diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,8652 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.996455696202531,
+  "eval_steps": 500,
+  "global_step": 1230,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.004050632911392405,
+      "grad_norm": 7.164759755587964,
+      "learning_rate": 3.2520325203252037e-07,
+      "loss": 1.0866,
+      "step": 1
+    },
+    {
+      "epoch": 0.00810126582278481,
+      "grad_norm": 7.390867054827321,
+      "learning_rate": 6.504065040650407e-07,
+      "loss": 1.1036,
+      "step": 2
+    },
+    {
+      "epoch": 0.012151898734177215,
+      "grad_norm": 7.140110649606754,
+      "learning_rate": 9.75609756097561e-07,
+      "loss": 1.0938,
+      "step": 3
+    },
+    {
+      "epoch": 0.01620253164556962,
+      "grad_norm": 7.284152181513857,
+      "learning_rate": 1.3008130081300815e-06,
+      "loss": 1.1019,
+      "step": 4
+    },
+    {
+      "epoch": 0.020253164556962026,
+      "grad_norm": 6.662595746840797,
+      "learning_rate": 1.6260162601626018e-06,
+      "loss": 1.0777,
+      "step": 5
+    },
+    {
+      "epoch": 0.02430379746835443,
+      "grad_norm": 6.39073781307896,
+      "learning_rate": 1.951219512195122e-06,
+      "loss": 1.0548,
+      "step": 6
+    },
+    {
+      "epoch": 0.028354430379746835,
+      "grad_norm": 5.239025853705934,
+      "learning_rate": 2.2764227642276426e-06,
+      "loss": 1.025,
+      "step": 7
+    },
+    {
+      "epoch": 0.03240506329113924,
+      "grad_norm": 5.1386996202956645,
+      "learning_rate": 2.601626016260163e-06,
+      "loss": 1.0252,
+      "step": 8
+    },
+    {
+      "epoch": 0.03645569620253165,
+      "grad_norm": 3.180257380822603,
+      "learning_rate": 2.926829268292683e-06,
+      "loss": 1.0045,
+      "step": 9
+    },
+    {
+      "epoch": 0.04050632911392405,
+      "grad_norm": 3.1224557762112766,
+      "learning_rate": 3.2520325203252037e-06,
+      "loss": 0.9901,
+      "step": 10
+    },
+    {
+      "epoch": 0.044556962025316456,
+      "grad_norm": 3.0184414822127676,
+      "learning_rate": 3.577235772357724e-06,
+      "loss": 0.9877,
+      "step": 11
+    },
+    {
+      "epoch": 0.04860759493670886,
+      "grad_norm": 4.19094823719288,
+      "learning_rate": 3.902439024390244e-06,
+      "loss": 0.9582,
+      "step": 12
+    },
+    {
+      "epoch": 0.052658227848101265,
+      "grad_norm": 4.511506706306519,
+      "learning_rate": 4.227642276422765e-06,
+      "loss": 0.9581,
+      "step": 13
+    },
+    {
+      "epoch": 0.05670886075949367,
+      "grad_norm": 4.172076896337365,
+      "learning_rate": 4.552845528455285e-06,
+      "loss": 0.9582,
+      "step": 14
+    },
+    {
+      "epoch": 0.060759493670886074,
+      "grad_norm": 3.858712071401548,
+      "learning_rate": 4.8780487804878055e-06,
+      "loss": 0.9303,
+      "step": 15
+    },
+    {
+      "epoch": 0.06481012658227848,
+      "grad_norm": 2.8425386099250844,
+      "learning_rate": 5.203252032520326e-06,
+      "loss": 0.8817,
+      "step": 16
+    },
+    {
+      "epoch": 0.06886075949367089,
+      "grad_norm": 2.766199909194919,
+      "learning_rate": 5.528455284552846e-06,
+      "loss": 0.8743,
+      "step": 17
+    },
+    {
+      "epoch": 0.0729113924050633,
+      "grad_norm": 2.201535129402159,
+      "learning_rate": 5.853658536585366e-06,
+      "loss": 0.8384,
+      "step": 18
+    },
+    {
+      "epoch": 0.0769620253164557,
+      "grad_norm": 1.7143247028287494,
+      "learning_rate": 6.178861788617887e-06,
+      "loss": 0.8256,
+      "step": 19
+    },
+    {
+      "epoch": 0.0810126582278481,
+      "grad_norm": 1.5552589982913902,
+      "learning_rate": 6.504065040650407e-06,
+      "loss": 0.829,
+      "step": 20
+    },
+    {
+      "epoch": 0.08506329113924051,
+      "grad_norm": 1.4503654713309844,
+      "learning_rate": 6.829268292682928e-06,
+      "loss": 0.8007,
+      "step": 21
+    },
+    {
+      "epoch": 0.08911392405063291,
+      "grad_norm": 1.5155729354420096,
+      "learning_rate": 7.154471544715448e-06,
+      "loss": 0.7848,
+      "step": 22
+    },
+    {
+      "epoch": 0.09316455696202532,
+      "grad_norm": 1.4133491269485978,
+      "learning_rate": 7.4796747967479676e-06,
+      "loss": 0.7779,
+      "step": 23
+    },
+    {
+      "epoch": 0.09721518987341772,
+      "grad_norm": 1.4231332404423471,
+      "learning_rate": 7.804878048780489e-06,
+      "loss": 0.7669,
+      "step": 24
+    },
+    {
+      "epoch": 0.10126582278481013,
+      "grad_norm": 1.0860534502367876,
+      "learning_rate": 8.130081300813009e-06,
+      "loss": 0.7725,
+      "step": 25
+    },
+    {
+      "epoch": 0.10531645569620253,
+      "grad_norm": 1.2956932934613654,
+      "learning_rate": 8.45528455284553e-06,
+      "loss": 0.7584,
+      "step": 26
+    },
+    {
+      "epoch": 0.10936708860759493,
+      "grad_norm": 1.374532851760313,
+      "learning_rate": 8.78048780487805e-06,
+      "loss": 0.7481,
+      "step": 27
+    },
+    {
+      "epoch": 0.11341772151898734,
+      "grad_norm": 0.9201436671091955,
+      "learning_rate": 9.10569105691057e-06,
+      "loss": 0.7374,
+      "step": 28
+    },
+    {
+      "epoch": 0.11746835443037974,
+      "grad_norm": 0.8268300385726199,
+      "learning_rate": 9.43089430894309e-06,
+      "loss": 0.7498,
+      "step": 29
+    },
+    {
+      "epoch": 0.12151898734177215,
+      "grad_norm": 1.141397696617225,
+      "learning_rate": 9.756097560975611e-06,
+      "loss": 0.7252,
+      "step": 30
+    },
+    {
+      "epoch": 0.12556962025316457,
+      "grad_norm": 1.1230695996726165,
+      "learning_rate": 1.008130081300813e-05,
+      "loss": 0.7367,
+      "step": 31
+    },
+    {
+      "epoch": 0.12962025316455697,
+      "grad_norm": 1.1010724211587886,
+      "learning_rate": 1.0406504065040652e-05,
+      "loss": 0.7317,
+      "step": 32
+    },
+    {
+      "epoch": 0.13367088607594937,
+      "grad_norm": 0.8863874497917904,
+      "learning_rate": 1.0731707317073172e-05,
+      "loss": 0.7231,
+      "step": 33
+    },
+    {
+      "epoch": 0.13772151898734178,
+      "grad_norm": 0.8831211969420624,
+      "learning_rate": 1.1056910569105692e-05,
+      "loss": 0.6992,
+      "step": 34
+    },
+    {
+      "epoch": 0.14177215189873418,
+      "grad_norm": 0.8680465544611914,
+      "learning_rate": 1.1382113821138213e-05,
+      "loss": 0.7109,
+      "step": 35
+    },
+    {
+      "epoch": 0.1458227848101266,
+      "grad_norm": 0.9461997961602754,
+      "learning_rate": 1.1707317073170731e-05,
+      "loss": 0.6977,
+      "step": 36
+    },
+    {
+      "epoch": 0.149873417721519,
+      "grad_norm": 0.6647558124595361,
+      "learning_rate": 1.2032520325203254e-05,
+      "loss": 0.7103,
+      "step": 37
+    },
+    {
+      "epoch": 0.1539240506329114,
+      "grad_norm": 0.8742696983073177,
+      "learning_rate": 1.2357723577235774e-05,
+      "loss": 0.6938,
+      "step": 38
+    },
+    {
+      "epoch": 0.1579746835443038,
+      "grad_norm": 0.8066977509185319,
+      "learning_rate": 1.2682926829268294e-05,
+      "loss": 0.6977,
+      "step": 39
+    },
+    {
+      "epoch": 0.1620253164556962,
+      "grad_norm": 0.6476639439585331,
+      "learning_rate": 1.3008130081300815e-05,
+      "loss": 0.6884,
+      "step": 40
+    },
+    {
+      "epoch": 0.1660759493670886,
+      "grad_norm": 0.8619642263909862,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.6865,
+      "step": 41
+    },
+    {
+      "epoch": 0.17012658227848101,
+      "grad_norm": 0.7506204975241192,
+      "learning_rate": 1.3658536585365855e-05,
+      "loss": 0.678,
+      "step": 42
+    },
+    {
+      "epoch": 0.17417721518987342,
+      "grad_norm": 0.7740100248962882,
+      "learning_rate": 1.3983739837398376e-05,
+      "loss": 0.6956,
+      "step": 43
+    },
+    {
+      "epoch": 0.17822784810126582,
+      "grad_norm": 0.7762930590451037,
+      "learning_rate": 1.4308943089430896e-05,
+      "loss": 0.6951,
+      "step": 44
+    },
+    {
+      "epoch": 0.18227848101265823,
+      "grad_norm": 0.7546998324371536,
+      "learning_rate": 1.4634146341463415e-05,
+      "loss": 0.6813,
+      "step": 45
+    },
+    {
+      "epoch": 0.18632911392405063,
+      "grad_norm": 0.9311703414812402,
+      "learning_rate": 1.4959349593495935e-05,
+      "loss": 0.675,
+      "step": 46
+    },
+    {
+      "epoch": 0.19037974683544304,
+      "grad_norm": 0.6922333525109576,
+      "learning_rate": 1.528455284552846e-05,
+      "loss": 0.6735,
+      "step": 47
+    },
+    {
+      "epoch": 0.19443037974683544,
+      "grad_norm": 0.8084242077177012,
+      "learning_rate": 1.5609756097560978e-05,
+      "loss": 0.6742,
+      "step": 48
+    },
+    {
+      "epoch": 0.19848101265822785,
+      "grad_norm": 0.7729394723837454,
+      "learning_rate": 1.5934959349593496e-05,
+      "loss": 0.6695,
+      "step": 49
+    },
+    {
+      "epoch": 0.20253164556962025,
+      "grad_norm": 0.7181741211040994,
+      "learning_rate": 1.6260162601626018e-05,
+      "loss": 0.6639,
+      "step": 50
+    },
+    {
+      "epoch": 0.20658227848101265,
+      "grad_norm": 0.7430212071438833,
+      "learning_rate": 1.6585365853658537e-05,
+      "loss": 0.6736,
+      "step": 51
+    },
+    {
+      "epoch": 0.21063291139240506,
+      "grad_norm": 0.5749568351214024,
+      "learning_rate": 1.691056910569106e-05,
+      "loss": 0.6732,
+      "step": 52
+    },
+    {
+      "epoch": 0.21468354430379746,
+      "grad_norm": 0.6508033294893169,
+      "learning_rate": 1.7235772357723578e-05,
+      "loss": 0.6552,
+      "step": 53
+    },
+    {
+      "epoch": 0.21873417721518987,
+      "grad_norm": 0.6872790123186038,
+      "learning_rate": 1.75609756097561e-05,
+      "loss": 0.6629,
+      "step": 54
+    },
+    {
+      "epoch": 0.22278481012658227,
+      "grad_norm": 0.6737038114187952,
+      "learning_rate": 1.788617886178862e-05,
+      "loss": 0.6667,
+      "step": 55
+    },
+    {
+      "epoch": 0.22683544303797468,
+      "grad_norm": 0.7969082982971626,
+      "learning_rate": 1.821138211382114e-05,
+      "loss": 0.675,
+      "step": 56
+    },
+    {
+      "epoch": 0.23088607594936708,
+      "grad_norm": 0.7420563970561559,
+      "learning_rate": 1.8536585365853663e-05,
+      "loss": 0.652,
+      "step": 57
+    },
+    {
+      "epoch": 0.23493670886075949,
+      "grad_norm": 0.6869975570897409,
+      "learning_rate": 1.886178861788618e-05,
+      "loss": 0.6593,
+      "step": 58
+    },
+    {
+      "epoch": 0.2389873417721519,
+      "grad_norm": 0.9288415399483133,
+      "learning_rate": 1.91869918699187e-05,
+      "loss": 0.6647,
+      "step": 59
+    },
+    {
+      "epoch": 0.2430379746835443,
+      "grad_norm": 0.8364640279924954,
+      "learning_rate": 1.9512195121951222e-05,
+      "loss": 0.6636,
+      "step": 60
+    },
+    {
+      "epoch": 0.2470886075949367,
+      "grad_norm": 0.6998989001362281,
+      "learning_rate": 1.983739837398374e-05,
+      "loss": 0.6592,
+      "step": 61
+    },
+    {
+      "epoch": 0.25113924050632913,
+      "grad_norm": 0.8598366155219836,
+      "learning_rate": 2.016260162601626e-05,
+      "loss": 0.6725,
+      "step": 62
+    },
+    {
+      "epoch": 0.25518987341772154,
+      "grad_norm": 0.9872958309290468,
+      "learning_rate": 2.048780487804878e-05,
+      "loss": 0.652,
+      "step": 63
+    },
+    {
+      "epoch": 0.25924050632911394,
+      "grad_norm": 0.7939835419182834,
+      "learning_rate": 2.0813008130081303e-05,
+      "loss": 0.6662,
+      "step": 64
+    },
+    {
+      "epoch": 0.26329113924050634,
+      "grad_norm": 1.0250071380843235,
+      "learning_rate": 2.1138211382113822e-05,
+      "loss": 0.6455,
+      "step": 65
+    },
+    {
+      "epoch": 0.26734177215189875,
+      "grad_norm": 0.8065172284447681,
+      "learning_rate": 2.1463414634146344e-05,
+      "loss": 0.6574,
+      "step": 66
+    },
+    {
+      "epoch": 0.27139240506329115,
+      "grad_norm": 0.7204063769341096,
+      "learning_rate": 2.1788617886178863e-05,
+      "loss": 0.6531,
+      "step": 67
+    },
+    {
+      "epoch": 0.27544303797468356,
+      "grad_norm": 0.7814281094305103,
+      "learning_rate": 2.2113821138211385e-05,
+      "loss": 0.6544,
+      "step": 68
+    },
+    {
+      "epoch": 0.27949367088607596,
+      "grad_norm": 1.19444239845307,
+      "learning_rate": 2.2439024390243907e-05,
+      "loss": 0.6544,
+      "step": 69
+    },
+    {
+      "epoch": 0.28354430379746837,
+      "grad_norm": 0.8245059146713083,
+      "learning_rate": 2.2764227642276426e-05,
+      "loss": 0.6388,
+      "step": 70
+    },
+    {
+      "epoch": 0.28759493670886077,
+      "grad_norm": 0.8608143177708192,
+      "learning_rate": 2.3089430894308948e-05,
+      "loss": 0.6472,
+      "step": 71
+    },
+    {
+      "epoch": 0.2916455696202532,
+      "grad_norm": 1.247263888540179,
+      "learning_rate": 2.3414634146341463e-05,
+      "loss": 0.6551,
+      "step": 72
+    },
+    {
+      "epoch": 0.2956962025316456,
+      "grad_norm": 0.8053496338007629,
+      "learning_rate": 2.3739837398373985e-05,
+      "loss": 0.6536,
+      "step": 73
+    },
+    {
+      "epoch": 0.299746835443038,
+      "grad_norm": 1.1126516769736994,
+      "learning_rate": 2.4065040650406507e-05,
+      "loss": 0.6524,
+      "step": 74
+    },
+    {
+      "epoch": 0.3037974683544304,
+      "grad_norm": 0.9425476366157906,
+      "learning_rate": 2.4390243902439026e-05,
+      "loss": 0.6456,
+      "step": 75
+    },
+    {
+      "epoch": 0.3078481012658228,
+      "grad_norm": 0.9331790833266544,
+      "learning_rate": 2.4715447154471548e-05,
+      "loss": 0.6469,
+      "step": 76
+    },
+    {
+      "epoch": 0.3118987341772152,
+      "grad_norm": 0.9187357404190338,
+      "learning_rate": 2.5040650406504066e-05,
+      "loss": 0.6358,
+      "step": 77
+    },
+    {
+      "epoch": 0.3159493670886076,
+      "grad_norm": 1.1731171986047029,
+      "learning_rate": 2.536585365853659e-05,
+      "loss": 0.6534,
+      "step": 78
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.118771015968267,
+      "learning_rate": 2.569105691056911e-05,
+      "loss": 0.6341,
+      "step": 79
+    },
+    {
+      "epoch": 0.3240506329113924,
+      "grad_norm": 0.8204071953607076,
+      "learning_rate": 2.601626016260163e-05,
+      "loss": 0.6349,
+      "step": 80
+    },
+    {
+      "epoch": 0.3281012658227848,
+      "grad_norm": 1.2526675960430151,
+      "learning_rate": 2.634146341463415e-05,
+      "loss": 0.6534,
+      "step": 81
+    },
+    {
+      "epoch": 0.3321518987341772,
+      "grad_norm": 0.9143304055032708,
+      "learning_rate": 2.6666666666666667e-05,
+      "loss": 0.6385,
+      "step": 82
+    },
+    {
+      "epoch": 0.3362025316455696,
+      "grad_norm": 0.9410091006210993,
+      "learning_rate": 2.699186991869919e-05,
+      "loss": 0.6217,
+      "step": 83
+    },
+    {
+      "epoch": 0.34025316455696203,
+      "grad_norm": 1.3203062789762847,
+      "learning_rate": 2.731707317073171e-05,
+      "loss": 0.6411,
+      "step": 84
+    },
+    {
+      "epoch": 0.34430379746835443,
+      "grad_norm": 0.8146297863220795,
+      "learning_rate": 2.764227642276423e-05,
+      "loss": 0.6347,
+      "step": 85
+    },
+    {
+      "epoch": 0.34835443037974684,
+      "grad_norm": 1.2315904676809015,
+      "learning_rate": 2.796747967479675e-05,
+      "loss": 0.6337,
+      "step": 86
+    },
+    {
+      "epoch": 0.35240506329113924,
+      "grad_norm": 1.175954307604315,
+      "learning_rate": 2.829268292682927e-05,
+      "loss": 0.6425,
+      "step": 87
+    },
+    {
+      "epoch": 0.35645569620253165,
+      "grad_norm": 0.8341326914482822,
+      "learning_rate": 2.8617886178861792e-05,
+      "loss": 0.6444,
+      "step": 88
+    },
+    {
+      "epoch": 0.36050632911392405,
+      "grad_norm": 1.043400044639378,
+      "learning_rate": 2.8943089430894314e-05,
+      "loss": 0.6362,
+      "step": 89
+    },
+    {
+      "epoch": 0.36455696202531646,
+      "grad_norm": 1.0710768063801273,
+      "learning_rate": 2.926829268292683e-05,
+      "loss": 0.6373,
+      "step": 90
+    },
+    {
+      "epoch": 0.36860759493670886,
+      "grad_norm": 0.8995816788165555,
+      "learning_rate": 2.959349593495935e-05,
+      "loss": 0.6481,
+      "step": 91
+    },
+    {
+      "epoch": 0.37265822784810126,
+      "grad_norm": 0.9305319327095318,
+      "learning_rate": 2.991869918699187e-05,
+      "loss": 0.6346,
+      "step": 92
+    },
+    {
+      "epoch": 0.37670886075949367,
+      "grad_norm": 1.1810494551692423,
+      "learning_rate": 3.0243902439024392e-05,
+      "loss": 0.6338,
+      "step": 93
+    },
+    {
+      "epoch": 0.3807594936708861,
+      "grad_norm": 0.8391915536551258,
+      "learning_rate": 3.056910569105692e-05,
+      "loss": 0.6438,
+      "step": 94
+    },
+    {
+      "epoch": 0.3848101265822785,
+      "grad_norm": 1.1259801887353662,
+      "learning_rate": 3.089430894308943e-05,
+      "loss": 0.6396,
+      "step": 95
+    },
+    {
+      "epoch": 0.3888607594936709,
+      "grad_norm": 0.8363638003547884,
+      "learning_rate": 3.1219512195121955e-05,
+      "loss": 0.6416,
+      "step": 96
+    },
+    {
+      "epoch": 0.3929113924050633,
+      "grad_norm": 0.9552476660417918,
+      "learning_rate": 3.154471544715447e-05,
+      "loss": 0.6402,
+      "step": 97
+    },
+    {
+      "epoch": 0.3969620253164557,
+      "grad_norm": 1.1005774390946101,
+      "learning_rate": 3.186991869918699e-05,
+      "loss": 0.6343,
+      "step": 98
+    },
+    {
+      "epoch": 0.4010126582278481,
+      "grad_norm": 0.8670967476218773,
+      "learning_rate": 3.2195121951219514e-05,
+      "loss": 0.6342,
+      "step": 99
+    },
+    {
+      "epoch": 0.4050632911392405,
+      "grad_norm": 0.9887928976369114,
+      "learning_rate": 3.2520325203252037e-05,
+      "loss": 0.6404,
+      "step": 100
+    },
+    {
+      "epoch": 0.4091139240506329,
+      "grad_norm": 0.9879473581696346,
+      "learning_rate": 3.284552845528456e-05,
+      "loss": 0.6326,
+      "step": 101
+    },
+    {
+      "epoch": 0.4131645569620253,
+      "grad_norm": 1.047469602319475,
+      "learning_rate": 3.3170731707317074e-05,
+      "loss": 0.6228,
+      "step": 102
+    },
+    {
+      "epoch": 0.4172151898734177,
+      "grad_norm": 0.8541272881513583,
+      "learning_rate": 3.3495934959349596e-05,
+      "loss": 0.6317,
+      "step": 103
+    },
+    {
+      "epoch": 0.4212658227848101,
+      "grad_norm": 0.7217928665348792,
+      "learning_rate": 3.382113821138212e-05,
+      "loss": 0.6437,
+      "step": 104
+    },
+    {
+      "epoch": 0.4253164556962025,
+      "grad_norm": 1.0212370994920323,
+      "learning_rate": 3.414634146341463e-05,
+      "loss": 0.6382,
+      "step": 105
+    },
+    {
+      "epoch": 0.4293670886075949,
+      "grad_norm": 0.9490541555935569,
+      "learning_rate": 3.4471544715447155e-05,
+      "loss": 0.6186,
+      "step": 106
+    },
+    {
+      "epoch": 0.43341772151898733,
+      "grad_norm": 0.7726120077725682,
+      "learning_rate": 3.479674796747968e-05,
+      "loss": 0.6391,
+      "step": 107
+    },
+    {
+      "epoch": 0.43746835443037974,
+      "grad_norm": 0.8713017461915067,
+      "learning_rate": 3.51219512195122e-05,
+      "loss": 0.6446,
+      "step": 108
+    },
+    {
+      "epoch": 0.44151898734177214,
+      "grad_norm": 0.9471023977179271,
+      "learning_rate": 3.544715447154472e-05,
+      "loss": 0.6337,
+      "step": 109
+    },
+    {
+      "epoch": 0.44556962025316454,
+      "grad_norm": 0.988920679395047,
+      "learning_rate": 3.577235772357724e-05,
+      "loss": 0.6246,
+      "step": 110
+    },
+    {
+      "epoch": 0.44962025316455695,
+      "grad_norm": 1.7207854192631866,
+      "learning_rate": 3.609756097560976e-05,
+      "loss": 0.6336,
+      "step": 111
+    },
+    {
+      "epoch": 0.45367088607594935,
+      "grad_norm": 0.8259445616264611,
+      "learning_rate": 3.642276422764228e-05,
+      "loss": 0.6357,
+      "step": 112
+    },
+    {
+      "epoch": 0.45772151898734176,
+      "grad_norm": 2.0184465594319594,
+      "learning_rate": 3.67479674796748e-05,
+      "loss": 0.6402,
+      "step": 113
+    },
+    {
+      "epoch": 0.46177215189873416,
+      "grad_norm": 1.0931796481815907,
+      "learning_rate": 3.7073170731707325e-05,
+      "loss": 0.6286,
+      "step": 114
+    },
+    {
+      "epoch": 0.46582278481012657,
+      "grad_norm": 2.40635893669466,
+      "learning_rate": 3.739837398373984e-05,
+      "loss": 0.6401,
+      "step": 115
+    },
+    {
+      "epoch": 0.46987341772151897,
+      "grad_norm": 2.1046697350703387,
+      "learning_rate": 3.772357723577236e-05,
+      "loss": 0.6515,
+      "step": 116
+    },
+    {
+      "epoch": 0.4739240506329114,
+      "grad_norm": 1.8080022910040814,
+      "learning_rate": 3.804878048780488e-05,
+      "loss": 0.6303,
+      "step": 117
+    },
+    {
+      "epoch": 0.4779746835443038,
+      "grad_norm": 1.707617898019629,
+      "learning_rate": 3.83739837398374e-05,
+      "loss": 0.6428,
+      "step": 118
+    },
+    {
+      "epoch": 0.4820253164556962,
+      "grad_norm": 1.86999261935865,
+      "learning_rate": 3.869918699186992e-05,
+      "loss": 0.6294,
+      "step": 119
+    },
+    {
+      "epoch": 0.4860759493670886,
+      "grad_norm": 1.4252121290172368,
+      "learning_rate": 3.9024390243902444e-05,
+      "loss": 0.6435,
+      "step": 120
+    },
+    {
+      "epoch": 0.490126582278481,
+      "grad_norm": 1.9778810059161434,
+      "learning_rate": 3.9349593495934966e-05,
+      "loss": 0.616,
+      "step": 121
+    },
+    {
+      "epoch": 0.4941772151898734,
+      "grad_norm": 1.9487470143963088,
+      "learning_rate": 3.967479674796748e-05,
+      "loss": 0.6421,
+      "step": 122
+    },
+    {
+      "epoch": 0.4982278481012658,
+      "grad_norm": 1.3259987654437193,
+      "learning_rate": 4e-05,
+      "loss": 0.6381,
+      "step": 123
+    },
+    {
+      "epoch": 0.5022784810126583,
+      "grad_norm": 1.4092694417451093,
+      "learning_rate": 3.999991946137476e-05,
+      "loss": 0.6258,
+      "step": 124
+    },
+    {
+      "epoch": 0.5063291139240507,
+      "grad_norm": 1.3252030030147437,
+      "learning_rate": 3.999967784614766e-05,
+      "loss": 0.629,
+      "step": 125
+    },
+    {
+      "epoch": 0.5103797468354431,
+      "grad_norm": 0.8083451231932557,
+      "learning_rate": 3.9999275156264656e-05,
+      "loss": 0.6317,
+      "step": 126
+    },
+    {
+      "epoch": 0.5144303797468355,
+      "grad_norm": 1.3664265015425423,
+      "learning_rate": 3.999871139496895e-05,
+      "loss": 0.6112,
+      "step": 127
+    },
+    {
+      "epoch": 0.5184810126582279,
+      "grad_norm": 0.6618504653697111,
+      "learning_rate": 3.9997986566800995e-05,
+      "loss": 0.6265,
+      "step": 128
+    },
+    {
+      "epoch": 0.5225316455696203,
+      "grad_norm": 1.3568668564099642,
+      "learning_rate": 3.999710067759846e-05,
+      "loss": 0.6356,
+      "step": 129
+    },
+    {
+      "epoch": 0.5265822784810127,
+      "grad_norm": 0.6456350980730462,
+      "learning_rate": 3.999605373449617e-05,
+      "loss": 0.6269,
+      "step": 130
+    },
+    {
+      "epoch": 0.5306329113924051,
+      "grad_norm": 1.1103199989116184,
+      "learning_rate": 3.9994845745926075e-05,
+      "loss": 0.6187,
+      "step": 131
+    },
+    {
+      "epoch": 0.5346835443037975,
+      "grad_norm": 0.8707265473591669,
+      "learning_rate": 3.999347672161713e-05,
+      "loss": 0.6306,
+      "step": 132
+    },
+    {
+      "epoch": 0.5387341772151899,
+      "grad_norm": 0.735316906400593,
+      "learning_rate": 3.999194667259528e-05,
+      "loss": 0.622,
+      "step": 133
+    },
+    {
+      "epoch": 0.5427848101265823,
+      "grad_norm": 1.3137039976756266,
+      "learning_rate": 3.999025561118334e-05,
+      "loss": 0.6224,
+      "step": 134
+    },
+    {
+      "epoch": 0.5468354430379747,
+      "grad_norm": 0.7255369352484277,
+      "learning_rate": 3.998840355100086e-05,
+      "loss": 0.6218,
+      "step": 135
+    },
+    {
+      "epoch": 0.5508860759493671,
+      "grad_norm": 1.176488532089197,
+      "learning_rate": 3.998639050696409e-05,
+      "loss": 0.6349,
+      "step": 136
+    },
+    {
+      "epoch": 0.5549367088607595,
+      "grad_norm": 0.8039976175677945,
+      "learning_rate": 3.998421649528582e-05,
+      "loss": 0.624,
+      "step": 137
+    },
+    {
+      "epoch": 0.5589873417721519,
+      "grad_norm": 0.9538868923033756,
+      "learning_rate": 3.9981881533475234e-05,
+      "loss": 0.6227,
+      "step": 138
+    },
+    {
+      "epoch": 0.5630379746835443,
+      "grad_norm": 0.7350773071124918,
+      "learning_rate": 3.997938564033779e-05,
+      "loss": 0.6258,
+      "step": 139
+    },
+    {
+      "epoch": 0.5670886075949367,
+      "grad_norm": 0.7519857688415897,
+      "learning_rate": 3.9976728835975064e-05,
+      "loss": 0.625,
+      "step": 140
+    },
+    {
+      "epoch": 0.5711392405063291,
+      "grad_norm": 1.1326250175813635,
+      "learning_rate": 3.9973911141784605e-05,
+      "loss": 0.6151,
+      "step": 141
+    },
+    {
+      "epoch": 0.5751898734177215,
+      "grad_norm": 0.7183529020396047,
+      "learning_rate": 3.997093258045973e-05,
+      "loss": 0.6109,
+      "step": 142
+    },
+    {
+      "epoch": 0.579240506329114,
+      "grad_norm": 0.7173691866075678,
+      "learning_rate": 3.996779317598936e-05,
+      "loss": 0.6135,
+      "step": 143
+    },
+    {
+      "epoch": 0.5832911392405064,
+      "grad_norm": 0.7189685828512047,
+      "learning_rate": 3.996449295365782e-05,
+      "loss": 0.6258,
+      "step": 144
+    },
+    {
+      "epoch": 0.5873417721518988,
+      "grad_norm": 0.6293347339372223,
+      "learning_rate": 3.996103194004467e-05,
+      "loss": 0.616,
+      "step": 145
+    },
+    {
+      "epoch": 0.5913924050632912,
+      "grad_norm": 0.7236533920114554,
+      "learning_rate": 3.995741016302441e-05,
+      "loss": 0.6199,
+      "step": 146
+    },
+    {
+      "epoch": 0.5954430379746836,
+      "grad_norm": 0.5936183988856741,
+      "learning_rate": 3.9953627651766364e-05,
+      "loss": 0.6146,
+      "step": 147
+    },
+    {
+      "epoch": 0.599493670886076,
+      "grad_norm": 0.8221670509382226,
+      "learning_rate": 3.9949684436734325e-05,
+      "loss": 0.615,
+      "step": 148
+    },
+    {
+      "epoch": 0.6035443037974684,
+      "grad_norm": 0.6722583504588668,
+      "learning_rate": 3.994558054968643e-05,
+      "loss": 0.6042,
+      "step": 149
+    },
+    {
+      "epoch": 0.6075949367088608,
+      "grad_norm": 0.5893746499600901,
+      "learning_rate": 3.994131602367481e-05,
+      "loss": 0.624,
+      "step": 150
+    },
+    {
+      "epoch": 0.6116455696202532,
+      "grad_norm": 0.7755316074556124,
+      "learning_rate": 3.9936890893045376e-05,
+      "loss": 0.6188,
+      "step": 151
+    },
+    {
+      "epoch": 0.6156962025316456,
+      "grad_norm": 0.697132139201966,
+      "learning_rate": 3.993230519343752e-05,
+      "loss": 0.6206,
+      "step": 152
+    },
+    {
+      "epoch": 0.619746835443038,
+      "grad_norm": 0.7156229849388253,
+      "learning_rate": 3.992755896178383e-05,
+      "loss": 0.6185,
+      "step": 153
+    },
+    {
+      "epoch": 0.6237974683544304,
+      "grad_norm": 0.7071853259982792,
+      "learning_rate": 3.992265223630981e-05,
+      "loss": 0.6183,
+      "step": 154
+    },
+    {
+      "epoch": 0.6278481012658228,
+      "grad_norm": 0.7132473578501581,
+      "learning_rate": 3.991758505653355e-05,
+      "loss": 0.6151,
+      "step": 155
+    },
+    {
+      "epoch": 0.6318987341772152,
+      "grad_norm": 0.6279576522553003,
+      "learning_rate": 3.991235746326543e-05,
+      "loss": 0.6236,
+      "step": 156
+    },
+    {
+      "epoch": 0.6359493670886076,
+      "grad_norm": 0.6273113803648434,
+      "learning_rate": 3.9906969498607745e-05,
+      "loss": 0.6296,
+      "step": 157
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.6349267562364772,
+      "learning_rate": 3.990142120595444e-05,
+      "loss": 0.6214,
+      "step": 158
+    },
+    {
+      "epoch": 0.6440506329113924,
+      "grad_norm": 0.8775633636400432,
+      "learning_rate": 3.98957126299907e-05,
+      "loss": 0.6216,
+      "step": 159
+    },
+    {
+      "epoch": 0.6481012658227848,
+      "grad_norm": 0.7448512080212946,
+      "learning_rate": 3.9889843816692596e-05,
+      "loss": 0.612,
+      "step": 160
+    },
+    {
+      "epoch": 0.6521518987341772,
+      "grad_norm": 0.7001030281221607,
+      "learning_rate": 3.9883814813326766e-05,
+      "loss": 0.6125,
+      "step": 161
+    },
+    {
+      "epoch": 0.6562025316455696,
+      "grad_norm": 0.5947598935544604,
+      "learning_rate": 3.9877625668449956e-05,
+      "loss": 0.6124,
+      "step": 162
+    },
+    {
+      "epoch": 0.660253164556962,
+      "grad_norm": 0.70506399124175,
+      "learning_rate": 3.98712764319087e-05,
+      "loss": 0.6179,
+      "step": 163
+    },
+    {
+      "epoch": 0.6643037974683544,
+      "grad_norm": 0.6181917570304543,
+      "learning_rate": 3.9864767154838864e-05,
+      "loss": 0.6254,
+      "step": 164
+    },
+    {
+      "epoch": 0.6683544303797468,
+      "grad_norm": 0.7676331459926508,
+      "learning_rate": 3.9858097889665277e-05,
+      "loss": 0.6101,
+      "step": 165
+    },
+    {
+      "epoch": 0.6724050632911392,
+      "grad_norm": 0.8840523629578689,
+      "learning_rate": 3.985126869010129e-05,
+      "loss": 0.6201,
+      "step": 166
+    },
+    {
+      "epoch": 0.6764556962025317,
+      "grad_norm": 0.8222576042145104,
+      "learning_rate": 3.984427961114833e-05,
+      "loss": 0.6115,
+      "step": 167
+    },
+    {
+      "epoch": 0.6805063291139241,
+      "grad_norm": 0.6832961000264165,
+      "learning_rate": 3.9837130709095475e-05,
+      "loss": 0.6203,
+      "step": 168
+    },
+    {
+      "epoch": 0.6845569620253165,
+      "grad_norm": 0.768415783502988,
+      "learning_rate": 3.982982204151901e-05,
+      "loss": 0.6289,
+      "step": 169
+    },
+    {
+      "epoch": 0.6886075949367089,
+      "grad_norm": 0.6709212736724036,
+      "learning_rate": 3.982235366728193e-05,
+      "loss": 0.6102,
+      "step": 170
+    },
+    {
+      "epoch": 0.6926582278481013,
+      "grad_norm": 0.8583243984195551,
+      "learning_rate": 3.9814725646533505e-05,
+      "loss": 0.5977,
+      "step": 171
+    },
+    {
+      "epoch": 0.6967088607594937,
+      "grad_norm": 0.7231043864060707,
+      "learning_rate": 3.9806938040708746e-05,
+      "loss": 0.615,
+      "step": 172
+    },
+    {
+      "epoch": 0.7007594936708861,
+      "grad_norm": 0.5847868159710016,
+      "learning_rate": 3.9798990912527976e-05,
+      "loss": 0.5961,
+      "step": 173
+    },
+    {
+      "epoch": 0.7048101265822785,
+      "grad_norm": 0.7672286833022892,
+      "learning_rate": 3.979088432599627e-05,
+      "loss": 0.6059,
+      "step": 174
+    },
+    {
+      "epoch": 0.7088607594936709,
+      "grad_norm": 0.8247912404861214,
+      "learning_rate": 3.9782618346402964e-05,
+      "loss": 0.614,
+      "step": 175
+    },
+    {
+      "epoch": 0.7129113924050633,
+      "grad_norm": 0.6258104903195489,
+      "learning_rate": 3.977419304032111e-05,
+      "loss": 0.6218,
+      "step": 176
+    },
+    {
+      "epoch": 0.7169620253164557,
+      "grad_norm": 0.6440084295669244,
+      "learning_rate": 3.976560847560697e-05,
+      "loss": 0.6143,
+      "step": 177
+    },
+    {
+      "epoch": 0.7210126582278481,
+      "grad_norm": 0.6146330092372728,
+      "learning_rate": 3.9756864721399456e-05,
+      "loss": 0.6142,
+      "step": 178
+    },
+    {
+      "epoch": 0.7250632911392405,
+      "grad_norm": 0.6688526396595473,
+      "learning_rate": 3.974796184811956e-05,
+      "loss": 0.6082,
+      "step": 179
+    },
+    {
+      "epoch": 0.7291139240506329,
+      "grad_norm": 0.7154318992501083,
+      "learning_rate": 3.973889992746979e-05,
+      "loss": 0.6144,
+      "step": 180
+    },
+    {
+      "epoch": 0.7331645569620253,
+      "grad_norm": 0.6522014589598161,
+      "learning_rate": 3.972967903243361e-05,
+      "loss": 0.6151,
+      "step": 181
+    },
+    {
+      "epoch": 0.7372151898734177,
+      "grad_norm": 0.9545443197531521,
+      "learning_rate": 3.972029923727486e-05,
+      "loss": 0.6095,
+      "step": 182
+    },
+    {
+      "epoch": 0.7412658227848101,
+      "grad_norm": 0.7157080455155103,
+      "learning_rate": 3.971076061753709e-05,
+      "loss": 0.6106,
+      "step": 183
+    },
+    {
+      "epoch": 0.7453164556962025,
+      "grad_norm": 0.6592044676467763,
+      "learning_rate": 3.9701063250043066e-05,
+      "loss": 0.61,
+      "step": 184
+    },
+    {
+      "epoch": 0.7493670886075949,
+      "grad_norm": 0.6073372083651385,
+      "learning_rate": 3.969120721289402e-05,
+      "loss": 0.6062,
+      "step": 185
+    },
+    {
+      "epoch": 0.7534177215189873,
+      "grad_norm": 0.6281520865893419,
+      "learning_rate": 3.9681192585469146e-05,
+      "loss": 0.6062,
+      "step": 186
+    },
+    {
+      "epoch": 0.7574683544303797,
+      "grad_norm": 0.6159255826807272,
+      "learning_rate": 3.9671019448424865e-05,
+      "loss": 0.6154,
+      "step": 187
+    },
+    {
+      "epoch": 0.7615189873417721,
+      "grad_norm": 0.7082343315050484,
+      "learning_rate": 3.966068788369422e-05,
+      "loss": 0.6069,
+      "step": 188
+    },
+    {
+      "epoch": 0.7655696202531646,
+      "grad_norm": 0.65691733381587,
+      "learning_rate": 3.965019797448622e-05,
+      "loss": 0.6042,
+      "step": 189
+    },
+    {
+      "epoch": 0.769620253164557,
+      "grad_norm": 0.6567339662337857,
+      "learning_rate": 3.963954980528515e-05,
+      "loss": 0.6063,
+      "step": 190
+    },
+    {
+      "epoch": 0.7736708860759494,
+      "grad_norm": 0.7170010354378551,
+      "learning_rate": 3.9628743461849905e-05,
+      "loss": 0.6147,
+      "step": 191
+    },
+    {
+      "epoch": 0.7777215189873418,
+      "grad_norm": 0.811562261171082,
+      "learning_rate": 3.961777903121329e-05,
+      "loss": 0.616,
+      "step": 192
+    },
+    {
+      "epoch": 0.7817721518987342,
+      "grad_norm": 0.49429091255877544,
+      "learning_rate": 3.960665660168131e-05,
+      "loss": 0.6144,
+      "step": 193
+    },
+    {
+      "epoch": 0.7858227848101266,
+      "grad_norm": 0.6983043467077465,
+      "learning_rate": 3.9595376262832485e-05,
+      "loss": 0.6123,
+      "step": 194
+    },
+    {
+      "epoch": 0.789873417721519,
+      "grad_norm": 0.5599764379839527,
+      "learning_rate": 3.9583938105517127e-05,
+      "loss": 0.5857,
+      "step": 195
+    },
+    {
+      "epoch": 0.7939240506329114,
+      "grad_norm": 0.6874190006410102,
+      "learning_rate": 3.957234222185657e-05,
+      "loss": 0.6014,
+      "step": 196
+    },
+    {
+      "epoch": 0.7979746835443038,
+      "grad_norm": 0.6241715172613774,
+      "learning_rate": 3.9560588705242474e-05,
+      "loss": 0.6173,
+      "step": 197
+    },
+    {
+      "epoch": 0.8020253164556962,
+      "grad_norm": 0.6514198865241718,
+      "learning_rate": 3.954867765033605e-05,
+      "loss": 0.6005,
+      "step": 198
+    },
+    {
+      "epoch": 0.8060759493670886,
+      "grad_norm": 0.6070654797102293,
+      "learning_rate": 3.953660915306728e-05,
+      "loss": 0.6191,
+      "step": 199
+    },
+    {
+      "epoch": 0.810126582278481,
+      "grad_norm": 0.6791134849658517,
+      "learning_rate": 3.952438331063419e-05,
+      "loss": 0.6097,
+      "step": 200
+    },
+    {
+      "epoch": 0.8141772151898734,
+      "grad_norm": 0.49757051148341747,
+      "learning_rate": 3.951200022150205e-05,
+      "loss": 0.6074,
+      "step": 201
+    },
+    {
+      "epoch": 0.8182278481012658,
+      "grad_norm": 0.5752299419528373,
+      "learning_rate": 3.949945998540253e-05,
+      "loss": 0.6022,
+      "step": 202
+    },
+    {
+      "epoch": 0.8222784810126582,
+      "grad_norm": 0.640011524975591,
+      "learning_rate": 3.9486762703332993e-05,
+      "loss": 0.6058,
+      "step": 203
+    },
+    {
+      "epoch": 0.8263291139240506,
+      "grad_norm": 0.5846596358930163,
+      "learning_rate": 3.947390847755559e-05,
+      "loss": 0.6101,
+      "step": 204
+    },
+    {
+      "epoch": 0.830379746835443,
+      "grad_norm": 0.6292287438120919,
+      "learning_rate": 3.946089741159648e-05,
+      "loss": 0.6007,
+      "step": 205
+    },
+    {
+      "epoch": 0.8344303797468354,
+      "grad_norm": 0.545961403300486,
+      "learning_rate": 3.944772961024501e-05,
+      "loss": 0.6,
+      "step": 206
+    },
+    {
+      "epoch": 0.8384810126582278,
+      "grad_norm": 0.6170293810247972,
+      "learning_rate": 3.943440517955285e-05,
+      "loss": 0.6074,
+      "step": 207
+    },
+    {
+      "epoch": 0.8425316455696202,
+      "grad_norm": 0.521635637333729,
+      "learning_rate": 3.9420924226833126e-05,
+      "loss": 0.6014,
+      "step": 208
+    },
+    {
+      "epoch": 0.8465822784810126,
+      "grad_norm": 0.6990800521684036,
+      "learning_rate": 3.9407286860659566e-05,
+      "loss": 0.6043,
+      "step": 209
+    },
+    {
+      "epoch": 0.850632911392405,
+      "grad_norm": 0.6716746028878288,
+      "learning_rate": 3.9393493190865657e-05,
+      "loss": 0.6033,
+      "step": 210
+    },
+    {
+      "epoch": 0.8546835443037974,
+      "grad_norm": 0.5220604823057909,
+      "learning_rate": 3.937954332854371e-05,
+      "loss": 0.6082,
+      "step": 211
+    },
+    {
+      "epoch": 0.8587341772151899,
+      "grad_norm": 0.7307126683380398,
+      "learning_rate": 3.9365437386044016e-05,
+      "loss": 0.6099,
+      "step": 212
+    },
+    {
+      "epoch": 0.8627848101265823,
+      "grad_norm": 0.5645040506206123,
+      "learning_rate": 3.935117547697387e-05,
+      "loss": 0.6041,
+      "step": 213
+    },
+    {
+      "epoch": 0.8668354430379747,
+      "grad_norm": 0.6687533901546738,
+      "learning_rate": 3.933675771619675e-05,
+      "loss": 0.6067,
+      "step": 214
+    },
+    {
+      "epoch": 0.8708860759493671,
+      "grad_norm": 0.6091274811248126,
+      "learning_rate": 3.932218421983131e-05,
+      "loss": 0.6168,
+      "step": 215
+    },
+    {
+      "epoch": 0.8749367088607595,
+      "grad_norm": 0.5458824256548308,
+      "learning_rate": 3.9307455105250484e-05,
+      "loss": 0.615,
+      "step": 216
+    },
+    {
+      "epoch": 0.8789873417721519,
+      "grad_norm": 0.6183235808241221,
+      "learning_rate": 3.929257049108054e-05,
+      "loss": 0.6095,
+      "step": 217
+    },
+    {
+      "epoch": 0.8830379746835443,
+      "grad_norm": 0.7021221374465482,
+      "learning_rate": 3.927753049720011e-05,
+      "loss": 0.5961,
+      "step": 218
+    },
+    {
+      "epoch": 0.8870886075949367,
+      "grad_norm": 0.5540169587483443,
+      "learning_rate": 3.9262335244739234e-05,
+      "loss": 0.6016,
+      "step": 219
+    },
+    {
+      "epoch": 0.8911392405063291,
+      "grad_norm": 0.6575064006090723,
+      "learning_rate": 3.92469848560784e-05,
+      "loss": 0.6028,
+      "step": 220
+    },
+    {
+      "epoch": 0.8951898734177215,
+      "grad_norm": 0.5154441182513175,
+      "learning_rate": 3.923147945484751e-05,
+      "loss": 0.5949,
+      "step": 221
+    },
+    {
+      "epoch": 0.8992405063291139,
+      "grad_norm": 0.5759759427597146,
+      "learning_rate": 3.9215819165924956e-05,
+      "loss": 0.6087,
+      "step": 222
+    },
+    {
+      "epoch": 0.9032911392405063,
+      "grad_norm": 0.618519351111325,
+      "learning_rate": 3.920000411543654e-05,
+      "loss": 0.6081,
+      "step": 223
+    },
+    {
+      "epoch": 0.9073417721518987,
+      "grad_norm": 0.5293465545324589,
+      "learning_rate": 3.9184034430754495e-05,
+      "loss": 0.6026,
+      "step": 224
+    },
+    {
+      "epoch": 0.9113924050632911,
+      "grad_norm": 0.6986839528162515,
+      "learning_rate": 3.916791024049648e-05,
+      "loss": 0.5908,
+      "step": 225
+    },
+    {
+      "epoch": 0.9154430379746835,
+      "grad_norm": 0.6439930333797015,
+      "learning_rate": 3.91516316745245e-05,
+      "loss": 0.5897,
+      "step": 226
+    },
+    {
+      "epoch": 0.9194936708860759,
+      "grad_norm": 0.6112230907940726,
+      "learning_rate": 3.913519886394389e-05,
+      "loss": 0.6048,
+      "step": 227
+    },
+    {
+      "epoch": 0.9235443037974683,
+      "grad_norm": 0.5799846024139732,
+      "learning_rate": 3.911861194110225e-05,
+      "loss": 0.5884,
+      "step": 228
+    },
+    {
+      "epoch": 0.9275949367088607,
+      "grad_norm": 0.5362394547396329,
+      "learning_rate": 3.910187103958837e-05,
+      "loss": 0.6014,
+      "step": 229
+    },
+    {
+      "epoch": 0.9316455696202531,
+      "grad_norm": 0.5510320237465745,
+      "learning_rate": 3.908497629423117e-05,
+      "loss": 0.6001,
+      "step": 230
+    },
+    {
+      "epoch": 0.9356962025316455,
+      "grad_norm": 0.5241041601719746,
+      "learning_rate": 3.9067927841098614e-05,
+      "loss": 0.6083,
+      "step": 231
+    },
+    {
+      "epoch": 0.9397468354430379,
+      "grad_norm": 0.5950224736376188,
+      "learning_rate": 3.9050725817496594e-05,
+      "loss": 0.5974,
+      "step": 232
+    },
+    {
+      "epoch": 0.9437974683544303,
+      "grad_norm": 0.5543485835029502,
+      "learning_rate": 3.9033370361967844e-05,
+      "loss": 0.5995,
+      "step": 233
+    },
+    {
+      "epoch": 0.9478481012658228,
+      "grad_norm": 0.6087849744190343,
+      "learning_rate": 3.901586161429081e-05,
+      "loss": 0.5998,
+      "step": 234
+    },
+    {
+      "epoch": 0.9518987341772152,
+      "grad_norm": 0.6537731473305335,
+      "learning_rate": 3.8998199715478545e-05,
+      "loss": 0.595,
+      "step": 235
+    },
+    {
+      "epoch": 0.9559493670886076,
+      "grad_norm": 0.5725841820866883,
+      "learning_rate": 3.8980384807777564e-05,
+      "loss": 0.5934,
+      "step": 236
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.5504584465014626,
+      "learning_rate": 3.896241703466667e-05,
+      "loss": 0.591,
+      "step": 237
+    },
+    {
+      "epoch": 0.9640506329113924,
+      "grad_norm": 0.518576089707281,
+      "learning_rate": 3.894429654085585e-05,
+      "loss": 0.5942,
+      "step": 238
+    },
+    {
+      "epoch": 0.9681012658227848,
+      "grad_norm": 0.561904406308489,
+      "learning_rate": 3.892602347228505e-05,
+      "loss": 0.61,
+      "step": 239
+    },
+    {
+      "epoch": 0.9721518987341772,
+      "grad_norm": 0.5839441430256872,
+      "learning_rate": 3.890759797612307e-05,
+      "loss": 0.5896,
+      "step": 240
+    },
+    {
+      "epoch": 0.9762025316455696,
+      "grad_norm": 0.6851510298366549,
+      "learning_rate": 3.888902020076632e-05,
+      "loss": 0.588,
+      "step": 241
+    },
+    {
+      "epoch": 0.980253164556962,
+      "grad_norm": 0.5086235365570487,
+      "learning_rate": 3.887029029583764e-05,
+      "loss": 0.6021,
+      "step": 242
+    },
+    {
+      "epoch": 0.9843037974683544,
+      "grad_norm": 0.6916257276541387,
+      "learning_rate": 3.8851408412185125e-05,
+      "loss": 0.613,
+      "step": 243
+    },
+    {
+      "epoch": 0.9883544303797468,
+      "grad_norm": 0.539925125748244,
+      "learning_rate": 3.8832374701880855e-05,
+      "loss": 0.6125,
+      "step": 244
+    },
+    {
+      "epoch": 0.9924050632911392,
+      "grad_norm": 0.5693346455904266,
+      "learning_rate": 3.881318931821972e-05,
+      "loss": 0.5922,
+      "step": 245
+    },
+    {
+      "epoch": 0.9964556962025316,
+      "grad_norm": 0.6498394979093176,
+      "learning_rate": 3.879385241571817e-05,
+      "loss": 0.6124,
+      "step": 246
+    },
+    {
+      "epoch": 1.000506329113924,
+      "grad_norm": 0.47647151054701814,
+      "learning_rate": 3.8774364150112955e-05,
+      "loss": 0.5945,
+      "step": 247
+    },
+    {
+      "epoch": 1.0045569620253165,
+      "grad_norm": 0.5317135072910572,
+      "learning_rate": 3.8754724678359884e-05,
+      "loss": 0.552,
+      "step": 248
+    },
+    {
+      "epoch": 1.0086075949367088,
+      "grad_norm": 0.44229719912696364,
+      "learning_rate": 3.873493415863256e-05,
+      "loss": 0.5611,
+      "step": 249
+    },
+    {
+      "epoch": 1.0126582278481013,
+      "grad_norm": 0.5026511092272193,
+      "learning_rate": 3.871499275032111e-05,
+      "loss": 0.5283,
+      "step": 250
+    },
+    {
+      "epoch": 1.0167088607594936,
+      "grad_norm": 0.5315621209123355,
+      "learning_rate": 3.869490061403091e-05,
+      "loss": 0.5351,
+      "step": 251
+    },
+    {
+      "epoch": 1.0207594936708861,
+      "grad_norm": 0.7490133533059018,
+      "learning_rate": 3.867465791158124e-05,
+      "loss": 0.5375,
+      "step": 252
+    },
+    {
+      "epoch": 1.0248101265822784,
+      "grad_norm": 0.8948667942381173,
+      "learning_rate": 3.865426480600407e-05,
+      "loss": 0.5348,
+      "step": 253
+    },
+    {
+      "epoch": 1.028860759493671,
+      "grad_norm": 0.6045569890206522,
+      "learning_rate": 3.863372146154264e-05,
+      "loss": 0.5362,
+      "step": 254
+    },
+    {
+      "epoch": 1.0329113924050632,
+      "grad_norm": 0.7467690822474323,
+      "learning_rate": 3.861302804365024e-05,
+      "loss": 0.5462,
+      "step": 255
+    },
+    {
+      "epoch": 1.0369620253164558,
+      "grad_norm": 0.5224238692966838,
+      "learning_rate": 3.85921847189888e-05,
+      "loss": 0.5425,
+      "step": 256
+    },
+    {
+      "epoch": 1.041012658227848,
+      "grad_norm": 0.6481718125151902,
+      "learning_rate": 3.85711916554276e-05,
+      "loss": 0.531,
+      "step": 257
+    },
+    {
+      "epoch": 1.0450632911392406,
+      "grad_norm": 0.5353094056270788,
+      "learning_rate": 3.85500490220419e-05,
+      "loss": 0.5337,
+      "step": 258
+    },
+    {
+      "epoch": 1.0491139240506329,
+      "grad_norm": 0.5953711864531605,
+      "learning_rate": 3.852875698911154e-05,
+      "loss": 0.5448,
+      "step": 259
+    },
+    {
+      "epoch": 1.0531645569620254,
+      "grad_norm": 0.622665107466701,
+      "learning_rate": 3.850731572811963e-05,
+      "loss": 0.5171,
+      "step": 260
+    },
+    {
+      "epoch": 1.0572151898734177,
+      "grad_norm": 0.6187928225946869,
+      "learning_rate": 3.848572541175116e-05,
+      "loss": 0.5397,
+      "step": 261
+    },
+    {
+      "epoch": 1.0612658227848102,
+      "grad_norm": 0.5708476664574291,
+      "learning_rate": 3.846398621389154e-05,
+      "loss": 0.5423,
+      "step": 262
+    },
+    {
+      "epoch": 1.0653164556962025,
+      "grad_norm": 0.6173504689157465,
+      "learning_rate": 3.84420983096253e-05,
+      "loss": 0.5279,
+      "step": 263
+    },
+    {
+      "epoch": 1.069367088607595,
+      "grad_norm": 0.5913600849642681,
+      "learning_rate": 3.8420061875234606e-05,
+      "loss": 0.5417,
+      "step": 264
+    },
+    {
+      "epoch": 1.0734177215189873,
+      "grad_norm": 0.4894581213948145,
+      "learning_rate": 3.839787708819787e-05,
+      "loss": 0.5348,
+      "step": 265
+    },
+    {
+      "epoch": 1.0774683544303798,
+      "grad_norm": 0.6207562932401871,
+      "learning_rate": 3.8375544127188325e-05,
+      "loss": 0.5355,
+      "step": 266
+    },
+    {
+      "epoch": 1.081518987341772,
+      "grad_norm": 0.5363393067367218,
+      "learning_rate": 3.8353063172072564e-05,
+      "loss": 0.5231,
+      "step": 267
+    },
+    {
+      "epoch": 1.0855696202531646,
+      "grad_norm": 0.4606486052401517,
+      "learning_rate": 3.8330434403909105e-05,
+      "loss": 0.52,
+      "step": 268
+    },
+    {
+      "epoch": 1.089620253164557,
+      "grad_norm": 0.5105013900712762,
+      "learning_rate": 3.8307658004946934e-05,
+      "loss": 0.5202,
+      "step": 269
+    },
+    {
+      "epoch": 1.0936708860759494,
+      "grad_norm": 0.5011182701645943,
+      "learning_rate": 3.8284734158624046e-05,
+      "loss": 0.5359,
+      "step": 270
+    },
+    {
+      "epoch": 1.0977215189873417,
+      "grad_norm": 0.5473683844806705,
+      "learning_rate": 3.826166304956594e-05,
+      "loss": 0.521,
+      "step": 271
+    },
+    {
+      "epoch": 1.1017721518987342,
+      "grad_norm": 0.5224421299502534,
+      "learning_rate": 3.8238444863584164e-05,
+      "loss": 0.5358,
+      "step": 272
+    },
+    {
+      "epoch": 1.1058227848101265,
+      "grad_norm": 0.44912053346957215,
+      "learning_rate": 3.821507978767479e-05,
+      "loss": 0.5195,
+      "step": 273
+    },
+    {
+      "epoch": 1.109873417721519,
+      "grad_norm": 0.6053115377001748,
+      "learning_rate": 3.819156801001693e-05,
+      "loss": 0.5341,
+      "step": 274
+    },
+    {
+      "epoch": 1.1139240506329113,
+      "grad_norm": 0.7160979160686912,
+      "learning_rate": 3.816790971997121e-05,
+      "loss": 0.5435,
+      "step": 275
+    },
+    {
+      "epoch": 1.1179746835443038,
+      "grad_norm": 0.5526720305106354,
+      "learning_rate": 3.8144105108078246e-05,
+      "loss": 0.5175,
+      "step": 276
+    },
+    {
+      "epoch": 1.1220253164556961,
+      "grad_norm": 0.505756483585839,
+      "learning_rate": 3.81201543660571e-05,
+      "loss": 0.5195,
+      "step": 277
+    },
+    {
+      "epoch": 1.1260759493670887,
+      "grad_norm": 0.6230930586638845,
+      "learning_rate": 3.809605768680377e-05,
+      "loss": 0.5389,
+      "step": 278
+    },
+    {
+      "epoch": 1.130126582278481,
+      "grad_norm": 0.508763771142595,
+      "learning_rate": 3.807181526438958e-05,
+      "loss": 0.5369,
+      "step": 279
+    },
+    {
+      "epoch": 1.1341772151898735,
+      "grad_norm": 0.5704422886186568,
+      "learning_rate": 3.8047427294059697e-05,
+      "loss": 0.5307,
+      "step": 280
+    },
+    {
+      "epoch": 1.1382278481012658,
+      "grad_norm": 0.4343559306286027,
+      "learning_rate": 3.802289397223145e-05,
+      "loss": 0.5387,
+      "step": 281
+    },
+    {
+      "epoch": 1.1422784810126583,
+      "grad_norm": 0.4970813096063536,
+      "learning_rate": 3.7998215496492854e-05,
+      "loss": 0.5213,
+      "step": 282
+    },
+    {
+      "epoch": 1.1463291139240506,
+      "grad_norm": 0.49190064555263463,
+      "learning_rate": 3.797339206560096e-05,
+      "loss": 0.5492,
+      "step": 283
+    },
+    {
+      "epoch": 1.150379746835443,
+      "grad_norm": 0.48133925535715805,
+      "learning_rate": 3.794842387948027e-05,
+      "loss": 0.5338,
+      "step": 284
+    },
+    {
+      "epoch": 1.1544303797468354,
+      "grad_norm": 0.42134075917382857,
+      "learning_rate": 3.7923311139221114e-05,
+      "loss": 0.5267,
+      "step": 285
+    },
+    {
+      "epoch": 1.158481012658228,
+      "grad_norm": 0.5151548804156758,
+      "learning_rate": 3.7898054047078054e-05,
+      "loss": 0.5392,
+      "step": 286
+    },
+    {
+      "epoch": 1.1625316455696202,
+      "grad_norm": 0.4814074121829109,
+      "learning_rate": 3.787265280646825e-05,
+      "loss": 0.5562,
+      "step": 287
+    },
+    {
+      "epoch": 1.1665822784810127,
+      "grad_norm": 0.4790826431064566,
+      "learning_rate": 3.7847107621969786e-05,
+      "loss": 0.5302,
+      "step": 288
+    },
+    {
+      "epoch": 1.170632911392405,
+      "grad_norm": 0.5104840241870794,
+      "learning_rate": 3.7821418699320064e-05,
+      "loss": 0.5475,
+      "step": 289
+    },
+    {
+      "epoch": 1.1746835443037975,
+      "grad_norm": 0.5519134259513058,
+      "learning_rate": 3.7795586245414145e-05,
+      "loss": 0.5175,
+      "step": 290
+    },
+    {
+      "epoch": 1.1787341772151898,
+      "grad_norm": 0.5081096131702956,
+      "learning_rate": 3.776961046830306e-05,
+      "loss": 0.5329,
+      "step": 291
+    },
+    {
+      "epoch": 1.1827848101265823,
+      "grad_norm": 0.5691397024115394,
+      "learning_rate": 3.774349157719215e-05,
+      "loss": 0.5392,
+      "step": 292
+    },
+    {
+      "epoch": 1.1868354430379746,
+      "grad_norm": 0.5015968024885753,
+      "learning_rate": 3.7717229782439365e-05,
+      "loss": 0.5527,
+      "step": 293
+    },
+    {
+      "epoch": 1.1908860759493671,
+      "grad_norm": 0.5104440205509849,
+      "learning_rate": 3.769082529555359e-05,
+      "loss": 0.5334,
+      "step": 294
+    },
+    {
+      "epoch": 1.1949367088607594,
+      "grad_norm": 0.6033867847587765,
+      "learning_rate": 3.766427832919294e-05,
+      "loss": 0.5429,
+      "step": 295
+    },
+    {
+      "epoch": 1.198987341772152,
+      "grad_norm": 0.6432406532537754,
+      "learning_rate": 3.7637589097163024e-05,
+      "loss": 0.5355,
+      "step": 296
+    },
+    {
+      "epoch": 1.2030379746835442,
+      "grad_norm": 0.5683367932351043,
+      "learning_rate": 3.761075781441526e-05,
+      "loss": 0.5394,
+      "step": 297
+    },
+    {
+      "epoch": 1.2070886075949367,
+      "grad_norm": 0.5814424565297173,
+      "learning_rate": 3.75837846970451e-05,
+      "loss": 0.5206,
+      "step": 298
+    },
+    {
+      "epoch": 1.211139240506329,
+      "grad_norm": 0.4970350670517452,
+      "learning_rate": 3.755666996229032e-05,
+      "loss": 0.542,
+      "step": 299
+    },
+    {
+      "epoch": 1.2151898734177216,
+      "grad_norm": 0.49179287012868833,
+      "learning_rate": 3.752941382852927e-05,
+      "loss": 0.5374,
+      "step": 300
+    },
+    {
+      "epoch": 1.2192405063291138,
+      "grad_norm": 0.5213732016717519,
+      "learning_rate": 3.7502016515279115e-05,
+      "loss": 0.5533,
+      "step": 301
+    },
+    {
+      "epoch": 1.2232911392405064,
+      "grad_norm": 0.5414956729875842,
+      "learning_rate": 3.7474478243194043e-05,
+      "loss": 0.5273,
+      "step": 302
+    },
+    {
+      "epoch": 1.2273417721518987,
+      "grad_norm": 0.5596354400831836,
+      "learning_rate": 3.744679923406351e-05,
+      "loss": 0.5299,
+      "step": 303
+    },
+    {
+      "epoch": 1.2313924050632912,
+      "grad_norm": 0.4916263823215435,
+      "learning_rate": 3.741897971081043e-05,
+      "loss": 0.5288,
+      "step": 304
+    },
+    {
+      "epoch": 1.2354430379746835,
+      "grad_norm": 0.49352947461894664,
+      "learning_rate": 3.739101989748946e-05,
+      "loss": 0.5392,
+      "step": 305
+    },
+    {
+      "epoch": 1.239493670886076,
+      "grad_norm": 0.4976661912882745,
+      "learning_rate": 3.7362920019285066e-05,
+      "loss": 0.5256,
+      "step": 306
+    },
+    {
+      "epoch": 1.2435443037974683,
+      "grad_norm": 0.4548952635264936,
+      "learning_rate": 3.73346803025098e-05,
+      "loss": 0.5437,
+      "step": 307
+    },
+    {
+      "epoch": 1.2475949367088608,
+      "grad_norm": 0.5326448172724431,
+      "learning_rate": 3.730630097460247e-05,
+      "loss": 0.5341,
+      "step": 308
+    },
+    {
+      "epoch": 1.251645569620253,
+      "grad_norm": 0.4923223783083126,
+      "learning_rate": 3.727778226412628e-05,
+      "loss": 0.5336,
+      "step": 309
+    },
+    {
+      "epoch": 1.2556962025316456,
+      "grad_norm": 0.5187507286686032,
+      "learning_rate": 3.7249124400767006e-05,
+      "loss": 0.5259,
+      "step": 310
+    },
+    {
+      "epoch": 1.259746835443038,
+      "grad_norm": 0.5198366009432919,
+      "learning_rate": 3.722032761533114e-05,
+      "loss": 0.5401,
+      "step": 311
+    },
+    {
+      "epoch": 1.2637974683544304,
+      "grad_norm": 0.6294919993372073,
+      "learning_rate": 3.719139213974403e-05,
+      "loss": 0.5279,
+      "step": 312
+    },
+    {
+      "epoch": 1.2678481012658227,
+      "grad_norm": 0.3969712755535583,
+      "learning_rate": 3.7162318207048006e-05,
+      "loss": 0.5527,
+      "step": 313
+    },
+    {
+      "epoch": 1.2718987341772152,
+      "grad_norm": 0.7080014102034156,
+      "learning_rate": 3.713310605140055e-05,
+      "loss": 0.5378,
+      "step": 314
+    },
+    {
+      "epoch": 1.2759493670886077,
+      "grad_norm": 0.4726863416942518,
+      "learning_rate": 3.710375590807233e-05,
+      "loss": 0.5329,
+      "step": 315
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.5541128057742213,
+      "learning_rate": 3.7074268013445365e-05,
+      "loss": 0.5299,
+      "step": 316
+    },
+    {
+      "epoch": 1.2840506329113923,
+      "grad_norm": 0.5609004947097687,
+      "learning_rate": 3.7044642605011114e-05,
+      "loss": 0.5353,
+      "step": 317
+    },
+    {
+      "epoch": 1.2881012658227848,
+      "grad_norm": 0.4290085758455709,
+      "learning_rate": 3.701487992136854e-05,
+      "loss": 0.5307,
+      "step": 318
+    },
+    {
+      "epoch": 1.2921518987341774,
+      "grad_norm": 0.5160335951692513,
+      "learning_rate": 3.69849802022222e-05,
+      "loss": 0.5564,
+      "step": 319
+    },
+    {
+      "epoch": 1.2962025316455696,
+      "grad_norm": 0.492471103155741,
+      "learning_rate": 3.6954943688380334e-05,
+      "loss": 0.5513,
+      "step": 320
+    },
+    {
+      "epoch": 1.300253164556962,
+      "grad_norm": 0.601526750861106,
+      "learning_rate": 3.692477062175289e-05,
+      "loss": 0.5283,
+      "step": 321
+    },
+    {
+      "epoch": 1.3043037974683545,
+      "grad_norm": 0.4163494246942371,
+      "learning_rate": 3.689446124534958e-05,
+      "loss": 0.5289,
+      "step": 322
+    },
+    {
+      "epoch": 1.308354430379747,
+      "grad_norm": 0.5486955923764949,
+      "learning_rate": 3.686401580327799e-05,
+      "loss": 0.5292,
+      "step": 323
+    },
+    {
+      "epoch": 1.3124050632911393,
+      "grad_norm": 0.46187387020444404,
+      "learning_rate": 3.683343454074149e-05,
+      "loss": 0.5277,
+      "step": 324
+    },
+    {
+      "epoch": 1.3164556962025316,
+      "grad_norm": 0.5493078925523197,
+      "learning_rate": 3.6802717704037386e-05,
+      "loss": 0.5369,
+      "step": 325
+    },
+    {
+      "epoch": 1.320506329113924,
+      "grad_norm": 0.570375313440836,
+      "learning_rate": 3.6771865540554855e-05,
+      "loss": 0.5343,
+      "step": 326
+    },
+    {
+      "epoch": 1.3245569620253166,
+      "grad_norm": 0.5291812076263362,
+      "learning_rate": 3.674087829877297e-05,
+      "loss": 0.5281,
+      "step": 327
+    },
+    {
+      "epoch": 1.3286075949367089,
+      "grad_norm": 0.4529103252597112,
+      "learning_rate": 3.6709756228258735e-05,
+      "loss": 0.5288,
+      "step": 328
+    },
+    {
+      "epoch": 1.3326582278481012,
+      "grad_norm": 0.5728463072354482,
+      "learning_rate": 3.667849957966501e-05,
+      "loss": 0.525,
+      "step": 329
+    },
+    {
+      "epoch": 1.3367088607594937,
+      "grad_norm": 0.5170611304457063,
+      "learning_rate": 3.6647108604728546e-05,
+      "loss": 0.534,
+      "step": 330
+    },
+    {
+      "epoch": 1.3407594936708862,
+      "grad_norm": 0.5490170671659294,
+      "learning_rate": 3.661558355626795e-05,
+      "loss": 0.5435,
+      "step": 331
+    },
+    {
+      "epoch": 1.3448101265822785,
+      "grad_norm": 0.518368277602232,
+      "learning_rate": 3.658392468818163e-05,
+      "loss": 0.5417,
+      "step": 332
+    },
+    {
+      "epoch": 1.3488607594936708,
+      "grad_norm": 0.5694234857214804,
+      "learning_rate": 3.655213225544574e-05,
+      "loss": 0.5344,
+      "step": 333
+    },
+    {
+      "epoch": 1.3529113924050633,
+      "grad_norm": 0.6243578050696611,
+      "learning_rate": 3.652020651411218e-05,
+      "loss": 0.5464,
+      "step": 334
+    },
+    {
+      "epoch": 1.3569620253164558,
+      "grad_norm": 0.6053735055573898,
+      "learning_rate": 3.6488147721306474e-05,
+      "loss": 0.5372,
+      "step": 335
+    },
+    {
+      "epoch": 1.3610126582278481,
+      "grad_norm": 0.49571712398077694,
+      "learning_rate": 3.645595613522574e-05,
+      "loss": 0.5258,
+      "step": 336
+    },
+    {
+      "epoch": 1.3650632911392404,
+      "grad_norm": 0.7631708507540599,
+      "learning_rate": 3.642363201513657e-05,
+      "loss": 0.5388,
+      "step": 337
+    },
+    {
+      "epoch": 1.369113924050633,
+      "grad_norm": 0.49535820931325814,
+      "learning_rate": 3.6391175621373006e-05,
+      "loss": 0.5503,
+      "step": 338
+    },
+    {
+      "epoch": 1.3731645569620254,
+      "grad_norm": 0.6119269891217433,
+      "learning_rate": 3.6358587215334355e-05,
+      "loss": 0.5297,
+      "step": 339
+    },
+    {
+      "epoch": 1.3772151898734177,
+      "grad_norm": 0.501600544394518,
+      "learning_rate": 3.632586705948318e-05,
+      "loss": 0.5475,
+      "step": 340
+    },
+    {
+      "epoch": 1.38126582278481,
+      "grad_norm": 0.5215850521691904,
+      "learning_rate": 3.629301541734311e-05,
+      "loss": 0.5462,
+      "step": 341
+    },
+    {
+      "epoch": 1.3853164556962025,
+      "grad_norm": 0.5165209027371007,
+      "learning_rate": 3.626003255349676e-05,
+      "loss": 0.5472,
+      "step": 342
+    },
+    {
+      "epoch": 1.389367088607595,
+      "grad_norm": 0.5596515213737597,
+      "learning_rate": 3.622691873358357e-05,
+      "loss": 0.5368,
+      "step": 343
+    },
+    {
+      "epoch": 1.3934177215189874,
+      "grad_norm": 0.5244111621042852,
+      "learning_rate": 3.61936742242977e-05,
+      "loss": 0.5395,
+      "step": 344
+    },
+    {
+      "epoch": 1.3974683544303796,
+      "grad_norm": 0.6104033168397763,
+      "learning_rate": 3.6160299293385864e-05,
+      "loss": 0.531,
+      "step": 345
+    },
+    {
+      "epoch": 1.4015189873417722,
+      "grad_norm": 0.4856843732819136,
+      "learning_rate": 3.612679420964516e-05,
+      "loss": 0.54,
+      "step": 346
+    },
+    {
+      "epoch": 1.4055696202531647,
+      "grad_norm": 0.6315911405846727,
+      "learning_rate": 3.609315924292092e-05,
+      "loss": 0.5411,
+      "step": 347
+    },
+    {
+      "epoch": 1.409620253164557,
+      "grad_norm": 0.4966858638536871,
+      "learning_rate": 3.6059394664104554e-05,
+      "loss": 0.5405,
+      "step": 348
+    },
+    {
+      "epoch": 1.4136708860759493,
+      "grad_norm": 0.5468352945546916,
+      "learning_rate": 3.602550074513133e-05,
+      "loss": 0.5399,
+      "step": 349
+    },
+    {
+      "epoch": 1.4177215189873418,
+      "grad_norm": 0.5085153077506748,
+      "learning_rate": 3.599147775897822e-05,
+      "loss": 0.5329,
+      "step": 350
+    },
+    {
+      "epoch": 1.4217721518987343,
+      "grad_norm": 0.5603516210162783,
+      "learning_rate": 3.595732597966167e-05,
+      "loss": 0.5417,
+      "step": 351
+    },
+    {
+      "epoch": 1.4258227848101266,
+      "grad_norm": 0.5199875623486893,
+      "learning_rate": 3.592304568223542e-05,
+      "loss": 0.5221,
+      "step": 352
+    },
+    {
+      "epoch": 1.4298734177215189,
+      "grad_norm": 0.5840228600700326,
+      "learning_rate": 3.588863714278826e-05,
+      "loss": 0.5307,
+      "step": 353
+    },
+    {
+      "epoch": 1.4339240506329114,
+      "grad_norm": 0.42374731263608295,
+      "learning_rate": 3.585410063844186e-05,
+      "loss": 0.5368,
+      "step": 354
+    },
+    {
+      "epoch": 1.437974683544304,
+      "grad_norm": 0.4953553884140005,
+      "learning_rate": 3.581943644734846e-05,
+      "loss": 0.5205,
+      "step": 355
+    },
+    {
+      "epoch": 1.4420253164556962,
+      "grad_norm": 0.48777377865316485,
+      "learning_rate": 3.578464484868869e-05,
+      "loss": 0.5292,
+      "step": 356
+    },
+    {
+      "epoch": 1.4460759493670885,
+      "grad_norm": 0.47746369419567036,
+      "learning_rate": 3.5749726122669316e-05,
+      "loss": 0.5184,
+      "step": 357
+    },
+    {
+      "epoch": 1.450126582278481,
+      "grad_norm": 0.4938036694373018,
+      "learning_rate": 3.5714680550520943e-05,
+      "loss": 0.5293,
+      "step": 358
+    },
+    {
+      "epoch": 1.4541772151898735,
+      "grad_norm": 0.4923828350388037,
+      "learning_rate": 3.5679508414495794e-05,
+      "loss": 0.5328,
+      "step": 359
+    },
+    {
+      "epoch": 1.4582278481012658,
+      "grad_norm": 0.45564606917392037,
+      "learning_rate": 3.564420999786543e-05,
+      "loss": 0.5386,
+      "step": 360
+    },
+    {
+      "epoch": 1.4622784810126581,
+      "grad_norm": 0.4854775429118622,
+      "learning_rate": 3.560878558491842e-05,
+      "loss": 0.5337,
+      "step": 361
+    },
+    {
+      "epoch": 1.4663291139240506,
+      "grad_norm": 0.47039064633388555,
+      "learning_rate": 3.5573235460958145e-05,
+      "loss": 0.5283,
+      "step": 362
+    },
+    {
+      "epoch": 1.4703797468354431,
+      "grad_norm": 0.44543016174585937,
+      "learning_rate": 3.553755991230039e-05,
+      "loss": 0.5406,
+      "step": 363
+    },
+    {
+      "epoch": 1.4744303797468354,
+      "grad_norm": 0.48248441501444084,
+      "learning_rate": 3.5501759226271144e-05,
+      "loss": 0.5352,
+      "step": 364
+    },
+    {
+      "epoch": 1.4784810126582277,
+      "grad_norm": 0.4533606198339921,
+      "learning_rate": 3.546583369120419e-05,
+      "loss": 0.5274,
+      "step": 365
+    },
+    {
+      "epoch": 1.4825316455696202,
+      "grad_norm": 0.46405702006824134,
+      "learning_rate": 3.5429783596438864e-05,
+      "loss": 0.538,
+      "step": 366
+    },
+    {
+      "epoch": 1.4865822784810128,
+      "grad_norm": 0.5011528504453227,
+      "learning_rate": 3.539360923231766e-05,
+      "loss": 0.5322,
+      "step": 367
+    },
+    {
+      "epoch": 1.490632911392405,
+      "grad_norm": 0.46934063452742525,
+      "learning_rate": 3.535731089018394e-05,
+      "loss": 0.533,
+      "step": 368
+    },
+    {
+      "epoch": 1.4946835443037974,
+      "grad_norm": 0.4067328500506999,
+      "learning_rate": 3.532088886237956e-05,
+      "loss": 0.5442,
+      "step": 369
+    },
+    {
+      "epoch": 1.4987341772151899,
+      "grad_norm": 0.5046417808252452,
+      "learning_rate": 3.528434344224253e-05,
+      "loss": 0.5532,
+      "step": 370
+    },
+    {
+      "epoch": 1.5027848101265824,
+      "grad_norm": 0.4214580587662184,
+      "learning_rate": 3.524767492410464e-05,
+      "loss": 0.5419,
+      "step": 371
+    },
+    {
+      "epoch": 1.5068354430379747,
+      "grad_norm": 0.4932188183816339,
+      "learning_rate": 3.521088360328908e-05,
+      "loss": 0.5389,
+      "step": 372
+    },
+    {
+      "epoch": 1.510886075949367,
+      "grad_norm": 0.5050993654603906,
+      "learning_rate": 3.517396977610811e-05,
+      "loss": 0.5132,
+      "step": 373
+    },
+    {
+      "epoch": 1.5149367088607595,
+      "grad_norm": 0.48220707436427046,
+      "learning_rate": 3.5136933739860595e-05,
+      "loss": 0.542,
+      "step": 374
+    },
+    {
+      "epoch": 1.518987341772152,
+      "grad_norm": 0.4391851412326274,
+      "learning_rate": 3.509977579282971e-05,
+      "loss": 0.5324,
+      "step": 375
+    },
+    {
+      "epoch": 1.5230379746835443,
+      "grad_norm": 0.46086219385868377,
+      "learning_rate": 3.5062496234280424e-05,
+      "loss": 0.5265,
+      "step": 376
+    },
+    {
+      "epoch": 1.5270886075949366,
+      "grad_norm": 0.4101521097045103,
+      "learning_rate": 3.502509536445719e-05,
+      "loss": 0.539,
+      "step": 377
+    },
+    {
+      "epoch": 1.531139240506329,
+      "grad_norm": 0.45089551987465615,
+      "learning_rate": 3.498757348458147e-05,
+      "loss": 0.5354,
+      "step": 378
+    },
+    {
+      "epoch": 1.5351898734177216,
+      "grad_norm": 0.4111739465253182,
+      "learning_rate": 3.4949930896849324e-05,
+      "loss": 0.5477,
+      "step": 379
+    },
+    {
+      "epoch": 1.539240506329114,
+      "grad_norm": 0.490577712043314,
+      "learning_rate": 3.491216790442899e-05,
+      "loss": 0.5369,
+      "step": 380
+    },
+    {
+      "epoch": 1.5432911392405062,
+      "grad_norm": 0.5077089789809147,
+      "learning_rate": 3.487428481145839e-05,
+      "loss": 0.5319,
+      "step": 381
+    },
+    {
+      "epoch": 1.5473417721518987,
+      "grad_norm": 0.43533326148189694,
+      "learning_rate": 3.483628192304278e-05,
+      "loss": 0.5332,
+      "step": 382
+    },
+    {
+      "epoch": 1.5513924050632912,
+      "grad_norm": 0.5455458450811859,
+      "learning_rate": 3.479815954525219e-05,
+      "loss": 0.5382,
+      "step": 383
+    },
+    {
+      "epoch": 1.5554430379746835,
+      "grad_norm": 0.46015967466631696,
+      "learning_rate": 3.475991798511899e-05,
+      "loss": 0.5327,
+      "step": 384
+    },
+    {
+      "epoch": 1.5594936708860758,
+      "grad_norm": 0.5469628123139261,
+      "learning_rate": 3.4721557550635464e-05,
+      "loss": 0.5406,
+      "step": 385
+    },
+    {
+      "epoch": 1.5635443037974683,
+      "grad_norm": 0.4832530048241379,
+      "learning_rate": 3.468307855075128e-05,
+      "loss": 0.5467,
+      "step": 386
+    },
+    {
+      "epoch": 1.5675949367088609,
+      "grad_norm": 0.4391786668513969,
+      "learning_rate": 3.4644481295371005e-05,
+      "loss": 0.5372,
+      "step": 387
+    },
+    {
+      "epoch": 1.5716455696202531,
+      "grad_norm": 0.499576090020655,
+      "learning_rate": 3.460576609535163e-05,
+      "loss": 0.536,
+      "step": 388
+    },
+    {
+      "epoch": 1.5756962025316454,
+      "grad_norm": 0.43335991145247055,
+      "learning_rate": 3.456693326250006e-05,
+      "loss": 0.5156,
+      "step": 389
+    },
+    {
+      "epoch": 1.579746835443038,
+      "grad_norm": 0.4933146130429338,
+      "learning_rate": 3.452798310957058e-05,
+      "loss": 0.5424,
+      "step": 390
+    },
+    {
+      "epoch": 1.5837974683544305,
+      "grad_norm": 0.4064110496139781,
+      "learning_rate": 3.4488915950262386e-05,
+      "loss": 0.5256,
+      "step": 391
+    },
+    {
+      "epoch": 1.5878481012658228,
+      "grad_norm": 0.447251225659645,
+      "learning_rate": 3.4449732099216985e-05,
+      "loss": 0.5264,
+      "step": 392
+    },
+    {
+      "epoch": 1.591898734177215,
+      "grad_norm": 0.40766110130251637,
+      "learning_rate": 3.441043187201574e-05,
+      "loss": 0.5352,
+      "step": 393
+    },
+    {
+      "epoch": 1.5959493670886076,
+      "grad_norm": 0.44390011193365964,
+      "learning_rate": 3.437101558517728e-05,
+      "loss": 0.541,
+      "step": 394
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.4663305819802249,
+      "learning_rate": 3.433148355615496e-05,
+      "loss": 0.5311,
+      "step": 395
+    },
+    {
+      "epoch": 1.6040506329113924,
+      "grad_norm": 0.42184316981714365,
+      "learning_rate": 3.4291836103334294e-05,
+      "loss": 0.5528,
+      "step": 396
+    },
+    {
+      "epoch": 1.6081012658227847,
+      "grad_norm": 0.4242124435232366,
+      "learning_rate": 3.425207354603043e-05,
+      "loss": 0.54,
+      "step": 397
+    },
+    {
+      "epoch": 1.6121518987341772,
+      "grad_norm": 0.46204081274356934,
+      "learning_rate": 3.421219620448553e-05,
+      "loss": 0.5284,
+      "step": 398
+    },
+    {
+      "epoch": 1.6162025316455697,
+      "grad_norm": 0.43301038092773003,
+      "learning_rate": 3.417220439986623e-05,
+      "loss": 0.5408,
+      "step": 399
+    },
+    {
+      "epoch": 1.620253164556962,
+      "grad_norm": 0.4163878312498914,
+      "learning_rate": 3.4132098454261024e-05,
+      "loss": 0.51,
+      "step": 400
+    },
+    {
+      "epoch": 1.6243037974683543,
+      "grad_norm": 0.38803322832471926,
+      "learning_rate": 3.4091878690677676e-05,
+      "loss": 0.5372,
+      "step": 401
+    },
+    {
+      "epoch": 1.6283544303797468,
+      "grad_norm": 0.4760543123357105,
+      "learning_rate": 3.405154543304065e-05,
+      "loss": 0.5331,
+      "step": 402
+    },
+    {
+      "epoch": 1.6324050632911393,
+      "grad_norm": 0.3674218626097115,
+      "learning_rate": 3.401109900618843e-05,
+      "loss": 0.5147,
+      "step": 403
+    },
+    {
+      "epoch": 1.6364556962025316,
+      "grad_norm": 0.4730676799359698,
+      "learning_rate": 3.3970539735870996e-05,
+      "loss": 0.529,
+      "step": 404
+    },
+    {
+      "epoch": 1.640506329113924,
+      "grad_norm": 0.4544856423656204,
+      "learning_rate": 3.392986794874714e-05,
+      "loss": 0.5467,
+      "step": 405
+    },
+    {
+      "epoch": 1.6445569620253164,
+      "grad_norm": 0.4587164457493161,
+      "learning_rate": 3.388908397238184e-05,
+      "loss": 0.5343,
+      "step": 406
+    },
+    {
+      "epoch": 1.648607594936709,
+      "grad_norm": 0.530724814986304,
+      "learning_rate": 3.384818813524362e-05,
+      "loss": 0.5324,
+      "step": 407
+    },
+    {
+      "epoch": 1.6526582278481012,
+      "grad_norm": 0.37637188407422517,
+      "learning_rate": 3.380718076670195e-05,
+      "loss": 0.5261,
+      "step": 408
+    },
+    {
+      "epoch": 1.6567088607594935,
+      "grad_norm": 0.5394236911639888,
+      "learning_rate": 3.376606219702454e-05,
+      "loss": 0.5171,
+      "step": 409
+    },
+    {
+      "epoch": 1.660759493670886,
+      "grad_norm": 0.4162281851748918,
+      "learning_rate": 3.372483275737468e-05,
+      "loss": 0.5456,
+      "step": 410
+    },
+    {
+      "epoch": 1.6648101265822786,
+      "grad_norm": 0.5090674202780153,
+      "learning_rate": 3.368349277980861e-05,
+      "loss": 0.5296,
+      "step": 411
+    },
+    {
+      "epoch": 1.6688607594936709,
+      "grad_norm": 0.5307651614950271,
+      "learning_rate": 3.3642042597272844e-05,
+      "loss": 0.5426,
+      "step": 412
+    },
+    {
+      "epoch": 1.6729113924050631,
+      "grad_norm": 0.4885096041996164,
+      "learning_rate": 3.360048254360144e-05,
+      "loss": 0.5268,
+      "step": 413
+    },
+    {
+      "epoch": 1.6769620253164557,
+      "grad_norm": 0.45702469011808683,
+      "learning_rate": 3.355881295351336e-05,
+      "loss": 0.5269,
+      "step": 414
+    },
+    {
+      "epoch": 1.6810126582278482,
+      "grad_norm": 0.5410421653574236,
+      "learning_rate": 3.351703416260975e-05,
+      "loss": 0.533,
+      "step": 415
+    },
+    {
+      "epoch": 1.6850632911392405,
+      "grad_norm": 0.49341617021290207,
+      "learning_rate": 3.347514650737126e-05,
+      "loss": 0.524,
+      "step": 416
+    },
+    {
+      "epoch": 1.6891139240506328,
+      "grad_norm": 0.41433744166780373,
+      "learning_rate": 3.3433150325155295e-05,
+      "loss": 0.5429,
+      "step": 417
+    },
+    {
+      "epoch": 1.6931645569620253,
+      "grad_norm": 0.49233220433208813,
+      "learning_rate": 3.339104595419334e-05,
+      "loss": 0.5257,
+      "step": 418
+    },
+    {
+      "epoch": 1.6972151898734178,
+      "grad_norm": 0.4324858979845979,
+      "learning_rate": 3.3348833733588204e-05,
+      "loss": 0.5418,
+      "step": 419
+    },
+    {
+      "epoch": 1.70126582278481,
+      "grad_norm": 0.5091922919118782,
+      "learning_rate": 3.3306514003311305e-05,
+      "loss": 0.533,
+      "step": 420
+    },
+    {
+      "epoch": 1.7053164556962024,
+      "grad_norm": 0.47429267797841657,
+      "learning_rate": 3.326408710419996e-05,
+      "loss": 0.5274,
+      "step": 421
+    },
+    {
+      "epoch": 1.709367088607595,
+      "grad_norm": 0.5266903682224798,
+      "learning_rate": 3.322155337795454e-05,
+      "loss": 0.5363,
+      "step": 422
+    },
+    {
+      "epoch": 1.7134177215189874,
+      "grad_norm": 0.4486544351399723,
+      "learning_rate": 3.317891316713587e-05,
+      "loss": 0.5148,
+      "step": 423
+    },
+    {
+      "epoch": 1.7174683544303797,
+      "grad_norm": 0.45986655174127256,
+      "learning_rate": 3.313616681516231e-05,
+      "loss": 0.5374,
+      "step": 424
+    },
+    {
+      "epoch": 1.721518987341772,
+      "grad_norm": 0.40040510658845807,
+      "learning_rate": 3.309331466630713e-05,
+      "loss": 0.5352,
+      "step": 425
+    },
+    {
+      "epoch": 1.7255696202531645,
+      "grad_norm": 0.461826696144098,
+      "learning_rate": 3.305035706569563e-05,
+      "loss": 0.5296,
+      "step": 426
+    },
+    {
+      "epoch": 1.729620253164557,
+      "grad_norm": 0.4516785242772318,
+      "learning_rate": 3.3007294359302433e-05,
+      "loss": 0.5251,
+      "step": 427
+    },
+    {
+      "epoch": 1.7336708860759493,
+      "grad_norm": 0.4151808846623754,
+      "learning_rate": 3.296412689394864e-05,
+      "loss": 0.5228,
+      "step": 428
+    },
+    {
+      "epoch": 1.7377215189873416,
+      "grad_norm": 0.4510358252159107,
+      "learning_rate": 3.292085501729909e-05,
+      "loss": 0.5287,
+      "step": 429
+    },
+    {
+      "epoch": 1.7417721518987341,
+      "grad_norm": 0.46765646331437555,
+      "learning_rate": 3.2877479077859534e-05,
+      "loss": 0.5188,
+      "step": 430
+    },
+    {
+      "epoch": 1.7458227848101266,
+      "grad_norm": 0.4973162426896262,
+      "learning_rate": 3.283399942497381e-05,
+      "loss": 0.5239,
+      "step": 431
+    },
+    {
+      "epoch": 1.749873417721519,
+      "grad_norm": 0.4601053766427905,
+      "learning_rate": 3.279041640882108e-05,
+      "loss": 0.533,
+      "step": 432
+    },
+    {
+      "epoch": 1.7539240506329112,
+      "grad_norm": 0.4861480717729161,
+      "learning_rate": 3.2746730380412964e-05,
+      "loss": 0.5378,
+      "step": 433
+    },
+    {
+      "epoch": 1.7579746835443038,
+      "grad_norm": 0.422536087233182,
+      "learning_rate": 3.2702941691590726e-05,
+      "loss": 0.5269,
+      "step": 434
+    },
+    {
+      "epoch": 1.7620253164556963,
+      "grad_norm": 0.4919314166232016,
+      "learning_rate": 3.265905069502244e-05,
+      "loss": 0.5361,
+      "step": 435
+    },
+    {
+      "epoch": 1.7660759493670886,
+      "grad_norm": 0.4304639026089148,
+      "learning_rate": 3.261505774420016e-05,
+      "loss": 0.541,
+      "step": 436
+    },
+    {
+      "epoch": 1.7701265822784809,
+      "grad_norm": 0.445351458932588,
+      "learning_rate": 3.257096319343707e-05,
+      "loss": 0.5291,
+      "step": 437
+    },
+    {
+      "epoch": 1.7741772151898734,
+      "grad_norm": 0.41031459144605387,
+      "learning_rate": 3.2526767397864614e-05,
+      "loss": 0.5229,
+      "step": 438
+    },
+    {
+      "epoch": 1.7782278481012659,
+      "grad_norm": 0.4033169237829173,
+      "learning_rate": 3.248247071342966e-05,
+      "loss": 0.5277,
+      "step": 439
+    },
+    {
+      "epoch": 1.7822784810126582,
+      "grad_norm": 0.4474126746226738,
+      "learning_rate": 3.243807349689161e-05,
+      "loss": 0.5308,
+      "step": 440
+    },
+    {
+      "epoch": 1.7863291139240505,
+      "grad_norm": 0.442506115984171,
+      "learning_rate": 3.2393576105819544e-05,
+      "loss": 0.5295,
+      "step": 441
+    },
+    {
+      "epoch": 1.790379746835443,
+      "grad_norm": 0.38252811108676976,
+      "learning_rate": 3.2348978898589333e-05,
+      "loss": 0.5249,
+      "step": 442
+    },
+    {
+      "epoch": 1.7944303797468355,
+      "grad_norm": 0.5556485032893761,
+      "learning_rate": 3.230428223438075e-05,
+      "loss": 0.5331,
+      "step": 443
+    },
+    {
+      "epoch": 1.7984810126582278,
+      "grad_norm": 0.3788220058636528,
+      "learning_rate": 3.225948647317459e-05,
+      "loss": 0.5291,
+      "step": 444
+    },
+    {
+      "epoch": 1.80253164556962,
+      "grad_norm": 0.5256558555387171,
+      "learning_rate": 3.2214591975749745e-05,
+      "loss": 0.5385,
+      "step": 445
+    },
+    {
+      "epoch": 1.8065822784810126,
+      "grad_norm": 0.47170527103206705,
+      "learning_rate": 3.216959910368034e-05,
+      "loss": 0.5342,
+      "step": 446
+    },
+    {
+      "epoch": 1.8106329113924051,
+      "grad_norm": 0.4747292098863455,
+      "learning_rate": 3.212450821933277e-05,
+      "loss": 0.527,
+      "step": 447
+    },
+    {
+      "epoch": 1.8146835443037974,
+      "grad_norm": 0.4717752579035982,
+      "learning_rate": 3.207931968586281e-05,
+      "loss": 0.5515,
+      "step": 448
+    },
+    {
+      "epoch": 1.81873417721519,
+      "grad_norm": 0.4811368317814634,
+      "learning_rate": 3.203403386721272e-05,
+      "loss": 0.5379,
+      "step": 449
+    },
+    {
+      "epoch": 1.8227848101265822,
+      "grad_norm": 0.4998489791505645,
+      "learning_rate": 3.1988651128108245e-05,
+      "loss": 0.5311,
+      "step": 450
+    },
+    {
+      "epoch": 1.8268354430379747,
+      "grad_norm": 0.47106252395143666,
+      "learning_rate": 3.194317183405573e-05,
+      "loss": 0.5325,
+      "step": 451
+    },
+    {
+      "epoch": 1.8308860759493673,
+      "grad_norm": 0.5659283147510659,
+      "learning_rate": 3.189759635133914e-05,
+      "loss": 0.5352,
+      "step": 452
+    },
+    {
+      "epoch": 1.8349367088607595,
+      "grad_norm": 0.4651233570125052,
+      "learning_rate": 3.185192504701718e-05,
+      "loss": 0.5213,
+      "step": 453
+    },
+    {
+      "epoch": 1.8389873417721518,
+      "grad_norm": 0.5342534155780265,
+      "learning_rate": 3.1806158288920234e-05,
+      "loss": 0.5443,
+      "step": 454
+    },
+    {
+      "epoch": 1.8430379746835444,
+      "grad_norm": 0.47641283015816316,
+      "learning_rate": 3.1760296445647477e-05,
+      "loss": 0.5228,
+      "step": 455
+    },
+    {
+      "epoch": 1.8470886075949369,
+      "grad_norm": 0.5201672755815242,
+      "learning_rate": 3.1714339886563896e-05,
+      "loss": 0.5367,
+      "step": 456
+    },
+    {
+      "epoch": 1.8511392405063292,
+      "grad_norm": 0.4617999886061925,
+      "learning_rate": 3.166828898179731e-05,
+      "loss": 0.5326,
+      "step": 457
+    },
+    {
+      "epoch": 1.8551898734177215,
+      "grad_norm": 0.480192793562171,
+      "learning_rate": 3.162214410223536e-05,
+      "loss": 0.5451,
+      "step": 458
+    },
+    {
+      "epoch": 1.859240506329114,
+      "grad_norm": 0.5260761319360927,
+      "learning_rate": 3.157590561952257e-05,
+      "loss": 0.5271,
+      "step": 459
+    },
+    {
+      "epoch": 1.8632911392405065,
+      "grad_norm": 0.44772366461349794,
+      "learning_rate": 3.152957390605732e-05,
+      "loss": 0.5225,
+      "step": 460
+    },
+    {
+      "epoch": 1.8673417721518988,
+      "grad_norm": 0.4549122408845582,
+      "learning_rate": 3.148314933498886e-05,
+      "loss": 0.5193,
+      "step": 461
+    },
+    {
+      "epoch": 1.871392405063291,
+      "grad_norm": 0.4516896354507125,
+      "learning_rate": 3.143663228021431e-05,
+      "loss": 0.5268,
+      "step": 462
+    },
+    {
+      "epoch": 1.8754430379746836,
+      "grad_norm": 0.40489127008905645,
+      "learning_rate": 3.1390023116375624e-05,
+      "loss": 0.5338,
+      "step": 463
+    },
+    {
+      "epoch": 1.879493670886076,
+      "grad_norm": 0.4292210673477139,
+      "learning_rate": 3.134332221885661e-05,
+      "loss": 0.545,
+      "step": 464
+    },
+    {
+      "epoch": 1.8835443037974684,
+      "grad_norm": 0.449039977679048,
+      "learning_rate": 3.129652996377987e-05,
+      "loss": 0.5457,
+      "step": 465
+    },
+    {
+      "epoch": 1.8875949367088607,
+      "grad_norm": 0.4215539842105942,
+      "learning_rate": 3.12496467280038e-05,
+      "loss": 0.533,
+      "step": 466
+    },
+    {
+      "epoch": 1.8916455696202532,
+      "grad_norm": 0.46942379340969476,
+      "learning_rate": 3.120267288911952e-05,
+      "loss": 0.5294,
+      "step": 467
+    },
+    {
+      "epoch": 1.8956962025316457,
+      "grad_norm": 0.41083911228825154,
+      "learning_rate": 3.11556088254479e-05,
+      "loss": 0.5354,
+      "step": 468
+    },
+    {
+      "epoch": 1.899746835443038,
+      "grad_norm": 0.4080037422428436,
+      "learning_rate": 3.11084549160364e-05,
+      "loss": 0.5368,
+      "step": 469
+    },
+    {
+      "epoch": 1.9037974683544303,
+      "grad_norm": 0.43734503727402935,
+      "learning_rate": 3.106121154065615e-05,
+      "loss": 0.5261,
+      "step": 470
+    },
+    {
+      "epoch": 1.9078481012658228,
+      "grad_norm": 0.40670277885822687,
+      "learning_rate": 3.1013879079798805e-05,
+      "loss": 0.5225,
+      "step": 471
+    },
+    {
+      "epoch": 1.9118987341772153,
+      "grad_norm": 0.42955658734842045,
+      "learning_rate": 3.096645791467348e-05,
+      "loss": 0.5366,
+      "step": 472
+    },
+    {
+      "epoch": 1.9159493670886076,
+      "grad_norm": 0.3728306954542638,
+      "learning_rate": 3.091894842720373e-05,
+      "loss": 0.5113,
+      "step": 473
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.43303651590683956,
+      "learning_rate": 3.0871351000024425e-05,
+      "loss": 0.527,
+      "step": 474
+    },
+    {
+      "epoch": 1.9240506329113924,
+      "grad_norm": 0.42343470413776335,
+      "learning_rate": 3.0823666016478716e-05,
+      "loss": 0.536,
+      "step": 475
+    },
+    {
+      "epoch": 1.928101265822785,
+      "grad_norm": 0.38323988842100454,
+      "learning_rate": 3.0775893860614896e-05,
+      "loss": 0.5301,
+      "step": 476
+    },
+    {
+      "epoch": 1.9321518987341773,
+      "grad_norm": 0.4050328085472597,
+      "learning_rate": 3.0728034917183336e-05,
+      "loss": 0.5315,
+      "step": 477
+    },
+    {
+      "epoch": 1.9362025316455695,
+      "grad_norm": 0.4188743747310876,
+      "learning_rate": 3.06800895716334e-05,
+      "loss": 0.5353,
+      "step": 478
+    },
+    {
+      "epoch": 1.940253164556962,
+      "grad_norm": 0.43020323953510137,
+      "learning_rate": 3.063205821011029e-05,
+      "loss": 0.5283,
+      "step": 479
+    },
+    {
+      "epoch": 1.9443037974683546,
+      "grad_norm": 0.4596811380953446,
+      "learning_rate": 3.0583941219452016e-05,
+      "loss": 0.519,
+      "step": 480
+    },
+    {
+      "epoch": 1.9483544303797469,
+      "grad_norm": 0.5487548889430699,
+      "learning_rate": 3.053573898718618e-05,
+      "loss": 0.5221,
+      "step": 481
+    },
+    {
+      "epoch": 1.9524050632911392,
+      "grad_norm": 0.44543019264896694,
+      "learning_rate": 3.0487451901526956e-05,
+      "loss": 0.5357,
+      "step": 482
+    },
+    {
+      "epoch": 1.9564556962025317,
+      "grad_norm": 0.558923817892189,
+      "learning_rate": 3.0439080351371875e-05,
+      "loss": 0.5289,
+      "step": 483
+    },
+    {
+      "epoch": 1.9605063291139242,
+      "grad_norm": 0.43326025955387454,
+      "learning_rate": 3.0390624726298764e-05,
+      "loss": 0.5336,
+      "step": 484
+    },
+    {
+      "epoch": 1.9645569620253165,
+      "grad_norm": 0.5056482652338182,
+      "learning_rate": 3.034208541656255e-05,
+      "loss": 0.5116,
+      "step": 485
+    },
+    {
+      "epoch": 1.9686075949367088,
+      "grad_norm": 0.3921301627978623,
+      "learning_rate": 3.029346281309218e-05,
+      "loss": 0.5392,
+      "step": 486
+    },
+    {
+      "epoch": 1.9726582278481013,
+      "grad_norm": 0.47746905391379274,
+      "learning_rate": 3.0244757307487415e-05,
+      "loss": 0.5267,
+      "step": 487
+    },
+    {
+      "epoch": 1.9767088607594938,
+      "grad_norm": 0.39894923300662993,
+      "learning_rate": 3.019596929201569e-05,
+      "loss": 0.5279,
+      "step": 488
+    },
+    {
+      "epoch": 1.980759493670886,
+      "grad_norm": 0.45526461150963027,
+      "learning_rate": 3.0147099159608985e-05,
+      "loss": 0.529,
+      "step": 489
+    },
+    {
+      "epoch": 1.9848101265822784,
+      "grad_norm": 0.42551463960637625,
+      "learning_rate": 3.0098147303860616e-05,
+      "loss": 0.5182,
+      "step": 490
+    },
+    {
+      "epoch": 1.988860759493671,
+      "grad_norm": 0.539768579642405,
+      "learning_rate": 3.0049114119022117e-05,
+      "loss": 0.5319,
+      "step": 491
+    },
+    {
+      "epoch": 1.9929113924050634,
+      "grad_norm": 0.40271191483593805,
+      "learning_rate": 3.0000000000000004e-05,
+      "loss": 0.5371,
+      "step": 492
+    },
+    {
+      "epoch": 1.9969620253164557,
+      "grad_norm": 0.6338436528474023,
+      "learning_rate": 2.995080534235264e-05,
+      "loss": 0.5262,
+      "step": 493
+    },
+    {
+      "epoch": 2.001012658227848,
+      "grad_norm": 0.39059202322744996,
+      "learning_rate": 2.9901530542287044e-05,
+      "loss": 0.5164,
+      "step": 494
+    },
+    {
+      "epoch": 2.0050632911392405,
+      "grad_norm": 0.6515591741780411,
+      "learning_rate": 2.9852175996655676e-05,
+      "loss": 0.4384,
+      "step": 495
+    },
+    {
+      "epoch": 2.009113924050633,
+      "grad_norm": 0.714819769824571,
+      "learning_rate": 2.980274210295326e-05,
+      "loss": 0.4481,
+      "step": 496
+    },
+    {
+      "epoch": 2.013164556962025,
+      "grad_norm": 0.6941470215578036,
+      "learning_rate": 2.9753229259313578e-05,
+      "loss": 0.4581,
+      "step": 497
+    },
+    {
+      "epoch": 2.0172151898734176,
+      "grad_norm": 0.43111243912006514,
+      "learning_rate": 2.9703637864506274e-05,
+      "loss": 0.4482,
+      "step": 498
+    },
+    {
+      "epoch": 2.02126582278481,
+      "grad_norm": 0.5846912074258437,
+      "learning_rate": 2.965396831793362e-05,
+      "loss": 0.442,
+      "step": 499
+    },
+    {
+      "epoch": 2.0253164556962027,
+      "grad_norm": 0.46201666160255517,
+      "learning_rate": 2.9604221019627316e-05,
+      "loss": 0.4416,
+      "step": 500
+    },
+    {
+      "epoch": 2.0293670886075947,
+      "grad_norm": 0.54810276645589,
+      "learning_rate": 2.955439637024526e-05,
+      "loss": 0.4417,
+      "step": 501
+    },
+    {
+      "epoch": 2.0334177215189873,
+      "grad_norm": 0.4342574941020381,
+      "learning_rate": 2.9504494771068334e-05,
+      "loss": 0.4564,
+      "step": 502
+    },
+    {
+      "epoch": 2.0374683544303798,
+      "grad_norm": 0.5755083877994038,
+      "learning_rate": 2.9454516623997156e-05,
+      "loss": 0.4449,
+      "step": 503
+    },
+    {
+      "epoch": 2.0415189873417723,
+      "grad_norm": 0.41334535549943463,
+      "learning_rate": 2.9404462331548847e-05,
+      "loss": 0.4447,
+      "step": 504
+    },
+    {
+      "epoch": 2.0455696202531644,
+      "grad_norm": 0.5404122103885102,
+      "learning_rate": 2.93543322968538e-05,
+      "loss": 0.4261,
+      "step": 505
+    },
+    {
+      "epoch": 2.049620253164557,
+      "grad_norm": 0.48362180432122065,
+      "learning_rate": 2.9304126923652428e-05,
+      "loss": 0.4345,
+      "step": 506
+    },
+    {
+      "epoch": 2.0536708860759494,
+      "grad_norm": 0.4857588052663968,
+      "learning_rate": 2.9253846616291896e-05,
+      "loss": 0.4523,
+      "step": 507
+    },
+    {
+      "epoch": 2.057721518987342,
+      "grad_norm": 0.42821800875500793,
+      "learning_rate": 2.9203491779722896e-05,
+      "loss": 0.4322,
+      "step": 508
+    },
+    {
+      "epoch": 2.061772151898734,
+      "grad_norm": 0.4520016743406533,
+      "learning_rate": 2.9153062819496357e-05,
+      "loss": 0.4405,
+      "step": 509
+    },
+    {
+      "epoch": 2.0658227848101265,
+      "grad_norm": 0.4522237536161529,
+      "learning_rate": 2.9102560141760178e-05,
+      "loss": 0.4472,
+      "step": 510
+    },
+    {
+      "epoch": 2.069873417721519,
+      "grad_norm": 0.4358696246335823,
+      "learning_rate": 2.9051984153256004e-05,
+      "loss": 0.4322,
+      "step": 511
+    },
+    {
+      "epoch": 2.0739240506329115,
+      "grad_norm": 0.4272557802889532,
+      "learning_rate": 2.900133526131588e-05,
+      "loss": 0.4506,
+      "step": 512
+    },
+    {
+      "epoch": 2.0779746835443036,
+      "grad_norm": 0.4701219166435722,
+      "learning_rate": 2.8950613873859025e-05,
+      "loss": 0.4616,
+      "step": 513
+    },
+    {
+      "epoch": 2.082025316455696,
+      "grad_norm": 0.4334836340135577,
+      "learning_rate": 2.8899820399388515e-05,
+      "loss": 0.4289,
+      "step": 514
+    },
+    {
+      "epoch": 2.0860759493670886,
+      "grad_norm": 0.48913223005586887,
+      "learning_rate": 2.8848955246988012e-05,
+      "loss": 0.4709,
+      "step": 515
+    },
+    {
+      "epoch": 2.090126582278481,
+      "grad_norm": 0.41108391187767684,
+      "learning_rate": 2.879801882631847e-05,
+      "loss": 0.4582,
+      "step": 516
+    },
+    {
+      "epoch": 2.094177215189873,
+      "grad_norm": 0.4260310368465769,
+      "learning_rate": 2.8747011547614808e-05,
+      "loss": 0.4596,
+      "step": 517
+    },
+    {
+      "epoch": 2.0982278481012657,
+      "grad_norm": 0.43060088759546494,
+      "learning_rate": 2.8695933821682635e-05,
+      "loss": 0.4319,
+      "step": 518
+    },
+    {
+      "epoch": 2.1022784810126582,
+      "grad_norm": 0.3911164344118625,
+      "learning_rate": 2.864478605989494e-05,
+      "loss": 0.4559,
+      "step": 519
+    },
+    {
+      "epoch": 2.1063291139240508,
+      "grad_norm": 0.38036783311987077,
+      "learning_rate": 2.8593568674188765e-05,
+      "loss": 0.44,
+      "step": 520
+    },
+    {
+      "epoch": 2.110379746835443,
+      "grad_norm": 0.3775385802199869,
+      "learning_rate": 2.8542282077061892e-05,
+      "loss": 0.4315,
+      "step": 521
+    },
+    {
+      "epoch": 2.1144303797468353,
+      "grad_norm": 0.45401161996405487,
+      "learning_rate": 2.8490926681569523e-05,
+      "loss": 0.4259,
+      "step": 522
+    },
+    {
+      "epoch": 2.118481012658228,
+      "grad_norm": 0.4118113243706397,
+      "learning_rate": 2.8439502901320956e-05,
+      "loss": 0.4402,
+      "step": 523
+    },
+    {
+      "epoch": 2.1225316455696204,
+      "grad_norm": 0.45304954929152796,
+      "learning_rate": 2.8388011150476237e-05,
+      "loss": 0.4214,
+      "step": 524
+    },
+    {
+      "epoch": 2.1265822784810124,
+      "grad_norm": 0.4366821210444535,
+      "learning_rate": 2.8336451843742866e-05,
+      "loss": 0.4495,
+      "step": 525
+    },
+    {
+      "epoch": 2.130632911392405,
+      "grad_norm": 0.4110696120366791,
+      "learning_rate": 2.8284825396372387e-05,
+      "loss": 0.4322,
+      "step": 526
+    },
+    {
+      "epoch": 2.1346835443037975,
+      "grad_norm": 0.4206794614041197,
+      "learning_rate": 2.8233132224157132e-05,
+      "loss": 0.4472,
+      "step": 527
+    },
+    {
+      "epoch": 2.13873417721519,
+      "grad_norm": 0.38192591942006054,
+      "learning_rate": 2.8181372743426805e-05,
+      "loss": 0.4379,
+      "step": 528
+    },
+    {
+      "epoch": 2.1427848101265825,
+      "grad_norm": 0.438036531151284,
+      "learning_rate": 2.8129547371045128e-05,
+      "loss": 0.4483,
+      "step": 529
+    },
+    {
+      "epoch": 2.1468354430379746,
+      "grad_norm": 0.4006984088445925,
+      "learning_rate": 2.8077656524406534e-05,
+      "loss": 0.471,
+      "step": 530
+    },
+    {
+      "epoch": 2.150886075949367,
+      "grad_norm": 0.39510614419673795,
+      "learning_rate": 2.802570062143278e-05,
+      "loss": 0.4282,
+      "step": 531
+    },
+    {
+      "epoch": 2.1549367088607596,
+      "grad_norm": 0.3854861388333837,
+      "learning_rate": 2.7973680080569555e-05,
+      "loss": 0.4462,
+      "step": 532
+    },
+    {
+      "epoch": 2.1589873417721517,
+      "grad_norm": 0.4150747979257623,
+      "learning_rate": 2.792159532078314e-05,
+      "loss": 0.4615,
+      "step": 533
+    },
+    {
+      "epoch": 2.163037974683544,
+      "grad_norm": 0.3802776526931151,
+      "learning_rate": 2.7869446761557033e-05,
+      "loss": 0.4487,
+      "step": 534
+    },
+    {
+      "epoch": 2.1670886075949367,
+      "grad_norm": 0.39924450303538683,
+      "learning_rate": 2.781723482288857e-05,
+      "loss": 0.4353,
+      "step": 535
+    },
+    {
+      "epoch": 2.1711392405063292,
+      "grad_norm": 0.39594960684502367,
+      "learning_rate": 2.7764959925285517e-05,
+      "loss": 0.4373,
+      "step": 536
+    },
+    {
+      "epoch": 2.1751898734177217,
+      "grad_norm": 0.3890765250840381,
+      "learning_rate": 2.771262248976272e-05,
+      "loss": 0.4624,
+      "step": 537
+    },
+    {
+      "epoch": 2.179240506329114,
+      "grad_norm": 0.374696296640861,
+      "learning_rate": 2.7660222937838677e-05,
+      "loss": 0.4407,
+      "step": 538
+    },
+    {
+      "epoch": 2.1832911392405063,
+      "grad_norm": 0.42168236746081494,
+      "learning_rate": 2.7607761691532186e-05,
+      "loss": 0.4505,
+      "step": 539
+    },
+    {
+      "epoch": 2.187341772151899,
+      "grad_norm": 0.39708322628420933,
+      "learning_rate": 2.7555239173358916e-05,
+      "loss": 0.4512,
+      "step": 540
+    },
+    {
+      "epoch": 2.191392405063291,
+      "grad_norm": 0.42199320605768886,
+      "learning_rate": 2.7502655806328e-05,
+      "loss": 0.4318,
+      "step": 541
+    },
+    {
+      "epoch": 2.1954430379746834,
+      "grad_norm": 0.4185513648379478,
+      "learning_rate": 2.7450012013938648e-05,
+      "loss": 0.4362,
+      "step": 542
+    },
+    {
+      "epoch": 2.199493670886076,
+      "grad_norm": 0.4290627184476055,
+      "learning_rate": 2.739730822017673e-05,
+      "loss": 0.4442,
+      "step": 543
+    },
+    {
+      "epoch": 2.2035443037974685,
+      "grad_norm": 0.4019274205308736,
+      "learning_rate": 2.7344544849511355e-05,
+      "loss": 0.4515,
+      "step": 544
+    },
+    {
+      "epoch": 2.207594936708861,
+      "grad_norm": 0.4109938369756479,
+      "learning_rate": 2.7291722326891456e-05,
+      "loss": 0.4239,
+      "step": 545
+    },
+    {
+      "epoch": 2.211645569620253,
+      "grad_norm": 0.4123309354071297,
+      "learning_rate": 2.723884107774236e-05,
+      "loss": 0.4608,
+      "step": 546
+    },
+    {
+      "epoch": 2.2156962025316456,
+      "grad_norm": 0.4208279209151056,
+      "learning_rate": 2.718590152796239e-05,
+      "loss": 0.4416,
+      "step": 547
+    },
+    {
+      "epoch": 2.219746835443038,
+      "grad_norm": 0.3710728162973027,
+      "learning_rate": 2.71329041039194e-05,
+      "loss": 0.4479,
+      "step": 548
+    },
+    {
+      "epoch": 2.22379746835443,
+      "grad_norm": 0.4598254135594956,
+      "learning_rate": 2.7079849232447357e-05,
+      "loss": 0.4354,
+      "step": 549
+    },
+    {
+      "epoch": 2.2278481012658227,
+      "grad_norm": 0.39672590947517883,
+      "learning_rate": 2.7026737340842895e-05,
+      "loss": 0.4413,
+      "step": 550
+    },
+    {
+      "epoch": 2.231898734177215,
+      "grad_norm": 0.4433791313529699,
+      "learning_rate": 2.697356885686189e-05,
+      "loss": 0.4445,
+      "step": 551
+    },
+    {
+      "epoch": 2.2359493670886077,
+      "grad_norm": 0.5608384449569562,
+      "learning_rate": 2.6920344208716014e-05,
+      "loss": 0.4554,
+      "step": 552
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 0.4298000034864174,
+      "learning_rate": 2.6867063825069252e-05,
+      "loss": 0.4265,
+      "step": 553
+    },
+    {
+      "epoch": 2.2440506329113923,
+      "grad_norm": 0.480915915793418,
+      "learning_rate": 2.6813728135034494e-05,
+      "loss": 0.4367,
+      "step": 554
+    },
+    {
+      "epoch": 2.248101265822785,
+      "grad_norm": 0.4328511144466288,
+      "learning_rate": 2.6760337568170056e-05,
+      "loss": 0.4505,
+      "step": 555
+    },
+    {
+      "epoch": 2.2521518987341773,
+      "grad_norm": 0.4446061978387312,
+      "learning_rate": 2.6706892554476226e-05,
+      "loss": 0.4414,
+      "step": 556
+    },
+    {
+      "epoch": 2.2562025316455694,
+      "grad_norm": 0.4107548130864555,
+      "learning_rate": 2.6653393524391795e-05,
+      "loss": 0.4442,
+      "step": 557
+    },
+    {
+      "epoch": 2.260253164556962,
+      "grad_norm": 0.4536069829576191,
+      "learning_rate": 2.6599840908790592e-05,
+      "loss": 0.4479,
+      "step": 558
+    },
+    {
+      "epoch": 2.2643037974683544,
+      "grad_norm": 0.4089205087353318,
+      "learning_rate": 2.6546235138978028e-05,
+      "loss": 0.4462,
+      "step": 559
+    },
+    {
+      "epoch": 2.268354430379747,
+      "grad_norm": 0.4500614857617038,
+      "learning_rate": 2.6492576646687597e-05,
+      "loss": 0.4567,
+      "step": 560
+    },
+    {
+      "epoch": 2.2724050632911394,
+      "grad_norm": 0.4321337685105182,
+      "learning_rate": 2.6438865864077425e-05,
+      "loss": 0.4344,
+      "step": 561
+    },
+    {
+      "epoch": 2.2764556962025315,
+      "grad_norm": 0.3654577040503411,
+      "learning_rate": 2.6385103223726766e-05,
+      "loss": 0.4527,
+      "step": 562
+    },
+    {
+      "epoch": 2.280506329113924,
+      "grad_norm": 0.45527344180177653,
+      "learning_rate": 2.6331289158632537e-05,
+      "loss": 0.4677,
+      "step": 563
+    },
+    {
+      "epoch": 2.2845569620253166,
+      "grad_norm": 0.416226269779077,
+      "learning_rate": 2.6277424102205817e-05,
+      "loss": 0.4606,
+      "step": 564
+    },
+    {
+      "epoch": 2.2886075949367086,
+      "grad_norm": 0.37715340626939114,
+      "learning_rate": 2.6223508488268374e-05,
+      "loss": 0.4543,
+      "step": 565
+    },
+    {
+      "epoch": 2.292658227848101,
+      "grad_norm": 0.432829848378625,
+      "learning_rate": 2.6169542751049148e-05,
+      "loss": 0.4709,
+      "step": 566
+    },
+    {
+      "epoch": 2.2967088607594937,
+      "grad_norm": 0.3788227200507674,
+      "learning_rate": 2.6115527325180754e-05,
+      "loss": 0.43,
+      "step": 567
+    },
+    {
+      "epoch": 2.300759493670886,
+      "grad_norm": 0.4372765233131344,
+      "learning_rate": 2.606146264569603e-05,
+      "loss": 0.4716,
+      "step": 568
+    },
+    {
+      "epoch": 2.3048101265822787,
+      "grad_norm": 0.39810542291692413,
+      "learning_rate": 2.6007349148024447e-05,
+      "loss": 0.4413,
+      "step": 569
+    },
+    {
+      "epoch": 2.3088607594936708,
+      "grad_norm": 0.37504698756154703,
+      "learning_rate": 2.5953187267988694e-05,
+      "loss": 0.4371,
+      "step": 570
+    },
+    {
+      "epoch": 2.3129113924050633,
+      "grad_norm": 0.39947484177907927,
+      "learning_rate": 2.5898977441801097e-05,
+      "loss": 0.4487,
+      "step": 571
+    },
+    {
+      "epoch": 2.316962025316456,
+      "grad_norm": 0.37909552467574625,
+      "learning_rate": 2.584472010606015e-05,
+      "loss": 0.4389,
+      "step": 572
+    },
+    {
+      "epoch": 2.321012658227848,
+      "grad_norm": 0.4016510663529906,
+      "learning_rate": 2.5790415697746976e-05,
+      "loss": 0.437,
+      "step": 573
+    },
+    {
+      "epoch": 2.3250632911392404,
+      "grad_norm": 0.37517344012310344,
+      "learning_rate": 2.5736064654221808e-05,
+      "loss": 0.4408,
+      "step": 574
+    },
+    {
+      "epoch": 2.329113924050633,
+      "grad_norm": 0.39080818833355796,
+      "learning_rate": 2.568166741322048e-05,
+      "loss": 0.4398,
+      "step": 575
+    },
+    {
+      "epoch": 2.3331645569620254,
+      "grad_norm": 0.4149395927396412,
+      "learning_rate": 2.56272244128509e-05,
+      "loss": 0.4453,
+      "step": 576
+    },
+    {
+      "epoch": 2.337215189873418,
+      "grad_norm": 0.3574957165177111,
+      "learning_rate": 2.55727360915895e-05,
+      "loss": 0.4452,
+      "step": 577
+    },
+    {
+      "epoch": 2.34126582278481,
+      "grad_norm": 0.4187321892433438,
+      "learning_rate": 2.5518202888277734e-05,
+      "loss": 0.4445,
+      "step": 578
+    },
+    {
+      "epoch": 2.3453164556962025,
+      "grad_norm": 0.37853801931381786,
+      "learning_rate": 2.5463625242118523e-05,
+      "loss": 0.4527,
+      "step": 579
+    },
+    {
+      "epoch": 2.349367088607595,
+      "grad_norm": 0.3629573011369568,
+      "learning_rate": 2.5409003592672723e-05,
+      "loss": 0.4566,
+      "step": 580
+    },
+    {
+      "epoch": 2.353417721518987,
+      "grad_norm": 0.4380487354475527,
+      "learning_rate": 2.535433837985559e-05,
+      "loss": 0.456,
+      "step": 581
+    },
+    {
+      "epoch": 2.3574683544303796,
+      "grad_norm": 0.33210758793447237,
+      "learning_rate": 2.529963004393324e-05,
+      "loss": 0.4568,
+      "step": 582
+    },
+    {
+      "epoch": 2.361518987341772,
+      "grad_norm": 0.4020725712740055,
+      "learning_rate": 2.524487902551908e-05,
+      "loss": 0.4423,
+      "step": 583
+    },
+    {
+      "epoch": 2.3655696202531646,
+      "grad_norm": 0.3527621153206325,
+      "learning_rate": 2.519008576557029e-05,
+      "loss": 0.4468,
+      "step": 584
+    },
+    {
+      "epoch": 2.369620253164557,
+      "grad_norm": 0.3504633646763965,
+      "learning_rate": 2.5135250705384254e-05,
+      "loss": 0.4534,
+      "step": 585
+    },
+    {
+      "epoch": 2.3736708860759492,
+      "grad_norm": 0.3256794785828873,
+      "learning_rate": 2.5080374286595007e-05,
+      "loss": 0.4552,
+      "step": 586
+    },
+    {
+      "epoch": 2.3777215189873417,
+      "grad_norm": 0.3514758769800106,
+      "learning_rate": 2.5025456951169677e-05,
+      "loss": 0.4751,
+      "step": 587
+    },
+    {
+      "epoch": 2.3817721518987343,
+      "grad_norm": 0.30738734717785604,
+      "learning_rate": 2.4970499141404942e-05,
+      "loss": 0.4598,
+      "step": 588
+    },
+    {
+      "epoch": 2.3858227848101268,
+      "grad_norm": 0.356386750378579,
+      "learning_rate": 2.491550129992345e-05,
+      "loss": 0.4563,
+      "step": 589
+    },
+    {
+      "epoch": 2.389873417721519,
+      "grad_norm": 0.3559598955480336,
+      "learning_rate": 2.486046386967024e-05,
+      "loss": 0.451,
+      "step": 590
+    },
+    {
+      "epoch": 2.3939240506329114,
+      "grad_norm": 0.367320522631459,
+      "learning_rate": 2.4805387293909214e-05,
+      "loss": 0.4681,
+      "step": 591
+    },
+    {
+      "epoch": 2.397974683544304,
+      "grad_norm": 0.35663248839498696,
+      "learning_rate": 2.4750272016219552e-05,
+      "loss": 0.454,
+      "step": 592
+    },
+    {
+      "epoch": 2.4020253164556964,
+      "grad_norm": 0.359765118074405,
+      "learning_rate": 2.4695118480492114e-05,
+      "loss": 0.4527,
+      "step": 593
+    },
+    {
+      "epoch": 2.4060759493670885,
+      "grad_norm": 0.35862634892517276,
+      "learning_rate": 2.4639927130925898e-05,
+      "loss": 0.444,
+      "step": 594
+    },
+    {
+      "epoch": 2.410126582278481,
+      "grad_norm": 0.36627153456799355,
+      "learning_rate": 2.458469841202444e-05,
+      "loss": 0.4549,
+      "step": 595
+    },
+    {
+      "epoch": 2.4141772151898735,
+      "grad_norm": 0.3434816868406632,
+      "learning_rate": 2.452943276859226e-05,
+      "loss": 0.452,
+      "step": 596
+    },
+    {
+      "epoch": 2.418227848101266,
+      "grad_norm": 0.37960095771942315,
+      "learning_rate": 2.447413064573125e-05,
+      "loss": 0.4418,
+      "step": 597
+    },
+    {
+      "epoch": 2.422278481012658,
+      "grad_norm": 0.34198670636936307,
+      "learning_rate": 2.4418792488837095e-05,
+      "loss": 0.4568,
+      "step": 598
+    },
+    {
+      "epoch": 2.4263291139240506,
+      "grad_norm": 0.3961331206814585,
+      "learning_rate": 2.4363418743595713e-05,
+      "loss": 0.4264,
+      "step": 599
+    },
+    {
+      "epoch": 2.430379746835443,
+      "grad_norm": 0.3344072411935779,
+      "learning_rate": 2.430800985597963e-05,
+      "loss": 0.4564,
+      "step": 600
+    },
+    {
+      "epoch": 2.4344303797468356,
+      "grad_norm": 0.3979318553936251,
+      "learning_rate": 2.4252566272244415e-05,
+      "loss": 0.4738,
+      "step": 601
+    },
+    {
+      "epoch": 2.4384810126582277,
+      "grad_norm": 0.38553731677340203,
+      "learning_rate": 2.4197088438925063e-05,
+      "loss": 0.4414,
+      "step": 602
+    },
+    {
+      "epoch": 2.44253164556962,
+      "grad_norm": 0.42168747155785896,
+      "learning_rate": 2.4141576802832417e-05,
+      "loss": 0.4483,
+      "step": 603
+    },
+    {
+      "epoch": 2.4465822784810127,
+      "grad_norm": 0.3764003286130598,
+      "learning_rate": 2.408603181104957e-05,
+      "loss": 0.4592,
+      "step": 604
+    },
+    {
+      "epoch": 2.4506329113924052,
+      "grad_norm": 0.4111087654637082,
+      "learning_rate": 2.4030453910928245e-05,
+      "loss": 0.4679,
+      "step": 605
+    },
+    {
+      "epoch": 2.4546835443037973,
+      "grad_norm": 0.4084148244199903,
+      "learning_rate": 2.397484355008521e-05,
+      "loss": 0.4635,
+      "step": 606
+    },
+    {
+      "epoch": 2.45873417721519,
+      "grad_norm": 0.399196330315307,
+      "learning_rate": 2.3919201176398662e-05,
+      "loss": 0.4527,
+      "step": 607
+    },
+    {
+      "epoch": 2.4627848101265823,
+      "grad_norm": 0.4124725406472348,
+      "learning_rate": 2.3863527238004633e-05,
+      "loss": 0.4398,
+      "step": 608
+    },
+    {
+      "epoch": 2.466835443037975,
+      "grad_norm": 0.36203890268966776,
+      "learning_rate": 2.380782218329337e-05,
+      "loss": 0.461,
+      "step": 609
+    },
+    {
+      "epoch": 2.470886075949367,
+      "grad_norm": 0.4056408225265219,
+      "learning_rate": 2.3752086460905725e-05,
+      "loss": 0.4299,
+      "step": 610
+    },
+    {
+      "epoch": 2.4749367088607594,
+      "grad_norm": 0.37571033631829864,
+      "learning_rate": 2.3696320519729544e-05,
+      "loss": 0.4603,
+      "step": 611
+    },
+    {
+      "epoch": 2.478987341772152,
+      "grad_norm": 0.3863297874679425,
+      "learning_rate": 2.3640524808896045e-05,
+      "loss": 0.4428,
+      "step": 612
+    },
+    {
+      "epoch": 2.4830379746835445,
+      "grad_norm": 0.38961760684279795,
+      "learning_rate": 2.3584699777776222e-05,
+      "loss": 0.4422,
+      "step": 613
+    },
+    {
+      "epoch": 2.4870886075949366,
+      "grad_norm": 0.39821593510887626,
+      "learning_rate": 2.3528845875977195e-05,
+      "loss": 0.4398,
+      "step": 614
+    },
+    {
+      "epoch": 2.491139240506329,
+      "grad_norm": 0.37130518213268043,
+      "learning_rate": 2.3472963553338614e-05,
+      "loss": 0.4413,
+      "step": 615
+    },
+    {
+      "epoch": 2.4951898734177216,
+      "grad_norm": 0.3791520672551331,
+      "learning_rate": 2.341705325992901e-05,
+      "loss": 0.4524,
+      "step": 616
+    },
+    {
+      "epoch": 2.499240506329114,
+      "grad_norm": 0.39864216078985476,
+      "learning_rate": 2.336111544604222e-05,
+      "loss": 0.4383,
+      "step": 617
+    },
+    {
+      "epoch": 2.503291139240506,
+      "grad_norm": 0.3742410871599069,
+      "learning_rate": 2.33051505621937e-05,
+      "loss": 0.4389,
+      "step": 618
+    },
+    {
+      "epoch": 2.5073417721518987,
+      "grad_norm": 0.3522862389430058,
+      "learning_rate": 2.324915905911693e-05,
+      "loss": 0.4575,
+      "step": 619
+    },
+    {
+      "epoch": 2.511392405063291,
+      "grad_norm": 0.39883004815966816,
+      "learning_rate": 2.319314138775977e-05,
+      "loss": 0.4474,
+      "step": 620
+    },
+    {
+      "epoch": 2.5154430379746833,
+      "grad_norm": 0.38070745625429087,
+      "learning_rate": 2.3137097999280856e-05,
+      "loss": 0.4535,
+      "step": 621
+    },
+    {
+      "epoch": 2.519493670886076,
+      "grad_norm": 0.3961767599269952,
+      "learning_rate": 2.308102934504593e-05,
+      "loss": 0.447,
+      "step": 622
+    },
+    {
+      "epoch": 2.5235443037974683,
+      "grad_norm": 0.4416348644500427,
+      "learning_rate": 2.3024935876624222e-05,
+      "loss": 0.4405,
+      "step": 623
+    },
+    {
+      "epoch": 2.527594936708861,
+      "grad_norm": 0.406312934858738,
+      "learning_rate": 2.2968818045784813e-05,
+      "loss": 0.4504,
+      "step": 624
+    },
+    {
+      "epoch": 2.5316455696202533,
+      "grad_norm": 0.4153124958959821,
+      "learning_rate": 2.2912676304493006e-05,
+      "loss": 0.4516,
+      "step": 625
+    },
+    {
+      "epoch": 2.5356962025316454,
+      "grad_norm": 0.4391079399859738,
+      "learning_rate": 2.2856511104906668e-05,
+      "loss": 0.4499,
+      "step": 626
+    },
+    {
+      "epoch": 2.539746835443038,
+      "grad_norm": 0.39529845135659575,
+      "learning_rate": 2.2800322899372586e-05,
+      "loss": 0.4326,
+      "step": 627
+    },
+    {
+      "epoch": 2.5437974683544304,
+      "grad_norm": 0.4481371230177085,
+      "learning_rate": 2.2744112140422844e-05,
+      "loss": 0.4475,
+      "step": 628
+    },
+    {
+      "epoch": 2.547848101265823,
+      "grad_norm": 0.39161417434495555,
+      "learning_rate": 2.2687879280771177e-05,
+      "loss": 0.455,
+      "step": 629
+    },
+    {
+      "epoch": 2.5518987341772155,
+      "grad_norm": 0.4450336524237636,
+      "learning_rate": 2.26316247733093e-05,
+      "loss": 0.4404,
+      "step": 630
+    },
+    {
+      "epoch": 2.5559493670886075,
+      "grad_norm": 0.36420932787322535,
+      "learning_rate": 2.257534907110328e-05,
+      "loss": 0.4503,
+      "step": 631
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 0.394863391271882,
+      "learning_rate": 2.2519052627389882e-05,
+      "loss": 0.4444,
+      "step": 632
+    },
+    {
+      "epoch": 2.5640506329113926,
+      "grad_norm": 0.45018083508621276,
+      "learning_rate": 2.246273589557294e-05,
+      "loss": 0.4258,
+      "step": 633
+    },
+    {
+      "epoch": 2.5681012658227846,
+      "grad_norm": 0.39540337792205765,
+      "learning_rate": 2.240639932921966e-05,
+      "loss": 0.4359,
+      "step": 634
+    },
+    {
+      "epoch": 2.572151898734177,
+      "grad_norm": 0.4342557931715197,
+      "learning_rate": 2.2350043382056995e-05,
+      "loss": 0.4479,
+      "step": 635
+    },
+    {
+      "epoch": 2.5762025316455697,
+      "grad_norm": 0.4184562502729378,
+      "learning_rate": 2.2293668507968015e-05,
+      "loss": 0.4568,
+      "step": 636
+    },
+    {
+      "epoch": 2.580253164556962,
+      "grad_norm": 0.433691680194859,
+      "learning_rate": 2.2237275160988186e-05,
+      "loss": 0.4483,
+      "step": 637
+    },
+    {
+      "epoch": 2.5843037974683547,
+      "grad_norm": 0.4690637724291503,
+      "learning_rate": 2.2180863795301787e-05,
+      "loss": 0.4393,
+      "step": 638
+    },
+    {
+      "epoch": 2.5883544303797468,
+      "grad_norm": 0.4341457523661746,
+      "learning_rate": 2.212443486523819e-05,
+      "loss": 0.4169,
+      "step": 639
+    },
+    {
+      "epoch": 2.5924050632911393,
+      "grad_norm": 0.41745804738460607,
+      "learning_rate": 2.2067988825268243e-05,
+      "loss": 0.4379,
+      "step": 640
+    },
+    {
+      "epoch": 2.596455696202532,
+      "grad_norm": 0.40094525842498524,
+      "learning_rate": 2.2011526130000596e-05,
+      "loss": 0.4604,
+      "step": 641
+    },
+    {
+      "epoch": 2.600506329113924,
+      "grad_norm": 0.36829785110112084,
+      "learning_rate": 2.1955047234178038e-05,
+      "loss": 0.4385,
+      "step": 642
+    },
+    {
+      "epoch": 2.6045569620253164,
+      "grad_norm": 0.404878414273811,
+      "learning_rate": 2.1898552592673825e-05,
+      "loss": 0.4266,
+      "step": 643
+    },
+    {
+      "epoch": 2.608607594936709,
+      "grad_norm": 0.39580229226704877,
+      "learning_rate": 2.184204266048803e-05,
+      "loss": 0.446,
+      "step": 644
+    },
+    {
+      "epoch": 2.6126582278481014,
+      "grad_norm": 0.3905385453234181,
+      "learning_rate": 2.1785517892743887e-05,
+      "loss": 0.4554,
+      "step": 645
+    },
+    {
+      "epoch": 2.616708860759494,
+      "grad_norm": 0.4464998813767474,
+      "learning_rate": 2.17289787446841e-05,
+      "loss": 0.4385,
+      "step": 646
+    },
+    {
+      "epoch": 2.620759493670886,
+      "grad_norm": 0.35602829881828946,
+      "learning_rate": 2.1672425671667198e-05,
+      "loss": 0.4505,
+      "step": 647
+    },
+    {
+      "epoch": 2.6248101265822785,
+      "grad_norm": 0.36491017062855086,
+      "learning_rate": 2.161585912916385e-05,
+      "loss": 0.439,
+      "step": 648
+    },
+    {
+      "epoch": 2.628860759493671,
+      "grad_norm": 0.3837747288547875,
+      "learning_rate": 2.1559279572753214e-05,
+      "loss": 0.4509,
+      "step": 649
+    },
+    {
+      "epoch": 2.632911392405063,
+      "grad_norm": 0.3679964911280036,
+      "learning_rate": 2.1502687458119268e-05,
+      "loss": 0.4437,
+      "step": 650
+    },
+    {
+      "epoch": 2.6369620253164556,
+      "grad_norm": 0.44260824353138345,
+      "learning_rate": 2.1446083241047116e-05,
+      "loss": 0.4224,
+      "step": 651
+    },
+    {
+      "epoch": 2.641012658227848,
+      "grad_norm": 0.36682096650586304,
+      "learning_rate": 2.1389467377419333e-05,
+      "loss": 0.4639,
+      "step": 652
+    },
+    {
+      "epoch": 2.6450632911392407,
+      "grad_norm": 0.4384285343203088,
+      "learning_rate": 2.133284032321232e-05,
+      "loss": 0.4463,
+      "step": 653
+    },
+    {
+      "epoch": 2.649113924050633,
+      "grad_norm": 0.3709946203060811,
+      "learning_rate": 2.1276202534492566e-05,
+      "loss": 0.4208,
+      "step": 654
+    },
+    {
+      "epoch": 2.6531645569620252,
+      "grad_norm": 0.44066284878274914,
+      "learning_rate": 2.121955446741306e-05,
+      "loss": 0.4332,
+      "step": 655
+    },
+    {
+      "epoch": 2.6572151898734178,
+      "grad_norm": 0.3533743480491943,
+      "learning_rate": 2.1162896578209517e-05,
+      "loss": 0.4612,
+      "step": 656
+    },
+    {
+      "epoch": 2.6612658227848103,
+      "grad_norm": 0.381661482596036,
+      "learning_rate": 2.1106229323196813e-05,
+      "loss": 0.455,
+      "step": 657
+    },
+    {
+      "epoch": 2.6653164556962023,
+      "grad_norm": 0.35738992665595776,
+      "learning_rate": 2.1049553158765214e-05,
+      "loss": 0.4367,
+      "step": 658
+    },
+    {
+      "epoch": 2.669367088607595,
+      "grad_norm": 0.3887645710048796,
+      "learning_rate": 2.0992868541376764e-05,
+      "loss": 0.4404,
+      "step": 659
+    },
+    {
+      "epoch": 2.6734177215189874,
+      "grad_norm": 0.37507821270673886,
+      "learning_rate": 2.093617592756158e-05,
+      "loss": 0.4643,
+      "step": 660
+    },
+    {
+      "epoch": 2.67746835443038,
+      "grad_norm": 0.40929353216275904,
+      "learning_rate": 2.0879475773914167e-05,
+      "loss": 0.4422,
+      "step": 661
+    },
+    {
+      "epoch": 2.6815189873417724,
+      "grad_norm": 0.3844316794752405,
+      "learning_rate": 2.082276853708978e-05,
+      "loss": 0.4397,
+      "step": 662
+    },
+    {
+      "epoch": 2.6855696202531645,
+      "grad_norm": 0.35977160084426585,
+      "learning_rate": 2.076605467380071e-05,
+      "loss": 0.4427,
+      "step": 663
+    },
+    {
+      "epoch": 2.689620253164557,
+      "grad_norm": 0.3461849340546685,
+      "learning_rate": 2.0709334640812613e-05,
+      "loss": 0.4417,
+      "step": 664
+    },
+    {
+      "epoch": 2.6936708860759495,
+      "grad_norm": 0.37924830546520405,
+      "learning_rate": 2.0652608894940824e-05,
+      "loss": 0.4416,
+      "step": 665
+    },
+    {
+      "epoch": 2.6977215189873416,
+      "grad_norm": 0.38707845677326513,
+      "learning_rate": 2.0595877893046722e-05,
+      "loss": 0.4441,
+      "step": 666
+    },
+    {
+      "epoch": 2.701772151898734,
+      "grad_norm": 0.3587571411153537,
+      "learning_rate": 2.0539142092033985e-05,
+      "loss": 0.4435,
+      "step": 667
+    },
+    {
+      "epoch": 2.7058227848101266,
+      "grad_norm": 0.3458977696925745,
+      "learning_rate": 2.048240194884496e-05,
+      "loss": 0.4448,
+      "step": 668
+    },
+    {
+      "epoch": 2.709873417721519,
+      "grad_norm": 0.3554944010630844,
+      "learning_rate": 2.042565792045695e-05,
+      "loss": 0.4739,
+      "step": 669
+    },
+    {
+      "epoch": 2.7139240506329116,
+      "grad_norm": 0.3757754405661183,
+      "learning_rate": 2.036891046387857e-05,
+      "loss": 0.4388,
+      "step": 670
+    },
+    {
+      "epoch": 2.7179746835443037,
+      "grad_norm": 0.3871295953332396,
+      "learning_rate": 2.0312160036146036e-05,
+      "loss": 0.4458,
+      "step": 671
+    },
+    {
+      "epoch": 2.7220253164556962,
+      "grad_norm": 0.3648200128746818,
+      "learning_rate": 2.025540709431948e-05,
+      "loss": 0.4458,
+      "step": 672
+    },
+    {
+      "epoch": 2.7260759493670887,
+      "grad_norm": 0.35529738221678575,
+      "learning_rate": 2.0198652095479298e-05,
+      "loss": 0.4477,
+      "step": 673
+    },
+    {
+      "epoch": 2.730126582278481,
+      "grad_norm": 0.34478485946486015,
+      "learning_rate": 2.014189549672245e-05,
+      "loss": 0.4487,
+      "step": 674
+    },
+    {
+      "epoch": 2.7341772151898733,
+      "grad_norm": 0.3594358438341782,
+      "learning_rate": 2.0085137755158776e-05,
+      "loss": 0.4349,
+      "step": 675
+    },
+    {
+      "epoch": 2.738227848101266,
+      "grad_norm": 0.31119279044809206,
+      "learning_rate": 2.0028379327907327e-05,
+      "loss": 0.4496,
+      "step": 676
+    },
+    {
+      "epoch": 2.7422784810126584,
+      "grad_norm": 0.34229033214132226,
+      "learning_rate": 1.9971620672092676e-05,
+      "loss": 0.45,
+      "step": 677
+    },
+    {
+      "epoch": 2.746329113924051,
+      "grad_norm": 0.35158473134765017,
+      "learning_rate": 1.991486224484123e-05,
+      "loss": 0.4549,
+      "step": 678
+    },
+    {
+      "epoch": 2.750379746835443,
+      "grad_norm": 0.327912707640582,
+      "learning_rate": 1.985810450327756e-05,
+      "loss": 0.4609,
+      "step": 679
+    },
+    {
+      "epoch": 2.7544303797468355,
+      "grad_norm": 0.34383867860355255,
+      "learning_rate": 1.9801347904520706e-05,
+      "loss": 0.4628,
+      "step": 680
+    },
+    {
+      "epoch": 2.758481012658228,
+      "grad_norm": 0.35264455758605684,
+      "learning_rate": 1.974459290568053e-05,
+      "loss": 0.4291,
+      "step": 681
+    },
+    {
+      "epoch": 2.76253164556962,
+      "grad_norm": 0.3686380367848887,
+      "learning_rate": 1.968783996385397e-05,
+      "loss": 0.4418,
+      "step": 682
+    },
+    {
+      "epoch": 2.7665822784810126,
+      "grad_norm": 0.34686346584333444,
+      "learning_rate": 1.963108953612143e-05,
+      "loss": 0.4401,
+      "step": 683
+    },
+    {
+      "epoch": 2.770632911392405,
+      "grad_norm": 0.3625487149985722,
+      "learning_rate": 1.9574342079543056e-05,
+      "loss": 0.4426,
+      "step": 684
+    },
+    {
+      "epoch": 2.7746835443037976,
+      "grad_norm": 0.35256505023800677,
+      "learning_rate": 1.9517598051155046e-05,
+      "loss": 0.4573,
+      "step": 685
+    },
+    {
+      "epoch": 2.77873417721519,
+      "grad_norm": 0.34469756234623194,
+      "learning_rate": 1.9460857907966025e-05,
+      "loss": 0.4477,
+      "step": 686
+    },
+    {
+      "epoch": 2.782784810126582,
+      "grad_norm": 0.3492838795594039,
+      "learning_rate": 1.9404122106953285e-05,
+      "loss": 0.4391,
+      "step": 687
+    },
+    {
+      "epoch": 2.7868354430379747,
+      "grad_norm": 0.35605390893963257,
+      "learning_rate": 1.9347391105059176e-05,
+      "loss": 0.4598,
+      "step": 688
+    },
+    {
+      "epoch": 2.790886075949367,
+      "grad_norm": 0.3376884661560383,
+      "learning_rate": 1.92906653591874e-05,
+      "loss": 0.4452,
+      "step": 689
+    },
+    {
+      "epoch": 2.7949367088607593,
+      "grad_norm": 0.33055598937867736,
+      "learning_rate": 1.9233945326199295e-05,
+      "loss": 0.4562,
+      "step": 690
+    },
+    {
+      "epoch": 2.798987341772152,
+      "grad_norm": 0.3464633567357682,
+      "learning_rate": 1.917723146291022e-05,
+      "loss": 0.4589,
+      "step": 691
+    },
+    {
+      "epoch": 2.8030379746835443,
+      "grad_norm": 0.37797864050878277,
+      "learning_rate": 1.912052422608584e-05,
+      "loss": 0.4489,
+      "step": 692
+    },
+    {
+      "epoch": 2.807088607594937,
+      "grad_norm": 0.32572119431489766,
+      "learning_rate": 1.9063824072438428e-05,
+      "loss": 0.4412,
+      "step": 693
+    },
+    {
+      "epoch": 2.8111392405063294,
+      "grad_norm": 0.3734375503792976,
+      "learning_rate": 1.9007131458623246e-05,
+      "loss": 0.433,
+      "step": 694
+    },
+    {
+      "epoch": 2.8151898734177214,
+      "grad_norm": 0.3535702205679094,
+      "learning_rate": 1.895044684123479e-05,
+      "loss": 0.4574,
+      "step": 695
+    },
+    {
+      "epoch": 2.819240506329114,
+      "grad_norm": 0.38204703712219745,
+      "learning_rate": 1.8893770676803194e-05,
+      "loss": 0.4568,
+      "step": 696
+    },
+    {
+      "epoch": 2.8232911392405065,
+      "grad_norm": 0.3556768018773698,
+      "learning_rate": 1.8837103421790486e-05,
+      "loss": 0.4631,
+      "step": 697
+    },
+    {
+      "epoch": 2.8273417721518985,
+      "grad_norm": 0.3364520203887203,
+      "learning_rate": 1.8780445532586952e-05,
+      "loss": 0.4294,
+      "step": 698
+    },
+    {
+      "epoch": 2.831392405063291,
+      "grad_norm": 0.35603975913928176,
+      "learning_rate": 1.872379746550743e-05,
+      "loss": 0.4482,
+      "step": 699
+    },
+    {
+      "epoch": 2.8354430379746836,
+      "grad_norm": 0.37486473689271815,
+      "learning_rate": 1.866715967678769e-05,
+      "loss": 0.4485,
+      "step": 700
+    },
+    {
+      "epoch": 2.839493670886076,
+      "grad_norm": 0.3617338379441558,
+      "learning_rate": 1.861053262258067e-05,
+      "loss": 0.4491,
+      "step": 701
+    },
+    {
+      "epoch": 2.8435443037974686,
+      "grad_norm": 0.36133010359982376,
+      "learning_rate": 1.8553916758952897e-05,
+      "loss": 0.4398,
+      "step": 702
+    },
+    {
+      "epoch": 2.8475949367088607,
+      "grad_norm": 0.3587284989414955,
+      "learning_rate": 1.8497312541880735e-05,
+      "loss": 0.4304,
+      "step": 703
+    },
+    {
+      "epoch": 2.851645569620253,
+      "grad_norm": 0.4105346496164975,
+      "learning_rate": 1.8440720427246786e-05,
+      "loss": 0.4375,
+      "step": 704
+    },
+    {
+      "epoch": 2.8556962025316457,
+      "grad_norm": 0.30967055790479037,
+      "learning_rate": 1.8384140870836157e-05,
+      "loss": 0.4453,
+      "step": 705
+    },
+    {
+      "epoch": 2.8597468354430378,
+      "grad_norm": 0.4059818085063748,
+      "learning_rate": 1.8327574328332806e-05,
+      "loss": 0.4608,
+      "step": 706
+    },
+    {
+      "epoch": 2.8637974683544303,
+      "grad_norm": 0.35997434637769893,
+      "learning_rate": 1.8271021255315906e-05,
+      "loss": 0.4489,
+      "step": 707
+    },
+    {
+      "epoch": 2.867848101265823,
+      "grad_norm": 0.3392723484461712,
+      "learning_rate": 1.8214482107256117e-05,
+      "loss": 0.4477,
+      "step": 708
+    },
+    {
+      "epoch": 2.8718987341772153,
+      "grad_norm": 0.3779647450838903,
+      "learning_rate": 1.8157957339511968e-05,
+      "loss": 0.4306,
+      "step": 709
+    },
+    {
+      "epoch": 2.875949367088608,
+      "grad_norm": 0.3456113908853255,
+      "learning_rate": 1.8101447407326182e-05,
+      "loss": 0.4591,
+      "step": 710
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 0.3394191115891613,
+      "learning_rate": 1.8044952765821966e-05,
+      "loss": 0.4545,
+      "step": 711
+    },
+    {
+      "epoch": 2.8840506329113924,
+      "grad_norm": 0.3605152891227635,
+      "learning_rate": 1.7988473869999407e-05,
+      "loss": 0.4737,
+      "step": 712
+    },
+    {
+      "epoch": 2.888101265822785,
+      "grad_norm": 0.3732677371744172,
+      "learning_rate": 1.7932011174731764e-05,
+      "loss": 0.4454,
+      "step": 713
+    },
+    {
+      "epoch": 2.892151898734177,
+      "grad_norm": 0.35003861244196904,
+      "learning_rate": 1.7875565134761817e-05,
+      "loss": 0.444,
+      "step": 714
+    },
+    {
+      "epoch": 2.8962025316455695,
+      "grad_norm": 0.3511044837748538,
+      "learning_rate": 1.7819136204698226e-05,
+      "loss": 0.451,
+      "step": 715
+    },
+    {
+      "epoch": 2.900253164556962,
+      "grad_norm": 0.3682783410596163,
+      "learning_rate": 1.776272483901182e-05,
+      "loss": 0.468,
+      "step": 716
+    },
+    {
+      "epoch": 2.9043037974683545,
+      "grad_norm": 0.3487168266398176,
+      "learning_rate": 1.7706331492031995e-05,
+      "loss": 0.4434,
+      "step": 717
+    },
+    {
+      "epoch": 2.908354430379747,
+      "grad_norm": 0.34378492760879165,
+      "learning_rate": 1.764995661794301e-05,
+      "loss": 0.4559,
+      "step": 718
+    },
+    {
+      "epoch": 2.912405063291139,
+      "grad_norm": 0.35240528602618676,
+      "learning_rate": 1.759360067078035e-05,
+      "loss": 0.4293,
+      "step": 719
+    },
+    {
+      "epoch": 2.9164556962025316,
+      "grad_norm": 0.3438055427303087,
+      "learning_rate": 1.7537264104427064e-05,
+      "loss": 0.4416,
+      "step": 720
+    },
+    {
+      "epoch": 2.920506329113924,
+      "grad_norm": 0.3543154987834358,
+      "learning_rate": 1.748094737261012e-05,
+      "loss": 0.4503,
+      "step": 721
+    },
+    {
+      "epoch": 2.9245569620253162,
+      "grad_norm": 0.3679482871591542,
+      "learning_rate": 1.7424650928896726e-05,
+      "loss": 0.4381,
+      "step": 722
+    },
+    {
+      "epoch": 2.9286075949367087,
+      "grad_norm": 0.34039471516427894,
+      "learning_rate": 1.7368375226690712e-05,
+      "loss": 0.4354,
+      "step": 723
+    },
+    {
+      "epoch": 2.9326582278481013,
+      "grad_norm": 0.3440535165337069,
+      "learning_rate": 1.731212071922883e-05,
+      "loss": 0.4336,
+      "step": 724
+    },
+    {
+      "epoch": 2.9367088607594938,
+      "grad_norm": 0.3644799472866355,
+      "learning_rate": 1.7255887859577156e-05,
+      "loss": 0.4514,
+      "step": 725
+    },
+    {
+      "epoch": 2.9407594936708863,
+      "grad_norm": 0.32991918355639815,
+      "learning_rate": 1.7199677100627427e-05,
+      "loss": 0.453,
+      "step": 726
+    },
+    {
+      "epoch": 2.9448101265822784,
+      "grad_norm": 0.38450809287163057,
+      "learning_rate": 1.7143488895093343e-05,
+      "loss": 0.4524,
+      "step": 727
+    },
+    {
+      "epoch": 2.948860759493671,
+      "grad_norm": 0.38107702307438757,
+      "learning_rate": 1.7087323695506994e-05,
+      "loss": 0.4556,
+      "step": 728
+    },
+    {
+      "epoch": 2.9529113924050634,
+      "grad_norm": 0.339670253149443,
+      "learning_rate": 1.7031181954215194e-05,
+      "loss": 0.4436,
+      "step": 729
+    },
+    {
+      "epoch": 2.9569620253164555,
+      "grad_norm": 0.3697444770511359,
+      "learning_rate": 1.6975064123375788e-05,
+      "loss": 0.4195,
+      "step": 730
+    },
+    {
+      "epoch": 2.961012658227848,
+      "grad_norm": 0.39502166597242794,
+      "learning_rate": 1.6918970654954084e-05,
+      "loss": 0.4334,
+      "step": 731
+    },
+    {
+      "epoch": 2.9650632911392405,
+      "grad_norm": 0.33486304770768743,
+      "learning_rate": 1.686290200071915e-05,
+      "loss": 0.4659,
+      "step": 732
+    },
+    {
+      "epoch": 2.969113924050633,
+      "grad_norm": 0.3455726787814247,
+      "learning_rate": 1.6806858612240234e-05,
+      "loss": 0.4503,
+      "step": 733
+    },
+    {
+      "epoch": 2.9731645569620255,
+      "grad_norm": 0.3948977376722803,
+      "learning_rate": 1.6750840940883078e-05,
+      "loss": 0.4548,
+      "step": 734
+    },
+    {
+      "epoch": 2.9772151898734176,
+      "grad_norm": 0.36386483110531614,
+      "learning_rate": 1.6694849437806305e-05,
+      "loss": 0.4244,
+      "step": 735
+    },
+    {
+      "epoch": 2.98126582278481,
+      "grad_norm": 0.39724040753219797,
+      "learning_rate": 1.663888455395778e-05,
+      "loss": 0.441,
+      "step": 736
+    },
+    {
+      "epoch": 2.9853164556962026,
+      "grad_norm": 0.3489948559365254,
+      "learning_rate": 1.6582946740070995e-05,
+      "loss": 0.4458,
+      "step": 737
+    },
+    {
+      "epoch": 2.9893670886075947,
+      "grad_norm": 0.37126464487316385,
+      "learning_rate": 1.6527036446661396e-05,
+      "loss": 0.4598,
+      "step": 738
+    },
+    {
+      "epoch": 2.993417721518987,
+      "grad_norm": 0.3186178633462075,
+      "learning_rate": 1.6471154124022818e-05,
+      "loss": 0.4413,
+      "step": 739
+    },
+    {
+      "epoch": 2.9974683544303797,
+      "grad_norm": 0.3373870177943284,
+      "learning_rate": 1.6415300222223788e-05,
+      "loss": 0.4601,
+      "step": 740
+    },
+    {
+      "epoch": 3.0015189873417722,
+      "grad_norm": 0.3493656167371995,
+      "learning_rate": 1.6359475191103958e-05,
+      "loss": 0.4161,
+      "step": 741
+    },
+    {
+      "epoch": 3.0055696202531648,
+      "grad_norm": 0.3716227515404071,
+      "learning_rate": 1.6303679480270466e-05,
+      "loss": 0.379,
+      "step": 742
+    },
+    {
+      "epoch": 3.009620253164557,
+      "grad_norm": 0.47456618549032914,
+      "learning_rate": 1.624791353909428e-05,
+      "loss": 0.359,
+      "step": 743
+    },
+    {
+      "epoch": 3.0136708860759494,
+      "grad_norm": 0.529733358140152,
+      "learning_rate": 1.619217781670663e-05,
+      "loss": 0.3544,
+      "step": 744
+    },
+    {
+      "epoch": 3.017721518987342,
+      "grad_norm": 0.46710057577995917,
+      "learning_rate": 1.6136472761995373e-05,
+      "loss": 0.3854,
+      "step": 745
+    },
+    {
+      "epoch": 3.0217721518987344,
+      "grad_norm": 0.50978456603637,
+      "learning_rate": 1.608079882360134e-05,
+      "loss": 0.3742,
+      "step": 746
+    },
+    {
+      "epoch": 3.0258227848101265,
+      "grad_norm": 0.4186653380745891,
+      "learning_rate": 1.60251564499148e-05,
+      "loss": 0.3756,
+      "step": 747
+    },
+    {
+      "epoch": 3.029873417721519,
+      "grad_norm": 0.4256991484130199,
+      "learning_rate": 1.596954608907176e-05,
+      "loss": 0.3922,
+      "step": 748
+    },
+    {
+      "epoch": 3.0339240506329115,
+      "grad_norm": 0.49203380306301847,
+      "learning_rate": 1.591396818895043e-05,
+      "loss": 0.3729,
+      "step": 749
+    },
+    {
+      "epoch": 3.037974683544304,
+      "grad_norm": 0.3892184984141726,
+      "learning_rate": 1.585842319716759e-05,
+      "loss": 0.3646,
+      "step": 750
+    },
+    {
+      "epoch": 3.042025316455696,
+      "grad_norm": 0.3797025453333974,
+      "learning_rate": 1.5802911561074944e-05,
+      "loss": 0.3544,
+      "step": 751
+    },
+    {
+      "epoch": 3.0460759493670886,
+      "grad_norm": 0.4347131597370357,
+      "learning_rate": 1.5747433727755595e-05,
+      "loss": 0.4036,
+      "step": 752
+    },
+    {
+      "epoch": 3.050126582278481,
+      "grad_norm": 0.37475454085245796,
+      "learning_rate": 1.5691990144020376e-05,
+      "loss": 0.3644,
+      "step": 753
+    },
+    {
+      "epoch": 3.0541772151898736,
+      "grad_norm": 0.4649895451447964,
+      "learning_rate": 1.5636581256404297e-05,
+      "loss": 0.3634,
+      "step": 754
+    },
+    {
+      "epoch": 3.0582278481012657,
+      "grad_norm": 0.38023973465292854,
+      "learning_rate": 1.558120751116291e-05,
+      "loss": 0.3656,
+      "step": 755
+    },
+    {
+      "epoch": 3.062278481012658,
+      "grad_norm": 0.4354911921801458,
+      "learning_rate": 1.552586935426876e-05,
+      "loss": 0.3484,
+      "step": 756
+    },
+    {
+      "epoch": 3.0663291139240507,
+      "grad_norm": 0.36269749032790716,
+      "learning_rate": 1.547056723140774e-05,
+      "loss": 0.3768,
+      "step": 757
+    },
+    {
+      "epoch": 3.0703797468354432,
+      "grad_norm": 0.39333819267228975,
+      "learning_rate": 1.5415301587975565e-05,
+      "loss": 0.3772,
+      "step": 758
+    },
+    {
+      "epoch": 3.0744303797468353,
+      "grad_norm": 0.3448339914350114,
+      "learning_rate": 1.536007286907411e-05,
+      "loss": 0.3793,
+      "step": 759
+    },
+    {
+      "epoch": 3.078481012658228,
+      "grad_norm": 0.33346730210234676,
+      "learning_rate": 1.5304881519507896e-05,
+      "loss": 0.3603,
+      "step": 760
+    },
+    {
+      "epoch": 3.0825316455696203,
+      "grad_norm": 0.371993659744108,
+      "learning_rate": 1.5249727983780453e-05,
+      "loss": 0.3773,
+      "step": 761
+    },
+    {
+      "epoch": 3.086582278481013,
+      "grad_norm": 0.3385441982582641,
+      "learning_rate": 1.5194612706090786e-05,
+      "loss": 0.3767,
+      "step": 762
+    },
+    {
+      "epoch": 3.090632911392405,
+      "grad_norm": 0.37735707173697935,
+      "learning_rate": 1.5139536130329771e-05,
+      "loss": 0.3675,
+      "step": 763
+    },
+    {
+      "epoch": 3.0946835443037974,
+      "grad_norm": 0.3698058622343602,
+      "learning_rate": 1.508449870007656e-05,
+      "loss": 0.3594,
+      "step": 764
+    },
+    {
+      "epoch": 3.09873417721519,
+      "grad_norm": 0.34407746231643876,
+      "learning_rate": 1.5029500858595056e-05,
+      "loss": 0.3774,
+      "step": 765
+    },
+    {
+      "epoch": 3.1027848101265825,
+      "grad_norm": 0.35593367973706763,
+      "learning_rate": 1.4974543048830328e-05,
+      "loss": 0.3732,
+      "step": 766
+    },
+    {
+      "epoch": 3.1068354430379745,
+      "grad_norm": 0.33790494835207735,
+      "learning_rate": 1.4919625713405e-05,
+      "loss": 0.3622,
+      "step": 767
+    },
+    {
+      "epoch": 3.110886075949367,
+      "grad_norm": 0.3343063478565219,
+      "learning_rate": 1.4864749294615756e-05,
+      "loss": 0.3609,
+      "step": 768
+    },
+    {
+      "epoch": 3.1149367088607596,
+      "grad_norm": 0.33285982678168924,
+      "learning_rate": 1.4809914234429716e-05,
+      "loss": 0.3401,
+      "step": 769
+    },
+    {
+      "epoch": 3.118987341772152,
+      "grad_norm": 0.32839142133108234,
+      "learning_rate": 1.4755120974480923e-05,
+      "loss": 0.3753,
+      "step": 770
+    },
+    {
+      "epoch": 3.123037974683544,
+      "grad_norm": 0.3146563539038781,
+      "learning_rate": 1.4700369956066771e-05,
+      "loss": 0.3651,
+      "step": 771
+    },
+    {
+      "epoch": 3.1270886075949367,
+      "grad_norm": 0.31746454462495627,
+      "learning_rate": 1.4645661620144413e-05,
+      "loss": 0.3423,
+      "step": 772
+    },
+    {
+      "epoch": 3.131139240506329,
+      "grad_norm": 0.3657742299977599,
+      "learning_rate": 1.4590996407327284e-05,
+      "loss": 0.3317,
+      "step": 773
+    },
+    {
+      "epoch": 3.1351898734177217,
+      "grad_norm": 0.31409731155271403,
+      "learning_rate": 1.4536374757881487e-05,
+      "loss": 0.3531,
+      "step": 774
+    },
+    {
+      "epoch": 3.1392405063291138,
+      "grad_norm": 0.3376175812481046,
+      "learning_rate": 1.4481797111722271e-05,
+      "loss": 0.3479,
+      "step": 775
+    },
+    {
+      "epoch": 3.1432911392405063,
+      "grad_norm": 0.29578060158354014,
+      "learning_rate": 1.4427263908410507e-05,
+      "loss": 0.371,
+      "step": 776
+    },
+    {
+      "epoch": 3.147341772151899,
+      "grad_norm": 0.32487713018657943,
+      "learning_rate": 1.4372775587149108e-05,
+      "loss": 0.361,
+      "step": 777
+    },
+    {
+      "epoch": 3.1513924050632913,
+      "grad_norm": 0.34158550204498384,
+      "learning_rate": 1.4318332586779522e-05,
+      "loss": 0.36,
+      "step": 778
+    },
+    {
+      "epoch": 3.1554430379746834,
+      "grad_norm": 0.3119238916023475,
+      "learning_rate": 1.4263935345778202e-05,
+      "loss": 0.3947,
+      "step": 779
+    },
+    {
+      "epoch": 3.159493670886076,
+      "grad_norm": 0.36419768485094245,
+      "learning_rate": 1.420958430225303e-05,
+      "loss": 0.3638,
+      "step": 780
+    },
+    {
+      "epoch": 3.1635443037974684,
+      "grad_norm": 0.3548147114119493,
+      "learning_rate": 1.415527989393985e-05,
+      "loss": 0.3914,
+      "step": 781
+    },
+    {
+      "epoch": 3.167594936708861,
+      "grad_norm": 0.34463353240673167,
+      "learning_rate": 1.410102255819891e-05,
+      "loss": 0.3603,
+      "step": 782
+    },
+    {
+      "epoch": 3.171645569620253,
+      "grad_norm": 0.35688575252168375,
+      "learning_rate": 1.404681273201131e-05,
+      "loss": 0.3991,
+      "step": 783
+    },
+    {
+      "epoch": 3.1756962025316455,
+      "grad_norm": 0.3108808585933113,
+      "learning_rate": 1.399265085197556e-05,
+      "loss": 0.3654,
+      "step": 784
+    },
+    {
+      "epoch": 3.179746835443038,
+      "grad_norm": 0.33912281784137877,
+      "learning_rate": 1.393853735430398e-05,
+      "loss": 0.3505,
+      "step": 785
+    },
+    {
+      "epoch": 3.1837974683544306,
+      "grad_norm": 0.3419758440899405,
+      "learning_rate": 1.3884472674819246e-05,
+      "loss": 0.4058,
+      "step": 786
+    },
+    {
+      "epoch": 3.1878481012658226,
+      "grad_norm": 0.3075760264137183,
+      "learning_rate": 1.3830457248950864e-05,
+      "loss": 0.3678,
+      "step": 787
+    },
+    {
+      "epoch": 3.191898734177215,
+      "grad_norm": 0.3055104074809774,
+      "learning_rate": 1.377649151173163e-05,
+      "loss": 0.3726,
+      "step": 788
+    },
+    {
+      "epoch": 3.1959493670886077,
+      "grad_norm": 0.3276894493308634,
+      "learning_rate": 1.3722575897794181e-05,
+      "loss": 0.3725,
+      "step": 789
+    },
+    {
+      "epoch": 3.2,
+      "grad_norm": 0.3188005283532247,
+      "learning_rate": 1.3668710841367472e-05,
+      "loss": 0.3933,
+      "step": 790
+    },
+    {
+      "epoch": 3.2040506329113922,
+      "grad_norm": 0.3403005716179843,
+      "learning_rate": 1.361489677627324e-05,
+      "loss": 0.3821,
+      "step": 791
+    },
+    {
+      "epoch": 3.2081012658227848,
+      "grad_norm": 0.3299116598081741,
+      "learning_rate": 1.3561134135922585e-05,
+      "loss": 0.3854,
+      "step": 792
+    },
+    {
+      "epoch": 3.2121518987341773,
+      "grad_norm": 0.31451547581995304,
+      "learning_rate": 1.350742335331241e-05,
+      "loss": 0.3773,
+      "step": 793
+    },
+    {
+      "epoch": 3.21620253164557,
+      "grad_norm": 0.32358958093539225,
+      "learning_rate": 1.345376486102198e-05,
+      "loss": 0.3546,
+      "step": 794
+    },
+    {
+      "epoch": 3.220253164556962,
+      "grad_norm": 0.3213380121122191,
+      "learning_rate": 1.3400159091209414e-05,
+      "loss": 0.3807,
+      "step": 795
+    },
+    {
+      "epoch": 3.2243037974683544,
+      "grad_norm": 0.3067449208904906,
+      "learning_rate": 1.3346606475608216e-05,
+      "loss": 0.3903,
+      "step": 796
+    },
+    {
+      "epoch": 3.228354430379747,
+      "grad_norm": 0.3101957920424751,
+      "learning_rate": 1.3293107445523781e-05,
+      "loss": 0.3755,
+      "step": 797
+    },
+    {
+      "epoch": 3.2324050632911394,
+      "grad_norm": 0.33187922280732596,
+      "learning_rate": 1.3239662431829949e-05,
+      "loss": 0.3554,
+      "step": 798
+    },
+    {
+      "epoch": 3.2364556962025315,
+      "grad_norm": 0.3138768351590569,
+      "learning_rate": 1.3186271864965509e-05,
+      "loss": 0.3583,
+      "step": 799
+    },
+    {
+      "epoch": 3.240506329113924,
+      "grad_norm": 0.32451583729143063,
+      "learning_rate": 1.3132936174930756e-05,
+      "loss": 0.3576,
+      "step": 800
+    },
+    {
+      "epoch": 3.2445569620253165,
+      "grad_norm": 0.33682516237295745,
+      "learning_rate": 1.3079655791283995e-05,
+      "loss": 0.3578,
+      "step": 801
+    },
+    {
+      "epoch": 3.248607594936709,
+      "grad_norm": 0.30587604238663335,
+      "learning_rate": 1.3026431143138108e-05,
+      "loss": 0.3987,
+      "step": 802
+    },
+    {
+      "epoch": 3.252658227848101,
+      "grad_norm": 0.3472079081844575,
+      "learning_rate": 1.2973262659157114e-05,
+      "loss": 0.3443,
+      "step": 803
+    },
+    {
+      "epoch": 3.2567088607594936,
+      "grad_norm": 0.35277029806538607,
+      "learning_rate": 1.2920150767552651e-05,
+      "loss": 0.3814,
+      "step": 804
+    },
+    {
+      "epoch": 3.260759493670886,
+      "grad_norm": 0.3594882377359774,
+      "learning_rate": 1.2867095896080607e-05,
+      "loss": 0.3702,
+      "step": 805
+    },
+    {
+      "epoch": 3.2648101265822786,
+      "grad_norm": 0.3224350983315474,
+      "learning_rate": 1.2814098472037612e-05,
+      "loss": 0.3768,
+      "step": 806
+    },
+    {
+      "epoch": 3.2688607594936707,
+      "grad_norm": 0.3130003048730583,
+      "learning_rate": 1.276115892225764e-05,
+      "loss": 0.378,
+      "step": 807
+    },
+    {
+      "epoch": 3.2729113924050632,
+      "grad_norm": 0.31306087827774726,
+      "learning_rate": 1.2708277673108555e-05,
+      "loss": 0.3608,
+      "step": 808
+    },
+    {
+      "epoch": 3.2769620253164558,
+      "grad_norm": 0.3324090464468117,
+      "learning_rate": 1.2655455150488649e-05,
+      "loss": 0.3814,
+      "step": 809
+    },
+    {
+      "epoch": 3.2810126582278483,
+      "grad_norm": 0.3005213182021047,
+      "learning_rate": 1.2602691779823272e-05,
+      "loss": 0.4055,
+      "step": 810
+    },
+    {
+      "epoch": 3.2850632911392403,
+      "grad_norm": 0.313353044890625,
+      "learning_rate": 1.2549987986061355e-05,
+      "loss": 0.3688,
+      "step": 811
+    },
+    {
+      "epoch": 3.289113924050633,
+      "grad_norm": 0.3062728626367921,
+      "learning_rate": 1.2497344193672005e-05,
+      "loss": 0.4088,
+      "step": 812
+    },
+    {
+      "epoch": 3.2931645569620254,
+      "grad_norm": 0.2999917365626548,
+      "learning_rate": 1.2444760826641092e-05,
+      "loss": 0.383,
+      "step": 813
+    },
+    {
+      "epoch": 3.297215189873418,
+      "grad_norm": 0.31934051338588915,
+      "learning_rate": 1.2392238308467817e-05,
+      "loss": 0.3521,
+      "step": 814
+    },
+    {
+      "epoch": 3.30126582278481,
+      "grad_norm": 0.31245403815814676,
+      "learning_rate": 1.2339777062161326e-05,
+      "loss": 0.3843,
+      "step": 815
+    },
+    {
+      "epoch": 3.3053164556962025,
+      "grad_norm": 0.29450048444735927,
+      "learning_rate": 1.2287377510237293e-05,
+      "loss": 0.3599,
+      "step": 816
+    },
+    {
+      "epoch": 3.309367088607595,
+      "grad_norm": 0.3096515436807607,
+      "learning_rate": 1.2235040074714488e-05,
+      "loss": 0.3794,
+      "step": 817
+    },
+    {
+      "epoch": 3.3134177215189875,
+      "grad_norm": 0.3246422528608137,
+      "learning_rate": 1.2182765177111434e-05,
+      "loss": 0.3549,
+      "step": 818
+    },
+    {
+      "epoch": 3.3174683544303796,
+      "grad_norm": 0.29171632790272434,
+      "learning_rate": 1.213055323844297e-05,
+      "loss": 0.363,
+      "step": 819
+    },
+    {
+      "epoch": 3.321518987341772,
+      "grad_norm": 0.31481224466873453,
+      "learning_rate": 1.2078404679216864e-05,
+      "loss": 0.3515,
+      "step": 820
+    },
+    {
+      "epoch": 3.3255696202531646,
+      "grad_norm": 0.31549209851326404,
+      "learning_rate": 1.2026319919430458e-05,
+      "loss": 0.3549,
+      "step": 821
+    },
+    {
+      "epoch": 3.329620253164557,
+      "grad_norm": 0.3054194520371187,
+      "learning_rate": 1.1974299378567227e-05,
+      "loss": 0.3517,
+      "step": 822
+    },
+    {
+      "epoch": 3.333670886075949,
+      "grad_norm": 0.312662409486909,
+      "learning_rate": 1.1922343475593462e-05,
+      "loss": 0.3578,
+      "step": 823
+    },
+    {
+      "epoch": 3.3377215189873417,
+      "grad_norm": 0.31915212841067725,
+      "learning_rate": 1.187045262895488e-05,
+      "loss": 0.3763,
+      "step": 824
+    },
+    {
+      "epoch": 3.3417721518987342,
+      "grad_norm": 0.30078213179369273,
+      "learning_rate": 1.1818627256573203e-05,
+      "loss": 0.3478,
+      "step": 825
+    },
+    {
+      "epoch": 3.3458227848101267,
+      "grad_norm": 0.2977826018352105,
+      "learning_rate": 1.1766867775842864e-05,
+      "loss": 0.386,
+      "step": 826
+    },
+    {
+      "epoch": 3.349873417721519,
+      "grad_norm": 0.3207264932499584,
+      "learning_rate": 1.1715174603627615e-05,
+      "loss": 0.3731,
+      "step": 827
+    },
+    {
+      "epoch": 3.3539240506329113,
+      "grad_norm": 0.3286181922135335,
+      "learning_rate": 1.1663548156257147e-05,
+      "loss": 0.3466,
+      "step": 828
+    },
+    {
+      "epoch": 3.357974683544304,
+      "grad_norm": 0.35633356574930464,
+      "learning_rate": 1.161198884952377e-05,
+      "loss": 0.3866,
+      "step": 829
+    },
+    {
+      "epoch": 3.3620253164556964,
+      "grad_norm": 0.305896007916683,
+      "learning_rate": 1.1560497098679056e-05,
+      "loss": 0.3637,
+      "step": 830
+    },
+    {
+      "epoch": 3.3660759493670884,
+      "grad_norm": 0.3258577402437654,
+      "learning_rate": 1.1509073318430479e-05,
+      "loss": 0.3642,
+      "step": 831
+    },
+    {
+      "epoch": 3.370126582278481,
+      "grad_norm": 0.3010577781462627,
+      "learning_rate": 1.1457717922938116e-05,
+      "loss": 0.3797,
+      "step": 832
+    },
+    {
+      "epoch": 3.3741772151898735,
+      "grad_norm": 0.3257973985017279,
+      "learning_rate": 1.1406431325811233e-05,
+      "loss": 0.3724,
+      "step": 833
+    },
+    {
+      "epoch": 3.378227848101266,
+      "grad_norm": 0.35811211365755935,
+      "learning_rate": 1.135521394010506e-05,
+      "loss": 0.3747,
+      "step": 834
+    },
+    {
+      "epoch": 3.382278481012658,
+      "grad_norm": 0.3118558731304108,
+      "learning_rate": 1.1304066178317367e-05,
+      "loss": 0.3601,
+      "step": 835
+    },
+    {
+      "epoch": 3.3863291139240506,
+      "grad_norm": 0.31104641352878853,
+      "learning_rate": 1.1252988452385199e-05,
+      "loss": 0.3569,
+      "step": 836
+    },
+    {
+      "epoch": 3.390379746835443,
+      "grad_norm": 0.3042123428063967,
+      "learning_rate": 1.1201981173681536e-05,
+      "loss": 0.3773,
+      "step": 837
+    },
+    {
+      "epoch": 3.3944303797468356,
+      "grad_norm": 0.31085805761028923,
+      "learning_rate": 1.1151044753011991e-05,
+      "loss": 0.3435,
+      "step": 838
+    },
+    {
+      "epoch": 3.3984810126582277,
+      "grad_norm": 0.31147093630481515,
+      "learning_rate": 1.1100179600611491e-05,
+      "loss": 0.3608,
+      "step": 839
+    },
+    {
+      "epoch": 3.40253164556962,
+      "grad_norm": 0.3011112754115436,
+      "learning_rate": 1.1049386126140985e-05,
+      "loss": 0.3585,
+      "step": 840
+    },
+    {
+      "epoch": 3.4065822784810127,
+      "grad_norm": 0.29174640333465257,
+      "learning_rate": 1.0998664738684128e-05,
+      "loss": 0.3859,
+      "step": 841
+    },
+    {
+      "epoch": 3.410632911392405,
+      "grad_norm": 0.3218979872656447,
+      "learning_rate": 1.0948015846744e-05,
+      "loss": 0.3716,
+      "step": 842
+    },
+    {
+      "epoch": 3.4146835443037973,
+      "grad_norm": 0.3199395047756437,
+      "learning_rate": 1.0897439858239832e-05,
+      "loss": 0.3686,
+      "step": 843
+    },
+    {
+      "epoch": 3.41873417721519,
+      "grad_norm": 0.2915938602618768,
+      "learning_rate": 1.0846937180503652e-05,
+      "loss": 0.3618,
+      "step": 844
+    },
+    {
+      "epoch": 3.4227848101265823,
+      "grad_norm": 0.3179577427617852,
+      "learning_rate": 1.0796508220277117e-05,
+      "loss": 0.3681,
+      "step": 845
+    },
+    {
+      "epoch": 3.426835443037975,
+      "grad_norm": 0.33253729341891913,
+      "learning_rate": 1.0746153383708107e-05,
+      "loss": 0.3673,
+      "step": 846
+    },
+    {
+      "epoch": 3.430886075949367,
+      "grad_norm": 0.29986316680683817,
+      "learning_rate": 1.0695873076347579e-05,
+      "loss": 0.3905,
+      "step": 847
+    },
+    {
+      "epoch": 3.4349367088607594,
+      "grad_norm": 0.2989766820999163,
+      "learning_rate": 1.0645667703146205e-05,
+      "loss": 0.3566,
+      "step": 848
+    },
+    {
+      "epoch": 3.438987341772152,
+      "grad_norm": 0.3409747627321669,
+      "learning_rate": 1.0595537668451161e-05,
+      "loss": 0.3656,
+      "step": 849
+    },
+    {
+      "epoch": 3.4430379746835444,
+      "grad_norm": 0.3404796494206864,
+      "learning_rate": 1.0545483376002854e-05,
+      "loss": 0.3973,
+      "step": 850
+    },
+    {
+      "epoch": 3.4470886075949365,
+      "grad_norm": 0.32529707336990316,
+      "learning_rate": 1.0495505228931676e-05,
+      "loss": 0.366,
+      "step": 851
+    },
+    {
+      "epoch": 3.451139240506329,
+      "grad_norm": 0.3380261947574518,
+      "learning_rate": 1.044560362975474e-05,
+      "loss": 0.3678,
+      "step": 852
+    },
+    {
+      "epoch": 3.4551898734177215,
+      "grad_norm": 0.3581631307557633,
+      "learning_rate": 1.0395778980372695e-05,
+      "loss": 0.3844,
+      "step": 853
+    },
+    {
+      "epoch": 3.459240506329114,
+      "grad_norm": 0.31357950315330996,
+      "learning_rate": 1.0346031682066381e-05,
+      "loss": 0.3593,
+      "step": 854
+    },
+    {
+      "epoch": 3.463291139240506,
+      "grad_norm": 0.3148292090027833,
+      "learning_rate": 1.0296362135493724e-05,
+      "loss": 0.3676,
+      "step": 855
+    },
+    {
+      "epoch": 3.4673417721518986,
+      "grad_norm": 0.33780614406292564,
+      "learning_rate": 1.0246770740686422e-05,
+      "loss": 0.3975,
+      "step": 856
+    },
+    {
+      "epoch": 3.471392405063291,
+      "grad_norm": 0.31464306102828377,
+      "learning_rate": 1.0197257897046743e-05,
+      "loss": 0.3597,
+      "step": 857
+    },
+    {
+      "epoch": 3.4754430379746837,
+      "grad_norm": 0.306505651938787,
+      "learning_rate": 1.014782400334433e-05,
+      "loss": 0.3801,
+      "step": 858
+    },
+    {
+      "epoch": 3.479493670886076,
+      "grad_norm": 0.2882519530183802,
+      "learning_rate": 1.009846945771296e-05,
+      "loss": 0.3892,
+      "step": 859
+    },
+    {
+      "epoch": 3.4835443037974683,
+      "grad_norm": 0.34101727963574346,
+      "learning_rate": 1.0049194657647363e-05,
+      "loss": 0.3873,
+      "step": 860
+    },
+    {
+      "epoch": 3.487594936708861,
+      "grad_norm": 0.3042703208656616,
+      "learning_rate": 1.0000000000000006e-05,
+      "loss": 0.3822,
+      "step": 861
+    },
+    {
+      "epoch": 3.4916455696202533,
+      "grad_norm": 0.30110190924120495,
+      "learning_rate": 9.950885880977891e-06,
+      "loss": 0.361,
+      "step": 862
+    },
+    {
+      "epoch": 3.4956962025316454,
+      "grad_norm": 0.2937942486006718,
+      "learning_rate": 9.901852696139382e-06,
+      "loss": 0.3866,
+      "step": 863
+    },
+    {
+      "epoch": 3.499746835443038,
+      "grad_norm": 0.30355970728158926,
+      "learning_rate": 9.852900840391027e-06,
+      "loss": 0.379,
+      "step": 864
+    },
+    {
+      "epoch": 3.5037974683544304,
+      "grad_norm": 0.27604181347116125,
+      "learning_rate": 9.804030707984313e-06,
+      "loss": 0.3813,
+      "step": 865
+    },
+    {
+      "epoch": 3.507848101265823,
+      "grad_norm": 0.291901450414897,
+      "learning_rate": 9.755242692512599e-06,
+      "loss": 0.3769,
+      "step": 866
+    },
+    {
+      "epoch": 3.5118987341772154,
+      "grad_norm": 0.2953671876255077,
+      "learning_rate": 9.70653718690782e-06,
+      "loss": 0.3767,
+      "step": 867
+    },
+    {
+      "epoch": 3.5159493670886075,
+      "grad_norm": 0.28345380953216576,
+      "learning_rate": 9.657914583437454e-06,
+      "loss": 0.3994,
+      "step": 868
+    },
+    {
+      "epoch": 3.52,
+      "grad_norm": 0.28587062452947604,
+      "learning_rate": 9.609375273701246e-06,
+      "loss": 0.3789,
+      "step": 869
+    },
+    {
+      "epoch": 3.5240506329113925,
+      "grad_norm": 0.3025718951166699,
+      "learning_rate": 9.560919648628133e-06,
+      "loss": 0.3521,
+      "step": 870
+    },
+    {
+      "epoch": 3.5281012658227846,
+      "grad_norm": 0.29262762374598444,
+      "learning_rate": 9.512548098473047e-06,
+      "loss": 0.3647,
+      "step": 871
+    },
+    {
+      "epoch": 3.532151898734177,
+      "grad_norm": 0.31044251778277154,
+      "learning_rate": 9.464261012813825e-06,
+      "loss": 0.3669,
+      "step": 872
+    },
+    {
+      "epoch": 3.5362025316455696,
+      "grad_norm": 0.277271904025459,
+      "learning_rate": 9.416058780547987e-06,
+      "loss": 0.3942,
+      "step": 873
+    },
+    {
+      "epoch": 3.540253164556962,
+      "grad_norm": 0.3114642800876992,
+      "learning_rate": 9.367941789889714e-06,
+      "loss": 0.3467,
+      "step": 874
+    },
+    {
+      "epoch": 3.5443037974683547,
+      "grad_norm": 0.3096110066800523,
+      "learning_rate": 9.319910428366607e-06,
+      "loss": 0.3647,
+      "step": 875
+    },
+    {
+      "epoch": 3.5483544303797467,
+      "grad_norm": 0.28959088964099156,
+      "learning_rate": 9.271965082816667e-06,
+      "loss": 0.3846,
+      "step": 876
+    },
+    {
+      "epoch": 3.5524050632911393,
+      "grad_norm": 0.30083251974389597,
+      "learning_rate": 9.224106139385111e-06,
+      "loss": 0.3539,
+      "step": 877
+    },
+    {
+      "epoch": 3.5564556962025318,
+      "grad_norm": 0.2763475676004709,
+      "learning_rate": 9.176333983521291e-06,
+      "loss": 0.3847,
+      "step": 878
+    },
+    {
+      "epoch": 3.560506329113924,
+      "grad_norm": 0.2950781952952933,
+      "learning_rate": 9.12864899997558e-06,
+      "loss": 0.3747,
+      "step": 879
+    },
+    {
+      "epoch": 3.5645569620253164,
+      "grad_norm": 0.3089783105485836,
+      "learning_rate": 9.08105157279628e-06,
+      "loss": 0.3339,
+      "step": 880
+    },
+    {
+      "epoch": 3.568607594936709,
+      "grad_norm": 0.2913295489668948,
+      "learning_rate": 9.03354208532653e-06,
+      "loss": 0.3974,
+      "step": 881
+    },
+    {
+      "epoch": 3.5726582278481014,
+      "grad_norm": 0.2787461825671875,
+      "learning_rate": 8.986120920201205e-06,
+      "loss": 0.3746,
+      "step": 882
+    },
+    {
+      "epoch": 3.576708860759494,
+      "grad_norm": 0.2911734856574514,
+      "learning_rate": 8.938788459343852e-06,
+      "loss": 0.3898,
+      "step": 883
+    },
+    {
+      "epoch": 3.580759493670886,
+      "grad_norm": 0.2972170948793507,
+      "learning_rate": 8.8915450839636e-06,
+      "loss": 0.3504,
+      "step": 884
+    },
+    {
+      "epoch": 3.5848101265822785,
+      "grad_norm": 0.28388427053773546,
+      "learning_rate": 8.844391174552116e-06,
+      "loss": 0.3517,
+      "step": 885
+    },
+    {
+      "epoch": 3.588860759493671,
+      "grad_norm": 0.28509962247419773,
+      "learning_rate": 8.797327110880479e-06,
+      "loss": 0.3552,
+      "step": 886
+    },
+    {
+      "epoch": 3.592911392405063,
+      "grad_norm": 0.30344938726069004,
+      "learning_rate": 8.750353271996206e-06,
+      "loss": 0.3612,
+      "step": 887
+    },
+    {
+      "epoch": 3.5969620253164556,
+      "grad_norm": 0.3028111495277901,
+      "learning_rate": 8.703470036220132e-06,
+      "loss": 0.3659,
+      "step": 888
+    },
+    {
+      "epoch": 3.601012658227848,
+      "grad_norm": 0.2930815045245661,
+      "learning_rate": 8.656677781143394e-06,
+      "loss": 0.3646,
+      "step": 889
+    },
+    {
+      "epoch": 3.6050632911392406,
+      "grad_norm": 0.31148373564422904,
+      "learning_rate": 8.609976883624377e-06,
+      "loss": 0.3675,
+      "step": 890
+    },
+    {
+      "epoch": 3.609113924050633,
+      "grad_norm": 0.27602732910010547,
+      "learning_rate": 8.563367719785698e-06,
+      "loss": 0.3747,
+      "step": 891
+    },
+    {
+      "epoch": 3.613164556962025,
+      "grad_norm": 0.2891068169127576,
+      "learning_rate": 8.516850665011138e-06,
+      "loss": 0.385,
+      "step": 892
+    },
+    {
+      "epoch": 3.6172151898734177,
+      "grad_norm": 0.29094105928931574,
+      "learning_rate": 8.47042609394269e-06,
+      "loss": 0.39,
+      "step": 893
+    },
+    {
+      "epoch": 3.6212658227848102,
+      "grad_norm": 0.3003695244789969,
+      "learning_rate": 8.424094380477432e-06,
+      "loss": 0.3557,
+      "step": 894
+    },
+    {
+      "epoch": 3.6253164556962023,
+      "grad_norm": 0.3065070687210877,
+      "learning_rate": 8.37785589776465e-06,
+      "loss": 0.3704,
+      "step": 895
+    },
+    {
+      "epoch": 3.629367088607595,
+      "grad_norm": 0.30084397123954093,
+      "learning_rate": 8.331711018202694e-06,
+      "loss": 0.3518,
+      "step": 896
+    },
+    {
+      "epoch": 3.6334177215189873,
+      "grad_norm": 0.29207784638455647,
+      "learning_rate": 8.285660113436104e-06,
+      "loss": 0.3746,
+      "step": 897
+    },
+    {
+      "epoch": 3.63746835443038,
+      "grad_norm": 0.30010491988105403,
+      "learning_rate": 8.239703554352527e-06,
+      "loss": 0.3866,
+      "step": 898
+    },
+    {
+      "epoch": 3.6415189873417724,
+      "grad_norm": 0.2971534379998443,
+      "learning_rate": 8.193841711079775e-06,
+      "loss": 0.3925,
+      "step": 899
+    },
+    {
+      "epoch": 3.6455696202531644,
+      "grad_norm": 0.306121761729776,
+      "learning_rate": 8.148074952982828e-06,
+      "loss": 0.3936,
+      "step": 900
+    },
+    {
+      "epoch": 3.649620253164557,
+      "grad_norm": 0.3047114483432978,
+      "learning_rate": 8.102403648660859e-06,
+      "loss": 0.3624,
+      "step": 901
+    },
+    {
+      "epoch": 3.6536708860759495,
+      "grad_norm": 0.3023527119463433,
+      "learning_rate": 8.056828165944282e-06,
+      "loss": 0.3681,
+      "step": 902
+    },
+    {
+      "epoch": 3.6577215189873415,
+      "grad_norm": 0.3020114601728021,
+      "learning_rate": 8.011348871891762e-06,
+      "loss": 0.3676,
+      "step": 903
+    },
+    {
+      "epoch": 3.661772151898734,
+      "grad_norm": 0.2962280539038721,
+      "learning_rate": 7.965966132787287e-06,
+      "loss": 0.3753,
+      "step": 904
+    },
+    {
+      "epoch": 3.6658227848101266,
+      "grad_norm": 0.28298981837187204,
+      "learning_rate": 7.920680314137189e-06,
+      "loss": 0.3717,
+      "step": 905
+    },
+    {
+      "epoch": 3.669873417721519,
+      "grad_norm": 0.30075620107927376,
+      "learning_rate": 7.875491780667246e-06,
+      "loss": 0.373,
+      "step": 906
+    },
+    {
+      "epoch": 3.6739240506329116,
+      "grad_norm": 0.30312990096628994,
+      "learning_rate": 7.830400896319667e-06,
+      "loss": 0.3833,
+      "step": 907
+    },
+    {
+      "epoch": 3.6779746835443037,
+      "grad_norm": 0.2979886431310191,
+      "learning_rate": 7.785408024250259e-06,
+      "loss": 0.3584,
+      "step": 908
+    },
+    {
+      "epoch": 3.682025316455696,
+      "grad_norm": 0.2823083436109584,
+      "learning_rate": 7.74051352682542e-06,
+      "loss": 0.3685,
+      "step": 909
+    },
+    {
+      "epoch": 3.6860759493670887,
+      "grad_norm": 0.32171498945562593,
+      "learning_rate": 7.695717765619257e-06,
+      "loss": 0.3458,
+      "step": 910
+    },
+    {
+      "epoch": 3.690126582278481,
+      "grad_norm": 0.3421067942673589,
+      "learning_rate": 7.651021101410673e-06,
+      "loss": 0.3572,
+      "step": 911
+    },
+    {
+      "epoch": 3.6941772151898733,
+      "grad_norm": 0.31682316400285954,
+      "learning_rate": 7.606423894180464e-06,
+      "loss": 0.3499,
+      "step": 912
+    },
+    {
+      "epoch": 3.698227848101266,
+      "grad_norm": 0.27563795907745825,
+      "learning_rate": 7.56192650310839e-06,
+      "loss": 0.3745,
+      "step": 913
+    },
+    {
+      "epoch": 3.7022784810126583,
+      "grad_norm": 0.30880206761967044,
+      "learning_rate": 7.517529286570349e-06,
+      "loss": 0.3544,
+      "step": 914
+    },
+    {
+      "epoch": 3.706329113924051,
+      "grad_norm": 0.3013558334169142,
+      "learning_rate": 7.473232602135387e-06,
+      "loss": 0.3578,
+      "step": 915
+    },
+    {
+      "epoch": 3.710379746835443,
+      "grad_norm": 0.3041033039468408,
+      "learning_rate": 7.429036806562935e-06,
+      "loss": 0.3641,
+      "step": 916
+    },
+    {
+      "epoch": 3.7144303797468354,
+      "grad_norm": 0.2787075674642778,
+      "learning_rate": 7.3849422557998455e-06,
+      "loss": 0.3655,
+      "step": 917
+    },
+    {
+      "epoch": 3.718481012658228,
+      "grad_norm": 0.3070956861912738,
+      "learning_rate": 7.340949304977567e-06,
+      "loss": 0.3381,
+      "step": 918
+    },
+    {
+      "epoch": 3.72253164556962,
+      "grad_norm": 0.307847293043792,
+      "learning_rate": 7.297058308409282e-06,
+      "loss": 0.3503,
+      "step": 919
+    },
+    {
+      "epoch": 3.7265822784810125,
+      "grad_norm": 0.30712794052860704,
+      "learning_rate": 7.25326961958704e-06,
+      "loss": 0.3655,
+      "step": 920
+    },
+    {
+      "epoch": 3.730632911392405,
+      "grad_norm": 0.2929125765616298,
+      "learning_rate": 7.209583591178921e-06,
+      "loss": 0.3615,
+      "step": 921
+    },
+    {
+      "epoch": 3.7346835443037976,
+      "grad_norm": 0.2854460007906495,
+      "learning_rate": 7.1660005750261925e-06,
+      "loss": 0.3742,
+      "step": 922
+    },
+    {
+      "epoch": 3.73873417721519,
+      "grad_norm": 0.2776575236337104,
+      "learning_rate": 7.1225209221404765e-06,
+      "loss": 0.3695,
+      "step": 923
+    },
+    {
+      "epoch": 3.742784810126582,
+      "grad_norm": 0.28597873478831165,
+      "learning_rate": 7.079144982700909e-06,
+      "loss": 0.372,
+      "step": 924
+    },
+    {
+      "epoch": 3.7468354430379747,
+      "grad_norm": 0.29239353504418586,
+      "learning_rate": 7.0358731060513695e-06,
+      "loss": 0.3702,
+      "step": 925
+    },
+    {
+      "epoch": 3.750886075949367,
+      "grad_norm": 0.2900471766324087,
+      "learning_rate": 6.99270564069757e-06,
+      "loss": 0.3709,
+      "step": 926
+    },
+    {
+      "epoch": 3.7549367088607593,
+      "grad_norm": 0.2651047038457451,
+      "learning_rate": 6.949642934304375e-06,
+      "loss": 0.3821,
+      "step": 927
+    },
+    {
+      "epoch": 3.7589873417721518,
+      "grad_norm": 0.29769680692989525,
+      "learning_rate": 6.906685333692871e-06,
+      "loss": 0.3509,
+      "step": 928
+    },
+    {
+      "epoch": 3.7630379746835443,
+      "grad_norm": 0.2847789792962804,
+      "learning_rate": 6.86383318483769e-06,
+      "loss": 0.3896,
+      "step": 929
+    },
+    {
+      "epoch": 3.767088607594937,
+      "grad_norm": 0.28574203833067807,
+      "learning_rate": 6.821086832864139e-06,
+      "loss": 0.3664,
+      "step": 930
+    },
+    {
+      "epoch": 3.7711392405063293,
+      "grad_norm": 0.29185575546718423,
+      "learning_rate": 6.77844662204546e-06,
+      "loss": 0.3449,
+      "step": 931
+    },
+    {
+      "epoch": 3.7751898734177214,
+      "grad_norm": 0.2958781419522794,
+      "learning_rate": 6.7359128958000455e-06,
+      "loss": 0.3489,
+      "step": 932
+    },
+    {
+      "epoch": 3.779240506329114,
+      "grad_norm": 0.28938205254199595,
+      "learning_rate": 6.693485996688695e-06,
+      "loss": 0.3669,
+      "step": 933
+    },
+    {
+      "epoch": 3.7832911392405064,
+      "grad_norm": 0.276326917045053,
+      "learning_rate": 6.651166266411801e-06,
+      "loss": 0.3728,
+      "step": 934
+    },
+    {
+      "epoch": 3.7873417721518985,
+      "grad_norm": 0.2840785195180916,
+      "learning_rate": 6.6089540458066725e-06,
+      "loss": 0.3735,
+      "step": 935
+    },
+    {
+      "epoch": 3.791392405063291,
+      "grad_norm": 0.33424682437191794,
+      "learning_rate": 6.566849674844711e-06,
+      "loss": 0.3782,
+      "step": 936
+    },
+    {
+      "epoch": 3.7954430379746835,
+      "grad_norm": 0.2981108484657493,
+      "learning_rate": 6.524853492628747e-06,
+      "loss": 0.36,
+      "step": 937
+    },
+    {
+      "epoch": 3.799493670886076,
+      "grad_norm": 0.28709770172457044,
+      "learning_rate": 6.4829658373902536e-06,
+      "loss": 0.3515,
+      "step": 938
+    },
+    {
+      "epoch": 3.8035443037974686,
+      "grad_norm": 0.30589736016838626,
+      "learning_rate": 6.441187046486648e-06,
+      "loss": 0.3551,
+      "step": 939
+    },
+    {
+      "epoch": 3.8075949367088606,
+      "grad_norm": 0.29049194179307275,
+      "learning_rate": 6.399517456398567e-06,
+      "loss": 0.3541,
+      "step": 940
+    },
+    {
+      "epoch": 3.811645569620253,
+      "grad_norm": 0.31994188185022887,
+      "learning_rate": 6.357957402727164e-06,
+      "loss": 0.3412,
+      "step": 941
+    },
+    {
+      "epoch": 3.8156962025316457,
+      "grad_norm": 0.28691262734027695,
+      "learning_rate": 6.316507220191395e-06,
+      "loss": 0.3691,
+      "step": 942
+    },
+    {
+      "epoch": 3.8197468354430377,
+      "grad_norm": 0.2716588226833492,
+      "learning_rate": 6.275167242625331e-06,
+      "loss": 0.3764,
+      "step": 943
+    },
+    {
+      "epoch": 3.8237974683544302,
+      "grad_norm": 0.2944808568315611,
+      "learning_rate": 6.233937802975471e-06,
+      "loss": 0.3489,
+      "step": 944
+    },
+    {
+      "epoch": 3.8278481012658228,
+      "grad_norm": 0.29510750751818193,
+      "learning_rate": 6.192819233298046e-06,
+      "loss": 0.3642,
+      "step": 945
+    },
+    {
+      "epoch": 3.8318987341772153,
+      "grad_norm": 0.2870513089616419,
+      "learning_rate": 6.151811864756383e-06,
+      "loss": 0.3635,
+      "step": 946
+    },
+    {
+      "epoch": 3.835949367088608,
+      "grad_norm": 0.30056777690978964,
+      "learning_rate": 6.1109160276181655e-06,
+      "loss": 0.3728,
+      "step": 947
+    },
+    {
+      "epoch": 3.84,
+      "grad_norm": 0.30836327320933377,
+      "learning_rate": 6.070132051252868e-06,
+      "loss": 0.3161,
+      "step": 948
+    },
+    {
+      "epoch": 3.8440506329113924,
+      "grad_norm": 0.27554205594374653,
+      "learning_rate": 6.0294602641290034e-06,
+      "loss": 0.3726,
+      "step": 949
+    },
+    {
+      "epoch": 3.848101265822785,
+      "grad_norm": 0.26584193849907367,
+      "learning_rate": 5.988900993811575e-06,
+      "loss": 0.3857,
+      "step": 950
+    },
+    {
+      "epoch": 3.852151898734177,
+      "grad_norm": 0.271412626509287,
+      "learning_rate": 5.948454566959363e-06,
+      "loss": 0.3734,
+      "step": 951
+    },
+    {
+      "epoch": 3.8562025316455695,
+      "grad_norm": 0.3058185290020724,
+      "learning_rate": 5.908121309322328e-06,
+      "loss": 0.3591,
+      "step": 952
+    },
+    {
+      "epoch": 3.860253164556962,
+      "grad_norm": 0.2885014715196981,
+      "learning_rate": 5.867901545738976e-06,
+      "loss": 0.3712,
+      "step": 953
+    },
+    {
+      "epoch": 3.8643037974683545,
+      "grad_norm": 0.2732854863521488,
+      "learning_rate": 5.827795600133774e-06,
+      "loss": 0.3645,
+      "step": 954
+    },
+    {
+      "epoch": 3.868354430379747,
+      "grad_norm": 0.2703747171384053,
+      "learning_rate": 5.787803795514466e-06,
+      "loss": 0.3817,
+      "step": 955
+    },
+    {
+      "epoch": 3.872405063291139,
+      "grad_norm": 0.2837959013546624,
+      "learning_rate": 5.747926453969576e-06,
+      "loss": 0.3693,
+      "step": 956
+    },
+    {
+      "epoch": 3.8764556962025316,
+      "grad_norm": 0.2721748695135868,
+      "learning_rate": 5.708163896665708e-06,
+      "loss": 0.3729,
+      "step": 957
+    },
+    {
+      "epoch": 3.880506329113924,
+      "grad_norm": 0.2682486113755075,
+      "learning_rate": 5.668516443845047e-06,
+      "loss": 0.3784,
+      "step": 958
+    },
+    {
+      "epoch": 3.884556962025316,
+      "grad_norm": 0.2814986362612556,
+      "learning_rate": 5.6289844148227225e-06,
+      "loss": 0.3782,
+      "step": 959
+    },
+    {
+      "epoch": 3.8886075949367087,
+      "grad_norm": 0.2962431117833889,
+      "learning_rate": 5.5895681279842615e-06,
+      "loss": 0.3578,
+      "step": 960
+    },
+    {
+      "epoch": 3.8926582278481012,
+      "grad_norm": 0.28100543440889575,
+      "learning_rate": 5.550267900783019e-06,
+      "loss": 0.372,
+      "step": 961
+    },
+    {
+      "epoch": 3.8967088607594937,
+      "grad_norm": 0.2807520051266444,
+      "learning_rate": 5.511084049737623e-06,
+      "loss": 0.3569,
+      "step": 962
+    },
+    {
+      "epoch": 3.9007594936708863,
+      "grad_norm": 0.28774209290042724,
+      "learning_rate": 5.4720168904294215e-06,
+      "loss": 0.3614,
+      "step": 963
+    },
+    {
+      "epoch": 3.9048101265822783,
+      "grad_norm": 0.28058592269946964,
+      "learning_rate": 5.433066737499948e-06,
+      "loss": 0.3495,
+      "step": 964
+    },
+    {
+      "epoch": 3.908860759493671,
+      "grad_norm": 0.28017466226287907,
+      "learning_rate": 5.394233904648376e-06,
+      "loss": 0.3661,
+      "step": 965
+    },
+    {
+      "epoch": 3.9129113924050634,
+      "grad_norm": 0.2855718626693182,
+      "learning_rate": 5.355518704628997e-06,
+      "loss": 0.3616,
+      "step": 966
+    },
+    {
+      "epoch": 3.9169620253164554,
+      "grad_norm": 0.2711036363617598,
+      "learning_rate": 5.316921449248731e-06,
+      "loss": 0.377,
+      "step": 967
+    },
+    {
+      "epoch": 3.921012658227848,
+      "grad_norm": 0.2824782490362998,
+      "learning_rate": 5.278442449364538e-06,
+      "loss": 0.3704,
+      "step": 968
+    },
+    {
+      "epoch": 3.9250632911392405,
+      "grad_norm": 0.290540612738727,
+      "learning_rate": 5.240082014881016e-06,
+      "loss": 0.3752,
+      "step": 969
+    },
+    {
+      "epoch": 3.929113924050633,
+      "grad_norm": 0.2911611470869701,
+      "learning_rate": 5.201840454747822e-06,
+      "loss": 0.3517,
+      "step": 970
+    },
+    {
+      "epoch": 3.9331645569620255,
+      "grad_norm": 0.2996710313741473,
+      "learning_rate": 5.163718076957223e-06,
+      "loss": 0.3793,
+      "step": 971
+    },
+    {
+      "epoch": 3.9372151898734176,
+      "grad_norm": 0.2750518256628816,
+      "learning_rate": 5.125715188541609e-06,
+      "loss": 0.3818,
+      "step": 972
+    },
+    {
+      "epoch": 3.94126582278481,
+      "grad_norm": 0.28512717499927404,
+      "learning_rate": 5.087832095571021e-06,
+      "loss": 0.355,
+      "step": 973
+    },
+    {
+      "epoch": 3.9453164556962026,
+      "grad_norm": 0.29576861995092646,
+      "learning_rate": 5.0500691031506766e-06,
+      "loss": 0.3372,
+      "step": 974
+    },
+    {
+      "epoch": 3.9493670886075947,
+      "grad_norm": 0.29135261708962734,
+      "learning_rate": 5.01242651541854e-06,
+      "loss": 0.3629,
+      "step": 975
+    },
+    {
+      "epoch": 3.953417721518987,
+      "grad_norm": 0.28645701086790754,
+      "learning_rate": 4.974904635542815e-06,
+      "loss": 0.3855,
+      "step": 976
+    },
+    {
+      "epoch": 3.9574683544303797,
+      "grad_norm": 0.2816144858561018,
+      "learning_rate": 4.937503765719582e-06,
+      "loss": 0.376,
+      "step": 977
+    },
+    {
+      "epoch": 3.961518987341772,
+      "grad_norm": 0.27279517676742415,
+      "learning_rate": 4.900224207170299e-06,
+      "loss": 0.3665,
+      "step": 978
+    },
+    {
+      "epoch": 3.9655696202531647,
+      "grad_norm": 0.2764347671740021,
+      "learning_rate": 4.8630662601394065e-06,
+      "loss": 0.3682,
+      "step": 979
+    },
+    {
+      "epoch": 3.969620253164557,
+      "grad_norm": 0.28659076487249235,
+      "learning_rate": 4.8260302238918995e-06,
+      "loss": 0.3601,
+      "step": 980
+    },
+    {
+      "epoch": 3.9736708860759493,
+      "grad_norm": 0.2828197183252901,
+      "learning_rate": 4.789116396710924e-06,
+      "loss": 0.3721,
+      "step": 981
+    },
+    {
+      "epoch": 3.977721518987342,
+      "grad_norm": 0.27760667044777804,
+      "learning_rate": 4.752325075895368e-06,
+      "loss": 0.3779,
+      "step": 982
+    },
+    {
+      "epoch": 3.981772151898734,
+      "grad_norm": 0.27589198017309546,
+      "learning_rate": 4.715656557757473e-06,
+      "loss": 0.3656,
+      "step": 983
+    },
+    {
+      "epoch": 3.9858227848101264,
+      "grad_norm": 0.2843751140136988,
+      "learning_rate": 4.679111137620442e-06,
+      "loss": 0.3449,
+      "step": 984
+    },
+    {
+      "epoch": 3.989873417721519,
+      "grad_norm": 0.27656179083611293,
+      "learning_rate": 4.6426891098160585e-06,
+      "loss": 0.3702,
+      "step": 985
+    },
+    {
+      "epoch": 3.9939240506329114,
+      "grad_norm": 0.2933116471980363,
+      "learning_rate": 4.6063907676823474e-06,
+      "loss": 0.3453,
+      "step": 986
+    },
+    {
+      "epoch": 3.997974683544304,
+      "grad_norm": 0.26169048610092965,
+      "learning_rate": 4.570216403561141e-06,
+      "loss": 0.3757,
+      "step": 987
+    },
+    {
+      "epoch": 4.01620253164557,
+      "grad_norm": 0.397119050057016,
+      "learning_rate": 4.534166308795815e-06,
+      "loss": 0.3427,
+      "step": 988
+    },
+    {
+      "epoch": 4.020253164556962,
+      "grad_norm": 0.36742091816564376,
+      "learning_rate": 4.498240773728859e-06,
+      "loss": 0.3351,
+      "step": 989
+    },
+    {
+      "epoch": 4.024303797468354,
+      "grad_norm": 0.2947391164338875,
+      "learning_rate": 4.462440087699609e-06,
+      "loss": 0.3007,
+      "step": 990
+    },
+    {
+      "epoch": 4.028354430379747,
+      "grad_norm": 0.35229504382499965,
+      "learning_rate": 4.426764539041861e-06,
+      "loss": 0.314,
+      "step": 991
+    },
+    {
+      "epoch": 4.032405063291139,
+      "grad_norm": 0.4838814890431431,
+      "learning_rate": 4.391214415081582e-06,
+      "loss": 0.3272,
+      "step": 992
+    },
+    {
+      "epoch": 4.036455696202531,
+      "grad_norm": 0.4381960294808827,
+      "learning_rate": 4.355790002134579e-06,
+      "loss": 0.3322,
+      "step": 993
+    },
+    {
+      "epoch": 4.040506329113924,
+      "grad_norm": 0.32460058164693595,
+      "learning_rate": 4.320491585504207e-06,
+      "loss": 0.3327,
+      "step": 994
+    },
+    {
+      "epoch": 4.044556962025316,
+      "grad_norm": 0.3601252488977855,
+      "learning_rate": 4.2853194494790615e-06,
+      "loss": 0.3353,
+      "step": 995
+    },
+    {
+      "epoch": 4.048607594936709,
+      "grad_norm": 0.362142886978251,
+      "learning_rate": 4.250273877330691e-06,
+      "loss": 0.3502,
+      "step": 996
+    },
+    {
+      "epoch": 4.052658227848101,
+      "grad_norm": 0.35113896625193886,
+      "learning_rate": 4.215355151311313e-06,
+      "loss": 0.3271,
+      "step": 997
+    },
+    {
+      "epoch": 4.056708860759493,
+      "grad_norm": 0.29529430567402576,
+      "learning_rate": 4.180563552651542e-06,
+      "loss": 0.2928,
+      "step": 998
+    },
+    {
+      "epoch": 4.060759493670886,
+      "grad_norm": 0.2927625563150051,
+      "learning_rate": 4.145899361558147e-06,
+      "loss": 0.3253,
+      "step": 999
+    },
+    {
+      "epoch": 4.0648101265822785,
+      "grad_norm": 0.34372268581782534,
+      "learning_rate": 4.111362857211738e-06,
+      "loss": 0.3028,
+      "step": 1000
+    },
+    {
+      "epoch": 4.0688607594936705,
+      "grad_norm": 0.3763440502439793,
+      "learning_rate": 4.076954317764592e-06,
+      "loss": 0.3182,
+      "step": 1001
+    },
+    {
+      "epoch": 4.0729113924050635,
+      "grad_norm": 0.2957390827453549,
+      "learning_rate": 4.042674020338335e-06,
+      "loss": 0.3158,
+      "step": 1002
+    },
+    {
+      "epoch": 4.076962025316456,
+      "grad_norm": 0.28116137342989367,
+      "learning_rate": 4.0085222410217835e-06,
+      "loss": 0.3196,
+      "step": 1003
+    },
+    {
+      "epoch": 4.0810126582278485,
+      "grad_norm": 0.28875641751138453,
+      "learning_rate": 3.974499254868674e-06,
+      "loss": 0.3201,
+      "step": 1004
+    },
+    {
+      "epoch": 4.085063291139241,
+      "grad_norm": 0.3110917331499229,
+      "learning_rate": 3.940605335895451e-06,
+      "loss": 0.3115,
+      "step": 1005
+    },
+    {
+      "epoch": 4.089113924050633,
+      "grad_norm": 0.3104697123678754,
+      "learning_rate": 3.90684075707908e-06,
+      "loss": 0.3019,
+      "step": 1006
+    },
+    {
+      "epoch": 4.093164556962026,
+      "grad_norm": 0.2950751278754521,
+      "learning_rate": 3.8732057903548505e-06,
+      "loss": 0.2961,
+      "step": 1007
+    },
+    {
+      "epoch": 4.097215189873418,
+      "grad_norm": 0.3063299040245558,
+      "learning_rate": 3.8397007066141375e-06,
+      "loss": 0.3199,
+      "step": 1008
+    },
+    {
+      "epoch": 4.10126582278481,
+      "grad_norm": 0.2945398625903202,
+      "learning_rate": 3.806325775702304e-06,
+      "loss": 0.302,
+      "step": 1009
+    },
+    {
+      "epoch": 4.105316455696203,
+      "grad_norm": 0.2897563507148937,
+      "learning_rate": 3.773081266416434e-06,
+      "loss": 0.3367,
+      "step": 1010
+    },
+    {
+      "epoch": 4.109367088607595,
+      "grad_norm": 0.27885704301049413,
+      "learning_rate": 3.739967446503245e-06,
+      "loss": 0.3054,
+      "step": 1011
+    },
+    {
+      "epoch": 4.113417721518988,
+      "grad_norm": 0.27607827989028333,
+      "learning_rate": 3.706984582656894e-06,
+      "loss": 0.3224,
+      "step": 1012
+    },
+    {
+      "epoch": 4.11746835443038,
+      "grad_norm": 0.283634484441117,
+      "learning_rate": 3.6741329405168237e-06,
+      "loss": 0.3075,
+      "step": 1013
+    },
+    {
+      "epoch": 4.121518987341772,
+      "grad_norm": 0.2530101655369655,
+      "learning_rate": 3.641412784665648e-06,
+      "loss": 0.3128,
+      "step": 1014
+    },
+    {
+      "epoch": 4.125569620253165,
+      "grad_norm": 0.2610352863555636,
+      "learning_rate": 3.608824378627005e-06,
+      "loss": 0.3147,
+      "step": 1015
+    },
+    {
+      "epoch": 4.129620253164557,
+      "grad_norm": 0.2582801240930403,
+      "learning_rate": 3.5763679848634337e-06,
+      "loss": 0.3028,
+      "step": 1016
+    },
+    {
+      "epoch": 4.133670886075949,
+      "grad_norm": 0.2614306672313173,
+      "learning_rate": 3.544043864774269e-06,
+      "loss": 0.3279,
+      "step": 1017
+    },
+    {
+      "epoch": 4.137721518987342,
+      "grad_norm": 0.2860704684432012,
+      "learning_rate": 3.5118522786935282e-06,
+      "loss": 0.3427,
+      "step": 1018
+    },
+    {
+      "epoch": 4.141772151898734,
+      "grad_norm": 0.2455312757267927,
+      "learning_rate": 3.479793485887819e-06,
+      "loss": 0.3313,
+      "step": 1019
+    },
+    {
+      "epoch": 4.145822784810127,
+      "grad_norm": 0.24769652872055514,
+      "learning_rate": 3.4478677445542653e-06,
+      "loss": 0.338,
+      "step": 1020
+    },
+    {
+      "epoch": 4.149873417721519,
+      "grad_norm": 0.2503642516042581,
+      "learning_rate": 3.4160753118183767e-06,
+      "loss": 0.3208,
+      "step": 1021
+    },
+    {
+      "epoch": 4.153924050632911,
+      "grad_norm": 0.251831225808,
+      "learning_rate": 3.3844164437320527e-06,
+      "loss": 0.3219,
+      "step": 1022
+    },
+    {
+      "epoch": 4.157974683544304,
+      "grad_norm": 0.25765724878488194,
+      "learning_rate": 3.3528913952714558e-06,
+      "loss": 0.3397,
+      "step": 1023
+    },
+    {
+      "epoch": 4.162025316455696,
+      "grad_norm": 0.24059934261981294,
+      "learning_rate": 3.321500420335e-06,
+      "loss": 0.2994,
+      "step": 1024
+    },
+    {
+      "epoch": 4.166075949367088,
+      "grad_norm": 0.2455308715154884,
+      "learning_rate": 3.290243771741275e-06,
+      "loss": 0.3371,
+      "step": 1025
+    },
+    {
+      "epoch": 4.170126582278481,
+      "grad_norm": 0.2630449643185088,
+      "learning_rate": 3.2591217012270325e-06,
+      "loss": 0.2813,
+      "step": 1026
+    },
+    {
+      "epoch": 4.174177215189873,
+      "grad_norm": 0.25709278264291874,
+      "learning_rate": 3.228134459445149e-06,
+      "loss": 0.3138,
+      "step": 1027
+    },
+    {
+      "epoch": 4.178227848101266,
+      "grad_norm": 0.2514350479478436,
+      "learning_rate": 3.1972822959626205e-06,
+      "loss": 0.3195,
+      "step": 1028
+    },
+    {
+      "epoch": 4.182278481012658,
+      "grad_norm": 0.2519057519685664,
+      "learning_rate": 3.166565459258513e-06,
+      "loss": 0.3225,
+      "step": 1029
+    },
+    {
+      "epoch": 4.18632911392405,
+      "grad_norm": 0.25012728954562297,
+      "learning_rate": 3.1359841967220193e-06,
+      "loss": 0.3301,
+      "step": 1030
+    },
+    {
+      "epoch": 4.190379746835443,
+      "grad_norm": 0.2638076767157157,
+      "learning_rate": 3.105538754650419e-06,
+      "loss": 0.2908,
+      "step": 1031
+    },
+    {
+      "epoch": 4.194430379746835,
+      "grad_norm": 0.2554928428045576,
+      "learning_rate": 3.07522937824712e-06,
+      "loss": 0.3122,
+      "step": 1032
+    },
+    {
+      "epoch": 4.1984810126582275,
+      "grad_norm": 0.2731846409449159,
+      "learning_rate": 3.0450563116196697e-06,
+      "loss": 0.3017,
+      "step": 1033
+    },
+    {
+      "epoch": 4.2025316455696204,
+      "grad_norm": 0.2532234988567094,
+      "learning_rate": 3.0150197977778008e-06,
+      "loss": 0.3071,
+      "step": 1034
+    },
+    {
+      "epoch": 4.2065822784810125,
+      "grad_norm": 0.25456786295708667,
+      "learning_rate": 2.985120078631465e-06,
+      "loss": 0.3127,
+      "step": 1035
+    },
+    {
+      "epoch": 4.2106329113924055,
+      "grad_norm": 0.2544422007422183,
+      "learning_rate": 2.9553573949888893e-06,
+      "loss": 0.358,
+      "step": 1036
+    },
+    {
+      "epoch": 4.2146835443037975,
+      "grad_norm": 0.2562164567537424,
+      "learning_rate": 2.9257319865546384e-06,
+      "loss": 0.3115,
+      "step": 1037
+    },
+    {
+      "epoch": 4.21873417721519,
+      "grad_norm": 0.2511307999635332,
+      "learning_rate": 2.896244091927678e-06,
+      "loss": 0.3185,
+      "step": 1038
+    },
+    {
+      "epoch": 4.222784810126583,
+      "grad_norm": 0.25782215985298046,
+      "learning_rate": 2.8668939485994584e-06,
+      "loss": 0.2887,
+      "step": 1039
+    },
+    {
+      "epoch": 4.226835443037975,
+      "grad_norm": 0.2550252580623202,
+      "learning_rate": 2.837681792951994e-06,
+      "loss": 0.2887,
+      "step": 1040
+    },
+    {
+      "epoch": 4.230886075949367,
+      "grad_norm": 0.2555161554513018,
+      "learning_rate": 2.808607860255981e-06,
+      "loss": 0.2986,
+      "step": 1041
+    },
+    {
+      "epoch": 4.23493670886076,
+      "grad_norm": 0.2611637410316661,
+      "learning_rate": 2.7796723846688634e-06,
+      "loss": 0.3345,
+      "step": 1042
+    },
+    {
+      "epoch": 4.238987341772152,
+      "grad_norm": 0.25223782479392337,
+      "learning_rate": 2.7508755992329937e-06,
+      "loss": 0.3387,
+      "step": 1043
+    },
+    {
+      "epoch": 4.243037974683545,
+      "grad_norm": 0.2535618834280767,
+      "learning_rate": 2.722217735873718e-06,
+      "loss": 0.2995,
+      "step": 1044
+    },
+    {
+      "epoch": 4.247088607594937,
+      "grad_norm": 0.22956425289304497,
+      "learning_rate": 2.6936990253975315e-06,
+      "loss": 0.3355,
+      "step": 1045
+    },
+    {
+      "epoch": 4.251139240506329,
+      "grad_norm": 0.22877176669864052,
+      "learning_rate": 2.665319697490205e-06,
+      "loss": 0.3309,
+      "step": 1046
+    },
+    {
+      "epoch": 4.255189873417722,
+      "grad_norm": 0.26514259181599886,
+      "learning_rate": 2.637079980714945e-06,
+      "loss": 0.3412,
+      "step": 1047
+    },
+    {
+      "epoch": 4.259240506329114,
+      "grad_norm": 0.24107746465913032,
+      "learning_rate": 2.6089801025105453e-06,
+      "loss": 0.3139,
+      "step": 1048
+    },
+    {
+      "epoch": 4.263291139240506,
+      "grad_norm": 0.2502845552277692,
+      "learning_rate": 2.581020289189571e-06,
+      "loss": 0.333,
+      "step": 1049
+    },
+    {
+      "epoch": 4.267341772151899,
+      "grad_norm": 0.24755398842313994,
+      "learning_rate": 2.553200765936501e-06,
+      "loss": 0.2957,
+      "step": 1050
+    },
+    {
+      "epoch": 4.271392405063291,
+      "grad_norm": 0.25720575471269735,
+      "learning_rate": 2.525521756805962e-06,
+      "loss": 0.3179,
+      "step": 1051
+    },
+    {
+      "epoch": 4.275443037974684,
+      "grad_norm": 0.26215959804707345,
+      "learning_rate": 2.497983484720885e-06,
+      "loss": 0.3087,
+      "step": 1052
+    },
+    {
+      "epoch": 4.279493670886076,
+      "grad_norm": 0.24836213524100006,
+      "learning_rate": 2.470586171470728e-06,
+      "loss": 0.3032,
+      "step": 1053
+    },
+    {
+      "epoch": 4.283544303797468,
+      "grad_norm": 0.25439744590967783,
+      "learning_rate": 2.4433300377096836e-06,
+      "loss": 0.3207,
+      "step": 1054
+    },
+    {
+      "epoch": 4.287594936708861,
+      "grad_norm": 0.24995671886263754,
+      "learning_rate": 2.4162153029549073e-06,
+      "loss": 0.3138,
+      "step": 1055
+    },
+    {
+      "epoch": 4.291645569620253,
+      "grad_norm": 0.2481933206472738,
+      "learning_rate": 2.3892421855847458e-06,
+      "loss": 0.3239,
+      "step": 1056
+    },
+    {
+      "epoch": 4.295696202531645,
+      "grad_norm": 0.24166735901504,
+      "learning_rate": 2.362410902836978e-06,
+      "loss": 0.3137,
+      "step": 1057
+    },
+    {
+      "epoch": 4.299746835443038,
+      "grad_norm": 0.25504367186027677,
+      "learning_rate": 2.3357216708070653e-06,
+      "loss": 0.3029,
+      "step": 1058
+    },
+    {
+      "epoch": 4.30379746835443,
+      "grad_norm": 0.2588330463386714,
+      "learning_rate": 2.309174704446411e-06,
+      "loss": 0.2958,
+      "step": 1059
+    },
+    {
+      "epoch": 4.307848101265823,
+      "grad_norm": 0.23947781442260802,
+      "learning_rate": 2.2827702175606437e-06,
+      "loss": 0.3368,
+      "step": 1060
+    },
+    {
+      "epoch": 4.311898734177215,
+      "grad_norm": 0.23560872098307692,
+      "learning_rate": 2.256508422807855e-06,
+      "loss": 0.3498,
+      "step": 1061
+    },
+    {
+      "epoch": 4.315949367088607,
+      "grad_norm": 0.2563887672023255,
+      "learning_rate": 2.230389531696946e-06,
+      "loss": 0.2783,
+      "step": 1062
+    },
+    {
+      "epoch": 4.32,
+      "grad_norm": 0.24223859896860991,
+      "learning_rate": 2.204413754585857e-06,
+      "loss": 0.3371,
+      "step": 1063
+    },
+    {
+      "epoch": 4.324050632911392,
+      "grad_norm": 0.25113861780361474,
+      "learning_rate": 2.1785813006799406e-06,
+      "loss": 0.2884,
+      "step": 1064
+    },
+    {
+      "epoch": 4.328101265822784,
+      "grad_norm": 0.24960077324152555,
+      "learning_rate": 2.1528923780302224e-06,
+      "loss": 0.3137,
+      "step": 1065
+    },
+    {
+      "epoch": 4.332151898734177,
+      "grad_norm": 0.2485111613281755,
+      "learning_rate": 2.127347193531757e-06,
+      "loss": 0.2945,
+      "step": 1066
+    },
+    {
+      "epoch": 4.3362025316455695,
+      "grad_norm": 0.23666601179263394,
+      "learning_rate": 2.101945952921942e-06,
+      "loss": 0.3345,
+      "step": 1067
+    },
+    {
+      "epoch": 4.340253164556962,
+      "grad_norm": 0.24869121738143674,
+      "learning_rate": 2.0766888607788906e-06,
+      "loss": 0.3017,
+      "step": 1068
+    },
+    {
+      "epoch": 4.3443037974683545,
+      "grad_norm": 0.23863304067145946,
+      "learning_rate": 2.0515761205197337e-06,
+      "loss": 0.3474,
+      "step": 1069
+    },
+    {
+      "epoch": 4.348354430379747,
+      "grad_norm": 0.2621884624133366,
+      "learning_rate": 2.0266079343990453e-06,
+      "loss": 0.2818,
+      "step": 1070
+    },
+    {
+      "epoch": 4.3524050632911395,
+      "grad_norm": 0.24030245919095614,
+      "learning_rate": 2.0017845035071494e-06,
+      "loss": 0.2993,
+      "step": 1071
+    },
+    {
+      "epoch": 4.356455696202532,
+      "grad_norm": 0.23000080858406455,
+      "learning_rate": 1.9771060277685537e-06,
+      "loss": 0.3307,
+      "step": 1072
+    },
+    {
+      "epoch": 4.360506329113924,
+      "grad_norm": 0.2297561774589047,
+      "learning_rate": 1.95257270594031e-06,
+      "loss": 0.3291,
+      "step": 1073
+    },
+    {
+      "epoch": 4.364556962025317,
+      "grad_norm": 0.26129284627364674,
+      "learning_rate": 1.9281847356104188e-06,
+      "loss": 0.3174,
+      "step": 1074
+    },
+    {
+      "epoch": 4.368607594936709,
+      "grad_norm": 0.24119611178729974,
+      "learning_rate": 1.9039423131962365e-06,
+      "loss": 0.3304,
+      "step": 1075
+    },
+    {
+      "epoch": 4.372658227848102,
+      "grad_norm": 0.24257776321266633,
+      "learning_rate": 1.8798456339429027e-06,
+      "loss": 0.3161,
+      "step": 1076
+    },
+    {
+      "epoch": 4.376708860759494,
+      "grad_norm": 0.24171421737726567,
+      "learning_rate": 1.8558948919217612e-06,
+      "loss": 0.3177,
+      "step": 1077
+    },
+    {
+      "epoch": 4.380759493670886,
+      "grad_norm": 0.24841645292002343,
+      "learning_rate": 1.8320902800287954e-06,
+      "loss": 0.2795,
+      "step": 1078
+    },
+    {
+      "epoch": 4.384810126582279,
+      "grad_norm": 0.24311968797058933,
+      "learning_rate": 1.8084319899830726e-06,
+      "loss": 0.3204,
+      "step": 1079
+    },
+    {
+      "epoch": 4.388860759493671,
+      "grad_norm": 0.2318171616551859,
+      "learning_rate": 1.7849202123252097e-06,
+      "loss": 0.3011,
+      "step": 1080
+    },
+    {
+      "epoch": 4.392911392405063,
+      "grad_norm": 0.2554406689469906,
+      "learning_rate": 1.7615551364158401e-06,
+      "loss": 0.3066,
+      "step": 1081
+    },
+    {
+      "epoch": 4.396962025316456,
+      "grad_norm": 0.25378780325139333,
+      "learning_rate": 1.738336950434061e-06,
+      "loss": 0.3294,
+      "step": 1082
+    },
+    {
+      "epoch": 4.401012658227848,
+      "grad_norm": 0.2489238084288099,
+      "learning_rate": 1.715265841375957e-06,
+      "loss": 0.3086,
+      "step": 1083
+    },
+    {
+      "epoch": 4.405063291139241,
+      "grad_norm": 0.24920763989372294,
+      "learning_rate": 1.6923419950530684e-06,
+      "loss": 0.3292,
+      "step": 1084
+    },
+    {
+      "epoch": 4.409113924050633,
+      "grad_norm": 0.24979036369634483,
+      "learning_rate": 1.6695655960909008e-06,
+      "loss": 0.3418,
+      "step": 1085
+    },
+    {
+      "epoch": 4.413164556962025,
+      "grad_norm": 0.23329817340467807,
+      "learning_rate": 1.646936827927441e-06,
+      "loss": 0.3359,
+      "step": 1086
+    },
+    {
+      "epoch": 4.417215189873418,
+      "grad_norm": 0.23298724078656627,
+      "learning_rate": 1.6244558728116766e-06,
+      "loss": 0.3295,
+      "step": 1087
+    },
+    {
+      "epoch": 4.42126582278481,
+      "grad_norm": 0.23465390556525123,
+      "learning_rate": 1.6021229118021265e-06,
+      "loss": 0.3351,
+      "step": 1088
+    },
+    {
+      "epoch": 4.425316455696202,
+      "grad_norm": 0.23076606599952046,
+      "learning_rate": 1.5799381247653967e-06,
+      "loss": 0.3254,
+      "step": 1089
+    },
+    {
+      "epoch": 4.429367088607595,
+      "grad_norm": 0.2539764638081751,
+      "learning_rate": 1.5579016903747013e-06,
+      "loss": 0.3147,
+      "step": 1090
+    },
+    {
+      "epoch": 4.433417721518987,
+      "grad_norm": 0.24619711743607883,
+      "learning_rate": 1.5360137861084656e-06,
+      "loss": 0.3117,
+      "step": 1091
+    },
+    {
+      "epoch": 4.43746835443038,
+      "grad_norm": 0.23857363313239555,
+      "learning_rate": 1.5142745882488475e-06,
+      "loss": 0.3033,
+      "step": 1092
+    },
+    {
+      "epoch": 4.441518987341772,
+      "grad_norm": 0.2533241289780377,
+      "learning_rate": 1.4926842718803691e-06,
+      "loss": 0.2913,
+      "step": 1093
+    },
+    {
+      "epoch": 4.445569620253164,
+      "grad_norm": 0.2559208957756302,
+      "learning_rate": 1.4712430108884657e-06,
+      "loss": 0.2973,
+      "step": 1094
+    },
+    {
+      "epoch": 4.449620253164557,
+      "grad_norm": 0.2454686663953579,
+      "learning_rate": 1.4499509779581078e-06,
+      "loss": 0.3457,
+      "step": 1095
+    },
+    {
+      "epoch": 4.453670886075949,
+      "grad_norm": 0.23340946842587731,
+      "learning_rate": 1.4288083445723988e-06,
+      "loss": 0.3238,
+      "step": 1096
+    },
+    {
+      "epoch": 4.457721518987341,
+      "grad_norm": 0.243931405392362,
+      "learning_rate": 1.4078152810112045e-06,
+      "loss": 0.3104,
+      "step": 1097
+    },
+    {
+      "epoch": 4.461772151898734,
+      "grad_norm": 0.2550501639693127,
+      "learning_rate": 1.3869719563497697e-06,
+      "loss": 0.3095,
+      "step": 1098
+    },
+    {
+      "epoch": 4.465822784810126,
+      "grad_norm": 0.25287687703089456,
+      "learning_rate": 1.3662785384573663e-06,
+      "loss": 0.3189,
+      "step": 1099
+    },
+    {
+      "epoch": 4.469873417721519,
+      "grad_norm": 0.2529935621038048,
+      "learning_rate": 1.3457351939959383e-06,
+      "loss": 0.3121,
+      "step": 1100
+    },
+    {
+      "epoch": 4.473924050632911,
+      "grad_norm": 0.24865752586981088,
+      "learning_rate": 1.3253420884187551e-06,
+      "loss": 0.3013,
+      "step": 1101
+    },
+    {
+      "epoch": 4.4779746835443035,
+      "grad_norm": 0.23896766029546243,
+      "learning_rate": 1.3050993859690953e-06,
+      "loss": 0.3211,
+      "step": 1102
+    },
+    {
+      "epoch": 4.4820253164556965,
+      "grad_norm": 0.24098909578133745,
+      "learning_rate": 1.2850072496788869e-06,
+      "loss": 0.3035,
+      "step": 1103
+    },
+    {
+      "epoch": 4.4860759493670885,
+      "grad_norm": 0.23901771094012042,
+      "learning_rate": 1.2650658413674434e-06,
+      "loss": 0.3237,
+      "step": 1104
+    },
+    {
+      "epoch": 4.490126582278481,
+      "grad_norm": 0.235934549911032,
+      "learning_rate": 1.2452753216401226e-06,
+      "loss": 0.3446,
+      "step": 1105
+    },
+    {
+      "epoch": 4.494177215189874,
+      "grad_norm": 0.23323299730538719,
+      "learning_rate": 1.2256358498870503e-06,
+      "loss": 0.3441,
+      "step": 1106
+    },
+    {
+      "epoch": 4.498227848101266,
+      "grad_norm": 0.2534493816489681,
+      "learning_rate": 1.2061475842818337e-06,
+      "loss": 0.3057,
+      "step": 1107
+    },
+    {
+      "epoch": 4.502278481012659,
+      "grad_norm": 0.24612012746955747,
+      "learning_rate": 1.1868106817802816e-06,
+      "loss": 0.3149,
+      "step": 1108
+    },
+    {
+      "epoch": 4.506329113924051,
+      "grad_norm": 0.22945872873990536,
+      "learning_rate": 1.1676252981191482e-06,
+      "loss": 0.3273,
+      "step": 1109
+    },
+    {
+      "epoch": 4.510379746835443,
+      "grad_norm": 0.2514057994954423,
+      "learning_rate": 1.1485915878148823e-06,
+      "loss": 0.2839,
+      "step": 1110
+    },
+    {
+      "epoch": 4.514430379746836,
+      "grad_norm": 0.236576928748132,
+      "learning_rate": 1.1297097041623584e-06,
+      "loss": 0.3269,
+      "step": 1111
+    },
+    {
+      "epoch": 4.518481012658228,
+      "grad_norm": 0.2394247244003654,
+      "learning_rate": 1.1109797992336847e-06,
+      "loss": 0.3141,
+      "step": 1112
+    },
+    {
+      "epoch": 4.52253164556962,
+      "grad_norm": 0.2412009790522203,
+      "learning_rate": 1.092402023876933e-06,
+      "loss": 0.3151,
+      "step": 1113
+    },
+    {
+      "epoch": 4.526582278481013,
+      "grad_norm": 0.23516528560928548,
+      "learning_rate": 1.0739765277149527e-06,
+      "loss": 0.3149,
+      "step": 1114
+    },
+    {
+      "epoch": 4.530632911392405,
+      "grad_norm": 0.23274906134242382,
+      "learning_rate": 1.0557034591441596e-06,
+      "loss": 0.3348,
+      "step": 1115
+    },
+    {
+      "epoch": 4.534683544303798,
+      "grad_norm": 0.25058652525954117,
+      "learning_rate": 1.0375829653333324e-06,
+      "loss": 0.3068,
+      "step": 1116
+    },
+    {
+      "epoch": 4.53873417721519,
+      "grad_norm": 0.2468377546544449,
+      "learning_rate": 1.0196151922224385e-06,
+      "loss": 0.2939,
+      "step": 1117
+    },
+    {
+      "epoch": 4.542784810126582,
+      "grad_norm": 0.2436211772607353,
+      "learning_rate": 1.0018002845214526e-06,
+      "loss": 0.3309,
+      "step": 1118
+    },
+    {
+      "epoch": 4.546835443037975,
+      "grad_norm": 0.24309128360736726,
+      "learning_rate": 9.841383857091947e-07,
+      "loss": 0.3403,
+      "step": 1119
+    },
+    {
+      "epoch": 4.550886075949367,
+      "grad_norm": 0.23754745822296883,
+      "learning_rate": 9.666296380321616e-07,
+      "loss": 0.3362,
+      "step": 1120
+    },
+    {
+      "epoch": 4.55493670886076,
+      "grad_norm": 0.25447465335456293,
+      "learning_rate": 9.492741825034124e-07,
+      "loss": 0.3338,
+      "step": 1121
+    },
+    {
+      "epoch": 4.558987341772152,
+      "grad_norm": 0.24548979284269073,
+      "learning_rate": 9.320721589013892e-07,
+      "loss": 0.3046,
+      "step": 1122
+    },
+    {
+      "epoch": 4.563037974683544,
+      "grad_norm": 0.2337336208651659,
+      "learning_rate": 9.150237057688339e-07,
+      "loss": 0.3265,
+      "step": 1123
+    },
+    {
+      "epoch": 4.567088607594937,
+      "grad_norm": 0.2575834525771249,
+      "learning_rate": 8.981289604116328e-07,
+      "loss": 0.3063,
+      "step": 1124
+    },
+    {
+      "epoch": 4.571139240506329,
+      "grad_norm": 0.23899546147804512,
+      "learning_rate": 8.813880588977542e-07,
+      "loss": 0.3217,
+      "step": 1125
+    },
+    {
+      "epoch": 4.575189873417721,
+      "grad_norm": 0.25386759852320406,
+      "learning_rate": 8.648011360561126e-07,
+      "loss": 0.3039,
+      "step": 1126
+    },
+    {
+      "epoch": 4.579240506329114,
+      "grad_norm": 0.24468543920891803,
+      "learning_rate": 8.483683254755037e-07,
+      "loss": 0.3384,
+      "step": 1127
+    },
+    {
+      "epoch": 4.583291139240506,
+      "grad_norm": 0.2342016330750236,
+      "learning_rate": 8.320897595035227e-07,
+      "loss": 0.3296,
+      "step": 1128
+    },
+    {
+      "epoch": 4.587341772151898,
+      "grad_norm": 0.24179717982980012,
+      "learning_rate": 8.159655692455093e-07,
+      "loss": 0.3465,
+      "step": 1129
+    },
+    {
+      "epoch": 4.591392405063291,
+      "grad_norm": 0.24378264990647946,
+      "learning_rate": 7.999958845634648e-07,
+      "loss": 0.299,
+      "step": 1130
+    },
+    {
+      "epoch": 4.595443037974683,
+      "grad_norm": 0.23524440181184123,
+      "learning_rate": 7.841808340750478e-07,
+      "loss": 0.3427,
+      "step": 1131
+    },
+    {
+      "epoch": 4.599493670886076,
+      "grad_norm": 0.24259350501479493,
+      "learning_rate": 7.685205451524869e-07,
+      "loss": 0.3262,
+      "step": 1132
+    },
+    {
+      "epoch": 4.603544303797468,
+      "grad_norm": 0.253657166266368,
+      "learning_rate": 7.530151439216027e-07,
+      "loss": 0.3071,
+      "step": 1133
+    },
+    {
+      "epoch": 4.6075949367088604,
+      "grad_norm": 0.24599795757425072,
+      "learning_rate": 7.376647552607675e-07,
+      "loss": 0.3075,
+      "step": 1134
+    },
+    {
+      "epoch": 4.611645569620253,
+      "grad_norm": 0.24438642687229675,
+      "learning_rate": 7.224695027998963e-07,
+      "loss": 0.2974,
+      "step": 1135
+    },
+    {
+      "epoch": 4.6156962025316455,
+      "grad_norm": 0.2611781845705243,
+      "learning_rate": 7.07429508919466e-07,
+      "loss": 0.302,
+      "step": 1136
+    },
+    {
+      "epoch": 4.619746835443038,
+      "grad_norm": 0.2398192660956819,
+      "learning_rate": 6.925448947495206e-07,
+      "loss": 0.3131,
+      "step": 1137
+    },
+    {
+      "epoch": 4.6237974683544305,
+      "grad_norm": 0.238716592942517,
+      "learning_rate": 6.778157801686936e-07,
+      "loss": 0.3241,
+      "step": 1138
+    },
+    {
+      "epoch": 4.627848101265823,
+      "grad_norm": 0.239671733414706,
+      "learning_rate": 6.632422838032515e-07,
+      "loss": 0.3034,
+      "step": 1139
+    },
+    {
+      "epoch": 4.6318987341772155,
+      "grad_norm": 0.25004346651093184,
+      "learning_rate": 6.488245230261281e-07,
+      "loss": 0.3117,
+      "step": 1140
+    },
+    {
+      "epoch": 4.635949367088608,
+      "grad_norm": 0.24772461180268454,
+      "learning_rate": 6.345626139559868e-07,
+      "loss": 0.3257,
+      "step": 1141
+    },
+    {
+      "epoch": 4.64,
+      "grad_norm": 0.2237823884004297,
+      "learning_rate": 6.204566714562866e-07,
+      "loss": 0.3319,
+      "step": 1142
+    },
+    {
+      "epoch": 4.644050632911393,
+      "grad_norm": 0.23676509245938213,
+      "learning_rate": 6.06506809134344e-07,
+      "loss": 0.3214,
+      "step": 1143
+    },
+    {
+      "epoch": 4.648101265822785,
+      "grad_norm": 0.24044966120245245,
+      "learning_rate": 5.927131393404373e-07,
+      "loss": 0.3004,
+      "step": 1144
+    },
+    {
+      "epoch": 4.652151898734177,
+      "grad_norm": 0.23868660438146355,
+      "learning_rate": 5.790757731668817e-07,
+      "loss": 0.3259,
+      "step": 1145
+    },
+    {
+      "epoch": 4.65620253164557,
+      "grad_norm": 0.23101837191428276,
+      "learning_rate": 5.655948204471507e-07,
+      "loss": 0.3277,
+      "step": 1146
+    },
+    {
+      "epoch": 4.660253164556962,
+      "grad_norm": 0.23967594511141235,
+      "learning_rate": 5.522703897549875e-07,
+      "loss": 0.3207,
+      "step": 1147
+    },
+    {
+      "epoch": 4.664303797468355,
+      "grad_norm": 0.23365900516572158,
+      "learning_rate": 5.391025884035239e-07,
+      "loss": 0.336,
+      "step": 1148
+    },
+    {
+      "epoch": 4.668354430379747,
+      "grad_norm": 0.24502431498770627,
+      "learning_rate": 5.260915224444207e-07,
+      "loss": 0.3091,
+      "step": 1149
+    },
+    {
+      "epoch": 4.672405063291139,
+      "grad_norm": 0.22721802103796368,
+      "learning_rate": 5.132372966670129e-07,
+      "loss": 0.332,
+      "step": 1150
+    },
+    {
+      "epoch": 4.676455696202532,
+      "grad_norm": 0.2609367814625271,
+      "learning_rate": 5.005400145974704e-07,
+      "loss": 0.3008,
+      "step": 1151
+    },
+    {
+      "epoch": 4.680506329113924,
+      "grad_norm": 0.23404657165426968,
+      "learning_rate": 4.879997784979562e-07,
+      "loss": 0.2827,
+      "step": 1152
+    },
+    {
+      "epoch": 4.684556962025317,
+      "grad_norm": 0.23079201199089633,
+      "learning_rate": 4.7561668936580984e-07,
+      "loss": 0.3314,
+      "step": 1153
+    },
+    {
+      "epoch": 4.688607594936709,
+      "grad_norm": 0.2433165569210696,
+      "learning_rate": 4.6339084693272306e-07,
+      "loss": 0.3058,
+      "step": 1154
+    },
+    {
+      "epoch": 4.692658227848101,
+      "grad_norm": 0.23939400562990004,
+      "learning_rate": 4.5132234966395847e-07,
+      "loss": 0.3002,
+      "step": 1155
+    },
+    {
+      "epoch": 4.696708860759494,
+      "grad_norm": 0.2357134320097386,
+      "learning_rate": 4.3941129475752795e-07,
+      "loss": 0.3401,
+      "step": 1156
+    },
+    {
+      "epoch": 4.700759493670886,
+      "grad_norm": 0.2332278640940389,
+      "learning_rate": 4.27657778143431e-07,
+      "loss": 0.3542,
+      "step": 1157
+    },
+    {
+      "epoch": 4.704810126582278,
+      "grad_norm": 0.2378434808629925,
+      "learning_rate": 4.1606189448287757e-07,
+      "loss": 0.3044,
+      "step": 1158
+    },
+    {
+      "epoch": 4.708860759493671,
+      "grad_norm": 0.2308059871402735,
+      "learning_rate": 4.046237371675177e-07,
+      "loss": 0.3339,
+      "step": 1159
+    },
+    {
+      "epoch": 4.712911392405063,
+      "grad_norm": 0.23497408676495904,
+      "learning_rate": 3.9334339831869963e-07,
+      "loss": 0.3471,
+      "step": 1160
+    },
+    {
+      "epoch": 4.716962025316455,
+      "grad_norm": 0.24343370598699862,
+      "learning_rate": 3.8222096878671955e-07,
+      "loss": 0.3254,
+      "step": 1161
+    },
+    {
+      "epoch": 4.721012658227848,
+      "grad_norm": 0.2374501617473106,
+      "learning_rate": 3.7125653815009545e-07,
+      "loss": 0.3176,
+      "step": 1162
+    },
+    {
+      "epoch": 4.72506329113924,
+      "grad_norm": 0.23505426510790256,
+      "learning_rate": 3.6045019471484974e-07,
+      "loss": 0.3121,
+      "step": 1163
+    },
+    {
+      "epoch": 4.729113924050633,
+      "grad_norm": 0.2422806702354837,
+      "learning_rate": 3.498020255137813e-07,
+      "loss": 0.3297,
+      "step": 1164
+    },
+    {
+      "epoch": 4.733164556962025,
+      "grad_norm": 0.22817645850601023,
+      "learning_rate": 3.393121163057811e-07,
+      "loss": 0.3327,
+      "step": 1165
+    },
+    {
+      "epoch": 4.737215189873417,
+      "grad_norm": 0.2333351947581364,
+      "learning_rate": 3.289805515751399e-07,
+      "loss": 0.3139,
+      "step": 1166
+    },
+    {
+      "epoch": 4.74126582278481,
+      "grad_norm": 0.22807657112766413,
+      "learning_rate": 3.188074145308573e-07,
+      "loss": 0.3079,
+      "step": 1167
+    },
+    {
+      "epoch": 4.745316455696202,
+      "grad_norm": 0.24912550644498896,
+      "learning_rate": 3.087927871059804e-07,
+      "loss": 0.2905,
+      "step": 1168
+    },
+    {
+      "epoch": 4.749367088607595,
+      "grad_norm": 0.23261505818805092,
+      "learning_rate": 2.989367499569418e-07,
+      "loss": 0.3408,
+      "step": 1169
+    },
+    {
+      "epoch": 4.7534177215189874,
+      "grad_norm": 0.253857310302098,
+      "learning_rate": 2.8923938246290917e-07,
+      "loss": 0.3145,
+      "step": 1170
+    },
+    {
+      "epoch": 4.7574683544303795,
+      "grad_norm": 0.23281271464627715,
+      "learning_rate": 2.7970076272514804e-07,
+      "loss": 0.3125,
+      "step": 1171
+    },
+    {
+      "epoch": 4.7615189873417725,
+      "grad_norm": 0.23890532909887577,
+      "learning_rate": 2.703209675663887e-07,
+      "loss": 0.3225,
+      "step": 1172
+    },
+    {
+      "epoch": 4.7655696202531646,
+      "grad_norm": 0.2813977547121189,
+      "learning_rate": 2.6110007253021374e-07,
+      "loss": 0.2689,
+      "step": 1173
+    },
+    {
+      "epoch": 4.769620253164557,
+      "grad_norm": 0.24563055220549226,
+      "learning_rate": 2.520381518804471e-07,
+      "loss": 0.3281,
+      "step": 1174
+    },
+    {
+      "epoch": 4.77367088607595,
+      "grad_norm": 0.24509696129940067,
+      "learning_rate": 2.4313527860054585e-07,
+      "loss": 0.3042,
+      "step": 1175
+    },
+    {
+      "epoch": 4.777721518987342,
+      "grad_norm": 0.24241528967489484,
+      "learning_rate": 2.343915243930317e-07,
+      "loss": 0.3492,
+      "step": 1176
+    },
+    {
+      "epoch": 4.781772151898734,
+      "grad_norm": 0.2333236168574974,
+      "learning_rate": 2.2580695967889367e-07,
+      "loss": 0.3294,
+      "step": 1177
+    },
+    {
+      "epoch": 4.785822784810127,
+      "grad_norm": 0.23983980415028053,
+      "learning_rate": 2.1738165359704189e-07,
+      "loss": 0.3269,
+      "step": 1178
+    },
+    {
+      "epoch": 4.789873417721519,
+      "grad_norm": 0.2471757727085521,
+      "learning_rate": 2.0911567400373257e-07,
+      "loss": 0.3198,
+      "step": 1179
+    },
+    {
+      "epoch": 4.793924050632912,
+      "grad_norm": 0.24760932029330943,
+      "learning_rate": 2.0100908747202607e-07,
+      "loss": 0.2966,
+      "step": 1180
+    },
+    {
+      "epoch": 4.797974683544304,
+      "grad_norm": 0.24443502266143868,
+      "learning_rate": 1.9306195929125638e-07,
+      "loss": 0.2941,
+      "step": 1181
+    },
+    {
+      "epoch": 4.802025316455696,
+      "grad_norm": 0.2338516480284218,
+      "learning_rate": 1.8527435346650247e-07,
+      "loss": 0.3096,
+      "step": 1182
+    },
+    {
+      "epoch": 4.806075949367089,
+      "grad_norm": 0.24398305556685643,
+      "learning_rate": 1.7764633271807108e-07,
+      "loss": 0.3279,
+      "step": 1183
+    },
+    {
+      "epoch": 4.810126582278481,
+      "grad_norm": 0.2296320348878043,
+      "learning_rate": 1.7017795848099262e-07,
+      "loss": 0.315,
+      "step": 1184
+    },
+    {
+      "epoch": 4.814177215189874,
+      "grad_norm": 0.2464231956370064,
+      "learning_rate": 1.6286929090452596e-07,
+      "loss": 0.3084,
+      "step": 1185
+    },
+    {
+      "epoch": 4.818227848101266,
+      "grad_norm": 0.24405191658115793,
+      "learning_rate": 1.557203888516745e-07,
+      "loss": 0.2966,
+      "step": 1186
+    },
+    {
+      "epoch": 4.822278481012658,
+      "grad_norm": 0.22393440934214012,
+      "learning_rate": 1.487313098987131e-07,
+      "loss": 0.3168,
+      "step": 1187
+    },
+    {
+      "epoch": 4.826329113924051,
+      "grad_norm": 0.2530041776190803,
+      "learning_rate": 1.4190211033472402e-07,
+      "loss": 0.3106,
+      "step": 1188
+    },
+    {
+      "epoch": 4.830379746835443,
+      "grad_norm": 0.24788678665660496,
+      "learning_rate": 1.3523284516113955e-07,
+      "loss": 0.2873,
+      "step": 1189
+    },
+    {
+      "epoch": 4.834430379746835,
+      "grad_norm": 0.2332099637153288,
+      "learning_rate": 1.2872356809130682e-07,
+      "loss": 0.3172,
+      "step": 1190
+    },
+    {
+      "epoch": 4.838481012658228,
+      "grad_norm": 0.24481522743683518,
+      "learning_rate": 1.2237433155004807e-07,
+      "loss": 0.3011,
+      "step": 1191
+    },
+    {
+      "epoch": 4.84253164556962,
+      "grad_norm": 0.24010160868602262,
+      "learning_rate": 1.1618518667323886e-07,
+      "loss": 0.3116,
+      "step": 1192
+    },
+    {
+      "epoch": 4.846582278481012,
+      "grad_norm": 0.23377559428540196,
+      "learning_rate": 1.1015618330740385e-07,
+      "loss": 0.3226,
+      "step": 1193
+    },
+    {
+      "epoch": 4.850632911392405,
+      "grad_norm": 0.235747990886001,
+      "learning_rate": 1.042873700093061e-07,
+      "loss": 0.3516,
+      "step": 1194
+    },
+    {
+      "epoch": 4.854683544303797,
+      "grad_norm": 0.2351865621942812,
+      "learning_rate": 9.857879404556291e-08,
+      "loss": 0.315,
+      "step": 1195
+    },
+    {
+      "epoch": 4.85873417721519,
+      "grad_norm": 0.2504742402820027,
+      "learning_rate": 9.303050139225722e-08,
+      "loss": 0.2806,
+      "step": 1196
+    },
+    {
+      "epoch": 4.862784810126582,
+      "grad_norm": 0.24340671224628196,
+      "learning_rate": 8.76425367345779e-08,
+      "loss": 0.3165,
+      "step": 1197
+    },
+    {
+      "epoch": 4.866835443037974,
+      "grad_norm": 0.23108263010368443,
+      "learning_rate": 8.241494346644897e-08,
+      "loss": 0.3207,
+      "step": 1198
+    },
+    {
+      "epoch": 4.870886075949367,
+      "grad_norm": 0.23733900158440496,
+      "learning_rate": 7.734776369019204e-08,
+      "loss": 0.3268,
+      "step": 1199
+    },
+    {
+      "epoch": 4.874936708860759,
+      "grad_norm": 0.23481853333959044,
+      "learning_rate": 7.244103821617332e-08,
+      "loss": 0.3292,
+      "step": 1200
+    },
+    {
+      "epoch": 4.878987341772152,
+      "grad_norm": 0.2477778642392466,
+      "learning_rate": 6.769480656248606e-08,
+      "loss": 0.314,
+      "step": 1201
+    },
+    {
+      "epoch": 4.883037974683544,
+      "grad_norm": 0.2335570997169446,
+      "learning_rate": 6.310910695462635e-08,
+      "loss": 0.3297,
+      "step": 1202
+    },
+    {
+      "epoch": 4.8870886075949365,
+      "grad_norm": 0.22738087685090824,
+      "learning_rate": 5.8683976325191185e-08,
+      "loss": 0.3318,
+      "step": 1203
+    },
+    {
+      "epoch": 4.891139240506329,
+      "grad_norm": 0.24570333922126789,
+      "learning_rate": 5.4419450313571984e-08,
+      "loss": 0.305,
+      "step": 1204
+    },
+    {
+      "epoch": 4.8951898734177215,
+      "grad_norm": 0.2611871837401194,
+      "learning_rate": 5.031556326567488e-08,
+      "loss": 0.3209,
+      "step": 1205
+    },
+    {
+      "epoch": 4.899240506329114,
+      "grad_norm": 0.23529491255971383,
+      "learning_rate": 4.637234823364312e-08,
+      "loss": 0.3117,
+      "step": 1206
+    },
+    {
+      "epoch": 4.9032911392405065,
+      "grad_norm": 0.23525949119902875,
+      "learning_rate": 4.258983697558838e-08,
+      "loss": 0.322,
+      "step": 1207
+    },
+    {
+      "epoch": 4.907341772151899,
+      "grad_norm": 0.259169055264744,
+      "learning_rate": 3.896805995533548e-08,
+      "loss": 0.2926,
+      "step": 1208
+    },
+    {
+      "epoch": 4.911392405063291,
+      "grad_norm": 0.2318707442264157,
+      "learning_rate": 3.550704634218028e-08,
+      "loss": 0.326,
+      "step": 1209
+    },
+    {
+      "epoch": 4.915443037974684,
+      "grad_norm": 0.23988939050075136,
+      "learning_rate": 3.2206824010647676e-08,
+      "loss": 0.3042,
+      "step": 1210
+    },
+    {
+      "epoch": 4.919493670886076,
+      "grad_norm": 0.2358209828619115,
+      "learning_rate": 2.9067419540278476e-08,
+      "loss": 0.3087,
+      "step": 1211
+    },
+    {
+      "epoch": 4.923544303797469,
+      "grad_norm": 0.22802965927365484,
+      "learning_rate": 2.6088858215400638e-08,
+      "loss": 0.327,
+      "step": 1212
+    },
+    {
+      "epoch": 4.927594936708861,
+      "grad_norm": 0.24263457001021801,
+      "learning_rate": 2.3271164024940564e-08,
+      "loss": 0.335,
+      "step": 1213
+    },
+    {
+      "epoch": 4.931645569620253,
+      "grad_norm": 0.23627161130990684,
+      "learning_rate": 2.061435966221881e-08,
+      "loss": 0.3169,
+      "step": 1214
+    },
+    {
+      "epoch": 4.935696202531646,
+      "grad_norm": 0.22592539335566103,
+      "learning_rate": 1.811846652477245e-08,
+      "loss": 0.3062,
+      "step": 1215
+    },
+    {
+      "epoch": 4.939746835443038,
+      "grad_norm": 0.23594970424889608,
+      "learning_rate": 1.5783504714184106e-08,
+      "loss": 0.3241,
+      "step": 1216
+    },
+    {
+      "epoch": 4.943797468354431,
+      "grad_norm": 0.2368129574035504,
+      "learning_rate": 1.360949303591097e-08,
+      "loss": 0.3097,
+      "step": 1217
+    },
+    {
+      "epoch": 4.947848101265823,
+      "grad_norm": 0.22643851360512568,
+      "learning_rate": 1.1596448999144916e-08,
+      "loss": 0.3286,
+      "step": 1218
+    },
+    {
+      "epoch": 4.951898734177215,
+      "grad_norm": 0.24346323509784984,
+      "learning_rate": 9.744388816668172e-09,
+      "loss": 0.3436,
+      "step": 1219
+    },
+    {
+      "epoch": 4.955949367088608,
+      "grad_norm": 0.23700974097229843,
+      "learning_rate": 8.05332740472009e-09,
+      "loss": 0.3239,
+      "step": 1220
+    },
+    {
+      "epoch": 4.96,
+      "grad_norm": 0.22786168868480058,
+      "learning_rate": 6.523278382872811e-09,
+      "loss": 0.3178,
+      "step": 1221
+    },
+    {
+      "epoch": 4.964050632911392,
+      "grad_norm": 0.222380251636523,
+      "learning_rate": 5.15425407393133e-09,
+      "loss": 0.3198,
+      "step": 1222
+    },
+    {
+      "epoch": 4.968101265822785,
+      "grad_norm": 0.24004804350271752,
+      "learning_rate": 3.94626550383137e-09,
+      "loss": 0.2969,
+      "step": 1223
+    },
+    {
+      "epoch": 4.972151898734177,
+      "grad_norm": 0.24710756175884965,
+      "learning_rate": 2.899322401546112e-09,
+      "loss": 0.3176,
+      "step": 1224
+    },
+    {
+      "epoch": 4.976202531645569,
+      "grad_norm": 0.24129996921922164,
+      "learning_rate": 2.013433199010706e-09,
+      "loss": 0.3084,
+      "step": 1225
+    },
+    {
+      "epoch": 4.980253164556962,
+      "grad_norm": 0.23801628116095974,
+      "learning_rate": 1.2886050310556563e-09,
+      "loss": 0.3075,
+      "step": 1226
+    },
+    {
+      "epoch": 4.984303797468354,
+      "grad_norm": 0.23497337170476226,
+      "learning_rate": 7.248437353468695e-10,
+      "loss": 0.3191,
+      "step": 1227
+    },
+    {
+      "epoch": 4.988354430379747,
+      "grad_norm": 0.23161585091788064,
+      "learning_rate": 3.221538523412449e-10,
+      "loss": 0.3276,
+      "step": 1228
+    },
+    {
+      "epoch": 4.992405063291139,
+      "grad_norm": 0.2267458057381176,
+      "learning_rate": 8.053862524670663e-11,
+      "loss": 0.309,
+      "step": 1229
+    },
+    {
+      "epoch": 4.996455696202531,
+      "grad_norm": 0.23311525308740766,
+      "learning_rate": 0.0,
+      "loss": 0.3261,
+      "step": 1230
+    },
+    {
+      "epoch": 4.996455696202531,
+      "step": 1230,
+      "total_flos": 3.8606127231439667e+18,
+      "train_loss": 0.06271887501565422,
+      "train_runtime": 10791.9406,
+      "train_samples_per_second": 14.641,
+      "train_steps_per_second": 0.114
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 1230,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 3.8606127231439667e+18,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}