{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 2199,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00136542072025943,
      "grad_norm": 3.824646472930908,
      "learning_rate": 0.0,
      "loss": 2.0623,
      "num_tokens": 810917.0,
      "step": 1
    },
    {
      "epoch": 0.00273084144051886,
      "grad_norm": 3.6299078464508057,
      "learning_rate": 6.060606060606061e-07,
      "loss": 2.1473,
      "num_tokens": 1637352.0,
      "step": 2
    },
    {
      "epoch": 0.00409626216077829,
      "grad_norm": 3.4623773097991943,
      "learning_rate": 1.2121212121212122e-06,
      "loss": 1.9384,
      "num_tokens": 2400944.0,
      "step": 3
    },
    {
      "epoch": 0.00546168288103772,
      "grad_norm": 2.7458271980285645,
      "learning_rate": 1.8181818181818183e-06,
      "loss": 1.6936,
      "num_tokens": 3074555.0,
      "step": 4
    },
    {
      "epoch": 0.006827103601297149,
      "grad_norm": 3.574291944503784,
      "learning_rate": 2.4242424242424244e-06,
      "loss": 2.0138,
      "num_tokens": 3896440.0,
      "step": 5
    },
    {
      "epoch": 0.00819252432155658,
      "grad_norm": 3.3688788414001465,
      "learning_rate": 3.0303030303030305e-06,
      "loss": 2.0293,
      "num_tokens": 4696744.0,
      "step": 6
    },
    {
      "epoch": 0.009557945041816009,
      "grad_norm": 3.3908255100250244,
      "learning_rate": 3.6363636363636366e-06,
      "loss": 2.0567,
      "num_tokens": 5449532.0,
      "step": 7
    },
    {
      "epoch": 0.01092336576207544,
      "grad_norm": 3.6228058338165283,
      "learning_rate": 4.242424242424243e-06,
      "loss": 2.1691,
      "num_tokens": 6279890.0,
      "step": 8
    },
    {
      "epoch": 0.01228878648233487,
      "grad_norm": 3.157053232192993,
      "learning_rate": 4.848484848484849e-06,
      "loss": 2.1514,
      "num_tokens": 7141872.0,
      "step": 9
    },
    {
      "epoch": 0.013654207202594299,
      "grad_norm": 2.5038750171661377,
      "learning_rate": 5.4545454545454545e-06,
      "loss": 1.8428,
      "num_tokens": 7944488.0,
      "step": 10
    },
    {
      "epoch": 0.01501962792285373,
      "grad_norm": 2.3904664516448975,
      "learning_rate": 6.060606060606061e-06,
      "loss": 1.829,
      "num_tokens": 8687257.0,
      "step": 11
    },
    {
      "epoch": 0.01638504864311316,
      "grad_norm": 1.9590299129486084,
      "learning_rate": 6.666666666666667e-06,
      "loss": 1.7447,
      "num_tokens": 9460642.0,
      "step": 12
    },
    {
      "epoch": 0.01775046936337259,
      "grad_norm": 1.7829296588897705,
      "learning_rate": 7.272727272727273e-06,
      "loss": 1.6933,
      "num_tokens": 10243996.0,
      "step": 13
    },
    {
      "epoch": 0.019115890083632018,
      "grad_norm": 1.5945770740509033,
      "learning_rate": 7.87878787878788e-06,
      "loss": 1.6003,
      "num_tokens": 10963865.0,
      "step": 14
    },
    {
      "epoch": 0.02048131080389145,
      "grad_norm": 1.7092599868774414,
      "learning_rate": 8.484848484848486e-06,
      "loss": 1.7995,
      "num_tokens": 11786756.0,
      "step": 15
    },
    {
      "epoch": 0.02184673152415088,
      "grad_norm": 1.2106293439865112,
      "learning_rate": 9.090909090909091e-06,
      "loss": 1.4204,
      "num_tokens": 12463989.0,
      "step": 16
    },
    {
      "epoch": 0.023212152244410308,
      "grad_norm": 1.5587540864944458,
      "learning_rate": 9.696969696969698e-06,
      "loss": 1.6501,
      "num_tokens": 13328035.0,
      "step": 17
    },
    {
      "epoch": 0.02457757296466974,
      "grad_norm": 1.29849112033844,
      "learning_rate": 1.0303030303030304e-05,
      "loss": 1.3735,
      "num_tokens": 14090112.0,
      "step": 18
    },
    {
      "epoch": 0.02594299368492917,
      "grad_norm": 1.3268945217132568,
      "learning_rate": 1.0909090909090909e-05,
      "loss": 1.4973,
      "num_tokens": 14858605.0,
      "step": 19
    },
    {
      "epoch": 0.027308414405188598,
      "grad_norm": 1.213194489479065,
      "learning_rate": 1.1515151515151517e-05,
      "loss": 1.5288,
      "num_tokens": 15678311.0,
      "step": 20
    },
    {
      "epoch": 0.02867383512544803,
      "grad_norm": 1.0691707134246826,
      "learning_rate": 1.2121212121212122e-05,
      "loss": 1.5023,
      "num_tokens": 16492154.0,
      "step": 21
    },
    {
      "epoch": 0.03003925584570746,
      "grad_norm": 1.0144366025924683,
      "learning_rate": 1.2727272727272728e-05,
      "loss": 1.4564,
      "num_tokens": 17329278.0,
      "step": 22
    },
    {
      "epoch": 0.03140467656596689,
      "grad_norm": 0.8399860262870789,
      "learning_rate": 1.3333333333333333e-05,
      "loss": 1.1626,
      "num_tokens": 18047779.0,
      "step": 23
    },
    {
      "epoch": 0.03277009728622632,
      "grad_norm": 0.8519062995910645,
      "learning_rate": 1.3939393939393942e-05,
      "loss": 1.2491,
      "num_tokens": 18845314.0,
      "step": 24
    },
    {
      "epoch": 0.034135518006485746,
      "grad_norm": 1.817383885383606,
      "learning_rate": 1.4545454545454546e-05,
      "loss": 1.1614,
      "num_tokens": 19608652.0,
      "step": 25
    },
    {
      "epoch": 0.03550093872674518,
      "grad_norm": 0.7562612295150757,
      "learning_rate": 1.5151515151515153e-05,
      "loss": 1.1902,
      "num_tokens": 20391088.0,
      "step": 26
    },
    {
      "epoch": 0.03686635944700461,
      "grad_norm": 0.7145614624023438,
      "learning_rate": 1.575757575757576e-05,
      "loss": 1.165,
      "num_tokens": 21209788.0,
      "step": 27
    },
    {
      "epoch": 0.038231780167264036,
      "grad_norm": 0.6634610295295715,
      "learning_rate": 1.6363636363636366e-05,
      "loss": 1.0699,
      "num_tokens": 21957235.0,
      "step": 28
    },
    {
      "epoch": 0.03959720088752347,
      "grad_norm": 0.5545756220817566,
      "learning_rate": 1.6969696969696972e-05,
      "loss": 1.021,
      "num_tokens": 22717810.0,
      "step": 29
    },
    {
      "epoch": 0.0409626216077829,
      "grad_norm": 0.49826163053512573,
      "learning_rate": 1.7575757575757576e-05,
      "loss": 1.0105,
      "num_tokens": 23483922.0,
      "step": 30
    },
    {
      "epoch": 0.042328042328042326,
      "grad_norm": 0.5375344753265381,
      "learning_rate": 1.8181818181818182e-05,
      "loss": 1.0678,
      "num_tokens": 24260390.0,
      "step": 31
    },
    {
      "epoch": 0.04369346304830176,
      "grad_norm": 0.5378962755203247,
      "learning_rate": 1.8787878787878792e-05,
      "loss": 1.1897,
      "num_tokens": 25131243.0,
      "step": 32
    },
    {
      "epoch": 0.04505888376856119,
      "grad_norm": 0.4427362382411957,
      "learning_rate": 1.9393939393939395e-05,
      "loss": 1.0236,
      "num_tokens": 25931856.0,
      "step": 33
    },
    {
      "epoch": 0.046424304488820615,
      "grad_norm": 0.4433201551437378,
      "learning_rate": 2e-05,
      "loss": 1.0578,
      "num_tokens": 26778000.0,
      "step": 34
    },
    {
      "epoch": 0.04778972520908005,
      "grad_norm": 0.4565609395503998,
      "learning_rate": 2.0606060606060608e-05,
      "loss": 1.0128,
      "num_tokens": 27622031.0,
      "step": 35
    },
    {
      "epoch": 0.04915514592933948,
      "grad_norm": 0.3476058542728424,
      "learning_rate": 2.121212121212121e-05,
      "loss": 0.8428,
      "num_tokens": 28357435.0,
      "step": 36
    },
    {
      "epoch": 0.050520566649598905,
      "grad_norm": 0.36521291732788086,
      "learning_rate": 2.1818181818181818e-05,
      "loss": 0.8803,
      "num_tokens": 29103274.0,
      "step": 37
    },
    {
      "epoch": 0.05188598736985834,
      "grad_norm": 0.36516472697257996,
      "learning_rate": 2.2424242424242424e-05,
      "loss": 0.9117,
      "num_tokens": 29917333.0,
      "step": 38
    },
    {
      "epoch": 0.05325140809011777,
      "grad_norm": 0.44692981243133545,
      "learning_rate": 2.3030303030303034e-05,
      "loss": 0.9111,
      "num_tokens": 30783071.0,
      "step": 39
    },
    {
      "epoch": 0.054616828810377195,
      "grad_norm": 0.32400602102279663,
      "learning_rate": 2.363636363636364e-05,
      "loss": 0.8313,
      "num_tokens": 31543642.0,
      "step": 40
    },
    {
      "epoch": 0.05598224953063663,
      "grad_norm": 0.34645065665245056,
      "learning_rate": 2.4242424242424244e-05,
      "loss": 0.8262,
      "num_tokens": 32329323.0,
      "step": 41
    },
    {
      "epoch": 0.05734767025089606,
      "grad_norm": 0.327103853225708,
      "learning_rate": 2.484848484848485e-05,
      "loss": 0.8285,
      "num_tokens": 33125319.0,
      "step": 42
    },
    {
      "epoch": 0.058713090971155485,
      "grad_norm": 0.31019967794418335,
      "learning_rate": 2.5454545454545457e-05,
      "loss": 0.8395,
      "num_tokens": 33954951.0,
      "step": 43
    },
    {
      "epoch": 0.06007851169141492,
      "grad_norm": 0.25041714310646057,
      "learning_rate": 2.6060606060606063e-05,
      "loss": 0.7829,
      "num_tokens": 34727155.0,
      "step": 44
    },
    {
      "epoch": 0.06144393241167435,
      "grad_norm": 0.24972213804721832,
      "learning_rate": 2.6666666666666667e-05,
      "loss": 0.7448,
      "num_tokens": 35495619.0,
      "step": 45
    },
    {
      "epoch": 0.06280935313193378,
      "grad_norm": 0.25105634331703186,
      "learning_rate": 2.7272727272727273e-05,
      "loss": 0.7304,
      "num_tokens": 36253648.0,
      "step": 46
    },
    {
      "epoch": 0.06417477385219321,
      "grad_norm": 0.23696695268154144,
      "learning_rate": 2.7878787878787883e-05,
      "loss": 0.7433,
      "num_tokens": 37013786.0,
      "step": 47
    },
    {
      "epoch": 0.06554019457245264,
      "grad_norm": 0.31701555848121643,
      "learning_rate": 2.848484848484849e-05,
      "loss": 0.7515,
      "num_tokens": 37777217.0,
      "step": 48
    },
    {
      "epoch": 0.06690561529271206,
      "grad_norm": 0.22433985769748688,
      "learning_rate": 2.9090909090909093e-05,
      "loss": 0.7592,
      "num_tokens": 38609357.0,
      "step": 49
    },
    {
      "epoch": 0.06827103601297149,
      "grad_norm": 0.21854479610919952,
      "learning_rate": 2.96969696969697e-05,
      "loss": 0.6692,
      "num_tokens": 39364320.0,
      "step": 50
    },
    {
      "epoch": 0.06963645673323093,
      "grad_norm": 0.2285614311695099,
      "learning_rate": 3.0303030303030306e-05,
      "loss": 0.7431,
      "num_tokens": 40168688.0,
      "step": 51
    },
    {
      "epoch": 0.07100187745349036,
      "grad_norm": 0.1951545625925064,
      "learning_rate": 3.090909090909091e-05,
      "loss": 0.6291,
      "num_tokens": 40887680.0,
      "step": 52
    },
    {
      "epoch": 0.07236729817374979,
      "grad_norm": 0.26640599966049194,
      "learning_rate": 3.151515151515152e-05,
      "loss": 0.7165,
      "num_tokens": 41717726.0,
      "step": 53
    },
    {
      "epoch": 0.07373271889400922,
      "grad_norm": 0.21340061724185944,
      "learning_rate": 3.212121212121212e-05,
      "loss": 0.6754,
      "num_tokens": 42491245.0,
      "step": 54
    },
    {
      "epoch": 0.07509813961426864,
      "grad_norm": 0.29345715045928955,
      "learning_rate": 3.272727272727273e-05,
      "loss": 0.726,
      "num_tokens": 43374525.0,
      "step": 55
    },
    {
      "epoch": 0.07646356033452807,
      "grad_norm": 0.2064027339220047,
      "learning_rate": 3.3333333333333335e-05,
      "loss": 0.6481,
      "num_tokens": 44119621.0,
      "step": 56
    },
    {
      "epoch": 0.07782898105478751,
      "grad_norm": 0.1900351643562317,
      "learning_rate": 3.3939393939393945e-05,
      "loss": 0.6397,
      "num_tokens": 44884679.0,
      "step": 57
    },
    {
      "epoch": 0.07919440177504694,
      "grad_norm": 0.20051197707653046,
      "learning_rate": 3.454545454545455e-05,
      "loss": 0.6669,
      "num_tokens": 45638858.0,
      "step": 58
    },
    {
      "epoch": 0.08055982249530637,
      "grad_norm": 0.18236449360847473,
      "learning_rate": 3.515151515151515e-05,
      "loss": 0.6444,
      "num_tokens": 46418214.0,
      "step": 59
    },
    {
      "epoch": 0.0819252432155658,
      "grad_norm": 0.18849390745162964,
      "learning_rate": 3.575757575757576e-05,
      "loss": 0.6524,
      "num_tokens": 47224980.0,
      "step": 60
    },
    {
      "epoch": 0.08329066393582522,
      "grad_norm": 0.18535448610782623,
      "learning_rate": 3.6363636363636364e-05,
      "loss": 0.639,
      "num_tokens": 47952103.0,
      "step": 61
    },
    {
      "epoch": 0.08465608465608465,
      "grad_norm": 0.1997687816619873,
      "learning_rate": 3.6969696969696974e-05,
      "loss": 0.6326,
      "num_tokens": 48790470.0,
      "step": 62
    },
    {
      "epoch": 0.08602150537634409,
      "grad_norm": 0.16598014533519745,
      "learning_rate": 3.7575757575757584e-05,
      "loss": 0.6488,
      "num_tokens": 49622523.0,
      "step": 63
    },
    {
      "epoch": 0.08738692609660352,
      "grad_norm": 0.22898410260677338,
      "learning_rate": 3.818181818181819e-05,
      "loss": 0.6283,
      "num_tokens": 50395407.0,
      "step": 64
    },
    {
      "epoch": 0.08875234681686295,
      "grad_norm": 0.180471271276474,
      "learning_rate": 3.878787878787879e-05,
      "loss": 0.6248,
      "num_tokens": 51189042.0,
      "step": 65
    },
    {
      "epoch": 0.09011776753712238,
      "grad_norm": 0.20928871631622314,
      "learning_rate": 3.93939393939394e-05,
      "loss": 0.628,
      "num_tokens": 51945385.0,
      "step": 66
    },
    {
      "epoch": 0.0914831882573818,
      "grad_norm": 0.18013833463191986,
      "learning_rate": 4e-05,
      "loss": 0.5938,
      "num_tokens": 52685034.0,
      "step": 67
    },
    {
      "epoch": 0.09284860897764123,
      "grad_norm": 0.2176145613193512,
      "learning_rate": 3.999998047637464e-05,
      "loss": 0.6304,
      "num_tokens": 53493841.0,
      "step": 68
    },
    {
      "epoch": 0.09421402969790067,
      "grad_norm": 0.18821479380130768,
      "learning_rate": 3.99999219055409e-05,
      "loss": 0.5901,
      "num_tokens": 54207375.0,
      "step": 69
    },
    {
      "epoch": 0.0955794504181601,
      "grad_norm": 0.2127874195575714,
      "learning_rate": 3.9999824287625855e-05,
      "loss": 0.618,
      "num_tokens": 54969994.0,
      "step": 70
    },
    {
      "epoch": 0.09694487113841953,
      "grad_norm": 0.19096501171588898,
      "learning_rate": 3.999968762284124e-05,
      "loss": 0.5865,
      "num_tokens": 55762965.0,
      "step": 71
    },
    {
      "epoch": 0.09831029185867896,
      "grad_norm": 0.21292489767074585,
      "learning_rate": 3.999951191148356e-05,
      "loss": 0.6034,
      "num_tokens": 56561625.0,
      "step": 72
    },
    {
      "epoch": 0.09967571257893838,
      "grad_norm": 0.15662604570388794,
      "learning_rate": 3.999929715393395e-05,
      "loss": 0.5758,
      "num_tokens": 57354498.0,
      "step": 73
    },
    {
      "epoch": 0.10104113329919781,
      "grad_norm": 0.1839398294687271,
      "learning_rate": 3.999904335065829e-05,
      "loss": 0.6216,
      "num_tokens": 58152251.0,
      "step": 74
    },
    {
      "epoch": 0.10240655401945725,
      "grad_norm": 0.17141106724739075,
      "learning_rate": 3.999875050220715e-05,
      "loss": 0.6213,
      "num_tokens": 59011068.0,
      "step": 75
    },
    {
      "epoch": 0.10377197473971668,
      "grad_norm": 0.17079652845859528,
      "learning_rate": 3.9998418609215826e-05,
      "loss": 0.5525,
      "num_tokens": 59784522.0,
      "step": 76
    },
    {
      "epoch": 0.10513739545997611,
      "grad_norm": 0.17049561440944672,
      "learning_rate": 3.9998047672404265e-05,
      "loss": 0.5733,
      "num_tokens": 60587642.0,
      "step": 77
    },
    {
      "epoch": 0.10650281618023553,
      "grad_norm": 0.17044956982135773,
      "learning_rate": 3.9997637692577146e-05,
      "loss": 0.5729,
      "num_tokens": 61327592.0,
      "step": 78
    },
    {
      "epoch": 0.10786823690049496,
      "grad_norm": 0.1717107594013214,
      "learning_rate": 3.999718867062384e-05,
      "loss": 0.5555,
      "num_tokens": 62099441.0,
      "step": 79
    },
    {
      "epoch": 0.10923365762075439,
      "grad_norm": 0.1577945053577423,
      "learning_rate": 3.99967006075184e-05,
      "loss": 0.5874,
      "num_tokens": 62888444.0,
      "step": 80
    },
    {
      "epoch": 0.11059907834101383,
      "grad_norm": 0.158148854970932,
      "learning_rate": 3.999617350431958e-05,
      "loss": 0.562,
      "num_tokens": 63682672.0,
      "step": 81
    },
    {
      "epoch": 0.11196449906127326,
      "grad_norm": 0.16743303835391998,
      "learning_rate": 3.9995607362170823e-05,
      "loss": 0.5618,
      "num_tokens": 64573108.0,
      "step": 82
    },
    {
      "epoch": 0.11332991978153269,
      "grad_norm": 0.15337343513965607,
      "learning_rate": 3.999500218230025e-05,
      "loss": 0.5711,
      "num_tokens": 65411389.0,
      "step": 83
    },
    {
      "epoch": 0.11469534050179211,
      "grad_norm": 0.1706647276878357,
      "learning_rate": 3.999435796602068e-05,
      "loss": 0.5549,
      "num_tokens": 66170606.0,
      "step": 84
    },
    {
      "epoch": 0.11606076122205154,
      "grad_norm": 0.1641642451286316,
      "learning_rate": 3.99936747147296e-05,
      "loss": 0.5661,
      "num_tokens": 66943810.0,
      "step": 85
    },
    {
      "epoch": 0.11742618194231097,
      "grad_norm": 0.16418936848640442,
      "learning_rate": 3.999295242990918e-05,
      "loss": 0.5683,
      "num_tokens": 67765182.0,
      "step": 86
    },
    {
      "epoch": 0.11879160266257041,
      "grad_norm": 0.1762625128030777,
      "learning_rate": 3.999219111312627e-05,
      "loss": 0.5792,
      "num_tokens": 68529643.0,
      "step": 87
    },
    {
      "epoch": 0.12015702338282984,
      "grad_norm": 0.13459305465221405,
      "learning_rate": 3.99913907660324e-05,
      "loss": 0.5625,
      "num_tokens": 69306401.0,
      "step": 88
    },
    {
      "epoch": 0.12152244410308927,
      "grad_norm": 0.19044966995716095,
      "learning_rate": 3.9990551390363735e-05,
      "loss": 0.5951,
      "num_tokens": 70150195.0,
      "step": 89
    },
    {
      "epoch": 0.1228878648233487,
      "grad_norm": 0.1444804072380066,
      "learning_rate": 3.998967298794113e-05,
      "loss": 0.563,
      "num_tokens": 70946010.0,
      "step": 90
    },
    {
      "epoch": 0.12425328554360812,
      "grad_norm": 0.258836954832077,
      "learning_rate": 3.998875556067011e-05,
      "loss": 0.5764,
      "num_tokens": 71769941.0,
      "step": 91
    },
    {
      "epoch": 0.12561870626386756,
      "grad_norm": 0.16031533479690552,
      "learning_rate": 3.9987799110540836e-05,
      "loss": 0.5783,
      "num_tokens": 72564346.0,
      "step": 92
    },
    {
      "epoch": 0.12698412698412698,
      "grad_norm": 0.17529574036598206,
      "learning_rate": 3.9986803639628124e-05,
      "loss": 0.5707,
      "num_tokens": 73334106.0,
      "step": 93
    },
    {
      "epoch": 0.12834954770438642,
      "grad_norm": 0.17634113132953644,
      "learning_rate": 3.9985769150091434e-05,
      "loss": 0.5951,
      "num_tokens": 74100260.0,
      "step": 94
    },
    {
      "epoch": 0.12971496842464583,
      "grad_norm": 0.22044458985328674,
      "learning_rate": 3.998469564417489e-05,
      "loss": 0.5407,
      "num_tokens": 74906036.0,
      "step": 95
    },
    {
      "epoch": 0.13108038914490527,
      "grad_norm": 0.15553663671016693,
      "learning_rate": 3.998358312420723e-05,
      "loss": 0.5245,
      "num_tokens": 75710176.0,
      "step": 96
    },
    {
      "epoch": 0.13244580986516472,
      "grad_norm": 0.16881951689720154,
      "learning_rate": 3.998243159260184e-05,
      "loss": 0.5716,
      "num_tokens": 76589725.0,
      "step": 97
    },
    {
      "epoch": 0.13381123058542413,
      "grad_norm": 0.14743050932884216,
      "learning_rate": 3.9981241051856734e-05,
      "loss": 0.5334,
      "num_tokens": 77363332.0,
      "step": 98
    },
    {
      "epoch": 0.13517665130568357,
      "grad_norm": 0.14990708231925964,
      "learning_rate": 3.998001150455452e-05,
      "loss": 0.5625,
      "num_tokens": 78157499.0,
      "step": 99
    },
    {
      "epoch": 0.13654207202594298,
      "grad_norm": 0.13977673649787903,
      "learning_rate": 3.997874295336246e-05,
      "loss": 0.5612,
      "num_tokens": 78950774.0,
      "step": 100
    },
    {
      "epoch": 0.13790749274620243,
      "grad_norm": 0.15616829693317413,
      "learning_rate": 3.9977435401032415e-05,
      "loss": 0.5598,
      "num_tokens": 79771176.0,
      "step": 101
    },
    {
      "epoch": 0.13927291346646187,
      "grad_norm": 0.14105041325092316,
      "learning_rate": 3.997608885040083e-05,
      "loss": 0.5636,
      "num_tokens": 80574592.0,
      "step": 102
    },
    {
      "epoch": 0.14063833418672128,
      "grad_norm": 0.1602727472782135,
      "learning_rate": 3.997470330438879e-05,
      "loss": 0.5651,
      "num_tokens": 81458112.0,
      "step": 103
    },
    {
      "epoch": 0.14200375490698072,
      "grad_norm": 0.1497201770544052,
      "learning_rate": 3.997327876600194e-05,
      "loss": 0.5618,
      "num_tokens": 82283035.0,
      "step": 104
    },
    {
      "epoch": 0.14336917562724014,
      "grad_norm": 0.13994167745113373,
      "learning_rate": 3.997181523833051e-05,
      "loss": 0.5611,
      "num_tokens": 83107054.0,
      "step": 105
    },
    {
      "epoch": 0.14473459634749958,
      "grad_norm": 0.16725604236125946,
      "learning_rate": 3.9970312724549325e-05,
      "loss": 0.5347,
      "num_tokens": 83870064.0,
      "step": 106
    },
    {
      "epoch": 0.146100017067759,
      "grad_norm": 0.13310487568378448,
      "learning_rate": 3.996877122791778e-05,
      "loss": 0.5551,
      "num_tokens": 84629069.0,
      "step": 107
    },
    {
      "epoch": 0.14746543778801843,
      "grad_norm": 0.160286083817482,
      "learning_rate": 3.9967190751779815e-05,
      "loss": 0.5652,
      "num_tokens": 85399380.0,
      "step": 108
    },
    {
      "epoch": 0.14883085850827787,
      "grad_norm": 0.1426922231912613,
      "learning_rate": 3.9965571299563966e-05,
      "loss": 0.5583,
      "num_tokens": 86218939.0,
      "step": 109
    },
    {
      "epoch": 0.1501962792285373,
      "grad_norm": 0.14385412633419037,
      "learning_rate": 3.9963912874783276e-05,
      "loss": 0.5372,
      "num_tokens": 86997053.0,
      "step": 110
    },
    {
      "epoch": 0.15156169994879673,
      "grad_norm": 0.1570274531841278,
      "learning_rate": 3.9962215481035374e-05,
      "loss": 0.5221,
      "num_tokens": 87689120.0,
      "step": 111
    },
    {
      "epoch": 0.15292712066905614,
      "grad_norm": 0.15103031694889069,
      "learning_rate": 3.9960479122002384e-05,
      "loss": 0.5253,
      "num_tokens": 88488438.0,
      "step": 112
    },
    {
      "epoch": 0.15429254138931559,
      "grad_norm": 0.13849900662899017,
      "learning_rate": 3.995870380145098e-05,
      "loss": 0.5483,
      "num_tokens": 89273873.0,
      "step": 113
    },
    {
      "epoch": 0.15565796210957503,
      "grad_norm": 0.18940778076648712,
      "learning_rate": 3.995688952323236e-05,
      "loss": 0.5936,
      "num_tokens": 90108832.0,
      "step": 114
    },
    {
      "epoch": 0.15702338282983444,
      "grad_norm": 0.1479174941778183,
      "learning_rate": 3.99550362912822e-05,
      "loss": 0.5581,
      "num_tokens": 90928601.0,
      "step": 115
    },
    {
      "epoch": 0.15838880355009388,
      "grad_norm": 0.1770520955324173,
      "learning_rate": 3.995314410962073e-05,
      "loss": 0.5289,
      "num_tokens": 91689223.0,
      "step": 116
    },
    {
      "epoch": 0.1597542242703533,
      "grad_norm": 0.14382274448871613,
      "learning_rate": 3.995121298235262e-05,
      "loss": 0.5236,
      "num_tokens": 92516553.0,
      "step": 117
    },
    {
      "epoch": 0.16111964499061274,
      "grad_norm": 0.15445734560489655,
      "learning_rate": 3.9949242913667066e-05,
      "loss": 0.5522,
      "num_tokens": 93292360.0,
      "step": 118
    },
    {
      "epoch": 0.16248506571087215,
      "grad_norm": 0.16529829800128937,
      "learning_rate": 3.994723390783771e-05,
      "loss": 0.5157,
      "num_tokens": 94084233.0,
      "step": 119
    },
    {
      "epoch": 0.1638504864311316,
      "grad_norm": 0.1528000831604004,
      "learning_rate": 3.994518596922267e-05,
      "loss": 0.5454,
      "num_tokens": 94912958.0,
      "step": 120
    },
    {
      "epoch": 0.16521590715139103,
      "grad_norm": 0.16148272156715393,
      "learning_rate": 3.994309910226453e-05,
      "loss": 0.5314,
      "num_tokens": 95684230.0,
      "step": 121
    },
    {
      "epoch": 0.16658132787165045,
      "grad_norm": 0.16350367665290833,
      "learning_rate": 3.994097331149031e-05,
      "loss": 0.5597,
      "num_tokens": 96567534.0,
      "step": 122
    },
    {
      "epoch": 0.1679467485919099,
      "grad_norm": 0.15823043882846832,
      "learning_rate": 3.9938808601511475e-05,
      "loss": 0.5182,
      "num_tokens": 97354656.0,
      "step": 123
    },
    {
      "epoch": 0.1693121693121693,
      "grad_norm": 0.15626828372478485,
      "learning_rate": 3.99366049770239e-05,
      "loss": 0.5712,
      "num_tokens": 98139538.0,
      "step": 124
    },
    {
      "epoch": 0.17067759003242874,
      "grad_norm": 0.16974851489067078,
      "learning_rate": 3.9934362442807904e-05,
      "loss": 0.552,
      "num_tokens": 98963905.0,
      "step": 125
    },
    {
      "epoch": 0.17204301075268819,
      "grad_norm": 0.13545623421669006,
      "learning_rate": 3.993208100372818e-05,
      "loss": 0.516,
      "num_tokens": 99674602.0,
      "step": 126
    },
    {
      "epoch": 0.1734084314729476,
      "grad_norm": 0.14402753114700317,
      "learning_rate": 3.992976066473386e-05,
      "loss": 0.5651,
      "num_tokens": 100467258.0,
      "step": 127
    },
    {
      "epoch": 0.17477385219320704,
      "grad_norm": 0.15770253539085388,
      "learning_rate": 3.992740143085842e-05,
      "loss": 0.5416,
      "num_tokens": 101369120.0,
      "step": 128
    },
    {
      "epoch": 0.17613927291346645,
      "grad_norm": 0.14714737236499786,
      "learning_rate": 3.992500330721973e-05,
      "loss": 0.5447,
      "num_tokens": 102174323.0,
      "step": 129
    },
    {
      "epoch": 0.1775046936337259,
      "grad_norm": 0.14332175254821777,
      "learning_rate": 3.9922566299020015e-05,
      "loss": 0.5265,
      "num_tokens": 103057513.0,
      "step": 130
    },
    {
      "epoch": 0.1788701143539853,
      "grad_norm": 0.1409868597984314,
      "learning_rate": 3.9920090411545855e-05,
      "loss": 0.5462,
      "num_tokens": 103874672.0,
      "step": 131
    },
    {
      "epoch": 0.18023553507424475,
      "grad_norm": 0.13499683141708374,
      "learning_rate": 3.991757565016818e-05,
      "loss": 0.5307,
      "num_tokens": 104702175.0,
      "step": 132
    },
    {
      "epoch": 0.1816009557945042,
      "grad_norm": 0.12215153872966766,
      "learning_rate": 3.991502202034225e-05,
      "loss": 0.5353,
      "num_tokens": 105508889.0,
      "step": 133
    },
    {
      "epoch": 0.1829663765147636,
      "grad_norm": 0.1402658075094223,
      "learning_rate": 3.991242952760761e-05,
      "loss": 0.5307,
      "num_tokens": 106286985.0,
      "step": 134
    },
    {
      "epoch": 0.18433179723502305,
      "grad_norm": 0.12332189083099365,
      "learning_rate": 3.990979817758815e-05,
      "loss": 0.5192,
      "num_tokens": 107034477.0,
      "step": 135
    },
    {
      "epoch": 0.18569721795528246,
      "grad_norm": 0.15967614948749542,
      "learning_rate": 3.990712797599202e-05,
      "loss": 0.5443,
      "num_tokens": 107815602.0,
      "step": 136
    },
    {
      "epoch": 0.1870626386755419,
      "grad_norm": 0.13816246390342712,
      "learning_rate": 3.9904418928611686e-05,
      "loss": 0.5435,
      "num_tokens": 108544217.0,
      "step": 137
    },
    {
      "epoch": 0.18842805939580134,
      "grad_norm": 0.13450150191783905,
      "learning_rate": 3.9901671041323856e-05,
      "loss": 0.5276,
      "num_tokens": 109292340.0,
      "step": 138
    },
    {
      "epoch": 0.18979348011606076,
      "grad_norm": 0.14420445263385773,
      "learning_rate": 3.989888432008948e-05,
      "loss": 0.5236,
      "num_tokens": 110016634.0,
      "step": 139
    },
    {
      "epoch": 0.1911589008363202,
      "grad_norm": 0.13923604786396027,
      "learning_rate": 3.98960587709538e-05,
      "loss": 0.5592,
      "num_tokens": 110851610.0,
      "step": 140
    },
    {
      "epoch": 0.19252432155657961,
      "grad_norm": 0.13730958104133606,
      "learning_rate": 3.989319440004623e-05,
      "loss": 0.491,
      "num_tokens": 111596782.0,
      "step": 141
    },
    {
      "epoch": 0.19388974227683906,
      "grad_norm": 0.1341806799173355,
      "learning_rate": 3.9890291213580444e-05,
      "loss": 0.5212,
      "num_tokens": 112403338.0,
      "step": 142
    },
    {
      "epoch": 0.19525516299709847,
      "grad_norm": 0.137551411986351,
      "learning_rate": 3.988734921785429e-05,
      "loss": 0.53,
      "num_tokens": 113269004.0,
      "step": 143
    },
    {
      "epoch": 0.1966205837173579,
      "grad_norm": 0.12640979886054993,
      "learning_rate": 3.988436841924982e-05,
      "loss": 0.5183,
      "num_tokens": 114083746.0,
      "step": 144
    },
    {
      "epoch": 0.19798600443761735,
      "grad_norm": 0.14309902489185333,
      "learning_rate": 3.988134882423326e-05,
      "loss": 0.536,
      "num_tokens": 114930696.0,
      "step": 145
    },
    {
      "epoch": 0.19935142515787677,
      "grad_norm": 0.15520532429218292,
      "learning_rate": 3.987829043935498e-05,
      "loss": 0.5222,
      "num_tokens": 115682051.0,
      "step": 146
    },
    {
      "epoch": 0.2007168458781362,
      "grad_norm": 0.1504029482603073,
      "learning_rate": 3.987519327124951e-05,
      "loss": 0.5272,
      "num_tokens": 116451759.0,
      "step": 147
    },
    {
      "epoch": 0.20208226659839562,
      "grad_norm": 0.1612902283668518,
      "learning_rate": 3.9872057326635525e-05,
      "loss": 0.5485,
      "num_tokens": 117254700.0,
      "step": 148
    },
    {
      "epoch": 0.20344768731865506,
      "grad_norm": 0.130841463804245,
      "learning_rate": 3.986888261231579e-05,
      "loss": 0.5185,
      "num_tokens": 118030789.0,
      "step": 149
    },
    {
      "epoch": 0.2048131080389145,
      "grad_norm": 0.15020939707756042,
      "learning_rate": 3.98656691351772e-05,
      "loss": 0.5155,
      "num_tokens": 118743654.0,
      "step": 150
    },
    {
      "epoch": 0.20617852875917392,
      "grad_norm": 0.1403353065252304,
      "learning_rate": 3.9862416902190706e-05,
      "loss": 0.5291,
      "num_tokens": 119532223.0,
      "step": 151
    },
    {
      "epoch": 0.20754394947943336,
      "grad_norm": 0.1320616751909256,
      "learning_rate": 3.985912592041136e-05,
      "loss": 0.5013,
      "num_tokens": 120322536.0,
      "step": 152
    },
    {
      "epoch": 0.20890937019969277,
      "grad_norm": 0.13357800245285034,
      "learning_rate": 3.9855796196978265e-05,
      "loss": 0.5097,
      "num_tokens": 121125785.0,
      "step": 153
    },
    {
      "epoch": 0.21027479091995221,
      "grad_norm": 0.15188884735107422,
      "learning_rate": 3.985242773911456e-05,
      "loss": 0.5194,
      "num_tokens": 121927858.0,
      "step": 154
    },
    {
      "epoch": 0.21164021164021163,
      "grad_norm": 0.12923157215118408,
      "learning_rate": 3.9849020554127415e-05,
      "loss": 0.51,
      "num_tokens": 122808775.0,
      "step": 155
    },
    {
      "epoch": 0.21300563236047107,
      "grad_norm": 0.14845870435237885,
      "learning_rate": 3.9845574649407996e-05,
      "loss": 0.5378,
      "num_tokens": 123584876.0,
      "step": 156
    },
    {
      "epoch": 0.2143710530807305,
      "grad_norm": 0.14386294782161713,
      "learning_rate": 3.984209003243149e-05,
      "loss": 0.4892,
      "num_tokens": 124317070.0,
      "step": 157
    },
    {
      "epoch": 0.21573647380098993,
      "grad_norm": 0.1292552947998047,
      "learning_rate": 3.9838566710757044e-05,
      "loss": 0.5562,
      "num_tokens": 125173487.0,
      "step": 158
    },
    {
      "epoch": 0.21710189452124937,
      "grad_norm": 0.13404200971126556,
      "learning_rate": 3.983500469202776e-05,
      "loss": 0.532,
      "num_tokens": 126034658.0,
      "step": 159
    },
    {
      "epoch": 0.21846731524150878,
      "grad_norm": 0.13287687301635742,
      "learning_rate": 3.983140398397071e-05,
      "loss": 0.5036,
      "num_tokens": 126822301.0,
      "step": 160
    },
    {
      "epoch": 0.21983273596176822,
      "grad_norm": 0.13076184689998627,
      "learning_rate": 3.982776459439687e-05,
      "loss": 0.5143,
      "num_tokens": 127596381.0,
      "step": 161
    },
    {
      "epoch": 0.22119815668202766,
      "grad_norm": 0.1249568983912468,
      "learning_rate": 3.982408653120115e-05,
      "loss": 0.5187,
      "num_tokens": 128376625.0,
      "step": 162
    },
    {
      "epoch": 0.22256357740228708,
      "grad_norm": 0.13266579806804657,
      "learning_rate": 3.9820369802362325e-05,
      "loss": 0.5503,
      "num_tokens": 129165656.0,
      "step": 163
    },
    {
      "epoch": 0.22392899812254652,
      "grad_norm": 0.13949643075466156,
      "learning_rate": 3.981661441594308e-05,
      "loss": 0.5268,
      "num_tokens": 129962103.0,
      "step": 164
    },
    {
      "epoch": 0.22529441884280593,
      "grad_norm": 0.15097415447235107,
      "learning_rate": 3.981282038008993e-05,
      "loss": 0.4968,
      "num_tokens": 130743246.0,
      "step": 165
    },
    {
      "epoch": 0.22665983956306537,
      "grad_norm": 0.1465708613395691,
      "learning_rate": 3.9808987703033256e-05,
      "loss": 0.5,
      "num_tokens": 131566020.0,
      "step": 166
    },
    {
      "epoch": 0.2280252602833248,
      "grad_norm": 0.13422159850597382,
      "learning_rate": 3.980511639308724e-05,
      "loss": 0.5127,
      "num_tokens": 132411125.0,
      "step": 167
    },
    {
      "epoch": 0.22939068100358423,
      "grad_norm": 0.1332809031009674,
      "learning_rate": 3.980120645864991e-05,
      "loss": 0.5248,
      "num_tokens": 133168196.0,
      "step": 168
    },
    {
      "epoch": 0.23075610172384367,
      "grad_norm": 0.13762716948986053,
      "learning_rate": 3.979725790820302e-05,
      "loss": 0.5221,
      "num_tokens": 134042762.0,
      "step": 169
    },
    {
      "epoch": 0.23212152244410308,
      "grad_norm": 0.15409940481185913,
      "learning_rate": 3.979327075031215e-05,
      "loss": 0.5382,
      "num_tokens": 134763575.0,
      "step": 170
    },
    {
      "epoch": 0.23348694316436253,
      "grad_norm": 0.14578698575496674,
      "learning_rate": 3.97892449936266e-05,
      "loss": 0.5141,
      "num_tokens": 135525003.0,
      "step": 171
    },
    {
      "epoch": 0.23485236388462194,
      "grad_norm": 0.13474570214748383,
      "learning_rate": 3.9785180646879406e-05,
      "loss": 0.5218,
      "num_tokens": 136288630.0,
      "step": 172
    },
    {
      "epoch": 0.23621778460488138,
      "grad_norm": 0.1379004269838333,
      "learning_rate": 3.978107771888733e-05,
      "loss": 0.5616,
      "num_tokens": 137111606.0,
      "step": 173
    },
    {
      "epoch": 0.23758320532514082,
      "grad_norm": 0.12900549173355103,
      "learning_rate": 3.9776936218550825e-05,
      "loss": 0.5258,
      "num_tokens": 137958048.0,
      "step": 174
    },
    {
      "epoch": 0.23894862604540024,
      "grad_norm": 0.1291351616382599,
      "learning_rate": 3.977275615485401e-05,
      "loss": 0.5292,
      "num_tokens": 138707213.0,
      "step": 175
    },
    {
      "epoch": 0.24031404676565968,
      "grad_norm": 0.12777993083000183,
      "learning_rate": 3.9768537536864646e-05,
      "loss": 0.4923,
      "num_tokens": 139433684.0,
      "step": 176
    },
    {
      "epoch": 0.2416794674859191,
      "grad_norm": 0.13202154636383057,
      "learning_rate": 3.9764280373734165e-05,
      "loss": 0.5059,
      "num_tokens": 140184059.0,
      "step": 177
    },
    {
      "epoch": 0.24304488820617853,
      "grad_norm": 0.1311146467924118,
      "learning_rate": 3.9759984674697584e-05,
      "loss": 0.4958,
      "num_tokens": 140920103.0,
      "step": 178
    },
    {
      "epoch": 0.24441030892643795,
      "grad_norm": 0.12218151986598969,
      "learning_rate": 3.975565044907354e-05,
      "loss": 0.5023,
      "num_tokens": 141782005.0,
      "step": 179
    },
    {
      "epoch": 0.2457757296466974,
      "grad_norm": 0.1345020979642868,
      "learning_rate": 3.975127770626422e-05,
      "loss": 0.5451,
      "num_tokens": 142604020.0,
      "step": 180
    },
    {
      "epoch": 0.24714115036695683,
      "grad_norm": 0.12720581889152527,
      "learning_rate": 3.9746866455755385e-05,
      "loss": 0.5026,
      "num_tokens": 143376701.0,
      "step": 181
    },
    {
      "epoch": 0.24850657108721624,
      "grad_norm": 0.13804341852664948,
      "learning_rate": 3.974241670711632e-05,
      "loss": 0.502,
      "num_tokens": 144138031.0,
      "step": 182
    },
    {
      "epoch": 0.24987199180747569,
      "grad_norm": 0.1312447190284729,
      "learning_rate": 3.973792846999984e-05,
      "loss": 0.5098,
      "num_tokens": 144965455.0,
      "step": 183
    },
    {
      "epoch": 0.2512374125277351,
      "grad_norm": 0.13266997039318085,
      "learning_rate": 3.973340175414222e-05,
      "loss": 0.4828,
      "num_tokens": 145646325.0,
      "step": 184
    },
    {
      "epoch": 0.25260283324799454,
      "grad_norm": 0.1402488797903061,
      "learning_rate": 3.972883656936325e-05,
      "loss": 0.5123,
      "num_tokens": 146411168.0,
      "step": 185
    },
    {
      "epoch": 0.25396825396825395,
      "grad_norm": 0.12218815833330154,
      "learning_rate": 3.972423292556613e-05,
      "loss": 0.5036,
      "num_tokens": 147313146.0,
      "step": 186
    },
    {
      "epoch": 0.2553336746885134,
      "grad_norm": 0.15058304369449615,
      "learning_rate": 3.9719590832737516e-05,
      "loss": 0.5645,
      "num_tokens": 148150636.0,
      "step": 187
    },
    {
      "epoch": 0.25669909540877284,
      "grad_norm": 0.130282923579216,
      "learning_rate": 3.9714910300947454e-05,
      "loss": 0.5263,
      "num_tokens": 148961511.0,
      "step": 188
    },
    {
      "epoch": 0.25806451612903225,
      "grad_norm": 0.13835476338863373,
      "learning_rate": 3.9710191340349385e-05,
      "loss": 0.4973,
      "num_tokens": 149754009.0,
      "step": 189
    },
    {
      "epoch": 0.25942993684929166,
      "grad_norm": 0.1281115561723709,
      "learning_rate": 3.9705433961180115e-05,
      "loss": 0.496,
      "num_tokens": 150492877.0,
      "step": 190
    },
    {
      "epoch": 0.26079535756955113,
      "grad_norm": 0.12777435779571533,
      "learning_rate": 3.970063817375979e-05,
      "loss": 0.5025,
      "num_tokens": 151284637.0,
      "step": 191
    },
    {
      "epoch": 0.26216077828981055,
      "grad_norm": 0.12798313796520233,
      "learning_rate": 3.969580398849186e-05,
      "loss": 0.5049,
      "num_tokens": 152097333.0,
      "step": 192
    },
    {
      "epoch": 0.26352619901006996,
      "grad_norm": 0.13077402114868164,
      "learning_rate": 3.96909314158631e-05,
      "loss": 0.4931,
      "num_tokens": 152807242.0,
      "step": 193
    },
    {
      "epoch": 0.26489161973032943,
      "grad_norm": 0.1272391825914383,
      "learning_rate": 3.9686020466443514e-05,
      "loss": 0.4985,
      "num_tokens": 153601328.0,
      "step": 194
    },
    {
      "epoch": 0.26625704045058884,
      "grad_norm": 0.13038583099842072,
      "learning_rate": 3.968107115088641e-05,
      "loss": 0.5244,
      "num_tokens": 154434231.0,
      "step": 195
    },
    {
      "epoch": 0.26762246117084826,
      "grad_norm": 0.1343611180782318,
      "learning_rate": 3.96760834799283e-05,
      "loss": 0.5263,
      "num_tokens": 155275964.0,
      "step": 196
    },
    {
      "epoch": 0.26898788189110767,
      "grad_norm": 0.12146608531475067,
      "learning_rate": 3.967105746438887e-05,
      "loss": 0.5131,
      "num_tokens": 156100553.0,
      "step": 197
    },
    {
      "epoch": 0.27035330261136714,
      "grad_norm": 0.13174134492874146,
      "learning_rate": 3.966599311517103e-05,
      "loss": 0.4991,
      "num_tokens": 156837817.0,
      "step": 198
    },
    {
      "epoch": 0.27171872333162655,
      "grad_norm": 0.1273331195116043,
      "learning_rate": 3.9660890443260836e-05,
      "loss": 0.5121,
      "num_tokens": 157651844.0,
      "step": 199
    },
    {
      "epoch": 0.27308414405188597,
      "grad_norm": 0.1284930408000946,
      "learning_rate": 3.9655749459727465e-05,
      "loss": 0.509,
      "num_tokens": 158442369.0,
      "step": 200
    },
    {
      "epoch": 0.27444956477214544,
      "grad_norm": 0.12409322708845139,
      "learning_rate": 3.965057017572321e-05,
      "loss": 0.4789,
      "num_tokens": 159225029.0,
      "step": 201
    },
    {
      "epoch": 0.27581498549240485,
      "grad_norm": 0.12448086589574814,
      "learning_rate": 3.9645352602483456e-05,
      "loss": 0.4961,
      "num_tokens": 159992340.0,
      "step": 202
    },
    {
      "epoch": 0.27718040621266427,
      "grad_norm": 0.1381082981824875,
      "learning_rate": 3.964009675132662e-05,
      "loss": 0.5328,
      "num_tokens": 160873144.0,
      "step": 203
    },
    {
      "epoch": 0.27854582693292373,
      "grad_norm": 0.12574738264083862,
      "learning_rate": 3.963480263365421e-05,
      "loss": 0.5017,
      "num_tokens": 161617185.0,
      "step": 204
    },
    {
      "epoch": 0.27991124765318315,
      "grad_norm": 0.1312534362077713,
      "learning_rate": 3.962947026095069e-05,
      "loss": 0.5217,
      "num_tokens": 162454763.0,
      "step": 205
    },
    {
      "epoch": 0.28127666837344256,
      "grad_norm": 0.1470540165901184,
      "learning_rate": 3.9624099644783526e-05,
      "loss": 0.5189,
      "num_tokens": 163196513.0,
      "step": 206
    },
    {
      "epoch": 0.282642089093702,
      "grad_norm": 0.14042791724205017,
      "learning_rate": 3.9618690796803155e-05,
      "loss": 0.4874,
      "num_tokens": 163950509.0,
      "step": 207
    },
    {
      "epoch": 0.28400750981396145,
      "grad_norm": 0.1321112960577011,
      "learning_rate": 3.961324372874296e-05,
      "loss": 0.4916,
      "num_tokens": 164755296.0,
      "step": 208
    },
    {
      "epoch": 0.28537293053422086,
      "grad_norm": 0.17003978788852692,
      "learning_rate": 3.960775845241921e-05,
      "loss": 0.5136,
      "num_tokens": 165537274.0,
      "step": 209
    },
    {
      "epoch": 0.2867383512544803,
      "grad_norm": 0.1251276731491089,
      "learning_rate": 3.960223497973108e-05,
      "loss": 0.5234,
      "num_tokens": 166343431.0,
      "step": 210
    },
    {
      "epoch": 0.28810377197473974,
      "grad_norm": 0.15385177731513977,
      "learning_rate": 3.959667332266057e-05,
      "loss": 0.5017,
      "num_tokens": 167109702.0,
      "step": 211
    },
    {
      "epoch": 0.28946919269499916,
      "grad_norm": 0.1455739140510559,
      "learning_rate": 3.959107349327256e-05,
      "loss": 0.5051,
      "num_tokens": 167939138.0,
      "step": 212
    },
    {
      "epoch": 0.29083461341525857,
      "grad_norm": 0.13237203657627106,
      "learning_rate": 3.95854355037147e-05,
      "loss": 0.5176,
      "num_tokens": 168719781.0,
      "step": 213
    },
    {
      "epoch": 0.292200034135518,
      "grad_norm": 0.17229105532169342,
      "learning_rate": 3.9579759366217444e-05,
      "loss": 0.5261,
      "num_tokens": 169538687.0,
      "step": 214
    },
    {
      "epoch": 0.29356545485577745,
      "grad_norm": 0.14317819476127625,
      "learning_rate": 3.9574045093093983e-05,
      "loss": 0.5058,
      "num_tokens": 170421781.0,
      "step": 215
    },
    {
      "epoch": 0.29493087557603687,
      "grad_norm": 0.123870350420475,
      "learning_rate": 3.9568292696740236e-05,
      "loss": 0.483,
      "num_tokens": 171173402.0,
      "step": 216
    },
    {
      "epoch": 0.2962962962962963,
      "grad_norm": 0.14159098267555237,
      "learning_rate": 3.9562502189634846e-05,
      "loss": 0.4881,
      "num_tokens": 171991522.0,
      "step": 217
    },
    {
      "epoch": 0.29766171701655575,
      "grad_norm": 0.11987169086933136,
      "learning_rate": 3.95566735843391e-05,
      "loss": 0.4907,
      "num_tokens": 172832867.0,
      "step": 218
    },
    {
      "epoch": 0.29902713773681516,
      "grad_norm": 0.12954303622245789,
      "learning_rate": 3.9550806893496944e-05,
      "loss": 0.4705,
      "num_tokens": 173628377.0,
      "step": 219
    },
    {
      "epoch": 0.3003925584570746,
      "grad_norm": 0.12343389540910721,
      "learning_rate": 3.9544902129834945e-05,
      "loss": 0.5131,
      "num_tokens": 174421681.0,
      "step": 220
    },
    {
      "epoch": 0.301757979177334,
      "grad_norm": 0.12530112266540527,
      "learning_rate": 3.953895930616226e-05,
      "loss": 0.5045,
      "num_tokens": 175253289.0,
      "step": 221
    },
    {
      "epoch": 0.30312339989759346,
      "grad_norm": 0.12804356217384338,
      "learning_rate": 3.9532978435370594e-05,
      "loss": 0.5321,
      "num_tokens": 176094038.0,
      "step": 222
    },
    {
      "epoch": 0.3044888206178529,
      "grad_norm": 0.12803159654140472,
      "learning_rate": 3.952695953043422e-05,
      "loss": 0.5114,
      "num_tokens": 176868819.0,
      "step": 223
    },
    {
      "epoch": 0.3058542413381123,
      "grad_norm": 0.12620675563812256,
      "learning_rate": 3.952090260440989e-05,
      "loss": 0.4878,
      "num_tokens": 177629175.0,
      "step": 224
    },
    {
      "epoch": 0.30721966205837176,
      "grad_norm": 0.13803231716156006,
      "learning_rate": 3.951480767043683e-05,
      "loss": 0.4896,
      "num_tokens": 178471729.0,
      "step": 225
    },
    {
      "epoch": 0.30858508277863117,
      "grad_norm": 0.12397008389234543,
      "learning_rate": 3.950867474173675e-05,
      "loss": 0.4984,
      "num_tokens": 179322474.0,
      "step": 226
    },
    {
      "epoch": 0.3099505034988906,
      "grad_norm": 0.13756783306598663,
      "learning_rate": 3.950250383161375e-05,
      "loss": 0.5154,
      "num_tokens": 180109609.0,
      "step": 227
    },
    {
      "epoch": 0.31131592421915005,
      "grad_norm": 0.11957614123821259,
      "learning_rate": 3.9496294953454336e-05,
      "loss": 0.5021,
      "num_tokens": 180923069.0,
      "step": 228
    },
    {
      "epoch": 0.31268134493940947,
      "grad_norm": 0.13213731348514557,
      "learning_rate": 3.949004812072738e-05,
      "loss": 0.5109,
      "num_tokens": 181814846.0,
      "step": 229
    },
    {
      "epoch": 0.3140467656596689,
      "grad_norm": 0.12807555496692657,
      "learning_rate": 3.948376334698409e-05,
      "loss": 0.4812,
      "num_tokens": 182571366.0,
      "step": 230
    },
    {
      "epoch": 0.3154121863799283,
      "grad_norm": 0.13743239641189575,
      "learning_rate": 3.947744064585795e-05,
      "loss": 0.4967,
      "num_tokens": 183413056.0,
      "step": 231
    },
    {
      "epoch": 0.31677760710018776,
      "grad_norm": 0.13192838430404663,
      "learning_rate": 3.9471080031064765e-05,
      "loss": 0.4641,
      "num_tokens": 184197165.0,
      "step": 232
    },
    {
      "epoch": 0.3181430278204472,
      "grad_norm": 0.15293429791927338,
      "learning_rate": 3.946468151640255e-05,
      "loss": 0.4826,
      "num_tokens": 184963380.0,
      "step": 233
    },
    {
      "epoch": 0.3195084485407066,
      "grad_norm": 0.11773902922868729,
      "learning_rate": 3.945824511575157e-05,
      "loss": 0.4877,
      "num_tokens": 185712248.0,
      "step": 234
    },
    {
      "epoch": 0.32087386926096606,
      "grad_norm": 0.1344311535358429,
      "learning_rate": 3.945177084307423e-05,
      "loss": 0.5069,
      "num_tokens": 186543537.0,
      "step": 235
    },
    {
      "epoch": 0.3222392899812255,
      "grad_norm": 0.1352429986000061,
      "learning_rate": 3.944525871241514e-05,
      "loss": 0.5152,
      "num_tokens": 187344004.0,
      "step": 236
    },
    {
      "epoch": 0.3236047107014849,
      "grad_norm": 0.1438681036233902,
      "learning_rate": 3.9438708737900986e-05,
      "loss": 0.531,
      "num_tokens": 188181249.0,
      "step": 237
    },
    {
      "epoch": 0.3249701314217443,
      "grad_norm": 0.12606382369995117,
      "learning_rate": 3.94321209337406e-05,
      "loss": 0.4744,
      "num_tokens": 188952112.0,
      "step": 238
    },
    {
      "epoch": 0.32633555214200377,
      "grad_norm": 0.13270947337150574,
      "learning_rate": 3.942549531422483e-05,
      "loss": 0.5003,
      "num_tokens": 189709028.0,
      "step": 239
    },
    {
      "epoch": 0.3277009728622632,
      "grad_norm": 0.1238170862197876,
      "learning_rate": 3.941883189372659e-05,
      "loss": 0.4989,
      "num_tokens": 190513403.0,
      "step": 240
    },
    {
      "epoch": 0.3290663935825226,
      "grad_norm": 0.1332513839006424,
      "learning_rate": 3.941213068670078e-05,
      "loss": 0.491,
      "num_tokens": 191274413.0,
      "step": 241
    },
    {
      "epoch": 0.33043181430278207,
      "grad_norm": 0.12888021767139435,
      "learning_rate": 3.940539170768426e-05,
      "loss": 0.5153,
      "num_tokens": 192106428.0,
      "step": 242
    },
    {
      "epoch": 0.3317972350230415,
      "grad_norm": 0.15177400410175323,
      "learning_rate": 3.939861497129586e-05,
      "loss": 0.5327,
      "num_tokens": 192972804.0,
      "step": 243
    },
    {
      "epoch": 0.3331626557433009,
      "grad_norm": 0.12233538925647736,
      "learning_rate": 3.939180049223629e-05,
      "loss": 0.4963,
      "num_tokens": 193797629.0,
      "step": 244
    },
    {
      "epoch": 0.3345280764635603,
      "grad_norm": 0.15920040011405945,
      "learning_rate": 3.9384948285288145e-05,
      "loss": 0.5153,
      "num_tokens": 194603220.0,
      "step": 245
    },
    {
      "epoch": 0.3358934971838198,
      "grad_norm": 0.13951385021209717,
      "learning_rate": 3.937805836531585e-05,
      "loss": 0.4882,
      "num_tokens": 195362273.0,
      "step": 246
    },
    {
      "epoch": 0.3372589179040792,
      "grad_norm": 0.13732312619686127,
      "learning_rate": 3.937113074726567e-05,
      "loss": 0.4982,
      "num_tokens": 196180166.0,
      "step": 247
    },
    {
      "epoch": 0.3386243386243386,
      "grad_norm": 0.12930165231227875,
      "learning_rate": 3.9364165446165615e-05,
      "loss": 0.4687,
      "num_tokens": 196918032.0,
      "step": 248
    },
    {
      "epoch": 0.3399897593445981,
      "grad_norm": 0.1411222517490387,
      "learning_rate": 3.935716247712546e-05,
      "loss": 0.5127,
      "num_tokens": 197792328.0,
      "step": 249
    },
    {
      "epoch": 0.3413551800648575,
      "grad_norm": 0.15770424902439117,
      "learning_rate": 3.9350121855336694e-05,
      "loss": 0.4892,
      "num_tokens": 198560912.0,
      "step": 250
    },
    {
      "epoch": 0.3427206007851169,
      "grad_norm": 0.12383218854665756,
      "learning_rate": 3.9343043596072464e-05,
      "loss": 0.4866,
      "num_tokens": 199383511.0,
      "step": 251
    },
    {
      "epoch": 0.34408602150537637,
      "grad_norm": 0.14918413758277893,
      "learning_rate": 3.9335927714687584e-05,
      "loss": 0.4924,
      "num_tokens": 200146973.0,
      "step": 252
    },
    {
      "epoch": 0.3454514422256358,
      "grad_norm": 0.1356574147939682,
      "learning_rate": 3.9328774226618485e-05,
      "loss": 0.4911,
      "num_tokens": 200943710.0,
      "step": 253
    },
    {
      "epoch": 0.3468168629458952,
      "grad_norm": 0.13283181190490723,
      "learning_rate": 3.932158314738317e-05,
      "loss": 0.4973,
      "num_tokens": 201723952.0,
      "step": 254
    },
    {
      "epoch": 0.3481822836661546,
      "grad_norm": 0.13168925046920776,
      "learning_rate": 3.931435449258118e-05,
      "loss": 0.4978,
      "num_tokens": 202554154.0,
      "step": 255
    },
    {
      "epoch": 0.3495477043864141,
      "grad_norm": 0.14067862927913666,
      "learning_rate": 3.930708827789357e-05,
      "loss": 0.4744,
      "num_tokens": 203349315.0,
      "step": 256
    },
    {
      "epoch": 0.3509131251066735,
      "grad_norm": 0.13338786363601685,
      "learning_rate": 3.9299784519082894e-05,
      "loss": 0.4812,
      "num_tokens": 204137301.0,
      "step": 257
    },
    {
      "epoch": 0.3522785458269329,
      "grad_norm": 0.13378150761127472,
      "learning_rate": 3.929244323199313e-05,
      "loss": 0.503,
      "num_tokens": 204932127.0,
      "step": 258
    },
    {
      "epoch": 0.3536439665471924,
      "grad_norm": 0.13969217240810394,
      "learning_rate": 3.928506443254967e-05,
      "loss": 0.4853,
      "num_tokens": 205725045.0,
      "step": 259
    },
    {
      "epoch": 0.3550093872674518,
      "grad_norm": 0.11506263166666031,
      "learning_rate": 3.927764813675928e-05,
      "loss": 0.464,
      "num_tokens": 206541541.0,
      "step": 260
    },
    {
      "epoch": 0.3563748079877112,
      "grad_norm": 0.1311144381761551,
      "learning_rate": 3.9270194360710064e-05,
      "loss": 0.4849,
      "num_tokens": 207346244.0,
      "step": 261
    },
    {
      "epoch": 0.3577402287079706,
      "grad_norm": 0.13580457866191864,
      "learning_rate": 3.926270312057145e-05,
      "loss": 0.5353,
      "num_tokens": 208117201.0,
      "step": 262
    },
    {
      "epoch": 0.3591056494282301,
      "grad_norm": 0.12431619316339493,
      "learning_rate": 3.925517443259412e-05,
      "loss": 0.4918,
      "num_tokens": 208848612.0,
      "step": 263
    },
    {
      "epoch": 0.3604710701484895,
      "grad_norm": 0.13317179679870605,
      "learning_rate": 3.924760831310999e-05,
      "loss": 0.5084,
      "num_tokens": 209629191.0,
      "step": 264
    },
    {
      "epoch": 0.3618364908687489,
      "grad_norm": 0.1348157525062561,
      "learning_rate": 3.924000477853218e-05,
      "loss": 0.4773,
      "num_tokens": 210368355.0,
      "step": 265
    },
    {
      "epoch": 0.3632019115890084,
      "grad_norm": 0.12406640499830246,
      "learning_rate": 3.9232363845354974e-05,
      "loss": 0.4768,
      "num_tokens": 211064793.0,
      "step": 266
    },
    {
      "epoch": 0.3645673323092678,
      "grad_norm": 0.1336199939250946,
      "learning_rate": 3.922468553015379e-05,
      "loss": 0.4882,
      "num_tokens": 211871464.0,
      "step": 267
    },
    {
      "epoch": 0.3659327530295272,
      "grad_norm": 0.12421881407499313,
      "learning_rate": 3.9216969849585135e-05,
      "loss": 0.4991,
      "num_tokens": 212717177.0,
      "step": 268
    },
    {
      "epoch": 0.3672981737497866,
      "grad_norm": 0.13326798379421234,
      "learning_rate": 3.920921682038656e-05,
      "loss": 0.485,
      "num_tokens": 213529589.0,
      "step": 269
    },
    {
      "epoch": 0.3686635944700461,
      "grad_norm": 0.12430132925510406,
      "learning_rate": 3.9201426459376655e-05,
      "loss": 0.4935,
      "num_tokens": 214363269.0,
      "step": 270
    },
    {
      "epoch": 0.3700290151903055,
      "grad_norm": 0.12998712062835693,
      "learning_rate": 3.9193598783454984e-05,
      "loss": 0.487,
      "num_tokens": 215172877.0,
      "step": 271
    },
    {
      "epoch": 0.3713944359105649,
      "grad_norm": 0.13348448276519775,
      "learning_rate": 3.918573380960206e-05,
      "loss": 0.4903,
      "num_tokens": 216019108.0,
      "step": 272
    },
    {
      "epoch": 0.3727598566308244,
      "grad_norm": 0.1209225282073021,
      "learning_rate": 3.91778315548793e-05,
      "loss": 0.48,
      "num_tokens": 216834505.0,
      "step": 273
    },
    {
      "epoch": 0.3741252773510838,
      "grad_norm": 0.12041261792182922,
      "learning_rate": 3.9169892036429014e-05,
      "loss": 0.4903,
      "num_tokens": 217575081.0,
      "step": 274
    },
    {
      "epoch": 0.3754906980713432,
      "grad_norm": 0.13472823798656464,
      "learning_rate": 3.916191527147431e-05,
      "loss": 0.4721,
      "num_tokens": 218330904.0,
      "step": 275
    },
    {
      "epoch": 0.3768561187916027,
      "grad_norm": 0.12635375559329987,
      "learning_rate": 3.915390127731914e-05,
      "loss": 0.4854,
      "num_tokens": 219058547.0,
      "step": 276
    },
    {
      "epoch": 0.3782215395118621,
      "grad_norm": 0.13131175935268402,
      "learning_rate": 3.91458500713482e-05,
      "loss": 0.495,
      "num_tokens": 219874010.0,
      "step": 277
    },
    {
      "epoch": 0.3795869602321215,
      "grad_norm": 0.13184237480163574,
      "learning_rate": 3.913776167102688e-05,
      "loss": 0.4887,
      "num_tokens": 220660471.0,
      "step": 278
    },
    {
      "epoch": 0.38095238095238093,
      "grad_norm": 0.1258932501077652,
      "learning_rate": 3.9129636093901284e-05,
      "loss": 0.4672,
      "num_tokens": 221372658.0,
      "step": 279
    },
    {
      "epoch": 0.3823178016726404,
      "grad_norm": 0.13474169373512268,
      "learning_rate": 3.9121473357598175e-05,
      "loss": 0.5056,
      "num_tokens": 222145732.0,
      "step": 280
    },
    {
      "epoch": 0.3836832223928998,
      "grad_norm": 0.1323632299900055,
      "learning_rate": 3.91132734798249e-05,
      "loss": 0.4865,
      "num_tokens": 222988223.0,
      "step": 281
    },
    {
      "epoch": 0.38504864311315923,
      "grad_norm": 0.14055615663528442,
      "learning_rate": 3.910503647836939e-05,
      "loss": 0.5245,
      "num_tokens": 223759249.0,
      "step": 282
    },
    {
      "epoch": 0.3864140638334187,
      "grad_norm": 0.11983688175678253,
      "learning_rate": 3.9096762371100094e-05,
      "loss": 0.4985,
      "num_tokens": 224582759.0,
      "step": 283
    },
    {
      "epoch": 0.3877794845536781,
      "grad_norm": 0.1420964002609253,
      "learning_rate": 3.908845117596597e-05,
      "loss": 0.493,
      "num_tokens": 225371765.0,
      "step": 284
    },
    {
      "epoch": 0.3891449052739375,
      "grad_norm": 0.1425657719373703,
      "learning_rate": 3.908010291099643e-05,
      "loss": 0.4931,
      "num_tokens": 226168287.0,
      "step": 285
    },
    {
      "epoch": 0.39051032599419694,
      "grad_norm": 0.1410464197397232,
      "learning_rate": 3.907171759430128e-05,
      "loss": 0.5001,
      "num_tokens": 226997883.0,
      "step": 286
    },
    {
      "epoch": 0.3918757467144564,
      "grad_norm": 0.14300286769866943,
      "learning_rate": 3.9063295244070743e-05,
      "loss": 0.4814,
      "num_tokens": 227839102.0,
      "step": 287
    },
    {
      "epoch": 0.3932411674347158,
      "grad_norm": 0.13014666736125946,
      "learning_rate": 3.9054835878575336e-05,
      "loss": 0.4753,
      "num_tokens": 228610228.0,
      "step": 288
    },
    {
      "epoch": 0.39460658815497524,
      "grad_norm": 0.12609681487083435,
      "learning_rate": 3.904633951616589e-05,
      "loss": 0.4955,
      "num_tokens": 229452496.0,
      "step": 289
    },
    {
      "epoch": 0.3959720088752347,
      "grad_norm": 0.13189776241779327,
      "learning_rate": 3.90378061752735e-05,
      "loss": 0.5059,
      "num_tokens": 230218550.0,
      "step": 290
    },
    {
      "epoch": 0.3973374295954941,
      "grad_norm": 0.14208318293094635,
      "learning_rate": 3.902923587440947e-05,
      "loss": 0.501,
      "num_tokens": 231028034.0,
      "step": 291
    },
    {
      "epoch": 0.39870285031575353,
      "grad_norm": 0.13137570023536682,
      "learning_rate": 3.902062863216528e-05,
      "loss": 0.4766,
      "num_tokens": 231746624.0,
      "step": 292
    },
    {
      "epoch": 0.40006827103601295,
      "grad_norm": 0.1392088532447815,
      "learning_rate": 3.901198446721256e-05,
      "loss": 0.4878,
      "num_tokens": 232528366.0,
      "step": 293
    },
    {
      "epoch": 0.4014336917562724,
      "grad_norm": 0.12913568317890167,
      "learning_rate": 3.900330339830301e-05,
      "loss": 0.4733,
      "num_tokens": 233298687.0,
      "step": 294
    },
    {
      "epoch": 0.40279911247653183,
      "grad_norm": 0.13415347039699554,
      "learning_rate": 3.899458544426841e-05,
      "loss": 0.4848,
      "num_tokens": 234130385.0,
      "step": 295
    },
    {
      "epoch": 0.40416453319679124,
      "grad_norm": 0.12311084568500519,
      "learning_rate": 3.8985830624020533e-05,
      "loss": 0.4547,
      "num_tokens": 234846290.0,
      "step": 296
    },
    {
      "epoch": 0.4055299539170507,
      "grad_norm": 0.15119503438472748,
      "learning_rate": 3.897703895655115e-05,
      "loss": 0.5098,
      "num_tokens": 235669164.0,
      "step": 297
    },
    {
      "epoch": 0.4068953746373101,
      "grad_norm": 0.13264413177967072,
      "learning_rate": 3.8968210460931957e-05,
      "loss": 0.4728,
      "num_tokens": 236474167.0,
      "step": 298
    },
    {
      "epoch": 0.40826079535756954,
      "grad_norm": 0.13744403421878815,
      "learning_rate": 3.8959345156314524e-05,
      "loss": 0.4942,
      "num_tokens": 237347244.0,
      "step": 299
    },
    {
      "epoch": 0.409626216077829,
      "grad_norm": 0.11688806116580963,
      "learning_rate": 3.8950443061930295e-05,
      "loss": 0.4813,
      "num_tokens": 238150290.0,
      "step": 300
    },
    {
      "epoch": 0.4109916367980884,
      "grad_norm": 0.13775864243507385,
      "learning_rate": 3.89415041970905e-05,
      "loss": 0.4724,
      "num_tokens": 238979776.0,
      "step": 301
    },
    {
      "epoch": 0.41235705751834784,
      "grad_norm": 0.12806487083435059,
      "learning_rate": 3.8932528581186145e-05,
      "loss": 0.5006,
      "num_tokens": 239735656.0,
      "step": 302
    },
    {
      "epoch": 0.41372247823860725,
      "grad_norm": 0.15095582604408264,
      "learning_rate": 3.892351623368796e-05,
      "loss": 0.4968,
      "num_tokens": 240549603.0,
      "step": 303
    },
    {
      "epoch": 0.4150878989588667,
      "grad_norm": 0.1303720325231552,
      "learning_rate": 3.891446717414635e-05,
      "loss": 0.5057,
      "num_tokens": 241375585.0,
      "step": 304
    },
    {
      "epoch": 0.41645331967912613,
      "grad_norm": 0.1322101354598999,
      "learning_rate": 3.8905381422191385e-05,
      "loss": 0.5163,
      "num_tokens": 242195243.0,
      "step": 305
    },
    {
      "epoch": 0.41781874039938555,
      "grad_norm": 0.12502868473529816,
      "learning_rate": 3.8896258997532685e-05,
      "loss": 0.4753,
      "num_tokens": 242980105.0,
      "step": 306
    },
    {
      "epoch": 0.419184161119645,
      "grad_norm": 0.13660453259944916,
      "learning_rate": 3.8887099919959475e-05,
      "loss": 0.5135,
      "num_tokens": 243750825.0,
      "step": 307
    },
    {
      "epoch": 0.42054958183990443,
      "grad_norm": 0.12836667895317078,
      "learning_rate": 3.887790420934045e-05,
      "loss": 0.4733,
      "num_tokens": 244494702.0,
      "step": 308
    },
    {
      "epoch": 0.42191500256016384,
      "grad_norm": 0.11835846304893494,
      "learning_rate": 3.886867188562379e-05,
      "loss": 0.5001,
      "num_tokens": 245349162.0,
      "step": 309
    },
    {
      "epoch": 0.42328042328042326,
      "grad_norm": 0.13893607258796692,
      "learning_rate": 3.8859402968837106e-05,
      "loss": 0.5056,
      "num_tokens": 246109777.0,
      "step": 310
    },
    {
      "epoch": 0.4246458440006827,
      "grad_norm": 0.1339644491672516,
      "learning_rate": 3.885009747908737e-05,
      "loss": 0.4853,
      "num_tokens": 246917096.0,
      "step": 311
    },
    {
      "epoch": 0.42601126472094214,
      "grad_norm": 0.1216612160205841,
      "learning_rate": 3.884075543656092e-05,
      "loss": 0.4981,
      "num_tokens": 247691991.0,
      "step": 312
    },
    {
      "epoch": 0.42737668544120155,
      "grad_norm": 0.1269971877336502,
      "learning_rate": 3.883137686152336e-05,
      "loss": 0.4817,
      "num_tokens": 248462046.0,
      "step": 313
    },
    {
      "epoch": 0.428742106161461,
      "grad_norm": 0.1269690841436386,
      "learning_rate": 3.8821961774319555e-05,
      "loss": 0.4871,
      "num_tokens": 249240298.0,
      "step": 314
    },
    {
      "epoch": 0.43010752688172044,
      "grad_norm": 0.11924248933792114,
      "learning_rate": 3.881251019537358e-05,
      "loss": 0.4899,
      "num_tokens": 249970068.0,
      "step": 315
    },
    {
      "epoch": 0.43147294760197985,
      "grad_norm": 0.12083810567855835,
      "learning_rate": 3.880302214518866e-05,
      "loss": 0.5107,
      "num_tokens": 250746191.0,
      "step": 316
    },
    {
      "epoch": 0.43283836832223926,
      "grad_norm": 0.12901411950588226,
      "learning_rate": 3.879349764434716e-05,
      "loss": 0.5098,
      "num_tokens": 251526613.0,
      "step": 317
    },
    {
      "epoch": 0.43420378904249873,
      "grad_norm": 0.12913751602172852,
      "learning_rate": 3.878393671351048e-05,
      "loss": 0.4942,
      "num_tokens": 252361320.0,
      "step": 318
    },
    {
      "epoch": 0.43556920976275815,
      "grad_norm": 0.11597591638565063,
      "learning_rate": 3.877433937341908e-05,
      "loss": 0.4825,
      "num_tokens": 253162257.0,
      "step": 319
    },
    {
      "epoch": 0.43693463048301756,
      "grad_norm": 0.12332484871149063,
      "learning_rate": 3.876470564489239e-05,
      "loss": 0.4547,
      "num_tokens": 253965539.0,
      "step": 320
    },
    {
      "epoch": 0.43830005120327703,
      "grad_norm": 0.13576579093933105,
      "learning_rate": 3.875503554882878e-05,
      "loss": 0.5103,
      "num_tokens": 254851788.0,
      "step": 321
    },
    {
      "epoch": 0.43966547192353644,
      "grad_norm": 0.14030450582504272,
      "learning_rate": 3.8745329106205486e-05,
      "loss": 0.5313,
      "num_tokens": 255674573.0,
      "step": 322
    },
    {
      "epoch": 0.44103089264379586,
      "grad_norm": 0.13076677918434143,
      "learning_rate": 3.873558633807865e-05,
      "loss": 0.4868,
      "num_tokens": 256448118.0,
      "step": 323
    },
    {
      "epoch": 0.4423963133640553,
      "grad_norm": 0.13764405250549316,
      "learning_rate": 3.872580726558315e-05,
      "loss": 0.4574,
      "num_tokens": 257145651.0,
      "step": 324
    },
    {
      "epoch": 0.44376173408431474,
      "grad_norm": 0.13082413375377655,
      "learning_rate": 3.871599190993266e-05,
      "loss": 0.4842,
      "num_tokens": 257887720.0,
      "step": 325
    },
    {
      "epoch": 0.44512715480457415,
      "grad_norm": 0.13689535856246948,
      "learning_rate": 3.8706140292419546e-05,
      "loss": 0.489,
      "num_tokens": 258702814.0,
      "step": 326
    },
    {
      "epoch": 0.44649257552483357,
      "grad_norm": 0.13559181988239288,
      "learning_rate": 3.869625243441484e-05,
      "loss": 0.4904,
      "num_tokens": 259480497.0,
      "step": 327
    },
    {
      "epoch": 0.44785799624509304,
      "grad_norm": 0.1292814463376999,
      "learning_rate": 3.868632835736819e-05,
      "loss": 0.4759,
      "num_tokens": 260243718.0,
      "step": 328
    },
    {
      "epoch": 0.44922341696535245,
      "grad_norm": 0.13478286564350128,
      "learning_rate": 3.867636808280781e-05,
      "loss": 0.5025,
      "num_tokens": 261058418.0,
      "step": 329
    },
    {
      "epoch": 0.45058883768561186,
      "grad_norm": 0.12874124944210052,
      "learning_rate": 3.866637163234045e-05,
      "loss": 0.4616,
      "num_tokens": 261882902.0,
      "step": 330
    },
    {
      "epoch": 0.45195425840587133,
      "grad_norm": 0.13023746013641357,
      "learning_rate": 3.865633902765133e-05,
      "loss": 0.5064,
      "num_tokens": 262715524.0,
      "step": 331
    },
    {
      "epoch": 0.45331967912613075,
      "grad_norm": 0.13462327420711517,
      "learning_rate": 3.8646270290504085e-05,
      "loss": 0.4685,
      "num_tokens": 263420416.0,
      "step": 332
    },
    {
      "epoch": 0.45468509984639016,
      "grad_norm": 0.11657001823186874,
      "learning_rate": 3.8636165442740746e-05,
      "loss": 0.4823,
      "num_tokens": 264172768.0,
      "step": 333
    },
    {
      "epoch": 0.4560505205666496,
      "grad_norm": 0.19176708161830902,
      "learning_rate": 3.862602450628167e-05,
      "loss": 0.4843,
      "num_tokens": 265016520.0,
      "step": 334
    },
    {
      "epoch": 0.45741594128690904,
      "grad_norm": 0.12497951090335846,
      "learning_rate": 3.861584750312552e-05,
      "loss": 0.4993,
      "num_tokens": 265835105.0,
      "step": 335
    },
    {
      "epoch": 0.45878136200716846,
      "grad_norm": 0.1430606245994568,
      "learning_rate": 3.860563445534918e-05,
      "loss": 0.4637,
      "num_tokens": 266617436.0,
      "step": 336
    },
    {
      "epoch": 0.46014678272742787,
      "grad_norm": 0.12785018980503082,
      "learning_rate": 3.859538538510772e-05,
      "loss": 0.4903,
      "num_tokens": 267457455.0,
      "step": 337
    },
    {
      "epoch": 0.46151220344768734,
      "grad_norm": 0.11555064469575882,
      "learning_rate": 3.858510031463438e-05,
      "loss": 0.4664,
      "num_tokens": 268242258.0,
      "step": 338
    },
    {
      "epoch": 0.46287762416794676,
      "grad_norm": 0.14547286927700043,
      "learning_rate": 3.857477926624046e-05,
      "loss": 0.4943,
      "num_tokens": 269131393.0,
      "step": 339
    },
    {
      "epoch": 0.46424304488820617,
      "grad_norm": 0.12551918625831604,
      "learning_rate": 3.8564422262315336e-05,
      "loss": 0.5085,
      "num_tokens": 270011275.0,
      "step": 340
    },
    {
      "epoch": 0.4656084656084656,
      "grad_norm": 0.12917302548885345,
      "learning_rate": 3.855402932532637e-05,
      "loss": 0.4681,
      "num_tokens": 270767380.0,
      "step": 341
    },
    {
      "epoch": 0.46697388632872505,
      "grad_norm": 0.13373717665672302,
      "learning_rate": 3.854360047781886e-05,
      "loss": 0.4642,
      "num_tokens": 271529966.0,
      "step": 342
    },
    {
      "epoch": 0.46833930704898447,
      "grad_norm": 0.12656982243061066,
      "learning_rate": 3.8533135742416045e-05,
      "loss": 0.4679,
      "num_tokens": 272309263.0,
      "step": 343
    },
    {
      "epoch": 0.4697047277692439,
      "grad_norm": 0.1245032399892807,
      "learning_rate": 3.852263514181897e-05,
      "loss": 0.4736,
      "num_tokens": 273129564.0,
      "step": 344
    },
    {
      "epoch": 0.47107014848950335,
      "grad_norm": 0.13621291518211365,
      "learning_rate": 3.8512098698806506e-05,
      "loss": 0.511,
      "num_tokens": 273989173.0,
      "step": 345
    },
    {
      "epoch": 0.47243556920976276,
      "grad_norm": 0.11677691340446472,
      "learning_rate": 3.850152643623527e-05,
      "loss": 0.4775,
      "num_tokens": 274740053.0,
      "step": 346
    },
    {
      "epoch": 0.4738009899300222,
      "grad_norm": 0.14074920117855072,
      "learning_rate": 3.8490918377039576e-05,
      "loss": 0.4935,
      "num_tokens": 275525377.0,
      "step": 347
    },
    {
      "epoch": 0.47516641065028165,
      "grad_norm": 0.1210772916674614,
      "learning_rate": 3.8480274544231415e-05,
      "loss": 0.5006,
      "num_tokens": 276343288.0,
      "step": 348
    },
    {
      "epoch": 0.47653183137054106,
      "grad_norm": 0.1354205161333084,
      "learning_rate": 3.846959496090036e-05,
      "loss": 0.469,
      "num_tokens": 277146721.0,
      "step": 349
    },
    {
      "epoch": 0.4778972520908005,
      "grad_norm": 0.11725703626871109,
      "learning_rate": 3.8458879650213534e-05,
      "loss": 0.4861,
      "num_tokens": 277938948.0,
      "step": 350
    },
    {
      "epoch": 0.4792626728110599,
      "grad_norm": 0.1442546546459198,
      "learning_rate": 3.844812863541557e-05,
      "loss": 0.4968,
      "num_tokens": 278729851.0,
      "step": 351
    },
    {
      "epoch": 0.48062809353131936,
      "grad_norm": 0.12948621809482574,
      "learning_rate": 3.843734193982857e-05,
      "loss": 0.4966,
      "num_tokens": 279556826.0,
      "step": 352
    },
    {
      "epoch": 0.48199351425157877,
      "grad_norm": 0.13450227677822113,
      "learning_rate": 3.842651958685202e-05,
      "loss": 0.4837,
      "num_tokens": 280362910.0,
      "step": 353
    },
    {
      "epoch": 0.4833589349718382,
      "grad_norm": 0.12182062119245529,
      "learning_rate": 3.8415661599962746e-05,
      "loss": 0.4993,
      "num_tokens": 281155713.0,
      "step": 354
    },
    {
      "epoch": 0.48472435569209765,
      "grad_norm": 0.13660092651844025,
      "learning_rate": 3.8404768002714904e-05,
      "loss": 0.4916,
      "num_tokens": 281956614.0,
      "step": 355
    },
    {
      "epoch": 0.48608977641235707,
      "grad_norm": 0.12395047396421432,
      "learning_rate": 3.8393838818739874e-05,
      "loss": 0.5049,
      "num_tokens": 282810430.0,
      "step": 356
    },
    {
      "epoch": 0.4874551971326165,
      "grad_norm": 0.12018755078315735,
      "learning_rate": 3.838287407174626e-05,
      "loss": 0.4977,
      "num_tokens": 283636081.0,
      "step": 357
    },
    {
      "epoch": 0.4888206178528759,
      "grad_norm": 0.12301836907863617,
      "learning_rate": 3.8371873785519785e-05,
      "loss": 0.4932,
      "num_tokens": 284474091.0,
      "step": 358
    },
    {
      "epoch": 0.49018603857313536,
      "grad_norm": 0.12773624062538147,
      "learning_rate": 3.836083798392327e-05,
      "loss": 0.4794,
      "num_tokens": 285219505.0,
      "step": 359
    },
    {
      "epoch": 0.4915514592933948,
      "grad_norm": 0.1354844719171524,
      "learning_rate": 3.83497666908966e-05,
      "loss": 0.5002,
      "num_tokens": 285977446.0,
      "step": 360
    },
    {
      "epoch": 0.4929168800136542,
      "grad_norm": 0.12325770407915115,
      "learning_rate": 3.833865993045665e-05,
      "loss": 0.4596,
      "num_tokens": 286710703.0,
      "step": 361
    },
    {
      "epoch": 0.49428230073391366,
      "grad_norm": 0.1314113289117813,
      "learning_rate": 3.8327517726697194e-05,
      "loss": 0.4545,
      "num_tokens": 287451186.0,
      "step": 362
    },
    {
      "epoch": 0.4956477214541731,
      "grad_norm": 0.1299847513437271,
      "learning_rate": 3.8316340103788947e-05,
      "loss": 0.4847,
      "num_tokens": 288254717.0,
      "step": 363
    },
    {
      "epoch": 0.4970131421744325,
      "grad_norm": 0.13236990571022034,
      "learning_rate": 3.830512708597942e-05,
      "loss": 0.4716,
      "num_tokens": 289009025.0,
      "step": 364
    },
    {
      "epoch": 0.4983785628946919,
      "grad_norm": 0.11719946563243866,
      "learning_rate": 3.829387869759293e-05,
      "loss": 0.4866,
      "num_tokens": 289767068.0,
      "step": 365
    },
    {
      "epoch": 0.49974398361495137,
      "grad_norm": 0.14981193840503693,
      "learning_rate": 3.82825949630305e-05,
      "loss": 0.4791,
      "num_tokens": 290633366.0,
      "step": 366
    },
    {
      "epoch": 0.5011094043352108,
      "grad_norm": 0.11282334476709366,
      "learning_rate": 3.827127590676986e-05,
      "loss": 0.4619,
      "num_tokens": 291387788.0,
      "step": 367
    },
    {
      "epoch": 0.5024748250554703,
      "grad_norm": 0.13857316970825195,
      "learning_rate": 3.8259921553365324e-05,
      "loss": 0.4778,
      "num_tokens": 292209554.0,
      "step": 368
    },
    {
      "epoch": 0.5038402457757296,
      "grad_norm": 0.13223221898078918,
      "learning_rate": 3.8248531927447805e-05,
      "loss": 0.4964,
      "num_tokens": 293091921.0,
      "step": 369
    },
    {
      "epoch": 0.5052056664959891,
      "grad_norm": 0.12255145609378815,
      "learning_rate": 3.823710705372473e-05,
      "loss": 0.4645,
      "num_tokens": 293866345.0,
      "step": 370
    },
    {
      "epoch": 0.5065710872162486,
      "grad_norm": 0.1371115893125534,
      "learning_rate": 3.822564695697998e-05,
      "loss": 0.4494,
      "num_tokens": 294644548.0,
      "step": 371
    },
    {
      "epoch": 0.5079365079365079,
      "grad_norm": 0.11735193431377411,
      "learning_rate": 3.8214151662073854e-05,
      "loss": 0.4902,
      "num_tokens": 295474477.0,
      "step": 372
    },
    {
      "epoch": 0.5093019286567674,
      "grad_norm": 0.1430472880601883,
      "learning_rate": 3.8202621193942976e-05,
      "loss": 0.506,
      "num_tokens": 296307758.0,
      "step": 373
    },
    {
      "epoch": 0.5106673493770268,
      "grad_norm": 0.13349537551403046,
      "learning_rate": 3.819105557760032e-05,
      "loss": 0.4681,
      "num_tokens": 297089759.0,
      "step": 374
    },
    {
      "epoch": 0.5120327700972862,
      "grad_norm": 0.13145381212234497,
      "learning_rate": 3.8179454838135066e-05,
      "loss": 0.5087,
      "num_tokens": 297915462.0,
      "step": 375
    },
    {
      "epoch": 0.5133981908175457,
      "grad_norm": 0.11589576303958893,
      "learning_rate": 3.8167819000712615e-05,
      "loss": 0.453,
      "num_tokens": 298677243.0,
      "step": 376
    },
    {
      "epoch": 0.514763611537805,
      "grad_norm": 0.12885381281375885,
      "learning_rate": 3.815614809057448e-05,
      "loss": 0.4548,
      "num_tokens": 299450788.0,
      "step": 377
    },
    {
      "epoch": 0.5161290322580645,
      "grad_norm": 0.11193668097257614,
      "learning_rate": 3.814444213303828e-05,
      "loss": 0.4887,
      "num_tokens": 300260294.0,
      "step": 378
    },
    {
      "epoch": 0.517494452978324,
      "grad_norm": 0.11762429028749466,
      "learning_rate": 3.8132701153497655e-05,
      "loss": 0.47,
      "num_tokens": 301013018.0,
      "step": 379
    },
    {
      "epoch": 0.5188598736985833,
      "grad_norm": 0.13105058670043945,
      "learning_rate": 3.8120925177422196e-05,
      "loss": 0.4841,
      "num_tokens": 301735232.0,
      "step": 380
    },
    {
      "epoch": 0.5202252944188428,
      "grad_norm": 0.13202178478240967,
      "learning_rate": 3.810911423035745e-05,
      "loss": 0.5094,
      "num_tokens": 302537556.0,
      "step": 381
    },
    {
      "epoch": 0.5215907151391023,
      "grad_norm": 0.12210100144147873,
      "learning_rate": 3.809726833792479e-05,
      "loss": 0.4648,
      "num_tokens": 303362755.0,
      "step": 382
    },
    {
      "epoch": 0.5229561358593616,
      "grad_norm": 0.12170364707708359,
      "learning_rate": 3.808538752582142e-05,
      "loss": 0.4873,
      "num_tokens": 304173699.0,
      "step": 383
    },
    {
      "epoch": 0.5243215565796211,
      "grad_norm": 0.12187743186950684,
      "learning_rate": 3.8073471819820294e-05,
      "loss": 0.4735,
      "num_tokens": 304909615.0,
      "step": 384
    },
    {
      "epoch": 0.5256869772998806,
      "grad_norm": 0.11484185606241226,
      "learning_rate": 3.8061521245770044e-05,
      "loss": 0.4644,
      "num_tokens": 305695795.0,
      "step": 385
    },
    {
      "epoch": 0.5270523980201399,
      "grad_norm": 0.13224932551383972,
      "learning_rate": 3.8049535829594956e-05,
      "loss": 0.4985,
      "num_tokens": 306501621.0,
      "step": 386
    },
    {
      "epoch": 0.5284178187403994,
      "grad_norm": 0.11923787742853165,
      "learning_rate": 3.803751559729489e-05,
      "loss": 0.4607,
      "num_tokens": 307268646.0,
      "step": 387
    },
    {
      "epoch": 0.5297832394606589,
      "grad_norm": 0.13633660972118378,
      "learning_rate": 3.802546057494524e-05,
      "loss": 0.4993,
      "num_tokens": 308088492.0,
      "step": 388
    },
    {
      "epoch": 0.5311486601809182,
      "grad_norm": 0.12569768726825714,
      "learning_rate": 3.801337078869686e-05,
      "loss": 0.4813,
      "num_tokens": 308904659.0,
      "step": 389
    },
    {
      "epoch": 0.5325140809011777,
      "grad_norm": 0.11979793757200241,
      "learning_rate": 3.8001246264776025e-05,
      "loss": 0.4821,
      "num_tokens": 309630742.0,
      "step": 390
    },
    {
      "epoch": 0.5338795016214372,
      "grad_norm": 0.11963630467653275,
      "learning_rate": 3.798908702948437e-05,
      "loss": 0.4853,
      "num_tokens": 310445167.0,
      "step": 391
    },
    {
      "epoch": 0.5352449223416965,
      "grad_norm": 0.12224148213863373,
      "learning_rate": 3.797689310919883e-05,
      "loss": 0.4814,
      "num_tokens": 311256291.0,
      "step": 392
    },
    {
      "epoch": 0.536610343061956,
      "grad_norm": 0.11214417964220047,
      "learning_rate": 3.7964664530371565e-05,
      "loss": 0.4676,
      "num_tokens": 311991214.0,
      "step": 393
    },
    {
      "epoch": 0.5379757637822153,
      "grad_norm": 0.11460646241903305,
      "learning_rate": 3.795240131952992e-05,
      "loss": 0.4536,
      "num_tokens": 312728136.0,
      "step": 394
    },
    {
      "epoch": 0.5393411845024748,
      "grad_norm": 0.11602908372879028,
      "learning_rate": 3.7940103503276383e-05,
      "loss": 0.4485,
      "num_tokens": 313484090.0,
      "step": 395
    },
    {
      "epoch": 0.5407066052227343,
      "grad_norm": 0.6019019484519958,
      "learning_rate": 3.792777110828852e-05,
      "loss": 0.4895,
      "num_tokens": 314299944.0,
      "step": 396
    },
    {
      "epoch": 0.5420720259429936,
      "grad_norm": 0.13061603903770447,
      "learning_rate": 3.791540416131887e-05,
      "loss": 0.5064,
      "num_tokens": 315074368.0,
      "step": 397
    },
    {
      "epoch": 0.5434374466632531,
      "grad_norm": 0.11932376027107239,
      "learning_rate": 3.7903002689194976e-05,
      "loss": 0.4743,
      "num_tokens": 315835589.0,
      "step": 398
    },
    {
      "epoch": 0.5448028673835126,
      "grad_norm": 0.1272733360528946,
      "learning_rate": 3.7890566718819216e-05,
      "loss": 0.4886,
      "num_tokens": 316600285.0,
      "step": 399
    },
    {
      "epoch": 0.5461682881037719,
      "grad_norm": 0.11871752142906189,
      "learning_rate": 3.787809627716887e-05,
      "loss": 0.4739,
      "num_tokens": 317360708.0,
      "step": 400
    },
    {
      "epoch": 0.5475337088240314,
      "grad_norm": 0.1296294778585434,
      "learning_rate": 3.786559139129594e-05,
      "loss": 0.4958,
      "num_tokens": 318183562.0,
      "step": 401
    },
    {
      "epoch": 0.5488991295442909,
      "grad_norm": 0.1361265778541565,
      "learning_rate": 3.785305208832717e-05,
      "loss": 0.4964,
      "num_tokens": 318924725.0,
      "step": 402
    },
    {
      "epoch": 0.5502645502645502,
      "grad_norm": 0.12215608358383179,
      "learning_rate": 3.784047839546399e-05,
      "loss": 0.493,
      "num_tokens": 319715516.0,
      "step": 403
    },
    {
      "epoch": 0.5516299709848097,
      "grad_norm": 0.12307325750589371,
      "learning_rate": 3.7827870339982385e-05,
      "loss": 0.4704,
      "num_tokens": 320524087.0,
      "step": 404
    },
    {
      "epoch": 0.5529953917050692,
      "grad_norm": 0.1168593093752861,
      "learning_rate": 3.781522794923291e-05,
      "loss": 0.4671,
      "num_tokens": 321320871.0,
      "step": 405
    },
    {
      "epoch": 0.5543608124253285,
      "grad_norm": 0.13736839592456818,
      "learning_rate": 3.780255125064061e-05,
      "loss": 0.4528,
      "num_tokens": 322123574.0,
      "step": 406
    },
    {
      "epoch": 0.555726233145588,
      "grad_norm": 0.13100117444992065,
      "learning_rate": 3.7789840271704925e-05,
      "loss": 0.4945,
      "num_tokens": 322942375.0,
      "step": 407
    },
    {
      "epoch": 0.5570916538658475,
      "grad_norm": 0.1321336179971695,
      "learning_rate": 3.777709503999968e-05,
      "loss": 0.464,
      "num_tokens": 323711214.0,
      "step": 408
    },
    {
      "epoch": 0.5584570745861068,
      "grad_norm": 0.1456555426120758,
      "learning_rate": 3.776431558317302e-05,
      "loss": 0.4781,
      "num_tokens": 324410229.0,
      "step": 409
    },
    {
      "epoch": 0.5598224953063663,
      "grad_norm": 0.1266820728778839,
      "learning_rate": 3.775150192894729e-05,
      "loss": 0.4813,
      "num_tokens": 325227916.0,
      "step": 410
    },
    {
      "epoch": 0.5611879160266257,
      "grad_norm": 0.13523218035697937,
      "learning_rate": 3.773865410511906e-05,
      "loss": 0.4836,
      "num_tokens": 326060576.0,
      "step": 411
    },
    {
      "epoch": 0.5625533367468851,
      "grad_norm": 0.12450605630874634,
      "learning_rate": 3.7725772139559e-05,
      "loss": 0.4643,
      "num_tokens": 326847413.0,
      "step": 412
    },
    {
      "epoch": 0.5639187574671446,
      "grad_norm": 0.13270050287246704,
      "learning_rate": 3.771285606021184e-05,
      "loss": 0.4863,
      "num_tokens": 327647516.0,
      "step": 413
    },
    {
      "epoch": 0.565284178187404,
      "grad_norm": 0.1361543834209442,
      "learning_rate": 3.769990589509635e-05,
      "loss": 0.4844,
      "num_tokens": 328483286.0,
      "step": 414
    },
    {
      "epoch": 0.5666495989076634,
      "grad_norm": 0.12508603930473328,
      "learning_rate": 3.768692167230519e-05,
      "loss": 0.4586,
      "num_tokens": 329171505.0,
      "step": 415
    },
    {
      "epoch": 0.5680150196279229,
      "grad_norm": 0.12660188972949982,
      "learning_rate": 3.7673903420004946e-05,
      "loss": 0.4882,
      "num_tokens": 329980802.0,
      "step": 416
    },
    {
      "epoch": 0.5693804403481822,
      "grad_norm": 0.130965918302536,
      "learning_rate": 3.7660851166436e-05,
      "loss": 0.4894,
      "num_tokens": 330854004.0,
      "step": 417
    },
    {
      "epoch": 0.5707458610684417,
      "grad_norm": 0.12567926943302155,
      "learning_rate": 3.764776493991249e-05,
      "loss": 0.4465,
      "num_tokens": 331543497.0,
      "step": 418
    },
    {
      "epoch": 0.5721112817887012,
      "grad_norm": 0.13441585004329681,
      "learning_rate": 3.763464476882226e-05,
      "loss": 0.5024,
      "num_tokens": 332409815.0,
      "step": 419
    },
    {
      "epoch": 0.5734767025089605,
      "grad_norm": 0.13385187089443207,
      "learning_rate": 3.76214906816268e-05,
      "loss": 0.4622,
      "num_tokens": 333197876.0,
      "step": 420
    },
    {
      "epoch": 0.57484212322922,
      "grad_norm": 0.12185902148485184,
      "learning_rate": 3.7608302706861154e-05,
      "loss": 0.4518,
      "num_tokens": 333955160.0,
      "step": 421
    },
    {
      "epoch": 0.5762075439494795,
      "grad_norm": 0.13713252544403076,
      "learning_rate": 3.759508087313388e-05,
      "loss": 0.4804,
      "num_tokens": 334753387.0,
      "step": 422
    },
    {
      "epoch": 0.5775729646697388,
      "grad_norm": 0.12268492579460144,
      "learning_rate": 3.7581825209127e-05,
      "loss": 0.4702,
      "num_tokens": 335521237.0,
      "step": 423
    },
    {
      "epoch": 0.5789383853899983,
      "grad_norm": 0.13871823251247406,
      "learning_rate": 3.756853574359592e-05,
      "loss": 0.5016,
      "num_tokens": 336317735.0,
      "step": 424
    },
    {
      "epoch": 0.5803038061102577,
      "grad_norm": 0.1284119188785553,
      "learning_rate": 3.7555212505369355e-05,
      "loss": 0.4524,
      "num_tokens": 337027844.0,
      "step": 425
    },
    {
      "epoch": 0.5816692268305171,
      "grad_norm": 0.1342606395483017,
      "learning_rate": 3.7541855523349306e-05,
      "loss": 0.4413,
      "num_tokens": 337737812.0,
      "step": 426
    },
    {
      "epoch": 0.5830346475507766,
      "grad_norm": 0.12945838272571564,
      "learning_rate": 3.7528464826510965e-05,
      "loss": 0.4982,
      "num_tokens": 338501994.0,
      "step": 427
    },
    {
      "epoch": 0.584400068271036,
      "grad_norm": 0.13038478791713715,
      "learning_rate": 3.751504044390264e-05,
      "loss": 0.488,
      "num_tokens": 339341284.0,
      "step": 428
    },
    {
      "epoch": 0.5857654889912954,
      "grad_norm": 0.13434827327728271,
      "learning_rate": 3.750158240464576e-05,
      "loss": 0.4935,
      "num_tokens": 340175933.0,
      "step": 429
    },
    {
      "epoch": 0.5871309097115549,
      "grad_norm": 0.12579165399074554,
      "learning_rate": 3.7488090737934714e-05,
      "loss": 0.4633,
      "num_tokens": 340897665.0,
      "step": 430
    },
    {
      "epoch": 0.5884963304318143,
      "grad_norm": 0.1354549378156662,
      "learning_rate": 3.7474565473036886e-05,
      "loss": 0.4963,
      "num_tokens": 341765677.0,
      "step": 431
    },
    {
      "epoch": 0.5898617511520737,
      "grad_norm": 0.12968935072422028,
      "learning_rate": 3.74610066392925e-05,
      "loss": 0.4587,
      "num_tokens": 342537361.0,
      "step": 432
    },
    {
      "epoch": 0.5912271718723332,
      "grad_norm": 0.12954206764698029,
      "learning_rate": 3.744741426611464e-05,
      "loss": 0.4713,
      "num_tokens": 343396334.0,
      "step": 433
    },
    {
      "epoch": 0.5925925925925926,
      "grad_norm": 0.11439257860183716,
      "learning_rate": 3.7433788382989106e-05,
      "loss": 0.4405,
      "num_tokens": 344178200.0,
      "step": 434
    },
    {
      "epoch": 0.593958013312852,
      "grad_norm": 0.15426005423069,
      "learning_rate": 3.742012901947443e-05,
      "loss": 0.4866,
      "num_tokens": 344973474.0,
      "step": 435
    },
    {
      "epoch": 0.5953234340331115,
      "grad_norm": 0.1279030740261078,
      "learning_rate": 3.7406436205201755e-05,
      "loss": 0.4855,
      "num_tokens": 345799515.0,
      "step": 436
    },
    {
      "epoch": 0.5966888547533709,
      "grad_norm": 0.11287161707878113,
      "learning_rate": 3.7392709969874785e-05,
      "loss": 0.4718,
      "num_tokens": 346597671.0,
      "step": 437
    },
    {
      "epoch": 0.5980542754736303,
      "grad_norm": 0.12305533140897751,
      "learning_rate": 3.737895034326973e-05,
      "loss": 0.4714,
      "num_tokens": 347369229.0,
      "step": 438
    },
    {
      "epoch": 0.5994196961938898,
      "grad_norm": 0.12582069635391235,
      "learning_rate": 3.736515735523523e-05,
      "loss": 0.4595,
      "num_tokens": 348179454.0,
      "step": 439
    },
    {
      "epoch": 0.6007851169141492,
      "grad_norm": 0.11667640507221222,
      "learning_rate": 3.735133103569231e-05,
      "loss": 0.4362,
      "num_tokens": 348940907.0,
      "step": 440
    },
    {
      "epoch": 0.6021505376344086,
      "grad_norm": 0.11790656298398972,
      "learning_rate": 3.733747141463428e-05,
      "loss": 0.4432,
      "num_tokens": 349655955.0,
      "step": 441
    },
    {
      "epoch": 0.603515958354668,
      "grad_norm": 0.13484777510166168,
      "learning_rate": 3.7323578522126704e-05,
      "loss": 0.4883,
      "num_tokens": 350450446.0,
      "step": 442
    },
    {
      "epoch": 0.6048813790749274,
      "grad_norm": 0.1377308964729309,
      "learning_rate": 3.730965238830732e-05,
      "loss": 0.4922,
      "num_tokens": 351250119.0,
      "step": 443
    },
    {
      "epoch": 0.6062467997951869,
      "grad_norm": 0.13560475409030914,
      "learning_rate": 3.729569304338597e-05,
      "loss": 0.476,
      "num_tokens": 352064552.0,
      "step": 444
    },
    {
      "epoch": 0.6076122205154463,
      "grad_norm": 0.1435183733701706,
      "learning_rate": 3.7281700517644554e-05,
      "loss": 0.4754,
      "num_tokens": 352843500.0,
      "step": 445
    },
    {
      "epoch": 0.6089776412357057,
      "grad_norm": 0.1499176323413849,
      "learning_rate": 3.726767484143694e-05,
      "loss": 0.49,
      "num_tokens": 353626154.0,
      "step": 446
    },
    {
      "epoch": 0.6103430619559652,
      "grad_norm": 0.1420065313577652,
      "learning_rate": 3.72536160451889e-05,
      "loss": 0.4789,
      "num_tokens": 354390317.0,
      "step": 447
    },
    {
      "epoch": 0.6117084826762246,
      "grad_norm": 0.11884896457195282,
      "learning_rate": 3.7239524159398076e-05,
      "loss": 0.4543,
      "num_tokens": 355143363.0,
      "step": 448
    },
    {
      "epoch": 0.613073903396484,
      "grad_norm": 0.12743939459323883,
      "learning_rate": 3.722539921463387e-05,
      "loss": 0.4554,
      "num_tokens": 355990666.0,
      "step": 449
    },
    {
      "epoch": 0.6144393241167435,
      "grad_norm": 0.14746026694774628,
      "learning_rate": 3.7211241241537414e-05,
      "loss": 0.4963,
      "num_tokens": 356868742.0,
      "step": 450
    },
    {
      "epoch": 0.6158047448370029,
      "grad_norm": 0.11990251392126083,
      "learning_rate": 3.7197050270821485e-05,
      "loss": 0.4738,
      "num_tokens": 357657534.0,
      "step": 451
    },
    {
      "epoch": 0.6171701655572623,
      "grad_norm": 0.13858865201473236,
      "learning_rate": 3.7182826333270436e-05,
      "loss": 0.513,
      "num_tokens": 358519845.0,
      "step": 452
    },
    {
      "epoch": 0.6185355862775218,
      "grad_norm": 0.13692766427993774,
      "learning_rate": 3.7168569459740126e-05,
      "loss": 0.4827,
      "num_tokens": 359296634.0,
      "step": 453
    },
    {
      "epoch": 0.6199010069977812,
      "grad_norm": 0.11928263306617737,
      "learning_rate": 3.715427968115789e-05,
      "loss": 0.4572,
      "num_tokens": 360083788.0,
      "step": 454
    },
    {
      "epoch": 0.6212664277180406,
      "grad_norm": 0.1335475742816925,
      "learning_rate": 3.713995702852241e-05,
      "loss": 0.4775,
      "num_tokens": 360850110.0,
      "step": 455
    },
    {
      "epoch": 0.6226318484383001,
      "grad_norm": 0.12674498558044434,
      "learning_rate": 3.71256015329037e-05,
      "loss": 0.4945,
      "num_tokens": 361661721.0,
      "step": 456
    },
    {
      "epoch": 0.6239972691585595,
      "grad_norm": 0.12268147617578506,
      "learning_rate": 3.711121322544302e-05,
      "loss": 0.4568,
      "num_tokens": 362368444.0,
      "step": 457
    },
    {
      "epoch": 0.6253626898788189,
      "grad_norm": 0.12003988772630692,
      "learning_rate": 3.709679213735283e-05,
      "loss": 0.462,
      "num_tokens": 363131376.0,
      "step": 458
    },
    {
      "epoch": 0.6267281105990783,
      "grad_norm": 0.12207403033971786,
      "learning_rate": 3.708233829991663e-05,
      "loss": 0.4894,
      "num_tokens": 363968705.0,
      "step": 459
    },
    {
      "epoch": 0.6280935313193378,
      "grad_norm": 0.12085213512182236,
      "learning_rate": 3.706785174448905e-05,
      "loss": 0.4422,
      "num_tokens": 364723139.0,
      "step": 460
    },
    {
      "epoch": 0.6294589520395972,
      "grad_norm": 0.11593151837587357,
      "learning_rate": 3.7053332502495636e-05,
      "loss": 0.467,
      "num_tokens": 365546943.0,
      "step": 461
    },
    {
      "epoch": 0.6308243727598566,
      "grad_norm": 0.12324142456054688,
      "learning_rate": 3.703878060543286e-05,
      "loss": 0.4661,
      "num_tokens": 366323327.0,
      "step": 462
    },
    {
      "epoch": 0.6321897934801161,
      "grad_norm": 0.1111968606710434,
      "learning_rate": 3.702419608486803e-05,
      "loss": 0.47,
      "num_tokens": 367094074.0,
      "step": 463
    },
    {
      "epoch": 0.6335552142003755,
      "grad_norm": 0.11477546393871307,
      "learning_rate": 3.700957897243924e-05,
      "loss": 0.4512,
      "num_tokens": 367895419.0,
      "step": 464
    },
    {
      "epoch": 0.6349206349206349,
      "grad_norm": 0.11551906168460846,
      "learning_rate": 3.6994929299855254e-05,
      "loss": 0.4705,
      "num_tokens": 368683099.0,
      "step": 465
    },
    {
      "epoch": 0.6362860556408944,
      "grad_norm": 0.12202318012714386,
      "learning_rate": 3.6980247098895485e-05,
      "loss": 0.4627,
      "num_tokens": 369514780.0,
      "step": 466
    },
    {
      "epoch": 0.6376514763611538,
      "grad_norm": 0.11507921665906906,
      "learning_rate": 3.696553240140993e-05,
      "loss": 0.4776,
      "num_tokens": 370222242.0,
      "step": 467
    },
    {
      "epoch": 0.6390168970814132,
      "grad_norm": 0.12866154313087463,
      "learning_rate": 3.695078523931903e-05,
      "loss": 0.5156,
      "num_tokens": 371051620.0,
      "step": 468
    },
    {
      "epoch": 0.6403823178016727,
      "grad_norm": 0.12623411417007446,
      "learning_rate": 3.693600564461371e-05,
      "loss": 0.4888,
      "num_tokens": 371832582.0,
      "step": 469
    },
    {
      "epoch": 0.6417477385219321,
      "grad_norm": 0.11496389657258987,
      "learning_rate": 3.6921193649355206e-05,
      "loss": 0.4728,
      "num_tokens": 372625383.0,
      "step": 470
    },
    {
      "epoch": 0.6431131592421915,
      "grad_norm": 0.1152229830622673,
      "learning_rate": 3.6906349285675065e-05,
      "loss": 0.4394,
      "num_tokens": 373417309.0,
      "step": 471
    },
    {
      "epoch": 0.644478579962451,
      "grad_norm": 0.11782772094011307,
      "learning_rate": 3.6891472585775034e-05,
      "loss": 0.4699,
      "num_tokens": 374217424.0,
      "step": 472
    },
    {
      "epoch": 0.6458440006827103,
      "grad_norm": 0.13930511474609375,
      "learning_rate": 3.687656358192702e-05,
      "loss": 0.46,
      "num_tokens": 374934019.0,
      "step": 473
    },
    {
      "epoch": 0.6472094214029698,
      "grad_norm": 0.1148938462138176,
      "learning_rate": 3.6861622306473e-05,
      "loss": 0.4769,
      "num_tokens": 375737112.0,
      "step": 474
    },
    {
      "epoch": 0.6485748421232292,
      "grad_norm": 0.1323624849319458,
      "learning_rate": 3.684664879182496e-05,
      "loss": 0.4653,
      "num_tokens": 376514011.0,
      "step": 475
    },
    {
      "epoch": 0.6499402628434886,
      "grad_norm": 0.12943492829799652,
      "learning_rate": 3.6831643070464806e-05,
      "loss": 0.4778,
      "num_tokens": 377275352.0,
      "step": 476
    },
    {
      "epoch": 0.6513056835637481,
      "grad_norm": 0.12270896881818771,
      "learning_rate": 3.681660517494436e-05,
      "loss": 0.4668,
      "num_tokens": 378083152.0,
      "step": 477
    },
    {
      "epoch": 0.6526711042840075,
      "grad_norm": 0.11453115195035934,
      "learning_rate": 3.680153513788517e-05,
      "loss": 0.4564,
      "num_tokens": 378819068.0,
      "step": 478
    },
    {
      "epoch": 0.6540365250042669,
      "grad_norm": 0.114867202937603,
      "learning_rate": 3.678643299197855e-05,
      "loss": 0.4656,
      "num_tokens": 379545024.0,
      "step": 479
    },
    {
      "epoch": 0.6554019457245264,
      "grad_norm": 0.14610740542411804,
      "learning_rate": 3.677129876998548e-05,
      "loss": 0.4567,
      "num_tokens": 380337327.0,
      "step": 480
    },
    {
      "epoch": 0.6567673664447858,
      "grad_norm": 0.11601844429969788,
      "learning_rate": 3.675613250473648e-05,
      "loss": 0.4728,
      "num_tokens": 381179608.0,
      "step": 481
    },
    {
      "epoch": 0.6581327871650452,
      "grad_norm": 0.11403773725032806,
      "learning_rate": 3.6740934229131606e-05,
      "loss": 0.4827,
      "num_tokens": 381977225.0,
      "step": 482
    },
    {
      "epoch": 0.6594982078853047,
      "grad_norm": 0.11540230363607407,
      "learning_rate": 3.672570397614036e-05,
      "loss": 0.4703,
      "num_tokens": 382738745.0,
      "step": 483
    },
    {
      "epoch": 0.6608636286055641,
      "grad_norm": 0.12612389028072357,
      "learning_rate": 3.67104417788016e-05,
      "loss": 0.4817,
      "num_tokens": 383526241.0,
      "step": 484
    },
    {
      "epoch": 0.6622290493258235,
      "grad_norm": 0.12272551655769348,
      "learning_rate": 3.669514767022349e-05,
      "loss": 0.4713,
      "num_tokens": 384332792.0,
      "step": 485
    },
    {
      "epoch": 0.663594470046083,
      "grad_norm": 0.1126057356595993,
      "learning_rate": 3.6679821683583396e-05,
      "loss": 0.4412,
      "num_tokens": 385039566.0,
      "step": 486
    },
    {
      "epoch": 0.6649598907663424,
      "grad_norm": 0.10870335251092911,
      "learning_rate": 3.666446385212787e-05,
      "loss": 0.4578,
      "num_tokens": 385819984.0,
      "step": 487
    },
    {
      "epoch": 0.6663253114866018,
      "grad_norm": 0.11402326822280884,
      "learning_rate": 3.664907420917252e-05,
      "loss": 0.4628,
      "num_tokens": 386634372.0,
      "step": 488
    },
    {
      "epoch": 0.6676907322068613,
      "grad_norm": 0.13265350461006165,
      "learning_rate": 3.6633652788101975e-05,
      "loss": 0.4891,
      "num_tokens": 387500745.0,
      "step": 489
    },
    {
      "epoch": 0.6690561529271206,
      "grad_norm": 0.10678058862686157,
      "learning_rate": 3.66181996223698e-05,
      "loss": 0.4745,
      "num_tokens": 388318419.0,
      "step": 490
    },
    {
      "epoch": 0.6704215736473801,
      "grad_norm": 0.11934733390808105,
      "learning_rate": 3.66027147454984e-05,
      "loss": 0.4673,
      "num_tokens": 389132686.0,
      "step": 491
    },
    {
      "epoch": 0.6717869943676396,
      "grad_norm": 0.12125112116336823,
      "learning_rate": 3.6587198191079014e-05,
      "loss": 0.4483,
      "num_tokens": 389859616.0,
      "step": 492
    },
    {
      "epoch": 0.6731524150878989,
      "grad_norm": 0.1077544316649437,
      "learning_rate": 3.657164999277156e-05,
      "loss": 0.453,
      "num_tokens": 390677062.0,
      "step": 493
    },
    {
      "epoch": 0.6745178358081584,
      "grad_norm": 0.11687429249286652,
      "learning_rate": 3.6556070184304624e-05,
      "loss": 0.4396,
      "num_tokens": 391459546.0,
      "step": 494
    },
    {
      "epoch": 0.6758832565284179,
      "grad_norm": 0.11729860305786133,
      "learning_rate": 3.654045879947535e-05,
      "loss": 0.4831,
      "num_tokens": 392221878.0,
      "step": 495
    },
    {
      "epoch": 0.6772486772486772,
      "grad_norm": 0.12118169665336609,
      "learning_rate": 3.652481587214939e-05,
      "loss": 0.4586,
      "num_tokens": 393026231.0,
      "step": 496
    },
    {
      "epoch": 0.6786140979689367,
      "grad_norm": 0.10897283256053925,
      "learning_rate": 3.650914143626081e-05,
      "loss": 0.4278,
      "num_tokens": 393796556.0,
      "step": 497
    },
    {
      "epoch": 0.6799795186891961,
      "grad_norm": 0.11320905387401581,
      "learning_rate": 3.6493435525812034e-05,
      "loss": 0.4685,
      "num_tokens": 394576697.0,
      "step": 498
    },
    {
      "epoch": 0.6813449394094555,
      "grad_norm": 0.12515385448932648,
      "learning_rate": 3.647769817487377e-05,
      "loss": 0.5031,
      "num_tokens": 395371861.0,
      "step": 499
    },
    {
      "epoch": 0.682710360129715,
      "grad_norm": 0.11675345152616501,
      "learning_rate": 3.6461929417584915e-05,
      "loss": 0.4445,
      "num_tokens": 396138886.0,
      "step": 500
    },
    {
      "epoch": 0.6840757808499744,
      "grad_norm": 0.1181105449795723,
      "learning_rate": 3.6446129288152514e-05,
      "loss": 0.495,
      "num_tokens": 396977902.0,
      "step": 501
    },
    {
      "epoch": 0.6854412015702338,
      "grad_norm": 0.12377340346574783,
      "learning_rate": 3.643029782085165e-05,
      "loss": 0.4823,
      "num_tokens": 397750134.0,
      "step": 502
    },
    {
      "epoch": 0.6868066222904933,
      "grad_norm": 0.11062133312225342,
      "learning_rate": 3.64144350500254e-05,
      "loss": 0.454,
      "num_tokens": 398488595.0,
      "step": 503
    },
    {
      "epoch": 0.6881720430107527,
      "grad_norm": 0.11000312119722366,
      "learning_rate": 3.639854101008472e-05,
      "loss": 0.4603,
      "num_tokens": 399316702.0,
      "step": 504
    },
    {
      "epoch": 0.6895374637310121,
      "grad_norm": 0.11922267079353333,
      "learning_rate": 3.638261573550844e-05,
      "loss": 0.4626,
      "num_tokens": 400012857.0,
      "step": 505
    },
    {
      "epoch": 0.6909028844512716,
      "grad_norm": 0.11231259256601334,
      "learning_rate": 3.636665926084313e-05,
      "loss": 0.4754,
      "num_tokens": 400820194.0,
      "step": 506
    },
    {
      "epoch": 0.6922683051715309,
      "grad_norm": 0.12647895514965057,
      "learning_rate": 3.6350671620703025e-05,
      "loss": 0.476,
      "num_tokens": 401637120.0,
      "step": 507
    },
    {
      "epoch": 0.6936337258917904,
      "grad_norm": 0.1150556281208992,
      "learning_rate": 3.633465284976998e-05,
      "loss": 0.4446,
      "num_tokens": 402317695.0,
      "step": 508
    },
    {
      "epoch": 0.6949991466120499,
      "grad_norm": 0.11447141319513321,
      "learning_rate": 3.631860298279338e-05,
      "loss": 0.4709,
      "num_tokens": 403145111.0,
      "step": 509
    },
    {
      "epoch": 0.6963645673323092,
      "grad_norm": 0.10791081190109253,
      "learning_rate": 3.630252205459009e-05,
      "loss": 0.4476,
      "num_tokens": 403908423.0,
      "step": 510
    },
    {
      "epoch": 0.6977299880525687,
      "grad_norm": 0.1311180293560028,
      "learning_rate": 3.628641010004431e-05,
      "loss": 0.4543,
      "num_tokens": 404690467.0,
      "step": 511
    },
    {
      "epoch": 0.6990954087728282,
      "grad_norm": 0.12326809763908386,
      "learning_rate": 3.6270267154107576e-05,
      "loss": 0.4979,
      "num_tokens": 405551385.0,
      "step": 512
    },
    {
      "epoch": 0.7004608294930875,
      "grad_norm": 0.1108599305152893,
      "learning_rate": 3.625409325179865e-05,
      "loss": 0.4852,
      "num_tokens": 406383570.0,
      "step": 513
    },
    {
      "epoch": 0.701826250213347,
      "grad_norm": 0.11727479845285416,
      "learning_rate": 3.623788842820344e-05,
      "loss": 0.4557,
      "num_tokens": 407156016.0,
      "step": 514
    },
    {
      "epoch": 0.7031916709336065,
      "grad_norm": 0.11476174741983414,
      "learning_rate": 3.622165271847494e-05,
      "loss": 0.4596,
      "num_tokens": 407892245.0,
      "step": 515
    },
    {
      "epoch": 0.7045570916538658,
      "grad_norm": 0.1211637407541275,
      "learning_rate": 3.6205386157833144e-05,
      "loss": 0.487,
      "num_tokens": 408630930.0,
      "step": 516
    },
    {
      "epoch": 0.7059225123741253,
      "grad_norm": 0.12017212808132172,
      "learning_rate": 3.618908878156496e-05,
      "loss": 0.4585,
      "num_tokens": 409404146.0,
      "step": 517
    },
    {
      "epoch": 0.7072879330943848,
      "grad_norm": 0.13110977411270142,
      "learning_rate": 3.617276062502415e-05,
      "loss": 0.4898,
      "num_tokens": 410233454.0,
      "step": 518
    },
    {
      "epoch": 0.7086533538146441,
      "grad_norm": 0.13033100962638855,
      "learning_rate": 3.615640172363126e-05,
      "loss": 0.4621,
      "num_tokens": 410984862.0,
      "step": 519
    },
    {
      "epoch": 0.7100187745349036,
      "grad_norm": 0.12565091252326965,
      "learning_rate": 3.61400121128735e-05,
      "loss": 0.4987,
      "num_tokens": 411761902.0,
      "step": 520
    },
    {
      "epoch": 0.7113841952551629,
      "grad_norm": 0.116177998483181,
      "learning_rate": 3.6123591828304726e-05,
      "loss": 0.464,
      "num_tokens": 412588366.0,
      "step": 521
    },
    {
      "epoch": 0.7127496159754224,
      "grad_norm": 0.11946090310811996,
      "learning_rate": 3.610714090554533e-05,
      "loss": 0.4409,
      "num_tokens": 413372551.0,
      "step": 522
    },
    {
      "epoch": 0.7141150366956819,
      "grad_norm": 0.12250180542469025,
      "learning_rate": 3.609065938028216e-05,
      "loss": 0.4686,
      "num_tokens": 414123255.0,
      "step": 523
    },
    {
      "epoch": 0.7154804574159412,
      "grad_norm": 0.12352269142866135,
      "learning_rate": 3.6074147288268456e-05,
      "loss": 0.476,
      "num_tokens": 414911593.0,
      "step": 524
    },
    {
      "epoch": 0.7168458781362007,
      "grad_norm": 0.12229899317026138,
      "learning_rate": 3.605760466532374e-05,
      "loss": 0.4769,
      "num_tokens": 415667737.0,
      "step": 525
    },
    {
      "epoch": 0.7182112988564602,
      "grad_norm": 0.1086830198764801,
      "learning_rate": 3.604103154733382e-05,
      "loss": 0.4573,
      "num_tokens": 416432559.0,
      "step": 526
    },
    {
      "epoch": 0.7195767195767195,
      "grad_norm": 0.11078007519245148,
      "learning_rate": 3.602442797025061e-05,
      "loss": 0.4434,
      "num_tokens": 417204561.0,
      "step": 527
    },
    {
      "epoch": 0.720942140296979,
      "grad_norm": 0.11286323517560959,
      "learning_rate": 3.6007793970092095e-05,
      "loss": 0.473,
      "num_tokens": 417988051.0,
      "step": 528
    },
    {
      "epoch": 0.7223075610172385,
      "grad_norm": 0.1103808581829071,
      "learning_rate": 3.599112958294229e-05,
      "loss": 0.4485,
      "num_tokens": 418779302.0,
      "step": 529
    },
    {
      "epoch": 0.7236729817374978,
      "grad_norm": 0.11847126483917236,
      "learning_rate": 3.597443484495112e-05,
      "loss": 0.4662,
      "num_tokens": 419567457.0,
      "step": 530
    },
    {
      "epoch": 0.7250384024577573,
      "grad_norm": 0.10949163883924484,
      "learning_rate": 3.595770979233433e-05,
      "loss": 0.4558,
      "num_tokens": 420368051.0,
      "step": 531
    },
    {
      "epoch": 0.7264038231780168,
      "grad_norm": 0.11041102558374405,
      "learning_rate": 3.594095446137343e-05,
      "loss": 0.4594,
      "num_tokens": 421099868.0,
      "step": 532
    },
    {
      "epoch": 0.7277692438982761,
      "grad_norm": 0.11672098934650421,
      "learning_rate": 3.592416888841564e-05,
      "loss": 0.4991,
      "num_tokens": 421942092.0,
      "step": 533
    },
    {
      "epoch": 0.7291346646185356,
      "grad_norm": 0.117094025015831,
      "learning_rate": 3.5907353109873745e-05,
      "loss": 0.4673,
      "num_tokens": 422785753.0,
      "step": 534
    },
    {
      "epoch": 0.7305000853387951,
      "grad_norm": 0.10962095111608505,
      "learning_rate": 3.589050716222609e-05,
      "loss": 0.4618,
      "num_tokens": 423606299.0,
      "step": 535
    },
    {
      "epoch": 0.7318655060590544,
      "grad_norm": 0.11803168803453445,
      "learning_rate": 3.5873631082016445e-05,
      "loss": 0.4848,
      "num_tokens": 424360117.0,
      "step": 536
    },
    {
      "epoch": 0.7332309267793139,
      "grad_norm": 0.11563384532928467,
      "learning_rate": 3.585672490585394e-05,
      "loss": 0.469,
      "num_tokens": 425166636.0,
      "step": 537
    },
    {
      "epoch": 0.7345963474995733,
      "grad_norm": 0.12476540356874466,
      "learning_rate": 3.583978867041302e-05,
      "loss": 0.4785,
      "num_tokens": 425973427.0,
      "step": 538
    },
    {
      "epoch": 0.7359617682198327,
      "grad_norm": 0.10442819446325302,
      "learning_rate": 3.5822822412433304e-05,
      "loss": 0.4656,
      "num_tokens": 426720036.0,
      "step": 539
    },
    {
      "epoch": 0.7373271889400922,
      "grad_norm": 0.10755571722984314,
      "learning_rate": 3.5805826168719564e-05,
      "loss": 0.4698,
      "num_tokens": 427545403.0,
      "step": 540
    },
    {
      "epoch": 0.7386926096603516,
      "grad_norm": 0.11853408068418503,
      "learning_rate": 3.578879997614161e-05,
      "loss": 0.49,
      "num_tokens": 428384320.0,
      "step": 541
    },
    {
      "epoch": 0.740058030380611,
      "grad_norm": 0.11008844524621964,
      "learning_rate": 3.577174387163421e-05,
      "loss": 0.4764,
      "num_tokens": 429204114.0,
      "step": 542
    },
    {
      "epoch": 0.7414234511008705,
      "grad_norm": 0.12466719001531601,
      "learning_rate": 3.5754657892197056e-05,
      "loss": 0.4683,
      "num_tokens": 429964635.0,
      "step": 543
    },
    {
      "epoch": 0.7427888718211298,
      "grad_norm": 0.11853611469268799,
      "learning_rate": 3.5737542074894595e-05,
      "loss": 0.4579,
      "num_tokens": 430734728.0,
      "step": 544
    },
    {
      "epoch": 0.7441542925413893,
      "grad_norm": 0.1129520907998085,
      "learning_rate": 3.572039645685603e-05,
      "loss": 0.4881,
      "num_tokens": 431673075.0,
      "step": 545
    },
    {
      "epoch": 0.7455197132616488,
      "grad_norm": 0.11060653626918793,
      "learning_rate": 3.5703221075275227e-05,
      "loss": 0.4753,
      "num_tokens": 432436049.0,
      "step": 546
    },
    {
      "epoch": 0.7468851339819081,
      "grad_norm": 0.11273801326751709,
      "learning_rate": 3.5686015967410584e-05,
      "loss": 0.4618,
      "num_tokens": 433225490.0,
      "step": 547
    },
    {
      "epoch": 0.7482505547021676,
      "grad_norm": 0.11641700565814972,
      "learning_rate": 3.5668781170584996e-05,
      "loss": 0.4419,
      "num_tokens": 434000184.0,
      "step": 548
    },
    {
      "epoch": 0.7496159754224271,
      "grad_norm": 0.11713509261608124,
      "learning_rate": 3.565151672218578e-05,
      "loss": 0.4456,
      "num_tokens": 434704866.0,
      "step": 549
    },
    {
      "epoch": 0.7509813961426864,
      "grad_norm": 0.11676023155450821,
      "learning_rate": 3.563422265966454e-05,
      "loss": 0.4712,
      "num_tokens": 435493023.0,
      "step": 550
    },
    {
      "epoch": 0.7523468168629459,
      "grad_norm": 0.11953166127204895,
      "learning_rate": 3.5616899020537166e-05,
      "loss": 0.448,
      "num_tokens": 436250412.0,
      "step": 551
    },
    {
      "epoch": 0.7537122375832054,
      "grad_norm": 0.10632363706827164,
      "learning_rate": 3.559954584238366e-05,
      "loss": 0.4522,
      "num_tokens": 437007876.0,
      "step": 552
    },
    {
      "epoch": 0.7550776583034647,
      "grad_norm": 0.11048779636621475,
      "learning_rate": 3.558216316284815e-05,
      "loss": 0.4686,
      "num_tokens": 437732327.0,
      "step": 553
    },
    {
      "epoch": 0.7564430790237242,
      "grad_norm": 0.11666207015514374,
      "learning_rate": 3.556475101963872e-05,
      "loss": 0.4503,
      "num_tokens": 438486918.0,
      "step": 554
    },
    {
      "epoch": 0.7578084997439836,
      "grad_norm": 0.11541448533535004,
      "learning_rate": 3.5547309450527394e-05,
      "loss": 0.4615,
      "num_tokens": 439267058.0,
      "step": 555
    },
    {
      "epoch": 0.759173920464243,
      "grad_norm": 0.11679370701313019,
      "learning_rate": 3.5529838493350034e-05,
      "loss": 0.4623,
      "num_tokens": 440066254.0,
      "step": 556
    },
    {
      "epoch": 0.7605393411845025,
      "grad_norm": 0.11054755747318268,
      "learning_rate": 3.551233818600623e-05,
      "loss": 0.4346,
      "num_tokens": 440813273.0,
      "step": 557
    },
    {
      "epoch": 0.7619047619047619,
      "grad_norm": 0.11531034111976624,
      "learning_rate": 3.5494808566459267e-05,
      "loss": 0.4588,
      "num_tokens": 441545387.0,
      "step": 558
    },
    {
      "epoch": 0.7632701826250213,
      "grad_norm": 0.11614120751619339,
      "learning_rate": 3.5477249672735976e-05,
      "loss": 0.452,
      "num_tokens": 442378053.0,
      "step": 559
    },
    {
      "epoch": 0.7646356033452808,
      "grad_norm": 0.10858071595430374,
      "learning_rate": 3.5459661542926756e-05,
      "loss": 0.4474,
      "num_tokens": 443126194.0,
      "step": 560
    },
    {
      "epoch": 0.7660010240655402,
      "grad_norm": 0.1246400699019432,
      "learning_rate": 3.5442044215185363e-05,
      "loss": 0.4538,
      "num_tokens": 443881072.0,
      "step": 561
    },
    {
      "epoch": 0.7673664447857996,
      "grad_norm": 0.11067893356084824,
      "learning_rate": 3.542439772772893e-05,
      "loss": 0.4556,
      "num_tokens": 444662023.0,
      "step": 562
    },
    {
      "epoch": 0.7687318655060591,
      "grad_norm": 0.11047633737325668,
      "learning_rate": 3.540672211883783e-05,
      "loss": 0.4645,
      "num_tokens": 445555182.0,
      "step": 563
    },
    {
      "epoch": 0.7700972862263185,
      "grad_norm": 0.11735224723815918,
      "learning_rate": 3.538901742685563e-05,
      "loss": 0.4556,
      "num_tokens": 446347138.0,
      "step": 564
    },
    {
      "epoch": 0.7714627069465779,
      "grad_norm": 0.10870572924613953,
      "learning_rate": 3.537128369018895e-05,
      "loss": 0.4592,
      "num_tokens": 447154379.0,
      "step": 565
    },
    {
      "epoch": 0.7728281276668374,
      "grad_norm": 0.11547549068927765,
      "learning_rate": 3.535352094730746e-05,
      "loss": 0.4551,
      "num_tokens": 447926931.0,
      "step": 566
    },
    {
      "epoch": 0.7741935483870968,
      "grad_norm": 0.11634697020053864,
      "learning_rate": 3.533572923674372e-05,
      "loss": 0.4745,
      "num_tokens": 448755226.0,
      "step": 567
    },
    {
      "epoch": 0.7755589691073562,
      "grad_norm": 0.10786948353052139,
      "learning_rate": 3.531790859709313e-05,
      "loss": 0.4657,
      "num_tokens": 449550443.0,
      "step": 568
    },
    {
      "epoch": 0.7769243898276156,
      "grad_norm": 0.12652099132537842,
      "learning_rate": 3.5300059067013885e-05,
      "loss": 0.4631,
      "num_tokens": 450363415.0,
      "step": 569
    },
    {
      "epoch": 0.778289810547875,
      "grad_norm": 0.10690262913703918,
      "learning_rate": 3.528218068522679e-05,
      "loss": 0.435,
      "num_tokens": 451085842.0,
      "step": 570
    },
    {
      "epoch": 0.7796552312681345,
      "grad_norm": 0.11398256570100784,
      "learning_rate": 3.526427349051528e-05,
      "loss": 0.4681,
      "num_tokens": 451920933.0,
      "step": 571
    },
    {
      "epoch": 0.7810206519883939,
      "grad_norm": 0.11256309598684311,
      "learning_rate": 3.52463375217253e-05,
      "loss": 0.4625,
      "num_tokens": 452759021.0,
      "step": 572
    },
    {
      "epoch": 0.7823860727086533,
      "grad_norm": 0.12646156549453735,
      "learning_rate": 3.522837281776519e-05,
      "loss": 0.4435,
      "num_tokens": 453509716.0,
      "step": 573
    },
    {
      "epoch": 0.7837514934289128,
      "grad_norm": 0.11686700582504272,
      "learning_rate": 3.521037941760562e-05,
      "loss": 0.4605,
      "num_tokens": 454339195.0,
      "step": 574
    },
    {
      "epoch": 0.7851169141491722,
      "grad_norm": 0.11004216223955154,
      "learning_rate": 3.5192357360279536e-05,
      "loss": 0.4662,
      "num_tokens": 455156513.0,
      "step": 575
    },
    {
      "epoch": 0.7864823348694316,
      "grad_norm": 0.12423640489578247,
      "learning_rate": 3.517430668488203e-05,
      "loss": 0.4849,
      "num_tokens": 456063494.0,
      "step": 576
    },
    {
      "epoch": 0.7878477555896911,
      "grad_norm": 0.12538117170333862,
      "learning_rate": 3.5156227430570294e-05,
      "loss": 0.4731,
      "num_tokens": 456916086.0,
      "step": 577
    },
    {
      "epoch": 0.7892131763099505,
      "grad_norm": 0.11302205920219421,
      "learning_rate": 3.5138119636563495e-05,
      "loss": 0.4671,
      "num_tokens": 457760380.0,
      "step": 578
    },
    {
      "epoch": 0.7905785970302099,
      "grad_norm": 0.12370152026414871,
      "learning_rate": 3.5119983342142726e-05,
      "loss": 0.4782,
      "num_tokens": 458598544.0,
      "step": 579
    },
    {
      "epoch": 0.7919440177504694,
      "grad_norm": 0.13028262555599213,
      "learning_rate": 3.51018185866509e-05,
      "loss": 0.4696,
      "num_tokens": 459338363.0,
      "step": 580
    },
    {
      "epoch": 0.7933094384707288,
      "grad_norm": 0.11211229860782623,
      "learning_rate": 3.508362540949266e-05,
      "loss": 0.4558,
      "num_tokens": 460114221.0,
      "step": 581
    },
    {
      "epoch": 0.7946748591909882,
      "grad_norm": 0.1160685122013092,
      "learning_rate": 3.506540385013433e-05,
      "loss": 0.4657,
      "num_tokens": 460897942.0,
      "step": 582
    },
    {
      "epoch": 0.7960402799112477,
      "grad_norm": 0.1304933875799179,
      "learning_rate": 3.5047153948103775e-05,
      "loss": 0.4675,
      "num_tokens": 461654501.0,
      "step": 583
    },
    {
      "epoch": 0.7974057006315071,
      "grad_norm": 0.11600954830646515,
      "learning_rate": 3.502887574299036e-05,
      "loss": 0.4717,
      "num_tokens": 462514849.0,
      "step": 584
    },
    {
      "epoch": 0.7987711213517665,
      "grad_norm": 0.11589249968528748,
      "learning_rate": 3.501056927444485e-05,
      "loss": 0.4475,
      "num_tokens": 463281466.0,
      "step": 585
    },
    {
      "epoch": 0.8001365420720259,
      "grad_norm": 0.1238078847527504,
      "learning_rate": 3.499223458217931e-05,
      "loss": 0.4526,
      "num_tokens": 464011338.0,
      "step": 586
    },
    {
      "epoch": 0.8015019627922854,
      "grad_norm": 0.12256167829036713,
      "learning_rate": 3.497387170596703e-05,
      "loss": 0.4701,
      "num_tokens": 464778947.0,
      "step": 587
    },
    {
      "epoch": 0.8028673835125448,
      "grad_norm": 0.1173655241727829,
      "learning_rate": 3.4955480685642454e-05,
      "loss": 0.4656,
      "num_tokens": 465587117.0,
      "step": 588
    },
    {
      "epoch": 0.8042328042328042,
      "grad_norm": 0.13252049684524536,
      "learning_rate": 3.493706156110107e-05,
      "loss": 0.489,
      "num_tokens": 466364468.0,
      "step": 589
    },
    {
      "epoch": 0.8055982249530637,
      "grad_norm": 0.12360338866710663,
      "learning_rate": 3.491861437229933e-05,
      "loss": 0.497,
      "num_tokens": 467263988.0,
      "step": 590
    },
    {
      "epoch": 0.8069636456733231,
      "grad_norm": 0.13858993351459503,
      "learning_rate": 3.490013915925456e-05,
      "loss": 0.4637,
      "num_tokens": 468074575.0,
      "step": 591
    },
    {
      "epoch": 0.8083290663935825,
      "grad_norm": 0.12826184928417206,
      "learning_rate": 3.4881635962044894e-05,
      "loss": 0.4729,
      "num_tokens": 468856827.0,
      "step": 592
    },
    {
      "epoch": 0.809694487113842,
      "grad_norm": 0.10866866260766983,
      "learning_rate": 3.486310482080917e-05,
      "loss": 0.4477,
      "num_tokens": 469617883.0,
      "step": 593
    },
    {
      "epoch": 0.8110599078341014,
      "grad_norm": 0.1250615417957306,
      "learning_rate": 3.4844545775746835e-05,
      "loss": 0.4503,
      "num_tokens": 470454979.0,
      "step": 594
    },
    {
      "epoch": 0.8124253285543608,
      "grad_norm": 0.13922300934791565,
      "learning_rate": 3.482595886711786e-05,
      "loss": 0.4643,
      "num_tokens": 471212857.0,
      "step": 595
    },
    {
      "epoch": 0.8137907492746203,
      "grad_norm": 0.1148633137345314,
      "learning_rate": 3.480734413524268e-05,
      "loss": 0.4536,
      "num_tokens": 471992155.0,
      "step": 596
    },
    {
      "epoch": 0.8151561699948797,
      "grad_norm": 0.12613527476787567,
      "learning_rate": 3.478870162050209e-05,
      "loss": 0.4368,
      "num_tokens": 472772198.0,
      "step": 597
    },
    {
      "epoch": 0.8165215907151391,
      "grad_norm": 0.11643815785646439,
      "learning_rate": 3.4770031363337126e-05,
      "loss": 0.4567,
      "num_tokens": 473563244.0,
      "step": 598
    },
    {
      "epoch": 0.8178870114353985,
      "grad_norm": 0.12114600837230682,
      "learning_rate": 3.4751333404249026e-05,
      "loss": 0.4682,
      "num_tokens": 474356498.0,
      "step": 599
    },
    {
      "epoch": 0.819252432155658,
      "grad_norm": 0.11650528013706207,
      "learning_rate": 3.4732607783799125e-05,
      "loss": 0.473,
      "num_tokens": 475192924.0,
      "step": 600
    },
    {
      "epoch": 0.8206178528759174,
      "grad_norm": 0.13819915056228638,
      "learning_rate": 3.4713854542608744e-05,
      "loss": 0.4714,
      "num_tokens": 475964234.0,
      "step": 601
    },
    {
      "epoch": 0.8219832735961768,
      "grad_norm": 0.12915681302547455,
      "learning_rate": 3.469507372135914e-05,
      "loss": 0.4738,
      "num_tokens": 476765905.0,
      "step": 602
    },
    {
      "epoch": 0.8233486943164362,
      "grad_norm": 0.14539968967437744,
      "learning_rate": 3.4676265360791396e-05,
      "loss": 0.468,
      "num_tokens": 477497369.0,
      "step": 603
    },
    {
      "epoch": 0.8247141150366957,
      "grad_norm": 0.12354116886854172,
      "learning_rate": 3.465742950170633e-05,
      "loss": 0.4483,
      "num_tokens": 478247746.0,
      "step": 604
    },
    {
      "epoch": 0.8260795357569551,
      "grad_norm": 0.1283135712146759,
      "learning_rate": 3.4638566184964414e-05,
      "loss": 0.447,
      "num_tokens": 479061479.0,
      "step": 605
    },
    {
      "epoch": 0.8274449564772145,
      "grad_norm": 0.11911921203136444,
      "learning_rate": 3.4619675451485686e-05,
      "loss": 0.4726,
      "num_tokens": 479936071.0,
      "step": 606
    },
    {
      "epoch": 0.828810377197474,
      "grad_norm": 0.12042238563299179,
      "learning_rate": 3.460075734224966e-05,
      "loss": 0.4416,
      "num_tokens": 480702703.0,
      "step": 607
    },
    {
      "epoch": 0.8301757979177334,
      "grad_norm": 0.13676589727401733,
      "learning_rate": 3.458181189829523e-05,
      "loss": 0.4628,
      "num_tokens": 481474943.0,
      "step": 608
    },
    {
      "epoch": 0.8315412186379928,
      "grad_norm": 0.12655797600746155,
      "learning_rate": 3.4562839160720586e-05,
      "loss": 0.4548,
      "num_tokens": 482259302.0,
      "step": 609
    },
    {
      "epoch": 0.8329066393582523,
      "grad_norm": 0.13091017305850983,
      "learning_rate": 3.4543839170683135e-05,
      "loss": 0.4728,
      "num_tokens": 483122182.0,
      "step": 610
    },
    {
      "epoch": 0.8342720600785117,
      "grad_norm": 0.12149649858474731,
      "learning_rate": 3.45248119693994e-05,
      "loss": 0.4727,
      "num_tokens": 483942345.0,
      "step": 611
    },
    {
      "epoch": 0.8356374807987711,
      "grad_norm": 0.14493805170059204,
      "learning_rate": 3.4505757598144926e-05,
      "loss": 0.4693,
      "num_tokens": 484789857.0,
      "step": 612
    },
    {
      "epoch": 0.8370029015190306,
      "grad_norm": 0.10913131386041641,
      "learning_rate": 3.4486676098254206e-05,
      "loss": 0.4169,
      "num_tokens": 485485142.0,
      "step": 613
    },
    {
      "epoch": 0.83836832223929,
      "grad_norm": 0.1453300416469574,
      "learning_rate": 3.4467567511120577e-05,
      "loss": 0.4537,
      "num_tokens": 486274385.0,
      "step": 614
    },
    {
      "epoch": 0.8397337429595494,
      "grad_norm": 0.11809556186199188,
      "learning_rate": 3.444843187819614e-05,
      "loss": 0.4598,
      "num_tokens": 487076335.0,
      "step": 615
    },
    {
      "epoch": 0.8410991636798089,
      "grad_norm": 0.15078119933605194,
      "learning_rate": 3.442926924099167e-05,
      "loss": 0.4436,
      "num_tokens": 487799125.0,
      "step": 616
    },
    {
      "epoch": 0.8424645844000682,
      "grad_norm": 0.11497651040554047,
      "learning_rate": 3.441007964107651e-05,
      "loss": 0.457,
      "num_tokens": 488613366.0,
      "step": 617
    },
    {
      "epoch": 0.8438300051203277,
      "grad_norm": 0.12549902498722076,
      "learning_rate": 3.43908631200785e-05,
      "loss": 0.4657,
      "num_tokens": 489356121.0,
      "step": 618
    },
    {
      "epoch": 0.8451954258405872,
      "grad_norm": 0.13588614761829376,
      "learning_rate": 3.437161971968388e-05,
      "loss": 0.4636,
      "num_tokens": 490218511.0,
      "step": 619
    },
    {
      "epoch": 0.8465608465608465,
      "grad_norm": 0.12390116602182388,
      "learning_rate": 3.435234948163721e-05,
      "loss": 0.445,
      "num_tokens": 491003541.0,
      "step": 620
    },
    {
      "epoch": 0.847926267281106,
      "grad_norm": 0.12032484263181686,
      "learning_rate": 3.433305244774124e-05,
      "loss": 0.4777,
      "num_tokens": 491859418.0,
      "step": 621
    },
    {
      "epoch": 0.8492916880013655,
      "grad_norm": 0.11749201267957687,
      "learning_rate": 3.4313728659856873e-05,
      "loss": 0.4586,
      "num_tokens": 492623443.0,
      "step": 622
    },
    {
      "epoch": 0.8506571087216248,
      "grad_norm": 0.12673687934875488,
      "learning_rate": 3.4294378159903044e-05,
      "loss": 0.4786,
      "num_tokens": 493417315.0,
      "step": 623
    },
    {
      "epoch": 0.8520225294418843,
      "grad_norm": 0.111882284283638,
      "learning_rate": 3.4275000989856634e-05,
      "loss": 0.452,
      "num_tokens": 494139379.0,
      "step": 624
    },
    {
      "epoch": 0.8533879501621437,
      "grad_norm": 0.12103655934333801,
      "learning_rate": 3.4255597191752365e-05,
      "loss": 0.4817,
      "num_tokens": 494971383.0,
      "step": 625
    },
    {
      "epoch": 0.8547533708824031,
      "grad_norm": 0.12442655116319656,
      "learning_rate": 3.423616680768276e-05,
      "loss": 0.4631,
      "num_tokens": 495752309.0,
      "step": 626
    },
    {
      "epoch": 0.8561187916026626,
      "grad_norm": 0.12896881997585297,
      "learning_rate": 3.421670987979796e-05,
      "loss": 0.4605,
      "num_tokens": 496580248.0,
      "step": 627
    },
    {
      "epoch": 0.857484212322922,
      "grad_norm": 0.11374063044786453,
      "learning_rate": 3.419722645030574e-05,
      "loss": 0.4846,
      "num_tokens": 497477553.0,
      "step": 628
    },
    {
      "epoch": 0.8588496330431814,
      "grad_norm": 0.1132168397307396,
      "learning_rate": 3.417771656147133e-05,
      "loss": 0.4621,
      "num_tokens": 498286581.0,
      "step": 629
    },
    {
      "epoch": 0.8602150537634409,
      "grad_norm": 0.11758428066968918,
      "learning_rate": 3.4158180255617365e-05,
      "loss": 0.4693,
      "num_tokens": 499052435.0,
      "step": 630
    },
    {
      "epoch": 0.8615804744837003,
      "grad_norm": 0.11080723255872726,
      "learning_rate": 3.413861757512381e-05,
      "loss": 0.4481,
      "num_tokens": 499816873.0,
      "step": 631
    },
    {
      "epoch": 0.8629458952039597,
      "grad_norm": 0.11279977858066559,
      "learning_rate": 3.411902856242782e-05,
      "loss": 0.4602,
      "num_tokens": 500612310.0,
      "step": 632
    },
    {
      "epoch": 0.8643113159242192,
      "grad_norm": 0.11113811284303665,
      "learning_rate": 3.4099413260023654e-05,
      "loss": 0.4596,
      "num_tokens": 501372246.0,
      "step": 633
    },
    {
      "epoch": 0.8656767366444785,
      "grad_norm": 0.11998709291219711,
      "learning_rate": 3.407977171046267e-05,
      "loss": 0.4866,
      "num_tokens": 502236216.0,
      "step": 634
    },
    {
      "epoch": 0.867042157364738,
      "grad_norm": 0.11701316386461258,
      "learning_rate": 3.406010395635307e-05,
      "loss": 0.4794,
      "num_tokens": 503012542.0,
      "step": 635
    },
    {
      "epoch": 0.8684075780849975,
      "grad_norm": 0.10782507061958313,
      "learning_rate": 3.404041004035998e-05,
      "loss": 0.454,
      "num_tokens": 503810590.0,
      "step": 636
    },
    {
      "epoch": 0.8697729988052568,
      "grad_norm": 0.10437658429145813,
      "learning_rate": 3.4020690005205245e-05,
      "loss": 0.4588,
      "num_tokens": 504650413.0,
      "step": 637
    },
    {
      "epoch": 0.8711384195255163,
      "grad_norm": 0.12532836198806763,
      "learning_rate": 3.400094389366737e-05,
      "loss": 0.456,
      "num_tokens": 505380996.0,
      "step": 638
    },
    {
      "epoch": 0.8725038402457758,
      "grad_norm": 0.10351867973804474,
      "learning_rate": 3.3981171748581434e-05,
      "loss": 0.4445,
      "num_tokens": 506230685.0,
      "step": 639
    },
    {
      "epoch": 0.8738692609660351,
      "grad_norm": 0.1168542131781578,
      "learning_rate": 3.396137361283898e-05,
      "loss": 0.4447,
      "num_tokens": 507038971.0,
      "step": 640
    },
    {
      "epoch": 0.8752346816862946,
      "grad_norm": 0.11234838515520096,
      "learning_rate": 3.3941549529387936e-05,
      "loss": 0.4583,
      "num_tokens": 507830039.0,
      "step": 641
    },
    {
      "epoch": 0.8766001024065541,
      "grad_norm": 0.10551413893699646,
      "learning_rate": 3.392169954123253e-05,
      "loss": 0.445,
      "num_tokens": 508602614.0,
      "step": 642
    },
    {
      "epoch": 0.8779655231268134,
      "grad_norm": 0.10385875403881073,
      "learning_rate": 3.390182369143319e-05,
      "loss": 0.4497,
      "num_tokens": 509459114.0,
      "step": 643
    },
    {
      "epoch": 0.8793309438470729,
      "grad_norm": 0.11604557186365128,
      "learning_rate": 3.3881922023106414e-05,
      "loss": 0.4559,
      "num_tokens": 510286640.0,
      "step": 644
    },
    {
      "epoch": 0.8806963645673324,
      "grad_norm": 0.1087288185954094,
      "learning_rate": 3.3861994579424725e-05,
      "loss": 0.4545,
      "num_tokens": 511009781.0,
      "step": 645
    },
    {
      "epoch": 0.8820617852875917,
      "grad_norm": 0.11322581022977829,
      "learning_rate": 3.384204140361657e-05,
      "loss": 0.4673,
      "num_tokens": 511872083.0,
      "step": 646
    },
    {
      "epoch": 0.8834272060078512,
      "grad_norm": 0.11553996801376343,
      "learning_rate": 3.38220625389662e-05,
      "loss": 0.4868,
      "num_tokens": 512695192.0,
      "step": 647
    },
    {
      "epoch": 0.8847926267281107,
      "grad_norm": 0.1159779354929924,
      "learning_rate": 3.380205802881362e-05,
      "loss": 0.4746,
      "num_tokens": 513581628.0,
      "step": 648
    },
    {
      "epoch": 0.88615804744837,
      "grad_norm": 0.10663741827011108,
      "learning_rate": 3.378202791655441e-05,
      "loss": 0.4311,
      "num_tokens": 514382407.0,
      "step": 649
    },
    {
      "epoch": 0.8875234681686295,
      "grad_norm": 0.10537584871053696,
      "learning_rate": 3.376197224563977e-05,
      "loss": 0.4581,
      "num_tokens": 515166821.0,
      "step": 650
    },
    {
      "epoch": 0.8888888888888888,
      "grad_norm": 0.12934701144695282,
      "learning_rate": 3.374189105957627e-05,
      "loss": 0.4625,
      "num_tokens": 515974328.0,
      "step": 651
    },
    {
      "epoch": 0.8902543096091483,
      "grad_norm": 0.11699160933494568,
      "learning_rate": 3.372178440192588e-05,
      "loss": 0.4701,
      "num_tokens": 516806639.0,
      "step": 652
    },
    {
      "epoch": 0.8916197303294078,
      "grad_norm": 0.10877185314893723,
      "learning_rate": 3.3701652316305784e-05,
      "loss": 0.4556,
      "num_tokens": 517660816.0,
      "step": 653
    },
    {
      "epoch": 0.8929851510496671,
      "grad_norm": 0.1083533763885498,
      "learning_rate": 3.3681494846388365e-05,
      "loss": 0.4576,
      "num_tokens": 518459802.0,
      "step": 654
    },
    {
      "epoch": 0.8943505717699266,
      "grad_norm": 0.11289810389280319,
      "learning_rate": 3.366131203590106e-05,
      "loss": 0.4516,
      "num_tokens": 519265238.0,
      "step": 655
    },
    {
      "epoch": 0.8957159924901861,
      "grad_norm": 0.10493972897529602,
      "learning_rate": 3.364110392862625e-05,
      "loss": 0.4751,
      "num_tokens": 520153394.0,
      "step": 656
    },
    {
      "epoch": 0.8970814132104454,
      "grad_norm": 0.12332845479249954,
      "learning_rate": 3.362087056840123e-05,
      "loss": 0.4618,
      "num_tokens": 520924635.0,
      "step": 657
    },
    {
      "epoch": 0.8984468339307049,
      "grad_norm": 0.10880496352910995,
      "learning_rate": 3.360061199911807e-05,
      "loss": 0.4563,
      "num_tokens": 521809493.0,
      "step": 658
    },
    {
      "epoch": 0.8998122546509644,
      "grad_norm": 0.11193115264177322,
      "learning_rate": 3.358032826472349e-05,
      "loss": 0.463,
      "num_tokens": 522574517.0,
      "step": 659
    },
    {
      "epoch": 0.9011776753712237,
      "grad_norm": 0.11564835906028748,
      "learning_rate": 3.3560019409218845e-05,
      "loss": 0.449,
      "num_tokens": 523350113.0,
      "step": 660
    },
    {
      "epoch": 0.9025430960914832,
      "grad_norm": 0.11182774603366852,
      "learning_rate": 3.3539685476659964e-05,
      "loss": 0.4572,
      "num_tokens": 524203752.0,
      "step": 661
    },
    {
      "epoch": 0.9039085168117427,
      "grad_norm": 0.11104340851306915,
      "learning_rate": 3.351932651115708e-05,
      "loss": 0.4613,
      "num_tokens": 524969300.0,
      "step": 662
    },
    {
      "epoch": 0.905273937532002,
      "grad_norm": 0.10952315479516983,
      "learning_rate": 3.3498942556874724e-05,
      "loss": 0.4549,
      "num_tokens": 525763253.0,
      "step": 663
    },
    {
      "epoch": 0.9066393582522615,
      "grad_norm": 0.10629931092262268,
      "learning_rate": 3.3478533658031645e-05,
      "loss": 0.45,
      "num_tokens": 526548055.0,
      "step": 664
    },
    {
      "epoch": 0.9080047789725209,
      "grad_norm": 0.11691020429134369,
      "learning_rate": 3.345809985890069e-05,
      "loss": 0.4473,
      "num_tokens": 527285929.0,
      "step": 665
    },
    {
      "epoch": 0.9093701996927803,
      "grad_norm": 0.11251469701528549,
      "learning_rate": 3.343764120380873e-05,
      "loss": 0.4599,
      "num_tokens": 528039458.0,
      "step": 666
    },
    {
      "epoch": 0.9107356204130398,
      "grad_norm": 0.10950896888971329,
      "learning_rate": 3.341715773713657e-05,
      "loss": 0.4563,
      "num_tokens": 528869150.0,
      "step": 667
    },
    {
      "epoch": 0.9121010411332992,
      "grad_norm": 0.12076958268880844,
      "learning_rate": 3.339664950331881e-05,
      "loss": 0.4583,
      "num_tokens": 529679423.0,
      "step": 668
    },
    {
      "epoch": 0.9134664618535586,
      "grad_norm": 0.11457866430282593,
      "learning_rate": 3.3376116546843784e-05,
      "loss": 0.4613,
      "num_tokens": 530493277.0,
      "step": 669
    },
    {
      "epoch": 0.9148318825738181,
      "grad_norm": 0.11846800893545151,
      "learning_rate": 3.335555891225349e-05,
      "loss": 0.4947,
      "num_tokens": 531301060.0,
      "step": 670
    },
    {
      "epoch": 0.9161973032940774,
      "grad_norm": 0.11642908304929733,
      "learning_rate": 3.333497664414341e-05,
      "loss": 0.4539,
      "num_tokens": 532055240.0,
      "step": 671
    },
    {
      "epoch": 0.9175627240143369,
      "grad_norm": 0.10209578275680542,
      "learning_rate": 3.33143697871625e-05,
      "loss": 0.4247,
      "num_tokens": 532836986.0,
      "step": 672
    },
    {
      "epoch": 0.9189281447345964,
      "grad_norm": 0.10477438569068909,
      "learning_rate": 3.3293738386013046e-05,
      "loss": 0.4494,
      "num_tokens": 533621672.0,
      "step": 673
    },
    {
      "epoch": 0.9202935654548557,
      "grad_norm": 0.11298660933971405,
      "learning_rate": 3.327308248545057e-05,
      "loss": 0.4458,
      "num_tokens": 534369867.0,
      "step": 674
    },
    {
      "epoch": 0.9216589861751152,
      "grad_norm": 0.11382252722978592,
      "learning_rate": 3.325240213028374e-05,
      "loss": 0.4958,
      "num_tokens": 535181213.0,
      "step": 675
    },
    {
      "epoch": 0.9230244068953747,
      "grad_norm": 0.10414505004882812,
      "learning_rate": 3.323169736537429e-05,
      "loss": 0.4484,
      "num_tokens": 535969004.0,
      "step": 676
    },
    {
      "epoch": 0.924389827615634,
      "grad_norm": 0.1114826649427414,
      "learning_rate": 3.321096823563689e-05,
      "loss": 0.4853,
      "num_tokens": 536746754.0,
      "step": 677
    },
    {
      "epoch": 0.9257552483358935,
      "grad_norm": 0.11756595224142075,
      "learning_rate": 3.3190214786039065e-05,
      "loss": 0.4636,
      "num_tokens": 537545303.0,
      "step": 678
    },
    {
      "epoch": 0.927120669056153,
      "grad_norm": 0.10061848908662796,
      "learning_rate": 3.316943706160111e-05,
      "loss": 0.4433,
      "num_tokens": 538290548.0,
      "step": 679
    },
    {
      "epoch": 0.9284860897764123,
      "grad_norm": 0.11385159939527512,
      "learning_rate": 3.314863510739597e-05,
      "loss": 0.4775,
      "num_tokens": 539068033.0,
      "step": 680
    },
    {
      "epoch": 0.9298515104966718,
      "grad_norm": 0.10928046703338623,
      "learning_rate": 3.312780896854915e-05,
      "loss": 0.4356,
      "num_tokens": 539811481.0,
      "step": 681
    },
    {
      "epoch": 0.9312169312169312,
      "grad_norm": 0.11159523576498032,
      "learning_rate": 3.310695869023861e-05,
      "loss": 0.4555,
      "num_tokens": 540597155.0,
      "step": 682
    },
    {
      "epoch": 0.9325823519371906,
      "grad_norm": 0.12235260009765625,
      "learning_rate": 3.308608431769469e-05,
      "loss": 0.4504,
      "num_tokens": 541380221.0,
      "step": 683
    },
    {
      "epoch": 0.9339477726574501,
      "grad_norm": 0.10428246855735779,
      "learning_rate": 3.30651858962e-05,
      "loss": 0.4311,
      "num_tokens": 542184640.0,
      "step": 684
    },
    {
      "epoch": 0.9353131933777095,
      "grad_norm": 0.10955273360013962,
      "learning_rate": 3.304426347108932e-05,
      "loss": 0.4734,
      "num_tokens": 543036383.0,
      "step": 685
    },
    {
      "epoch": 0.9366786140979689,
      "grad_norm": 0.11666237562894821,
      "learning_rate": 3.302331708774947e-05,
      "loss": 0.4595,
      "num_tokens": 543806341.0,
      "step": 686
    },
    {
      "epoch": 0.9380440348182284,
      "grad_norm": 0.12301596254110336,
      "learning_rate": 3.300234679161928e-05,
      "loss": 0.4909,
      "num_tokens": 544661203.0,
      "step": 687
    },
    {
      "epoch": 0.9394094555384878,
      "grad_norm": 0.10711503773927689,
      "learning_rate": 3.298135262818944e-05,
      "loss": 0.4581,
      "num_tokens": 545504542.0,
      "step": 688
    },
    {
      "epoch": 0.9407748762587472,
      "grad_norm": 0.11774254590272903,
      "learning_rate": 3.296033464300241e-05,
      "loss": 0.4554,
      "num_tokens": 546304403.0,
      "step": 689
    },
    {
      "epoch": 0.9421402969790067,
      "grad_norm": 0.10866866260766983,
      "learning_rate": 3.293929288165234e-05,
      "loss": 0.4574,
      "num_tokens": 547060229.0,
      "step": 690
    },
    {
      "epoch": 0.943505717699266,
      "grad_norm": 0.10932304710149765,
      "learning_rate": 3.291822738978494e-05,
      "loss": 0.4535,
      "num_tokens": 547916549.0,
      "step": 691
    },
    {
      "epoch": 0.9448711384195255,
      "grad_norm": 0.11324688792228699,
      "learning_rate": 3.289713821309741e-05,
      "loss": 0.4764,
      "num_tokens": 548722315.0,
      "step": 692
    },
    {
      "epoch": 0.946236559139785,
      "grad_norm": 0.11028104275465012,
      "learning_rate": 3.287602539733832e-05,
      "loss": 0.4569,
      "num_tokens": 549493024.0,
      "step": 693
    },
    {
      "epoch": 0.9476019798600444,
      "grad_norm": 0.11328945308923721,
      "learning_rate": 3.2854888988307536e-05,
      "loss": 0.4723,
      "num_tokens": 550260246.0,
      "step": 694
    },
    {
      "epoch": 0.9489674005803038,
      "grad_norm": 0.1130959764122963,
      "learning_rate": 3.2833729031856086e-05,
      "loss": 0.4451,
      "num_tokens": 550998721.0,
      "step": 695
    },
    {
      "epoch": 0.9503328213005633,
      "grad_norm": 0.10879845917224884,
      "learning_rate": 3.281254557388609e-05,
      "loss": 0.4655,
      "num_tokens": 551779998.0,
      "step": 696
    },
    {
      "epoch": 0.9516982420208226,
      "grad_norm": 0.11777359992265701,
      "learning_rate": 3.279133866035065e-05,
      "loss": 0.4726,
      "num_tokens": 552566383.0,
      "step": 697
    },
    {
      "epoch": 0.9530636627410821,
      "grad_norm": 0.11379359662532806,
      "learning_rate": 3.277010833725375e-05,
      "loss": 0.4664,
      "num_tokens": 553369153.0,
      "step": 698
    },
    {
      "epoch": 0.9544290834613415,
      "grad_norm": 0.11348449438810349,
      "learning_rate": 3.2748854650650156e-05,
      "loss": 0.4573,
      "num_tokens": 554144592.0,
      "step": 699
    },
    {
      "epoch": 0.955794504181601,
      "grad_norm": 0.11871406435966492,
      "learning_rate": 3.27275776466453e-05,
      "loss": 0.4427,
      "num_tokens": 554927484.0,
      "step": 700
    },
    {
      "epoch": 0.9571599249018604,
      "grad_norm": 0.10621630400419235,
      "learning_rate": 3.270627737139522e-05,
      "loss": 0.4263,
      "num_tokens": 555711746.0,
      "step": 701
    },
    {
      "epoch": 0.9585253456221198,
      "grad_norm": 0.1147707998752594,
      "learning_rate": 3.268495387110644e-05,
      "loss": 0.4644,
      "num_tokens": 556568564.0,
      "step": 702
    },
    {
      "epoch": 0.9598907663423792,
      "grad_norm": 0.12247543036937714,
      "learning_rate": 3.266360719203582e-05,
      "loss": 0.444,
      "num_tokens": 557390732.0,
      "step": 703
    },
    {
      "epoch": 0.9612561870626387,
      "grad_norm": 0.1106138676404953,
      "learning_rate": 3.264223738049056e-05,
      "loss": 0.4747,
      "num_tokens": 558211621.0,
      "step": 704
    },
    {
      "epoch": 0.9626216077828981,
      "grad_norm": 0.10380808264017105,
      "learning_rate": 3.262084448282801e-05,
      "loss": 0.4311,
      "num_tokens": 558954534.0,
      "step": 705
    },
    {
      "epoch": 0.9639870285031575,
      "grad_norm": 0.11856523156166077,
      "learning_rate": 3.259942854545559e-05,
      "loss": 0.4667,
      "num_tokens": 559792223.0,
      "step": 706
    },
    {
      "epoch": 0.965352449223417,
      "grad_norm": 0.11219717562198639,
      "learning_rate": 3.257798961483074e-05,
      "loss": 0.4349,
      "num_tokens": 560465831.0,
      "step": 707
    },
    {
      "epoch": 0.9667178699436764,
      "grad_norm": 0.12004894018173218,
      "learning_rate": 3.2556527737460744e-05,
      "loss": 0.4745,
      "num_tokens": 561230502.0,
      "step": 708
    },
    {
      "epoch": 0.9680832906639358,
      "grad_norm": 0.1109546348452568,
      "learning_rate": 3.2535042959902663e-05,
      "loss": 0.4509,
      "num_tokens": 561984119.0,
      "step": 709
    },
    {
      "epoch": 0.9694487113841953,
      "grad_norm": 0.12040089070796967,
      "learning_rate": 3.2513535328763264e-05,
      "loss": 0.4505,
      "num_tokens": 562775181.0,
      "step": 710
    },
    {
      "epoch": 0.9708141321044547,
      "grad_norm": 0.11058896780014038,
      "learning_rate": 3.249200489069886e-05,
      "loss": 0.4561,
      "num_tokens": 563597668.0,
      "step": 711
    },
    {
      "epoch": 0.9721795528247141,
      "grad_norm": 0.12361027300357819,
      "learning_rate": 3.247045169241526e-05,
      "loss": 0.5027,
      "num_tokens": 564399768.0,
      "step": 712
    },
    {
      "epoch": 0.9735449735449735,
      "grad_norm": 0.11781793087720871,
      "learning_rate": 3.2448875780667637e-05,
      "loss": 0.4398,
      "num_tokens": 565164650.0,
      "step": 713
    },
    {
      "epoch": 0.974910394265233,
      "grad_norm": 0.11737298220396042,
      "learning_rate": 3.242727720226044e-05,
      "loss": 0.4533,
      "num_tokens": 565870717.0,
      "step": 714
    },
    {
      "epoch": 0.9762758149854924,
      "grad_norm": 0.12257419526576996,
      "learning_rate": 3.240565600404727e-05,
      "loss": 0.4688,
      "num_tokens": 566665118.0,
      "step": 715
    },
    {
      "epoch": 0.9776412357057518,
      "grad_norm": 0.11644288897514343,
      "learning_rate": 3.2384012232930835e-05,
      "loss": 0.4411,
      "num_tokens": 567407714.0,
      "step": 716
    },
    {
      "epoch": 0.9790066564260113,
      "grad_norm": 0.10999421775341034,
      "learning_rate": 3.236234593586278e-05,
      "loss": 0.4631,
      "num_tokens": 568286277.0,
      "step": 717
    },
    {
      "epoch": 0.9803720771462707,
      "grad_norm": 0.11726582050323486,
      "learning_rate": 3.234065715984361e-05,
      "loss": 0.4602,
      "num_tokens": 569034498.0,
      "step": 718
    },
    {
      "epoch": 0.9817374978665301,
      "grad_norm": 0.11371101438999176,
      "learning_rate": 3.2318945951922626e-05,
      "loss": 0.4533,
      "num_tokens": 569782220.0,
      "step": 719
    },
    {
      "epoch": 0.9831029185867896,
      "grad_norm": 0.11760714650154114,
      "learning_rate": 3.2297212359197765e-05,
      "loss": 0.4523,
      "num_tokens": 570604213.0,
      "step": 720
    },
    {
      "epoch": 0.984468339307049,
      "grad_norm": 0.11869388818740845,
      "learning_rate": 3.227545642881553e-05,
      "loss": 0.4398,
      "num_tokens": 571330997.0,
      "step": 721
    },
    {
      "epoch": 0.9858337600273084,
      "grad_norm": 0.11762310564517975,
      "learning_rate": 3.2253678207970874e-05,
      "loss": 0.479,
      "num_tokens": 572227227.0,
      "step": 722
    },
    {
      "epoch": 0.9871991807475679,
      "grad_norm": 0.11605551838874817,
      "learning_rate": 3.223187774390712e-05,
      "loss": 0.4444,
      "num_tokens": 572980517.0,
      "step": 723
    },
    {
      "epoch": 0.9885646014678273,
      "grad_norm": 0.11197071522474289,
      "learning_rate": 3.221005508391583e-05,
      "loss": 0.4411,
      "num_tokens": 573776549.0,
      "step": 724
    },
    {
      "epoch": 0.9899300221880867,
      "grad_norm": 0.10938811302185059,
      "learning_rate": 3.218821027533671e-05,
      "loss": 0.4515,
      "num_tokens": 574592007.0,
      "step": 725
    },
    {
      "epoch": 0.9912954429083461,
      "grad_norm": 0.1079869419336319,
      "learning_rate": 3.2166343365557534e-05,
      "loss": 0.4487,
      "num_tokens": 575339898.0,
      "step": 726
    },
    {
      "epoch": 0.9926608636286056,
      "grad_norm": 0.11661006510257721,
      "learning_rate": 3.214445440201401e-05,
      "loss": 0.4705,
      "num_tokens": 576108560.0,
      "step": 727
    },
    {
      "epoch": 0.994026284348865,
      "grad_norm": 0.11490686237812042,
      "learning_rate": 3.2122543432189685e-05,
      "loss": 0.4861,
      "num_tokens": 576901545.0,
      "step": 728
    },
    {
      "epoch": 0.9953917050691244,
      "grad_norm": 0.11197572946548462,
      "learning_rate": 3.210061050361583e-05,
      "loss": 0.4466,
      "num_tokens": 577667593.0,
      "step": 729
    },
    {
      "epoch": 0.9967571257893838,
      "grad_norm": 0.11252836883068085,
      "learning_rate": 3.207865566387138e-05,
      "loss": 0.4511,
      "num_tokens": 578505417.0,
      "step": 730
    },
    {
      "epoch": 0.9981225465096433,
      "grad_norm": 0.10709596425294876,
      "learning_rate": 3.2056678960582776e-05,
      "loss": 0.4585,
      "num_tokens": 579272925.0,
      "step": 731
    },
    {
      "epoch": 0.9994879672299027,
      "grad_norm": 0.10882391780614853,
      "learning_rate": 3.203468044142391e-05,
      "loss": 0.449,
      "num_tokens": 580063320.0,
      "step": 732
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.10882391780614853,
      "learning_rate": 3.2012660154115974e-05,
      "loss": 0.4664,
      "num_tokens": 580282057.0,
      "step": 733
    },
    {
      "epoch": 1.0013654207202594,
      "grad_norm": 0.17368696630001068,
      "learning_rate": 3.19906181464274e-05,
      "loss": 0.4282,
      "num_tokens": 581067400.0,
      "step": 734
    },
    {
      "epoch": 1.002730841440519,
      "grad_norm": 0.1297268271446228,
      "learning_rate": 3.1968554466173736e-05,
      "loss": 0.4487,
      "num_tokens": 581868341.0,
      "step": 735
    },
    {
      "epoch": 1.0040962621607783,
      "grad_norm": 0.12596215307712555,
      "learning_rate": 3.1946469161217534e-05,
      "loss": 0.4261,
      "num_tokens": 582584415.0,
      "step": 736
    },
    {
      "epoch": 1.0054616828810377,
      "grad_norm": 0.1078345775604248,
      "learning_rate": 3.192436227946828e-05,
      "loss": 0.4471,
      "num_tokens": 583391225.0,
      "step": 737
    },
    {
      "epoch": 1.0068271036012972,
      "grad_norm": 0.13533489406108856,
      "learning_rate": 3.190223386888223e-05,
      "loss": 0.4275,
      "num_tokens": 584163598.0,
      "step": 738
    },
    {
      "epoch": 1.0081925243215566,
      "grad_norm": 0.12527130544185638,
      "learning_rate": 3.188008397746236e-05,
      "loss": 0.4214,
      "num_tokens": 584922674.0,
      "step": 739
    },
    {
      "epoch": 1.009557945041816,
      "grad_norm": 0.11368127912282944,
      "learning_rate": 3.185791265325826e-05,
      "loss": 0.4423,
      "num_tokens": 585730191.0,
      "step": 740
    },
    {
      "epoch": 1.0109233657620755,
      "grad_norm": 0.13352465629577637,
      "learning_rate": 3.1835719944365994e-05,
      "loss": 0.4444,
      "num_tokens": 586537712.0,
      "step": 741
    },
    {
      "epoch": 1.012288786482335,
      "grad_norm": 0.11526013165712357,
      "learning_rate": 3.1813505898928017e-05,
      "loss": 0.4231,
      "num_tokens": 587407480.0,
      "step": 742
    },
    {
      "epoch": 1.0136542072025942,
      "grad_norm": 0.11485767364501953,
      "learning_rate": 3.179127056513309e-05,
      "loss": 0.4731,
      "num_tokens": 588195784.0,
      "step": 743
    },
    {
      "epoch": 1.0150196279228538,
      "grad_norm": 0.1315630078315735,
      "learning_rate": 3.176901399121611e-05,
      "loss": 0.4311,
      "num_tokens": 588969328.0,
      "step": 744
    },
    {
      "epoch": 1.0163850486431132,
      "grad_norm": 0.11521752178668976,
      "learning_rate": 3.1746736225458104e-05,
      "loss": 0.4347,
      "num_tokens": 589710448.0,
      "step": 745
    },
    {
      "epoch": 1.0177504693633725,
      "grad_norm": 0.12397229671478271,
      "learning_rate": 3.172443731618602e-05,
      "loss": 0.4183,
      "num_tokens": 590567266.0,
      "step": 746
    },
    {
      "epoch": 1.0191158900836321,
      "grad_norm": 0.1110033318400383,
      "learning_rate": 3.1702117311772725e-05,
      "loss": 0.4355,
      "num_tokens": 591362903.0,
      "step": 747
    },
    {
      "epoch": 1.0204813108038915,
      "grad_norm": 0.13571684062480927,
      "learning_rate": 3.16797762606368e-05,
      "loss": 0.447,
      "num_tokens": 592198261.0,
      "step": 748
    },
    {
      "epoch": 1.0218467315241508,
      "grad_norm": 0.12022636830806732,
      "learning_rate": 3.165741421124251e-05,
      "loss": 0.4591,
      "num_tokens": 593018491.0,
      "step": 749
    },
    {
      "epoch": 1.0232121522444102,
      "grad_norm": 0.11951100081205368,
      "learning_rate": 3.163503121209966e-05,
      "loss": 0.4451,
      "num_tokens": 593832905.0,
      "step": 750
    },
    {
      "epoch": 1.0245775729646698,
      "grad_norm": 0.12160904705524445,
      "learning_rate": 3.161262731176351e-05,
      "loss": 0.4297,
      "num_tokens": 594630145.0,
      "step": 751
    },
    {
      "epoch": 1.0259429936849291,
      "grad_norm": 0.1131938025355339,
      "learning_rate": 3.159020255883464e-05,
      "loss": 0.4157,
      "num_tokens": 595353817.0,
      "step": 752
    },
    {
      "epoch": 1.0273084144051885,
      "grad_norm": 0.11528072506189346,
      "learning_rate": 3.15677570019589e-05,
      "loss": 0.4309,
      "num_tokens": 596113722.0,
      "step": 753
    },
    {
      "epoch": 1.028673835125448,
      "grad_norm": 0.10757298767566681,
      "learning_rate": 3.154529068982724e-05,
      "loss": 0.4602,
      "num_tokens": 596946163.0,
      "step": 754
    },
    {
      "epoch": 1.0300392558457074,
      "grad_norm": 0.11627766489982605,
      "learning_rate": 3.152280367117565e-05,
      "loss": 0.4342,
      "num_tokens": 597795740.0,
      "step": 755
    },
    {
      "epoch": 1.0314046765659668,
      "grad_norm": 0.10605552047491074,
      "learning_rate": 3.150029599478502e-05,
      "loss": 0.4311,
      "num_tokens": 598620339.0,
      "step": 756
    },
    {
      "epoch": 1.0327700972862264,
      "grad_norm": 0.11296802759170532,
      "learning_rate": 3.147776770948108e-05,
      "loss": 0.4345,
      "num_tokens": 599444342.0,
      "step": 757
    },
    {
      "epoch": 1.0341355180064857,
      "grad_norm": 0.1058763861656189,
      "learning_rate": 3.145521886413425e-05,
      "loss": 0.4543,
      "num_tokens": 600228411.0,
      "step": 758
    },
    {
      "epoch": 1.035500938726745,
      "grad_norm": 0.11110397428274155,
      "learning_rate": 3.143264950765954e-05,
      "loss": 0.4284,
      "num_tokens": 601133346.0,
      "step": 759
    },
    {
      "epoch": 1.0368663594470047,
      "grad_norm": 0.10711853206157684,
      "learning_rate": 3.1410059689016485e-05,
      "loss": 0.4311,
      "num_tokens": 601917178.0,
      "step": 760
    },
    {
      "epoch": 1.038231780167264,
      "grad_norm": 0.11388956755399704,
      "learning_rate": 3.138744945720899e-05,
      "loss": 0.4351,
      "num_tokens": 602679326.0,
      "step": 761
    },
    {
      "epoch": 1.0395972008875234,
      "grad_norm": 0.12199167907238007,
      "learning_rate": 3.136481886128523e-05,
      "loss": 0.4634,
      "num_tokens": 603469224.0,
      "step": 762
    },
    {
      "epoch": 1.040962621607783,
      "grad_norm": 0.12296682596206665,
      "learning_rate": 3.134216795033758e-05,
      "loss": 0.4242,
      "num_tokens": 604267016.0,
      "step": 763
    },
    {
      "epoch": 1.0423280423280423,
      "grad_norm": 0.10790324211120605,
      "learning_rate": 3.131949677350248e-05,
      "loss": 0.4198,
      "num_tokens": 605042770.0,
      "step": 764
    },
    {
      "epoch": 1.0436934630483017,
      "grad_norm": 0.119001105427742,
      "learning_rate": 3.12968053799603e-05,
      "loss": 0.4328,
      "num_tokens": 605829265.0,
      "step": 765
    },
    {
      "epoch": 1.0450588837685613,
      "grad_norm": 0.11803442984819412,
      "learning_rate": 3.127409381893531e-05,
      "loss": 0.438,
      "num_tokens": 606628098.0,
      "step": 766
    },
    {
      "epoch": 1.0464243044888206,
      "grad_norm": 0.10707368701696396,
      "learning_rate": 3.125136213969552e-05,
      "loss": 0.4581,
      "num_tokens": 607451366.0,
      "step": 767
    },
    {
      "epoch": 1.04778972520908,
      "grad_norm": 0.1050037369132042,
      "learning_rate": 3.1228610391552556e-05,
      "loss": 0.4443,
      "num_tokens": 608238948.0,
      "step": 768
    },
    {
      "epoch": 1.0491551459293396,
      "grad_norm": 0.11791109293699265,
      "learning_rate": 3.12058386238616e-05,
      "loss": 0.423,
      "num_tokens": 608989128.0,
      "step": 769
    },
    {
      "epoch": 1.050520566649599,
      "grad_norm": 0.12111836671829224,
      "learning_rate": 3.1183046886021266e-05,
      "loss": 0.4631,
      "num_tokens": 609812552.0,
      "step": 770
    },
    {
      "epoch": 1.0518859873698583,
      "grad_norm": 0.10765352100133896,
      "learning_rate": 3.116023522747348e-05,
      "loss": 0.41,
      "num_tokens": 610528969.0,
      "step": 771
    },
    {
      "epoch": 1.0532514080901179,
      "grad_norm": 0.13161171972751617,
      "learning_rate": 3.1137403697703376e-05,
      "loss": 0.4206,
      "num_tokens": 611323222.0,
      "step": 772
    },
    {
      "epoch": 1.0546168288103772,
      "grad_norm": 0.11381059139966965,
      "learning_rate": 3.111455234623921e-05,
      "loss": 0.428,
      "num_tokens": 612116787.0,
      "step": 773
    },
    {
      "epoch": 1.0559822495306366,
      "grad_norm": 0.1220988780260086,
      "learning_rate": 3.109168122265222e-05,
      "loss": 0.4192,
      "num_tokens": 612940436.0,
      "step": 774
    },
    {
      "epoch": 1.0573476702508962,
      "grad_norm": 0.12938185036182404,
      "learning_rate": 3.106879037655656e-05,
      "loss": 0.4266,
      "num_tokens": 613738093.0,
      "step": 775
    },
    {
      "epoch": 1.0587130909711555,
      "grad_norm": 0.11425158381462097,
      "learning_rate": 3.1045879857609146e-05,
      "loss": 0.4036,
      "num_tokens": 614440034.0,
      "step": 776
    },
    {
      "epoch": 1.0600785116914149,
      "grad_norm": 0.11348623037338257,
      "learning_rate": 3.102294971550957e-05,
      "loss": 0.4724,
      "num_tokens": 615320263.0,
      "step": 777
    },
    {
      "epoch": 1.0614439324116745,
      "grad_norm": 0.12750263512134552,
      "learning_rate": 3.1e-05,
      "loss": 0.4348,
      "num_tokens": 616161957.0,
      "step": 778
    },
    {
      "epoch": 1.0628093531319338,
      "grad_norm": 0.10571657866239548,
      "learning_rate": 3.097703076086508e-05,
      "loss": 0.4155,
      "num_tokens": 616983170.0,
      "step": 779
    },
    {
      "epoch": 1.0641747738521932,
      "grad_norm": 0.1338396966457367,
      "learning_rate": 3.095404204793176e-05,
      "loss": 0.4425,
      "num_tokens": 617763729.0,
      "step": 780
    },
    {
      "epoch": 1.0655401945724527,
      "grad_norm": 0.10509327799081802,
      "learning_rate": 3.093103391106929e-05,
      "loss": 0.4442,
      "num_tokens": 618562540.0,
      "step": 781
    },
    {
      "epoch": 1.066905615292712,
      "grad_norm": 0.12091317027807236,
      "learning_rate": 3.090800640018903e-05,
      "loss": 0.4433,
      "num_tokens": 619458953.0,
      "step": 782
    },
    {
      "epoch": 1.0682710360129715,
      "grad_norm": 0.10914302617311478,
      "learning_rate": 3.088495956524435e-05,
      "loss": 0.4335,
      "num_tokens": 620291119.0,
      "step": 783
    },
    {
      "epoch": 1.0696364567332308,
      "grad_norm": 0.12517289817333221,
      "learning_rate": 3.086189345623058e-05,
      "loss": 0.4477,
      "num_tokens": 621102388.0,
      "step": 784
    },
    {
      "epoch": 1.0710018774534904,
      "grad_norm": 0.1077839583158493,
      "learning_rate": 3.083880812318482e-05,
      "loss": 0.415,
      "num_tokens": 621881412.0,
      "step": 785
    },
    {
      "epoch": 1.0723672981737498,
      "grad_norm": 0.1311974972486496,
      "learning_rate": 3.0815703616185886e-05,
      "loss": 0.422,
      "num_tokens": 622587541.0,
      "step": 786
    },
    {
      "epoch": 1.0737327188940091,
      "grad_norm": 0.10586967319250107,
      "learning_rate": 3.079257998535423e-05,
      "loss": 0.4175,
      "num_tokens": 623380285.0,
      "step": 787
    },
    {
      "epoch": 1.0750981396142687,
      "grad_norm": 0.11513744294643402,
      "learning_rate": 3.076943728085171e-05,
      "loss": 0.4447,
      "num_tokens": 624120468.0,
      "step": 788
    },
    {
      "epoch": 1.076463560334528,
      "grad_norm": 0.11171440035104752,
      "learning_rate": 3.074627555288163e-05,
      "loss": 0.4397,
      "num_tokens": 624926232.0,
      "step": 789
    },
    {
      "epoch": 1.0778289810547874,
      "grad_norm": 0.1062481477856636,
      "learning_rate": 3.072309485168853e-05,
      "loss": 0.4587,
      "num_tokens": 625745715.0,
      "step": 790
    },
    {
      "epoch": 1.079194401775047,
      "grad_norm": 0.09897271543741226,
      "learning_rate": 3.0699895227558105e-05,
      "loss": 0.4395,
      "num_tokens": 626535481.0,
      "step": 791
    },
    {
      "epoch": 1.0805598224953064,
      "grad_norm": 0.11536641418933868,
      "learning_rate": 3.067667673081711e-05,
      "loss": 0.4216,
      "num_tokens": 627292591.0,
      "step": 792
    },
    {
      "epoch": 1.0819252432155657,
      "grad_norm": 0.10827899724245071,
      "learning_rate": 3.065343941183324e-05,
      "loss": 0.4441,
      "num_tokens": 628060331.0,
      "step": 793
    },
    {
      "epoch": 1.0832906639358253,
      "grad_norm": 0.11871815472841263,
      "learning_rate": 3.063018332101502e-05,
      "loss": 0.4388,
      "num_tokens": 628821908.0,
      "step": 794
    },
    {
      "epoch": 1.0846560846560847,
      "grad_norm": 0.1107751727104187,
      "learning_rate": 3.060690850881169e-05,
      "loss": 0.4221,
      "num_tokens": 629639920.0,
      "step": 795
    },
    {
      "epoch": 1.086021505376344,
      "grad_norm": 0.1242925301194191,
      "learning_rate": 3.058361502571312e-05,
      "loss": 0.4207,
      "num_tokens": 630372733.0,
      "step": 796
    },
    {
      "epoch": 1.0873869260966036,
      "grad_norm": 0.1030455008149147,
      "learning_rate": 3.056030292224964e-05,
      "loss": 0.444,
      "num_tokens": 631146173.0,
      "step": 797
    },
    {
      "epoch": 1.088752346816863,
      "grad_norm": 0.11843367666006088,
      "learning_rate": 3.053697224899203e-05,
      "loss": 0.4508,
      "num_tokens": 631939303.0,
      "step": 798
    },
    {
      "epoch": 1.0901177675371223,
      "grad_norm": 0.12166637927293777,
      "learning_rate": 3.0513623056551323e-05,
      "loss": 0.4468,
      "num_tokens": 632761525.0,
      "step": 799
    },
    {
      "epoch": 1.0914831882573819,
      "grad_norm": 0.10679758340120316,
      "learning_rate": 3.0490255395578713e-05,
      "loss": 0.4108,
      "num_tokens": 633492889.0,
      "step": 800
    },
    {
      "epoch": 1.0928486089776412,
      "grad_norm": 0.1207294836640358,
      "learning_rate": 3.0466869316765487e-05,
      "loss": 0.4237,
      "num_tokens": 634231405.0,
      "step": 801
    },
    {
      "epoch": 1.0942140296979006,
      "grad_norm": 0.10565589368343353,
      "learning_rate": 3.0443464870842873e-05,
      "loss": 0.4157,
      "num_tokens": 635034883.0,
      "step": 802
    },
    {
      "epoch": 1.0955794504181602,
      "grad_norm": 0.12499801814556122,
      "learning_rate": 3.0420042108581934e-05,
      "loss": 0.4387,
      "num_tokens": 635851850.0,
      "step": 803
    },
    {
      "epoch": 1.0969448711384195,
      "grad_norm": 0.10761891305446625,
      "learning_rate": 3.039660108079348e-05,
      "loss": 0.4605,
      "num_tokens": 636660356.0,
      "step": 804
    },
    {
      "epoch": 1.098310291858679,
      "grad_norm": 0.11971672624349594,
      "learning_rate": 3.037314183832793e-05,
      "loss": 0.4136,
      "num_tokens": 637411850.0,
      "step": 805
    },
    {
      "epoch": 1.0996757125789385,
      "grad_norm": 0.10417275875806808,
      "learning_rate": 3.0349664432075227e-05,
      "loss": 0.4425,
      "num_tokens": 638337522.0,
      "step": 806
    },
    {
      "epoch": 1.1010411332991978,
      "grad_norm": 0.11194221675395966,
      "learning_rate": 3.0326168912964725e-05,
      "loss": 0.4061,
      "num_tokens": 639129354.0,
      "step": 807
    },
    {
      "epoch": 1.1024065540194572,
      "grad_norm": 0.1077108085155487,
      "learning_rate": 3.0302655331965044e-05,
      "loss": 0.4335,
      "num_tokens": 639930835.0,
      "step": 808
    },
    {
      "epoch": 1.1037719747397168,
      "grad_norm": 0.11484408378601074,
      "learning_rate": 3.0279123740084005e-05,
      "loss": 0.4494,
      "num_tokens": 640722963.0,
      "step": 809
    },
    {
      "epoch": 1.1051373954599761,
      "grad_norm": 0.12186506390571594,
      "learning_rate": 3.02555741883685e-05,
      "loss": 0.4286,
      "num_tokens": 641485325.0,
      "step": 810
    },
    {
      "epoch": 1.1065028161802355,
      "grad_norm": 0.11648596823215485,
      "learning_rate": 3.0232006727904368e-05,
      "loss": 0.421,
      "num_tokens": 642206871.0,
      "step": 811
    },
    {
      "epoch": 1.107868236900495,
      "grad_norm": 0.10721496492624283,
      "learning_rate": 3.0208421409816303e-05,
      "loss": 0.4457,
      "num_tokens": 643062873.0,
      "step": 812
    },
    {
      "epoch": 1.1092336576207544,
      "grad_norm": 0.11282586306333542,
      "learning_rate": 3.018481828526776e-05,
      "loss": 0.4279,
      "num_tokens": 643898720.0,
      "step": 813
    },
    {
      "epoch": 1.1105990783410138,
      "grad_norm": 0.11510273069143295,
      "learning_rate": 3.0161197405460774e-05,
      "loss": 0.4247,
      "num_tokens": 644625027.0,
      "step": 814
    },
    {
      "epoch": 1.1119644990612731,
      "grad_norm": 0.1229609027504921,
      "learning_rate": 3.013755882163595e-05,
      "loss": 0.4254,
      "num_tokens": 645391645.0,
      "step": 815
    },
    {
      "epoch": 1.1133299197815327,
      "grad_norm": 0.10373496264219284,
      "learning_rate": 3.0113902585072252e-05,
      "loss": 0.4144,
      "num_tokens": 646171211.0,
      "step": 816
    },
    {
      "epoch": 1.114695340501792,
      "grad_norm": 0.1307738721370697,
      "learning_rate": 3.0090228747086968e-05,
      "loss": 0.4584,
      "num_tokens": 646974194.0,
      "step": 817
    },
    {
      "epoch": 1.1160607612220514,
      "grad_norm": 0.11937116831541061,
      "learning_rate": 3.0066537359035554e-05,
      "loss": 0.4411,
      "num_tokens": 647853935.0,
      "step": 818
    },
    {
      "epoch": 1.117426181942311,
      "grad_norm": 0.1376524120569229,
      "learning_rate": 3.0042828472311548e-05,
      "loss": 0.4538,
      "num_tokens": 648683508.0,
      "step": 819
    },
    {
      "epoch": 1.1187916026625704,
      "grad_norm": 0.10297880321741104,
      "learning_rate": 3.0019102138346427e-05,
      "loss": 0.4409,
      "num_tokens": 649516172.0,
      "step": 820
    },
    {
      "epoch": 1.1201570233828297,
      "grad_norm": 0.1131322905421257,
      "learning_rate": 2.9995358408609547e-05,
      "loss": 0.4158,
      "num_tokens": 650271262.0,
      "step": 821
    },
    {
      "epoch": 1.1215224441030893,
      "grad_norm": 0.12518459558486938,
      "learning_rate": 2.9971597334607977e-05,
      "loss": 0.4376,
      "num_tokens": 651069702.0,
      "step": 822
    },
    {
      "epoch": 1.1228878648233487,
      "grad_norm": 0.10958273708820343,
      "learning_rate": 2.9947818967886416e-05,
      "loss": 0.4324,
      "num_tokens": 651875285.0,
      "step": 823
    },
    {
      "epoch": 1.124253285543608,
      "grad_norm": 0.11421357095241547,
      "learning_rate": 2.9924023360027078e-05,
      "loss": 0.4336,
      "num_tokens": 652646822.0,
      "step": 824
    },
    {
      "epoch": 1.1256187062638676,
      "grad_norm": 0.12892921268939972,
      "learning_rate": 2.9900210562649587e-05,
      "loss": 0.4128,
      "num_tokens": 653452644.0,
      "step": 825
    },
    {
      "epoch": 1.126984126984127,
      "grad_norm": 0.11748889833688736,
      "learning_rate": 2.9876380627410832e-05,
      "loss": 0.4549,
      "num_tokens": 654200606.0,
      "step": 826
    },
    {
      "epoch": 1.1283495477043863,
      "grad_norm": 0.10513462871313095,
      "learning_rate": 2.98525336060049e-05,
      "loss": 0.427,
      "num_tokens": 654981082.0,
      "step": 827
    },
    {
      "epoch": 1.129714968424646,
      "grad_norm": 0.11560460925102234,
      "learning_rate": 2.9828669550162942e-05,
      "loss": 0.4264,
      "num_tokens": 655759983.0,
      "step": 828
    },
    {
      "epoch": 1.1310803891449053,
      "grad_norm": 0.11454298347234726,
      "learning_rate": 2.9804788511653054e-05,
      "loss": 0.4092,
      "num_tokens": 656463429.0,
      "step": 829
    },
    {
      "epoch": 1.1324458098651646,
      "grad_norm": 0.10877662152051926,
      "learning_rate": 2.9780890542280173e-05,
      "loss": 0.3918,
      "num_tokens": 657203221.0,
      "step": 830
    },
    {
      "epoch": 1.1338112305854242,
      "grad_norm": 0.10893647372722626,
      "learning_rate": 2.975697569388596e-05,
      "loss": 0.4093,
      "num_tokens": 657915956.0,
      "step": 831
    },
    {
      "epoch": 1.1351766513056836,
      "grad_norm": 0.11820916831493378,
      "learning_rate": 2.9733044018348707e-05,
      "loss": 0.4249,
      "num_tokens": 658671188.0,
      "step": 832
    },
    {
      "epoch": 1.136542072025943,
      "grad_norm": 0.11217863857746124,
      "learning_rate": 2.9709095567583194e-05,
      "loss": 0.4284,
      "num_tokens": 659417387.0,
      "step": 833
    },
    {
      "epoch": 1.1379074927462025,
      "grad_norm": 0.12120455503463745,
      "learning_rate": 2.96851303935406e-05,
      "loss": 0.4417,
      "num_tokens": 660166645.0,
      "step": 834
    },
    {
      "epoch": 1.1392729134664619,
      "grad_norm": 0.1150188073515892,
      "learning_rate": 2.966114854820837e-05,
      "loss": 0.4466,
      "num_tokens": 660921528.0,
      "step": 835
    },
    {
      "epoch": 1.1406383341867212,
      "grad_norm": 0.11354726552963257,
      "learning_rate": 2.9637150083610136e-05,
      "loss": 0.4217,
      "num_tokens": 661723902.0,
      "step": 836
    },
    {
      "epoch": 1.1420037549069808,
      "grad_norm": 0.10787733644247055,
      "learning_rate": 2.961313505180556e-05,
      "loss": 0.4298,
      "num_tokens": 662547258.0,
      "step": 837
    },
    {
      "epoch": 1.1433691756272402,
      "grad_norm": 0.11528708785772324,
      "learning_rate": 2.9589103504890254e-05,
      "loss": 0.4251,
      "num_tokens": 663342588.0,
      "step": 838
    },
    {
      "epoch": 1.1447345963474995,
      "grad_norm": 0.12706610560417175,
      "learning_rate": 2.9565055494995645e-05,
      "loss": 0.4415,
      "num_tokens": 664195675.0,
      "step": 839
    },
    {
      "epoch": 1.1461000170677589,
      "grad_norm": 0.10292981564998627,
      "learning_rate": 2.9540991074288895e-05,
      "loss": 0.4389,
      "num_tokens": 664964128.0,
      "step": 840
    },
    {
      "epoch": 1.1474654377880185,
      "grad_norm": 0.11292129755020142,
      "learning_rate": 2.9516910294972745e-05,
      "loss": 0.4308,
      "num_tokens": 665778002.0,
      "step": 841
    },
    {
      "epoch": 1.1488308585082778,
      "grad_norm": 0.11062616854906082,
      "learning_rate": 2.9492813209285428e-05,
      "loss": 0.4284,
      "num_tokens": 666591923.0,
      "step": 842
    },
    {
      "epoch": 1.1501962792285374,
      "grad_norm": 0.09791098535060883,
      "learning_rate": 2.9468699869500552e-05,
      "loss": 0.4335,
      "num_tokens": 667407564.0,
      "step": 843
    },
    {
      "epoch": 1.1515616999487968,
      "grad_norm": 0.11473805457353592,
      "learning_rate": 2.9444570327926983e-05,
      "loss": 0.454,
      "num_tokens": 668220160.0,
      "step": 844
    },
    {
      "epoch": 1.1529271206690561,
      "grad_norm": 0.10932657867670059,
      "learning_rate": 2.9420424636908755e-05,
      "loss": 0.4434,
      "num_tokens": 669076230.0,
      "step": 845
    },
    {
      "epoch": 1.1542925413893155,
      "grad_norm": 0.09519892930984497,
      "learning_rate": 2.9396262848824887e-05,
      "loss": 0.4226,
      "num_tokens": 669905056.0,
      "step": 846
    },
    {
      "epoch": 1.155657962109575,
      "grad_norm": 0.11010502278804779,
      "learning_rate": 2.937208501608936e-05,
      "loss": 0.4317,
      "num_tokens": 670758927.0,
      "step": 847
    },
    {
      "epoch": 1.1570233828298344,
      "grad_norm": 0.10503517836332321,
      "learning_rate": 2.934789119115095e-05,
      "loss": 0.4287,
      "num_tokens": 671499026.0,
      "step": 848
    },
    {
      "epoch": 1.158388803550094,
      "grad_norm": 0.11325167864561081,
      "learning_rate": 2.9323681426493117e-05,
      "loss": 0.4331,
      "num_tokens": 672272538.0,
      "step": 849
    },
    {
      "epoch": 1.1597542242703534,
      "grad_norm": 0.10296265780925751,
      "learning_rate": 2.9299455774633894e-05,
      "loss": 0.4418,
      "num_tokens": 673064366.0,
      "step": 850
    },
    {
      "epoch": 1.1611196449906127,
      "grad_norm": 0.13018940389156342,
      "learning_rate": 2.9275214288125794e-05,
      "loss": 0.4452,
      "num_tokens": 673825106.0,
      "step": 851
    },
    {
      "epoch": 1.162485065710872,
      "grad_norm": 0.11285494267940521,
      "learning_rate": 2.9250957019555674e-05,
      "loss": 0.4403,
      "num_tokens": 674615478.0,
      "step": 852
    },
    {
      "epoch": 1.1638504864311316,
      "grad_norm": 0.11723152548074722,
      "learning_rate": 2.9226684021544627e-05,
      "loss": 0.4225,
      "num_tokens": 675369115.0,
      "step": 853
    },
    {
      "epoch": 1.165215907151391,
      "grad_norm": 0.12146797776222229,
      "learning_rate": 2.9202395346747865e-05,
      "loss": 0.433,
      "num_tokens": 676141893.0,
      "step": 854
    },
    {
      "epoch": 1.1665813278716504,
      "grad_norm": 0.12038953602313995,
      "learning_rate": 2.9178091047854603e-05,
      "loss": 0.4506,
      "num_tokens": 676887538.0,
      "step": 855
    },
    {
      "epoch": 1.16794674859191,
      "grad_norm": 0.10361883789300919,
      "learning_rate": 2.915377117758796e-05,
      "loss": 0.4323,
      "num_tokens": 677687153.0,
      "step": 856
    },
    {
      "epoch": 1.1693121693121693,
      "grad_norm": 0.11310066282749176,
      "learning_rate": 2.9129435788704827e-05,
      "loss": 0.4543,
      "num_tokens": 678491181.0,
      "step": 857
    },
    {
      "epoch": 1.1706775900324287,
      "grad_norm": 0.12228952348232269,
      "learning_rate": 2.9105084933995763e-05,
      "loss": 0.42,
      "num_tokens": 679314820.0,
      "step": 858
    },
    {
      "epoch": 1.1720430107526882,
      "grad_norm": 0.10873158276081085,
      "learning_rate": 2.908071866628487e-05,
      "loss": 0.4262,
      "num_tokens": 680096424.0,
      "step": 859
    },
    {
      "epoch": 1.1734084314729476,
      "grad_norm": 0.102928526699543,
      "learning_rate": 2.9056337038429708e-05,
      "loss": 0.4449,
      "num_tokens": 680903665.0,
      "step": 860
    },
    {
      "epoch": 1.174773852193207,
      "grad_norm": 0.13183657824993134,
      "learning_rate": 2.9031940103321115e-05,
      "loss": 0.4524,
      "num_tokens": 681712526.0,
      "step": 861
    },
    {
      "epoch": 1.1761392729134665,
      "grad_norm": 0.10158593207597733,
      "learning_rate": 2.9007527913883176e-05,
      "loss": 0.4316,
      "num_tokens": 682565997.0,
      "step": 862
    },
    {
      "epoch": 1.177504693633726,
      "grad_norm": 0.10301385819911957,
      "learning_rate": 2.8983100523073044e-05,
      "loss": 0.4274,
      "num_tokens": 683391965.0,
      "step": 863
    },
    {
      "epoch": 1.1788701143539853,
      "grad_norm": 0.11103726923465729,
      "learning_rate": 2.895865798388086e-05,
      "loss": 0.4049,
      "num_tokens": 684162155.0,
      "step": 864
    },
    {
      "epoch": 1.1802355350742448,
      "grad_norm": 0.1082831546664238,
      "learning_rate": 2.8934200349329623e-05,
      "loss": 0.4245,
      "num_tokens": 684948818.0,
      "step": 865
    },
    {
      "epoch": 1.1816009557945042,
      "grad_norm": 0.09826285392045975,
      "learning_rate": 2.8909727672475072e-05,
      "loss": 0.4604,
      "num_tokens": 685791713.0,
      "step": 866
    },
    {
      "epoch": 1.1829663765147636,
      "grad_norm": 0.11977633833885193,
      "learning_rate": 2.888524000640558e-05,
      "loss": 0.4423,
      "num_tokens": 686591354.0,
      "step": 867
    },
    {
      "epoch": 1.1843317972350231,
      "grad_norm": 0.09989691525697708,
      "learning_rate": 2.886073740424205e-05,
      "loss": 0.4517,
      "num_tokens": 687372756.0,
      "step": 868
    },
    {
      "epoch": 1.1856972179552825,
      "grad_norm": 0.11419820040464401,
      "learning_rate": 2.8836219919137764e-05,
      "loss": 0.4133,
      "num_tokens": 688122631.0,
      "step": 869
    },
    {
      "epoch": 1.1870626386755418,
      "grad_norm": 0.10000214725732803,
      "learning_rate": 2.8811687604278297e-05,
      "loss": 0.4325,
      "num_tokens": 688884022.0,
      "step": 870
    },
    {
      "epoch": 1.1884280593958014,
      "grad_norm": 0.11071143299341202,
      "learning_rate": 2.8787140512881398e-05,
      "loss": 0.4346,
      "num_tokens": 689784422.0,
      "step": 871
    },
    {
      "epoch": 1.1897934801160608,
      "grad_norm": 0.10273734480142593,
      "learning_rate": 2.8762578698196876e-05,
      "loss": 0.4279,
      "num_tokens": 690579553.0,
      "step": 872
    },
    {
      "epoch": 1.1911589008363201,
      "grad_norm": 0.11361464858055115,
      "learning_rate": 2.8738002213506462e-05,
      "loss": 0.4586,
      "num_tokens": 691313002.0,
      "step": 873
    },
    {
      "epoch": 1.1925243215565797,
      "grad_norm": 0.10171885043382645,
      "learning_rate": 2.8713411112123722e-05,
      "loss": 0.4164,
      "num_tokens": 692071531.0,
      "step": 874
    },
    {
      "epoch": 1.193889742276839,
      "grad_norm": 0.11136980354785919,
      "learning_rate": 2.868880544739394e-05,
      "loss": 0.4281,
      "num_tokens": 692858578.0,
      "step": 875
    },
    {
      "epoch": 1.1952551629970984,
      "grad_norm": 0.09982937574386597,
      "learning_rate": 2.866418527269396e-05,
      "loss": 0.409,
      "num_tokens": 693638406.0,
      "step": 876
    },
    {
      "epoch": 1.1966205837173578,
      "grad_norm": 0.10511600226163864,
      "learning_rate": 2.8639550641432145e-05,
      "loss": 0.4457,
      "num_tokens": 694516711.0,
      "step": 877
    },
    {
      "epoch": 1.1979860044376174,
      "grad_norm": 0.13396698236465454,
      "learning_rate": 2.861490160704818e-05,
      "loss": 0.4429,
      "num_tokens": 695304100.0,
      "step": 878
    },
    {
      "epoch": 1.1993514251578767,
      "grad_norm": 0.0956239178776741,
      "learning_rate": 2.859023822301302e-05,
      "loss": 0.415,
      "num_tokens": 696084594.0,
      "step": 879
    },
    {
      "epoch": 1.2007168458781363,
      "grad_norm": 0.10699908435344696,
      "learning_rate": 2.856556054282873e-05,
      "loss": 0.4359,
      "num_tokens": 696922679.0,
      "step": 880
    },
    {
      "epoch": 1.2020822665983957,
      "grad_norm": 0.11450411379337311,
      "learning_rate": 2.854086862002841e-05,
      "loss": 0.4312,
      "num_tokens": 697733834.0,
      "step": 881
    },
    {
      "epoch": 1.203447687318655,
      "grad_norm": 0.11562763154506683,
      "learning_rate": 2.8516162508176034e-05,
      "loss": 0.4443,
      "num_tokens": 698545731.0,
      "step": 882
    },
    {
      "epoch": 1.2048131080389144,
      "grad_norm": 0.10740004479885101,
      "learning_rate": 2.8491442260866372e-05,
      "loss": 0.4204,
      "num_tokens": 699257938.0,
      "step": 883
    },
    {
      "epoch": 1.206178528759174,
      "grad_norm": 0.10913646221160889,
      "learning_rate": 2.8466707931724836e-05,
      "loss": 0.4142,
      "num_tokens": 700080664.0,
      "step": 884
    },
    {
      "epoch": 1.2075439494794333,
      "grad_norm": 0.12578341364860535,
      "learning_rate": 2.8441959574407415e-05,
      "loss": 0.4212,
      "num_tokens": 700868458.0,
      "step": 885
    },
    {
      "epoch": 1.2089093701996927,
      "grad_norm": 0.10489172488451004,
      "learning_rate": 2.841719724260052e-05,
      "loss": 0.4544,
      "num_tokens": 701616434.0,
      "step": 886
    },
    {
      "epoch": 1.2102747909199523,
      "grad_norm": 0.11545940488576889,
      "learning_rate": 2.8392420990020864e-05,
      "loss": 0.4489,
      "num_tokens": 702347503.0,
      "step": 887
    },
    {
      "epoch": 1.2116402116402116,
      "grad_norm": 0.11579377949237823,
      "learning_rate": 2.8367630870415364e-05,
      "loss": 0.4194,
      "num_tokens": 703076915.0,
      "step": 888
    },
    {
      "epoch": 1.213005632360471,
      "grad_norm": 0.11395800113677979,
      "learning_rate": 2.834282693756102e-05,
      "loss": 0.4492,
      "num_tokens": 703901744.0,
      "step": 889
    },
    {
      "epoch": 1.2143710530807306,
      "grad_norm": 0.10612054914236069,
      "learning_rate": 2.8318009245264814e-05,
      "loss": 0.4326,
      "num_tokens": 704671295.0,
      "step": 890
    },
    {
      "epoch": 1.21573647380099,
      "grad_norm": 0.10997232049703598,
      "learning_rate": 2.829317784736353e-05,
      "loss": 0.4349,
      "num_tokens": 705384296.0,
      "step": 891
    },
    {
      "epoch": 1.2171018945212493,
      "grad_norm": 0.12213108688592911,
      "learning_rate": 2.826833279772375e-05,
      "loss": 0.4358,
      "num_tokens": 706281848.0,
      "step": 892
    },
    {
      "epoch": 1.2184673152415089,
      "grad_norm": 0.10173822939395905,
      "learning_rate": 2.8243474150241602e-05,
      "loss": 0.4405,
      "num_tokens": 707091739.0,
      "step": 893
    },
    {
      "epoch": 1.2198327359617682,
      "grad_norm": 0.1089099571108818,
      "learning_rate": 2.821860195884276e-05,
      "loss": 0.4293,
      "num_tokens": 707873489.0,
      "step": 894
    },
    {
      "epoch": 1.2211981566820276,
      "grad_norm": 0.10236304998397827,
      "learning_rate": 2.819371627748226e-05,
      "loss": 0.417,
      "num_tokens": 708615231.0,
      "step": 895
    },
    {
      "epoch": 1.2225635774022872,
      "grad_norm": 0.11349616199731827,
      "learning_rate": 2.816881716014441e-05,
      "loss": 0.4309,
      "num_tokens": 709382455.0,
      "step": 896
    },
    {
      "epoch": 1.2239289981225465,
      "grad_norm": 0.10458794981241226,
      "learning_rate": 2.814390466084264e-05,
      "loss": 0.4358,
      "num_tokens": 710214956.0,
      "step": 897
    },
    {
      "epoch": 1.2252944188428059,
      "grad_norm": 0.11855491250753403,
      "learning_rate": 2.8118978833619458e-05,
      "loss": 0.4338,
      "num_tokens": 710936541.0,
      "step": 898
    },
    {
      "epoch": 1.2266598395630655,
      "grad_norm": 0.10450224578380585,
      "learning_rate": 2.8094039732546225e-05,
      "loss": 0.4416,
      "num_tokens": 711824581.0,
      "step": 899
    },
    {
      "epoch": 1.2280252602833248,
      "grad_norm": 0.11006946116685867,
      "learning_rate": 2.806908741172314e-05,
      "loss": 0.4093,
      "num_tokens": 712627527.0,
      "step": 900
    },
    {
      "epoch": 1.2293906810035842,
      "grad_norm": 0.10745418816804886,
      "learning_rate": 2.8044121925279068e-05,
      "loss": 0.433,
      "num_tokens": 713373680.0,
      "step": 901
    },
    {
      "epoch": 1.2307561017238438,
      "grad_norm": 0.12005268782377243,
      "learning_rate": 2.801914332737143e-05,
      "loss": 0.4305,
      "num_tokens": 714163581.0,
      "step": 902
    },
    {
      "epoch": 1.2321215224441031,
      "grad_norm": 0.10837270319461823,
      "learning_rate": 2.799415167218608e-05,
      "loss": 0.4369,
      "num_tokens": 714902320.0,
      "step": 903
    },
    {
      "epoch": 1.2334869431643625,
      "grad_norm": 0.10878144949674606,
      "learning_rate": 2.7969147013937227e-05,
      "loss": 0.4539,
      "num_tokens": 715689161.0,
      "step": 904
    },
    {
      "epoch": 1.234852363884622,
      "grad_norm": 0.12281865626573563,
      "learning_rate": 2.794412940686725e-05,
      "loss": 0.4434,
      "num_tokens": 716469432.0,
      "step": 905
    },
    {
      "epoch": 1.2362177846048814,
      "grad_norm": 0.11455006897449493,
      "learning_rate": 2.7919098905246645e-05,
      "loss": 0.4472,
      "num_tokens": 717283736.0,
      "step": 906
    },
    {
      "epoch": 1.2375832053251408,
      "grad_norm": 0.10644208639860153,
      "learning_rate": 2.7894055563373883e-05,
      "loss": 0.4312,
      "num_tokens": 717997368.0,
      "step": 907
    },
    {
      "epoch": 1.2389486260454001,
      "grad_norm": 0.11796961724758148,
      "learning_rate": 2.7868999435575256e-05,
      "loss": 0.4286,
      "num_tokens": 718901181.0,
      "step": 908
    },
    {
      "epoch": 1.2403140467656597,
      "grad_norm": 0.10295096784830093,
      "learning_rate": 2.7843930576204833e-05,
      "loss": 0.4436,
      "num_tokens": 719746864.0,
      "step": 909
    },
    {
      "epoch": 1.241679467485919,
      "grad_norm": 0.10588818788528442,
      "learning_rate": 2.7818849039644272e-05,
      "loss": 0.4191,
      "num_tokens": 720532690.0,
      "step": 910
    },
    {
      "epoch": 1.2430448882061786,
      "grad_norm": 0.11232714354991913,
      "learning_rate": 2.7793754880302747e-05,
      "loss": 0.43,
      "num_tokens": 721320795.0,
      "step": 911
    },
    {
      "epoch": 1.244410308926438,
      "grad_norm": 0.106940858066082,
      "learning_rate": 2.7768648152616802e-05,
      "loss": 0.4268,
      "num_tokens": 722158114.0,
      "step": 912
    },
    {
      "epoch": 1.2457757296466974,
      "grad_norm": 0.10274654626846313,
      "learning_rate": 2.7743528911050267e-05,
      "loss": 0.4214,
      "num_tokens": 722988391.0,
      "step": 913
    },
    {
      "epoch": 1.2471411503669567,
      "grad_norm": 0.11170939356088638,
      "learning_rate": 2.7718397210094092e-05,
      "loss": 0.4145,
      "num_tokens": 723752755.0,
      "step": 914
    },
    {
      "epoch": 1.2485065710872163,
      "grad_norm": 0.1023930087685585,
      "learning_rate": 2.7693253104266276e-05,
      "loss": 0.417,
      "num_tokens": 724534617.0,
      "step": 915
    },
    {
      "epoch": 1.2498719918074757,
      "grad_norm": 0.10137946158647537,
      "learning_rate": 2.7668096648111714e-05,
      "loss": 0.4037,
      "num_tokens": 725269594.0,
      "step": 916
    },
    {
      "epoch": 1.2512374125277352,
      "grad_norm": 0.11302996426820755,
      "learning_rate": 2.764292789620209e-05,
      "loss": 0.4058,
      "num_tokens": 726008186.0,
      "step": 917
    },
    {
      "epoch": 1.2526028332479946,
      "grad_norm": 0.09767593443393707,
      "learning_rate": 2.761774690313579e-05,
      "loss": 0.4515,
      "num_tokens": 726869901.0,
      "step": 918
    },
    {
      "epoch": 1.253968253968254,
      "grad_norm": 0.12046097964048386,
      "learning_rate": 2.7592553723537718e-05,
      "loss": 0.402,
      "num_tokens": 727611984.0,
      "step": 919
    },
    {
      "epoch": 1.2553336746885133,
      "grad_norm": 0.10207499563694,
      "learning_rate": 2.7567348412059226e-05,
      "loss": 0.4094,
      "num_tokens": 728398784.0,
      "step": 920
    },
    {
      "epoch": 1.256699095408773,
      "grad_norm": 0.1163308173418045,
      "learning_rate": 2.7542131023378004e-05,
      "loss": 0.4165,
      "num_tokens": 729218165.0,
      "step": 921
    },
    {
      "epoch": 1.2580645161290323,
      "grad_norm": 0.11352997273206711,
      "learning_rate": 2.751690161219791e-05,
      "loss": 0.4343,
      "num_tokens": 730044037.0,
      "step": 922
    },
    {
      "epoch": 1.2594299368492916,
      "grad_norm": 0.10292612016201019,
      "learning_rate": 2.74916602332489e-05,
      "loss": 0.4367,
      "num_tokens": 730836404.0,
      "step": 923
    },
    {
      "epoch": 1.2607953575695512,
      "grad_norm": 0.10747306793928146,
      "learning_rate": 2.7466406941286886e-05,
      "loss": 0.4401,
      "num_tokens": 731671400.0,
      "step": 924
    },
    {
      "epoch": 1.2621607782898105,
      "grad_norm": 0.1126956194639206,
      "learning_rate": 2.7441141791093627e-05,
      "loss": 0.4467,
      "num_tokens": 732455095.0,
      "step": 925
    },
    {
      "epoch": 1.26352619901007,
      "grad_norm": 0.11229176819324493,
      "learning_rate": 2.7415864837476614e-05,
      "loss": 0.414,
      "num_tokens": 733130498.0,
      "step": 926
    },
    {
      "epoch": 1.2648916197303295,
      "grad_norm": 0.10936608165502548,
      "learning_rate": 2.7390576135268915e-05,
      "loss": 0.438,
      "num_tokens": 733961193.0,
      "step": 927
    },
    {
      "epoch": 1.2662570404505888,
      "grad_norm": 0.1064135730266571,
      "learning_rate": 2.7365275739329112e-05,
      "loss": 0.4131,
      "num_tokens": 734717776.0,
      "step": 928
    },
    {
      "epoch": 1.2676224611708482,
      "grad_norm": 0.10358226299285889,
      "learning_rate": 2.7339963704541143e-05,
      "loss": 0.4413,
      "num_tokens": 735571959.0,
      "step": 929
    },
    {
      "epoch": 1.2689878818911078,
      "grad_norm": 0.10645900666713715,
      "learning_rate": 2.7314640085814196e-05,
      "loss": 0.4289,
      "num_tokens": 736403603.0,
      "step": 930
    },
    {
      "epoch": 1.2703533026113671,
      "grad_norm": 0.10581892728805542,
      "learning_rate": 2.7289304938082583e-05,
      "loss": 0.4167,
      "num_tokens": 737150162.0,
      "step": 931
    },
    {
      "epoch": 1.2717187233316265,
      "grad_norm": 0.10588841140270233,
      "learning_rate": 2.7263958316305627e-05,
      "loss": 0.4411,
      "num_tokens": 737898062.0,
      "step": 932
    },
    {
      "epoch": 1.2730841440518859,
      "grad_norm": 0.10500168055295944,
      "learning_rate": 2.723860027546756e-05,
      "loss": 0.4028,
      "num_tokens": 738680801.0,
      "step": 933
    },
    {
      "epoch": 1.2744495647721454,
      "grad_norm": 0.11681793630123138,
      "learning_rate": 2.721323087057735e-05,
      "loss": 0.4245,
      "num_tokens": 739488521.0,
      "step": 934
    },
    {
      "epoch": 1.2758149854924048,
      "grad_norm": 0.10773719847202301,
      "learning_rate": 2.7187850156668648e-05,
      "loss": 0.448,
      "num_tokens": 740340395.0,
      "step": 935
    },
    {
      "epoch": 1.2771804062126644,
      "grad_norm": 0.11045579612255096,
      "learning_rate": 2.716245818879963e-05,
      "loss": 0.425,
      "num_tokens": 741163522.0,
      "step": 936
    },
    {
      "epoch": 1.2785458269329237,
      "grad_norm": 0.11172912269830704,
      "learning_rate": 2.713705502205287e-05,
      "loss": 0.4375,
      "num_tokens": 741924879.0,
      "step": 937
    },
    {
      "epoch": 1.279911247653183,
      "grad_norm": 0.1052454262971878,
      "learning_rate": 2.711164071153525e-05,
      "loss": 0.4239,
      "num_tokens": 742750165.0,
      "step": 938
    },
    {
      "epoch": 1.2812766683734425,
      "grad_norm": 0.11461296677589417,
      "learning_rate": 2.7086215312377836e-05,
      "loss": 0.4285,
      "num_tokens": 743546062.0,
      "step": 939
    },
    {
      "epoch": 1.282642089093702,
      "grad_norm": 0.09977599233388901,
      "learning_rate": 2.7060778879735714e-05,
      "loss": 0.4214,
      "num_tokens": 744321077.0,
      "step": 940
    },
    {
      "epoch": 1.2840075098139614,
      "grad_norm": 0.11312580108642578,
      "learning_rate": 2.7035331468787945e-05,
      "loss": 0.4526,
      "num_tokens": 745081867.0,
      "step": 941
    },
    {
      "epoch": 1.285372930534221,
      "grad_norm": 0.11729885637760162,
      "learning_rate": 2.7009873134737375e-05,
      "loss": 0.4332,
      "num_tokens": 745967640.0,
      "step": 942
    },
    {
      "epoch": 1.2867383512544803,
      "grad_norm": 0.11115650832653046,
      "learning_rate": 2.698440393281056e-05,
      "loss": 0.4018,
      "num_tokens": 746740620.0,
      "step": 943
    },
    {
      "epoch": 1.2881037719747397,
      "grad_norm": 0.12337054312229156,
      "learning_rate": 2.6958923918257634e-05,
      "loss": 0.4376,
      "num_tokens": 747507874.0,
      "step": 944
    },
    {
      "epoch": 1.289469192694999,
      "grad_norm": 0.10796622931957245,
      "learning_rate": 2.693343314635217e-05,
      "loss": 0.4145,
      "num_tokens": 748287145.0,
      "step": 945
    },
    {
      "epoch": 1.2908346134152586,
      "grad_norm": 0.11556454747915268,
      "learning_rate": 2.6907931672391106e-05,
      "loss": 0.4285,
      "num_tokens": 749079098.0,
      "step": 946
    },
    {
      "epoch": 1.292200034135518,
      "grad_norm": 0.10363607853651047,
      "learning_rate": 2.688241955169456e-05,
      "loss": 0.4133,
      "num_tokens": 749815874.0,
      "step": 947
    },
    {
      "epoch": 1.2935654548557776,
      "grad_norm": 0.10397264361381531,
      "learning_rate": 2.6856896839605772e-05,
      "loss": 0.4613,
      "num_tokens": 750661799.0,
      "step": 948
    },
    {
      "epoch": 1.294930875576037,
      "grad_norm": 0.11002306640148163,
      "learning_rate": 2.6831363591490945e-05,
      "loss": 0.4312,
      "num_tokens": 751496654.0,
      "step": 949
    },
    {
      "epoch": 1.2962962962962963,
      "grad_norm": 0.10395850986242294,
      "learning_rate": 2.6805819862739146e-05,
      "loss": 0.4506,
      "num_tokens": 752329975.0,
      "step": 950
    },
    {
      "epoch": 1.2976617170165556,
      "grad_norm": 0.10316610336303711,
      "learning_rate": 2.6780265708762177e-05,
      "loss": 0.4369,
      "num_tokens": 753126327.0,
      "step": 951
    },
    {
      "epoch": 1.2990271377368152,
      "grad_norm": 0.10796497762203217,
      "learning_rate": 2.6754701184994445e-05,
      "loss": 0.4524,
      "num_tokens": 753979352.0,
      "step": 952
    },
    {
      "epoch": 1.3003925584570746,
      "grad_norm": 0.12164391577243805,
      "learning_rate": 2.6729126346892863e-05,
      "loss": 0.4495,
      "num_tokens": 754737562.0,
      "step": 953
    },
    {
      "epoch": 1.301757979177334,
      "grad_norm": 0.09945181757211685,
      "learning_rate": 2.670354124993672e-05,
      "loss": 0.4328,
      "num_tokens": 755479271.0,
      "step": 954
    },
    {
      "epoch": 1.3031233998975935,
      "grad_norm": 0.11143320798873901,
      "learning_rate": 2.6677945949627547e-05,
      "loss": 0.4423,
      "num_tokens": 756245955.0,
      "step": 955
    },
    {
      "epoch": 1.3044888206178529,
      "grad_norm": 0.11009370535612106,
      "learning_rate": 2.6652340501489026e-05,
      "loss": 0.4198,
      "num_tokens": 757100865.0,
      "step": 956
    },
    {
      "epoch": 1.3058542413381122,
      "grad_norm": 0.1033826693892479,
      "learning_rate": 2.662672496106683e-05,
      "loss": 0.4298,
      "num_tokens": 757863941.0,
      "step": 957
    },
    {
      "epoch": 1.3072196620583718,
      "grad_norm": 0.12719859182834625,
      "learning_rate": 2.6601099383928545e-05,
      "loss": 0.4292,
      "num_tokens": 758655530.0,
      "step": 958
    },
    {
      "epoch": 1.3085850827786312,
      "grad_norm": 0.0997733473777771,
      "learning_rate": 2.657546382566353e-05,
      "loss": 0.4493,
      "num_tokens": 759470723.0,
      "step": 959
    },
    {
      "epoch": 1.3099505034988905,
      "grad_norm": 0.12012312561273575,
      "learning_rate": 2.6549818341882776e-05,
      "loss": 0.4038,
      "num_tokens": 760164706.0,
      "step": 960
    },
    {
      "epoch": 1.31131592421915,
      "grad_norm": 0.10330194234848022,
      "learning_rate": 2.652416298821883e-05,
      "loss": 0.4425,
      "num_tokens": 760993351.0,
      "step": 961
    },
    {
      "epoch": 1.3126813449394095,
      "grad_norm": 0.10536931455135345,
      "learning_rate": 2.6498497820325625e-05,
      "loss": 0.4511,
      "num_tokens": 761830986.0,
      "step": 962
    },
    {
      "epoch": 1.3140467656596688,
      "grad_norm": 0.09635301679372787,
      "learning_rate": 2.6472822893878415e-05,
      "loss": 0.4422,
      "num_tokens": 762657356.0,
      "step": 963
    },
    {
      "epoch": 1.3154121863799282,
      "grad_norm": 0.11277693510055542,
      "learning_rate": 2.644713826457359e-05,
      "loss": 0.4195,
      "num_tokens": 763394392.0,
      "step": 964
    },
    {
      "epoch": 1.3167776071001878,
      "grad_norm": 0.10904648154973984,
      "learning_rate": 2.6421443988128607e-05,
      "loss": 0.4462,
      "num_tokens": 764218209.0,
      "step": 965
    },
    {
      "epoch": 1.3181430278204471,
      "grad_norm": 0.11771729588508606,
      "learning_rate": 2.639574012028185e-05,
      "loss": 0.4408,
      "num_tokens": 764963041.0,
      "step": 966
    },
    {
      "epoch": 1.3195084485407067,
      "grad_norm": 0.09998705238103867,
      "learning_rate": 2.6370026716792505e-05,
      "loss": 0.4327,
      "num_tokens": 765821849.0,
      "step": 967
    },
    {
      "epoch": 1.320873869260966,
      "grad_norm": 0.11514997482299805,
      "learning_rate": 2.6344303833440443e-05,
      "loss": 0.4385,
      "num_tokens": 766639170.0,
      "step": 968
    },
    {
      "epoch": 1.3222392899812254,
      "grad_norm": 0.11636712402105331,
      "learning_rate": 2.63185715260261e-05,
      "loss": 0.4574,
      "num_tokens": 767470588.0,
      "step": 969
    },
    {
      "epoch": 1.3236047107014848,
      "grad_norm": 0.10752368718385696,
      "learning_rate": 2.6292829850370355e-05,
      "loss": 0.4117,
      "num_tokens": 768196145.0,
      "step": 970
    },
    {
      "epoch": 1.3249701314217444,
      "grad_norm": 0.11279142647981644,
      "learning_rate": 2.626707886231443e-05,
      "loss": 0.4497,
      "num_tokens": 769017646.0,
      "step": 971
    },
    {
      "epoch": 1.3263355521420037,
      "grad_norm": 0.09887406975030899,
      "learning_rate": 2.6241318617719703e-05,
      "loss": 0.3945,
      "num_tokens": 769768116.0,
      "step": 972
    },
    {
      "epoch": 1.3277009728622633,
      "grad_norm": 0.10203395783901215,
      "learning_rate": 2.621554917246768e-05,
      "loss": 0.4217,
      "num_tokens": 770621828.0,
      "step": 973
    },
    {
      "epoch": 1.3290663935825227,
      "grad_norm": 0.10714080929756165,
      "learning_rate": 2.618977058245979e-05,
      "loss": 0.4336,
      "num_tokens": 771403348.0,
      "step": 974
    },
    {
      "epoch": 1.330431814302782,
      "grad_norm": 0.10714375972747803,
      "learning_rate": 2.616398290361733e-05,
      "loss": 0.4472,
      "num_tokens": 772176413.0,
      "step": 975
    },
    {
      "epoch": 1.3317972350230414,
      "grad_norm": 0.10936625301837921,
      "learning_rate": 2.613818619188129e-05,
      "loss": 0.4272,
      "num_tokens": 772963982.0,
      "step": 976
    },
    {
      "epoch": 1.333162655743301,
      "grad_norm": 0.1090153306722641,
      "learning_rate": 2.6112380503212255e-05,
      "loss": 0.4457,
      "num_tokens": 773820909.0,
      "step": 977
    },
    {
      "epoch": 1.3345280764635603,
      "grad_norm": 0.11082867532968521,
      "learning_rate": 2.6086565893590302e-05,
      "loss": 0.4109,
      "num_tokens": 774613420.0,
      "step": 978
    },
    {
      "epoch": 1.33589349718382,
      "grad_norm": 0.10827858000993729,
      "learning_rate": 2.606074241901484e-05,
      "loss": 0.4359,
      "num_tokens": 775458462.0,
      "step": 979
    },
    {
      "epoch": 1.3372589179040792,
      "grad_norm": 0.10969540476799011,
      "learning_rate": 2.6034910135504534e-05,
      "loss": 0.4135,
      "num_tokens": 776185415.0,
      "step": 980
    },
    {
      "epoch": 1.3386243386243386,
      "grad_norm": 0.09657879918813705,
      "learning_rate": 2.6009069099097125e-05,
      "loss": 0.4658,
      "num_tokens": 777041110.0,
      "step": 981
    },
    {
      "epoch": 1.339989759344598,
      "grad_norm": 0.11153651773929596,
      "learning_rate": 2.5983219365849367e-05,
      "loss": 0.426,
      "num_tokens": 777862793.0,
      "step": 982
    },
    {
      "epoch": 1.3413551800648575,
      "grad_norm": 0.10435855388641357,
      "learning_rate": 2.595736099183688e-05,
      "loss": 0.4147,
      "num_tokens": 778615797.0,
      "step": 983
    },
    {
      "epoch": 1.342720600785117,
      "grad_norm": 0.10750438272953033,
      "learning_rate": 2.5931494033154002e-05,
      "loss": 0.4287,
      "num_tokens": 779370721.0,
      "step": 984
    },
    {
      "epoch": 1.3440860215053765,
      "grad_norm": 0.10663682967424393,
      "learning_rate": 2.5905618545913728e-05,
      "loss": 0.4208,
      "num_tokens": 780153197.0,
      "step": 985
    },
    {
      "epoch": 1.3454514422256358,
      "grad_norm": 0.10064613074064255,
      "learning_rate": 2.5879734586247534e-05,
      "loss": 0.4222,
      "num_tokens": 780904279.0,
      "step": 986
    },
    {
      "epoch": 1.3468168629458952,
      "grad_norm": 0.11982409656047821,
      "learning_rate": 2.5853842210305268e-05,
      "loss": 0.4471,
      "num_tokens": 781684419.0,
      "step": 987
    },
    {
      "epoch": 1.3481822836661546,
      "grad_norm": 0.10777048766613007,
      "learning_rate": 2.582794147425506e-05,
      "loss": 0.4496,
      "num_tokens": 782601802.0,
      "step": 988
    },
    {
      "epoch": 1.3495477043864141,
      "grad_norm": 0.10021175444126129,
      "learning_rate": 2.5802032434283156e-05,
      "loss": 0.3977,
      "num_tokens": 783290883.0,
      "step": 989
    },
    {
      "epoch": 1.3509131251066735,
      "grad_norm": 0.11055293679237366,
      "learning_rate": 2.5776115146593827e-05,
      "loss": 0.4369,
      "num_tokens": 784140587.0,
      "step": 990
    },
    {
      "epoch": 1.3522785458269329,
      "grad_norm": 0.10336604714393616,
      "learning_rate": 2.5750189667409222e-05,
      "loss": 0.424,
      "num_tokens": 784926011.0,
      "step": 991
    },
    {
      "epoch": 1.3536439665471924,
      "grad_norm": 0.12045895308256149,
      "learning_rate": 2.572425605296928e-05,
      "loss": 0.4137,
      "num_tokens": 785678668.0,
      "step": 992
    },
    {
      "epoch": 1.3550093872674518,
      "grad_norm": 0.10642347484827042,
      "learning_rate": 2.5698314359531557e-05,
      "loss": 0.4257,
      "num_tokens": 786441586.0,
      "step": 993
    },
    {
      "epoch": 1.3563748079877112,
      "grad_norm": 0.11747383326292038,
      "learning_rate": 2.567236464337117e-05,
      "loss": 0.4215,
      "num_tokens": 787165696.0,
      "step": 994
    },
    {
      "epoch": 1.3577402287079705,
      "grad_norm": 0.1291494369506836,
      "learning_rate": 2.5646406960780614e-05,
      "loss": 0.4289,
      "num_tokens": 788005743.0,
      "step": 995
    },
    {
      "epoch": 1.35910564942823,
      "grad_norm": 0.11023978143930435,
      "learning_rate": 2.5620441368069675e-05,
      "loss": 0.4137,
      "num_tokens": 788767131.0,
      "step": 996
    },
    {
      "epoch": 1.3604710701484894,
      "grad_norm": 0.11628345400094986,
      "learning_rate": 2.5594467921565308e-05,
      "loss": 0.43,
      "num_tokens": 789535844.0,
      "step": 997
    },
    {
      "epoch": 1.361836490868749,
      "grad_norm": 0.11895844340324402,
      "learning_rate": 2.5568486677611487e-05,
      "loss": 0.436,
      "num_tokens": 790268727.0,
      "step": 998
    },
    {
      "epoch": 1.3632019115890084,
      "grad_norm": 0.11845717579126358,
      "learning_rate": 2.5542497692569103e-05,
      "loss": 0.4245,
      "num_tokens": 791069579.0,
      "step": 999
    },
    {
      "epoch": 1.3645673323092677,
      "grad_norm": 0.11205568164587021,
      "learning_rate": 2.5516501022815857e-05,
      "loss": 0.4456,
      "num_tokens": 791937177.0,
      "step": 1000
    },
    {
      "epoch": 1.365932753029527,
      "grad_norm": 0.10688168555498123,
      "learning_rate": 2.5490496724746104e-05,
      "loss": 0.4432,
      "num_tokens": 792738424.0,
      "step": 1001
    },
    {
      "epoch": 1.3672981737497867,
      "grad_norm": 0.11356765776872635,
      "learning_rate": 2.5464484854770755e-05,
      "loss": 0.4062,
      "num_tokens": 793499030.0,
      "step": 1002
    },
    {
      "epoch": 1.368663594470046,
      "grad_norm": 0.10751923173666,
      "learning_rate": 2.5438465469317142e-05,
      "loss": 0.4365,
      "num_tokens": 794307172.0,
      "step": 1003
    },
    {
      "epoch": 1.3700290151903056,
      "grad_norm": 0.11118900775909424,
      "learning_rate": 2.54124386248289e-05,
      "loss": 0.4399,
      "num_tokens": 795173341.0,
      "step": 1004
    },
    {
      "epoch": 1.371394435910565,
      "grad_norm": 0.09990012645721436,
      "learning_rate": 2.538640437776585e-05,
      "loss": 0.4299,
      "num_tokens": 795916087.0,
      "step": 1005
    },
    {
      "epoch": 1.3727598566308243,
      "grad_norm": 0.09950078278779984,
      "learning_rate": 2.5360362784603866e-05,
      "loss": 0.4315,
      "num_tokens": 796716961.0,
      "step": 1006
    },
    {
      "epoch": 1.3741252773510837,
      "grad_norm": 0.11290010809898376,
      "learning_rate": 2.533431390183476e-05,
      "loss": 0.4362,
      "num_tokens": 797494957.0,
      "step": 1007
    },
    {
      "epoch": 1.3754906980713433,
      "grad_norm": 0.1041722372174263,
      "learning_rate": 2.530825778596616e-05,
      "loss": 0.4324,
      "num_tokens": 798310851.0,
      "step": 1008
    },
    {
      "epoch": 1.3768561187916026,
      "grad_norm": 0.10744116455316544,
      "learning_rate": 2.528219449352138e-05,
      "loss": 0.4254,
      "num_tokens": 799054528.0,
      "step": 1009
    },
    {
      "epoch": 1.3782215395118622,
      "grad_norm": 0.09962642937898636,
      "learning_rate": 2.5256124081039307e-05,
      "loss": 0.426,
      "num_tokens": 799887367.0,
      "step": 1010
    },
    {
      "epoch": 1.3795869602321216,
      "grad_norm": 0.1215762346982956,
      "learning_rate": 2.5230046605074262e-05,
      "loss": 0.4426,
      "num_tokens": 800762199.0,
      "step": 1011
    },
    {
      "epoch": 1.380952380952381,
      "grad_norm": 0.11345130205154419,
      "learning_rate": 2.5203962122195915e-05,
      "loss": 0.4559,
      "num_tokens": 801571455.0,
      "step": 1012
    },
    {
      "epoch": 1.3823178016726403,
      "grad_norm": 0.10933373123407364,
      "learning_rate": 2.517787068898909e-05,
      "loss": 0.4373,
      "num_tokens": 802428953.0,
      "step": 1013
    },
    {
      "epoch": 1.3836832223928999,
      "grad_norm": 0.1384706199169159,
      "learning_rate": 2.5151772362053742e-05,
      "loss": 0.4293,
      "num_tokens": 803220231.0,
      "step": 1014
    },
    {
      "epoch": 1.3850486431131592,
      "grad_norm": 0.10726287215948105,
      "learning_rate": 2.512566719800475e-05,
      "loss": 0.4269,
      "num_tokens": 804011086.0,
      "step": 1015
    },
    {
      "epoch": 1.3864140638334188,
      "grad_norm": 0.10914111137390137,
      "learning_rate": 2.5099555253471824e-05,
      "loss": 0.426,
      "num_tokens": 804763110.0,
      "step": 1016
    },
    {
      "epoch": 1.3877794845536782,
      "grad_norm": 0.10863643884658813,
      "learning_rate": 2.5073436585099384e-05,
      "loss": 0.4314,
      "num_tokens": 805564626.0,
      "step": 1017
    },
    {
      "epoch": 1.3891449052739375,
      "grad_norm": 0.10587038099765778,
      "learning_rate": 2.5047311249546464e-05,
      "loss": 0.4118,
      "num_tokens": 806330230.0,
      "step": 1018
    },
    {
      "epoch": 1.3905103259941969,
      "grad_norm": 0.1147788017988205,
      "learning_rate": 2.502117930348651e-05,
      "loss": 0.4274,
      "num_tokens": 807151952.0,
      "step": 1019
    },
    {
      "epoch": 1.3918757467144565,
      "grad_norm": 0.10346291959285736,
      "learning_rate": 2.499504080360735e-05,
      "loss": 0.4291,
      "num_tokens": 807990493.0,
      "step": 1020
    },
    {
      "epoch": 1.3932411674347158,
      "grad_norm": 0.10195817053318024,
      "learning_rate": 2.4968895806611014e-05,
      "loss": 0.4285,
      "num_tokens": 808822125.0,
      "step": 1021
    },
    {
      "epoch": 1.3946065881549752,
      "grad_norm": 0.10600725561380386,
      "learning_rate": 2.4942744369213626e-05,
      "loss": 0.4225,
      "num_tokens": 809561550.0,
      "step": 1022
    },
    {
      "epoch": 1.3959720088752348,
      "grad_norm": 0.1048147976398468,
      "learning_rate": 2.4916586548145286e-05,
      "loss": 0.4219,
      "num_tokens": 810378731.0,
      "step": 1023
    },
    {
      "epoch": 1.3973374295954941,
      "grad_norm": 0.1058591827750206,
      "learning_rate": 2.489042240014993e-05,
      "loss": 0.4363,
      "num_tokens": 811231037.0,
      "step": 1024
    },
    {
      "epoch": 1.3987028503157535,
      "grad_norm": 0.10640069097280502,
      "learning_rate": 2.4864251981985233e-05,
      "loss": 0.4315,
      "num_tokens": 812050064.0,
      "step": 1025
    },
    {
      "epoch": 1.4000682710360128,
      "grad_norm": 0.11103836447000504,
      "learning_rate": 2.483807535042246e-05,
      "loss": 0.4666,
      "num_tokens": 812916074.0,
      "step": 1026
    },
    {
      "epoch": 1.4014336917562724,
      "grad_norm": 0.10110757499933243,
      "learning_rate": 2.4811892562246373e-05,
      "loss": 0.42,
      "num_tokens": 813764303.0,
      "step": 1027
    },
    {
      "epoch": 1.4027991124765318,
      "grad_norm": 0.1108081266283989,
      "learning_rate": 2.4785703674255062e-05,
      "loss": 0.4209,
      "num_tokens": 814490556.0,
      "step": 1028
    },
    {
      "epoch": 1.4041645331967914,
      "grad_norm": 0.12750664353370667,
      "learning_rate": 2.475950874325988e-05,
      "loss": 0.436,
      "num_tokens": 815366820.0,
      "step": 1029
    },
    {
      "epoch": 1.4055299539170507,
      "grad_norm": 0.09626328945159912,
      "learning_rate": 2.4733307826085265e-05,
      "loss": 0.4157,
      "num_tokens": 816195802.0,
      "step": 1030
    },
    {
      "epoch": 1.40689537463731,
      "grad_norm": 0.11903474479913712,
      "learning_rate": 2.4707100979568646e-05,
      "loss": 0.4371,
      "num_tokens": 816983376.0,
      "step": 1031
    },
    {
      "epoch": 1.4082607953575694,
      "grad_norm": 0.1088799461722374,
      "learning_rate": 2.468088826056032e-05,
      "loss": 0.4416,
      "num_tokens": 817813401.0,
      "step": 1032
    },
    {
      "epoch": 1.409626216077829,
      "grad_norm": 0.10537111014127731,
      "learning_rate": 2.465466972592333e-05,
      "loss": 0.4192,
      "num_tokens": 818645727.0,
      "step": 1033
    },
    {
      "epoch": 1.4109916367980884,
      "grad_norm": 0.11401479691267014,
      "learning_rate": 2.4628445432533315e-05,
      "loss": 0.4165,
      "num_tokens": 819384258.0,
      "step": 1034
    },
    {
      "epoch": 1.412357057518348,
      "grad_norm": 0.12183521687984467,
      "learning_rate": 2.4602215437278422e-05,
      "loss": 0.4486,
      "num_tokens": 820154980.0,
      "step": 1035
    },
    {
      "epoch": 1.4137224782386073,
      "grad_norm": 0.11336731165647507,
      "learning_rate": 2.4575979797059165e-05,
      "loss": 0.4302,
      "num_tokens": 820885309.0,
      "step": 1036
    },
    {
      "epoch": 1.4150878989588667,
      "grad_norm": 0.10727894306182861,
      "learning_rate": 2.4549738568788293e-05,
      "loss": 0.4164,
      "num_tokens": 821614200.0,
      "step": 1037
    },
    {
      "epoch": 1.416453319679126,
      "grad_norm": 0.10895020514726639,
      "learning_rate": 2.4523491809390686e-05,
      "loss": 0.4326,
      "num_tokens": 822436539.0,
      "step": 1038
    },
    {
      "epoch": 1.4178187403993856,
      "grad_norm": 0.13270831108093262,
      "learning_rate": 2.4497239575803233e-05,
      "loss": 0.4343,
      "num_tokens": 823179229.0,
      "step": 1039
    },
    {
      "epoch": 1.419184161119645,
      "grad_norm": 0.11718117445707321,
      "learning_rate": 2.447098192497467e-05,
      "loss": 0.4252,
      "num_tokens": 823894968.0,
      "step": 1040
    },
    {
      "epoch": 1.4205495818399045,
      "grad_norm": 0.11471572518348694,
      "learning_rate": 2.4444718913865506e-05,
      "loss": 0.4492,
      "num_tokens": 824610375.0,
      "step": 1041
    },
    {
      "epoch": 1.421915002560164,
      "grad_norm": 0.09908491373062134,
      "learning_rate": 2.441845059944789e-05,
      "loss": 0.4271,
      "num_tokens": 825447947.0,
      "step": 1042
    },
    {
      "epoch": 1.4232804232804233,
      "grad_norm": 0.11874699592590332,
      "learning_rate": 2.4392177038705432e-05,
      "loss": 0.4361,
      "num_tokens": 826245089.0,
      "step": 1043
    },
    {
      "epoch": 1.4246458440006826,
      "grad_norm": 0.10499081760644913,
      "learning_rate": 2.436589828863317e-05,
      "loss": 0.423,
      "num_tokens": 827058741.0,
      "step": 1044
    },
    {
      "epoch": 1.4260112647209422,
      "grad_norm": 0.10959479212760925,
      "learning_rate": 2.433961440623737e-05,
      "loss": 0.4228,
      "num_tokens": 827873766.0,
      "step": 1045
    },
    {
      "epoch": 1.4273766854412016,
      "grad_norm": 0.11674372106790543,
      "learning_rate": 2.431332544853544e-05,
      "loss": 0.409,
      "num_tokens": 828704087.0,
      "step": 1046
    },
    {
      "epoch": 1.4287421061614611,
      "grad_norm": 0.10263203829526901,
      "learning_rate": 2.4287031472555796e-05,
      "loss": 0.4311,
      "num_tokens": 829533650.0,
      "step": 1047
    },
    {
      "epoch": 1.4301075268817205,
      "grad_norm": 0.1011975109577179,
      "learning_rate": 2.426073253533775e-05,
      "loss": 0.426,
      "num_tokens": 830354588.0,
      "step": 1048
    },
    {
      "epoch": 1.4314729476019799,
      "grad_norm": 0.11933902651071548,
      "learning_rate": 2.4234428693931355e-05,
      "loss": 0.442,
      "num_tokens": 831157835.0,
      "step": 1049
    },
    {
      "epoch": 1.4328383683222392,
      "grad_norm": 0.09962167590856552,
      "learning_rate": 2.420812000539734e-05,
      "loss": 0.4025,
      "num_tokens": 831929780.0,
      "step": 1050
    },
    {
      "epoch": 1.4342037890424988,
      "grad_norm": 0.11247876286506653,
      "learning_rate": 2.4181806526806893e-05,
      "loss": 0.429,
      "num_tokens": 832797366.0,
      "step": 1051
    },
    {
      "epoch": 1.4355692097627581,
      "grad_norm": 0.10259748995304108,
      "learning_rate": 2.415548831524165e-05,
      "loss": 0.4212,
      "num_tokens": 833716848.0,
      "step": 1052
    },
    {
      "epoch": 1.4369346304830175,
      "grad_norm": 0.10765533894300461,
      "learning_rate": 2.4129165427793475e-05,
      "loss": 0.436,
      "num_tokens": 834524447.0,
      "step": 1053
    },
    {
      "epoch": 1.438300051203277,
      "grad_norm": 0.10601310431957245,
      "learning_rate": 2.41028379215644e-05,
      "loss": 0.455,
      "num_tokens": 835303579.0,
      "step": 1054
    },
    {
      "epoch": 1.4396654719235364,
      "grad_norm": 0.09174465388059616,
      "learning_rate": 2.4076505853666458e-05,
      "loss": 0.4147,
      "num_tokens": 836053825.0,
      "step": 1055
    },
    {
      "epoch": 1.4410308926437958,
      "grad_norm": 0.11375413835048676,
      "learning_rate": 2.405016928122159e-05,
      "loss": 0.4193,
      "num_tokens": 836795144.0,
      "step": 1056
    },
    {
      "epoch": 1.4423963133640554,
      "grad_norm": 0.10399957001209259,
      "learning_rate": 2.40238282613615e-05,
      "loss": 0.4456,
      "num_tokens": 837573897.0,
      "step": 1057
    },
    {
      "epoch": 1.4437617340843147,
      "grad_norm": 0.11164519190788269,
      "learning_rate": 2.399748285122755e-05,
      "loss": 0.441,
      "num_tokens": 838425493.0,
      "step": 1058
    },
    {
      "epoch": 1.445127154804574,
      "grad_norm": 0.09152604639530182,
      "learning_rate": 2.3971133107970617e-05,
      "loss": 0.4021,
      "num_tokens": 839236719.0,
      "step": 1059
    },
    {
      "epoch": 1.4464925755248337,
      "grad_norm": 0.10764876008033752,
      "learning_rate": 2.3944779088750975e-05,
      "loss": 0.4243,
      "num_tokens": 840025819.0,
      "step": 1060
    },
    {
      "epoch": 1.447857996245093,
      "grad_norm": 0.11159835755825043,
      "learning_rate": 2.391842085073818e-05,
      "loss": 0.4398,
      "num_tokens": 840804892.0,
      "step": 1061
    },
    {
      "epoch": 1.4492234169653524,
      "grad_norm": 0.10140714049339294,
      "learning_rate": 2.389205845111095e-05,
      "loss": 0.4191,
      "num_tokens": 841637742.0,
      "step": 1062
    },
    {
      "epoch": 1.4505888376856118,
      "grad_norm": 0.11281415075063705,
      "learning_rate": 2.386569194705701e-05,
      "loss": 0.4154,
      "num_tokens": 842420275.0,
      "step": 1063
    },
    {
      "epoch": 1.4519542584058713,
      "grad_norm": 0.11447974294424057,
      "learning_rate": 2.383932139577301e-05,
      "loss": 0.4262,
      "num_tokens": 843195985.0,
      "step": 1064
    },
    {
      "epoch": 1.4533196791261307,
      "grad_norm": 0.10813387483358383,
      "learning_rate": 2.381294685446435e-05,
      "loss": 0.4285,
      "num_tokens": 843971035.0,
      "step": 1065
    },
    {
      "epoch": 1.4546850998463903,
      "grad_norm": 0.09846271574497223,
      "learning_rate": 2.378656838034511e-05,
      "loss": 0.4268,
      "num_tokens": 844761965.0,
      "step": 1066
    },
    {
      "epoch": 1.4560505205666496,
      "grad_norm": 0.09854397177696228,
      "learning_rate": 2.3760186030637905e-05,
      "loss": 0.4161,
      "num_tokens": 845538267.0,
      "step": 1067
    },
    {
      "epoch": 1.457415941286909,
      "grad_norm": 0.1177617534995079,
      "learning_rate": 2.3733799862573746e-05,
      "loss": 0.4479,
      "num_tokens": 846387110.0,
      "step": 1068
    },
    {
      "epoch": 1.4587813620071683,
      "grad_norm": 0.09591616690158844,
      "learning_rate": 2.3707409933391918e-05,
      "loss": 0.4259,
      "num_tokens": 847148130.0,
      "step": 1069
    },
    {
      "epoch": 1.460146782727428,
      "grad_norm": 0.11032189428806305,
      "learning_rate": 2.3681016300339875e-05,
      "loss": 0.4225,
      "num_tokens": 847919331.0,
      "step": 1070
    },
    {
      "epoch": 1.4615122034476873,
      "grad_norm": 0.10382021218538284,
      "learning_rate": 2.3654619020673124e-05,
      "loss": 0.4343,
      "num_tokens": 848697077.0,
      "step": 1071
    },
    {
      "epoch": 1.4628776241679469,
      "grad_norm": 0.10325926542282104,
      "learning_rate": 2.3628218151655043e-05,
      "loss": 0.3978,
      "num_tokens": 849444665.0,
      "step": 1072
    },
    {
      "epoch": 1.4642430448882062,
      "grad_norm": 0.10231555253267288,
      "learning_rate": 2.3601813750556834e-05,
      "loss": 0.4034,
      "num_tokens": 850227439.0,
      "step": 1073
    },
    {
      "epoch": 1.4656084656084656,
      "grad_norm": 0.10483427345752716,
      "learning_rate": 2.3575405874657343e-05,
      "loss": 0.4254,
      "num_tokens": 851079480.0,
      "step": 1074
    },
    {
      "epoch": 1.466973886328725,
      "grad_norm": 0.10400961339473724,
      "learning_rate": 2.3548994581242948e-05,
      "loss": 0.4637,
      "num_tokens": 851931761.0,
      "step": 1075
    },
    {
      "epoch": 1.4683393070489845,
      "grad_norm": 0.11214058846235275,
      "learning_rate": 2.352257992760745e-05,
      "loss": 0.4164,
      "num_tokens": 852702836.0,
      "step": 1076
    },
    {
      "epoch": 1.4697047277692439,
      "grad_norm": 0.10605745017528534,
      "learning_rate": 2.349616197105195e-05,
      "loss": 0.4515,
      "num_tokens": 853502013.0,
      "step": 1077
    },
    {
      "epoch": 1.4710701484895035,
      "grad_norm": 0.10598117858171463,
      "learning_rate": 2.346974076888469e-05,
      "loss": 0.4008,
      "num_tokens": 854227415.0,
      "step": 1078
    },
    {
      "epoch": 1.4724355692097628,
      "grad_norm": 0.11889506876468658,
      "learning_rate": 2.3443316378420975e-05,
      "loss": 0.4231,
      "num_tokens": 854974112.0,
      "step": 1079
    },
    {
      "epoch": 1.4738009899300222,
      "grad_norm": 0.10076722502708435,
      "learning_rate": 2.3416888856983003e-05,
      "loss": 0.4262,
      "num_tokens": 855738735.0,
      "step": 1080
    },
    {
      "epoch": 1.4751664106502815,
      "grad_norm": 0.10548829287290573,
      "learning_rate": 2.3390458261899783e-05,
      "loss": 0.4336,
      "num_tokens": 856538382.0,
      "step": 1081
    },
    {
      "epoch": 1.4765318313705411,
      "grad_norm": 0.11030023545026779,
      "learning_rate": 2.3364024650506986e-05,
      "loss": 0.4437,
      "num_tokens": 857328763.0,
      "step": 1082
    },
    {
      "epoch": 1.4778972520908005,
      "grad_norm": 0.10678166896104813,
      "learning_rate": 2.333758808014684e-05,
      "loss": 0.4074,
      "num_tokens": 858085107.0,
      "step": 1083
    },
    {
      "epoch": 1.4792626728110598,
      "grad_norm": 0.11350295692682266,
      "learning_rate": 2.3311148608167944e-05,
      "loss": 0.425,
      "num_tokens": 858880828.0,
      "step": 1084
    },
    {
      "epoch": 1.4806280935313194,
      "grad_norm": 0.11764386296272278,
      "learning_rate": 2.3284706291925257e-05,
      "loss": 0.4296,
      "num_tokens": 859693219.0,
      "step": 1085
    },
    {
      "epoch": 1.4819935142515788,
      "grad_norm": 0.10533306747674942,
      "learning_rate": 2.3258261188779863e-05,
      "loss": 0.4554,
      "num_tokens": 860535848.0,
      "step": 1086
    },
    {
      "epoch": 1.4833589349718381,
      "grad_norm": 0.11193782836198807,
      "learning_rate": 2.323181335609891e-05,
      "loss": 0.449,
      "num_tokens": 861328797.0,
      "step": 1087
    },
    {
      "epoch": 1.4847243556920977,
      "grad_norm": 0.09792253375053406,
      "learning_rate": 2.3205362851255453e-05,
      "loss": 0.4081,
      "num_tokens": 862032708.0,
      "step": 1088
    },
    {
      "epoch": 1.486089776412357,
      "grad_norm": 0.11309471726417542,
      "learning_rate": 2.3178909731628365e-05,
      "loss": 0.3933,
      "num_tokens": 862726194.0,
      "step": 1089
    },
    {
      "epoch": 1.4874551971326164,
      "grad_norm": 0.10348217189311981,
      "learning_rate": 2.3152454054602166e-05,
      "loss": 0.4365,
      "num_tokens": 863565182.0,
      "step": 1090
    },
    {
      "epoch": 1.488820617852876,
      "grad_norm": 0.10750546306371689,
      "learning_rate": 2.312599587756695e-05,
      "loss": 0.4354,
      "num_tokens": 864360781.0,
      "step": 1091
    },
    {
      "epoch": 1.4901860385731354,
      "grad_norm": 0.10602644830942154,
      "learning_rate": 2.3099535257918212e-05,
      "loss": 0.4298,
      "num_tokens": 865189088.0,
      "step": 1092
    },
    {
      "epoch": 1.4915514592933947,
      "grad_norm": 0.10414795577526093,
      "learning_rate": 2.307307225305676e-05,
      "loss": 0.4248,
      "num_tokens": 865982262.0,
      "step": 1093
    },
    {
      "epoch": 1.492916880013654,
      "grad_norm": 0.10582312941551208,
      "learning_rate": 2.304660692038857e-05,
      "loss": 0.4433,
      "num_tokens": 866800240.0,
      "step": 1094
    },
    {
      "epoch": 1.4942823007339137,
      "grad_norm": 0.10097209364175797,
      "learning_rate": 2.302013931732466e-05,
      "loss": 0.4632,
      "num_tokens": 867602247.0,
      "step": 1095
    },
    {
      "epoch": 1.495647721454173,
      "grad_norm": 0.10782213509082794,
      "learning_rate": 2.2993669501280995e-05,
      "loss": 0.4327,
      "num_tokens": 868454523.0,
      "step": 1096
    },
    {
      "epoch": 1.4970131421744326,
      "grad_norm": 0.09826157242059708,
      "learning_rate": 2.2967197529678323e-05,
      "loss": 0.4447,
      "num_tokens": 869281295.0,
      "step": 1097
    },
    {
      "epoch": 1.498378562894692,
      "grad_norm": 0.10459816455841064,
      "learning_rate": 2.2940723459942067e-05,
      "loss": 0.4319,
      "num_tokens": 870123860.0,
      "step": 1098
    },
    {
      "epoch": 1.4997439836149513,
      "grad_norm": 0.09983882308006287,
      "learning_rate": 2.2914247349502206e-05,
      "loss": 0.4298,
      "num_tokens": 870914316.0,
      "step": 1099
    },
    {
      "epoch": 1.5011094043352107,
      "grad_norm": 0.1023237481713295,
      "learning_rate": 2.2887769255793165e-05,
      "loss": 0.442,
      "num_tokens": 871795860.0,
      "step": 1100
    },
    {
      "epoch": 1.5024748250554703,
      "grad_norm": 0.10290241241455078,
      "learning_rate": 2.2861289236253627e-05,
      "loss": 0.4378,
      "num_tokens": 872593862.0,
      "step": 1101
    },
    {
      "epoch": 1.5038402457757296,
      "grad_norm": 0.10564558207988739,
      "learning_rate": 2.2834807348326497e-05,
      "loss": 0.414,
      "num_tokens": 873328198.0,
      "step": 1102
    },
    {
      "epoch": 1.5052056664959892,
      "grad_norm": 0.11021730303764343,
      "learning_rate": 2.2808323649458703e-05,
      "loss": 0.4281,
      "num_tokens": 874148499.0,
      "step": 1103
    },
    {
      "epoch": 1.5065710872162486,
      "grad_norm": 0.09020983427762985,
      "learning_rate": 2.278183819710113e-05,
      "loss": 0.4175,
      "num_tokens": 874923135.0,
      "step": 1104
    },
    {
      "epoch": 1.507936507936508,
      "grad_norm": 0.10958003997802734,
      "learning_rate": 2.2755351048708428e-05,
      "loss": 0.4209,
      "num_tokens": 875690940.0,
      "step": 1105
    },
    {
      "epoch": 1.5093019286567673,
      "grad_norm": 0.10897855460643768,
      "learning_rate": 2.2728862261738972e-05,
      "loss": 0.4302,
      "num_tokens": 876534127.0,
      "step": 1106
    },
    {
      "epoch": 1.5106673493770268,
      "grad_norm": 0.10475944727659225,
      "learning_rate": 2.2702371893654646e-05,
      "loss": 0.4456,
      "num_tokens": 877329693.0,
      "step": 1107
    },
    {
      "epoch": 1.5120327700972862,
      "grad_norm": 0.0993519127368927,
      "learning_rate": 2.2675880001920806e-05,
      "loss": 0.4136,
      "num_tokens": 878154771.0,
      "step": 1108
    },
    {
      "epoch": 1.5133981908175458,
      "grad_norm": 0.10209588706493378,
      "learning_rate": 2.264938664400608e-05,
      "loss": 0.4374,
      "num_tokens": 878890917.0,
      "step": 1109
    },
    {
      "epoch": 1.5147636115378051,
      "grad_norm": 0.11019439250230789,
      "learning_rate": 2.2622891877382285e-05,
      "loss": 0.4339,
      "num_tokens": 879621489.0,
      "step": 1110
    },
    {
      "epoch": 1.5161290322580645,
      "grad_norm": 0.1123914122581482,
      "learning_rate": 2.2596395759524316e-05,
      "loss": 0.4157,
      "num_tokens": 880343260.0,
      "step": 1111
    },
    {
      "epoch": 1.5174944529783239,
      "grad_norm": 0.10034352540969849,
      "learning_rate": 2.2569898347909967e-05,
      "loss": 0.4328,
      "num_tokens": 881186694.0,
      "step": 1112
    },
    {
      "epoch": 1.5188598736985832,
      "grad_norm": 0.10904011875391006,
      "learning_rate": 2.2543399700019858e-05,
      "loss": 0.431,
      "num_tokens": 881982089.0,
      "step": 1113
    },
    {
      "epoch": 1.5202252944188428,
      "grad_norm": 0.11889615654945374,
      "learning_rate": 2.2516899873337286e-05,
      "loss": 0.4241,
      "num_tokens": 882802813.0,
      "step": 1114
    },
    {
      "epoch": 1.5215907151391024,
      "grad_norm": 0.10656171292066574,
      "learning_rate": 2.2490398925348108e-05,
      "loss": 0.441,
      "num_tokens": 883575179.0,
      "step": 1115
    },
    {
      "epoch": 1.5229561358593617,
      "grad_norm": 0.11328525096178055,
      "learning_rate": 2.2463896913540605e-05,
      "loss": 0.432,
      "num_tokens": 884389083.0,
      "step": 1116
    },
    {
      "epoch": 1.524321556579621,
      "grad_norm": 0.10749693959951401,
      "learning_rate": 2.2437393895405377e-05,
      "loss": 0.4397,
      "num_tokens": 885191883.0,
      "step": 1117
    },
    {
      "epoch": 1.5256869772998805,
      "grad_norm": 0.12348347902297974,
      "learning_rate": 2.24108899284352e-05,
      "loss": 0.4298,
      "num_tokens": 885934055.0,
      "step": 1118
    },
    {
      "epoch": 1.5270523980201398,
      "grad_norm": 0.10909584909677505,
      "learning_rate": 2.2384385070124907e-05,
      "loss": 0.4348,
      "num_tokens": 886736653.0,
      "step": 1119
    },
    {
      "epoch": 1.5284178187403994,
      "grad_norm": 0.10454745590686798,
      "learning_rate": 2.2357879377971265e-05,
      "loss": 0.4027,
      "num_tokens": 887460928.0,
      "step": 1120
    },
    {
      "epoch": 1.529783239460659,
      "grad_norm": 0.1159176453948021,
      "learning_rate": 2.2331372909472863e-05,
      "loss": 0.4359,
      "num_tokens": 888271581.0,
      "step": 1121
    },
    {
      "epoch": 1.5311486601809183,
      "grad_norm": 0.10885307192802429,
      "learning_rate": 2.2304865722129948e-05,
      "loss": 0.4418,
      "num_tokens": 889134574.0,
      "step": 1122
    },
    {
      "epoch": 1.5325140809011777,
      "grad_norm": 0.11641599982976913,
      "learning_rate": 2.227835787344436e-05,
      "loss": 0.4364,
      "num_tokens": 889961661.0,
      "step": 1123
    },
    {
      "epoch": 1.533879501621437,
      "grad_norm": 0.11138612776994705,
      "learning_rate": 2.2251849420919342e-05,
      "loss": 0.4076,
      "num_tokens": 890727978.0,
      "step": 1124
    },
    {
      "epoch": 1.5352449223416964,
      "grad_norm": 0.11445411294698715,
      "learning_rate": 2.2225340422059463e-05,
      "loss": 0.4541,
      "num_tokens": 891530109.0,
      "step": 1125
    },
    {
      "epoch": 1.536610343061956,
      "grad_norm": 0.11296062916517258,
      "learning_rate": 2.2198830934370474e-05,
      "loss": 0.4119,
      "num_tokens": 892325031.0,
      "step": 1126
    },
    {
      "epoch": 1.5379757637822153,
      "grad_norm": 0.11561346054077148,
      "learning_rate": 2.2172321015359196e-05,
      "loss": 0.452,
      "num_tokens": 893220154.0,
      "step": 1127
    },
    {
      "epoch": 1.539341184502475,
      "grad_norm": 0.10182221978902817,
      "learning_rate": 2.214581072253336e-05,
      "loss": 0.4125,
      "num_tokens": 893942080.0,
      "step": 1128
    },
    {
      "epoch": 1.5407066052227343,
      "grad_norm": 0.10316316038370132,
      "learning_rate": 2.2119300113401527e-05,
      "loss": 0.4189,
      "num_tokens": 894735447.0,
      "step": 1129
    },
    {
      "epoch": 1.5420720259429936,
      "grad_norm": 0.1174209713935852,
      "learning_rate": 2.209278924547296e-05,
      "loss": 0.4417,
      "num_tokens": 895549785.0,
      "step": 1130
    },
    {
      "epoch": 1.543437446663253,
      "grad_norm": 0.10003872960805893,
      "learning_rate": 2.206627817625744e-05,
      "loss": 0.4147,
      "num_tokens": 896380901.0,
      "step": 1131
    },
    {
      "epoch": 1.5448028673835126,
      "grad_norm": 0.10912458598613739,
      "learning_rate": 2.2039766963265237e-05,
      "loss": 0.4383,
      "num_tokens": 897207503.0,
      "step": 1132
    },
    {
      "epoch": 1.546168288103772,
      "grad_norm": 0.10656042397022247,
      "learning_rate": 2.201325566400688e-05,
      "loss": 0.4341,
      "num_tokens": 897996296.0,
      "step": 1133
    },
    {
      "epoch": 1.5475337088240315,
      "grad_norm": 0.10212191939353943,
      "learning_rate": 2.1986744335993126e-05,
      "loss": 0.4239,
      "num_tokens": 898804208.0,
      "step": 1134
    },
    {
      "epoch": 1.5488991295442909,
      "grad_norm": 0.11562422662973404,
      "learning_rate": 2.1960233036734766e-05,
      "loss": 0.4402,
      "num_tokens": 899564990.0,
      "step": 1135
    },
    {
      "epoch": 1.5502645502645502,
      "grad_norm": 0.1037953719496727,
      "learning_rate": 2.1933721823742554e-05,
      "loss": 0.4614,
      "num_tokens": 900377004.0,
      "step": 1136
    },
    {
      "epoch": 1.5516299709848096,
      "grad_norm": 0.11243940889835358,
      "learning_rate": 2.190721075452705e-05,
      "loss": 0.4217,
      "num_tokens": 901116101.0,
      "step": 1137
    },
    {
      "epoch": 1.5529953917050692,
      "grad_norm": 0.10382035374641418,
      "learning_rate": 2.188069988659848e-05,
      "loss": 0.4577,
      "num_tokens": 901976746.0,
      "step": 1138
    },
    {
      "epoch": 1.5543608124253285,
      "grad_norm": 0.10912242531776428,
      "learning_rate": 2.185418927746665e-05,
      "loss": 0.4291,
      "num_tokens": 902769540.0,
      "step": 1139
    },
    {
      "epoch": 1.5557262331455881,
      "grad_norm": 0.0986318290233612,
      "learning_rate": 2.1827678984640816e-05,
      "loss": 0.4415,
      "num_tokens": 903610115.0,
      "step": 1140
    },
    {
      "epoch": 1.5570916538658475,
      "grad_norm": 0.10446079075336456,
      "learning_rate": 2.180116906562953e-05,
      "loss": 0.417,
      "num_tokens": 904481626.0,
      "step": 1141
    },
    {
      "epoch": 1.5584570745861068,
      "grad_norm": 0.10122212767601013,
      "learning_rate": 2.177465957794054e-05,
      "loss": 0.4478,
      "num_tokens": 905342129.0,
      "step": 1142
    },
    {
      "epoch": 1.5598224953063662,
      "grad_norm": 0.09834959357976913,
      "learning_rate": 2.1748150579080667e-05,
      "loss": 0.4269,
      "num_tokens": 906100800.0,
      "step": 1143
    },
    {
      "epoch": 1.5611879160266255,
      "grad_norm": 0.10471849888563156,
      "learning_rate": 2.1721642126555647e-05,
      "loss": 0.4106,
      "num_tokens": 906867834.0,
      "step": 1144
    },
    {
      "epoch": 1.5625533367468851,
      "grad_norm": 0.102863609790802,
      "learning_rate": 2.1695134277870054e-05,
      "loss": 0.4498,
      "num_tokens": 907606847.0,
      "step": 1145
    },
    {
      "epoch": 1.5639187574671447,
      "grad_norm": 0.11569855362176895,
      "learning_rate": 2.1668627090527143e-05,
      "loss": 0.453,
      "num_tokens": 908418404.0,
      "step": 1146
    },
    {
      "epoch": 1.565284178187404,
      "grad_norm": 0.10664054751396179,
      "learning_rate": 2.1642120622028737e-05,
      "loss": 0.4357,
      "num_tokens": 909256031.0,
      "step": 1147
    },
    {
      "epoch": 1.5666495989076634,
      "grad_norm": 0.104103684425354,
      "learning_rate": 2.16156149298751e-05,
      "loss": 0.4206,
      "num_tokens": 910012655.0,
      "step": 1148
    },
    {
      "epoch": 1.5680150196279228,
      "grad_norm": 0.09594161808490753,
      "learning_rate": 2.158911007156481e-05,
      "loss": 0.4114,
      "num_tokens": 910830084.0,
      "step": 1149
    },
    {
      "epoch": 1.5693804403481821,
      "grad_norm": 0.11522035300731659,
      "learning_rate": 2.1562606104594633e-05,
      "loss": 0.4412,
      "num_tokens": 911651424.0,
      "step": 1150
    },
    {
      "epoch": 1.5707458610684417,
      "grad_norm": 0.10276751965284348,
      "learning_rate": 2.15361030864594e-05,
      "loss": 0.4121,
      "num_tokens": 912334147.0,
      "step": 1151
    },
    {
      "epoch": 1.5721112817887013,
      "grad_norm": 0.10709182173013687,
      "learning_rate": 2.1509601074651898e-05,
      "loss": 0.4451,
      "num_tokens": 913148278.0,
      "step": 1152
    },
    {
      "epoch": 1.5734767025089607,
      "grad_norm": 0.09857240319252014,
      "learning_rate": 2.1483100126662716e-05,
      "loss": 0.4359,
      "num_tokens": 913952788.0,
      "step": 1153
    },
    {
      "epoch": 1.57484212322922,
      "grad_norm": 0.10830063372850418,
      "learning_rate": 2.1456600299980148e-05,
      "loss": 0.4303,
      "num_tokens": 914772360.0,
      "step": 1154
    },
    {
      "epoch": 1.5762075439494794,
      "grad_norm": 0.10211794823408127,
      "learning_rate": 2.1430101652090042e-05,
      "loss": 0.4304,
      "num_tokens": 915562794.0,
      "step": 1155
    },
    {
      "epoch": 1.5775729646697387,
      "grad_norm": 0.10725120455026627,
      "learning_rate": 2.140360424047569e-05,
      "loss": 0.4541,
      "num_tokens": 916296484.0,
      "step": 1156
    },
    {
      "epoch": 1.5789383853899983,
      "grad_norm": 0.10814740508794785,
      "learning_rate": 2.1377108122617717e-05,
      "loss": 0.4566,
      "num_tokens": 917129759.0,
      "step": 1157
    },
    {
      "epoch": 1.5803038061102577,
      "grad_norm": 0.10133656114339828,
      "learning_rate": 2.135061335599393e-05,
      "loss": 0.4342,
      "num_tokens": 917924737.0,
      "step": 1158
    },
    {
      "epoch": 1.5816692268305172,
      "grad_norm": 0.10980561375617981,
      "learning_rate": 2.1324119998079196e-05,
      "loss": 0.4622,
      "num_tokens": 918677239.0,
      "step": 1159
    },
    {
      "epoch": 1.5830346475507766,
      "grad_norm": 0.10683796554803848,
      "learning_rate": 2.1297628106345357e-05,
      "loss": 0.4432,
      "num_tokens": 919493236.0,
      "step": 1160
    },
    {
      "epoch": 1.584400068271036,
      "grad_norm": 0.10316112637519836,
      "learning_rate": 2.1271137738261033e-05,
      "loss": 0.4092,
      "num_tokens": 920237123.0,
      "step": 1161
    },
    {
      "epoch": 1.5857654889912953,
      "grad_norm": 0.10805463045835495,
      "learning_rate": 2.1244648951291578e-05,
      "loss": 0.4203,
      "num_tokens": 921026469.0,
      "step": 1162
    },
    {
      "epoch": 1.587130909711555,
      "grad_norm": 0.09288125485181808,
      "learning_rate": 2.121816180289888e-05,
      "loss": 0.4228,
      "num_tokens": 921871168.0,
      "step": 1163
    },
    {
      "epoch": 1.5884963304318143,
      "grad_norm": 0.10428325086832047,
      "learning_rate": 2.11916763505413e-05,
      "loss": 0.4206,
      "num_tokens": 922680274.0,
      "step": 1164
    },
    {
      "epoch": 1.5898617511520738,
      "grad_norm": 0.10803043842315674,
      "learning_rate": 2.116519265167351e-05,
      "loss": 0.4296,
      "num_tokens": 923455809.0,
      "step": 1165
    },
    {
      "epoch": 1.5912271718723332,
      "grad_norm": 0.1155373603105545,
      "learning_rate": 2.1138710763746385e-05,
      "loss": 0.4288,
      "num_tokens": 924237147.0,
      "step": 1166
    },
    {
      "epoch": 1.5925925925925926,
      "grad_norm": 0.10335394740104675,
      "learning_rate": 2.1112230744206847e-05,
      "loss": 0.4279,
      "num_tokens": 925142574.0,
      "step": 1167
    },
    {
      "epoch": 1.593958013312852,
      "grad_norm": 0.10030847787857056,
      "learning_rate": 2.1085752650497796e-05,
      "loss": 0.4251,
      "num_tokens": 925929169.0,
      "step": 1168
    },
    {
      "epoch": 1.5953234340331115,
      "grad_norm": 0.10802744328975677,
      "learning_rate": 2.105927654005794e-05,
      "loss": 0.4112,
      "num_tokens": 926694420.0,
      "step": 1169
    },
    {
      "epoch": 1.5966888547533709,
      "grad_norm": 0.10934530943632126,
      "learning_rate": 2.1032802470321686e-05,
      "loss": 0.4335,
      "num_tokens": 927479350.0,
      "step": 1170
    },
    {
      "epoch": 1.5980542754736304,
      "grad_norm": 0.10250721126794815,
      "learning_rate": 2.1006330498719008e-05,
      "loss": 0.4136,
      "num_tokens": 928242692.0,
      "step": 1171
    },
    {
      "epoch": 1.5994196961938898,
      "grad_norm": 0.09642814844846725,
      "learning_rate": 2.0979860682675344e-05,
      "loss": 0.4305,
      "num_tokens": 929044980.0,
      "step": 1172
    },
    {
      "epoch": 1.6007851169141492,
      "grad_norm": 0.10491377115249634,
      "learning_rate": 2.095339307961144e-05,
      "loss": 0.4343,
      "num_tokens": 929865577.0,
      "step": 1173
    },
    {
      "epoch": 1.6021505376344085,
      "grad_norm": 0.10596424341201782,
      "learning_rate": 2.0926927746943243e-05,
      "loss": 0.4332,
      "num_tokens": 930689945.0,
      "step": 1174
    },
    {
      "epoch": 1.6035159583546679,
      "grad_norm": 0.09632733464241028,
      "learning_rate": 2.0900464742081793e-05,
      "loss": 0.4257,
      "num_tokens": 931497109.0,
      "step": 1175
    },
    {
      "epoch": 1.6048813790749274,
      "grad_norm": 0.10278981178998947,
      "learning_rate": 2.087400412243305e-05,
      "loss": 0.4196,
      "num_tokens": 932269630.0,
      "step": 1176
    },
    {
      "epoch": 1.606246799795187,
      "grad_norm": 0.09672658145427704,
      "learning_rate": 2.0847545945397836e-05,
      "loss": 0.4242,
      "num_tokens": 933024387.0,
      "step": 1177
    },
    {
      "epoch": 1.6076122205154464,
      "grad_norm": 0.11659767478704453,
      "learning_rate": 2.0821090268371648e-05,
      "loss": 0.4405,
      "num_tokens": 933831869.0,
      "step": 1178
    },
    {
      "epoch": 1.6089776412357057,
      "grad_norm": 0.09981337189674377,
      "learning_rate": 2.0794637148744553e-05,
      "loss": 0.4242,
      "num_tokens": 934627778.0,
      "step": 1179
    },
    {
      "epoch": 1.610343061955965,
      "grad_norm": 0.10197538882493973,
      "learning_rate": 2.0768186643901094e-05,
      "loss": 0.4319,
      "num_tokens": 935360207.0,
      "step": 1180
    },
    {
      "epoch": 1.6117084826762245,
      "grad_norm": 0.09995868057012558,
      "learning_rate": 2.0741738811220142e-05,
      "loss": 0.4285,
      "num_tokens": 936090972.0,
      "step": 1181
    },
    {
      "epoch": 1.613073903396484,
      "grad_norm": 0.1071997880935669,
      "learning_rate": 2.0715293708074745e-05,
      "loss": 0.4304,
      "num_tokens": 936943765.0,
      "step": 1182
    },
    {
      "epoch": 1.6144393241167436,
      "grad_norm": 0.09958726912736893,
      "learning_rate": 2.0688851391832058e-05,
      "loss": 0.4241,
      "num_tokens": 937720450.0,
      "step": 1183
    },
    {
      "epoch": 1.615804744837003,
      "grad_norm": 0.0974750965833664,
      "learning_rate": 2.0662411919853177e-05,
      "loss": 0.4523,
      "num_tokens": 938528702.0,
      "step": 1184
    },
    {
      "epoch": 1.6171701655572623,
      "grad_norm": 0.10775598138570786,
      "learning_rate": 2.063597534949302e-05,
      "loss": 0.414,
      "num_tokens": 939328579.0,
      "step": 1185
    },
    {
      "epoch": 1.6185355862775217,
      "grad_norm": 0.11130537837743759,
      "learning_rate": 2.0609541738100223e-05,
      "loss": 0.4507,
      "num_tokens": 940098976.0,
      "step": 1186
    },
    {
      "epoch": 1.619901006997781,
      "grad_norm": 0.09744404256343842,
      "learning_rate": 2.0583111143017007e-05,
      "loss": 0.4058,
      "num_tokens": 940869277.0,
      "step": 1187
    },
    {
      "epoch": 1.6212664277180406,
      "grad_norm": 0.105192169547081,
      "learning_rate": 2.0556683621579037e-05,
      "loss": 0.4436,
      "num_tokens": 941730902.0,
      "step": 1188
    },
    {
      "epoch": 1.6226318484383002,
      "grad_norm": 0.09250310063362122,
      "learning_rate": 2.0530259231115308e-05,
      "loss": 0.4349,
      "num_tokens": 942580096.0,
      "step": 1189
    },
    {
      "epoch": 1.6239972691585596,
      "grad_norm": 0.10239614546298981,
      "learning_rate": 2.0503838028948055e-05,
      "loss": 0.4356,
      "num_tokens": 943384404.0,
      "step": 1190
    },
    {
      "epoch": 1.625362689878819,
      "grad_norm": 0.10445614904165268,
      "learning_rate": 2.0477420072392555e-05,
      "loss": 0.4363,
      "num_tokens": 944187628.0,
      "step": 1191
    },
    {
      "epoch": 1.6267281105990783,
      "grad_norm": 0.10482870042324066,
      "learning_rate": 2.0451005418757058e-05,
      "loss": 0.4194,
      "num_tokens": 944996026.0,
      "step": 1192
    },
    {
      "epoch": 1.6280935313193377,
      "grad_norm": 0.09876703470945358,
      "learning_rate": 2.0424594125342666e-05,
      "loss": 0.4255,
      "num_tokens": 945830258.0,
      "step": 1193
    },
    {
      "epoch": 1.6294589520395972,
      "grad_norm": 0.09807577729225159,
      "learning_rate": 2.0398186249443172e-05,
      "loss": 0.418,
      "num_tokens": 946630900.0,
      "step": 1194
    },
    {
      "epoch": 1.6308243727598566,
      "grad_norm": 0.09917954355478287,
      "learning_rate": 2.037178184834496e-05,
      "loss": 0.4407,
      "num_tokens": 947433484.0,
      "step": 1195
    },
    {
      "epoch": 1.6321897934801162,
      "grad_norm": 0.1095096543431282,
      "learning_rate": 2.034538097932689e-05,
      "loss": 0.4456,
      "num_tokens": 948339313.0,
      "step": 1196
    },
    {
      "epoch": 1.6335552142003755,
      "grad_norm": 0.10962647944688797,
      "learning_rate": 2.031898369966013e-05,
      "loss": 0.434,
      "num_tokens": 949106309.0,
      "step": 1197
    },
    {
      "epoch": 1.6349206349206349,
      "grad_norm": 0.09392847865819931,
      "learning_rate": 2.029259006660809e-05,
      "loss": 0.4188,
      "num_tokens": 949885415.0,
      "step": 1198
    },
    {
      "epoch": 1.6362860556408942,
      "grad_norm": 0.10660086572170258,
      "learning_rate": 2.0266200137426263e-05,
      "loss": 0.4443,
      "num_tokens": 950717090.0,
      "step": 1199
    },
    {
      "epoch": 1.6376514763611538,
      "grad_norm": 0.0942782461643219,
      "learning_rate": 2.0239813969362097e-05,
      "loss": 0.4133,
      "num_tokens": 951505245.0,
      "step": 1200
    },
    {
      "epoch": 1.6390168970814132,
      "grad_norm": 0.09615947306156158,
      "learning_rate": 2.0213431619654894e-05,
      "loss": 0.4033,
      "num_tokens": 952250524.0,
      "step": 1201
    },
    {
      "epoch": 1.6403823178016728,
      "grad_norm": 0.11451907455921173,
      "learning_rate": 2.0187053145535663e-05,
      "loss": 0.417,
      "num_tokens": 952984198.0,
      "step": 1202
    },
    {
      "epoch": 1.6417477385219321,
      "grad_norm": 0.09906096011400223,
      "learning_rate": 2.0160678604227e-05,
      "loss": 0.4464,
      "num_tokens": 953833244.0,
      "step": 1203
    },
    {
      "epoch": 1.6431131592421915,
      "grad_norm": 0.11343998461961746,
      "learning_rate": 2.0134308052942995e-05,
      "loss": 0.4247,
      "num_tokens": 954537461.0,
      "step": 1204
    },
    {
      "epoch": 1.6444785799624508,
      "grad_norm": 0.10869438201189041,
      "learning_rate": 2.0107941548889056e-05,
      "loss": 0.4122,
      "num_tokens": 955386831.0,
      "step": 1205
    },
    {
      "epoch": 1.6458440006827102,
      "grad_norm": 0.09882889688014984,
      "learning_rate": 2.008157914926182e-05,
      "loss": 0.4303,
      "num_tokens": 956215017.0,
      "step": 1206
    },
    {
      "epoch": 1.6472094214029698,
      "grad_norm": 0.0987955778837204,
      "learning_rate": 2.0055220911249034e-05,
      "loss": 0.4338,
      "num_tokens": 957009672.0,
      "step": 1207
    },
    {
      "epoch": 1.6485748421232294,
      "grad_norm": 0.10438631474971771,
      "learning_rate": 2.0028866892029395e-05,
      "loss": 0.4217,
      "num_tokens": 957769231.0,
      "step": 1208
    },
    {
      "epoch": 1.6499402628434887,
      "grad_norm": 0.11413824558258057,
      "learning_rate": 2.000251714877246e-05,
      "loss": 0.4362,
      "num_tokens": 958550093.0,
      "step": 1209
    },
    {
      "epoch": 1.651305683563748,
      "grad_norm": 0.10120825469493866,
      "learning_rate": 1.9976171738638504e-05,
      "loss": 0.4181,
      "num_tokens": 959279883.0,
      "step": 1210
    },
    {
      "epoch": 1.6526711042840074,
      "grad_norm": 0.10767576098442078,
      "learning_rate": 1.9949830718778415e-05,
      "loss": 0.4267,
      "num_tokens": 960157449.0,
      "step": 1211
    },
    {
      "epoch": 1.6540365250042668,
      "grad_norm": 0.1008666604757309,
      "learning_rate": 1.9923494146333544e-05,
      "loss": 0.4194,
      "num_tokens": 960970246.0,
      "step": 1212
    },
    {
      "epoch": 1.6554019457245264,
      "grad_norm": 0.09907978028059006,
      "learning_rate": 1.9897162078435613e-05,
      "loss": 0.4345,
      "num_tokens": 961782156.0,
      "step": 1213
    },
    {
      "epoch": 1.656767366444786,
      "grad_norm": 0.10368235409259796,
      "learning_rate": 1.9870834572206527e-05,
      "loss": 0.4248,
      "num_tokens": 962607862.0,
      "step": 1214
    },
    {
      "epoch": 1.6581327871650453,
      "grad_norm": 0.09835489839315414,
      "learning_rate": 1.984451168475836e-05,
      "loss": 0.4253,
      "num_tokens": 963402788.0,
      "step": 1215
    },
    {
      "epoch": 1.6594982078853047,
      "grad_norm": 0.11004816740751266,
      "learning_rate": 1.9818193473193113e-05,
      "loss": 0.3962,
      "num_tokens": 964120512.0,
      "step": 1216
    },
    {
      "epoch": 1.660863628605564,
      "grad_norm": 0.09124414622783661,
      "learning_rate": 1.9791879994602674e-05,
      "loss": 0.4277,
      "num_tokens": 964941873.0,
      "step": 1217
    },
    {
      "epoch": 1.6622290493258234,
      "grad_norm": 0.12096381932497025,
      "learning_rate": 1.9765571306068647e-05,
      "loss": 0.4377,
      "num_tokens": 965698041.0,
      "step": 1218
    },
    {
      "epoch": 1.663594470046083,
      "grad_norm": 0.10291089117527008,
      "learning_rate": 1.973926746466226e-05,
      "loss": 0.4084,
      "num_tokens": 966446979.0,
      "step": 1219
    },
    {
      "epoch": 1.6649598907663425,
      "grad_norm": 0.10862986743450165,
      "learning_rate": 1.9712968527444216e-05,
      "loss": 0.412,
      "num_tokens": 967230354.0,
      "step": 1220
    },
    {
      "epoch": 1.666325311486602,
      "grad_norm": 0.09565671533346176,
      "learning_rate": 1.9686674551464566e-05,
      "loss": 0.4252,
      "num_tokens": 967994746.0,
      "step": 1221
    },
    {
      "epoch": 1.6676907322068613,
      "grad_norm": 0.1098107248544693,
      "learning_rate": 1.966038559376264e-05,
      "loss": 0.4312,
      "num_tokens": 968765497.0,
      "step": 1222
    },
    {
      "epoch": 1.6690561529271206,
      "grad_norm": 0.10779411345720291,
      "learning_rate": 1.9634101711366833e-05,
      "loss": 0.4196,
      "num_tokens": 969495054.0,
      "step": 1223
    },
    {
      "epoch": 1.67042157364738,
      "grad_norm": 0.10671643912792206,
      "learning_rate": 1.9607822961294567e-05,
      "loss": 0.4203,
      "num_tokens": 970254151.0,
      "step": 1224
    },
    {
      "epoch": 1.6717869943676396,
      "grad_norm": 0.10291077941656113,
      "learning_rate": 1.958154940055212e-05,
      "loss": 0.39,
      "num_tokens": 970933870.0,
      "step": 1225
    },
    {
      "epoch": 1.673152415087899,
      "grad_norm": 0.11375992000102997,
      "learning_rate": 1.9555281086134496e-05,
      "loss": 0.4327,
      "num_tokens": 971766168.0,
      "step": 1226
    },
    {
      "epoch": 1.6745178358081585,
      "grad_norm": 0.09608227759599686,
      "learning_rate": 1.952901807502533e-05,
      "loss": 0.4264,
      "num_tokens": 972619577.0,
      "step": 1227
    },
    {
      "epoch": 1.6758832565284179,
      "grad_norm": 0.11247794330120087,
      "learning_rate": 1.9502760424196773e-05,
      "loss": 0.4291,
      "num_tokens": 973452752.0,
      "step": 1228
    },
    {
      "epoch": 1.6772486772486772,
      "grad_norm": 0.09650498628616333,
      "learning_rate": 1.9476508190609316e-05,
      "loss": 0.4132,
      "num_tokens": 974254390.0,
      "step": 1229
    },
    {
      "epoch": 1.6786140979689366,
      "grad_norm": 0.09668194502592087,
      "learning_rate": 1.945026143121171e-05,
      "loss": 0.4089,
      "num_tokens": 974988187.0,
      "step": 1230
    },
    {
      "epoch": 1.6799795186891961,
      "grad_norm": 0.11596614122390747,
      "learning_rate": 1.9424020202940844e-05,
      "loss": 0.4202,
      "num_tokens": 975742863.0,
      "step": 1231
    },
    {
      "epoch": 1.6813449394094555,
      "grad_norm": 0.11060704290866852,
      "learning_rate": 1.939778456272158e-05,
      "loss": 0.4347,
      "num_tokens": 976557115.0,
      "step": 1232
    },
    {
      "epoch": 1.682710360129715,
      "grad_norm": 0.09841381758451462,
      "learning_rate": 1.9371554567466687e-05,
      "loss": 0.4387,
      "num_tokens": 977330538.0,
      "step": 1233
    },
    {
      "epoch": 1.6840757808499744,
      "grad_norm": 0.10936102271080017,
      "learning_rate": 1.9345330274076675e-05,
      "loss": 0.4084,
      "num_tokens": 978106762.0,
      "step": 1234
    },
    {
      "epoch": 1.6854412015702338,
      "grad_norm": 0.10843499004840851,
      "learning_rate": 1.931911173943968e-05,
      "loss": 0.45,
      "num_tokens": 978908908.0,
      "step": 1235
    },
    {
      "epoch": 1.6868066222904932,
      "grad_norm": 0.09672868996858597,
      "learning_rate": 1.9292899020431356e-05,
      "loss": 0.417,
      "num_tokens": 979757591.0,
      "step": 1236
    },
    {
      "epoch": 1.6881720430107527,
      "grad_norm": 0.10432808101177216,
      "learning_rate": 1.9266692173914747e-05,
      "loss": 0.4311,
      "num_tokens": 980626835.0,
      "step": 1237
    },
    {
      "epoch": 1.689537463731012,
      "grad_norm": 0.09221287816762924,
      "learning_rate": 1.9240491256740127e-05,
      "loss": 0.4035,
      "num_tokens": 981404618.0,
      "step": 1238
    },
    {
      "epoch": 1.6909028844512717,
      "grad_norm": 0.1071370393037796,
      "learning_rate": 1.921429632574494e-05,
      "loss": 0.4243,
      "num_tokens": 982201485.0,
      "step": 1239
    },
    {
      "epoch": 1.692268305171531,
      "grad_norm": 0.09250541776418686,
      "learning_rate": 1.9188107437753633e-05,
      "loss": 0.4388,
      "num_tokens": 983013541.0,
      "step": 1240
    },
    {
      "epoch": 1.6936337258917904,
      "grad_norm": 0.10701955109834671,
      "learning_rate": 1.9161924649577542e-05,
      "loss": 0.3974,
      "num_tokens": 983742301.0,
      "step": 1241
    },
    {
      "epoch": 1.6949991466120498,
      "grad_norm": 0.10282711684703827,
      "learning_rate": 1.913574801801478e-05,
      "loss": 0.4256,
      "num_tokens": 984490034.0,
      "step": 1242
    },
    {
      "epoch": 1.6963645673323091,
      "grad_norm": 0.10613784939050674,
      "learning_rate": 1.9109577599850083e-05,
      "loss": 0.4159,
      "num_tokens": 985271325.0,
      "step": 1243
    },
    {
      "epoch": 1.6977299880525687,
      "grad_norm": 0.10933897644281387,
      "learning_rate": 1.908341345185472e-05,
      "loss": 0.4297,
      "num_tokens": 986102589.0,
      "step": 1244
    },
    {
      "epoch": 1.6990954087728283,
      "grad_norm": 0.10248047858476639,
      "learning_rate": 1.9057255630786376e-05,
      "loss": 0.4196,
      "num_tokens": 986905815.0,
      "step": 1245
    },
    {
      "epoch": 1.7004608294930876,
      "grad_norm": 0.08874692022800446,
      "learning_rate": 1.903110419338899e-05,
      "loss": 0.4336,
      "num_tokens": 987683747.0,
      "step": 1246
    },
    {
      "epoch": 1.701826250213347,
      "grad_norm": 0.11156293749809265,
      "learning_rate": 1.9004959196392652e-05,
      "loss": 0.4337,
      "num_tokens": 988499119.0,
      "step": 1247
    },
    {
      "epoch": 1.7031916709336063,
      "grad_norm": 0.11304923892021179,
      "learning_rate": 1.89788206965135e-05,
      "loss": 0.4334,
      "num_tokens": 989278731.0,
      "step": 1248
    },
    {
      "epoch": 1.7045570916538657,
      "grad_norm": 0.09855560958385468,
      "learning_rate": 1.895268875045355e-05,
      "loss": 0.416,
      "num_tokens": 990123804.0,
      "step": 1249
    },
    {
      "epoch": 1.7059225123741253,
      "grad_norm": 0.10234690457582474,
      "learning_rate": 1.892656341490062e-05,
      "loss": 0.4232,
      "num_tokens": 990912624.0,
      "step": 1250
    },
    {
      "epoch": 1.7072879330943849,
      "grad_norm": 0.11822590976953506,
      "learning_rate": 1.890044474652818e-05,
      "loss": 0.4153,
      "num_tokens": 991641761.0,
      "step": 1251
    },
    {
      "epoch": 1.7086533538146442,
      "grad_norm": 0.10367462038993835,
      "learning_rate": 1.8874332801995258e-05,
      "loss": 0.4369,
      "num_tokens": 992427975.0,
      "step": 1252
    },
    {
      "epoch": 1.7100187745349036,
      "grad_norm": 0.11220160126686096,
      "learning_rate": 1.8848227637946257e-05,
      "loss": 0.4052,
      "num_tokens": 993173677.0,
      "step": 1253
    },
    {
      "epoch": 1.711384195255163,
      "grad_norm": 0.10756947100162506,
      "learning_rate": 1.8822129311010918e-05,
      "loss": 0.422,
      "num_tokens": 993969869.0,
      "step": 1254
    },
    {
      "epoch": 1.7127496159754223,
      "grad_norm": 0.11073581129312515,
      "learning_rate": 1.8796037877804098e-05,
      "loss": 0.4231,
      "num_tokens": 994735532.0,
      "step": 1255
    },
    {
      "epoch": 1.7141150366956819,
      "grad_norm": 0.0940244048833847,
      "learning_rate": 1.8769953394925747e-05,
      "loss": 0.3874,
      "num_tokens": 995472010.0,
      "step": 1256
    },
    {
      "epoch": 1.7154804574159412,
      "grad_norm": 0.10193836688995361,
      "learning_rate": 1.8743875918960702e-05,
      "loss": 0.4255,
      "num_tokens": 996229055.0,
      "step": 1257
    },
    {
      "epoch": 1.7168458781362008,
      "grad_norm": 0.11459581553936005,
      "learning_rate": 1.8717805506478624e-05,
      "loss": 0.4371,
      "num_tokens": 997075833.0,
      "step": 1258
    },
    {
      "epoch": 1.7182112988564602,
      "grad_norm": 0.09934771060943604,
      "learning_rate": 1.869174221403385e-05,
      "loss": 0.408,
      "num_tokens": 997834457.0,
      "step": 1259
    },
    {
      "epoch": 1.7195767195767195,
      "grad_norm": 0.09764434397220612,
      "learning_rate": 1.866568609816525e-05,
      "loss": 0.4272,
      "num_tokens": 998687422.0,
      "step": 1260
    },
    {
      "epoch": 1.720942140296979,
      "grad_norm": 0.10948468744754791,
      "learning_rate": 1.863963721539614e-05,
      "loss": 0.4096,
      "num_tokens": 999459349.0,
      "step": 1261
    },
    {
      "epoch": 1.7223075610172385,
      "grad_norm": 0.09778808057308197,
      "learning_rate": 1.8613595622234158e-05,
      "loss": 0.4382,
      "num_tokens": 1000275200.0,
      "step": 1262
    },
    {
      "epoch": 1.7236729817374978,
      "grad_norm": 0.09645889699459076,
      "learning_rate": 1.8587561375171104e-05,
      "loss": 0.4249,
      "num_tokens": 1001022661.0,
      "step": 1263
    },
    {
      "epoch": 1.7250384024577574,
      "grad_norm": 0.10575588792562485,
      "learning_rate": 1.856153453068286e-05,
      "loss": 0.4134,
      "num_tokens": 1001795746.0,
      "step": 1264
    },
    {
      "epoch": 1.7264038231780168,
      "grad_norm": 0.09924504160881042,
      "learning_rate": 1.8535515145229247e-05,
      "loss": 0.433,
      "num_tokens": 1002641080.0,
      "step": 1265
    },
    {
      "epoch": 1.7277692438982761,
      "grad_norm": 0.10560528188943863,
      "learning_rate": 1.85095032752539e-05,
      "loss": 0.4157,
      "num_tokens": 1003431199.0,
      "step": 1266
    },
    {
      "epoch": 1.7291346646185355,
      "grad_norm": 0.09562241286039352,
      "learning_rate": 1.8483498977184152e-05,
      "loss": 0.4013,
      "num_tokens": 1004199642.0,
      "step": 1267
    },
    {
      "epoch": 1.730500085338795,
      "grad_norm": 0.0954812541604042,
      "learning_rate": 1.84575023074309e-05,
      "loss": 0.4029,
      "num_tokens": 1004955479.0,
      "step": 1268
    },
    {
      "epoch": 1.7318655060590544,
      "grad_norm": 0.09561867266893387,
      "learning_rate": 1.8431513322388522e-05,
      "loss": 0.4052,
      "num_tokens": 1005731414.0,
      "step": 1269
    },
    {
      "epoch": 1.733230926779314,
      "grad_norm": 0.10708153247833252,
      "learning_rate": 1.8405532078434695e-05,
      "loss": 0.4164,
      "num_tokens": 1006506654.0,
      "step": 1270
    },
    {
      "epoch": 1.7345963474995734,
      "grad_norm": 0.10320819914340973,
      "learning_rate": 1.8379558631930324e-05,
      "loss": 0.422,
      "num_tokens": 1007327791.0,
      "step": 1271
    },
    {
      "epoch": 1.7359617682198327,
      "grad_norm": 0.10671532154083252,
      "learning_rate": 1.835359303921939e-05,
      "loss": 0.4555,
      "num_tokens": 1008149678.0,
      "step": 1272
    },
    {
      "epoch": 1.737327188940092,
      "grad_norm": 0.10446067899465561,
      "learning_rate": 1.832763535662884e-05,
      "loss": 0.4289,
      "num_tokens": 1008946649.0,
      "step": 1273
    },
    {
      "epoch": 1.7386926096603514,
      "grad_norm": 0.10558211803436279,
      "learning_rate": 1.830168564046845e-05,
      "loss": 0.4304,
      "num_tokens": 1009730603.0,
      "step": 1274
    },
    {
      "epoch": 1.740058030380611,
      "grad_norm": 0.10401894152164459,
      "learning_rate": 1.827574394703073e-05,
      "loss": 0.4384,
      "num_tokens": 1010490882.0,
      "step": 1275
    },
    {
      "epoch": 1.7414234511008706,
      "grad_norm": 0.11288337409496307,
      "learning_rate": 1.824981033259078e-05,
      "loss": 0.4401,
      "num_tokens": 1011213415.0,
      "step": 1276
    },
    {
      "epoch": 1.74278887182113,
      "grad_norm": 0.09810886532068253,
      "learning_rate": 1.8223884853406172e-05,
      "loss": 0.4195,
      "num_tokens": 1011991059.0,
      "step": 1277
    },
    {
      "epoch": 1.7441542925413893,
      "grad_norm": 0.09039241075515747,
      "learning_rate": 1.8197967565716847e-05,
      "loss": 0.4188,
      "num_tokens": 1012823541.0,
      "step": 1278
    },
    {
      "epoch": 1.7455197132616487,
      "grad_norm": 0.10296230763196945,
      "learning_rate": 1.8172058525744948e-05,
      "loss": 0.4338,
      "num_tokens": 1013657151.0,
      "step": 1279
    },
    {
      "epoch": 1.746885133981908,
      "grad_norm": 0.10214242339134216,
      "learning_rate": 1.8146157789694737e-05,
      "loss": 0.4312,
      "num_tokens": 1014457375.0,
      "step": 1280
    },
    {
      "epoch": 1.7482505547021676,
      "grad_norm": 0.10315077006816864,
      "learning_rate": 1.8120265413752475e-05,
      "loss": 0.4089,
      "num_tokens": 1015316709.0,
      "step": 1281
    },
    {
      "epoch": 1.7496159754224272,
      "grad_norm": 0.10303769260644913,
      "learning_rate": 1.8094381454086278e-05,
      "loss": 0.4246,
      "num_tokens": 1016091820.0,
      "step": 1282
    },
    {
      "epoch": 1.7509813961426866,
      "grad_norm": 0.09959693253040314,
      "learning_rate": 1.8068505966846007e-05,
      "loss": 0.4349,
      "num_tokens": 1016938282.0,
      "step": 1283
    },
    {
      "epoch": 1.752346816862946,
      "grad_norm": 0.10206132382154465,
      "learning_rate": 1.8042639008163133e-05,
      "loss": 0.4269,
      "num_tokens": 1017732359.0,
      "step": 1284
    },
    {
      "epoch": 1.7537122375832053,
      "grad_norm": 0.09955953806638718,
      "learning_rate": 1.8016780634150635e-05,
      "loss": 0.4239,
      "num_tokens": 1018603105.0,
      "step": 1285
    },
    {
      "epoch": 1.7550776583034646,
      "grad_norm": 0.10602343082427979,
      "learning_rate": 1.7990930900902884e-05,
      "loss": 0.4424,
      "num_tokens": 1019496785.0,
      "step": 1286
    },
    {
      "epoch": 1.7564430790237242,
      "grad_norm": 0.09596966207027435,
      "learning_rate": 1.7965089864495472e-05,
      "loss": 0.4205,
      "num_tokens": 1020244646.0,
      "step": 1287
    },
    {
      "epoch": 1.7578084997439836,
      "grad_norm": 0.10619444400072098,
      "learning_rate": 1.7939257580985162e-05,
      "loss": 0.4415,
      "num_tokens": 1021067981.0,
      "step": 1288
    },
    {
      "epoch": 1.7591739204642431,
      "grad_norm": 0.10434774309396744,
      "learning_rate": 1.791343410640971e-05,
      "loss": 0.42,
      "num_tokens": 1021773796.0,
      "step": 1289
    },
    {
      "epoch": 1.7605393411845025,
      "grad_norm": 0.10765963792800903,
      "learning_rate": 1.7887619496787754e-05,
      "loss": 0.4462,
      "num_tokens": 1022592246.0,
      "step": 1290
    },
    {
      "epoch": 1.7619047619047619,
      "grad_norm": 0.10404103994369507,
      "learning_rate": 1.786181380811872e-05,
      "loss": 0.4177,
      "num_tokens": 1023286591.0,
      "step": 1291
    },
    {
      "epoch": 1.7632701826250212,
      "grad_norm": 0.10079611092805862,
      "learning_rate": 1.7836017096382673e-05,
      "loss": 0.4319,
      "num_tokens": 1024030669.0,
      "step": 1292
    },
    {
      "epoch": 1.7646356033452808,
      "grad_norm": 0.09758203476667404,
      "learning_rate": 1.7810229417540213e-05,
      "loss": 0.426,
      "num_tokens": 1024788545.0,
      "step": 1293
    },
    {
      "epoch": 1.7660010240655402,
      "grad_norm": 0.10819488763809204,
      "learning_rate": 1.7784450827532323e-05,
      "loss": 0.4103,
      "num_tokens": 1025578420.0,
      "step": 1294
    },
    {
      "epoch": 1.7673664447857997,
      "grad_norm": 0.10110345482826233,
      "learning_rate": 1.7758681382280302e-05,
      "loss": 0.4404,
      "num_tokens": 1026412530.0,
      "step": 1295
    },
    {
      "epoch": 1.768731865506059,
      "grad_norm": 0.099916011095047,
      "learning_rate": 1.773292113768558e-05,
      "loss": 0.4397,
      "num_tokens": 1027337955.0,
      "step": 1296
    },
    {
      "epoch": 1.7700972862263185,
      "grad_norm": 0.09616987407207489,
      "learning_rate": 1.770717014962965e-05,
      "loss": 0.4379,
      "num_tokens": 1028168892.0,
      "step": 1297
    },
    {
      "epoch": 1.7714627069465778,
      "grad_norm": 0.11512433737516403,
      "learning_rate": 1.7681428473973908e-05,
      "loss": 0.4132,
      "num_tokens": 1028944499.0,
      "step": 1298
    },
    {
      "epoch": 1.7728281276668374,
      "grad_norm": 0.09603319317102432,
      "learning_rate": 1.7655696166559563e-05,
      "loss": 0.4076,
      "num_tokens": 1029673074.0,
      "step": 1299
    },
    {
      "epoch": 1.7741935483870968,
      "grad_norm": 0.10091511905193329,
      "learning_rate": 1.76299732832075e-05,
      "loss": 0.4004,
      "num_tokens": 1030458876.0,
      "step": 1300
    },
    {
      "epoch": 1.7755589691073563,
      "grad_norm": 0.12591512501239777,
      "learning_rate": 1.760425987971816e-05,
      "loss": 0.457,
      "num_tokens": 1031241488.0,
      "step": 1301
    },
    {
      "epoch": 1.7769243898276157,
      "grad_norm": 0.09369759261608124,
      "learning_rate": 1.7578556011871398e-05,
      "loss": 0.4221,
      "num_tokens": 1032011071.0,
      "step": 1302
    },
    {
      "epoch": 1.778289810547875,
      "grad_norm": 0.10140180587768555,
      "learning_rate": 1.7552861735426415e-05,
      "loss": 0.4144,
      "num_tokens": 1032764805.0,
      "step": 1303
    },
    {
      "epoch": 1.7796552312681344,
      "grad_norm": 0.1144343689084053,
      "learning_rate": 1.7527177106121588e-05,
      "loss": 0.4082,
      "num_tokens": 1033471368.0,
      "step": 1304
    },
    {
      "epoch": 1.7810206519883938,
      "grad_norm": 0.10481277108192444,
      "learning_rate": 1.7501502179674378e-05,
      "loss": 0.4073,
      "num_tokens": 1034188086.0,
      "step": 1305
    },
    {
      "epoch": 1.7823860727086533,
      "grad_norm": 0.10635516792535782,
      "learning_rate": 1.7475837011781175e-05,
      "loss": 0.4341,
      "num_tokens": 1034978588.0,
      "step": 1306
    },
    {
      "epoch": 1.783751493428913,
      "grad_norm": 0.1039450466632843,
      "learning_rate": 1.745018165811723e-05,
      "loss": 0.426,
      "num_tokens": 1035721266.0,
      "step": 1307
    },
    {
      "epoch": 1.7851169141491723,
      "grad_norm": 0.09656086564064026,
      "learning_rate": 1.7424536174336478e-05,
      "loss": 0.4083,
      "num_tokens": 1036502290.0,
      "step": 1308
    },
    {
      "epoch": 1.7864823348694316,
      "grad_norm": 0.10273008048534393,
      "learning_rate": 1.7398900616071457e-05,
      "loss": 0.4115,
      "num_tokens": 1037255022.0,
      "step": 1309
    },
    {
      "epoch": 1.787847755589691,
      "grad_norm": 0.10038214176893234,
      "learning_rate": 1.737327503893318e-05,
      "loss": 0.4071,
      "num_tokens": 1038019772.0,
      "step": 1310
    },
    {
      "epoch": 1.7892131763099504,
      "grad_norm": 0.10686325281858444,
      "learning_rate": 1.7347659498510983e-05,
      "loss": 0.4235,
      "num_tokens": 1038775929.0,
      "step": 1311
    },
    {
      "epoch": 1.79057859703021,
      "grad_norm": 0.1119302287697792,
      "learning_rate": 1.7322054050372455e-05,
      "loss": 0.425,
      "num_tokens": 1039558152.0,
      "step": 1312
    },
    {
      "epoch": 1.7919440177504695,
      "grad_norm": 0.10013668984174728,
      "learning_rate": 1.7296458750063285e-05,
      "loss": 0.4597,
      "num_tokens": 1040345161.0,
      "step": 1313
    },
    {
      "epoch": 1.7933094384707289,
      "grad_norm": 0.11377568542957306,
      "learning_rate": 1.7270873653107143e-05,
      "loss": 0.4124,
      "num_tokens": 1041099101.0,
      "step": 1314
    },
    {
      "epoch": 1.7946748591909882,
      "grad_norm": 0.10531895607709885,
      "learning_rate": 1.724529881500556e-05,
      "loss": 0.4248,
      "num_tokens": 1041891210.0,
      "step": 1315
    },
    {
      "epoch": 1.7960402799112476,
      "grad_norm": 0.09123875200748444,
      "learning_rate": 1.7219734291237832e-05,
      "loss": 0.4133,
      "num_tokens": 1042594045.0,
      "step": 1316
    },
    {
      "epoch": 1.797405700631507,
      "grad_norm": 0.11393178254365921,
      "learning_rate": 1.7194180137260853e-05,
      "loss": 0.4343,
      "num_tokens": 1043352629.0,
      "step": 1317
    },
    {
      "epoch": 1.7987711213517665,
      "grad_norm": 0.11243606358766556,
      "learning_rate": 1.7168636408509064e-05,
      "loss": 0.4148,
      "num_tokens": 1044151121.0,
      "step": 1318
    },
    {
      "epoch": 1.800136542072026,
      "grad_norm": 0.10705243796110153,
      "learning_rate": 1.714310316039424e-05,
      "loss": 0.4304,
      "num_tokens": 1044859590.0,
      "step": 1319
    },
    {
      "epoch": 1.8015019627922855,
      "grad_norm": 0.09881531447172165,
      "learning_rate": 1.7117580448305454e-05,
      "loss": 0.4095,
      "num_tokens": 1045606465.0,
      "step": 1320
    },
    {
      "epoch": 1.8028673835125448,
      "grad_norm": 0.10583215951919556,
      "learning_rate": 1.7092068327608903e-05,
      "loss": 0.4145,
      "num_tokens": 1046290193.0,
      "step": 1321
    },
    {
      "epoch": 1.8042328042328042,
      "grad_norm": 0.11174498498439789,
      "learning_rate": 1.706656685364783e-05,
      "loss": 0.4086,
      "num_tokens": 1046976064.0,
      "step": 1322
    },
    {
      "epoch": 1.8055982249530635,
      "grad_norm": 0.09245815873146057,
      "learning_rate": 1.704107608174237e-05,
      "loss": 0.4345,
      "num_tokens": 1047747108.0,
      "step": 1323
    },
    {
      "epoch": 1.8069636456733231,
      "grad_norm": 0.09690052270889282,
      "learning_rate": 1.7015596067189444e-05,
      "loss": 0.4294,
      "num_tokens": 1048553007.0,
      "step": 1324
    },
    {
      "epoch": 1.8083290663935825,
      "grad_norm": 0.10531186312437057,
      "learning_rate": 1.6990126865262634e-05,
      "loss": 0.4475,
      "num_tokens": 1049325794.0,
      "step": 1325
    },
    {
      "epoch": 1.809694487113842,
      "grad_norm": 0.10867897421121597,
      "learning_rate": 1.6964668531212064e-05,
      "loss": 0.4123,
      "num_tokens": 1050040355.0,
      "step": 1326
    },
    {
      "epoch": 1.8110599078341014,
      "grad_norm": 0.10652785003185272,
      "learning_rate": 1.693922112026429e-05,
      "loss": 0.4192,
      "num_tokens": 1050864685.0,
      "step": 1327
    },
    {
      "epoch": 1.8124253285543608,
      "grad_norm": 0.09433171153068542,
      "learning_rate": 1.691378468762217e-05,
      "loss": 0.4259,
      "num_tokens": 1051685988.0,
      "step": 1328
    },
    {
      "epoch": 1.8137907492746201,
      "grad_norm": 0.10933739691972733,
      "learning_rate": 1.688835928846475e-05,
      "loss": 0.4105,
      "num_tokens": 1052438936.0,
      "step": 1329
    },
    {
      "epoch": 1.8151561699948797,
      "grad_norm": 0.10533864051103592,
      "learning_rate": 1.686294497794714e-05,
      "loss": 0.4325,
      "num_tokens": 1053223670.0,
      "step": 1330
    },
    {
      "epoch": 1.816521590715139,
      "grad_norm": 0.10096635669469833,
      "learning_rate": 1.6837541811200382e-05,
      "loss": 0.4101,
      "num_tokens": 1053981128.0,
      "step": 1331
    },
    {
      "epoch": 1.8178870114353987,
      "grad_norm": 0.0914192795753479,
      "learning_rate": 1.6812149843331358e-05,
      "loss": 0.4129,
      "num_tokens": 1054765567.0,
      "step": 1332
    },
    {
      "epoch": 1.819252432155658,
      "grad_norm": 0.09378449618816376,
      "learning_rate": 1.678676912942266e-05,
      "loss": 0.4365,
      "num_tokens": 1055586607.0,
      "step": 1333
    },
    {
      "epoch": 1.8206178528759174,
      "grad_norm": 0.10227354615926743,
      "learning_rate": 1.6761399724532446e-05,
      "loss": 0.4168,
      "num_tokens": 1056409531.0,
      "step": 1334
    },
    {
      "epoch": 1.8219832735961767,
      "grad_norm": 0.09901878982782364,
      "learning_rate": 1.6736041683694372e-05,
      "loss": 0.4399,
      "num_tokens": 1057182961.0,
      "step": 1335
    },
    {
      "epoch": 1.823348694316436,
      "grad_norm": 0.10653631389141083,
      "learning_rate": 1.671069506191743e-05,
      "loss": 0.4128,
      "num_tokens": 1057969537.0,
      "step": 1336
    },
    {
      "epoch": 1.8247141150366957,
      "grad_norm": 0.09764046221971512,
      "learning_rate": 1.668535991418582e-05,
      "loss": 0.4133,
      "num_tokens": 1058771263.0,
      "step": 1337
    },
    {
      "epoch": 1.8260795357569553,
      "grad_norm": 0.0965229868888855,
      "learning_rate": 1.6660036295458863e-05,
      "loss": 0.4161,
      "num_tokens": 1059585123.0,
      "step": 1338
    },
    {
      "epoch": 1.8274449564772146,
      "grad_norm": 0.0965578556060791,
      "learning_rate": 1.6634724260670894e-05,
      "loss": 0.4192,
      "num_tokens": 1060371810.0,
      "step": 1339
    },
    {
      "epoch": 1.828810377197474,
      "grad_norm": 0.1004619300365448,
      "learning_rate": 1.660942386473109e-05,
      "loss": 0.4135,
      "num_tokens": 1061181439.0,
      "step": 1340
    },
    {
      "epoch": 1.8301757979177333,
      "grad_norm": 0.09298712015151978,
      "learning_rate": 1.658413516252339e-05,
      "loss": 0.4319,
      "num_tokens": 1061943758.0,
      "step": 1341
    },
    {
      "epoch": 1.8315412186379927,
      "grad_norm": 0.10247784852981567,
      "learning_rate": 1.6558858208906375e-05,
      "loss": 0.4462,
      "num_tokens": 1062808721.0,
      "step": 1342
    },
    {
      "epoch": 1.8329066393582523,
      "grad_norm": 0.10446092486381531,
      "learning_rate": 1.653359305871312e-05,
      "loss": 0.4249,
      "num_tokens": 1063615244.0,
      "step": 1343
    },
    {
      "epoch": 1.8342720600785118,
      "grad_norm": 0.09736216068267822,
      "learning_rate": 1.6508339766751106e-05,
      "loss": 0.4507,
      "num_tokens": 1064386752.0,
      "step": 1344
    },
    {
      "epoch": 1.8356374807987712,
      "grad_norm": 0.10410141199827194,
      "learning_rate": 1.6483098387802098e-05,
      "loss": 0.4097,
      "num_tokens": 1065113745.0,
      "step": 1345
    },
    {
      "epoch": 1.8370029015190306,
      "grad_norm": 0.0993262529373169,
      "learning_rate": 1.6457868976622e-05,
      "loss": 0.4143,
      "num_tokens": 1065929542.0,
      "step": 1346
    },
    {
      "epoch": 1.83836832223929,
      "grad_norm": 0.0939323753118515,
      "learning_rate": 1.643265158794077e-05,
      "loss": 0.439,
      "num_tokens": 1066742446.0,
      "step": 1347
    },
    {
      "epoch": 1.8397337429595493,
      "grad_norm": 0.09184844046831131,
      "learning_rate": 1.640744627646229e-05,
      "loss": 0.4099,
      "num_tokens": 1067526789.0,
      "step": 1348
    },
    {
      "epoch": 1.8410991636798089,
      "grad_norm": 0.10677743703126907,
      "learning_rate": 1.6382253096864217e-05,
      "loss": 0.445,
      "num_tokens": 1068292661.0,
      "step": 1349
    },
    {
      "epoch": 1.8424645844000682,
      "grad_norm": 0.09920361638069153,
      "learning_rate": 1.635707210379791e-05,
      "loss": 0.4278,
      "num_tokens": 1069124833.0,
      "step": 1350
    },
    {
      "epoch": 1.8438300051203278,
      "grad_norm": 0.09593260288238525,
      "learning_rate": 1.6331903351888295e-05,
      "loss": 0.4304,
      "num_tokens": 1069941702.0,
      "step": 1351
    },
    {
      "epoch": 1.8451954258405872,
      "grad_norm": 0.09732770919799805,
      "learning_rate": 1.630674689573373e-05,
      "loss": 0.4278,
      "num_tokens": 1070713181.0,
      "step": 1352
    },
    {
      "epoch": 1.8465608465608465,
      "grad_norm": 0.10098724067211151,
      "learning_rate": 1.628160278990591e-05,
      "loss": 0.4265,
      "num_tokens": 1071520917.0,
      "step": 1353
    },
    {
      "epoch": 1.8479262672811059,
      "grad_norm": 0.10264068841934204,
      "learning_rate": 1.6256471088949742e-05,
      "loss": 0.4368,
      "num_tokens": 1072306312.0,
      "step": 1354
    },
    {
      "epoch": 1.8492916880013655,
      "grad_norm": 0.10062193125486374,
      "learning_rate": 1.62313518473832e-05,
      "loss": 0.4136,
      "num_tokens": 1073069991.0,
      "step": 1355
    },
    {
      "epoch": 1.8506571087216248,
      "grad_norm": 0.09308270364999771,
      "learning_rate": 1.6206245119697262e-05,
      "loss": 0.4485,
      "num_tokens": 1073889448.0,
      "step": 1356
    },
    {
      "epoch": 1.8520225294418844,
      "grad_norm": 0.10951436311006546,
      "learning_rate": 1.6181150960355734e-05,
      "loss": 0.4324,
      "num_tokens": 1074693586.0,
      "step": 1357
    },
    {
      "epoch": 1.8533879501621437,
      "grad_norm": 0.0969163328409195,
      "learning_rate": 1.615606942379517e-05,
      "loss": 0.4317,
      "num_tokens": 1075502984.0,
      "step": 1358
    },
    {
      "epoch": 1.854753370882403,
      "grad_norm": 0.10211966186761856,
      "learning_rate": 1.6131000564424747e-05,
      "loss": 0.4223,
      "num_tokens": 1076236906.0,
      "step": 1359
    },
    {
      "epoch": 1.8561187916026625,
      "grad_norm": 0.09796388447284698,
      "learning_rate": 1.6105944436626123e-05,
      "loss": 0.4463,
      "num_tokens": 1077113329.0,
      "step": 1360
    },
    {
      "epoch": 1.857484212322922,
      "grad_norm": 0.09188685566186905,
      "learning_rate": 1.6080901094753357e-05,
      "loss": 0.4226,
      "num_tokens": 1077909759.0,
      "step": 1361
    },
    {
      "epoch": 1.8588496330431814,
      "grad_norm": 0.1016664057970047,
      "learning_rate": 1.6055870593132756e-05,
      "loss": 0.4284,
      "num_tokens": 1078762966.0,
      "step": 1362
    },
    {
      "epoch": 1.860215053763441,
      "grad_norm": 0.09719672054052353,
      "learning_rate": 1.6030852986062782e-05,
      "loss": 0.412,
      "num_tokens": 1079529774.0,
      "step": 1363
    },
    {
      "epoch": 1.8615804744837003,
      "grad_norm": 0.09942573308944702,
      "learning_rate": 1.6005848327813924e-05,
      "loss": 0.433,
      "num_tokens": 1080252465.0,
      "step": 1364
    },
    {
      "epoch": 1.8629458952039597,
      "grad_norm": 0.09560124576091766,
      "learning_rate": 1.5980856672628583e-05,
      "loss": 0.4238,
      "num_tokens": 1081043315.0,
      "step": 1365
    },
    {
      "epoch": 1.864311315924219,
      "grad_norm": 0.10372166335582733,
      "learning_rate": 1.5955878074720938e-05,
      "loss": 0.4216,
      "num_tokens": 1081790261.0,
      "step": 1366
    },
    {
      "epoch": 1.8656767366444784,
      "grad_norm": 0.09641843289136887,
      "learning_rate": 1.5930912588276863e-05,
      "loss": 0.4284,
      "num_tokens": 1082588616.0,
      "step": 1367
    },
    {
      "epoch": 1.867042157364738,
      "grad_norm": 0.09407299011945724,
      "learning_rate": 1.590596026745378e-05,
      "loss": 0.4365,
      "num_tokens": 1083400860.0,
      "step": 1368
    },
    {
      "epoch": 1.8684075780849976,
      "grad_norm": 0.11026313900947571,
      "learning_rate": 1.5881021166380547e-05,
      "loss": 0.4398,
      "num_tokens": 1084171509.0,
      "step": 1369
    },
    {
      "epoch": 1.869772998805257,
      "grad_norm": 0.10270657390356064,
      "learning_rate": 1.5856095339157357e-05,
      "loss": 0.4136,
      "num_tokens": 1084999465.0,
      "step": 1370
    },
    {
      "epoch": 1.8711384195255163,
      "grad_norm": 0.10813093185424805,
      "learning_rate": 1.58311828398556e-05,
      "loss": 0.4192,
      "num_tokens": 1085785619.0,
      "step": 1371
    },
    {
      "epoch": 1.8725038402457757,
      "grad_norm": 0.09777439385652542,
      "learning_rate": 1.5806283722517747e-05,
      "loss": 0.4202,
      "num_tokens": 1086561771.0,
      "step": 1372
    },
    {
      "epoch": 1.873869260966035,
      "grad_norm": 0.10611224174499512,
      "learning_rate": 1.5781398041157245e-05,
      "loss": 0.4346,
      "num_tokens": 1087387505.0,
      "step": 1373
    },
    {
      "epoch": 1.8752346816862946,
      "grad_norm": 0.10068873316049576,
      "learning_rate": 1.5756525849758407e-05,
      "loss": 0.4312,
      "num_tokens": 1088192198.0,
      "step": 1374
    },
    {
      "epoch": 1.8766001024065542,
      "grad_norm": 0.09615612030029297,
      "learning_rate": 1.573166720227626e-05,
      "loss": 0.4263,
      "num_tokens": 1089050181.0,
      "step": 1375
    },
    {
      "epoch": 1.8779655231268135,
      "grad_norm": 0.10142555832862854,
      "learning_rate": 1.5706822152636472e-05,
      "loss": 0.4231,
      "num_tokens": 1089789085.0,
      "step": 1376
    },
    {
      "epoch": 1.8793309438470729,
      "grad_norm": 0.09990958124399185,
      "learning_rate": 1.56819907547352e-05,
      "loss": 0.4341,
      "num_tokens": 1090478922.0,
      "step": 1377
    },
    {
      "epoch": 1.8806963645673322,
      "grad_norm": 0.10351818799972534,
      "learning_rate": 1.5657173062438986e-05,
      "loss": 0.405,
      "num_tokens": 1091281811.0,
      "step": 1378
    },
    {
      "epoch": 1.8820617852875916,
      "grad_norm": 0.10006029903888702,
      "learning_rate": 1.5632369129584645e-05,
      "loss": 0.4098,
      "num_tokens": 1092062502.0,
      "step": 1379
    },
    {
      "epoch": 1.8834272060078512,
      "grad_norm": 0.10143296420574188,
      "learning_rate": 1.5607579009979145e-05,
      "loss": 0.4245,
      "num_tokens": 1092844030.0,
      "step": 1380
    },
    {
      "epoch": 1.8847926267281108,
      "grad_norm": 0.09553175419569016,
      "learning_rate": 1.5582802757399485e-05,
      "loss": 0.4482,
      "num_tokens": 1093703152.0,
      "step": 1381
    },
    {
      "epoch": 1.8861580474483701,
      "grad_norm": 0.09664008766412735,
      "learning_rate": 1.5558040425592584e-05,
      "loss": 0.4118,
      "num_tokens": 1094496330.0,
      "step": 1382
    },
    {
      "epoch": 1.8875234681686295,
      "grad_norm": 0.09819834679365158,
      "learning_rate": 1.5533292068275173e-05,
      "loss": 0.4245,
      "num_tokens": 1095287277.0,
      "step": 1383
    },
    {
      "epoch": 1.8888888888888888,
      "grad_norm": 0.09456814080476761,
      "learning_rate": 1.5508557739133647e-05,
      "loss": 0.4025,
      "num_tokens": 1096043409.0,
      "step": 1384
    },
    {
      "epoch": 1.8902543096091482,
      "grad_norm": 0.09407369792461395,
      "learning_rate": 1.5483837491823975e-05,
      "loss": 0.4102,
      "num_tokens": 1096778795.0,
      "step": 1385
    },
    {
      "epoch": 1.8916197303294078,
      "grad_norm": 0.09829109162092209,
      "learning_rate": 1.545913137997159e-05,
      "loss": 0.4174,
      "num_tokens": 1097587021.0,
      "step": 1386
    },
    {
      "epoch": 1.8929851510496671,
      "grad_norm": 0.1076306700706482,
      "learning_rate": 1.5434439457171274e-05,
      "loss": 0.4151,
      "num_tokens": 1098390570.0,
      "step": 1387
    },
    {
      "epoch": 1.8943505717699267,
      "grad_norm": 0.08842550218105316,
      "learning_rate": 1.540976177698698e-05,
      "loss": 0.4236,
      "num_tokens": 1099202976.0,
      "step": 1388
    },
    {
      "epoch": 1.895715992490186,
      "grad_norm": 0.10222078859806061,
      "learning_rate": 1.5385098392951826e-05,
      "loss": 0.4211,
      "num_tokens": 1100046373.0,
      "step": 1389
    },
    {
      "epoch": 1.8970814132104454,
      "grad_norm": 0.10492260009050369,
      "learning_rate": 1.5360449358567864e-05,
      "loss": 0.3983,
      "num_tokens": 1100791999.0,
      "step": 1390
    },
    {
      "epoch": 1.8984468339307048,
      "grad_norm": 0.09411447495222092,
      "learning_rate": 1.5335814727306044e-05,
      "loss": 0.4177,
      "num_tokens": 1101480290.0,
      "step": 1391
    },
    {
      "epoch": 1.8998122546509644,
      "grad_norm": 0.11141517758369446,
      "learning_rate": 1.5311194552606067e-05,
      "loss": 0.4363,
      "num_tokens": 1102342832.0,
      "step": 1392
    },
    {
      "epoch": 1.9011776753712237,
      "grad_norm": 0.09969846904277802,
      "learning_rate": 1.528658888787628e-05,
      "loss": 0.4398,
      "num_tokens": 1103150856.0,
      "step": 1393
    },
    {
      "epoch": 1.9025430960914833,
      "grad_norm": 0.09772596508264542,
      "learning_rate": 1.5261997786493547e-05,
      "loss": 0.43,
      "num_tokens": 1103992434.0,
      "step": 1394
    },
    {
      "epoch": 1.9039085168117427,
      "grad_norm": 0.1031450554728508,
      "learning_rate": 1.5237421301803136e-05,
      "loss": 0.4256,
      "num_tokens": 1104776554.0,
      "step": 1395
    },
    {
      "epoch": 1.905273937532002,
      "grad_norm": 0.10192961245775223,
      "learning_rate": 1.521285948711861e-05,
      "loss": 0.4258,
      "num_tokens": 1105572947.0,
      "step": 1396
    },
    {
      "epoch": 1.9066393582522614,
      "grad_norm": 0.09039384126663208,
      "learning_rate": 1.5188312395721712e-05,
      "loss": 0.424,
      "num_tokens": 1106400789.0,
      "step": 1397
    },
    {
      "epoch": 1.9080047789725207,
      "grad_norm": 0.10476740449666977,
      "learning_rate": 1.5163780080862247e-05,
      "loss": 0.4214,
      "num_tokens": 1107197584.0,
      "step": 1398
    },
    {
      "epoch": 1.9093701996927803,
      "grad_norm": 0.09453436732292175,
      "learning_rate": 1.5139262595757954e-05,
      "loss": 0.4082,
      "num_tokens": 1107991173.0,
      "step": 1399
    },
    {
      "epoch": 1.91073562041304,
      "grad_norm": 0.10254891961812973,
      "learning_rate": 1.5114759993594419e-05,
      "loss": 0.4341,
      "num_tokens": 1108789190.0,
      "step": 1400
    },
    {
      "epoch": 1.9121010411332993,
      "grad_norm": 0.10639569908380508,
      "learning_rate": 1.5090272327524935e-05,
      "loss": 0.4254,
      "num_tokens": 1109622837.0,
      "step": 1401
    },
    {
      "epoch": 1.9134664618535586,
      "grad_norm": 0.10742893815040588,
      "learning_rate": 1.5065799650670383e-05,
      "loss": 0.3961,
      "num_tokens": 1110366039.0,
      "step": 1402
    },
    {
      "epoch": 1.914831882573818,
      "grad_norm": 0.09521877765655518,
      "learning_rate": 1.5041342016119145e-05,
      "loss": 0.4267,
      "num_tokens": 1111197489.0,
      "step": 1403
    },
    {
      "epoch": 1.9161973032940773,
      "grad_norm": 0.12256337702274323,
      "learning_rate": 1.5016899476926962e-05,
      "loss": 0.4386,
      "num_tokens": 1111995722.0,
      "step": 1404
    },
    {
      "epoch": 1.917562724014337,
      "grad_norm": 0.1161620020866394,
      "learning_rate": 1.4992472086116828e-05,
      "loss": 0.4191,
      "num_tokens": 1112746863.0,
      "step": 1405
    },
    {
      "epoch": 1.9189281447345965,
      "grad_norm": 0.10387642681598663,
      "learning_rate": 1.4968059896678898e-05,
      "loss": 0.4435,
      "num_tokens": 1113560910.0,
      "step": 1406
    },
    {
      "epoch": 1.9202935654548559,
      "grad_norm": 0.10705823451280594,
      "learning_rate": 1.4943662961570304e-05,
      "loss": 0.4529,
      "num_tokens": 1114384817.0,
      "step": 1407
    },
    {
      "epoch": 1.9216589861751152,
      "grad_norm": 0.11539414525032043,
      "learning_rate": 1.4919281333715136e-05,
      "loss": 0.4148,
      "num_tokens": 1115141182.0,
      "step": 1408
    },
    {
      "epoch": 1.9230244068953746,
      "grad_norm": 0.10361826419830322,
      "learning_rate": 1.489491506600424e-05,
      "loss": 0.41,
      "num_tokens": 1115905625.0,
      "step": 1409
    },
    {
      "epoch": 1.924389827615634,
      "grad_norm": 0.09953330457210541,
      "learning_rate": 1.487056421129518e-05,
      "loss": 0.4173,
      "num_tokens": 1116731071.0,
      "step": 1410
    },
    {
      "epoch": 1.9257552483358935,
      "grad_norm": 0.10519306361675262,
      "learning_rate": 1.4846228822412046e-05,
      "loss": 0.414,
      "num_tokens": 1117529232.0,
      "step": 1411
    },
    {
      "epoch": 1.927120669056153,
      "grad_norm": 0.10571619123220444,
      "learning_rate": 1.4821908952145409e-05,
      "loss": 0.4199,
      "num_tokens": 1118303056.0,
      "step": 1412
    },
    {
      "epoch": 1.9284860897764124,
      "grad_norm": 0.09096406400203705,
      "learning_rate": 1.4797604653252142e-05,
      "loss": 0.4147,
      "num_tokens": 1119127789.0,
      "step": 1413
    },
    {
      "epoch": 1.9298515104966718,
      "grad_norm": 0.10239739716053009,
      "learning_rate": 1.477331597845538e-05,
      "loss": 0.4006,
      "num_tokens": 1119863146.0,
      "step": 1414
    },
    {
      "epoch": 1.9312169312169312,
      "grad_norm": 0.09798164665699005,
      "learning_rate": 1.4749042980444332e-05,
      "loss": 0.4351,
      "num_tokens": 1120643766.0,
      "step": 1415
    },
    {
      "epoch": 1.9325823519371905,
      "grad_norm": 0.110104501247406,
      "learning_rate": 1.4724785711874208e-05,
      "loss": 0.4394,
      "num_tokens": 1121446634.0,
      "step": 1416
    },
    {
      "epoch": 1.93394777265745,
      "grad_norm": 0.10180757194757462,
      "learning_rate": 1.4700544225366114e-05,
      "loss": 0.4259,
      "num_tokens": 1122153577.0,
      "step": 1417
    },
    {
      "epoch": 1.9353131933777095,
      "grad_norm": 0.09743893891572952,
      "learning_rate": 1.4676318573506896e-05,
      "loss": 0.4323,
      "num_tokens": 1122982545.0,
      "step": 1418
    },
    {
      "epoch": 1.936678614097969,
      "grad_norm": 0.0992962047457695,
      "learning_rate": 1.465210880884906e-05,
      "loss": 0.4168,
      "num_tokens": 1123741571.0,
      "step": 1419
    },
    {
      "epoch": 1.9380440348182284,
      "grad_norm": 0.10541314631700516,
      "learning_rate": 1.4627914983910643e-05,
      "loss": 0.4309,
      "num_tokens": 1124493814.0,
      "step": 1420
    },
    {
      "epoch": 1.9394094555384878,
      "grad_norm": 0.09932273626327515,
      "learning_rate": 1.4603737151175122e-05,
      "loss": 0.4293,
      "num_tokens": 1125295050.0,
      "step": 1421
    },
    {
      "epoch": 1.9407748762587471,
      "grad_norm": 0.09662323445081711,
      "learning_rate": 1.4579575363091252e-05,
      "loss": 0.4046,
      "num_tokens": 1126090909.0,
      "step": 1422
    },
    {
      "epoch": 1.9421402969790067,
      "grad_norm": 0.09414529800415039,
      "learning_rate": 1.4555429672073014e-05,
      "loss": 0.4386,
      "num_tokens": 1126965037.0,
      "step": 1423
    },
    {
      "epoch": 1.943505717699266,
      "grad_norm": 0.10634944587945938,
      "learning_rate": 1.4531300130499455e-05,
      "loss": 0.4373,
      "num_tokens": 1127809105.0,
      "step": 1424
    },
    {
      "epoch": 1.9448711384195256,
      "grad_norm": 0.09423484653234482,
      "learning_rate": 1.4507186790714581e-05,
      "loss": 0.424,
      "num_tokens": 1128581251.0,
      "step": 1425
    },
    {
      "epoch": 1.946236559139785,
      "grad_norm": 0.09344486147165298,
      "learning_rate": 1.4483089705027267e-05,
      "loss": 0.4225,
      "num_tokens": 1129368507.0,
      "step": 1426
    },
    {
      "epoch": 1.9476019798600444,
      "grad_norm": 0.10422907024621964,
      "learning_rate": 1.4459008925711104e-05,
      "loss": 0.422,
      "num_tokens": 1130109086.0,
      "step": 1427
    },
    {
      "epoch": 1.9489674005803037,
      "grad_norm": 0.09968835115432739,
      "learning_rate": 1.4434944505004355e-05,
      "loss": 0.4217,
      "num_tokens": 1130920279.0,
      "step": 1428
    },
    {
      "epoch": 1.9503328213005633,
      "grad_norm": 0.09430572390556335,
      "learning_rate": 1.4410896495109752e-05,
      "loss": 0.413,
      "num_tokens": 1131734499.0,
      "step": 1429
    },
    {
      "epoch": 1.9516982420208226,
      "grad_norm": 0.09552576392889023,
      "learning_rate": 1.4386864948194447e-05,
      "loss": 0.4387,
      "num_tokens": 1132580298.0,
      "step": 1430
    },
    {
      "epoch": 1.9530636627410822,
      "grad_norm": 0.10360225290060043,
      "learning_rate": 1.4362849916389867e-05,
      "loss": 0.4247,
      "num_tokens": 1133365599.0,
      "step": 1431
    },
    {
      "epoch": 1.9544290834613416,
      "grad_norm": 0.10563445836305618,
      "learning_rate": 1.4338851451791631e-05,
      "loss": 0.4581,
      "num_tokens": 1134183758.0,
      "step": 1432
    },
    {
      "epoch": 1.955794504181601,
      "grad_norm": 0.0875878781080246,
      "learning_rate": 1.4314869606459408e-05,
      "loss": 0.4382,
      "num_tokens": 1135048421.0,
      "step": 1433
    },
    {
      "epoch": 1.9571599249018603,
      "grad_norm": 0.10272813588380814,
      "learning_rate": 1.4290904432416813e-05,
      "loss": 0.4198,
      "num_tokens": 1135754552.0,
      "step": 1434
    },
    {
      "epoch": 1.9585253456221197,
      "grad_norm": 0.0880967453122139,
      "learning_rate": 1.42669559816513e-05,
      "loss": 0.4161,
      "num_tokens": 1136541054.0,
      "step": 1435
    },
    {
      "epoch": 1.9598907663423792,
      "grad_norm": 0.09873216599225998,
      "learning_rate": 1.4243024306114048e-05,
      "loss": 0.4026,
      "num_tokens": 1137292289.0,
      "step": 1436
    },
    {
      "epoch": 1.9612561870626388,
      "grad_norm": 0.09773841500282288,
      "learning_rate": 1.4219109457719839e-05,
      "loss": 0.4278,
      "num_tokens": 1138155806.0,
      "step": 1437
    },
    {
      "epoch": 1.9626216077828982,
      "grad_norm": 0.10267940908670425,
      "learning_rate": 1.4195211488346953e-05,
      "loss": 0.4239,
      "num_tokens": 1138982292.0,
      "step": 1438
    },
    {
      "epoch": 1.9639870285031575,
      "grad_norm": 0.09426815807819366,
      "learning_rate": 1.4171330449837063e-05,
      "loss": 0.4262,
      "num_tokens": 1139809406.0,
      "step": 1439
    },
    {
      "epoch": 1.965352449223417,
      "grad_norm": 0.09345131367444992,
      "learning_rate": 1.4147466393995109e-05,
      "loss": 0.4266,
      "num_tokens": 1140647980.0,
      "step": 1440
    },
    {
      "epoch": 1.9667178699436763,
      "grad_norm": 0.09879738092422485,
      "learning_rate": 1.412361937258917e-05,
      "loss": 0.4287,
      "num_tokens": 1141389332.0,
      "step": 1441
    },
    {
      "epoch": 1.9680832906639358,
      "grad_norm": 0.1022033840417862,
      "learning_rate": 1.4099789437350424e-05,
      "loss": 0.4121,
      "num_tokens": 1142186348.0,
      "step": 1442
    },
    {
      "epoch": 1.9694487113841954,
      "grad_norm": 0.10338587313890457,
      "learning_rate": 1.4075976639972928e-05,
      "loss": 0.412,
      "num_tokens": 1142893343.0,
      "step": 1443
    },
    {
      "epoch": 1.9708141321044548,
      "grad_norm": 0.10997879505157471,
      "learning_rate": 1.4052181032113596e-05,
      "loss": 0.4413,
      "num_tokens": 1143720657.0,
      "step": 1444
    },
    {
      "epoch": 1.9721795528247141,
      "grad_norm": 0.09204406291246414,
      "learning_rate": 1.4028402665392027e-05,
      "loss": 0.4159,
      "num_tokens": 1144589859.0,
      "step": 1445
    },
    {
      "epoch": 1.9735449735449735,
      "grad_norm": 0.09977439790964127,
      "learning_rate": 1.4004641591390455e-05,
      "loss": 0.4235,
      "num_tokens": 1145370654.0,
      "step": 1446
    },
    {
      "epoch": 1.9749103942652328,
      "grad_norm": 0.10130681842565536,
      "learning_rate": 1.3980897861653575e-05,
      "loss": 0.4038,
      "num_tokens": 1146127463.0,
      "step": 1447
    },
    {
      "epoch": 1.9762758149854924,
      "grad_norm": 0.09759507328271866,
      "learning_rate": 1.3957171527688466e-05,
      "loss": 0.4211,
      "num_tokens": 1146922781.0,
      "step": 1448
    },
    {
      "epoch": 1.9776412357057518,
      "grad_norm": 0.11624299734830856,
      "learning_rate": 1.393346264096445e-05,
      "loss": 0.429,
      "num_tokens": 1147702644.0,
      "step": 1449
    },
    {
      "epoch": 1.9790066564260114,
      "grad_norm": 0.0953071117401123,
      "learning_rate": 1.3909771252913038e-05,
      "loss": 0.3937,
      "num_tokens": 1148484245.0,
      "step": 1450
    },
    {
      "epoch": 1.9803720771462707,
      "grad_norm": 0.09296579658985138,
      "learning_rate": 1.3886097414927754e-05,
      "loss": 0.4371,
      "num_tokens": 1149298374.0,
      "step": 1451
    },
    {
      "epoch": 1.98173749786653,
      "grad_norm": 0.10784582793712616,
      "learning_rate": 1.386244117836406e-05,
      "loss": 0.4349,
      "num_tokens": 1150101704.0,
      "step": 1452
    },
    {
      "epoch": 1.9831029185867894,
      "grad_norm": 0.10239841043949127,
      "learning_rate": 1.3838802594539228e-05,
      "loss": 0.4262,
      "num_tokens": 1150912652.0,
      "step": 1453
    },
    {
      "epoch": 1.984468339307049,
      "grad_norm": 0.08905389904975891,
      "learning_rate": 1.3815181714732251e-05,
      "loss": 0.4193,
      "num_tokens": 1151705083.0,
      "step": 1454
    },
    {
      "epoch": 1.9858337600273084,
      "grad_norm": 0.09325070679187775,
      "learning_rate": 1.3791578590183701e-05,
      "loss": 0.4379,
      "num_tokens": 1152570247.0,
      "step": 1455
    },
    {
      "epoch": 1.987199180747568,
      "grad_norm": 0.09890599548816681,
      "learning_rate": 1.3767993272095643e-05,
      "loss": 0.4244,
      "num_tokens": 1153286126.0,
      "step": 1456
    },
    {
      "epoch": 1.9885646014678273,
      "grad_norm": 0.09974310547113419,
      "learning_rate": 1.3744425811631512e-05,
      "loss": 0.43,
      "num_tokens": 1154062224.0,
      "step": 1457
    },
    {
      "epoch": 1.9899300221880867,
      "grad_norm": 0.10331180691719055,
      "learning_rate": 1.3720876259915996e-05,
      "loss": 0.4282,
      "num_tokens": 1154896258.0,
      "step": 1458
    },
    {
      "epoch": 1.991295442908346,
      "grad_norm": 0.10293087363243103,
      "learning_rate": 1.3697344668034967e-05,
      "loss": 0.424,
      "num_tokens": 1155692079.0,
      "step": 1459
    },
    {
      "epoch": 1.9926608636286056,
      "grad_norm": 0.09727304428815842,
      "learning_rate": 1.3673831087035286e-05,
      "loss": 0.4204,
      "num_tokens": 1156490029.0,
      "step": 1460
    },
    {
      "epoch": 1.994026284348865,
      "grad_norm": 0.0941903218626976,
      "learning_rate": 1.3650335567924779e-05,
      "loss": 0.4144,
      "num_tokens": 1157269822.0,
      "step": 1461
    },
    {
      "epoch": 1.9953917050691246,
      "grad_norm": 0.09818943589925766,
      "learning_rate": 1.3626858161672074e-05,
      "loss": 0.408,
      "num_tokens": 1158094009.0,
      "step": 1462
    },
    {
      "epoch": 1.996757125789384,
      "grad_norm": 0.10969363898038864,
      "learning_rate": 1.3603398919206523e-05,
      "loss": 0.4228,
      "num_tokens": 1158893953.0,
      "step": 1463
    },
    {
      "epoch": 1.9981225465096433,
      "grad_norm": 0.0924169048666954,
      "learning_rate": 1.3579957891418068e-05,
      "loss": 0.4253,
      "num_tokens": 1159659316.0,
      "step": 1464
    },
    {
      "epoch": 1.9994879672299026,
      "grad_norm": 0.09941767901182175,
      "learning_rate": 1.3556535129157138e-05,
      "loss": 0.4091,
      "num_tokens": 1160383886.0,
      "step": 1465
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.09941767901182175,
      "learning_rate": 1.3533130683234515e-05,
      "loss": 0.4191,
      "num_tokens": 1160565010.0,
      "step": 1466
    },
    {
      "epoch": 2.0013654207202594,
      "grad_norm": 0.186760812997818,
      "learning_rate": 1.3509744604421289e-05,
      "loss": 0.411,
      "num_tokens": 1161377956.0,
      "step": 1467
    },
    {
      "epoch": 2.0027308414405187,
      "grad_norm": 0.10184840857982635,
      "learning_rate": 1.3486376943448686e-05,
      "loss": 0.3953,
      "num_tokens": 1162182485.0,
      "step": 1468
    },
    {
      "epoch": 2.004096262160778,
      "grad_norm": 0.10738896578550339,
      "learning_rate": 1.3463027751007975e-05,
      "loss": 0.3957,
      "num_tokens": 1162949504.0,
      "step": 1469
    },
    {
      "epoch": 2.005461682881038,
      "grad_norm": 0.10919803380966187,
      "learning_rate": 1.3439697077750367e-05,
      "loss": 0.4034,
      "num_tokens": 1163748131.0,
      "step": 1470
    },
    {
      "epoch": 2.0068271036012972,
      "grad_norm": 0.09961043298244476,
      "learning_rate": 1.3416384974286895e-05,
      "loss": 0.4091,
      "num_tokens": 1164541124.0,
      "step": 1471
    },
    {
      "epoch": 2.0081925243215566,
      "grad_norm": 0.10359957814216614,
      "learning_rate": 1.3393091491188317e-05,
      "loss": 0.3833,
      "num_tokens": 1165317960.0,
      "step": 1472
    },
    {
      "epoch": 2.009557945041816,
      "grad_norm": 0.10492849349975586,
      "learning_rate": 1.3369816678984986e-05,
      "loss": 0.3953,
      "num_tokens": 1166135026.0,
      "step": 1473
    },
    {
      "epoch": 2.0109233657620753,
      "grad_norm": 0.10861695557832718,
      "learning_rate": 1.3346560588166764e-05,
      "loss": 0.4253,
      "num_tokens": 1167000518.0,
      "step": 1474
    },
    {
      "epoch": 2.0122887864823347,
      "grad_norm": 0.10022366791963577,
      "learning_rate": 1.3323323269182891e-05,
      "loss": 0.3838,
      "num_tokens": 1167778823.0,
      "step": 1475
    },
    {
      "epoch": 2.0136542072025945,
      "grad_norm": 0.10842650383710861,
      "learning_rate": 1.3300104772441897e-05,
      "loss": 0.4262,
      "num_tokens": 1168618434.0,
      "step": 1476
    },
    {
      "epoch": 2.015019627922854,
      "grad_norm": 0.09651152789592743,
      "learning_rate": 1.3276905148311481e-05,
      "loss": 0.408,
      "num_tokens": 1169400262.0,
      "step": 1477
    },
    {
      "epoch": 2.016385048643113,
      "grad_norm": 0.10497045516967773,
      "learning_rate": 1.325372444711838e-05,
      "loss": 0.4129,
      "num_tokens": 1170260551.0,
      "step": 1478
    },
    {
      "epoch": 2.0177504693633725,
      "grad_norm": 0.10679692775011063,
      "learning_rate": 1.3230562719148292e-05,
      "loss": 0.3932,
      "num_tokens": 1170997996.0,
      "step": 1479
    },
    {
      "epoch": 2.019115890083632,
      "grad_norm": 0.09802696108818054,
      "learning_rate": 1.320742001464578e-05,
      "loss": 0.397,
      "num_tokens": 1171705341.0,
      "step": 1480
    },
    {
      "epoch": 2.0204813108038913,
      "grad_norm": 0.10369493812322617,
      "learning_rate": 1.3184296383814114e-05,
      "loss": 0.3904,
      "num_tokens": 1172436562.0,
      "step": 1481
    },
    {
      "epoch": 2.021846731524151,
      "grad_norm": 0.10279261320829391,
      "learning_rate": 1.3161191876815194e-05,
      "loss": 0.414,
      "num_tokens": 1173298027.0,
      "step": 1482
    },
    {
      "epoch": 2.0232121522444104,
      "grad_norm": 0.10075166821479797,
      "learning_rate": 1.3138106543769433e-05,
      "loss": 0.4178,
      "num_tokens": 1174097622.0,
      "step": 1483
    },
    {
      "epoch": 2.02457757296467,
      "grad_norm": 0.102852001786232,
      "learning_rate": 1.3115040434755656e-05,
      "loss": 0.3921,
      "num_tokens": 1174801389.0,
      "step": 1484
    },
    {
      "epoch": 2.025942993684929,
      "grad_norm": 0.10267363488674164,
      "learning_rate": 1.309199359981098e-05,
      "loss": 0.3984,
      "num_tokens": 1175644296.0,
      "step": 1485
    },
    {
      "epoch": 2.0273084144051885,
      "grad_norm": 0.09447620809078217,
      "learning_rate": 1.3068966088930713e-05,
      "loss": 0.4134,
      "num_tokens": 1176456618.0,
      "step": 1486
    },
    {
      "epoch": 2.028673835125448,
      "grad_norm": 0.10669664293527603,
      "learning_rate": 1.3045957952068242e-05,
      "loss": 0.4222,
      "num_tokens": 1177271381.0,
      "step": 1487
    },
    {
      "epoch": 2.0300392558457077,
      "grad_norm": 0.10022611916065216,
      "learning_rate": 1.3022969239134927e-05,
      "loss": 0.422,
      "num_tokens": 1178072478.0,
      "step": 1488
    },
    {
      "epoch": 2.031404676565967,
      "grad_norm": 0.1078900545835495,
      "learning_rate": 1.3000000000000006e-05,
      "loss": 0.4293,
      "num_tokens": 1178929939.0,
      "step": 1489
    },
    {
      "epoch": 2.0327700972862264,
      "grad_norm": 0.101734958589077,
      "learning_rate": 1.297705028449044e-05,
      "loss": 0.3993,
      "num_tokens": 1179775652.0,
      "step": 1490
    },
    {
      "epoch": 2.0341355180064857,
      "grad_norm": 0.09091786295175552,
      "learning_rate": 1.2954120142390863e-05,
      "loss": 0.4012,
      "num_tokens": 1180579777.0,
      "step": 1491
    },
    {
      "epoch": 2.035500938726745,
      "grad_norm": 0.10008620470762253,
      "learning_rate": 1.293120962344344e-05,
      "loss": 0.4012,
      "num_tokens": 1181397450.0,
      "step": 1492
    },
    {
      "epoch": 2.0368663594470044,
      "grad_norm": 0.09611472487449646,
      "learning_rate": 1.2908318777347778e-05,
      "loss": 0.4259,
      "num_tokens": 1182211514.0,
      "step": 1493
    },
    {
      "epoch": 2.0382317801672643,
      "grad_norm": 0.10578350722789764,
      "learning_rate": 1.2885447653760804e-05,
      "loss": 0.3967,
      "num_tokens": 1182958855.0,
      "step": 1494
    },
    {
      "epoch": 2.0395972008875236,
      "grad_norm": 0.09627925604581833,
      "learning_rate": 1.2862596302296635e-05,
      "loss": 0.392,
      "num_tokens": 1183729426.0,
      "step": 1495
    },
    {
      "epoch": 2.040962621607783,
      "grad_norm": 0.09960373491048813,
      "learning_rate": 1.2839764772526527e-05,
      "loss": 0.3837,
      "num_tokens": 1184508441.0,
      "step": 1496
    },
    {
      "epoch": 2.0423280423280423,
      "grad_norm": 0.09442532062530518,
      "learning_rate": 1.2816953113978738e-05,
      "loss": 0.3869,
      "num_tokens": 1185245526.0,
      "step": 1497
    },
    {
      "epoch": 2.0436934630483017,
      "grad_norm": 0.10098136961460114,
      "learning_rate": 1.2794161376138403e-05,
      "loss": 0.4016,
      "num_tokens": 1186010738.0,
      "step": 1498
    },
    {
      "epoch": 2.045058883768561,
      "grad_norm": 0.09880411624908447,
      "learning_rate": 1.2771389608447452e-05,
      "loss": 0.3969,
      "num_tokens": 1186795066.0,
      "step": 1499
    },
    {
      "epoch": 2.0464243044888204,
      "grad_norm": 0.09314557909965515,
      "learning_rate": 1.2748637860304486e-05,
      "loss": 0.4033,
      "num_tokens": 1187561632.0,
      "step": 1500
    },
    {
      "epoch": 2.04778972520908,
      "grad_norm": 0.10952898859977722,
      "learning_rate": 1.2725906181064693e-05,
      "loss": 0.419,
      "num_tokens": 1188476085.0,
      "step": 1501
    },
    {
      "epoch": 2.0491551459293396,
      "grad_norm": 0.08929389715194702,
      "learning_rate": 1.2703194620039706e-05,
      "loss": 0.4099,
      "num_tokens": 1189306269.0,
      "step": 1502
    },
    {
      "epoch": 2.050520566649599,
      "grad_norm": 0.10028892755508423,
      "learning_rate": 1.2680503226497533e-05,
      "loss": 0.4254,
      "num_tokens": 1190136857.0,
      "step": 1503
    },
    {
      "epoch": 2.0518859873698583,
      "grad_norm": 0.10909068584442139,
      "learning_rate": 1.2657832049662426e-05,
      "loss": 0.3939,
      "num_tokens": 1190899604.0,
      "step": 1504
    },
    {
      "epoch": 2.0532514080901176,
      "grad_norm": 0.10118990391492844,
      "learning_rate": 1.2635181138714771e-05,
      "loss": 0.3977,
      "num_tokens": 1191671835.0,
      "step": 1505
    },
    {
      "epoch": 2.054616828810377,
      "grad_norm": 0.09613123536109924,
      "learning_rate": 1.261255054279102e-05,
      "loss": 0.3856,
      "num_tokens": 1192452454.0,
      "step": 1506
    },
    {
      "epoch": 2.055982249530637,
      "grad_norm": 0.09849090129137039,
      "learning_rate": 1.2589940310983519e-05,
      "loss": 0.3894,
      "num_tokens": 1193201420.0,
      "step": 1507
    },
    {
      "epoch": 2.057347670250896,
      "grad_norm": 0.10794316977262497,
      "learning_rate": 1.2567350492340466e-05,
      "loss": 0.4112,
      "num_tokens": 1193980512.0,
      "step": 1508
    },
    {
      "epoch": 2.0587130909711555,
      "grad_norm": 0.09509319812059402,
      "learning_rate": 1.2544781135865755e-05,
      "loss": 0.4128,
      "num_tokens": 1194783119.0,
      "step": 1509
    },
    {
      "epoch": 2.060078511691415,
      "grad_norm": 0.10389722883701324,
      "learning_rate": 1.2522232290518924e-05,
      "loss": 0.3936,
      "num_tokens": 1195551226.0,
      "step": 1510
    },
    {
      "epoch": 2.0614439324116742,
      "grad_norm": 0.09738654643297195,
      "learning_rate": 1.2499704005214983e-05,
      "loss": 0.4301,
      "num_tokens": 1196349889.0,
      "step": 1511
    },
    {
      "epoch": 2.0628093531319336,
      "grad_norm": 0.09773007780313492,
      "learning_rate": 1.2477196328824364e-05,
      "loss": 0.421,
      "num_tokens": 1197304362.0,
      "step": 1512
    },
    {
      "epoch": 2.0641747738521934,
      "grad_norm": 0.09242887049913406,
      "learning_rate": 1.2454709310172763e-05,
      "loss": 0.3959,
      "num_tokens": 1198117058.0,
      "step": 1513
    },
    {
      "epoch": 2.0655401945724527,
      "grad_norm": 0.09369007498025894,
      "learning_rate": 1.2432242998041101e-05,
      "loss": 0.3908,
      "num_tokens": 1198910387.0,
      "step": 1514
    },
    {
      "epoch": 2.066905615292712,
      "grad_norm": 0.09117814153432846,
      "learning_rate": 1.2409797441165363e-05,
      "loss": 0.3956,
      "num_tokens": 1199701646.0,
      "step": 1515
    },
    {
      "epoch": 2.0682710360129715,
      "grad_norm": 0.09868895262479782,
      "learning_rate": 1.2387372688236499e-05,
      "loss": 0.4124,
      "num_tokens": 1200499867.0,
      "step": 1516
    },
    {
      "epoch": 2.069636456733231,
      "grad_norm": 0.09686566144227982,
      "learning_rate": 1.2364968787900344e-05,
      "loss": 0.4024,
      "num_tokens": 1201291264.0,
      "step": 1517
    },
    {
      "epoch": 2.07100187745349,
      "grad_norm": 0.10280191153287888,
      "learning_rate": 1.2342585788757492e-05,
      "loss": 0.4141,
      "num_tokens": 1202090256.0,
      "step": 1518
    },
    {
      "epoch": 2.07236729817375,
      "grad_norm": 0.09863436222076416,
      "learning_rate": 1.2320223739363204e-05,
      "loss": 0.3897,
      "num_tokens": 1202781901.0,
      "step": 1519
    },
    {
      "epoch": 2.0737327188940093,
      "grad_norm": 0.09169047325849533,
      "learning_rate": 1.229788268822728e-05,
      "loss": 0.423,
      "num_tokens": 1203600558.0,
      "step": 1520
    },
    {
      "epoch": 2.0750981396142687,
      "grad_norm": 0.09922443330287933,
      "learning_rate": 1.2275562683813984e-05,
      "loss": 0.4103,
      "num_tokens": 1204405366.0,
      "step": 1521
    },
    {
      "epoch": 2.076463560334528,
      "grad_norm": 0.09708257019519806,
      "learning_rate": 1.2253263774541901e-05,
      "loss": 0.3931,
      "num_tokens": 1205196198.0,
      "step": 1522
    },
    {
      "epoch": 2.0778289810547874,
      "grad_norm": 0.09546215832233429,
      "learning_rate": 1.2230986008783891e-05,
      "loss": 0.4358,
      "num_tokens": 1205984404.0,
      "step": 1523
    },
    {
      "epoch": 2.0791944017750468,
      "grad_norm": 0.1052326112985611,
      "learning_rate": 1.2208729434866923e-05,
      "loss": 0.3894,
      "num_tokens": 1206712276.0,
      "step": 1524
    },
    {
      "epoch": 2.0805598224953066,
      "grad_norm": 0.09288535267114639,
      "learning_rate": 1.218649410107199e-05,
      "loss": 0.4175,
      "num_tokens": 1207531544.0,
      "step": 1525
    },
    {
      "epoch": 2.081925243215566,
      "grad_norm": 0.09636424481868744,
      "learning_rate": 1.216428005563401e-05,
      "loss": 0.3913,
      "num_tokens": 1208324596.0,
      "step": 1526
    },
    {
      "epoch": 2.0832906639358253,
      "grad_norm": 0.09829853475093842,
      "learning_rate": 1.2142087346741742e-05,
      "loss": 0.3966,
      "num_tokens": 1209076513.0,
      "step": 1527
    },
    {
      "epoch": 2.0846560846560847,
      "grad_norm": 0.09319709241390228,
      "learning_rate": 1.2119916022537644e-05,
      "loss": 0.4037,
      "num_tokens": 1209911090.0,
      "step": 1528
    },
    {
      "epoch": 2.086021505376344,
      "grad_norm": 0.09405887126922607,
      "learning_rate": 1.2097766131117781e-05,
      "loss": 0.4147,
      "num_tokens": 1210692356.0,
      "step": 1529
    },
    {
      "epoch": 2.0873869260966034,
      "grad_norm": 0.09597503393888474,
      "learning_rate": 1.207563772053173e-05,
      "loss": 0.4028,
      "num_tokens": 1211500251.0,
      "step": 1530
    },
    {
      "epoch": 2.0887523468168627,
      "grad_norm": 0.10275600850582123,
      "learning_rate": 1.2053530838782468e-05,
      "loss": 0.4026,
      "num_tokens": 1212266823.0,
      "step": 1531
    },
    {
      "epoch": 2.0901177675371225,
      "grad_norm": 0.09315111488103867,
      "learning_rate": 1.203144553382627e-05,
      "loss": 0.397,
      "num_tokens": 1213006548.0,
      "step": 1532
    },
    {
      "epoch": 2.091483188257382,
      "grad_norm": 0.09816258400678635,
      "learning_rate": 1.2009381853572608e-05,
      "loss": 0.4154,
      "num_tokens": 1213854037.0,
      "step": 1533
    },
    {
      "epoch": 2.0928486089776412,
      "grad_norm": 0.09696178138256073,
      "learning_rate": 1.1987339845884036e-05,
      "loss": 0.3965,
      "num_tokens": 1214593373.0,
      "step": 1534
    },
    {
      "epoch": 2.0942140296979006,
      "grad_norm": 0.09413173794746399,
      "learning_rate": 1.1965319558576103e-05,
      "loss": 0.4183,
      "num_tokens": 1215378962.0,
      "step": 1535
    },
    {
      "epoch": 2.09557945041816,
      "grad_norm": 0.10094869136810303,
      "learning_rate": 1.1943321039417234e-05,
      "loss": 0.4106,
      "num_tokens": 1216151621.0,
      "step": 1536
    },
    {
      "epoch": 2.0969448711384193,
      "grad_norm": 0.37168505787849426,
      "learning_rate": 1.1921344336128633e-05,
      "loss": 0.4379,
      "num_tokens": 1216980686.0,
      "step": 1537
    },
    {
      "epoch": 2.098310291858679,
      "grad_norm": 0.0974922627210617,
      "learning_rate": 1.189938949638418e-05,
      "loss": 0.3983,
      "num_tokens": 1217770716.0,
      "step": 1538
    },
    {
      "epoch": 2.0996757125789385,
      "grad_norm": 0.0992826372385025,
      "learning_rate": 1.187745656781032e-05,
      "loss": 0.4071,
      "num_tokens": 1218606727.0,
      "step": 1539
    },
    {
      "epoch": 2.101041133299198,
      "grad_norm": 0.09375310689210892,
      "learning_rate": 1.1855545597985989e-05,
      "loss": 0.4093,
      "num_tokens": 1219348248.0,
      "step": 1540
    },
    {
      "epoch": 2.102406554019457,
      "grad_norm": 0.10654960572719574,
      "learning_rate": 1.183365663444247e-05,
      "loss": 0.4013,
      "num_tokens": 1220160533.0,
      "step": 1541
    },
    {
      "epoch": 2.1037719747397166,
      "grad_norm": 0.09849007427692413,
      "learning_rate": 1.1811789724663298e-05,
      "loss": 0.3982,
      "num_tokens": 1220948277.0,
      "step": 1542
    },
    {
      "epoch": 2.105137395459976,
      "grad_norm": 0.09713562577962875,
      "learning_rate": 1.1789944916084177e-05,
      "loss": 0.4039,
      "num_tokens": 1221737008.0,
      "step": 1543
    },
    {
      "epoch": 2.1065028161802357,
      "grad_norm": 0.09663241356611252,
      "learning_rate": 1.1768122256092887e-05,
      "loss": 0.3943,
      "num_tokens": 1222490829.0,
      "step": 1544
    },
    {
      "epoch": 2.107868236900495,
      "grad_norm": 0.09749586135149002,
      "learning_rate": 1.174632179202913e-05,
      "loss": 0.3974,
      "num_tokens": 1223275614.0,
      "step": 1545
    },
    {
      "epoch": 2.1092336576207544,
      "grad_norm": 0.09357865154743195,
      "learning_rate": 1.1724543571184476e-05,
      "loss": 0.416,
      "num_tokens": 1224113420.0,
      "step": 1546
    },
    {
      "epoch": 2.110599078341014,
      "grad_norm": 0.10085227340459824,
      "learning_rate": 1.1702787640802238e-05,
      "loss": 0.3941,
      "num_tokens": 1224913037.0,
      "step": 1547
    },
    {
      "epoch": 2.111964499061273,
      "grad_norm": 0.09974703192710876,
      "learning_rate": 1.1681054048077378e-05,
      "loss": 0.427,
      "num_tokens": 1225756916.0,
      "step": 1548
    },
    {
      "epoch": 2.1133299197815325,
      "grad_norm": 0.09222148358821869,
      "learning_rate": 1.1659342840156394e-05,
      "loss": 0.4178,
      "num_tokens": 1226539614.0,
      "step": 1549
    },
    {
      "epoch": 2.1146953405017923,
      "grad_norm": 0.1035853922367096,
      "learning_rate": 1.163765406413723e-05,
      "loss": 0.3972,
      "num_tokens": 1227293759.0,
      "step": 1550
    },
    {
      "epoch": 2.1160607612220517,
      "grad_norm": 0.09526224434375763,
      "learning_rate": 1.1615987767069173e-05,
      "loss": 0.4076,
      "num_tokens": 1228115787.0,
      "step": 1551
    },
    {
      "epoch": 2.117426181942311,
      "grad_norm": 0.1010228842496872,
      "learning_rate": 1.1594343995952728e-05,
      "loss": 0.4229,
      "num_tokens": 1228847049.0,
      "step": 1552
    },
    {
      "epoch": 2.1187916026625704,
      "grad_norm": 0.1088263988494873,
      "learning_rate": 1.1572722797739572e-05,
      "loss": 0.4274,
      "num_tokens": 1229655500.0,
      "step": 1553
    },
    {
      "epoch": 2.1201570233828297,
      "grad_norm": 0.1025305911898613,
      "learning_rate": 1.1551124219332367e-05,
      "loss": 0.4006,
      "num_tokens": 1230409124.0,
      "step": 1554
    },
    {
      "epoch": 2.121522444103089,
      "grad_norm": 0.10329543799161911,
      "learning_rate": 1.1529548307584748e-05,
      "loss": 0.4276,
      "num_tokens": 1231159279.0,
      "step": 1555
    },
    {
      "epoch": 2.122887864823349,
      "grad_norm": 0.09708493947982788,
      "learning_rate": 1.150799510930114e-05,
      "loss": 0.4259,
      "num_tokens": 1232015199.0,
      "step": 1556
    },
    {
      "epoch": 2.1242532855436083,
      "grad_norm": 0.10361675918102264,
      "learning_rate": 1.1486464671236738e-05,
      "loss": 0.394,
      "num_tokens": 1232744565.0,
      "step": 1557
    },
    {
      "epoch": 2.1256187062638676,
      "grad_norm": 0.09690624475479126,
      "learning_rate": 1.1464957040097337e-05,
      "loss": 0.3828,
      "num_tokens": 1233520955.0,
      "step": 1558
    },
    {
      "epoch": 2.126984126984127,
      "grad_norm": 0.09309430420398712,
      "learning_rate": 1.144347226253927e-05,
      "loss": 0.3736,
      "num_tokens": 1234228944.0,
      "step": 1559
    },
    {
      "epoch": 2.1283495477043863,
      "grad_norm": 0.10124411433935165,
      "learning_rate": 1.1422010385169262e-05,
      "loss": 0.403,
      "num_tokens": 1235023266.0,
      "step": 1560
    },
    {
      "epoch": 2.1297149684246457,
      "grad_norm": 0.09351851046085358,
      "learning_rate": 1.1400571454544412e-05,
      "loss": 0.4335,
      "num_tokens": 1235882722.0,
      "step": 1561
    },
    {
      "epoch": 2.1310803891449055,
      "grad_norm": 0.09701912105083466,
      "learning_rate": 1.1379155517172e-05,
      "loss": 0.4014,
      "num_tokens": 1236660063.0,
      "step": 1562
    },
    {
      "epoch": 2.132445809865165,
      "grad_norm": 0.1032654196023941,
      "learning_rate": 1.1357762619509445e-05,
      "loss": 0.4005,
      "num_tokens": 1237456857.0,
      "step": 1563
    },
    {
      "epoch": 2.133811230585424,
      "grad_norm": 0.10620412230491638,
      "learning_rate": 1.1336392807964185e-05,
      "loss": 0.3824,
      "num_tokens": 1238189659.0,
      "step": 1564
    },
    {
      "epoch": 2.1351766513056836,
      "grad_norm": 0.09708909690380096,
      "learning_rate": 1.1315046128893571e-05,
      "loss": 0.4031,
      "num_tokens": 1238942355.0,
      "step": 1565
    },
    {
      "epoch": 2.136542072025943,
      "grad_norm": 0.09779118001461029,
      "learning_rate": 1.1293722628604782e-05,
      "loss": 0.3938,
      "num_tokens": 1239785025.0,
      "step": 1566
    },
    {
      "epoch": 2.1379074927462023,
      "grad_norm": 0.10045590996742249,
      "learning_rate": 1.1272422353354706e-05,
      "loss": 0.3949,
      "num_tokens": 1240514295.0,
      "step": 1567
    },
    {
      "epoch": 2.1392729134664616,
      "grad_norm": 0.10228996723890305,
      "learning_rate": 1.1251145349349855e-05,
      "loss": 0.4105,
      "num_tokens": 1241336593.0,
      "step": 1568
    },
    {
      "epoch": 2.1406383341867214,
      "grad_norm": 0.10208886861801147,
      "learning_rate": 1.1229891662746248e-05,
      "loss": 0.3998,
      "num_tokens": 1242211551.0,
      "step": 1569
    },
    {
      "epoch": 2.142003754906981,
      "grad_norm": 0.0933549702167511,
      "learning_rate": 1.1208661339649348e-05,
      "loss": 0.3978,
      "num_tokens": 1242913890.0,
      "step": 1570
    },
    {
      "epoch": 2.14336917562724,
      "grad_norm": 0.10039491951465607,
      "learning_rate": 1.1187454426113917e-05,
      "loss": 0.4209,
      "num_tokens": 1243810208.0,
      "step": 1571
    },
    {
      "epoch": 2.1447345963474995,
      "grad_norm": 0.0932595431804657,
      "learning_rate": 1.1166270968143923e-05,
      "loss": 0.4173,
      "num_tokens": 1244601601.0,
      "step": 1572
    },
    {
      "epoch": 2.146100017067759,
      "grad_norm": 0.0970836654305458,
      "learning_rate": 1.1145111011692471e-05,
      "loss": 0.3837,
      "num_tokens": 1245345849.0,
      "step": 1573
    },
    {
      "epoch": 2.1474654377880182,
      "grad_norm": 0.10047222673892975,
      "learning_rate": 1.1123974602661685e-05,
      "loss": 0.4189,
      "num_tokens": 1246156817.0,
      "step": 1574
    },
    {
      "epoch": 2.148830858508278,
      "grad_norm": 0.09958304464817047,
      "learning_rate": 1.1102861786902595e-05,
      "loss": 0.4023,
      "num_tokens": 1246964727.0,
      "step": 1575
    },
    {
      "epoch": 2.1501962792285374,
      "grad_norm": 0.09364857524633408,
      "learning_rate": 1.108177261021507e-05,
      "loss": 0.3945,
      "num_tokens": 1247775569.0,
      "step": 1576
    },
    {
      "epoch": 2.1515616999487968,
      "grad_norm": 0.09441335499286652,
      "learning_rate": 1.1060707118347662e-05,
      "loss": 0.4215,
      "num_tokens": 1248599657.0,
      "step": 1577
    },
    {
      "epoch": 2.152927120669056,
      "grad_norm": 0.10512730479240417,
      "learning_rate": 1.103966535699759e-05,
      "loss": 0.4079,
      "num_tokens": 1249392768.0,
      "step": 1578
    },
    {
      "epoch": 2.1542925413893155,
      "grad_norm": 0.1045491173863411,
      "learning_rate": 1.1018647371810562e-05,
      "loss": 0.4156,
      "num_tokens": 1250194838.0,
      "step": 1579
    },
    {
      "epoch": 2.155657962109575,
      "grad_norm": 0.10127020627260208,
      "learning_rate": 1.0997653208380723e-05,
      "loss": 0.4108,
      "num_tokens": 1251072381.0,
      "step": 1580
    },
    {
      "epoch": 2.1570233828298346,
      "grad_norm": 0.09860903024673462,
      "learning_rate": 1.0976682912250537e-05,
      "loss": 0.3812,
      "num_tokens": 1251828392.0,
      "step": 1581
    },
    {
      "epoch": 2.158388803550094,
      "grad_norm": 0.09554125368595123,
      "learning_rate": 1.0955736528910692e-05,
      "loss": 0.3891,
      "num_tokens": 1252549978.0,
      "step": 1582
    },
    {
      "epoch": 2.1597542242703534,
      "grad_norm": 0.0992458313703537,
      "learning_rate": 1.0934814103800004e-05,
      "loss": 0.4185,
      "num_tokens": 1253372571.0,
      "step": 1583
    },
    {
      "epoch": 2.1611196449906127,
      "grad_norm": 0.10090472549200058,
      "learning_rate": 1.0913915682305319e-05,
      "loss": 0.4202,
      "num_tokens": 1254250220.0,
      "step": 1584
    },
    {
      "epoch": 2.162485065710872,
      "grad_norm": 0.10184525698423386,
      "learning_rate": 1.0893041309761403e-05,
      "loss": 0.3914,
      "num_tokens": 1254999065.0,
      "step": 1585
    },
    {
      "epoch": 2.1638504864311314,
      "grad_norm": 0.09753662347793579,
      "learning_rate": 1.0872191031450861e-05,
      "loss": 0.3915,
      "num_tokens": 1255778563.0,
      "step": 1586
    },
    {
      "epoch": 2.1652159071513912,
      "grad_norm": 0.09987428784370422,
      "learning_rate": 1.0851364892604032e-05,
      "loss": 0.4194,
      "num_tokens": 1256500330.0,
      "step": 1587
    },
    {
      "epoch": 2.1665813278716506,
      "grad_norm": 0.10557276010513306,
      "learning_rate": 1.0830562938398893e-05,
      "loss": 0.407,
      "num_tokens": 1257358876.0,
      "step": 1588
    },
    {
      "epoch": 2.16794674859191,
      "grad_norm": 0.10442834347486496,
      "learning_rate": 1.080978521396094e-05,
      "loss": 0.4094,
      "num_tokens": 1258113679.0,
      "step": 1589
    },
    {
      "epoch": 2.1693121693121693,
      "grad_norm": 0.10067546367645264,
      "learning_rate": 1.0789031764363114e-05,
      "loss": 0.4054,
      "num_tokens": 1258887915.0,
      "step": 1590
    },
    {
      "epoch": 2.1706775900324287,
      "grad_norm": 0.1051938459277153,
      "learning_rate": 1.0768302634625716e-05,
      "loss": 0.4404,
      "num_tokens": 1259728618.0,
      "step": 1591
    },
    {
      "epoch": 2.172043010752688,
      "grad_norm": 0.09482062608003616,
      "learning_rate": 1.0747597869716265e-05,
      "loss": 0.4093,
      "num_tokens": 1260546418.0,
      "step": 1592
    },
    {
      "epoch": 2.1734084314729474,
      "grad_norm": 0.09855875372886658,
      "learning_rate": 1.072691751454944e-05,
      "loss": 0.4288,
      "num_tokens": 1261372269.0,
      "step": 1593
    },
    {
      "epoch": 2.174773852193207,
      "grad_norm": 0.10455437749624252,
      "learning_rate": 1.0706261613986961e-05,
      "loss": 0.3977,
      "num_tokens": 1262097079.0,
      "step": 1594
    },
    {
      "epoch": 2.1761392729134665,
      "grad_norm": 0.10095681250095367,
      "learning_rate": 1.0685630212837503e-05,
      "loss": 0.3888,
      "num_tokens": 1262868819.0,
      "step": 1595
    },
    {
      "epoch": 2.177504693633726,
      "grad_norm": 0.09713822603225708,
      "learning_rate": 1.0665023355856596e-05,
      "loss": 0.4152,
      "num_tokens": 1263651092.0,
      "step": 1596
    },
    {
      "epoch": 2.1788701143539853,
      "grad_norm": 0.10276846587657928,
      "learning_rate": 1.064444108774652e-05,
      "loss": 0.4178,
      "num_tokens": 1264404731.0,
      "step": 1597
    },
    {
      "epoch": 2.1802355350742446,
      "grad_norm": 0.10070840269327164,
      "learning_rate": 1.0623883453156218e-05,
      "loss": 0.4362,
      "num_tokens": 1265270335.0,
      "step": 1598
    },
    {
      "epoch": 2.1816009557945044,
      "grad_norm": 0.09889353066682816,
      "learning_rate": 1.0603350496681197e-05,
      "loss": 0.3796,
      "num_tokens": 1266096089.0,
      "step": 1599
    },
    {
      "epoch": 2.1829663765147638,
      "grad_norm": 0.09896312654018402,
      "learning_rate": 1.0582842262863442e-05,
      "loss": 0.4078,
      "num_tokens": 1266883383.0,
      "step": 1600
    },
    {
      "epoch": 2.184331797235023,
      "grad_norm": 0.09334592521190643,
      "learning_rate": 1.0562358796191274e-05,
      "loss": 0.4032,
      "num_tokens": 1267665521.0,
      "step": 1601
    },
    {
      "epoch": 2.1856972179552825,
      "grad_norm": 0.09068528562784195,
      "learning_rate": 1.0541900141099319e-05,
      "loss": 0.3801,
      "num_tokens": 1268426436.0,
      "step": 1602
    },
    {
      "epoch": 2.187062638675542,
      "grad_norm": 0.09869010001420975,
      "learning_rate": 1.052146634196836e-05,
      "loss": 0.3829,
      "num_tokens": 1269159212.0,
      "step": 1603
    },
    {
      "epoch": 2.188428059395801,
      "grad_norm": 0.10397112369537354,
      "learning_rate": 1.0501057443125277e-05,
      "loss": 0.391,
      "num_tokens": 1269970955.0,
      "step": 1604
    },
    {
      "epoch": 2.1897934801160606,
      "grad_norm": 0.09869799762964249,
      "learning_rate": 1.0480673488842923e-05,
      "loss": 0.3995,
      "num_tokens": 1270735994.0,
      "step": 1605
    },
    {
      "epoch": 2.1911589008363204,
      "grad_norm": 0.10335088521242142,
      "learning_rate": 1.0460314523340045e-05,
      "loss": 0.3889,
      "num_tokens": 1271505147.0,
      "step": 1606
    },
    {
      "epoch": 2.1925243215565797,
      "grad_norm": 0.08854160457849503,
      "learning_rate": 1.0439980590781161e-05,
      "loss": 0.4112,
      "num_tokens": 1272347644.0,
      "step": 1607
    },
    {
      "epoch": 2.193889742276839,
      "grad_norm": 0.09081967920064926,
      "learning_rate": 1.0419671735276518e-05,
      "loss": 0.3984,
      "num_tokens": 1273111001.0,
      "step": 1608
    },
    {
      "epoch": 2.1952551629970984,
      "grad_norm": 0.09805253893136978,
      "learning_rate": 1.0399388000881942e-05,
      "loss": 0.4169,
      "num_tokens": 1273905616.0,
      "step": 1609
    },
    {
      "epoch": 2.196620583717358,
      "grad_norm": 0.0982927531003952,
      "learning_rate": 1.0379129431598775e-05,
      "loss": 0.3922,
      "num_tokens": 1274679387.0,
      "step": 1610
    },
    {
      "epoch": 2.197986004437617,
      "grad_norm": 0.09682530909776688,
      "learning_rate": 1.0358896071373758e-05,
      "loss": 0.425,
      "num_tokens": 1275521234.0,
      "step": 1611
    },
    {
      "epoch": 2.199351425157877,
      "grad_norm": 0.10083907097578049,
      "learning_rate": 1.0338687964098953e-05,
      "loss": 0.3918,
      "num_tokens": 1276273212.0,
      "step": 1612
    },
    {
      "epoch": 2.2007168458781363,
      "grad_norm": 0.09620651602745056,
      "learning_rate": 1.0318505153611642e-05,
      "loss": 0.4007,
      "num_tokens": 1276998518.0,
      "step": 1613
    },
    {
      "epoch": 2.2020822665983957,
      "grad_norm": 0.09567209333181381,
      "learning_rate": 1.0298347683694224e-05,
      "loss": 0.4,
      "num_tokens": 1277792468.0,
      "step": 1614
    },
    {
      "epoch": 2.203447687318655,
      "grad_norm": 0.09632347524166107,
      "learning_rate": 1.0278215598074133e-05,
      "loss": 0.4259,
      "num_tokens": 1278599468.0,
      "step": 1615
    },
    {
      "epoch": 2.2048131080389144,
      "grad_norm": 0.10429361462593079,
      "learning_rate": 1.025810894042373e-05,
      "loss": 0.4095,
      "num_tokens": 1279392460.0,
      "step": 1616
    },
    {
      "epoch": 2.2061785287591738,
      "grad_norm": 0.08928142488002777,
      "learning_rate": 1.0238027754360238e-05,
      "loss": 0.4055,
      "num_tokens": 1280195113.0,
      "step": 1617
    },
    {
      "epoch": 2.2075439494794336,
      "grad_norm": 0.09861229360103607,
      "learning_rate": 1.0217972083445591e-05,
      "loss": 0.4102,
      "num_tokens": 1281009568.0,
      "step": 1618
    },
    {
      "epoch": 2.208909370199693,
      "grad_norm": 0.1006210669875145,
      "learning_rate": 1.0197941971186393e-05,
      "loss": 0.4126,
      "num_tokens": 1281778806.0,
      "step": 1619
    },
    {
      "epoch": 2.2102747909199523,
      "grad_norm": 0.09274760633707047,
      "learning_rate": 1.0177937461033798e-05,
      "loss": 0.4069,
      "num_tokens": 1282611524.0,
      "step": 1620
    },
    {
      "epoch": 2.2116402116402116,
      "grad_norm": 0.09846462309360504,
      "learning_rate": 1.0157958596383433e-05,
      "loss": 0.3891,
      "num_tokens": 1283340447.0,
      "step": 1621
    },
    {
      "epoch": 2.213005632360471,
      "grad_norm": 0.09642168134450912,
      "learning_rate": 1.0138005420575281e-05,
      "loss": 0.4248,
      "num_tokens": 1284151754.0,
      "step": 1622
    },
    {
      "epoch": 2.2143710530807303,
      "grad_norm": 0.09564360231161118,
      "learning_rate": 1.01180779768936e-05,
      "loss": 0.4257,
      "num_tokens": 1285004244.0,
      "step": 1623
    },
    {
      "epoch": 2.21573647380099,
      "grad_norm": 0.09428383409976959,
      "learning_rate": 1.0098176308566817e-05,
      "loss": 0.4062,
      "num_tokens": 1285760951.0,
      "step": 1624
    },
    {
      "epoch": 2.2171018945212495,
      "grad_norm": 0.0915636196732521,
      "learning_rate": 1.0078300458767467e-05,
      "loss": 0.3835,
      "num_tokens": 1286556350.0,
      "step": 1625
    },
    {
      "epoch": 2.218467315241509,
      "grad_norm": 0.09107786417007446,
      "learning_rate": 1.0058450470612068e-05,
      "loss": 0.4025,
      "num_tokens": 1287319034.0,
      "step": 1626
    },
    {
      "epoch": 2.219832735961768,
      "grad_norm": 0.09939936548471451,
      "learning_rate": 1.003862638716103e-05,
      "loss": 0.4127,
      "num_tokens": 1288110356.0,
      "step": 1627
    },
    {
      "epoch": 2.2211981566820276,
      "grad_norm": 0.09531965851783752,
      "learning_rate": 1.0018828251418576e-05,
      "loss": 0.3971,
      "num_tokens": 1288908202.0,
      "step": 1628
    },
    {
      "epoch": 2.222563577402287,
      "grad_norm": 0.09365642815828323,
      "learning_rate": 9.999056106332635e-06,
      "loss": 0.4187,
      "num_tokens": 1289765305.0,
      "step": 1629
    },
    {
      "epoch": 2.2239289981225463,
      "grad_norm": 0.09418322890996933,
      "learning_rate": 9.979309994794757e-06,
      "loss": 0.4036,
      "num_tokens": 1290581671.0,
      "step": 1630
    },
    {
      "epoch": 2.225294418842806,
      "grad_norm": 0.09542888402938843,
      "learning_rate": 9.959589959640021e-06,
      "loss": 0.4164,
      "num_tokens": 1291330271.0,
      "step": 1631
    },
    {
      "epoch": 2.2266598395630655,
      "grad_norm": 0.10436046123504639,
      "learning_rate": 9.939896043646932e-06,
      "loss": 0.4259,
      "num_tokens": 1292135959.0,
      "step": 1632
    },
    {
      "epoch": 2.228025260283325,
      "grad_norm": 0.09666141867637634,
      "learning_rate": 9.920228289537337e-06,
      "loss": 0.4168,
      "num_tokens": 1293014531.0,
      "step": 1633
    },
    {
      "epoch": 2.229390681003584,
      "grad_norm": 0.09237797558307648,
      "learning_rate": 9.90058673997634e-06,
      "loss": 0.4217,
      "num_tokens": 1293863410.0,
      "step": 1634
    },
    {
      "epoch": 2.2307561017238435,
      "grad_norm": 0.09510931372642517,
      "learning_rate": 9.880971437572194e-06,
      "loss": 0.3937,
      "num_tokens": 1294653550.0,
      "step": 1635
    },
    {
      "epoch": 2.232121522444103,
      "grad_norm": 0.09263675659894943,
      "learning_rate": 9.861382424876196e-06,
      "loss": 0.3964,
      "num_tokens": 1295400034.0,
      "step": 1636
    },
    {
      "epoch": 2.2334869431643627,
      "grad_norm": 0.09794460982084274,
      "learning_rate": 9.841819744382634e-06,
      "loss": 0.4108,
      "num_tokens": 1296147375.0,
      "step": 1637
    },
    {
      "epoch": 2.234852363884622,
      "grad_norm": 0.09825485944747925,
      "learning_rate": 9.822283438528677e-06,
      "loss": 0.3797,
      "num_tokens": 1296839819.0,
      "step": 1638
    },
    {
      "epoch": 2.2362177846048814,
      "grad_norm": 0.09483623504638672,
      "learning_rate": 9.802773549694267e-06,
      "loss": 0.4137,
      "num_tokens": 1297638888.0,
      "step": 1639
    },
    {
      "epoch": 2.2375832053251408,
      "grad_norm": 0.11944864690303802,
      "learning_rate": 9.783290120202043e-06,
      "loss": 0.4069,
      "num_tokens": 1298370118.0,
      "step": 1640
    },
    {
      "epoch": 2.2389486260454,
      "grad_norm": 0.10375841706991196,
      "learning_rate": 9.763833192317248e-06,
      "loss": 0.4038,
      "num_tokens": 1299173013.0,
      "step": 1641
    },
    {
      "epoch": 2.2403140467656595,
      "grad_norm": 0.09504275023937225,
      "learning_rate": 9.744402808247635e-06,
      "loss": 0.4137,
      "num_tokens": 1300039414.0,
      "step": 1642
    },
    {
      "epoch": 2.2416794674859193,
      "grad_norm": 0.10177642852067947,
      "learning_rate": 9.724999010143372e-06,
      "loss": 0.4018,
      "num_tokens": 1300829563.0,
      "step": 1643
    },
    {
      "epoch": 2.2430448882061786,
      "grad_norm": 0.09937194734811783,
      "learning_rate": 9.70562184009696e-06,
      "loss": 0.4186,
      "num_tokens": 1301608766.0,
      "step": 1644
    },
    {
      "epoch": 2.244410308926438,
      "grad_norm": 0.09489071369171143,
      "learning_rate": 9.686271340143132e-06,
      "loss": 0.4033,
      "num_tokens": 1302370100.0,
      "step": 1645
    },
    {
      "epoch": 2.2457757296466974,
      "grad_norm": 0.09602957963943481,
      "learning_rate": 9.666947552258764e-06,
      "loss": 0.3837,
      "num_tokens": 1303237479.0,
      "step": 1646
    },
    {
      "epoch": 2.2471411503669567,
      "grad_norm": 0.09830103069543839,
      "learning_rate": 9.647650518362799e-06,
      "loss": 0.4252,
      "num_tokens": 1304079944.0,
      "step": 1647
    },
    {
      "epoch": 2.248506571087216,
      "grad_norm": 0.09430351108312607,
      "learning_rate": 9.628380280316124e-06,
      "loss": 0.41,
      "num_tokens": 1304851869.0,
      "step": 1648
    },
    {
      "epoch": 2.249871991807476,
      "grad_norm": 0.09818066656589508,
      "learning_rate": 9.609136879921506e-06,
      "loss": 0.4074,
      "num_tokens": 1305663248.0,
      "step": 1649
    },
    {
      "epoch": 2.2512374125277352,
      "grad_norm": 0.09799886494874954,
      "learning_rate": 9.589920358923492e-06,
      "loss": 0.4281,
      "num_tokens": 1306472256.0,
      "step": 1650
    },
    {
      "epoch": 2.2526028332479946,
      "grad_norm": 0.09831734746694565,
      "learning_rate": 9.570730759008331e-06,
      "loss": 0.3918,
      "num_tokens": 1307281646.0,
      "step": 1651
    },
    {
      "epoch": 2.253968253968254,
      "grad_norm": 0.09582869708538055,
      "learning_rate": 9.551568121803859e-06,
      "loss": 0.3927,
      "num_tokens": 1308089995.0,
      "step": 1652
    },
    {
      "epoch": 2.2553336746885133,
      "grad_norm": 0.09847426414489746,
      "learning_rate": 9.53243248887943e-06,
      "loss": 0.4065,
      "num_tokens": 1308832199.0,
      "step": 1653
    },
    {
      "epoch": 2.2566990954087727,
      "grad_norm": 0.09598053246736526,
      "learning_rate": 9.5133239017458e-06,
      "loss": 0.41,
      "num_tokens": 1309604520.0,
      "step": 1654
    },
    {
      "epoch": 2.258064516129032,
      "grad_norm": 0.10019654035568237,
      "learning_rate": 9.494242401855081e-06,
      "loss": 0.4199,
      "num_tokens": 1310433095.0,
      "step": 1655
    },
    {
      "epoch": 2.259429936849292,
      "grad_norm": 0.09726449102163315,
      "learning_rate": 9.475188030600608e-06,
      "loss": 0.4256,
      "num_tokens": 1311197541.0,
      "step": 1656
    },
    {
      "epoch": 2.260795357569551,
      "grad_norm": 0.09670136123895645,
      "learning_rate": 9.456160829316872e-06,
      "loss": 0.3849,
      "num_tokens": 1311989075.0,
      "step": 1657
    },
    {
      "epoch": 2.2621607782898105,
      "grad_norm": 0.0928812175989151,
      "learning_rate": 9.437160839279422e-06,
      "loss": 0.398,
      "num_tokens": 1312776150.0,
      "step": 1658
    },
    {
      "epoch": 2.26352619901007,
      "grad_norm": 0.0954815223813057,
      "learning_rate": 9.418188101704779e-06,
      "loss": 0.4025,
      "num_tokens": 1313537368.0,
      "step": 1659
    },
    {
      "epoch": 2.2648916197303293,
      "grad_norm": 0.09644950181245804,
      "learning_rate": 9.399242657750346e-06,
      "loss": 0.405,
      "num_tokens": 1314334621.0,
      "step": 1660
    },
    {
      "epoch": 2.266257040450589,
      "grad_norm": 0.10169140994548798,
      "learning_rate": 9.380324548514316e-06,
      "loss": 0.4018,
      "num_tokens": 1315175054.0,
      "step": 1661
    },
    {
      "epoch": 2.2676224611708484,
      "grad_norm": 0.09833645820617676,
      "learning_rate": 9.361433815035588e-06,
      "loss": 0.393,
      "num_tokens": 1315941092.0,
      "step": 1662
    },
    {
      "epoch": 2.268987881891108,
      "grad_norm": 0.09277669340372086,
      "learning_rate": 9.342570498293668e-06,
      "loss": 0.3854,
      "num_tokens": 1316719695.0,
      "step": 1663
    },
    {
      "epoch": 2.270353302611367,
      "grad_norm": 0.09858205169439316,
      "learning_rate": 9.323734639208607e-06,
      "loss": 0.3855,
      "num_tokens": 1317408352.0,
      "step": 1664
    },
    {
      "epoch": 2.2717187233316265,
      "grad_norm": 0.09545305371284485,
      "learning_rate": 9.304926278640865e-06,
      "loss": 0.4012,
      "num_tokens": 1318242125.0,
      "step": 1665
    },
    {
      "epoch": 2.273084144051886,
      "grad_norm": 0.09298548847436905,
      "learning_rate": 9.286145457391265e-06,
      "loss": 0.4062,
      "num_tokens": 1319031497.0,
      "step": 1666
    },
    {
      "epoch": 2.274449564772145,
      "grad_norm": 0.09414782375097275,
      "learning_rate": 9.26739221620088e-06,
      "loss": 0.3944,
      "num_tokens": 1319864254.0,
      "step": 1667
    },
    {
      "epoch": 2.275814985492405,
      "grad_norm": 0.0928754210472107,
      "learning_rate": 9.248666595750976e-06,
      "loss": 0.3934,
      "num_tokens": 1320645796.0,
      "step": 1668
    },
    {
      "epoch": 2.2771804062126644,
      "grad_norm": 0.0943174734711647,
      "learning_rate": 9.229968636662878e-06,
      "loss": 0.4155,
      "num_tokens": 1321551414.0,
      "step": 1669
    },
    {
      "epoch": 2.2785458269329237,
      "grad_norm": 0.10365728288888931,
      "learning_rate": 9.21129837949792e-06,
      "loss": 0.4024,
      "num_tokens": 1322299632.0,
      "step": 1670
    },
    {
      "epoch": 2.279911247653183,
      "grad_norm": 0.09317129850387573,
      "learning_rate": 9.192655864757318e-06,
      "loss": 0.3866,
      "num_tokens": 1323086753.0,
      "step": 1671
    },
    {
      "epoch": 2.2812766683734425,
      "grad_norm": 0.09492946416139603,
      "learning_rate": 9.174041132882142e-06,
      "loss": 0.4036,
      "num_tokens": 1323797011.0,
      "step": 1672
    },
    {
      "epoch": 2.282642089093702,
      "grad_norm": 0.09937416017055511,
      "learning_rate": 9.155454224253172e-06,
      "loss": 0.4298,
      "num_tokens": 1324680335.0,
      "step": 1673
    },
    {
      "epoch": 2.2840075098139616,
      "grad_norm": 0.09772209823131561,
      "learning_rate": 9.136895179190832e-06,
      "loss": 0.4126,
      "num_tokens": 1325491488.0,
      "step": 1674
    },
    {
      "epoch": 2.285372930534221,
      "grad_norm": 0.09296298772096634,
      "learning_rate": 9.118364037955105e-06,
      "loss": 0.4073,
      "num_tokens": 1326235879.0,
      "step": 1675
    },
    {
      "epoch": 2.2867383512544803,
      "grad_norm": 0.0975380539894104,
      "learning_rate": 9.099860840745443e-06,
      "loss": 0.3955,
      "num_tokens": 1327108376.0,
      "step": 1676
    },
    {
      "epoch": 2.2881037719747397,
      "grad_norm": 0.09035859256982803,
      "learning_rate": 9.081385627700679e-06,
      "loss": 0.4145,
      "num_tokens": 1327864471.0,
      "step": 1677
    },
    {
      "epoch": 2.289469192694999,
      "grad_norm": 0.10155180841684341,
      "learning_rate": 9.062938438898936e-06,
      "loss": 0.4041,
      "num_tokens": 1328657954.0,
      "step": 1678
    },
    {
      "epoch": 2.2908346134152584,
      "grad_norm": 0.09217248111963272,
      "learning_rate": 9.04451931435755e-06,
      "loss": 0.4106,
      "num_tokens": 1329427349.0,
      "step": 1679
    },
    {
      "epoch": 2.2922000341355178,
      "grad_norm": 0.10207211971282959,
      "learning_rate": 9.02612829403297e-06,
      "loss": 0.4137,
      "num_tokens": 1330249911.0,
      "step": 1680
    },
    {
      "epoch": 2.2935654548557776,
      "grad_norm": 0.09454300254583359,
      "learning_rate": 9.007765417820693e-06,
      "loss": 0.3998,
      "num_tokens": 1331098066.0,
      "step": 1681
    },
    {
      "epoch": 2.294930875576037,
      "grad_norm": 0.09491771459579468,
      "learning_rate": 8.989430725555157e-06,
      "loss": 0.4215,
      "num_tokens": 1331899449.0,
      "step": 1682
    },
    {
      "epoch": 2.2962962962962963,
      "grad_norm": 0.09497112035751343,
      "learning_rate": 8.971124257009644e-06,
      "loss": 0.3983,
      "num_tokens": 1332668878.0,
      "step": 1683
    },
    {
      "epoch": 2.2976617170165556,
      "grad_norm": 0.09928029775619507,
      "learning_rate": 8.952846051896229e-06,
      "loss": 0.4007,
      "num_tokens": 1333492846.0,
      "step": 1684
    },
    {
      "epoch": 2.299027137736815,
      "grad_norm": 0.09489702433347702,
      "learning_rate": 8.934596149865675e-06,
      "loss": 0.3966,
      "num_tokens": 1334350439.0,
      "step": 1685
    },
    {
      "epoch": 2.300392558457075,
      "grad_norm": 0.09581439197063446,
      "learning_rate": 8.916374590507344e-06,
      "loss": 0.4025,
      "num_tokens": 1335094810.0,
      "step": 1686
    },
    {
      "epoch": 2.301757979177334,
      "grad_norm": 0.10056094080209732,
      "learning_rate": 8.898181413349107e-06,
      "loss": 0.4,
      "num_tokens": 1335890238.0,
      "step": 1687
    },
    {
      "epoch": 2.3031233998975935,
      "grad_norm": 0.09339898824691772,
      "learning_rate": 8.880016657857278e-06,
      "loss": 0.4034,
      "num_tokens": 1336672281.0,
      "step": 1688
    },
    {
      "epoch": 2.304488820617853,
      "grad_norm": 0.09601354598999023,
      "learning_rate": 8.861880363436507e-06,
      "loss": 0.4222,
      "num_tokens": 1337526505.0,
      "step": 1689
    },
    {
      "epoch": 2.3058542413381122,
      "grad_norm": 0.09306097775697708,
      "learning_rate": 8.843772569429712e-06,
      "loss": 0.414,
      "num_tokens": 1338313843.0,
      "step": 1690
    },
    {
      "epoch": 2.3072196620583716,
      "grad_norm": 0.10127237439155579,
      "learning_rate": 8.825693315117974e-06,
      "loss": 0.4186,
      "num_tokens": 1339134220.0,
      "step": 1691
    },
    {
      "epoch": 2.308585082778631,
      "grad_norm": 0.10448286682367325,
      "learning_rate": 8.807642639720475e-06,
      "loss": 0.4004,
      "num_tokens": 1339905417.0,
      "step": 1692
    },
    {
      "epoch": 2.3099505034988908,
      "grad_norm": 0.09598113596439362,
      "learning_rate": 8.789620582394385e-06,
      "loss": 0.4018,
      "num_tokens": 1340747998.0,
      "step": 1693
    },
    {
      "epoch": 2.31131592421915,
      "grad_norm": 0.0916651263833046,
      "learning_rate": 8.771627182234824e-06,
      "loss": 0.3875,
      "num_tokens": 1341502591.0,
      "step": 1694
    },
    {
      "epoch": 2.3126813449394095,
      "grad_norm": 0.09759789705276489,
      "learning_rate": 8.753662478274704e-06,
      "loss": 0.4079,
      "num_tokens": 1342274617.0,
      "step": 1695
    },
    {
      "epoch": 2.314046765659669,
      "grad_norm": 0.09644515812397003,
      "learning_rate": 8.735726509484722e-06,
      "loss": 0.4051,
      "num_tokens": 1343027036.0,
      "step": 1696
    },
    {
      "epoch": 2.315412186379928,
      "grad_norm": 0.0972147285938263,
      "learning_rate": 8.717819314773213e-06,
      "loss": 0.3915,
      "num_tokens": 1343869808.0,
      "step": 1697
    },
    {
      "epoch": 2.316777607100188,
      "grad_norm": 0.09392126649618149,
      "learning_rate": 8.699940932986121e-06,
      "loss": 0.3999,
      "num_tokens": 1344656396.0,
      "step": 1698
    },
    {
      "epoch": 2.3181430278204473,
      "grad_norm": 0.0958828330039978,
      "learning_rate": 8.682091402906871e-06,
      "loss": 0.3845,
      "num_tokens": 1345433620.0,
      "step": 1699
    },
    {
      "epoch": 2.3195084485407067,
      "grad_norm": 0.09382796287536621,
      "learning_rate": 8.66427076325629e-06,
      "loss": 0.3963,
      "num_tokens": 1346138604.0,
      "step": 1700
    },
    {
      "epoch": 2.320873869260966,
      "grad_norm": 0.09693051874637604,
      "learning_rate": 8.646479052692541e-06,
      "loss": 0.4045,
      "num_tokens": 1346962455.0,
      "step": 1701
    },
    {
      "epoch": 2.3222392899812254,
      "grad_norm": 0.10147445648908615,
      "learning_rate": 8.62871630981105e-06,
      "loss": 0.4044,
      "num_tokens": 1347709933.0,
      "step": 1702
    },
    {
      "epoch": 2.3236047107014848,
      "grad_norm": 0.09062594175338745,
      "learning_rate": 8.610982573144376e-06,
      "loss": 0.4245,
      "num_tokens": 1348556651.0,
      "step": 1703
    },
    {
      "epoch": 2.324970131421744,
      "grad_norm": 0.09791932255029678,
      "learning_rate": 8.593277881162171e-06,
      "loss": 0.4217,
      "num_tokens": 1349271651.0,
      "step": 1704
    },
    {
      "epoch": 2.326335552142004,
      "grad_norm": 0.09933177381753922,
      "learning_rate": 8.575602272271077e-06,
      "loss": 0.4117,
      "num_tokens": 1350145163.0,
      "step": 1705
    },
    {
      "epoch": 2.3277009728622633,
      "grad_norm": 0.0937393382191658,
      "learning_rate": 8.557955784814647e-06,
      "loss": 0.431,
      "num_tokens": 1350961322.0,
      "step": 1706
    },
    {
      "epoch": 2.3290663935825227,
      "grad_norm": 0.1009640023112297,
      "learning_rate": 8.540338457073253e-06,
      "loss": 0.4003,
      "num_tokens": 1351705998.0,
      "step": 1707
    },
    {
      "epoch": 2.330431814302782,
      "grad_norm": 0.10209198296070099,
      "learning_rate": 8.522750327264024e-06,
      "loss": 0.3965,
      "num_tokens": 1352435688.0,
      "step": 1708
    },
    {
      "epoch": 2.3317972350230414,
      "grad_norm": 0.09848228096961975,
      "learning_rate": 8.505191433540743e-06,
      "loss": 0.4011,
      "num_tokens": 1353186083.0,
      "step": 1709
    },
    {
      "epoch": 2.3331626557433007,
      "grad_norm": 0.09884539246559143,
      "learning_rate": 8.487661813993767e-06,
      "loss": 0.4218,
      "num_tokens": 1354000253.0,
      "step": 1710
    },
    {
      "epoch": 2.3345280764635605,
      "grad_norm": 0.09319755434989929,
      "learning_rate": 8.47016150664997e-06,
      "loss": 0.4018,
      "num_tokens": 1354802805.0,
      "step": 1711
    },
    {
      "epoch": 2.33589349718382,
      "grad_norm": 0.09683693200349808,
      "learning_rate": 8.452690549472608e-06,
      "loss": 0.3847,
      "num_tokens": 1355569674.0,
      "step": 1712
    },
    {
      "epoch": 2.3372589179040792,
      "grad_norm": 0.09437060356140137,
      "learning_rate": 8.43524898036129e-06,
      "loss": 0.391,
      "num_tokens": 1356348293.0,
      "step": 1713
    },
    {
      "epoch": 2.3386243386243386,
      "grad_norm": 0.09812674671411514,
      "learning_rate": 8.417836837151857e-06,
      "loss": 0.4154,
      "num_tokens": 1357133277.0,
      "step": 1714
    },
    {
      "epoch": 2.339989759344598,
      "grad_norm": 0.09293098747730255,
      "learning_rate": 8.400454157616343e-06,
      "loss": 0.3959,
      "num_tokens": 1357935470.0,
      "step": 1715
    },
    {
      "epoch": 2.3413551800648573,
      "grad_norm": 0.09371138364076614,
      "learning_rate": 8.383100979462843e-06,
      "loss": 0.407,
      "num_tokens": 1358657615.0,
      "step": 1716
    },
    {
      "epoch": 2.3427206007851167,
      "grad_norm": 0.09427204728126526,
      "learning_rate": 8.365777340335466e-06,
      "loss": 0.4092,
      "num_tokens": 1359451699.0,
      "step": 1717
    },
    {
      "epoch": 2.3440860215053765,
      "grad_norm": 0.09673437476158142,
      "learning_rate": 8.348483277814228e-06,
      "loss": 0.401,
      "num_tokens": 1360207796.0,
      "step": 1718
    },
    {
      "epoch": 2.345451442225636,
      "grad_norm": 0.09722506999969482,
      "learning_rate": 8.331218829415008e-06,
      "loss": 0.3988,
      "num_tokens": 1360991712.0,
      "step": 1719
    },
    {
      "epoch": 2.346816862945895,
      "grad_norm": 0.09315294772386551,
      "learning_rate": 8.313984032589423e-06,
      "loss": 0.3879,
      "num_tokens": 1361687998.0,
      "step": 1720
    },
    {
      "epoch": 2.3481822836661546,
      "grad_norm": 0.09436554461717606,
      "learning_rate": 8.296778924724777e-06,
      "loss": 0.4173,
      "num_tokens": 1362478611.0,
      "step": 1721
    },
    {
      "epoch": 2.349547704386414,
      "grad_norm": 0.09866146743297577,
      "learning_rate": 8.27960354314397e-06,
      "loss": 0.3932,
      "num_tokens": 1363219247.0,
      "step": 1722
    },
    {
      "epoch": 2.3509131251066737,
      "grad_norm": 0.0959412157535553,
      "learning_rate": 8.262457925105414e-06,
      "loss": 0.3775,
      "num_tokens": 1363929189.0,
      "step": 1723
    },
    {
      "epoch": 2.352278545826933,
      "grad_norm": 0.10344033688306808,
      "learning_rate": 8.245342107802954e-06,
      "loss": 0.4153,
      "num_tokens": 1364782567.0,
      "step": 1724
    },
    {
      "epoch": 2.3536439665471924,
      "grad_norm": 0.09278857707977295,
      "learning_rate": 8.22825612836579e-06,
      "loss": 0.3969,
      "num_tokens": 1365621876.0,
      "step": 1725
    },
    {
      "epoch": 2.355009387267452,
      "grad_norm": 0.08986662328243256,
      "learning_rate": 8.2112000238584e-06,
      "loss": 0.3991,
      "num_tokens": 1366376013.0,
      "step": 1726
    },
    {
      "epoch": 2.356374807987711,
      "grad_norm": 0.10116857290267944,
      "learning_rate": 8.194173831280438e-06,
      "loss": 0.401,
      "num_tokens": 1367095704.0,
      "step": 1727
    },
    {
      "epoch": 2.3577402287079705,
      "grad_norm": 0.09166115522384644,
      "learning_rate": 8.177177587566698e-06,
      "loss": 0.3916,
      "num_tokens": 1367826636.0,
      "step": 1728
    },
    {
      "epoch": 2.35910564942823,
      "grad_norm": 0.09865598380565643,
      "learning_rate": 8.160211329586991e-06,
      "loss": 0.3967,
      "num_tokens": 1368667661.0,
      "step": 1729
    },
    {
      "epoch": 2.3604710701484897,
      "grad_norm": 0.09915482997894287,
      "learning_rate": 8.143275094146068e-06,
      "loss": 0.4066,
      "num_tokens": 1369420989.0,
      "step": 1730
    },
    {
      "epoch": 2.361836490868749,
      "grad_norm": 0.09218109399080276,
      "learning_rate": 8.12636891798356e-06,
      "loss": 0.3918,
      "num_tokens": 1370254348.0,
      "step": 1731
    },
    {
      "epoch": 2.3632019115890084,
      "grad_norm": 0.09546741098165512,
      "learning_rate": 8.109492837773912e-06,
      "loss": 0.3838,
      "num_tokens": 1371029173.0,
      "step": 1732
    },
    {
      "epoch": 2.3645673323092677,
      "grad_norm": 0.09043113142251968,
      "learning_rate": 8.092646890126255e-06,
      "loss": 0.4087,
      "num_tokens": 1371752256.0,
      "step": 1733
    },
    {
      "epoch": 2.365932753029527,
      "grad_norm": 0.10270420461893082,
      "learning_rate": 8.075831111584364e-06,
      "loss": 0.4235,
      "num_tokens": 1372569219.0,
      "step": 1734
    },
    {
      "epoch": 2.3672981737497865,
      "grad_norm": 0.09740057587623596,
      "learning_rate": 8.05904553862657e-06,
      "loss": 0.4205,
      "num_tokens": 1373430175.0,
      "step": 1735
    },
    {
      "epoch": 2.3686635944700463,
      "grad_norm": 0.10265360027551651,
      "learning_rate": 8.042290207665677e-06,
      "loss": 0.3747,
      "num_tokens": 1374187197.0,
      "step": 1736
    },
    {
      "epoch": 2.3700290151903056,
      "grad_norm": 0.0907999724149704,
      "learning_rate": 8.025565155048884e-06,
      "loss": 0.3951,
      "num_tokens": 1374917106.0,
      "step": 1737
    },
    {
      "epoch": 2.371394435910565,
      "grad_norm": 0.09388679265975952,
      "learning_rate": 8.008870417057709e-06,
      "loss": 0.4175,
      "num_tokens": 1375736124.0,
      "step": 1738
    },
    {
      "epoch": 2.3727598566308243,
      "grad_norm": 0.0967550203204155,
      "learning_rate": 7.99220602990791e-06,
      "loss": 0.4022,
      "num_tokens": 1376519049.0,
      "step": 1739
    },
    {
      "epoch": 2.3741252773510837,
      "grad_norm": 0.10198497027158737,
      "learning_rate": 7.975572029749402e-06,
      "loss": 0.3844,
      "num_tokens": 1377326429.0,
      "step": 1740
    },
    {
      "epoch": 2.375490698071343,
      "grad_norm": 0.09259793907403946,
      "learning_rate": 7.958968452666184e-06,
      "loss": 0.4,
      "num_tokens": 1378091477.0,
      "step": 1741
    },
    {
      "epoch": 2.376856118791603,
      "grad_norm": 0.10230544954538345,
      "learning_rate": 7.94239533467626e-06,
      "loss": 0.4065,
      "num_tokens": 1378869160.0,
      "step": 1742
    },
    {
      "epoch": 2.378221539511862,
      "grad_norm": 0.09618833661079407,
      "learning_rate": 7.925852711731557e-06,
      "loss": 0.3882,
      "num_tokens": 1379639295.0,
      "step": 1743
    },
    {
      "epoch": 2.3795869602321216,
      "grad_norm": 0.09133503586053848,
      "learning_rate": 7.909340619717848e-06,
      "loss": 0.3835,
      "num_tokens": 1380422846.0,
      "step": 1744
    },
    {
      "epoch": 2.380952380952381,
      "grad_norm": 0.08841606229543686,
      "learning_rate": 7.892859094454672e-06,
      "loss": 0.4054,
      "num_tokens": 1381294741.0,
      "step": 1745
    },
    {
      "epoch": 2.3823178016726403,
      "grad_norm": 0.09453877061605453,
      "learning_rate": 7.87640817169528e-06,
      "loss": 0.3874,
      "num_tokens": 1382017672.0,
      "step": 1746
    },
    {
      "epoch": 2.3836832223928996,
      "grad_norm": 0.10645560175180435,
      "learning_rate": 7.859987887126512e-06,
      "loss": 0.4098,
      "num_tokens": 1382835216.0,
      "step": 1747
    },
    {
      "epoch": 2.3850486431131595,
      "grad_norm": 0.09965003281831741,
      "learning_rate": 7.843598276368756e-06,
      "loss": 0.389,
      "num_tokens": 1383522147.0,
      "step": 1748
    },
    {
      "epoch": 2.386414063833419,
      "grad_norm": 0.09883282333612442,
      "learning_rate": 7.827239374975853e-06,
      "loss": 0.4212,
      "num_tokens": 1384307226.0,
      "step": 1749
    },
    {
      "epoch": 2.387779484553678,
      "grad_norm": 0.09670928120613098,
      "learning_rate": 7.810911218435044e-06,
      "loss": 0.4045,
      "num_tokens": 1385145567.0,
      "step": 1750
    },
    {
      "epoch": 2.3891449052739375,
      "grad_norm": 0.09608610719442368,
      "learning_rate": 7.794613842166859e-06,
      "loss": 0.4181,
      "num_tokens": 1385939791.0,
      "step": 1751
    },
    {
      "epoch": 2.390510325994197,
      "grad_norm": 0.09865006804466248,
      "learning_rate": 7.778347281525058e-06,
      "loss": 0.4103,
      "num_tokens": 1386752292.0,
      "step": 1752
    },
    {
      "epoch": 2.3918757467144562,
      "grad_norm": 0.09749456495046616,
      "learning_rate": 7.762111571796562e-06,
      "loss": 0.3806,
      "num_tokens": 1387522814.0,
      "step": 1753
    },
    {
      "epoch": 2.3932411674347156,
      "grad_norm": 0.09666375815868378,
      "learning_rate": 7.745906748201354e-06,
      "loss": 0.4062,
      "num_tokens": 1388308036.0,
      "step": 1754
    },
    {
      "epoch": 2.3946065881549754,
      "grad_norm": 0.09653221070766449,
      "learning_rate": 7.72973284589243e-06,
      "loss": 0.4008,
      "num_tokens": 1389069283.0,
      "step": 1755
    },
    {
      "epoch": 2.3959720088752348,
      "grad_norm": 0.09443670511245728,
      "learning_rate": 7.713589899955698e-06,
      "loss": 0.4136,
      "num_tokens": 1389868793.0,
      "step": 1756
    },
    {
      "epoch": 2.397337429595494,
      "grad_norm": 0.0979897677898407,
      "learning_rate": 7.69747794540992e-06,
      "loss": 0.429,
      "num_tokens": 1390709024.0,
      "step": 1757
    },
    {
      "epoch": 2.3987028503157535,
      "grad_norm": 0.10439679771661758,
      "learning_rate": 7.681397017206621e-06,
      "loss": 0.4076,
      "num_tokens": 1391504759.0,
      "step": 1758
    },
    {
      "epoch": 2.400068271036013,
      "grad_norm": 0.09746696054935455,
      "learning_rate": 7.665347150230029e-06,
      "loss": 0.4221,
      "num_tokens": 1392285188.0,
      "step": 1759
    },
    {
      "epoch": 2.4014336917562726,
      "grad_norm": 0.09530731290578842,
      "learning_rate": 7.649328379296984e-06,
      "loss": 0.4126,
      "num_tokens": 1393105374.0,
      "step": 1760
    },
    {
      "epoch": 2.402799112476532,
      "grad_norm": 0.09623304009437561,
      "learning_rate": 7.633340739156879e-06,
      "loss": 0.4032,
      "num_tokens": 1393823494.0,
      "step": 1761
    },
    {
      "epoch": 2.4041645331967914,
      "grad_norm": 0.10252279043197632,
      "learning_rate": 7.617384264491558e-06,
      "loss": 0.4106,
      "num_tokens": 1394667198.0,
      "step": 1762
    },
    {
      "epoch": 2.4055299539170507,
      "grad_norm": 0.09577364474534988,
      "learning_rate": 7.601458989915283e-06,
      "loss": 0.4307,
      "num_tokens": 1395563568.0,
      "step": 1763
    },
    {
      "epoch": 2.40689537463731,
      "grad_norm": 0.0991474911570549,
      "learning_rate": 7.585564949974615e-06,
      "loss": 0.4178,
      "num_tokens": 1396329196.0,
      "step": 1764
    },
    {
      "epoch": 2.4082607953575694,
      "grad_norm": 0.09657761454582214,
      "learning_rate": 7.569702179148357e-06,
      "loss": 0.4047,
      "num_tokens": 1397104045.0,
      "step": 1765
    },
    {
      "epoch": 2.409626216077829,
      "grad_norm": 0.09525848925113678,
      "learning_rate": 7.553870711847489e-06,
      "loss": 0.4221,
      "num_tokens": 1397901951.0,
      "step": 1766
    },
    {
      "epoch": 2.4109916367980886,
      "grad_norm": 0.10029143840074539,
      "learning_rate": 7.538070582415087e-06,
      "loss": 0.4221,
      "num_tokens": 1398816207.0,
      "step": 1767
    },
    {
      "epoch": 2.412357057518348,
      "grad_norm": 0.09553137421607971,
      "learning_rate": 7.522301825126235e-06,
      "loss": 0.3822,
      "num_tokens": 1399594570.0,
      "step": 1768
    },
    {
      "epoch": 2.4137224782386073,
      "grad_norm": 0.09421303868293762,
      "learning_rate": 7.506564474187972e-06,
      "loss": 0.4034,
      "num_tokens": 1400391587.0,
      "step": 1769
    },
    {
      "epoch": 2.4150878989588667,
      "grad_norm": 0.09511613845825195,
      "learning_rate": 7.490858563739198e-06,
      "loss": 0.3909,
      "num_tokens": 1401148721.0,
      "step": 1770
    },
    {
      "epoch": 2.416453319679126,
      "grad_norm": 0.10411738604307175,
      "learning_rate": 7.475184127850619e-06,
      "loss": 0.4013,
      "num_tokens": 1401943821.0,
      "step": 1771
    },
    {
      "epoch": 2.4178187403993854,
      "grad_norm": 0.09129701554775238,
      "learning_rate": 7.459541200524655e-06,
      "loss": 0.4012,
      "num_tokens": 1402668535.0,
      "step": 1772
    },
    {
      "epoch": 2.419184161119645,
      "grad_norm": 0.09189926832914352,
      "learning_rate": 7.443929815695381e-06,
      "loss": 0.4071,
      "num_tokens": 1403473500.0,
      "step": 1773
    },
    {
      "epoch": 2.4205495818399045,
      "grad_norm": 0.09532161056995392,
      "learning_rate": 7.428350007228442e-06,
      "loss": 0.4256,
      "num_tokens": 1404316948.0,
      "step": 1774
    },
    {
      "epoch": 2.421915002560164,
      "grad_norm": 0.0950179472565651,
      "learning_rate": 7.412801808920987e-06,
      "loss": 0.3957,
      "num_tokens": 1405042674.0,
      "step": 1775
    },
    {
      "epoch": 2.4232804232804233,
      "grad_norm": 0.10107088088989258,
      "learning_rate": 7.397285254501603e-06,
      "loss": 0.408,
      "num_tokens": 1405839141.0,
      "step": 1776
    },
    {
      "epoch": 2.4246458440006826,
      "grad_norm": 0.08784332871437073,
      "learning_rate": 7.3818003776302105e-06,
      "loss": 0.4229,
      "num_tokens": 1406631957.0,
      "step": 1777
    },
    {
      "epoch": 2.426011264720942,
      "grad_norm": 0.09807129949331284,
      "learning_rate": 7.366347211898031e-06,
      "loss": 0.3903,
      "num_tokens": 1407385791.0,
      "step": 1778
    },
    {
      "epoch": 2.4273766854412013,
      "grad_norm": 0.10051941126585007,
      "learning_rate": 7.350925790827481e-06,
      "loss": 0.3932,
      "num_tokens": 1408181054.0,
      "step": 1779
    },
    {
      "epoch": 2.428742106161461,
      "grad_norm": 0.09349891543388367,
      "learning_rate": 7.335536147872131e-06,
      "loss": 0.4258,
      "num_tokens": 1409032244.0,
      "step": 1780
    },
    {
      "epoch": 2.4301075268817205,
      "grad_norm": 0.0910247340798378,
      "learning_rate": 7.320178316416604e-06,
      "loss": 0.4025,
      "num_tokens": 1409879239.0,
      "step": 1781
    },
    {
      "epoch": 2.43147294760198,
      "grad_norm": 0.09298503398895264,
      "learning_rate": 7.30485232977652e-06,
      "loss": 0.421,
      "num_tokens": 1410690046.0,
      "step": 1782
    },
    {
      "epoch": 2.432838368322239,
      "grad_norm": 0.09487801790237427,
      "learning_rate": 7.289558221198401e-06,
      "loss": 0.3983,
      "num_tokens": 1411505870.0,
      "step": 1783
    },
    {
      "epoch": 2.4342037890424986,
      "grad_norm": 0.09647859632968903,
      "learning_rate": 7.2742960238596415e-06,
      "loss": 0.3847,
      "num_tokens": 1412260793.0,
      "step": 1784
    },
    {
      "epoch": 2.4355692097627584,
      "grad_norm": 0.09321960061788559,
      "learning_rate": 7.259065770868397e-06,
      "loss": 0.4075,
      "num_tokens": 1413061116.0,
      "step": 1785
    },
    {
      "epoch": 2.4369346304830177,
      "grad_norm": 0.0924951434135437,
      "learning_rate": 7.243867495263527e-06,
      "loss": 0.4101,
      "num_tokens": 1413907558.0,
      "step": 1786
    },
    {
      "epoch": 2.438300051203277,
      "grad_norm": 0.09694798290729523,
      "learning_rate": 7.228701230014528e-06,
      "loss": 0.3905,
      "num_tokens": 1414652146.0,
      "step": 1787
    },
    {
      "epoch": 2.4396654719235364,
      "grad_norm": 0.09072563052177429,
      "learning_rate": 7.2135670080214516e-06,
      "loss": 0.4187,
      "num_tokens": 1415502490.0,
      "step": 1788
    },
    {
      "epoch": 2.441030892643796,
      "grad_norm": 0.09537357091903687,
      "learning_rate": 7.198464862114838e-06,
      "loss": 0.4156,
      "num_tokens": 1416299429.0,
      "step": 1789
    },
    {
      "epoch": 2.442396313364055,
      "grad_norm": 0.10175148397684097,
      "learning_rate": 7.183394825055652e-06,
      "loss": 0.4201,
      "num_tokens": 1417177003.0,
      "step": 1790
    },
    {
      "epoch": 2.4437617340843145,
      "grad_norm": 0.09121620655059814,
      "learning_rate": 7.168356929535196e-06,
      "loss": 0.4051,
      "num_tokens": 1417968963.0,
      "step": 1791
    },
    {
      "epoch": 2.4451271548045743,
      "grad_norm": 0.09385012835264206,
      "learning_rate": 7.153351208175048e-06,
      "loss": 0.3994,
      "num_tokens": 1418716921.0,
      "step": 1792
    },
    {
      "epoch": 2.4464925755248337,
      "grad_norm": 0.09638749063014984,
      "learning_rate": 7.138377693527011e-06,
      "loss": 0.4106,
      "num_tokens": 1419547958.0,
      "step": 1793
    },
    {
      "epoch": 2.447857996245093,
      "grad_norm": 0.09490463137626648,
      "learning_rate": 7.123436418072989e-06,
      "loss": 0.4121,
      "num_tokens": 1420349607.0,
      "step": 1794
    },
    {
      "epoch": 2.4492234169653524,
      "grad_norm": 0.0922735333442688,
      "learning_rate": 7.108527414224974e-06,
      "loss": 0.3952,
      "num_tokens": 1421121810.0,
      "step": 1795
    },
    {
      "epoch": 2.4505888376856118,
      "grad_norm": 0.10679720342159271,
      "learning_rate": 7.09365071432494e-06,
      "loss": 0.427,
      "num_tokens": 1421910199.0,
      "step": 1796
    },
    {
      "epoch": 2.4519542584058716,
      "grad_norm": 0.10029077529907227,
      "learning_rate": 7.078806350644796e-06,
      "loss": 0.3748,
      "num_tokens": 1422706147.0,
      "step": 1797
    },
    {
      "epoch": 2.453319679126131,
      "grad_norm": 0.09536296129226685,
      "learning_rate": 7.063994355386293e-06,
      "loss": 0.4135,
      "num_tokens": 1423529479.0,
      "step": 1798
    },
    {
      "epoch": 2.4546850998463903,
      "grad_norm": 0.09225652366876602,
      "learning_rate": 7.049214760680973e-06,
      "loss": 0.4137,
      "num_tokens": 1424345864.0,
      "step": 1799
    },
    {
      "epoch": 2.4560505205666496,
      "grad_norm": 0.09622188657522202,
      "learning_rate": 7.03446759859008e-06,
      "loss": 0.3999,
      "num_tokens": 1425043344.0,
      "step": 1800
    },
    {
      "epoch": 2.457415941286909,
      "grad_norm": 0.10075096040964127,
      "learning_rate": 7.0197529011045165e-06,
      "loss": 0.4138,
      "num_tokens": 1425827500.0,
      "step": 1801
    },
    {
      "epoch": 2.4587813620071683,
      "grad_norm": 0.09771601110696793,
      "learning_rate": 7.005070700144754e-06,
      "loss": 0.3906,
      "num_tokens": 1426578235.0,
      "step": 1802
    },
    {
      "epoch": 2.4601467827274277,
      "grad_norm": 0.09154694527387619,
      "learning_rate": 6.990421027560767e-06,
      "loss": 0.3853,
      "num_tokens": 1427308406.0,
      "step": 1803
    },
    {
      "epoch": 2.4615122034476875,
      "grad_norm": 0.09345058351755142,
      "learning_rate": 6.975803915131972e-06,
      "loss": 0.3969,
      "num_tokens": 1428065875.0,
      "step": 1804
    },
    {
      "epoch": 2.462877624167947,
      "grad_norm": 0.1008749008178711,
      "learning_rate": 6.961219394567148e-06,
      "loss": 0.3943,
      "num_tokens": 1428838663.0,
      "step": 1805
    },
    {
      "epoch": 2.4642430448882062,
      "grad_norm": 0.0851283073425293,
      "learning_rate": 6.946667497504371e-06,
      "loss": 0.4013,
      "num_tokens": 1429636120.0,
      "step": 1806
    },
    {
      "epoch": 2.4656084656084656,
      "grad_norm": 0.09406868368387222,
      "learning_rate": 6.932148255510955e-06,
      "loss": 0.3918,
      "num_tokens": 1430422929.0,
      "step": 1807
    },
    {
      "epoch": 2.466973886328725,
      "grad_norm": 0.09537248313426971,
      "learning_rate": 6.917661700083371e-06,
      "loss": 0.3768,
      "num_tokens": 1431212078.0,
      "step": 1808
    },
    {
      "epoch": 2.4683393070489843,
      "grad_norm": 0.0944947674870491,
      "learning_rate": 6.9032078626471765e-06,
      "loss": 0.4116,
      "num_tokens": 1432025261.0,
      "step": 1809
    },
    {
      "epoch": 2.469704727769244,
      "grad_norm": 0.09532343596220016,
      "learning_rate": 6.888786774556972e-06,
      "loss": 0.4352,
      "num_tokens": 1432828284.0,
      "step": 1810
    },
    {
      "epoch": 2.4710701484895035,
      "grad_norm": 0.0970771536231041,
      "learning_rate": 6.8743984670963045e-06,
      "loss": 0.3764,
      "num_tokens": 1433533612.0,
      "step": 1811
    },
    {
      "epoch": 2.472435569209763,
      "grad_norm": 0.0880197286605835,
      "learning_rate": 6.860042971477601e-06,
      "loss": 0.4137,
      "num_tokens": 1434344180.0,
      "step": 1812
    },
    {
      "epoch": 2.473800989930022,
      "grad_norm": 0.09898258745670319,
      "learning_rate": 6.845720318842117e-06,
      "loss": 0.3974,
      "num_tokens": 1435153086.0,
      "step": 1813
    },
    {
      "epoch": 2.4751664106502815,
      "grad_norm": 0.08922412246465683,
      "learning_rate": 6.831430540259875e-06,
      "loss": 0.3852,
      "num_tokens": 1435959542.0,
      "step": 1814
    },
    {
      "epoch": 2.476531831370541,
      "grad_norm": 0.09473318606615067,
      "learning_rate": 6.8171736667295695e-06,
      "loss": 0.4037,
      "num_tokens": 1436752252.0,
      "step": 1815
    },
    {
      "epoch": 2.4778972520908003,
      "grad_norm": 0.104718416929245,
      "learning_rate": 6.802949729178516e-06,
      "loss": 0.4123,
      "num_tokens": 1437495623.0,
      "step": 1816
    },
    {
      "epoch": 2.47926267281106,
      "grad_norm": 0.09608372300863266,
      "learning_rate": 6.788758758462586e-06,
      "loss": 0.4192,
      "num_tokens": 1438314049.0,
      "step": 1817
    },
    {
      "epoch": 2.4806280935313194,
      "grad_norm": 0.09347930550575256,
      "learning_rate": 6.774600785366134e-06,
      "loss": 0.3905,
      "num_tokens": 1439121216.0,
      "step": 1818
    },
    {
      "epoch": 2.4819935142515788,
      "grad_norm": 0.08853622525930405,
      "learning_rate": 6.7604758406019325e-06,
      "loss": 0.3955,
      "num_tokens": 1439910675.0,
      "step": 1819
    },
    {
      "epoch": 2.483358934971838,
      "grad_norm": 0.09452494233846664,
      "learning_rate": 6.746383954811109e-06,
      "loss": 0.3979,
      "num_tokens": 1440689562.0,
      "step": 1820
    },
    {
      "epoch": 2.4847243556920975,
      "grad_norm": 0.09481710940599442,
      "learning_rate": 6.732325158563071e-06,
      "loss": 0.4392,
      "num_tokens": 1441520996.0,
      "step": 1821
    },
    {
      "epoch": 2.4860897764123573,
      "grad_norm": 0.09924156218767166,
      "learning_rate": 6.718299482355448e-06,
      "loss": 0.3844,
      "num_tokens": 1442245109.0,
      "step": 1822
    },
    {
      "epoch": 2.4874551971326166,
      "grad_norm": 0.09648500382900238,
      "learning_rate": 6.704306956614035e-06,
      "loss": 0.3848,
      "num_tokens": 1443015929.0,
      "step": 1823
    },
    {
      "epoch": 2.488820617852876,
      "grad_norm": 0.09641363471746445,
      "learning_rate": 6.690347611692688e-06,
      "loss": 0.4188,
      "num_tokens": 1443804949.0,
      "step": 1824
    },
    {
      "epoch": 2.4901860385731354,
      "grad_norm": 0.08825608342885971,
      "learning_rate": 6.676421477873305e-06,
      "loss": 0.3861,
      "num_tokens": 1444571298.0,
      "step": 1825
    },
    {
      "epoch": 2.4915514592933947,
      "grad_norm": 0.09441713243722916,
      "learning_rate": 6.662528585365724e-06,
      "loss": 0.3863,
      "num_tokens": 1445285635.0,
      "step": 1826
    },
    {
      "epoch": 2.492916880013654,
      "grad_norm": 0.0955725833773613,
      "learning_rate": 6.648668964307694e-06,
      "loss": 0.4001,
      "num_tokens": 1446119786.0,
      "step": 1827
    },
    {
      "epoch": 2.4942823007339134,
      "grad_norm": 0.0926135778427124,
      "learning_rate": 6.6348426447647756e-06,
      "loss": 0.3933,
      "num_tokens": 1446924803.0,
      "step": 1828
    },
    {
      "epoch": 2.4956477214541732,
      "grad_norm": 0.09665083885192871,
      "learning_rate": 6.621049656730278e-06,
      "loss": 0.4185,
      "num_tokens": 1447701397.0,
      "step": 1829
    },
    {
      "epoch": 2.4970131421744326,
      "grad_norm": 0.09904329478740692,
      "learning_rate": 6.607290030125219e-06,
      "loss": 0.3942,
      "num_tokens": 1448526996.0,
      "step": 1830
    },
    {
      "epoch": 2.498378562894692,
      "grad_norm": 0.09437855333089828,
      "learning_rate": 6.5935637947982485e-06,
      "loss": 0.379,
      "num_tokens": 1449287847.0,
      "step": 1831
    },
    {
      "epoch": 2.4997439836149513,
      "grad_norm": 0.09524792432785034,
      "learning_rate": 6.579870980525573e-06,
      "loss": 0.4282,
      "num_tokens": 1450188772.0,
      "step": 1832
    },
    {
      "epoch": 2.5011094043352107,
      "grad_norm": 0.0906117856502533,
      "learning_rate": 6.566211617010899e-06,
      "loss": 0.4026,
      "num_tokens": 1450935495.0,
      "step": 1833
    },
    {
      "epoch": 2.5024748250554705,
      "grad_norm": 0.09287602454423904,
      "learning_rate": 6.552585733885369e-06,
      "loss": 0.3901,
      "num_tokens": 1451703993.0,
      "step": 1834
    },
    {
      "epoch": 2.50384024577573,
      "grad_norm": 0.09984180331230164,
      "learning_rate": 6.538993360707504e-06,
      "loss": 0.4184,
      "num_tokens": 1452414043.0,
      "step": 1835
    },
    {
      "epoch": 2.505205666495989,
      "grad_norm": 0.09817738085985184,
      "learning_rate": 6.525434526963119e-06,
      "loss": 0.4075,
      "num_tokens": 1453237503.0,
      "step": 1836
    },
    {
      "epoch": 2.5065710872162486,
      "grad_norm": 0.0983203873038292,
      "learning_rate": 6.511909262065287e-06,
      "loss": 0.4038,
      "num_tokens": 1454064859.0,
      "step": 1837
    },
    {
      "epoch": 2.507936507936508,
      "grad_norm": 0.0887639969587326,
      "learning_rate": 6.498417595354247e-06,
      "loss": 0.4008,
      "num_tokens": 1454823949.0,
      "step": 1838
    },
    {
      "epoch": 2.5093019286567673,
      "grad_norm": 0.09347344189882278,
      "learning_rate": 6.484959556097359e-06,
      "loss": 0.4143,
      "num_tokens": 1455623248.0,
      "step": 1839
    },
    {
      "epoch": 2.5106673493770266,
      "grad_norm": 0.09773072600364685,
      "learning_rate": 6.4715351734890465e-06,
      "loss": 0.4147,
      "num_tokens": 1456436991.0,
      "step": 1840
    },
    {
      "epoch": 2.512032770097286,
      "grad_norm": 0.09596148133277893,
      "learning_rate": 6.458144476650699e-06,
      "loss": 0.4141,
      "num_tokens": 1457253396.0,
      "step": 1841
    },
    {
      "epoch": 2.513398190817546,
      "grad_norm": 0.09724052995443344,
      "learning_rate": 6.44478749463065e-06,
      "loss": 0.4181,
      "num_tokens": 1458010697.0,
      "step": 1842
    },
    {
      "epoch": 2.514763611537805,
      "grad_norm": 0.09504944086074829,
      "learning_rate": 6.431464256404084e-06,
      "loss": 0.3783,
      "num_tokens": 1458764233.0,
      "step": 1843
    },
    {
      "epoch": 2.5161290322580645,
      "grad_norm": 0.09069196879863739,
      "learning_rate": 6.418174790873002e-06,
      "loss": 0.3832,
      "num_tokens": 1459454120.0,
      "step": 1844
    },
    {
      "epoch": 2.517494452978324,
      "grad_norm": 0.0960589349269867,
      "learning_rate": 6.404919126866124e-06,
      "loss": 0.4036,
      "num_tokens": 1460285800.0,
      "step": 1845
    },
    {
      "epoch": 2.518859873698583,
      "grad_norm": 0.09708927571773529,
      "learning_rate": 6.391697293138856e-06,
      "loss": 0.4134,
      "num_tokens": 1461130936.0,
      "step": 1846
    },
    {
      "epoch": 2.520225294418843,
      "grad_norm": 0.09632696211338043,
      "learning_rate": 6.378509318373205e-06,
      "loss": 0.4032,
      "num_tokens": 1461946526.0,
      "step": 1847
    },
    {
      "epoch": 2.5215907151391024,
      "grad_norm": 0.09884379059076309,
      "learning_rate": 6.36535523117774e-06,
      "loss": 0.4137,
      "num_tokens": 1462687772.0,
      "step": 1848
    },
    {
      "epoch": 2.5229561358593617,
      "grad_norm": 0.09286358952522278,
      "learning_rate": 6.352235060087516e-06,
      "loss": 0.413,
      "num_tokens": 1463617569.0,
      "step": 1849
    },
    {
      "epoch": 2.524321556579621,
      "grad_norm": 0.09459643810987473,
      "learning_rate": 6.3391488335640075e-06,
      "loss": 0.4225,
      "num_tokens": 1464370713.0,
      "step": 1850
    },
    {
      "epoch": 2.5256869772998805,
      "grad_norm": 0.09824322164058685,
      "learning_rate": 6.326096579995058e-06,
      "loss": 0.4251,
      "num_tokens": 1465114050.0,
      "step": 1851
    },
    {
      "epoch": 2.52705239802014,
      "grad_norm": 0.09841996431350708,
      "learning_rate": 6.313078327694814e-06,
      "loss": 0.4023,
      "num_tokens": 1465884916.0,
      "step": 1852
    },
    {
      "epoch": 2.528417818740399,
      "grad_norm": 0.09185584634542465,
      "learning_rate": 6.30009410490366e-06,
      "loss": 0.4222,
      "num_tokens": 1466689313.0,
      "step": 1853
    },
    {
      "epoch": 2.529783239460659,
      "grad_norm": 0.09746100753545761,
      "learning_rate": 6.2871439397881664e-06,
      "loss": 0.3716,
      "num_tokens": 1467423211.0,
      "step": 1854
    },
    {
      "epoch": 2.5311486601809183,
      "grad_norm": 0.0921073704957962,
      "learning_rate": 6.2742278604410135e-06,
      "loss": 0.4324,
      "num_tokens": 1468245492.0,
      "step": 1855
    },
    {
      "epoch": 2.5325140809011777,
      "grad_norm": 0.1032097339630127,
      "learning_rate": 6.261345894880947e-06,
      "loss": 0.3849,
      "num_tokens": 1468983572.0,
      "step": 1856
    },
    {
      "epoch": 2.533879501621437,
      "grad_norm": 0.09687773883342743,
      "learning_rate": 6.248498071052713e-06,
      "loss": 0.3885,
      "num_tokens": 1469781507.0,
      "step": 1857
    },
    {
      "epoch": 2.5352449223416964,
      "grad_norm": 0.09321971237659454,
      "learning_rate": 6.235684416826989e-06,
      "loss": 0.3984,
      "num_tokens": 1470568649.0,
      "step": 1858
    },
    {
      "epoch": 2.536610343061956,
      "grad_norm": 0.08858620375394821,
      "learning_rate": 6.222904960000322e-06,
      "loss": 0.4191,
      "num_tokens": 1471480425.0,
      "step": 1859
    },
    {
      "epoch": 2.5379757637822156,
      "grad_norm": 0.09058038890361786,
      "learning_rate": 6.210159728295081e-06,
      "loss": 0.4082,
      "num_tokens": 1472318093.0,
      "step": 1860
    },
    {
      "epoch": 2.539341184502475,
      "grad_norm": 0.09291671961545944,
      "learning_rate": 6.197448749359399e-06,
      "loss": 0.4233,
      "num_tokens": 1473069778.0,
      "step": 1861
    },
    {
      "epoch": 2.5407066052227343,
      "grad_norm": 0.10256543755531311,
      "learning_rate": 6.184772050767094e-06,
      "loss": 0.3932,
      "num_tokens": 1473920662.0,
      "step": 1862
    },
    {
      "epoch": 2.5420720259429936,
      "grad_norm": 0.09115435928106308,
      "learning_rate": 6.172129660017622e-06,
      "loss": 0.4223,
      "num_tokens": 1474762728.0,
      "step": 1863
    },
    {
      "epoch": 2.543437446663253,
      "grad_norm": 0.0955607146024704,
      "learning_rate": 6.159521604536017e-06,
      "loss": 0.4094,
      "num_tokens": 1475525542.0,
      "step": 1864
    },
    {
      "epoch": 2.5448028673835124,
      "grad_norm": 0.09765505790710449,
      "learning_rate": 6.1469479116728315e-06,
      "loss": 0.3994,
      "num_tokens": 1476322593.0,
      "step": 1865
    },
    {
      "epoch": 2.5461682881037717,
      "grad_norm": 0.0914231538772583,
      "learning_rate": 6.13440860870407e-06,
      "loss": 0.4025,
      "num_tokens": 1477148378.0,
      "step": 1866
    },
    {
      "epoch": 2.5475337088240315,
      "grad_norm": 0.09129006415605545,
      "learning_rate": 6.121903722831139e-06,
      "loss": 0.3936,
      "num_tokens": 1478083181.0,
      "step": 1867
    },
    {
      "epoch": 2.548899129544291,
      "grad_norm": 0.09144231677055359,
      "learning_rate": 6.109433281180784e-06,
      "loss": 0.3922,
      "num_tokens": 1478838008.0,
      "step": 1868
    },
    {
      "epoch": 2.5502645502645502,
      "grad_norm": 0.0864814892411232,
      "learning_rate": 6.096997310805033e-06,
      "loss": 0.4107,
      "num_tokens": 1479703039.0,
      "step": 1869
    },
    {
      "epoch": 2.5516299709848096,
      "grad_norm": 0.09731002897024155,
      "learning_rate": 6.0845958386811285e-06,
      "loss": 0.418,
      "num_tokens": 1480533293.0,
      "step": 1870
    },
    {
      "epoch": 2.5529953917050694,
      "grad_norm": 0.08816472440958023,
      "learning_rate": 6.072228891711487e-06,
      "loss": 0.3854,
      "num_tokens": 1481310785.0,
      "step": 1871
    },
    {
      "epoch": 2.5543608124253288,
      "grad_norm": 0.08814267814159393,
      "learning_rate": 6.05989649672362e-06,
      "loss": 0.4073,
      "num_tokens": 1482058492.0,
      "step": 1872
    },
    {
      "epoch": 2.555726233145588,
      "grad_norm": 0.09146502614021301,
      "learning_rate": 6.047598680470086e-06,
      "loss": 0.3954,
      "num_tokens": 1482870871.0,
      "step": 1873
    },
    {
      "epoch": 2.5570916538658475,
      "grad_norm": 0.0928448811173439,
      "learning_rate": 6.035335469628445e-06,
      "loss": 0.3887,
      "num_tokens": 1483675014.0,
      "step": 1874
    },
    {
      "epoch": 2.558457074586107,
      "grad_norm": 0.09056254476308823,
      "learning_rate": 6.023106890801177e-06,
      "loss": 0.4171,
      "num_tokens": 1484492262.0,
      "step": 1875
    },
    {
      "epoch": 2.559822495306366,
      "grad_norm": 0.09000557661056519,
      "learning_rate": 6.010912970515631e-06,
      "loss": 0.408,
      "num_tokens": 1485364477.0,
      "step": 1876
    },
    {
      "epoch": 2.5611879160266255,
      "grad_norm": 0.09254024177789688,
      "learning_rate": 5.9987537352239755e-06,
      "loss": 0.3967,
      "num_tokens": 1486108199.0,
      "step": 1877
    },
    {
      "epoch": 2.562553336746885,
      "grad_norm": 0.09373597800731659,
      "learning_rate": 5.986629211303147e-06,
      "loss": 0.4035,
      "num_tokens": 1486872220.0,
      "step": 1878
    },
    {
      "epoch": 2.5639187574671447,
      "grad_norm": 0.09113255888223648,
      "learning_rate": 5.9745394250547695e-06,
      "loss": 0.4008,
      "num_tokens": 1487716900.0,
      "step": 1879
    },
    {
      "epoch": 2.565284178187404,
      "grad_norm": 0.08792366832494736,
      "learning_rate": 5.962484402705118e-06,
      "loss": 0.4075,
      "num_tokens": 1488516312.0,
      "step": 1880
    },
    {
      "epoch": 2.5666495989076634,
      "grad_norm": 0.09624229371547699,
      "learning_rate": 5.950464170405052e-06,
      "loss": 0.4062,
      "num_tokens": 1489308517.0,
      "step": 1881
    },
    {
      "epoch": 2.568015019627923,
      "grad_norm": 0.08952644467353821,
      "learning_rate": 5.938478754229964e-06,
      "loss": 0.4152,
      "num_tokens": 1490120630.0,
      "step": 1882
    },
    {
      "epoch": 2.569380440348182,
      "grad_norm": 0.09385248273611069,
      "learning_rate": 5.926528180179712e-06,
      "loss": 0.3949,
      "num_tokens": 1490924723.0,
      "step": 1883
    },
    {
      "epoch": 2.570745861068442,
      "grad_norm": 0.09714731574058533,
      "learning_rate": 5.914612474178582e-06,
      "loss": 0.4235,
      "num_tokens": 1491683948.0,
      "step": 1884
    },
    {
      "epoch": 2.5721112817887013,
      "grad_norm": 0.09526418149471283,
      "learning_rate": 5.902731662075215e-06,
      "loss": 0.3995,
      "num_tokens": 1492459019.0,
      "step": 1885
    },
    {
      "epoch": 2.5734767025089607,
      "grad_norm": 0.09683454781770706,
      "learning_rate": 5.890885769642553e-06,
      "loss": 0.4048,
      "num_tokens": 1493237631.0,
      "step": 1886
    },
    {
      "epoch": 2.57484212322922,
      "grad_norm": 0.09574704617261887,
      "learning_rate": 5.879074822577806e-06,
      "loss": 0.4046,
      "num_tokens": 1494015382.0,
      "step": 1887
    },
    {
      "epoch": 2.5762075439494794,
      "grad_norm": 0.08780235797166824,
      "learning_rate": 5.8672988465023515e-06,
      "loss": 0.4178,
      "num_tokens": 1494835052.0,
      "step": 1888
    },
    {
      "epoch": 2.5775729646697387,
      "grad_norm": 0.09431976079940796,
      "learning_rate": 5.855557866961721e-06,
      "loss": 0.3933,
      "num_tokens": 1495583344.0,
      "step": 1889
    },
    {
      "epoch": 2.578938385389998,
      "grad_norm": 0.09019556641578674,
      "learning_rate": 5.843851909425517e-06,
      "loss": 0.396,
      "num_tokens": 1496377377.0,
      "step": 1890
    },
    {
      "epoch": 2.5803038061102574,
      "grad_norm": 0.092668317258358,
      "learning_rate": 5.8321809992873865e-06,
      "loss": 0.4094,
      "num_tokens": 1497205363.0,
      "step": 1891
    },
    {
      "epoch": 2.5816692268305172,
      "grad_norm": 0.09556429088115692,
      "learning_rate": 5.820545161864934e-06,
      "loss": 0.3866,
      "num_tokens": 1497974242.0,
      "step": 1892
    },
    {
      "epoch": 2.5830346475507766,
      "grad_norm": 0.09507754445075989,
      "learning_rate": 5.808944422399689e-06,
      "loss": 0.4174,
      "num_tokens": 1498773633.0,
      "step": 1893
    },
    {
      "epoch": 2.584400068271036,
      "grad_norm": 0.0908997505903244,
      "learning_rate": 5.797378806057029e-06,
      "loss": 0.4002,
      "num_tokens": 1499550674.0,
      "step": 1894
    },
    {
      "epoch": 2.5857654889912953,
      "grad_norm": 0.09233848750591278,
      "learning_rate": 5.785848337926157e-06,
      "loss": 0.4313,
      "num_tokens": 1500399990.0,
      "step": 1895
    },
    {
      "epoch": 2.587130909711555,
      "grad_norm": 0.1018880233168602,
      "learning_rate": 5.774353043020022e-06,
      "loss": 0.4154,
      "num_tokens": 1501211656.0,
      "step": 1896
    },
    {
      "epoch": 2.5884963304318145,
      "grad_norm": 0.09463141858577728,
      "learning_rate": 5.762892946275271e-06,
      "loss": 0.4089,
      "num_tokens": 1501994740.0,
      "step": 1897
    },
    {
      "epoch": 2.589861751152074,
      "grad_norm": 0.09042040258646011,
      "learning_rate": 5.751468072552196e-06,
      "loss": 0.4167,
      "num_tokens": 1502754279.0,
      "step": 1898
    },
    {
      "epoch": 2.591227171872333,
      "grad_norm": 0.10007559508085251,
      "learning_rate": 5.740078446634683e-06,
      "loss": 0.4106,
      "num_tokens": 1503517437.0,
      "step": 1899
    },
    {
      "epoch": 2.5925925925925926,
      "grad_norm": 0.09376079589128494,
      "learning_rate": 5.72872409323015e-06,
      "loss": 0.4165,
      "num_tokens": 1504363030.0,
      "step": 1900
    },
    {
      "epoch": 2.593958013312852,
      "grad_norm": 0.08763644844293594,
      "learning_rate": 5.717405036969504e-06,
      "loss": 0.3924,
      "num_tokens": 1505148136.0,
      "step": 1901
    },
    {
      "epoch": 2.5953234340331113,
      "grad_norm": 0.091402567923069,
      "learning_rate": 5.706121302407077e-06,
      "loss": 0.4103,
      "num_tokens": 1505920947.0,
      "step": 1902
    },
    {
      "epoch": 2.5966888547533706,
      "grad_norm": 0.09371322393417358,
      "learning_rate": 5.694872914020581e-06,
      "loss": 0.4087,
      "num_tokens": 1506701932.0,
      "step": 1903
    },
    {
      "epoch": 2.5980542754736304,
      "grad_norm": 0.09919822961091995,
      "learning_rate": 5.683659896211058e-06,
      "loss": 0.3954,
      "num_tokens": 1507438519.0,
      "step": 1904
    },
    {
      "epoch": 2.59941969619389,
      "grad_norm": 0.09252594411373138,
      "learning_rate": 5.672482273302811e-06,
      "loss": 0.4313,
      "num_tokens": 1508194430.0,
      "step": 1905
    },
    {
      "epoch": 2.600785116914149,
      "grad_norm": 0.09268096834421158,
      "learning_rate": 5.6613400695433605e-06,
      "loss": 0.422,
      "num_tokens": 1509043296.0,
      "step": 1906
    },
    {
      "epoch": 2.6021505376344085,
      "grad_norm": 0.09786684066057205,
      "learning_rate": 5.6502333091033975e-06,
      "loss": 0.387,
      "num_tokens": 1509785666.0,
      "step": 1907
    },
    {
      "epoch": 2.603515958354668,
      "grad_norm": 0.10275842249393463,
      "learning_rate": 5.6391620160767315e-06,
      "loss": 0.4136,
      "num_tokens": 1510664592.0,
      "step": 1908
    },
    {
      "epoch": 2.6048813790749277,
      "grad_norm": 0.09375209361314774,
      "learning_rate": 5.628126214480222e-06,
      "loss": 0.4112,
      "num_tokens": 1511459184.0,
      "step": 1909
    },
    {
      "epoch": 2.606246799795187,
      "grad_norm": 0.08990532159805298,
      "learning_rate": 5.617125928253745e-06,
      "loss": 0.4006,
      "num_tokens": 1512239123.0,
      "step": 1910
    },
    {
      "epoch": 2.6076122205154464,
      "grad_norm": 0.10065276175737381,
      "learning_rate": 5.606161181260123e-06,
      "loss": 0.3882,
      "num_tokens": 1513001998.0,
      "step": 1911
    },
    {
      "epoch": 2.6089776412357057,
      "grad_norm": 0.08765577524900436,
      "learning_rate": 5.595231997285099e-06,
      "loss": 0.3954,
      "num_tokens": 1513808401.0,
      "step": 1912
    },
    {
      "epoch": 2.610343061955965,
      "grad_norm": 0.09887242317199707,
      "learning_rate": 5.584338400037258e-06,
      "loss": 0.3891,
      "num_tokens": 1514534790.0,
      "step": 1913
    },
    {
      "epoch": 2.6117084826762245,
      "grad_norm": 0.09907309710979462,
      "learning_rate": 5.5734804131479894e-06,
      "loss": 0.3981,
      "num_tokens": 1515289686.0,
      "step": 1914
    },
    {
      "epoch": 2.613073903396484,
      "grad_norm": 0.09412331879138947,
      "learning_rate": 5.562658060171433e-06,
      "loss": 0.3991,
      "num_tokens": 1516068561.0,
      "step": 1915
    },
    {
      "epoch": 2.6144393241167436,
      "grad_norm": 0.09290080517530441,
      "learning_rate": 5.551871364584433e-06,
      "loss": 0.3865,
      "num_tokens": 1516912171.0,
      "step": 1916
    },
    {
      "epoch": 2.615804744837003,
      "grad_norm": 0.08947597444057465,
      "learning_rate": 5.541120349786477e-06,
      "loss": 0.4263,
      "num_tokens": 1517728395.0,
      "step": 1917
    },
    {
      "epoch": 2.6171701655572623,
      "grad_norm": 0.09919590502977371,
      "learning_rate": 5.53040503909965e-06,
      "loss": 0.3981,
      "num_tokens": 1518471172.0,
      "step": 1918
    },
    {
      "epoch": 2.6185355862775217,
      "grad_norm": 0.10242892056703568,
      "learning_rate": 5.51972545576859e-06,
      "loss": 0.372,
      "num_tokens": 1519131142.0,
      "step": 1919
    },
    {
      "epoch": 2.619901006997781,
      "grad_norm": 0.09684967249631882,
      "learning_rate": 5.509081622960423e-06,
      "loss": 0.4055,
      "num_tokens": 1519969270.0,
      "step": 1920
    },
    {
      "epoch": 2.621266427718041,
      "grad_norm": 0.0893520936369896,
      "learning_rate": 5.498473563764734e-06,
      "loss": 0.4044,
      "num_tokens": 1520796973.0,
      "step": 1921
    },
    {
      "epoch": 2.6226318484383,
      "grad_norm": 0.08478132635354996,
      "learning_rate": 5.487901301193501e-06,
      "loss": 0.3932,
      "num_tokens": 1521570678.0,
      "step": 1922
    },
    {
      "epoch": 2.6239972691585596,
      "grad_norm": 0.09486637264490128,
      "learning_rate": 5.477364858181034e-06,
      "loss": 0.4182,
      "num_tokens": 1522383638.0,
      "step": 1923
    },
    {
      "epoch": 2.625362689878819,
      "grad_norm": 0.09728098660707474,
      "learning_rate": 5.466864257583956e-06,
      "loss": 0.3931,
      "num_tokens": 1523180444.0,
      "step": 1924
    },
    {
      "epoch": 2.6267281105990783,
      "grad_norm": 0.09627032279968262,
      "learning_rate": 5.456399522181137e-06,
      "loss": 0.4085,
      "num_tokens": 1524012063.0,
      "step": 1925
    },
    {
      "epoch": 2.6280935313193377,
      "grad_norm": 0.09322833269834518,
      "learning_rate": 5.445970674673636e-06,
      "loss": 0.3966,
      "num_tokens": 1524768115.0,
      "step": 1926
    },
    {
      "epoch": 2.629458952039597,
      "grad_norm": 0.09049064666032791,
      "learning_rate": 5.435577737684668e-06,
      "loss": 0.3941,
      "num_tokens": 1525624462.0,
      "step": 1927
    },
    {
      "epoch": 2.6308243727598564,
      "grad_norm": 0.08866015821695328,
      "learning_rate": 5.4252207337595445e-06,
      "loss": 0.4028,
      "num_tokens": 1526424038.0,
      "step": 1928
    },
    {
      "epoch": 2.632189793480116,
      "grad_norm": 0.09025109559297562,
      "learning_rate": 5.414899685365627e-06,
      "loss": 0.3954,
      "num_tokens": 1527181680.0,
      "step": 1929
    },
    {
      "epoch": 2.6335552142003755,
      "grad_norm": 0.0975562185049057,
      "learning_rate": 5.404614614892283e-06,
      "loss": 0.3968,
      "num_tokens": 1527900395.0,
      "step": 1930
    },
    {
      "epoch": 2.634920634920635,
      "grad_norm": 0.09448160231113434,
      "learning_rate": 5.394365544650826e-06,
      "loss": 0.3943,
      "num_tokens": 1528673960.0,
      "step": 1931
    },
    {
      "epoch": 2.6362860556408942,
      "grad_norm": 0.09414353966712952,
      "learning_rate": 5.384152496874483e-06,
      "loss": 0.4281,
      "num_tokens": 1529529147.0,
      "step": 1932
    },
    {
      "epoch": 2.637651476361154,
      "grad_norm": 0.09328985214233398,
      "learning_rate": 5.373975493718329e-06,
      "loss": 0.3901,
      "num_tokens": 1530348806.0,
      "step": 1933
    },
    {
      "epoch": 2.6390168970814134,
      "grad_norm": 0.09540776163339615,
      "learning_rate": 5.363834557259262e-06,
      "loss": 0.4007,
      "num_tokens": 1531139701.0,
      "step": 1934
    },
    {
      "epoch": 2.6403823178016728,
      "grad_norm": 0.09686370939016342,
      "learning_rate": 5.3537297094959236e-06,
      "loss": 0.4084,
      "num_tokens": 1531899693.0,
      "step": 1935
    },
    {
      "epoch": 2.641747738521932,
      "grad_norm": 0.11829400062561035,
      "learning_rate": 5.3436609723486765e-06,
      "loss": 0.4389,
      "num_tokens": 1532694767.0,
      "step": 1936
    },
    {
      "epoch": 2.6431131592421915,
      "grad_norm": 0.09832534939050674,
      "learning_rate": 5.333628367659548e-06,
      "loss": 0.4101,
      "num_tokens": 1533531871.0,
      "step": 1937
    },
    {
      "epoch": 2.644478579962451,
      "grad_norm": 0.09270016849040985,
      "learning_rate": 5.32363191719219e-06,
      "loss": 0.4096,
      "num_tokens": 1534292481.0,
      "step": 1938
    },
    {
      "epoch": 2.64584400068271,
      "grad_norm": 0.09857146441936493,
      "learning_rate": 5.313671642631817e-06,
      "loss": 0.4134,
      "num_tokens": 1535087317.0,
      "step": 1939
    },
    {
      "epoch": 2.6472094214029696,
      "grad_norm": 0.10009737312793732,
      "learning_rate": 5.303747565585169e-06,
      "loss": 0.4042,
      "num_tokens": 1535892499.0,
      "step": 1940
    },
    {
      "epoch": 2.6485748421232294,
      "grad_norm": 0.09206550568342209,
      "learning_rate": 5.2938597075804606e-06,
      "loss": 0.3961,
      "num_tokens": 1536664872.0,
      "step": 1941
    },
    {
      "epoch": 2.6499402628434887,
      "grad_norm": 0.09256400167942047,
      "learning_rate": 5.284008090067348e-06,
      "loss": 0.4125,
      "num_tokens": 1537360127.0,
      "step": 1942
    },
    {
      "epoch": 2.651305683563748,
      "grad_norm": 0.09796247631311417,
      "learning_rate": 5.274192734416857e-06,
      "loss": 0.3932,
      "num_tokens": 1538105143.0,
      "step": 1943
    },
    {
      "epoch": 2.6526711042840074,
      "grad_norm": 0.0905158668756485,
      "learning_rate": 5.2644136619213595e-06,
      "loss": 0.4143,
      "num_tokens": 1538957311.0,
      "step": 1944
    },
    {
      "epoch": 2.654036525004267,
      "grad_norm": 0.09744814783334732,
      "learning_rate": 5.254670893794515e-06,
      "loss": 0.4146,
      "num_tokens": 1539786554.0,
      "step": 1945
    },
    {
      "epoch": 2.6554019457245266,
      "grad_norm": 0.09042836725711823,
      "learning_rate": 5.2449644511712325e-06,
      "loss": 0.3751,
      "num_tokens": 1540582121.0,
      "step": 1946
    },
    {
      "epoch": 2.656767366444786,
      "grad_norm": 0.0917704850435257,
      "learning_rate": 5.235294355107616e-06,
      "loss": 0.4025,
      "num_tokens": 1541370901.0,
      "step": 1947
    },
    {
      "epoch": 2.6581327871650453,
      "grad_norm": 0.09489761292934418,
      "learning_rate": 5.225660626580924e-06,
      "loss": 0.4069,
      "num_tokens": 1542144660.0,
      "step": 1948
    },
    {
      "epoch": 2.6594982078853047,
      "grad_norm": 0.09600350260734558,
      "learning_rate": 5.216063286489525e-06,
      "loss": 0.3841,
      "num_tokens": 1542896064.0,
      "step": 1949
    },
    {
      "epoch": 2.660863628605564,
      "grad_norm": 0.10126474499702454,
      "learning_rate": 5.206502355652844e-06,
      "loss": 0.3964,
      "num_tokens": 1543668149.0,
      "step": 1950
    },
    {
      "epoch": 2.6622290493258234,
      "grad_norm": 0.09096404910087585,
      "learning_rate": 5.196977854811341e-06,
      "loss": 0.3946,
      "num_tokens": 1544485209.0,
      "step": 1951
    },
    {
      "epoch": 2.6635944700460827,
      "grad_norm": 0.09853465110063553,
      "learning_rate": 5.1874898046264255e-06,
      "loss": 0.4288,
      "num_tokens": 1545336605.0,
      "step": 1952
    },
    {
      "epoch": 2.6649598907663425,
      "grad_norm": 0.09498508274555206,
      "learning_rate": 5.178038225680451e-06,
      "loss": 0.4165,
      "num_tokens": 1546100006.0,
      "step": 1953
    },
    {
      "epoch": 2.666325311486602,
      "grad_norm": 0.0963786169886589,
      "learning_rate": 5.168623138476645e-06,
      "loss": 0.3792,
      "num_tokens": 1546833199.0,
      "step": 1954
    },
    {
      "epoch": 2.6676907322068613,
      "grad_norm": 0.08840270340442657,
      "learning_rate": 5.159244563439085e-06,
      "loss": 0.4093,
      "num_tokens": 1547641926.0,
      "step": 1955
    },
    {
      "epoch": 2.6690561529271206,
      "grad_norm": 0.09721392393112183,
      "learning_rate": 5.149902520912631e-06,
      "loss": 0.384,
      "num_tokens": 1548390921.0,
      "step": 1956
    },
    {
      "epoch": 2.67042157364738,
      "grad_norm": 0.09264185279607773,
      "learning_rate": 5.140597031162901e-06,
      "loss": 0.3898,
      "num_tokens": 1549217833.0,
      "step": 1957
    },
    {
      "epoch": 2.67178699436764,
      "grad_norm": 0.09000653773546219,
      "learning_rate": 5.131328114376213e-06,
      "loss": 0.4129,
      "num_tokens": 1550005315.0,
      "step": 1958
    },
    {
      "epoch": 2.673152415087899,
      "grad_norm": 0.09940768033266068,
      "learning_rate": 5.1220957906595556e-06,
      "loss": 0.3791,
      "num_tokens": 1550760419.0,
      "step": 1959
    },
    {
      "epoch": 2.6745178358081585,
      "grad_norm": 0.09012605249881744,
      "learning_rate": 5.112900080040529e-06,
      "loss": 0.4002,
      "num_tokens": 1551515702.0,
      "step": 1960
    },
    {
      "epoch": 2.675883256528418,
      "grad_norm": 0.10186552256345749,
      "learning_rate": 5.103741002467314e-06,
      "loss": 0.3879,
      "num_tokens": 1552258243.0,
      "step": 1961
    },
    {
      "epoch": 2.677248677248677,
      "grad_norm": 0.0952875092625618,
      "learning_rate": 5.094618577808621e-06,
      "loss": 0.3988,
      "num_tokens": 1553044171.0,
      "step": 1962
    },
    {
      "epoch": 2.6786140979689366,
      "grad_norm": 0.09206876158714294,
      "learning_rate": 5.085532825853651e-06,
      "loss": 0.4386,
      "num_tokens": 1553820306.0,
      "step": 1963
    },
    {
      "epoch": 2.679979518689196,
      "grad_norm": 0.12153801321983337,
      "learning_rate": 5.076483766312049e-06,
      "loss": 0.4102,
      "num_tokens": 1554657148.0,
      "step": 1964
    },
    {
      "epoch": 2.6813449394094553,
      "grad_norm": 0.08835148066282272,
      "learning_rate": 5.0674714188138645e-06,
      "loss": 0.4103,
      "num_tokens": 1555521429.0,
      "step": 1965
    },
    {
      "epoch": 2.682710360129715,
      "grad_norm": 0.089751698076725,
      "learning_rate": 5.05849580290951e-06,
      "loss": 0.4095,
      "num_tokens": 1556331459.0,
      "step": 1966
    },
    {
      "epoch": 2.6840757808499744,
      "grad_norm": 0.0921187773346901,
      "learning_rate": 5.0495569380697085e-06,
      "loss": 0.3904,
      "num_tokens": 1557032842.0,
      "step": 1967
    },
    {
      "epoch": 2.685441201570234,
      "grad_norm": 0.0981219932436943,
      "learning_rate": 5.040654843685476e-06,
      "loss": 0.4175,
      "num_tokens": 1557883362.0,
      "step": 1968
    },
    {
      "epoch": 2.686806622290493,
      "grad_norm": 0.08752515912055969,
      "learning_rate": 5.031789539068047e-06,
      "loss": 0.4169,
      "num_tokens": 1558757501.0,
      "step": 1969
    },
    {
      "epoch": 2.688172043010753,
      "grad_norm": 0.09159765392541885,
      "learning_rate": 5.022961043448851e-06,
      "loss": 0.3937,
      "num_tokens": 1559486217.0,
      "step": 1970
    },
    {
      "epoch": 2.6895374637310123,
      "grad_norm": 0.09410414844751358,
      "learning_rate": 5.014169375979469e-06,
      "loss": 0.388,
      "num_tokens": 1560164807.0,
      "step": 1971
    },
    {
      "epoch": 2.6909028844512717,
      "grad_norm": 0.10924460738897324,
      "learning_rate": 5.005414555731598e-06,
      "loss": 0.412,
      "num_tokens": 1560932496.0,
      "step": 1972
    },
    {
      "epoch": 2.692268305171531,
      "grad_norm": 0.09052724391222,
      "learning_rate": 4.996696601696996e-06,
      "loss": 0.3713,
      "num_tokens": 1561606265.0,
      "step": 1973
    },
    {
      "epoch": 2.6936337258917904,
      "grad_norm": 0.09827476739883423,
      "learning_rate": 4.9880155327874445e-06,
      "loss": 0.4158,
      "num_tokens": 1562405033.0,
      "step": 1974
    },
    {
      "epoch": 2.6949991466120498,
      "grad_norm": 0.09681303054094315,
      "learning_rate": 4.97937136783472e-06,
      "loss": 0.3918,
      "num_tokens": 1563247776.0,
      "step": 1975
    },
    {
      "epoch": 2.696364567332309,
      "grad_norm": 0.09270733594894409,
      "learning_rate": 4.97076412559053e-06,
      "loss": 0.3733,
      "num_tokens": 1563973578.0,
      "step": 1976
    },
    {
      "epoch": 2.6977299880525685,
      "grad_norm": 0.10020172595977783,
      "learning_rate": 4.9621938247265e-06,
      "loss": 0.3876,
      "num_tokens": 1564751727.0,
      "step": 1977
    },
    {
      "epoch": 2.6990954087728283,
      "grad_norm": 0.08932332694530487,
      "learning_rate": 4.953660483834112e-06,
      "loss": 0.4165,
      "num_tokens": 1565569657.0,
      "step": 1978
    },
    {
      "epoch": 2.7004608294930876,
      "grad_norm": 0.09466245770454407,
      "learning_rate": 4.9451641214246696e-06,
      "loss": 0.3955,
      "num_tokens": 1566323577.0,
      "step": 1979
    },
    {
      "epoch": 2.701826250213347,
      "grad_norm": 0.08565118163824081,
      "learning_rate": 4.936704755929259e-06,
      "loss": 0.4167,
      "num_tokens": 1567106515.0,
      "step": 1980
    },
    {
      "epoch": 2.7031916709336063,
      "grad_norm": 0.09783415496349335,
      "learning_rate": 4.928282405698719e-06,
      "loss": 0.3943,
      "num_tokens": 1567954858.0,
      "step": 1981
    },
    {
      "epoch": 2.7045570916538657,
      "grad_norm": 0.098345547914505,
      "learning_rate": 4.919897089003578e-06,
      "loss": 0.3786,
      "num_tokens": 1568726012.0,
      "step": 1982
    },
    {
      "epoch": 2.7059225123741255,
      "grad_norm": 0.08808200061321259,
      "learning_rate": 4.911548824034037e-06,
      "loss": 0.4242,
      "num_tokens": 1569646395.0,
      "step": 1983
    },
    {
      "epoch": 2.707287933094385,
      "grad_norm": 0.09163957834243774,
      "learning_rate": 4.903237628899912e-06,
      "loss": 0.409,
      "num_tokens": 1570404897.0,
      "step": 1984
    },
    {
      "epoch": 2.7086533538146442,
      "grad_norm": 0.09441593289375305,
      "learning_rate": 4.8949635216306176e-06,
      "loss": 0.4231,
      "num_tokens": 1571212375.0,
      "step": 1985
    },
    {
      "epoch": 2.7100187745349036,
      "grad_norm": 0.09321702271699905,
      "learning_rate": 4.886726520175104e-06,
      "loss": 0.376,
      "num_tokens": 1571975179.0,
      "step": 1986
    },
    {
      "epoch": 2.711384195255163,
      "grad_norm": 0.08818022906780243,
      "learning_rate": 4.878526642401829e-06,
      "loss": 0.4095,
      "num_tokens": 1572783156.0,
      "step": 1987
    },
    {
      "epoch": 2.7127496159754223,
      "grad_norm": 0.09201285243034363,
      "learning_rate": 4.870363906098718e-06,
      "loss": 0.3929,
      "num_tokens": 1573570494.0,
      "step": 1988
    },
    {
      "epoch": 2.7141150366956817,
      "grad_norm": 0.09344325214624405,
      "learning_rate": 4.862238328973129e-06,
      "loss": 0.4043,
      "num_tokens": 1574401284.0,
      "step": 1989
    },
    {
      "epoch": 2.715480457415941,
      "grad_norm": 0.09428045153617859,
      "learning_rate": 4.854149928651809e-06,
      "loss": 0.4078,
      "num_tokens": 1575246582.0,
      "step": 1990
    },
    {
      "epoch": 2.716845878136201,
      "grad_norm": 0.08725274354219437,
      "learning_rate": 4.846098722680857e-06,
      "loss": 0.403,
      "num_tokens": 1575994728.0,
      "step": 1991
    },
    {
      "epoch": 2.71821129885646,
      "grad_norm": 0.09388323128223419,
      "learning_rate": 4.838084728525688e-06,
      "loss": 0.4058,
      "num_tokens": 1576820749.0,
      "step": 1992
    },
    {
      "epoch": 2.7195767195767195,
      "grad_norm": 0.09039080142974854,
      "learning_rate": 4.830107963570993e-06,
      "loss": 0.4064,
      "num_tokens": 1577624753.0,
      "step": 1993
    },
    {
      "epoch": 2.720942140296979,
      "grad_norm": 0.09470582008361816,
      "learning_rate": 4.822168445120702e-06,
      "loss": 0.4081,
      "num_tokens": 1578381136.0,
      "step": 1994
    },
    {
      "epoch": 2.7223075610172387,
      "grad_norm": 0.09191529452800751,
      "learning_rate": 4.814266190397946e-06,
      "loss": 0.4012,
      "num_tokens": 1579170655.0,
      "step": 1995
    },
    {
      "epoch": 2.723672981737498,
      "grad_norm": 0.09312351047992706,
      "learning_rate": 4.806401216545022e-06,
      "loss": 0.403,
      "num_tokens": 1580011988.0,
      "step": 1996
    },
    {
      "epoch": 2.7250384024577574,
      "grad_norm": 0.09498925507068634,
      "learning_rate": 4.798573540623349e-06,
      "loss": 0.3761,
      "num_tokens": 1580705371.0,
      "step": 1997
    },
    {
      "epoch": 2.7264038231780168,
      "grad_norm": 0.09520500153303146,
      "learning_rate": 4.790783179613446e-06,
      "loss": 0.3889,
      "num_tokens": 1581452864.0,
      "step": 1998
    },
    {
      "epoch": 2.727769243898276,
      "grad_norm": 0.09486683458089828,
      "learning_rate": 4.783030150414873e-06,
      "loss": 0.4166,
      "num_tokens": 1582211888.0,
      "step": 1999
    },
    {
      "epoch": 2.7291346646185355,
      "grad_norm": 0.08806447684764862,
      "learning_rate": 4.775314469846213e-06,
      "loss": 0.4226,
      "num_tokens": 1583069974.0,
      "step": 2000
    },
    {
      "epoch": 2.730500085338795,
      "grad_norm": 0.09183834493160248,
      "learning_rate": 4.767636154645027e-06,
      "loss": 0.4043,
      "num_tokens": 1583861996.0,
      "step": 2001
    },
    {
      "epoch": 2.731865506059054,
      "grad_norm": 0.09277243167161942,
      "learning_rate": 4.759995221467824e-06,
      "loss": 0.4106,
      "num_tokens": 1584609658.0,
      "step": 2002
    },
    {
      "epoch": 2.733230926779314,
      "grad_norm": 0.09872997552156448,
      "learning_rate": 4.752391686890016e-06,
      "loss": 0.4023,
      "num_tokens": 1585374258.0,
      "step": 2003
    },
    {
      "epoch": 2.7345963474995734,
      "grad_norm": 0.09036573022603989,
      "learning_rate": 4.744825567405886e-06,
      "loss": 0.4252,
      "num_tokens": 1586256964.0,
      "step": 2004
    },
    {
      "epoch": 2.7359617682198327,
      "grad_norm": 0.0909745991230011,
      "learning_rate": 4.737296879428551e-06,
      "loss": 0.4216,
      "num_tokens": 1587082041.0,
      "step": 2005
    },
    {
      "epoch": 2.737327188940092,
      "grad_norm": 0.09217943996191025,
      "learning_rate": 4.729805639289938e-06,
      "loss": 0.4122,
      "num_tokens": 1587909953.0,
      "step": 2006
    },
    {
      "epoch": 2.7386926096603514,
      "grad_norm": 0.09452023357152939,
      "learning_rate": 4.7223518632407304e-06,
      "loss": 0.4113,
      "num_tokens": 1588735122.0,
      "step": 2007
    },
    {
      "epoch": 2.7400580303806112,
      "grad_norm": 0.08990570157766342,
      "learning_rate": 4.714935567450339e-06,
      "loss": 0.4274,
      "num_tokens": 1589595156.0,
      "step": 2008
    },
    {
      "epoch": 2.7414234511008706,
      "grad_norm": 0.0898444727063179,
      "learning_rate": 4.7075567680068766e-06,
      "loss": 0.3845,
      "num_tokens": 1590388813.0,
      "step": 2009
    },
    {
      "epoch": 2.74278887182113,
      "grad_norm": 0.09289334714412689,
      "learning_rate": 4.700215480917109e-06,
      "loss": 0.3835,
      "num_tokens": 1591118640.0,
      "step": 2010
    },
    {
      "epoch": 2.7441542925413893,
      "grad_norm": 0.09624753147363663,
      "learning_rate": 4.692911722106433e-06,
      "loss": 0.401,
      "num_tokens": 1591870386.0,
      "step": 2011
    },
    {
      "epoch": 2.7455197132616487,
      "grad_norm": 0.09598765522241592,
      "learning_rate": 4.685645507418829e-06,
      "loss": 0.3984,
      "num_tokens": 1592619405.0,
      "step": 2012
    },
    {
      "epoch": 2.746885133981908,
      "grad_norm": 0.08954776078462601,
      "learning_rate": 4.678416852616837e-06,
      "loss": 0.3827,
      "num_tokens": 1593377849.0,
      "step": 2013
    },
    {
      "epoch": 2.7482505547021674,
      "grad_norm": 0.09620246291160583,
      "learning_rate": 4.671225773381514e-06,
      "loss": 0.3871,
      "num_tokens": 1594145767.0,
      "step": 2014
    },
    {
      "epoch": 2.749615975422427,
      "grad_norm": 0.08828806132078171,
      "learning_rate": 4.664072285312418e-06,
      "loss": 0.3842,
      "num_tokens": 1594937059.0,
      "step": 2015
    },
    {
      "epoch": 2.7509813961426866,
      "grad_norm": 0.09120579063892365,
      "learning_rate": 4.656956403927546e-06,
      "loss": 0.4031,
      "num_tokens": 1595875001.0,
      "step": 2016
    },
    {
      "epoch": 2.752346816862946,
      "grad_norm": 0.08983155339956284,
      "learning_rate": 4.649878144663317e-06,
      "loss": 0.4153,
      "num_tokens": 1596733404.0,
      "step": 2017
    },
    {
      "epoch": 2.7537122375832053,
      "grad_norm": 0.09379485994577408,
      "learning_rate": 4.642837522874542e-06,
      "loss": 0.4224,
      "num_tokens": 1597639120.0,
      "step": 2018
    },
    {
      "epoch": 2.7550776583034646,
      "grad_norm": 0.09259969741106033,
      "learning_rate": 4.635834553834389e-06,
      "loss": 0.4048,
      "num_tokens": 1598434991.0,
      "step": 2019
    },
    {
      "epoch": 2.7564430790237244,
      "grad_norm": 0.09457537531852722,
      "learning_rate": 4.628869252734334e-06,
      "loss": 0.4183,
      "num_tokens": 1599240307.0,
      "step": 2020
    },
    {
      "epoch": 2.757808499743984,
      "grad_norm": 0.0943383276462555,
      "learning_rate": 4.621941634684151e-06,
      "loss": 0.3917,
      "num_tokens": 1600031957.0,
      "step": 2021
    },
    {
      "epoch": 2.759173920464243,
      "grad_norm": 0.09482277929782867,
      "learning_rate": 4.615051714711862e-06,
      "loss": 0.4035,
      "num_tokens": 1600774714.0,
      "step": 2022
    },
    {
      "epoch": 2.7605393411845025,
      "grad_norm": 0.09461604058742523,
      "learning_rate": 4.608199507763714e-06,
      "loss": 0.423,
      "num_tokens": 1601611106.0,
      "step": 2023
    },
    {
      "epoch": 2.761904761904762,
      "grad_norm": 0.09789746254682541,
      "learning_rate": 4.6013850287041444e-06,
      "loss": 0.4202,
      "num_tokens": 1602451588.0,
      "step": 2024
    },
    {
      "epoch": 2.763270182625021,
      "grad_norm": 0.09093061089515686,
      "learning_rate": 4.594608292315743e-06,
      "loss": 0.3952,
      "num_tokens": 1603324342.0,
      "step": 2025
    },
    {
      "epoch": 2.7646356033452806,
      "grad_norm": 0.09152773022651672,
      "learning_rate": 4.5878693132992285e-06,
      "loss": 0.399,
      "num_tokens": 1604063459.0,
      "step": 2026
    },
    {
      "epoch": 2.76600102406554,
      "grad_norm": 0.09745840728282928,
      "learning_rate": 4.581168106273413e-06,
      "loss": 0.3962,
      "num_tokens": 1604814176.0,
      "step": 2027
    },
    {
      "epoch": 2.7673664447857997,
      "grad_norm": 0.09289680421352386,
      "learning_rate": 4.574504685775173e-06,
      "loss": 0.4061,
      "num_tokens": 1605661321.0,
      "step": 2028
    },
    {
      "epoch": 2.768731865506059,
      "grad_norm": 0.0916365459561348,
      "learning_rate": 4.5678790662594055e-06,
      "loss": 0.3981,
      "num_tokens": 1606531858.0,
      "step": 2029
    },
    {
      "epoch": 2.7700972862263185,
      "grad_norm": 0.0859002098441124,
      "learning_rate": 4.5612912620990145e-06,
      "loss": 0.42,
      "num_tokens": 1607369362.0,
      "step": 2030
    },
    {
      "epoch": 2.771462706946578,
      "grad_norm": 0.09188824146986008,
      "learning_rate": 4.5547412875848645e-06,
      "loss": 0.4117,
      "num_tokens": 1608190617.0,
      "step": 2031
    },
    {
      "epoch": 2.7728281276668376,
      "grad_norm": 0.09359250962734222,
      "learning_rate": 4.548229156925768e-06,
      "loss": 0.4155,
      "num_tokens": 1609030832.0,
      "step": 2032
    },
    {
      "epoch": 2.774193548387097,
      "grad_norm": 0.08950609713792801,
      "learning_rate": 4.541754884248436e-06,
      "loss": 0.4284,
      "num_tokens": 1609917716.0,
      "step": 2033
    },
    {
      "epoch": 2.7755589691073563,
      "grad_norm": 0.08520414680242538,
      "learning_rate": 4.535318483597449e-06,
      "loss": 0.4116,
      "num_tokens": 1610745993.0,
      "step": 2034
    },
    {
      "epoch": 2.7769243898276157,
      "grad_norm": 0.09726717323064804,
      "learning_rate": 4.52891996893524e-06,
      "loss": 0.3855,
      "num_tokens": 1611456094.0,
      "step": 2035
    },
    {
      "epoch": 2.778289810547875,
      "grad_norm": 0.09210415184497833,
      "learning_rate": 4.522559354142054e-06,
      "loss": 0.3978,
      "num_tokens": 1612211571.0,
      "step": 2036
    },
    {
      "epoch": 2.7796552312681344,
      "grad_norm": 0.09478045254945755,
      "learning_rate": 4.51623665301592e-06,
      "loss": 0.3905,
      "num_tokens": 1613006719.0,
      "step": 2037
    },
    {
      "epoch": 2.7810206519883938,
      "grad_norm": 0.08838328719139099,
      "learning_rate": 4.50995187927262e-06,
      "loss": 0.398,
      "num_tokens": 1613829458.0,
      "step": 2038
    },
    {
      "epoch": 2.782386072708653,
      "grad_norm": 0.09552766382694244,
      "learning_rate": 4.503705046545664e-06,
      "loss": 0.4039,
      "num_tokens": 1614634838.0,
      "step": 2039
    },
    {
      "epoch": 2.783751493428913,
      "grad_norm": 0.09035251289606094,
      "learning_rate": 4.497496168386252e-06,
      "loss": 0.3815,
      "num_tokens": 1615427165.0,
      "step": 2040
    },
    {
      "epoch": 2.7851169141491723,
      "grad_norm": 0.08838033676147461,
      "learning_rate": 4.491325258263255e-06,
      "loss": 0.4053,
      "num_tokens": 1616196805.0,
      "step": 2041
    },
    {
      "epoch": 2.7864823348694316,
      "grad_norm": 0.09871675074100494,
      "learning_rate": 4.485192329563172e-06,
      "loss": 0.4138,
      "num_tokens": 1616985104.0,
      "step": 2042
    },
    {
      "epoch": 2.787847755589691,
      "grad_norm": 0.0934433713555336,
      "learning_rate": 4.479097395590118e-06,
      "loss": 0.4058,
      "num_tokens": 1617769020.0,
      "step": 2043
    },
    {
      "epoch": 2.7892131763099504,
      "grad_norm": 0.09257467091083527,
      "learning_rate": 4.4730404695657845e-06,
      "loss": 0.4134,
      "num_tokens": 1618598837.0,
      "step": 2044
    },
    {
      "epoch": 2.79057859703021,
      "grad_norm": 0.0954659953713417,
      "learning_rate": 4.467021564629407e-06,
      "loss": 0.3987,
      "num_tokens": 1619424184.0,
      "step": 2045
    },
    {
      "epoch": 2.7919440177504695,
      "grad_norm": 0.09435628354549408,
      "learning_rate": 4.461040693837746e-06,
      "loss": 0.4134,
      "num_tokens": 1620226696.0,
      "step": 2046
    },
    {
      "epoch": 2.793309438470729,
      "grad_norm": 0.09834734350442886,
      "learning_rate": 4.455097870165058e-06,
      "loss": 0.4214,
      "num_tokens": 1621053327.0,
      "step": 2047
    },
    {
      "epoch": 2.7946748591909882,
      "grad_norm": 0.09361906349658966,
      "learning_rate": 4.449193106503059e-06,
      "loss": 0.4028,
      "num_tokens": 1621848103.0,
      "step": 2048
    },
    {
      "epoch": 2.7960402799112476,
      "grad_norm": 0.09105750173330307,
      "learning_rate": 4.4433264156609035e-06,
      "loss": 0.4088,
      "num_tokens": 1622612972.0,
      "step": 2049
    },
    {
      "epoch": 2.797405700631507,
      "grad_norm": 0.09357636421918869,
      "learning_rate": 4.437497810365158e-06,
      "loss": 0.4145,
      "num_tokens": 1623418948.0,
      "step": 2050
    },
    {
      "epoch": 2.7987711213517663,
      "grad_norm": 0.09467504918575287,
      "learning_rate": 4.431707303259766e-06,
      "loss": 0.4054,
      "num_tokens": 1624145656.0,
      "step": 2051
    },
    {
      "epoch": 2.8001365420720257,
      "grad_norm": 0.09104820340871811,
      "learning_rate": 4.425954906906022e-06,
      "loss": 0.4063,
      "num_tokens": 1624914231.0,
      "step": 2052
    },
    {
      "epoch": 2.8015019627922855,
      "grad_norm": 0.09943435341119766,
      "learning_rate": 4.420240633782562e-06,
      "loss": 0.4376,
      "num_tokens": 1625750164.0,
      "step": 2053
    },
    {
      "epoch": 2.802867383512545,
      "grad_norm": 0.09640801697969437,
      "learning_rate": 4.414564496285303e-06,
      "loss": 0.4341,
      "num_tokens": 1626530965.0,
      "step": 2054
    },
    {
      "epoch": 2.804232804232804,
      "grad_norm": 0.09917979687452316,
      "learning_rate": 4.408926506727444e-06,
      "loss": 0.3982,
      "num_tokens": 1627352395.0,
      "step": 2055
    },
    {
      "epoch": 2.8055982249530635,
      "grad_norm": 0.09077587723731995,
      "learning_rate": 4.403326677339432e-06,
      "loss": 0.3997,
      "num_tokens": 1628148642.0,
      "step": 2056
    },
    {
      "epoch": 2.8069636456733233,
      "grad_norm": 0.08447649329900742,
      "learning_rate": 4.397765020268929e-06,
      "loss": 0.3902,
      "num_tokens": 1629027708.0,
      "step": 2057
    },
    {
      "epoch": 2.8083290663935827,
      "grad_norm": 0.09747743606567383,
      "learning_rate": 4.392241547580792e-06,
      "loss": 0.4068,
      "num_tokens": 1629810178.0,
      "step": 2058
    },
    {
      "epoch": 2.809694487113842,
      "grad_norm": 0.09433713555335999,
      "learning_rate": 4.386756271257042e-06,
      "loss": 0.4073,
      "num_tokens": 1630624847.0,
      "step": 2059
    },
    {
      "epoch": 2.8110599078341014,
      "grad_norm": 0.08769740909337997,
      "learning_rate": 4.381309203196847e-06,
      "loss": 0.3959,
      "num_tokens": 1631419698.0,
      "step": 2060
    },
    {
      "epoch": 2.812425328554361,
      "grad_norm": 0.08841016888618469,
      "learning_rate": 4.3759003552164834e-06,
      "loss": 0.39,
      "num_tokens": 1632214402.0,
      "step": 2061
    },
    {
      "epoch": 2.81379074927462,
      "grad_norm": 0.09541250765323639,
      "learning_rate": 4.3705297390493205e-06,
      "loss": 0.4073,
      "num_tokens": 1633037954.0,
      "step": 2062
    },
    {
      "epoch": 2.8151561699948795,
      "grad_norm": 0.09084053337574005,
      "learning_rate": 4.365197366345796e-06,
      "loss": 0.4115,
      "num_tokens": 1633906816.0,
      "step": 2063
    },
    {
      "epoch": 2.816521590715139,
      "grad_norm": 0.09596242755651474,
      "learning_rate": 4.359903248673378e-06,
      "loss": 0.3973,
      "num_tokens": 1634701822.0,
      "step": 2064
    },
    {
      "epoch": 2.8178870114353987,
      "grad_norm": 0.09038528054952621,
      "learning_rate": 4.354647397516551e-06,
      "loss": 0.4209,
      "num_tokens": 1635485228.0,
      "step": 2065
    },
    {
      "epoch": 2.819252432155658,
      "grad_norm": 0.09371691197156906,
      "learning_rate": 4.349429824276789e-06,
      "loss": 0.3972,
      "num_tokens": 1636262854.0,
      "step": 2066
    },
    {
      "epoch": 2.8206178528759174,
      "grad_norm": 0.09717525541782379,
      "learning_rate": 4.344250540272537e-06,
      "loss": 0.4046,
      "num_tokens": 1637049203.0,
      "step": 2067
    },
    {
      "epoch": 2.8219832735961767,
      "grad_norm": 0.09631655365228653,
      "learning_rate": 4.339109556739164e-06,
      "loss": 0.3945,
      "num_tokens": 1637795380.0,
      "step": 2068
    },
    {
      "epoch": 2.823348694316436,
      "grad_norm": 0.09134913235902786,
      "learning_rate": 4.334006884828971e-06,
      "loss": 0.4266,
      "num_tokens": 1638625099.0,
      "step": 2069
    },
    {
      "epoch": 2.824714115036696,
      "grad_norm": 0.0972280278801918,
      "learning_rate": 4.328942535611134e-06,
      "loss": 0.4149,
      "num_tokens": 1639424844.0,
      "step": 2070
    },
    {
      "epoch": 2.8260795357569553,
      "grad_norm": 0.09396138787269592,
      "learning_rate": 4.323916520071709e-06,
      "loss": 0.3928,
      "num_tokens": 1640115370.0,
      "step": 2071
    },
    {
      "epoch": 2.8274449564772146,
      "grad_norm": 0.1011878252029419,
      "learning_rate": 4.318928849113587e-06,
      "loss": 0.3956,
      "num_tokens": 1640843999.0,
      "step": 2072
    },
    {
      "epoch": 2.828810377197474,
      "grad_norm": 0.09429088979959488,
      "learning_rate": 4.313979533556485e-06,
      "loss": 0.3917,
      "num_tokens": 1641623960.0,
      "step": 2073
    },
    {
      "epoch": 2.8301757979177333,
      "grad_norm": 0.09032998979091644,
      "learning_rate": 4.309068584136909e-06,
      "loss": 0.397,
      "num_tokens": 1642414713.0,
      "step": 2074
    },
    {
      "epoch": 2.8315412186379927,
      "grad_norm": 0.0930265337228775,
      "learning_rate": 4.304196011508142e-06,
      "loss": 0.3897,
      "num_tokens": 1643200280.0,
      "step": 2075
    },
    {
      "epoch": 2.832906639358252,
      "grad_norm": 0.09327401965856552,
      "learning_rate": 4.299361826240214e-06,
      "loss": 0.3764,
      "num_tokens": 1643990454.0,
      "step": 2076
    },
    {
      "epoch": 2.834272060078512,
      "grad_norm": 0.10129429399967194,
      "learning_rate": 4.2945660388198866e-06,
      "loss": 0.415,
      "num_tokens": 1644708896.0,
      "step": 2077
    },
    {
      "epoch": 2.835637480798771,
      "grad_norm": 0.09968750178813934,
      "learning_rate": 4.289808659650617e-06,
      "loss": 0.3977,
      "num_tokens": 1645539266.0,
      "step": 2078
    },
    {
      "epoch": 2.8370029015190306,
      "grad_norm": 0.08821015059947968,
      "learning_rate": 4.285089699052549e-06,
      "loss": 0.4024,
      "num_tokens": 1646352931.0,
      "step": 2079
    },
    {
      "epoch": 2.83836832223929,
      "grad_norm": 0.09446611255407333,
      "learning_rate": 4.28040916726249e-06,
      "loss": 0.3779,
      "num_tokens": 1647087485.0,
      "step": 2080
    },
    {
      "epoch": 2.8397337429595493,
      "grad_norm": 0.09654194116592407,
      "learning_rate": 4.275767074433873e-06,
      "loss": 0.396,
      "num_tokens": 1647777620.0,
      "step": 2081
    },
    {
      "epoch": 2.841099163679809,
      "grad_norm": 0.09113694727420807,
      "learning_rate": 4.271163430636755e-06,
      "loss": 0.4043,
      "num_tokens": 1648511710.0,
      "step": 2082
    },
    {
      "epoch": 2.8424645844000684,
      "grad_norm": 0.11268550902605057,
      "learning_rate": 4.2665982458577795e-06,
      "loss": 0.419,
      "num_tokens": 1649334843.0,
      "step": 2083
    },
    {
      "epoch": 2.843830005120328,
      "grad_norm": 0.09686917066574097,
      "learning_rate": 4.262071530000165e-06,
      "loss": 0.4291,
      "num_tokens": 1650222339.0,
      "step": 2084
    },
    {
      "epoch": 2.845195425840587,
      "grad_norm": 0.09184122085571289,
      "learning_rate": 4.2575832928836795e-06,
      "loss": 0.3968,
      "num_tokens": 1651003198.0,
      "step": 2085
    },
    {
      "epoch": 2.8465608465608465,
      "grad_norm": 0.09295312315225601,
      "learning_rate": 4.253133544244619e-06,
      "loss": 0.4047,
      "num_tokens": 1651763189.0,
      "step": 2086
    },
    {
      "epoch": 2.847926267281106,
      "grad_norm": 0.09598802775144577,
      "learning_rate": 4.2487222937357844e-06,
      "loss": 0.407,
      "num_tokens": 1652508824.0,
      "step": 2087
    },
    {
      "epoch": 2.8492916880013652,
      "grad_norm": 0.09482987970113754,
      "learning_rate": 4.2443495509264666e-06,
      "loss": 0.3857,
      "num_tokens": 1653246535.0,
      "step": 2088
    },
    {
      "epoch": 2.8506571087216246,
      "grad_norm": 0.09073173999786377,
      "learning_rate": 4.2400153253024175e-06,
      "loss": 0.4026,
      "num_tokens": 1654071255.0,
      "step": 2089
    },
    {
      "epoch": 2.8520225294418844,
      "grad_norm": 0.09238149970769882,
      "learning_rate": 4.235719626265841e-06,
      "loss": 0.3969,
      "num_tokens": 1654872631.0,
      "step": 2090
    },
    {
      "epoch": 2.8533879501621437,
      "grad_norm": 0.09922967106103897,
      "learning_rate": 4.2314624631353575e-06,
      "loss": 0.4126,
      "num_tokens": 1655625565.0,
      "step": 2091
    },
    {
      "epoch": 2.854753370882403,
      "grad_norm": 0.09538991749286652,
      "learning_rate": 4.227243845146e-06,
      "loss": 0.3938,
      "num_tokens": 1656424783.0,
      "step": 2092
    },
    {
      "epoch": 2.8561187916026625,
      "grad_norm": 0.0923151969909668,
      "learning_rate": 4.223063781449177e-06,
      "loss": 0.4231,
      "num_tokens": 1657248064.0,
      "step": 2093
    },
    {
      "epoch": 2.8574842123229223,
      "grad_norm": 0.0908510833978653,
      "learning_rate": 4.218922281112668e-06,
      "loss": 0.4071,
      "num_tokens": 1658111288.0,
      "step": 2094
    },
    {
      "epoch": 2.8588496330431816,
      "grad_norm": 0.08928512781858444,
      "learning_rate": 4.214819353120597e-06,
      "loss": 0.3884,
      "num_tokens": 1658962442.0,
      "step": 2095
    },
    {
      "epoch": 2.860215053763441,
      "grad_norm": 0.09004784375429153,
      "learning_rate": 4.210755006373406e-06,
      "loss": 0.4086,
      "num_tokens": 1659802633.0,
      "step": 2096
    },
    {
      "epoch": 2.8615804744837003,
      "grad_norm": 0.08389522135257721,
      "learning_rate": 4.2067292496878556e-06,
      "loss": 0.4299,
      "num_tokens": 1660718232.0,
      "step": 2097
    },
    {
      "epoch": 2.8629458952039597,
      "grad_norm": 0.09778501838445663,
      "learning_rate": 4.202742091796985e-06,
      "loss": 0.401,
      "num_tokens": 1661510885.0,
      "step": 2098
    },
    {
      "epoch": 2.864311315924219,
      "grad_norm": 0.0959913432598114,
      "learning_rate": 4.198793541350099e-06,
      "loss": 0.4121,
      "num_tokens": 1662363145.0,
      "step": 2099
    },
    {
      "epoch": 2.8656767366444784,
      "grad_norm": 0.0920393243432045,
      "learning_rate": 4.194883606912757e-06,
      "loss": 0.4043,
      "num_tokens": 1663105905.0,
      "step": 2100
    },
    {
      "epoch": 2.8670421573647378,
      "grad_norm": 0.09331068396568298,
      "learning_rate": 4.191012296966747e-06,
      "loss": 0.4087,
      "num_tokens": 1663971513.0,
      "step": 2101
    },
    {
      "epoch": 2.8684075780849976,
      "grad_norm": 0.08895860612392426,
      "learning_rate": 4.187179619910074e-06,
      "loss": 0.4107,
      "num_tokens": 1664794511.0,
      "step": 2102
    },
    {
      "epoch": 2.869772998805257,
      "grad_norm": 0.09558247774839401,
      "learning_rate": 4.183385584056927e-06,
      "loss": 0.408,
      "num_tokens": 1665578051.0,
      "step": 2103
    },
    {
      "epoch": 2.8711384195255163,
      "grad_norm": 0.09352470189332962,
      "learning_rate": 4.1796301976376785e-06,
      "loss": 0.4314,
      "num_tokens": 1666449291.0,
      "step": 2104
    },
    {
      "epoch": 2.8725038402457757,
      "grad_norm": 0.09197687357664108,
      "learning_rate": 4.175913468798858e-06,
      "loss": 0.4047,
      "num_tokens": 1667238143.0,
      "step": 2105
    },
    {
      "epoch": 2.873869260966035,
      "grad_norm": 0.09440501034259796,
      "learning_rate": 4.172235405603132e-06,
      "loss": 0.3951,
      "num_tokens": 1667995602.0,
      "step": 2106
    },
    {
      "epoch": 2.875234681686295,
      "grad_norm": 0.09551522880792618,
      "learning_rate": 4.168596016029295e-06,
      "loss": 0.4173,
      "num_tokens": 1668839510.0,
      "step": 2107
    },
    {
      "epoch": 2.876600102406554,
      "grad_norm": 0.09023937582969666,
      "learning_rate": 4.164995307972244e-06,
      "loss": 0.409,
      "num_tokens": 1669647788.0,
      "step": 2108
    },
    {
      "epoch": 2.8779655231268135,
      "grad_norm": 0.08833608031272888,
      "learning_rate": 4.161433289242963e-06,
      "loss": 0.4148,
      "num_tokens": 1670454175.0,
      "step": 2109
    },
    {
      "epoch": 2.879330943847073,
      "grad_norm": 0.0948411077260971,
      "learning_rate": 4.157909967568514e-06,
      "loss": 0.4068,
      "num_tokens": 1671243937.0,
      "step": 2110
    },
    {
      "epoch": 2.8806963645673322,
      "grad_norm": 0.09104492515325546,
      "learning_rate": 4.154425350592006e-06,
      "loss": 0.4023,
      "num_tokens": 1672014627.0,
      "step": 2111
    },
    {
      "epoch": 2.8820617852875916,
      "grad_norm": 0.09658384323120117,
      "learning_rate": 4.150979445872593e-06,
      "loss": 0.3989,
      "num_tokens": 1672809026.0,
      "step": 2112
    },
    {
      "epoch": 2.883427206007851,
      "grad_norm": 0.08753466606140137,
      "learning_rate": 4.1475722608854405e-06,
      "loss": 0.3932,
      "num_tokens": 1673566281.0,
      "step": 2113
    },
    {
      "epoch": 2.8847926267281108,
      "grad_norm": 0.09601560980081558,
      "learning_rate": 4.144203803021736e-06,
      "loss": 0.3983,
      "num_tokens": 1674360354.0,
      "step": 2114
    },
    {
      "epoch": 2.88615804744837,
      "grad_norm": 0.0948871597647667,
      "learning_rate": 4.140874079588641e-06,
      "loss": 0.4015,
      "num_tokens": 1675158182.0,
      "step": 2115
    },
    {
      "epoch": 2.8875234681686295,
      "grad_norm": 0.09435639530420303,
      "learning_rate": 4.137583097809299e-06,
      "loss": 0.3948,
      "num_tokens": 1675951277.0,
      "step": 2116
    },
    {
      "epoch": 2.888888888888889,
      "grad_norm": 0.09594330936670303,
      "learning_rate": 4.134330864822808e-06,
      "loss": 0.3994,
      "num_tokens": 1676751624.0,
      "step": 2117
    },
    {
      "epoch": 2.890254309609148,
      "grad_norm": 0.08383087813854218,
      "learning_rate": 4.131117387684213e-06,
      "loss": 0.401,
      "num_tokens": 1677535455.0,
      "step": 2118
    },
    {
      "epoch": 2.891619730329408,
      "grad_norm": 0.09217392653226852,
      "learning_rate": 4.127942673364479e-06,
      "loss": 0.3858,
      "num_tokens": 1678286661.0,
      "step": 2119
    },
    {
      "epoch": 2.8929851510496674,
      "grad_norm": 0.0944477990269661,
      "learning_rate": 4.124806728750492e-06,
      "loss": 0.4006,
      "num_tokens": 1679045179.0,
      "step": 2120
    },
    {
      "epoch": 2.8943505717699267,
      "grad_norm": 0.09479736536741257,
      "learning_rate": 4.121709560645026e-06,
      "loss": 0.4101,
      "num_tokens": 1679839310.0,
      "step": 2121
    },
    {
      "epoch": 2.895715992490186,
      "grad_norm": 0.08741678297519684,
      "learning_rate": 4.118651175766747e-06,
      "loss": 0.405,
      "num_tokens": 1680599507.0,
      "step": 2122
    },
    {
      "epoch": 2.8970814132104454,
      "grad_norm": 0.10161706805229187,
      "learning_rate": 4.115631580750181e-06,
      "loss": 0.4283,
      "num_tokens": 1681327300.0,
      "step": 2123
    },
    {
      "epoch": 2.898446833930705,
      "grad_norm": 0.09556473791599274,
      "learning_rate": 4.112650782145711e-06,
      "loss": 0.4058,
      "num_tokens": 1682163714.0,
      "step": 2124
    },
    {
      "epoch": 2.899812254650964,
      "grad_norm": 0.09553345292806625,
      "learning_rate": 4.1097087864195576e-06,
      "loss": 0.4048,
      "num_tokens": 1682951560.0,
      "step": 2125
    },
    {
      "epoch": 2.9011776753712235,
      "grad_norm": 0.09193833917379379,
      "learning_rate": 4.106805599953769e-06,
      "loss": 0.4112,
      "num_tokens": 1683701534.0,
      "step": 2126
    },
    {
      "epoch": 2.9025430960914833,
      "grad_norm": 0.0909915640950203,
      "learning_rate": 4.103941229046205e-06,
      "loss": 0.3983,
      "num_tokens": 1684492435.0,
      "step": 2127
    },
    {
      "epoch": 2.9039085168117427,
      "grad_norm": 0.09048466384410858,
      "learning_rate": 4.101115679910519e-06,
      "loss": 0.4033,
      "num_tokens": 1685308869.0,
      "step": 2128
    },
    {
      "epoch": 2.905273937532002,
      "grad_norm": 0.0956258624792099,
      "learning_rate": 4.098328958676152e-06,
      "loss": 0.4278,
      "num_tokens": 1686142500.0,
      "step": 2129
    },
    {
      "epoch": 2.9066393582522614,
      "grad_norm": 0.0872219130396843,
      "learning_rate": 4.095581071388315e-06,
      "loss": 0.3954,
      "num_tokens": 1686974546.0,
      "step": 2130
    },
    {
      "epoch": 2.9080047789725207,
      "grad_norm": 0.09244464337825775,
      "learning_rate": 4.092872024007979e-06,
      "loss": 0.4114,
      "num_tokens": 1687727196.0,
      "step": 2131
    },
    {
      "epoch": 2.9093701996927805,
      "grad_norm": 0.1011604517698288,
      "learning_rate": 4.090201822411856e-06,
      "loss": 0.3979,
      "num_tokens": 1688493639.0,
      "step": 2132
    },
    {
      "epoch": 2.91073562041304,
      "grad_norm": 0.09081840515136719,
      "learning_rate": 4.087570472392394e-06,
      "loss": 0.3835,
      "num_tokens": 1689249823.0,
      "step": 2133
    },
    {
      "epoch": 2.9121010411332993,
      "grad_norm": 0.09519986063241959,
      "learning_rate": 4.084977979657756e-06,
      "loss": 0.3887,
      "num_tokens": 1690059560.0,
      "step": 2134
    },
    {
      "epoch": 2.9134664618535586,
      "grad_norm": 0.09231436997652054,
      "learning_rate": 4.082424349831818e-06,
      "loss": 0.4113,
      "num_tokens": 1690828208.0,
      "step": 2135
    },
    {
      "epoch": 2.914831882573818,
      "grad_norm": 0.09134180098772049,
      "learning_rate": 4.079909588454145e-06,
      "loss": 0.4275,
      "num_tokens": 1691684162.0,
      "step": 2136
    },
    {
      "epoch": 2.9161973032940773,
      "grad_norm": 0.09938810020685196,
      "learning_rate": 4.077433700979991e-06,
      "loss": 0.4035,
      "num_tokens": 1692517952.0,
      "step": 2137
    },
    {
      "epoch": 2.9175627240143367,
      "grad_norm": 0.09261612594127655,
      "learning_rate": 4.074996692780277e-06,
      "loss": 0.414,
      "num_tokens": 1693340282.0,
      "step": 2138
    },
    {
      "epoch": 2.9189281447345965,
      "grad_norm": 0.09513507038354874,
      "learning_rate": 4.072598569141582e-06,
      "loss": 0.4169,
      "num_tokens": 1694142182.0,
      "step": 2139
    },
    {
      "epoch": 2.920293565454856,
      "grad_norm": 0.09059052169322968,
      "learning_rate": 4.070239335266142e-06,
      "loss": 0.3704,
      "num_tokens": 1694923772.0,
      "step": 2140
    },
    {
      "epoch": 2.921658986175115,
      "grad_norm": 0.08932448923587799,
      "learning_rate": 4.067918996271815e-06,
      "loss": 0.4119,
      "num_tokens": 1695667705.0,
      "step": 2141
    },
    {
      "epoch": 2.9230244068953746,
      "grad_norm": 0.08721384406089783,
      "learning_rate": 4.0656375571921006e-06,
      "loss": 0.3853,
      "num_tokens": 1696553776.0,
      "step": 2142
    },
    {
      "epoch": 2.924389827615634,
      "grad_norm": 0.08202825486660004,
      "learning_rate": 4.063395022976103e-06,
      "loss": 0.3977,
      "num_tokens": 1697315471.0,
      "step": 2143
    },
    {
      "epoch": 2.9257552483358937,
      "grad_norm": 0.11364448815584183,
      "learning_rate": 4.061191398488533e-06,
      "loss": 0.3916,
      "num_tokens": 1698081277.0,
      "step": 2144
    },
    {
      "epoch": 2.927120669056153,
      "grad_norm": 0.09417442977428436,
      "learning_rate": 4.059026688509696e-06,
      "loss": 0.3983,
      "num_tokens": 1698831346.0,
      "step": 2145
    },
    {
      "epoch": 2.9284860897764124,
      "grad_norm": 0.09235160797834396,
      "learning_rate": 4.0569008977354756e-06,
      "loss": 0.4116,
      "num_tokens": 1699651656.0,
      "step": 2146
    },
    {
      "epoch": 2.929851510496672,
      "grad_norm": 0.09709472954273224,
      "learning_rate": 4.054814030777336e-06,
      "loss": 0.3858,
      "num_tokens": 1700380507.0,
      "step": 2147
    },
    {
      "epoch": 2.931216931216931,
      "grad_norm": 0.0914941132068634,
      "learning_rate": 4.052766092162299e-06,
      "loss": 0.4287,
      "num_tokens": 1701208034.0,
      "step": 2148
    },
    {
      "epoch": 2.9325823519371905,
      "grad_norm": 0.08866321295499802,
      "learning_rate": 4.050757086332941e-06,
      "loss": 0.3988,
      "num_tokens": 1701992867.0,
      "step": 2149
    },
    {
      "epoch": 2.93394777265745,
      "grad_norm": 0.09565865993499756,
      "learning_rate": 4.048787017647381e-06,
      "loss": 0.4246,
      "num_tokens": 1702791943.0,
      "step": 2150
    },
    {
      "epoch": 2.9353131933777092,
      "grad_norm": 0.09348441660404205,
      "learning_rate": 4.046855890379276e-06,
      "loss": 0.4066,
      "num_tokens": 1703597518.0,
      "step": 2151
    },
    {
      "epoch": 2.936678614097969,
      "grad_norm": 0.09682600200176239,
      "learning_rate": 4.0449637087178e-06,
      "loss": 0.3954,
      "num_tokens": 1704359180.0,
      "step": 2152
    },
    {
      "epoch": 2.9380440348182284,
      "grad_norm": 0.09087349474430084,
      "learning_rate": 4.043110476767649e-06,
      "loss": 0.3849,
      "num_tokens": 1705051695.0,
      "step": 2153
    },
    {
      "epoch": 2.9394094555384878,
      "grad_norm": 0.09314211457967758,
      "learning_rate": 4.041296198549024e-06,
      "loss": 0.3943,
      "num_tokens": 1705792572.0,
      "step": 2154
    },
    {
      "epoch": 2.940774876258747,
      "grad_norm": 0.09515301883220673,
      "learning_rate": 4.0395208779976204e-06,
      "loss": 0.405,
      "num_tokens": 1706623182.0,
      "step": 2155
    },
    {
      "epoch": 2.942140296979007,
      "grad_norm": 0.08748797327280045,
      "learning_rate": 4.037784518964631e-06,
      "loss": 0.3988,
      "num_tokens": 1707457312.0,
      "step": 2156
    },
    {
      "epoch": 2.9435057176992663,
      "grad_norm": 0.09419576078653336,
      "learning_rate": 4.0360871252167235e-06,
      "loss": 0.3854,
      "num_tokens": 1708194415.0,
      "step": 2157
    },
    {
      "epoch": 2.9448711384195256,
      "grad_norm": 0.08856400102376938,
      "learning_rate": 4.03442870043604e-06,
      "loss": 0.3996,
      "num_tokens": 1708998796.0,
      "step": 2158
    },
    {
      "epoch": 2.946236559139785,
      "grad_norm": 0.0944274440407753,
      "learning_rate": 4.032809248220187e-06,
      "loss": 0.4098,
      "num_tokens": 1709861251.0,
      "step": 2159
    },
    {
      "epoch": 2.9476019798600444,
      "grad_norm": 0.0910477265715599,
      "learning_rate": 4.031228772082228e-06,
      "loss": 0.4065,
      "num_tokens": 1710645651.0,
      "step": 2160
    },
    {
      "epoch": 2.9489674005803037,
      "grad_norm": 0.09331275522708893,
      "learning_rate": 4.0296872754506785e-06,
      "loss": 0.4175,
      "num_tokens": 1711455263.0,
      "step": 2161
    },
    {
      "epoch": 2.950332821300563,
      "grad_norm": 0.08734722435474396,
      "learning_rate": 4.028184761669494e-06,
      "loss": 0.3923,
      "num_tokens": 1712205340.0,
      "step": 2162
    },
    {
      "epoch": 2.9516982420208224,
      "grad_norm": 0.0937071442604065,
      "learning_rate": 4.026721233998066e-06,
      "loss": 0.3993,
      "num_tokens": 1712971039.0,
      "step": 2163
    },
    {
      "epoch": 2.9530636627410822,
      "grad_norm": 0.09147994220256805,
      "learning_rate": 4.02529669561121e-06,
      "loss": 0.3986,
      "num_tokens": 1713756974.0,
      "step": 2164
    },
    {
      "epoch": 2.9544290834613416,
      "grad_norm": 0.09801165759563446,
      "learning_rate": 4.023911149599168e-06,
      "loss": 0.4154,
      "num_tokens": 1714520674.0,
      "step": 2165
    },
    {
      "epoch": 2.955794504181601,
      "grad_norm": 0.09826762229204178,
      "learning_rate": 4.022564598967591e-06,
      "loss": 0.4019,
      "num_tokens": 1715338937.0,
      "step": 2166
    },
    {
      "epoch": 2.9571599249018603,
      "grad_norm": 0.09634470194578171,
      "learning_rate": 4.021257046637543e-06,
      "loss": 0.4199,
      "num_tokens": 1716076794.0,
      "step": 2167
    },
    {
      "epoch": 2.9585253456221197,
      "grad_norm": 0.10245173424482346,
      "learning_rate": 4.019988495445487e-06,
      "loss": 0.4026,
      "num_tokens": 1716892891.0,
      "step": 2168
    },
    {
      "epoch": 2.9598907663423795,
      "grad_norm": 0.09090927988290787,
      "learning_rate": 4.018758948143276e-06,
      "loss": 0.385,
      "num_tokens": 1717663195.0,
      "step": 2169
    },
    {
      "epoch": 2.961256187062639,
      "grad_norm": 0.08775265514850616,
      "learning_rate": 4.017568407398161e-06,
      "loss": 0.3808,
      "num_tokens": 1718386031.0,
      "step": 2170
    },
    {
      "epoch": 2.962621607782898,
      "grad_norm": 0.09342020004987717,
      "learning_rate": 4.016416875792769e-06,
      "loss": 0.4084,
      "num_tokens": 1719146222.0,
      "step": 2171
    },
    {
      "epoch": 2.9639870285031575,
      "grad_norm": 0.09705204516649246,
      "learning_rate": 4.015304355825112e-06,
      "loss": 0.3945,
      "num_tokens": 1719918485.0,
      "step": 2172
    },
    {
      "epoch": 2.965352449223417,
      "grad_norm": 0.09527206420898438,
      "learning_rate": 4.014230849908567e-06,
      "loss": 0.4113,
      "num_tokens": 1720692597.0,
      "step": 2173
    },
    {
      "epoch": 2.9667178699436763,
      "grad_norm": 0.09667731821537018,
      "learning_rate": 4.0131963603718825e-06,
      "loss": 0.3803,
      "num_tokens": 1721426604.0,
      "step": 2174
    },
    {
      "epoch": 2.9680832906639356,
      "grad_norm": 0.10198929905891418,
      "learning_rate": 4.012200889459169e-06,
      "loss": 0.4247,
      "num_tokens": 1722280213.0,
      "step": 2175
    },
    {
      "epoch": 2.9694487113841954,
      "grad_norm": 0.09083274751901627,
      "learning_rate": 4.01124443932989e-06,
      "loss": 0.4025,
      "num_tokens": 1723103374.0,
      "step": 2176
    },
    {
      "epoch": 2.9708141321044548,
      "grad_norm": 0.09526471793651581,
      "learning_rate": 4.01032701205887e-06,
      "loss": 0.3942,
      "num_tokens": 1723862459.0,
      "step": 2177
    },
    {
      "epoch": 2.972179552824714,
      "grad_norm": 0.09120840579271317,
      "learning_rate": 4.0094486096362705e-06,
      "loss": 0.4166,
      "num_tokens": 1724626937.0,
      "step": 2178
    },
    {
      "epoch": 2.9735449735449735,
      "grad_norm": 0.0939774364233017,
      "learning_rate": 4.008609233967607e-06,
      "loss": 0.4045,
      "num_tokens": 1725453645.0,
      "step": 2179
    },
    {
      "epoch": 2.974910394265233,
      "grad_norm": 0.08881974965333939,
      "learning_rate": 4.007808886873731e-06,
      "loss": 0.3803,
      "num_tokens": 1726229406.0,
      "step": 2180
    },
    {
      "epoch": 2.9762758149854927,
      "grad_norm": 0.09174589067697525,
      "learning_rate": 4.007047570090823e-06,
      "loss": 0.406,
      "num_tokens": 1727025429.0,
      "step": 2181
    },
    {
      "epoch": 2.977641235705752,
      "grad_norm": 0.09273260831832886,
      "learning_rate": 4.006325285270407e-06,
      "loss": 0.4189,
      "num_tokens": 1727929729.0,
      "step": 2182
    },
    {
      "epoch": 2.9790066564260114,
      "grad_norm": 0.09305782616138458,
      "learning_rate": 4.0056420339793265e-06,
      "loss": 0.3955,
      "num_tokens": 1728716985.0,
      "step": 2183
    },
    {
      "epoch": 2.9803720771462707,
      "grad_norm": 0.09278541803359985,
      "learning_rate": 4.004997817699755e-06,
      "loss": 0.4064,
      "num_tokens": 1729525589.0,
      "step": 2184
    },
    {
      "epoch": 2.98173749786653,
      "grad_norm": 0.09437207877635956,
      "learning_rate": 4.004392637829182e-06,
      "loss": 0.4022,
      "num_tokens": 1730283312.0,
      "step": 2185
    },
    {
      "epoch": 2.9831029185867894,
      "grad_norm": 0.09564542770385742,
      "learning_rate": 4.003826495680423e-06,
      "loss": 0.4009,
      "num_tokens": 1731154422.0,
      "step": 2186
    },
    {
      "epoch": 2.984468339307049,
      "grad_norm": 0.08786697685718536,
      "learning_rate": 4.003299392481602e-06,
      "loss": 0.4132,
      "num_tokens": 1732022738.0,
      "step": 2187
    },
    {
      "epoch": 2.985833760027308,
      "grad_norm": 0.0956958681344986,
      "learning_rate": 4.002811329376164e-06,
      "loss": 0.4063,
      "num_tokens": 1732813003.0,
      "step": 2188
    },
    {
      "epoch": 2.987199180747568,
      "grad_norm": 0.0881400853395462,
      "learning_rate": 4.002362307422857e-06,
      "loss": 0.3883,
      "num_tokens": 1733578594.0,
      "step": 2189
    },
    {
      "epoch": 2.9885646014678273,
      "grad_norm": 0.0908321887254715,
      "learning_rate": 4.00195232759574e-06,
      "loss": 0.4091,
      "num_tokens": 1734401521.0,
      "step": 2190
    },
    {
      "epoch": 2.9899300221880867,
      "grad_norm": 0.09049002826213837,
      "learning_rate": 4.00158139078418e-06,
      "loss": 0.421,
      "num_tokens": 1735202088.0,
      "step": 2191
    },
    {
      "epoch": 2.991295442908346,
      "grad_norm": 0.09417341649532318,
      "learning_rate": 4.001249497792849e-06,
      "loss": 0.4082,
      "num_tokens": 1736034050.0,
      "step": 2192
    },
    {
      "epoch": 2.992660863628606,
      "grad_norm": 0.09101802110671997,
      "learning_rate": 4.000956649341716e-06,
      "loss": 0.3951,
      "num_tokens": 1736815019.0,
      "step": 2193
    },
    {
      "epoch": 2.994026284348865,
      "grad_norm": 0.09183265268802643,
      "learning_rate": 4.000702846066058e-06,
      "loss": 0.4056,
      "num_tokens": 1737625996.0,
      "step": 2194
    },
    {
      "epoch": 2.9953917050691246,
      "grad_norm": 0.09251420944929123,
      "learning_rate": 4.000488088516446e-06,
      "loss": 0.3915,
      "num_tokens": 1738415506.0,
      "step": 2195
    },
    {
      "epoch": 2.996757125789384,
      "grad_norm": 0.09131419658660889,
      "learning_rate": 4.000312377158754e-06,
      "loss": 0.3965,
      "num_tokens": 1739198066.0,
      "step": 2196
    },
    {
      "epoch": 2.9981225465096433,
      "grad_norm": 0.09083900600671768,
      "learning_rate": 4.000175712374149e-06,
      "loss": 0.4042,
      "num_tokens": 1739947353.0,
      "step": 2197
    },
    {
      "epoch": 2.9994879672299026,
      "grad_norm": 0.09619448333978653,
      "learning_rate": 4.000078094459098e-06,
      "loss": 0.3959,
      "num_tokens": 1740648522.0,
      "step": 2198
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.14404502511024475,
      "learning_rate": 4.000019523625363e-06,
      "loss": 0.3991,
      "num_tokens": 1740826744.0,
      "step": 2199
    },
    {
      "epoch": 3.0,
      "step": 2199,
      "total_flos": 1.8348642473640722e+19,
      "train_loss": 0.460054492993808,
      "train_runtime": 22248.7222,
      "train_samples_per_second": 12.639,
      "train_steps_per_second": 0.099
    }
  ],
  "logging_steps": 1,
  "max_steps": 2199,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.8348642473640722e+19,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}