{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 4.0,
  "eval_steps": 500,
  "global_step": 1672,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0023923444976076554,
      "grad_norm": 6.9691807760923075,
      "learning_rate": 0.0,
      "loss": 1.7686,
      "num_tokens": 224939.0,
      "step": 1
    },
    {
      "epoch": 0.004784688995215311,
      "grad_norm": 5.696067520260917,
      "learning_rate": 1.9607843137254904e-07,
      "loss": 1.7074,
      "num_tokens": 466553.0,
      "step": 2
    },
    {
      "epoch": 0.007177033492822967,
      "grad_norm": 7.218883969227019,
      "learning_rate": 3.921568627450981e-07,
      "loss": 1.7878,
      "num_tokens": 649823.0,
      "step": 3
    },
    {
      "epoch": 0.009569377990430622,
      "grad_norm": 6.599319930344236,
      "learning_rate": 5.882352941176471e-07,
      "loss": 1.776,
      "num_tokens": 862983.0,
      "step": 4
    },
    {
      "epoch": 0.011961722488038277,
      "grad_norm": 6.600002893691311,
      "learning_rate": 7.843137254901962e-07,
      "loss": 1.8101,
      "num_tokens": 1090598.0,
      "step": 5
    },
    {
      "epoch": 0.014354066985645933,
      "grad_norm": 6.1947461987357215,
      "learning_rate": 9.80392156862745e-07,
      "loss": 1.7604,
      "num_tokens": 1309816.0,
      "step": 6
    },
    {
      "epoch": 0.01674641148325359,
      "grad_norm": 7.322579343479808,
      "learning_rate": 1.1764705882352942e-06,
      "loss": 1.7636,
      "num_tokens": 1501898.0,
      "step": 7
    },
    {
      "epoch": 0.019138755980861243,
      "grad_norm": 5.565852999878923,
      "learning_rate": 1.3725490196078434e-06,
      "loss": 1.747,
      "num_tokens": 1690771.0,
      "step": 8
    },
    {
      "epoch": 0.0215311004784689,
      "grad_norm": 6.293532003483547,
      "learning_rate": 1.5686274509803923e-06,
      "loss": 1.745,
      "num_tokens": 1900964.0,
      "step": 9
    },
    {
      "epoch": 0.023923444976076555,
      "grad_norm": 6.163171111681885,
      "learning_rate": 1.7647058823529414e-06,
      "loss": 1.744,
      "num_tokens": 2072545.0,
      "step": 10
    },
    {
      "epoch": 0.02631578947368421,
      "grad_norm": 4.642248066242216,
      "learning_rate": 1.96078431372549e-06,
      "loss": 1.6107,
      "num_tokens": 2230117.0,
      "step": 11
    },
    {
      "epoch": 0.028708133971291867,
      "grad_norm": 2.938871677963385,
      "learning_rate": 2.1568627450980393e-06,
      "loss": 1.6337,
      "num_tokens": 2418536.0,
      "step": 12
    },
    {
      "epoch": 0.03110047846889952,
      "grad_norm": 3.352461014645353,
      "learning_rate": 2.3529411764705885e-06,
      "loss": 1.6248,
      "num_tokens": 2610436.0,
      "step": 13
    },
    {
      "epoch": 0.03349282296650718,
      "grad_norm": 2.4423792859075024,
      "learning_rate": 2.549019607843137e-06,
      "loss": 1.5582,
      "num_tokens": 2840410.0,
      "step": 14
    },
    {
      "epoch": 0.03588516746411483,
      "grad_norm": 3.6326614999386093,
      "learning_rate": 2.7450980392156867e-06,
      "loss": 1.4153,
      "num_tokens": 3039241.0,
      "step": 15
    },
    {
      "epoch": 0.03827751196172249,
      "grad_norm": 2.309721427859542,
      "learning_rate": 2.9411764705882355e-06,
      "loss": 1.4088,
      "num_tokens": 3284539.0,
      "step": 16
    },
    {
      "epoch": 0.04066985645933014,
      "grad_norm": 2.8925752012878223,
      "learning_rate": 3.1372549019607846e-06,
      "loss": 1.5011,
      "num_tokens": 3485634.0,
      "step": 17
    },
    {
      "epoch": 0.0430622009569378,
      "grad_norm": 2.4261545230518626,
      "learning_rate": 3.3333333333333333e-06,
      "loss": 1.4527,
      "num_tokens": 3672944.0,
      "step": 18
    },
    {
      "epoch": 0.045454545454545456,
      "grad_norm": 4.355994016148595,
      "learning_rate": 3.529411764705883e-06,
      "loss": 1.2766,
      "num_tokens": 3843273.0,
      "step": 19
    },
    {
      "epoch": 0.04784688995215311,
      "grad_norm": 4.628861544958714,
      "learning_rate": 3.7254901960784316e-06,
      "loss": 1.2748,
      "num_tokens": 4009616.0,
      "step": 20
    },
    {
      "epoch": 0.050239234449760764,
      "grad_norm": 2.2063220407104116,
      "learning_rate": 3.92156862745098e-06,
      "loss": 1.284,
      "num_tokens": 4196040.0,
      "step": 21
    },
    {
      "epoch": 0.05263157894736842,
      "grad_norm": 2.083766163581579,
      "learning_rate": 4.11764705882353e-06,
      "loss": 1.3234,
      "num_tokens": 4426192.0,
      "step": 22
    },
    {
      "epoch": 0.05502392344497608,
      "grad_norm": 1.4244871277633704,
      "learning_rate": 4.313725490196079e-06,
      "loss": 1.2551,
      "num_tokens": 4692793.0,
      "step": 23
    },
    {
      "epoch": 0.05741626794258373,
      "grad_norm": 1.4458640545451162,
      "learning_rate": 4.509803921568628e-06,
      "loss": 1.2925,
      "num_tokens": 4922578.0,
      "step": 24
    },
    {
      "epoch": 0.05980861244019139,
      "grad_norm": 1.2401785717364968,
      "learning_rate": 4.705882352941177e-06,
      "loss": 1.2988,
      "num_tokens": 5103273.0,
      "step": 25
    },
    {
      "epoch": 0.06220095693779904,
      "grad_norm": 1.2308788644379343,
      "learning_rate": 4.901960784313726e-06,
      "loss": 1.1625,
      "num_tokens": 5321035.0,
      "step": 26
    },
    {
      "epoch": 0.0645933014354067,
      "grad_norm": 1.070148340321906,
      "learning_rate": 5.098039215686274e-06,
      "loss": 1.1461,
      "num_tokens": 5559536.0,
      "step": 27
    },
    {
      "epoch": 0.06698564593301436,
      "grad_norm": 1.03341304889324,
      "learning_rate": 5.294117647058824e-06,
      "loss": 1.1232,
      "num_tokens": 5746590.0,
      "step": 28
    },
    {
      "epoch": 0.06937799043062201,
      "grad_norm": 1.0269837046140493,
      "learning_rate": 5.4901960784313735e-06,
      "loss": 1.1695,
      "num_tokens": 5978411.0,
      "step": 29
    },
    {
      "epoch": 0.07177033492822966,
      "grad_norm": 1.188053126912019,
      "learning_rate": 5.686274509803922e-06,
      "loss": 1.1733,
      "num_tokens": 6204023.0,
      "step": 30
    },
    {
      "epoch": 0.07416267942583732,
      "grad_norm": 1.3792751984185245,
      "learning_rate": 5.882352941176471e-06,
      "loss": 1.1271,
      "num_tokens": 6418366.0,
      "step": 31
    },
    {
      "epoch": 0.07655502392344497,
      "grad_norm": 1.2984724277479058,
      "learning_rate": 6.07843137254902e-06,
      "loss": 1.0451,
      "num_tokens": 6613243.0,
      "step": 32
    },
    {
      "epoch": 0.07894736842105263,
      "grad_norm": 0.9146841559197153,
      "learning_rate": 6.274509803921569e-06,
      "loss": 1.2392,
      "num_tokens": 6833041.0,
      "step": 33
    },
    {
      "epoch": 0.08133971291866028,
      "grad_norm": 1.0033903976372076,
      "learning_rate": 6.470588235294119e-06,
      "loss": 1.079,
      "num_tokens": 7030942.0,
      "step": 34
    },
    {
      "epoch": 0.08373205741626795,
      "grad_norm": 1.1388077906522334,
      "learning_rate": 6.666666666666667e-06,
      "loss": 1.0549,
      "num_tokens": 7237519.0,
      "step": 35
    },
    {
      "epoch": 0.0861244019138756,
      "grad_norm": 0.9842459198671744,
      "learning_rate": 6.862745098039216e-06,
      "loss": 1.0202,
      "num_tokens": 7435884.0,
      "step": 36
    },
    {
      "epoch": 0.08851674641148326,
      "grad_norm": 284.7649201492139,
      "learning_rate": 7.058823529411766e-06,
      "loss": 1.0647,
      "num_tokens": 7635319.0,
      "step": 37
    },
    {
      "epoch": 0.09090909090909091,
      "grad_norm": 0.9557715780751038,
      "learning_rate": 7.2549019607843145e-06,
      "loss": 1.1426,
      "num_tokens": 7862270.0,
      "step": 38
    },
    {
      "epoch": 0.09330143540669857,
      "grad_norm": 0.915320008451191,
      "learning_rate": 7.450980392156863e-06,
      "loss": 1.1608,
      "num_tokens": 8052102.0,
      "step": 39
    },
    {
      "epoch": 0.09569377990430622,
      "grad_norm": 0.8989126142994759,
      "learning_rate": 7.647058823529411e-06,
      "loss": 1.0364,
      "num_tokens": 8239372.0,
      "step": 40
    },
    {
      "epoch": 0.09808612440191387,
      "grad_norm": 0.8329537974290705,
      "learning_rate": 7.84313725490196e-06,
      "loss": 1.1433,
      "num_tokens": 8465829.0,
      "step": 41
    },
    {
      "epoch": 0.10047846889952153,
      "grad_norm": 0.8398997141542524,
      "learning_rate": 8.03921568627451e-06,
      "loss": 1.156,
      "num_tokens": 8642972.0,
      "step": 42
    },
    {
      "epoch": 0.10287081339712918,
      "grad_norm": 0.7585422097649339,
      "learning_rate": 8.23529411764706e-06,
      "loss": 1.0503,
      "num_tokens": 8847697.0,
      "step": 43
    },
    {
      "epoch": 0.10526315789473684,
      "grad_norm": 0.7622670839009824,
      "learning_rate": 8.43137254901961e-06,
      "loss": 0.9813,
      "num_tokens": 9066935.0,
      "step": 44
    },
    {
      "epoch": 0.1076555023923445,
      "grad_norm": 0.8193765838996492,
      "learning_rate": 8.627450980392157e-06,
      "loss": 1.0981,
      "num_tokens": 9284399.0,
      "step": 45
    },
    {
      "epoch": 0.11004784688995216,
      "grad_norm": 0.7925794880440407,
      "learning_rate": 8.823529411764707e-06,
      "loss": 1.0236,
      "num_tokens": 9508366.0,
      "step": 46
    },
    {
      "epoch": 0.11244019138755981,
      "grad_norm": 1.2045774927138393,
      "learning_rate": 9.019607843137256e-06,
      "loss": 1.0526,
      "num_tokens": 9694057.0,
      "step": 47
    },
    {
      "epoch": 0.11483253588516747,
      "grad_norm": 0.9084407634375898,
      "learning_rate": 9.215686274509804e-06,
      "loss": 1.0955,
      "num_tokens": 9929798.0,
      "step": 48
    },
    {
      "epoch": 0.11722488038277512,
      "grad_norm": 0.8240735702666425,
      "learning_rate": 9.411764705882354e-06,
      "loss": 0.9807,
      "num_tokens": 10180029.0,
      "step": 49
    },
    {
      "epoch": 0.11961722488038277,
      "grad_norm": 0.8097524297757899,
      "learning_rate": 9.607843137254903e-06,
      "loss": 1.1265,
      "num_tokens": 10419750.0,
      "step": 50
    },
    {
      "epoch": 0.12200956937799043,
      "grad_norm": 0.8353633432968253,
      "learning_rate": 9.803921568627451e-06,
      "loss": 1.1419,
      "num_tokens": 10605541.0,
      "step": 51
    },
    {
      "epoch": 0.12440191387559808,
      "grad_norm": 0.817154674246075,
      "learning_rate": 1e-05,
      "loss": 1.0961,
      "num_tokens": 10814676.0,
      "step": 52
    },
    {
      "epoch": 0.12679425837320574,
      "grad_norm": 0.8612859746077591,
      "learning_rate": 9.999991548844401e-06,
      "loss": 1.1244,
      "num_tokens": 11035241.0,
      "step": 53
    },
    {
      "epoch": 0.1291866028708134,
      "grad_norm": 0.8154431514013905,
      "learning_rate": 9.999966195409343e-06,
      "loss": 0.9722,
      "num_tokens": 11195242.0,
      "step": 54
    },
    {
      "epoch": 0.13157894736842105,
      "grad_norm": 0.7860112239586271,
      "learning_rate": 9.99992393979006e-06,
      "loss": 1.0957,
      "num_tokens": 11401885.0,
      "step": 55
    },
    {
      "epoch": 0.1339712918660287,
      "grad_norm": 0.7767268725169872,
      "learning_rate": 9.999864782145264e-06,
      "loss": 0.901,
      "num_tokens": 11587966.0,
      "step": 56
    },
    {
      "epoch": 0.13636363636363635,
      "grad_norm": 0.9737729561858515,
      "learning_rate": 9.999788722697156e-06,
      "loss": 1.2394,
      "num_tokens": 11765601.0,
      "step": 57
    },
    {
      "epoch": 0.13875598086124402,
      "grad_norm": 0.8847294438412543,
      "learning_rate": 9.999695761731418e-06,
      "loss": 0.9228,
      "num_tokens": 11942390.0,
      "step": 58
    },
    {
      "epoch": 0.14114832535885166,
      "grad_norm": 0.7337378919952546,
      "learning_rate": 9.999585899597224e-06,
      "loss": 1.0978,
      "num_tokens": 12165372.0,
      "step": 59
    },
    {
      "epoch": 0.14354066985645933,
      "grad_norm": 0.7638475691265573,
      "learning_rate": 9.999459136707217e-06,
      "loss": 0.9959,
      "num_tokens": 12414971.0,
      "step": 60
    },
    {
      "epoch": 0.145933014354067,
      "grad_norm": 0.7949985672748343,
      "learning_rate": 9.999315473537532e-06,
      "loss": 1.0083,
      "num_tokens": 12595239.0,
      "step": 61
    },
    {
      "epoch": 0.14832535885167464,
      "grad_norm": 0.8626640964385176,
      "learning_rate": 9.999154910627776e-06,
      "loss": 1.1222,
      "num_tokens": 12789753.0,
      "step": 62
    },
    {
      "epoch": 0.1507177033492823,
      "grad_norm": 0.7389387597807686,
      "learning_rate": 9.998977448581031e-06,
      "loss": 0.9927,
      "num_tokens": 13029643.0,
      "step": 63
    },
    {
      "epoch": 0.15311004784688995,
      "grad_norm": 0.9337215647236915,
      "learning_rate": 9.998783088063864e-06,
      "loss": 0.6957,
      "num_tokens": 13155991.0,
      "step": 64
    },
    {
      "epoch": 0.15550239234449761,
      "grad_norm": 0.8247995409977257,
      "learning_rate": 9.998571829806301e-06,
      "loss": 1.0184,
      "num_tokens": 13310880.0,
      "step": 65
    },
    {
      "epoch": 0.15789473684210525,
      "grad_norm": 0.8340681998808398,
      "learning_rate": 9.998343674601845e-06,
      "loss": 1.0168,
      "num_tokens": 13489712.0,
      "step": 66
    },
    {
      "epoch": 0.16028708133971292,
      "grad_norm": 0.8370767723991338,
      "learning_rate": 9.998098623307461e-06,
      "loss": 0.9239,
      "num_tokens": 13684647.0,
      "step": 67
    },
    {
      "epoch": 0.16267942583732056,
      "grad_norm": 0.7288813191624135,
      "learning_rate": 9.99783667684358e-06,
      "loss": 1.1175,
      "num_tokens": 13907099.0,
      "step": 68
    },
    {
      "epoch": 0.16507177033492823,
      "grad_norm": 0.8504439815246466,
      "learning_rate": 9.997557836194091e-06,
      "loss": 0.9208,
      "num_tokens": 14130272.0,
      "step": 69
    },
    {
      "epoch": 0.1674641148325359,
      "grad_norm": 0.8451053143388385,
      "learning_rate": 9.997262102406337e-06,
      "loss": 1.0591,
      "num_tokens": 14346591.0,
      "step": 70
    },
    {
      "epoch": 0.16985645933014354,
      "grad_norm": 0.7361210479126754,
      "learning_rate": 9.996949476591117e-06,
      "loss": 0.8635,
      "num_tokens": 14560238.0,
      "step": 71
    },
    {
      "epoch": 0.1722488038277512,
      "grad_norm": 0.8577939566262116,
      "learning_rate": 9.996619959922675e-06,
      "loss": 0.8466,
      "num_tokens": 14755604.0,
      "step": 72
    },
    {
      "epoch": 0.17464114832535885,
      "grad_norm": 0.8381521666289012,
      "learning_rate": 9.996273553638695e-06,
      "loss": 0.7695,
      "num_tokens": 14933946.0,
      "step": 73
    },
    {
      "epoch": 0.17703349282296652,
      "grad_norm": 0.7754686982261173,
      "learning_rate": 9.995910259040306e-06,
      "loss": 0.9558,
      "num_tokens": 15113851.0,
      "step": 74
    },
    {
      "epoch": 0.17942583732057416,
      "grad_norm": 0.7725675853421875,
      "learning_rate": 9.99553007749207e-06,
      "loss": 0.9318,
      "num_tokens": 15317677.0,
      "step": 75
    },
    {
      "epoch": 0.18181818181818182,
      "grad_norm": 0.7639475931469095,
      "learning_rate": 9.995133010421968e-06,
      "loss": 1.0368,
      "num_tokens": 15509536.0,
      "step": 76
    },
    {
      "epoch": 0.18421052631578946,
      "grad_norm": 0.8319629992606972,
      "learning_rate": 9.99471905932142e-06,
      "loss": 0.9003,
      "num_tokens": 15715725.0,
      "step": 77
    },
    {
      "epoch": 0.18660287081339713,
      "grad_norm": 0.7882397528734394,
      "learning_rate": 9.994288225745249e-06,
      "loss": 0.9092,
      "num_tokens": 15952916.0,
      "step": 78
    },
    {
      "epoch": 0.18899521531100477,
      "grad_norm": 0.770940813293932,
      "learning_rate": 9.993840511311695e-06,
      "loss": 0.9546,
      "num_tokens": 16118037.0,
      "step": 79
    },
    {
      "epoch": 0.19138755980861244,
      "grad_norm": 0.7714491446159815,
      "learning_rate": 9.993375917702409e-06,
      "loss": 0.9709,
      "num_tokens": 16304381.0,
      "step": 80
    },
    {
      "epoch": 0.1937799043062201,
      "grad_norm": 0.7786320649310523,
      "learning_rate": 9.992894446662434e-06,
      "loss": 1.0758,
      "num_tokens": 16485104.0,
      "step": 81
    },
    {
      "epoch": 0.19617224880382775,
      "grad_norm": 0.7798209889697421,
      "learning_rate": 9.992396100000208e-06,
      "loss": 0.9808,
      "num_tokens": 16658375.0,
      "step": 82
    },
    {
      "epoch": 0.19856459330143542,
      "grad_norm": 0.7546944385505975,
      "learning_rate": 9.99188087958756e-06,
      "loss": 1.0092,
      "num_tokens": 16839821.0,
      "step": 83
    },
    {
      "epoch": 0.20095693779904306,
      "grad_norm": 0.774709240553806,
      "learning_rate": 9.991348787359686e-06,
      "loss": 0.9415,
      "num_tokens": 17010092.0,
      "step": 84
    },
    {
      "epoch": 0.20334928229665072,
      "grad_norm": 0.860230274595423,
      "learning_rate": 9.990799825315166e-06,
      "loss": 0.8845,
      "num_tokens": 17171471.0,
      "step": 85
    },
    {
      "epoch": 0.20574162679425836,
      "grad_norm": 0.821110135970871,
      "learning_rate": 9.990233995515939e-06,
      "loss": 0.7484,
      "num_tokens": 17349168.0,
      "step": 86
    },
    {
      "epoch": 0.20813397129186603,
      "grad_norm": 0.7685272559332614,
      "learning_rate": 9.989651300087301e-06,
      "loss": 1.0519,
      "num_tokens": 17560749.0,
      "step": 87
    },
    {
      "epoch": 0.21052631578947367,
      "grad_norm": 0.8151856217138742,
      "learning_rate": 9.989051741217895e-06,
      "loss": 1.1785,
      "num_tokens": 17765678.0,
      "step": 88
    },
    {
      "epoch": 0.21291866028708134,
      "grad_norm": 0.6934729593230965,
      "learning_rate": 9.988435321159706e-06,
      "loss": 0.8942,
      "num_tokens": 18017748.0,
      "step": 89
    },
    {
      "epoch": 0.215311004784689,
      "grad_norm": 0.8466118922234218,
      "learning_rate": 9.987802042228052e-06,
      "loss": 0.9131,
      "num_tokens": 18166766.0,
      "step": 90
    },
    {
      "epoch": 0.21770334928229665,
      "grad_norm": 0.7725523726706883,
      "learning_rate": 9.98715190680157e-06,
      "loss": 1.2444,
      "num_tokens": 18358833.0,
      "step": 91
    },
    {
      "epoch": 0.22009569377990432,
      "grad_norm": 0.7889212005298246,
      "learning_rate": 9.986484917322216e-06,
      "loss": 1.0743,
      "num_tokens": 18578155.0,
      "step": 92
    },
    {
      "epoch": 0.22248803827751196,
      "grad_norm": 0.7757930472253841,
      "learning_rate": 9.985801076295245e-06,
      "loss": 1.0136,
      "num_tokens": 18807884.0,
      "step": 93
    },
    {
      "epoch": 0.22488038277511962,
      "grad_norm": 0.7619186356991171,
      "learning_rate": 9.985100386289217e-06,
      "loss": 0.9538,
      "num_tokens": 18987783.0,
      "step": 94
    },
    {
      "epoch": 0.22727272727272727,
      "grad_norm": 0.7560512807570609,
      "learning_rate": 9.984382849935966e-06,
      "loss": 1.0655,
      "num_tokens": 19191375.0,
      "step": 95
    },
    {
      "epoch": 0.22966507177033493,
      "grad_norm": 0.8273044076025551,
      "learning_rate": 9.983648469930613e-06,
      "loss": 0.9171,
      "num_tokens": 19385570.0,
      "step": 96
    },
    {
      "epoch": 0.23205741626794257,
      "grad_norm": 0.8441992541867371,
      "learning_rate": 9.982897249031536e-06,
      "loss": 0.8167,
      "num_tokens": 19567794.0,
      "step": 97
    },
    {
      "epoch": 0.23444976076555024,
      "grad_norm": 0.7796712666919974,
      "learning_rate": 9.982129190060376e-06,
      "loss": 0.8066,
      "num_tokens": 19762696.0,
      "step": 98
    },
    {
      "epoch": 0.23684210526315788,
      "grad_norm": 0.6861892375802735,
      "learning_rate": 9.981344295902013e-06,
      "loss": 0.9534,
      "num_tokens": 20017766.0,
      "step": 99
    },
    {
      "epoch": 0.23923444976076555,
      "grad_norm": 0.7613445310920244,
      "learning_rate": 9.980542569504566e-06,
      "loss": 1.0992,
      "num_tokens": 20233456.0,
      "step": 100
    },
    {
      "epoch": 0.24162679425837322,
      "grad_norm": 0.7178060201919961,
      "learning_rate": 9.979724013879373e-06,
      "loss": 0.9877,
      "num_tokens": 20465617.0,
      "step": 101
    },
    {
      "epoch": 0.24401913875598086,
      "grad_norm": 0.7365437362689727,
      "learning_rate": 9.978888632100986e-06,
      "loss": 1.0293,
      "num_tokens": 20665957.0,
      "step": 102
    },
    {
      "epoch": 0.24641148325358853,
      "grad_norm": 0.676827848041371,
      "learning_rate": 9.978036427307158e-06,
      "loss": 1.1481,
      "num_tokens": 20888246.0,
      "step": 103
    },
    {
      "epoch": 0.24880382775119617,
      "grad_norm": 0.7580628326295507,
      "learning_rate": 9.977167402698828e-06,
      "loss": 0.987,
      "num_tokens": 21099480.0,
      "step": 104
    },
    {
      "epoch": 0.2511961722488038,
      "grad_norm": 0.7221725341556634,
      "learning_rate": 9.976281561540112e-06,
      "loss": 1.0031,
      "num_tokens": 21299349.0,
      "step": 105
    },
    {
      "epoch": 0.2535885167464115,
      "grad_norm": 0.7374394457768372,
      "learning_rate": 9.975378907158291e-06,
      "loss": 0.924,
      "num_tokens": 21516131.0,
      "step": 106
    },
    {
      "epoch": 0.25598086124401914,
      "grad_norm": 0.8372803563503946,
      "learning_rate": 9.9744594429438e-06,
      "loss": 1.1085,
      "num_tokens": 21657586.0,
      "step": 107
    },
    {
      "epoch": 0.2583732057416268,
      "grad_norm": 0.8474797225142054,
      "learning_rate": 9.973523172350205e-06,
      "loss": 0.7941,
      "num_tokens": 21808026.0,
      "step": 108
    },
    {
      "epoch": 0.2607655502392344,
      "grad_norm": 0.716388521294015,
      "learning_rate": 9.972570098894208e-06,
      "loss": 0.8706,
      "num_tokens": 22050205.0,
      "step": 109
    },
    {
      "epoch": 0.2631578947368421,
      "grad_norm": 0.7391113763396667,
      "learning_rate": 9.971600226155614e-06,
      "loss": 0.84,
      "num_tokens": 22268351.0,
      "step": 110
    },
    {
      "epoch": 0.26555023923444976,
      "grad_norm": 0.7735135057092313,
      "learning_rate": 9.970613557777336e-06,
      "loss": 0.8856,
      "num_tokens": 22464789.0,
      "step": 111
    },
    {
      "epoch": 0.2679425837320574,
      "grad_norm": 0.8223949250475577,
      "learning_rate": 9.96961009746537e-06,
      "loss": 0.719,
      "num_tokens": 22620151.0,
      "step": 112
    },
    {
      "epoch": 0.2703349282296651,
      "grad_norm": 0.7173732870433329,
      "learning_rate": 9.968589848988778e-06,
      "loss": 0.9263,
      "num_tokens": 22852967.0,
      "step": 113
    },
    {
      "epoch": 0.2727272727272727,
      "grad_norm": 0.658820584842679,
      "learning_rate": 9.967552816179684e-06,
      "loss": 0.8595,
      "num_tokens": 23082712.0,
      "step": 114
    },
    {
      "epoch": 0.2751196172248804,
      "grad_norm": 0.8291768755324592,
      "learning_rate": 9.966499002933257e-06,
      "loss": 0.7979,
      "num_tokens": 23236129.0,
      "step": 115
    },
    {
      "epoch": 0.27751196172248804,
      "grad_norm": 0.6891612480479086,
      "learning_rate": 9.965428413207696e-06,
      "loss": 1.116,
      "num_tokens": 23482677.0,
      "step": 116
    },
    {
      "epoch": 0.2799043062200957,
      "grad_norm": 0.7316914455895902,
      "learning_rate": 9.964341051024202e-06,
      "loss": 0.7551,
      "num_tokens": 23687754.0,
      "step": 117
    },
    {
      "epoch": 0.2822966507177033,
      "grad_norm": 0.6565088792008686,
      "learning_rate": 9.963236920466987e-06,
      "loss": 0.8169,
      "num_tokens": 23937964.0,
      "step": 118
    },
    {
      "epoch": 0.284688995215311,
      "grad_norm": 0.7421437162178479,
      "learning_rate": 9.962116025683241e-06,
      "loss": 0.8436,
      "num_tokens": 24127868.0,
      "step": 119
    },
    {
      "epoch": 0.28708133971291866,
      "grad_norm": 0.7006473705081104,
      "learning_rate": 9.960978370883123e-06,
      "loss": 0.9349,
      "num_tokens": 24354281.0,
      "step": 120
    },
    {
      "epoch": 0.2894736842105263,
      "grad_norm": 0.708154430058184,
      "learning_rate": 9.959823960339744e-06,
      "loss": 0.9994,
      "num_tokens": 24591341.0,
      "step": 121
    },
    {
      "epoch": 0.291866028708134,
      "grad_norm": 0.7523976805120025,
      "learning_rate": 9.958652798389148e-06,
      "loss": 0.9759,
      "num_tokens": 24808421.0,
      "step": 122
    },
    {
      "epoch": 0.2942583732057416,
      "grad_norm": 0.695503522009102,
      "learning_rate": 9.957464889430301e-06,
      "loss": 1.0527,
      "num_tokens": 25047341.0,
      "step": 123
    },
    {
      "epoch": 0.2966507177033493,
      "grad_norm": 0.7321236634606988,
      "learning_rate": 9.956260237925069e-06,
      "loss": 1.0856,
      "num_tokens": 25262803.0,
      "step": 124
    },
    {
      "epoch": 0.29904306220095694,
      "grad_norm": 0.6124241419692257,
      "learning_rate": 9.955038848398211e-06,
      "loss": 0.8271,
      "num_tokens": 25485650.0,
      "step": 125
    },
    {
      "epoch": 0.3014354066985646,
      "grad_norm": 0.8281763110793705,
      "learning_rate": 9.953800725437347e-06,
      "loss": 0.73,
      "num_tokens": 25664886.0,
      "step": 126
    },
    {
      "epoch": 0.3038277511961722,
      "grad_norm": 0.8118511520964532,
      "learning_rate": 9.952545873692953e-06,
      "loss": 0.9294,
      "num_tokens": 25847077.0,
      "step": 127
    },
    {
      "epoch": 0.3062200956937799,
      "grad_norm": 0.6694662698941429,
      "learning_rate": 9.951274297878341e-06,
      "loss": 0.9563,
      "num_tokens": 26088392.0,
      "step": 128
    },
    {
      "epoch": 0.30861244019138756,
      "grad_norm": 0.6923316125065374,
      "learning_rate": 9.949986002769633e-06,
      "loss": 0.9151,
      "num_tokens": 26307138.0,
      "step": 129
    },
    {
      "epoch": 0.31100478468899523,
      "grad_norm": 0.6309626818691494,
      "learning_rate": 9.94868099320576e-06,
      "loss": 0.7817,
      "num_tokens": 26592132.0,
      "step": 130
    },
    {
      "epoch": 0.3133971291866029,
      "grad_norm": 0.7429883494359573,
      "learning_rate": 9.947359274088427e-06,
      "loss": 1.0797,
      "num_tokens": 26763373.0,
      "step": 131
    },
    {
      "epoch": 0.3157894736842105,
      "grad_norm": 0.6943111995834695,
      "learning_rate": 9.946020850382098e-06,
      "loss": 0.951,
      "num_tokens": 26994962.0,
      "step": 132
    },
    {
      "epoch": 0.3181818181818182,
      "grad_norm": 0.6817616892645713,
      "learning_rate": 9.94466572711399e-06,
      "loss": 0.8817,
      "num_tokens": 27205603.0,
      "step": 133
    },
    {
      "epoch": 0.32057416267942584,
      "grad_norm": 0.6955201958805303,
      "learning_rate": 9.943293909374036e-06,
      "loss": 1.0069,
      "num_tokens": 27412792.0,
      "step": 134
    },
    {
      "epoch": 0.3229665071770335,
      "grad_norm": 0.694019942449106,
      "learning_rate": 9.94190540231488e-06,
      "loss": 0.7741,
      "num_tokens": 27625644.0,
      "step": 135
    },
    {
      "epoch": 0.3253588516746411,
      "grad_norm": 0.7987750635682109,
      "learning_rate": 9.940500211151853e-06,
      "loss": 0.9964,
      "num_tokens": 27809952.0,
      "step": 136
    },
    {
      "epoch": 0.3277511961722488,
      "grad_norm": 0.788259855652086,
      "learning_rate": 9.939078341162947e-06,
      "loss": 0.9231,
      "num_tokens": 27971074.0,
      "step": 137
    },
    {
      "epoch": 0.33014354066985646,
      "grad_norm": 0.6494876305365771,
      "learning_rate": 9.937639797688807e-06,
      "loss": 0.9048,
      "num_tokens": 28168165.0,
      "step": 138
    },
    {
      "epoch": 0.33253588516746413,
      "grad_norm": 0.7100895039199577,
      "learning_rate": 9.936184586132696e-06,
      "loss": 0.9842,
      "num_tokens": 28361420.0,
      "step": 139
    },
    {
      "epoch": 0.3349282296650718,
      "grad_norm": 0.6206399951244527,
      "learning_rate": 9.934712711960496e-06,
      "loss": 1.0349,
      "num_tokens": 28679853.0,
      "step": 140
    },
    {
      "epoch": 0.3373205741626794,
      "grad_norm": 0.7407205278350535,
      "learning_rate": 9.933224180700661e-06,
      "loss": 0.8706,
      "num_tokens": 28855036.0,
      "step": 141
    },
    {
      "epoch": 0.3397129186602871,
      "grad_norm": 0.6760210344304235,
      "learning_rate": 9.931718997944226e-06,
      "loss": 0.7435,
      "num_tokens": 29055967.0,
      "step": 142
    },
    {
      "epoch": 0.34210526315789475,
      "grad_norm": 0.730294279383326,
      "learning_rate": 9.930197169344753e-06,
      "loss": 0.8819,
      "num_tokens": 29265413.0,
      "step": 143
    },
    {
      "epoch": 0.3444976076555024,
      "grad_norm": 0.7982518259307247,
      "learning_rate": 9.928658700618343e-06,
      "loss": 1.0037,
      "num_tokens": 29459699.0,
      "step": 144
    },
    {
      "epoch": 0.34688995215311,
      "grad_norm": 0.6642614534122017,
      "learning_rate": 9.927103597543584e-06,
      "loss": 0.6234,
      "num_tokens": 29670531.0,
      "step": 145
    },
    {
      "epoch": 0.3492822966507177,
      "grad_norm": 0.7726991449535662,
      "learning_rate": 9.925531865961557e-06,
      "loss": 0.802,
      "num_tokens": 29841104.0,
      "step": 146
    },
    {
      "epoch": 0.35167464114832536,
      "grad_norm": 0.7190746964648677,
      "learning_rate": 9.923943511775788e-06,
      "loss": 0.8837,
      "num_tokens": 30083772.0,
      "step": 147
    },
    {
      "epoch": 0.35406698564593303,
      "grad_norm": 0.6773961708968047,
      "learning_rate": 9.922338540952251e-06,
      "loss": 0.9064,
      "num_tokens": 30318998.0,
      "step": 148
    },
    {
      "epoch": 0.35645933014354064,
      "grad_norm": 0.7165474452525965,
      "learning_rate": 9.920716959519326e-06,
      "loss": 0.8625,
      "num_tokens": 30511589.0,
      "step": 149
    },
    {
      "epoch": 0.3588516746411483,
      "grad_norm": 0.6962013623862424,
      "learning_rate": 9.919078773567784e-06,
      "loss": 1.0303,
      "num_tokens": 30704626.0,
      "step": 150
    },
    {
      "epoch": 0.361244019138756,
      "grad_norm": 0.701112474981306,
      "learning_rate": 9.917423989250762e-06,
      "loss": 1.1225,
      "num_tokens": 30941731.0,
      "step": 151
    },
    {
      "epoch": 0.36363636363636365,
      "grad_norm": 0.574854019614758,
      "learning_rate": 9.91575261278375e-06,
      "loss": 1.0329,
      "num_tokens": 31203750.0,
      "step": 152
    },
    {
      "epoch": 0.3660287081339713,
      "grad_norm": 0.7243129631207466,
      "learning_rate": 9.914064650444549e-06,
      "loss": 0.9106,
      "num_tokens": 31415131.0,
      "step": 153
    },
    {
      "epoch": 0.3684210526315789,
      "grad_norm": 0.7656940105326371,
      "learning_rate": 9.912360108573265e-06,
      "loss": 1.0737,
      "num_tokens": 31621021.0,
      "step": 154
    },
    {
      "epoch": 0.3708133971291866,
      "grad_norm": 0.7642923662971779,
      "learning_rate": 9.910638993572275e-06,
      "loss": 1.1397,
      "num_tokens": 31825774.0,
      "step": 155
    },
    {
      "epoch": 0.37320574162679426,
      "grad_norm": 0.8550286241707264,
      "learning_rate": 9.908901311906202e-06,
      "loss": 0.993,
      "num_tokens": 31998156.0,
      "step": 156
    },
    {
      "epoch": 0.37559808612440193,
      "grad_norm": 0.7186526140526422,
      "learning_rate": 9.907147070101906e-06,
      "loss": 1.0599,
      "num_tokens": 32219085.0,
      "step": 157
    },
    {
      "epoch": 0.37799043062200954,
      "grad_norm": 0.7167027687359431,
      "learning_rate": 9.905376274748433e-06,
      "loss": 0.8874,
      "num_tokens": 32451875.0,
      "step": 158
    },
    {
      "epoch": 0.3803827751196172,
      "grad_norm": 0.652018878770897,
      "learning_rate": 9.90358893249702e-06,
      "loss": 0.8651,
      "num_tokens": 32728142.0,
      "step": 159
    },
    {
      "epoch": 0.3827751196172249,
      "grad_norm": 0.7805318753632378,
      "learning_rate": 9.901785050061041e-06,
      "loss": 1.03,
      "num_tokens": 32954822.0,
      "step": 160
    },
    {
      "epoch": 0.38516746411483255,
      "grad_norm": 0.7527518503203819,
      "learning_rate": 9.899964634216006e-06,
      "loss": 1.0222,
      "num_tokens": 33154057.0,
      "step": 161
    },
    {
      "epoch": 0.3875598086124402,
      "grad_norm": 0.7433980271420518,
      "learning_rate": 9.898127691799521e-06,
      "loss": 0.8789,
      "num_tokens": 33317780.0,
      "step": 162
    },
    {
      "epoch": 0.38995215311004783,
      "grad_norm": 0.7680691692838955,
      "learning_rate": 9.896274229711276e-06,
      "loss": 1.0284,
      "num_tokens": 33552472.0,
      "step": 163
    },
    {
      "epoch": 0.3923444976076555,
      "grad_norm": 0.7722033859950767,
      "learning_rate": 9.894404254912994e-06,
      "loss": 1.037,
      "num_tokens": 33756438.0,
      "step": 164
    },
    {
      "epoch": 0.39473684210526316,
      "grad_norm": 0.737004315990203,
      "learning_rate": 9.892517774428434e-06,
      "loss": 0.9836,
      "num_tokens": 33979635.0,
      "step": 165
    },
    {
      "epoch": 0.39712918660287083,
      "grad_norm": 0.6691182389869809,
      "learning_rate": 9.89061479534335e-06,
      "loss": 1.0604,
      "num_tokens": 34273884.0,
      "step": 166
    },
    {
      "epoch": 0.39952153110047844,
      "grad_norm": 0.6790421569992825,
      "learning_rate": 9.888695324805453e-06,
      "loss": 0.9092,
      "num_tokens": 34464735.0,
      "step": 167
    },
    {
      "epoch": 0.4019138755980861,
      "grad_norm": 0.8288654161356374,
      "learning_rate": 9.88675937002442e-06,
      "loss": 1.1103,
      "num_tokens": 34686097.0,
      "step": 168
    },
    {
      "epoch": 0.4043062200956938,
      "grad_norm": 0.7965114072662832,
      "learning_rate": 9.88480693827182e-06,
      "loss": 0.8898,
      "num_tokens": 34854045.0,
      "step": 169
    },
    {
      "epoch": 0.40669856459330145,
      "grad_norm": 0.7434096681118705,
      "learning_rate": 9.882838036881128e-06,
      "loss": 0.8313,
      "num_tokens": 35069557.0,
      "step": 170
    },
    {
      "epoch": 0.4090909090909091,
      "grad_norm": 0.7239308188160857,
      "learning_rate": 9.880852673247671e-06,
      "loss": 0.9248,
      "num_tokens": 35263232.0,
      "step": 171
    },
    {
      "epoch": 0.41148325358851673,
      "grad_norm": 0.6944813811953502,
      "learning_rate": 9.878850854828614e-06,
      "loss": 1.0155,
      "num_tokens": 35508355.0,
      "step": 172
    },
    {
      "epoch": 0.4138755980861244,
      "grad_norm": 0.7508945276214297,
      "learning_rate": 9.876832589142925e-06,
      "loss": 0.8899,
      "num_tokens": 35718913.0,
      "step": 173
    },
    {
      "epoch": 0.41626794258373206,
      "grad_norm": 0.8067676235321118,
      "learning_rate": 9.87479788377135e-06,
      "loss": 0.8972,
      "num_tokens": 35917104.0,
      "step": 174
    },
    {
      "epoch": 0.41866028708133973,
      "grad_norm": 0.7477163711045044,
      "learning_rate": 9.87274674635638e-06,
      "loss": 0.8003,
      "num_tokens": 36084425.0,
      "step": 175
    },
    {
      "epoch": 0.42105263157894735,
      "grad_norm": 0.7262231697306871,
      "learning_rate": 9.870679184602233e-06,
      "loss": 0.9849,
      "num_tokens": 36311216.0,
      "step": 176
    },
    {
      "epoch": 0.423444976076555,
      "grad_norm": 0.8119435832767165,
      "learning_rate": 9.868595206274812e-06,
      "loss": 1.0225,
      "num_tokens": 36478396.0,
      "step": 177
    },
    {
      "epoch": 0.4258373205741627,
      "grad_norm": 0.767066338561648,
      "learning_rate": 9.866494819201682e-06,
      "loss": 0.9216,
      "num_tokens": 36672268.0,
      "step": 178
    },
    {
      "epoch": 0.42822966507177035,
      "grad_norm": 0.7011545360538196,
      "learning_rate": 9.864378031272045e-06,
      "loss": 0.9498,
      "num_tokens": 36904693.0,
      "step": 179
    },
    {
      "epoch": 0.430622009569378,
      "grad_norm": 0.7773922051063414,
      "learning_rate": 9.862244850436702e-06,
      "loss": 0.6955,
      "num_tokens": 37057970.0,
      "step": 180
    },
    {
      "epoch": 0.43301435406698563,
      "grad_norm": 0.7869339535084815,
      "learning_rate": 9.860095284708028e-06,
      "loss": 0.878,
      "num_tokens": 37226986.0,
      "step": 181
    },
    {
      "epoch": 0.4354066985645933,
      "grad_norm": 0.6978575252288622,
      "learning_rate": 9.857929342159938e-06,
      "loss": 0.8166,
      "num_tokens": 37454734.0,
      "step": 182
    },
    {
      "epoch": 0.43779904306220097,
      "grad_norm": 0.7859244195562889,
      "learning_rate": 9.855747030927866e-06,
      "loss": 0.6955,
      "num_tokens": 37598443.0,
      "step": 183
    },
    {
      "epoch": 0.44019138755980863,
      "grad_norm": 0.7367509293500684,
      "learning_rate": 9.853548359208722e-06,
      "loss": 0.827,
      "num_tokens": 37772689.0,
      "step": 184
    },
    {
      "epoch": 0.44258373205741625,
      "grad_norm": 0.7415296718711833,
      "learning_rate": 9.851333335260869e-06,
      "loss": 0.9512,
      "num_tokens": 38004549.0,
      "step": 185
    },
    {
      "epoch": 0.4449760765550239,
      "grad_norm": 0.8674192800405747,
      "learning_rate": 9.84910196740409e-06,
      "loss": 0.8545,
      "num_tokens": 38159721.0,
      "step": 186
    },
    {
      "epoch": 0.4473684210526316,
      "grad_norm": 0.8172487094325894,
      "learning_rate": 9.846854264019556e-06,
      "loss": 0.9048,
      "num_tokens": 38361874.0,
      "step": 187
    },
    {
      "epoch": 0.44976076555023925,
      "grad_norm": 0.749271142770708,
      "learning_rate": 9.844590233549797e-06,
      "loss": 0.7697,
      "num_tokens": 38532356.0,
      "step": 188
    },
    {
      "epoch": 0.45215311004784686,
      "grad_norm": 0.7006067463192553,
      "learning_rate": 9.842309884498672e-06,
      "loss": 0.8987,
      "num_tokens": 38771369.0,
      "step": 189
    },
    {
      "epoch": 0.45454545454545453,
      "grad_norm": 0.5978765943982213,
      "learning_rate": 9.840013225431322e-06,
      "loss": 1.0488,
      "num_tokens": 39038278.0,
      "step": 190
    },
    {
      "epoch": 0.4569377990430622,
      "grad_norm": 0.7887229239731655,
      "learning_rate": 9.837700264974163e-06,
      "loss": 0.8302,
      "num_tokens": 39240942.0,
      "step": 191
    },
    {
      "epoch": 0.45933014354066987,
      "grad_norm": 0.7502651179572393,
      "learning_rate": 9.835371011814832e-06,
      "loss": 1.0493,
      "num_tokens": 39391888.0,
      "step": 192
    },
    {
      "epoch": 0.46172248803827753,
      "grad_norm": 0.648210560375548,
      "learning_rate": 9.833025474702169e-06,
      "loss": 0.8346,
      "num_tokens": 39645186.0,
      "step": 193
    },
    {
      "epoch": 0.46411483253588515,
      "grad_norm": 0.6635996305419883,
      "learning_rate": 9.830663662446166e-06,
      "loss": 0.9651,
      "num_tokens": 39872583.0,
      "step": 194
    },
    {
      "epoch": 0.4665071770334928,
      "grad_norm": 0.7512488474508062,
      "learning_rate": 9.828285583917962e-06,
      "loss": 0.724,
      "num_tokens": 40030081.0,
      "step": 195
    },
    {
      "epoch": 0.4688995215311005,
      "grad_norm": 0.7620024583744265,
      "learning_rate": 9.825891248049776e-06,
      "loss": 0.9252,
      "num_tokens": 40225498.0,
      "step": 196
    },
    {
      "epoch": 0.47129186602870815,
      "grad_norm": 0.7381021014120716,
      "learning_rate": 9.823480663834905e-06,
      "loss": 0.9246,
      "num_tokens": 40419013.0,
      "step": 197
    },
    {
      "epoch": 0.47368421052631576,
      "grad_norm": 0.7310934363198373,
      "learning_rate": 9.821053840327668e-06,
      "loss": 0.8483,
      "num_tokens": 40638483.0,
      "step": 198
    },
    {
      "epoch": 0.47607655502392343,
      "grad_norm": 0.6369502362340661,
      "learning_rate": 9.818610786643383e-06,
      "loss": 1.1045,
      "num_tokens": 40851402.0,
      "step": 199
    },
    {
      "epoch": 0.4784688995215311,
      "grad_norm": 0.6662173138687252,
      "learning_rate": 9.816151511958326e-06,
      "loss": 0.8642,
      "num_tokens": 41076074.0,
      "step": 200
    },
    {
      "epoch": 0.48086124401913877,
      "grad_norm": 0.7069311757824951,
      "learning_rate": 9.813676025509705e-06,
      "loss": 0.8348,
      "num_tokens": 41281733.0,
      "step": 201
    },
    {
      "epoch": 0.48325358851674644,
      "grad_norm": 0.7715896724886065,
      "learning_rate": 9.811184336595618e-06,
      "loss": 0.7638,
      "num_tokens": 41469321.0,
      "step": 202
    },
    {
      "epoch": 0.48564593301435405,
      "grad_norm": 0.7300586617529341,
      "learning_rate": 9.808676454575021e-06,
      "loss": 0.8712,
      "num_tokens": 41655589.0,
      "step": 203
    },
    {
      "epoch": 0.4880382775119617,
      "grad_norm": 0.7433895285131304,
      "learning_rate": 9.806152388867691e-06,
      "loss": 0.9764,
      "num_tokens": 41884531.0,
      "step": 204
    },
    {
      "epoch": 0.4904306220095694,
      "grad_norm": 0.729313387708453,
      "learning_rate": 9.803612148954196e-06,
      "loss": 0.89,
      "num_tokens": 42067360.0,
      "step": 205
    },
    {
      "epoch": 0.49282296650717705,
      "grad_norm": 0.7368107493572161,
      "learning_rate": 9.80105574437585e-06,
      "loss": 0.6557,
      "num_tokens": 42220024.0,
      "step": 206
    },
    {
      "epoch": 0.49521531100478466,
      "grad_norm": 0.7182944346849723,
      "learning_rate": 9.798483184734685e-06,
      "loss": 0.999,
      "num_tokens": 42412973.0,
      "step": 207
    },
    {
      "epoch": 0.49760765550239233,
      "grad_norm": 0.6838259373624938,
      "learning_rate": 9.795894479693416e-06,
      "loss": 0.7809,
      "num_tokens": 42635600.0,
      "step": 208
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6596178695787379,
      "learning_rate": 9.793289638975396e-06,
      "loss": 0.8985,
      "num_tokens": 42829803.0,
      "step": 209
    },
    {
      "epoch": 0.5023923444976076,
      "grad_norm": 0.7819562505016303,
      "learning_rate": 9.790668672364588e-06,
      "loss": 1.0058,
      "num_tokens": 42987359.0,
      "step": 210
    },
    {
      "epoch": 0.5047846889952153,
      "grad_norm": 0.6376009810615048,
      "learning_rate": 9.788031589705522e-06,
      "loss": 0.825,
      "num_tokens": 43222234.0,
      "step": 211
    },
    {
      "epoch": 0.507177033492823,
      "grad_norm": 0.7170548940637805,
      "learning_rate": 9.785378400903265e-06,
      "loss": 0.8579,
      "num_tokens": 43415580.0,
      "step": 212
    },
    {
      "epoch": 0.5095693779904307,
      "grad_norm": 0.6943966856575938,
      "learning_rate": 9.782709115923377e-06,
      "loss": 0.8678,
      "num_tokens": 43636143.0,
      "step": 213
    },
    {
      "epoch": 0.5119617224880383,
      "grad_norm": 0.7777442170195412,
      "learning_rate": 9.780023744791875e-06,
      "loss": 0.9754,
      "num_tokens": 43788003.0,
      "step": 214
    },
    {
      "epoch": 0.5143540669856459,
      "grad_norm": 0.7565902699333324,
      "learning_rate": 9.777322297595203e-06,
      "loss": 0.8631,
      "num_tokens": 43936330.0,
      "step": 215
    },
    {
      "epoch": 0.5167464114832536,
      "grad_norm": 0.6723572251989621,
      "learning_rate": 9.77460478448018e-06,
      "loss": 0.8151,
      "num_tokens": 44164196.0,
      "step": 216
    },
    {
      "epoch": 0.5191387559808612,
      "grad_norm": 0.7222983115738448,
      "learning_rate": 9.771871215653973e-06,
      "loss": 0.9845,
      "num_tokens": 44362899.0,
      "step": 217
    },
    {
      "epoch": 0.5215311004784688,
      "grad_norm": 0.7125065875645107,
      "learning_rate": 9.769121601384057e-06,
      "loss": 1.0482,
      "num_tokens": 44601312.0,
      "step": 218
    },
    {
      "epoch": 0.5239234449760766,
      "grad_norm": 0.707677277074084,
      "learning_rate": 9.766355951998172e-06,
      "loss": 1.0536,
      "num_tokens": 44795812.0,
      "step": 219
    },
    {
      "epoch": 0.5263157894736842,
      "grad_norm": 0.6953123625587964,
      "learning_rate": 9.76357427788429e-06,
      "loss": 1.0325,
      "num_tokens": 45011494.0,
      "step": 220
    },
    {
      "epoch": 0.5287081339712919,
      "grad_norm": 0.6209070758822427,
      "learning_rate": 9.760776589490571e-06,
      "loss": 0.8775,
      "num_tokens": 45271878.0,
      "step": 221
    },
    {
      "epoch": 0.5311004784688995,
      "grad_norm": 0.8181333695786353,
      "learning_rate": 9.757962897325327e-06,
      "loss": 0.7584,
      "num_tokens": 45395391.0,
      "step": 222
    },
    {
      "epoch": 0.5334928229665071,
      "grad_norm": 0.7353051768668135,
      "learning_rate": 9.755133211956976e-06,
      "loss": 0.997,
      "num_tokens": 45590083.0,
      "step": 223
    },
    {
      "epoch": 0.5358851674641149,
      "grad_norm": 0.6917954967989495,
      "learning_rate": 9.752287544014016e-06,
      "loss": 0.8478,
      "num_tokens": 45809050.0,
      "step": 224
    },
    {
      "epoch": 0.5382775119617225,
      "grad_norm": 0.6310463226895536,
      "learning_rate": 9.749425904184973e-06,
      "loss": 0.8349,
      "num_tokens": 46026582.0,
      "step": 225
    },
    {
      "epoch": 0.5406698564593302,
      "grad_norm": 0.7496878034969545,
      "learning_rate": 9.746548303218361e-06,
      "loss": 0.7081,
      "num_tokens": 46217058.0,
      "step": 226
    },
    {
      "epoch": 0.5430622009569378,
      "grad_norm": 0.7613261565472779,
      "learning_rate": 9.74365475192265e-06,
      "loss": 0.9549,
      "num_tokens": 46405604.0,
      "step": 227
    },
    {
      "epoch": 0.5454545454545454,
      "grad_norm": 0.7118103212558673,
      "learning_rate": 9.74074526116622e-06,
      "loss": 0.7397,
      "num_tokens": 46609383.0,
      "step": 228
    },
    {
      "epoch": 0.5478468899521531,
      "grad_norm": 0.7886609868852241,
      "learning_rate": 9.737819841877319e-06,
      "loss": 0.6614,
      "num_tokens": 46774027.0,
      "step": 229
    },
    {
      "epoch": 0.5502392344497608,
      "grad_norm": 0.7512297618411289,
      "learning_rate": 9.734878505044022e-06,
      "loss": 0.8708,
      "num_tokens": 46934136.0,
      "step": 230
    },
    {
      "epoch": 0.5526315789473685,
      "grad_norm": 0.6467942759269715,
      "learning_rate": 9.731921261714195e-06,
      "loss": 0.9094,
      "num_tokens": 47166689.0,
      "step": 231
    },
    {
      "epoch": 0.5550239234449761,
      "grad_norm": 0.6186151914065757,
      "learning_rate": 9.728948122995448e-06,
      "loss": 0.74,
      "num_tokens": 47407182.0,
      "step": 232
    },
    {
      "epoch": 0.5574162679425837,
      "grad_norm": 0.7546119364082358,
      "learning_rate": 9.725959100055095e-06,
      "loss": 0.8622,
      "num_tokens": 47621230.0,
      "step": 233
    },
    {
      "epoch": 0.5598086124401914,
      "grad_norm": 0.7016789780653074,
      "learning_rate": 9.722954204120116e-06,
      "loss": 1.1317,
      "num_tokens": 47839958.0,
      "step": 234
    },
    {
      "epoch": 0.562200956937799,
      "grad_norm": 0.6830572301668759,
      "learning_rate": 9.719933446477104e-06,
      "loss": 1.0093,
      "num_tokens": 48044769.0,
      "step": 235
    },
    {
      "epoch": 0.5645933014354066,
      "grad_norm": 0.6848757588607501,
      "learning_rate": 9.716896838472237e-06,
      "loss": 1.0054,
      "num_tokens": 48254718.0,
      "step": 236
    },
    {
      "epoch": 0.5669856459330144,
      "grad_norm": 0.7931488360108984,
      "learning_rate": 9.71384439151122e-06,
      "loss": 0.5908,
      "num_tokens": 48403707.0,
      "step": 237
    },
    {
      "epoch": 0.569377990430622,
      "grad_norm": 0.7468278674122727,
      "learning_rate": 9.710776117059262e-06,
      "loss": 0.8507,
      "num_tokens": 48580654.0,
      "step": 238
    },
    {
      "epoch": 0.5717703349282297,
      "grad_norm": 0.7928730950381853,
      "learning_rate": 9.707692026641007e-06,
      "loss": 1.0811,
      "num_tokens": 48757864.0,
      "step": 239
    },
    {
      "epoch": 0.5741626794258373,
      "grad_norm": 0.8167247736000246,
      "learning_rate": 9.704592131840517e-06,
      "loss": 0.8996,
      "num_tokens": 48948795.0,
      "step": 240
    },
    {
      "epoch": 0.5765550239234449,
      "grad_norm": 0.7576343998163653,
      "learning_rate": 9.701476444301208e-06,
      "loss": 0.8986,
      "num_tokens": 49137680.0,
      "step": 241
    },
    {
      "epoch": 0.5789473684210527,
      "grad_norm": 0.7136634589206409,
      "learning_rate": 9.69834497572582e-06,
      "loss": 0.969,
      "num_tokens": 49330369.0,
      "step": 242
    },
    {
      "epoch": 0.5813397129186603,
      "grad_norm": 0.6736332498636337,
      "learning_rate": 9.695197737876364e-06,
      "loss": 0.9318,
      "num_tokens": 49545446.0,
      "step": 243
    },
    {
      "epoch": 0.583732057416268,
      "grad_norm": 0.6714841653229451,
      "learning_rate": 9.692034742574082e-06,
      "loss": 0.8462,
      "num_tokens": 49779498.0,
      "step": 244
    },
    {
      "epoch": 0.5861244019138756,
      "grad_norm": 0.8175525625058889,
      "learning_rate": 9.688856001699407e-06,
      "loss": 0.9611,
      "num_tokens": 49952841.0,
      "step": 245
    },
    {
      "epoch": 0.5885167464114832,
      "grad_norm": 0.7409257480013611,
      "learning_rate": 9.685661527191905e-06,
      "loss": 0.9577,
      "num_tokens": 50151322.0,
      "step": 246
    },
    {
      "epoch": 0.5909090909090909,
      "grad_norm": 0.7504645386701574,
      "learning_rate": 9.682451331050246e-06,
      "loss": 1.0545,
      "num_tokens": 50349864.0,
      "step": 247
    },
    {
      "epoch": 0.5933014354066986,
      "grad_norm": 0.693369456244369,
      "learning_rate": 9.679225425332147e-06,
      "loss": 0.9629,
      "num_tokens": 50584303.0,
      "step": 248
    },
    {
      "epoch": 0.5956937799043063,
      "grad_norm": 0.646410500745444,
      "learning_rate": 9.675983822154336e-06,
      "loss": 0.8581,
      "num_tokens": 50818111.0,
      "step": 249
    },
    {
      "epoch": 0.5980861244019139,
      "grad_norm": 0.7492761839244249,
      "learning_rate": 9.672726533692496e-06,
      "loss": 0.9365,
      "num_tokens": 51034934.0,
      "step": 250
    },
    {
      "epoch": 0.6004784688995215,
      "grad_norm": 0.7875923284327673,
      "learning_rate": 9.66945357218123e-06,
      "loss": 0.8799,
      "num_tokens": 51196696.0,
      "step": 251
    },
    {
      "epoch": 0.6028708133971292,
      "grad_norm": 0.6890802350922746,
      "learning_rate": 9.666164949914003e-06,
      "loss": 0.8263,
      "num_tokens": 51394657.0,
      "step": 252
    },
    {
      "epoch": 0.6052631578947368,
      "grad_norm": 0.6609546018572526,
      "learning_rate": 9.662860679243112e-06,
      "loss": 0.7737,
      "num_tokens": 51629548.0,
      "step": 253
    },
    {
      "epoch": 0.6076555023923444,
      "grad_norm": 0.6388346290233738,
      "learning_rate": 9.659540772579626e-06,
      "loss": 0.7746,
      "num_tokens": 51859831.0,
      "step": 254
    },
    {
      "epoch": 0.6100478468899522,
      "grad_norm": 0.7630946831333012,
      "learning_rate": 9.656205242393341e-06,
      "loss": 0.8517,
      "num_tokens": 52014662.0,
      "step": 255
    },
    {
      "epoch": 0.6124401913875598,
      "grad_norm": 0.7099062445428919,
      "learning_rate": 9.652854101212744e-06,
      "loss": 1.1071,
      "num_tokens": 52240815.0,
      "step": 256
    },
    {
      "epoch": 0.6148325358851675,
      "grad_norm": 0.6754326767149219,
      "learning_rate": 9.64948736162495e-06,
      "loss": 1.019,
      "num_tokens": 52474992.0,
      "step": 257
    },
    {
      "epoch": 0.6172248803827751,
      "grad_norm": 0.739542831715403,
      "learning_rate": 9.646105036275663e-06,
      "loss": 0.7532,
      "num_tokens": 52651126.0,
      "step": 258
    },
    {
      "epoch": 0.6196172248803827,
      "grad_norm": 0.728643865556642,
      "learning_rate": 9.642707137869136e-06,
      "loss": 0.8722,
      "num_tokens": 52820774.0,
      "step": 259
    },
    {
      "epoch": 0.6220095693779905,
      "grad_norm": 0.7086257685596635,
      "learning_rate": 9.639293679168105e-06,
      "loss": 0.6187,
      "num_tokens": 53018322.0,
      "step": 260
    },
    {
      "epoch": 0.6244019138755981,
      "grad_norm": 0.7399683012373918,
      "learning_rate": 9.635864672993764e-06,
      "loss": 0.872,
      "num_tokens": 53219194.0,
      "step": 261
    },
    {
      "epoch": 0.6267942583732058,
      "grad_norm": 0.7661507351405599,
      "learning_rate": 9.632420132225691e-06,
      "loss": 0.6345,
      "num_tokens": 53387444.0,
      "step": 262
    },
    {
      "epoch": 0.6291866028708134,
      "grad_norm": 0.7069621149226814,
      "learning_rate": 9.628960069801825e-06,
      "loss": 0.8452,
      "num_tokens": 53626979.0,
      "step": 263
    },
    {
      "epoch": 0.631578947368421,
      "grad_norm": 0.797463821355159,
      "learning_rate": 9.625484498718392e-06,
      "loss": 0.7205,
      "num_tokens": 53774781.0,
      "step": 264
    },
    {
      "epoch": 0.6339712918660287,
      "grad_norm": 0.7005373002761818,
      "learning_rate": 9.621993432029884e-06,
      "loss": 1.0427,
      "num_tokens": 53957662.0,
      "step": 265
    },
    {
      "epoch": 0.6363636363636364,
      "grad_norm": 0.7101631369249962,
      "learning_rate": 9.618486882848988e-06,
      "loss": 0.7358,
      "num_tokens": 54144418.0,
      "step": 266
    },
    {
      "epoch": 0.638755980861244,
      "grad_norm": 0.7439018504530075,
      "learning_rate": 9.614964864346541e-06,
      "loss": 0.7923,
      "num_tokens": 54321948.0,
      "step": 267
    },
    {
      "epoch": 0.6411483253588517,
      "grad_norm": 0.704791131563084,
      "learning_rate": 9.611427389751494e-06,
      "loss": 0.6775,
      "num_tokens": 54511012.0,
      "step": 268
    },
    {
      "epoch": 0.6435406698564593,
      "grad_norm": 0.718086612467513,
      "learning_rate": 9.607874472350842e-06,
      "loss": 0.6723,
      "num_tokens": 54715882.0,
      "step": 269
    },
    {
      "epoch": 0.645933014354067,
      "grad_norm": 0.7143312089646857,
      "learning_rate": 9.60430612548959e-06,
      "loss": 1.0111,
      "num_tokens": 54914319.0,
      "step": 270
    },
    {
      "epoch": 0.6483253588516746,
      "grad_norm": 0.6755764277181526,
      "learning_rate": 9.600722362570694e-06,
      "loss": 1.0113,
      "num_tokens": 55122172.0,
      "step": 271
    },
    {
      "epoch": 0.6507177033492823,
      "grad_norm": 0.7761302927431148,
      "learning_rate": 9.59712319705502e-06,
      "loss": 0.8245,
      "num_tokens": 55323698.0,
      "step": 272
    },
    {
      "epoch": 0.65311004784689,
      "grad_norm": 0.7340335032058501,
      "learning_rate": 9.593508642461276e-06,
      "loss": 0.9043,
      "num_tokens": 55531439.0,
      "step": 273
    },
    {
      "epoch": 0.6555023923444976,
      "grad_norm": 0.7588348203486196,
      "learning_rate": 9.589878712365982e-06,
      "loss": 0.624,
      "num_tokens": 55692316.0,
      "step": 274
    },
    {
      "epoch": 0.6578947368421053,
      "grad_norm": 0.7550464009356935,
      "learning_rate": 9.586233420403407e-06,
      "loss": 1.1432,
      "num_tokens": 55928859.0,
      "step": 275
    },
    {
      "epoch": 0.6602870813397129,
      "grad_norm": 0.7135616341229885,
      "learning_rate": 9.582572780265521e-06,
      "loss": 0.7436,
      "num_tokens": 56121111.0,
      "step": 276
    },
    {
      "epoch": 0.6626794258373205,
      "grad_norm": 0.6371530652389988,
      "learning_rate": 9.578896805701939e-06,
      "loss": 0.9986,
      "num_tokens": 56359401.0,
      "step": 277
    },
    {
      "epoch": 0.6650717703349283,
      "grad_norm": 0.683186850513734,
      "learning_rate": 9.575205510519878e-06,
      "loss": 0.8857,
      "num_tokens": 56581344.0,
      "step": 278
    },
    {
      "epoch": 0.6674641148325359,
      "grad_norm": 0.6810286701061241,
      "learning_rate": 9.571498908584097e-06,
      "loss": 1.0477,
      "num_tokens": 56788478.0,
      "step": 279
    },
    {
      "epoch": 0.6698564593301436,
      "grad_norm": 0.6142445432056164,
      "learning_rate": 9.567777013816846e-06,
      "loss": 1.1127,
      "num_tokens": 57018493.0,
      "step": 280
    },
    {
      "epoch": 0.6722488038277512,
      "grad_norm": 0.7105090692349689,
      "learning_rate": 9.564039840197824e-06,
      "loss": 0.9211,
      "num_tokens": 57201946.0,
      "step": 281
    },
    {
      "epoch": 0.6746411483253588,
      "grad_norm": 0.7441838628071439,
      "learning_rate": 9.560287401764109e-06,
      "loss": 0.8588,
      "num_tokens": 57378278.0,
      "step": 282
    },
    {
      "epoch": 0.6770334928229665,
      "grad_norm": 0.616622351957538,
      "learning_rate": 9.556519712610122e-06,
      "loss": 0.9045,
      "num_tokens": 57597843.0,
      "step": 283
    },
    {
      "epoch": 0.6794258373205742,
      "grad_norm": 0.6650978269716845,
      "learning_rate": 9.552736786887566e-06,
      "loss": 0.8173,
      "num_tokens": 57812696.0,
      "step": 284
    },
    {
      "epoch": 0.6818181818181818,
      "grad_norm": 0.6669101745672817,
      "learning_rate": 9.548938638805369e-06,
      "loss": 0.8258,
      "num_tokens": 58031586.0,
      "step": 285
    },
    {
      "epoch": 0.6842105263157895,
      "grad_norm": 0.712052565123226,
      "learning_rate": 9.545125282629639e-06,
      "loss": 0.6606,
      "num_tokens": 58219867.0,
      "step": 286
    },
    {
      "epoch": 0.6866028708133971,
      "grad_norm": 0.6903410307218643,
      "learning_rate": 9.541296732683601e-06,
      "loss": 0.9946,
      "num_tokens": 58455804.0,
      "step": 287
    },
    {
      "epoch": 0.6889952153110048,
      "grad_norm": 0.7028207533903529,
      "learning_rate": 9.537453003347562e-06,
      "loss": 0.7778,
      "num_tokens": 58647209.0,
      "step": 288
    },
    {
      "epoch": 0.6913875598086124,
      "grad_norm": 0.5935012950917158,
      "learning_rate": 9.533594109058827e-06,
      "loss": 0.6958,
      "num_tokens": 58876476.0,
      "step": 289
    },
    {
      "epoch": 0.69377990430622,
      "grad_norm": 0.7147877252504475,
      "learning_rate": 9.529720064311674e-06,
      "loss": 0.8571,
      "num_tokens": 59054211.0,
      "step": 290
    },
    {
      "epoch": 0.6961722488038278,
      "grad_norm": 0.6165616093448105,
      "learning_rate": 9.525830883657283e-06,
      "loss": 0.9089,
      "num_tokens": 59274415.0,
      "step": 291
    },
    {
      "epoch": 0.6985645933014354,
      "grad_norm": 0.7262837935686104,
      "learning_rate": 9.521926581703682e-06,
      "loss": 0.8738,
      "num_tokens": 59494463.0,
      "step": 292
    },
    {
      "epoch": 0.7009569377990431,
      "grad_norm": 0.7227392632343834,
      "learning_rate": 9.518007173115705e-06,
      "loss": 0.9422,
      "num_tokens": 59683825.0,
      "step": 293
    },
    {
      "epoch": 0.7033492822966507,
      "grad_norm": 0.657716296887696,
      "learning_rate": 9.514072672614915e-06,
      "loss": 0.9122,
      "num_tokens": 59900990.0,
      "step": 294
    },
    {
      "epoch": 0.7057416267942583,
      "grad_norm": 0.6950627976694991,
      "learning_rate": 9.510123094979575e-06,
      "loss": 0.7463,
      "num_tokens": 60100020.0,
      "step": 295
    },
    {
      "epoch": 0.7081339712918661,
      "grad_norm": 0.7417107599439859,
      "learning_rate": 9.506158455044567e-06,
      "loss": 0.6647,
      "num_tokens": 60267424.0,
      "step": 296
    },
    {
      "epoch": 0.7105263157894737,
      "grad_norm": 0.7175495863527874,
      "learning_rate": 9.502178767701353e-06,
      "loss": 0.9446,
      "num_tokens": 60452992.0,
      "step": 297
    },
    {
      "epoch": 0.7129186602870813,
      "grad_norm": 0.7618820197933083,
      "learning_rate": 9.498184047897915e-06,
      "loss": 0.919,
      "num_tokens": 60618749.0,
      "step": 298
    },
    {
      "epoch": 0.715311004784689,
      "grad_norm": 0.6170834471553733,
      "learning_rate": 9.494174310638697e-06,
      "loss": 0.8629,
      "num_tokens": 60883079.0,
      "step": 299
    },
    {
      "epoch": 0.7177033492822966,
      "grad_norm": 0.7814822668163934,
      "learning_rate": 9.490149570984549e-06,
      "loss": 0.8538,
      "num_tokens": 61050703.0,
      "step": 300
    },
    {
      "epoch": 0.7200956937799043,
      "grad_norm": 0.7973106331385097,
      "learning_rate": 9.48610984405267e-06,
      "loss": 0.9464,
      "num_tokens": 61281941.0,
      "step": 301
    },
    {
      "epoch": 0.722488038277512,
      "grad_norm": 0.7386439891400786,
      "learning_rate": 9.482055145016557e-06,
      "loss": 1.0107,
      "num_tokens": 61500702.0,
      "step": 302
    },
    {
      "epoch": 0.7248803827751196,
      "grad_norm": 0.7576189557606913,
      "learning_rate": 9.47798548910594e-06,
      "loss": 0.7948,
      "num_tokens": 61677831.0,
      "step": 303
    },
    {
      "epoch": 0.7272727272727273,
      "grad_norm": 0.582639751482507,
      "learning_rate": 9.473900891606724e-06,
      "loss": 1.0505,
      "num_tokens": 61943094.0,
      "step": 304
    },
    {
      "epoch": 0.7296650717703349,
      "grad_norm": 0.6975691338736785,
      "learning_rate": 9.469801367860944e-06,
      "loss": 0.8864,
      "num_tokens": 62162858.0,
      "step": 305
    },
    {
      "epoch": 0.7320574162679426,
      "grad_norm": 0.6001353096390548,
      "learning_rate": 9.46568693326669e-06,
      "loss": 0.8886,
      "num_tokens": 62440988.0,
      "step": 306
    },
    {
      "epoch": 0.7344497607655502,
      "grad_norm": 0.6694234470054652,
      "learning_rate": 9.461557603278066e-06,
      "loss": 0.8452,
      "num_tokens": 62675134.0,
      "step": 307
    },
    {
      "epoch": 0.7368421052631579,
      "grad_norm": 0.6920337040450307,
      "learning_rate": 9.457413393405122e-06,
      "loss": 0.8468,
      "num_tokens": 62839459.0,
      "step": 308
    },
    {
      "epoch": 0.7392344497607656,
      "grad_norm": 0.6664718554483587,
      "learning_rate": 9.453254319213794e-06,
      "loss": 0.9225,
      "num_tokens": 63032332.0,
      "step": 309
    },
    {
      "epoch": 0.7416267942583732,
      "grad_norm": 0.7044180002802695,
      "learning_rate": 9.449080396325855e-06,
      "loss": 0.6374,
      "num_tokens": 63244272.0,
      "step": 310
    },
    {
      "epoch": 0.7440191387559809,
      "grad_norm": 0.7003159969901962,
      "learning_rate": 9.444891640418847e-06,
      "loss": 0.9323,
      "num_tokens": 63466930.0,
      "step": 311
    },
    {
      "epoch": 0.7464114832535885,
      "grad_norm": 0.6724163707523774,
      "learning_rate": 9.440688067226025e-06,
      "loss": 0.957,
      "num_tokens": 63666983.0,
      "step": 312
    },
    {
      "epoch": 0.7488038277511961,
      "grad_norm": 0.6990957415189266,
      "learning_rate": 9.436469692536303e-06,
      "loss": 1.1225,
      "num_tokens": 63921481.0,
      "step": 313
    },
    {
      "epoch": 0.7511961722488039,
      "grad_norm": 0.7267577309004162,
      "learning_rate": 9.432236532194188e-06,
      "loss": 0.9527,
      "num_tokens": 64108035.0,
      "step": 314
    },
    {
      "epoch": 0.7535885167464115,
      "grad_norm": 0.6565759412319756,
      "learning_rate": 9.427988602099722e-06,
      "loss": 0.8312,
      "num_tokens": 64347582.0,
      "step": 315
    },
    {
      "epoch": 0.7559808612440191,
      "grad_norm": 0.6989421029736691,
      "learning_rate": 9.423725918208422e-06,
      "loss": 1.0384,
      "num_tokens": 64565695.0,
      "step": 316
    },
    {
      "epoch": 0.7583732057416268,
      "grad_norm": 0.8075590633967065,
      "learning_rate": 9.41944849653123e-06,
      "loss": 0.7874,
      "num_tokens": 64736228.0,
      "step": 317
    },
    {
      "epoch": 0.7607655502392344,
      "grad_norm": 0.7420969172241604,
      "learning_rate": 9.41515635313443e-06,
      "loss": 0.8776,
      "num_tokens": 64919765.0,
      "step": 318
    },
    {
      "epoch": 0.7631578947368421,
      "grad_norm": 0.8074410953534803,
      "learning_rate": 9.410849504139615e-06,
      "loss": 0.8007,
      "num_tokens": 65093331.0,
      "step": 319
    },
    {
      "epoch": 0.7655502392344498,
      "grad_norm": 0.7359810070319559,
      "learning_rate": 9.406527965723606e-06,
      "loss": 0.775,
      "num_tokens": 65281368.0,
      "step": 320
    },
    {
      "epoch": 0.7679425837320574,
      "grad_norm": 0.671250608829433,
      "learning_rate": 9.402191754118398e-06,
      "loss": 0.8672,
      "num_tokens": 65474208.0,
      "step": 321
    },
    {
      "epoch": 0.7703349282296651,
      "grad_norm": 0.70087002173681,
      "learning_rate": 9.397840885611105e-06,
      "loss": 0.8645,
      "num_tokens": 65700409.0,
      "step": 322
    },
    {
      "epoch": 0.7727272727272727,
      "grad_norm": 0.6634659621815793,
      "learning_rate": 9.393475376543888e-06,
      "loss": 0.9032,
      "num_tokens": 65968264.0,
      "step": 323
    },
    {
      "epoch": 0.7751196172248804,
      "grad_norm": 0.6855924238506387,
      "learning_rate": 9.389095243313901e-06,
      "loss": 0.7714,
      "num_tokens": 66153787.0,
      "step": 324
    },
    {
      "epoch": 0.777511961722488,
      "grad_norm": 0.6892245239453134,
      "learning_rate": 9.384700502373227e-06,
      "loss": 0.6299,
      "num_tokens": 66343926.0,
      "step": 325
    },
    {
      "epoch": 0.7799043062200957,
      "grad_norm": 0.7020557300227046,
      "learning_rate": 9.380291170228819e-06,
      "loss": 0.7912,
      "num_tokens": 66543908.0,
      "step": 326
    },
    {
      "epoch": 0.7822966507177034,
      "grad_norm": 0.7371763474019104,
      "learning_rate": 9.375867263442433e-06,
      "loss": 0.9272,
      "num_tokens": 66771074.0,
      "step": 327
    },
    {
      "epoch": 0.784688995215311,
      "grad_norm": 0.7796597325368293,
      "learning_rate": 9.371428798630567e-06,
      "loss": 0.8503,
      "num_tokens": 66937799.0,
      "step": 328
    },
    {
      "epoch": 0.7870813397129187,
      "grad_norm": 0.750139825760097,
      "learning_rate": 9.366975792464402e-06,
      "loss": 0.8718,
      "num_tokens": 67141282.0,
      "step": 329
    },
    {
      "epoch": 0.7894736842105263,
      "grad_norm": 0.7573777835614758,
      "learning_rate": 9.362508261669739e-06,
      "loss": 0.7777,
      "num_tokens": 67344223.0,
      "step": 330
    },
    {
      "epoch": 0.7918660287081339,
      "grad_norm": 0.7196830972315318,
      "learning_rate": 9.358026223026931e-06,
      "loss": 0.9235,
      "num_tokens": 67523494.0,
      "step": 331
    },
    {
      "epoch": 0.7942583732057417,
      "grad_norm": 0.7148268968783641,
      "learning_rate": 9.353529693370827e-06,
      "loss": 0.5793,
      "num_tokens": 67707250.0,
      "step": 332
    },
    {
      "epoch": 0.7966507177033493,
      "grad_norm": 0.7183862252835895,
      "learning_rate": 9.3490186895907e-06,
      "loss": 0.9153,
      "num_tokens": 67887746.0,
      "step": 333
    },
    {
      "epoch": 0.7990430622009569,
      "grad_norm": 0.6783981868743665,
      "learning_rate": 9.344493228630194e-06,
      "loss": 1.0557,
      "num_tokens": 68117792.0,
      "step": 334
    },
    {
      "epoch": 0.8014354066985646,
      "grad_norm": 0.6366624212064868,
      "learning_rate": 9.339953327487254e-06,
      "loss": 0.7942,
      "num_tokens": 68366492.0,
      "step": 335
    },
    {
      "epoch": 0.8038277511961722,
      "grad_norm": 0.7078016914757955,
      "learning_rate": 9.335399003214062e-06,
      "loss": 0.5402,
      "num_tokens": 68548241.0,
      "step": 336
    },
    {
      "epoch": 0.80622009569378,
      "grad_norm": 0.6522725955130363,
      "learning_rate": 9.330830272916974e-06,
      "loss": 0.9855,
      "num_tokens": 68763750.0,
      "step": 337
    },
    {
      "epoch": 0.8086124401913876,
      "grad_norm": 0.6214168725386654,
      "learning_rate": 9.326247153756459e-06,
      "loss": 0.86,
      "num_tokens": 69001165.0,
      "step": 338
    },
    {
      "epoch": 0.8110047846889952,
      "grad_norm": 0.7314720468232042,
      "learning_rate": 9.321649662947027e-06,
      "loss": 0.5351,
      "num_tokens": 69178248.0,
      "step": 339
    },
    {
      "epoch": 0.8133971291866029,
      "grad_norm": 0.65272447368887,
      "learning_rate": 9.317037817757175e-06,
      "loss": 0.8351,
      "num_tokens": 69428803.0,
      "step": 340
    },
    {
      "epoch": 0.8157894736842105,
      "grad_norm": 0.5839621781348253,
      "learning_rate": 9.312411635509309e-06,
      "loss": 1.0737,
      "num_tokens": 69712302.0,
      "step": 341
    },
    {
      "epoch": 0.8181818181818182,
      "grad_norm": 0.7016886773959953,
      "learning_rate": 9.307771133579691e-06,
      "loss": 0.9811,
      "num_tokens": 69928563.0,
      "step": 342
    },
    {
      "epoch": 0.8205741626794258,
      "grad_norm": 0.6965166415390514,
      "learning_rate": 9.303116329398367e-06,
      "loss": 0.842,
      "num_tokens": 70161341.0,
      "step": 343
    },
    {
      "epoch": 0.8229665071770335,
      "grad_norm": 0.6875678411924595,
      "learning_rate": 9.298447240449104e-06,
      "loss": 0.7473,
      "num_tokens": 70381060.0,
      "step": 344
    },
    {
      "epoch": 0.8253588516746412,
      "grad_norm": 0.6438807693082683,
      "learning_rate": 9.293763884269321e-06,
      "loss": 0.8741,
      "num_tokens": 70598036.0,
      "step": 345
    },
    {
      "epoch": 0.8277511961722488,
      "grad_norm": 0.7613247244695108,
      "learning_rate": 9.28906627845003e-06,
      "loss": 0.7014,
      "num_tokens": 70784049.0,
      "step": 346
    },
    {
      "epoch": 0.8301435406698564,
      "grad_norm": 0.6738748732421641,
      "learning_rate": 9.284354440635763e-06,
      "loss": 0.7834,
      "num_tokens": 70976591.0,
      "step": 347
    },
    {
      "epoch": 0.8325358851674641,
      "grad_norm": 0.699827251012226,
      "learning_rate": 9.279628388524506e-06,
      "loss": 0.5531,
      "num_tokens": 71141435.0,
      "step": 348
    },
    {
      "epoch": 0.8349282296650717,
      "grad_norm": 0.6478605379617053,
      "learning_rate": 9.274888139867642e-06,
      "loss": 0.6004,
      "num_tokens": 71392389.0,
      "step": 349
    },
    {
      "epoch": 0.8373205741626795,
      "grad_norm": 0.6628070766800735,
      "learning_rate": 9.27013371246987e-06,
      "loss": 0.6012,
      "num_tokens": 71589359.0,
      "step": 350
    },
    {
      "epoch": 0.8397129186602871,
      "grad_norm": 0.6747088798563603,
      "learning_rate": 9.265365124189147e-06,
      "loss": 0.793,
      "num_tokens": 71781509.0,
      "step": 351
    },
    {
      "epoch": 0.8421052631578947,
      "grad_norm": 0.6633362054226132,
      "learning_rate": 9.26058239293662e-06,
      "loss": 0.841,
      "num_tokens": 72024732.0,
      "step": 352
    },
    {
      "epoch": 0.8444976076555024,
      "grad_norm": 0.6728638102656455,
      "learning_rate": 9.255785536676563e-06,
      "loss": 0.9799,
      "num_tokens": 72239477.0,
      "step": 353
    },
    {
      "epoch": 0.84688995215311,
      "grad_norm": 0.7031237520856609,
      "learning_rate": 9.250974573426298e-06,
      "loss": 0.6896,
      "num_tokens": 72394017.0,
      "step": 354
    },
    {
      "epoch": 0.8492822966507177,
      "grad_norm": 0.6852896746420252,
      "learning_rate": 9.24614952125613e-06,
      "loss": 0.8719,
      "num_tokens": 72580756.0,
      "step": 355
    },
    {
      "epoch": 0.8516746411483254,
      "grad_norm": 0.6138862436489377,
      "learning_rate": 9.241310398289298e-06,
      "loss": 0.5569,
      "num_tokens": 72818717.0,
      "step": 356
    },
    {
      "epoch": 0.854066985645933,
      "grad_norm": 0.6957775906573956,
      "learning_rate": 9.236457222701877e-06,
      "loss": 0.9885,
      "num_tokens": 73026311.0,
      "step": 357
    },
    {
      "epoch": 0.8564593301435407,
      "grad_norm": 0.6881414684962115,
      "learning_rate": 9.23159001272273e-06,
      "loss": 0.9276,
      "num_tokens": 73229574.0,
      "step": 358
    },
    {
      "epoch": 0.8588516746411483,
      "grad_norm": 0.6506924307578823,
      "learning_rate": 9.226708786633436e-06,
      "loss": 0.8401,
      "num_tokens": 73470348.0,
      "step": 359
    },
    {
      "epoch": 0.861244019138756,
      "grad_norm": 0.6825146769213832,
      "learning_rate": 9.221813562768215e-06,
      "loss": 0.8945,
      "num_tokens": 73651355.0,
      "step": 360
    },
    {
      "epoch": 0.8636363636363636,
      "grad_norm": 0.6636653793892663,
      "learning_rate": 9.21690435951387e-06,
      "loss": 1.0136,
      "num_tokens": 73867323.0,
      "step": 361
    },
    {
      "epoch": 0.8660287081339713,
      "grad_norm": 0.682534402054643,
      "learning_rate": 9.211981195309706e-06,
      "loss": 0.745,
      "num_tokens": 74056459.0,
      "step": 362
    },
    {
      "epoch": 0.868421052631579,
      "grad_norm": 0.675060886089477,
      "learning_rate": 9.207044088647469e-06,
      "loss": 0.9638,
      "num_tokens": 74297192.0,
      "step": 363
    },
    {
      "epoch": 0.8708133971291866,
      "grad_norm": 0.6614026961374179,
      "learning_rate": 9.202093058071271e-06,
      "loss": 0.7406,
      "num_tokens": 74493323.0,
      "step": 364
    },
    {
      "epoch": 0.8732057416267942,
      "grad_norm": 0.7020816511860245,
      "learning_rate": 9.197128122177529e-06,
      "loss": 0.7018,
      "num_tokens": 74657888.0,
      "step": 365
    },
    {
      "epoch": 0.8755980861244019,
      "grad_norm": 0.7344090939912941,
      "learning_rate": 9.192149299614881e-06,
      "loss": 0.9503,
      "num_tokens": 74843306.0,
      "step": 366
    },
    {
      "epoch": 0.8779904306220095,
      "grad_norm": 0.7127202448586417,
      "learning_rate": 9.187156609084132e-06,
      "loss": 0.8734,
      "num_tokens": 75041637.0,
      "step": 367
    },
    {
      "epoch": 0.8803827751196173,
      "grad_norm": 0.7554609285292794,
      "learning_rate": 9.182150069338173e-06,
      "loss": 0.8169,
      "num_tokens": 75216247.0,
      "step": 368
    },
    {
      "epoch": 0.8827751196172249,
      "grad_norm": 0.6837024175271184,
      "learning_rate": 9.177129699181911e-06,
      "loss": 0.8137,
      "num_tokens": 75397921.0,
      "step": 369
    },
    {
      "epoch": 0.8851674641148325,
      "grad_norm": 0.6828943305704646,
      "learning_rate": 9.172095517472207e-06,
      "loss": 0.7669,
      "num_tokens": 75582214.0,
      "step": 370
    },
    {
      "epoch": 0.8875598086124402,
      "grad_norm": 0.7170012830419972,
      "learning_rate": 9.167047543117792e-06,
      "loss": 0.7986,
      "num_tokens": 75749744.0,
      "step": 371
    },
    {
      "epoch": 0.8899521531100478,
      "grad_norm": 0.6160468357211861,
      "learning_rate": 9.16198579507921e-06,
      "loss": 0.7408,
      "num_tokens": 75977317.0,
      "step": 372
    },
    {
      "epoch": 0.8923444976076556,
      "grad_norm": 0.5876983959606766,
      "learning_rate": 9.156910292368735e-06,
      "loss": 0.7504,
      "num_tokens": 76181889.0,
      "step": 373
    },
    {
      "epoch": 0.8947368421052632,
      "grad_norm": 0.5773610843835477,
      "learning_rate": 9.151821054050304e-06,
      "loss": 0.8454,
      "num_tokens": 76438027.0,
      "step": 374
    },
    {
      "epoch": 0.8971291866028708,
      "grad_norm": 0.6552968917204041,
      "learning_rate": 9.146718099239453e-06,
      "loss": 0.8435,
      "num_tokens": 76623156.0,
      "step": 375
    },
    {
      "epoch": 0.8995215311004785,
      "grad_norm": 0.5899716503345066,
      "learning_rate": 9.14160144710323e-06,
      "loss": 0.8287,
      "num_tokens": 76850856.0,
      "step": 376
    },
    {
      "epoch": 0.9019138755980861,
      "grad_norm": 0.6987492577459424,
      "learning_rate": 9.136471116860135e-06,
      "loss": 0.9,
      "num_tokens": 77008225.0,
      "step": 377
    },
    {
      "epoch": 0.9043062200956937,
      "grad_norm": 0.7290004187885213,
      "learning_rate": 9.131327127780042e-06,
      "loss": 0.7641,
      "num_tokens": 77219739.0,
      "step": 378
    },
    {
      "epoch": 0.9066985645933014,
      "grad_norm": 0.6480714471754585,
      "learning_rate": 9.126169499184133e-06,
      "loss": 0.6829,
      "num_tokens": 77397964.0,
      "step": 379
    },
    {
      "epoch": 0.9090909090909091,
      "grad_norm": 0.690068715632962,
      "learning_rate": 9.120998250444814e-06,
      "loss": 0.8802,
      "num_tokens": 77587726.0,
      "step": 380
    },
    {
      "epoch": 0.9114832535885168,
      "grad_norm": 0.6432928968716475,
      "learning_rate": 9.115813400985655e-06,
      "loss": 0.7432,
      "num_tokens": 77811303.0,
      "step": 381
    },
    {
      "epoch": 0.9138755980861244,
      "grad_norm": 0.6914007018250378,
      "learning_rate": 9.110614970281308e-06,
      "loss": 1.0123,
      "num_tokens": 78007317.0,
      "step": 382
    },
    {
      "epoch": 0.916267942583732,
      "grad_norm": 0.613689907821979,
      "learning_rate": 9.105402977857438e-06,
      "loss": 0.9744,
      "num_tokens": 78274135.0,
      "step": 383
    },
    {
      "epoch": 0.9186602870813397,
      "grad_norm": 0.7804443703479781,
      "learning_rate": 9.100177443290651e-06,
      "loss": 1.0173,
      "num_tokens": 78412125.0,
      "step": 384
    },
    {
      "epoch": 0.9210526315789473,
      "grad_norm": 0.6498222123812953,
      "learning_rate": 9.094938386208413e-06,
      "loss": 0.8756,
      "num_tokens": 78639268.0,
      "step": 385
    },
    {
      "epoch": 0.9234449760765551,
      "grad_norm": 0.7035590974736037,
      "learning_rate": 9.089685826288988e-06,
      "loss": 1.0546,
      "num_tokens": 78874435.0,
      "step": 386
    },
    {
      "epoch": 0.9258373205741627,
      "grad_norm": 0.6759814498789215,
      "learning_rate": 9.084419783261352e-06,
      "loss": 0.739,
      "num_tokens": 79069513.0,
      "step": 387
    },
    {
      "epoch": 0.9282296650717703,
      "grad_norm": 0.7725332697456008,
      "learning_rate": 9.07914027690513e-06,
      "loss": 0.8059,
      "num_tokens": 79220713.0,
      "step": 388
    },
    {
      "epoch": 0.930622009569378,
      "grad_norm": 0.6396922845104327,
      "learning_rate": 9.073847327050508e-06,
      "loss": 0.8805,
      "num_tokens": 79421216.0,
      "step": 389
    },
    {
      "epoch": 0.9330143540669856,
      "grad_norm": 0.6816021616837575,
      "learning_rate": 9.068540953578177e-06,
      "loss": 0.889,
      "num_tokens": 79617679.0,
      "step": 390
    },
    {
      "epoch": 0.9354066985645934,
      "grad_norm": 0.6625189742716404,
      "learning_rate": 9.063221176419239e-06,
      "loss": 0.7465,
      "num_tokens": 79815971.0,
      "step": 391
    },
    {
      "epoch": 0.937799043062201,
      "grad_norm": 0.7346141410671172,
      "learning_rate": 9.057888015555146e-06,
      "loss": 0.9849,
      "num_tokens": 79998530.0,
      "step": 392
    },
    {
      "epoch": 0.9401913875598086,
      "grad_norm": 0.7256211133967554,
      "learning_rate": 9.052541491017617e-06,
      "loss": 0.9619,
      "num_tokens": 80177175.0,
      "step": 393
    },
    {
      "epoch": 0.9425837320574163,
      "grad_norm": 0.6160901991706935,
      "learning_rate": 9.047181622888573e-06,
      "loss": 0.9088,
      "num_tokens": 80414212.0,
      "step": 394
    },
    {
      "epoch": 0.9449760765550239,
      "grad_norm": 0.6347176991976088,
      "learning_rate": 9.041808431300045e-06,
      "loss": 0.6943,
      "num_tokens": 80641688.0,
      "step": 395
    },
    {
      "epoch": 0.9473684210526315,
      "grad_norm": 0.7262184312989681,
      "learning_rate": 9.036421936434113e-06,
      "loss": 0.6596,
      "num_tokens": 80799917.0,
      "step": 396
    },
    {
      "epoch": 0.9497607655502392,
      "grad_norm": 0.7600322039283162,
      "learning_rate": 9.031022158522825e-06,
      "loss": 0.7362,
      "num_tokens": 80966302.0,
      "step": 397
    },
    {
      "epoch": 0.9521531100478469,
      "grad_norm": 0.6116472259730488,
      "learning_rate": 9.025609117848121e-06,
      "loss": 0.8165,
      "num_tokens": 81200995.0,
      "step": 398
    },
    {
      "epoch": 0.9545454545454546,
      "grad_norm": 0.6381990133240583,
      "learning_rate": 9.020182834741753e-06,
      "loss": 0.7013,
      "num_tokens": 81402899.0,
      "step": 399
    },
    {
      "epoch": 0.9569377990430622,
      "grad_norm": 0.6650882585682993,
      "learning_rate": 9.014743329585219e-06,
      "loss": 0.7528,
      "num_tokens": 81596933.0,
      "step": 400
    },
    {
      "epoch": 0.9593301435406698,
      "grad_norm": 0.7178489094483574,
      "learning_rate": 9.009290622809675e-06,
      "loss": 0.8122,
      "num_tokens": 81787868.0,
      "step": 401
    },
    {
      "epoch": 0.9617224880382775,
      "grad_norm": 0.678302286870584,
      "learning_rate": 9.003824734895865e-06,
      "loss": 0.8968,
      "num_tokens": 81985553.0,
      "step": 402
    },
    {
      "epoch": 0.9641148325358851,
      "grad_norm": 0.5812234524449984,
      "learning_rate": 8.998345686374041e-06,
      "loss": 0.4995,
      "num_tokens": 82196566.0,
      "step": 403
    },
    {
      "epoch": 0.9665071770334929,
      "grad_norm": 0.6680187444668858,
      "learning_rate": 8.992853497823892e-06,
      "loss": 0.9803,
      "num_tokens": 82416154.0,
      "step": 404
    },
    {
      "epoch": 0.9688995215311005,
      "grad_norm": 0.5808799825851646,
      "learning_rate": 8.987348189874452e-06,
      "loss": 0.8295,
      "num_tokens": 82679639.0,
      "step": 405
    },
    {
      "epoch": 0.9712918660287081,
      "grad_norm": 0.7747877111516284,
      "learning_rate": 8.981829783204043e-06,
      "loss": 0.8987,
      "num_tokens": 82865891.0,
      "step": 406
    },
    {
      "epoch": 0.9736842105263158,
      "grad_norm": 0.6578557909185174,
      "learning_rate": 8.97629829854018e-06,
      "loss": 0.7227,
      "num_tokens": 83036757.0,
      "step": 407
    },
    {
      "epoch": 0.9760765550239234,
      "grad_norm": 0.7083961242499112,
      "learning_rate": 8.970753756659503e-06,
      "loss": 0.823,
      "num_tokens": 83209010.0,
      "step": 408
    },
    {
      "epoch": 0.9784688995215312,
      "grad_norm": 0.679889621867238,
      "learning_rate": 8.965196178387694e-06,
      "loss": 1.0539,
      "num_tokens": 83414309.0,
      "step": 409
    },
    {
      "epoch": 0.9808612440191388,
      "grad_norm": 0.7007964954333357,
      "learning_rate": 8.959625584599402e-06,
      "loss": 0.85,
      "num_tokens": 83592167.0,
      "step": 410
    },
    {
      "epoch": 0.9832535885167464,
      "grad_norm": 0.6438718189253033,
      "learning_rate": 8.954041996218161e-06,
      "loss": 0.7298,
      "num_tokens": 83805708.0,
      "step": 411
    },
    {
      "epoch": 0.9856459330143541,
      "grad_norm": 0.7316281539767336,
      "learning_rate": 8.948445434216318e-06,
      "loss": 0.7838,
      "num_tokens": 83949063.0,
      "step": 412
    },
    {
      "epoch": 0.9880382775119617,
      "grad_norm": 0.6926173532035703,
      "learning_rate": 8.942835919614944e-06,
      "loss": 0.6193,
      "num_tokens": 84162009.0,
      "step": 413
    },
    {
      "epoch": 0.9904306220095693,
      "grad_norm": 0.7817319996989844,
      "learning_rate": 8.937213473483764e-06,
      "loss": 0.8445,
      "num_tokens": 84364009.0,
      "step": 414
    },
    {
      "epoch": 0.992822966507177,
      "grad_norm": 0.7552374300980621,
      "learning_rate": 8.931578116941076e-06,
      "loss": 0.7978,
      "num_tokens": 84546258.0,
      "step": 415
    },
    {
      "epoch": 0.9952153110047847,
      "grad_norm": 0.6591622274548676,
      "learning_rate": 8.925929871153669e-06,
      "loss": 0.8502,
      "num_tokens": 84772093.0,
      "step": 416
    },
    {
      "epoch": 0.9976076555023924,
      "grad_norm": 0.6609823889041686,
      "learning_rate": 8.920268757336743e-06,
      "loss": 0.9196,
      "num_tokens": 85012261.0,
      "step": 417
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6485021614358057,
      "learning_rate": 8.914594796753835e-06,
      "loss": 0.7938,
      "num_tokens": 85251468.0,
      "step": 418
    },
    {
      "epoch": 1.0,
      "eval_loss": 0.5268629193305969,
      "eval_num_tokens": 85251468.0,
      "eval_runtime": 110.895,
      "eval_samples_per_second": 26.8,
      "eval_steps_per_second": 3.355,
      "step": 418
    },
    {
      "epoch": 1.0023923444976077,
      "grad_norm": 0.8833626539066562,
      "learning_rate": 8.908908010716731e-06,
      "loss": 0.7925,
      "num_tokens": 85384390.0,
      "step": 419
    },
    {
      "epoch": 1.0047846889952152,
      "grad_norm": 0.6501794385479605,
      "learning_rate": 8.903208420585396e-06,
      "loss": 0.7715,
      "num_tokens": 85609801.0,
      "step": 420
    },
    {
      "epoch": 1.007177033492823,
      "grad_norm": 0.6943673036447846,
      "learning_rate": 8.89749604776788e-06,
      "loss": 0.7712,
      "num_tokens": 85790579.0,
      "step": 421
    },
    {
      "epoch": 1.0095693779904307,
      "grad_norm": 0.7449979227556588,
      "learning_rate": 8.891770913720254e-06,
      "loss": 0.5588,
      "num_tokens": 85982217.0,
      "step": 422
    },
    {
      "epoch": 1.0119617224880382,
      "grad_norm": 0.7280868590547691,
      "learning_rate": 8.886033039946517e-06,
      "loss": 0.675,
      "num_tokens": 86236063.0,
      "step": 423
    },
    {
      "epoch": 1.014354066985646,
      "grad_norm": 0.6303280123936105,
      "learning_rate": 8.88028244799852e-06,
      "loss": 0.6639,
      "num_tokens": 86421835.0,
      "step": 424
    },
    {
      "epoch": 1.0167464114832536,
      "grad_norm": 0.6853812289146318,
      "learning_rate": 8.874519159475883e-06,
      "loss": 0.7649,
      "num_tokens": 86622859.0,
      "step": 425
    },
    {
      "epoch": 1.0191387559808613,
      "grad_norm": 0.59208146860945,
      "learning_rate": 8.868743196025916e-06,
      "loss": 0.7123,
      "num_tokens": 86881615.0,
      "step": 426
    },
    {
      "epoch": 1.0215311004784688,
      "grad_norm": 0.7469783342201638,
      "learning_rate": 8.862954579343536e-06,
      "loss": 0.7227,
      "num_tokens": 87041292.0,
      "step": 427
    },
    {
      "epoch": 1.0239234449760766,
      "grad_norm": 0.5881653131003812,
      "learning_rate": 8.85715333117119e-06,
      "loss": 0.6107,
      "num_tokens": 87285738.0,
      "step": 428
    },
    {
      "epoch": 1.0263157894736843,
      "grad_norm": 0.5975544652225052,
      "learning_rate": 8.851339473298768e-06,
      "loss": 0.7007,
      "num_tokens": 87490754.0,
      "step": 429
    },
    {
      "epoch": 1.0287081339712918,
      "grad_norm": 0.6413104534073729,
      "learning_rate": 8.845513027563519e-06,
      "loss": 0.6546,
      "num_tokens": 87651942.0,
      "step": 430
    },
    {
      "epoch": 1.0311004784688995,
      "grad_norm": 0.6314558446671377,
      "learning_rate": 8.839674015849979e-06,
      "loss": 0.7543,
      "num_tokens": 87890353.0,
      "step": 431
    },
    {
      "epoch": 1.0334928229665072,
      "grad_norm": 0.7629826412960697,
      "learning_rate": 8.833822460089878e-06,
      "loss": 0.824,
      "num_tokens": 88067568.0,
      "step": 432
    },
    {
      "epoch": 1.0358851674641147,
      "grad_norm": 0.6075050455387743,
      "learning_rate": 8.827958382262065e-06,
      "loss": 0.6635,
      "num_tokens": 88287133.0,
      "step": 433
    },
    {
      "epoch": 1.0382775119617225,
      "grad_norm": 0.6196621512852998,
      "learning_rate": 8.822081804392422e-06,
      "loss": 0.5884,
      "num_tokens": 88486840.0,
      "step": 434
    },
    {
      "epoch": 1.0406698564593302,
      "grad_norm": 0.6961183655503131,
      "learning_rate": 8.81619274855378e-06,
      "loss": 0.7386,
      "num_tokens": 88673475.0,
      "step": 435
    },
    {
      "epoch": 1.0430622009569377,
      "grad_norm": 0.6729951014404428,
      "learning_rate": 8.810291236865841e-06,
      "loss": 0.6455,
      "num_tokens": 88845660.0,
      "step": 436
    },
    {
      "epoch": 1.0454545454545454,
      "grad_norm": 0.6342083555311104,
      "learning_rate": 8.804377291495093e-06,
      "loss": 0.8274,
      "num_tokens": 89088548.0,
      "step": 437
    },
    {
      "epoch": 1.0478468899521531,
      "grad_norm": 0.6701006397404995,
      "learning_rate": 8.798450934654722e-06,
      "loss": 0.6701,
      "num_tokens": 89260368.0,
      "step": 438
    },
    {
      "epoch": 1.0502392344497609,
      "grad_norm": 0.7404737999856115,
      "learning_rate": 8.792512188604536e-06,
      "loss": 0.7418,
      "num_tokens": 89419609.0,
      "step": 439
    },
    {
      "epoch": 1.0526315789473684,
      "grad_norm": 0.6448113889129491,
      "learning_rate": 8.78656107565087e-06,
      "loss": 0.655,
      "num_tokens": 89596692.0,
      "step": 440
    },
    {
      "epoch": 1.055023923444976,
      "grad_norm": 0.558939448736452,
      "learning_rate": 8.780597618146524e-06,
      "loss": 0.7539,
      "num_tokens": 89894246.0,
      "step": 441
    },
    {
      "epoch": 1.0574162679425838,
      "grad_norm": 0.7842683298759312,
      "learning_rate": 8.774621838490652e-06,
      "loss": 0.5933,
      "num_tokens": 90034436.0,
      "step": 442
    },
    {
      "epoch": 1.0598086124401913,
      "grad_norm": 0.7153205331680785,
      "learning_rate": 8.768633759128696e-06,
      "loss": 0.4877,
      "num_tokens": 90204583.0,
      "step": 443
    },
    {
      "epoch": 1.062200956937799,
      "grad_norm": 0.6230069749775586,
      "learning_rate": 8.762633402552296e-06,
      "loss": 0.5801,
      "num_tokens": 90371035.0,
      "step": 444
    },
    {
      "epoch": 1.0645933014354068,
      "grad_norm": 0.5450852716876841,
      "learning_rate": 8.75662079129921e-06,
      "loss": 0.6347,
      "num_tokens": 90592327.0,
      "step": 445
    },
    {
      "epoch": 1.0669856459330143,
      "grad_norm": 0.6532394649960671,
      "learning_rate": 8.750595947953215e-06,
      "loss": 0.5904,
      "num_tokens": 90824557.0,
      "step": 446
    },
    {
      "epoch": 1.069377990430622,
      "grad_norm": 0.6187711188149183,
      "learning_rate": 8.744558895144044e-06,
      "loss": 0.7607,
      "num_tokens": 91073154.0,
      "step": 447
    },
    {
      "epoch": 1.0717703349282297,
      "grad_norm": 0.6980624446279815,
      "learning_rate": 8.738509655547284e-06,
      "loss": 0.7618,
      "num_tokens": 91258963.0,
      "step": 448
    },
    {
      "epoch": 1.0741626794258372,
      "grad_norm": 0.7024003253964036,
      "learning_rate": 8.732448251884296e-06,
      "loss": 0.7303,
      "num_tokens": 91423256.0,
      "step": 449
    },
    {
      "epoch": 1.076555023923445,
      "grad_norm": 0.6634897373813881,
      "learning_rate": 8.726374706922133e-06,
      "loss": 0.7738,
      "num_tokens": 91637291.0,
      "step": 450
    },
    {
      "epoch": 1.0789473684210527,
      "grad_norm": 0.6804572080651244,
      "learning_rate": 8.72028904347345e-06,
      "loss": 0.7228,
      "num_tokens": 91841448.0,
      "step": 451
    },
    {
      "epoch": 1.0813397129186604,
      "grad_norm": 0.6518235040560235,
      "learning_rate": 8.714191284396417e-06,
      "loss": 0.8062,
      "num_tokens": 92053059.0,
      "step": 452
    },
    {
      "epoch": 1.0837320574162679,
      "grad_norm": 0.6080056046652248,
      "learning_rate": 8.708081452594644e-06,
      "loss": 0.6634,
      "num_tokens": 92261100.0,
      "step": 453
    },
    {
      "epoch": 1.0861244019138756,
      "grad_norm": 0.6544718325563127,
      "learning_rate": 8.701959571017075e-06,
      "loss": 0.6247,
      "num_tokens": 92464515.0,
      "step": 454
    },
    {
      "epoch": 1.0885167464114833,
      "grad_norm": 0.7381644122745519,
      "learning_rate": 8.695825662657926e-06,
      "loss": 0.8531,
      "num_tokens": 92644031.0,
      "step": 455
    },
    {
      "epoch": 1.0909090909090908,
      "grad_norm": 0.6061504580235114,
      "learning_rate": 8.68967975055658e-06,
      "loss": 0.687,
      "num_tokens": 92853081.0,
      "step": 456
    },
    {
      "epoch": 1.0933014354066986,
      "grad_norm": 0.657190432984543,
      "learning_rate": 8.683521857797506e-06,
      "loss": 0.8088,
      "num_tokens": 93084673.0,
      "step": 457
    },
    {
      "epoch": 1.0956937799043063,
      "grad_norm": 0.6531620668793157,
      "learning_rate": 8.677352007510175e-06,
      "loss": 0.6926,
      "num_tokens": 93272082.0,
      "step": 458
    },
    {
      "epoch": 1.0980861244019138,
      "grad_norm": 0.5469254561243377,
      "learning_rate": 8.671170222868973e-06,
      "loss": 0.8201,
      "num_tokens": 93532996.0,
      "step": 459
    },
    {
      "epoch": 1.1004784688995215,
      "grad_norm": 0.7224234205027193,
      "learning_rate": 8.664976527093111e-06,
      "loss": 0.6561,
      "num_tokens": 93700551.0,
      "step": 460
    },
    {
      "epoch": 1.1028708133971292,
      "grad_norm": 0.6103043834106922,
      "learning_rate": 8.658770943446537e-06,
      "loss": 0.6106,
      "num_tokens": 93897552.0,
      "step": 461
    },
    {
      "epoch": 1.1052631578947367,
      "grad_norm": 0.6388200976277861,
      "learning_rate": 8.652553495237854e-06,
      "loss": 0.6085,
      "num_tokens": 94091727.0,
      "step": 462
    },
    {
      "epoch": 1.1076555023923444,
      "grad_norm": 0.6688215830036501,
      "learning_rate": 8.646324205820225e-06,
      "loss": 0.7937,
      "num_tokens": 94323488.0,
      "step": 463
    },
    {
      "epoch": 1.1100478468899522,
      "grad_norm": 0.6499235099896005,
      "learning_rate": 8.640083098591295e-06,
      "loss": 0.6834,
      "num_tokens": 94504378.0,
      "step": 464
    },
    {
      "epoch": 1.11244019138756,
      "grad_norm": 0.6141690793822103,
      "learning_rate": 8.633830196993091e-06,
      "loss": 0.6812,
      "num_tokens": 94704650.0,
      "step": 465
    },
    {
      "epoch": 1.1148325358851674,
      "grad_norm": 0.711438564340125,
      "learning_rate": 8.627565524511947e-06,
      "loss": 0.5216,
      "num_tokens": 94940873.0,
      "step": 466
    },
    {
      "epoch": 1.1172248803827751,
      "grad_norm": 0.6167376352678712,
      "learning_rate": 8.621289104678402e-06,
      "loss": 0.7062,
      "num_tokens": 95151213.0,
      "step": 467
    },
    {
      "epoch": 1.1196172248803828,
      "grad_norm": 0.7384760378511186,
      "learning_rate": 8.615000961067127e-06,
      "loss": 0.5648,
      "num_tokens": 95285914.0,
      "step": 468
    },
    {
      "epoch": 1.1220095693779903,
      "grad_norm": 0.700556827023194,
      "learning_rate": 8.608701117296823e-06,
      "loss": 0.5138,
      "num_tokens": 95444044.0,
      "step": 469
    },
    {
      "epoch": 1.124401913875598,
      "grad_norm": 0.671132609284084,
      "learning_rate": 8.602389597030137e-06,
      "loss": 0.7428,
      "num_tokens": 95680518.0,
      "step": 470
    },
    {
      "epoch": 1.1267942583732058,
      "grad_norm": 0.5442552642709221,
      "learning_rate": 8.596066423973581e-06,
      "loss": 0.595,
      "num_tokens": 95923658.0,
      "step": 471
    },
    {
      "epoch": 1.1291866028708135,
      "grad_norm": 0.6531706451873329,
      "learning_rate": 8.589731621877422e-06,
      "loss": 0.6666,
      "num_tokens": 96114735.0,
      "step": 472
    },
    {
      "epoch": 1.131578947368421,
      "grad_norm": 0.6088153066934165,
      "learning_rate": 8.583385214535623e-06,
      "loss": 0.6571,
      "num_tokens": 96344220.0,
      "step": 473
    },
    {
      "epoch": 1.1339712918660287,
      "grad_norm": 0.6312706973960832,
      "learning_rate": 8.577027225785722e-06,
      "loss": 0.707,
      "num_tokens": 96532999.0,
      "step": 474
    },
    {
      "epoch": 1.1363636363636362,
      "grad_norm": 0.7723559287093527,
      "learning_rate": 8.57065767950877e-06,
      "loss": 0.8366,
      "num_tokens": 96724160.0,
      "step": 475
    },
    {
      "epoch": 1.138755980861244,
      "grad_norm": 0.6292499138689865,
      "learning_rate": 8.56427659962922e-06,
      "loss": 0.6214,
      "num_tokens": 96921531.0,
      "step": 476
    },
    {
      "epoch": 1.1411483253588517,
      "grad_norm": 0.6214894130873831,
      "learning_rate": 8.55788401011485e-06,
      "loss": 0.8551,
      "num_tokens": 97135552.0,
      "step": 477
    },
    {
      "epoch": 1.1435406698564594,
      "grad_norm": 0.6259093670614493,
      "learning_rate": 8.551479934976668e-06,
      "loss": 0.6917,
      "num_tokens": 97345841.0,
      "step": 478
    },
    {
      "epoch": 1.145933014354067,
      "grad_norm": 0.7168111716307588,
      "learning_rate": 8.545064398268826e-06,
      "loss": 0.8033,
      "num_tokens": 97509658.0,
      "step": 479
    },
    {
      "epoch": 1.1483253588516746,
      "grad_norm": 0.5987584068142854,
      "learning_rate": 8.53863742408852e-06,
      "loss": 0.5887,
      "num_tokens": 97731018.0,
      "step": 480
    },
    {
      "epoch": 1.1507177033492824,
      "grad_norm": 0.6900729370961642,
      "learning_rate": 8.53219903657591e-06,
      "loss": 0.8195,
      "num_tokens": 97933426.0,
      "step": 481
    },
    {
      "epoch": 1.1531100478468899,
      "grad_norm": 0.6355834178889574,
      "learning_rate": 8.525749259914024e-06,
      "loss": 0.7962,
      "num_tokens": 98169526.0,
      "step": 482
    },
    {
      "epoch": 1.1555023923444976,
      "grad_norm": 0.6731062920620434,
      "learning_rate": 8.519288118328672e-06,
      "loss": 0.7054,
      "num_tokens": 98324262.0,
      "step": 483
    },
    {
      "epoch": 1.1578947368421053,
      "grad_norm": 0.6674576791981732,
      "learning_rate": 8.512815636088347e-06,
      "loss": 0.669,
      "num_tokens": 98515046.0,
      "step": 484
    },
    {
      "epoch": 1.160287081339713,
      "grad_norm": 0.6869492423073407,
      "learning_rate": 8.50633183750414e-06,
      "loss": 0.7411,
      "num_tokens": 98716661.0,
      "step": 485
    },
    {
      "epoch": 1.1626794258373205,
      "grad_norm": 0.6834157769460931,
      "learning_rate": 8.499836746929648e-06,
      "loss": 0.6594,
      "num_tokens": 98880085.0,
      "step": 486
    },
    {
      "epoch": 1.1650717703349283,
      "grad_norm": 0.6164602298440709,
      "learning_rate": 8.493330388760878e-06,
      "loss": 0.7071,
      "num_tokens": 99126250.0,
      "step": 487
    },
    {
      "epoch": 1.167464114832536,
      "grad_norm": 0.5616797077224229,
      "learning_rate": 8.486812787436162e-06,
      "loss": 0.7958,
      "num_tokens": 99369635.0,
      "step": 488
    },
    {
      "epoch": 1.1698564593301435,
      "grad_norm": 0.715505280608796,
      "learning_rate": 8.480283967436063e-06,
      "loss": 0.7958,
      "num_tokens": 99542434.0,
      "step": 489
    },
    {
      "epoch": 1.1722488038277512,
      "grad_norm": 0.6040191308287042,
      "learning_rate": 8.473743953283279e-06,
      "loss": 0.5818,
      "num_tokens": 99728793.0,
      "step": 490
    },
    {
      "epoch": 1.174641148325359,
      "grad_norm": 0.633301702571846,
      "learning_rate": 8.467192769542555e-06,
      "loss": 0.703,
      "num_tokens": 99929158.0,
      "step": 491
    },
    {
      "epoch": 1.1770334928229664,
      "grad_norm": 0.670950489427574,
      "learning_rate": 8.460630440820593e-06,
      "loss": 0.7055,
      "num_tokens": 100157655.0,
      "step": 492
    },
    {
      "epoch": 1.1794258373205742,
      "grad_norm": 0.616738962710494,
      "learning_rate": 8.454056991765949e-06,
      "loss": 0.5882,
      "num_tokens": 100334982.0,
      "step": 493
    },
    {
      "epoch": 1.1818181818181819,
      "grad_norm": 0.701693251512557,
      "learning_rate": 8.447472447068957e-06,
      "loss": 0.7114,
      "num_tokens": 100520642.0,
      "step": 494
    },
    {
      "epoch": 1.1842105263157894,
      "grad_norm": 0.6910368950769709,
      "learning_rate": 8.440876831461615e-06,
      "loss": 0.817,
      "num_tokens": 100729788.0,
      "step": 495
    },
    {
      "epoch": 1.186602870813397,
      "grad_norm": 0.6007792355517747,
      "learning_rate": 8.434270169717521e-06,
      "loss": 0.6787,
      "num_tokens": 100979908.0,
      "step": 496
    },
    {
      "epoch": 1.1889952153110048,
      "grad_norm": 0.7122991248891785,
      "learning_rate": 8.427652486651749e-06,
      "loss": 0.7728,
      "num_tokens": 101151147.0,
      "step": 497
    },
    {
      "epoch": 1.1913875598086126,
      "grad_norm": 0.6917781088203647,
      "learning_rate": 8.42102380712077e-06,
      "loss": 0.7039,
      "num_tokens": 101312071.0,
      "step": 498
    },
    {
      "epoch": 1.19377990430622,
      "grad_norm": 0.58665825021754,
      "learning_rate": 8.41438415602237e-06,
      "loss": 0.5133,
      "num_tokens": 101493495.0,
      "step": 499
    },
    {
      "epoch": 1.1961722488038278,
      "grad_norm": 0.6161205811479572,
      "learning_rate": 8.407733558295534e-06,
      "loss": 0.6248,
      "num_tokens": 101690177.0,
      "step": 500
    },
    {
      "epoch": 1.1985645933014355,
      "grad_norm": 0.6348422730130996,
      "learning_rate": 8.401072038920364e-06,
      "loss": 0.9154,
      "num_tokens": 101940187.0,
      "step": 501
    },
    {
      "epoch": 1.200956937799043,
      "grad_norm": 0.5887797500918408,
      "learning_rate": 8.394399622917993e-06,
      "loss": 0.8369,
      "num_tokens": 102137576.0,
      "step": 502
    },
    {
      "epoch": 1.2033492822966507,
      "grad_norm": 0.622168916271061,
      "learning_rate": 8.387716335350476e-06,
      "loss": 0.7483,
      "num_tokens": 102334478.0,
      "step": 503
    },
    {
      "epoch": 1.2057416267942584,
      "grad_norm": 0.5362188064080355,
      "learning_rate": 8.3810222013207e-06,
      "loss": 0.6779,
      "num_tokens": 102580656.0,
      "step": 504
    },
    {
      "epoch": 1.208133971291866,
      "grad_norm": 0.6236816878968677,
      "learning_rate": 8.374317245972297e-06,
      "loss": 0.5656,
      "num_tokens": 102754539.0,
      "step": 505
    },
    {
      "epoch": 1.2105263157894737,
      "grad_norm": 0.6710884082075333,
      "learning_rate": 8.367601494489544e-06,
      "loss": 0.7807,
      "num_tokens": 102956836.0,
      "step": 506
    },
    {
      "epoch": 1.2129186602870814,
      "grad_norm": 0.5160236694619329,
      "learning_rate": 8.360874972097267e-06,
      "loss": 0.627,
      "num_tokens": 103222597.0,
      "step": 507
    },
    {
      "epoch": 1.215311004784689,
      "grad_norm": 0.6392481521941529,
      "learning_rate": 8.354137704060751e-06,
      "loss": 0.7102,
      "num_tokens": 103432011.0,
      "step": 508
    },
    {
      "epoch": 1.2177033492822966,
      "grad_norm": 0.5609407707750648,
      "learning_rate": 8.347389715685638e-06,
      "loss": 0.5622,
      "num_tokens": 103662118.0,
      "step": 509
    },
    {
      "epoch": 1.2200956937799043,
      "grad_norm": 0.5857654279746968,
      "learning_rate": 8.34063103231784e-06,
      "loss": 0.7978,
      "num_tokens": 103935876.0,
      "step": 510
    },
    {
      "epoch": 1.222488038277512,
      "grad_norm": 0.6427090354154541,
      "learning_rate": 8.33386167934344e-06,
      "loss": 0.8944,
      "num_tokens": 104179417.0,
      "step": 511
    },
    {
      "epoch": 1.2248803827751196,
      "grad_norm": 0.5388067729124021,
      "learning_rate": 8.327081682188594e-06,
      "loss": 0.6181,
      "num_tokens": 104404008.0,
      "step": 512
    },
    {
      "epoch": 1.2272727272727273,
      "grad_norm": 0.5968370274704392,
      "learning_rate": 8.320291066319442e-06,
      "loss": 0.8699,
      "num_tokens": 104627408.0,
      "step": 513
    },
    {
      "epoch": 1.229665071770335,
      "grad_norm": 0.5878293652309775,
      "learning_rate": 8.313489857242006e-06,
      "loss": 0.792,
      "num_tokens": 104875829.0,
      "step": 514
    },
    {
      "epoch": 1.2320574162679425,
      "grad_norm": 0.6157999004770994,
      "learning_rate": 8.306678080502099e-06,
      "loss": 0.679,
      "num_tokens": 105046140.0,
      "step": 515
    },
    {
      "epoch": 1.2344497607655502,
      "grad_norm": 0.5526340028957575,
      "learning_rate": 8.299855761685223e-06,
      "loss": 0.6366,
      "num_tokens": 105270581.0,
      "step": 516
    },
    {
      "epoch": 1.236842105263158,
      "grad_norm": 0.5659621965437514,
      "learning_rate": 8.293022926416482e-06,
      "loss": 0.6995,
      "num_tokens": 105485976.0,
      "step": 517
    },
    {
      "epoch": 1.2392344497607655,
      "grad_norm": 0.611032283336834,
      "learning_rate": 8.286179600360476e-06,
      "loss": 0.6958,
      "num_tokens": 105700505.0,
      "step": 518
    },
    {
      "epoch": 1.2416267942583732,
      "grad_norm": 0.6139639547074152,
      "learning_rate": 8.27932580922121e-06,
      "loss": 0.7304,
      "num_tokens": 105903407.0,
      "step": 519
    },
    {
      "epoch": 1.244019138755981,
      "grad_norm": 0.7850060555839244,
      "learning_rate": 8.272461578742001e-06,
      "loss": 0.6765,
      "num_tokens": 106052358.0,
      "step": 520
    },
    {
      "epoch": 1.2464114832535884,
      "grad_norm": 0.6629378613315833,
      "learning_rate": 8.265586934705371e-06,
      "loss": 0.5863,
      "num_tokens": 106237968.0,
      "step": 521
    },
    {
      "epoch": 1.2488038277511961,
      "grad_norm": 0.628395091568339,
      "learning_rate": 8.258701902932957e-06,
      "loss": 0.9289,
      "num_tokens": 106450419.0,
      "step": 522
    },
    {
      "epoch": 1.2511961722488039,
      "grad_norm": 0.633356030726603,
      "learning_rate": 8.251806509285418e-06,
      "loss": 0.9009,
      "num_tokens": 106702311.0,
      "step": 523
    },
    {
      "epoch": 1.2535885167464116,
      "grad_norm": 0.6917776003797732,
      "learning_rate": 8.244900779662326e-06,
      "loss": 0.8736,
      "num_tokens": 106917705.0,
      "step": 524
    },
    {
      "epoch": 1.255980861244019,
      "grad_norm": 0.5759491864770921,
      "learning_rate": 8.237984740002084e-06,
      "loss": 0.6075,
      "num_tokens": 107147203.0,
      "step": 525
    },
    {
      "epoch": 1.2583732057416268,
      "grad_norm": 0.6458032921809065,
      "learning_rate": 8.23105841628181e-06,
      "loss": 0.6169,
      "num_tokens": 107333669.0,
      "step": 526
    },
    {
      "epoch": 1.2607655502392343,
      "grad_norm": 0.6915332882979576,
      "learning_rate": 8.224121834517253e-06,
      "loss": 0.6451,
      "num_tokens": 107483189.0,
      "step": 527
    },
    {
      "epoch": 1.263157894736842,
      "grad_norm": 0.761378206128293,
      "learning_rate": 8.217175020762702e-06,
      "loss": 0.725,
      "num_tokens": 107631808.0,
      "step": 528
    },
    {
      "epoch": 1.2655502392344498,
      "grad_norm": 0.7381435867769686,
      "learning_rate": 8.210218001110862e-06,
      "loss": 0.6566,
      "num_tokens": 107795891.0,
      "step": 529
    },
    {
      "epoch": 1.2679425837320575,
      "grad_norm": 0.6783225501913807,
      "learning_rate": 8.203250801692786e-06,
      "loss": 0.7918,
      "num_tokens": 107979143.0,
      "step": 530
    },
    {
      "epoch": 1.2703349282296652,
      "grad_norm": 0.6484130153813725,
      "learning_rate": 8.196273448677753e-06,
      "loss": 0.669,
      "num_tokens": 108196590.0,
      "step": 531
    },
    {
      "epoch": 1.2727272727272727,
      "grad_norm": 0.6554392135157858,
      "learning_rate": 8.189285968273185e-06,
      "loss": 0.9,
      "num_tokens": 108381814.0,
      "step": 532
    },
    {
      "epoch": 1.2751196172248804,
      "grad_norm": 0.5849041476113271,
      "learning_rate": 8.18228838672454e-06,
      "loss": 0.7717,
      "num_tokens": 108619346.0,
      "step": 533
    },
    {
      "epoch": 1.277511961722488,
      "grad_norm": 0.5350083618643547,
      "learning_rate": 8.175280730315222e-06,
      "loss": 0.595,
      "num_tokens": 108875870.0,
      "step": 534
    },
    {
      "epoch": 1.2799043062200957,
      "grad_norm": 0.6159114405886904,
      "learning_rate": 8.168263025366468e-06,
      "loss": 0.8042,
      "num_tokens": 109079053.0,
      "step": 535
    },
    {
      "epoch": 1.2822966507177034,
      "grad_norm": 0.6158840971619612,
      "learning_rate": 8.161235298237268e-06,
      "loss": 0.6988,
      "num_tokens": 109262309.0,
      "step": 536
    },
    {
      "epoch": 1.284688995215311,
      "grad_norm": 0.6938016995261597,
      "learning_rate": 8.154197575324247e-06,
      "loss": 0.7512,
      "num_tokens": 109483891.0,
      "step": 537
    },
    {
      "epoch": 1.2870813397129186,
      "grad_norm": 0.7241914849049002,
      "learning_rate": 8.147149883061581e-06,
      "loss": 0.6713,
      "num_tokens": 109636457.0,
      "step": 538
    },
    {
      "epoch": 1.2894736842105263,
      "grad_norm": 0.5664246048988882,
      "learning_rate": 8.14009224792089e-06,
      "loss": 0.7083,
      "num_tokens": 109869798.0,
      "step": 539
    },
    {
      "epoch": 1.291866028708134,
      "grad_norm": 0.6406286160132263,
      "learning_rate": 8.133024696411138e-06,
      "loss": 0.6811,
      "num_tokens": 110055693.0,
      "step": 540
    },
    {
      "epoch": 1.2942583732057416,
      "grad_norm": 0.6123102900846681,
      "learning_rate": 8.125947255078537e-06,
      "loss": 0.6826,
      "num_tokens": 110294193.0,
      "step": 541
    },
    {
      "epoch": 1.2966507177033493,
      "grad_norm": 0.7129531959007127,
      "learning_rate": 8.11885995050645e-06,
      "loss": 0.4591,
      "num_tokens": 110438211.0,
      "step": 542
    },
    {
      "epoch": 1.299043062200957,
      "grad_norm": 0.6384794663335771,
      "learning_rate": 8.111762809315278e-06,
      "loss": 1.0005,
      "num_tokens": 110698383.0,
      "step": 543
    },
    {
      "epoch": 1.3014354066985647,
      "grad_norm": 0.7042025683566201,
      "learning_rate": 8.10465585816238e-06,
      "loss": 0.7055,
      "num_tokens": 110867049.0,
      "step": 544
    },
    {
      "epoch": 1.3038277511961722,
      "grad_norm": 0.5768164534769485,
      "learning_rate": 8.097539123741949e-06,
      "loss": 0.6007,
      "num_tokens": 111108415.0,
      "step": 545
    },
    {
      "epoch": 1.30622009569378,
      "grad_norm": 0.6313351384712172,
      "learning_rate": 8.090412632784934e-06,
      "loss": 0.7005,
      "num_tokens": 111323074.0,
      "step": 546
    },
    {
      "epoch": 1.3086124401913874,
      "grad_norm": 0.6206830248756131,
      "learning_rate": 8.083276412058932e-06,
      "loss": 0.6337,
      "num_tokens": 111537304.0,
      "step": 547
    },
    {
      "epoch": 1.3110047846889952,
      "grad_norm": 0.6758816496685596,
      "learning_rate": 8.076130488368078e-06,
      "loss": 0.7958,
      "num_tokens": 111747591.0,
      "step": 548
    },
    {
      "epoch": 1.313397129186603,
      "grad_norm": 0.6241461189184837,
      "learning_rate": 8.068974888552955e-06,
      "loss": 0.7754,
      "num_tokens": 112006485.0,
      "step": 549
    },
    {
      "epoch": 1.3157894736842106,
      "grad_norm": 0.6991048663837968,
      "learning_rate": 8.061809639490493e-06,
      "loss": 0.6889,
      "num_tokens": 112182235.0,
      "step": 550
    },
    {
      "epoch": 1.3181818181818181,
      "grad_norm": 0.6561488497102201,
      "learning_rate": 8.054634768093861e-06,
      "loss": 0.8882,
      "num_tokens": 112411061.0,
      "step": 551
    },
    {
      "epoch": 1.3205741626794258,
      "grad_norm": 0.6114674626884281,
      "learning_rate": 8.047450301312373e-06,
      "loss": 0.6171,
      "num_tokens": 112594695.0,
      "step": 552
    },
    {
      "epoch": 1.3229665071770336,
      "grad_norm": 0.6947222688278742,
      "learning_rate": 8.040256266131384e-06,
      "loss": 0.7327,
      "num_tokens": 112796582.0,
      "step": 553
    },
    {
      "epoch": 1.325358851674641,
      "grad_norm": 0.7174977428549918,
      "learning_rate": 8.033052689572185e-06,
      "loss": 0.9901,
      "num_tokens": 113018372.0,
      "step": 554
    },
    {
      "epoch": 1.3277511961722488,
      "grad_norm": 0.5459671536037778,
      "learning_rate": 8.02583959869191e-06,
      "loss": 0.4489,
      "num_tokens": 113224353.0,
      "step": 555
    },
    {
      "epoch": 1.3301435406698565,
      "grad_norm": 0.6098137823161369,
      "learning_rate": 8.018617020583429e-06,
      "loss": 0.829,
      "num_tokens": 113427127.0,
      "step": 556
    },
    {
      "epoch": 1.3325358851674642,
      "grad_norm": 0.6568359830867574,
      "learning_rate": 8.011384982375238e-06,
      "loss": 0.5615,
      "num_tokens": 113626089.0,
      "step": 557
    },
    {
      "epoch": 1.3349282296650717,
      "grad_norm": 0.6281135823952779,
      "learning_rate": 8.00414351123138e-06,
      "loss": 0.7058,
      "num_tokens": 113865385.0,
      "step": 558
    },
    {
      "epoch": 1.3373205741626795,
      "grad_norm": 0.6010690345416376,
      "learning_rate": 7.996892634351319e-06,
      "loss": 0.6422,
      "num_tokens": 114094828.0,
      "step": 559
    },
    {
      "epoch": 1.339712918660287,
      "grad_norm": 0.638515263025877,
      "learning_rate": 7.989632378969849e-06,
      "loss": 0.6424,
      "num_tokens": 114285775.0,
      "step": 560
    },
    {
      "epoch": 1.3421052631578947,
      "grad_norm": 0.6605018239443282,
      "learning_rate": 7.982362772356989e-06,
      "loss": 0.709,
      "num_tokens": 114464592.0,
      "step": 561
    },
    {
      "epoch": 1.3444976076555024,
      "grad_norm": 0.633121921756574,
      "learning_rate": 7.97508384181789e-06,
      "loss": 0.6402,
      "num_tokens": 114661703.0,
      "step": 562
    },
    {
      "epoch": 1.3468899521531101,
      "grad_norm": 0.6930520796230434,
      "learning_rate": 7.967795614692714e-06,
      "loss": 0.7155,
      "num_tokens": 114854790.0,
      "step": 563
    },
    {
      "epoch": 1.3492822966507176,
      "grad_norm": 0.5462776230436035,
      "learning_rate": 7.96049811835655e-06,
      "loss": 0.4381,
      "num_tokens": 115063770.0,
      "step": 564
    },
    {
      "epoch": 1.3516746411483254,
      "grad_norm": 0.7349664493374579,
      "learning_rate": 7.953191380219294e-06,
      "loss": 0.6836,
      "num_tokens": 115230976.0,
      "step": 565
    },
    {
      "epoch": 1.354066985645933,
      "grad_norm": 0.6404975561020254,
      "learning_rate": 7.945875427725563e-06,
      "loss": 0.8089,
      "num_tokens": 115449531.0,
      "step": 566
    },
    {
      "epoch": 1.3564593301435406,
      "grad_norm": 0.598351041926338,
      "learning_rate": 7.93855028835458e-06,
      "loss": 0.8505,
      "num_tokens": 115683622.0,
      "step": 567
    },
    {
      "epoch": 1.3588516746411483,
      "grad_norm": 0.5816191942301666,
      "learning_rate": 7.931215989620075e-06,
      "loss": 0.71,
      "num_tokens": 115850770.0,
      "step": 568
    },
    {
      "epoch": 1.361244019138756,
      "grad_norm": 0.5989480503872827,
      "learning_rate": 7.923872559070179e-06,
      "loss": 0.5141,
      "num_tokens": 116051148.0,
      "step": 569
    },
    {
      "epoch": 1.3636363636363638,
      "grad_norm": 0.7146437669553671,
      "learning_rate": 7.916520024287329e-06,
      "loss": 0.7548,
      "num_tokens": 116247421.0,
      "step": 570
    },
    {
      "epoch": 1.3660287081339713,
      "grad_norm": 0.6512219935518021,
      "learning_rate": 7.909158412888152e-06,
      "loss": 0.7714,
      "num_tokens": 116456221.0,
      "step": 571
    },
    {
      "epoch": 1.368421052631579,
      "grad_norm": 0.6640687941436083,
      "learning_rate": 7.901787752523368e-06,
      "loss": 0.7303,
      "num_tokens": 116616642.0,
      "step": 572
    },
    {
      "epoch": 1.3708133971291865,
      "grad_norm": 0.685283033422343,
      "learning_rate": 7.89440807087769e-06,
      "loss": 0.6986,
      "num_tokens": 116828595.0,
      "step": 573
    },
    {
      "epoch": 1.3732057416267942,
      "grad_norm": 0.7264640201047767,
      "learning_rate": 7.887019395669711e-06,
      "loss": 0.8794,
      "num_tokens": 117000928.0,
      "step": 574
    },
    {
      "epoch": 1.375598086124402,
      "grad_norm": 0.6035933154970914,
      "learning_rate": 7.879621754651806e-06,
      "loss": 0.6661,
      "num_tokens": 117224707.0,
      "step": 575
    },
    {
      "epoch": 1.3779904306220097,
      "grad_norm": 0.7356154161250829,
      "learning_rate": 7.872215175610027e-06,
      "loss": 0.7089,
      "num_tokens": 117406833.0,
      "step": 576
    },
    {
      "epoch": 1.3803827751196172,
      "grad_norm": 0.6784361033034778,
      "learning_rate": 7.864799686363998e-06,
      "loss": 0.9435,
      "num_tokens": 117608333.0,
      "step": 577
    },
    {
      "epoch": 1.3827751196172249,
      "grad_norm": 0.601646584885362,
      "learning_rate": 7.857375314766805e-06,
      "loss": 0.6242,
      "num_tokens": 117824859.0,
      "step": 578
    },
    {
      "epoch": 1.3851674641148326,
      "grad_norm": 0.6286606987208951,
      "learning_rate": 7.849942088704908e-06,
      "loss": 0.7129,
      "num_tokens": 118009711.0,
      "step": 579
    },
    {
      "epoch": 1.38755980861244,
      "grad_norm": 0.6712984784969586,
      "learning_rate": 7.842500036098011e-06,
      "loss": 0.7308,
      "num_tokens": 118195660.0,
      "step": 580
    },
    {
      "epoch": 1.3899521531100478,
      "grad_norm": 0.6986289059083648,
      "learning_rate": 7.835049184898984e-06,
      "loss": 0.7085,
      "num_tokens": 118374150.0,
      "step": 581
    },
    {
      "epoch": 1.3923444976076556,
      "grad_norm": 0.6365250045967269,
      "learning_rate": 7.827589563093737e-06,
      "loss": 0.7445,
      "num_tokens": 118578238.0,
      "step": 582
    },
    {
      "epoch": 1.3947368421052633,
      "grad_norm": 0.5980300561132873,
      "learning_rate": 7.820121198701123e-06,
      "loss": 0.732,
      "num_tokens": 118786622.0,
      "step": 583
    },
    {
      "epoch": 1.3971291866028708,
      "grad_norm": 0.5947783020696292,
      "learning_rate": 7.812644119772838e-06,
      "loss": 0.7338,
      "num_tokens": 119008588.0,
      "step": 584
    },
    {
      "epoch": 1.3995215311004785,
      "grad_norm": 0.5903656266151648,
      "learning_rate": 7.80515835439331e-06,
      "loss": 0.7805,
      "num_tokens": 119262356.0,
      "step": 585
    },
    {
      "epoch": 1.401913875598086,
      "grad_norm": 0.6206154261533207,
      "learning_rate": 7.797663930679586e-06,
      "loss": 0.836,
      "num_tokens": 119472323.0,
      "step": 586
    },
    {
      "epoch": 1.4043062200956937,
      "grad_norm": 0.6626657461296346,
      "learning_rate": 7.790160876781242e-06,
      "loss": 0.8119,
      "num_tokens": 119652941.0,
      "step": 587
    },
    {
      "epoch": 1.4066985645933014,
      "grad_norm": 0.6868149066567649,
      "learning_rate": 7.782649220880268e-06,
      "loss": 0.6947,
      "num_tokens": 119833083.0,
      "step": 588
    },
    {
      "epoch": 1.4090909090909092,
      "grad_norm": 0.6204817485919448,
      "learning_rate": 7.775128991190963e-06,
      "loss": 0.7922,
      "num_tokens": 120030493.0,
      "step": 589
    },
    {
      "epoch": 1.4114832535885167,
      "grad_norm": 0.6724390520172234,
      "learning_rate": 7.76760021595983e-06,
      "loss": 0.7076,
      "num_tokens": 120259396.0,
      "step": 590
    },
    {
      "epoch": 1.4138755980861244,
      "grad_norm": 0.6440599092231826,
      "learning_rate": 7.760062923465467e-06,
      "loss": 0.4596,
      "num_tokens": 120416796.0,
      "step": 591
    },
    {
      "epoch": 1.4162679425837321,
      "grad_norm": 0.6218562163131692,
      "learning_rate": 7.75251714201847e-06,
      "loss": 0.6655,
      "num_tokens": 120616896.0,
      "step": 592
    },
    {
      "epoch": 1.4186602870813396,
      "grad_norm": 0.6849812328433857,
      "learning_rate": 7.744962899961311e-06,
      "loss": 0.7767,
      "num_tokens": 120787459.0,
      "step": 593
    },
    {
      "epoch": 1.4210526315789473,
      "grad_norm": 0.6170197137081643,
      "learning_rate": 7.73740022566825e-06,
      "loss": 0.7519,
      "num_tokens": 120998500.0,
      "step": 594
    },
    {
      "epoch": 1.423444976076555,
      "grad_norm": 0.6454119314614041,
      "learning_rate": 7.729829147545214e-06,
      "loss": 0.6476,
      "num_tokens": 121209444.0,
      "step": 595
    },
    {
      "epoch": 1.4258373205741628,
      "grad_norm": 0.6253962654706593,
      "learning_rate": 7.722249694029692e-06,
      "loss": 0.4759,
      "num_tokens": 121377586.0,
      "step": 596
    },
    {
      "epoch": 1.4282296650717703,
      "grad_norm": 0.6096531645355685,
      "learning_rate": 7.714661893590641e-06,
      "loss": 0.7043,
      "num_tokens": 121605803.0,
      "step": 597
    },
    {
      "epoch": 1.430622009569378,
      "grad_norm": 0.6005996949073303,
      "learning_rate": 7.707065774728359e-06,
      "loss": 0.6528,
      "num_tokens": 121832216.0,
      "step": 598
    },
    {
      "epoch": 1.4330143540669855,
      "grad_norm": 0.6389038973889795,
      "learning_rate": 7.699461365974398e-06,
      "loss": 0.551,
      "num_tokens": 122025954.0,
      "step": 599
    },
    {
      "epoch": 1.4354066985645932,
      "grad_norm": 0.6974149384728023,
      "learning_rate": 7.69184869589144e-06,
      "loss": 0.738,
      "num_tokens": 122208176.0,
      "step": 600
    },
    {
      "epoch": 1.437799043062201,
      "grad_norm": 0.6150640365254872,
      "learning_rate": 7.684227793073205e-06,
      "loss": 0.7368,
      "num_tokens": 122410734.0,
      "step": 601
    },
    {
      "epoch": 1.4401913875598087,
      "grad_norm": 0.696837292155682,
      "learning_rate": 7.676598686144326e-06,
      "loss": 0.8521,
      "num_tokens": 122611097.0,
      "step": 602
    },
    {
      "epoch": 1.4425837320574162,
      "grad_norm": 0.6565300019075943,
      "learning_rate": 7.668961403760257e-06,
      "loss": 0.6649,
      "num_tokens": 122803282.0,
      "step": 603
    },
    {
      "epoch": 1.444976076555024,
      "grad_norm": 0.6692653954984873,
      "learning_rate": 7.661315974607161e-06,
      "loss": 0.6806,
      "num_tokens": 123012117.0,
      "step": 604
    },
    {
      "epoch": 1.4473684210526316,
      "grad_norm": 0.6617024095318141,
      "learning_rate": 7.653662427401797e-06,
      "loss": 0.8588,
      "num_tokens": 123255081.0,
      "step": 605
    },
    {
      "epoch": 1.4497607655502391,
      "grad_norm": 0.5964810181267309,
      "learning_rate": 7.646000790891418e-06,
      "loss": 0.5307,
      "num_tokens": 123453656.0,
      "step": 606
    },
    {
      "epoch": 1.4521531100478469,
      "grad_norm": 0.7899864065988431,
      "learning_rate": 7.638331093853662e-06,
      "loss": 0.5959,
      "num_tokens": 123633504.0,
      "step": 607
    },
    {
      "epoch": 1.4545454545454546,
      "grad_norm": 0.6887408990548677,
      "learning_rate": 7.63065336509644e-06,
      "loss": 0.7329,
      "num_tokens": 123801896.0,
      "step": 608
    },
    {
      "epoch": 1.4569377990430623,
      "grad_norm": 0.633552365230913,
      "learning_rate": 7.622967633457831e-06,
      "loss": 0.6681,
      "num_tokens": 123974366.0,
      "step": 609
    },
    {
      "epoch": 1.4593301435406698,
      "grad_norm": 0.6343850458718804,
      "learning_rate": 7.615273927805976e-06,
      "loss": 0.9128,
      "num_tokens": 124190706.0,
      "step": 610
    },
    {
      "epoch": 1.4617224880382775,
      "grad_norm": 0.5748629243715516,
      "learning_rate": 7.6075722770389634e-06,
      "loss": 0.6477,
      "num_tokens": 124403746.0,
      "step": 611
    },
    {
      "epoch": 1.464114832535885,
      "grad_norm": 0.5661320410441629,
      "learning_rate": 7.599862710084731e-06,
      "loss": 0.5624,
      "num_tokens": 124578707.0,
      "step": 612
    },
    {
      "epoch": 1.4665071770334928,
      "grad_norm": 0.5900248167722216,
      "learning_rate": 7.5921452559009424e-06,
      "loss": 0.8237,
      "num_tokens": 124820690.0,
      "step": 613
    },
    {
      "epoch": 1.4688995215311005,
      "grad_norm": 0.6386434549167093,
      "learning_rate": 7.584419943474888e-06,
      "loss": 0.716,
      "num_tokens": 125023721.0,
      "step": 614
    },
    {
      "epoch": 1.4712918660287082,
      "grad_norm": 0.669335586967033,
      "learning_rate": 7.576686801823377e-06,
      "loss": 0.7834,
      "num_tokens": 125226122.0,
      "step": 615
    },
    {
      "epoch": 1.4736842105263157,
      "grad_norm": 0.6165452625138145,
      "learning_rate": 7.5689458599926244e-06,
      "loss": 0.8478,
      "num_tokens": 125465984.0,
      "step": 616
    },
    {
      "epoch": 1.4760765550239234,
      "grad_norm": 0.6508737359095208,
      "learning_rate": 7.5611971470581425e-06,
      "loss": 0.7474,
      "num_tokens": 125660386.0,
      "step": 617
    },
    {
      "epoch": 1.4784688995215312,
      "grad_norm": 0.5374864935102592,
      "learning_rate": 7.553440692124634e-06,
      "loss": 0.4226,
      "num_tokens": 125893050.0,
      "step": 618
    },
    {
      "epoch": 1.4808612440191387,
      "grad_norm": 0.6796724151735416,
      "learning_rate": 7.545676524325878e-06,
      "loss": 0.6615,
      "num_tokens": 126094266.0,
      "step": 619
    },
    {
      "epoch": 1.4832535885167464,
      "grad_norm": 0.5751206841528997,
      "learning_rate": 7.537904672824627e-06,
      "loss": 0.8344,
      "num_tokens": 126333416.0,
      "step": 620
    },
    {
      "epoch": 1.485645933014354,
      "grad_norm": 0.6976735779434823,
      "learning_rate": 7.530125166812493e-06,
      "loss": 0.8913,
      "num_tokens": 126513194.0,
      "step": 621
    },
    {
      "epoch": 1.4880382775119618,
      "grad_norm": 0.7134664678115061,
      "learning_rate": 7.5223380355098376e-06,
      "loss": 0.5069,
      "num_tokens": 126659843.0,
      "step": 622
    },
    {
      "epoch": 1.4904306220095693,
      "grad_norm": 0.6281529318513822,
      "learning_rate": 7.514543308165665e-06,
      "loss": 0.5912,
      "num_tokens": 126854255.0,
      "step": 623
    },
    {
      "epoch": 1.492822966507177,
      "grad_norm": 0.6753085882032426,
      "learning_rate": 7.5067410140575105e-06,
      "loss": 0.644,
      "num_tokens": 127064662.0,
      "step": 624
    },
    {
      "epoch": 1.4952153110047846,
      "grad_norm": 0.6543948938709288,
      "learning_rate": 7.498931182491329e-06,
      "loss": 0.8564,
      "num_tokens": 127279305.0,
      "step": 625
    },
    {
      "epoch": 1.4976076555023923,
      "grad_norm": 0.5694667262337427,
      "learning_rate": 7.491113842801391e-06,
      "loss": 0.6843,
      "num_tokens": 127503404.0,
      "step": 626
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.6154345937543192,
      "learning_rate": 7.483289024350161e-06,
      "loss": 0.7261,
      "num_tokens": 127707884.0,
      "step": 627
    },
    {
      "epoch": 1.5023923444976077,
      "grad_norm": 0.5870500440697171,
      "learning_rate": 7.4754567565282e-06,
      "loss": 0.7015,
      "num_tokens": 127928802.0,
      "step": 628
    },
    {
      "epoch": 1.5047846889952154,
      "grad_norm": 0.6189074200886605,
      "learning_rate": 7.46761706875405e-06,
      "loss": 0.6552,
      "num_tokens": 128132676.0,
      "step": 629
    },
    {
      "epoch": 1.507177033492823,
      "grad_norm": 0.678134402555645,
      "learning_rate": 7.4597699904741164e-06,
      "loss": 0.6806,
      "num_tokens": 128294864.0,
      "step": 630
    },
    {
      "epoch": 1.5095693779904307,
      "grad_norm": 0.6742324121181981,
      "learning_rate": 7.451915551162571e-06,
      "loss": 0.6127,
      "num_tokens": 128467254.0,
      "step": 631
    },
    {
      "epoch": 1.5119617224880382,
      "grad_norm": 0.6561738843878836,
      "learning_rate": 7.444053780321229e-06,
      "loss": 0.669,
      "num_tokens": 128683604.0,
      "step": 632
    },
    {
      "epoch": 1.514354066985646,
      "grad_norm": 0.6712644537160007,
      "learning_rate": 7.436184707479446e-06,
      "loss": 0.5979,
      "num_tokens": 128886445.0,
      "step": 633
    },
    {
      "epoch": 1.5167464114832536,
      "grad_norm": 0.5858800119893577,
      "learning_rate": 7.4283083621940046e-06,
      "loss": 0.9023,
      "num_tokens": 129172873.0,
      "step": 634
    },
    {
      "epoch": 1.5191387559808613,
      "grad_norm": 0.6693350217386079,
      "learning_rate": 7.420424774049001e-06,
      "loss": 0.5293,
      "num_tokens": 129335516.0,
      "step": 635
    },
    {
      "epoch": 1.5215311004784688,
      "grad_norm": 0.5630177602648866,
      "learning_rate": 7.412533972655738e-06,
      "loss": 0.9111,
      "num_tokens": 129643366.0,
      "step": 636
    },
    {
      "epoch": 1.5239234449760766,
      "grad_norm": 0.5465767654675465,
      "learning_rate": 7.4046359876526105e-06,
      "loss": 0.6228,
      "num_tokens": 129873409.0,
      "step": 637
    },
    {
      "epoch": 1.526315789473684,
      "grad_norm": 0.6278118175282995,
      "learning_rate": 7.396730848704998e-06,
      "loss": 0.7113,
      "num_tokens": 130084999.0,
      "step": 638
    },
    {
      "epoch": 1.5287081339712918,
      "grad_norm": 0.5683146907948892,
      "learning_rate": 7.3888185855051496e-06,
      "loss": 0.8121,
      "num_tokens": 130319504.0,
      "step": 639
    },
    {
      "epoch": 1.5311004784688995,
      "grad_norm": 0.5769236478312393,
      "learning_rate": 7.380899227772071e-06,
      "loss": 0.6355,
      "num_tokens": 130571172.0,
      "step": 640
    },
    {
      "epoch": 1.5334928229665072,
      "grad_norm": 0.6040589218978154,
      "learning_rate": 7.372972805251418e-06,
      "loss": 0.7999,
      "num_tokens": 130790304.0,
      "step": 641
    },
    {
      "epoch": 1.535885167464115,
      "grad_norm": 0.6889509880785355,
      "learning_rate": 7.365039347715382e-06,
      "loss": 0.9458,
      "num_tokens": 130983743.0,
      "step": 642
    },
    {
      "epoch": 1.5382775119617225,
      "grad_norm": 0.5518524219982203,
      "learning_rate": 7.3570988849625776e-06,
      "loss": 0.6396,
      "num_tokens": 131223506.0,
      "step": 643
    },
    {
      "epoch": 1.5406698564593302,
      "grad_norm": 0.6263391549174677,
      "learning_rate": 7.349151446817933e-06,
      "loss": 0.7415,
      "num_tokens": 131422092.0,
      "step": 644
    },
    {
      "epoch": 1.5430622009569377,
      "grad_norm": 0.6254985326192385,
      "learning_rate": 7.341197063132578e-06,
      "loss": 0.5479,
      "num_tokens": 131611001.0,
      "step": 645
    },
    {
      "epoch": 1.5454545454545454,
      "grad_norm": 0.7025669116044877,
      "learning_rate": 7.333235763783723e-06,
      "loss": 0.8543,
      "num_tokens": 131798367.0,
      "step": 646
    },
    {
      "epoch": 1.5478468899521531,
      "grad_norm": 0.6769673977558561,
      "learning_rate": 7.325267578674562e-06,
      "loss": 0.7417,
      "num_tokens": 131982165.0,
      "step": 647
    },
    {
      "epoch": 1.5502392344497609,
      "grad_norm": 0.6594417737311544,
      "learning_rate": 7.317292537734146e-06,
      "loss": 0.7702,
      "num_tokens": 132179543.0,
      "step": 648
    },
    {
      "epoch": 1.5526315789473686,
      "grad_norm": 0.5825105050658751,
      "learning_rate": 7.309310670917284e-06,
      "loss": 0.869,
      "num_tokens": 132418034.0,
      "step": 649
    },
    {
      "epoch": 1.555023923444976,
      "grad_norm": 0.6076102425806628,
      "learning_rate": 7.301322008204417e-06,
      "loss": 0.6586,
      "num_tokens": 132649014.0,
      "step": 650
    },
    {
      "epoch": 1.5574162679425836,
      "grad_norm": 0.711423037588177,
      "learning_rate": 7.293326579601516e-06,
      "loss": 0.7468,
      "num_tokens": 132803863.0,
      "step": 651
    },
    {
      "epoch": 1.5598086124401913,
      "grad_norm": 0.7368594597294233,
      "learning_rate": 7.285324415139963e-06,
      "loss": 0.8176,
      "num_tokens": 133026746.0,
      "step": 652
    },
    {
      "epoch": 1.562200956937799,
      "grad_norm": 0.6550397000696636,
      "learning_rate": 7.277315544876443e-06,
      "loss": 0.4705,
      "num_tokens": 133160363.0,
      "step": 653
    },
    {
      "epoch": 1.5645933014354068,
      "grad_norm": 0.6169864021795348,
      "learning_rate": 7.269299998892826e-06,
      "loss": 0.7017,
      "num_tokens": 133372416.0,
      "step": 654
    },
    {
      "epoch": 1.5669856459330145,
      "grad_norm": 0.6851837929301584,
      "learning_rate": 7.261277807296054e-06,
      "loss": 0.6455,
      "num_tokens": 133547830.0,
      "step": 655
    },
    {
      "epoch": 1.569377990430622,
      "grad_norm": 0.5247519311165174,
      "learning_rate": 7.253249000218039e-06,
      "loss": 0.5979,
      "num_tokens": 133786320.0,
      "step": 656
    },
    {
      "epoch": 1.5717703349282297,
      "grad_norm": 0.6589549645314666,
      "learning_rate": 7.245213607815529e-06,
      "loss": 0.5587,
      "num_tokens": 133976402.0,
      "step": 657
    },
    {
      "epoch": 1.5741626794258372,
      "grad_norm": 0.6447471357403248,
      "learning_rate": 7.237171660270019e-06,
      "loss": 0.7355,
      "num_tokens": 134193941.0,
      "step": 658
    },
    {
      "epoch": 1.576555023923445,
      "grad_norm": 0.5905891091552951,
      "learning_rate": 7.229123187787618e-06,
      "loss": 0.7559,
      "num_tokens": 134406719.0,
      "step": 659
    },
    {
      "epoch": 1.5789473684210527,
      "grad_norm": 0.5849610167995951,
      "learning_rate": 7.221068220598944e-06,
      "loss": 0.5986,
      "num_tokens": 134606887.0,
      "step": 660
    },
    {
      "epoch": 1.5813397129186604,
      "grad_norm": 0.6367055646406621,
      "learning_rate": 7.213006788959013e-06,
      "loss": 0.7577,
      "num_tokens": 134802016.0,
      "step": 661
    },
    {
      "epoch": 1.583732057416268,
      "grad_norm": 0.5923713608410193,
      "learning_rate": 7.20493892314712e-06,
      "loss": 0.7655,
      "num_tokens": 135021088.0,
      "step": 662
    },
    {
      "epoch": 1.5861244019138756,
      "grad_norm": 0.6459428419359479,
      "learning_rate": 7.1968646534667245e-06,
      "loss": 0.9141,
      "num_tokens": 135244745.0,
      "step": 663
    },
    {
      "epoch": 1.588516746411483,
      "grad_norm": 0.6394239767985107,
      "learning_rate": 7.188784010245342e-06,
      "loss": 0.7753,
      "num_tokens": 135464277.0,
      "step": 664
    },
    {
      "epoch": 1.5909090909090908,
      "grad_norm": 0.5549326186185076,
      "learning_rate": 7.180697023834429e-06,
      "loss": 0.6716,
      "num_tokens": 135684135.0,
      "step": 665
    },
    {
      "epoch": 1.5933014354066986,
      "grad_norm": 0.6647510124923258,
      "learning_rate": 7.172603724609265e-06,
      "loss": 0.881,
      "num_tokens": 135891765.0,
      "step": 666
    },
    {
      "epoch": 1.5956937799043063,
      "grad_norm": 0.5867761913813149,
      "learning_rate": 7.164504142968842e-06,
      "loss": 0.6829,
      "num_tokens": 136089551.0,
      "step": 667
    },
    {
      "epoch": 1.598086124401914,
      "grad_norm": 0.7373727808155365,
      "learning_rate": 7.156398309335748e-06,
      "loss": 0.656,
      "num_tokens": 136218480.0,
      "step": 668
    },
    {
      "epoch": 1.6004784688995215,
      "grad_norm": 0.6654353752622224,
      "learning_rate": 7.148286254156055e-06,
      "loss": 0.6759,
      "num_tokens": 136394386.0,
      "step": 669
    },
    {
      "epoch": 1.6028708133971292,
      "grad_norm": 0.5865986644953781,
      "learning_rate": 7.140168007899204e-06,
      "loss": 0.6801,
      "num_tokens": 136626576.0,
      "step": 670
    },
    {
      "epoch": 1.6052631578947367,
      "grad_norm": 0.6059106005903319,
      "learning_rate": 7.1320436010578885e-06,
      "loss": 0.748,
      "num_tokens": 136876727.0,
      "step": 671
    },
    {
      "epoch": 1.6076555023923444,
      "grad_norm": 0.6863180911732465,
      "learning_rate": 7.123913064147942e-06,
      "loss": 0.7255,
      "num_tokens": 137073722.0,
      "step": 672
    },
    {
      "epoch": 1.6100478468899522,
      "grad_norm": 0.677113903994993,
      "learning_rate": 7.115776427708225e-06,
      "loss": 0.7247,
      "num_tokens": 137277798.0,
      "step": 673
    },
    {
      "epoch": 1.61244019138756,
      "grad_norm": 0.6030904942096611,
      "learning_rate": 7.107633722300504e-06,
      "loss": 0.8239,
      "num_tokens": 137541312.0,
      "step": 674
    },
    {
      "epoch": 1.6148325358851676,
      "grad_norm": 0.658973988174715,
      "learning_rate": 7.099484978509347e-06,
      "loss": 0.7388,
      "num_tokens": 137732486.0,
      "step": 675
    },
    {
      "epoch": 1.6172248803827751,
      "grad_norm": 0.5504898493296739,
      "learning_rate": 7.091330226941997e-06,
      "loss": 0.6264,
      "num_tokens": 137966847.0,
      "step": 676
    },
    {
      "epoch": 1.6196172248803826,
      "grad_norm": 0.5982217124976851,
      "learning_rate": 7.083169498228265e-06,
      "loss": 0.7082,
      "num_tokens": 138167570.0,
      "step": 677
    },
    {
      "epoch": 1.6220095693779903,
      "grad_norm": 0.6922794170445625,
      "learning_rate": 7.075002823020415e-06,
      "loss": 0.8387,
      "num_tokens": 138382643.0,
      "step": 678
    },
    {
      "epoch": 1.624401913875598,
      "grad_norm": 0.572655561281994,
      "learning_rate": 7.0668302319930385e-06,
      "loss": 0.6965,
      "num_tokens": 138585117.0,
      "step": 679
    },
    {
      "epoch": 1.6267942583732058,
      "grad_norm": 0.6552294365015775,
      "learning_rate": 7.0586517558429565e-06,
      "loss": 0.8051,
      "num_tokens": 138793586.0,
      "step": 680
    },
    {
      "epoch": 1.6291866028708135,
      "grad_norm": 0.681546332536658,
      "learning_rate": 7.05046742528909e-06,
      "loss": 0.5143,
      "num_tokens": 138946513.0,
      "step": 681
    },
    {
      "epoch": 1.631578947368421,
      "grad_norm": 0.663097578505039,
      "learning_rate": 7.042277271072353e-06,
      "loss": 0.7839,
      "num_tokens": 139128215.0,
      "step": 682
    },
    {
      "epoch": 1.6339712918660287,
      "grad_norm": 0.6866957270444947,
      "learning_rate": 7.034081323955526e-06,
      "loss": 0.6465,
      "num_tokens": 139352216.0,
      "step": 683
    },
    {
      "epoch": 1.6363636363636362,
      "grad_norm": 0.6068063835514473,
      "learning_rate": 7.025879614723159e-06,
      "loss": 0.7776,
      "num_tokens": 139582137.0,
      "step": 684
    },
    {
      "epoch": 1.638755980861244,
      "grad_norm": 0.6668218142807971,
      "learning_rate": 7.017672174181433e-06,
      "loss": 0.5697,
      "num_tokens": 139767821.0,
      "step": 685
    },
    {
      "epoch": 1.6411483253588517,
      "grad_norm": 0.5761416700134772,
      "learning_rate": 7.009459033158067e-06,
      "loss": 0.7146,
      "num_tokens": 140016403.0,
      "step": 686
    },
    {
      "epoch": 1.6435406698564594,
      "grad_norm": 0.6212350178196538,
      "learning_rate": 7.001240222502185e-06,
      "loss": 0.5901,
      "num_tokens": 140215130.0,
      "step": 687
    },
    {
      "epoch": 1.6459330143540671,
      "grad_norm": 0.6932671836785006,
      "learning_rate": 6.993015773084209e-06,
      "loss": 0.8066,
      "num_tokens": 140435213.0,
      "step": 688
    },
    {
      "epoch": 1.6483253588516746,
      "grad_norm": 0.6152901109158736,
      "learning_rate": 6.984785715795736e-06,
      "loss": 0.7012,
      "num_tokens": 140638974.0,
      "step": 689
    },
    {
      "epoch": 1.6507177033492821,
      "grad_norm": 0.6046502480861653,
      "learning_rate": 6.976550081549437e-06,
      "loss": 0.6277,
      "num_tokens": 140828485.0,
      "step": 690
    },
    {
      "epoch": 1.6531100478468899,
      "grad_norm": 0.624543382041799,
      "learning_rate": 6.968308901278918e-06,
      "loss": 0.7345,
      "num_tokens": 141024806.0,
      "step": 691
    },
    {
      "epoch": 1.6555023923444976,
      "grad_norm": 0.5853674718023154,
      "learning_rate": 6.960062205938625e-06,
      "loss": 0.5641,
      "num_tokens": 141226224.0,
      "step": 692
    },
    {
      "epoch": 1.6578947368421053,
      "grad_norm": 0.5897568318638701,
      "learning_rate": 6.951810026503716e-06,
      "loss": 0.6473,
      "num_tokens": 141456777.0,
      "step": 693
    },
    {
      "epoch": 1.660287081339713,
      "grad_norm": 0.5965536569799941,
      "learning_rate": 6.943552393969945e-06,
      "loss": 0.6568,
      "num_tokens": 141698392.0,
      "step": 694
    },
    {
      "epoch": 1.6626794258373205,
      "grad_norm": 0.6216221057717923,
      "learning_rate": 6.935289339353554e-06,
      "loss": 0.7735,
      "num_tokens": 141952515.0,
      "step": 695
    },
    {
      "epoch": 1.6650717703349283,
      "grad_norm": 0.5628621673381613,
      "learning_rate": 6.927020893691144e-06,
      "loss": 0.701,
      "num_tokens": 142193916.0,
      "step": 696
    },
    {
      "epoch": 1.6674641148325358,
      "grad_norm": 0.721909783452347,
      "learning_rate": 6.918747088039573e-06,
      "loss": 0.7463,
      "num_tokens": 142382031.0,
      "step": 697
    },
    {
      "epoch": 1.6698564593301435,
      "grad_norm": 0.6160712988562029,
      "learning_rate": 6.910467953475825e-06,
      "loss": 0.6198,
      "num_tokens": 142589032.0,
      "step": 698
    },
    {
      "epoch": 1.6722488038277512,
      "grad_norm": 0.5777273958814299,
      "learning_rate": 6.902183521096903e-06,
      "loss": 0.6441,
      "num_tokens": 142816199.0,
      "step": 699
    },
    {
      "epoch": 1.674641148325359,
      "grad_norm": 0.66250696939839,
      "learning_rate": 6.8938938220197045e-06,
      "loss": 0.6611,
      "num_tokens": 142981069.0,
      "step": 700
    },
    {
      "epoch": 1.6770334928229667,
      "grad_norm": 0.613928495672255,
      "learning_rate": 6.885598887380916e-06,
      "loss": 0.5948,
      "num_tokens": 143161674.0,
      "step": 701
    },
    {
      "epoch": 1.6794258373205742,
      "grad_norm": 0.5576396758289864,
      "learning_rate": 6.877298748336886e-06,
      "loss": 0.6836,
      "num_tokens": 143401819.0,
      "step": 702
    },
    {
      "epoch": 1.6818181818181817,
      "grad_norm": 0.6883792449494041,
      "learning_rate": 6.868993436063508e-06,
      "loss": 0.4856,
      "num_tokens": 143603641.0,
      "step": 703
    },
    {
      "epoch": 1.6842105263157894,
      "grad_norm": 0.6392186575702383,
      "learning_rate": 6.860682981756112e-06,
      "loss": 0.7784,
      "num_tokens": 143827071.0,
      "step": 704
    },
    {
      "epoch": 1.686602870813397,
      "grad_norm": 0.6598165642309378,
      "learning_rate": 6.852367416629336e-06,
      "loss": 0.8908,
      "num_tokens": 144047846.0,
      "step": 705
    },
    {
      "epoch": 1.6889952153110048,
      "grad_norm": 0.6584885917359966,
      "learning_rate": 6.844046771917021e-06,
      "loss": 0.4543,
      "num_tokens": 144225181.0,
      "step": 706
    },
    {
      "epoch": 1.6913875598086126,
      "grad_norm": 0.5207686938231294,
      "learning_rate": 6.8357210788720815e-06,
      "loss": 0.6706,
      "num_tokens": 144457553.0,
      "step": 707
    },
    {
      "epoch": 1.69377990430622,
      "grad_norm": 0.5857688524496493,
      "learning_rate": 6.827390368766399e-06,
      "loss": 0.5963,
      "num_tokens": 144700567.0,
      "step": 708
    },
    {
      "epoch": 1.6961722488038278,
      "grad_norm": 0.6269445216008943,
      "learning_rate": 6.819054672890694e-06,
      "loss": 0.6396,
      "num_tokens": 144898391.0,
      "step": 709
    },
    {
      "epoch": 1.6985645933014353,
      "grad_norm": 0.681265371552301,
      "learning_rate": 6.810714022554416e-06,
      "loss": 0.6671,
      "num_tokens": 145085594.0,
      "step": 710
    },
    {
      "epoch": 1.700956937799043,
      "grad_norm": 0.6385132137297136,
      "learning_rate": 6.802368449085629e-06,
      "loss": 0.7816,
      "num_tokens": 145313346.0,
      "step": 711
    },
    {
      "epoch": 1.7033492822966507,
      "grad_norm": 0.5699021834675809,
      "learning_rate": 6.79401798383088e-06,
      "loss": 0.5681,
      "num_tokens": 145537646.0,
      "step": 712
    },
    {
      "epoch": 1.7057416267942584,
      "grad_norm": 0.6766184855175188,
      "learning_rate": 6.785662658155096e-06,
      "loss": 0.7845,
      "num_tokens": 145746945.0,
      "step": 713
    },
    {
      "epoch": 1.7081339712918662,
      "grad_norm": 0.6563572860619684,
      "learning_rate": 6.777302503441456e-06,
      "loss": 0.7179,
      "num_tokens": 145958251.0,
      "step": 714
    },
    {
      "epoch": 1.7105263157894737,
      "grad_norm": 0.6546511561234387,
      "learning_rate": 6.768937551091281e-06,
      "loss": 0.6195,
      "num_tokens": 146148849.0,
      "step": 715
    },
    {
      "epoch": 1.7129186602870812,
      "grad_norm": 0.6292286462040894,
      "learning_rate": 6.760567832523911e-06,
      "loss": 0.5005,
      "num_tokens": 146356588.0,
      "step": 716
    },
    {
      "epoch": 1.715311004784689,
      "grad_norm": 0.6810222536271355,
      "learning_rate": 6.7521933791765845e-06,
      "loss": 0.8664,
      "num_tokens": 146568777.0,
      "step": 717
    },
    {
      "epoch": 1.7177033492822966,
      "grad_norm": 0.558428668986471,
      "learning_rate": 6.7438142225043326e-06,
      "loss": 0.7028,
      "num_tokens": 146825389.0,
      "step": 718
    },
    {
      "epoch": 1.7200956937799043,
      "grad_norm": 0.6455640793650751,
      "learning_rate": 6.735430393979844e-06,
      "loss": 0.7942,
      "num_tokens": 147063089.0,
      "step": 719
    },
    {
      "epoch": 1.722488038277512,
      "grad_norm": 0.6213200830644523,
      "learning_rate": 6.727041925093356e-06,
      "loss": 0.5403,
      "num_tokens": 147245237.0,
      "step": 720
    },
    {
      "epoch": 1.7248803827751196,
      "grad_norm": 0.6664752095444129,
      "learning_rate": 6.718648847352544e-06,
      "loss": 0.6582,
      "num_tokens": 147420404.0,
      "step": 721
    },
    {
      "epoch": 1.7272727272727273,
      "grad_norm": 0.7118439039143982,
      "learning_rate": 6.710251192282383e-06,
      "loss": 0.7656,
      "num_tokens": 147611537.0,
      "step": 722
    },
    {
      "epoch": 1.7296650717703348,
      "grad_norm": 0.7033855462925209,
      "learning_rate": 6.701848991425048e-06,
      "loss": 0.8086,
      "num_tokens": 147792423.0,
      "step": 723
    },
    {
      "epoch": 1.7320574162679425,
      "grad_norm": 0.5338887128010329,
      "learning_rate": 6.69344227633979e-06,
      "loss": 0.6233,
      "num_tokens": 148036593.0,
      "step": 724
    },
    {
      "epoch": 1.7344497607655502,
      "grad_norm": 0.627987489446626,
      "learning_rate": 6.685031078602806e-06,
      "loss": 0.539,
      "num_tokens": 148247028.0,
      "step": 725
    },
    {
      "epoch": 1.736842105263158,
      "grad_norm": 0.6380003491433351,
      "learning_rate": 6.676615429807139e-06,
      "loss": 0.7812,
      "num_tokens": 148445651.0,
      "step": 726
    },
    {
      "epoch": 1.7392344497607657,
      "grad_norm": 0.6304048306139799,
      "learning_rate": 6.66819536156255e-06,
      "loss": 0.7064,
      "num_tokens": 148649603.0,
      "step": 727
    },
    {
      "epoch": 1.7416267942583732,
      "grad_norm": 0.6961464516761647,
      "learning_rate": 6.659770905495395e-06,
      "loss": 0.8016,
      "num_tokens": 148826982.0,
      "step": 728
    },
    {
      "epoch": 1.744019138755981,
      "grad_norm": 0.6297522222022299,
      "learning_rate": 6.651342093248514e-06,
      "loss": 0.5682,
      "num_tokens": 148985475.0,
      "step": 729
    },
    {
      "epoch": 1.7464114832535884,
      "grad_norm": 0.6740019228321092,
      "learning_rate": 6.64290895648111e-06,
      "loss": 0.759,
      "num_tokens": 149191208.0,
      "step": 730
    },
    {
      "epoch": 1.7488038277511961,
      "grad_norm": 0.6670699466136719,
      "learning_rate": 6.634471526868626e-06,
      "loss": 0.6014,
      "num_tokens": 149354730.0,
      "step": 731
    },
    {
      "epoch": 1.7511961722488039,
      "grad_norm": 0.6738246947970548,
      "learning_rate": 6.626029836102632e-06,
      "loss": 0.7066,
      "num_tokens": 149585290.0,
      "step": 732
    },
    {
      "epoch": 1.7535885167464116,
      "grad_norm": 0.7257178534956213,
      "learning_rate": 6.6175839158907044e-06,
      "loss": 0.7729,
      "num_tokens": 149748808.0,
      "step": 733
    },
    {
      "epoch": 1.755980861244019,
      "grad_norm": 0.5813926473198574,
      "learning_rate": 6.609133797956299e-06,
      "loss": 0.6473,
      "num_tokens": 149986613.0,
      "step": 734
    },
    {
      "epoch": 1.7583732057416268,
      "grad_norm": 0.5696351716269552,
      "learning_rate": 6.600679514038648e-06,
      "loss": 0.7636,
      "num_tokens": 150243571.0,
      "step": 735
    },
    {
      "epoch": 1.7607655502392343,
      "grad_norm": 0.6984568231501039,
      "learning_rate": 6.592221095892624e-06,
      "loss": 0.7487,
      "num_tokens": 150445035.0,
      "step": 736
    },
    {
      "epoch": 1.763157894736842,
      "grad_norm": 0.7068183140491251,
      "learning_rate": 6.583758575288629e-06,
      "loss": 0.7338,
      "num_tokens": 150613995.0,
      "step": 737
    },
    {
      "epoch": 1.7655502392344498,
      "grad_norm": 0.5681101022881475,
      "learning_rate": 6.575291984012479e-06,
      "loss": 0.6505,
      "num_tokens": 150888694.0,
      "step": 738
    },
    {
      "epoch": 1.7679425837320575,
      "grad_norm": 0.6844383241481893,
      "learning_rate": 6.566821353865274e-06,
      "loss": 0.8043,
      "num_tokens": 151074729.0,
      "step": 739
    },
    {
      "epoch": 1.7703349282296652,
      "grad_norm": 0.6698154744229244,
      "learning_rate": 6.558346716663288e-06,
      "loss": 0.564,
      "num_tokens": 151278234.0,
      "step": 740
    },
    {
      "epoch": 1.7727272727272727,
      "grad_norm": 0.6931357382617912,
      "learning_rate": 6.5498681042378436e-06,
      "loss": 0.8781,
      "num_tokens": 151454406.0,
      "step": 741
    },
    {
      "epoch": 1.7751196172248804,
      "grad_norm": 0.547487238351369,
      "learning_rate": 6.541385548435194e-06,
      "loss": 0.7641,
      "num_tokens": 151721146.0,
      "step": 742
    },
    {
      "epoch": 1.777511961722488,
      "grad_norm": 0.5211188643543073,
      "learning_rate": 6.532899081116407e-06,
      "loss": 0.5243,
      "num_tokens": 151907808.0,
      "step": 743
    },
    {
      "epoch": 1.7799043062200957,
      "grad_norm": 0.5903546234259213,
      "learning_rate": 6.524408734157243e-06,
      "loss": 0.7527,
      "num_tokens": 152097386.0,
      "step": 744
    },
    {
      "epoch": 1.7822966507177034,
      "grad_norm": 0.7281834521077791,
      "learning_rate": 6.515914539448027e-06,
      "loss": 0.675,
      "num_tokens": 152251887.0,
      "step": 745
    },
    {
      "epoch": 1.784688995215311,
      "grad_norm": 0.7070506737237091,
      "learning_rate": 6.5074165288935465e-06,
      "loss": 0.7677,
      "num_tokens": 152427986.0,
      "step": 746
    },
    {
      "epoch": 1.7870813397129188,
      "grad_norm": 0.6725409977391403,
      "learning_rate": 6.498914734412914e-06,
      "loss": 0.7322,
      "num_tokens": 152632145.0,
      "step": 747
    },
    {
      "epoch": 1.7894736842105263,
      "grad_norm": 0.6286109645379147,
      "learning_rate": 6.490409187939461e-06,
      "loss": 0.533,
      "num_tokens": 152800360.0,
      "step": 748
    },
    {
      "epoch": 1.7918660287081338,
      "grad_norm": 0.6577712868510451,
      "learning_rate": 6.481899921420605e-06,
      "loss": 0.7657,
      "num_tokens": 152999064.0,
      "step": 749
    },
    {
      "epoch": 1.7942583732057416,
      "grad_norm": 0.5492567256359078,
      "learning_rate": 6.473386966817739e-06,
      "loss": 0.4693,
      "num_tokens": 153209749.0,
      "step": 750
    },
    {
      "epoch": 1.7966507177033493,
      "grad_norm": 0.635209254410753,
      "learning_rate": 6.464870356106111e-06,
      "loss": 0.5932,
      "num_tokens": 153409227.0,
      "step": 751
    },
    {
      "epoch": 1.799043062200957,
      "grad_norm": 0.5793755424272056,
      "learning_rate": 6.456350121274701e-06,
      "loss": 0.5852,
      "num_tokens": 153636309.0,
      "step": 752
    },
    {
      "epoch": 1.8014354066985647,
      "grad_norm": 0.5893424243716154,
      "learning_rate": 6.447826294326096e-06,
      "loss": 0.5136,
      "num_tokens": 153827119.0,
      "step": 753
    },
    {
      "epoch": 1.8038277511961722,
      "grad_norm": 0.6072000657757534,
      "learning_rate": 6.439298907276383e-06,
      "loss": 0.6568,
      "num_tokens": 154010369.0,
      "step": 754
    },
    {
      "epoch": 1.80622009569378,
      "grad_norm": 0.6585123683717504,
      "learning_rate": 6.430767992155018e-06,
      "loss": 0.7205,
      "num_tokens": 154239396.0,
      "step": 755
    },
    {
      "epoch": 1.8086124401913874,
      "grad_norm": 0.562448012001657,
      "learning_rate": 6.422233581004705e-06,
      "loss": 0.8302,
      "num_tokens": 154529668.0,
      "step": 756
    },
    {
      "epoch": 1.8110047846889952,
      "grad_norm": 0.669204687243846,
      "learning_rate": 6.413695705881285e-06,
      "loss": 0.755,
      "num_tokens": 154716614.0,
      "step": 757
    },
    {
      "epoch": 1.813397129186603,
      "grad_norm": 0.575016512956893,
      "learning_rate": 6.405154398853607e-06,
      "loss": 0.587,
      "num_tokens": 154938736.0,
      "step": 758
    },
    {
      "epoch": 1.8157894736842106,
      "grad_norm": 0.5777414948415983,
      "learning_rate": 6.396609692003409e-06,
      "loss": 0.5674,
      "num_tokens": 155140701.0,
      "step": 759
    },
    {
      "epoch": 1.8181818181818183,
      "grad_norm": 0.5963378705676244,
      "learning_rate": 6.3880616174252045e-06,
      "loss": 0.5978,
      "num_tokens": 155310866.0,
      "step": 760
    },
    {
      "epoch": 1.8205741626794258,
      "grad_norm": 0.6637288549764916,
      "learning_rate": 6.379510207226149e-06,
      "loss": 0.7328,
      "num_tokens": 155490254.0,
      "step": 761
    },
    {
      "epoch": 1.8229665071770333,
      "grad_norm": 0.6737242450951861,
      "learning_rate": 6.370955493525934e-06,
      "loss": 0.597,
      "num_tokens": 155671785.0,
      "step": 762
    },
    {
      "epoch": 1.825358851674641,
      "grad_norm": 0.6542151896826175,
      "learning_rate": 6.362397508456652e-06,
      "loss": 0.656,
      "num_tokens": 155902312.0,
      "step": 763
    },
    {
      "epoch": 1.8277511961722488,
      "grad_norm": 0.5973672724693193,
      "learning_rate": 6.353836284162689e-06,
      "loss": 0.6636,
      "num_tokens": 156085880.0,
      "step": 764
    },
    {
      "epoch": 1.8301435406698565,
      "grad_norm": 0.6154944777126776,
      "learning_rate": 6.345271852800594e-06,
      "loss": 0.7481,
      "num_tokens": 156281186.0,
      "step": 765
    },
    {
      "epoch": 1.8325358851674642,
      "grad_norm": 0.6338112989470145,
      "learning_rate": 6.336704246538967e-06,
      "loss": 0.6602,
      "num_tokens": 156491235.0,
      "step": 766
    },
    {
      "epoch": 1.8349282296650717,
      "grad_norm": 0.6531681516448476,
      "learning_rate": 6.328133497558326e-06,
      "loss": 0.6314,
      "num_tokens": 156685969.0,
      "step": 767
    },
    {
      "epoch": 1.8373205741626795,
      "grad_norm": 0.6708022721442888,
      "learning_rate": 6.319559638050998e-06,
      "loss": 0.758,
      "num_tokens": 156859214.0,
      "step": 768
    },
    {
      "epoch": 1.839712918660287,
      "grad_norm": 0.6556721794330259,
      "learning_rate": 6.310982700220993e-06,
      "loss": 0.7323,
      "num_tokens": 157060937.0,
      "step": 769
    },
    {
      "epoch": 1.8421052631578947,
      "grad_norm": 0.6578595057293558,
      "learning_rate": 6.302402716283883e-06,
      "loss": 0.9535,
      "num_tokens": 157261078.0,
      "step": 770
    },
    {
      "epoch": 1.8444976076555024,
      "grad_norm": 0.7148099674142656,
      "learning_rate": 6.293819718466679e-06,
      "loss": 0.6388,
      "num_tokens": 157414683.0,
      "step": 771
    },
    {
      "epoch": 1.8468899521531101,
      "grad_norm": 0.7238377114336406,
      "learning_rate": 6.2852337390077165e-06,
      "loss": 0.6809,
      "num_tokens": 157602306.0,
      "step": 772
    },
    {
      "epoch": 1.8492822966507179,
      "grad_norm": 0.7749663077031087,
      "learning_rate": 6.276644810156526e-06,
      "loss": 0.8122,
      "num_tokens": 157779076.0,
      "step": 773
    },
    {
      "epoch": 1.8516746411483254,
      "grad_norm": 0.6612435155051035,
      "learning_rate": 6.26805296417372e-06,
      "loss": 0.6311,
      "num_tokens": 157976364.0,
      "step": 774
    },
    {
      "epoch": 1.8540669856459329,
      "grad_norm": 0.5573801868109426,
      "learning_rate": 6.259458233330865e-06,
      "loss": 0.7612,
      "num_tokens": 158225906.0,
      "step": 775
    },
    {
      "epoch": 1.8564593301435406,
      "grad_norm": 0.6210661879260847,
      "learning_rate": 6.250860649910364e-06,
      "loss": 0.7606,
      "num_tokens": 158456010.0,
      "step": 776
    },
    {
      "epoch": 1.8588516746411483,
      "grad_norm": 0.5965280962073838,
      "learning_rate": 6.242260246205336e-06,
      "loss": 0.8011,
      "num_tokens": 158725023.0,
      "step": 777
    },
    {
      "epoch": 1.861244019138756,
      "grad_norm": 0.5411360641638349,
      "learning_rate": 6.233657054519493e-06,
      "loss": 0.7206,
      "num_tokens": 158972556.0,
      "step": 778
    },
    {
      "epoch": 1.8636363636363638,
      "grad_norm": 0.6880950857454029,
      "learning_rate": 6.225051107167016e-06,
      "loss": 0.5751,
      "num_tokens": 159137291.0,
      "step": 779
    },
    {
      "epoch": 1.8660287081339713,
      "grad_norm": 0.6732364239510688,
      "learning_rate": 6.21644243647244e-06,
      "loss": 0.8581,
      "num_tokens": 159304892.0,
      "step": 780
    },
    {
      "epoch": 1.868421052631579,
      "grad_norm": 0.7033409732295164,
      "learning_rate": 6.207831074770525e-06,
      "loss": 0.7957,
      "num_tokens": 159477762.0,
      "step": 781
    },
    {
      "epoch": 1.8708133971291865,
      "grad_norm": 0.6669767550877167,
      "learning_rate": 6.199217054406144e-06,
      "loss": 0.7279,
      "num_tokens": 159692236.0,
      "step": 782
    },
    {
      "epoch": 1.8732057416267942,
      "grad_norm": 0.6329334013965955,
      "learning_rate": 6.190600407734152e-06,
      "loss": 0.4734,
      "num_tokens": 159861359.0,
      "step": 783
    },
    {
      "epoch": 1.875598086124402,
      "grad_norm": 0.6108528654038816,
      "learning_rate": 6.181981167119268e-06,
      "loss": 0.7286,
      "num_tokens": 160078902.0,
      "step": 784
    },
    {
      "epoch": 1.8779904306220097,
      "grad_norm": 0.7284756956486506,
      "learning_rate": 6.173359364935962e-06,
      "loss": 0.8924,
      "num_tokens": 160288294.0,
      "step": 785
    },
    {
      "epoch": 1.8803827751196174,
      "grad_norm": 0.7028581335759021,
      "learning_rate": 6.164735033568311e-06,
      "loss": 0.6249,
      "num_tokens": 160447101.0,
      "step": 786
    },
    {
      "epoch": 1.8827751196172249,
      "grad_norm": 0.5681307594359682,
      "learning_rate": 6.156108205409905e-06,
      "loss": 0.6184,
      "num_tokens": 160648475.0,
      "step": 787
    },
    {
      "epoch": 1.8851674641148324,
      "grad_norm": 0.6548663136064242,
      "learning_rate": 6.147478912863706e-06,
      "loss": 0.8074,
      "num_tokens": 160834767.0,
      "step": 788
    },
    {
      "epoch": 1.88755980861244,
      "grad_norm": 0.5766678861349332,
      "learning_rate": 6.138847188341933e-06,
      "loss": 0.6363,
      "num_tokens": 161064024.0,
      "step": 789
    },
    {
      "epoch": 1.8899521531100478,
      "grad_norm": 0.6277507803468657,
      "learning_rate": 6.130213064265942e-06,
      "loss": 0.6889,
      "num_tokens": 161292692.0,
      "step": 790
    },
    {
      "epoch": 1.8923444976076556,
      "grad_norm": 0.6427100164026258,
      "learning_rate": 6.121576573066097e-06,
      "loss": 0.5882,
      "num_tokens": 161490801.0,
      "step": 791
    },
    {
      "epoch": 1.8947368421052633,
      "grad_norm": 0.5690909727573779,
      "learning_rate": 6.112937747181658e-06,
      "loss": 0.4301,
      "num_tokens": 161670780.0,
      "step": 792
    },
    {
      "epoch": 1.8971291866028708,
      "grad_norm": 0.636924652289299,
      "learning_rate": 6.104296619060654e-06,
      "loss": 0.5616,
      "num_tokens": 161876992.0,
      "step": 793
    },
    {
      "epoch": 1.8995215311004785,
      "grad_norm": 0.6100826828491932,
      "learning_rate": 6.095653221159758e-06,
      "loss": 0.6796,
      "num_tokens": 162076563.0,
      "step": 794
    },
    {
      "epoch": 1.901913875598086,
      "grad_norm": 0.6373158305391043,
      "learning_rate": 6.08700758594417e-06,
      "loss": 0.7603,
      "num_tokens": 162295731.0,
      "step": 795
    },
    {
      "epoch": 1.9043062200956937,
      "grad_norm": 0.6270935536574143,
      "learning_rate": 6.078359745887496e-06,
      "loss": 0.7324,
      "num_tokens": 162463047.0,
      "step": 796
    },
    {
      "epoch": 1.9066985645933014,
      "grad_norm": 0.6888041390063913,
      "learning_rate": 6.069709733471618e-06,
      "loss": 0.7396,
      "num_tokens": 162648928.0,
      "step": 797
    },
    {
      "epoch": 1.9090909090909092,
      "grad_norm": 0.5453051524568404,
      "learning_rate": 6.061057581186583e-06,
      "loss": 0.7578,
      "num_tokens": 162890874.0,
      "step": 798
    },
    {
      "epoch": 1.911483253588517,
      "grad_norm": 0.5878553603380681,
      "learning_rate": 6.052403321530473e-06,
      "loss": 0.7789,
      "num_tokens": 163147285.0,
      "step": 799
    },
    {
      "epoch": 1.9138755980861244,
      "grad_norm": 0.5748854305021395,
      "learning_rate": 6.043746987009287e-06,
      "loss": 0.3847,
      "num_tokens": 163293805.0,
      "step": 800
    },
    {
      "epoch": 1.916267942583732,
      "grad_norm": 0.6041052961662696,
      "learning_rate": 6.035088610136814e-06,
      "loss": 0.7391,
      "num_tokens": 163512727.0,
      "step": 801
    },
    {
      "epoch": 1.9186602870813396,
      "grad_norm": 0.623206001139848,
      "learning_rate": 6.026428223434515e-06,
      "loss": 0.569,
      "num_tokens": 163731841.0,
      "step": 802
    },
    {
      "epoch": 1.9210526315789473,
      "grad_norm": 0.595797010000766,
      "learning_rate": 6.017765859431405e-06,
      "loss": 0.6446,
      "num_tokens": 163944213.0,
      "step": 803
    },
    {
      "epoch": 1.923444976076555,
      "grad_norm": 0.6770597646109086,
      "learning_rate": 6.00910155066392e-06,
      "loss": 0.6816,
      "num_tokens": 164125249.0,
      "step": 804
    },
    {
      "epoch": 1.9258373205741628,
      "grad_norm": 0.6216999025378627,
      "learning_rate": 6.000435329675804e-06,
      "loss": 0.5176,
      "num_tokens": 164287594.0,
      "step": 805
    },
    {
      "epoch": 1.9282296650717703,
      "grad_norm": 0.5779829552728587,
      "learning_rate": 5.991767229017982e-06,
      "loss": 0.6527,
      "num_tokens": 164504811.0,
      "step": 806
    },
    {
      "epoch": 1.930622009569378,
      "grad_norm": 0.5999659117860371,
      "learning_rate": 5.983097281248439e-06,
      "loss": 0.6139,
      "num_tokens": 164700832.0,
      "step": 807
    },
    {
      "epoch": 1.9330143540669855,
      "grad_norm": 0.6925949959790602,
      "learning_rate": 5.974425518932099e-06,
      "loss": 0.736,
      "num_tokens": 164855500.0,
      "step": 808
    },
    {
      "epoch": 1.9354066985645932,
      "grad_norm": 0.664280629821679,
      "learning_rate": 5.9657519746407e-06,
      "loss": 0.7058,
      "num_tokens": 165081012.0,
      "step": 809
    },
    {
      "epoch": 1.937799043062201,
      "grad_norm": 0.6294916529959589,
      "learning_rate": 5.957076680952676e-06,
      "loss": 0.7708,
      "num_tokens": 165321183.0,
      "step": 810
    },
    {
      "epoch": 1.9401913875598087,
      "grad_norm": 0.613260914286205,
      "learning_rate": 5.948399670453029e-06,
      "loss": 0.6023,
      "num_tokens": 165510499.0,
      "step": 811
    },
    {
      "epoch": 1.9425837320574164,
      "grad_norm": 0.7432131661007875,
      "learning_rate": 5.93972097573321e-06,
      "loss": 0.6179,
      "num_tokens": 165680826.0,
      "step": 812
    },
    {
      "epoch": 1.944976076555024,
      "grad_norm": 0.595133384251671,
      "learning_rate": 5.9310406293909976e-06,
      "loss": 0.5926,
      "num_tokens": 165855759.0,
      "step": 813
    },
    {
      "epoch": 1.9473684210526314,
      "grad_norm": 0.5642473636840877,
      "learning_rate": 5.922358664030371e-06,
      "loss": 0.685,
      "num_tokens": 166088776.0,
      "step": 814
    },
    {
      "epoch": 1.9497607655502391,
      "grad_norm": 0.5440878709758462,
      "learning_rate": 5.913675112261395e-06,
      "loss": 0.5341,
      "num_tokens": 166310945.0,
      "step": 815
    },
    {
      "epoch": 1.9521531100478469,
      "grad_norm": 0.604649800399483,
      "learning_rate": 5.9049900067000886e-06,
      "loss": 0.5747,
      "num_tokens": 166528617.0,
      "step": 816
    },
    {
      "epoch": 1.9545454545454546,
      "grad_norm": 0.6751299088807625,
      "learning_rate": 5.896303379968311e-06,
      "loss": 0.7819,
      "num_tokens": 166715195.0,
      "step": 817
    },
    {
      "epoch": 1.9569377990430623,
      "grad_norm": 0.5878703194902412,
      "learning_rate": 5.8876152646936315e-06,
      "loss": 0.5685,
      "num_tokens": 166872201.0,
      "step": 818
    },
    {
      "epoch": 1.9593301435406698,
      "grad_norm": 0.6850872373141852,
      "learning_rate": 5.878925693509211e-06,
      "loss": 0.7066,
      "num_tokens": 167083751.0,
      "step": 819
    },
    {
      "epoch": 1.9617224880382775,
      "grad_norm": 0.721599338200044,
      "learning_rate": 5.870234699053681e-06,
      "loss": 0.7546,
      "num_tokens": 167316838.0,
      "step": 820
    },
    {
      "epoch": 1.964114832535885,
      "grad_norm": 0.5264357129866483,
      "learning_rate": 5.8615423139710184e-06,
      "loss": 0.3273,
      "num_tokens": 167537637.0,
      "step": 821
    },
    {
      "epoch": 1.9665071770334928,
      "grad_norm": 0.7601910916603852,
      "learning_rate": 5.852848570910421e-06,
      "loss": 0.863,
      "num_tokens": 167727717.0,
      "step": 822
    },
    {
      "epoch": 1.9688995215311005,
      "grad_norm": 0.5993187893093678,
      "learning_rate": 5.8441535025261896e-06,
      "loss": 0.7109,
      "num_tokens": 167936009.0,
      "step": 823
    },
    {
      "epoch": 1.9712918660287082,
      "grad_norm": 0.7913031835854982,
      "learning_rate": 5.835457141477604e-06,
      "loss": 0.8485,
      "num_tokens": 168096748.0,
      "step": 824
    },
    {
      "epoch": 1.973684210526316,
      "grad_norm": 0.5170452980689251,
      "learning_rate": 5.826759520428796e-06,
      "loss": 0.53,
      "num_tokens": 168321952.0,
      "step": 825
    },
    {
      "epoch": 1.9760765550239234,
      "grad_norm": 0.5512288566093851,
      "learning_rate": 5.818060672048633e-06,
      "loss": 0.3594,
      "num_tokens": 168487009.0,
      "step": 826
    },
    {
      "epoch": 1.9784688995215312,
      "grad_norm": 0.6451763029149299,
      "learning_rate": 5.809360629010587e-06,
      "loss": 0.6353,
      "num_tokens": 168700356.0,
      "step": 827
    },
    {
      "epoch": 1.9808612440191387,
      "grad_norm": 0.6873439326093063,
      "learning_rate": 5.800659423992627e-06,
      "loss": 0.7417,
      "num_tokens": 168901226.0,
      "step": 828
    },
    {
      "epoch": 1.9832535885167464,
      "grad_norm": 0.7057939062681675,
      "learning_rate": 5.791957089677077e-06,
      "loss": 0.7176,
      "num_tokens": 169106405.0,
      "step": 829
    },
    {
      "epoch": 1.985645933014354,
      "grad_norm": 0.6718548592817507,
      "learning_rate": 5.783253658750508e-06,
      "loss": 0.3632,
      "num_tokens": 169241118.0,
      "step": 830
    },
    {
      "epoch": 1.9880382775119618,
      "grad_norm": 0.5635631634019334,
      "learning_rate": 5.774549163903609e-06,
      "loss": 0.5848,
      "num_tokens": 169443391.0,
      "step": 831
    },
    {
      "epoch": 1.9904306220095693,
      "grad_norm": 0.5979096434267269,
      "learning_rate": 5.765843637831064e-06,
      "loss": 0.6669,
      "num_tokens": 169656881.0,
      "step": 832
    },
    {
      "epoch": 1.992822966507177,
      "grad_norm": 0.6955815014808082,
      "learning_rate": 5.7571371132314295e-06,
      "loss": 0.7672,
      "num_tokens": 169884159.0,
      "step": 833
    },
    {
      "epoch": 1.9952153110047846,
      "grad_norm": 0.544509544292357,
      "learning_rate": 5.748429622807016e-06,
      "loss": 0.5541,
      "num_tokens": 170118387.0,
      "step": 834
    },
    {
      "epoch": 1.9976076555023923,
      "grad_norm": 0.5592361506310813,
      "learning_rate": 5.739721199263759e-06,
      "loss": 0.6119,
      "num_tokens": 170306703.0,
      "step": 835
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.6908089161991803,
      "learning_rate": 5.731011875311099e-06,
      "loss": 0.8147,
      "num_tokens": 170479057.0,
      "step": 836
    },
    {
      "epoch": 2.0,
      "eval_loss": 0.45707619190216064,
      "eval_num_tokens": 170479057.0,
      "eval_runtime": 109.3619,
      "eval_samples_per_second": 27.176,
      "eval_steps_per_second": 3.402,
      "step": 836
    },
    {
      "epoch": 2.0023923444976077,
      "grad_norm": 0.7010101792981096,
      "learning_rate": 5.722301683661859e-06,
      "loss": 0.5802,
      "num_tokens": 170662391.0,
      "step": 837
    },
    {
      "epoch": 2.0047846889952154,
      "grad_norm": 0.536459345119045,
      "learning_rate": 5.713590657032124e-06,
      "loss": 0.6956,
      "num_tokens": 170945306.0,
      "step": 838
    },
    {
      "epoch": 2.007177033492823,
      "grad_norm": 0.6865598520551047,
      "learning_rate": 5.704878828141108e-06,
      "loss": 0.4844,
      "num_tokens": 171123685.0,
      "step": 839
    },
    {
      "epoch": 2.0095693779904304,
      "grad_norm": 0.6556001284101048,
      "learning_rate": 5.696166229711045e-06,
      "loss": 0.4734,
      "num_tokens": 171366956.0,
      "step": 840
    },
    {
      "epoch": 2.011961722488038,
      "grad_norm": 0.7304372391236456,
      "learning_rate": 5.687452894467058e-06,
      "loss": 0.4759,
      "num_tokens": 171503225.0,
      "step": 841
    },
    {
      "epoch": 2.014354066985646,
      "grad_norm": 0.5893145169524098,
      "learning_rate": 5.678738855137037e-06,
      "loss": 0.5807,
      "num_tokens": 171715062.0,
      "step": 842
    },
    {
      "epoch": 2.0167464114832536,
      "grad_norm": 0.6891485855481355,
      "learning_rate": 5.6700241444515146e-06,
      "loss": 0.7269,
      "num_tokens": 171932247.0,
      "step": 843
    },
    {
      "epoch": 2.0191387559808613,
      "grad_norm": 0.7002346069882126,
      "learning_rate": 5.661308795143546e-06,
      "loss": 0.5333,
      "num_tokens": 172113046.0,
      "step": 844
    },
    {
      "epoch": 2.021531100478469,
      "grad_norm": 0.6197702947732776,
      "learning_rate": 5.6525928399485895e-06,
      "loss": 0.5762,
      "num_tokens": 172357016.0,
      "step": 845
    },
    {
      "epoch": 2.0239234449760763,
      "grad_norm": 0.6387029634051665,
      "learning_rate": 5.643876311604371e-06,
      "loss": 0.7473,
      "num_tokens": 172559033.0,
      "step": 846
    },
    {
      "epoch": 2.026315789473684,
      "grad_norm": 0.7098536589133576,
      "learning_rate": 5.635159242850779e-06,
      "loss": 0.5927,
      "num_tokens": 172732819.0,
      "step": 847
    },
    {
      "epoch": 2.028708133971292,
      "grad_norm": 0.678135657260688,
      "learning_rate": 5.626441666429724e-06,
      "loss": 0.5892,
      "num_tokens": 172932324.0,
      "step": 848
    },
    {
      "epoch": 2.0311004784688995,
      "grad_norm": 0.583893760377583,
      "learning_rate": 5.6177236150850246e-06,
      "loss": 0.5486,
      "num_tokens": 173186930.0,
      "step": 849
    },
    {
      "epoch": 2.0334928229665072,
      "grad_norm": 0.757431329558552,
      "learning_rate": 5.609005121562285e-06,
      "loss": 0.6904,
      "num_tokens": 173388905.0,
      "step": 850
    },
    {
      "epoch": 2.035885167464115,
      "grad_norm": 0.6637010404138924,
      "learning_rate": 5.600286218608773e-06,
      "loss": 0.6561,
      "num_tokens": 173594963.0,
      "step": 851
    },
    {
      "epoch": 2.0382775119617227,
      "grad_norm": 0.6278760783639207,
      "learning_rate": 5.591566938973287e-06,
      "loss": 0.5484,
      "num_tokens": 173764506.0,
      "step": 852
    },
    {
      "epoch": 2.04066985645933,
      "grad_norm": 0.7164186834391592,
      "learning_rate": 5.582847315406045e-06,
      "loss": 0.604,
      "num_tokens": 173941988.0,
      "step": 853
    },
    {
      "epoch": 2.0430622009569377,
      "grad_norm": 0.5957046275665503,
      "learning_rate": 5.5741273806585575e-06,
      "loss": 0.652,
      "num_tokens": 174185773.0,
      "step": 854
    },
    {
      "epoch": 2.0454545454545454,
      "grad_norm": 0.6429557814193617,
      "learning_rate": 5.565407167483501e-06,
      "loss": 0.6522,
      "num_tokens": 174410841.0,
      "step": 855
    },
    {
      "epoch": 2.047846889952153,
      "grad_norm": 0.6124880169897995,
      "learning_rate": 5.556686708634602e-06,
      "loss": 0.5845,
      "num_tokens": 174594347.0,
      "step": 856
    },
    {
      "epoch": 2.050239234449761,
      "grad_norm": 0.6247892510841981,
      "learning_rate": 5.547966036866503e-06,
      "loss": 0.499,
      "num_tokens": 174806968.0,
      "step": 857
    },
    {
      "epoch": 2.0526315789473686,
      "grad_norm": 0.5731204912599471,
      "learning_rate": 5.539245184934654e-06,
      "loss": 0.5916,
      "num_tokens": 175015537.0,
      "step": 858
    },
    {
      "epoch": 2.055023923444976,
      "grad_norm": 0.6406016610173918,
      "learning_rate": 5.530524185595178e-06,
      "loss": 0.7588,
      "num_tokens": 175285596.0,
      "step": 859
    },
    {
      "epoch": 2.0574162679425836,
      "grad_norm": 0.6939904180549464,
      "learning_rate": 5.521803071604747e-06,
      "loss": 0.4302,
      "num_tokens": 175407601.0,
      "step": 860
    },
    {
      "epoch": 2.0598086124401913,
      "grad_norm": 0.5840171644741489,
      "learning_rate": 5.513081875720472e-06,
      "loss": 0.5561,
      "num_tokens": 175617977.0,
      "step": 861
    },
    {
      "epoch": 2.062200956937799,
      "grad_norm": 0.5280383888463869,
      "learning_rate": 5.5043606306997685e-06,
      "loss": 0.4372,
      "num_tokens": 175862507.0,
      "step": 862
    },
    {
      "epoch": 2.0645933014354068,
      "grad_norm": 0.6746604030251611,
      "learning_rate": 5.495639369300233e-06,
      "loss": 0.5175,
      "num_tokens": 176069325.0,
      "step": 863
    },
    {
      "epoch": 2.0669856459330145,
      "grad_norm": 0.5327219079193565,
      "learning_rate": 5.4869181242795285e-06,
      "loss": 0.5431,
      "num_tokens": 176319089.0,
      "step": 864
    },
    {
      "epoch": 2.069377990430622,
      "grad_norm": 0.654611728144275,
      "learning_rate": 5.4781969283952554e-06,
      "loss": 0.5751,
      "num_tokens": 176544855.0,
      "step": 865
    },
    {
      "epoch": 2.0717703349282295,
      "grad_norm": 0.6697005720590413,
      "learning_rate": 5.469475814404825e-06,
      "loss": 0.5246,
      "num_tokens": 176743145.0,
      "step": 866
    },
    {
      "epoch": 2.074162679425837,
      "grad_norm": 0.6312841831223603,
      "learning_rate": 5.4607548150653464e-06,
      "loss": 0.4315,
      "num_tokens": 176942006.0,
      "step": 867
    },
    {
      "epoch": 2.076555023923445,
      "grad_norm": 0.5152891194242731,
      "learning_rate": 5.452033963133498e-06,
      "loss": 0.4911,
      "num_tokens": 177145857.0,
      "step": 868
    },
    {
      "epoch": 2.0789473684210527,
      "grad_norm": 0.5983945019880107,
      "learning_rate": 5.4433132913654e-06,
      "loss": 0.5751,
      "num_tokens": 177354826.0,
      "step": 869
    },
    {
      "epoch": 2.0813397129186604,
      "grad_norm": 0.5695605383121004,
      "learning_rate": 5.4345928325164985e-06,
      "loss": 0.6437,
      "num_tokens": 177568066.0,
      "step": 870
    },
    {
      "epoch": 2.083732057416268,
      "grad_norm": 0.577493281390095,
      "learning_rate": 5.425872619341445e-06,
      "loss": 0.6113,
      "num_tokens": 177785516.0,
      "step": 871
    },
    {
      "epoch": 2.0861244019138754,
      "grad_norm": 0.577110713494551,
      "learning_rate": 5.4171526845939584e-06,
      "loss": 0.5294,
      "num_tokens": 177989703.0,
      "step": 872
    },
    {
      "epoch": 2.088516746411483,
      "grad_norm": 0.5129404785846571,
      "learning_rate": 5.4084330610267156e-06,
      "loss": 0.6171,
      "num_tokens": 178259977.0,
      "step": 873
    },
    {
      "epoch": 2.090909090909091,
      "grad_norm": 0.6305465119747049,
      "learning_rate": 5.39971378139123e-06,
      "loss": 0.6896,
      "num_tokens": 178511083.0,
      "step": 874
    },
    {
      "epoch": 2.0933014354066986,
      "grad_norm": 0.5840406916165961,
      "learning_rate": 5.390994878437716e-06,
      "loss": 0.6089,
      "num_tokens": 178723973.0,
      "step": 875
    },
    {
      "epoch": 2.0956937799043063,
      "grad_norm": 0.5161608116725473,
      "learning_rate": 5.382276384914978e-06,
      "loss": 0.6667,
      "num_tokens": 178986671.0,
      "step": 876
    },
    {
      "epoch": 2.098086124401914,
      "grad_norm": 0.6030252688774056,
      "learning_rate": 5.373558333570278e-06,
      "loss": 0.5266,
      "num_tokens": 179184109.0,
      "step": 877
    },
    {
      "epoch": 2.1004784688995217,
      "grad_norm": 0.5056079079906192,
      "learning_rate": 5.364840757149222e-06,
      "loss": 0.5745,
      "num_tokens": 179477427.0,
      "step": 878
    },
    {
      "epoch": 2.102870813397129,
      "grad_norm": 0.5469146485955257,
      "learning_rate": 5.356123688395629e-06,
      "loss": 0.4426,
      "num_tokens": 179663693.0,
      "step": 879
    },
    {
      "epoch": 2.1052631578947367,
      "grad_norm": 0.5298879105875276,
      "learning_rate": 5.347407160051414e-06,
      "loss": 0.456,
      "num_tokens": 179868101.0,
      "step": 880
    },
    {
      "epoch": 2.1076555023923444,
      "grad_norm": 0.6377883062253229,
      "learning_rate": 5.338691204856455e-06,
      "loss": 0.5953,
      "num_tokens": 180104805.0,
      "step": 881
    },
    {
      "epoch": 2.110047846889952,
      "grad_norm": 0.632649188788054,
      "learning_rate": 5.3299758555484885e-06,
      "loss": 0.747,
      "num_tokens": 180330338.0,
      "step": 882
    },
    {
      "epoch": 2.11244019138756,
      "grad_norm": 0.6906228668304801,
      "learning_rate": 5.321261144862964e-06,
      "loss": 0.6373,
      "num_tokens": 180518149.0,
      "step": 883
    },
    {
      "epoch": 2.1148325358851676,
      "grad_norm": 0.5765219468106015,
      "learning_rate": 5.312547105532943e-06,
      "loss": 0.6644,
      "num_tokens": 180744284.0,
      "step": 884
    },
    {
      "epoch": 2.117224880382775,
      "grad_norm": 0.5228583600905762,
      "learning_rate": 5.303833770288955e-06,
      "loss": 0.5296,
      "num_tokens": 180984196.0,
      "step": 885
    },
    {
      "epoch": 2.1196172248803826,
      "grad_norm": 0.5378436640957062,
      "learning_rate": 5.295121171858893e-06,
      "loss": 0.3444,
      "num_tokens": 181190172.0,
      "step": 886
    },
    {
      "epoch": 2.1220095693779903,
      "grad_norm": 0.6509631304472094,
      "learning_rate": 5.286409342967878e-06,
      "loss": 0.7,
      "num_tokens": 181401483.0,
      "step": 887
    },
    {
      "epoch": 2.124401913875598,
      "grad_norm": 0.7172404422129441,
      "learning_rate": 5.277698316338142e-06,
      "loss": 0.6491,
      "num_tokens": 181608904.0,
      "step": 888
    },
    {
      "epoch": 2.126794258373206,
      "grad_norm": 0.7620587247682503,
      "learning_rate": 5.268988124688903e-06,
      "loss": 0.7506,
      "num_tokens": 181778389.0,
      "step": 889
    },
    {
      "epoch": 2.1291866028708135,
      "grad_norm": 0.6994356608870241,
      "learning_rate": 5.260278800736242e-06,
      "loss": 0.7422,
      "num_tokens": 181977883.0,
      "step": 890
    },
    {
      "epoch": 2.1315789473684212,
      "grad_norm": 0.6228038727199776,
      "learning_rate": 5.251570377192987e-06,
      "loss": 0.5658,
      "num_tokens": 182162343.0,
      "step": 891
    },
    {
      "epoch": 2.1339712918660285,
      "grad_norm": 0.7318448075698833,
      "learning_rate": 5.242862886768572e-06,
      "loss": 0.7979,
      "num_tokens": 182343690.0,
      "step": 892
    },
    {
      "epoch": 2.1363636363636362,
      "grad_norm": 0.6176672020032611,
      "learning_rate": 5.234156362168939e-06,
      "loss": 0.5408,
      "num_tokens": 182542228.0,
      "step": 893
    },
    {
      "epoch": 2.138755980861244,
      "grad_norm": 0.6830141376665197,
      "learning_rate": 5.225450836096392e-06,
      "loss": 0.7719,
      "num_tokens": 182755030.0,
      "step": 894
    },
    {
      "epoch": 2.1411483253588517,
      "grad_norm": 0.5870134491180863,
      "learning_rate": 5.216746341249493e-06,
      "loss": 0.4887,
      "num_tokens": 182947312.0,
      "step": 895
    },
    {
      "epoch": 2.1435406698564594,
      "grad_norm": 0.6325304186483628,
      "learning_rate": 5.208042910322924e-06,
      "loss": 0.5166,
      "num_tokens": 183110659.0,
      "step": 896
    },
    {
      "epoch": 2.145933014354067,
      "grad_norm": 0.6979115952850606,
      "learning_rate": 5.199340576007374e-06,
      "loss": 0.7542,
      "num_tokens": 183316493.0,
      "step": 897
    },
    {
      "epoch": 2.1483253588516744,
      "grad_norm": 0.7444990538919707,
      "learning_rate": 5.1906393709894135e-06,
      "loss": 0.5085,
      "num_tokens": 183475125.0,
      "step": 898
    },
    {
      "epoch": 2.150717703349282,
      "grad_norm": 0.6378799593192597,
      "learning_rate": 5.18193932795137e-06,
      "loss": 0.7381,
      "num_tokens": 183711701.0,
      "step": 899
    },
    {
      "epoch": 2.15311004784689,
      "grad_norm": 0.6736618720341914,
      "learning_rate": 5.173240479571206e-06,
      "loss": 0.5898,
      "num_tokens": 183881633.0,
      "step": 900
    },
    {
      "epoch": 2.1555023923444976,
      "grad_norm": 0.6220359607877941,
      "learning_rate": 5.164542858522398e-06,
      "loss": 0.6882,
      "num_tokens": 184115049.0,
      "step": 901
    },
    {
      "epoch": 2.1578947368421053,
      "grad_norm": 0.538351043125533,
      "learning_rate": 5.155846497473811e-06,
      "loss": 0.387,
      "num_tokens": 184287354.0,
      "step": 902
    },
    {
      "epoch": 2.160287081339713,
      "grad_norm": 0.5642599124571198,
      "learning_rate": 5.147151429089581e-06,
      "loss": 0.484,
      "num_tokens": 184508937.0,
      "step": 903
    },
    {
      "epoch": 2.1626794258373208,
      "grad_norm": 0.6001870881012696,
      "learning_rate": 5.138457686028985e-06,
      "loss": 0.5965,
      "num_tokens": 184748958.0,
      "step": 904
    },
    {
      "epoch": 2.165071770334928,
      "grad_norm": 0.6114874191151893,
      "learning_rate": 5.12976530094632e-06,
      "loss": 0.4673,
      "num_tokens": 184918082.0,
      "step": 905
    },
    {
      "epoch": 2.1674641148325358,
      "grad_norm": 0.5601252936577793,
      "learning_rate": 5.121074306490792e-06,
      "loss": 0.6114,
      "num_tokens": 185142287.0,
      "step": 906
    },
    {
      "epoch": 2.1698564593301435,
      "grad_norm": 0.6169147349809473,
      "learning_rate": 5.112384735306371e-06,
      "loss": 0.5461,
      "num_tokens": 185293698.0,
      "step": 907
    },
    {
      "epoch": 2.172248803827751,
      "grad_norm": 0.6195660313799543,
      "learning_rate": 5.103696620031692e-06,
      "loss": 0.6711,
      "num_tokens": 185499307.0,
      "step": 908
    },
    {
      "epoch": 2.174641148325359,
      "grad_norm": 0.6031245154149687,
      "learning_rate": 5.095009993299912e-06,
      "loss": 0.6294,
      "num_tokens": 185720997.0,
      "step": 909
    },
    {
      "epoch": 2.1770334928229667,
      "grad_norm": 0.45150169395352036,
      "learning_rate": 5.086324887738608e-06,
      "loss": 0.4616,
      "num_tokens": 185996237.0,
      "step": 910
    },
    {
      "epoch": 2.179425837320574,
      "grad_norm": 0.5937895068578424,
      "learning_rate": 5.07764133596963e-06,
      "loss": 0.6337,
      "num_tokens": 186217389.0,
      "step": 911
    },
    {
      "epoch": 2.1818181818181817,
      "grad_norm": 0.6066999652282524,
      "learning_rate": 5.068959370609005e-06,
      "loss": 0.5495,
      "num_tokens": 186427020.0,
      "step": 912
    },
    {
      "epoch": 2.1842105263157894,
      "grad_norm": 0.6371757752607746,
      "learning_rate": 5.060279024266791e-06,
      "loss": 0.5121,
      "num_tokens": 186621807.0,
      "step": 913
    },
    {
      "epoch": 2.186602870813397,
      "grad_norm": 0.5354988636020794,
      "learning_rate": 5.051600329546973e-06,
      "loss": 0.4376,
      "num_tokens": 186845555.0,
      "step": 914
    },
    {
      "epoch": 2.188995215311005,
      "grad_norm": 0.6984023668454643,
      "learning_rate": 5.042923319047325e-06,
      "loss": 0.6887,
      "num_tokens": 187042423.0,
      "step": 915
    },
    {
      "epoch": 2.1913875598086126,
      "grad_norm": 0.5999427476898559,
      "learning_rate": 5.034248025359302e-06,
      "loss": 0.5246,
      "num_tokens": 187225861.0,
      "step": 916
    },
    {
      "epoch": 2.1937799043062203,
      "grad_norm": 0.5612113652706427,
      "learning_rate": 5.025574481067904e-06,
      "loss": 0.5317,
      "num_tokens": 187424793.0,
      "step": 917
    },
    {
      "epoch": 2.1961722488038276,
      "grad_norm": 0.5751247397201221,
      "learning_rate": 5.016902718751563e-06,
      "loss": 0.5002,
      "num_tokens": 187616947.0,
      "step": 918
    },
    {
      "epoch": 2.1985645933014353,
      "grad_norm": 0.5933499014582654,
      "learning_rate": 5.008232770982021e-06,
      "loss": 0.6113,
      "num_tokens": 187859226.0,
      "step": 919
    },
    {
      "epoch": 2.200956937799043,
      "grad_norm": 0.6725228864556276,
      "learning_rate": 4.999564670324197e-06,
      "loss": 0.5584,
      "num_tokens": 188017748.0,
      "step": 920
    },
    {
      "epoch": 2.2033492822966507,
      "grad_norm": 0.6885249368514662,
      "learning_rate": 4.990898449336082e-06,
      "loss": 0.7148,
      "num_tokens": 188233000.0,
      "step": 921
    },
    {
      "epoch": 2.2057416267942584,
      "grad_norm": 0.7099706677726074,
      "learning_rate": 4.9822341405685955e-06,
      "loss": 0.7334,
      "num_tokens": 188417806.0,
      "step": 922
    },
    {
      "epoch": 2.208133971291866,
      "grad_norm": 0.5668827507723063,
      "learning_rate": 4.973571776565487e-06,
      "loss": 0.5281,
      "num_tokens": 188647820.0,
      "step": 923
    },
    {
      "epoch": 2.2105263157894735,
      "grad_norm": 0.5687915556773702,
      "learning_rate": 4.964911389863187e-06,
      "loss": 0.571,
      "num_tokens": 188857361.0,
      "step": 924
    },
    {
      "epoch": 2.212918660287081,
      "grad_norm": 0.5716920970399307,
      "learning_rate": 4.956253012990715e-06,
      "loss": 0.5454,
      "num_tokens": 189077250.0,
      "step": 925
    },
    {
      "epoch": 2.215311004784689,
      "grad_norm": 0.5520991021275099,
      "learning_rate": 4.947596678469527e-06,
      "loss": 0.6517,
      "num_tokens": 189332723.0,
      "step": 926
    },
    {
      "epoch": 2.2177033492822966,
      "grad_norm": 0.6837998974956843,
      "learning_rate": 4.938942418813417e-06,
      "loss": 0.6381,
      "num_tokens": 189525225.0,
      "step": 927
    },
    {
      "epoch": 2.2200956937799043,
      "grad_norm": 0.5976614086012115,
      "learning_rate": 4.930290266528383e-06,
      "loss": 0.5666,
      "num_tokens": 189744774.0,
      "step": 928
    },
    {
      "epoch": 2.222488038277512,
      "grad_norm": 0.6648157340151367,
      "learning_rate": 4.9216402541125064e-06,
      "loss": 0.5748,
      "num_tokens": 189898670.0,
      "step": 929
    },
    {
      "epoch": 2.22488038277512,
      "grad_norm": 0.5695867945010721,
      "learning_rate": 4.9129924140558305e-06,
      "loss": 0.5266,
      "num_tokens": 190078320.0,
      "step": 930
    },
    {
      "epoch": 2.227272727272727,
      "grad_norm": 0.640940270681438,
      "learning_rate": 4.904346778840243e-06,
      "loss": 0.722,
      "num_tokens": 190324800.0,
      "step": 931
    },
    {
      "epoch": 2.229665071770335,
      "grad_norm": 0.6115300277298661,
      "learning_rate": 4.895703380939348e-06,
      "loss": 0.6223,
      "num_tokens": 190526042.0,
      "step": 932
    },
    {
      "epoch": 2.2320574162679425,
      "grad_norm": 0.6078403150041939,
      "learning_rate": 4.887062252818343e-06,
      "loss": 0.6288,
      "num_tokens": 190763868.0,
      "step": 933
    },
    {
      "epoch": 2.2344497607655502,
      "grad_norm": 0.491734778374236,
      "learning_rate": 4.878423426933906e-06,
      "loss": 0.4818,
      "num_tokens": 190990352.0,
      "step": 934
    },
    {
      "epoch": 2.236842105263158,
      "grad_norm": 0.7432657101202276,
      "learning_rate": 4.869786935734061e-06,
      "loss": 0.567,
      "num_tokens": 191134359.0,
      "step": 935
    },
    {
      "epoch": 2.2392344497607657,
      "grad_norm": 0.6351576528189151,
      "learning_rate": 4.861152811658069e-06,
      "loss": 0.4386,
      "num_tokens": 191303415.0,
      "step": 936
    },
    {
      "epoch": 2.2416267942583734,
      "grad_norm": 0.7554013281643714,
      "learning_rate": 4.852521087136296e-06,
      "loss": 0.7855,
      "num_tokens": 191507996.0,
      "step": 937
    },
    {
      "epoch": 2.2440191387559807,
      "grad_norm": 0.601895734080878,
      "learning_rate": 4.843891794590097e-06,
      "loss": 0.5065,
      "num_tokens": 191719942.0,
      "step": 938
    },
    {
      "epoch": 2.2464114832535884,
      "grad_norm": 0.7135527988113347,
      "learning_rate": 4.835264966431691e-06,
      "loss": 0.5396,
      "num_tokens": 191853658.0,
      "step": 939
    },
    {
      "epoch": 2.248803827751196,
      "grad_norm": 0.631633953168551,
      "learning_rate": 4.8266406350640406e-06,
      "loss": 0.7803,
      "num_tokens": 192082886.0,
      "step": 940
    },
    {
      "epoch": 2.251196172248804,
      "grad_norm": 0.528159042893983,
      "learning_rate": 4.818018832880731e-06,
      "loss": 0.5049,
      "num_tokens": 192286179.0,
      "step": 941
    },
    {
      "epoch": 2.2535885167464116,
      "grad_norm": 0.5692998771536943,
      "learning_rate": 4.80939959226585e-06,
      "loss": 0.5496,
      "num_tokens": 192482338.0,
      "step": 942
    },
    {
      "epoch": 2.2559808612440193,
      "grad_norm": 0.5998091248077541,
      "learning_rate": 4.800782945593856e-06,
      "loss": 0.608,
      "num_tokens": 192680750.0,
      "step": 943
    },
    {
      "epoch": 2.258373205741627,
      "grad_norm": 0.7033907861591749,
      "learning_rate": 4.792168925229476e-06,
      "loss": 0.6006,
      "num_tokens": 192864346.0,
      "step": 944
    },
    {
      "epoch": 2.2607655502392343,
      "grad_norm": 0.7629806480330503,
      "learning_rate": 4.783557563527561e-06,
      "loss": 0.626,
      "num_tokens": 193033493.0,
      "step": 945
    },
    {
      "epoch": 2.263157894736842,
      "grad_norm": 0.5707192285167687,
      "learning_rate": 4.7749488928329855e-06,
      "loss": 0.6194,
      "num_tokens": 193294902.0,
      "step": 946
    },
    {
      "epoch": 2.2655502392344498,
      "grad_norm": 0.5905874972965276,
      "learning_rate": 4.766342945480508e-06,
      "loss": 0.5512,
      "num_tokens": 193490577.0,
      "step": 947
    },
    {
      "epoch": 2.2679425837320575,
      "grad_norm": 0.5177620597135223,
      "learning_rate": 4.757739753794664e-06,
      "loss": 0.6391,
      "num_tokens": 193791362.0,
      "step": 948
    },
    {
      "epoch": 2.270334928229665,
      "grad_norm": 0.5934058427993907,
      "learning_rate": 4.749139350089639e-06,
      "loss": 0.517,
      "num_tokens": 193983233.0,
      "step": 949
    },
    {
      "epoch": 2.2727272727272725,
      "grad_norm": 0.5978699434451642,
      "learning_rate": 4.740541766669136e-06,
      "loss": 0.5101,
      "num_tokens": 194192656.0,
      "step": 950
    },
    {
      "epoch": 2.27511961722488,
      "grad_norm": 0.6275089312864266,
      "learning_rate": 4.731947035826283e-06,
      "loss": 0.4238,
      "num_tokens": 194381428.0,
      "step": 951
    },
    {
      "epoch": 2.277511961722488,
      "grad_norm": 0.6170245058787546,
      "learning_rate": 4.723355189843476e-06,
      "loss": 0.5306,
      "num_tokens": 194561183.0,
      "step": 952
    },
    {
      "epoch": 2.2799043062200957,
      "grad_norm": 0.697796398431545,
      "learning_rate": 4.7147662609922866e-06,
      "loss": 0.5723,
      "num_tokens": 194774084.0,
      "step": 953
    },
    {
      "epoch": 2.2822966507177034,
      "grad_norm": 0.5708474589048195,
      "learning_rate": 4.706180281533322e-06,
      "loss": 0.5441,
      "num_tokens": 195022872.0,
      "step": 954
    },
    {
      "epoch": 2.284688995215311,
      "grad_norm": 0.6221089024679085,
      "learning_rate": 4.697597283716119e-06,
      "loss": 0.4427,
      "num_tokens": 195188063.0,
      "step": 955
    },
    {
      "epoch": 2.287081339712919,
      "grad_norm": 0.6966142472427688,
      "learning_rate": 4.689017299779007e-06,
      "loss": 0.6812,
      "num_tokens": 195357841.0,
      "step": 956
    },
    {
      "epoch": 2.2894736842105265,
      "grad_norm": 0.5858764677824216,
      "learning_rate": 4.6804403619490025e-06,
      "loss": 0.6256,
      "num_tokens": 195574109.0,
      "step": 957
    },
    {
      "epoch": 2.291866028708134,
      "grad_norm": 0.6720579646318956,
      "learning_rate": 4.671866502441674e-06,
      "loss": 0.6995,
      "num_tokens": 195763960.0,
      "step": 958
    },
    {
      "epoch": 2.2942583732057416,
      "grad_norm": 0.6739432961084355,
      "learning_rate": 4.663295753461035e-06,
      "loss": 0.7389,
      "num_tokens": 195972654.0,
      "step": 959
    },
    {
      "epoch": 2.2966507177033493,
      "grad_norm": 0.5337406968936691,
      "learning_rate": 4.654728147199405e-06,
      "loss": 0.5339,
      "num_tokens": 196221210.0,
      "step": 960
    },
    {
      "epoch": 2.299043062200957,
      "grad_norm": 0.7543750707624769,
      "learning_rate": 4.646163715837313e-06,
      "loss": 0.604,
      "num_tokens": 196356480.0,
      "step": 961
    },
    {
      "epoch": 2.3014354066985647,
      "grad_norm": 0.64221875053758,
      "learning_rate": 4.637602491543349e-06,
      "loss": 0.7221,
      "num_tokens": 196586279.0,
      "step": 962
    },
    {
      "epoch": 2.303827751196172,
      "grad_norm": 0.6051407759973162,
      "learning_rate": 4.629044506474069e-06,
      "loss": 0.4194,
      "num_tokens": 196759153.0,
      "step": 963
    },
    {
      "epoch": 2.3062200956937797,
      "grad_norm": 0.6673074288198911,
      "learning_rate": 4.620489792773854e-06,
      "loss": 0.5546,
      "num_tokens": 196928715.0,
      "step": 964
    },
    {
      "epoch": 2.3086124401913874,
      "grad_norm": 0.6667667122541079,
      "learning_rate": 4.611938382574799e-06,
      "loss": 0.6447,
      "num_tokens": 197132410.0,
      "step": 965
    },
    {
      "epoch": 2.311004784688995,
      "grad_norm": 0.7382401821510317,
      "learning_rate": 4.603390307996592e-06,
      "loss": 0.5502,
      "num_tokens": 197303154.0,
      "step": 966
    },
    {
      "epoch": 2.313397129186603,
      "grad_norm": 0.5355460347305746,
      "learning_rate": 4.5948456011463946e-06,
      "loss": 0.5119,
      "num_tokens": 197538989.0,
      "step": 967
    },
    {
      "epoch": 2.3157894736842106,
      "grad_norm": 0.588745542689267,
      "learning_rate": 4.586304294118718e-06,
      "loss": 0.5423,
      "num_tokens": 197766883.0,
      "step": 968
    },
    {
      "epoch": 2.3181818181818183,
      "grad_norm": 0.5842131197012101,
      "learning_rate": 4.577766418995296e-06,
      "loss": 0.603,
      "num_tokens": 197987061.0,
      "step": 969
    },
    {
      "epoch": 2.320574162679426,
      "grad_norm": 0.6979657515490693,
      "learning_rate": 4.569232007844985e-06,
      "loss": 0.5722,
      "num_tokens": 198185962.0,
      "step": 970
    },
    {
      "epoch": 2.3229665071770333,
      "grad_norm": 0.5616019882938909,
      "learning_rate": 4.560701092723617e-06,
      "loss": 0.7079,
      "num_tokens": 198419002.0,
      "step": 971
    },
    {
      "epoch": 2.325358851674641,
      "grad_norm": 0.5897872136033281,
      "learning_rate": 4.552173705673905e-06,
      "loss": 0.6447,
      "num_tokens": 198618480.0,
      "step": 972
    },
    {
      "epoch": 2.327751196172249,
      "grad_norm": 0.6208653472110455,
      "learning_rate": 4.5436498787253015e-06,
      "loss": 0.4777,
      "num_tokens": 198776227.0,
      "step": 973
    },
    {
      "epoch": 2.3301435406698565,
      "grad_norm": 0.4775412883065485,
      "learning_rate": 4.535129643893891e-06,
      "loss": 0.4726,
      "num_tokens": 199043473.0,
      "step": 974
    },
    {
      "epoch": 2.3325358851674642,
      "grad_norm": 0.507525873030941,
      "learning_rate": 4.526613033182261e-06,
      "loss": 0.5436,
      "num_tokens": 199306828.0,
      "step": 975
    },
    {
      "epoch": 2.334928229665072,
      "grad_norm": 0.6560516411422593,
      "learning_rate": 4.518100078579397e-06,
      "loss": 0.5843,
      "num_tokens": 199503531.0,
      "step": 976
    },
    {
      "epoch": 2.3373205741626792,
      "grad_norm": 0.6356188122044742,
      "learning_rate": 4.50959081206054e-06,
      "loss": 0.5826,
      "num_tokens": 199688436.0,
      "step": 977
    },
    {
      "epoch": 2.339712918660287,
      "grad_norm": 0.6420147098738003,
      "learning_rate": 4.501085265587086e-06,
      "loss": 0.5274,
      "num_tokens": 199876619.0,
      "step": 978
    },
    {
      "epoch": 2.3421052631578947,
      "grad_norm": 0.6400044423826036,
      "learning_rate": 4.492583471106454e-06,
      "loss": 0.5228,
      "num_tokens": 200078177.0,
      "step": 979
    },
    {
      "epoch": 2.3444976076555024,
      "grad_norm": 0.6337573455616772,
      "learning_rate": 4.484085460551974e-06,
      "loss": 0.4544,
      "num_tokens": 200277105.0,
      "step": 980
    },
    {
      "epoch": 2.34688995215311,
      "grad_norm": 0.5146634318287221,
      "learning_rate": 4.4755912658427605e-06,
      "loss": 0.6146,
      "num_tokens": 200556359.0,
      "step": 981
    },
    {
      "epoch": 2.349282296650718,
      "grad_norm": 0.557609526377004,
      "learning_rate": 4.467100918883594e-06,
      "loss": 0.5273,
      "num_tokens": 200789036.0,
      "step": 982
    },
    {
      "epoch": 2.3516746411483256,
      "grad_norm": 0.6103882750369611,
      "learning_rate": 4.458614451564809e-06,
      "loss": 0.5557,
      "num_tokens": 201016291.0,
      "step": 983
    },
    {
      "epoch": 2.354066985645933,
      "grad_norm": 0.5615569932242273,
      "learning_rate": 4.4501318957621595e-06,
      "loss": 0.531,
      "num_tokens": 201206656.0,
      "step": 984
    },
    {
      "epoch": 2.3564593301435406,
      "grad_norm": 0.5897719971401356,
      "learning_rate": 4.441653283336715e-06,
      "loss": 0.5324,
      "num_tokens": 201412902.0,
      "step": 985
    },
    {
      "epoch": 2.3588516746411483,
      "grad_norm": 0.6345523830064598,
      "learning_rate": 4.433178646134727e-06,
      "loss": 0.5531,
      "num_tokens": 201669487.0,
      "step": 986
    },
    {
      "epoch": 2.361244019138756,
      "grad_norm": 0.45138951192895616,
      "learning_rate": 4.424708015987524e-06,
      "loss": 0.5122,
      "num_tokens": 201917216.0,
      "step": 987
    },
    {
      "epoch": 2.3636363636363638,
      "grad_norm": 0.5682611787081613,
      "learning_rate": 4.416241424711372e-06,
      "loss": 0.4912,
      "num_tokens": 202090530.0,
      "step": 988
    },
    {
      "epoch": 2.3660287081339715,
      "grad_norm": 0.5261641148966557,
      "learning_rate": 4.4077789041073785e-06,
      "loss": 0.4621,
      "num_tokens": 202317413.0,
      "step": 989
    },
    {
      "epoch": 2.3684210526315788,
      "grad_norm": 0.5631165113167252,
      "learning_rate": 4.399320485961354e-06,
      "loss": 0.5311,
      "num_tokens": 202518648.0,
      "step": 990
    },
    {
      "epoch": 2.3708133971291865,
      "grad_norm": 0.4967545289756969,
      "learning_rate": 4.390866202043702e-06,
      "loss": 0.4564,
      "num_tokens": 202760261.0,
      "step": 991
    },
    {
      "epoch": 2.373205741626794,
      "grad_norm": 0.7465162438884122,
      "learning_rate": 4.382416084109298e-06,
      "loss": 0.5281,
      "num_tokens": 202889486.0,
      "step": 992
    },
    {
      "epoch": 2.375598086124402,
      "grad_norm": 0.5801443612563177,
      "learning_rate": 4.37397016389737e-06,
      "loss": 0.6409,
      "num_tokens": 203141411.0,
      "step": 993
    },
    {
      "epoch": 2.3779904306220097,
      "grad_norm": 0.5997099906097679,
      "learning_rate": 4.365528473131378e-06,
      "loss": 0.5756,
      "num_tokens": 203352106.0,
      "step": 994
    },
    {
      "epoch": 2.3803827751196174,
      "grad_norm": 0.6006398336798263,
      "learning_rate": 4.3570910435188915e-06,
      "loss": 0.4975,
      "num_tokens": 203544738.0,
      "step": 995
    },
    {
      "epoch": 2.382775119617225,
      "grad_norm": 0.616118763392102,
      "learning_rate": 4.348657906751488e-06,
      "loss": 0.6145,
      "num_tokens": 203738759.0,
      "step": 996
    },
    {
      "epoch": 2.3851674641148324,
      "grad_norm": 0.6640491937457687,
      "learning_rate": 4.340229094504607e-06,
      "loss": 0.5607,
      "num_tokens": 203898058.0,
      "step": 997
    },
    {
      "epoch": 2.38755980861244,
      "grad_norm": 0.6225754729256734,
      "learning_rate": 4.3318046384374526e-06,
      "loss": 0.554,
      "num_tokens": 204115504.0,
      "step": 998
    },
    {
      "epoch": 2.389952153110048,
      "grad_norm": 0.6377150799357686,
      "learning_rate": 4.3233845701928615e-06,
      "loss": 0.5783,
      "num_tokens": 204332997.0,
      "step": 999
    },
    {
      "epoch": 2.3923444976076556,
      "grad_norm": 0.5853455958148285,
      "learning_rate": 4.314968921397197e-06,
      "loss": 0.4981,
      "num_tokens": 204535990.0,
      "step": 1000
    },
    {
      "epoch": 2.3947368421052633,
      "grad_norm": 0.6248730722883572,
      "learning_rate": 4.306557723660212e-06,
      "loss": 0.5558,
      "num_tokens": 204744939.0,
      "step": 1001
    },
    {
      "epoch": 2.397129186602871,
      "grad_norm": 0.5151699139022012,
      "learning_rate": 4.298151008574952e-06,
      "loss": 0.3998,
      "num_tokens": 204938198.0,
      "step": 1002
    },
    {
      "epoch": 2.3995215311004783,
      "grad_norm": 0.6334922426435169,
      "learning_rate": 4.289748807717618e-06,
      "loss": 0.6982,
      "num_tokens": 205170249.0,
      "step": 1003
    },
    {
      "epoch": 2.401913875598086,
      "grad_norm": 0.7135525185350413,
      "learning_rate": 4.2813511526474585e-06,
      "loss": 0.5034,
      "num_tokens": 205304624.0,
      "step": 1004
    },
    {
      "epoch": 2.4043062200956937,
      "grad_norm": 0.5902081724362165,
      "learning_rate": 4.2729580749066435e-06,
      "loss": 0.6034,
      "num_tokens": 205516948.0,
      "step": 1005
    },
    {
      "epoch": 2.4066985645933014,
      "grad_norm": 0.6436090687074414,
      "learning_rate": 4.264569606020159e-06,
      "loss": 0.6051,
      "num_tokens": 205713236.0,
      "step": 1006
    },
    {
      "epoch": 2.409090909090909,
      "grad_norm": 0.5528964704740937,
      "learning_rate": 4.256185777495668e-06,
      "loss": 0.457,
      "num_tokens": 205918094.0,
      "step": 1007
    },
    {
      "epoch": 2.411483253588517,
      "grad_norm": 0.689153042053297,
      "learning_rate": 4.247806620823416e-06,
      "loss": 0.6527,
      "num_tokens": 206126263.0,
      "step": 1008
    },
    {
      "epoch": 2.4138755980861246,
      "grad_norm": 0.6370805657969235,
      "learning_rate": 4.239432167476091e-06,
      "loss": 0.7036,
      "num_tokens": 206339096.0,
      "step": 1009
    },
    {
      "epoch": 2.416267942583732,
      "grad_norm": 0.6004855383256495,
      "learning_rate": 4.231062448908721e-06,
      "loss": 0.587,
      "num_tokens": 206552352.0,
      "step": 1010
    },
    {
      "epoch": 2.4186602870813396,
      "grad_norm": 0.6173172520580517,
      "learning_rate": 4.222697496558547e-06,
      "loss": 0.6095,
      "num_tokens": 206777192.0,
      "step": 1011
    },
    {
      "epoch": 2.4210526315789473,
      "grad_norm": 0.6835207209010378,
      "learning_rate": 4.214337341844906e-06,
      "loss": 0.5918,
      "num_tokens": 206978187.0,
      "step": 1012
    },
    {
      "epoch": 2.423444976076555,
      "grad_norm": 0.6131323747610118,
      "learning_rate": 4.205982016169122e-06,
      "loss": 0.6657,
      "num_tokens": 207195050.0,
      "step": 1013
    },
    {
      "epoch": 2.425837320574163,
      "grad_norm": 0.5850555875038469,
      "learning_rate": 4.197631550914372e-06,
      "loss": 0.57,
      "num_tokens": 207406444.0,
      "step": 1014
    },
    {
      "epoch": 2.4282296650717705,
      "grad_norm": 0.4939100102339527,
      "learning_rate": 4.189285977445584e-06,
      "loss": 0.3718,
      "num_tokens": 207620232.0,
      "step": 1015
    },
    {
      "epoch": 2.430622009569378,
      "grad_norm": 0.653808161167263,
      "learning_rate": 4.180945327109308e-06,
      "loss": 0.5829,
      "num_tokens": 207826050.0,
      "step": 1016
    },
    {
      "epoch": 2.4330143540669855,
      "grad_norm": 0.43981969392432796,
      "learning_rate": 4.172609631233603e-06,
      "loss": 0.3208,
      "num_tokens": 208017777.0,
      "step": 1017
    },
    {
      "epoch": 2.4354066985645932,
      "grad_norm": 0.6261020734046672,
      "learning_rate": 4.164278921127919e-06,
      "loss": 0.5125,
      "num_tokens": 208207732.0,
      "step": 1018
    },
    {
      "epoch": 2.437799043062201,
      "grad_norm": 0.6317188642592404,
      "learning_rate": 4.15595322808298e-06,
      "loss": 0.4381,
      "num_tokens": 208386520.0,
      "step": 1019
    },
    {
      "epoch": 2.4401913875598087,
      "grad_norm": 0.6202931009885635,
      "learning_rate": 4.147632583370664e-06,
      "loss": 0.6438,
      "num_tokens": 208641359.0,
      "step": 1020
    },
    {
      "epoch": 2.4425837320574164,
      "grad_norm": 0.6022603778669238,
      "learning_rate": 4.1393170182438905e-06,
      "loss": 0.5467,
      "num_tokens": 208841020.0,
      "step": 1021
    },
    {
      "epoch": 2.444976076555024,
      "grad_norm": 0.6327040046052304,
      "learning_rate": 4.131006563936492e-06,
      "loss": 0.708,
      "num_tokens": 209071062.0,
      "step": 1022
    },
    {
      "epoch": 2.4473684210526314,
      "grad_norm": 0.5266051069162052,
      "learning_rate": 4.122701251663116e-06,
      "loss": 0.5639,
      "num_tokens": 209307718.0,
      "step": 1023
    },
    {
      "epoch": 2.449760765550239,
      "grad_norm": 0.5136191658605107,
      "learning_rate": 4.114401112619084e-06,
      "loss": 0.5557,
      "num_tokens": 209584037.0,
      "step": 1024
    },
    {
      "epoch": 2.452153110047847,
      "grad_norm": 0.6058510179396281,
      "learning_rate": 4.106106177980297e-06,
      "loss": 0.6519,
      "num_tokens": 209800026.0,
      "step": 1025
    },
    {
      "epoch": 2.4545454545454546,
      "grad_norm": 0.7748377168030689,
      "learning_rate": 4.0978164789031e-06,
      "loss": 0.6554,
      "num_tokens": 209945028.0,
      "step": 1026
    },
    {
      "epoch": 2.4569377990430623,
      "grad_norm": 0.539487028193741,
      "learning_rate": 4.089532046524176e-06,
      "loss": 0.7053,
      "num_tokens": 210227001.0,
      "step": 1027
    },
    {
      "epoch": 2.45933014354067,
      "grad_norm": 0.5917748409979356,
      "learning_rate": 4.081252911960428e-06,
      "loss": 0.5759,
      "num_tokens": 210417942.0,
      "step": 1028
    },
    {
      "epoch": 2.4617224880382773,
      "grad_norm": 0.5992588126824665,
      "learning_rate": 4.072979106308858e-06,
      "loss": 0.6159,
      "num_tokens": 210628053.0,
      "step": 1029
    },
    {
      "epoch": 2.464114832535885,
      "grad_norm": 0.6651607322996815,
      "learning_rate": 4.064710660646448e-06,
      "loss": 0.6818,
      "num_tokens": 210829550.0,
      "step": 1030
    },
    {
      "epoch": 2.4665071770334928,
      "grad_norm": 0.6540903157509698,
      "learning_rate": 4.056447606030056e-06,
      "loss": 0.61,
      "num_tokens": 211056840.0,
      "step": 1031
    },
    {
      "epoch": 2.4688995215311005,
      "grad_norm": 0.7991178220481605,
      "learning_rate": 4.048189973496287e-06,
      "loss": 0.6707,
      "num_tokens": 211211372.0,
      "step": 1032
    },
    {
      "epoch": 2.471291866028708,
      "grad_norm": 0.5935027675340312,
      "learning_rate": 4.039937794061376e-06,
      "loss": 0.621,
      "num_tokens": 211414390.0,
      "step": 1033
    },
    {
      "epoch": 2.473684210526316,
      "grad_norm": 0.6755369914687083,
      "learning_rate": 4.031691098721083e-06,
      "loss": 0.6413,
      "num_tokens": 211588059.0,
      "step": 1034
    },
    {
      "epoch": 2.4760765550239237,
      "grad_norm": 0.6526118866886425,
      "learning_rate": 4.023449918450565e-06,
      "loss": 0.6944,
      "num_tokens": 211807518.0,
      "step": 1035
    },
    {
      "epoch": 2.478468899521531,
      "grad_norm": 0.564198257094253,
      "learning_rate": 4.015214284204265e-06,
      "loss": 0.6394,
      "num_tokens": 212025980.0,
      "step": 1036
    },
    {
      "epoch": 2.4808612440191387,
      "grad_norm": 0.627883039676126,
      "learning_rate": 4.0069842269157935e-06,
      "loss": 0.5591,
      "num_tokens": 212199416.0,
      "step": 1037
    },
    {
      "epoch": 2.4832535885167464,
      "grad_norm": 0.6055641095041324,
      "learning_rate": 3.998759777497816e-06,
      "loss": 0.5301,
      "num_tokens": 212379337.0,
      "step": 1038
    },
    {
      "epoch": 2.485645933014354,
      "grad_norm": 0.7117718176806141,
      "learning_rate": 3.9905409668419346e-06,
      "loss": 0.6314,
      "num_tokens": 212556989.0,
      "step": 1039
    },
    {
      "epoch": 2.488038277511962,
      "grad_norm": 0.7179326451594609,
      "learning_rate": 3.982327825818568e-06,
      "loss": 0.6056,
      "num_tokens": 212743946.0,
      "step": 1040
    },
    {
      "epoch": 2.4904306220095696,
      "grad_norm": 0.6535767607430862,
      "learning_rate": 3.974120385276844e-06,
      "loss": 0.7189,
      "num_tokens": 212935011.0,
      "step": 1041
    },
    {
      "epoch": 2.492822966507177,
      "grad_norm": 0.6050383476235132,
      "learning_rate": 3.965918676044476e-06,
      "loss": 0.5952,
      "num_tokens": 213123277.0,
      "step": 1042
    },
    {
      "epoch": 2.4952153110047846,
      "grad_norm": 0.6607651606385861,
      "learning_rate": 3.95772272892765e-06,
      "loss": 0.6589,
      "num_tokens": 213334844.0,
      "step": 1043
    },
    {
      "epoch": 2.4976076555023923,
      "grad_norm": 0.6241385293512635,
      "learning_rate": 3.94953257471091e-06,
      "loss": 0.654,
      "num_tokens": 213535142.0,
      "step": 1044
    },
    {
      "epoch": 2.5,
      "grad_norm": 0.6213474009144981,
      "learning_rate": 3.941348244157045e-06,
      "loss": 0.6657,
      "num_tokens": 213749790.0,
      "step": 1045
    },
    {
      "epoch": 2.5023923444976077,
      "grad_norm": 0.5820164738008969,
      "learning_rate": 3.933169768006962e-06,
      "loss": 0.6203,
      "num_tokens": 213993242.0,
      "step": 1046
    },
    {
      "epoch": 2.5047846889952154,
      "grad_norm": 0.6881833790841918,
      "learning_rate": 3.924997176979589e-06,
      "loss": 0.7844,
      "num_tokens": 214184984.0,
      "step": 1047
    },
    {
      "epoch": 2.507177033492823,
      "grad_norm": 0.6146739381057159,
      "learning_rate": 3.916830501771736e-06,
      "loss": 0.4875,
      "num_tokens": 214360349.0,
      "step": 1048
    },
    {
      "epoch": 2.509569377990431,
      "grad_norm": 0.6025043525167425,
      "learning_rate": 3.908669773058005e-06,
      "loss": 0.6133,
      "num_tokens": 214575876.0,
      "step": 1049
    },
    {
      "epoch": 2.511961722488038,
      "grad_norm": 0.693733393553052,
      "learning_rate": 3.900515021490654e-06,
      "loss": 0.7332,
      "num_tokens": 214788140.0,
      "step": 1050
    },
    {
      "epoch": 2.514354066985646,
      "grad_norm": 0.5594177626974491,
      "learning_rate": 3.8923662776994965e-06,
      "loss": 0.433,
      "num_tokens": 214982831.0,
      "step": 1051
    },
    {
      "epoch": 2.5167464114832536,
      "grad_norm": 0.5817776587828116,
      "learning_rate": 3.884223572291778e-06,
      "loss": 0.5643,
      "num_tokens": 215190443.0,
      "step": 1052
    },
    {
      "epoch": 2.5191387559808613,
      "grad_norm": 0.6656717449668278,
      "learning_rate": 3.876086935852059e-06,
      "loss": 0.687,
      "num_tokens": 215387454.0,
      "step": 1053
    },
    {
      "epoch": 2.5215311004784686,
      "grad_norm": 0.46271635043906406,
      "learning_rate": 3.867956398942113e-06,
      "loss": 0.3984,
      "num_tokens": 215598001.0,
      "step": 1054
    },
    {
      "epoch": 2.5239234449760763,
      "grad_norm": 0.6034680377399717,
      "learning_rate": 3.859831992100798e-06,
      "loss": 0.7469,
      "num_tokens": 215814084.0,
      "step": 1055
    },
    {
      "epoch": 2.526315789473684,
      "grad_norm": 0.5789325890650581,
      "learning_rate": 3.8517137458439465e-06,
      "loss": 0.5048,
      "num_tokens": 215979272.0,
      "step": 1056
    },
    {
      "epoch": 2.528708133971292,
      "grad_norm": 0.4948180098998025,
      "learning_rate": 3.843601690664253e-06,
      "loss": 0.4004,
      "num_tokens": 216194524.0,
      "step": 1057
    },
    {
      "epoch": 2.5311004784688995,
      "grad_norm": 0.5754107413539873,
      "learning_rate": 3.83549585703116e-06,
      "loss": 0.4014,
      "num_tokens": 216365812.0,
      "step": 1058
    },
    {
      "epoch": 2.5334928229665072,
      "grad_norm": 0.6750744543039905,
      "learning_rate": 3.827396275390736e-06,
      "loss": 0.6902,
      "num_tokens": 216545039.0,
      "step": 1059
    },
    {
      "epoch": 2.535885167464115,
      "grad_norm": 0.6577431658509755,
      "learning_rate": 3.819302976165573e-06,
      "loss": 0.664,
      "num_tokens": 216746551.0,
      "step": 1060
    },
    {
      "epoch": 2.5382775119617227,
      "grad_norm": 0.627785069016739,
      "learning_rate": 3.8112159897546597e-06,
      "loss": 0.6174,
      "num_tokens": 216955697.0,
      "step": 1061
    },
    {
      "epoch": 2.5406698564593304,
      "grad_norm": 0.5995174609182009,
      "learning_rate": 3.8031353465332786e-06,
      "loss": 0.7305,
      "num_tokens": 217192047.0,
      "step": 1062
    },
    {
      "epoch": 2.5430622009569377,
      "grad_norm": 0.5695080880257041,
      "learning_rate": 3.795061076852882e-06,
      "loss": 0.6676,
      "num_tokens": 217418481.0,
      "step": 1063
    },
    {
      "epoch": 2.5454545454545454,
      "grad_norm": 0.6765456404371507,
      "learning_rate": 3.786993211040988e-06,
      "loss": 0.5479,
      "num_tokens": 217582242.0,
      "step": 1064
    },
    {
      "epoch": 2.547846889952153,
      "grad_norm": 0.5558441353632505,
      "learning_rate": 3.7789317794010562e-06,
      "loss": 0.4938,
      "num_tokens": 217821253.0,
      "step": 1065
    },
    {
      "epoch": 2.550239234449761,
      "grad_norm": 0.5652778076960531,
      "learning_rate": 3.770876812212384e-06,
      "loss": 0.4471,
      "num_tokens": 218028715.0,
      "step": 1066
    },
    {
      "epoch": 2.5526315789473686,
      "grad_norm": 0.7041275199284726,
      "learning_rate": 3.7628283397299825e-06,
      "loss": 0.4983,
      "num_tokens": 218188513.0,
      "step": 1067
    },
    {
      "epoch": 2.555023923444976,
      "grad_norm": 0.6547316059280318,
      "learning_rate": 3.7547863921844717e-06,
      "loss": 0.4126,
      "num_tokens": 218376107.0,
      "step": 1068
    },
    {
      "epoch": 2.5574162679425836,
      "grad_norm": 0.6336547930033282,
      "learning_rate": 3.7467509997819627e-06,
      "loss": 0.6674,
      "num_tokens": 218570758.0,
      "step": 1069
    },
    {
      "epoch": 2.5598086124401913,
      "grad_norm": 0.5478760136946561,
      "learning_rate": 3.738722192703946e-06,
      "loss": 0.4566,
      "num_tokens": 218786920.0,
      "step": 1070
    },
    {
      "epoch": 2.562200956937799,
      "grad_norm": 0.7139683911439062,
      "learning_rate": 3.730700001107177e-06,
      "loss": 0.7549,
      "num_tokens": 218985901.0,
      "step": 1071
    },
    {
      "epoch": 2.5645933014354068,
      "grad_norm": 0.6975615168028549,
      "learning_rate": 3.7226844551235575e-06,
      "loss": 0.6629,
      "num_tokens": 219159140.0,
      "step": 1072
    },
    {
      "epoch": 2.5669856459330145,
      "grad_norm": 0.6668143922240283,
      "learning_rate": 3.714675584860038e-06,
      "loss": 0.6375,
      "num_tokens": 219371386.0,
      "step": 1073
    },
    {
      "epoch": 2.569377990430622,
      "grad_norm": 0.5982014271194838,
      "learning_rate": 3.7066734203984855e-06,
      "loss": 0.3727,
      "num_tokens": 219552579.0,
      "step": 1074
    },
    {
      "epoch": 2.57177033492823,
      "grad_norm": 0.6955525411313587,
      "learning_rate": 3.698677991795585e-06,
      "loss": 0.7192,
      "num_tokens": 219775671.0,
      "step": 1075
    },
    {
      "epoch": 2.574162679425837,
      "grad_norm": 0.5369077452815465,
      "learning_rate": 3.6906893290827185e-06,
      "loss": 0.4529,
      "num_tokens": 219965861.0,
      "step": 1076
    },
    {
      "epoch": 2.576555023923445,
      "grad_norm": 0.6721059340974544,
      "learning_rate": 3.6827074622658566e-06,
      "loss": 0.5365,
      "num_tokens": 220165003.0,
      "step": 1077
    },
    {
      "epoch": 2.5789473684210527,
      "grad_norm": 0.7135684792977236,
      "learning_rate": 3.6747324213254397e-06,
      "loss": 0.5682,
      "num_tokens": 220392321.0,
      "step": 1078
    },
    {
      "epoch": 2.5813397129186604,
      "grad_norm": 0.6494117986406011,
      "learning_rate": 3.6667642362162793e-06,
      "loss": 0.5806,
      "num_tokens": 220598874.0,
      "step": 1079
    },
    {
      "epoch": 2.583732057416268,
      "grad_norm": 0.6222063527010798,
      "learning_rate": 3.6588029368674237e-06,
      "loss": 0.4917,
      "num_tokens": 220776263.0,
      "step": 1080
    },
    {
      "epoch": 2.5861244019138754,
      "grad_norm": 0.5934688030422492,
      "learning_rate": 3.650848553182067e-06,
      "loss": 0.5072,
      "num_tokens": 220960288.0,
      "step": 1081
    },
    {
      "epoch": 2.588516746411483,
      "grad_norm": 0.5684846998173526,
      "learning_rate": 3.6429011150374217e-06,
      "loss": 0.4913,
      "num_tokens": 221184618.0,
      "step": 1082
    },
    {
      "epoch": 2.590909090909091,
      "grad_norm": 0.549834773214977,
      "learning_rate": 3.634960652284619e-06,
      "loss": 0.3509,
      "num_tokens": 221386351.0,
      "step": 1083
    },
    {
      "epoch": 2.5933014354066986,
      "grad_norm": 0.6483735196222553,
      "learning_rate": 3.6270271947485835e-06,
      "loss": 0.5319,
      "num_tokens": 221579057.0,
      "step": 1084
    },
    {
      "epoch": 2.5956937799043063,
      "grad_norm": 0.5153676973173746,
      "learning_rate": 3.6191007722279315e-06,
      "loss": 0.4309,
      "num_tokens": 221781696.0,
      "step": 1085
    },
    {
      "epoch": 2.598086124401914,
      "grad_norm": 0.5489517271767583,
      "learning_rate": 3.611181414494852e-06,
      "loss": 0.4592,
      "num_tokens": 222005363.0,
      "step": 1086
    },
    {
      "epoch": 2.6004784688995217,
      "grad_norm": 0.49214928466990493,
      "learning_rate": 3.6032691512950024e-06,
      "loss": 0.4458,
      "num_tokens": 222259186.0,
      "step": 1087
    },
    {
      "epoch": 2.6028708133971294,
      "grad_norm": 0.6196094337292347,
      "learning_rate": 3.595364012347391e-06,
      "loss": 0.663,
      "num_tokens": 222513076.0,
      "step": 1088
    },
    {
      "epoch": 2.6052631578947367,
      "grad_norm": 0.5427019877705024,
      "learning_rate": 3.587466027344264e-06,
      "loss": 0.5428,
      "num_tokens": 222750527.0,
      "step": 1089
    },
    {
      "epoch": 2.6076555023923444,
      "grad_norm": 0.5771856967931067,
      "learning_rate": 3.579575225951003e-06,
      "loss": 0.5553,
      "num_tokens": 223006934.0,
      "step": 1090
    },
    {
      "epoch": 2.610047846889952,
      "grad_norm": 0.6034575663203413,
      "learning_rate": 3.5716916378059973e-06,
      "loss": 0.7012,
      "num_tokens": 223224028.0,
      "step": 1091
    },
    {
      "epoch": 2.61244019138756,
      "grad_norm": 0.6660365069148427,
      "learning_rate": 3.5638152925205573e-06,
      "loss": 0.7283,
      "num_tokens": 223426112.0,
      "step": 1092
    },
    {
      "epoch": 2.6148325358851676,
      "grad_norm": 0.643954882493234,
      "learning_rate": 3.555946219678772e-06,
      "loss": 0.527,
      "num_tokens": 223597677.0,
      "step": 1093
    },
    {
      "epoch": 2.617224880382775,
      "grad_norm": 0.5294855404946633,
      "learning_rate": 3.5480844488374313e-06,
      "loss": 0.3729,
      "num_tokens": 223791816.0,
      "step": 1094
    },
    {
      "epoch": 2.6196172248803826,
      "grad_norm": 0.6803208173614663,
      "learning_rate": 3.5402300095258833e-06,
      "loss": 0.6289,
      "num_tokens": 224011960.0,
      "step": 1095
    },
    {
      "epoch": 2.6220095693779903,
      "grad_norm": 0.5981932778163507,
      "learning_rate": 3.5323829312459524e-06,
      "loss": 0.5048,
      "num_tokens": 224234695.0,
      "step": 1096
    },
    {
      "epoch": 2.624401913875598,
      "grad_norm": 0.6101452118273694,
      "learning_rate": 3.524543243471801e-06,
      "loss": 0.4449,
      "num_tokens": 224410706.0,
      "step": 1097
    },
    {
      "epoch": 2.626794258373206,
      "grad_norm": 0.5389924597744387,
      "learning_rate": 3.516710975649841e-06,
      "loss": 0.5182,
      "num_tokens": 224593832.0,
      "step": 1098
    },
    {
      "epoch": 2.6291866028708135,
      "grad_norm": 0.6975855703487497,
      "learning_rate": 3.508886157198611e-06,
      "loss": 0.6577,
      "num_tokens": 224798557.0,
      "step": 1099
    },
    {
      "epoch": 2.6315789473684212,
      "grad_norm": 0.6351790408256973,
      "learning_rate": 3.5010688175086718e-06,
      "loss": 0.7309,
      "num_tokens": 225033322.0,
      "step": 1100
    },
    {
      "epoch": 2.633971291866029,
      "grad_norm": 0.6330427861367581,
      "learning_rate": 3.4932589859424913e-06,
      "loss": 0.7808,
      "num_tokens": 225255549.0,
      "step": 1101
    },
    {
      "epoch": 2.6363636363636362,
      "grad_norm": 0.6801017341307927,
      "learning_rate": 3.485456691834337e-06,
      "loss": 0.6036,
      "num_tokens": 225432118.0,
      "step": 1102
    },
    {
      "epoch": 2.638755980861244,
      "grad_norm": 0.6658896469366345,
      "learning_rate": 3.4776619644901655e-06,
      "loss": 0.6647,
      "num_tokens": 225616023.0,
      "step": 1103
    },
    {
      "epoch": 2.6411483253588517,
      "grad_norm": 0.4996367213764907,
      "learning_rate": 3.469874833187508e-06,
      "loss": 0.5414,
      "num_tokens": 225893120.0,
      "step": 1104
    },
    {
      "epoch": 2.6435406698564594,
      "grad_norm": 0.6616958516454566,
      "learning_rate": 3.4620953271753756e-06,
      "loss": 0.6326,
      "num_tokens": 226088868.0,
      "step": 1105
    },
    {
      "epoch": 2.645933014354067,
      "grad_norm": 0.6362043584410174,
      "learning_rate": 3.454323475674123e-06,
      "loss": 0.671,
      "num_tokens": 226310547.0,
      "step": 1106
    },
    {
      "epoch": 2.6483253588516744,
      "grad_norm": 0.47501498356513666,
      "learning_rate": 3.4465593078753688e-06,
      "loss": 0.4813,
      "num_tokens": 226550492.0,
      "step": 1107
    },
    {
      "epoch": 2.650717703349282,
      "grad_norm": 0.5031158510563932,
      "learning_rate": 3.438802852941858e-06,
      "loss": 0.5583,
      "num_tokens": 226832459.0,
      "step": 1108
    },
    {
      "epoch": 2.65311004784689,
      "grad_norm": 0.5435062765868961,
      "learning_rate": 3.431054140007377e-06,
      "loss": 0.5455,
      "num_tokens": 227038778.0,
      "step": 1109
    },
    {
      "epoch": 2.6555023923444976,
      "grad_norm": 0.5388254458858796,
      "learning_rate": 3.4233131981766243e-06,
      "loss": 0.5894,
      "num_tokens": 227249531.0,
      "step": 1110
    },
    {
      "epoch": 2.6578947368421053,
      "grad_norm": 0.5922470838904194,
      "learning_rate": 3.4155800565251135e-06,
      "loss": 0.4353,
      "num_tokens": 227462703.0,
      "step": 1111
    },
    {
      "epoch": 2.660287081339713,
      "grad_norm": 0.6846109348473851,
      "learning_rate": 3.40785474409906e-06,
      "loss": 0.5729,
      "num_tokens": 227635078.0,
      "step": 1112
    },
    {
      "epoch": 2.6626794258373208,
      "grad_norm": 0.4764323015116,
      "learning_rate": 3.4001372899152703e-06,
      "loss": 0.4876,
      "num_tokens": 227906297.0,
      "step": 1113
    },
    {
      "epoch": 2.6650717703349285,
      "grad_norm": 0.5853206610805425,
      "learning_rate": 3.392427722961037e-06,
      "loss": 0.5074,
      "num_tokens": 228140226.0,
      "step": 1114
    },
    {
      "epoch": 2.6674641148325358,
      "grad_norm": 0.6793627537042475,
      "learning_rate": 3.384726072194026e-06,
      "loss": 0.7223,
      "num_tokens": 228357361.0,
      "step": 1115
    },
    {
      "epoch": 2.6698564593301435,
      "grad_norm": 0.5463734944701496,
      "learning_rate": 3.3770323665421713e-06,
      "loss": 0.6404,
      "num_tokens": 228594109.0,
      "step": 1116
    },
    {
      "epoch": 2.672248803827751,
      "grad_norm": 0.6738633775521184,
      "learning_rate": 3.3693466349035627e-06,
      "loss": 0.5112,
      "num_tokens": 228755387.0,
      "step": 1117
    },
    {
      "epoch": 2.674641148325359,
      "grad_norm": 0.6531030945413439,
      "learning_rate": 3.361668906146341e-06,
      "loss": 0.6901,
      "num_tokens": 228956043.0,
      "step": 1118
    },
    {
      "epoch": 2.6770334928229667,
      "grad_norm": 0.6923041244581303,
      "learning_rate": 3.353999209108583e-06,
      "loss": 0.5057,
      "num_tokens": 229125917.0,
      "step": 1119
    },
    {
      "epoch": 2.679425837320574,
      "grad_norm": 0.6775907584116359,
      "learning_rate": 3.3463375725982056e-06,
      "loss": 0.5541,
      "num_tokens": 229281789.0,
      "step": 1120
    },
    {
      "epoch": 2.6818181818181817,
      "grad_norm": 0.6173853747715191,
      "learning_rate": 3.3386840253928398e-06,
      "loss": 0.4948,
      "num_tokens": 229477784.0,
      "step": 1121
    },
    {
      "epoch": 2.6842105263157894,
      "grad_norm": 0.6725183241705002,
      "learning_rate": 3.3310385962397453e-06,
      "loss": 0.7163,
      "num_tokens": 229673326.0,
      "step": 1122
    },
    {
      "epoch": 2.686602870813397,
      "grad_norm": 0.6499836969575646,
      "learning_rate": 3.3234013138556746e-06,
      "loss": 0.5034,
      "num_tokens": 229824970.0,
      "step": 1123
    },
    {
      "epoch": 2.688995215311005,
      "grad_norm": 0.7328340711385909,
      "learning_rate": 3.315772206926798e-06,
      "loss": 0.6574,
      "num_tokens": 229981855.0,
      "step": 1124
    },
    {
      "epoch": 2.6913875598086126,
      "grad_norm": 0.6549771664676625,
      "learning_rate": 3.308151304108561e-06,
      "loss": 0.5855,
      "num_tokens": 230203169.0,
      "step": 1125
    },
    {
      "epoch": 2.6937799043062203,
      "grad_norm": 0.7356607114309072,
      "learning_rate": 3.300538634025604e-06,
      "loss": 0.6899,
      "num_tokens": 230361759.0,
      "step": 1126
    },
    {
      "epoch": 2.696172248803828,
      "grad_norm": 0.6495199581004002,
      "learning_rate": 3.2929342252716424e-06,
      "loss": 0.4668,
      "num_tokens": 230559686.0,
      "step": 1127
    },
    {
      "epoch": 2.6985645933014353,
      "grad_norm": 0.5919753120066145,
      "learning_rate": 3.2853381064093614e-06,
      "loss": 0.7145,
      "num_tokens": 230825813.0,
      "step": 1128
    },
    {
      "epoch": 2.700956937799043,
      "grad_norm": 0.6280734437638849,
      "learning_rate": 3.27775030597031e-06,
      "loss": 0.6184,
      "num_tokens": 231018150.0,
      "step": 1129
    },
    {
      "epoch": 2.7033492822966507,
      "grad_norm": 0.5629872166072551,
      "learning_rate": 3.2701708524547883e-06,
      "loss": 0.5601,
      "num_tokens": 231240035.0,
      "step": 1130
    },
    {
      "epoch": 2.7057416267942584,
      "grad_norm": 0.56395118769966,
      "learning_rate": 3.2625997743317508e-06,
      "loss": 0.5507,
      "num_tokens": 231467233.0,
      "step": 1131
    },
    {
      "epoch": 2.708133971291866,
      "grad_norm": 0.5779718638481424,
      "learning_rate": 3.255037100038689e-06,
      "loss": 0.6199,
      "num_tokens": 231674686.0,
      "step": 1132
    },
    {
      "epoch": 2.7105263157894735,
      "grad_norm": 0.5112483312693801,
      "learning_rate": 3.247482857981533e-06,
      "loss": 0.3537,
      "num_tokens": 231885671.0,
      "step": 1133
    },
    {
      "epoch": 2.712918660287081,
      "grad_norm": 0.6686531419352844,
      "learning_rate": 3.2399370765345328e-06,
      "loss": 0.6546,
      "num_tokens": 232078705.0,
      "step": 1134
    },
    {
      "epoch": 2.715311004784689,
      "grad_norm": 0.5485950632598388,
      "learning_rate": 3.2323997840401733e-06,
      "loss": 0.4732,
      "num_tokens": 232297544.0,
      "step": 1135
    },
    {
      "epoch": 2.7177033492822966,
      "grad_norm": 0.6605191743413368,
      "learning_rate": 3.2248710088090373e-06,
      "loss": 0.5781,
      "num_tokens": 232480907.0,
      "step": 1136
    },
    {
      "epoch": 2.7200956937799043,
      "grad_norm": 0.6745067883216485,
      "learning_rate": 3.2173507791197338e-06,
      "loss": 0.6685,
      "num_tokens": 232645016.0,
      "step": 1137
    },
    {
      "epoch": 2.722488038277512,
      "grad_norm": 0.5925129825038307,
      "learning_rate": 3.2098391232187597e-06,
      "loss": 0.4972,
      "num_tokens": 232853233.0,
      "step": 1138
    },
    {
      "epoch": 2.72488038277512,
      "grad_norm": 0.6467934814235833,
      "learning_rate": 3.2023360693204166e-06,
      "loss": 0.6755,
      "num_tokens": 233072460.0,
      "step": 1139
    },
    {
      "epoch": 2.7272727272727275,
      "grad_norm": 0.6202378830396424,
      "learning_rate": 3.1948416456066934e-06,
      "loss": 0.5755,
      "num_tokens": 233276759.0,
      "step": 1140
    },
    {
      "epoch": 2.729665071770335,
      "grad_norm": 0.6243928806510726,
      "learning_rate": 3.187355880227163e-06,
      "loss": 0.6521,
      "num_tokens": 233492573.0,
      "step": 1141
    },
    {
      "epoch": 2.7320574162679425,
      "grad_norm": 0.6702870073976731,
      "learning_rate": 3.179878801298879e-06,
      "loss": 0.5395,
      "num_tokens": 233720969.0,
      "step": 1142
    },
    {
      "epoch": 2.7344497607655502,
      "grad_norm": 0.7130159451718587,
      "learning_rate": 3.172410436906266e-06,
      "loss": 0.5496,
      "num_tokens": 233895280.0,
      "step": 1143
    },
    {
      "epoch": 2.736842105263158,
      "grad_norm": 0.6528019038883675,
      "learning_rate": 3.1649508151010178e-06,
      "loss": 0.5347,
      "num_tokens": 234096477.0,
      "step": 1144
    },
    {
      "epoch": 2.7392344497607657,
      "grad_norm": 0.6158483720195869,
      "learning_rate": 3.15749996390199e-06,
      "loss": 0.4725,
      "num_tokens": 234270137.0,
      "step": 1145
    },
    {
      "epoch": 2.741626794258373,
      "grad_norm": 0.5818298212230908,
      "learning_rate": 3.150057911295094e-06,
      "loss": 0.5322,
      "num_tokens": 234494284.0,
      "step": 1146
    },
    {
      "epoch": 2.7440191387559807,
      "grad_norm": 0.6045623627858717,
      "learning_rate": 3.142624685233195e-06,
      "loss": 0.3617,
      "num_tokens": 234638165.0,
      "step": 1147
    },
    {
      "epoch": 2.7464114832535884,
      "grad_norm": 0.7193182995338183,
      "learning_rate": 3.135200313636003e-06,
      "loss": 0.6732,
      "num_tokens": 234845386.0,
      "step": 1148
    },
    {
      "epoch": 2.748803827751196,
      "grad_norm": 0.6826295810846958,
      "learning_rate": 3.1277848243899735e-06,
      "loss": 0.6403,
      "num_tokens": 235068606.0,
      "step": 1149
    },
    {
      "epoch": 2.751196172248804,
      "grad_norm": 0.667979745521155,
      "learning_rate": 3.1203782453481957e-06,
      "loss": 0.5665,
      "num_tokens": 235285213.0,
      "step": 1150
    },
    {
      "epoch": 2.7535885167464116,
      "grad_norm": 0.6627017024547665,
      "learning_rate": 3.112980604330291e-06,
      "loss": 0.6244,
      "num_tokens": 235484153.0,
      "step": 1151
    },
    {
      "epoch": 2.7559808612440193,
      "grad_norm": 0.6369823056389959,
      "learning_rate": 3.1055919291223118e-06,
      "loss": 0.5083,
      "num_tokens": 235660671.0,
      "step": 1152
    },
    {
      "epoch": 2.758373205741627,
      "grad_norm": 0.6595674482265206,
      "learning_rate": 3.0982122474766334e-06,
      "loss": 0.51,
      "num_tokens": 235845956.0,
      "step": 1153
    },
    {
      "epoch": 2.7607655502392343,
      "grad_norm": 0.6180105058916081,
      "learning_rate": 3.09084158711185e-06,
      "loss": 0.3644,
      "num_tokens": 236032609.0,
      "step": 1154
    },
    {
      "epoch": 2.763157894736842,
      "grad_norm": 0.7079586191256172,
      "learning_rate": 3.0834799757126733e-06,
      "loss": 0.6025,
      "num_tokens": 236240870.0,
      "step": 1155
    },
    {
      "epoch": 2.7655502392344498,
      "grad_norm": 0.6307140834614965,
      "learning_rate": 3.0761274409298225e-06,
      "loss": 0.6264,
      "num_tokens": 236448909.0,
      "step": 1156
    },
    {
      "epoch": 2.7679425837320575,
      "grad_norm": 0.6165485458109691,
      "learning_rate": 3.0687840103799273e-06,
      "loss": 0.5498,
      "num_tokens": 236637469.0,
      "step": 1157
    },
    {
      "epoch": 2.770334928229665,
      "grad_norm": 0.7029934436003829,
      "learning_rate": 3.061449711645422e-06,
      "loss": 0.4857,
      "num_tokens": 236807888.0,
      "step": 1158
    },
    {
      "epoch": 2.7727272727272725,
      "grad_norm": 0.6048751465681457,
      "learning_rate": 3.0541245722744386e-06,
      "loss": 0.5999,
      "num_tokens": 237043548.0,
      "step": 1159
    },
    {
      "epoch": 2.77511961722488,
      "grad_norm": 0.5304879535734554,
      "learning_rate": 3.0468086197807067e-06,
      "loss": 0.4942,
      "num_tokens": 237306083.0,
      "step": 1160
    },
    {
      "epoch": 2.777511961722488,
      "grad_norm": 0.5216652603990098,
      "learning_rate": 3.0395018816434518e-06,
      "loss": 0.5044,
      "num_tokens": 237542954.0,
      "step": 1161
    },
    {
      "epoch": 2.7799043062200957,
      "grad_norm": 0.5560558785906827,
      "learning_rate": 3.032204385307286e-06,
      "loss": 0.6172,
      "num_tokens": 237805450.0,
      "step": 1162
    },
    {
      "epoch": 2.7822966507177034,
      "grad_norm": 0.5940789140636036,
      "learning_rate": 3.024916158182111e-06,
      "loss": 0.5746,
      "num_tokens": 238027547.0,
      "step": 1163
    },
    {
      "epoch": 2.784688995215311,
      "grad_norm": 0.5980952368279757,
      "learning_rate": 3.017637227643012e-06,
      "loss": 0.6859,
      "num_tokens": 238267470.0,
      "step": 1164
    },
    {
      "epoch": 2.787081339712919,
      "grad_norm": 0.6098108550239246,
      "learning_rate": 3.010367621030154e-06,
      "loss": 0.6507,
      "num_tokens": 238499989.0,
      "step": 1165
    },
    {
      "epoch": 2.7894736842105265,
      "grad_norm": 0.5189097810260411,
      "learning_rate": 3.0031073656486837e-06,
      "loss": 0.523,
      "num_tokens": 238739324.0,
      "step": 1166
    },
    {
      "epoch": 2.791866028708134,
      "grad_norm": 0.6670038458917087,
      "learning_rate": 2.9958564887686215e-06,
      "loss": 0.5835,
      "num_tokens": 238934431.0,
      "step": 1167
    },
    {
      "epoch": 2.7942583732057416,
      "grad_norm": 0.702165953457099,
      "learning_rate": 2.988615017624763e-06,
      "loss": 0.7271,
      "num_tokens": 239127556.0,
      "step": 1168
    },
    {
      "epoch": 2.7966507177033493,
      "grad_norm": 0.6075352549844745,
      "learning_rate": 2.981382979416574e-06,
      "loss": 0.5385,
      "num_tokens": 239336182.0,
      "step": 1169
    },
    {
      "epoch": 2.799043062200957,
      "grad_norm": 0.6577445889508683,
      "learning_rate": 2.974160401308091e-06,
      "loss": 0.5115,
      "num_tokens": 239498071.0,
      "step": 1170
    },
    {
      "epoch": 2.8014354066985647,
      "grad_norm": 0.5991280998688097,
      "learning_rate": 2.966947310427816e-06,
      "loss": 0.3523,
      "num_tokens": 239663863.0,
      "step": 1171
    },
    {
      "epoch": 2.803827751196172,
      "grad_norm": 0.5288753605341616,
      "learning_rate": 2.959743733868618e-06,
      "loss": 0.6265,
      "num_tokens": 239917379.0,
      "step": 1172
    },
    {
      "epoch": 2.8062200956937797,
      "grad_norm": 0.6582699532838108,
      "learning_rate": 2.952549698687628e-06,
      "loss": 0.5765,
      "num_tokens": 240092500.0,
      "step": 1173
    },
    {
      "epoch": 2.8086124401913874,
      "grad_norm": 0.6417080651634351,
      "learning_rate": 2.9453652319061406e-06,
      "loss": 0.6619,
      "num_tokens": 240292939.0,
      "step": 1174
    },
    {
      "epoch": 2.811004784688995,
      "grad_norm": 0.6397601763793894,
      "learning_rate": 2.9381903605095084e-06,
      "loss": 0.5063,
      "num_tokens": 240496298.0,
      "step": 1175
    },
    {
      "epoch": 2.813397129186603,
      "grad_norm": 0.694398445810493,
      "learning_rate": 2.9310251114470455e-06,
      "loss": 0.6727,
      "num_tokens": 240710446.0,
      "step": 1176
    },
    {
      "epoch": 2.8157894736842106,
      "grad_norm": 0.6298183779915213,
      "learning_rate": 2.923869511631925e-06,
      "loss": 0.5447,
      "num_tokens": 240901068.0,
      "step": 1177
    },
    {
      "epoch": 2.8181818181818183,
      "grad_norm": 0.6481343638854814,
      "learning_rate": 2.916723587941068e-06,
      "loss": 0.5114,
      "num_tokens": 241084940.0,
      "step": 1178
    },
    {
      "epoch": 2.820574162679426,
      "grad_norm": 0.6356530269175816,
      "learning_rate": 2.9095873672150664e-06,
      "loss": 0.5139,
      "num_tokens": 241234540.0,
      "step": 1179
    },
    {
      "epoch": 2.8229665071770333,
      "grad_norm": 0.4917846106897839,
      "learning_rate": 2.902460876258053e-06,
      "loss": 0.3871,
      "num_tokens": 241457101.0,
      "step": 1180
    },
    {
      "epoch": 2.825358851674641,
      "grad_norm": 0.508873897219866,
      "learning_rate": 2.895344141837623e-06,
      "loss": 0.6379,
      "num_tokens": 241712060.0,
      "step": 1181
    },
    {
      "epoch": 2.827751196172249,
      "grad_norm": 0.6102406723360337,
      "learning_rate": 2.888237190684723e-06,
      "loss": 0.5976,
      "num_tokens": 241897217.0,
      "step": 1182
    },
    {
      "epoch": 2.8301435406698565,
      "grad_norm": 0.6716288629809128,
      "learning_rate": 2.8811400494935514e-06,
      "loss": 0.596,
      "num_tokens": 242076528.0,
      "step": 1183
    },
    {
      "epoch": 2.8325358851674642,
      "grad_norm": 0.6693125442801139,
      "learning_rate": 2.874052744921463e-06,
      "loss": 0.6404,
      "num_tokens": 242281353.0,
      "step": 1184
    },
    {
      "epoch": 2.8349282296650715,
      "grad_norm": 0.7474600972985558,
      "learning_rate": 2.866975303588863e-06,
      "loss": 0.431,
      "num_tokens": 242390556.0,
      "step": 1185
    },
    {
      "epoch": 2.8373205741626792,
      "grad_norm": 0.70776528403428,
      "learning_rate": 2.8599077520791113e-06,
      "loss": 0.6298,
      "num_tokens": 242574039.0,
      "step": 1186
    },
    {
      "epoch": 2.839712918660287,
      "grad_norm": 0.6671876432775635,
      "learning_rate": 2.85285011693842e-06,
      "loss": 0.6645,
      "num_tokens": 242774348.0,
      "step": 1187
    },
    {
      "epoch": 2.8421052631578947,
      "grad_norm": 0.6586262202984541,
      "learning_rate": 2.8458024246757542e-06,
      "loss": 0.6766,
      "num_tokens": 242980501.0,
      "step": 1188
    },
    {
      "epoch": 2.8444976076555024,
      "grad_norm": 0.7257360375805328,
      "learning_rate": 2.838764701762733e-06,
      "loss": 0.4131,
      "num_tokens": 243075975.0,
      "step": 1189
    },
    {
      "epoch": 2.84688995215311,
      "grad_norm": 0.6177579979073565,
      "learning_rate": 2.8317369746335336e-06,
      "loss": 0.549,
      "num_tokens": 243251454.0,
      "step": 1190
    },
    {
      "epoch": 2.849282296650718,
      "grad_norm": 0.6230229958426546,
      "learning_rate": 2.8247192696847793e-06,
      "loss": 0.5025,
      "num_tokens": 243413685.0,
      "step": 1191
    },
    {
      "epoch": 2.8516746411483256,
      "grad_norm": 0.6701426528255272,
      "learning_rate": 2.8177116132754613e-06,
      "loss": 0.7372,
      "num_tokens": 243621688.0,
      "step": 1192
    },
    {
      "epoch": 2.854066985645933,
      "grad_norm": 0.6496796729542728,
      "learning_rate": 2.8107140317268155e-06,
      "loss": 0.5084,
      "num_tokens": 243773491.0,
      "step": 1193
    },
    {
      "epoch": 2.8564593301435406,
      "grad_norm": 0.581030373061313,
      "learning_rate": 2.803726551322249e-06,
      "loss": 0.49,
      "num_tokens": 243962139.0,
      "step": 1194
    },
    {
      "epoch": 2.8588516746411483,
      "grad_norm": 0.6196073988069656,
      "learning_rate": 2.796749198307216e-06,
      "loss": 0.5849,
      "num_tokens": 244175734.0,
      "step": 1195
    },
    {
      "epoch": 2.861244019138756,
      "grad_norm": 0.6856361894478656,
      "learning_rate": 2.7897819988891395e-06,
      "loss": 0.5156,
      "num_tokens": 244385159.0,
      "step": 1196
    },
    {
      "epoch": 2.8636363636363638,
      "grad_norm": 0.7476507018765085,
      "learning_rate": 2.782824979237301e-06,
      "loss": 0.6144,
      "num_tokens": 244576046.0,
      "step": 1197
    },
    {
      "epoch": 2.866028708133971,
      "grad_norm": 0.6475977052467857,
      "learning_rate": 2.775878165482748e-06,
      "loss": 0.3768,
      "num_tokens": 244751373.0,
      "step": 1198
    },
    {
      "epoch": 2.8684210526315788,
      "grad_norm": 0.6369236071255224,
      "learning_rate": 2.7689415837181938e-06,
      "loss": 0.5896,
      "num_tokens": 244949211.0,
      "step": 1199
    },
    {
      "epoch": 2.8708133971291865,
      "grad_norm": 0.668885759771659,
      "learning_rate": 2.7620152599979187e-06,
      "loss": 0.4775,
      "num_tokens": 245111385.0,
      "step": 1200
    },
    {
      "epoch": 2.873205741626794,
      "grad_norm": 0.6643818203128119,
      "learning_rate": 2.7550992203376746e-06,
      "loss": 0.5528,
      "num_tokens": 245269535.0,
      "step": 1201
    },
    {
      "epoch": 2.875598086124402,
      "grad_norm": 0.550651881589533,
      "learning_rate": 2.7481934907145834e-06,
      "loss": 0.389,
      "num_tokens": 245470303.0,
      "step": 1202
    },
    {
      "epoch": 2.8779904306220097,
      "grad_norm": 0.5848540423797766,
      "learning_rate": 2.741298097067045e-06,
      "loss": 0.4821,
      "num_tokens": 245648787.0,
      "step": 1203
    },
    {
      "epoch": 2.8803827751196174,
      "grad_norm": 0.634574634635918,
      "learning_rate": 2.7344130652946297e-06,
      "loss": 0.5777,
      "num_tokens": 245840189.0,
      "step": 1204
    },
    {
      "epoch": 2.882775119617225,
      "grad_norm": 0.5812504617218902,
      "learning_rate": 2.727538421258001e-06,
      "loss": 0.5581,
      "num_tokens": 246026826.0,
      "step": 1205
    },
    {
      "epoch": 2.8851674641148324,
      "grad_norm": 0.7546932260743351,
      "learning_rate": 2.7206741907787894e-06,
      "loss": 0.6422,
      "num_tokens": 246221707.0,
      "step": 1206
    },
    {
      "epoch": 2.88755980861244,
      "grad_norm": 0.6438170658948759,
      "learning_rate": 2.713820399639526e-06,
      "loss": 0.5042,
      "num_tokens": 246407305.0,
      "step": 1207
    },
    {
      "epoch": 2.889952153110048,
      "grad_norm": 0.6527785852932717,
      "learning_rate": 2.7069770735835187e-06,
      "loss": 0.4221,
      "num_tokens": 246554288.0,
      "step": 1208
    },
    {
      "epoch": 2.8923444976076556,
      "grad_norm": 0.6539575911422356,
      "learning_rate": 2.700144238314778e-06,
      "loss": 0.6612,
      "num_tokens": 246761962.0,
      "step": 1209
    },
    {
      "epoch": 2.8947368421052633,
      "grad_norm": 0.6864455493850431,
      "learning_rate": 2.6933219194979026e-06,
      "loss": 0.6308,
      "num_tokens": 246962519.0,
      "step": 1210
    },
    {
      "epoch": 2.8971291866028706,
      "grad_norm": 0.6117976724684638,
      "learning_rate": 2.6865101427579947e-06,
      "loss": 0.6078,
      "num_tokens": 247163580.0,
      "step": 1211
    },
    {
      "epoch": 2.8995215311004783,
      "grad_norm": 0.5331232631649402,
      "learning_rate": 2.6797089336805593e-06,
      "loss": 0.3885,
      "num_tokens": 247358140.0,
      "step": 1212
    },
    {
      "epoch": 2.901913875598086,
      "grad_norm": 0.568828690618135,
      "learning_rate": 2.6729183178114073e-06,
      "loss": 0.5382,
      "num_tokens": 247571321.0,
      "step": 1213
    },
    {
      "epoch": 2.9043062200956937,
      "grad_norm": 0.6189893931640706,
      "learning_rate": 2.6661383206565618e-06,
      "loss": 0.5882,
      "num_tokens": 247767989.0,
      "step": 1214
    },
    {
      "epoch": 2.9066985645933014,
      "grad_norm": 0.5178577121697557,
      "learning_rate": 2.6593689676821615e-06,
      "loss": 0.3739,
      "num_tokens": 247952838.0,
      "step": 1215
    },
    {
      "epoch": 2.909090909090909,
      "grad_norm": 0.6986764619040865,
      "learning_rate": 2.652610284314365e-06,
      "loss": 0.6257,
      "num_tokens": 248123176.0,
      "step": 1216
    },
    {
      "epoch": 2.911483253588517,
      "grad_norm": 0.5886690805293942,
      "learning_rate": 2.6458622959392503e-06,
      "loss": 0.5471,
      "num_tokens": 248335747.0,
      "step": 1217
    },
    {
      "epoch": 2.9138755980861246,
      "grad_norm": 0.585596258764202,
      "learning_rate": 2.639125027902735e-06,
      "loss": 0.5411,
      "num_tokens": 248603711.0,
      "step": 1218
    },
    {
      "epoch": 2.916267942583732,
      "grad_norm": 0.48717043479237854,
      "learning_rate": 2.6323985055104563e-06,
      "loss": 0.5637,
      "num_tokens": 248894392.0,
      "step": 1219
    },
    {
      "epoch": 2.9186602870813396,
      "grad_norm": 0.6566559983795414,
      "learning_rate": 2.625682754027705e-06,
      "loss": 0.6663,
      "num_tokens": 249135781.0,
      "step": 1220
    },
    {
      "epoch": 2.9210526315789473,
      "grad_norm": 0.5735215705319195,
      "learning_rate": 2.618977798679301e-06,
      "loss": 0.5486,
      "num_tokens": 249335710.0,
      "step": 1221
    },
    {
      "epoch": 2.923444976076555,
      "grad_norm": 0.6211751646592965,
      "learning_rate": 2.6122836646495265e-06,
      "loss": 0.5461,
      "num_tokens": 249549041.0,
      "step": 1222
    },
    {
      "epoch": 2.925837320574163,
      "grad_norm": 0.642131723486152,
      "learning_rate": 2.6056003770820064e-06,
      "loss": 0.6271,
      "num_tokens": 249748547.0,
      "step": 1223
    },
    {
      "epoch": 2.92822966507177,
      "grad_norm": 0.7207742139976621,
      "learning_rate": 2.598927961079637e-06,
      "loss": 0.4905,
      "num_tokens": 249881783.0,
      "step": 1224
    },
    {
      "epoch": 2.930622009569378,
      "grad_norm": 0.6573176241763478,
      "learning_rate": 2.5922664417044687e-06,
      "loss": 0.5394,
      "num_tokens": 250060894.0,
      "step": 1225
    },
    {
      "epoch": 2.9330143540669855,
      "grad_norm": 0.47731017053274466,
      "learning_rate": 2.5856158439776313e-06,
      "loss": 0.3004,
      "num_tokens": 250217743.0,
      "step": 1226
    },
    {
      "epoch": 2.9354066985645932,
      "grad_norm": 0.6630055860064705,
      "learning_rate": 2.57897619287923e-06,
      "loss": 0.5711,
      "num_tokens": 250453281.0,
      "step": 1227
    },
    {
      "epoch": 2.937799043062201,
      "grad_norm": 0.5738916532045819,
      "learning_rate": 2.5723475133482536e-06,
      "loss": 0.5421,
      "num_tokens": 250666240.0,
      "step": 1228
    },
    {
      "epoch": 2.9401913875598087,
      "grad_norm": 0.5811112867173626,
      "learning_rate": 2.5657298302824807e-06,
      "loss": 0.4771,
      "num_tokens": 250890460.0,
      "step": 1229
    },
    {
      "epoch": 2.9425837320574164,
      "grad_norm": 0.6548400631555413,
      "learning_rate": 2.559123168538384e-06,
      "loss": 0.6436,
      "num_tokens": 251071802.0,
      "step": 1230
    },
    {
      "epoch": 2.944976076555024,
      "grad_norm": 0.6206799035843503,
      "learning_rate": 2.552527552931047e-06,
      "loss": 0.5424,
      "num_tokens": 251287330.0,
      "step": 1231
    },
    {
      "epoch": 2.9473684210526314,
      "grad_norm": 0.7116397379248429,
      "learning_rate": 2.545943008234052e-06,
      "loss": 0.3903,
      "num_tokens": 251421447.0,
      "step": 1232
    },
    {
      "epoch": 2.949760765550239,
      "grad_norm": 0.6222630079338823,
      "learning_rate": 2.53936955917941e-06,
      "loss": 0.6054,
      "num_tokens": 251640901.0,
      "step": 1233
    },
    {
      "epoch": 2.952153110047847,
      "grad_norm": 0.5614733325607167,
      "learning_rate": 2.5328072304574447e-06,
      "loss": 0.5067,
      "num_tokens": 251839968.0,
      "step": 1234
    },
    {
      "epoch": 2.9545454545454546,
      "grad_norm": 0.5829510657829275,
      "learning_rate": 2.526256046716723e-06,
      "loss": 0.3963,
      "num_tokens": 252001429.0,
      "step": 1235
    },
    {
      "epoch": 2.9569377990430623,
      "grad_norm": 0.705061218204218,
      "learning_rate": 2.519716032563937e-06,
      "loss": 0.6263,
      "num_tokens": 252201921.0,
      "step": 1236
    },
    {
      "epoch": 2.9593301435406696,
      "grad_norm": 0.7565419642872887,
      "learning_rate": 2.5131872125638395e-06,
      "loss": 0.4781,
      "num_tokens": 252323614.0,
      "step": 1237
    },
    {
      "epoch": 2.9617224880382773,
      "grad_norm": 0.6684300894500792,
      "learning_rate": 2.5066696112391242e-06,
      "loss": 0.6936,
      "num_tokens": 252532906.0,
      "step": 1238
    },
    {
      "epoch": 2.964114832535885,
      "grad_norm": 0.5483020414647526,
      "learning_rate": 2.5001632530703548e-06,
      "loss": 0.4427,
      "num_tokens": 252733652.0,
      "step": 1239
    },
    {
      "epoch": 2.9665071770334928,
      "grad_norm": 0.6557740442678165,
      "learning_rate": 2.493668162495861e-06,
      "loss": 0.5387,
      "num_tokens": 252911982.0,
      "step": 1240
    },
    {
      "epoch": 2.9688995215311005,
      "grad_norm": 0.6463982314075196,
      "learning_rate": 2.4871843639116546e-06,
      "loss": 0.5954,
      "num_tokens": 253079059.0,
      "step": 1241
    },
    {
      "epoch": 2.971291866028708,
      "grad_norm": 0.6447361732169392,
      "learning_rate": 2.480711881671331e-06,
      "loss": 0.4672,
      "num_tokens": 253237391.0,
      "step": 1242
    },
    {
      "epoch": 2.973684210526316,
      "grad_norm": 0.5759297479374669,
      "learning_rate": 2.474250740085977e-06,
      "loss": 0.4308,
      "num_tokens": 253440527.0,
      "step": 1243
    },
    {
      "epoch": 2.9760765550239237,
      "grad_norm": 0.5509943045604462,
      "learning_rate": 2.467800963424094e-06,
      "loss": 0.4426,
      "num_tokens": 253691730.0,
      "step": 1244
    },
    {
      "epoch": 2.9784688995215314,
      "grad_norm": 0.6422357151167679,
      "learning_rate": 2.461362575911482e-06,
      "loss": 0.613,
      "num_tokens": 253892279.0,
      "step": 1245
    },
    {
      "epoch": 2.9808612440191387,
      "grad_norm": 0.6064370988052857,
      "learning_rate": 2.454935601731177e-06,
      "loss": 0.5947,
      "num_tokens": 254101835.0,
      "step": 1246
    },
    {
      "epoch": 2.9832535885167464,
      "grad_norm": 0.6820673517757627,
      "learning_rate": 2.4485200650233317e-06,
      "loss": 0.5324,
      "num_tokens": 254297612.0,
      "step": 1247
    },
    {
      "epoch": 2.985645933014354,
      "grad_norm": 0.6652085972974082,
      "learning_rate": 2.4421159898851517e-06,
      "loss": 0.5012,
      "num_tokens": 254474125.0,
      "step": 1248
    },
    {
      "epoch": 2.988038277511962,
      "grad_norm": 0.5996508902229061,
      "learning_rate": 2.4357234003707806e-06,
      "loss": 0.6659,
      "num_tokens": 254713889.0,
      "step": 1249
    },
    {
      "epoch": 2.990430622009569,
      "grad_norm": 0.7053309145898407,
      "learning_rate": 2.429342320491232e-06,
      "loss": 0.7754,
      "num_tokens": 254904822.0,
      "step": 1250
    },
    {
      "epoch": 2.992822966507177,
      "grad_norm": 0.5968448525444648,
      "learning_rate": 2.4229727742142786e-06,
      "loss": 0.4946,
      "num_tokens": 255079070.0,
      "step": 1251
    },
    {
      "epoch": 2.9952153110047846,
      "grad_norm": 0.6978601159410557,
      "learning_rate": 2.4166147854643795e-06,
      "loss": 0.8096,
      "num_tokens": 255288793.0,
      "step": 1252
    },
    {
      "epoch": 2.9976076555023923,
      "grad_norm": 0.6134744705998697,
      "learning_rate": 2.4102683781225787e-06,
      "loss": 0.625,
      "num_tokens": 255522740.0,
      "step": 1253
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.6378341238106746,
      "learning_rate": 2.4039335760264217e-06,
      "loss": 0.5745,
      "num_tokens": 255716206.0,
      "step": 1254
    },
    {
      "epoch": 3.0,
      "eval_loss": 0.42938756942749023,
      "eval_num_tokens": 255716206.0,
      "eval_runtime": 109.1806,
      "eval_samples_per_second": 27.221,
      "eval_steps_per_second": 3.407,
      "step": 1254
    },
    {
      "epoch": 3.0023923444976077,
      "grad_norm": 0.7402586261049895,
      "learning_rate": 2.3976104029698633e-06,
      "loss": 0.4572,
      "num_tokens": 255913889.0,
      "step": 1255
    },
    {
      "epoch": 3.0047846889952154,
      "grad_norm": 0.7387412356378916,
      "learning_rate": 2.3912988827031782e-06,
      "loss": 0.6147,
      "num_tokens": 256166684.0,
      "step": 1256
    },
    {
      "epoch": 3.007177033492823,
      "grad_norm": 0.6999477349602615,
      "learning_rate": 2.3849990389328756e-06,
      "loss": 0.4733,
      "num_tokens": 256312393.0,
      "step": 1257
    },
    {
      "epoch": 3.0095693779904304,
      "grad_norm": 0.5690209540504421,
      "learning_rate": 2.378710895321599e-06,
      "loss": 0.3537,
      "num_tokens": 256512256.0,
      "step": 1258
    },
    {
      "epoch": 3.011961722488038,
      "grad_norm": 0.7890373783734057,
      "learning_rate": 2.3724344754880567e-06,
      "loss": 0.5304,
      "num_tokens": 256727440.0,
      "step": 1259
    },
    {
      "epoch": 3.014354066985646,
      "grad_norm": 0.9614164329801103,
      "learning_rate": 2.3661698030069102e-06,
      "loss": 0.6293,
      "num_tokens": 256908452.0,
      "step": 1260
    },
    {
      "epoch": 3.0167464114832536,
      "grad_norm": 0.6512936329535286,
      "learning_rate": 2.359916901408708e-06,
      "loss": 0.4104,
      "num_tokens": 257118864.0,
      "step": 1261
    },
    {
      "epoch": 3.0191387559808613,
      "grad_norm": 0.6700153979446825,
      "learning_rate": 2.3536757941797756e-06,
      "loss": 0.5046,
      "num_tokens": 257330088.0,
      "step": 1262
    },
    {
      "epoch": 3.021531100478469,
      "grad_norm": 0.677137498964303,
      "learning_rate": 2.3474465047621483e-06,
      "loss": 0.5112,
      "num_tokens": 257491666.0,
      "step": 1263
    },
    {
      "epoch": 3.0239234449760763,
      "grad_norm": 0.6832888725075986,
      "learning_rate": 2.341229056553464e-06,
      "loss": 0.6466,
      "num_tokens": 257705251.0,
      "step": 1264
    },
    {
      "epoch": 3.026315789473684,
      "grad_norm": 0.6843703156850178,
      "learning_rate": 2.3350234729068903e-06,
      "loss": 0.5112,
      "num_tokens": 257901651.0,
      "step": 1265
    },
    {
      "epoch": 3.028708133971292,
      "grad_norm": 0.6746922650910443,
      "learning_rate": 2.3288297771310275e-06,
      "loss": 0.6056,
      "num_tokens": 258111668.0,
      "step": 1266
    },
    {
      "epoch": 3.0311004784688995,
      "grad_norm": 0.63011570756557,
      "learning_rate": 2.3226479924898256e-06,
      "loss": 0.5336,
      "num_tokens": 258323599.0,
      "step": 1267
    },
    {
      "epoch": 3.0334928229665072,
      "grad_norm": 0.6593826509477446,
      "learning_rate": 2.316478142202496e-06,
      "loss": 0.4867,
      "num_tokens": 258494973.0,
      "step": 1268
    },
    {
      "epoch": 3.035885167464115,
      "grad_norm": 0.5949720867828437,
      "learning_rate": 2.310320249443422e-06,
      "loss": 0.4177,
      "num_tokens": 258689669.0,
      "step": 1269
    },
    {
      "epoch": 3.0382775119617227,
      "grad_norm": 0.5964321271477145,
      "learning_rate": 2.3041743373420745e-06,
      "loss": 0.4252,
      "num_tokens": 258906775.0,
      "step": 1270
    },
    {
      "epoch": 3.04066985645933,
      "grad_norm": 0.6582924002907377,
      "learning_rate": 2.298040428982925e-06,
      "loss": 0.5591,
      "num_tokens": 259077744.0,
      "step": 1271
    },
    {
      "epoch": 3.0430622009569377,
      "grad_norm": 0.6279206160144934,
      "learning_rate": 2.291918547405359e-06,
      "loss": 0.4881,
      "num_tokens": 259270234.0,
      "step": 1272
    },
    {
      "epoch": 3.0454545454545454,
      "grad_norm": 0.553921649997124,
      "learning_rate": 2.285808715603583e-06,
      "loss": 0.3983,
      "num_tokens": 259472134.0,
      "step": 1273
    },
    {
      "epoch": 3.047846889952153,
      "grad_norm": 0.5492785480976846,
      "learning_rate": 2.279710956526553e-06,
      "loss": 0.4185,
      "num_tokens": 259678929.0,
      "step": 1274
    },
    {
      "epoch": 3.050239234449761,
      "grad_norm": 0.5846959421173259,
      "learning_rate": 2.273625293077867e-06,
      "loss": 0.5988,
      "num_tokens": 259932212.0,
      "step": 1275
    },
    {
      "epoch": 3.0526315789473686,
      "grad_norm": 0.589709064028512,
      "learning_rate": 2.2675517481157055e-06,
      "loss": 0.4188,
      "num_tokens": 260146724.0,
      "step": 1276
    },
    {
      "epoch": 3.055023923444976,
      "grad_norm": 0.6274338008637919,
      "learning_rate": 2.2614903444527188e-06,
      "loss": 0.4442,
      "num_tokens": 260350406.0,
      "step": 1277
    },
    {
      "epoch": 3.0574162679425836,
      "grad_norm": 0.7915974724556752,
      "learning_rate": 2.255441104855958e-06,
      "loss": 0.5775,
      "num_tokens": 260538578.0,
      "step": 1278
    },
    {
      "epoch": 3.0598086124401913,
      "grad_norm": 0.584101511052562,
      "learning_rate": 2.2494040520467874e-06,
      "loss": 0.4285,
      "num_tokens": 260755344.0,
      "step": 1279
    },
    {
      "epoch": 3.062200956937799,
      "grad_norm": 0.5949543375784039,
      "learning_rate": 2.243379208700793e-06,
      "loss": 0.4389,
      "num_tokens": 260929211.0,
      "step": 1280
    },
    {
      "epoch": 3.0645933014354068,
      "grad_norm": 0.7097892851187321,
      "learning_rate": 2.237366597447705e-06,
      "loss": 0.6095,
      "num_tokens": 261139280.0,
      "step": 1281
    },
    {
      "epoch": 3.0669856459330145,
      "grad_norm": 0.6198718642387921,
      "learning_rate": 2.2313662408713054e-06,
      "loss": 0.5429,
      "num_tokens": 261341146.0,
      "step": 1282
    },
    {
      "epoch": 3.069377990430622,
      "grad_norm": 0.6087302978836517,
      "learning_rate": 2.225378161509349e-06,
      "loss": 0.5559,
      "num_tokens": 261581286.0,
      "step": 1283
    },
    {
      "epoch": 3.0717703349282295,
      "grad_norm": 0.6220061356000266,
      "learning_rate": 2.2194023818534767e-06,
      "loss": 0.4997,
      "num_tokens": 261799499.0,
      "step": 1284
    },
    {
      "epoch": 3.074162679425837,
      "grad_norm": 0.5368146821622435,
      "learning_rate": 2.2134389243491296e-06,
      "loss": 0.5151,
      "num_tokens": 262026022.0,
      "step": 1285
    },
    {
      "epoch": 3.076555023923445,
      "grad_norm": 0.7153515807441471,
      "learning_rate": 2.207487811395466e-06,
      "loss": 0.6361,
      "num_tokens": 262256556.0,
      "step": 1286
    },
    {
      "epoch": 3.0789473684210527,
      "grad_norm": 0.6213730994225999,
      "learning_rate": 2.201549065345278e-06,
      "loss": 0.5283,
      "num_tokens": 262477701.0,
      "step": 1287
    },
    {
      "epoch": 3.0813397129186604,
      "grad_norm": 0.6574330655368827,
      "learning_rate": 2.195622708504907e-06,
      "loss": 0.4982,
      "num_tokens": 262690340.0,
      "step": 1288
    },
    {
      "epoch": 3.083732057416268,
      "grad_norm": 0.6575370913279897,
      "learning_rate": 2.1897087631341595e-06,
      "loss": 0.5313,
      "num_tokens": 262928530.0,
      "step": 1289
    },
    {
      "epoch": 3.0861244019138754,
      "grad_norm": 0.6051544279189891,
      "learning_rate": 2.1838072514462224e-06,
      "loss": 0.4823,
      "num_tokens": 263167965.0,
      "step": 1290
    },
    {
      "epoch": 3.088516746411483,
      "grad_norm": 0.6999294105552621,
      "learning_rate": 2.1779181956075805e-06,
      "loss": 0.4199,
      "num_tokens": 263323188.0,
      "step": 1291
    },
    {
      "epoch": 3.090909090909091,
      "grad_norm": 0.5627138982473117,
      "learning_rate": 2.1720416177379366e-06,
      "loss": 0.4115,
      "num_tokens": 263537362.0,
      "step": 1292
    },
    {
      "epoch": 3.0933014354066986,
      "grad_norm": 0.629390025884304,
      "learning_rate": 2.1661775399101234e-06,
      "loss": 0.4305,
      "num_tokens": 263749336.0,
      "step": 1293
    },
    {
      "epoch": 3.0956937799043063,
      "grad_norm": 0.7247727661733878,
      "learning_rate": 2.1603259841500228e-06,
      "loss": 0.5312,
      "num_tokens": 263914691.0,
      "step": 1294
    },
    {
      "epoch": 3.098086124401914,
      "grad_norm": 0.5586270865087458,
      "learning_rate": 2.1544869724364816e-06,
      "loss": 0.3326,
      "num_tokens": 264052504.0,
      "step": 1295
    },
    {
      "epoch": 3.1004784688995217,
      "grad_norm": 0.5532261224348151,
      "learning_rate": 2.148660526701233e-06,
      "loss": 0.4917,
      "num_tokens": 264268596.0,
      "step": 1296
    },
    {
      "epoch": 3.102870813397129,
      "grad_norm": 0.549079831877891,
      "learning_rate": 2.1428466688288102e-06,
      "loss": 0.5321,
      "num_tokens": 264529822.0,
      "step": 1297
    },
    {
      "epoch": 3.1052631578947367,
      "grad_norm": 0.5671370864801888,
      "learning_rate": 2.1370454206564646e-06,
      "loss": 0.4896,
      "num_tokens": 264748064.0,
      "step": 1298
    },
    {
      "epoch": 3.1076555023923444,
      "grad_norm": 0.5904040731943784,
      "learning_rate": 2.131256803974086e-06,
      "loss": 0.4935,
      "num_tokens": 264969358.0,
      "step": 1299
    },
    {
      "epoch": 3.110047846889952,
      "grad_norm": 0.5626029492726947,
      "learning_rate": 2.1254808405241183e-06,
      "loss": 0.4896,
      "num_tokens": 265201219.0,
      "step": 1300
    },
    {
      "epoch": 3.11244019138756,
      "grad_norm": 0.6848742442647878,
      "learning_rate": 2.1197175520014806e-06,
      "loss": 0.5598,
      "num_tokens": 265406095.0,
      "step": 1301
    },
    {
      "epoch": 3.1148325358851676,
      "grad_norm": 0.675172611025597,
      "learning_rate": 2.1139669600534825e-06,
      "loss": 0.5134,
      "num_tokens": 265634192.0,
      "step": 1302
    },
    {
      "epoch": 3.117224880382775,
      "grad_norm": 0.5877000187862749,
      "learning_rate": 2.1082290862797464e-06,
      "loss": 0.5396,
      "num_tokens": 265876633.0,
      "step": 1303
    },
    {
      "epoch": 3.1196172248803826,
      "grad_norm": 0.5662213441307106,
      "learning_rate": 2.102503952232121e-06,
      "loss": 0.5159,
      "num_tokens": 266111430.0,
      "step": 1304
    },
    {
      "epoch": 3.1220095693779903,
      "grad_norm": 0.5985612354820727,
      "learning_rate": 2.0967915794146066e-06,
      "loss": 0.4746,
      "num_tokens": 266329504.0,
      "step": 1305
    },
    {
      "epoch": 3.124401913875598,
      "grad_norm": 0.6527589826039932,
      "learning_rate": 2.0910919892832697e-06,
      "loss": 0.3463,
      "num_tokens": 266471386.0,
      "step": 1306
    },
    {
      "epoch": 3.126794258373206,
      "grad_norm": 0.47596331322234914,
      "learning_rate": 2.0854052032461667e-06,
      "loss": 0.3295,
      "num_tokens": 266684099.0,
      "step": 1307
    },
    {
      "epoch": 3.1291866028708135,
      "grad_norm": 0.6108469881663725,
      "learning_rate": 2.0797312426632584e-06,
      "loss": 0.6519,
      "num_tokens": 266933331.0,
      "step": 1308
    },
    {
      "epoch": 3.1315789473684212,
      "grad_norm": 0.6646226322119491,
      "learning_rate": 2.0740701288463327e-06,
      "loss": 0.5259,
      "num_tokens": 267130637.0,
      "step": 1309
    },
    {
      "epoch": 3.1339712918660285,
      "grad_norm": 0.5508619091258276,
      "learning_rate": 2.068421883058925e-06,
      "loss": 0.3927,
      "num_tokens": 267299054.0,
      "step": 1310
    },
    {
      "epoch": 3.1363636363636362,
      "grad_norm": 0.6293553226109324,
      "learning_rate": 2.0627865265162374e-06,
      "loss": 0.433,
      "num_tokens": 267510099.0,
      "step": 1311
    },
    {
      "epoch": 3.138755980861244,
      "grad_norm": 0.7440115581163417,
      "learning_rate": 2.057164080385058e-06,
      "loss": 0.5481,
      "num_tokens": 267726029.0,
      "step": 1312
    },
    {
      "epoch": 3.1411483253588517,
      "grad_norm": 0.6236043688054738,
      "learning_rate": 2.051554565783684e-06,
      "loss": 0.4325,
      "num_tokens": 267925976.0,
      "step": 1313
    },
    {
      "epoch": 3.1435406698564594,
      "grad_norm": 0.671585593115855,
      "learning_rate": 2.04595800378184e-06,
      "loss": 0.4412,
      "num_tokens": 268098208.0,
      "step": 1314
    },
    {
      "epoch": 3.145933014354067,
      "grad_norm": 0.6949266221749684,
      "learning_rate": 2.0403744154005988e-06,
      "loss": 0.4432,
      "num_tokens": 268308006.0,
      "step": 1315
    },
    {
      "epoch": 3.1483253588516744,
      "grad_norm": 0.6473691472293852,
      "learning_rate": 2.034803821612308e-06,
      "loss": 0.5865,
      "num_tokens": 268514308.0,
      "step": 1316
    },
    {
      "epoch": 3.150717703349282,
      "grad_norm": 0.5536706707358635,
      "learning_rate": 2.0292462433404976e-06,
      "loss": 0.4216,
      "num_tokens": 268725648.0,
      "step": 1317
    },
    {
      "epoch": 3.15311004784689,
      "grad_norm": 0.6221734201066783,
      "learning_rate": 2.0237017014598215e-06,
      "loss": 0.4751,
      "num_tokens": 268940579.0,
      "step": 1318
    },
    {
      "epoch": 3.1555023923444976,
      "grad_norm": 0.6392719861236702,
      "learning_rate": 2.018170216795959e-06,
      "loss": 0.4724,
      "num_tokens": 269130369.0,
      "step": 1319
    },
    {
      "epoch": 3.1578947368421053,
      "grad_norm": 0.7162837149510706,
      "learning_rate": 2.0126518101255495e-06,
      "loss": 0.5957,
      "num_tokens": 269316540.0,
      "step": 1320
    },
    {
      "epoch": 3.160287081339713,
      "grad_norm": 0.5580788330841133,
      "learning_rate": 2.0071465021761106e-06,
      "loss": 0.4358,
      "num_tokens": 269506411.0,
      "step": 1321
    },
    {
      "epoch": 3.1626794258373208,
      "grad_norm": 0.6598206261361217,
      "learning_rate": 2.0016543136259595e-06,
      "loss": 0.5602,
      "num_tokens": 269701502.0,
      "step": 1322
    },
    {
      "epoch": 3.165071770334928,
      "grad_norm": 0.6581789669102812,
      "learning_rate": 1.996175265104136e-06,
      "loss": 0.529,
      "num_tokens": 269900469.0,
      "step": 1323
    },
    {
      "epoch": 3.1674641148325358,
      "grad_norm": 0.6059113124115182,
      "learning_rate": 1.9907093771903265e-06,
      "loss": 0.4354,
      "num_tokens": 270096044.0,
      "step": 1324
    },
    {
      "epoch": 3.1698564593301435,
      "grad_norm": 0.8188363685332233,
      "learning_rate": 1.9852566704147823e-06,
      "loss": 0.6595,
      "num_tokens": 270254500.0,
      "step": 1325
    },
    {
      "epoch": 3.172248803827751,
      "grad_norm": 0.6548140241893287,
      "learning_rate": 1.979817165258248e-06,
      "loss": 0.5036,
      "num_tokens": 270443820.0,
      "step": 1326
    },
    {
      "epoch": 3.174641148325359,
      "grad_norm": 0.5207871010872199,
      "learning_rate": 1.9743908821518813e-06,
      "loss": 0.4581,
      "num_tokens": 270676004.0,
      "step": 1327
    },
    {
      "epoch": 3.1770334928229667,
      "grad_norm": 0.5652418638260854,
      "learning_rate": 1.968977841477176e-06,
      "loss": 0.3529,
      "num_tokens": 270893341.0,
      "step": 1328
    },
    {
      "epoch": 3.179425837320574,
      "grad_norm": 0.596366758247755,
      "learning_rate": 1.9635780635658886e-06,
      "loss": 0.5376,
      "num_tokens": 271106483.0,
      "step": 1329
    },
    {
      "epoch": 3.1818181818181817,
      "grad_norm": 0.645025175672765,
      "learning_rate": 1.9581915686999563e-06,
      "loss": 0.4806,
      "num_tokens": 271302176.0,
      "step": 1330
    },
    {
      "epoch": 3.1842105263157894,
      "grad_norm": 0.7117928160952178,
      "learning_rate": 1.9528183771114295e-06,
      "loss": 0.4697,
      "num_tokens": 271514293.0,
      "step": 1331
    },
    {
      "epoch": 3.186602870813397,
      "grad_norm": 0.6221638039635599,
      "learning_rate": 1.9474585089823827e-06,
      "loss": 0.4587,
      "num_tokens": 271732762.0,
      "step": 1332
    },
    {
      "epoch": 3.188995215311005,
      "grad_norm": 0.7078194657393896,
      "learning_rate": 1.9421119844448565e-06,
      "loss": 0.5736,
      "num_tokens": 271912358.0,
      "step": 1333
    },
    {
      "epoch": 3.1913875598086126,
      "grad_norm": 0.6440523574621861,
      "learning_rate": 1.9367788235807632e-06,
      "loss": 0.507,
      "num_tokens": 272100294.0,
      "step": 1334
    },
    {
      "epoch": 3.1937799043062203,
      "grad_norm": 0.5978274244815054,
      "learning_rate": 1.931459046421825e-06,
      "loss": 0.5187,
      "num_tokens": 272301185.0,
      "step": 1335
    },
    {
      "epoch": 3.1961722488038276,
      "grad_norm": 0.5143790337279445,
      "learning_rate": 1.9261526729494923e-06,
      "loss": 0.3528,
      "num_tokens": 272529698.0,
      "step": 1336
    },
    {
      "epoch": 3.1985645933014353,
      "grad_norm": 0.6269390757214239,
      "learning_rate": 1.9208597230948717e-06,
      "loss": 0.4604,
      "num_tokens": 272750346.0,
      "step": 1337
    },
    {
      "epoch": 3.200956937799043,
      "grad_norm": 0.6020267987482506,
      "learning_rate": 1.9155802167386485e-06,
      "loss": 0.4629,
      "num_tokens": 272972518.0,
      "step": 1338
    },
    {
      "epoch": 3.2033492822966507,
      "grad_norm": 0.6342687191296817,
      "learning_rate": 1.9103141737110132e-06,
      "loss": 0.6085,
      "num_tokens": 273188653.0,
      "step": 1339
    },
    {
      "epoch": 3.2057416267942584,
      "grad_norm": 0.6471878633504836,
      "learning_rate": 1.9050616137915876e-06,
      "loss": 0.4324,
      "num_tokens": 273376449.0,
      "step": 1340
    },
    {
      "epoch": 3.208133971291866,
      "grad_norm": 0.6475159896178175,
      "learning_rate": 1.8998225567093506e-06,
      "loss": 0.4442,
      "num_tokens": 273542078.0,
      "step": 1341
    },
    {
      "epoch": 3.2105263157894735,
      "grad_norm": 0.6651285488992867,
      "learning_rate": 1.894597022142563e-06,
      "loss": 0.4446,
      "num_tokens": 273740563.0,
      "step": 1342
    },
    {
      "epoch": 3.212918660287081,
      "grad_norm": 0.7333478802010834,
      "learning_rate": 1.8893850297186938e-06,
      "loss": 0.4411,
      "num_tokens": 273911486.0,
      "step": 1343
    },
    {
      "epoch": 3.215311004784689,
      "grad_norm": 0.7195592143270082,
      "learning_rate": 1.8841865990143476e-06,
      "loss": 0.4759,
      "num_tokens": 274101029.0,
      "step": 1344
    },
    {
      "epoch": 3.2177033492822966,
      "grad_norm": 0.6670742198623395,
      "learning_rate": 1.8790017495551871e-06,
      "loss": 0.4744,
      "num_tokens": 274258804.0,
      "step": 1345
    },
    {
      "epoch": 3.2200956937799043,
      "grad_norm": 0.594034860458389,
      "learning_rate": 1.8738305008158692e-06,
      "loss": 0.5643,
      "num_tokens": 274480204.0,
      "step": 1346
    },
    {
      "epoch": 3.222488038277512,
      "grad_norm": 0.5442037919610911,
      "learning_rate": 1.8686728722199576e-06,
      "loss": 0.4505,
      "num_tokens": 274743758.0,
      "step": 1347
    },
    {
      "epoch": 3.22488038277512,
      "grad_norm": 0.6333611210516911,
      "learning_rate": 1.8635288831398666e-06,
      "loss": 0.5109,
      "num_tokens": 274953215.0,
      "step": 1348
    },
    {
      "epoch": 3.227272727272727,
      "grad_norm": 0.5057598137437262,
      "learning_rate": 1.8583985528967715e-06,
      "loss": 0.5229,
      "num_tokens": 275221291.0,
      "step": 1349
    },
    {
      "epoch": 3.229665071770335,
      "grad_norm": 0.7028594855989811,
      "learning_rate": 1.8532819007605482e-06,
      "loss": 0.5113,
      "num_tokens": 275395942.0,
      "step": 1350
    },
    {
      "epoch": 3.2320574162679425,
      "grad_norm": 0.6257706921052196,
      "learning_rate": 1.8481789459496962e-06,
      "loss": 0.5542,
      "num_tokens": 275628282.0,
      "step": 1351
    },
    {
      "epoch": 3.2344497607655502,
      "grad_norm": 0.6279792884558518,
      "learning_rate": 1.843089707631267e-06,
      "loss": 0.4097,
      "num_tokens": 275800544.0,
      "step": 1352
    },
    {
      "epoch": 3.236842105263158,
      "grad_norm": 0.6539270252609328,
      "learning_rate": 1.8380142049207916e-06,
      "loss": 0.5232,
      "num_tokens": 275973348.0,
      "step": 1353
    },
    {
      "epoch": 3.2392344497607657,
      "grad_norm": 0.6739143040433802,
      "learning_rate": 1.8329524568822087e-06,
      "loss": 0.5698,
      "num_tokens": 276166783.0,
      "step": 1354
    },
    {
      "epoch": 3.2416267942583734,
      "grad_norm": 0.674231860453957,
      "learning_rate": 1.8279044825277941e-06,
      "loss": 0.5229,
      "num_tokens": 276344559.0,
      "step": 1355
    },
    {
      "epoch": 3.2440191387559807,
      "grad_norm": 0.7948154225401068,
      "learning_rate": 1.8228703008180892e-06,
      "loss": 0.6472,
      "num_tokens": 276514471.0,
      "step": 1356
    },
    {
      "epoch": 3.2464114832535884,
      "grad_norm": 0.5373150672084231,
      "learning_rate": 1.817849930661829e-06,
      "loss": 0.4819,
      "num_tokens": 276767291.0,
      "step": 1357
    },
    {
      "epoch": 3.248803827751196,
      "grad_norm": 0.5376958767027991,
      "learning_rate": 1.8128433909158689e-06,
      "loss": 0.4425,
      "num_tokens": 276976997.0,
      "step": 1358
    },
    {
      "epoch": 3.251196172248804,
      "grad_norm": 0.587439310427447,
      "learning_rate": 1.8078507003851209e-06,
      "loss": 0.4103,
      "num_tokens": 277181334.0,
      "step": 1359
    },
    {
      "epoch": 3.2535885167464116,
      "grad_norm": 0.5554877966420847,
      "learning_rate": 1.8028718778224727e-06,
      "loss": 0.4608,
      "num_tokens": 277438461.0,
      "step": 1360
    },
    {
      "epoch": 3.2559808612440193,
      "grad_norm": 0.7349112175964104,
      "learning_rate": 1.7979069419287304e-06,
      "loss": 0.5519,
      "num_tokens": 277603370.0,
      "step": 1361
    },
    {
      "epoch": 3.258373205741627,
      "grad_norm": 0.5577771336621236,
      "learning_rate": 1.7929559113525322e-06,
      "loss": 0.5025,
      "num_tokens": 277816222.0,
      "step": 1362
    },
    {
      "epoch": 3.2607655502392343,
      "grad_norm": 0.6032365353365063,
      "learning_rate": 1.7880188046902957e-06,
      "loss": 0.4066,
      "num_tokens": 278008701.0,
      "step": 1363
    },
    {
      "epoch": 3.263157894736842,
      "grad_norm": 0.6015228385731863,
      "learning_rate": 1.783095640486131e-06,
      "loss": 0.5296,
      "num_tokens": 278240487.0,
      "step": 1364
    },
    {
      "epoch": 3.2655502392344498,
      "grad_norm": 0.5754578129860267,
      "learning_rate": 1.778186437231786e-06,
      "loss": 0.6699,
      "num_tokens": 278485002.0,
      "step": 1365
    },
    {
      "epoch": 3.2679425837320575,
      "grad_norm": 0.5289445323732264,
      "learning_rate": 1.7732912133665658e-06,
      "loss": 0.202,
      "num_tokens": 278625829.0,
      "step": 1366
    },
    {
      "epoch": 3.270334928229665,
      "grad_norm": 0.5179852711199288,
      "learning_rate": 1.7684099872772714e-06,
      "loss": 0.4492,
      "num_tokens": 278887561.0,
      "step": 1367
    },
    {
      "epoch": 3.2727272727272725,
      "grad_norm": 0.656713592521766,
      "learning_rate": 1.7635427772981245e-06,
      "loss": 0.5281,
      "num_tokens": 279094325.0,
      "step": 1368
    },
    {
      "epoch": 3.27511961722488,
      "grad_norm": 0.6915296583387971,
      "learning_rate": 1.7586896017107024e-06,
      "loss": 0.4877,
      "num_tokens": 279256505.0,
      "step": 1369
    },
    {
      "epoch": 3.277511961722488,
      "grad_norm": 0.5676187090907497,
      "learning_rate": 1.7538504787438698e-06,
      "loss": 0.4877,
      "num_tokens": 279501359.0,
      "step": 1370
    },
    {
      "epoch": 3.2799043062200957,
      "grad_norm": 0.5059203117134556,
      "learning_rate": 1.749025426573704e-06,
      "loss": 0.4648,
      "num_tokens": 279742742.0,
      "step": 1371
    },
    {
      "epoch": 3.2822966507177034,
      "grad_norm": 0.679642013673189,
      "learning_rate": 1.7442144633234385e-06,
      "loss": 0.5408,
      "num_tokens": 279936081.0,
      "step": 1372
    },
    {
      "epoch": 3.284688995215311,
      "grad_norm": 0.5728793788182743,
      "learning_rate": 1.7394176070633801e-06,
      "loss": 0.5537,
      "num_tokens": 280175806.0,
      "step": 1373
    },
    {
      "epoch": 3.287081339712919,
      "grad_norm": 0.5101464525299488,
      "learning_rate": 1.7346348758108558e-06,
      "loss": 0.3603,
      "num_tokens": 280390741.0,
      "step": 1374
    },
    {
      "epoch": 3.2894736842105265,
      "grad_norm": 0.6298630050286675,
      "learning_rate": 1.729866287530132e-06,
      "loss": 0.6027,
      "num_tokens": 280626762.0,
      "step": 1375
    },
    {
      "epoch": 3.291866028708134,
      "grad_norm": 0.610266116505584,
      "learning_rate": 1.7251118601323606e-06,
      "loss": 0.3322,
      "num_tokens": 280826020.0,
      "step": 1376
    },
    {
      "epoch": 3.2942583732057416,
      "grad_norm": 0.6684549544777105,
      "learning_rate": 1.7203716114754937e-06,
      "loss": 0.4599,
      "num_tokens": 281012635.0,
      "step": 1377
    },
    {
      "epoch": 3.2966507177033493,
      "grad_norm": 0.5036480775036766,
      "learning_rate": 1.7156455593642387e-06,
      "loss": 0.3533,
      "num_tokens": 281221620.0,
      "step": 1378
    },
    {
      "epoch": 3.299043062200957,
      "grad_norm": 0.7343363629973206,
      "learning_rate": 1.7109337215499711e-06,
      "loss": 0.5138,
      "num_tokens": 281411026.0,
      "step": 1379
    },
    {
      "epoch": 3.3014354066985647,
      "grad_norm": 0.7136580742476932,
      "learning_rate": 1.7062361157306795e-06,
      "loss": 0.6031,
      "num_tokens": 281635408.0,
      "step": 1380
    },
    {
      "epoch": 3.303827751196172,
      "grad_norm": 0.6779875945232949,
      "learning_rate": 1.7015527595508973e-06,
      "loss": 0.435,
      "num_tokens": 281823429.0,
      "step": 1381
    },
    {
      "epoch": 3.3062200956937797,
      "grad_norm": 0.6165374543507615,
      "learning_rate": 1.6968836706016334e-06,
      "loss": 0.5343,
      "num_tokens": 282042936.0,
      "step": 1382
    },
    {
      "epoch": 3.3086124401913874,
      "grad_norm": 0.6671429016147503,
      "learning_rate": 1.6922288664203106e-06,
      "loss": 0.6049,
      "num_tokens": 282244064.0,
      "step": 1383
    },
    {
      "epoch": 3.311004784688995,
      "grad_norm": 0.6971528883690388,
      "learning_rate": 1.6875883644906916e-06,
      "loss": 0.4373,
      "num_tokens": 282421524.0,
      "step": 1384
    },
    {
      "epoch": 3.313397129186603,
      "grad_norm": 0.6347144039685346,
      "learning_rate": 1.6829621822428273e-06,
      "loss": 0.4571,
      "num_tokens": 282591812.0,
      "step": 1385
    },
    {
      "epoch": 3.3157894736842106,
      "grad_norm": 0.596451326446268,
      "learning_rate": 1.678350337052973e-06,
      "loss": 0.5014,
      "num_tokens": 282841560.0,
      "step": 1386
    },
    {
      "epoch": 3.3181818181818183,
      "grad_norm": 0.5504535023805781,
      "learning_rate": 1.6737528462435432e-06,
      "loss": 0.4702,
      "num_tokens": 283075035.0,
      "step": 1387
    },
    {
      "epoch": 3.320574162679426,
      "grad_norm": 0.6372384711768105,
      "learning_rate": 1.6691697270830264e-06,
      "loss": 0.4023,
      "num_tokens": 283262952.0,
      "step": 1388
    },
    {
      "epoch": 3.3229665071770333,
      "grad_norm": 0.44016337227270674,
      "learning_rate": 1.6646009967859406e-06,
      "loss": 0.3404,
      "num_tokens": 283507482.0,
      "step": 1389
    },
    {
      "epoch": 3.325358851674641,
      "grad_norm": 0.6430682309007208,
      "learning_rate": 1.660046672512747e-06,
      "loss": 0.6049,
      "num_tokens": 283741511.0,
      "step": 1390
    },
    {
      "epoch": 3.327751196172249,
      "grad_norm": 0.6662078531560585,
      "learning_rate": 1.6555067713698078e-06,
      "loss": 0.4163,
      "num_tokens": 283910656.0,
      "step": 1391
    },
    {
      "epoch": 3.3301435406698565,
      "grad_norm": 0.5236951135874593,
      "learning_rate": 1.6509813104093025e-06,
      "loss": 0.433,
      "num_tokens": 284141437.0,
      "step": 1392
    },
    {
      "epoch": 3.3325358851674642,
      "grad_norm": 0.5891915715828933,
      "learning_rate": 1.646470306629176e-06,
      "loss": 0.4364,
      "num_tokens": 284302071.0,
      "step": 1393
    },
    {
      "epoch": 3.334928229665072,
      "grad_norm": 0.5659081142660974,
      "learning_rate": 1.64197377697307e-06,
      "loss": 0.3732,
      "num_tokens": 284465884.0,
      "step": 1394
    },
    {
      "epoch": 3.3373205741626792,
      "grad_norm": 0.581006107542102,
      "learning_rate": 1.6374917383302616e-06,
      "loss": 0.3888,
      "num_tokens": 284650977.0,
      "step": 1395
    },
    {
      "epoch": 3.339712918660287,
      "grad_norm": 0.6364252639645998,
      "learning_rate": 1.6330242075355996e-06,
      "loss": 0.4256,
      "num_tokens": 284823907.0,
      "step": 1396
    },
    {
      "epoch": 3.3421052631578947,
      "grad_norm": 0.6938031314089937,
      "learning_rate": 1.6285712013694338e-06,
      "loss": 0.4122,
      "num_tokens": 284984075.0,
      "step": 1397
    },
    {
      "epoch": 3.3444976076555024,
      "grad_norm": 0.6895442975483198,
      "learning_rate": 1.6241327365575688e-06,
      "loss": 0.564,
      "num_tokens": 285203372.0,
      "step": 1398
    },
    {
      "epoch": 3.34688995215311,
      "grad_norm": 0.5894084030392284,
      "learning_rate": 1.6197088297711806e-06,
      "loss": 0.4046,
      "num_tokens": 285398044.0,
      "step": 1399
    },
    {
      "epoch": 3.349282296650718,
      "grad_norm": 0.72395557559556,
      "learning_rate": 1.6152994976267732e-06,
      "loss": 0.5514,
      "num_tokens": 285592197.0,
      "step": 1400
    },
    {
      "epoch": 3.3516746411483256,
      "grad_norm": 0.6491839671223758,
      "learning_rate": 1.6109047566860997e-06,
      "loss": 0.3798,
      "num_tokens": 285738915.0,
      "step": 1401
    },
    {
      "epoch": 3.354066985645933,
      "grad_norm": 0.7049883361490701,
      "learning_rate": 1.6065246234561137e-06,
      "loss": 0.6441,
      "num_tokens": 285928806.0,
      "step": 1402
    },
    {
      "epoch": 3.3564593301435406,
      "grad_norm": 0.6036528727339665,
      "learning_rate": 1.602159114388896e-06,
      "loss": 0.4551,
      "num_tokens": 286166116.0,
      "step": 1403
    },
    {
      "epoch": 3.3588516746411483,
      "grad_norm": 0.6062845296574271,
      "learning_rate": 1.5978082458816032e-06,
      "loss": 0.5571,
      "num_tokens": 286423925.0,
      "step": 1404
    },
    {
      "epoch": 3.361244019138756,
      "grad_norm": 0.6306014380075803,
      "learning_rate": 1.593472034276396e-06,
      "loss": 0.4419,
      "num_tokens": 286624999.0,
      "step": 1405
    },
    {
      "epoch": 3.3636363636363638,
      "grad_norm": 0.7035613097127984,
      "learning_rate": 1.5891504958603867e-06,
      "loss": 0.4917,
      "num_tokens": 286794117.0,
      "step": 1406
    },
    {
      "epoch": 3.3660287081339715,
      "grad_norm": 0.665456708111723,
      "learning_rate": 1.584843646865571e-06,
      "loss": 0.5867,
      "num_tokens": 287043237.0,
      "step": 1407
    },
    {
      "epoch": 3.3684210526315788,
      "grad_norm": 0.6523486514014031,
      "learning_rate": 1.5805515034687725e-06,
      "loss": 0.5675,
      "num_tokens": 287262346.0,
      "step": 1408
    },
    {
      "epoch": 3.3708133971291865,
      "grad_norm": 0.6800989600061507,
      "learning_rate": 1.5762740817915783e-06,
      "loss": 0.3928,
      "num_tokens": 287435821.0,
      "step": 1409
    },
    {
      "epoch": 3.373205741626794,
      "grad_norm": 0.6085689835831343,
      "learning_rate": 1.5720113979002805e-06,
      "loss": 0.3708,
      "num_tokens": 287621476.0,
      "step": 1410
    },
    {
      "epoch": 3.375598086124402,
      "grad_norm": 0.7554742958605225,
      "learning_rate": 1.5677634678058146e-06,
      "loss": 0.4084,
      "num_tokens": 287759069.0,
      "step": 1411
    },
    {
      "epoch": 3.3779904306220097,
      "grad_norm": 0.6596430075419027,
      "learning_rate": 1.5635303074636979e-06,
      "loss": 0.5614,
      "num_tokens": 287942413.0,
      "step": 1412
    },
    {
      "epoch": 3.3803827751196174,
      "grad_norm": 0.6869062135938377,
      "learning_rate": 1.5593119327739773e-06,
      "loss": 0.5973,
      "num_tokens": 288154785.0,
      "step": 1413
    },
    {
      "epoch": 3.382775119617225,
      "grad_norm": 0.6454421371119977,
      "learning_rate": 1.5551083595811551e-06,
      "loss": 0.5079,
      "num_tokens": 288343799.0,
      "step": 1414
    },
    {
      "epoch": 3.3851674641148324,
      "grad_norm": 0.5860523042624308,
      "learning_rate": 1.5509196036741466e-06,
      "loss": 0.3886,
      "num_tokens": 288535767.0,
      "step": 1415
    },
    {
      "epoch": 3.38755980861244,
      "grad_norm": 0.6020591732561443,
      "learning_rate": 1.5467456807862063e-06,
      "loss": 0.4034,
      "num_tokens": 288737701.0,
      "step": 1416
    },
    {
      "epoch": 3.389952153110048,
      "grad_norm": 0.6740634893238907,
      "learning_rate": 1.5425866065948797e-06,
      "loss": 0.5953,
      "num_tokens": 288922136.0,
      "step": 1417
    },
    {
      "epoch": 3.3923444976076556,
      "grad_norm": 0.6082129324791387,
      "learning_rate": 1.5384423967219347e-06,
      "loss": 0.397,
      "num_tokens": 289111549.0,
      "step": 1418
    },
    {
      "epoch": 3.3947368421052633,
      "grad_norm": 0.6587824796591607,
      "learning_rate": 1.5343130667333116e-06,
      "loss": 0.5051,
      "num_tokens": 289319059.0,
      "step": 1419
    },
    {
      "epoch": 3.397129186602871,
      "grad_norm": 0.6464490795588403,
      "learning_rate": 1.5301986321390588e-06,
      "loss": 0.4862,
      "num_tokens": 289517343.0,
      "step": 1420
    },
    {
      "epoch": 3.3995215311004783,
      "grad_norm": 0.6097300879288532,
      "learning_rate": 1.5260991083932772e-06,
      "loss": 0.4842,
      "num_tokens": 289715665.0,
      "step": 1421
    },
    {
      "epoch": 3.401913875598086,
      "grad_norm": 0.6548853735664123,
      "learning_rate": 1.5220145108940614e-06,
      "loss": 0.4464,
      "num_tokens": 289876828.0,
      "step": 1422
    },
    {
      "epoch": 3.4043062200956937,
      "grad_norm": 0.5754235894763253,
      "learning_rate": 1.5179448549834427e-06,
      "loss": 0.3191,
      "num_tokens": 290072817.0,
      "step": 1423
    },
    {
      "epoch": 3.4066985645933014,
      "grad_norm": 0.6774434347134051,
      "learning_rate": 1.5138901559473295e-06,
      "loss": 0.5724,
      "num_tokens": 290318208.0,
      "step": 1424
    },
    {
      "epoch": 3.409090909090909,
      "grad_norm": 0.6309179267047327,
      "learning_rate": 1.5098504290154522e-06,
      "loss": 0.5446,
      "num_tokens": 290561443.0,
      "step": 1425
    },
    {
      "epoch": 3.411483253588517,
      "grad_norm": 0.7245178380869103,
      "learning_rate": 1.5058256893613052e-06,
      "loss": 0.529,
      "num_tokens": 290728218.0,
      "step": 1426
    },
    {
      "epoch": 3.4138755980861246,
      "grad_norm": 0.6793714067868013,
      "learning_rate": 1.501815952102086e-06,
      "loss": 0.4026,
      "num_tokens": 290879223.0,
      "step": 1427
    },
    {
      "epoch": 3.416267942583732,
      "grad_norm": 0.677049622272289,
      "learning_rate": 1.4978212322986494e-06,
      "loss": 0.4234,
      "num_tokens": 291044967.0,
      "step": 1428
    },
    {
      "epoch": 3.4186602870813396,
      "grad_norm": 0.5891529086179643,
      "learning_rate": 1.4938415449554347e-06,
      "loss": 0.3909,
      "num_tokens": 291244108.0,
      "step": 1429
    },
    {
      "epoch": 3.4210526315789473,
      "grad_norm": 0.7680086416040134,
      "learning_rate": 1.489876905020427e-06,
      "loss": 0.5968,
      "num_tokens": 291422309.0,
      "step": 1430
    },
    {
      "epoch": 3.423444976076555,
      "grad_norm": 0.60541747237417,
      "learning_rate": 1.4859273273850853e-06,
      "loss": 0.4644,
      "num_tokens": 291607211.0,
      "step": 1431
    },
    {
      "epoch": 3.425837320574163,
      "grad_norm": 0.5129408106770269,
      "learning_rate": 1.481992826884297e-06,
      "loss": 0.4663,
      "num_tokens": 291847315.0,
      "step": 1432
    },
    {
      "epoch": 3.4282296650717705,
      "grad_norm": 0.6418602073158258,
      "learning_rate": 1.4780734182963186e-06,
      "loss": 0.3039,
      "num_tokens": 292003446.0,
      "step": 1433
    },
    {
      "epoch": 3.430622009569378,
      "grad_norm": 0.5414698213649558,
      "learning_rate": 1.474169116342719e-06,
      "loss": 0.4259,
      "num_tokens": 292221228.0,
      "step": 1434
    },
    {
      "epoch": 3.4330143540669855,
      "grad_norm": 0.5861235141143475,
      "learning_rate": 1.4702799356883276e-06,
      "loss": 0.4508,
      "num_tokens": 292392595.0,
      "step": 1435
    },
    {
      "epoch": 3.4354066985645932,
      "grad_norm": 0.6492147392943652,
      "learning_rate": 1.4664058909411743e-06,
      "loss": 0.5558,
      "num_tokens": 292616491.0,
      "step": 1436
    },
    {
      "epoch": 3.437799043062201,
      "grad_norm": 0.5867621281367356,
      "learning_rate": 1.4625469966524402e-06,
      "loss": 0.3956,
      "num_tokens": 292811254.0,
      "step": 1437
    },
    {
      "epoch": 3.4401913875598087,
      "grad_norm": 0.6082196218062246,
      "learning_rate": 1.4587032673163986e-06,
      "loss": 0.3441,
      "num_tokens": 293001343.0,
      "step": 1438
    },
    {
      "epoch": 3.4425837320574164,
      "grad_norm": 0.6414375078998078,
      "learning_rate": 1.454874717370363e-06,
      "loss": 0.5404,
      "num_tokens": 293221621.0,
      "step": 1439
    },
    {
      "epoch": 3.444976076555024,
      "grad_norm": 0.613280918544221,
      "learning_rate": 1.4510613611946317e-06,
      "loss": 0.5266,
      "num_tokens": 293420371.0,
      "step": 1440
    },
    {
      "epoch": 3.4473684210526314,
      "grad_norm": 0.687184407411569,
      "learning_rate": 1.4472632131124343e-06,
      "loss": 0.5471,
      "num_tokens": 293613114.0,
      "step": 1441
    },
    {
      "epoch": 3.449760765550239,
      "grad_norm": 0.7132696555661605,
      "learning_rate": 1.4434802873898774e-06,
      "loss": 0.6045,
      "num_tokens": 293828237.0,
      "step": 1442
    },
    {
      "epoch": 3.452153110047847,
      "grad_norm": 0.663988889436373,
      "learning_rate": 1.4397125982358924e-06,
      "loss": 0.3951,
      "num_tokens": 294030397.0,
      "step": 1443
    },
    {
      "epoch": 3.4545454545454546,
      "grad_norm": 0.6111538223995584,
      "learning_rate": 1.435960159802179e-06,
      "loss": 0.5118,
      "num_tokens": 294266833.0,
      "step": 1444
    },
    {
      "epoch": 3.4569377990430623,
      "grad_norm": 0.7058152631528316,
      "learning_rate": 1.4322229861831558e-06,
      "loss": 0.5049,
      "num_tokens": 294451025.0,
      "step": 1445
    },
    {
      "epoch": 3.45933014354067,
      "grad_norm": 0.5699602285006586,
      "learning_rate": 1.4285010914159065e-06,
      "loss": 0.5684,
      "num_tokens": 294672913.0,
      "step": 1446
    },
    {
      "epoch": 3.4617224880382773,
      "grad_norm": 0.5623025715207065,
      "learning_rate": 1.4247944894801236e-06,
      "loss": 0.5162,
      "num_tokens": 294903672.0,
      "step": 1447
    },
    {
      "epoch": 3.464114832535885,
      "grad_norm": 0.712647341968248,
      "learning_rate": 1.4211031942980613e-06,
      "loss": 0.5999,
      "num_tokens": 295087849.0,
      "step": 1448
    },
    {
      "epoch": 3.4665071770334928,
      "grad_norm": 0.6172051693085459,
      "learning_rate": 1.4174272197344793e-06,
      "loss": 0.4612,
      "num_tokens": 295287798.0,
      "step": 1449
    },
    {
      "epoch": 3.4688995215311005,
      "grad_norm": 0.6311358310811482,
      "learning_rate": 1.4137665795965927e-06,
      "loss": 0.5761,
      "num_tokens": 295492370.0,
      "step": 1450
    },
    {
      "epoch": 3.471291866028708,
      "grad_norm": 0.7251245358729828,
      "learning_rate": 1.4101212876340187e-06,
      "loss": 0.5706,
      "num_tokens": 295672740.0,
      "step": 1451
    },
    {
      "epoch": 3.473684210526316,
      "grad_norm": 0.560049433607437,
      "learning_rate": 1.4064913575387251e-06,
      "loss": 0.4304,
      "num_tokens": 295869423.0,
      "step": 1452
    },
    {
      "epoch": 3.4760765550239237,
      "grad_norm": 0.6974882287107438,
      "learning_rate": 1.4028768029449824e-06,
      "loss": 0.5892,
      "num_tokens": 296062015.0,
      "step": 1453
    },
    {
      "epoch": 3.478468899521531,
      "grad_norm": 0.6545967283254652,
      "learning_rate": 1.399277637429306e-06,
      "loss": 0.5022,
      "num_tokens": 296274972.0,
      "step": 1454
    },
    {
      "epoch": 3.4808612440191387,
      "grad_norm": 0.8022540603329968,
      "learning_rate": 1.3956938745104109e-06,
      "loss": 0.5512,
      "num_tokens": 296429164.0,
      "step": 1455
    },
    {
      "epoch": 3.4832535885167464,
      "grad_norm": 0.7082878808081615,
      "learning_rate": 1.3921255276491586e-06,
      "loss": 0.5427,
      "num_tokens": 296614072.0,
      "step": 1456
    },
    {
      "epoch": 3.485645933014354,
      "grad_norm": 0.7641750032011092,
      "learning_rate": 1.3885726102485073e-06,
      "loss": 0.6203,
      "num_tokens": 296769349.0,
      "step": 1457
    },
    {
      "epoch": 3.488038277511962,
      "grad_norm": 0.7398101140526424,
      "learning_rate": 1.3850351356534592e-06,
      "loss": 0.5137,
      "num_tokens": 296956478.0,
      "step": 1458
    },
    {
      "epoch": 3.4904306220095696,
      "grad_norm": 0.6930674531073416,
      "learning_rate": 1.3815131171510138e-06,
      "loss": 0.7111,
      "num_tokens": 297202360.0,
      "step": 1459
    },
    {
      "epoch": 3.492822966507177,
      "grad_norm": 0.7104167967522225,
      "learning_rate": 1.3780065679701163e-06,
      "loss": 0.4507,
      "num_tokens": 297368352.0,
      "step": 1460
    },
    {
      "epoch": 3.4952153110047846,
      "grad_norm": 0.6783264118382551,
      "learning_rate": 1.3745155012816083e-06,
      "loss": 0.591,
      "num_tokens": 297571077.0,
      "step": 1461
    },
    {
      "epoch": 3.4976076555023923,
      "grad_norm": 0.6499466582199835,
      "learning_rate": 1.3710399301981774e-06,
      "loss": 0.3173,
      "num_tokens": 297713503.0,
      "step": 1462
    },
    {
      "epoch": 3.5,
      "grad_norm": 0.6482663233430564,
      "learning_rate": 1.3675798677743094e-06,
      "loss": 0.425,
      "num_tokens": 297877390.0,
      "step": 1463
    },
    {
      "epoch": 3.5023923444976077,
      "grad_norm": 0.7096180478132679,
      "learning_rate": 1.364135327006237e-06,
      "loss": 0.6449,
      "num_tokens": 298080783.0,
      "step": 1464
    },
    {
      "epoch": 3.5047846889952154,
      "grad_norm": 0.820714681833925,
      "learning_rate": 1.3607063208318944e-06,
      "loss": 0.4437,
      "num_tokens": 298294526.0,
      "step": 1465
    },
    {
      "epoch": 3.507177033492823,
      "grad_norm": 0.5435495568523284,
      "learning_rate": 1.3572928621308662e-06,
      "loss": 0.4597,
      "num_tokens": 298534343.0,
      "step": 1466
    },
    {
      "epoch": 3.509569377990431,
      "grad_norm": 0.7369001406669367,
      "learning_rate": 1.3538949637243382e-06,
      "loss": 0.5622,
      "num_tokens": 298709480.0,
      "step": 1467
    },
    {
      "epoch": 3.511961722488038,
      "grad_norm": 0.6616177825442939,
      "learning_rate": 1.3505126383750528e-06,
      "loss": 0.5391,
      "num_tokens": 298918347.0,
      "step": 1468
    },
    {
      "epoch": 3.514354066985646,
      "grad_norm": 0.5800915812997612,
      "learning_rate": 1.3471458987872574e-06,
      "loss": 0.5072,
      "num_tokens": 299167577.0,
      "step": 1469
    },
    {
      "epoch": 3.5167464114832536,
      "grad_norm": 0.5776807179015478,
      "learning_rate": 1.3437947576066593e-06,
      "loss": 0.3321,
      "num_tokens": 299349965.0,
      "step": 1470
    },
    {
      "epoch": 3.5191387559808613,
      "grad_norm": 0.5576868780501636,
      "learning_rate": 1.3404592274203752e-06,
      "loss": 0.5429,
      "num_tokens": 299589593.0,
      "step": 1471
    },
    {
      "epoch": 3.5215311004784686,
      "grad_norm": 0.7277519126409697,
      "learning_rate": 1.3371393207568894e-06,
      "loss": 0.5165,
      "num_tokens": 299747218.0,
      "step": 1472
    },
    {
      "epoch": 3.5239234449760763,
      "grad_norm": 0.6846297716168203,
      "learning_rate": 1.3338350500859988e-06,
      "loss": 0.4876,
      "num_tokens": 299950882.0,
      "step": 1473
    },
    {
      "epoch": 3.526315789473684,
      "grad_norm": 0.5504630676093619,
      "learning_rate": 1.330546427818773e-06,
      "loss": 0.5165,
      "num_tokens": 300172705.0,
      "step": 1474
    },
    {
      "epoch": 3.528708133971292,
      "grad_norm": 0.6077413153293856,
      "learning_rate": 1.3272734663075054e-06,
      "loss": 0.5361,
      "num_tokens": 300408847.0,
      "step": 1475
    },
    {
      "epoch": 3.5311004784688995,
      "grad_norm": 0.6151817478419886,
      "learning_rate": 1.324016177845665e-06,
      "loss": 0.5336,
      "num_tokens": 300627309.0,
      "step": 1476
    },
    {
      "epoch": 3.5334928229665072,
      "grad_norm": 0.6657148941500219,
      "learning_rate": 1.3207745746678533e-06,
      "loss": 0.6072,
      "num_tokens": 300893676.0,
      "step": 1477
    },
    {
      "epoch": 3.535885167464115,
      "grad_norm": 0.5864974741212591,
      "learning_rate": 1.3175486689497558e-06,
      "loss": 0.4759,
      "num_tokens": 301119831.0,
      "step": 1478
    },
    {
      "epoch": 3.5382775119617227,
      "grad_norm": 0.669995482447796,
      "learning_rate": 1.3143384728080966e-06,
      "loss": 0.397,
      "num_tokens": 301288802.0,
      "step": 1479
    },
    {
      "epoch": 3.5406698564593304,
      "grad_norm": 0.6229464873082609,
      "learning_rate": 1.3111439983005952e-06,
      "loss": 0.6313,
      "num_tokens": 301518507.0,
      "step": 1480
    },
    {
      "epoch": 3.5430622009569377,
      "grad_norm": 0.6175779598804242,
      "learning_rate": 1.3079652574259187e-06,
      "loss": 0.4716,
      "num_tokens": 301730836.0,
      "step": 1481
    },
    {
      "epoch": 3.5454545454545454,
      "grad_norm": 0.6557616851250915,
      "learning_rate": 1.3048022621236382e-06,
      "loss": 0.5501,
      "num_tokens": 301944446.0,
      "step": 1482
    },
    {
      "epoch": 3.547846889952153,
      "grad_norm": 0.6431228077639217,
      "learning_rate": 1.301655024274182e-06,
      "loss": 0.4078,
      "num_tokens": 302125230.0,
      "step": 1483
    },
    {
      "epoch": 3.550239234449761,
      "grad_norm": 0.6165365783638824,
      "learning_rate": 1.2985235556987926e-06,
      "loss": 0.4014,
      "num_tokens": 302326057.0,
      "step": 1484
    },
    {
      "epoch": 3.5526315789473686,
      "grad_norm": 0.6022448407744763,
      "learning_rate": 1.2954078681594845e-06,
      "loss": 0.372,
      "num_tokens": 302506769.0,
      "step": 1485
    },
    {
      "epoch": 3.555023923444976,
      "grad_norm": 0.5938105239263732,
      "learning_rate": 1.2923079733589928e-06,
      "loss": 0.5537,
      "num_tokens": 302720143.0,
      "step": 1486
    },
    {
      "epoch": 3.5574162679425836,
      "grad_norm": 0.6873490122710798,
      "learning_rate": 1.2892238829407395e-06,
      "loss": 0.4151,
      "num_tokens": 302877685.0,
      "step": 1487
    },
    {
      "epoch": 3.5598086124401913,
      "grad_norm": 0.5681553037684327,
      "learning_rate": 1.2861556084887797e-06,
      "loss": 0.355,
      "num_tokens": 303118737.0,
      "step": 1488
    },
    {
      "epoch": 3.562200956937799,
      "grad_norm": 0.6948437599501222,
      "learning_rate": 1.2831031615277648e-06,
      "loss": 0.5261,
      "num_tokens": 303302096.0,
      "step": 1489
    },
    {
      "epoch": 3.5645933014354068,
      "grad_norm": 0.6419899066739717,
      "learning_rate": 1.2800665535228967e-06,
      "loss": 0.3954,
      "num_tokens": 303492631.0,
      "step": 1490
    },
    {
      "epoch": 3.5669856459330145,
      "grad_norm": 0.6901526120395478,
      "learning_rate": 1.2770457958798858e-06,
      "loss": 0.4241,
      "num_tokens": 303694151.0,
      "step": 1491
    },
    {
      "epoch": 3.569377990430622,
      "grad_norm": 0.6328577003617731,
      "learning_rate": 1.2740408999449055e-06,
      "loss": 0.6208,
      "num_tokens": 303941252.0,
      "step": 1492
    },
    {
      "epoch": 3.57177033492823,
      "grad_norm": 0.621056467226815,
      "learning_rate": 1.2710518770045535e-06,
      "loss": 0.367,
      "num_tokens": 304122539.0,
      "step": 1493
    },
    {
      "epoch": 3.574162679425837,
      "grad_norm": 0.6484720088529732,
      "learning_rate": 1.2680787382858068e-06,
      "loss": 0.5649,
      "num_tokens": 304340159.0,
      "step": 1494
    },
    {
      "epoch": 3.576555023923445,
      "grad_norm": 0.6765968336575626,
      "learning_rate": 1.2651214949559796e-06,
      "loss": 0.5138,
      "num_tokens": 304564017.0,
      "step": 1495
    },
    {
      "epoch": 3.5789473684210527,
      "grad_norm": 0.6151693720513433,
      "learning_rate": 1.2621801581226832e-06,
      "loss": 0.477,
      "num_tokens": 304782069.0,
      "step": 1496
    },
    {
      "epoch": 3.5813397129186604,
      "grad_norm": 0.6951947406065941,
      "learning_rate": 1.2592547388337797e-06,
      "loss": 0.5591,
      "num_tokens": 304964038.0,
      "step": 1497
    },
    {
      "epoch": 3.583732057416268,
      "grad_norm": 0.7379327952571767,
      "learning_rate": 1.25634524807735e-06,
      "loss": 0.5745,
      "num_tokens": 305144017.0,
      "step": 1498
    },
    {
      "epoch": 3.5861244019138754,
      "grad_norm": 0.6546876395903558,
      "learning_rate": 1.253451696781639e-06,
      "loss": 0.5772,
      "num_tokens": 305325548.0,
      "step": 1499
    },
    {
      "epoch": 3.588516746411483,
      "grad_norm": 0.5931090788351342,
      "learning_rate": 1.2505740958150287e-06,
      "loss": 0.5098,
      "num_tokens": 305545495.0,
      "step": 1500
    },
    {
      "epoch": 3.590909090909091,
      "grad_norm": 0.6700019687712389,
      "learning_rate": 1.2477124559859837e-06,
      "loss": 0.6396,
      "num_tokens": 305763102.0,
      "step": 1501
    },
    {
      "epoch": 3.5933014354066986,
      "grad_norm": 0.528208222358471,
      "learning_rate": 1.2448667880430252e-06,
      "loss": 0.4368,
      "num_tokens": 306006879.0,
      "step": 1502
    },
    {
      "epoch": 3.5956937799043063,
      "grad_norm": 0.5929577605264226,
      "learning_rate": 1.242037102674675e-06,
      "loss": 0.4719,
      "num_tokens": 306202979.0,
      "step": 1503
    },
    {
      "epoch": 3.598086124401914,
      "grad_norm": 0.5935852989033888,
      "learning_rate": 1.2392234105094301e-06,
      "loss": 0.5019,
      "num_tokens": 306423290.0,
      "step": 1504
    },
    {
      "epoch": 3.6004784688995217,
      "grad_norm": 0.6272016449415893,
      "learning_rate": 1.236425722115711e-06,
      "loss": 0.6179,
      "num_tokens": 306674700.0,
      "step": 1505
    },
    {
      "epoch": 3.6028708133971294,
      "grad_norm": 0.6795819406741945,
      "learning_rate": 1.2336440480018292e-06,
      "loss": 0.4432,
      "num_tokens": 306864232.0,
      "step": 1506
    },
    {
      "epoch": 3.6052631578947367,
      "grad_norm": 0.6346248295282361,
      "learning_rate": 1.2308783986159451e-06,
      "loss": 0.6093,
      "num_tokens": 307116714.0,
      "step": 1507
    },
    {
      "epoch": 3.6076555023923444,
      "grad_norm": 0.5839403967153011,
      "learning_rate": 1.228128784346029e-06,
      "loss": 0.3204,
      "num_tokens": 307290174.0,
      "step": 1508
    },
    {
      "epoch": 3.610047846889952,
      "grad_norm": 0.5847207999870764,
      "learning_rate": 1.225395215519823e-06,
      "loss": 0.47,
      "num_tokens": 307496199.0,
      "step": 1509
    },
    {
      "epoch": 3.61244019138756,
      "grad_norm": 0.6125888173508722,
      "learning_rate": 1.2226777024047985e-06,
      "loss": 0.393,
      "num_tokens": 307673185.0,
      "step": 1510
    },
    {
      "epoch": 3.6148325358851676,
      "grad_norm": 0.6713056636085881,
      "learning_rate": 1.2199762552081254e-06,
      "loss": 0.5381,
      "num_tokens": 307901253.0,
      "step": 1511
    },
    {
      "epoch": 3.617224880382775,
      "grad_norm": 0.6509525755677749,
      "learning_rate": 1.2172908840766242e-06,
      "loss": 0.2484,
      "num_tokens": 308052165.0,
      "step": 1512
    },
    {
      "epoch": 3.6196172248803826,
      "grad_norm": 0.6804638610041888,
      "learning_rate": 1.2146215990967363e-06,
      "loss": 0.6382,
      "num_tokens": 308286709.0,
      "step": 1513
    },
    {
      "epoch": 3.6220095693779903,
      "grad_norm": 0.5726101216910606,
      "learning_rate": 1.2119684102944782e-06,
      "loss": 0.5203,
      "num_tokens": 308518482.0,
      "step": 1514
    },
    {
      "epoch": 3.624401913875598,
      "grad_norm": 0.7457351396502082,
      "learning_rate": 1.2093313276354139e-06,
      "loss": 0.5756,
      "num_tokens": 308701930.0,
      "step": 1515
    },
    {
      "epoch": 3.626794258373206,
      "grad_norm": 0.618254544803711,
      "learning_rate": 1.2067103610246046e-06,
      "loss": 0.5782,
      "num_tokens": 308920981.0,
      "step": 1516
    },
    {
      "epoch": 3.6291866028708135,
      "grad_norm": 0.6140635887344684,
      "learning_rate": 1.2041055203065852e-06,
      "loss": 0.5024,
      "num_tokens": 309109111.0,
      "step": 1517
    },
    {
      "epoch": 3.6315789473684212,
      "grad_norm": 0.5547321147206474,
      "learning_rate": 1.2015168152653159e-06,
      "loss": 0.3661,
      "num_tokens": 309314804.0,
      "step": 1518
    },
    {
      "epoch": 3.633971291866029,
      "grad_norm": 0.7611178589551898,
      "learning_rate": 1.198944255624152e-06,
      "loss": 0.622,
      "num_tokens": 309496156.0,
      "step": 1519
    },
    {
      "epoch": 3.6363636363636362,
      "grad_norm": 0.6207298025870296,
      "learning_rate": 1.1963878510458058e-06,
      "loss": 0.5098,
      "num_tokens": 309713174.0,
      "step": 1520
    },
    {
      "epoch": 3.638755980861244,
      "grad_norm": 0.5547121537829205,
      "learning_rate": 1.1938476111323094e-06,
      "loss": 0.4402,
      "num_tokens": 309921616.0,
      "step": 1521
    },
    {
      "epoch": 3.6411483253588517,
      "grad_norm": 0.7199324671101245,
      "learning_rate": 1.1913235454249807e-06,
      "loss": 0.3728,
      "num_tokens": 310061923.0,
      "step": 1522
    },
    {
      "epoch": 3.6435406698564594,
      "grad_norm": 0.5715673360964941,
      "learning_rate": 1.1888156634043825e-06,
      "loss": 0.5396,
      "num_tokens": 310282339.0,
      "step": 1523
    },
    {
      "epoch": 3.645933014354067,
      "grad_norm": 0.4864991883545776,
      "learning_rate": 1.1863239744902962e-06,
      "loss": 0.3361,
      "num_tokens": 310516447.0,
      "step": 1524
    },
    {
      "epoch": 3.6483253588516744,
      "grad_norm": 0.7242060944351417,
      "learning_rate": 1.1838484880416746e-06,
      "loss": 0.6207,
      "num_tokens": 310711662.0,
      "step": 1525
    },
    {
      "epoch": 3.650717703349282,
      "grad_norm": 0.6919148397543631,
      "learning_rate": 1.181389213356619e-06,
      "loss": 0.5954,
      "num_tokens": 310902943.0,
      "step": 1526
    },
    {
      "epoch": 3.65311004784689,
      "grad_norm": 0.7059381091150269,
      "learning_rate": 1.1789461596723325e-06,
      "loss": 0.4793,
      "num_tokens": 311096238.0,
      "step": 1527
    },
    {
      "epoch": 3.6555023923444976,
      "grad_norm": 0.7806938857436164,
      "learning_rate": 1.1765193361650964e-06,
      "loss": 0.6368,
      "num_tokens": 311260692.0,
      "step": 1528
    },
    {
      "epoch": 3.6578947368421053,
      "grad_norm": 0.621657260828602,
      "learning_rate": 1.1741087519502246e-06,
      "loss": 0.328,
      "num_tokens": 311444390.0,
      "step": 1529
    },
    {
      "epoch": 3.660287081339713,
      "grad_norm": 0.6199793090742324,
      "learning_rate": 1.171714416082041e-06,
      "loss": 0.4227,
      "num_tokens": 311632736.0,
      "step": 1530
    },
    {
      "epoch": 3.6626794258373208,
      "grad_norm": 0.7466730988608571,
      "learning_rate": 1.1693363375538352e-06,
      "loss": 0.4312,
      "num_tokens": 311787085.0,
      "step": 1531
    },
    {
      "epoch": 3.6650717703349285,
      "grad_norm": 0.706574319299492,
      "learning_rate": 1.166974525297834e-06,
      "loss": 0.4932,
      "num_tokens": 311960841.0,
      "step": 1532
    },
    {
      "epoch": 3.6674641148325358,
      "grad_norm": 0.6489663598451897,
      "learning_rate": 1.1646289881851686e-06,
      "loss": 0.4557,
      "num_tokens": 312140508.0,
      "step": 1533
    },
    {
      "epoch": 3.6698564593301435,
      "grad_norm": 0.5629965502266332,
      "learning_rate": 1.1622997350258386e-06,
      "loss": 0.3826,
      "num_tokens": 312368815.0,
      "step": 1534
    },
    {
      "epoch": 3.672248803827751,
      "grad_norm": 0.7502937179319551,
      "learning_rate": 1.1599867745686797e-06,
      "loss": 0.4459,
      "num_tokens": 312510042.0,
      "step": 1535
    },
    {
      "epoch": 3.674641148325359,
      "grad_norm": 0.6450025468017979,
      "learning_rate": 1.157690115501331e-06,
      "loss": 0.5155,
      "num_tokens": 312753628.0,
      "step": 1536
    },
    {
      "epoch": 3.6770334928229667,
      "grad_norm": 0.6516806356437004,
      "learning_rate": 1.1554097664502042e-06,
      "loss": 0.6561,
      "num_tokens": 313017066.0,
      "step": 1537
    },
    {
      "epoch": 3.679425837320574,
      "grad_norm": 0.5838091230332593,
      "learning_rate": 1.1531457359804453e-06,
      "loss": 0.5613,
      "num_tokens": 313258011.0,
      "step": 1538
    },
    {
      "epoch": 3.6818181818181817,
      "grad_norm": 0.6930340123423862,
      "learning_rate": 1.1508980325959124e-06,
      "loss": 0.5582,
      "num_tokens": 313462213.0,
      "step": 1539
    },
    {
      "epoch": 3.6842105263157894,
      "grad_norm": 0.6056783363354274,
      "learning_rate": 1.1486666647391325e-06,
      "loss": 0.4607,
      "num_tokens": 313715622.0,
      "step": 1540
    },
    {
      "epoch": 3.686602870813397,
      "grad_norm": 0.5370198813826438,
      "learning_rate": 1.1464516407912792e-06,
      "loss": 0.4582,
      "num_tokens": 313970687.0,
      "step": 1541
    },
    {
      "epoch": 3.688995215311005,
      "grad_norm": 0.6121865084573093,
      "learning_rate": 1.1442529690721341e-06,
      "loss": 0.6653,
      "num_tokens": 314228905.0,
      "step": 1542
    },
    {
      "epoch": 3.6913875598086126,
      "grad_norm": 0.6261892366272805,
      "learning_rate": 1.142070657840062e-06,
      "loss": 0.4181,
      "num_tokens": 314417597.0,
      "step": 1543
    },
    {
      "epoch": 3.6937799043062203,
      "grad_norm": 0.5999298045009195,
      "learning_rate": 1.1399047152919733e-06,
      "loss": 0.4485,
      "num_tokens": 314612517.0,
      "step": 1544
    },
    {
      "epoch": 3.696172248803828,
      "grad_norm": 0.6510228815523412,
      "learning_rate": 1.1377551495632985e-06,
      "loss": 0.4576,
      "num_tokens": 314810459.0,
      "step": 1545
    },
    {
      "epoch": 3.6985645933014353,
      "grad_norm": 0.7103858911919039,
      "learning_rate": 1.1356219687279554e-06,
      "loss": 0.497,
      "num_tokens": 314975980.0,
      "step": 1546
    },
    {
      "epoch": 3.700956937799043,
      "grad_norm": 0.6483471228722542,
      "learning_rate": 1.1335051807983188e-06,
      "loss": 0.3976,
      "num_tokens": 315153202.0,
      "step": 1547
    },
    {
      "epoch": 3.7033492822966507,
      "grad_norm": 0.5767066645524299,
      "learning_rate": 1.1314047937251894e-06,
      "loss": 0.4727,
      "num_tokens": 315378174.0,
      "step": 1548
    },
    {
      "epoch": 3.7057416267942584,
      "grad_norm": 0.6178224515003014,
      "learning_rate": 1.1293208153977677e-06,
      "loss": 0.4304,
      "num_tokens": 315577510.0,
      "step": 1549
    },
    {
      "epoch": 3.708133971291866,
      "grad_norm": 0.6414850819255512,
      "learning_rate": 1.1272532536436205e-06,
      "loss": 0.3087,
      "num_tokens": 315722493.0,
      "step": 1550
    },
    {
      "epoch": 3.7105263157894735,
      "grad_norm": 0.5916056531201479,
      "learning_rate": 1.1252021162286513e-06,
      "loss": 0.4199,
      "num_tokens": 315915892.0,
      "step": 1551
    },
    {
      "epoch": 3.712918660287081,
      "grad_norm": 0.638094656338588,
      "learning_rate": 1.1231674108570756e-06,
      "loss": 0.5788,
      "num_tokens": 316148867.0,
      "step": 1552
    },
    {
      "epoch": 3.715311004784689,
      "grad_norm": 0.7083497582287557,
      "learning_rate": 1.121149145171386e-06,
      "loss": 0.583,
      "num_tokens": 316320260.0,
      "step": 1553
    },
    {
      "epoch": 3.7177033492822966,
      "grad_norm": 0.650630130615892,
      "learning_rate": 1.1191473267523297e-06,
      "loss": 0.5426,
      "num_tokens": 316501062.0,
      "step": 1554
    },
    {
      "epoch": 3.7200956937799043,
      "grad_norm": 0.6113182049236018,
      "learning_rate": 1.1171619631188733e-06,
      "loss": 0.5281,
      "num_tokens": 316760839.0,
      "step": 1555
    },
    {
      "epoch": 3.722488038277512,
      "grad_norm": 0.6124521857276314,
      "learning_rate": 1.1151930617281814e-06,
      "loss": 0.4915,
      "num_tokens": 316978471.0,
      "step": 1556
    },
    {
      "epoch": 3.72488038277512,
      "grad_norm": 0.5855301464032153,
      "learning_rate": 1.1132406299755828e-06,
      "loss": 0.3729,
      "num_tokens": 317175826.0,
      "step": 1557
    },
    {
      "epoch": 3.7272727272727275,
      "grad_norm": 0.4900955113250577,
      "learning_rate": 1.111304675194547e-06,
      "loss": 0.3394,
      "num_tokens": 317410638.0,
      "step": 1558
    },
    {
      "epoch": 3.729665071770335,
      "grad_norm": 0.6800914502081618,
      "learning_rate": 1.1093852046566534e-06,
      "loss": 0.5626,
      "num_tokens": 317596615.0,
      "step": 1559
    },
    {
      "epoch": 3.7320574162679425,
      "grad_norm": 0.580145555361245,
      "learning_rate": 1.1074822255715664e-06,
      "loss": 0.3454,
      "num_tokens": 317798272.0,
      "step": 1560
    },
    {
      "epoch": 3.7344497607655502,
      "grad_norm": 0.586689431375617,
      "learning_rate": 1.1055957450870066e-06,
      "loss": 0.4209,
      "num_tokens": 318011177.0,
      "step": 1561
    },
    {
      "epoch": 3.736842105263158,
      "grad_norm": 0.5508088873557061,
      "learning_rate": 1.103725770288726e-06,
      "loss": 0.3773,
      "num_tokens": 318216392.0,
      "step": 1562
    },
    {
      "epoch": 3.7392344497607657,
      "grad_norm": 0.6811545208859248,
      "learning_rate": 1.1018723082004782e-06,
      "loss": 0.5659,
      "num_tokens": 318416875.0,
      "step": 1563
    },
    {
      "epoch": 3.741626794258373,
      "grad_norm": 0.72071013278369,
      "learning_rate": 1.1000353657839962e-06,
      "loss": 0.5061,
      "num_tokens": 318594496.0,
      "step": 1564
    },
    {
      "epoch": 3.7440191387559807,
      "grad_norm": 0.7606517545104856,
      "learning_rate": 1.0982149499389617e-06,
      "loss": 0.6404,
      "num_tokens": 318787094.0,
      "step": 1565
    },
    {
      "epoch": 3.7464114832535884,
      "grad_norm": 0.5877892564705959,
      "learning_rate": 1.0964110675029828e-06,
      "loss": 0.4668,
      "num_tokens": 318981017.0,
      "step": 1566
    },
    {
      "epoch": 3.748803827751196,
      "grad_norm": 0.5670748324028697,
      "learning_rate": 1.0946237252515671e-06,
      "loss": 0.4134,
      "num_tokens": 319161348.0,
      "step": 1567
    },
    {
      "epoch": 3.751196172248804,
      "grad_norm": 0.6047984846668623,
      "learning_rate": 1.092852929898095e-06,
      "loss": 0.5675,
      "num_tokens": 319404004.0,
      "step": 1568
    },
    {
      "epoch": 3.7535885167464116,
      "grad_norm": 0.774410044947522,
      "learning_rate": 1.0910986880937972e-06,
      "loss": 0.6418,
      "num_tokens": 319566164.0,
      "step": 1569
    },
    {
      "epoch": 3.7559808612440193,
      "grad_norm": 0.6901217036526389,
      "learning_rate": 1.0893610064277268e-06,
      "loss": 0.5677,
      "num_tokens": 319784161.0,
      "step": 1570
    },
    {
      "epoch": 3.758373205741627,
      "grad_norm": 0.601635981880848,
      "learning_rate": 1.0876398914267358e-06,
      "loss": 0.497,
      "num_tokens": 320014206.0,
      "step": 1571
    },
    {
      "epoch": 3.7607655502392343,
      "grad_norm": 0.5909062617346802,
      "learning_rate": 1.0859353495554517e-06,
      "loss": 0.5804,
      "num_tokens": 320272986.0,
      "step": 1572
    },
    {
      "epoch": 3.763157894736842,
      "grad_norm": 0.6020618990919157,
      "learning_rate": 1.0842473872162517e-06,
      "loss": 0.2992,
      "num_tokens": 320400622.0,
      "step": 1573
    },
    {
      "epoch": 3.7655502392344498,
      "grad_norm": 0.5821329545334476,
      "learning_rate": 1.0825760107492384e-06,
      "loss": 0.5617,
      "num_tokens": 320646872.0,
      "step": 1574
    },
    {
      "epoch": 3.7679425837320575,
      "grad_norm": 0.491139430544993,
      "learning_rate": 1.0809212264322185e-06,
      "loss": 0.4149,
      "num_tokens": 320920941.0,
      "step": 1575
    },
    {
      "epoch": 3.770334928229665,
      "grad_norm": 0.6017226268390595,
      "learning_rate": 1.0792830404806756e-06,
      "loss": 0.4497,
      "num_tokens": 321121661.0,
      "step": 1576
    },
    {
      "epoch": 3.7727272727272725,
      "grad_norm": 0.6654757497149889,
      "learning_rate": 1.0776614590477493e-06,
      "loss": 0.3742,
      "num_tokens": 321281938.0,
      "step": 1577
    },
    {
      "epoch": 3.77511961722488,
      "grad_norm": 0.5210291647393411,
      "learning_rate": 1.0760564882242122e-06,
      "loss": 0.3463,
      "num_tokens": 321483427.0,
      "step": 1578
    },
    {
      "epoch": 3.777511961722488,
      "grad_norm": 0.7436218210948605,
      "learning_rate": 1.0744681340384452e-06,
      "loss": 0.4782,
      "num_tokens": 321632939.0,
      "step": 1579
    },
    {
      "epoch": 3.7799043062200957,
      "grad_norm": 0.6545805936948292,
      "learning_rate": 1.0728964024564165e-06,
      "loss": 0.565,
      "num_tokens": 321812453.0,
      "step": 1580
    },
    {
      "epoch": 3.7822966507177034,
      "grad_norm": 0.4607908620988967,
      "learning_rate": 1.071341299381659e-06,
      "loss": 0.3077,
      "num_tokens": 321996465.0,
      "step": 1581
    },
    {
      "epoch": 3.784688995215311,
      "grad_norm": 0.5033378978914038,
      "learning_rate": 1.069802830655248e-06,
      "loss": 0.4661,
      "num_tokens": 322267207.0,
      "step": 1582
    },
    {
      "epoch": 3.787081339712919,
      "grad_norm": 0.6235361901036778,
      "learning_rate": 1.0682810020557762e-06,
      "loss": 0.5563,
      "num_tokens": 322476579.0,
      "step": 1583
    },
    {
      "epoch": 3.7894736842105265,
      "grad_norm": 0.5893893559866515,
      "learning_rate": 1.0667758192993393e-06,
      "loss": 0.4772,
      "num_tokens": 322683747.0,
      "step": 1584
    },
    {
      "epoch": 3.791866028708134,
      "grad_norm": 0.5758781713440533,
      "learning_rate": 1.0652872880395065e-06,
      "loss": 0.3445,
      "num_tokens": 322872058.0,
      "step": 1585
    },
    {
      "epoch": 3.7942583732057416,
      "grad_norm": 0.5335020907001772,
      "learning_rate": 1.063815413867305e-06,
      "loss": 0.3899,
      "num_tokens": 323127466.0,
      "step": 1586
    },
    {
      "epoch": 3.7966507177033493,
      "grad_norm": 0.6289334799509361,
      "learning_rate": 1.0623602023111953e-06,
      "loss": 0.5584,
      "num_tokens": 323326737.0,
      "step": 1587
    },
    {
      "epoch": 3.799043062200957,
      "grad_norm": 0.6211221471702003,
      "learning_rate": 1.0609216588370533e-06,
      "loss": 0.563,
      "num_tokens": 323569246.0,
      "step": 1588
    },
    {
      "epoch": 3.8014354066985647,
      "grad_norm": 0.5346284527419118,
      "learning_rate": 1.0594997888481473e-06,
      "loss": 0.4427,
      "num_tokens": 323829962.0,
      "step": 1589
    },
    {
      "epoch": 3.803827751196172,
      "grad_norm": 0.6105884009837914,
      "learning_rate": 1.058094597685119e-06,
      "loss": 0.4559,
      "num_tokens": 324028942.0,
      "step": 1590
    },
    {
      "epoch": 3.8062200956937797,
      "grad_norm": 0.5468954320316094,
      "learning_rate": 1.0567060906259646e-06,
      "loss": 0.4562,
      "num_tokens": 324256000.0,
      "step": 1591
    },
    {
      "epoch": 3.8086124401913874,
      "grad_norm": 0.684261812822514,
      "learning_rate": 1.055334272886012e-06,
      "loss": 0.5445,
      "num_tokens": 324434459.0,
      "step": 1592
    },
    {
      "epoch": 3.811004784688995,
      "grad_norm": 0.6526354730563445,
      "learning_rate": 1.0539791496179034e-06,
      "loss": 0.5026,
      "num_tokens": 324612716.0,
      "step": 1593
    },
    {
      "epoch": 3.813397129186603,
      "grad_norm": 0.808016517252592,
      "learning_rate": 1.052640725911575e-06,
      "loss": 0.4559,
      "num_tokens": 324755406.0,
      "step": 1594
    },
    {
      "epoch": 3.8157894736842106,
      "grad_norm": 0.6847153846248437,
      "learning_rate": 1.0513190067942405e-06,
      "loss": 0.5469,
      "num_tokens": 324945448.0,
      "step": 1595
    },
    {
      "epoch": 3.8181818181818183,
      "grad_norm": 0.5032406865472859,
      "learning_rate": 1.0500139972303667e-06,
      "loss": 0.3701,
      "num_tokens": 325177483.0,
      "step": 1596
    },
    {
      "epoch": 3.820574162679426,
      "grad_norm": 0.6356968965119758,
      "learning_rate": 1.048725702121661e-06,
      "loss": 0.6085,
      "num_tokens": 325386402.0,
      "step": 1597
    },
    {
      "epoch": 3.8229665071770333,
      "grad_norm": 0.5246325645094293,
      "learning_rate": 1.047454126307048e-06,
      "loss": 0.3515,
      "num_tokens": 325610116.0,
      "step": 1598
    },
    {
      "epoch": 3.825358851674641,
      "grad_norm": 0.5578264875928621,
      "learning_rate": 1.0461992745626539e-06,
      "loss": 0.385,
      "num_tokens": 325809437.0,
      "step": 1599
    },
    {
      "epoch": 3.827751196172249,
      "grad_norm": 0.5877031056698316,
      "learning_rate": 1.04496115160179e-06,
      "loss": 0.4319,
      "num_tokens": 326016149.0,
      "step": 1600
    },
    {
      "epoch": 3.8301435406698565,
      "grad_norm": 0.7046844882013413,
      "learning_rate": 1.0437397620749312e-06,
      "loss": 0.6778,
      "num_tokens": 326230889.0,
      "step": 1601
    },
    {
      "epoch": 3.8325358851674642,
      "grad_norm": 0.6223264724390841,
      "learning_rate": 1.0425351105697009e-06,
      "loss": 0.4787,
      "num_tokens": 326414427.0,
      "step": 1602
    },
    {
      "epoch": 3.8349282296650715,
      "grad_norm": 0.5851815349219407,
      "learning_rate": 1.0413472016108534e-06,
      "loss": 0.3812,
      "num_tokens": 326602846.0,
      "step": 1603
    },
    {
      "epoch": 3.8373205741626792,
      "grad_norm": 0.7261081213176126,
      "learning_rate": 1.040176039660257e-06,
      "loss": 0.6145,
      "num_tokens": 326792181.0,
      "step": 1604
    },
    {
      "epoch": 3.839712918660287,
      "grad_norm": 0.7167146631031324,
      "learning_rate": 1.0390216291168773e-06,
      "loss": 0.6231,
      "num_tokens": 326996430.0,
      "step": 1605
    },
    {
      "epoch": 3.8421052631578947,
      "grad_norm": 0.69933271543258,
      "learning_rate": 1.0378839743167598e-06,
      "loss": 0.4485,
      "num_tokens": 327146520.0,
      "step": 1606
    },
    {
      "epoch": 3.8444976076555024,
      "grad_norm": 0.6096694228023282,
      "learning_rate": 1.0367630795330143e-06,
      "loss": 0.4307,
      "num_tokens": 327368158.0,
      "step": 1607
    },
    {
      "epoch": 3.84688995215311,
      "grad_norm": 0.6284560072949098,
      "learning_rate": 1.0356589489757998e-06,
      "loss": 0.4751,
      "num_tokens": 327581854.0,
      "step": 1608
    },
    {
      "epoch": 3.849282296650718,
      "grad_norm": 0.6986455090501255,
      "learning_rate": 1.0345715867923069e-06,
      "loss": 0.5876,
      "num_tokens": 327758594.0,
      "step": 1609
    },
    {
      "epoch": 3.8516746411483256,
      "grad_norm": 0.6279890443357188,
      "learning_rate": 1.0335009970667424e-06,
      "loss": 0.452,
      "num_tokens": 328005472.0,
      "step": 1610
    },
    {
      "epoch": 3.854066985645933,
      "grad_norm": 0.5038022028163734,
      "learning_rate": 1.0324471838203165e-06,
      "loss": 0.3595,
      "num_tokens": 328237457.0,
      "step": 1611
    },
    {
      "epoch": 3.8564593301435406,
      "grad_norm": 0.6174001923472133,
      "learning_rate": 1.031410151011224e-06,
      "loss": 0.5795,
      "num_tokens": 328452327.0,
      "step": 1612
    },
    {
      "epoch": 3.8588516746411483,
      "grad_norm": 0.6609200384872597,
      "learning_rate": 1.0303899025346317e-06,
      "loss": 0.4913,
      "num_tokens": 328640854.0,
      "step": 1613
    },
    {
      "epoch": 3.861244019138756,
      "grad_norm": 0.6438960049381821,
      "learning_rate": 1.0293864422226636e-06,
      "loss": 0.4904,
      "num_tokens": 328846549.0,
      "step": 1614
    },
    {
      "epoch": 3.8636363636363638,
      "grad_norm": 0.6478362817028315,
      "learning_rate": 1.028399773844386e-06,
      "loss": 0.5504,
      "num_tokens": 329126737.0,
      "step": 1615
    },
    {
      "epoch": 3.866028708133971,
      "grad_norm": 0.5035472376476936,
      "learning_rate": 1.0274299011057936e-06,
      "loss": 0.4404,
      "num_tokens": 329338983.0,
      "step": 1616
    },
    {
      "epoch": 3.8684210526315788,
      "grad_norm": 0.7034481106903326,
      "learning_rate": 1.0264768276497958e-06,
      "loss": 0.5546,
      "num_tokens": 329546794.0,
      "step": 1617
    },
    {
      "epoch": 3.8708133971291865,
      "grad_norm": 0.631611590849145,
      "learning_rate": 1.0255405570562023e-06,
      "loss": 0.4272,
      "num_tokens": 329731632.0,
      "step": 1618
    },
    {
      "epoch": 3.873205741626794,
      "grad_norm": 0.6622179754285995,
      "learning_rate": 1.024621092841709e-06,
      "loss": 0.397,
      "num_tokens": 329926478.0,
      "step": 1619
    },
    {
      "epoch": 3.875598086124402,
      "grad_norm": 0.533233637209565,
      "learning_rate": 1.023718438459889e-06,
      "loss": 0.3988,
      "num_tokens": 330152207.0,
      "step": 1620
    },
    {
      "epoch": 3.8779904306220097,
      "grad_norm": 0.7196831072594635,
      "learning_rate": 1.0228325973011735e-06,
      "loss": 0.685,
      "num_tokens": 330344567.0,
      "step": 1621
    },
    {
      "epoch": 3.8803827751196174,
      "grad_norm": 0.6799866497463672,
      "learning_rate": 1.0219635726928432e-06,
      "loss": 0.5472,
      "num_tokens": 330553058.0,
      "step": 1622
    },
    {
      "epoch": 3.882775119617225,
      "grad_norm": 0.6452372012120255,
      "learning_rate": 1.0211113678990146e-06,
      "loss": 0.586,
      "num_tokens": 330770601.0,
      "step": 1623
    },
    {
      "epoch": 3.8851674641148324,
      "grad_norm": 0.7358669796048254,
      "learning_rate": 1.0202759861206288e-06,
      "loss": 0.5635,
      "num_tokens": 330949102.0,
      "step": 1624
    },
    {
      "epoch": 3.88755980861244,
      "grad_norm": 0.6914992060273657,
      "learning_rate": 1.0194574304954356e-06,
      "loss": 0.5109,
      "num_tokens": 331138604.0,
      "step": 1625
    },
    {
      "epoch": 3.889952153110048,
      "grad_norm": 0.6049799861182562,
      "learning_rate": 1.0186557040979883e-06,
      "loss": 0.4627,
      "num_tokens": 331323330.0,
      "step": 1626
    },
    {
      "epoch": 3.8923444976076556,
      "grad_norm": 0.5709437841289162,
      "learning_rate": 1.0178708099396263e-06,
      "loss": 0.5568,
      "num_tokens": 331588980.0,
      "step": 1627
    },
    {
      "epoch": 3.8947368421052633,
      "grad_norm": 0.6835800952526491,
      "learning_rate": 1.0171027509684653e-06,
      "loss": 0.4222,
      "num_tokens": 331795142.0,
      "step": 1628
    },
    {
      "epoch": 3.8971291866028706,
      "grad_norm": 0.6336779564251682,
      "learning_rate": 1.016351530069389e-06,
      "loss": 0.6262,
      "num_tokens": 332018389.0,
      "step": 1629
    },
    {
      "epoch": 3.8995215311004783,
      "grad_norm": 0.6138106246476915,
      "learning_rate": 1.0156171500640349e-06,
      "loss": 0.5274,
      "num_tokens": 332265698.0,
      "step": 1630
    },
    {
      "epoch": 3.901913875598086,
      "grad_norm": 0.6367531709575587,
      "learning_rate": 1.0148996137107847e-06,
      "loss": 0.4006,
      "num_tokens": 332416740.0,
      "step": 1631
    },
    {
      "epoch": 3.9043062200956937,
      "grad_norm": 0.6968162515814991,
      "learning_rate": 1.0141989237047547e-06,
      "loss": 0.7096,
      "num_tokens": 332651113.0,
      "step": 1632
    },
    {
      "epoch": 3.9066985645933014,
      "grad_norm": 0.6117743697368456,
      "learning_rate": 1.013515082677786e-06,
      "loss": 0.5262,
      "num_tokens": 332879068.0,
      "step": 1633
    },
    {
      "epoch": 3.909090909090909,
      "grad_norm": 0.6863788690669168,
      "learning_rate": 1.0128480931984307e-06,
      "loss": 0.5488,
      "num_tokens": 333095663.0,
      "step": 1634
    },
    {
      "epoch": 3.911483253588517,
      "grad_norm": 0.692371309851935,
      "learning_rate": 1.0121979577719493e-06,
      "loss": 0.6368,
      "num_tokens": 333279591.0,
      "step": 1635
    },
    {
      "epoch": 3.9138755980861246,
      "grad_norm": 0.5543815742967236,
      "learning_rate": 1.0115646788402948e-06,
      "loss": 0.4403,
      "num_tokens": 333515081.0,
      "step": 1636
    },
    {
      "epoch": 3.916267942583732,
      "grad_norm": 0.42712256054935754,
      "learning_rate": 1.010948258782106e-06,
      "loss": 0.2922,
      "num_tokens": 333771678.0,
      "step": 1637
    },
    {
      "epoch": 3.9186602870813396,
      "grad_norm": 0.6438096039933913,
      "learning_rate": 1.0103486999127003e-06,
      "loss": 0.4995,
      "num_tokens": 333951246.0,
      "step": 1638
    },
    {
      "epoch": 3.9210526315789473,
      "grad_norm": 0.7556968161493737,
      "learning_rate": 1.0097660044840622e-06,
      "loss": 0.5535,
      "num_tokens": 334119009.0,
      "step": 1639
    },
    {
      "epoch": 3.923444976076555,
      "grad_norm": 0.681944719510437,
      "learning_rate": 1.009200174684835e-06,
      "loss": 0.4946,
      "num_tokens": 334355751.0,
      "step": 1640
    },
    {
      "epoch": 3.925837320574163,
      "grad_norm": 0.6353703842610504,
      "learning_rate": 1.0086512126403156e-06,
      "loss": 0.4986,
      "num_tokens": 334568267.0,
      "step": 1641
    },
    {
      "epoch": 3.92822966507177,
      "grad_norm": 0.659594484115981,
      "learning_rate": 1.0081191204124426e-06,
      "loss": 0.5352,
      "num_tokens": 334796021.0,
      "step": 1642
    },
    {
      "epoch": 3.930622009569378,
      "grad_norm": 0.6454639384667783,
      "learning_rate": 1.0076038999997921e-06,
      "loss": 0.5735,
      "num_tokens": 335010213.0,
      "step": 1643
    },
    {
      "epoch": 3.9330143540669855,
      "grad_norm": 0.6662553736968893,
      "learning_rate": 1.007105553337567e-06,
      "loss": 0.5289,
      "num_tokens": 335231088.0,
      "step": 1644
    },
    {
      "epoch": 3.9354066985645932,
      "grad_norm": 0.7481341829695102,
      "learning_rate": 1.006624082297592e-06,
      "loss": 0.4365,
      "num_tokens": 335413772.0,
      "step": 1645
    },
    {
      "epoch": 3.937799043062201,
      "grad_norm": 0.5629931719433228,
      "learning_rate": 1.0061594886883055e-06,
      "loss": 0.2927,
      "num_tokens": 335607882.0,
      "step": 1646
    },
    {
      "epoch": 3.9401913875598087,
      "grad_norm": 0.634796826062313,
      "learning_rate": 1.005711774254753e-06,
      "loss": 0.4054,
      "num_tokens": 335767032.0,
      "step": 1647
    },
    {
      "epoch": 3.9425837320574164,
      "grad_norm": 0.6001806697380168,
      "learning_rate": 1.0052809406785818e-06,
      "loss": 0.4927,
      "num_tokens": 335966589.0,
      "step": 1648
    },
    {
      "epoch": 3.944976076555024,
      "grad_norm": 0.6228056915445697,
      "learning_rate": 1.0048669895780313e-06,
      "loss": 0.4874,
      "num_tokens": 336173798.0,
      "step": 1649
    },
    {
      "epoch": 3.9473684210526314,
      "grad_norm": 0.6936880601170302,
      "learning_rate": 1.0044699225079318e-06,
      "loss": 0.2793,
      "num_tokens": 336328091.0,
      "step": 1650
    },
    {
      "epoch": 3.949760765550239,
      "grad_norm": 0.7405485802310904,
      "learning_rate": 1.0040897409596937e-06,
      "loss": 0.5016,
      "num_tokens": 336476142.0,
      "step": 1651
    },
    {
      "epoch": 3.952153110047847,
      "grad_norm": 0.6722368227288898,
      "learning_rate": 1.0037264463613057e-06,
      "loss": 0.558,
      "num_tokens": 336689820.0,
      "step": 1652
    },
    {
      "epoch": 3.9545454545454546,
      "grad_norm": 0.6758549587597688,
      "learning_rate": 1.0033800400773267e-06,
      "loss": 0.509,
      "num_tokens": 336926238.0,
      "step": 1653
    },
    {
      "epoch": 3.9569377990430623,
      "grad_norm": 0.5580208536403065,
      "learning_rate": 1.003050523408884e-06,
      "loss": 0.4084,
      "num_tokens": 337164449.0,
      "step": 1654
    },
    {
      "epoch": 3.9593301435406696,
      "grad_norm": 0.628066217543992,
      "learning_rate": 1.0027378975936637e-06,
      "loss": 0.38,
      "num_tokens": 337359565.0,
      "step": 1655
    },
    {
      "epoch": 3.9617224880382773,
      "grad_norm": 0.6580546118467593,
      "learning_rate": 1.0024421638059106e-06,
      "loss": 0.5665,
      "num_tokens": 337600405.0,
      "step": 1656
    },
    {
      "epoch": 3.964114832535885,
      "grad_norm": 0.6799976508185673,
      "learning_rate": 1.0021633231564209e-06,
      "loss": 0.5376,
      "num_tokens": 337851152.0,
      "step": 1657
    },
    {
      "epoch": 3.9665071770334928,
      "grad_norm": 0.6386070227564089,
      "learning_rate": 1.00190137669254e-06,
      "loss": 0.4064,
      "num_tokens": 338007553.0,
      "step": 1658
    },
    {
      "epoch": 3.9688995215311005,
      "grad_norm": 0.671364821882183,
      "learning_rate": 1.0016563253981562e-06,
      "loss": 0.5927,
      "num_tokens": 338245439.0,
      "step": 1659
    },
    {
      "epoch": 3.971291866028708,
      "grad_norm": 0.6417827038307478,
      "learning_rate": 1.0014281701936995e-06,
      "loss": 0.5757,
      "num_tokens": 338450116.0,
      "step": 1660
    },
    {
      "epoch": 3.973684210526316,
      "grad_norm": 0.6704182296093534,
      "learning_rate": 1.001216911936137e-06,
      "loss": 0.601,
      "num_tokens": 338673938.0,
      "step": 1661
    },
    {
      "epoch": 3.9760765550239237,
      "grad_norm": 0.7305158742473202,
      "learning_rate": 1.0010225514189683e-06,
      "loss": 0.599,
      "num_tokens": 338866942.0,
      "step": 1662
    },
    {
      "epoch": 3.9784688995215314,
      "grad_norm": 0.6042213634249859,
      "learning_rate": 1.0008450893722257e-06,
      "loss": 0.4336,
      "num_tokens": 339041961.0,
      "step": 1663
    },
    {
      "epoch": 3.9808612440191387,
      "grad_norm": 0.7572508371907327,
      "learning_rate": 1.000684526462469e-06,
      "loss": 0.6808,
      "num_tokens": 339208183.0,
      "step": 1664
    },
    {
      "epoch": 3.9832535885167464,
      "grad_norm": 0.5604458795361834,
      "learning_rate": 1.0005408632927834e-06,
      "loss": 0.3913,
      "num_tokens": 339420138.0,
      "step": 1665
    },
    {
      "epoch": 3.985645933014354,
      "grad_norm": 0.5880986784981255,
      "learning_rate": 1.0004141004027772e-06,
      "loss": 0.6053,
      "num_tokens": 339646260.0,
      "step": 1666
    },
    {
      "epoch": 3.988038277511962,
      "grad_norm": 0.6648043812580797,
      "learning_rate": 1.0003042382685818e-06,
      "loss": 0.5238,
      "num_tokens": 339835650.0,
      "step": 1667
    },
    {
      "epoch": 3.990430622009569,
      "grad_norm": 0.6181701791737245,
      "learning_rate": 1.0002112773028455e-06,
      "loss": 0.506,
      "num_tokens": 340078350.0,
      "step": 1668
    },
    {
      "epoch": 3.992822966507177,
      "grad_norm": 0.5772600494577129,
      "learning_rate": 1.0001352178547368e-06,
      "loss": 0.4216,
      "num_tokens": 340306707.0,
      "step": 1669
    },
    {
      "epoch": 3.9952153110047846,
      "grad_norm": 0.735348445818812,
      "learning_rate": 1.0000760602099404e-06,
      "loss": 0.5963,
      "num_tokens": 340485329.0,
      "step": 1670
    },
    {
      "epoch": 3.9976076555023923,
      "grad_norm": 0.5903693208196457,
      "learning_rate": 1.0000338045906564e-06,
      "loss": 0.5158,
      "num_tokens": 340697021.0,
      "step": 1671
    },
    {
      "epoch": 4.0,
      "grad_norm": 0.4716699799293331,
      "learning_rate": 1.0000084511556e-06,
      "loss": 0.3813,
      "num_tokens": 340953374.0,
      "step": 1672
    },
    {
      "epoch": 4.0,
      "eval_loss": 0.4249289631843567,
      "eval_num_tokens": 340953374.0,
      "eval_runtime": 109.3301,
      "eval_samples_per_second": 27.184,
      "eval_steps_per_second": 3.403,
      "step": 1672
    },
    {
      "epoch": 4.0,
      "step": 1672,
      "total_flos": 1933127778631680.0,
      "train_loss": 0.6745891945267265,
      "train_runtime": 14540.5824,
      "train_samples_per_second": 7.357,
      "train_steps_per_second": 0.115
    }
  ],
  "logging_steps": 1,
  "max_steps": 1672,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 4,
  "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": 1933127778631680.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}