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": 4.988354430379747,
+  "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": 7.929403323450866,
+      "learning_rate": 3.2520325203252037e-07,
+      "loss": 0.8181,
+      "step": 1
+    },
+    {
+      "epoch": 0.00810126582278481,
+      "grad_norm": 8.068385469828149,
+      "learning_rate": 6.504065040650407e-07,
+      "loss": 0.8239,
+      "step": 2
+    },
+    {
+      "epoch": 0.012151898734177215,
+      "grad_norm": 7.866788314962798,
+      "learning_rate": 9.75609756097561e-07,
+      "loss": 0.8192,
+      "step": 3
+    },
+    {
+      "epoch": 0.01620253164556962,
+      "grad_norm": 8.000703467850785,
+      "learning_rate": 1.3008130081300815e-06,
+      "loss": 0.8298,
+      "step": 4
+    },
+    {
+      "epoch": 0.020253164556962026,
+      "grad_norm": 7.4514753864364405,
+      "learning_rate": 1.6260162601626018e-06,
+      "loss": 0.8072,
+      "step": 5
+    },
+    {
+      "epoch": 0.02430379746835443,
+      "grad_norm": 6.832472743851544,
+      "learning_rate": 1.951219512195122e-06,
+      "loss": 0.7757,
+      "step": 6
+    },
+    {
+      "epoch": 0.028354430379746835,
+      "grad_norm": 5.689704956615217,
+      "learning_rate": 2.2764227642276426e-06,
+      "loss": 0.7452,
+      "step": 7
+    },
+    {
+      "epoch": 0.03240506329113924,
+      "grad_norm": 5.303099217946958,
+      "learning_rate": 2.601626016260163e-06,
+      "loss": 0.7312,
+      "step": 8
+    },
+    {
+      "epoch": 0.03645569620253165,
+      "grad_norm": 3.2230208171330603,
+      "learning_rate": 2.926829268292683e-06,
+      "loss": 0.6808,
+      "step": 9
+    },
+    {
+      "epoch": 0.04050632911392405,
+      "grad_norm": 2.8962145107561756,
+      "learning_rate": 3.2520325203252037e-06,
+      "loss": 0.6806,
+      "step": 10
+    },
+    {
+      "epoch": 0.044556962025316456,
+      "grad_norm": 2.653229643551163,
+      "learning_rate": 3.577235772357724e-06,
+      "loss": 0.665,
+      "step": 11
+    },
+    {
+      "epoch": 0.04860759493670886,
+      "grad_norm": 4.035840992622092,
+      "learning_rate": 3.902439024390244e-06,
+      "loss": 0.6439,
+      "step": 12
+    },
+    {
+      "epoch": 0.052658227848101265,
+      "grad_norm": 3.843526847507736,
+      "learning_rate": 4.227642276422765e-06,
+      "loss": 0.6187,
+      "step": 13
+    },
+    {
+      "epoch": 0.05670886075949367,
+      "grad_norm": 3.795009674035192,
+      "learning_rate": 4.552845528455285e-06,
+      "loss": 0.6223,
+      "step": 14
+    },
+    {
+      "epoch": 0.060759493670886074,
+      "grad_norm": 3.4848594829111805,
+      "learning_rate": 4.8780487804878055e-06,
+      "loss": 0.5871,
+      "step": 15
+    },
+    {
+      "epoch": 0.06481012658227848,
+      "grad_norm": 3.180880124341412,
+      "learning_rate": 5.203252032520326e-06,
+      "loss": 0.5785,
+      "step": 16
+    },
+    {
+      "epoch": 0.06886075949367089,
+      "grad_norm": 2.903601174944415,
+      "learning_rate": 5.528455284552846e-06,
+      "loss": 0.5572,
+      "step": 17
+    },
+    {
+      "epoch": 0.0729113924050633,
+      "grad_norm": 2.240422369423512,
+      "learning_rate": 5.853658536585366e-06,
+      "loss": 0.5338,
+      "step": 18
+    },
+    {
+      "epoch": 0.0769620253164557,
+      "grad_norm": 2.0917670835066757,
+      "learning_rate": 6.178861788617887e-06,
+      "loss": 0.5067,
+      "step": 19
+    },
+    {
+      "epoch": 0.0810126582278481,
+      "grad_norm": 1.533966149318591,
+      "learning_rate": 6.504065040650407e-06,
+      "loss": 0.5038,
+      "step": 20
+    },
+    {
+      "epoch": 0.08506329113924051,
+      "grad_norm": 1.447180902987296,
+      "learning_rate": 6.829268292682928e-06,
+      "loss": 0.5108,
+      "step": 21
+    },
+    {
+      "epoch": 0.08911392405063291,
+      "grad_norm": 1.559124934439477,
+      "learning_rate": 7.154471544715448e-06,
+      "loss": 0.4919,
+      "step": 22
+    },
+    {
+      "epoch": 0.09316455696202532,
+      "grad_norm": 1.5897097998475778,
+      "learning_rate": 7.4796747967479676e-06,
+      "loss": 0.4939,
+      "step": 23
+    },
+    {
+      "epoch": 0.09721518987341772,
+      "grad_norm": 1.2500975851053497,
+      "learning_rate": 7.804878048780489e-06,
+      "loss": 0.4795,
+      "step": 24
+    },
+    {
+      "epoch": 0.10126582278481013,
+      "grad_norm": 1.0045399266912127,
+      "learning_rate": 8.130081300813009e-06,
+      "loss": 0.4799,
+      "step": 25
+    },
+    {
+      "epoch": 0.10531645569620253,
+      "grad_norm": 1.0546016067618889,
+      "learning_rate": 8.45528455284553e-06,
+      "loss": 0.474,
+      "step": 26
+    },
+    {
+      "epoch": 0.10936708860759493,
+      "grad_norm": 1.0015413069064079,
+      "learning_rate": 8.78048780487805e-06,
+      "loss": 0.4699,
+      "step": 27
+    },
+    {
+      "epoch": 0.11341772151898734,
+      "grad_norm": 0.8606421929876965,
+      "learning_rate": 9.10569105691057e-06,
+      "loss": 0.4516,
+      "step": 28
+    },
+    {
+      "epoch": 0.11746835443037974,
+      "grad_norm": 0.9419575830150634,
+      "learning_rate": 9.43089430894309e-06,
+      "loss": 0.4682,
+      "step": 29
+    },
+    {
+      "epoch": 0.12151898734177215,
+      "grad_norm": 0.8053949080185298,
+      "learning_rate": 9.756097560975611e-06,
+      "loss": 0.449,
+      "step": 30
+    },
+    {
+      "epoch": 0.12556962025316457,
+      "grad_norm": 0.9255143746771395,
+      "learning_rate": 1.008130081300813e-05,
+      "loss": 0.435,
+      "step": 31
+    },
+    {
+      "epoch": 0.12962025316455697,
+      "grad_norm": 0.8111882763101464,
+      "learning_rate": 1.0406504065040652e-05,
+      "loss": 0.4364,
+      "step": 32
+    },
+    {
+      "epoch": 0.13367088607594937,
+      "grad_norm": 0.7099154427456007,
+      "learning_rate": 1.0731707317073172e-05,
+      "loss": 0.431,
+      "step": 33
+    },
+    {
+      "epoch": 0.13772151898734178,
+      "grad_norm": 0.7389930049124497,
+      "learning_rate": 1.1056910569105692e-05,
+      "loss": 0.4285,
+      "step": 34
+    },
+    {
+      "epoch": 0.14177215189873418,
+      "grad_norm": 0.6379408100830813,
+      "learning_rate": 1.1382113821138213e-05,
+      "loss": 0.4316,
+      "step": 35
+    },
+    {
+      "epoch": 0.1458227848101266,
+      "grad_norm": 0.7531974001415243,
+      "learning_rate": 1.1707317073170731e-05,
+      "loss": 0.4263,
+      "step": 36
+    },
+    {
+      "epoch": 0.149873417721519,
+      "grad_norm": 0.6369856003620411,
+      "learning_rate": 1.2032520325203254e-05,
+      "loss": 0.4387,
+      "step": 37
+    },
+    {
+      "epoch": 0.1539240506329114,
+      "grad_norm": 0.5996689912213102,
+      "learning_rate": 1.2357723577235774e-05,
+      "loss": 0.4222,
+      "step": 38
+    },
+    {
+      "epoch": 0.1579746835443038,
+      "grad_norm": 0.6807876365923533,
+      "learning_rate": 1.2682926829268294e-05,
+      "loss": 0.4016,
+      "step": 39
+    },
+    {
+      "epoch": 0.1620253164556962,
+      "grad_norm": 0.5763951318788307,
+      "learning_rate": 1.3008130081300815e-05,
+      "loss": 0.4241,
+      "step": 40
+    },
+    {
+      "epoch": 0.1660759493670886,
+      "grad_norm": 0.5690729760215962,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.4136,
+      "step": 41
+    },
+    {
+      "epoch": 0.17012658227848101,
+      "grad_norm": 0.6301468113398263,
+      "learning_rate": 1.3658536585365855e-05,
+      "loss": 0.4202,
+      "step": 42
+    },
+    {
+      "epoch": 0.17417721518987342,
+      "grad_norm": 0.5934870316452963,
+      "learning_rate": 1.3983739837398376e-05,
+      "loss": 0.4112,
+      "step": 43
+    },
+    {
+      "epoch": 0.17822784810126582,
+      "grad_norm": 0.6036429447061717,
+      "learning_rate": 1.4308943089430896e-05,
+      "loss": 0.4199,
+      "step": 44
+    },
+    {
+      "epoch": 0.18227848101265823,
+      "grad_norm": 0.577236748031787,
+      "learning_rate": 1.4634146341463415e-05,
+      "loss": 0.4209,
+      "step": 45
+    },
+    {
+      "epoch": 0.18632911392405063,
+      "grad_norm": 0.6968522451809518,
+      "learning_rate": 1.4959349593495935e-05,
+      "loss": 0.4291,
+      "step": 46
+    },
+    {
+      "epoch": 0.19037974683544304,
+      "grad_norm": 0.49754794784424305,
+      "learning_rate": 1.528455284552846e-05,
+      "loss": 0.4132,
+      "step": 47
+    },
+    {
+      "epoch": 0.19443037974683544,
+      "grad_norm": 0.7278618733307458,
+      "learning_rate": 1.5609756097560978e-05,
+      "loss": 0.4155,
+      "step": 48
+    },
+    {
+      "epoch": 0.19848101265822785,
+      "grad_norm": 0.57109712722284,
+      "learning_rate": 1.5934959349593496e-05,
+      "loss": 0.4071,
+      "step": 49
+    },
+    {
+      "epoch": 0.20253164556962025,
+      "grad_norm": 0.5567670146280373,
+      "learning_rate": 1.6260162601626018e-05,
+      "loss": 0.4015,
+      "step": 50
+    },
+    {
+      "epoch": 0.20658227848101265,
+      "grad_norm": 0.5417382185689416,
+      "learning_rate": 1.6585365853658537e-05,
+      "loss": 0.3799,
+      "step": 51
+    },
+    {
+      "epoch": 0.21063291139240506,
+      "grad_norm": 0.6255000493519758,
+      "learning_rate": 1.691056910569106e-05,
+      "loss": 0.4105,
+      "step": 52
+    },
+    {
+      "epoch": 0.21468354430379746,
+      "grad_norm": 0.5239445507155484,
+      "learning_rate": 1.7235772357723578e-05,
+      "loss": 0.3789,
+      "step": 53
+    },
+    {
+      "epoch": 0.21873417721518987,
+      "grad_norm": 0.5524318549074562,
+      "learning_rate": 1.75609756097561e-05,
+      "loss": 0.4144,
+      "step": 54
+    },
+    {
+      "epoch": 0.22278481012658227,
+      "grad_norm": 0.5898597463295152,
+      "learning_rate": 1.788617886178862e-05,
+      "loss": 0.3817,
+      "step": 55
+    },
+    {
+      "epoch": 0.22683544303797468,
+      "grad_norm": 0.631343047405675,
+      "learning_rate": 1.821138211382114e-05,
+      "loss": 0.3953,
+      "step": 56
+    },
+    {
+      "epoch": 0.23088607594936708,
+      "grad_norm": 0.6292253928674758,
+      "learning_rate": 1.8536585365853663e-05,
+      "loss": 0.4095,
+      "step": 57
+    },
+    {
+      "epoch": 0.23493670886075949,
+      "grad_norm": 0.7289332762025695,
+      "learning_rate": 1.886178861788618e-05,
+      "loss": 0.3985,
+      "step": 58
+    },
+    {
+      "epoch": 0.2389873417721519,
+      "grad_norm": 0.7389853583556427,
+      "learning_rate": 1.91869918699187e-05,
+      "loss": 0.38,
+      "step": 59
+    },
+    {
+      "epoch": 0.2430379746835443,
+      "grad_norm": 0.6276159832551357,
+      "learning_rate": 1.9512195121951222e-05,
+      "loss": 0.3858,
+      "step": 60
+    },
+    {
+      "epoch": 0.2470886075949367,
+      "grad_norm": 0.7305830182611137,
+      "learning_rate": 1.983739837398374e-05,
+      "loss": 0.379,
+      "step": 61
+    },
+    {
+      "epoch": 0.25113924050632913,
+      "grad_norm": 0.6618120588382487,
+      "learning_rate": 2.016260162601626e-05,
+      "loss": 0.3843,
+      "step": 62
+    },
+    {
+      "epoch": 0.25518987341772154,
+      "grad_norm": 0.6764630091230303,
+      "learning_rate": 2.048780487804878e-05,
+      "loss": 0.3972,
+      "step": 63
+    },
+    {
+      "epoch": 0.25924050632911394,
+      "grad_norm": 0.8152954944378757,
+      "learning_rate": 2.0813008130081303e-05,
+      "loss": 0.385,
+      "step": 64
+    },
+    {
+      "epoch": 0.26329113924050634,
+      "grad_norm": 0.7337677699232661,
+      "learning_rate": 2.1138211382113822e-05,
+      "loss": 0.3946,
+      "step": 65
+    },
+    {
+      "epoch": 0.26734177215189875,
+      "grad_norm": 0.7590248680716263,
+      "learning_rate": 2.1463414634146344e-05,
+      "loss": 0.3784,
+      "step": 66
+    },
+    {
+      "epoch": 0.27139240506329115,
+      "grad_norm": 0.8121748890019888,
+      "learning_rate": 2.1788617886178863e-05,
+      "loss": 0.3853,
+      "step": 67
+    },
+    {
+      "epoch": 0.27544303797468356,
+      "grad_norm": 0.7438242682714056,
+      "learning_rate": 2.2113821138211385e-05,
+      "loss": 0.3748,
+      "step": 68
+    },
+    {
+      "epoch": 0.27949367088607596,
+      "grad_norm": 0.6674823945996519,
+      "learning_rate": 2.2439024390243907e-05,
+      "loss": 0.3779,
+      "step": 69
+    },
+    {
+      "epoch": 0.28354430379746837,
+      "grad_norm": 1.0270496337932657,
+      "learning_rate": 2.2764227642276426e-05,
+      "loss": 0.384,
+      "step": 70
+    },
+    {
+      "epoch": 0.28759493670886077,
+      "grad_norm": 0.6855012774896265,
+      "learning_rate": 2.3089430894308948e-05,
+      "loss": 0.3868,
+      "step": 71
+    },
+    {
+      "epoch": 0.2916455696202532,
+      "grad_norm": 0.7703890138720862,
+      "learning_rate": 2.3414634146341463e-05,
+      "loss": 0.3805,
+      "step": 72
+    },
+    {
+      "epoch": 0.2956962025316456,
+      "grad_norm": 0.7845658223895129,
+      "learning_rate": 2.3739837398373985e-05,
+      "loss": 0.3849,
+      "step": 73
+    },
+    {
+      "epoch": 0.299746835443038,
+      "grad_norm": 0.7100777997126864,
+      "learning_rate": 2.4065040650406507e-05,
+      "loss": 0.4091,
+      "step": 74
+    },
+    {
+      "epoch": 0.3037974683544304,
+      "grad_norm": 0.783534858228256,
+      "learning_rate": 2.4390243902439026e-05,
+      "loss": 0.3705,
+      "step": 75
+    },
+    {
+      "epoch": 0.3078481012658228,
+      "grad_norm": 0.6635900701515034,
+      "learning_rate": 2.4715447154471548e-05,
+      "loss": 0.388,
+      "step": 76
+    },
+    {
+      "epoch": 0.3118987341772152,
+      "grad_norm": 0.7821024137299016,
+      "learning_rate": 2.5040650406504066e-05,
+      "loss": 0.3802,
+      "step": 77
+    },
+    {
+      "epoch": 0.3159493670886076,
+      "grad_norm": 0.6805344250311997,
+      "learning_rate": 2.536585365853659e-05,
+      "loss": 0.3834,
+      "step": 78
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.8972185143873374,
+      "learning_rate": 2.569105691056911e-05,
+      "loss": 0.3757,
+      "step": 79
+    },
+    {
+      "epoch": 0.3240506329113924,
+      "grad_norm": 0.8378178852458943,
+      "learning_rate": 2.601626016260163e-05,
+      "loss": 0.3853,
+      "step": 80
+    },
+    {
+      "epoch": 0.3281012658227848,
+      "grad_norm": 0.6889715691982597,
+      "learning_rate": 2.634146341463415e-05,
+      "loss": 0.3795,
+      "step": 81
+    },
+    {
+      "epoch": 0.3321518987341772,
+      "grad_norm": 0.6839448275955223,
+      "learning_rate": 2.6666666666666667e-05,
+      "loss": 0.3897,
+      "step": 82
+    },
+    {
+      "epoch": 0.3362025316455696,
+      "grad_norm": 0.72703877548494,
+      "learning_rate": 2.699186991869919e-05,
+      "loss": 0.3777,
+      "step": 83
+    },
+    {
+      "epoch": 0.34025316455696203,
+      "grad_norm": 0.8651511907162992,
+      "learning_rate": 2.731707317073171e-05,
+      "loss": 0.375,
+      "step": 84
+    },
+    {
+      "epoch": 0.34430379746835443,
+      "grad_norm": 0.6308269742063808,
+      "learning_rate": 2.764227642276423e-05,
+      "loss": 0.3775,
+      "step": 85
+    },
+    {
+      "epoch": 0.34835443037974684,
+      "grad_norm": 0.7190618734589886,
+      "learning_rate": 2.796747967479675e-05,
+      "loss": 0.39,
+      "step": 86
+    },
+    {
+      "epoch": 0.35240506329113924,
+      "grad_norm": 0.5854781201574748,
+      "learning_rate": 2.829268292682927e-05,
+      "loss": 0.3912,
+      "step": 87
+    },
+    {
+      "epoch": 0.35645569620253165,
+      "grad_norm": 0.6558827892334383,
+      "learning_rate": 2.8617886178861792e-05,
+      "loss": 0.3611,
+      "step": 88
+    },
+    {
+      "epoch": 0.36050632911392405,
+      "grad_norm": 0.6756238378276024,
+      "learning_rate": 2.8943089430894314e-05,
+      "loss": 0.3657,
+      "step": 89
+    },
+    {
+      "epoch": 0.36455696202531646,
+      "grad_norm": 0.6178323621390194,
+      "learning_rate": 2.926829268292683e-05,
+      "loss": 0.3876,
+      "step": 90
+    },
+    {
+      "epoch": 0.36860759493670886,
+      "grad_norm": 0.6311266321891061,
+      "learning_rate": 2.959349593495935e-05,
+      "loss": 0.368,
+      "step": 91
+    },
+    {
+      "epoch": 0.37265822784810126,
+      "grad_norm": 0.7379396806997098,
+      "learning_rate": 2.991869918699187e-05,
+      "loss": 0.3668,
+      "step": 92
+    },
+    {
+      "epoch": 0.37670886075949367,
+      "grad_norm": 0.6960440352323919,
+      "learning_rate": 3.0243902439024392e-05,
+      "loss": 0.3708,
+      "step": 93
+    },
+    {
+      "epoch": 0.3807594936708861,
+      "grad_norm": 0.8121170429697133,
+      "learning_rate": 3.056910569105692e-05,
+      "loss": 0.3805,
+      "step": 94
+    },
+    {
+      "epoch": 0.3848101265822785,
+      "grad_norm": 0.7646124203220518,
+      "learning_rate": 3.089430894308943e-05,
+      "loss": 0.3867,
+      "step": 95
+    },
+    {
+      "epoch": 0.3888607594936709,
+      "grad_norm": 0.7709011685168649,
+      "learning_rate": 3.1219512195121955e-05,
+      "loss": 0.3667,
+      "step": 96
+    },
+    {
+      "epoch": 0.3929113924050633,
+      "grad_norm": 0.7731520461616286,
+      "learning_rate": 3.154471544715447e-05,
+      "loss": 0.3784,
+      "step": 97
+    },
+    {
+      "epoch": 0.3969620253164557,
+      "grad_norm": 0.696029326947107,
+      "learning_rate": 3.186991869918699e-05,
+      "loss": 0.3708,
+      "step": 98
+    },
+    {
+      "epoch": 0.4010126582278481,
+      "grad_norm": 0.7490657138393501,
+      "learning_rate": 3.2195121951219514e-05,
+      "loss": 0.3714,
+      "step": 99
+    },
+    {
+      "epoch": 0.4050632911392405,
+      "grad_norm": 0.5129550466531104,
+      "learning_rate": 3.2520325203252037e-05,
+      "loss": 0.3762,
+      "step": 100
+    },
+    {
+      "epoch": 0.4091139240506329,
+      "grad_norm": 0.7264867653075008,
+      "learning_rate": 3.284552845528456e-05,
+      "loss": 0.3879,
+      "step": 101
+    },
+    {
+      "epoch": 0.4131645569620253,
+      "grad_norm": 0.5908546205636361,
+      "learning_rate": 3.3170731707317074e-05,
+      "loss": 0.3817,
+      "step": 102
+    },
+    {
+      "epoch": 0.4172151898734177,
+      "grad_norm": 0.6951199334440988,
+      "learning_rate": 3.3495934959349596e-05,
+      "loss": 0.3795,
+      "step": 103
+    },
+    {
+      "epoch": 0.4212658227848101,
+      "grad_norm": 0.6989493772454484,
+      "learning_rate": 3.382113821138212e-05,
+      "loss": 0.3823,
+      "step": 104
+    },
+    {
+      "epoch": 0.4253164556962025,
+      "grad_norm": 0.7522110944102618,
+      "learning_rate": 3.414634146341463e-05,
+      "loss": 0.3615,
+      "step": 105
+    },
+    {
+      "epoch": 0.4293670886075949,
+      "grad_norm": 0.7708569094484375,
+      "learning_rate": 3.4471544715447155e-05,
+      "loss": 0.3753,
+      "step": 106
+    },
+    {
+      "epoch": 0.43341772151898733,
+      "grad_norm": 0.685047376617551,
+      "learning_rate": 3.479674796747968e-05,
+      "loss": 0.3547,
+      "step": 107
+    },
+    {
+      "epoch": 0.43746835443037974,
+      "grad_norm": 0.6726650801757061,
+      "learning_rate": 3.51219512195122e-05,
+      "loss": 0.3793,
+      "step": 108
+    },
+    {
+      "epoch": 0.44151898734177214,
+      "grad_norm": 0.7053039517237943,
+      "learning_rate": 3.544715447154472e-05,
+      "loss": 0.3693,
+      "step": 109
+    },
+    {
+      "epoch": 0.44556962025316454,
+      "grad_norm": 0.6356416910174751,
+      "learning_rate": 3.577235772357724e-05,
+      "loss": 0.3805,
+      "step": 110
+    },
+    {
+      "epoch": 0.44962025316455695,
+      "grad_norm": 0.6802973243635959,
+      "learning_rate": 3.609756097560976e-05,
+      "loss": 0.3838,
+      "step": 111
+    },
+    {
+      "epoch": 0.45367088607594935,
+      "grad_norm": 0.7086114020945177,
+      "learning_rate": 3.642276422764228e-05,
+      "loss": 0.3769,
+      "step": 112
+    },
+    {
+      "epoch": 0.45772151898734176,
+      "grad_norm": 0.6722197550706972,
+      "learning_rate": 3.67479674796748e-05,
+      "loss": 0.3723,
+      "step": 113
+    },
+    {
+      "epoch": 0.46177215189873416,
+      "grad_norm": 0.7972232215230091,
+      "learning_rate": 3.7073170731707325e-05,
+      "loss": 0.3663,
+      "step": 114
+    },
+    {
+      "epoch": 0.46582278481012657,
+      "grad_norm": 0.8080057449998354,
+      "learning_rate": 3.739837398373984e-05,
+      "loss": 0.3867,
+      "step": 115
+    },
+    {
+      "epoch": 0.46987341772151897,
+      "grad_norm": 0.7452077477306999,
+      "learning_rate": 3.772357723577236e-05,
+      "loss": 0.3647,
+      "step": 116
+    },
+    {
+      "epoch": 0.4739240506329114,
+      "grad_norm": 0.8647759250209857,
+      "learning_rate": 3.804878048780488e-05,
+      "loss": 0.3799,
+      "step": 117
+    },
+    {
+      "epoch": 0.4779746835443038,
+      "grad_norm": 1.1351892400568382,
+      "learning_rate": 3.83739837398374e-05,
+      "loss": 0.3668,
+      "step": 118
+    },
+    {
+      "epoch": 0.4820253164556962,
+      "grad_norm": 0.6388175844511421,
+      "learning_rate": 3.869918699186992e-05,
+      "loss": 0.3815,
+      "step": 119
+    },
+    {
+      "epoch": 0.4860759493670886,
+      "grad_norm": 1.0156007918071417,
+      "learning_rate": 3.9024390243902444e-05,
+      "loss": 0.3716,
+      "step": 120
+    },
+    {
+      "epoch": 0.490126582278481,
+      "grad_norm": 0.9283241495079131,
+      "learning_rate": 3.9349593495934966e-05,
+      "loss": 0.3865,
+      "step": 121
+    },
+    {
+      "epoch": 0.4941772151898734,
+      "grad_norm": 0.6151477196936285,
+      "learning_rate": 3.967479674796748e-05,
+      "loss": 0.3832,
+      "step": 122
+    },
+    {
+      "epoch": 0.4982278481012658,
+      "grad_norm": 0.8046575858788497,
+      "learning_rate": 4e-05,
+      "loss": 0.3803,
+      "step": 123
+    },
+    {
+      "epoch": 0.5022784810126583,
+      "grad_norm": 0.7179848838774415,
+      "learning_rate": 3.999991946137476e-05,
+      "loss": 0.3698,
+      "step": 124
+    },
+    {
+      "epoch": 0.5063291139240507,
+      "grad_norm": 0.7750915988170466,
+      "learning_rate": 3.999967784614766e-05,
+      "loss": 0.3769,
+      "step": 125
+    },
+    {
+      "epoch": 0.5103797468354431,
+      "grad_norm": 0.7281794224711874,
+      "learning_rate": 3.9999275156264656e-05,
+      "loss": 0.3857,
+      "step": 126
+    },
+    {
+      "epoch": 0.5144303797468355,
+      "grad_norm": 0.6321201547286188,
+      "learning_rate": 3.999871139496895e-05,
+      "loss": 0.3688,
+      "step": 127
+    },
+    {
+      "epoch": 0.5184810126582279,
+      "grad_norm": 0.7339980169090614,
+      "learning_rate": 3.9997986566800995e-05,
+      "loss": 0.3661,
+      "step": 128
+    },
+    {
+      "epoch": 0.5225316455696203,
+      "grad_norm": 0.6254549451613145,
+      "learning_rate": 3.999710067759846e-05,
+      "loss": 0.3889,
+      "step": 129
+    },
+    {
+      "epoch": 0.5265822784810127,
+      "grad_norm": 0.7870509032985993,
+      "learning_rate": 3.999605373449617e-05,
+      "loss": 0.3615,
+      "step": 130
+    },
+    {
+      "epoch": 0.5306329113924051,
+      "grad_norm": 0.544096495045824,
+      "learning_rate": 3.9994845745926075e-05,
+      "loss": 0.3714,
+      "step": 131
+    },
+    {
+      "epoch": 0.5346835443037975,
+      "grad_norm": 0.6993073364560833,
+      "learning_rate": 3.999347672161713e-05,
+      "loss": 0.3745,
+      "step": 132
+    },
+    {
+      "epoch": 0.5387341772151899,
+      "grad_norm": 0.6404688380915519,
+      "learning_rate": 3.999194667259528e-05,
+      "loss": 0.3728,
+      "step": 133
+    },
+    {
+      "epoch": 0.5427848101265823,
+      "grad_norm": 0.5339110854716141,
+      "learning_rate": 3.999025561118334e-05,
+      "loss": 0.3723,
+      "step": 134
+    },
+    {
+      "epoch": 0.5468354430379747,
+      "grad_norm": 0.7225173510561741,
+      "learning_rate": 3.998840355100086e-05,
+      "loss": 0.3849,
+      "step": 135
+    },
+    {
+      "epoch": 0.5508860759493671,
+      "grad_norm": 0.5106088510081688,
+      "learning_rate": 3.998639050696409e-05,
+      "loss": 0.3692,
+      "step": 136
+    },
+    {
+      "epoch": 0.5549367088607595,
+      "grad_norm": 0.6076484495236667,
+      "learning_rate": 3.998421649528582e-05,
+      "loss": 0.3634,
+      "step": 137
+    },
+    {
+      "epoch": 0.5589873417721519,
+      "grad_norm": 0.6048673185528904,
+      "learning_rate": 3.9981881533475234e-05,
+      "loss": 0.3708,
+      "step": 138
+    },
+    {
+      "epoch": 0.5630379746835443,
+      "grad_norm": 0.583497085745526,
+      "learning_rate": 3.997938564033779e-05,
+      "loss": 0.3667,
+      "step": 139
+    },
+    {
+      "epoch": 0.5670886075949367,
+      "grad_norm": 0.6478876493598718,
+      "learning_rate": 3.9976728835975064e-05,
+      "loss": 0.3962,
+      "step": 140
+    },
+    {
+      "epoch": 0.5711392405063291,
+      "grad_norm": 0.6824634268033497,
+      "learning_rate": 3.9973911141784605e-05,
+      "loss": 0.3751,
+      "step": 141
+    },
+    {
+      "epoch": 0.5751898734177215,
+      "grad_norm": 0.5421623549403886,
+      "learning_rate": 3.997093258045973e-05,
+      "loss": 0.362,
+      "step": 142
+    },
+    {
+      "epoch": 0.579240506329114,
+      "grad_norm": 0.6399414982056923,
+      "learning_rate": 3.996779317598936e-05,
+      "loss": 0.3734,
+      "step": 143
+    },
+    {
+      "epoch": 0.5832911392405064,
+      "grad_norm": 0.6240161735136511,
+      "learning_rate": 3.996449295365782e-05,
+      "loss": 0.3611,
+      "step": 144
+    },
+    {
+      "epoch": 0.5873417721518988,
+      "grad_norm": 0.5688208031618256,
+      "learning_rate": 3.996103194004467e-05,
+      "loss": 0.3774,
+      "step": 145
+    },
+    {
+      "epoch": 0.5913924050632912,
+      "grad_norm": 0.6475851205478446,
+      "learning_rate": 3.995741016302441e-05,
+      "loss": 0.3764,
+      "step": 146
+    },
+    {
+      "epoch": 0.5954430379746836,
+      "grad_norm": 0.6695969633503395,
+      "learning_rate": 3.9953627651766364e-05,
+      "loss": 0.3652,
+      "step": 147
+    },
+    {
+      "epoch": 0.599493670886076,
+      "grad_norm": 0.5495661328011012,
+      "learning_rate": 3.9949684436734325e-05,
+      "loss": 0.3778,
+      "step": 148
+    },
+    {
+      "epoch": 0.6035443037974684,
+      "grad_norm": 0.6377165854820763,
+      "learning_rate": 3.994558054968643e-05,
+      "loss": 0.3681,
+      "step": 149
+    },
+    {
+      "epoch": 0.6075949367088608,
+      "grad_norm": 0.6233492479261898,
+      "learning_rate": 3.994131602367481e-05,
+      "loss": 0.3639,
+      "step": 150
+    },
+    {
+      "epoch": 0.6116455696202532,
+      "grad_norm": 0.5437726257459187,
+      "learning_rate": 3.9936890893045376e-05,
+      "loss": 0.3848,
+      "step": 151
+    },
+    {
+      "epoch": 0.6156962025316456,
+      "grad_norm": 0.5722884627663332,
+      "learning_rate": 3.993230519343752e-05,
+      "loss": 0.3808,
+      "step": 152
+    },
+    {
+      "epoch": 0.619746835443038,
+      "grad_norm": 0.5399886556365208,
+      "learning_rate": 3.992755896178383e-05,
+      "loss": 0.3742,
+      "step": 153
+    },
+    {
+      "epoch": 0.6237974683544304,
+      "grad_norm": 0.49574566493614375,
+      "learning_rate": 3.992265223630981e-05,
+      "loss": 0.3755,
+      "step": 154
+    },
+    {
+      "epoch": 0.6278481012658228,
+      "grad_norm": 0.5600574361020272,
+      "learning_rate": 3.991758505653355e-05,
+      "loss": 0.365,
+      "step": 155
+    },
+    {
+      "epoch": 0.6318987341772152,
+      "grad_norm": 0.46766566075298527,
+      "learning_rate": 3.991235746326543e-05,
+      "loss": 0.3634,
+      "step": 156
+    },
+    {
+      "epoch": 0.6359493670886076,
+      "grad_norm": 0.5349592875553506,
+      "learning_rate": 3.9906969498607745e-05,
+      "loss": 0.3667,
+      "step": 157
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.5004688663290637,
+      "learning_rate": 3.990142120595444e-05,
+      "loss": 0.3543,
+      "step": 158
+    },
+    {
+      "epoch": 0.6440506329113924,
+      "grad_norm": 0.5753741608036556,
+      "learning_rate": 3.98957126299907e-05,
+      "loss": 0.3681,
+      "step": 159
+    },
+    {
+      "epoch": 0.6481012658227848,
+      "grad_norm": 0.48226993495058806,
+      "learning_rate": 3.9889843816692596e-05,
+      "loss": 0.3684,
+      "step": 160
+    },
+    {
+      "epoch": 0.6521518987341772,
+      "grad_norm": 0.5363678103898167,
+      "learning_rate": 3.9883814813326766e-05,
+      "loss": 0.3618,
+      "step": 161
+    },
+    {
+      "epoch": 0.6562025316455696,
+      "grad_norm": 0.5029572184223613,
+      "learning_rate": 3.9877625668449956e-05,
+      "loss": 0.3576,
+      "step": 162
+    },
+    {
+      "epoch": 0.660253164556962,
+      "grad_norm": 0.43855701053191515,
+      "learning_rate": 3.98712764319087e-05,
+      "loss": 0.3685,
+      "step": 163
+    },
+    {
+      "epoch": 0.6643037974683544,
+      "grad_norm": 0.4767591314132275,
+      "learning_rate": 3.9864767154838864e-05,
+      "loss": 0.3537,
+      "step": 164
+    },
+    {
+      "epoch": 0.6683544303797468,
+      "grad_norm": 0.4734656745224945,
+      "learning_rate": 3.9858097889665277e-05,
+      "loss": 0.3642,
+      "step": 165
+    },
+    {
+      "epoch": 0.6724050632911392,
+      "grad_norm": 0.44379385270287813,
+      "learning_rate": 3.985126869010129e-05,
+      "loss": 0.3695,
+      "step": 166
+    },
+    {
+      "epoch": 0.6764556962025317,
+      "grad_norm": 0.45144997721187424,
+      "learning_rate": 3.984427961114833e-05,
+      "loss": 0.3487,
+      "step": 167
+    },
+    {
+      "epoch": 0.6805063291139241,
+      "grad_norm": 0.5375728748827706,
+      "learning_rate": 3.9837130709095475e-05,
+      "loss": 0.3568,
+      "step": 168
+    },
+    {
+      "epoch": 0.6845569620253165,
+      "grad_norm": 0.4587580157238261,
+      "learning_rate": 3.982982204151901e-05,
+      "loss": 0.3654,
+      "step": 169
+    },
+    {
+      "epoch": 0.6886075949367089,
+      "grad_norm": 0.49119988530605946,
+      "learning_rate": 3.982235366728193e-05,
+      "loss": 0.3687,
+      "step": 170
+    },
+    {
+      "epoch": 0.6926582278481013,
+      "grad_norm": 0.45391998441953163,
+      "learning_rate": 3.9814725646533505e-05,
+      "loss": 0.3452,
+      "step": 171
+    },
+    {
+      "epoch": 0.6967088607594937,
+      "grad_norm": 0.4668632922214047,
+      "learning_rate": 3.9806938040708746e-05,
+      "loss": 0.3398,
+      "step": 172
+    },
+    {
+      "epoch": 0.7007594936708861,
+      "grad_norm": 0.5020782641684365,
+      "learning_rate": 3.9798990912527976e-05,
+      "loss": 0.3547,
+      "step": 173
+    },
+    {
+      "epoch": 0.7048101265822785,
+      "grad_norm": 0.5895215344427097,
+      "learning_rate": 3.979088432599627e-05,
+      "loss": 0.357,
+      "step": 174
+    },
+    {
+      "epoch": 0.7088607594936709,
+      "grad_norm": 0.549828745397567,
+      "learning_rate": 3.9782618346402964e-05,
+      "loss": 0.3708,
+      "step": 175
+    },
+    {
+      "epoch": 0.7129113924050633,
+      "grad_norm": 0.5112458901666622,
+      "learning_rate": 3.977419304032111e-05,
+      "loss": 0.3689,
+      "step": 176
+    },
+    {
+      "epoch": 0.7169620253164557,
+      "grad_norm": 0.6015965009142736,
+      "learning_rate": 3.976560847560697e-05,
+      "loss": 0.3755,
+      "step": 177
+    },
+    {
+      "epoch": 0.7210126582278481,
+      "grad_norm": 0.487944600791812,
+      "learning_rate": 3.9756864721399456e-05,
+      "loss": 0.3649,
+      "step": 178
+    },
+    {
+      "epoch": 0.7250632911392405,
+      "grad_norm": 0.5429317153689766,
+      "learning_rate": 3.974796184811956e-05,
+      "loss": 0.3651,
+      "step": 179
+    },
+    {
+      "epoch": 0.7291139240506329,
+      "grad_norm": 0.5887761158726933,
+      "learning_rate": 3.973889992746979e-05,
+      "loss": 0.3556,
+      "step": 180
+    },
+    {
+      "epoch": 0.7331645569620253,
+      "grad_norm": 0.5336222008412589,
+      "learning_rate": 3.972967903243361e-05,
+      "loss": 0.364,
+      "step": 181
+    },
+    {
+      "epoch": 0.7372151898734177,
+      "grad_norm": 0.5370830071330063,
+      "learning_rate": 3.972029923727486e-05,
+      "loss": 0.3496,
+      "step": 182
+    },
+    {
+      "epoch": 0.7412658227848101,
+      "grad_norm": 0.54620922307627,
+      "learning_rate": 3.971076061753709e-05,
+      "loss": 0.3538,
+      "step": 183
+    },
+    {
+      "epoch": 0.7453164556962025,
+      "grad_norm": 0.5631831478736016,
+      "learning_rate": 3.9701063250043066e-05,
+      "loss": 0.3564,
+      "step": 184
+    },
+    {
+      "epoch": 0.7493670886075949,
+      "grad_norm": 0.5480503710860208,
+      "learning_rate": 3.969120721289402e-05,
+      "loss": 0.3571,
+      "step": 185
+    },
+    {
+      "epoch": 0.7534177215189873,
+      "grad_norm": 0.538449964904298,
+      "learning_rate": 3.9681192585469146e-05,
+      "loss": 0.3701,
+      "step": 186
+    },
+    {
+      "epoch": 0.7574683544303797,
+      "grad_norm": 0.789882377905188,
+      "learning_rate": 3.9671019448424865e-05,
+      "loss": 0.3513,
+      "step": 187
+    },
+    {
+      "epoch": 0.7615189873417721,
+      "grad_norm": 0.448131407738137,
+      "learning_rate": 3.966068788369422e-05,
+      "loss": 0.3647,
+      "step": 188
+    },
+    {
+      "epoch": 0.7655696202531646,
+      "grad_norm": 0.787497365696159,
+      "learning_rate": 3.965019797448622e-05,
+      "loss": 0.3583,
+      "step": 189
+    },
+    {
+      "epoch": 0.769620253164557,
+      "grad_norm": 0.7093073003250262,
+      "learning_rate": 3.963954980528515e-05,
+      "loss": 0.3653,
+      "step": 190
+    },
+    {
+      "epoch": 0.7736708860759494,
+      "grad_norm": 0.49267863828732245,
+      "learning_rate": 3.9628743461849905e-05,
+      "loss": 0.3603,
+      "step": 191
+    },
+    {
+      "epoch": 0.7777215189873418,
+      "grad_norm": 0.6909077728448914,
+      "learning_rate": 3.961777903121329e-05,
+      "loss": 0.3533,
+      "step": 192
+    },
+    {
+      "epoch": 0.7817721518987342,
+      "grad_norm": 0.44423076566896835,
+      "learning_rate": 3.960665660168131e-05,
+      "loss": 0.3409,
+      "step": 193
+    },
+    {
+      "epoch": 0.7858227848101266,
+      "grad_norm": 0.5747845711286524,
+      "learning_rate": 3.9595376262832485e-05,
+      "loss": 0.3642,
+      "step": 194
+    },
+    {
+      "epoch": 0.789873417721519,
+      "grad_norm": 0.44833896777044735,
+      "learning_rate": 3.9583938105517127e-05,
+      "loss": 0.3543,
+      "step": 195
+    },
+    {
+      "epoch": 0.7939240506329114,
+      "grad_norm": 0.45440565128838634,
+      "learning_rate": 3.957234222185657e-05,
+      "loss": 0.3488,
+      "step": 196
+    },
+    {
+      "epoch": 0.7979746835443038,
+      "grad_norm": 0.46378331931210476,
+      "learning_rate": 3.9560588705242474e-05,
+      "loss": 0.3645,
+      "step": 197
+    },
+    {
+      "epoch": 0.8020253164556962,
+      "grad_norm": 0.44908330427822546,
+      "learning_rate": 3.954867765033605e-05,
+      "loss": 0.353,
+      "step": 198
+    },
+    {
+      "epoch": 0.8060759493670886,
+      "grad_norm": 0.5014507258089634,
+      "learning_rate": 3.953660915306728e-05,
+      "loss": 0.3464,
+      "step": 199
+    },
+    {
+      "epoch": 0.810126582278481,
+      "grad_norm": 0.48066974675209506,
+      "learning_rate": 3.952438331063419e-05,
+      "loss": 0.3576,
+      "step": 200
+    },
+    {
+      "epoch": 0.8141772151898734,
+      "grad_norm": 0.4676361130186093,
+      "learning_rate": 3.951200022150205e-05,
+      "loss": 0.3388,
+      "step": 201
+    },
+    {
+      "epoch": 0.8182278481012658,
+      "grad_norm": 0.554363727789509,
+      "learning_rate": 3.949945998540253e-05,
+      "loss": 0.3643,
+      "step": 202
+    },
+    {
+      "epoch": 0.8222784810126582,
+      "grad_norm": 0.5253534062971379,
+      "learning_rate": 3.9486762703332993e-05,
+      "loss": 0.3599,
+      "step": 203
+    },
+    {
+      "epoch": 0.8263291139240506,
+      "grad_norm": 0.48850250503999826,
+      "learning_rate": 3.947390847755559e-05,
+      "loss": 0.3521,
+      "step": 204
+    },
+    {
+      "epoch": 0.830379746835443,
+      "grad_norm": 0.49899462246348236,
+      "learning_rate": 3.946089741159648e-05,
+      "loss": 0.3568,
+      "step": 205
+    },
+    {
+      "epoch": 0.8344303797468354,
+      "grad_norm": 0.5149665324223205,
+      "learning_rate": 3.944772961024501e-05,
+      "loss": 0.3511,
+      "step": 206
+    },
+    {
+      "epoch": 0.8384810126582278,
+      "grad_norm": 0.526072995124016,
+      "learning_rate": 3.943440517955285e-05,
+      "loss": 0.3612,
+      "step": 207
+    },
+    {
+      "epoch": 0.8425316455696202,
+      "grad_norm": 0.5313429199666861,
+      "learning_rate": 3.9420924226833126e-05,
+      "loss": 0.3656,
+      "step": 208
+    },
+    {
+      "epoch": 0.8465822784810126,
+      "grad_norm": 0.5223384889133448,
+      "learning_rate": 3.9407286860659566e-05,
+      "loss": 0.3551,
+      "step": 209
+    },
+    {
+      "epoch": 0.850632911392405,
+      "grad_norm": 0.4872283514423054,
+      "learning_rate": 3.9393493190865657e-05,
+      "loss": 0.3593,
+      "step": 210
+    },
+    {
+      "epoch": 0.8546835443037974,
+      "grad_norm": 0.5272197937049422,
+      "learning_rate": 3.937954332854371e-05,
+      "loss": 0.3588,
+      "step": 211
+    },
+    {
+      "epoch": 0.8587341772151899,
+      "grad_norm": 0.49722423302521024,
+      "learning_rate": 3.9365437386044016e-05,
+      "loss": 0.3378,
+      "step": 212
+    },
+    {
+      "epoch": 0.8627848101265823,
+      "grad_norm": 0.5194954934300087,
+      "learning_rate": 3.935117547697387e-05,
+      "loss": 0.351,
+      "step": 213
+    },
+    {
+      "epoch": 0.8668354430379747,
+      "grad_norm": 0.5320043744041031,
+      "learning_rate": 3.933675771619675e-05,
+      "loss": 0.3693,
+      "step": 214
+    },
+    {
+      "epoch": 0.8708860759493671,
+      "grad_norm": 0.5321053338069895,
+      "learning_rate": 3.932218421983131e-05,
+      "loss": 0.361,
+      "step": 215
+    },
+    {
+      "epoch": 0.8749367088607595,
+      "grad_norm": 0.4263122283176262,
+      "learning_rate": 3.9307455105250484e-05,
+      "loss": 0.3534,
+      "step": 216
+    },
+    {
+      "epoch": 0.8789873417721519,
+      "grad_norm": 0.519231317675057,
+      "learning_rate": 3.929257049108054e-05,
+      "loss": 0.3451,
+      "step": 217
+    },
+    {
+      "epoch": 0.8830379746835443,
+      "grad_norm": 0.40642468059976805,
+      "learning_rate": 3.927753049720011e-05,
+      "loss": 0.3392,
+      "step": 218
+    },
+    {
+      "epoch": 0.8870886075949367,
+      "grad_norm": 0.4631215493833277,
+      "learning_rate": 3.9262335244739234e-05,
+      "loss": 0.3616,
+      "step": 219
+    },
+    {
+      "epoch": 0.8911392405063291,
+      "grad_norm": 0.43814979645018776,
+      "learning_rate": 3.92469848560784e-05,
+      "loss": 0.3341,
+      "step": 220
+    },
+    {
+      "epoch": 0.8951898734177215,
+      "grad_norm": 0.45737615733250975,
+      "learning_rate": 3.923147945484751e-05,
+      "loss": 0.3469,
+      "step": 221
+    },
+    {
+      "epoch": 0.8992405063291139,
+      "grad_norm": 0.45532065987027875,
+      "learning_rate": 3.9215819165924956e-05,
+      "loss": 0.3641,
+      "step": 222
+    },
+    {
+      "epoch": 0.9032911392405063,
+      "grad_norm": 0.48411110159907034,
+      "learning_rate": 3.920000411543654e-05,
+      "loss": 0.3466,
+      "step": 223
+    },
+    {
+      "epoch": 0.9073417721518987,
+      "grad_norm": 0.46179009400052584,
+      "learning_rate": 3.9184034430754495e-05,
+      "loss": 0.3539,
+      "step": 224
+    },
+    {
+      "epoch": 0.9113924050632911,
+      "grad_norm": 0.4301846224693135,
+      "learning_rate": 3.916791024049648e-05,
+      "loss": 0.353,
+      "step": 225
+    },
+    {
+      "epoch": 0.9154430379746835,
+      "grad_norm": 0.42150134662749117,
+      "learning_rate": 3.91516316745245e-05,
+      "loss": 0.3441,
+      "step": 226
+    },
+    {
+      "epoch": 0.9194936708860759,
+      "grad_norm": 0.42389229015949703,
+      "learning_rate": 3.913519886394389e-05,
+      "loss": 0.3602,
+      "step": 227
+    },
+    {
+      "epoch": 0.9235443037974683,
+      "grad_norm": 0.42581328511596567,
+      "learning_rate": 3.911861194110225e-05,
+      "loss": 0.3386,
+      "step": 228
+    },
+    {
+      "epoch": 0.9275949367088607,
+      "grad_norm": 0.4247415074754598,
+      "learning_rate": 3.910187103958837e-05,
+      "loss": 0.3428,
+      "step": 229
+    },
+    {
+      "epoch": 0.9316455696202531,
+      "grad_norm": 0.4618434822693623,
+      "learning_rate": 3.908497629423117e-05,
+      "loss": 0.344,
+      "step": 230
+    },
+    {
+      "epoch": 0.9356962025316455,
+      "grad_norm": 0.4508257340116673,
+      "learning_rate": 3.9067927841098614e-05,
+      "loss": 0.3406,
+      "step": 231
+    },
+    {
+      "epoch": 0.9397468354430379,
+      "grad_norm": 0.4289747615867386,
+      "learning_rate": 3.9050725817496594e-05,
+      "loss": 0.3433,
+      "step": 232
+    },
+    {
+      "epoch": 0.9437974683544303,
+      "grad_norm": 0.48130985658005576,
+      "learning_rate": 3.9033370361967844e-05,
+      "loss": 0.3433,
+      "step": 233
+    },
+    {
+      "epoch": 0.9478481012658228,
+      "grad_norm": 0.4146306160774892,
+      "learning_rate": 3.901586161429081e-05,
+      "loss": 0.353,
+      "step": 234
+    },
+    {
+      "epoch": 0.9518987341772152,
+      "grad_norm": 0.41640799127770517,
+      "learning_rate": 3.8998199715478545e-05,
+      "loss": 0.3523,
+      "step": 235
+    },
+    {
+      "epoch": 0.9559493670886076,
+      "grad_norm": 0.4028495802646592,
+      "learning_rate": 3.8980384807777564e-05,
+      "loss": 0.3404,
+      "step": 236
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.49544102490874636,
+      "learning_rate": 3.896241703466667e-05,
+      "loss": 0.3431,
+      "step": 237
+    },
+    {
+      "epoch": 0.9640506329113924,
+      "grad_norm": 0.4476747802735973,
+      "learning_rate": 3.894429654085585e-05,
+      "loss": 0.3485,
+      "step": 238
+    },
+    {
+      "epoch": 0.9681012658227848,
+      "grad_norm": 0.5352718051092262,
+      "learning_rate": 3.892602347228505e-05,
+      "loss": 0.3589,
+      "step": 239
+    },
+    {
+      "epoch": 0.9721518987341772,
+      "grad_norm": 0.40890023395973357,
+      "learning_rate": 3.890759797612307e-05,
+      "loss": 0.3592,
+      "step": 240
+    },
+    {
+      "epoch": 0.9762025316455696,
+      "grad_norm": 0.5071773938228529,
+      "learning_rate": 3.888902020076632e-05,
+      "loss": 0.3465,
+      "step": 241
+    },
+    {
+      "epoch": 0.980253164556962,
+      "grad_norm": 0.4038189512867058,
+      "learning_rate": 3.887029029583764e-05,
+      "loss": 0.3453,
+      "step": 242
+    },
+    {
+      "epoch": 0.9843037974683544,
+      "grad_norm": 0.5063912734188458,
+      "learning_rate": 3.8851408412185125e-05,
+      "loss": 0.3568,
+      "step": 243
+    },
+    {
+      "epoch": 0.9883544303797468,
+      "grad_norm": 0.41510313233404605,
+      "learning_rate": 3.8832374701880855e-05,
+      "loss": 0.3466,
+      "step": 244
+    },
+    {
+      "epoch": 0.9924050632911392,
+      "grad_norm": 0.46171980248621214,
+      "learning_rate": 3.881318931821972e-05,
+      "loss": 0.3631,
+      "step": 245
+    },
+    {
+      "epoch": 0.9964556962025316,
+      "grad_norm": 0.5245029808119696,
+      "learning_rate": 3.879385241571817e-05,
+      "loss": 0.3602,
+      "step": 246
+    },
+    {
+      "epoch": 1.0020253164556963,
+      "grad_norm": 0.4266625895520189,
+      "learning_rate": 3.8774364150112955e-05,
+      "loss": 0.3183,
+      "step": 247
+    },
+    {
+      "epoch": 1.0060759493670886,
+      "grad_norm": 0.44825977929393396,
+      "learning_rate": 3.8754724678359884e-05,
+      "loss": 0.2848,
+      "step": 248
+    },
+    {
+      "epoch": 1.010126582278481,
+      "grad_norm": 0.46583900792173005,
+      "learning_rate": 3.873493415863256e-05,
+      "loss": 0.2803,
+      "step": 249
+    },
+    {
+      "epoch": 1.0141772151898734,
+      "grad_norm": 0.4257794224148312,
+      "learning_rate": 3.871499275032111e-05,
+      "loss": 0.2958,
+      "step": 250
+    },
+    {
+      "epoch": 1.0182278481012659,
+      "grad_norm": 0.42843498296600957,
+      "learning_rate": 3.869490061403091e-05,
+      "loss": 0.2805,
+      "step": 251
+    },
+    {
+      "epoch": 1.0222784810126582,
+      "grad_norm": 0.4614264803206258,
+      "learning_rate": 3.867465791158124e-05,
+      "loss": 0.2855,
+      "step": 252
+    },
+    {
+      "epoch": 1.0263291139240507,
+      "grad_norm": 0.42804622155019706,
+      "learning_rate": 3.865426480600407e-05,
+      "loss": 0.2853,
+      "step": 253
+    },
+    {
+      "epoch": 1.030379746835443,
+      "grad_norm": 0.47532638233563707,
+      "learning_rate": 3.863372146154264e-05,
+      "loss": 0.2955,
+      "step": 254
+    },
+    {
+      "epoch": 1.0344303797468355,
+      "grad_norm": 0.4481934917571885,
+      "learning_rate": 3.861302804365024e-05,
+      "loss": 0.269,
+      "step": 255
+    },
+    {
+      "epoch": 1.0384810126582278,
+      "grad_norm": 0.45792364969024807,
+      "learning_rate": 3.85921847189888e-05,
+      "loss": 0.278,
+      "step": 256
+    },
+    {
+      "epoch": 1.0425316455696203,
+      "grad_norm": 0.4604879229231146,
+      "learning_rate": 3.85711916554276e-05,
+      "loss": 0.2868,
+      "step": 257
+    },
+    {
+      "epoch": 1.0465822784810126,
+      "grad_norm": 0.44051855982432864,
+      "learning_rate": 3.85500490220419e-05,
+      "loss": 0.2806,
+      "step": 258
+    },
+    {
+      "epoch": 1.0506329113924051,
+      "grad_norm": 0.5375266479903489,
+      "learning_rate": 3.852875698911154e-05,
+      "loss": 0.2936,
+      "step": 259
+    },
+    {
+      "epoch": 1.0546835443037974,
+      "grad_norm": 0.5033203114153285,
+      "learning_rate": 3.850731572811963e-05,
+      "loss": 0.2912,
+      "step": 260
+    },
+    {
+      "epoch": 1.05873417721519,
+      "grad_norm": 0.46602037346813013,
+      "learning_rate": 3.848572541175116e-05,
+      "loss": 0.2742,
+      "step": 261
+    },
+    {
+      "epoch": 1.0627848101265822,
+      "grad_norm": 0.4731053044021018,
+      "learning_rate": 3.846398621389154e-05,
+      "loss": 0.2752,
+      "step": 262
+    },
+    {
+      "epoch": 1.0668354430379747,
+      "grad_norm": 0.5228427155863514,
+      "learning_rate": 3.84420983096253e-05,
+      "loss": 0.2807,
+      "step": 263
+    },
+    {
+      "epoch": 1.070886075949367,
+      "grad_norm": 0.45084125382316076,
+      "learning_rate": 3.8420061875234606e-05,
+      "loss": 0.2725,
+      "step": 264
+    },
+    {
+      "epoch": 1.0749367088607595,
+      "grad_norm": 0.5034464714498813,
+      "learning_rate": 3.839787708819787e-05,
+      "loss": 0.2773,
+      "step": 265
+    },
+    {
+      "epoch": 1.0789873417721518,
+      "grad_norm": 0.4373651634029232,
+      "learning_rate": 3.8375544127188325e-05,
+      "loss": 0.2778,
+      "step": 266
+    },
+    {
+      "epoch": 1.0830379746835443,
+      "grad_norm": 0.5413247625869282,
+      "learning_rate": 3.8353063172072564e-05,
+      "loss": 0.2758,
+      "step": 267
+    },
+    {
+      "epoch": 1.0870886075949366,
+      "grad_norm": 0.44718533244110564,
+      "learning_rate": 3.8330434403909105e-05,
+      "loss": 0.274,
+      "step": 268
+    },
+    {
+      "epoch": 1.0911392405063292,
+      "grad_norm": 0.45333404801323546,
+      "learning_rate": 3.8307658004946934e-05,
+      "loss": 0.282,
+      "step": 269
+    },
+    {
+      "epoch": 1.0951898734177214,
+      "grad_norm": 0.47326246806602423,
+      "learning_rate": 3.8284734158624046e-05,
+      "loss": 0.268,
+      "step": 270
+    },
+    {
+      "epoch": 1.099240506329114,
+      "grad_norm": 0.4478133770098197,
+      "learning_rate": 3.826166304956594e-05,
+      "loss": 0.2913,
+      "step": 271
+    },
+    {
+      "epoch": 1.1032911392405063,
+      "grad_norm": 0.5022238672758715,
+      "learning_rate": 3.8238444863584164e-05,
+      "loss": 0.2856,
+      "step": 272
+    },
+    {
+      "epoch": 1.1073417721518988,
+      "grad_norm": 0.4878877080676492,
+      "learning_rate": 3.821507978767479e-05,
+      "loss": 0.2707,
+      "step": 273
+    },
+    {
+      "epoch": 1.111392405063291,
+      "grad_norm": 0.4287161217111746,
+      "learning_rate": 3.819156801001693e-05,
+      "loss": 0.2795,
+      "step": 274
+    },
+    {
+      "epoch": 1.1154430379746836,
+      "grad_norm": 0.5713188327432861,
+      "learning_rate": 3.816790971997121e-05,
+      "loss": 0.2862,
+      "step": 275
+    },
+    {
+      "epoch": 1.1194936708860759,
+      "grad_norm": 0.4037418089733278,
+      "learning_rate": 3.8144105108078246e-05,
+      "loss": 0.2662,
+      "step": 276
+    },
+    {
+      "epoch": 1.1235443037974684,
+      "grad_norm": 0.548810471001094,
+      "learning_rate": 3.81201543660571e-05,
+      "loss": 0.2915,
+      "step": 277
+    },
+    {
+      "epoch": 1.1275949367088607,
+      "grad_norm": 0.4665587327896116,
+      "learning_rate": 3.809605768680377e-05,
+      "loss": 0.2724,
+      "step": 278
+    },
+    {
+      "epoch": 1.1316455696202532,
+      "grad_norm": 0.4335246867788456,
+      "learning_rate": 3.807181526438958e-05,
+      "loss": 0.2745,
+      "step": 279
+    },
+    {
+      "epoch": 1.1356962025316455,
+      "grad_norm": 0.5193065334230033,
+      "learning_rate": 3.8047427294059697e-05,
+      "loss": 0.2744,
+      "step": 280
+    },
+    {
+      "epoch": 1.139746835443038,
+      "grad_norm": 0.5031274584595459,
+      "learning_rate": 3.802289397223145e-05,
+      "loss": 0.2809,
+      "step": 281
+    },
+    {
+      "epoch": 1.1437974683544303,
+      "grad_norm": 0.46954667300072617,
+      "learning_rate": 3.7998215496492854e-05,
+      "loss": 0.2795,
+      "step": 282
+    },
+    {
+      "epoch": 1.1478481012658228,
+      "grad_norm": 0.5617659373343593,
+      "learning_rate": 3.797339206560096e-05,
+      "loss": 0.2752,
+      "step": 283
+    },
+    {
+      "epoch": 1.1518987341772151,
+      "grad_norm": 0.4086677790558437,
+      "learning_rate": 3.794842387948027e-05,
+      "loss": 0.2751,
+      "step": 284
+    },
+    {
+      "epoch": 1.1559493670886076,
+      "grad_norm": 0.5187084465252801,
+      "learning_rate": 3.7923311139221114e-05,
+      "loss": 0.2782,
+      "step": 285
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.43742980242869817,
+      "learning_rate": 3.7898054047078054e-05,
+      "loss": 0.2737,
+      "step": 286
+    },
+    {
+      "epoch": 1.1640506329113924,
+      "grad_norm": 0.5015186103977192,
+      "learning_rate": 3.787265280646825e-05,
+      "loss": 0.2999,
+      "step": 287
+    },
+    {
+      "epoch": 1.1681012658227847,
+      "grad_norm": 0.4534905173794621,
+      "learning_rate": 3.7847107621969786e-05,
+      "loss": 0.2892,
+      "step": 288
+    },
+    {
+      "epoch": 1.1721518987341772,
+      "grad_norm": 0.4092755078012933,
+      "learning_rate": 3.7821418699320064e-05,
+      "loss": 0.2812,
+      "step": 289
+    },
+    {
+      "epoch": 1.1762025316455695,
+      "grad_norm": 0.44959802282993777,
+      "learning_rate": 3.7795586245414145e-05,
+      "loss": 0.2817,
+      "step": 290
+    },
+    {
+      "epoch": 1.180253164556962,
+      "grad_norm": 0.38525190172356577,
+      "learning_rate": 3.776961046830306e-05,
+      "loss": 0.2843,
+      "step": 291
+    },
+    {
+      "epoch": 1.1843037974683543,
+      "grad_norm": 0.4703193726685085,
+      "learning_rate": 3.774349157719215e-05,
+      "loss": 0.272,
+      "step": 292
+    },
+    {
+      "epoch": 1.1883544303797469,
+      "grad_norm": 0.42595972836640816,
+      "learning_rate": 3.7717229782439365e-05,
+      "loss": 0.2973,
+      "step": 293
+    },
+    {
+      "epoch": 1.1924050632911392,
+      "grad_norm": 0.44880109023322284,
+      "learning_rate": 3.769082529555359e-05,
+      "loss": 0.2869,
+      "step": 294
+    },
+    {
+      "epoch": 1.1964556962025317,
+      "grad_norm": 0.451939365201627,
+      "learning_rate": 3.766427832919294e-05,
+      "loss": 0.3041,
+      "step": 295
+    },
+    {
+      "epoch": 1.200506329113924,
+      "grad_norm": 0.46925330746202804,
+      "learning_rate": 3.7637589097163024e-05,
+      "loss": 0.296,
+      "step": 296
+    },
+    {
+      "epoch": 1.2045569620253165,
+      "grad_norm": 0.501702857671787,
+      "learning_rate": 3.761075781441526e-05,
+      "loss": 0.2875,
+      "step": 297
+    },
+    {
+      "epoch": 1.2086075949367088,
+      "grad_norm": 0.44848829870253976,
+      "learning_rate": 3.75837846970451e-05,
+      "loss": 0.2651,
+      "step": 298
+    },
+    {
+      "epoch": 1.2126582278481013,
+      "grad_norm": 0.46012288237964544,
+      "learning_rate": 3.755666996229032e-05,
+      "loss": 0.2853,
+      "step": 299
+    },
+    {
+      "epoch": 1.2167088607594936,
+      "grad_norm": 0.5182821787639246,
+      "learning_rate": 3.752941382852927e-05,
+      "loss": 0.2939,
+      "step": 300
+    },
+    {
+      "epoch": 1.220759493670886,
+      "grad_norm": 0.42929540240780983,
+      "learning_rate": 3.7502016515279115e-05,
+      "loss": 0.2908,
+      "step": 301
+    },
+    {
+      "epoch": 1.2248101265822784,
+      "grad_norm": 0.519563851694771,
+      "learning_rate": 3.7474478243194043e-05,
+      "loss": 0.2877,
+      "step": 302
+    },
+    {
+      "epoch": 1.228860759493671,
+      "grad_norm": 0.39705018498123495,
+      "learning_rate": 3.744679923406351e-05,
+      "loss": 0.2789,
+      "step": 303
+    },
+    {
+      "epoch": 1.2329113924050632,
+      "grad_norm": 0.475284363757787,
+      "learning_rate": 3.741897971081043e-05,
+      "loss": 0.2815,
+      "step": 304
+    },
+    {
+      "epoch": 1.2369620253164557,
+      "grad_norm": 0.39882513133120384,
+      "learning_rate": 3.739101989748946e-05,
+      "loss": 0.281,
+      "step": 305
+    },
+    {
+      "epoch": 1.241012658227848,
+      "grad_norm": 0.3776020273632585,
+      "learning_rate": 3.7362920019285066e-05,
+      "loss": 0.2758,
+      "step": 306
+    },
+    {
+      "epoch": 1.2450632911392405,
+      "grad_norm": 0.4648184375532975,
+      "learning_rate": 3.73346803025098e-05,
+      "loss": 0.2766,
+      "step": 307
+    },
+    {
+      "epoch": 1.2491139240506328,
+      "grad_norm": 0.3956220157318331,
+      "learning_rate": 3.730630097460247e-05,
+      "loss": 0.2771,
+      "step": 308
+    },
+    {
+      "epoch": 1.2531645569620253,
+      "grad_norm": 0.40953026057906944,
+      "learning_rate": 3.727778226412628e-05,
+      "loss": 0.2796,
+      "step": 309
+    },
+    {
+      "epoch": 1.2572151898734178,
+      "grad_norm": 0.40256754707878833,
+      "learning_rate": 3.7249124400767006e-05,
+      "loss": 0.2899,
+      "step": 310
+    },
+    {
+      "epoch": 1.2612658227848101,
+      "grad_norm": 0.4596891823731273,
+      "learning_rate": 3.722032761533114e-05,
+      "loss": 0.2909,
+      "step": 311
+    },
+    {
+      "epoch": 1.2653164556962024,
+      "grad_norm": 0.4219038208684509,
+      "learning_rate": 3.719139213974403e-05,
+      "loss": 0.2872,
+      "step": 312
+    },
+    {
+      "epoch": 1.269367088607595,
+      "grad_norm": 0.37045992557641105,
+      "learning_rate": 3.7162318207048006e-05,
+      "loss": 0.2699,
+      "step": 313
+    },
+    {
+      "epoch": 1.2734177215189875,
+      "grad_norm": 0.4026805043319528,
+      "learning_rate": 3.713310605140055e-05,
+      "loss": 0.2854,
+      "step": 314
+    },
+    {
+      "epoch": 1.2774683544303798,
+      "grad_norm": 0.41405061396725756,
+      "learning_rate": 3.710375590807233e-05,
+      "loss": 0.2931,
+      "step": 315
+    },
+    {
+      "epoch": 1.281518987341772,
+      "grad_norm": 0.4565033870978405,
+      "learning_rate": 3.7074268013445365e-05,
+      "loss": 0.3045,
+      "step": 316
+    },
+    {
+      "epoch": 1.2855696202531646,
+      "grad_norm": 0.36794935396827255,
+      "learning_rate": 3.7044642605011114e-05,
+      "loss": 0.2819,
+      "step": 317
+    },
+    {
+      "epoch": 1.289620253164557,
+      "grad_norm": 0.36833280364152393,
+      "learning_rate": 3.701487992136854e-05,
+      "loss": 0.2857,
+      "step": 318
+    },
+    {
+      "epoch": 1.2936708860759494,
+      "grad_norm": 0.42003702102257456,
+      "learning_rate": 3.69849802022222e-05,
+      "loss": 0.2868,
+      "step": 319
+    },
+    {
+      "epoch": 1.2977215189873417,
+      "grad_norm": 0.4412920407159841,
+      "learning_rate": 3.6954943688380334e-05,
+      "loss": 0.2892,
+      "step": 320
+    },
+    {
+      "epoch": 1.3017721518987342,
+      "grad_norm": 0.39040269115900467,
+      "learning_rate": 3.692477062175289e-05,
+      "loss": 0.2854,
+      "step": 321
+    },
+    {
+      "epoch": 1.3058227848101267,
+      "grad_norm": 0.41017671691221486,
+      "learning_rate": 3.689446124534958e-05,
+      "loss": 0.2811,
+      "step": 322
+    },
+    {
+      "epoch": 1.309873417721519,
+      "grad_norm": 0.44866820690820697,
+      "learning_rate": 3.686401580327799e-05,
+      "loss": 0.288,
+      "step": 323
+    },
+    {
+      "epoch": 1.3139240506329113,
+      "grad_norm": 0.42714247028953417,
+      "learning_rate": 3.683343454074149e-05,
+      "loss": 0.2787,
+      "step": 324
+    },
+    {
+      "epoch": 1.3179746835443038,
+      "grad_norm": 0.4439476283900815,
+      "learning_rate": 3.6802717704037386e-05,
+      "loss": 0.2798,
+      "step": 325
+    },
+    {
+      "epoch": 1.3220253164556963,
+      "grad_norm": 0.4124148417425249,
+      "learning_rate": 3.6771865540554855e-05,
+      "loss": 0.2863,
+      "step": 326
+    },
+    {
+      "epoch": 1.3260759493670886,
+      "grad_norm": 0.4119989829018879,
+      "learning_rate": 3.674087829877297e-05,
+      "loss": 0.2717,
+      "step": 327
+    },
+    {
+      "epoch": 1.330126582278481,
+      "grad_norm": 0.4272990690423661,
+      "learning_rate": 3.6709756228258735e-05,
+      "loss": 0.2791,
+      "step": 328
+    },
+    {
+      "epoch": 1.3341772151898734,
+      "grad_norm": 0.4008701435333682,
+      "learning_rate": 3.667849957966501e-05,
+      "loss": 0.2805,
+      "step": 329
+    },
+    {
+      "epoch": 1.338227848101266,
+      "grad_norm": 0.4643625207852802,
+      "learning_rate": 3.6647108604728546e-05,
+      "loss": 0.2951,
+      "step": 330
+    },
+    {
+      "epoch": 1.3422784810126582,
+      "grad_norm": 0.3797109294766969,
+      "learning_rate": 3.661558355626795e-05,
+      "loss": 0.283,
+      "step": 331
+    },
+    {
+      "epoch": 1.3463291139240505,
+      "grad_norm": 0.3913606555570985,
+      "learning_rate": 3.658392468818163e-05,
+      "loss": 0.2797,
+      "step": 332
+    },
+    {
+      "epoch": 1.350379746835443,
+      "grad_norm": 0.4235500511125955,
+      "learning_rate": 3.655213225544574e-05,
+      "loss": 0.2819,
+      "step": 333
+    },
+    {
+      "epoch": 1.3544303797468356,
+      "grad_norm": 0.42030673987437767,
+      "learning_rate": 3.652020651411218e-05,
+      "loss": 0.2776,
+      "step": 334
+    },
+    {
+      "epoch": 1.3584810126582278,
+      "grad_norm": 0.4411703635824486,
+      "learning_rate": 3.6488147721306474e-05,
+      "loss": 0.2896,
+      "step": 335
+    },
+    {
+      "epoch": 1.3625316455696201,
+      "grad_norm": 0.43482217610839535,
+      "learning_rate": 3.645595613522574e-05,
+      "loss": 0.292,
+      "step": 336
+    },
+    {
+      "epoch": 1.3665822784810127,
+      "grad_norm": 0.39981847645551233,
+      "learning_rate": 3.642363201513657e-05,
+      "loss": 0.2782,
+      "step": 337
+    },
+    {
+      "epoch": 1.3706329113924052,
+      "grad_norm": 0.3947145615824429,
+      "learning_rate": 3.6391175621373006e-05,
+      "loss": 0.2859,
+      "step": 338
+    },
+    {
+      "epoch": 1.3746835443037975,
+      "grad_norm": 0.3987291062390215,
+      "learning_rate": 3.6358587215334355e-05,
+      "loss": 0.2694,
+      "step": 339
+    },
+    {
+      "epoch": 1.3787341772151898,
+      "grad_norm": 0.36205810252671816,
+      "learning_rate": 3.632586705948318e-05,
+      "loss": 0.2781,
+      "step": 340
+    },
+    {
+      "epoch": 1.3827848101265823,
+      "grad_norm": 0.4134940763862734,
+      "learning_rate": 3.629301541734311e-05,
+      "loss": 0.2668,
+      "step": 341
+    },
+    {
+      "epoch": 1.3868354430379748,
+      "grad_norm": 0.38993218906261246,
+      "learning_rate": 3.626003255349676e-05,
+      "loss": 0.2911,
+      "step": 342
+    },
+    {
+      "epoch": 1.390886075949367,
+      "grad_norm": 0.400132611281598,
+      "learning_rate": 3.622691873358357e-05,
+      "loss": 0.2888,
+      "step": 343
+    },
+    {
+      "epoch": 1.3949367088607594,
+      "grad_norm": 0.373548024748819,
+      "learning_rate": 3.61936742242977e-05,
+      "loss": 0.2689,
+      "step": 344
+    },
+    {
+      "epoch": 1.398987341772152,
+      "grad_norm": 0.42681500452146365,
+      "learning_rate": 3.6160299293385864e-05,
+      "loss": 0.2741,
+      "step": 345
+    },
+    {
+      "epoch": 1.4030379746835444,
+      "grad_norm": 0.38368290116346054,
+      "learning_rate": 3.612679420964516e-05,
+      "loss": 0.3007,
+      "step": 346
+    },
+    {
+      "epoch": 1.4070886075949367,
+      "grad_norm": 0.4666316552877006,
+      "learning_rate": 3.609315924292092e-05,
+      "loss": 0.2924,
+      "step": 347
+    },
+    {
+      "epoch": 1.411139240506329,
+      "grad_norm": 0.3914114839836464,
+      "learning_rate": 3.6059394664104554e-05,
+      "loss": 0.2717,
+      "step": 348
+    },
+    {
+      "epoch": 1.4151898734177215,
+      "grad_norm": 0.37730574749033735,
+      "learning_rate": 3.602550074513133e-05,
+      "loss": 0.2826,
+      "step": 349
+    },
+    {
+      "epoch": 1.419240506329114,
+      "grad_norm": 0.446853988977003,
+      "learning_rate": 3.599147775897822e-05,
+      "loss": 0.2807,
+      "step": 350
+    },
+    {
+      "epoch": 1.4232911392405063,
+      "grad_norm": 0.405995860674348,
+      "learning_rate": 3.595732597966167e-05,
+      "loss": 0.2917,
+      "step": 351
+    },
+    {
+      "epoch": 1.4273417721518986,
+      "grad_norm": 0.3649404905301226,
+      "learning_rate": 3.592304568223542e-05,
+      "loss": 0.2856,
+      "step": 352
+    },
+    {
+      "epoch": 1.4313924050632911,
+      "grad_norm": 0.38586387767670033,
+      "learning_rate": 3.588863714278826e-05,
+      "loss": 0.2809,
+      "step": 353
+    },
+    {
+      "epoch": 1.4354430379746836,
+      "grad_norm": 0.3600775470698098,
+      "learning_rate": 3.585410063844186e-05,
+      "loss": 0.2839,
+      "step": 354
+    },
+    {
+      "epoch": 1.439493670886076,
+      "grad_norm": 0.3799733807570866,
+      "learning_rate": 3.581943644734846e-05,
+      "loss": 0.2884,
+      "step": 355
+    },
+    {
+      "epoch": 1.4435443037974682,
+      "grad_norm": 0.3894317816116,
+      "learning_rate": 3.578464484868869e-05,
+      "loss": 0.2721,
+      "step": 356
+    },
+    {
+      "epoch": 1.4475949367088607,
+      "grad_norm": 0.36841745705742507,
+      "learning_rate": 3.5749726122669316e-05,
+      "loss": 0.2762,
+      "step": 357
+    },
+    {
+      "epoch": 1.4516455696202533,
+      "grad_norm": 0.41261579402374887,
+      "learning_rate": 3.5714680550520943e-05,
+      "loss": 0.2827,
+      "step": 358
+    },
+    {
+      "epoch": 1.4556962025316456,
+      "grad_norm": 0.3923821037221912,
+      "learning_rate": 3.5679508414495794e-05,
+      "loss": 0.294,
+      "step": 359
+    },
+    {
+      "epoch": 1.4597468354430378,
+      "grad_norm": 0.40044025559416496,
+      "learning_rate": 3.564420999786543e-05,
+      "loss": 0.2849,
+      "step": 360
+    },
+    {
+      "epoch": 1.4637974683544304,
+      "grad_norm": 0.4034597128037802,
+      "learning_rate": 3.560878558491842e-05,
+      "loss": 0.2621,
+      "step": 361
+    },
+    {
+      "epoch": 1.4678481012658229,
+      "grad_norm": 0.39061108189284743,
+      "learning_rate": 3.5573235460958145e-05,
+      "loss": 0.2783,
+      "step": 362
+    },
+    {
+      "epoch": 1.4718987341772152,
+      "grad_norm": 0.3883680741786202,
+      "learning_rate": 3.553755991230039e-05,
+      "loss": 0.2811,
+      "step": 363
+    },
+    {
+      "epoch": 1.4759493670886075,
+      "grad_norm": 0.38966697904717884,
+      "learning_rate": 3.5501759226271144e-05,
+      "loss": 0.2769,
+      "step": 364
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.3708596832498937,
+      "learning_rate": 3.546583369120419e-05,
+      "loss": 0.2839,
+      "step": 365
+    },
+    {
+      "epoch": 1.4840506329113925,
+      "grad_norm": 0.40242573333486586,
+      "learning_rate": 3.5429783596438864e-05,
+      "loss": 0.2832,
+      "step": 366
+    },
+    {
+      "epoch": 1.4881012658227848,
+      "grad_norm": 0.38175094424552686,
+      "learning_rate": 3.539360923231766e-05,
+      "loss": 0.288,
+      "step": 367
+    },
+    {
+      "epoch": 1.492151898734177,
+      "grad_norm": 0.44703622414175576,
+      "learning_rate": 3.535731089018394e-05,
+      "loss": 0.2779,
+      "step": 368
+    },
+    {
+      "epoch": 1.4962025316455696,
+      "grad_norm": 0.38363759054557406,
+      "learning_rate": 3.532088886237956e-05,
+      "loss": 0.2779,
+      "step": 369
+    },
+    {
+      "epoch": 1.5002531645569621,
+      "grad_norm": 0.443854342630533,
+      "learning_rate": 3.528434344224253e-05,
+      "loss": 0.2887,
+      "step": 370
+    },
+    {
+      "epoch": 1.5043037974683544,
+      "grad_norm": 0.4233746000086867,
+      "learning_rate": 3.524767492410464e-05,
+      "loss": 0.2816,
+      "step": 371
+    },
+    {
+      "epoch": 1.5083544303797467,
+      "grad_norm": 0.3875451562177697,
+      "learning_rate": 3.521088360328908e-05,
+      "loss": 0.2825,
+      "step": 372
+    },
+    {
+      "epoch": 1.5124050632911392,
+      "grad_norm": 0.4403419237245663,
+      "learning_rate": 3.517396977610811e-05,
+      "loss": 0.2822,
+      "step": 373
+    },
+    {
+      "epoch": 1.5164556962025317,
+      "grad_norm": 0.40868292156283126,
+      "learning_rate": 3.5136933739860595e-05,
+      "loss": 0.2965,
+      "step": 374
+    },
+    {
+      "epoch": 1.520506329113924,
+      "grad_norm": 0.44740547386426444,
+      "learning_rate": 3.509977579282971e-05,
+      "loss": 0.2933,
+      "step": 375
+    },
+    {
+      "epoch": 1.5245569620253163,
+      "grad_norm": 0.37675859328315303,
+      "learning_rate": 3.5062496234280424e-05,
+      "loss": 0.296,
+      "step": 376
+    },
+    {
+      "epoch": 1.5286075949367088,
+      "grad_norm": 0.4827681376839125,
+      "learning_rate": 3.502509536445719e-05,
+      "loss": 0.3003,
+      "step": 377
+    },
+    {
+      "epoch": 1.5326582278481014,
+      "grad_norm": 0.4072622127476221,
+      "learning_rate": 3.498757348458147e-05,
+      "loss": 0.2754,
+      "step": 378
+    },
+    {
+      "epoch": 1.5367088607594936,
+      "grad_norm": 0.4781459717647683,
+      "learning_rate": 3.4949930896849324e-05,
+      "loss": 0.2849,
+      "step": 379
+    },
+    {
+      "epoch": 1.540759493670886,
+      "grad_norm": 0.4115864222239874,
+      "learning_rate": 3.491216790442899e-05,
+      "loss": 0.2819,
+      "step": 380
+    },
+    {
+      "epoch": 1.5448101265822785,
+      "grad_norm": 0.3895661364356451,
+      "learning_rate": 3.487428481145839e-05,
+      "loss": 0.2932,
+      "step": 381
+    },
+    {
+      "epoch": 1.548860759493671,
+      "grad_norm": 0.4021889062786402,
+      "learning_rate": 3.483628192304278e-05,
+      "loss": 0.2734,
+      "step": 382
+    },
+    {
+      "epoch": 1.5529113924050633,
+      "grad_norm": 0.37622704500885457,
+      "learning_rate": 3.479815954525219e-05,
+      "loss": 0.2953,
+      "step": 383
+    },
+    {
+      "epoch": 1.5569620253164556,
+      "grad_norm": 0.3748663259812325,
+      "learning_rate": 3.475991798511899e-05,
+      "loss": 0.2773,
+      "step": 384
+    },
+    {
+      "epoch": 1.561012658227848,
+      "grad_norm": 0.3924258380897512,
+      "learning_rate": 3.4721557550635464e-05,
+      "loss": 0.2815,
+      "step": 385
+    },
+    {
+      "epoch": 1.5650632911392406,
+      "grad_norm": 0.4229107097282921,
+      "learning_rate": 3.468307855075128e-05,
+      "loss": 0.2886,
+      "step": 386
+    },
+    {
+      "epoch": 1.5691139240506329,
+      "grad_norm": 0.4457800983536041,
+      "learning_rate": 3.4644481295371005e-05,
+      "loss": 0.2923,
+      "step": 387
+    },
+    {
+      "epoch": 1.5731645569620252,
+      "grad_norm": 0.42804020935239645,
+      "learning_rate": 3.460576609535163e-05,
+      "loss": 0.2841,
+      "step": 388
+    },
+    {
+      "epoch": 1.5772151898734177,
+      "grad_norm": 0.4159182695003156,
+      "learning_rate": 3.456693326250006e-05,
+      "loss": 0.2633,
+      "step": 389
+    },
+    {
+      "epoch": 1.5812658227848102,
+      "grad_norm": 0.464310876571883,
+      "learning_rate": 3.452798310957058e-05,
+      "loss": 0.2777,
+      "step": 390
+    },
+    {
+      "epoch": 1.5853164556962025,
+      "grad_norm": 0.47043963245918047,
+      "learning_rate": 3.4488915950262386e-05,
+      "loss": 0.2825,
+      "step": 391
+    },
+    {
+      "epoch": 1.5893670886075948,
+      "grad_norm": 0.3560658085384851,
+      "learning_rate": 3.4449732099216985e-05,
+      "loss": 0.2812,
+      "step": 392
+    },
+    {
+      "epoch": 1.5934177215189873,
+      "grad_norm": 0.44599473816969776,
+      "learning_rate": 3.441043187201574e-05,
+      "loss": 0.279,
+      "step": 393
+    },
+    {
+      "epoch": 1.5974683544303798,
+      "grad_norm": 0.3687860478743435,
+      "learning_rate": 3.437101558517728e-05,
+      "loss": 0.293,
+      "step": 394
+    },
+    {
+      "epoch": 1.6015189873417721,
+      "grad_norm": 0.45857618154757285,
+      "learning_rate": 3.433148355615496e-05,
+      "loss": 0.2702,
+      "step": 395
+    },
+    {
+      "epoch": 1.6055696202531644,
+      "grad_norm": 0.3995383999684415,
+      "learning_rate": 3.4291836103334294e-05,
+      "loss": 0.2853,
+      "step": 396
+    },
+    {
+      "epoch": 1.609620253164557,
+      "grad_norm": 0.4452663354244655,
+      "learning_rate": 3.425207354603043e-05,
+      "loss": 0.2903,
+      "step": 397
+    },
+    {
+      "epoch": 1.6136708860759494,
+      "grad_norm": 0.4280834990396901,
+      "learning_rate": 3.421219620448553e-05,
+      "loss": 0.286,
+      "step": 398
+    },
+    {
+      "epoch": 1.6177215189873417,
+      "grad_norm": 0.41515678010935303,
+      "learning_rate": 3.417220439986623e-05,
+      "loss": 0.2743,
+      "step": 399
+    },
+    {
+      "epoch": 1.621772151898734,
+      "grad_norm": 0.4461982766026146,
+      "learning_rate": 3.4132098454261024e-05,
+      "loss": 0.2866,
+      "step": 400
+    },
+    {
+      "epoch": 1.6258227848101265,
+      "grad_norm": 0.3624596480476297,
+      "learning_rate": 3.4091878690677676e-05,
+      "loss": 0.2804,
+      "step": 401
+    },
+    {
+      "epoch": 1.629873417721519,
+      "grad_norm": 0.44236159823930893,
+      "learning_rate": 3.405154543304065e-05,
+      "loss": 0.2767,
+      "step": 402
+    },
+    {
+      "epoch": 1.6339240506329114,
+      "grad_norm": 0.4023018123831375,
+      "learning_rate": 3.401109900618843e-05,
+      "loss": 0.2766,
+      "step": 403
+    },
+    {
+      "epoch": 1.6379746835443036,
+      "grad_norm": 0.44888738235265274,
+      "learning_rate": 3.3970539735870996e-05,
+      "loss": 0.2878,
+      "step": 404
+    },
+    {
+      "epoch": 1.6420253164556962,
+      "grad_norm": 0.3692550564683675,
+      "learning_rate": 3.392986794874714e-05,
+      "loss": 0.2782,
+      "step": 405
+    },
+    {
+      "epoch": 1.6460759493670887,
+      "grad_norm": 0.4434879828020414,
+      "learning_rate": 3.388908397238184e-05,
+      "loss": 0.2811,
+      "step": 406
+    },
+    {
+      "epoch": 1.650126582278481,
+      "grad_norm": 0.3839018232261325,
+      "learning_rate": 3.384818813524362e-05,
+      "loss": 0.2856,
+      "step": 407
+    },
+    {
+      "epoch": 1.6541772151898733,
+      "grad_norm": 0.37903416991089983,
+      "learning_rate": 3.380718076670195e-05,
+      "loss": 0.2776,
+      "step": 408
+    },
+    {
+      "epoch": 1.6582278481012658,
+      "grad_norm": 0.444153272815382,
+      "learning_rate": 3.376606219702454e-05,
+      "loss": 0.2813,
+      "step": 409
+    },
+    {
+      "epoch": 1.6622784810126583,
+      "grad_norm": 0.36319661269493647,
+      "learning_rate": 3.372483275737468e-05,
+      "loss": 0.2684,
+      "step": 410
+    },
+    {
+      "epoch": 1.6663291139240506,
+      "grad_norm": 0.3860447465008665,
+      "learning_rate": 3.368349277980861e-05,
+      "loss": 0.2863,
+      "step": 411
+    },
+    {
+      "epoch": 1.6703797468354429,
+      "grad_norm": 0.36769108088811153,
+      "learning_rate": 3.3642042597272844e-05,
+      "loss": 0.2836,
+      "step": 412
+    },
+    {
+      "epoch": 1.6744303797468354,
+      "grad_norm": 0.43169248622233664,
+      "learning_rate": 3.360048254360144e-05,
+      "loss": 0.3034,
+      "step": 413
+    },
+    {
+      "epoch": 1.678481012658228,
+      "grad_norm": 0.4037548451407967,
+      "learning_rate": 3.355881295351336e-05,
+      "loss": 0.2762,
+      "step": 414
+    },
+    {
+      "epoch": 1.6825316455696202,
+      "grad_norm": 0.39466368182621836,
+      "learning_rate": 3.351703416260975e-05,
+      "loss": 0.2859,
+      "step": 415
+    },
+    {
+      "epoch": 1.6865822784810125,
+      "grad_norm": 0.4178941462421841,
+      "learning_rate": 3.347514650737126e-05,
+      "loss": 0.2767,
+      "step": 416
+    },
+    {
+      "epoch": 1.690632911392405,
+      "grad_norm": 0.3633221527685955,
+      "learning_rate": 3.3433150325155295e-05,
+      "loss": 0.2693,
+      "step": 417
+    },
+    {
+      "epoch": 1.6946835443037975,
+      "grad_norm": 0.4451748949249502,
+      "learning_rate": 3.339104595419334e-05,
+      "loss": 0.2819,
+      "step": 418
+    },
+    {
+      "epoch": 1.6987341772151898,
+      "grad_norm": 0.38092744356352193,
+      "learning_rate": 3.3348833733588204e-05,
+      "loss": 0.2882,
+      "step": 419
+    },
+    {
+      "epoch": 1.7027848101265821,
+      "grad_norm": 0.39651235347602226,
+      "learning_rate": 3.3306514003311305e-05,
+      "loss": 0.2664,
+      "step": 420
+    },
+    {
+      "epoch": 1.7068354430379746,
+      "grad_norm": 0.3562360054487312,
+      "learning_rate": 3.326408710419996e-05,
+      "loss": 0.278,
+      "step": 421
+    },
+    {
+      "epoch": 1.7108860759493671,
+      "grad_norm": 0.41183600605483217,
+      "learning_rate": 3.322155337795454e-05,
+      "loss": 0.2815,
+      "step": 422
+    },
+    {
+      "epoch": 1.7149367088607594,
+      "grad_norm": 0.401241127431055,
+      "learning_rate": 3.317891316713587e-05,
+      "loss": 0.292,
+      "step": 423
+    },
+    {
+      "epoch": 1.7189873417721517,
+      "grad_norm": 0.4443197581197659,
+      "learning_rate": 3.313616681516231e-05,
+      "loss": 0.292,
+      "step": 424
+    },
+    {
+      "epoch": 1.7230379746835442,
+      "grad_norm": 0.5242489210597426,
+      "learning_rate": 3.309331466630713e-05,
+      "loss": 0.2969,
+      "step": 425
+    },
+    {
+      "epoch": 1.7270886075949368,
+      "grad_norm": 0.43846898953042646,
+      "learning_rate": 3.305035706569563e-05,
+      "loss": 0.2903,
+      "step": 426
+    },
+    {
+      "epoch": 1.7311392405063293,
+      "grad_norm": 0.4531005789972409,
+      "learning_rate": 3.3007294359302433e-05,
+      "loss": 0.2862,
+      "step": 427
+    },
+    {
+      "epoch": 1.7351898734177216,
+      "grad_norm": 0.35006866840124573,
+      "learning_rate": 3.296412689394864e-05,
+      "loss": 0.2821,
+      "step": 428
+    },
+    {
+      "epoch": 1.7392405063291139,
+      "grad_norm": 0.4143133094864929,
+      "learning_rate": 3.292085501729909e-05,
+      "loss": 0.2883,
+      "step": 429
+    },
+    {
+      "epoch": 1.7432911392405064,
+      "grad_norm": 0.36285369181209504,
+      "learning_rate": 3.2877479077859534e-05,
+      "loss": 0.2793,
+      "step": 430
+    },
+    {
+      "epoch": 1.747341772151899,
+      "grad_norm": 0.3586371679575702,
+      "learning_rate": 3.283399942497381e-05,
+      "loss": 0.2748,
+      "step": 431
+    },
+    {
+      "epoch": 1.7513924050632912,
+      "grad_norm": 0.4097690426614037,
+      "learning_rate": 3.279041640882108e-05,
+      "loss": 0.2974,
+      "step": 432
+    },
+    {
+      "epoch": 1.7554430379746835,
+      "grad_norm": 0.4057920426274522,
+      "learning_rate": 3.2746730380412964e-05,
+      "loss": 0.2682,
+      "step": 433
+    },
+    {
+      "epoch": 1.759493670886076,
+      "grad_norm": 0.40773735865245336,
+      "learning_rate": 3.2702941691590726e-05,
+      "loss": 0.2909,
+      "step": 434
+    },
+    {
+      "epoch": 1.7635443037974685,
+      "grad_norm": 0.4046634730226732,
+      "learning_rate": 3.265905069502244e-05,
+      "loss": 0.2889,
+      "step": 435
+    },
+    {
+      "epoch": 1.7675949367088608,
+      "grad_norm": 0.45560560291922064,
+      "learning_rate": 3.261505774420016e-05,
+      "loss": 0.2792,
+      "step": 436
+    },
+    {
+      "epoch": 1.771645569620253,
+      "grad_norm": 0.40013555427261305,
+      "learning_rate": 3.257096319343707e-05,
+      "loss": 0.2847,
+      "step": 437
+    },
+    {
+      "epoch": 1.7756962025316456,
+      "grad_norm": 0.39087449535031976,
+      "learning_rate": 3.2526767397864614e-05,
+      "loss": 0.2748,
+      "step": 438
+    },
+    {
+      "epoch": 1.7797468354430381,
+      "grad_norm": 0.5214504654856412,
+      "learning_rate": 3.248247071342966e-05,
+      "loss": 0.2948,
+      "step": 439
+    },
+    {
+      "epoch": 1.7837974683544304,
+      "grad_norm": 0.3687024137884713,
+      "learning_rate": 3.243807349689161e-05,
+      "loss": 0.2797,
+      "step": 440
+    },
+    {
+      "epoch": 1.7878481012658227,
+      "grad_norm": 0.4807819385108898,
+      "learning_rate": 3.2393576105819544e-05,
+      "loss": 0.308,
+      "step": 441
+    },
+    {
+      "epoch": 1.7918987341772152,
+      "grad_norm": 0.39122716712896827,
+      "learning_rate": 3.2348978898589333e-05,
+      "loss": 0.2778,
+      "step": 442
+    },
+    {
+      "epoch": 1.7959493670886078,
+      "grad_norm": 0.5747259638957977,
+      "learning_rate": 3.230428223438075e-05,
+      "loss": 0.2867,
+      "step": 443
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.3645339666378964,
+      "learning_rate": 3.225948647317459e-05,
+      "loss": 0.2736,
+      "step": 444
+    },
+    {
+      "epoch": 1.8040506329113923,
+      "grad_norm": 0.4967648018348401,
+      "learning_rate": 3.2214591975749745e-05,
+      "loss": 0.2716,
+      "step": 445
+    },
+    {
+      "epoch": 1.8081012658227849,
+      "grad_norm": 0.36056720286457467,
+      "learning_rate": 3.216959910368034e-05,
+      "loss": 0.2764,
+      "step": 446
+    },
+    {
+      "epoch": 1.8121518987341774,
+      "grad_norm": 0.46742065942119937,
+      "learning_rate": 3.212450821933277e-05,
+      "loss": 0.2665,
+      "step": 447
+    },
+    {
+      "epoch": 1.8162025316455697,
+      "grad_norm": 0.357575228946387,
+      "learning_rate": 3.207931968586281e-05,
+      "loss": 0.2841,
+      "step": 448
+    },
+    {
+      "epoch": 1.820253164556962,
+      "grad_norm": 0.42328707376500146,
+      "learning_rate": 3.203403386721272e-05,
+      "loss": 0.2822,
+      "step": 449
+    },
+    {
+      "epoch": 1.8243037974683545,
+      "grad_norm": 0.4391707111009717,
+      "learning_rate": 3.1988651128108245e-05,
+      "loss": 0.2816,
+      "step": 450
+    },
+    {
+      "epoch": 1.828354430379747,
+      "grad_norm": 0.37397528345174147,
+      "learning_rate": 3.194317183405573e-05,
+      "loss": 0.2879,
+      "step": 451
+    },
+    {
+      "epoch": 1.8324050632911393,
+      "grad_norm": 0.3759433462825667,
+      "learning_rate": 3.189759635133914e-05,
+      "loss": 0.2909,
+      "step": 452
+    },
+    {
+      "epoch": 1.8364556962025316,
+      "grad_norm": 0.3875942139294186,
+      "learning_rate": 3.185192504701718e-05,
+      "loss": 0.2783,
+      "step": 453
+    },
+    {
+      "epoch": 1.840506329113924,
+      "grad_norm": 0.4350612084736872,
+      "learning_rate": 3.1806158288920234e-05,
+      "loss": 0.2856,
+      "step": 454
+    },
+    {
+      "epoch": 1.8445569620253166,
+      "grad_norm": 0.3595597218079778,
+      "learning_rate": 3.1760296445647477e-05,
+      "loss": 0.2817,
+      "step": 455
+    },
+    {
+      "epoch": 1.848607594936709,
+      "grad_norm": 0.3764106232923682,
+      "learning_rate": 3.1714339886563896e-05,
+      "loss": 0.2912,
+      "step": 456
+    },
+    {
+      "epoch": 1.8526582278481012,
+      "grad_norm": 0.3838024347690083,
+      "learning_rate": 3.166828898179731e-05,
+      "loss": 0.2866,
+      "step": 457
+    },
+    {
+      "epoch": 1.8567088607594937,
+      "grad_norm": 0.36553643814355896,
+      "learning_rate": 3.162214410223536e-05,
+      "loss": 0.2815,
+      "step": 458
+    },
+    {
+      "epoch": 1.8607594936708862,
+      "grad_norm": 0.3932147654195156,
+      "learning_rate": 3.157590561952257e-05,
+      "loss": 0.2695,
+      "step": 459
+    },
+    {
+      "epoch": 1.8648101265822785,
+      "grad_norm": 0.40031022991451426,
+      "learning_rate": 3.152957390605732e-05,
+      "loss": 0.2871,
+      "step": 460
+    },
+    {
+      "epoch": 1.8688607594936708,
+      "grad_norm": 0.37734852317406636,
+      "learning_rate": 3.148314933498886e-05,
+      "loss": 0.2719,
+      "step": 461
+    },
+    {
+      "epoch": 1.8729113924050633,
+      "grad_norm": 0.3992868194058598,
+      "learning_rate": 3.143663228021431e-05,
+      "loss": 0.2814,
+      "step": 462
+    },
+    {
+      "epoch": 1.8769620253164558,
+      "grad_norm": 0.42590101807148933,
+      "learning_rate": 3.1390023116375624e-05,
+      "loss": 0.2956,
+      "step": 463
+    },
+    {
+      "epoch": 1.8810126582278481,
+      "grad_norm": 0.34991017705972977,
+      "learning_rate": 3.134332221885661e-05,
+      "loss": 0.2937,
+      "step": 464
+    },
+    {
+      "epoch": 1.8850632911392404,
+      "grad_norm": 0.42054425127097844,
+      "learning_rate": 3.129652996377987e-05,
+      "loss": 0.3048,
+      "step": 465
+    },
+    {
+      "epoch": 1.889113924050633,
+      "grad_norm": 0.37456467626658535,
+      "learning_rate": 3.12496467280038e-05,
+      "loss": 0.2676,
+      "step": 466
+    },
+    {
+      "epoch": 1.8931645569620255,
+      "grad_norm": 0.36156872007322755,
+      "learning_rate": 3.120267288911952e-05,
+      "loss": 0.2803,
+      "step": 467
+    },
+    {
+      "epoch": 1.8972151898734178,
+      "grad_norm": 0.38177336302063897,
+      "learning_rate": 3.11556088254479e-05,
+      "loss": 0.2853,
+      "step": 468
+    },
+    {
+      "epoch": 1.90126582278481,
+      "grad_norm": 0.38205231113609195,
+      "learning_rate": 3.11084549160364e-05,
+      "loss": 0.2816,
+      "step": 469
+    },
+    {
+      "epoch": 1.9053164556962026,
+      "grad_norm": 0.3787901580146895,
+      "learning_rate": 3.106121154065615e-05,
+      "loss": 0.2775,
+      "step": 470
+    },
+    {
+      "epoch": 1.909367088607595,
+      "grad_norm": 0.42909812405162984,
+      "learning_rate": 3.1013879079798805e-05,
+      "loss": 0.2827,
+      "step": 471
+    },
+    {
+      "epoch": 1.9134177215189874,
+      "grad_norm": 0.3706067448091144,
+      "learning_rate": 3.096645791467348e-05,
+      "loss": 0.2846,
+      "step": 472
+    },
+    {
+      "epoch": 1.9174683544303797,
+      "grad_norm": 0.37394987038791694,
+      "learning_rate": 3.091894842720373e-05,
+      "loss": 0.2784,
+      "step": 473
+    },
+    {
+      "epoch": 1.9215189873417722,
+      "grad_norm": 0.39122397544385595,
+      "learning_rate": 3.0871351000024425e-05,
+      "loss": 0.2904,
+      "step": 474
+    },
+    {
+      "epoch": 1.9255696202531647,
+      "grad_norm": 0.3597412209530651,
+      "learning_rate": 3.0823666016478716e-05,
+      "loss": 0.2773,
+      "step": 475
+    },
+    {
+      "epoch": 1.929620253164557,
+      "grad_norm": 0.4111617619544528,
+      "learning_rate": 3.0775893860614896e-05,
+      "loss": 0.2802,
+      "step": 476
+    },
+    {
+      "epoch": 1.9336708860759493,
+      "grad_norm": 0.3809851578959392,
+      "learning_rate": 3.0728034917183336e-05,
+      "loss": 0.2814,
+      "step": 477
+    },
+    {
+      "epoch": 1.9377215189873418,
+      "grad_norm": 0.34807573481460313,
+      "learning_rate": 3.06800895716334e-05,
+      "loss": 0.2842,
+      "step": 478
+    },
+    {
+      "epoch": 1.9417721518987343,
+      "grad_norm": 0.3676288872962272,
+      "learning_rate": 3.063205821011029e-05,
+      "loss": 0.2871,
+      "step": 479
+    },
+    {
+      "epoch": 1.9458227848101266,
+      "grad_norm": 0.3800475989293683,
+      "learning_rate": 3.0583941219452016e-05,
+      "loss": 0.2684,
+      "step": 480
+    },
+    {
+      "epoch": 1.949873417721519,
+      "grad_norm": 0.33369561946096643,
+      "learning_rate": 3.053573898718618e-05,
+      "loss": 0.2775,
+      "step": 481
+    },
+    {
+      "epoch": 1.9539240506329114,
+      "grad_norm": 0.38033412686542045,
+      "learning_rate": 3.0487451901526956e-05,
+      "loss": 0.2799,
+      "step": 482
+    },
+    {
+      "epoch": 1.957974683544304,
+      "grad_norm": 0.3384185745896955,
+      "learning_rate": 3.0439080351371875e-05,
+      "loss": 0.2625,
+      "step": 483
+    },
+    {
+      "epoch": 1.9620253164556962,
+      "grad_norm": 0.34187429825748206,
+      "learning_rate": 3.0390624726298764e-05,
+      "loss": 0.2848,
+      "step": 484
+    },
+    {
+      "epoch": 1.9660759493670885,
+      "grad_norm": 0.3874469785270207,
+      "learning_rate": 3.034208541656255e-05,
+      "loss": 0.2923,
+      "step": 485
+    },
+    {
+      "epoch": 1.970126582278481,
+      "grad_norm": 0.3757558645622096,
+      "learning_rate": 3.029346281309218e-05,
+      "loss": 0.2852,
+      "step": 486
+    },
+    {
+      "epoch": 1.9741772151898735,
+      "grad_norm": 0.3899587637715491,
+      "learning_rate": 3.0244757307487415e-05,
+      "loss": 0.279,
+      "step": 487
+    },
+    {
+      "epoch": 1.9782278481012658,
+      "grad_norm": 0.4195009633405334,
+      "learning_rate": 3.019596929201569e-05,
+      "loss": 0.2712,
+      "step": 488
+    },
+    {
+      "epoch": 1.9822784810126581,
+      "grad_norm": 0.4891229953426986,
+      "learning_rate": 3.0147099159608985e-05,
+      "loss": 0.2954,
+      "step": 489
+    },
+    {
+      "epoch": 1.9863291139240506,
+      "grad_norm": 0.38370901827948845,
+      "learning_rate": 3.0098147303860616e-05,
+      "loss": 0.2854,
+      "step": 490
+    },
+    {
+      "epoch": 1.9903797468354432,
+      "grad_norm": 0.35161642734304976,
+      "learning_rate": 3.0049114119022117e-05,
+      "loss": 0.267,
+      "step": 491
+    },
+    {
+      "epoch": 1.9944303797468355,
+      "grad_norm": 0.3853425723757686,
+      "learning_rate": 3.0000000000000004e-05,
+      "loss": 0.2761,
+      "step": 492
+    },
+    {
+      "epoch": 1.9984810126582278,
+      "grad_norm": 0.3590194174905781,
+      "learning_rate": 2.995080534235264e-05,
+      "loss": 0.2759,
+      "step": 493
+    },
+    {
+      "epoch": 2.0040506329113925,
+      "grad_norm": 0.5067968978326327,
+      "learning_rate": 2.9901530542287044e-05,
+      "loss": 0.2122,
+      "step": 494
+    },
+    {
+      "epoch": 2.008101265822785,
+      "grad_norm": 0.3560573608285091,
+      "learning_rate": 2.9852175996655676e-05,
+      "loss": 0.1936,
+      "step": 495
+    },
+    {
+      "epoch": 2.012151898734177,
+      "grad_norm": 0.5986668862382186,
+      "learning_rate": 2.980274210295326e-05,
+      "loss": 0.1922,
+      "step": 496
+    },
+    {
+      "epoch": 2.0162025316455696,
+      "grad_norm": 0.4085218624798357,
+      "learning_rate": 2.9753229259313578e-05,
+      "loss": 0.2038,
+      "step": 497
+    },
+    {
+      "epoch": 2.020253164556962,
+      "grad_norm": 0.3995984456257672,
+      "learning_rate": 2.9703637864506274e-05,
+      "loss": 0.2014,
+      "step": 498
+    },
+    {
+      "epoch": 2.0243037974683546,
+      "grad_norm": 0.3819067568703095,
+      "learning_rate": 2.965396831793362e-05,
+      "loss": 0.1925,
+      "step": 499
+    },
+    {
+      "epoch": 2.0283544303797467,
+      "grad_norm": 0.4010082329351815,
+      "learning_rate": 2.9604221019627316e-05,
+      "loss": 0.1986,
+      "step": 500
+    },
+    {
+      "epoch": 2.0324050632911392,
+      "grad_norm": 0.36809904499464635,
+      "learning_rate": 2.955439637024526e-05,
+      "loss": 0.191,
+      "step": 501
+    },
+    {
+      "epoch": 2.0364556962025318,
+      "grad_norm": 0.42171985935107353,
+      "learning_rate": 2.9504494771068334e-05,
+      "loss": 0.2024,
+      "step": 502
+    },
+    {
+      "epoch": 2.0405063291139243,
+      "grad_norm": 0.37801082553191967,
+      "learning_rate": 2.9454516623997156e-05,
+      "loss": 0.1846,
+      "step": 503
+    },
+    {
+      "epoch": 2.0445569620253163,
+      "grad_norm": 0.4010889015046548,
+      "learning_rate": 2.9404462331548847e-05,
+      "loss": 0.1877,
+      "step": 504
+    },
+    {
+      "epoch": 2.048607594936709,
+      "grad_norm": 0.374022878880801,
+      "learning_rate": 2.93543322968538e-05,
+      "loss": 0.176,
+      "step": 505
+    },
+    {
+      "epoch": 2.0526582278481014,
+      "grad_norm": 0.43230991217997705,
+      "learning_rate": 2.9304126923652428e-05,
+      "loss": 0.1918,
+      "step": 506
+    },
+    {
+      "epoch": 2.056708860759494,
+      "grad_norm": 0.39530144470710715,
+      "learning_rate": 2.9253846616291896e-05,
+      "loss": 0.181,
+      "step": 507
+    },
+    {
+      "epoch": 2.060759493670886,
+      "grad_norm": 0.4043333737145639,
+      "learning_rate": 2.9203491779722896e-05,
+      "loss": 0.1829,
+      "step": 508
+    },
+    {
+      "epoch": 2.0648101265822785,
+      "grad_norm": 0.4000785739006497,
+      "learning_rate": 2.9153062819496357e-05,
+      "loss": 0.1978,
+      "step": 509
+    },
+    {
+      "epoch": 2.068860759493671,
+      "grad_norm": 0.3869961911074061,
+      "learning_rate": 2.9102560141760178e-05,
+      "loss": 0.179,
+      "step": 510
+    },
+    {
+      "epoch": 2.0729113924050635,
+      "grad_norm": 0.3864128449488987,
+      "learning_rate": 2.9051984153256004e-05,
+      "loss": 0.1888,
+      "step": 511
+    },
+    {
+      "epoch": 2.0769620253164556,
+      "grad_norm": 0.4323823447774737,
+      "learning_rate": 2.900133526131588e-05,
+      "loss": 0.1985,
+      "step": 512
+    },
+    {
+      "epoch": 2.081012658227848,
+      "grad_norm": 0.36880504028164107,
+      "learning_rate": 2.8950613873859025e-05,
+      "loss": 0.1869,
+      "step": 513
+    },
+    {
+      "epoch": 2.0850632911392406,
+      "grad_norm": 0.4334929191337118,
+      "learning_rate": 2.8899820399388515e-05,
+      "loss": 0.1844,
+      "step": 514
+    },
+    {
+      "epoch": 2.089113924050633,
+      "grad_norm": 0.4116226145983687,
+      "learning_rate": 2.8848955246988012e-05,
+      "loss": 0.1923,
+      "step": 515
+    },
+    {
+      "epoch": 2.093164556962025,
+      "grad_norm": 0.4020617674471567,
+      "learning_rate": 2.879801882631847e-05,
+      "loss": 0.2064,
+      "step": 516
+    },
+    {
+      "epoch": 2.0972151898734177,
+      "grad_norm": 0.4382424390592082,
+      "learning_rate": 2.8747011547614808e-05,
+      "loss": 0.1898,
+      "step": 517
+    },
+    {
+      "epoch": 2.1012658227848102,
+      "grad_norm": 0.3549374388702027,
+      "learning_rate": 2.8695933821682635e-05,
+      "loss": 0.1858,
+      "step": 518
+    },
+    {
+      "epoch": 2.1053164556962027,
+      "grad_norm": 0.41169854484397866,
+      "learning_rate": 2.864478605989494e-05,
+      "loss": 0.2135,
+      "step": 519
+    },
+    {
+      "epoch": 2.109367088607595,
+      "grad_norm": 0.394069744395246,
+      "learning_rate": 2.8593568674188765e-05,
+      "loss": 0.1858,
+      "step": 520
+    },
+    {
+      "epoch": 2.1134177215189873,
+      "grad_norm": 0.365832465644085,
+      "learning_rate": 2.8542282077061892e-05,
+      "loss": 0.2108,
+      "step": 521
+    },
+    {
+      "epoch": 2.11746835443038,
+      "grad_norm": 0.38620290002317886,
+      "learning_rate": 2.8490926681569523e-05,
+      "loss": 0.2029,
+      "step": 522
+    },
+    {
+      "epoch": 2.1215189873417724,
+      "grad_norm": 0.3978430959915937,
+      "learning_rate": 2.8439502901320956e-05,
+      "loss": 0.195,
+      "step": 523
+    },
+    {
+      "epoch": 2.1255696202531644,
+      "grad_norm": 0.3613714597495287,
+      "learning_rate": 2.8388011150476237e-05,
+      "loss": 0.1864,
+      "step": 524
+    },
+    {
+      "epoch": 2.129620253164557,
+      "grad_norm": 0.3917729993428087,
+      "learning_rate": 2.8336451843742866e-05,
+      "loss": 0.1939,
+      "step": 525
+    },
+    {
+      "epoch": 2.1336708860759495,
+      "grad_norm": 0.3449201401352711,
+      "learning_rate": 2.8284825396372387e-05,
+      "loss": 0.1958,
+      "step": 526
+    },
+    {
+      "epoch": 2.137721518987342,
+      "grad_norm": 0.3827716975573245,
+      "learning_rate": 2.8233132224157132e-05,
+      "loss": 0.1864,
+      "step": 527
+    },
+    {
+      "epoch": 2.141772151898734,
+      "grad_norm": 0.3914384746241429,
+      "learning_rate": 2.8181372743426805e-05,
+      "loss": 0.1778,
+      "step": 528
+    },
+    {
+      "epoch": 2.1458227848101266,
+      "grad_norm": 0.39319737220402196,
+      "learning_rate": 2.8129547371045128e-05,
+      "loss": 0.1973,
+      "step": 529
+    },
+    {
+      "epoch": 2.149873417721519,
+      "grad_norm": 0.3812334938569515,
+      "learning_rate": 2.8077656524406534e-05,
+      "loss": 0.1847,
+      "step": 530
+    },
+    {
+      "epoch": 2.1539240506329116,
+      "grad_norm": 0.402561398769873,
+      "learning_rate": 2.802570062143278e-05,
+      "loss": 0.1928,
+      "step": 531
+    },
+    {
+      "epoch": 2.1579746835443037,
+      "grad_norm": 0.37287450690830337,
+      "learning_rate": 2.7973680080569555e-05,
+      "loss": 0.1974,
+      "step": 532
+    },
+    {
+      "epoch": 2.162025316455696,
+      "grad_norm": 0.37868340992731553,
+      "learning_rate": 2.792159532078314e-05,
+      "loss": 0.1843,
+      "step": 533
+    },
+    {
+      "epoch": 2.1660759493670887,
+      "grad_norm": 0.4509185239209555,
+      "learning_rate": 2.7869446761557033e-05,
+      "loss": 0.187,
+      "step": 534
+    },
+    {
+      "epoch": 2.170126582278481,
+      "grad_norm": 0.3472787220041161,
+      "learning_rate": 2.781723482288857e-05,
+      "loss": 0.1867,
+      "step": 535
+    },
+    {
+      "epoch": 2.1741772151898733,
+      "grad_norm": 0.4291980019776311,
+      "learning_rate": 2.7764959925285517e-05,
+      "loss": 0.1867,
+      "step": 536
+    },
+    {
+      "epoch": 2.178227848101266,
+      "grad_norm": 0.3526663342569369,
+      "learning_rate": 2.771262248976272e-05,
+      "loss": 0.1983,
+      "step": 537
+    },
+    {
+      "epoch": 2.1822784810126583,
+      "grad_norm": 0.3853767278044529,
+      "learning_rate": 2.7660222937838677e-05,
+      "loss": 0.1741,
+      "step": 538
+    },
+    {
+      "epoch": 2.186329113924051,
+      "grad_norm": 0.45546251429744505,
+      "learning_rate": 2.7607761691532186e-05,
+      "loss": 0.1985,
+      "step": 539
+    },
+    {
+      "epoch": 2.190379746835443,
+      "grad_norm": 0.38030854970582967,
+      "learning_rate": 2.7555239173358916e-05,
+      "loss": 0.1926,
+      "step": 540
+    },
+    {
+      "epoch": 2.1944303797468354,
+      "grad_norm": 0.40822521638510656,
+      "learning_rate": 2.7502655806328e-05,
+      "loss": 0.1818,
+      "step": 541
+    },
+    {
+      "epoch": 2.198481012658228,
+      "grad_norm": 0.4030889446088792,
+      "learning_rate": 2.7450012013938648e-05,
+      "loss": 0.1992,
+      "step": 542
+    },
+    {
+      "epoch": 2.2025316455696204,
+      "grad_norm": 0.37781999177708,
+      "learning_rate": 2.739730822017673e-05,
+      "loss": 0.1912,
+      "step": 543
+    },
+    {
+      "epoch": 2.2065822784810125,
+      "grad_norm": 0.3799446258649155,
+      "learning_rate": 2.7344544849511355e-05,
+      "loss": 0.2103,
+      "step": 544
+    },
+    {
+      "epoch": 2.210632911392405,
+      "grad_norm": 0.34279797473781004,
+      "learning_rate": 2.7291722326891456e-05,
+      "loss": 0.2033,
+      "step": 545
+    },
+    {
+      "epoch": 2.2146835443037975,
+      "grad_norm": 0.31866951616221734,
+      "learning_rate": 2.723884107774236e-05,
+      "loss": 0.1945,
+      "step": 546
+    },
+    {
+      "epoch": 2.21873417721519,
+      "grad_norm": 0.35807400692629565,
+      "learning_rate": 2.718590152796239e-05,
+      "loss": 0.1957,
+      "step": 547
+    },
+    {
+      "epoch": 2.222784810126582,
+      "grad_norm": 0.3433517045868625,
+      "learning_rate": 2.71329041039194e-05,
+      "loss": 0.2081,
+      "step": 548
+    },
+    {
+      "epoch": 2.2268354430379746,
+      "grad_norm": 0.35238840161471363,
+      "learning_rate": 2.7079849232447357e-05,
+      "loss": 0.1881,
+      "step": 549
+    },
+    {
+      "epoch": 2.230886075949367,
+      "grad_norm": 0.3496556827004259,
+      "learning_rate": 2.7026737340842895e-05,
+      "loss": 0.2011,
+      "step": 550
+    },
+    {
+      "epoch": 2.2349367088607597,
+      "grad_norm": 0.33780250444658483,
+      "learning_rate": 2.697356885686189e-05,
+      "loss": 0.2068,
+      "step": 551
+    },
+    {
+      "epoch": 2.2389873417721518,
+      "grad_norm": 0.4130537371208423,
+      "learning_rate": 2.6920344208716014e-05,
+      "loss": 0.2093,
+      "step": 552
+    },
+    {
+      "epoch": 2.2430379746835443,
+      "grad_norm": 0.3448124823883893,
+      "learning_rate": 2.6867063825069252e-05,
+      "loss": 0.1914,
+      "step": 553
+    },
+    {
+      "epoch": 2.247088607594937,
+      "grad_norm": 0.40395490882672497,
+      "learning_rate": 2.6813728135034494e-05,
+      "loss": 0.2081,
+      "step": 554
+    },
+    {
+      "epoch": 2.2511392405063293,
+      "grad_norm": 0.3465537831303768,
+      "learning_rate": 2.6760337568170056e-05,
+      "loss": 0.1876,
+      "step": 555
+    },
+    {
+      "epoch": 2.2551898734177214,
+      "grad_norm": 0.368854340850213,
+      "learning_rate": 2.6706892554476226e-05,
+      "loss": 0.2,
+      "step": 556
+    },
+    {
+      "epoch": 2.259240506329114,
+      "grad_norm": 0.3104649789752454,
+      "learning_rate": 2.6653393524391795e-05,
+      "loss": 0.1963,
+      "step": 557
+    },
+    {
+      "epoch": 2.2632911392405064,
+      "grad_norm": 0.3606226260342972,
+      "learning_rate": 2.6599840908790592e-05,
+      "loss": 0.2093,
+      "step": 558
+    },
+    {
+      "epoch": 2.267341772151899,
+      "grad_norm": 0.3622411336033707,
+      "learning_rate": 2.6546235138978028e-05,
+      "loss": 0.1886,
+      "step": 559
+    },
+    {
+      "epoch": 2.271392405063291,
+      "grad_norm": 0.373455206550521,
+      "learning_rate": 2.6492576646687597e-05,
+      "loss": 0.1889,
+      "step": 560
+    },
+    {
+      "epoch": 2.2754430379746835,
+      "grad_norm": 0.3892117132313476,
+      "learning_rate": 2.6438865864077425e-05,
+      "loss": 0.2046,
+      "step": 561
+    },
+    {
+      "epoch": 2.279493670886076,
+      "grad_norm": 0.38261599808177327,
+      "learning_rate": 2.6385103223726766e-05,
+      "loss": 0.1902,
+      "step": 562
+    },
+    {
+      "epoch": 2.2835443037974685,
+      "grad_norm": 0.36406876585648146,
+      "learning_rate": 2.6331289158632537e-05,
+      "loss": 0.199,
+      "step": 563
+    },
+    {
+      "epoch": 2.2875949367088606,
+      "grad_norm": 0.39060917527738037,
+      "learning_rate": 2.6277424102205817e-05,
+      "loss": 0.1883,
+      "step": 564
+    },
+    {
+      "epoch": 2.291645569620253,
+      "grad_norm": 0.3437726543442497,
+      "learning_rate": 2.6223508488268374e-05,
+      "loss": 0.1884,
+      "step": 565
+    },
+    {
+      "epoch": 2.2956962025316456,
+      "grad_norm": 0.3954895966951088,
+      "learning_rate": 2.6169542751049148e-05,
+      "loss": 0.2136,
+      "step": 566
+    },
+    {
+      "epoch": 2.299746835443038,
+      "grad_norm": 0.38053198850672537,
+      "learning_rate": 2.6115527325180754e-05,
+      "loss": 0.1905,
+      "step": 567
+    },
+    {
+      "epoch": 2.3037974683544302,
+      "grad_norm": 0.362245117948773,
+      "learning_rate": 2.606146264569603e-05,
+      "loss": 0.1924,
+      "step": 568
+    },
+    {
+      "epoch": 2.3078481012658227,
+      "grad_norm": 0.41104744764113577,
+      "learning_rate": 2.6007349148024447e-05,
+      "loss": 0.1958,
+      "step": 569
+    },
+    {
+      "epoch": 2.3118987341772153,
+      "grad_norm": 0.35947218273144277,
+      "learning_rate": 2.5953187267988694e-05,
+      "loss": 0.1993,
+      "step": 570
+    },
+    {
+      "epoch": 2.3159493670886078,
+      "grad_norm": 0.3651808982013425,
+      "learning_rate": 2.5898977441801097e-05,
+      "loss": 0.2081,
+      "step": 571
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 0.38242910554021137,
+      "learning_rate": 2.584472010606015e-05,
+      "loss": 0.201,
+      "step": 572
+    },
+    {
+      "epoch": 2.3240506329113924,
+      "grad_norm": 0.37501107963852537,
+      "learning_rate": 2.5790415697746976e-05,
+      "loss": 0.1851,
+      "step": 573
+    },
+    {
+      "epoch": 2.328101265822785,
+      "grad_norm": 0.35993648006120443,
+      "learning_rate": 2.5736064654221808e-05,
+      "loss": 0.211,
+      "step": 574
+    },
+    {
+      "epoch": 2.3321518987341774,
+      "grad_norm": 0.3799105152759867,
+      "learning_rate": 2.568166741322048e-05,
+      "loss": 0.2047,
+      "step": 575
+    },
+    {
+      "epoch": 2.3362025316455695,
+      "grad_norm": 0.36306070342584695,
+      "learning_rate": 2.56272244128509e-05,
+      "loss": 0.2042,
+      "step": 576
+    },
+    {
+      "epoch": 2.340253164556962,
+      "grad_norm": 0.3922956733579463,
+      "learning_rate": 2.55727360915895e-05,
+      "loss": 0.1965,
+      "step": 577
+    },
+    {
+      "epoch": 2.3443037974683545,
+      "grad_norm": 0.4086945188690434,
+      "learning_rate": 2.5518202888277734e-05,
+      "loss": 0.1987,
+      "step": 578
+    },
+    {
+      "epoch": 2.348354430379747,
+      "grad_norm": 0.35437521665471183,
+      "learning_rate": 2.5463625242118523e-05,
+      "loss": 0.2032,
+      "step": 579
+    },
+    {
+      "epoch": 2.352405063291139,
+      "grad_norm": 0.3499814737518172,
+      "learning_rate": 2.5409003592672723e-05,
+      "loss": 0.1884,
+      "step": 580
+    },
+    {
+      "epoch": 2.3564556962025316,
+      "grad_norm": 0.45230116810363835,
+      "learning_rate": 2.535433837985559e-05,
+      "loss": 0.2022,
+      "step": 581
+    },
+    {
+      "epoch": 2.360506329113924,
+      "grad_norm": 0.3569210159678209,
+      "learning_rate": 2.529963004393324e-05,
+      "loss": 0.2024,
+      "step": 582
+    },
+    {
+      "epoch": 2.3645569620253166,
+      "grad_norm": 0.34618914696938685,
+      "learning_rate": 2.524487902551908e-05,
+      "loss": 0.1959,
+      "step": 583
+    },
+    {
+      "epoch": 2.3686075949367087,
+      "grad_norm": 0.38833589841582794,
+      "learning_rate": 2.519008576557029e-05,
+      "loss": 0.1762,
+      "step": 584
+    },
+    {
+      "epoch": 2.372658227848101,
+      "grad_norm": 0.3325760129670432,
+      "learning_rate": 2.5135250705384254e-05,
+      "loss": 0.1967,
+      "step": 585
+    },
+    {
+      "epoch": 2.3767088607594937,
+      "grad_norm": 0.35644347573184443,
+      "learning_rate": 2.5080374286595007e-05,
+      "loss": 0.1935,
+      "step": 586
+    },
+    {
+      "epoch": 2.3807594936708862,
+      "grad_norm": 0.38290344550109473,
+      "learning_rate": 2.5025456951169677e-05,
+      "loss": 0.2027,
+      "step": 587
+    },
+    {
+      "epoch": 2.3848101265822783,
+      "grad_norm": 0.33374798638900777,
+      "learning_rate": 2.4970499141404942e-05,
+      "loss": 0.2083,
+      "step": 588
+    },
+    {
+      "epoch": 2.388860759493671,
+      "grad_norm": 0.33760487295112207,
+      "learning_rate": 2.491550129992345e-05,
+      "loss": 0.2052,
+      "step": 589
+    },
+    {
+      "epoch": 2.3929113924050633,
+      "grad_norm": 0.32379276284824415,
+      "learning_rate": 2.486046386967024e-05,
+      "loss": 0.1928,
+      "step": 590
+    },
+    {
+      "epoch": 2.396962025316456,
+      "grad_norm": 0.31183985603448294,
+      "learning_rate": 2.4805387293909214e-05,
+      "loss": 0.1949,
+      "step": 591
+    },
+    {
+      "epoch": 2.401012658227848,
+      "grad_norm": 0.336682340969239,
+      "learning_rate": 2.4750272016219552e-05,
+      "loss": 0.1961,
+      "step": 592
+    },
+    {
+      "epoch": 2.4050632911392404,
+      "grad_norm": 0.33184977594090836,
+      "learning_rate": 2.4695118480492114e-05,
+      "loss": 0.1946,
+      "step": 593
+    },
+    {
+      "epoch": 2.409113924050633,
+      "grad_norm": 0.36537229419636374,
+      "learning_rate": 2.4639927130925898e-05,
+      "loss": 0.1894,
+      "step": 594
+    },
+    {
+      "epoch": 2.4131645569620255,
+      "grad_norm": 0.3589861005152909,
+      "learning_rate": 2.458469841202444e-05,
+      "loss": 0.2029,
+      "step": 595
+    },
+    {
+      "epoch": 2.4172151898734175,
+      "grad_norm": 0.32671895887973634,
+      "learning_rate": 2.452943276859226e-05,
+      "loss": 0.1937,
+      "step": 596
+    },
+    {
+      "epoch": 2.42126582278481,
+      "grad_norm": 0.3423627656889005,
+      "learning_rate": 2.447413064573125e-05,
+      "loss": 0.1986,
+      "step": 597
+    },
+    {
+      "epoch": 2.4253164556962026,
+      "grad_norm": 0.3498489896335323,
+      "learning_rate": 2.4418792488837095e-05,
+      "loss": 0.2084,
+      "step": 598
+    },
+    {
+      "epoch": 2.429367088607595,
+      "grad_norm": 0.3405012755372256,
+      "learning_rate": 2.4363418743595713e-05,
+      "loss": 0.1953,
+      "step": 599
+    },
+    {
+      "epoch": 2.433417721518987,
+      "grad_norm": 0.32343647123039493,
+      "learning_rate": 2.430800985597963e-05,
+      "loss": 0.1952,
+      "step": 600
+    },
+    {
+      "epoch": 2.4374683544303797,
+      "grad_norm": 0.3593789741656094,
+      "learning_rate": 2.4252566272244415e-05,
+      "loss": 0.1977,
+      "step": 601
+    },
+    {
+      "epoch": 2.441518987341772,
+      "grad_norm": 0.34007089412717423,
+      "learning_rate": 2.4197088438925063e-05,
+      "loss": 0.1932,
+      "step": 602
+    },
+    {
+      "epoch": 2.4455696202531647,
+      "grad_norm": 0.31735817399915317,
+      "learning_rate": 2.4141576802832417e-05,
+      "loss": 0.1912,
+      "step": 603
+    },
+    {
+      "epoch": 2.449620253164557,
+      "grad_norm": 0.35343849069393735,
+      "learning_rate": 2.408603181104957e-05,
+      "loss": 0.1961,
+      "step": 604
+    },
+    {
+      "epoch": 2.4536708860759493,
+      "grad_norm": 0.3454539954601134,
+      "learning_rate": 2.4030453910928245e-05,
+      "loss": 0.1965,
+      "step": 605
+    },
+    {
+      "epoch": 2.457721518987342,
+      "grad_norm": 0.3707344263029008,
+      "learning_rate": 2.397484355008521e-05,
+      "loss": 0.1965,
+      "step": 606
+    },
+    {
+      "epoch": 2.4617721518987343,
+      "grad_norm": 0.35166573233504816,
+      "learning_rate": 2.3919201176398662e-05,
+      "loss": 0.1968,
+      "step": 607
+    },
+    {
+      "epoch": 2.4658227848101264,
+      "grad_norm": 0.33818133751821666,
+      "learning_rate": 2.3863527238004633e-05,
+      "loss": 0.1761,
+      "step": 608
+    },
+    {
+      "epoch": 2.469873417721519,
+      "grad_norm": 0.36776885790844943,
+      "learning_rate": 2.380782218329337e-05,
+      "loss": 0.2061,
+      "step": 609
+    },
+    {
+      "epoch": 2.4739240506329114,
+      "grad_norm": 0.35500649912396737,
+      "learning_rate": 2.3752086460905725e-05,
+      "loss": 0.1882,
+      "step": 610
+    },
+    {
+      "epoch": 2.477974683544304,
+      "grad_norm": 0.3285743051684861,
+      "learning_rate": 2.3696320519729544e-05,
+      "loss": 0.209,
+      "step": 611
+    },
+    {
+      "epoch": 2.482025316455696,
+      "grad_norm": 0.34943681722945247,
+      "learning_rate": 2.3640524808896045e-05,
+      "loss": 0.1819,
+      "step": 612
+    },
+    {
+      "epoch": 2.4860759493670885,
+      "grad_norm": 0.3382353165116399,
+      "learning_rate": 2.3584699777776222e-05,
+      "loss": 0.1884,
+      "step": 613
+    },
+    {
+      "epoch": 2.490126582278481,
+      "grad_norm": 0.3660648383103512,
+      "learning_rate": 2.3528845875977195e-05,
+      "loss": 0.2064,
+      "step": 614
+    },
+    {
+      "epoch": 2.4941772151898736,
+      "grad_norm": 0.33526007029874966,
+      "learning_rate": 2.3472963553338614e-05,
+      "loss": 0.2189,
+      "step": 615
+    },
+    {
+      "epoch": 2.4982278481012656,
+      "grad_norm": 0.3462936235411421,
+      "learning_rate": 2.341705325992901e-05,
+      "loss": 0.1974,
+      "step": 616
+    },
+    {
+      "epoch": 2.502278481012658,
+      "grad_norm": 0.3466482659146112,
+      "learning_rate": 2.336111544604222e-05,
+      "loss": 0.1809,
+      "step": 617
+    },
+    {
+      "epoch": 2.5063291139240507,
+      "grad_norm": 0.32583669956109584,
+      "learning_rate": 2.33051505621937e-05,
+      "loss": 0.1847,
+      "step": 618
+    },
+    {
+      "epoch": 2.510379746835443,
+      "grad_norm": 0.35255484856559327,
+      "learning_rate": 2.324915905911693e-05,
+      "loss": 0.1804,
+      "step": 619
+    },
+    {
+      "epoch": 2.5144303797468357,
+      "grad_norm": 0.3677045376298333,
+      "learning_rate": 2.319314138775977e-05,
+      "loss": 0.1927,
+      "step": 620
+    },
+    {
+      "epoch": 2.5184810126582278,
+      "grad_norm": 0.32912620470088944,
+      "learning_rate": 2.3137097999280856e-05,
+      "loss": 0.1963,
+      "step": 621
+    },
+    {
+      "epoch": 2.5225316455696203,
+      "grad_norm": 0.3470753841256499,
+      "learning_rate": 2.308102934504593e-05,
+      "loss": 0.1952,
+      "step": 622
+    },
+    {
+      "epoch": 2.526582278481013,
+      "grad_norm": 0.36772449545826114,
+      "learning_rate": 2.3024935876624222e-05,
+      "loss": 0.1946,
+      "step": 623
+    },
+    {
+      "epoch": 2.530632911392405,
+      "grad_norm": 0.31622112515991974,
+      "learning_rate": 2.2968818045784813e-05,
+      "loss": 0.1899,
+      "step": 624
+    },
+    {
+      "epoch": 2.5346835443037974,
+      "grad_norm": 0.3677244055624019,
+      "learning_rate": 2.2912676304493006e-05,
+      "loss": 0.1867,
+      "step": 625
+    },
+    {
+      "epoch": 2.53873417721519,
+      "grad_norm": 0.36369536249441164,
+      "learning_rate": 2.2856511104906668e-05,
+      "loss": 0.1967,
+      "step": 626
+    },
+    {
+      "epoch": 2.5427848101265824,
+      "grad_norm": 0.3238004152086756,
+      "learning_rate": 2.2800322899372586e-05,
+      "loss": 0.1952,
+      "step": 627
+    },
+    {
+      "epoch": 2.546835443037975,
+      "grad_norm": 0.35053159156613733,
+      "learning_rate": 2.2744112140422844e-05,
+      "loss": 0.1855,
+      "step": 628
+    },
+    {
+      "epoch": 2.550886075949367,
+      "grad_norm": 0.3449560189229714,
+      "learning_rate": 2.2687879280771177e-05,
+      "loss": 0.1819,
+      "step": 629
+    },
+    {
+      "epoch": 2.5549367088607595,
+      "grad_norm": 0.3759909706220713,
+      "learning_rate": 2.26316247733093e-05,
+      "loss": 0.2074,
+      "step": 630
+    },
+    {
+      "epoch": 2.558987341772152,
+      "grad_norm": 0.3398132430184106,
+      "learning_rate": 2.257534907110328e-05,
+      "loss": 0.1987,
+      "step": 631
+    },
+    {
+      "epoch": 2.563037974683544,
+      "grad_norm": 0.35627546817339256,
+      "learning_rate": 2.2519052627389882e-05,
+      "loss": 0.1934,
+      "step": 632
+    },
+    {
+      "epoch": 2.5670886075949366,
+      "grad_norm": 0.3580755379385013,
+      "learning_rate": 2.246273589557294e-05,
+      "loss": 0.1946,
+      "step": 633
+    },
+    {
+      "epoch": 2.571139240506329,
+      "grad_norm": 0.31782377262468975,
+      "learning_rate": 2.240639932921966e-05,
+      "loss": 0.195,
+      "step": 634
+    },
+    {
+      "epoch": 2.5751898734177217,
+      "grad_norm": 0.3728357078746841,
+      "learning_rate": 2.2350043382056995e-05,
+      "loss": 0.1945,
+      "step": 635
+    },
+    {
+      "epoch": 2.579240506329114,
+      "grad_norm": 0.3540370649341306,
+      "learning_rate": 2.2293668507968015e-05,
+      "loss": 0.2074,
+      "step": 636
+    },
+    {
+      "epoch": 2.5832911392405062,
+      "grad_norm": 0.3528159019759957,
+      "learning_rate": 2.2237275160988186e-05,
+      "loss": 0.1867,
+      "step": 637
+    },
+    {
+      "epoch": 2.5873417721518988,
+      "grad_norm": 0.3592187091150007,
+      "learning_rate": 2.2180863795301787e-05,
+      "loss": 0.1937,
+      "step": 638
+    },
+    {
+      "epoch": 2.5913924050632913,
+      "grad_norm": 0.34847533545603837,
+      "learning_rate": 2.212443486523819e-05,
+      "loss": 0.1866,
+      "step": 639
+    },
+    {
+      "epoch": 2.5954430379746833,
+      "grad_norm": 0.3326134845067451,
+      "learning_rate": 2.2067988825268243e-05,
+      "loss": 0.198,
+      "step": 640
+    },
+    {
+      "epoch": 2.599493670886076,
+      "grad_norm": 0.3232208564283459,
+      "learning_rate": 2.2011526130000596e-05,
+      "loss": 0.1908,
+      "step": 641
+    },
+    {
+      "epoch": 2.6035443037974684,
+      "grad_norm": 0.3400559491508242,
+      "learning_rate": 2.1955047234178038e-05,
+      "loss": 0.1992,
+      "step": 642
+    },
+    {
+      "epoch": 2.607594936708861,
+      "grad_norm": 0.33038972374833203,
+      "learning_rate": 2.1898552592673825e-05,
+      "loss": 0.1994,
+      "step": 643
+    },
+    {
+      "epoch": 2.6116455696202534,
+      "grad_norm": 0.3297903532623937,
+      "learning_rate": 2.184204266048803e-05,
+      "loss": 0.2038,
+      "step": 644
+    },
+    {
+      "epoch": 2.6156962025316455,
+      "grad_norm": 0.33297667489746025,
+      "learning_rate": 2.1785517892743887e-05,
+      "loss": 0.2087,
+      "step": 645
+    },
+    {
+      "epoch": 2.619746835443038,
+      "grad_norm": 0.313577215146083,
+      "learning_rate": 2.17289787446841e-05,
+      "loss": 0.2057,
+      "step": 646
+    },
+    {
+      "epoch": 2.6237974683544305,
+      "grad_norm": 0.3475796715510358,
+      "learning_rate": 2.1672425671667198e-05,
+      "loss": 0.1887,
+      "step": 647
+    },
+    {
+      "epoch": 2.6278481012658226,
+      "grad_norm": 0.34386896056831545,
+      "learning_rate": 2.161585912916385e-05,
+      "loss": 0.2046,
+      "step": 648
+    },
+    {
+      "epoch": 2.631898734177215,
+      "grad_norm": 0.3515232126915773,
+      "learning_rate": 2.1559279572753214e-05,
+      "loss": 0.1944,
+      "step": 649
+    },
+    {
+      "epoch": 2.6359493670886076,
+      "grad_norm": 0.30073395886971477,
+      "learning_rate": 2.1502687458119268e-05,
+      "loss": 0.1995,
+      "step": 650
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 0.33916506446091366,
+      "learning_rate": 2.1446083241047116e-05,
+      "loss": 0.2001,
+      "step": 651
+    },
+    {
+      "epoch": 2.6440506329113926,
+      "grad_norm": 0.33219094096610313,
+      "learning_rate": 2.1389467377419333e-05,
+      "loss": 0.2012,
+      "step": 652
+    },
+    {
+      "epoch": 2.6481012658227847,
+      "grad_norm": 0.3138368182627072,
+      "learning_rate": 2.133284032321232e-05,
+      "loss": 0.2081,
+      "step": 653
+    },
+    {
+      "epoch": 2.6521518987341772,
+      "grad_norm": 0.32465279261602803,
+      "learning_rate": 2.1276202534492566e-05,
+      "loss": 0.2014,
+      "step": 654
+    },
+    {
+      "epoch": 2.6562025316455697,
+      "grad_norm": 0.3348108347581183,
+      "learning_rate": 2.121955446741306e-05,
+      "loss": 0.2021,
+      "step": 655
+    },
+    {
+      "epoch": 2.660253164556962,
+      "grad_norm": 0.3141483980472192,
+      "learning_rate": 2.1162896578209517e-05,
+      "loss": 0.1956,
+      "step": 656
+    },
+    {
+      "epoch": 2.6643037974683543,
+      "grad_norm": 0.3263049319800991,
+      "learning_rate": 2.1106229323196813e-05,
+      "loss": 0.1932,
+      "step": 657
+    },
+    {
+      "epoch": 2.668354430379747,
+      "grad_norm": 0.3207611663015678,
+      "learning_rate": 2.1049553158765214e-05,
+      "loss": 0.1981,
+      "step": 658
+    },
+    {
+      "epoch": 2.6724050632911394,
+      "grad_norm": 0.33007089647826554,
+      "learning_rate": 2.0992868541376764e-05,
+      "loss": 0.1975,
+      "step": 659
+    },
+    {
+      "epoch": 2.676455696202532,
+      "grad_norm": 0.30306424420488653,
+      "learning_rate": 2.093617592756158e-05,
+      "loss": 0.1969,
+      "step": 660
+    },
+    {
+      "epoch": 2.680506329113924,
+      "grad_norm": 0.32011756788711165,
+      "learning_rate": 2.0879475773914167e-05,
+      "loss": 0.1922,
+      "step": 661
+    },
+    {
+      "epoch": 2.6845569620253165,
+      "grad_norm": 0.33406356523175645,
+      "learning_rate": 2.082276853708978e-05,
+      "loss": 0.2017,
+      "step": 662
+    },
+    {
+      "epoch": 2.688607594936709,
+      "grad_norm": 0.3283333583674527,
+      "learning_rate": 2.076605467380071e-05,
+      "loss": 0.19,
+      "step": 663
+    },
+    {
+      "epoch": 2.692658227848101,
+      "grad_norm": 0.3083235645459152,
+      "learning_rate": 2.0709334640812613e-05,
+      "loss": 0.2051,
+      "step": 664
+    },
+    {
+      "epoch": 2.6967088607594936,
+      "grad_norm": 0.3322441958455594,
+      "learning_rate": 2.0652608894940824e-05,
+      "loss": 0.1884,
+      "step": 665
+    },
+    {
+      "epoch": 2.700759493670886,
+      "grad_norm": 0.3564324018418401,
+      "learning_rate": 2.0595877893046722e-05,
+      "loss": 0.1942,
+      "step": 666
+    },
+    {
+      "epoch": 2.7048101265822786,
+      "grad_norm": 0.322198933160248,
+      "learning_rate": 2.0539142092033985e-05,
+      "loss": 0.1859,
+      "step": 667
+    },
+    {
+      "epoch": 2.708860759493671,
+      "grad_norm": 0.31524683626983574,
+      "learning_rate": 2.048240194884496e-05,
+      "loss": 0.2097,
+      "step": 668
+    },
+    {
+      "epoch": 2.712911392405063,
+      "grad_norm": 0.3715156374496271,
+      "learning_rate": 2.042565792045695e-05,
+      "loss": 0.2046,
+      "step": 669
+    },
+    {
+      "epoch": 2.7169620253164557,
+      "grad_norm": 0.32186977211364903,
+      "learning_rate": 2.036891046387857e-05,
+      "loss": 0.2029,
+      "step": 670
+    },
+    {
+      "epoch": 2.721012658227848,
+      "grad_norm": 0.3227788583923027,
+      "learning_rate": 2.0312160036146036e-05,
+      "loss": 0.1856,
+      "step": 671
+    },
+    {
+      "epoch": 2.7250632911392403,
+      "grad_norm": 0.338291266878291,
+      "learning_rate": 2.025540709431948e-05,
+      "loss": 0.1917,
+      "step": 672
+    },
+    {
+      "epoch": 2.729113924050633,
+      "grad_norm": 0.3288841552657687,
+      "learning_rate": 2.0198652095479298e-05,
+      "loss": 0.1958,
+      "step": 673
+    },
+    {
+      "epoch": 2.7331645569620253,
+      "grad_norm": 0.3089444796818637,
+      "learning_rate": 2.014189549672245e-05,
+      "loss": 0.2,
+      "step": 674
+    },
+    {
+      "epoch": 2.737215189873418,
+      "grad_norm": 0.3428144224149456,
+      "learning_rate": 2.0085137755158776e-05,
+      "loss": 0.2029,
+      "step": 675
+    },
+    {
+      "epoch": 2.7412658227848103,
+      "grad_norm": 0.3372740008298548,
+      "learning_rate": 2.0028379327907327e-05,
+      "loss": 0.1821,
+      "step": 676
+    },
+    {
+      "epoch": 2.7453164556962024,
+      "grad_norm": 0.3528274796721654,
+      "learning_rate": 1.9971620672092676e-05,
+      "loss": 0.1972,
+      "step": 677
+    },
+    {
+      "epoch": 2.749367088607595,
+      "grad_norm": 0.33507757759985896,
+      "learning_rate": 1.991486224484123e-05,
+      "loss": 0.1925,
+      "step": 678
+    },
+    {
+      "epoch": 2.7534177215189874,
+      "grad_norm": 0.33865863416443204,
+      "learning_rate": 1.985810450327756e-05,
+      "loss": 0.2017,
+      "step": 679
+    },
+    {
+      "epoch": 2.7574683544303795,
+      "grad_norm": 0.32717966081012284,
+      "learning_rate": 1.9801347904520706e-05,
+      "loss": 0.1926,
+      "step": 680
+    },
+    {
+      "epoch": 2.761518987341772,
+      "grad_norm": 0.34516218528817266,
+      "learning_rate": 1.974459290568053e-05,
+      "loss": 0.1908,
+      "step": 681
+    },
+    {
+      "epoch": 2.7655696202531646,
+      "grad_norm": 0.31559010110283975,
+      "learning_rate": 1.968783996385397e-05,
+      "loss": 0.1927,
+      "step": 682
+    },
+    {
+      "epoch": 2.769620253164557,
+      "grad_norm": 0.3263804318515319,
+      "learning_rate": 1.963108953612143e-05,
+      "loss": 0.1917,
+      "step": 683
+    },
+    {
+      "epoch": 2.7736708860759496,
+      "grad_norm": 0.34375006453057244,
+      "learning_rate": 1.9574342079543056e-05,
+      "loss": 0.1923,
+      "step": 684
+    },
+    {
+      "epoch": 2.7777215189873417,
+      "grad_norm": 0.31653624279105247,
+      "learning_rate": 1.9517598051155046e-05,
+      "loss": 0.2009,
+      "step": 685
+    },
+    {
+      "epoch": 2.781772151898734,
+      "grad_norm": 0.30736221050586876,
+      "learning_rate": 1.9460857907966025e-05,
+      "loss": 0.1984,
+      "step": 686
+    },
+    {
+      "epoch": 2.7858227848101267,
+      "grad_norm": 0.31944244697342583,
+      "learning_rate": 1.9404122106953285e-05,
+      "loss": 0.1882,
+      "step": 687
+    },
+    {
+      "epoch": 2.7898734177215188,
+      "grad_norm": 0.3307886541457952,
+      "learning_rate": 1.9347391105059176e-05,
+      "loss": 0.1888,
+      "step": 688
+    },
+    {
+      "epoch": 2.7939240506329113,
+      "grad_norm": 0.3323117369179801,
+      "learning_rate": 1.92906653591874e-05,
+      "loss": 0.1897,
+      "step": 689
+    },
+    {
+      "epoch": 2.797974683544304,
+      "grad_norm": 0.33699865061122797,
+      "learning_rate": 1.9233945326199295e-05,
+      "loss": 0.1926,
+      "step": 690
+    },
+    {
+      "epoch": 2.8020253164556963,
+      "grad_norm": 0.31329785665894916,
+      "learning_rate": 1.917723146291022e-05,
+      "loss": 0.1888,
+      "step": 691
+    },
+    {
+      "epoch": 2.806075949367089,
+      "grad_norm": 0.31150446114555064,
+      "learning_rate": 1.912052422608584e-05,
+      "loss": 0.1951,
+      "step": 692
+    },
+    {
+      "epoch": 2.810126582278481,
+      "grad_norm": 0.32741553680633667,
+      "learning_rate": 1.9063824072438428e-05,
+      "loss": 0.192,
+      "step": 693
+    },
+    {
+      "epoch": 2.8141772151898734,
+      "grad_norm": 0.31332912861586,
+      "learning_rate": 1.9007131458623246e-05,
+      "loss": 0.205,
+      "step": 694
+    },
+    {
+      "epoch": 2.818227848101266,
+      "grad_norm": 0.3087004413871022,
+      "learning_rate": 1.895044684123479e-05,
+      "loss": 0.2007,
+      "step": 695
+    },
+    {
+      "epoch": 2.822278481012658,
+      "grad_norm": 0.33323109793953826,
+      "learning_rate": 1.8893770676803194e-05,
+      "loss": 0.1899,
+      "step": 696
+    },
+    {
+      "epoch": 2.8263291139240505,
+      "grad_norm": 0.3287514983236516,
+      "learning_rate": 1.8837103421790486e-05,
+      "loss": 0.2004,
+      "step": 697
+    },
+    {
+      "epoch": 2.830379746835443,
+      "grad_norm": 0.3449421555323112,
+      "learning_rate": 1.8780445532586952e-05,
+      "loss": 0.1943,
+      "step": 698
+    },
+    {
+      "epoch": 2.8344303797468355,
+      "grad_norm": 0.324820363778185,
+      "learning_rate": 1.872379746550743e-05,
+      "loss": 0.2037,
+      "step": 699
+    },
+    {
+      "epoch": 2.838481012658228,
+      "grad_norm": 0.31674402485110065,
+      "learning_rate": 1.866715967678769e-05,
+      "loss": 0.1975,
+      "step": 700
+    },
+    {
+      "epoch": 2.84253164556962,
+      "grad_norm": 0.3400692083028216,
+      "learning_rate": 1.861053262258067e-05,
+      "loss": 0.1893,
+      "step": 701
+    },
+    {
+      "epoch": 2.8465822784810126,
+      "grad_norm": 0.33850437309289844,
+      "learning_rate": 1.8553916758952897e-05,
+      "loss": 0.1841,
+      "step": 702
+    },
+    {
+      "epoch": 2.850632911392405,
+      "grad_norm": 0.3120135638085117,
+      "learning_rate": 1.8497312541880735e-05,
+      "loss": 0.1946,
+      "step": 703
+    },
+    {
+      "epoch": 2.8546835443037972,
+      "grad_norm": 0.3305581159799596,
+      "learning_rate": 1.8440720427246786e-05,
+      "loss": 0.1919,
+      "step": 704
+    },
+    {
+      "epoch": 2.8587341772151897,
+      "grad_norm": 0.33262119407323026,
+      "learning_rate": 1.8384140870836157e-05,
+      "loss": 0.2147,
+      "step": 705
+    },
+    {
+      "epoch": 2.8627848101265823,
+      "grad_norm": 0.3276530505805219,
+      "learning_rate": 1.8327574328332806e-05,
+      "loss": 0.1883,
+      "step": 706
+    },
+    {
+      "epoch": 2.8668354430379748,
+      "grad_norm": 0.31474464165712357,
+      "learning_rate": 1.8271021255315906e-05,
+      "loss": 0.2183,
+      "step": 707
+    },
+    {
+      "epoch": 2.8708860759493673,
+      "grad_norm": 0.32448882017365205,
+      "learning_rate": 1.8214482107256117e-05,
+      "loss": 0.1894,
+      "step": 708
+    },
+    {
+      "epoch": 2.8749367088607594,
+      "grad_norm": 0.32754685748572276,
+      "learning_rate": 1.8157957339511968e-05,
+      "loss": 0.1984,
+      "step": 709
+    },
+    {
+      "epoch": 2.878987341772152,
+      "grad_norm": 0.3151622532010612,
+      "learning_rate": 1.8101447407326182e-05,
+      "loss": 0.1937,
+      "step": 710
+    },
+    {
+      "epoch": 2.8830379746835444,
+      "grad_norm": 0.31625803880818654,
+      "learning_rate": 1.8044952765821966e-05,
+      "loss": 0.1889,
+      "step": 711
+    },
+    {
+      "epoch": 2.8870886075949365,
+      "grad_norm": 0.3215236632221963,
+      "learning_rate": 1.7988473869999407e-05,
+      "loss": 0.1942,
+      "step": 712
+    },
+    {
+      "epoch": 2.891139240506329,
+      "grad_norm": 0.33894772305673126,
+      "learning_rate": 1.7932011174731764e-05,
+      "loss": 0.2071,
+      "step": 713
+    },
+    {
+      "epoch": 2.8951898734177215,
+      "grad_norm": 0.3166331808241484,
+      "learning_rate": 1.7875565134761817e-05,
+      "loss": 0.1944,
+      "step": 714
+    },
+    {
+      "epoch": 2.899240506329114,
+      "grad_norm": 0.3411504494242672,
+      "learning_rate": 1.7819136204698226e-05,
+      "loss": 0.2054,
+      "step": 715
+    },
+    {
+      "epoch": 2.9032911392405065,
+      "grad_norm": 0.3514324232464818,
+      "learning_rate": 1.776272483901182e-05,
+      "loss": 0.2076,
+      "step": 716
+    },
+    {
+      "epoch": 2.9073417721518986,
+      "grad_norm": 0.33050860219196626,
+      "learning_rate": 1.7706331492031995e-05,
+      "loss": 0.1927,
+      "step": 717
+    },
+    {
+      "epoch": 2.911392405063291,
+      "grad_norm": 0.3312196797209112,
+      "learning_rate": 1.764995661794301e-05,
+      "loss": 0.1924,
+      "step": 718
+    },
+    {
+      "epoch": 2.9154430379746836,
+      "grad_norm": 0.3224176784769984,
+      "learning_rate": 1.759360067078035e-05,
+      "loss": 0.1876,
+      "step": 719
+    },
+    {
+      "epoch": 2.9194936708860757,
+      "grad_norm": 0.3228823581736586,
+      "learning_rate": 1.7537264104427064e-05,
+      "loss": 0.182,
+      "step": 720
+    },
+    {
+      "epoch": 2.923544303797468,
+      "grad_norm": 0.32436509637995165,
+      "learning_rate": 1.748094737261012e-05,
+      "loss": 0.1903,
+      "step": 721
+    },
+    {
+      "epoch": 2.9275949367088607,
+      "grad_norm": 0.336695420988062,
+      "learning_rate": 1.7424650928896726e-05,
+      "loss": 0.2083,
+      "step": 722
+    },
+    {
+      "epoch": 2.9316455696202532,
+      "grad_norm": 0.3164816723093164,
+      "learning_rate": 1.7368375226690712e-05,
+      "loss": 0.1975,
+      "step": 723
+    },
+    {
+      "epoch": 2.9356962025316458,
+      "grad_norm": 0.3342117421770553,
+      "learning_rate": 1.731212071922883e-05,
+      "loss": 0.1892,
+      "step": 724
+    },
+    {
+      "epoch": 2.939746835443038,
+      "grad_norm": 0.3347533416668081,
+      "learning_rate": 1.7255887859577156e-05,
+      "loss": 0.2032,
+      "step": 725
+    },
+    {
+      "epoch": 2.9437974683544303,
+      "grad_norm": 0.3016300011033397,
+      "learning_rate": 1.7199677100627427e-05,
+      "loss": 0.1901,
+      "step": 726
+    },
+    {
+      "epoch": 2.947848101265823,
+      "grad_norm": 0.3451808050412577,
+      "learning_rate": 1.7143488895093343e-05,
+      "loss": 0.1866,
+      "step": 727
+    },
+    {
+      "epoch": 2.951898734177215,
+      "grad_norm": 0.3333258979226123,
+      "learning_rate": 1.7087323695506994e-05,
+      "loss": 0.1888,
+      "step": 728
+    },
+    {
+      "epoch": 2.9559493670886074,
+      "grad_norm": 0.29743842182390684,
+      "learning_rate": 1.7031181954215194e-05,
+      "loss": 0.1881,
+      "step": 729
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 0.3312662856158668,
+      "learning_rate": 1.6975064123375788e-05,
+      "loss": 0.1978,
+      "step": 730
+    },
+    {
+      "epoch": 2.9640506329113925,
+      "grad_norm": 0.3271128430009879,
+      "learning_rate": 1.6918970654954084e-05,
+      "loss": 0.1824,
+      "step": 731
+    },
+    {
+      "epoch": 2.968101265822785,
+      "grad_norm": 0.33227824863209715,
+      "learning_rate": 1.686290200071915e-05,
+      "loss": 0.1874,
+      "step": 732
+    },
+    {
+      "epoch": 2.972151898734177,
+      "grad_norm": 0.3228225612824928,
+      "learning_rate": 1.6806858612240234e-05,
+      "loss": 0.1882,
+      "step": 733
+    },
+    {
+      "epoch": 2.9762025316455696,
+      "grad_norm": 0.3264448991463743,
+      "learning_rate": 1.6750840940883078e-05,
+      "loss": 0.2001,
+      "step": 734
+    },
+    {
+      "epoch": 2.980253164556962,
+      "grad_norm": 0.3144916881941647,
+      "learning_rate": 1.6694849437806305e-05,
+      "loss": 0.1989,
+      "step": 735
+    },
+    {
+      "epoch": 2.984303797468354,
+      "grad_norm": 0.3133873660631297,
+      "learning_rate": 1.663888455395778e-05,
+      "loss": 0.1903,
+      "step": 736
+    },
+    {
+      "epoch": 2.9883544303797467,
+      "grad_norm": 0.31058862637812307,
+      "learning_rate": 1.6582946740070995e-05,
+      "loss": 0.1924,
+      "step": 737
+    },
+    {
+      "epoch": 2.992405063291139,
+      "grad_norm": 0.32868526614300503,
+      "learning_rate": 1.6527036446661396e-05,
+      "loss": 0.1992,
+      "step": 738
+    },
+    {
+      "epoch": 2.9964556962025317,
+      "grad_norm": 0.3128912690073749,
+      "learning_rate": 1.6471154124022818e-05,
+      "loss": 0.1928,
+      "step": 739
+    },
+    {
+      "epoch": 3.002025316455696,
+      "grad_norm": 0.3407581977738386,
+      "learning_rate": 1.6415300222223788e-05,
+      "loss": 0.1592,
+      "step": 740
+    },
+    {
+      "epoch": 3.0060759493670886,
+      "grad_norm": 0.36325657851538645,
+      "learning_rate": 1.6359475191103958e-05,
+      "loss": 0.1267,
+      "step": 741
+    },
+    {
+      "epoch": 3.010126582278481,
+      "grad_norm": 0.3121664852835936,
+      "learning_rate": 1.6303679480270466e-05,
+      "loss": 0.1184,
+      "step": 742
+    },
+    {
+      "epoch": 3.0141772151898736,
+      "grad_norm": 0.40782535388869484,
+      "learning_rate": 1.624791353909428e-05,
+      "loss": 0.1352,
+      "step": 743
+    },
+    {
+      "epoch": 3.0182278481012657,
+      "grad_norm": 0.48510251424497025,
+      "learning_rate": 1.619217781670663e-05,
+      "loss": 0.1276,
+      "step": 744
+    },
+    {
+      "epoch": 3.022278481012658,
+      "grad_norm": 0.3440403738392954,
+      "learning_rate": 1.6136472761995373e-05,
+      "loss": 0.1148,
+      "step": 745
+    },
+    {
+      "epoch": 3.0263291139240507,
+      "grad_norm": 0.39173249005000454,
+      "learning_rate": 1.608079882360134e-05,
+      "loss": 0.1241,
+      "step": 746
+    },
+    {
+      "epoch": 3.030379746835443,
+      "grad_norm": 0.44324575571421304,
+      "learning_rate": 1.60251564499148e-05,
+      "loss": 0.1286,
+      "step": 747
+    },
+    {
+      "epoch": 3.0344303797468353,
+      "grad_norm": 0.3168312069939277,
+      "learning_rate": 1.596954608907176e-05,
+      "loss": 0.1174,
+      "step": 748
+    },
+    {
+      "epoch": 3.038481012658228,
+      "grad_norm": 0.35589268280124264,
+      "learning_rate": 1.591396818895043e-05,
+      "loss": 0.1293,
+      "step": 749
+    },
+    {
+      "epoch": 3.0425316455696203,
+      "grad_norm": 0.37163292044952617,
+      "learning_rate": 1.585842319716759e-05,
+      "loss": 0.1284,
+      "step": 750
+    },
+    {
+      "epoch": 3.046582278481013,
+      "grad_norm": 0.32516324577570027,
+      "learning_rate": 1.5802911561074944e-05,
+      "loss": 0.1134,
+      "step": 751
+    },
+    {
+      "epoch": 3.050632911392405,
+      "grad_norm": 0.3606787255471758,
+      "learning_rate": 1.5747433727755595e-05,
+      "loss": 0.1173,
+      "step": 752
+    },
+    {
+      "epoch": 3.0546835443037974,
+      "grad_norm": 0.3835141381833492,
+      "learning_rate": 1.5691990144020376e-05,
+      "loss": 0.1182,
+      "step": 753
+    },
+    {
+      "epoch": 3.05873417721519,
+      "grad_norm": 0.3279903712229184,
+      "learning_rate": 1.5636581256404297e-05,
+      "loss": 0.1249,
+      "step": 754
+    },
+    {
+      "epoch": 3.0627848101265824,
+      "grad_norm": 0.3611839933423646,
+      "learning_rate": 1.558120751116291e-05,
+      "loss": 0.1194,
+      "step": 755
+    },
+    {
+      "epoch": 3.0668354430379745,
+      "grad_norm": 0.35261659970593406,
+      "learning_rate": 1.552586935426876e-05,
+      "loss": 0.1181,
+      "step": 756
+    },
+    {
+      "epoch": 3.070886075949367,
+      "grad_norm": 0.34046559536182935,
+      "learning_rate": 1.547056723140774e-05,
+      "loss": 0.1317,
+      "step": 757
+    },
+    {
+      "epoch": 3.0749367088607595,
+      "grad_norm": 0.3295656904690491,
+      "learning_rate": 1.5415301587975565e-05,
+      "loss": 0.1189,
+      "step": 758
+    },
+    {
+      "epoch": 3.078987341772152,
+      "grad_norm": 0.3543591005482601,
+      "learning_rate": 1.536007286907411e-05,
+      "loss": 0.1203,
+      "step": 759
+    },
+    {
+      "epoch": 3.083037974683544,
+      "grad_norm": 0.34039178748965687,
+      "learning_rate": 1.5304881519507896e-05,
+      "loss": 0.1079,
+      "step": 760
+    },
+    {
+      "epoch": 3.0870886075949366,
+      "grad_norm": 0.3417803044358889,
+      "learning_rate": 1.5249727983780453e-05,
+      "loss": 0.1232,
+      "step": 761
+    },
+    {
+      "epoch": 3.091139240506329,
+      "grad_norm": 0.3307231244135652,
+      "learning_rate": 1.5194612706090786e-05,
+      "loss": 0.1201,
+      "step": 762
+    },
+    {
+      "epoch": 3.0951898734177217,
+      "grad_norm": 0.3051624022172646,
+      "learning_rate": 1.5139536130329771e-05,
+      "loss": 0.1242,
+      "step": 763
+    },
+    {
+      "epoch": 3.0992405063291137,
+      "grad_norm": 0.3177831596381515,
+      "learning_rate": 1.508449870007656e-05,
+      "loss": 0.118,
+      "step": 764
+    },
+    {
+      "epoch": 3.1032911392405063,
+      "grad_norm": 0.32018151131618583,
+      "learning_rate": 1.5029500858595056e-05,
+      "loss": 0.1211,
+      "step": 765
+    },
+    {
+      "epoch": 3.1073417721518988,
+      "grad_norm": 0.3264575361901864,
+      "learning_rate": 1.4974543048830328e-05,
+      "loss": 0.1221,
+      "step": 766
+    },
+    {
+      "epoch": 3.1113924050632913,
+      "grad_norm": 0.33486502733483503,
+      "learning_rate": 1.4919625713405e-05,
+      "loss": 0.122,
+      "step": 767
+    },
+    {
+      "epoch": 3.1154430379746834,
+      "grad_norm": 0.3460940077337307,
+      "learning_rate": 1.4864749294615756e-05,
+      "loss": 0.1209,
+      "step": 768
+    },
+    {
+      "epoch": 3.119493670886076,
+      "grad_norm": 0.2995856066054969,
+      "learning_rate": 1.4809914234429716e-05,
+      "loss": 0.1248,
+      "step": 769
+    },
+    {
+      "epoch": 3.1235443037974684,
+      "grad_norm": 0.3321297272125938,
+      "learning_rate": 1.4755120974480923e-05,
+      "loss": 0.1263,
+      "step": 770
+    },
+    {
+      "epoch": 3.127594936708861,
+      "grad_norm": 0.3088248947281187,
+      "learning_rate": 1.4700369956066771e-05,
+      "loss": 0.1194,
+      "step": 771
+    },
+    {
+      "epoch": 3.131645569620253,
+      "grad_norm": 0.30295299615124366,
+      "learning_rate": 1.4645661620144413e-05,
+      "loss": 0.1285,
+      "step": 772
+    },
+    {
+      "epoch": 3.1356962025316455,
+      "grad_norm": 0.3069948243123267,
+      "learning_rate": 1.4590996407327284e-05,
+      "loss": 0.1192,
+      "step": 773
+    },
+    {
+      "epoch": 3.139746835443038,
+      "grad_norm": 0.3079211917917274,
+      "learning_rate": 1.4536374757881487e-05,
+      "loss": 0.1206,
+      "step": 774
+    },
+    {
+      "epoch": 3.1437974683544305,
+      "grad_norm": 0.30615315684131006,
+      "learning_rate": 1.4481797111722271e-05,
+      "loss": 0.1168,
+      "step": 775
+    },
+    {
+      "epoch": 3.1478481012658226,
+      "grad_norm": 0.291133826926988,
+      "learning_rate": 1.4427263908410507e-05,
+      "loss": 0.116,
+      "step": 776
+    },
+    {
+      "epoch": 3.151898734177215,
+      "grad_norm": 0.30621731096615923,
+      "learning_rate": 1.4372775587149108e-05,
+      "loss": 0.1192,
+      "step": 777
+    },
+    {
+      "epoch": 3.1559493670886076,
+      "grad_norm": 0.282489372101,
+      "learning_rate": 1.4318332586779522e-05,
+      "loss": 0.1134,
+      "step": 778
+    },
+    {
+      "epoch": 3.16,
+      "grad_norm": 0.2887038433712594,
+      "learning_rate": 1.4263935345778202e-05,
+      "loss": 0.1206,
+      "step": 779
+    },
+    {
+      "epoch": 3.164050632911392,
+      "grad_norm": 0.2892457222783059,
+      "learning_rate": 1.420958430225303e-05,
+      "loss": 0.1106,
+      "step": 780
+    },
+    {
+      "epoch": 3.1681012658227847,
+      "grad_norm": 0.30392927061905406,
+      "learning_rate": 1.415527989393985e-05,
+      "loss": 0.1234,
+      "step": 781
+    },
+    {
+      "epoch": 3.1721518987341772,
+      "grad_norm": 0.29319960859481226,
+      "learning_rate": 1.410102255819891e-05,
+      "loss": 0.1279,
+      "step": 782
+    },
+    {
+      "epoch": 3.1762025316455698,
+      "grad_norm": 0.2968748078563776,
+      "learning_rate": 1.404681273201131e-05,
+      "loss": 0.1264,
+      "step": 783
+    },
+    {
+      "epoch": 3.180253164556962,
+      "grad_norm": 0.2929781183842439,
+      "learning_rate": 1.399265085197556e-05,
+      "loss": 0.121,
+      "step": 784
+    },
+    {
+      "epoch": 3.1843037974683543,
+      "grad_norm": 0.28086656122974146,
+      "learning_rate": 1.393853735430398e-05,
+      "loss": 0.1231,
+      "step": 785
+    },
+    {
+      "epoch": 3.188354430379747,
+      "grad_norm": 0.2921182697250289,
+      "learning_rate": 1.3884472674819246e-05,
+      "loss": 0.1406,
+      "step": 786
+    },
+    {
+      "epoch": 3.1924050632911394,
+      "grad_norm": 0.31598960558217076,
+      "learning_rate": 1.3830457248950864e-05,
+      "loss": 0.1195,
+      "step": 787
+    },
+    {
+      "epoch": 3.1964556962025314,
+      "grad_norm": 0.2884265839200278,
+      "learning_rate": 1.377649151173163e-05,
+      "loss": 0.1138,
+      "step": 788
+    },
+    {
+      "epoch": 3.200506329113924,
+      "grad_norm": 0.28041047728139223,
+      "learning_rate": 1.3722575897794181e-05,
+      "loss": 0.1135,
+      "step": 789
+    },
+    {
+      "epoch": 3.2045569620253165,
+      "grad_norm": 0.30684393389884085,
+      "learning_rate": 1.3668710841367472e-05,
+      "loss": 0.1201,
+      "step": 790
+    },
+    {
+      "epoch": 3.208607594936709,
+      "grad_norm": 0.30342694463869313,
+      "learning_rate": 1.361489677627324e-05,
+      "loss": 0.124,
+      "step": 791
+    },
+    {
+      "epoch": 3.212658227848101,
+      "grad_norm": 0.3017226234856027,
+      "learning_rate": 1.3561134135922585e-05,
+      "loss": 0.136,
+      "step": 792
+    },
+    {
+      "epoch": 3.2167088607594936,
+      "grad_norm": 0.3078700848386741,
+      "learning_rate": 1.350742335331241e-05,
+      "loss": 0.1317,
+      "step": 793
+    },
+    {
+      "epoch": 3.220759493670886,
+      "grad_norm": 0.3090057804203604,
+      "learning_rate": 1.345376486102198e-05,
+      "loss": 0.1317,
+      "step": 794
+    },
+    {
+      "epoch": 3.2248101265822786,
+      "grad_norm": 0.295315617790507,
+      "learning_rate": 1.3400159091209414e-05,
+      "loss": 0.1228,
+      "step": 795
+    },
+    {
+      "epoch": 3.2288607594936707,
+      "grad_norm": 0.29530837960969497,
+      "learning_rate": 1.3346606475608216e-05,
+      "loss": 0.1205,
+      "step": 796
+    },
+    {
+      "epoch": 3.232911392405063,
+      "grad_norm": 0.3010338941974923,
+      "learning_rate": 1.3293107445523781e-05,
+      "loss": 0.1222,
+      "step": 797
+    },
+    {
+      "epoch": 3.2369620253164557,
+      "grad_norm": 0.30307166292834936,
+      "learning_rate": 1.3239662431829949e-05,
+      "loss": 0.1087,
+      "step": 798
+    },
+    {
+      "epoch": 3.2410126582278482,
+      "grad_norm": 0.30323429541216407,
+      "learning_rate": 1.3186271864965509e-05,
+      "loss": 0.1188,
+      "step": 799
+    },
+    {
+      "epoch": 3.2450632911392403,
+      "grad_norm": 0.29690072970758036,
+      "learning_rate": 1.3132936174930756e-05,
+      "loss": 0.1228,
+      "step": 800
+    },
+    {
+      "epoch": 3.249113924050633,
+      "grad_norm": 0.31140285399982626,
+      "learning_rate": 1.3079655791283995e-05,
+      "loss": 0.1213,
+      "step": 801
+    },
+    {
+      "epoch": 3.2531645569620253,
+      "grad_norm": 0.3142774327832572,
+      "learning_rate": 1.3026431143138108e-05,
+      "loss": 0.1101,
+      "step": 802
+    },
+    {
+      "epoch": 3.257215189873418,
+      "grad_norm": 0.29248319222180497,
+      "learning_rate": 1.2973262659157114e-05,
+      "loss": 0.1276,
+      "step": 803
+    },
+    {
+      "epoch": 3.26126582278481,
+      "grad_norm": 0.27077414664023225,
+      "learning_rate": 1.2920150767552651e-05,
+      "loss": 0.1178,
+      "step": 804
+    },
+    {
+      "epoch": 3.2653164556962024,
+      "grad_norm": 0.3167628306152777,
+      "learning_rate": 1.2867095896080607e-05,
+      "loss": 0.1218,
+      "step": 805
+    },
+    {
+      "epoch": 3.269367088607595,
+      "grad_norm": 0.3173758515382908,
+      "learning_rate": 1.2814098472037612e-05,
+      "loss": 0.1199,
+      "step": 806
+    },
+    {
+      "epoch": 3.2734177215189875,
+      "grad_norm": 0.3036490354198925,
+      "learning_rate": 1.276115892225764e-05,
+      "loss": 0.1123,
+      "step": 807
+    },
+    {
+      "epoch": 3.27746835443038,
+      "grad_norm": 0.31591443785998913,
+      "learning_rate": 1.2708277673108555e-05,
+      "loss": 0.1124,
+      "step": 808
+    },
+    {
+      "epoch": 3.281518987341772,
+      "grad_norm": 0.315404594635598,
+      "learning_rate": 1.2655455150488649e-05,
+      "loss": 0.1205,
+      "step": 809
+    },
+    {
+      "epoch": 3.2855696202531646,
+      "grad_norm": 0.3064459182784286,
+      "learning_rate": 1.2602691779823272e-05,
+      "loss": 0.1172,
+      "step": 810
+    },
+    {
+      "epoch": 3.289620253164557,
+      "grad_norm": 0.29684549050244136,
+      "learning_rate": 1.2549987986061355e-05,
+      "loss": 0.1192,
+      "step": 811
+    },
+    {
+      "epoch": 3.293670886075949,
+      "grad_norm": 0.29454842090597244,
+      "learning_rate": 1.2497344193672005e-05,
+      "loss": 0.1276,
+      "step": 812
+    },
+    {
+      "epoch": 3.2977215189873417,
+      "grad_norm": 0.29274188502643944,
+      "learning_rate": 1.2444760826641092e-05,
+      "loss": 0.1039,
+      "step": 813
+    },
+    {
+      "epoch": 3.301772151898734,
+      "grad_norm": 0.3104923228474782,
+      "learning_rate": 1.2392238308467817e-05,
+      "loss": 0.1159,
+      "step": 814
+    },
+    {
+      "epoch": 3.3058227848101267,
+      "grad_norm": 0.28999708652202844,
+      "learning_rate": 1.2339777062161326e-05,
+      "loss": 0.1161,
+      "step": 815
+    },
+    {
+      "epoch": 3.309873417721519,
+      "grad_norm": 0.29568683150498604,
+      "learning_rate": 1.2287377510237293e-05,
+      "loss": 0.1239,
+      "step": 816
+    },
+    {
+      "epoch": 3.3139240506329113,
+      "grad_norm": 0.3293786292585372,
+      "learning_rate": 1.2235040074714488e-05,
+      "loss": 0.1031,
+      "step": 817
+    },
+    {
+      "epoch": 3.317974683544304,
+      "grad_norm": 0.29485389190977,
+      "learning_rate": 1.2182765177111434e-05,
+      "loss": 0.1278,
+      "step": 818
+    },
+    {
+      "epoch": 3.3220253164556963,
+      "grad_norm": 0.3061033184305001,
+      "learning_rate": 1.213055323844297e-05,
+      "loss": 0.1202,
+      "step": 819
+    },
+    {
+      "epoch": 3.3260759493670884,
+      "grad_norm": 0.30813600291518234,
+      "learning_rate": 1.2078404679216864e-05,
+      "loss": 0.1148,
+      "step": 820
+    },
+    {
+      "epoch": 3.330126582278481,
+      "grad_norm": 0.2825066435575841,
+      "learning_rate": 1.2026319919430458e-05,
+      "loss": 0.1237,
+      "step": 821
+    },
+    {
+      "epoch": 3.3341772151898734,
+      "grad_norm": 0.2984609891582971,
+      "learning_rate": 1.1974299378567227e-05,
+      "loss": 0.1149,
+      "step": 822
+    },
+    {
+      "epoch": 3.338227848101266,
+      "grad_norm": 0.29730496870780093,
+      "learning_rate": 1.1922343475593462e-05,
+      "loss": 0.1285,
+      "step": 823
+    },
+    {
+      "epoch": 3.3422784810126585,
+      "grad_norm": 0.3098946183075653,
+      "learning_rate": 1.187045262895488e-05,
+      "loss": 0.1207,
+      "step": 824
+    },
+    {
+      "epoch": 3.3463291139240505,
+      "grad_norm": 0.29805790913334124,
+      "learning_rate": 1.1818627256573203e-05,
+      "loss": 0.1239,
+      "step": 825
+    },
+    {
+      "epoch": 3.350379746835443,
+      "grad_norm": 0.3097371104822833,
+      "learning_rate": 1.1766867775842864e-05,
+      "loss": 0.1267,
+      "step": 826
+    },
+    {
+      "epoch": 3.3544303797468356,
+      "grad_norm": 0.29305863524369036,
+      "learning_rate": 1.1715174603627615e-05,
+      "loss": 0.1159,
+      "step": 827
+    },
+    {
+      "epoch": 3.3584810126582276,
+      "grad_norm": 0.2939846856571145,
+      "learning_rate": 1.1663548156257147e-05,
+      "loss": 0.1287,
+      "step": 828
+    },
+    {
+      "epoch": 3.36253164556962,
+      "grad_norm": 0.32411879744573757,
+      "learning_rate": 1.161198884952377e-05,
+      "loss": 0.1268,
+      "step": 829
+    },
+    {
+      "epoch": 3.3665822784810127,
+      "grad_norm": 0.3251628800936934,
+      "learning_rate": 1.1560497098679056e-05,
+      "loss": 0.1214,
+      "step": 830
+    },
+    {
+      "epoch": 3.370632911392405,
+      "grad_norm": 0.30291601275098573,
+      "learning_rate": 1.1509073318430479e-05,
+      "loss": 0.1368,
+      "step": 831
+    },
+    {
+      "epoch": 3.3746835443037977,
+      "grad_norm": 0.3147905330241073,
+      "learning_rate": 1.1457717922938116e-05,
+      "loss": 0.1188,
+      "step": 832
+    },
+    {
+      "epoch": 3.3787341772151898,
+      "grad_norm": 0.3208438999289737,
+      "learning_rate": 1.1406431325811233e-05,
+      "loss": 0.1255,
+      "step": 833
+    },
+    {
+      "epoch": 3.3827848101265823,
+      "grad_norm": 0.3018182856111235,
+      "learning_rate": 1.135521394010506e-05,
+      "loss": 0.1377,
+      "step": 834
+    },
+    {
+      "epoch": 3.386835443037975,
+      "grad_norm": 0.29543838500234515,
+      "learning_rate": 1.1304066178317367e-05,
+      "loss": 0.1211,
+      "step": 835
+    },
+    {
+      "epoch": 3.390886075949367,
+      "grad_norm": 0.2866700631373677,
+      "learning_rate": 1.1252988452385199e-05,
+      "loss": 0.1228,
+      "step": 836
+    },
+    {
+      "epoch": 3.3949367088607594,
+      "grad_norm": 0.2911719909894494,
+      "learning_rate": 1.1201981173681536e-05,
+      "loss": 0.1223,
+      "step": 837
+    },
+    {
+      "epoch": 3.398987341772152,
+      "grad_norm": 0.2951899593029513,
+      "learning_rate": 1.1151044753011991e-05,
+      "loss": 0.114,
+      "step": 838
+    },
+    {
+      "epoch": 3.4030379746835444,
+      "grad_norm": 0.28322892558896695,
+      "learning_rate": 1.1100179600611491e-05,
+      "loss": 0.1202,
+      "step": 839
+    },
+    {
+      "epoch": 3.407088607594937,
+      "grad_norm": 0.29448052828359117,
+      "learning_rate": 1.1049386126140985e-05,
+      "loss": 0.1223,
+      "step": 840
+    },
+    {
+      "epoch": 3.411139240506329,
+      "grad_norm": 0.3058968516560319,
+      "learning_rate": 1.0998664738684128e-05,
+      "loss": 0.1013,
+      "step": 841
+    },
+    {
+      "epoch": 3.4151898734177215,
+      "grad_norm": 0.2848116174766817,
+      "learning_rate": 1.0948015846744e-05,
+      "loss": 0.1342,
+      "step": 842
+    },
+    {
+      "epoch": 3.419240506329114,
+      "grad_norm": 0.2858459195114275,
+      "learning_rate": 1.0897439858239832e-05,
+      "loss": 0.1424,
+      "step": 843
+    },
+    {
+      "epoch": 3.423291139240506,
+      "grad_norm": 0.2896584838655389,
+      "learning_rate": 1.0846937180503652e-05,
+      "loss": 0.1255,
+      "step": 844
+    },
+    {
+      "epoch": 3.4273417721518986,
+      "grad_norm": 0.2962162393039344,
+      "learning_rate": 1.0796508220277117e-05,
+      "loss": 0.1163,
+      "step": 845
+    },
+    {
+      "epoch": 3.431392405063291,
+      "grad_norm": 0.2858956918778084,
+      "learning_rate": 1.0746153383708107e-05,
+      "loss": 0.1164,
+      "step": 846
+    },
+    {
+      "epoch": 3.4354430379746836,
+      "grad_norm": 0.292367583588525,
+      "learning_rate": 1.0695873076347579e-05,
+      "loss": 0.1148,
+      "step": 847
+    },
+    {
+      "epoch": 3.439493670886076,
+      "grad_norm": 0.28322465478441133,
+      "learning_rate": 1.0645667703146205e-05,
+      "loss": 0.1097,
+      "step": 848
+    },
+    {
+      "epoch": 3.4435443037974682,
+      "grad_norm": 0.3248741359848856,
+      "learning_rate": 1.0595537668451161e-05,
+      "loss": 0.1232,
+      "step": 849
+    },
+    {
+      "epoch": 3.4475949367088607,
+      "grad_norm": 0.29730671998826763,
+      "learning_rate": 1.0545483376002854e-05,
+      "loss": 0.1244,
+      "step": 850
+    },
+    {
+      "epoch": 3.4516455696202533,
+      "grad_norm": 0.28812385164505827,
+      "learning_rate": 1.0495505228931676e-05,
+      "loss": 0.1187,
+      "step": 851
+    },
+    {
+      "epoch": 3.4556962025316453,
+      "grad_norm": 0.2965904354257529,
+      "learning_rate": 1.044560362975474e-05,
+      "loss": 0.1208,
+      "step": 852
+    },
+    {
+      "epoch": 3.459746835443038,
+      "grad_norm": 0.2778103460126865,
+      "learning_rate": 1.0395778980372695e-05,
+      "loss": 0.1254,
+      "step": 853
+    },
+    {
+      "epoch": 3.4637974683544304,
+      "grad_norm": 0.2816639358160425,
+      "learning_rate": 1.0346031682066381e-05,
+      "loss": 0.1237,
+      "step": 854
+    },
+    {
+      "epoch": 3.467848101265823,
+      "grad_norm": 0.2990851346893387,
+      "learning_rate": 1.0296362135493724e-05,
+      "loss": 0.1093,
+      "step": 855
+    },
+    {
+      "epoch": 3.4718987341772154,
+      "grad_norm": 0.2934256375975188,
+      "learning_rate": 1.0246770740686422e-05,
+      "loss": 0.1197,
+      "step": 856
+    },
+    {
+      "epoch": 3.4759493670886075,
+      "grad_norm": 0.28757616522179114,
+      "learning_rate": 1.0197257897046743e-05,
+      "loss": 0.116,
+      "step": 857
+    },
+    {
+      "epoch": 3.48,
+      "grad_norm": 0.2987139977971812,
+      "learning_rate": 1.014782400334433e-05,
+      "loss": 0.1255,
+      "step": 858
+    },
+    {
+      "epoch": 3.4840506329113925,
+      "grad_norm": 0.2968393133559736,
+      "learning_rate": 1.009846945771296e-05,
+      "loss": 0.1198,
+      "step": 859
+    },
+    {
+      "epoch": 3.4881012658227846,
+      "grad_norm": 0.28865728653950967,
+      "learning_rate": 1.0049194657647363e-05,
+      "loss": 0.1188,
+      "step": 860
+    },
+    {
+      "epoch": 3.492151898734177,
+      "grad_norm": 0.2984573957423054,
+      "learning_rate": 1.0000000000000006e-05,
+      "loss": 0.1147,
+      "step": 861
+    },
+    {
+      "epoch": 3.4962025316455696,
+      "grad_norm": 0.28367415727124967,
+      "learning_rate": 9.950885880977891e-06,
+      "loss": 0.1081,
+      "step": 862
+    },
+    {
+      "epoch": 3.500253164556962,
+      "grad_norm": 0.2952798828382643,
+      "learning_rate": 9.901852696139382e-06,
+      "loss": 0.1264,
+      "step": 863
+    },
+    {
+      "epoch": 3.5043037974683546,
+      "grad_norm": 0.28847091985279244,
+      "learning_rate": 9.852900840391027e-06,
+      "loss": 0.1246,
+      "step": 864
+    },
+    {
+      "epoch": 3.5083544303797467,
+      "grad_norm": 0.2983167613602088,
+      "learning_rate": 9.804030707984313e-06,
+      "loss": 0.1336,
+      "step": 865
+    },
+    {
+      "epoch": 3.512405063291139,
+      "grad_norm": 0.2976258317270268,
+      "learning_rate": 9.755242692512599e-06,
+      "loss": 0.1142,
+      "step": 866
+    },
+    {
+      "epoch": 3.5164556962025317,
+      "grad_norm": 0.2832843593093659,
+      "learning_rate": 9.70653718690782e-06,
+      "loss": 0.1208,
+      "step": 867
+    },
+    {
+      "epoch": 3.520506329113924,
+      "grad_norm": 0.2915538355811435,
+      "learning_rate": 9.657914583437454e-06,
+      "loss": 0.1157,
+      "step": 868
+    },
+    {
+      "epoch": 3.5245569620253163,
+      "grad_norm": 0.2867312640710937,
+      "learning_rate": 9.609375273701246e-06,
+      "loss": 0.1147,
+      "step": 869
+    },
+    {
+      "epoch": 3.528607594936709,
+      "grad_norm": 0.3042429453666043,
+      "learning_rate": 9.560919648628133e-06,
+      "loss": 0.122,
+      "step": 870
+    },
+    {
+      "epoch": 3.5326582278481014,
+      "grad_norm": 0.27982042597424606,
+      "learning_rate": 9.512548098473047e-06,
+      "loss": 0.1189,
+      "step": 871
+    },
+    {
+      "epoch": 3.536708860759494,
+      "grad_norm": 0.2889953645696949,
+      "learning_rate": 9.464261012813825e-06,
+      "loss": 0.1182,
+      "step": 872
+    },
+    {
+      "epoch": 3.540759493670886,
+      "grad_norm": 0.2774141819262852,
+      "learning_rate": 9.416058780547987e-06,
+      "loss": 0.1114,
+      "step": 873
+    },
+    {
+      "epoch": 3.5448101265822785,
+      "grad_norm": 0.2982215954380378,
+      "learning_rate": 9.367941789889714e-06,
+      "loss": 0.1163,
+      "step": 874
+    },
+    {
+      "epoch": 3.548860759493671,
+      "grad_norm": 0.27910882055946623,
+      "learning_rate": 9.319910428366607e-06,
+      "loss": 0.1196,
+      "step": 875
+    },
+    {
+      "epoch": 3.552911392405063,
+      "grad_norm": 0.2927036048417553,
+      "learning_rate": 9.271965082816667e-06,
+      "loss": 0.1197,
+      "step": 876
+    },
+    {
+      "epoch": 3.5569620253164556,
+      "grad_norm": 0.28828961505301504,
+      "learning_rate": 9.224106139385111e-06,
+      "loss": 0.1236,
+      "step": 877
+    },
+    {
+      "epoch": 3.561012658227848,
+      "grad_norm": 0.29680098800434523,
+      "learning_rate": 9.176333983521291e-06,
+      "loss": 0.1172,
+      "step": 878
+    },
+    {
+      "epoch": 3.5650632911392406,
+      "grad_norm": 0.284789294969407,
+      "learning_rate": 9.12864899997558e-06,
+      "loss": 0.1213,
+      "step": 879
+    },
+    {
+      "epoch": 3.569113924050633,
+      "grad_norm": 0.27970610310220567,
+      "learning_rate": 9.08105157279628e-06,
+      "loss": 0.1108,
+      "step": 880
+    },
+    {
+      "epoch": 3.573164556962025,
+      "grad_norm": 0.29143268791067745,
+      "learning_rate": 9.03354208532653e-06,
+      "loss": 0.1224,
+      "step": 881
+    },
+    {
+      "epoch": 3.5772151898734177,
+      "grad_norm": 0.30370818738178235,
+      "learning_rate": 8.986120920201205e-06,
+      "loss": 0.1114,
+      "step": 882
+    },
+    {
+      "epoch": 3.58126582278481,
+      "grad_norm": 0.2797006679176373,
+      "learning_rate": 8.938788459343852e-06,
+      "loss": 0.1164,
+      "step": 883
+    },
+    {
+      "epoch": 3.5853164556962023,
+      "grad_norm": 0.28776678908442505,
+      "learning_rate": 8.8915450839636e-06,
+      "loss": 0.1108,
+      "step": 884
+    },
+    {
+      "epoch": 3.589367088607595,
+      "grad_norm": 0.292348411036141,
+      "learning_rate": 8.844391174552116e-06,
+      "loss": 0.121,
+      "step": 885
+    },
+    {
+      "epoch": 3.5934177215189873,
+      "grad_norm": 0.2908579512120188,
+      "learning_rate": 8.797327110880479e-06,
+      "loss": 0.1152,
+      "step": 886
+    },
+    {
+      "epoch": 3.59746835443038,
+      "grad_norm": 0.2835103316103552,
+      "learning_rate": 8.750353271996206e-06,
+      "loss": 0.1117,
+      "step": 887
+    },
+    {
+      "epoch": 3.6015189873417723,
+      "grad_norm": 0.28304603542164597,
+      "learning_rate": 8.703470036220132e-06,
+      "loss": 0.1192,
+      "step": 888
+    },
+    {
+      "epoch": 3.6055696202531644,
+      "grad_norm": 0.28516957703873425,
+      "learning_rate": 8.656677781143394e-06,
+      "loss": 0.1157,
+      "step": 889
+    },
+    {
+      "epoch": 3.609620253164557,
+      "grad_norm": 0.29962831183229355,
+      "learning_rate": 8.609976883624377e-06,
+      "loss": 0.1201,
+      "step": 890
+    },
+    {
+      "epoch": 3.6136708860759494,
+      "grad_norm": 0.28820163012627387,
+      "learning_rate": 8.563367719785698e-06,
+      "loss": 0.131,
+      "step": 891
+    },
+    {
+      "epoch": 3.6177215189873415,
+      "grad_norm": 0.2911940277300189,
+      "learning_rate": 8.516850665011138e-06,
+      "loss": 0.1241,
+      "step": 892
+    },
+    {
+      "epoch": 3.621772151898734,
+      "grad_norm": 0.290555582710059,
+      "learning_rate": 8.47042609394269e-06,
+      "loss": 0.1207,
+      "step": 893
+    },
+    {
+      "epoch": 3.6258227848101265,
+      "grad_norm": 0.30551122782176293,
+      "learning_rate": 8.424094380477432e-06,
+      "loss": 0.1281,
+      "step": 894
+    },
+    {
+      "epoch": 3.629873417721519,
+      "grad_norm": 0.2905878042065041,
+      "learning_rate": 8.37785589776465e-06,
+      "loss": 0.1221,
+      "step": 895
+    },
+    {
+      "epoch": 3.6339240506329116,
+      "grad_norm": 0.2806764442940441,
+      "learning_rate": 8.331711018202694e-06,
+      "loss": 0.1299,
+      "step": 896
+    },
+    {
+      "epoch": 3.6379746835443036,
+      "grad_norm": 0.29779539192140597,
+      "learning_rate": 8.285660113436104e-06,
+      "loss": 0.1082,
+      "step": 897
+    },
+    {
+      "epoch": 3.642025316455696,
+      "grad_norm": 0.28954650969237006,
+      "learning_rate": 8.239703554352527e-06,
+      "loss": 0.1116,
+      "step": 898
+    },
+    {
+      "epoch": 3.6460759493670887,
+      "grad_norm": 0.27610791619730346,
+      "learning_rate": 8.193841711079775e-06,
+      "loss": 0.123,
+      "step": 899
+    },
+    {
+      "epoch": 3.6501265822784807,
+      "grad_norm": 0.2957682631287175,
+      "learning_rate": 8.148074952982828e-06,
+      "loss": 0.1239,
+      "step": 900
+    },
+    {
+      "epoch": 3.6541772151898733,
+      "grad_norm": 0.2875784820075001,
+      "learning_rate": 8.102403648660859e-06,
+      "loss": 0.1093,
+      "step": 901
+    },
+    {
+      "epoch": 3.6582278481012658,
+      "grad_norm": 0.2935952297746027,
+      "learning_rate": 8.056828165944282e-06,
+      "loss": 0.1204,
+      "step": 902
+    },
+    {
+      "epoch": 3.6622784810126583,
+      "grad_norm": 0.2994468831151663,
+      "learning_rate": 8.011348871891762e-06,
+      "loss": 0.1131,
+      "step": 903
+    },
+    {
+      "epoch": 3.666329113924051,
+      "grad_norm": 0.2890210524994184,
+      "learning_rate": 7.965966132787287e-06,
+      "loss": 0.1069,
+      "step": 904
+    },
+    {
+      "epoch": 3.670379746835443,
+      "grad_norm": 0.28926566230233486,
+      "learning_rate": 7.920680314137189e-06,
+      "loss": 0.1249,
+      "step": 905
+    },
+    {
+      "epoch": 3.6744303797468354,
+      "grad_norm": 0.27826225166410357,
+      "learning_rate": 7.875491780667246e-06,
+      "loss": 0.1193,
+      "step": 906
+    },
+    {
+      "epoch": 3.678481012658228,
+      "grad_norm": 0.2880036948154761,
+      "learning_rate": 7.830400896319667e-06,
+      "loss": 0.1117,
+      "step": 907
+    },
+    {
+      "epoch": 3.68253164556962,
+      "grad_norm": 0.28959891619381667,
+      "learning_rate": 7.785408024250259e-06,
+      "loss": 0.1235,
+      "step": 908
+    },
+    {
+      "epoch": 3.6865822784810125,
+      "grad_norm": 0.2853153269069459,
+      "learning_rate": 7.74051352682542e-06,
+      "loss": 0.1188,
+      "step": 909
+    },
+    {
+      "epoch": 3.690632911392405,
+      "grad_norm": 0.28742063424491204,
+      "learning_rate": 7.695717765619257e-06,
+      "loss": 0.131,
+      "step": 910
+    },
+    {
+      "epoch": 3.6946835443037975,
+      "grad_norm": 0.28198603077248674,
+      "learning_rate": 7.651021101410673e-06,
+      "loss": 0.1246,
+      "step": 911
+    },
+    {
+      "epoch": 3.69873417721519,
+      "grad_norm": 0.3030096851674664,
+      "learning_rate": 7.606423894180464e-06,
+      "loss": 0.1066,
+      "step": 912
+    },
+    {
+      "epoch": 3.702784810126582,
+      "grad_norm": 0.2935479868146617,
+      "learning_rate": 7.56192650310839e-06,
+      "loss": 0.1107,
+      "step": 913
+    },
+    {
+      "epoch": 3.7068354430379746,
+      "grad_norm": 0.29102996023877725,
+      "learning_rate": 7.517529286570349e-06,
+      "loss": 0.1221,
+      "step": 914
+    },
+    {
+      "epoch": 3.710886075949367,
+      "grad_norm": 0.297831961559691,
+      "learning_rate": 7.473232602135387e-06,
+      "loss": 0.1039,
+      "step": 915
+    },
+    {
+      "epoch": 3.714936708860759,
+      "grad_norm": 0.2859041324393695,
+      "learning_rate": 7.429036806562935e-06,
+      "loss": 0.1173,
+      "step": 916
+    },
+    {
+      "epoch": 3.7189873417721517,
+      "grad_norm": 0.29124024581719415,
+      "learning_rate": 7.3849422557998455e-06,
+      "loss": 0.1132,
+      "step": 917
+    },
+    {
+      "epoch": 3.7230379746835442,
+      "grad_norm": 0.28335791634191265,
+      "learning_rate": 7.340949304977567e-06,
+      "loss": 0.1219,
+      "step": 918
+    },
+    {
+      "epoch": 3.7270886075949368,
+      "grad_norm": 0.28338293128272307,
+      "learning_rate": 7.297058308409282e-06,
+      "loss": 0.1264,
+      "step": 919
+    },
+    {
+      "epoch": 3.7311392405063293,
+      "grad_norm": 0.2834952319056188,
+      "learning_rate": 7.25326961958704e-06,
+      "loss": 0.1227,
+      "step": 920
+    },
+    {
+      "epoch": 3.7351898734177214,
+      "grad_norm": 0.30122360499168754,
+      "learning_rate": 7.209583591178921e-06,
+      "loss": 0.1311,
+      "step": 921
+    },
+    {
+      "epoch": 3.739240506329114,
+      "grad_norm": 0.27642747503869397,
+      "learning_rate": 7.1660005750261925e-06,
+      "loss": 0.1153,
+      "step": 922
+    },
+    {
+      "epoch": 3.7432911392405064,
+      "grad_norm": 0.281336735131628,
+      "learning_rate": 7.1225209221404765e-06,
+      "loss": 0.1166,
+      "step": 923
+    },
+    {
+      "epoch": 3.747341772151899,
+      "grad_norm": 0.2798155071576694,
+      "learning_rate": 7.079144982700909e-06,
+      "loss": 0.1076,
+      "step": 924
+    },
+    {
+      "epoch": 3.7513924050632914,
+      "grad_norm": 0.26894058240995744,
+      "learning_rate": 7.0358731060513695e-06,
+      "loss": 0.1243,
+      "step": 925
+    },
+    {
+      "epoch": 3.7554430379746835,
+      "grad_norm": 0.292641234118352,
+      "learning_rate": 6.99270564069757e-06,
+      "loss": 0.1121,
+      "step": 926
+    },
+    {
+      "epoch": 3.759493670886076,
+      "grad_norm": 0.27342980796011546,
+      "learning_rate": 6.949642934304375e-06,
+      "loss": 0.1281,
+      "step": 927
+    },
+    {
+      "epoch": 3.7635443037974685,
+      "grad_norm": 0.27518044982576506,
+      "learning_rate": 6.906685333692871e-06,
+      "loss": 0.1228,
+      "step": 928
+    },
+    {
+      "epoch": 3.7675949367088606,
+      "grad_norm": 0.2909559654571805,
+      "learning_rate": 6.86383318483769e-06,
+      "loss": 0.1144,
+      "step": 929
+    },
+    {
+      "epoch": 3.771645569620253,
+      "grad_norm": 0.28190841104202624,
+      "learning_rate": 6.821086832864139e-06,
+      "loss": 0.1196,
+      "step": 930
+    },
+    {
+      "epoch": 3.7756962025316456,
+      "grad_norm": 0.2816684638854699,
+      "learning_rate": 6.77844662204546e-06,
+      "loss": 0.1216,
+      "step": 931
+    },
+    {
+      "epoch": 3.779746835443038,
+      "grad_norm": 0.27434031728950636,
+      "learning_rate": 6.7359128958000455e-06,
+      "loss": 0.116,
+      "step": 932
+    },
+    {
+      "epoch": 3.7837974683544306,
+      "grad_norm": 0.2786145000461609,
+      "learning_rate": 6.693485996688695e-06,
+      "loss": 0.1116,
+      "step": 933
+    },
+    {
+      "epoch": 3.7878481012658227,
+      "grad_norm": 0.2744994101942207,
+      "learning_rate": 6.651166266411801e-06,
+      "loss": 0.1258,
+      "step": 934
+    },
+    {
+      "epoch": 3.7918987341772152,
+      "grad_norm": 0.2693385992140348,
+      "learning_rate": 6.6089540458066725e-06,
+      "loss": 0.1259,
+      "step": 935
+    },
+    {
+      "epoch": 3.7959493670886078,
+      "grad_norm": 0.26815199469907885,
+      "learning_rate": 6.566849674844711e-06,
+      "loss": 0.1144,
+      "step": 936
+    },
+    {
+      "epoch": 3.8,
+      "grad_norm": 0.2802595838432032,
+      "learning_rate": 6.524853492628747e-06,
+      "loss": 0.1219,
+      "step": 937
+    },
+    {
+      "epoch": 3.8040506329113923,
+      "grad_norm": 0.29378563909401745,
+      "learning_rate": 6.4829658373902536e-06,
+      "loss": 0.1103,
+      "step": 938
+    },
+    {
+      "epoch": 3.808101265822785,
+      "grad_norm": 0.2799623914956454,
+      "learning_rate": 6.441187046486648e-06,
+      "loss": 0.1154,
+      "step": 939
+    },
+    {
+      "epoch": 3.8121518987341774,
+      "grad_norm": 0.2876837372603903,
+      "learning_rate": 6.399517456398567e-06,
+      "loss": 0.1148,
+      "step": 940
+    },
+    {
+      "epoch": 3.81620253164557,
+      "grad_norm": 0.2775463461914315,
+      "learning_rate": 6.357957402727164e-06,
+      "loss": 0.1183,
+      "step": 941
+    },
+    {
+      "epoch": 3.820253164556962,
+      "grad_norm": 0.2861776437574836,
+      "learning_rate": 6.316507220191395e-06,
+      "loss": 0.1096,
+      "step": 942
+    },
+    {
+      "epoch": 3.8243037974683545,
+      "grad_norm": 0.27659013899309676,
+      "learning_rate": 6.275167242625331e-06,
+      "loss": 0.1193,
+      "step": 943
+    },
+    {
+      "epoch": 3.828354430379747,
+      "grad_norm": 0.28146941812332327,
+      "learning_rate": 6.233937802975471e-06,
+      "loss": 0.1161,
+      "step": 944
+    },
+    {
+      "epoch": 3.832405063291139,
+      "grad_norm": 0.2734277161565118,
+      "learning_rate": 6.192819233298046e-06,
+      "loss": 0.1157,
+      "step": 945
+    },
+    {
+      "epoch": 3.8364556962025316,
+      "grad_norm": 0.2802267995700749,
+      "learning_rate": 6.151811864756383e-06,
+      "loss": 0.1098,
+      "step": 946
+    },
+    {
+      "epoch": 3.840506329113924,
+      "grad_norm": 0.27842245304520197,
+      "learning_rate": 6.1109160276181655e-06,
+      "loss": 0.1068,
+      "step": 947
+    },
+    {
+      "epoch": 3.8445569620253166,
+      "grad_norm": 0.275214044160783,
+      "learning_rate": 6.070132051252868e-06,
+      "loss": 0.1125,
+      "step": 948
+    },
+    {
+      "epoch": 3.848607594936709,
+      "grad_norm": 0.28848770080732566,
+      "learning_rate": 6.0294602641290034e-06,
+      "loss": 0.1296,
+      "step": 949
+    },
+    {
+      "epoch": 3.852658227848101,
+      "grad_norm": 0.27963040774112413,
+      "learning_rate": 5.988900993811575e-06,
+      "loss": 0.1167,
+      "step": 950
+    },
+    {
+      "epoch": 3.8567088607594937,
+      "grad_norm": 0.2698730171579938,
+      "learning_rate": 5.948454566959363e-06,
+      "loss": 0.1119,
+      "step": 951
+    },
+    {
+      "epoch": 3.8607594936708862,
+      "grad_norm": 0.2683646637193907,
+      "learning_rate": 5.908121309322328e-06,
+      "loss": 0.1276,
+      "step": 952
+    },
+    {
+      "epoch": 3.8648101265822783,
+      "grad_norm": 0.27695554961958346,
+      "learning_rate": 5.867901545738976e-06,
+      "loss": 0.1292,
+      "step": 953
+    },
+    {
+      "epoch": 3.868860759493671,
+      "grad_norm": 0.270615496146439,
+      "learning_rate": 5.827795600133774e-06,
+      "loss": 0.1059,
+      "step": 954
+    },
+    {
+      "epoch": 3.8729113924050633,
+      "grad_norm": 0.2790900955204084,
+      "learning_rate": 5.787803795514466e-06,
+      "loss": 0.1134,
+      "step": 955
+    },
+    {
+      "epoch": 3.876962025316456,
+      "grad_norm": 0.2795600397292093,
+      "learning_rate": 5.747926453969576e-06,
+      "loss": 0.1138,
+      "step": 956
+    },
+    {
+      "epoch": 3.8810126582278484,
+      "grad_norm": 0.2662768950099552,
+      "learning_rate": 5.708163896665708e-06,
+      "loss": 0.1151,
+      "step": 957
+    },
+    {
+      "epoch": 3.8850632911392404,
+      "grad_norm": 0.27581700832692163,
+      "learning_rate": 5.668516443845047e-06,
+      "loss": 0.1159,
+      "step": 958
+    },
+    {
+      "epoch": 3.889113924050633,
+      "grad_norm": 0.286018430099024,
+      "learning_rate": 5.6289844148227225e-06,
+      "loss": 0.1074,
+      "step": 959
+    },
+    {
+      "epoch": 3.8931645569620255,
+      "grad_norm": 0.27986275583450143,
+      "learning_rate": 5.5895681279842615e-06,
+      "loss": 0.1154,
+      "step": 960
+    },
+    {
+      "epoch": 3.8972151898734175,
+      "grad_norm": 0.28135645414744465,
+      "learning_rate": 5.550267900783019e-06,
+      "loss": 0.1183,
+      "step": 961
+    },
+    {
+      "epoch": 3.90126582278481,
+      "grad_norm": 0.27761863745117965,
+      "learning_rate": 5.511084049737623e-06,
+      "loss": 0.1014,
+      "step": 962
+    },
+    {
+      "epoch": 3.9053164556962026,
+      "grad_norm": 0.2856860995475159,
+      "learning_rate": 5.4720168904294215e-06,
+      "loss": 0.1083,
+      "step": 963
+    },
+    {
+      "epoch": 3.909367088607595,
+      "grad_norm": 0.27549826683237333,
+      "learning_rate": 5.433066737499948e-06,
+      "loss": 0.1228,
+      "step": 964
+    },
+    {
+      "epoch": 3.9134177215189876,
+      "grad_norm": 0.27134354064225585,
+      "learning_rate": 5.394233904648376e-06,
+      "loss": 0.1215,
+      "step": 965
+    },
+    {
+      "epoch": 3.9174683544303797,
+      "grad_norm": 0.271881020199358,
+      "learning_rate": 5.355518704628997e-06,
+      "loss": 0.1167,
+      "step": 966
+    },
+    {
+      "epoch": 3.921518987341772,
+      "grad_norm": 0.27089450268063014,
+      "learning_rate": 5.316921449248731e-06,
+      "loss": 0.1116,
+      "step": 967
+    },
+    {
+      "epoch": 3.9255696202531647,
+      "grad_norm": 0.2696985740874705,
+      "learning_rate": 5.278442449364538e-06,
+      "loss": 0.12,
+      "step": 968
+    },
+    {
+      "epoch": 3.9296202531645568,
+      "grad_norm": 0.27373565620715845,
+      "learning_rate": 5.240082014881016e-06,
+      "loss": 0.1126,
+      "step": 969
+    },
+    {
+      "epoch": 3.9336708860759493,
+      "grad_norm": 0.27635021876216875,
+      "learning_rate": 5.201840454747822e-06,
+      "loss": 0.1222,
+      "step": 970
+    },
+    {
+      "epoch": 3.937721518987342,
+      "grad_norm": 0.26995235836063103,
+      "learning_rate": 5.163718076957223e-06,
+      "loss": 0.1154,
+      "step": 971
+    },
+    {
+      "epoch": 3.9417721518987343,
+      "grad_norm": 0.27442972807522215,
+      "learning_rate": 5.125715188541609e-06,
+      "loss": 0.1175,
+      "step": 972
+    },
+    {
+      "epoch": 3.945822784810127,
+      "grad_norm": 0.27721362853070874,
+      "learning_rate": 5.087832095571021e-06,
+      "loss": 0.1188,
+      "step": 973
+    },
+    {
+      "epoch": 3.949873417721519,
+      "grad_norm": 0.27415640861050844,
+      "learning_rate": 5.0500691031506766e-06,
+      "loss": 0.1192,
+      "step": 974
+    },
+    {
+      "epoch": 3.9539240506329114,
+      "grad_norm": 0.26905157902942506,
+      "learning_rate": 5.01242651541854e-06,
+      "loss": 0.1224,
+      "step": 975
+    },
+    {
+      "epoch": 3.957974683544304,
+      "grad_norm": 0.28087768566450777,
+      "learning_rate": 4.974904635542815e-06,
+      "loss": 0.1262,
+      "step": 976
+    },
+    {
+      "epoch": 3.962025316455696,
+      "grad_norm": 0.2736212952857948,
+      "learning_rate": 4.937503765719582e-06,
+      "loss": 0.1054,
+      "step": 977
+    },
+    {
+      "epoch": 3.9660759493670885,
+      "grad_norm": 0.27313212203894843,
+      "learning_rate": 4.900224207170299e-06,
+      "loss": 0.1123,
+      "step": 978
+    },
+    {
+      "epoch": 3.970126582278481,
+      "grad_norm": 0.27409242959481117,
+      "learning_rate": 4.8630662601394065e-06,
+      "loss": 0.1158,
+      "step": 979
+    },
+    {
+      "epoch": 3.9741772151898735,
+      "grad_norm": 0.2754095731731679,
+      "learning_rate": 4.8260302238918995e-06,
+      "loss": 0.1248,
+      "step": 980
+    },
+    {
+      "epoch": 3.978227848101266,
+      "grad_norm": 0.2646765629499829,
+      "learning_rate": 4.789116396710924e-06,
+      "loss": 0.1146,
+      "step": 981
+    },
+    {
+      "epoch": 3.982278481012658,
+      "grad_norm": 0.28801657763216887,
+      "learning_rate": 4.752325075895368e-06,
+      "loss": 0.1223,
+      "step": 982
+    },
+    {
+      "epoch": 3.9863291139240506,
+      "grad_norm": 0.2795947022130087,
+      "learning_rate": 4.715656557757473e-06,
+      "loss": 0.1228,
+      "step": 983
+    },
+    {
+      "epoch": 3.990379746835443,
+      "grad_norm": 0.27067734357337214,
+      "learning_rate": 4.679111137620442e-06,
+      "loss": 0.1112,
+      "step": 984
+    },
+    {
+      "epoch": 3.9944303797468352,
+      "grad_norm": 0.27422215430182084,
+      "learning_rate": 4.6426891098160585e-06,
+      "loss": 0.107,
+      "step": 985
+    },
+    {
+      "epoch": 3.9984810126582278,
+      "grad_norm": 0.27799258599551485,
+      "learning_rate": 4.6063907676823474e-06,
+      "loss": 0.1158,
+      "step": 986
+    },
+    {
+      "epoch": 4.004050632911392,
+      "grad_norm": 0.3141943430196263,
+      "learning_rate": 4.570216403561141e-06,
+      "loss": 0.097,
+      "step": 987
+    },
+    {
+      "epoch": 4.008101265822785,
+      "grad_norm": 0.28163401696085344,
+      "learning_rate": 4.534166308795815e-06,
+      "loss": 0.0705,
+      "step": 988
+    },
+    {
+      "epoch": 4.012151898734177,
+      "grad_norm": 0.2614364352416061,
+      "learning_rate": 4.498240773728859e-06,
+      "loss": 0.0711,
+      "step": 989
+    },
+    {
+      "epoch": 4.01620253164557,
+      "grad_norm": 0.23317355902904877,
+      "learning_rate": 4.462440087699609e-06,
+      "loss": 0.0773,
+      "step": 990
+    },
+    {
+      "epoch": 4.020253164556962,
+      "grad_norm": 0.22040339273136772,
+      "learning_rate": 4.426764539041861e-06,
+      "loss": 0.0653,
+      "step": 991
+    },
+    {
+      "epoch": 4.024303797468354,
+      "grad_norm": 0.3193273068203288,
+      "learning_rate": 4.391214415081582e-06,
+      "loss": 0.0824,
+      "step": 992
+    },
+    {
+      "epoch": 4.028354430379747,
+      "grad_norm": 0.33373023041847627,
+      "learning_rate": 4.355790002134579e-06,
+      "loss": 0.0914,
+      "step": 993
+    },
+    {
+      "epoch": 4.032405063291139,
+      "grad_norm": 0.2932512865696009,
+      "learning_rate": 4.320491585504207e-06,
+      "loss": 0.0615,
+      "step": 994
+    },
+    {
+      "epoch": 4.036455696202531,
+      "grad_norm": 0.2888930390551226,
+      "learning_rate": 4.2853194494790615e-06,
+      "loss": 0.0664,
+      "step": 995
+    },
+    {
+      "epoch": 4.040506329113924,
+      "grad_norm": 0.28703433724634375,
+      "learning_rate": 4.250273877330691e-06,
+      "loss": 0.083,
+      "step": 996
+    },
+    {
+      "epoch": 4.044556962025316,
+      "grad_norm": 0.2497937957510591,
+      "learning_rate": 4.215355151311313e-06,
+      "loss": 0.0747,
+      "step": 997
+    },
+    {
+      "epoch": 4.048607594936709,
+      "grad_norm": 0.2573718881645155,
+      "learning_rate": 4.180563552651542e-06,
+      "loss": 0.0732,
+      "step": 998
+    },
+    {
+      "epoch": 4.052658227848101,
+      "grad_norm": 0.2671407663193828,
+      "learning_rate": 4.145899361558147e-06,
+      "loss": 0.0686,
+      "step": 999
+    },
+    {
+      "epoch": 4.056708860759493,
+      "grad_norm": 0.2724893648257238,
+      "learning_rate": 4.111362857211738e-06,
+      "loss": 0.0787,
+      "step": 1000
+    },
+    {
+      "epoch": 4.060759493670886,
+      "grad_norm": 0.23648895684741564,
+      "learning_rate": 4.076954317764592e-06,
+      "loss": 0.0658,
+      "step": 1001
+    },
+    {
+      "epoch": 4.0648101265822785,
+      "grad_norm": 0.23699386298267666,
+      "learning_rate": 4.042674020338335e-06,
+      "loss": 0.0772,
+      "step": 1002
+    },
+    {
+      "epoch": 4.0688607594936705,
+      "grad_norm": 0.23241803424260163,
+      "learning_rate": 4.0085222410217835e-06,
+      "loss": 0.066,
+      "step": 1003
+    },
+    {
+      "epoch": 4.0729113924050635,
+      "grad_norm": 0.2423183039566781,
+      "learning_rate": 3.974499254868674e-06,
+      "loss": 0.0722,
+      "step": 1004
+    },
+    {
+      "epoch": 4.076962025316456,
+      "grad_norm": 0.2591481228096467,
+      "learning_rate": 3.940605335895451e-06,
+      "loss": 0.0808,
+      "step": 1005
+    },
+    {
+      "epoch": 4.0810126582278485,
+      "grad_norm": 0.23907268635353118,
+      "learning_rate": 3.90684075707908e-06,
+      "loss": 0.069,
+      "step": 1006
+    },
+    {
+      "epoch": 4.085063291139241,
+      "grad_norm": 0.23637379771931205,
+      "learning_rate": 3.8732057903548505e-06,
+      "loss": 0.0805,
+      "step": 1007
+    },
+    {
+      "epoch": 4.089113924050633,
+      "grad_norm": 0.22931693871588965,
+      "learning_rate": 3.8397007066141375e-06,
+      "loss": 0.0685,
+      "step": 1008
+    },
+    {
+      "epoch": 4.093164556962026,
+      "grad_norm": 0.2452292420656254,
+      "learning_rate": 3.806325775702304e-06,
+      "loss": 0.0797,
+      "step": 1009
+    },
+    {
+      "epoch": 4.097215189873418,
+      "grad_norm": 0.23084440342895926,
+      "learning_rate": 3.773081266416434e-06,
+      "loss": 0.0633,
+      "step": 1010
+    },
+    {
+      "epoch": 4.10126582278481,
+      "grad_norm": 0.24108656415245977,
+      "learning_rate": 3.739967446503245e-06,
+      "loss": 0.0751,
+      "step": 1011
+    },
+    {
+      "epoch": 4.105316455696203,
+      "grad_norm": 0.22983421585858402,
+      "learning_rate": 3.706984582656894e-06,
+      "loss": 0.0645,
+      "step": 1012
+    },
+    {
+      "epoch": 4.109367088607595,
+      "grad_norm": 0.24681418279793418,
+      "learning_rate": 3.6741329405168237e-06,
+      "loss": 0.0714,
+      "step": 1013
+    },
+    {
+      "epoch": 4.113417721518988,
+      "grad_norm": 0.2317121890160988,
+      "learning_rate": 3.641412784665648e-06,
+      "loss": 0.0792,
+      "step": 1014
+    },
+    {
+      "epoch": 4.11746835443038,
+      "grad_norm": 0.23979032671469558,
+      "learning_rate": 3.608824378627005e-06,
+      "loss": 0.0753,
+      "step": 1015
+    },
+    {
+      "epoch": 4.121518987341772,
+      "grad_norm": 0.23058330547563133,
+      "learning_rate": 3.5763679848634337e-06,
+      "loss": 0.0716,
+      "step": 1016
+    },
+    {
+      "epoch": 4.125569620253165,
+      "grad_norm": 0.2463205749282097,
+      "learning_rate": 3.544043864774269e-06,
+      "loss": 0.066,
+      "step": 1017
+    },
+    {
+      "epoch": 4.129620253164557,
+      "grad_norm": 0.23767010384535395,
+      "learning_rate": 3.5118522786935282e-06,
+      "loss": 0.077,
+      "step": 1018
+    },
+    {
+      "epoch": 4.133670886075949,
+      "grad_norm": 0.22617331829971196,
+      "learning_rate": 3.479793485887819e-06,
+      "loss": 0.0731,
+      "step": 1019
+    },
+    {
+      "epoch": 4.137721518987342,
+      "grad_norm": 0.23041520147846287,
+      "learning_rate": 3.4478677445542653e-06,
+      "loss": 0.0763,
+      "step": 1020
+    },
+    {
+      "epoch": 4.141772151898734,
+      "grad_norm": 0.23544663983004532,
+      "learning_rate": 3.4160753118183767e-06,
+      "loss": 0.0719,
+      "step": 1021
+    },
+    {
+      "epoch": 4.145822784810127,
+      "grad_norm": 0.23872863904315703,
+      "learning_rate": 3.3844164437320527e-06,
+      "loss": 0.0705,
+      "step": 1022
+    },
+    {
+      "epoch": 4.149873417721519,
+      "grad_norm": 0.23026799944014098,
+      "learning_rate": 3.3528913952714558e-06,
+      "loss": 0.072,
+      "step": 1023
+    },
+    {
+      "epoch": 4.153924050632911,
+      "grad_norm": 0.23172792308772103,
+      "learning_rate": 3.321500420335e-06,
+      "loss": 0.0618,
+      "step": 1024
+    },
+    {
+      "epoch": 4.157974683544304,
+      "grad_norm": 0.22300637265921386,
+      "learning_rate": 3.290243771741275e-06,
+      "loss": 0.0866,
+      "step": 1025
+    },
+    {
+      "epoch": 4.162025316455696,
+      "grad_norm": 0.22070073276687366,
+      "learning_rate": 3.2591217012270325e-06,
+      "loss": 0.0723,
+      "step": 1026
+    },
+    {
+      "epoch": 4.166075949367088,
+      "grad_norm": 0.23416649756535082,
+      "learning_rate": 3.228134459445149e-06,
+      "loss": 0.0679,
+      "step": 1027
+    },
+    {
+      "epoch": 4.170126582278481,
+      "grad_norm": 0.22632568804139408,
+      "learning_rate": 3.1972822959626205e-06,
+      "loss": 0.0776,
+      "step": 1028
+    },
+    {
+      "epoch": 4.174177215189873,
+      "grad_norm": 0.23444986876521764,
+      "learning_rate": 3.166565459258513e-06,
+      "loss": 0.0711,
+      "step": 1029
+    },
+    {
+      "epoch": 4.178227848101266,
+      "grad_norm": 0.22532407526550732,
+      "learning_rate": 3.1359841967220193e-06,
+      "loss": 0.0976,
+      "step": 1030
+    },
+    {
+      "epoch": 4.182278481012658,
+      "grad_norm": 0.22718656326252865,
+      "learning_rate": 3.105538754650419e-06,
+      "loss": 0.0676,
+      "step": 1031
+    },
+    {
+      "epoch": 4.18632911392405,
+      "grad_norm": 0.23122447136102475,
+      "learning_rate": 3.07522937824712e-06,
+      "loss": 0.0679,
+      "step": 1032
+    },
+    {
+      "epoch": 4.190379746835443,
+      "grad_norm": 0.2418584730334152,
+      "learning_rate": 3.0450563116196697e-06,
+      "loss": 0.0714,
+      "step": 1033
+    },
+    {
+      "epoch": 4.194430379746835,
+      "grad_norm": 0.22885727181977056,
+      "learning_rate": 3.0150197977778008e-06,
+      "loss": 0.0868,
+      "step": 1034
+    },
+    {
+      "epoch": 4.1984810126582275,
+      "grad_norm": 0.24406165962783163,
+      "learning_rate": 2.985120078631465e-06,
+      "loss": 0.0803,
+      "step": 1035
+    },
+    {
+      "epoch": 4.2025316455696204,
+      "grad_norm": 0.23778672267848214,
+      "learning_rate": 2.9553573949888893e-06,
+      "loss": 0.0674,
+      "step": 1036
+    },
+    {
+      "epoch": 4.2065822784810125,
+      "grad_norm": 0.23960240469857155,
+      "learning_rate": 2.9257319865546384e-06,
+      "loss": 0.0787,
+      "step": 1037
+    },
+    {
+      "epoch": 4.2106329113924055,
+      "grad_norm": 0.2339080221449331,
+      "learning_rate": 2.896244091927678e-06,
+      "loss": 0.0891,
+      "step": 1038
+    },
+    {
+      "epoch": 4.2146835443037975,
+      "grad_norm": 0.22752587917059697,
+      "learning_rate": 2.8668939485994584e-06,
+      "loss": 0.0825,
+      "step": 1039
+    },
+    {
+      "epoch": 4.21873417721519,
+      "grad_norm": 0.22606483698266205,
+      "learning_rate": 2.837681792951994e-06,
+      "loss": 0.0737,
+      "step": 1040
+    },
+    {
+      "epoch": 4.222784810126583,
+      "grad_norm": 0.23101270983832545,
+      "learning_rate": 2.808607860255981e-06,
+      "loss": 0.0834,
+      "step": 1041
+    },
+    {
+      "epoch": 4.226835443037975,
+      "grad_norm": 0.22637697210671806,
+      "learning_rate": 2.7796723846688634e-06,
+      "loss": 0.0776,
+      "step": 1042
+    },
+    {
+      "epoch": 4.230886075949367,
+      "grad_norm": 0.23124801835490533,
+      "learning_rate": 2.7508755992329937e-06,
+      "loss": 0.0818,
+      "step": 1043
+    },
+    {
+      "epoch": 4.23493670886076,
+      "grad_norm": 0.2217772259871419,
+      "learning_rate": 2.722217735873718e-06,
+      "loss": 0.0741,
+      "step": 1044
+    },
+    {
+      "epoch": 4.238987341772152,
+      "grad_norm": 0.23247388883359257,
+      "learning_rate": 2.6936990253975315e-06,
+      "loss": 0.0673,
+      "step": 1045
+    },
+    {
+      "epoch": 4.243037974683545,
+      "grad_norm": 0.23854948942055945,
+      "learning_rate": 2.665319697490205e-06,
+      "loss": 0.072,
+      "step": 1046
+    },
+    {
+      "epoch": 4.247088607594937,
+      "grad_norm": 0.2320170324832033,
+      "learning_rate": 2.637079980714945e-06,
+      "loss": 0.0757,
+      "step": 1047
+    },
+    {
+      "epoch": 4.251139240506329,
+      "grad_norm": 0.2366160797133911,
+      "learning_rate": 2.6089801025105453e-06,
+      "loss": 0.0766,
+      "step": 1048
+    },
+    {
+      "epoch": 4.255189873417722,
+      "grad_norm": 0.22585522593617371,
+      "learning_rate": 2.581020289189571e-06,
+      "loss": 0.0843,
+      "step": 1049
+    },
+    {
+      "epoch": 4.259240506329114,
+      "grad_norm": 0.23033569059515768,
+      "learning_rate": 2.553200765936501e-06,
+      "loss": 0.0911,
+      "step": 1050
+    },
+    {
+      "epoch": 4.263291139240506,
+      "grad_norm": 0.2306509992651539,
+      "learning_rate": 2.525521756805962e-06,
+      "loss": 0.0725,
+      "step": 1051
+    },
+    {
+      "epoch": 4.267341772151899,
+      "grad_norm": 0.24017081586121786,
+      "learning_rate": 2.497983484720885e-06,
+      "loss": 0.071,
+      "step": 1052
+    },
+    {
+      "epoch": 4.271392405063291,
+      "grad_norm": 0.23382419654169606,
+      "learning_rate": 2.470586171470728e-06,
+      "loss": 0.0762,
+      "step": 1053
+    },
+    {
+      "epoch": 4.275443037974684,
+      "grad_norm": 0.22813231375743165,
+      "learning_rate": 2.4433300377096836e-06,
+      "loss": 0.0728,
+      "step": 1054
+    },
+    {
+      "epoch": 4.279493670886076,
+      "grad_norm": 0.22833254501366135,
+      "learning_rate": 2.4162153029549073e-06,
+      "loss": 0.0634,
+      "step": 1055
+    },
+    {
+      "epoch": 4.283544303797468,
+      "grad_norm": 0.21988586276595595,
+      "learning_rate": 2.3892421855847458e-06,
+      "loss": 0.0859,
+      "step": 1056
+    },
+    {
+      "epoch": 4.287594936708861,
+      "grad_norm": 0.23007884772104983,
+      "learning_rate": 2.362410902836978e-06,
+      "loss": 0.0742,
+      "step": 1057
+    },
+    {
+      "epoch": 4.291645569620253,
+      "grad_norm": 0.22494929462901375,
+      "learning_rate": 2.3357216708070653e-06,
+      "loss": 0.0688,
+      "step": 1058
+    },
+    {
+      "epoch": 4.295696202531645,
+      "grad_norm": 0.22551123290879277,
+      "learning_rate": 2.309174704446411e-06,
+      "loss": 0.068,
+      "step": 1059
+    },
+    {
+      "epoch": 4.299746835443038,
+      "grad_norm": 0.23653198104963452,
+      "learning_rate": 2.2827702175606437e-06,
+      "loss": 0.0674,
+      "step": 1060
+    },
+    {
+      "epoch": 4.30379746835443,
+      "grad_norm": 0.23054312483844933,
+      "learning_rate": 2.256508422807855e-06,
+      "loss": 0.0795,
+      "step": 1061
+    },
+    {
+      "epoch": 4.307848101265823,
+      "grad_norm": 0.24145317954775194,
+      "learning_rate": 2.230389531696946e-06,
+      "loss": 0.0677,
+      "step": 1062
+    },
+    {
+      "epoch": 4.311898734177215,
+      "grad_norm": 0.21837857958563162,
+      "learning_rate": 2.204413754585857e-06,
+      "loss": 0.074,
+      "step": 1063
+    },
+    {
+      "epoch": 4.315949367088607,
+      "grad_norm": 0.23723937686072197,
+      "learning_rate": 2.1785813006799406e-06,
+      "loss": 0.0683,
+      "step": 1064
+    },
+    {
+      "epoch": 4.32,
+      "grad_norm": 0.23338720322499099,
+      "learning_rate": 2.1528923780302224e-06,
+      "loss": 0.0753,
+      "step": 1065
+    },
+    {
+      "epoch": 4.324050632911392,
+      "grad_norm": 0.23149410128509243,
+      "learning_rate": 2.127347193531757e-06,
+      "loss": 0.0748,
+      "step": 1066
+    },
+    {
+      "epoch": 4.328101265822784,
+      "grad_norm": 0.22391833213047865,
+      "learning_rate": 2.101945952921942e-06,
+      "loss": 0.0664,
+      "step": 1067
+    },
+    {
+      "epoch": 4.332151898734177,
+      "grad_norm": 0.23433318436267303,
+      "learning_rate": 2.0766888607788906e-06,
+      "loss": 0.0645,
+      "step": 1068
+    },
+    {
+      "epoch": 4.3362025316455695,
+      "grad_norm": 0.2251735011834636,
+      "learning_rate": 2.0515761205197337e-06,
+      "loss": 0.0778,
+      "step": 1069
+    },
+    {
+      "epoch": 4.340253164556962,
+      "grad_norm": 0.24094313790811897,
+      "learning_rate": 2.0266079343990453e-06,
+      "loss": 0.0689,
+      "step": 1070
+    },
+    {
+      "epoch": 4.3443037974683545,
+      "grad_norm": 0.22302857586925365,
+      "learning_rate": 2.0017845035071494e-06,
+      "loss": 0.0745,
+      "step": 1071
+    },
+    {
+      "epoch": 4.348354430379747,
+      "grad_norm": 0.23304966537854382,
+      "learning_rate": 1.9771060277685537e-06,
+      "loss": 0.0744,
+      "step": 1072
+    },
+    {
+      "epoch": 4.3524050632911395,
+      "grad_norm": 0.22899357698185985,
+      "learning_rate": 1.95257270594031e-06,
+      "loss": 0.0702,
+      "step": 1073
+    },
+    {
+      "epoch": 4.356455696202532,
+      "grad_norm": 0.24562164309701776,
+      "learning_rate": 1.9281847356104188e-06,
+      "loss": 0.0753,
+      "step": 1074
+    },
+    {
+      "epoch": 4.360506329113924,
+      "grad_norm": 0.23896359031103415,
+      "learning_rate": 1.9039423131962365e-06,
+      "loss": 0.0749,
+      "step": 1075
+    },
+    {
+      "epoch": 4.364556962025317,
+      "grad_norm": 0.24292639469431102,
+      "learning_rate": 1.8798456339429027e-06,
+      "loss": 0.0735,
+      "step": 1076
+    },
+    {
+      "epoch": 4.368607594936709,
+      "grad_norm": 0.2378818653623834,
+      "learning_rate": 1.8558948919217612e-06,
+      "loss": 0.0814,
+      "step": 1077
+    },
+    {
+      "epoch": 4.372658227848102,
+      "grad_norm": 0.2294948107168792,
+      "learning_rate": 1.8320902800287954e-06,
+      "loss": 0.0731,
+      "step": 1078
+    },
+    {
+      "epoch": 4.376708860759494,
+      "grad_norm": 0.22745176352371702,
+      "learning_rate": 1.8084319899830726e-06,
+      "loss": 0.0736,
+      "step": 1079
+    },
+    {
+      "epoch": 4.380759493670886,
+      "grad_norm": 0.22866307006259196,
+      "learning_rate": 1.7849202123252097e-06,
+      "loss": 0.0849,
+      "step": 1080
+    },
+    {
+      "epoch": 4.384810126582279,
+      "grad_norm": 0.22557189253037532,
+      "learning_rate": 1.7615551364158401e-06,
+      "loss": 0.0789,
+      "step": 1081
+    },
+    {
+      "epoch": 4.388860759493671,
+      "grad_norm": 0.22702758184993022,
+      "learning_rate": 1.738336950434061e-06,
+      "loss": 0.0744,
+      "step": 1082
+    },
+    {
+      "epoch": 4.392911392405063,
+      "grad_norm": 0.2350360018284904,
+      "learning_rate": 1.715265841375957e-06,
+      "loss": 0.0748,
+      "step": 1083
+    },
+    {
+      "epoch": 4.396962025316456,
+      "grad_norm": 0.23638046298268078,
+      "learning_rate": 1.6923419950530684e-06,
+      "loss": 0.0647,
+      "step": 1084
+    },
+    {
+      "epoch": 4.401012658227848,
+      "grad_norm": 0.22467244749851462,
+      "learning_rate": 1.6695655960909008e-06,
+      "loss": 0.0718,
+      "step": 1085
+    },
+    {
+      "epoch": 4.405063291139241,
+      "grad_norm": 0.2297443750839427,
+      "learning_rate": 1.646936827927441e-06,
+      "loss": 0.0724,
+      "step": 1086
+    },
+    {
+      "epoch": 4.409113924050633,
+      "grad_norm": 0.23772441814861495,
+      "learning_rate": 1.6244558728116766e-06,
+      "loss": 0.0765,
+      "step": 1087
+    },
+    {
+      "epoch": 4.413164556962025,
+      "grad_norm": 0.23474792582469628,
+      "learning_rate": 1.6021229118021265e-06,
+      "loss": 0.0727,
+      "step": 1088
+    },
+    {
+      "epoch": 4.417215189873418,
+      "grad_norm": 0.2240717232933449,
+      "learning_rate": 1.5799381247653967e-06,
+      "loss": 0.08,
+      "step": 1089
+    },
+    {
+      "epoch": 4.42126582278481,
+      "grad_norm": 0.2268745724216572,
+      "learning_rate": 1.5579016903747013e-06,
+      "loss": 0.0725,
+      "step": 1090
+    },
+    {
+      "epoch": 4.425316455696202,
+      "grad_norm": 0.22958491093682049,
+      "learning_rate": 1.5360137861084656e-06,
+      "loss": 0.0808,
+      "step": 1091
+    },
+    {
+      "epoch": 4.429367088607595,
+      "grad_norm": 0.22520477896404104,
+      "learning_rate": 1.5142745882488475e-06,
+      "loss": 0.0756,
+      "step": 1092
+    },
+    {
+      "epoch": 4.433417721518987,
+      "grad_norm": 0.23999798714539694,
+      "learning_rate": 1.4926842718803691e-06,
+      "loss": 0.0786,
+      "step": 1093
+    },
+    {
+      "epoch": 4.43746835443038,
+      "grad_norm": 0.219675639468252,
+      "learning_rate": 1.4712430108884657e-06,
+      "loss": 0.0741,
+      "step": 1094
+    },
+    {
+      "epoch": 4.441518987341772,
+      "grad_norm": 0.23137785717304035,
+      "learning_rate": 1.4499509779581078e-06,
+      "loss": 0.0547,
+      "step": 1095
+    },
+    {
+      "epoch": 4.445569620253164,
+      "grad_norm": 0.2353640091033897,
+      "learning_rate": 1.4288083445723988e-06,
+      "loss": 0.0741,
+      "step": 1096
+    },
+    {
+      "epoch": 4.449620253164557,
+      "grad_norm": 0.23557999612961292,
+      "learning_rate": 1.4078152810112045e-06,
+      "loss": 0.0763,
+      "step": 1097
+    },
+    {
+      "epoch": 4.453670886075949,
+      "grad_norm": 0.22606862247423296,
+      "learning_rate": 1.3869719563497697e-06,
+      "loss": 0.0717,
+      "step": 1098
+    },
+    {
+      "epoch": 4.457721518987341,
+      "grad_norm": 0.22887032682805075,
+      "learning_rate": 1.3662785384573663e-06,
+      "loss": 0.0653,
+      "step": 1099
+    },
+    {
+      "epoch": 4.461772151898734,
+      "grad_norm": 0.21480864629840124,
+      "learning_rate": 1.3457351939959383e-06,
+      "loss": 0.0704,
+      "step": 1100
+    },
+    {
+      "epoch": 4.465822784810126,
+      "grad_norm": 0.23443378508953658,
+      "learning_rate": 1.3253420884187551e-06,
+      "loss": 0.0839,
+      "step": 1101
+    },
+    {
+      "epoch": 4.469873417721519,
+      "grad_norm": 0.22796502161479426,
+      "learning_rate": 1.3050993859690953e-06,
+      "loss": 0.0738,
+      "step": 1102
+    },
+    {
+      "epoch": 4.473924050632911,
+      "grad_norm": 0.2267468775199094,
+      "learning_rate": 1.2850072496788869e-06,
+      "loss": 0.0807,
+      "step": 1103
+    },
+    {
+      "epoch": 4.4779746835443035,
+      "grad_norm": 0.23415203728162196,
+      "learning_rate": 1.2650658413674434e-06,
+      "loss": 0.0863,
+      "step": 1104
+    },
+    {
+      "epoch": 4.4820253164556965,
+      "grad_norm": 0.23853169480111538,
+      "learning_rate": 1.2452753216401226e-06,
+      "loss": 0.0579,
+      "step": 1105
+    },
+    {
+      "epoch": 4.4860759493670885,
+      "grad_norm": 0.23116016944035533,
+      "learning_rate": 1.2256358498870503e-06,
+      "loss": 0.0736,
+      "step": 1106
+    },
+    {
+      "epoch": 4.490126582278481,
+      "grad_norm": 0.23024372318914815,
+      "learning_rate": 1.2061475842818337e-06,
+      "loss": 0.0681,
+      "step": 1107
+    },
+    {
+      "epoch": 4.494177215189874,
+      "grad_norm": 0.23530675904190823,
+      "learning_rate": 1.1868106817802816e-06,
+      "loss": 0.0658,
+      "step": 1108
+    },
+    {
+      "epoch": 4.498227848101266,
+      "grad_norm": 0.2207402192358052,
+      "learning_rate": 1.1676252981191482e-06,
+      "loss": 0.0708,
+      "step": 1109
+    },
+    {
+      "epoch": 4.502278481012659,
+      "grad_norm": 0.22702492007732486,
+      "learning_rate": 1.1485915878148823e-06,
+      "loss": 0.0793,
+      "step": 1110
+    },
+    {
+      "epoch": 4.506329113924051,
+      "grad_norm": 0.22405085204861686,
+      "learning_rate": 1.1297097041623584e-06,
+      "loss": 0.0783,
+      "step": 1111
+    },
+    {
+      "epoch": 4.510379746835443,
+      "grad_norm": 0.22688763493542635,
+      "learning_rate": 1.1109797992336847e-06,
+      "loss": 0.0773,
+      "step": 1112
+    },
+    {
+      "epoch": 4.514430379746836,
+      "grad_norm": 0.2273869394063846,
+      "learning_rate": 1.092402023876933e-06,
+      "loss": 0.0778,
+      "step": 1113
+    },
+    {
+      "epoch": 4.518481012658228,
+      "grad_norm": 0.2305389721018135,
+      "learning_rate": 1.0739765277149527e-06,
+      "loss": 0.0654,
+      "step": 1114
+    },
+    {
+      "epoch": 4.52253164556962,
+      "grad_norm": 0.21714336802096523,
+      "learning_rate": 1.0557034591441596e-06,
+      "loss": 0.0672,
+      "step": 1115
+    },
+    {
+      "epoch": 4.526582278481013,
+      "grad_norm": 0.23179380503611682,
+      "learning_rate": 1.0375829653333324e-06,
+      "loss": 0.0818,
+      "step": 1116
+    },
+    {
+      "epoch": 4.530632911392405,
+      "grad_norm": 0.2296342806877945,
+      "learning_rate": 1.0196151922224385e-06,
+      "loss": 0.0741,
+      "step": 1117
+    },
+    {
+      "epoch": 4.534683544303798,
+      "grad_norm": 0.2372606269794857,
+      "learning_rate": 1.0018002845214526e-06,
+      "loss": 0.0778,
+      "step": 1118
+    },
+    {
+      "epoch": 4.53873417721519,
+      "grad_norm": 0.22287691784486063,
+      "learning_rate": 9.841383857091947e-07,
+      "loss": 0.0746,
+      "step": 1119
+    },
+    {
+      "epoch": 4.542784810126582,
+      "grad_norm": 0.2355185518672048,
+      "learning_rate": 9.666296380321616e-07,
+      "loss": 0.0717,
+      "step": 1120
+    },
+    {
+      "epoch": 4.546835443037975,
+      "grad_norm": 0.23450179782376493,
+      "learning_rate": 9.492741825034124e-07,
+      "loss": 0.0766,
+      "step": 1121
+    },
+    {
+      "epoch": 4.550886075949367,
+      "grad_norm": 0.2268902697710501,
+      "learning_rate": 9.320721589013892e-07,
+      "loss": 0.0633,
+      "step": 1122
+    },
+    {
+      "epoch": 4.55493670886076,
+      "grad_norm": 0.22209112100149853,
+      "learning_rate": 9.150237057688339e-07,
+      "loss": 0.0785,
+      "step": 1123
+    },
+    {
+      "epoch": 4.558987341772152,
+      "grad_norm": 0.2231733247823616,
+      "learning_rate": 8.981289604116328e-07,
+      "loss": 0.0717,
+      "step": 1124
+    },
+    {
+      "epoch": 4.563037974683544,
+      "grad_norm": 0.22807906181384469,
+      "learning_rate": 8.813880588977542e-07,
+      "loss": 0.0843,
+      "step": 1125
+    },
+    {
+      "epoch": 4.567088607594937,
+      "grad_norm": 0.24403878971254678,
+      "learning_rate": 8.648011360561126e-07,
+      "loss": 0.0855,
+      "step": 1126
+    },
+    {
+      "epoch": 4.571139240506329,
+      "grad_norm": 0.2203291220052404,
+      "learning_rate": 8.483683254755037e-07,
+      "loss": 0.0763,
+      "step": 1127
+    },
+    {
+      "epoch": 4.575189873417721,
+      "grad_norm": 0.23556731992226115,
+      "learning_rate": 8.320897595035227e-07,
+      "loss": 0.0729,
+      "step": 1128
+    },
+    {
+      "epoch": 4.579240506329114,
+      "grad_norm": 0.22221544302797402,
+      "learning_rate": 8.159655692455093e-07,
+      "loss": 0.0728,
+      "step": 1129
+    },
+    {
+      "epoch": 4.583291139240506,
+      "grad_norm": 0.23281037482976163,
+      "learning_rate": 7.999958845634648e-07,
+      "loss": 0.066,
+      "step": 1130
+    },
+    {
+      "epoch": 4.587341772151898,
+      "grad_norm": 0.23258666787468232,
+      "learning_rate": 7.841808340750478e-07,
+      "loss": 0.0825,
+      "step": 1131
+    },
+    {
+      "epoch": 4.591392405063291,
+      "grad_norm": 0.22547374299906256,
+      "learning_rate": 7.685205451524869e-07,
+      "loss": 0.0671,
+      "step": 1132
+    },
+    {
+      "epoch": 4.595443037974683,
+      "grad_norm": 0.22839664815267097,
+      "learning_rate": 7.530151439216027e-07,
+      "loss": 0.0706,
+      "step": 1133
+    },
+    {
+      "epoch": 4.599493670886076,
+      "grad_norm": 0.22512527359542828,
+      "learning_rate": 7.376647552607675e-07,
+      "loss": 0.0754,
+      "step": 1134
+    },
+    {
+      "epoch": 4.603544303797468,
+      "grad_norm": 0.22994699400965846,
+      "learning_rate": 7.224695027998963e-07,
+      "loss": 0.0678,
+      "step": 1135
+    },
+    {
+      "epoch": 4.6075949367088604,
+      "grad_norm": 0.22083079489475116,
+      "learning_rate": 7.07429508919466e-07,
+      "loss": 0.0751,
+      "step": 1136
+    },
+    {
+      "epoch": 4.611645569620253,
+      "grad_norm": 0.22059002073074965,
+      "learning_rate": 6.925448947495206e-07,
+      "loss": 0.0744,
+      "step": 1137
+    },
+    {
+      "epoch": 4.6156962025316455,
+      "grad_norm": 0.21616028536609982,
+      "learning_rate": 6.778157801686936e-07,
+      "loss": 0.0619,
+      "step": 1138
+    },
+    {
+      "epoch": 4.619746835443038,
+      "grad_norm": 0.23240181228112583,
+      "learning_rate": 6.632422838032515e-07,
+      "loss": 0.0802,
+      "step": 1139
+    },
+    {
+      "epoch": 4.6237974683544305,
+      "grad_norm": 0.22996199375271537,
+      "learning_rate": 6.488245230261281e-07,
+      "loss": 0.0749,
+      "step": 1140
+    },
+    {
+      "epoch": 4.627848101265823,
+      "grad_norm": 0.24308052928641347,
+      "learning_rate": 6.345626139559868e-07,
+      "loss": 0.0689,
+      "step": 1141
+    },
+    {
+      "epoch": 4.6318987341772155,
+      "grad_norm": 0.23351843857388113,
+      "learning_rate": 6.204566714562866e-07,
+      "loss": 0.0746,
+      "step": 1142
+    },
+    {
+      "epoch": 4.635949367088608,
+      "grad_norm": 0.22567868321949774,
+      "learning_rate": 6.06506809134344e-07,
+      "loss": 0.0713,
+      "step": 1143
+    },
+    {
+      "epoch": 4.64,
+      "grad_norm": 0.2178548605735714,
+      "learning_rate": 5.927131393404373e-07,
+      "loss": 0.0885,
+      "step": 1144
+    },
+    {
+      "epoch": 4.644050632911393,
+      "grad_norm": 0.2193869854140966,
+      "learning_rate": 5.790757731668817e-07,
+      "loss": 0.0669,
+      "step": 1145
+    },
+    {
+      "epoch": 4.648101265822785,
+      "grad_norm": 0.22267090735241174,
+      "learning_rate": 5.655948204471507e-07,
+      "loss": 0.0662,
+      "step": 1146
+    },
+    {
+      "epoch": 4.652151898734177,
+      "grad_norm": 0.22612503477181822,
+      "learning_rate": 5.522703897549875e-07,
+      "loss": 0.0678,
+      "step": 1147
+    },
+    {
+      "epoch": 4.65620253164557,
+      "grad_norm": 0.2241256966810181,
+      "learning_rate": 5.391025884035239e-07,
+      "loss": 0.0657,
+      "step": 1148
+    },
+    {
+      "epoch": 4.660253164556962,
+      "grad_norm": 0.22501688118222446,
+      "learning_rate": 5.260915224444207e-07,
+      "loss": 0.0824,
+      "step": 1149
+    },
+    {
+      "epoch": 4.664303797468355,
+      "grad_norm": 0.23477557436030896,
+      "learning_rate": 5.132372966670129e-07,
+      "loss": 0.0747,
+      "step": 1150
+    },
+    {
+      "epoch": 4.668354430379747,
+      "grad_norm": 0.2271761318728923,
+      "learning_rate": 5.005400145974704e-07,
+      "loss": 0.0666,
+      "step": 1151
+    },
+    {
+      "epoch": 4.672405063291139,
+      "grad_norm": 0.23027689886524813,
+      "learning_rate": 4.879997784979562e-07,
+      "loss": 0.0659,
+      "step": 1152
+    },
+    {
+      "epoch": 4.676455696202532,
+      "grad_norm": 0.23179603794057016,
+      "learning_rate": 4.7561668936580984e-07,
+      "loss": 0.0757,
+      "step": 1153
+    },
+    {
+      "epoch": 4.680506329113924,
+      "grad_norm": 0.22845673362229962,
+      "learning_rate": 4.6339084693272306e-07,
+      "loss": 0.0791,
+      "step": 1154
+    },
+    {
+      "epoch": 4.684556962025317,
+      "grad_norm": 0.22725422280822674,
+      "learning_rate": 4.5132234966395847e-07,
+      "loss": 0.0746,
+      "step": 1155
+    },
+    {
+      "epoch": 4.688607594936709,
+      "grad_norm": 0.22825986998454606,
+      "learning_rate": 4.3941129475752795e-07,
+      "loss": 0.0676,
+      "step": 1156
+    },
+    {
+      "epoch": 4.692658227848101,
+      "grad_norm": 0.2279612673277111,
+      "learning_rate": 4.27657778143431e-07,
+      "loss": 0.0791,
+      "step": 1157
+    },
+    {
+      "epoch": 4.696708860759494,
+      "grad_norm": 0.22507238134599716,
+      "learning_rate": 4.1606189448287757e-07,
+      "loss": 0.0686,
+      "step": 1158
+    },
+    {
+      "epoch": 4.700759493670886,
+      "grad_norm": 0.22530257280628646,
+      "learning_rate": 4.046237371675177e-07,
+      "loss": 0.0732,
+      "step": 1159
+    },
+    {
+      "epoch": 4.704810126582278,
+      "grad_norm": 0.23246465821221476,
+      "learning_rate": 3.9334339831869963e-07,
+      "loss": 0.0713,
+      "step": 1160
+    },
+    {
+      "epoch": 4.708860759493671,
+      "grad_norm": 0.23239851607287815,
+      "learning_rate": 3.8222096878671955e-07,
+      "loss": 0.0705,
+      "step": 1161
+    },
+    {
+      "epoch": 4.712911392405063,
+      "grad_norm": 0.2196876020083495,
+      "learning_rate": 3.7125653815009545e-07,
+      "loss": 0.0717,
+      "step": 1162
+    },
+    {
+      "epoch": 4.716962025316455,
+      "grad_norm": 0.23166609082610043,
+      "learning_rate": 3.6045019471484974e-07,
+      "loss": 0.0759,
+      "step": 1163
+    },
+    {
+      "epoch": 4.721012658227848,
+      "grad_norm": 0.2300708283164102,
+      "learning_rate": 3.498020255137813e-07,
+      "loss": 0.0758,
+      "step": 1164
+    },
+    {
+      "epoch": 4.72506329113924,
+      "grad_norm": 0.22925007925494906,
+      "learning_rate": 3.393121163057811e-07,
+      "loss": 0.0651,
+      "step": 1165
+    },
+    {
+      "epoch": 4.729113924050633,
+      "grad_norm": 0.2231293859192535,
+      "learning_rate": 3.289805515751399e-07,
+      "loss": 0.0576,
+      "step": 1166
+    },
+    {
+      "epoch": 4.733164556962025,
+      "grad_norm": 0.2324309928855117,
+      "learning_rate": 3.188074145308573e-07,
+      "loss": 0.0832,
+      "step": 1167
+    },
+    {
+      "epoch": 4.737215189873417,
+      "grad_norm": 0.22221562159211014,
+      "learning_rate": 3.087927871059804e-07,
+      "loss": 0.0693,
+      "step": 1168
+    },
+    {
+      "epoch": 4.74126582278481,
+      "grad_norm": 0.2317598768575415,
+      "learning_rate": 2.989367499569418e-07,
+      "loss": 0.0739,
+      "step": 1169
+    },
+    {
+      "epoch": 4.745316455696202,
+      "grad_norm": 0.22884525184193447,
+      "learning_rate": 2.8923938246290917e-07,
+      "loss": 0.0663,
+      "step": 1170
+    },
+    {
+      "epoch": 4.749367088607595,
+      "grad_norm": 0.23914424219601724,
+      "learning_rate": 2.7970076272514804e-07,
+      "loss": 0.0784,
+      "step": 1171
+    },
+    {
+      "epoch": 4.7534177215189874,
+      "grad_norm": 0.22229253903979035,
+      "learning_rate": 2.703209675663887e-07,
+      "loss": 0.0602,
+      "step": 1172
+    },
+    {
+      "epoch": 4.7574683544303795,
+      "grad_norm": 0.23016589345099833,
+      "learning_rate": 2.6110007253021374e-07,
+      "loss": 0.0614,
+      "step": 1173
+    },
+    {
+      "epoch": 4.7615189873417725,
+      "grad_norm": 0.23217796868260332,
+      "learning_rate": 2.520381518804471e-07,
+      "loss": 0.0589,
+      "step": 1174
+    },
+    {
+      "epoch": 4.7655696202531646,
+      "grad_norm": 0.23381517465888357,
+      "learning_rate": 2.4313527860054585e-07,
+      "loss": 0.0804,
+      "step": 1175
+    },
+    {
+      "epoch": 4.769620253164557,
+      "grad_norm": 0.22566505415789925,
+      "learning_rate": 2.343915243930317e-07,
+      "loss": 0.0736,
+      "step": 1176
+    },
+    {
+      "epoch": 4.77367088607595,
+      "grad_norm": 0.2290133383333956,
+      "learning_rate": 2.2580695967889367e-07,
+      "loss": 0.0782,
+      "step": 1177
+    },
+    {
+      "epoch": 4.777721518987342,
+      "grad_norm": 0.22792564854655945,
+      "learning_rate": 2.1738165359704189e-07,
+      "loss": 0.0902,
+      "step": 1178
+    },
+    {
+      "epoch": 4.781772151898734,
+      "grad_norm": 0.22549742829191122,
+      "learning_rate": 2.0911567400373257e-07,
+      "loss": 0.0735,
+      "step": 1179
+    },
+    {
+      "epoch": 4.785822784810127,
+      "grad_norm": 0.22954937372910578,
+      "learning_rate": 2.0100908747202607e-07,
+      "loss": 0.073,
+      "step": 1180
+    },
+    {
+      "epoch": 4.789873417721519,
+      "grad_norm": 0.23319002640650782,
+      "learning_rate": 1.9306195929125638e-07,
+      "loss": 0.0631,
+      "step": 1181
+    },
+    {
+      "epoch": 4.793924050632912,
+      "grad_norm": 0.2295503304600173,
+      "learning_rate": 1.8527435346650247e-07,
+      "loss": 0.0679,
+      "step": 1182
+    },
+    {
+      "epoch": 4.797974683544304,
+      "grad_norm": 0.22457360918221378,
+      "learning_rate": 1.7764633271807108e-07,
+      "loss": 0.0602,
+      "step": 1183
+    },
+    {
+      "epoch": 4.802025316455696,
+      "grad_norm": 0.2303413038884243,
+      "learning_rate": 1.7017795848099262e-07,
+      "loss": 0.0777,
+      "step": 1184
+    },
+    {
+      "epoch": 4.806075949367089,
+      "grad_norm": 0.23254396838617797,
+      "learning_rate": 1.6286929090452596e-07,
+      "loss": 0.0783,
+      "step": 1185
+    },
+    {
+      "epoch": 4.810126582278481,
+      "grad_norm": 0.2285469389607588,
+      "learning_rate": 1.557203888516745e-07,
+      "loss": 0.0659,
+      "step": 1186
+    },
+    {
+      "epoch": 4.814177215189874,
+      "grad_norm": 0.23100175456122982,
+      "learning_rate": 1.487313098987131e-07,
+      "loss": 0.0667,
+      "step": 1187
+    },
+    {
+      "epoch": 4.818227848101266,
+      "grad_norm": 0.23885995984834033,
+      "learning_rate": 1.4190211033472402e-07,
+      "loss": 0.0796,
+      "step": 1188
+    },
+    {
+      "epoch": 4.822278481012658,
+      "grad_norm": 0.22460682782827573,
+      "learning_rate": 1.3523284516113955e-07,
+      "loss": 0.0755,
+      "step": 1189
+    },
+    {
+      "epoch": 4.826329113924051,
+      "grad_norm": 0.22998749484055897,
+      "learning_rate": 1.2872356809130682e-07,
+      "loss": 0.0695,
+      "step": 1190
+    },
+    {
+      "epoch": 4.830379746835443,
+      "grad_norm": 0.23347544903178227,
+      "learning_rate": 1.2237433155004807e-07,
+      "loss": 0.0762,
+      "step": 1191
+    },
+    {
+      "epoch": 4.834430379746835,
+      "grad_norm": 0.22621230413709884,
+      "learning_rate": 1.1618518667323886e-07,
+      "loss": 0.0622,
+      "step": 1192
+    },
+    {
+      "epoch": 4.838481012658228,
+      "grad_norm": 0.22744997045896975,
+      "learning_rate": 1.1015618330740385e-07,
+      "loss": 0.0855,
+      "step": 1193
+    },
+    {
+      "epoch": 4.84253164556962,
+      "grad_norm": 0.219789973273217,
+      "learning_rate": 1.042873700093061e-07,
+      "loss": 0.0684,
+      "step": 1194
+    },
+    {
+      "epoch": 4.846582278481012,
+      "grad_norm": 0.23465149875841637,
+      "learning_rate": 9.857879404556291e-08,
+      "loss": 0.0671,
+      "step": 1195
+    },
+    {
+      "epoch": 4.850632911392405,
+      "grad_norm": 0.22910917128702676,
+      "learning_rate": 9.303050139225722e-08,
+      "loss": 0.0704,
+      "step": 1196
+    },
+    {
+      "epoch": 4.854683544303797,
+      "grad_norm": 0.22476040628119925,
+      "learning_rate": 8.76425367345779e-08,
+      "loss": 0.0702,
+      "step": 1197
+    },
+    {
+      "epoch": 4.85873417721519,
+      "grad_norm": 0.2282836588777298,
+      "learning_rate": 8.241494346644897e-08,
+      "loss": 0.0759,
+      "step": 1198
+    },
+    {
+      "epoch": 4.862784810126582,
+      "grad_norm": 0.22725603790267304,
+      "learning_rate": 7.734776369019204e-08,
+      "loss": 0.07,
+      "step": 1199
+    },
+    {
+      "epoch": 4.866835443037974,
+      "grad_norm": 0.22151614947365447,
+      "learning_rate": 7.244103821617332e-08,
+      "loss": 0.0683,
+      "step": 1200
+    },
+    {
+      "epoch": 4.870886075949367,
+      "grad_norm": 0.23241139237740377,
+      "learning_rate": 6.769480656248606e-08,
+      "loss": 0.0698,
+      "step": 1201
+    },
+    {
+      "epoch": 4.874936708860759,
+      "grad_norm": 0.22383023800843957,
+      "learning_rate": 6.310910695462635e-08,
+      "loss": 0.0729,
+      "step": 1202
+    },
+    {
+      "epoch": 4.878987341772152,
+      "grad_norm": 0.22100448026500355,
+      "learning_rate": 5.8683976325191185e-08,
+      "loss": 0.0707,
+      "step": 1203
+    },
+    {
+      "epoch": 4.883037974683544,
+      "grad_norm": 0.23218584496867917,
+      "learning_rate": 5.4419450313571984e-08,
+      "loss": 0.0686,
+      "step": 1204
+    },
+    {
+      "epoch": 4.8870886075949365,
+      "grad_norm": 0.22852687200289729,
+      "learning_rate": 5.031556326567488e-08,
+      "loss": 0.0752,
+      "step": 1205
+    },
+    {
+      "epoch": 4.891139240506329,
+      "grad_norm": 0.22818199688905633,
+      "learning_rate": 4.637234823364312e-08,
+      "loss": 0.0798,
+      "step": 1206
+    },
+    {
+      "epoch": 4.8951898734177215,
+      "grad_norm": 0.23286676957269953,
+      "learning_rate": 4.258983697558838e-08,
+      "loss": 0.0612,
+      "step": 1207
+    },
+    {
+      "epoch": 4.899240506329114,
+      "grad_norm": 0.22176580057159812,
+      "learning_rate": 3.896805995533548e-08,
+      "loss": 0.08,
+      "step": 1208
+    },
+    {
+      "epoch": 4.9032911392405065,
+      "grad_norm": 0.22805160890427828,
+      "learning_rate": 3.550704634218028e-08,
+      "loss": 0.0813,
+      "step": 1209
+    },
+    {
+      "epoch": 4.907341772151899,
+      "grad_norm": 0.22828312489111935,
+      "learning_rate": 3.2206824010647676e-08,
+      "loss": 0.0737,
+      "step": 1210
+    },
+    {
+      "epoch": 4.911392405063291,
+      "grad_norm": 0.24302997203438922,
+      "learning_rate": 2.9067419540278476e-08,
+      "loss": 0.0761,
+      "step": 1211
+    },
+    {
+      "epoch": 4.915443037974684,
+      "grad_norm": 0.23320658201772612,
+      "learning_rate": 2.6088858215400638e-08,
+      "loss": 0.0784,
+      "step": 1212
+    },
+    {
+      "epoch": 4.919493670886076,
+      "grad_norm": 0.23291279860484987,
+      "learning_rate": 2.3271164024940564e-08,
+      "loss": 0.0676,
+      "step": 1213
+    },
+    {
+      "epoch": 4.923544303797469,
+      "grad_norm": 0.23367617427631296,
+      "learning_rate": 2.061435966221881e-08,
+      "loss": 0.079,
+      "step": 1214
+    },
+    {
+      "epoch": 4.927594936708861,
+      "grad_norm": 0.2269971071370791,
+      "learning_rate": 1.811846652477245e-08,
+      "loss": 0.0755,
+      "step": 1215
+    },
+    {
+      "epoch": 4.931645569620253,
+      "grad_norm": 0.22125308826234522,
+      "learning_rate": 1.5783504714184106e-08,
+      "loss": 0.0549,
+      "step": 1216
+    },
+    {
+      "epoch": 4.935696202531646,
+      "grad_norm": 0.22629435180681093,
+      "learning_rate": 1.360949303591097e-08,
+      "loss": 0.0723,
+      "step": 1217
+    },
+    {
+      "epoch": 4.939746835443038,
+      "grad_norm": 0.22358099737142573,
+      "learning_rate": 1.1596448999144916e-08,
+      "loss": 0.0655,
+      "step": 1218
+    },
+    {
+      "epoch": 4.943797468354431,
+      "grad_norm": 0.2281071216684399,
+      "learning_rate": 9.744388816668172e-09,
+      "loss": 0.0702,
+      "step": 1219
+    },
+    {
+      "epoch": 4.947848101265823,
+      "grad_norm": 0.23783512759316733,
+      "learning_rate": 8.05332740472009e-09,
+      "loss": 0.0691,
+      "step": 1220
+    },
+    {
+      "epoch": 4.951898734177215,
+      "grad_norm": 0.22815813126667328,
+      "learning_rate": 6.523278382872811e-09,
+      "loss": 0.0747,
+      "step": 1221
+    },
+    {
+      "epoch": 4.955949367088608,
+      "grad_norm": 0.23833122251004704,
+      "learning_rate": 5.15425407393133e-09,
+      "loss": 0.0618,
+      "step": 1222
+    },
+    {
+      "epoch": 4.96,
+      "grad_norm": 0.22941493931615423,
+      "learning_rate": 3.94626550383137e-09,
+      "loss": 0.0769,
+      "step": 1223
+    },
+    {
+      "epoch": 4.964050632911392,
+      "grad_norm": 0.22496897395762616,
+      "learning_rate": 2.899322401546112e-09,
+      "loss": 0.0776,
+      "step": 1224
+    },
+    {
+      "epoch": 4.968101265822785,
+      "grad_norm": 0.23541550171993667,
+      "learning_rate": 2.013433199010706e-09,
+      "loss": 0.0723,
+      "step": 1225
+    },
+    {
+      "epoch": 4.972151898734177,
+      "grad_norm": 0.23103511272295585,
+      "learning_rate": 1.2886050310556563e-09,
+      "loss": 0.0741,
+      "step": 1226
+    },
+    {
+      "epoch": 4.976202531645569,
+      "grad_norm": 0.22657570148104478,
+      "learning_rate": 7.248437353468695e-10,
+      "loss": 0.0821,
+      "step": 1227
+    },
+    {
+      "epoch": 4.980253164556962,
+      "grad_norm": 0.2306749727274773,
+      "learning_rate": 3.221538523412449e-10,
+      "loss": 0.0659,
+      "step": 1228
+    },
+    {
+      "epoch": 4.984303797468354,
+      "grad_norm": 0.2272518020636281,
+      "learning_rate": 8.053862524670663e-11,
+      "loss": 0.0768,
+      "step": 1229
+    },
+    {
+      "epoch": 4.988354430379747,
+      "grad_norm": 0.22637537779767702,
+      "learning_rate": 0.0,
+      "loss": 0.0713,
+      "step": 1230
+    },
+    {
+      "epoch": 4.988354430379747,
+      "step": 1230,
+      "total_flos": 1.4390160925078323e+18,
+      "train_loss": 0.21472142088582846,
+      "train_runtime": 51117.6952,
+      "train_samples_per_second": 3.091,
+      "train_steps_per_second": 0.024
+    }
+  ],
+  "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": 1.4390160925078323e+18,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}