
    },
    {
      "epoch": 0.5801719473336139,
      "flos": 22484150968320.0,
      "grad_norm": 3.0581609388056603,
      "learning_rate": 1.5809262105720416e-06,
      "loss": 0.986,
      "num_input_tokens_seen": 102724490,
      "step": 4825
    },
    {
      "epoch": 0.580292190224253,
      "flos": 14462340403200.0,
      "grad_norm": 2.204941778300965,
      "learning_rate": 1.5801645633236644e-06,
      "loss": 0.9861,
      "num_input_tokens_seen": 102745195,
      "step": 4826
    },
    {
      "epoch": 0.5804124331148921,
      "flos": 18946529894400.0,
      "grad_norm": 3.6069373945022787,
      "learning_rate": 1.579402979763989e-06,
      "loss": 0.9717,
      "num_input_tokens_seen": 102765250,
      "step": 4827
    },
    {
      "epoch": 0.5805326760055312,
      "flos": 9505035325440.0,
      "grad_norm": 6.180420187087998,
      "learning_rate": 1.578641460008548e-06,
      "loss": 0.9924,
      "num_input_tokens_seen": 102782705,
      "step": 4828
    },
    {
      "epoch": 0.5806529188961702,
      "flos": 8508048015360.0,
      "grad_norm": 5.602152390324539,
      "learning_rate": 1.5778800041728613e-06,
      "loss": 0.8522,
      "num_input_tokens_seen": 102798715,
      "step": 4829
    },
    {
      "epoch": 0.5807731617868094,
      "flos": 18658158243840.0,
      "grad_norm": 20.84301793129716,
      "learning_rate": 1.577118612372443e-06,
      "loss": 0.8524,
      "num_input_tokens_seen": 102820275,
      "step": 4830
    },
    {
      "epoch": 0.5808934046774484,
      "flos": 27101193891840.0,
      "grad_norm": 4.302691375891591,
      "learning_rate": 1.5763572847227943e-06,
      "loss": 0.8886,
      "num_input_tokens_seen": 102840880,
      "step": 4831
    },
    {
      "epoch": 0.5810136475680875,
      "flos": 14541371842560.0,
      "grad_norm": 3.4462573773426546,
      "learning_rate": 1.5755960213394091e-06,
      "loss": 0.9962,
      "num_input_tokens_seen": 102857700,
      "step": 4832
    },
    {
      "epoch": 0.5811338904587267,
      "flos": 12417119109120.0,
      "grad_norm": 3.134845330957085,
      "learning_rate": 1.5748348223377703e-06,
      "loss": 0.9706,
      "num_input_tokens_seen": 102874975,
      "step": 4833
    },
    {
      "epoch": 0.5812541333493657,
      "flos": 13801844244480.0,
      "grad_norm": 3.4105635462043153,
      "learning_rate": 1.5740736878333507e-06,
      "loss": 0.974,
      "num_input_tokens_seen": 102892535,
      "step": 4834
    },
    {
      "epoch": 0.5813743762400048,
      "flos": 14619728793600.0,
      "grad_norm": 3.6180320066994245,
      "learning_rate": 1.5733126179416143e-06,
      "loss": 0.9754,
      "num_input_tokens_seen": 102906740,
      "step": 4835
    },
    {
      "epoch": 0.5814946191306439,
      "flos": 23662509772800.0,
      "grad_norm": 5.262475042216,
      "learning_rate": 1.5725516127780137e-06,
      "loss": 0.9064,
      "num_input_tokens_seen": 102928595,
      "step": 4836
    },
    {
      "epoch": 0.581614862021283,
      "flos": 11420346408960.0,
      "grad_norm": 4.749579762390724,
      "learning_rate": 1.5717906724579943e-06,
      "loss": 1.0645,
      "num_input_tokens_seen": 102945375,
      "step": 4837
    },
    {
      "epoch": 0.581735104911922,
      "flos": 24111826206720.0,
      "grad_norm": 4.346085174568544,
      "learning_rate": 1.571029797096989e-06,
      "loss": 0.8678,
      "num_input_tokens_seen": 102966200,
      "step": 4838
    },
    {
      "epoch": 0.5818553478025612,
      "flos": 16586163916800.0,
      "grad_norm": 3.8668281773565063,
      "learning_rate": 1.570268986810423e-06,
      "loss": 0.9758,
      "num_input_tokens_seen": 102985815,
      "step": 4839
    },
    {
      "epoch": 0.5819755906932003,
      "flos": 14908039127040.0,
      "grad_norm": 3.0941268229023215,
      "learning_rate": 1.5695082417137096e-06,
      "loss": 0.9505,
      "num_input_tokens_seen": 103003410,
      "step": 4840
    },
    {
      "epoch": 0.5820958335838393,
      "flos": 15220731125760.0,
      "grad_norm": 3.0031948561785407,
      "learning_rate": 1.5687475619222539e-06,
      "loss": 0.9478,
      "num_input_tokens_seen": 103023085,
      "step": 4841
    },
    {
      "epoch": 0.5822160764744785,
      "flos": 12731344035840.0,
      "grad_norm": 11.524131335044032,
      "learning_rate": 1.5679869475514496e-06,
      "loss": 0.9327,
      "num_input_tokens_seen": 103039740,
      "step": 4842
    },
    {
      "epoch": 0.5823363193651175,
      "flos": 16371765043200.0,
      "grad_norm": 5.888224874203553,
      "learning_rate": 1.567226398716682e-06,
      "loss": 0.9984,
      "num_input_tokens_seen": 103059375,
      "step": 4843
    },
    {
      "epoch": 0.5824565622557566,
      "flos": 23457431101440.0,
      "grad_norm": 3.431674668344584,
      "learning_rate": 1.566465915533326e-06,
      "loss": 0.8225,
      "num_input_tokens_seen": 103081125,
      "step": 4844
    },
    {
      "epoch": 0.5825768051463958,
      "flos": 15793611448320.0,
      "grad_norm": 12.464068841472312,
      "learning_rate": 1.5657054981167458e-06,
      "loss": 1.0696,
      "num_input_tokens_seen": 103099740,
      "step": 4845
    },
    {
      "epoch": 0.5826970480370348,
      "flos": 19942720081920.0,
      "grad_norm": 4.192142020258567,
      "learning_rate": 1.5649451465822965e-06,
      "loss": 0.8733,
      "num_input_tokens_seen": 103120850,
      "step": 4846
    },
    {
      "epoch": 0.5828172909276739,
      "flos": 12652711157760.0,
      "grad_norm": 3.6769374281741602,
      "learning_rate": 1.5641848610453218e-06,
      "loss": 1.0268,
      "num_input_tokens_seen": 103139230,
      "step": 4847
    },
    {
      "epoch": 0.582937533818313,
      "flos": 14095305216000.0,
      "grad_norm": 3.584653803284816,
      "learning_rate": 1.563424641621158e-06,
      "loss": 1.0482,
      "num_input_tokens_seen": 103158130,
      "step": 4848
    },
    {
      "epoch": 0.5830577767089521,
      "flos": 19130047488000.0,
      "grad_norm": 4.275299838075025,
      "learning_rate": 1.5626644884251282e-06,
      "loss": 0.8926,
      "num_input_tokens_seen": 103177370,
      "step": 4849
    },
    {
      "epoch": 0.5831780195995911,
      "flos": 17999409623040.0,
      "grad_norm": 4.185409479560713,
      "learning_rate": 1.5619044015725488e-06,
      "loss": 1.0762,
      "num_input_tokens_seen": 103196780,
      "step": 4850
    },
    {
      "epoch": 0.5832982624902303,
      "flos": 10424064245760.0,
      "grad_norm": 12.957761369616051,
      "learning_rate": 1.5611443811787224e-06,
      "loss": 1.0566,
      "num_input_tokens_seen": 103210625,
      "step": 4851
    },
    {
      "epoch": 0.5834185053808694,
      "flos": 14511042416640.0,
      "grad_norm": 3.9377350509967206,
      "learning_rate": 1.560384427358945e-06,
      "loss": 0.8973,
      "num_input_tokens_seen": 103229890,
      "step": 4852
    },
    {
      "epoch": 0.5835387482715084,
      "flos": 19367141806080.0,
      "grad_norm": 2.5204812321523207,
      "learning_rate": 1.5596245402284998e-06,
      "loss": 0.9292,
      "num_input_tokens_seen": 103253135,
      "step": 4853
    },
    {
      "epoch": 0.5836589911621476,
      "flos": 11787902791680.0,
      "grad_norm": 6.569337316620377,
      "learning_rate": 1.5588647199026619e-06,
      "loss": 1.0088,
      "num_input_tokens_seen": 103270590,
      "step": 4854
    },
    {
      "epoch": 0.5837792340527866,
      "flos": 14512973905920.0,
      "grad_norm": 4.014555436126328,
      "learning_rate": 1.5581049664966956e-06,
      "loss": 1.0688,
      "num_input_tokens_seen": 103288070,
      "step": 4855
    },
    {
      "epoch": 0.5838994769434257,
      "flos": 47249881743360.0,
      "grad_norm": 1.0659307181154616,
      "learning_rate": 1.5573452801258545e-06,
      "loss": 0.8829,
      "num_input_tokens_seen": 103334960,
      "step": 4856
    },
    {
      "epoch": 0.5840197198340649,
      "flos": 15248975769600.0,
      "grad_norm": 4.487657150024608,
      "learning_rate": 1.5565856609053824e-06,
      "loss": 0.8244,
      "num_input_tokens_seen": 103353475,
      "step": 4857
    },
    {
      "epoch": 0.5841399627247039,
      "flos": 13570544394240.0,
      "grad_norm": 2.8483771940827127,
      "learning_rate": 1.5558261089505127e-06,
      "loss": 0.9936,
      "num_input_tokens_seen": 103371925,
      "step": 4858
    },
    {
      "epoch": 0.584260205615343,
      "flos": 18809230970880.0,
      "grad_norm": 3.799978236085907,
      "learning_rate": 1.5550666243764697e-06,
      "loss": 0.9933,
      "num_input_tokens_seen": 103389805,
      "step": 4859
    },
    {
      "epoch": 0.584380448505982,
      "flos": 9795154513920.0,
      "grad_norm": 4.157325548100681,
      "learning_rate": 1.554307207298465e-06,
      "loss": 0.9599,
      "num_input_tokens_seen": 103407785,
      "step": 4860
    },
    {
      "epoch": 0.5845006913966212,
      "flos": 15301602078720.0,
      "grad_norm": 3.1851045171575274,
      "learning_rate": 1.553547857831704e-06,
      "loss": 0.9814,
      "num_input_tokens_seen": 103424015,
      "step": 4861
    },
    {
      "epoch": 0.5846209342872603,
      "flos": 41774254141440.0,
      "grad_norm": 0.9898657780509265,
      "learning_rate": 1.5527885760913771e-06,
      "loss": 0.9123,
      "num_input_tokens_seen": 103473625,
      "step": 4862
    },
    {
      "epoch": 0.5847411771778993,
      "flos": 13177380003840.0,
      "grad_norm": 3.092204160486399,
      "learning_rate": 1.552029362192668e-06,
      "loss": 0.9572,
      "num_input_tokens_seen": 103492605,
      "step": 4863
    },
    {
      "epoch": 0.5848614200685385,
      "flos": 17239547289600.0,
      "grad_norm": 3.1217714034753796,
      "learning_rate": 1.5512702162507478e-06,
      "loss": 0.9056,
      "num_input_tokens_seen": 103512640,
      "step": 4864
    },
    {
      "epoch": 0.5849816629591775,
      "flos": 51322319646720.0,
      "grad_norm": 1.0037378548425364,
      "learning_rate": 1.5505111383807792e-06,
      "loss": 0.7624,
      "num_input_tokens_seen": 103575030,
      "step": 4865
    },
    {
      "epoch": 0.5851019058498166,
      "flos": 16924157337600.0,
      "grad_norm": 2.5401735662496736,
      "learning_rate": 1.5497521286979138e-06,
      "loss": 1.0025,
      "num_input_tokens_seen": 103594990,
      "step": 4866
    },
    {
      "epoch": 0.5852221487404557,
      "flos": 17345964933120.0,
      "grad_norm": 5.492597974303619,
      "learning_rate": 1.5489931873172927e-06,
      "loss": 0.9459,
      "num_input_tokens_seen": 103616030,
      "step": 4867
    },
    {
      "epoch": 0.5853423916310948,
      "flos": 19417928601600.0,
      "grad_norm": 3.265477811425774,
      "learning_rate": 1.5482343143540467e-06,
      "loss": 0.9929,
      "num_input_tokens_seen": 103637015,
      "step": 4868
    },
    {
      "epoch": 0.5854626345217339,
      "flos": 8430978723840.0,
      "grad_norm": 6.006430041485684,
      "learning_rate": 1.547475509923295e-06,
      "loss": 1.0329,
      "num_input_tokens_seen": 103653775,
      "step": 4869
    },
    {
      "epoch": 0.585582877412373,
      "flos": 46065199165440.0,
      "grad_norm": 0.7753083692299931,
      "learning_rate": 1.5467167741401495e-06,
      "loss": 0.7857,
      "num_input_tokens_seen": 103714975,
      "step": 4870
    },
    {
      "epoch": 0.5857031203030121,
      "flos": 12043676282880.0,
      "grad_norm": 2.983489059845007,
      "learning_rate": 1.5459581071197083e-06,
      "loss": 0.9049,
      "num_input_tokens_seen": 103730355,
      "step": 4871
    },
    {
      "epoch": 0.5858233631936511,
      "flos": 14828517150720.0,
      "grad_norm": 5.730368896113659,
      "learning_rate": 1.5451995089770624e-06,
      "loss": 1.0131,
      "num_input_tokens_seen": 103749860,
      "step": 4872
    },
    {
      "epoch": 0.5859436060842903,
      "flos": 16486981693440.0,
      "grad_norm": 2.1006018275782066,
      "learning_rate": 1.5444409798272885e-06,
      "loss": 0.9164,
      "num_input_tokens_seen": 103773670,
      "step": 4873
    },
    {
      "epoch": 0.5860638489749294,
      "flos": 15983628656640.0,
      "grad_norm": 19.62213463730716,
      "learning_rate": 1.543682519785456e-06,
      "loss": 0.9972,
      "num_input_tokens_seen": 103791870,
      "step": 4874
    },
    {
      "epoch": 0.5861840918655684,
      "flos": 12443003043840.0,
      "grad_norm": 4.573401817127837,
      "learning_rate": 1.5429241289666219e-06,
      "loss": 0.9828,
      "num_input_tokens_seen": 103809090,
      "step": 4875
    },
    {
      "epoch": 0.5863043347562076,
      "flos": 18185134632960.0,
      "grad_norm": 3.2713636871395324,
      "learning_rate": 1.5421658074858342e-06,
      "loss": 0.8919,
      "num_input_tokens_seen": 103826915,
      "step": 4876
    },
    {
      "epoch": 0.5864245776468466,
      "flos": 14671833907200.0,
      "grad_norm": 5.139793259140303,
      "learning_rate": 1.5414075554581298e-06,
      "loss": 0.8431,
      "num_input_tokens_seen": 103844680,
      "step": 4877
    },
    {
      "epoch": 0.5865448205374857,
      "flos": 20598433505280.0,
      "grad_norm": 2.800347585902092,
      "learning_rate": 1.5406493729985348e-06,
      "loss": 0.964,
      "num_input_tokens_seen": 103863595,
      "step": 4878
    },
    {
      "epoch": 0.5866650634281249,
      "flos": 18390243962880.0,
      "grad_norm": 3.3435151204833633,
      "learning_rate": 1.5398912602220644e-06,
      "loss": 0.9093,
      "num_input_tokens_seen": 103882590,
      "step": 4879
    },
    {
      "epoch": 0.5867853063187639,
      "flos": 12073239244800.0,
      "grad_norm": 3.531205978794334,
      "learning_rate": 1.539133217243724e-06,
      "loss": 0.9768,
      "num_input_tokens_seen": 103899330,
      "step": 4880
    },
    {
      "epoch": 0.586905549209403,
      "flos": 17530463600640.0,
      "grad_norm": 3.894852023368224,
      "learning_rate": 1.5383752441785081e-06,
      "loss": 0.9409,
      "num_input_tokens_seen": 103918275,
      "step": 4881
    },
    {
      "epoch": 0.5870257921000421,
      "flos": 10397873725440.0,
      "grad_norm": 4.66226320048909,
      "learning_rate": 1.5376173411414003e-06,
      "loss": 1.0513,
      "num_input_tokens_seen": 103936035,
      "step": 4882
    },
    {
      "epoch": 0.5871460349906812,
      "flos": 17005764096000.0,
      "grad_norm": 3.4184244952855494,
      "learning_rate": 1.5368595082473753e-06,
      "loss": 0.9743,
      "num_input_tokens_seen": 103954055,
      "step": 4883
    },
    {
      "epoch": 0.5872662778813202,
      "flos": 15747546071040.0,
      "grad_norm": 2.4951041921139074,
      "learning_rate": 1.5361017456113935e-06,
      "loss": 0.9796,
      "num_input_tokens_seen": 103974125,
      "step": 4884
    },
    {
      "epoch": 0.5873865207719594,
      "flos": 13072341995520.0,
      "grad_norm": 2.8993260751532377,
      "learning_rate": 1.5353440533484085e-06,
      "loss": 1.0451,
      "num_input_tokens_seen": 103992700,
      "step": 4885
    },
    {
      "epoch": 0.5875067636625985,
      "flos": 38638236672000.0,
      "grad_norm": 3.694216930742465,
      "learning_rate": 1.534586431573361e-06,
      "loss": 0.8591,
      "num_input_tokens_seen": 104017360,
      "step": 4886
    },
    {
      "epoch": 0.5876270065532375,
      "flos": 19937722736640.0,
      "grad_norm": 2.6931181447616233,
      "learning_rate": 1.5338288804011817e-06,
      "loss": 0.9716,
      "num_input_tokens_seen": 104036580,
      "step": 4887
    },
    {
      "epoch": 0.5877472494438767,
      "flos": 15170649477120.0,
      "grad_norm": 4.753531699051938,
      "learning_rate": 1.533071399946791e-06,
      "loss": 0.8845,
      "num_input_tokens_seen": 104055045,
      "step": 4888
    },
    {
      "epoch": 0.5878674923345157,
      "flos": 15904811827200.0,
      "grad_norm": 3.920706160641945,
      "learning_rate": 1.5323139903250977e-06,
      "loss": 0.7664,
      "num_input_tokens_seen": 104075370,
      "step": 4889
    },
    {
      "epoch": 0.5879877352251548,
      "flos": 15534680125440.0,
      "grad_norm": 3.3908674443490834,
      "learning_rate": 1.5315566516510002e-06,
      "loss": 0.9676,
      "num_input_tokens_seen": 104093260,
      "step": 4890
    },
    {
      "epoch": 0.5881079781157939,
      "flos": 12391020564480.0,
      "grad_norm": 2.582541272552769,
      "learning_rate": 1.5307993840393857e-06,
      "loss": 0.8757,
      "num_input_tokens_seen": 104111060,
      "step": 4891
    },
    {
      "epoch": 0.588228221006433,
      "flos": 16062077583360.0,
      "grad_norm": 4.3014570739441735,
      "learning_rate": 1.530042187605132e-06,
      "loss": 1.0,
      "num_input_tokens_seen": 104130035,
      "step": 4892
    },
    {
      "epoch": 0.5883484638970721,
      "flos": 18631354552320.0,
      "grad_norm": 3.0061265059581608,
      "learning_rate": 1.5292850624631044e-06,
      "loss": 1.0342,
      "num_input_tokens_seen": 104151950,
      "step": 4893
    },
    {
      "epoch": 0.5884687067877111,
      "flos": 21697546260480.0,
      "grad_norm": 2.490905910526617,
      "learning_rate": 1.5285280087281593e-06,
      "loss": 0.9762,
      "num_input_tokens_seen": 104172400,
      "step": 4894
    },
    {
      "epoch": 0.5885889496783503,
      "flos": 50495903907840.0,
      "grad_norm": 0.6470135830583957,
      "learning_rate": 1.5277710265151398e-06,
      "loss": 0.7611,
      "num_input_tokens_seen": 104241600,
      "step": 4895
    },
    {
      "epoch": 0.5887091925689893,
      "flos": 13544415191040.0,
      "grad_norm": 5.831410154574033,
      "learning_rate": 1.5270141159388803e-06,
      "loss": 0.945,
      "num_input_tokens_seen": 104258340,
      "step": 4896
    },
    {
      "epoch": 0.5888294354596284,
      "flos": 16559605493760.0,
      "grad_norm": 3.650672636369497,
      "learning_rate": 1.526257277114203e-06,
      "loss": 0.9866,
      "num_input_tokens_seen": 104279135,
      "step": 4897
    },
    {
      "epoch": 0.5889496783502676,
      "flos": 15616256225280.0,
      "grad_norm": 2.75724159793817,
      "learning_rate": 1.5255005101559201e-06,
      "loss": 0.9844,
      "num_input_tokens_seen": 104296465,
      "step": 4898
    },
    {
      "epoch": 0.5890699212409066,
      "flos": 15402715791360.0,
      "grad_norm": 3.7913930745331164,
      "learning_rate": 1.524743815178833e-06,
      "loss": 0.9586,
      "num_input_tokens_seen": 104314145,
      "step": 4899
    },
    {
      "epoch": 0.5891901641315457,
      "flos": 13807424102400.0,
      "grad_norm": 2.7292535394329573,
      "learning_rate": 1.5239871922977315e-06,
      "loss": 1.0027,
      "num_input_tokens_seen": 104333780,
      "step": 4900
    },
    {
      "epoch": 0.5893104070221848,
      "flos": 13911787622400.0,
      "grad_norm": 6.954063643630821,
      "learning_rate": 1.523230641627394e-06,
      "loss": 1.0841,
      "num_input_tokens_seen": 104352485,
      "step": 4901
    },
    {
      "epoch": 0.5894306499128239,
      "flos": 20703502172160.0,
      "grad_norm": 3.977350363035592,
      "learning_rate": 1.5224741632825888e-06,
      "loss": 0.9126,
      "num_input_tokens_seen": 104372395,
      "step": 4902
    },
    {
      "epoch": 0.589550892803463,
      "flos": 30217252638720.0,
      "grad_norm": 4.115067748836542,
      "learning_rate": 1.521717757378074e-06,
      "loss": 0.8845,
      "num_input_tokens_seen": 104392660,
      "step": 4903
    },
    {
      "epoch": 0.5896711356941021,
      "flos": 9978580131840.0,
      "grad_norm": 6.691922445701766,
      "learning_rate": 1.5209614240285943e-06,
      "loss": 0.8786,
      "num_input_tokens_seen": 104410035,
      "step": 4904
    },
    {
      "epoch": 0.5897913785847412,
      "flos": 12180760596480.0,
      "grad_norm": 7.889659014200269,
      "learning_rate": 1.520205163348887e-06,
      "loss": 1.0368,
      "num_input_tokens_seen": 104427690,
      "step": 4905
    },
    {
      "epoch": 0.5899116214753802,
      "flos": 34890316861440.0,
      "grad_norm": 0.7508160998843122,
      "learning_rate": 1.519448975453674e-06,
      "loss": 0.7723,
      "num_input_tokens_seen": 104482510,
      "step": 4906
    },
    {
      "epoch": 0.5900318643660194,
      "flos": 14985047101440.0,
      "grad_norm": 16.640919341586965,
      "learning_rate": 1.5186928604576696e-06,
      "loss": 0.9588,
      "num_input_tokens_seen": 104499425,
      "step": 4907
    },
    {
      "epoch": 0.5901521072566585,
      "flos": 15038991728640.0,
      "grad_norm": 3.0026254292000214,
      "learning_rate": 1.5179368184755752e-06,
      "loss": 0.9686,
      "num_input_tokens_seen": 104517230,
      "step": 4908
    },
    {
      "epoch": 0.5902723501472975,
      "flos": 14353991270400.0,
      "grad_norm": 2.7844713393476397,
      "learning_rate": 1.5171808496220821e-06,
      "loss": 1.0202,
      "num_input_tokens_seen": 104535705,
      "step": 4909
    },
    {
      "epoch": 0.5903925930379367,
      "flos": 16322327224320.0,
      "grad_norm": 3.6692442215044037,
      "learning_rate": 1.5164249540118708e-06,
      "loss": 1.0045,
      "num_input_tokens_seen": 104554550,
      "step": 4910
    },
    {
      "epoch": 0.5905128359285757,
      "flos": 16612446412800.0,
      "grad_norm": 2.834847401005441,
      "learning_rate": 1.5156691317596093e-06,
      "loss": 1.026,
      "num_input_tokens_seen": 104575695,
      "step": 4911
    },
    {
      "epoch": 0.5906330788192148,
      "flos": 19964066549760.0,
      "grad_norm": 3.9100163727244346,
      "learning_rate": 1.5149133829799556e-06,
      "loss": 0.8771,
      "num_input_tokens_seen": 104593410,
      "step": 4912
    },
    {
      "epoch": 0.590753321709854,
      "flos": 13098164613120.0,
      "grad_norm": 3.175710379359903,
      "learning_rate": 1.5141577077875556e-06,
      "loss": 0.9959,
      "num_input_tokens_seen": 104610455,
      "step": 4913
    },
    {
      "epoch": 0.590873564600493,
      "flos": 11945383157760.0,
      "grad_norm": 3.6048324205280795,
      "learning_rate": 1.5134021062970451e-06,
      "loss": 0.9062,
      "num_input_tokens_seen": 104628555,
      "step": 4914
    },
    {
      "epoch": 0.5909938074911321,
      "flos": 9532268236800.0,
      "grad_norm": 3.0414474268942286,
      "learning_rate": 1.5126465786230483e-06,
      "loss": 1.0021,
      "num_input_tokens_seen": 104645050,
      "step": 4915
    },
    {
      "epoch": 0.5911140503817712,
      "flos": 18520552734720.0,
      "grad_norm": 22.682537282204425,
      "learning_rate": 1.5118911248801787e-06,
      "loss": 0.9986,
      "num_input_tokens_seen": 104662780,
      "step": 4916
    },
    {
      "epoch": 0.5912342932724103,
      "flos": 16534181437440.0,
      "grad_norm": 5.341541034492662,
      "learning_rate": 1.5111357451830364e-06,
      "loss": 0.9747,
      "num_input_tokens_seen": 104681195,
      "step": 4917
    },
    {
      "epoch": 0.5913545361630493,
      "flos": 13885259857920.0,
      "grad_norm": 5.672908972451153,
      "learning_rate": 1.5103804396462131e-06,
      "loss": 0.905,
      "num_input_tokens_seen": 104700850,
      "step": 4918
    },
    {
      "epoch": 0.5914747790536885,
      "flos": 18657667706880.0,
      "grad_norm": 4.049757943160722,
      "learning_rate": 1.5096252083842877e-06,
      "loss": 0.995,
      "num_input_tokens_seen": 104719780,
      "step": 4919
    },
    {
      "epoch": 0.5915950219443276,
      "flos": 19522966609920.0,
      "grad_norm": 2.9280446239989915,
      "learning_rate": 1.5088700515118285e-06,
      "loss": 1.0465,
      "num_input_tokens_seen": 104738820,
      "step": 4920
    },
    {
      "epoch": 0.5917152648349666,
      "flos": 15563231354880.0,
      "grad_norm": 3.1797551328322355,
      "learning_rate": 1.508114969143392e-06,
      "loss": 0.8581,
      "num_input_tokens_seen": 104758525,
      "step": 4921
    },
    {
      "epoch": 0.5918355077256057,
      "flos": 20019820032000.0,
      "grad_norm": 2.3833415460333898,
      "learning_rate": 1.5073599613935238e-06,
      "loss": 0.9735,
      "num_input_tokens_seen": 104780365,
      "step": 4922
    },
    {
      "epoch": 0.5919557506162448,
      "flos": 20073672683520.0,
      "grad_norm": 2.9255902338763597,
      "learning_rate": 1.5066050283767574e-06,
      "loss": 0.7745,
      "num_input_tokens_seen": 104800765,
      "step": 4923
    },
    {
      "epoch": 0.5920759935068839,
      "flos": 8510592675840.0,
      "grad_norm": 3.693790565952161,
      "learning_rate": 1.505850170207616e-06,
      "loss": 1.0186,
      "num_input_tokens_seen": 104817350,
      "step": 4924
    },
    {
      "epoch": 0.592196236397523,
      "flos": 20965591326720.0,
      "grad_norm": 3.9716102926074224,
      "learning_rate": 1.505095387000611e-06,
      "loss": 0.9609,
      "num_input_tokens_seen": 104839370,
      "step": 4925
    },
    {
      "epoch": 0.5923164792881621,
      "flos": 17342715125760.0,
      "grad_norm": 2.9847633947916514,
      "learning_rate": 1.504340678870242e-06,
      "loss": 0.9461,
      "num_input_tokens_seen": 104857305,
      "step": 4926
    },
    {
      "epoch": 0.5924367221788012,
      "flos": 17084243681280.0,
      "grad_norm": 3.6676034308724628,
      "learning_rate": 1.5035860459309989e-06,
      "loss": 1.091,
      "num_input_tokens_seen": 104874740,
      "step": 4927
    },
    {
      "epoch": 0.5925569650694402,
      "flos": 19129924853760.0,
      "grad_norm": 3.9984452086922566,
      "learning_rate": 1.5028314882973568e-06,
      "loss": 0.8299,
      "num_input_tokens_seen": 104894865,
      "step": 4928
    },
    {
      "epoch": 0.5926772079600794,
      "flos": 15846176440320.0,
      "grad_norm": 4.292115839195769,
      "learning_rate": 1.502077006083783e-06,
      "loss": 1.0353,
      "num_input_tokens_seen": 104913245,
      "step": 4929
    },
    {
      "epoch": 0.5927974508507184,
      "flos": 14095305216000.0,
      "grad_norm": 3.4104694515393055,
      "learning_rate": 1.5013225994047315e-06,
      "loss": 0.9675,
      "num_input_tokens_seen": 104930595,
      "step": 4930
    },
    {
      "epoch": 0.5929176937413575,
      "flos": 11156693667840.0,
      "grad_norm": 2.718158175190775,
      "learning_rate": 1.5005682683746452e-06,
      "loss": 1.0016,
      "num_input_tokens_seen": 104948830,
      "step": 4931
    },
    {
      "epoch": 0.5930379366319967,
      "flos": 12468151173120.0,
      "grad_norm": 4.519506139528882,
      "learning_rate": 1.4998140131079553e-06,
      "loss": 0.9237,
      "num_input_tokens_seen": 104964640,
      "step": 4932
    },
    {
      "epoch": 0.5931581795226357,
      "flos": 12541142876160.0,
      "grad_norm": 3.42459458227956,
      "learning_rate": 1.4990598337190821e-06,
      "loss": 0.9375,
      "num_input_tokens_seen": 104980715,
      "step": 4933
    },
    {
      "epoch": 0.5932784224132748,
      "flos": 17268190494720.0,
      "grad_norm": 2.7447762727154945,
      "learning_rate": 1.4983057303224338e-06,
      "loss": 0.8679,
      "num_input_tokens_seen": 105000250,
      "step": 4934
    },
    {
      "epoch": 0.5933986653039139,
      "flos": 16295063654400.0,
      "grad_norm": 5.519057634879711,
      "learning_rate": 1.4975517030324072e-06,
      "loss": 1.0658,
      "num_input_tokens_seen": 105017980,
      "step": 4935
    },
    {
      "epoch": 0.593518908194553,
      "flos": 50936482652160.0,
      "grad_norm": 0.8488543433095858,
      "learning_rate": 1.4967977519633882e-06,
      "loss": 0.8419,
      "num_input_tokens_seen": 105075160,
      "step": 4936
    },
    {
      "epoch": 0.593639151085192,
      "flos": 14514537492480.0,
      "grad_norm": 4.164962928854286,
      "learning_rate": 1.4960438772297494e-06,
      "loss": 0.9811,
      "num_input_tokens_seen": 105091925,
      "step": 4937
    },
    {
      "epoch": 0.5937593939758312,
      "flos": 22013794652160.0,
      "grad_norm": 6.2608456922158915,
      "learning_rate": 1.495290078945855e-06,
      "loss": 0.932,
      "num_input_tokens_seen": 105111410,
      "step": 4938
    },
    {
      "epoch": 0.5938796368664703,
      "flos": 26335659724800.0,
      "grad_norm": 4.116291911458879,
      "learning_rate": 1.4945363572260529e-06,
      "loss": 0.9343,
      "num_input_tokens_seen": 105132125,
      "step": 4939
    },
    {
      "epoch": 0.5939998797571093,
      "flos": 16954088202240.0,
      "grad_norm": 3.2877293147765694,
      "learning_rate": 1.4937827121846845e-06,
      "loss": 0.8557,
      "num_input_tokens_seen": 105152100,
      "step": 4940
    },
    {
      "epoch": 0.5941201226477485,
      "flos": 17923014819840.0,
      "grad_norm": 2.6261045765910356,
      "learning_rate": 1.4930291439360755e-06,
      "loss": 0.9323,
      "num_input_tokens_seen": 105174385,
      "step": 4941
    },
    {
      "epoch": 0.5942403655383875,
      "flos": 15930603786240.0,
      "grad_norm": 3.7178703998880556,
      "learning_rate": 1.4922756525945427e-06,
      "loss": 0.99,
      "num_input_tokens_seen": 105193415,
      "step": 4942
    },
    {
      "epoch": 0.5943606084290266,
      "flos": 48427925852160.0,
      "grad_norm": 0.7821699067170671,
      "learning_rate": 1.4915222382743894e-06,
      "loss": 0.8094,
      "num_input_tokens_seen": 105251970,
      "step": 4943
    },
    {
      "epoch": 0.5944808513196658,
      "flos": 12915137556480.0,
      "grad_norm": 5.208106969364752,
      "learning_rate": 1.4907689010899085e-06,
      "loss": 0.9198,
      "num_input_tokens_seen": 105269270,
      "step": 4944
    },
    {
      "epoch": 0.5946010942103048,
      "flos": 17635225681920.0,
      "grad_norm": 3.9878858469290726,
      "learning_rate": 1.4900156411553804e-06,
      "loss": 0.8125,
      "num_input_tokens_seen": 105288820,
      "step": 4945
    },
    {
      "epoch": 0.5947213371009439,
      "flos": 10765215498240.0,
      "grad_norm": 9.976037115424155,
      "learning_rate": 1.4892624585850739e-06,
      "loss": 1.0378,
      "num_input_tokens_seen": 105306895,
      "step": 4946
    },
    {
      "epoch": 0.594841579991583,
      "flos": 18394934722560.0,
      "grad_norm": 3.502811120405903,
      "learning_rate": 1.4885093534932465e-06,
      "loss": 0.9782,
      "num_input_tokens_seen": 105324580,
      "step": 4947
    },
    {
      "epoch": 0.5949618228822221,
      "flos": 17052350668800.0,
      "grad_norm": 3.6746427453189376,
      "learning_rate": 1.4877563259941433e-06,
      "loss": 0.914,
      "num_input_tokens_seen": 105342155,
      "step": 4948
    },
    {
      "epoch": 0.5950820657728612,
      "flos": 28958544076800.0,
      "grad_norm": 6.246929787895999,
      "learning_rate": 1.4870033762019988e-06,
      "loss": 0.8708,
      "num_input_tokens_seen": 105362040,
      "step": 4949
    },
    {
      "epoch": 0.5952023086635003,
      "flos": 16480635371520.0,
      "grad_norm": 4.508905288226107,
      "learning_rate": 1.4862505042310334e-06,
      "loss": 0.9261,
      "num_input_tokens_seen": 105381045,
      "step": 4950
    },
    {
      "epoch": 0.5953225515541394,
      "flos": 24004028928000.0,
      "grad_norm": 2.4608963251453755,
      "learning_rate": 1.4854977101954587e-06,
      "loss": 0.8943,
      "num_input_tokens_seen": 105402985,
      "step": 4951
    },
    {
      "epoch": 0.5954427944447784,
      "flos": 17396813045760.0,
      "grad_norm": 5.799609914355568,
      "learning_rate": 1.4847449942094716e-06,
      "loss": 1.039,
      "num_input_tokens_seen": 105421585,
      "step": 4952
    },
    {
      "epoch": 0.5955630373354175,
      "flos": 13151404093440.0,
      "grad_norm": 2.824273706431718,
      "learning_rate": 1.4839923563872598e-06,
      "loss": 1.0533,
      "num_input_tokens_seen": 105439845,
      "step": 4953
    },
    {
      "epoch": 0.5956832802260567,
      "flos": 14042372321280.0,
      "grad_norm": 3.529563973922034,
      "learning_rate": 1.483239796842997e-06,
      "loss": 0.959,
      "num_input_tokens_seen": 105457595,
      "step": 4954
    },
    {
      "epoch": 0.5958035231166957,
      "flos": 14012441456640.0,
      "grad_norm": 2.636958629857942,
      "learning_rate": 1.4824873156908462e-06,
      "loss": 1.0241,
      "num_input_tokens_seen": 105475240,
      "step": 4955
    },
    {
      "epoch": 0.5959237660073348,
      "flos": 15379959029760.0,
      "grad_norm": 2.6822525070936747,
      "learning_rate": 1.4817349130449584e-06,
      "loss": 0.9487,
      "num_input_tokens_seen": 105494680,
      "step": 4956
    },
    {
      "epoch": 0.5960440088979739,
      "flos": 15032860016640.0,
      "grad_norm": 4.912775062062663,
      "learning_rate": 1.4809825890194717e-06,
      "loss": 1.0172,
      "num_input_tokens_seen": 105513070,
      "step": 4957
    },
    {
      "epoch": 0.596164251788613,
      "flos": 10424524124160.0,
      "grad_norm": 2.3382134720563394,
      "learning_rate": 1.4802303437285139e-06,
      "loss": 0.9609,
      "num_input_tokens_seen": 105530060,
      "step": 4958
    },
    {
      "epoch": 0.596284494679252,
      "flos": 14541433159680.0,
      "grad_norm": 4.901740267262748,
      "learning_rate": 1.4794781772861994e-06,
      "loss": 0.9792,
      "num_input_tokens_seen": 105546275,
      "step": 4959
    },
    {
      "epoch": 0.5964047375698912,
      "flos": 22250183823360.0,
      "grad_norm": 4.039648489437279,
      "learning_rate": 1.4787260898066324e-06,
      "loss": 0.8585,
      "num_input_tokens_seen": 105565995,
      "step": 4960
    },
    {
      "epoch": 0.5965249804605303,
      "flos": 19569522524160.0,
      "grad_norm": 3.830253040729563,
      "learning_rate": 1.4779740814039023e-06,
      "loss": 1.0463,
      "num_input_tokens_seen": 105585800,
      "step": 4961
    },
    {
      "epoch": 0.5966452233511693,
      "flos": 21935100456960.0,
      "grad_norm": 3.513976257270062,
      "learning_rate": 1.4772221521920894e-06,
      "loss": 0.8751,
      "num_input_tokens_seen": 105605545,
      "step": 4962
    },
    {
      "epoch": 0.5967654662418085,
      "flos": 18130975395840.0,
      "grad_norm": 3.020169011875085,
      "learning_rate": 1.4764703022852598e-06,
      "loss": 0.9328,
      "num_input_tokens_seen": 105625785,
      "step": 4963
    },
    {
      "epoch": 0.5968857091324475,
      "flos": 13544721776640.0,
      "grad_norm": 3.9196568025489253,
      "learning_rate": 1.4757185317974696e-06,
      "loss": 0.958,
      "num_input_tokens_seen": 105643890,
      "step": 4964
    },
    {
      "epoch": 0.5970059520230866,
      "flos": 16845555118080.0,
      "grad_norm": 3.4854274922119375,
      "learning_rate": 1.474966840842761e-06,
      "loss": 0.9147,
      "num_input_tokens_seen": 105663190,
      "step": 4965
    },
    {
      "epoch": 0.5971261949137258,
      "flos": 16481555128320.0,
      "grad_norm": 6.499449193064934,
      "learning_rate": 1.4742152295351655e-06,
      "loss": 1.0635,
      "num_input_tokens_seen": 105682065,
      "step": 4966
    },
    {
      "epoch": 0.5972464378043648,
      "flos": 14592863784960.0,
      "grad_norm": 3.364112388003506,
      "learning_rate": 1.4734636979887016e-06,
      "loss": 0.8326,
      "num_input_tokens_seen": 105699245,
      "step": 4967
    },
    {
      "epoch": 0.5973666806950039,
      "flos": 20938297098240.0,
      "grad_norm": 2.797260790595512,
      "learning_rate": 1.4727122463173755e-06,
      "loss": 1.0972,
      "num_input_tokens_seen": 105717495,
      "step": 4968
    },
    {
      "epoch": 0.597486923585643,
      "flos": 15826117632000.0,
      "grad_norm": 3.8740798103725522,
      "learning_rate": 1.471960874635183e-06,
      "loss": 0.8349,
      "num_input_tokens_seen": 105736775,
      "step": 4969
    },
    {
      "epoch": 0.5976071664762821,
      "flos": 9716582952960.0,
      "grad_norm": 3.4621242443248756,
      "learning_rate": 1.4712095830561055e-06,
      "loss": 0.8845,
      "num_input_tokens_seen": 105754985,
      "step": 4970
    },
    {
      "epoch": 0.5977274093669211,
      "flos": 13544292556800.0,
      "grad_norm": 4.2673490624309,
      "learning_rate": 1.4704583716941147e-06,
      "loss": 0.9921,
      "num_input_tokens_seen": 105773570,
      "step": 4971
    },
    {
      "epoch": 0.5978476522575603,
      "flos": 14462371061760.0,
      "grad_norm": 14.612467294726553,
      "learning_rate": 1.4697072406631672e-06,
      "loss": 0.9108,
      "num_input_tokens_seen": 105793195,
      "step": 4972
    },
    {
      "epoch": 0.5979678951481994,
      "flos": 20676913090560.0,
      "grad_norm": 2.9772568741322725,
      "learning_rate": 1.4689561900772097e-06,
      "loss": 0.9302,
      "num_input_tokens_seen": 105812975,
      "step": 4973
    },
    {
      "epoch": 0.5980881380388384,
      "flos": 12731527987200.0,
      "grad_norm": 3.858732347611761,
      "learning_rate": 1.4682052200501758e-06,
      "loss": 0.9258,
      "num_input_tokens_seen": 105829900,
      "step": 4974
    },
    {
      "epoch": 0.5982083809294776,
      "flos": 16320978247680.0,
      "grad_norm": 2.572893418015272,
      "learning_rate": 1.4674543306959876e-06,
      "loss": 0.9738,
      "num_input_tokens_seen": 105849090,
      "step": 4975
    },
    {
      "epoch": 0.5983286238201166,
      "flos": 14904452075520.0,
      "grad_norm": 4.674516932371236,
      "learning_rate": 1.4667035221285535e-06,
      "loss": 1.0431,
      "num_input_tokens_seen": 105866450,
      "step": 4976
    },
    {
      "epoch": 0.5984488667107557,
      "flos": 20073366097920.0,
      "grad_norm": 2.78748189124824,
      "learning_rate": 1.4659527944617715e-06,
      "loss": 0.9429,
      "num_input_tokens_seen": 105886115,
      "step": 4977
    },
    {
      "epoch": 0.5985691096013949,
      "flos": 11656459653120.0,
      "grad_norm": 3.830324249685273,
      "learning_rate": 1.465202147809526e-06,
      "loss": 0.9493,
      "num_input_tokens_seen": 105904330,
      "step": 4978
    },
    {
      "epoch": 0.5986893524920339,
      "flos": 19020257402880.0,
      "grad_norm": 4.261247897339244,
      "learning_rate": 1.4644515822856888e-06,
      "loss": 0.9594,
      "num_input_tokens_seen": 105922485,
      "step": 4979
    },
    {
      "epoch": 0.598809595382673,
      "flos": 44100419604480.0,
      "grad_norm": 0.7696817318711414,
      "learning_rate": 1.4637010980041215e-06,
      "loss": 0.7806,
      "num_input_tokens_seen": 105984315,
      "step": 4980
    },
    {
      "epoch": 0.5989298382733121,
      "flos": 8300179415040.0,
      "grad_norm": 4.696465831403652,
      "learning_rate": 1.4629506950786707e-06,
      "loss": 1.0928,
      "num_input_tokens_seen": 106000215,
      "step": 4981
    },
    {
      "epoch": 0.5990500811639512,
      "flos": 42961473269760.0,
      "grad_norm": 0.813472056290509,
      "learning_rate": 1.4622003736231733e-06,
      "loss": 0.7773,
      "num_input_tokens_seen": 106058925,
      "step": 4982
    },
    {
      "epoch": 0.5991703240545903,
      "flos": 12914769653760.0,
      "grad_norm": 3.509743208009588,
      "learning_rate": 1.461450133751451e-06,
      "loss": 0.9845,
      "num_input_tokens_seen": 106076715,
      "step": 4983
    },
    {
      "epoch": 0.5992905669452293,
      "flos": 19733134602240.0,
      "grad_norm": 3.1613365958508157,
      "learning_rate": 1.4606999755773153e-06,
      "loss": 0.9587,
      "num_input_tokens_seen": 106097640,
      "step": 4984
    },
    {
      "epoch": 0.5994108098358685,
      "flos": 14514844078080.0,
      "grad_norm": 3.8352958935079933,
      "learning_rate": 1.4599498992145643e-06,
      "loss": 1.0077,
      "num_input_tokens_seen": 106117385,
      "step": 4985
    },
    {
      "epoch": 0.5995310527265075,
      "flos": 15823082434560.0,
      "grad_norm": 9.687243648951963,
      "learning_rate": 1.4591999047769846e-06,
      "loss": 0.9121,
      "num_input_tokens_seen": 106135960,
      "step": 4986
    },
    {
      "epoch": 0.5996512956171466,
      "flos": 13413830492160.0,
      "grad_norm": 2.779458830049377,
      "learning_rate": 1.4584499923783486e-06,
      "loss": 0.9483,
      "num_input_tokens_seen": 106154260,
      "step": 4987
    },
    {
      "epoch": 0.5997715385077858,
      "flos": 10864428380160.0,
      "grad_norm": 4.149763969235625,
      "learning_rate": 1.457700162132419e-06,
      "loss": 0.9563,
      "num_input_tokens_seen": 106170970,
      "step": 4988
    },
    {
      "epoch": 0.5998917813984248,
      "flos": 17975855738880.0,
      "grad_norm": 2.979456178340662,
      "learning_rate": 1.4569504141529433e-06,
      "loss": 0.9154,
      "num_input_tokens_seen": 106188525,
      "step": 4989
    },
    {
      "epoch": 0.6000120242890639,
      "flos": 15668698583040.0,
      "grad_norm": 4.01463715392327,
      "learning_rate": 1.456200748553658e-06,
      "loss": 0.9169,
      "num_input_tokens_seen": 106206240,
      "step": 4990
    },
    {
      "epoch": 0.600132267179703,
      "flos": 21280950620160.0,
      "grad_norm": 2.891610632215175,
      "learning_rate": 1.455451165448287e-06,
      "loss": 0.9732,
      "num_input_tokens_seen": 106228615,
      "step": 4991
    },
    {
      "epoch": 0.6002525100703421,
      "flos": 17818038128640.0,
      "grad_norm": 5.381427027396756,
      "learning_rate": 1.4547016649505407e-06,
      "loss": 0.9082,
      "num_input_tokens_seen": 106246345,
      "step": 4992
    },
    {
      "epoch": 0.6003727529609811,
      "flos": 14802633216000.0,
      "grad_norm": 3.523540486862757,
      "learning_rate": 1.4539522471741193e-06,
      "loss": 1.0317,
      "num_input_tokens_seen": 106263490,
      "step": 4993
    },
    {
      "epoch": 0.6004929958516203,
      "flos": 11025833041920.0,
      "grad_norm": 4.143778050300398,
      "learning_rate": 1.4532029122327067e-06,
      "loss": 0.9011,
      "num_input_tokens_seen": 106279995,
      "step": 4994
    },
    {
      "epoch": 0.6006132387422594,
      "flos": 15459481006080.0,
      "grad_norm": 3.811437490023475,
      "learning_rate": 1.4524536602399783e-06,
      "loss": 0.9513,
      "num_input_tokens_seen": 106298805,
      "step": 4995
    },
    {
      "epoch": 0.6007334816328984,
      "flos": 16246729543680.0,
      "grad_norm": 4.889416627625027,
      "learning_rate": 1.4517044913095938e-06,
      "loss": 0.9705,
      "num_input_tokens_seen": 106318945,
      "step": 4996
    },
    {
      "epoch": 0.6008537245235376,
      "flos": 20174142566400.0,
      "grad_norm": 3.570231831179754,
      "learning_rate": 1.4509554055552022e-06,
      "loss": 1.0094,
      "num_input_tokens_seen": 106338895,
      "step": 4997
    },
    {
      "epoch": 0.6009739674141766,
      "flos": 14828885053440.0,
      "grad_norm": 3.684204956996982,
      "learning_rate": 1.450206403090439e-06,
      "loss": 1.0258,
      "num_input_tokens_seen": 106356810,
      "step": 4998
    },
    {
      "epoch": 0.6010942103048157,
      "flos": 14537232936960.0,
      "grad_norm": 3.5358084904174607,
      "learning_rate": 1.4494574840289274e-06,
      "loss": 1.0504,
      "num_input_tokens_seen": 106373645,
      "step": 4999
    },
    {
      "epoch": 0.6012144531954549,
      "flos": 16927468462080.0,
      "grad_norm": 3.2731346409065485,
      "learning_rate": 1.4487086484842782e-06,
      "loss": 0.9411,
      "num_input_tokens_seen": 106392010,
      "step": 5000
    },
    {
      "epoch": 0.6013346960860939,
      "flos": 13464954531840.0,
      "grad_norm": 3.0481749280259005,
      "learning_rate": 1.4479598965700878e-06,
      "loss": 0.7888,
      "num_input_tokens_seen": 106408995,
      "step": 5001
    },
    {
      "epoch": 0.601454938976733,
      "flos": 17085163438080.0,
      "grad_norm": 3.989670648773048,
      "learning_rate": 1.4472112283999427e-06,
      "loss": 0.872,
      "num_input_tokens_seen": 106427370,
      "step": 5002
    },
    {
      "epoch": 0.6015751818673721,
      "flos": 18810457313280.0,
      "grad_norm": 7.930328734079465,
      "learning_rate": 1.4464626440874143e-06,
      "loss": 0.8919,
      "num_input_tokens_seen": 106446205,
      "step": 5003
    },
    {
      "epoch": 0.6016954247580112,
      "flos": 9244203171840.0,
      "grad_norm": 4.356811207069107,
      "learning_rate": 1.4457141437460636e-06,
      "loss": 0.9233,
      "num_input_tokens_seen": 106463150,
      "step": 5004
    },
    {
      "epoch": 0.6018156676486502,
      "flos": 16900358184960.0,
      "grad_norm": 3.2581994721086107,
      "learning_rate": 1.444965727489436e-06,
      "loss": 0.9238,
      "num_input_tokens_seen": 106482315,
      "step": 5005
    },
    {
      "epoch": 0.6019359105392894,
      "flos": 18841583861760.0,
      "grad_norm": 6.088722698390886,
      "learning_rate": 1.444217395431066e-06,
      "loss": 0.8282,
      "num_input_tokens_seen": 106504575,
      "step": 5006
    },
    {
      "epoch": 0.6020561534299285,
      "flos": 49549182197760.0,
      "grad_norm": 0.8184315870376584,
      "learning_rate": 1.4434691476844755e-06,
      "loss": 0.7764,
      "num_input_tokens_seen": 106565270,
      "step": 5007
    },
    {
      "epoch": 0.6021763963205675,
      "flos": 15511126241280.0,
      "grad_norm": 7.381151123418236,
      "learning_rate": 1.4427209843631729e-06,
      "loss": 0.8645,
      "num_input_tokens_seen": 106582040,
      "step": 5008
    },
    {
      "epoch": 0.6022966392112067,
      "flos": 18920094105600.0,
      "grad_norm": 3.381555900909243,
      "learning_rate": 1.4419729055806534e-06,
      "loss": 1.009,
      "num_input_tokens_seen": 106601195,
      "step": 5009
    },
    {
      "epoch": 0.6024168821018457,
      "flos": 14697809817600.0,
      "grad_norm": 2.9155068589262143,
      "learning_rate": 1.441224911450401e-06,
      "loss": 1.0274,
      "num_input_tokens_seen": 106616870,
      "step": 5010
    },
    {
      "epoch": 0.6025371249924848,
      "flos": 17556224901120.0,
      "grad_norm": 8.030665751183184,
      "learning_rate": 1.4404770020858851e-06,
      "loss": 1.0266,
      "num_input_tokens_seen": 106636075,
      "step": 5011
    },
    {
      "epoch": 0.602657367883124,
      "flos": 18472126648320.0,
      "grad_norm": 2.845019214478087,
      "learning_rate": 1.439729177600563e-06,
      "loss": 1.0616,
      "num_input_tokens_seen": 106656290,
      "step": 5012
    },
    {
      "epoch": 0.602777610773763,
      "flos": 11813725409280.0,
      "grad_norm": 4.5608596988536005,
      "learning_rate": 1.4389814381078793e-06,
      "loss": 0.9169,
      "num_input_tokens_seen": 106675250,
      "step": 5013
    },
    {
      "epoch": 0.6028978536644021,
      "flos": 9401591562240.0,
      "grad_norm": 4.223108145700074,
      "learning_rate": 1.438233783721265e-06,
      "loss": 0.9911,
      "num_input_tokens_seen": 106691135,
      "step": 5014
    },
    {
      "epoch": 0.6030180965550412,
      "flos": 13936690483200.0,
      "grad_norm": 3.641195751850743,
      "learning_rate": 1.43748621455414e-06,
      "loss": 0.9729,
      "num_input_tokens_seen": 106707290,
      "step": 5015
    },
    {
      "epoch": 0.6031383394456803,
      "flos": 10208561664000.0,
      "grad_norm": 4.477244167122366,
      "learning_rate": 1.4367387307199082e-06,
      "loss": 1.0031,
      "num_input_tokens_seen": 106723860,
      "step": 5016
    },
    {
      "epoch": 0.6032585823363193,
      "flos": 9820946472960.0,
      "grad_norm": 2.9632025034644673,
      "learning_rate": 1.4359913323319632e-06,
      "loss": 1.019,
      "num_input_tokens_seen": 106740750,
      "step": 5017
    },
    {
      "epoch": 0.6033788252269584,
      "flos": 17320786145280.0,
      "grad_norm": 2.9187041840643437,
      "learning_rate": 1.4352440195036847e-06,
      "loss": 0.9727,
      "num_input_tokens_seen": 106760645,
      "step": 5018
    },
    {
      "epoch": 0.6034990681175976,
      "flos": 18160262430720.0,
      "grad_norm": 2.5723735117608384,
      "learning_rate": 1.4344967923484395e-06,
      "loss": 0.986,
      "num_input_tokens_seen": 106782335,
      "step": 5019
    },
    {
      "epoch": 0.6036193110082366,
      "flos": 18474334064640.0,
      "grad_norm": 3.5741688542453094,
      "learning_rate": 1.433749650979581e-06,
      "loss": 0.9162,
      "num_input_tokens_seen": 106802040,
      "step": 5020
    },
    {
      "epoch": 0.6037395538988757,
      "flos": 18212030300160.0,
      "grad_norm": 4.125137789187794,
      "learning_rate": 1.433002595510451e-06,
      "loss": 0.8781,
      "num_input_tokens_seen": 106820540,
      "step": 5021
    },
    {
      "epoch": 0.6038597967895148,
      "flos": 12621952512000.0,
      "grad_norm": 4.266043506255606,
      "learning_rate": 1.4322556260543757e-06,
      "loss": 0.9137,
      "num_input_tokens_seen": 106836835,
      "step": 5022
    },
    {
      "epoch": 0.6039800396801539,
      "flos": 46657801052160.0,
      "grad_norm": 0.933061296264845,
      "learning_rate": 1.4315087427246703e-06,
      "loss": 0.8479,
      "num_input_tokens_seen": 106890380,
      "step": 5023
    },
    {
      "epoch": 0.604100282570793,
      "flos": 48250118860800.0,
      "grad_norm": 0.9009772187633096,
      "learning_rate": 1.4307619456346372e-06,
      "loss": 0.8104,
      "num_input_tokens_seen": 106934405,
      "step": 5024
    },
    {
      "epoch": 0.6042205254614321,
      "flos": 25186005442560.0,
      "grad_norm": 3.810794261134521,
      "learning_rate": 1.430015234897564e-06,
      "loss": 0.9272,
      "num_input_tokens_seen": 106957405,
      "step": 5025
    },
    {
      "epoch": 0.6043407683520712,
      "flos": 32631931146240.0,
      "grad_norm": 6.905554827056434,
      "learning_rate": 1.4292686106267274e-06,
      "loss": 0.8542,
      "num_input_tokens_seen": 106979975,
      "step": 5026
    },
    {
      "epoch": 0.6044610112427102,
      "flos": 11446782197760.0,
      "grad_norm": 2.611437593389645,
      "learning_rate": 1.4285220729353876e-06,
      "loss": 0.9627,
      "num_input_tokens_seen": 106998235,
      "step": 5027
    },
    {
      "epoch": 0.6045812541333494,
      "flos": 9738634567680.0,
      "grad_norm": 3.2982517640094584,
      "learning_rate": 1.4277756219367957e-06,
      "loss": 0.9689,
      "num_input_tokens_seen": 107014980,
      "step": 5028
    },
    {
      "epoch": 0.6047014970239885,
      "flos": 14174275338240.0,
      "grad_norm": 3.72175784242918,
      "learning_rate": 1.4270292577441864e-06,
      "loss": 0.9914,
      "num_input_tokens_seen": 107034205,
      "step": 5029
    },
    {
      "epoch": 0.6048217399146275,
      "flos": 18027684925440.0,
      "grad_norm": 3.1428764206431294,
      "learning_rate": 1.4262829804707836e-06,
      "loss": 0.9134,
      "num_input_tokens_seen": 107055915,
      "step": 5030
    },
    {
      "epoch": 0.6049419828052667,
      "flos": 18526316544000.0,
      "grad_norm": 3.5483559360981762,
      "learning_rate": 1.4255367902297958e-06,
      "loss": 0.886,
      "num_input_tokens_seen": 107076965,
      "step": 5031
    },
    {
      "epoch": 0.6050622256959057,
      "flos": 10345891246080.0,
      "grad_norm": 4.058386730388512,
      "learning_rate": 1.4247906871344215e-06,
      "loss": 0.9905,
      "num_input_tokens_seen": 107092080,
      "step": 5032
    },
    {
      "epoch": 0.6051824685865448,
      "flos": 16586409185280.0,
      "grad_norm": 4.662936740905094,
      "learning_rate": 1.4240446712978415e-06,
      "loss": 0.9495,
      "num_input_tokens_seen": 107110785,
      "step": 5033
    },
    {
      "epoch": 0.605302711477184,
      "flos": 19627698032640.0,
      "grad_norm": 3.220599041535648,
      "learning_rate": 1.423298742833227e-06,
      "loss": 0.9343,
      "num_input_tokens_seen": 107129165,
      "step": 5034
    },
    {
      "epoch": 0.605422954367823,
      "flos": 10709400698880.0,
      "grad_norm": 3.2536565311590446,
      "learning_rate": 1.4225529018537352e-06,
      "loss": 0.9282,
      "num_input_tokens_seen": 107144390,
      "step": 5035
    },
    {
      "epoch": 0.6055431972584621,
      "flos": 19707342643200.0,
      "grad_norm": 3.5310984891272845,
      "learning_rate": 1.4218071484725082e-06,
      "loss": 0.9772,
      "num_input_tokens_seen": 107166230,
      "step": 5036
    },
    {
      "epoch": 0.6056634401491012,
      "flos": 13751486668800.0,
      "grad_norm": 4.564604809337791,
      "learning_rate": 1.4210614828026786e-06,
      "loss": 0.9537,
      "num_input_tokens_seen": 107183800,
      "step": 5037
    },
    {
      "epoch": 0.6057836830397403,
      "flos": 17634060656640.0,
      "grad_norm": 2.3704002409395115,
      "learning_rate": 1.4203159049573605e-06,
      "loss": 0.9402,
      "num_input_tokens_seen": 107204755,
      "step": 5038
    },
    {
      "epoch": 0.6059039259303793,
      "flos": 14593323663360.0,
      "grad_norm": 4.741601043586751,
      "learning_rate": 1.4195704150496593e-06,
      "loss": 1.0634,
      "num_input_tokens_seen": 107222190,
      "step": 5039
    },
    {
      "epoch": 0.6060241688210185,
      "flos": 14961002680320.0,
      "grad_norm": 3.1577629768757887,
      "learning_rate": 1.4188250131926639e-06,
      "loss": 0.9272,
      "num_input_tokens_seen": 107240710,
      "step": 5040
    },
    {
      "epoch": 0.6061444117116576,
      "flos": 11576723066880.0,
      "grad_norm": 3.55562530555127,
      "learning_rate": 1.4180796994994525e-06,
      "loss": 1.0077,
      "num_input_tokens_seen": 107257845,
      "step": 5041
    },
    {
      "epoch": 0.6062646546022966,
      "flos": 15275442216960.0,
      "grad_norm": 3.527355375724889,
      "learning_rate": 1.4173344740830877e-06,
      "loss": 0.9118,
      "num_input_tokens_seen": 107276695,
      "step": 5042
    },
    {
      "epoch": 0.6063848974929358,
      "flos": 27834896363520.0,
      "grad_norm": 3.1230567993998815,
      "learning_rate": 1.4165893370566206e-06,
      "loss": 0.898,
      "num_input_tokens_seen": 107300170,
      "step": 5043
    },
    {
      "epoch": 0.6065051403835748,
      "flos": 13938192752640.0,
      "grad_norm": 3.0010040510354288,
      "learning_rate": 1.4158442885330865e-06,
      "loss": 0.9598,
      "num_input_tokens_seen": 107318460,
      "step": 5044
    },
    {
      "epoch": 0.6066253832742139,
      "flos": 16716840591360.0,
      "grad_norm": 5.739734608433054,
      "learning_rate": 1.4150993286255094e-06,
      "loss": 0.9828,
      "num_input_tokens_seen": 107337430,
      "step": 5045
    },
    {
      "epoch": 0.6067456261648531,
      "flos": 13569318051840.0,
      "grad_norm": 3.6072555858905364,
      "learning_rate": 1.4143544574468993e-06,
      "loss": 0.9782,
      "num_input_tokens_seen": 107355510,
      "step": 5046
    },
    {
      "epoch": 0.6068658690554921,
      "flos": 14566121410560.0,
      "grad_norm": 2.1543590713058203,
      "learning_rate": 1.4136096751102523e-06,
      "loss": 1.0138,
      "num_input_tokens_seen": 107373560,
      "step": 5047
    },
    {
      "epoch": 0.6069861119461312,
      "flos": 19491196231680.0,
      "grad_norm": 3.7597401328275395,
      "learning_rate": 1.4128649817285516e-06,
      "loss": 1.0185,
      "num_input_tokens_seen": 107391415,
      "step": 5048
    },
    {
      "epoch": 0.6071063548367702,
      "flos": 18235338915840.0,
      "grad_norm": 2.642655263650327,
      "learning_rate": 1.412120377414766e-06,
      "loss": 0.8218,
      "num_input_tokens_seen": 107411325,
      "step": 5049
    },
    {
      "epoch": 0.6072265977274094,
      "flos": 17397579509760.0,
      "grad_norm": 2.241846038656816,
      "learning_rate": 1.4113758622818522e-06,
      "loss": 0.8949,
      "num_input_tokens_seen": 107431110,
      "step": 5050
    },
    {
      "epoch": 0.6073468406180484,
      "flos": 12861898076160.0,
      "grad_norm": 3.2504919120008453,
      "learning_rate": 1.410631436442751e-06,
      "loss": 1.0159,
      "num_input_tokens_seen": 107449625,
      "step": 5051
    },
    {
      "epoch": 0.6074670835086875,
      "flos": 14692597862400.0,
      "grad_norm": 3.7460472891654177,
      "learning_rate": 1.4098871000103936e-06,
      "loss": 1.0597,
      "num_input_tokens_seen": 107467945,
      "step": 5052
    },
    {
      "epoch": 0.6075873263993267,
      "flos": 16901645844480.0,
      "grad_norm": 3.253180488164501,
      "learning_rate": 1.409142853097693e-06,
      "loss": 1.0227,
      "num_input_tokens_seen": 107487905,
      "step": 5053
    },
    {
      "epoch": 0.6077075692899657,
      "flos": 17392428871680.0,
      "grad_norm": 4.378219291984034,
      "learning_rate": 1.408398695817553e-06,
      "loss": 0.9889,
      "num_input_tokens_seen": 107504850,
      "step": 5054
    },
    {
      "epoch": 0.6078278121806048,
      "flos": 19497634529280.0,
      "grad_norm": 2.9765753157532506,
      "learning_rate": 1.4076546282828593e-06,
      "loss": 0.8897,
      "num_input_tokens_seen": 107527425,
      "step": 5055
    },
    {
      "epoch": 0.6079480550712439,
      "flos": 27442069217280.0,
      "grad_norm": 5.033849767165639,
      "learning_rate": 1.4069106506064874e-06,
      "loss": 0.8496,
      "num_input_tokens_seen": 107548570,
      "step": 5056
    },
    {
      "epoch": 0.608068297961883,
      "flos": 18026979778560.0,
      "grad_norm": 3.2727105328102857,
      "learning_rate": 1.4061667629012989e-06,
      "loss": 0.9717,
      "num_input_tokens_seen": 107568960,
      "step": 5057
    },
    {
      "epoch": 0.608188540852522,
      "flos": 17211762524160.0,
      "grad_norm": 2.3044595896645115,
      "learning_rate": 1.40542296528014e-06,
      "loss": 1.0225,
      "num_input_tokens_seen": 107588340,
      "step": 5058
    },
    {
      "epoch": 0.6083087837431612,
      "flos": 15114497433600.0,
      "grad_norm": 3.506072090517571,
      "learning_rate": 1.4046792578558452e-06,
      "loss": 0.9548,
      "num_input_tokens_seen": 107605955,
      "step": 5059
    },
    {
      "epoch": 0.6084290266338003,
      "flos": 11656153067520.0,
      "grad_norm": 4.348207450809853,
      "learning_rate": 1.4039356407412325e-06,
      "loss": 0.9562,
      "num_input_tokens_seen": 107618915,
      "step": 5060
    },
    {
      "epoch": 0.6085492695244393,
      "flos": 47573304238080.0,
      "grad_norm": 0.8284149741869982,
      "learning_rate": 1.40319211404911e-06,
      "loss": 0.81,
      "num_input_tokens_seen": 107673635,
      "step": 5061
    },
    {
      "epoch": 0.6086695124150785,
      "flos": 16792468930560.0,
      "grad_norm": 3.181704169868806,
      "learning_rate": 1.4024486778922691e-06,
      "loss": 1.0907,
      "num_input_tokens_seen": 107691670,
      "step": 5062
    },
    {
      "epoch": 0.6087897553057176,
      "flos": 14305013329920.0,
      "grad_norm": 3.847838669533562,
      "learning_rate": 1.4017053323834884e-06,
      "loss": 0.9705,
      "num_input_tokens_seen": 107711220,
      "step": 5063
    },
    {
      "epoch": 0.6089099981963566,
      "flos": 18132140421120.0,
      "grad_norm": 105.06763074919887,
      "learning_rate": 1.4009620776355333e-06,
      "loss": 0.95,
      "num_input_tokens_seen": 107732540,
      "step": 5064
    },
    {
      "epoch": 0.6090302410869958,
      "flos": 18024097873920.0,
      "grad_norm": 2.478461082345341,
      "learning_rate": 1.4002189137611553e-06,
      "loss": 0.9871,
      "num_input_tokens_seen": 107751600,
      "step": 5065
    },
    {
      "epoch": 0.6091504839776348,
      "flos": 17057685258240.0,
      "grad_norm": 8.459075243754024,
      "learning_rate": 1.3994758408730901e-06,
      "loss": 0.8881,
      "num_input_tokens_seen": 107770505,
      "step": 5066
    },
    {
      "epoch": 0.6092707268682739,
      "flos": 21122366545920.0,
      "grad_norm": 4.2159737953804886,
      "learning_rate": 1.3987328590840629e-06,
      "loss": 0.9733,
      "num_input_tokens_seen": 107791170,
      "step": 5067
    },
    {
      "epoch": 0.609390969758913,
      "flos": 17084856852480.0,
      "grad_norm": 4.698539247162771,
      "learning_rate": 1.397989968506783e-06,
      "loss": 1.0584,
      "num_input_tokens_seen": 107809900,
      "step": 5068
    },
    {
      "epoch": 0.6095112126495521,
      "flos": 7796550451200.0,
      "grad_norm": 3.430081216723638,
      "learning_rate": 1.3972471692539458e-06,
      "loss": 0.9202,
      "num_input_tokens_seen": 107824335,
      "step": 5069
    },
    {
      "epoch": 0.6096314555401912,
      "flos": 12225660948480.0,
      "grad_norm": 5.282066507818468,
      "learning_rate": 1.3965044614382348e-06,
      "loss": 0.9535,
      "num_input_tokens_seen": 107839505,
      "step": 5070
    },
    {
      "epoch": 0.6097516984308303,
      "flos": 15374072586240.0,
      "grad_norm": 2.986789066525703,
      "learning_rate": 1.3957618451723162e-06,
      "loss": 0.9433,
      "num_input_tokens_seen": 107855255,
      "step": 5071
    },
    {
      "epoch": 0.6098719413214694,
      "flos": 19366130073600.0,
      "grad_norm": 5.73587823025085,
      "learning_rate": 1.3950193205688457e-06,
      "loss": 0.9006,
      "num_input_tokens_seen": 107874700,
      "step": 5072
    },
    {
      "epoch": 0.6099921842121084,
      "flos": 14488990801920.0,
      "grad_norm": 3.6486855530805595,
      "learning_rate": 1.3942768877404627e-06,
      "loss": 1.0337,
      "num_input_tokens_seen": 107893385,
      "step": 5073
    },
    {
      "epoch": 0.6101124271027476,
      "flos": 16611097436160.0,
      "grad_norm": 2.7119910751578518,
      "learning_rate": 1.393534546799795e-06,
      "loss": 0.9439,
      "num_input_tokens_seen": 107912805,
      "step": 5074
    },
    {
      "epoch": 0.6102326699933867,
      "flos": 18998543032320.0,
      "grad_norm": 2.893856572275288,
      "learning_rate": 1.3927922978594536e-06,
      "loss": 0.8599,
      "num_input_tokens_seen": 107933610,
      "step": 5075
    },
    {
      "epoch": 0.6103529128840257,
      "flos": 43406006968320.0,
      "grad_norm": 0.8344447988968509,
      "learning_rate": 1.3920501410320387e-06,
      "loss": 0.8161,
      "num_input_tokens_seen": 107989445,
      "step": 5076
    },
    {
      "epoch": 0.6104731557746649,
      "flos": 13488140513280.0,
      "grad_norm": 7.343912585467523,
      "learning_rate": 1.3913080764301333e-06,
      "loss": 0.9593,
      "num_input_tokens_seen": 108006125,
      "step": 5077
    },
    {
      "epoch": 0.6105933986653039,
      "flos": 16611741265920.0,
      "grad_norm": 3.129246230115004,
      "learning_rate": 1.3905661041663085e-06,
      "loss": 0.909,
      "num_input_tokens_seen": 108027030,
      "step": 5078
    },
    {
      "epoch": 0.610713641555943,
      "flos": 24711295610880.0,
      "grad_norm": 3.6692809986256165,
      "learning_rate": 1.389824224353122e-06,
      "loss": 0.84,
      "num_input_tokens_seen": 108048340,
      "step": 5079
    },
    {
      "epoch": 0.610833884446582,
      "flos": 18968152289280.0,
      "grad_norm": 3.1638861713020043,
      "learning_rate": 1.389082437103115e-06,
      "loss": 0.9654,
      "num_input_tokens_seen": 108067330,
      "step": 5080
    },
    {
      "epoch": 0.6109541273372212,
      "flos": 15065427517440.0,
      "grad_norm": 4.581351643640819,
      "learning_rate": 1.3883407425288172e-06,
      "loss": 0.9641,
      "num_input_tokens_seen": 108087385,
      "step": 5081
    },
    {
      "epoch": 0.6110743702278603,
      "flos": 14252570972160.0,
      "grad_norm": 5.830367324276089,
      "learning_rate": 1.3875991407427417e-06,
      "loss": 0.975,
      "num_input_tokens_seen": 108105330,
      "step": 5082
    },
    {
      "epoch": 0.6111946131184993,
      "flos": 48909419335680.0,
      "grad_norm": 0.7610483594033262,
      "learning_rate": 1.38685763185739e-06,
      "loss": 0.8032,
      "num_input_tokens_seen": 108158710,
      "step": 5083
    },
    {
      "epoch": 0.6113148560091385,
      "flos": 14147287695360.0,
      "grad_norm": 4.857772513857222,
      "learning_rate": 1.3861162159852476e-06,
      "loss": 0.8632,
      "num_input_tokens_seen": 108176565,
      "step": 5084
    },
    {
      "epoch": 0.6114350988997775,
      "flos": 16874045030400.0,
      "grad_norm": 2.938468286534956,
      "learning_rate": 1.3853748932387875e-06,
      "loss": 0.9897,
      "num_input_tokens_seen": 108196925,
      "step": 5085
    },
    {
      "epoch": 0.6115553417904166,
      "flos": 17083783802880.0,
      "grad_norm": 3.7820282442600637,
      "learning_rate": 1.3846336637304671e-06,
      "loss": 0.9472,
      "num_input_tokens_seen": 108214915,
      "step": 5086
    },
    {
      "epoch": 0.6116755846810558,
      "flos": 16875485982720.0,
      "grad_norm": 3.1721131183711258,
      "learning_rate": 1.3838925275727316e-06,
      "loss": 1.0105,
      "num_input_tokens_seen": 108235375,
      "step": 5087
    },
    {
      "epoch": 0.6117958275716948,
      "flos": 13230343557120.0,
      "grad_norm": 3.0928411208354154,
      "learning_rate": 1.3831514848780089e-06,
      "loss": 0.9835,
      "num_input_tokens_seen": 108254670,
      "step": 5088
    },
    {
      "epoch": 0.6119160704623339,
      "flos": 11655938457600.0,
      "grad_norm": 6.335518142237084,
      "learning_rate": 1.3824105357587152e-06,
      "loss": 1.1105,
      "num_input_tokens_seen": 108271495,
      "step": 5089
    },
    {
      "epoch": 0.612036313352973,
      "flos": 17006162657280.0,
      "grad_norm": 2.633807891223947,
      "learning_rate": 1.381669680327253e-06,
      "loss": 1.0203,
      "num_input_tokens_seen": 108292895,
      "step": 5090
    },
    {
      "epoch": 0.6121565562436121,
      "flos": 19204081582080.0,
      "grad_norm": 10.268494830498986,
      "learning_rate": 1.380928918696008e-06,
      "loss": 0.8998,
      "num_input_tokens_seen": 108311385,
      "step": 5091
    },
    {
      "epoch": 0.6122767991342511,
      "flos": 11079379107840.0,
      "grad_norm": 4.933007429716252,
      "learning_rate": 1.3801882509773548e-06,
      "loss": 0.9063,
      "num_input_tokens_seen": 108328965,
      "step": 5092
    },
    {
      "epoch": 0.6123970420248903,
      "flos": 19915855073280.0,
      "grad_norm": 4.134973433385008,
      "learning_rate": 1.3794476772836503e-06,
      "loss": 1.0037,
      "num_input_tokens_seen": 108349785,
      "step": 5093
    },
    {
      "epoch": 0.6125172849155294,
      "flos": 15247442841600.0,
      "grad_norm": 4.324375552924317,
      "learning_rate": 1.3787071977272402e-06,
      "loss": 1.0317,
      "num_input_tokens_seen": 108368765,
      "step": 5094
    },
    {
      "epoch": 0.6126375278061684,
      "flos": 11495238942720.0,
      "grad_norm": 6.0495946322744425,
      "learning_rate": 1.3779668124204535e-06,
      "loss": 0.8985,
      "num_input_tokens_seen": 108384900,
      "step": 5095
    },
    {
      "epoch": 0.6127577706968076,
      "flos": 14514721443840.0,
      "grad_norm": 3.6740382538508327,
      "learning_rate": 1.3772265214756074e-06,
      "loss": 1.0069,
      "num_input_tokens_seen": 108404380,
      "step": 5096
    },
    {
      "epoch": 0.6128780135874466,
      "flos": 12942002565120.0,
      "grad_norm": 3.811701331359757,
      "learning_rate": 1.3764863250050025e-06,
      "loss": 0.9333,
      "num_input_tokens_seen": 108422340,
      "step": 5097
    },
    {
      "epoch": 0.6129982564780857,
      "flos": 17740386324480.0,
      "grad_norm": 3.107256450361867,
      "learning_rate": 1.3757462231209272e-06,
      "loss": 0.993,
      "num_input_tokens_seen": 108442365,
      "step": 5098
    },
    {
      "epoch": 0.6131184993687249,
      "flos": 15983107461120.0,
      "grad_norm": 4.591025238559342,
      "learning_rate": 1.3750062159356525e-06,
      "loss": 1.077,
      "num_input_tokens_seen": 108461435,
      "step": 5099
    },
    {
      "epoch": 0.6132387422593639,
      "flos": 11235050618880.0,
      "grad_norm": 4.749797013613553,
      "learning_rate": 1.3742663035614382e-06,
      "loss": 1.0222,
      "num_input_tokens_seen": 108478525,
      "step": 5100
    },
    {
      "epoch": 0.613358985150003,
      "flos": 18081384284160.0,
      "grad_norm": 3.425493948186524,
      "learning_rate": 1.3735264861105283e-06,
      "loss": 1.0007,
      "num_input_tokens_seen": 108498885,
      "step": 5101
    },
    {
      "epoch": 0.6134792280406421,
      "flos": 15170496184320.0,
      "grad_norm": 4.950706685801813,
      "learning_rate": 1.372786763695152e-06,
      "loss": 0.9837,
      "num_input_tokens_seen": 108517365,
      "step": 5102
    },
    {
      "epoch": 0.6135994709312812,
      "flos": 15062116392960.0,
      "grad_norm": 4.31399709174598,
      "learning_rate": 1.3720471364275257e-06,
      "loss": 0.967,
      "num_input_tokens_seen": 108536730,
      "step": 5103
    },
    {
      "epoch": 0.6137197138219203,
      "flos": 10450775961600.0,
      "grad_norm": 3.097651478016895,
      "learning_rate": 1.3713076044198486e-06,
      "loss": 0.9729,
      "num_input_tokens_seen": 108553260,
      "step": 5104
    },
    {
      "epoch": 0.6138399567125594,
      "flos": 14252938874880.0,
      "grad_norm": 3.677505396897993,
      "learning_rate": 1.3705681677843086e-06,
      "loss": 0.9968,
      "num_input_tokens_seen": 108571575,
      "step": 5105
    },
    {
      "epoch": 0.6139601996031985,
      "flos": 43035315302400.0,
      "grad_norm": 0.817583467381315,
      "learning_rate": 1.3698288266330768e-06,
      "loss": 0.8209,
      "num_input_tokens_seen": 108631920,
      "step": 5106
    },
    {
      "epoch": 0.6140804424938375,
      "flos": 16769834803200.0,
      "grad_norm": 5.462664024282246,
      "learning_rate": 1.3690895810783113e-06,
      "loss": 0.9231,
      "num_input_tokens_seen": 108650435,
      "step": 5107
    },
    {
      "epoch": 0.6142006853844767,
      "flos": 15197085265920.0,
      "grad_norm": 4.263684848096388,
      "learning_rate": 1.3683504312321543e-06,
      "loss": 0.9085,
      "num_input_tokens_seen": 108670490,
      "step": 5108
    },
    {
      "epoch": 0.6143209282751158,
      "flos": 8483942277120.0,
      "grad_norm": 3.2553619755407963,
      "learning_rate": 1.3676113772067355e-06,
      "loss": 0.986,
      "num_input_tokens_seen": 108687265,
      "step": 5109
    },
    {
      "epoch": 0.6144411711657548,
      "flos": 17844627210240.0,
      "grad_norm": 3.7516785997302367,
      "learning_rate": 1.3668724191141671e-06,
      "loss": 0.9111,
      "num_input_tokens_seen": 108706255,
      "step": 5110
    },
    {
      "epoch": 0.6145614140563939,
      "flos": 14273242951680.0,
      "grad_norm": 4.792587501391716,
      "learning_rate": 1.3661335570665493e-06,
      "loss": 0.8574,
      "num_input_tokens_seen": 108723885,
      "step": 5111
    },
    {
      "epoch": 0.614681656947033,
      "flos": 11892634214400.0,
      "grad_norm": 4.5325643803547395,
      "learning_rate": 1.3653947911759676e-06,
      "loss": 0.8824,
      "num_input_tokens_seen": 108741155,
      "step": 5112
    },
    {
      "epoch": 0.6148018998376721,
      "flos": 27704526274560.0,
      "grad_norm": 2.6872130839645694,
      "learning_rate": 1.3646561215544904e-06,
      "loss": 0.9335,
      "num_input_tokens_seen": 108765515,
      "step": 5113
    },
    {
      "epoch": 0.6149221427283111,
      "flos": 16582638182400.0,
      "grad_norm": 5.871427383131027,
      "learning_rate": 1.363917548314176e-06,
      "loss": 0.9835,
      "num_input_tokens_seen": 108784500,
      "step": 5114
    },
    {
      "epoch": 0.6150423856189503,
      "flos": 15901807288320.0,
      "grad_norm": 2.755862143425257,
      "learning_rate": 1.3631790715670626e-06,
      "loss": 0.9191,
      "num_input_tokens_seen": 108802625,
      "step": 5115
    },
    {
      "epoch": 0.6151626285095894,
      "flos": 13251414097920.0,
      "grad_norm": 4.376535854510441,
      "learning_rate": 1.3624406914251783e-06,
      "loss": 1.0508,
      "num_input_tokens_seen": 108819465,
      "step": 5116
    },
    {
      "epoch": 0.6152828714002284,
      "flos": 11210822246400.0,
      "grad_norm": 4.651177222429109,
      "learning_rate": 1.3617024080005335e-06,
      "loss": 1.07,
      "num_input_tokens_seen": 108836085,
      "step": 5117
    },
    {
      "epoch": 0.6154031142908676,
      "flos": 17692788019200.0,
      "grad_norm": 2.9205497143260715,
      "learning_rate": 1.3609642214051266e-06,
      "loss": 0.9326,
      "num_input_tokens_seen": 108860030,
      "step": 5118
    },
    {
      "epoch": 0.6155233571815066,
      "flos": 13649330565120.0,
      "grad_norm": 3.501638437696281,
      "learning_rate": 1.3602261317509385e-06,
      "loss": 0.859,
      "num_input_tokens_seen": 108876410,
      "step": 5119
    },
    {
      "epoch": 0.6156436000721457,
      "flos": 13308148654080.0,
      "grad_norm": 6.601758676356597,
      "learning_rate": 1.3594881391499387e-06,
      "loss": 1.0097,
      "num_input_tokens_seen": 108895045,
      "step": 5120
    },
    {
      "epoch": 0.6157638429627849,
      "flos": 12784430223360.0,
      "grad_norm": 3.1191198339945774,
      "learning_rate": 1.3587502437140778e-06,
      "loss": 0.9873,
      "num_input_tokens_seen": 108912930,
      "step": 5121
    },
    {
      "epoch": 0.6158840858534239,
      "flos": 18185625169920.0,
      "grad_norm": 7.054473366766783,
      "learning_rate": 1.3580124455552952e-06,
      "loss": 1.0421,
      "num_input_tokens_seen": 108932015,
      "step": 5122
    },
    {
      "epoch": 0.616004328744063,
      "flos": 17526845890560.0,
      "grad_norm": 3.7547968384582013,
      "learning_rate": 1.3572747447855148e-06,
      "loss": 1.0582,
      "num_input_tokens_seen": 108952145,
      "step": 5123
    },
    {
      "epoch": 0.6161245716347021,
      "flos": 15406517452800.0,
      "grad_norm": 3.5491165942454383,
      "learning_rate": 1.356537141516644e-06,
      "loss": 0.8795,
      "num_input_tokens_seen": 108969285,
      "step": 5124
    },
    {
      "epoch": 0.6162448145253412,
      "flos": 25108353638400.0,
      "grad_norm": 3.987804776684872,
      "learning_rate": 1.3557996358605775e-06,
      "loss": 0.8106,
      "num_input_tokens_seen": 108988925,
      "step": 5125
    },
    {
      "epoch": 0.6163650574159802,
      "flos": 15352848752640.0,
      "grad_norm": 4.128345404494226,
      "learning_rate": 1.3550622279291941e-06,
      "loss": 0.898,
      "num_input_tokens_seen": 109006790,
      "step": 5126
    },
    {
      "epoch": 0.6164853003066194,
      "flos": 17478113218560.0,
      "grad_norm": 2.461718335480021,
      "learning_rate": 1.354324917834358e-06,
      "loss": 1.0295,
      "num_input_tokens_seen": 109027755,
      "step": 5127
    },
    {
      "epoch": 0.6166055431972585,
      "flos": 15511708753920.0,
      "grad_norm": 2.993497954969853,
      "learning_rate": 1.353587705687918e-06,
      "loss": 0.9603,
      "num_input_tokens_seen": 109045650,
      "step": 5128
    },
    {
      "epoch": 0.6167257860878975,
      "flos": 12600422092800.0,
      "grad_norm": 5.527424669515891,
      "learning_rate": 1.3528505916017096e-06,
      "loss": 0.8956,
      "num_input_tokens_seen": 109063070,
      "step": 5129
    },
    {
      "epoch": 0.6168460289785367,
      "flos": 16501613936640.0,
      "grad_norm": 4.844463074402368,
      "learning_rate": 1.3521135756875514e-06,
      "loss": 1.0676,
      "num_input_tokens_seen": 109079105,
      "step": 5130
    },
    {
      "epoch": 0.6169662718691757,
      "flos": 18657453096960.0,
      "grad_norm": 2.7070832130542093,
      "learning_rate": 1.3513766580572496e-06,
      "loss": 1.053,
      "num_input_tokens_seen": 109101645,
      "step": 5131
    },
    {
      "epoch": 0.6170865147598148,
      "flos": 13492248760320.0,
      "grad_norm": 4.048170905866682,
      "learning_rate": 1.3506398388225924e-06,
      "loss": 0.9601,
      "num_input_tokens_seen": 109118685,
      "step": 5132
    },
    {
      "epoch": 0.617206757650454,
      "flos": 12941971906560.0,
      "grad_norm": 3.2809121094162754,
      "learning_rate": 1.349903118095355e-06,
      "loss": 0.9046,
      "num_input_tokens_seen": 109137540,
      "step": 5133
    },
    {
      "epoch": 0.617327000541093,
      "flos": 12889161646080.0,
      "grad_norm": 4.91005971930521,
      "learning_rate": 1.349166495987298e-06,
      "loss": 0.9176,
      "num_input_tokens_seen": 109155825,
      "step": 5134
    },
    {
      "epoch": 0.6174472434317321,
      "flos": 44263571804160.0,
      "grad_norm": 0.9146286930720265,
      "learning_rate": 1.348429972610166e-06,
      "loss": 0.8699,
      "num_input_tokens_seen": 109219850,
      "step": 5135
    },
    {
      "epoch": 0.6175674863223712,
      "flos": 51014962237440.0,
      "grad_norm": 0.9751172166446962,
      "learning_rate": 1.3476935480756897e-06,
      "loss": 0.7991,
      "num_input_tokens_seen": 109276320,
      "step": 5136
    },
    {
      "epoch": 0.6176877292130103,
      "flos": 15511248875520.0,
      "grad_norm": 3.377928840524612,
      "learning_rate": 1.346957222495583e-06,
      "loss": 0.9434,
      "num_input_tokens_seen": 109293835,
      "step": 5137
    },
    {
      "epoch": 0.6178079721036493,
      "flos": 12568191836160.0,
      "grad_norm": 5.151221950707348,
      "learning_rate": 1.3462209959815466e-06,
      "loss": 0.9104,
      "num_input_tokens_seen": 109308295,
      "step": 5138
    },
    {
      "epoch": 0.6179282149942885,
      "flos": 16083853271040.0,
      "grad_norm": 4.136196316013011,
      "learning_rate": 1.345484868645265e-06,
      "loss": 0.932,
      "num_input_tokens_seen": 109326825,
      "step": 5139
    },
    {
      "epoch": 0.6180484578849276,
      "flos": 15852614737920.0,
      "grad_norm": 4.8766445418242945,
      "learning_rate": 1.3447488405984088e-06,
      "loss": 0.9743,
      "num_input_tokens_seen": 109344805,
      "step": 5140
    },
    {
      "epoch": 0.6181687007755666,
      "flos": 25134360207360.0,
      "grad_norm": 8.15828705704072,
      "learning_rate": 1.3440129119526322e-06,
      "loss": 0.8874,
      "num_input_tokens_seen": 109366950,
      "step": 5141
    },
    {
      "epoch": 0.6182889436662057,
      "flos": 44053534556160.0,
      "grad_norm": 0.9158146033716204,
      "learning_rate": 1.3432770828195762e-06,
      "loss": 0.7594,
      "num_input_tokens_seen": 109427655,
      "step": 5142
    },
    {
      "epoch": 0.6184091865568448,
      "flos": 13911879598080.0,
      "grad_norm": 9.03790017240549,
      "learning_rate": 1.3425413533108635e-06,
      "loss": 0.911,
      "num_input_tokens_seen": 109445975,
      "step": 5143
    },
    {
      "epoch": 0.6185294294474839,
      "flos": 16769988096000.0,
      "grad_norm": 4.086285723987789,
      "learning_rate": 1.341805723538105e-06,
      "loss": 0.8996,
      "num_input_tokens_seen": 109465800,
      "step": 5144
    },
    {
      "epoch": 0.618649672338123,
      "flos": 19052027781120.0,
      "grad_norm": 5.621637653389434,
      "learning_rate": 1.3410701936128948e-06,
      "loss": 0.9646,
      "num_input_tokens_seen": 109488300,
      "step": 5145
    },
    {
      "epoch": 0.6187699152287621,
      "flos": 10207059394560.0,
      "grad_norm": 5.108967336087689,
      "learning_rate": 1.340334763646812e-06,
      "loss": 1.0454,
      "num_input_tokens_seen": 109502155,
      "step": 5146
    },
    {
      "epoch": 0.6188901581194012,
      "flos": 14253153484800.0,
      "grad_norm": 3.6247800562989982,
      "learning_rate": 1.3395994337514218e-06,
      "loss": 0.9392,
      "num_input_tokens_seen": 109522045,
      "step": 5147
    },
    {
      "epoch": 0.6190104010100402,
      "flos": 18288854323200.0,
      "grad_norm": 2.672449596704085,
      "learning_rate": 1.3388642040382725e-06,
      "loss": 0.9826,
      "num_input_tokens_seen": 109542190,
      "step": 5148
    },
    {
      "epoch": 0.6191306439006794,
      "flos": 21696381235200.0,
      "grad_norm": 5.7644690205064775,
      "learning_rate": 1.3381290746188975e-06,
      "loss": 1.0275,
      "num_input_tokens_seen": 109561280,
      "step": 5149
    },
    {
      "epoch": 0.6192508867913185,
      "flos": 18996795494400.0,
      "grad_norm": 3.9317827649107495,
      "learning_rate": 1.3373940456048152e-06,
      "loss": 0.8648,
      "num_input_tokens_seen": 109581025,
      "step": 5150
    },
    {
      "epoch": 0.6193711296819575,
      "flos": 26210011054080.0,
      "grad_norm": 2.6117225183658874,
      "learning_rate": 1.3366591171075299e-06,
      "loss": 0.7824,
      "num_input_tokens_seen": 109604250,
      "step": 5151
    },
    {
      "epoch": 0.6194913725725967,
      "flos": 17923995893760.0,
      "grad_norm": 3.213768372455369,
      "learning_rate": 1.335924289238529e-06,
      "loss": 1.1021,
      "num_input_tokens_seen": 109623180,
      "step": 5152
    },
    {
      "epoch": 0.6196116154632357,
      "flos": 15039359631360.0,
      "grad_norm": 2.715981926853684,
      "learning_rate": 1.3351895621092859e-06,
      "loss": 0.9564,
      "num_input_tokens_seen": 109643245,
      "step": 5153
    },
    {
      "epoch": 0.6197318583538748,
      "flos": 11499868385280.0,
      "grad_norm": 3.806277215168188,
      "learning_rate": 1.3344549358312567e-06,
      "loss": 0.9559,
      "num_input_tokens_seen": 109661365,
      "step": 5154
    },
    {
      "epoch": 0.619852101244514,
      "flos": 17372707307520.0,
      "grad_norm": 6.120546551402956,
      "learning_rate": 1.3337204105158852e-06,
      "loss": 0.9663,
      "num_input_tokens_seen": 109679955,
      "step": 5155
    },
    {
      "epoch": 0.619972344135153,
      "flos": 11840099880960.0,
      "grad_norm": 3.1765026291413476,
      "learning_rate": 1.332985986274597e-06,
      "loss": 0.906,
      "num_input_tokens_seen": 109697305,
      "step": 5156
    },
    {
      "epoch": 0.6200925870257921,
      "flos": 8798596423680.0,
      "grad_norm": 6.1246966851317035,
      "learning_rate": 1.3322516632188047e-06,
      "loss": 0.9456,
      "num_input_tokens_seen": 109713920,
      "step": 5157
    },
    {
      "epoch": 0.6202128299164312,
      "flos": 18891481559040.0,
      "grad_norm": 3.152448725695574,
      "learning_rate": 1.3315174414599045e-06,
      "loss": 0.8682,
      "num_input_tokens_seen": 109734960,
      "step": 5158
    },
    {
      "epoch": 0.6203330728070703,
      "flos": 13308516556800.0,
      "grad_norm": 4.289265056336661,
      "learning_rate": 1.3307833211092768e-06,
      "loss": 0.9455,
      "num_input_tokens_seen": 109753345,
      "step": 5159
    },
    {
      "epoch": 0.6204533156977093,
      "flos": 14644539678720.0,
      "grad_norm": 3.800350152999114,
      "learning_rate": 1.3300493022782873e-06,
      "loss": 0.9367,
      "num_input_tokens_seen": 109773635,
      "step": 5160
    },
    {
      "epoch": 0.6205735585883485,
      "flos": 12286289141760.0,
      "grad_norm": 3.29747430497831,
      "learning_rate": 1.3293153850782855e-06,
      "loss": 0.9085,
      "num_input_tokens_seen": 109791675,
      "step": 5161
    },
    {
      "epoch": 0.6206938014789876,
      "flos": 16322848419840.0,
      "grad_norm": 3.385463037417802,
      "learning_rate": 1.3285815696206069e-06,
      "loss": 0.9043,
      "num_input_tokens_seen": 109812940,
      "step": 5162
    },
    {
      "epoch": 0.6208140443696266,
      "flos": 17054128865280.0,
      "grad_norm": 3.2433475487277614,
      "learning_rate": 1.32784785601657e-06,
      "loss": 0.9606,
      "num_input_tokens_seen": 109832070,
      "step": 5163
    },
    {
      "epoch": 0.6209342872602658,
      "flos": 25027574661120.0,
      "grad_norm": 3.849934510503842,
      "learning_rate": 1.3271142443774798e-06,
      "loss": 0.9392,
      "num_input_tokens_seen": 109854025,
      "step": 5164
    },
    {
      "epoch": 0.6210545301509048,
      "flos": 19207975219200.0,
      "grad_norm": 3.5570905339063623,
      "learning_rate": 1.3263807348146228e-06,
      "loss": 1.012,
      "num_input_tokens_seen": 109873600,
      "step": 5165
    },
    {
      "epoch": 0.6211747730415439,
      "flos": 23979034091520.0,
      "grad_norm": 9.796971954888065,
      "learning_rate": 1.3256473274392733e-06,
      "loss": 0.9281,
      "num_input_tokens_seen": 109894665,
      "step": 5166
    },
    {
      "epoch": 0.6212950159321831,
      "flos": 24373700751360.0,
      "grad_norm": 3.6180043297230893,
      "learning_rate": 1.3249140223626873e-06,
      "loss": 0.8945,
      "num_input_tokens_seen": 109916005,
      "step": 5167
    },
    {
      "epoch": 0.6214152588228221,
      "flos": 19917081415680.0,
      "grad_norm": 2.8647598674593246,
      "learning_rate": 1.3241808196961077e-06,
      "loss": 0.9598,
      "num_input_tokens_seen": 109936850,
      "step": 5168
    },
    {
      "epoch": 0.6215355017134612,
      "flos": 14357608980480.0,
      "grad_norm": 3.99065009746262,
      "learning_rate": 1.3234477195507608e-06,
      "loss": 0.9002,
      "num_input_tokens_seen": 109955400,
      "step": 5169
    },
    {
      "epoch": 0.6216557446041003,
      "flos": 29592113909760.0,
      "grad_norm": 4.399278035320758,
      "learning_rate": 1.322714722037857e-06,
      "loss": 0.8232,
      "num_input_tokens_seen": 109976565,
      "step": 5170
    },
    {
      "epoch": 0.6217759874947394,
      "flos": 19890431016960.0,
      "grad_norm": 3.817741418185995,
      "learning_rate": 1.321981827268591e-06,
      "loss": 0.9573,
      "num_input_tokens_seen": 109996940,
      "step": 5171
    },
    {
      "epoch": 0.6218962303853784,
      "flos": 15461535129600.0,
      "grad_norm": 3.846208855895814,
      "learning_rate": 1.3212490353541426e-06,
      "loss": 1.0138,
      "num_input_tokens_seen": 110018920,
      "step": 5172
    },
    {
      "epoch": 0.6220164732760175,
      "flos": 15087080570880.0,
      "grad_norm": 3.2958343454416474,
      "learning_rate": 1.3205163464056762e-06,
      "loss": 0.9962,
      "num_input_tokens_seen": 110035245,
      "step": 5173
    },
    {
      "epoch": 0.6221367161666567,
      "flos": 18601638297600.0,
      "grad_norm": 4.099468866696474,
      "learning_rate": 1.319783760534339e-06,
      "loss": 0.9141,
      "num_input_tokens_seen": 110054210,
      "step": 5174
    },
    {
      "epoch": 0.6222569590572957,
      "flos": 11520938926080.0,
      "grad_norm": 5.490799751605539,
      "learning_rate": 1.319051277851266e-06,
      "loss": 0.9513,
      "num_input_tokens_seen": 110070215,
      "step": 5175
    },
    {
      "epoch": 0.6223772019479348,
      "flos": 12915720069120.0,
      "grad_norm": 3.9150103184745513,
      "learning_rate": 1.3183188984675716e-06,
      "loss": 1.0321,
      "num_input_tokens_seen": 110088300,
      "step": 5176
    },
    {
      "epoch": 0.6224974448385739,
      "flos": 19574795796480.0,
      "grad_norm": 5.268151104647756,
      "learning_rate": 1.3175866224943586e-06,
      "loss": 0.9198,
      "num_input_tokens_seen": 110106740,
      "step": 5177
    },
    {
      "epoch": 0.622617687729213,
      "flos": 14042586931200.0,
      "grad_norm": 5.267490497973417,
      "learning_rate": 1.316854450042712e-06,
      "loss": 0.929,
      "num_input_tokens_seen": 110124400,
      "step": 5178
    },
    {
      "epoch": 0.622737930619852,
      "flos": 16376057241600.0,
      "grad_norm": 4.084239366142877,
      "learning_rate": 1.3161223812237024e-06,
      "loss": 0.9402,
      "num_input_tokens_seen": 110143475,
      "step": 5179
    },
    {
      "epoch": 0.6228581735104912,
      "flos": 8897625354240.0,
      "grad_norm": 4.031528197159507,
      "learning_rate": 1.3153904161483842e-06,
      "loss": 1.0417,
      "num_input_tokens_seen": 110158495,
      "step": 5180
    },
    {
      "epoch": 0.6229784164011303,
      "flos": 16924770508800.0,
      "grad_norm": 4.160659859849199,
      "learning_rate": 1.3146585549277953e-06,
      "loss": 1.0425,
      "num_input_tokens_seen": 110176855,
      "step": 5181
    },
    {
      "epoch": 0.6230986592917693,
      "flos": 15926526197760.0,
      "grad_norm": 3.950565036969791,
      "learning_rate": 1.3139267976729591e-06,
      "loss": 0.9721,
      "num_input_tokens_seen": 110196765,
      "step": 5182
    },
    {
      "epoch": 0.6232189021824085,
      "flos": 24632724049920.0,
      "grad_norm": 3.7240129942539233,
      "learning_rate": 1.3131951444948815e-06,
      "loss": 0.9106,
      "num_input_tokens_seen": 110215885,
      "step": 5183
    },
    {
      "epoch": 0.6233391450730476,
      "flos": 15799865794560.0,
      "grad_norm": 6.556437571327053,
      "learning_rate": 1.3124635955045546e-06,
      "loss": 0.9357,
      "num_input_tokens_seen": 110235420,
      "step": 5184
    },
    {
      "epoch": 0.6234593879636866,
      "flos": 14430324756480.0,
      "grad_norm": 3.6619889303109883,
      "learning_rate": 1.3117321508129537e-06,
      "loss": 1.0358,
      "num_input_tokens_seen": 110253220,
      "step": 5185
    },
    {
      "epoch": 0.6235796308543258,
      "flos": 14669595832320.0,
      "grad_norm": 2.2848820448465834,
      "learning_rate": 1.3110008105310388e-06,
      "loss": 0.9617,
      "num_input_tokens_seen": 110272760,
      "step": 5186
    },
    {
      "epoch": 0.6236998737449648,
      "flos": 18949319823360.0,
      "grad_norm": 3.673679898408635,
      "learning_rate": 1.3102695747697526e-06,
      "loss": 0.9736,
      "num_input_tokens_seen": 110295350,
      "step": 5187
    },
    {
      "epoch": 0.6238201166356039,
      "flos": 8927126999040.0,
      "grad_norm": 7.740470934594783,
      "learning_rate": 1.3095384436400237e-06,
      "loss": 1.0863,
      "num_input_tokens_seen": 110306600,
      "step": 5188
    },
    {
      "epoch": 0.623940359526243,
      "flos": 7330333040640.0,
      "grad_norm": 3.2741912483838895,
      "learning_rate": 1.3088074172527633e-06,
      "loss": 1.012,
      "num_input_tokens_seen": 110323450,
      "step": 5189
    },
    {
      "epoch": 0.6240606024168821,
      "flos": 20703042293760.0,
      "grad_norm": 3.9448006175124073,
      "learning_rate": 1.3080764957188684e-06,
      "loss": 0.9093,
      "num_input_tokens_seen": 110343415,
      "step": 5190
    },
    {
      "epoch": 0.6241808453075212,
      "flos": 15642538721280.0,
      "grad_norm": 3.5058875292654337,
      "learning_rate": 1.3073456791492192e-06,
      "loss": 0.8937,
      "num_input_tokens_seen": 110362845,
      "step": 5191
    },
    {
      "epoch": 0.6243010881981603,
      "flos": 15010041937920.0,
      "grad_norm": 3.606087512714164,
      "learning_rate": 1.3066149676546801e-06,
      "loss": 0.9814,
      "num_input_tokens_seen": 110380745,
      "step": 5192
    },
    {
      "epoch": 0.6244213310887994,
      "flos": 15877241671680.0,
      "grad_norm": 3.509175806685114,
      "learning_rate": 1.3058843613460985e-06,
      "loss": 0.8531,
      "num_input_tokens_seen": 110398405,
      "step": 5193
    },
    {
      "epoch": 0.6245415739794384,
      "flos": 10765522083840.0,
      "grad_norm": 3.78955414718677,
      "learning_rate": 1.3051538603343075e-06,
      "loss": 0.9351,
      "num_input_tokens_seen": 110416055,
      "step": 5194
    },
    {
      "epoch": 0.6246618168700776,
      "flos": 13387118776320.0,
      "grad_norm": 4.244594502306866,
      "learning_rate": 1.3044234647301235e-06,
      "loss": 0.8712,
      "num_input_tokens_seen": 110433800,
      "step": 5195
    },
    {
      "epoch": 0.6247820597607167,
      "flos": 10109440757760.0,
      "grad_norm": 3.083070915671115,
      "learning_rate": 1.303693174644347e-06,
      "loss": 0.9102,
      "num_input_tokens_seen": 110450995,
      "step": 5196
    },
    {
      "epoch": 0.6249023026513557,
      "flos": 16087992176640.0,
      "grad_norm": 4.047326022315005,
      "learning_rate": 1.3029629901877625e-06,
      "loss": 1.0007,
      "num_input_tokens_seen": 110470090,
      "step": 5197
    },
    {
      "epoch": 0.6250225455419949,
      "flos": 14383830159360.0,
      "grad_norm": 6.001844440522632,
      "learning_rate": 1.3022329114711376e-06,
      "loss": 0.9711,
      "num_input_tokens_seen": 110488520,
      "step": 5198
    },
    {
      "epoch": 0.6251427884326339,
      "flos": 16662466744320.0,
      "grad_norm": 8.5578663115092,
      "learning_rate": 1.3015029386052256e-06,
      "loss": 0.8824,
      "num_input_tokens_seen": 110508410,
      "step": 5199
    },
    {
      "epoch": 0.625263031323273,
      "flos": 22617280327680.0,
      "grad_norm": 4.27133255188575,
      "learning_rate": 1.3007730717007622e-06,
      "loss": 0.9205,
      "num_input_tokens_seen": 110528945,
      "step": 5200
    },
    {
      "epoch": 0.6253832742139122,
      "flos": 17163459072000.0,
      "grad_norm": 3.362905944371034,
      "learning_rate": 1.3000433108684676e-06,
      "loss": 0.9526,
      "num_input_tokens_seen": 110549165,
      "step": 5201
    },
    {
      "epoch": 0.6255035171045512,
      "flos": 19702989127680.0,
      "grad_norm": 5.63728076990669,
      "learning_rate": 1.2993136562190467e-06,
      "loss": 0.9884,
      "num_input_tokens_seen": 110568005,
      "step": 5202
    },
    {
      "epoch": 0.6256237599951903,
      "flos": 14355554856960.0,
      "grad_norm": 2.4765516214703407,
      "learning_rate": 1.2985841078631871e-06,
      "loss": 0.8913,
      "num_input_tokens_seen": 110587045,
      "step": 5203
    },
    {
      "epoch": 0.6257440028858293,
      "flos": 17189465640960.0,
      "grad_norm": 2.8662254112177146,
      "learning_rate": 1.2978546659115608e-06,
      "loss": 0.9767,
      "num_input_tokens_seen": 110604845,
      "step": 5204
    },
    {
      "epoch": 0.6258642457764685,
      "flos": 11211036856320.0,
      "grad_norm": 3.739185634423517,
      "learning_rate": 1.2971253304748228e-06,
      "loss": 1.0486,
      "num_input_tokens_seen": 110622280,
      "step": 5205
    },
    {
      "epoch": 0.6259844886671075,
      "flos": 8378996244480.0,
      "grad_norm": 5.108276062354465,
      "learning_rate": 1.296396101663614e-06,
      "loss": 0.9493,
      "num_input_tokens_seen": 110638560,
      "step": 5206
    },
    {
      "epoch": 0.6261047315577466,
      "flos": 11236859473920.0,
      "grad_norm": 4.703463824160917,
      "learning_rate": 1.2956669795885565e-06,
      "loss": 1.0318,
      "num_input_tokens_seen": 110654910,
      "step": 5207
    },
    {
      "epoch": 0.6262249744483858,
      "flos": 22276558295040.0,
      "grad_norm": 8.30113188597669,
      "learning_rate": 1.294937964360259e-06,
      "loss": 0.8795,
      "num_input_tokens_seen": 110674900,
      "step": 5208
    },
    {
      "epoch": 0.6263452173390248,
      "flos": 19365608878080.0,
      "grad_norm": 10.585390827097616,
      "learning_rate": 1.2942090560893108e-06,
      "loss": 0.8859,
      "num_input_tokens_seen": 110694025,
      "step": 5209
    },
    {
      "epoch": 0.6264654602296639,
      "flos": 26655587143680.0,
      "grad_norm": 3.72279855590248,
      "learning_rate": 1.2934802548862882e-06,
      "loss": 0.7998,
      "num_input_tokens_seen": 110716530,
      "step": 5210
    },
    {
      "epoch": 0.626585703120303,
      "flos": 10424033587200.0,
      "grad_norm": 3.3493300607851566,
      "learning_rate": 1.292751560861749e-06,
      "loss": 1.0281,
      "num_input_tokens_seen": 110731155,
      "step": 5211
    },
    {
      "epoch": 0.6267059460109421,
      "flos": 15879265136640.0,
      "grad_norm": 5.618988724539232,
      "learning_rate": 1.2920229741262354e-06,
      "loss": 0.9889,
      "num_input_tokens_seen": 110748880,
      "step": 5212
    },
    {
      "epoch": 0.6268261889015811,
      "flos": 12574231572480.0,
      "grad_norm": 4.2958967725261035,
      "learning_rate": 1.2912944947902739e-06,
      "loss": 0.9417,
      "num_input_tokens_seen": 110765085,
      "step": 5213
    },
    {
      "epoch": 0.6269464317922203,
      "flos": 23423637258240.0,
      "grad_norm": 7.265582472153762,
      "learning_rate": 1.2905661229643742e-06,
      "loss": 0.9075,
      "num_input_tokens_seen": 110784565,
      "step": 5214
    },
    {
      "epoch": 0.6270666746828594,
      "flos": 12703589928960.0,
      "grad_norm": 10.742669023298662,
      "learning_rate": 1.2898378587590299e-06,
      "loss": 1.0167,
      "num_input_tokens_seen": 110800885,
      "step": 5215
    },
    {
      "epoch": 0.6271869175734984,
      "flos": 12364799385600.0,
      "grad_norm": 4.96648120064472,
      "learning_rate": 1.2891097022847173e-06,
      "loss": 1.0647,
      "num_input_tokens_seen": 110817950,
      "step": 5216
    },
    {
      "epoch": 0.6273071604641376,
      "flos": 19128238632960.0,
      "grad_norm": 7.134865071678963,
      "learning_rate": 1.2883816536518978e-06,
      "loss": 0.8632,
      "num_input_tokens_seen": 110838810,
      "step": 5217
    },
    {
      "epoch": 0.6274274033547766,
      "flos": 18548337500160.0,
      "grad_norm": 2.9332436099924473,
      "learning_rate": 1.2876537129710155e-06,
      "loss": 1.0111,
      "num_input_tokens_seen": 110856260,
      "step": 5218
    },
    {
      "epoch": 0.6275476462454157,
      "flos": 14383462256640.0,
      "grad_norm": 3.337067245258543,
      "learning_rate": 1.286925880352499e-06,
      "loss": 0.9401,
      "num_input_tokens_seen": 110874840,
      "step": 5219
    },
    {
      "epoch": 0.6276678891360549,
      "flos": 18734123827200.0,
      "grad_norm": 2.6237007740595226,
      "learning_rate": 1.2861981559067592e-06,
      "loss": 0.8931,
      "num_input_tokens_seen": 110895165,
      "step": 5220
    },
    {
      "epoch": 0.6277881320266939,
      "flos": 9816439664640.0,
      "grad_norm": 3.8081560245253745,
      "learning_rate": 1.2854705397441917e-06,
      "loss": 0.994,
      "num_input_tokens_seen": 110910425,
      "step": 5221
    },
    {
      "epoch": 0.627908374917333,
      "flos": 19257535672320.0,
      "grad_norm": 8.298582889042478,
      "learning_rate": 1.2847430319751747e-06,
      "loss": 0.9586,
      "num_input_tokens_seen": 110928240,
      "step": 5222
    },
    {
      "epoch": 0.6280286178079721,
      "flos": 16901002014720.0,
      "grad_norm": 3.735473726556244,
      "learning_rate": 1.2840156327100712e-06,
      "loss": 0.8765,
      "num_input_tokens_seen": 110945085,
      "step": 5223
    },
    {
      "epoch": 0.6281488606986112,
      "flos": 18762276495360.0,
      "grad_norm": 5.75486748377922,
      "learning_rate": 1.2832883420592272e-06,
      "loss": 0.913,
      "num_input_tokens_seen": 110963700,
      "step": 5224
    },
    {
      "epoch": 0.6282691035892503,
      "flos": 25789552435200.0,
      "grad_norm": 2.902231042247155,
      "learning_rate": 1.282561160132972e-06,
      "loss": 0.8312,
      "num_input_tokens_seen": 110983940,
      "step": 5225
    },
    {
      "epoch": 0.6283893464798894,
      "flos": 18889672704000.0,
      "grad_norm": 2.794217856059907,
      "learning_rate": 1.2818340870416186e-06,
      "loss": 1.0073,
      "num_input_tokens_seen": 111004795,
      "step": 5226
    },
    {
      "epoch": 0.6285095893705285,
      "flos": 15800049745920.0,
      "grad_norm": 4.158282937233106,
      "learning_rate": 1.2811071228954626e-06,
      "loss": 0.9508,
      "num_input_tokens_seen": 111023150,
      "step": 5227
    },
    {
      "epoch": 0.6286298322611675,
      "flos": 18893964902400.0,
      "grad_norm": 4.71449259514611,
      "learning_rate": 1.2803802678047846e-06,
      "loss": 1.0035,
      "num_input_tokens_seen": 111043020,
      "step": 5228
    },
    {
      "epoch": 0.6287500751518067,
      "flos": 15481839206400.0,
      "grad_norm": 3.3818737251534543,
      "learning_rate": 1.279653521879848e-06,
      "loss": 0.9228,
      "num_input_tokens_seen": 111062805,
      "step": 5229
    },
    {
      "epoch": 0.6288703180424458,
      "flos": 14198840954880.0,
      "grad_norm": 3.9234561240644066,
      "learning_rate": 1.2789268852308997e-06,
      "loss": 1.0349,
      "num_input_tokens_seen": 111077735,
      "step": 5230
    },
    {
      "epoch": 0.6289905609330848,
      "flos": 15718964183040.0,
      "grad_norm": 4.109100448661773,
      "learning_rate": 1.2782003579681688e-06,
      "loss": 0.8922,
      "num_input_tokens_seen": 111096985,
      "step": 5231
    },
    {
      "epoch": 0.629110803823724,
      "flos": 18157901721600.0,
      "grad_norm": 2.9067329907797688,
      "learning_rate": 1.2774739402018701e-06,
      "loss": 0.9378,
      "num_input_tokens_seen": 111117540,
      "step": 5232
    },
    {
      "epoch": 0.629231046714363,
      "flos": 14829314273280.0,
      "grad_norm": 3.3410849244811502,
      "learning_rate": 1.2767476320422002e-06,
      "loss": 0.9378,
      "num_input_tokens_seen": 111137185,
      "step": 5233
    },
    {
      "epoch": 0.6293512896050021,
      "flos": 46575756963840.0,
      "grad_norm": 0.7475160553014913,
      "learning_rate": 1.2760214335993392e-06,
      "loss": 0.7892,
      "num_input_tokens_seen": 111203550,
      "step": 5234
    },
    {
      "epoch": 0.6294715324956413,
      "flos": 24740766597120.0,
      "grad_norm": 3.0249694570465424,
      "learning_rate": 1.2752953449834514e-06,
      "loss": 0.7738,
      "num_input_tokens_seen": 111225720,
      "step": 5235
    },
    {
      "epoch": 0.6295917753862803,
      "flos": 16193122160640.0,
      "grad_norm": 3.618983043486522,
      "learning_rate": 1.2745693663046836e-06,
      "loss": 0.9924,
      "num_input_tokens_seen": 111244510,
      "step": 5236
    },
    {
      "epoch": 0.6297120182769194,
      "flos": 14803369021440.0,
      "grad_norm": 4.240970893758186,
      "learning_rate": 1.2738434976731662e-06,
      "loss": 0.9943,
      "num_input_tokens_seen": 111262415,
      "step": 5237
    },
    {
      "epoch": 0.6298322611675584,
      "flos": 13830763376640.0,
      "grad_norm": 2.3940337568180152,
      "learning_rate": 1.2731177391990125e-06,
      "loss": 0.9438,
      "num_input_tokens_seen": 111282060,
      "step": 5238
    },
    {
      "epoch": 0.6299525040581976,
      "flos": 8877106667520.0,
      "grad_norm": 3.9517722246499907,
      "learning_rate": 1.2723920909923203e-06,
      "loss": 1.0102,
      "num_input_tokens_seen": 111297525,
      "step": 5239
    },
    {
      "epoch": 0.6300727469488366,
      "flos": 41308220682240.0,
      "grad_norm": 0.9144287027897018,
      "learning_rate": 1.2716665531631688e-06,
      "loss": 0.8465,
      "num_input_tokens_seen": 111351530,
      "step": 5240
    },
    {
      "epoch": 0.6301929898394757,
      "flos": 16008286248960.0,
      "grad_norm": 2.383603224333842,
      "learning_rate": 1.270941125821623e-06,
      "loss": 0.9589,
      "num_input_tokens_seen": 111371675,
      "step": 5241
    },
    {
      "epoch": 0.6303132327301149,
      "flos": 20152581488640.0,
      "grad_norm": 3.5641981394242372,
      "learning_rate": 1.2702158090777278e-06,
      "loss": 0.9517,
      "num_input_tokens_seen": 111392485,
      "step": 5242
    },
    {
      "epoch": 0.6304334756207539,
      "flos": 17975579811840.0,
      "grad_norm": 4.08689877815989,
      "learning_rate": 1.2694906030415148e-06,
      "loss": 0.9422,
      "num_input_tokens_seen": 111409950,
      "step": 5243
    },
    {
      "epoch": 0.630553718511393,
      "flos": 12777899950080.0,
      "grad_norm": 7.072094097443536,
      "learning_rate": 1.2687655078229958e-06,
      "loss": 0.9996,
      "num_input_tokens_seen": 111424000,
      "step": 5244
    },
    {
      "epoch": 0.6306739614020321,
      "flos": 19440899973120.0,
      "grad_norm": 6.01856158031915,
      "learning_rate": 1.2680405235321678e-06,
      "loss": 0.8988,
      "num_input_tokens_seen": 111445055,
      "step": 5245
    },
    {
      "epoch": 0.6307942042926712,
      "flos": 10844185620480.0,
      "grad_norm": 4.679176977774088,
      "learning_rate": 1.267315650279011e-06,
      "loss": 0.982,
      "num_input_tokens_seen": 111463245,
      "step": 5246
    },
    {
      "epoch": 0.6309144471833102,
      "flos": 13908813742080.0,
      "grad_norm": 3.744334964554083,
      "learning_rate": 1.2665908881734874e-06,
      "loss": 0.9257,
      "num_input_tokens_seen": 111481800,
      "step": 5247
    },
    {
      "epoch": 0.6310346900739494,
      "flos": 12390622003200.0,
      "grad_norm": 5.347380715315337,
      "learning_rate": 1.2658662373255432e-06,
      "loss": 1.0418,
      "num_input_tokens_seen": 111499910,
      "step": 5248
    },
    {
      "epoch": 0.6311549329645885,
      "flos": 39400819507200.0,
      "grad_norm": 0.8073352842955742,
      "learning_rate": 1.2651416978451063e-06,
      "loss": 0.7632,
      "num_input_tokens_seen": 111565015,
      "step": 5249
    },
    {
      "epoch": 0.6312751758552275,
      "flos": 29933081210880.0,
      "grad_norm": 3.824337451873302,
      "learning_rate": 1.2644172698420903e-06,
      "loss": 0.8466,
      "num_input_tokens_seen": 111586840,
      "step": 5250
    },
    {
      "epoch": 0.6313954187458667,
      "flos": 13938376704000.0,
      "grad_norm": 2.9252422330274133,
      "learning_rate": 1.2636929534263892e-06,
      "loss": 1.0319,
      "num_input_tokens_seen": 111605545,
      "step": 5251
    },
    {
      "epoch": 0.6315156616365057,
      "flos": 16272306892800.0,
      "grad_norm": 3.4285544781038806,
      "learning_rate": 1.2629687487078821e-06,
      "loss": 0.9626,
      "num_input_tokens_seen": 111624075,
      "step": 5252
    },
    {
      "epoch": 0.6316359045271448,
      "flos": 16869262295040.0,
      "grad_norm": 5.333577428636042,
      "learning_rate": 1.2622446557964293e-06,
      "loss": 0.9549,
      "num_input_tokens_seen": 111641800,
      "step": 5253
    },
    {
      "epoch": 0.631756147417784,
      "flos": 23612765368320.0,
      "grad_norm": 3.2317221261574764,
      "learning_rate": 1.261520674801876e-06,
      "loss": 0.9032,
      "num_input_tokens_seen": 111662115,
      "step": 5254
    },
    {
      "epoch": 0.631876390308423,
      "flos": 22276129075200.0,
      "grad_norm": 3.0966444251086758,
      "learning_rate": 1.2607968058340488e-06,
      "loss": 0.9099,
      "num_input_tokens_seen": 111681530,
      "step": 5255
    },
    {
      "epoch": 0.6319966331990621,
      "flos": 17556378193920.0,
      "grad_norm": 4.155895476171448,
      "learning_rate": 1.2600730490027583e-06,
      "loss": 0.9222,
      "num_input_tokens_seen": 111701490,
      "step": 5256
    },
    {
      "epoch": 0.6321168760897012,
      "flos": 12389273026560.0,
      "grad_norm": 2.595503432699104,
      "learning_rate": 1.2593494044177984e-06,
      "loss": 0.9999,
      "num_input_tokens_seen": 111719515,
      "step": 5257
    },
    {
      "epoch": 0.6322371189803403,
      "flos": 12966936084480.0,
      "grad_norm": 4.270420829886799,
      "learning_rate": 1.2586258721889448e-06,
      "loss": 0.9921,
      "num_input_tokens_seen": 111736585,
      "step": 5258
    },
    {
      "epoch": 0.6323573618709794,
      "flos": 14305197281280.0,
      "grad_norm": 4.170436882716799,
      "learning_rate": 1.2579024524259573e-06,
      "loss": 1.0054,
      "num_input_tokens_seen": 111752565,
      "step": 5259
    },
    {
      "epoch": 0.6324776047616185,
      "flos": 14221904302080.0,
      "grad_norm": 3.7928583696767215,
      "learning_rate": 1.2571791452385768e-06,
      "loss": 1.1049,
      "num_input_tokens_seen": 111769550,
      "step": 5260
    },
    {
      "epoch": 0.6325978476522576,
      "flos": 21988677181440.0,
      "grad_norm": 3.995922633112325,
      "learning_rate": 1.2564559507365301e-06,
      "loss": 0.9643,
      "num_input_tokens_seen": 111791675,
      "step": 5261
    },
    {
      "epoch": 0.6327180905428966,
      "flos": 17451401502720.0,
      "grad_norm": 3.2776484023889165,
      "learning_rate": 1.2557328690295244e-06,
      "loss": 0.9806,
      "num_input_tokens_seen": 111809585,
      "step": 5262
    },
    {
      "epoch": 0.6328383334335358,
      "flos": 15323623034880.0,
      "grad_norm": 7.027539168350136,
      "learning_rate": 1.255009900227251e-06,
      "loss": 0.9539,
      "num_input_tokens_seen": 111828330,
      "step": 5263
    },
    {
      "epoch": 0.6329585763241748,
      "flos": 16297761607680.0,
      "grad_norm": 3.2981540314172126,
      "learning_rate": 1.254287044439383e-06,
      "loss": 0.9869,
      "num_input_tokens_seen": 111847655,
      "step": 5264
    },
    {
      "epoch": 0.6330788192148139,
      "flos": 50803322634240.0,
      "grad_norm": 1.053315464081473,
      "learning_rate": 1.2535643017755776e-06,
      "loss": 0.7633,
      "num_input_tokens_seen": 111909565,
      "step": 5265
    },
    {
      "epoch": 0.6331990621054531,
      "flos": 15085578301440.0,
      "grad_norm": 3.8832801214891934,
      "learning_rate": 1.2528416723454737e-06,
      "loss": 0.917,
      "num_input_tokens_seen": 111925955,
      "step": 5266
    },
    {
      "epoch": 0.6333193049960921,
      "flos": 24583838085120.0,
      "grad_norm": 7.407311472094913,
      "learning_rate": 1.2521191562586945e-06,
      "loss": 0.9005,
      "num_input_tokens_seen": 111949325,
      "step": 5267
    },
    {
      "epoch": 0.6334395478867312,
      "flos": 12991471042560.0,
      "grad_norm": 3.5168709814612393,
      "learning_rate": 1.2513967536248445e-06,
      "loss": 0.964,
      "num_input_tokens_seen": 111965365,
      "step": 5268
    },
    {
      "epoch": 0.6335597907773702,
      "flos": 16796025323520.0,
      "grad_norm": 3.558911574206021,
      "learning_rate": 1.2506744645535117e-06,
      "loss": 1.0108,
      "num_input_tokens_seen": 111985515,
      "step": 5269
    },
    {
      "epoch": 0.6336800336680094,
      "flos": 16140403875840.0,
      "grad_norm": 12.4392352403296,
      "learning_rate": 1.249952289154267e-06,
      "loss": 0.7879,
      "num_input_tokens_seen": 112005275,
      "step": 5270
    },
    {
      "epoch": 0.6338002765586485,
      "flos": 16795688079360.0,
      "grad_norm": 3.141410160133416,
      "learning_rate": 1.2492302275366635e-06,
      "loss": 0.9618,
      "num_input_tokens_seen": 112024670,
      "step": 5271
    },
    {
      "epoch": 0.6339205194492875,
      "flos": 18867467796480.0,
      "grad_norm": 3.6292150261361775,
      "learning_rate": 1.2485082798102377e-06,
      "loss": 0.8248,
      "num_input_tokens_seen": 112044805,
      "step": 5272
    },
    {
      "epoch": 0.6340407623399267,
      "flos": 13144965795840.0,
      "grad_norm": 4.9992335685776235,
      "learning_rate": 1.2477864460845084e-06,
      "loss": 0.874,
      "num_input_tokens_seen": 112060925,
      "step": 5273
    },
    {
      "epoch": 0.6341610052305657,
      "flos": 12548225003520.0,
      "grad_norm": 10.95218566262343,
      "learning_rate": 1.2470647264689776e-06,
      "loss": 0.9223,
      "num_input_tokens_seen": 112079125,
      "step": 5274
    },
    {
      "epoch": 0.6342812481212048,
      "flos": 16770110730240.0,
      "grad_norm": 4.967009233441957,
      "learning_rate": 1.2463431210731282e-06,
      "loss": 0.8975,
      "num_input_tokens_seen": 112098430,
      "step": 5275
    },
    {
      "epoch": 0.634401491011844,
      "flos": 12626796564480.0,
      "grad_norm": 4.14527580052309,
      "learning_rate": 1.2456216300064289e-06,
      "loss": 0.943,
      "num_input_tokens_seen": 112115700,
      "step": 5276
    },
    {
      "epoch": 0.634521733902483,
      "flos": 15167828889600.0,
      "grad_norm": 2.269781890003098,
      "learning_rate": 1.244900253378328e-06,
      "loss": 0.9603,
      "num_input_tokens_seen": 112135475,
      "step": 5277
    },
    {
      "epoch": 0.6346419767931221,
      "flos": 11709147279360.0,
      "grad_norm": 3.418648230451881,
      "learning_rate": 1.2441789912982583e-06,
      "loss": 0.8864,
      "num_input_tokens_seen": 112152280,
      "step": 5278
    },
    {
      "epoch": 0.6347622196837612,
      "flos": 17318854656000.0,
      "grad_norm": 3.131428324785909,
      "learning_rate": 1.2434578438756346e-06,
      "loss": 0.832,
      "num_input_tokens_seen": 112172430,
      "step": 5279
    },
    {
      "epoch": 0.6348824625744003,
      "flos": 46190028165120.0,
      "grad_norm": 3.854704961903408,
      "learning_rate": 1.242736811219855e-06,
      "loss": 0.972,
      "num_input_tokens_seen": 112198110,
      "step": 5280
    },
    {
      "epoch": 0.6350027054650393,
      "flos": 20388756049920.0,
      "grad_norm": 2.67305858787324,
      "learning_rate": 1.2420158934402988e-06,
      "loss": 1.0177,
      "num_input_tokens_seen": 112218445,
      "step": 5281
    },
    {
      "epoch": 0.6351229483556785,
      "flos": 16349958696960.0,
      "grad_norm": 3.913246054404452,
      "learning_rate": 1.2412950906463286e-06,
      "loss": 1.027,
      "num_input_tokens_seen": 112235470,
      "step": 5282
    },
    {
      "epoch": 0.6352431912463176,
      "flos": 15585129676800.0,
      "grad_norm": 3.2478914022941394,
      "learning_rate": 1.2405744029472902e-06,
      "loss": 1.0849,
      "num_input_tokens_seen": 112254675,
      "step": 5283
    },
    {
      "epoch": 0.6353634341369566,
      "flos": 9689748602880.0,
      "grad_norm": 21.623204151181476,
      "learning_rate": 1.2398538304525108e-06,
      "loss": 0.9361,
      "num_input_tokens_seen": 112273020,
      "step": 5284
    },
    {
      "epoch": 0.6354836770275958,
      "flos": 13700454604800.0,
      "grad_norm": 3.3660566727870256,
      "learning_rate": 1.2391333732713016e-06,
      "loss": 0.9437,
      "num_input_tokens_seen": 112290545,
      "step": 5285
    },
    {
      "epoch": 0.6356039199182348,
      "flos": 15351223848960.0,
      "grad_norm": 10.92074993925966,
      "learning_rate": 1.2384130315129543e-06,
      "loss": 0.9632,
      "num_input_tokens_seen": 112308590,
      "step": 5286
    },
    {
      "epoch": 0.6357241628088739,
      "flos": 12835002408960.0,
      "grad_norm": 4.325480510501523,
      "learning_rate": 1.2376928052867447e-06,
      "loss": 0.9176,
      "num_input_tokens_seen": 112327430,
      "step": 5287
    },
    {
      "epoch": 0.6358444056995131,
      "flos": 17738822737920.0,
      "grad_norm": 3.006033710780556,
      "learning_rate": 1.2369726947019299e-06,
      "loss": 0.983,
      "num_input_tokens_seen": 112347625,
      "step": 5288
    },
    {
      "epoch": 0.6359646485901521,
      "flos": 16558808371200.0,
      "grad_norm": 5.29217032290955,
      "learning_rate": 1.2362526998677511e-06,
      "loss": 0.8605,
      "num_input_tokens_seen": 112363710,
      "step": 5289
    },
    {
      "epoch": 0.6360848914807912,
      "flos": 14830019420160.0,
      "grad_norm": 3.1363266715282627,
      "learning_rate": 1.2355328208934301e-06,
      "loss": 1.0454,
      "num_input_tokens_seen": 112382305,
      "step": 5290
    },
    {
      "epoch": 0.6362051343714303,
      "flos": 13099728199680.0,
      "grad_norm": 2.9392978766622933,
      "learning_rate": 1.2348130578881728e-06,
      "loss": 0.9253,
      "num_input_tokens_seen": 112400245,
      "step": 5291
    },
    {
      "epoch": 0.6363253772620694,
      "flos": 17346884689920.0,
      "grad_norm": 3.891647265846747,
      "learning_rate": 1.2340934109611664e-06,
      "loss": 0.9384,
      "num_input_tokens_seen": 112420725,
      "step": 5292
    },
    {
      "epoch": 0.6364456201527084,
      "flos": 18474211430400.0,
      "grad_norm": 3.8885716914439823,
      "learning_rate": 1.2333738802215798e-06,
      "loss": 0.8614,
      "num_input_tokens_seen": 112440665,
      "step": 5293
    },
    {
      "epoch": 0.6365658630433476,
      "flos": 14724460216320.0,
      "grad_norm": 4.359484380628219,
      "learning_rate": 1.2326544657785668e-06,
      "loss": 0.9903,
      "num_input_tokens_seen": 112460075,
      "step": 5294
    },
    {
      "epoch": 0.6366861059339867,
      "flos": 15222846566400.0,
      "grad_norm": 4.356101280092872,
      "learning_rate": 1.2319351677412608e-06,
      "loss": 0.9332,
      "num_input_tokens_seen": 112476840,
      "step": 5295
    },
    {
      "epoch": 0.6368063488246257,
      "flos": 15820629749760.0,
      "grad_norm": 3.0457166768567943,
      "learning_rate": 1.2312159862187796e-06,
      "loss": 0.9386,
      "num_input_tokens_seen": 112494970,
      "step": 5296
    },
    {
      "epoch": 0.6369265917152649,
      "flos": 15932075397120.0,
      "grad_norm": 5.662266013593092,
      "learning_rate": 1.2304969213202217e-06,
      "loss": 0.9508,
      "num_input_tokens_seen": 112515950,
      "step": 5297
    },
    {
      "epoch": 0.6370468346059039,
      "flos": 17583059251200.0,
      "grad_norm": 3.1028365970213354,
      "learning_rate": 1.2297779731546692e-06,
      "loss": 0.9755,
      "num_input_tokens_seen": 112534765,
      "step": 5298
    },
    {
      "epoch": 0.637167077496543,
      "flos": 17998029987840.0,
      "grad_norm": 3.281772826185632,
      "learning_rate": 1.2290591418311853e-06,
      "loss": 0.9702,
      "num_input_tokens_seen": 112551880,
      "step": 5299
    },
    {
      "epoch": 0.637287320387182,
      "flos": 19704951275520.0,
      "grad_norm": 2.9600103934853026,
      "learning_rate": 1.2283404274588172e-06,
      "loss": 0.9092,
      "num_input_tokens_seen": 112570545,
      "step": 5300
    },
    {
      "epoch": 0.6374075632778212,
      "flos": 45224925757440.0,
      "grad_norm": 0.7567841149092517,
      "learning_rate": 1.227621830146592e-06,
      "loss": 0.7406,
      "num_input_tokens_seen": 112625630,
      "step": 5301
    },
    {
      "epoch": 0.6375278061684603,
      "flos": 18186820853760.0,
      "grad_norm": 7.096977138414202,
      "learning_rate": 1.2269033500035217e-06,
      "loss": 0.9848,
      "num_input_tokens_seen": 112645485,
      "step": 5302
    },
    {
      "epoch": 0.6376480490590993,
      "flos": 18264441999360.0,
      "grad_norm": 3.683389344515925,
      "learning_rate": 1.2261849871385988e-06,
      "loss": 0.9343,
      "num_input_tokens_seen": 112666310,
      "step": 5303
    },
    {
      "epoch": 0.6377682919497385,
      "flos": 22483660431360.0,
      "grad_norm": 3.107173120886833,
      "learning_rate": 1.2254667416607972e-06,
      "loss": 0.8015,
      "num_input_tokens_seen": 112687630,
      "step": 5304
    },
    {
      "epoch": 0.6378885348403776,
      "flos": 16376333168640.0,
      "grad_norm": 3.539107097151057,
      "learning_rate": 1.2247486136790756e-06,
      "loss": 1.0295,
      "num_input_tokens_seen": 112706830,
      "step": 5305
    },
    {
      "epoch": 0.6380087777310166,
      "flos": 13255675637760.0,
      "grad_norm": 3.9014021431475316,
      "learning_rate": 1.2240306033023726e-06,
      "loss": 0.9955,
      "num_input_tokens_seen": 112724375,
      "step": 5306
    },
    {
      "epoch": 0.6381290206216558,
      "flos": 16586501160960.0,
      "grad_norm": 5.357306523320693,
      "learning_rate": 1.223312710639611e-06,
      "loss": 0.9219,
      "num_input_tokens_seen": 112742815,
      "step": 5307
    },
    {
      "epoch": 0.6382492635122948,
      "flos": 13387609313280.0,
      "grad_norm": 2.984938004199755,
      "learning_rate": 1.2225949357996928e-06,
      "loss": 1.051,
      "num_input_tokens_seen": 112760660,
      "step": 5308
    },
    {
      "epoch": 0.6383695064029339,
      "flos": 19811614187520.0,
      "grad_norm": 2.6978269486011124,
      "learning_rate": 1.221877278891505e-06,
      "loss": 1.0054,
      "num_input_tokens_seen": 112779635,
      "step": 5309
    },
    {
      "epoch": 0.638489749293573,
      "flos": 18788497674240.0,
      "grad_norm": 5.293160455476659,
      "learning_rate": 1.221159740023915e-06,
      "loss": 0.8902,
      "num_input_tokens_seen": 112799185,
      "step": 5310
    },
    {
      "epoch": 0.6386099921842121,
      "flos": 17058513039360.0,
      "grad_norm": 5.777305123159412,
      "learning_rate": 1.2204423193057735e-06,
      "loss": 0.9191,
      "num_input_tokens_seen": 112817735,
      "step": 5311
    },
    {
      "epoch": 0.6387302350748512,
      "flos": 51373014466560.0,
      "grad_norm": 0.9403264354460473,
      "learning_rate": 1.2197250168459122e-06,
      "loss": 0.8824,
      "num_input_tokens_seen": 112873855,
      "step": 5312
    },
    {
      "epoch": 0.6388504779654903,
      "flos": 10264775024640.0,
      "grad_norm": 3.2291369200946343,
      "learning_rate": 1.2190078327531454e-06,
      "loss": 0.9371,
      "num_input_tokens_seen": 112889820,
      "step": 5313
    },
    {
      "epoch": 0.6389707208561294,
      "flos": 15878253404160.0,
      "grad_norm": 4.739938620640715,
      "learning_rate": 1.2182907671362697e-06,
      "loss": 0.9239,
      "num_input_tokens_seen": 112910235,
      "step": 5314
    },
    {
      "epoch": 0.6390909637467684,
      "flos": 13780191191040.0,
      "grad_norm": 3.148654826120242,
      "learning_rate": 1.2175738201040626e-06,
      "loss": 0.9725,
      "num_input_tokens_seen": 112926995,
      "step": 5315
    },
    {
      "epoch": 0.6392112066374076,
      "flos": 17131259473920.0,
      "grad_norm": 3.3274948168226266,
      "learning_rate": 1.2168569917652855e-06,
      "loss": 0.9819,
      "num_input_tokens_seen": 112946570,
      "step": 5316
    },
    {
      "epoch": 0.6393314495280467,
      "flos": 19075367055360.0,
      "grad_norm": 3.8113841941081796,
      "learning_rate": 1.2161402822286797e-06,
      "loss": 0.8445,
      "num_input_tokens_seen": 112966975,
      "step": 5317
    },
    {
      "epoch": 0.6394516924186857,
      "flos": 14378710179840.0,
      "grad_norm": 4.308717699939674,
      "learning_rate": 1.2154236916029703e-06,
      "loss": 0.9726,
      "num_input_tokens_seen": 112984670,
      "step": 5318
    },
    {
      "epoch": 0.6395719353093249,
      "flos": 13019807662080.0,
      "grad_norm": 4.604981064151762,
      "learning_rate": 1.2147072199968627e-06,
      "loss": 0.9168,
      "num_input_tokens_seen": 113003025,
      "step": 5319
    },
    {
      "epoch": 0.6396921781999639,
      "flos": 12443064360960.0,
      "grad_norm": 2.6404476670531247,
      "learning_rate": 1.2139908675190454e-06,
      "loss": 0.9022,
      "num_input_tokens_seen": 113021955,
      "step": 5320
    },
    {
      "epoch": 0.639812421090603,
      "flos": 15091342110720.0,
      "grad_norm": 3.1250975021410428,
      "learning_rate": 1.2132746342781883e-06,
      "loss": 0.9515,
      "num_input_tokens_seen": 113042835,
      "step": 5321
    },
    {
      "epoch": 0.6399326639812422,
      "flos": 7853990154240.0,
      "grad_norm": 19.978725562399024,
      "learning_rate": 1.2125585203829442e-06,
      "loss": 0.982,
      "num_input_tokens_seen": 113058195,
      "step": 5322
    },
    {
      "epoch": 0.6400529068718812,
      "flos": 17002391654400.0,
      "grad_norm": 3.869773909201596,
      "learning_rate": 1.211842525941946e-06,
      "loss": 0.9272,
      "num_input_tokens_seen": 113077710,
      "step": 5323
    },
    {
      "epoch": 0.6401731497625203,
      "flos": 31453940244480.0,
      "grad_norm": 5.066525024810039,
      "learning_rate": 1.2111266510638105e-06,
      "loss": 0.9952,
      "num_input_tokens_seen": 113100355,
      "step": 5324
    },
    {
      "epoch": 0.6402933926531594,
      "flos": 14667296440320.0,
      "grad_norm": 4.869175292257755,
      "learning_rate": 1.2104108958571346e-06,
      "loss": 1.006,
      "num_input_tokens_seen": 113118345,
      "step": 5325
    },
    {
      "epoch": 0.6404136355437985,
      "flos": 17687269478400.0,
      "grad_norm": 5.831240069512746,
      "learning_rate": 1.2096952604304975e-06,
      "loss": 0.9462,
      "num_input_tokens_seen": 113138495,
      "step": 5326
    },
    {
      "epoch": 0.6405338784344375,
      "flos": 28910393917440.0,
      "grad_norm": 2.5224294627282835,
      "learning_rate": 1.2089797448924616e-06,
      "loss": 0.8886,
      "num_input_tokens_seen": 113162090,
      "step": 5327
    },
    {
      "epoch": 0.6406541213250767,
      "flos": 14828823736320.0,
      "grad_norm": 9.660605624151657,
      "learning_rate": 1.2082643493515692e-06,
      "loss": 0.8515,
      "num_input_tokens_seen": 113180130,
      "step": 5328
    },
    {
      "epoch": 0.6407743642157158,
      "flos": 16561046446080.0,
      "grad_norm": 3.5741541788907796,
      "learning_rate": 1.207549073916346e-06,
      "loss": 1.0094,
      "num_input_tokens_seen": 113200785,
      "step": 5329
    },
    {
      "epoch": 0.6408946071063548,
      "flos": 10607827107840.0,
      "grad_norm": 5.827042629994016,
      "learning_rate": 1.2068339186952976e-06,
      "loss": 0.9784,
      "num_input_tokens_seen": 113218045,
      "step": 5330
    },
    {
      "epoch": 0.6410148499969939,
      "flos": 16009389957120.0,
      "grad_norm": 3.702268090876677,
      "learning_rate": 1.2061188837969136e-06,
      "loss": 0.9191,
      "num_input_tokens_seen": 113237375,
      "step": 5331
    },
    {
      "epoch": 0.641135092887633,
      "flos": 8745970114560.0,
      "grad_norm": 4.272004551517191,
      "learning_rate": 1.2054039693296631e-06,
      "loss": 1.0167,
      "num_input_tokens_seen": 113255090,
      "step": 5332
    },
    {
      "epoch": 0.6412553357782721,
      "flos": 15721018306560.0,
      "grad_norm": 3.277406166764474,
      "learning_rate": 1.2046891754019992e-06,
      "loss": 1.0117,
      "num_input_tokens_seen": 113275420,
      "step": 5333
    },
    {
      "epoch": 0.6413755786689112,
      "flos": 11237411328000.0,
      "grad_norm": 4.724231491967478,
      "learning_rate": 1.2039745021223548e-06,
      "loss": 1.0194,
      "num_input_tokens_seen": 113292560,
      "step": 5334
    },
    {
      "epoch": 0.6414958215595503,
      "flos": 48723624898560.0,
      "grad_norm": 0.9099200355587834,
      "learning_rate": 1.2032599495991456e-06,
      "loss": 0.8122,
      "num_input_tokens_seen": 113357020,
      "step": 5335
    },
    {
      "epoch": 0.6416160644501894,
      "flos": 31506045358080.0,
      "grad_norm": 3.9418526463791874,
      "learning_rate": 1.2025455179407685e-06,
      "loss": 0.8859,
      "num_input_tokens_seen": 113377900,
      "step": 5336
    },
    {
      "epoch": 0.6417363073408284,
      "flos": 14881971240960.0,
      "grad_norm": 5.963189056055637,
      "learning_rate": 1.2018312072556022e-06,
      "loss": 0.9317,
      "num_input_tokens_seen": 113396120,
      "step": 5337
    },
    {
      "epoch": 0.6418565502314676,
      "flos": 15956518379520.0,
      "grad_norm": 3.954112898230232,
      "learning_rate": 1.2011170176520077e-06,
      "loss": 0.9402,
      "num_input_tokens_seen": 113416755,
      "step": 5338
    },
    {
      "epoch": 0.6419767931221066,
      "flos": 17818160762880.0,
      "grad_norm": 2.2613776020509304,
      "learning_rate": 1.2004029492383256e-06,
      "loss": 0.9991,
      "num_input_tokens_seen": 113437815,
      "step": 5339
    },
    {
      "epoch": 0.6420970360127457,
      "flos": 13806688296960.0,
      "grad_norm": 3.541365995920985,
      "learning_rate": 1.1996890021228814e-06,
      "loss": 0.9317,
      "num_input_tokens_seen": 113454310,
      "step": 5340
    },
    {
      "epoch": 0.6422172789033849,
      "flos": 28857583656960.0,
      "grad_norm": 3.0045043191204237,
      "learning_rate": 1.1989751764139785e-06,
      "loss": 0.8942,
      "num_input_tokens_seen": 113477680,
      "step": 5341
    },
    {
      "epoch": 0.6423375217940239,
      "flos": 19706453544960.0,
      "grad_norm": 2.5335201867946013,
      "learning_rate": 1.1982614722199044e-06,
      "loss": 1.0077,
      "num_input_tokens_seen": 113498575,
      "step": 5342
    },
    {
      "epoch": 0.642457764684663,
      "flos": 13020144906240.0,
      "grad_norm": 3.509866431756971,
      "learning_rate": 1.1975478896489276e-06,
      "loss": 0.9731,
      "num_input_tokens_seen": 113516130,
      "step": 5343
    },
    {
      "epoch": 0.6425780075753021,
      "flos": 14012134871040.0,
      "grad_norm": 3.1128107000107734,
      "learning_rate": 1.1968344288092981e-06,
      "loss": 0.9588,
      "num_input_tokens_seen": 113532430,
      "step": 5344
    },
    {
      "epoch": 0.6426982504659412,
      "flos": 14593568931840.0,
      "grad_norm": 3.8312482276899957,
      "learning_rate": 1.1961210898092468e-06,
      "loss": 0.8391,
      "num_input_tokens_seen": 113551100,
      "step": 5345
    },
    {
      "epoch": 0.6428184933565803,
      "flos": 12647529861120.0,
      "grad_norm": 5.994592455070868,
      "learning_rate": 1.1954078727569874e-06,
      "loss": 0.9834,
      "num_input_tokens_seen": 113568120,
      "step": 5346
    },
    {
      "epoch": 0.6429387362472194,
      "flos": 15958081966080.0,
      "grad_norm": 6.47104416123249,
      "learning_rate": 1.1946947777607141e-06,
      "loss": 0.9752,
      "num_input_tokens_seen": 113588975,
      "step": 5347
    },
    {
      "epoch": 0.6430589791378585,
      "flos": 17607502233600.0,
      "grad_norm": 5.4560068816983405,
      "learning_rate": 1.1939818049286024e-06,
      "loss": 0.9841,
      "num_input_tokens_seen": 113606855,
      "step": 5348
    },
    {
      "epoch": 0.6431792220284975,
      "flos": 17714716999680.0,
      "grad_norm": 3.4179994418264865,
      "learning_rate": 1.1932689543688101e-06,
      "loss": 0.9372,
      "num_input_tokens_seen": 113627680,
      "step": 5349
    },
    {
      "epoch": 0.6432994649191367,
      "flos": 14933984378880.0,
      "grad_norm": 3.487411886216415,
      "learning_rate": 1.1925562261894756e-06,
      "loss": 0.9204,
      "num_input_tokens_seen": 113646480,
      "step": 5350
    },
    {
      "epoch": 0.6434197078097758,
      "flos": 22014653091840.0,
      "grad_norm": 3.625700389681263,
      "learning_rate": 1.1918436204987207e-06,
      "loss": 0.9749,
      "num_input_tokens_seen": 113668060,
      "step": 5351
    },
    {
      "epoch": 0.6435399507004148,
      "flos": 10607704473600.0,
      "grad_norm": 7.629545514131053,
      "learning_rate": 1.191131137404645e-06,
      "loss": 1.0012,
      "num_input_tokens_seen": 113684520,
      "step": 5352
    },
    {
      "epoch": 0.643660193591054,
      "flos": 14121741004800.0,
      "grad_norm": 9.235627839743126,
      "learning_rate": 1.190418777015333e-06,
      "loss": 0.9661,
      "num_input_tokens_seen": 113703150,
      "step": 5353
    },
    {
      "epoch": 0.643780436481693,
      "flos": 17162845900800.0,
      "grad_norm": 6.971332157191941,
      "learning_rate": 1.1897065394388487e-06,
      "loss": 0.9308,
      "num_input_tokens_seen": 113723310,
      "step": 5354
    },
    {
      "epoch": 0.6439006793723321,
      "flos": 16455119339520.0,
      "grad_norm": 2.6192592335128895,
      "learning_rate": 1.1889944247832385e-06,
      "loss": 0.9572,
      "num_input_tokens_seen": 113743270,
      "step": 5355
    },
    {
      "epoch": 0.6440209222629713,
      "flos": 16791549173760.0,
      "grad_norm": 5.239505633264064,
      "learning_rate": 1.1882824331565283e-06,
      "loss": 0.896,
      "num_input_tokens_seen": 113762450,
      "step": 5356
    },
    {
      "epoch": 0.6441411651536103,
      "flos": 11708993986560.0,
      "grad_norm": 9.510572874381742,
      "learning_rate": 1.1875705646667287e-06,
      "loss": 1.0777,
      "num_input_tokens_seen": 113780060,
      "step": 5357
    },
    {
      "epoch": 0.6442614080442494,
      "flos": 18080893747200.0,
      "grad_norm": 4.106411289432479,
      "learning_rate": 1.1868588194218282e-06,
      "loss": 0.9411,
      "num_input_tokens_seen": 113800160,
      "step": 5358
    },
    {
      "epoch": 0.6443816509348885,
      "flos": 20152980049920.0,
      "grad_norm": 3.7944340650032684,
      "learning_rate": 1.1861471975297979e-06,
      "loss": 0.9321,
      "num_input_tokens_seen": 113821575,
      "step": 5359
    },
    {
      "epoch": 0.6445018938255276,
      "flos": 26187652853760.0,
      "grad_norm": 4.400272074870569,
      "learning_rate": 1.185435699098591e-06,
      "loss": 0.9077,
      "num_input_tokens_seen": 113847490,
      "step": 5360
    },
    {
      "epoch": 0.6446221367161666,
      "flos": 10240362700800.0,
      "grad_norm": 5.840829700946517,
      "learning_rate": 1.1847243242361403e-06,
      "loss": 0.9611,
      "num_input_tokens_seen": 113865800,
      "step": 5361
    },
    {
      "epoch": 0.6447423796068057,
      "flos": 17504947568640.0,
      "grad_norm": 3.5036302540758193,
      "learning_rate": 1.1840130730503624e-06,
      "loss": 0.9736,
      "num_input_tokens_seen": 113886800,
      "step": 5362
    },
    {
      "epoch": 0.6448626224974449,
      "flos": 17819693690880.0,
      "grad_norm": 3.2912678698305617,
      "learning_rate": 1.1833019456491518e-06,
      "loss": 0.9345,
      "num_input_tokens_seen": 113908050,
      "step": 5363
    },
    {
      "epoch": 0.6449828653880839,
      "flos": 16010892226560.0,
      "grad_norm": 5.280248061624109,
      "learning_rate": 1.1825909421403871e-06,
      "loss": 0.9601,
      "num_input_tokens_seen": 113926865,
      "step": 5364
    },
    {
      "epoch": 0.645103108278723,
      "flos": 18284592783360.0,
      "grad_norm": 3.2288641675139975,
      "learning_rate": 1.181880062631926e-06,
      "loss": 0.9478,
      "num_input_tokens_seen": 113945920,
      "step": 5365
    },
    {
      "epoch": 0.6452233511693621,
      "flos": 19546029957120.0,
      "grad_norm": 3.311868859389948,
      "learning_rate": 1.1811693072316093e-06,
      "loss": 1.0448,
      "num_input_tokens_seen": 113963320,
      "step": 5366
    },
    {
      "epoch": 0.6453435940600012,
      "flos": 13623415971840.0,
      "grad_norm": 4.979391542514707,
      "learning_rate": 1.1804586760472574e-06,
      "loss": 1.0171,
      "num_input_tokens_seen": 113979505,
      "step": 5367
    },
    {
      "epoch": 0.6454638369506402,
      "flos": 18315474063360.0,
      "grad_norm": 4.993870411210123,
      "learning_rate": 1.1797481691866736e-06,
      "loss": 0.9787,
      "num_input_tokens_seen": 113996450,
      "step": 5368
    },
    {
      "epoch": 0.6455840798412794,
      "flos": 14902091366400.0,
      "grad_norm": 11.357924839733668,
      "learning_rate": 1.1790377867576393e-06,
      "loss": 1.0256,
      "num_input_tokens_seen": 114013920,
      "step": 5369
    },
    {
      "epoch": 0.6457043227319185,
      "flos": 18552752332800.0,
      "grad_norm": 2.517934177692769,
      "learning_rate": 1.1783275288679203e-06,
      "loss": 0.9505,
      "num_input_tokens_seen": 114030805,
      "step": 5370
    },
    {
      "epoch": 0.6458245656225575,
      "flos": 43211689451520.0,
      "grad_norm": 0.9259668156195948,
      "learning_rate": 1.177617395625262e-06,
      "loss": 0.8459,
      "num_input_tokens_seen": 114088500,
      "step": 5371
    },
    {
      "epoch": 0.6459448085131967,
      "flos": 16403474104320.0,
      "grad_norm": 2.752300860471161,
      "learning_rate": 1.1769073871373908e-06,
      "loss": 0.9439,
      "num_input_tokens_seen": 114108425,
      "step": 5372
    },
    {
      "epoch": 0.6460650514038357,
      "flos": 16058950410240.0,
      "grad_norm": 3.223670827507979,
      "learning_rate": 1.176197503512015e-06,
      "loss": 1.0313,
      "num_input_tokens_seen": 114127860,
      "step": 5373
    },
    {
      "epoch": 0.6461852942944748,
      "flos": 14383860817920.0,
      "grad_norm": 5.485952992581727,
      "learning_rate": 1.1754877448568223e-06,
      "loss": 1.0009,
      "num_input_tokens_seen": 114147035,
      "step": 5374
    },
    {
      "epoch": 0.646305537185114,
      "flos": 16611925217280.0,
      "grad_norm": 2.9982938683375657,
      "learning_rate": 1.1747781112794837e-06,
      "loss": 1.0956,
      "num_input_tokens_seen": 114163250,
      "step": 5375
    },
    {
      "epoch": 0.646425780075753,
      "flos": 17265952419840.0,
      "grad_norm": 2.6455990144393833,
      "learning_rate": 1.1740686028876487e-06,
      "loss": 1.02,
      "num_input_tokens_seen": 114181835,
      "step": 5376
    },
    {
      "epoch": 0.6465460229663921,
      "flos": 14777147842560.0,
      "grad_norm": 5.492569325443221,
      "learning_rate": 1.1733592197889507e-06,
      "loss": 0.9356,
      "num_input_tokens_seen": 114198465,
      "step": 5377
    },
    {
      "epoch": 0.6466662658570312,
      "flos": 16242253393920.0,
      "grad_norm": 2.864752377232861,
      "learning_rate": 1.1726499620910014e-06,
      "loss": 0.9186,
      "num_input_tokens_seen": 114218465,
      "step": 5378
    },
    {
      "epoch": 0.6467865087476703,
      "flos": 10817627197440.0,
      "grad_norm": 7.953153299157658,
      "learning_rate": 1.1719408299013955e-06,
      "loss": 0.9487,
      "num_input_tokens_seen": 114236910,
      "step": 5379
    },
    {
      "epoch": 0.6469067516383094,
      "flos": 13885781053440.0,
      "grad_norm": 9.35832501414182,
      "learning_rate": 1.1712318233277067e-06,
      "loss": 0.9362,
      "num_input_tokens_seen": 114255650,
      "step": 5380
    },
    {
      "epoch": 0.6470269945289485,
      "flos": 46606860963840.0,
      "grad_norm": 0.7898458992073257,
      "learning_rate": 1.1705229424774916e-06,
      "loss": 0.7829,
      "num_input_tokens_seen": 114309640,
      "step": 5381
    },
    {
      "epoch": 0.6471472374195876,
      "flos": 21879224340480.0,
      "grad_norm": 2.9524666188522795,
      "learning_rate": 1.1698141874582867e-06,
      "loss": 0.832,
      "num_input_tokens_seen": 114330405,
      "step": 5382
    },
    {
      "epoch": 0.6472674803102266,
      "flos": 14567163801600.0,
      "grad_norm": 3.6129782389838847,
      "learning_rate": 1.169105558377609e-06,
      "loss": 0.9119,
      "num_input_tokens_seen": 114350215,
      "step": 5383
    },
    {
      "epoch": 0.6473877232008658,
      "flos": 17580729200640.0,
      "grad_norm": 3.307188129556555,
      "learning_rate": 1.1683970553429587e-06,
      "loss": 0.9717,
      "num_input_tokens_seen": 114371390,
      "step": 5384
    },
    {
      "epoch": 0.6475079660915048,
      "flos": 11234805350400.0,
      "grad_norm": 4.103980174725035,
      "learning_rate": 1.1676886784618128e-06,
      "loss": 1.0126,
      "num_input_tokens_seen": 114387775,
      "step": 5385
    },
    {
      "epoch": 0.6476282089821439,
      "flos": 12309996318720.0,
      "grad_norm": 3.998932338324391,
      "learning_rate": 1.1669804278416332e-06,
      "loss": 1.0122,
      "num_input_tokens_seen": 114402220,
      "step": 5386
    },
    {
      "epoch": 0.6477484518727831,
      "flos": 14331694387200.0,
      "grad_norm": 2.876330026560216,
      "learning_rate": 1.1662723035898602e-06,
      "loss": 0.9048,
      "num_input_tokens_seen": 114421700,
      "step": 5387
    },
    {
      "epoch": 0.6478686947634221,
      "flos": 18080433868800.0,
      "grad_norm": 4.313727604938608,
      "learning_rate": 1.165564305813915e-06,
      "loss": 1.0125,
      "num_input_tokens_seen": 114440420,
      "step": 5388
    },
    {
      "epoch": 0.6479889376540612,
      "flos": 14252141752320.0,
      "grad_norm": 2.3544137088033037,
      "learning_rate": 1.1648564346212019e-06,
      "loss": 1.0039,
      "num_input_tokens_seen": 114459260,
      "step": 5389
    },
    {
      "epoch": 0.6481091805447003,
      "flos": 19048655339520.0,
      "grad_norm": 3.4005811923822504,
      "learning_rate": 1.164148690119104e-06,
      "loss": 0.959,
      "num_input_tokens_seen": 114480260,
      "step": 5390
    },
    {
      "epoch": 0.6482294234353394,
      "flos": 17032506470400.0,
      "grad_norm": 2.53910944968612,
      "learning_rate": 1.163441072414985e-06,
      "loss": 0.9322,
      "num_input_tokens_seen": 114500185,
      "step": 5391
    },
    {
      "epoch": 0.6483496663259785,
      "flos": 18654509875200.0,
      "grad_norm": 4.9035841387086965,
      "learning_rate": 1.16273358161619e-06,
      "loss": 0.8931,
      "num_input_tokens_seen": 114520240,
      "step": 5392
    },
    {
      "epoch": 0.6484699092166175,
      "flos": 14854799646720.0,
      "grad_norm": 3.4936332014347173,
      "learning_rate": 1.1620262178300446e-06,
      "loss": 1.0377,
      "num_input_tokens_seen": 114538575,
      "step": 5393
    },
    {
      "epoch": 0.6485901521072567,
      "flos": 23587341312000.0,
      "grad_norm": 6.754898554381144,
      "learning_rate": 1.1613189811638563e-06,
      "loss": 0.9419,
      "num_input_tokens_seen": 114560020,
      "step": 5394
    },
    {
      "epoch": 0.6487103949978957,
      "flos": 15826577510400.0,
      "grad_norm": 4.726666813479602,
      "learning_rate": 1.1606118717249117e-06,
      "loss": 0.9705,
      "num_input_tokens_seen": 114579840,
      "step": 5395
    },
    {
      "epoch": 0.6488306378885348,
      "flos": 16298221486080.0,
      "grad_norm": 2.7119254035424096,
      "learning_rate": 1.1599048896204787e-06,
      "loss": 0.8565,
      "num_input_tokens_seen": 114599440,
      "step": 5396
    },
    {
      "epoch": 0.648950880779174,
      "flos": 14462463037440.0,
      "grad_norm": 2.833691929582718,
      "learning_rate": 1.1591980349578061e-06,
      "loss": 0.9939,
      "num_input_tokens_seen": 114617830,
      "step": 5397
    },
    {
      "epoch": 0.649071123669813,
      "flos": 42895226449920.0,
      "grad_norm": 0.7808803065110371,
      "learning_rate": 1.158491307844123e-06,
      "loss": 0.7663,
      "num_input_tokens_seen": 114677470,
      "step": 5398
    },
    {
      "epoch": 0.6491913665604521,
      "flos": 14512759296000.0,
      "grad_norm": 3.081792979343242,
      "learning_rate": 1.1577847083866387e-06,
      "loss": 1.0405,
      "num_input_tokens_seen": 114696225,
      "step": 5399
    },
    {
      "epoch": 0.6493116094510912,
      "flos": 11997948149760.0,
      "grad_norm": 5.249306587566452,
      "learning_rate": 1.1570782366925453e-06,
      "loss": 0.9121,
      "num_input_tokens_seen": 114714460,
      "step": 5400
    },
    {
      "epoch": 0.6494318523417303,
      "flos": 13331273318400.0,
      "grad_norm": 3.6418308340353303,
      "learning_rate": 1.1563718928690132e-06,
      "loss": 0.9549,
      "num_input_tokens_seen": 114731615,
      "step": 5401
    },
    {
      "epoch": 0.6495520952323693,
      "flos": 13460233113600.0,
      "grad_norm": 3.7218573914951327,
      "learning_rate": 1.1556656770231942e-06,
      "loss": 0.9158,
      "num_input_tokens_seen": 114747530,
      "step": 5402
    },
    {
      "epoch": 0.6496723381230085,
      "flos": 16164908175360.0,
      "grad_norm": 3.798970087924209,
      "learning_rate": 1.1549595892622207e-06,
      "loss": 0.9517,
      "num_input_tokens_seen": 114766020,
      "step": 5403
    },
    {
      "epoch": 0.6497925810136476,
      "flos": 44485152890880.0,
      "grad_norm": 0.8340686692959793,
      "learning_rate": 1.1542536296932047e-06,
      "loss": 0.8243,
      "num_input_tokens_seen": 114829275,
      "step": 5404
    },
    {
      "epoch": 0.6499128239042866,
      "flos": 14304737402880.0,
      "grad_norm": 3.6236098682353832,
      "learning_rate": 1.1535477984232414e-06,
      "loss": 0.8865,
      "num_input_tokens_seen": 114848870,
      "step": 5405
    },
    {
      "epoch": 0.6500330667949258,
      "flos": 17399204413440.0,
      "grad_norm": 2.977646037251026,
      "learning_rate": 1.152842095559404e-06,
      "loss": 0.9582,
      "num_input_tokens_seen": 114869250,
      "step": 5406
    },
    {
      "epoch": 0.6501533096855648,
      "flos": 18127756247040.0,
      "grad_norm": 2.8600779852205025,
      "learning_rate": 1.1521365212087474e-06,
      "loss": 0.9667,
      "num_input_tokens_seen": 114888955,
      "step": 5407
    },
    {
      "epoch": 0.6502735525762039,
      "flos": 32028966666240.0,
      "grad_norm": 2.8728528505079036,
      "learning_rate": 1.1514310754783062e-06,
      "loss": 0.8873,
      "num_input_tokens_seen": 114911625,
      "step": 5408
    },
    {
      "epoch": 0.6503937954668431,
      "flos": 20414915911680.0,
      "grad_norm": 4.69731211807812,
      "learning_rate": 1.1507257584750964e-06,
      "loss": 0.9128,
      "num_input_tokens_seen": 114931525,
      "step": 5409
    },
    {
      "epoch": 0.6505140383574821,
      "flos": 14855075573760.0,
      "grad_norm": 3.1921659240142874,
      "learning_rate": 1.150020570306113e-06,
      "loss": 0.9492,
      "num_input_tokens_seen": 114950385,
      "step": 5410
    },
    {
      "epoch": 0.6506342812481212,
      "flos": 14619974062080.0,
      "grad_norm": 4.677288274657448,
      "learning_rate": 1.1493155110783338e-06,
      "loss": 0.9369,
      "num_input_tokens_seen": 114968630,
      "step": 5411
    },
    {
      "epoch": 0.6507545241387603,
      "flos": 21541629480960.0,
      "grad_norm": 3.774581072469373,
      "learning_rate": 1.1486105808987155e-06,
      "loss": 0.8956,
      "num_input_tokens_seen": 114989840,
      "step": 5412
    },
    {
      "epoch": 0.6508747670293994,
      "flos": 12128195604480.0,
      "grad_norm": 3.166102375816422,
      "learning_rate": 1.1479057798741947e-06,
      "loss": 1.0028,
      "num_input_tokens_seen": 115007615,
      "step": 5413
    },
    {
      "epoch": 0.6509950099200384,
      "flos": 49094224588800.0,
      "grad_norm": 0.8118476082495645,
      "learning_rate": 1.14720110811169e-06,
      "loss": 0.772,
      "num_input_tokens_seen": 115064565,
      "step": 5414
    },
    {
      "epoch": 0.6511152528106776,
      "flos": 15879387770880.0,
      "grad_norm": 3.0603684773209876,
      "learning_rate": 1.146496565718098e-06,
      "loss": 0.9595,
      "num_input_tokens_seen": 115084855,
      "step": 5415
    },
    {
      "epoch": 0.6512354957013167,
      "flos": 14567439728640.0,
      "grad_norm": 4.406819408230037,
      "learning_rate": 1.1457921528002996e-06,
      "loss": 0.9466,
      "num_input_tokens_seen": 115103010,
      "step": 5416
    },
    {
      "epoch": 0.6513557385919557,
      "flos": 23057951047680.0,
      "grad_norm": 3.4057789048065485,
      "learning_rate": 1.1450878694651522e-06,
      "loss": 0.9178,
      "num_input_tokens_seen": 115123295,
      "step": 5417
    },
    {
      "epoch": 0.6514759814825949,
      "flos": 8509611601920.0,
      "grad_norm": 3.995180087082766,
      "learning_rate": 1.1443837158194954e-06,
      "loss": 0.8197,
      "num_input_tokens_seen": 115138160,
      "step": 5418
    },
    {
      "epoch": 0.651596224373234,
      "flos": 16007949004800.0,
      "grad_norm": 6.58102244448345,
      "learning_rate": 1.1436796919701484e-06,
      "loss": 0.9397,
      "num_input_tokens_seen": 115156595,
      "step": 5419
    },
    {
      "epoch": 0.651716467263873,
      "flos": 19811460894720.0,
      "grad_norm": 3.4811546600271597,
      "learning_rate": 1.1429757980239115e-06,
      "loss": 0.8111,
      "num_input_tokens_seen": 115176740,
      "step": 5420
    },
    {
      "epoch": 0.6518367101545122,
      "flos": 17294166405120.0,
      "grad_norm": 5.107498617671531,
      "learning_rate": 1.1422720340875636e-06,
      "loss": 0.998,
      "num_input_tokens_seen": 115195210,
      "step": 5421
    },
    {
      "epoch": 0.6519569530451512,
      "flos": 14200833761280.0,
      "grad_norm": 3.6801083722950247,
      "learning_rate": 1.1415684002678671e-06,
      "loss": 0.9848,
      "num_input_tokens_seen": 115213690,
      "step": 5422
    },
    {
      "epoch": 0.6520771959357903,
      "flos": 15324297523200.0,
      "grad_norm": 11.703162158861234,
      "learning_rate": 1.1408648966715617e-06,
      "loss": 0.9778,
      "num_input_tokens_seen": 115230930,
      "step": 5423
    },
    {
      "epoch": 0.6521974388264293,
      "flos": 16140495851520.0,
      "grad_norm": 3.72976578884796,
      "learning_rate": 1.1401615234053683e-06,
      "loss": 0.9186,
      "num_input_tokens_seen": 115249470,
      "step": 5424
    },
    {
      "epoch": 0.6523176817170685,
      "flos": 16350295941120.0,
      "grad_norm": 4.145580162612171,
      "learning_rate": 1.1394582805759885e-06,
      "loss": 0.9459,
      "num_input_tokens_seen": 115268470,
      "step": 5425
    },
    {
      "epoch": 0.6524379246077076,
      "flos": 15405291110400.0,
      "grad_norm": 4.717271788009277,
      "learning_rate": 1.1387551682901022e-06,
      "loss": 0.9557,
      "num_input_tokens_seen": 115288795,
      "step": 5426
    },
    {
      "epoch": 0.6525581674983466,
      "flos": 13754429890560.0,
      "grad_norm": 4.323562144918433,
      "learning_rate": 1.138052186654373e-06,
      "loss": 0.9087,
      "num_input_tokens_seen": 115305985,
      "step": 5427
    },
    {
      "epoch": 0.6526784103889858,
      "flos": 12155029954560.0,
      "grad_norm": 3.274492990399773,
      "learning_rate": 1.1373493357754417e-06,
      "loss": 1.069,
      "num_input_tokens_seen": 115324610,
      "step": 5428
    },
    {
      "epoch": 0.6527986532796248,
      "flos": 13465629020160.0,
      "grad_norm": 3.6321930387931105,
      "learning_rate": 1.1366466157599303e-06,
      "loss": 0.9666,
      "num_input_tokens_seen": 115343605,
      "step": 5429
    },
    {
      "epoch": 0.6529188961702639,
      "flos": 10050468126720.0,
      "grad_norm": 4.7105048694251375,
      "learning_rate": 1.1359440267144412e-06,
      "loss": 0.9488,
      "num_input_tokens_seen": 115360780,
      "step": 5430
    },
    {
      "epoch": 0.653039139060903,
      "flos": 26181551800320.0,
      "grad_norm": 4.640415790607216,
      "learning_rate": 1.1352415687455556e-06,
      "loss": 0.9392,
      "num_input_tokens_seen": 115381760,
      "step": 5431
    },
    {
      "epoch": 0.6531593819515421,
      "flos": 18056205496320.0,
      "grad_norm": 4.461233366392534,
      "learning_rate": 1.1345392419598362e-06,
      "loss": 0.8262,
      "num_input_tokens_seen": 115400360,
      "step": 5432
    },
    {
      "epoch": 0.6532796248421812,
      "flos": 15039114362880.0,
      "grad_norm": 2.9728431191777807,
      "learning_rate": 1.1338370464638263e-06,
      "loss": 0.9066,
      "num_input_tokens_seen": 115419480,
      "step": 5433
    },
    {
      "epoch": 0.6533998677328203,
      "flos": 12521881190400.0,
      "grad_norm": 3.894107347871032,
      "learning_rate": 1.1331349823640474e-06,
      "loss": 0.8267,
      "num_input_tokens_seen": 115436630,
      "step": 5434
    },
    {
      "epoch": 0.6535201106234594,
      "flos": 20229006950400.0,
      "grad_norm": 6.413766907584358,
      "learning_rate": 1.132433049767003e-06,
      "loss": 0.9735,
      "num_input_tokens_seen": 115454265,
      "step": 5435
    },
    {
      "epoch": 0.6536403535140984,
      "flos": 16769804144640.0,
      "grad_norm": 2.4888731343853068,
      "learning_rate": 1.1317312487791748e-06,
      "loss": 0.9983,
      "num_input_tokens_seen": 115475635,
      "step": 5436
    },
    {
      "epoch": 0.6537605964047376,
      "flos": 15327210086400.0,
      "grad_norm": 8.13842087558419,
      "learning_rate": 1.1310295795070253e-06,
      "loss": 0.9095,
      "num_input_tokens_seen": 115495295,
      "step": 5437
    },
    {
      "epoch": 0.6538808392953767,
      "flos": 19103611699200.0,
      "grad_norm": 3.7933320520686,
      "learning_rate": 1.1303280420569982e-06,
      "loss": 0.9898,
      "num_input_tokens_seen": 115516900,
      "step": 5438
    },
    {
      "epoch": 0.6540010821860157,
      "flos": 21909553766400.0,
      "grad_norm": 7.715085538141482,
      "learning_rate": 1.1296266365355158e-06,
      "loss": 0.9621,
      "num_input_tokens_seen": 115540005,
      "step": 5439
    },
    {
      "epoch": 0.6541213250766549,
      "flos": 19156207349760.0,
      "grad_norm": 3.4639679534081678,
      "learning_rate": 1.1289253630489806e-06,
      "loss": 0.9047,
      "num_input_tokens_seen": 115560775,
      "step": 5440
    },
    {
      "epoch": 0.6542415679672939,
      "flos": 13597409402880.0,
      "grad_norm": 5.5464087061929375,
      "learning_rate": 1.1282242217037753e-06,
      "loss": 0.9428,
      "num_input_tokens_seen": 115577995,
      "step": 5441
    },
    {
      "epoch": 0.654361810857933,
      "flos": 34439291658240.0,
      "grad_norm": 3.982211494220842,
      "learning_rate": 1.127523212606262e-06,
      "loss": 0.82,
      "num_input_tokens_seen": 115600540,
      "step": 5442
    },
    {
      "epoch": 0.6544820537485722,
      "flos": 19182029967360.0,
      "grad_norm": 2.958080191370596,
      "learning_rate": 1.1268223358627835e-06,
      "loss": 0.9173,
      "num_input_tokens_seen": 115622750,
      "step": 5443
    },
    {
      "epoch": 0.6546022966392112,
      "flos": 14829069004800.0,
      "grad_norm": 4.07754430619334,
      "learning_rate": 1.126121591579663e-06,
      "loss": 0.8996,
      "num_input_tokens_seen": 115641675,
      "step": 5444
    },
    {
      "epoch": 0.6547225395298503,
      "flos": 17740049080320.0,
      "grad_norm": 2.584494565621875,
      "learning_rate": 1.1254209798632018e-06,
      "loss": 0.8815,
      "num_input_tokens_seen": 115662415,
      "step": 5445
    },
    {
      "epoch": 0.6548427824204894,
      "flos": 16036346941440.0,
      "grad_norm": 3.546843003452182,
      "learning_rate": 1.124720500819683e-06,
      "loss": 1.0381,
      "num_input_tokens_seen": 115680290,
      "step": 5446
    },
    {
      "epoch": 0.6549630253111285,
      "flos": 13072832532480.0,
      "grad_norm": 3.333792146259705,
      "learning_rate": 1.1240201545553682e-06,
      "loss": 1.0133,
      "num_input_tokens_seen": 115697810,
      "step": 5447
    },
    {
      "epoch": 0.6550832682017675,
      "flos": 17919519744000.0,
      "grad_norm": 3.6765056873216846,
      "learning_rate": 1.1233199411764987e-06,
      "loss": 0.9324,
      "num_input_tokens_seen": 115716965,
      "step": 5448
    },
    {
      "epoch": 0.6552035110924067,
      "flos": 16167268884480.0,
      "grad_norm": 3.696745580502246,
      "learning_rate": 1.1226198607892978e-06,
      "loss": 0.8837,
      "num_input_tokens_seen": 115737245,
      "step": 5449
    },
    {
      "epoch": 0.6553237539830458,
      "flos": 15485211648000.0,
      "grad_norm": 3.235496437152144,
      "learning_rate": 1.1219199134999664e-06,
      "loss": 0.9935,
      "num_input_tokens_seen": 115755465,
      "step": 5450
    },
    {
      "epoch": 0.6554439968736848,
      "flos": 14829835468800.0,
      "grad_norm": 3.877977642299817,
      "learning_rate": 1.1212200994146863e-06,
      "loss": 0.9647,
      "num_input_tokens_seen": 115772940,
      "step": 5451
    },
    {
      "epoch": 0.655564239764324,
      "flos": 11416728698880.0,
      "grad_norm": 3.5331816664719433,
      "learning_rate": 1.120520418639618e-06,
      "loss": 0.9513,
      "num_input_tokens_seen": 115791195,
      "step": 5452
    },
    {
      "epoch": 0.655684482654963,
      "flos": 21070169456640.0,
      "grad_norm": 4.420693391292191,
      "learning_rate": 1.119820871280903e-06,
      "loss": 1.0285,
      "num_input_tokens_seen": 115811990,
      "step": 5453
    },
    {
      "epoch": 0.6558047255456021,
      "flos": 21227619164160.0,
      "grad_norm": 4.099057549807133,
      "learning_rate": 1.1191214574446614e-06,
      "loss": 0.9071,
      "num_input_tokens_seen": 115831955,
      "step": 5454
    },
    {
      "epoch": 0.6559249684362413,
      "flos": 20702735708160.0,
      "grad_norm": 2.6887939354478294,
      "learning_rate": 1.118422177236995e-06,
      "loss": 1.0012,
      "num_input_tokens_seen": 115853500,
      "step": 5455
    },
    {
      "epoch": 0.6560452113268803,
      "flos": 14353562050560.0,
      "grad_norm": 3.1543792619368407,
      "learning_rate": 1.1177230307639835e-06,
      "loss": 1.0423,
      "num_input_tokens_seen": 115870760,
      "step": 5456
    },
    {
      "epoch": 0.6561654542175194,
      "flos": 17818191421440.0,
      "grad_norm": 3.126435369817049,
      "learning_rate": 1.1170240181316865e-06,
      "loss": 0.9795,
      "num_input_tokens_seen": 115891925,
      "step": 5457
    },
    {
      "epoch": 0.6562856971081584,
      "flos": 16245165957120.0,
      "grad_norm": 4.300529119184761,
      "learning_rate": 1.1163251394461442e-06,
      "loss": 0.9884,
      "num_input_tokens_seen": 115910125,
      "step": 5458
    },
    {
      "epoch": 0.6564059399987976,
      "flos": 13380864430080.0,
      "grad_norm": 4.321975029417365,
      "learning_rate": 1.1156263948133746e-06,
      "loss": 1.0163,
      "num_input_tokens_seen": 115926500,
      "step": 5459
    },
    {
      "epoch": 0.6565261828894366,
      "flos": 18133550714880.0,
      "grad_norm": 4.458965553782004,
      "learning_rate": 1.1149277843393787e-06,
      "loss": 0.971,
      "num_input_tokens_seen": 115947380,
      "step": 5460
    },
    {
      "epoch": 0.6566464257800757,
      "flos": 13964781834240.0,
      "grad_norm": 4.133351974035089,
      "learning_rate": 1.1142293081301342e-06,
      "loss": 0.8255,
      "num_input_tokens_seen": 115964980,
      "step": 5461
    },
    {
      "epoch": 0.6567666686707149,
      "flos": 16717729689600.0,
      "grad_norm": 3.043728499757627,
      "learning_rate": 1.1135309662915995e-06,
      "loss": 0.8615,
      "num_input_tokens_seen": 115984865,
      "step": 5462
    },
    {
      "epoch": 0.6568869115613539,
      "flos": 23377878466560.0,
      "grad_norm": 5.898699419888371,
      "learning_rate": 1.112832758929712e-06,
      "loss": 0.7901,
      "num_input_tokens_seen": 116007195,
      "step": 5463
    },
    {
      "epoch": 0.657007154451993,
      "flos": 13072771215360.0,
      "grad_norm": 5.923431433511302,
      "learning_rate": 1.11213468615039e-06,
      "loss": 0.9351,
      "num_input_tokens_seen": 116026345,
      "step": 5464
    },
    {
      "epoch": 0.6571273973426321,
      "flos": 17898326568960.0,
      "grad_norm": 3.40914486004303,
      "learning_rate": 1.1114367480595292e-06,
      "loss": 0.9515,
      "num_input_tokens_seen": 116047145,
      "step": 5465
    },
    {
      "epoch": 0.6572476402332712,
      "flos": 12416843182080.0,
      "grad_norm": 3.5005792410961587,
      "learning_rate": 1.1107389447630086e-06,
      "loss": 1.0096,
      "num_input_tokens_seen": 116065565,
      "step": 5466
    },
    {
      "epoch": 0.6573678831239103,
      "flos": 12045669089280.0,
      "grad_norm": 3.6189414070911656,
      "learning_rate": 1.1100412763666818e-06,
      "loss": 0.9787,
      "num_input_tokens_seen": 116080545,
      "step": 5467
    },
    {
      "epoch": 0.6574881260145494,
      "flos": 17001563873280.0,
      "grad_norm": 3.258801499966901,
      "learning_rate": 1.1093437429763865e-06,
      "loss": 0.9912,
      "num_input_tokens_seen": 116100530,
      "step": 5468
    },
    {
      "epoch": 0.6576083689051885,
      "flos": 7881069772800.0,
      "grad_norm": 4.24177810632474,
      "learning_rate": 1.1086463446979361e-06,
      "loss": 0.9295,
      "num_input_tokens_seen": 116118600,
      "step": 5469
    },
    {
      "epoch": 0.6577286117958275,
      "flos": 15956886282240.0,
      "grad_norm": 3.315751117748966,
      "learning_rate": 1.1079490816371277e-06,
      "loss": 0.9607,
      "num_input_tokens_seen": 116138085,
      "step": 5470
    },
    {
      "epoch": 0.6578488546864667,
      "flos": 15537684664320.0,
      "grad_norm": 3.9171588399087414,
      "learning_rate": 1.1072519538997352e-06,
      "loss": 0.9218,
      "num_input_tokens_seen": 116156945,
      "step": 5471
    },
    {
      "epoch": 0.6579690975771058,
      "flos": 16738095083520.0,
      "grad_norm": 3.1277429368326053,
      "learning_rate": 1.1065549615915095e-06,
      "loss": 1.0141,
      "num_input_tokens_seen": 116176495,
      "step": 5472
    },
    {
      "epoch": 0.6580893404677448,
      "flos": 23350921482240.0,
      "grad_norm": 3.2374344513491278,
      "learning_rate": 1.105858104818187e-06,
      "loss": 0.9587,
      "num_input_tokens_seen": 116197370,
      "step": 5473
    },
    {
      "epoch": 0.658209583358384,
      "flos": 11237012766720.0,
      "grad_norm": 6.575749274084665,
      "learning_rate": 1.105161383685478e-06,
      "loss": 0.9393,
      "num_input_tokens_seen": 116213475,
      "step": 5474
    },
    {
      "epoch": 0.658329826249023,
      "flos": 44881321820160.0,
      "grad_norm": 0.7624661031599143,
      "learning_rate": 1.1044647982990771e-06,
      "loss": 0.7901,
      "num_input_tokens_seen": 116275080,
      "step": 5475
    },
    {
      "epoch": 0.6584500691396621,
      "flos": 22616912424960.0,
      "grad_norm": 5.479251565988613,
      "learning_rate": 1.1037683487646536e-06,
      "loss": 0.8309,
      "num_input_tokens_seen": 116295085,
      "step": 5476
    },
    {
      "epoch": 0.6585703120303013,
      "flos": 13046672670720.0,
      "grad_norm": 2.5457599281762526,
      "learning_rate": 1.1030720351878583e-06,
      "loss": 0.9565,
      "num_input_tokens_seen": 116312925,
      "step": 5477
    },
    {
      "epoch": 0.6586905549209403,
      "flos": 41728127447040.0,
      "grad_norm": 0.8081934578466292,
      "learning_rate": 1.102375857674323e-06,
      "loss": 0.8058,
      "num_input_tokens_seen": 116374560,
      "step": 5478
    },
    {
      "epoch": 0.6588107978115794,
      "flos": 16191773184000.0,
      "grad_norm": 2.855672805529585,
      "learning_rate": 1.1016798163296561e-06,
      "loss": 1.0977,
      "num_input_tokens_seen": 116393480,
      "step": 5479
    },
    {
      "epoch": 0.6589310407022185,
      "flos": 14671833907200.0,
      "grad_norm": 11.971982673752288,
      "learning_rate": 1.1009839112594471e-06,
      "loss": 0.851,
      "num_input_tokens_seen": 116411225,
      "step": 5480
    },
    {
      "epoch": 0.6590512835928576,
      "flos": 18238833991680.0,
      "grad_norm": 4.689606053576797,
      "learning_rate": 1.1002881425692638e-06,
      "loss": 0.9124,
      "num_input_tokens_seen": 116431375,
      "step": 5481
    },
    {
      "epoch": 0.6591715264834966,
      "flos": 16869139660800.0,
      "grad_norm": 16.132932619104878,
      "learning_rate": 1.0995925103646532e-06,
      "loss": 0.9581,
      "num_input_tokens_seen": 116449695,
      "step": 5482
    },
    {
      "epoch": 0.6592917693741358,
      "flos": 25528628305920.0,
      "grad_norm": 2.765318341683878,
      "learning_rate": 1.0988970147511437e-06,
      "loss": 0.8652,
      "num_input_tokens_seen": 116471295,
      "step": 5483
    },
    {
      "epoch": 0.6594120122647749,
      "flos": 15195920240640.0,
      "grad_norm": 2.8066182835803986,
      "learning_rate": 1.0982016558342405e-06,
      "loss": 0.9908,
      "num_input_tokens_seen": 116489985,
      "step": 5484
    },
    {
      "epoch": 0.6595322551554139,
      "flos": 13725970636800.0,
      "grad_norm": 3.120009748492539,
      "learning_rate": 1.0975064337194291e-06,
      "loss": 0.908,
      "num_input_tokens_seen": 116507750,
      "step": 5485
    },
    {
      "epoch": 0.6596524980460531,
      "flos": 11919070003200.0,
      "grad_norm": 2.736012257459299,
      "learning_rate": 1.0968113485121743e-06,
      "loss": 0.8994,
      "num_input_tokens_seen": 116527060,
      "step": 5486
    },
    {
      "epoch": 0.6597727409366921,
      "flos": 15484690452480.0,
      "grad_norm": 2.7930871471572436,
      "learning_rate": 1.0961164003179185e-06,
      "loss": 0.9819,
      "num_input_tokens_seen": 116545290,
      "step": 5487
    },
    {
      "epoch": 0.6598929838273312,
      "flos": 16873063956480.0,
      "grad_norm": 5.778869138657538,
      "learning_rate": 1.0954215892420884e-06,
      "loss": 1.0231,
      "num_input_tokens_seen": 116565710,
      "step": 5488
    },
    {
      "epoch": 0.6600132267179702,
      "flos": 14173815459840.0,
      "grad_norm": 3.4987098495834528,
      "learning_rate": 1.094726915390082e-06,
      "loss": 0.8908,
      "num_input_tokens_seen": 116583765,
      "step": 5489
    },
    {
      "epoch": 0.6601334696086094,
      "flos": 15875279523840.0,
      "grad_norm": 4.339200278578668,
      "learning_rate": 1.0940323788672836e-06,
      "loss": 0.8919,
      "num_input_tokens_seen": 116602660,
      "step": 5490
    },
    {
      "epoch": 0.6602537124992485,
      "flos": 18290877788160.0,
      "grad_norm": 2.870259066329923,
      "learning_rate": 1.0933379797790522e-06,
      "loss": 0.9365,
      "num_input_tokens_seen": 116621795,
      "step": 5491
    },
    {
      "epoch": 0.6603739553898875,
      "flos": 18395363942400.0,
      "grad_norm": 5.37606126820756,
      "learning_rate": 1.0926437182307293e-06,
      "loss": 0.901,
      "num_input_tokens_seen": 116640325,
      "step": 5492
    },
    {
      "epoch": 0.6604941982805267,
      "flos": 17476518973440.0,
      "grad_norm": 2.5395146832928837,
      "learning_rate": 1.0919495943276338e-06,
      "loss": 0.9649,
      "num_input_tokens_seen": 116661065,
      "step": 5493
    },
    {
      "epoch": 0.6606144411711657,
      "flos": 9349578424320.0,
      "grad_norm": 4.450449785796761,
      "learning_rate": 1.0912556081750611e-06,
      "loss": 0.9369,
      "num_input_tokens_seen": 116678715,
      "step": 5494
    },
    {
      "epoch": 0.6607346840618048,
      "flos": 17897100226560.0,
      "grad_norm": 4.3325437168704655,
      "learning_rate": 1.0905617598782909e-06,
      "loss": 0.9531,
      "num_input_tokens_seen": 116698640,
      "step": 5495
    },
    {
      "epoch": 0.660854926952444,
      "flos": 12494433669120.0,
      "grad_norm": 3.6015942752123244,
      "learning_rate": 1.0898680495425775e-06,
      "loss": 1.002,
      "num_input_tokens_seen": 116716650,
      "step": 5496
    },
    {
      "epoch": 0.660975169843083,
      "flos": 11918119587840.0,
      "grad_norm": 2.6787622928737838,
      "learning_rate": 1.0891744772731594e-06,
      "loss": 1.0003,
      "num_input_tokens_seen": 116734185,
      "step": 5497
    },
    {
      "epoch": 0.6610954127337221,
      "flos": 19129771560960.0,
      "grad_norm": 3.1461086700062935,
      "learning_rate": 1.088481043175248e-06,
      "loss": 0.8471,
      "num_input_tokens_seen": 116754475,
      "step": 5498
    },
    {
      "epoch": 0.6612156556243612,
      "flos": 18838456688640.0,
      "grad_norm": 2.3493517245281117,
      "learning_rate": 1.0877877473540368e-06,
      "loss": 0.9507,
      "num_input_tokens_seen": 116774780,
      "step": 5499
    },
    {
      "epoch": 0.6613358985150003,
      "flos": 14042556272640.0,
      "grad_norm": 2.8118582947135797,
      "learning_rate": 1.0870945899147002e-06,
      "loss": 0.918,
      "num_input_tokens_seen": 116791145,
      "step": 5500
    },
    {
      "epoch": 0.6614561414056394,
      "flos": 19101588234240.0,
      "grad_norm": 4.588345417007256,
      "learning_rate": 1.0864015709623879e-06,
      "loss": 0.9425,
      "num_input_tokens_seen": 116811735,
      "step": 5501
    },
    {
      "epoch": 0.6615763842962785,
      "flos": 16272582819840.0,
      "grad_norm": 3.5773418518981095,
      "learning_rate": 1.0857086906022313e-06,
      "loss": 0.9846,
      "num_input_tokens_seen": 116829790,
      "step": 5502
    },
    {
      "epoch": 0.6616966271869176,
      "flos": 17634796462080.0,
      "grad_norm": 4.782366852808738,
      "learning_rate": 1.0850159489393388e-06,
      "loss": 0.9213,
      "num_input_tokens_seen": 116848770,
      "step": 5503
    },
    {
      "epoch": 0.6618168700775566,
      "flos": 12181404426240.0,
      "grad_norm": 3.650381481144915,
      "learning_rate": 1.0843233460787992e-06,
      "loss": 1.0108,
      "num_input_tokens_seen": 116865705,
      "step": 5504
    },
    {
      "epoch": 0.6619371129681958,
      "flos": 18106808340480.0,
      "grad_norm": 4.346515022472161,
      "learning_rate": 1.0836308821256805e-06,
      "loss": 0.9743,
      "num_input_tokens_seen": 116886225,
      "step": 5505
    },
    {
      "epoch": 0.6620573558588349,
      "flos": 12783786393600.0,
      "grad_norm": 4.047918825775581,
      "learning_rate": 1.0829385571850282e-06,
      "loss": 0.9779,
      "num_input_tokens_seen": 116902925,
      "step": 5506
    },
    {
      "epoch": 0.6621775987494739,
      "flos": 12600698019840.0,
      "grad_norm": 6.320960993557868,
      "learning_rate": 1.0822463713618679e-06,
      "loss": 1.0184,
      "num_input_tokens_seen": 116919500,
      "step": 5507
    },
    {
      "epoch": 0.6622978416401131,
      "flos": 12389978173440.0,
      "grad_norm": 2.9691809090540304,
      "learning_rate": 1.0815543247612034e-06,
      "loss": 1.0447,
      "num_input_tokens_seen": 116936290,
      "step": 5508
    },
    {
      "epoch": 0.6624180845307521,
      "flos": 15376157368320.0,
      "grad_norm": 2.8005020236508766,
      "learning_rate": 1.0808624174880168e-06,
      "loss": 1.0222,
      "num_input_tokens_seen": 116956660,
      "step": 5509
    },
    {
      "epoch": 0.6625383274213912,
      "flos": 16927407144960.0,
      "grad_norm": 2.6205145342284943,
      "learning_rate": 1.080170649647272e-06,
      "loss": 0.988,
      "num_input_tokens_seen": 116976185,
      "step": 5510
    },
    {
      "epoch": 0.6626585703120303,
      "flos": 23724977479680.0,
      "grad_norm": 2.450859043782452,
      "learning_rate": 1.0794790213439068e-06,
      "loss": 0.8614,
      "num_input_tokens_seen": 117002805,
      "step": 5511
    },
    {
      "epoch": 0.6627788132026694,
      "flos": 15690474270720.0,
      "grad_norm": 4.2335562389490855,
      "learning_rate": 1.078787532682843e-06,
      "loss": 0.9739,
      "num_input_tokens_seen": 117020000,
      "step": 5512
    },
    {
      "epoch": 0.6628990560933085,
      "flos": 25815712296960.0,
      "grad_norm": 6.788823412163054,
      "learning_rate": 1.0780961837689773e-06,
      "loss": 0.9473,
      "num_input_tokens_seen": 117039230,
      "step": 5513
    },
    {
      "epoch": 0.6630192989839476,
      "flos": 13123649986560.0,
      "grad_norm": 2.777257046103993,
      "learning_rate": 1.0774049747071883e-06,
      "loss": 0.8966,
      "num_input_tokens_seen": 117056830,
      "step": 5514
    },
    {
      "epoch": 0.6631395418745867,
      "flos": 25554144337920.0,
      "grad_norm": 3.48597783057232,
      "learning_rate": 1.076713905602332e-06,
      "loss": 0.8731,
      "num_input_tokens_seen": 117077125,
      "step": 5515
    },
    {
      "epoch": 0.6632597847652257,
      "flos": 14226472427520.0,
      "grad_norm": 4.405013972941135,
      "learning_rate": 1.07602297655924e-06,
      "loss": 1.0037,
      "num_input_tokens_seen": 117095165,
      "step": 5516
    },
    {
      "epoch": 0.6633800276558649,
      "flos": 15064875663360.0,
      "grad_norm": 2.8281884792221907,
      "learning_rate": 1.0753321876827292e-06,
      "loss": 0.9977,
      "num_input_tokens_seen": 117114170,
      "step": 5517
    },
    {
      "epoch": 0.663500270546504,
      "flos": 17058022502400.0,
      "grad_norm": 12.591910632580992,
      "learning_rate": 1.0746415390775893e-06,
      "loss": 0.9339,
      "num_input_tokens_seen": 117132020,
      "step": 5518
    },
    {
      "epoch": 0.663620513437143,
      "flos": 12706318540800.0,
      "grad_norm": 3.5418043820402687,
      "learning_rate": 1.0739510308485939e-06,
      "loss": 0.9731,
      "num_input_tokens_seen": 117148955,
      "step": 5519
    },
    {
      "epoch": 0.6637407563277821,
      "flos": 41391237734400.0,
      "grad_norm": 0.8952952624620547,
      "learning_rate": 1.07326066310049e-06,
      "loss": 0.8353,
      "num_input_tokens_seen": 117212800,
      "step": 5520
    },
    {
      "epoch": 0.6638609992184212,
      "flos": 19442524876800.0,
      "grad_norm": 3.686151678275253,
      "learning_rate": 1.0725704359380059e-06,
      "loss": 0.9811,
      "num_input_tokens_seen": 117232375,
      "step": 5521
    },
    {
      "epoch": 0.6639812421090603,
      "flos": 13203877109760.0,
      "grad_norm": 3.154157039034691,
      "learning_rate": 1.0718803494658497e-06,
      "loss": 0.9148,
      "num_input_tokens_seen": 117250985,
      "step": 5522
    },
    {
      "epoch": 0.6641014849996993,
      "flos": 11263111311360.0,
      "grad_norm": 3.9922895026566474,
      "learning_rate": 1.071190403788707e-06,
      "loss": 1.0259,
      "num_input_tokens_seen": 117266010,
      "step": 5523
    },
    {
      "epoch": 0.6642217278903385,
      "flos": 18867498455040.0,
      "grad_norm": 5.013481846868684,
      "learning_rate": 1.0705005990112415e-06,
      "loss": 0.9433,
      "num_input_tokens_seen": 117285510,
      "step": 5524
    },
    {
      "epoch": 0.6643419707809776,
      "flos": 10869885603840.0,
      "grad_norm": 3.7648361456121973,
      "learning_rate": 1.0698109352380957e-06,
      "loss": 0.9258,
      "num_input_tokens_seen": 117302830,
      "step": 5525
    },
    {
      "epoch": 0.6644622136716166,
      "flos": 17869867315200.0,
      "grad_norm": 2.7869374276299967,
      "learning_rate": 1.0691214125738909e-06,
      "loss": 0.9756,
      "num_input_tokens_seen": 117322755,
      "step": 5526
    },
    {
      "epoch": 0.6645824565622558,
      "flos": 47401314263040.0,
      "grad_norm": 0.7854809824937874,
      "learning_rate": 1.0684320311232287e-06,
      "loss": 0.8071,
      "num_input_tokens_seen": 117385380,
      "step": 5527
    },
    {
      "epoch": 0.6647026994528948,
      "flos": 17844504576000.0,
      "grad_norm": 3.5104382039814017,
      "learning_rate": 1.0677427909906865e-06,
      "loss": 0.9992,
      "num_input_tokens_seen": 117405550,
      "step": 5528
    },
    {
      "epoch": 0.6648229423435339,
      "flos": 12914309775360.0,
      "grad_norm": 4.529151377905081,
      "learning_rate": 1.0670536922808216e-06,
      "loss": 0.9014,
      "num_input_tokens_seen": 117425395,
      "step": 5529
    },
    {
      "epoch": 0.6649431852341731,
      "flos": 12968223744000.0,
      "grad_norm": 4.154081538994774,
      "learning_rate": 1.06636473509817e-06,
      "loss": 0.9028,
      "num_input_tokens_seen": 117441495,
      "step": 5530
    },
    {
      "epoch": 0.6650634281248121,
      "flos": 12050206556160.0,
      "grad_norm": 4.745981022002306,
      "learning_rate": 1.0656759195472447e-06,
      "loss": 0.9972,
      "num_input_tokens_seen": 117458505,
      "step": 5531
    },
    {
      "epoch": 0.6651836710154512,
      "flos": 49623154974720.0,
      "grad_norm": 0.8047263126827496,
      "learning_rate": 1.0649872457325414e-06,
      "loss": 0.8348,
      "num_input_tokens_seen": 117519510,
      "step": 5532
    },
    {
      "epoch": 0.6653039139060903,
      "flos": 47889705922560.0,
      "grad_norm": 0.8941040684871011,
      "learning_rate": 1.0642987137585278e-06,
      "loss": 0.8046,
      "num_input_tokens_seen": 117578755,
      "step": 5533
    },
    {
      "epoch": 0.6654241567967294,
      "flos": 15248577208320.0,
      "grad_norm": 2.8287239915677866,
      "learning_rate": 1.0636103237296561e-06,
      "loss": 1.0153,
      "num_input_tokens_seen": 117597400,
      "step": 5534
    },
    {
      "epoch": 0.6655443996873684,
      "flos": 17871492218880.0,
      "grad_norm": 3.3170739030905616,
      "learning_rate": 1.062922075750353e-06,
      "loss": 1.0253,
      "num_input_tokens_seen": 117617135,
      "step": 5535
    },
    {
      "epoch": 0.6656646425780076,
      "flos": 12364799385600.0,
      "grad_norm": 4.225457404844044,
      "learning_rate": 1.0622339699250267e-06,
      "loss": 0.9026,
      "num_input_tokens_seen": 117634775,
      "step": 5536
    },
    {
      "epoch": 0.6657848854686467,
      "flos": 16659339571200.0,
      "grad_norm": 2.7133447694029753,
      "learning_rate": 1.0615460063580624e-06,
      "loss": 0.9966,
      "num_input_tokens_seen": 117652970,
      "step": 5537
    },
    {
      "epoch": 0.6659051283592857,
      "flos": 8090471301120.0,
      "grad_norm": 3.253380850955625,
      "learning_rate": 1.060858185153821e-06,
      "loss": 0.9212,
      "num_input_tokens_seen": 117670790,
      "step": 5538
    },
    {
      "epoch": 0.6660253712499249,
      "flos": 14619391549440.0,
      "grad_norm": 5.015617014085348,
      "learning_rate": 1.0601705064166474e-06,
      "loss": 0.9497,
      "num_input_tokens_seen": 117688905,
      "step": 5539
    },
    {
      "epoch": 0.666145614140564,
      "flos": 15091556720640.0,
      "grad_norm": 3.4940139807261454,
      "learning_rate": 1.0594829702508596e-06,
      "loss": 0.9245,
      "num_input_tokens_seen": 117706340,
      "step": 5540
    },
    {
      "epoch": 0.666265857031203,
      "flos": 24056409968640.0,
      "grad_norm": 3.1198073924352374,
      "learning_rate": 1.0587955767607592e-06,
      "loss": 0.753,
      "num_input_tokens_seen": 117727920,
      "step": 5541
    },
    {
      "epoch": 0.6663860999218422,
      "flos": 12364216872960.0,
      "grad_norm": 4.486580867731064,
      "learning_rate": 1.0581083260506206e-06,
      "loss": 0.9635,
      "num_input_tokens_seen": 117744425,
      "step": 5542
    },
    {
      "epoch": 0.6665063428124812,
      "flos": 12522555678720.0,
      "grad_norm": 3.002880938374632,
      "learning_rate": 1.0574212182246993e-06,
      "loss": 0.953,
      "num_input_tokens_seen": 117762840,
      "step": 5543
    },
    {
      "epoch": 0.6666265857031203,
      "flos": 19706790789120.0,
      "grad_norm": 4.051910919731631,
      "learning_rate": 1.0567342533872303e-06,
      "loss": 0.934,
      "num_input_tokens_seen": 117782590,
      "step": 5544
    },
    {
      "epoch": 0.6667468285937594,
      "flos": 17819295129600.0,
      "grad_norm": 2.7536754322042554,
      "learning_rate": 1.0560474316424255e-06,
      "loss": 1.0048,
      "num_input_tokens_seen": 117802070,
      "step": 5545
    },
    {
      "epoch": 0.6668670714843985,
      "flos": 16189933670400.0,
      "grad_norm": 4.050053706813796,
      "learning_rate": 1.0553607530944746e-06,
      "loss": 0.924,
      "num_input_tokens_seen": 117819845,
      "step": 5546
    },
    {
      "epoch": 0.6669873143750376,
      "flos": 15747300802560.0,
      "grad_norm": 7.942814286989683,
      "learning_rate": 1.0546742178475463e-06,
      "loss": 1.0818,
      "num_input_tokens_seen": 117838560,
      "step": 5547
    },
    {
      "epoch": 0.6671075572656767,
      "flos": 14561185382400.0,
      "grad_norm": 3.378257168734777,
      "learning_rate": 1.0539878260057868e-06,
      "loss": 1.0502,
      "num_input_tokens_seen": 117857320,
      "step": 5548
    },
    {
      "epoch": 0.6672278001563158,
      "flos": 12705306808320.0,
      "grad_norm": 5.2715938255704025,
      "learning_rate": 1.0533015776733226e-06,
      "loss": 0.8734,
      "num_input_tokens_seen": 117873190,
      "step": 5549
    },
    {
      "epoch": 0.6673480430469548,
      "flos": 15874850304000.0,
      "grad_norm": 3.3467492173438247,
      "learning_rate": 1.0526154729542566e-06,
      "loss": 0.9759,
      "num_input_tokens_seen": 117892970,
      "step": 5550
    },
    {
      "epoch": 0.6674682859375939,
      "flos": 14697901793280.0,
      "grad_norm": 17.069600063794425,
      "learning_rate": 1.0519295119526699e-06,
      "loss": 0.9764,
      "num_input_tokens_seen": 117908995,
      "step": 5551
    },
    {
      "epoch": 0.667588528828233,
      "flos": 18651811921920.0,
      "grad_norm": 2.7041518184996782,
      "learning_rate": 1.0512436947726227e-06,
      "loss": 1.0249,
      "num_input_tokens_seen": 117930130,
      "step": 5552
    },
    {
      "epoch": 0.6677087717188721,
      "flos": 16398568734720.0,
      "grad_norm": 3.858977015434512,
      "learning_rate": 1.0505580215181517e-06,
      "loss": 0.8519,
      "num_input_tokens_seen": 117948090,
      "step": 5553
    },
    {
      "epoch": 0.6678290146095112,
      "flos": 50807093637120.0,
      "grad_norm": 0.8162285278206686,
      "learning_rate": 1.0498724922932753e-06,
      "loss": 0.7844,
      "num_input_tokens_seen": 118005925,
      "step": 5554
    },
    {
      "epoch": 0.6679492575001503,
      "flos": 13229883678720.0,
      "grad_norm": 4.7107028681334615,
      "learning_rate": 1.0491871072019851e-06,
      "loss": 1.0553,
      "num_input_tokens_seen": 118023535,
      "step": 5555
    },
    {
      "epoch": 0.6680695003907894,
      "flos": 21171375144960.0,
      "grad_norm": 4.8445506756963175,
      "learning_rate": 1.0485018663482555e-06,
      "loss": 0.83,
      "num_input_tokens_seen": 118043275,
      "step": 5556
    },
    {
      "epoch": 0.6681897432814284,
      "flos": 20098759495680.0,
      "grad_norm": 3.573050819896081,
      "learning_rate": 1.0478167698360354e-06,
      "loss": 0.8961,
      "num_input_tokens_seen": 118062295,
      "step": 5557
    },
    {
      "epoch": 0.6683099861720676,
      "flos": 17819049861120.0,
      "grad_norm": 5.278282460404286,
      "learning_rate": 1.0471318177692556e-06,
      "loss": 0.887,
      "num_input_tokens_seen": 118082315,
      "step": 5558
    },
    {
      "epoch": 0.6684302290627067,
      "flos": 16345206620160.0,
      "grad_norm": 8.473999648568869,
      "learning_rate": 1.046447010251821e-06,
      "loss": 0.9293,
      "num_input_tokens_seen": 118099365,
      "step": 5559
    },
    {
      "epoch": 0.6685504719533457,
      "flos": 18915985858560.0,
      "grad_norm": 4.377682762491882,
      "learning_rate": 1.0457623473876157e-06,
      "loss": 0.9523,
      "num_input_tokens_seen": 118118590,
      "step": 5560
    },
    {
      "epoch": 0.6686707148439849,
      "flos": 20650201374720.0,
      "grad_norm": 2.6962326469446576,
      "learning_rate": 1.0450778292805046e-06,
      "loss": 0.898,
      "num_input_tokens_seen": 118138295,
      "step": 5561
    },
    {
      "epoch": 0.6687909577346239,
      "flos": 16796025323520.0,
      "grad_norm": 4.693547937587475,
      "learning_rate": 1.0443934560343267e-06,
      "loss": 0.9843,
      "num_input_tokens_seen": 118159425,
      "step": 5562
    },
    {
      "epoch": 0.668911200625263,
      "flos": 16454935388160.0,
      "grad_norm": 6.049257898136272,
      "learning_rate": 1.0437092277529034e-06,
      "loss": 0.9649,
      "num_input_tokens_seen": 118178400,
      "step": 5563
    },
    {
      "epoch": 0.6690314435159022,
      "flos": 13151465410560.0,
      "grad_norm": 4.35834231159946,
      "learning_rate": 1.0430251445400292e-06,
      "loss": 0.9243,
      "num_input_tokens_seen": 118196165,
      "step": 5564
    },
    {
      "epoch": 0.6691516864065412,
      "flos": 22643501506560.0,
      "grad_norm": 2.987362948669773,
      "learning_rate": 1.0423412064994787e-06,
      "loss": 0.8142,
      "num_input_tokens_seen": 118216655,
      "step": 5565
    },
    {
      "epoch": 0.6692719292971803,
      "flos": 24924836044800.0,
      "grad_norm": 2.9184947324950024,
      "learning_rate": 1.0416574137350064e-06,
      "loss": 0.9335,
      "num_input_tokens_seen": 118237080,
      "step": 5566
    },
    {
      "epoch": 0.6693921721878194,
      "flos": 14514997370880.0,
      "grad_norm": 4.843703140524344,
      "learning_rate": 1.0409737663503428e-06,
      "loss": 0.9986,
      "num_input_tokens_seen": 118255180,
      "step": 5567
    },
    {
      "epoch": 0.6695124150784585,
      "flos": 11758615756800.0,
      "grad_norm": 3.1619284132129244,
      "learning_rate": 1.040290264449196e-06,
      "loss": 1.0265,
      "num_input_tokens_seen": 118273005,
      "step": 5568
    },
    {
      "epoch": 0.6696326579690975,
      "flos": 18972996341760.0,
      "grad_norm": 12.651518815728275,
      "learning_rate": 1.0396069081352532e-06,
      "loss": 0.8307,
      "num_input_tokens_seen": 118291880,
      "step": 5569
    },
    {
      "epoch": 0.6697529008597367,
      "flos": 47949138432000.0,
      "grad_norm": 0.8320016631834206,
      "learning_rate": 1.0389236975121782e-06,
      "loss": 0.785,
      "num_input_tokens_seen": 118346450,
      "step": 5570
    },
    {
      "epoch": 0.6698731437503758,
      "flos": 14829590200320.0,
      "grad_norm": 10.2357886815859,
      "learning_rate": 1.0382406326836147e-06,
      "loss": 0.905,
      "num_input_tokens_seen": 118365315,
      "step": 5571
    },
    {
      "epoch": 0.6699933866410148,
      "flos": 14486016921600.0,
      "grad_norm": 3.4542390571128805,
      "learning_rate": 1.0375577137531828e-06,
      "loss": 0.932,
      "num_input_tokens_seen": 118383595,
      "step": 5572
    },
    {
      "epoch": 0.670113629531654,
      "flos": 20676851773440.0,
      "grad_norm": 3.5076923307297845,
      "learning_rate": 1.0368749408244802e-06,
      "loss": 0.9118,
      "num_input_tokens_seen": 118406235,
      "step": 5573
    },
    {
      "epoch": 0.670233872422293,
      "flos": 14042372321280.0,
      "grad_norm": 4.891986693918593,
      "learning_rate": 1.0361923140010836e-06,
      "loss": 0.9693,
      "num_input_tokens_seen": 118424440,
      "step": 5574
    },
    {
      "epoch": 0.6703541153129321,
      "flos": 17242030632960.0,
      "grad_norm": 6.465534092672903,
      "learning_rate": 1.0355098333865455e-06,
      "loss": 0.8212,
      "num_input_tokens_seen": 118443390,
      "step": 5575
    },
    {
      "epoch": 0.6704743582035713,
      "flos": 18999125544960.0,
      "grad_norm": 2.6829932816770015,
      "learning_rate": 1.0348274990844006e-06,
      "loss": 0.8898,
      "num_input_tokens_seen": 118465870,
      "step": 5576
    },
    {
      "epoch": 0.6705946010942103,
      "flos": 16717944299520.0,
      "grad_norm": 3.8348673087071248,
      "learning_rate": 1.034145311198155e-06,
      "loss": 0.9132,
      "num_input_tokens_seen": 118485605,
      "step": 5577
    },
    {
      "epoch": 0.6707148439848494,
      "flos": 17110710128640.0,
      "grad_norm": 4.453728457604063,
      "learning_rate": 1.0334632698312989e-06,
      "loss": 0.8372,
      "num_input_tokens_seen": 118506120,
      "step": 5578
    },
    {
      "epoch": 0.6708350868754885,
      "flos": 16009451274240.0,
      "grad_norm": 3.6963444049275043,
      "learning_rate": 1.032781375087295e-06,
      "loss": 0.945,
      "num_input_tokens_seen": 118525740,
      "step": 5579
    },
    {
      "epoch": 0.6709553297661276,
      "flos": 17949113364480.0,
      "grad_norm": 2.7079424845054683,
      "learning_rate": 1.0320996270695891e-06,
      "loss": 0.859,
      "num_input_tokens_seen": 118546530,
      "step": 5580
    },
    {
      "epoch": 0.6710755726567667,
      "flos": 14514629468160.0,
      "grad_norm": 4.483106905715742,
      "learning_rate": 1.0314180258815998e-06,
      "loss": 0.9206,
      "num_input_tokens_seen": 118564890,
      "step": 5581
    },
    {
      "epoch": 0.6711958155474057,
      "flos": 18501015121920.0,
      "grad_norm": 3.014579473113827,
      "learning_rate": 1.0307365716267247e-06,
      "loss": 0.9338,
      "num_input_tokens_seen": 118585055,
      "step": 5582
    },
    {
      "epoch": 0.6713160584380449,
      "flos": 14147318353920.0,
      "grad_norm": 6.3456060390964995,
      "learning_rate": 1.0300552644083423e-06,
      "loss": 0.9702,
      "num_input_tokens_seen": 118603700,
      "step": 5583
    },
    {
      "epoch": 0.6714363013286839,
      "flos": 12915321507840.0,
      "grad_norm": 4.196115313957344,
      "learning_rate": 1.0293741043298036e-06,
      "loss": 0.909,
      "num_input_tokens_seen": 118621770,
      "step": 5584
    },
    {
      "epoch": 0.671556544219323,
      "flos": 18369388032000.0,
      "grad_norm": 3.6002377537525105,
      "learning_rate": 1.0286930914944436e-06,
      "loss": 0.9131,
      "num_input_tokens_seen": 118641305,
      "step": 5585
    },
    {
      "epoch": 0.6716767871099621,
      "flos": 11209749196800.0,
      "grad_norm": 8.337831376795233,
      "learning_rate": 1.0280122260055684e-06,
      "loss": 0.9662,
      "num_input_tokens_seen": 118656735,
      "step": 5586
    },
    {
      "epoch": 0.6717970300006012,
      "flos": 14016917606400.0,
      "grad_norm": 3.3952849399731386,
      "learning_rate": 1.0273315079664652e-06,
      "loss": 0.9991,
      "num_input_tokens_seen": 118674410,
      "step": 5587
    },
    {
      "epoch": 0.6719172728912403,
      "flos": 18134286520320.0,
      "grad_norm": 4.454751878672606,
      "learning_rate": 1.0266509374803992e-06,
      "loss": 0.9395,
      "num_input_tokens_seen": 118695290,
      "step": 5588
    },
    {
      "epoch": 0.6720375157818794,
      "flos": 11230359859200.0,
      "grad_norm": 6.3699365590528,
      "learning_rate": 1.0259705146506123e-06,
      "loss": 1.0418,
      "num_input_tokens_seen": 118709905,
      "step": 5589
    },
    {
      "epoch": 0.6721577586725185,
      "flos": 22823554682880.0,
      "grad_norm": 5.098822940305626,
      "learning_rate": 1.025290239580324e-06,
      "loss": 0.9677,
      "num_input_tokens_seen": 118730295,
      "step": 5590
    },
    {
      "epoch": 0.6722780015631575,
      "flos": 14722038190080.0,
      "grad_norm": 2.5517089465338803,
      "learning_rate": 1.0246101123727313e-06,
      "loss": 0.9453,
      "num_input_tokens_seen": 118748995,
      "step": 5591
    },
    {
      "epoch": 0.6723982444537967,
      "flos": 11971113799680.0,
      "grad_norm": 4.039863315196289,
      "learning_rate": 1.0239301331310085e-06,
      "loss": 0.9803,
      "num_input_tokens_seen": 118766335,
      "step": 5592
    },
    {
      "epoch": 0.6725184873444358,
      "flos": 14672079175680.0,
      "grad_norm": 3.8075878829982863,
      "learning_rate": 1.0232503019583088e-06,
      "loss": 1.0759,
      "num_input_tokens_seen": 118785665,
      "step": 5593
    },
    {
      "epoch": 0.6726387302350748,
      "flos": 16870427320320.0,
      "grad_norm": 3.6306836525357093,
      "learning_rate": 1.0225706189577619e-06,
      "loss": 0.8888,
      "num_input_tokens_seen": 118803910,
      "step": 5594
    },
    {
      "epoch": 0.672758973125714,
      "flos": 10733138534400.0,
      "grad_norm": 6.278342110054127,
      "learning_rate": 1.021891084232475e-06,
      "loss": 0.9268,
      "num_input_tokens_seen": 118821565,
      "step": 5595
    },
    {
      "epoch": 0.672879216016353,
      "flos": 12810283499520.0,
      "grad_norm": 5.071823576774029,
      "learning_rate": 1.0212116978855325e-06,
      "loss": 0.9874,
      "num_input_tokens_seen": 118839300,
      "step": 5596
    },
    {
      "epoch": 0.6729994589069921,
      "flos": 16690404802560.0,
      "grad_norm": 3.541643494174921,
      "learning_rate": 1.020532460019997e-06,
      "loss": 0.9933,
      "num_input_tokens_seen": 118858270,
      "step": 5597
    },
    {
      "epoch": 0.6731197017976313,
      "flos": 18735503462400.0,
      "grad_norm": 2.4001327636872913,
      "learning_rate": 1.0198533707389096e-06,
      "loss": 0.8941,
      "num_input_tokens_seen": 118878865,
      "step": 5598
    },
    {
      "epoch": 0.6732399446882703,
      "flos": 15353829826560.0,
      "grad_norm": 3.26631062442243,
      "learning_rate": 1.0191744301452853e-06,
      "loss": 0.9235,
      "num_input_tokens_seen": 118897885,
      "step": 5599
    },
    {
      "epoch": 0.6733601875789094,
      "flos": 18417415557120.0,
      "grad_norm": 2.711618123650663,
      "learning_rate": 1.0184956383421208e-06,
      "loss": 0.8976,
      "num_input_tokens_seen": 118916255,
      "step": 5600
    },
    {
      "epoch": 0.6734804304695485,
      "flos": 16297516339200.0,
      "grad_norm": 4.161759704193554,
      "learning_rate": 1.017816995432387e-06,
      "loss": 0.8522,
      "num_input_tokens_seen": 118935075,
      "step": 5601
    },
    {
      "epoch": 0.6736006733601876,
      "flos": 13256442101760.0,
      "grad_norm": 3.40666854106854,
      "learning_rate": 1.0171385015190353e-06,
      "loss": 0.9302,
      "num_input_tokens_seen": 118954655,
      "step": 5602
    },
    {
      "epoch": 0.6737209162508266,
      "flos": 13780926996480.0,
      "grad_norm": 3.7179567148070336,
      "learning_rate": 1.0164601567049908e-06,
      "loss": 0.9194,
      "num_input_tokens_seen": 118972905,
      "step": 5603
    },
    {
      "epoch": 0.6738411591414658,
      "flos": 14305871769600.0,
      "grad_norm": 3.3882136757946357,
      "learning_rate": 1.015781961093158e-06,
      "loss": 0.9925,
      "num_input_tokens_seen": 118991945,
      "step": 5604
    },
    {
      "epoch": 0.6739614020321049,
      "flos": 15380848128000.0,
      "grad_norm": 2.5502616972432985,
      "learning_rate": 1.0151039147864197e-06,
      "loss": 0.9669,
      "num_input_tokens_seen": 119011640,
      "step": 5605
    },
    {
      "epoch": 0.6740816449227439,
      "flos": 13596826890240.0,
      "grad_norm": 4.441555695045093,
      "learning_rate": 1.0144260178876336e-06,
      "loss": 0.8575,
      "num_input_tokens_seen": 119030705,
      "step": 5606
    },
    {
      "epoch": 0.6742018878133831,
      "flos": 14979375267840.0,
      "grad_norm": 4.266314216798847,
      "learning_rate": 1.0137482704996388e-06,
      "loss": 0.8729,
      "num_input_tokens_seen": 119044775,
      "step": 5607
    },
    {
      "epoch": 0.6743221307040221,
      "flos": 16743399014400.0,
      "grad_norm": 3.4619389929902757,
      "learning_rate": 1.0130706727252461e-06,
      "loss": 0.9867,
      "num_input_tokens_seen": 119061550,
      "step": 5608
    },
    {
      "epoch": 0.6744423735946612,
      "flos": 11496281333760.0,
      "grad_norm": 4.958538959417436,
      "learning_rate": 1.0123932246672468e-06,
      "loss": 0.8726,
      "num_input_tokens_seen": 119075415,
      "step": 5609
    },
    {
      "epoch": 0.6745626164853004,
      "flos": 41393813053440.0,
      "grad_norm": 0.8072605291124704,
      "learning_rate": 1.0117159264284114e-06,
      "loss": 0.7731,
      "num_input_tokens_seen": 119138305,
      "step": 5610
    },
    {
      "epoch": 0.6746828593759394,
      "flos": 14540666695680.0,
      "grad_norm": 2.7317212987917454,
      "learning_rate": 1.0110387781114837e-06,
      "loss": 0.9665,
      "num_input_tokens_seen": 119156640,
      "step": 5611
    },
    {
      "epoch": 0.6748031022665785,
      "flos": 13623078727680.0,
      "grad_norm": 3.522350067348253,
      "learning_rate": 1.0103617798191872e-06,
      "loss": 0.9578,
      "num_input_tokens_seen": 119175835,
      "step": 5612
    },
    {
      "epoch": 0.6749233451572175,
      "flos": 10738902343680.0,
      "grad_norm": 3.8830190561412965,
      "learning_rate": 1.0096849316542217e-06,
      "loss": 0.9938,
      "num_input_tokens_seen": 119192105,
      "step": 5613
    },
    {
      "epoch": 0.6750435880478567,
      "flos": 18862378475520.0,
      "grad_norm": 4.428157373927838,
      "learning_rate": 1.0090082337192643e-06,
      "loss": 0.9309,
      "num_input_tokens_seen": 119211470,
      "step": 5614
    },
    {
      "epoch": 0.6751638309384957,
      "flos": 16638759567360.0,
      "grad_norm": 3.7487788437321647,
      "learning_rate": 1.0083316861169705e-06,
      "loss": 0.9678,
      "num_input_tokens_seen": 119229925,
      "step": 5615
    },
    {
      "epoch": 0.6752840738291348,
      "flos": 16665501941760.0,
      "grad_norm": 3.523636433806894,
      "learning_rate": 1.0076552889499713e-06,
      "loss": 0.9123,
      "num_input_tokens_seen": 119250410,
      "step": 5616
    },
    {
      "epoch": 0.675404316719774,
      "flos": 21620201041920.0,
      "grad_norm": 3.7426495037910694,
      "learning_rate": 1.006979042320876e-06,
      "loss": 0.9278,
      "num_input_tokens_seen": 119270345,
      "step": 5617
    },
    {
      "epoch": 0.675524559610413,
      "flos": 16795565445120.0,
      "grad_norm": 3.853296350039855,
      "learning_rate": 1.0063029463322702e-06,
      "loss": 0.8302,
      "num_input_tokens_seen": 119290340,
      "step": 5618
    },
    {
      "epoch": 0.6756448025010521,
      "flos": 15089104035840.0,
      "grad_norm": 3.394847016743147,
      "learning_rate": 1.0056270010867164e-06,
      "loss": 0.9439,
      "num_input_tokens_seen": 119307630,
      "step": 5619
    },
    {
      "epoch": 0.6757650453916912,
      "flos": 15374593781760.0,
      "grad_norm": 5.2844201679882365,
      "learning_rate": 1.004951206686758e-06,
      "loss": 0.9658,
      "num_input_tokens_seen": 119325625,
      "step": 5620
    },
    {
      "epoch": 0.6758852882823303,
      "flos": 15482207109120.0,
      "grad_norm": 2.9018593713823035,
      "learning_rate": 1.0042755632349087e-06,
      "loss": 0.9004,
      "num_input_tokens_seen": 119342235,
      "step": 5621
    },
    {
      "epoch": 0.6760055311729694,
      "flos": 19286638755840.0,
      "grad_norm": 3.8131568217829392,
      "learning_rate": 1.0036000708336653e-06,
      "loss": 0.818,
      "num_input_tokens_seen": 119361085,
      "step": 5622
    },
    {
      "epoch": 0.6761257740636085,
      "flos": 12754008821760.0,
      "grad_norm": 5.312550802548912,
      "learning_rate": 1.0029247295854984e-06,
      "loss": 0.9971,
      "num_input_tokens_seen": 119377425,
      "step": 5623
    },
    {
      "epoch": 0.6762460169542476,
      "flos": 10686337351680.0,
      "grad_norm": 5.0086434994507085,
      "learning_rate": 1.0022495395928588e-06,
      "loss": 0.9059,
      "num_input_tokens_seen": 119395625,
      "step": 5624
    },
    {
      "epoch": 0.6763662598448866,
      "flos": 48613405593600.0,
      "grad_norm": 0.784987929791999,
      "learning_rate": 1.0015745009581697e-06,
      "loss": 0.8302,
      "num_input_tokens_seen": 119456950,
      "step": 5625
    },
    {
      "epoch": 0.6764865027355258,
      "flos": 14646287216640.0,
      "grad_norm": 3.8019900933246484,
      "learning_rate": 1.0008996137838343e-06,
      "loss": 0.8616,
      "num_input_tokens_seen": 119475645,
      "step": 5626
    },
    {
      "epoch": 0.6766067456261649,
      "flos": 15066010030080.0,
      "grad_norm": 8.585015980745082,
      "learning_rate": 1.000224878172234e-06,
      "loss": 0.9898,
      "num_input_tokens_seen": 119494490,
      "step": 5627
    },
    {
      "epoch": 0.6767269885168039,
      "flos": 14147778232320.0,
      "grad_norm": 4.00912555207002,
      "learning_rate": 9.99550294225724e-07,
      "loss": 0.9274,
      "num_input_tokens_seen": 119513365,
      "step": 5628
    },
    {
      "epoch": 0.6768472314074431,
      "flos": 14777239818240.0,
      "grad_norm": 3.52645589103681,
      "learning_rate": 9.988758620466402e-07,
      "loss": 0.9163,
      "num_input_tokens_seen": 119531955,
      "step": 5629
    },
    {
      "epoch": 0.6769674742980821,
      "flos": 16481953689600.0,
      "grad_norm": 2.9042412163147717,
      "learning_rate": 9.982015817372917e-07,
      "loss": 0.9487,
      "num_input_tokens_seen": 119552115,
      "step": 5630
    },
    {
      "epoch": 0.6770877171887212,
      "flos": 17241386803200.0,
      "grad_norm": 3.8595399396608663,
      "learning_rate": 9.975274533999657e-07,
      "loss": 1.0085,
      "num_input_tokens_seen": 119571365,
      "step": 5631
    },
    {
      "epoch": 0.6772079600793603,
      "flos": 12855337144320.0,
      "grad_norm": 5.44307554426863,
      "learning_rate": 9.96853477136929e-07,
      "loss": 1.0151,
      "num_input_tokens_seen": 119585830,
      "step": 5632
    },
    {
      "epoch": 0.6773282029699994,
      "flos": 15953912401920.0,
      "grad_norm": 6.280143070899483,
      "learning_rate": 9.96179653050422e-07,
      "loss": 0.937,
      "num_input_tokens_seen": 119605710,
      "step": 5633
    },
    {
      "epoch": 0.6774484458606385,
      "flos": 13252824391680.0,
      "grad_norm": 4.095463983615149,
      "learning_rate": 9.955059812426635e-07,
      "loss": 0.9246,
      "num_input_tokens_seen": 119622960,
      "step": 5634
    },
    {
      "epoch": 0.6775686887512776,
      "flos": 18499850096640.0,
      "grad_norm": 2.89947589873855,
      "learning_rate": 9.948324618158493e-07,
      "loss": 1.02,
      "num_input_tokens_seen": 119643020,
      "step": 5635
    },
    {
      "epoch": 0.6776889316419167,
      "flos": 9583729520640.0,
      "grad_norm": 4.216846630090901,
      "learning_rate": 9.941590948721502e-07,
      "loss": 0.9663,
      "num_input_tokens_seen": 119659940,
      "step": 5636
    },
    {
      "epoch": 0.6778091745325557,
      "flos": 19655053578240.0,
      "grad_norm": 3.990618193881236,
      "learning_rate": 9.934858805137188e-07,
      "loss": 0.9477,
      "num_input_tokens_seen": 119680310,
      "step": 5637
    },
    {
      "epoch": 0.6779294174231949,
      "flos": 13282326036480.0,
      "grad_norm": 3.09950907111652,
      "learning_rate": 9.92812818842677e-07,
      "loss": 1.0022,
      "num_input_tokens_seen": 119699205,
      "step": 5638
    },
    {
      "epoch": 0.678049660313834,
      "flos": 32786376314880.0,
      "grad_norm": 2.7125535704776746,
      "learning_rate": 9.921399099611306e-07,
      "loss": 0.8228,
      "num_input_tokens_seen": 119720090,
      "step": 5639
    },
    {
      "epoch": 0.678169903204473,
      "flos": 14173846118400.0,
      "grad_norm": 3.2297176424270724,
      "learning_rate": 9.914671539711588e-07,
      "loss": 0.8893,
      "num_input_tokens_seen": 119739330,
      "step": 5640
    },
    {
      "epoch": 0.6782901460951122,
      "flos": 15194877849600.0,
      "grad_norm": 4.075903510347023,
      "learning_rate": 9.90794550974817e-07,
      "loss": 0.9683,
      "num_input_tokens_seen": 119759445,
      "step": 5641
    },
    {
      "epoch": 0.6784103889857512,
      "flos": 15222723932160.0,
      "grad_norm": 4.222597529367383,
      "learning_rate": 9.901221010741407e-07,
      "loss": 0.999,
      "num_input_tokens_seen": 119778485,
      "step": 5642
    },
    {
      "epoch": 0.6785306318763903,
      "flos": 23299245588480.0,
      "grad_norm": 6.437944415409195,
      "learning_rate": 9.894498043711375e-07,
      "loss": 0.9382,
      "num_input_tokens_seen": 119799950,
      "step": 5643
    },
    {
      "epoch": 0.6786508747670293,
      "flos": 18240152309760.0,
      "grad_norm": 3.601140127839642,
      "learning_rate": 9.887776609677962e-07,
      "loss": 0.8884,
      "num_input_tokens_seen": 119821040,
      "step": 5644
    },
    {
      "epoch": 0.6787711176576685,
      "flos": 13596980183040.0,
      "grad_norm": 3.3677595674861545,
      "learning_rate": 9.88105670966079e-07,
      "loss": 0.9177,
      "num_input_tokens_seen": 119839220,
      "step": 5645
    },
    {
      "epoch": 0.6788913605483076,
      "flos": 9868606095360.0,
      "grad_norm": 2.8794915317332057,
      "learning_rate": 9.874338344679283e-07,
      "loss": 0.9797,
      "num_input_tokens_seen": 119854785,
      "step": 5646
    },
    {
      "epoch": 0.6790116034389466,
      "flos": 15642109501440.0,
      "grad_norm": 2.5769292659444316,
      "learning_rate": 9.86762151575259e-07,
      "loss": 0.9373,
      "num_input_tokens_seen": 119874500,
      "step": 5647
    },
    {
      "epoch": 0.6791318463295858,
      "flos": 14854983598080.0,
      "grad_norm": 2.1021852386040543,
      "learning_rate": 9.860906223899651e-07,
      "loss": 0.9976,
      "num_input_tokens_seen": 119893615,
      "step": 5648
    },
    {
      "epoch": 0.6792520892202248,
      "flos": 20310153830400.0,
      "grad_norm": 3.121450315501362,
      "learning_rate": 9.854192470139184e-07,
      "loss": 0.9426,
      "num_input_tokens_seen": 119914815,
      "step": 5649
    },
    {
      "epoch": 0.6793723321108639,
      "flos": 14200803102720.0,
      "grad_norm": 3.3102521647030003,
      "learning_rate": 9.847480255489645e-07,
      "loss": 0.9085,
      "num_input_tokens_seen": 119933560,
      "step": 5650
    },
    {
      "epoch": 0.6794925750015031,
      "flos": 18970727608320.0,
      "grad_norm": 4.744560138344973,
      "learning_rate": 9.840769580969295e-07,
      "loss": 0.8787,
      "num_input_tokens_seen": 119953720,
      "step": 5651
    },
    {
      "epoch": 0.6796128178921421,
      "flos": 15327976550400.0,
      "grad_norm": 2.5993843009839197,
      "learning_rate": 9.834060447596114e-07,
      "loss": 0.9863,
      "num_input_tokens_seen": 119972710,
      "step": 5652
    },
    {
      "epoch": 0.6797330607827812,
      "flos": 15983076802560.0,
      "grad_norm": 5.10927066054826,
      "learning_rate": 9.827352856387868e-07,
      "loss": 0.9717,
      "num_input_tokens_seen": 119992140,
      "step": 5653
    },
    {
      "epoch": 0.6798533036734203,
      "flos": 47476114821120.0,
      "grad_norm": 0.8626560691939712,
      "learning_rate": 9.820646808362118e-07,
      "loss": 0.8477,
      "num_input_tokens_seen": 120058115,
      "step": 5654
    },
    {
      "epoch": 0.6799735465640594,
      "flos": 11446904832000.0,
      "grad_norm": 4.332169718889901,
      "learning_rate": 9.813942304536154e-07,
      "loss": 0.8972,
      "num_input_tokens_seen": 120075805,
      "step": 5655
    },
    {
      "epoch": 0.6800937894546984,
      "flos": 15719669329920.0,
      "grad_norm": 2.4339253550221254,
      "learning_rate": 9.807239345927043e-07,
      "loss": 0.8212,
      "num_input_tokens_seen": 120095535,
      "step": 5656
    },
    {
      "epoch": 0.6802140323453376,
      "flos": 22538064936960.0,
      "grad_norm": 4.03344524848296,
      "learning_rate": 9.80053793355162e-07,
      "loss": 0.9103,
      "num_input_tokens_seen": 120113950,
      "step": 5657
    },
    {
      "epoch": 0.6803342752359767,
      "flos": 12548531589120.0,
      "grad_norm": 7.434014280915623,
      "learning_rate": 9.793838068426472e-07,
      "loss": 0.9319,
      "num_input_tokens_seen": 120131365,
      "step": 5658
    },
    {
      "epoch": 0.6804545181266157,
      "flos": 7958782894080.0,
      "grad_norm": 4.330785931803762,
      "learning_rate": 9.78713975156799e-07,
      "loss": 0.8063,
      "num_input_tokens_seen": 120146950,
      "step": 5659
    },
    {
      "epoch": 0.6805747610172549,
      "flos": 20912443822080.0,
      "grad_norm": 2.995102699824351,
      "learning_rate": 9.780442983992273e-07,
      "loss": 0.9137,
      "num_input_tokens_seen": 120165185,
      "step": 5660
    },
    {
      "epoch": 0.680695003907894,
      "flos": 26863578378240.0,
      "grad_norm": 2.570902469721397,
      "learning_rate": 9.773747766715238e-07,
      "loss": 0.9083,
      "num_input_tokens_seen": 120185725,
      "step": 5661
    },
    {
      "epoch": 0.680815246798533,
      "flos": 15721048965120.0,
      "grad_norm": 3.9641645479280143,
      "learning_rate": 9.767054100752536e-07,
      "loss": 0.9942,
      "num_input_tokens_seen": 120205395,
      "step": 5662
    },
    {
      "epoch": 0.6809354896891722,
      "flos": 12181159157760.0,
      "grad_norm": 3.0646622478836907,
      "learning_rate": 9.760361987119584e-07,
      "loss": 0.9981,
      "num_input_tokens_seen": 120222850,
      "step": 5663
    },
    {
      "epoch": 0.6810557325798112,
      "flos": 8771915366400.0,
      "grad_norm": 4.837302167923144,
      "learning_rate": 9.753671426831592e-07,
      "loss": 0.8699,
      "num_input_tokens_seen": 120238585,
      "step": 5664
    },
    {
      "epoch": 0.6811759754704503,
      "flos": 15740709212160.0,
      "grad_norm": 3.1714517846601917,
      "learning_rate": 9.746982420903483e-07,
      "loss": 0.981,
      "num_input_tokens_seen": 120256500,
      "step": 5665
    },
    {
      "epoch": 0.6812962183610894,
      "flos": 12413041520640.0,
      "grad_norm": 3.251175741119818,
      "learning_rate": 9.740294970349993e-07,
      "loss": 0.947,
      "num_input_tokens_seen": 120272635,
      "step": 5666
    },
    {
      "epoch": 0.6814164612517285,
      "flos": 43139288371200.0,
      "grad_norm": 0.9650168321554727,
      "learning_rate": 9.733609076185594e-07,
      "loss": 0.8692,
      "num_input_tokens_seen": 120328760,
      "step": 5667
    },
    {
      "epoch": 0.6815367041423676,
      "flos": 13701803581440.0,
      "grad_norm": 3.3188450341707494,
      "learning_rate": 9.72692473942455e-07,
      "loss": 1.0343,
      "num_input_tokens_seen": 120345705,
      "step": 5668
    },
    {
      "epoch": 0.6816569470330067,
      "flos": 15745430630400.0,
      "grad_norm": 4.178226736195346,
      "learning_rate": 9.720241961080849e-07,
      "loss": 0.9642,
      "num_input_tokens_seen": 120364740,
      "step": 5669
    },
    {
      "epoch": 0.6817771899236458,
      "flos": 29617384673280.0,
      "grad_norm": 5.099661924339074,
      "learning_rate": 9.713560742168259e-07,
      "loss": 0.9061,
      "num_input_tokens_seen": 120387085,
      "step": 5670
    },
    {
      "epoch": 0.6818974328142848,
      "flos": 14987162542080.0,
      "grad_norm": 3.2796756418857953,
      "learning_rate": 9.706881083700333e-07,
      "loss": 0.9093,
      "num_input_tokens_seen": 120406490,
      "step": 5671
    },
    {
      "epoch": 0.682017675704924,
      "flos": 14508773683200.0,
      "grad_norm": 4.629699381550101,
      "learning_rate": 9.700202986690357e-07,
      "loss": 1.0147,
      "num_input_tokens_seen": 120424510,
      "step": 5672
    },
    {
      "epoch": 0.682137918595563,
      "flos": 14223651840000.0,
      "grad_norm": 6.306949705959757,
      "learning_rate": 9.693526452151413e-07,
      "loss": 0.849,
      "num_input_tokens_seen": 120443280,
      "step": 5673
    },
    {
      "epoch": 0.6822581614862021,
      "flos": 22589342269440.0,
      "grad_norm": 3.235348172679706,
      "learning_rate": 9.686851481096305e-07,
      "loss": 0.9555,
      "num_input_tokens_seen": 120464310,
      "step": 5674
    },
    {
      "epoch": 0.6823784043768413,
      "flos": 16691447193600.0,
      "grad_norm": 4.370266342601265,
      "learning_rate": 9.68017807453762e-07,
      "loss": 0.9304,
      "num_input_tokens_seen": 120482775,
      "step": 5675
    },
    {
      "epoch": 0.6824986472674803,
      "flos": 9978641448960.0,
      "grad_norm": 2.622073131254304,
      "learning_rate": 9.673506233487721e-07,
      "loss": 0.9159,
      "num_input_tokens_seen": 120500460,
      "step": 5676
    },
    {
      "epoch": 0.6826188901581194,
      "flos": 15273480069120.0,
      "grad_norm": 2.807706435402156,
      "learning_rate": 9.666835958958717e-07,
      "loss": 1.0556,
      "num_input_tokens_seen": 120519500,
      "step": 5677
    },
    {
      "epoch": 0.6827391330487584,
      "flos": 14773867376640.0,
      "grad_norm": 3.4813635432630887,
      "learning_rate": 9.660167251962484e-07,
      "loss": 0.9961,
      "num_input_tokens_seen": 120537580,
      "step": 5678
    },
    {
      "epoch": 0.6828593759393976,
      "flos": 15405750988800.0,
      "grad_norm": 3.7919203844524962,
      "learning_rate": 9.653500113510654e-07,
      "loss": 0.9676,
      "num_input_tokens_seen": 120556415,
      "step": 5679
    },
    {
      "epoch": 0.6829796188300367,
      "flos": 18028206120960.0,
      "grad_norm": 6.346657373330714,
      "learning_rate": 9.646834544614627e-07,
      "loss": 0.8627,
      "num_input_tokens_seen": 120576635,
      "step": 5680
    },
    {
      "epoch": 0.6830998617206757,
      "flos": 14698729574400.0,
      "grad_norm": 5.620527057301252,
      "learning_rate": 9.64017054628558e-07,
      "loss": 0.9462,
      "num_input_tokens_seen": 120595180,
      "step": 5681
    },
    {
      "epoch": 0.6832201046113149,
      "flos": 15433137192960.0,
      "grad_norm": 3.609975789588625,
      "learning_rate": 9.63350811953441e-07,
      "loss": 0.9878,
      "num_input_tokens_seen": 120615275,
      "step": 5682
    },
    {
      "epoch": 0.6833403475019539,
      "flos": 13859038679040.0,
      "grad_norm": 4.787307193548924,
      "learning_rate": 9.626847265371826e-07,
      "loss": 0.8892,
      "num_input_tokens_seen": 120634315,
      "step": 5683
    },
    {
      "epoch": 0.683460590392593,
      "flos": 13726246563840.0,
      "grad_norm": 2.715357362752165,
      "learning_rate": 9.620187984808262e-07,
      "loss": 0.9714,
      "num_input_tokens_seen": 120652835,
      "step": 5684
    },
    {
      "epoch": 0.6835808332832322,
      "flos": 16554669465600.0,
      "grad_norm": 2.8097014172845434,
      "learning_rate": 9.613530278853919e-07,
      "loss": 1.0536,
      "num_input_tokens_seen": 120672530,
      "step": 5685
    },
    {
      "epoch": 0.6837010761738712,
      "flos": 15380480225280.0,
      "grad_norm": 4.3117237221671365,
      "learning_rate": 9.60687414851879e-07,
      "loss": 0.9314,
      "num_input_tokens_seen": 120693255,
      "step": 5686
    },
    {
      "epoch": 0.6838213190645103,
      "flos": 12443125678080.0,
      "grad_norm": 4.742862358024737,
      "learning_rate": 9.600219594812575e-07,
      "loss": 0.9518,
      "num_input_tokens_seen": 120710915,
      "step": 5687
    },
    {
      "epoch": 0.6839415619551494,
      "flos": 16429082112000.0,
      "grad_norm": 2.539647994393082,
      "learning_rate": 9.593566618744786e-07,
      "loss": 0.9193,
      "num_input_tokens_seen": 120730785,
      "step": 5688
    },
    {
      "epoch": 0.6840618048457885,
      "flos": 15721478184960.0,
      "grad_norm": 3.815034117939828,
      "learning_rate": 9.58691522132466e-07,
      "loss": 0.9227,
      "num_input_tokens_seen": 120749315,
      "step": 5689
    },
    {
      "epoch": 0.6841820477364275,
      "flos": 15640668549120.0,
      "grad_norm": 4.547038806180042,
      "learning_rate": 9.58026540356123e-07,
      "loss": 1.044,
      "num_input_tokens_seen": 120767300,
      "step": 5690
    },
    {
      "epoch": 0.6843022906270667,
      "flos": 17714103828480.0,
      "grad_norm": 3.378714742360133,
      "learning_rate": 9.573617166463246e-07,
      "loss": 1.0506,
      "num_input_tokens_seen": 120788235,
      "step": 5691
    },
    {
      "epoch": 0.6844225335177058,
      "flos": 14169891164160.0,
      "grad_norm": 4.882404865797138,
      "learning_rate": 9.56697051103924e-07,
      "loss": 0.7948,
      "num_input_tokens_seen": 120805395,
      "step": 5692
    },
    {
      "epoch": 0.6845427764083448,
      "flos": 18420113510400.0,
      "grad_norm": 4.573927898590968,
      "learning_rate": 9.560325438297522e-07,
      "loss": 0.9993,
      "num_input_tokens_seen": 120823425,
      "step": 5693
    },
    {
      "epoch": 0.684663019298984,
      "flos": 13387333386240.0,
      "grad_norm": 3.581276384559357,
      "learning_rate": 9.553681949246127e-07,
      "loss": 1.075,
      "num_input_tokens_seen": 120840770,
      "step": 5694
    },
    {
      "epoch": 0.684783262189623,
      "flos": 38765908807680.0,
      "grad_norm": 6.8132598173252505,
      "learning_rate": 9.547040044892886e-07,
      "loss": 0.9562,
      "num_input_tokens_seen": 120868005,
      "step": 5695
    },
    {
      "epoch": 0.6849035050802621,
      "flos": 45796825006080.0,
      "grad_norm": 0.861830653432165,
      "learning_rate": 9.540399726245354e-07,
      "loss": 0.8271,
      "num_input_tokens_seen": 120924430,
      "step": 5696
    },
    {
      "epoch": 0.6850237479709013,
      "flos": 17946016849920.0,
      "grad_norm": 3.057629507570596,
      "learning_rate": 9.533760994310859e-07,
      "loss": 0.8763,
      "num_input_tokens_seen": 120944550,
      "step": 5697
    },
    {
      "epoch": 0.6851439908615403,
      "flos": 13727932784640.0,
      "grad_norm": 3.742947113095597,
      "learning_rate": 9.527123850096508e-07,
      "loss": 0.9434,
      "num_input_tokens_seen": 120962630,
      "step": 5698
    },
    {
      "epoch": 0.6852642337521794,
      "flos": 16478642565120.0,
      "grad_norm": 3.5586673130174407,
      "learning_rate": 9.520488294609142e-07,
      "loss": 0.9104,
      "num_input_tokens_seen": 120981130,
      "step": 5699
    },
    {
      "epoch": 0.6853844766428185,
      "flos": 44844676730880.0,
      "grad_norm": 0.9243790514111154,
      "learning_rate": 9.513854328855368e-07,
      "loss": 0.7885,
      "num_input_tokens_seen": 121038725,
      "step": 5700
    },
    {
      "epoch": 0.6855047195334576,
      "flos": 16661945548800.0,
      "grad_norm": 5.929327550525103,
      "learning_rate": 9.507221953841558e-07,
      "loss": 1.0026,
      "num_input_tokens_seen": 121056075,
      "step": 5701
    },
    {
      "epoch": 0.6856249624240967,
      "flos": 14669534515200.0,
      "grad_norm": 2.7721277717328436,
      "learning_rate": 9.500591170573824e-07,
      "loss": 0.9687,
      "num_input_tokens_seen": 121075815,
      "step": 5702
    },
    {
      "epoch": 0.6857452053147358,
      "flos": 12098172764160.0,
      "grad_norm": 3.023233411736649,
      "learning_rate": 9.493961980058078e-07,
      "loss": 0.9361,
      "num_input_tokens_seen": 121093130,
      "step": 5703
    },
    {
      "epoch": 0.6858654482053749,
      "flos": 21987849400320.0,
      "grad_norm": 5.662550646826999,
      "learning_rate": 9.48733438329993e-07,
      "loss": 0.8681,
      "num_input_tokens_seen": 121113115,
      "step": 5704
    },
    {
      "epoch": 0.6859856910960139,
      "flos": 20572120350720.0,
      "grad_norm": 4.951519534453363,
      "learning_rate": 9.480708381304807e-07,
      "loss": 0.9252,
      "num_input_tokens_seen": 121134130,
      "step": 5705
    },
    {
      "epoch": 0.6861059339866531,
      "flos": 13728270028800.0,
      "grad_norm": 3.7255447245426905,
      "learning_rate": 9.474083975077858e-07,
      "loss": 1.0369,
      "num_input_tokens_seen": 121150975,
      "step": 5706
    },
    {
      "epoch": 0.6862261768772921,
      "flos": 15773184737280.0,
      "grad_norm": 4.528067056246974,
      "learning_rate": 9.467461165623994e-07,
      "loss": 0.9864,
      "num_input_tokens_seen": 121169745,
      "step": 5707
    },
    {
      "epoch": 0.6863464197679312,
      "flos": 18709404917760.0,
      "grad_norm": 6.97764614869278,
      "learning_rate": 9.46083995394791e-07,
      "loss": 0.9878,
      "num_input_tokens_seen": 121187275,
      "step": 5708
    },
    {
      "epoch": 0.6864666626585703,
      "flos": 26995481395200.0,
      "grad_norm": 3.5544988896562315,
      "learning_rate": 9.454220341054012e-07,
      "loss": 0.8198,
      "num_input_tokens_seen": 121211780,
      "step": 5709
    },
    {
      "epoch": 0.6865869055492094,
      "flos": 13754613841920.0,
      "grad_norm": 8.79099978649151,
      "learning_rate": 9.447602327946512e-07,
      "loss": 0.9998,
      "num_input_tokens_seen": 121230140,
      "step": 5710
    },
    {
      "epoch": 0.6867071484398485,
      "flos": 14462340403200.0,
      "grad_norm": 4.027426629087803,
      "learning_rate": 9.440985915629338e-07,
      "loss": 0.9637,
      "num_input_tokens_seen": 121247190,
      "step": 5711
    },
    {
      "epoch": 0.6868273913304875,
      "flos": 11237871206400.0,
      "grad_norm": 5.155353815052273,
      "learning_rate": 9.434371105106223e-07,
      "loss": 0.9146,
      "num_input_tokens_seen": 121264510,
      "step": 5712
    },
    {
      "epoch": 0.6869476342211267,
      "flos": 17399265730560.0,
      "grad_norm": 8.431318264989578,
      "learning_rate": 9.427757897380602e-07,
      "loss": 0.8781,
      "num_input_tokens_seen": 121283630,
      "step": 5713
    },
    {
      "epoch": 0.6870678771117658,
      "flos": 13072893849600.0,
      "grad_norm": 4.307468298442815,
      "learning_rate": 9.421146293455695e-07,
      "loss": 1.0336,
      "num_input_tokens_seen": 121299090,
      "step": 5714
    },
    {
      "epoch": 0.6871881200024048,
      "flos": 15773521981440.0,
      "grad_norm": 3.4354537422087286,
      "learning_rate": 9.414536294334489e-07,
      "loss": 0.8787,
      "num_input_tokens_seen": 121318830,
      "step": 5715
    },
    {
      "epoch": 0.687308362893044,
      "flos": 15721355550720.0,
      "grad_norm": 3.620066870603918,
      "learning_rate": 9.407927901019708e-07,
      "loss": 0.8865,
      "num_input_tokens_seen": 121337680,
      "step": 5716
    },
    {
      "epoch": 0.687428605783683,
      "flos": 17813929881600.0,
      "grad_norm": 4.852921005451897,
      "learning_rate": 9.401321114513854e-07,
      "loss": 0.9716,
      "num_input_tokens_seen": 121356295,
      "step": 5717
    },
    {
      "epoch": 0.6875488486743221,
      "flos": 16901737820160.0,
      "grad_norm": 3.997131272630977,
      "learning_rate": 9.394715935819155e-07,
      "loss": 0.9421,
      "num_input_tokens_seen": 121376405,
      "step": 5718
    },
    {
      "epoch": 0.6876690915649613,
      "flos": 18156583403520.0,
      "grad_norm": 4.426429802022882,
      "learning_rate": 9.388112365937608e-07,
      "loss": 0.8015,
      "num_input_tokens_seen": 121395590,
      "step": 5719
    },
    {
      "epoch": 0.6877893344556003,
      "flos": 13781141606400.0,
      "grad_norm": 5.385794512580134,
      "learning_rate": 9.381510405870985e-07,
      "loss": 1.0063,
      "num_input_tokens_seen": 121414325,
      "step": 5720
    },
    {
      "epoch": 0.6879095773462394,
      "flos": 13230067630080.0,
      "grad_norm": 3.726089287548566,
      "learning_rate": 9.374910056620791e-07,
      "loss": 0.9555,
      "num_input_tokens_seen": 121433110,
      "step": 5721
    },
    {
      "epoch": 0.6880298202368785,
      "flos": 14826340392960.0,
      "grad_norm": 3.272019254315346,
      "learning_rate": 9.368311319188293e-07,
      "loss": 1.0005,
      "num_input_tokens_seen": 121450645,
      "step": 5722
    },
    {
      "epoch": 0.6881500631275176,
      "flos": 21489003171840.0,
      "grad_norm": 5.220901685278955,
      "learning_rate": 9.361714194574515e-07,
      "loss": 0.9881,
      "num_input_tokens_seen": 121472700,
      "step": 5723
    },
    {
      "epoch": 0.6882703060181566,
      "flos": 47389012070400.0,
      "grad_norm": 0.7818283610533122,
      "learning_rate": 9.355118683780228e-07,
      "loss": 0.7909,
      "num_input_tokens_seen": 121542490,
      "step": 5724
    },
    {
      "epoch": 0.6883905489087958,
      "flos": 12907871477760.0,
      "grad_norm": 4.156991763358437,
      "learning_rate": 9.348524787805987e-07,
      "loss": 0.9767,
      "num_input_tokens_seen": 121557400,
      "step": 5725
    },
    {
      "epoch": 0.6885107917994349,
      "flos": 9919454208000.0,
      "grad_norm": 5.760829512947428,
      "learning_rate": 9.341932507652053e-07,
      "loss": 1.0423,
      "num_input_tokens_seen": 121571610,
      "step": 5726
    },
    {
      "epoch": 0.6886310346900739,
      "flos": 20437458063360.0,
      "grad_norm": 2.7594947763024797,
      "learning_rate": 9.335341844318489e-07,
      "loss": 0.9617,
      "num_input_tokens_seen": 121591470,
      "step": 5727
    },
    {
      "epoch": 0.6887512775807131,
      "flos": 17451585454080.0,
      "grad_norm": 2.735003764203984,
      "learning_rate": 9.328752798805091e-07,
      "loss": 0.9237,
      "num_input_tokens_seen": 121609660,
      "step": 5728
    },
    {
      "epoch": 0.6888715204713521,
      "flos": 15926648832000.0,
      "grad_norm": 3.6361584421264648,
      "learning_rate": 9.322165372111399e-07,
      "loss": 0.9412,
      "num_input_tokens_seen": 121627525,
      "step": 5729
    },
    {
      "epoch": 0.6889917633619912,
      "flos": 15668545290240.0,
      "grad_norm": 5.04570045621736,
      "learning_rate": 9.315579565236747e-07,
      "loss": 0.9463,
      "num_input_tokens_seen": 121646350,
      "step": 5730
    },
    {
      "epoch": 0.6891120062526304,
      "flos": 17030483005440.0,
      "grad_norm": 3.8824737830913425,
      "learning_rate": 9.308995379180162e-07,
      "loss": 0.9329,
      "num_input_tokens_seen": 121665625,
      "step": 5731
    },
    {
      "epoch": 0.6892322491432694,
      "flos": 45902690795520.0,
      "grad_norm": 0.7958389728073375,
      "learning_rate": 9.302412814940488e-07,
      "loss": 0.8279,
      "num_input_tokens_seen": 121728120,
      "step": 5732
    },
    {
      "epoch": 0.6893524920339085,
      "flos": 16349866721280.0,
      "grad_norm": 5.568328033112865,
      "learning_rate": 9.295831873516276e-07,
      "loss": 0.8899,
      "num_input_tokens_seen": 121747115,
      "step": 5733
    },
    {
      "epoch": 0.6894727349245476,
      "flos": 15195429703680.0,
      "grad_norm": 2.342385975964313,
      "learning_rate": 9.289252555905873e-07,
      "loss": 0.9579,
      "num_input_tokens_seen": 121766915,
      "step": 5734
    },
    {
      "epoch": 0.6895929778151867,
      "flos": 14095366533120.0,
      "grad_norm": 3.905698935739382,
      "learning_rate": 9.282674863107334e-07,
      "loss": 0.9367,
      "num_input_tokens_seen": 121784450,
      "step": 5735
    },
    {
      "epoch": 0.6897132207058257,
      "flos": 12882508738560.0,
      "grad_norm": 3.6358379276610036,
      "learning_rate": 9.276098796118488e-07,
      "loss": 0.9541,
      "num_input_tokens_seen": 121800655,
      "step": 5736
    },
    {
      "epoch": 0.6898334635964649,
      "flos": 23219876904960.0,
      "grad_norm": 4.293510534724701,
      "learning_rate": 9.269524355936938e-07,
      "loss": 0.8581,
      "num_input_tokens_seen": 121823555,
      "step": 5737
    },
    {
      "epoch": 0.689953706487104,
      "flos": 16217902387200.0,
      "grad_norm": 4.024832652149356,
      "learning_rate": 9.262951543560002e-07,
      "loss": 1.0427,
      "num_input_tokens_seen": 121842500,
      "step": 5738
    },
    {
      "epoch": 0.690073949377743,
      "flos": 13125182914560.0,
      "grad_norm": 5.50269979624615,
      "learning_rate": 9.256380359984795e-07,
      "loss": 1.0419,
      "num_input_tokens_seen": 121859330,
      "step": 5739
    },
    {
      "epoch": 0.6901941922683821,
      "flos": 24869235855360.0,
      "grad_norm": 3.6037773554170895,
      "learning_rate": 9.249810806208139e-07,
      "loss": 0.9393,
      "num_input_tokens_seen": 121878315,
      "step": 5740
    },
    {
      "epoch": 0.6903144351590212,
      "flos": 11499807068160.0,
      "grad_norm": 3.697739689133586,
      "learning_rate": 9.243242883226627e-07,
      "loss": 0.9931,
      "num_input_tokens_seen": 121897130,
      "step": 5741
    },
    {
      "epoch": 0.6904346780496603,
      "flos": 19966457917440.0,
      "grad_norm": 3.3538629777428475,
      "learning_rate": 9.236676592036628e-07,
      "loss": 0.8927,
      "num_input_tokens_seen": 121916525,
      "step": 5742
    },
    {
      "epoch": 0.6905549209402994,
      "flos": 16796975738880.0,
      "grad_norm": 2.51203596090587,
      "learning_rate": 9.230111933634228e-07,
      "loss": 0.9184,
      "num_input_tokens_seen": 121937840,
      "step": 5743
    },
    {
      "epoch": 0.6906751638309385,
      "flos": 16431136235520.0,
      "grad_norm": 3.5040988082453923,
      "learning_rate": 9.223548909015288e-07,
      "loss": 1.0049,
      "num_input_tokens_seen": 121959250,
      "step": 5744
    },
    {
      "epoch": 0.6907954067215776,
      "flos": 19442064998400.0,
      "grad_norm": 3.4734050685725877,
      "learning_rate": 9.216987519175407e-07,
      "loss": 0.9047,
      "num_input_tokens_seen": 121979145,
      "step": 5745
    },
    {
      "epoch": 0.6909156496122166,
      "flos": 15406272184320.0,
      "grad_norm": 3.594515909594589,
      "learning_rate": 9.210427765109942e-07,
      "loss": 0.8802,
      "num_input_tokens_seen": 121998540,
      "step": 5746
    },
    {
      "epoch": 0.6910358925028558,
      "flos": 16032514621440.0,
      "grad_norm": 3.4813392997533756,
      "learning_rate": 9.20386964781402e-07,
      "loss": 0.9952,
      "num_input_tokens_seen": 122016280,
      "step": 5747
    },
    {
      "epoch": 0.6911561353934949,
      "flos": 15668361338880.0,
      "grad_norm": 4.415501182855778,
      "learning_rate": 9.197313168282472e-07,
      "loss": 1.0266,
      "num_input_tokens_seen": 122033445,
      "step": 5748
    },
    {
      "epoch": 0.6912763782841339,
      "flos": 17215257600000.0,
      "grad_norm": 3.2897891875094967,
      "learning_rate": 9.190758327509935e-07,
      "loss": 0.9139,
      "num_input_tokens_seen": 122051910,
      "step": 5749
    },
    {
      "epoch": 0.6913966211747731,
      "flos": 37430705356800.0,
      "grad_norm": 0.8832055626236195,
      "learning_rate": 9.184205126490767e-07,
      "loss": 0.8617,
      "num_input_tokens_seen": 122100525,
      "step": 5750
    },
    {
      "epoch": 0.6915168640654121,
      "flos": 47788040355840.0,
      "grad_norm": 0.9606850765506185,
      "learning_rate": 9.177653566219075e-07,
      "loss": 0.8296,
      "num_input_tokens_seen": 122154970,
      "step": 5751
    },
    {
      "epoch": 0.6916371069560512,
      "flos": 12967917158400.0,
      "grad_norm": 4.454267816968162,
      "learning_rate": 9.171103647688744e-07,
      "loss": 0.942,
      "num_input_tokens_seen": 122173430,
      "step": 5752
    },
    {
      "epoch": 0.6917573498466904,
      "flos": 13937518264320.0,
      "grad_norm": 3.514721834943182,
      "learning_rate": 9.164555371893367e-07,
      "loss": 0.8774,
      "num_input_tokens_seen": 122193080,
      "step": 5753
    },
    {
      "epoch": 0.6918775927373294,
      "flos": 10031267758080.0,
      "grad_norm": 4.063590035553633,
      "learning_rate": 9.158008739826333e-07,
      "loss": 0.9484,
      "num_input_tokens_seen": 122210400,
      "step": 5754
    },
    {
      "epoch": 0.6919978356279685,
      "flos": 17056060354560.0,
      "grad_norm": 2.4707142222762144,
      "learning_rate": 9.151463752480744e-07,
      "loss": 1.0541,
      "num_input_tokens_seen": 122228850,
      "step": 5755
    },
    {
      "epoch": 0.6921180785186076,
      "flos": 16795841372160.0,
      "grad_norm": 2.1248295659550416,
      "learning_rate": 9.144920410849493e-07,
      "loss": 0.9923,
      "num_input_tokens_seen": 122249805,
      "step": 5756
    },
    {
      "epoch": 0.6922383214092467,
      "flos": 15013383720960.0,
      "grad_norm": 3.32868172014366,
      "learning_rate": 9.138378715925176e-07,
      "loss": 0.9908,
      "num_input_tokens_seen": 122268620,
      "step": 5757
    },
    {
      "epoch": 0.6923585642998857,
      "flos": 15249098403840.0,
      "grad_norm": 2.909549767707891,
      "learning_rate": 9.131838668700167e-07,
      "loss": 0.9929,
      "num_input_tokens_seen": 122288410,
      "step": 5758
    },
    {
      "epoch": 0.6924788071905249,
      "flos": 14986365419520.0,
      "grad_norm": 11.521810663766338,
      "learning_rate": 9.125300270166598e-07,
      "loss": 1.063,
      "num_input_tokens_seen": 122308735,
      "step": 5759
    },
    {
      "epoch": 0.692599050081164,
      "flos": 18683888885760.0,
      "grad_norm": 4.527827160545814,
      "learning_rate": 9.118763521316324e-07,
      "loss": 1.0581,
      "num_input_tokens_seen": 122329030,
      "step": 5760
    },
    {
      "epoch": 0.692719292971803,
      "flos": 14828118589440.0,
      "grad_norm": 3.703982225824475,
      "learning_rate": 9.112228423140987e-07,
      "loss": 0.9562,
      "num_input_tokens_seen": 122347670,
      "step": 5761
    },
    {
      "epoch": 0.6928395358624422,
      "flos": 18448143544320.0,
      "grad_norm": 4.822132527464043,
      "learning_rate": 9.105694976631932e-07,
      "loss": 1.0661,
      "num_input_tokens_seen": 122365300,
      "step": 5762
    },
    {
      "epoch": 0.6929597787530812,
      "flos": 16769712168960.0,
      "grad_norm": 4.2049726658135675,
      "learning_rate": 9.099163182780283e-07,
      "loss": 0.9252,
      "num_input_tokens_seen": 122383175,
      "step": 5763
    },
    {
      "epoch": 0.6930800216437203,
      "flos": 12937955635200.0,
      "grad_norm": 4.09153116679054,
      "learning_rate": 9.092633042576916e-07,
      "loss": 0.6889,
      "num_input_tokens_seen": 122400160,
      "step": 5764
    },
    {
      "epoch": 0.6932002645343595,
      "flos": 20781399244800.0,
      "grad_norm": 7.41288440859069,
      "learning_rate": 9.086104557012446e-07,
      "loss": 0.7585,
      "num_input_tokens_seen": 122420450,
      "step": 5765
    },
    {
      "epoch": 0.6933205074249985,
      "flos": 16953934909440.0,
      "grad_norm": 3.3383505891026064,
      "learning_rate": 9.079577727077239e-07,
      "loss": 0.8587,
      "num_input_tokens_seen": 122439000,
      "step": 5766
    },
    {
      "epoch": 0.6934407503156376,
      "flos": 17186154516480.0,
      "grad_norm": 3.975820415931324,
      "learning_rate": 9.073052553761404e-07,
      "loss": 0.8951,
      "num_input_tokens_seen": 122458085,
      "step": 5767
    },
    {
      "epoch": 0.6935609932062767,
      "flos": 14645857996800.0,
      "grad_norm": 4.675323540410816,
      "learning_rate": 9.066529038054805e-07,
      "loss": 0.9661,
      "num_input_tokens_seen": 122477870,
      "step": 5768
    },
    {
      "epoch": 0.6936812360969158,
      "flos": 12936698634240.0,
      "grad_norm": 3.5719461856133106,
      "learning_rate": 9.060007180947071e-07,
      "loss": 0.9327,
      "num_input_tokens_seen": 122495645,
      "step": 5769
    },
    {
      "epoch": 0.6938014789875548,
      "flos": 22325107015680.0,
      "grad_norm": 4.835110348662193,
      "learning_rate": 9.053486983427534e-07,
      "loss": 0.9186,
      "num_input_tokens_seen": 122516615,
      "step": 5770
    },
    {
      "epoch": 0.6939217218781939,
      "flos": 12415647498240.0,
      "grad_norm": 3.4237462331549877,
      "learning_rate": 9.046968446485326e-07,
      "loss": 0.8935,
      "num_input_tokens_seen": 122534740,
      "step": 5771
    },
    {
      "epoch": 0.6940419647688331,
      "flos": 13151342776320.0,
      "grad_norm": 3.305369205368936,
      "learning_rate": 9.040451571109295e-07,
      "loss": 0.8935,
      "num_input_tokens_seen": 122550080,
      "step": 5772
    },
    {
      "epoch": 0.6941622076594721,
      "flos": 47921353666560.0,
      "grad_norm": 0.903420173692636,
      "learning_rate": 9.033936358288042e-07,
      "loss": 0.8382,
      "num_input_tokens_seen": 122603535,
      "step": 5773
    },
    {
      "epoch": 0.6942824505501112,
      "flos": 18919603568640.0,
      "grad_norm": 6.137695234810063,
      "learning_rate": 9.027422809009937e-07,
      "loss": 1.0148,
      "num_input_tokens_seen": 122623200,
      "step": 5774
    },
    {
      "epoch": 0.6944026934407503,
      "flos": 15088828108800.0,
      "grad_norm": 2.8559311102351,
      "learning_rate": 9.020910924263054e-07,
      "loss": 1.0146,
      "num_input_tokens_seen": 122641445,
      "step": 5775
    },
    {
      "epoch": 0.6945229363313894,
      "flos": 50616095354880.0,
      "grad_norm": 0.8606374311138039,
      "learning_rate": 9.014400705035261e-07,
      "loss": 0.8163,
      "num_input_tokens_seen": 122698070,
      "step": 5776
    },
    {
      "epoch": 0.6946431792220285,
      "flos": 13439622451200.0,
      "grad_norm": 2.925761692608199,
      "learning_rate": 9.00789215231414e-07,
      "loss": 0.9667,
      "num_input_tokens_seen": 122716185,
      "step": 5777
    },
    {
      "epoch": 0.6947634221126676,
      "flos": 14435567370240.0,
      "grad_norm": 3.432007719498069,
      "learning_rate": 9.001385267087056e-07,
      "loss": 1.0155,
      "num_input_tokens_seen": 122735050,
      "step": 5778
    },
    {
      "epoch": 0.6948836650033067,
      "flos": 15509777264640.0,
      "grad_norm": 2.459825343157833,
      "learning_rate": 8.994880050341072e-07,
      "loss": 0.8918,
      "num_input_tokens_seen": 122754875,
      "step": 5779
    },
    {
      "epoch": 0.6950039078939457,
      "flos": 16820437647360.0,
      "grad_norm": 3.8258300130025527,
      "learning_rate": 8.988376503063026e-07,
      "loss": 0.9721,
      "num_input_tokens_seen": 122775855,
      "step": 5780
    },
    {
      "epoch": 0.6951241507845849,
      "flos": 15479478497280.0,
      "grad_norm": 2.9606670970794746,
      "learning_rate": 8.981874626239521e-07,
      "loss": 1.0107,
      "num_input_tokens_seen": 122794150,
      "step": 5781
    },
    {
      "epoch": 0.695244393675224,
      "flos": 10345431367680.0,
      "grad_norm": 3.7009742382094,
      "learning_rate": 8.975374420856872e-07,
      "loss": 1.0662,
      "num_input_tokens_seen": 122810765,
      "step": 5782
    },
    {
      "epoch": 0.695364636565863,
      "flos": 11944953937920.0,
      "grad_norm": 3.05562785862621,
      "learning_rate": 8.968875887901157e-07,
      "loss": 0.9098,
      "num_input_tokens_seen": 122827865,
      "step": 5783
    },
    {
      "epoch": 0.6954848794565022,
      "flos": 13728270028800.0,
      "grad_norm": 4.045149151630326,
      "learning_rate": 8.9623790283582e-07,
      "loss": 0.8033,
      "num_input_tokens_seen": 122845465,
      "step": 5784
    },
    {
      "epoch": 0.6956051223471412,
      "flos": 13466518118400.0,
      "grad_norm": 4.23308563308738,
      "learning_rate": 8.955883843213561e-07,
      "loss": 0.9523,
      "num_input_tokens_seen": 122864200,
      "step": 5785
    },
    {
      "epoch": 0.6957253652377803,
      "flos": 11394493132800.0,
      "grad_norm": 3.373713778074849,
      "learning_rate": 8.949390333452569e-07,
      "loss": 1.0741,
      "num_input_tokens_seen": 122881865,
      "step": 5786
    },
    {
      "epoch": 0.6958456081284194,
      "flos": 20939676733440.0,
      "grad_norm": 3.3946245222369815,
      "learning_rate": 8.942898500060279e-07,
      "loss": 0.8626,
      "num_input_tokens_seen": 122901300,
      "step": 5787
    },
    {
      "epoch": 0.6959658510190585,
      "flos": 17896732323840.0,
      "grad_norm": 4.4165179149601625,
      "learning_rate": 8.936408344021493e-07,
      "loss": 0.8923,
      "num_input_tokens_seen": 122917935,
      "step": 5788
    },
    {
      "epoch": 0.6960860939096976,
      "flos": 30588365414400.0,
      "grad_norm": 3.4525661102528513,
      "learning_rate": 8.929919866320765e-07,
      "loss": 0.9028,
      "num_input_tokens_seen": 122938470,
      "step": 5789
    },
    {
      "epoch": 0.6962063368003367,
      "flos": 12443278970880.0,
      "grad_norm": 3.6052006614593877,
      "learning_rate": 8.923433067942385e-07,
      "loss": 1.0006,
      "num_input_tokens_seen": 122956755,
      "step": 5790
    },
    {
      "epoch": 0.6963265796909758,
      "flos": 15091679354880.0,
      "grad_norm": 3.9852548861865205,
      "learning_rate": 8.916947949870417e-07,
      "loss": 0.8755,
      "num_input_tokens_seen": 122976140,
      "step": 5791
    },
    {
      "epoch": 0.6964468225816148,
      "flos": 49288932556800.0,
      "grad_norm": 0.8012705726779491,
      "learning_rate": 8.910464513088615e-07,
      "loss": 0.8244,
      "num_input_tokens_seen": 123039900,
      "step": 5792
    },
    {
      "epoch": 0.696567065472254,
      "flos": 13437874913280.0,
      "grad_norm": 2.9204891406541322,
      "learning_rate": 8.903982758580542e-07,
      "loss": 0.9836,
      "num_input_tokens_seen": 123058560,
      "step": 5793
    },
    {
      "epoch": 0.696687308362893,
      "flos": 16245135298560.0,
      "grad_norm": 4.637493682610252,
      "learning_rate": 8.897502687329457e-07,
      "loss": 1.0053,
      "num_input_tokens_seen": 123078080,
      "step": 5794
    },
    {
      "epoch": 0.6968075512535321,
      "flos": 17766055649280.0,
      "grad_norm": 3.105513124928633,
      "learning_rate": 8.891024300318382e-07,
      "loss": 0.9945,
      "num_input_tokens_seen": 123096370,
      "step": 5795
    },
    {
      "epoch": 0.6969277941441713,
      "flos": 14932390133760.0,
      "grad_norm": 3.106232094665132,
      "learning_rate": 8.884547598530103e-07,
      "loss": 0.954,
      "num_input_tokens_seen": 123116660,
      "step": 5796
    },
    {
      "epoch": 0.6970480370348103,
      "flos": 15327148769280.0,
      "grad_norm": 4.856788462043521,
      "learning_rate": 8.8780725829471e-07,
      "loss": 0.944,
      "num_input_tokens_seen": 123134285,
      "step": 5797
    },
    {
      "epoch": 0.6971682799254494,
      "flos": 15930757079040.0,
      "grad_norm": 4.075300034787029,
      "learning_rate": 8.87159925455165e-07,
      "loss": 0.9592,
      "num_input_tokens_seen": 123153835,
      "step": 5798
    },
    {
      "epoch": 0.6972885228160886,
      "flos": 14195621806080.0,
      "grad_norm": 2.303710213046993,
      "learning_rate": 8.865127614325738e-07,
      "loss": 0.929,
      "num_input_tokens_seen": 123171985,
      "step": 5799
    },
    {
      "epoch": 0.6974087657067276,
      "flos": 27023174184960.0,
      "grad_norm": 2.983638812431779,
      "learning_rate": 8.85865766325113e-07,
      "loss": 0.8568,
      "num_input_tokens_seen": 123195635,
      "step": 5800
    },
    {
      "epoch": 0.6975290085973667,
      "flos": 21010798264320.0,
      "grad_norm": 12.429885143384647,
      "learning_rate": 8.852189402309287e-07,
      "loss": 0.9043,
      "num_input_tokens_seen": 123214540,
      "step": 5801
    },
    {
      "epoch": 0.6976492514880057,
      "flos": 9086262927360.0,
      "grad_norm": 5.495015746936676,
      "learning_rate": 8.845722832481441e-07,
      "loss": 0.9025,
      "num_input_tokens_seen": 123229690,
      "step": 5802
    },
    {
      "epoch": 0.6977694943786449,
      "flos": 17320510218240.0,
      "grad_norm": 10.581818199336473,
      "learning_rate": 8.83925795474858e-07,
      "loss": 0.9633,
      "num_input_tokens_seen": 123249535,
      "step": 5803
    },
    {
      "epoch": 0.6978897372692839,
      "flos": 21305945456640.0,
      "grad_norm": 5.206708893810255,
      "learning_rate": 8.832794770091414e-07,
      "loss": 0.7942,
      "num_input_tokens_seen": 123270090,
      "step": 5804
    },
    {
      "epoch": 0.698009980159923,
      "flos": 15458040053760.0,
      "grad_norm": 4.059497604104416,
      "learning_rate": 8.826333279490401e-07,
      "loss": 1.0139,
      "num_input_tokens_seen": 123290445,
      "step": 5805
    },
    {
      "epoch": 0.6981302230505622,
      "flos": 14088407040000.0,
      "grad_norm": 4.880573900902726,
      "learning_rate": 8.819873483925748e-07,
      "loss": 0.8683,
      "num_input_tokens_seen": 123307285,
      "step": 5806
    },
    {
      "epoch": 0.6982504659412012,
      "flos": 15772050370560.0,
      "grad_norm": 8.356280001669917,
      "learning_rate": 8.81341538437739e-07,
      "loss": 0.9412,
      "num_input_tokens_seen": 123325295,
      "step": 5807
    },
    {
      "epoch": 0.6983707088318403,
      "flos": 25395008409600.0,
      "grad_norm": 2.8846712400457304,
      "learning_rate": 8.80695898182503e-07,
      "loss": 0.8827,
      "num_input_tokens_seen": 123345995,
      "step": 5808
    },
    {
      "epoch": 0.6984909517224794,
      "flos": 46854532485120.0,
      "grad_norm": 0.9899818956280285,
      "learning_rate": 8.800504277248093e-07,
      "loss": 0.8784,
      "num_input_tokens_seen": 123410465,
      "step": 5809
    },
    {
      "epoch": 0.6986111946131185,
      "flos": 13147203870720.0,
      "grad_norm": 3.775578809410464,
      "learning_rate": 8.794051271625753e-07,
      "loss": 0.9383,
      "num_input_tokens_seen": 123427820,
      "step": 5810
    },
    {
      "epoch": 0.6987314375037575,
      "flos": 16376302510080.0,
      "grad_norm": 3.0154521675193537,
      "learning_rate": 8.787599965936925e-07,
      "loss": 1.0318,
      "num_input_tokens_seen": 123448470,
      "step": 5811
    },
    {
      "epoch": 0.6988516803943967,
      "flos": 27416215941120.0,
      "grad_norm": 5.868417176710609,
      "learning_rate": 8.781150361160261e-07,
      "loss": 0.9215,
      "num_input_tokens_seen": 123470100,
      "step": 5812
    },
    {
      "epoch": 0.6989719232850358,
      "flos": 17136502087680.0,
      "grad_norm": 4.485945263852237,
      "learning_rate": 8.774702458274181e-07,
      "loss": 0.9292,
      "num_input_tokens_seen": 123490225,
      "step": 5813
    },
    {
      "epoch": 0.6990921661756748,
      "flos": 10503065026560.0,
      "grad_norm": 8.326413549521561,
      "learning_rate": 8.768256258256799e-07,
      "loss": 0.8803,
      "num_input_tokens_seen": 123506570,
      "step": 5814
    },
    {
      "epoch": 0.699212409066314,
      "flos": 14331387801600.0,
      "grad_norm": 6.396127733378177,
      "learning_rate": 8.76181176208602e-07,
      "loss": 0.9388,
      "num_input_tokens_seen": 123524390,
      "step": 5815
    },
    {
      "epoch": 0.699332651956953,
      "flos": 13781049630720.0,
      "grad_norm": 5.254266004760256,
      "learning_rate": 8.755368970739461e-07,
      "loss": 0.924,
      "num_input_tokens_seen": 123543470,
      "step": 5816
    },
    {
      "epoch": 0.6994528948475921,
      "flos": 11420714311680.0,
      "grad_norm": 6.731215441658474,
      "learning_rate": 8.748927885194479e-07,
      "loss": 0.8003,
      "num_input_tokens_seen": 123561495,
      "step": 5817
    },
    {
      "epoch": 0.6995731377382313,
      "flos": 46502282833920.0,
      "grad_norm": 0.7532898750119992,
      "learning_rate": 8.742488506428209e-07,
      "loss": 0.7746,
      "num_input_tokens_seen": 123620305,
      "step": 5818
    },
    {
      "epoch": 0.6996933806288703,
      "flos": 17713766584320.0,
      "grad_norm": 4.068209582651074,
      "learning_rate": 8.736050835417466e-07,
      "loss": 0.9811,
      "num_input_tokens_seen": 123640065,
      "step": 5819
    },
    {
      "epoch": 0.6998136235195094,
      "flos": 14751202590720.0,
      "grad_norm": 3.5835263991279693,
      "learning_rate": 8.729614873138862e-07,
      "loss": 0.8022,
      "num_input_tokens_seen": 123657420,
      "step": 5820
    },
    {
      "epoch": 0.6999338664101485,
      "flos": 16875424665600.0,
      "grad_norm": 4.689389877814326,
      "learning_rate": 8.723180620568716e-07,
      "loss": 0.9743,
      "num_input_tokens_seen": 123676395,
      "step": 5821
    },
    {
      "epoch": 0.7000541093007876,
      "flos": 14094753361920.0,
      "grad_norm": 2.8364220407559473,
      "learning_rate": 8.716748078683116e-07,
      "loss": 1.0555,
      "num_input_tokens_seen": 123692890,
      "step": 5822
    },
    {
      "epoch": 0.7001743521914267,
      "flos": 21148189163520.0,
      "grad_norm": 4.423210023389206,
      "learning_rate": 8.710317248457855e-07,
      "loss": 0.8772,
      "num_input_tokens_seen": 123712225,
      "step": 5823
    },
    {
      "epoch": 0.7002945950820658,
      "flos": 19574765137920.0,
      "grad_norm": 5.439878272489821,
      "learning_rate": 8.703888130868482e-07,
      "loss": 0.9109,
      "num_input_tokens_seen": 123733795,
      "step": 5824
    },
    {
      "epoch": 0.7004148379727049,
      "flos": 15742671360000.0,
      "grad_norm": 2.687185135960218,
      "learning_rate": 8.697460726890307e-07,
      "loss": 1.0101,
      "num_input_tokens_seen": 123750660,
      "step": 5825
    },
    {
      "epoch": 0.7005350808633439,
      "flos": 13777523896320.0,
      "grad_norm": 4.064240302948671,
      "learning_rate": 8.691035037498354e-07,
      "loss": 1.0884,
      "num_input_tokens_seen": 123766370,
      "step": 5826
    },
    {
      "epoch": 0.7006553237539831,
      "flos": 16690711388160.0,
      "grad_norm": 2.565677254780071,
      "learning_rate": 8.684611063667391e-07,
      "loss": 0.9235,
      "num_input_tokens_seen": 123786555,
      "step": 5827
    },
    {
      "epoch": 0.7007755666446221,
      "flos": 22250705018880.0,
      "grad_norm": 3.686289960528838,
      "learning_rate": 8.678188806371935e-07,
      "loss": 0.9539,
      "num_input_tokens_seen": 123808310,
      "step": 5828
    },
    {
      "epoch": 0.7008958095352612,
      "flos": 12862449930240.0,
      "grad_norm": 2.5752837274770437,
      "learning_rate": 8.671768266586228e-07,
      "loss": 1.0498,
      "num_input_tokens_seen": 123826155,
      "step": 5829
    },
    {
      "epoch": 0.7010160524259004,
      "flos": 19547716177920.0,
      "grad_norm": 4.4054995499704726,
      "learning_rate": 8.665349445284275e-07,
      "loss": 0.9762,
      "num_input_tokens_seen": 123845615,
      "step": 5830
    },
    {
      "epoch": 0.7011362953165394,
      "flos": 16953352396800.0,
      "grad_norm": 2.8801377870130063,
      "learning_rate": 8.658932343439799e-07,
      "loss": 1.0019,
      "num_input_tokens_seen": 123865120,
      "step": 5831
    },
    {
      "epoch": 0.7012565382071785,
      "flos": 17658135736320.0,
      "grad_norm": 3.603173944251676,
      "learning_rate": 8.65251696202627e-07,
      "loss": 0.9578,
      "num_input_tokens_seen": 123881220,
      "step": 5832
    },
    {
      "epoch": 0.7013767810978175,
      "flos": 15196441436160.0,
      "grad_norm": 5.102494112391722,
      "learning_rate": 8.646103302016896e-07,
      "loss": 1.0664,
      "num_input_tokens_seen": 123899910,
      "step": 5833
    },
    {
      "epoch": 0.7014970239884567,
      "flos": 11810751528960.0,
      "grad_norm": 3.1001548359257587,
      "learning_rate": 8.639691364384614e-07,
      "loss": 1.0747,
      "num_input_tokens_seen": 123917255,
      "step": 5834
    },
    {
      "epoch": 0.7016172668790958,
      "flos": 8851130757120.0,
      "grad_norm": 3.790839779296905,
      "learning_rate": 8.633281150102136e-07,
      "loss": 0.9131,
      "num_input_tokens_seen": 123933825,
      "step": 5835
    },
    {
      "epoch": 0.7017375097697348,
      "flos": 12360875089920.0,
      "grad_norm": 11.094385091306288,
      "learning_rate": 8.626872660141855e-07,
      "loss": 0.8716,
      "num_input_tokens_seen": 123951455,
      "step": 5836
    },
    {
      "epoch": 0.701857752660374,
      "flos": 13123190108160.0,
      "grad_norm": 3.138437752576022,
      "learning_rate": 8.620465895475957e-07,
      "loss": 0.9409,
      "num_input_tokens_seen": 123969395,
      "step": 5837
    },
    {
      "epoch": 0.701977995551013,
      "flos": 17373013893120.0,
      "grad_norm": 2.4013856543489815,
      "learning_rate": 8.614060857076333e-07,
      "loss": 0.947,
      "num_input_tokens_seen": 123989785,
      "step": 5838
    },
    {
      "epoch": 0.7020982384416521,
      "flos": 16350142648320.0,
      "grad_norm": 3.5245348792588653,
      "learning_rate": 8.60765754591462e-07,
      "loss": 0.9398,
      "num_input_tokens_seen": 124009200,
      "step": 5839
    },
    {
      "epoch": 0.7022184813322913,
      "flos": 14514905395200.0,
      "grad_norm": 3.606287044024156,
      "learning_rate": 8.601255962962211e-07,
      "loss": 0.9291,
      "num_input_tokens_seen": 124027930,
      "step": 5840
    },
    {
      "epoch": 0.7023387242229303,
      "flos": 14041360588800.0,
      "grad_norm": 5.156648007839765,
      "learning_rate": 8.594856109190194e-07,
      "loss": 0.9224,
      "num_input_tokens_seen": 124044680,
      "step": 5841
    },
    {
      "epoch": 0.7024589671135694,
      "flos": 23721850306560.0,
      "grad_norm": 3.6042563072626206,
      "learning_rate": 8.588457985569446e-07,
      "loss": 0.8917,
      "num_input_tokens_seen": 124067310,
      "step": 5842
    },
    {
      "epoch": 0.7025792100042085,
      "flos": 13544599142400.0,
      "grad_norm": 3.3486100288121095,
      "learning_rate": 8.582061593070542e-07,
      "loss": 0.9067,
      "num_input_tokens_seen": 124087760,
      "step": 5843
    },
    {
      "epoch": 0.7026994528948476,
      "flos": 13439346524160.0,
      "grad_norm": 4.2912743151426165,
      "learning_rate": 8.57566693266383e-07,
      "loss": 0.9512,
      "num_input_tokens_seen": 124105170,
      "step": 5844
    },
    {
      "epoch": 0.7028196957854866,
      "flos": 13859130654720.0,
      "grad_norm": 4.80654042808763,
      "learning_rate": 8.569274005319354e-07,
      "loss": 0.8953,
      "num_input_tokens_seen": 124123290,
      "step": 5845
    },
    {
      "epoch": 0.7029399386761258,
      "flos": 14798984847360.0,
      "grad_norm": 2.9790856588238386,
      "learning_rate": 8.562882812006913e-07,
      "loss": 0.9954,
      "num_input_tokens_seen": 124140500,
      "step": 5846
    },
    {
      "epoch": 0.7030601815667649,
      "flos": 15669127802880.0,
      "grad_norm": 2.722145328170982,
      "learning_rate": 8.556493353696066e-07,
      "loss": 0.9596,
      "num_input_tokens_seen": 124159220,
      "step": 5847
    },
    {
      "epoch": 0.7031804244574039,
      "flos": 19365731512320.0,
      "grad_norm": 4.2903688314028035,
      "learning_rate": 8.550105631356077e-07,
      "loss": 0.86,
      "num_input_tokens_seen": 124178665,
      "step": 5848
    },
    {
      "epoch": 0.7033006673480431,
      "flos": 15902113873920.0,
      "grad_norm": 9.40914760835541,
      "learning_rate": 8.543719645955961e-07,
      "loss": 0.9564,
      "num_input_tokens_seen": 124196715,
      "step": 5849
    },
    {
      "epoch": 0.7034209102386821,
      "flos": 17581189079040.0,
      "grad_norm": 4.895149064098172,
      "learning_rate": 8.537335398464467e-07,
      "loss": 0.9366,
      "num_input_tokens_seen": 124216755,
      "step": 5850
    },
    {
      "epoch": 0.7035411531293212,
      "flos": 16028007813120.0,
      "grad_norm": 4.736302955796527,
      "learning_rate": 8.53095288985007e-07,
      "loss": 1.0457,
      "num_input_tokens_seen": 124230210,
      "step": 5851
    },
    {
      "epoch": 0.7036613960199604,
      "flos": 16114642575360.0,
      "grad_norm": 4.754817527374518,
      "learning_rate": 8.524572121081009e-07,
      "loss": 1.0076,
      "num_input_tokens_seen": 124250030,
      "step": 5852
    },
    {
      "epoch": 0.7037816389105994,
      "flos": 15983199436800.0,
      "grad_norm": 4.958803790343789,
      "learning_rate": 8.518193093125232e-07,
      "loss": 0.8162,
      "num_input_tokens_seen": 124268805,
      "step": 5853
    },
    {
      "epoch": 0.7039018818012385,
      "flos": 19285535047680.0,
      "grad_norm": 2.4148627377630243,
      "learning_rate": 8.511815806950436e-07,
      "loss": 0.9979,
      "num_input_tokens_seen": 124289555,
      "step": 5854
    },
    {
      "epoch": 0.7040221246918776,
      "flos": 12575090012160.0,
      "grad_norm": 6.75702273654655,
      "learning_rate": 8.505440263524044e-07,
      "loss": 0.9674,
      "num_input_tokens_seen": 124308120,
      "step": 5855
    },
    {
      "epoch": 0.7041423675825167,
      "flos": 11517229240320.0,
      "grad_norm": 5.3409595007518895,
      "learning_rate": 8.49906646381322e-07,
      "loss": 1.0566,
      "num_input_tokens_seen": 124320675,
      "step": 5856
    },
    {
      "epoch": 0.7042626104731557,
      "flos": 18132692275200.0,
      "grad_norm": 3.8709698300202184,
      "learning_rate": 8.492694408784884e-07,
      "loss": 0.9134,
      "num_input_tokens_seen": 124340650,
      "step": 5857
    },
    {
      "epoch": 0.7043828533637949,
      "flos": 12652343255040.0,
      "grad_norm": 4.862334598189534,
      "learning_rate": 8.486324099405642e-07,
      "loss": 0.7879,
      "num_input_tokens_seen": 124357215,
      "step": 5858
    },
    {
      "epoch": 0.704503096254434,
      "flos": 21015059804160.0,
      "grad_norm": 4.61588729908577,
      "learning_rate": 8.479955536641887e-07,
      "loss": 0.9538,
      "num_input_tokens_seen": 124378430,
      "step": 5859
    },
    {
      "epoch": 0.704623339145073,
      "flos": 22040720977920.0,
      "grad_norm": 3.5273475351239383,
      "learning_rate": 8.473588721459716e-07,
      "loss": 0.8581,
      "num_input_tokens_seen": 124398060,
      "step": 5860
    },
    {
      "epoch": 0.7047435820357122,
      "flos": 17005549486080.0,
      "grad_norm": 3.529484940686559,
      "learning_rate": 8.467223654824967e-07,
      "loss": 0.8886,
      "num_input_tokens_seen": 124417235,
      "step": 5861
    },
    {
      "epoch": 0.7048638249263512,
      "flos": 33232105697280.0,
      "grad_norm": 3.7887634788583027,
      "learning_rate": 8.460860337703233e-07,
      "loss": 0.8132,
      "num_input_tokens_seen": 124437560,
      "step": 5862
    },
    {
      "epoch": 0.7049840678169903,
      "flos": 15406118891520.0,
      "grad_norm": 3.160406265459608,
      "learning_rate": 8.454498771059797e-07,
      "loss": 0.8951,
      "num_input_tokens_seen": 124456655,
      "step": 5863
    },
    {
      "epoch": 0.7051043107076294,
      "flos": 13045814231040.0,
      "grad_norm": 4.277354675000318,
      "learning_rate": 8.448138955859725e-07,
      "loss": 1.0178,
      "num_input_tokens_seen": 124472960,
      "step": 5864
    },
    {
      "epoch": 0.7052245535982685,
      "flos": 13702907289600.0,
      "grad_norm": 4.899541849009986,
      "learning_rate": 8.44178089306778e-07,
      "loss": 1.0949,
      "num_input_tokens_seen": 124490615,
      "step": 5865
    },
    {
      "epoch": 0.7053447964889076,
      "flos": 13518071377920.0,
      "grad_norm": 4.303212434078477,
      "learning_rate": 8.4354245836485e-07,
      "loss": 0.9696,
      "num_input_tokens_seen": 124508780,
      "step": 5866
    },
    {
      "epoch": 0.7054650393795466,
      "flos": 19495181844480.0,
      "grad_norm": 4.350358724601322,
      "learning_rate": 8.429070028566108e-07,
      "loss": 0.9199,
      "num_input_tokens_seen": 124529810,
      "step": 5867
    },
    {
      "epoch": 0.7055852822701858,
      "flos": 11390200934400.0,
      "grad_norm": 3.2539918299820365,
      "learning_rate": 8.422717228784586e-07,
      "loss": 0.9376,
      "num_input_tokens_seen": 124546405,
      "step": 5868
    },
    {
      "epoch": 0.7057055251608249,
      "flos": 8221546536960.0,
      "grad_norm": 3.5327442349870144,
      "learning_rate": 8.416366185267663e-07,
      "loss": 0.8934,
      "num_input_tokens_seen": 124563625,
      "step": 5869
    },
    {
      "epoch": 0.7058257680514639,
      "flos": 16140955729920.0,
      "grad_norm": 3.201620204362691,
      "learning_rate": 8.410016898978778e-07,
      "loss": 0.9693,
      "num_input_tokens_seen": 124580820,
      "step": 5870
    },
    {
      "epoch": 0.7059460109421031,
      "flos": 12416597913600.0,
      "grad_norm": 3.0875358981003127,
      "learning_rate": 8.403669370881115e-07,
      "loss": 0.9932,
      "num_input_tokens_seen": 124599275,
      "step": 5871
    },
    {
      "epoch": 0.7060662538327421,
      "flos": 16738892206080.0,
      "grad_norm": 2.9628827391024752,
      "learning_rate": 8.397323601937587e-07,
      "loss": 0.9703,
      "num_input_tokens_seen": 124618895,
      "step": 5872
    },
    {
      "epoch": 0.7061864967233812,
      "flos": 21565949829120.0,
      "grad_norm": 3.4560001883153513,
      "learning_rate": 8.390979593110838e-07,
      "loss": 0.968,
      "num_input_tokens_seen": 124640745,
      "step": 5873
    },
    {
      "epoch": 0.7063067396140204,
      "flos": 14695878328320.0,
      "grad_norm": 4.991184665205508,
      "learning_rate": 8.384637345363262e-07,
      "loss": 1.0025,
      "num_input_tokens_seen": 124659655,
      "step": 5874
    },
    {
      "epoch": 0.7064269825046594,
      "flos": 23007041617920.0,
      "grad_norm": 3.653361426567972,
      "learning_rate": 8.378296859656964e-07,
      "loss": 0.962,
      "num_input_tokens_seen": 124680530,
      "step": 5875
    },
    {
      "epoch": 0.7065472253952985,
      "flos": 21542181335040.0,
      "grad_norm": 8.675401345239111,
      "learning_rate": 8.371958136953792e-07,
      "loss": 0.8723,
      "num_input_tokens_seen": 124700280,
      "step": 5876
    },
    {
      "epoch": 0.7066674682859376,
      "flos": 11473187328000.0,
      "grad_norm": 3.4370397086181828,
      "learning_rate": 8.365621178215326e-07,
      "loss": 0.8504,
      "num_input_tokens_seen": 124716470,
      "step": 5877
    },
    {
      "epoch": 0.7067877111765767,
      "flos": 10476935823360.0,
      "grad_norm": 5.244096615853511,
      "learning_rate": 8.359285984402871e-07,
      "loss": 0.9431,
      "num_input_tokens_seen": 124733455,
      "step": 5878
    },
    {
      "epoch": 0.7069079540672157,
      "flos": 18100921896960.0,
      "grad_norm": 4.967184527919152,
      "learning_rate": 8.352952556477489e-07,
      "loss": 0.937,
      "num_input_tokens_seen": 124751085,
      "step": 5879
    },
    {
      "epoch": 0.7070281969578549,
      "flos": 17503567933440.0,
      "grad_norm": 4.179100836821395,
      "learning_rate": 8.34662089539993e-07,
      "loss": 0.9615,
      "num_input_tokens_seen": 124770315,
      "step": 5880
    },
    {
      "epoch": 0.707148439848494,
      "flos": 19024886845440.0,
      "grad_norm": 7.424565456017837,
      "learning_rate": 8.340291002130722e-07,
      "loss": 0.9667,
      "num_input_tokens_seen": 124789225,
      "step": 5881
    },
    {
      "epoch": 0.707268682739133,
      "flos": 10660392099840.0,
      "grad_norm": 5.143117932110699,
      "learning_rate": 8.3339628776301e-07,
      "loss": 0.989,
      "num_input_tokens_seen": 124807085,
      "step": 5882
    },
    {
      "epoch": 0.7073889256297722,
      "flos": 24478677442560.0,
      "grad_norm": 3.453878191109379,
      "learning_rate": 8.327636522858033e-07,
      "loss": 0.7559,
      "num_input_tokens_seen": 124826410,
      "step": 5883
    },
    {
      "epoch": 0.7075091685204112,
      "flos": 14252448337920.0,
      "grad_norm": 5.276224126293707,
      "learning_rate": 8.321311938774225e-07,
      "loss": 0.9546,
      "num_input_tokens_seen": 124845220,
      "step": 5884
    },
    {
      "epoch": 0.7076294114110503,
      "flos": 14750620078080.0,
      "grad_norm": 3.0496328539388546,
      "learning_rate": 8.314989126338104e-07,
      "loss": 0.9803,
      "num_input_tokens_seen": 124864950,
      "step": 5885
    },
    {
      "epoch": 0.7077496543016895,
      "flos": 12731527987200.0,
      "grad_norm": 3.008913309366638,
      "learning_rate": 8.308668086508847e-07,
      "loss": 1.0402,
      "num_input_tokens_seen": 124882750,
      "step": 5886
    },
    {
      "epoch": 0.7078698971923285,
      "flos": 32502756741120.0,
      "grad_norm": 3.4724366812703913,
      "learning_rate": 8.302348820245342e-07,
      "loss": 0.9303,
      "num_input_tokens_seen": 124905035,
      "step": 5887
    },
    {
      "epoch": 0.7079901400829676,
      "flos": 19182489845760.0,
      "grad_norm": 4.272005890945773,
      "learning_rate": 8.296031328506232e-07,
      "loss": 0.8864,
      "num_input_tokens_seen": 124924505,
      "step": 5888
    },
    {
      "epoch": 0.7081103829736067,
      "flos": 17370653184000.0,
      "grad_norm": 3.503820105677056,
      "learning_rate": 8.289715612249857e-07,
      "loss": 0.9599,
      "num_input_tokens_seen": 124944840,
      "step": 5889
    },
    {
      "epoch": 0.7082306258642458,
      "flos": 13144107356160.0,
      "grad_norm": 7.144786465165526,
      "learning_rate": 8.283401672434305e-07,
      "loss": 0.9646,
      "num_input_tokens_seen": 124959785,
      "step": 5890
    },
    {
      "epoch": 0.7083508687548848,
      "flos": 16691416535040.0,
      "grad_norm": 4.832896223801436,
      "learning_rate": 8.277089510017412e-07,
      "loss": 0.8967,
      "num_input_tokens_seen": 124980310,
      "step": 5891
    },
    {
      "epoch": 0.708471111645524,
      "flos": 15931002347520.0,
      "grad_norm": 3.788996052896755,
      "learning_rate": 8.270779125956719e-07,
      "loss": 1.0036,
      "num_input_tokens_seen": 125000410,
      "step": 5892
    },
    {
      "epoch": 0.7085913545361631,
      "flos": 14855351500800.0,
      "grad_norm": 5.469321607961847,
      "learning_rate": 8.264470521209505e-07,
      "loss": 0.9932,
      "num_input_tokens_seen": 125018495,
      "step": 5893
    },
    {
      "epoch": 0.7087115974268021,
      "flos": 10607612497920.0,
      "grad_norm": 6.042749066789029,
      "learning_rate": 8.258163696732785e-07,
      "loss": 0.9612,
      "num_input_tokens_seen": 125035805,
      "step": 5894
    },
    {
      "epoch": 0.7088318403174413,
      "flos": 15297616465920.0,
      "grad_norm": 2.6997537041402127,
      "learning_rate": 8.251858653483288e-07,
      "loss": 0.9566,
      "num_input_tokens_seen": 125053690,
      "step": 5895
    },
    {
      "epoch": 0.7089520832080803,
      "flos": 10969251778560.0,
      "grad_norm": 5.523612919332526,
      "learning_rate": 8.245555392417501e-07,
      "loss": 1.0476,
      "num_input_tokens_seen": 125068065,
      "step": 5896
    },
    {
      "epoch": 0.7090723260987194,
      "flos": 14488745533440.0,
      "grad_norm": 4.641887659990667,
      "learning_rate": 8.239253914491613e-07,
      "loss": 0.9804,
      "num_input_tokens_seen": 125086110,
      "step": 5897
    },
    {
      "epoch": 0.7091925689893585,
      "flos": 18266128220160.0,
      "grad_norm": 3.4205152344338625,
      "learning_rate": 8.232954220661556e-07,
      "loss": 0.9383,
      "num_input_tokens_seen": 125108565,
      "step": 5898
    },
    {
      "epoch": 0.7093128118799976,
      "flos": 17215318917120.0,
      "grad_norm": 9.21324733151138,
      "learning_rate": 8.226656311882989e-07,
      "loss": 0.8605,
      "num_input_tokens_seen": 125127595,
      "step": 5899
    },
    {
      "epoch": 0.7094330547706367,
      "flos": 11781219225600.0,
      "grad_norm": 4.054351381104006,
      "learning_rate": 8.22036018911129e-07,
      "loss": 0.9566,
      "num_input_tokens_seen": 125145345,
      "step": 5900
    },
    {
      "epoch": 0.7095532976612757,
      "flos": 11520233779200.0,
      "grad_norm": 4.225533689841942,
      "learning_rate": 8.214065853301599e-07,
      "loss": 0.9997,
      "num_input_tokens_seen": 125160595,
      "step": 5901
    },
    {
      "epoch": 0.7096735405519149,
      "flos": 50649490636800.0,
      "grad_norm": 0.7827012125733276,
      "learning_rate": 8.207773305408734e-07,
      "loss": 0.7958,
      "num_input_tokens_seen": 125227535,
      "step": 5902
    },
    {
      "epoch": 0.709793783442554,
      "flos": 16794615029760.0,
      "grad_norm": 4.282823927654286,
      "learning_rate": 8.201482546387288e-07,
      "loss": 0.9942,
      "num_input_tokens_seen": 125246730,
      "step": 5903
    },
    {
      "epoch": 0.709914026333193,
      "flos": 18499359559680.0,
      "grad_norm": 3.060566350099572,
      "learning_rate": 8.195193577191553e-07,
      "loss": 1.1078,
      "num_input_tokens_seen": 125268280,
      "step": 5904
    },
    {
      "epoch": 0.7100342692238322,
      "flos": 17685736550400.0,
      "grad_norm": 4.065722786433311,
      "learning_rate": 8.188906398775579e-07,
      "loss": 1.0368,
      "num_input_tokens_seen": 125288545,
      "step": 5905
    },
    {
      "epoch": 0.7101545121144712,
      "flos": 17736216760320.0,
      "grad_norm": 4.613051367578934,
      "learning_rate": 8.18262101209311e-07,
      "loss": 0.8723,
      "num_input_tokens_seen": 125307475,
      "step": 5906
    },
    {
      "epoch": 0.7102747550051103,
      "flos": 16900695429120.0,
      "grad_norm": 3.5257394380328084,
      "learning_rate": 8.176337418097626e-07,
      "loss": 0.9026,
      "num_input_tokens_seen": 125327665,
      "step": 5907
    },
    {
      "epoch": 0.7103949978957494,
      "flos": 10817075343360.0,
      "grad_norm": 5.061006113678428,
      "learning_rate": 8.170055617742364e-07,
      "loss": 0.9957,
      "num_input_tokens_seen": 125344665,
      "step": 5908
    },
    {
      "epoch": 0.7105152407863885,
      "flos": 16088697323520.0,
      "grad_norm": 4.622657801408815,
      "learning_rate": 8.163775611980252e-07,
      "loss": 0.8963,
      "num_input_tokens_seen": 125363495,
      "step": 5909
    },
    {
      "epoch": 0.7106354836770276,
      "flos": 12207625605120.0,
      "grad_norm": 6.308055257106503,
      "learning_rate": 8.157497401763982e-07,
      "loss": 0.9839,
      "num_input_tokens_seen": 125380880,
      "step": 5910
    },
    {
      "epoch": 0.7107557265676667,
      "flos": 14331326484480.0,
      "grad_norm": 3.3768805456883775,
      "learning_rate": 8.151220988045935e-07,
      "loss": 0.9697,
      "num_input_tokens_seen": 125399855,
      "step": 5911
    },
    {
      "epoch": 0.7108759694583058,
      "flos": 15275258265600.0,
      "grad_norm": 3.476030424742124,
      "learning_rate": 8.144946371778234e-07,
      "loss": 1.028,
      "num_input_tokens_seen": 125419685,
      "step": 5912
    },
    {
      "epoch": 0.7109962123489448,
      "flos": 17111476592640.0,
      "grad_norm": 4.818941041793542,
      "learning_rate": 8.138673553912751e-07,
      "loss": 0.9759,
      "num_input_tokens_seen": 125439965,
      "step": 5913
    },
    {
      "epoch": 0.711116455239584,
      "flos": 21723982049280.0,
      "grad_norm": 4.591175006406653,
      "learning_rate": 8.132402535401059e-07,
      "loss": 0.7627,
      "num_input_tokens_seen": 125460940,
      "step": 5914
    },
    {
      "epoch": 0.711236698130223,
      "flos": 17818252738560.0,
      "grad_norm": 3.0066115322541735,
      "learning_rate": 8.126133317194465e-07,
      "loss": 0.935,
      "num_input_tokens_seen": 125480850,
      "step": 5915
    },
    {
      "epoch": 0.7113569410208621,
      "flos": 17215380234240.0,
      "grad_norm": 2.8374071753018795,
      "learning_rate": 8.11986590024401e-07,
      "loss": 0.9331,
      "num_input_tokens_seen": 125500310,
      "step": 5916
    },
    {
      "epoch": 0.7114771839115013,
      "flos": 25287364423680.0,
      "grad_norm": 6.822523949267608,
      "learning_rate": 8.113600285500442e-07,
      "loss": 0.8838,
      "num_input_tokens_seen": 125520240,
      "step": 5917
    },
    {
      "epoch": 0.7115974268021403,
      "flos": 14981460049920.0,
      "grad_norm": 4.381820376024017,
      "learning_rate": 8.107336473914268e-07,
      "loss": 0.9362,
      "num_input_tokens_seen": 125538590,
      "step": 5918
    },
    {
      "epoch": 0.7117176696927794,
      "flos": 40610734080000.0,
      "grad_norm": 0.8205090241944234,
      "learning_rate": 8.101074466435694e-07,
      "loss": 0.7993,
      "num_input_tokens_seen": 125597785,
      "step": 5919
    },
    {
      "epoch": 0.7118379125834186,
      "flos": 11263785799680.0,
      "grad_norm": 2.6599266188402164,
      "learning_rate": 8.094814264014662e-07,
      "loss": 0.8736,
      "num_input_tokens_seen": 125616260,
      "step": 5920
    },
    {
      "epoch": 0.7119581554740576,
      "flos": 14331081216000.0,
      "grad_norm": 2.993450963768298,
      "learning_rate": 8.088555867600844e-07,
      "loss": 1.0105,
      "num_input_tokens_seen": 125632145,
      "step": 5921
    },
    {
      "epoch": 0.7120783983646967,
      "flos": 24767876874240.0,
      "grad_norm": 3.1724531276357575,
      "learning_rate": 8.08229927814362e-07,
      "loss": 0.814,
      "num_input_tokens_seen": 125654755,
      "step": 5922
    },
    {
      "epoch": 0.7121986412553358,
      "flos": 18762245836800.0,
      "grad_norm": 5.501628807944541,
      "learning_rate": 8.076044496592134e-07,
      "loss": 0.8367,
      "num_input_tokens_seen": 125676325,
      "step": 5923
    },
    {
      "epoch": 0.7123188841459749,
      "flos": 7828504780800.0,
      "grad_norm": 4.288262853793206,
      "learning_rate": 8.069791523895204e-07,
      "loss": 0.9709,
      "num_input_tokens_seen": 125692495,
      "step": 5924
    },
    {
      "epoch": 0.7124391270366139,
      "flos": 14775185694720.0,
      "grad_norm": 3.9835590798694147,
      "learning_rate": 8.063540361001422e-07,
      "loss": 0.9677,
      "num_input_tokens_seen": 125710785,
      "step": 5925
    },
    {
      "epoch": 0.7125593699272531,
      "flos": 12470082662400.0,
      "grad_norm": 3.6175883588037543,
      "learning_rate": 8.057291008859069e-07,
      "loss": 0.984,
      "num_input_tokens_seen": 125728665,
      "step": 5926
    },
    {
      "epoch": 0.7126796128178922,
      "flos": 20411083591680.0,
      "grad_norm": 3.3982121535381076,
      "learning_rate": 8.051043468416187e-07,
      "loss": 0.8723,
      "num_input_tokens_seen": 125749635,
      "step": 5927
    },
    {
      "epoch": 0.7127998557085312,
      "flos": 11342204067840.0,
      "grad_norm": 3.117936964453682,
      "learning_rate": 8.044797740620506e-07,
      "loss": 1.0134,
      "num_input_tokens_seen": 125767960,
      "step": 5928
    },
    {
      "epoch": 0.7129200985991703,
      "flos": 16637870469120.0,
      "grad_norm": 3.006177264501458,
      "learning_rate": 8.038553826419494e-07,
      "loss": 0.9801,
      "num_input_tokens_seen": 125786390,
      "step": 5929
    },
    {
      "epoch": 0.7130403414898094,
      "flos": 15196441436160.0,
      "grad_norm": 2.8752197513354067,
      "learning_rate": 8.032311726760364e-07,
      "loss": 1.0001,
      "num_input_tokens_seen": 125807330,
      "step": 5930
    },
    {
      "epoch": 0.7131605843804485,
      "flos": 53715506503680.0,
      "grad_norm": 2.700664922290704,
      "learning_rate": 8.026071442590022e-07,
      "loss": 0.8865,
      "num_input_tokens_seen": 125833980,
      "step": 5931
    },
    {
      "epoch": 0.7132808272710875,
      "flos": 13019715686400.0,
      "grad_norm": 4.7563636716074305,
      "learning_rate": 8.019832974855134e-07,
      "loss": 1.005,
      "num_input_tokens_seen": 125851660,
      "step": 5932
    },
    {
      "epoch": 0.7134010701617267,
      "flos": 16529613312000.0,
      "grad_norm": 4.246187407330127,
      "learning_rate": 8.013596324502052e-07,
      "loss": 1.0233,
      "num_input_tokens_seen": 125869845,
      "step": 5933
    },
    {
      "epoch": 0.7135213130523658,
      "flos": 16816697303040.0,
      "grad_norm": 3.277111095222762,
      "learning_rate": 8.007361492476872e-07,
      "loss": 0.9759,
      "num_input_tokens_seen": 125888890,
      "step": 5934
    },
    {
      "epoch": 0.7136415559430048,
      "flos": 17635072389120.0,
      "grad_norm": 3.164326673240756,
      "learning_rate": 8.001128479725426e-07,
      "loss": 0.977,
      "num_input_tokens_seen": 125910515,
      "step": 5935
    },
    {
      "epoch": 0.713761798833644,
      "flos": 12968285061120.0,
      "grad_norm": 3.374870297800723,
      "learning_rate": 7.994897287193248e-07,
      "loss": 0.9955,
      "num_input_tokens_seen": 125929615,
      "step": 5936
    },
    {
      "epoch": 0.713882041724283,
      "flos": 10999734497280.0,
      "grad_norm": 3.752350007273762,
      "learning_rate": 7.988667915825605e-07,
      "loss": 1.0191,
      "num_input_tokens_seen": 125946400,
      "step": 5937
    },
    {
      "epoch": 0.7140022846149221,
      "flos": 17110710128640.0,
      "grad_norm": 4.240288803053488,
      "learning_rate": 7.982440366567491e-07,
      "loss": 0.9386,
      "num_input_tokens_seen": 125964610,
      "step": 5938
    },
    {
      "epoch": 0.7141225275055613,
      "flos": 19863719301120.0,
      "grad_norm": 3.309061965329913,
      "learning_rate": 7.97621464036361e-07,
      "loss": 0.9455,
      "num_input_tokens_seen": 125986090,
      "step": 5939
    },
    {
      "epoch": 0.7142427703962003,
      "flos": 13964015370240.0,
      "grad_norm": 3.7419123857069567,
      "learning_rate": 7.969990738158417e-07,
      "loss": 0.8693,
      "num_input_tokens_seen": 126004220,
      "step": 5940
    },
    {
      "epoch": 0.7143630132868394,
      "flos": 14934045696000.0,
      "grad_norm": 3.0523224477656807,
      "learning_rate": 7.963768660896062e-07,
      "loss": 1.0393,
      "num_input_tokens_seen": 126022350,
      "step": 5941
    },
    {
      "epoch": 0.7144832561774785,
      "flos": 17159688069120.0,
      "grad_norm": 4.986142983979646,
      "learning_rate": 7.957548409520432e-07,
      "loss": 1.02,
      "num_input_tokens_seen": 126041295,
      "step": 5942
    },
    {
      "epoch": 0.7146034990681176,
      "flos": 11551544279040.0,
      "grad_norm": 3.15927553105194,
      "learning_rate": 7.951329984975135e-07,
      "loss": 1.0283,
      "num_input_tokens_seen": 126057955,
      "step": 5943
    },
    {
      "epoch": 0.7147237419587567,
      "flos": 49862548684800.0,
      "grad_norm": 0.7265828447672423,
      "learning_rate": 7.94511338820349e-07,
      "loss": 0.7546,
      "num_input_tokens_seen": 126119980,
      "step": 5944
    },
    {
      "epoch": 0.7148439848493958,
      "flos": 15820384481280.0,
      "grad_norm": 4.198643220007627,
      "learning_rate": 7.938898620148575e-07,
      "loss": 0.98,
      "num_input_tokens_seen": 126137460,
      "step": 5945
    },
    {
      "epoch": 0.7149642277400349,
      "flos": 12705460101120.0,
      "grad_norm": 4.262923394707044,
      "learning_rate": 7.932685681753135e-07,
      "loss": 0.9088,
      "num_input_tokens_seen": 126154460,
      "step": 5946
    },
    {
      "epoch": 0.7150844706306739,
      "flos": 22587104194560.0,
      "grad_norm": 5.159475531427532,
      "learning_rate": 7.92647457395969e-07,
      "loss": 0.8133,
      "num_input_tokens_seen": 126176005,
      "step": 5947
    },
    {
      "epoch": 0.7152047135213131,
      "flos": 7670993756160.0,
      "grad_norm": 6.164848093851435,
      "learning_rate": 7.920265297710444e-07,
      "loss": 0.9361,
      "num_input_tokens_seen": 126193115,
      "step": 5948
    },
    {
      "epoch": 0.7153249564119522,
      "flos": 14907609907200.0,
      "grad_norm": 6.249848936162659,
      "learning_rate": 7.914057853947363e-07,
      "loss": 0.9158,
      "num_input_tokens_seen": 126212015,
      "step": 5949
    },
    {
      "epoch": 0.7154451993025912,
      "flos": 17241540096000.0,
      "grad_norm": 3.672749386831571,
      "learning_rate": 7.907852243612089e-07,
      "loss": 0.8298,
      "num_input_tokens_seen": 126232140,
      "step": 5950
    },
    {
      "epoch": 0.7155654421932304,
      "flos": 16585979965440.0,
      "grad_norm": 2.7249213146083475,
      "learning_rate": 7.901648467646009e-07,
      "loss": 0.9199,
      "num_input_tokens_seen": 126250800,
      "step": 5951
    },
    {
      "epoch": 0.7156856850838694,
      "flos": 16141231656960.0,
      "grad_norm": 2.9682897913754136,
      "learning_rate": 7.895446526990244e-07,
      "loss": 0.9152,
      "num_input_tokens_seen": 126270535,
      "step": 5952
    },
    {
      "epoch": 0.7158059279745085,
      "flos": 14095458508800.0,
      "grad_norm": 2.4879804636013447,
      "learning_rate": 7.889246422585609e-07,
      "loss": 0.9434,
      "num_input_tokens_seen": 126289640,
      "step": 5953
    },
    {
      "epoch": 0.7159261708651476,
      "flos": 17163826974720.0,
      "grad_norm": 3.5897746779943627,
      "learning_rate": 7.883048155372675e-07,
      "loss": 0.9327,
      "num_input_tokens_seen": 126307875,
      "step": 5954
    },
    {
      "epoch": 0.7160464137557867,
      "flos": 12024077352960.0,
      "grad_norm": 6.687139198338778,
      "learning_rate": 7.876851726291698e-07,
      "loss": 0.8919,
      "num_input_tokens_seen": 126325895,
      "step": 5955
    },
    {
      "epoch": 0.7161666566464258,
      "flos": 17949542584320.0,
      "grad_norm": 4.074738130170793,
      "learning_rate": 7.870657136282666e-07,
      "loss": 0.9675,
      "num_input_tokens_seen": 126344475,
      "step": 5956
    },
    {
      "epoch": 0.7162868995370649,
      "flos": 18840756080640.0,
      "grad_norm": 2.363833528411577,
      "learning_rate": 7.86446438628531e-07,
      "loss": 1.0136,
      "num_input_tokens_seen": 126365265,
      "step": 5957
    },
    {
      "epoch": 0.716407142427704,
      "flos": 50130646917120.0,
      "grad_norm": 0.8043121277158146,
      "learning_rate": 7.858273477239059e-07,
      "loss": 0.7901,
      "num_input_tokens_seen": 126433405,
      "step": 5958
    },
    {
      "epoch": 0.716527385318343,
      "flos": 14724245606400.0,
      "grad_norm": 4.027361983545655,
      "learning_rate": 7.852084410083067e-07,
      "loss": 0.9023,
      "num_input_tokens_seen": 126451945,
      "step": 5959
    },
    {
      "epoch": 0.7166476282089821,
      "flos": 18052526469120.0,
      "grad_norm": 2.4229210348136365,
      "learning_rate": 7.84589718575621e-07,
      "loss": 0.8276,
      "num_input_tokens_seen": 126472110,
      "step": 5960
    },
    {
      "epoch": 0.7167678710996213,
      "flos": 17162937876480.0,
      "grad_norm": 6.6940147539881725,
      "learning_rate": 7.83971180519708e-07,
      "loss": 0.8801,
      "num_input_tokens_seen": 126490685,
      "step": 5961
    },
    {
      "epoch": 0.7168881139902603,
      "flos": 21541721456640.0,
      "grad_norm": 5.1401726891864445,
      "learning_rate": 7.833528269344008e-07,
      "loss": 0.9385,
      "num_input_tokens_seen": 126510310,
      "step": 5962
    },
    {
      "epoch": 0.7170083568808994,
      "flos": 10315224576000.0,
      "grad_norm": 4.28021762401581,
      "learning_rate": 7.827346579135023e-07,
      "loss": 0.9728,
      "num_input_tokens_seen": 126527370,
      "step": 5963
    },
    {
      "epoch": 0.7171285997715385,
      "flos": 16586102599680.0,
      "grad_norm": 5.905117027163152,
      "learning_rate": 7.821166735507885e-07,
      "loss": 1.0284,
      "num_input_tokens_seen": 126546120,
      "step": 5964
    },
    {
      "epoch": 0.7172488426621776,
      "flos": 11707859619840.0,
      "grad_norm": 3.4006957520337497,
      "learning_rate": 7.81498873940007e-07,
      "loss": 0.8767,
      "num_input_tokens_seen": 126563055,
      "step": 5965
    },
    {
      "epoch": 0.7173690855528166,
      "flos": 18893413048320.0,
      "grad_norm": 4.819904922244074,
      "learning_rate": 7.808812591748768e-07,
      "loss": 0.9537,
      "num_input_tokens_seen": 126583155,
      "step": 5966
    },
    {
      "epoch": 0.7174893284434558,
      "flos": 16193030184960.0,
      "grad_norm": 3.7840926149256653,
      "learning_rate": 7.802638293490915e-07,
      "loss": 0.8387,
      "num_input_tokens_seen": 126602520,
      "step": 5967
    },
    {
      "epoch": 0.7176095713340949,
      "flos": 16558992322560.0,
      "grad_norm": 3.548877839583674,
      "learning_rate": 7.796465845563123e-07,
      "loss": 0.9676,
      "num_input_tokens_seen": 126621115,
      "step": 5968
    },
    {
      "epoch": 0.7177298142247339,
      "flos": 18210497372160.0,
      "grad_norm": 3.7661449025984335,
      "learning_rate": 7.790295248901766e-07,
      "loss": 0.9879,
      "num_input_tokens_seen": 126641965,
      "step": 5969
    },
    {
      "epoch": 0.7178500571153731,
      "flos": 22567137361920.0,
      "grad_norm": 2.877770788739856,
      "learning_rate": 7.784126504442902e-07,
      "loss": 0.8114,
      "num_input_tokens_seen": 126664915,
      "step": 5970
    },
    {
      "epoch": 0.7179703000060121,
      "flos": 13780497776640.0,
      "grad_norm": 2.1858961492721543,
      "learning_rate": 7.777959613122351e-07,
      "loss": 0.8652,
      "num_input_tokens_seen": 126684460,
      "step": 5971
    },
    {
      "epoch": 0.7180905428966512,
      "flos": 20544550195200.0,
      "grad_norm": 2.688813975063872,
      "learning_rate": 7.771794575875604e-07,
      "loss": 0.9628,
      "num_input_tokens_seen": 126706050,
      "step": 5972
    },
    {
      "epoch": 0.7182107857872904,
      "flos": 14226319134720.0,
      "grad_norm": 3.8161023740331657,
      "learning_rate": 7.765631393637888e-07,
      "loss": 0.9571,
      "num_input_tokens_seen": 126723965,
      "step": 5973
    },
    {
      "epoch": 0.7183310286779294,
      "flos": 16166870323200.0,
      "grad_norm": 4.640983387886734,
      "learning_rate": 7.75947006734417e-07,
      "loss": 0.6633,
      "num_input_tokens_seen": 126741465,
      "step": 5974
    },
    {
      "epoch": 0.7184512715685685,
      "flos": 12148898242560.0,
      "grad_norm": 4.8532544808392375,
      "learning_rate": 7.753310597929101e-07,
      "loss": 1.0283,
      "num_input_tokens_seen": 126757825,
      "step": 5975
    },
    {
      "epoch": 0.7185715144592076,
      "flos": 46904062279680.0,
      "grad_norm": 0.7773125416994218,
      "learning_rate": 7.747152986327095e-07,
      "loss": 0.7677,
      "num_input_tokens_seen": 126818090,
      "step": 5976
    },
    {
      "epoch": 0.7186917573498467,
      "flos": 11447211417600.0,
      "grad_norm": 3.816452728894302,
      "learning_rate": 7.740997233472228e-07,
      "loss": 0.867,
      "num_input_tokens_seen": 126835430,
      "step": 5977
    },
    {
      "epoch": 0.7188120002404857,
      "flos": 20834424115200.0,
      "grad_norm": 3.869910312372325,
      "learning_rate": 7.734843340298329e-07,
      "loss": 0.895,
      "num_input_tokens_seen": 126854975,
      "step": 5978
    },
    {
      "epoch": 0.7189322431311249,
      "flos": 23823270604800.0,
      "grad_norm": 3.755392012775324,
      "learning_rate": 7.72869130773895e-07,
      "loss": 0.9393,
      "num_input_tokens_seen": 126875295,
      "step": 5979
    },
    {
      "epoch": 0.719052486021764,
      "flos": 43915675668480.0,
      "grad_norm": 0.7935695157647916,
      "learning_rate": 7.722541136727343e-07,
      "loss": 0.8031,
      "num_input_tokens_seen": 126931030,
      "step": 5980
    },
    {
      "epoch": 0.719172728912403,
      "flos": 11027703214080.0,
      "grad_norm": 3.9044038606214824,
      "learning_rate": 7.716392828196483e-07,
      "loss": 0.9957,
      "num_input_tokens_seen": 126948550,
      "step": 5981
    },
    {
      "epoch": 0.7192929718030422,
      "flos": 10995380981760.0,
      "grad_norm": 5.3420857297358735,
      "learning_rate": 7.710246383079064e-07,
      "loss": 0.9556,
      "num_input_tokens_seen": 126963655,
      "step": 5982
    },
    {
      "epoch": 0.7194132146936812,
      "flos": 15529406853120.0,
      "grad_norm": 5.790734671199032,
      "learning_rate": 7.704101802307492e-07,
      "loss": 1.107,
      "num_input_tokens_seen": 126975675,
      "step": 5983
    },
    {
      "epoch": 0.7195334575843203,
      "flos": 19465680199680.0,
      "grad_norm": 6.163623093099469,
      "learning_rate": 7.697959086813912e-07,
      "loss": 1.0647,
      "num_input_tokens_seen": 126991560,
      "step": 5984
    },
    {
      "epoch": 0.7196537004749595,
      "flos": 13308608532480.0,
      "grad_norm": 5.068857980652776,
      "learning_rate": 7.691818237530145e-07,
      "loss": 0.9961,
      "num_input_tokens_seen": 127010140,
      "step": 5985
    },
    {
      "epoch": 0.7197739433655985,
      "flos": 17447906426880.0,
      "grad_norm": 3.0791532861180557,
      "learning_rate": 7.685679255387774e-07,
      "loss": 0.9697,
      "num_input_tokens_seen": 127028175,
      "step": 5986
    },
    {
      "epoch": 0.7198941862562376,
      "flos": 12783909027840.0,
      "grad_norm": 3.367299513934056,
      "learning_rate": 7.679542141318065e-07,
      "loss": 0.9553,
      "num_input_tokens_seen": 127045000,
      "step": 5987
    },
    {
      "epoch": 0.7200144291468767,
      "flos": 20676269260800.0,
      "grad_norm": 3.287866515408515,
      "learning_rate": 7.673406896252013e-07,
      "loss": 0.9579,
      "num_input_tokens_seen": 127066095,
      "step": 5988
    },
    {
      "epoch": 0.7201346720375158,
      "flos": 18054488616960.0,
      "grad_norm": 2.598090618787928,
      "learning_rate": 7.667273521120347e-07,
      "loss": 0.9745,
      "num_input_tokens_seen": 127085375,
      "step": 5989
    },
    {
      "epoch": 0.7202549149281549,
      "flos": 10135600619520.0,
      "grad_norm": 3.4255596511139377,
      "learning_rate": 7.661142016853468e-07,
      "loss": 0.9956,
      "num_input_tokens_seen": 127102455,
      "step": 5990
    },
    {
      "epoch": 0.7203751578187939,
      "flos": 16349222891520.0,
      "grad_norm": 2.8293376872789904,
      "learning_rate": 7.655012384381543e-07,
      "loss": 0.9406,
      "num_input_tokens_seen": 127121660,
      "step": 5991
    },
    {
      "epoch": 0.7204954007094331,
      "flos": 16845309849600.0,
      "grad_norm": 4.6295949431638785,
      "learning_rate": 7.648884624634415e-07,
      "loss": 1.005,
      "num_input_tokens_seen": 127139930,
      "step": 5992
    },
    {
      "epoch": 0.7206156436000721,
      "flos": 11499255214080.0,
      "grad_norm": 3.5736586501890875,
      "learning_rate": 7.642758738541683e-07,
      "loss": 1.0818,
      "num_input_tokens_seen": 127156230,
      "step": 5993
    },
    {
      "epoch": 0.7207358864907112,
      "flos": 38903352913920.0,
      "grad_norm": 0.7748116895101077,
      "learning_rate": 7.636634727032621e-07,
      "loss": 0.8111,
      "num_input_tokens_seen": 127213055,
      "step": 5994
    },
    {
      "epoch": 0.7208561293813504,
      "flos": 13570881638400.0,
      "grad_norm": 4.097829169596781,
      "learning_rate": 7.630512591036231e-07,
      "loss": 0.989,
      "num_input_tokens_seen": 127232085,
      "step": 5995
    },
    {
      "epoch": 0.7209763722719894,
      "flos": 12573986304000.0,
      "grad_norm": 6.1369844861107214,
      "learning_rate": 7.624392331481255e-07,
      "loss": 0.8401,
      "num_input_tokens_seen": 127249460,
      "step": 5996
    },
    {
      "epoch": 0.7210966151626285,
      "flos": 47844958863360.0,
      "grad_norm": 0.7612524708424324,
      "learning_rate": 7.618273949296115e-07,
      "loss": 0.7326,
      "num_input_tokens_seen": 127308690,
      "step": 5997
    },
    {
      "epoch": 0.7212168580532676,
      "flos": 15012739891200.0,
      "grad_norm": 3.652632268433728,
      "learning_rate": 7.612157445408987e-07,
      "loss": 0.8827,
      "num_input_tokens_seen": 127326220,
      "step": 5998
    },
    {
      "epoch": 0.7213371009439067,
      "flos": 15877916160000.0,
      "grad_norm": 4.024392379483871,
      "learning_rate": 7.606042820747716e-07,
      "loss": 0.9318,
      "num_input_tokens_seen": 127342345,
      "step": 5999
    },
    {
      "epoch": 0.7214573438345457,
      "flos": 13125642792960.0,
      "grad_norm": 4.483152641616414,
      "learning_rate": 7.599930076239889e-07,
      "loss": 1.0433,
      "num_input_tokens_seen": 127359350,
      "step": 6000
    },
    {
      "epoch": 0.7215775867251849,
      "flos": 25501334077440.0,
      "grad_norm": 3.522138064154889,
      "learning_rate": 7.593819212812818e-07,
      "loss": 0.9016,
      "num_input_tokens_seen": 127380650,
      "step": 6001
    },
    {
      "epoch": 0.721697829615824,
      "flos": 14459305205760.0,
      "grad_norm": 4.242883221251182,
      "learning_rate": 7.587710231393508e-07,
      "loss": 0.9152,
      "num_input_tokens_seen": 127398725,
      "step": 6002
    },
    {
      "epoch": 0.721818072506463,
      "flos": 14357271736320.0,
      "grad_norm": 4.019309880247124,
      "learning_rate": 7.581603132908685e-07,
      "loss": 1.0224,
      "num_input_tokens_seen": 127416415,
      "step": 6003
    },
    {
      "epoch": 0.7219383153971022,
      "flos": 12889100328960.0,
      "grad_norm": 3.5653337953158726,
      "learning_rate": 7.575497918284795e-07,
      "loss": 0.9746,
      "num_input_tokens_seen": 127433680,
      "step": 6004
    },
    {
      "epoch": 0.7220585582877412,
      "flos": 12281230479360.0,
      "grad_norm": 8.319434922909334,
      "learning_rate": 7.569394588447984e-07,
      "loss": 0.9413,
      "num_input_tokens_seen": 127450415,
      "step": 6005
    },
    {
      "epoch": 0.7221788011783803,
      "flos": 12018221568000.0,
      "grad_norm": 4.691330832098317,
      "learning_rate": 7.563293144324146e-07,
      "loss": 0.97,
      "num_input_tokens_seen": 127465685,
      "step": 6006
    },
    {
      "epoch": 0.7222990440690195,
      "flos": 18709987430400.0,
      "grad_norm": 4.208424457029312,
      "learning_rate": 7.557193586838834e-07,
      "loss": 0.9967,
      "num_input_tokens_seen": 127480770,
      "step": 6007
    },
    {
      "epoch": 0.7224192869596585,
      "flos": 12468335124480.0,
      "grad_norm": 3.13417871988324,
      "learning_rate": 7.551095916917371e-07,
      "loss": 0.8909,
      "num_input_tokens_seen": 127497820,
      "step": 6008
    },
    {
      "epoch": 0.7225395298502976,
      "flos": 9112668057600.0,
      "grad_norm": 4.083862711527709,
      "learning_rate": 7.545000135484758e-07,
      "loss": 0.8495,
      "num_input_tokens_seen": 127514975,
      "step": 6009
    },
    {
      "epoch": 0.7226597727409367,
      "flos": 21123102351360.0,
      "grad_norm": 3.859297655079604,
      "learning_rate": 7.538906243465714e-07,
      "loss": 0.8268,
      "num_input_tokens_seen": 127534830,
      "step": 6010
    },
    {
      "epoch": 0.7227800156315758,
      "flos": 9715877806080.0,
      "grad_norm": 3.5201087672595293,
      "learning_rate": 7.5328142417847e-07,
      "loss": 0.9755,
      "num_input_tokens_seen": 127551315,
      "step": 6011
    },
    {
      "epoch": 0.7229002585222148,
      "flos": 14408947630080.0,
      "grad_norm": 9.407913415782863,
      "learning_rate": 7.526724131365838e-07,
      "loss": 0.8853,
      "num_input_tokens_seen": 127571990,
      "step": 6012
    },
    {
      "epoch": 0.723020501412854,
      "flos": 11735123189760.0,
      "grad_norm": 9.267606732371547,
      "learning_rate": 7.520635913133017e-07,
      "loss": 0.8917,
      "num_input_tokens_seen": 127590340,
      "step": 6013
    },
    {
      "epoch": 0.7231407443034931,
      "flos": 20335577886720.0,
      "grad_norm": 3.397130956192373,
      "learning_rate": 7.514549588009798e-07,
      "loss": 1.0136,
      "num_input_tokens_seen": 127610935,
      "step": 6014
    },
    {
      "epoch": 0.7232609871941321,
      "flos": 21385130188800.0,
      "grad_norm": 4.683213779285571,
      "learning_rate": 7.508465156919492e-07,
      "loss": 0.9093,
      "num_input_tokens_seen": 127634165,
      "step": 6015
    },
    {
      "epoch": 0.7233812300847713,
      "flos": 11787381596160.0,
      "grad_norm": 4.46761494339773,
      "learning_rate": 7.502382620785083e-07,
      "loss": 0.8,
      "num_input_tokens_seen": 127650435,
      "step": 6016
    },
    {
      "epoch": 0.7235014729754103,
      "flos": 48160870010880.0,
      "grad_norm": 0.8940895022621728,
      "learning_rate": 7.496301980529289e-07,
      "loss": 0.8854,
      "num_input_tokens_seen": 127713365,
      "step": 6017
    },
    {
      "epoch": 0.7236217158660494,
      "flos": 19182091284480.0,
      "grad_norm": 4.662022436807222,
      "learning_rate": 7.490223237074547e-07,
      "loss": 0.9357,
      "num_input_tokens_seen": 127732795,
      "step": 6018
    },
    {
      "epoch": 0.7237419587566886,
      "flos": 20964395642880.0,
      "grad_norm": 3.768427142584461,
      "learning_rate": 7.484146391342989e-07,
      "loss": 0.8529,
      "num_input_tokens_seen": 127752310,
      "step": 6019
    },
    {
      "epoch": 0.7238622016473276,
      "flos": 12626428661760.0,
      "grad_norm": 4.366227209870986,
      "learning_rate": 7.478071444256484e-07,
      "loss": 0.7537,
      "num_input_tokens_seen": 127769790,
      "step": 6020
    },
    {
      "epoch": 0.7239824445379667,
      "flos": 18316700405760.0,
      "grad_norm": 4.962961724082408,
      "learning_rate": 7.471998396736579e-07,
      "loss": 0.9689,
      "num_input_tokens_seen": 127789890,
      "step": 6021
    },
    {
      "epoch": 0.7241026874286057,
      "flos": 16455211315200.0,
      "grad_norm": 3.2422005894408215,
      "learning_rate": 7.465927249704549e-07,
      "loss": 0.9482,
      "num_input_tokens_seen": 127807495,
      "step": 6022
    },
    {
      "epoch": 0.7242229303192449,
      "flos": 14384014110720.0,
      "grad_norm": 3.540632628039554,
      "learning_rate": 7.459858004081398e-07,
      "loss": 0.9627,
      "num_input_tokens_seen": 127825185,
      "step": 6023
    },
    {
      "epoch": 0.724343173209884,
      "flos": 44607482327040.0,
      "grad_norm": 0.6674890189749897,
      "learning_rate": 7.453790660787815e-07,
      "loss": 0.7928,
      "num_input_tokens_seen": 127893000,
      "step": 6024
    },
    {
      "epoch": 0.724463416100523,
      "flos": 24976910499840.0,
      "grad_norm": 2.9037133087848495,
      "learning_rate": 7.447725220744214e-07,
      "loss": 0.8317,
      "num_input_tokens_seen": 127914965,
      "step": 6025
    },
    {
      "epoch": 0.7245836589911622,
      "flos": 15537500712960.0,
      "grad_norm": 3.4010103856157814,
      "learning_rate": 7.441661684870717e-07,
      "loss": 0.9474,
      "num_input_tokens_seen": 127934940,
      "step": 6026
    },
    {
      "epoch": 0.7247039018818012,
      "flos": 16353331138560.0,
      "grad_norm": 2.8979616828528294,
      "learning_rate": 7.435600054087152e-07,
      "loss": 1.0089,
      "num_input_tokens_seen": 127956825,
      "step": 6027
    },
    {
      "epoch": 0.7248241447724403,
      "flos": 22616881766400.0,
      "grad_norm": 7.029908997555801,
      "learning_rate": 7.42954032931308e-07,
      "loss": 0.9409,
      "num_input_tokens_seen": 127977585,
      "step": 6028
    },
    {
      "epoch": 0.7249443876630794,
      "flos": 24898277621760.0,
      "grad_norm": 4.383257457429389,
      "learning_rate": 7.423482511467733e-07,
      "loss": 0.9404,
      "num_input_tokens_seen": 127998075,
      "step": 6029
    },
    {
      "epoch": 0.7250646305537185,
      "flos": 18762399129600.0,
      "grad_norm": 3.451291801644372,
      "learning_rate": 7.417426601470099e-07,
      "loss": 0.8502,
      "num_input_tokens_seen": 128018155,
      "step": 6030
    },
    {
      "epoch": 0.7251848734443576,
      "flos": 21437664522240.0,
      "grad_norm": 5.420384270816097,
      "learning_rate": 7.411372600238841e-07,
      "loss": 0.9688,
      "num_input_tokens_seen": 128038490,
      "step": 6031
    },
    {
      "epoch": 0.7253051163349967,
      "flos": 12600422092800.0,
      "grad_norm": 3.793107789434741,
      "learning_rate": 7.405320508692346e-07,
      "loss": 0.9199,
      "num_input_tokens_seen": 128056950,
      "step": 6032
    },
    {
      "epoch": 0.7254253592256358,
      "flos": 8903665090560.0,
      "grad_norm": 3.237726511450845,
      "learning_rate": 7.399270327748727e-07,
      "loss": 0.9317,
      "num_input_tokens_seen": 128074330,
      "step": 6033
    },
    {
      "epoch": 0.7255456021162748,
      "flos": 19366130073600.0,
      "grad_norm": 8.94051112956147,
      "learning_rate": 7.39322205832577e-07,
      "loss": 0.9342,
      "num_input_tokens_seen": 128094940,
      "step": 6034
    },
    {
      "epoch": 0.725665845006914,
      "flos": 15117961850880.0,
      "grad_norm": 2.660674774227663,
      "learning_rate": 7.387175701341009e-07,
      "loss": 0.9981,
      "num_input_tokens_seen": 128113330,
      "step": 6035
    },
    {
      "epoch": 0.7257860878975531,
      "flos": 11341560238080.0,
      "grad_norm": 5.1988150126745944,
      "learning_rate": 7.381131257711659e-07,
      "loss": 0.9118,
      "num_input_tokens_seen": 128130155,
      "step": 6036
    },
    {
      "epoch": 0.7259063307881921,
      "flos": 8536047390720.0,
      "grad_norm": 2.90887218026936,
      "learning_rate": 7.375088728354677e-07,
      "loss": 1.0306,
      "num_input_tokens_seen": 128144905,
      "step": 6037
    },
    {
      "epoch": 0.7260265736788313,
      "flos": 21697638236160.0,
      "grad_norm": 2.7838730176543742,
      "learning_rate": 7.369048114186691e-07,
      "loss": 0.8563,
      "num_input_tokens_seen": 128165670,
      "step": 6038
    },
    {
      "epoch": 0.7261468165694703,
      "flos": 15013077135360.0,
      "grad_norm": 2.614402541841511,
      "learning_rate": 7.363009416124055e-07,
      "loss": 1.029,
      "num_input_tokens_seen": 128184715,
      "step": 6039
    },
    {
      "epoch": 0.7262670594601094,
      "flos": 15849180979200.0,
      "grad_norm": 10.954485625078824,
      "learning_rate": 7.356972635082852e-07,
      "loss": 0.8218,
      "num_input_tokens_seen": 128203290,
      "step": 6040
    },
    {
      "epoch": 0.7263873023507486,
      "flos": 18025937387520.0,
      "grad_norm": 4.598068802181491,
      "learning_rate": 7.35093777197884e-07,
      "loss": 0.9487,
      "num_input_tokens_seen": 128223080,
      "step": 6041
    },
    {
      "epoch": 0.7265075452413876,
      "flos": 16979634892800.0,
      "grad_norm": 4.355740909027665,
      "learning_rate": 7.344904827727525e-07,
      "loss": 1.0426,
      "num_input_tokens_seen": 128239980,
      "step": 6042
    },
    {
      "epoch": 0.7266277881320267,
      "flos": 20461410508800.0,
      "grad_norm": 4.505508018919167,
      "learning_rate": 7.338873803244076e-07,
      "loss": 0.9035,
      "num_input_tokens_seen": 128254935,
      "step": 6043
    },
    {
      "epoch": 0.7267480310226658,
      "flos": 17687422771200.0,
      "grad_norm": 3.2138251988957984,
      "learning_rate": 7.332844699443401e-07,
      "loss": 0.9892,
      "num_input_tokens_seen": 128273255,
      "step": 6044
    },
    {
      "epoch": 0.7268682739133049,
      "flos": 19365516902400.0,
      "grad_norm": 4.0272926009118555,
      "learning_rate": 7.326817517240121e-07,
      "loss": 0.9434,
      "num_input_tokens_seen": 128294680,
      "step": 6045
    },
    {
      "epoch": 0.7269885168039439,
      "flos": 23899818700800.0,
      "grad_norm": 4.468813595619376,
      "learning_rate": 7.320792257548545e-07,
      "loss": 1.0243,
      "num_input_tokens_seen": 128315575,
      "step": 6046
    },
    {
      "epoch": 0.7271087596945831,
      "flos": 17291989647360.0,
      "grad_norm": 4.001955269719411,
      "learning_rate": 7.314768921282704e-07,
      "loss": 0.9505,
      "num_input_tokens_seen": 128335950,
      "step": 6047
    },
    {
      "epoch": 0.7272290025852222,
      "flos": 16926763315200.0,
      "grad_norm": 5.003854029172281,
      "learning_rate": 7.30874750935633e-07,
      "loss": 0.9075,
      "num_input_tokens_seen": 128355355,
      "step": 6048
    },
    {
      "epoch": 0.7273492454758612,
      "flos": 11834182778880.0,
      "grad_norm": 19.135154671104946,
      "learning_rate": 7.30272802268286e-07,
      "loss": 0.9845,
      "num_input_tokens_seen": 128372070,
      "step": 6049
    },
    {
      "epoch": 0.7274694883665004,
      "flos": 19963208110080.0,
      "grad_norm": 2.622804541055909,
      "learning_rate": 7.29671046217547e-07,
      "loss": 0.9561,
      "num_input_tokens_seen": 128390900,
      "step": 6050
    },
    {
      "epoch": 0.7275897312571394,
      "flos": 21646759464960.0,
      "grad_norm": 4.088012401883001,
      "learning_rate": 7.290694828746988e-07,
      "loss": 1.0099,
      "num_input_tokens_seen": 128410285,
      "step": 6051
    },
    {
      "epoch": 0.7277099741477785,
      "flos": 13620135505920.0,
      "grad_norm": 3.1168951600170316,
      "learning_rate": 7.284681123310004e-07,
      "loss": 1.0526,
      "num_input_tokens_seen": 128428720,
      "step": 6052
    },
    {
      "epoch": 0.7278302170384175,
      "flos": 14672140492800.0,
      "grad_norm": 2.4591803196168573,
      "learning_rate": 7.27866934677678e-07,
      "loss": 0.9818,
      "num_input_tokens_seen": 128448110,
      "step": 6053
    },
    {
      "epoch": 0.7279504599290567,
      "flos": 13539663114240.0,
      "grad_norm": 2.4960452270567326,
      "learning_rate": 7.272659500059297e-07,
      "loss": 0.9652,
      "num_input_tokens_seen": 128465170,
      "step": 6054
    },
    {
      "epoch": 0.7280707028196958,
      "flos": 13518102036480.0,
      "grad_norm": 4.8140115717268195,
      "learning_rate": 7.266651584069264e-07,
      "loss": 0.9985,
      "num_input_tokens_seen": 128482555,
      "step": 6055
    },
    {
      "epoch": 0.7281909457103348,
      "flos": 26550855720960.0,
      "grad_norm": 2.8556467566994086,
      "learning_rate": 7.260645599718045e-07,
      "loss": 0.769,
      "num_input_tokens_seen": 128508630,
      "step": 6056
    },
    {
      "epoch": 0.728311188600974,
      "flos": 14671466004480.0,
      "grad_norm": 4.267737679368584,
      "learning_rate": 7.254641547916767e-07,
      "loss": 0.8617,
      "num_input_tokens_seen": 128525845,
      "step": 6057
    },
    {
      "epoch": 0.728431431491613,
      "flos": 20545776537600.0,
      "grad_norm": 3.0417526616172403,
      "learning_rate": 7.248639429576226e-07,
      "loss": 0.8802,
      "num_input_tokens_seen": 128545020,
      "step": 6058
    },
    {
      "epoch": 0.7285516743822521,
      "flos": 18497979924480.0,
      "grad_norm": 3.288508642685553,
      "learning_rate": 7.242639245606959e-07,
      "loss": 0.9155,
      "num_input_tokens_seen": 128564530,
      "step": 6059
    },
    {
      "epoch": 0.7286719172728913,
      "flos": 11604783759360.0,
      "grad_norm": 3.6617594884938454,
      "learning_rate": 7.236640996919168e-07,
      "loss": 1.016,
      "num_input_tokens_seen": 128583295,
      "step": 6060
    },
    {
      "epoch": 0.7287921601635303,
      "flos": 15642600038400.0,
      "grad_norm": 2.6791633115503197,
      "learning_rate": 7.230644684422782e-07,
      "loss": 0.8949,
      "num_input_tokens_seen": 128603245,
      "step": 6061
    },
    {
      "epoch": 0.7289124030541694,
      "flos": 17497742807040.0,
      "grad_norm": 3.1402291575109818,
      "learning_rate": 7.224650309027451e-07,
      "loss": 1.0138,
      "num_input_tokens_seen": 128622715,
      "step": 6062
    },
    {
      "epoch": 0.7290326459448085,
      "flos": 15193191628800.0,
      "grad_norm": 4.353470498728662,
      "learning_rate": 7.218657871642506e-07,
      "loss": 0.8773,
      "num_input_tokens_seen": 128641240,
      "step": 6063
    },
    {
      "epoch": 0.7291528888354476,
      "flos": 13177073418240.0,
      "grad_norm": 3.476448794444979,
      "learning_rate": 7.212667373177012e-07,
      "loss": 0.807,
      "num_input_tokens_seen": 128655955,
      "step": 6064
    },
    {
      "epoch": 0.7292731317260867,
      "flos": 13438028206080.0,
      "grad_norm": 3.5284336978823285,
      "learning_rate": 7.206678814539704e-07,
      "loss": 0.9454,
      "num_input_tokens_seen": 128673975,
      "step": 6065
    },
    {
      "epoch": 0.7293933746167258,
      "flos": 14959163166720.0,
      "grad_norm": 3.0971944111106176,
      "learning_rate": 7.20069219663904e-07,
      "loss": 0.9266,
      "num_input_tokens_seen": 128693580,
      "step": 6066
    },
    {
      "epoch": 0.7295136175073649,
      "flos": 15954985451520.0,
      "grad_norm": 2.982567364554646,
      "learning_rate": 7.1947075203832e-07,
      "loss": 0.9762,
      "num_input_tokens_seen": 128713280,
      "step": 6067
    },
    {
      "epoch": 0.7296338603980039,
      "flos": 40158903644160.0,
      "grad_norm": 0.8667787843099707,
      "learning_rate": 7.188724786680049e-07,
      "loss": 0.8046,
      "num_input_tokens_seen": 128773470,
      "step": 6068
    },
    {
      "epoch": 0.7297541032886431,
      "flos": 17949573242880.0,
      "grad_norm": 2.490014734492694,
      "learning_rate": 7.182743996437162e-07,
      "loss": 0.9398,
      "num_input_tokens_seen": 128792725,
      "step": 6069
    },
    {
      "epoch": 0.7298743461792822,
      "flos": 18840081592320.0,
      "grad_norm": 6.617474005935972,
      "learning_rate": 7.176765150561819e-07,
      "loss": 0.8768,
      "num_input_tokens_seen": 128811050,
      "step": 6070
    },
    {
      "epoch": 0.7299945890699212,
      "flos": 13881856757760.0,
      "grad_norm": 3.5547659729061416,
      "learning_rate": 7.170788249961002e-07,
      "loss": 0.9801,
      "num_input_tokens_seen": 128829280,
      "step": 6071
    },
    {
      "epoch": 0.7301148319605604,
      "flos": 16296995143680.0,
      "grad_norm": 3.3771654881896356,
      "learning_rate": 7.164813295541418e-07,
      "loss": 1.0658,
      "num_input_tokens_seen": 128848565,
      "step": 6072
    },
    {
      "epoch": 0.7302350748511994,
      "flos": 18050594979840.0,
      "grad_norm": 3.102510936634789,
      "learning_rate": 7.15884028820944e-07,
      "loss": 0.8846,
      "num_input_tokens_seen": 128867340,
      "step": 6073
    },
    {
      "epoch": 0.7303553177418385,
      "flos": 19811644846080.0,
      "grad_norm": 3.817483865020569,
      "learning_rate": 7.152869228871185e-07,
      "loss": 0.7757,
      "num_input_tokens_seen": 128889545,
      "step": 6074
    },
    {
      "epoch": 0.7304755606324776,
      "flos": 17373473771520.0,
      "grad_norm": 2.911672476238823,
      "learning_rate": 7.146900118432457e-07,
      "loss": 0.9082,
      "num_input_tokens_seen": 128909010,
      "step": 6075
    },
    {
      "epoch": 0.7305958035231167,
      "flos": 16952003420160.0,
      "grad_norm": 4.331637833310828,
      "learning_rate": 7.140932957798753e-07,
      "loss": 1.0514,
      "num_input_tokens_seen": 128927170,
      "step": 6076
    },
    {
      "epoch": 0.7307160464137558,
      "flos": 11839793295360.0,
      "grad_norm": 4.920692907402863,
      "learning_rate": 7.134967747875309e-07,
      "loss": 0.8977,
      "num_input_tokens_seen": 128945100,
      "step": 6077
    },
    {
      "epoch": 0.7308362893043949,
      "flos": 15484445184000.0,
      "grad_norm": 3.506450431113278,
      "learning_rate": 7.129004489567014e-07,
      "loss": 1.0112,
      "num_input_tokens_seen": 128962300,
      "step": 6078
    },
    {
      "epoch": 0.730956532195034,
      "flos": 7513728000000.0,
      "grad_norm": 3.706281515826954,
      "learning_rate": 7.123043183778512e-07,
      "loss": 0.9774,
      "num_input_tokens_seen": 128979350,
      "step": 6079
    },
    {
      "epoch": 0.731076775085673,
      "flos": 14042617589760.0,
      "grad_norm": 2.901283203881685,
      "learning_rate": 7.117083831414114e-07,
      "loss": 0.8425,
      "num_input_tokens_seen": 128998345,
      "step": 6080
    },
    {
      "epoch": 0.7311970179763122,
      "flos": 14513157857280.0,
      "grad_norm": 4.3878434353823215,
      "learning_rate": 7.11112643337787e-07,
      "loss": 0.8938,
      "num_input_tokens_seen": 129017110,
      "step": 6081
    },
    {
      "epoch": 0.7313172608669513,
      "flos": 13123864596480.0,
      "grad_norm": 4.560617554992981,
      "learning_rate": 7.10517099057349e-07,
      "loss": 0.9659,
      "num_input_tokens_seen": 129033780,
      "step": 6082
    },
    {
      "epoch": 0.7314375037575903,
      "flos": 11447395368960.0,
      "grad_norm": 4.725201775011856,
      "learning_rate": 7.099217503904411e-07,
      "loss": 0.7806,
      "num_input_tokens_seen": 129051355,
      "step": 6083
    },
    {
      "epoch": 0.7315577466482295,
      "flos": 12731742597120.0,
      "grad_norm": 3.399177558384571,
      "learning_rate": 7.093265974273788e-07,
      "loss": 1.0967,
      "num_input_tokens_seen": 129068970,
      "step": 6084
    },
    {
      "epoch": 0.7316779895388685,
      "flos": 13046120816640.0,
      "grad_norm": 4.775971734198118,
      "learning_rate": 7.087316402584447e-07,
      "loss": 0.9168,
      "num_input_tokens_seen": 129087515,
      "step": 6085
    },
    {
      "epoch": 0.7317982324295076,
      "flos": 12702823464960.0,
      "grad_norm": 3.491911533725159,
      "learning_rate": 7.081368789738953e-07,
      "loss": 1.0462,
      "num_input_tokens_seen": 129104435,
      "step": 6086
    },
    {
      "epoch": 0.7319184753201466,
      "flos": 19387353907200.0,
      "grad_norm": 2.9724706524169386,
      "learning_rate": 7.075423136639537e-07,
      "loss": 0.9718,
      "num_input_tokens_seen": 129123410,
      "step": 6087
    },
    {
      "epoch": 0.7320387182107858,
      "flos": 26731706019840.0,
      "grad_norm": 4.535438022846529,
      "learning_rate": 7.069479444188149e-07,
      "loss": 0.9314,
      "num_input_tokens_seen": 129143720,
      "step": 6088
    },
    {
      "epoch": 0.7321589611014249,
      "flos": 12653538938880.0,
      "grad_norm": 4.014126627978896,
      "learning_rate": 7.063537713286453e-07,
      "loss": 1.0135,
      "num_input_tokens_seen": 129161120,
      "step": 6089
    },
    {
      "epoch": 0.7322792039920639,
      "flos": 18576030289920.0,
      "grad_norm": 7.569307652286366,
      "learning_rate": 7.057597944835803e-07,
      "loss": 1.0028,
      "num_input_tokens_seen": 129180115,
      "step": 6090
    },
    {
      "epoch": 0.7323994468827031,
      "flos": 18050073784320.0,
      "grad_norm": 2.596942453621542,
      "learning_rate": 7.051660139737253e-07,
      "loss": 0.9397,
      "num_input_tokens_seen": 129198055,
      "step": 6091
    },
    {
      "epoch": 0.7325196897733421,
      "flos": 19156146032640.0,
      "grad_norm": 5.233517522466031,
      "learning_rate": 7.045724298891565e-07,
      "loss": 0.9533,
      "num_input_tokens_seen": 129217245,
      "step": 6092
    },
    {
      "epoch": 0.7326399326639812,
      "flos": 18343074877440.0,
      "grad_norm": 4.831609067614119,
      "learning_rate": 7.039790423199192e-07,
      "loss": 0.8838,
      "num_input_tokens_seen": 129236605,
      "step": 6093
    },
    {
      "epoch": 0.7327601755546204,
      "flos": 14933953720320.0,
      "grad_norm": 4.365679594885006,
      "learning_rate": 7.033858513560322e-07,
      "loss": 0.9644,
      "num_input_tokens_seen": 129252620,
      "step": 6094
    },
    {
      "epoch": 0.7328804184452594,
      "flos": 11525231124480.0,
      "grad_norm": 3.0905745422132376,
      "learning_rate": 7.027928570874794e-07,
      "loss": 0.9507,
      "num_input_tokens_seen": 129270530,
      "step": 6095
    },
    {
      "epoch": 0.7330006613358985,
      "flos": 12652864450560.0,
      "grad_norm": 5.111500895651883,
      "learning_rate": 7.022000596042194e-07,
      "loss": 1.0402,
      "num_input_tokens_seen": 129287350,
      "step": 6096
    },
    {
      "epoch": 0.7331209042265376,
      "flos": 15983138119680.0,
      "grad_norm": 3.1483575323281165,
      "learning_rate": 7.016074589961784e-07,
      "loss": 1.0025,
      "num_input_tokens_seen": 129305635,
      "step": 6097
    },
    {
      "epoch": 0.7332411471171767,
      "flos": 23587249336320.0,
      "grad_norm": 2.9088970967775936,
      "learning_rate": 7.01015055353253e-07,
      "loss": 0.862,
      "num_input_tokens_seen": 129327780,
      "step": 6098
    },
    {
      "epoch": 0.7333613900078157,
      "flos": 16162884710400.0,
      "grad_norm": 3.412274833209294,
      "learning_rate": 7.004228487653123e-07,
      "loss": 0.9672,
      "num_input_tokens_seen": 129348305,
      "step": 6099
    },
    {
      "epoch": 0.7334816328984549,
      "flos": 15878253404160.0,
      "grad_norm": 3.3503566609017947,
      "learning_rate": 6.998308393221906e-07,
      "loss": 0.9737,
      "num_input_tokens_seen": 129366430,
      "step": 6100
    },
    {
      "epoch": 0.733601875789094,
      "flos": 14720566579200.0,
      "grad_norm": 4.725187041602163,
      "learning_rate": 6.992390271136977e-07,
      "loss": 0.9109,
      "num_input_tokens_seen": 129381860,
      "step": 6101
    },
    {
      "epoch": 0.733722118679733,
      "flos": 16034997964800.0,
      "grad_norm": 5.1992710262797335,
      "learning_rate": 6.986474122296094e-07,
      "loss": 1.0506,
      "num_input_tokens_seen": 129400695,
      "step": 6102
    },
    {
      "epoch": 0.7338423615703722,
      "flos": 14252662947840.0,
      "grad_norm": 3.1067724022724104,
      "learning_rate": 6.980559947596751e-07,
      "loss": 0.9176,
      "num_input_tokens_seen": 129418955,
      "step": 6103
    },
    {
      "epoch": 0.7339626044610112,
      "flos": 15404800573440.0,
      "grad_norm": 8.03156395680724,
      "learning_rate": 6.974647747936109e-07,
      "loss": 0.9521,
      "num_input_tokens_seen": 129437060,
      "step": 6104
    },
    {
      "epoch": 0.7340828473516503,
      "flos": 10791559311360.0,
      "grad_norm": 3.9641837937890747,
      "learning_rate": 6.968737524211039e-07,
      "loss": 1.0169,
      "num_input_tokens_seen": 129453590,
      "step": 6105
    },
    {
      "epoch": 0.7342030902422895,
      "flos": 16297976217600.0,
      "grad_norm": 4.7615561203257615,
      "learning_rate": 6.962829277318132e-07,
      "loss": 0.9774,
      "num_input_tokens_seen": 129472905,
      "step": 6106
    },
    {
      "epoch": 0.7343233331329285,
      "flos": 18394781429760.0,
      "grad_norm": 2.939855077895562,
      "learning_rate": 6.956923008153652e-07,
      "loss": 1.018,
      "num_input_tokens_seen": 129492390,
      "step": 6107
    },
    {
      "epoch": 0.7344435760235676,
      "flos": 13099053711360.0,
      "grad_norm": 4.281479432578393,
      "learning_rate": 6.951018717613593e-07,
      "loss": 1.0301,
      "num_input_tokens_seen": 129511125,
      "step": 6108
    },
    {
      "epoch": 0.7345638189142067,
      "flos": 12496150548480.0,
      "grad_norm": 2.9779389170165067,
      "learning_rate": 6.945116406593614e-07,
      "loss": 0.9599,
      "num_input_tokens_seen": 129529700,
      "step": 6109
    },
    {
      "epoch": 0.7346840618048458,
      "flos": 14378311618560.0,
      "grad_norm": 3.7950098299444495,
      "learning_rate": 6.939216075989089e-07,
      "loss": 0.9173,
      "num_input_tokens_seen": 129547350,
      "step": 6110
    },
    {
      "epoch": 0.7348043046954849,
      "flos": 20676943749120.0,
      "grad_norm": 3.323917505261993,
      "learning_rate": 6.933317726695109e-07,
      "loss": 0.8615,
      "num_input_tokens_seen": 129568300,
      "step": 6111
    },
    {
      "epoch": 0.734924547586124,
      "flos": 12705306808320.0,
      "grad_norm": 4.345074348823246,
      "learning_rate": 6.92742135960644e-07,
      "loss": 0.9751,
      "num_input_tokens_seen": 129585720,
      "step": 6112
    },
    {
      "epoch": 0.7350447904767631,
      "flos": 45524947660800.0,
      "grad_norm": 0.8795016836136171,
      "learning_rate": 6.921526975617556e-07,
      "loss": 0.7891,
      "num_input_tokens_seen": 129644900,
      "step": 6113
    },
    {
      "epoch": 0.7351650333674021,
      "flos": 15327669964800.0,
      "grad_norm": 4.211618003991798,
      "learning_rate": 6.915634575622631e-07,
      "loss": 0.9398,
      "num_input_tokens_seen": 129663135,
      "step": 6114
    },
    {
      "epoch": 0.7352852762580413,
      "flos": 12888640450560.0,
      "grad_norm": 5.122058489470387,
      "learning_rate": 6.909744160515532e-07,
      "loss": 0.9026,
      "num_input_tokens_seen": 129680995,
      "step": 6115
    },
    {
      "epoch": 0.7354055191486804,
      "flos": 27782362030080.0,
      "grad_norm": 3.6513756759312272,
      "learning_rate": 6.903855731189849e-07,
      "loss": 0.8645,
      "num_input_tokens_seen": 129703350,
      "step": 6116
    },
    {
      "epoch": 0.7355257620393194,
      "flos": 11525874954240.0,
      "grad_norm": 4.65281745299066,
      "learning_rate": 6.897969288538825e-07,
      "loss": 0.999,
      "num_input_tokens_seen": 129721015,
      "step": 6117
    },
    {
      "epoch": 0.7356460049299585,
      "flos": 12836933898240.0,
      "grad_norm": 3.586434390146662,
      "learning_rate": 6.892084833455452e-07,
      "loss": 1.0036,
      "num_input_tokens_seen": 129740305,
      "step": 6118
    },
    {
      "epoch": 0.7357662478205976,
      "flos": 15144612249600.0,
      "grad_norm": 2.861790205063154,
      "learning_rate": 6.886202366832384e-07,
      "loss": 1.0282,
      "num_input_tokens_seen": 129761710,
      "step": 6119
    },
    {
      "epoch": 0.7358864907112367,
      "flos": 10057519595520.0,
      "grad_norm": 3.9352076078952085,
      "learning_rate": 6.880321889561987e-07,
      "loss": 0.9241,
      "num_input_tokens_seen": 129779405,
      "step": 6120
    },
    {
      "epoch": 0.7360067336018757,
      "flos": 15852062883840.0,
      "grad_norm": 4.1129387896820075,
      "learning_rate": 6.874443402536338e-07,
      "loss": 0.836,
      "num_input_tokens_seen": 129798215,
      "step": 6121
    },
    {
      "epoch": 0.7361269764925149,
      "flos": 18183816314880.0,
      "grad_norm": 3.0208263704066214,
      "learning_rate": 6.868566906647177e-07,
      "loss": 1.0061,
      "num_input_tokens_seen": 129818885,
      "step": 6122
    },
    {
      "epoch": 0.736247219383154,
      "flos": 14462800281600.0,
      "grad_norm": 3.319490506017973,
      "learning_rate": 6.862692402785984e-07,
      "loss": 1.0209,
      "num_input_tokens_seen": 129838855,
      "step": 6123
    },
    {
      "epoch": 0.736367462273793,
      "flos": 50377092096000.0,
      "grad_norm": 0.7054874996426936,
      "learning_rate": 6.856819891843899e-07,
      "loss": 0.7052,
      "num_input_tokens_seen": 129903280,
      "step": 6124
    },
    {
      "epoch": 0.7364877051644322,
      "flos": 15925238538240.0,
      "grad_norm": 3.529946685907123,
      "learning_rate": 6.8509493747118e-07,
      "loss": 0.9113,
      "num_input_tokens_seen": 129921810,
      "step": 6125
    },
    {
      "epoch": 0.7366079480550712,
      "flos": 8536323317760.0,
      "grad_norm": 3.8189557558559835,
      "learning_rate": 6.845080852280221e-07,
      "loss": 1.0594,
      "num_input_tokens_seen": 129938600,
      "step": 6126
    },
    {
      "epoch": 0.7367281909457103,
      "flos": 10634232238080.0,
      "grad_norm": 3.4503925266042534,
      "learning_rate": 6.839214325439409e-07,
      "loss": 0.9413,
      "num_input_tokens_seen": 129956015,
      "step": 6127
    },
    {
      "epoch": 0.7368484338363495,
      "flos": 16714847784960.0,
      "grad_norm": 5.299160534576757,
      "learning_rate": 6.833349795079327e-07,
      "loss": 0.9098,
      "num_input_tokens_seen": 129974845,
      "step": 6128
    },
    {
      "epoch": 0.7369686767269885,
      "flos": 19523211878400.0,
      "grad_norm": 3.4217918072346403,
      "learning_rate": 6.827487262089613e-07,
      "loss": 0.8702,
      "num_input_tokens_seen": 129995070,
      "step": 6129
    },
    {
      "epoch": 0.7370889196176276,
      "flos": 50342593105920.0,
      "grad_norm": 0.869387439444077,
      "learning_rate": 6.821626727359606e-07,
      "loss": 0.79,
      "num_input_tokens_seen": 130060350,
      "step": 6130
    },
    {
      "epoch": 0.7372091625082667,
      "flos": 12784123637760.0,
      "grad_norm": 4.904805353406959,
      "learning_rate": 6.815768191778348e-07,
      "loss": 0.9619,
      "num_input_tokens_seen": 130078150,
      "step": 6131
    },
    {
      "epoch": 0.7373294053989058,
      "flos": 24056532602880.0,
      "grad_norm": 3.907762158484095,
      "learning_rate": 6.809911656234569e-07,
      "loss": 0.9235,
      "num_input_tokens_seen": 130099845,
      "step": 6132
    },
    {
      "epoch": 0.7374496482895448,
      "flos": 15274614435840.0,
      "grad_norm": 9.919136017737578,
      "learning_rate": 6.804057121616707e-07,
      "loss": 0.9736,
      "num_input_tokens_seen": 130117770,
      "step": 6133
    },
    {
      "epoch": 0.737569891180184,
      "flos": 17740233031680.0,
      "grad_norm": 3.0696798745415665,
      "learning_rate": 6.798204588812888e-07,
      "loss": 0.9084,
      "num_input_tokens_seen": 130136905,
      "step": 6134
    },
    {
      "epoch": 0.7376901340708231,
      "flos": 14669289246720.0,
      "grad_norm": 4.5293219311914745,
      "learning_rate": 6.792354058710937e-07,
      "loss": 0.9435,
      "num_input_tokens_seen": 130154095,
      "step": 6135
    },
    {
      "epoch": 0.7378103769614621,
      "flos": 16927315169280.0,
      "grad_norm": 6.836410697907839,
      "learning_rate": 6.786505532198374e-07,
      "loss": 0.8477,
      "num_input_tokens_seen": 130172760,
      "step": 6136
    },
    {
      "epoch": 0.7379306198521013,
      "flos": 15800049745920.0,
      "grad_norm": 4.4755713412441995,
      "learning_rate": 6.780659010162411e-07,
      "loss": 1.0479,
      "num_input_tokens_seen": 130191430,
      "step": 6137
    },
    {
      "epoch": 0.7380508627427403,
      "flos": 10529102254080.0,
      "grad_norm": 2.7310853188316533,
      "learning_rate": 6.774814493489975e-07,
      "loss": 1.0344,
      "num_input_tokens_seen": 130208825,
      "step": 6138
    },
    {
      "epoch": 0.7381711056333794,
      "flos": 15403236986880.0,
      "grad_norm": 2.480250167226848,
      "learning_rate": 6.768971983067655e-07,
      "loss": 0.8678,
      "num_input_tokens_seen": 130228875,
      "step": 6139
    },
    {
      "epoch": 0.7382913485240186,
      "flos": 37484558008320.0,
      "grad_norm": 1.0425136873549665,
      "learning_rate": 6.763131479781772e-07,
      "loss": 0.8958,
      "num_input_tokens_seen": 130278355,
      "step": 6140
    },
    {
      "epoch": 0.7384115914146576,
      "flos": 15484905062400.0,
      "grad_norm": 3.580667716683893,
      "learning_rate": 6.757292984518316e-07,
      "loss": 0.9658,
      "num_input_tokens_seen": 130297475,
      "step": 6141
    },
    {
      "epoch": 0.7385318343052967,
      "flos": 44018107699200.0,
      "grad_norm": 0.8222727650674448,
      "learning_rate": 6.751456498162981e-07,
      "loss": 0.8045,
      "num_input_tokens_seen": 130356230,
      "step": 6142
    },
    {
      "epoch": 0.7386520771959358,
      "flos": 12045239869440.0,
      "grad_norm": 3.444760094470945,
      "learning_rate": 6.745622021601174e-07,
      "loss": 1.048,
      "num_input_tokens_seen": 130372975,
      "step": 6143
    },
    {
      "epoch": 0.7387723200865749,
      "flos": 13308731166720.0,
      "grad_norm": 3.7285840451340797,
      "learning_rate": 6.739789555717954e-07,
      "loss": 0.8901,
      "num_input_tokens_seen": 130389670,
      "step": 6144
    },
    {
      "epoch": 0.738892562977214,
      "flos": 16006783979520.0,
      "grad_norm": 3.802300615067961,
      "learning_rate": 6.733959101398124e-07,
      "loss": 0.9751,
      "num_input_tokens_seen": 130407520,
      "step": 6145
    },
    {
      "epoch": 0.7390128058678531,
      "flos": 15270475530240.0,
      "grad_norm": 5.459771507987637,
      "learning_rate": 6.728130659526143e-07,
      "loss": 1.0111,
      "num_input_tokens_seen": 130425050,
      "step": 6146
    },
    {
      "epoch": 0.7391330487584922,
      "flos": 18343412121600.0,
      "grad_norm": 4.779878905929174,
      "learning_rate": 6.7223042309862e-07,
      "loss": 0.8934,
      "num_input_tokens_seen": 130444970,
      "step": 6147
    },
    {
      "epoch": 0.7392532916491312,
      "flos": 20205207797760.0,
      "grad_norm": 4.35416683701997,
      "learning_rate": 6.716479816662144e-07,
      "loss": 0.9264,
      "num_input_tokens_seen": 130466420,
      "step": 6148
    },
    {
      "epoch": 0.7393735345397703,
      "flos": 16769129656320.0,
      "grad_norm": 3.9841624483746974,
      "learning_rate": 6.710657417437531e-07,
      "loss": 0.9226,
      "num_input_tokens_seen": 130485845,
      "step": 6149
    },
    {
      "epoch": 0.7394937774304094,
      "flos": 14174030069760.0,
      "grad_norm": 6.421885989288803,
      "learning_rate": 6.704837034195628e-07,
      "loss": 0.9723,
      "num_input_tokens_seen": 130504030,
      "step": 6150
    },
    {
      "epoch": 0.7396140203210485,
      "flos": 16691845754880.0,
      "grad_norm": 5.05557379992946,
      "learning_rate": 6.699018667819376e-07,
      "loss": 1.0381,
      "num_input_tokens_seen": 130523150,
      "step": 6151
    },
    {
      "epoch": 0.7397342632116876,
      "flos": 18184521461760.0,
      "grad_norm": 4.716481655163222,
      "learning_rate": 6.693202319191415e-07,
      "loss": 0.913,
      "num_input_tokens_seen": 130544605,
      "step": 6152
    },
    {
      "epoch": 0.7398545061023267,
      "flos": 17609587015680.0,
      "grad_norm": 2.9838662238112423,
      "learning_rate": 6.687387989194084e-07,
      "loss": 0.9275,
      "num_input_tokens_seen": 130563840,
      "step": 6153
    },
    {
      "epoch": 0.7399747489929658,
      "flos": 11682956759040.0,
      "grad_norm": 4.905463866407428,
      "learning_rate": 6.681575678709404e-07,
      "loss": 0.9851,
      "num_input_tokens_seen": 130582250,
      "step": 6154
    },
    {
      "epoch": 0.7400949918836048,
      "flos": 17136563404800.0,
      "grad_norm": 3.7259327738292884,
      "learning_rate": 6.67576538861911e-07,
      "loss": 0.9033,
      "num_input_tokens_seen": 130600545,
      "step": 6155
    },
    {
      "epoch": 0.740215234774244,
      "flos": 15488062894080.0,
      "grad_norm": 7.695664698872392,
      "learning_rate": 6.669957119804612e-07,
      "loss": 1.0074,
      "num_input_tokens_seen": 130621900,
      "step": 6156
    },
    {
      "epoch": 0.7403354776648831,
      "flos": 13281590231040.0,
      "grad_norm": 5.64724064515555,
      "learning_rate": 6.66415087314702e-07,
      "loss": 0.8898,
      "num_input_tokens_seen": 130636575,
      "step": 6157
    },
    {
      "epoch": 0.7404557205555221,
      "flos": 11971083141120.0,
      "grad_norm": 3.651778396668399,
      "learning_rate": 6.65834664952714e-07,
      "loss": 0.9126,
      "num_input_tokens_seen": 130653745,
      "step": 6158
    },
    {
      "epoch": 0.7405759634461613,
      "flos": 15064814346240.0,
      "grad_norm": 5.752380873314103,
      "learning_rate": 6.652544449825457e-07,
      "loss": 0.9468,
      "num_input_tokens_seen": 130673720,
      "step": 6159
    },
    {
      "epoch": 0.7406962063368003,
      "flos": 14536466472960.0,
      "grad_norm": 3.6900338987535326,
      "learning_rate": 6.646744274922182e-07,
      "loss": 0.9508,
      "num_input_tokens_seen": 130691885,
      "step": 6160
    },
    {
      "epoch": 0.7408164492274394,
      "flos": 14042556272640.0,
      "grad_norm": 7.957837580340411,
      "learning_rate": 6.640946125697171e-07,
      "loss": 0.9405,
      "num_input_tokens_seen": 130709135,
      "step": 6161
    },
    {
      "epoch": 0.7409366921180786,
      "flos": 20807405813760.0,
      "grad_norm": 4.872600062494207,
      "learning_rate": 6.635150003030017e-07,
      "loss": 0.9473,
      "num_input_tokens_seen": 130727380,
      "step": 6162
    },
    {
      "epoch": 0.7410569350087176,
      "flos": 16297884241920.0,
      "grad_norm": 5.340273791662922,
      "learning_rate": 6.629355907799981e-07,
      "loss": 1.0525,
      "num_input_tokens_seen": 130746905,
      "step": 6163
    },
    {
      "epoch": 0.7411771778993567,
      "flos": 21694695014400.0,
      "grad_norm": 3.0658026811941257,
      "learning_rate": 6.623563840886015e-07,
      "loss": 0.8858,
      "num_input_tokens_seen": 130767550,
      "step": 6164
    },
    {
      "epoch": 0.7412974207899958,
      "flos": 14567347752960.0,
      "grad_norm": 8.789101996439033,
      "learning_rate": 6.617773803166795e-07,
      "loss": 0.8962,
      "num_input_tokens_seen": 130785595,
      "step": 6165
    },
    {
      "epoch": 0.7414176636806349,
      "flos": 15694674493440.0,
      "grad_norm": 5.347850988282218,
      "learning_rate": 6.611985795520634e-07,
      "loss": 1.0153,
      "num_input_tokens_seen": 130803860,
      "step": 6166
    },
    {
      "epoch": 0.7415379065712739,
      "flos": 17897069568000.0,
      "grad_norm": 14.365199189496337,
      "learning_rate": 6.606199818825588e-07,
      "loss": 0.968,
      "num_input_tokens_seen": 130824035,
      "step": 6167
    },
    {
      "epoch": 0.7416581494619131,
      "flos": 11943696936960.0,
      "grad_norm": 3.6431178125247414,
      "learning_rate": 6.600415873959377e-07,
      "loss": 1.0032,
      "num_input_tokens_seen": 130841630,
      "step": 6168
    },
    {
      "epoch": 0.7417783923525522,
      "flos": 20256945008640.0,
      "grad_norm": 8.424642982049743,
      "learning_rate": 6.594633961799437e-07,
      "loss": 0.8327,
      "num_input_tokens_seen": 130860390,
      "step": 6169
    },
    {
      "epoch": 0.7418986352431912,
      "flos": 14252754923520.0,
      "grad_norm": 2.9753846568983895,
      "learning_rate": 6.588854083222857e-07,
      "loss": 1.0127,
      "num_input_tokens_seen": 130879545,
      "step": 6170
    },
    {
      "epoch": 0.7420188781338304,
      "flos": 12941297418240.0,
      "grad_norm": 3.725278877213449,
      "learning_rate": 6.583076239106444e-07,
      "loss": 0.9879,
      "num_input_tokens_seen": 130897770,
      "step": 6171
    },
    {
      "epoch": 0.7421391210244694,
      "flos": 9716184391680.0,
      "grad_norm": 4.121422343238758,
      "learning_rate": 6.577300430326707e-07,
      "loss": 0.9456,
      "num_input_tokens_seen": 130912435,
      "step": 6172
    },
    {
      "epoch": 0.7422593639151085,
      "flos": 11289945661440.0,
      "grad_norm": 3.6380303432128227,
      "learning_rate": 6.571526657759821e-07,
      "loss": 0.917,
      "num_input_tokens_seen": 130927895,
      "step": 6173
    },
    {
      "epoch": 0.7423796068057477,
      "flos": 21460727869440.0,
      "grad_norm": 6.3422717547218195,
      "learning_rate": 6.565754922281663e-07,
      "loss": 0.8993,
      "num_input_tokens_seen": 130949860,
      "step": 6174
    },
    {
      "epoch": 0.7424998496963867,
      "flos": 14567225118720.0,
      "grad_norm": 2.9987636243694347,
      "learning_rate": 6.559985224767801e-07,
      "loss": 0.9851,
      "num_input_tokens_seen": 130967455,
      "step": 6175
    },
    {
      "epoch": 0.7426200925870258,
      "flos": 15537163468800.0,
      "grad_norm": 4.260235746066237,
      "learning_rate": 6.55421756609349e-07,
      "loss": 0.9478,
      "num_input_tokens_seen": 130985430,
      "step": 6176
    },
    {
      "epoch": 0.7427403354776649,
      "flos": 18814657536000.0,
      "grad_norm": 2.9475820686232863,
      "learning_rate": 6.54845194713369e-07,
      "loss": 0.9822,
      "num_input_tokens_seen": 131006100,
      "step": 6177
    },
    {
      "epoch": 0.742860578368304,
      "flos": 14118675148800.0,
      "grad_norm": 3.4498985717901403,
      "learning_rate": 6.542688368763034e-07,
      "loss": 0.9953,
      "num_input_tokens_seen": 131024225,
      "step": 6178
    },
    {
      "epoch": 0.742980821258943,
      "flos": 17661354885120.0,
      "grad_norm": 2.695198565300242,
      "learning_rate": 6.536926831855854e-07,
      "loss": 0.9706,
      "num_input_tokens_seen": 131043110,
      "step": 6179
    },
    {
      "epoch": 0.7431010641495821,
      "flos": 17949879828480.0,
      "grad_norm": 6.379203756744909,
      "learning_rate": 6.531167337286165e-07,
      "loss": 0.916,
      "num_input_tokens_seen": 131062850,
      "step": 6180
    },
    {
      "epoch": 0.7432213070402213,
      "flos": 15458714542080.0,
      "grad_norm": 2.5439834562579433,
      "learning_rate": 6.52540988592768e-07,
      "loss": 0.9889,
      "num_input_tokens_seen": 131083590,
      "step": 6181
    },
    {
      "epoch": 0.7433415499308603,
      "flos": 10450592010240.0,
      "grad_norm": 3.3759830597276212,
      "learning_rate": 6.519654478653814e-07,
      "loss": 1.0251,
      "num_input_tokens_seen": 131101675,
      "step": 6182
    },
    {
      "epoch": 0.7434617928214994,
      "flos": 48087541063680.0,
      "grad_norm": 0.7599937273067813,
      "learning_rate": 6.51390111633763e-07,
      "loss": 0.7672,
      "num_input_tokens_seen": 131166670,
      "step": 6183
    },
    {
      "epoch": 0.7435820357121385,
      "flos": 19493771550720.0,
      "grad_norm": 2.4681515511547856,
      "learning_rate": 6.508149799851932e-07,
      "loss": 0.9498,
      "num_input_tokens_seen": 131188055,
      "step": 6184
    },
    {
      "epoch": 0.7437022786027776,
      "flos": 17057930526720.0,
      "grad_norm": 4.538783683683384,
      "learning_rate": 6.502400530069183e-07,
      "loss": 0.813,
      "num_input_tokens_seen": 131207660,
      "step": 6185
    },
    {
      "epoch": 0.7438225214934167,
      "flos": 15533055221760.0,
      "grad_norm": 4.3614186517978295,
      "learning_rate": 6.496653307861535e-07,
      "loss": 0.8777,
      "num_input_tokens_seen": 131228050,
      "step": 6186
    },
    {
      "epoch": 0.7439427643840558,
      "flos": 14357517004800.0,
      "grad_norm": 4.420068378523707,
      "learning_rate": 6.490908134100857e-07,
      "loss": 0.8437,
      "num_input_tokens_seen": 131246235,
      "step": 6187
    },
    {
      "epoch": 0.7440630072746949,
      "flos": 14802909143040.0,
      "grad_norm": 4.924262736174121,
      "learning_rate": 6.48516500965866e-07,
      "loss": 0.8847,
      "num_input_tokens_seen": 131265890,
      "step": 6188
    },
    {
      "epoch": 0.7441832501653339,
      "flos": 18866026844160.0,
      "grad_norm": 2.882835987688023,
      "learning_rate": 6.479423935406192e-07,
      "loss": 1.0085,
      "num_input_tokens_seen": 131285595,
      "step": 6189
    },
    {
      "epoch": 0.7443034930559731,
      "flos": 49126025625600.0,
      "grad_norm": 0.9308204159285732,
      "learning_rate": 6.473684912214357e-07,
      "loss": 0.9022,
      "num_input_tokens_seen": 131348875,
      "step": 6190
    },
    {
      "epoch": 0.7444237359466122,
      "flos": 13224671723520.0,
      "grad_norm": 2.9538008305086305,
      "learning_rate": 6.467947940953778e-07,
      "loss": 0.8932,
      "num_input_tokens_seen": 131367120,
      "step": 6191
    },
    {
      "epoch": 0.7445439788372512,
      "flos": 16216706703360.0,
      "grad_norm": 2.6333482689574765,
      "learning_rate": 6.462213022494732e-07,
      "loss": 0.9203,
      "num_input_tokens_seen": 131386085,
      "step": 6192
    },
    {
      "epoch": 0.7446642217278904,
      "flos": 48008509624320.0,
      "grad_norm": 0.779504160453357,
      "learning_rate": 6.456480157707201e-07,
      "loss": 0.8405,
      "num_input_tokens_seen": 131450580,
      "step": 6193
    },
    {
      "epoch": 0.7447844646185294,
      "flos": 12335696302080.0,
      "grad_norm": 3.251657210166342,
      "learning_rate": 6.450749347460866e-07,
      "loss": 1.0479,
      "num_input_tokens_seen": 131467275,
      "step": 6194
    },
    {
      "epoch": 0.7449047075091685,
      "flos": 18946959114240.0,
      "grad_norm": 18.581653102758377,
      "learning_rate": 6.445020592625083e-07,
      "loss": 0.9852,
      "num_input_tokens_seen": 131487645,
      "step": 6195
    },
    {
      "epoch": 0.7450249503998077,
      "flos": 10002164674560.0,
      "grad_norm": 4.090268817079347,
      "learning_rate": 6.4392938940689e-07,
      "loss": 0.9923,
      "num_input_tokens_seen": 131502780,
      "step": 6196
    },
    {
      "epoch": 0.7451451932904467,
      "flos": 13909059010560.0,
      "grad_norm": 4.106354157913665,
      "learning_rate": 6.433569252661049e-07,
      "loss": 0.8928,
      "num_input_tokens_seen": 131520500,
      "step": 6197
    },
    {
      "epoch": 0.7452654361810858,
      "flos": 8798780375040.0,
      "grad_norm": 6.028763807116031,
      "learning_rate": 6.427846669269952e-07,
      "loss": 0.8951,
      "num_input_tokens_seen": 131537840,
      "step": 6198
    },
    {
      "epoch": 0.7453856790717249,
      "flos": 15721018306560.0,
      "grad_norm": 3.5227491294205167,
      "learning_rate": 6.422126144763729e-07,
      "loss": 1.0037,
      "num_input_tokens_seen": 131556950,
      "step": 6199
    },
    {
      "epoch": 0.745505921962364,
      "flos": 14199760711680.0,
      "grad_norm": 3.9498302616396264,
      "learning_rate": 6.416407680010174e-07,
      "loss": 0.9558,
      "num_input_tokens_seen": 131571030,
      "step": 6200
    },
    {
      "epoch": 0.745626164853003,
      "flos": 17553588264960.0,
      "grad_norm": 4.698173679446217,
      "learning_rate": 6.410691275876774e-07,
      "loss": 1.001,
      "num_input_tokens_seen": 131590170,
      "step": 6201
    },
    {
      "epoch": 0.7457464077436422,
      "flos": 10267381002240.0,
      "grad_norm": 4.647346124836014,
      "learning_rate": 6.404976933230704e-07,
      "loss": 0.9543,
      "num_input_tokens_seen": 131606410,
      "step": 6202
    },
    {
      "epoch": 0.7458666506342813,
      "flos": 24268846694400.0,
      "grad_norm": 3.390013608877047,
      "learning_rate": 6.399264652938813e-07,
      "loss": 0.9196,
      "num_input_tokens_seen": 131627035,
      "step": 6203
    },
    {
      "epoch": 0.7459868935249203,
      "flos": 17267270737920.0,
      "grad_norm": 3.488439814775166,
      "learning_rate": 6.393554435867679e-07,
      "loss": 0.9346,
      "num_input_tokens_seen": 131647605,
      "step": 6204
    },
    {
      "epoch": 0.7461071364155595,
      "flos": 15563997818880.0,
      "grad_norm": 3.4437980523884915,
      "learning_rate": 6.387846282883502e-07,
      "loss": 1.0168,
      "num_input_tokens_seen": 131663855,
      "step": 6205
    },
    {
      "epoch": 0.7462273793061985,
      "flos": 16268413255680.0,
      "grad_norm": 4.317341174201215,
      "learning_rate": 6.38214019485223e-07,
      "loss": 0.9598,
      "num_input_tokens_seen": 131682400,
      "step": 6206
    },
    {
      "epoch": 0.7463476221968376,
      "flos": 14168634163200.0,
      "grad_norm": 13.76816039275349,
      "learning_rate": 6.376436172639461e-07,
      "loss": 0.9062,
      "num_input_tokens_seen": 131699965,
      "step": 6207
    },
    {
      "epoch": 0.7464678650874768,
      "flos": 11918640783360.0,
      "grad_norm": 7.929228550450128,
      "learning_rate": 6.370734217110487e-07,
      "loss": 0.8392,
      "num_input_tokens_seen": 131718430,
      "step": 6208
    },
    {
      "epoch": 0.7465881079781158,
      "flos": 34387125227520.0,
      "grad_norm": 3.4570425604511303,
      "learning_rate": 6.36503432913031e-07,
      "loss": 0.8392,
      "num_input_tokens_seen": 131741295,
      "step": 6209
    },
    {
      "epoch": 0.7467083508687549,
      "flos": 13959539220480.0,
      "grad_norm": 9.02879643358197,
      "learning_rate": 6.359336509563569e-07,
      "loss": 0.8756,
      "num_input_tokens_seen": 131757035,
      "step": 6210
    },
    {
      "epoch": 0.7468285937593939,
      "flos": 12679698800640.0,
      "grad_norm": 2.9947314093235904,
      "learning_rate": 6.353640759274641e-07,
      "loss": 0.9955,
      "num_input_tokens_seen": 131775645,
      "step": 6211
    },
    {
      "epoch": 0.7469488366500331,
      "flos": 16448834334720.0,
      "grad_norm": 4.531142240097965,
      "learning_rate": 6.347947079127556e-07,
      "loss": 0.9317,
      "num_input_tokens_seen": 131793265,
      "step": 6212
    },
    {
      "epoch": 0.7470690795406721,
      "flos": 11813909360640.0,
      "grad_norm": 3.7696611609333686,
      "learning_rate": 6.342255469986053e-07,
      "loss": 0.9519,
      "num_input_tokens_seen": 131811730,
      "step": 6213
    },
    {
      "epoch": 0.7471893224313112,
      "flos": 17924118528000.0,
      "grad_norm": 2.886936453119567,
      "learning_rate": 6.336565932713533e-07,
      "loss": 0.9638,
      "num_input_tokens_seen": 131830875,
      "step": 6214
    },
    {
      "epoch": 0.7473095653219504,
      "flos": 16007366492160.0,
      "grad_norm": 2.9303954222822246,
      "learning_rate": 6.330878468173088e-07,
      "loss": 0.9618,
      "num_input_tokens_seen": 131850660,
      "step": 6215
    },
    {
      "epoch": 0.7474298082125894,
      "flos": 12836719288320.0,
      "grad_norm": 3.7115898121169915,
      "learning_rate": 6.32519307722752e-07,
      "loss": 0.9289,
      "num_input_tokens_seen": 131868275,
      "step": 6216
    },
    {
      "epoch": 0.7475500511032285,
      "flos": 44444851322880.0,
      "grad_norm": 0.7954398677140568,
      "learning_rate": 6.31950976073929e-07,
      "loss": 0.7687,
      "num_input_tokens_seen": 131922085,
      "step": 6217
    },
    {
      "epoch": 0.7476702939938676,
      "flos": 12600698019840.0,
      "grad_norm": 5.083371688614691,
      "learning_rate": 6.31382851957055e-07,
      "loss": 0.9874,
      "num_input_tokens_seen": 131938625,
      "step": 6218
    },
    {
      "epoch": 0.7477905368845067,
      "flos": 19889725870080.0,
      "grad_norm": 2.698388508858346,
      "learning_rate": 6.308149354583143e-07,
      "loss": 0.8938,
      "num_input_tokens_seen": 131957750,
      "step": 6219
    },
    {
      "epoch": 0.7479107797751458,
      "flos": 19129740902400.0,
      "grad_norm": 5.808745453424154,
      "learning_rate": 6.302472266638586e-07,
      "loss": 0.9999,
      "num_input_tokens_seen": 131978010,
      "step": 6220
    },
    {
      "epoch": 0.7480310226657849,
      "flos": 24212265431040.0,
      "grad_norm": 3.641373193177902,
      "learning_rate": 6.296797256598101e-07,
      "loss": 0.902,
      "num_input_tokens_seen": 131999210,
      "step": 6221
    },
    {
      "epoch": 0.748151265556424,
      "flos": 17660803031040.0,
      "grad_norm": 2.5483558889637856,
      "learning_rate": 6.291124325322576e-07,
      "loss": 1.0054,
      "num_input_tokens_seen": 132019055,
      "step": 6222
    },
    {
      "epoch": 0.748271508447063,
      "flos": 27415326842880.0,
      "grad_norm": 2.943880658985786,
      "learning_rate": 6.285453473672595e-07,
      "loss": 0.8233,
      "num_input_tokens_seen": 132041345,
      "step": 6223
    },
    {
      "epoch": 0.7483917513377022,
      "flos": 15300007833600.0,
      "grad_norm": 3.2187821469970066,
      "learning_rate": 6.279784702508415e-07,
      "loss": 0.9285,
      "num_input_tokens_seen": 132061815,
      "step": 6224
    },
    {
      "epoch": 0.7485119942283412,
      "flos": 44609935011840.0,
      "grad_norm": 0.8594481003621955,
      "learning_rate": 6.274118012689979e-07,
      "loss": 0.8347,
      "num_input_tokens_seen": 132123435,
      "step": 6225
    },
    {
      "epoch": 0.7486322371189803,
      "flos": 21334435368960.0,
      "grad_norm": 2.579868998343618,
      "learning_rate": 6.268453405076943e-07,
      "loss": 0.8718,
      "num_input_tokens_seen": 132145550,
      "step": 6226
    },
    {
      "epoch": 0.7487524800096195,
      "flos": 13436617912320.0,
      "grad_norm": 5.0791296331960805,
      "learning_rate": 6.262790880528592e-07,
      "loss": 1.0186,
      "num_input_tokens_seen": 132162890,
      "step": 6227
    },
    {
      "epoch": 0.7488727229002585,
      "flos": 13255951564800.0,
      "grad_norm": 6.625464369135907,
      "learning_rate": 6.257130439903951e-07,
      "loss": 0.97,
      "num_input_tokens_seen": 132179105,
      "step": 6228
    },
    {
      "epoch": 0.7489929657908976,
      "flos": 16796209274880.0,
      "grad_norm": 3.9965555142389246,
      "learning_rate": 6.251472084061695e-07,
      "loss": 1.0006,
      "num_input_tokens_seen": 132197745,
      "step": 6229
    },
    {
      "epoch": 0.7491132086815367,
      "flos": 14587590512640.0,
      "grad_norm": 4.604541792051202,
      "learning_rate": 6.245815813860191e-07,
      "loss": 1.0904,
      "num_input_tokens_seen": 132212975,
      "step": 6230
    },
    {
      "epoch": 0.7492334515721758,
      "flos": 16350357258240.0,
      "grad_norm": 3.891560637331861,
      "learning_rate": 6.240161630157495e-07,
      "loss": 0.8797,
      "num_input_tokens_seen": 132232050,
      "step": 6231
    },
    {
      "epoch": 0.7493536944628149,
      "flos": 11604017295360.0,
      "grad_norm": 3.880895252570698,
      "learning_rate": 6.23450953381133e-07,
      "loss": 0.8872,
      "num_input_tokens_seen": 132249860,
      "step": 6232
    },
    {
      "epoch": 0.749473937353454,
      "flos": 10841732935680.0,
      "grad_norm": 4.049865795784743,
      "learning_rate": 6.228859525679131e-07,
      "loss": 0.8712,
      "num_input_tokens_seen": 132263995,
      "step": 6233
    },
    {
      "epoch": 0.7495941802440931,
      "flos": 13438089523200.0,
      "grad_norm": 7.376598378914148,
      "learning_rate": 6.223211606617986e-07,
      "loss": 0.9961,
      "num_input_tokens_seen": 132282135,
      "step": 6234
    },
    {
      "epoch": 0.7497144231347321,
      "flos": 15983475363840.0,
      "grad_norm": 3.1853578138351497,
      "learning_rate": 6.217565777484701e-07,
      "loss": 1.0208,
      "num_input_tokens_seen": 132300950,
      "step": 6235
    },
    {
      "epoch": 0.7498346660253713,
      "flos": 17241877340160.0,
      "grad_norm": 2.6638426927740504,
      "learning_rate": 6.211922039135722e-07,
      "loss": 0.987,
      "num_input_tokens_seen": 132320815,
      "step": 6236
    },
    {
      "epoch": 0.7499549089160104,
      "flos": 17344738590720.0,
      "grad_norm": 3.0111849815281952,
      "learning_rate": 6.206280392427201e-07,
      "loss": 1.0054,
      "num_input_tokens_seen": 132340120,
      "step": 6237
    },
    {
      "epoch": 0.7500751518066494,
      "flos": 24294761287680.0,
      "grad_norm": 3.5559478202312698,
      "learning_rate": 6.200640838214983e-07,
      "loss": 0.9249,
      "num_input_tokens_seen": 132362615,
      "step": 6238
    },
    {
      "epoch": 0.7501953946972886,
      "flos": 13361050890240.0,
      "grad_norm": 6.356686785458894,
      "learning_rate": 6.195003377354578e-07,
      "loss": 0.8567,
      "num_input_tokens_seen": 132381605,
      "step": 6239
    },
    {
      "epoch": 0.7503156375879276,
      "flos": 14376134860800.0,
      "grad_norm": 6.751843518599806,
      "learning_rate": 6.189368010701183e-07,
      "loss": 0.9064,
      "num_input_tokens_seen": 132398385,
      "step": 6240
    },
    {
      "epoch": 0.7504358804785667,
      "flos": 9506353643520.0,
      "grad_norm": 3.124171795770019,
      "learning_rate": 6.183734739109683e-07,
      "loss": 0.9481,
      "num_input_tokens_seen": 132415925,
      "step": 6241
    },
    {
      "epoch": 0.7505561233692057,
      "flos": 20991628554240.0,
      "grad_norm": 6.133259090029516,
      "learning_rate": 6.178103563434629e-07,
      "loss": 0.8832,
      "num_input_tokens_seen": 132434645,
      "step": 6242
    },
    {
      "epoch": 0.7506763662598449,
      "flos": 14409438167040.0,
      "grad_norm": 3.3516837144819958,
      "learning_rate": 6.172474484530283e-07,
      "loss": 1.0404,
      "num_input_tokens_seen": 132453100,
      "step": 6243
    },
    {
      "epoch": 0.750796609150484,
      "flos": 26575513313280.0,
      "grad_norm": 3.0707757420792046,
      "learning_rate": 6.166847503250563e-07,
      "loss": 0.9475,
      "num_input_tokens_seen": 132475060,
      "step": 6244
    },
    {
      "epoch": 0.750916852041123,
      "flos": 13911389061120.0,
      "grad_norm": 3.699992081917203,
      "learning_rate": 6.161222620449078e-07,
      "loss": 0.98,
      "num_input_tokens_seen": 132493555,
      "step": 6245
    },
    {
      "epoch": 0.7510370949317622,
      "flos": 17865667092480.0,
      "grad_norm": 4.552245858067743,
      "learning_rate": 6.155599836979117e-07,
      "loss": 0.9869,
      "num_input_tokens_seen": 132511960,
      "step": 6246
    },
    {
      "epoch": 0.7511573378224012,
      "flos": 13518194012160.0,
      "grad_norm": 2.9332670189846484,
      "learning_rate": 6.149979153693649e-07,
      "loss": 1.0031,
      "num_input_tokens_seen": 132528935,
      "step": 6247
    },
    {
      "epoch": 0.7512775807130403,
      "flos": 14147287695360.0,
      "grad_norm": 4.257953382488621,
      "learning_rate": 6.144360571445343e-07,
      "loss": 0.95,
      "num_input_tokens_seen": 132547800,
      "step": 6248
    },
    {
      "epoch": 0.7513978236036795,
      "flos": 14723785728000.0,
      "grad_norm": 6.469244454324976,
      "learning_rate": 6.138744091086509e-07,
      "loss": 0.9972,
      "num_input_tokens_seen": 132567105,
      "step": 6249
    },
    {
      "epoch": 0.7515180664943185,
      "flos": 19627575398400.0,
      "grad_norm": 4.241506729158412,
      "learning_rate": 6.133129713469183e-07,
      "loss": 0.9216,
      "num_input_tokens_seen": 132586030,
      "step": 6250
    },
    {
      "epoch": 0.7516383093849576,
      "flos": 24112286085120.0,
      "grad_norm": 4.074102646955485,
      "learning_rate": 6.127517439445053e-07,
      "loss": 0.8277,
      "num_input_tokens_seen": 132606595,
      "step": 6251
    },
    {
      "epoch": 0.7517585522755967,
      "flos": 21195634176000.0,
      "grad_norm": 3.305587321395046,
      "learning_rate": 6.121907269865498e-07,
      "loss": 1.0129,
      "num_input_tokens_seen": 132625805,
      "step": 6252
    },
    {
      "epoch": 0.7518787951662358,
      "flos": 49992274944000.0,
      "grad_norm": 0.9691011346215398,
      "learning_rate": 6.116299205581577e-07,
      "loss": 0.9204,
      "num_input_tokens_seen": 132680355,
      "step": 6253
    },
    {
      "epoch": 0.7519990380568748,
      "flos": 24399615344640.0,
      "grad_norm": 3.2624705287727642,
      "learning_rate": 6.110693247444018e-07,
      "loss": 0.8696,
      "num_input_tokens_seen": 132701910,
      "step": 6254
    },
    {
      "epoch": 0.752119280947514,
      "flos": 15432340070400.0,
      "grad_norm": 6.088204232277838,
      "learning_rate": 6.105089396303258e-07,
      "loss": 1.0125,
      "num_input_tokens_seen": 132720020,
      "step": 6255
    },
    {
      "epoch": 0.7522395238381531,
      "flos": 23349879091200.0,
      "grad_norm": 3.6061433138346777,
      "learning_rate": 6.099487653009383e-07,
      "loss": 0.9496,
      "num_input_tokens_seen": 132739085,
      "step": 6256
    },
    {
      "epoch": 0.7523597667287921,
      "flos": 16690128875520.0,
      "grad_norm": 5.5740343155586505,
      "learning_rate": 6.093888018412192e-07,
      "loss": 1.0277,
      "num_input_tokens_seen": 132754995,
      "step": 6257
    },
    {
      "epoch": 0.7524800096194313,
      "flos": 48226096988160.0,
      "grad_norm": 0.7304063658161126,
      "learning_rate": 6.088290493361125e-07,
      "loss": 0.7675,
      "num_input_tokens_seen": 132819600,
      "step": 6258
    },
    {
      "epoch": 0.7526002525100703,
      "flos": 9165754245120.0,
      "grad_norm": 4.378525103979837,
      "learning_rate": 6.082695078705322e-07,
      "loss": 0.8951,
      "num_input_tokens_seen": 132836800,
      "step": 6259
    },
    {
      "epoch": 0.7527204954007094,
      "flos": 15196257484800.0,
      "grad_norm": 6.046175342072858,
      "learning_rate": 6.077101775293618e-07,
      "loss": 0.8729,
      "num_input_tokens_seen": 132855345,
      "step": 6260
    },
    {
      "epoch": 0.7528407382913486,
      "flos": 13434717081600.0,
      "grad_norm": 9.385694583640724,
      "learning_rate": 6.071510583974504e-07,
      "loss": 1.0213,
      "num_input_tokens_seen": 132870250,
      "step": 6261
    },
    {
      "epoch": 0.7529609811819876,
      "flos": 10765368791040.0,
      "grad_norm": 5.654265840804297,
      "learning_rate": 6.065921505596161e-07,
      "loss": 0.9071,
      "num_input_tokens_seen": 132888250,
      "step": 6262
    },
    {
      "epoch": 0.7530812240726267,
      "flos": 13728239370240.0,
      "grad_norm": 2.9835857526098137,
      "learning_rate": 6.060334541006445e-07,
      "loss": 0.9629,
      "num_input_tokens_seen": 132906465,
      "step": 6263
    },
    {
      "epoch": 0.7532014669632658,
      "flos": 19760950026240.0,
      "grad_norm": 2.6937190998898397,
      "learning_rate": 6.05474969105289e-07,
      "loss": 0.8855,
      "num_input_tokens_seen": 132929175,
      "step": 6264
    },
    {
      "epoch": 0.7533217098539049,
      "flos": 9978886717440.0,
      "grad_norm": 20.875293112884144,
      "learning_rate": 6.049166956582725e-07,
      "loss": 0.9293,
      "num_input_tokens_seen": 132947160,
      "step": 6265
    },
    {
      "epoch": 0.753441952744544,
      "flos": 18812112875520.0,
      "grad_norm": 3.5508024678787122,
      "learning_rate": 6.043586338442841e-07,
      "loss": 1.0629,
      "num_input_tokens_seen": 132965935,
      "step": 6266
    },
    {
      "epoch": 0.7535621956351831,
      "flos": 16979634892800.0,
      "grad_norm": 7.869404576036719,
      "learning_rate": 6.038007837479815e-07,
      "loss": 0.9223,
      "num_input_tokens_seen": 132986760,
      "step": 6267
    },
    {
      "epoch": 0.7536824385258222,
      "flos": 15482391060480.0,
      "grad_norm": 4.466382660068672,
      "learning_rate": 6.032431454539897e-07,
      "loss": 0.8229,
      "num_input_tokens_seen": 133005325,
      "step": 6268
    },
    {
      "epoch": 0.7538026814164612,
      "flos": 20596502016000.0,
      "grad_norm": 3.3080786221811938,
      "learning_rate": 6.026857190469014e-07,
      "loss": 1.003,
      "num_input_tokens_seen": 133026800,
      "step": 6269
    },
    {
      "epoch": 0.7539229243071004,
      "flos": 14985936199680.0,
      "grad_norm": 3.9262902684228047,
      "learning_rate": 6.0212850461128e-07,
      "loss": 0.9372,
      "num_input_tokens_seen": 133045640,
      "step": 6270
    },
    {
      "epoch": 0.7540431671977395,
      "flos": 10712619847680.0,
      "grad_norm": 4.282842632244762,
      "learning_rate": 6.015715022316516e-07,
      "loss": 0.936,
      "num_input_tokens_seen": 133063340,
      "step": 6271
    },
    {
      "epoch": 0.7541634100883785,
      "flos": 13308577873920.0,
      "grad_norm": 4.147190862152252,
      "learning_rate": 6.010147119925154e-07,
      "loss": 0.9667,
      "num_input_tokens_seen": 133080815,
      "step": 6272
    },
    {
      "epoch": 0.7542836529790176,
      "flos": 14619575500800.0,
      "grad_norm": 3.25447332196679,
      "learning_rate": 6.004581339783348e-07,
      "loss": 0.8381,
      "num_input_tokens_seen": 133098855,
      "step": 6273
    },
    {
      "epoch": 0.7544038958696567,
      "flos": 13541288017920.0,
      "grad_norm": 7.040653976062488,
      "learning_rate": 5.999017682735425e-07,
      "loss": 0.8636,
      "num_input_tokens_seen": 133114965,
      "step": 6274
    },
    {
      "epoch": 0.7545241387602958,
      "flos": 22617494937600.0,
      "grad_norm": 3.638152105338251,
      "learning_rate": 5.993456149625387e-07,
      "loss": 0.853,
      "num_input_tokens_seen": 133135835,
      "step": 6275
    },
    {
      "epoch": 0.7546443816509348,
      "flos": 14404348846080.0,
      "grad_norm": 3.1683299732407826,
      "learning_rate": 5.987896741296909e-07,
      "loss": 1.017,
      "num_input_tokens_seen": 133153295,
      "step": 6276
    },
    {
      "epoch": 0.754764624541574,
      "flos": 16848498339840.0,
      "grad_norm": 5.179147821680076,
      "learning_rate": 5.982339458593361e-07,
      "loss": 0.9736,
      "num_input_tokens_seen": 133172955,
      "step": 6277
    },
    {
      "epoch": 0.7548848674322131,
      "flos": 18027776901120.0,
      "grad_norm": 3.4644437986999566,
      "learning_rate": 5.976784302357767e-07,
      "loss": 1.04,
      "num_input_tokens_seen": 133193240,
      "step": 6278
    },
    {
      "epoch": 0.7550051103228521,
      "flos": 13885351833600.0,
      "grad_norm": 3.721363207319137,
      "learning_rate": 5.971231273432855e-07,
      "loss": 0.9188,
      "num_input_tokens_seen": 133212445,
      "step": 6279
    },
    {
      "epoch": 0.7551253532134913,
      "flos": 45925692825600.0,
      "grad_norm": 0.8261051737235555,
      "learning_rate": 5.965680372661e-07,
      "loss": 0.7765,
      "num_input_tokens_seen": 133269730,
      "step": 6280
    },
    {
      "epoch": 0.7552455961041303,
      "flos": 18553120235520.0,
      "grad_norm": 6.5058866401222435,
      "learning_rate": 5.960131600884266e-07,
      "loss": 0.7415,
      "num_input_tokens_seen": 133288720,
      "step": 6281
    },
    {
      "epoch": 0.7553658389947694,
      "flos": 17424843079680.0,
      "grad_norm": 4.296298123562694,
      "learning_rate": 5.954584958944413e-07,
      "loss": 0.9524,
      "num_input_tokens_seen": 133307105,
      "step": 6282
    },
    {
      "epoch": 0.7554860818854086,
      "flos": 15485456916480.0,
      "grad_norm": 3.2938219732133587,
      "learning_rate": 5.949040447682854e-07,
      "loss": 1.0036,
      "num_input_tokens_seen": 133326650,
      "step": 6283
    },
    {
      "epoch": 0.7556063247760476,
      "flos": 11577612165120.0,
      "grad_norm": 3.9650290760643414,
      "learning_rate": 5.943498067940686e-07,
      "loss": 0.8831,
      "num_input_tokens_seen": 133343395,
      "step": 6284
    },
    {
      "epoch": 0.7557265676666867,
      "flos": 19391891374080.0,
      "grad_norm": 3.3312809983863505,
      "learning_rate": 5.937957820558686e-07,
      "loss": 0.9992,
      "num_input_tokens_seen": 133362460,
      "step": 6285
    },
    {
      "epoch": 0.7558468105573258,
      "flos": 44517659074560.0,
      "grad_norm": 0.8551874612792746,
      "learning_rate": 5.932419706377296e-07,
      "loss": 0.8588,
      "num_input_tokens_seen": 133420485,
      "step": 6286
    },
    {
      "epoch": 0.7559670534479649,
      "flos": 23713143275520.0,
      "grad_norm": 4.1586466597939795,
      "learning_rate": 5.92688372623666e-07,
      "loss": 0.9336,
      "num_input_tokens_seen": 133438910,
      "step": 6287
    },
    {
      "epoch": 0.7560872963386039,
      "flos": 9926383042560.0,
      "grad_norm": 4.518552893511909,
      "learning_rate": 5.921349880976574e-07,
      "loss": 0.9295,
      "num_input_tokens_seen": 133456465,
      "step": 6288
    },
    {
      "epoch": 0.7562075392292431,
      "flos": 14488438947840.0,
      "grad_norm": 3.305168098872461,
      "learning_rate": 5.915818171436515e-07,
      "loss": 1.0018,
      "num_input_tokens_seen": 133475520,
      "step": 6289
    },
    {
      "epoch": 0.7563277821198822,
      "flos": 14463045550080.0,
      "grad_norm": 3.1507278328133714,
      "learning_rate": 5.910288598455642e-07,
      "loss": 0.9334,
      "num_input_tokens_seen": 133494590,
      "step": 6290
    },
    {
      "epoch": 0.7564480250105212,
      "flos": 13177563955200.0,
      "grad_norm": 5.958806088061365,
      "learning_rate": 5.90476116287278e-07,
      "loss": 0.9227,
      "num_input_tokens_seen": 133511910,
      "step": 6291
    },
    {
      "epoch": 0.7565682679011604,
      "flos": 15065764761600.0,
      "grad_norm": 2.9495564854734577,
      "learning_rate": 5.899235865526456e-07,
      "loss": 0.8717,
      "num_input_tokens_seen": 133530925,
      "step": 6292
    },
    {
      "epoch": 0.7566885107917994,
      "flos": 14514844078080.0,
      "grad_norm": 2.8576252904676314,
      "learning_rate": 5.893712707254825e-07,
      "loss": 1.0069,
      "num_input_tokens_seen": 133548105,
      "step": 6293
    },
    {
      "epoch": 0.7568087536824385,
      "flos": 13492003491840.0,
      "grad_norm": 4.568574806538627,
      "learning_rate": 5.888191688895769e-07,
      "loss": 0.8311,
      "num_input_tokens_seen": 133565085,
      "step": 6294
    },
    {
      "epoch": 0.7569289965730777,
      "flos": 10759942225920.0,
      "grad_norm": 5.268747190789212,
      "learning_rate": 5.882672811286813e-07,
      "loss": 0.7963,
      "num_input_tokens_seen": 133581085,
      "step": 6295
    },
    {
      "epoch": 0.7570492394637167,
      "flos": 14743997829120.0,
      "grad_norm": 3.570101742189564,
      "learning_rate": 5.877156075265166e-07,
      "loss": 0.8876,
      "num_input_tokens_seen": 133597070,
      "step": 6296
    },
    {
      "epoch": 0.7571694823543558,
      "flos": 11075301519360.0,
      "grad_norm": 4.878551656574842,
      "learning_rate": 5.871641481667715e-07,
      "loss": 0.8814,
      "num_input_tokens_seen": 133611235,
      "step": 6297
    },
    {
      "epoch": 0.7572897252449949,
      "flos": 18079422136320.0,
      "grad_norm": 2.75293930068271,
      "learning_rate": 5.866129031331011e-07,
      "loss": 1.0328,
      "num_input_tokens_seen": 133630610,
      "step": 6298
    },
    {
      "epoch": 0.757409968135634,
      "flos": 17267546664960.0,
      "grad_norm": 5.41703135901084,
      "learning_rate": 5.8606187250913e-07,
      "loss": 1.0122,
      "num_input_tokens_seen": 133648380,
      "step": 6299
    },
    {
      "epoch": 0.757530211026273,
      "flos": 17162968535040.0,
      "grad_norm": 3.2331458073404975,
      "learning_rate": 5.855110563784482e-07,
      "loss": 1.0219,
      "num_input_tokens_seen": 133666635,
      "step": 6300
    },
    {
      "epoch": 0.7576504539169122,
      "flos": 17032261201920.0,
      "grad_norm": 2.9770605116214814,
      "learning_rate": 5.849604548246156e-07,
      "loss": 0.8342,
      "num_input_tokens_seen": 133687465,
      "step": 6301
    },
    {
      "epoch": 0.7577706968075513,
      "flos": 15091434086400.0,
      "grad_norm": 3.7167591568140037,
      "learning_rate": 5.844100679311565e-07,
      "loss": 0.9895,
      "num_input_tokens_seen": 133706145,
      "step": 6302
    },
    {
      "epoch": 0.7578909396981903,
      "flos": 12967978475520.0,
      "grad_norm": 5.869779174394242,
      "learning_rate": 5.838598957815637e-07,
      "loss": 0.9631,
      "num_input_tokens_seen": 133723095,
      "step": 6303
    },
    {
      "epoch": 0.7580111825888295,
      "flos": 18054151372800.0,
      "grad_norm": 2.9241428651780472,
      "learning_rate": 5.833099384592996e-07,
      "loss": 1.0468,
      "num_input_tokens_seen": 133743390,
      "step": 6304
    },
    {
      "epoch": 0.7581314254794685,
      "flos": 16900542136320.0,
      "grad_norm": 4.14410622562957,
      "learning_rate": 5.827601960477913e-07,
      "loss": 0.9001,
      "num_input_tokens_seen": 133761035,
      "step": 6305
    },
    {
      "epoch": 0.7582516683701076,
      "flos": 15668790558720.0,
      "grad_norm": 3.3715233026316143,
      "learning_rate": 5.822106686304344e-07,
      "loss": 0.9079,
      "num_input_tokens_seen": 133780045,
      "step": 6306
    },
    {
      "epoch": 0.7583719112607467,
      "flos": 22564531384320.0,
      "grad_norm": 3.9535718823574317,
      "learning_rate": 5.816613562905919e-07,
      "loss": 0.7664,
      "num_input_tokens_seen": 133800950,
      "step": 6307
    },
    {
      "epoch": 0.7584921541513858,
      "flos": 23583508992000.0,
      "grad_norm": 3.7473976006237506,
      "learning_rate": 5.811122591115933e-07,
      "loss": 0.8994,
      "num_input_tokens_seen": 133821655,
      "step": 6308
    },
    {
      "epoch": 0.7586123970420249,
      "flos": 16582055669760.0,
      "grad_norm": 6.823860145933515,
      "learning_rate": 5.805633771767376e-07,
      "loss": 0.8877,
      "num_input_tokens_seen": 133838770,
      "step": 6309
    },
    {
      "epoch": 0.7587326399326639,
      "flos": 12995180728320.0,
      "grad_norm": 5.2522734306379695,
      "learning_rate": 5.800147105692888e-07,
      "loss": 0.9681,
      "num_input_tokens_seen": 133858065,
      "step": 6310
    },
    {
      "epoch": 0.7588528828233031,
      "flos": 12233846784000.0,
      "grad_norm": 3.262156857974158,
      "learning_rate": 5.794662593724795e-07,
      "loss": 0.979,
      "num_input_tokens_seen": 133876790,
      "step": 6311
    },
    {
      "epoch": 0.7589731257139422,
      "flos": 12549114101760.0,
      "grad_norm": 3.9369753533469485,
      "learning_rate": 5.789180236695091e-07,
      "loss": 0.9439,
      "num_input_tokens_seen": 133893365,
      "step": 6312
    },
    {
      "epoch": 0.7590933686045812,
      "flos": 11289424465920.0,
      "grad_norm": 4.023746100877411,
      "learning_rate": 5.78370003543544e-07,
      "loss": 1.051,
      "num_input_tokens_seen": 133911840,
      "step": 6313
    },
    {
      "epoch": 0.7592136114952204,
      "flos": 14959592386560.0,
      "grad_norm": 4.198598473447461,
      "learning_rate": 5.778221990777203e-07,
      "loss": 1.025,
      "num_input_tokens_seen": 133929300,
      "step": 6314
    },
    {
      "epoch": 0.7593338543858594,
      "flos": 17998949744640.0,
      "grad_norm": 3.7843360602771865,
      "learning_rate": 5.772746103551372e-07,
      "loss": 1.0204,
      "num_input_tokens_seen": 133944415,
      "step": 6315
    },
    {
      "epoch": 0.7594540972764985,
      "flos": 22695821230080.0,
      "grad_norm": 3.7027703610347777,
      "learning_rate": 5.767272374588648e-07,
      "loss": 0.9186,
      "num_input_tokens_seen": 133965540,
      "step": 6316
    },
    {
      "epoch": 0.7595743401671377,
      "flos": 26838951444480.0,
      "grad_norm": 4.7844413129819205,
      "learning_rate": 5.76180080471939e-07,
      "loss": 0.976,
      "num_input_tokens_seen": 133988430,
      "step": 6317
    },
    {
      "epoch": 0.7596945830577767,
      "flos": 12960405811200.0,
      "grad_norm": 9.710966961816732,
      "learning_rate": 5.756331394773631e-07,
      "loss": 0.9128,
      "num_input_tokens_seen": 134004365,
      "step": 6318
    },
    {
      "epoch": 0.7598148259484158,
      "flos": 15796554670080.0,
      "grad_norm": 2.795763700691634,
      "learning_rate": 5.750864145581071e-07,
      "loss": 0.9632,
      "num_input_tokens_seen": 134023305,
      "step": 6319
    },
    {
      "epoch": 0.7599350688390549,
      "flos": 19339050455040.0,
      "grad_norm": 5.0011852766872105,
      "learning_rate": 5.745399057971085e-07,
      "loss": 1.0471,
      "num_input_tokens_seen": 134044160,
      "step": 6320
    },
    {
      "epoch": 0.760055311729694,
      "flos": 11001819279360.0,
      "grad_norm": 5.999080587515708,
      "learning_rate": 5.739936132772738e-07,
      "loss": 0.9404,
      "num_input_tokens_seen": 134062445,
      "step": 6321
    },
    {
      "epoch": 0.760175554620333,
      "flos": 17897284177920.0,
      "grad_norm": 5.831556031270694,
      "learning_rate": 5.734475370814733e-07,
      "loss": 0.9423,
      "num_input_tokens_seen": 134081845,
      "step": 6322
    },
    {
      "epoch": 0.7602957975109722,
      "flos": 17321246023680.0,
      "grad_norm": 2.2750081408009177,
      "learning_rate": 5.729016772925483e-07,
      "loss": 0.9744,
      "num_input_tokens_seen": 134103140,
      "step": 6323
    },
    {
      "epoch": 0.7604160404016113,
      "flos": 17924087869440.0,
      "grad_norm": 5.093935910442178,
      "learning_rate": 5.723560339933038e-07,
      "loss": 0.9033,
      "num_input_tokens_seen": 134123195,
      "step": 6324
    },
    {
      "epoch": 0.7605362832922503,
      "flos": 21279601643520.0,
      "grad_norm": 4.98802238157095,
      "learning_rate": 5.71810607266513e-07,
      "loss": 0.8334,
      "num_input_tokens_seen": 134141500,
      "step": 6325
    },
    {
      "epoch": 0.7606565261828895,
      "flos": 9821069107200.0,
      "grad_norm": 3.4717252875873035,
      "learning_rate": 5.712653971949184e-07,
      "loss": 0.7918,
      "num_input_tokens_seen": 134159340,
      "step": 6326
    },
    {
      "epoch": 0.7607767690735285,
      "flos": 13150913556480.0,
      "grad_norm": 5.911062389227163,
      "learning_rate": 5.707204038612268e-07,
      "loss": 0.959,
      "num_input_tokens_seen": 134176490,
      "step": 6327
    },
    {
      "epoch": 0.7608970119641676,
      "flos": 14855443476480.0,
      "grad_norm": 4.399999635869792,
      "learning_rate": 5.701756273481138e-07,
      "loss": 0.9219,
      "num_input_tokens_seen": 134193630,
      "step": 6328
    },
    {
      "epoch": 0.7610172548548068,
      "flos": 16928602828800.0,
      "grad_norm": 2.1904859599395152,
      "learning_rate": 5.696310677382212e-07,
      "loss": 0.9248,
      "num_input_tokens_seen": 134214745,
      "step": 6329
    },
    {
      "epoch": 0.7611374977454458,
      "flos": 47613628354560.0,
      "grad_norm": 0.8307581371330164,
      "learning_rate": 5.690867251141576e-07,
      "loss": 0.8373,
      "num_input_tokens_seen": 134281120,
      "step": 6330
    },
    {
      "epoch": 0.7612577406360849,
      "flos": 11053801758720.0,
      "grad_norm": 5.051023874260022,
      "learning_rate": 5.685425995585013e-07,
      "loss": 1.1056,
      "num_input_tokens_seen": 134298765,
      "step": 6331
    },
    {
      "epoch": 0.761377983526724,
      "flos": 43324668026880.0,
      "grad_norm": 0.786100512486674,
      "learning_rate": 5.679986911537935e-07,
      "loss": 0.8121,
      "num_input_tokens_seen": 134366015,
      "step": 6332
    },
    {
      "epoch": 0.7614982264173631,
      "flos": 25527493939200.0,
      "grad_norm": 4.0718213467931275,
      "learning_rate": 5.674549999825462e-07,
      "loss": 0.8633,
      "num_input_tokens_seen": 134388550,
      "step": 6333
    },
    {
      "epoch": 0.7616184693080021,
      "flos": 48641374310400.0,
      "grad_norm": 0.9875893564478073,
      "learning_rate": 5.669115261272363e-07,
      "loss": 0.9657,
      "num_input_tokens_seen": 134448590,
      "step": 6334
    },
    {
      "epoch": 0.7617387121986413,
      "flos": 14567286435840.0,
      "grad_norm": 4.194141188908643,
      "learning_rate": 5.663682696703081e-07,
      "loss": 0.9099,
      "num_input_tokens_seen": 134466575,
      "step": 6335
    },
    {
      "epoch": 0.7618589550892804,
      "flos": 13203570524160.0,
      "grad_norm": 3.6973067662710344,
      "learning_rate": 5.658252306941746e-07,
      "loss": 1.0234,
      "num_input_tokens_seen": 134485615,
      "step": 6336
    },
    {
      "epoch": 0.7619791979799194,
      "flos": 12361733529600.0,
      "grad_norm": 4.443287044397552,
      "learning_rate": 5.65282409281212e-07,
      "loss": 0.9484,
      "num_input_tokens_seen": 134502800,
      "step": 6337
    },
    {
      "epoch": 0.7620994408705585,
      "flos": 9978886717440.0,
      "grad_norm": 3.0721532530704847,
      "learning_rate": 5.64739805513768e-07,
      "loss": 0.8842,
      "num_input_tokens_seen": 134520065,
      "step": 6338
    },
    {
      "epoch": 0.7622196837611976,
      "flos": 50638146969600.0,
      "grad_norm": 0.8153315136682017,
      "learning_rate": 5.641974194741541e-07,
      "loss": 0.7779,
      "num_input_tokens_seen": 134575470,
      "step": 6339
    },
    {
      "epoch": 0.7623399266518367,
      "flos": 43434834124800.0,
      "grad_norm": 0.7322580790153801,
      "learning_rate": 5.636552512446502e-07,
      "loss": 0.8421,
      "num_input_tokens_seen": 134636245,
      "step": 6340
    },
    {
      "epoch": 0.7624601695424758,
      "flos": 18841185300480.0,
      "grad_norm": 3.508968171395781,
      "learning_rate": 5.631133009075027e-07,
      "loss": 0.9764,
      "num_input_tokens_seen": 134655150,
      "step": 6341
    },
    {
      "epoch": 0.7625804124331149,
      "flos": 13571065589760.0,
      "grad_norm": 5.157565602168704,
      "learning_rate": 5.625715685449242e-07,
      "loss": 0.8935,
      "num_input_tokens_seen": 134672975,
      "step": 6342
    },
    {
      "epoch": 0.762700655323754,
      "flos": 18657299804160.0,
      "grad_norm": 2.167965780720095,
      "learning_rate": 5.620300542390966e-07,
      "loss": 0.9109,
      "num_input_tokens_seen": 134693740,
      "step": 6343
    },
    {
      "epoch": 0.762820898214393,
      "flos": 15904443924480.0,
      "grad_norm": 4.920927023042841,
      "learning_rate": 5.614887580721659e-07,
      "loss": 1.0384,
      "num_input_tokens_seen": 134713605,
      "step": 6344
    },
    {
      "epoch": 0.7629411411050322,
      "flos": 11101461381120.0,
      "grad_norm": 3.288576067475817,
      "learning_rate": 5.609476801262481e-07,
      "loss": 0.9238,
      "num_input_tokens_seen": 134728185,
      "step": 6345
    },
    {
      "epoch": 0.7630613839956712,
      "flos": 9714774097920.0,
      "grad_norm": 6.640450939814406,
      "learning_rate": 5.604068204834223e-07,
      "loss": 0.829,
      "num_input_tokens_seen": 134744800,
      "step": 6346
    },
    {
      "epoch": 0.7631816268863103,
      "flos": 10288696811520.0,
      "grad_norm": 4.521554821940402,
      "learning_rate": 5.598661792257367e-07,
      "loss": 0.9406,
      "num_input_tokens_seen": 134761565,
      "step": 6347
    },
    {
      "epoch": 0.7633018697769495,
      "flos": 13518132695040.0,
      "grad_norm": 3.464189848324831,
      "learning_rate": 5.593257564352071e-07,
      "loss": 0.9589,
      "num_input_tokens_seen": 134779725,
      "step": 6348
    },
    {
      "epoch": 0.7634221126675885,
      "flos": 15667380264960.0,
      "grad_norm": 6.412976710872218,
      "learning_rate": 5.58785552193815e-07,
      "loss": 0.9495,
      "num_input_tokens_seen": 134799690,
      "step": 6349
    },
    {
      "epoch": 0.7635423555582276,
      "flos": 20934618071040.0,
      "grad_norm": 3.5824150121155367,
      "learning_rate": 5.582455665835086e-07,
      "loss": 0.9462,
      "num_input_tokens_seen": 134819705,
      "step": 6350
    },
    {
      "epoch": 0.7636625984488667,
      "flos": 12600115507200.0,
      "grad_norm": 4.319404713624103,
      "learning_rate": 5.577057996862036e-07,
      "loss": 0.9163,
      "num_input_tokens_seen": 134837050,
      "step": 6351
    },
    {
      "epoch": 0.7637828413395058,
      "flos": 16876375080960.0,
      "grad_norm": 2.585631476328012,
      "learning_rate": 5.571662515837814e-07,
      "loss": 0.945,
      "num_input_tokens_seen": 134858730,
      "step": 6352
    },
    {
      "epoch": 0.7639030842301449,
      "flos": 25894559784960.0,
      "grad_norm": 2.887516639298226,
      "learning_rate": 5.566269223580926e-07,
      "loss": 1.03,
      "num_input_tokens_seen": 134880160,
      "step": 6353
    },
    {
      "epoch": 0.764023327120784,
      "flos": 20572794839040.0,
      "grad_norm": 2.552774348744979,
      "learning_rate": 5.560878120909511e-07,
      "loss": 0.9425,
      "num_input_tokens_seen": 134902480,
      "step": 6354
    },
    {
      "epoch": 0.7641435700114231,
      "flos": 46388008488960.0,
      "grad_norm": 0.9950060122514356,
      "learning_rate": 5.55548920864141e-07,
      "loss": 0.8433,
      "num_input_tokens_seen": 134962855,
      "step": 6355
    },
    {
      "epoch": 0.7642638129020621,
      "flos": 11917935636480.0,
      "grad_norm": 3.103241204336208,
      "learning_rate": 5.550102487594113e-07,
      "loss": 0.966,
      "num_input_tokens_seen": 134981245,
      "step": 6356
    },
    {
      "epoch": 0.7643840557927013,
      "flos": 21672766033920.0,
      "grad_norm": 2.8103319924967116,
      "learning_rate": 5.54471795858477e-07,
      "loss": 0.9041,
      "num_input_tokens_seen": 135001035,
      "step": 6357
    },
    {
      "epoch": 0.7645042986833404,
      "flos": 12024108011520.0,
      "grad_norm": 3.5121183412030335,
      "learning_rate": 5.539335622430235e-07,
      "loss": 1.0155,
      "num_input_tokens_seen": 135019375,
      "step": 6358
    },
    {
      "epoch": 0.7646245415739794,
      "flos": 12260006645760.0,
      "grad_norm": 5.247622905453183,
      "learning_rate": 5.533955479946975e-07,
      "loss": 0.9311,
      "num_input_tokens_seen": 135037875,
      "step": 6359
    },
    {
      "epoch": 0.7647447844646186,
      "flos": 50420858081280.0,
      "grad_norm": 0.8970390714707396,
      "learning_rate": 5.528577531951173e-07,
      "loss": 0.8646,
      "num_input_tokens_seen": 135098000,
      "step": 6360
    },
    {
      "epoch": 0.7648650273552576,
      "flos": 12521850531840.0,
      "grad_norm": 3.645625270174469,
      "learning_rate": 5.523201779258653e-07,
      "loss": 0.9266,
      "num_input_tokens_seen": 135116695,
      "step": 6361
    },
    {
      "epoch": 0.7649852702458967,
      "flos": 15745706557440.0,
      "grad_norm": 3.4190857549792963,
      "learning_rate": 5.517828222684912e-07,
      "loss": 1.0333,
      "num_input_tokens_seen": 135137070,
      "step": 6362
    },
    {
      "epoch": 0.7651055131365359,
      "flos": 50021807247360.0,
      "grad_norm": 0.9741749378132765,
      "learning_rate": 5.512456863045117e-07,
      "loss": 0.7974,
      "num_input_tokens_seen": 135197480,
      "step": 6363
    },
    {
      "epoch": 0.7652257560271749,
      "flos": 13806964224000.0,
      "grad_norm": 2.8699055191597873,
      "learning_rate": 5.507087701154089e-07,
      "loss": 0.9322,
      "num_input_tokens_seen": 135217120,
      "step": 6364
    },
    {
      "epoch": 0.765345998917814,
      "flos": 11289455124480.0,
      "grad_norm": 3.717052809723181,
      "learning_rate": 5.50172073782634e-07,
      "loss": 0.9351,
      "num_input_tokens_seen": 135234820,
      "step": 6365
    },
    {
      "epoch": 0.7654662418084531,
      "flos": 16822430453760.0,
      "grad_norm": 3.130431833185198,
      "learning_rate": 5.496355973876023e-07,
      "loss": 1.0666,
      "num_input_tokens_seen": 135253795,
      "step": 6366
    },
    {
      "epoch": 0.7655864846990922,
      "flos": 29617476648960.0,
      "grad_norm": 2.84509748456099,
      "learning_rate": 5.490993410116984e-07,
      "loss": 0.8925,
      "num_input_tokens_seen": 135276505,
      "step": 6367
    },
    {
      "epoch": 0.7657067275897312,
      "flos": 30824294707200.0,
      "grad_norm": 2.9745066193574146,
      "learning_rate": 5.485633047362704e-07,
      "loss": 0.903,
      "num_input_tokens_seen": 135298230,
      "step": 6368
    },
    {
      "epoch": 0.7658269704803703,
      "flos": 12260251914240.0,
      "grad_norm": 3.369637786008895,
      "learning_rate": 5.480274886426341e-07,
      "loss": 0.9821,
      "num_input_tokens_seen": 135314590,
      "step": 6369
    },
    {
      "epoch": 0.7659472133710095,
      "flos": 8851038781440.0,
      "grad_norm": 4.68895261509294,
      "learning_rate": 5.474918928120744e-07,
      "loss": 0.9734,
      "num_input_tokens_seen": 135330805,
      "step": 6370
    },
    {
      "epoch": 0.7660674562616485,
      "flos": 16137429995520.0,
      "grad_norm": 4.441552903732576,
      "learning_rate": 5.469565173258392e-07,
      "loss": 1.0688,
      "num_input_tokens_seen": 135349040,
      "step": 6371
    },
    {
      "epoch": 0.7661876991522876,
      "flos": 12076764979200.0,
      "grad_norm": 3.624543654557962,
      "learning_rate": 5.464213622651454e-07,
      "loss": 0.8267,
      "num_input_tokens_seen": 135366575,
      "step": 6372
    },
    {
      "epoch": 0.7663079420429267,
      "flos": 14252693606400.0,
      "grad_norm": 4.511791251784885,
      "learning_rate": 5.458864277111753e-07,
      "loss": 1.0274,
      "num_input_tokens_seen": 135384130,
      "step": 6373
    },
    {
      "epoch": 0.7664281849335658,
      "flos": 8929579683840.0,
      "grad_norm": 7.860224835235263,
      "learning_rate": 5.453517137450769e-07,
      "loss": 0.8945,
      "num_input_tokens_seen": 135400425,
      "step": 6374
    },
    {
      "epoch": 0.7665484278242048,
      "flos": 15877517598720.0,
      "grad_norm": 2.975427285932982,
      "learning_rate": 5.448172204479684e-07,
      "loss": 0.9551,
      "num_input_tokens_seen": 135419425,
      "step": 6375
    },
    {
      "epoch": 0.766668670714844,
      "flos": 16791549173760.0,
      "grad_norm": 3.3442610590756816,
      "learning_rate": 5.442829479009294e-07,
      "loss": 0.9398,
      "num_input_tokens_seen": 135437925,
      "step": 6376
    },
    {
      "epoch": 0.7667889136054831,
      "flos": 13780405800960.0,
      "grad_norm": 7.601218236359078,
      "learning_rate": 5.437488961850103e-07,
      "loss": 0.9083,
      "num_input_tokens_seen": 135457445,
      "step": 6377
    },
    {
      "epoch": 0.7669091564961221,
      "flos": 19126307143680.0,
      "grad_norm": 2.638472181736314,
      "learning_rate": 5.432150653812258e-07,
      "loss": 0.9479,
      "num_input_tokens_seen": 135477200,
      "step": 6378
    },
    {
      "epoch": 0.7670293993867613,
      "flos": 8719564984320.0,
      "grad_norm": 5.812402867972021,
      "learning_rate": 5.42681455570557e-07,
      "loss": 1.0184,
      "num_input_tokens_seen": 135493450,
      "step": 6379
    },
    {
      "epoch": 0.7671496422774003,
      "flos": 15459205079040.0,
      "grad_norm": 2.8777946489675,
      "learning_rate": 5.42148066833954e-07,
      "loss": 0.8413,
      "num_input_tokens_seen": 135512415,
      "step": 6380
    },
    {
      "epoch": 0.7672698851680394,
      "flos": 14960573460480.0,
      "grad_norm": 4.76118257177894,
      "learning_rate": 5.416148992523289e-07,
      "loss": 0.9426,
      "num_input_tokens_seen": 135530710,
      "step": 6381
    },
    {
      "epoch": 0.7673901280586786,
      "flos": 12018528153600.0,
      "grad_norm": 3.531953724645484,
      "learning_rate": 5.410819529065644e-07,
      "loss": 0.9774,
      "num_input_tokens_seen": 135548385,
      "step": 6382
    },
    {
      "epoch": 0.7675103709493176,
      "flos": 20834761359360.0,
      "grad_norm": 3.413835950095587,
      "learning_rate": 5.405492278775079e-07,
      "loss": 0.8396,
      "num_input_tokens_seen": 135567885,
      "step": 6383
    },
    {
      "epoch": 0.7676306138399567,
      "flos": 20676790456320.0,
      "grad_norm": 4.145363683632604,
      "learning_rate": 5.400167242459732e-07,
      "loss": 0.9833,
      "num_input_tokens_seen": 135586565,
      "step": 6384
    },
    {
      "epoch": 0.7677508567305958,
      "flos": 16035641794560.0,
      "grad_norm": 2.9158395048433943,
      "learning_rate": 5.394844420927405e-07,
      "loss": 1.0016,
      "num_input_tokens_seen": 135605895,
      "step": 6385
    },
    {
      "epoch": 0.7678710996212349,
      "flos": 18081292308480.0,
      "grad_norm": 3.803194791079795,
      "learning_rate": 5.389523814985562e-07,
      "loss": 0.9119,
      "num_input_tokens_seen": 135625035,
      "step": 6386
    },
    {
      "epoch": 0.767991342511874,
      "flos": 19047582289920.0,
      "grad_norm": 4.288249732659389,
      "learning_rate": 5.384205425441344e-07,
      "loss": 0.9531,
      "num_input_tokens_seen": 135645665,
      "step": 6387
    },
    {
      "epoch": 0.7681115854025131,
      "flos": 18762583080960.0,
      "grad_norm": 2.5294506129374805,
      "learning_rate": 5.378889253101537e-07,
      "loss": 1.0326,
      "num_input_tokens_seen": 135665940,
      "step": 6388
    },
    {
      "epoch": 0.7682318282931522,
      "flos": 16532740485120.0,
      "grad_norm": 3.756954864531377,
      "learning_rate": 5.373575298772617e-07,
      "loss": 0.9949,
      "num_input_tokens_seen": 135684780,
      "step": 6389
    },
    {
      "epoch": 0.7683520711837912,
      "flos": 50183579811840.0,
      "grad_norm": 0.7494822940580019,
      "learning_rate": 5.368263563260689e-07,
      "loss": 0.8158,
      "num_input_tokens_seen": 135749635,
      "step": 6390
    },
    {
      "epoch": 0.7684723140744304,
      "flos": 13203907768320.0,
      "grad_norm": 2.993933424775799,
      "learning_rate": 5.362954047371537e-07,
      "loss": 0.8232,
      "num_input_tokens_seen": 135768465,
      "step": 6391
    },
    {
      "epoch": 0.7685925569650695,
      "flos": 19548298690560.0,
      "grad_norm": 3.6870616070344395,
      "learning_rate": 5.357646751910627e-07,
      "loss": 0.9202,
      "num_input_tokens_seen": 135789365,
      "step": 6392
    },
    {
      "epoch": 0.7687127998557085,
      "flos": 17451922698240.0,
      "grad_norm": 3.8556282914843756,
      "learning_rate": 5.352341677683061e-07,
      "loss": 0.9891,
      "num_input_tokens_seen": 135810385,
      "step": 6393
    },
    {
      "epoch": 0.7688330427463477,
      "flos": 17897682739200.0,
      "grad_norm": 6.104195899589594,
      "learning_rate": 5.347038825493617e-07,
      "loss": 0.9765,
      "num_input_tokens_seen": 135831635,
      "step": 6394
    },
    {
      "epoch": 0.7689532856369867,
      "flos": 15062147051520.0,
      "grad_norm": 3.101780804463066,
      "learning_rate": 5.341738196146732e-07,
      "loss": 0.8578,
      "num_input_tokens_seen": 135849700,
      "step": 6395
    },
    {
      "epoch": 0.7690735285276258,
      "flos": 17871124316160.0,
      "grad_norm": 15.534157801237567,
      "learning_rate": 5.336439790446503e-07,
      "loss": 0.9117,
      "num_input_tokens_seen": 135868520,
      "step": 6396
    },
    {
      "epoch": 0.769193771418265,
      "flos": 39161801809920.0,
      "grad_norm": 3.2357901458124285,
      "learning_rate": 5.331143609196711e-07,
      "loss": 0.8125,
      "num_input_tokens_seen": 135892055,
      "step": 6397
    },
    {
      "epoch": 0.769314014308904,
      "flos": 26655433850880.0,
      "grad_norm": 3.10432338105662,
      "learning_rate": 5.325849653200758e-07,
      "loss": 0.9699,
      "num_input_tokens_seen": 135915725,
      "step": 6398
    },
    {
      "epoch": 0.7694342571995431,
      "flos": 14646164582400.0,
      "grad_norm": 4.784125367273014,
      "learning_rate": 5.32055792326175e-07,
      "loss": 0.9528,
      "num_input_tokens_seen": 135933870,
      "step": 6399
    },
    {
      "epoch": 0.7695545000901821,
      "flos": 17216238673920.0,
      "grad_norm": 3.1488956089736253,
      "learning_rate": 5.315268420182437e-07,
      "loss": 0.9166,
      "num_input_tokens_seen": 135952265,
      "step": 6400
    },
    {
      "epoch": 0.7696747429808213,
      "flos": 19942474813440.0,
      "grad_norm": 3.7222784917446705,
      "learning_rate": 5.309981144765221e-07,
      "loss": 0.9502,
      "num_input_tokens_seen": 135972130,
      "step": 6401
    },
    {
      "epoch": 0.7697949858714603,
      "flos": 8090164715520.0,
      "grad_norm": 3.4515237677401127,
      "learning_rate": 5.304696097812196e-07,
      "loss": 0.9403,
      "num_input_tokens_seen": 135988450,
      "step": 6402
    },
    {
      "epoch": 0.7699152287620994,
      "flos": 18998910935040.0,
      "grad_norm": 7.363821318473201,
      "learning_rate": 5.299413280125078e-07,
      "loss": 0.7844,
      "num_input_tokens_seen": 136006480,
      "step": 6403
    },
    {
      "epoch": 0.7700354716527386,
      "flos": 11709055303680.0,
      "grad_norm": 4.591814528690543,
      "learning_rate": 5.294132692505284e-07,
      "loss": 0.9227,
      "num_input_tokens_seen": 136024610,
      "step": 6404
    },
    {
      "epoch": 0.7701557145433776,
      "flos": 13647613685760.0,
      "grad_norm": 3.7401979130892458,
      "learning_rate": 5.288854335753861e-07,
      "loss": 0.9676,
      "num_input_tokens_seen": 136042590,
      "step": 6405
    },
    {
      "epoch": 0.7702759574340167,
      "flos": 22590354001920.0,
      "grad_norm": 3.416975720701963,
      "learning_rate": 5.283578210671551e-07,
      "loss": 0.9526,
      "num_input_tokens_seen": 136064550,
      "step": 6406
    },
    {
      "epoch": 0.7703962003246558,
      "flos": 11892388945920.0,
      "grad_norm": 5.242575345423412,
      "learning_rate": 5.278304318058719e-07,
      "loss": 0.9471,
      "num_input_tokens_seen": 136082125,
      "step": 6407
    },
    {
      "epoch": 0.7705164432152949,
      "flos": 25501303418880.0,
      "grad_norm": 2.9828522477216803,
      "learning_rate": 5.273032658715411e-07,
      "loss": 0.9828,
      "num_input_tokens_seen": 136104655,
      "step": 6408
    },
    {
      "epoch": 0.7706366861059339,
      "flos": 16610944143360.0,
      "grad_norm": 4.1061606944691915,
      "learning_rate": 5.267763233441347e-07,
      "loss": 0.961,
      "num_input_tokens_seen": 136125005,
      "step": 6409
    },
    {
      "epoch": 0.7707569289965731,
      "flos": 16297792266240.0,
      "grad_norm": 4.0055865376673845,
      "learning_rate": 5.26249604303588e-07,
      "loss": 0.8812,
      "num_input_tokens_seen": 136143230,
      "step": 6410
    },
    {
      "epoch": 0.7708771718872122,
      "flos": 12338424913920.0,
      "grad_norm": 4.286777464372595,
      "learning_rate": 5.257231088298057e-07,
      "loss": 0.9705,
      "num_input_tokens_seen": 136161360,
      "step": 6411
    },
    {
      "epoch": 0.7709974147778512,
      "flos": 51739559116800.0,
      "grad_norm": 0.8556965215534937,
      "learning_rate": 5.25196837002655e-07,
      "loss": 0.7568,
      "num_input_tokens_seen": 136220790,
      "step": 6412
    },
    {
      "epoch": 0.7711176576684904,
      "flos": 28464971120640.0,
      "grad_norm": 3.6872907676173377,
      "learning_rate": 5.24670788901971e-07,
      "loss": 0.8928,
      "num_input_tokens_seen": 136243600,
      "step": 6413
    },
    {
      "epoch": 0.7712379005591294,
      "flos": 26392670208000.0,
      "grad_norm": 4.655719496441019,
      "learning_rate": 5.241449646075557e-07,
      "loss": 0.8637,
      "num_input_tokens_seen": 136266545,
      "step": 6414
    },
    {
      "epoch": 0.7713581434497685,
      "flos": 16186775838720.0,
      "grad_norm": 7.545354148890411,
      "learning_rate": 5.236193641991762e-07,
      "loss": 0.9131,
      "num_input_tokens_seen": 136284195,
      "step": 6415
    },
    {
      "epoch": 0.7714783863404077,
      "flos": 17136900648960.0,
      "grad_norm": 3.5430388690777033,
      "learning_rate": 5.23093987756565e-07,
      "loss": 0.8936,
      "num_input_tokens_seen": 136302610,
      "step": 6416
    },
    {
      "epoch": 0.7715986292310467,
      "flos": 14955821383680.0,
      "grad_norm": 5.247737760423329,
      "learning_rate": 5.225688353594217e-07,
      "loss": 0.9501,
      "num_input_tokens_seen": 136321960,
      "step": 6417
    },
    {
      "epoch": 0.7717188721216858,
      "flos": 14619667476480.0,
      "grad_norm": 4.490425945375164,
      "learning_rate": 5.220439070874108e-07,
      "loss": 0.9613,
      "num_input_tokens_seen": 136340920,
      "step": 6418
    },
    {
      "epoch": 0.7718391150123249,
      "flos": 18684686008320.0,
      "grad_norm": 2.917801245767478,
      "learning_rate": 5.215192030201652e-07,
      "loss": 0.9026,
      "num_input_tokens_seen": 136361630,
      "step": 6419
    },
    {
      "epoch": 0.771959357902964,
      "flos": 15664651653120.0,
      "grad_norm": 3.0438969069478934,
      "learning_rate": 5.209947232372798e-07,
      "loss": 1.0524,
      "num_input_tokens_seen": 136378840,
      "step": 6420
    },
    {
      "epoch": 0.772079600793603,
      "flos": 21699477749760.0,
      "grad_norm": 2.9002416707331866,
      "learning_rate": 5.204704678183196e-07,
      "loss": 0.996,
      "num_input_tokens_seen": 136397295,
      "step": 6421
    },
    {
      "epoch": 0.7721998436842422,
      "flos": 9139778334720.0,
      "grad_norm": 4.222016376920784,
      "learning_rate": 5.19946436842813e-07,
      "loss": 1.0446,
      "num_input_tokens_seen": 136414145,
      "step": 6422
    },
    {
      "epoch": 0.7723200865748813,
      "flos": 23273085726720.0,
      "grad_norm": 3.217881465134666,
      "learning_rate": 5.194226303902546e-07,
      "loss": 0.8829,
      "num_input_tokens_seen": 136433600,
      "step": 6423
    },
    {
      "epoch": 0.7724403294655203,
      "flos": 14986794639360.0,
      "grad_norm": 2.412991184178993,
      "learning_rate": 5.188990485401072e-07,
      "loss": 0.8949,
      "num_input_tokens_seen": 136452525,
      "step": 6424
    },
    {
      "epoch": 0.7725605723561595,
      "flos": 15694705152000.0,
      "grad_norm": 4.607266425485428,
      "learning_rate": 5.183756913717954e-07,
      "loss": 1.0471,
      "num_input_tokens_seen": 136472020,
      "step": 6425
    },
    {
      "epoch": 0.7726808152467985,
      "flos": 24609813995520.0,
      "grad_norm": 3.1364216931062225,
      "learning_rate": 5.178525589647136e-07,
      "loss": 0.9251,
      "num_input_tokens_seen": 136493380,
      "step": 6426
    },
    {
      "epoch": 0.7728010581374376,
      "flos": 15849211637760.0,
      "grad_norm": 3.8339346262055094,
      "learning_rate": 5.173296513982197e-07,
      "loss": 0.9751,
      "num_input_tokens_seen": 136511625,
      "step": 6427
    },
    {
      "epoch": 0.7729213010280768,
      "flos": 19313871667200.0,
      "grad_norm": 4.210580106441827,
      "learning_rate": 5.168069687516398e-07,
      "loss": 0.8428,
      "num_input_tokens_seen": 136531115,
      "step": 6428
    },
    {
      "epoch": 0.7730415439187158,
      "flos": 12862725857280.0,
      "grad_norm": 3.329641936700663,
      "learning_rate": 5.16284511104263e-07,
      "loss": 0.9012,
      "num_input_tokens_seen": 136549970,
      "step": 6429
    },
    {
      "epoch": 0.7731617868093549,
      "flos": 8404910837760.0,
      "grad_norm": 6.011103211258833,
      "learning_rate": 5.157622785353457e-07,
      "loss": 0.9738,
      "num_input_tokens_seen": 136567805,
      "step": 6430
    },
    {
      "epoch": 0.7732820296999939,
      "flos": 45964269404160.0,
      "grad_norm": 0.6969324592884923,
      "learning_rate": 5.152402711241113e-07,
      "loss": 0.8106,
      "num_input_tokens_seen": 136635430,
      "step": 6431
    },
    {
      "epoch": 0.7734022725906331,
      "flos": 18002812723200.0,
      "grad_norm": 3.4178804397073748,
      "learning_rate": 5.147184889497465e-07,
      "loss": 1.0202,
      "num_input_tokens_seen": 136654620,
      "step": 6432
    },
    {
      "epoch": 0.7735225154812722,
      "flos": 12285767946240.0,
      "grad_norm": 4.638256560355166,
      "learning_rate": 5.141969320914072e-07,
      "loss": 0.9916,
      "num_input_tokens_seen": 136671845,
      "step": 6433
    },
    {
      "epoch": 0.7736427583719112,
      "flos": 23268456284160.0,
      "grad_norm": 5.959104403596966,
      "learning_rate": 5.136756006282113e-07,
      "loss": 0.8112,
      "num_input_tokens_seen": 136690230,
      "step": 6434
    },
    {
      "epoch": 0.7737630012625504,
      "flos": 14090890383360.0,
      "grad_norm": 5.75639849708362,
      "learning_rate": 5.131544946392446e-07,
      "loss": 1.0517,
      "num_input_tokens_seen": 136705230,
      "step": 6435
    },
    {
      "epoch": 0.7738832441531894,
      "flos": 25706535383040.0,
      "grad_norm": 6.005225131652929,
      "learning_rate": 5.126336142035592e-07,
      "loss": 0.8317,
      "num_input_tokens_seen": 136724985,
      "step": 6436
    },
    {
      "epoch": 0.7740034870438285,
      "flos": 9452378357760.0,
      "grad_norm": 5.586187439775422,
      "learning_rate": 5.121129594001721e-07,
      "loss": 0.9195,
      "num_input_tokens_seen": 136738970,
      "step": 6437
    },
    {
      "epoch": 0.7741237299344677,
      "flos": 15691271393280.0,
      "grad_norm": 2.8473052552890907,
      "learning_rate": 5.115925303080661e-07,
      "loss": 1.0067,
      "num_input_tokens_seen": 136758400,
      "step": 6438
    },
    {
      "epoch": 0.7742439728251067,
      "flos": 14094906654720.0,
      "grad_norm": 3.5315324788496456,
      "learning_rate": 5.110723270061899e-07,
      "loss": 0.9757,
      "num_input_tokens_seen": 136774610,
      "step": 6439
    },
    {
      "epoch": 0.7743642157157458,
      "flos": 11813326848000.0,
      "grad_norm": 2.940675844894104,
      "learning_rate": 5.105523495734572e-07,
      "loss": 0.9917,
      "num_input_tokens_seen": 136791730,
      "step": 6440
    },
    {
      "epoch": 0.7744844586063849,
      "flos": 14411063070720.0,
      "grad_norm": 2.716045130285901,
      "learning_rate": 5.100325980887499e-07,
      "loss": 0.9301,
      "num_input_tokens_seen": 136811375,
      "step": 6441
    },
    {
      "epoch": 0.774604701497024,
      "flos": 16324289372160.0,
      "grad_norm": 12.107161466035235,
      "learning_rate": 5.095130726309116e-07,
      "loss": 1.0155,
      "num_input_tokens_seen": 136831270,
      "step": 6442
    },
    {
      "epoch": 0.774724944387663,
      "flos": 43151275868160.0,
      "grad_norm": 2.364627070309529,
      "learning_rate": 5.089937732787559e-07,
      "loss": 0.8556,
      "num_input_tokens_seen": 136895550,
      "step": 6443
    },
    {
      "epoch": 0.7748451872783022,
      "flos": 19051169341440.0,
      "grad_norm": 3.984374760646439,
      "learning_rate": 5.084747001110592e-07,
      "loss": 0.8543,
      "num_input_tokens_seen": 136914895,
      "step": 6444
    },
    {
      "epoch": 0.7749654301689413,
      "flos": 21622592409600.0,
      "grad_norm": 4.793838185838353,
      "learning_rate": 5.07955853206564e-07,
      "loss": 0.8902,
      "num_input_tokens_seen": 136939320,
      "step": 6445
    },
    {
      "epoch": 0.7750856730595803,
      "flos": 30851098398720.0,
      "grad_norm": 15.454543863388224,
      "learning_rate": 5.074372326439807e-07,
      "loss": 0.9004,
      "num_input_tokens_seen": 136962050,
      "step": 6446
    },
    {
      "epoch": 0.7752059159502195,
      "flos": 12496150548480.0,
      "grad_norm": 3.3210620707219385,
      "learning_rate": 5.069188385019814e-07,
      "loss": 0.9189,
      "num_input_tokens_seen": 136979470,
      "step": 6447
    },
    {
      "epoch": 0.7753261588408585,
      "flos": 8929763635200.0,
      "grad_norm": 4.75975620799489,
      "learning_rate": 5.064006708592077e-07,
      "loss": 0.7913,
      "num_input_tokens_seen": 136995435,
      "step": 6448
    },
    {
      "epoch": 0.7754464017314976,
      "flos": 11813817384960.0,
      "grad_norm": 4.6251024801911695,
      "learning_rate": 5.058827297942641e-07,
      "loss": 0.9441,
      "num_input_tokens_seen": 137010260,
      "step": 6449
    },
    {
      "epoch": 0.7755666446221368,
      "flos": 13990113914880.0,
      "grad_norm": 4.364146475525182,
      "learning_rate": 5.053650153857237e-07,
      "loss": 0.921,
      "num_input_tokens_seen": 137028990,
      "step": 6450
    },
    {
      "epoch": 0.7756868875127758,
      "flos": 13252732416000.0,
      "grad_norm": 4.362688015709135,
      "learning_rate": 5.048475277121214e-07,
      "loss": 0.8999,
      "num_input_tokens_seen": 137045925,
      "step": 6451
    },
    {
      "epoch": 0.7758071304034149,
      "flos": 20232256757760.0,
      "grad_norm": 3.2255371533549786,
      "learning_rate": 5.043302668519598e-07,
      "loss": 0.9626,
      "num_input_tokens_seen": 137064980,
      "step": 6452
    },
    {
      "epoch": 0.775927373294054,
      "flos": 14620066037760.0,
      "grad_norm": 2.6685220303868875,
      "learning_rate": 5.038132328837079e-07,
      "loss": 0.9163,
      "num_input_tokens_seen": 137083090,
      "step": 6453
    },
    {
      "epoch": 0.7760476161846931,
      "flos": 16007489126400.0,
      "grad_norm": 4.184149697173189,
      "learning_rate": 5.032964258857993e-07,
      "loss": 0.9341,
      "num_input_tokens_seen": 137102905,
      "step": 6454
    },
    {
      "epoch": 0.7761678590753321,
      "flos": 34783294156800.0,
      "grad_norm": 2.627805300759543,
      "learning_rate": 5.027798459366329e-07,
      "loss": 0.8792,
      "num_input_tokens_seen": 137127990,
      "step": 6455
    },
    {
      "epoch": 0.7762881019659713,
      "flos": 18631446528000.0,
      "grad_norm": 3.3436488911681006,
      "learning_rate": 5.02263493114573e-07,
      "loss": 0.8324,
      "num_input_tokens_seen": 137149505,
      "step": 6456
    },
    {
      "epoch": 0.7764083448566104,
      "flos": 14614087618560.0,
      "grad_norm": 5.204716069725137,
      "learning_rate": 5.017473674979502e-07,
      "loss": 0.949,
      "num_input_tokens_seen": 137165250,
      "step": 6457
    },
    {
      "epoch": 0.7765285877472494,
      "flos": 48185221017600.0,
      "grad_norm": 0.8489822353126506,
      "learning_rate": 5.01231469165061e-07,
      "loss": 0.7907,
      "num_input_tokens_seen": 137220795,
      "step": 6458
    },
    {
      "epoch": 0.7766488306378886,
      "flos": 43909881200640.0,
      "grad_norm": 0.9368666098678956,
      "learning_rate": 5.007157981941663e-07,
      "loss": 0.8082,
      "num_input_tokens_seen": 137285875,
      "step": 6459
    },
    {
      "epoch": 0.7767690735285276,
      "flos": 45060516556800.0,
      "grad_norm": 0.8811185671143581,
      "learning_rate": 5.002003546634928e-07,
      "loss": 0.8912,
      "num_input_tokens_seen": 137341695,
      "step": 6460
    },
    {
      "epoch": 0.7768893164191667,
      "flos": 14829130321920.0,
      "grad_norm": 3.6383754345328776,
      "learning_rate": 4.996851386512331e-07,
      "loss": 0.9569,
      "num_input_tokens_seen": 137360120,
      "step": 6461
    },
    {
      "epoch": 0.7770095593098058,
      "flos": 14698545623040.0,
      "grad_norm": 5.6356001264097895,
      "learning_rate": 4.991701502355444e-07,
      "loss": 1.0233,
      "num_input_tokens_seen": 137380305,
      "step": 6462
    },
    {
      "epoch": 0.7771298022004449,
      "flos": 17582660689920.0,
      "grad_norm": 8.93292887706792,
      "learning_rate": 4.986553894945518e-07,
      "loss": 0.965,
      "num_input_tokens_seen": 137401235,
      "step": 6463
    },
    {
      "epoch": 0.777250045091084,
      "flos": 17792123535360.0,
      "grad_norm": 3.0760987094473533,
      "learning_rate": 4.981408565063416e-07,
      "loss": 1.0665,
      "num_input_tokens_seen": 137420900,
      "step": 6464
    },
    {
      "epoch": 0.777370287981723,
      "flos": 14278209638400.0,
      "grad_norm": 8.51414927723487,
      "learning_rate": 4.976265513489701e-07,
      "loss": 0.9462,
      "num_input_tokens_seen": 137440590,
      "step": 6465
    },
    {
      "epoch": 0.7774905308723622,
      "flos": 15426453626880.0,
      "grad_norm": 3.8360361991662013,
      "learning_rate": 4.971124741004562e-07,
      "loss": 1.0096,
      "num_input_tokens_seen": 137459310,
      "step": 6466
    },
    {
      "epoch": 0.7776107737630013,
      "flos": 11341928140800.0,
      "grad_norm": 3.915999430102111,
      "learning_rate": 4.965986248387846e-07,
      "loss": 0.955,
      "num_input_tokens_seen": 137477345,
      "step": 6467
    },
    {
      "epoch": 0.7777310166536403,
      "flos": 17635286999040.0,
      "grad_norm": 2.8123263835301904,
      "learning_rate": 4.960850036419073e-07,
      "loss": 0.9655,
      "num_input_tokens_seen": 137496165,
      "step": 6468
    },
    {
      "epoch": 0.7778512595442795,
      "flos": 12231670026240.0,
      "grad_norm": 3.45129705179396,
      "learning_rate": 4.955716105877378e-07,
      "loss": 0.9785,
      "num_input_tokens_seen": 137514655,
      "step": 6469
    },
    {
      "epoch": 0.7779715024349185,
      "flos": 12573894328320.0,
      "grad_norm": 3.2558024034738824,
      "learning_rate": 4.950584457541598e-07,
      "loss": 1.0319,
      "num_input_tokens_seen": 137532840,
      "step": 6470
    },
    {
      "epoch": 0.7780917453255576,
      "flos": 17294656942080.0,
      "grad_norm": 1.856402481443871,
      "learning_rate": 4.945455092190183e-07,
      "loss": 1.02,
      "num_input_tokens_seen": 137553815,
      "step": 6471
    },
    {
      "epoch": 0.7782119882161967,
      "flos": 40333460828160.0,
      "grad_norm": 0.7088568706773863,
      "learning_rate": 4.940328010601271e-07,
      "loss": 0.7699,
      "num_input_tokens_seen": 137618450,
      "step": 6472
    },
    {
      "epoch": 0.7783322311068358,
      "flos": 33445370204160.0,
      "grad_norm": 3.3552835788636832,
      "learning_rate": 4.935203213552621e-07,
      "loss": 0.9551,
      "num_input_tokens_seen": 137641910,
      "step": 6473
    },
    {
      "epoch": 0.7784524739974749,
      "flos": 13514423009280.0,
      "grad_norm": 7.045487700801036,
      "learning_rate": 4.930080701821662e-07,
      "loss": 0.8502,
      "num_input_tokens_seen": 137659095,
      "step": 6474
    },
    {
      "epoch": 0.778572716888114,
      "flos": 17635470950400.0,
      "grad_norm": 6.445448921378223,
      "learning_rate": 4.92496047618548e-07,
      "loss": 0.9606,
      "num_input_tokens_seen": 137678575,
      "step": 6475
    },
    {
      "epoch": 0.7786929597787531,
      "flos": 14247450992640.0,
      "grad_norm": 3.1315416150776105,
      "learning_rate": 4.919842537420811e-07,
      "loss": 0.9716,
      "num_input_tokens_seen": 137695410,
      "step": 6476
    },
    {
      "epoch": 0.7788132026693921,
      "flos": 15537715322880.0,
      "grad_norm": 2.710225388279255,
      "learning_rate": 4.91472688630404e-07,
      "loss": 0.9858,
      "num_input_tokens_seen": 137715870,
      "step": 6477
    },
    {
      "epoch": 0.7789334455600313,
      "flos": 7854817935360.0,
      "grad_norm": 2.553578453930805,
      "learning_rate": 4.909613523611202e-07,
      "loss": 0.9335,
      "num_input_tokens_seen": 137732470,
      "step": 6478
    },
    {
      "epoch": 0.7790536884506704,
      "flos": 20441474334720.0,
      "grad_norm": 3.2268295362314996,
      "learning_rate": 4.904502450117991e-07,
      "loss": 0.9315,
      "num_input_tokens_seen": 137753150,
      "step": 6479
    },
    {
      "epoch": 0.7791739313413094,
      "flos": 7776001105920.0,
      "grad_norm": 3.9319764404600037,
      "learning_rate": 4.899393666599762e-07,
      "loss": 0.9103,
      "num_input_tokens_seen": 137769445,
      "step": 6480
    },
    {
      "epoch": 0.7792941742319486,
      "flos": 10367574958080.0,
      "grad_norm": 2.9679639880008355,
      "learning_rate": 4.894287173831506e-07,
      "loss": 0.9134,
      "num_input_tokens_seen": 137785125,
      "step": 6481
    },
    {
      "epoch": 0.7794144171225876,
      "flos": 16534304071680.0,
      "grad_norm": 2.9800802755104314,
      "learning_rate": 4.889182972587877e-07,
      "loss": 1.0342,
      "num_input_tokens_seen": 137804140,
      "step": 6482
    },
    {
      "epoch": 0.7795346600132267,
      "flos": 15275258265600.0,
      "grad_norm": 3.9884731384803627,
      "learning_rate": 4.884081063643177e-07,
      "loss": 0.858,
      "num_input_tokens_seen": 137822520,
      "step": 6483
    },
    {
      "epoch": 0.7796549029038659,
      "flos": 50166349701120.0,
      "grad_norm": 1.5203228533589286,
      "learning_rate": 4.878981447771353e-07,
      "loss": 0.7586,
      "num_input_tokens_seen": 137876620,
      "step": 6484
    },
    {
      "epoch": 0.7797751457945049,
      "flos": 17059432796160.0,
      "grad_norm": 3.669328041614369,
      "learning_rate": 4.873884125746035e-07,
      "loss": 0.9281,
      "num_input_tokens_seen": 137898015,
      "step": 6485
    },
    {
      "epoch": 0.779895388685144,
      "flos": 16114458624000.0,
      "grad_norm": 5.838414985128602,
      "learning_rate": 4.868789098340456e-07,
      "loss": 0.9073,
      "num_input_tokens_seen": 137915640,
      "step": 6486
    },
    {
      "epoch": 0.7800156315757831,
      "flos": 16900572794880.0,
      "grad_norm": 4.2732455752720355,
      "learning_rate": 4.863696366327543e-07,
      "loss": 0.9075,
      "num_input_tokens_seen": 137934530,
      "step": 6487
    },
    {
      "epoch": 0.7801358744664222,
      "flos": 18812664729600.0,
      "grad_norm": 3.5109323336024656,
      "learning_rate": 4.85860593047986e-07,
      "loss": 0.9815,
      "num_input_tokens_seen": 137954315,
      "step": 6488
    },
    {
      "epoch": 0.7802561173570612,
      "flos": 18735626096640.0,
      "grad_norm": 3.029500239990346,
      "learning_rate": 4.853517791569613e-07,
      "loss": 0.9339,
      "num_input_tokens_seen": 137976215,
      "step": 6489
    },
    {
      "epoch": 0.7803763602477004,
      "flos": 28805631836160.0,
      "grad_norm": 3.542406266721577,
      "learning_rate": 4.848431950368684e-07,
      "loss": 0.8521,
      "num_input_tokens_seen": 137998495,
      "step": 6490
    },
    {
      "epoch": 0.7804966031383395,
      "flos": 47975666196480.0,
      "grad_norm": 0.7305076711148749,
      "learning_rate": 4.843348407648569e-07,
      "loss": 0.7756,
      "num_input_tokens_seen": 138059495,
      "step": 6491
    },
    {
      "epoch": 0.7806168460289785,
      "flos": 12567333396480.0,
      "grad_norm": 3.2527688776016275,
      "learning_rate": 4.838267164180457e-07,
      "loss": 1.0187,
      "num_input_tokens_seen": 138074885,
      "step": 6492
    },
    {
      "epoch": 0.7807370889196176,
      "flos": 17027509125120.0,
      "grad_norm": 4.716862180901427,
      "learning_rate": 4.833188220735156e-07,
      "loss": 1.03,
      "num_input_tokens_seen": 138094275,
      "step": 6493
    },
    {
      "epoch": 0.7808573318102567,
      "flos": 13465506385920.0,
      "grad_norm": 3.253504404302148,
      "learning_rate": 4.828111578083152e-07,
      "loss": 0.9372,
      "num_input_tokens_seen": 138110900,
      "step": 6494
    },
    {
      "epoch": 0.7809775747008958,
      "flos": 17052044083200.0,
      "grad_norm": 5.207767832255748,
      "learning_rate": 4.823037236994556e-07,
      "loss": 1.0008,
      "num_input_tokens_seen": 138128785,
      "step": 6495
    },
    {
      "epoch": 0.7810978175915348,
      "flos": 49078703247360.0,
      "grad_norm": 0.8078005673617368,
      "learning_rate": 4.817965198239136e-07,
      "loss": 0.7843,
      "num_input_tokens_seen": 138194965,
      "step": 6496
    },
    {
      "epoch": 0.781218060482174,
      "flos": 13934391091200.0,
      "grad_norm": 18.12096281142066,
      "learning_rate": 4.812895462586331e-07,
      "loss": 0.9133,
      "num_input_tokens_seen": 138212510,
      "step": 6497
    },
    {
      "epoch": 0.7813383033728131,
      "flos": 18236503941120.0,
      "grad_norm": 3.8911125153425807,
      "learning_rate": 4.807828030805207e-07,
      "loss": 1.0112,
      "num_input_tokens_seen": 138231220,
      "step": 6498
    },
    {
      "epoch": 0.7814585462634521,
      "flos": 14541433159680.0,
      "grad_norm": 9.592160170845984,
      "learning_rate": 4.802762903664495e-07,
      "loss": 0.8731,
      "num_input_tokens_seen": 138250120,
      "step": 6499
    },
    {
      "epoch": 0.7815787891540913,
      "flos": 15847464099840.0,
      "grad_norm": 4.188445710388529,
      "learning_rate": 4.797700081932565e-07,
      "loss": 0.908,
      "num_input_tokens_seen": 138267705,
      "step": 6500
    },
    {
      "epoch": 0.7816990320447303,
      "flos": 16061065850880.0,
      "grad_norm": 4.962395977784525,
      "learning_rate": 4.792639566377442e-07,
      "loss": 1.0126,
      "num_input_tokens_seen": 138284835,
      "step": 6501
    },
    {
      "epoch": 0.7818192749353694,
      "flos": 17739435909120.0,
      "grad_norm": 3.824152290304469,
      "learning_rate": 4.78758135776681e-07,
      "loss": 0.9696,
      "num_input_tokens_seen": 138304410,
      "step": 6502
    },
    {
      "epoch": 0.7819395178260086,
      "flos": 16875056762880.0,
      "grad_norm": 3.7036935876778805,
      "learning_rate": 4.782525456867989e-07,
      "loss": 0.9758,
      "num_input_tokens_seen": 138322985,
      "step": 6503
    },
    {
      "epoch": 0.7820597607166476,
      "flos": 16507469721600.0,
      "grad_norm": 3.0126775855434706,
      "learning_rate": 4.777471864447959e-07,
      "loss": 1.0229,
      "num_input_tokens_seen": 138343445,
      "step": 6504
    },
    {
      "epoch": 0.7821800036072867,
      "flos": 15851940249600.0,
      "grad_norm": 3.160658976738289,
      "learning_rate": 4.772420581273344e-07,
      "loss": 0.9873,
      "num_input_tokens_seen": 138360650,
      "step": 6505
    },
    {
      "epoch": 0.7823002464979258,
      "flos": 15302000640000.0,
      "grad_norm": 4.349318852205413,
      "learning_rate": 4.7673716081104134e-07,
      "loss": 0.9615,
      "num_input_tokens_seen": 138380545,
      "step": 6506
    },
    {
      "epoch": 0.7824204893885649,
      "flos": 17451861381120.0,
      "grad_norm": 2.914267525095235,
      "learning_rate": 4.762324945725109e-07,
      "loss": 1.0331,
      "num_input_tokens_seen": 138399710,
      "step": 6507
    },
    {
      "epoch": 0.782540732279204,
      "flos": 19520973803520.0,
      "grad_norm": 2.8431622243588204,
      "learning_rate": 4.7572805948829844e-07,
      "loss": 0.9504,
      "num_input_tokens_seen": 138419690,
      "step": 6508
    },
    {
      "epoch": 0.7826609751698431,
      "flos": 17320755486720.0,
      "grad_norm": 3.7787401681381563,
      "learning_rate": 4.7522385563492795e-07,
      "loss": 0.8981,
      "num_input_tokens_seen": 138439710,
      "step": 6509
    },
    {
      "epoch": 0.7827812180604822,
      "flos": 17058696990720.0,
      "grad_norm": 3.116842685835631,
      "learning_rate": 4.747198830888863e-07,
      "loss": 0.8784,
      "num_input_tokens_seen": 138459300,
      "step": 6510
    },
    {
      "epoch": 0.7829014609511212,
      "flos": 19549433057280.0,
      "grad_norm": 3.9222466646056637,
      "learning_rate": 4.742161419266251e-07,
      "loss": 0.8789,
      "num_input_tokens_seen": 138478180,
      "step": 6511
    },
    {
      "epoch": 0.7830217038417604,
      "flos": 20807129886720.0,
      "grad_norm": 3.5377063974078253,
      "learning_rate": 4.7371263222456304e-07,
      "loss": 0.8246,
      "num_input_tokens_seen": 138495220,
      "step": 6512
    },
    {
      "epoch": 0.7831419467323995,
      "flos": 43628070481920.0,
      "grad_norm": 0.8180001137157445,
      "learning_rate": 4.7320935405908004e-07,
      "loss": 0.8308,
      "num_input_tokens_seen": 138555810,
      "step": 6513
    },
    {
      "epoch": 0.7832621896230385,
      "flos": 13963954053120.0,
      "grad_norm": 4.320648676949333,
      "learning_rate": 4.7270630750652475e-07,
      "loss": 1.0131,
      "num_input_tokens_seen": 138571485,
      "step": 6514
    },
    {
      "epoch": 0.7833824325136777,
      "flos": 17792123535360.0,
      "grad_norm": 3.4343465038502305,
      "learning_rate": 4.7220349264320746e-07,
      "loss": 1.0049,
      "num_input_tokens_seen": 138590290,
      "step": 6515
    },
    {
      "epoch": 0.7835026754043167,
      "flos": 49269824163840.0,
      "grad_norm": 0.7919497402119205,
      "learning_rate": 4.71700909545407e-07,
      "loss": 0.7819,
      "num_input_tokens_seen": 138652955,
      "step": 6516
    },
    {
      "epoch": 0.7836229182949558,
      "flos": 14094753361920.0,
      "grad_norm": 4.4824578310785546,
      "learning_rate": 4.711985582893627e-07,
      "loss": 0.9594,
      "num_input_tokens_seen": 138671195,
      "step": 6517
    },
    {
      "epoch": 0.783743161185595,
      "flos": 16323553566720.0,
      "grad_norm": 3.430868653609863,
      "learning_rate": 4.706964389512811e-07,
      "loss": 0.9086,
      "num_input_tokens_seen": 138690950,
      "step": 6518
    },
    {
      "epoch": 0.783863404076234,
      "flos": 8772191293440.0,
      "grad_norm": 5.7746577438065945,
      "learning_rate": 4.701945516073345e-07,
      "loss": 1.0641,
      "num_input_tokens_seen": 138708145,
      "step": 6519
    },
    {
      "epoch": 0.7839836469668731,
      "flos": 17242245242880.0,
      "grad_norm": 3.3930847808652174,
      "learning_rate": 4.696928963336577e-07,
      "loss": 0.9485,
      "num_input_tokens_seen": 138727295,
      "step": 6520
    },
    {
      "epoch": 0.7841038898575122,
      "flos": 44470183403520.0,
      "grad_norm": 0.8616791440243587,
      "learning_rate": 4.6919147320635224e-07,
      "loss": 0.8228,
      "num_input_tokens_seen": 138789725,
      "step": 6521
    },
    {
      "epoch": 0.7842241327481513,
      "flos": 14331541094400.0,
      "grad_norm": 3.66561941161059,
      "learning_rate": 4.6869028230148286e-07,
      "loss": 0.9019,
      "num_input_tokens_seen": 138807240,
      "step": 6522
    },
    {
      "epoch": 0.7843443756387903,
      "flos": 19995162439680.0,
      "grad_norm": 4.769168124485588,
      "learning_rate": 4.6818932369507957e-07,
      "loss": 0.7884,
      "num_input_tokens_seen": 138826460,
      "step": 6523
    },
    {
      "epoch": 0.7844646185294295,
      "flos": 15143079321600.0,
      "grad_norm": 3.818504108310361,
      "learning_rate": 4.676885974631386e-07,
      "loss": 1.0737,
      "num_input_tokens_seen": 138844540,
      "step": 6524
    },
    {
      "epoch": 0.7845848614200686,
      "flos": 16819640524800.0,
      "grad_norm": 5.278396467564636,
      "learning_rate": 4.67188103681619e-07,
      "loss": 1.0031,
      "num_input_tokens_seen": 138864045,
      "step": 6525
    },
    {
      "epoch": 0.7847051043107076,
      "flos": 16636950712320.0,
      "grad_norm": 4.44792862592008,
      "learning_rate": 4.666878424264453e-07,
      "loss": 0.8872,
      "num_input_tokens_seen": 138883720,
      "step": 6526
    },
    {
      "epoch": 0.7848253472013467,
      "flos": 13488968294400.0,
      "grad_norm": 3.5235749955849838,
      "learning_rate": 4.661878137735069e-07,
      "loss": 0.9215,
      "num_input_tokens_seen": 138901630,
      "step": 6527
    },
    {
      "epoch": 0.7849455900919858,
      "flos": 15039666216960.0,
      "grad_norm": 4.283171730604849,
      "learning_rate": 4.656880177986571e-07,
      "loss": 0.9398,
      "num_input_tokens_seen": 138919895,
      "step": 6528
    },
    {
      "epoch": 0.7850658329826249,
      "flos": 13858977361920.0,
      "grad_norm": 5.813096866931272,
      "learning_rate": 4.6518845457771607e-07,
      "loss": 1.0001,
      "num_input_tokens_seen": 138938475,
      "step": 6529
    },
    {
      "epoch": 0.7851860758732639,
      "flos": 8798719057920.0,
      "grad_norm": 4.218040802296591,
      "learning_rate": 4.646891241864652e-07,
      "loss": 0.9742,
      "num_input_tokens_seen": 138956760,
      "step": 6530
    },
    {
      "epoch": 0.7853063187639031,
      "flos": 16319384002560.0,
      "grad_norm": 3.676881145217328,
      "learning_rate": 4.6419002670065397e-07,
      "loss": 0.9226,
      "num_input_tokens_seen": 138976060,
      "step": 6531
    },
    {
      "epoch": 0.7854265616545422,
      "flos": 12285614653440.0,
      "grad_norm": 3.6314113088683917,
      "learning_rate": 4.6369116219599445e-07,
      "loss": 1.0391,
      "num_input_tokens_seen": 138991765,
      "step": 6532
    },
    {
      "epoch": 0.7855468045451812,
      "flos": 16950133248000.0,
      "grad_norm": 3.726770167292555,
      "learning_rate": 4.631925307481637e-07,
      "loss": 0.9701,
      "num_input_tokens_seen": 139011300,
      "step": 6533
    },
    {
      "epoch": 0.7856670474358204,
      "flos": 18264809902080.0,
      "grad_norm": 3.685393556156157,
      "learning_rate": 4.6269413243280533e-07,
      "loss": 0.9326,
      "num_input_tokens_seen": 139030440,
      "step": 6534
    },
    {
      "epoch": 0.7857872903264594,
      "flos": 12857973780480.0,
      "grad_norm": 9.608498038616935,
      "learning_rate": 4.621959673255236e-07,
      "loss": 0.9233,
      "num_input_tokens_seen": 139046460,
      "step": 6535
    },
    {
      "epoch": 0.7859075332170985,
      "flos": 10156732477440.0,
      "grad_norm": 3.614967802123303,
      "learning_rate": 4.6169803550189135e-07,
      "loss": 1.0844,
      "num_input_tokens_seen": 139061875,
      "step": 6536
    },
    {
      "epoch": 0.7860277761077377,
      "flos": 14094293483520.0,
      "grad_norm": 5.111107768642788,
      "learning_rate": 4.6120033703744355e-07,
      "loss": 0.9561,
      "num_input_tokens_seen": 139080490,
      "step": 6537
    },
    {
      "epoch": 0.7861480189983767,
      "flos": 18788957552640.0,
      "grad_norm": 3.310870795592885,
      "learning_rate": 4.607028720076822e-07,
      "loss": 0.9781,
      "num_input_tokens_seen": 139096890,
      "step": 6538
    },
    {
      "epoch": 0.7862682618890158,
      "flos": 17235960238080.0,
      "grad_norm": 3.4529737896845574,
      "learning_rate": 4.6020564048807074e-07,
      "loss": 0.9179,
      "num_input_tokens_seen": 139114285,
      "step": 6539
    },
    {
      "epoch": 0.7863885047796549,
      "flos": 33992029347840.0,
      "grad_norm": 4.088585077853067,
      "learning_rate": 4.5970864255403883e-07,
      "loss": 0.9058,
      "num_input_tokens_seen": 139135530,
      "step": 6540
    },
    {
      "epoch": 0.786508747670294,
      "flos": 17346578104320.0,
      "grad_norm": 3.3709873788258427,
      "learning_rate": 4.59211878280982e-07,
      "loss": 1.0127,
      "num_input_tokens_seen": 139154765,
      "step": 6541
    },
    {
      "epoch": 0.786628990560933,
      "flos": 12784460881920.0,
      "grad_norm": 4.867457497151081,
      "learning_rate": 4.587153477442578e-07,
      "loss": 0.8895,
      "num_input_tokens_seen": 139170800,
      "step": 6542
    },
    {
      "epoch": 0.7867492334515722,
      "flos": 18395609210880.0,
      "grad_norm": 4.658848184830855,
      "learning_rate": 4.582190510191899e-07,
      "loss": 0.9946,
      "num_input_tokens_seen": 139189180,
      "step": 6543
    },
    {
      "epoch": 0.7868694763422113,
      "flos": 11734755287040.0,
      "grad_norm": 4.005772716639456,
      "learning_rate": 4.5772298818106625e-07,
      "loss": 1.0632,
      "num_input_tokens_seen": 139204690,
      "step": 6544
    },
    {
      "epoch": 0.7869897192328503,
      "flos": 20937653268480.0,
      "grad_norm": 5.434414744656812,
      "learning_rate": 4.572271593051384e-07,
      "loss": 0.9065,
      "num_input_tokens_seen": 139221765,
      "step": 6545
    },
    {
      "epoch": 0.7871099621234895,
      "flos": 12128778117120.0,
      "grad_norm": 2.98308131753623,
      "learning_rate": 4.567315644666245e-07,
      "loss": 0.9714,
      "num_input_tokens_seen": 139240280,
      "step": 6546
    },
    {
      "epoch": 0.7872302050141285,
      "flos": 16664459550720.0,
      "grad_norm": 3.5946453803485854,
      "learning_rate": 4.5623620374070507e-07,
      "loss": 1.0397,
      "num_input_tokens_seen": 139259315,
      "step": 6547
    },
    {
      "epoch": 0.7873504479047676,
      "flos": 47229232312320.0,
      "grad_norm": 0.8006338588632459,
      "learning_rate": 4.557410772025263e-07,
      "loss": 0.8184,
      "num_input_tokens_seen": 139320985,
      "step": 6548
    },
    {
      "epoch": 0.7874706907954068,
      "flos": 16533844193280.0,
      "grad_norm": 13.769441489069134,
      "learning_rate": 4.5524618492719803e-07,
      "loss": 0.8369,
      "num_input_tokens_seen": 139339925,
      "step": 6549
    },
    {
      "epoch": 0.7875909336860458,
      "flos": 20493395496960.0,
      "grad_norm": 3.474771308234723,
      "learning_rate": 4.54751526989795e-07,
      "loss": 0.9739,
      "num_input_tokens_seen": 139361485,
      "step": 6550
    },
    {
      "epoch": 0.7877111765766849,
      "flos": 13255859589120.0,
      "grad_norm": 3.7498396521300763,
      "learning_rate": 4.5425710346535775e-07,
      "loss": 0.9919,
      "num_input_tokens_seen": 139379150,
      "step": 6551
    },
    {
      "epoch": 0.787831419467324,
      "flos": 19648094085120.0,
      "grad_norm": 4.038337096368535,
      "learning_rate": 4.537629144288877e-07,
      "loss": 1.0113,
      "num_input_tokens_seen": 139396325,
      "step": 6552
    },
    {
      "epoch": 0.7879516623579631,
      "flos": 12862695198720.0,
      "grad_norm": 5.295246245194005,
      "learning_rate": 4.5326895995535477e-07,
      "loss": 0.9327,
      "num_input_tokens_seen": 139414945,
      "step": 6553
    },
    {
      "epoch": 0.7880719052486022,
      "flos": 14435873955840.0,
      "grad_norm": 3.7771025051016833,
      "learning_rate": 4.527752401196907e-07,
      "loss": 1.0341,
      "num_input_tokens_seen": 139432680,
      "step": 6554
    },
    {
      "epoch": 0.7881921481392413,
      "flos": 15380326932480.0,
      "grad_norm": 3.218920212476468,
      "learning_rate": 4.5228175499679254e-07,
      "loss": 0.8607,
      "num_input_tokens_seen": 139451985,
      "step": 6555
    },
    {
      "epoch": 0.7883123910298804,
      "flos": 49099375226880.0,
      "grad_norm": 0.8486277203686948,
      "learning_rate": 4.5178850466152174e-07,
      "loss": 0.7562,
      "num_input_tokens_seen": 139510535,
      "step": 6556
    },
    {
      "epoch": 0.7884326339205194,
      "flos": 13702202142720.0,
      "grad_norm": 3.947225035355386,
      "learning_rate": 4.512954891887031e-07,
      "loss": 1.0077,
      "num_input_tokens_seen": 139528555,
      "step": 6557
    },
    {
      "epoch": 0.7885528768111585,
      "flos": 12600115507200.0,
      "grad_norm": 5.790379259934271,
      "learning_rate": 4.5080270865312806e-07,
      "loss": 1.028,
      "num_input_tokens_seen": 139545470,
      "step": 6558
    },
    {
      "epoch": 0.7886731197017977,
      "flos": 13335289589760.0,
      "grad_norm": 4.834981549611436,
      "learning_rate": 4.5031016312954985e-07,
      "loss": 0.8996,
      "num_input_tokens_seen": 139563505,
      "step": 6559
    },
    {
      "epoch": 0.7887933625924367,
      "flos": 23797631938560.0,
      "grad_norm": 4.240260689476731,
      "learning_rate": 4.498178526926886e-07,
      "loss": 0.9392,
      "num_input_tokens_seen": 139584090,
      "step": 6560
    },
    {
      "epoch": 0.7889136054830758,
      "flos": 12050022604800.0,
      "grad_norm": 5.406731380133374,
      "learning_rate": 4.4932577741722635e-07,
      "loss": 0.9284,
      "num_input_tokens_seen": 139602340,
      "step": 6561
    },
    {
      "epoch": 0.7890338483737149,
      "flos": 20965468692480.0,
      "grad_norm": 2.838454125597262,
      "learning_rate": 4.4883393737780985e-07,
      "loss": 0.946,
      "num_input_tokens_seen": 139623010,
      "step": 6562
    },
    {
      "epoch": 0.789154091264354,
      "flos": 14171301457920.0,
      "grad_norm": 4.106692987291614,
      "learning_rate": 4.4834233264905254e-07,
      "loss": 0.9744,
      "num_input_tokens_seen": 139639745,
      "step": 6563
    },
    {
      "epoch": 0.789274334154993,
      "flos": 10266522562560.0,
      "grad_norm": 4.3734871564733,
      "learning_rate": 4.478509633055294e-07,
      "loss": 0.904,
      "num_input_tokens_seen": 139657175,
      "step": 6564
    },
    {
      "epoch": 0.7893945770456322,
      "flos": 15504381358080.0,
      "grad_norm": 4.921409366925327,
      "learning_rate": 4.473598294217813e-07,
      "loss": 0.9843,
      "num_input_tokens_seen": 139672320,
      "step": 6565
    },
    {
      "epoch": 0.7895148199362713,
      "flos": 14724398899200.0,
      "grad_norm": 3.6191796240478378,
      "learning_rate": 4.468689310723124e-07,
      "loss": 0.9133,
      "num_input_tokens_seen": 139689855,
      "step": 6566
    },
    {
      "epoch": 0.7896350628269103,
      "flos": 11813756067840.0,
      "grad_norm": 2.8037177404349722,
      "learning_rate": 4.463782683315913e-07,
      "loss": 0.9738,
      "num_input_tokens_seen": 139708580,
      "step": 6567
    },
    {
      "epoch": 0.7897553057175495,
      "flos": 16088329420800.0,
      "grad_norm": 3.122354684812569,
      "learning_rate": 4.458878412740523e-07,
      "loss": 0.9142,
      "num_input_tokens_seen": 139727080,
      "step": 6568
    },
    {
      "epoch": 0.7898755486081885,
      "flos": 10266307952640.0,
      "grad_norm": 4.090389357410226,
      "learning_rate": 4.453976499740919e-07,
      "loss": 0.9531,
      "num_input_tokens_seen": 139744445,
      "step": 6569
    },
    {
      "epoch": 0.7899957914988276,
      "flos": 12207533629440.0,
      "grad_norm": 2.9636533656911794,
      "learning_rate": 4.4490769450607215e-07,
      "loss": 0.9783,
      "num_input_tokens_seen": 139761790,
      "step": 6570
    },
    {
      "epoch": 0.7901160343894668,
      "flos": 29484807168000.0,
      "grad_norm": 2.893837952564574,
      "learning_rate": 4.4441797494431845e-07,
      "loss": 0.9245,
      "num_input_tokens_seen": 139783315,
      "step": 6571
    },
    {
      "epoch": 0.7902362772801058,
      "flos": 11918978027520.0,
      "grad_norm": 5.721914947720327,
      "learning_rate": 4.439284913631207e-07,
      "loss": 0.9751,
      "num_input_tokens_seen": 139800245,
      "step": 6572
    },
    {
      "epoch": 0.7903565201707449,
      "flos": 19313473105920.0,
      "grad_norm": 3.2017047990086653,
      "learning_rate": 4.434392438367347e-07,
      "loss": 1.0244,
      "num_input_tokens_seen": 139819390,
      "step": 6573
    },
    {
      "epoch": 0.790476763061384,
      "flos": 22115092316160.0,
      "grad_norm": 4.435639542115997,
      "learning_rate": 4.4295023243937677e-07,
      "loss": 0.9265,
      "num_input_tokens_seen": 139839315,
      "step": 6574
    },
    {
      "epoch": 0.7905970059520231,
      "flos": 15693294858240.0,
      "grad_norm": 3.52028906112162,
      "learning_rate": 4.4246145724523123e-07,
      "loss": 1.0011,
      "num_input_tokens_seen": 139856780,
      "step": 6575
    },
    {
      "epoch": 0.7907172488426621,
      "flos": 14593231687680.0,
      "grad_norm": 6.102658689843042,
      "learning_rate": 4.41972918328444e-07,
      "loss": 0.9533,
      "num_input_tokens_seen": 139873935,
      "step": 6576
    },
    {
      "epoch": 0.7908374917333013,
      "flos": 21437603205120.0,
      "grad_norm": 6.99193462575209,
      "learning_rate": 4.4148461576312646e-07,
      "loss": 0.9803,
      "num_input_tokens_seen": 139893320,
      "step": 6577
    },
    {
      "epoch": 0.7909577346239404,
      "flos": 14908223078400.0,
      "grad_norm": 3.1414833224465735,
      "learning_rate": 4.4099654962335343e-07,
      "loss": 0.9389,
      "num_input_tokens_seen": 139913490,
      "step": 6578
    },
    {
      "epoch": 0.7910779775145794,
      "flos": 18682110689280.0,
      "grad_norm": 3.110860651773232,
      "learning_rate": 4.405087199831636e-07,
      "loss": 0.9396,
      "num_input_tokens_seen": 139933450,
      "step": 6579
    },
    {
      "epoch": 0.7911982204052186,
      "flos": 16035059281920.0,
      "grad_norm": 3.3043621074380987,
      "learning_rate": 4.400211269165619e-07,
      "loss": 0.8696,
      "num_input_tokens_seen": 139949625,
      "step": 6580
    },
    {
      "epoch": 0.7913184632958576,
      "flos": 16429082112000.0,
      "grad_norm": 2.9431205708326,
      "learning_rate": 4.3953377049751416e-07,
      "loss": 0.9647,
      "num_input_tokens_seen": 139969770,
      "step": 6581
    },
    {
      "epoch": 0.7914387061864967,
      "flos": 8666632089600.0,
      "grad_norm": 6.689575443680953,
      "learning_rate": 4.390466507999537e-07,
      "loss": 0.965,
      "num_input_tokens_seen": 139985240,
      "step": 6582
    },
    {
      "epoch": 0.7915589490771359,
      "flos": 12469684101120.0,
      "grad_norm": 4.560756820649354,
      "learning_rate": 4.385597678977748e-07,
      "loss": 0.9579,
      "num_input_tokens_seen": 140003795,
      "step": 6583
    },
    {
      "epoch": 0.7916791919677749,
      "flos": 18210190786560.0,
      "grad_norm": 3.514794145527873,
      "learning_rate": 4.3807312186483726e-07,
      "loss": 0.936,
      "num_input_tokens_seen": 140024235,
      "step": 6584
    },
    {
      "epoch": 0.791799434858414,
      "flos": 13361694720000.0,
      "grad_norm": 5.204876395891529,
      "learning_rate": 4.375867127749655e-07,
      "loss": 0.9722,
      "num_input_tokens_seen": 140042230,
      "step": 6585
    },
    {
      "epoch": 0.7919196777490531,
      "flos": 18369081446400.0,
      "grad_norm": 3.544104615891205,
      "learning_rate": 4.3710054070194744e-07,
      "loss": 0.8663,
      "num_input_tokens_seen": 140061645,
      "step": 6586
    },
    {
      "epoch": 0.7920399206396922,
      "flos": 8405033472000.0,
      "grad_norm": 4.8130068821686445,
      "learning_rate": 4.3661460571953455e-07,
      "loss": 0.8391,
      "num_input_tokens_seen": 140078100,
      "step": 6587
    },
    {
      "epoch": 0.7921601635303313,
      "flos": 15327577989120.0,
      "grad_norm": 3.194254879694198,
      "learning_rate": 4.36128907901443e-07,
      "loss": 0.8796,
      "num_input_tokens_seen": 140097415,
      "step": 6588
    },
    {
      "epoch": 0.7922804064209703,
      "flos": 12836719288320.0,
      "grad_norm": 4.561518942508758,
      "learning_rate": 4.356434473213519e-07,
      "loss": 0.9186,
      "num_input_tokens_seen": 140114585,
      "step": 6589
    },
    {
      "epoch": 0.7924006493116095,
      "flos": 15379959029760.0,
      "grad_norm": 4.210395282664003,
      "learning_rate": 4.351582240529068e-07,
      "loss": 0.975,
      "num_input_tokens_seen": 140135135,
      "step": 6590
    },
    {
      "epoch": 0.7925208922022485,
      "flos": 45994966732800.0,
      "grad_norm": 0.6861611681561813,
      "learning_rate": 4.346732381697149e-07,
      "loss": 0.7998,
      "num_input_tokens_seen": 140198985,
      "step": 6591
    },
    {
      "epoch": 0.7926411350928876,
      "flos": 11993441341440.0,
      "grad_norm": 2.9217667993105167,
      "learning_rate": 4.3418848974534825e-07,
      "loss": 1.0019,
      "num_input_tokens_seen": 140215645,
      "step": 6592
    },
    {
      "epoch": 0.7927613779835267,
      "flos": 24584328622080.0,
      "grad_norm": 3.2161662277419327,
      "learning_rate": 4.3370397885334276e-07,
      "loss": 0.875,
      "num_input_tokens_seen": 140235995,
      "step": 6593
    },
    {
      "epoch": 0.7928816208741658,
      "flos": 13438120181760.0,
      "grad_norm": 3.512866077827265,
      "learning_rate": 4.3321970556719777e-07,
      "loss": 0.9452,
      "num_input_tokens_seen": 140254010,
      "step": 6594
    },
    {
      "epoch": 0.7930018637648049,
      "flos": 13202313523200.0,
      "grad_norm": 3.8913174055212583,
      "learning_rate": 4.3273566996037856e-07,
      "loss": 0.8906,
      "num_input_tokens_seen": 140270425,
      "step": 6595
    },
    {
      "epoch": 0.793122106655444,
      "flos": 17447599841280.0,
      "grad_norm": 3.6008192772573664,
      "learning_rate": 4.322518721063113e-07,
      "loss": 0.9963,
      "num_input_tokens_seen": 140288695,
      "step": 6596
    },
    {
      "epoch": 0.7932423495460831,
      "flos": 24556145295360.0,
      "grad_norm": 3.487323420166363,
      "learning_rate": 4.3176831207838906e-07,
      "loss": 0.8892,
      "num_input_tokens_seen": 140311825,
      "step": 6597
    },
    {
      "epoch": 0.7933625924367221,
      "flos": 19208833658880.0,
      "grad_norm": 5.013214768130463,
      "learning_rate": 4.3128498994996685e-07,
      "loss": 0.9327,
      "num_input_tokens_seen": 140331020,
      "step": 6598
    },
    {
      "epoch": 0.7934828353273613,
      "flos": 21068697845760.0,
      "grad_norm": 6.468370509418825,
      "learning_rate": 4.308019057943646e-07,
      "loss": 0.8995,
      "num_input_tokens_seen": 140352465,
      "step": 6599
    },
    {
      "epoch": 0.7936030782180004,
      "flos": 20382992240640.0,
      "grad_norm": 2.9424252714327865,
      "learning_rate": 4.3031905968486535e-07,
      "loss": 0.9364,
      "num_input_tokens_seen": 140373015,
      "step": 6600
    },
    {
      "epoch": 0.7937233211086394,
      "flos": 11598682705920.0,
      "grad_norm": 4.258721770743596,
      "learning_rate": 4.298364516947162e-07,
      "loss": 0.8732,
      "num_input_tokens_seen": 140389965,
      "step": 6601
    },
    {
      "epoch": 0.7938435639992786,
      "flos": 15638553108480.0,
      "grad_norm": 4.0875560698323685,
      "learning_rate": 4.293540818971295e-07,
      "loss": 0.853,
      "num_input_tokens_seen": 140407490,
      "step": 6602
    },
    {
      "epoch": 0.7939638068899176,
      "flos": 15770670735360.0,
      "grad_norm": 2.8315090776689003,
      "learning_rate": 4.2887195036527934e-07,
      "loss": 0.9595,
      "num_input_tokens_seen": 140426015,
      "step": 6603
    },
    {
      "epoch": 0.7940840497805567,
      "flos": 12571135057920.0,
      "grad_norm": 6.294076756894508,
      "learning_rate": 4.28390057172306e-07,
      "loss": 0.9007,
      "num_input_tokens_seen": 140442240,
      "step": 6604
    },
    {
      "epoch": 0.7942042926711959,
      "flos": 16927008583680.0,
      "grad_norm": 3.701635519896004,
      "learning_rate": 4.279084023913111e-07,
      "loss": 0.912,
      "num_input_tokens_seen": 140459835,
      "step": 6605
    },
    {
      "epoch": 0.7943245355618349,
      "flos": 13649514516480.0,
      "grad_norm": 3.171943701389985,
      "learning_rate": 4.2742698609536096e-07,
      "loss": 0.8831,
      "num_input_tokens_seen": 140477865,
      "step": 6606
    },
    {
      "epoch": 0.794444778452474,
      "flos": 17790743900160.0,
      "grad_norm": 4.367816586301145,
      "learning_rate": 4.2694580835748706e-07,
      "loss": 0.9706,
      "num_input_tokens_seen": 140497445,
      "step": 6607
    },
    {
      "epoch": 0.7945650213431131,
      "flos": 16507316428800.0,
      "grad_norm": 4.6761918043050175,
      "learning_rate": 4.264648692506836e-07,
      "loss": 0.9307,
      "num_input_tokens_seen": 140515955,
      "step": 6608
    },
    {
      "epoch": 0.7946852642337522,
      "flos": 18548766720000.0,
      "grad_norm": 3.072250569883249,
      "learning_rate": 4.2598416884790824e-07,
      "loss": 0.9035,
      "num_input_tokens_seen": 140534725,
      "step": 6609
    },
    {
      "epoch": 0.7948055071243912,
      "flos": 16901431234560.0,
      "grad_norm": 4.646720197160384,
      "learning_rate": 4.255037072220828e-07,
      "loss": 0.9932,
      "num_input_tokens_seen": 140555815,
      "step": 6610
    },
    {
      "epoch": 0.7949257500150304,
      "flos": 15615643054080.0,
      "grad_norm": 2.8211192828966483,
      "learning_rate": 4.2502348444609293e-07,
      "loss": 0.9134,
      "num_input_tokens_seen": 140575155,
      "step": 6611
    },
    {
      "epoch": 0.7950459929056695,
      "flos": 18341664583680.0,
      "grad_norm": 4.615614776496696,
      "learning_rate": 4.2454350059278844e-07,
      "loss": 0.8704,
      "num_input_tokens_seen": 140595935,
      "step": 6612
    },
    {
      "epoch": 0.7951662357963085,
      "flos": 15742334115840.0,
      "grad_norm": 2.7319557168591504,
      "learning_rate": 4.240637557349824e-07,
      "loss": 1.0388,
      "num_input_tokens_seen": 140612870,
      "step": 6613
    },
    {
      "epoch": 0.7952864786869477,
      "flos": 17527826964480.0,
      "grad_norm": 2.590699849813272,
      "learning_rate": 4.235842499454516e-07,
      "loss": 0.8456,
      "num_input_tokens_seen": 140632505,
      "step": 6614
    },
    {
      "epoch": 0.7954067215775867,
      "flos": 15507385896960.0,
      "grad_norm": 2.99457440901262,
      "learning_rate": 4.2310498329693687e-07,
      "loss": 1.0106,
      "num_input_tokens_seen": 140653125,
      "step": 6615
    },
    {
      "epoch": 0.7955269644682258,
      "flos": 17110403543040.0,
      "grad_norm": 15.790846401574711,
      "learning_rate": 4.2262595586214164e-07,
      "loss": 1.0025,
      "num_input_tokens_seen": 140673940,
      "step": 6616
    },
    {
      "epoch": 0.795647207358865,
      "flos": 17793411194880.0,
      "grad_norm": 3.4723932876833157,
      "learning_rate": 4.221471677137358e-07,
      "loss": 0.9582,
      "num_input_tokens_seen": 140694475,
      "step": 6617
    },
    {
      "epoch": 0.795767450249504,
      "flos": 10346105856000.0,
      "grad_norm": 3.6691743049712464,
      "learning_rate": 4.216686189243492e-07,
      "loss": 0.9008,
      "num_input_tokens_seen": 140712985,
      "step": 6618
    },
    {
      "epoch": 0.7958876931401431,
      "flos": 13148031651840.0,
      "grad_norm": 3.1006679061507696,
      "learning_rate": 4.211903095665785e-07,
      "loss": 0.923,
      "num_input_tokens_seen": 140732090,
      "step": 6619
    },
    {
      "epoch": 0.7960079360307821,
      "flos": 15301816688640.0,
      "grad_norm": 4.715984418252792,
      "learning_rate": 4.2071223971298277e-07,
      "loss": 0.9433,
      "num_input_tokens_seen": 140748995,
      "step": 6620
    },
    {
      "epoch": 0.7961281789214213,
      "flos": 18131465932800.0,
      "grad_norm": 4.329442233486963,
      "learning_rate": 4.2023440943608433e-07,
      "loss": 0.793,
      "num_input_tokens_seen": 140768680,
      "step": 6621
    },
    {
      "epoch": 0.7962484218120603,
      "flos": 15589513850880.0,
      "grad_norm": 2.5451617885344096,
      "learning_rate": 4.1975681880837023e-07,
      "loss": 0.9726,
      "num_input_tokens_seen": 140788405,
      "step": 6622
    },
    {
      "epoch": 0.7963686647026994,
      "flos": 13384022261760.0,
      "grad_norm": 3.1080288680808965,
      "learning_rate": 4.192794679022895e-07,
      "loss": 1.0183,
      "num_input_tokens_seen": 140806450,
      "step": 6623
    },
    {
      "epoch": 0.7964889075933386,
      "flos": 21175299440640.0,
      "grad_norm": 3.1941951672876274,
      "learning_rate": 4.1880235679025743e-07,
      "loss": 0.9166,
      "num_input_tokens_seen": 140826265,
      "step": 6624
    },
    {
      "epoch": 0.7966091504839776,
      "flos": 21012668436480.0,
      "grad_norm": 2.781054372015222,
      "learning_rate": 4.1832548554464986e-07,
      "loss": 0.8174,
      "num_input_tokens_seen": 140844280,
      "step": 6625
    },
    {
      "epoch": 0.7967293933746167,
      "flos": 48181480673280.0,
      "grad_norm": 0.7805859604687418,
      "learning_rate": 4.178488542378098e-07,
      "loss": 0.796,
      "num_input_tokens_seen": 140901580,
      "step": 6626
    },
    {
      "epoch": 0.7968496362652558,
      "flos": 18183877632000.0,
      "grad_norm": 4.651555504631854,
      "learning_rate": 4.173724629420401e-07,
      "loss": 1.0835,
      "num_input_tokens_seen": 140922660,
      "step": 6627
    },
    {
      "epoch": 0.7969698791558949,
      "flos": 10240086773760.0,
      "grad_norm": 5.909760254517026,
      "learning_rate": 4.168963117296087e-07,
      "loss": 0.8812,
      "num_input_tokens_seen": 140939715,
      "step": 6628
    },
    {
      "epoch": 0.797090122046534,
      "flos": 15721202257920.0,
      "grad_norm": 3.574718336200281,
      "learning_rate": 4.1642040067274876e-07,
      "loss": 0.937,
      "num_input_tokens_seen": 140959105,
      "step": 6629
    },
    {
      "epoch": 0.7972103649371731,
      "flos": 14118061977600.0,
      "grad_norm": 3.3795844130682267,
      "learning_rate": 4.1594472984365493e-07,
      "loss": 0.9217,
      "num_input_tokens_seen": 140977510,
      "step": 6630
    },
    {
      "epoch": 0.7973306078278122,
      "flos": 25732327342080.0,
      "grad_norm": 5.606736436576833,
      "learning_rate": 4.154692993144862e-07,
      "loss": 0.9656,
      "num_input_tokens_seen": 140997000,
      "step": 6631
    },
    {
      "epoch": 0.7974508507184512,
      "flos": 15169913671680.0,
      "grad_norm": 4.011257542561173,
      "learning_rate": 4.1499410915736476e-07,
      "loss": 0.9092,
      "num_input_tokens_seen": 141015650,
      "step": 6632
    },
    {
      "epoch": 0.7975710936090904,
      "flos": 48876721090560.0,
      "grad_norm": 0.8075172150256078,
      "learning_rate": 4.145191594443762e-07,
      "loss": 0.9004,
      "num_input_tokens_seen": 141079725,
      "step": 6633
    },
    {
      "epoch": 0.7976913364997295,
      "flos": 15983444705280.0,
      "grad_norm": 6.019349369615963,
      "learning_rate": 4.140444502475713e-07,
      "loss": 0.9026,
      "num_input_tokens_seen": 141098995,
      "step": 6634
    },
    {
      "epoch": 0.7978115793903685,
      "flos": 10787267112960.0,
      "grad_norm": 3.311923282729912,
      "learning_rate": 4.1356998163896216e-07,
      "loss": 0.8917,
      "num_input_tokens_seen": 141115765,
      "step": 6635
    },
    {
      "epoch": 0.7979318222810077,
      "flos": 13990359183360.0,
      "grad_norm": 3.3151843152941827,
      "learning_rate": 4.130957536905255e-07,
      "loss": 0.9326,
      "num_input_tokens_seen": 141133500,
      "step": 6636
    },
    {
      "epoch": 0.7980520651716467,
      "flos": 11001451376640.0,
      "grad_norm": 4.4070107094202005,
      "learning_rate": 4.1262176647420134e-07,
      "loss": 0.8936,
      "num_input_tokens_seen": 141151385,
      "step": 6637
    },
    {
      "epoch": 0.7981723080622858,
      "flos": 15851848273920.0,
      "grad_norm": 3.0883015136671563,
      "learning_rate": 4.121480200618923e-07,
      "loss": 0.9887,
      "num_input_tokens_seen": 141170760,
      "step": 6638
    },
    {
      "epoch": 0.798292550952925,
      "flos": 16297516339200.0,
      "grad_norm": 3.189067380416403,
      "learning_rate": 4.116745145254674e-07,
      "loss": 0.9934,
      "num_input_tokens_seen": 141190015,
      "step": 6639
    },
    {
      "epoch": 0.798412793843564,
      "flos": 46178331033600.0,
      "grad_norm": 0.7770320708383458,
      "learning_rate": 4.1120124993675476e-07,
      "loss": 0.7932,
      "num_input_tokens_seen": 141254165,
      "step": 6640
    },
    {
      "epoch": 0.7985330367342031,
      "flos": 9582809763840.0,
      "grad_norm": 17.8201206971265,
      "learning_rate": 4.107282263675498e-07,
      "loss": 0.8065,
      "num_input_tokens_seen": 141271555,
      "step": 6641
    },
    {
      "epoch": 0.7986532796248422,
      "flos": 48477394329600.0,
      "grad_norm": 0.7301317549387969,
      "learning_rate": 4.1025544388960907e-07,
      "loss": 0.7321,
      "num_input_tokens_seen": 141332315,
      "step": 6642
    },
    {
      "epoch": 0.7987735225154813,
      "flos": 15956732989440.0,
      "grad_norm": 3.2873168082016906,
      "learning_rate": 4.097829025746538e-07,
      "loss": 0.9009,
      "num_input_tokens_seen": 141353580,
      "step": 6643
    },
    {
      "epoch": 0.7988937654061203,
      "flos": 49314571223040.0,
      "grad_norm": 0.7001458526976673,
      "learning_rate": 4.0931060249436757e-07,
      "loss": 0.8158,
      "num_input_tokens_seen": 141417140,
      "step": 6644
    },
    {
      "epoch": 0.7990140082967595,
      "flos": 14771108106240.0,
      "grad_norm": 3.6430373160634066,
      "learning_rate": 4.088385437203978e-07,
      "loss": 0.893,
      "num_input_tokens_seen": 141433870,
      "step": 6645
    },
    {
      "epoch": 0.7991342511873986,
      "flos": 13462317895680.0,
      "grad_norm": 4.416880452482109,
      "learning_rate": 4.083667263243564e-07,
      "loss": 0.963,
      "num_input_tokens_seen": 141451935,
      "step": 6646
    },
    {
      "epoch": 0.7992544940780376,
      "flos": 14779569868800.0,
      "grad_norm": 5.56041121107996,
      "learning_rate": 4.0789515037781653e-07,
      "loss": 0.9094,
      "num_input_tokens_seen": 141472380,
      "step": 6647
    },
    {
      "epoch": 0.7993747369686768,
      "flos": 8903174553600.0,
      "grad_norm": 3.1435257151021987,
      "learning_rate": 4.0742381595231755e-07,
      "loss": 1.0156,
      "num_input_tokens_seen": 141488825,
      "step": 6648
    },
    {
      "epoch": 0.7994949798593158,
      "flos": 14248033505280.0,
      "grad_norm": 2.638989184520396,
      "learning_rate": 4.06952723119359e-07,
      "loss": 0.984,
      "num_input_tokens_seen": 141508420,
      "step": 6649
    },
    {
      "epoch": 0.7996152227499549,
      "flos": 27598415216640.0,
      "grad_norm": 3.290686996759876,
      "learning_rate": 4.0648187195040504e-07,
      "loss": 0.8538,
      "num_input_tokens_seen": 141530345,
      "step": 6650
    },
    {
      "epoch": 0.799735465640594,
      "flos": 50305242869760.0,
      "grad_norm": 0.9442099454634337,
      "learning_rate": 4.060112625168848e-07,
      "loss": 0.9184,
      "num_input_tokens_seen": 141595175,
      "step": 6651
    },
    {
      "epoch": 0.7998557085312331,
      "flos": 17239731240960.0,
      "grad_norm": 3.9794658262163685,
      "learning_rate": 4.055408948901886e-07,
      "loss": 0.9257,
      "num_input_tokens_seen": 141616295,
      "step": 6652
    },
    {
      "epoch": 0.7999759514218722,
      "flos": 19628311203840.0,
      "grad_norm": 2.856305084699014,
      "learning_rate": 4.050707691416708e-07,
      "loss": 0.9033,
      "num_input_tokens_seen": 141637325,
      "step": 6653
    },
    {
      "epoch": 0.8000961943125112,
      "flos": 48218156421120.0,
      "grad_norm": 0.7299365792747587,
      "learning_rate": 4.046008853426495e-07,
      "loss": 0.8147,
      "num_input_tokens_seen": 141700360,
      "step": 6654
    },
    {
      "epoch": 0.8002164372031504,
      "flos": 20469320417280.0,
      "grad_norm": 3.520333218841604,
      "learning_rate": 4.0413124356440464e-07,
      "loss": 0.8207,
      "num_input_tokens_seen": 141724125,
      "step": 6655
    },
    {
      "epoch": 0.8003366800937894,
      "flos": 12495629352960.0,
      "grad_norm": 4.963067794333696,
      "learning_rate": 4.0366184387818223e-07,
      "loss": 1.0099,
      "num_input_tokens_seen": 141742305,
      "step": 6656
    },
    {
      "epoch": 0.8004569229844285,
      "flos": 18500923146240.0,
      "grad_norm": 4.167343822131833,
      "learning_rate": 4.0319268635518797e-07,
      "loss": 1.0322,
      "num_input_tokens_seen": 141762600,
      "step": 6657
    },
    {
      "epoch": 0.8005771658750677,
      "flos": 14775768207360.0,
      "grad_norm": 2.5974725873453406,
      "learning_rate": 4.027237710665943e-07,
      "loss": 0.9487,
      "num_input_tokens_seen": 141780785,
      "step": 6658
    },
    {
      "epoch": 0.8006974087657067,
      "flos": 18369602641920.0,
      "grad_norm": 3.5427221782308735,
      "learning_rate": 4.022550980835344e-07,
      "loss": 0.8855,
      "num_input_tokens_seen": 141802750,
      "step": 6659
    },
    {
      "epoch": 0.8008176516563458,
      "flos": 12154386124800.0,
      "grad_norm": 4.466783424293806,
      "learning_rate": 4.017866674771051e-07,
      "loss": 1.0028,
      "num_input_tokens_seen": 141819955,
      "step": 6660
    },
    {
      "epoch": 0.8009378945469849,
      "flos": 17215901429760.0,
      "grad_norm": 2.9240976947061865,
      "learning_rate": 4.013184793183688e-07,
      "loss": 0.9309,
      "num_input_tokens_seen": 141841770,
      "step": 6661
    },
    {
      "epoch": 0.801058137437624,
      "flos": 14038631976960.0,
      "grad_norm": 3.5624886228145494,
      "learning_rate": 4.008505336783472e-07,
      "loss": 0.9178,
      "num_input_tokens_seen": 141859215,
      "step": 6662
    },
    {
      "epoch": 0.801178380328263,
      "flos": 13228442726400.0,
      "grad_norm": 4.1475709631553235,
      "learning_rate": 4.003828306280284e-07,
      "loss": 1.01,
      "num_input_tokens_seen": 141876610,
      "step": 6663
    },
    {
      "epoch": 0.8012986232189022,
      "flos": 11106642677760.0,
      "grad_norm": 3.1782014640679424,
      "learning_rate": 3.999153702383626e-07,
      "loss": 0.9694,
      "num_input_tokens_seen": 141894220,
      "step": 6664
    },
    {
      "epoch": 0.8014188661095413,
      "flos": 20361247211520.0,
      "grad_norm": 3.439532702000358,
      "learning_rate": 3.9944815258026263e-07,
      "loss": 0.9195,
      "num_input_tokens_seen": 141915760,
      "step": 6665
    },
    {
      "epoch": 0.8015391090001803,
      "flos": 20882696908800.0,
      "grad_norm": 3.7000734012643117,
      "learning_rate": 3.989811777246057e-07,
      "loss": 1.027,
      "num_input_tokens_seen": 141935650,
      "step": 6666
    },
    {
      "epoch": 0.8016593518908195,
      "flos": 50416596541440.0,
      "grad_norm": 0.9100183210519822,
      "learning_rate": 3.985144457422305e-07,
      "loss": 0.892,
      "num_input_tokens_seen": 141989655,
      "step": 6667
    },
    {
      "epoch": 0.8017795947814585,
      "flos": 18522300272640.0,
      "grad_norm": 2.946508837510726,
      "learning_rate": 3.9804795670394096e-07,
      "loss": 0.9461,
      "num_input_tokens_seen": 142009500,
      "step": 6668
    },
    {
      "epoch": 0.8018998376720976,
      "flos": 15691669954560.0,
      "grad_norm": 3.2234545649466493,
      "learning_rate": 3.975817106805022e-07,
      "loss": 0.906,
      "num_input_tokens_seen": 142027920,
      "step": 6669
    },
    {
      "epoch": 0.8020200805627368,
      "flos": 24662225694720.0,
      "grad_norm": 3.3381191865134388,
      "learning_rate": 3.97115707742645e-07,
      "loss": 0.8318,
      "num_input_tokens_seen": 142048315,
      "step": 6670
    },
    {
      "epoch": 0.8021403234533758,
      "flos": 14278608199680.0,
      "grad_norm": 4.36588820781518,
      "learning_rate": 3.966499479610599e-07,
      "loss": 0.8432,
      "num_input_tokens_seen": 142066130,
      "step": 6671
    },
    {
      "epoch": 0.8022605663440149,
      "flos": 19759202488320.0,
      "grad_norm": 4.082430499688885,
      "learning_rate": 3.9618443140640225e-07,
      "loss": 0.8451,
      "num_input_tokens_seen": 142084760,
      "step": 6672
    },
    {
      "epoch": 0.802380809234654,
      "flos": 43119045611520.0,
      "grad_norm": 0.7470306585421651,
      "learning_rate": 3.957191581492918e-07,
      "loss": 0.7366,
      "num_input_tokens_seen": 142145240,
      "step": 6673
    },
    {
      "epoch": 0.8025010521252931,
      "flos": 10656099901440.0,
      "grad_norm": 4.084260614686183,
      "learning_rate": 3.952541282603097e-07,
      "loss": 0.8843,
      "num_input_tokens_seen": 142160065,
      "step": 6674
    },
    {
      "epoch": 0.8026212950159322,
      "flos": 15746994216960.0,
      "grad_norm": 3.0900409698239324,
      "learning_rate": 3.9478934181000013e-07,
      "loss": 1.0296,
      "num_input_tokens_seen": 142179810,
      "step": 6675
    },
    {
      "epoch": 0.8027415379065713,
      "flos": 12521911848960.0,
      "grad_norm": 5.612263820241194,
      "learning_rate": 3.943247988688714e-07,
      "loss": 1.0337,
      "num_input_tokens_seen": 142198225,
      "step": 6676
    },
    {
      "epoch": 0.8028617807972104,
      "flos": 15614600663040.0,
      "grad_norm": 3.4699215671788672,
      "learning_rate": 3.938604995073933e-07,
      "loss": 0.9151,
      "num_input_tokens_seen": 142216415,
      "step": 6677
    },
    {
      "epoch": 0.8029820236878494,
      "flos": 18811591680000.0,
      "grad_norm": 3.383603609185662,
      "learning_rate": 3.9339644379600157e-07,
      "loss": 0.8578,
      "num_input_tokens_seen": 142235965,
      "step": 6678
    },
    {
      "epoch": 0.8031022665784886,
      "flos": 12522341068800.0,
      "grad_norm": 3.6940322174336138,
      "learning_rate": 3.929326318050907e-07,
      "loss": 0.9104,
      "num_input_tokens_seen": 142253355,
      "step": 6679
    },
    {
      "epoch": 0.8032225094691277,
      "flos": 10922327961600.0,
      "grad_norm": 3.690906567536383,
      "learning_rate": 3.924690636050225e-07,
      "loss": 0.9762,
      "num_input_tokens_seen": 142270485,
      "step": 6680
    },
    {
      "epoch": 0.8033427523597667,
      "flos": 18633286041600.0,
      "grad_norm": 4.944066570191818,
      "learning_rate": 3.9200573926611915e-07,
      "loss": 0.921,
      "num_input_tokens_seen": 142291620,
      "step": 6681
    },
    {
      "epoch": 0.8034629952504058,
      "flos": 15144366981120.0,
      "grad_norm": 3.2867565634338347,
      "learning_rate": 3.9154265885866613e-07,
      "loss": 0.9196,
      "num_input_tokens_seen": 142310650,
      "step": 6682
    },
    {
      "epoch": 0.8035832381410449,
      "flos": 15378916638720.0,
      "grad_norm": 5.982833783362672,
      "learning_rate": 3.9107982245291394e-07,
      "loss": 0.9241,
      "num_input_tokens_seen": 142328495,
      "step": 6683
    },
    {
      "epoch": 0.803703481031684,
      "flos": 14564588482560.0,
      "grad_norm": 3.707398986263438,
      "learning_rate": 3.9061723011907245e-07,
      "loss": 0.9547,
      "num_input_tokens_seen": 142347570,
      "step": 6684
    },
    {
      "epoch": 0.803823723922323,
      "flos": 16243541053440.0,
      "grad_norm": 3.161620905224415,
      "learning_rate": 3.901548819273179e-07,
      "loss": 0.9679,
      "num_input_tokens_seen": 142367305,
      "step": 6685
    },
    {
      "epoch": 0.8039439668129622,
      "flos": 15171231989760.0,
      "grad_norm": 3.3615352338504456,
      "learning_rate": 3.896927779477881e-07,
      "loss": 0.8896,
      "num_input_tokens_seen": 142386285,
      "step": 6686
    },
    {
      "epoch": 0.8040642097036013,
      "flos": 16924402606080.0,
      "grad_norm": 7.973643038591881,
      "learning_rate": 3.892309182505833e-07,
      "loss": 0.8729,
      "num_input_tokens_seen": 142403820,
      "step": 6687
    },
    {
      "epoch": 0.8041844525942403,
      "flos": 18448235520000.0,
      "grad_norm": 3.6725611268858263,
      "learning_rate": 3.887693029057675e-07,
      "loss": 1.0501,
      "num_input_tokens_seen": 142423050,
      "step": 6688
    },
    {
      "epoch": 0.8043046954848795,
      "flos": 17922769551360.0,
      "grad_norm": 2.8511554584273577,
      "learning_rate": 3.8830793198336684e-07,
      "loss": 1.0142,
      "num_input_tokens_seen": 142442360,
      "step": 6689
    },
    {
      "epoch": 0.8044249383755185,
      "flos": 29801178193920.0,
      "grad_norm": 3.274972487472402,
      "learning_rate": 3.878468055533721e-07,
      "loss": 0.8984,
      "num_input_tokens_seen": 142464620,
      "step": 6690
    },
    {
      "epoch": 0.8045451812661576,
      "flos": 14645949972480.0,
      "grad_norm": 10.496934715798435,
      "learning_rate": 3.8738592368573464e-07,
      "loss": 1.0318,
      "num_input_tokens_seen": 142481895,
      "step": 6691
    },
    {
      "epoch": 0.8046654241567968,
      "flos": 21171221852160.0,
      "grad_norm": 7.305630840240763,
      "learning_rate": 3.8692528645037137e-07,
      "loss": 1.063,
      "num_input_tokens_seen": 142500795,
      "step": 6692
    },
    {
      "epoch": 0.8047856670474358,
      "flos": 12517987553280.0,
      "grad_norm": 4.208782486819948,
      "learning_rate": 3.8646489391715907e-07,
      "loss": 0.9811,
      "num_input_tokens_seen": 142514810,
      "step": 6693
    },
    {
      "epoch": 0.8049059099380749,
      "flos": 12121757306880.0,
      "grad_norm": 4.0470917986088715,
      "learning_rate": 3.8600474615593903e-07,
      "loss": 1.0729,
      "num_input_tokens_seen": 142529145,
      "step": 6694
    },
    {
      "epoch": 0.805026152828714,
      "flos": 44533027123200.0,
      "grad_norm": 0.8272495680848531,
      "learning_rate": 3.8554484323651605e-07,
      "loss": 0.8491,
      "num_input_tokens_seen": 142590735,
      "step": 6695
    },
    {
      "epoch": 0.8051463957193531,
      "flos": 15405597696000.0,
      "grad_norm": 3.6053737904782333,
      "learning_rate": 3.85085185228657e-07,
      "loss": 0.9787,
      "num_input_tokens_seen": 142609425,
      "step": 6696
    },
    {
      "epoch": 0.8052666386099921,
      "flos": 22853301596160.0,
      "grad_norm": 4.494296804380986,
      "learning_rate": 3.8462577220209114e-07,
      "loss": 0.9335,
      "num_input_tokens_seen": 142629520,
      "step": 6697
    },
    {
      "epoch": 0.8053868815006313,
      "flos": 48089503211520.0,
      "grad_norm": 0.6785081519295686,
      "learning_rate": 3.8416660422651127e-07,
      "loss": 0.7993,
      "num_input_tokens_seen": 142698890,
      "step": 6698
    },
    {
      "epoch": 0.8055071243912704,
      "flos": 16949029539840.0,
      "grad_norm": 7.512765574402527,
      "learning_rate": 3.837076813715723e-07,
      "loss": 0.8805,
      "num_input_tokens_seen": 142718495,
      "step": 6699
    },
    {
      "epoch": 0.8056273672819094,
      "flos": 15144029736960.0,
      "grad_norm": 8.235152785106234,
      "learning_rate": 3.832490037068941e-07,
      "loss": 0.9413,
      "num_input_tokens_seen": 142737005,
      "step": 6700
    },
    {
      "epoch": 0.8057476101725486,
      "flos": 18336759214080.0,
      "grad_norm": 3.1520871952602194,
      "learning_rate": 3.827905713020554e-07,
      "loss": 0.9501,
      "num_input_tokens_seen": 142754370,
      "step": 6701
    },
    {
      "epoch": 0.8058678530631876,
      "flos": 17451554795520.0,
      "grad_norm": 3.531275926342878,
      "learning_rate": 3.823323842266017e-07,
      "loss": 0.8748,
      "num_input_tokens_seen": 142773485,
      "step": 6702
    },
    {
      "epoch": 0.8059880959538267,
      "flos": 17608483307520.0,
      "grad_norm": 6.069484187776097,
      "learning_rate": 3.818744425500393e-07,
      "loss": 0.9206,
      "num_input_tokens_seen": 142791220,
      "step": 6703
    },
    {
      "epoch": 0.8061083388444659,
      "flos": 15770088222720.0,
      "grad_norm": 3.8400973200386685,
      "learning_rate": 3.8141674634183675e-07,
      "loss": 1.0041,
      "num_input_tokens_seen": 142809970,
      "step": 6704
    },
    {
      "epoch": 0.8062285817351049,
      "flos": 21410646220800.0,
      "grad_norm": 3.7478270274891976,
      "learning_rate": 3.809592956714278e-07,
      "loss": 0.8391,
      "num_input_tokens_seen": 142832925,
      "step": 6705
    },
    {
      "epoch": 0.806348824625744,
      "flos": 16191742525440.0,
      "grad_norm": 3.449938378267469,
      "learning_rate": 3.805020906082057e-07,
      "loss": 0.9334,
      "num_input_tokens_seen": 142851220,
      "step": 6706
    },
    {
      "epoch": 0.8064690675163831,
      "flos": 16639004835840.0,
      "grad_norm": 5.071296678666291,
      "learning_rate": 3.8004513122152917e-07,
      "loss": 1.0104,
      "num_input_tokens_seen": 142869250,
      "step": 6707
    },
    {
      "epoch": 0.8065893104070222,
      "flos": 17110618152960.0,
      "grad_norm": 3.3944667726633195,
      "learning_rate": 3.79588417580718e-07,
      "loss": 0.8717,
      "num_input_tokens_seen": 142887080,
      "step": 6708
    },
    {
      "epoch": 0.8067095532976613,
      "flos": 15848414515200.0,
      "grad_norm": 4.226735335392906,
      "learning_rate": 3.791319497550558e-07,
      "loss": 0.9451,
      "num_input_tokens_seen": 142904630,
      "step": 6709
    },
    {
      "epoch": 0.8068297961883004,
      "flos": 12129268654080.0,
      "grad_norm": 3.5716219114014476,
      "learning_rate": 3.78675727813788e-07,
      "loss": 0.9099,
      "num_input_tokens_seen": 142921915,
      "step": 6710
    },
    {
      "epoch": 0.8069500390789395,
      "flos": 15643182551040.0,
      "grad_norm": 2.3933706037688487,
      "learning_rate": 3.782197518261225e-07,
      "loss": 0.9174,
      "num_input_tokens_seen": 142941075,
      "step": 6711
    },
    {
      "epoch": 0.8070702819695785,
      "flos": 13648901345280.0,
      "grad_norm": 4.449338353720547,
      "learning_rate": 3.777640218612319e-07,
      "loss": 1.1623,
      "num_input_tokens_seen": 142958780,
      "step": 6712
    },
    {
      "epoch": 0.8071905248602176,
      "flos": 15302123274240.0,
      "grad_norm": 5.026878780034887,
      "learning_rate": 3.773085379882488e-07,
      "loss": 0.9057,
      "num_input_tokens_seen": 142977555,
      "step": 6713
    },
    {
      "epoch": 0.8073107677508568,
      "flos": 26602470297600.0,
      "grad_norm": 3.3645718852853244,
      "learning_rate": 3.768533002762715e-07,
      "loss": 0.9595,
      "num_input_tokens_seen": 143000810,
      "step": 6714
    },
    {
      "epoch": 0.8074310106414958,
      "flos": 20204809236480.0,
      "grad_norm": 4.596892650785602,
      "learning_rate": 3.763983087943572e-07,
      "loss": 0.9543,
      "num_input_tokens_seen": 143019920,
      "step": 6715
    },
    {
      "epoch": 0.8075512535321349,
      "flos": 17269508812800.0,
      "grad_norm": 2.799214838841288,
      "learning_rate": 3.759435636115282e-07,
      "loss": 0.9913,
      "num_input_tokens_seen": 143040425,
      "step": 6716
    },
    {
      "epoch": 0.807671496422774,
      "flos": 18526347202560.0,
      "grad_norm": 3.19308461165034,
      "learning_rate": 3.7548906479676967e-07,
      "loss": 0.931,
      "num_input_tokens_seen": 143059740,
      "step": 6717
    },
    {
      "epoch": 0.8077917393134131,
      "flos": 16873094615040.0,
      "grad_norm": 4.148702092182881,
      "learning_rate": 3.7503481241902855e-07,
      "loss": 0.8999,
      "num_input_tokens_seen": 143079435,
      "step": 6718
    },
    {
      "epoch": 0.8079119822040521,
      "flos": 13043208253440.0,
      "grad_norm": 4.115896409802558,
      "learning_rate": 3.745808065472145e-07,
      "loss": 0.9841,
      "num_input_tokens_seen": 143096450,
      "step": 6719
    },
    {
      "epoch": 0.8080322250946913,
      "flos": 16794860298240.0,
      "grad_norm": 3.064789986315096,
      "learning_rate": 3.741270472501994e-07,
      "loss": 0.9565,
      "num_input_tokens_seen": 143116810,
      "step": 6720
    },
    {
      "epoch": 0.8081524679853304,
      "flos": 16219343339520.0,
      "grad_norm": 4.103657156724538,
      "learning_rate": 3.736735345968183e-07,
      "loss": 0.9277,
      "num_input_tokens_seen": 143136140,
      "step": 6721
    },
    {
      "epoch": 0.8082727108759694,
      "flos": 12496119889920.0,
      "grad_norm": 2.6965909275630287,
      "learning_rate": 3.7322026865586986e-07,
      "loss": 0.9844,
      "num_input_tokens_seen": 143154895,
      "step": 6722
    },
    {
      "epoch": 0.8083929537666086,
      "flos": 18474150113280.0,
      "grad_norm": 6.136236665998353,
      "learning_rate": 3.7276724949611206e-07,
      "loss": 0.9203,
      "num_input_tokens_seen": 143174725,
      "step": 6723
    },
    {
      "epoch": 0.8085131966572476,
      "flos": 19286976000000.0,
      "grad_norm": 3.2771105132013574,
      "learning_rate": 3.723144771862694e-07,
      "loss": 0.9312,
      "num_input_tokens_seen": 143195085,
      "step": 6724
    },
    {
      "epoch": 0.8086334395478867,
      "flos": 17058696990720.0,
      "grad_norm": 4.188233268646537,
      "learning_rate": 3.718619517950263e-07,
      "loss": 0.951,
      "num_input_tokens_seen": 143215400,
      "step": 6725
    },
    {
      "epoch": 0.8087536824385259,
      "flos": 14482951065600.0,
      "grad_norm": 3.8760716586894906,
      "learning_rate": 3.714096733910301e-07,
      "loss": 0.9527,
      "num_input_tokens_seen": 143232645,
      "step": 6726
    },
    {
      "epoch": 0.8088739253291649,
      "flos": 18446181396480.0,
      "grad_norm": 4.393324921232823,
      "learning_rate": 3.709576420428926e-07,
      "loss": 0.8883,
      "num_input_tokens_seen": 143253165,
      "step": 6727
    },
    {
      "epoch": 0.808994168219804,
      "flos": 20230447902720.0,
      "grad_norm": 3.7548100458245033,
      "learning_rate": 3.7050585781918463e-07,
      "loss": 0.9247,
      "num_input_tokens_seen": 143273185,
      "step": 6728
    },
    {
      "epoch": 0.8091144111104431,
      "flos": 12338762158080.0,
      "grad_norm": 3.179535733286866,
      "learning_rate": 3.700543207884428e-07,
      "loss": 0.8662,
      "num_input_tokens_seen": 143289815,
      "step": 6729
    },
    {
      "epoch": 0.8092346540010822,
      "flos": 22925894737920.0,
      "grad_norm": 4.45990523438638,
      "learning_rate": 3.6960303101916466e-07,
      "loss": 0.8996,
      "num_input_tokens_seen": 143309450,
      "step": 6730
    },
    {
      "epoch": 0.8093548968917212,
      "flos": 41532951490560.0,
      "grad_norm": 0.7877003081958066,
      "learning_rate": 3.6915198857981047e-07,
      "loss": 0.7754,
      "num_input_tokens_seen": 143374370,
      "step": 6731
    },
    {
      "epoch": 0.8094751397823604,
      "flos": 19497205309440.0,
      "grad_norm": 3.9177897175648835,
      "learning_rate": 3.687011935388027e-07,
      "loss": 0.8739,
      "num_input_tokens_seen": 143396985,
      "step": 6732
    },
    {
      "epoch": 0.8095953826729995,
      "flos": 17110771445760.0,
      "grad_norm": 6.428365800990007,
      "learning_rate": 3.6825064596452646e-07,
      "loss": 0.9208,
      "num_input_tokens_seen": 143417050,
      "step": 6733
    },
    {
      "epoch": 0.8097156255636385,
      "flos": 17031923957760.0,
      "grad_norm": 2.477839288199897,
      "learning_rate": 3.678003459253305e-07,
      "loss": 0.8999,
      "num_input_tokens_seen": 143437620,
      "step": 6734
    },
    {
      "epoch": 0.8098358684542777,
      "flos": 15485180989440.0,
      "grad_norm": 3.3643670893446265,
      "learning_rate": 3.673502934895236e-07,
      "loss": 0.9299,
      "num_input_tokens_seen": 143456845,
      "step": 6735
    },
    {
      "epoch": 0.8099561113449167,
      "flos": 49277826048000.0,
      "grad_norm": 0.6709754155625504,
      "learning_rate": 3.669004887253802e-07,
      "loss": 0.7817,
      "num_input_tokens_seen": 143522855,
      "step": 6736
    },
    {
      "epoch": 0.8100763542355558,
      "flos": 16769282949120.0,
      "grad_norm": 2.5462023074852693,
      "learning_rate": 3.664509317011335e-07,
      "loss": 0.9835,
      "num_input_tokens_seen": 143542910,
      "step": 6737
    },
    {
      "epoch": 0.810196597126195,
      "flos": 22565113896960.0,
      "grad_norm": 6.548802786168546,
      "learning_rate": 3.6600162248498134e-07,
      "loss": 0.9187,
      "num_input_tokens_seen": 143566260,
      "step": 6738
    },
    {
      "epoch": 0.810316840016834,
      "flos": 17714042511360.0,
      "grad_norm": 5.058949862284311,
      "learning_rate": 3.6555256114508426e-07,
      "loss": 0.9581,
      "num_input_tokens_seen": 143585775,
      "step": 6739
    },
    {
      "epoch": 0.8104370829074731,
      "flos": 19916437585920.0,
      "grad_norm": 3.5694971802242548,
      "learning_rate": 3.651037477495642e-07,
      "loss": 0.9242,
      "num_input_tokens_seen": 143606945,
      "step": 6740
    },
    {
      "epoch": 0.8105573257981122,
      "flos": 17526815232000.0,
      "grad_norm": 2.8491951927087737,
      "learning_rate": 3.6465518236650584e-07,
      "loss": 0.8594,
      "num_input_tokens_seen": 143626810,
      "step": 6741
    },
    {
      "epoch": 0.8106775686887513,
      "flos": 18762092544000.0,
      "grad_norm": 3.4057569234397245,
      "learning_rate": 3.642068650639558e-07,
      "loss": 0.9813,
      "num_input_tokens_seen": 143646275,
      "step": 6742
    },
    {
      "epoch": 0.8107978115793903,
      "flos": 19418327162880.0,
      "grad_norm": 2.921885934085073,
      "learning_rate": 3.6375879590992334e-07,
      "loss": 0.8381,
      "num_input_tokens_seen": 143666340,
      "step": 6743
    },
    {
      "epoch": 0.8109180544700295,
      "flos": 17738025615360.0,
      "grad_norm": 5.534433601711023,
      "learning_rate": 3.6331097497238173e-07,
      "loss": 0.9982,
      "num_input_tokens_seen": 143685505,
      "step": 6744
    },
    {
      "epoch": 0.8110382973606686,
      "flos": 14986518712320.0,
      "grad_norm": 3.9121233920352356,
      "learning_rate": 3.628634023192627e-07,
      "loss": 0.988,
      "num_input_tokens_seen": 143705470,
      "step": 6745
    },
    {
      "epoch": 0.8111585402513076,
      "flos": 10896352051200.0,
      "grad_norm": 2.7974019326755695,
      "learning_rate": 3.624160780184644e-07,
      "loss": 0.9427,
      "num_input_tokens_seen": 143722405,
      "step": 6746
    },
    {
      "epoch": 0.8112787831419467,
      "flos": 17135429038080.0,
      "grad_norm": 2.852795645181067,
      "learning_rate": 3.6196900213784496e-07,
      "loss": 0.9432,
      "num_input_tokens_seen": 143741440,
      "step": 6747
    },
    {
      "epoch": 0.8113990260325858,
      "flos": 14539256401920.0,
      "grad_norm": 2.8494934100268807,
      "learning_rate": 3.6152217474522527e-07,
      "loss": 1.0552,
      "num_input_tokens_seen": 143757975,
      "step": 6748
    },
    {
      "epoch": 0.8115192689232249,
      "flos": 17714625024000.0,
      "grad_norm": 2.688888191279101,
      "learning_rate": 3.6107559590838975e-07,
      "loss": 0.9236,
      "num_input_tokens_seen": 143776680,
      "step": 6749
    },
    {
      "epoch": 0.811639511813864,
      "flos": 17107460321280.0,
      "grad_norm": 7.252344738870718,
      "learning_rate": 3.606292656950822e-07,
      "loss": 0.8338,
      "num_input_tokens_seen": 143794810,
      "step": 6750
    },
    {
      "epoch": 0.8117597547045031,
      "flos": 16481217884160.0,
      "grad_norm": 3.5069180646015403,
      "learning_rate": 3.601831841730121e-07,
      "loss": 1.0456,
      "num_input_tokens_seen": 143812450,
      "step": 6751
    },
    {
      "epoch": 0.8118799975951422,
      "flos": 16377222266880.0,
      "grad_norm": 2.9698628799176396,
      "learning_rate": 3.5973735140984916e-07,
      "loss": 0.9318,
      "num_input_tokens_seen": 143832340,
      "step": 6752
    },
    {
      "epoch": 0.8120002404857812,
      "flos": 17526140743680.0,
      "grad_norm": 5.465327123569728,
      "learning_rate": 3.5929176747322607e-07,
      "loss": 0.9659,
      "num_input_tokens_seen": 143851165,
      "step": 6753
    },
    {
      "epoch": 0.8121204833764204,
      "flos": 41087957913600.0,
      "grad_norm": 0.8267255150001434,
      "learning_rate": 3.588464324307372e-07,
      "loss": 0.7664,
      "num_input_tokens_seen": 143914510,
      "step": 6754
    },
    {
      "epoch": 0.8122407262670595,
      "flos": 13807454760960.0,
      "grad_norm": 2.574566043471176,
      "learning_rate": 3.584013463499391e-07,
      "loss": 0.9466,
      "num_input_tokens_seen": 143932850,
      "step": 6755
    },
    {
      "epoch": 0.8123609691576985,
      "flos": 40375448616960.0,
      "grad_norm": 0.7230950337945954,
      "learning_rate": 3.579565092983521e-07,
      "loss": 0.8451,
      "num_input_tokens_seen": 143993690,
      "step": 6756
    },
    {
      "epoch": 0.8124812120483377,
      "flos": 14646409850880.0,
      "grad_norm": 3.03463901742274,
      "learning_rate": 3.575119213434565e-07,
      "loss": 1.0356,
      "num_input_tokens_seen": 144011925,
      "step": 6757
    },
    {
      "epoch": 0.8126014549389767,
      "flos": 15983414046720.0,
      "grad_norm": 3.187996320791766,
      "learning_rate": 3.5706758255269765e-07,
      "loss": 1.0009,
      "num_input_tokens_seen": 144030100,
      "step": 6758
    },
    {
      "epoch": 0.8127216978296158,
      "flos": 16554209587200.0,
      "grad_norm": 3.2430656990221194,
      "learning_rate": 3.566234929934795e-07,
      "loss": 0.8894,
      "num_input_tokens_seen": 144049020,
      "step": 6759
    },
    {
      "epoch": 0.812841940720255,
      "flos": 17896854958080.0,
      "grad_norm": 3.7769212141275927,
      "learning_rate": 3.561796527331706e-07,
      "loss": 0.9187,
      "num_input_tokens_seen": 144070415,
      "step": 6760
    },
    {
      "epoch": 0.812962183610894,
      "flos": 18969838510080.0,
      "grad_norm": 3.577545652171553,
      "learning_rate": 3.5573606183910163e-07,
      "loss": 0.9634,
      "num_input_tokens_seen": 144090140,
      "step": 6761
    },
    {
      "epoch": 0.8130824265015331,
      "flos": 17760966328320.0,
      "grad_norm": 2.5595799998624384,
      "learning_rate": 3.5529272037856493e-07,
      "loss": 0.9686,
      "num_input_tokens_seen": 144108075,
      "step": 6762
    },
    {
      "epoch": 0.8132026693921722,
      "flos": 48422468628480.0,
      "grad_norm": 0.7783012435663966,
      "learning_rate": 3.548496284188149e-07,
      "loss": 0.7624,
      "num_input_tokens_seen": 144168000,
      "step": 6763
    },
    {
      "epoch": 0.8133229122828113,
      "flos": 13828617277440.0,
      "grad_norm": 3.008178054592162,
      "learning_rate": 3.544067860270681e-07,
      "loss": 0.984,
      "num_input_tokens_seen": 144185295,
      "step": 6764
    },
    {
      "epoch": 0.8134431551734503,
      "flos": 14672355102720.0,
      "grad_norm": 2.8398932829135326,
      "learning_rate": 3.539641932705029e-07,
      "loss": 0.8915,
      "num_input_tokens_seen": 144203495,
      "step": 6765
    },
    {
      "epoch": 0.8135633980640895,
      "flos": 15275472875520.0,
      "grad_norm": 4.350227192174583,
      "learning_rate": 3.53521850216262e-07,
      "loss": 0.9366,
      "num_input_tokens_seen": 144222785,
      "step": 6766
    },
    {
      "epoch": 0.8136836409547286,
      "flos": 14590073856000.0,
      "grad_norm": 6.240893328862169,
      "learning_rate": 3.530797569314461e-07,
      "loss": 0.9643,
      "num_input_tokens_seen": 144241530,
      "step": 6767
    },
    {
      "epoch": 0.8138038838453676,
      "flos": 14406525603840.0,
      "grad_norm": 4.320462160467246,
      "learning_rate": 3.5263791348312235e-07,
      "loss": 0.9624,
      "num_input_tokens_seen": 144260445,
      "step": 6768
    },
    {
      "epoch": 0.8139241267360068,
      "flos": 21227987066880.0,
      "grad_norm": 3.009876365530453,
      "learning_rate": 3.521963199383171e-07,
      "loss": 0.9006,
      "num_input_tokens_seen": 144283120,
      "step": 6769
    },
    {
      "epoch": 0.8140443696266458,
      "flos": 13985269862400.0,
      "grad_norm": 3.6807031869769884,
      "learning_rate": 3.517549763640197e-07,
      "loss": 0.9466,
      "num_input_tokens_seen": 144300480,
      "step": 6770
    },
    {
      "epoch": 0.8141646125172849,
      "flos": 19338069381120.0,
      "grad_norm": 4.368701870152441,
      "learning_rate": 3.513138828271829e-07,
      "loss": 0.9032,
      "num_input_tokens_seen": 144320070,
      "step": 6771
    },
    {
      "epoch": 0.8142848554079241,
      "flos": 28330308833280.0,
      "grad_norm": 2.517042057090641,
      "learning_rate": 3.508730393947179e-07,
      "loss": 0.899,
      "num_input_tokens_seen": 144343045,
      "step": 6772
    },
    {
      "epoch": 0.8144050982985631,
      "flos": 15800111063040.0,
      "grad_norm": 3.9137542664155274,
      "learning_rate": 3.504324461335024e-07,
      "loss": 0.9004,
      "num_input_tokens_seen": 144362875,
      "step": 6773
    },
    {
      "epoch": 0.8145253411892022,
      "flos": 16375505387520.0,
      "grad_norm": 5.607926630669369,
      "learning_rate": 3.499921031103732e-07,
      "loss": 1.0703,
      "num_input_tokens_seen": 144383365,
      "step": 6774
    },
    {
      "epoch": 0.8146455840798413,
      "flos": 17661753446400.0,
      "grad_norm": 6.971838570315374,
      "learning_rate": 3.4955201039212987e-07,
      "loss": 0.9716,
      "num_input_tokens_seen": 144404005,
      "step": 6775
    },
    {
      "epoch": 0.8147658269704804,
      "flos": 14173815459840.0,
      "grad_norm": 4.023843274441194,
      "learning_rate": 3.4911216804553465e-07,
      "loss": 0.8393,
      "num_input_tokens_seen": 144422625,
      "step": 6776
    },
    {
      "epoch": 0.8148860698611194,
      "flos": 15038685143040.0,
      "grad_norm": 6.741150317637226,
      "learning_rate": 3.4867257613731017e-07,
      "loss": 0.8939,
      "num_input_tokens_seen": 144441540,
      "step": 6777
    },
    {
      "epoch": 0.8150063127517585,
      "flos": 13909242961920.0,
      "grad_norm": 2.9004721684665116,
      "learning_rate": 3.4823323473414343e-07,
      "loss": 1.0416,
      "num_input_tokens_seen": 144460780,
      "step": 6778
    },
    {
      "epoch": 0.8151265556423977,
      "flos": 16088329420800.0,
      "grad_norm": 2.7017674206775406,
      "learning_rate": 3.477941439026812e-07,
      "loss": 0.9392,
      "num_input_tokens_seen": 144478720,
      "step": 6779
    },
    {
      "epoch": 0.8152467985330367,
      "flos": 12731834572800.0,
      "grad_norm": 3.2055305194225974,
      "learning_rate": 3.473553037095349e-07,
      "loss": 0.9182,
      "num_input_tokens_seen": 144497465,
      "step": 6780
    },
    {
      "epoch": 0.8153670414236758,
      "flos": 17762713866240.0,
      "grad_norm": 3.681355545784729,
      "learning_rate": 3.469167142212743e-07,
      "loss": 1.0242,
      "num_input_tokens_seen": 144519030,
      "step": 6781
    },
    {
      "epoch": 0.8154872843143149,
      "flos": 22145084497920.0,
      "grad_norm": 5.236058206513005,
      "learning_rate": 3.4647837550443337e-07,
      "loss": 0.8235,
      "num_input_tokens_seen": 144537315,
      "step": 6782
    },
    {
      "epoch": 0.815607527204954,
      "flos": 13755043061760.0,
      "grad_norm": 12.584549261970137,
      "learning_rate": 3.460402876255086e-07,
      "loss": 0.9312,
      "num_input_tokens_seen": 144554425,
      "step": 6783
    },
    {
      "epoch": 0.815727770095593,
      "flos": 18605102714880.0,
      "grad_norm": 4.832735397360219,
      "learning_rate": 3.456024506509574e-07,
      "loss": 0.8969,
      "num_input_tokens_seen": 144575065,
      "step": 6784
    },
    {
      "epoch": 0.8158480129862322,
      "flos": 18028052828160.0,
      "grad_norm": 2.4189651235878786,
      "learning_rate": 3.4516486464719873e-07,
      "loss": 0.935,
      "num_input_tokens_seen": 144594175,
      "step": 6785
    },
    {
      "epoch": 0.8159682558768713,
      "flos": 24558230077440.0,
      "grad_norm": 3.0990959756667817,
      "learning_rate": 3.4472752968061445e-07,
      "loss": 0.8214,
      "num_input_tokens_seen": 144618325,
      "step": 6786
    },
    {
      "epoch": 0.8160884987675103,
      "flos": 13223169454080.0,
      "grad_norm": 3.7165919862323418,
      "learning_rate": 3.442904458175475e-07,
      "loss": 0.9258,
      "num_input_tokens_seen": 144635365,
      "step": 6787
    },
    {
      "epoch": 0.8162087416581495,
      "flos": 22406959042560.0,
      "grad_norm": 2.300241403764936,
      "learning_rate": 3.438536131243044e-07,
      "loss": 0.9503,
      "num_input_tokens_seen": 144656245,
      "step": 6788
    },
    {
      "epoch": 0.8163289845487885,
      "flos": 26836100198400.0,
      "grad_norm": 17.792545532151152,
      "learning_rate": 3.434170316671503e-07,
      "loss": 0.8124,
      "num_input_tokens_seen": 144680995,
      "step": 6789
    },
    {
      "epoch": 0.8164492274394276,
      "flos": 9846922383360.0,
      "grad_norm": 5.458341448052273,
      "learning_rate": 3.4298070151231583e-07,
      "loss": 1.082,
      "num_input_tokens_seen": 144696115,
      "step": 6790
    },
    {
      "epoch": 0.8165694703300668,
      "flos": 20650599936000.0,
      "grad_norm": 3.0167566261436862,
      "learning_rate": 3.425446227259916e-07,
      "loss": 0.7887,
      "num_input_tokens_seen": 144716800,
      "step": 6791
    },
    {
      "epoch": 0.8166897132207058,
      "flos": 17870511144960.0,
      "grad_norm": 6.753232676341794,
      "learning_rate": 3.421087953743296e-07,
      "loss": 1.0159,
      "num_input_tokens_seen": 144736285,
      "step": 6792
    },
    {
      "epoch": 0.8168099561113449,
      "flos": 16454935388160.0,
      "grad_norm": 2.3467268286203455,
      "learning_rate": 3.416732195234464e-07,
      "loss": 0.9827,
      "num_input_tokens_seen": 144756060,
      "step": 6793
    },
    {
      "epoch": 0.816930199001984,
      "flos": 13048021647360.0,
      "grad_norm": 3.972737509347461,
      "learning_rate": 3.4123789523941613e-07,
      "loss": 0.9835,
      "num_input_tokens_seen": 144775605,
      "step": 6794
    },
    {
      "epoch": 0.8170504418926231,
      "flos": 15091403427840.0,
      "grad_norm": 3.564794370437574,
      "learning_rate": 3.4080282258827884e-07,
      "loss": 0.8294,
      "num_input_tokens_seen": 144793700,
      "step": 6795
    },
    {
      "epoch": 0.8171706847832622,
      "flos": 13544813752320.0,
      "grad_norm": 6.358722025271881,
      "learning_rate": 3.403680016360342e-07,
      "loss": 0.9059,
      "num_input_tokens_seen": 144812025,
      "step": 6796
    },
    {
      "epoch": 0.8172909276739013,
      "flos": 15248638525440.0,
      "grad_norm": 6.666502028657585,
      "learning_rate": 3.3993343244864403e-07,
      "loss": 0.8599,
      "num_input_tokens_seen": 144831335,
      "step": 6797
    },
    {
      "epoch": 0.8174111705645404,
      "flos": 19653919211520.0,
      "grad_norm": 3.637560950748476,
      "learning_rate": 3.394991150920323e-07,
      "loss": 0.9165,
      "num_input_tokens_seen": 144854175,
      "step": 6798
    },
    {
      "epoch": 0.8175314134551794,
      "flos": 9926566993920.0,
      "grad_norm": 3.959730938912585,
      "learning_rate": 3.3906504963208396e-07,
      "loss": 0.9362,
      "num_input_tokens_seen": 144870590,
      "step": 6799
    },
    {
      "epoch": 0.8176516563458186,
      "flos": 16137981849600.0,
      "grad_norm": 2.6724392479181778,
      "learning_rate": 3.3863123613464774e-07,
      "loss": 0.8519,
      "num_input_tokens_seen": 144889210,
      "step": 6800
    },
    {
      "epoch": 0.8177718992364577,
      "flos": 15590341632000.0,
      "grad_norm": 3.772703869507851,
      "learning_rate": 3.381976746655317e-07,
      "loss": 0.9315,
      "num_input_tokens_seen": 144908685,
      "step": 6801
    },
    {
      "epoch": 0.8178921421270967,
      "flos": 15641986867200.0,
      "grad_norm": 4.7228840283159474,
      "learning_rate": 3.3776436529050756e-07,
      "loss": 0.8592,
      "num_input_tokens_seen": 144927955,
      "step": 6802
    },
    {
      "epoch": 0.8180123850177359,
      "flos": 23666679336960.0,
      "grad_norm": 2.7880755048513217,
      "learning_rate": 3.373313080753073e-07,
      "loss": 0.9111,
      "num_input_tokens_seen": 144951735,
      "step": 6803
    },
    {
      "epoch": 0.8181326279083749,
      "flos": 15695073054720.0,
      "grad_norm": 2.585143644521766,
      "learning_rate": 3.3689850308562527e-07,
      "loss": 0.9698,
      "num_input_tokens_seen": 144971900,
      "step": 6804
    },
    {
      "epoch": 0.818252870799014,
      "flos": 11105937530880.0,
      "grad_norm": 3.140680113351542,
      "learning_rate": 3.364659503871183e-07,
      "loss": 0.9623,
      "num_input_tokens_seen": 144989555,
      "step": 6805
    },
    {
      "epoch": 0.8183731136896532,
      "flos": 13308669849600.0,
      "grad_norm": 2.860760629250314,
      "learning_rate": 3.3603365004540417e-07,
      "loss": 1.0243,
      "num_input_tokens_seen": 145007570,
      "step": 6806
    },
    {
      "epoch": 0.8184933565802922,
      "flos": 18893566341120.0,
      "grad_norm": 3.2563501090038933,
      "learning_rate": 3.356016021260624e-07,
      "loss": 0.9526,
      "num_input_tokens_seen": 145027620,
      "step": 6807
    },
    {
      "epoch": 0.8186135994709313,
      "flos": 12417088450560.0,
      "grad_norm": 6.0474414806522985,
      "learning_rate": 3.35169806694634e-07,
      "loss": 0.8299,
      "num_input_tokens_seen": 145045590,
      "step": 6808
    },
    {
      "epoch": 0.8187338423615703,
      "flos": 45447510466560.0,
      "grad_norm": 0.7506951051447013,
      "learning_rate": 3.3473826381662186e-07,
      "loss": 0.8147,
      "num_input_tokens_seen": 145116450,
      "step": 6809
    },
    {
      "epoch": 0.8188540852522095,
      "flos": 12416965816320.0,
      "grad_norm": 4.14722834492463,
      "learning_rate": 3.3430697355749216e-07,
      "loss": 1.0075,
      "num_input_tokens_seen": 145133860,
      "step": 6810
    },
    {
      "epoch": 0.8189743281428485,
      "flos": 10162189701120.0,
      "grad_norm": 6.873975920265693,
      "learning_rate": 3.3387593598266907e-07,
      "loss": 0.9311,
      "num_input_tokens_seen": 145150190,
      "step": 6811
    },
    {
      "epoch": 0.8190945710334876,
      "flos": 17843094282240.0,
      "grad_norm": 2.57450862751647,
      "learning_rate": 3.3344515115754225e-07,
      "loss": 0.9715,
      "num_input_tokens_seen": 145168890,
      "step": 6812
    },
    {
      "epoch": 0.8192148139241268,
      "flos": 15275319582720.0,
      "grad_norm": 4.44837350159498,
      "learning_rate": 3.33014619147461e-07,
      "loss": 0.9839,
      "num_input_tokens_seen": 145186635,
      "step": 6813
    },
    {
      "epoch": 0.8193350568147658,
      "flos": 17032567787520.0,
      "grad_norm": 3.6442121261760887,
      "learning_rate": 3.325843400177362e-07,
      "loss": 0.8967,
      "num_input_tokens_seen": 145207695,
      "step": 6814
    },
    {
      "epoch": 0.8194552997054049,
      "flos": 14594366054400.0,
      "grad_norm": 4.159512952756655,
      "learning_rate": 3.32154313833642e-07,
      "loss": 0.9224,
      "num_input_tokens_seen": 145227570,
      "step": 6815
    },
    {
      "epoch": 0.819575542596044,
      "flos": 18527113666560.0,
      "grad_norm": 4.279002692706497,
      "learning_rate": 3.3172454066041164e-07,
      "loss": 0.7793,
      "num_input_tokens_seen": 145246795,
      "step": 6816
    },
    {
      "epoch": 0.8196957854866831,
      "flos": 20730060595200.0,
      "grad_norm": 2.892026983966223,
      "learning_rate": 3.3129502056324234e-07,
      "loss": 0.937,
      "num_input_tokens_seen": 145267880,
      "step": 6817
    },
    {
      "epoch": 0.8198160283773221,
      "flos": 49434670694400.0,
      "grad_norm": 0.7838517925103987,
      "learning_rate": 3.3086575360729165e-07,
      "loss": 0.8151,
      "num_input_tokens_seen": 145325135,
      "step": 6818
    },
    {
      "epoch": 0.8199362712679613,
      "flos": 11761620295680.0,
      "grad_norm": 4.710188418626673,
      "learning_rate": 3.3043673985767906e-07,
      "loss": 0.9045,
      "num_input_tokens_seen": 145343920,
      "step": 6819
    },
    {
      "epoch": 0.8200565141586004,
      "flos": 15454698270720.0,
      "grad_norm": 3.94756314884866,
      "learning_rate": 3.3000797937948564e-07,
      "loss": 0.9578,
      "num_input_tokens_seen": 145361935,
      "step": 6820
    },
    {
      "epoch": 0.8201767570492394,
      "flos": 49991539138560.0,
      "grad_norm": 0.9257494442139971,
      "learning_rate": 3.295794722377534e-07,
      "loss": 0.8696,
      "num_input_tokens_seen": 145425260,
      "step": 6821
    },
    {
      "epoch": 0.8202969999398786,
      "flos": 16428530257920.0,
      "grad_norm": 3.3481880753717412,
      "learning_rate": 3.291512184974876e-07,
      "loss": 0.9888,
      "num_input_tokens_seen": 145445370,
      "step": 6822
    },
    {
      "epoch": 0.8204172428305176,
      "flos": 20100169789440.0,
      "grad_norm": 3.377990104529803,
      "learning_rate": 3.2872321822365346e-07,
      "loss": 0.8606,
      "num_input_tokens_seen": 145465305,
      "step": 6823
    },
    {
      "epoch": 0.8205374857211567,
      "flos": 14829651517440.0,
      "grad_norm": 3.567120817867017,
      "learning_rate": 3.282954714811783e-07,
      "loss": 0.9255,
      "num_input_tokens_seen": 145483930,
      "step": 6824
    },
    {
      "epoch": 0.8206577286117959,
      "flos": 9270761594880.0,
      "grad_norm": 3.967442695189411,
      "learning_rate": 3.2786797833495093e-07,
      "loss": 0.9022,
      "num_input_tokens_seen": 145499005,
      "step": 6825
    },
    {
      "epoch": 0.8207779715024349,
      "flos": 17976530227200.0,
      "grad_norm": 2.8997829224468603,
      "learning_rate": 3.274407388498213e-07,
      "loss": 0.9221,
      "num_input_tokens_seen": 145516855,
      "step": 6826
    },
    {
      "epoch": 0.820898214393074,
      "flos": 13912094208000.0,
      "grad_norm": 5.650382185981427,
      "learning_rate": 3.270137530906021e-07,
      "loss": 0.9336,
      "num_input_tokens_seen": 145535810,
      "step": 6827
    },
    {
      "epoch": 0.8210184572837131,
      "flos": 11027549921280.0,
      "grad_norm": 3.421199923599369,
      "learning_rate": 3.265870211220665e-07,
      "loss": 1.0208,
      "num_input_tokens_seen": 145553365,
      "step": 6828
    },
    {
      "epoch": 0.8211387001743522,
      "flos": 14777117184000.0,
      "grad_norm": 3.1389878594053435,
      "learning_rate": 3.2616054300894934e-07,
      "loss": 1.0034,
      "num_input_tokens_seen": 145572535,
      "step": 6829
    },
    {
      "epoch": 0.8212589430649913,
      "flos": 19728597135360.0,
      "grad_norm": 5.766812168703896,
      "learning_rate": 3.2573431881594693e-07,
      "loss": 1.0426,
      "num_input_tokens_seen": 145591800,
      "step": 6830
    },
    {
      "epoch": 0.8213791859556304,
      "flos": 15956947599360.0,
      "grad_norm": 7.752743881311248,
      "learning_rate": 3.2530834860771663e-07,
      "loss": 0.8453,
      "num_input_tokens_seen": 145610900,
      "step": 6831
    },
    {
      "epoch": 0.8214994288462695,
      "flos": 11787780157440.0,
      "grad_norm": 4.429265077824838,
      "learning_rate": 3.248826324488794e-07,
      "loss": 0.9286,
      "num_input_tokens_seen": 145627915,
      "step": 6832
    },
    {
      "epoch": 0.8216196717369085,
      "flos": 17944024043520.0,
      "grad_norm": 3.1339511086472935,
      "learning_rate": 3.244571704040138e-07,
      "loss": 1.0693,
      "num_input_tokens_seen": 145647795,
      "step": 6833
    },
    {
      "epoch": 0.8217399146275477,
      "flos": 18052495810560.0,
      "grad_norm": 3.5218304608430366,
      "learning_rate": 3.2403196253766374e-07,
      "loss": 0.9366,
      "num_input_tokens_seen": 145666595,
      "step": 6834
    },
    {
      "epoch": 0.8218601575181868,
      "flos": 18238006210560.0,
      "grad_norm": 4.946078417435957,
      "learning_rate": 3.2360700891433254e-07,
      "loss": 0.98,
      "num_input_tokens_seen": 145685340,
      "step": 6835
    },
    {
      "epoch": 0.8219804004088258,
      "flos": 48449394954240.0,
      "grad_norm": 0.8007808034011712,
      "learning_rate": 3.231823095984847e-07,
      "loss": 0.7828,
      "num_input_tokens_seen": 145739700,
      "step": 6836
    },
    {
      "epoch": 0.822100643299465,
      "flos": 13807178833920.0,
      "grad_norm": 4.644240284396778,
      "learning_rate": 3.2275786465454814e-07,
      "loss": 0.9511,
      "num_input_tokens_seen": 145756070,
      "step": 6837
    },
    {
      "epoch": 0.822220886190104,
      "flos": 17556562145280.0,
      "grad_norm": 4.8995336738875075,
      "learning_rate": 3.2233367414690917e-07,
      "loss": 0.951,
      "num_input_tokens_seen": 145777980,
      "step": 6838
    },
    {
      "epoch": 0.8223411290807431,
      "flos": 19811552870400.0,
      "grad_norm": 3.9137266094881893,
      "learning_rate": 3.219097381399183e-07,
      "loss": 1.0274,
      "num_input_tokens_seen": 145794875,
      "step": 6839
    },
    {
      "epoch": 0.8224613719713821,
      "flos": 16504802426880.0,
      "grad_norm": 2.910683480449954,
      "learning_rate": 3.2148605669788584e-07,
      "loss": 0.9967,
      "num_input_tokens_seen": 145814485,
      "step": 6840
    },
    {
      "epoch": 0.8225816148620213,
      "flos": 11106366750720.0,
      "grad_norm": 4.14816139581119,
      "learning_rate": 3.2106262988508405e-07,
      "loss": 0.9563,
      "num_input_tokens_seen": 145832255,
      "step": 6841
    },
    {
      "epoch": 0.8227018577526604,
      "flos": 13124968304640.0,
      "grad_norm": 3.6634358274811847,
      "learning_rate": 3.206394577657465e-07,
      "loss": 0.9332,
      "num_input_tokens_seen": 145849755,
      "step": 6842
    },
    {
      "epoch": 0.8228221006432994,
      "flos": 15799681843200.0,
      "grad_norm": 3.6855424760671944,
      "learning_rate": 3.202165404040675e-07,
      "loss": 0.895,
      "num_input_tokens_seen": 145867395,
      "step": 6843
    },
    {
      "epoch": 0.8229423435339386,
      "flos": 17136961966080.0,
      "grad_norm": 4.188621940055448,
      "learning_rate": 3.1979387786420396e-07,
      "loss": 0.9333,
      "num_input_tokens_seen": 145887355,
      "step": 6844
    },
    {
      "epoch": 0.8230625864245776,
      "flos": 16979573575680.0,
      "grad_norm": 11.259744111807288,
      "learning_rate": 3.1937147021027346e-07,
      "loss": 1.0199,
      "num_input_tokens_seen": 145905530,
      "step": 6845
    },
    {
      "epoch": 0.8231828293152167,
      "flos": 11735000555520.0,
      "grad_norm": 3.9599730993570947,
      "learning_rate": 3.189493175063547e-07,
      "loss": 0.9583,
      "num_input_tokens_seen": 145922485,
      "step": 6846
    },
    {
      "epoch": 0.8233030722058559,
      "flos": 13387272069120.0,
      "grad_norm": 3.071769543348167,
      "learning_rate": 3.1852741981648776e-07,
      "loss": 0.8676,
      "num_input_tokens_seen": 145940855,
      "step": 6847
    },
    {
      "epoch": 0.8234233150964949,
      "flos": 20257067642880.0,
      "grad_norm": 3.3758038164058575,
      "learning_rate": 3.1810577720467404e-07,
      "loss": 0.8842,
      "num_input_tokens_seen": 145962305,
      "step": 6848
    },
    {
      "epoch": 0.823543557987134,
      "flos": 24085421076480.0,
      "grad_norm": 2.2924656486641575,
      "learning_rate": 3.176843897348769e-07,
      "loss": 0.7626,
      "num_input_tokens_seen": 145985220,
      "step": 6849
    },
    {
      "epoch": 0.8236638008777731,
      "flos": 12102495621120.0,
      "grad_norm": 3.594787580557245,
      "learning_rate": 3.1726325747102034e-07,
      "loss": 0.9495,
      "num_input_tokens_seen": 146003315,
      "step": 6850
    },
    {
      "epoch": 0.8237840437684122,
      "flos": 44117819228160.0,
      "grad_norm": 2.2632429557512848,
      "learning_rate": 3.1684238047698974e-07,
      "loss": 0.8356,
      "num_input_tokens_seen": 146031305,
      "step": 6851
    },
    {
      "epoch": 0.8239042866590512,
      "flos": 19445100195840.0,
      "grad_norm": 6.136066015970147,
      "learning_rate": 3.1642175881663155e-07,
      "loss": 0.7165,
      "num_input_tokens_seen": 146050755,
      "step": 6852
    },
    {
      "epoch": 0.8240245295496904,
      "flos": 15432891924480.0,
      "grad_norm": 4.755279217373115,
      "learning_rate": 3.160013925537537e-07,
      "loss": 1.0351,
      "num_input_tokens_seen": 146071310,
      "step": 6853
    },
    {
      "epoch": 0.8241447724403295,
      "flos": 14198595686400.0,
      "grad_norm": 3.1000794677393073,
      "learning_rate": 3.155812817521266e-07,
      "loss": 0.9378,
      "num_input_tokens_seen": 146091405,
      "step": 6854
    },
    {
      "epoch": 0.8242650153309685,
      "flos": 15825657753600.0,
      "grad_norm": 5.83239295282411,
      "learning_rate": 3.151614264754787e-07,
      "loss": 0.9624,
      "num_input_tokens_seen": 146109070,
      "step": 6855
    },
    {
      "epoch": 0.8243852582216077,
      "flos": 15852308152320.0,
      "grad_norm": 2.8900205237903913,
      "learning_rate": 3.147418267875035e-07,
      "loss": 0.986,
      "num_input_tokens_seen": 146126920,
      "step": 6856
    },
    {
      "epoch": 0.8245055011122467,
      "flos": 17530800844800.0,
      "grad_norm": 3.7754844657811772,
      "learning_rate": 3.1432248275185315e-07,
      "loss": 0.838,
      "num_input_tokens_seen": 146147150,
      "step": 6857
    },
    {
      "epoch": 0.8246257440028858,
      "flos": 12385440706560.0,
      "grad_norm": 3.5975238390484603,
      "learning_rate": 3.139033944321412e-07,
      "loss": 0.9667,
      "num_input_tokens_seen": 146164230,
      "step": 6858
    },
    {
      "epoch": 0.824745986893525,
      "flos": 17792889999360.0,
      "grad_norm": 4.80634935373069,
      "learning_rate": 3.1348456189194507e-07,
      "loss": 0.985,
      "num_input_tokens_seen": 146184410,
      "step": 6859
    },
    {
      "epoch": 0.824866229784164,
      "flos": 13310386728960.0,
      "grad_norm": 2.4011147929962675,
      "learning_rate": 3.1306598519479876e-07,
      "loss": 1.0248,
      "num_input_tokens_seen": 146203950,
      "step": 6860
    },
    {
      "epoch": 0.8249864726748031,
      "flos": 16953444372480.0,
      "grad_norm": 3.3535847287398175,
      "learning_rate": 3.1264766440420177e-07,
      "loss": 0.9721,
      "num_input_tokens_seen": 146226140,
      "step": 6861
    },
    {
      "epoch": 0.8251067155654422,
      "flos": 14409990021120.0,
      "grad_norm": 3.5484041791096392,
      "learning_rate": 3.122295995836124e-07,
      "loss": 0.8763,
      "num_input_tokens_seen": 146245730,
      "step": 6862
    },
    {
      "epoch": 0.8252269584560813,
      "flos": 17792522096640.0,
      "grad_norm": 2.717315646237857,
      "learning_rate": 3.118117907964508e-07,
      "loss": 0.9608,
      "num_input_tokens_seen": 146267395,
      "step": 6863
    },
    {
      "epoch": 0.8253472013467203,
      "flos": 12128471531520.0,
      "grad_norm": 22.219442922783372,
      "learning_rate": 3.1139423810609856e-07,
      "loss": 1.0045,
      "num_input_tokens_seen": 146283810,
      "step": 6864
    },
    {
      "epoch": 0.8254674442373595,
      "flos": 15927016734720.0,
      "grad_norm": 2.9507687165163556,
      "learning_rate": 3.1097694157589714e-07,
      "loss": 0.942,
      "num_input_tokens_seen": 146303415,
      "step": 6865
    },
    {
      "epoch": 0.8255876871279986,
      "flos": 17631761264640.0,
      "grad_norm": 3.5986644810767343,
      "learning_rate": 3.105599012691511e-07,
      "loss": 0.9382,
      "num_input_tokens_seen": 146321565,
      "step": 6866
    },
    {
      "epoch": 0.8257079300186376,
      "flos": 19550383472640.0,
      "grad_norm": 4.355249565322878,
      "learning_rate": 3.101431172491249e-07,
      "loss": 1.0136,
      "num_input_tokens_seen": 146342830,
      "step": 6867
    },
    {
      "epoch": 0.8258281729092768,
      "flos": 11656520970240.0,
      "grad_norm": 3.994143251450848,
      "learning_rate": 3.097265895790444e-07,
      "loss": 0.9093,
      "num_input_tokens_seen": 146360760,
      "step": 6868
    },
    {
      "epoch": 0.8259484157999158,
      "flos": 15222570639360.0,
      "grad_norm": 3.7497896771259938,
      "learning_rate": 3.093103183220962e-07,
      "loss": 1.0263,
      "num_input_tokens_seen": 146380525,
      "step": 6869
    },
    {
      "epoch": 0.8260686586905549,
      "flos": 41738765967360.0,
      "grad_norm": 0.9116680049421494,
      "learning_rate": 3.0889430354142796e-07,
      "loss": 0.8353,
      "num_input_tokens_seen": 146441755,
      "step": 6870
    },
    {
      "epoch": 0.826188901581194,
      "flos": 19601875415040.0,
      "grad_norm": 4.333763223500814,
      "learning_rate": 3.084785453001497e-07,
      "loss": 0.8903,
      "num_input_tokens_seen": 146462390,
      "step": 6871
    },
    {
      "epoch": 0.8263091444718331,
      "flos": 16848406364160.0,
      "grad_norm": 2.997851556137615,
      "learning_rate": 3.080630436613314e-07,
      "loss": 1.0049,
      "num_input_tokens_seen": 146479880,
      "step": 6872
    },
    {
      "epoch": 0.8264293873624722,
      "flos": 12154784686080.0,
      "grad_norm": 3.8699736451821343,
      "learning_rate": 3.076477986880039e-07,
      "loss": 1.0428,
      "num_input_tokens_seen": 146497395,
      "step": 6873
    },
    {
      "epoch": 0.8265496302531112,
      "flos": 17137912381440.0,
      "grad_norm": 3.555022037400242,
      "learning_rate": 3.0723281044315986e-07,
      "loss": 0.8851,
      "num_input_tokens_seen": 146519070,
      "step": 6874
    },
    {
      "epoch": 0.8266698731437504,
      "flos": 9951776440320.0,
      "grad_norm": 4.137049790519602,
      "learning_rate": 3.068180789897521e-07,
      "loss": 0.954,
      "num_input_tokens_seen": 146537200,
      "step": 6875
    },
    {
      "epoch": 0.8267901160343895,
      "flos": 21938196971520.0,
      "grad_norm": 2.305216169336799,
      "learning_rate": 3.064036043906966e-07,
      "loss": 1.0104,
      "num_input_tokens_seen": 146560360,
      "step": 6876
    },
    {
      "epoch": 0.8269103589250285,
      "flos": 29014788096000.0,
      "grad_norm": 3.142006489582511,
      "learning_rate": 3.059893867088668e-07,
      "loss": 0.8714,
      "num_input_tokens_seen": 146584225,
      "step": 6877
    },
    {
      "epoch": 0.8270306018156677,
      "flos": 21568433172480.0,
      "grad_norm": 5.612556426650336,
      "learning_rate": 3.055754260071004e-07,
      "loss": 0.8648,
      "num_input_tokens_seen": 146606240,
      "step": 6878
    },
    {
      "epoch": 0.8271508447063067,
      "flos": 17948132290560.0,
      "grad_norm": 6.490626692645422,
      "learning_rate": 3.051617223481948e-07,
      "loss": 0.9339,
      "num_input_tokens_seen": 146627280,
      "step": 6879
    },
    {
      "epoch": 0.8272710875969458,
      "flos": 12443309629440.0,
      "grad_norm": 3.586812097609738,
      "learning_rate": 3.047482757949078e-07,
      "loss": 0.9449,
      "num_input_tokens_seen": 146644630,
      "step": 6880
    },
    {
      "epoch": 0.827391330487585,
      "flos": 14016365752320.0,
      "grad_norm": 2.7842338072388215,
      "learning_rate": 3.043350864099605e-07,
      "loss": 1.0422,
      "num_input_tokens_seen": 146662910,
      "step": 6881
    },
    {
      "epoch": 0.827511573378224,
      "flos": 11917414440960.0,
      "grad_norm": 5.2475636596598445,
      "learning_rate": 3.039221542560315e-07,
      "loss": 0.9985,
      "num_input_tokens_seen": 146679195,
      "step": 6882
    },
    {
      "epoch": 0.8276318162688631,
      "flos": 13020144906240.0,
      "grad_norm": 4.1955034436295975,
      "learning_rate": 3.0350947939576356e-07,
      "loss": 0.9378,
      "num_input_tokens_seen": 146698070,
      "step": 6883
    },
    {
      "epoch": 0.8277520591595022,
      "flos": 13726706442240.0,
      "grad_norm": 3.774464974794079,
      "learning_rate": 3.0309706189175876e-07,
      "loss": 0.9192,
      "num_input_tokens_seen": 146717625,
      "step": 6884
    },
    {
      "epoch": 0.8278723020501413,
      "flos": 48635702476800.0,
      "grad_norm": 0.7864478954776203,
      "learning_rate": 3.0268490180658045e-07,
      "loss": 0.7872,
      "num_input_tokens_seen": 146780125,
      "step": 6885
    },
    {
      "epoch": 0.8279925449407803,
      "flos": 12888119255040.0,
      "grad_norm": 4.0482729113862215,
      "learning_rate": 3.0227299920275305e-07,
      "loss": 0.9692,
      "num_input_tokens_seen": 146796160,
      "step": 6886
    },
    {
      "epoch": 0.8281127878314195,
      "flos": 14645949972480.0,
      "grad_norm": 4.308990377256494,
      "learning_rate": 3.018613541427613e-07,
      "loss": 1.0432,
      "num_input_tokens_seen": 146815400,
      "step": 6887
    },
    {
      "epoch": 0.8282330307220586,
      "flos": 12757841141760.0,
      "grad_norm": 2.440505497897129,
      "learning_rate": 3.0144996668905243e-07,
      "loss": 0.9273,
      "num_input_tokens_seen": 146832500,
      "step": 6888
    },
    {
      "epoch": 0.8283532736126976,
      "flos": 14252601630720.0,
      "grad_norm": 5.34758366715945,
      "learning_rate": 3.010388369040331e-07,
      "loss": 1.0168,
      "num_input_tokens_seen": 146850880,
      "step": 6889
    },
    {
      "epoch": 0.8284735165033368,
      "flos": 22721674506240.0,
      "grad_norm": 4.196331674680373,
      "learning_rate": 3.0062796485007156e-07,
      "loss": 1.0262,
      "num_input_tokens_seen": 146871540,
      "step": 6890
    },
    {
      "epoch": 0.8285937593939758,
      "flos": 18972291194880.0,
      "grad_norm": 6.63872531421563,
      "learning_rate": 3.002173505894965e-07,
      "loss": 0.8394,
      "num_input_tokens_seen": 146890410,
      "step": 6891
    },
    {
      "epoch": 0.8287140022846149,
      "flos": 14331357143040.0,
      "grad_norm": 10.014783993551733,
      "learning_rate": 2.998069941845973e-07,
      "loss": 0.8084,
      "num_input_tokens_seen": 146909200,
      "step": 6892
    },
    {
      "epoch": 0.8288342451752541,
      "flos": 50673504399360.0,
      "grad_norm": 0.756554454018542,
      "learning_rate": 2.993968956976258e-07,
      "loss": 0.8109,
      "num_input_tokens_seen": 146976665,
      "step": 6893
    },
    {
      "epoch": 0.8289544880658931,
      "flos": 17478174535680.0,
      "grad_norm": 6.713181285986222,
      "learning_rate": 2.9898705519079313e-07,
      "loss": 0.8908,
      "num_input_tokens_seen": 146995490,
      "step": 6894
    },
    {
      "epoch": 0.8290747309565322,
      "flos": 15825994997760.0,
      "grad_norm": 4.430315245698609,
      "learning_rate": 2.985774727262715e-07,
      "loss": 0.9361,
      "num_input_tokens_seen": 147014055,
      "step": 6895
    },
    {
      "epoch": 0.8291949738471713,
      "flos": 16531483484160.0,
      "grad_norm": 4.346636355202735,
      "learning_rate": 2.981681483661949e-07,
      "loss": 1.0026,
      "num_input_tokens_seen": 147033360,
      "step": 6896
    },
    {
      "epoch": 0.8293152167378104,
      "flos": 37589266882560.0,
      "grad_norm": 2.3094975337039902,
      "learning_rate": 2.9775908217265633e-07,
      "loss": 0.8969,
      "num_input_tokens_seen": 147058315,
      "step": 6897
    },
    {
      "epoch": 0.8294354596284494,
      "flos": 45356307578880.0,
      "grad_norm": 0.8264530145428958,
      "learning_rate": 2.9735027420771253e-07,
      "loss": 0.7129,
      "num_input_tokens_seen": 147118370,
      "step": 6898
    },
    {
      "epoch": 0.8295557025190886,
      "flos": 17661201592320.0,
      "grad_norm": 3.9385471465361794,
      "learning_rate": 2.969417245333774e-07,
      "loss": 0.8994,
      "num_input_tokens_seen": 147137470,
      "step": 6899
    },
    {
      "epoch": 0.8296759454097277,
      "flos": 17870695096320.0,
      "grad_norm": 5.238259220859254,
      "learning_rate": 2.9653343321162915e-07,
      "loss": 0.9644,
      "num_input_tokens_seen": 147156700,
      "step": 6900
    },
    {
      "epoch": 0.8297961883003667,
      "flos": 17161956802560.0,
      "grad_norm": 3.460629141613723,
      "learning_rate": 2.9612540030440446e-07,
      "loss": 0.8386,
      "num_input_tokens_seen": 147176965,
      "step": 6901
    },
    {
      "epoch": 0.8299164311910058,
      "flos": 48296544030720.0,
      "grad_norm": 0.8504019011493292,
      "learning_rate": 2.9571762587360206e-07,
      "loss": 0.834,
      "num_input_tokens_seen": 147233070,
      "step": 6902
    },
    {
      "epoch": 0.8300366740816449,
      "flos": 17949910487040.0,
      "grad_norm": 2.4779564818224036,
      "learning_rate": 2.953101099810806e-07,
      "loss": 0.932,
      "num_input_tokens_seen": 147252395,
      "step": 6903
    },
    {
      "epoch": 0.830156916972284,
      "flos": 12784276930560.0,
      "grad_norm": 4.306098720512095,
      "learning_rate": 2.9490285268865965e-07,
      "loss": 1.0281,
      "num_input_tokens_seen": 147269605,
      "step": 6904
    },
    {
      "epoch": 0.830277159862923,
      "flos": 18735902023680.0,
      "grad_norm": 5.526979683791938,
      "learning_rate": 2.9449585405812085e-07,
      "loss": 0.9986,
      "num_input_tokens_seen": 147286705,
      "step": 6905
    },
    {
      "epoch": 0.8303974027535622,
      "flos": 14148299427840.0,
      "grad_norm": 3.521721466427629,
      "learning_rate": 2.940891141512043e-07,
      "loss": 0.9334,
      "num_input_tokens_seen": 147304445,
      "step": 6906
    },
    {
      "epoch": 0.8305176456442013,
      "flos": 12155121930240.0,
      "grad_norm": 4.343915977325805,
      "learning_rate": 2.9368263302961385e-07,
      "loss": 0.8987,
      "num_input_tokens_seen": 147322865,
      "step": 6907
    },
    {
      "epoch": 0.8306378885348403,
      "flos": 18236289331200.0,
      "grad_norm": 4.09180340094945,
      "learning_rate": 2.9327641075501075e-07,
      "loss": 0.9926,
      "num_input_tokens_seen": 147341575,
      "step": 6908
    },
    {
      "epoch": 0.8307581314254795,
      "flos": 24216097751040.0,
      "grad_norm": 4.468180373373374,
      "learning_rate": 2.9287044738901866e-07,
      "loss": 0.8512,
      "num_input_tokens_seen": 147359280,
      "step": 6909
    },
    {
      "epoch": 0.8308783743161186,
      "flos": 12439998504960.0,
      "grad_norm": 5.2933546763133155,
      "learning_rate": 2.9246474299322274e-07,
      "loss": 1.1128,
      "num_input_tokens_seen": 147374560,
      "step": 6910
    },
    {
      "epoch": 0.8309986172067576,
      "flos": 49709176565760.0,
      "grad_norm": 0.9004580127050086,
      "learning_rate": 2.920592976291678e-07,
      "loss": 0.8545,
      "num_input_tokens_seen": 147431610,
      "step": 6911
    },
    {
      "epoch": 0.8311188600973968,
      "flos": 15852124200960.0,
      "grad_norm": 2.9542873456559917,
      "learning_rate": 2.916541113583595e-07,
      "loss": 0.9998,
      "num_input_tokens_seen": 147449830,
      "step": 6912
    },
    {
      "epoch": 0.8312391029880358,
      "flos": 13308148654080.0,
      "grad_norm": 10.504648542012767,
      "learning_rate": 2.912491842422642e-07,
      "loss": 0.8462,
      "num_input_tokens_seen": 147467255,
      "step": 6913
    },
    {
      "epoch": 0.8313593458786749,
      "flos": 14462922915840.0,
      "grad_norm": 16.793977766121532,
      "learning_rate": 2.9084451634230857e-07,
      "loss": 0.8977,
      "num_input_tokens_seen": 147486275,
      "step": 6914
    },
    {
      "epoch": 0.831479588769314,
      "flos": 22905437368320.0,
      "grad_norm": 4.254842803956094,
      "learning_rate": 2.9044010771988125e-07,
      "loss": 0.9128,
      "num_input_tokens_seen": 147505810,
      "step": 6915
    },
    {
      "epoch": 0.8315998316599531,
      "flos": 32292742041600.0,
      "grad_norm": 3.0935040000444505,
      "learning_rate": 2.900359584363303e-07,
      "loss": 0.9298,
      "num_input_tokens_seen": 147528635,
      "step": 6916
    },
    {
      "epoch": 0.8317200745505922,
      "flos": 13015454146560.0,
      "grad_norm": 3.3326969174973717,
      "learning_rate": 2.8963206855296494e-07,
      "loss": 1.0314,
      "num_input_tokens_seen": 147544595,
      "step": 6917
    },
    {
      "epoch": 0.8318403174412313,
      "flos": 17215257600000.0,
      "grad_norm": 2.704827362446341,
      "learning_rate": 2.892284381310548e-07,
      "loss": 0.9695,
      "num_input_tokens_seen": 147565730,
      "step": 6918
    },
    {
      "epoch": 0.8319605603318704,
      "flos": 15930419834880.0,
      "grad_norm": 5.421753104211078,
      "learning_rate": 2.888250672318302e-07,
      "loss": 0.91,
      "num_input_tokens_seen": 147582850,
      "step": 6919
    },
    {
      "epoch": 0.8320808032225094,
      "flos": 26707569623040.0,
      "grad_norm": 3.012753240934159,
      "learning_rate": 2.884219559164831e-07,
      "loss": 0.8867,
      "num_input_tokens_seen": 147605715,
      "step": 6920
    },
    {
      "epoch": 0.8322010461131486,
      "flos": 9008641781760.0,
      "grad_norm": 3.8678986483119333,
      "learning_rate": 2.880191042461635e-07,
      "loss": 1.0023,
      "num_input_tokens_seen": 147621375,
      "step": 6921
    },
    {
      "epoch": 0.8323212890037877,
      "flos": 11184815677440.0,
      "grad_norm": 3.2951613759188136,
      "learning_rate": 2.876165122819849e-07,
      "loss": 0.9954,
      "num_input_tokens_seen": 147639075,
      "step": 6922
    },
    {
      "epoch": 0.8324415318944267,
      "flos": 15426852188160.0,
      "grad_norm": 2.5614344544685452,
      "learning_rate": 2.872141800850201e-07,
      "loss": 0.9827,
      "num_input_tokens_seen": 147655970,
      "step": 6923
    },
    {
      "epoch": 0.8325617747850659,
      "flos": 24395752366080.0,
      "grad_norm": 4.9424502510033,
      "learning_rate": 2.868121077163024e-07,
      "loss": 0.9249,
      "num_input_tokens_seen": 147675245,
      "step": 6924
    },
    {
      "epoch": 0.8326820176757049,
      "flos": 13020328857600.0,
      "grad_norm": 4.757050150913448,
      "learning_rate": 2.864102952368257e-07,
      "loss": 0.9165,
      "num_input_tokens_seen": 147692890,
      "step": 6925
    },
    {
      "epoch": 0.832802260566344,
      "flos": 25684729036800.0,
      "grad_norm": 2.432870037792519,
      "learning_rate": 2.860087427075444e-07,
      "loss": 0.7814,
      "num_input_tokens_seen": 147716860,
      "step": 6926
    },
    {
      "epoch": 0.8329225034569832,
      "flos": 10055465472000.0,
      "grad_norm": 7.650391871569339,
      "learning_rate": 2.856074501893744e-07,
      "loss": 1.0404,
      "num_input_tokens_seen": 147731780,
      "step": 6927
    },
    {
      "epoch": 0.8330427463476222,
      "flos": 12810467450880.0,
      "grad_norm": 2.6451249463400566,
      "learning_rate": 2.8520641774319054e-07,
      "loss": 1.0112,
      "num_input_tokens_seen": 147749590,
      "step": 6928
    },
    {
      "epoch": 0.8331629892382613,
      "flos": 12784123637760.0,
      "grad_norm": 3.5770428616176937,
      "learning_rate": 2.848056454298309e-07,
      "loss": 0.9405,
      "num_input_tokens_seen": 147766635,
      "step": 6929
    },
    {
      "epoch": 0.8332832321289004,
      "flos": 12364860702720.0,
      "grad_norm": 4.0414479514388715,
      "learning_rate": 2.844051333100905e-07,
      "loss": 0.8345,
      "num_input_tokens_seen": 147783900,
      "step": 6930
    },
    {
      "epoch": 0.8334034750195395,
      "flos": 10659625635840.0,
      "grad_norm": 3.288194990488124,
      "learning_rate": 2.840048814447269e-07,
      "loss": 1.0287,
      "num_input_tokens_seen": 147801785,
      "step": 6931
    },
    {
      "epoch": 0.8335237179101785,
      "flos": 13780436459520.0,
      "grad_norm": 6.100210667708918,
      "learning_rate": 2.836048898944587e-07,
      "loss": 0.9242,
      "num_input_tokens_seen": 147819930,
      "step": 6932
    },
    {
      "epoch": 0.8336439608008177,
      "flos": 15458438615040.0,
      "grad_norm": 10.771467907115698,
      "learning_rate": 2.832051587199642e-07,
      "loss": 0.9123,
      "num_input_tokens_seen": 147836905,
      "step": 6933
    },
    {
      "epoch": 0.8337642036914568,
      "flos": 42730050785280.0,
      "grad_norm": 0.8262893917833471,
      "learning_rate": 2.828056879818821e-07,
      "loss": 0.7879,
      "num_input_tokens_seen": 147895700,
      "step": 6934
    },
    {
      "epoch": 0.8338844465820958,
      "flos": 19339449016320.0,
      "grad_norm": 3.2048844134217314,
      "learning_rate": 2.824064777408117e-07,
      "loss": 1.0216,
      "num_input_tokens_seen": 147915210,
      "step": 6935
    },
    {
      "epoch": 0.8340046894727349,
      "flos": 21724656537600.0,
      "grad_norm": 3.059610365950322,
      "learning_rate": 2.8200752805731263e-07,
      "loss": 0.9572,
      "num_input_tokens_seen": 147937920,
      "step": 6936
    },
    {
      "epoch": 0.834124932363374,
      "flos": 19313626398720.0,
      "grad_norm": 2.3014419720925887,
      "learning_rate": 2.8160883899190625e-07,
      "loss": 0.9962,
      "num_input_tokens_seen": 147960910,
      "step": 6937
    },
    {
      "epoch": 0.8342451752540131,
      "flos": 17476151070720.0,
      "grad_norm": 4.060084637290569,
      "learning_rate": 2.8121041060507234e-07,
      "loss": 0.9396,
      "num_input_tokens_seen": 147979660,
      "step": 6938
    },
    {
      "epoch": 0.8343654181446521,
      "flos": 18946284625920.0,
      "grad_norm": 4.986147191806452,
      "learning_rate": 2.808122429572528e-07,
      "loss": 0.9083,
      "num_input_tokens_seen": 147999585,
      "step": 6939
    },
    {
      "epoch": 0.8344856610352913,
      "flos": 14750742712320.0,
      "grad_norm": 6.033891326785479,
      "learning_rate": 2.804143361088489e-07,
      "loss": 0.9493,
      "num_input_tokens_seen": 148018485,
      "step": 6940
    },
    {
      "epoch": 0.8346059039259304,
      "flos": 18572075335680.0,
      "grad_norm": 3.4610358712917386,
      "learning_rate": 2.8001669012022277e-07,
      "loss": 0.9472,
      "num_input_tokens_seen": 148036175,
      "step": 6941
    },
    {
      "epoch": 0.8347261468165694,
      "flos": 20680806727680.0,
      "grad_norm": 3.5056220269152325,
      "learning_rate": 2.7961930505169795e-07,
      "loss": 0.8869,
      "num_input_tokens_seen": 148060140,
      "step": 6942
    },
    {
      "epoch": 0.8348463897072086,
      "flos": 18789202821120.0,
      "grad_norm": 4.382790522936767,
      "learning_rate": 2.792221809635558e-07,
      "loss": 0.9429,
      "num_input_tokens_seen": 148080490,
      "step": 6943
    },
    {
      "epoch": 0.8349666325978476,
      "flos": 16612446412800.0,
      "grad_norm": 3.1217422289103576,
      "learning_rate": 2.788253179160411e-07,
      "loss": 0.9431,
      "num_input_tokens_seen": 148101370,
      "step": 6944
    },
    {
      "epoch": 0.8350868754884867,
      "flos": 9087029391360.0,
      "grad_norm": 3.6694429175384706,
      "learning_rate": 2.7842871596935725e-07,
      "loss": 0.8382,
      "num_input_tokens_seen": 148119605,
      "step": 6945
    },
    {
      "epoch": 0.8352071183791259,
      "flos": 18867529113600.0,
      "grad_norm": 6.123992214205407,
      "learning_rate": 2.780323751836682e-07,
      "loss": 0.8796,
      "num_input_tokens_seen": 148140540,
      "step": 6946
    },
    {
      "epoch": 0.8353273612697649,
      "flos": 14672293785600.0,
      "grad_norm": 3.31342130120806,
      "learning_rate": 2.7763629561909876e-07,
      "loss": 0.9793,
      "num_input_tokens_seen": 148161090,
      "step": 6947
    },
    {
      "epoch": 0.835447604160404,
      "flos": 14015231385600.0,
      "grad_norm": 5.182571643825465,
      "learning_rate": 2.772404773357335e-07,
      "loss": 0.9601,
      "num_input_tokens_seen": 148180215,
      "step": 6948
    },
    {
      "epoch": 0.8355678470510431,
      "flos": 16660749864960.0,
      "grad_norm": 2.4600546570265185,
      "learning_rate": 2.7684492039361853e-07,
      "loss": 0.966,
      "num_input_tokens_seen": 148199160,
      "step": 6949
    },
    {
      "epoch": 0.8356880899416822,
      "flos": 15064967639040.0,
      "grad_norm": 4.21143050837597,
      "learning_rate": 2.764496248527586e-07,
      "loss": 1.0281,
      "num_input_tokens_seen": 148217855,
      "step": 6950
    },
    {
      "epoch": 0.8358083328323213,
      "flos": 19968512040960.0,
      "grad_norm": 2.9917455600830856,
      "learning_rate": 2.760545907731211e-07,
      "loss": 0.9585,
      "num_input_tokens_seen": 148238150,
      "step": 6951
    },
    {
      "epoch": 0.8359285757229604,
      "flos": 19786251448320.0,
      "grad_norm": 5.459709149393554,
      "learning_rate": 2.75659818214631e-07,
      "loss": 0.8641,
      "num_input_tokens_seen": 148258975,
      "step": 6952
    },
    {
      "epoch": 0.8360488186135995,
      "flos": 15223674347520.0,
      "grad_norm": 3.0604939021021895,
      "learning_rate": 2.752653072371749e-07,
      "loss": 0.9742,
      "num_input_tokens_seen": 148278130,
      "step": 6953
    },
    {
      "epoch": 0.8361690615042385,
      "flos": 19676767948800.0,
      "grad_norm": 3.9881847642665984,
      "learning_rate": 2.7487105790060105e-07,
      "loss": 0.9431,
      "num_input_tokens_seen": 148297260,
      "step": 6954
    },
    {
      "epoch": 0.8362893043948777,
      "flos": 27992621998080.0,
      "grad_norm": 5.649491798703505,
      "learning_rate": 2.7447707026471587e-07,
      "loss": 0.8861,
      "num_input_tokens_seen": 148319955,
      "step": 6955
    },
    {
      "epoch": 0.8364095472855168,
      "flos": 17630534922240.0,
      "grad_norm": 2.796505013826794,
      "learning_rate": 2.740833443892874e-07,
      "loss": 0.9976,
      "num_input_tokens_seen": 148337845,
      "step": 6956
    },
    {
      "epoch": 0.8365297901761558,
      "flos": 16163221954560.0,
      "grad_norm": 3.8426515824072385,
      "learning_rate": 2.7368988033404327e-07,
      "loss": 0.9844,
      "num_input_tokens_seen": 148355080,
      "step": 6957
    },
    {
      "epoch": 0.836650033066795,
      "flos": 20225266606080.0,
      "grad_norm": 3.363492208105197,
      "learning_rate": 2.732966781586712e-07,
      "loss": 1.0443,
      "num_input_tokens_seen": 148374545,
      "step": 6958
    },
    {
      "epoch": 0.836770275957434,
      "flos": 15799743160320.0,
      "grad_norm": 4.804817508086102,
      "learning_rate": 2.729037379228205e-07,
      "loss": 0.8637,
      "num_input_tokens_seen": 148394450,
      "step": 6959
    },
    {
      "epoch": 0.8368905188480731,
      "flos": 15983015485440.0,
      "grad_norm": 3.3224674188237735,
      "learning_rate": 2.725110596860998e-07,
      "loss": 0.9972,
      "num_input_tokens_seen": 148414850,
      "step": 6960
    },
    {
      "epoch": 0.8370107617387123,
      "flos": 9427536814080.0,
      "grad_norm": 5.5588011712990895,
      "learning_rate": 2.7211864350807776e-07,
      "loss": 0.9012,
      "num_input_tokens_seen": 148432770,
      "step": 6961
    },
    {
      "epoch": 0.8371310046293513,
      "flos": 17972820541440.0,
      "grad_norm": 2.404532178555983,
      "learning_rate": 2.717264894482836e-07,
      "loss": 0.9273,
      "num_input_tokens_seen": 148452830,
      "step": 6962
    },
    {
      "epoch": 0.8372512475199904,
      "flos": 14043230760960.0,
      "grad_norm": 3.4497773527090354,
      "learning_rate": 2.7133459756620646e-07,
      "loss": 0.9943,
      "num_input_tokens_seen": 148469745,
      "step": 6963
    },
    {
      "epoch": 0.8373714904106295,
      "flos": 13754675159040.0,
      "grad_norm": 3.21661202247601,
      "learning_rate": 2.7094296792129733e-07,
      "loss": 0.9164,
      "num_input_tokens_seen": 148489065,
      "step": 6964
    },
    {
      "epoch": 0.8374917333012686,
      "flos": 10581421977600.0,
      "grad_norm": 4.098764391539665,
      "learning_rate": 2.7055160057296424e-07,
      "loss": 0.9485,
      "num_input_tokens_seen": 148506025,
      "step": 6965
    },
    {
      "epoch": 0.8376119761919076,
      "flos": 21616797941760.0,
      "grad_norm": 2.688663674302836,
      "learning_rate": 2.7016049558057896e-07,
      "loss": 0.9134,
      "num_input_tokens_seen": 148527705,
      "step": 6966
    },
    {
      "epoch": 0.8377322190825467,
      "flos": 20964364984320.0,
      "grad_norm": 3.102825378520332,
      "learning_rate": 2.6976965300347074e-07,
      "loss": 0.9081,
      "num_input_tokens_seen": 148550035,
      "step": 6967
    },
    {
      "epoch": 0.8378524619731859,
      "flos": 18997899202560.0,
      "grad_norm": 4.011068765047472,
      "learning_rate": 2.693790729009309e-07,
      "loss": 0.8732,
      "num_input_tokens_seen": 148571365,
      "step": 6968
    },
    {
      "epoch": 0.8379727048638249,
      "flos": 14697932451840.0,
      "grad_norm": 2.5447915571548205,
      "learning_rate": 2.6898875533220946e-07,
      "loss": 1.0652,
      "num_input_tokens_seen": 148590390,
      "step": 6969
    },
    {
      "epoch": 0.838092947754464,
      "flos": 14252570972160.0,
      "grad_norm": 2.598073917168456,
      "learning_rate": 2.685987003565171e-07,
      "loss": 1.0145,
      "num_input_tokens_seen": 148608150,
      "step": 6970
    },
    {
      "epoch": 0.8382131906451031,
      "flos": 12836106117120.0,
      "grad_norm": 3.8128739470648716,
      "learning_rate": 2.6820890803302566e-07,
      "loss": 0.9334,
      "num_input_tokens_seen": 148623395,
      "step": 6971
    },
    {
      "epoch": 0.8383334335357422,
      "flos": 12102740889600.0,
      "grad_norm": 3.9621948413493535,
      "learning_rate": 2.6781937842086557e-07,
      "loss": 1.0128,
      "num_input_tokens_seen": 148641905,
      "step": 6972
    },
    {
      "epoch": 0.8384536764263812,
      "flos": 14698668257280.0,
      "grad_norm": 3.681874396797303,
      "learning_rate": 2.6743011157912933e-07,
      "loss": 0.8763,
      "num_input_tokens_seen": 148661345,
      "step": 6973
    },
    {
      "epoch": 0.8385739193170204,
      "flos": 20650845204480.0,
      "grad_norm": 3.1750275678263833,
      "learning_rate": 2.6704110756686725e-07,
      "loss": 0.8379,
      "num_input_tokens_seen": 148681890,
      "step": 6974
    },
    {
      "epoch": 0.8386941622076595,
      "flos": 16662650695680.0,
      "grad_norm": 3.853870497312876,
      "learning_rate": 2.6665236644309085e-07,
      "loss": 1.0262,
      "num_input_tokens_seen": 148701920,
      "step": 6975
    },
    {
      "epoch": 0.8388144050982985,
      "flos": 16350111989760.0,
      "grad_norm": 3.919590987840809,
      "learning_rate": 2.662638882667727e-07,
      "loss": 0.9844,
      "num_input_tokens_seen": 148720580,
      "step": 6976
    },
    {
      "epoch": 0.8389346479889377,
      "flos": 17268405104640.0,
      "grad_norm": 3.470753271699136,
      "learning_rate": 2.658756730968443e-07,
      "loss": 0.9225,
      "num_input_tokens_seen": 148738765,
      "step": 6977
    },
    {
      "epoch": 0.8390548908795767,
      "flos": 15065335541760.0,
      "grad_norm": 3.48023528783492,
      "learning_rate": 2.654877209921975e-07,
      "loss": 1.0808,
      "num_input_tokens_seen": 148756020,
      "step": 6978
    },
    {
      "epoch": 0.8391751337702158,
      "flos": 25423222394880.0,
      "grad_norm": 3.7841596521647034,
      "learning_rate": 2.651000320116843e-07,
      "loss": 0.8219,
      "num_input_tokens_seen": 148776625,
      "step": 6979
    },
    {
      "epoch": 0.839295376660855,
      "flos": 15144642908160.0,
      "grad_norm": 3.3380791893783086,
      "learning_rate": 2.647126062141163e-07,
      "loss": 0.9427,
      "num_input_tokens_seen": 148795420,
      "step": 6980
    },
    {
      "epoch": 0.839415619551494,
      "flos": 13072863191040.0,
      "grad_norm": 6.098231775370039,
      "learning_rate": 2.643254436582669e-07,
      "loss": 1.0285,
      "num_input_tokens_seen": 148814630,
      "step": 6981
    },
    {
      "epoch": 0.8395358624421331,
      "flos": 16507714990080.0,
      "grad_norm": 7.237032111926995,
      "learning_rate": 2.6393854440286743e-07,
      "loss": 1.006,
      "num_input_tokens_seen": 148833520,
      "step": 6982
    },
    {
      "epoch": 0.8396561053327722,
      "flos": 17340109148160.0,
      "grad_norm": 3.6559359423745508,
      "learning_rate": 2.6355190850661045e-07,
      "loss": 0.9025,
      "num_input_tokens_seen": 148850075,
      "step": 6983
    },
    {
      "epoch": 0.8397763482234113,
      "flos": 15800509624320.0,
      "grad_norm": 4.485485404634101,
      "learning_rate": 2.631655360281486e-07,
      "loss": 1.0527,
      "num_input_tokens_seen": 148869470,
      "step": 6984
    },
    {
      "epoch": 0.8398965911140504,
      "flos": 15747178168320.0,
      "grad_norm": 8.52387385383396,
      "learning_rate": 2.6277942702609323e-07,
      "loss": 0.8497,
      "num_input_tokens_seen": 148888670,
      "step": 6985
    },
    {
      "epoch": 0.8400168340046895,
      "flos": 15300804956160.0,
      "grad_norm": 4.6157317214436135,
      "learning_rate": 2.623935815590186e-07,
      "loss": 1.0653,
      "num_input_tokens_seen": 148906770,
      "step": 6986
    },
    {
      "epoch": 0.8401370768953286,
      "flos": 15983046144000.0,
      "grad_norm": 3.370363229255734,
      "learning_rate": 2.6200799968545516e-07,
      "loss": 0.9968,
      "num_input_tokens_seen": 148926785,
      "step": 6987
    },
    {
      "epoch": 0.8402573197859676,
      "flos": 42398066442240.0,
      "grad_norm": 0.8009970025424379,
      "learning_rate": 2.616226814638969e-07,
      "loss": 0.7856,
      "num_input_tokens_seen": 148991610,
      "step": 6988
    },
    {
      "epoch": 0.8403775626766068,
      "flos": 16114335989760.0,
      "grad_norm": 5.103567760389558,
      "learning_rate": 2.612376269527954e-07,
      "loss": 0.9555,
      "num_input_tokens_seen": 149011035,
      "step": 6989
    },
    {
      "epoch": 0.8404978055672458,
      "flos": 13911634329600.0,
      "grad_norm": 3.267268009370543,
      "learning_rate": 2.608528362105635e-07,
      "loss": 0.865,
      "num_input_tokens_seen": 149030125,
      "step": 6990
    },
    {
      "epoch": 0.8406180484578849,
      "flos": 19601568829440.0,
      "grad_norm": 2.8026271041917346,
      "learning_rate": 2.6046830929557374e-07,
      "loss": 0.9176,
      "num_input_tokens_seen": 149049495,
      "step": 6991
    },
    {
      "epoch": 0.8407382913485241,
      "flos": 15721140940800.0,
      "grad_norm": 4.421560539144337,
      "learning_rate": 2.6008404626615776e-07,
      "loss": 1.0295,
      "num_input_tokens_seen": 149067715,
      "step": 6992
    },
    {
      "epoch": 0.8408585342391631,
      "flos": 9821528985600.0,
      "grad_norm": 6.176624025848542,
      "learning_rate": 2.597000471806092e-07,
      "loss": 0.9257,
      "num_input_tokens_seen": 149084000,
      "step": 6993
    },
    {
      "epoch": 0.8409787771298022,
      "flos": 14326390456320.0,
      "grad_norm": 6.734628765520413,
      "learning_rate": 2.593163120971793e-07,
      "loss": 0.9158,
      "num_input_tokens_seen": 149102585,
      "step": 6994
    },
    {
      "epoch": 0.8410990200204413,
      "flos": 16449754091520.0,
      "grad_norm": 3.2886803194303034,
      "learning_rate": 2.5893284107408165e-07,
      "loss": 0.8835,
      "num_input_tokens_seen": 149119675,
      "step": 6995
    },
    {
      "epoch": 0.8412192629110804,
      "flos": 17086788341760.0,
      "grad_norm": 9.390522630993608,
      "learning_rate": 2.5854963416948726e-07,
      "loss": 0.9778,
      "num_input_tokens_seen": 149141660,
      "step": 6996
    },
    {
      "epoch": 0.8413395058017195,
      "flos": 18210558689280.0,
      "grad_norm": 2.6564761794788736,
      "learning_rate": 2.5816669144152816e-07,
      "loss": 0.873,
      "num_input_tokens_seen": 149162560,
      "step": 6997
    },
    {
      "epoch": 0.8414597486923585,
      "flos": 45555285196800.0,
      "grad_norm": 0.9044433211518274,
      "learning_rate": 2.5778401294829777e-07,
      "loss": 0.9243,
      "num_input_tokens_seen": 149221020,
      "step": 6998
    },
    {
      "epoch": 0.8415799915829977,
      "flos": 13544384532480.0,
      "grad_norm": 4.227658959845474,
      "learning_rate": 2.574015987478473e-07,
      "loss": 0.8403,
      "num_input_tokens_seen": 149238870,
      "step": 6999
    },
    {
      "epoch": 0.8417002344736367,
      "flos": 14062982983680.0,
      "grad_norm": 3.4076377071729866,
      "learning_rate": 2.570194488981887e-07,
      "loss": 1.0478,
      "num_input_tokens_seen": 149255135,
      "step": 7000
    },
    {
      "epoch": 0.8418204773642758,
      "flos": 44499010560000.0,
      "grad_norm": 0.9307068757880048,
      "learning_rate": 2.566375634572939e-07,
      "loss": 0.8277,
      "num_input_tokens_seen": 149315495,
      "step": 7001
    },
    {
      "epoch": 0.841940720254915,
      "flos": 12102955499520.0,
      "grad_norm": 6.794447752663337,
      "learning_rate": 2.562559424830943e-07,
      "loss": 0.9355,
      "num_input_tokens_seen": 149333175,
      "step": 7002
    },
    {
      "epoch": 0.842060963145554,
      "flos": 11520417730560.0,
      "grad_norm": 8.013160371273134,
      "learning_rate": 2.5587458603348256e-07,
      "loss": 0.8772,
      "num_input_tokens_seen": 149350185,
      "step": 7003
    },
    {
      "epoch": 0.8421812060361931,
      "flos": 15563537940480.0,
      "grad_norm": 2.6672276463138167,
      "learning_rate": 2.554934941663085e-07,
      "loss": 1.0375,
      "num_input_tokens_seen": 149367440,
      "step": 7004
    },
    {
      "epoch": 0.8423014489268322,
      "flos": 19781162127360.0,
      "grad_norm": 3.889745896382916,
      "learning_rate": 2.5511266693938484e-07,
      "loss": 0.9129,
      "num_input_tokens_seen": 149385620,
      "step": 7005
    },
    {
      "epoch": 0.8424216918174713,
      "flos": 17869867315200.0,
      "grad_norm": 4.6564716055958835,
      "learning_rate": 2.547321044104822e-07,
      "loss": 0.9606,
      "num_input_tokens_seen": 149406835,
      "step": 7006
    },
    {
      "epoch": 0.8425419347081103,
      "flos": 17603976499200.0,
      "grad_norm": 7.111824301215366,
      "learning_rate": 2.5435180663733113e-07,
      "loss": 0.9551,
      "num_input_tokens_seen": 149426855,
      "step": 7007
    },
    {
      "epoch": 0.8426621775987495,
      "flos": 17655989637120.0,
      "grad_norm": 3.5785479566306324,
      "learning_rate": 2.539717736776241e-07,
      "loss": 0.8947,
      "num_input_tokens_seen": 149442800,
      "step": 7008
    },
    {
      "epoch": 0.8427824204893886,
      "flos": 16743490990080.0,
      "grad_norm": 6.649407106514389,
      "learning_rate": 2.535920055890097e-07,
      "loss": 0.9545,
      "num_input_tokens_seen": 149463815,
      "step": 7009
    },
    {
      "epoch": 0.8429026633800276,
      "flos": 11420836945920.0,
      "grad_norm": 4.138940547703226,
      "learning_rate": 2.5321250242910006e-07,
      "loss": 0.8351,
      "num_input_tokens_seen": 149481450,
      "step": 7010
    },
    {
      "epoch": 0.8430229062706668,
      "flos": 15771958394880.0,
      "grad_norm": 3.452957356383783,
      "learning_rate": 2.5283326425546493e-07,
      "loss": 1.0542,
      "num_input_tokens_seen": 149500280,
      "step": 7011
    },
    {
      "epoch": 0.8431431491613058,
      "flos": 25209099448320.0,
      "grad_norm": 3.934867099713436,
      "learning_rate": 2.5245429112563443e-07,
      "loss": 0.8796,
      "num_input_tokens_seen": 149520675,
      "step": 7012
    },
    {
      "epoch": 0.8432633920519449,
      "flos": 18369388032000.0,
      "grad_norm": 4.250335904479,
      "learning_rate": 2.5207558309709865e-07,
      "loss": 1.0133,
      "num_input_tokens_seen": 149540130,
      "step": 7013
    },
    {
      "epoch": 0.8433836349425841,
      "flos": 46508353228800.0,
      "grad_norm": 0.6915133980523979,
      "learning_rate": 2.516971402273065e-07,
      "loss": 0.7723,
      "num_input_tokens_seen": 149605915,
      "step": 7014
    },
    {
      "epoch": 0.8435038778332231,
      "flos": 14357425029120.0,
      "grad_norm": 4.106408967028514,
      "learning_rate": 2.513189625736687e-07,
      "loss": 0.8616,
      "num_input_tokens_seen": 149622530,
      "step": 7015
    },
    {
      "epoch": 0.8436241207238622,
      "flos": 14904697344000.0,
      "grad_norm": 4.024433612660291,
      "learning_rate": 2.509410501935534e-07,
      "loss": 0.9047,
      "num_input_tokens_seen": 149637885,
      "step": 7016
    },
    {
      "epoch": 0.8437443636145013,
      "flos": 10370119618560.0,
      "grad_norm": 4.078433438033736,
      "learning_rate": 2.5056340314429116e-07,
      "loss": 0.9361,
      "num_input_tokens_seen": 149655070,
      "step": 7017
    },
    {
      "epoch": 0.8438646065051404,
      "flos": 15347330211840.0,
      "grad_norm": 5.7658846830281405,
      "learning_rate": 2.5018602148316904e-07,
      "loss": 0.9861,
      "num_input_tokens_seen": 149670825,
      "step": 7018
    },
    {
      "epoch": 0.8439848493957794,
      "flos": 16555895808000.0,
      "grad_norm": 2.6616904698093604,
      "learning_rate": 2.498089052674359e-07,
      "loss": 0.9824,
      "num_input_tokens_seen": 149688520,
      "step": 7019
    },
    {
      "epoch": 0.8441050922864186,
      "flos": 13990451159040.0,
      "grad_norm": 3.409684715574028,
      "learning_rate": 2.494320545543007e-07,
      "loss": 0.9429,
      "num_input_tokens_seen": 149707810,
      "step": 7020
    },
    {
      "epoch": 0.8442253351770577,
      "flos": 15511095582720.0,
      "grad_norm": 2.8543694045928913,
      "learning_rate": 2.490554694009308e-07,
      "loss": 0.8541,
      "num_input_tokens_seen": 149728395,
      "step": 7021
    },
    {
      "epoch": 0.8443455780676967,
      "flos": 24502108692480.0,
      "grad_norm": 2.831537706186371,
      "learning_rate": 2.4867914986445426e-07,
      "loss": 0.9885,
      "num_input_tokens_seen": 149750505,
      "step": 7022
    },
    {
      "epoch": 0.8444658209583359,
      "flos": 34469437132800.0,
      "grad_norm": 3.1125992969775154,
      "learning_rate": 2.483030960019581e-07,
      "loss": 0.8901,
      "num_input_tokens_seen": 149774155,
      "step": 7023
    },
    {
      "epoch": 0.8445860638489749,
      "flos": 49040004034560.0,
      "grad_norm": 0.7418937905044142,
      "learning_rate": 2.479273078704891e-07,
      "loss": 0.7587,
      "num_input_tokens_seen": 149827240,
      "step": 7024
    },
    {
      "epoch": 0.844706306739614,
      "flos": 44978940456960.0,
      "grad_norm": 0.7888330938154973,
      "learning_rate": 2.475517855270552e-07,
      "loss": 0.858,
      "num_input_tokens_seen": 149887040,
      "step": 7025
    },
    {
      "epoch": 0.8448265496302532,
      "flos": 10581575270400.0,
      "grad_norm": 4.697669226423746,
      "learning_rate": 2.4717652902862143e-07,
      "loss": 0.9178,
      "num_input_tokens_seen": 149905735,
      "step": 7026
    },
    {
      "epoch": 0.8449467925208922,
      "flos": 16665379307520.0,
      "grad_norm": 4.516666912533987,
      "learning_rate": 2.4680153843211495e-07,
      "loss": 1.0133,
      "num_input_tokens_seen": 149925385,
      "step": 7027
    },
    {
      "epoch": 0.8450670354115313,
      "flos": 16167176908800.0,
      "grad_norm": 3.6696976069879925,
      "learning_rate": 2.464268137944212e-07,
      "loss": 0.909,
      "num_input_tokens_seen": 149946400,
      "step": 7028
    },
    {
      "epoch": 0.8451872783021703,
      "flos": 21253564416000.0,
      "grad_norm": 4.906944080022491,
      "learning_rate": 2.46052355172385e-07,
      "loss": 0.9755,
      "num_input_tokens_seen": 149964160,
      "step": 7029
    },
    {
      "epoch": 0.8453075211928095,
      "flos": 15535845150720.0,
      "grad_norm": 3.6350474056211812,
      "learning_rate": 2.456781626228128e-07,
      "loss": 0.9335,
      "num_input_tokens_seen": 149983385,
      "step": 7030
    },
    {
      "epoch": 0.8454277640834486,
      "flos": 42046337986560.0,
      "grad_norm": 1.0432715958518097,
      "learning_rate": 2.453042362024675e-07,
      "loss": 0.935,
      "num_input_tokens_seen": 150036350,
      "step": 7031
    },
    {
      "epoch": 0.8455480069740876,
      "flos": 19287742464000.0,
      "grad_norm": 4.396662677553887,
      "learning_rate": 2.449305759680751e-07,
      "loss": 0.928,
      "num_input_tokens_seen": 150057395,
      "step": 7032
    },
    {
      "epoch": 0.8456682498647268,
      "flos": 19365670195200.0,
      "grad_norm": 4.167299883728414,
      "learning_rate": 2.445571819763188e-07,
      "loss": 0.9365,
      "num_input_tokens_seen": 150079415,
      "step": 7033
    },
    {
      "epoch": 0.8457884927553658,
      "flos": 14646348533760.0,
      "grad_norm": 4.195558679273308,
      "learning_rate": 2.4418405428384227e-07,
      "loss": 0.778,
      "num_input_tokens_seen": 150099345,
      "step": 7034
    },
    {
      "epoch": 0.8459087356460049,
      "flos": 10812905779200.0,
      "grad_norm": 2.8109118746172057,
      "learning_rate": 2.4381119294724864e-07,
      "loss": 0.9163,
      "num_input_tokens_seen": 150116510,
      "step": 7035
    },
    {
      "epoch": 0.846028978536644,
      "flos": 13360682987520.0,
      "grad_norm": 4.271655392809897,
      "learning_rate": 2.434385980231004e-07,
      "loss": 0.7296,
      "num_input_tokens_seen": 150135070,
      "step": 7036
    },
    {
      "epoch": 0.8461492214272831,
      "flos": 37664956538880.0,
      "grad_norm": 2.292094231224631,
      "learning_rate": 2.4306626956792043e-07,
      "loss": 0.844,
      "num_input_tokens_seen": 150159735,
      "step": 7037
    },
    {
      "epoch": 0.8462694643179222,
      "flos": 13177686589440.0,
      "grad_norm": 5.334174427626155,
      "learning_rate": 2.4269420763819017e-07,
      "loss": 0.9482,
      "num_input_tokens_seen": 150177500,
      "step": 7038
    },
    {
      "epoch": 0.8463897072085613,
      "flos": 17344738590720.0,
      "grad_norm": 5.202040184044719,
      "learning_rate": 2.4232241229035223e-07,
      "loss": 1.0202,
      "num_input_tokens_seen": 150194975,
      "step": 7039
    },
    {
      "epoch": 0.8465099500992004,
      "flos": 49196012789760.0,
      "grad_norm": 0.8257631413577382,
      "learning_rate": 2.419508835808064e-07,
      "loss": 0.784,
      "num_input_tokens_seen": 150251250,
      "step": 7040
    },
    {
      "epoch": 0.8466301929898394,
      "flos": 9690147164160.0,
      "grad_norm": 5.71688261968068,
      "learning_rate": 2.415796215659134e-07,
      "loss": 0.8134,
      "num_input_tokens_seen": 150267675,
      "step": 7041
    },
    {
      "epoch": 0.8467504358804786,
      "flos": 13646356684800.0,
      "grad_norm": 5.001994307473965,
      "learning_rate": 2.412086263019939e-07,
      "loss": 0.963,
      "num_input_tokens_seen": 150285420,
      "step": 7042
    },
    {
      "epoch": 0.8468706787711177,
      "flos": 15143692492800.0,
      "grad_norm": 3.826862111323008,
      "learning_rate": 2.408378978453276e-07,
      "loss": 0.9956,
      "num_input_tokens_seen": 150305260,
      "step": 7043
    },
    {
      "epoch": 0.8469909216617567,
      "flos": 46448123596800.0,
      "grad_norm": 0.799210324061213,
      "learning_rate": 2.404674362521533e-07,
      "loss": 0.8532,
      "num_input_tokens_seen": 150363475,
      "step": 7044
    },
    {
      "epoch": 0.8471111645523959,
      "flos": 13674999889920.0,
      "grad_norm": 6.051713294197438,
      "learning_rate": 2.4009724157866997e-07,
      "loss": 0.9302,
      "num_input_tokens_seen": 150380255,
      "step": 7045
    },
    {
      "epoch": 0.8472314074430349,
      "flos": 15640607232000.0,
      "grad_norm": 2.933514590154099,
      "learning_rate": 2.3972731388103564e-07,
      "loss": 0.9542,
      "num_input_tokens_seen": 150398455,
      "step": 7046
    },
    {
      "epoch": 0.847351650333674,
      "flos": 41422456258560.0,
      "grad_norm": 0.8154151778157994,
      "learning_rate": 2.393576532153687e-07,
      "loss": 0.8408,
      "num_input_tokens_seen": 150461960,
      "step": 7047
    },
    {
      "epoch": 0.8474718932243132,
      "flos": 29493904650240.0,
      "grad_norm": 0.9556309946694226,
      "learning_rate": 2.389882596377453e-07,
      "loss": 0.8151,
      "num_input_tokens_seen": 150515945,
      "step": 7048
    },
    {
      "epoch": 0.8475921361149522,
      "flos": 27258398330880.0,
      "grad_norm": 3.522309319321488,
      "learning_rate": 2.386191332042031e-07,
      "loss": 0.9562,
      "num_input_tokens_seen": 150537560,
      "step": 7049
    },
    {
      "epoch": 0.8477123790055913,
      "flos": 18055163105280.0,
      "grad_norm": 3.830310804739072,
      "learning_rate": 2.3825027397073794e-07,
      "loss": 0.9178,
      "num_input_tokens_seen": 150557755,
      "step": 7050
    },
    {
      "epoch": 0.8478326218962304,
      "flos": 21540525772800.0,
      "grad_norm": 3.7212945262845,
      "learning_rate": 2.3788168199330515e-07,
      "loss": 0.8637,
      "num_input_tokens_seen": 150579035,
      "step": 7051
    },
    {
      "epoch": 0.8479528647868695,
      "flos": 27281553653760.0,
      "grad_norm": 3.0904665390447246,
      "learning_rate": 2.3751335732782074e-07,
      "loss": 0.9273,
      "num_input_tokens_seen": 150600015,
      "step": 7052
    },
    {
      "epoch": 0.8480731076775085,
      "flos": 14880223703040.0,
      "grad_norm": 4.250430141447934,
      "learning_rate": 2.371453000301582e-07,
      "loss": 0.9884,
      "num_input_tokens_seen": 150618420,
      "step": 7053
    },
    {
      "epoch": 0.8481933505681477,
      "flos": 23244718448640.0,
      "grad_norm": 2.8432444028672657,
      "learning_rate": 2.3677751015615222e-07,
      "loss": 0.9344,
      "num_input_tokens_seen": 150640215,
      "step": 7054
    },
    {
      "epoch": 0.8483135934587868,
      "flos": 14724858777600.0,
      "grad_norm": 3.5002132078353707,
      "learning_rate": 2.3640998776159593e-07,
      "loss": 1.0418,
      "num_input_tokens_seen": 150657440,
      "step": 7055
    },
    {
      "epoch": 0.8484338363494258,
      "flos": 15380173639680.0,
      "grad_norm": 3.0862210300425432,
      "learning_rate": 2.3604273290224253e-07,
      "loss": 1.0093,
      "num_input_tokens_seen": 150677875,
      "step": 7056
    },
    {
      "epoch": 0.848554079240065,
      "flos": 10608961474560.0,
      "grad_norm": 2.8996423237657996,
      "learning_rate": 2.356757456338039e-07,
      "loss": 0.9336,
      "num_input_tokens_seen": 150695080,
      "step": 7057
    },
    {
      "epoch": 0.848674322130704,
      "flos": 48736877506560.0,
      "grad_norm": 0.8077236045460987,
      "learning_rate": 2.3530902601195147e-07,
      "loss": 0.842,
      "num_input_tokens_seen": 150763290,
      "step": 7058
    },
    {
      "epoch": 0.8487945650213431,
      "flos": 13096325099520.0,
      "grad_norm": 3.824638678744597,
      "learning_rate": 2.34942574092317e-07,
      "loss": 0.9775,
      "num_input_tokens_seen": 150778260,
      "step": 7059
    },
    {
      "epoch": 0.8489148079119821,
      "flos": 16687706849280.0,
      "grad_norm": 3.5444127077085588,
      "learning_rate": 2.3457638993049045e-07,
      "loss": 0.9487,
      "num_input_tokens_seen": 150795970,
      "step": 7060
    },
    {
      "epoch": 0.8490350508026213,
      "flos": 14147226378240.0,
      "grad_norm": 5.332280750515611,
      "learning_rate": 2.3421047358202252e-07,
      "loss": 0.828,
      "num_input_tokens_seen": 150814540,
      "step": 7061
    },
    {
      "epoch": 0.8491552936932604,
      "flos": 17268067860480.0,
      "grad_norm": 7.114579192831045,
      "learning_rate": 2.3384482510242144e-07,
      "loss": 1.0037,
      "num_input_tokens_seen": 150832120,
      "step": 7062
    },
    {
      "epoch": 0.8492755365838994,
      "flos": 16004208660480.0,
      "grad_norm": 5.043817402934073,
      "learning_rate": 2.3347944454715575e-07,
      "loss": 0.9583,
      "num_input_tokens_seen": 150848230,
      "step": 7063
    },
    {
      "epoch": 0.8493957794745386,
      "flos": 19208925634560.0,
      "grad_norm": 2.7551910649835936,
      "learning_rate": 2.331143319716542e-07,
      "loss": 0.8517,
      "num_input_tokens_seen": 150867480,
      "step": 7064
    },
    {
      "epoch": 0.8495160223651776,
      "flos": 21279724277760.0,
      "grad_norm": 3.7683305952986488,
      "learning_rate": 2.3274948743130363e-07,
      "loss": 0.8403,
      "num_input_tokens_seen": 150887035,
      "step": 7065
    },
    {
      "epoch": 0.8496362652558167,
      "flos": 16402339737600.0,
      "grad_norm": 2.8161746285651783,
      "learning_rate": 2.3238491098145085e-07,
      "loss": 0.9917,
      "num_input_tokens_seen": 150906285,
      "step": 7066
    },
    {
      "epoch": 0.8497565081464559,
      "flos": 10317523968000.0,
      "grad_norm": 4.031963196586488,
      "learning_rate": 2.3202060267740141e-07,
      "loss": 0.91,
      "num_input_tokens_seen": 150923530,
      "step": 7067
    },
    {
      "epoch": 0.8498767510370949,
      "flos": 15008110448640.0,
      "grad_norm": 4.206681232538009,
      "learning_rate": 2.3165656257442044e-07,
      "loss": 0.9525,
      "num_input_tokens_seen": 150941770,
      "step": 7068
    },
    {
      "epoch": 0.849996993927734,
      "flos": 16817892986880.0,
      "grad_norm": 2.7318221900459303,
      "learning_rate": 2.31292790727734e-07,
      "loss": 1.0964,
      "num_input_tokens_seen": 150959055,
      "step": 7069
    },
    {
      "epoch": 0.8501172368183731,
      "flos": 14593170370560.0,
      "grad_norm": 4.61393341700312,
      "learning_rate": 2.3092928719252392e-07,
      "loss": 0.9922,
      "num_input_tokens_seen": 150977175,
      "step": 7070
    },
    {
      "epoch": 0.8502374797090122,
      "flos": 15825351168000.0,
      "grad_norm": 3.5836653149022437,
      "learning_rate": 2.3056605202393475e-07,
      "loss": 0.9733,
      "num_input_tokens_seen": 150994455,
      "step": 7071
    },
    {
      "epoch": 0.8503577225996513,
      "flos": 16822399795200.0,
      "grad_norm": 3.488627896059992,
      "learning_rate": 2.3020308527706888e-07,
      "loss": 0.865,
      "num_input_tokens_seen": 151013590,
      "step": 7072
    },
    {
      "epoch": 0.8504779654902904,
      "flos": 19049023242240.0,
      "grad_norm": 2.9318169344468585,
      "learning_rate": 2.2984038700698715e-07,
      "loss": 1.0746,
      "num_input_tokens_seen": 151032620,
      "step": 7073
    },
    {
      "epoch": 0.8505982083809295,
      "flos": 18840633446400.0,
      "grad_norm": 2.329503515335447,
      "learning_rate": 2.2947795726871222e-07,
      "loss": 0.9871,
      "num_input_tokens_seen": 151053365,
      "step": 7074
    },
    {
      "epoch": 0.8507184512715685,
      "flos": 14410235289600.0,
      "grad_norm": 2.9183117359813875,
      "learning_rate": 2.2911579611722253e-07,
      "loss": 1.0494,
      "num_input_tokens_seen": 151072230,
      "step": 7075
    },
    {
      "epoch": 0.8508386941622077,
      "flos": 13491451637760.0,
      "grad_norm": 4.748652417615465,
      "learning_rate": 2.2875390360745905e-07,
      "loss": 1.0673,
      "num_input_tokens_seen": 151091355,
      "step": 7076
    },
    {
      "epoch": 0.8509589370528468,
      "flos": 11628613570560.0,
      "grad_norm": 7.260371452055754,
      "learning_rate": 2.2839227979432008e-07,
      "loss": 0.969,
      "num_input_tokens_seen": 151108725,
      "step": 7077
    },
    {
      "epoch": 0.8510791799434858,
      "flos": 12941420052480.0,
      "grad_norm": 3.3782640850219825,
      "learning_rate": 2.2803092473266373e-07,
      "loss": 1.0364,
      "num_input_tokens_seen": 151125970,
      "step": 7078
    },
    {
      "epoch": 0.851199422834125,
      "flos": 16665471283200.0,
      "grad_norm": 4.9492981384220815,
      "learning_rate": 2.2766983847730724e-07,
      "loss": 1.0419,
      "num_input_tokens_seen": 151145360,
      "step": 7079
    },
    {
      "epoch": 0.851319665724764,
      "flos": 11525384417280.0,
      "grad_norm": 3.9013504429005494,
      "learning_rate": 2.2730902108302663e-07,
      "loss": 0.8613,
      "num_input_tokens_seen": 151161995,
      "step": 7080
    },
    {
      "epoch": 0.8514399086154031,
      "flos": 13465782312960.0,
      "grad_norm": 2.417465428733386,
      "learning_rate": 2.269484726045583e-07,
      "loss": 0.8881,
      "num_input_tokens_seen": 151180630,
      "step": 7081
    },
    {
      "epoch": 0.8515601515060423,
      "flos": 17477285437440.0,
      "grad_norm": 3.927132900474997,
      "learning_rate": 2.2658819309659672e-07,
      "loss": 0.9799,
      "num_input_tokens_seen": 151200550,
      "step": 7082
    },
    {
      "epoch": 0.8516803943966813,
      "flos": 13852968284160.0,
      "grad_norm": 4.284393938218879,
      "learning_rate": 2.2622818261379706e-07,
      "loss": 1.0352,
      "num_input_tokens_seen": 151217290,
      "step": 7083
    },
    {
      "epoch": 0.8518006372873204,
      "flos": 14383063695360.0,
      "grad_norm": 3.5684280597693188,
      "learning_rate": 2.2586844121077142e-07,
      "loss": 0.9504,
      "num_input_tokens_seen": 151235520,
      "step": 7084
    },
    {
      "epoch": 0.8519208801779595,
      "flos": 17162508656640.0,
      "grad_norm": 3.6352572838823294,
      "learning_rate": 2.2550896894209215e-07,
      "loss": 0.9094,
      "num_input_tokens_seen": 151254755,
      "step": 7085
    },
    {
      "epoch": 0.8520411230685986,
      "flos": 45126663290880.0,
      "grad_norm": 0.6885643219983871,
      "learning_rate": 2.2514976586229184e-07,
      "loss": 0.7814,
      "num_input_tokens_seen": 151322420,
      "step": 7086
    },
    {
      "epoch": 0.8521613659592376,
      "flos": 47139102474240.0,
      "grad_norm": 0.8630493681170568,
      "learning_rate": 2.247908320258609e-07,
      "loss": 0.8169,
      "num_input_tokens_seen": 151382230,
      "step": 7087
    },
    {
      "epoch": 0.8522816088498768,
      "flos": 16428898160640.0,
      "grad_norm": 3.335650322231722,
      "learning_rate": 2.2443216748724914e-07,
      "loss": 0.978,
      "num_input_tokens_seen": 151402660,
      "step": 7088
    },
    {
      "epoch": 0.8524018517405159,
      "flos": 22642305822720.0,
      "grad_norm": 4.845174469222655,
      "learning_rate": 2.2407377230086588e-07,
      "loss": 0.9423,
      "num_input_tokens_seen": 151424735,
      "step": 7089
    },
    {
      "epoch": 0.8525220946311549,
      "flos": 13249911828480.0,
      "grad_norm": 3.691982841873388,
      "learning_rate": 2.23715646521079e-07,
      "loss": 1.0271,
      "num_input_tokens_seen": 151441975,
      "step": 7090
    },
    {
      "epoch": 0.852642337521794,
      "flos": 15480122327040.0,
      "grad_norm": 7.3720231594771315,
      "learning_rate": 2.2335779020221724e-07,
      "loss": 1.0328,
      "num_input_tokens_seen": 151458315,
      "step": 7091
    },
    {
      "epoch": 0.8527625804124331,
      "flos": 49440219893760.0,
      "grad_norm": 0.9493804919427433,
      "learning_rate": 2.2300020339856497e-07,
      "loss": 0.8154,
      "num_input_tokens_seen": 151520720,
      "step": 7092
    },
    {
      "epoch": 0.8528828233030722,
      "flos": 19208036536320.0,
      "grad_norm": 5.8694464224620315,
      "learning_rate": 2.2264288616436966e-07,
      "loss": 0.9637,
      "num_input_tokens_seen": 151540695,
      "step": 7093
    },
    {
      "epoch": 0.8530030661937112,
      "flos": 12385716633600.0,
      "grad_norm": 4.19677549742724,
      "learning_rate": 2.222858385538351e-07,
      "loss": 0.9297,
      "num_input_tokens_seen": 151557215,
      "step": 7094
    },
    {
      "epoch": 0.8531233090843504,
      "flos": 15744664166400.0,
      "grad_norm": 3.32393486343812,
      "learning_rate": 2.2192906062112527e-07,
      "loss": 0.8691,
      "num_input_tokens_seen": 151576810,
      "step": 7095
    },
    {
      "epoch": 0.8532435519749895,
      "flos": 26866582917120.0,
      "grad_norm": 2.753058206914925,
      "learning_rate": 2.2157255242036377e-07,
      "loss": 0.893,
      "num_input_tokens_seen": 151600195,
      "step": 7096
    },
    {
      "epoch": 0.8533637948656285,
      "flos": 15197054607360.0,
      "grad_norm": 3.3519726482267793,
      "learning_rate": 2.2121631400563135e-07,
      "loss": 0.9307,
      "num_input_tokens_seen": 151619745,
      "step": 7097
    },
    {
      "epoch": 0.8534840377562677,
      "flos": 38159686410240.0,
      "grad_norm": 0.8554204639759934,
      "learning_rate": 2.208603454309701e-07,
      "loss": 0.7955,
      "num_input_tokens_seen": 151677555,
      "step": 7098
    },
    {
      "epoch": 0.8536042806469067,
      "flos": 14777301135360.0,
      "grad_norm": 4.377803667081219,
      "learning_rate": 2.2050464675037994e-07,
      "loss": 0.8959,
      "num_input_tokens_seen": 151695900,
      "step": 7099
    },
    {
      "epoch": 0.8537245235375458,
      "flos": 17556562145280.0,
      "grad_norm": 3.819358285259664,
      "learning_rate": 2.2014921801782016e-07,
      "loss": 0.9103,
      "num_input_tokens_seen": 151715110,
      "step": 7100
    },
    {
      "epoch": 0.853844766428185,
      "flos": 17342807101440.0,
      "grad_norm": 3.291949022633058,
      "learning_rate": 2.1979405928720872e-07,
      "loss": 0.9325,
      "num_input_tokens_seen": 151734485,
      "step": 7101
    },
    {
      "epoch": 0.853965009318824,
      "flos": 14878445506560.0,
      "grad_norm": 2.9685786147589885,
      "learning_rate": 2.1943917061242257e-07,
      "loss": 0.9872,
      "num_input_tokens_seen": 151754060,
      "step": 7102
    },
    {
      "epoch": 0.8540852522094631,
      "flos": 17211210670080.0,
      "grad_norm": 3.362370916108191,
      "learning_rate": 2.1908455204729903e-07,
      "loss": 0.8484,
      "num_input_tokens_seen": 151772930,
      "step": 7103
    },
    {
      "epoch": 0.8542054951001022,
      "flos": 17845179064320.0,
      "grad_norm": 2.786295140321332,
      "learning_rate": 2.1873020364563265e-07,
      "loss": 0.9635,
      "num_input_tokens_seen": 151791715,
      "step": 7104
    },
    {
      "epoch": 0.8543257379907413,
      "flos": 17294381015040.0,
      "grad_norm": 4.059924204076559,
      "learning_rate": 2.183761254611789e-07,
      "loss": 0.9495,
      "num_input_tokens_seen": 151811760,
      "step": 7105
    },
    {
      "epoch": 0.8544459808813804,
      "flos": 39551133880320.0,
      "grad_norm": 3.6993605396202747,
      "learning_rate": 2.1802231754764987e-07,
      "loss": 0.8882,
      "num_input_tokens_seen": 151836920,
      "step": 7106
    },
    {
      "epoch": 0.8545662237720195,
      "flos": 18343473438720.0,
      "grad_norm": 3.3960579795875416,
      "learning_rate": 2.17668779958718e-07,
      "loss": 0.9548,
      "num_input_tokens_seen": 151859220,
      "step": 7107
    },
    {
      "epoch": 0.8546864666626586,
      "flos": 7801823723520.0,
      "grad_norm": 4.796810385800062,
      "learning_rate": 2.1731551274801553e-07,
      "loss": 0.9841,
      "num_input_tokens_seen": 151875380,
      "step": 7108
    },
    {
      "epoch": 0.8548067095532976,
      "flos": 18159342673920.0,
      "grad_norm": 5.670252880667042,
      "learning_rate": 2.169625159691324e-07,
      "loss": 0.7943,
      "num_input_tokens_seen": 151894975,
      "step": 7109
    },
    {
      "epoch": 0.8549269524439368,
      "flos": 17582783324160.0,
      "grad_norm": 14.860446510229712,
      "learning_rate": 2.1660978967561784e-07,
      "loss": 0.9308,
      "num_input_tokens_seen": 151914030,
      "step": 7110
    },
    {
      "epoch": 0.8550471953345758,
      "flos": 14066386083840.0,
      "grad_norm": 4.218892865057838,
      "learning_rate": 2.1625733392098035e-07,
      "loss": 0.9815,
      "num_input_tokens_seen": 151929360,
      "step": 7111
    },
    {
      "epoch": 0.8551674382252149,
      "flos": 16219128729600.0,
      "grad_norm": 3.1212770792527778,
      "learning_rate": 2.159051487586867e-07,
      "loss": 0.9877,
      "num_input_tokens_seen": 151949210,
      "step": 7112
    },
    {
      "epoch": 0.8552876811158541,
      "flos": 14645796679680.0,
      "grad_norm": 7.895562830764554,
      "learning_rate": 2.155532342421642e-07,
      "loss": 0.9123,
      "num_input_tokens_seen": 151966930,
      "step": 7113
    },
    {
      "epoch": 0.8554079240064931,
      "flos": 16428928819200.0,
      "grad_norm": 10.304449464768734,
      "learning_rate": 2.1520159042479636e-07,
      "loss": 0.9756,
      "num_input_tokens_seen": 151984940,
      "step": 7114
    },
    {
      "epoch": 0.8555281668971322,
      "flos": 15721631477760.0,
      "grad_norm": 3.611740724173703,
      "learning_rate": 2.148502173599287e-07,
      "loss": 0.904,
      "num_input_tokens_seen": 152002800,
      "step": 7115
    },
    {
      "epoch": 0.8556484097877713,
      "flos": 22197526855680.0,
      "grad_norm": 3.166645719224679,
      "learning_rate": 2.1449911510086372e-07,
      "loss": 0.8447,
      "num_input_tokens_seen": 152021990,
      "step": 7116
    },
    {
      "epoch": 0.8557686526784104,
      "flos": 17294381015040.0,
      "grad_norm": 4.721413170315547,
      "learning_rate": 2.141482837008628e-07,
      "loss": 0.9637,
      "num_input_tokens_seen": 152042250,
      "step": 7117
    },
    {
      "epoch": 0.8558888955690495,
      "flos": 12548347637760.0,
      "grad_norm": 3.9279577494289355,
      "learning_rate": 2.1379772321314826e-07,
      "loss": 0.9057,
      "num_input_tokens_seen": 152060015,
      "step": 7118
    },
    {
      "epoch": 0.8560091384596886,
      "flos": 13596826890240.0,
      "grad_norm": 4.307081937777082,
      "learning_rate": 2.1344743369089802e-07,
      "loss": 1.0102,
      "num_input_tokens_seen": 152075515,
      "step": 7119
    },
    {
      "epoch": 0.8561293813503277,
      "flos": 17005641461760.0,
      "grad_norm": 2.411422223891167,
      "learning_rate": 2.130974151872522e-07,
      "loss": 1.0156,
      "num_input_tokens_seen": 152095570,
      "step": 7120
    },
    {
      "epoch": 0.8562496242409667,
      "flos": 16009911152640.0,
      "grad_norm": 3.6123354442036795,
      "learning_rate": 2.1274766775530773e-07,
      "loss": 0.9744,
      "num_input_tokens_seen": 152115155,
      "step": 7121
    },
    {
      "epoch": 0.8563698671316058,
      "flos": 10391220817920.0,
      "grad_norm": 3.0407901354462377,
      "learning_rate": 2.1239819144812077e-07,
      "loss": 0.9933,
      "num_input_tokens_seen": 152129335,
      "step": 7122
    },
    {
      "epoch": 0.856490110022245,
      "flos": 27967872430080.0,
      "grad_norm": 5.512219639436871,
      "learning_rate": 2.1204898631870716e-07,
      "loss": 0.8903,
      "num_input_tokens_seen": 152153945,
      "step": 7123
    },
    {
      "epoch": 0.856610352912884,
      "flos": 20702889000960.0,
      "grad_norm": 3.4767296911606387,
      "learning_rate": 2.1170005242004006e-07,
      "loss": 0.9484,
      "num_input_tokens_seen": 152175015,
      "step": 7124
    },
    {
      "epoch": 0.8567305958035231,
      "flos": 16979389624320.0,
      "grad_norm": 3.595247570315278,
      "learning_rate": 2.1135138980505384e-07,
      "loss": 0.9605,
      "num_input_tokens_seen": 152195405,
      "step": 7125
    },
    {
      "epoch": 0.8568508386941622,
      "flos": 15773307371520.0,
      "grad_norm": 3.2033898220843944,
      "learning_rate": 2.110029985266395e-07,
      "loss": 0.9117,
      "num_input_tokens_seen": 152214830,
      "step": 7126
    },
    {
      "epoch": 0.8569710815848013,
      "flos": 12256603545600.0,
      "grad_norm": 4.291311872810188,
      "learning_rate": 2.1065487863764787e-07,
      "loss": 0.9333,
      "num_input_tokens_seen": 152232895,
      "step": 7127
    },
    {
      "epoch": 0.8570913244754403,
      "flos": 16874412933120.0,
      "grad_norm": 4.8576784760203715,
      "learning_rate": 2.1030703019088846e-07,
      "loss": 1.0517,
      "num_input_tokens_seen": 152253245,
      "step": 7128
    },
    {
      "epoch": 0.8572115673660795,
      "flos": 14227330867200.0,
      "grad_norm": 6.879021318156696,
      "learning_rate": 2.099594532391291e-07,
      "loss": 0.8897,
      "num_input_tokens_seen": 152271650,
      "step": 7129
    },
    {
      "epoch": 0.8573318102567186,
      "flos": 19337916088320.0,
      "grad_norm": 2.5653720670632283,
      "learning_rate": 2.0961214783509806e-07,
      "loss": 0.9847,
      "num_input_tokens_seen": 152294250,
      "step": 7130
    },
    {
      "epoch": 0.8574520531473576,
      "flos": 17739957104640.0,
      "grad_norm": 7.09453722917103,
      "learning_rate": 2.0926511403148051e-07,
      "loss": 0.9483,
      "num_input_tokens_seen": 152312935,
      "step": 7131
    },
    {
      "epoch": 0.8575722960379968,
      "flos": 13309068410880.0,
      "grad_norm": 2.7671656223769907,
      "learning_rate": 2.0891835188092143e-07,
      "loss": 0.9368,
      "num_input_tokens_seen": 152329655,
      "step": 7132
    },
    {
      "epoch": 0.8576925389286358,
      "flos": 15773460664320.0,
      "grad_norm": 3.82706147036938,
      "learning_rate": 2.0857186143602434e-07,
      "loss": 1.0087,
      "num_input_tokens_seen": 152348020,
      "step": 7133
    },
    {
      "epoch": 0.8578127818192749,
      "flos": 16272368209920.0,
      "grad_norm": 3.1068606538197714,
      "learning_rate": 2.0822564274935094e-07,
      "loss": 0.8691,
      "num_input_tokens_seen": 152367165,
      "step": 7134
    },
    {
      "epoch": 0.8579330247099141,
      "flos": 24846540410880.0,
      "grad_norm": 4.3313212247114325,
      "learning_rate": 2.078796958734239e-07,
      "loss": 0.8651,
      "num_input_tokens_seen": 152389605,
      "step": 7135
    },
    {
      "epoch": 0.8580532676005531,
      "flos": 14016549703680.0,
      "grad_norm": 4.841431487385306,
      "learning_rate": 2.0753402086072124e-07,
      "loss": 0.9316,
      "num_input_tokens_seen": 152407955,
      "step": 7136
    },
    {
      "epoch": 0.8581735104911922,
      "flos": 16159726878720.0,
      "grad_norm": 3.5504810974365704,
      "learning_rate": 2.071886177636828e-07,
      "loss": 0.9483,
      "num_input_tokens_seen": 152424460,
      "step": 7137
    },
    {
      "epoch": 0.8582937533818313,
      "flos": 16455395266560.0,
      "grad_norm": 3.6925132423736793,
      "learning_rate": 2.0684348663470575e-07,
      "loss": 1.0167,
      "num_input_tokens_seen": 152444360,
      "step": 7138
    },
    {
      "epoch": 0.8584139962724704,
      "flos": 13831621816320.0,
      "grad_norm": 2.8355329240728233,
      "learning_rate": 2.0649862752614555e-07,
      "loss": 0.8092,
      "num_input_tokens_seen": 152462790,
      "step": 7139
    },
    {
      "epoch": 0.8585342391631094,
      "flos": 51048970690560.0,
      "grad_norm": 0.7744589332001808,
      "learning_rate": 2.0615404049031838e-07,
      "loss": 0.7861,
      "num_input_tokens_seen": 152519480,
      "step": 7140
    },
    {
      "epoch": 0.8586544820537486,
      "flos": 7591195852800.0,
      "grad_norm": 4.084843155313998,
      "learning_rate": 2.0580972557949616e-07,
      "loss": 0.9673,
      "num_input_tokens_seen": 152534290,
      "step": 7141
    },
    {
      "epoch": 0.8587747249443877,
      "flos": 46401414389760.0,
      "grad_norm": 0.8011652328238192,
      "learning_rate": 2.054656828459125e-07,
      "loss": 0.7402,
      "num_input_tokens_seen": 152598120,
      "step": 7142
    },
    {
      "epoch": 0.8588949678350267,
      "flos": 19104132894720.0,
      "grad_norm": 3.6440216067916174,
      "learning_rate": 2.051219123417578e-07,
      "loss": 0.9564,
      "num_input_tokens_seen": 152617900,
      "step": 7143
    },
    {
      "epoch": 0.8590152107256659,
      "flos": 18579157463040.0,
      "grad_norm": 8.526698648583167,
      "learning_rate": 2.0477841411918196e-07,
      "loss": 0.7922,
      "num_input_tokens_seen": 152637145,
      "step": 7144
    },
    {
      "epoch": 0.859135453616305,
      "flos": 18605501276160.0,
      "grad_norm": 3.0077547934805255,
      "learning_rate": 2.0443518823029326e-07,
      "loss": 0.9192,
      "num_input_tokens_seen": 152657405,
      "step": 7145
    },
    {
      "epoch": 0.859255696506944,
      "flos": 9139318456320.0,
      "grad_norm": 3.2475446082172357,
      "learning_rate": 2.0409223472715854e-07,
      "loss": 0.9459,
      "num_input_tokens_seen": 152674270,
      "step": 7146
    },
    {
      "epoch": 0.8593759393975832,
      "flos": 13096294440960.0,
      "grad_norm": 5.049451424291484,
      "learning_rate": 2.0374955366180434e-07,
      "loss": 0.9326,
      "num_input_tokens_seen": 152691630,
      "step": 7147
    },
    {
      "epoch": 0.8594961822882222,
      "flos": 15773338030080.0,
      "grad_norm": 3.4375589539067506,
      "learning_rate": 2.034071450862147e-07,
      "loss": 0.914,
      "num_input_tokens_seen": 152708820,
      "step": 7148
    },
    {
      "epoch": 0.8596164251788613,
      "flos": 16560004055040.0,
      "grad_norm": 4.5602212728387395,
      "learning_rate": 2.030650090523327e-07,
      "loss": 0.9579,
      "num_input_tokens_seen": 152727730,
      "step": 7149
    },
    {
      "epoch": 0.8597366680695004,
      "flos": 22564684677120.0,
      "grad_norm": 3.0132258067174567,
      "learning_rate": 2.0272314561205995e-07,
      "loss": 0.7996,
      "num_input_tokens_seen": 152747845,
      "step": 7150
    },
    {
      "epoch": 0.8598569109601395,
      "flos": 15117563289600.0,
      "grad_norm": 4.349256798406635,
      "learning_rate": 2.023815548172567e-07,
      "loss": 0.9189,
      "num_input_tokens_seen": 152767635,
      "step": 7151
    },
    {
      "epoch": 0.8599771538507786,
      "flos": 18107421511680.0,
      "grad_norm": 2.913302979678531,
      "learning_rate": 2.0204023671974267e-07,
      "loss": 0.8623,
      "num_input_tokens_seen": 152786740,
      "step": 7152
    },
    {
      "epoch": 0.8600973967414177,
      "flos": 11837248634880.0,
      "grad_norm": 3.7646630035791486,
      "learning_rate": 2.0169919137129532e-07,
      "loss": 1.0086,
      "num_input_tokens_seen": 152804900,
      "step": 7153
    },
    {
      "epoch": 0.8602176396320568,
      "flos": 17949266657280.0,
      "grad_norm": 5.352832593352882,
      "learning_rate": 2.013584188236508e-07,
      "loss": 0.8905,
      "num_input_tokens_seen": 152822525,
      "step": 7154
    },
    {
      "epoch": 0.8603378825226958,
      "flos": 14488377630720.0,
      "grad_norm": 3.0322240463958967,
      "learning_rate": 2.0101791912850396e-07,
      "loss": 0.9865,
      "num_input_tokens_seen": 152841785,
      "step": 7155
    },
    {
      "epoch": 0.8604581254133349,
      "flos": 24921708871680.0,
      "grad_norm": 4.070979966041485,
      "learning_rate": 2.006776923375082e-07,
      "loss": 0.828,
      "num_input_tokens_seen": 152863160,
      "step": 7156
    },
    {
      "epoch": 0.860578368303974,
      "flos": 16057540116480.0,
      "grad_norm": 7.101483536586574,
      "learning_rate": 2.003377385022764e-07,
      "loss": 0.9046,
      "num_input_tokens_seen": 152881705,
      "step": 7157
    },
    {
      "epoch": 0.8606986111946131,
      "flos": 15143876444160.0,
      "grad_norm": 3.923705870497776,
      "learning_rate": 1.9999805767437826e-07,
      "loss": 0.9599,
      "num_input_tokens_seen": 152900315,
      "step": 7158
    },
    {
      "epoch": 0.8608188540852522,
      "flos": 20572028375040.0,
      "grad_norm": 3.2732181156906948,
      "learning_rate": 1.9965864990534386e-07,
      "loss": 0.9111,
      "num_input_tokens_seen": 152920560,
      "step": 7159
    },
    {
      "epoch": 0.8609390969758913,
      "flos": 21174962196480.0,
      "grad_norm": 4.477539168995364,
      "learning_rate": 1.9931951524666092e-07,
      "loss": 0.9727,
      "num_input_tokens_seen": 152941370,
      "step": 7160
    },
    {
      "epoch": 0.8610593398665304,
      "flos": 15089717207040.0,
      "grad_norm": 2.796666142854274,
      "learning_rate": 1.9898065374977534e-07,
      "loss": 1.0019,
      "num_input_tokens_seen": 152961295,
      "step": 7161
    },
    {
      "epoch": 0.8611795827571694,
      "flos": 10476629237760.0,
      "grad_norm": 3.951660850088364,
      "learning_rate": 1.9864206546609342e-07,
      "loss": 0.9253,
      "num_input_tokens_seen": 152979855,
      "step": 7162
    },
    {
      "epoch": 0.8612998256478086,
      "flos": 17241662730240.0,
      "grad_norm": 3.032911179758517,
      "learning_rate": 1.983037504469771e-07,
      "loss": 1.0325,
      "num_input_tokens_seen": 152998285,
      "step": 7163
    },
    {
      "epoch": 0.8614200685384477,
      "flos": 15092537794560.0,
      "grad_norm": 2.9707076193670803,
      "learning_rate": 1.9796570874374984e-07,
      "loss": 0.8633,
      "num_input_tokens_seen": 153018110,
      "step": 7164
    },
    {
      "epoch": 0.8615403114290867,
      "flos": 14197246709760.0,
      "grad_norm": 2.84298477776337,
      "learning_rate": 1.976279404076917e-07,
      "loss": 0.9693,
      "num_input_tokens_seen": 153037230,
      "step": 7165
    },
    {
      "epoch": 0.8616605543197259,
      "flos": 21227588505600.0,
      "grad_norm": 3.037168093785266,
      "learning_rate": 1.9729044549004193e-07,
      "loss": 0.9524,
      "num_input_tokens_seen": 153058335,
      "step": 7166
    },
    {
      "epoch": 0.8617807972103649,
      "flos": 20596471357440.0,
      "grad_norm": 4.891406304591419,
      "learning_rate": 1.9695322404199822e-07,
      "loss": 0.8921,
      "num_input_tokens_seen": 153080100,
      "step": 7167
    },
    {
      "epoch": 0.861901040101004,
      "flos": 19707250667520.0,
      "grad_norm": 3.60601967748773,
      "learning_rate": 1.9661627611471654e-07,
      "loss": 1.0191,
      "num_input_tokens_seen": 153099615,
      "step": 7168
    },
    {
      "epoch": 0.8620212829916432,
      "flos": 35571646402560.0,
      "grad_norm": 4.809749263009711,
      "learning_rate": 1.9627960175931246e-07,
      "loss": 0.8894,
      "num_input_tokens_seen": 153124035,
      "step": 7169
    },
    {
      "epoch": 0.8621415258822822,
      "flos": 15010011279360.0,
      "grad_norm": 6.1977309597200385,
      "learning_rate": 1.9594320102685847e-07,
      "loss": 0.932,
      "num_input_tokens_seen": 153143025,
      "step": 7170
    },
    {
      "epoch": 0.8622617687729213,
      "flos": 15405965598720.0,
      "grad_norm": 5.823934485433795,
      "learning_rate": 1.956070739683864e-07,
      "loss": 0.8222,
      "num_input_tokens_seen": 153162080,
      "step": 7171
    },
    {
      "epoch": 0.8623820116635604,
      "flos": 18684379422720.0,
      "grad_norm": 3.081948012084944,
      "learning_rate": 1.9527122063488678e-07,
      "loss": 0.9308,
      "num_input_tokens_seen": 153182915,
      "step": 7172
    },
    {
      "epoch": 0.8625022545541995,
      "flos": 13939020533760.0,
      "grad_norm": 3.3362065012480455,
      "learning_rate": 1.9493564107730755e-07,
      "loss": 0.9947,
      "num_input_tokens_seen": 153202635,
      "step": 7173
    },
    {
      "epoch": 0.8626224974448385,
      "flos": 15563660574720.0,
      "grad_norm": 3.5448722394510583,
      "learning_rate": 1.9460033534655684e-07,
      "loss": 0.8059,
      "num_input_tokens_seen": 153221715,
      "step": 7174
    },
    {
      "epoch": 0.8627427403354777,
      "flos": 16586194575360.0,
      "grad_norm": 2.3480397956829,
      "learning_rate": 1.9426530349349978e-07,
      "loss": 1.0276,
      "num_input_tokens_seen": 153241885,
      "step": 7175
    },
    {
      "epoch": 0.8628629832261168,
      "flos": 11578255994880.0,
      "grad_norm": 3.71082275213216,
      "learning_rate": 1.9393054556896038e-07,
      "loss": 0.8445,
      "num_input_tokens_seen": 153259305,
      "step": 7176
    },
    {
      "epoch": 0.8629832261167558,
      "flos": 20015343882240.0,
      "grad_norm": 3.705766210775221,
      "learning_rate": 1.9359606162372133e-07,
      "loss": 0.8763,
      "num_input_tokens_seen": 153280630,
      "step": 7177
    },
    {
      "epoch": 0.863103469007395,
      "flos": 14357486346240.0,
      "grad_norm": 3.5164444032068425,
      "learning_rate": 1.9326185170852293e-07,
      "loss": 0.8998,
      "num_input_tokens_seen": 153299315,
      "step": 7178
    },
    {
      "epoch": 0.863223711898034,
      "flos": 17425119006720.0,
      "grad_norm": 3.4226037211894402,
      "learning_rate": 1.9292791587406598e-07,
      "loss": 0.9128,
      "num_input_tokens_seen": 153317895,
      "step": 7179
    },
    {
      "epoch": 0.8633439547886731,
      "flos": 12521390653440.0,
      "grad_norm": 3.793540463639155,
      "learning_rate": 1.9259425417100661e-07,
      "loss": 1.054,
      "num_input_tokens_seen": 153333730,
      "step": 7180
    },
    {
      "epoch": 0.8634641976793123,
      "flos": 9085772390400.0,
      "grad_norm": 3.497960722696768,
      "learning_rate": 1.9226086664996234e-07,
      "loss": 0.9296,
      "num_input_tokens_seen": 153351695,
      "step": 7181
    },
    {
      "epoch": 0.8635844405699513,
      "flos": 16979941478400.0,
      "grad_norm": 4.284610515328833,
      "learning_rate": 1.9192775336150712e-07,
      "loss": 0.9308,
      "num_input_tokens_seen": 153371715,
      "step": 7182
    },
    {
      "epoch": 0.8637046834605904,
      "flos": 43267389726720.0,
      "grad_norm": 0.7729368034132013,
      "learning_rate": 1.915949143561739e-07,
      "loss": 0.765,
      "num_input_tokens_seen": 153426110,
      "step": 7183
    },
    {
      "epoch": 0.8638249263512295,
      "flos": 14593752883200.0,
      "grad_norm": 3.631044806823534,
      "learning_rate": 1.9126234968445498e-07,
      "loss": 0.9549,
      "num_input_tokens_seen": 153445520,
      "step": 7184
    },
    {
      "epoch": 0.8639451692418686,
      "flos": 18946867138560.0,
      "grad_norm": 3.041446642995721,
      "learning_rate": 1.9093005939679884e-07,
      "loss": 0.8674,
      "num_input_tokens_seen": 153467195,
      "step": 7185
    },
    {
      "epoch": 0.8640654121325076,
      "flos": 10686797230080.0,
      "grad_norm": 3.115396774029168,
      "learning_rate": 1.9059804354361452e-07,
      "loss": 0.9563,
      "num_input_tokens_seen": 153484690,
      "step": 7186
    },
    {
      "epoch": 0.8641856550231467,
      "flos": 22722226360320.0,
      "grad_norm": 3.4128988626753336,
      "learning_rate": 1.902663021752684e-07,
      "loss": 0.9035,
      "num_input_tokens_seen": 153505840,
      "step": 7187
    },
    {
      "epoch": 0.8643058979137859,
      "flos": 10581943173120.0,
      "grad_norm": 5.046972315177774,
      "learning_rate": 1.8993483534208556e-07,
      "loss": 1.016,
      "num_input_tokens_seen": 153524470,
      "step": 7188
    },
    {
      "epoch": 0.8644261408044249,
      "flos": 9244601733120.0,
      "grad_norm": 9.168045726269304,
      "learning_rate": 1.8960364309434884e-07,
      "loss": 0.9313,
      "num_input_tokens_seen": 153541685,
      "step": 7189
    },
    {
      "epoch": 0.864546383695064,
      "flos": 14803767582720.0,
      "grad_norm": 3.274313579579609,
      "learning_rate": 1.8927272548229967e-07,
      "loss": 0.9726,
      "num_input_tokens_seen": 153561095,
      "step": 7190
    },
    {
      "epoch": 0.8646666265857031,
      "flos": 15144091054080.0,
      "grad_norm": 3.1827556232357472,
      "learning_rate": 1.8894208255613876e-07,
      "loss": 1.0224,
      "num_input_tokens_seen": 153580130,
      "step": 7191
    },
    {
      "epoch": 0.8647868694763422,
      "flos": 14012410798080.0,
      "grad_norm": 3.3720239304115363,
      "learning_rate": 1.8861171436602397e-07,
      "loss": 0.957,
      "num_input_tokens_seen": 153596965,
      "step": 7192
    },
    {
      "epoch": 0.8649071123669813,
      "flos": 18631201259520.0,
      "grad_norm": 3.58806498793793,
      "learning_rate": 1.882816209620719e-07,
      "loss": 0.9967,
      "num_input_tokens_seen": 153613395,
      "step": 7193
    },
    {
      "epoch": 0.8650273552576204,
      "flos": 14409806069760.0,
      "grad_norm": 2.9024288529540154,
      "learning_rate": 1.8795180239435738e-07,
      "loss": 0.9489,
      "num_input_tokens_seen": 153631970,
      "step": 7194
    },
    {
      "epoch": 0.8651475981482595,
      "flos": 17031648030720.0,
      "grad_norm": 4.8216438124310965,
      "learning_rate": 1.8762225871291348e-07,
      "loss": 0.9409,
      "num_input_tokens_seen": 153647565,
      "step": 7195
    },
    {
      "epoch": 0.8652678410388985,
      "flos": 15402194595840.0,
      "grad_norm": 2.216792680509523,
      "learning_rate": 1.8729298996773201e-07,
      "loss": 1.0013,
      "num_input_tokens_seen": 153666035,
      "step": 7196
    },
    {
      "epoch": 0.8653880839295377,
      "flos": 46700209950720.0,
      "grad_norm": 0.8405468149235684,
      "learning_rate": 1.8696399620876301e-07,
      "loss": 0.8233,
      "num_input_tokens_seen": 153722785,
      "step": 7197
    },
    {
      "epoch": 0.8655083268201768,
      "flos": 12574538158080.0,
      "grad_norm": 4.240777049926418,
      "learning_rate": 1.866352774859141e-07,
      "loss": 0.984,
      "num_input_tokens_seen": 153737730,
      "step": 7198
    },
    {
      "epoch": 0.8656285697108158,
      "flos": 14698024427520.0,
      "grad_norm": 4.272185370577472,
      "learning_rate": 1.8630683384905188e-07,
      "loss": 0.8745,
      "num_input_tokens_seen": 153756780,
      "step": 7199
    },
    {
      "epoch": 0.865748812601455,
      "flos": 13223690649600.0,
      "grad_norm": 6.082036267660101,
      "learning_rate": 1.8597866534800045e-07,
      "loss": 1.0837,
      "num_input_tokens_seen": 153771615,
      "step": 7200
    },
    {
      "epoch": 0.865869055492094,
      "flos": 50595063582720.0,
      "grad_norm": 3.245746763851689,
      "learning_rate": 1.8565077203254398e-07,
      "loss": 0.9347,
      "num_input_tokens_seen": 153796795,
      "step": 7201
    },
    {
      "epoch": 0.8659892983827331,
      "flos": 12312019783680.0,
      "grad_norm": 4.892121010020559,
      "learning_rate": 1.8532315395242203e-07,
      "loss": 0.9,
      "num_input_tokens_seen": 153812965,
      "step": 7202
    },
    {
      "epoch": 0.8661095412733723,
      "flos": 12679514849280.0,
      "grad_norm": 4.5556143444165365,
      "learning_rate": 1.849958111573353e-07,
      "loss": 0.9069,
      "num_input_tokens_seen": 153831290,
      "step": 7203
    },
    {
      "epoch": 0.8662297841640113,
      "flos": 12915995996160.0,
      "grad_norm": 2.5124156691029946,
      "learning_rate": 1.8466874369694074e-07,
      "loss": 0.8354,
      "num_input_tokens_seen": 153848705,
      "step": 7204
    },
    {
      "epoch": 0.8663500270546504,
      "flos": 11577642823680.0,
      "grad_norm": 18.683854682084476,
      "learning_rate": 1.843419516208542e-07,
      "loss": 0.8885,
      "num_input_tokens_seen": 153865350,
      "step": 7205
    },
    {
      "epoch": 0.8664702699452895,
      "flos": 12678840360960.0,
      "grad_norm": 6.3251298009045644,
      "learning_rate": 1.8401543497865047e-07,
      "loss": 0.972,
      "num_input_tokens_seen": 153883070,
      "step": 7206
    },
    {
      "epoch": 0.8665905128359286,
      "flos": 21907959521280.0,
      "grad_norm": 4.787619542952585,
      "learning_rate": 1.836891938198608e-07,
      "loss": 0.8252,
      "num_input_tokens_seen": 153903215,
      "step": 7207
    },
    {
      "epoch": 0.8667107557265676,
      "flos": 13225898065920.0,
      "grad_norm": 5.067779990065666,
      "learning_rate": 1.8336322819397677e-07,
      "loss": 0.9004,
      "num_input_tokens_seen": 153920470,
      "step": 7208
    },
    {
      "epoch": 0.8668309986172068,
      "flos": 14252264386560.0,
      "grad_norm": 3.1529289373682747,
      "learning_rate": 1.8303753815044654e-07,
      "loss": 0.8051,
      "num_input_tokens_seen": 153939495,
      "step": 7209
    },
    {
      "epoch": 0.8669512415078459,
      "flos": 15352695459840.0,
      "grad_norm": 5.4186331871668,
      "learning_rate": 1.827121237386773e-07,
      "loss": 0.8842,
      "num_input_tokens_seen": 153956660,
      "step": 7210
    },
    {
      "epoch": 0.8670714843984849,
      "flos": 12540560363520.0,
      "grad_norm": 4.605375151455583,
      "learning_rate": 1.8238698500803374e-07,
      "loss": 0.9406,
      "num_input_tokens_seen": 153969145,
      "step": 7211
    },
    {
      "epoch": 0.8671917272891241,
      "flos": 43453045309440.0,
      "grad_norm": 0.820583189852873,
      "learning_rate": 1.820621220078391e-07,
      "loss": 0.8118,
      "num_input_tokens_seen": 154032775,
      "step": 7212
    },
    {
      "epoch": 0.8673119701797631,
      "flos": 14517296762880.0,
      "grad_norm": 2.7792349384239796,
      "learning_rate": 1.8173753478737553e-07,
      "loss": 0.8647,
      "num_input_tokens_seen": 154052930,
      "step": 7213
    },
    {
      "epoch": 0.8674322130704022,
      "flos": 13938897899520.0,
      "grad_norm": 3.788910727066641,
      "learning_rate": 1.8141322339588205e-07,
      "loss": 0.9803,
      "num_input_tokens_seen": 154069990,
      "step": 7214
    },
    {
      "epoch": 0.8675524559610414,
      "flos": 18521349857280.0,
      "grad_norm": 4.2715118365458125,
      "learning_rate": 1.810891878825569e-07,
      "loss": 0.8919,
      "num_input_tokens_seen": 154089685,
      "step": 7215
    },
    {
      "epoch": 0.8676726988516804,
      "flos": 10634324213760.0,
      "grad_norm": 3.6416891613810365,
      "learning_rate": 1.8076542829655561e-07,
      "loss": 0.9003,
      "num_input_tokens_seen": 154108210,
      "step": 7216
    },
    {
      "epoch": 0.8677929417423195,
      "flos": 11524280709120.0,
      "grad_norm": 3.9341443007068486,
      "learning_rate": 1.8044194468699203e-07,
      "loss": 0.9949,
      "num_input_tokens_seen": 154125240,
      "step": 7217
    },
    {
      "epoch": 0.8679131846329585,
      "flos": 13361541427200.0,
      "grad_norm": 5.071479275179405,
      "learning_rate": 1.8011873710293912e-07,
      "loss": 0.924,
      "num_input_tokens_seen": 154143465,
      "step": 7218
    },
    {
      "epoch": 0.8680334275235977,
      "flos": 23981149532160.0,
      "grad_norm": 3.2179985280243972,
      "learning_rate": 1.7979580559342677e-07,
      "loss": 0.8788,
      "num_input_tokens_seen": 154163915,
      "step": 7219
    },
    {
      "epoch": 0.8681536704142367,
      "flos": 17556378193920.0,
      "grad_norm": 3.0873914938945246,
      "learning_rate": 1.7947315020744358e-07,
      "loss": 0.8607,
      "num_input_tokens_seen": 154184730,
      "step": 7220
    },
    {
      "epoch": 0.8682739133048758,
      "flos": 14200067297280.0,
      "grad_norm": 13.5190154481206,
      "learning_rate": 1.7915077099393594e-07,
      "loss": 0.987,
      "num_input_tokens_seen": 154201050,
      "step": 7221
    },
    {
      "epoch": 0.868394156195515,
      "flos": 11788025425920.0,
      "grad_norm": 3.601348804414753,
      "learning_rate": 1.788286680018083e-07,
      "loss": 0.9318,
      "num_input_tokens_seen": 154219480,
      "step": 7222
    },
    {
      "epoch": 0.868514399086154,
      "flos": 19942566789120.0,
      "grad_norm": 6.122267132867644,
      "learning_rate": 1.7850684127992443e-07,
      "loss": 0.9105,
      "num_input_tokens_seen": 154238945,
      "step": 7223
    },
    {
      "epoch": 0.8686346419767931,
      "flos": 14253000192000.0,
      "grad_norm": 3.8204329458022923,
      "learning_rate": 1.7818529087710378e-07,
      "loss": 0.8957,
      "num_input_tokens_seen": 154259020,
      "step": 7224
    },
    {
      "epoch": 0.8687548848674322,
      "flos": 12915444142080.0,
      "grad_norm": 4.351130739773239,
      "learning_rate": 1.7786401684212637e-07,
      "loss": 1.0313,
      "num_input_tokens_seen": 154277570,
      "step": 7225
    },
    {
      "epoch": 0.8688751277580713,
      "flos": 50458431037440.0,
      "grad_norm": 0.737276675469015,
      "learning_rate": 1.7754301922372883e-07,
      "loss": 0.7636,
      "num_input_tokens_seen": 154326935,
      "step": 7226
    },
    {
      "epoch": 0.8689953706487104,
      "flos": 19155900764160.0,
      "grad_norm": 4.067734383252588,
      "learning_rate": 1.7722229807060617e-07,
      "loss": 1.0062,
      "num_input_tokens_seen": 154345235,
      "step": 7227
    },
    {
      "epoch": 0.8691156135393495,
      "flos": 24711142318080.0,
      "grad_norm": 8.050186098210421,
      "learning_rate": 1.7690185343141172e-07,
      "loss": 1.0048,
      "num_input_tokens_seen": 154364870,
      "step": 7228
    },
    {
      "epoch": 0.8692358564299886,
      "flos": 13466180874240.0,
      "grad_norm": 4.376853114444761,
      "learning_rate": 1.7658168535475615e-07,
      "loss": 0.895,
      "num_input_tokens_seen": 154382375,
      "step": 7229
    },
    {
      "epoch": 0.8693560993206276,
      "flos": 21645165219840.0,
      "grad_norm": 2.331466427490357,
      "learning_rate": 1.7626179388920948e-07,
      "loss": 0.8408,
      "num_input_tokens_seen": 154403375,
      "step": 7230
    },
    {
      "epoch": 0.8694763422112668,
      "flos": 19367172464640.0,
      "grad_norm": 2.365582012960084,
      "learning_rate": 1.7594217908329866e-07,
      "loss": 0.9975,
      "num_input_tokens_seen": 154425280,
      "step": 7231
    },
    {
      "epoch": 0.8695965851019059,
      "flos": 18604029665280.0,
      "grad_norm": 3.853184356603525,
      "learning_rate": 1.7562284098550895e-07,
      "loss": 0.9275,
      "num_input_tokens_seen": 154444710,
      "step": 7232
    },
    {
      "epoch": 0.8697168279925449,
      "flos": 48214048174080.0,
      "grad_norm": 0.8741160082677136,
      "learning_rate": 1.753037796442838e-07,
      "loss": 0.8516,
      "num_input_tokens_seen": 154503870,
      "step": 7233
    },
    {
      "epoch": 0.8698370708831841,
      "flos": 13989899304960.0,
      "grad_norm": 4.440825814438636,
      "learning_rate": 1.74984995108024e-07,
      "loss": 0.9424,
      "num_input_tokens_seen": 154521520,
      "step": 7234
    },
    {
      "epoch": 0.8699573137738231,
      "flos": 9060746895360.0,
      "grad_norm": 3.149500059387914,
      "learning_rate": 1.7466648742508981e-07,
      "loss": 1.0124,
      "num_input_tokens_seen": 154537425,
      "step": 7235
    },
    {
      "epoch": 0.8700775566644622,
      "flos": 12338486231040.0,
      "grad_norm": 3.4154668577672287,
      "learning_rate": 1.7434825664379837e-07,
      "loss": 1.0317,
      "num_input_tokens_seen": 154555650,
      "step": 7236
    },
    {
      "epoch": 0.8701977995551013,
      "flos": 9689717944320.0,
      "grad_norm": 10.260238744028944,
      "learning_rate": 1.740303028124246e-07,
      "loss": 1.0392,
      "num_input_tokens_seen": 154571430,
      "step": 7237
    },
    {
      "epoch": 0.8703180424457404,
      "flos": 21778355896320.0,
      "grad_norm": 16.52328605131539,
      "learning_rate": 1.7371262597920212e-07,
      "loss": 0.9503,
      "num_input_tokens_seen": 154593210,
      "step": 7238
    },
    {
      "epoch": 0.8704382853363795,
      "flos": 13911021158400.0,
      "grad_norm": 2.781830116377835,
      "learning_rate": 1.7339522619232195e-07,
      "loss": 0.9483,
      "num_input_tokens_seen": 154611310,
      "step": 7239
    },
    {
      "epoch": 0.8705585282270186,
      "flos": 18945272893440.0,
      "grad_norm": 3.1053170413022166,
      "learning_rate": 1.730781034999338e-07,
      "loss": 0.9263,
      "num_input_tokens_seen": 154632610,
      "step": 7240
    },
    {
      "epoch": 0.8706787711176577,
      "flos": 24318131220480.0,
      "grad_norm": 3.4099129396662775,
      "learning_rate": 1.7276125795014497e-07,
      "loss": 0.9278,
      "num_input_tokens_seen": 154650780,
      "step": 7241
    },
    {
      "epoch": 0.8707990140082967,
      "flos": 10319578091520.0,
      "grad_norm": 5.373495024306352,
      "learning_rate": 1.7244468959102054e-07,
      "loss": 0.8582,
      "num_input_tokens_seen": 154667555,
      "step": 7242
    },
    {
      "epoch": 0.8709192568989359,
      "flos": 14252632289280.0,
      "grad_norm": 4.928517536792746,
      "learning_rate": 1.7212839847058348e-07,
      "loss": 1.0467,
      "num_input_tokens_seen": 154682405,
      "step": 7243
    },
    {
      "epoch": 0.871039499789575,
      "flos": 11839915929600.0,
      "grad_norm": 3.3673264193809076,
      "learning_rate": 1.718123846368147e-07,
      "loss": 0.9307,
      "num_input_tokens_seen": 154701170,
      "step": 7244
    },
    {
      "epoch": 0.871159742680214,
      "flos": 14960205557760.0,
      "grad_norm": 2.7653396373016004,
      "learning_rate": 1.714966481376543e-07,
      "loss": 0.9169,
      "num_input_tokens_seen": 154717570,
      "step": 7245
    },
    {
      "epoch": 0.8712799855708532,
      "flos": 20125900431360.0,
      "grad_norm": 5.257890675205885,
      "learning_rate": 1.7118118902099797e-07,
      "loss": 0.9949,
      "num_input_tokens_seen": 154735375,
      "step": 7246
    },
    {
      "epoch": 0.8714002284614922,
      "flos": 15799589867520.0,
      "grad_norm": 3.2272908494760455,
      "learning_rate": 1.7086600733470146e-07,
      "loss": 0.9886,
      "num_input_tokens_seen": 154755765,
      "step": 7247
    },
    {
      "epoch": 0.8715204713521313,
      "flos": 15220945735680.0,
      "grad_norm": 5.545532579910485,
      "learning_rate": 1.7055110312657738e-07,
      "loss": 0.9587,
      "num_input_tokens_seen": 154774980,
      "step": 7248
    },
    {
      "epoch": 0.8716407142427703,
      "flos": 16664674160640.0,
      "grad_norm": 4.6765215677241,
      "learning_rate": 1.702364764443962e-07,
      "loss": 0.9399,
      "num_input_tokens_seen": 154793775,
      "step": 7249
    },
    {
      "epoch": 0.8717609571334095,
      "flos": 19911041679360.0,
      "grad_norm": 3.137972038596687,
      "learning_rate": 1.6992212733588685e-07,
      "loss": 0.9192,
      "num_input_tokens_seen": 154813160,
      "step": 7250
    },
    {
      "epoch": 0.8718812000240486,
      "flos": 18129534443520.0,
      "grad_norm": 3.032652854746541,
      "learning_rate": 1.6960805584873538e-07,
      "loss": 0.9372,
      "num_input_tokens_seen": 154830880,
      "step": 7251
    },
    {
      "epoch": 0.8720014429146876,
      "flos": 16637686517760.0,
      "grad_norm": 2.5330875461538023,
      "learning_rate": 1.6929426203058684e-07,
      "loss": 0.9757,
      "num_input_tokens_seen": 154851025,
      "step": 7252
    },
    {
      "epoch": 0.8721216858053268,
      "flos": 17320418242560.0,
      "grad_norm": 4.2483757505597755,
      "learning_rate": 1.689807459290431e-07,
      "loss": 1.0046,
      "num_input_tokens_seen": 154869400,
      "step": 7253
    },
    {
      "epoch": 0.8722419286959658,
      "flos": 24159209902080.0,
      "grad_norm": 2.764486132256672,
      "learning_rate": 1.6866750759166437e-07,
      "loss": 0.9024,
      "num_input_tokens_seen": 154889100,
      "step": 7254
    },
    {
      "epoch": 0.8723621715866049,
      "flos": 13020267540480.0,
      "grad_norm": 5.4482399994674315,
      "learning_rate": 1.6835454706596865e-07,
      "loss": 0.958,
      "num_input_tokens_seen": 154906650,
      "step": 7255
    },
    {
      "epoch": 0.8724824144772441,
      "flos": 15638491791360.0,
      "grad_norm": 10.000316614860798,
      "learning_rate": 1.680418643994317e-07,
      "loss": 0.9337,
      "num_input_tokens_seen": 154924940,
      "step": 7256
    },
    {
      "epoch": 0.8726026573678831,
      "flos": 47756637880320.0,
      "grad_norm": 0.9771655242664266,
      "learning_rate": 1.6772945963948738e-07,
      "loss": 0.876,
      "num_input_tokens_seen": 154982825,
      "step": 7257
    },
    {
      "epoch": 0.8727229002585222,
      "flos": 9427966033920.0,
      "grad_norm": 4.495728478727709,
      "learning_rate": 1.6741733283352733e-07,
      "loss": 0.9463,
      "num_input_tokens_seen": 155000150,
      "step": 7258
    },
    {
      "epoch": 0.8728431431491613,
      "flos": 15483341475840.0,
      "grad_norm": 4.6295891752948055,
      "learning_rate": 1.6710548402890102e-07,
      "loss": 1.0332,
      "num_input_tokens_seen": 155020395,
      "step": 7259
    },
    {
      "epoch": 0.8729633860398004,
      "flos": 25815896248320.0,
      "grad_norm": 7.847320595033711,
      "learning_rate": 1.6679391327291527e-07,
      "loss": 0.8585,
      "num_input_tokens_seen": 155041320,
      "step": 7260
    },
    {
      "epoch": 0.8730836289304394,
      "flos": 11575527383040.0,
      "grad_norm": 4.266990806267513,
      "learning_rate": 1.6648262061283492e-07,
      "loss": 0.8604,
      "num_input_tokens_seen": 155056340,
      "step": 7261
    },
    {
      "epoch": 0.8732038718210786,
      "flos": 15065427517440.0,
      "grad_norm": 4.29010964206823,
      "learning_rate": 1.6617160609588353e-07,
      "loss": 0.9306,
      "num_input_tokens_seen": 155075235,
      "step": 7262
    },
    {
      "epoch": 0.8733241147117177,
      "flos": 11755335290880.0,
      "grad_norm": 5.640844789909368,
      "learning_rate": 1.6586086976924163e-07,
      "loss": 0.9049,
      "num_input_tokens_seen": 155090455,
      "step": 7263
    },
    {
      "epoch": 0.8734443576023567,
      "flos": 14331571752960.0,
      "grad_norm": 3.1735719548060577,
      "learning_rate": 1.6555041168004747e-07,
      "loss": 0.9766,
      "num_input_tokens_seen": 155109495,
      "step": 7264
    },
    {
      "epoch": 0.8735646004929959,
      "flos": 12784399564800.0,
      "grad_norm": 2.719465402771982,
      "learning_rate": 1.6524023187539715e-07,
      "loss": 0.8749,
      "num_input_tokens_seen": 155127500,
      "step": 7265
    },
    {
      "epoch": 0.873684843383635,
      "flos": 14380120473600.0,
      "grad_norm": 3.6095759348239587,
      "learning_rate": 1.649303304023446e-07,
      "loss": 0.9382,
      "num_input_tokens_seen": 155146975,
      "step": 7266
    },
    {
      "epoch": 0.873805086274274,
      "flos": 11998346711040.0,
      "grad_norm": 3.342014611094123,
      "learning_rate": 1.6462070730790246e-07,
      "loss": 0.9792,
      "num_input_tokens_seen": 155165855,
      "step": 7267
    },
    {
      "epoch": 0.8739253291649132,
      "flos": 12784491540480.0,
      "grad_norm": 6.013902452241649,
      "learning_rate": 1.6431136263903912e-07,
      "loss": 0.9654,
      "num_input_tokens_seen": 155184575,
      "step": 7268
    },
    {
      "epoch": 0.8740455720555522,
      "flos": 15144520273920.0,
      "grad_norm": 4.903766756867798,
      "learning_rate": 1.6400229644268282e-07,
      "loss": 0.9202,
      "num_input_tokens_seen": 155202650,
      "step": 7269
    },
    {
      "epoch": 0.8741658149461913,
      "flos": 10712803799040.0,
      "grad_norm": 3.8408215539980195,
      "learning_rate": 1.6369350876571852e-07,
      "loss": 0.9962,
      "num_input_tokens_seen": 155220525,
      "step": 7270
    },
    {
      "epoch": 0.8742860578368304,
      "flos": 16376885022720.0,
      "grad_norm": 4.532433716951586,
      "learning_rate": 1.6338499965498874e-07,
      "loss": 0.9885,
      "num_input_tokens_seen": 155238975,
      "step": 7271
    },
    {
      "epoch": 0.8744063007274695,
      "flos": 20045765283840.0,
      "grad_norm": 3.188853705681368,
      "learning_rate": 1.630767691572943e-07,
      "loss": 0.9648,
      "num_input_tokens_seen": 155258715,
      "step": 7272
    },
    {
      "epoch": 0.8745265436181086,
      "flos": 45841449431040.0,
      "grad_norm": 0.7625686208146194,
      "learning_rate": 1.6276881731939306e-07,
      "loss": 0.7512,
      "num_input_tokens_seen": 155320325,
      "step": 7273
    },
    {
      "epoch": 0.8746467865087477,
      "flos": 20414977228800.0,
      "grad_norm": 3.361773960618483,
      "learning_rate": 1.6246114418800193e-07,
      "loss": 0.944,
      "num_input_tokens_seen": 155340325,
      "step": 7274
    },
    {
      "epoch": 0.8747670293993868,
      "flos": 17056397598720.0,
      "grad_norm": 2.5731796459726626,
      "learning_rate": 1.6215374980979423e-07,
      "loss": 0.9574,
      "num_input_tokens_seen": 155360455,
      "step": 7275
    },
    {
      "epoch": 0.8748872722900258,
      "flos": 32318288732160.0,
      "grad_norm": 4.887841941013645,
      "learning_rate": 1.6184663423140133e-07,
      "loss": 0.872,
      "num_input_tokens_seen": 155380475,
      "step": 7276
    },
    {
      "epoch": 0.875007515180665,
      "flos": 14015354019840.0,
      "grad_norm": 2.949295386524981,
      "learning_rate": 1.615397974994126e-07,
      "loss": 0.8363,
      "num_input_tokens_seen": 155398000,
      "step": 7277
    },
    {
      "epoch": 0.875127758071304,
      "flos": 16140250583040.0,
      "grad_norm": 3.2264790004612585,
      "learning_rate": 1.6123323966037438e-07,
      "loss": 0.9967,
      "num_input_tokens_seen": 155416240,
      "step": 7278
    },
    {
      "epoch": 0.8752480009619431,
      "flos": 16638361006080.0,
      "grad_norm": 7.697206897216333,
      "learning_rate": 1.6092696076079216e-07,
      "loss": 0.9789,
      "num_input_tokens_seen": 155434335,
      "step": 7279
    },
    {
      "epoch": 0.8753682438525822,
      "flos": 18657299804160.0,
      "grad_norm": 2.705461583691193,
      "learning_rate": 1.6062096084712785e-07,
      "loss": 0.9303,
      "num_input_tokens_seen": 155455405,
      "step": 7280
    },
    {
      "epoch": 0.8754884867432213,
      "flos": 16582392913920.0,
      "grad_norm": 2.6666674812633544,
      "learning_rate": 1.6031523996580098e-07,
      "loss": 0.8964,
      "num_input_tokens_seen": 155472685,
      "step": 7281
    },
    {
      "epoch": 0.8756087296338604,
      "flos": 8798627082240.0,
      "grad_norm": 6.263265036873271,
      "learning_rate": 1.6000979816318981e-07,
      "loss": 0.8518,
      "num_input_tokens_seen": 155490870,
      "step": 7282
    },
    {
      "epoch": 0.8757289725244994,
      "flos": 13439591792640.0,
      "grad_norm": 4.079277957155335,
      "learning_rate": 1.5970463548562886e-07,
      "loss": 0.9364,
      "num_input_tokens_seen": 155509745,
      "step": 7283
    },
    {
      "epoch": 0.8758492154151386,
      "flos": 17976806154240.0,
      "grad_norm": 3.412709402289849,
      "learning_rate": 1.5939975197941192e-07,
      "loss": 0.907,
      "num_input_tokens_seen": 155531120,
      "step": 7284
    },
    {
      "epoch": 0.8759694583057777,
      "flos": 46947184435200.0,
      "grad_norm": 0.8114658523491969,
      "learning_rate": 1.5909514769078892e-07,
      "loss": 0.7513,
      "num_input_tokens_seen": 155595945,
      "step": 7285
    },
    {
      "epoch": 0.8760897011964167,
      "flos": 18106041876480.0,
      "grad_norm": 3.5020999058579627,
      "learning_rate": 1.5879082266596867e-07,
      "loss": 0.9597,
      "num_input_tokens_seen": 155617005,
      "step": 7286
    },
    {
      "epoch": 0.8762099440870559,
      "flos": 20649128325120.0,
      "grad_norm": 3.4344933972763814,
      "learning_rate": 1.5848677695111645e-07,
      "loss": 0.9244,
      "num_input_tokens_seen": 155638325,
      "step": 7287
    },
    {
      "epoch": 0.8763301869776949,
      "flos": 15347882065920.0,
      "grad_norm": 5.0570006465236315,
      "learning_rate": 1.5818301059235562e-07,
      "loss": 0.8982,
      "num_input_tokens_seen": 155653220,
      "step": 7288
    },
    {
      "epoch": 0.876450429868334,
      "flos": 17529942405120.0,
      "grad_norm": 2.7649371482507257,
      "learning_rate": 1.578795236357684e-07,
      "loss": 1.0064,
      "num_input_tokens_seen": 155674405,
      "step": 7289
    },
    {
      "epoch": 0.8765706727589732,
      "flos": 14378158325760.0,
      "grad_norm": 5.600152470011681,
      "learning_rate": 1.5757631612739218e-07,
      "loss": 1.0487,
      "num_input_tokens_seen": 155687670,
      "step": 7290
    },
    {
      "epoch": 0.8766909156496122,
      "flos": 51114941583360.0,
      "grad_norm": 0.8490520885785323,
      "learning_rate": 1.572733881132242e-07,
      "loss": 0.8607,
      "num_input_tokens_seen": 155748035,
      "step": 7291
    },
    {
      "epoch": 0.8768111585402513,
      "flos": 49788821176320.0,
      "grad_norm": 0.7585005635188404,
      "learning_rate": 1.5697073963921814e-07,
      "loss": 0.8079,
      "num_input_tokens_seen": 155806995,
      "step": 7292
    },
    {
      "epoch": 0.8769314014308904,
      "flos": 13356758691840.0,
      "grad_norm": 3.3711448126292773,
      "learning_rate": 1.566683707512857e-07,
      "loss": 1.0365,
      "num_input_tokens_seen": 155824390,
      "step": 7293
    },
    {
      "epoch": 0.8770516443215295,
      "flos": 10581820538880.0,
      "grad_norm": 4.242968857887589,
      "learning_rate": 1.5636628149529553e-07,
      "loss": 0.9896,
      "num_input_tokens_seen": 155841900,
      "step": 7294
    },
    {
      "epoch": 0.8771718872121685,
      "flos": 22564470067200.0,
      "grad_norm": 5.100782910260681,
      "learning_rate": 1.560644719170743e-07,
      "loss": 0.9909,
      "num_input_tokens_seen": 155862490,
      "step": 7295
    },
    {
      "epoch": 0.8772921301028077,
      "flos": 25758855106560.0,
      "grad_norm": 4.503960032609281,
      "learning_rate": 1.5576294206240692e-07,
      "loss": 0.9113,
      "num_input_tokens_seen": 155881735,
      "step": 7296
    },
    {
      "epoch": 0.8774123729934468,
      "flos": 40867374120960.0,
      "grad_norm": 3.0912533328681993,
      "learning_rate": 1.5546169197703507e-07,
      "loss": 0.8767,
      "num_input_tokens_seen": 155907730,
      "step": 7297
    },
    {
      "epoch": 0.8775326158840858,
      "flos": 17005334876160.0,
      "grad_norm": 5.714910166544171,
      "learning_rate": 1.5516072170665774e-07,
      "loss": 0.9634,
      "num_input_tokens_seen": 155925420,
      "step": 7298
    },
    {
      "epoch": 0.877652858774725,
      "flos": 12123964723200.0,
      "grad_norm": 3.1616641904036644,
      "learning_rate": 1.5486003129693214e-07,
      "loss": 1.0469,
      "num_input_tokens_seen": 155942505,
      "step": 7299
    },
    {
      "epoch": 0.877773101665364,
      "flos": 11683263344640.0,
      "grad_norm": 3.0883113953218997,
      "learning_rate": 1.545596207934725e-07,
      "loss": 0.9661,
      "num_input_tokens_seen": 155960710,
      "step": 7300
    },
    {
      "epoch": 0.8778933445560031,
      "flos": 15667594874880.0,
      "grad_norm": 3.1142398980846457,
      "learning_rate": 1.5425949024185147e-07,
      "loss": 0.9642,
      "num_input_tokens_seen": 155980455,
      "step": 7301
    },
    {
      "epoch": 0.8780135874466423,
      "flos": 16035335208960.0,
      "grad_norm": 3.4525438743195664,
      "learning_rate": 1.5395963968759818e-07,
      "loss": 0.8601,
      "num_input_tokens_seen": 156000450,
      "step": 7302
    },
    {
      "epoch": 0.8781338303372813,
      "flos": 44040412692480.0,
      "grad_norm": 3.7403573994535564,
      "learning_rate": 1.536600691761998e-07,
      "loss": 0.8376,
      "num_input_tokens_seen": 156026000,
      "step": 7303
    },
    {
      "epoch": 0.8782540732279204,
      "flos": 16114458624000.0,
      "grad_norm": 3.004173872079731,
      "learning_rate": 1.5336077875310084e-07,
      "loss": 0.9022,
      "num_input_tokens_seen": 156044945,
      "step": 7304
    },
    {
      "epoch": 0.8783743161185595,
      "flos": 11367934709760.0,
      "grad_norm": 2.99017760498854,
      "learning_rate": 1.5306176846370321e-07,
      "loss": 0.9265,
      "num_input_tokens_seen": 156062810,
      "step": 7305
    },
    {
      "epoch": 0.8784945590091986,
      "flos": 18553028259840.0,
      "grad_norm": 2.8241899628708786,
      "learning_rate": 1.5276303835336712e-07,
      "loss": 0.9219,
      "num_input_tokens_seen": 156083070,
      "step": 7306
    },
    {
      "epoch": 0.8786148018998376,
      "flos": 44897517649920.0,
      "grad_norm": 0.7844326423650838,
      "learning_rate": 1.524645884674094e-07,
      "loss": 0.7499,
      "num_input_tokens_seen": 156139720,
      "step": 7307
    },
    {
      "epoch": 0.8787350447904768,
      "flos": 15379897712640.0,
      "grad_norm": 3.280925480371577,
      "learning_rate": 1.521664188511047e-07,
      "loss": 0.9858,
      "num_input_tokens_seen": 156159465,
      "step": 7308
    },
    {
      "epoch": 0.8788552876811159,
      "flos": 18128676003840.0,
      "grad_norm": 3.0371313554268915,
      "learning_rate": 1.518685295496851e-07,
      "loss": 0.9971,
      "num_input_tokens_seen": 156177045,
      "step": 7309
    },
    {
      "epoch": 0.8789755305717549,
      "flos": 15852522762240.0,
      "grad_norm": 2.880416743008435,
      "learning_rate": 1.5157092060833975e-07,
      "loss": 1.039,
      "num_input_tokens_seen": 156196415,
      "step": 7310
    },
    {
      "epoch": 0.879095773462394,
      "flos": 20883218104320.0,
      "grad_norm": 2.938837761325068,
      "learning_rate": 1.5127359207221658e-07,
      "loss": 0.8434,
      "num_input_tokens_seen": 156215615,
      "step": 7311
    },
    {
      "epoch": 0.8792160163530331,
      "flos": 11813204213760.0,
      "grad_norm": 3.66066338639977,
      "learning_rate": 1.5097654398641923e-07,
      "loss": 0.9098,
      "num_input_tokens_seen": 156233240,
      "step": 7312
    },
    {
      "epoch": 0.8793362592436722,
      "flos": 17425762836480.0,
      "grad_norm": 2.5780535138940985,
      "learning_rate": 1.5067977639601014e-07,
      "loss": 0.9223,
      "num_input_tokens_seen": 156255720,
      "step": 7313
    },
    {
      "epoch": 0.8794565021343113,
      "flos": 10267013099520.0,
      "grad_norm": 4.3701471935073215,
      "learning_rate": 1.5038328934600864e-07,
      "loss": 0.8953,
      "num_input_tokens_seen": 156272075,
      "step": 7314
    },
    {
      "epoch": 0.8795767450249504,
      "flos": 28228091412480.0,
      "grad_norm": 3.2058495819920183,
      "learning_rate": 1.5008708288139161e-07,
      "loss": 0.8838,
      "num_input_tokens_seen": 156294155,
      "step": 7315
    },
    {
      "epoch": 0.8796969879155895,
      "flos": 16319138734080.0,
      "grad_norm": 15.552430598918605,
      "learning_rate": 1.497911570470931e-07,
      "loss": 0.9264,
      "num_input_tokens_seen": 156313880,
      "step": 7316
    },
    {
      "epoch": 0.8798172308062285,
      "flos": 20201528770560.0,
      "grad_norm": 2.8328928511554987,
      "learning_rate": 1.494955118880048e-07,
      "loss": 1.0468,
      "num_input_tokens_seen": 156334585,
      "step": 7317
    },
    {
      "epoch": 0.8799374736968677,
      "flos": 17059064893440.0,
      "grad_norm": 3.461642707983008,
      "learning_rate": 1.4920014744897634e-07,
      "loss": 0.9248,
      "num_input_tokens_seen": 156353720,
      "step": 7318
    },
    {
      "epoch": 0.8800577165875068,
      "flos": 18238558064640.0,
      "grad_norm": 3.962400267877229,
      "learning_rate": 1.4890506377481392e-07,
      "loss": 1.0494,
      "num_input_tokens_seen": 156372530,
      "step": 7319
    },
    {
      "epoch": 0.8801779594781458,
      "flos": 16664582184960.0,
      "grad_norm": 2.545222395677248,
      "learning_rate": 1.486102609102815e-07,
      "loss": 0.8262,
      "num_input_tokens_seen": 156392800,
      "step": 7320
    },
    {
      "epoch": 0.880298202368785,
      "flos": 8089336934400.0,
      "grad_norm": 4.327143699042486,
      "learning_rate": 1.483157389001004e-07,
      "loss": 1.0429,
      "num_input_tokens_seen": 156410080,
      "step": 7321
    },
    {
      "epoch": 0.880418445259424,
      "flos": 16111638036480.0,
      "grad_norm": 3.5937302298623996,
      "learning_rate": 1.4802149778894933e-07,
      "loss": 0.9674,
      "num_input_tokens_seen": 156428590,
      "step": 7322
    },
    {
      "epoch": 0.8805386881500631,
      "flos": 14567439728640.0,
      "grad_norm": 8.496840394634615,
      "learning_rate": 1.4772753762146484e-07,
      "loss": 1.07,
      "num_input_tokens_seen": 156447565,
      "step": 7323
    },
    {
      "epoch": 0.8806589310407023,
      "flos": 26078291988480.0,
      "grad_norm": 2.848683029532521,
      "learning_rate": 1.474338584422401e-07,
      "loss": 0.8878,
      "num_input_tokens_seen": 156472495,
      "step": 7324
    },
    {
      "epoch": 0.8807791739313413,
      "flos": 16664459550720.0,
      "grad_norm": 3.0551783955084106,
      "learning_rate": 1.4714046029582595e-07,
      "loss": 0.9418,
      "num_input_tokens_seen": 156491280,
      "step": 7325
    },
    {
      "epoch": 0.8808994168219804,
      "flos": 18473015746560.0,
      "grad_norm": 3.6192470808682082,
      "learning_rate": 1.46847343226731e-07,
      "loss": 0.9596,
      "num_input_tokens_seen": 156512040,
      "step": 7326
    },
    {
      "epoch": 0.8810196597126195,
      "flos": 12102250352640.0,
      "grad_norm": 10.504398333002177,
      "learning_rate": 1.465545072794203e-07,
      "loss": 0.884,
      "num_input_tokens_seen": 156529635,
      "step": 7327
    },
    {
      "epoch": 0.8811399026032586,
      "flos": 16350203965440.0,
      "grad_norm": 2.751849506233227,
      "learning_rate": 1.4626195249831774e-07,
      "loss": 0.9505,
      "num_input_tokens_seen": 156550255,
      "step": 7328
    },
    {
      "epoch": 0.8812601454938976,
      "flos": 10213191106560.0,
      "grad_norm": 3.388804848120502,
      "learning_rate": 1.4596967892780244e-07,
      "loss": 0.9133,
      "num_input_tokens_seen": 156566305,
      "step": 7329
    },
    {
      "epoch": 0.8813803883845368,
      "flos": 15984119193600.0,
      "grad_norm": 3.659579725264086,
      "learning_rate": 1.4567768661221314e-07,
      "loss": 0.9393,
      "num_input_tokens_seen": 156586595,
      "step": 7330
    },
    {
      "epoch": 0.8815006312751759,
      "flos": 15275104972800.0,
      "grad_norm": 3.670792959960425,
      "learning_rate": 1.4538597559584442e-07,
      "loss": 0.9293,
      "num_input_tokens_seen": 156604105,
      "step": 7331
    },
    {
      "epoch": 0.8816208741658149,
      "flos": 16193428746240.0,
      "grad_norm": 3.413379871552824,
      "learning_rate": 1.4509454592294823e-07,
      "loss": 0.9685,
      "num_input_tokens_seen": 156624310,
      "step": 7332
    },
    {
      "epoch": 0.8817411170564541,
      "flos": 12595424747520.0,
      "grad_norm": 4.241587671969531,
      "learning_rate": 1.448033976377354e-07,
      "loss": 0.9836,
      "num_input_tokens_seen": 156639015,
      "step": 7333
    },
    {
      "epoch": 0.8818613599470931,
      "flos": 13151373434880.0,
      "grad_norm": 5.899733404828296,
      "learning_rate": 1.445125307843713e-07,
      "loss": 0.9291,
      "num_input_tokens_seen": 156656960,
      "step": 7334
    },
    {
      "epoch": 0.8819816028377322,
      "flos": 19653919211520.0,
      "grad_norm": 3.2157111776387386,
      "learning_rate": 1.442219454069813e-07,
      "loss": 0.9348,
      "num_input_tokens_seen": 156677705,
      "step": 7335
    },
    {
      "epoch": 0.8821018457283714,
      "flos": 16639004835840.0,
      "grad_norm": 2.8775825511183823,
      "learning_rate": 1.4393164154964676e-07,
      "loss": 0.8528,
      "num_input_tokens_seen": 156696955,
      "step": 7336
    },
    {
      "epoch": 0.8822220886190104,
      "flos": 20755453992960.0,
      "grad_norm": 3.6469987732216036,
      "learning_rate": 1.4364161925640649e-07,
      "loss": 1.1325,
      "num_input_tokens_seen": 156718075,
      "step": 7337
    },
    {
      "epoch": 0.8823423315096495,
      "flos": 14541003939840.0,
      "grad_norm": 4.953935141750634,
      "learning_rate": 1.4335187857125663e-07,
      "loss": 1.0423,
      "num_input_tokens_seen": 156736495,
      "step": 7338
    },
    {
      "epoch": 0.8824625744002886,
      "flos": 17215104307200.0,
      "grad_norm": 2.916827079358059,
      "learning_rate": 1.4306241953815023e-07,
      "loss": 0.9313,
      "num_input_tokens_seen": 156757275,
      "step": 7339
    },
    {
      "epoch": 0.8825828172909277,
      "flos": 17554967900160.0,
      "grad_norm": 3.1576782620837003,
      "learning_rate": 1.4277324220099862e-07,
      "loss": 0.892,
      "num_input_tokens_seen": 156778905,
      "step": 7340
    },
    {
      "epoch": 0.8827030601815667,
      "flos": 15957407477760.0,
      "grad_norm": 6.081030144005593,
      "learning_rate": 1.4248434660366938e-07,
      "loss": 0.9335,
      "num_input_tokens_seen": 156798100,
      "step": 7341
    },
    {
      "epoch": 0.8828233030722058,
      "flos": 14095642460160.0,
      "grad_norm": 4.254770406587006,
      "learning_rate": 1.4219573278998808e-07,
      "loss": 0.8949,
      "num_input_tokens_seen": 156816280,
      "step": 7342
    },
    {
      "epoch": 0.882943545962845,
      "flos": 28096433664000.0,
      "grad_norm": 4.071929083440831,
      "learning_rate": 1.4190740080373685e-07,
      "loss": 0.8528,
      "num_input_tokens_seen": 156836280,
      "step": 7343
    },
    {
      "epoch": 0.883063788853484,
      "flos": 13511571763200.0,
      "grad_norm": 2.982873508954472,
      "learning_rate": 1.4161935068865538e-07,
      "loss": 1.0416,
      "num_input_tokens_seen": 156851145,
      "step": 7344
    },
    {
      "epoch": 0.8831840317441231,
      "flos": 13281590231040.0,
      "grad_norm": 4.562861807077155,
      "learning_rate": 1.4133158248844113e-07,
      "loss": 0.9508,
      "num_input_tokens_seen": 156869770,
      "step": 7345
    },
    {
      "epoch": 0.8833042746347622,
      "flos": 19098246451200.0,
      "grad_norm": 3.3119502061150685,
      "learning_rate": 1.4104409624674785e-07,
      "loss": 0.9188,
      "num_input_tokens_seen": 156889275,
      "step": 7346
    },
    {
      "epoch": 0.8834245175254013,
      "flos": 18579004170240.0,
      "grad_norm": 2.6300470787447083,
      "learning_rate": 1.407568920071873e-07,
      "loss": 0.9728,
      "num_input_tokens_seen": 156907860,
      "step": 7347
    },
    {
      "epoch": 0.8835447604160404,
      "flos": 21831104839680.0,
      "grad_norm": 3.314392988460784,
      "learning_rate": 1.4046996981332782e-07,
      "loss": 0.8609,
      "num_input_tokens_seen": 156927465,
      "step": 7348
    },
    {
      "epoch": 0.8836650033066795,
      "flos": 17583212544000.0,
      "grad_norm": 3.3314275697829543,
      "learning_rate": 1.4018332970869516e-07,
      "loss": 0.9689,
      "num_input_tokens_seen": 156945125,
      "step": 7349
    },
    {
      "epoch": 0.8837852461973186,
      "flos": 18082947870720.0,
      "grad_norm": 3.1562421439800126,
      "learning_rate": 1.398969717367733e-07,
      "loss": 1.03,
      "num_input_tokens_seen": 156966170,
      "step": 7350
    },
    {
      "epoch": 0.8839054890879576,
      "flos": 12627348418560.0,
      "grad_norm": 3.180260608958573,
      "learning_rate": 1.396108959410014e-07,
      "loss": 0.9512,
      "num_input_tokens_seen": 156985105,
      "step": 7351
    },
    {
      "epoch": 0.8840257319785968,
      "flos": 16900848721920.0,
      "grad_norm": 2.7420849033736303,
      "learning_rate": 1.3932510236477745e-07,
      "loss": 0.9929,
      "num_input_tokens_seen": 157005495,
      "step": 7352
    },
    {
      "epoch": 0.8841459748692359,
      "flos": 20703256903680.0,
      "grad_norm": 4.13233019028171,
      "learning_rate": 1.3903959105145636e-07,
      "loss": 0.7466,
      "num_input_tokens_seen": 157025705,
      "step": 7353
    },
    {
      "epoch": 0.8842662177598749,
      "flos": 17290088816640.0,
      "grad_norm": 2.598463991251479,
      "learning_rate": 1.387543620443492e-07,
      "loss": 1.02,
      "num_input_tokens_seen": 157042270,
      "step": 7354
    },
    {
      "epoch": 0.8843864606505141,
      "flos": 17790958510080.0,
      "grad_norm": 2.8947908642852282,
      "learning_rate": 1.3846941538672606e-07,
      "loss": 1.0281,
      "num_input_tokens_seen": 157060695,
      "step": 7355
    },
    {
      "epoch": 0.8845067035411531,
      "flos": 20073764659200.0,
      "grad_norm": 3.5140633738691807,
      "learning_rate": 1.3818475112181193e-07,
      "loss": 1.0015,
      "num_input_tokens_seen": 157079210,
      "step": 7356
    },
    {
      "epoch": 0.8846269464317922,
      "flos": 9054983086080.0,
      "grad_norm": 4.011275849227853,
      "learning_rate": 1.3790036929279091e-07,
      "loss": 0.9794,
      "num_input_tokens_seen": 157096085,
      "step": 7357
    },
    {
      "epoch": 0.8847471893224313,
      "flos": 13203447889920.0,
      "grad_norm": 3.3741193964737164,
      "learning_rate": 1.3761626994280363e-07,
      "loss": 0.7794,
      "num_input_tokens_seen": 157113275,
      "step": 7358
    },
    {
      "epoch": 0.8848674322130704,
      "flos": 25524857303040.0,
      "grad_norm": 7.270078020477368,
      "learning_rate": 1.3733245311494735e-07,
      "loss": 0.9274,
      "num_input_tokens_seen": 157135650,
      "step": 7359
    },
    {
      "epoch": 0.8849876751037095,
      "flos": 17242091950080.0,
      "grad_norm": 4.755439254095214,
      "learning_rate": 1.3704891885227676e-07,
      "loss": 0.9011,
      "num_input_tokens_seen": 157155415,
      "step": 7360
    },
    {
      "epoch": 0.8851079179943486,
      "flos": 15269709066240.0,
      "grad_norm": 4.388455869878949,
      "learning_rate": 1.367656671978037e-07,
      "loss": 0.9671,
      "num_input_tokens_seen": 157172600,
      "step": 7361
    },
    {
      "epoch": 0.8852281608849877,
      "flos": 10814346731520.0,
      "grad_norm": 3.8558654890223534,
      "learning_rate": 1.36482698194498e-07,
      "loss": 0.9316,
      "num_input_tokens_seen": 157188865,
      "step": 7362
    },
    {
      "epoch": 0.8853484037756267,
      "flos": 16560831836160.0,
      "grad_norm": 4.0066492128758995,
      "learning_rate": 1.3620001188528506e-07,
      "loss": 0.9034,
      "num_input_tokens_seen": 157209305,
      "step": 7363
    },
    {
      "epoch": 0.8854686466662659,
      "flos": 17867445288960.0,
      "grad_norm": 3.1786152299339823,
      "learning_rate": 1.3591760831304865e-07,
      "loss": 0.9096,
      "num_input_tokens_seen": 157226715,
      "step": 7364
    },
    {
      "epoch": 0.885588889556905,
      "flos": 15190830919680.0,
      "grad_norm": 2.984589733766812,
      "learning_rate": 1.356354875206287e-07,
      "loss": 0.9896,
      "num_input_tokens_seen": 157244270,
      "step": 7365
    },
    {
      "epoch": 0.885709132447544,
      "flos": 19155870105600.0,
      "grad_norm": 2.9023859732060755,
      "learning_rate": 1.3535364955082296e-07,
      "loss": 0.8789,
      "num_input_tokens_seen": 157263840,
      "step": 7366
    },
    {
      "epoch": 0.8858293753381832,
      "flos": 18578421657600.0,
      "grad_norm": 3.928994331882007,
      "learning_rate": 1.3507209444638613e-07,
      "loss": 0.8296,
      "num_input_tokens_seen": 157285560,
      "step": 7367
    },
    {
      "epoch": 0.8859496182288222,
      "flos": 16558348492800.0,
      "grad_norm": 3.0534643665943646,
      "learning_rate": 1.347908222500298e-07,
      "loss": 0.9315,
      "num_input_tokens_seen": 157305355,
      "step": 7368
    },
    {
      "epoch": 0.8860698611194613,
      "flos": 11944708669440.0,
      "grad_norm": 3.789054823405561,
      "learning_rate": 1.3450983300442276e-07,
      "loss": 0.88,
      "num_input_tokens_seen": 157324305,
      "step": 7369
    },
    {
      "epoch": 0.8861901040101005,
      "flos": 17556592803840.0,
      "grad_norm": 4.622379694807508,
      "learning_rate": 1.3422912675219068e-07,
      "loss": 0.9222,
      "num_input_tokens_seen": 157343780,
      "step": 7370
    },
    {
      "epoch": 0.8863103469007395,
      "flos": 17370499891200.0,
      "grad_norm": 2.5714565154479754,
      "learning_rate": 1.339487035359166e-07,
      "loss": 0.9734,
      "num_input_tokens_seen": 157363870,
      "step": 7371
    },
    {
      "epoch": 0.8864305897913786,
      "flos": 15668238704640.0,
      "grad_norm": 2.4539817723139015,
      "learning_rate": 1.336685633981409e-07,
      "loss": 1.0431,
      "num_input_tokens_seen": 157384675,
      "step": 7372
    },
    {
      "epoch": 0.8865508326820177,
      "flos": 13544997703680.0,
      "grad_norm": 3.743281895280421,
      "learning_rate": 1.333887063813597e-07,
      "loss": 0.9309,
      "num_input_tokens_seen": 157402500,
      "step": 7373
    },
    {
      "epoch": 0.8866710755726568,
      "flos": 10896444026880.0,
      "grad_norm": 3.2799024448773775,
      "learning_rate": 1.331091325280278e-07,
      "loss": 0.8531,
      "num_input_tokens_seen": 157421190,
      "step": 7374
    },
    {
      "epoch": 0.8867913184632958,
      "flos": 14252233728000.0,
      "grad_norm": 2.650418781083172,
      "learning_rate": 1.3282984188055625e-07,
      "loss": 0.9759,
      "num_input_tokens_seen": 157440700,
      "step": 7375
    },
    {
      "epoch": 0.8869115613539349,
      "flos": 16610852167680.0,
      "grad_norm": 5.628113287744615,
      "learning_rate": 1.3255083448131288e-07,
      "loss": 0.9836,
      "num_input_tokens_seen": 157459465,
      "step": 7376
    },
    {
      "epoch": 0.8870318042445741,
      "flos": 15116796825600.0,
      "grad_norm": 4.609152992203882,
      "learning_rate": 1.3227211037262365e-07,
      "loss": 0.9762,
      "num_input_tokens_seen": 157476425,
      "step": 7377
    },
    {
      "epoch": 0.8871520471352131,
      "flos": 14199914004480.0,
      "grad_norm": 6.1001506349068695,
      "learning_rate": 1.319936695967696e-07,
      "loss": 1.051,
      "num_input_tokens_seen": 157493970,
      "step": 7378
    },
    {
      "epoch": 0.8872722900258522,
      "flos": 16061709680640.0,
      "grad_norm": 4.3368739311149875,
      "learning_rate": 1.3171551219599097e-07,
      "loss": 1.0124,
      "num_input_tokens_seen": 157512215,
      "step": 7379
    },
    {
      "epoch": 0.8873925329164913,
      "flos": 15746779607040.0,
      "grad_norm": 4.8716573748574445,
      "learning_rate": 1.3143763821248377e-07,
      "loss": 0.9676,
      "num_input_tokens_seen": 157529020,
      "step": 7380
    },
    {
      "epoch": 0.8875127758071304,
      "flos": 13623078727680.0,
      "grad_norm": 2.769862834412556,
      "learning_rate": 1.3116004768840118e-07,
      "loss": 0.9158,
      "num_input_tokens_seen": 157547115,
      "step": 7381
    },
    {
      "epoch": 0.8876330186977694,
      "flos": 12834082652160.0,
      "grad_norm": 5.206109734485519,
      "learning_rate": 1.3088274066585348e-07,
      "loss": 0.9255,
      "num_input_tokens_seen": 157564445,
      "step": 7382
    },
    {
      "epoch": 0.8877532615884086,
      "flos": 15635150008320.0,
      "grad_norm": 6.113498398699262,
      "learning_rate": 1.3060571718690749e-07,
      "loss": 1.0954,
      "num_input_tokens_seen": 157581660,
      "step": 7383
    },
    {
      "epoch": 0.8878735044790477,
      "flos": 51665218437120.0,
      "grad_norm": 0.7598759588231448,
      "learning_rate": 1.3032897729358805e-07,
      "loss": 0.7912,
      "num_input_tokens_seen": 157642335,
      "step": 7384
    },
    {
      "epoch": 0.8879937473696867,
      "flos": 19601078292480.0,
      "grad_norm": 6.685713484814122,
      "learning_rate": 1.3005252102787645e-07,
      "loss": 0.9921,
      "num_input_tokens_seen": 157660995,
      "step": 7385
    },
    {
      "epoch": 0.8881139902603259,
      "flos": 15797382451200.0,
      "grad_norm": 5.935559086035481,
      "learning_rate": 1.297763484317105e-07,
      "loss": 0.9254,
      "num_input_tokens_seen": 157679010,
      "step": 7386
    },
    {
      "epoch": 0.888234233150965,
      "flos": 14406954823680.0,
      "grad_norm": 5.566450966103288,
      "learning_rate": 1.2950045954698551e-07,
      "loss": 0.8931,
      "num_input_tokens_seen": 157696565,
      "step": 7387
    },
    {
      "epoch": 0.888354476041604,
      "flos": 12860825026560.0,
      "grad_norm": 6.2167821483096075,
      "learning_rate": 1.2922485441555343e-07,
      "loss": 0.9523,
      "num_input_tokens_seen": 157715365,
      "step": 7388
    },
    {
      "epoch": 0.8884747189322432,
      "flos": 15639288913920.0,
      "grad_norm": 3.5067277008324504,
      "learning_rate": 1.2894953307922363e-07,
      "loss": 1.0091,
      "num_input_tokens_seen": 157734045,
      "step": 7389
    },
    {
      "epoch": 0.8885949618228822,
      "flos": 14038172098560.0,
      "grad_norm": 2.8447981359537926,
      "learning_rate": 1.2867449557976208e-07,
      "loss": 1.0358,
      "num_input_tokens_seen": 157751865,
      "step": 7390
    },
    {
      "epoch": 0.8887152047135213,
      "flos": 14226349793280.0,
      "grad_norm": 3.146509228447778,
      "learning_rate": 1.283997419588916e-07,
      "loss": 0.9454,
      "num_input_tokens_seen": 157771055,
      "step": 7391
    },
    {
      "epoch": 0.8888354476041604,
      "flos": 13177441320960.0,
      "grad_norm": 3.732954066176098,
      "learning_rate": 1.2812527225829216e-07,
      "loss": 0.812,
      "num_input_tokens_seen": 157789000,
      "step": 7392
    },
    {
      "epoch": 0.8889556904947995,
      "flos": 15406609428480.0,
      "grad_norm": 3.936834733257161,
      "learning_rate": 1.2785108651960052e-07,
      "loss": 0.9493,
      "num_input_tokens_seen": 157810355,
      "step": 7393
    },
    {
      "epoch": 0.8890759333854386,
      "flos": 19496868065280.0,
      "grad_norm": 8.762084028135845,
      "learning_rate": 1.2757718478441094e-07,
      "loss": 0.9912,
      "num_input_tokens_seen": 157830820,
      "step": 7394
    },
    {
      "epoch": 0.8891961762760777,
      "flos": 17425364275200.0,
      "grad_norm": 3.3815736682701427,
      "learning_rate": 1.2730356709427302e-07,
      "loss": 0.9621,
      "num_input_tokens_seen": 157849220,
      "step": 7395
    },
    {
      "epoch": 0.8893164191667168,
      "flos": 29643421900800.0,
      "grad_norm": 2.801082834263042,
      "learning_rate": 1.2703023349069542e-07,
      "loss": 0.7855,
      "num_input_tokens_seen": 157873790,
      "step": 7396
    },
    {
      "epoch": 0.8894366620573558,
      "flos": 23954407157760.0,
      "grad_norm": 3.152036063341359,
      "learning_rate": 1.2675718401514223e-07,
      "loss": 0.8013,
      "num_input_tokens_seen": 157897690,
      "step": 7397
    },
    {
      "epoch": 0.889556904947995,
      "flos": 11971297751040.0,
      "grad_norm": 4.4330927494327526,
      "learning_rate": 1.264844187090346e-07,
      "loss": 0.9335,
      "num_input_tokens_seen": 157914535,
      "step": 7398
    },
    {
      "epoch": 0.889677147838634,
      "flos": 18526347202560.0,
      "grad_norm": 2.739290103280041,
      "learning_rate": 1.262119376137516e-07,
      "loss": 0.9404,
      "num_input_tokens_seen": 157935315,
      "step": 7399
    },
    {
      "epoch": 0.8897973907292731,
      "flos": 18840664104960.0,
      "grad_norm": 2.6889063681760335,
      "learning_rate": 1.2593974077062707e-07,
      "loss": 1.032,
      "num_input_tokens_seen": 157956655,
      "step": 7400
    },
    {
      "epoch": 0.8899176336199123,
      "flos": 18683612958720.0,
      "grad_norm": 3.28605227775572,
      "learning_rate": 1.2566782822095423e-07,
      "loss": 0.8241,
      "num_input_tokens_seen": 157976630,
      "step": 7401
    },
    {
      "epoch": 0.8900378765105513,
      "flos": 14775247011840.0,
      "grad_norm": 3.0848932358430643,
      "learning_rate": 1.2539620000598162e-07,
      "loss": 0.9004,
      "num_input_tokens_seen": 157995685,
      "step": 7402
    },
    {
      "epoch": 0.8901581194011904,
      "flos": 11656919531520.0,
      "grad_norm": 3.4516044479452863,
      "learning_rate": 1.2512485616691492e-07,
      "loss": 0.9846,
      "num_input_tokens_seen": 158012460,
      "step": 7403
    },
    {
      "epoch": 0.8902783622918296,
      "flos": 25084278558720.0,
      "grad_norm": 2.636492503314155,
      "learning_rate": 1.2485379674491681e-07,
      "loss": 1.0005,
      "num_input_tokens_seen": 158038375,
      "step": 7404
    },
    {
      "epoch": 0.8903986051824686,
      "flos": 12180515328000.0,
      "grad_norm": 8.458826310648005,
      "learning_rate": 1.2458302178110657e-07,
      "loss": 0.9661,
      "num_input_tokens_seen": 158056460,
      "step": 7405
    },
    {
      "epoch": 0.8905188480731077,
      "flos": 18133274787840.0,
      "grad_norm": 4.221063954165036,
      "learning_rate": 1.2431253131656118e-07,
      "loss": 1.0125,
      "num_input_tokens_seen": 158075655,
      "step": 7406
    },
    {
      "epoch": 0.8906390909637467,
      "flos": 16610882826240.0,
      "grad_norm": 3.6148596373094115,
      "learning_rate": 1.240423253923133e-07,
      "loss": 0.9539,
      "num_input_tokens_seen": 158094980,
      "step": 7407
    },
    {
      "epoch": 0.8907593338543859,
      "flos": 14960205557760.0,
      "grad_norm": 4.343876459490129,
      "learning_rate": 1.237724040493533e-07,
      "loss": 0.8814,
      "num_input_tokens_seen": 158113325,
      "step": 7408
    },
    {
      "epoch": 0.8908795767450249,
      "flos": 15535446589440.0,
      "grad_norm": 4.273353143448719,
      "learning_rate": 1.2350276732862773e-07,
      "loss": 0.9056,
      "num_input_tokens_seen": 158134070,
      "step": 7409
    },
    {
      "epoch": 0.890999819635664,
      "flos": 47477157212160.0,
      "grad_norm": 0.9114763239946942,
      "learning_rate": 1.2323341527103993e-07,
      "loss": 0.8112,
      "num_input_tokens_seen": 158188990,
      "step": 7410
    },
    {
      "epoch": 0.8911200625263032,
      "flos": 19129342341120.0,
      "grad_norm": 3.972917906359736,
      "learning_rate": 1.2296434791745135e-07,
      "loss": 1.0278,
      "num_input_tokens_seen": 158207160,
      "step": 7411
    },
    {
      "epoch": 0.8912403054169422,
      "flos": 14828639784960.0,
      "grad_norm": 2.8110844758706484,
      "learning_rate": 1.2269556530867875e-07,
      "loss": 0.9541,
      "num_input_tokens_seen": 158225435,
      "step": 7412
    },
    {
      "epoch": 0.8913605483075813,
      "flos": 19234840227840.0,
      "grad_norm": 3.766586750724377,
      "learning_rate": 1.2242706748549614e-07,
      "loss": 1.0124,
      "num_input_tokens_seen": 158243150,
      "step": 7413
    },
    {
      "epoch": 0.8914807911982204,
      "flos": 16794615029760.0,
      "grad_norm": 6.638369906617878,
      "learning_rate": 1.2215885448863473e-07,
      "loss": 1.0019,
      "num_input_tokens_seen": 158263745,
      "step": 7414
    },
    {
      "epoch": 0.8916010340888595,
      "flos": 17399081779200.0,
      "grad_norm": 3.0395093260852772,
      "learning_rate": 1.2189092635878152e-07,
      "loss": 0.9859,
      "num_input_tokens_seen": 158284915,
      "step": 7415
    },
    {
      "epoch": 0.8917212769794985,
      "flos": 15065979371520.0,
      "grad_norm": 2.598065749853911,
      "learning_rate": 1.216232831365822e-07,
      "loss": 0.966,
      "num_input_tokens_seen": 158303580,
      "step": 7416
    },
    {
      "epoch": 0.8918415198701377,
      "flos": 18154069401600.0,
      "grad_norm": 5.601827064555067,
      "learning_rate": 1.2135592486263678e-07,
      "loss": 1.0029,
      "num_input_tokens_seen": 158322550,
      "step": 7417
    },
    {
      "epoch": 0.8919617627607768,
      "flos": 27022806282240.0,
      "grad_norm": 3.3352283018741846,
      "learning_rate": 1.2108885157750415e-07,
      "loss": 0.7935,
      "num_input_tokens_seen": 158344630,
      "step": 7418
    },
    {
      "epoch": 0.8920820056514158,
      "flos": 18657790341120.0,
      "grad_norm": 3.1307274784093524,
      "learning_rate": 1.2082206332169897e-07,
      "loss": 0.9882,
      "num_input_tokens_seen": 158364445,
      "step": 7419
    },
    {
      "epoch": 0.892202248542055,
      "flos": 12311314636800.0,
      "grad_norm": 7.021271402368156,
      "learning_rate": 1.2055556013569225e-07,
      "loss": 0.9218,
      "num_input_tokens_seen": 158379675,
      "step": 7420
    },
    {
      "epoch": 0.892322491432694,
      "flos": 15143784468480.0,
      "grad_norm": 2.5047108135697096,
      "learning_rate": 1.2028934205991315e-07,
      "loss": 1.0085,
      "num_input_tokens_seen": 158398715,
      "step": 7421
    },
    {
      "epoch": 0.8924427343233331,
      "flos": 17088627855360.0,
      "grad_norm": 2.7238751916006954,
      "learning_rate": 1.2002340913474607e-07,
      "loss": 0.9586,
      "num_input_tokens_seen": 158422070,
      "step": 7422
    },
    {
      "epoch": 0.8925629772139723,
      "flos": 21384731627520.0,
      "grad_norm": 5.275079135391578,
      "learning_rate": 1.1975776140053317e-07,
      "loss": 0.9386,
      "num_input_tokens_seen": 158441760,
      "step": 7423
    },
    {
      "epoch": 0.8926832201046113,
      "flos": 16061954949120.0,
      "grad_norm": 4.147426561069861,
      "learning_rate": 1.194923988975729e-07,
      "loss": 0.9087,
      "num_input_tokens_seen": 158461080,
      "step": 7424
    },
    {
      "epoch": 0.8928034629952504,
      "flos": 9375339724800.0,
      "grad_norm": 3.816104123388406,
      "learning_rate": 1.192273216661206e-07,
      "loss": 0.9172,
      "num_input_tokens_seen": 158478890,
      "step": 7425
    },
    {
      "epoch": 0.8929237058858895,
      "flos": 39246313021440.0,
      "grad_norm": 0.7628899202670716,
      "learning_rate": 1.189625297463881e-07,
      "loss": 0.7992,
      "num_input_tokens_seen": 158540300,
      "step": 7426
    },
    {
      "epoch": 0.8930439487765286,
      "flos": 20577700208640.0,
      "grad_norm": 3.2159460502483417,
      "learning_rate": 1.1869802317854394e-07,
      "loss": 0.9832,
      "num_input_tokens_seen": 158563805,
      "step": 7427
    },
    {
      "epoch": 0.8931641916671677,
      "flos": 15930849054720.0,
      "grad_norm": 5.479595481811327,
      "learning_rate": 1.1843380200271425e-07,
      "loss": 0.9134,
      "num_input_tokens_seen": 158582725,
      "step": 7428
    },
    {
      "epoch": 0.8932844345578068,
      "flos": 18390734499840.0,
      "grad_norm": 3.2665196703580306,
      "learning_rate": 1.181698662589805e-07,
      "loss": 0.9941,
      "num_input_tokens_seen": 158602030,
      "step": 7429
    },
    {
      "epoch": 0.8934046774484459,
      "flos": 16294297190400.0,
      "grad_norm": 4.108157829781438,
      "learning_rate": 1.1790621598738249e-07,
      "loss": 0.9497,
      "num_input_tokens_seen": 158620065,
      "step": 7430
    },
    {
      "epoch": 0.8935249203390849,
      "flos": 17398928486400.0,
      "grad_norm": 6.7862847970325815,
      "learning_rate": 1.1764285122791461e-07,
      "loss": 0.9464,
      "num_input_tokens_seen": 158640505,
      "step": 7431
    },
    {
      "epoch": 0.8936451632297241,
      "flos": 11132587929600.0,
      "grad_norm": 3.950858450209436,
      "learning_rate": 1.173797720205294e-07,
      "loss": 0.9604,
      "num_input_tokens_seen": 158658260,
      "step": 7432
    },
    {
      "epoch": 0.8937654061203631,
      "flos": 25055267450880.0,
      "grad_norm": 3.696794982714902,
      "learning_rate": 1.1711697840513602e-07,
      "loss": 0.9108,
      "num_input_tokens_seen": 158677415,
      "step": 7433
    },
    {
      "epoch": 0.8938856490110022,
      "flos": 11394677084160.0,
      "grad_norm": 5.935035595678652,
      "learning_rate": 1.1685447042160012e-07,
      "loss": 0.8944,
      "num_input_tokens_seen": 158695170,
      "step": 7434
    },
    {
      "epoch": 0.8940058919016414,
      "flos": 14698208378880.0,
      "grad_norm": 3.9641127039249207,
      "learning_rate": 1.1659224810974367e-07,
      "loss": 0.9033,
      "num_input_tokens_seen": 158714850,
      "step": 7435
    },
    {
      "epoch": 0.8941261347922804,
      "flos": 17950217072640.0,
      "grad_norm": 2.94079259218522,
      "learning_rate": 1.1633031150934591e-07,
      "loss": 0.8734,
      "num_input_tokens_seen": 158737600,
      "step": 7436
    },
    {
      "epoch": 0.8942463776829195,
      "flos": 13859621191680.0,
      "grad_norm": 4.802913480927326,
      "learning_rate": 1.1606866066014176e-07,
      "loss": 0.9832,
      "num_input_tokens_seen": 158756370,
      "step": 7437
    },
    {
      "epoch": 0.8943666205735585,
      "flos": 15845134049280.0,
      "grad_norm": 6.205916307588821,
      "learning_rate": 1.1580729560182434e-07,
      "loss": 0.9442,
      "num_input_tokens_seen": 158771945,
      "step": 7438
    },
    {
      "epoch": 0.8944868634641977,
      "flos": 13410366074880.0,
      "grad_norm": 3.5443740967958095,
      "learning_rate": 1.1554621637404171e-07,
      "loss": 0.91,
      "num_input_tokens_seen": 158789755,
      "step": 7439
    },
    {
      "epoch": 0.8946071063548368,
      "flos": 10210707763200.0,
      "grad_norm": 6.72214780543868,
      "learning_rate": 1.1528542301639999e-07,
      "loss": 0.7975,
      "num_input_tokens_seen": 158806265,
      "step": 7440
    },
    {
      "epoch": 0.8947273492454758,
      "flos": 14252570972160.0,
      "grad_norm": 4.105602548961487,
      "learning_rate": 1.1502491556846105e-07,
      "loss": 1.0081,
      "num_input_tokens_seen": 158824480,
      "step": 7441
    },
    {
      "epoch": 0.894847592136115,
      "flos": 13150300385280.0,
      "grad_norm": 4.394002789405767,
      "learning_rate": 1.1476469406974331e-07,
      "loss": 1.0025,
      "num_input_tokens_seen": 158839800,
      "step": 7442
    },
    {
      "epoch": 0.894967835026754,
      "flos": 16692213657600.0,
      "grad_norm": 3.017416782448435,
      "learning_rate": 1.1450475855972341e-07,
      "loss": 0.9636,
      "num_input_tokens_seen": 158860310,
      "step": 7443
    },
    {
      "epoch": 0.8950880779173931,
      "flos": 10733169192960.0,
      "grad_norm": 3.790865313907372,
      "learning_rate": 1.1424510907783158e-07,
      "loss": 0.8951,
      "num_input_tokens_seen": 158877310,
      "step": 7444
    },
    {
      "epoch": 0.8952083208080323,
      "flos": 15695747543040.0,
      "grad_norm": 3.7384424762777213,
      "learning_rate": 1.1398574566345787e-07,
      "loss": 1.0162,
      "num_input_tokens_seen": 158897665,
      "step": 7445
    },
    {
      "epoch": 0.8953285636986713,
      "flos": 16530165166080.0,
      "grad_norm": 4.0151069989798325,
      "learning_rate": 1.1372666835594702e-07,
      "loss": 1.0061,
      "num_input_tokens_seen": 158915710,
      "step": 7446
    },
    {
      "epoch": 0.8954488065893104,
      "flos": 11578072043520.0,
      "grad_norm": 5.587193573266142,
      "learning_rate": 1.1346787719460071e-07,
      "loss": 0.9019,
      "num_input_tokens_seen": 158934315,
      "step": 7447
    },
    {
      "epoch": 0.8955690494799495,
      "flos": 12939856465920.0,
      "grad_norm": 3.9181526422369832,
      "learning_rate": 1.1320937221867732e-07,
      "loss": 0.9218,
      "num_input_tokens_seen": 158951615,
      "step": 7448
    },
    {
      "epoch": 0.8956892923705886,
      "flos": 18107145584640.0,
      "grad_norm": 3.5202843202291865,
      "learning_rate": 1.1295115346739192e-07,
      "loss": 0.9905,
      "num_input_tokens_seen": 158971335,
      "step": 7449
    },
    {
      "epoch": 0.8958095352612276,
      "flos": 37665079173120.0,
      "grad_norm": 8.443036558786643,
      "learning_rate": 1.1269322097991629e-07,
      "loss": 0.9196,
      "num_input_tokens_seen": 158994340,
      "step": 7450
    },
    {
      "epoch": 0.8959297781518668,
      "flos": 16482229616640.0,
      "grad_norm": 5.273618522645309,
      "learning_rate": 1.1243557479537846e-07,
      "loss": 0.8818,
      "num_input_tokens_seen": 159013950,
      "step": 7451
    },
    {
      "epoch": 0.8960500210425059,
      "flos": 14431704391680.0,
      "grad_norm": 5.752522122747993,
      "learning_rate": 1.121782149528634e-07,
      "loss": 0.8894,
      "num_input_tokens_seen": 159030770,
      "step": 7452
    },
    {
      "epoch": 0.8961702639331449,
      "flos": 14121495736320.0,
      "grad_norm": 6.061829834697129,
      "learning_rate": 1.1192114149141208e-07,
      "loss": 0.9725,
      "num_input_tokens_seen": 159050125,
      "step": 7453
    },
    {
      "epoch": 0.8962905068237841,
      "flos": 9086876098560.0,
      "grad_norm": 3.4200415032082567,
      "learning_rate": 1.1166435445002197e-07,
      "loss": 0.8385,
      "num_input_tokens_seen": 159067515,
      "step": 7454
    },
    {
      "epoch": 0.8964107497144231,
      "flos": 16664582184960.0,
      "grad_norm": 4.067037073578422,
      "learning_rate": 1.1140785386764818e-07,
      "loss": 0.8837,
      "num_input_tokens_seen": 159085935,
      "step": 7455
    },
    {
      "epoch": 0.8965309926050622,
      "flos": 13832909475840.0,
      "grad_norm": 3.8142466061925733,
      "learning_rate": 1.1115163978320153e-07,
      "loss": 0.8873,
      "num_input_tokens_seen": 159104385,
      "step": 7456
    },
    {
      "epoch": 0.8966512354957014,
      "flos": 20414456033280.0,
      "grad_norm": 4.3133318554867355,
      "learning_rate": 1.1089571223554917e-07,
      "loss": 1.0083,
      "num_input_tokens_seen": 159124990,
      "step": 7457
    },
    {
      "epoch": 0.8967714783863404,
      "flos": 16615175024640.0,
      "grad_norm": 3.912485501686022,
      "learning_rate": 1.1064007126351537e-07,
      "loss": 1.0387,
      "num_input_tokens_seen": 159145425,
      "step": 7458
    },
    {
      "epoch": 0.8968917212769795,
      "flos": 17451462819840.0,
      "grad_norm": 3.341142011696267,
      "learning_rate": 1.1038471690588003e-07,
      "loss": 0.9456,
      "num_input_tokens_seen": 159164290,
      "step": 7459
    },
    {
      "epoch": 0.8970119641676186,
      "flos": 16689607680000.0,
      "grad_norm": 2.8283076780553467,
      "learning_rate": 1.1012964920138145e-07,
      "loss": 0.9839,
      "num_input_tokens_seen": 159183595,
      "step": 7460
    },
    {
      "epoch": 0.8971322070582577,
      "flos": 17454038138880.0,
      "grad_norm": 2.879305393908677,
      "learning_rate": 1.0987486818871205e-07,
      "loss": 0.946,
      "num_input_tokens_seen": 159206905,
      "step": 7461
    },
    {
      "epoch": 0.8972524499488967,
      "flos": 15483740037120.0,
      "grad_norm": 7.244109029487582,
      "learning_rate": 1.0962037390652245e-07,
      "loss": 0.9026,
      "num_input_tokens_seen": 159225645,
      "step": 7462
    },
    {
      "epoch": 0.8973726928395359,
      "flos": 15432799948800.0,
      "grad_norm": 10.043208044844597,
      "learning_rate": 1.0936616639341911e-07,
      "loss": 0.9096,
      "num_input_tokens_seen": 159245655,
      "step": 7463
    },
    {
      "epoch": 0.897492935730175,
      "flos": 38515645747200.0,
      "grad_norm": 0.7373523014685843,
      "learning_rate": 1.0911224568796473e-07,
      "loss": 0.7547,
      "num_input_tokens_seen": 159303570,
      "step": 7464
    },
    {
      "epoch": 0.897613178620814,
      "flos": 12962643886080.0,
      "grad_norm": 4.240533045964389,
      "learning_rate": 1.0885861182867984e-07,
      "loss": 0.8951,
      "num_input_tokens_seen": 159321395,
      "step": 7465
    },
    {
      "epoch": 0.8977334215114532,
      "flos": 23529625681920.0,
      "grad_norm": 4.6631451444212475,
      "learning_rate": 1.0860526485403942e-07,
      "loss": 0.8994,
      "num_input_tokens_seen": 159342390,
      "step": 7466
    },
    {
      "epoch": 0.8978536644020922,
      "flos": 10739423539200.0,
      "grad_norm": 3.0858523586454862,
      "learning_rate": 1.0835220480247675e-07,
      "loss": 0.968,
      "num_input_tokens_seen": 159360605,
      "step": 7467
    },
    {
      "epoch": 0.8979739072927313,
      "flos": 12757933117440.0,
      "grad_norm": 4.188190460217239,
      "learning_rate": 1.0809943171238067e-07,
      "loss": 1.022,
      "num_input_tokens_seen": 159378250,
      "step": 7468
    },
    {
      "epoch": 0.8980941501833704,
      "flos": 15824431411200.0,
      "grad_norm": 3.1223403293659797,
      "learning_rate": 1.078469456220965e-07,
      "loss": 0.8232,
      "num_input_tokens_seen": 159398125,
      "step": 7469
    },
    {
      "epoch": 0.8982143930740095,
      "flos": 26812699607040.0,
      "grad_norm": 3.214128454085236,
      "learning_rate": 1.0759474656992606e-07,
      "loss": 0.8792,
      "num_input_tokens_seen": 159420615,
      "step": 7470
    },
    {
      "epoch": 0.8983346359646486,
      "flos": 12809976913920.0,
      "grad_norm": 6.251467723170908,
      "learning_rate": 1.0734283459412785e-07,
      "loss": 0.974,
      "num_input_tokens_seen": 159437185,
      "step": 7471
    },
    {
      "epoch": 0.8984548788552876,
      "flos": 14593875517440.0,
      "grad_norm": 2.7575316894038417,
      "learning_rate": 1.0709120973291707e-07,
      "loss": 0.9836,
      "num_input_tokens_seen": 159456685,
      "step": 7472
    },
    {
      "epoch": 0.8985751217459268,
      "flos": 12600575385600.0,
      "grad_norm": 2.9753412259380605,
      "learning_rate": 1.0683987202446475e-07,
      "loss": 0.9534,
      "num_input_tokens_seen": 159474590,
      "step": 7473
    },
    {
      "epoch": 0.8986953646365659,
      "flos": 15354320363520.0,
      "grad_norm": 3.4535836998260376,
      "learning_rate": 1.0658882150689862e-07,
      "loss": 0.8931,
      "num_input_tokens_seen": 159493170,
      "step": 7474
    },
    {
      "epoch": 0.8988156075272049,
      "flos": 9899701985280.0,
      "grad_norm": 3.410490563643974,
      "learning_rate": 1.0633805821830288e-07,
      "loss": 0.9694,
      "num_input_tokens_seen": 159509575,
      "step": 7475
    },
    {
      "epoch": 0.8989358504178441,
      "flos": 20702766366720.0,
      "grad_norm": 6.2677184148953256,
      "learning_rate": 1.0608758219671753e-07,
      "loss": 1.0216,
      "num_input_tokens_seen": 159528335,
      "step": 7476
    },
    {
      "epoch": 0.8990560933084831,
      "flos": 14357792931840.0,
      "grad_norm": 2.9973013183822994,
      "learning_rate": 1.0583739348014065e-07,
      "loss": 0.8936,
      "num_input_tokens_seen": 159549140,
      "step": 7477
    },
    {
      "epoch": 0.8991763361991222,
      "flos": 18159771893760.0,
      "grad_norm": 6.258991030468283,
      "learning_rate": 1.0558749210652518e-07,
      "loss": 1.043,
      "num_input_tokens_seen": 159568790,
      "step": 7478
    },
    {
      "epoch": 0.8992965790897613,
      "flos": 17871216291840.0,
      "grad_norm": 2.4805032562499,
      "learning_rate": 1.053378781137808e-07,
      "loss": 1.0421,
      "num_input_tokens_seen": 159589430,
      "step": 7479
    },
    {
      "epoch": 0.8994168219804004,
      "flos": 11368333271040.0,
      "grad_norm": 3.8181525679289483,
      "learning_rate": 1.0508855153977392e-07,
      "loss": 0.9724,
      "num_input_tokens_seen": 159605615,
      "step": 7480
    },
    {
      "epoch": 0.8995370648710395,
      "flos": 17660128542720.0,
      "grad_norm": 4.430421798637549,
      "learning_rate": 1.0483951242232669e-07,
      "loss": 0.8446,
      "num_input_tokens_seen": 159625810,
      "step": 7481
    },
    {
      "epoch": 0.8996573077616786,
      "flos": 45185973166080.0,
      "grad_norm": 1.0113356170435723,
      "learning_rate": 1.0459076079921936e-07,
      "loss": 0.7985,
      "num_input_tokens_seen": 159678190,
      "step": 7482
    },
    {
      "epoch": 0.8997775506523177,
      "flos": 12912163676160.0,
      "grad_norm": 4.226396878253362,
      "learning_rate": 1.0434229670818618e-07,
      "loss": 1.0345,
      "num_input_tokens_seen": 159694585,
      "step": 7483
    },
    {
      "epoch": 0.8998977935429567,
      "flos": 17186215833600.0,
      "grad_norm": 3.457640030701458,
      "learning_rate": 1.0409412018691944e-07,
      "loss": 0.994,
      "num_input_tokens_seen": 159714770,
      "step": 7484
    },
    {
      "epoch": 0.9000180364335959,
      "flos": 14747125002240.0,
      "grad_norm": 2.9482512447070217,
      "learning_rate": 1.0384623127306724e-07,
      "loss": 0.9456,
      "num_input_tokens_seen": 159731835,
      "step": 7485
    },
    {
      "epoch": 0.900138279324235,
      "flos": 13620963287040.0,
      "grad_norm": 2.8700153427511417,
      "learning_rate": 1.0359863000423397e-07,
      "loss": 0.9798,
      "num_input_tokens_seen": 159749690,
      "step": 7486
    },
    {
      "epoch": 0.900258522214874,
      "flos": 20467051683840.0,
      "grad_norm": 2.6822088630018404,
      "learning_rate": 1.0335131641798112e-07,
      "loss": 0.9057,
      "num_input_tokens_seen": 159771370,
      "step": 7487
    },
    {
      "epoch": 0.9003787651055132,
      "flos": 41708129955840.0,
      "grad_norm": 0.9773412422373652,
      "learning_rate": 1.0310429055182512e-07,
      "loss": 0.8163,
      "num_input_tokens_seen": 159825410,
      "step": 7488
    },
    {
      "epoch": 0.9004990079961522,
      "flos": 18185410560000.0,
      "grad_norm": 2.2187987846061716,
      "learning_rate": 1.0285755244324024e-07,
      "loss": 0.9256,
      "num_input_tokens_seen": 159845875,
      "step": 7489
    },
    {
      "epoch": 0.9006192508867913,
      "flos": 16589597675520.0,
      "grad_norm": 2.3829269631831425,
      "learning_rate": 1.0261110212965629e-07,
      "loss": 0.8833,
      "num_input_tokens_seen": 159867390,
      "step": 7490
    },
    {
      "epoch": 0.9007394937774305,
      "flos": 12783663759360.0,
      "grad_norm": 3.8515622523928674,
      "learning_rate": 1.023649396484596e-07,
      "loss": 0.9704,
      "num_input_tokens_seen": 159886165,
      "step": 7491
    },
    {
      "epoch": 0.9008597366680695,
      "flos": 30770779299840.0,
      "grad_norm": 4.051299633085785,
      "learning_rate": 1.0211906503699275e-07,
      "loss": 0.8655,
      "num_input_tokens_seen": 159908860,
      "step": 7492
    },
    {
      "epoch": 0.9009799795587086,
      "flos": 10554863554560.0,
      "grad_norm": 6.115036314562487,
      "learning_rate": 1.0187347833255455e-07,
      "loss": 1.0116,
      "num_input_tokens_seen": 159924485,
      "step": 7493
    },
    {
      "epoch": 0.9011002224493477,
      "flos": 15327302062080.0,
      "grad_norm": 2.345886579036873,
      "learning_rate": 1.0162817957240056e-07,
      "loss": 0.988,
      "num_input_tokens_seen": 159944100,
      "step": 7494
    },
    {
      "epoch": 0.9012204653399868,
      "flos": 50969663324160.0,
      "grad_norm": 1.090359364161434,
      "learning_rate": 1.0138316879374253e-07,
      "loss": 0.8739,
      "num_input_tokens_seen": 160013110,
      "step": 7495
    },
    {
      "epoch": 0.9013407082306258,
      "flos": 11026139627520.0,
      "grad_norm": 8.325227788151468,
      "learning_rate": 1.0113844603374833e-07,
      "loss": 0.9298,
      "num_input_tokens_seen": 160029355,
      "step": 7496
    },
    {
      "epoch": 0.901460951121265,
      "flos": 10634078945280.0,
      "grad_norm": 5.176026121860724,
      "learning_rate": 1.0089401132954178e-07,
      "loss": 0.9152,
      "num_input_tokens_seen": 160047055,
      "step": 7497
    },
    {
      "epoch": 0.9015811940119041,
      "flos": 15799743160320.0,
      "grad_norm": 2.7333069377493056,
      "learning_rate": 1.006498647182037e-07,
      "loss": 0.9082,
      "num_input_tokens_seen": 160065430,
      "step": 7498
    },
    {
      "epoch": 0.9017014369025431,
      "flos": 17766822113280.0,
      "grad_norm": 4.159291008173608,
      "learning_rate": 1.004060062367713e-07,
      "loss": 0.9184,
      "num_input_tokens_seen": 160086245,
      "step": 7499
    },
    {
      "epoch": 0.9018216797931822,
      "flos": 12836872581120.0,
      "grad_norm": 10.826645611083254,
      "learning_rate": 1.0016243592223728e-07,
      "loss": 0.8812,
      "num_input_tokens_seen": 160106365,
      "step": 7500
    },
    {
      "epoch": 0.9019419226838213,
      "flos": 26602592931840.0,
      "grad_norm": 3.7923182392256423,
      "learning_rate": 9.991915381155114e-08,
      "loss": 0.8516,
      "num_input_tokens_seen": 160129065,
      "step": 7501
    },
    {
      "epoch": 0.9020621655744604,
      "flos": 16665256673280.0,
      "grad_norm": 4.427939418223231,
      "learning_rate": 9.967615994161871e-08,
      "loss": 0.94,
      "num_input_tokens_seen": 160148445,
      "step": 7502
    },
    {
      "epoch": 0.9021824084650995,
      "flos": 16245809786880.0,
      "grad_norm": 4.116082928206744,
      "learning_rate": 9.943345434930161e-08,
      "loss": 0.9698,
      "num_input_tokens_seen": 160168415,
      "step": 7503
    },
    {
      "epoch": 0.9023026513557386,
      "flos": 15721232916480.0,
      "grad_norm": 4.221721817501731,
      "learning_rate": 9.919103707141885e-08,
      "loss": 0.8719,
      "num_input_tokens_seen": 160187015,
      "step": 7504
    },
    {
      "epoch": 0.9024228942463777,
      "flos": 17208880619520.0,
      "grad_norm": 7.732140420569096,
      "learning_rate": 9.89489081447441e-08,
      "loss": 0.9576,
      "num_input_tokens_seen": 160203935,
      "step": 7505
    },
    {
      "epoch": 0.9025431371370167,
      "flos": 17791295754240.0,
      "grad_norm": 4.0983795313165965,
      "learning_rate": 9.870706760600844e-08,
      "loss": 1.0135,
      "num_input_tokens_seen": 160223605,
      "step": 7506
    },
    {
      "epoch": 0.9026633800276559,
      "flos": 13439561134080.0,
      "grad_norm": 3.865029829337031,
      "learning_rate": 9.846551549189918e-08,
      "loss": 0.9099,
      "num_input_tokens_seen": 160242930,
      "step": 7507
    },
    {
      "epoch": 0.902783622918295,
      "flos": 23115298775040.0,
      "grad_norm": 3.7272368966471703,
      "learning_rate": 9.822425183905902e-08,
      "loss": 0.8609,
      "num_input_tokens_seen": 160263175,
      "step": 7508
    },
    {
      "epoch": 0.902903865808934,
      "flos": 45615484170240.0,
      "grad_norm": 0.9254746070069787,
      "learning_rate": 9.798327668408823e-08,
      "loss": 0.9578,
      "num_input_tokens_seen": 160324530,
      "step": 7509
    },
    {
      "epoch": 0.9030241086995732,
      "flos": 16925812899840.0,
      "grad_norm": 3.5498080456913277,
      "learning_rate": 9.774259006354158e-08,
      "loss": 0.8751,
      "num_input_tokens_seen": 160344320,
      "step": 7510
    },
    {
      "epoch": 0.9031443515902122,
      "flos": 18788712284160.0,
      "grad_norm": 3.135765149256089,
      "learning_rate": 9.750219201393184e-08,
      "loss": 0.9481,
      "num_input_tokens_seen": 160364005,
      "step": 7511
    },
    {
      "epoch": 0.9032645944808513,
      "flos": 17742317813760.0,
      "grad_norm": 2.643634052304222,
      "learning_rate": 9.726208257172697e-08,
      "loss": 0.9744,
      "num_input_tokens_seen": 160385420,
      "step": 7512
    },
    {
      "epoch": 0.9033848373714904,
      "flos": 15039359631360.0,
      "grad_norm": 4.8137459071084265,
      "learning_rate": 9.702226177335115e-08,
      "loss": 0.9439,
      "num_input_tokens_seen": 160403635,
      "step": 7513
    },
    {
      "epoch": 0.9035050802621295,
      "flos": 18709987430400.0,
      "grad_norm": 2.534637915429974,
      "learning_rate": 9.67827296551853e-08,
      "loss": 0.9213,
      "num_input_tokens_seen": 160424640,
      "step": 7514
    },
    {
      "epoch": 0.9036253231527686,
      "flos": 17213540720640.0,
      "grad_norm": 3.2225082825879063,
      "learning_rate": 9.65434862535659e-08,
      "loss": 0.8582,
      "num_input_tokens_seen": 160443730,
      "step": 7515
    },
    {
      "epoch": 0.9037455660434077,
      "flos": 12806359203840.0,
      "grad_norm": 4.595971848218205,
      "learning_rate": 9.630453160478635e-08,
      "loss": 0.8448,
      "num_input_tokens_seen": 160458805,
      "step": 7516
    },
    {
      "epoch": 0.9038658089340468,
      "flos": 17110526177280.0,
      "grad_norm": 2.5295461877910457,
      "learning_rate": 9.60658657450959e-08,
      "loss": 0.9997,
      "num_input_tokens_seen": 160478825,
      "step": 7517
    },
    {
      "epoch": 0.9039860518246858,
      "flos": 15510267801600.0,
      "grad_norm": 2.3290109676759645,
      "learning_rate": 9.582748871069979e-08,
      "loss": 0.9818,
      "num_input_tokens_seen": 160497985,
      "step": 7518
    },
    {
      "epoch": 0.904106294715325,
      "flos": 18946100674560.0,
      "grad_norm": 3.6108618642098156,
      "learning_rate": 9.558940053775954e-08,
      "loss": 1.0211,
      "num_input_tokens_seen": 160516345,
      "step": 7519
    },
    {
      "epoch": 0.904226537605964,
      "flos": 12600789995520.0,
      "grad_norm": 3.2872706808312477,
      "learning_rate": 9.535160126239294e-08,
      "loss": 0.8768,
      "num_input_tokens_seen": 160532690,
      "step": 7520
    },
    {
      "epoch": 0.9043467804966031,
      "flos": 17634888437760.0,
      "grad_norm": 3.2285716011008736,
      "learning_rate": 9.511409092067424e-08,
      "loss": 0.9034,
      "num_input_tokens_seen": 160552765,
      "step": 7521
    },
    {
      "epoch": 0.9044670233872423,
      "flos": 16082994831360.0,
      "grad_norm": 2.6615158841255555,
      "learning_rate": 9.487686954863327e-08,
      "loss": 0.8709,
      "num_input_tokens_seen": 160572205,
      "step": 7522
    },
    {
      "epoch": 0.9045872662778813,
      "flos": 16902810869760.0,
      "grad_norm": 5.187031920123575,
      "learning_rate": 9.46399371822566e-08,
      "loss": 0.9591,
      "num_input_tokens_seen": 160591700,
      "step": 7523
    },
    {
      "epoch": 0.9047075091685204,
      "flos": 10735499243520.0,
      "grad_norm": 3.9797867030587533,
      "learning_rate": 9.440329385748657e-08,
      "loss": 0.8993,
      "num_input_tokens_seen": 160608490,
      "step": 7524
    },
    {
      "epoch": 0.9048277520591596,
      "flos": 12757350604800.0,
      "grad_norm": 3.808967454622517,
      "learning_rate": 9.416693961022137e-08,
      "loss": 0.903,
      "num_input_tokens_seen": 160626460,
      "step": 7525
    },
    {
      "epoch": 0.9049479949497986,
      "flos": 15537929932800.0,
      "grad_norm": 4.684723298686263,
      "learning_rate": 9.393087447631654e-08,
      "loss": 0.9658,
      "num_input_tokens_seen": 160644460,
      "step": 7526
    },
    {
      "epoch": 0.9050682378404377,
      "flos": 14747891466240.0,
      "grad_norm": 3.6440140172142326,
      "learning_rate": 9.36950984915823e-08,
      "loss": 0.9135,
      "num_input_tokens_seen": 160662535,
      "step": 7527
    },
    {
      "epoch": 0.9051884807310768,
      "flos": 15327669964800.0,
      "grad_norm": 4.17432528771609,
      "learning_rate": 9.345961169178607e-08,
      "loss": 0.8904,
      "num_input_tokens_seen": 160681940,
      "step": 7528
    },
    {
      "epoch": 0.9053087236217159,
      "flos": 15563384647680.0,
      "grad_norm": 4.402701606185169,
      "learning_rate": 9.322441411265081e-08,
      "loss": 0.9176,
      "num_input_tokens_seen": 160702645,
      "step": 7529
    },
    {
      "epoch": 0.9054289665123549,
      "flos": 12076213125120.0,
      "grad_norm": 3.5557084514274697,
      "learning_rate": 9.298950578985554e-08,
      "loss": 0.9224,
      "num_input_tokens_seen": 160719440,
      "step": 7530
    },
    {
      "epoch": 0.905549209402994,
      "flos": 14750742712320.0,
      "grad_norm": 3.0436055174576153,
      "learning_rate": 9.275488675903665e-08,
      "loss": 0.898,
      "num_input_tokens_seen": 160738105,
      "step": 7531
    },
    {
      "epoch": 0.9056694522936332,
      "flos": 15404064768000.0,
      "grad_norm": 3.7225800353311143,
      "learning_rate": 9.252055705578454e-08,
      "loss": 0.9244,
      "num_input_tokens_seen": 160757325,
      "step": 7532
    },
    {
      "epoch": 0.9057896951842722,
      "flos": 21069127065600.0,
      "grad_norm": 2.9248932126106384,
      "learning_rate": 9.228651671564747e-08,
      "loss": 0.9083,
      "num_input_tokens_seen": 160779075,
      "step": 7533
    },
    {
      "epoch": 0.9059099380749113,
      "flos": 19864179179520.0,
      "grad_norm": 2.1957068038282537,
      "learning_rate": 9.205276577412901e-08,
      "loss": 0.9789,
      "num_input_tokens_seen": 160801575,
      "step": 7534
    },
    {
      "epoch": 0.9060301809655504,
      "flos": 12574384865280.0,
      "grad_norm": 4.487631861202486,
      "learning_rate": 9.181930426668905e-08,
      "loss": 0.9549,
      "num_input_tokens_seen": 160818090,
      "step": 7535
    },
    {
      "epoch": 0.9061504238561895,
      "flos": 22643409530880.0,
      "grad_norm": 6.617940055993013,
      "learning_rate": 9.158613222874346e-08,
      "loss": 0.8868,
      "num_input_tokens_seen": 160839435,
      "step": 7536
    },
    {
      "epoch": 0.9062706667468285,
      "flos": 14226748354560.0,
      "grad_norm": 3.7844469411675856,
      "learning_rate": 9.135324969566394e-08,
      "loss": 1.0111,
      "num_input_tokens_seen": 160858655,
      "step": 7537
    },
    {
      "epoch": 0.9063909096374677,
      "flos": 13068387041280.0,
      "grad_norm": 3.023323628649405,
      "learning_rate": 9.112065670277913e-08,
      "loss": 0.942,
      "num_input_tokens_seen": 160874740,
      "step": 7538
    },
    {
      "epoch": 0.9065111525281068,
      "flos": 23928369930240.0,
      "grad_norm": 3.7278222727753403,
      "learning_rate": 9.088835328537303e-08,
      "loss": 0.9167,
      "num_input_tokens_seen": 160896050,
      "step": 7539
    },
    {
      "epoch": 0.9066313954187458,
      "flos": 16612569047040.0,
      "grad_norm": 2.9016850212987606,
      "learning_rate": 9.065633947868568e-08,
      "loss": 0.8956,
      "num_input_tokens_seen": 160915375,
      "step": 7540
    },
    {
      "epoch": 0.906751638309385,
      "flos": 18683551641600.0,
      "grad_norm": 7.664884221844277,
      "learning_rate": 9.042461531791379e-08,
      "loss": 0.9832,
      "num_input_tokens_seen": 160933515,
      "step": 7541
    },
    {
      "epoch": 0.906871881200024,
      "flos": 11708656742400.0,
      "grad_norm": 3.691239223031738,
      "learning_rate": 9.019318083820903e-08,
      "loss": 0.956,
      "num_input_tokens_seen": 160951815,
      "step": 7542
    },
    {
      "epoch": 0.9069921240906631,
      "flos": 17501329858560.0,
      "grad_norm": 3.4891833327570856,
      "learning_rate": 8.996203607468045e-08,
      "loss": 1.0376,
      "num_input_tokens_seen": 160970535,
      "step": 7543
    },
    {
      "epoch": 0.9071123669813023,
      "flos": 18054887178240.0,
      "grad_norm": 2.3975919046776357,
      "learning_rate": 8.973118106239241e-08,
      "loss": 0.9486,
      "num_input_tokens_seen": 160992860,
      "step": 7544
    },
    {
      "epoch": 0.9072326098719413,
      "flos": 19026021212160.0,
      "grad_norm": 3.5921298148097756,
      "learning_rate": 8.95006158363656e-08,
      "loss": 1.1301,
      "num_input_tokens_seen": 161012765,
      "step": 7545
    },
    {
      "epoch": 0.9073528527625804,
      "flos": 16979788185600.0,
      "grad_norm": 3.092506235743953,
      "learning_rate": 8.9270340431576e-08,
      "loss": 0.9583,
      "num_input_tokens_seen": 161031575,
      "step": 7546
    },
    {
      "epoch": 0.9074730956532195,
      "flos": 27022070476800.0,
      "grad_norm": 3.8719541823300903,
      "learning_rate": 8.904035488295658e-08,
      "loss": 0.9176,
      "num_input_tokens_seen": 161050795,
      "step": 7547
    },
    {
      "epoch": 0.9075933385438586,
      "flos": 46661296128000.0,
      "grad_norm": 0.6926143243046445,
      "learning_rate": 8.881065922539632e-08,
      "loss": 0.7563,
      "num_input_tokens_seen": 161110955,
      "step": 7548
    },
    {
      "epoch": 0.9077135814344977,
      "flos": 14142075740160.0,
      "grad_norm": 4.39439669932907,
      "learning_rate": 8.85812534937389e-08,
      "loss": 0.933,
      "num_input_tokens_seen": 161128775,
      "step": 7549
    },
    {
      "epoch": 0.9078338243251368,
      "flos": 12416689889280.0,
      "grad_norm": 6.459060562371952,
      "learning_rate": 8.835213772278583e-08,
      "loss": 0.8518,
      "num_input_tokens_seen": 161145350,
      "step": 7550
    },
    {
      "epoch": 0.9079540672157759,
      "flos": 20518880870400.0,
      "grad_norm": 3.6794513861514404,
      "learning_rate": 8.812331194729373e-08,
      "loss": 0.9789,
      "num_input_tokens_seen": 161164715,
      "step": 7551
    },
    {
      "epoch": 0.9080743101064149,
      "flos": 16717484421120.0,
      "grad_norm": 3.391707146580327,
      "learning_rate": 8.789477620197461e-08,
      "loss": 0.9074,
      "num_input_tokens_seen": 161183960,
      "step": 7552
    },
    {
      "epoch": 0.9081945529970541,
      "flos": 16187910205440.0,
      "grad_norm": 4.337376151391773,
      "learning_rate": 8.766653052149831e-08,
      "loss": 0.9858,
      "num_input_tokens_seen": 161198865,
      "step": 7553
    },
    {
      "epoch": 0.9083147958876931,
      "flos": 13381845504000.0,
      "grad_norm": 5.864588175670551,
      "learning_rate": 8.743857494048823e-08,
      "loss": 0.941,
      "num_input_tokens_seen": 161215400,
      "step": 7554
    },
    {
      "epoch": 0.9084350387783322,
      "flos": 13407361536000.0,
      "grad_norm": 3.731468058848493,
      "learning_rate": 8.721090949352605e-08,
      "loss": 0.8346,
      "num_input_tokens_seen": 161231360,
      "step": 7555
    },
    {
      "epoch": 0.9085552816689714,
      "flos": 14620188672000.0,
      "grad_norm": 3.20123309576717,
      "learning_rate": 8.698353421514793e-08,
      "loss": 0.9132,
      "num_input_tokens_seen": 161249455,
      "step": 7556
    },
    {
      "epoch": 0.9086755245596104,
      "flos": 12836014141440.0,
      "grad_norm": 14.452620345919327,
      "learning_rate": 8.67564491398467e-08,
      "loss": 0.9897,
      "num_input_tokens_seen": 161266180,
      "step": 7557
    },
    {
      "epoch": 0.9087957674502495,
      "flos": 13565976268800.0,
      "grad_norm": 4.408151851925451,
      "learning_rate": 8.652965430207104e-08,
      "loss": 0.9369,
      "num_input_tokens_seen": 161283805,
      "step": 7558
    },
    {
      "epoch": 0.9089160103408886,
      "flos": 12833254871040.0,
      "grad_norm": 3.6478887958140906,
      "learning_rate": 8.630314973622521e-08,
      "loss": 0.8595,
      "num_input_tokens_seen": 161301070,
      "step": 7559
    },
    {
      "epoch": 0.9090362532315277,
      "flos": 23797202718720.0,
      "grad_norm": 5.813571297969284,
      "learning_rate": 8.607693547666995e-08,
      "loss": 0.9023,
      "num_input_tokens_seen": 161323330,
      "step": 7560
    },
    {
      "epoch": 0.9091564961221668,
      "flos": 51193574461440.0,
      "grad_norm": 0.9356114120363115,
      "learning_rate": 8.585101155772201e-08,
      "loss": 0.8168,
      "num_input_tokens_seen": 161385170,
      "step": 7561
    },
    {
      "epoch": 0.9092767390128058,
      "flos": 17578276515840.0,
      "grad_norm": 5.266562358611966,
      "learning_rate": 8.562537801365377e-08,
      "loss": 0.8806,
      "num_input_tokens_seen": 161404625,
      "step": 7562
    },
    {
      "epoch": 0.909396981903445,
      "flos": 16769650851840.0,
      "grad_norm": 2.7853764985944216,
      "learning_rate": 8.540003487869362e-08,
      "loss": 0.8781,
      "num_input_tokens_seen": 161426015,
      "step": 7563
    },
    {
      "epoch": 0.909517224794084,
      "flos": 16638636933120.0,
      "grad_norm": 4.474669264180729,
      "learning_rate": 8.517498218702557e-08,
      "loss": 0.9837,
      "num_input_tokens_seen": 161443665,
      "step": 7564
    },
    {
      "epoch": 0.9096374676847231,
      "flos": 13623415971840.0,
      "grad_norm": 4.078955087184889,
      "learning_rate": 8.49502199727905e-08,
      "loss": 0.8944,
      "num_input_tokens_seen": 161461410,
      "step": 7565
    },
    {
      "epoch": 0.9097577105753623,
      "flos": 23745281556480.0,
      "grad_norm": 3.3122260502230034,
      "learning_rate": 8.472574827008428e-08,
      "loss": 0.83,
      "num_input_tokens_seen": 161482015,
      "step": 7566
    },
    {
      "epoch": 0.9098779534660013,
      "flos": 15562005012480.0,
      "grad_norm": 2.4584460044084344,
      "learning_rate": 8.450156711295942e-08,
      "loss": 1.0316,
      "num_input_tokens_seen": 161501905,
      "step": 7567
    },
    {
      "epoch": 0.9099981963566404,
      "flos": 18309219717120.0,
      "grad_norm": 3.778774491513729,
      "learning_rate": 8.427767653542383e-08,
      "loss": 1.0525,
      "num_input_tokens_seen": 161516795,
      "step": 7568
    },
    {
      "epoch": 0.9101184392472795,
      "flos": 14960757411840.0,
      "grad_norm": 3.205793357886125,
      "learning_rate": 8.405407657144125e-08,
      "loss": 0.8938,
      "num_input_tokens_seen": 161535675,
      "step": 7569
    },
    {
      "epoch": 0.9102386821379186,
      "flos": 17607532892160.0,
      "grad_norm": 3.4388726008256647,
      "learning_rate": 8.383076725493232e-08,
      "loss": 0.9074,
      "num_input_tokens_seen": 161552715,
      "step": 7570
    },
    {
      "epoch": 0.9103589250285576,
      "flos": 16033679646720.0,
      "grad_norm": 5.834745436004772,
      "learning_rate": 8.360774861977216e-08,
      "loss": 0.8738,
      "num_input_tokens_seen": 161571555,
      "step": 7571
    },
    {
      "epoch": 0.9104791679191968,
      "flos": 18053814128640.0,
      "grad_norm": 4.9011400181154645,
      "learning_rate": 8.338502069979281e-08,
      "loss": 0.9351,
      "num_input_tokens_seen": 161591585,
      "step": 7572
    },
    {
      "epoch": 0.9105994108098359,
      "flos": 10186418073600.0,
      "grad_norm": 4.6903940040998995,
      "learning_rate": 8.316258352878214e-08,
      "loss": 0.9755,
      "num_input_tokens_seen": 161607725,
      "step": 7573
    },
    {
      "epoch": 0.9107196537004749,
      "flos": 19019613573120.0,
      "grad_norm": 3.751170293666034,
      "learning_rate": 8.294043714048338e-08,
      "loss": 0.898,
      "num_input_tokens_seen": 161626525,
      "step": 7574
    },
    {
      "epoch": 0.9108398965911141,
      "flos": 47639481630720.0,
      "grad_norm": 0.7870111143813018,
      "learning_rate": 8.271858156859624e-08,
      "loss": 0.8345,
      "num_input_tokens_seen": 161691615,
      "step": 7575
    },
    {
      "epoch": 0.9109601394817531,
      "flos": 18080985722880.0,
      "grad_norm": 2.837458095157083,
      "learning_rate": 8.249701684677557e-08,
      "loss": 0.9275,
      "num_input_tokens_seen": 161712660,
      "step": 7576
    },
    {
      "epoch": 0.9110803823723922,
      "flos": 16166962298880.0,
      "grad_norm": 3.2563154041948517,
      "learning_rate": 8.227574300863294e-08,
      "loss": 1.0007,
      "num_input_tokens_seen": 161732550,
      "step": 7577
    },
    {
      "epoch": 0.9112006252630314,
      "flos": 34652924067840.0,
      "grad_norm": 2.444048179913257,
      "learning_rate": 8.205476008773548e-08,
      "loss": 0.8944,
      "num_input_tokens_seen": 161756270,
      "step": 7578
    },
    {
      "epoch": 0.9113208681536704,
      "flos": 21386080604160.0,
      "grad_norm": 4.204688026204213,
      "learning_rate": 8.183406811760596e-08,
      "loss": 1.0178,
      "num_input_tokens_seen": 161775720,
      "step": 7579
    },
    {
      "epoch": 0.9114411110443095,
      "flos": 18213348618240.0,
      "grad_norm": 3.110815433423789,
      "learning_rate": 8.161366713172313e-08,
      "loss": 0.9373,
      "num_input_tokens_seen": 161797830,
      "step": 7580
    },
    {
      "epoch": 0.9115613539349486,
      "flos": 13041246105600.0,
      "grad_norm": 5.299065511119644,
      "learning_rate": 8.139355716352137e-08,
      "loss": 1.0419,
      "num_input_tokens_seen": 161812390,
      "step": 7581
    },
    {
      "epoch": 0.9116815968255877,
      "flos": 15432738631680.0,
      "grad_norm": 3.6054629308254773,
      "learning_rate": 8.117373824639196e-08,
      "loss": 0.8882,
      "num_input_tokens_seen": 161832375,
      "step": 7582
    },
    {
      "epoch": 0.9118018397162267,
      "flos": 46799054929920.0,
      "grad_norm": 0.7398405209947191,
      "learning_rate": 8.095421041368067e-08,
      "loss": 0.8012,
      "num_input_tokens_seen": 161891510,
      "step": 7583
    },
    {
      "epoch": 0.9119220826068659,
      "flos": 14855351500800.0,
      "grad_norm": 4.577967722407752,
      "learning_rate": 8.073497369868999e-08,
      "loss": 0.8934,
      "num_input_tokens_seen": 161909690,
      "step": 7584
    },
    {
      "epoch": 0.912042325497505,
      "flos": 20204962529280.0,
      "grad_norm": 2.9589540788465536,
      "learning_rate": 8.051602813467772e-08,
      "loss": 0.9418,
      "num_input_tokens_seen": 161931265,
      "step": 7585
    },
    {
      "epoch": 0.912162568388144,
      "flos": 12443462922240.0,
      "grad_norm": 4.587367807921158,
      "learning_rate": 8.029737375485756e-08,
      "loss": 0.9021,
      "num_input_tokens_seen": 161950215,
      "step": 7586
    },
    {
      "epoch": 0.9122828112787832,
      "flos": 14068348231680.0,
      "grad_norm": 7.8551144892303135,
      "learning_rate": 8.007901059239986e-08,
      "loss": 0.9257,
      "num_input_tokens_seen": 161969215,
      "step": 7587
    },
    {
      "epoch": 0.9124030541694222,
      "flos": 14776994549760.0,
      "grad_norm": 2.1914124514242133,
      "learning_rate": 7.986093868042964e-08,
      "loss": 0.996,
      "num_input_tokens_seen": 161989180,
      "step": 7588
    },
    {
      "epoch": 0.9125232970600613,
      "flos": 17923781283840.0,
      "grad_norm": 3.394621853338107,
      "learning_rate": 7.964315805202826e-08,
      "loss": 0.8718,
      "num_input_tokens_seen": 162009480,
      "step": 7589
    },
    {
      "epoch": 0.9126435399507005,
      "flos": 13990665768960.0,
      "grad_norm": 2.5568467094294136,
      "learning_rate": 7.942566874023304e-08,
      "loss": 0.9247,
      "num_input_tokens_seen": 162028385,
      "step": 7590
    },
    {
      "epoch": 0.9127637828413395,
      "flos": 13885750394880.0,
      "grad_norm": 4.747127065690113,
      "learning_rate": 7.920847077803649e-08,
      "loss": 0.8837,
      "num_input_tokens_seen": 162045895,
      "step": 7591
    },
    {
      "epoch": 0.9128840257319786,
      "flos": 14357976883200.0,
      "grad_norm": 3.4328037433746985,
      "learning_rate": 7.899156419838826e-08,
      "loss": 0.9963,
      "num_input_tokens_seen": 162064585,
      "step": 7592
    },
    {
      "epoch": 0.9130042686226177,
      "flos": 17683283865600.0,
      "grad_norm": 3.088593317964807,
      "learning_rate": 7.87749490341918e-08,
      "loss": 0.8518,
      "num_input_tokens_seen": 162084580,
      "step": 7593
    },
    {
      "epoch": 0.9131245115132568,
      "flos": 16765603921920.0,
      "grad_norm": 3.4448536677625894,
      "learning_rate": 7.855862531830836e-08,
      "loss": 1.0386,
      "num_input_tokens_seen": 162100410,
      "step": 7594
    },
    {
      "epoch": 0.9132447544038959,
      "flos": 14142106398720.0,
      "grad_norm": 2.9393507625968383,
      "learning_rate": 7.834259308355373e-08,
      "loss": 0.919,
      "num_input_tokens_seen": 162118895,
      "step": 7595
    },
    {
      "epoch": 0.9133649972945349,
      "flos": 15616440176640.0,
      "grad_norm": 4.6734898050608615,
      "learning_rate": 7.812685236269989e-08,
      "loss": 0.9344,
      "num_input_tokens_seen": 162137275,
      "step": 7596
    },
    {
      "epoch": 0.9134852401851741,
      "flos": 51023117414400.0,
      "grad_norm": 0.8387679195621741,
      "learning_rate": 7.791140318847445e-08,
      "loss": 0.8077,
      "num_input_tokens_seen": 162195130,
      "step": 7597
    },
    {
      "epoch": 0.9136054830758131,
      "flos": 16799244472320.0,
      "grad_norm": 3.5673921690953123,
      "learning_rate": 7.769624559356081e-08,
      "loss": 0.9938,
      "num_input_tokens_seen": 162218245,
      "step": 7598
    },
    {
      "epoch": 0.9137257259664522,
      "flos": 16663325184000.0,
      "grad_norm": 3.9977458324314035,
      "learning_rate": 7.748137961059842e-08,
      "loss": 0.9059,
      "num_input_tokens_seen": 162231945,
      "step": 7599
    },
    {
      "epoch": 0.9138459688570914,
      "flos": 13564320706560.0,
      "grad_norm": 5.707284575785923,
      "learning_rate": 7.726680527218211e-08,
      "loss": 0.8561,
      "num_input_tokens_seen": 162248705,
      "step": 7600
    },
    {
      "epoch": 0.9139662117477304,
      "flos": 33080021237760.0,
      "grad_norm": 2.860463086497336,
      "learning_rate": 7.70525226108627e-08,
      "loss": 0.939,
      "num_input_tokens_seen": 162272095,
      "step": 7601
    },
    {
      "epoch": 0.9140864546383695,
      "flos": 15825841704960.0,
      "grad_norm": 2.880933361073252,
      "learning_rate": 7.683853165914666e-08,
      "loss": 0.9961,
      "num_input_tokens_seen": 162289585,
      "step": 7602
    },
    {
      "epoch": 0.9142066975290086,
      "flos": 12416996474880.0,
      "grad_norm": 2.877289565252265,
      "learning_rate": 7.662483244949602e-08,
      "loss": 0.9583,
      "num_input_tokens_seen": 162306565,
      "step": 7603
    },
    {
      "epoch": 0.9143269404196477,
      "flos": 12547857100800.0,
      "grad_norm": 5.986397264588498,
      "learning_rate": 7.641142501432951e-08,
      "loss": 0.9996,
      "num_input_tokens_seen": 162322480,
      "step": 7604
    },
    {
      "epoch": 0.9144471833102867,
      "flos": 23766750658560.0,
      "grad_norm": 2.98744722583078,
      "learning_rate": 7.619830938602013e-08,
      "loss": 0.9333,
      "num_input_tokens_seen": 162343425,
      "step": 7605
    },
    {
      "epoch": 0.9145674262009259,
      "flos": 14960358850560.0,
      "grad_norm": 5.878466293753574,
      "learning_rate": 7.598548559689777e-08,
      "loss": 1.0046,
      "num_input_tokens_seen": 162361545,
      "step": 7606
    },
    {
      "epoch": 0.914687669091565,
      "flos": 11892542238720.0,
      "grad_norm": 3.5578138080874298,
      "learning_rate": 7.577295367924751e-08,
      "loss": 0.9994,
      "num_input_tokens_seen": 162377665,
      "step": 7607
    },
    {
      "epoch": 0.914807911982204,
      "flos": 18341235363840.0,
      "grad_norm": 4.280501697049352,
      "learning_rate": 7.556071366531002e-08,
      "loss": 1.0272,
      "num_input_tokens_seen": 162398355,
      "step": 7608
    },
    {
      "epoch": 0.9149281548728432,
      "flos": 13622925434880.0,
      "grad_norm": 3.7677588530719883,
      "learning_rate": 7.53487655872822e-08,
      "loss": 0.9712,
      "num_input_tokens_seen": 162417245,
      "step": 7609
    },
    {
      "epoch": 0.9150483977634822,
      "flos": 19129894195200.0,
      "grad_norm": 3.366153142182265,
      "learning_rate": 7.513710947731656e-08,
      "loss": 0.9349,
      "num_input_tokens_seen": 162438175,
      "step": 7610
    },
    {
      "epoch": 0.9151686406541213,
      "flos": 15564151111680.0,
      "grad_norm": 3.4525788164361453,
      "learning_rate": 7.492574536752095e-08,
      "loss": 1.0367,
      "num_input_tokens_seen": 162457885,
      "step": 7611
    },
    {
      "epoch": 0.9152888835447605,
      "flos": 19444701634560.0,
      "grad_norm": 3.013538013978642,
      "learning_rate": 7.471467328995907e-08,
      "loss": 0.9772,
      "num_input_tokens_seen": 162476415,
      "step": 7612
    },
    {
      "epoch": 0.9154091264353995,
      "flos": 9427812741120.0,
      "grad_norm": 4.642454255617704,
      "learning_rate": 7.450389327665018e-08,
      "loss": 0.7872,
      "num_input_tokens_seen": 162493970,
      "step": 7613
    },
    {
      "epoch": 0.9155293693260386,
      "flos": 14330805288960.0,
      "grad_norm": 4.263861546685896,
      "learning_rate": 7.429340535957029e-08,
      "loss": 0.8536,
      "num_input_tokens_seen": 162508885,
      "step": 7614
    },
    {
      "epoch": 0.9156496122166777,
      "flos": 13728668590080.0,
      "grad_norm": 4.1969631932834925,
      "learning_rate": 7.40832095706494e-08,
      "loss": 0.9082,
      "num_input_tokens_seen": 162525300,
      "step": 7615
    },
    {
      "epoch": 0.9157698551073168,
      "flos": 14988143616000.0,
      "grad_norm": 3.617271603622006,
      "learning_rate": 7.387330594177443e-08,
      "loss": 0.9947,
      "num_input_tokens_seen": 162547095,
      "step": 7616
    },
    {
      "epoch": 0.9158900979979558,
      "flos": 17924517089280.0,
      "grad_norm": 5.472963987488828,
      "learning_rate": 7.366369450478749e-08,
      "loss": 0.987,
      "num_input_tokens_seen": 162567925,
      "step": 7617
    },
    {
      "epoch": 0.916010340888595,
      "flos": 21483331338240.0,
      "grad_norm": 3.2393718948952896,
      "learning_rate": 7.345437529148646e-08,
      "loss": 0.8496,
      "num_input_tokens_seen": 162586655,
      "step": 7618
    },
    {
      "epoch": 0.9161305837792341,
      "flos": 12102035742720.0,
      "grad_norm": 3.281714706211353,
      "learning_rate": 7.324534833362483e-08,
      "loss": 0.9281,
      "num_input_tokens_seen": 162603950,
      "step": 7619
    },
    {
      "epoch": 0.9162508266698731,
      "flos": 16271816355840.0,
      "grad_norm": 2.8868021662613716,
      "learning_rate": 7.303661366291192e-08,
      "loss": 0.876,
      "num_input_tokens_seen": 162624340,
      "step": 7620
    },
    {
      "epoch": 0.9163710695605123,
      "flos": 14174060728320.0,
      "grad_norm": 2.727347099850689,
      "learning_rate": 7.28281713110126e-08,
      "loss": 1.0125,
      "num_input_tokens_seen": 162642135,
      "step": 7621
    },
    {
      "epoch": 0.9164913124511513,
      "flos": 16192784916480.0,
      "grad_norm": 3.015108054329839,
      "learning_rate": 7.262002130954759e-08,
      "loss": 0.9629,
      "num_input_tokens_seen": 162660310,
      "step": 7622
    },
    {
      "epoch": 0.9166115553417904,
      "flos": 17714379755520.0,
      "grad_norm": 4.120336034459475,
      "learning_rate": 7.241216369009296e-08,
      "loss": 0.981,
      "num_input_tokens_seen": 162680215,
      "step": 7623
    },
    {
      "epoch": 0.9167317982324296,
      "flos": 18290724495360.0,
      "grad_norm": 3.9673196212445268,
      "learning_rate": 7.220459848418037e-08,
      "loss": 0.8643,
      "num_input_tokens_seen": 162700010,
      "step": 7624
    },
    {
      "epoch": 0.9168520411230686,
      "flos": 11053525831680.0,
      "grad_norm": 3.3386359635404843,
      "learning_rate": 7.199732572329708e-08,
      "loss": 0.9807,
      "num_input_tokens_seen": 162717630,
      "step": 7625
    },
    {
      "epoch": 0.9169722840137077,
      "flos": 21563742412800.0,
      "grad_norm": 4.160987162105648,
      "learning_rate": 7.179034543888684e-08,
      "loss": 0.9399,
      "num_input_tokens_seen": 162736855,
      "step": 7626
    },
    {
      "epoch": 0.9170925269043467,
      "flos": 15982770216960.0,
      "grad_norm": 5.050895294294446,
      "learning_rate": 7.158365766234808e-08,
      "loss": 0.9679,
      "num_input_tokens_seen": 162755425,
      "step": 7627
    },
    {
      "epoch": 0.9172127697949859,
      "flos": 16272000307200.0,
      "grad_norm": 4.199781030668909,
      "learning_rate": 7.137726242503527e-08,
      "loss": 0.9107,
      "num_input_tokens_seen": 162774065,
      "step": 7628
    },
    {
      "epoch": 0.917333012685625,
      "flos": 12360139284480.0,
      "grad_norm": 8.931727539280265,
      "learning_rate": 7.11711597582585e-08,
      "loss": 0.9719,
      "num_input_tokens_seen": 162791145,
      "step": 7629
    },
    {
      "epoch": 0.917453255576264,
      "flos": 10109226147840.0,
      "grad_norm": 2.4607622356674486,
      "learning_rate": 7.096534969328271e-08,
      "loss": 0.9933,
      "num_input_tokens_seen": 162808310,
      "step": 7630
    },
    {
      "epoch": 0.9175734984669032,
      "flos": 14724858777600.0,
      "grad_norm": 2.9927767579253755,
      "learning_rate": 7.075983226132987e-08,
      "loss": 1.0299,
      "num_input_tokens_seen": 162826960,
      "step": 7631
    },
    {
      "epoch": 0.9176937413575422,
      "flos": 10476966481920.0,
      "grad_norm": 7.641072078800706,
      "learning_rate": 7.055460749357656e-08,
      "loss": 0.9752,
      "num_input_tokens_seen": 162842960,
      "step": 7632
    },
    {
      "epoch": 0.9178139842481813,
      "flos": 13095313367040.0,
      "grad_norm": 5.365898588356144,
      "learning_rate": 7.034967542115521e-08,
      "loss": 0.9008,
      "num_input_tokens_seen": 162860945,
      "step": 7633
    },
    {
      "epoch": 0.9179342271388204,
      "flos": 14226227159040.0,
      "grad_norm": 3.4097221945982548,
      "learning_rate": 7.014503607515388e-08,
      "loss": 0.945,
      "num_input_tokens_seen": 162879970,
      "step": 7634
    },
    {
      "epoch": 0.9180544700294595,
      "flos": 17552177971200.0,
      "grad_norm": 4.225200102091019,
      "learning_rate": 6.994068948661592e-08,
      "loss": 0.869,
      "num_input_tokens_seen": 162897845,
      "step": 7635
    },
    {
      "epoch": 0.9181747129200986,
      "flos": 11840253173760.0,
      "grad_norm": 3.324625530524136,
      "learning_rate": 6.973663568654142e-08,
      "loss": 0.9575,
      "num_input_tokens_seen": 162915270,
      "step": 7636
    },
    {
      "epoch": 0.9182949558107377,
      "flos": 17260955074560.0,
      "grad_norm": 4.022020523770102,
      "learning_rate": 6.953287470588386e-08,
      "loss": 0.8449,
      "num_input_tokens_seen": 162932945,
      "step": 7637
    },
    {
      "epoch": 0.9184151987013768,
      "flos": 15687347097600.0,
      "grad_norm": 3.968367070024346,
      "learning_rate": 6.932940657555452e-08,
      "loss": 1.0429,
      "num_input_tokens_seen": 162948795,
      "step": 7638
    },
    {
      "epoch": 0.9185354415920158,
      "flos": 22936931819520.0,
      "grad_norm": 3.4377564768044633,
      "learning_rate": 6.912623132641938e-08,
      "loss": 0.9548,
      "num_input_tokens_seen": 162973605,
      "step": 7639
    },
    {
      "epoch": 0.918655684482655,
      "flos": 14909480079360.0,
      "grad_norm": 3.213137427593329,
      "learning_rate": 6.892334898929952e-08,
      "loss": 0.9509,
      "num_input_tokens_seen": 162993570,
      "step": 7640
    },
    {
      "epoch": 0.918775927373294,
      "flos": 11001819279360.0,
      "grad_norm": 4.054464286290324,
      "learning_rate": 6.872075959497236e-08,
      "loss": 1.0313,
      "num_input_tokens_seen": 163012065,
      "step": 7641
    },
    {
      "epoch": 0.9188961702639331,
      "flos": 21331400171520.0,
      "grad_norm": 3.174228640594208,
      "learning_rate": 6.85184631741702e-08,
      "loss": 1.0161,
      "num_input_tokens_seen": 163032350,
      "step": 7642
    },
    {
      "epoch": 0.9190164131545723,
      "flos": 14695694376960.0,
      "grad_norm": 9.248239710568743,
      "learning_rate": 6.831645975758161e-08,
      "loss": 0.9694,
      "num_input_tokens_seen": 163050010,
      "step": 7643
    },
    {
      "epoch": 0.9191366560452113,
      "flos": 18238251479040.0,
      "grad_norm": 5.181322579053816,
      "learning_rate": 6.811474937585026e-08,
      "loss": 0.8615,
      "num_input_tokens_seen": 163069520,
      "step": 7644
    },
    {
      "epoch": 0.9192568989358504,
      "flos": 15222999859200.0,
      "grad_norm": 7.291346806367971,
      "learning_rate": 6.79133320595755e-08,
      "loss": 0.9753,
      "num_input_tokens_seen": 163089160,
      "step": 7645
    },
    {
      "epoch": 0.9193771418264896,
      "flos": 16481156567040.0,
      "grad_norm": 6.89467123175884,
      "learning_rate": 6.771220783931198e-08,
      "loss": 0.9289,
      "num_input_tokens_seen": 163109040,
      "step": 7646
    },
    {
      "epoch": 0.9194973847171286,
      "flos": 50550431047680.0,
      "grad_norm": 0.8593579377301234,
      "learning_rate": 6.751137674556994e-08,
      "loss": 0.8661,
      "num_input_tokens_seen": 163169145,
      "step": 7647
    },
    {
      "epoch": 0.9196176276077677,
      "flos": 10397996359680.0,
      "grad_norm": 5.686712860301422,
      "learning_rate": 6.731083880881572e-08,
      "loss": 0.9716,
      "num_input_tokens_seen": 163185085,
      "step": 7648
    },
    {
      "epoch": 0.9197378704984068,
      "flos": 16560126689280.0,
      "grad_norm": 9.384413405526224,
      "learning_rate": 6.711059405947072e-08,
      "loss": 1.0014,
      "num_input_tokens_seen": 163202995,
      "step": 7649
    },
    {
      "epoch": 0.9198581133890459,
      "flos": 14409530142720.0,
      "grad_norm": 2.6750173550336913,
      "learning_rate": 6.691064252791156e-08,
      "loss": 0.9649,
      "num_input_tokens_seen": 163222190,
      "step": 7650
    },
    {
      "epoch": 0.9199783562796849,
      "flos": 12521635921920.0,
      "grad_norm": 2.9115557077100758,
      "learning_rate": 6.67109842444713e-08,
      "loss": 0.9709,
      "num_input_tokens_seen": 163240840,
      "step": 7651
    },
    {
      "epoch": 0.9200985991703241,
      "flos": 12521942507520.0,
      "grad_norm": 3.8313754236995754,
      "learning_rate": 6.651161923943704e-08,
      "loss": 0.9525,
      "num_input_tokens_seen": 163258465,
      "step": 7652
    },
    {
      "epoch": 0.9202188420609632,
      "flos": 14908407029760.0,
      "grad_norm": 5.406776358498686,
      "learning_rate": 6.631254754305326e-08,
      "loss": 0.9563,
      "num_input_tokens_seen": 163277645,
      "step": 7653
    },
    {
      "epoch": 0.9203390849516022,
      "flos": 9821314375680.0,
      "grad_norm": 8.806431188946208,
      "learning_rate": 6.611376918551848e-08,
      "loss": 0.9758,
      "num_input_tokens_seen": 163296150,
      "step": 7654
    },
    {
      "epoch": 0.9204593278422414,
      "flos": 15036876288000.0,
      "grad_norm": 5.100703075041833,
      "learning_rate": 6.591528419698744e-08,
      "loss": 0.9856,
      "num_input_tokens_seen": 163315655,
      "step": 7655
    },
    {
      "epoch": 0.9205795707328804,
      "flos": 10240332042240.0,
      "grad_norm": 3.6257504968832888,
      "learning_rate": 6.571709260756986e-08,
      "loss": 1.0124,
      "num_input_tokens_seen": 163332020,
      "step": 7656
    },
    {
      "epoch": 0.9206998136235195,
      "flos": 15929377443840.0,
      "grad_norm": 35.984926246863246,
      "learning_rate": 6.551919444733122e-08,
      "loss": 0.9347,
      "num_input_tokens_seen": 163349555,
      "step": 7657
    },
    {
      "epoch": 0.9208200565141585,
      "flos": 38166102159360.0,
      "grad_norm": 3.1956533236249083,
      "learning_rate": 6.53215897462931e-08,
      "loss": 0.8494,
      "num_input_tokens_seen": 163373030,
      "step": 7658
    },
    {
      "epoch": 0.9209402994047977,
      "flos": 21801971097600.0,
      "grad_norm": 3.9672439000258706,
      "learning_rate": 6.512427853443103e-08,
      "loss": 0.9465,
      "num_input_tokens_seen": 163394830,
      "step": 7659
    },
    {
      "epoch": 0.9210605422954368,
      "flos": 20755668602880.0,
      "grad_norm": 3.9888960974323258,
      "learning_rate": 6.492726084167799e-08,
      "loss": 0.9468,
      "num_input_tokens_seen": 163416665,
      "step": 7660
    },
    {
      "epoch": 0.9211807851860758,
      "flos": 39245699850240.0,
      "grad_norm": 0.8016913475895363,
      "learning_rate": 6.473053669792072e-08,
      "loss": 0.772,
      "num_input_tokens_seen": 163471075,
      "step": 7661
    },
    {
      "epoch": 0.921301028076715,
      "flos": 13617744138240.0,
      "grad_norm": 3.993801201386199,
      "learning_rate": 6.453410613300248e-08,
      "loss": 0.916,
      "num_input_tokens_seen": 163488725,
      "step": 7662
    },
    {
      "epoch": 0.921421270967354,
      "flos": 19601538170880.0,
      "grad_norm": 3.9479432654512245,
      "learning_rate": 6.43379691767214e-08,
      "loss": 0.7692,
      "num_input_tokens_seen": 163507650,
      "step": 7663
    },
    {
      "epoch": 0.9215415138579931,
      "flos": 50280707911680.0,
      "grad_norm": 0.8453678585406841,
      "learning_rate": 6.414212585883105e-08,
      "loss": 0.7992,
      "num_input_tokens_seen": 163570000,
      "step": 7664
    },
    {
      "epoch": 0.9216617567486323,
      "flos": 25369737646080.0,
      "grad_norm": 2.330743226305505,
      "learning_rate": 6.394657620904143e-08,
      "loss": 0.885,
      "num_input_tokens_seen": 163592830,
      "step": 7665
    },
    {
      "epoch": 0.9217819996392713,
      "flos": 21043488399360.0,
      "grad_norm": 3.0410121753553008,
      "learning_rate": 6.375132025701657e-08,
      "loss": 0.9065,
      "num_input_tokens_seen": 163614850,
      "step": 7666
    },
    {
      "epoch": 0.9219022425299104,
      "flos": 10293050327040.0,
      "grad_norm": 4.292835014835721,
      "learning_rate": 6.355635803237724e-08,
      "loss": 0.874,
      "num_input_tokens_seen": 163630270,
      "step": 7667
    },
    {
      "epoch": 0.9220224854205495,
      "flos": 12810498109440.0,
      "grad_norm": 3.6906923602850727,
      "learning_rate": 6.336168956469867e-08,
      "loss": 0.9907,
      "num_input_tokens_seen": 163648465,
      "step": 7668
    },
    {
      "epoch": 0.9221427283111886,
      "flos": 17634919096320.0,
      "grad_norm": 2.628265529508675,
      "learning_rate": 6.316731488351168e-08,
      "loss": 0.9126,
      "num_input_tokens_seen": 163669375,
      "step": 7669
    },
    {
      "epoch": 0.9222629712018277,
      "flos": 9768688066560.0,
      "grad_norm": 7.213336714898314,
      "learning_rate": 6.297323401830334e-08,
      "loss": 0.8324,
      "num_input_tokens_seen": 163687880,
      "step": 7670
    },
    {
      "epoch": 0.9223832140924668,
      "flos": 15353615216640.0,
      "grad_norm": 4.24090747946588,
      "learning_rate": 6.277944699851523e-08,
      "loss": 0.8716,
      "num_input_tokens_seen": 163707120,
      "step": 7671
    },
    {
      "epoch": 0.9225034569831059,
      "flos": 15012923842560.0,
      "grad_norm": 5.913272295912431,
      "learning_rate": 6.25859538535447e-08,
      "loss": 0.9294,
      "num_input_tokens_seen": 163727635,
      "step": 7672
    },
    {
      "epoch": 0.9226236998737449,
      "flos": 8798473789440.0,
      "grad_norm": 4.390411263147259,
      "learning_rate": 6.239275461274474e-08,
      "loss": 0.9714,
      "num_input_tokens_seen": 163743730,
      "step": 7673
    },
    {
      "epoch": 0.9227439427643841,
      "flos": 18658127585280.0,
      "grad_norm": 3.776583100887324,
      "learning_rate": 6.219984930542299e-08,
      "loss": 1.0585,
      "num_input_tokens_seen": 163764190,
      "step": 7674
    },
    {
      "epoch": 0.9228641856550232,
      "flos": 12731436011520.0,
      "grad_norm": 3.443568058365511,
      "learning_rate": 6.200723796084383e-08,
      "loss": 0.9354,
      "num_input_tokens_seen": 163782005,
      "step": 7675
    },
    {
      "epoch": 0.9229844285456622,
      "flos": 50431535370240.0,
      "grad_norm": 0.8775800405444227,
      "learning_rate": 6.181492060822546e-08,
      "loss": 0.8352,
      "num_input_tokens_seen": 163841900,
      "step": 7676
    },
    {
      "epoch": 0.9231046714363014,
      "flos": 12731650621440.0,
      "grad_norm": 3.0350555442491074,
      "learning_rate": 6.162289727674274e-08,
      "loss": 0.9994,
      "num_input_tokens_seen": 163859300,
      "step": 7677
    },
    {
      "epoch": 0.9232249143269404,
      "flos": 12652864450560.0,
      "grad_norm": 4.477500191522932,
      "learning_rate": 6.143116799552527e-08,
      "loss": 1.0725,
      "num_input_tokens_seen": 163875265,
      "step": 7678
    },
    {
      "epoch": 0.9233451572175795,
      "flos": 16638820884480.0,
      "grad_norm": 13.168287990234342,
      "learning_rate": 6.123973279365802e-08,
      "loss": 0.7443,
      "num_input_tokens_seen": 163893960,
      "step": 7679
    },
    {
      "epoch": 0.9234654001082186,
      "flos": 12753610260480.0,
      "grad_norm": 4.2042654539410504,
      "learning_rate": 6.10485917001824e-08,
      "loss": 0.9687,
      "num_input_tokens_seen": 163911535,
      "step": 7680
    },
    {
      "epoch": 0.9235856429988577,
      "flos": 17605938647040.0,
      "grad_norm": 3.0228925504627298,
      "learning_rate": 6.085774474409322e-08,
      "loss": 1.0038,
      "num_input_tokens_seen": 163931815,
      "step": 7681
    },
    {
      "epoch": 0.9237058858894968,
      "flos": 9951776440320.0,
      "grad_norm": 4.207364011619333,
      "learning_rate": 6.066719195434267e-08,
      "loss": 0.8952,
      "num_input_tokens_seen": 163949335,
      "step": 7682
    },
    {
      "epoch": 0.9238261287801359,
      "flos": 20440707870720.0,
      "grad_norm": 4.623365371105374,
      "learning_rate": 6.047693335983717e-08,
      "loss": 0.8513,
      "num_input_tokens_seen": 163971400,
      "step": 7683
    },
    {
      "epoch": 0.923946371670775,
      "flos": 16428499599360.0,
      "grad_norm": 7.651385322748929,
      "learning_rate": 6.028696898943853e-08,
      "loss": 0.9965,
      "num_input_tokens_seen": 163990180,
      "step": 7684
    },
    {
      "epoch": 0.924066614561414,
      "flos": 15533361807360.0,
      "grad_norm": 3.153959595212796,
      "learning_rate": 6.00972988719648e-08,
      "loss": 0.8946,
      "num_input_tokens_seen": 164008135,
      "step": 7685
    },
    {
      "epoch": 0.9241868574520532,
      "flos": 20309418024960.0,
      "grad_norm": 7.26043581492573,
      "learning_rate": 5.990792303618807e-08,
      "loss": 0.898,
      "num_input_tokens_seen": 164027435,
      "step": 7686
    },
    {
      "epoch": 0.9243071003426923,
      "flos": 21751368253440.0,
      "grad_norm": 5.289264444076794,
      "learning_rate": 5.971884151083695e-08,
      "loss": 0.8814,
      "num_input_tokens_seen": 164049565,
      "step": 7687
    },
    {
      "epoch": 0.9244273432333313,
      "flos": 20414731960320.0,
      "grad_norm": 3.157184122675245,
      "learning_rate": 5.9530054324595124e-08,
      "loss": 0.9354,
      "num_input_tokens_seen": 164069400,
      "step": 7688
    },
    {
      "epoch": 0.9245475861239704,
      "flos": 47421035827200.0,
      "grad_norm": 0.752962935489164,
      "learning_rate": 5.934156150610103e-08,
      "loss": 0.7884,
      "num_input_tokens_seen": 164126485,
      "step": 7689
    },
    {
      "epoch": 0.9246678290146095,
      "flos": 17238811484160.0,
      "grad_norm": 4.954312443921173,
      "learning_rate": 5.915336308394914e-08,
      "loss": 0.9604,
      "num_input_tokens_seen": 164145040,
      "step": 7690
    },
    {
      "epoch": 0.9247880719052486,
      "flos": 13465322434560.0,
      "grad_norm": 2.5541076935457308,
      "learning_rate": 5.89654590866886e-08,
      "loss": 0.9653,
      "num_input_tokens_seen": 164164260,
      "step": 7691
    },
    {
      "epoch": 0.9249083147958876,
      "flos": 17084121047040.0,
      "grad_norm": 3.514194451785287,
      "learning_rate": 5.877784954282483e-08,
      "loss": 1.0751,
      "num_input_tokens_seen": 164183320,
      "step": 7692
    },
    {
      "epoch": 0.9250285576865268,
      "flos": 21935131115520.0,
      "grad_norm": 4.011519771644644,
      "learning_rate": 5.8590534480817963e-08,
      "loss": 0.9152,
      "num_input_tokens_seen": 164204765,
      "step": 7693
    },
    {
      "epoch": 0.9251488005771659,
      "flos": 7460580495360.0,
      "grad_norm": 7.310260658052616,
      "learning_rate": 5.840351392908349e-08,
      "loss": 0.9135,
      "num_input_tokens_seen": 164220205,
      "step": 7694
    },
    {
      "epoch": 0.9252690434678049,
      "flos": 16769589534720.0,
      "grad_norm": 4.956267025161884,
      "learning_rate": 5.821678791599205e-08,
      "loss": 0.9056,
      "num_input_tokens_seen": 164239370,
      "step": 7695
    },
    {
      "epoch": 0.9253892863584441,
      "flos": 15247810744320.0,
      "grad_norm": 2.8491453194262477,
      "learning_rate": 5.803035646986965e-08,
      "loss": 0.9925,
      "num_input_tokens_seen": 164258425,
      "step": 7696
    },
    {
      "epoch": 0.9255095292490831,
      "flos": 12364247531520.0,
      "grad_norm": 3.495247884535949,
      "learning_rate": 5.7844219618998766e-08,
      "loss": 0.8638,
      "num_input_tokens_seen": 164272470,
      "step": 7697
    },
    {
      "epoch": 0.9256297721397222,
      "flos": 17605877329920.0,
      "grad_norm": 5.232850174356765,
      "learning_rate": 5.765837739161505e-08,
      "loss": 0.8901,
      "num_input_tokens_seen": 164291310,
      "step": 7698
    },
    {
      "epoch": 0.9257500150303614,
      "flos": 16428806184960.0,
      "grad_norm": 2.8714028120749435,
      "learning_rate": 5.7472829815911504e-08,
      "loss": 0.9432,
      "num_input_tokens_seen": 164309855,
      "step": 7699
    },
    {
      "epoch": 0.9258702579210004,
      "flos": 16035273891840.0,
      "grad_norm": 3.033962334053936,
      "learning_rate": 5.7287576920035164e-08,
      "loss": 1.0066,
      "num_input_tokens_seen": 164328590,
      "step": 7700
    },
    {
      "epoch": 0.9259905008116395,
      "flos": 21381512478720.0,
      "grad_norm": 2.6569557598710802,
      "learning_rate": 5.7102618732088435e-08,
      "loss": 0.9568,
      "num_input_tokens_seen": 164347640,
      "step": 7701
    },
    {
      "epoch": 0.9261107437022786,
      "flos": 17478082560000.0,
      "grad_norm": 4.161862821181638,
      "learning_rate": 5.6917955280130216e-08,
      "loss": 0.9366,
      "num_input_tokens_seen": 164368840,
      "step": 7702
    },
    {
      "epoch": 0.9262309865929177,
      "flos": 15642845306880.0,
      "grad_norm": 3.0203731493280164,
      "learning_rate": 5.6733586592172755e-08,
      "loss": 0.9068,
      "num_input_tokens_seen": 164388055,
      "step": 7703
    },
    {
      "epoch": 0.9263512294835567,
      "flos": 14436119224320.0,
      "grad_norm": 5.747947658313778,
      "learning_rate": 5.6549512696185244e-08,
      "loss": 1.0027,
      "num_input_tokens_seen": 164406275,
      "step": 7704
    },
    {
      "epoch": 0.9264714723741959,
      "flos": 15065642127360.0,
      "grad_norm": 3.1380463447430746,
      "learning_rate": 5.636573362009156e-08,
      "loss": 0.8642,
      "num_input_tokens_seen": 164426055,
      "step": 7705
    },
    {
      "epoch": 0.926591715264835,
      "flos": 12758423654400.0,
      "grad_norm": 3.184141446609142,
      "learning_rate": 5.618224939177074e-08,
      "loss": 0.954,
      "num_input_tokens_seen": 164443680,
      "step": 7706
    },
    {
      "epoch": 0.926711958155474,
      "flos": 25810898903040.0,
      "grad_norm": 4.525618006196799,
      "learning_rate": 5.599906003905719e-08,
      "loss": 0.8953,
      "num_input_tokens_seen": 164465945,
      "step": 7707
    },
    {
      "epoch": 0.9268322010461132,
      "flos": 14934352281600.0,
      "grad_norm": 4.550666617722003,
      "learning_rate": 5.581616558974023e-08,
      "loss": 1.0094,
      "num_input_tokens_seen": 164484230,
      "step": 7708
    },
    {
      "epoch": 0.9269524439367522,
      "flos": 16322695127040.0,
      "grad_norm": 3.3613759313948575,
      "learning_rate": 5.5633566071565444e-08,
      "loss": 0.9823,
      "num_input_tokens_seen": 164503190,
      "step": 7709
    },
    {
      "epoch": 0.9270726868273913,
      "flos": 29985554227200.0,
      "grad_norm": 4.247603919075403,
      "learning_rate": 5.5451261512232896e-08,
      "loss": 0.8858,
      "num_input_tokens_seen": 164525590,
      "step": 7710
    },
    {
      "epoch": 0.9271929297180305,
      "flos": 14042433638400.0,
      "grad_norm": 3.3745487052691057,
      "learning_rate": 5.5269251939397576e-08,
      "loss": 0.8176,
      "num_input_tokens_seen": 164544825,
      "step": 7711
    },
    {
      "epoch": 0.9273131726086695,
      "flos": 14168542187520.0,
      "grad_norm": 3.108173387382093,
      "learning_rate": 5.508753738067073e-08,
      "loss": 0.962,
      "num_input_tokens_seen": 164564085,
      "step": 7712
    },
    {
      "epoch": 0.9274334154993086,
      "flos": 16534058803200.0,
      "grad_norm": 3.694990535202603,
      "learning_rate": 5.4906117863617875e-08,
      "loss": 0.9808,
      "num_input_tokens_seen": 164583190,
      "step": 7713
    },
    {
      "epoch": 0.9275536583899477,
      "flos": 22669140172800.0,
      "grad_norm": 6.081779420866254,
      "learning_rate": 5.4724993415760533e-08,
      "loss": 0.973,
      "num_input_tokens_seen": 164601265,
      "step": 7714
    },
    {
      "epoch": 0.9276739012805868,
      "flos": 13255614320640.0,
      "grad_norm": 3.9562169222782266,
      "learning_rate": 5.454416406457496e-08,
      "loss": 0.9335,
      "num_input_tokens_seen": 164620080,
      "step": 7715
    },
    {
      "epoch": 0.9277941441712259,
      "flos": 9792824463360.0,
      "grad_norm": 4.120491801046947,
      "learning_rate": 5.436362983749299e-08,
      "loss": 0.926,
      "num_input_tokens_seen": 164634970,
      "step": 7716
    },
    {
      "epoch": 0.927914387061865,
      "flos": 16534120120320.0,
      "grad_norm": 3.0546583091336226,
      "learning_rate": 5.418339076190137e-08,
      "loss": 0.8277,
      "num_input_tokens_seen": 164654200,
      "step": 7717
    },
    {
      "epoch": 0.9280346299525041,
      "flos": 12807278960640.0,
      "grad_norm": 3.75677247436324,
      "learning_rate": 5.400344686514202e-08,
      "loss": 1.0833,
      "num_input_tokens_seen": 164671505,
      "step": 7718
    },
    {
      "epoch": 0.9281548728431431,
      "flos": 15875494133760.0,
      "grad_norm": 4.502090816088519,
      "learning_rate": 5.38237981745131e-08,
      "loss": 0.8602,
      "num_input_tokens_seen": 164689340,
      "step": 7719
    },
    {
      "epoch": 0.9282751157337822,
      "flos": 13361204183040.0,
      "grad_norm": 3.6207676877537507,
      "learning_rate": 5.364444471726592e-08,
      "loss": 1.002,
      "num_input_tokens_seen": 164708265,
      "step": 7720
    },
    {
      "epoch": 0.9283953586244214,
      "flos": 18185717145600.0,
      "grad_norm": 4.876389525528852,
      "learning_rate": 5.346538652060939e-08,
      "loss": 0.9797,
      "num_input_tokens_seen": 164729340,
      "step": 7721
    },
    {
      "epoch": 0.9285156015150604,
      "flos": 12915413483520.0,
      "grad_norm": 4.296309444314805,
      "learning_rate": 5.3286623611705994e-08,
      "loss": 0.9015,
      "num_input_tokens_seen": 164747105,
      "step": 7722
    },
    {
      "epoch": 0.9286358444056995,
      "flos": 47545971240960.0,
      "grad_norm": 1.0850162028496013,
      "learning_rate": 5.3108156017673824e-08,
      "loss": 0.842,
      "num_input_tokens_seen": 164808585,
      "step": 7723
    },
    {
      "epoch": 0.9287560872963386,
      "flos": 15877824184320.0,
      "grad_norm": 3.6659036333703203,
      "learning_rate": 5.2929983765586775e-08,
      "loss": 0.9137,
      "num_input_tokens_seen": 164827085,
      "step": 7724
    },
    {
      "epoch": 0.9288763301869777,
      "flos": 18288425103360.0,
      "grad_norm": 9.578800445188843,
      "learning_rate": 5.275210688247278e-08,
      "loss": 0.8236,
      "num_input_tokens_seen": 164847130,
      "step": 7725
    },
    {
      "epoch": 0.9289965730776167,
      "flos": 8666877358080.0,
      "grad_norm": 3.2413287760572933,
      "learning_rate": 5.257452539531604e-08,
      "loss": 1.034,
      "num_input_tokens_seen": 164863920,
      "step": 7726
    },
    {
      "epoch": 0.9291168159682559,
      "flos": 18996948787200.0,
      "grad_norm": 2.6388065012959436,
      "learning_rate": 5.2397239331055445e-08,
      "loss": 0.8877,
      "num_input_tokens_seen": 164885640,
      "step": 7727
    },
    {
      "epoch": 0.929237058858895,
      "flos": 10266890465280.0,
      "grad_norm": 5.221625535212236,
      "learning_rate": 5.2220248716585036e-08,
      "loss": 0.9916,
      "num_input_tokens_seen": 164903040,
      "step": 7728
    },
    {
      "epoch": 0.929357301749534,
      "flos": 16947711221760.0,
      "grad_norm": 5.132989192342611,
      "learning_rate": 5.204355357875445e-08,
      "loss": 0.9411,
      "num_input_tokens_seen": 164921105,
      "step": 7729
    },
    {
      "epoch": 0.9294775446401732,
      "flos": 9139318456320.0,
      "grad_norm": 3.46659976882454,
      "learning_rate": 5.1867153944367584e-08,
      "loss": 0.8932,
      "num_input_tokens_seen": 164937215,
      "step": 7730
    },
    {
      "epoch": 0.9295977875308122,
      "flos": 18656196096000.0,
      "grad_norm": 2.4955287049036907,
      "learning_rate": 5.16910498401848e-08,
      "loss": 0.9293,
      "num_input_tokens_seen": 164956385,
      "step": 7731
    },
    {
      "epoch": 0.9297180304214513,
      "flos": 11656888872960.0,
      "grad_norm": 6.83637749700045,
      "learning_rate": 5.151524129292073e-08,
      "loss": 1.0112,
      "num_input_tokens_seen": 164974000,
      "step": 7732
    },
    {
      "epoch": 0.9298382733120905,
      "flos": 17110219591680.0,
      "grad_norm": 3.338058333539178,
      "learning_rate": 5.1339728329245155e-08,
      "loss": 0.8637,
      "num_input_tokens_seen": 164994285,
      "step": 7733
    },
    {
      "epoch": 0.9299585162027295,
      "flos": 15721294233600.0,
      "grad_norm": 4.961704272989528,
      "learning_rate": 5.116451097578367e-08,
      "loss": 0.9767,
      "num_input_tokens_seen": 165013045,
      "step": 7734
    },
    {
      "epoch": 0.9300787590933686,
      "flos": 15250048819200.0,
      "grad_norm": 2.6382985899278872,
      "learning_rate": 5.0989589259115895e-08,
      "loss": 0.9359,
      "num_input_tokens_seen": 165033650,
      "step": 7735
    },
    {
      "epoch": 0.9301990019840077,
      "flos": 12595578040320.0,
      "grad_norm": 4.0058612795582995,
      "learning_rate": 5.081496320577816e-08,
      "loss": 0.8961,
      "num_input_tokens_seen": 165050490,
      "step": 7736
    },
    {
      "epoch": 0.9303192448746468,
      "flos": 42149781749760.0,
      "grad_norm": 0.9338662615686225,
      "learning_rate": 5.0640632842260835e-08,
      "loss": 0.8497,
      "num_input_tokens_seen": 165110470,
      "step": 7737
    },
    {
      "epoch": 0.9304394877652858,
      "flos": 41260078632960.0,
      "grad_norm": 2.244759602918816,
      "learning_rate": 5.0466598195009426e-08,
      "loss": 0.9207,
      "num_input_tokens_seen": 165137060,
      "step": 7738
    },
    {
      "epoch": 0.930559730655925,
      "flos": 14908039127040.0,
      "grad_norm": 3.5423880945593167,
      "learning_rate": 5.0292859290425036e-08,
      "loss": 0.8838,
      "num_input_tokens_seen": 165154650,
      "step": 7739
    },
    {
      "epoch": 0.9306799735465641,
      "flos": 16533782876160.0,
      "grad_norm": 3.19086353632083,
      "learning_rate": 5.011941615486348e-08,
      "loss": 0.9612,
      "num_input_tokens_seen": 165173485,
      "step": 7740
    },
    {
      "epoch": 0.9308002164372031,
      "flos": 10765154181120.0,
      "grad_norm": 4.1563986844880585,
      "learning_rate": 4.994626881463659e-08,
      "loss": 1.0459,
      "num_input_tokens_seen": 165189460,
      "step": 7741
    },
    {
      "epoch": 0.9309204593278423,
      "flos": 21988033351680.0,
      "grad_norm": 4.524050550804055,
      "learning_rate": 4.9773417296009814e-08,
      "loss": 0.9044,
      "num_input_tokens_seen": 165210700,
      "step": 7742
    },
    {
      "epoch": 0.9310407022184813,
      "flos": 16374677606400.0,
      "grad_norm": 7.684347110928829,
      "learning_rate": 4.960086162520527e-08,
      "loss": 0.8464,
      "num_input_tokens_seen": 165230510,
      "step": 7743
    },
    {
      "epoch": 0.9311609451091204,
      "flos": 15721570160640.0,
      "grad_norm": 3.2095379900379855,
      "learning_rate": 4.942860182839936e-08,
      "loss": 1.0269,
      "num_input_tokens_seen": 165248575,
      "step": 7744
    },
    {
      "epoch": 0.9312811879997596,
      "flos": 14981245440000.0,
      "grad_norm": 4.385129265266385,
      "learning_rate": 4.925663793172341e-08,
      "loss": 0.9825,
      "num_input_tokens_seen": 165266255,
      "step": 7745
    },
    {
      "epoch": 0.9314014308903986,
      "flos": 48081225400320.0,
      "grad_norm": 0.8473881943695816,
      "learning_rate": 4.908496996126477e-08,
      "loss": 0.8047,
      "num_input_tokens_seen": 165329435,
      "step": 7746
    },
    {
      "epoch": 0.9315216737810377,
      "flos": 16036040355840.0,
      "grad_norm": 2.6362818829348056,
      "learning_rate": 4.89135979430646e-08,
      "loss": 0.9521,
      "num_input_tokens_seen": 165349200,
      "step": 7747
    },
    {
      "epoch": 0.9316419166716768,
      "flos": 17054864670720.0,
      "grad_norm": 2.7829156238568995,
      "learning_rate": 4.874252190312078e-08,
      "loss": 1.0336,
      "num_input_tokens_seen": 165369305,
      "step": 7748
    },
    {
      "epoch": 0.9317621595623159,
      "flos": 21545032581120.0,
      "grad_norm": 5.0879875305670765,
      "learning_rate": 4.857174186738477e-08,
      "loss": 0.8411,
      "num_input_tokens_seen": 165392375,
      "step": 7749
    },
    {
      "epoch": 0.931882402452955,
      "flos": 11132587929600.0,
      "grad_norm": 10.035152828799534,
      "learning_rate": 4.840125786176408e-08,
      "loss": 0.9267,
      "num_input_tokens_seen": 165408300,
      "step": 7750
    },
    {
      "epoch": 0.932002645343594,
      "flos": 20205023846400.0,
      "grad_norm": 2.655962760204621,
      "learning_rate": 4.823106991212067e-08,
      "loss": 0.9655,
      "num_input_tokens_seen": 165427260,
      "step": 7751
    },
    {
      "epoch": 0.9321228882342332,
      "flos": 10843940352000.0,
      "grad_norm": 6.040063614668167,
      "learning_rate": 4.806117804427212e-08,
      "loss": 1.037,
      "num_input_tokens_seen": 165444915,
      "step": 7752
    },
    {
      "epoch": 0.9322431311248722,
      "flos": 12679637483520.0,
      "grad_norm": 3.828764445557526,
      "learning_rate": 4.7891582283990926e-08,
      "loss": 0.8439,
      "num_input_tokens_seen": 165463360,
      "step": 7753
    },
    {
      "epoch": 0.9323633740155113,
      "flos": 17188729835520.0,
      "grad_norm": 3.3304893441114976,
      "learning_rate": 4.772228265700473e-08,
      "loss": 0.9167,
      "num_input_tokens_seen": 165483940,
      "step": 7754
    },
    {
      "epoch": 0.9324836169061504,
      "flos": 10629234892800.0,
      "grad_norm": 6.28973362344981,
      "learning_rate": 4.75532791889961e-08,
      "loss": 0.9433,
      "num_input_tokens_seen": 165500360,
      "step": 7755
    },
    {
      "epoch": 0.9326038597967895,
      "flos": 13203907768320.0,
      "grad_norm": 3.270347429536423,
      "learning_rate": 4.738457190560252e-08,
      "loss": 0.8437,
      "num_input_tokens_seen": 165519190,
      "step": 7756
    },
    {
      "epoch": 0.9327241026874286,
      "flos": 13439377182720.0,
      "grad_norm": 7.33530801138072,
      "learning_rate": 4.721616083241664e-08,
      "loss": 0.9766,
      "num_input_tokens_seen": 165537165,
      "step": 7757
    },
    {
      "epoch": 0.9328443455780677,
      "flos": 21070108139520.0,
      "grad_norm": 6.7065531894090755,
      "learning_rate": 4.7048045994986684e-08,
      "loss": 0.9652,
      "num_input_tokens_seen": 165557745,
      "step": 7758
    },
    {
      "epoch": 0.9329645884687068,
      "flos": 21437756497920.0,
      "grad_norm": 2.8168490375264383,
      "learning_rate": 4.688022741881559e-08,
      "loss": 1.0929,
      "num_input_tokens_seen": 165577990,
      "step": 7759
    },
    {
      "epoch": 0.9330848313593458,
      "flos": 15534097612800.0,
      "grad_norm": 2.5836687075408733,
      "learning_rate": 4.671270512936076e-08,
      "loss": 0.953,
      "num_input_tokens_seen": 165596870,
      "step": 7760
    },
    {
      "epoch": 0.933205074249985,
      "flos": 15720926330880.0,
      "grad_norm": 3.6507703360148094,
      "learning_rate": 4.6545479152035884e-08,
      "loss": 1.0134,
      "num_input_tokens_seen": 165615760,
      "step": 7761
    },
    {
      "epoch": 0.9333253171406241,
      "flos": 10844032327680.0,
      "grad_norm": 3.7579041150637007,
      "learning_rate": 4.637854951220821e-08,
      "loss": 0.9634,
      "num_input_tokens_seen": 165632265,
      "step": 7762
    },
    {
      "epoch": 0.9334455600312631,
      "flos": 11106274775040.0,
      "grad_norm": 2.980423633238962,
      "learning_rate": 4.621191623520171e-08,
      "loss": 0.938,
      "num_input_tokens_seen": 165650415,
      "step": 7763
    },
    {
      "epoch": 0.9335658029219023,
      "flos": 15747024875520.0,
      "grad_norm": 5.445694284724013,
      "learning_rate": 4.604557934629372e-08,
      "loss": 1.0371,
      "num_input_tokens_seen": 165669210,
      "step": 7764
    },
    {
      "epoch": 0.9336860458125413,
      "flos": 14383952793600.0,
      "grad_norm": 5.114441392512607,
      "learning_rate": 4.587953887071805e-08,
      "loss": 1.0085,
      "num_input_tokens_seen": 165688750,
      "step": 7765
    },
    {
      "epoch": 0.9338062887031804,
      "flos": 14852622888960.0,
      "grad_norm": 3.1312666049296185,
      "learning_rate": 4.5713794833662554e-08,
      "loss": 1.0477,
      "num_input_tokens_seen": 165707685,
      "step": 7766
    },
    {
      "epoch": 0.9339265315938196,
      "flos": 16507439063040.0,
      "grad_norm": 4.844421143334801,
      "learning_rate": 4.5548347260270236e-08,
      "loss": 0.8237,
      "num_input_tokens_seen": 165726695,
      "step": 7767
    },
    {
      "epoch": 0.9340467744844586,
      "flos": 16218975436800.0,
      "grad_norm": 3.990696817228306,
      "learning_rate": 4.538319617564012e-08,
      "loss": 0.8816,
      "num_input_tokens_seen": 165745435,
      "step": 7768
    },
    {
      "epoch": 0.9341670173750977,
      "flos": 16822920990720.0,
      "grad_norm": 4.320040316570204,
      "learning_rate": 4.521834160482485e-08,
      "loss": 0.9392,
      "num_input_tokens_seen": 165763895,
      "step": 7769
    },
    {
      "epoch": 0.9342872602657368,
      "flos": 17658840883200.0,
      "grad_norm": 4.115939043325839,
      "learning_rate": 4.5053783572832846e-08,
      "loss": 1.0125,
      "num_input_tokens_seen": 165783795,
      "step": 7770
    },
    {
      "epoch": 0.9344075031563759,
      "flos": 18339702435840.0,
      "grad_norm": 3.423343848336429,
      "learning_rate": 4.488952210462771e-08,
      "loss": 0.9576,
      "num_input_tokens_seen": 165803720,
      "step": 7771
    },
    {
      "epoch": 0.9345277460470149,
      "flos": 18180995727360.0,
      "grad_norm": 4.219813686072646,
      "learning_rate": 4.4725557225127495e-08,
      "loss": 1.0439,
      "num_input_tokens_seen": 165821780,
      "step": 7772
    },
    {
      "epoch": 0.9346479889376541,
      "flos": 24478646784000.0,
      "grad_norm": 2.6227969052639604,
      "learning_rate": 4.456188895920565e-08,
      "loss": 0.9902,
      "num_input_tokens_seen": 165843255,
      "step": 7773
    },
    {
      "epoch": 0.9347682318282932,
      "flos": 13540122992640.0,
      "grad_norm": 2.8414715033341884,
      "learning_rate": 4.439851733169031e-08,
      "loss": 1.0328,
      "num_input_tokens_seen": 165860765,
      "step": 7774
    },
    {
      "epoch": 0.9348884747189322,
      "flos": 18683520983040.0,
      "grad_norm": 4.3351836777859365,
      "learning_rate": 4.4235442367365204e-08,
      "loss": 0.8722,
      "num_input_tokens_seen": 165880795,
      "step": 7775
    },
    {
      "epoch": 0.9350087176095714,
      "flos": 12994199654400.0,
      "grad_norm": 3.968903906533807,
      "learning_rate": 4.4072664090968545e-08,
      "loss": 0.9687,
      "num_input_tokens_seen": 165898900,
      "step": 7776
    },
    {
      "epoch": 0.9351289605002104,
      "flos": 13702232801280.0,
      "grad_norm": 5.0831531221725115,
      "learning_rate": 4.391018252719347e-08,
      "loss": 1.0235,
      "num_input_tokens_seen": 165918415,
      "step": 7777
    },
    {
      "epoch": 0.9352492033908495,
      "flos": 13328759316480.0,
      "grad_norm": 5.805193502928676,
      "learning_rate": 4.374799770068849e-08,
      "loss": 0.8784,
      "num_input_tokens_seen": 165934810,
      "step": 7778
    },
    {
      "epoch": 0.9353694462814887,
      "flos": 21040759787520.0,
      "grad_norm": 6.683233879027959,
      "learning_rate": 4.358610963605658e-08,
      "loss": 0.9461,
      "num_input_tokens_seen": 165954980,
      "step": 7779
    },
    {
      "epoch": 0.9354896891721277,
      "flos": 21857264701440.0,
      "grad_norm": 4.277100271323672,
      "learning_rate": 4.342451835785677e-08,
      "loss": 0.8733,
      "num_input_tokens_seen": 165975610,
      "step": 7780
    },
    {
      "epoch": 0.9356099320627668,
      "flos": 13806872248320.0,
      "grad_norm": 4.528734232248333,
      "learning_rate": 4.3263223890601665e-08,
      "loss": 0.9423,
      "num_input_tokens_seen": 165994040,
      "step": 7781
    },
    {
      "epoch": 0.9357301749534058,
      "flos": 13543893995520.0,
      "grad_norm": 4.325962889986125,
      "learning_rate": 4.31022262587597e-08,
      "loss": 0.9816,
      "num_input_tokens_seen": 166012435,
      "step": 7782
    },
    {
      "epoch": 0.935850417844045,
      "flos": 16743736258560.0,
      "grad_norm": 2.980817341948683,
      "learning_rate": 4.2941525486754225e-08,
      "loss": 0.8552,
      "num_input_tokens_seen": 166032475,
      "step": 7783
    },
    {
      "epoch": 0.935970660734684,
      "flos": 13177625272320.0,
      "grad_norm": 3.5539351453128467,
      "learning_rate": 4.278112159896286e-08,
      "loss": 0.985,
      "num_input_tokens_seen": 166050035,
      "step": 7784
    },
    {
      "epoch": 0.9360909036253231,
      "flos": 14645796679680.0,
      "grad_norm": 2.6991882764142225,
      "learning_rate": 4.2621014619719896e-08,
      "loss": 0.8701,
      "num_input_tokens_seen": 166067520,
      "step": 7785
    },
    {
      "epoch": 0.9362111465159623,
      "flos": 44232322621440.0,
      "grad_norm": 0.7716723249350191,
      "learning_rate": 4.246120457331215e-08,
      "loss": 0.8202,
      "num_input_tokens_seen": 166129415,
      "step": 7786
    },
    {
      "epoch": 0.9363313894066013,
      "flos": 17189312348160.0,
      "grad_norm": 6.064293035829821,
      "learning_rate": 4.2301691483983325e-08,
      "loss": 0.9163,
      "num_input_tokens_seen": 166149255,
      "step": 7787
    },
    {
      "epoch": 0.9364516322972404,
      "flos": 14278608199680.0,
      "grad_norm": 2.919134639905881,
      "learning_rate": 4.214247537593163e-08,
      "loss": 0.9483,
      "num_input_tokens_seen": 166168225,
      "step": 7788
    },
    {
      "epoch": 0.9365718751878795,
      "flos": 14697840476160.0,
      "grad_norm": 4.985453234914508,
      "learning_rate": 4.1983556273309293e-08,
      "loss": 1.0023,
      "num_input_tokens_seen": 166186695,
      "step": 7789
    },
    {
      "epoch": 0.9366921180785186,
      "flos": 13224916992000.0,
      "grad_norm": 5.603140168560408,
      "learning_rate": 4.182493420022526e-08,
      "loss": 0.8824,
      "num_input_tokens_seen": 166202085,
      "step": 7790
    },
    {
      "epoch": 0.9368123609691577,
      "flos": 18342369730560.0,
      "grad_norm": 3.0345585650658404,
      "learning_rate": 4.166660918074139e-08,
      "loss": 0.9789,
      "num_input_tokens_seen": 166221710,
      "step": 7791
    },
    {
      "epoch": 0.9369326038597968,
      "flos": 18182773923840.0,
      "grad_norm": 2.1949137804133643,
      "learning_rate": 4.15085812388758e-08,
      "loss": 0.9256,
      "num_input_tokens_seen": 166243650,
      "step": 7792
    },
    {
      "epoch": 0.9370528467504359,
      "flos": 16506059427840.0,
      "grad_norm": 3.330566943137418,
      "learning_rate": 4.135085039860153e-08,
      "loss": 0.9831,
      "num_input_tokens_seen": 166262770,
      "step": 7793
    },
    {
      "epoch": 0.9371730896410749,
      "flos": 17761395548160.0,
      "grad_norm": 4.936035132099182,
      "learning_rate": 4.1193416683845906e-08,
      "loss": 0.9748,
      "num_input_tokens_seen": 166281420,
      "step": 7794
    },
    {
      "epoch": 0.9372933325317141,
      "flos": 10995319664640.0,
      "grad_norm": 4.303068157587985,
      "learning_rate": 4.103628011849136e-08,
      "loss": 1.0255,
      "num_input_tokens_seen": 166296500,
      "step": 7795
    },
    {
      "epoch": 0.9374135754223532,
      "flos": 15537408737280.0,
      "grad_norm": 3.9499658317536728,
      "learning_rate": 4.0879440726375506e-08,
      "loss": 0.9382,
      "num_input_tokens_seen": 166314005,
      "step": 7796
    },
    {
      "epoch": 0.9375338183129922,
      "flos": 16082075074560.0,
      "grad_norm": 5.584011188561453,
      "learning_rate": 4.0722898531291074e-08,
      "loss": 0.7449,
      "num_input_tokens_seen": 166330965,
      "step": 7797
    },
    {
      "epoch": 0.9376540612036314,
      "flos": 18579188121600.0,
      "grad_norm": 3.1217023616555157,
      "learning_rate": 4.0566653556985295e-08,
      "loss": 0.9543,
      "num_input_tokens_seen": 166351230,
      "step": 7798
    },
    {
      "epoch": 0.9377743040942704,
      "flos": 13988948889600.0,
      "grad_norm": 5.522526818379137,
      "learning_rate": 4.0410705827159886e-08,
      "loss": 0.9859,
      "num_input_tokens_seen": 166368245,
      "step": 7799
    },
    {
      "epoch": 0.9378945469849095,
      "flos": 10790946140160.0,
      "grad_norm": 5.0111358612832735,
      "learning_rate": 4.0255055365472356e-08,
      "loss": 0.9074,
      "num_input_tokens_seen": 166386060,
      "step": 7800
    },
    {
      "epoch": 0.9380147898755486,
      "flos": 14616693596160.0,
      "grad_norm": 4.987356221127183,
      "learning_rate": 4.009970219553471e-08,
      "loss": 0.9323,
      "num_input_tokens_seen": 166402730,
      "step": 7801
    },
    {
      "epoch": 0.9381350327661877,
      "flos": 18705511280640.0,
      "grad_norm": 4.652508147415497,
      "learning_rate": 3.99446463409141e-08,
      "loss": 0.9493,
      "num_input_tokens_seen": 166420305,
      "step": 7802
    },
    {
      "epoch": 0.9382552756568268,
      "flos": 16769528217600.0,
      "grad_norm": 3.2189627271558083,
      "learning_rate": 3.978988782513215e-08,
      "loss": 0.8722,
      "num_input_tokens_seen": 166437520,
      "step": 7803
    },
    {
      "epoch": 0.9383755185474659,
      "flos": 20256883691520.0,
      "grad_norm": 3.2733836021629332,
      "learning_rate": 3.963542667166586e-08,
      "loss": 0.9519,
      "num_input_tokens_seen": 166457345,
      "step": 7804
    },
    {
      "epoch": 0.938495761438105,
      "flos": 14515334615040.0,
      "grad_norm": 3.2927241658778756,
      "learning_rate": 3.9481262903946486e-08,
      "loss": 0.87,
      "num_input_tokens_seen": 166476510,
      "step": 7805
    },
    {
      "epoch": 0.938616004328744,
      "flos": 49629899857920.0,
      "grad_norm": 0.7947864088998642,
      "learning_rate": 3.932739654536066e-08,
      "loss": 0.7636,
      "num_input_tokens_seen": 166538930,
      "step": 7806
    },
    {
      "epoch": 0.9387362472193832,
      "flos": 13409293025280.0,
      "grad_norm": 3.6642280329745742,
      "learning_rate": 3.917382761925014e-08,
      "loss": 0.9339,
      "num_input_tokens_seen": 166554485,
      "step": 7807
    },
    {
      "epoch": 0.9388564901100223,
      "flos": 18864493916160.0,
      "grad_norm": 3.077415888091632,
      "learning_rate": 3.9020556148910754e-08,
      "loss": 0.9798,
      "num_input_tokens_seen": 166573560,
      "step": 7808
    },
    {
      "epoch": 0.9389767330006613,
      "flos": 42183238348800.0,
      "grad_norm": 0.7168217792394328,
      "learning_rate": 3.8867582157593895e-08,
      "loss": 0.7795,
      "num_input_tokens_seen": 166627485,
      "step": 7809
    },
    {
      "epoch": 0.9390969758913005,
      "flos": 22171305676800.0,
      "grad_norm": 4.577418979196508,
      "learning_rate": 3.871490566850544e-08,
      "loss": 0.9515,
      "num_input_tokens_seen": 166651415,
      "step": 7810
    },
    {
      "epoch": 0.9392172187819395,
      "flos": 15930941030400.0,
      "grad_norm": 2.8536202309069107,
      "learning_rate": 3.856252670480642e-08,
      "loss": 0.8965,
      "num_input_tokens_seen": 166669795,
      "step": 7811
    },
    {
      "epoch": 0.9393374616725786,
      "flos": 13990573793280.0,
      "grad_norm": 3.114327938023579,
      "learning_rate": 3.841044528961279e-08,
      "loss": 1.0032,
      "num_input_tokens_seen": 166687310,
      "step": 7812
    },
    {
      "epoch": 0.9394577045632178,
      "flos": 17189159055360.0,
      "grad_norm": 3.168652480692068,
      "learning_rate": 3.825866144599477e-08,
      "loss": 0.9769,
      "num_input_tokens_seen": 166706085,
      "step": 7813
    },
    {
      "epoch": 0.9395779474538568,
      "flos": 13490378588160.0,
      "grad_norm": 3.699107505793324,
      "learning_rate": 3.8107175196978145e-08,
      "loss": 0.9518,
      "num_input_tokens_seen": 166722110,
      "step": 7814
    },
    {
      "epoch": 0.9396981903444959,
      "flos": 10109379440640.0,
      "grad_norm": 4.500350938464237,
      "learning_rate": 3.7955986565542996e-08,
      "loss": 0.9386,
      "num_input_tokens_seen": 166739910,
      "step": 7815
    },
    {
      "epoch": 0.9398184332351349,
      "flos": 24819644743680.0,
      "grad_norm": 5.157682832407111,
      "learning_rate": 3.780509557462497e-08,
      "loss": 0.8793,
      "num_input_tokens_seen": 166759830,
      "step": 7816
    },
    {
      "epoch": 0.9399386761257741,
      "flos": 18052771737600.0,
      "grad_norm": 3.325746071082591,
      "learning_rate": 3.765450224711375e-08,
      "loss": 0.9433,
      "num_input_tokens_seen": 166780055,
      "step": 7817
    },
    {
      "epoch": 0.9400589190164131,
      "flos": 19441635778560.0,
      "grad_norm": 3.301154662911066,
      "learning_rate": 3.750420660585396e-08,
      "loss": 0.987,
      "num_input_tokens_seen": 166801715,
      "step": 7818
    },
    {
      "epoch": 0.9401791619070522,
      "flos": 16634957905920.0,
      "grad_norm": 3.640887336938725,
      "learning_rate": 3.735420867364603e-08,
      "loss": 0.994,
      "num_input_tokens_seen": 166822415,
      "step": 7819
    },
    {
      "epoch": 0.9402994047976914,
      "flos": 25605666938880.0,
      "grad_norm": 2.7594056971258443,
      "learning_rate": 3.7204508473244186e-08,
      "loss": 0.8147,
      "num_input_tokens_seen": 166845760,
      "step": 7820
    },
    {
      "epoch": 0.9404196476883304,
      "flos": 15799712501760.0,
      "grad_norm": 2.970373573056625,
      "learning_rate": 3.7055106027357395e-08,
      "loss": 0.8869,
      "num_input_tokens_seen": 166865345,
      "step": 7821
    },
    {
      "epoch": 0.9405398905789695,
      "flos": 13412849418240.0,
      "grad_norm": 3.670073111738758,
      "learning_rate": 3.690600135865063e-08,
      "loss": 0.9138,
      "num_input_tokens_seen": 166881990,
      "step": 7822
    },
    {
      "epoch": 0.9406601334696086,
      "flos": 50326190776320.0,
      "grad_norm": 0.7511300871320177,
      "learning_rate": 3.675719448974246e-08,
      "loss": 0.7915,
      "num_input_tokens_seen": 166946800,
      "step": 7823
    },
    {
      "epoch": 0.9407803763602477,
      "flos": 15748312535040.0,
      "grad_norm": 3.168151624413042,
      "learning_rate": 3.6608685443207054e-08,
      "loss": 0.7871,
      "num_input_tokens_seen": 166965670,
      "step": 7824
    },
    {
      "epoch": 0.9409006192508867,
      "flos": 13386781532160.0,
      "grad_norm": 4.158458610487391,
      "learning_rate": 3.646047424157306e-08,
      "loss": 0.8592,
      "num_input_tokens_seen": 166982365,
      "step": 7825
    },
    {
      "epoch": 0.9410208621415259,
      "flos": 16612906291200.0,
      "grad_norm": 7.280600326434713,
      "learning_rate": 3.631256090732382e-08,
      "loss": 0.8624,
      "num_input_tokens_seen": 167002545,
      "step": 7826
    },
    {
      "epoch": 0.941141105032165,
      "flos": 16162210222080.0,
      "grad_norm": 2.7347120676734167,
      "learning_rate": 3.6164945462897833e-08,
      "loss": 1.0203,
      "num_input_tokens_seen": 167021555,
      "step": 7827
    },
    {
      "epoch": 0.941261347922804,
      "flos": 14698821550080.0,
      "grad_norm": 2.925748328315279,
      "learning_rate": 3.6017627930687856e-08,
      "loss": 0.9433,
      "num_input_tokens_seen": 167041100,
      "step": 7828
    },
    {
      "epoch": 0.9413815908134432,
      "flos": 13776021626880.0,
      "grad_norm": 4.113528397281844,
      "learning_rate": 3.587060833304267e-08,
      "loss": 0.961,
      "num_input_tokens_seen": 167059010,
      "step": 7829
    },
    {
      "epoch": 0.9415018337040822,
      "flos": 12390713978880.0,
      "grad_norm": 21.07645703185488,
      "learning_rate": 3.5723886692264225e-08,
      "loss": 0.828,
      "num_input_tokens_seen": 167076270,
      "step": 7830
    },
    {
      "epoch": 0.9416220765947213,
      "flos": 22695177400320.0,
      "grad_norm": 4.176723220535677,
      "learning_rate": 3.557746303061071e-08,
      "loss": 0.8034,
      "num_input_tokens_seen": 167097745,
      "step": 7831
    },
    {
      "epoch": 0.9417423194853605,
      "flos": 16715890176000.0,
      "grad_norm": 3.279832079473499,
      "learning_rate": 3.543133737029391e-08,
      "loss": 0.9191,
      "num_input_tokens_seen": 167117975,
      "step": 7832
    },
    {
      "epoch": 0.9418625623759995,
      "flos": 17005856071680.0,
      "grad_norm": 5.361531438255614,
      "learning_rate": 3.5285509733481214e-08,
      "loss": 0.8752,
      "num_input_tokens_seen": 167137420,
      "step": 7833
    },
    {
      "epoch": 0.9419828052666386,
      "flos": 12810038231040.0,
      "grad_norm": 3.4770554096676998,
      "learning_rate": 3.513998014229469e-08,
      "loss": 0.9487,
      "num_input_tokens_seen": 167156090,
      "step": 7834
    },
    {
      "epoch": 0.9421030481572777,
      "flos": 12548010393600.0,
      "grad_norm": 6.051705099642341,
      "learning_rate": 3.499474861881069e-08,
      "loss": 1.0523,
      "num_input_tokens_seen": 167173035,
      "step": 7835
    },
    {
      "epoch": 0.9422232910479168,
      "flos": 14331725045760.0,
      "grad_norm": 3.173897385061603,
      "learning_rate": 3.4849815185061136e-08,
      "loss": 0.8655,
      "num_input_tokens_seen": 167192645,
      "step": 7836
    },
    {
      "epoch": 0.9423435339385559,
      "flos": 13072893849600.0,
      "grad_norm": 3.748195213892974,
      "learning_rate": 3.470517986303223e-08,
      "loss": 0.9588,
      "num_input_tokens_seen": 167211350,
      "step": 7837
    },
    {
      "epoch": 0.942463776829195,
      "flos": 14249627750400.0,
      "grad_norm": 4.014043711866671,
      "learning_rate": 3.4560842674664856e-08,
      "loss": 0.9829,
      "num_input_tokens_seen": 167229585,
      "step": 7838
    },
    {
      "epoch": 0.9425840197198341,
      "flos": 16086520565760.0,
      "grad_norm": 3.6310663436169888,
      "learning_rate": 3.441680364185506e-08,
      "loss": 0.946,
      "num_input_tokens_seen": 167249175,
      "step": 7839
    },
    {
      "epoch": 0.9427042626104731,
      "flos": 14147471646720.0,
      "grad_norm": 6.356891718966129,
      "learning_rate": 3.427306278645314e-08,
      "loss": 0.9436,
      "num_input_tokens_seen": 167267350,
      "step": 7840
    },
    {
      "epoch": 0.9428245055011123,
      "flos": 16245411225600.0,
      "grad_norm": 11.633322741076467,
      "learning_rate": 3.4129620130264767e-08,
      "loss": 0.9108,
      "num_input_tokens_seen": 167285430,
      "step": 7841
    },
    {
      "epoch": 0.9429447483917514,
      "flos": 14875073064960.0,
      "grad_norm": 3.033877462971626,
      "learning_rate": 3.398647569505009e-08,
      "loss": 0.9774,
      "num_input_tokens_seen": 167302575,
      "step": 7842
    },
    {
      "epoch": 0.9430649912823904,
      "flos": 13361112207360.0,
      "grad_norm": 4.961641228691468,
      "learning_rate": 3.384362950252373e-08,
      "loss": 0.9277,
      "num_input_tokens_seen": 167319265,
      "step": 7843
    },
    {
      "epoch": 0.9431852341730296,
      "flos": 23215094169600.0,
      "grad_norm": 3.5744084542381964,
      "learning_rate": 3.3701081574355473e-08,
      "loss": 0.758,
      "num_input_tokens_seen": 167340945,
      "step": 7844
    },
    {
      "epoch": 0.9433054770636686,
      "flos": 47907671838720.0,
      "grad_norm": 0.6500009050729759,
      "learning_rate": 3.3558831932169796e-08,
      "loss": 0.7262,
      "num_input_tokens_seen": 167409335,
      "step": 7845
    },
    {
      "epoch": 0.9434257199543077,
      "flos": 18605041397760.0,
      "grad_norm": 6.256548693192835,
      "learning_rate": 3.341688059754588e-08,
      "loss": 1.0759,
      "num_input_tokens_seen": 167424710,
      "step": 7846
    },
    {
      "epoch": 0.9435459628449467,
      "flos": 17787218165760.0,
      "grad_norm": 3.142089654052539,
      "learning_rate": 3.327522759201762e-08,
      "loss": 0.9603,
      "num_input_tokens_seen": 167444300,
      "step": 7847
    },
    {
      "epoch": 0.9436662057355859,
      "flos": 15746748948480.0,
      "grad_norm": 3.7323638735845233,
      "learning_rate": 3.313387293707359e-08,
      "loss": 0.8523,
      "num_input_tokens_seen": 167462725,
      "step": 7848
    },
    {
      "epoch": 0.943786448626225,
      "flos": 14277044613120.0,
      "grad_norm": 3.8317960614467146,
      "learning_rate": 3.29928166541571e-08,
      "loss": 0.8674,
      "num_input_tokens_seen": 167481400,
      "step": 7849
    },
    {
      "epoch": 0.943906691516864,
      "flos": 15694766469120.0,
      "grad_norm": 3.5079810562927207,
      "learning_rate": 3.2852058764666346e-08,
      "loss": 0.9877,
      "num_input_tokens_seen": 167500220,
      "step": 7850
    },
    {
      "epoch": 0.9440269344075032,
      "flos": 25186863882240.0,
      "grad_norm": 4.0409056481228625,
      "learning_rate": 3.2711599289954264e-08,
      "loss": 0.8777,
      "num_input_tokens_seen": 167523975,
      "step": 7851
    },
    {
      "epoch": 0.9441471772981422,
      "flos": 13643689390080.0,
      "grad_norm": 2.6693507534561296,
      "learning_rate": 3.257143825132847e-08,
      "loss": 0.962,
      "num_input_tokens_seen": 167541865,
      "step": 7852
    },
    {
      "epoch": 0.9442674201887813,
      "flos": 18317282918400.0,
      "grad_norm": 4.457675041488859,
      "learning_rate": 3.243157567005106e-08,
      "loss": 0.9453,
      "num_input_tokens_seen": 167559765,
      "step": 7853
    },
    {
      "epoch": 0.9443876630794205,
      "flos": 10974831636480.0,
      "grad_norm": 4.52006106951483,
      "learning_rate": 3.2292011567339296e-08,
      "loss": 0.8261,
      "num_input_tokens_seen": 167577290,
      "step": 7854
    },
    {
      "epoch": 0.9445079059700595,
      "flos": 9448576696320.0,
      "grad_norm": 4.09196235121843,
      "learning_rate": 3.21527459643649e-08,
      "loss": 0.7337,
      "num_input_tokens_seen": 167593895,
      "step": 7855
    },
    {
      "epoch": 0.9446281488606986,
      "flos": 16821909258240.0,
      "grad_norm": 3.2822670949687875,
      "learning_rate": 3.2013778882254536e-08,
      "loss": 0.9299,
      "num_input_tokens_seen": 167612410,
      "step": 7856
    },
    {
      "epoch": 0.9447483917513377,
      "flos": 18186146365440.0,
      "grad_norm": 3.4735271104243632,
      "learning_rate": 3.1875110342088676e-08,
      "loss": 0.9455,
      "num_input_tokens_seen": 167633580,
      "step": 7857
    },
    {
      "epoch": 0.9448686346419768,
      "flos": 17451708088320.0,
      "grad_norm": 2.6017619105257306,
      "learning_rate": 3.1736740364904035e-08,
      "loss": 0.8419,
      "num_input_tokens_seen": 167653830,
      "step": 7858
    },
    {
      "epoch": 0.9449888775326158,
      "flos": 10398486896640.0,
      "grad_norm": 3.687575387992287,
      "learning_rate": 3.159866897169094e-08,
      "loss": 0.9424,
      "num_input_tokens_seen": 167671750,
      "step": 7859
    },
    {
      "epoch": 0.945109120423255,
      "flos": 10920273838080.0,
      "grad_norm": 5.970498033769062,
      "learning_rate": 3.146089618339487e-08,
      "loss": 0.9524,
      "num_input_tokens_seen": 167688325,
      "step": 7860
    },
    {
      "epoch": 0.9452293633138941,
      "flos": 18107666780160.0,
      "grad_norm": 3.378607199781087,
      "learning_rate": 3.132342202091554e-08,
      "loss": 0.8703,
      "num_input_tokens_seen": 167708270,
      "step": 7861
    },
    {
      "epoch": 0.9453496062045331,
      "flos": 15065611468800.0,
      "grad_norm": 6.559280041520253,
      "learning_rate": 3.1186246505107595e-08,
      "loss": 0.867,
      "num_input_tokens_seen": 167727130,
      "step": 7862
    },
    {
      "epoch": 0.9454698490951723,
      "flos": 14199944663040.0,
      "grad_norm": 3.8700131968586184,
      "learning_rate": 3.104936965678084e-08,
      "loss": 1.0268,
      "num_input_tokens_seen": 167745180,
      "step": 7863
    },
    {
      "epoch": 0.9455900919858113,
      "flos": 14960941363200.0,
      "grad_norm": 3.236433175959858,
      "learning_rate": 3.091279149669956e-08,
      "loss": 0.9947,
      "num_input_tokens_seen": 167763690,
      "step": 7864
    },
    {
      "epoch": 0.9457103348764504,
      "flos": 14724582850560.0,
      "grad_norm": 3.780412060918994,
      "learning_rate": 3.0776512045581624e-08,
      "loss": 0.9297,
      "num_input_tokens_seen": 167782200,
      "step": 7865
    },
    {
      "epoch": 0.9458305777670896,
      "flos": 15217910538240.0,
      "grad_norm": 3.081076044894281,
      "learning_rate": 3.0640531324101384e-08,
      "loss": 0.9611,
      "num_input_tokens_seen": 167799685,
      "step": 7866
    },
    {
      "epoch": 0.9459508206577286,
      "flos": 14200189931520.0,
      "grad_norm": 3.8869350354001786,
      "learning_rate": 3.0504849352886554e-08,
      "loss": 0.9439,
      "num_input_tokens_seen": 167817550,
      "step": 7867
    },
    {
      "epoch": 0.9460710635483677,
      "flos": 8562268569600.0,
      "grad_norm": 3.8286858595439477,
      "learning_rate": 3.036946615252023e-08,
      "loss": 0.9058,
      "num_input_tokens_seen": 167832800,
      "step": 7868
    },
    {
      "epoch": 0.9461913064390068,
      "flos": 24452057702400.0,
      "grad_norm": 3.0130522035874137,
      "learning_rate": 3.0234381743539984e-08,
      "loss": 0.8557,
      "num_input_tokens_seen": 167850135,
      "step": 7869
    },
    {
      "epoch": 0.9463115493296459,
      "flos": 13806749614080.0,
      "grad_norm": 3.1929376632860937,
      "learning_rate": 3.0099596146437863e-08,
      "loss": 0.9937,
      "num_input_tokens_seen": 167866960,
      "step": 7870
    },
    {
      "epoch": 0.946431792220285,
      "flos": 50540068454400.0,
      "grad_norm": 0.8206516109896288,
      "learning_rate": 2.996510938166086e-08,
      "loss": 0.8187,
      "num_input_tokens_seen": 167929655,
      "step": 7871
    },
    {
      "epoch": 0.9465520351109241,
      "flos": 13435330252800.0,
      "grad_norm": 4.545131703095778,
      "learning_rate": 2.983092146960997e-08,
      "loss": 0.9355,
      "num_input_tokens_seen": 167946720,
      "step": 7872
    },
    {
      "epoch": 0.9466722780015632,
      "flos": 13570973614080.0,
      "grad_norm": 3.1584939061985167,
      "learning_rate": 2.9697032430642256e-08,
      "loss": 0.9823,
      "num_input_tokens_seen": 167964655,
      "step": 7873
    },
    {
      "epoch": 0.9467925208922022,
      "flos": 12207043092480.0,
      "grad_norm": 4.136424341432762,
      "learning_rate": 2.9563442285067906e-08,
      "loss": 0.9156,
      "num_input_tokens_seen": 167981420,
      "step": 7874
    },
    {
      "epoch": 0.9469127637828414,
      "flos": 20781736488960.0,
      "grad_norm": 3.5900218699434636,
      "learning_rate": 2.943015105315294e-08,
      "loss": 0.9884,
      "num_input_tokens_seen": 168001335,
      "step": 7875
    },
    {
      "epoch": 0.9470330066734804,
      "flos": 18526009958400.0,
      "grad_norm": 5.575318557464848,
      "learning_rate": 2.929715875511718e-08,
      "loss": 0.8657,
      "num_input_tokens_seen": 168020090,
      "step": 7876
    },
    {
      "epoch": 0.9471532495641195,
      "flos": 16665072721920.0,
      "grad_norm": 3.234237741699955,
      "learning_rate": 2.9164465411135375e-08,
      "loss": 0.8929,
      "num_input_tokens_seen": 168039580,
      "step": 7877
    },
    {
      "epoch": 0.9472734924547586,
      "flos": 11185152921600.0,
      "grad_norm": 3.7463044076506535,
      "learning_rate": 2.9032071041337426e-08,
      "loss": 0.996,
      "num_input_tokens_seen": 168057535,
      "step": 7878
    },
    {
      "epoch": 0.9473937353453977,
      "flos": 7854695301120.0,
      "grad_norm": 2.866987424101057,
      "learning_rate": 2.889997566580704e-08,
      "loss": 0.9152,
      "num_input_tokens_seen": 168075410,
      "step": 7879
    },
    {
      "epoch": 0.9475139782360368,
      "flos": 18342860267520.0,
      "grad_norm": 5.650758553956046,
      "learning_rate": 2.8768179304583086e-08,
      "loss": 0.8974,
      "num_input_tokens_seen": 168097185,
      "step": 7880
    },
    {
      "epoch": 0.9476342211266758,
      "flos": 16218791485440.0,
      "grad_norm": 2.8359457388934213,
      "learning_rate": 2.8636681977659117e-08,
      "loss": 0.9302,
      "num_input_tokens_seen": 168116555,
      "step": 7881
    },
    {
      "epoch": 0.947754464017315,
      "flos": 14331295825920.0,
      "grad_norm": 3.2526230130694405,
      "learning_rate": 2.850548370498318e-08,
      "loss": 0.9738,
      "num_input_tokens_seen": 168134115,
      "step": 7882
    },
    {
      "epoch": 0.9478747069079541,
      "flos": 17582844641280.0,
      "grad_norm": 3.461574246148229,
      "learning_rate": 2.8374584506457798e-08,
      "loss": 0.894,
      "num_input_tokens_seen": 168155110,
      "step": 7883
    },
    {
      "epoch": 0.9479949497985931,
      "flos": 15170557501440.0,
      "grad_norm": 3.8312785958098816,
      "learning_rate": 2.824398440193998e-08,
      "loss": 0.8521,
      "num_input_tokens_seen": 168173630,
      "step": 7884
    },
    {
      "epoch": 0.9481151926892323,
      "flos": 12861346222080.0,
      "grad_norm": 3.8245496596968516,
      "learning_rate": 2.811368341124232e-08,
      "loss": 0.8948,
      "num_input_tokens_seen": 168192420,
      "step": 7885
    },
    {
      "epoch": 0.9482354355798713,
      "flos": 15721600819200.0,
      "grad_norm": 40.48709013714665,
      "learning_rate": 2.7983681554131222e-08,
      "loss": 0.8734,
      "num_input_tokens_seen": 168212400,
      "step": 7886
    },
    {
      "epoch": 0.9483556784705104,
      "flos": 13518868500480.0,
      "grad_norm": 3.885176800371072,
      "learning_rate": 2.7853978850327365e-08,
      "loss": 0.8665,
      "num_input_tokens_seen": 168231290,
      "step": 7887
    },
    {
      "epoch": 0.9484759213611496,
      "flos": 18344209244160.0,
      "grad_norm": 3.11305257036356,
      "learning_rate": 2.7724575319507225e-08,
      "loss": 1.0684,
      "num_input_tokens_seen": 168250720,
      "step": 7888
    },
    {
      "epoch": 0.9485961642517886,
      "flos": 14672017858560.0,
      "grad_norm": 2.7778537029910244,
      "learning_rate": 2.759547098130044e-08,
      "loss": 0.9726,
      "num_input_tokens_seen": 168269170,
      "step": 7889
    },
    {
      "epoch": 0.9487164071424277,
      "flos": 16113876111360.0,
      "grad_norm": 3.3070466839425463,
      "learning_rate": 2.746666585529267e-08,
      "loss": 0.9609,
      "num_input_tokens_seen": 168289165,
      "step": 7890
    },
    {
      "epoch": 0.9488366500330668,
      "flos": 27650244403200.0,
      "grad_norm": 3.8657595069076116,
      "learning_rate": 2.73381599610234e-08,
      "loss": 0.9276,
      "num_input_tokens_seen": 168309285,
      "step": 7891
    },
    {
      "epoch": 0.9489568929237059,
      "flos": 19863044812800.0,
      "grad_norm": 3.494507976602022,
      "learning_rate": 2.7209953317987033e-08,
      "loss": 0.9075,
      "num_input_tokens_seen": 168330045,
      "step": 7892
    },
    {
      "epoch": 0.9490771358143449,
      "flos": 23951862497280.0,
      "grad_norm": 3.9376340570789794,
      "learning_rate": 2.7082045945631793e-08,
      "loss": 0.9663,
      "num_input_tokens_seen": 168351980,
      "step": 7893
    },
    {
      "epoch": 0.9491973787049841,
      "flos": 10450561351680.0,
      "grad_norm": 4.377063591519331,
      "learning_rate": 2.6954437863361712e-08,
      "loss": 0.8758,
      "num_input_tokens_seen": 168369615,
      "step": 7894
    },
    {
      "epoch": 0.9493176215956232,
      "flos": 18023791288320.0,
      "grad_norm": 3.51839707334153,
      "learning_rate": 2.6827129090534862e-08,
      "loss": 0.9032,
      "num_input_tokens_seen": 168389635,
      "step": 7895
    },
    {
      "epoch": 0.9494378644862622,
      "flos": 14931991572480.0,
      "grad_norm": 3.329398900901002,
      "learning_rate": 2.670011964646335e-08,
      "loss": 0.9757,
      "num_input_tokens_seen": 168408035,
      "step": 7896
    },
    {
      "epoch": 0.9495581073769014,
      "flos": 10738994319360.0,
      "grad_norm": 3.533989982646913,
      "learning_rate": 2.657340955041487e-08,
      "loss": 0.8708,
      "num_input_tokens_seen": 168426530,
      "step": 7897
    },
    {
      "epoch": 0.9496783502675404,
      "flos": 20384003973120.0,
      "grad_norm": 4.538500647643681,
      "learning_rate": 2.6446998821611167e-08,
      "loss": 0.9001,
      "num_input_tokens_seen": 168446445,
      "step": 7898
    },
    {
      "epoch": 0.9497985931581795,
      "flos": 10503095685120.0,
      "grad_norm": 7.320004516766931,
      "learning_rate": 2.6320887479228228e-08,
      "loss": 0.899,
      "num_input_tokens_seen": 168462765,
      "step": 7899
    },
    {
      "epoch": 0.9499188360488187,
      "flos": 19361102069760.0,
      "grad_norm": 4.356173306943487,
      "learning_rate": 2.619507554239786e-08,
      "loss": 0.9088,
      "num_input_tokens_seen": 168481045,
      "step": 7900
    },
    {
      "epoch": 0.9500390789394577,
      "flos": 17294319697920.0,
      "grad_norm": 3.7407643074731323,
      "learning_rate": 2.606956303020502e-08,
      "loss": 0.9037,
      "num_input_tokens_seen": 168501570,
      "step": 7901
    },
    {
      "epoch": 0.9501593218300968,
      "flos": 10134864814080.0,
      "grad_norm": 5.10304918577429,
      "learning_rate": 2.5944349961690036e-08,
      "loss": 1.0316,
      "num_input_tokens_seen": 168518310,
      "step": 7902
    },
    {
      "epoch": 0.9502795647207359,
      "flos": 27651654696960.0,
      "grad_norm": 3.209641095118371,
      "learning_rate": 2.581943635584749e-08,
      "loss": 0.9142,
      "num_input_tokens_seen": 168540860,
      "step": 7903
    },
    {
      "epoch": 0.950399807611375,
      "flos": 29146476503040.0,
      "grad_norm": 3.638802657589965,
      "learning_rate": 2.569482223162689e-08,
      "loss": 0.8474,
      "num_input_tokens_seen": 168564555,
      "step": 7904
    },
    {
      "epoch": 0.950520050502014,
      "flos": 16664643502080.0,
      "grad_norm": 3.5183972088683446,
      "learning_rate": 2.5570507607932e-08,
      "loss": 0.9257,
      "num_input_tokens_seen": 168584190,
      "step": 7905
    },
    {
      "epoch": 0.9506402933926532,
      "flos": 12599103774720.0,
      "grad_norm": 7.325076512272383,
      "learning_rate": 2.54464925036213e-08,
      "loss": 0.8029,
      "num_input_tokens_seen": 168601200,
      "step": 7906
    },
    {
      "epoch": 0.9507605362832923,
      "flos": 23220122173440.0,
      "grad_norm": 3.3743296769940794,
      "learning_rate": 2.532277693750773e-08,
      "loss": 0.7945,
      "num_input_tokens_seen": 168621845,
      "step": 7907
    },
    {
      "epoch": 0.9508807791739313,
      "flos": 13905104056320.0,
      "grad_norm": 3.510466593190184,
      "learning_rate": 2.5199360928358948e-08,
      "loss": 0.9469,
      "num_input_tokens_seen": 168638800,
      "step": 7908
    },
    {
      "epoch": 0.9510010220645704,
      "flos": 15250110136320.0,
      "grad_norm": 2.749354199961457,
      "learning_rate": 2.507624449489665e-08,
      "loss": 1.0643,
      "num_input_tokens_seen": 168657150,
      "step": 7909
    },
    {
      "epoch": 0.9511212649552095,
      "flos": 13387210752000.0,
      "grad_norm": 5.054405239897328,
      "learning_rate": 2.495342765579811e-08,
      "loss": 0.843,
      "num_input_tokens_seen": 168675530,
      "step": 7910
    },
    {
      "epoch": 0.9512415078458486,
      "flos": 14774204620800.0,
      "grad_norm": 3.2823666080469414,
      "learning_rate": 2.4830910429693984e-08,
      "loss": 0.9025,
      "num_input_tokens_seen": 168695210,
      "step": 7911
    },
    {
      "epoch": 0.9513617507364877,
      "flos": 13020420833280.0,
      "grad_norm": 3.292703312394633,
      "learning_rate": 2.470869283517052e-08,
      "loss": 0.9853,
      "num_input_tokens_seen": 168712965,
      "step": 7912
    },
    {
      "epoch": 0.9514819936271268,
      "flos": 17792920657920.0,
      "grad_norm": 3.306207402032411,
      "learning_rate": 2.458677489076777e-08,
      "loss": 0.9654,
      "num_input_tokens_seen": 168733695,
      "step": 7913
    },
    {
      "epoch": 0.9516022365177659,
      "flos": 13230772776960.0,
      "grad_norm": 2.846098380827193,
      "learning_rate": 2.446515661498072e-08,
      "loss": 1.0242,
      "num_input_tokens_seen": 168752745,
      "step": 7914
    },
    {
      "epoch": 0.9517224794084049,
      "flos": 18052649103360.0,
      "grad_norm": 3.4666433724818964,
      "learning_rate": 2.434383802625861e-08,
      "loss": 0.9285,
      "num_input_tokens_seen": 168771420,
      "step": 7915
    },
    {
      "epoch": 0.9518427222990441,
      "flos": 15249313013760.0,
      "grad_norm": 3.919518967522364,
      "learning_rate": 2.4222819143005168e-08,
      "loss": 0.9436,
      "num_input_tokens_seen": 168790735,
      "step": 7916
    },
    {
      "epoch": 0.9519629651896832,
      "flos": 14935118745600.0,
      "grad_norm": 4.126554716334717,
      "learning_rate": 2.4102099983579706e-08,
      "loss": 0.9995,
      "num_input_tokens_seen": 168809605,
      "step": 7917
    },
    {
      "epoch": 0.9520832080803222,
      "flos": 15406118891520.0,
      "grad_norm": 3.2155323426712936,
      "learning_rate": 2.3981680566294236e-08,
      "loss": 0.9586,
      "num_input_tokens_seen": 168828925,
      "step": 7918
    },
    {
      "epoch": 0.9522034509709614,
      "flos": 16452114800640.0,
      "grad_norm": 2.9846718525668585,
      "learning_rate": 2.3861560909416822e-08,
      "loss": 0.9378,
      "num_input_tokens_seen": 168848195,
      "step": 7919
    },
    {
      "epoch": 0.9523236938616004,
      "flos": 17554906583040.0,
      "grad_norm": 3.8232485482466245,
      "learning_rate": 2.3741741031169325e-08,
      "loss": 1.019,
      "num_input_tokens_seen": 168867485,
      "step": 7920
    },
    {
      "epoch": 0.9524439367522395,
      "flos": 16112373841920.0,
      "grad_norm": 2.776010786293907,
      "learning_rate": 2.3622220949728544e-08,
      "loss": 0.8871,
      "num_input_tokens_seen": 168886090,
      "step": 7921
    },
    {
      "epoch": 0.9525641796428787,
      "flos": 24293228359680.0,
      "grad_norm": 5.131604474888369,
      "learning_rate": 2.3503000683225526e-08,
      "loss": 0.8068,
      "num_input_tokens_seen": 168903525,
      "step": 7922
    },
    {
      "epoch": 0.9526844225335177,
      "flos": 11839915929600.0,
      "grad_norm": 3.9919585220154405,
      "learning_rate": 2.3384080249745585e-08,
      "loss": 1.0274,
      "num_input_tokens_seen": 168921135,
      "step": 7923
    },
    {
      "epoch": 0.9528046654241568,
      "flos": 26365989150720.0,
      "grad_norm": 6.016643963011001,
      "learning_rate": 2.3265459667329178e-08,
      "loss": 1.0151,
      "num_input_tokens_seen": 168940345,
      "step": 7924
    },
    {
      "epoch": 0.9529249083147959,
      "flos": 12937403781120.0,
      "grad_norm": 5.212381829036035,
      "learning_rate": 2.31471389539708e-08,
      "loss": 1.0495,
      "num_input_tokens_seen": 168957190,
      "step": 7925
    },
    {
      "epoch": 0.953045151205435,
      "flos": 20414823936000.0,
      "grad_norm": 3.4052283478517538,
      "learning_rate": 2.3029118127619872e-08,
      "loss": 0.9123,
      "num_input_tokens_seen": 168976625,
      "step": 7926
    },
    {
      "epoch": 0.953165394096074,
      "flos": 15511064924160.0,
      "grad_norm": 3.5785804691858583,
      "learning_rate": 2.2911397206179628e-08,
      "loss": 1.0582,
      "num_input_tokens_seen": 168993095,
      "step": 7927
    },
    {
      "epoch": 0.9532856369867132,
      "flos": 14174275338240.0,
      "grad_norm": 3.951607635348879,
      "learning_rate": 2.279397620750845e-08,
      "loss": 0.8136,
      "num_input_tokens_seen": 169011860,
      "step": 7928
    },
    {
      "epoch": 0.9534058798773523,
      "flos": 10634446848000.0,
      "grad_norm": 6.232362197595044,
      "learning_rate": 2.2676855149419195e-08,
      "loss": 0.9675,
      "num_input_tokens_seen": 169028750,
      "step": 7929
    },
    {
      "epoch": 0.9535261227679913,
      "flos": 12469285539840.0,
      "grad_norm": 7.841826792050653,
      "learning_rate": 2.2560034049678988e-08,
      "loss": 0.9364,
      "num_input_tokens_seen": 169042820,
      "step": 7930
    },
    {
      "epoch": 0.9536463656586305,
      "flos": 16450121994240.0,
      "grad_norm": 5.535853824526829,
      "learning_rate": 2.2443512926008988e-08,
      "loss": 0.9451,
      "num_input_tokens_seen": 169061870,
      "step": 7931
    },
    {
      "epoch": 0.9537666085492695,
      "flos": 13204030402560.0,
      "grad_norm": 4.371883917493363,
      "learning_rate": 2.2327291796085946e-08,
      "loss": 0.8957,
      "num_input_tokens_seen": 169079950,
      "step": 7932
    },
    {
      "epoch": 0.9538868514399086,
      "flos": 13465782312960.0,
      "grad_norm": 5.737180267932829,
      "learning_rate": 2.2211370677540197e-08,
      "loss": 0.9675,
      "num_input_tokens_seen": 169096195,
      "step": 7933
    },
    {
      "epoch": 0.9540070943305478,
      "flos": 11892787507200.0,
      "grad_norm": 4.42325249773189,
      "learning_rate": 2.2095749587957012e-08,
      "loss": 0.9621,
      "num_input_tokens_seen": 169113820,
      "step": 7934
    },
    {
      "epoch": 0.9541273372211868,
      "flos": 14304890695680.0,
      "grad_norm": 2.775501096004709,
      "learning_rate": 2.1980428544876138e-08,
      "loss": 0.8755,
      "num_input_tokens_seen": 169132180,
      "step": 7935
    },
    {
      "epoch": 0.9542475801118259,
      "flos": 19025775943680.0,
      "grad_norm": 3.538022054695148,
      "learning_rate": 2.1865407565791584e-08,
      "loss": 0.9283,
      "num_input_tokens_seen": 169153470,
      "step": 7936
    },
    {
      "epoch": 0.954367823002465,
      "flos": 16585949306880.0,
      "grad_norm": 5.005965488840676,
      "learning_rate": 2.175068666815183e-08,
      "loss": 0.9591,
      "num_input_tokens_seen": 169174030,
      "step": 7937
    },
    {
      "epoch": 0.9544880658931041,
      "flos": 10528765009920.0,
      "grad_norm": 3.4947241119510912,
      "learning_rate": 2.163626586935985e-08,
      "loss": 0.9754,
      "num_input_tokens_seen": 169190290,
      "step": 7938
    },
    {
      "epoch": 0.9546083087837431,
      "flos": 20729171496960.0,
      "grad_norm": 4.903076117170383,
      "learning_rate": 2.1522145186773755e-08,
      "loss": 0.8108,
      "num_input_tokens_seen": 169208930,
      "step": 7939
    },
    {
      "epoch": 0.9547285516743822,
      "flos": 15013138452480.0,
      "grad_norm": 3.7721230565181227,
      "learning_rate": 2.140832463770481e-08,
      "loss": 1.0425,
      "num_input_tokens_seen": 169227845,
      "step": 7940
    },
    {
      "epoch": 0.9548487945650214,
      "flos": 19575102382080.0,
      "grad_norm": 3.214090177855425,
      "learning_rate": 2.129480423941987e-08,
      "loss": 0.9421,
      "num_input_tokens_seen": 169244235,
      "step": 7941
    },
    {
      "epoch": 0.9549690374556604,
      "flos": 15825841704960.0,
      "grad_norm": 2.6087510510795755,
      "learning_rate": 2.1181584009140052e-08,
      "loss": 1.0054,
      "num_input_tokens_seen": 169263495,
      "step": 7942
    },
    {
      "epoch": 0.9550892803462995,
      "flos": 12462877900800.0,
      "grad_norm": 5.365144432377701,
      "learning_rate": 2.10686639640405e-08,
      "loss": 1.023,
      "num_input_tokens_seen": 169277305,
      "step": 7943
    },
    {
      "epoch": 0.9552095232369386,
      "flos": 17320694169600.0,
      "grad_norm": 3.316006513932,
      "learning_rate": 2.0956044121251294e-08,
      "loss": 1.0043,
      "num_input_tokens_seen": 169295810,
      "step": 7944
    },
    {
      "epoch": 0.9553297661275777,
      "flos": 16166165176320.0,
      "grad_norm": 3.0428002914398484,
      "learning_rate": 2.084372449785654e-08,
      "loss": 1.0002,
      "num_input_tokens_seen": 169315365,
      "step": 7945
    },
    {
      "epoch": 0.9554500090182168,
      "flos": 10896168099840.0,
      "grad_norm": 3.531058787342398,
      "learning_rate": 2.0731705110895282e-08,
      "loss": 0.8794,
      "num_input_tokens_seen": 169332575,
      "step": 7946
    },
    {
      "epoch": 0.9555702519088559,
      "flos": 16717208494080.0,
      "grad_norm": 3.8721673517947512,
      "learning_rate": 2.0619985977360587e-08,
      "loss": 1.0544,
      "num_input_tokens_seen": 169350615,
      "step": 7947
    },
    {
      "epoch": 0.955690494799495,
      "flos": 16320334417920.0,
      "grad_norm": 6.755064583108933,
      "learning_rate": 2.0508567114200237e-08,
      "loss": 0.9743,
      "num_input_tokens_seen": 169370250,
      "step": 7948
    },
    {
      "epoch": 0.955810737690134,
      "flos": 18526684446720.0,
      "grad_norm": 3.79677879933749,
      "learning_rate": 2.0397448538316485e-08,
      "loss": 0.9774,
      "num_input_tokens_seen": 169391010,
      "step": 7949
    },
    {
      "epoch": 0.9559309805807732,
      "flos": 14802939801600.0,
      "grad_norm": 3.202102274560512,
      "learning_rate": 2.028663026656563e-08,
      "loss": 0.8504,
      "num_input_tokens_seen": 169409585,
      "step": 7950
    },
    {
      "epoch": 0.9560512234714122,
      "flos": 15326137036800.0,
      "grad_norm": 3.1706979756949685,
      "learning_rate": 2.0176112315758885e-08,
      "loss": 0.915,
      "num_input_tokens_seen": 169427095,
      "step": 7951
    },
    {
      "epoch": 0.9561714663620513,
      "flos": 12358667673600.0,
      "grad_norm": 3.763468397863049,
      "learning_rate": 2.0065894702661957e-08,
      "loss": 0.8859,
      "num_input_tokens_seen": 169443490,
      "step": 7952
    },
    {
      "epoch": 0.9562917092526905,
      "flos": 18574252093440.0,
      "grad_norm": 3.9774104265428623,
      "learning_rate": 1.9955977443994577e-08,
      "loss": 0.9715,
      "num_input_tokens_seen": 169463200,
      "step": 7953
    },
    {
      "epoch": 0.9564119521433295,
      "flos": 17136287477760.0,
      "grad_norm": 5.123315673943296,
      "learning_rate": 1.9846360556430965e-08,
      "loss": 0.8177,
      "num_input_tokens_seen": 169481220,
      "step": 7954
    },
    {
      "epoch": 0.9565321950339686,
      "flos": 22821899120640.0,
      "grad_norm": 10.958426548859915,
      "learning_rate": 1.973704405660004e-08,
      "loss": 0.7988,
      "num_input_tokens_seen": 169502055,
      "step": 7955
    },
    {
      "epoch": 0.9566524379246077,
      "flos": 16771490365440.0,
      "grad_norm": 2.166052388875897,
      "learning_rate": 1.9628027961085203e-08,
      "loss": 0.9645,
      "num_input_tokens_seen": 169525005,
      "step": 7956
    },
    {
      "epoch": 0.9567726808152468,
      "flos": 27176546304000.0,
      "grad_norm": 2.436963731510736,
      "learning_rate": 1.9519312286423894e-08,
      "loss": 1.027,
      "num_input_tokens_seen": 169547920,
      "step": 7957
    },
    {
      "epoch": 0.9568929237058859,
      "flos": 16163773808640.0,
      "grad_norm": 2.9250974785627637,
      "learning_rate": 1.9410897049108255e-08,
      "loss": 0.9685,
      "num_input_tokens_seen": 169566920,
      "step": 7958
    },
    {
      "epoch": 0.957013166596525,
      "flos": 16952984494080.0,
      "grad_norm": 3.824777938280946,
      "learning_rate": 1.9302782265584905e-08,
      "loss": 1.0925,
      "num_input_tokens_seen": 169587305,
      "step": 7959
    },
    {
      "epoch": 0.9571334094871641,
      "flos": 12495506718720.0,
      "grad_norm": 7.004188919107678,
      "learning_rate": 1.9194967952254282e-08,
      "loss": 1.0466,
      "num_input_tokens_seen": 169605600,
      "step": 7960
    },
    {
      "epoch": 0.9572536523778031,
      "flos": 10863324672000.0,
      "grad_norm": 4.3937015275736835,
      "learning_rate": 1.9087454125472635e-08,
      "loss": 0.9877,
      "num_input_tokens_seen": 169619795,
      "step": 7961
    },
    {
      "epoch": 0.9573738952684423,
      "flos": 17763419013120.0,
      "grad_norm": 3.375656064163734,
      "learning_rate": 1.8980240801548696e-08,
      "loss": 0.9706,
      "num_input_tokens_seen": 169638705,
      "step": 7962
    },
    {
      "epoch": 0.9574941381590814,
      "flos": 18337678970880.0,
      "grad_norm": 2.5429169024032374,
      "learning_rate": 1.8873327996747458e-08,
      "loss": 0.9291,
      "num_input_tokens_seen": 169656925,
      "step": 7963
    },
    {
      "epoch": 0.9576143810497204,
      "flos": 23037125775360.0,
      "grad_norm": 5.13562659194994,
      "learning_rate": 1.8766715727287053e-08,
      "loss": 0.8562,
      "num_input_tokens_seen": 169678350,
      "step": 7964
    },
    {
      "epoch": 0.9577346239403596,
      "flos": 19758803927040.0,
      "grad_norm": 2.502047749143761,
      "learning_rate": 1.8660404009340546e-08,
      "loss": 0.9813,
      "num_input_tokens_seen": 169698520,
      "step": 7965
    },
    {
      "epoch": 0.9578548668309986,
      "flos": 41122986209280.0,
      "grad_norm": 0.9129129337691214,
      "learning_rate": 1.8554392859035485e-08,
      "loss": 0.8579,
      "num_input_tokens_seen": 169755990,
      "step": 7966
    },
    {
      "epoch": 0.9579751097216377,
      "flos": 14016856289280.0,
      "grad_norm": 3.1171636365392965,
      "learning_rate": 1.8448682292453444e-08,
      "loss": 0.9837,
      "num_input_tokens_seen": 169774785,
      "step": 7967
    },
    {
      "epoch": 0.9580953526122769,
      "flos": 12784491540480.0,
      "grad_norm": 3.6457861470393254,
      "learning_rate": 1.8343272325631154e-08,
      "loss": 0.8542,
      "num_input_tokens_seen": 169793450,
      "step": 7968
    },
    {
      "epoch": 0.9582155955029159,
      "flos": 17259698073600.0,
      "grad_norm": 4.506671199081519,
      "learning_rate": 1.8238162974558492e-08,
      "loss": 0.9844,
      "num_input_tokens_seen": 169807100,
      "step": 7969
    },
    {
      "epoch": 0.958335838393555,
      "flos": 16192355696640.0,
      "grad_norm": 3.384690818246126,
      "learning_rate": 1.8133354255181144e-08,
      "loss": 0.9396,
      "num_input_tokens_seen": 169827135,
      "step": 7970
    },
    {
      "epoch": 0.958456081284194,
      "flos": 11972708044800.0,
      "grad_norm": 3.393967909868591,
      "learning_rate": 1.802884618339795e-08,
      "loss": 0.9344,
      "num_input_tokens_seen": 169845660,
      "step": 7971
    },
    {
      "epoch": 0.9585763241748332,
      "flos": 14174183362560.0,
      "grad_norm": 55.59752437651226,
      "learning_rate": 1.7924638775062894e-08,
      "loss": 0.9967,
      "num_input_tokens_seen": 169864500,
      "step": 7972
    },
    {
      "epoch": 0.9586965670654722,
      "flos": 15194908508160.0,
      "grad_norm": 2.9417170022290744,
      "learning_rate": 1.7820732045984444e-08,
      "loss": 1.0054,
      "num_input_tokens_seen": 169884365,
      "step": 7973
    },
    {
      "epoch": 0.9588168099561113,
      "flos": 15223674347520.0,
      "grad_norm": 3.2533112310482153,
      "learning_rate": 1.7717126011924655e-08,
      "loss": 0.9333,
      "num_input_tokens_seen": 169905670,
      "step": 7974
    },
    {
      "epoch": 0.9589370528467505,
      "flos": 8272701235200.0,
      "grad_norm": 10.018591193108385,
      "learning_rate": 1.7613820688600957e-08,
      "loss": 0.9498,
      "num_input_tokens_seen": 169921295,
      "step": 7975
    },
    {
      "epoch": 0.9590572957373895,
      "flos": 16665593917440.0,
      "grad_norm": 3.451149906418831,
      "learning_rate": 1.7510816091684588e-08,
      "loss": 0.9845,
      "num_input_tokens_seen": 169940940,
      "step": 7976
    },
    {
      "epoch": 0.9591775386280286,
      "flos": 16009114030080.0,
      "grad_norm": 3.895166159771807,
      "learning_rate": 1.740811223680083e-08,
      "loss": 0.9695,
      "num_input_tokens_seen": 169957515,
      "step": 7977
    },
    {
      "epoch": 0.9592977815186677,
      "flos": 12889161646080.0,
      "grad_norm": 3.713005822655643,
      "learning_rate": 1.7305709139530334e-08,
      "loss": 0.9216,
      "num_input_tokens_seen": 169976015,
      "step": 7978
    },
    {
      "epoch": 0.9594180244093068,
      "flos": 11702954250240.0,
      "grad_norm": 42.89542465175846,
      "learning_rate": 1.7203606815407334e-08,
      "loss": 0.9393,
      "num_input_tokens_seen": 169990330,
      "step": 7979
    },
    {
      "epoch": 0.9595382672999458,
      "flos": 14589981880320.0,
      "grad_norm": 3.2014834780084276,
      "learning_rate": 1.7101805279920557e-08,
      "loss": 0.9872,
      "num_input_tokens_seen": 170008210,
      "step": 7980
    },
    {
      "epoch": 0.959658510190585,
      "flos": 16088237445120.0,
      "grad_norm": 3.029902522467397,
      "learning_rate": 1.7000304548513643e-08,
      "loss": 1.007,
      "num_input_tokens_seen": 170028035,
      "step": 7981
    },
    {
      "epoch": 0.9597787530812241,
      "flos": 13570636369920.0,
      "grad_norm": 3.306112212282741,
      "learning_rate": 1.6899104636583394e-08,
      "loss": 1.0133,
      "num_input_tokens_seen": 170045805,
      "step": 7982
    },
    {
      "epoch": 0.9598989959718631,
      "flos": 43013363773440.0,
      "grad_norm": 0.7217426195180213,
      "learning_rate": 1.6798205559482638e-08,
      "loss": 0.8408,
      "num_input_tokens_seen": 170107905,
      "step": 7983
    },
    {
      "epoch": 0.9600192388625023,
      "flos": 14829467566080.0,
      "grad_norm": 2.964598795103677,
      "learning_rate": 1.669760733251713e-08,
      "loss": 0.9514,
      "num_input_tokens_seen": 170126500,
      "step": 7984
    },
    {
      "epoch": 0.9601394817531413,
      "flos": 14512054149120.0,
      "grad_norm": 2.7875474934316737,
      "learning_rate": 1.659730997094755e-08,
      "loss": 1.0255,
      "num_input_tokens_seen": 170144710,
      "step": 7985
    },
    {
      "epoch": 0.9602597246437804,
      "flos": 15275319582720.0,
      "grad_norm": 3.605500094028504,
      "learning_rate": 1.6497313489989283e-08,
      "loss": 0.8154,
      "num_input_tokens_seen": 170164255,
      "step": 7986
    },
    {
      "epoch": 0.9603799675344196,
      "flos": 21332289269760.0,
      "grad_norm": 4.448588741131251,
      "learning_rate": 1.639761790481131e-08,
      "loss": 0.8766,
      "num_input_tokens_seen": 170184855,
      "step": 7987
    },
    {
      "epoch": 0.9605002104250586,
      "flos": 19942689423360.0,
      "grad_norm": 4.784824406542045,
      "learning_rate": 1.6298223230537754e-08,
      "loss": 0.985,
      "num_input_tokens_seen": 170202375,
      "step": 7988
    },
    {
      "epoch": 0.9606204533156977,
      "flos": 25396725288960.0,
      "grad_norm": 3.9933444920183874,
      "learning_rate": 1.619912948224611e-08,
      "loss": 0.8809,
      "num_input_tokens_seen": 170223300,
      "step": 7989
    },
    {
      "epoch": 0.9607406962063368,
      "flos": 18916660346880.0,
      "grad_norm": 4.201974931238156,
      "learning_rate": 1.6100336674969682e-08,
      "loss": 0.7863,
      "num_input_tokens_seen": 170241860,
      "step": 7990
    },
    {
      "epoch": 0.9608609390969759,
      "flos": 18023147458560.0,
      "grad_norm": 2.8535176302611025,
      "learning_rate": 1.600184482369449e-08,
      "loss": 0.9518,
      "num_input_tokens_seen": 170261495,
      "step": 7991
    },
    {
      "epoch": 0.960981181987615,
      "flos": 14960481484800.0,
      "grad_norm": 4.737439061077396,
      "learning_rate": 1.5903653943362126e-08,
      "loss": 1.0834,
      "num_input_tokens_seen": 170280210,
      "step": 7992
    },
    {
      "epoch": 0.9611014248782541,
      "flos": 12627930931200.0,
      "grad_norm": 4.2043600430484105,
      "learning_rate": 1.580576404886802e-08,
      "loss": 0.953,
      "num_input_tokens_seen": 170298460,
      "step": 7993
    },
    {
      "epoch": 0.9612216677688932,
      "flos": 13728362004480.0,
      "grad_norm": 5.460965965535164,
      "learning_rate": 1.570817515506162e-08,
      "loss": 0.995,
      "num_input_tokens_seen": 170316870,
      "step": 7994
    },
    {
      "epoch": 0.9613419106595322,
      "flos": 11179082526720.0,
      "grad_norm": 7.649583304033089,
      "learning_rate": 1.561088727674753e-08,
      "loss": 1.0052,
      "num_input_tokens_seen": 170330800,
      "step": 7995
    },
    {
      "epoch": 0.9614621535501714,
      "flos": 18290356592640.0,
      "grad_norm": 10.586600532741553,
      "learning_rate": 1.551390042868417e-08,
      "loss": 0.891,
      "num_input_tokens_seen": 170352290,
      "step": 7996
    },
    {
      "epoch": 0.9615823964408104,
      "flos": 12624343879680.0,
      "grad_norm": 3.432804021186758,
      "learning_rate": 1.5417214625584207e-08,
      "loss": 0.9111,
      "num_input_tokens_seen": 170369665,
      "step": 7997
    },
    {
      "epoch": 0.9617026393314495,
      "flos": 14328229969920.0,
      "grad_norm": 5.210828788572732,
      "learning_rate": 1.5320829882114806e-08,
      "loss": 1.0441,
      "num_input_tokens_seen": 170387460,
      "step": 7998
    },
    {
      "epoch": 0.9618228822220887,
      "flos": 14384106086400.0,
      "grad_norm": 5.257471308754612,
      "learning_rate": 1.5224746212897378e-08,
      "loss": 0.9708,
      "num_input_tokens_seen": 170406475,
      "step": 7999
    },
    {
      "epoch": 0.9619431251127277,
      "flos": 14933125939200.0,
      "grad_norm": 2.391609338301467,
      "learning_rate": 1.512896363250804e-08,
      "loss": 0.9645,
      "num_input_tokens_seen": 170426305,
      "step": 8000
    },
    {
      "epoch": 0.9620633680033668,
      "flos": 15904505241600.0,
      "grad_norm": 3.70887548076882,
      "learning_rate": 1.503348215547673e-08,
      "loss": 0.9475,
      "num_input_tokens_seen": 170447115,
      "step": 8001
    },
    {
      "epoch": 0.962183610894006,
      "flos": 13092860682240.0,
      "grad_norm": 2.9589474716655046,
      "learning_rate": 1.4938301796288078e-08,
      "loss": 0.9958,
      "num_input_tokens_seen": 170463405,
      "step": 8002
    },
    {
      "epoch": 0.962303853784645,
      "flos": 13066639503360.0,
      "grad_norm": 10.418157323354267,
      "learning_rate": 1.4843422569380537e-08,
      "loss": 1.0076,
      "num_input_tokens_seen": 170479880,
      "step": 8003
    },
    {
      "epoch": 0.9624240966752841,
      "flos": 18784604037120.0,
      "grad_norm": 4.085218785752584,
      "learning_rate": 1.4748844489147483e-08,
      "loss": 1.014,
      "num_input_tokens_seen": 170496590,
      "step": 8004
    },
    {
      "epoch": 0.9625443395659231,
      "flos": 10345523343360.0,
      "grad_norm": 3.690719750566504,
      "learning_rate": 1.4654567569936326e-08,
      "loss": 0.8952,
      "num_input_tokens_seen": 170513885,
      "step": 8005
    },
    {
      "epoch": 0.9626645824565623,
      "flos": 13018765271040.0,
      "grad_norm": 2.9725394708500876,
      "learning_rate": 1.456059182604874e-08,
      "loss": 1.0093,
      "num_input_tokens_seen": 170532410,
      "step": 8006
    },
    {
      "epoch": 0.9627848253472013,
      "flos": 11734418042880.0,
      "grad_norm": 3.023731779222599,
      "learning_rate": 1.4466917271740653e-08,
      "loss": 0.9601,
      "num_input_tokens_seen": 170550330,
      "step": 8007
    },
    {
      "epoch": 0.9629050682378404,
      "flos": 14829528883200.0,
      "grad_norm": 3.2939674609672656,
      "learning_rate": 1.4373543921222697e-08,
      "loss": 0.8603,
      "num_input_tokens_seen": 170569635,
      "step": 8008
    },
    {
      "epoch": 0.9630253111284796,
      "flos": 12049930629120.0,
      "grad_norm": 3.2995022253822976,
      "learning_rate": 1.428047178865932e-08,
      "loss": 0.9672,
      "num_input_tokens_seen": 170586145,
      "step": 8009
    },
    {
      "epoch": 0.9631455540191186,
      "flos": 14435659345920.0,
      "grad_norm": 4.868932812139224,
      "learning_rate": 1.4187700888169451e-08,
      "loss": 0.9275,
      "num_input_tokens_seen": 170605040,
      "step": 8010
    },
    {
      "epoch": 0.9632657969097577,
      "flos": 47226626334720.0,
      "grad_norm": 0.8451502448595211,
      "learning_rate": 1.40952312338265e-08,
      "loss": 0.8374,
      "num_input_tokens_seen": 170669405,
      "step": 8011
    },
    {
      "epoch": 0.9633860398003968,
      "flos": 31741913333760.0,
      "grad_norm": 3.273332762578952,
      "learning_rate": 1.4003062839657909e-08,
      "loss": 0.8723,
      "num_input_tokens_seen": 170691605,
      "step": 8012
    },
    {
      "epoch": 0.9635062826910359,
      "flos": 17661784104960.0,
      "grad_norm": 3.3277593420753147,
      "learning_rate": 1.391119571964583e-08,
      "loss": 0.9909,
      "num_input_tokens_seen": 170712265,
      "step": 8013
    },
    {
      "epoch": 0.9636265255816749,
      "flos": 11289669734400.0,
      "grad_norm": 5.212242958079826,
      "learning_rate": 1.3819629887726225e-08,
      "loss": 0.9196,
      "num_input_tokens_seen": 170730075,
      "step": 8014
    },
    {
      "epoch": 0.9637467684723141,
      "flos": 16062016266240.0,
      "grad_norm": 2.9769656891977516,
      "learning_rate": 1.3728365357789317e-08,
      "loss": 0.9591,
      "num_input_tokens_seen": 170749160,
      "step": 8015
    },
    {
      "epoch": 0.9638670113629532,
      "flos": 12442635141120.0,
      "grad_norm": 5.129200028477044,
      "learning_rate": 1.3637402143680254e-08,
      "loss": 0.9474,
      "num_input_tokens_seen": 170763780,
      "step": 8016
    },
    {
      "epoch": 0.9639872542535922,
      "flos": 39402505728000.0,
      "grad_norm": 0.7656016054277119,
      "learning_rate": 1.3546740259197998e-08,
      "loss": 0.7623,
      "num_input_tokens_seen": 170816310,
      "step": 8017
    },
    {
      "epoch": 0.9641074971442314,
      "flos": 17163581706240.0,
      "grad_norm": 15.968955439908926,
      "learning_rate": 1.3456379718095989e-08,
      "loss": 0.8769,
      "num_input_tokens_seen": 170836445,
      "step": 8018
    },
    {
      "epoch": 0.9642277400348704,
      "flos": 47793282969600.0,
      "grad_norm": 0.8710065036754233,
      "learning_rate": 1.3366320534081487e-08,
      "loss": 0.8409,
      "num_input_tokens_seen": 170898845,
      "step": 8019
    },
    {
      "epoch": 0.9643479829255095,
      "flos": 22040322416640.0,
      "grad_norm": 3.3306996590522893,
      "learning_rate": 1.3276562720816675e-08,
      "loss": 0.938,
      "num_input_tokens_seen": 170920075,
      "step": 8020
    },
    {
      "epoch": 0.9644682258161487,
      "flos": 14226901647360.0,
      "grad_norm": 3.272269903359833,
      "learning_rate": 1.3187106291917549e-08,
      "loss": 1.015,
      "num_input_tokens_seen": 170936785,
      "step": 8021
    },
    {
      "epoch": 0.9645884687067877,
      "flos": 15039267655680.0,
      "grad_norm": 5.091026092708202,
      "learning_rate": 1.309795126095503e-08,
      "loss": 0.9007,
      "num_input_tokens_seen": 170954805,
      "step": 8022
    },
    {
      "epoch": 0.9647087115974268,
      "flos": 13433674690560.0,
      "grad_norm": 4.562436874161975,
      "learning_rate": 1.3009097641453192e-08,
      "loss": 0.9907,
      "num_input_tokens_seen": 170972375,
      "step": 8023
    },
    {
      "epoch": 0.9648289544880659,
      "flos": 11709177937920.0,
      "grad_norm": 5.282076799128317,
      "learning_rate": 1.2920545446891474e-08,
      "loss": 0.9489,
      "num_input_tokens_seen": 170988815,
      "step": 8024
    },
    {
      "epoch": 0.964949197378705,
      "flos": 17107736248320.0,
      "grad_norm": 16.659255452640377,
      "learning_rate": 1.2832294690703127e-08,
      "loss": 0.8958,
      "num_input_tokens_seen": 171007510,
      "step": 8025
    },
    {
      "epoch": 0.965069440269344,
      "flos": 16610606899200.0,
      "grad_norm": 5.698093148891765,
      "learning_rate": 1.2744345386275668e-08,
      "loss": 0.9655,
      "num_input_tokens_seen": 171026770,
      "step": 8026
    },
    {
      "epoch": 0.9651896831599832,
      "flos": 18055009812480.0,
      "grad_norm": 2.9767182076645744,
      "learning_rate": 1.265669754695109e-08,
      "loss": 0.9665,
      "num_input_tokens_seen": 171046060,
      "step": 8027
    },
    {
      "epoch": 0.9653099260506223,
      "flos": 15825565777920.0,
      "grad_norm": 2.9129309208566236,
      "learning_rate": 1.2569351186025201e-08,
      "loss": 1.0139,
      "num_input_tokens_seen": 171064235,
      "step": 8028
    },
    {
      "epoch": 0.9654301689412613,
      "flos": 19051016048640.0,
      "grad_norm": 3.7978444058896956,
      "learning_rate": 1.2482306316748737e-08,
      "loss": 0.9499,
      "num_input_tokens_seen": 171084400,
      "step": 8029
    },
    {
      "epoch": 0.9655504118319005,
      "flos": 12331618713600.0,
      "grad_norm": 4.030471371706963,
      "learning_rate": 1.2395562952326021e-08,
      "loss": 0.9699,
      "num_input_tokens_seen": 171101280,
      "step": 8030
    },
    {
      "epoch": 0.9656706547225395,
      "flos": 15720711720960.0,
      "grad_norm": 4.743097610746416,
      "learning_rate": 1.2309121105916309e-08,
      "loss": 1.004,
      "num_input_tokens_seen": 171119290,
      "step": 8031
    },
    {
      "epoch": 0.9657908976131786,
      "flos": 26445051248640.0,
      "grad_norm": 9.866821378921134,
      "learning_rate": 1.222298079063222e-08,
      "loss": 0.8826,
      "num_input_tokens_seen": 171140150,
      "step": 8032
    },
    {
      "epoch": 0.9659111405038178,
      "flos": 17347068641280.0,
      "grad_norm": 7.477595370547591,
      "learning_rate": 1.2137142019541524e-08,
      "loss": 0.9121,
      "num_input_tokens_seen": 171158425,
      "step": 8033
    },
    {
      "epoch": 0.9660313833944568,
      "flos": 17792276828160.0,
      "grad_norm": 3.3575130536811986,
      "learning_rate": 1.2051604805666027e-08,
      "loss": 0.9317,
      "num_input_tokens_seen": 171175270,
      "step": 8034
    },
    {
      "epoch": 0.9661516262850959,
      "flos": 8300056780800.0,
      "grad_norm": 3.050952550008332,
      "learning_rate": 1.196636916198135e-08,
      "loss": 0.965,
      "num_input_tokens_seen": 171192530,
      "step": 8035
    },
    {
      "epoch": 0.9662718691757349,
      "flos": 14226503086080.0,
      "grad_norm": 6.1537073761624015,
      "learning_rate": 1.1881435101418036e-08,
      "loss": 0.9695,
      "num_input_tokens_seen": 171211665,
      "step": 8036
    },
    {
      "epoch": 0.9663921120663741,
      "flos": 48713844817920.0,
      "grad_norm": 0.7757406525683372,
      "learning_rate": 1.1796802636860003e-08,
      "loss": 0.895,
      "num_input_tokens_seen": 171279915,
      "step": 8037
    },
    {
      "epoch": 0.9665123549570132,
      "flos": 18736269926400.0,
      "grad_norm": 3.683155409944537,
      "learning_rate": 1.1712471781146316e-08,
      "loss": 0.9117,
      "num_input_tokens_seen": 171298970,
      "step": 8038
    },
    {
      "epoch": 0.9666325978476522,
      "flos": 31397542932480.0,
      "grad_norm": 4.57053242993803,
      "learning_rate": 1.1628442547069628e-08,
      "loss": 0.8704,
      "num_input_tokens_seen": 171320890,
      "step": 8039
    },
    {
      "epoch": 0.9667528407382914,
      "flos": 15301356810240.0,
      "grad_norm": 4.8730302401394425,
      "learning_rate": 1.1544714947377521e-08,
      "loss": 0.9576,
      "num_input_tokens_seen": 171338295,
      "step": 8040
    },
    {
      "epoch": 0.9668730836289304,
      "flos": 16979604234240.0,
      "grad_norm": 3.2166498239764754,
      "learning_rate": 1.1461288994770945e-08,
      "loss": 0.8916,
      "num_input_tokens_seen": 171357090,
      "step": 8041
    },
    {
      "epoch": 0.9669933265195695,
      "flos": 20152121610240.0,
      "grad_norm": 3.017496901718357,
      "learning_rate": 1.1378164701906002e-08,
      "loss": 0.9576,
      "num_input_tokens_seen": 171378575,
      "step": 8042
    },
    {
      "epoch": 0.9671135694102087,
      "flos": 15956211793920.0,
      "grad_norm": 3.4858143660317875,
      "learning_rate": 1.1295342081392156e-08,
      "loss": 0.8573,
      "num_input_tokens_seen": 171397655,
      "step": 8043
    },
    {
      "epoch": 0.9672338123008477,
      "flos": 14304062914560.0,
      "grad_norm": 2.6193291399287775,
      "learning_rate": 1.1212821145793804e-08,
      "loss": 0.8834,
      "num_input_tokens_seen": 171416990,
      "step": 8044
    },
    {
      "epoch": 0.9673540551914868,
      "flos": 12019202641920.0,
      "grad_norm": 4.036283204576203,
      "learning_rate": 1.1130601907629156e-08,
      "loss": 0.9754,
      "num_input_tokens_seen": 171434440,
      "step": 8045
    },
    {
      "epoch": 0.9674742980821259,
      "flos": 44304179957760.0,
      "grad_norm": 0.8278570641416862,
      "learning_rate": 1.1048684379370899e-08,
      "loss": 0.86,
      "num_input_tokens_seen": 171494845,
      "step": 8046
    },
    {
      "epoch": 0.967594540972765,
      "flos": 13256319467520.0,
      "grad_norm": 3.1518551283525182,
      "learning_rate": 1.0967068573445759e-08,
      "loss": 0.9311,
      "num_input_tokens_seen": 171512050,
      "step": 8047
    },
    {
      "epoch": 0.967714783863404,
      "flos": 14750865346560.0,
      "grad_norm": 5.270999508777553,
      "learning_rate": 1.0885754502234945e-08,
      "loss": 0.8351,
      "num_input_tokens_seen": 171531430,
      "step": 8048
    },
    {
      "epoch": 0.9678350267540432,
      "flos": 16481493811200.0,
      "grad_norm": 2.6713549012939093,
      "learning_rate": 1.08047421780737e-08,
      "loss": 0.9739,
      "num_input_tokens_seen": 171550340,
      "step": 8049
    },
    {
      "epoch": 0.9679552696446823,
      "flos": 15432646656000.0,
      "grad_norm": 10.004631495821698,
      "learning_rate": 1.0724031613251305e-08,
      "loss": 0.9265,
      "num_input_tokens_seen": 171567960,
      "step": 8050
    },
    {
      "epoch": 0.9680755125353213,
      "flos": 19128606535680.0,
      "grad_norm": 4.095912216735978,
      "learning_rate": 1.0643622820011744e-08,
      "loss": 0.8444,
      "num_input_tokens_seen": 171588735,
      "step": 8051
    },
    {
      "epoch": 0.9681957554259605,
      "flos": 20174510469120.0,
      "grad_norm": 4.413835205707677,
      "learning_rate": 1.0563515810552814e-08,
      "loss": 0.8754,
      "num_input_tokens_seen": 171605425,
      "step": 8052
    },
    {
      "epoch": 0.9683159983165995,
      "flos": 14592833126400.0,
      "grad_norm": 2.6489118753235226,
      "learning_rate": 1.0483710597026795e-08,
      "loss": 0.9309,
      "num_input_tokens_seen": 171625005,
      "step": 8053
    },
    {
      "epoch": 0.9684362412072386,
      "flos": 17216453283840.0,
      "grad_norm": 3.66511699411716,
      "learning_rate": 1.0404207191540227e-08,
      "loss": 0.9233,
      "num_input_tokens_seen": 171645180,
      "step": 8054
    },
    {
      "epoch": 0.9685564840978778,
      "flos": 15878345379840.0,
      "grad_norm": 3.7015521737337496,
      "learning_rate": 1.0325005606153236e-08,
      "loss": 0.9361,
      "num_input_tokens_seen": 171664360,
      "step": 8055
    },
    {
      "epoch": 0.9686767269885168,
      "flos": 10156947087360.0,
      "grad_norm": 4.626152307426636,
      "learning_rate": 1.0246105852881104e-08,
      "loss": 0.9685,
      "num_input_tokens_seen": 171679180,
      "step": 8056
    },
    {
      "epoch": 0.9687969698791559,
      "flos": 15117471313920.0,
      "grad_norm": 3.5631810925907472,
      "learning_rate": 1.0167507943692476e-08,
      "loss": 0.9735,
      "num_input_tokens_seen": 171697985,
      "step": 8057
    },
    {
      "epoch": 0.968917212769795,
      "flos": 14068716134400.0,
      "grad_norm": 9.290201616183362,
      "learning_rate": 1.008921189051093e-08,
      "loss": 0.9049,
      "num_input_tokens_seen": 171715050,
      "step": 8058
    },
    {
      "epoch": 0.9690374556604341,
      "flos": 15399741911040.0,
      "grad_norm": 8.031317922104563,
      "learning_rate": 1.0011217705213848e-08,
      "loss": 0.9594,
      "num_input_tokens_seen": 171732645,
      "step": 8059
    },
    {
      "epoch": 0.9691576985510731,
      "flos": 23273024409600.0,
      "grad_norm": 2.9585263542998756,
      "learning_rate": 9.933525399632658e-09,
      "loss": 0.9468,
      "num_input_tokens_seen": 171750600,
      "step": 8060
    },
    {
      "epoch": 0.9692779414417123,
      "flos": 25448922378240.0,
      "grad_norm": 4.0033024030174,
      "learning_rate": 9.856134985553488e-09,
      "loss": 0.8423,
      "num_input_tokens_seen": 171770045,
      "step": 8061
    },
    {
      "epoch": 0.9693981843323514,
      "flos": 20205391749120.0,
      "grad_norm": 2.7005062758894236,
      "learning_rate": 9.77904647471628e-09,
      "loss": 0.9193,
      "num_input_tokens_seen": 171792945,
      "step": 8062
    },
    {
      "epoch": 0.9695184272229904,
      "flos": 16795780055040.0,
      "grad_norm": 2.9455271675541894,
      "learning_rate": 9.702259878815454e-09,
      "loss": 0.9388,
      "num_input_tokens_seen": 171812990,
      "step": 8063
    },
    {
      "epoch": 0.9696386701136296,
      "flos": 16559942737920.0,
      "grad_norm": 5.071489053591965,
      "learning_rate": 9.625775209499254e-09,
      "loss": 0.9323,
      "num_input_tokens_seen": 171832715,
      "step": 8064
    },
    {
      "epoch": 0.9697589130042686,
      "flos": 10686306693120.0,
      "grad_norm": 3.945401332346154,
      "learning_rate": 9.549592478370172e-09,
      "loss": 0.9325,
      "num_input_tokens_seen": 171850615,
      "step": 8065
    },
    {
      "epoch": 0.9698791558949077,
      "flos": 13386812190720.0,
      "grad_norm": 3.250756982530785,
      "learning_rate": 9.473711696985632e-09,
      "loss": 0.9866,
      "num_input_tokens_seen": 171869665,
      "step": 8066
    },
    {
      "epoch": 0.9699993987855468,
      "flos": 12705521418240.0,
      "grad_norm": 4.5397656629808845,
      "learning_rate": 9.398132876856201e-09,
      "loss": 0.9442,
      "num_input_tokens_seen": 171888350,
      "step": 8067
    },
    {
      "epoch": 0.9701196416761859,
      "flos": 48105331138560.0,
      "grad_norm": 0.7710376846662795,
      "learning_rate": 9.322856029447379e-09,
      "loss": 0.8358,
      "num_input_tokens_seen": 171949255,
      "step": 8068
    },
    {
      "epoch": 0.970239884566825,
      "flos": 17266228346880.0,
      "grad_norm": 3.811739110843409,
      "learning_rate": 9.247881166178695e-09,
      "loss": 0.9919,
      "num_input_tokens_seen": 171967685,
      "step": 8069
    },
    {
      "epoch": 0.970360127457464,
      "flos": 18002383503360.0,
      "grad_norm": 4.469222677348006,
      "learning_rate": 9.173208298423274e-09,
      "loss": 0.961,
      "num_input_tokens_seen": 171988610,
      "step": 8070
    },
    {
      "epoch": 0.9704803703481032,
      "flos": 21071978311680.0,
      "grad_norm": 2.551090145831665,
      "learning_rate": 9.09883743750961e-09,
      "loss": 0.9584,
      "num_input_tokens_seen": 172011220,
      "step": 8071
    },
    {
      "epoch": 0.9706006132387422,
      "flos": 12308678000640.0,
      "grad_norm": 2.870978320649454,
      "learning_rate": 9.024768594719124e-09,
      "loss": 1.0229,
      "num_input_tokens_seen": 172029320,
      "step": 8072
    },
    {
      "epoch": 0.9707208561293813,
      "flos": 12888701767680.0,
      "grad_norm": 3.35128349085418,
      "learning_rate": 8.95100178128816e-09,
      "loss": 0.9117,
      "num_input_tokens_seen": 172048180,
      "step": 8073
    },
    {
      "epoch": 0.9708410990200205,
      "flos": 22276098416640.0,
      "grad_norm": 3.6131921705657986,
      "learning_rate": 8.877537008407321e-09,
      "loss": 0.9054,
      "num_input_tokens_seen": 172067950,
      "step": 8074
    },
    {
      "epoch": 0.9709613419106595,
      "flos": 21777282846720.0,
      "grad_norm": 2.830094805366216,
      "learning_rate": 8.804374287221028e-09,
      "loss": 0.8718,
      "num_input_tokens_seen": 172088905,
      "step": 8075
    },
    {
      "epoch": 0.9710815848012986,
      "flos": 16872604078080.0,
      "grad_norm": 4.456049872100455,
      "learning_rate": 8.731513628827958e-09,
      "loss": 1.0361,
      "num_input_tokens_seen": 172107990,
      "step": 8076
    },
    {
      "epoch": 0.9712018276919377,
      "flos": 16894747668480.0,
      "grad_norm": 3.719552346233626,
      "learning_rate": 8.658955044280825e-09,
      "loss": 1.0179,
      "num_input_tokens_seen": 172126635,
      "step": 8077
    },
    {
      "epoch": 0.9713220705825768,
      "flos": 16586010624000.0,
      "grad_norm": 2.7349682845485743,
      "learning_rate": 8.586698544587268e-09,
      "loss": 0.9577,
      "num_input_tokens_seen": 172147965,
      "step": 8078
    },
    {
      "epoch": 0.9714423134732159,
      "flos": 15774012518400.0,
      "grad_norm": 3.71792157346225,
      "learning_rate": 8.514744140707853e-09,
      "loss": 0.9416,
      "num_input_tokens_seen": 172166825,
      "step": 8079
    },
    {
      "epoch": 0.971562556363855,
      "flos": 14357363712000.0,
      "grad_norm": 3.102474818346496,
      "learning_rate": 8.443091843558515e-09,
      "loss": 0.9532,
      "num_input_tokens_seen": 172185630,
      "step": 8080
    },
    {
      "epoch": 0.9716827992544941,
      "flos": 17763725598720.0,
      "grad_norm": 4.594346195899199,
      "learning_rate": 8.37174166400878e-09,
      "loss": 0.8402,
      "num_input_tokens_seen": 172200925,
      "step": 8081
    },
    {
      "epoch": 0.9718030421451331,
      "flos": 17556654120960.0,
      "grad_norm": 3.383969573783716,
      "learning_rate": 8.300693612881992e-09,
      "loss": 1.0351,
      "num_input_tokens_seen": 172220710,
      "step": 8082
    },
    {
      "epoch": 0.9719232850357723,
      "flos": 15694674493440.0,
      "grad_norm": 4.710219801453317,
      "learning_rate": 8.22994770095664e-09,
      "loss": 0.9937,
      "num_input_tokens_seen": 172239005,
      "step": 8083
    },
    {
      "epoch": 0.9720435279264114,
      "flos": 16819916451840.0,
      "grad_norm": 4.910851914099712,
      "learning_rate": 8.159503938964585e-09,
      "loss": 0.9388,
      "num_input_tokens_seen": 172256045,
      "step": 8084
    },
    {
      "epoch": 0.9721637708170504,
      "flos": 20204502650880.0,
      "grad_norm": 4.892618863665223,
      "learning_rate": 8.089362337592164e-09,
      "loss": 0.9014,
      "num_input_tokens_seen": 172279390,
      "step": 8085
    },
    {
      "epoch": 0.9722840137076896,
      "flos": 20754043699200.0,
      "grad_norm": 5.589136021540821,
      "learning_rate": 8.019522907479536e-09,
      "loss": 0.9099,
      "num_input_tokens_seen": 172299470,
      "step": 8086
    },
    {
      "epoch": 0.9724042565983286,
      "flos": 13648656076800.0,
      "grad_norm": 4.153441433874915,
      "learning_rate": 7.949985659221558e-09,
      "loss": 0.957,
      "num_input_tokens_seen": 172316455,
      "step": 8087
    },
    {
      "epoch": 0.9725244994889677,
      "flos": 16376731729920.0,
      "grad_norm": 3.547239713360997,
      "learning_rate": 7.880750603366904e-09,
      "loss": 0.9754,
      "num_input_tokens_seen": 172335045,
      "step": 8088
    },
    {
      "epoch": 0.9726447423796069,
      "flos": 16612231802880.0,
      "grad_norm": 4.01137284939219,
      "learning_rate": 7.811817750418282e-09,
      "loss": 0.9803,
      "num_input_tokens_seen": 172353525,
      "step": 8089
    },
    {
      "epoch": 0.9727649852702459,
      "flos": 18893443706880.0,
      "grad_norm": 2.5498493823280217,
      "learning_rate": 7.743187110833105e-09,
      "loss": 0.9884,
      "num_input_tokens_seen": 172376005,
      "step": 8090
    },
    {
      "epoch": 0.972885228160885,
      "flos": 14567593021440.0,
      "grad_norm": 13.995838091615619,
      "learning_rate": 7.674858695022602e-09,
      "loss": 0.999,
      "num_input_tokens_seen": 172394080,
      "step": 8091
    },
    {
      "epoch": 0.9730054710515241,
      "flos": 12442849751040.0,
      "grad_norm": 5.238476595269332,
      "learning_rate": 7.606832513351591e-09,
      "loss": 0.9466,
      "num_input_tokens_seen": 172411750,
      "step": 8092
    },
    {
      "epoch": 0.9731257139421632,
      "flos": 49391916441600.0,
      "grad_norm": 0.7956084489561719,
      "learning_rate": 7.539108576140264e-09,
      "loss": 0.8667,
      "num_input_tokens_seen": 172475580,
      "step": 8093
    },
    {
      "epoch": 0.9732459568328022,
      "flos": 13098931077120.0,
      "grad_norm": 4.132938260506819,
      "learning_rate": 7.471686893661732e-09,
      "loss": 0.8809,
      "num_input_tokens_seen": 172493595,
      "step": 8094
    },
    {
      "epoch": 0.9733661997234414,
      "flos": 14826984222720.0,
      "grad_norm": 3.5667618807148735,
      "learning_rate": 7.4045674761442636e-09,
      "loss": 0.8324,
      "num_input_tokens_seen": 172510645,
      "step": 8095
    },
    {
      "epoch": 0.9734864426140805,
      "flos": 16898947891200.0,
      "grad_norm": 3.151946656129113,
      "learning_rate": 7.337750333769488e-09,
      "loss": 0.9235,
      "num_input_tokens_seen": 172530170,
      "step": 8096
    },
    {
      "epoch": 0.9736066855047195,
      "flos": 25002733117440.0,
      "grad_norm": 3.6591493245420956,
      "learning_rate": 7.2712354766737425e-09,
      "loss": 0.9268,
      "num_input_tokens_seen": 172550220,
      "step": 8097
    },
    {
      "epoch": 0.9737269283953586,
      "flos": 14486997995520.0,
      "grad_norm": 2.848011554587424,
      "learning_rate": 7.2050229149469565e-09,
      "loss": 0.9998,
      "num_input_tokens_seen": 172569950,
      "step": 8098
    },
    {
      "epoch": 0.9738471712859977,
      "flos": 20596532674560.0,
      "grad_norm": 3.3879687896335837,
      "learning_rate": 7.139112658633984e-09,
      "loss": 0.8187,
      "num_input_tokens_seen": 172589820,
      "step": 8099
    },
    {
      "epoch": 0.9739674141766368,
      "flos": 19785117081600.0,
      "grad_norm": 4.294726025121903,
      "learning_rate": 7.073504717733048e-09,
      "loss": 0.8857,
      "num_input_tokens_seen": 172609105,
      "step": 8100
    },
    {
      "epoch": 0.9740876570672758,
      "flos": 49313682124800.0,
      "grad_norm": 0.7255295003759892,
      "learning_rate": 7.008199102196855e-09,
      "loss": 0.7826,
      "num_input_tokens_seen": 172670250,
      "step": 8101
    },
    {
      "epoch": 0.974207899957915,
      "flos": 41675439820800.0,
      "grad_norm": 0.8050464459164469,
      "learning_rate": 6.9431958219321464e-09,
      "loss": 0.8103,
      "num_input_tokens_seen": 172726135,
      "step": 8102
    },
    {
      "epoch": 0.9743281428485541,
      "flos": 16060544655360.0,
      "grad_norm": 3.0951574957837047,
      "learning_rate": 6.878494886800146e-09,
      "loss": 0.9761,
      "num_input_tokens_seen": 172746630,
      "step": 8103
    },
    {
      "epoch": 0.9744483857391931,
      "flos": 14198166466560.0,
      "grad_norm": 3.2544106118905973,
      "learning_rate": 6.814096306615669e-09,
      "loss": 0.9453,
      "num_input_tokens_seen": 172764490,
      "step": 8104
    },
    {
      "epoch": 0.9745686286298323,
      "flos": 12521298677760.0,
      "grad_norm": 3.26187838774763,
      "learning_rate": 6.750000091148011e-09,
      "loss": 0.8316,
      "num_input_tokens_seen": 172781505,
      "step": 8105
    },
    {
      "epoch": 0.9746888715204713,
      "flos": 20990954065920.0,
      "grad_norm": 5.2762008095702875,
      "learning_rate": 6.686206250120729e-09,
      "loss": 0.9181,
      "num_input_tokens_seen": 172802720,
      "step": 8106
    },
    {
      "epoch": 0.9748091144111104,
      "flos": 13099329638400.0,
      "grad_norm": 14.181651250039058,
      "learning_rate": 6.622714793210749e-09,
      "loss": 0.9284,
      "num_input_tokens_seen": 172821360,
      "step": 8107
    },
    {
      "epoch": 0.9749293573017496,
      "flos": 14669810442240.0,
      "grad_norm": 2.9768029463763934,
      "learning_rate": 6.559525730050364e-09,
      "loss": 0.977,
      "num_input_tokens_seen": 172841180,
      "step": 8108
    },
    {
      "epoch": 0.9750496001923886,
      "flos": 13098716467200.0,
      "grad_norm": 3.2800171488802072,
      "learning_rate": 6.496639070224574e-09,
      "loss": 0.9512,
      "num_input_tokens_seen": 172859385,
      "step": 8109
    },
    {
      "epoch": 0.9751698430830277,
      "flos": 13885198540800.0,
      "grad_norm": 6.715010769083416,
      "learning_rate": 6.4340548232739714e-09,
      "loss": 1.0303,
      "num_input_tokens_seen": 172875305,
      "step": 8110
    },
    {
      "epoch": 0.9752900859736668,
      "flos": 16743705600000.0,
      "grad_norm": 2.7206404679443863,
      "learning_rate": 6.371772998692071e-09,
      "loss": 0.9825,
      "num_input_tokens_seen": 172894280,
      "step": 8111
    },
    {
      "epoch": 0.9754103288643059,
      "flos": 14410081996800.0,
      "grad_norm": 5.629973522821913,
      "learning_rate": 6.309793605927094e-09,
      "loss": 0.8342,
      "num_input_tokens_seen": 172912320,
      "step": 8112
    },
    {
      "epoch": 0.975530571754945,
      "flos": 13724989562880.0,
      "grad_norm": 4.283279494801946,
      "learning_rate": 6.248116654381297e-09,
      "loss": 0.991,
      "num_input_tokens_seen": 172930510,
      "step": 8113
    },
    {
      "epoch": 0.9756508146455841,
      "flos": 16767014215680.0,
      "grad_norm": 3.71896002681264,
      "learning_rate": 6.186742153410751e-09,
      "loss": 0.9114,
      "num_input_tokens_seen": 172949725,
      "step": 8114
    },
    {
      "epoch": 0.9757710575362232,
      "flos": 16324473323520.0,
      "grad_norm": 4.965849407639691,
      "learning_rate": 6.125670112326453e-09,
      "loss": 1.0573,
      "num_input_tokens_seen": 172968705,
      "step": 8115
    },
    {
      "epoch": 0.9758913004268622,
      "flos": 19917296025600.0,
      "grad_norm": 8.144548579867655,
      "learning_rate": 6.064900540392548e-09,
      "loss": 0.8888,
      "num_input_tokens_seen": 172990520,
      "step": 8116
    },
    {
      "epoch": 0.9760115433175014,
      "flos": 15773338030080.0,
      "grad_norm": 3.6644298637716433,
      "learning_rate": 6.0044334468278835e-09,
      "loss": 0.9818,
      "num_input_tokens_seen": 173009585,
      "step": 8117
    },
    {
      "epoch": 0.9761317862081405,
      "flos": 18684502056960.0,
      "grad_norm": 2.9886011687921754,
      "learning_rate": 5.944268840805345e-09,
      "loss": 0.9096,
      "num_input_tokens_seen": 173030050,
      "step": 8118
    },
    {
      "epoch": 0.9762520290987795,
      "flos": 18917365493760.0,
      "grad_norm": 3.807756568645056,
      "learning_rate": 5.88440673145163e-09,
      "loss": 0.8335,
      "num_input_tokens_seen": 173050820,
      "step": 8119
    },
    {
      "epoch": 0.9763722719894187,
      "flos": 12758484971520.0,
      "grad_norm": 3.3008789048054235,
      "learning_rate": 5.824847127848142e-09,
      "loss": 1.0138,
      "num_input_tokens_seen": 173069065,
      "step": 8120
    },
    {
      "epoch": 0.9764925148800577,
      "flos": 15956978257920.0,
      "grad_norm": 3.0511622856437777,
      "learning_rate": 5.765590039029433e-09,
      "loss": 0.9693,
      "num_input_tokens_seen": 173088105,
      "step": 8121
    },
    {
      "epoch": 0.9766127577706968,
      "flos": 26235005890560.0,
      "grad_norm": 2.942418465080217,
      "learning_rate": 5.706635473985422e-09,
      "loss": 0.9078,
      "num_input_tokens_seen": 173111695,
      "step": 8122
    },
    {
      "epoch": 0.976733000661336,
      "flos": 15852308152320.0,
      "grad_norm": 3.7190560607623624,
      "learning_rate": 5.6479834416591764e-09,
      "loss": 1.0433,
      "num_input_tokens_seen": 173130775,
      "step": 8123
    },
    {
      "epoch": 0.976853243551975,
      "flos": 18368161689600.0,
      "grad_norm": 3.020586271158099,
      "learning_rate": 5.589633950947803e-09,
      "loss": 0.8778,
      "num_input_tokens_seen": 173147995,
      "step": 8124
    },
    {
      "epoch": 0.9769734864426141,
      "flos": 15196441436160.0,
      "grad_norm": 7.068428994199837,
      "learning_rate": 5.5315870107035535e-09,
      "loss": 0.8817,
      "num_input_tokens_seen": 173165765,
      "step": 8125
    },
    {
      "epoch": 0.9770937293332532,
      "flos": 9873572782080.0,
      "grad_norm": 3.734676843678578,
      "learning_rate": 5.473842629731607e-09,
      "loss": 0.9667,
      "num_input_tokens_seen": 173183985,
      "step": 8126
    },
    {
      "epoch": 0.9772139722238923,
      "flos": 12705981296640.0,
      "grad_norm": 4.110469309796309,
      "learning_rate": 5.416400816792066e-09,
      "loss": 0.9729,
      "num_input_tokens_seen": 173201220,
      "step": 8127
    },
    {
      "epoch": 0.9773342151145313,
      "flos": 14512422051840.0,
      "grad_norm": 5.867571090573142,
      "learning_rate": 5.359261580598407e-09,
      "loss": 0.9561,
      "num_input_tokens_seen": 173216780,
      "step": 8128
    },
    {
      "epoch": 0.9774544580051704,
      "flos": 8325634129920.0,
      "grad_norm": 5.0583545053825585,
      "learning_rate": 5.302424929819027e-09,
      "loss": 0.9697,
      "num_input_tokens_seen": 173230510,
      "step": 8129
    },
    {
      "epoch": 0.9775747008958096,
      "flos": 9506261667840.0,
      "grad_norm": 6.349231775626481,
      "learning_rate": 5.24589087307592e-09,
      "loss": 0.9143,
      "num_input_tokens_seen": 173247850,
      "step": 8130
    },
    {
      "epoch": 0.9776949437864486,
      "flos": 42604226273280.0,
      "grad_norm": 2.501694200564575,
      "learning_rate": 5.189659418944891e-09,
      "loss": 0.8445,
      "num_input_tokens_seen": 173277745,
      "step": 8131
    },
    {
      "epoch": 0.9778151866770877,
      "flos": 15039175680000.0,
      "grad_norm": 4.0610184535565645,
      "learning_rate": 5.133730575956674e-09,
      "loss": 0.9643,
      "num_input_tokens_seen": 173297135,
      "step": 8132
    },
    {
      "epoch": 0.9779354295677268,
      "flos": 14829467566080.0,
      "grad_norm": 3.270232386366425,
      "learning_rate": 5.0781043525953696e-09,
      "loss": 0.9071,
      "num_input_tokens_seen": 173314920,
      "step": 8133
    },
    {
      "epoch": 0.9780556724583659,
      "flos": 16664827453440.0,
      "grad_norm": 4.927632576113583,
      "learning_rate": 5.0227807572995605e-09,
      "loss": 0.9191,
      "num_input_tokens_seen": 173336615,
      "step": 8134
    },
    {
      "epoch": 0.9781759153490049,
      "flos": 14855658086400.0,
      "grad_norm": 4.392668007778701,
      "learning_rate": 4.967759798461646e-09,
      "loss": 0.8678,
      "num_input_tokens_seen": 173354680,
      "step": 8135
    },
    {
      "epoch": 0.9782961582396441,
      "flos": 20152274903040.0,
      "grad_norm": 4.343260154273538,
      "learning_rate": 4.913041484428282e-09,
      "loss": 0.9375,
      "num_input_tokens_seen": 173374875,
      "step": 8136
    },
    {
      "epoch": 0.9784164011302832,
      "flos": 18182406021120.0,
      "grad_norm": 7.791093965042338,
      "learning_rate": 4.858625823500384e-09,
      "loss": 0.9453,
      "num_input_tokens_seen": 173392295,
      "step": 8137
    },
    {
      "epoch": 0.9785366440209222,
      "flos": 21353697054720.0,
      "grad_norm": 4.917538818764689,
      "learning_rate": 4.80451282393246e-09,
      "loss": 0.9339,
      "num_input_tokens_seen": 173412000,
      "step": 8138
    },
    {
      "epoch": 0.9786568869115614,
      "flos": 23063285637120.0,
      "grad_norm": 4.250654506774446,
      "learning_rate": 4.750702493933722e-09,
      "loss": 0.8707,
      "num_input_tokens_seen": 173431605,
      "step": 8139
    },
    {
      "epoch": 0.9787771298022004,
      "flos": 16586378526720.0,
      "grad_norm": 3.8477770791227703,
      "learning_rate": 4.697194841666974e-09,
      "loss": 1.0395,
      "num_input_tokens_seen": 173450250,
      "step": 8140
    },
    {
      "epoch": 0.9788973726928395,
      "flos": 15247289548800.0,
      "grad_norm": 8.428796418814741,
      "learning_rate": 4.6439898752492764e-09,
      "loss": 1.0091,
      "num_input_tokens_seen": 173470110,
      "step": 8141
    },
    {
      "epoch": 0.9790176155834787,
      "flos": 49338125107200.0,
      "grad_norm": 0.7423469372204974,
      "learning_rate": 4.591087602751731e-09,
      "loss": 0.829,
      "num_input_tokens_seen": 173531690,
      "step": 8142
    },
    {
      "epoch": 0.9791378584741177,
      "flos": 15220056637440.0,
      "grad_norm": 2.9535291284222955,
      "learning_rate": 4.538488032199916e-09,
      "loss": 0.9132,
      "num_input_tokens_seen": 173549510,
      "step": 8143
    },
    {
      "epoch": 0.9792581013647568,
      "flos": 14301579571200.0,
      "grad_norm": 4.547780998654042,
      "learning_rate": 4.486191171572784e-09,
      "loss": 0.8694,
      "num_input_tokens_seen": 173566500,
      "step": 8144
    },
    {
      "epoch": 0.9793783442553959,
      "flos": 16870917857280.0,
      "grad_norm": 2.403329042421162,
      "learning_rate": 4.434197028803766e-09,
      "loss": 0.9738,
      "num_input_tokens_seen": 173585445,
      "step": 8145
    },
    {
      "epoch": 0.979498587146035,
      "flos": 16375934607360.0,
      "grad_norm": 4.597617461138946,
      "learning_rate": 4.3825056117805514e-09,
      "loss": 1.0031,
      "num_input_tokens_seen": 173601050,
      "step": 8146
    },
    {
      "epoch": 0.979618830036674,
      "flos": 10109134172160.0,
      "grad_norm": 4.262906392432812,
      "learning_rate": 4.331116928344425e-09,
      "loss": 0.9882,
      "num_input_tokens_seen": 173617085,
      "step": 8147
    },
    {
      "epoch": 0.9797390729273132,
      "flos": 11840283832320.0,
      "grad_norm": 3.637493178108871,
      "learning_rate": 4.28003098629115e-09,
      "loss": 0.819,
      "num_input_tokens_seen": 173632940,
      "step": 8148
    },
    {
      "epoch": 0.9798593158179523,
      "flos": 17449316720640.0,
      "grad_norm": 4.2681289422132505,
      "learning_rate": 4.229247793370305e-09,
      "loss": 0.9872,
      "num_input_tokens_seen": 173651785,
      "step": 8149
    },
    {
      "epoch": 0.9799795587085913,
      "flos": 19444916244480.0,
      "grad_norm": 3.588569024604117,
      "learning_rate": 4.178767357285951e-09,
      "loss": 0.8919,
      "num_input_tokens_seen": 173673135,
      "step": 8150
    },
    {
      "epoch": 0.9800998015992305,
      "flos": 18710171381760.0,
      "grad_norm": 3.113587406282351,
      "learning_rate": 4.128589685695516e-09,
      "loss": 0.8897,
      "num_input_tokens_seen": 173693280,
      "step": 8151
    },
    {
      "epoch": 0.9802200444898695,
      "flos": 11836972707840.0,
      "grad_norm": 3.3313385398793014,
      "learning_rate": 4.078714786211135e-09,
      "loss": 1.0439,
      "num_input_tokens_seen": 173708850,
      "step": 8152
    },
    {
      "epoch": 0.9803402873805086,
      "flos": 17714042511360.0,
      "grad_norm": 2.7965270097650077,
      "learning_rate": 4.029142666398977e-09,
      "loss": 0.9597,
      "num_input_tokens_seen": 173728735,
      "step": 8153
    },
    {
      "epoch": 0.9804605302711478,
      "flos": 16035488501760.0,
      "grad_norm": 3.0350381050043054,
      "learning_rate": 3.979873333778805e-09,
      "loss": 0.991,
      "num_input_tokens_seen": 173746630,
      "step": 8154
    },
    {
      "epoch": 0.9805807731617868,
      "flos": 27778345758720.0,
      "grad_norm": 3.65148315082305,
      "learning_rate": 3.930906795824862e-09,
      "loss": 0.9265,
      "num_input_tokens_seen": 173767025,
      "step": 8155
    },
    {
      "epoch": 0.9807010160524259,
      "flos": 12626857881600.0,
      "grad_norm": 4.8870807516671375,
      "learning_rate": 3.882243059965207e-09,
      "loss": 0.9683,
      "num_input_tokens_seen": 173784460,
      "step": 8156
    },
    {
      "epoch": 0.980821258943065,
      "flos": 9558765342720.0,
      "grad_norm": 3.8191605217723508,
      "learning_rate": 3.833882133582156e-09,
      "loss": 0.8553,
      "num_input_tokens_seen": 173799840,
      "step": 8157
    },
    {
      "epoch": 0.9809415018337041,
      "flos": 15406609428480.0,
      "grad_norm": 5.184118248650508,
      "learning_rate": 3.785824024012285e-09,
      "loss": 0.9658,
      "num_input_tokens_seen": 173818560,
      "step": 8158
    },
    {
      "epoch": 0.9810617447243432,
      "flos": 16560096030720.0,
      "grad_norm": 2.5399810537998326,
      "learning_rate": 3.738068738545541e-09,
      "loss": 0.9712,
      "num_input_tokens_seen": 173837365,
      "step": 8159
    },
    {
      "epoch": 0.9811819876149822,
      "flos": 12993310556160.0,
      "grad_norm": 4.972987065844397,
      "learning_rate": 3.6906162844265733e-09,
      "loss": 0.9701,
      "num_input_tokens_seen": 173854170,
      "step": 8160
    },
    {
      "epoch": 0.9813022305056214,
      "flos": 16062169559040.0,
      "grad_norm": 3.447460102074498,
      "learning_rate": 3.643466668853845e-09,
      "loss": 0.9059,
      "num_input_tokens_seen": 173871915,
      "step": 8161
    },
    {
      "epoch": 0.9814224733962604,
      "flos": 18082917212160.0,
      "grad_norm": 4.567650800782676,
      "learning_rate": 3.59661989898008e-09,
      "loss": 0.9347,
      "num_input_tokens_seen": 173892690,
      "step": 8162
    },
    {
      "epoch": 0.9815427162868995,
      "flos": 17790375997440.0,
      "grad_norm": 3.1071787990251427,
      "learning_rate": 3.5500759819115934e-09,
      "loss": 0.9609,
      "num_input_tokens_seen": 173912775,
      "step": 8163
    },
    {
      "epoch": 0.9816629591775387,
      "flos": 14667633684480.0,
      "grad_norm": 7.338336653831759,
      "learning_rate": 3.5038349247094034e-09,
      "loss": 1.0123,
      "num_input_tokens_seen": 173929755,
      "step": 8164
    },
    {
      "epoch": 0.9817832020681777,
      "flos": 12547918417920.0,
      "grad_norm": 3.97948703508102,
      "learning_rate": 3.4578967343878994e-09,
      "loss": 0.964,
      "num_input_tokens_seen": 173945680,
      "step": 8165
    },
    {
      "epoch": 0.9819034449588168,
      "flos": 16010616299520.0,
      "grad_norm": 3.47849369202799,
      "learning_rate": 3.4122614179161733e-09,
      "loss": 0.999,
      "num_input_tokens_seen": 173965360,
      "step": 8166
    },
    {
      "epoch": 0.9820236878494559,
      "flos": 14200220590080.0,
      "grad_norm": 2.397884242956872,
      "learning_rate": 3.36692898221691e-09,
      "loss": 0.9557,
      "num_input_tokens_seen": 173983445,
      "step": 8167
    },
    {
      "epoch": 0.982143930740095,
      "flos": 13335013662720.0,
      "grad_norm": 3.075137164770593,
      "learning_rate": 3.3218994341668305e-09,
      "loss": 0.9204,
      "num_input_tokens_seen": 174002095,
      "step": 8168
    },
    {
      "epoch": 0.982264173630734,
      "flos": 18919572910080.0,
      "grad_norm": 2.627893669479806,
      "learning_rate": 3.2771727805971373e-09,
      "loss": 0.9408,
      "num_input_tokens_seen": 174023200,
      "step": 8169
    },
    {
      "epoch": 0.9823844165213732,
      "flos": 15639442206720.0,
      "grad_norm": 2.916479504347628,
      "learning_rate": 3.232749028292847e-09,
      "loss": 0.9629,
      "num_input_tokens_seen": 174039885,
      "step": 8170
    },
    {
      "epoch": 0.9825046594120123,
      "flos": 15563476623360.0,
      "grad_norm": 3.1545768306491997,
      "learning_rate": 3.188628183992792e-09,
      "loss": 1.0746,
      "num_input_tokens_seen": 174059870,
      "step": 8171
    },
    {
      "epoch": 0.9826249023026513,
      "flos": 42581890621440.0,
      "grad_norm": 0.7789299658069909,
      "learning_rate": 3.1448102543902844e-09,
      "loss": 0.849,
      "num_input_tokens_seen": 174123505,
      "step": 8172
    },
    {
      "epoch": 0.9827451451932905,
      "flos": 11368517222400.0,
      "grad_norm": 4.009327027887939,
      "learning_rate": 3.1012952461324515e-09,
      "loss": 0.8573,
      "num_input_tokens_seen": 174142200,
      "step": 8173
    },
    {
      "epoch": 0.9828653880839295,
      "flos": 14380488376320.0,
      "grad_norm": 4.411522058284141,
      "learning_rate": 3.0580831658204575e-09,
      "loss": 0.9387,
      "num_input_tokens_seen": 174159500,
      "step": 8174
    },
    {
      "epoch": 0.9829856309745686,
      "flos": 15353584558080.0,
      "grad_norm": 5.899633829424421,
      "learning_rate": 3.015174020009281e-09,
      "loss": 0.9763,
      "num_input_tokens_seen": 174178545,
      "step": 8175
    },
    {
      "epoch": 0.9831058738652078,
      "flos": 16896740474880.0,
      "grad_norm": 3.788772414596439,
      "learning_rate": 2.9725678152086043e-09,
      "loss": 0.9428,
      "num_input_tokens_seen": 174196835,
      "step": 8176
    },
    {
      "epoch": 0.9832261167558468,
      "flos": 7953754890240.0,
      "grad_norm": 5.392171355991299,
      "learning_rate": 2.930264557881257e-09,
      "loss": 1.0094,
      "num_input_tokens_seen": 174211740,
      "step": 8177
    },
    {
      "epoch": 0.9833463596464859,
      "flos": 42944633610240.0,
      "grad_norm": 0.8009325210231973,
      "learning_rate": 2.8882642544452163e-09,
      "loss": 0.8208,
      "num_input_tokens_seen": 174276185,
      "step": 8178
    },
    {
      "epoch": 0.983466602537125,
      "flos": 9611851530240.0,
      "grad_norm": 3.2683533582863378,
      "learning_rate": 2.8465669112716083e-09,
      "loss": 0.9348,
      "num_input_tokens_seen": 174293430,
      "step": 8179
    },
    {
      "epoch": 0.9835868454277641,
      "flos": 16295186288640.0,
      "grad_norm": 6.046477548711342,
      "learning_rate": 2.8051725346858177e-09,
      "loss": 0.9391,
      "num_input_tokens_seen": 174313410,
      "step": 8180
    },
    {
      "epoch": 0.9837070883184031,
      "flos": 19706913423360.0,
      "grad_norm": 4.417097442540127,
      "learning_rate": 2.7640811309674883e-09,
      "loss": 0.8913,
      "num_input_tokens_seen": 174332630,
      "step": 8181
    },
    {
      "epoch": 0.9838273312090423,
      "flos": 20834148188160.0,
      "grad_norm": 2.4607052648047953,
      "learning_rate": 2.7232927063498557e-09,
      "loss": 1.0004,
      "num_input_tokens_seen": 174352725,
      "step": 8182
    },
    {
      "epoch": 0.9839475740996814,
      "flos": 28644349808640.0,
      "grad_norm": 3.4802909146311127,
      "learning_rate": 2.682807267020859e-09,
      "loss": 0.8725,
      "num_input_tokens_seen": 174375205,
      "step": 8183
    },
    {
      "epoch": 0.9840678169903204,
      "flos": 17188453908480.0,
      "grad_norm": 2.6321527124121693,
      "learning_rate": 2.642624819121808e-09,
      "loss": 0.8172,
      "num_input_tokens_seen": 174395075,
      "step": 8184
    },
    {
      "epoch": 0.9841880598809596,
      "flos": 10372020449280.0,
      "grad_norm": 2.9162537055353788,
      "learning_rate": 2.6027453687487154e-09,
      "loss": 0.8054,
      "num_input_tokens_seen": 174411885,
      "step": 8185
    },
    {
      "epoch": 0.9843083027715986,
      "flos": 15877088378880.0,
      "grad_norm": 4.039779746972938,
      "learning_rate": 2.5631689219509643e-09,
      "loss": 0.7022,
      "num_input_tokens_seen": 174430285,
      "step": 8186
    },
    {
      "epoch": 0.9844285456622377,
      "flos": 15486008770560.0,
      "grad_norm": 3.1135682628249066,
      "learning_rate": 2.523895484732197e-09,
      "loss": 1.03,
      "num_input_tokens_seen": 174449460,
      "step": 8187
    },
    {
      "epoch": 0.9845487885528769,
      "flos": 12779463536640.0,
      "grad_norm": 3.2674056313743054,
      "learning_rate": 2.4849250630505357e-09,
      "loss": 0.9457,
      "num_input_tokens_seen": 174467425,
      "step": 8188
    },
    {
      "epoch": 0.9846690314435159,
      "flos": 17949849169920.0,
      "grad_norm": 2.8131980029848243,
      "learning_rate": 2.4462576628172528e-09,
      "loss": 0.9206,
      "num_input_tokens_seen": 174485775,
      "step": 8189
    },
    {
      "epoch": 0.984789274334155,
      "flos": 13151680020480.0,
      "grad_norm": 3.515123255211489,
      "learning_rate": 2.407893289898766e-09,
      "loss": 0.9359,
      "num_input_tokens_seen": 174504525,
      "step": 8190
    },
    {
      "epoch": 0.984909517224794,
      "flos": 19471260057600.0,
      "grad_norm": 5.686400590211991,
      "learning_rate": 2.3698319501144202e-09,
      "loss": 1.0317,
      "num_input_tokens_seen": 174525230,
      "step": 8191
    },
    {
      "epoch": 0.9850297601154332,
      "flos": 13282326036480.0,
      "grad_norm": 4.526354440549743,
      "learning_rate": 2.3320736492382644e-09,
      "loss": 0.9228,
      "num_input_tokens_seen": 174543785,
      "step": 8192
    },
    {
      "epoch": 0.9851500030060723,
      "flos": 15850836541440.0,
      "grad_norm": 3.8962235824771247,
      "learning_rate": 2.29461839299816e-09,
      "loss": 0.8765,
      "num_input_tokens_seen": 174563220,
      "step": 8193
    },
    {
      "epoch": 0.9852702458967113,
      "flos": 18757616394240.0,
      "grad_norm": 2.1827824540688363,
      "learning_rate": 2.257466187076229e-09,
      "loss": 0.9952,
      "num_input_tokens_seen": 174582145,
      "step": 8194
    },
    {
      "epoch": 0.9853904887873505,
      "flos": 14826738954240.0,
      "grad_norm": 2.816674419445075,
      "learning_rate": 2.2206170371081854e-09,
      "loss": 0.8895,
      "num_input_tokens_seen": 174600450,
      "step": 8195
    },
    {
      "epoch": 0.9855107316779895,
      "flos": 17974660055040.0,
      "grad_norm": 3.489240730130659,
      "learning_rate": 2.1840709486842247e-09,
      "loss": 1.0421,
      "num_input_tokens_seen": 174619790,
      "step": 8196
    },
    {
      "epoch": 0.9856309745686286,
      "flos": 13518040719360.0,
      "grad_norm": 3.181739838366401,
      "learning_rate": 2.1478279273481335e-09,
      "loss": 0.973,
      "num_input_tokens_seen": 174637995,
      "step": 8197
    },
    {
      "epoch": 0.9857512174592677,
      "flos": 24347602206720.0,
      "grad_norm": 12.481536657167608,
      "learning_rate": 2.1118879785981815e-09,
      "loss": 0.9833,
      "num_input_tokens_seen": 174657855,
      "step": 8198
    },
    {
      "epoch": 0.9858714603499068,
      "flos": 18500401950720.0,
      "grad_norm": 8.045217991561092,
      "learning_rate": 2.0762511078862288e-09,
      "loss": 0.9847,
      "num_input_tokens_seen": 174677920,
      "step": 8199
    },
    {
      "epoch": 0.9859917032405459,
      "flos": 16848559656960.0,
      "grad_norm": 6.846518039495457,
      "learning_rate": 2.0409173206186183e-09,
      "loss": 0.8521,
      "num_input_tokens_seen": 174696880,
      "step": 8200
    },
    {
      "epoch": 0.986111946131185,
      "flos": 14147808890880.0,
      "grad_norm": 3.1194605749824165,
      "learning_rate": 2.0058866221550617e-09,
      "loss": 1.0562,
      "num_input_tokens_seen": 174714840,
      "step": 8201
    },
    {
      "epoch": 0.9862321890218241,
      "flos": 14069145354240.0,
      "grad_norm": 4.6270646306873235,
      "learning_rate": 1.971159017809976e-09,
      "loss": 0.9465,
      "num_input_tokens_seen": 174732850,
      "step": 8202
    },
    {
      "epoch": 0.9863524319124631,
      "flos": 15379315200000.0,
      "grad_norm": 7.613308557624833,
      "learning_rate": 1.93673451285159e-09,
      "loss": 0.9652,
      "num_input_tokens_seen": 174751620,
      "step": 8203
    },
    {
      "epoch": 0.9864726748031023,
      "flos": 37747505602560.0,
      "grad_norm": 0.7620971131230421,
      "learning_rate": 1.9026131125019495e-09,
      "loss": 0.7902,
      "num_input_tokens_seen": 174808710,
      "step": 8204
    },
    {
      "epoch": 0.9865929176937414,
      "flos": 16609809776640.0,
      "grad_norm": 2.613778516047396,
      "learning_rate": 1.8687948219371363e-09,
      "loss": 1.0516,
      "num_input_tokens_seen": 174827655,
      "step": 8205
    },
    {
      "epoch": 0.9867131605843804,
      "flos": 15354044436480.0,
      "grad_norm": 6.283214963949835,
      "learning_rate": 1.835279646287491e-09,
      "loss": 1.0682,
      "num_input_tokens_seen": 174845385,
      "step": 8206
    },
    {
      "epoch": 0.9868334034750196,
      "flos": 15824308776960.0,
      "grad_norm": 3.013256506685989,
      "learning_rate": 1.8020675906371685e-09,
      "loss": 0.9656,
      "num_input_tokens_seen": 174864500,
      "step": 8207
    },
    {
      "epoch": 0.9869536463656586,
      "flos": 18366996664320.0,
      "grad_norm": 3.2944804530742746,
      "learning_rate": 1.7691586600243612e-09,
      "loss": 0.9387,
      "num_input_tokens_seen": 174883120,
      "step": 8208
    },
    {
      "epoch": 0.9870738892562977,
      "flos": 11603741368320.0,
      "grad_norm": 4.700207993795224,
      "learning_rate": 1.7365528594415202e-09,
      "loss": 1.0579,
      "num_input_tokens_seen": 174896910,
      "step": 8209
    },
    {
      "epoch": 0.9871941321469369,
      "flos": 25318337679360.0,
      "grad_norm": 5.652862546109711,
      "learning_rate": 1.7042501938346888e-09,
      "loss": 0.8671,
      "num_input_tokens_seen": 174919360,
      "step": 8210
    },
    {
      "epoch": 0.9873143750375759,
      "flos": 15222907883520.0,
      "grad_norm": 3.6724845218007673,
      "learning_rate": 1.6722506681043913e-09,
      "loss": 0.9642,
      "num_input_tokens_seen": 174938040,
      "step": 8211
    },
    {
      "epoch": 0.987434617928215,
      "flos": 11552126791680.0,
      "grad_norm": 4.437710179469012,
      "learning_rate": 1.640554287104745e-09,
      "loss": 0.8804,
      "num_input_tokens_seen": 174956035,
      "step": 8212
    },
    {
      "epoch": 0.9875548608188541,
      "flos": 12647529861120.0,
      "grad_norm": 4.862801181124768,
      "learning_rate": 1.609161055644348e-09,
      "loss": 0.972,
      "num_input_tokens_seen": 174971680,
      "step": 8213
    },
    {
      "epoch": 0.9876751037094932,
      "flos": 18598664417280.0,
      "grad_norm": 4.023706993780953,
      "learning_rate": 1.5780709784849467e-09,
      "loss": 0.8661,
      "num_input_tokens_seen": 174988420,
      "step": 8214
    },
    {
      "epoch": 0.9877953466001322,
      "flos": 11310341713920.0,
      "grad_norm": 5.217870032765574,
      "learning_rate": 1.5472840603436565e-09,
      "loss": 1.0131,
      "num_input_tokens_seen": 175005370,
      "step": 8215
    },
    {
      "epoch": 0.9879155894907714,
      "flos": 13334553784320.0,
      "grad_norm": 3.658369705146834,
      "learning_rate": 1.5168003058900757e-09,
      "loss": 0.9861,
      "num_input_tokens_seen": 175023090,
      "step": 8216
    },
    {
      "epoch": 0.9880358323814105,
      "flos": 15903984046080.0,
      "grad_norm": 3.4528512975698624,
      "learning_rate": 1.4866197197491715e-09,
      "loss": 1.1108,
      "num_input_tokens_seen": 175042170,
      "step": 8217
    },
    {
      "epoch": 0.9881560752720495,
      "flos": 11079348449280.0,
      "grad_norm": 7.036142774650304,
      "learning_rate": 1.4567423064988371e-09,
      "loss": 0.9436,
      "num_input_tokens_seen": 175059240,
      "step": 8218
    },
    {
      "epoch": 0.9882763181626887,
      "flos": 15269310504960.0,
      "grad_norm": 4.029605027146543,
      "learning_rate": 1.4271680706718913e-09,
      "loss": 0.9758,
      "num_input_tokens_seen": 175076635,
      "step": 8219
    },
    {
      "epoch": 0.9883965610533277,
      "flos": 19965998039040.0,
      "grad_norm": 3.900581458346409,
      "learning_rate": 1.3978970167543013e-09,
      "loss": 1.013,
      "num_input_tokens_seen": 175096535,
      "step": 8220
    },
    {
      "epoch": 0.9885168039439668,
      "flos": 9950672732160.0,
      "grad_norm": 4.7905388652028025,
      "learning_rate": 1.3689291491867372e-09,
      "loss": 0.9573,
      "num_input_tokens_seen": 175114570,
      "step": 8221
    },
    {
      "epoch": 0.988637046834606,
      "flos": 18814626877440.0,
      "grad_norm": 2.987542034795154,
      "learning_rate": 1.3402644723636836e-09,
      "loss": 0.932,
      "num_input_tokens_seen": 175136320,
      "step": 8222
    },
    {
      "epoch": 0.988757289725245,
      "flos": 17950278389760.0,
      "grad_norm": 5.599556748695414,
      "learning_rate": 1.311902990633218e-09,
      "loss": 1.0137,
      "num_input_tokens_seen": 175155005,
      "step": 8223
    },
    {
      "epoch": 0.9888775326158841,
      "flos": 18762184519680.0,
      "grad_norm": 3.2497611324854154,
      "learning_rate": 1.2838447082978987e-09,
      "loss": 0.8986,
      "num_input_tokens_seen": 175175880,
      "step": 8224
    },
    {
      "epoch": 0.9889977755065231,
      "flos": 17294013112320.0,
      "grad_norm": 4.880498084304851,
      "learning_rate": 1.2560896296143208e-09,
      "loss": 1.0189,
      "num_input_tokens_seen": 175194065,
      "step": 8225
    },
    {
      "epoch": 0.9891180183971623,
      "flos": 13438028206080.0,
      "grad_norm": 3.8529114749895883,
      "learning_rate": 1.2286377587926722e-09,
      "loss": 0.9976,
      "num_input_tokens_seen": 175210575,
      "step": 8226
    },
    {
      "epoch": 0.9892382612878013,
      "flos": 18631047966720.0,
      "grad_norm": 7.096430738310127,
      "learning_rate": 1.2014890999973992e-09,
      "loss": 0.9514,
      "num_input_tokens_seen": 175227215,
      "step": 8227
    },
    {
      "epoch": 0.9893585041784404,
      "flos": 18108954439680.0,
      "grad_norm": 3.309709489180432,
      "learning_rate": 1.1746436573472073e-09,
      "loss": 0.9714,
      "num_input_tokens_seen": 175248670,
      "step": 8228
    },
    {
      "epoch": 0.9894787470690796,
      "flos": 14327616798720.0,
      "grad_norm": 3.6640233279739105,
      "learning_rate": 1.1481014349141726e-09,
      "loss": 0.8823,
      "num_input_tokens_seen": 175265610,
      "step": 8229
    },
    {
      "epoch": 0.9895989899597186,
      "flos": 17529850429440.0,
      "grad_norm": 3.257639782029758,
      "learning_rate": 1.121862436724852e-09,
      "loss": 1.0281,
      "num_input_tokens_seen": 175284170,
      "step": 8230
    },
    {
      "epoch": 0.9897192328503577,
      "flos": 15485089013760.0,
      "grad_norm": 3.7346187496041705,
      "learning_rate": 1.0959266667598388e-09,
      "loss": 0.9061,
      "num_input_tokens_seen": 175302705,
      "step": 8231
    },
    {
      "epoch": 0.9898394757409968,
      "flos": 15144642908160.0,
      "grad_norm": 6.010062680544377,
      "learning_rate": 1.0702941289533196e-09,
      "loss": 0.9339,
      "num_input_tokens_seen": 175321100,
      "step": 8232
    },
    {
      "epoch": 0.9899597186316359,
      "flos": 13145578967040.0,
      "grad_norm": 2.9512058056661283,
      "learning_rate": 1.0449648271939615e-09,
      "loss": 1.0765,
      "num_input_tokens_seen": 175337165,
      "step": 8233
    },
    {
      "epoch": 0.990079961522275,
      "flos": 16898886574080.0,
      "grad_norm": 2.4110856451378284,
      "learning_rate": 1.0199387653240243e-09,
      "loss": 0.9289,
      "num_input_tokens_seen": 175356575,
      "step": 8234
    },
    {
      "epoch": 0.9902002044129141,
      "flos": 11604323880960.0,
      "grad_norm": 2.5375038410026156,
      "learning_rate": 9.952159471400267e-10,
      "loss": 0.8994,
      "num_input_tokens_seen": 175373335,
      "step": 8235
    },
    {
      "epoch": 0.9903204473035532,
      "flos": 16030981693440.0,
      "grad_norm": 4.335962133939032,
      "learning_rate": 9.707963763923022e-10,
      "loss": 1.0276,
      "num_input_tokens_seen": 175392105,
      "step": 8236
    },
    {
      "epoch": 0.9904406901941922,
      "flos": 11420959580160.0,
      "grad_norm": 2.8937512787060147,
      "learning_rate": 9.466800567854427e-10,
      "loss": 0.9816,
      "num_input_tokens_seen": 175410425,
      "step": 8237
    },
    {
      "epoch": 0.9905609330848314,
      "flos": 18972444487680.0,
      "grad_norm": 5.529196665188271,
      "learning_rate": 9.228669919778553e-10,
      "loss": 0.8789,
      "num_input_tokens_seen": 175429070,
      "step": 8238
    },
    {
      "epoch": 0.9906811759754705,
      "flos": 16428468940800.0,
      "grad_norm": 4.381384849921883,
      "learning_rate": 8.993571855817617e-10,
      "loss": 0.9857,
      "num_input_tokens_seen": 175447620,
      "step": 8239
    },
    {
      "epoch": 0.9908014188661095,
      "flos": 15694582517760.0,
      "grad_norm": 7.493563115046118,
      "learning_rate": 8.761506411638642e-10,
      "loss": 0.9278,
      "num_input_tokens_seen": 175466805,
      "step": 8240
    },
    {
      "epoch": 0.9909216617567487,
      "flos": 13647797637120.0,
      "grad_norm": 4.867040543151623,
      "learning_rate": 8.53247362244236e-10,
      "loss": 0.9253,
      "num_input_tokens_seen": 175485335,
      "step": 8241
    },
    {
      "epoch": 0.9910419046473877,
      "flos": 16794369761280.0,
      "grad_norm": 2.8264568319649648,
      "learning_rate": 8.306473522976532e-10,
      "loss": 0.8769,
      "num_input_tokens_seen": 175504460,
      "step": 8242
    },
    {
      "epoch": 0.9911621475380268,
      "flos": 16140802437120.0,
      "grad_norm": 7.294574644513878,
      "learning_rate": 8.083506147522623e-10,
      "loss": 0.9147,
      "num_input_tokens_seen": 175523575,
      "step": 8243
    },
    {
      "epoch": 0.991282390428666,
      "flos": 9532942725120.0,
      "grad_norm": 4.408388092527791,
      "learning_rate": 7.863571529906909e-10,
      "loss": 1.0456,
      "num_input_tokens_seen": 175538880,
      "step": 8244
    },
    {
      "epoch": 0.991402633319305,
      "flos": 44260751216640.0,
      "grad_norm": 0.7819015456342807,
      "learning_rate": 7.646669703489372e-10,
      "loss": 0.8425,
      "num_input_tokens_seen": 175602910,
      "step": 8245
    },
    {
      "epoch": 0.9915228762099441,
      "flos": 13308209971200.0,
      "grad_norm": 3.1733878901410626,
      "learning_rate": 7.432800701177023e-10,
      "loss": 0.7694,
      "num_input_tokens_seen": 175620630,
      "step": 8246
    },
    {
      "epoch": 0.9916431191005832,
      "flos": 47209518858240.0,
      "grad_norm": 0.8142790025061584,
      "learning_rate": 7.221964555415017e-10,
      "loss": 0.7931,
      "num_input_tokens_seen": 175680010,
      "step": 8247
    },
    {
      "epoch": 0.9917633619912223,
      "flos": 11735399116800.0,
      "grad_norm": 4.983124674805779,
      "learning_rate": 7.01416129818222e-10,
      "loss": 0.9402,
      "num_input_tokens_seen": 175697350,
      "step": 8248
    },
    {
      "epoch": 0.9918836048818613,
      "flos": 18081016381440.0,
      "grad_norm": 3.1916296144198952,
      "learning_rate": 6.809390961006745e-10,
      "loss": 0.7755,
      "num_input_tokens_seen": 175717200,
      "step": 8249
    },
    {
      "epoch": 0.9920038477725005,
      "flos": 17819325788160.0,
      "grad_norm": 2.904740300347585,
      "learning_rate": 6.607653574948191e-10,
      "loss": 0.881,
      "num_input_tokens_seen": 175737700,
      "step": 8250
    },
    {
      "epoch": 0.9921240906631396,
      "flos": 15506404823040.0,
      "grad_norm": 5.107913306902348,
      "learning_rate": 6.408949170613187e-10,
      "loss": 1.0082,
      "num_input_tokens_seen": 175756685,
      "step": 8251
    },
    {
      "epoch": 0.9922443335537786,
      "flos": 17688035942400.0,
      "grad_norm": 2.215471128385389,
      "learning_rate": 6.213277778144288e-10,
      "loss": 1.0106,
      "num_input_tokens_seen": 175778050,
      "step": 8252
    },
    {
      "epoch": 0.9923645764444178,
      "flos": 15350978580480.0,
      "grad_norm": 4.662744076964819,
      "learning_rate": 6.020639427224416e-10,
      "loss": 0.8656,
      "num_input_tokens_seen": 175795415,
      "step": 8253
    },
    {
      "epoch": 0.9924848193350568,
      "flos": 17785777213440.0,
      "grad_norm": 3.4788884643675724,
      "learning_rate": 5.831034147076864e-10,
      "loss": 0.9122,
      "num_input_tokens_seen": 175812385,
      "step": 8254
    },
    {
      "epoch": 0.9926050622256959,
      "flos": 49351001702400.0,
      "grad_norm": 0.7100594096781775,
      "learning_rate": 5.644461966463065e-10,
      "loss": 0.7663,
      "num_input_tokens_seen": 175879715,
      "step": 8255
    },
    {
      "epoch": 0.9927253051163349,
      "flos": 14856025989120.0,
      "grad_norm": 3.487690532770987,
      "learning_rate": 5.460922913687049e-10,
      "loss": 0.9471,
      "num_input_tokens_seen": 175898525,
      "step": 8256
    },
    {
      "epoch": 0.9928455480069741,
      "flos": 15850805882880.0,
      "grad_norm": 3.8021030930654485,
      "learning_rate": 5.280417016593208e-10,
      "loss": 0.9371,
      "num_input_tokens_seen": 175918035,
      "step": 8257
    },
    {
      "epoch": 0.9929657908976132,
      "flos": 12311958466560.0,
      "grad_norm": 2.5305929390799964,
      "learning_rate": 5.102944302559642e-10,
      "loss": 0.9419,
      "num_input_tokens_seen": 175935250,
      "step": 8258
    },
    {
      "epoch": 0.9930860337882522,
      "flos": 16114397306880.0,
      "grad_norm": 4.195494351272862,
      "learning_rate": 4.9285047985137e-10,
      "loss": 0.9693,
      "num_input_tokens_seen": 175954390,
      "step": 8259
    },
    {
      "epoch": 0.9932062766788914,
      "flos": 20047850065920.0,
      "grad_norm": 3.0924726218859657,
      "learning_rate": 4.757098530916436e-10,
      "loss": 0.9419,
      "num_input_tokens_seen": 175974555,
      "step": 8260
    },
    {
      "epoch": 0.9933265195695304,
      "flos": 14304676085760.0,
      "grad_norm": 6.848607956272625,
      "learning_rate": 4.5887255257670563e-10,
      "loss": 0.9767,
      "num_input_tokens_seen": 175991315,
      "step": 8261
    },
    {
      "epoch": 0.9934467624601695,
      "flos": 15171140014080.0,
      "grad_norm": 8.394129527381677,
      "learning_rate": 4.4233858086117906e-10,
      "loss": 0.9664,
      "num_input_tokens_seen": 176009560,
      "step": 8262
    },
    {
      "epoch": 0.9935670053508087,
      "flos": 14016856289280.0,
      "grad_norm": 3.8369529607582225,
      "learning_rate": 4.261079404528356e-10,
      "loss": 0.8631,
      "num_input_tokens_seen": 176028760,
      "step": 8263
    },
    {
      "epoch": 0.9936872482414477,
      "flos": 15613956833280.0,
      "grad_norm": 5.130798595055399,
      "learning_rate": 4.1018063381437205e-10,
      "loss": 0.8774,
      "num_input_tokens_seen": 176048865,
      "step": 8264
    },
    {
      "epoch": 0.9938074911320868,
      "flos": 49994574336000.0,
      "grad_norm": 0.9117838505349187,
      "learning_rate": 3.9455666336141167e-10,
      "loss": 0.8518,
      "num_input_tokens_seen": 176112365,
      "step": 8265
    },
    {
      "epoch": 0.9939277340227259,
      "flos": 10659043123200.0,
      "grad_norm": 7.533664173974738,
      "learning_rate": 3.7923603146450267e-10,
      "loss": 1.0038,
      "num_input_tokens_seen": 176128145,
      "step": 8266
    },
    {
      "epoch": 0.994047976913365,
      "flos": 12548041052160.0,
      "grad_norm": 3.4206882317482252,
      "learning_rate": 3.642187404473418e-10,
      "loss": 0.9812,
      "num_input_tokens_seen": 176146025,
      "step": 8267
    },
    {
      "epoch": 0.994168219804004,
      "flos": 13596704256000.0,
      "grad_norm": 4.420817218335281,
      "learning_rate": 3.495047925885508e-10,
      "loss": 1.0402,
      "num_input_tokens_seen": 176164080,
      "step": 8268
    },
    {
      "epoch": 0.9942884626946432,
      "flos": 12647192616960.0,
      "grad_norm": 3.0308367260660547,
      "learning_rate": 3.350941901199e-10,
      "loss": 1.0278,
      "num_input_tokens_seen": 176180720,
      "step": 8269
    },
    {
      "epoch": 0.9944087055852823,
      "flos": 13325846753280.0,
      "grad_norm": 4.9174678386805954,
      "learning_rate": 3.2098693522764066e-10,
      "loss": 1.026,
      "num_input_tokens_seen": 176193640,
      "step": 8270
    },
    {
      "epoch": 0.9945289484759213,
      "flos": 14908744273920.0,
      "grad_norm": 6.785406710889019,
      "learning_rate": 3.071830300516165e-10,
      "loss": 1.0133,
      "num_input_tokens_seen": 176211190,
      "step": 8271
    },
    {
      "epoch": 0.9946491913665605,
      "flos": 10424033587200.0,
      "grad_norm": 4.140611641340479,
      "learning_rate": 2.9368247668615234e-10,
      "loss": 0.8882,
      "num_input_tokens_seen": 176229500,
      "step": 8272
    },
    {
      "epoch": 0.9947694342571995,
      "flos": 8923141386240.0,
      "grad_norm": 6.898956728583757,
      "learning_rate": 2.804852771789434e-10,
      "loss": 0.7999,
      "num_input_tokens_seen": 176242520,
      "step": 8273
    },
    {
      "epoch": 0.9948896771478386,
      "flos": 13360928256000.0,
      "grad_norm": 5.669566964327018,
      "learning_rate": 2.675914335321661e-10,
      "loss": 0.754,
      "num_input_tokens_seen": 176260995,
      "step": 8274
    },
    {
      "epoch": 0.9950099200384778,
      "flos": 17713827901440.0,
      "grad_norm": 5.20504716073107,
      "learning_rate": 2.550009477018111e-10,
      "loss": 0.9761,
      "num_input_tokens_seen": 176279485,
      "step": 8275
    },
    {
      "epoch": 0.9951301629291168,
      "flos": 16875547299840.0,
      "grad_norm": 2.607968710945691,
      "learning_rate": 2.4271382159790634e-10,
      "loss": 0.823,
      "num_input_tokens_seen": 176296635,
      "step": 8276
    },
    {
      "epoch": 0.9952504058197559,
      "flos": 15799773818880.0,
      "grad_norm": 2.9474764294122724,
      "learning_rate": 2.3073005708429406e-10,
      "loss": 1.0501,
      "num_input_tokens_seen": 176316000,
      "step": 8277
    },
    {
      "epoch": 0.995370648710395,
      "flos": 15061932441600.0,
      "grad_norm": 3.1974805271489273,
      "learning_rate": 2.190496559788535e-10,
      "loss": 0.9114,
      "num_input_tokens_seen": 176334005,
      "step": 8278
    },
    {
      "epoch": 0.9954908916010341,
      "flos": 10502973050880.0,
      "grad_norm": 5.722904218177138,
      "learning_rate": 2.0767262005372265e-10,
      "loss": 0.9594,
      "num_input_tokens_seen": 176351240,
      "step": 8279
    },
    {
      "epoch": 0.9956111344916732,
      "flos": 13623109386240.0,
      "grad_norm": 3.7866538031005863,
      "learning_rate": 1.965989510346322e-10,
      "loss": 0.9427,
      "num_input_tokens_seen": 176370080,
      "step": 8280
    },
    {
      "epoch": 0.9957313773823123,
      "flos": 14225889914880.0,
      "grad_norm": 3.5585625288061666,
      "learning_rate": 1.8582865060134955e-10,
      "loss": 0.891,
      "num_input_tokens_seen": 176387990,
      "step": 8281
    },
    {
      "epoch": 0.9958516202729514,
      "flos": 41135740170240.0,
      "grad_norm": 0.7894437265568093,
      "learning_rate": 1.7536172038790098e-10,
      "loss": 0.78,
      "num_input_tokens_seen": 176448020,
      "step": 8282
    },
    {
      "epoch": 0.9959718631635904,
      "flos": 19785607618560.0,
      "grad_norm": 19.278158623228094,
      "learning_rate": 1.651981619819054e-10,
      "loss": 0.8844,
      "num_input_tokens_seen": 176464890,
      "step": 8283
    },
    {
      "epoch": 0.9960921060542296,
      "flos": 17084121047040.0,
      "grad_norm": 4.292183162880044,
      "learning_rate": 1.5533797692546257e-10,
      "loss": 0.8837,
      "num_input_tokens_seen": 176483345,
      "step": 8284
    },
    {
      "epoch": 0.9962123489448687,
      "flos": 13255951564800.0,
      "grad_norm": 7.897077343321785,
      "learning_rate": 1.4578116671404296e-10,
      "loss": 1.0159,
      "num_input_tokens_seen": 176501345,
      "step": 8285
    },
    {
      "epoch": 0.9963325918355077,
      "flos": 14199822028800.0,
      "grad_norm": 6.145826781533407,
      "learning_rate": 1.3652773279759777e-10,
      "loss": 0.9095,
      "num_input_tokens_seen": 176517715,
      "step": 8286
    },
    {
      "epoch": 0.9964528347261468,
      "flos": 23612826685440.0,
      "grad_norm": 2.5123089086637553,
      "learning_rate": 1.2757767657989305e-10,
      "loss": 0.8154,
      "num_input_tokens_seen": 176541225,
      "step": 8287
    },
    {
      "epoch": 0.9965730776167859,
      "flos": 16426353500160.0,
      "grad_norm": 4.8650436455865345,
      "learning_rate": 1.1893099941850948e-10,
      "loss": 1.0576,
      "num_input_tokens_seen": 176559840,
      "step": 8288
    },
    {
      "epoch": 0.996693320507425,
      "flos": 16323553566720.0,
      "grad_norm": 4.283049713323645,
      "learning_rate": 1.105877026252866e-10,
      "loss": 0.9564,
      "num_input_tokens_seen": 176577890,
      "step": 8289
    },
    {
      "epoch": 0.996813563398064,
      "flos": 9321180487680.0,
      "grad_norm": 4.105996951778895,
      "learning_rate": 1.0254778746565663e-10,
      "loss": 0.9127,
      "num_input_tokens_seen": 176592885,
      "step": 8290
    },
    {
      "epoch": 0.9969338062887032,
      "flos": 10345676636160.0,
      "grad_norm": 6.017663387399047,
      "learning_rate": 9.481125515953259e-11,
      "loss": 0.9282,
      "num_input_tokens_seen": 176610665,
      "step": 8291
    },
    {
      "epoch": 0.9970540491793423,
      "flos": 18313818501120.0,
      "grad_norm": 3.042013192142571,
      "learning_rate": 8.737810688064228e-11,
      "loss": 0.9935,
      "num_input_tokens_seen": 176630220,
      "step": 8292
    },
    {
      "epoch": 0.9971742920699813,
      "flos": 15248730501120.0,
      "grad_norm": 3.2057530484625683,
      "learning_rate": 8.024834375608414e-11,
      "loss": 0.982,
      "num_input_tokens_seen": 176648530,
      "step": 8293
    },
    {
      "epoch": 0.9972945349606205,
      "flos": 51718795161600.0,
      "grad_norm": 0.8351667897743201,
      "learning_rate": 7.342196686788149e-11,
      "loss": 0.8539,
      "num_input_tokens_seen": 176701415,
      "step": 8294
    },
    {
      "epoch": 0.9974147778512595,
      "flos": 13960336343040.0,
      "grad_norm": 3.3004119847071864,
      "learning_rate": 6.689897725142834e-11,
      "loss": 0.8652,
      "num_input_tokens_seen": 176720610,
      "step": 8295
    },
    {
      "epoch": 0.9975350207418986,
      "flos": 11289209856000.0,
      "grad_norm": 4.7624036563930385,
      "learning_rate": 6.067937589615545e-11,
      "loss": 1.0648,
      "num_input_tokens_seen": 176738405,
      "step": 8296
    },
    {
      "epoch": 0.9976552636325378,
      "flos": 42916756869120.0,
      "grad_norm": 0.7559463490366001,
      "learning_rate": 5.476316374575241e-11,
      "loss": 0.7727,
      "num_input_tokens_seen": 176801610,
      "step": 8297
    },
    {
      "epoch": 0.9977755065231768,
      "flos": 15979551068160.0,
      "grad_norm": 3.6294021185533,
      "learning_rate": 4.9150341697723476e-11,
      "loss": 0.928,
      "num_input_tokens_seen": 176821220,
      "step": 8298
    },
    {
      "epoch": 0.9978957494138159,
      "flos": 18525856665600.0,
      "grad_norm": 3.1709145280331663,
      "learning_rate": 4.384091060338768e-11,
      "loss": 0.8522,
      "num_input_tokens_seen": 176841410,
      "step": 8299
    },
    {
      "epoch": 0.998015992304455,
      "flos": 16087869542400.0,
      "grad_norm": 7.069308349791625,
      "learning_rate": 3.883487126810081e-11,
      "loss": 0.9215,
      "num_input_tokens_seen": 176860390,
      "step": 8300
    },
    {
      "epoch": 0.9981362351950941,
      "flos": 12915382824960.0,
      "grad_norm": 4.066689077430218,
      "learning_rate": 3.41322244516995e-11,
      "loss": 0.9914,
      "num_input_tokens_seen": 176878055,
      "step": 8301
    },
    {
      "epoch": 0.9982564780857331,
      "flos": 23875682304000.0,
      "grad_norm": 2.685453123363579,
      "learning_rate": 2.9732970866946925e-11,
      "loss": 0.8169,
      "num_input_tokens_seen": 176897655,
      "step": 8302
    },
    {
      "epoch": 0.9983767209763723,
      "flos": 10974954270720.0,
      "grad_norm": 5.627009223676883,
      "learning_rate": 2.563711118175327e-11,
      "loss": 0.9708,
      "num_input_tokens_seen": 176914260,
      "step": 8303
    },
    {
      "epoch": 0.9984969638670114,
      "flos": 14173692825600.0,
      "grad_norm": 3.6722746935885526,
      "learning_rate": 2.184464601717728e-11,
      "loss": 1.0204,
      "num_input_tokens_seen": 176932295,
      "step": 8304
    },
    {
      "epoch": 0.9986172067576504,
      "flos": 14462156451840.0,
      "grad_norm": 3.9696053611955953,
      "learning_rate": 1.8355575948758585e-11,
      "loss": 0.9603,
      "num_input_tokens_seen": 176950000,
      "step": 8305
    },
    {
      "epoch": 0.9987374496482896,
      "flos": 16870672588800.0,
      "grad_norm": 3.424974126787681,
      "learning_rate": 1.5169901505407424e-11,
      "loss": 0.9238,
      "num_input_tokens_seen": 176966785,
      "step": 8306
    },
    {
      "epoch": 0.9988576925389286,
      "flos": 17816903761920.0,
      "grad_norm": 2.9116190875627206,
      "learning_rate": 1.228762317073695e-11,
      "loss": 0.931,
      "num_input_tokens_seen": 176985335,
      "step": 8307
    },
    {
      "epoch": 0.9989779354295677,
      "flos": 22302963425280.0,
      "grad_norm": 12.738876576858326,
      "learning_rate": 9.70874138195299e-12,
      "loss": 0.9731,
      "num_input_tokens_seen": 177006965,
      "step": 8308
    },
    {
      "epoch": 0.9990981783202069,
      "flos": 13885566443520.0,
      "grad_norm": 3.0918832406604477,
      "learning_rate": 7.433256530076093e-12,
      "loss": 0.9335,
      "num_input_tokens_seen": 177026640,
      "step": 8309
    },
    {
      "epoch": 0.9992184212108459,
      "flos": 12179319644160.0,
      "grad_norm": 4.172858990127663,
      "learning_rate": 5.46116896038562e-12,
      "loss": 0.9443,
      "num_input_tokens_seen": 177040770,
      "step": 8310
    },
    {
      "epoch": 0.999338664101485,
      "flos": 33235386163200.0,
      "grad_norm": 3.268094967623681,
      "learning_rate": 3.792478972197699e-12,
      "loss": 0.8226,
      "num_input_tokens_seen": 177061075,
      "step": 8311
    },
    {
      "epoch": 0.9994589069921241,
      "flos": 10712435896320.0,
      "grad_norm": 4.4113668397005,
      "learning_rate": 2.4271868181990895e-12,
      "loss": 0.8749,
      "num_input_tokens_seen": 177077960,
      "step": 8312
    },
    {
      "epoch": 0.9995791498827632,
      "flos": 8824633651200.0,
      "grad_norm": 3.4663349714295175,
      "learning_rate": 1.3652927060014973e-12,
      "loss": 0.9954,
      "num_input_tokens_seen": 177093275,
      "step": 8313
    },
    {
      "epoch": 0.9996993927734023,
      "flos": 13649698467840.0,
      "grad_norm": 4.977373042190333,
      "learning_rate": 6.067967965872612e-13,
      "loss": 0.8326,
      "num_input_tokens_seen": 177112605,
      "step": 8314
    },
    {
      "epoch": 0.9998196356640414,
      "flos": 45061229813760.0,
      "grad_norm": 3.149551535274832,
      "learning_rate": 1.5169920497548615e-13,
      "loss": 0.9626,
      "num_input_tokens_seen": 177136945,
      "step": 8315
    },
    {
      "epoch": 0.9999398785546805,
      "flos": 36421446512640.0,
      "grad_norm": 1.7974536793222466,
      "learning_rate": 0.0,
      "loss": 0.7646,
      "num_input_tokens_seen": 177185545,
      "step": 8316
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 8316,
  "num_input_tokens_seen": 177185545,
  "num_train_epochs": 1,
  "save_steps": 832,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 4.968105716188447e+17,
  "train_batch_size": 5,
  "trial_name": null,
  "trial_params": null
}