{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 875,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "grad_norm": 8.812040662498683,
      "learning_rate": 3.7037037037037036e-07,
      "loss": 0.6324,
      "step": 1
    },
    {
      "epoch": 0.0,
      "grad_norm": 8.231021188121307,
      "learning_rate": 7.407407407407407e-07,
      "loss": 0.5088,
      "step": 2
    },
    {
      "epoch": 0.0,
      "grad_norm": 10.084697585340875,
      "learning_rate": 1.111111111111111e-06,
      "loss": 0.5952,
      "step": 3
    },
    {
      "epoch": 0.0,
      "grad_norm": 8.504686522060355,
      "learning_rate": 1.4814814814814815e-06,
      "loss": 0.5506,
      "step": 4
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.78998673730679,
      "learning_rate": 1.8518518518518519e-06,
      "loss": 0.5389,
      "step": 5
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.564344020278276,
      "learning_rate": 2.222222222222222e-06,
      "loss": 0.4419,
      "step": 6
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.335657208121732,
      "learning_rate": 2.5925925925925925e-06,
      "loss": 0.4503,
      "step": 7
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.736176328651262,
      "learning_rate": 2.962962962962963e-06,
      "loss": 0.4152,
      "step": 8
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.646697141146249,
      "learning_rate": 3.3333333333333333e-06,
      "loss": 0.4766,
      "step": 9
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.271376765925272,
      "learning_rate": 3.7037037037037037e-06,
      "loss": 0.3847,
      "step": 10
    },
    {
      "epoch": 0.01,
      "grad_norm": 11.816959033855433,
      "learning_rate": 4.074074074074074e-06,
      "loss": 0.3234,
      "step": 11
    },
    {
      "epoch": 0.01,
      "grad_norm": 17.376716600598993,
      "learning_rate": 4.444444444444444e-06,
      "loss": 0.2575,
      "step": 12
    },
    {
      "epoch": 0.01,
      "grad_norm": 13.166527782235265,
      "learning_rate": 4.814814814814815e-06,
      "loss": 0.3037,
      "step": 13
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.131001484797347,
      "learning_rate": 5.185185185185185e-06,
      "loss": 0.1747,
      "step": 14
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.174937982191583,
      "learning_rate": 5.555555555555557e-06,
      "loss": 0.2399,
      "step": 15
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.592965454381307,
      "learning_rate": 5.925925925925926e-06,
      "loss": 0.1949,
      "step": 16
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.872465363012901,
      "learning_rate": 6.296296296296297e-06,
      "loss": 0.1805,
      "step": 17
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.792592320141116,
      "learning_rate": 6.666666666666667e-06,
      "loss": 0.1909,
      "step": 18
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.288697729735019,
      "learning_rate": 7.0370370370370375e-06,
      "loss": 0.1502,
      "step": 19
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.6947246687987911,
      "learning_rate": 7.4074074074074075e-06,
      "loss": 0.1174,
      "step": 20
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.72465565098873,
      "learning_rate": 7.77777777777778e-06,
      "loss": 0.1739,
      "step": 21
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.765453075575737,
      "learning_rate": 8.148148148148148e-06,
      "loss": 0.1894,
      "step": 22
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.2240061796287334,
      "learning_rate": 8.518518518518519e-06,
      "loss": 0.1376,
      "step": 23
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.983671039035109,
      "learning_rate": 8.888888888888888e-06,
      "loss": 0.1445,
      "step": 24
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.056090509557002,
      "learning_rate": 9.25925925925926e-06,
      "loss": 0.1628,
      "step": 25
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.667505927161921,
      "learning_rate": 9.62962962962963e-06,
      "loss": 0.1411,
      "step": 26
    },
    {
      "epoch": 0.03,
      "grad_norm": 9.96893801537625,
      "learning_rate": 1e-05,
      "loss": 0.1942,
      "step": 27
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.081996759418418,
      "learning_rate": 9.999965687879942e-06,
      "loss": 0.1401,
      "step": 28
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.992334648293423,
      "learning_rate": 9.999862751990697e-06,
      "loss": 0.1495,
      "step": 29
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.446786827543344,
      "learning_rate": 9.999691193745045e-06,
      "loss": 0.1356,
      "step": 30
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.452927122228086,
      "learning_rate": 9.999451015497595e-06,
      "loss": 0.1576,
      "step": 31
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.8167704565956715,
      "learning_rate": 9.999142220544759e-06,
      "loss": 0.1237,
      "step": 32
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.128302969077741,
      "learning_rate": 9.9987648131247e-06,
      "loss": 0.124,
      "step": 33
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.7257757203533584,
      "learning_rate": 9.998318798417276e-06,
      "loss": 0.1622,
      "step": 34
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.811332863461092,
      "learning_rate": 9.997804182543973e-06,
      "loss": 0.1373,
      "step": 35
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.006945777469291,
      "learning_rate": 9.997220972567815e-06,
      "loss": 0.1617,
      "step": 36
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.2290233962608275,
      "learning_rate": 9.996569176493269e-06,
      "loss": 0.1317,
      "step": 37
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.887606275239368,
      "learning_rate": 9.995848803266141e-06,
      "loss": 0.1559,
      "step": 38
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.649617040738647,
      "learning_rate": 9.99505986277344e-06,
      "loss": 0.1359,
      "step": 39
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.6342992270813492,
      "learning_rate": 9.994202365843256e-06,
      "loss": 0.1444,
      "step": 40
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.977502599851725,
      "learning_rate": 9.993276324244606e-06,
      "loss": 0.1367,
      "step": 41
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.007957356864356,
      "learning_rate": 9.992281750687265e-06,
      "loss": 0.1297,
      "step": 42
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8406129718892825,
      "learning_rate": 9.991218658821609e-06,
      "loss": 0.129,
      "step": 43
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.0044983411925275,
      "learning_rate": 9.990087063238408e-06,
      "loss": 0.107,
      "step": 44
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8795658720176402,
      "learning_rate": 9.988886979468644e-06,
      "loss": 0.1121,
      "step": 45
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.356084791458523,
      "learning_rate": 9.98761842398328e-06,
      "loss": 0.0981,
      "step": 46
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.534924570867148,
      "learning_rate": 9.98628141419305e-06,
      "loss": 0.1794,
      "step": 47
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.3579321421491866,
      "learning_rate": 9.984875968448212e-06,
      "loss": 0.108,
      "step": 48
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.2510325224717254,
      "learning_rate": 9.983402106038292e-06,
      "loss": 0.1117,
      "step": 49
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.4788953336363653,
      "learning_rate": 9.981859847191827e-06,
      "loss": 0.1282,
      "step": 50
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8906655279037572,
      "learning_rate": 9.980249213076085e-06,
      "loss": 0.1208,
      "step": 51
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.1073556178878188,
      "learning_rate": 9.978570225796777e-06,
      "loss": 0.1644,
      "step": 52
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.9452742167713701,
      "learning_rate": 9.97682290839775e-06,
      "loss": 0.1285,
      "step": 53
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.670121793189728,
      "learning_rate": 9.975007284860664e-06,
      "loss": 0.142,
      "step": 54
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8244790066486611,
      "learning_rate": 9.97312338010468e-06,
      "loss": 0.1182,
      "step": 55
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1114450183499889,
      "learning_rate": 9.971171219986104e-06,
      "loss": 0.1013,
      "step": 56
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.8735563274335463,
      "learning_rate": 9.969150831298037e-06,
      "loss": 0.1351,
      "step": 57
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.222351118905723,
      "learning_rate": 9.967062241770008e-06,
      "loss": 0.1122,
      "step": 58
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7829363422208375,
      "learning_rate": 9.964905480067585e-06,
      "loss": 0.1299,
      "step": 59
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.187968659410326,
      "learning_rate": 9.962680575792e-06,
      "loss": 0.0957,
      "step": 60
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.846988563675764,
      "learning_rate": 9.960387559479727e-06,
      "loss": 0.1187,
      "step": 61
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.643896260993509,
      "learning_rate": 9.958026462602062e-06,
      "loss": 0.1086,
      "step": 62
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.3501454089832894,
      "learning_rate": 9.955597317564705e-06,
      "loss": 0.1181,
      "step": 63
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.3392801570363164,
      "learning_rate": 9.953100157707299e-06,
      "loss": 0.1283,
      "step": 64
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.215077177644609,
      "learning_rate": 9.950535017302984e-06,
      "loss": 0.1194,
      "step": 65
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.298364445576207,
      "learning_rate": 9.947901931557924e-06,
      "loss": 0.1317,
      "step": 66
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.8138148663788862,
      "learning_rate": 9.945200936610821e-06,
      "loss": 0.1241,
      "step": 67
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.5107542165471934,
      "learning_rate": 9.942432069532417e-06,
      "loss": 0.1021,
      "step": 68
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.5847605764691766,
      "learning_rate": 9.939595368324996e-06,
      "loss": 0.1432,
      "step": 69
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.534356782579585,
      "learning_rate": 9.936690871921849e-06,
      "loss": 0.1407,
      "step": 70
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.2605138628608623,
      "learning_rate": 9.933718620186745e-06,
      "loss": 0.1323,
      "step": 71
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.239825098653746,
      "learning_rate": 9.930678653913392e-06,
      "loss": 0.1329,
      "step": 72
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.4950058767667134,
      "learning_rate": 9.927571014824862e-06,
      "loss": 0.1259,
      "step": 73
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.538742560460438,
      "learning_rate": 9.924395745573029e-06,
      "loss": 0.1276,
      "step": 74
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.8351129913937987,
      "learning_rate": 9.921152889737985e-06,
      "loss": 0.1333,
      "step": 75
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.6729191949488227,
      "learning_rate": 9.91784249182743e-06,
      "loss": 0.1209,
      "step": 76
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.5760153753569086,
      "learning_rate": 9.914464597276072e-06,
      "loss": 0.1443,
      "step": 77
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.057116589062098,
      "learning_rate": 9.911019252445e-06,
      "loss": 0.1082,
      "step": 78
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.6837432634831067,
      "learning_rate": 9.907506504621052e-06,
      "loss": 0.1133,
      "step": 79
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.118680599811244,
      "learning_rate": 9.903926402016153e-06,
      "loss": 0.1193,
      "step": 80
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.1410450154804663,
      "learning_rate": 9.90027899376667e-06,
      "loss": 0.1212,
      "step": 81
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.110344600127584,
      "learning_rate": 9.896564329932727e-06,
      "loss": 0.1318,
      "step": 82
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.33027322583564,
      "learning_rate": 9.892782461497521e-06,
      "loss": 0.1054,
      "step": 83
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9832010503747017,
      "learning_rate": 9.88893344036662e-06,
      "loss": 0.1113,
      "step": 84
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.6705177259808988,
      "learning_rate": 9.885017319367253e-06,
      "loss": 0.1181,
      "step": 85
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6879356709331421,
      "learning_rate": 9.88103415224759e-06,
      "loss": 0.0843,
      "step": 86
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3641501934361187,
      "learning_rate": 9.87698399367599e-06,
      "loss": 0.0985,
      "step": 87
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0451464275509053,
      "learning_rate": 9.872866899240265e-06,
      "loss": 0.1057,
      "step": 88
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.2974724008225484,
      "learning_rate": 9.86868292544691e-06,
      "loss": 0.1166,
      "step": 89
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.1974043469355458,
      "learning_rate": 9.864432129720332e-06,
      "loss": 0.1007,
      "step": 90
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5880779234771585,
      "learning_rate": 9.860114570402055e-06,
      "loss": 0.1059,
      "step": 91
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6210019246715581,
      "learning_rate": 9.855730306749924e-06,
      "loss": 0.0955,
      "step": 92
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.053943766699023,
      "learning_rate": 9.85127939893729e-06,
      "loss": 0.1223,
      "step": 93
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5230542823178097,
      "learning_rate": 9.846761908052188e-06,
      "loss": 0.0969,
      "step": 94
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9898055550048976,
      "learning_rate": 9.842177896096495e-06,
      "loss": 0.1359,
      "step": 95
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5219298537997954,
      "learning_rate": 9.837527425985074e-06,
      "loss": 0.1049,
      "step": 96
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5694384883915047,
      "learning_rate": 9.832810561544925e-06,
      "loss": 0.0885,
      "step": 97
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8382074625592275,
      "learning_rate": 9.828027367514296e-06,
      "loss": 0.0789,
      "step": 98
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.094156783497333,
      "learning_rate": 9.823177909541795e-06,
      "loss": 0.125,
      "step": 99
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.1177801350294474,
      "learning_rate": 9.818262254185495e-06,
      "loss": 0.1286,
      "step": 100
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.051759324600107,
      "learning_rate": 9.813280468912024e-06,
      "loss": 0.0867,
      "step": 101
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.309052314502647,
      "learning_rate": 9.808232622095621e-06,
      "loss": 0.0836,
      "step": 102
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9691513403830945,
      "learning_rate": 9.803118783017221e-06,
      "loss": 0.1088,
      "step": 103
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8816504929896276,
      "learning_rate": 9.797939021863487e-06,
      "loss": 0.1263,
      "step": 104
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.5814461055271196,
      "learning_rate": 9.792693409725853e-06,
      "loss": 0.1266,
      "step": 105
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.804380821700463,
      "learning_rate": 9.78738201859955e-06,
      "loss": 0.1548,
      "step": 106
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.0086077809415332,
      "learning_rate": 9.782004921382612e-06,
      "loss": 0.1312,
      "step": 107
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7232244244365362,
      "learning_rate": 9.776562191874884e-06,
      "loss": 0.1197,
      "step": 108
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3610013356293815,
      "learning_rate": 9.771053904776998e-06,
      "loss": 0.1139,
      "step": 109
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.0085151960644643,
      "learning_rate": 9.765480135689357e-06,
      "loss": 0.1049,
      "step": 110
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.94972805872663,
      "learning_rate": 9.759840961111098e-06,
      "loss": 0.0992,
      "step": 111
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2797796597040898,
      "learning_rate": 9.754136458439033e-06,
      "loss": 0.0779,
      "step": 112
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.080098377998089,
      "learning_rate": 9.748366705966595e-06,
      "loss": 0.1192,
      "step": 113
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.026465124623598,
      "learning_rate": 9.742531782882758e-06,
      "loss": 0.1248,
      "step": 114
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.419083049509824,
      "learning_rate": 9.736631769270958e-06,
      "loss": 0.1284,
      "step": 115
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9998224613269435,
      "learning_rate": 9.730666746107982e-06,
      "loss": 0.1106,
      "step": 116
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7572924951035829,
      "learning_rate": 9.724636795262868e-06,
      "loss": 0.0874,
      "step": 117
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.448977492181066,
      "learning_rate": 9.718541999495773e-06,
      "loss": 0.0823,
      "step": 118
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.110748776288409,
      "learning_rate": 9.712382442456845e-06,
      "loss": 0.1604,
      "step": 119
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.09981079064145,
      "learning_rate": 9.706158208685066e-06,
      "loss": 0.0911,
      "step": 120
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9370008400745906,
      "learning_rate": 9.6998693836071e-06,
      "loss": 0.1122,
      "step": 121
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9316495646897083,
      "learning_rate": 9.693516053536118e-06,
      "loss": 0.0807,
      "step": 122
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.16334665341287,
      "learning_rate": 9.687098305670606e-06,
      "loss": 0.1276,
      "step": 123
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.341234757750837,
      "learning_rate": 9.680616228093178e-06,
      "loss": 0.1187,
      "step": 124
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.223298454459867,
      "learning_rate": 9.674069909769365e-06,
      "loss": 0.1045,
      "step": 125
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.4945449518796585,
      "learning_rate": 9.66745944054639e-06,
      "loss": 0.12,
      "step": 126
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.143984048508036,
      "learning_rate": 9.66078491115194e-06,
      "loss": 0.1013,
      "step": 127
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.7705806851123493,
      "learning_rate": 9.654046413192916e-06,
      "loss": 0.1123,
      "step": 128
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.2313407498327578,
      "learning_rate": 9.647244039154178e-06,
      "loss": 0.0846,
      "step": 129
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7722835766792302,
      "learning_rate": 9.640377882397276e-06,
      "loss": 0.0965,
      "step": 130
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.4978175367401065,
      "learning_rate": 9.633448037159167e-06,
      "loss": 0.1218,
      "step": 131
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.9772864819418463,
      "learning_rate": 9.626454598550927e-06,
      "loss": 0.0919,
      "step": 132
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9729154551618289,
      "learning_rate": 9.619397662556434e-06,
      "loss": 0.1104,
      "step": 133
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9749670072044205,
      "learning_rate": 9.612277326031065e-06,
      "loss": 0.0879,
      "step": 134
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.46650550465165,
      "learning_rate": 9.605093686700356e-06,
      "loss": 0.1151,
      "step": 135
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.1700672021295397,
      "learning_rate": 9.597846843158663e-06,
      "loss": 0.1192,
      "step": 136
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5484117589367528,
      "learning_rate": 9.590536894867814e-06,
      "loss": 0.1093,
      "step": 137
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.159315025435744,
      "learning_rate": 9.58316394215574e-06,
      "loss": 0.1246,
      "step": 138
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.3062290760264044,
      "learning_rate": 9.575728086215093e-06,
      "loss": 0.1044,
      "step": 139
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.1924424597130647,
      "learning_rate": 9.568229429101867e-06,
      "loss": 0.1018,
      "step": 140
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4292889067167893,
      "learning_rate": 9.560668073733993e-06,
      "loss": 0.1102,
      "step": 141
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8452890182403634,
      "learning_rate": 9.553044123889922e-06,
      "loss": 0.0851,
      "step": 142
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.686312012130693,
      "learning_rate": 9.54535768420721e-06,
      "loss": 0.0877,
      "step": 143
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9544040435460412,
      "learning_rate": 9.537608860181069e-06,
      "loss": 0.1401,
      "step": 144
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.5638709576265777,
      "learning_rate": 9.529797758162935e-06,
      "loss": 0.1092,
      "step": 145
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.282994763913454,
      "learning_rate": 9.521924485358993e-06,
      "loss": 0.1153,
      "step": 146
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.5304046735735803,
      "learning_rate": 9.513989149828718e-06,
      "loss": 0.1249,
      "step": 147
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.084105660790946,
      "learning_rate": 9.505991860483384e-06,
      "loss": 0.1127,
      "step": 148
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.0807266175742334,
      "learning_rate": 9.497932727084571e-06,
      "loss": 0.0923,
      "step": 149
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.535491149619429,
      "learning_rate": 9.489811860242658e-06,
      "loss": 0.1171,
      "step": 150
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.8577061917206694,
      "learning_rate": 9.481629371415315e-06,
      "loss": 0.1027,
      "step": 151
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.366848987135643,
      "learning_rate": 9.47338537290595e-06,
      "loss": 0.1461,
      "step": 152
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.370601085821976,
      "learning_rate": 9.465079977862193e-06,
      "loss": 0.1176,
      "step": 153
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.418308466667954,
      "learning_rate": 9.45671330027433e-06,
      "loss": 0.14,
      "step": 154
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.3789531359503155,
      "learning_rate": 9.448285454973739e-06,
      "loss": 0.1272,
      "step": 155
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.837535532086006,
      "learning_rate": 9.439796557631311e-06,
      "loss": 0.1018,
      "step": 156
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.26111781111186,
      "learning_rate": 9.431246724755879e-06,
      "loss": 0.1089,
      "step": 157
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.149991918092674,
      "learning_rate": 9.422636073692595e-06,
      "loss": 0.1224,
      "step": 158
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7257130483628491,
      "learning_rate": 9.413964722621339e-06,
      "loss": 0.0829,
      "step": 159
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8121675919179852,
      "learning_rate": 9.405232790555083e-06,
      "loss": 0.1082,
      "step": 160
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1269467977377563,
      "learning_rate": 9.396440397338273e-06,
      "loss": 0.0735,
      "step": 161
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.66841905851172,
      "learning_rate": 9.387587663645163e-06,
      "loss": 0.1171,
      "step": 162
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9979957509268915,
      "learning_rate": 9.378674710978185e-06,
      "loss": 0.0867,
      "step": 163
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3467231541181564,
      "learning_rate": 9.369701661666255e-06,
      "loss": 0.084,
      "step": 164
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5661741867693126,
      "learning_rate": 9.36066863886311e-06,
      "loss": 0.0879,
      "step": 165
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4781065347920526,
      "learning_rate": 9.35157576654562e-06,
      "loss": 0.081,
      "step": 166
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.747486649844575,
      "learning_rate": 9.342423169512072e-06,
      "loss": 0.0998,
      "step": 167
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.183425315531194,
      "learning_rate": 9.333210973380469e-06,
      "loss": 0.113,
      "step": 168
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.6187200801222126,
      "learning_rate": 9.323939304586806e-06,
      "loss": 0.1211,
      "step": 169
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.298579486613351,
      "learning_rate": 9.314608290383324e-06,
      "loss": 0.098,
      "step": 170
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.3760059355851286,
      "learning_rate": 9.305218058836778e-06,
      "loss": 0.1072,
      "step": 171
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.7212461615360892,
      "learning_rate": 9.295768738826668e-06,
      "loss": 0.1128,
      "step": 172
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7379020422811915,
      "learning_rate": 9.286260460043475e-06,
      "loss": 0.125,
      "step": 173
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.1792312593675724,
      "learning_rate": 9.27669335298688e-06,
      "loss": 0.124,
      "step": 174
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.0277111530618916,
      "learning_rate": 9.267067548963975e-06,
      "loss": 0.0942,
      "step": 175
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4469039936621195,
      "learning_rate": 9.257383180087454e-06,
      "loss": 0.0909,
      "step": 176
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.051541742743329,
      "learning_rate": 9.24764037927381e-06,
      "loss": 0.1199,
      "step": 177
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4918383860547757,
      "learning_rate": 9.237839280241504e-06,
      "loss": 0.0948,
      "step": 178
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.64559721928272,
      "learning_rate": 9.22798001750913e-06,
      "loss": 0.0926,
      "step": 179
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.7118844975893306,
      "learning_rate": 9.218062726393572e-06,
      "loss": 0.1006,
      "step": 180
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.9102131867075969,
      "learning_rate": 9.208087543008142e-06,
      "loss": 0.0971,
      "step": 181
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1482279190562423,
      "learning_rate": 9.198054604260714e-06,
      "loss": 0.0856,
      "step": 182
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8541863535744467,
      "learning_rate": 9.187964047851851e-06,
      "loss": 0.0744,
      "step": 183
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6754271851342437,
      "learning_rate": 9.177816012272904e-06,
      "loss": 0.1079,
      "step": 184
    },
    {
      "epoch": 0.21,
      "grad_norm": 7.473478229947605,
      "learning_rate": 9.16761063680412e-06,
      "loss": 0.0957,
      "step": 185
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.0632873310708706,
      "learning_rate": 9.157348061512728e-06,
      "loss": 0.1259,
      "step": 186
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.173245370297517,
      "learning_rate": 9.14702842725101e-06,
      "loss": 0.1064,
      "step": 187
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.5577392058893906,
      "learning_rate": 9.136651875654381e-06,
      "loss": 0.1018,
      "step": 188
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.4914753007974073,
      "learning_rate": 9.126218549139434e-06,
      "loss": 0.0766,
      "step": 189
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.607548007532628,
      "learning_rate": 9.115728590901988e-06,
      "loss": 0.1145,
      "step": 190
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.3652955179863704,
      "learning_rate": 9.10518214491513e-06,
      "loss": 0.1075,
      "step": 191
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.944852075113964,
      "learning_rate": 9.094579355927224e-06,
      "loss": 0.1237,
      "step": 192
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.531422123532763,
      "learning_rate": 9.083920369459941e-06,
      "loss": 0.0996,
      "step": 193
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6840091229079706,
      "learning_rate": 9.073205331806248e-06,
      "loss": 0.0978,
      "step": 194
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.1180165313908608,
      "learning_rate": 9.062434390028407e-06,
      "loss": 0.1071,
      "step": 195
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3064591316530667,
      "learning_rate": 9.05160769195596e-06,
      "loss": 0.0989,
      "step": 196
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.467504354977469,
      "learning_rate": 9.040725386183691e-06,
      "loss": 0.1042,
      "step": 197
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.050946499964558,
      "learning_rate": 9.029787622069594e-06,
      "loss": 0.0975,
      "step": 198
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4499865119155375,
      "learning_rate": 9.018794549732819e-06,
      "loss": 0.085,
      "step": 199
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3533643803726048,
      "learning_rate": 9.007746320051612e-06,
      "loss": 0.1035,
      "step": 200
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6026335880678133,
      "learning_rate": 8.996643084661245e-06,
      "loss": 0.1027,
      "step": 201
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.9397686652721016,
      "learning_rate": 8.985484995951941e-06,
      "loss": 0.1186,
      "step": 202
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.985748151579965,
      "learning_rate": 8.974272207066767e-06,
      "loss": 0.1106,
      "step": 203
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.0494865642540487,
      "learning_rate": 8.96300487189955e-06,
      "loss": 0.1064,
      "step": 204
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5961078519634195,
      "learning_rate": 8.951683145092749e-06,
      "loss": 0.107,
      "step": 205
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.1110307819462288,
      "learning_rate": 8.940307182035348e-06,
      "loss": 0.0572,
      "step": 206
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5666032654284987,
      "learning_rate": 8.928877138860708e-06,
      "loss": 0.0855,
      "step": 207
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.5628141099754145,
      "learning_rate": 8.917393172444435e-06,
      "loss": 0.1166,
      "step": 208
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.1688262602344186,
      "learning_rate": 8.905855440402225e-06,
      "loss": 0.1282,
      "step": 209
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.9861109578632914,
      "learning_rate": 8.894264101087692e-06,
      "loss": 0.114,
      "step": 210
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.501921488076984,
      "learning_rate": 8.882619313590212e-06,
      "loss": 0.1097,
      "step": 211
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.268674705211975,
      "learning_rate": 8.870921237732722e-06,
      "loss": 0.1071,
      "step": 212
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.067986488761189,
      "learning_rate": 8.859170034069533e-06,
      "loss": 0.078,
      "step": 213
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.848889166285347,
      "learning_rate": 8.847365863884131e-06,
      "loss": 0.0939,
      "step": 214
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.633820214871211,
      "learning_rate": 8.835508889186957e-06,
      "loss": 0.0668,
      "step": 215
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.762882965258614,
      "learning_rate": 8.823599272713188e-06,
      "loss": 0.1218,
      "step": 216
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.320523619911371,
      "learning_rate": 8.8116371779205e-06,
      "loss": 0.0862,
      "step": 217
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.2237720260978944,
      "learning_rate": 8.799622768986826e-06,
      "loss": 0.0818,
      "step": 218
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.4364710210798948,
      "learning_rate": 8.787556210808101e-06,
      "loss": 0.1199,
      "step": 219
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6849135689019348,
      "learning_rate": 8.775437668996004e-06,
      "loss": 0.1075,
      "step": 220
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.220372344984879,
      "learning_rate": 8.76326730987568e-06,
      "loss": 0.0939,
      "step": 221
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5555946268423595,
      "learning_rate": 8.751045300483457e-06,
      "loss": 0.0936,
      "step": 222
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.1402335410022744,
      "learning_rate": 8.738771808564555e-06,
      "loss": 0.0934,
      "step": 223
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0488112883529606,
      "learning_rate": 8.726447002570791e-06,
      "loss": 0.0984,
      "step": 224
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6002022312116384,
      "learning_rate": 8.714071051658247e-06,
      "loss": 0.1182,
      "step": 225
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1565321002590416,
      "learning_rate": 8.701644125684972e-06,
      "loss": 0.0936,
      "step": 226
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.185324721324382,
      "learning_rate": 8.689166395208638e-06,
      "loss": 0.083,
      "step": 227
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.030700715820133,
      "learning_rate": 8.676638031484196e-06,
      "loss": 0.117,
      "step": 228
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.984880186535839,
      "learning_rate": 8.664059206461537e-06,
      "loss": 0.1018,
      "step": 229
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5970035045867854,
      "learning_rate": 8.65143009278312e-06,
      "loss": 0.0949,
      "step": 230
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1816220314739803,
      "learning_rate": 8.638750863781614e-06,
      "loss": 0.0813,
      "step": 231
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1265530666409853,
      "learning_rate": 8.626021693477506e-06,
      "loss": 0.0807,
      "step": 232
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4798060764319763,
      "learning_rate": 8.613242756576729e-06,
      "loss": 0.1005,
      "step": 233
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.2654194933943224,
      "learning_rate": 8.600414228468245e-06,
      "loss": 0.1188,
      "step": 234
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.069029364915602,
      "learning_rate": 8.587536285221656e-06,
      "loss": 0.1174,
      "step": 235
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.575346902419333,
      "learning_rate": 8.574609103584776e-06,
      "loss": 0.1085,
      "step": 236
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4098282704868057,
      "learning_rate": 8.561632860981205e-06,
      "loss": 0.0866,
      "step": 237
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.322192828067451,
      "learning_rate": 8.548607735507903e-06,
      "loss": 0.0989,
      "step": 238
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.503973593981987,
      "learning_rate": 8.535533905932739e-06,
      "loss": 0.0959,
      "step": 239
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.9979798111827445,
      "learning_rate": 8.522411551692034e-06,
      "loss": 0.1173,
      "step": 240
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.331120139565699,
      "learning_rate": 8.509240852888106e-06,
      "loss": 0.1207,
      "step": 241
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.6853298265250545,
      "learning_rate": 8.496021990286799e-06,
      "loss": 0.1062,
      "step": 242
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.9768348877886996,
      "learning_rate": 8.482755145314987e-06,
      "loss": 0.1072,
      "step": 243
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.6831710659010493,
      "learning_rate": 8.469440500058104e-06,
      "loss": 0.1415,
      "step": 244
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3075146767802872,
      "learning_rate": 8.456078237257632e-06,
      "loss": 0.0927,
      "step": 245
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6603378900652728,
      "learning_rate": 8.442668540308597e-06,
      "loss": 0.1199,
      "step": 246
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.732311993009954,
      "learning_rate": 8.429211593257054e-06,
      "loss": 0.1162,
      "step": 247
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0678226132557302,
      "learning_rate": 8.415707580797552e-06,
      "loss": 0.0811,
      "step": 248
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.637868487774632,
      "learning_rate": 8.402156688270613e-06,
      "loss": 0.1134,
      "step": 249
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.133244302298436,
      "learning_rate": 8.388559101660177e-06,
      "loss": 0.0926,
      "step": 250
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.0183612078901554,
      "learning_rate": 8.374915007591053e-06,
      "loss": 0.1111,
      "step": 251
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.9309545129494814,
      "learning_rate": 8.361224593326359e-06,
      "loss": 0.0935,
      "step": 252
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.9872264464717466,
      "learning_rate": 8.347488046764948e-06,
      "loss": 0.0943,
      "step": 253
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.421297893280941,
      "learning_rate": 8.33370555643884e-06,
      "loss": 0.1065,
      "step": 254
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.897377603677577,
      "learning_rate": 8.319877311510614e-06,
      "loss": 0.1002,
      "step": 255
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.3046015963660937,
      "learning_rate": 8.30600350177083e-06,
      "loss": 0.0855,
      "step": 256
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.124569836934495,
      "learning_rate": 8.29208431763542e-06,
      "loss": 0.1393,
      "step": 257
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.1578644869843313,
      "learning_rate": 8.278119950143074e-06,
      "loss": 0.0727,
      "step": 258
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.854683458939737,
      "learning_rate": 8.264110590952609e-06,
      "loss": 0.0943,
      "step": 259
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4832089308281162,
      "learning_rate": 8.250056432340352e-06,
      "loss": 0.0874,
      "step": 260
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.258489452559218,
      "learning_rate": 8.235957667197494e-06,
      "loss": 0.0813,
      "step": 261
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8439098458745395,
      "learning_rate": 8.221814489027447e-06,
      "loss": 0.1109,
      "step": 262
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.044799826590439,
      "learning_rate": 8.207627091943178e-06,
      "loss": 0.1031,
      "step": 263
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0827682370362863,
      "learning_rate": 8.193395670664555e-06,
      "loss": 0.1009,
      "step": 264
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.323956787231295,
      "learning_rate": 8.179120420515675e-06,
      "loss": 0.124,
      "step": 265
    },
    {
      "epoch": 0.3,
      "grad_norm": 10.023030336667428,
      "learning_rate": 8.164801537422179e-06,
      "loss": 0.0942,
      "step": 266
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7106775924129292,
      "learning_rate": 8.150439217908557e-06,
      "loss": 0.1294,
      "step": 267
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.496815910955381,
      "learning_rate": 8.136033659095462e-06,
      "loss": 0.1046,
      "step": 268
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.646786991911152,
      "learning_rate": 8.121585058697e-06,
      "loss": 0.0861,
      "step": 269
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.8266200161202253,
      "learning_rate": 8.107093615018017e-06,
      "loss": 0.0949,
      "step": 270
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.5916169021461823,
      "learning_rate": 8.092559526951374e-06,
      "loss": 0.1169,
      "step": 271
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3232237535990277,
      "learning_rate": 8.077982993975222e-06,
      "loss": 0.0972,
      "step": 272
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5105056997202508,
      "learning_rate": 8.063364216150258e-06,
      "loss": 0.0838,
      "step": 273
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.422000688554186,
      "learning_rate": 8.04870339411699e-06,
      "loss": 0.0906,
      "step": 274
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6597229035037346,
      "learning_rate": 8.034000729092967e-06,
      "loss": 0.1005,
      "step": 275
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.0117580401941204,
      "learning_rate": 8.019256422870036e-06,
      "loss": 0.1114,
      "step": 276
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1092429217293147,
      "learning_rate": 8.00447067781156e-06,
      "loss": 0.0879,
      "step": 277
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7979548894678264,
      "learning_rate": 7.98964369684964e-06,
      "loss": 0.0802,
      "step": 278
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.002883666778827,
      "learning_rate": 7.974775683482337e-06,
      "loss": 0.1201,
      "step": 279
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.9158794261385665,
      "learning_rate": 7.959866841770875e-06,
      "loss": 0.0928,
      "step": 280
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.8564270859008971,
      "learning_rate": 7.94491737633684e-06,
      "loss": 0.1002,
      "step": 281
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9683845320254458,
      "learning_rate": 7.929927492359377e-06,
      "loss": 0.076,
      "step": 282
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4081959120743777,
      "learning_rate": 7.914897395572362e-06,
      "loss": 0.1062,
      "step": 283
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.5215551080759475,
      "learning_rate": 7.899827292261589e-06,
      "loss": 0.1165,
      "step": 284
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2748703118190907,
      "learning_rate": 7.884717389261935e-06,
      "loss": 0.0815,
      "step": 285
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7476333672006885,
      "learning_rate": 7.869567893954523e-06,
      "loss": 0.1036,
      "step": 286
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.0867624034561008,
      "learning_rate": 7.854379014263877e-06,
      "loss": 0.0975,
      "step": 287
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.505523226821333,
      "learning_rate": 7.839150958655058e-06,
      "loss": 0.0909,
      "step": 288
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1729694098449017,
      "learning_rate": 7.823883936130817e-06,
      "loss": 0.0973,
      "step": 289
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.8384149448507516,
      "learning_rate": 7.808578156228718e-06,
      "loss": 0.0897,
      "step": 290
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.62588050433828,
      "learning_rate": 7.793233829018263e-06,
      "loss": 0.0743,
      "step": 291
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3669908142605618,
      "learning_rate": 7.777851165098012e-06,
      "loss": 0.0905,
      "step": 292
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.808747017133476,
      "learning_rate": 7.762430375592689e-06,
      "loss": 0.1093,
      "step": 293
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6645286286403973,
      "learning_rate": 7.746971672150286e-06,
      "loss": 0.0904,
      "step": 294
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5994325863369407,
      "learning_rate": 7.731475266939159e-06,
      "loss": 0.1028,
      "step": 295
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5901303888596037,
      "learning_rate": 7.715941372645115e-06,
      "loss": 0.0956,
      "step": 296
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.328737208260307,
      "learning_rate": 7.70037020246849e-06,
      "loss": 0.1105,
      "step": 297
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.9563743002247425,
      "learning_rate": 7.68476197012123e-06,
      "loss": 0.1004,
      "step": 298
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.0781455779725273,
      "learning_rate": 7.669116889823955e-06,
      "loss": 0.0996,
      "step": 299
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9436403770848232,
      "learning_rate": 7.653435176303007e-06,
      "loss": 0.0761,
      "step": 300
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6406510415551603,
      "learning_rate": 7.637717044787527e-06,
      "loss": 0.1147,
      "step": 301
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7565088827574487,
      "learning_rate": 7.621962711006479e-06,
      "loss": 0.1064,
      "step": 302
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.744343584962407,
      "learning_rate": 7.6061723911857e-06,
      "loss": 0.1237,
      "step": 303
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.8449164911800016,
      "learning_rate": 7.59034630204493e-06,
      "loss": 0.0866,
      "step": 304
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.950623541879017,
      "learning_rate": 7.574484660794836e-06,
      "loss": 0.0933,
      "step": 305
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.1686948556536465,
      "learning_rate": 7.558587685134035e-06,
      "loss": 0.1025,
      "step": 306
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7581227603244038,
      "learning_rate": 7.542655593246103e-06,
      "loss": 0.0875,
      "step": 307
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.067970224647301,
      "learning_rate": 7.526688603796578e-06,
      "loss": 0.0984,
      "step": 308
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.589283294901552,
      "learning_rate": 7.510686935929963e-06,
      "loss": 0.086,
      "step": 309
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.0302397726328767,
      "learning_rate": 7.494650809266719e-06,
      "loss": 0.099,
      "step": 310
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5989723290365598,
      "learning_rate": 7.478580443900247e-06,
      "loss": 0.0939,
      "step": 311
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7617760156145226,
      "learning_rate": 7.4624760603938704e-06,
      "loss": 0.1073,
      "step": 312
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.399035341306831,
      "learning_rate": 7.446337879777803e-06,
      "loss": 0.0875,
      "step": 313
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7044340741166755,
      "learning_rate": 7.430166123546122e-06,
      "loss": 0.1059,
      "step": 314
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.0742510793830444,
      "learning_rate": 7.413961013653725e-06,
      "loss": 0.121,
      "step": 315
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3710310648899238,
      "learning_rate": 7.397722772513282e-06,
      "loss": 0.099,
      "step": 316
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4093925149964166,
      "learning_rate": 7.381451622992184e-06,
      "loss": 0.0809,
      "step": 317
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0871938974349964,
      "learning_rate": 7.3651477884094855e-06,
      "loss": 0.0873,
      "step": 318
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.6023982808593678,
      "learning_rate": 7.34881149253284e-06,
      "loss": 0.0827,
      "step": 319
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.467883768091289,
      "learning_rate": 7.332442959575424e-06,
      "loss": 0.1005,
      "step": 320
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.1212388969162672,
      "learning_rate": 7.316042414192866e-06,
      "loss": 0.0914,
      "step": 321
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.8358020056483744,
      "learning_rate": 7.299610081480155e-06,
      "loss": 0.1146,
      "step": 322
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3636541396641189,
      "learning_rate": 7.283146186968566e-06,
      "loss": 0.1085,
      "step": 323
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.07569146070376,
      "learning_rate": 7.266650956622546e-06,
      "loss": 0.089,
      "step": 324
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.8768958532173836,
      "learning_rate": 7.2501246168366235e-06,
      "loss": 0.096,
      "step": 325
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.8347653173611815,
      "learning_rate": 7.233567394432302e-06,
      "loss": 0.1067,
      "step": 326
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.202389845113578,
      "learning_rate": 7.216979516654944e-06,
      "loss": 0.0754,
      "step": 327
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0247024117157115,
      "learning_rate": 7.200361211170648e-06,
      "loss": 0.0958,
      "step": 328
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9758438803344642,
      "learning_rate": 7.183712706063133e-06,
      "loss": 0.073,
      "step": 329
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0927982452541785,
      "learning_rate": 7.1670342298306e-06,
      "loss": 0.0891,
      "step": 330
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3702951110134494,
      "learning_rate": 7.1503260113826035e-06,
      "loss": 0.102,
      "step": 331
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6194293721517825,
      "learning_rate": 7.1335882800369e-06,
      "loss": 0.0877,
      "step": 332
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3330024442939468,
      "learning_rate": 7.116821265516307e-06,
      "loss": 0.1038,
      "step": 333
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5082089283588793,
      "learning_rate": 7.100025197945555e-06,
      "loss": 0.0903,
      "step": 334
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.918479171265335,
      "learning_rate": 7.083200307848116e-06,
      "loss": 0.0782,
      "step": 335
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.270870459177953,
      "learning_rate": 7.066346826143049e-06,
      "loss": 0.078,
      "step": 336
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.808167569407084,
      "learning_rate": 7.049464984141831e-06,
      "loss": 0.1034,
      "step": 337
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4850123445239072,
      "learning_rate": 7.032555013545176e-06,
      "loss": 0.0987,
      "step": 338
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0857390354428968,
      "learning_rate": 7.015617146439863e-06,
      "loss": 0.0846,
      "step": 339
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4019142617304745,
      "learning_rate": 6.998651615295541e-06,
      "loss": 0.0946,
      "step": 340
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4419350588759805,
      "learning_rate": 6.9816586529615475e-06,
      "loss": 0.0925,
      "step": 341
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.122469313937723,
      "learning_rate": 6.9646384926637076e-06,
      "loss": 0.101,
      "step": 342
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.023493438132179,
      "learning_rate": 6.947591368001138e-06,
      "loss": 0.1049,
      "step": 343
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.8085392355883325,
      "learning_rate": 6.930517512943029e-06,
      "loss": 0.0906,
      "step": 344
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.8476500146983619,
      "learning_rate": 6.913417161825449e-06,
      "loss": 0.075,
      "step": 345
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3169871494739973,
      "learning_rate": 6.896290549348119e-06,
      "loss": 0.0999,
      "step": 346
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5793887545549115,
      "learning_rate": 6.879137910571191e-06,
      "loss": 0.0951,
      "step": 347
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.8170746345841167,
      "learning_rate": 6.861959480912027e-06,
      "loss": 0.0872,
      "step": 348
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.9489463937624985,
      "learning_rate": 6.8447554961419615e-06,
      "loss": 0.1034,
      "step": 349
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.1832874089959704,
      "learning_rate": 6.82752619238307e-06,
      "loss": 0.1158,
      "step": 350
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3585943935815048,
      "learning_rate": 6.810271806104931e-06,
      "loss": 0.0801,
      "step": 351
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.249666680648082,
      "learning_rate": 6.792992574121374e-06,
      "loss": 0.097,
      "step": 352
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.597130012884428,
      "learning_rate": 6.7756887335872275e-06,
      "loss": 0.1166,
      "step": 353
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4936397575141562,
      "learning_rate": 6.758360521995079e-06,
      "loss": 0.1212,
      "step": 354
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.171091229180981,
      "learning_rate": 6.741008177171995e-06,
      "loss": 0.0908,
      "step": 355
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9253783941058572,
      "learning_rate": 6.723631937276271e-06,
      "loss": 0.0715,
      "step": 356
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.1914008404922645,
      "learning_rate": 6.706232040794162e-06,
      "loss": 0.0753,
      "step": 357
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6494093340509537,
      "learning_rate": 6.6888087265365995e-06,
      "loss": 0.1066,
      "step": 358
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.1812819049319463,
      "learning_rate": 6.671362233635926e-06,
      "loss": 0.0927,
      "step": 359
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.4559536958240527,
      "learning_rate": 6.653892801542604e-06,
      "loss": 0.0928,
      "step": 360
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.25389273102752,
      "learning_rate": 6.636400670021934e-06,
      "loss": 0.0999,
      "step": 361
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.7179424428138383,
      "learning_rate": 6.618886079150764e-06,
      "loss": 0.0921,
      "step": 362
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.135878355877357,
      "learning_rate": 6.601349269314188e-06,
      "loss": 0.082,
      "step": 363
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.362482025104223,
      "learning_rate": 6.583790481202261e-06,
      "loss": 0.0999,
      "step": 364
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.9432112152221168,
      "learning_rate": 6.566209955806679e-06,
      "loss": 0.1061,
      "step": 365
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6175027458716444,
      "learning_rate": 6.54860793441748e-06,
      "loss": 0.1199,
      "step": 366
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7771666465783587,
      "learning_rate": 6.530984658619735e-06,
      "loss": 0.1045,
      "step": 367
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4987987433398124,
      "learning_rate": 6.5133403702902245e-06,
      "loss": 0.1011,
      "step": 368
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.024080574349673,
      "learning_rate": 6.495675311594123e-06,
      "loss": 0.0962,
      "step": 369
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.233698410707572,
      "learning_rate": 6.477989724981679e-06,
      "loss": 0.102,
      "step": 370
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.860410897978535,
      "learning_rate": 6.46028385318488e-06,
      "loss": 0.0882,
      "step": 371
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.056573638183693,
      "learning_rate": 6.442557939214125e-06,
      "loss": 0.0768,
      "step": 372
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.8634932101277937,
      "learning_rate": 6.42481222635489e-06,
      "loss": 0.1102,
      "step": 373
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.624917843958734,
      "learning_rate": 6.407046958164387e-06,
      "loss": 0.1014,
      "step": 374
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3310298175257878,
      "learning_rate": 6.389262378468219e-06,
      "loss": 0.0828,
      "step": 375
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0614121621946018,
      "learning_rate": 6.371458731357047e-06,
      "loss": 0.0987,
      "step": 376
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.7534907444752217,
      "learning_rate": 6.353636261183214e-06,
      "loss": 0.0884,
      "step": 377
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3293066218185532,
      "learning_rate": 6.335795212557416e-06,
      "loss": 0.1089,
      "step": 378
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.8227195971889414,
      "learning_rate": 6.3179358303453386e-06,
      "loss": 0.1146,
      "step": 379
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.8879414641081136,
      "learning_rate": 6.300058359664285e-06,
      "loss": 0.0771,
      "step": 380
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.0221109980415646,
      "learning_rate": 6.2821630458798236e-06,
      "loss": 0.113,
      "step": 381
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.9690496853348498,
      "learning_rate": 6.264250134602415e-06,
      "loss": 0.1076,
      "step": 382
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3353353790171576,
      "learning_rate": 6.246319871684048e-06,
      "loss": 0.0983,
      "step": 383
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8840122811422151,
      "learning_rate": 6.228372503214853e-06,
      "loss": 0.0784,
      "step": 384
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.9223687815447972,
      "learning_rate": 6.210408275519735e-06,
      "loss": 0.1164,
      "step": 385
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3646046422183316,
      "learning_rate": 6.192427435154991e-06,
      "loss": 0.1062,
      "step": 386
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3010904867560287,
      "learning_rate": 6.17443022890492e-06,
      "loss": 0.0733,
      "step": 387
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9588966018297974,
      "learning_rate": 6.156416903778443e-06,
      "loss": 0.0937,
      "step": 388
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8176889337157417,
      "learning_rate": 6.138387707005711e-06,
      "loss": 0.0785,
      "step": 389
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3798947208822552,
      "learning_rate": 6.1203428860347085e-06,
      "loss": 0.1029,
      "step": 390
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2627266387294376,
      "learning_rate": 6.10228268852786e-06,
      "loss": 0.0853,
      "step": 391
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3636634430274013,
      "learning_rate": 6.084207362358635e-06,
      "loss": 0.1015,
      "step": 392
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.7109591987525221,
      "learning_rate": 6.066117155608136e-06,
      "loss": 0.1118,
      "step": 393
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.517961417989287,
      "learning_rate": 6.048012316561699e-06,
      "loss": 0.0852,
      "step": 394
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.342629048761873,
      "learning_rate": 6.029893093705492e-06,
      "loss": 0.1187,
      "step": 395
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9294142148345402,
      "learning_rate": 6.011759735723096e-06,
      "loss": 0.0864,
      "step": 396
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.1950075847793311,
      "learning_rate": 5.993612491492088e-06,
      "loss": 0.092,
      "step": 397
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3551545691072298,
      "learning_rate": 5.975451610080643e-06,
      "loss": 0.0693,
      "step": 398
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.216161129705157,
      "learning_rate": 5.957277340744094e-06,
      "loss": 0.0895,
      "step": 399
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4076300298899016,
      "learning_rate": 5.939089932921527e-06,
      "loss": 0.085,
      "step": 400
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.465518309640694,
      "learning_rate": 5.920889636232352e-06,
      "loss": 0.0957,
      "step": 401
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3444282299556731,
      "learning_rate": 5.902676700472874e-06,
      "loss": 0.1023,
      "step": 402
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.172246680213261,
      "learning_rate": 5.884451375612865e-06,
      "loss": 0.0859,
      "step": 403
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0096049350842804,
      "learning_rate": 5.8662139117921435e-06,
      "loss": 0.0952,
      "step": 404
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3924332642607742,
      "learning_rate": 5.847964559317129e-06,
      "loss": 0.075,
      "step": 405
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.555222595194896,
      "learning_rate": 5.829703568657406e-06,
      "loss": 0.1194,
      "step": 406
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.814069025220444,
      "learning_rate": 5.8114311904423e-06,
      "loss": 0.1114,
      "step": 407
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.63589934143905,
      "learning_rate": 5.793147675457425e-06,
      "loss": 0.0997,
      "step": 408
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.672874605967912,
      "learning_rate": 5.774853274641243e-06,
      "loss": 0.0936,
      "step": 409
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.293398379507954,
      "learning_rate": 5.756548239081626e-06,
      "loss": 0.0936,
      "step": 410
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3509632172781578,
      "learning_rate": 5.738232820012407e-06,
      "loss": 0.1014,
      "step": 411
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4424133826280734,
      "learning_rate": 5.719907268809927e-06,
      "loss": 0.0942,
      "step": 412
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6011921811511678,
      "learning_rate": 5.701571836989591e-06,
      "loss": 0.0839,
      "step": 413
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.1072468499647248,
      "learning_rate": 5.683226776202416e-06,
      "loss": 0.1096,
      "step": 414
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3764568994495563,
      "learning_rate": 5.664872338231572e-06,
      "loss": 0.1001,
      "step": 415
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.53101947882973,
      "learning_rate": 5.64650877498893e-06,
      "loss": 0.0885,
      "step": 416
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.9958749325869098,
      "learning_rate": 5.628136338511607e-06,
      "loss": 0.1067,
      "step": 417
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1068155120562846,
      "learning_rate": 5.6097552809584996e-06,
      "loss": 0.0862,
      "step": 418
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.625671593043311,
      "learning_rate": 5.591365854606829e-06,
      "loss": 0.0954,
      "step": 419
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8652769478201479,
      "learning_rate": 5.5729683118486785e-06,
      "loss": 0.0848,
      "step": 420
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.0731466858910217,
      "learning_rate": 5.554562905187527e-06,
      "loss": 0.089,
      "step": 421
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6660858543441386,
      "learning_rate": 5.536149887234781e-06,
      "loss": 0.0982,
      "step": 422
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.552222550169945,
      "learning_rate": 5.517729510706316e-06,
      "loss": 0.1154,
      "step": 423
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3368802229091306,
      "learning_rate": 5.499302028418998e-06,
      "loss": 0.0976,
      "step": 424
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.181126697333057,
      "learning_rate": 5.480867693287224e-06,
      "loss": 0.1062,
      "step": 425
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.9205645782423137,
      "learning_rate": 5.462426758319439e-06,
      "loss": 0.0911,
      "step": 426
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.9632409923514775,
      "learning_rate": 5.443979476614674e-06,
      "loss": 0.1095,
      "step": 427
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1362611723204445,
      "learning_rate": 5.425526101359068e-06,
      "loss": 0.0777,
      "step": 428
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3238503828238188,
      "learning_rate": 5.407066885822391e-06,
      "loss": 0.0824,
      "step": 429
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5813044674729215,
      "learning_rate": 5.388602083354572e-06,
      "loss": 0.1091,
      "step": 430
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2656866672687868,
      "learning_rate": 5.370131947382215e-06,
      "loss": 0.1061,
      "step": 431
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3312297948060596,
      "learning_rate": 5.351656731405132e-06,
      "loss": 0.0823,
      "step": 432
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4157388997754832,
      "learning_rate": 5.333176688992856e-06,
      "loss": 0.0792,
      "step": 433
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4094464298882508,
      "learning_rate": 5.314692073781157e-06,
      "loss": 0.101,
      "step": 434
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4099498232807148,
      "learning_rate": 5.296203139468572e-06,
      "loss": 0.0865,
      "step": 435
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.8502414411111636,
      "learning_rate": 5.2777101398129135e-06,
      "loss": 0.0889,
      "step": 436
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.7560344587273293,
      "learning_rate": 5.2592133286277925e-06,
      "loss": 0.0916,
      "step": 437
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6710267518378974,
      "learning_rate": 5.2407129597791295e-06,
      "loss": 0.1047,
      "step": 438
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2201280272734387,
      "learning_rate": 5.222209287181677e-06,
      "loss": 0.0891,
      "step": 439
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4358152308072225,
      "learning_rate": 5.203702564795528e-06,
      "loss": 0.0802,
      "step": 440
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8914507091018575,
      "learning_rate": 5.185193046622636e-06,
      "loss": 0.0782,
      "step": 441
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6165198764693776,
      "learning_rate": 5.166680986703323e-06,
      "loss": 0.1246,
      "step": 442
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4298748482799117,
      "learning_rate": 5.148166639112799e-06,
      "loss": 0.105,
      "step": 443
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.535559818853895,
      "learning_rate": 5.129650257957671e-06,
      "loss": 0.1119,
      "step": 444
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.9197842071730942,
      "learning_rate": 5.111132097372459e-06,
      "loss": 0.1033,
      "step": 445
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1599006332138146,
      "learning_rate": 5.0926124115160976e-06,
      "loss": 0.0896,
      "step": 446
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3974819169152706,
      "learning_rate": 5.074091454568464e-06,
      "loss": 0.0969,
      "step": 447
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9408016749032659,
      "learning_rate": 5.055569480726876e-06,
      "loss": 0.0895,
      "step": 448
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.9952244551309546,
      "learning_rate": 5.0370467442026115e-06,
      "loss": 0.0871,
      "step": 449
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.36586105745886,
      "learning_rate": 5.0185234992174114e-06,
      "loss": 0.0844,
      "step": 450
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.7479502067142012,
      "learning_rate": 5e-06,
      "loss": 0.1044,
      "step": 451
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3573048964331627,
      "learning_rate": 4.981476500782591e-06,
      "loss": 0.0948,
      "step": 452
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.473812452772602,
      "learning_rate": 4.96295325579739e-06,
      "loss": 0.0797,
      "step": 453
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.0296959916896,
      "learning_rate": 4.944430519273126e-06,
      "loss": 0.0895,
      "step": 454
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2103951372488675,
      "learning_rate": 4.925908545431537e-06,
      "loss": 0.0915,
      "step": 455
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.434571480202637,
      "learning_rate": 4.907387588483903e-06,
      "loss": 0.0962,
      "step": 456
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2049199119312812,
      "learning_rate": 4.8888679026275436e-06,
      "loss": 0.0909,
      "step": 457
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.7639545815530557,
      "learning_rate": 4.870349742042329e-06,
      "loss": 0.1117,
      "step": 458
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.1494030416261993,
      "learning_rate": 4.8518333608872015e-06,
      "loss": 0.0752,
      "step": 459
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6445773991618207,
      "learning_rate": 4.8333190132966794e-06,
      "loss": 0.0861,
      "step": 460
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6414731186074902,
      "learning_rate": 4.814806953377366e-06,
      "loss": 0.0829,
      "step": 461
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.8968387997428107,
      "learning_rate": 4.796297435204473e-06,
      "loss": 0.0887,
      "step": 462
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.25439526816161,
      "learning_rate": 4.777790712818324e-06,
      "loss": 0.075,
      "step": 463
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.1014332021036057,
      "learning_rate": 4.759287040220872e-06,
      "loss": 0.0719,
      "step": 464
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2580923566775155,
      "learning_rate": 4.740786671372209e-06,
      "loss": 0.0955,
      "step": 465
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.7123882846797445,
      "learning_rate": 4.7222898601870864e-06,
      "loss": 0.0935,
      "step": 466
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.178774620445043,
      "learning_rate": 4.703796860531429e-06,
      "loss": 0.0987,
      "step": 467
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6539278144919904,
      "learning_rate": 4.6853079262188445e-06,
      "loss": 0.1124,
      "step": 468
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0935968959871103,
      "learning_rate": 4.666823311007145e-06,
      "loss": 0.0871,
      "step": 469
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3166304048687854,
      "learning_rate": 4.648343268594869e-06,
      "loss": 0.0749,
      "step": 470
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.8561652305018204,
      "learning_rate": 4.629868052617786e-06,
      "loss": 0.0847,
      "step": 471
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.414637359104844,
      "learning_rate": 4.61139791664543e-06,
      "loss": 0.0887,
      "step": 472
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2510852736375488,
      "learning_rate": 4.59293311417761e-06,
      "loss": 0.0713,
      "step": 473
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.404573593610514,
      "learning_rate": 4.574473898640933e-06,
      "loss": 0.0863,
      "step": 474
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.1678998858209273,
      "learning_rate": 4.556020523385326e-06,
      "loss": 0.1042,
      "step": 475
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4292796447931189,
      "learning_rate": 4.537573241680563e-06,
      "loss": 0.0898,
      "step": 476
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.8305004593870045,
      "learning_rate": 4.519132306712778e-06,
      "loss": 0.0883,
      "step": 477
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.706985080173823,
      "learning_rate": 4.500697971581003e-06,
      "loss": 0.0628,
      "step": 478
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5783162958105084,
      "learning_rate": 4.482270489293685e-06,
      "loss": 0.0899,
      "step": 479
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2211595747222304,
      "learning_rate": 4.463850112765221e-06,
      "loss": 0.0754,
      "step": 480
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.995983190967539,
      "learning_rate": 4.445437094812476e-06,
      "loss": 0.0887,
      "step": 481
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.428898296054049,
      "learning_rate": 4.4270316881513215e-06,
      "loss": 0.1133,
      "step": 482
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4225799563091075,
      "learning_rate": 4.408634145393172e-06,
      "loss": 0.1184,
      "step": 483
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3918275141803165,
      "learning_rate": 4.390244719041502e-06,
      "loss": 0.0648,
      "step": 484
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.367999116075585,
      "learning_rate": 4.371863661488394e-06,
      "loss": 0.0853,
      "step": 485
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.7079182185930346,
      "learning_rate": 4.353491225011071e-06,
      "loss": 0.0949,
      "step": 486
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5010142904615251,
      "learning_rate": 4.335127661768429e-06,
      "loss": 0.0641,
      "step": 487
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.7284173282169324,
      "learning_rate": 4.316773223797585e-06,
      "loss": 0.0703,
      "step": 488
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5621635512490386,
      "learning_rate": 4.298428163010411e-06,
      "loss": 0.0869,
      "step": 489
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.422341370982355,
      "learning_rate": 4.280092731190075e-06,
      "loss": 0.0799,
      "step": 490
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.7501446138446721,
      "learning_rate": 4.261767179987595e-06,
      "loss": 0.0897,
      "step": 491
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2772834345032038,
      "learning_rate": 4.243451760918376e-06,
      "loss": 0.1065,
      "step": 492
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7832782732498063,
      "learning_rate": 4.225146725358759e-06,
      "loss": 0.0806,
      "step": 493
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9590851368962775,
      "learning_rate": 4.206852324542578e-06,
      "loss": 0.0834,
      "step": 494
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5291845240945432,
      "learning_rate": 4.188568809557701e-06,
      "loss": 0.0935,
      "step": 495
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.774265127813176,
      "learning_rate": 4.170296431342595e-06,
      "loss": 0.0776,
      "step": 496
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6751404930422793,
      "learning_rate": 4.152035440682874e-06,
      "loss": 0.1102,
      "step": 497
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1017368315986533,
      "learning_rate": 4.1337860882078564e-06,
      "loss": 0.0842,
      "step": 498
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0095647614965828,
      "learning_rate": 4.115548624387136e-06,
      "loss": 0.1033,
      "step": 499
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1850618393493437,
      "learning_rate": 4.097323299527129e-06,
      "loss": 0.0935,
      "step": 500
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6016802129348722,
      "learning_rate": 4.079110363767649e-06,
      "loss": 0.0957,
      "step": 501
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.7614328270624333,
      "learning_rate": 4.060910067078475e-06,
      "loss": 0.0932,
      "step": 502
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1784018017105278,
      "learning_rate": 4.042722659255907e-06,
      "loss": 0.0947,
      "step": 503
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9536882155687711,
      "learning_rate": 4.02454838991936e-06,
      "loss": 0.0864,
      "step": 504
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3412379988333587,
      "learning_rate": 4.006387508507914e-06,
      "loss": 0.0815,
      "step": 505
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9937123395901131,
      "learning_rate": 3.988240264276906e-06,
      "loss": 0.0837,
      "step": 506
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1118836347253336,
      "learning_rate": 3.970106906294509e-06,
      "loss": 0.1172,
      "step": 507
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2375186671145577,
      "learning_rate": 3.951987683438301e-06,
      "loss": 0.0799,
      "step": 508
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.9232301509219243,
      "learning_rate": 3.933882844391866e-06,
      "loss": 0.1145,
      "step": 509
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5256990396302608,
      "learning_rate": 3.9157926376413676e-06,
      "loss": 0.1008,
      "step": 510
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.737700461488523,
      "learning_rate": 3.897717311472141e-06,
      "loss": 0.1239,
      "step": 511
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.36273898878739,
      "learning_rate": 3.879657113965294e-06,
      "loss": 0.0945,
      "step": 512
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0718267007797795,
      "learning_rate": 3.861612292994293e-06,
      "loss": 0.0899,
      "step": 513
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6888992492780912,
      "learning_rate": 3.843583096221559e-06,
      "loss": 0.0994,
      "step": 514
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0582879934749914,
      "learning_rate": 3.825569771095082e-06,
      "loss": 0.1048,
      "step": 515
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9288548490683142,
      "learning_rate": 3.80757256484501e-06,
      "loss": 0.0913,
      "step": 516
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.488206708121423,
      "learning_rate": 3.789591724480266e-06,
      "loss": 0.092,
      "step": 517
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.396811130524263,
      "learning_rate": 3.7716274967851485e-06,
      "loss": 0.0923,
      "step": 518
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1865029532635987,
      "learning_rate": 3.7536801283159523e-06,
      "loss": 0.1021,
      "step": 519
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1926573230493913,
      "learning_rate": 3.7357498653975854e-06,
      "loss": 0.1041,
      "step": 520
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.7369114629342763,
      "learning_rate": 3.7178369541201785e-06,
      "loss": 0.098,
      "step": 521
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3911362868676322,
      "learning_rate": 3.6999416403357158e-06,
      "loss": 0.0876,
      "step": 522
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3865651156084833,
      "learning_rate": 3.682064169654663e-06,
      "loss": 0.0948,
      "step": 523
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5188940499590486,
      "learning_rate": 3.6642047874425833e-06,
      "loss": 0.1049,
      "step": 524
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.8578901647350845,
      "learning_rate": 3.646363738816788e-06,
      "loss": 0.1042,
      "step": 525
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9682155137321418,
      "learning_rate": 3.6285412686429567e-06,
      "loss": 0.0926,
      "step": 526
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5329919136198829,
      "learning_rate": 3.6107376215317813e-06,
      "loss": 0.0918,
      "step": 527
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3283757375600345,
      "learning_rate": 3.592953041835615e-06,
      "loss": 0.0979,
      "step": 528
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.7426984409997923,
      "learning_rate": 3.5751877736451123e-06,
      "loss": 0.0857,
      "step": 529
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.8343173094126104,
      "learning_rate": 3.557442060785876e-06,
      "loss": 0.1074,
      "step": 530
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3960147690089932,
      "learning_rate": 3.539716146815122e-06,
      "loss": 0.1034,
      "step": 531
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.952266082891725,
      "learning_rate": 3.5220102750183218e-06,
      "loss": 0.0934,
      "step": 532
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1119714305902637,
      "learning_rate": 3.504324688405878e-06,
      "loss": 0.0911,
      "step": 533
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.273976446680227,
      "learning_rate": 3.4866596297097776e-06,
      "loss": 0.0886,
      "step": 534
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0318179315692617,
      "learning_rate": 3.469015341380266e-06,
      "loss": 0.0809,
      "step": 535
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0847335063342978,
      "learning_rate": 3.451392065582521e-06,
      "loss": 0.0707,
      "step": 536
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.319209798209738,
      "learning_rate": 3.433790044193323e-06,
      "loss": 0.0874,
      "step": 537
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2541175082857459,
      "learning_rate": 3.4162095187977395e-06,
      "loss": 0.0764,
      "step": 538
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0523487335558295,
      "learning_rate": 3.398650730685813e-06,
      "loss": 0.0864,
      "step": 539
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.7334821654184849,
      "learning_rate": 3.3811139208492384e-06,
      "loss": 0.0868,
      "step": 540
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5858715871111615,
      "learning_rate": 3.3635993299780667e-06,
      "loss": 0.0993,
      "step": 541
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.295933466624895,
      "learning_rate": 3.3461071984573983e-06,
      "loss": 0.1178,
      "step": 542
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5080133595230978,
      "learning_rate": 3.3286377663640753e-06,
      "loss": 0.1269,
      "step": 543
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.7043750113015401,
      "learning_rate": 3.311191273463401e-06,
      "loss": 0.0982,
      "step": 544
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3460953104218514,
      "learning_rate": 3.29376795920584e-06,
      "loss": 0.0918,
      "step": 545
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.424318713178504,
      "learning_rate": 3.276368062723729e-06,
      "loss": 0.0935,
      "step": 546
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6851360859318019,
      "learning_rate": 3.258991822828007e-06,
      "loss": 0.0879,
      "step": 547
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.036620519893431,
      "learning_rate": 3.241639478004922e-06,
      "loss": 0.0961,
      "step": 548
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.320247516635966,
      "learning_rate": 3.224311266412773e-06,
      "loss": 0.0901,
      "step": 549
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3883416513760776,
      "learning_rate": 3.2070074258786287e-06,
      "loss": 0.0819,
      "step": 550
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1731384518668249,
      "learning_rate": 3.1897281938950693e-06,
      "loss": 0.0688,
      "step": 551
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4524331811565157,
      "learning_rate": 3.1724738076169314e-06,
      "loss": 0.102,
      "step": 552
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3664712580781437,
      "learning_rate": 3.1552445038580415e-06,
      "loss": 0.1001,
      "step": 553
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.9253928104639182,
      "learning_rate": 3.138040519087975e-06,
      "loss": 0.0782,
      "step": 554
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.7784718748270076,
      "learning_rate": 3.1208620894288105e-06,
      "loss": 0.1103,
      "step": 555
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.9332032343565215,
      "learning_rate": 3.1037094506518806e-06,
      "loss": 0.1249,
      "step": 556
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2481471142608356,
      "learning_rate": 3.0865828381745515e-06,
      "loss": 0.0929,
      "step": 557
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.7804865693694558,
      "learning_rate": 3.069482487056974e-06,
      "loss": 0.1004,
      "step": 558
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6441870832230248,
      "learning_rate": 3.0524086319988635e-06,
      "loss": 0.0635,
      "step": 559
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4112352863599653,
      "learning_rate": 3.0353615073362937e-06,
      "loss": 0.0852,
      "step": 560
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.679818456248062,
      "learning_rate": 3.0183413470384537e-06,
      "loss": 0.1063,
      "step": 561
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0605912128021044,
      "learning_rate": 3.001348384704461e-06,
      "loss": 0.0903,
      "step": 562
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6370195416483042,
      "learning_rate": 2.98438285356014e-06,
      "loss": 0.0886,
      "step": 563
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2563461066733568,
      "learning_rate": 2.967444986454825e-06,
      "loss": 0.0855,
      "step": 564
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3138148878284275,
      "learning_rate": 2.95053501585817e-06,
      "loss": 0.0954,
      "step": 565
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.102694770789112,
      "learning_rate": 2.933653173856953e-06,
      "loss": 0.0772,
      "step": 566
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0199400816615287,
      "learning_rate": 2.9167996921518848e-06,
      "loss": 0.0925,
      "step": 567
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8547500043718359,
      "learning_rate": 2.899974802054446e-06,
      "loss": 0.0944,
      "step": 568
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3931067354085072,
      "learning_rate": 2.8831787344836926e-06,
      "loss": 0.0993,
      "step": 569
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.2303435090125667,
      "learning_rate": 2.8664117199631023e-06,
      "loss": 0.1219,
      "step": 570
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9215955759772718,
      "learning_rate": 2.8496739886173994e-06,
      "loss": 0.0861,
      "step": 571
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6413872734862562,
      "learning_rate": 2.8329657701694015e-06,
      "loss": 0.0893,
      "step": 572
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.193727034493147,
      "learning_rate": 2.816287293936868e-06,
      "loss": 0.0795,
      "step": 573
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2289155652249402,
      "learning_rate": 2.799638788829354e-06,
      "loss": 0.0737,
      "step": 574
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.830345925001274,
      "learning_rate": 2.7830204833450577e-06,
      "loss": 0.0831,
      "step": 575
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.068546819825466,
      "learning_rate": 2.766432605567699e-06,
      "loss": 0.0668,
      "step": 576
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9453172607728504,
      "learning_rate": 2.7498753831633773e-06,
      "loss": 0.0901,
      "step": 577
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.586371621963873,
      "learning_rate": 2.733349043377457e-06,
      "loss": 0.1053,
      "step": 578
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5289078110540708,
      "learning_rate": 2.716853813031435e-06,
      "loss": 0.1089,
      "step": 579
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.633488144896007,
      "learning_rate": 2.7003899185198445e-06,
      "loss": 0.1071,
      "step": 580
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4665200970556465,
      "learning_rate": 2.6839575858071367e-06,
      "loss": 0.0867,
      "step": 581
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.141767875960879,
      "learning_rate": 2.6675570404245783e-06,
      "loss": 0.0814,
      "step": 582
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5201605416934159,
      "learning_rate": 2.651188507467161e-06,
      "loss": 0.0977,
      "step": 583
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4871770110745552,
      "learning_rate": 2.634852211590516e-06,
      "loss": 0.0781,
      "step": 584
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3369457391490027,
      "learning_rate": 2.6185483770078173e-06,
      "loss": 0.0733,
      "step": 585
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3086109100448613,
      "learning_rate": 2.602277227486719e-06,
      "loss": 0.1085,
      "step": 586
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3624853079150907,
      "learning_rate": 2.5860389863462765e-06,
      "loss": 0.0865,
      "step": 587
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.887651265201604,
      "learning_rate": 2.569833876453879e-06,
      "loss": 0.075,
      "step": 588
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6592764745581379,
      "learning_rate": 2.553662120222199e-06,
      "loss": 0.0772,
      "step": 589
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.145073711872565,
      "learning_rate": 2.537523939606133e-06,
      "loss": 0.0831,
      "step": 590
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.0550075999705046,
      "learning_rate": 2.5214195560997546e-06,
      "loss": 0.089,
      "step": 591
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.316393420334233,
      "learning_rate": 2.505349190733282e-06,
      "loss": 0.0991,
      "step": 592
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0938355449466903,
      "learning_rate": 2.489313064070037e-06,
      "loss": 0.0888,
      "step": 593
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9546003971132232,
      "learning_rate": 2.4733113962034234e-06,
      "loss": 0.0863,
      "step": 594
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.274927581713751,
      "learning_rate": 2.457344406753899e-06,
      "loss": 0.0866,
      "step": 595
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5245485802819234,
      "learning_rate": 2.441412314865965e-06,
      "loss": 0.0941,
      "step": 596
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3490782992266146,
      "learning_rate": 2.4255153392051655e-06,
      "loss": 0.0825,
      "step": 597
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.973182972399078,
      "learning_rate": 2.4096536979550716e-06,
      "loss": 0.0931,
      "step": 598
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2896193150614899,
      "learning_rate": 2.3938276088143003e-06,
      "loss": 0.0848,
      "step": 599
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0831644395119333,
      "learning_rate": 2.3780372889935223e-06,
      "loss": 0.0768,
      "step": 600
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.965364721243873,
      "learning_rate": 2.362282955212473e-06,
      "loss": 0.0818,
      "step": 601
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0958775941489052,
      "learning_rate": 2.346564823696994e-06,
      "loss": 0.0853,
      "step": 602
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2297992304308838,
      "learning_rate": 2.330883110176049e-06,
      "loss": 0.0843,
      "step": 603
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2258534468894116,
      "learning_rate": 2.31523802987877e-06,
      "loss": 0.0833,
      "step": 604
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2676625578470901,
      "learning_rate": 2.29962979753151e-06,
      "loss": 0.0885,
      "step": 605
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.8759537272989633,
      "learning_rate": 2.284058627354887e-06,
      "loss": 0.1015,
      "step": 606
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5621051081143618,
      "learning_rate": 2.2685247330608417e-06,
      "loss": 0.0877,
      "step": 607
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.070142424838425,
      "learning_rate": 2.253028327849716e-06,
      "loss": 0.0975,
      "step": 608
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2968662084013451,
      "learning_rate": 2.2375696244073126e-06,
      "loss": 0.1027,
      "step": 609
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1236440141208273,
      "learning_rate": 2.2221488349019903e-06,
      "loss": 0.084,
      "step": 610
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2823682175649511,
      "learning_rate": 2.2067661709817384e-06,
      "loss": 0.0632,
      "step": 611
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2915537305568103,
      "learning_rate": 2.1914218437712824e-06,
      "loss": 0.098,
      "step": 612
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7053759254724352,
      "learning_rate": 2.1761160638691843e-06,
      "loss": 0.0902,
      "step": 613
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1559649931446625,
      "learning_rate": 2.1608490413449428e-06,
      "loss": 0.0765,
      "step": 614
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.9421156314939874,
      "learning_rate": 2.145620985736125e-06,
      "loss": 0.0868,
      "step": 615
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0296488915584479,
      "learning_rate": 2.130432106045477e-06,
      "loss": 0.0916,
      "step": 616
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1090554359149103,
      "learning_rate": 2.1152826107380652e-06,
      "loss": 0.0916,
      "step": 617
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0976260165303318,
      "learning_rate": 2.1001727077384133e-06,
      "loss": 0.1014,
      "step": 618
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1628149430896515,
      "learning_rate": 2.0851026044276405e-06,
      "loss": 0.0848,
      "step": 619
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2655603395728712,
      "learning_rate": 2.0700725076406235e-06,
      "loss": 0.0943,
      "step": 620
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6340092128585122,
      "learning_rate": 2.05508262366316e-06,
      "loss": 0.0789,
      "step": 621
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3978109379704229,
      "learning_rate": 2.0401331582291258e-06,
      "loss": 0.095,
      "step": 622
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9953586497854784,
      "learning_rate": 2.025224316517663e-06,
      "loss": 0.0683,
      "step": 623
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1514532520993712,
      "learning_rate": 2.0103563031503613e-06,
      "loss": 0.0813,
      "step": 624
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0382979576594575,
      "learning_rate": 1.9955293221884403e-06,
      "loss": 0.0881,
      "step": 625
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1852249996471935,
      "learning_rate": 1.9807435771299644e-06,
      "loss": 0.0782,
      "step": 626
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.788558970046365,
      "learning_rate": 1.9659992709070346e-06,
      "loss": 0.0869,
      "step": 627
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3959129722582788,
      "learning_rate": 1.9512966058830125e-06,
      "loss": 0.1102,
      "step": 628
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.8621826836279993,
      "learning_rate": 1.9366357838497423e-06,
      "loss": 0.1095,
      "step": 629
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.437311145687304,
      "learning_rate": 1.9220170060247783e-06,
      "loss": 0.0796,
      "step": 630
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4068547957795265,
      "learning_rate": 1.9074404730486264e-06,
      "loss": 0.0857,
      "step": 631
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0094769073015202,
      "learning_rate": 1.8929063849819846e-06,
      "loss": 0.1138,
      "step": 632
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3064844865279543,
      "learning_rate": 1.8784149413030006e-06,
      "loss": 0.0844,
      "step": 633
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2958603015342525,
      "learning_rate": 1.8639663409045405e-06,
      "loss": 0.081,
      "step": 634
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1833983355282296,
      "learning_rate": 1.8495607820914451e-06,
      "loss": 0.1007,
      "step": 635
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.142884197140577,
      "learning_rate": 1.835198462577822e-06,
      "loss": 0.0764,
      "step": 636
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4586827243798255,
      "learning_rate": 1.820879579484325e-06,
      "loss": 0.0958,
      "step": 637
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.5290567861521294,
      "learning_rate": 1.8066043293354447e-06,
      "loss": 0.1029,
      "step": 638
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.534517648492461,
      "learning_rate": 1.7923729080568242e-06,
      "loss": 0.1011,
      "step": 639
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.513440785236101,
      "learning_rate": 1.778185510972556e-06,
      "loss": 0.0819,
      "step": 640
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4095512013043103,
      "learning_rate": 1.7640423328025063e-06,
      "loss": 0.0867,
      "step": 641
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0547200042580467,
      "learning_rate": 1.7499435676596482e-06,
      "loss": 0.0692,
      "step": 642
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2896781158690789,
      "learning_rate": 1.7358894090473928e-06,
      "loss": 0.0826,
      "step": 643
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.0540774389675582,
      "learning_rate": 1.721880049856927e-06,
      "loss": 0.1067,
      "step": 644
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.249361727592556,
      "learning_rate": 1.7079156823645805e-06,
      "loss": 0.0964,
      "step": 645
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3933011021656423,
      "learning_rate": 1.6939964982291713e-06,
      "loss": 0.0828,
      "step": 646
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1240586077405685,
      "learning_rate": 1.6801226884893895e-06,
      "loss": 0.0992,
      "step": 647
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8382368133177154,
      "learning_rate": 1.6662944435611622e-06,
      "loss": 0.0695,
      "step": 648
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.7300003312511045,
      "learning_rate": 1.652511953235051e-06,
      "loss": 0.1086,
      "step": 649
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2259941514107384,
      "learning_rate": 1.6387754066736422e-06,
      "loss": 0.0795,
      "step": 650
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3255066128333746,
      "learning_rate": 1.6250849924089485e-06,
      "loss": 0.0893,
      "step": 651
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8199741391521307,
      "learning_rate": 1.611440898339824e-06,
      "loss": 0.0606,
      "step": 652
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9774400053985267,
      "learning_rate": 1.5978433117293884e-06,
      "loss": 0.0878,
      "step": 653
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5814733538185348,
      "learning_rate": 1.5842924192024489e-06,
      "loss": 0.0922,
      "step": 654
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0642541089350566,
      "learning_rate": 1.5707884067429474e-06,
      "loss": 0.0868,
      "step": 655
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4337445102024018,
      "learning_rate": 1.5573314596914046e-06,
      "loss": 0.0849,
      "step": 656
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.882421180731485,
      "learning_rate": 1.5439217627423696e-06,
      "loss": 0.0663,
      "step": 657
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5004750860759757,
      "learning_rate": 1.5305594999418989e-06,
      "loss": 0.0865,
      "step": 658
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.412572230637191,
      "learning_rate": 1.5172448546850166e-06,
      "loss": 0.0879,
      "step": 659
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.727267767004979,
      "learning_rate": 1.5039780097132046e-06,
      "loss": 0.087,
      "step": 660
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6199501511796164,
      "learning_rate": 1.4907591471118942e-06,
      "loss": 0.0633,
      "step": 661
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.196467314248605,
      "learning_rate": 1.4775884483079666e-06,
      "loss": 0.0905,
      "step": 662
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6865314637877673,
      "learning_rate": 1.4644660940672628e-06,
      "loss": 0.095,
      "step": 663
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9948746939122228,
      "learning_rate": 1.4513922644920985e-06,
      "loss": 0.0672,
      "step": 664
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3057924965780205,
      "learning_rate": 1.4383671390187965e-06,
      "loss": 0.0918,
      "step": 665
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3739151217740653,
      "learning_rate": 1.4253908964152252e-06,
      "loss": 0.0784,
      "step": 666
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3048795820201535,
      "learning_rate": 1.4124637147783431e-06,
      "loss": 0.099,
      "step": 667
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8009570850923019,
      "learning_rate": 1.399585771531755e-06,
      "loss": 0.0735,
      "step": 668
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9489677584418744,
      "learning_rate": 1.386757243423273e-06,
      "loss": 0.0867,
      "step": 669
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2187793313789825,
      "learning_rate": 1.3739783065224944e-06,
      "loss": 0.0753,
      "step": 670
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2771944302953684,
      "learning_rate": 1.3612491362183887e-06,
      "loss": 0.108,
      "step": 671
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.114716915199913,
      "learning_rate": 1.3485699072168812e-06,
      "loss": 0.0946,
      "step": 672
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.104197818175097,
      "learning_rate": 1.3359407935384644e-06,
      "loss": 0.0828,
      "step": 673
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4388633932295818,
      "learning_rate": 1.3233619685158056e-06,
      "loss": 0.0751,
      "step": 674
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1502743745660746,
      "learning_rate": 1.3108336047913633e-06,
      "loss": 0.0823,
      "step": 675
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.302715374352943,
      "learning_rate": 1.2983558743150287e-06,
      "loss": 0.0747,
      "step": 676
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2518301338351348,
      "learning_rate": 1.2859289483417558e-06,
      "loss": 0.0764,
      "step": 677
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.250777735934068,
      "learning_rate": 1.2735529974292122e-06,
      "loss": 0.1011,
      "step": 678
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1136692815348772,
      "learning_rate": 1.2612281914354452e-06,
      "loss": 0.0955,
      "step": 679
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.538562671017371,
      "learning_rate": 1.2489546995165452e-06,
      "loss": 0.097,
      "step": 680
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1759591419244437,
      "learning_rate": 1.2367326901243215e-06,
      "loss": 0.0805,
      "step": 681
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.313266929061223,
      "learning_rate": 1.2245623310039973e-06,
      "loss": 0.0992,
      "step": 682
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9182273134449173,
      "learning_rate": 1.2124437891918995e-06,
      "loss": 0.0708,
      "step": 683
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.906205039600781,
      "learning_rate": 1.200377231013176e-06,
      "loss": 0.0784,
      "step": 684
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1516839769243379,
      "learning_rate": 1.1883628220795008e-06,
      "loss": 0.0928,
      "step": 685
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2813185479825682,
      "learning_rate": 1.1764007272868116e-06,
      "loss": 0.0726,
      "step": 686
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9213558082977095,
      "learning_rate": 1.1644911108130436e-06,
      "loss": 0.097,
      "step": 687
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0500113768635098,
      "learning_rate": 1.152634136115871e-06,
      "loss": 0.0733,
      "step": 688
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4068990440454445,
      "learning_rate": 1.1408299659304684e-06,
      "loss": 0.1059,
      "step": 689
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3695680323420723,
      "learning_rate": 1.1290787622672806e-06,
      "loss": 0.08,
      "step": 690
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2133479596639074,
      "learning_rate": 1.1173806864097885e-06,
      "loss": 0.1077,
      "step": 691
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.98289755771727,
      "learning_rate": 1.1057358989123075e-06,
      "loss": 0.0828,
      "step": 692
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9289925027488103,
      "learning_rate": 1.0941445595977768e-06,
      "loss": 0.0715,
      "step": 693
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2183961669443226,
      "learning_rate": 1.0826068275555652e-06,
      "loss": 0.0759,
      "step": 694
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7469515955726049,
      "learning_rate": 1.0711228611392937e-06,
      "loss": 0.0601,
      "step": 695
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8905767402220537,
      "learning_rate": 1.0596928179646544e-06,
      "loss": 0.0673,
      "step": 696
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.377778311548707,
      "learning_rate": 1.048316854907252e-06,
      "loss": 0.0906,
      "step": 697
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8788045612493512,
      "learning_rate": 1.0369951281004514e-06,
      "loss": 0.0703,
      "step": 698
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.33875040003829,
      "learning_rate": 1.0257277929332332e-06,
      "loss": 0.0969,
      "step": 699
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.081307813191564,
      "learning_rate": 1.0145150040480606e-06,
      "loss": 0.11,
      "step": 700
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8860649979755456,
      "learning_rate": 1.0033569153387563e-06,
      "loss": 0.0852,
      "step": 701
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0736324612290742,
      "learning_rate": 9.9225367994839e-07,
      "loss": 0.0934,
      "step": 702
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9872829827731274,
      "learning_rate": 9.812054502671835e-07,
      "loss": 0.0799,
      "step": 703
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9282388618215119,
      "learning_rate": 9.702123779304074e-07,
      "loss": 0.1006,
      "step": 704
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9435099327920515,
      "learning_rate": 9.592746138163096e-07,
      "loss": 0.082,
      "step": 705
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8088817691415676,
      "learning_rate": 9.48392308044041e-07,
      "loss": 0.0873,
      "step": 706
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5352041214674343,
      "learning_rate": 9.375656099715935e-07,
      "loss": 0.0996,
      "step": 707
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9852893987107155,
      "learning_rate": 9.267946681937545e-07,
      "loss": 0.0818,
      "step": 708
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5229064672434325,
      "learning_rate": 9.16079630540061e-07,
      "loss": 0.1036,
      "step": 709
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2208544307050118,
      "learning_rate": 9.054206440727764e-07,
      "loss": 0.1008,
      "step": 710
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0719293706407658,
      "learning_rate": 8.948178550848702e-07,
      "loss": 0.1068,
      "step": 711
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6953289616059486,
      "learning_rate": 8.84271409098012e-07,
      "loss": 0.0933,
      "step": 712
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7642497146531543,
      "learning_rate": 8.737814508605674e-07,
      "loss": 0.0801,
      "step": 713
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4547072421888465,
      "learning_rate": 8.63348124345621e-07,
      "loss": 0.1064,
      "step": 714
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9751721208506822,
      "learning_rate": 8.529715727489912e-07,
      "loss": 0.095,
      "step": 715
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0346592348396086,
      "learning_rate": 8.426519384872733e-07,
      "loss": 0.052,
      "step": 716
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.149521366909892,
      "learning_rate": 8.323893631958807e-07,
      "loss": 0.0937,
      "step": 717
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9997697767501313,
      "learning_rate": 8.221839877270966e-07,
      "loss": 0.0712,
      "step": 718
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3762790723922584,
      "learning_rate": 8.120359521481502e-07,
      "loss": 0.0939,
      "step": 719
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1717426287446866,
      "learning_rate": 8.019453957392865e-07,
      "loss": 0.0991,
      "step": 720
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5011600992405558,
      "learning_rate": 7.919124569918602e-07,
      "loss": 0.0705,
      "step": 721
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.044248767738908,
      "learning_rate": 7.81937273606429e-07,
      "loss": 0.0829,
      "step": 722
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.193338514115385,
      "learning_rate": 7.720199824908692e-07,
      "loss": 0.085,
      "step": 723
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.932214944981213,
      "learning_rate": 7.621607197584963e-07,
      "loss": 0.104,
      "step": 724
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1568238510064872,
      "learning_rate": 7.523596207261907e-07,
      "loss": 0.0808,
      "step": 725
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0930033851467154,
      "learning_rate": 7.426168199125466e-07,
      "loss": 0.0903,
      "step": 726
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9226197490862279,
      "learning_rate": 7.329324510360269e-07,
      "loss": 0.0862,
      "step": 727
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9031890089209195,
      "learning_rate": 7.233066470131205e-07,
      "loss": 0.0736,
      "step": 728
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9118942141479596,
      "learning_rate": 7.13739539956525e-07,
      "loss": 0.0766,
      "step": 729
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.538007969459405,
      "learning_rate": 7.042312611733327e-07,
      "loss": 0.079,
      "step": 730
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8504785007825396,
      "learning_rate": 6.947819411632223e-07,
      "loss": 0.0712,
      "step": 731
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3292424584047728,
      "learning_rate": 6.853917096166768e-07,
      "loss": 0.1028,
      "step": 732
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.8593939850427263,
      "learning_rate": 6.760606954131966e-07,
      "loss": 0.1032,
      "step": 733
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8862139675569493,
      "learning_rate": 6.667890266195321e-07,
      "loss": 0.0605,
      "step": 734
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.0966949364511906,
      "learning_rate": 6.575768304879293e-07,
      "loss": 0.1037,
      "step": 735
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.705445088627103,
      "learning_rate": 6.484242334543806e-07,
      "loss": 0.0803,
      "step": 736
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6749697226888411,
      "learning_rate": 6.393313611368901e-07,
      "loss": 0.065,
      "step": 737
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9923237133863795,
      "learning_rate": 6.302983383337474e-07,
      "loss": 0.0816,
      "step": 738
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4666632314163908,
      "learning_rate": 6.213252890218163e-07,
      "loss": 0.0904,
      "step": 739
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.9654914667201568,
      "learning_rate": 6.124123363548374e-07,
      "loss": 0.1037,
      "step": 740
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9823530441556243,
      "learning_rate": 6.035596026617291e-07,
      "loss": 0.0821,
      "step": 741
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2947434558738595,
      "learning_rate": 5.947672094449169e-07,
      "loss": 0.0734,
      "step": 742
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3295655580056323,
      "learning_rate": 5.860352773786632e-07,
      "loss": 0.0956,
      "step": 743
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0700520145965227,
      "learning_rate": 5.77363926307406e-07,
      "loss": 0.0883,
      "step": 744
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2221120631868163,
      "learning_rate": 5.687532752441232e-07,
      "loss": 0.0916,
      "step": 745
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2132238049137432,
      "learning_rate": 5.602034423686903e-07,
      "loss": 0.0914,
      "step": 746
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.966731842559717,
      "learning_rate": 5.517145450262639e-07,
      "loss": 0.0757,
      "step": 747
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2637487646994567,
      "learning_rate": 5.432866997256708e-07,
      "loss": 0.082,
      "step": 748
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.1793858464928524,
      "learning_rate": 5.349200221378076e-07,
      "loss": 0.0766,
      "step": 749
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6875498904508052,
      "learning_rate": 5.266146270940509e-07,
      "loss": 0.0628,
      "step": 750
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9079148678981082,
      "learning_rate": 5.183706285846873e-07,
      "loss": 0.0967,
      "step": 751
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0095386679421636,
      "learning_rate": 5.101881397573417e-07,
      "loss": 0.0818,
      "step": 752
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2870739096748065,
      "learning_rate": 5.020672729154308e-07,
      "loss": 0.0951,
      "step": 753
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9850619846545391,
      "learning_rate": 4.940081395166174e-07,
      "loss": 0.0499,
      "step": 754
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0382004505125342,
      "learning_rate": 4.860108501712824e-07,
      "loss": 0.0831,
      "step": 755
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8522871618719096,
      "learning_rate": 4.780755146410077e-07,
      "loss": 0.0884,
      "step": 756
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6984622083006453,
      "learning_rate": 4.7020224183706717e-07,
      "loss": 0.0826,
      "step": 757
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2324234124736468,
      "learning_rate": 4.623911398189318e-07,
      "loss": 0.0867,
      "step": 758
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9551492511223306,
      "learning_rate": 4.5464231579279206e-07,
      "loss": 0.0668,
      "step": 759
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.973090902672149,
      "learning_rate": 4.4695587611007786e-07,
      "loss": 0.0741,
      "step": 760
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.221155278221863,
      "learning_rate": 4.393319262660073e-07,
      "loss": 0.0867,
      "step": 761
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.927123352010104,
      "learning_rate": 4.317705708981329e-07,
      "loss": 0.0936,
      "step": 762
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.164545511434888,
      "learning_rate": 4.242719137849077e-07,
      "loss": 0.0849,
      "step": 763
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9999441318021076,
      "learning_rate": 4.168360578442615e-07,
      "loss": 0.0834,
      "step": 764
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4159092264260025,
      "learning_rate": 4.0946310513218733e-07,
      "loss": 0.0862,
      "step": 765
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.126512488995191,
      "learning_rate": 4.021531568413384e-07,
      "loss": 0.0854,
      "step": 766
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1346899479313985,
      "learning_rate": 3.949063132996456e-07,
      "loss": 0.0851,
      "step": 767
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2301055402134775,
      "learning_rate": 3.877226739689355e-07,
      "loss": 0.1078,
      "step": 768
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8267592190940007,
      "learning_rate": 3.8060233744356634e-07,
      "loss": 0.0835,
      "step": 769
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6989450862265247,
      "learning_rate": 3.735454014490747e-07,
      "loss": 0.0621,
      "step": 770
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.091037054539574,
      "learning_rate": 3.665519628408332e-07,
      "loss": 0.0876,
      "step": 771
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1571797362725371,
      "learning_rate": 3.596221176027248e-07,
      "loss": 0.0828,
      "step": 772
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7724986008317762,
      "learning_rate": 3.5275596084582254e-07,
      "loss": 0.0681,
      "step": 773
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1797550383983542,
      "learning_rate": 3.459535868070851e-07,
      "loss": 0.0956,
      "step": 774
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5194629663196104,
      "learning_rate": 3.392150888480611e-07,
      "loss": 0.0766,
      "step": 775
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1717270963830988,
      "learning_rate": 3.3254055945361076e-07,
      "loss": 0.0827,
      "step": 776
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.1502924320409957,
      "learning_rate": 3.259300902306367e-07,
      "loss": 0.0782,
      "step": 777
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9421484362361487,
      "learning_rate": 3.193837719068232e-07,
      "loss": 0.0612,
      "step": 778
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6220845244438815,
      "learning_rate": 3.1290169432939556e-07,
      "loss": 0.0851,
      "step": 779
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2624256131518634,
      "learning_rate": 3.0648394646388304e-07,
      "loss": 0.0744,
      "step": 780
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1192439687232085,
      "learning_rate": 3.0013061639289855e-07,
      "loss": 0.1133,
      "step": 781
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2185909194304914,
      "learning_rate": 2.938417913149344e-07,
      "loss": 0.0892,
      "step": 782
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4523300793441352,
      "learning_rate": 2.8761755754315666e-07,
      "loss": 0.0769,
      "step": 783
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.295435971693865,
      "learning_rate": 2.814580005042283e-07,
      "loss": 0.0659,
      "step": 784
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8957520877828618,
      "learning_rate": 2.753632047371335e-07,
      "loss": 0.0858,
      "step": 785
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6932007112743931,
      "learning_rate": 2.693332538920196e-07,
      "loss": 0.0798,
      "step": 786
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4343179975918245,
      "learning_rate": 2.6336823072904305e-07,
      "loss": 0.0787,
      "step": 787
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7646585417594924,
      "learning_rate": 2.5746821711724215e-07,
      "loss": 0.0842,
      "step": 788
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2296867976725898,
      "learning_rate": 2.51633294033406e-07,
      "loss": 0.0893,
      "step": 789
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8344753721377653,
      "learning_rate": 2.4586354156096783e-07,
      "loss": 0.0856,
      "step": 790
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.5235942296187353,
      "learning_rate": 2.401590388889025e-07,
      "loss": 0.0926,
      "step": 791
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2341731250729187,
      "learning_rate": 2.3451986431064276e-07,
      "loss": 0.0676,
      "step": 792
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.380821266380361,
      "learning_rate": 2.289460952230038e-07,
      "loss": 0.0804,
      "step": 793
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9195607263018777,
      "learning_rate": 2.2343780812511819e-07,
      "loss": 0.0807,
      "step": 794
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3508840786940255,
      "learning_rate": 2.179950786173879e-07,
      "loss": 0.102,
      "step": 795
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2203512555545983,
      "learning_rate": 2.1261798140045086e-07,
      "loss": 0.0804,
      "step": 796
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.947476542013538,
      "learning_rate": 2.0730659027414723e-07,
      "loss": 0.0627,
      "step": 797
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2875569576022898,
      "learning_rate": 2.020609781365135e-07,
      "loss": 0.0697,
      "step": 798
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0738316446109282,
      "learning_rate": 1.9688121698277995e-07,
      "loss": 0.0928,
      "step": 799
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.100302688905556,
      "learning_rate": 1.9176737790437937e-07,
      "loss": 0.0922,
      "step": 800
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9144336671031666,
      "learning_rate": 1.8671953108797825e-07,
      "loss": 0.0847,
      "step": 801
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8267226116425094,
      "learning_rate": 1.8173774581450586e-07,
      "loss": 0.068,
      "step": 802
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2445599352791576,
      "learning_rate": 1.7682209045820687e-07,
      "loss": 0.0826,
      "step": 803
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9745484531589232,
      "learning_rate": 1.7197263248570517e-07,
      "loss": 0.0767,
      "step": 804
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8839595990343955,
      "learning_rate": 1.671894384550743e-07,
      "loss": 0.0772,
      "step": 805
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9753602503357406,
      "learning_rate": 1.6247257401492633e-07,
      "loss": 0.0851,
      "step": 806
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3608102557409236,
      "learning_rate": 1.5782210390350717e-07,
      "loss": 0.0838,
      "step": 807
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.003710736260583,
      "learning_rate": 1.532380919478127e-07,
      "loss": 0.0753,
      "step": 808
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9986605939833132,
      "learning_rate": 1.487206010627118e-07,
      "loss": 0.0852,
      "step": 809
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0489713487182382,
      "learning_rate": 1.4426969325007812e-07,
      "loss": 0.0829,
      "step": 810
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1380778539184384,
      "learning_rate": 1.3988542959794627e-07,
      "loss": 0.0924,
      "step": 811
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.289574338056299,
      "learning_rate": 1.3556787027966867e-07,
      "loss": 0.081,
      "step": 812
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5232518215408282,
      "learning_rate": 1.3131707455309006e-07,
      "loss": 0.0927,
      "step": 813
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3190542373254943,
      "learning_rate": 1.271331007597365e-07,
      "loss": 0.085,
      "step": 814
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.823982217741129,
      "learning_rate": 1.2301600632401212e-07,
      "loss": 0.0583,
      "step": 815
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.9601283751011347,
      "learning_rate": 1.1896584775241194e-07,
      "loss": 0.0951,
      "step": 816
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1845946794586921,
      "learning_rate": 1.1498268063274697e-07,
      "loss": 0.0807,
      "step": 817
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.048858092638161,
      "learning_rate": 1.1106655963338154e-07,
      "loss": 0.0735,
      "step": 818
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0582211099338623,
      "learning_rate": 1.0721753850247984e-07,
      "loss": 0.0627,
      "step": 819
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.285727079837848,
      "learning_rate": 1.034356700672734e-07,
      "loss": 0.1099,
      "step": 820
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5296721973274716,
      "learning_rate": 9.972100623333036e-08,
      "loss": 0.085,
      "step": 821
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4156755488399408,
      "learning_rate": 9.607359798384785e-08,
      "loss": 0.0887,
      "step": 822
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9732631646078624,
      "learning_rate": 9.249349537894969e-08,
      "loss": 0.0848,
      "step": 823
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.20599096937309,
      "learning_rate": 8.89807475550003e-08,
      "loss": 0.0869,
      "step": 824
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8664219455219898,
      "learning_rate": 8.553540272392969e-08,
      "loss": 0.0682,
      "step": 825
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.766439476917645,
      "learning_rate": 8.215750817257173e-08,
      "loss": 0.0781,
      "step": 826
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9177212304579919,
      "learning_rate": 7.884711026201586e-08,
      "loss": 0.0838,
      "step": 827
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.625903722521225,
      "learning_rate": 7.560425442697028e-08,
      "loss": 0.1111,
      "step": 828
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0497108005642395,
      "learning_rate": 7.242898517513864e-08,
      "loss": 0.0921,
      "step": 829
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0158040826453472,
      "learning_rate": 6.932134608660934e-08,
      "loss": 0.0764,
      "step": 830
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.8820874252726552,
      "learning_rate": 6.628137981325611e-08,
      "loss": 0.0859,
      "step": 831
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1111712419679023,
      "learning_rate": 6.330912807815338e-08,
      "loss": 0.0755,
      "step": 832
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9119864524126255,
      "learning_rate": 6.04046316750051e-08,
      "loss": 0.0828,
      "step": 833
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.400402904520679,
      "learning_rate": 5.756793046758302e-08,
      "loss": 0.0862,
      "step": 834
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1178616108463992,
      "learning_rate": 5.479906338917984e-08,
      "loss": 0.0848,
      "step": 835
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7790435488564559,
      "learning_rate": 5.2098068442076325e-08,
      "loss": 0.0657,
      "step": 836
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9151934981942192,
      "learning_rate": 4.946498269701616e-08,
      "loss": 0.0808,
      "step": 837
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6466988196620609,
      "learning_rate": 4.6899842292702504e-08,
      "loss": 0.096,
      "step": 838
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9366718134612106,
      "learning_rate": 4.4402682435296666e-08,
      "loss": 0.0633,
      "step": 839
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2794085716988426,
      "learning_rate": 4.1973537397938524e-08,
      "loss": 0.083,
      "step": 840
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4323100350640852,
      "learning_rate": 3.961244052027413e-08,
      "loss": 0.0726,
      "step": 841
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.442592489577927,
      "learning_rate": 3.731942420799939e-08,
      "loss": 0.108,
      "step": 842
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9558342121619565,
      "learning_rate": 3.5094519932415417e-08,
      "loss": 0.074,
      "step": 843
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2267293855463146,
      "learning_rate": 3.2937758229994455e-08,
      "loss": 0.1133,
      "step": 844
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.377404962474636,
      "learning_rate": 3.084916870196297e-08,
      "loss": 0.0706,
      "step": 845
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0880792886802064,
      "learning_rate": 2.8828780013895885e-08,
      "loss": 0.0749,
      "step": 846
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.127974831685117,
      "learning_rate": 2.6876619895319646e-08,
      "loss": 0.0868,
      "step": 847
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.177548020934705,
      "learning_rate": 2.4992715139335877e-08,
      "loss": 0.0752,
      "step": 848
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.7602271025103597,
      "learning_rate": 2.3177091602251677e-08,
      "loss": 0.0966,
      "step": 849
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0131384087763258,
      "learning_rate": 2.1429774203222674e-08,
      "loss": 0.0825,
      "step": 850
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1479626099193676,
      "learning_rate": 1.975078692391552e-08,
      "loss": 0.0751,
      "step": 851
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0591293526686105,
      "learning_rate": 1.814015280817538e-08,
      "loss": 0.0802,
      "step": 852
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3768999915261486,
      "learning_rate": 1.659789396171063e-08,
      "loss": 0.0772,
      "step": 853
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.9181306911762186,
      "learning_rate": 1.5124031551789208e-08,
      "loss": 0.0696,
      "step": 854
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.955991797622,
      "learning_rate": 1.3718585806949403e-08,
      "loss": 0.0624,
      "step": 855
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9424787636427376,
      "learning_rate": 1.2381576016720076e-08,
      "loss": 0.09,
      "step": 856
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2768624565283,
      "learning_rate": 1.1113020531357543e-08,
      "loss": 0.0679,
      "step": 857
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8292588184596877,
      "learning_rate": 9.91293676159244e-09,
      "loss": 0.0762,
      "step": 858
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4980250112396316,
      "learning_rate": 8.781341178393244e-09,
      "loss": 0.0896,
      "step": 859
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0751217382850593,
      "learning_rate": 7.718249312735903e-09,
      "loss": 0.0669,
      "step": 860
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6280685312977554,
      "learning_rate": 6.7236757553962306e-09,
      "loss": 0.0888,
      "step": 861
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.717122292617126,
      "learning_rate": 5.797634156744502e-09,
      "loss": 0.0992,
      "step": 862
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.9477118843467882,
      "learning_rate": 4.940137226560615e-09,
      "loss": 0.0892,
      "step": 863
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3961547990365242,
      "learning_rate": 4.151196733859775e-09,
      "loss": 0.0694,
      "step": 864
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.039172931689318,
      "learning_rate": 3.4308235067309624e-09,
      "loss": 0.0766,
      "step": 865
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5917603991318112,
      "learning_rate": 2.7790274321859435e-09,
      "loss": 0.1037,
      "step": 866
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0235372378595196,
      "learning_rate": 2.19581745602826e-09,
      "loss": 0.1025,
      "step": 867
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9414557507341034,
      "learning_rate": 1.6812015827255557e-09,
      "loss": 0.0913,
      "step": 868
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9517346204680242,
      "learning_rate": 1.2351868753018858e-09,
      "loss": 0.084,
      "step": 869
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0166083917147921,
      "learning_rate": 8.577794552422359e-10,
      "loss": 0.0556,
      "step": 870
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.1209124381768403,
      "learning_rate": 5.489845024053698e-10,
      "loss": 0.0786,
      "step": 871
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.4314007730116183,
      "learning_rate": 3.088062549555515e-10,
      "loss": 0.085,
      "step": 872
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.876419163965729,
      "learning_rate": 1.3724800930314807e-10,
      "loss": 0.0863,
      "step": 873
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.1111355248850858,
      "learning_rate": 3.4312120057999886e-11,
      "loss": 0.0886,
      "step": 874
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9385207351245615,
      "learning_rate": 0.0,
      "loss": 0.0745,
      "step": 875
    },
    {
      "epoch": 1.0,
      "step": 875,
      "total_flos": 48930844000256.0,
      "train_loss": 0.10205818943466459,
      "train_runtime": 2840.5403,
      "train_samples_per_second": 2.464,
      "train_steps_per_second": 0.308
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 875,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 500,
  "total_flos": 48930844000256.0,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}