diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,8652 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.996455696202531,
+  "eval_steps": 500,
+  "global_step": 1230,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.004050632911392405,
+      "grad_norm": 6.140918888827733,
+      "learning_rate": 3.2520325203252037e-07,
+      "loss": 1.0239,
+      "step": 1
+    },
+    {
+      "epoch": 0.00810126582278481,
+      "grad_norm": 6.246667155317643,
+      "learning_rate": 6.504065040650407e-07,
+      "loss": 1.0381,
+      "step": 2
+    },
+    {
+      "epoch": 0.012151898734177215,
+      "grad_norm": 6.243309154792987,
+      "learning_rate": 9.75609756097561e-07,
+      "loss": 1.042,
+      "step": 3
+    },
+    {
+      "epoch": 0.01620253164556962,
+      "grad_norm": 6.123895612745012,
+      "learning_rate": 1.3008130081300815e-06,
+      "loss": 1.043,
+      "step": 4
+    },
+    {
+      "epoch": 0.020253164556962026,
+      "grad_norm": 5.770273230127823,
+      "learning_rate": 1.6260162601626018e-06,
+      "loss": 1.0271,
+      "step": 5
+    },
+    {
+      "epoch": 0.02430379746835443,
+      "grad_norm": 5.6798724992317124,
+      "learning_rate": 1.951219512195122e-06,
+      "loss": 1.0336,
+      "step": 6
+    },
+    {
+      "epoch": 0.028354430379746835,
+      "grad_norm": 4.362217402422049,
+      "learning_rate": 2.2764227642276426e-06,
+      "loss": 0.9918,
+      "step": 7
+    },
+    {
+      "epoch": 0.03240506329113924,
+      "grad_norm": 4.030398868690057,
+      "learning_rate": 2.601626016260163e-06,
+      "loss": 0.9884,
+      "step": 8
+    },
+    {
+      "epoch": 0.03645569620253165,
+      "grad_norm": 2.6931862545071157,
+      "learning_rate": 2.926829268292683e-06,
+      "loss": 0.9627,
+      "step": 9
+    },
+    {
+      "epoch": 0.04050632911392405,
+      "grad_norm": 2.42962354700195,
+      "learning_rate": 3.2520325203252037e-06,
+      "loss": 0.9501,
+      "step": 10
+    },
+    {
+      "epoch": 0.044556962025316456,
+      "grad_norm": 2.4412268463590325,
+      "learning_rate": 3.577235772357724e-06,
+      "loss": 0.9476,
+      "step": 11
+    },
+    {
+      "epoch": 0.04860759493670886,
+      "grad_norm": 3.936472574815198,
+      "learning_rate": 3.902439024390244e-06,
+      "loss": 0.9518,
+      "step": 12
+    },
+    {
+      "epoch": 0.052658227848101265,
+      "grad_norm": 4.1665564001820705,
+      "learning_rate": 4.227642276422765e-06,
+      "loss": 0.9341,
+      "step": 13
+    },
+    {
+      "epoch": 0.05670886075949367,
+      "grad_norm": 4.038368062588068,
+      "learning_rate": 4.552845528455285e-06,
+      "loss": 0.932,
+      "step": 14
+    },
+    {
+      "epoch": 0.060759493670886074,
+      "grad_norm": 3.587972954540083,
+      "learning_rate": 4.8780487804878055e-06,
+      "loss": 0.927,
+      "step": 15
+    },
+    {
+      "epoch": 0.06481012658227848,
+      "grad_norm": 2.724703858582525,
+      "learning_rate": 5.203252032520326e-06,
+      "loss": 0.8782,
+      "step": 16
+    },
+    {
+      "epoch": 0.06886075949367089,
+      "grad_norm": 2.453068806118883,
+      "learning_rate": 5.528455284552846e-06,
+      "loss": 0.8663,
+      "step": 17
+    },
+    {
+      "epoch": 0.0729113924050633,
+      "grad_norm": 2.045273397507766,
+      "learning_rate": 5.853658536585366e-06,
+      "loss": 0.8625,
+      "step": 18
+    },
+    {
+      "epoch": 0.0769620253164557,
+      "grad_norm": 1.4057133120609766,
+      "learning_rate": 6.178861788617887e-06,
+      "loss": 0.8437,
+      "step": 19
+    },
+    {
+      "epoch": 0.0810126582278481,
+      "grad_norm": 1.0413245156379682,
+      "learning_rate": 6.504065040650407e-06,
+      "loss": 0.8448,
+      "step": 20
+    },
+    {
+      "epoch": 0.08506329113924051,
+      "grad_norm": 0.939433695501421,
+      "learning_rate": 6.829268292682928e-06,
+      "loss": 0.8339,
+      "step": 21
+    },
+    {
+      "epoch": 0.08911392405063291,
+      "grad_norm": 1.2328502874656528,
+      "learning_rate": 7.154471544715448e-06,
+      "loss": 0.7979,
+      "step": 22
+    },
+    {
+      "epoch": 0.09316455696202532,
+      "grad_norm": 1.045293963325136,
+      "learning_rate": 7.4796747967479676e-06,
+      "loss": 0.808,
+      "step": 23
+    },
+    {
+      "epoch": 0.09721518987341772,
+      "grad_norm": 0.8243475932463831,
+      "learning_rate": 7.804878048780489e-06,
+      "loss": 0.8003,
+      "step": 24
+    },
+    {
+      "epoch": 0.10126582278481013,
+      "grad_norm": 0.875261024544733,
+      "learning_rate": 8.130081300813009e-06,
+      "loss": 0.791,
+      "step": 25
+    },
+    {
+      "epoch": 0.10531645569620253,
+      "grad_norm": 0.7997303068980246,
+      "learning_rate": 8.45528455284553e-06,
+      "loss": 0.7921,
+      "step": 26
+    },
+    {
+      "epoch": 0.10936708860759493,
+      "grad_norm": 0.7896851960035628,
+      "learning_rate": 8.78048780487805e-06,
+      "loss": 0.7833,
+      "step": 27
+    },
+    {
+      "epoch": 0.11341772151898734,
+      "grad_norm": 0.8264968348771252,
+      "learning_rate": 9.10569105691057e-06,
+      "loss": 0.7796,
+      "step": 28
+    },
+    {
+      "epoch": 0.11746835443037974,
+      "grad_norm": 0.8208609147120469,
+      "learning_rate": 9.43089430894309e-06,
+      "loss": 0.7582,
+      "step": 29
+    },
+    {
+      "epoch": 0.12151898734177215,
+      "grad_norm": 0.7068416168582553,
+      "learning_rate": 9.756097560975611e-06,
+      "loss": 0.7623,
+      "step": 30
+    },
+    {
+      "epoch": 0.12556962025316457,
+      "grad_norm": 0.7035956596103234,
+      "learning_rate": 1.008130081300813e-05,
+      "loss": 0.7489,
+      "step": 31
+    },
+    {
+      "epoch": 0.12962025316455697,
+      "grad_norm": 0.7166061766778842,
+      "learning_rate": 1.0406504065040652e-05,
+      "loss": 0.7512,
+      "step": 32
+    },
+    {
+      "epoch": 0.13367088607594937,
+      "grad_norm": 0.6872423361447135,
+      "learning_rate": 1.0731707317073172e-05,
+      "loss": 0.7439,
+      "step": 33
+    },
+    {
+      "epoch": 0.13772151898734178,
+      "grad_norm": 0.6887572008161434,
+      "learning_rate": 1.1056910569105692e-05,
+      "loss": 0.7342,
+      "step": 34
+    },
+    {
+      "epoch": 0.14177215189873418,
+      "grad_norm": 0.5858282297195299,
+      "learning_rate": 1.1382113821138213e-05,
+      "loss": 0.7427,
+      "step": 35
+    },
+    {
+      "epoch": 0.1458227848101266,
+      "grad_norm": 0.6953514165852698,
+      "learning_rate": 1.1707317073170731e-05,
+      "loss": 0.749,
+      "step": 36
+    },
+    {
+      "epoch": 0.149873417721519,
+      "grad_norm": 0.5108170802468917,
+      "learning_rate": 1.2032520325203254e-05,
+      "loss": 0.7557,
+      "step": 37
+    },
+    {
+      "epoch": 0.1539240506329114,
+      "grad_norm": 0.5727417461362477,
+      "learning_rate": 1.2357723577235774e-05,
+      "loss": 0.7297,
+      "step": 38
+    },
+    {
+      "epoch": 0.1579746835443038,
+      "grad_norm": 0.540738890013823,
+      "learning_rate": 1.2682926829268294e-05,
+      "loss": 0.7291,
+      "step": 39
+    },
+    {
+      "epoch": 0.1620253164556962,
+      "grad_norm": 0.5406565078339417,
+      "learning_rate": 1.3008130081300815e-05,
+      "loss": 0.7236,
+      "step": 40
+    },
+    {
+      "epoch": 0.1660759493670886,
+      "grad_norm": 0.4774281436155498,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.719,
+      "step": 41
+    },
+    {
+      "epoch": 0.17012658227848101,
+      "grad_norm": 0.49121289044390753,
+      "learning_rate": 1.3658536585365855e-05,
+      "loss": 0.7122,
+      "step": 42
+    },
+    {
+      "epoch": 0.17417721518987342,
+      "grad_norm": 0.4826008139366319,
+      "learning_rate": 1.3983739837398376e-05,
+      "loss": 0.7193,
+      "step": 43
+    },
+    {
+      "epoch": 0.17822784810126582,
+      "grad_norm": 0.48068512944206326,
+      "learning_rate": 1.4308943089430896e-05,
+      "loss": 0.7233,
+      "step": 44
+    },
+    {
+      "epoch": 0.18227848101265823,
+      "grad_norm": 0.450053865759097,
+      "learning_rate": 1.4634146341463415e-05,
+      "loss": 0.6882,
+      "step": 45
+    },
+    {
+      "epoch": 0.18632911392405063,
+      "grad_norm": 0.5017753888814318,
+      "learning_rate": 1.4959349593495935e-05,
+      "loss": 0.7144,
+      "step": 46
+    },
+    {
+      "epoch": 0.19037974683544304,
+      "grad_norm": 0.3958241199989004,
+      "learning_rate": 1.528455284552846e-05,
+      "loss": 0.7028,
+      "step": 47
+    },
+    {
+      "epoch": 0.19443037974683544,
+      "grad_norm": 0.45931432394175414,
+      "learning_rate": 1.5609756097560978e-05,
+      "loss": 0.7012,
+      "step": 48
+    },
+    {
+      "epoch": 0.19848101265822785,
+      "grad_norm": 0.5128046546290616,
+      "learning_rate": 1.5934959349593496e-05,
+      "loss": 0.7024,
+      "step": 49
+    },
+    {
+      "epoch": 0.20253164556962025,
+      "grad_norm": 0.4887738220344158,
+      "learning_rate": 1.6260162601626018e-05,
+      "loss": 0.7055,
+      "step": 50
+    },
+    {
+      "epoch": 0.20658227848101265,
+      "grad_norm": 0.4459364824766297,
+      "learning_rate": 1.6585365853658537e-05,
+      "loss": 0.6952,
+      "step": 51
+    },
+    {
+      "epoch": 0.21063291139240506,
+      "grad_norm": 0.41954813487801257,
+      "learning_rate": 1.691056910569106e-05,
+      "loss": 0.6875,
+      "step": 52
+    },
+    {
+      "epoch": 0.21468354430379746,
+      "grad_norm": 0.5502141137215872,
+      "learning_rate": 1.7235772357723578e-05,
+      "loss": 0.6864,
+      "step": 53
+    },
+    {
+      "epoch": 0.21873417721518987,
+      "grad_norm": 0.426768065079474,
+      "learning_rate": 1.75609756097561e-05,
+      "loss": 0.7028,
+      "step": 54
+    },
+    {
+      "epoch": 0.22278481012658227,
+      "grad_norm": 0.5655446353139504,
+      "learning_rate": 1.788617886178862e-05,
+      "loss": 0.7091,
+      "step": 55
+    },
+    {
+      "epoch": 0.22683544303797468,
+      "grad_norm": 0.5630735873568579,
+      "learning_rate": 1.821138211382114e-05,
+      "loss": 0.697,
+      "step": 56
+    },
+    {
+      "epoch": 0.23088607594936708,
+      "grad_norm": 0.5421010994439103,
+      "learning_rate": 1.8536585365853663e-05,
+      "loss": 0.6743,
+      "step": 57
+    },
+    {
+      "epoch": 0.23493670886075949,
+      "grad_norm": 0.5683479779611681,
+      "learning_rate": 1.886178861788618e-05,
+      "loss": 0.6893,
+      "step": 58
+    },
+    {
+      "epoch": 0.2389873417721519,
+      "grad_norm": 0.7180352621962132,
+      "learning_rate": 1.91869918699187e-05,
+      "loss": 0.696,
+      "step": 59
+    },
+    {
+      "epoch": 0.2430379746835443,
+      "grad_norm": 0.8585019502208456,
+      "learning_rate": 1.9512195121951222e-05,
+      "loss": 0.699,
+      "step": 60
+    },
+    {
+      "epoch": 0.2470886075949367,
+      "grad_norm": 0.9762152433193898,
+      "learning_rate": 1.983739837398374e-05,
+      "loss": 0.6883,
+      "step": 61
+    },
+    {
+      "epoch": 0.25113924050632913,
+      "grad_norm": 0.9227695045359097,
+      "learning_rate": 2.016260162601626e-05,
+      "loss": 0.6806,
+      "step": 62
+    },
+    {
+      "epoch": 0.25518987341772154,
+      "grad_norm": 0.8002249280939675,
+      "learning_rate": 2.048780487804878e-05,
+      "loss": 0.6837,
+      "step": 63
+    },
+    {
+      "epoch": 0.25924050632911394,
+      "grad_norm": 0.7954777996581135,
+      "learning_rate": 2.0813008130081303e-05,
+      "loss": 0.6858,
+      "step": 64
+    },
+    {
+      "epoch": 0.26329113924050634,
+      "grad_norm": 0.5570037169831571,
+      "learning_rate": 2.1138211382113822e-05,
+      "loss": 0.6892,
+      "step": 65
+    },
+    {
+      "epoch": 0.26734177215189875,
+      "grad_norm": 0.8400063226559213,
+      "learning_rate": 2.1463414634146344e-05,
+      "loss": 0.6771,
+      "step": 66
+    },
+    {
+      "epoch": 0.27139240506329115,
+      "grad_norm": 0.8854983577003646,
+      "learning_rate": 2.1788617886178863e-05,
+      "loss": 0.6845,
+      "step": 67
+    },
+    {
+      "epoch": 0.27544303797468356,
+      "grad_norm": 0.5516763695259643,
+      "learning_rate": 2.2113821138211385e-05,
+      "loss": 0.6813,
+      "step": 68
+    },
+    {
+      "epoch": 0.27949367088607596,
+      "grad_norm": 0.6836420248304126,
+      "learning_rate": 2.2439024390243907e-05,
+      "loss": 0.6611,
+      "step": 69
+    },
+    {
+      "epoch": 0.28354430379746837,
+      "grad_norm": 0.762033090965938,
+      "learning_rate": 2.2764227642276426e-05,
+      "loss": 0.6797,
+      "step": 70
+    },
+    {
+      "epoch": 0.28759493670886077,
+      "grad_norm": 0.747595402885718,
+      "learning_rate": 2.3089430894308948e-05,
+      "loss": 0.6777,
+      "step": 71
+    },
+    {
+      "epoch": 0.2916455696202532,
+      "grad_norm": 0.6290995439542029,
+      "learning_rate": 2.3414634146341463e-05,
+      "loss": 0.6688,
+      "step": 72
+    },
+    {
+      "epoch": 0.2956962025316456,
+      "grad_norm": 0.851720138673897,
+      "learning_rate": 2.3739837398373985e-05,
+      "loss": 0.6544,
+      "step": 73
+    },
+    {
+      "epoch": 0.299746835443038,
+      "grad_norm": 1.0947131985530276,
+      "learning_rate": 2.4065040650406507e-05,
+      "loss": 0.659,
+      "step": 74
+    },
+    {
+      "epoch": 0.3037974683544304,
+      "grad_norm": 1.0809702442566147,
+      "learning_rate": 2.4390243902439026e-05,
+      "loss": 0.6627,
+      "step": 75
+    },
+    {
+      "epoch": 0.3078481012658228,
+      "grad_norm": 0.881564044752804,
+      "learning_rate": 2.4715447154471548e-05,
+      "loss": 0.6794,
+      "step": 76
+    },
+    {
+      "epoch": 0.3118987341772152,
+      "grad_norm": 0.6911002507006814,
+      "learning_rate": 2.5040650406504066e-05,
+      "loss": 0.6768,
+      "step": 77
+    },
+    {
+      "epoch": 0.3159493670886076,
+      "grad_norm": 0.5929023187459714,
+      "learning_rate": 2.536585365853659e-05,
+      "loss": 0.687,
+      "step": 78
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.651320002819318,
+      "learning_rate": 2.569105691056911e-05,
+      "loss": 0.6675,
+      "step": 79
+    },
+    {
+      "epoch": 0.3240506329113924,
+      "grad_norm": 1.0598743847957752,
+      "learning_rate": 2.601626016260163e-05,
+      "loss": 0.6789,
+      "step": 80
+    },
+    {
+      "epoch": 0.3281012658227848,
+      "grad_norm": 1.6761911939339438,
+      "learning_rate": 2.634146341463415e-05,
+      "loss": 0.6722,
+      "step": 81
+    },
+    {
+      "epoch": 0.3321518987341772,
+      "grad_norm": 0.5830047664941159,
+      "learning_rate": 2.6666666666666667e-05,
+      "loss": 0.6535,
+      "step": 82
+    },
+    {
+      "epoch": 0.3362025316455696,
+      "grad_norm": 1.5408272250786492,
+      "learning_rate": 2.699186991869919e-05,
+      "loss": 0.6775,
+      "step": 83
+    },
+    {
+      "epoch": 0.34025316455696203,
+      "grad_norm": 1.042303717105932,
+      "learning_rate": 2.731707317073171e-05,
+      "loss": 0.6589,
+      "step": 84
+    },
+    {
+      "epoch": 0.34430379746835443,
+      "grad_norm": 0.7959018371533267,
+      "learning_rate": 2.764227642276423e-05,
+      "loss": 0.6604,
+      "step": 85
+    },
+    {
+      "epoch": 0.34835443037974684,
+      "grad_norm": 0.8908113075513283,
+      "learning_rate": 2.796747967479675e-05,
+      "loss": 0.6681,
+      "step": 86
+    },
+    {
+      "epoch": 0.35240506329113924,
+      "grad_norm": 1.247417745453985,
+      "learning_rate": 2.829268292682927e-05,
+      "loss": 0.6665,
+      "step": 87
+    },
+    {
+      "epoch": 0.35645569620253165,
+      "grad_norm": 1.011676383672481,
+      "learning_rate": 2.8617886178861792e-05,
+      "loss": 0.6738,
+      "step": 88
+    },
+    {
+      "epoch": 0.36050632911392405,
+      "grad_norm": 1.047360646256462,
+      "learning_rate": 2.8943089430894314e-05,
+      "loss": 0.6559,
+      "step": 89
+    },
+    {
+      "epoch": 0.36455696202531646,
+      "grad_norm": 0.936670743062065,
+      "learning_rate": 2.926829268292683e-05,
+      "loss": 0.6801,
+      "step": 90
+    },
+    {
+      "epoch": 0.36860759493670886,
+      "grad_norm": 0.778254630386846,
+      "learning_rate": 2.959349593495935e-05,
+      "loss": 0.6748,
+      "step": 91
+    },
+    {
+      "epoch": 0.37265822784810126,
+      "grad_norm": 1.1363742767117666,
+      "learning_rate": 2.991869918699187e-05,
+      "loss": 0.6827,
+      "step": 92
+    },
+    {
+      "epoch": 0.37670886075949367,
+      "grad_norm": 1.1221973338494988,
+      "learning_rate": 3.0243902439024392e-05,
+      "loss": 0.663,
+      "step": 93
+    },
+    {
+      "epoch": 0.3807594936708861,
+      "grad_norm": 1.0202616966657243,
+      "learning_rate": 3.056910569105692e-05,
+      "loss": 0.6503,
+      "step": 94
+    },
+    {
+      "epoch": 0.3848101265822785,
+      "grad_norm": 1.534099024826461,
+      "learning_rate": 3.089430894308943e-05,
+      "loss": 0.6548,
+      "step": 95
+    },
+    {
+      "epoch": 0.3888607594936709,
+      "grad_norm": 0.6461197714871487,
+      "learning_rate": 3.1219512195121955e-05,
+      "loss": 0.6656,
+      "step": 96
+    },
+    {
+      "epoch": 0.3929113924050633,
+      "grad_norm": 1.302510392461573,
+      "learning_rate": 3.154471544715447e-05,
+      "loss": 0.673,
+      "step": 97
+    },
+    {
+      "epoch": 0.3969620253164557,
+      "grad_norm": 1.3561973264531044,
+      "learning_rate": 3.186991869918699e-05,
+      "loss": 0.6602,
+      "step": 98
+    },
+    {
+      "epoch": 0.4010126582278481,
+      "grad_norm": 1.0592034697182116,
+      "learning_rate": 3.2195121951219514e-05,
+      "loss": 0.6696,
+      "step": 99
+    },
+    {
+      "epoch": 0.4050632911392405,
+      "grad_norm": 1.4831103831575274,
+      "learning_rate": 3.2520325203252037e-05,
+      "loss": 0.6653,
+      "step": 100
+    },
+    {
+      "epoch": 0.4091139240506329,
+      "grad_norm": 0.9156679645939271,
+      "learning_rate": 3.284552845528456e-05,
+      "loss": 0.6552,
+      "step": 101
+    },
+    {
+      "epoch": 0.4131645569620253,
+      "grad_norm": 1.5081801403878143,
+      "learning_rate": 3.3170731707317074e-05,
+      "loss": 0.6629,
+      "step": 102
+    },
+    {
+      "epoch": 0.4172151898734177,
+      "grad_norm": 1.149030005902453,
+      "learning_rate": 3.3495934959349596e-05,
+      "loss": 0.653,
+      "step": 103
+    },
+    {
+      "epoch": 0.4212658227848101,
+      "grad_norm": 1.3109344885475176,
+      "learning_rate": 3.382113821138212e-05,
+      "loss": 0.6652,
+      "step": 104
+    },
+    {
+      "epoch": 0.4253164556962025,
+      "grad_norm": 1.1974454586135963,
+      "learning_rate": 3.414634146341463e-05,
+      "loss": 0.6617,
+      "step": 105
+    },
+    {
+      "epoch": 0.4293670886075949,
+      "grad_norm": 1.3968661795068182,
+      "learning_rate": 3.4471544715447155e-05,
+      "loss": 0.6688,
+      "step": 106
+    },
+    {
+      "epoch": 0.43341772151898733,
+      "grad_norm": 0.8192564976682359,
+      "learning_rate": 3.479674796747968e-05,
+      "loss": 0.6471,
+      "step": 107
+    },
+    {
+      "epoch": 0.43746835443037974,
+      "grad_norm": 1.1604582784022137,
+      "learning_rate": 3.51219512195122e-05,
+      "loss": 0.6657,
+      "step": 108
+    },
+    {
+      "epoch": 0.44151898734177214,
+      "grad_norm": 1.2284997875002734,
+      "learning_rate": 3.544715447154472e-05,
+      "loss": 0.6576,
+      "step": 109
+    },
+    {
+      "epoch": 0.44556962025316454,
+      "grad_norm": 1.1112723763500196,
+      "learning_rate": 3.577235772357724e-05,
+      "loss": 0.658,
+      "step": 110
+    },
+    {
+      "epoch": 0.44962025316455695,
+      "grad_norm": 0.9332308919884056,
+      "learning_rate": 3.609756097560976e-05,
+      "loss": 0.6492,
+      "step": 111
+    },
+    {
+      "epoch": 0.45367088607594935,
+      "grad_norm": 1.2058377067306505,
+      "learning_rate": 3.642276422764228e-05,
+      "loss": 0.6559,
+      "step": 112
+    },
+    {
+      "epoch": 0.45772151898734176,
+      "grad_norm": 0.8762156223930023,
+      "learning_rate": 3.67479674796748e-05,
+      "loss": 0.6524,
+      "step": 113
+    },
+    {
+      "epoch": 0.46177215189873416,
+      "grad_norm": 0.9974766276776971,
+      "learning_rate": 3.7073170731707325e-05,
+      "loss": 0.6986,
+      "step": 114
+    },
+    {
+      "epoch": 0.46582278481012657,
+      "grad_norm": 1.2825488758103822,
+      "learning_rate": 3.739837398373984e-05,
+      "loss": 0.646,
+      "step": 115
+    },
+    {
+      "epoch": 0.46987341772151897,
+      "grad_norm": 1.3791615104180788,
+      "learning_rate": 3.772357723577236e-05,
+      "loss": 0.6315,
+      "step": 116
+    },
+    {
+      "epoch": 0.4739240506329114,
+      "grad_norm": 0.7927348096869162,
+      "learning_rate": 3.804878048780488e-05,
+      "loss": 0.6445,
+      "step": 117
+    },
+    {
+      "epoch": 0.4779746835443038,
+      "grad_norm": 1.5967845354807566,
+      "learning_rate": 3.83739837398374e-05,
+      "loss": 0.6474,
+      "step": 118
+    },
+    {
+      "epoch": 0.4820253164556962,
+      "grad_norm": 0.8821004758141643,
+      "learning_rate": 3.869918699186992e-05,
+      "loss": 0.6429,
+      "step": 119
+    },
+    {
+      "epoch": 0.4860759493670886,
+      "grad_norm": 1.3701405274791276,
+      "learning_rate": 3.9024390243902444e-05,
+      "loss": 0.6581,
+      "step": 120
+    },
+    {
+      "epoch": 0.490126582278481,
+      "grad_norm": 1.3274824660431699,
+      "learning_rate": 3.9349593495934966e-05,
+      "loss": 0.6626,
+      "step": 121
+    },
+    {
+      "epoch": 0.4941772151898734,
+      "grad_norm": 1.1233105513540989,
+      "learning_rate": 3.967479674796748e-05,
+      "loss": 0.6627,
+      "step": 122
+    },
+    {
+      "epoch": 0.4982278481012658,
+      "grad_norm": 0.974173872199073,
+      "learning_rate": 4e-05,
+      "loss": 0.6454,
+      "step": 123
+    },
+    {
+      "epoch": 0.5022784810126583,
+      "grad_norm": 1.2033219032586282,
+      "learning_rate": 3.999991946137476e-05,
+      "loss": 0.6496,
+      "step": 124
+    },
+    {
+      "epoch": 0.5063291139240507,
+      "grad_norm": 1.0770853766098971,
+      "learning_rate": 3.999967784614766e-05,
+      "loss": 0.6518,
+      "step": 125
+    },
+    {
+      "epoch": 0.5103797468354431,
+      "grad_norm": 1.001027659949386,
+      "learning_rate": 3.9999275156264656e-05,
+      "loss": 0.6438,
+      "step": 126
+    },
+    {
+      "epoch": 0.5144303797468355,
+      "grad_norm": 0.8763645487074053,
+      "learning_rate": 3.999871139496895e-05,
+      "loss": 0.6423,
+      "step": 127
+    },
+    {
+      "epoch": 0.5184810126582279,
+      "grad_norm": 1.0006640351318323,
+      "learning_rate": 3.9997986566800995e-05,
+      "loss": 0.645,
+      "step": 128
+    },
+    {
+      "epoch": 0.5225316455696203,
+      "grad_norm": 1.655850724767539,
+      "learning_rate": 3.999710067759846e-05,
+      "loss": 0.6925,
+      "step": 129
+    },
+    {
+      "epoch": 0.5265822784810127,
+      "grad_norm": 0.8534925970272823,
+      "learning_rate": 3.999605373449617e-05,
+      "loss": 0.6618,
+      "step": 130
+    },
+    {
+      "epoch": 0.5306329113924051,
+      "grad_norm": 3.12404171991767,
+      "learning_rate": 3.9994845745926075e-05,
+      "loss": 0.6511,
+      "step": 131
+    },
+    {
+      "epoch": 0.5346835443037975,
+      "grad_norm": 2.1464765762725566,
+      "learning_rate": 3.999347672161713e-05,
+      "loss": 0.6415,
+      "step": 132
+    },
+    {
+      "epoch": 0.5387341772151899,
+      "grad_norm": 1.2362459828014085,
+      "learning_rate": 3.999194667259528e-05,
+      "loss": 0.6562,
+      "step": 133
+    },
+    {
+      "epoch": 0.5427848101265823,
+      "grad_norm": 1.4015876307210802,
+      "learning_rate": 3.999025561118334e-05,
+      "loss": 0.6597,
+      "step": 134
+    },
+    {
+      "epoch": 0.5468354430379747,
+      "grad_norm": 1.1016966758388884,
+      "learning_rate": 3.998840355100086e-05,
+      "loss": 0.6403,
+      "step": 135
+    },
+    {
+      "epoch": 0.5508860759493671,
+      "grad_norm": 1.3800844338779508,
+      "learning_rate": 3.998639050696409e-05,
+      "loss": 0.665,
+      "step": 136
+    },
+    {
+      "epoch": 0.5549367088607595,
+      "grad_norm": 1.001117312019488,
+      "learning_rate": 3.998421649528582e-05,
+      "loss": 0.6459,
+      "step": 137
+    },
+    {
+      "epoch": 0.5589873417721519,
+      "grad_norm": 0.921424288211298,
+      "learning_rate": 3.9981881533475234e-05,
+      "loss": 0.6548,
+      "step": 138
+    },
+    {
+      "epoch": 0.5630379746835443,
+      "grad_norm": 1.1240667762086187,
+      "learning_rate": 3.997938564033779e-05,
+      "loss": 0.6373,
+      "step": 139
+    },
+    {
+      "epoch": 0.5670886075949367,
+      "grad_norm": 0.9427654582883883,
+      "learning_rate": 3.9976728835975064e-05,
+      "loss": 0.6432,
+      "step": 140
+    },
+    {
+      "epoch": 0.5711392405063291,
+      "grad_norm": 0.9320470605268965,
+      "learning_rate": 3.9973911141784605e-05,
+      "loss": 0.6444,
+      "step": 141
+    },
+    {
+      "epoch": 0.5751898734177215,
+      "grad_norm": 0.9933307556975449,
+      "learning_rate": 3.997093258045973e-05,
+      "loss": 0.6609,
+      "step": 142
+    },
+    {
+      "epoch": 0.579240506329114,
+      "grad_norm": 1.051309986063521,
+      "learning_rate": 3.996779317598936e-05,
+      "loss": 0.645,
+      "step": 143
+    },
+    {
+      "epoch": 0.5832911392405064,
+      "grad_norm": 0.9009827373999945,
+      "learning_rate": 3.996449295365782e-05,
+      "loss": 0.6409,
+      "step": 144
+    },
+    {
+      "epoch": 0.5873417721518988,
+      "grad_norm": 0.8471661013444758,
+      "learning_rate": 3.996103194004467e-05,
+      "loss": 0.6556,
+      "step": 145
+    },
+    {
+      "epoch": 0.5913924050632912,
+      "grad_norm": 0.9626653475859239,
+      "learning_rate": 3.995741016302441e-05,
+      "loss": 0.6362,
+      "step": 146
+    },
+    {
+      "epoch": 0.5954430379746836,
+      "grad_norm": 0.6755597552484974,
+      "learning_rate": 3.9953627651766364e-05,
+      "loss": 0.647,
+      "step": 147
+    },
+    {
+      "epoch": 0.599493670886076,
+      "grad_norm": 0.6828315418937321,
+      "learning_rate": 3.9949684436734325e-05,
+      "loss": 0.6465,
+      "step": 148
+    },
+    {
+      "epoch": 0.6035443037974684,
+      "grad_norm": 0.7933824548292066,
+      "learning_rate": 3.994558054968643e-05,
+      "loss": 0.6498,
+      "step": 149
+    },
+    {
+      "epoch": 0.6075949367088608,
+      "grad_norm": 0.7219662563509037,
+      "learning_rate": 3.994131602367481e-05,
+      "loss": 0.6419,
+      "step": 150
+    },
+    {
+      "epoch": 0.6116455696202532,
+      "grad_norm": 0.7739325873294521,
+      "learning_rate": 3.9936890893045376e-05,
+      "loss": 0.6423,
+      "step": 151
+    },
+    {
+      "epoch": 0.6156962025316456,
+      "grad_norm": 0.7304112553298716,
+      "learning_rate": 3.993230519343752e-05,
+      "loss": 0.6364,
+      "step": 152
+    },
+    {
+      "epoch": 0.619746835443038,
+      "grad_norm": 0.9108757923078489,
+      "learning_rate": 3.992755896178383e-05,
+      "loss": 0.6346,
+      "step": 153
+    },
+    {
+      "epoch": 0.6237974683544304,
+      "grad_norm": 0.7125701045897886,
+      "learning_rate": 3.992265223630981e-05,
+      "loss": 0.6306,
+      "step": 154
+    },
+    {
+      "epoch": 0.6278481012658228,
+      "grad_norm": 3.3720661608969436,
+      "learning_rate": 3.991758505653355e-05,
+      "loss": 0.6501,
+      "step": 155
+    },
+    {
+      "epoch": 0.6318987341772152,
+      "grad_norm": 0.7803044214755777,
+      "learning_rate": 3.991235746326543e-05,
+      "loss": 0.6389,
+      "step": 156
+    },
+    {
+      "epoch": 0.6359493670886076,
+      "grad_norm": 0.8499518621474861,
+      "learning_rate": 3.9906969498607745e-05,
+      "loss": 0.6353,
+      "step": 157
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.7619826232162474,
+      "learning_rate": 3.990142120595444e-05,
+      "loss": 0.6547,
+      "step": 158
+    },
+    {
+      "epoch": 0.6440506329113924,
+      "grad_norm": 0.7884478331368974,
+      "learning_rate": 3.98957126299907e-05,
+      "loss": 0.6444,
+      "step": 159
+    },
+    {
+      "epoch": 0.6481012658227848,
+      "grad_norm": 0.8357124603230247,
+      "learning_rate": 3.9889843816692596e-05,
+      "loss": 0.6355,
+      "step": 160
+    },
+    {
+      "epoch": 0.6521518987341772,
+      "grad_norm": 0.7419416584080845,
+      "learning_rate": 3.9883814813326766e-05,
+      "loss": 0.6401,
+      "step": 161
+    },
+    {
+      "epoch": 0.6562025316455696,
+      "grad_norm": 0.7127694591986538,
+      "learning_rate": 3.9877625668449956e-05,
+      "loss": 0.6294,
+      "step": 162
+    },
+    {
+      "epoch": 0.660253164556962,
+      "grad_norm": 0.6986141605791567,
+      "learning_rate": 3.98712764319087e-05,
+      "loss": 0.6538,
+      "step": 163
+    },
+    {
+      "epoch": 0.6643037974683544,
+      "grad_norm": 0.6439002856946543,
+      "learning_rate": 3.9864767154838864e-05,
+      "loss": 0.6372,
+      "step": 164
+    },
+    {
+      "epoch": 0.6683544303797468,
+      "grad_norm": 0.7971587787196176,
+      "learning_rate": 3.9858097889665277e-05,
+      "loss": 0.6481,
+      "step": 165
+    },
+    {
+      "epoch": 0.6724050632911392,
+      "grad_norm": 0.9215342819313256,
+      "learning_rate": 3.985126869010129e-05,
+      "loss": 0.637,
+      "step": 166
+    },
+    {
+      "epoch": 0.6764556962025317,
+      "grad_norm": 0.706993002734517,
+      "learning_rate": 3.984427961114833e-05,
+      "loss": 0.6306,
+      "step": 167
+    },
+    {
+      "epoch": 0.6805063291139241,
+      "grad_norm": 0.7767684378889671,
+      "learning_rate": 3.9837130709095475e-05,
+      "loss": 0.6216,
+      "step": 168
+    },
+    {
+      "epoch": 0.6845569620253165,
+      "grad_norm": 0.7152021391110636,
+      "learning_rate": 3.982982204151901e-05,
+      "loss": 0.6335,
+      "step": 169
+    },
+    {
+      "epoch": 0.6886075949367089,
+      "grad_norm": 0.7988691364751567,
+      "learning_rate": 3.982235366728193e-05,
+      "loss": 0.6338,
+      "step": 170
+    },
+    {
+      "epoch": 0.6926582278481013,
+      "grad_norm": 0.9436462522014369,
+      "learning_rate": 3.9814725646533505e-05,
+      "loss": 0.6225,
+      "step": 171
+    },
+    {
+      "epoch": 0.6967088607594937,
+      "grad_norm": 0.7915170149477645,
+      "learning_rate": 3.9806938040708746e-05,
+      "loss": 0.6484,
+      "step": 172
+    },
+    {
+      "epoch": 0.7007594936708861,
+      "grad_norm": 0.7939834547356556,
+      "learning_rate": 3.9798990912527976e-05,
+      "loss": 0.6388,
+      "step": 173
+    },
+    {
+      "epoch": 0.7048101265822785,
+      "grad_norm": 0.5268456350129127,
+      "learning_rate": 3.979088432599627e-05,
+      "loss": 0.6242,
+      "step": 174
+    },
+    {
+      "epoch": 0.7088607594936709,
+      "grad_norm": 0.8952143133675022,
+      "learning_rate": 3.9782618346402964e-05,
+      "loss": 0.6296,
+      "step": 175
+    },
+    {
+      "epoch": 0.7129113924050633,
+      "grad_norm": 0.5845067011748345,
+      "learning_rate": 3.977419304032111e-05,
+      "loss": 0.6412,
+      "step": 176
+    },
+    {
+      "epoch": 0.7169620253164557,
+      "grad_norm": 0.5251208658507742,
+      "learning_rate": 3.976560847560697e-05,
+      "loss": 0.6286,
+      "step": 177
+    },
+    {
+      "epoch": 0.7210126582278481,
+      "grad_norm": 0.6483295070761234,
+      "learning_rate": 3.9756864721399456e-05,
+      "loss": 0.6242,
+      "step": 178
+    },
+    {
+      "epoch": 0.7250632911392405,
+      "grad_norm": 0.521984614823933,
+      "learning_rate": 3.974796184811956e-05,
+      "loss": 0.6397,
+      "step": 179
+    },
+    {
+      "epoch": 0.7291139240506329,
+      "grad_norm": 0.501747811034605,
+      "learning_rate": 3.973889992746979e-05,
+      "loss": 0.627,
+      "step": 180
+    },
+    {
+      "epoch": 0.7331645569620253,
+      "grad_norm": 2.345191251694919,
+      "learning_rate": 3.972967903243361e-05,
+      "loss": 0.6839,
+      "step": 181
+    },
+    {
+      "epoch": 0.7372151898734177,
+      "grad_norm": 0.669531825540114,
+      "learning_rate": 3.972029923727486e-05,
+      "loss": 0.6278,
+      "step": 182
+    },
+    {
+      "epoch": 0.7412658227848101,
+      "grad_norm": 0.8071894011208748,
+      "learning_rate": 3.971076061753709e-05,
+      "loss": 0.6326,
+      "step": 183
+    },
+    {
+      "epoch": 0.7453164556962025,
+      "grad_norm": 0.7226315617563667,
+      "learning_rate": 3.9701063250043066e-05,
+      "loss": 0.634,
+      "step": 184
+    },
+    {
+      "epoch": 0.7493670886075949,
+      "grad_norm": 0.6496693167643978,
+      "learning_rate": 3.969120721289402e-05,
+      "loss": 0.6313,
+      "step": 185
+    },
+    {
+      "epoch": 0.7534177215189873,
+      "grad_norm": 0.6522377509409054,
+      "learning_rate": 3.9681192585469146e-05,
+      "loss": 0.629,
+      "step": 186
+    },
+    {
+      "epoch": 0.7574683544303797,
+      "grad_norm": 0.7827889212114946,
+      "learning_rate": 3.9671019448424865e-05,
+      "loss": 0.6257,
+      "step": 187
+    },
+    {
+      "epoch": 0.7615189873417721,
+      "grad_norm": 0.5743381251091988,
+      "learning_rate": 3.966068788369422e-05,
+      "loss": 0.6279,
+      "step": 188
+    },
+    {
+      "epoch": 0.7655696202531646,
+      "grad_norm": 0.6023555076034441,
+      "learning_rate": 3.965019797448622e-05,
+      "loss": 0.6321,
+      "step": 189
+    },
+    {
+      "epoch": 0.769620253164557,
+      "grad_norm": 0.5702150060346968,
+      "learning_rate": 3.963954980528515e-05,
+      "loss": 0.6307,
+      "step": 190
+    },
+    {
+      "epoch": 0.7736708860759494,
+      "grad_norm": 0.4207757111857299,
+      "learning_rate": 3.9628743461849905e-05,
+      "loss": 0.6284,
+      "step": 191
+    },
+    {
+      "epoch": 0.7777215189873418,
+      "grad_norm": 0.7125947500866271,
+      "learning_rate": 3.961777903121329e-05,
+      "loss": 0.6384,
+      "step": 192
+    },
+    {
+      "epoch": 0.7817721518987342,
+      "grad_norm": 0.670773941376741,
+      "learning_rate": 3.960665660168131e-05,
+      "loss": 0.6358,
+      "step": 193
+    },
+    {
+      "epoch": 0.7858227848101266,
+      "grad_norm": 0.4411828249174791,
+      "learning_rate": 3.9595376262832485e-05,
+      "loss": 0.6206,
+      "step": 194
+    },
+    {
+      "epoch": 0.789873417721519,
+      "grad_norm": 0.5544985671068478,
+      "learning_rate": 3.9583938105517127e-05,
+      "loss": 0.6296,
+      "step": 195
+    },
+    {
+      "epoch": 0.7939240506329114,
+      "grad_norm": 0.5509582470713575,
+      "learning_rate": 3.957234222185657e-05,
+      "loss": 0.6326,
+      "step": 196
+    },
+    {
+      "epoch": 0.7979746835443038,
+      "grad_norm": 0.5028847992676232,
+      "learning_rate": 3.9560588705242474e-05,
+      "loss": 0.633,
+      "step": 197
+    },
+    {
+      "epoch": 0.8020253164556962,
+      "grad_norm": 0.48321685964184674,
+      "learning_rate": 3.954867765033605e-05,
+      "loss": 0.6274,
+      "step": 198
+    },
+    {
+      "epoch": 0.8060759493670886,
+      "grad_norm": 0.42253496804490304,
+      "learning_rate": 3.953660915306728e-05,
+      "loss": 0.6602,
+      "step": 199
+    },
+    {
+      "epoch": 0.810126582278481,
+      "grad_norm": 0.46571783695745805,
+      "learning_rate": 3.952438331063419e-05,
+      "loss": 0.6273,
+      "step": 200
+    },
+    {
+      "epoch": 0.8141772151898734,
+      "grad_norm": 0.6551524130051822,
+      "learning_rate": 3.951200022150205e-05,
+      "loss": 0.6248,
+      "step": 201
+    },
+    {
+      "epoch": 0.8182278481012658,
+      "grad_norm": 0.4111310570637657,
+      "learning_rate": 3.949945998540253e-05,
+      "loss": 0.6356,
+      "step": 202
+    },
+    {
+      "epoch": 0.8222784810126582,
+      "grad_norm": 0.5805701824902954,
+      "learning_rate": 3.9486762703332993e-05,
+      "loss": 0.6266,
+      "step": 203
+    },
+    {
+      "epoch": 0.8263291139240506,
+      "grad_norm": 0.6258319920508982,
+      "learning_rate": 3.947390847755559e-05,
+      "loss": 0.6289,
+      "step": 204
+    },
+    {
+      "epoch": 0.830379746835443,
+      "grad_norm": 0.4309311481149233,
+      "learning_rate": 3.946089741159648e-05,
+      "loss": 0.6239,
+      "step": 205
+    },
+    {
+      "epoch": 0.8344303797468354,
+      "grad_norm": 0.6211206655413966,
+      "learning_rate": 3.944772961024501e-05,
+      "loss": 0.6166,
+      "step": 206
+    },
+    {
+      "epoch": 0.8384810126582278,
+      "grad_norm": 0.5708160548844384,
+      "learning_rate": 3.943440517955285e-05,
+      "loss": 0.6285,
+      "step": 207
+    },
+    {
+      "epoch": 0.8425316455696202,
+      "grad_norm": 0.35327375774618724,
+      "learning_rate": 3.9420924226833126e-05,
+      "loss": 0.623,
+      "step": 208
+    },
+    {
+      "epoch": 0.8465822784810126,
+      "grad_norm": 0.4341428542212234,
+      "learning_rate": 3.9407286860659566e-05,
+      "loss": 0.6265,
+      "step": 209
+    },
+    {
+      "epoch": 0.850632911392405,
+      "grad_norm": 0.3978150159806719,
+      "learning_rate": 3.9393493190865657e-05,
+      "loss": 0.6298,
+      "step": 210
+    },
+    {
+      "epoch": 0.8546835443037974,
+      "grad_norm": 0.39552318723689256,
+      "learning_rate": 3.937954332854371e-05,
+      "loss": 0.6207,
+      "step": 211
+    },
+    {
+      "epoch": 0.8587341772151899,
+      "grad_norm": 0.4443901946067092,
+      "learning_rate": 3.9365437386044016e-05,
+      "loss": 0.6233,
+      "step": 212
+    },
+    {
+      "epoch": 0.8627848101265823,
+      "grad_norm": 0.3945219810865154,
+      "learning_rate": 3.935117547697387e-05,
+      "loss": 0.6132,
+      "step": 213
+    },
+    {
+      "epoch": 0.8668354430379747,
+      "grad_norm": 0.4307547519179813,
+      "learning_rate": 3.933675771619675e-05,
+      "loss": 0.6317,
+      "step": 214
+    },
+    {
+      "epoch": 0.8708860759493671,
+      "grad_norm": 0.48486342378651276,
+      "learning_rate": 3.932218421983131e-05,
+      "loss": 0.6257,
+      "step": 215
+    },
+    {
+      "epoch": 0.8749367088607595,
+      "grad_norm": 0.561239382098453,
+      "learning_rate": 3.9307455105250484e-05,
+      "loss": 0.6248,
+      "step": 216
+    },
+    {
+      "epoch": 0.8789873417721519,
+      "grad_norm": 0.4878460725532662,
+      "learning_rate": 3.929257049108054e-05,
+      "loss": 0.625,
+      "step": 217
+    },
+    {
+      "epoch": 0.8830379746835443,
+      "grad_norm": 0.4637731426866663,
+      "learning_rate": 3.927753049720011e-05,
+      "loss": 0.6186,
+      "step": 218
+    },
+    {
+      "epoch": 0.8870886075949367,
+      "grad_norm": 0.5118702953317623,
+      "learning_rate": 3.9262335244739234e-05,
+      "loss": 0.6315,
+      "step": 219
+    },
+    {
+      "epoch": 0.8911392405063291,
+      "grad_norm": 0.47229670109149624,
+      "learning_rate": 3.92469848560784e-05,
+      "loss": 0.6328,
+      "step": 220
+    },
+    {
+      "epoch": 0.8951898734177215,
+      "grad_norm": 0.5477612670389427,
+      "learning_rate": 3.923147945484751e-05,
+      "loss": 0.624,
+      "step": 221
+    },
+    {
+      "epoch": 0.8992405063291139,
+      "grad_norm": 0.4476303476536526,
+      "learning_rate": 3.9215819165924956e-05,
+      "loss": 0.6166,
+      "step": 222
+    },
+    {
+      "epoch": 0.9032911392405063,
+      "grad_norm": 0.5345154956516384,
+      "learning_rate": 3.920000411543654e-05,
+      "loss": 0.6157,
+      "step": 223
+    },
+    {
+      "epoch": 0.9073417721518987,
+      "grad_norm": 0.5165178265999331,
+      "learning_rate": 3.9184034430754495e-05,
+      "loss": 0.6284,
+      "step": 224
+    },
+    {
+      "epoch": 0.9113924050632911,
+      "grad_norm": 0.7352801781842193,
+      "learning_rate": 3.916791024049648e-05,
+      "loss": 0.6324,
+      "step": 225
+    },
+    {
+      "epoch": 0.9154430379746835,
+      "grad_norm": 0.6895314110453129,
+      "learning_rate": 3.91516316745245e-05,
+      "loss": 0.6057,
+      "step": 226
+    },
+    {
+      "epoch": 0.9194936708860759,
+      "grad_norm": 0.6039766684879408,
+      "learning_rate": 3.913519886394389e-05,
+      "loss": 0.6257,
+      "step": 227
+    },
+    {
+      "epoch": 0.9235443037974683,
+      "grad_norm": 0.47662543930611173,
+      "learning_rate": 3.911861194110225e-05,
+      "loss": 0.6369,
+      "step": 228
+    },
+    {
+      "epoch": 0.9275949367088607,
+      "grad_norm": 2.266800218450399,
+      "learning_rate": 3.910187103958837e-05,
+      "loss": 0.6176,
+      "step": 229
+    },
+    {
+      "epoch": 0.9316455696202531,
+      "grad_norm": 0.6007436852218514,
+      "learning_rate": 3.908497629423117e-05,
+      "loss": 0.6245,
+      "step": 230
+    },
+    {
+      "epoch": 0.9356962025316455,
+      "grad_norm": 0.8609306969802487,
+      "learning_rate": 3.9067927841098614e-05,
+      "loss": 0.6255,
+      "step": 231
+    },
+    {
+      "epoch": 0.9397468354430379,
+      "grad_norm": 0.9728316991106356,
+      "learning_rate": 3.9050725817496594e-05,
+      "loss": 0.6273,
+      "step": 232
+    },
+    {
+      "epoch": 0.9437974683544303,
+      "grad_norm": 0.9363584828501078,
+      "learning_rate": 3.9033370361967844e-05,
+      "loss": 0.6243,
+      "step": 233
+    },
+    {
+      "epoch": 0.9478481012658228,
+      "grad_norm": 0.8261651192371651,
+      "learning_rate": 3.901586161429081e-05,
+      "loss": 0.6205,
+      "step": 234
+    },
+    {
+      "epoch": 0.9518987341772152,
+      "grad_norm": 0.7174236759114364,
+      "learning_rate": 3.8998199715478545e-05,
+      "loss": 0.6192,
+      "step": 235
+    },
+    {
+      "epoch": 0.9559493670886076,
+      "grad_norm": 0.7567382635940536,
+      "learning_rate": 3.8980384807777564e-05,
+      "loss": 0.5984,
+      "step": 236
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.7823413777821112,
+      "learning_rate": 3.896241703466667e-05,
+      "loss": 0.6121,
+      "step": 237
+    },
+    {
+      "epoch": 0.9640506329113924,
+      "grad_norm": 0.6891454472499935,
+      "learning_rate": 3.894429654085585e-05,
+      "loss": 0.6262,
+      "step": 238
+    },
+    {
+      "epoch": 0.9681012658227848,
+      "grad_norm": 0.5824562863666468,
+      "learning_rate": 3.892602347228505e-05,
+      "loss": 0.6378,
+      "step": 239
+    },
+    {
+      "epoch": 0.9721518987341772,
+      "grad_norm": 0.7325048062270555,
+      "learning_rate": 3.890759797612307e-05,
+      "loss": 0.6067,
+      "step": 240
+    },
+    {
+      "epoch": 0.9762025316455696,
+      "grad_norm": 0.8005181915638786,
+      "learning_rate": 3.888902020076632e-05,
+      "loss": 0.6329,
+      "step": 241
+    },
+    {
+      "epoch": 0.980253164556962,
+      "grad_norm": 0.7468677639028247,
+      "learning_rate": 3.887029029583764e-05,
+      "loss": 0.6229,
+      "step": 242
+    },
+    {
+      "epoch": 0.9843037974683544,
+      "grad_norm": 0.5493186212692502,
+      "learning_rate": 3.8851408412185125e-05,
+      "loss": 0.6152,
+      "step": 243
+    },
+    {
+      "epoch": 0.9883544303797468,
+      "grad_norm": 0.5451027544082364,
+      "learning_rate": 3.8832374701880855e-05,
+      "loss": 0.6242,
+      "step": 244
+    },
+    {
+      "epoch": 0.9924050632911392,
+      "grad_norm": 0.9354396135307061,
+      "learning_rate": 3.881318931821972e-05,
+      "loss": 0.6178,
+      "step": 245
+    },
+    {
+      "epoch": 0.9964556962025316,
+      "grad_norm": 0.8128372622462632,
+      "learning_rate": 3.879385241571817e-05,
+      "loss": 0.6231,
+      "step": 246
+    },
+    {
+      "epoch": 1.000506329113924,
+      "grad_norm": 0.8234209711154441,
+      "learning_rate": 3.8774364150112955e-05,
+      "loss": 0.6155,
+      "step": 247
+    },
+    {
+      "epoch": 1.0045569620253165,
+      "grad_norm": 0.6353032872489213,
+      "learning_rate": 3.8754724678359884e-05,
+      "loss": 0.5708,
+      "step": 248
+    },
+    {
+      "epoch": 1.0086075949367088,
+      "grad_norm": 0.7396959893959472,
+      "learning_rate": 3.873493415863256e-05,
+      "loss": 0.5685,
+      "step": 249
+    },
+    {
+      "epoch": 1.0126582278481013,
+      "grad_norm": 0.9039061156593761,
+      "learning_rate": 3.871499275032111e-05,
+      "loss": 0.5731,
+      "step": 250
+    },
+    {
+      "epoch": 1.0167088607594936,
+      "grad_norm": 0.7570711081933043,
+      "learning_rate": 3.869490061403091e-05,
+      "loss": 0.566,
+      "step": 251
+    },
+    {
+      "epoch": 1.0207594936708861,
+      "grad_norm": 0.741162484815298,
+      "learning_rate": 3.867465791158124e-05,
+      "loss": 0.5769,
+      "step": 252
+    },
+    {
+      "epoch": 1.0248101265822784,
+      "grad_norm": 0.488573977425084,
+      "learning_rate": 3.865426480600407e-05,
+      "loss": 0.5634,
+      "step": 253
+    },
+    {
+      "epoch": 1.028860759493671,
+      "grad_norm": 0.7011470356187858,
+      "learning_rate": 3.863372146154264e-05,
+      "loss": 0.5765,
+      "step": 254
+    },
+    {
+      "epoch": 1.0329113924050632,
+      "grad_norm": 3.276032650061963,
+      "learning_rate": 3.861302804365024e-05,
+      "loss": 0.6044,
+      "step": 255
+    },
+    {
+      "epoch": 1.0369620253164558,
+      "grad_norm": 1.3592046273125458,
+      "learning_rate": 3.85921847189888e-05,
+      "loss": 0.5853,
+      "step": 256
+    },
+    {
+      "epoch": 1.041012658227848,
+      "grad_norm": 1.0908279681585735,
+      "learning_rate": 3.85711916554276e-05,
+      "loss": 0.6167,
+      "step": 257
+    },
+    {
+      "epoch": 1.0450632911392406,
+      "grad_norm": 1.01964481164059,
+      "learning_rate": 3.85500490220419e-05,
+      "loss": 0.571,
+      "step": 258
+    },
+    {
+      "epoch": 1.0491139240506329,
+      "grad_norm": 1.0378558391688495,
+      "learning_rate": 3.852875698911154e-05,
+      "loss": 0.577,
+      "step": 259
+    },
+    {
+      "epoch": 1.0531645569620254,
+      "grad_norm": 1.0372432166554024,
+      "learning_rate": 3.850731572811963e-05,
+      "loss": 0.5743,
+      "step": 260
+    },
+    {
+      "epoch": 1.0572151898734177,
+      "grad_norm": 3.8121769352347443,
+      "learning_rate": 3.848572541175116e-05,
+      "loss": 0.5853,
+      "step": 261
+    },
+    {
+      "epoch": 1.0612658227848102,
+      "grad_norm": 2.0572809327974166,
+      "learning_rate": 3.846398621389154e-05,
+      "loss": 0.5879,
+      "step": 262
+    },
+    {
+      "epoch": 1.0653164556962025,
+      "grad_norm": 0.42270328456285344,
+      "learning_rate": 3.84420983096253e-05,
+      "loss": 0.5631,
+      "step": 263
+    },
+    {
+      "epoch": 1.069367088607595,
+      "grad_norm": 1.5749478265299717,
+      "learning_rate": 3.8420061875234606e-05,
+      "loss": 0.5918,
+      "step": 264
+    },
+    {
+      "epoch": 1.0734177215189873,
+      "grad_norm": 0.6481968686355974,
+      "learning_rate": 3.839787708819787e-05,
+      "loss": 0.5861,
+      "step": 265
+    },
+    {
+      "epoch": 1.0774683544303798,
+      "grad_norm": 1.6927507364965457,
+      "learning_rate": 3.8375544127188325e-05,
+      "loss": 0.5717,
+      "step": 266
+    },
+    {
+      "epoch": 1.081518987341772,
+      "grad_norm": 1.0459880461976456,
+      "learning_rate": 3.8353063172072564e-05,
+      "loss": 0.5742,
+      "step": 267
+    },
+    {
+      "epoch": 1.0855696202531646,
+      "grad_norm": 1.6379309496309788,
+      "learning_rate": 3.8330434403909105e-05,
+      "loss": 0.5772,
+      "step": 268
+    },
+    {
+      "epoch": 1.089620253164557,
+      "grad_norm": 1.5622869434688782,
+      "learning_rate": 3.8307658004946934e-05,
+      "loss": 0.5791,
+      "step": 269
+    },
+    {
+      "epoch": 1.0936708860759494,
+      "grad_norm": 0.8559535202890913,
+      "learning_rate": 3.8284734158624046e-05,
+      "loss": 0.5849,
+      "step": 270
+    },
+    {
+      "epoch": 1.0977215189873417,
+      "grad_norm": 1.1579943969547808,
+      "learning_rate": 3.826166304956594e-05,
+      "loss": 0.5751,
+      "step": 271
+    },
+    {
+      "epoch": 1.1017721518987342,
+      "grad_norm": 0.7067291265227951,
+      "learning_rate": 3.8238444863584164e-05,
+      "loss": 0.565,
+      "step": 272
+    },
+    {
+      "epoch": 1.1058227848101265,
+      "grad_norm": 1.0318431421143763,
+      "learning_rate": 3.821507978767479e-05,
+      "loss": 0.5722,
+      "step": 273
+    },
+    {
+      "epoch": 1.109873417721519,
+      "grad_norm": 0.5594233046819611,
+      "learning_rate": 3.819156801001693e-05,
+      "loss": 0.5612,
+      "step": 274
+    },
+    {
+      "epoch": 1.1139240506329113,
+      "grad_norm": 0.8332740729183064,
+      "learning_rate": 3.816790971997121e-05,
+      "loss": 0.5783,
+      "step": 275
+    },
+    {
+      "epoch": 1.1179746835443038,
+      "grad_norm": 0.7546032492743715,
+      "learning_rate": 3.8144105108078246e-05,
+      "loss": 0.5694,
+      "step": 276
+    },
+    {
+      "epoch": 1.1220253164556961,
+      "grad_norm": 0.6531359502718055,
+      "learning_rate": 3.81201543660571e-05,
+      "loss": 0.5578,
+      "step": 277
+    },
+    {
+      "epoch": 1.1260759493670887,
+      "grad_norm": 0.7918492261836714,
+      "learning_rate": 3.809605768680377e-05,
+      "loss": 0.5858,
+      "step": 278
+    },
+    {
+      "epoch": 1.130126582278481,
+      "grad_norm": 0.8602298456236489,
+      "learning_rate": 3.807181526438958e-05,
+      "loss": 0.5631,
+      "step": 279
+    },
+    {
+      "epoch": 1.1341772151898735,
+      "grad_norm": 0.534208717535231,
+      "learning_rate": 3.8047427294059697e-05,
+      "loss": 0.5556,
+      "step": 280
+    },
+    {
+      "epoch": 1.1382278481012658,
+      "grad_norm": 0.8276463284080177,
+      "learning_rate": 3.802289397223145e-05,
+      "loss": 0.5781,
+      "step": 281
+    },
+    {
+      "epoch": 1.1422784810126583,
+      "grad_norm": 0.5459940053921946,
+      "learning_rate": 3.7998215496492854e-05,
+      "loss": 0.5555,
+      "step": 282
+    },
+    {
+      "epoch": 1.1463291139240506,
+      "grad_norm": 0.6091900955809786,
+      "learning_rate": 3.797339206560096e-05,
+      "loss": 0.5579,
+      "step": 283
+    },
+    {
+      "epoch": 1.150379746835443,
+      "grad_norm": 0.6170578561119218,
+      "learning_rate": 3.794842387948027e-05,
+      "loss": 0.5596,
+      "step": 284
+    },
+    {
+      "epoch": 1.1544303797468354,
+      "grad_norm": 0.5184165881567075,
+      "learning_rate": 3.7923311139221114e-05,
+      "loss": 0.5762,
+      "step": 285
+    },
+    {
+      "epoch": 1.158481012658228,
+      "grad_norm": 0.48843526422221817,
+      "learning_rate": 3.7898054047078054e-05,
+      "loss": 0.5684,
+      "step": 286
+    },
+    {
+      "epoch": 1.1625316455696202,
+      "grad_norm": 0.43182716251115744,
+      "learning_rate": 3.787265280646825e-05,
+      "loss": 0.5673,
+      "step": 287
+    },
+    {
+      "epoch": 1.1665822784810127,
+      "grad_norm": 0.4566153748098503,
+      "learning_rate": 3.7847107621969786e-05,
+      "loss": 0.5526,
+      "step": 288
+    },
+    {
+      "epoch": 1.170632911392405,
+      "grad_norm": 0.46111019215633325,
+      "learning_rate": 3.7821418699320064e-05,
+      "loss": 0.5742,
+      "step": 289
+    },
+    {
+      "epoch": 1.1746835443037975,
+      "grad_norm": 0.5093852287546266,
+      "learning_rate": 3.7795586245414145e-05,
+      "loss": 0.5619,
+      "step": 290
+    },
+    {
+      "epoch": 1.1787341772151898,
+      "grad_norm": 0.4393004699618283,
+      "learning_rate": 3.776961046830306e-05,
+      "loss": 0.5701,
+      "step": 291
+    },
+    {
+      "epoch": 1.1827848101265823,
+      "grad_norm": 0.90533965050441,
+      "learning_rate": 3.774349157719215e-05,
+      "loss": 0.5681,
+      "step": 292
+    },
+    {
+      "epoch": 1.1868354430379746,
+      "grad_norm": 0.4672123660166234,
+      "learning_rate": 3.7717229782439365e-05,
+      "loss": 0.5729,
+      "step": 293
+    },
+    {
+      "epoch": 1.1908860759493671,
+      "grad_norm": 0.4526362574176795,
+      "learning_rate": 3.769082529555359e-05,
+      "loss": 0.5748,
+      "step": 294
+    },
+    {
+      "epoch": 1.1949367088607594,
+      "grad_norm": 0.49547532759016727,
+      "learning_rate": 3.766427832919294e-05,
+      "loss": 0.5748,
+      "step": 295
+    },
+    {
+      "epoch": 1.198987341772152,
+      "grad_norm": 1.5380124471969918,
+      "learning_rate": 3.7637589097163024e-05,
+      "loss": 0.5849,
+      "step": 296
+    },
+    {
+      "epoch": 1.2030379746835442,
+      "grad_norm": 0.49194106566564133,
+      "learning_rate": 3.761075781441526e-05,
+      "loss": 0.5697,
+      "step": 297
+    },
+    {
+      "epoch": 1.2070886075949367,
+      "grad_norm": 0.5369986839676262,
+      "learning_rate": 3.75837846970451e-05,
+      "loss": 0.5499,
+      "step": 298
+    },
+    {
+      "epoch": 1.211139240506329,
+      "grad_norm": 0.4203046179735023,
+      "learning_rate": 3.755666996229032e-05,
+      "loss": 0.5597,
+      "step": 299
+    },
+    {
+      "epoch": 1.2151898734177216,
+      "grad_norm": 0.5650800247129706,
+      "learning_rate": 3.752941382852927e-05,
+      "loss": 0.5772,
+      "step": 300
+    },
+    {
+      "epoch": 1.2192405063291138,
+      "grad_norm": 0.3476068899603692,
+      "learning_rate": 3.7502016515279115e-05,
+      "loss": 0.5513,
+      "step": 301
+    },
+    {
+      "epoch": 1.2232911392405064,
+      "grad_norm": 0.5435101022173598,
+      "learning_rate": 3.7474478243194043e-05,
+      "loss": 0.566,
+      "step": 302
+    },
+    {
+      "epoch": 1.2273417721518987,
+      "grad_norm": 0.47784129756957805,
+      "learning_rate": 3.744679923406351e-05,
+      "loss": 0.5761,
+      "step": 303
+    },
+    {
+      "epoch": 1.2313924050632912,
+      "grad_norm": 0.434425694087252,
+      "learning_rate": 3.741897971081043e-05,
+      "loss": 0.5791,
+      "step": 304
+    },
+    {
+      "epoch": 1.2354430379746835,
+      "grad_norm": 0.46098124580032557,
+      "learning_rate": 3.739101989748946e-05,
+      "loss": 0.5706,
+      "step": 305
+    },
+    {
+      "epoch": 1.239493670886076,
+      "grad_norm": 0.43092340402799423,
+      "learning_rate": 3.7362920019285066e-05,
+      "loss": 0.5744,
+      "step": 306
+    },
+    {
+      "epoch": 1.2435443037974683,
+      "grad_norm": 0.5213267611696205,
+      "learning_rate": 3.73346803025098e-05,
+      "loss": 0.5763,
+      "step": 307
+    },
+    {
+      "epoch": 1.2475949367088608,
+      "grad_norm": 0.5496196616065996,
+      "learning_rate": 3.730630097460247e-05,
+      "loss": 0.5574,
+      "step": 308
+    },
+    {
+      "epoch": 1.251645569620253,
+      "grad_norm": 0.42127192635845007,
+      "learning_rate": 3.727778226412628e-05,
+      "loss": 0.5712,
+      "step": 309
+    },
+    {
+      "epoch": 1.2556962025316456,
+      "grad_norm": 0.7339077269993177,
+      "learning_rate": 3.7249124400767006e-05,
+      "loss": 0.5705,
+      "step": 310
+    },
+    {
+      "epoch": 1.259746835443038,
+      "grad_norm": 0.47911240300216057,
+      "learning_rate": 3.722032761533114e-05,
+      "loss": 0.5705,
+      "step": 311
+    },
+    {
+      "epoch": 1.2637974683544304,
+      "grad_norm": 0.4749894190418205,
+      "learning_rate": 3.719139213974403e-05,
+      "loss": 0.5655,
+      "step": 312
+    },
+    {
+      "epoch": 1.2678481012658227,
+      "grad_norm": 0.4857459411692775,
+      "learning_rate": 3.7162318207048006e-05,
+      "loss": 0.5674,
+      "step": 313
+    },
+    {
+      "epoch": 1.2718987341772152,
+      "grad_norm": 0.34330024578067697,
+      "learning_rate": 3.713310605140055e-05,
+      "loss": 0.5663,
+      "step": 314
+    },
+    {
+      "epoch": 1.2759493670886077,
+      "grad_norm": 0.41606777384825283,
+      "learning_rate": 3.710375590807233e-05,
+      "loss": 0.569,
+      "step": 315
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.4401300979787435,
+      "learning_rate": 3.7074268013445365e-05,
+      "loss": 0.5749,
+      "step": 316
+    },
+    {
+      "epoch": 1.2840506329113923,
+      "grad_norm": 0.37886794158566567,
+      "learning_rate": 3.7044642605011114e-05,
+      "loss": 0.5696,
+      "step": 317
+    },
+    {
+      "epoch": 1.2881012658227848,
+      "grad_norm": 0.49305024957383353,
+      "learning_rate": 3.701487992136854e-05,
+      "loss": 0.5598,
+      "step": 318
+    },
+    {
+      "epoch": 1.2921518987341774,
+      "grad_norm": 0.4221727710231363,
+      "learning_rate": 3.69849802022222e-05,
+      "loss": 0.5686,
+      "step": 319
+    },
+    {
+      "epoch": 1.2962025316455696,
+      "grad_norm": 0.4307545531753929,
+      "learning_rate": 3.6954943688380334e-05,
+      "loss": 0.5696,
+      "step": 320
+    },
+    {
+      "epoch": 1.300253164556962,
+      "grad_norm": 0.4275766728557898,
+      "learning_rate": 3.692477062175289e-05,
+      "loss": 0.5611,
+      "step": 321
+    },
+    {
+      "epoch": 1.3043037974683545,
+      "grad_norm": 0.4566295489759074,
+      "learning_rate": 3.689446124534958e-05,
+      "loss": 0.5727,
+      "step": 322
+    },
+    {
+      "epoch": 1.308354430379747,
+      "grad_norm": 0.4531839628366296,
+      "learning_rate": 3.686401580327799e-05,
+      "loss": 0.5685,
+      "step": 323
+    },
+    {
+      "epoch": 1.3124050632911393,
+      "grad_norm": 0.4996686681161687,
+      "learning_rate": 3.683343454074149e-05,
+      "loss": 0.5679,
+      "step": 324
+    },
+    {
+      "epoch": 1.3164556962025316,
+      "grad_norm": 0.33727387090912975,
+      "learning_rate": 3.6802717704037386e-05,
+      "loss": 0.5634,
+      "step": 325
+    },
+    {
+      "epoch": 1.320506329113924,
+      "grad_norm": 0.3963675267900204,
+      "learning_rate": 3.6771865540554855e-05,
+      "loss": 0.5595,
+      "step": 326
+    },
+    {
+      "epoch": 1.3245569620253166,
+      "grad_norm": 0.38319038811494466,
+      "learning_rate": 3.674087829877297e-05,
+      "loss": 0.5627,
+      "step": 327
+    },
+    {
+      "epoch": 1.3286075949367089,
+      "grad_norm": 0.37103007539860966,
+      "learning_rate": 3.6709756228258735e-05,
+      "loss": 0.5918,
+      "step": 328
+    },
+    {
+      "epoch": 1.3326582278481012,
+      "grad_norm": 0.4720948781775961,
+      "learning_rate": 3.667849957966501e-05,
+      "loss": 0.5665,
+      "step": 329
+    },
+    {
+      "epoch": 1.3367088607594937,
+      "grad_norm": 0.41296532258612395,
+      "learning_rate": 3.6647108604728546e-05,
+      "loss": 0.5687,
+      "step": 330
+    },
+    {
+      "epoch": 1.3407594936708862,
+      "grad_norm": 0.4000916766586861,
+      "learning_rate": 3.661558355626795e-05,
+      "loss": 0.5749,
+      "step": 331
+    },
+    {
+      "epoch": 1.3448101265822785,
+      "grad_norm": 0.3826222840971011,
+      "learning_rate": 3.658392468818163e-05,
+      "loss": 0.5533,
+      "step": 332
+    },
+    {
+      "epoch": 1.3488607594936708,
+      "grad_norm": 0.34264443406126893,
+      "learning_rate": 3.655213225544574e-05,
+      "loss": 0.5601,
+      "step": 333
+    },
+    {
+      "epoch": 1.3529113924050633,
+      "grad_norm": 0.34741998951196895,
+      "learning_rate": 3.652020651411218e-05,
+      "loss": 0.5666,
+      "step": 334
+    },
+    {
+      "epoch": 1.3569620253164558,
+      "grad_norm": 0.3363407518102006,
+      "learning_rate": 3.6488147721306474e-05,
+      "loss": 0.5613,
+      "step": 335
+    },
+    {
+      "epoch": 1.3610126582278481,
+      "grad_norm": 2.097122773490873,
+      "learning_rate": 3.645595613522574e-05,
+      "loss": 0.5983,
+      "step": 336
+    },
+    {
+      "epoch": 1.3650632911392404,
+      "grad_norm": 0.38986048250221966,
+      "learning_rate": 3.642363201513657e-05,
+      "loss": 0.5606,
+      "step": 337
+    },
+    {
+      "epoch": 1.369113924050633,
+      "grad_norm": 0.3727306220605748,
+      "learning_rate": 3.6391175621373006e-05,
+      "loss": 0.58,
+      "step": 338
+    },
+    {
+      "epoch": 1.3731645569620254,
+      "grad_norm": 0.3839221119887089,
+      "learning_rate": 3.6358587215334355e-05,
+      "loss": 0.563,
+      "step": 339
+    },
+    {
+      "epoch": 1.3772151898734177,
+      "grad_norm": 0.3445787267373247,
+      "learning_rate": 3.632586705948318e-05,
+      "loss": 0.5646,
+      "step": 340
+    },
+    {
+      "epoch": 1.38126582278481,
+      "grad_norm": 0.35012172188361035,
+      "learning_rate": 3.629301541734311e-05,
+      "loss": 0.5601,
+      "step": 341
+    },
+    {
+      "epoch": 1.3853164556962025,
+      "grad_norm": 0.3406378733108128,
+      "learning_rate": 3.626003255349676e-05,
+      "loss": 0.5599,
+      "step": 342
+    },
+    {
+      "epoch": 1.389367088607595,
+      "grad_norm": 0.37801853132965807,
+      "learning_rate": 3.622691873358357e-05,
+      "loss": 0.5786,
+      "step": 343
+    },
+    {
+      "epoch": 1.3934177215189874,
+      "grad_norm": 1.6700996901265937,
+      "learning_rate": 3.61936742242977e-05,
+      "loss": 0.6032,
+      "step": 344
+    },
+    {
+      "epoch": 1.3974683544303796,
+      "grad_norm": 0.37593906130692617,
+      "learning_rate": 3.6160299293385864e-05,
+      "loss": 0.5652,
+      "step": 345
+    },
+    {
+      "epoch": 1.4015189873417722,
+      "grad_norm": 0.41809290285926454,
+      "learning_rate": 3.612679420964516e-05,
+      "loss": 0.5617,
+      "step": 346
+    },
+    {
+      "epoch": 1.4055696202531647,
+      "grad_norm": 0.49729405274176053,
+      "learning_rate": 3.609315924292092e-05,
+      "loss": 0.5631,
+      "step": 347
+    },
+    {
+      "epoch": 1.409620253164557,
+      "grad_norm": 0.4522810762300436,
+      "learning_rate": 3.6059394664104554e-05,
+      "loss": 0.5768,
+      "step": 348
+    },
+    {
+      "epoch": 1.4136708860759493,
+      "grad_norm": 0.42467372150270394,
+      "learning_rate": 3.602550074513133e-05,
+      "loss": 0.5775,
+      "step": 349
+    },
+    {
+      "epoch": 1.4177215189873418,
+      "grad_norm": 0.4690363255199558,
+      "learning_rate": 3.599147775897822e-05,
+      "loss": 0.5855,
+      "step": 350
+    },
+    {
+      "epoch": 1.4217721518987343,
+      "grad_norm": 0.4654090083823508,
+      "learning_rate": 3.595732597966167e-05,
+      "loss": 0.5906,
+      "step": 351
+    },
+    {
+      "epoch": 1.4258227848101266,
+      "grad_norm": 0.4662088650001763,
+      "learning_rate": 3.592304568223542e-05,
+      "loss": 0.5695,
+      "step": 352
+    },
+    {
+      "epoch": 1.4298734177215189,
+      "grad_norm": 0.4205543496790651,
+      "learning_rate": 3.588863714278826e-05,
+      "loss": 0.5712,
+      "step": 353
+    },
+    {
+      "epoch": 1.4339240506329114,
+      "grad_norm": 0.4479766843699324,
+      "learning_rate": 3.585410063844186e-05,
+      "loss": 0.5653,
+      "step": 354
+    },
+    {
+      "epoch": 1.437974683544304,
+      "grad_norm": 0.40862760905975254,
+      "learning_rate": 3.581943644734846e-05,
+      "loss": 0.565,
+      "step": 355
+    },
+    {
+      "epoch": 1.4420253164556962,
+      "grad_norm": 0.3877095182678462,
+      "learning_rate": 3.578464484868869e-05,
+      "loss": 0.5512,
+      "step": 356
+    },
+    {
+      "epoch": 1.4460759493670885,
+      "grad_norm": 0.3493165700707726,
+      "learning_rate": 3.5749726122669316e-05,
+      "loss": 0.5751,
+      "step": 357
+    },
+    {
+      "epoch": 1.450126582278481,
+      "grad_norm": 0.39508533874961227,
+      "learning_rate": 3.5714680550520943e-05,
+      "loss": 0.5672,
+      "step": 358
+    },
+    {
+      "epoch": 1.4541772151898735,
+      "grad_norm": 0.5595278708398115,
+      "learning_rate": 3.5679508414495794e-05,
+      "loss": 0.5568,
+      "step": 359
+    },
+    {
+      "epoch": 1.4582278481012658,
+      "grad_norm": 0.6488202066101495,
+      "learning_rate": 3.564420999786543e-05,
+      "loss": 0.5762,
+      "step": 360
+    },
+    {
+      "epoch": 1.4622784810126581,
+      "grad_norm": 0.4392797795976331,
+      "learning_rate": 3.560878558491842e-05,
+      "loss": 0.5648,
+      "step": 361
+    },
+    {
+      "epoch": 1.4663291139240506,
+      "grad_norm": 0.5998217543756692,
+      "learning_rate": 3.5573235460958145e-05,
+      "loss": 0.5701,
+      "step": 362
+    },
+    {
+      "epoch": 1.4703797468354431,
+      "grad_norm": 0.6200391574989828,
+      "learning_rate": 3.553755991230039e-05,
+      "loss": 0.5648,
+      "step": 363
+    },
+    {
+      "epoch": 1.4744303797468354,
+      "grad_norm": 0.5628375987934299,
+      "learning_rate": 3.5501759226271144e-05,
+      "loss": 0.5644,
+      "step": 364
+    },
+    {
+      "epoch": 1.4784810126582277,
+      "grad_norm": 0.3709866451874852,
+      "learning_rate": 3.546583369120419e-05,
+      "loss": 0.5704,
+      "step": 365
+    },
+    {
+      "epoch": 1.4825316455696202,
+      "grad_norm": 0.43573315510835015,
+      "learning_rate": 3.5429783596438864e-05,
+      "loss": 0.5611,
+      "step": 366
+    },
+    {
+      "epoch": 1.4865822784810128,
+      "grad_norm": 0.44677815862761955,
+      "learning_rate": 3.539360923231766e-05,
+      "loss": 0.5734,
+      "step": 367
+    },
+    {
+      "epoch": 1.490632911392405,
+      "grad_norm": 0.43211914487556247,
+      "learning_rate": 3.535731089018394e-05,
+      "loss": 0.5654,
+      "step": 368
+    },
+    {
+      "epoch": 1.4946835443037974,
+      "grad_norm": 0.44216831793826106,
+      "learning_rate": 3.532088886237956e-05,
+      "loss": 0.5703,
+      "step": 369
+    },
+    {
+      "epoch": 1.4987341772151899,
+      "grad_norm": 0.5202690694252641,
+      "learning_rate": 3.528434344224253e-05,
+      "loss": 0.5579,
+      "step": 370
+    },
+    {
+      "epoch": 1.5027848101265824,
+      "grad_norm": 0.4161871013972273,
+      "learning_rate": 3.524767492410464e-05,
+      "loss": 0.5602,
+      "step": 371
+    },
+    {
+      "epoch": 1.5068354430379747,
+      "grad_norm": 0.37178807188227797,
+      "learning_rate": 3.521088360328908e-05,
+      "loss": 0.5522,
+      "step": 372
+    },
+    {
+      "epoch": 1.510886075949367,
+      "grad_norm": 0.38236510559620307,
+      "learning_rate": 3.517396977610811e-05,
+      "loss": 0.5634,
+      "step": 373
+    },
+    {
+      "epoch": 1.5149367088607595,
+      "grad_norm": 0.36338065153437016,
+      "learning_rate": 3.5136933739860595e-05,
+      "loss": 0.5636,
+      "step": 374
+    },
+    {
+      "epoch": 1.518987341772152,
+      "grad_norm": 0.38355924876265984,
+      "learning_rate": 3.509977579282971e-05,
+      "loss": 0.5615,
+      "step": 375
+    },
+    {
+      "epoch": 1.5230379746835443,
+      "grad_norm": 0.40176275948961365,
+      "learning_rate": 3.5062496234280424e-05,
+      "loss": 0.5707,
+      "step": 376
+    },
+    {
+      "epoch": 1.5270886075949366,
+      "grad_norm": 0.3137514448494699,
+      "learning_rate": 3.502509536445719e-05,
+      "loss": 0.5681,
+      "step": 377
+    },
+    {
+      "epoch": 1.531139240506329,
+      "grad_norm": 0.4786215102910475,
+      "learning_rate": 3.498757348458147e-05,
+      "loss": 0.5616,
+      "step": 378
+    },
+    {
+      "epoch": 1.5351898734177216,
+      "grad_norm": 0.36707060276644926,
+      "learning_rate": 3.4949930896849324e-05,
+      "loss": 0.5643,
+      "step": 379
+    },
+    {
+      "epoch": 1.539240506329114,
+      "grad_norm": 0.4510915192877146,
+      "learning_rate": 3.491216790442899e-05,
+      "loss": 0.5707,
+      "step": 380
+    },
+    {
+      "epoch": 1.5432911392405062,
+      "grad_norm": 0.43594462183123467,
+      "learning_rate": 3.487428481145839e-05,
+      "loss": 0.5584,
+      "step": 381
+    },
+    {
+      "epoch": 1.5473417721518987,
+      "grad_norm": 0.35672136023164186,
+      "learning_rate": 3.483628192304278e-05,
+      "loss": 0.556,
+      "step": 382
+    },
+    {
+      "epoch": 1.5513924050632912,
+      "grad_norm": 0.4232696166127617,
+      "learning_rate": 3.479815954525219e-05,
+      "loss": 0.5604,
+      "step": 383
+    },
+    {
+      "epoch": 1.5554430379746835,
+      "grad_norm": 0.4432004308394549,
+      "learning_rate": 3.475991798511899e-05,
+      "loss": 0.5566,
+      "step": 384
+    },
+    {
+      "epoch": 1.5594936708860758,
+      "grad_norm": 0.329925997849822,
+      "learning_rate": 3.4721557550635464e-05,
+      "loss": 0.5499,
+      "step": 385
+    },
+    {
+      "epoch": 1.5635443037974683,
+      "grad_norm": 0.5010918344630433,
+      "learning_rate": 3.468307855075128e-05,
+      "loss": 0.5591,
+      "step": 386
+    },
+    {
+      "epoch": 1.5675949367088609,
+      "grad_norm": 0.43417346657842304,
+      "learning_rate": 3.4644481295371005e-05,
+      "loss": 0.5548,
+      "step": 387
+    },
+    {
+      "epoch": 1.5716455696202531,
+      "grad_norm": 0.3520197697560343,
+      "learning_rate": 3.460576609535163e-05,
+      "loss": 0.5617,
+      "step": 388
+    },
+    {
+      "epoch": 1.5756962025316454,
+      "grad_norm": 0.4225652605234586,
+      "learning_rate": 3.456693326250006e-05,
+      "loss": 0.5658,
+      "step": 389
+    },
+    {
+      "epoch": 1.579746835443038,
+      "grad_norm": 0.3726074534004623,
+      "learning_rate": 3.452798310957058e-05,
+      "loss": 0.5643,
+      "step": 390
+    },
+    {
+      "epoch": 1.5837974683544305,
+      "grad_norm": 0.3802246597922696,
+      "learning_rate": 3.4488915950262386e-05,
+      "loss": 0.5563,
+      "step": 391
+    },
+    {
+      "epoch": 1.5878481012658228,
+      "grad_norm": 0.4256886708759718,
+      "learning_rate": 3.4449732099216985e-05,
+      "loss": 0.5767,
+      "step": 392
+    },
+    {
+      "epoch": 1.591898734177215,
+      "grad_norm": 0.44549839539831504,
+      "learning_rate": 3.441043187201574e-05,
+      "loss": 0.5675,
+      "step": 393
+    },
+    {
+      "epoch": 1.5959493670886076,
+      "grad_norm": 0.3974965617049389,
+      "learning_rate": 3.437101558517728e-05,
+      "loss": 0.5772,
+      "step": 394
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.3823102294300647,
+      "learning_rate": 3.433148355615496e-05,
+      "loss": 0.5583,
+      "step": 395
+    },
+    {
+      "epoch": 1.6040506329113924,
+      "grad_norm": 0.4777399487009456,
+      "learning_rate": 3.4291836103334294e-05,
+      "loss": 0.5565,
+      "step": 396
+    },
+    {
+      "epoch": 1.6081012658227847,
+      "grad_norm": 0.3613789109790527,
+      "learning_rate": 3.425207354603043e-05,
+      "loss": 0.5667,
+      "step": 397
+    },
+    {
+      "epoch": 1.6121518987341772,
+      "grad_norm": 0.3987893016077345,
+      "learning_rate": 3.421219620448553e-05,
+      "loss": 0.5571,
+      "step": 398
+    },
+    {
+      "epoch": 1.6162025316455697,
+      "grad_norm": 0.41514108178236325,
+      "learning_rate": 3.417220439986623e-05,
+      "loss": 0.564,
+      "step": 399
+    },
+    {
+      "epoch": 1.620253164556962,
+      "grad_norm": 0.40646139124085096,
+      "learning_rate": 3.4132098454261024e-05,
+      "loss": 0.553,
+      "step": 400
+    },
+    {
+      "epoch": 1.6243037974683543,
+      "grad_norm": 0.4230017896868895,
+      "learning_rate": 3.4091878690677676e-05,
+      "loss": 0.5533,
+      "step": 401
+    },
+    {
+      "epoch": 1.6283544303797468,
+      "grad_norm": 0.41970492285038064,
+      "learning_rate": 3.405154543304065e-05,
+      "loss": 0.5591,
+      "step": 402
+    },
+    {
+      "epoch": 1.6324050632911393,
+      "grad_norm": 0.41907337151796376,
+      "learning_rate": 3.401109900618843e-05,
+      "loss": 0.5631,
+      "step": 403
+    },
+    {
+      "epoch": 1.6364556962025316,
+      "grad_norm": 0.37144712851906575,
+      "learning_rate": 3.3970539735870996e-05,
+      "loss": 0.5659,
+      "step": 404
+    },
+    {
+      "epoch": 1.640506329113924,
+      "grad_norm": 0.31739024726120096,
+      "learning_rate": 3.392986794874714e-05,
+      "loss": 0.561,
+      "step": 405
+    },
+    {
+      "epoch": 1.6445569620253164,
+      "grad_norm": 0.35670361433705866,
+      "learning_rate": 3.388908397238184e-05,
+      "loss": 0.5712,
+      "step": 406
+    },
+    {
+      "epoch": 1.648607594936709,
+      "grad_norm": 0.31944903318365464,
+      "learning_rate": 3.384818813524362e-05,
+      "loss": 0.5495,
+      "step": 407
+    },
+    {
+      "epoch": 1.6526582278481012,
+      "grad_norm": 0.31693103327585165,
+      "learning_rate": 3.380718076670195e-05,
+      "loss": 0.552,
+      "step": 408
+    },
+    {
+      "epoch": 1.6567088607594935,
+      "grad_norm": 0.33430219079863965,
+      "learning_rate": 3.376606219702454e-05,
+      "loss": 0.5588,
+      "step": 409
+    },
+    {
+      "epoch": 1.660759493670886,
+      "grad_norm": 0.29533999186035187,
+      "learning_rate": 3.372483275737468e-05,
+      "loss": 0.5507,
+      "step": 410
+    },
+    {
+      "epoch": 1.6648101265822786,
+      "grad_norm": 0.32394533287175103,
+      "learning_rate": 3.368349277980861e-05,
+      "loss": 0.5715,
+      "step": 411
+    },
+    {
+      "epoch": 1.6688607594936709,
+      "grad_norm": 0.2937618424292873,
+      "learning_rate": 3.3642042597272844e-05,
+      "loss": 0.5531,
+      "step": 412
+    },
+    {
+      "epoch": 1.6729113924050631,
+      "grad_norm": 0.3571342983970591,
+      "learning_rate": 3.360048254360144e-05,
+      "loss": 0.5514,
+      "step": 413
+    },
+    {
+      "epoch": 1.6769620253164557,
+      "grad_norm": 0.3736614380034126,
+      "learning_rate": 3.355881295351336e-05,
+      "loss": 0.5715,
+      "step": 414
+    },
+    {
+      "epoch": 1.6810126582278482,
+      "grad_norm": 0.8315922244283974,
+      "learning_rate": 3.351703416260975e-05,
+      "loss": 0.5615,
+      "step": 415
+    },
+    {
+      "epoch": 1.6850632911392405,
+      "grad_norm": 0.3036184951145606,
+      "learning_rate": 3.347514650737126e-05,
+      "loss": 0.5551,
+      "step": 416
+    },
+    {
+      "epoch": 1.6891139240506328,
+      "grad_norm": 0.3620408574232247,
+      "learning_rate": 3.3433150325155295e-05,
+      "loss": 0.5635,
+      "step": 417
+    },
+    {
+      "epoch": 1.6931645569620253,
+      "grad_norm": 0.3183530624308286,
+      "learning_rate": 3.339104595419334e-05,
+      "loss": 0.5531,
+      "step": 418
+    },
+    {
+      "epoch": 1.6972151898734178,
+      "grad_norm": 0.3215427875750439,
+      "learning_rate": 3.3348833733588204e-05,
+      "loss": 0.5614,
+      "step": 419
+    },
+    {
+      "epoch": 1.70126582278481,
+      "grad_norm": 0.40045808111133274,
+      "learning_rate": 3.3306514003311305e-05,
+      "loss": 0.5541,
+      "step": 420
+    },
+    {
+      "epoch": 1.7053164556962024,
+      "grad_norm": 0.3601126389162235,
+      "learning_rate": 3.326408710419996e-05,
+      "loss": 0.563,
+      "step": 421
+    },
+    {
+      "epoch": 1.709367088607595,
+      "grad_norm": 1.1927219034930208,
+      "learning_rate": 3.322155337795454e-05,
+      "loss": 0.5896,
+      "step": 422
+    },
+    {
+      "epoch": 1.7134177215189874,
+      "grad_norm": 0.8689245156852687,
+      "learning_rate": 3.317891316713587e-05,
+      "loss": 0.5704,
+      "step": 423
+    },
+    {
+      "epoch": 1.7174683544303797,
+      "grad_norm": 0.36211605459820295,
+      "learning_rate": 3.313616681516231e-05,
+      "loss": 0.5537,
+      "step": 424
+    },
+    {
+      "epoch": 1.721518987341772,
+      "grad_norm": 0.366192429002904,
+      "learning_rate": 3.309331466630713e-05,
+      "loss": 0.5628,
+      "step": 425
+    },
+    {
+      "epoch": 1.7255696202531645,
+      "grad_norm": 0.3528743282243074,
+      "learning_rate": 3.305035706569563e-05,
+      "loss": 0.5507,
+      "step": 426
+    },
+    {
+      "epoch": 1.729620253164557,
+      "grad_norm": 0.366856338057958,
+      "learning_rate": 3.3007294359302433e-05,
+      "loss": 0.5663,
+      "step": 427
+    },
+    {
+      "epoch": 1.7336708860759493,
+      "grad_norm": 0.40642980533351114,
+      "learning_rate": 3.296412689394864e-05,
+      "loss": 0.5503,
+      "step": 428
+    },
+    {
+      "epoch": 1.7377215189873416,
+      "grad_norm": 0.325176194016976,
+      "learning_rate": 3.292085501729909e-05,
+      "loss": 0.5631,
+      "step": 429
+    },
+    {
+      "epoch": 1.7417721518987341,
+      "grad_norm": 0.39773110372096976,
+      "learning_rate": 3.2877479077859534e-05,
+      "loss": 0.5738,
+      "step": 430
+    },
+    {
+      "epoch": 1.7458227848101266,
+      "grad_norm": 0.41657103057462547,
+      "learning_rate": 3.283399942497381e-05,
+      "loss": 0.5608,
+      "step": 431
+    },
+    {
+      "epoch": 1.749873417721519,
+      "grad_norm": 0.4081810290255033,
+      "learning_rate": 3.279041640882108e-05,
+      "loss": 0.5627,
+      "step": 432
+    },
+    {
+      "epoch": 1.7539240506329112,
+      "grad_norm": 0.3871614519863413,
+      "learning_rate": 3.2746730380412964e-05,
+      "loss": 0.5661,
+      "step": 433
+    },
+    {
+      "epoch": 1.7579746835443038,
+      "grad_norm": 0.44725141605089314,
+      "learning_rate": 3.2702941691590726e-05,
+      "loss": 0.5579,
+      "step": 434
+    },
+    {
+      "epoch": 1.7620253164556963,
+      "grad_norm": 0.6086898832572759,
+      "learning_rate": 3.265905069502244e-05,
+      "loss": 0.5685,
+      "step": 435
+    },
+    {
+      "epoch": 1.7660759493670886,
+      "grad_norm": 0.39557547007086324,
+      "learning_rate": 3.261505774420016e-05,
+      "loss": 0.5617,
+      "step": 436
+    },
+    {
+      "epoch": 1.7701265822784809,
+      "grad_norm": 0.44666421329994843,
+      "learning_rate": 3.257096319343707e-05,
+      "loss": 0.5617,
+      "step": 437
+    },
+    {
+      "epoch": 1.7741772151898734,
+      "grad_norm": 0.3632338374646441,
+      "learning_rate": 3.2526767397864614e-05,
+      "loss": 0.554,
+      "step": 438
+    },
+    {
+      "epoch": 1.7782278481012659,
+      "grad_norm": 0.4706462627364949,
+      "learning_rate": 3.248247071342966e-05,
+      "loss": 0.5601,
+      "step": 439
+    },
+    {
+      "epoch": 1.7822784810126582,
+      "grad_norm": 0.4425388269386228,
+      "learning_rate": 3.243807349689161e-05,
+      "loss": 0.566,
+      "step": 440
+    },
+    {
+      "epoch": 1.7863291139240505,
+      "grad_norm": 0.30411235551886673,
+      "learning_rate": 3.2393576105819544e-05,
+      "loss": 0.5617,
+      "step": 441
+    },
+    {
+      "epoch": 1.790379746835443,
+      "grad_norm": 0.42902095190620043,
+      "learning_rate": 3.2348978898589333e-05,
+      "loss": 0.5667,
+      "step": 442
+    },
+    {
+      "epoch": 1.7944303797468355,
+      "grad_norm": 0.5100613861677228,
+      "learning_rate": 3.230428223438075e-05,
+      "loss": 0.5668,
+      "step": 443
+    },
+    {
+      "epoch": 1.7984810126582278,
+      "grad_norm": 0.3728054829313011,
+      "learning_rate": 3.225948647317459e-05,
+      "loss": 0.5678,
+      "step": 444
+    },
+    {
+      "epoch": 1.80253164556962,
+      "grad_norm": 0.466266110558158,
+      "learning_rate": 3.2214591975749745e-05,
+      "loss": 0.5601,
+      "step": 445
+    },
+    {
+      "epoch": 1.8065822784810126,
+      "grad_norm": 0.3517377524022992,
+      "learning_rate": 3.216959910368034e-05,
+      "loss": 0.5664,
+      "step": 446
+    },
+    {
+      "epoch": 1.8106329113924051,
+      "grad_norm": 2.3947173451860655,
+      "learning_rate": 3.212450821933277e-05,
+      "loss": 0.5604,
+      "step": 447
+    },
+    {
+      "epoch": 1.8146835443037974,
+      "grad_norm": 0.38764722566701904,
+      "learning_rate": 3.207931968586281e-05,
+      "loss": 0.5589,
+      "step": 448
+    },
+    {
+      "epoch": 1.81873417721519,
+      "grad_norm": 0.35423869725193285,
+      "learning_rate": 3.203403386721272e-05,
+      "loss": 0.5623,
+      "step": 449
+    },
+    {
+      "epoch": 1.8227848101265822,
+      "grad_norm": 0.36797308832882974,
+      "learning_rate": 3.1988651128108245e-05,
+      "loss": 0.5602,
+      "step": 450
+    },
+    {
+      "epoch": 1.8268354430379747,
+      "grad_norm": 0.33327581409332585,
+      "learning_rate": 3.194317183405573e-05,
+      "loss": 0.5723,
+      "step": 451
+    },
+    {
+      "epoch": 1.8308860759493673,
+      "grad_norm": 0.44394652419445785,
+      "learning_rate": 3.189759635133914e-05,
+      "loss": 0.5572,
+      "step": 452
+    },
+    {
+      "epoch": 1.8349367088607595,
+      "grad_norm": 0.3232571787282512,
+      "learning_rate": 3.185192504701718e-05,
+      "loss": 0.5623,
+      "step": 453
+    },
+    {
+      "epoch": 1.8389873417721518,
+      "grad_norm": 3.3610588528370706,
+      "learning_rate": 3.1806158288920234e-05,
+      "loss": 0.57,
+      "step": 454
+    },
+    {
+      "epoch": 1.8430379746835444,
+      "grad_norm": 0.4992103842263646,
+      "learning_rate": 3.1760296445647477e-05,
+      "loss": 0.5604,
+      "step": 455
+    },
+    {
+      "epoch": 1.8470886075949369,
+      "grad_norm": 0.4709894423752063,
+      "learning_rate": 3.1714339886563896e-05,
+      "loss": 0.5766,
+      "step": 456
+    },
+    {
+      "epoch": 1.8511392405063292,
+      "grad_norm": 0.34260783354459534,
+      "learning_rate": 3.166828898179731e-05,
+      "loss": 0.5672,
+      "step": 457
+    },
+    {
+      "epoch": 1.8551898734177215,
+      "grad_norm": 0.39487640738637164,
+      "learning_rate": 3.162214410223536e-05,
+      "loss": 0.5648,
+      "step": 458
+    },
+    {
+      "epoch": 1.859240506329114,
+      "grad_norm": 0.3624443979671359,
+      "learning_rate": 3.157590561952257e-05,
+      "loss": 0.5516,
+      "step": 459
+    },
+    {
+      "epoch": 1.8632911392405065,
+      "grad_norm": 0.3750462481095353,
+      "learning_rate": 3.152957390605732e-05,
+      "loss": 0.5503,
+      "step": 460
+    },
+    {
+      "epoch": 1.8673417721518988,
+      "grad_norm": 0.340821081254788,
+      "learning_rate": 3.148314933498886e-05,
+      "loss": 0.5536,
+      "step": 461
+    },
+    {
+      "epoch": 1.871392405063291,
+      "grad_norm": 0.39488073138132074,
+      "learning_rate": 3.143663228021431e-05,
+      "loss": 0.5609,
+      "step": 462
+    },
+    {
+      "epoch": 1.8754430379746836,
+      "grad_norm": 0.3462203871520774,
+      "learning_rate": 3.1390023116375624e-05,
+      "loss": 0.5642,
+      "step": 463
+    },
+    {
+      "epoch": 1.879493670886076,
+      "grad_norm": 0.35969510100452035,
+      "learning_rate": 3.134332221885661e-05,
+      "loss": 0.5689,
+      "step": 464
+    },
+    {
+      "epoch": 1.8835443037974684,
+      "grad_norm": 0.32400275553461594,
+      "learning_rate": 3.129652996377987e-05,
+      "loss": 0.5698,
+      "step": 465
+    },
+    {
+      "epoch": 1.8875949367088607,
+      "grad_norm": 0.3299980169417157,
+      "learning_rate": 3.12496467280038e-05,
+      "loss": 0.5435,
+      "step": 466
+    },
+    {
+      "epoch": 1.8916455696202532,
+      "grad_norm": 0.8165264480798752,
+      "learning_rate": 3.120267288911952e-05,
+      "loss": 0.5625,
+      "step": 467
+    },
+    {
+      "epoch": 1.8956962025316457,
+      "grad_norm": 0.35079245220896893,
+      "learning_rate": 3.11556088254479e-05,
+      "loss": 0.5603,
+      "step": 468
+    },
+    {
+      "epoch": 1.899746835443038,
+      "grad_norm": 0.34420024318702247,
+      "learning_rate": 3.11084549160364e-05,
+      "loss": 0.5667,
+      "step": 469
+    },
+    {
+      "epoch": 1.9037974683544303,
+      "grad_norm": 0.34075618768398924,
+      "learning_rate": 3.106121154065615e-05,
+      "loss": 0.5595,
+      "step": 470
+    },
+    {
+      "epoch": 1.9078481012658228,
+      "grad_norm": 0.3447162955753632,
+      "learning_rate": 3.1013879079798805e-05,
+      "loss": 0.5526,
+      "step": 471
+    },
+    {
+      "epoch": 1.9118987341772153,
+      "grad_norm": 0.3391742149410081,
+      "learning_rate": 3.096645791467348e-05,
+      "loss": 0.5504,
+      "step": 472
+    },
+    {
+      "epoch": 1.9159493670886076,
+      "grad_norm": 0.31997510378717275,
+      "learning_rate": 3.091894842720373e-05,
+      "loss": 0.5626,
+      "step": 473
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.3093428638013072,
+      "learning_rate": 3.0871351000024425e-05,
+      "loss": 0.5409,
+      "step": 474
+    },
+    {
+      "epoch": 1.9240506329113924,
+      "grad_norm": 0.33940697024958016,
+      "learning_rate": 3.0823666016478716e-05,
+      "loss": 0.5628,
+      "step": 475
+    },
+    {
+      "epoch": 1.928101265822785,
+      "grad_norm": 0.4127001032111777,
+      "learning_rate": 3.0775893860614896e-05,
+      "loss": 0.5541,
+      "step": 476
+    },
+    {
+      "epoch": 1.9321518987341773,
+      "grad_norm": 0.3820692585285003,
+      "learning_rate": 3.0728034917183336e-05,
+      "loss": 0.5628,
+      "step": 477
+    },
+    {
+      "epoch": 1.9362025316455695,
+      "grad_norm": 0.32225546932061705,
+      "learning_rate": 3.06800895716334e-05,
+      "loss": 0.5607,
+      "step": 478
+    },
+    {
+      "epoch": 1.940253164556962,
+      "grad_norm": 2.2692288071467908,
+      "learning_rate": 3.063205821011029e-05,
+      "loss": 0.5935,
+      "step": 479
+    },
+    {
+      "epoch": 1.9443037974683546,
+      "grad_norm": 0.4410649630094251,
+      "learning_rate": 3.0583941219452016e-05,
+      "loss": 0.5653,
+      "step": 480
+    },
+    {
+      "epoch": 1.9483544303797469,
+      "grad_norm": 0.46639735542820754,
+      "learning_rate": 3.053573898718618e-05,
+      "loss": 0.559,
+      "step": 481
+    },
+    {
+      "epoch": 1.9524050632911392,
+      "grad_norm": 0.35926762322705735,
+      "learning_rate": 3.0487451901526956e-05,
+      "loss": 0.5666,
+      "step": 482
+    },
+    {
+      "epoch": 1.9564556962025317,
+      "grad_norm": 2.115485989736596,
+      "learning_rate": 3.0439080351371875e-05,
+      "loss": 0.5873,
+      "step": 483
+    },
+    {
+      "epoch": 1.9605063291139242,
+      "grad_norm": 0.3986719039188294,
+      "learning_rate": 3.0390624726298764e-05,
+      "loss": 0.55,
+      "step": 484
+    },
+    {
+      "epoch": 1.9645569620253165,
+      "grad_norm": 0.3157152217267981,
+      "learning_rate": 3.034208541656255e-05,
+      "loss": 0.5573,
+      "step": 485
+    },
+    {
+      "epoch": 1.9686075949367088,
+      "grad_norm": 0.3234659189044682,
+      "learning_rate": 3.029346281309218e-05,
+      "loss": 0.5696,
+      "step": 486
+    },
+    {
+      "epoch": 1.9726582278481013,
+      "grad_norm": 0.31970148095743267,
+      "learning_rate": 3.0244757307487415e-05,
+      "loss": 0.5485,
+      "step": 487
+    },
+    {
+      "epoch": 1.9767088607594938,
+      "grad_norm": 0.39225704139349443,
+      "learning_rate": 3.019596929201569e-05,
+      "loss": 0.5607,
+      "step": 488
+    },
+    {
+      "epoch": 1.980759493670886,
+      "grad_norm": 0.34114045478701727,
+      "learning_rate": 3.0147099159608985e-05,
+      "loss": 0.5515,
+      "step": 489
+    },
+    {
+      "epoch": 1.9848101265822784,
+      "grad_norm": 0.3440368714619485,
+      "learning_rate": 3.0098147303860616e-05,
+      "loss": 0.558,
+      "step": 490
+    },
+    {
+      "epoch": 1.988860759493671,
+      "grad_norm": 0.3144271573526958,
+      "learning_rate": 3.0049114119022117e-05,
+      "loss": 0.5671,
+      "step": 491
+    },
+    {
+      "epoch": 1.9929113924050634,
+      "grad_norm": 0.34969493835726656,
+      "learning_rate": 3.0000000000000004e-05,
+      "loss": 0.5598,
+      "step": 492
+    },
+    {
+      "epoch": 1.9969620253164557,
+      "grad_norm": 0.30268473749099023,
+      "learning_rate": 2.995080534235264e-05,
+      "loss": 0.546,
+      "step": 493
+    },
+    {
+      "epoch": 2.001012658227848,
+      "grad_norm": 0.3484502025777852,
+      "learning_rate": 2.9901530542287044e-05,
+      "loss": 0.5405,
+      "step": 494
+    },
+    {
+      "epoch": 2.0050632911392405,
+      "grad_norm": 1.4873207698996302,
+      "learning_rate": 2.9852175996655676e-05,
+      "loss": 0.507,
+      "step": 495
+    },
+    {
+      "epoch": 2.009113924050633,
+      "grad_norm": 0.40220714909115707,
+      "learning_rate": 2.980274210295326e-05,
+      "loss": 0.5081,
+      "step": 496
+    },
+    {
+      "epoch": 2.013164556962025,
+      "grad_norm": 0.3532274420639015,
+      "learning_rate": 2.9753229259313578e-05,
+      "loss": 0.5043,
+      "step": 497
+    },
+    {
+      "epoch": 2.0172151898734176,
+      "grad_norm": 0.44227303366137194,
+      "learning_rate": 2.9703637864506274e-05,
+      "loss": 0.4989,
+      "step": 498
+    },
+    {
+      "epoch": 2.02126582278481,
+      "grad_norm": 0.3652837442990965,
+      "learning_rate": 2.965396831793362e-05,
+      "loss": 0.5109,
+      "step": 499
+    },
+    {
+      "epoch": 2.0253164556962027,
+      "grad_norm": 0.4143855584619444,
+      "learning_rate": 2.9604221019627316e-05,
+      "loss": 0.4936,
+      "step": 500
+    },
+    {
+      "epoch": 2.0293670886075947,
+      "grad_norm": 0.4026586074509516,
+      "learning_rate": 2.955439637024526e-05,
+      "loss": 0.4974,
+      "step": 501
+    },
+    {
+      "epoch": 2.0334177215189873,
+      "grad_norm": 0.436739832158227,
+      "learning_rate": 2.9504494771068334e-05,
+      "loss": 0.5385,
+      "step": 502
+    },
+    {
+      "epoch": 2.0374683544303798,
+      "grad_norm": 0.430994480037579,
+      "learning_rate": 2.9454516623997156e-05,
+      "loss": 0.4891,
+      "step": 503
+    },
+    {
+      "epoch": 2.0415189873417723,
+      "grad_norm": 0.35670262635482064,
+      "learning_rate": 2.9404462331548847e-05,
+      "loss": 0.4953,
+      "step": 504
+    },
+    {
+      "epoch": 2.0455696202531644,
+      "grad_norm": 0.4249675358813206,
+      "learning_rate": 2.93543322968538e-05,
+      "loss": 0.5092,
+      "step": 505
+    },
+    {
+      "epoch": 2.049620253164557,
+      "grad_norm": 0.4366258752903941,
+      "learning_rate": 2.9304126923652428e-05,
+      "loss": 0.5019,
+      "step": 506
+    },
+    {
+      "epoch": 2.0536708860759494,
+      "grad_norm": 0.32233437653012087,
+      "learning_rate": 2.9253846616291896e-05,
+      "loss": 0.4978,
+      "step": 507
+    },
+    {
+      "epoch": 2.057721518987342,
+      "grad_norm": 0.3736314100503556,
+      "learning_rate": 2.9203491779722896e-05,
+      "loss": 0.4907,
+      "step": 508
+    },
+    {
+      "epoch": 2.061772151898734,
+      "grad_norm": 0.30185447813292304,
+      "learning_rate": 2.9153062819496357e-05,
+      "loss": 0.494,
+      "step": 509
+    },
+    {
+      "epoch": 2.0658227848101265,
+      "grad_norm": 0.3419690241400411,
+      "learning_rate": 2.9102560141760178e-05,
+      "loss": 0.507,
+      "step": 510
+    },
+    {
+      "epoch": 2.069873417721519,
+      "grad_norm": 0.37438963953331,
+      "learning_rate": 2.9051984153256004e-05,
+      "loss": 0.4962,
+      "step": 511
+    },
+    {
+      "epoch": 2.0739240506329115,
+      "grad_norm": 0.30120141608242595,
+      "learning_rate": 2.900133526131588e-05,
+      "loss": 0.4926,
+      "step": 512
+    },
+    {
+      "epoch": 2.0779746835443036,
+      "grad_norm": 0.43734054952859197,
+      "learning_rate": 2.8950613873859025e-05,
+      "loss": 0.4933,
+      "step": 513
+    },
+    {
+      "epoch": 2.082025316455696,
+      "grad_norm": 0.5779350471884324,
+      "learning_rate": 2.8899820399388515e-05,
+      "loss": 0.5225,
+      "step": 514
+    },
+    {
+      "epoch": 2.0860759493670886,
+      "grad_norm": 0.31816128956536077,
+      "learning_rate": 2.8848955246988012e-05,
+      "loss": 0.4912,
+      "step": 515
+    },
+    {
+      "epoch": 2.090126582278481,
+      "grad_norm": 0.4639729447304105,
+      "learning_rate": 2.879801882631847e-05,
+      "loss": 0.5018,
+      "step": 516
+    },
+    {
+      "epoch": 2.094177215189873,
+      "grad_norm": 0.31814031110958296,
+      "learning_rate": 2.8747011547614808e-05,
+      "loss": 0.4955,
+      "step": 517
+    },
+    {
+      "epoch": 2.0982278481012657,
+      "grad_norm": 0.35676078383279114,
+      "learning_rate": 2.8695933821682635e-05,
+      "loss": 0.4936,
+      "step": 518
+    },
+    {
+      "epoch": 2.1022784810126582,
+      "grad_norm": 0.33795372732418694,
+      "learning_rate": 2.864478605989494e-05,
+      "loss": 0.5058,
+      "step": 519
+    },
+    {
+      "epoch": 2.1063291139240508,
+      "grad_norm": 0.3136195521986097,
+      "learning_rate": 2.8593568674188765e-05,
+      "loss": 0.5008,
+      "step": 520
+    },
+    {
+      "epoch": 2.110379746835443,
+      "grad_norm": 0.32948359967853336,
+      "learning_rate": 2.8542282077061892e-05,
+      "loss": 0.5076,
+      "step": 521
+    },
+    {
+      "epoch": 2.1144303797468353,
+      "grad_norm": 0.2943158290342833,
+      "learning_rate": 2.8490926681569523e-05,
+      "loss": 0.5017,
+      "step": 522
+    },
+    {
+      "epoch": 2.118481012658228,
+      "grad_norm": 0.3419292815317538,
+      "learning_rate": 2.8439502901320956e-05,
+      "loss": 0.5027,
+      "step": 523
+    },
+    {
+      "epoch": 2.1225316455696204,
+      "grad_norm": 0.38699666975291763,
+      "learning_rate": 2.8388011150476237e-05,
+      "loss": 0.5211,
+      "step": 524
+    },
+    {
+      "epoch": 2.1265822784810124,
+      "grad_norm": 0.3606498471928407,
+      "learning_rate": 2.8336451843742866e-05,
+      "loss": 0.4947,
+      "step": 525
+    },
+    {
+      "epoch": 2.130632911392405,
+      "grad_norm": 0.37445162377594154,
+      "learning_rate": 2.8284825396372387e-05,
+      "loss": 0.4921,
+      "step": 526
+    },
+    {
+      "epoch": 2.1346835443037975,
+      "grad_norm": 0.4381290668895678,
+      "learning_rate": 2.8233132224157132e-05,
+      "loss": 0.4933,
+      "step": 527
+    },
+    {
+      "epoch": 2.13873417721519,
+      "grad_norm": 0.325646035461865,
+      "learning_rate": 2.8181372743426805e-05,
+      "loss": 0.4998,
+      "step": 528
+    },
+    {
+      "epoch": 2.1427848101265825,
+      "grad_norm": 0.38314613071181247,
+      "learning_rate": 2.8129547371045128e-05,
+      "loss": 0.514,
+      "step": 529
+    },
+    {
+      "epoch": 2.1468354430379746,
+      "grad_norm": 0.37175009840703477,
+      "learning_rate": 2.8077656524406534e-05,
+      "loss": 0.495,
+      "step": 530
+    },
+    {
+      "epoch": 2.150886075949367,
+      "grad_norm": 0.318613371349883,
+      "learning_rate": 2.802570062143278e-05,
+      "loss": 0.4764,
+      "step": 531
+    },
+    {
+      "epoch": 2.1549367088607596,
+      "grad_norm": 0.37430365465350396,
+      "learning_rate": 2.7973680080569555e-05,
+      "loss": 0.4969,
+      "step": 532
+    },
+    {
+      "epoch": 2.1589873417721517,
+      "grad_norm": 0.3288590390608046,
+      "learning_rate": 2.792159532078314e-05,
+      "loss": 0.4943,
+      "step": 533
+    },
+    {
+      "epoch": 2.163037974683544,
+      "grad_norm": 0.4116723554692981,
+      "learning_rate": 2.7869446761557033e-05,
+      "loss": 0.5059,
+      "step": 534
+    },
+    {
+      "epoch": 2.1670886075949367,
+      "grad_norm": 0.3011294096975146,
+      "learning_rate": 2.781723482288857e-05,
+      "loss": 0.4948,
+      "step": 535
+    },
+    {
+      "epoch": 2.1711392405063292,
+      "grad_norm": 0.3673945172248939,
+      "learning_rate": 2.7764959925285517e-05,
+      "loss": 0.507,
+      "step": 536
+    },
+    {
+      "epoch": 2.1751898734177217,
+      "grad_norm": 0.33594710331701655,
+      "learning_rate": 2.771262248976272e-05,
+      "loss": 0.4992,
+      "step": 537
+    },
+    {
+      "epoch": 2.179240506329114,
+      "grad_norm": 0.3364005506863424,
+      "learning_rate": 2.7660222937838677e-05,
+      "loss": 0.4985,
+      "step": 538
+    },
+    {
+      "epoch": 2.1832911392405063,
+      "grad_norm": 0.31265479878526675,
+      "learning_rate": 2.7607761691532186e-05,
+      "loss": 0.4898,
+      "step": 539
+    },
+    {
+      "epoch": 2.187341772151899,
+      "grad_norm": 0.380973445249096,
+      "learning_rate": 2.7555239173358916e-05,
+      "loss": 0.4984,
+      "step": 540
+    },
+    {
+      "epoch": 2.191392405063291,
+      "grad_norm": 0.29746811536421247,
+      "learning_rate": 2.7502655806328e-05,
+      "loss": 0.4846,
+      "step": 541
+    },
+    {
+      "epoch": 2.1954430379746834,
+      "grad_norm": 0.3820806343519624,
+      "learning_rate": 2.7450012013938648e-05,
+      "loss": 0.4942,
+      "step": 542
+    },
+    {
+      "epoch": 2.199493670886076,
+      "grad_norm": 0.2872910660797289,
+      "learning_rate": 2.739730822017673e-05,
+      "loss": 0.484,
+      "step": 543
+    },
+    {
+      "epoch": 2.2035443037974685,
+      "grad_norm": 0.3835829770319144,
+      "learning_rate": 2.7344544849511355e-05,
+      "loss": 0.5016,
+      "step": 544
+    },
+    {
+      "epoch": 2.207594936708861,
+      "grad_norm": 0.32449001917013826,
+      "learning_rate": 2.7291722326891456e-05,
+      "loss": 0.4968,
+      "step": 545
+    },
+    {
+      "epoch": 2.211645569620253,
+      "grad_norm": 0.31219934790562875,
+      "learning_rate": 2.723884107774236e-05,
+      "loss": 0.4936,
+      "step": 546
+    },
+    {
+      "epoch": 2.2156962025316456,
+      "grad_norm": 0.2905518883190544,
+      "learning_rate": 2.718590152796239e-05,
+      "loss": 0.5049,
+      "step": 547
+    },
+    {
+      "epoch": 2.219746835443038,
+      "grad_norm": 0.328824334261568,
+      "learning_rate": 2.71329041039194e-05,
+      "loss": 0.501,
+      "step": 548
+    },
+    {
+      "epoch": 2.22379746835443,
+      "grad_norm": 0.3096866003553902,
+      "learning_rate": 2.7079849232447357e-05,
+      "loss": 0.4979,
+      "step": 549
+    },
+    {
+      "epoch": 2.2278481012658227,
+      "grad_norm": 0.3139396251891686,
+      "learning_rate": 2.7026737340842895e-05,
+      "loss": 0.4927,
+      "step": 550
+    },
+    {
+      "epoch": 2.231898734177215,
+      "grad_norm": 0.3447634462435707,
+      "learning_rate": 2.697356885686189e-05,
+      "loss": 0.5025,
+      "step": 551
+    },
+    {
+      "epoch": 2.2359493670886077,
+      "grad_norm": 0.3160614056756411,
+      "learning_rate": 2.6920344208716014e-05,
+      "loss": 0.5016,
+      "step": 552
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 0.2860361377348481,
+      "learning_rate": 2.6867063825069252e-05,
+      "loss": 0.4986,
+      "step": 553
+    },
+    {
+      "epoch": 2.2440506329113923,
+      "grad_norm": 0.32742844108305985,
+      "learning_rate": 2.6813728135034494e-05,
+      "loss": 0.5053,
+      "step": 554
+    },
+    {
+      "epoch": 2.248101265822785,
+      "grad_norm": 0.29147805405715066,
+      "learning_rate": 2.6760337568170056e-05,
+      "loss": 0.5088,
+      "step": 555
+    },
+    {
+      "epoch": 2.2521518987341773,
+      "grad_norm": 0.30923210394253975,
+      "learning_rate": 2.6706892554476226e-05,
+      "loss": 0.5019,
+      "step": 556
+    },
+    {
+      "epoch": 2.2562025316455694,
+      "grad_norm": 0.30682935405838035,
+      "learning_rate": 2.6653393524391795e-05,
+      "loss": 0.4961,
+      "step": 557
+    },
+    {
+      "epoch": 2.260253164556962,
+      "grad_norm": 0.3368471105460133,
+      "learning_rate": 2.6599840908790592e-05,
+      "loss": 0.4999,
+      "step": 558
+    },
+    {
+      "epoch": 2.2643037974683544,
+      "grad_norm": 0.3825410903368047,
+      "learning_rate": 2.6546235138978028e-05,
+      "loss": 0.4963,
+      "step": 559
+    },
+    {
+      "epoch": 2.268354430379747,
+      "grad_norm": 0.28949643451685286,
+      "learning_rate": 2.6492576646687597e-05,
+      "loss": 0.4972,
+      "step": 560
+    },
+    {
+      "epoch": 2.2724050632911394,
+      "grad_norm": 0.345125369116229,
+      "learning_rate": 2.6438865864077425e-05,
+      "loss": 0.5004,
+      "step": 561
+    },
+    {
+      "epoch": 2.2764556962025315,
+      "grad_norm": 0.3308130345414538,
+      "learning_rate": 2.6385103223726766e-05,
+      "loss": 0.4989,
+      "step": 562
+    },
+    {
+      "epoch": 2.280506329113924,
+      "grad_norm": 0.3180541118296476,
+      "learning_rate": 2.6331289158632537e-05,
+      "loss": 0.5043,
+      "step": 563
+    },
+    {
+      "epoch": 2.2845569620253166,
+      "grad_norm": 0.40368993068673553,
+      "learning_rate": 2.6277424102205817e-05,
+      "loss": 0.506,
+      "step": 564
+    },
+    {
+      "epoch": 2.2886075949367086,
+      "grad_norm": 0.3523338313710627,
+      "learning_rate": 2.6223508488268374e-05,
+      "loss": 0.5059,
+      "step": 565
+    },
+    {
+      "epoch": 2.292658227848101,
+      "grad_norm": 0.4072791973999593,
+      "learning_rate": 2.6169542751049148e-05,
+      "loss": 0.5145,
+      "step": 566
+    },
+    {
+      "epoch": 2.2967088607594937,
+      "grad_norm": 2.073556787290978,
+      "learning_rate": 2.6115527325180754e-05,
+      "loss": 0.5132,
+      "step": 567
+    },
+    {
+      "epoch": 2.300759493670886,
+      "grad_norm": 0.5425198074619388,
+      "learning_rate": 2.606146264569603e-05,
+      "loss": 0.5015,
+      "step": 568
+    },
+    {
+      "epoch": 2.3048101265822787,
+      "grad_norm": 0.4973742821999527,
+      "learning_rate": 2.6007349148024447e-05,
+      "loss": 0.4947,
+      "step": 569
+    },
+    {
+      "epoch": 2.3088607594936708,
+      "grad_norm": 0.33107302005686706,
+      "learning_rate": 2.5953187267988694e-05,
+      "loss": 0.4973,
+      "step": 570
+    },
+    {
+      "epoch": 2.3129113924050633,
+      "grad_norm": 0.4835292101792158,
+      "learning_rate": 2.5898977441801097e-05,
+      "loss": 0.4982,
+      "step": 571
+    },
+    {
+      "epoch": 2.316962025316456,
+      "grad_norm": 0.34368718646822033,
+      "learning_rate": 2.584472010606015e-05,
+      "loss": 0.4962,
+      "step": 572
+    },
+    {
+      "epoch": 2.321012658227848,
+      "grad_norm": 0.36073610333663036,
+      "learning_rate": 2.5790415697746976e-05,
+      "loss": 0.4913,
+      "step": 573
+    },
+    {
+      "epoch": 2.3250632911392404,
+      "grad_norm": 0.669877181960211,
+      "learning_rate": 2.5736064654221808e-05,
+      "loss": 0.4997,
+      "step": 574
+    },
+    {
+      "epoch": 2.329113924050633,
+      "grad_norm": 0.3664776516607763,
+      "learning_rate": 2.568166741322048e-05,
+      "loss": 0.5012,
+      "step": 575
+    },
+    {
+      "epoch": 2.3331645569620254,
+      "grad_norm": 0.4089147171935835,
+      "learning_rate": 2.56272244128509e-05,
+      "loss": 0.4999,
+      "step": 576
+    },
+    {
+      "epoch": 2.337215189873418,
+      "grad_norm": 0.3872512694011354,
+      "learning_rate": 2.55727360915895e-05,
+      "loss": 0.5036,
+      "step": 577
+    },
+    {
+      "epoch": 2.34126582278481,
+      "grad_norm": 0.3836992801754127,
+      "learning_rate": 2.5518202888277734e-05,
+      "loss": 0.5062,
+      "step": 578
+    },
+    {
+      "epoch": 2.3453164556962025,
+      "grad_norm": 0.43819732960596824,
+      "learning_rate": 2.5463625242118523e-05,
+      "loss": 0.5013,
+      "step": 579
+    },
+    {
+      "epoch": 2.349367088607595,
+      "grad_norm": 0.36794285867016724,
+      "learning_rate": 2.5409003592672723e-05,
+      "loss": 0.5026,
+      "step": 580
+    },
+    {
+      "epoch": 2.353417721518987,
+      "grad_norm": 0.3306805051540774,
+      "learning_rate": 2.535433837985559e-05,
+      "loss": 0.4981,
+      "step": 581
+    },
+    {
+      "epoch": 2.3574683544303796,
+      "grad_norm": 0.31296153244308256,
+      "learning_rate": 2.529963004393324e-05,
+      "loss": 0.4971,
+      "step": 582
+    },
+    {
+      "epoch": 2.361518987341772,
+      "grad_norm": 0.3857752965363398,
+      "learning_rate": 2.524487902551908e-05,
+      "loss": 0.5102,
+      "step": 583
+    },
+    {
+      "epoch": 2.3655696202531646,
+      "grad_norm": 0.32621895054882744,
+      "learning_rate": 2.519008576557029e-05,
+      "loss": 0.497,
+      "step": 584
+    },
+    {
+      "epoch": 2.369620253164557,
+      "grad_norm": 0.3818750345362715,
+      "learning_rate": 2.5135250705384254e-05,
+      "loss": 0.4827,
+      "step": 585
+    },
+    {
+      "epoch": 2.3736708860759492,
+      "grad_norm": 0.3172280412928033,
+      "learning_rate": 2.5080374286595007e-05,
+      "loss": 0.4954,
+      "step": 586
+    },
+    {
+      "epoch": 2.3777215189873417,
+      "grad_norm": 0.3671330772364744,
+      "learning_rate": 2.5025456951169677e-05,
+      "loss": 0.4861,
+      "step": 587
+    },
+    {
+      "epoch": 2.3817721518987343,
+      "grad_norm": 0.3367445419918956,
+      "learning_rate": 2.4970499141404942e-05,
+      "loss": 0.4966,
+      "step": 588
+    },
+    {
+      "epoch": 2.3858227848101268,
+      "grad_norm": 0.31573361232457375,
+      "learning_rate": 2.491550129992345e-05,
+      "loss": 0.494,
+      "step": 589
+    },
+    {
+      "epoch": 2.389873417721519,
+      "grad_norm": 0.3103681170451071,
+      "learning_rate": 2.486046386967024e-05,
+      "loss": 0.4976,
+      "step": 590
+    },
+    {
+      "epoch": 2.3939240506329114,
+      "grad_norm": 0.31016571776308294,
+      "learning_rate": 2.4805387293909214e-05,
+      "loss": 0.5053,
+      "step": 591
+    },
+    {
+      "epoch": 2.397974683544304,
+      "grad_norm": 0.30358928434877325,
+      "learning_rate": 2.4750272016219552e-05,
+      "loss": 0.4987,
+      "step": 592
+    },
+    {
+      "epoch": 2.4020253164556964,
+      "grad_norm": 0.33309293268792084,
+      "learning_rate": 2.4695118480492114e-05,
+      "loss": 0.501,
+      "step": 593
+    },
+    {
+      "epoch": 2.4060759493670885,
+      "grad_norm": 0.2746742750488384,
+      "learning_rate": 2.4639927130925898e-05,
+      "loss": 0.5062,
+      "step": 594
+    },
+    {
+      "epoch": 2.410126582278481,
+      "grad_norm": 0.33580086506298534,
+      "learning_rate": 2.458469841202444e-05,
+      "loss": 0.4918,
+      "step": 595
+    },
+    {
+      "epoch": 2.4141772151898735,
+      "grad_norm": 0.31499929555819944,
+      "learning_rate": 2.452943276859226e-05,
+      "loss": 0.5103,
+      "step": 596
+    },
+    {
+      "epoch": 2.418227848101266,
+      "grad_norm": 0.3034272623739805,
+      "learning_rate": 2.447413064573125e-05,
+      "loss": 0.5039,
+      "step": 597
+    },
+    {
+      "epoch": 2.422278481012658,
+      "grad_norm": 0.3188340376098387,
+      "learning_rate": 2.4418792488837095e-05,
+      "loss": 0.503,
+      "step": 598
+    },
+    {
+      "epoch": 2.4263291139240506,
+      "grad_norm": 0.28776091080565863,
+      "learning_rate": 2.4363418743595713e-05,
+      "loss": 0.4964,
+      "step": 599
+    },
+    {
+      "epoch": 2.430379746835443,
+      "grad_norm": 0.37094203278098015,
+      "learning_rate": 2.430800985597963e-05,
+      "loss": 0.5058,
+      "step": 600
+    },
+    {
+      "epoch": 2.4344303797468356,
+      "grad_norm": 0.33792928707950287,
+      "learning_rate": 2.4252566272244415e-05,
+      "loss": 0.4903,
+      "step": 601
+    },
+    {
+      "epoch": 2.4384810126582277,
+      "grad_norm": 0.2790654877820239,
+      "learning_rate": 2.4197088438925063e-05,
+      "loss": 0.5121,
+      "step": 602
+    },
+    {
+      "epoch": 2.44253164556962,
+      "grad_norm": 0.313989515823712,
+      "learning_rate": 2.4141576802832417e-05,
+      "loss": 0.5066,
+      "step": 603
+    },
+    {
+      "epoch": 2.4465822784810127,
+      "grad_norm": 0.29232690863680283,
+      "learning_rate": 2.408603181104957e-05,
+      "loss": 0.5048,
+      "step": 604
+    },
+    {
+      "epoch": 2.4506329113924052,
+      "grad_norm": 0.3806345200126955,
+      "learning_rate": 2.4030453910928245e-05,
+      "loss": 0.5004,
+      "step": 605
+    },
+    {
+      "epoch": 2.4546835443037973,
+      "grad_norm": 0.29608099968882123,
+      "learning_rate": 2.397484355008521e-05,
+      "loss": 0.5009,
+      "step": 606
+    },
+    {
+      "epoch": 2.45873417721519,
+      "grad_norm": 0.32668692410133393,
+      "learning_rate": 2.3919201176398662e-05,
+      "loss": 0.5085,
+      "step": 607
+    },
+    {
+      "epoch": 2.4627848101265823,
+      "grad_norm": 0.32108839465597083,
+      "learning_rate": 2.3863527238004633e-05,
+      "loss": 0.5065,
+      "step": 608
+    },
+    {
+      "epoch": 2.466835443037975,
+      "grad_norm": 0.30435042135475443,
+      "learning_rate": 2.380782218329337e-05,
+      "loss": 0.4948,
+      "step": 609
+    },
+    {
+      "epoch": 2.470886075949367,
+      "grad_norm": 0.33754281874893854,
+      "learning_rate": 2.3752086460905725e-05,
+      "loss": 0.5035,
+      "step": 610
+    },
+    {
+      "epoch": 2.4749367088607594,
+      "grad_norm": 0.2947212975916069,
+      "learning_rate": 2.3696320519729544e-05,
+      "loss": 0.4999,
+      "step": 611
+    },
+    {
+      "epoch": 2.478987341772152,
+      "grad_norm": 0.290873906688026,
+      "learning_rate": 2.3640524808896045e-05,
+      "loss": 0.4961,
+      "step": 612
+    },
+    {
+      "epoch": 2.4830379746835445,
+      "grad_norm": 0.3308557046164479,
+      "learning_rate": 2.3584699777776222e-05,
+      "loss": 0.498,
+      "step": 613
+    },
+    {
+      "epoch": 2.4870886075949366,
+      "grad_norm": 0.3290540550211233,
+      "learning_rate": 2.3528845875977195e-05,
+      "loss": 0.4939,
+      "step": 614
+    },
+    {
+      "epoch": 2.491139240506329,
+      "grad_norm": 0.30904414539378894,
+      "learning_rate": 2.3472963553338614e-05,
+      "loss": 0.5016,
+      "step": 615
+    },
+    {
+      "epoch": 2.4951898734177216,
+      "grad_norm": 0.35390183484680765,
+      "learning_rate": 2.341705325992901e-05,
+      "loss": 0.5031,
+      "step": 616
+    },
+    {
+      "epoch": 2.499240506329114,
+      "grad_norm": 0.3225264203394793,
+      "learning_rate": 2.336111544604222e-05,
+      "loss": 0.4906,
+      "step": 617
+    },
+    {
+      "epoch": 2.503291139240506,
+      "grad_norm": 0.3532517718903853,
+      "learning_rate": 2.33051505621937e-05,
+      "loss": 0.4977,
+      "step": 618
+    },
+    {
+      "epoch": 2.5073417721518987,
+      "grad_norm": 0.2814031896161318,
+      "learning_rate": 2.324915905911693e-05,
+      "loss": 0.4993,
+      "step": 619
+    },
+    {
+      "epoch": 2.511392405063291,
+      "grad_norm": 0.346396509952919,
+      "learning_rate": 2.319314138775977e-05,
+      "loss": 0.5052,
+      "step": 620
+    },
+    {
+      "epoch": 2.5154430379746833,
+      "grad_norm": 0.29302444044292386,
+      "learning_rate": 2.3137097999280856e-05,
+      "loss": 0.4931,
+      "step": 621
+    },
+    {
+      "epoch": 2.519493670886076,
+      "grad_norm": 0.3168028047969663,
+      "learning_rate": 2.308102934504593e-05,
+      "loss": 0.4841,
+      "step": 622
+    },
+    {
+      "epoch": 2.5235443037974683,
+      "grad_norm": 0.2803635554859486,
+      "learning_rate": 2.3024935876624222e-05,
+      "loss": 0.4938,
+      "step": 623
+    },
+    {
+      "epoch": 2.527594936708861,
+      "grad_norm": 0.28190267394403196,
+      "learning_rate": 2.2968818045784813e-05,
+      "loss": 0.4993,
+      "step": 624
+    },
+    {
+      "epoch": 2.5316455696202533,
+      "grad_norm": 0.2740188951190504,
+      "learning_rate": 2.2912676304493006e-05,
+      "loss": 0.4877,
+      "step": 625
+    },
+    {
+      "epoch": 2.5356962025316454,
+      "grad_norm": 0.29243733439391145,
+      "learning_rate": 2.2856511104906668e-05,
+      "loss": 0.4971,
+      "step": 626
+    },
+    {
+      "epoch": 2.539746835443038,
+      "grad_norm": 0.3010886115452139,
+      "learning_rate": 2.2800322899372586e-05,
+      "loss": 0.4893,
+      "step": 627
+    },
+    {
+      "epoch": 2.5437974683544304,
+      "grad_norm": 0.27278058877598765,
+      "learning_rate": 2.2744112140422844e-05,
+      "loss": 0.5038,
+      "step": 628
+    },
+    {
+      "epoch": 2.547848101265823,
+      "grad_norm": 0.30210146064380594,
+      "learning_rate": 2.2687879280771177e-05,
+      "loss": 0.5039,
+      "step": 629
+    },
+    {
+      "epoch": 2.5518987341772155,
+      "grad_norm": 0.2769365749083288,
+      "learning_rate": 2.26316247733093e-05,
+      "loss": 0.4908,
+      "step": 630
+    },
+    {
+      "epoch": 2.5559493670886075,
+      "grad_norm": 0.28596045656926206,
+      "learning_rate": 2.257534907110328e-05,
+      "loss": 0.4996,
+      "step": 631
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 0.29914389577690503,
+      "learning_rate": 2.2519052627389882e-05,
+      "loss": 0.5035,
+      "step": 632
+    },
+    {
+      "epoch": 2.5640506329113926,
+      "grad_norm": 0.2929550134414236,
+      "learning_rate": 2.246273589557294e-05,
+      "loss": 0.492,
+      "step": 633
+    },
+    {
+      "epoch": 2.5681012658227846,
+      "grad_norm": 0.2803523769007011,
+      "learning_rate": 2.240639932921966e-05,
+      "loss": 0.4898,
+      "step": 634
+    },
+    {
+      "epoch": 2.572151898734177,
+      "grad_norm": 0.3198672221426174,
+      "learning_rate": 2.2350043382056995e-05,
+      "loss": 0.4967,
+      "step": 635
+    },
+    {
+      "epoch": 2.5762025316455697,
+      "grad_norm": 0.31434440946592646,
+      "learning_rate": 2.2293668507968015e-05,
+      "loss": 0.5003,
+      "step": 636
+    },
+    {
+      "epoch": 2.580253164556962,
+      "grad_norm": 0.7014461771652335,
+      "learning_rate": 2.2237275160988186e-05,
+      "loss": 0.5123,
+      "step": 637
+    },
+    {
+      "epoch": 2.5843037974683547,
+      "grad_norm": 0.30376728299291383,
+      "learning_rate": 2.2180863795301787e-05,
+      "loss": 0.5011,
+      "step": 638
+    },
+    {
+      "epoch": 2.5883544303797468,
+      "grad_norm": 0.3370529877250139,
+      "learning_rate": 2.212443486523819e-05,
+      "loss": 0.4781,
+      "step": 639
+    },
+    {
+      "epoch": 2.5924050632911393,
+      "grad_norm": 0.28363016617298786,
+      "learning_rate": 2.2067988825268243e-05,
+      "loss": 0.4895,
+      "step": 640
+    },
+    {
+      "epoch": 2.596455696202532,
+      "grad_norm": 0.29623382388584313,
+      "learning_rate": 2.2011526130000596e-05,
+      "loss": 0.502,
+      "step": 641
+    },
+    {
+      "epoch": 2.600506329113924,
+      "grad_norm": 0.3045941513626773,
+      "learning_rate": 2.1955047234178038e-05,
+      "loss": 0.4928,
+      "step": 642
+    },
+    {
+      "epoch": 2.6045569620253164,
+      "grad_norm": 3.735490700991824,
+      "learning_rate": 2.1898552592673825e-05,
+      "loss": 0.572,
+      "step": 643
+    },
+    {
+      "epoch": 2.608607594936709,
+      "grad_norm": 0.4125920629588283,
+      "learning_rate": 2.184204266048803e-05,
+      "loss": 0.4986,
+      "step": 644
+    },
+    {
+      "epoch": 2.6126582278481014,
+      "grad_norm": 0.289955171072708,
+      "learning_rate": 2.1785517892743887e-05,
+      "loss": 0.5195,
+      "step": 645
+    },
+    {
+      "epoch": 2.616708860759494,
+      "grad_norm": 0.34711464913455536,
+      "learning_rate": 2.17289787446841e-05,
+      "loss": 0.4958,
+      "step": 646
+    },
+    {
+      "epoch": 2.620759493670886,
+      "grad_norm": 0.339230143600048,
+      "learning_rate": 2.1672425671667198e-05,
+      "loss": 0.4825,
+      "step": 647
+    },
+    {
+      "epoch": 2.6248101265822785,
+      "grad_norm": 0.6552102055306429,
+      "learning_rate": 2.161585912916385e-05,
+      "loss": 0.5016,
+      "step": 648
+    },
+    {
+      "epoch": 2.628860759493671,
+      "grad_norm": 0.30504219439099645,
+      "learning_rate": 2.1559279572753214e-05,
+      "loss": 0.5006,
+      "step": 649
+    },
+    {
+      "epoch": 2.632911392405063,
+      "grad_norm": 0.32391489325732314,
+      "learning_rate": 2.1502687458119268e-05,
+      "loss": 0.5001,
+      "step": 650
+    },
+    {
+      "epoch": 2.6369620253164556,
+      "grad_norm": 0.2833567155734703,
+      "learning_rate": 2.1446083241047116e-05,
+      "loss": 0.5046,
+      "step": 651
+    },
+    {
+      "epoch": 2.641012658227848,
+      "grad_norm": 0.3317922825504114,
+      "learning_rate": 2.1389467377419333e-05,
+      "loss": 0.4912,
+      "step": 652
+    },
+    {
+      "epoch": 2.6450632911392407,
+      "grad_norm": 0.3173315570244207,
+      "learning_rate": 2.133284032321232e-05,
+      "loss": 0.4965,
+      "step": 653
+    },
+    {
+      "epoch": 2.649113924050633,
+      "grad_norm": 0.31277803048082975,
+      "learning_rate": 2.1276202534492566e-05,
+      "loss": 0.5091,
+      "step": 654
+    },
+    {
+      "epoch": 2.6531645569620252,
+      "grad_norm": 0.3367289794296316,
+      "learning_rate": 2.121955446741306e-05,
+      "loss": 0.4928,
+      "step": 655
+    },
+    {
+      "epoch": 2.6572151898734178,
+      "grad_norm": 0.36912683557203896,
+      "learning_rate": 2.1162896578209517e-05,
+      "loss": 0.5149,
+      "step": 656
+    },
+    {
+      "epoch": 2.6612658227848103,
+      "grad_norm": 0.30213239105156964,
+      "learning_rate": 2.1106229323196813e-05,
+      "loss": 0.4945,
+      "step": 657
+    },
+    {
+      "epoch": 2.6653164556962023,
+      "grad_norm": 0.2545774070646931,
+      "learning_rate": 2.1049553158765214e-05,
+      "loss": 0.5043,
+      "step": 658
+    },
+    {
+      "epoch": 2.669367088607595,
+      "grad_norm": 0.2764981951984016,
+      "learning_rate": 2.0992868541376764e-05,
+      "loss": 0.4942,
+      "step": 659
+    },
+    {
+      "epoch": 2.6734177215189874,
+      "grad_norm": 0.259720877414413,
+      "learning_rate": 2.093617592756158e-05,
+      "loss": 0.5084,
+      "step": 660
+    },
+    {
+      "epoch": 2.67746835443038,
+      "grad_norm": 0.2633886605652148,
+      "learning_rate": 2.0879475773914167e-05,
+      "loss": 0.5081,
+      "step": 661
+    },
+    {
+      "epoch": 2.6815189873417724,
+      "grad_norm": 1.8457154109361442,
+      "learning_rate": 2.082276853708978e-05,
+      "loss": 0.5192,
+      "step": 662
+    },
+    {
+      "epoch": 2.6855696202531645,
+      "grad_norm": 0.29323422986799735,
+      "learning_rate": 2.076605467380071e-05,
+      "loss": 0.4958,
+      "step": 663
+    },
+    {
+      "epoch": 2.689620253164557,
+      "grad_norm": 0.2816034888025096,
+      "learning_rate": 2.0709334640812613e-05,
+      "loss": 0.498,
+      "step": 664
+    },
+    {
+      "epoch": 2.6936708860759495,
+      "grad_norm": 0.2763981748959946,
+      "learning_rate": 2.0652608894940824e-05,
+      "loss": 0.5058,
+      "step": 665
+    },
+    {
+      "epoch": 2.6977215189873416,
+      "grad_norm": 0.29639998620911534,
+      "learning_rate": 2.0595877893046722e-05,
+      "loss": 0.4975,
+      "step": 666
+    },
+    {
+      "epoch": 2.701772151898734,
+      "grad_norm": 0.3159084760223744,
+      "learning_rate": 2.0539142092033985e-05,
+      "loss": 0.4862,
+      "step": 667
+    },
+    {
+      "epoch": 2.7058227848101266,
+      "grad_norm": 0.3229740145973752,
+      "learning_rate": 2.048240194884496e-05,
+      "loss": 0.4872,
+      "step": 668
+    },
+    {
+      "epoch": 2.709873417721519,
+      "grad_norm": 0.5778711664989139,
+      "learning_rate": 2.042565792045695e-05,
+      "loss": 0.5186,
+      "step": 669
+    },
+    {
+      "epoch": 2.7139240506329116,
+      "grad_norm": 0.38784907696428345,
+      "learning_rate": 2.036891046387857e-05,
+      "loss": 0.5088,
+      "step": 670
+    },
+    {
+      "epoch": 2.7179746835443037,
+      "grad_norm": 0.37504888006721276,
+      "learning_rate": 2.0312160036146036e-05,
+      "loss": 0.5021,
+      "step": 671
+    },
+    {
+      "epoch": 2.7220253164556962,
+      "grad_norm": 0.2800576635728949,
+      "learning_rate": 2.025540709431948e-05,
+      "loss": 0.4995,
+      "step": 672
+    },
+    {
+      "epoch": 2.7260759493670887,
+      "grad_norm": 0.31078239144712505,
+      "learning_rate": 2.0198652095479298e-05,
+      "loss": 0.4936,
+      "step": 673
+    },
+    {
+      "epoch": 2.730126582278481,
+      "grad_norm": 0.27162263463808856,
+      "learning_rate": 2.014189549672245e-05,
+      "loss": 0.4928,
+      "step": 674
+    },
+    {
+      "epoch": 2.7341772151898733,
+      "grad_norm": 0.28795135089215923,
+      "learning_rate": 2.0085137755158776e-05,
+      "loss": 0.4924,
+      "step": 675
+    },
+    {
+      "epoch": 2.738227848101266,
+      "grad_norm": 1.1084192697854593,
+      "learning_rate": 2.0028379327907327e-05,
+      "loss": 0.5051,
+      "step": 676
+    },
+    {
+      "epoch": 2.7422784810126584,
+      "grad_norm": 0.2807583822986574,
+      "learning_rate": 1.9971620672092676e-05,
+      "loss": 0.4997,
+      "step": 677
+    },
+    {
+      "epoch": 2.746329113924051,
+      "grad_norm": 0.3549098157055366,
+      "learning_rate": 1.991486224484123e-05,
+      "loss": 0.5008,
+      "step": 678
+    },
+    {
+      "epoch": 2.750379746835443,
+      "grad_norm": 0.2598999694003001,
+      "learning_rate": 1.985810450327756e-05,
+      "loss": 0.4988,
+      "step": 679
+    },
+    {
+      "epoch": 2.7544303797468355,
+      "grad_norm": 0.3945034010151704,
+      "learning_rate": 1.9801347904520706e-05,
+      "loss": 0.5047,
+      "step": 680
+    },
+    {
+      "epoch": 2.758481012658228,
+      "grad_norm": 0.2760565355999657,
+      "learning_rate": 1.974459290568053e-05,
+      "loss": 0.5072,
+      "step": 681
+    },
+    {
+      "epoch": 2.76253164556962,
+      "grad_norm": 0.4377293044524199,
+      "learning_rate": 1.968783996385397e-05,
+      "loss": 0.5022,
+      "step": 682
+    },
+    {
+      "epoch": 2.7665822784810126,
+      "grad_norm": 0.25288888132803583,
+      "learning_rate": 1.963108953612143e-05,
+      "loss": 0.4929,
+      "step": 683
+    },
+    {
+      "epoch": 2.770632911392405,
+      "grad_norm": 0.36309752943639034,
+      "learning_rate": 1.9574342079543056e-05,
+      "loss": 0.4973,
+      "step": 684
+    },
+    {
+      "epoch": 2.7746835443037976,
+      "grad_norm": 1.9913056691818236,
+      "learning_rate": 1.9517598051155046e-05,
+      "loss": 0.5318,
+      "step": 685
+    },
+    {
+      "epoch": 2.77873417721519,
+      "grad_norm": 0.4209863892760475,
+      "learning_rate": 1.9460857907966025e-05,
+      "loss": 0.494,
+      "step": 686
+    },
+    {
+      "epoch": 2.782784810126582,
+      "grad_norm": 0.2512911797207737,
+      "learning_rate": 1.9404122106953285e-05,
+      "loss": 0.4954,
+      "step": 687
+    },
+    {
+      "epoch": 2.7868354430379747,
+      "grad_norm": 0.408738783539645,
+      "learning_rate": 1.9347391105059176e-05,
+      "loss": 0.5044,
+      "step": 688
+    },
+    {
+      "epoch": 2.790886075949367,
+      "grad_norm": 3.07329121338879,
+      "learning_rate": 1.92906653591874e-05,
+      "loss": 0.5177,
+      "step": 689
+    },
+    {
+      "epoch": 2.7949367088607593,
+      "grad_norm": 0.4615817609813334,
+      "learning_rate": 1.9233945326199295e-05,
+      "loss": 0.4997,
+      "step": 690
+    },
+    {
+      "epoch": 2.798987341772152,
+      "grad_norm": 0.2870404260580923,
+      "learning_rate": 1.917723146291022e-05,
+      "loss": 0.5114,
+      "step": 691
+    },
+    {
+      "epoch": 2.8030379746835443,
+      "grad_norm": 0.37431597230412494,
+      "learning_rate": 1.912052422608584e-05,
+      "loss": 0.5194,
+      "step": 692
+    },
+    {
+      "epoch": 2.807088607594937,
+      "grad_norm": 0.3242301761635135,
+      "learning_rate": 1.9063824072438428e-05,
+      "loss": 0.5044,
+      "step": 693
+    },
+    {
+      "epoch": 2.8111392405063294,
+      "grad_norm": 0.29329840651222105,
+      "learning_rate": 1.9007131458623246e-05,
+      "loss": 0.5195,
+      "step": 694
+    },
+    {
+      "epoch": 2.8151898734177214,
+      "grad_norm": 0.3126593003971115,
+      "learning_rate": 1.895044684123479e-05,
+      "loss": 0.5001,
+      "step": 695
+    },
+    {
+      "epoch": 2.819240506329114,
+      "grad_norm": 0.2625892309189569,
+      "learning_rate": 1.8893770676803194e-05,
+      "loss": 0.5009,
+      "step": 696
+    },
+    {
+      "epoch": 2.8232911392405065,
+      "grad_norm": 0.3128256653018164,
+      "learning_rate": 1.8837103421790486e-05,
+      "loss": 0.4955,
+      "step": 697
+    },
+    {
+      "epoch": 2.8273417721518985,
+      "grad_norm": 0.27931823587792415,
+      "learning_rate": 1.8780445532586952e-05,
+      "loss": 0.5032,
+      "step": 698
+    },
+    {
+      "epoch": 2.831392405063291,
+      "grad_norm": 0.4358999303049613,
+      "learning_rate": 1.872379746550743e-05,
+      "loss": 0.4991,
+      "step": 699
+    },
+    {
+      "epoch": 2.8354430379746836,
+      "grad_norm": 0.264283085722407,
+      "learning_rate": 1.866715967678769e-05,
+      "loss": 0.4966,
+      "step": 700
+    },
+    {
+      "epoch": 2.839493670886076,
+      "grad_norm": 0.34195600085604644,
+      "learning_rate": 1.861053262258067e-05,
+      "loss": 0.502,
+      "step": 701
+    },
+    {
+      "epoch": 2.8435443037974686,
+      "grad_norm": 0.27015562385534153,
+      "learning_rate": 1.8553916758952897e-05,
+      "loss": 0.4881,
+      "step": 702
+    },
+    {
+      "epoch": 2.8475949367088607,
+      "grad_norm": 0.3372511403508146,
+      "learning_rate": 1.8497312541880735e-05,
+      "loss": 0.499,
+      "step": 703
+    },
+    {
+      "epoch": 2.851645569620253,
+      "grad_norm": 0.2784045719887347,
+      "learning_rate": 1.8440720427246786e-05,
+      "loss": 0.4924,
+      "step": 704
+    },
+    {
+      "epoch": 2.8556962025316457,
+      "grad_norm": 0.39300097693682906,
+      "learning_rate": 1.8384140870836157e-05,
+      "loss": 0.4916,
+      "step": 705
+    },
+    {
+      "epoch": 2.8597468354430378,
+      "grad_norm": 0.2454630489557861,
+      "learning_rate": 1.8327574328332806e-05,
+      "loss": 0.4934,
+      "step": 706
+    },
+    {
+      "epoch": 2.8637974683544303,
+      "grad_norm": 0.27371858942958355,
+      "learning_rate": 1.8271021255315906e-05,
+      "loss": 0.4872,
+      "step": 707
+    },
+    {
+      "epoch": 2.867848101265823,
+      "grad_norm": 0.24507771655872482,
+      "learning_rate": 1.8214482107256117e-05,
+      "loss": 0.4847,
+      "step": 708
+    },
+    {
+      "epoch": 2.8718987341772153,
+      "grad_norm": 0.26943159336216044,
+      "learning_rate": 1.8157957339511968e-05,
+      "loss": 0.4989,
+      "step": 709
+    },
+    {
+      "epoch": 2.875949367088608,
+      "grad_norm": 0.2510573990382002,
+      "learning_rate": 1.8101447407326182e-05,
+      "loss": 0.4931,
+      "step": 710
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 1.02038111694285,
+      "learning_rate": 1.8044952765821966e-05,
+      "loss": 0.5157,
+      "step": 711
+    },
+    {
+      "epoch": 2.8840506329113924,
+      "grad_norm": 0.26328871862656755,
+      "learning_rate": 1.7988473869999407e-05,
+      "loss": 0.4986,
+      "step": 712
+    },
+    {
+      "epoch": 2.888101265822785,
+      "grad_norm": 0.25603118343727566,
+      "learning_rate": 1.7932011174731764e-05,
+      "loss": 0.4938,
+      "step": 713
+    },
+    {
+      "epoch": 2.892151898734177,
+      "grad_norm": 0.25259733617054314,
+      "learning_rate": 1.7875565134761817e-05,
+      "loss": 0.4931,
+      "step": 714
+    },
+    {
+      "epoch": 2.8962025316455695,
+      "grad_norm": 0.3041887316435713,
+      "learning_rate": 1.7819136204698226e-05,
+      "loss": 0.5043,
+      "step": 715
+    },
+    {
+      "epoch": 2.900253164556962,
+      "grad_norm": 0.2508858141705034,
+      "learning_rate": 1.776272483901182e-05,
+      "loss": 0.504,
+      "step": 716
+    },
+    {
+      "epoch": 2.9043037974683545,
+      "grad_norm": 0.26489639622027716,
+      "learning_rate": 1.7706331492031995e-05,
+      "loss": 0.4914,
+      "step": 717
+    },
+    {
+      "epoch": 2.908354430379747,
+      "grad_norm": 0.24523410249215852,
+      "learning_rate": 1.764995661794301e-05,
+      "loss": 0.4925,
+      "step": 718
+    },
+    {
+      "epoch": 2.912405063291139,
+      "grad_norm": 0.2738442889423671,
+      "learning_rate": 1.759360067078035e-05,
+      "loss": 0.4942,
+      "step": 719
+    },
+    {
+      "epoch": 2.9164556962025316,
+      "grad_norm": 0.2563091028045858,
+      "learning_rate": 1.7537264104427064e-05,
+      "loss": 0.4884,
+      "step": 720
+    },
+    {
+      "epoch": 2.920506329113924,
+      "grad_norm": 0.26771987541427206,
+      "learning_rate": 1.748094737261012e-05,
+      "loss": 0.4851,
+      "step": 721
+    },
+    {
+      "epoch": 2.9245569620253162,
+      "grad_norm": 0.9087036741073158,
+      "learning_rate": 1.7424650928896726e-05,
+      "loss": 0.5028,
+      "step": 722
+    },
+    {
+      "epoch": 2.9286075949367087,
+      "grad_norm": 0.4111438382438639,
+      "learning_rate": 1.7368375226690712e-05,
+      "loss": 0.5099,
+      "step": 723
+    },
+    {
+      "epoch": 2.9326582278481013,
+      "grad_norm": 0.24913106472541177,
+      "learning_rate": 1.731212071922883e-05,
+      "loss": 0.5,
+      "step": 724
+    },
+    {
+      "epoch": 2.9367088607594938,
+      "grad_norm": 0.2798352903751036,
+      "learning_rate": 1.7255887859577156e-05,
+      "loss": 0.5013,
+      "step": 725
+    },
+    {
+      "epoch": 2.9407594936708863,
+      "grad_norm": 0.2606052276932628,
+      "learning_rate": 1.7199677100627427e-05,
+      "loss": 0.4953,
+      "step": 726
+    },
+    {
+      "epoch": 2.9448101265822784,
+      "grad_norm": 0.2798031751347415,
+      "learning_rate": 1.7143488895093343e-05,
+      "loss": 0.5052,
+      "step": 727
+    },
+    {
+      "epoch": 2.948860759493671,
+      "grad_norm": 0.23751237243672865,
+      "learning_rate": 1.7087323695506994e-05,
+      "loss": 0.4901,
+      "step": 728
+    },
+    {
+      "epoch": 2.9529113924050634,
+      "grad_norm": 0.2511650937754486,
+      "learning_rate": 1.7031181954215194e-05,
+      "loss": 0.4817,
+      "step": 729
+    },
+    {
+      "epoch": 2.9569620253164555,
+      "grad_norm": 0.274600015847451,
+      "learning_rate": 1.6975064123375788e-05,
+      "loss": 0.4965,
+      "step": 730
+    },
+    {
+      "epoch": 2.961012658227848,
+      "grad_norm": 0.2508102700810794,
+      "learning_rate": 1.6918970654954084e-05,
+      "loss": 0.4936,
+      "step": 731
+    },
+    {
+      "epoch": 2.9650632911392405,
+      "grad_norm": 0.2749466458291213,
+      "learning_rate": 1.686290200071915e-05,
+      "loss": 0.5071,
+      "step": 732
+    },
+    {
+      "epoch": 2.969113924050633,
+      "grad_norm": 0.2689193874508499,
+      "learning_rate": 1.6806858612240234e-05,
+      "loss": 0.5004,
+      "step": 733
+    },
+    {
+      "epoch": 2.9731645569620255,
+      "grad_norm": 0.25220251896829765,
+      "learning_rate": 1.6750840940883078e-05,
+      "loss": 0.5104,
+      "step": 734
+    },
+    {
+      "epoch": 2.9772151898734176,
+      "grad_norm": 0.25216923996399926,
+      "learning_rate": 1.6694849437806305e-05,
+      "loss": 0.4856,
+      "step": 735
+    },
+    {
+      "epoch": 2.98126582278481,
+      "grad_norm": 0.2672746249087662,
+      "learning_rate": 1.663888455395778e-05,
+      "loss": 0.5077,
+      "step": 736
+    },
+    {
+      "epoch": 2.9853164556962026,
+      "grad_norm": 1.2621731497153623,
+      "learning_rate": 1.6582946740070995e-05,
+      "loss": 0.5266,
+      "step": 737
+    },
+    {
+      "epoch": 2.9893670886075947,
+      "grad_norm": 0.31301906508386823,
+      "learning_rate": 1.6527036446661396e-05,
+      "loss": 0.4952,
+      "step": 738
+    },
+    {
+      "epoch": 2.993417721518987,
+      "grad_norm": 0.2795930522047099,
+      "learning_rate": 1.6471154124022818e-05,
+      "loss": 0.4907,
+      "step": 739
+    },
+    {
+      "epoch": 2.9974683544303797,
+      "grad_norm": 0.38583921900101475,
+      "learning_rate": 1.6415300222223788e-05,
+      "loss": 0.4931,
+      "step": 740
+    },
+    {
+      "epoch": 3.0015189873417722,
+      "grad_norm": 0.45121417413343595,
+      "learning_rate": 1.6359475191103958e-05,
+      "loss": 0.4916,
+      "step": 741
+    },
+    {
+      "epoch": 3.0055696202531648,
+      "grad_norm": 0.3352286466499682,
+      "learning_rate": 1.6303679480270466e-05,
+      "loss": 0.4421,
+      "step": 742
+    },
+    {
+      "epoch": 3.009620253164557,
+      "grad_norm": 0.3108695840840613,
+      "learning_rate": 1.624791353909428e-05,
+      "loss": 0.4526,
+      "step": 743
+    },
+    {
+      "epoch": 3.0136708860759494,
+      "grad_norm": 0.4104235924594916,
+      "learning_rate": 1.619217781670663e-05,
+      "loss": 0.4462,
+      "step": 744
+    },
+    {
+      "epoch": 3.017721518987342,
+      "grad_norm": 0.36065199400990067,
+      "learning_rate": 1.6136472761995373e-05,
+      "loss": 0.4343,
+      "step": 745
+    },
+    {
+      "epoch": 3.0217721518987344,
+      "grad_norm": 0.3873463334359871,
+      "learning_rate": 1.608079882360134e-05,
+      "loss": 0.4514,
+      "step": 746
+    },
+    {
+      "epoch": 3.0258227848101265,
+      "grad_norm": 0.35998451707213275,
+      "learning_rate": 1.60251564499148e-05,
+      "loss": 0.4491,
+      "step": 747
+    },
+    {
+      "epoch": 3.029873417721519,
+      "grad_norm": 0.35207118729316195,
+      "learning_rate": 1.596954608907176e-05,
+      "loss": 0.4416,
+      "step": 748
+    },
+    {
+      "epoch": 3.0339240506329115,
+      "grad_norm": 0.3736336946589617,
+      "learning_rate": 1.591396818895043e-05,
+      "loss": 0.4498,
+      "step": 749
+    },
+    {
+      "epoch": 3.037974683544304,
+      "grad_norm": 0.3090122808286461,
+      "learning_rate": 1.585842319716759e-05,
+      "loss": 0.4435,
+      "step": 750
+    },
+    {
+      "epoch": 3.042025316455696,
+      "grad_norm": 0.32909415734728364,
+      "learning_rate": 1.5802911561074944e-05,
+      "loss": 0.4386,
+      "step": 751
+    },
+    {
+      "epoch": 3.0460759493670886,
+      "grad_norm": 0.31650010363371284,
+      "learning_rate": 1.5747433727755595e-05,
+      "loss": 0.4457,
+      "step": 752
+    },
+    {
+      "epoch": 3.050126582278481,
+      "grad_norm": 0.28668085145473804,
+      "learning_rate": 1.5691990144020376e-05,
+      "loss": 0.437,
+      "step": 753
+    },
+    {
+      "epoch": 3.0541772151898736,
+      "grad_norm": 0.29793289728099787,
+      "learning_rate": 1.5636581256404297e-05,
+      "loss": 0.4391,
+      "step": 754
+    },
+    {
+      "epoch": 3.0582278481012657,
+      "grad_norm": 0.28094266456582984,
+      "learning_rate": 1.558120751116291e-05,
+      "loss": 0.4391,
+      "step": 755
+    },
+    {
+      "epoch": 3.062278481012658,
+      "grad_norm": 0.28739191063663627,
+      "learning_rate": 1.552586935426876e-05,
+      "loss": 0.4389,
+      "step": 756
+    },
+    {
+      "epoch": 3.0663291139240507,
+      "grad_norm": 0.2616803154385566,
+      "learning_rate": 1.547056723140774e-05,
+      "loss": 0.432,
+      "step": 757
+    },
+    {
+      "epoch": 3.0703797468354432,
+      "grad_norm": 0.25932517655028514,
+      "learning_rate": 1.5415301587975565e-05,
+      "loss": 0.4386,
+      "step": 758
+    },
+    {
+      "epoch": 3.0744303797468353,
+      "grad_norm": 0.31407429999168546,
+      "learning_rate": 1.536007286907411e-05,
+      "loss": 0.4494,
+      "step": 759
+    },
+    {
+      "epoch": 3.078481012658228,
+      "grad_norm": 0.2562845956367167,
+      "learning_rate": 1.5304881519507896e-05,
+      "loss": 0.4335,
+      "step": 760
+    },
+    {
+      "epoch": 3.0825316455696203,
+      "grad_norm": 0.2950018590785276,
+      "learning_rate": 1.5249727983780453e-05,
+      "loss": 0.4366,
+      "step": 761
+    },
+    {
+      "epoch": 3.086582278481013,
+      "grad_norm": 0.26240927342373627,
+      "learning_rate": 1.5194612706090786e-05,
+      "loss": 0.4387,
+      "step": 762
+    },
+    {
+      "epoch": 3.090632911392405,
+      "grad_norm": 0.2989059263148206,
+      "learning_rate": 1.5139536130329771e-05,
+      "loss": 0.4431,
+      "step": 763
+    },
+    {
+      "epoch": 3.0946835443037974,
+      "grad_norm": 0.2801746909447191,
+      "learning_rate": 1.508449870007656e-05,
+      "loss": 0.4462,
+      "step": 764
+    },
+    {
+      "epoch": 3.09873417721519,
+      "grad_norm": 0.305688367994578,
+      "learning_rate": 1.5029500858595056e-05,
+      "loss": 0.4314,
+      "step": 765
+    },
+    {
+      "epoch": 3.1027848101265825,
+      "grad_norm": 0.25450510065787746,
+      "learning_rate": 1.4974543048830328e-05,
+      "loss": 0.4375,
+      "step": 766
+    },
+    {
+      "epoch": 3.1068354430379745,
+      "grad_norm": 0.25974311081008133,
+      "learning_rate": 1.4919625713405e-05,
+      "loss": 0.4452,
+      "step": 767
+    },
+    {
+      "epoch": 3.110886075949367,
+      "grad_norm": 0.2825186087098998,
+      "learning_rate": 1.4864749294615756e-05,
+      "loss": 0.4345,
+      "step": 768
+    },
+    {
+      "epoch": 3.1149367088607596,
+      "grad_norm": 0.2666762686816631,
+      "learning_rate": 1.4809914234429716e-05,
+      "loss": 0.4425,
+      "step": 769
+    },
+    {
+      "epoch": 3.118987341772152,
+      "grad_norm": 1.1082562237707059,
+      "learning_rate": 1.4755120974480923e-05,
+      "loss": 0.4694,
+      "step": 770
+    },
+    {
+      "epoch": 3.123037974683544,
+      "grad_norm": 0.30934578185831174,
+      "learning_rate": 1.4700369956066771e-05,
+      "loss": 0.4325,
+      "step": 771
+    },
+    {
+      "epoch": 3.1270886075949367,
+      "grad_norm": 0.3225118339112458,
+      "learning_rate": 1.4645661620144413e-05,
+      "loss": 0.4342,
+      "step": 772
+    },
+    {
+      "epoch": 3.131139240506329,
+      "grad_norm": 0.2708091894661945,
+      "learning_rate": 1.4590996407327284e-05,
+      "loss": 0.4364,
+      "step": 773
+    },
+    {
+      "epoch": 3.1351898734177217,
+      "grad_norm": 0.31909071007699646,
+      "learning_rate": 1.4536374757881487e-05,
+      "loss": 0.4397,
+      "step": 774
+    },
+    {
+      "epoch": 3.1392405063291138,
+      "grad_norm": 0.2907463048168282,
+      "learning_rate": 1.4481797111722271e-05,
+      "loss": 0.4525,
+      "step": 775
+    },
+    {
+      "epoch": 3.1432911392405063,
+      "grad_norm": 0.29920640839815743,
+      "learning_rate": 1.4427263908410507e-05,
+      "loss": 0.4494,
+      "step": 776
+    },
+    {
+      "epoch": 3.147341772151899,
+      "grad_norm": 0.3101155549334292,
+      "learning_rate": 1.4372775587149108e-05,
+      "loss": 0.4426,
+      "step": 777
+    },
+    {
+      "epoch": 3.1513924050632913,
+      "grad_norm": 0.2972624300356401,
+      "learning_rate": 1.4318332586779522e-05,
+      "loss": 0.4444,
+      "step": 778
+    },
+    {
+      "epoch": 3.1554430379746834,
+      "grad_norm": 0.2766281327120839,
+      "learning_rate": 1.4263935345778202e-05,
+      "loss": 0.4473,
+      "step": 779
+    },
+    {
+      "epoch": 3.159493670886076,
+      "grad_norm": 0.2712364025954586,
+      "learning_rate": 1.420958430225303e-05,
+      "loss": 0.42,
+      "step": 780
+    },
+    {
+      "epoch": 3.1635443037974684,
+      "grad_norm": 0.32583412752715,
+      "learning_rate": 1.415527989393985e-05,
+      "loss": 0.4424,
+      "step": 781
+    },
+    {
+      "epoch": 3.167594936708861,
+      "grad_norm": 0.2233983344084385,
+      "learning_rate": 1.410102255819891e-05,
+      "loss": 0.4341,
+      "step": 782
+    },
+    {
+      "epoch": 3.171645569620253,
+      "grad_norm": 0.36740439845953465,
+      "learning_rate": 1.404681273201131e-05,
+      "loss": 0.438,
+      "step": 783
+    },
+    {
+      "epoch": 3.1756962025316455,
+      "grad_norm": 0.2620327059981999,
+      "learning_rate": 1.399265085197556e-05,
+      "loss": 0.4236,
+      "step": 784
+    },
+    {
+      "epoch": 3.179746835443038,
+      "grad_norm": 0.3452601261674081,
+      "learning_rate": 1.393853735430398e-05,
+      "loss": 0.4532,
+      "step": 785
+    },
+    {
+      "epoch": 3.1837974683544306,
+      "grad_norm": 0.2580905963555383,
+      "learning_rate": 1.3884472674819246e-05,
+      "loss": 0.4474,
+      "step": 786
+    },
+    {
+      "epoch": 3.1878481012658226,
+      "grad_norm": 0.260385317329407,
+      "learning_rate": 1.3830457248950864e-05,
+      "loss": 0.4428,
+      "step": 787
+    },
+    {
+      "epoch": 3.191898734177215,
+      "grad_norm": 0.277970038108189,
+      "learning_rate": 1.377649151173163e-05,
+      "loss": 0.4443,
+      "step": 788
+    },
+    {
+      "epoch": 3.1959493670886077,
+      "grad_norm": 0.23368078022308109,
+      "learning_rate": 1.3722575897794181e-05,
+      "loss": 0.437,
+      "step": 789
+    },
+    {
+      "epoch": 3.2,
+      "grad_norm": 0.2841706236688083,
+      "learning_rate": 1.3668710841367472e-05,
+      "loss": 0.453,
+      "step": 790
+    },
+    {
+      "epoch": 3.2040506329113922,
+      "grad_norm": 0.27098350913841524,
+      "learning_rate": 1.361489677627324e-05,
+      "loss": 0.4407,
+      "step": 791
+    },
+    {
+      "epoch": 3.2081012658227848,
+      "grad_norm": 0.2661210793623429,
+      "learning_rate": 1.3561134135922585e-05,
+      "loss": 0.4345,
+      "step": 792
+    },
+    {
+      "epoch": 3.2121518987341773,
+      "grad_norm": 0.2497439364407498,
+      "learning_rate": 1.350742335331241e-05,
+      "loss": 0.4317,
+      "step": 793
+    },
+    {
+      "epoch": 3.21620253164557,
+      "grad_norm": 0.2737294884512297,
+      "learning_rate": 1.345376486102198e-05,
+      "loss": 0.4331,
+      "step": 794
+    },
+    {
+      "epoch": 3.220253164556962,
+      "grad_norm": 0.28833279535326894,
+      "learning_rate": 1.3400159091209414e-05,
+      "loss": 0.4542,
+      "step": 795
+    },
+    {
+      "epoch": 3.2243037974683544,
+      "grad_norm": 0.24435052847809421,
+      "learning_rate": 1.3346606475608216e-05,
+      "loss": 0.4543,
+      "step": 796
+    },
+    {
+      "epoch": 3.228354430379747,
+      "grad_norm": 0.26221512890671833,
+      "learning_rate": 1.3293107445523781e-05,
+      "loss": 0.4321,
+      "step": 797
+    },
+    {
+      "epoch": 3.2324050632911394,
+      "grad_norm": 0.251566109565453,
+      "learning_rate": 1.3239662431829949e-05,
+      "loss": 0.4544,
+      "step": 798
+    },
+    {
+      "epoch": 3.2364556962025315,
+      "grad_norm": 0.23719125976843064,
+      "learning_rate": 1.3186271864965509e-05,
+      "loss": 0.4424,
+      "step": 799
+    },
+    {
+      "epoch": 3.240506329113924,
+      "grad_norm": 0.2733556195279977,
+      "learning_rate": 1.3132936174930756e-05,
+      "loss": 0.4402,
+      "step": 800
+    },
+    {
+      "epoch": 3.2445569620253165,
+      "grad_norm": 0.26949453628824943,
+      "learning_rate": 1.3079655791283995e-05,
+      "loss": 0.4441,
+      "step": 801
+    },
+    {
+      "epoch": 3.248607594936709,
+      "grad_norm": 0.28892345065342845,
+      "learning_rate": 1.3026431143138108e-05,
+      "loss": 0.4498,
+      "step": 802
+    },
+    {
+      "epoch": 3.252658227848101,
+      "grad_norm": 0.25019975816107914,
+      "learning_rate": 1.2973262659157114e-05,
+      "loss": 0.439,
+      "step": 803
+    },
+    {
+      "epoch": 3.2567088607594936,
+      "grad_norm": 0.2858824639576326,
+      "learning_rate": 1.2920150767552651e-05,
+      "loss": 0.4332,
+      "step": 804
+    },
+    {
+      "epoch": 3.260759493670886,
+      "grad_norm": 0.2679979666932312,
+      "learning_rate": 1.2867095896080607e-05,
+      "loss": 0.4264,
+      "step": 805
+    },
+    {
+      "epoch": 3.2648101265822786,
+      "grad_norm": 0.27696486105891815,
+      "learning_rate": 1.2814098472037612e-05,
+      "loss": 0.4362,
+      "step": 806
+    },
+    {
+      "epoch": 3.2688607594936707,
+      "grad_norm": 0.24728992942218017,
+      "learning_rate": 1.276115892225764e-05,
+      "loss": 0.4275,
+      "step": 807
+    },
+    {
+      "epoch": 3.2729113924050632,
+      "grad_norm": 0.29472388035967767,
+      "learning_rate": 1.2708277673108555e-05,
+      "loss": 0.4401,
+      "step": 808
+    },
+    {
+      "epoch": 3.2769620253164558,
+      "grad_norm": 0.2468998660693589,
+      "learning_rate": 1.2655455150488649e-05,
+      "loss": 0.4497,
+      "step": 809
+    },
+    {
+      "epoch": 3.2810126582278483,
+      "grad_norm": 0.26718309808665375,
+      "learning_rate": 1.2602691779823272e-05,
+      "loss": 0.4315,
+      "step": 810
+    },
+    {
+      "epoch": 3.2850632911392403,
+      "grad_norm": 0.23077485295386982,
+      "learning_rate": 1.2549987986061355e-05,
+      "loss": 0.4304,
+      "step": 811
+    },
+    {
+      "epoch": 3.289113924050633,
+      "grad_norm": 0.27147482178999754,
+      "learning_rate": 1.2497344193672005e-05,
+      "loss": 0.442,
+      "step": 812
+    },
+    {
+      "epoch": 3.2931645569620254,
+      "grad_norm": 0.23610093389091447,
+      "learning_rate": 1.2444760826641092e-05,
+      "loss": 0.4366,
+      "step": 813
+    },
+    {
+      "epoch": 3.297215189873418,
+      "grad_norm": 0.24367087651406416,
+      "learning_rate": 1.2392238308467817e-05,
+      "loss": 0.4379,
+      "step": 814
+    },
+    {
+      "epoch": 3.30126582278481,
+      "grad_norm": 0.2486333451291113,
+      "learning_rate": 1.2339777062161326e-05,
+      "loss": 0.4504,
+      "step": 815
+    },
+    {
+      "epoch": 3.3053164556962025,
+      "grad_norm": 0.47004038935296233,
+      "learning_rate": 1.2287377510237293e-05,
+      "loss": 0.4426,
+      "step": 816
+    },
+    {
+      "epoch": 3.309367088607595,
+      "grad_norm": 0.25641474763464484,
+      "learning_rate": 1.2235040074714488e-05,
+      "loss": 0.4423,
+      "step": 817
+    },
+    {
+      "epoch": 3.3134177215189875,
+      "grad_norm": 0.2417981250999771,
+      "learning_rate": 1.2182765177111434e-05,
+      "loss": 0.4542,
+      "step": 818
+    },
+    {
+      "epoch": 3.3174683544303796,
+      "grad_norm": 0.24422634439692198,
+      "learning_rate": 1.213055323844297e-05,
+      "loss": 0.4288,
+      "step": 819
+    },
+    {
+      "epoch": 3.321518987341772,
+      "grad_norm": 0.24485964106041805,
+      "learning_rate": 1.2078404679216864e-05,
+      "loss": 0.4332,
+      "step": 820
+    },
+    {
+      "epoch": 3.3255696202531646,
+      "grad_norm": 0.247449784815094,
+      "learning_rate": 1.2026319919430458e-05,
+      "loss": 0.4588,
+      "step": 821
+    },
+    {
+      "epoch": 3.329620253164557,
+      "grad_norm": 0.2575792122082809,
+      "learning_rate": 1.1974299378567227e-05,
+      "loss": 0.4398,
+      "step": 822
+    },
+    {
+      "epoch": 3.333670886075949,
+      "grad_norm": 0.23191207115842696,
+      "learning_rate": 1.1922343475593462e-05,
+      "loss": 0.4217,
+      "step": 823
+    },
+    {
+      "epoch": 3.3377215189873417,
+      "grad_norm": 0.2658723444614187,
+      "learning_rate": 1.187045262895488e-05,
+      "loss": 0.447,
+      "step": 824
+    },
+    {
+      "epoch": 3.3417721518987342,
+      "grad_norm": 0.2659561016741752,
+      "learning_rate": 1.1818627256573203e-05,
+      "loss": 0.446,
+      "step": 825
+    },
+    {
+      "epoch": 3.3458227848101267,
+      "grad_norm": 0.2251823838052862,
+      "learning_rate": 1.1766867775842864e-05,
+      "loss": 0.428,
+      "step": 826
+    },
+    {
+      "epoch": 3.349873417721519,
+      "grad_norm": 0.26390124140531984,
+      "learning_rate": 1.1715174603627615e-05,
+      "loss": 0.4345,
+      "step": 827
+    },
+    {
+      "epoch": 3.3539240506329113,
+      "grad_norm": 0.22954528159439458,
+      "learning_rate": 1.1663548156257147e-05,
+      "loss": 0.4481,
+      "step": 828
+    },
+    {
+      "epoch": 3.357974683544304,
+      "grad_norm": 0.24031739010155342,
+      "learning_rate": 1.161198884952377e-05,
+      "loss": 0.4417,
+      "step": 829
+    },
+    {
+      "epoch": 3.3620253164556964,
+      "grad_norm": 0.2557158038856814,
+      "learning_rate": 1.1560497098679056e-05,
+      "loss": 0.4504,
+      "step": 830
+    },
+    {
+      "epoch": 3.3660759493670884,
+      "grad_norm": 0.23914434471707105,
+      "learning_rate": 1.1509073318430479e-05,
+      "loss": 0.4377,
+      "step": 831
+    },
+    {
+      "epoch": 3.370126582278481,
+      "grad_norm": 0.2784460973444432,
+      "learning_rate": 1.1457717922938116e-05,
+      "loss": 0.4442,
+      "step": 832
+    },
+    {
+      "epoch": 3.3741772151898735,
+      "grad_norm": 0.24157280969632744,
+      "learning_rate": 1.1406431325811233e-05,
+      "loss": 0.4523,
+      "step": 833
+    },
+    {
+      "epoch": 3.378227848101266,
+      "grad_norm": 0.2570842575636363,
+      "learning_rate": 1.135521394010506e-05,
+      "loss": 0.4438,
+      "step": 834
+    },
+    {
+      "epoch": 3.382278481012658,
+      "grad_norm": 0.9485809102167648,
+      "learning_rate": 1.1304066178317367e-05,
+      "loss": 0.4481,
+      "step": 835
+    },
+    {
+      "epoch": 3.3863291139240506,
+      "grad_norm": 0.25362819223163174,
+      "learning_rate": 1.1252988452385199e-05,
+      "loss": 0.4386,
+      "step": 836
+    },
+    {
+      "epoch": 3.390379746835443,
+      "grad_norm": 0.27830120577342393,
+      "learning_rate": 1.1201981173681536e-05,
+      "loss": 0.4474,
+      "step": 837
+    },
+    {
+      "epoch": 3.3944303797468356,
+      "grad_norm": 0.28638396658621845,
+      "learning_rate": 1.1151044753011991e-05,
+      "loss": 0.4442,
+      "step": 838
+    },
+    {
+      "epoch": 3.3984810126582277,
+      "grad_norm": 0.2406726297634227,
+      "learning_rate": 1.1100179600611491e-05,
+      "loss": 0.4511,
+      "step": 839
+    },
+    {
+      "epoch": 3.40253164556962,
+      "grad_norm": 0.25177042530618543,
+      "learning_rate": 1.1049386126140985e-05,
+      "loss": 0.4439,
+      "step": 840
+    },
+    {
+      "epoch": 3.4065822784810127,
+      "grad_norm": 0.24951187428668178,
+      "learning_rate": 1.0998664738684128e-05,
+      "loss": 0.439,
+      "step": 841
+    },
+    {
+      "epoch": 3.410632911392405,
+      "grad_norm": 0.27017648419029144,
+      "learning_rate": 1.0948015846744e-05,
+      "loss": 0.4296,
+      "step": 842
+    },
+    {
+      "epoch": 3.4146835443037973,
+      "grad_norm": 0.23485592566617547,
+      "learning_rate": 1.0897439858239832e-05,
+      "loss": 0.4193,
+      "step": 843
+    },
+    {
+      "epoch": 3.41873417721519,
+      "grad_norm": 0.25046630997710706,
+      "learning_rate": 1.0846937180503652e-05,
+      "loss": 0.4366,
+      "step": 844
+    },
+    {
+      "epoch": 3.4227848101265823,
+      "grad_norm": 0.26822503994751884,
+      "learning_rate": 1.0796508220277117e-05,
+      "loss": 0.4496,
+      "step": 845
+    },
+    {
+      "epoch": 3.426835443037975,
+      "grad_norm": 0.2451036340861361,
+      "learning_rate": 1.0746153383708107e-05,
+      "loss": 0.4367,
+      "step": 846
+    },
+    {
+      "epoch": 3.430886075949367,
+      "grad_norm": 0.23716992423063563,
+      "learning_rate": 1.0695873076347579e-05,
+      "loss": 0.4428,
+      "step": 847
+    },
+    {
+      "epoch": 3.4349367088607594,
+      "grad_norm": 0.6868531210463256,
+      "learning_rate": 1.0645667703146205e-05,
+      "loss": 0.4294,
+      "step": 848
+    },
+    {
+      "epoch": 3.438987341772152,
+      "grad_norm": 0.23813957260483462,
+      "learning_rate": 1.0595537668451161e-05,
+      "loss": 0.4297,
+      "step": 849
+    },
+    {
+      "epoch": 3.4430379746835444,
+      "grad_norm": 0.229493008491044,
+      "learning_rate": 1.0545483376002854e-05,
+      "loss": 0.4585,
+      "step": 850
+    },
+    {
+      "epoch": 3.4470886075949365,
+      "grad_norm": 0.25217817845386675,
+      "learning_rate": 1.0495505228931676e-05,
+      "loss": 0.4482,
+      "step": 851
+    },
+    {
+      "epoch": 3.451139240506329,
+      "grad_norm": 0.26218171311534294,
+      "learning_rate": 1.044560362975474e-05,
+      "loss": 0.4856,
+      "step": 852
+    },
+    {
+      "epoch": 3.4551898734177215,
+      "grad_norm": 0.237534168071189,
+      "learning_rate": 1.0395778980372695e-05,
+      "loss": 0.4387,
+      "step": 853
+    },
+    {
+      "epoch": 3.459240506329114,
+      "grad_norm": 0.25758903840662645,
+      "learning_rate": 1.0346031682066381e-05,
+      "loss": 0.4451,
+      "step": 854
+    },
+    {
+      "epoch": 3.463291139240506,
+      "grad_norm": 0.24177341523644627,
+      "learning_rate": 1.0296362135493724e-05,
+      "loss": 0.451,
+      "step": 855
+    },
+    {
+      "epoch": 3.4673417721518986,
+      "grad_norm": 0.24090450317808407,
+      "learning_rate": 1.0246770740686422e-05,
+      "loss": 0.4552,
+      "step": 856
+    },
+    {
+      "epoch": 3.471392405063291,
+      "grad_norm": 0.22778722312438937,
+      "learning_rate": 1.0197257897046743e-05,
+      "loss": 0.4322,
+      "step": 857
+    },
+    {
+      "epoch": 3.4754430379746837,
+      "grad_norm": 0.22007436083903384,
+      "learning_rate": 1.014782400334433e-05,
+      "loss": 0.4474,
+      "step": 858
+    },
+    {
+      "epoch": 3.479493670886076,
+      "grad_norm": 0.22936013406460887,
+      "learning_rate": 1.009846945771296e-05,
+      "loss": 0.4393,
+      "step": 859
+    },
+    {
+      "epoch": 3.4835443037974683,
+      "grad_norm": 0.2286553140947832,
+      "learning_rate": 1.0049194657647363e-05,
+      "loss": 0.4436,
+      "step": 860
+    },
+    {
+      "epoch": 3.487594936708861,
+      "grad_norm": 1.1860415338805586,
+      "learning_rate": 1.0000000000000006e-05,
+      "loss": 0.4415,
+      "step": 861
+    },
+    {
+      "epoch": 3.4916455696202533,
+      "grad_norm": 0.2307609089206592,
+      "learning_rate": 9.950885880977891e-06,
+      "loss": 0.4426,
+      "step": 862
+    },
+    {
+      "epoch": 3.4956962025316454,
+      "grad_norm": 0.23044705478261443,
+      "learning_rate": 9.901852696139382e-06,
+      "loss": 0.4401,
+      "step": 863
+    },
+    {
+      "epoch": 3.499746835443038,
+      "grad_norm": 0.21755381569653534,
+      "learning_rate": 9.852900840391027e-06,
+      "loss": 0.429,
+      "step": 864
+    },
+    {
+      "epoch": 3.5037974683544304,
+      "grad_norm": 4.14045927060773,
+      "learning_rate": 9.804030707984313e-06,
+      "loss": 0.4751,
+      "step": 865
+    },
+    {
+      "epoch": 3.507848101265823,
+      "grad_norm": 0.24509790422153643,
+      "learning_rate": 9.755242692512599e-06,
+      "loss": 0.4185,
+      "step": 866
+    },
+    {
+      "epoch": 3.5118987341772154,
+      "grad_norm": 0.23505080444173176,
+      "learning_rate": 9.70653718690782e-06,
+      "loss": 0.4536,
+      "step": 867
+    },
+    {
+      "epoch": 3.5159493670886075,
+      "grad_norm": 0.2069297170195969,
+      "learning_rate": 9.657914583437454e-06,
+      "loss": 0.4397,
+      "step": 868
+    },
+    {
+      "epoch": 3.52,
+      "grad_norm": 0.2159664323435058,
+      "learning_rate": 9.609375273701246e-06,
+      "loss": 0.4229,
+      "step": 869
+    },
+    {
+      "epoch": 3.5240506329113925,
+      "grad_norm": 0.23715535204866214,
+      "learning_rate": 9.560919648628133e-06,
+      "loss": 0.4436,
+      "step": 870
+    },
+    {
+      "epoch": 3.5281012658227846,
+      "grad_norm": 0.21919965790097823,
+      "learning_rate": 9.512548098473047e-06,
+      "loss": 0.43,
+      "step": 871
+    },
+    {
+      "epoch": 3.532151898734177,
+      "grad_norm": 0.3532258867326395,
+      "learning_rate": 9.464261012813825e-06,
+      "loss": 0.4318,
+      "step": 872
+    },
+    {
+      "epoch": 3.5362025316455696,
+      "grad_norm": 0.22883148832946118,
+      "learning_rate": 9.416058780547987e-06,
+      "loss": 0.4449,
+      "step": 873
+    },
+    {
+      "epoch": 3.540253164556962,
+      "grad_norm": 0.23303757426386837,
+      "learning_rate": 9.367941789889714e-06,
+      "loss": 0.4443,
+      "step": 874
+    },
+    {
+      "epoch": 3.5443037974683547,
+      "grad_norm": 0.20994715359250835,
+      "learning_rate": 9.319910428366607e-06,
+      "loss": 0.4588,
+      "step": 875
+    },
+    {
+      "epoch": 3.5483544303797467,
+      "grad_norm": 0.22479848263389648,
+      "learning_rate": 9.271965082816667e-06,
+      "loss": 0.4537,
+      "step": 876
+    },
+    {
+      "epoch": 3.5524050632911393,
+      "grad_norm": 0.21728741406491808,
+      "learning_rate": 9.224106139385111e-06,
+      "loss": 0.4378,
+      "step": 877
+    },
+    {
+      "epoch": 3.5564556962025318,
+      "grad_norm": 0.22975281235470255,
+      "learning_rate": 9.176333983521291e-06,
+      "loss": 0.4358,
+      "step": 878
+    },
+    {
+      "epoch": 3.560506329113924,
+      "grad_norm": 0.22206280896333408,
+      "learning_rate": 9.12864899997558e-06,
+      "loss": 0.4524,
+      "step": 879
+    },
+    {
+      "epoch": 3.5645569620253164,
+      "grad_norm": 0.206969306676907,
+      "learning_rate": 9.08105157279628e-06,
+      "loss": 0.4396,
+      "step": 880
+    },
+    {
+      "epoch": 3.568607594936709,
+      "grad_norm": 0.23515691794684046,
+      "learning_rate": 9.03354208532653e-06,
+      "loss": 0.4319,
+      "step": 881
+    },
+    {
+      "epoch": 3.5726582278481014,
+      "grad_norm": 0.2160351882399224,
+      "learning_rate": 8.986120920201205e-06,
+      "loss": 0.454,
+      "step": 882
+    },
+    {
+      "epoch": 3.576708860759494,
+      "grad_norm": 0.25279461596232977,
+      "learning_rate": 8.938788459343852e-06,
+      "loss": 0.4328,
+      "step": 883
+    },
+    {
+      "epoch": 3.580759493670886,
+      "grad_norm": 0.2174897882358163,
+      "learning_rate": 8.8915450839636e-06,
+      "loss": 0.4385,
+      "step": 884
+    },
+    {
+      "epoch": 3.5848101265822785,
+      "grad_norm": 0.22214439794119725,
+      "learning_rate": 8.844391174552116e-06,
+      "loss": 0.4489,
+      "step": 885
+    },
+    {
+      "epoch": 3.588860759493671,
+      "grad_norm": 0.21723149512286818,
+      "learning_rate": 8.797327110880479e-06,
+      "loss": 0.4301,
+      "step": 886
+    },
+    {
+      "epoch": 3.592911392405063,
+      "grad_norm": 0.23002160088875379,
+      "learning_rate": 8.750353271996206e-06,
+      "loss": 0.4319,
+      "step": 887
+    },
+    {
+      "epoch": 3.5969620253164556,
+      "grad_norm": 0.23688780224298792,
+      "learning_rate": 8.703470036220132e-06,
+      "loss": 0.4426,
+      "step": 888
+    },
+    {
+      "epoch": 3.601012658227848,
+      "grad_norm": 0.23048264902547552,
+      "learning_rate": 8.656677781143394e-06,
+      "loss": 0.4414,
+      "step": 889
+    },
+    {
+      "epoch": 3.6050632911392406,
+      "grad_norm": 0.22624461979386218,
+      "learning_rate": 8.609976883624377e-06,
+      "loss": 0.4628,
+      "step": 890
+    },
+    {
+      "epoch": 3.609113924050633,
+      "grad_norm": 0.23455092656052673,
+      "learning_rate": 8.563367719785698e-06,
+      "loss": 0.4373,
+      "step": 891
+    },
+    {
+      "epoch": 3.613164556962025,
+      "grad_norm": 0.2515170792446118,
+      "learning_rate": 8.516850665011138e-06,
+      "loss": 0.4401,
+      "step": 892
+    },
+    {
+      "epoch": 3.6172151898734177,
+      "grad_norm": 0.2115296208037574,
+      "learning_rate": 8.47042609394269e-06,
+      "loss": 0.4495,
+      "step": 893
+    },
+    {
+      "epoch": 3.6212658227848102,
+      "grad_norm": 0.2401442491005213,
+      "learning_rate": 8.424094380477432e-06,
+      "loss": 0.4422,
+      "step": 894
+    },
+    {
+      "epoch": 3.6253164556962023,
+      "grad_norm": 0.7682954624302767,
+      "learning_rate": 8.37785589776465e-06,
+      "loss": 0.4548,
+      "step": 895
+    },
+    {
+      "epoch": 3.629367088607595,
+      "grad_norm": 0.21437882742820943,
+      "learning_rate": 8.331711018202694e-06,
+      "loss": 0.4447,
+      "step": 896
+    },
+    {
+      "epoch": 3.6334177215189873,
+      "grad_norm": 0.21117472450611,
+      "learning_rate": 8.285660113436104e-06,
+      "loss": 0.4415,
+      "step": 897
+    },
+    {
+      "epoch": 3.63746835443038,
+      "grad_norm": 0.2573159365120494,
+      "learning_rate": 8.239703554352527e-06,
+      "loss": 0.4583,
+      "step": 898
+    },
+    {
+      "epoch": 3.6415189873417724,
+      "grad_norm": 0.22701329050542945,
+      "learning_rate": 8.193841711079775e-06,
+      "loss": 0.4375,
+      "step": 899
+    },
+    {
+      "epoch": 3.6455696202531644,
+      "grad_norm": 0.21846256770800526,
+      "learning_rate": 8.148074952982828e-06,
+      "loss": 0.4336,
+      "step": 900
+    },
+    {
+      "epoch": 3.649620253164557,
+      "grad_norm": 0.22447198374345922,
+      "learning_rate": 8.102403648660859e-06,
+      "loss": 0.4313,
+      "step": 901
+    },
+    {
+      "epoch": 3.6536708860759495,
+      "grad_norm": 0.22778809631781957,
+      "learning_rate": 8.056828165944282e-06,
+      "loss": 0.4466,
+      "step": 902
+    },
+    {
+      "epoch": 3.6577215189873415,
+      "grad_norm": 0.2294390723341458,
+      "learning_rate": 8.011348871891762e-06,
+      "loss": 0.4529,
+      "step": 903
+    },
+    {
+      "epoch": 3.661772151898734,
+      "grad_norm": 0.22367446784667905,
+      "learning_rate": 7.965966132787287e-06,
+      "loss": 0.4515,
+      "step": 904
+    },
+    {
+      "epoch": 3.6658227848101266,
+      "grad_norm": 0.22061179784890464,
+      "learning_rate": 7.920680314137189e-06,
+      "loss": 0.4431,
+      "step": 905
+    },
+    {
+      "epoch": 3.669873417721519,
+      "grad_norm": 0.23954012060933585,
+      "learning_rate": 7.875491780667246e-06,
+      "loss": 0.4402,
+      "step": 906
+    },
+    {
+      "epoch": 3.6739240506329116,
+      "grad_norm": 0.23354253821842086,
+      "learning_rate": 7.830400896319667e-06,
+      "loss": 0.4387,
+      "step": 907
+    },
+    {
+      "epoch": 3.6779746835443037,
+      "grad_norm": 0.22339929583068074,
+      "learning_rate": 7.785408024250259e-06,
+      "loss": 0.4307,
+      "step": 908
+    },
+    {
+      "epoch": 3.682025316455696,
+      "grad_norm": 0.23454056042784424,
+      "learning_rate": 7.74051352682542e-06,
+      "loss": 0.4351,
+      "step": 909
+    },
+    {
+      "epoch": 3.6860759493670887,
+      "grad_norm": 0.2140394421670661,
+      "learning_rate": 7.695717765619257e-06,
+      "loss": 0.4422,
+      "step": 910
+    },
+    {
+      "epoch": 3.690126582278481,
+      "grad_norm": 0.2423559045574754,
+      "learning_rate": 7.651021101410673e-06,
+      "loss": 0.4342,
+      "step": 911
+    },
+    {
+      "epoch": 3.6941772151898733,
+      "grad_norm": 0.23141321022943273,
+      "learning_rate": 7.606423894180464e-06,
+      "loss": 0.4453,
+      "step": 912
+    },
+    {
+      "epoch": 3.698227848101266,
+      "grad_norm": 0.2303831898623387,
+      "learning_rate": 7.56192650310839e-06,
+      "loss": 0.4363,
+      "step": 913
+    },
+    {
+      "epoch": 3.7022784810126583,
+      "grad_norm": 0.2379463617105292,
+      "learning_rate": 7.517529286570349e-06,
+      "loss": 0.4532,
+      "step": 914
+    },
+    {
+      "epoch": 3.706329113924051,
+      "grad_norm": 0.23798481210775274,
+      "learning_rate": 7.473232602135387e-06,
+      "loss": 0.444,
+      "step": 915
+    },
+    {
+      "epoch": 3.710379746835443,
+      "grad_norm": 0.21620803698657426,
+      "learning_rate": 7.429036806562935e-06,
+      "loss": 0.4453,
+      "step": 916
+    },
+    {
+      "epoch": 3.7144303797468354,
+      "grad_norm": 0.23003101132890638,
+      "learning_rate": 7.3849422557998455e-06,
+      "loss": 0.45,
+      "step": 917
+    },
+    {
+      "epoch": 3.718481012658228,
+      "grad_norm": 0.25673451504239553,
+      "learning_rate": 7.340949304977567e-06,
+      "loss": 0.4513,
+      "step": 918
+    },
+    {
+      "epoch": 3.72253164556962,
+      "grad_norm": 0.23423564426169666,
+      "learning_rate": 7.297058308409282e-06,
+      "loss": 0.4477,
+      "step": 919
+    },
+    {
+      "epoch": 3.7265822784810125,
+      "grad_norm": 0.22503735135894054,
+      "learning_rate": 7.25326961958704e-06,
+      "loss": 0.4278,
+      "step": 920
+    },
+    {
+      "epoch": 3.730632911392405,
+      "grad_norm": 0.23443123573603264,
+      "learning_rate": 7.209583591178921e-06,
+      "loss": 0.4461,
+      "step": 921
+    },
+    {
+      "epoch": 3.7346835443037976,
+      "grad_norm": 0.2366011800858292,
+      "learning_rate": 7.1660005750261925e-06,
+      "loss": 0.4428,
+      "step": 922
+    },
+    {
+      "epoch": 3.73873417721519,
+      "grad_norm": 0.22028942179066022,
+      "learning_rate": 7.1225209221404765e-06,
+      "loss": 0.4566,
+      "step": 923
+    },
+    {
+      "epoch": 3.742784810126582,
+      "grad_norm": 0.2197533680808418,
+      "learning_rate": 7.079144982700909e-06,
+      "loss": 0.4336,
+      "step": 924
+    },
+    {
+      "epoch": 3.7468354430379747,
+      "grad_norm": 0.34670351691687024,
+      "learning_rate": 7.0358731060513695e-06,
+      "loss": 0.4349,
+      "step": 925
+    },
+    {
+      "epoch": 3.750886075949367,
+      "grad_norm": 0.22360450880073382,
+      "learning_rate": 6.99270564069757e-06,
+      "loss": 0.4388,
+      "step": 926
+    },
+    {
+      "epoch": 3.7549367088607593,
+      "grad_norm": 0.22192933895167635,
+      "learning_rate": 6.949642934304375e-06,
+      "loss": 0.4345,
+      "step": 927
+    },
+    {
+      "epoch": 3.7589873417721518,
+      "grad_norm": 0.22120484597553983,
+      "learning_rate": 6.906685333692871e-06,
+      "loss": 0.4552,
+      "step": 928
+    },
+    {
+      "epoch": 3.7630379746835443,
+      "grad_norm": 6.18368016891881,
+      "learning_rate": 6.86383318483769e-06,
+      "loss": 0.4913,
+      "step": 929
+    },
+    {
+      "epoch": 3.767088607594937,
+      "grad_norm": 0.2459859713024362,
+      "learning_rate": 6.821086832864139e-06,
+      "loss": 0.4445,
+      "step": 930
+    },
+    {
+      "epoch": 3.7711392405063293,
+      "grad_norm": 0.23534546172339393,
+      "learning_rate": 6.77844662204546e-06,
+      "loss": 0.4424,
+      "step": 931
+    },
+    {
+      "epoch": 3.7751898734177214,
+      "grad_norm": 0.2207223145790432,
+      "learning_rate": 6.7359128958000455e-06,
+      "loss": 0.4375,
+      "step": 932
+    },
+    {
+      "epoch": 3.779240506329114,
+      "grad_norm": 0.23196292731063067,
+      "learning_rate": 6.693485996688695e-06,
+      "loss": 0.4693,
+      "step": 933
+    },
+    {
+      "epoch": 3.7832911392405064,
+      "grad_norm": 0.21443055064839212,
+      "learning_rate": 6.651166266411801e-06,
+      "loss": 0.4429,
+      "step": 934
+    },
+    {
+      "epoch": 3.7873417721518985,
+      "grad_norm": 0.2269346284192372,
+      "learning_rate": 6.6089540458066725e-06,
+      "loss": 0.4592,
+      "step": 935
+    },
+    {
+      "epoch": 3.791392405063291,
+      "grad_norm": 0.24359484998833658,
+      "learning_rate": 6.566849674844711e-06,
+      "loss": 0.4369,
+      "step": 936
+    },
+    {
+      "epoch": 3.7954430379746835,
+      "grad_norm": 0.2175012753830777,
+      "learning_rate": 6.524853492628747e-06,
+      "loss": 0.4288,
+      "step": 937
+    },
+    {
+      "epoch": 3.799493670886076,
+      "grad_norm": 0.21613655652227481,
+      "learning_rate": 6.4829658373902536e-06,
+      "loss": 0.4313,
+      "step": 938
+    },
+    {
+      "epoch": 3.8035443037974686,
+      "grad_norm": 0.2264199936358381,
+      "learning_rate": 6.441187046486648e-06,
+      "loss": 0.4417,
+      "step": 939
+    },
+    {
+      "epoch": 3.8075949367088606,
+      "grad_norm": 0.24855183751668666,
+      "learning_rate": 6.399517456398567e-06,
+      "loss": 0.4466,
+      "step": 940
+    },
+    {
+      "epoch": 3.811645569620253,
+      "grad_norm": 0.21470128270241134,
+      "learning_rate": 6.357957402727164e-06,
+      "loss": 0.427,
+      "step": 941
+    },
+    {
+      "epoch": 3.8156962025316457,
+      "grad_norm": 0.22866449092510105,
+      "learning_rate": 6.316507220191395e-06,
+      "loss": 0.4314,
+      "step": 942
+    },
+    {
+      "epoch": 3.8197468354430377,
+      "grad_norm": 0.2354262936682115,
+      "learning_rate": 6.275167242625331e-06,
+      "loss": 0.4436,
+      "step": 943
+    },
+    {
+      "epoch": 3.8237974683544302,
+      "grad_norm": 0.22152397528684803,
+      "learning_rate": 6.233937802975471e-06,
+      "loss": 0.4396,
+      "step": 944
+    },
+    {
+      "epoch": 3.8278481012658228,
+      "grad_norm": 0.2125091082816343,
+      "learning_rate": 6.192819233298046e-06,
+      "loss": 0.4476,
+      "step": 945
+    },
+    {
+      "epoch": 3.8318987341772153,
+      "grad_norm": 0.21842653031719708,
+      "learning_rate": 6.151811864756383e-06,
+      "loss": 0.4322,
+      "step": 946
+    },
+    {
+      "epoch": 3.835949367088608,
+      "grad_norm": 0.2223962461804476,
+      "learning_rate": 6.1109160276181655e-06,
+      "loss": 0.4429,
+      "step": 947
+    },
+    {
+      "epoch": 3.84,
+      "grad_norm": 0.20938933670792725,
+      "learning_rate": 6.070132051252868e-06,
+      "loss": 0.4451,
+      "step": 948
+    },
+    {
+      "epoch": 3.8440506329113924,
+      "grad_norm": 0.2075073427261809,
+      "learning_rate": 6.0294602641290034e-06,
+      "loss": 0.4239,
+      "step": 949
+    },
+    {
+      "epoch": 3.848101265822785,
+      "grad_norm": 0.22486211157664104,
+      "learning_rate": 5.988900993811575e-06,
+      "loss": 0.4382,
+      "step": 950
+    },
+    {
+      "epoch": 3.852151898734177,
+      "grad_norm": 0.21583568581971158,
+      "learning_rate": 5.948454566959363e-06,
+      "loss": 0.4429,
+      "step": 951
+    },
+    {
+      "epoch": 3.8562025316455695,
+      "grad_norm": 0.23094939824977917,
+      "learning_rate": 5.908121309322328e-06,
+      "loss": 0.4475,
+      "step": 952
+    },
+    {
+      "epoch": 3.860253164556962,
+      "grad_norm": 0.24082490567634832,
+      "learning_rate": 5.867901545738976e-06,
+      "loss": 0.4595,
+      "step": 953
+    },
+    {
+      "epoch": 3.8643037974683545,
+      "grad_norm": 0.21441419017355598,
+      "learning_rate": 5.827795600133774e-06,
+      "loss": 0.4419,
+      "step": 954
+    },
+    {
+      "epoch": 3.868354430379747,
+      "grad_norm": 0.22935797899476443,
+      "learning_rate": 5.787803795514466e-06,
+      "loss": 0.4438,
+      "step": 955
+    },
+    {
+      "epoch": 3.872405063291139,
+      "grad_norm": 0.20837718385923812,
+      "learning_rate": 5.747926453969576e-06,
+      "loss": 0.4493,
+      "step": 956
+    },
+    {
+      "epoch": 3.8764556962025316,
+      "grad_norm": 0.20821737793652306,
+      "learning_rate": 5.708163896665708e-06,
+      "loss": 0.4463,
+      "step": 957
+    },
+    {
+      "epoch": 3.880506329113924,
+      "grad_norm": 0.2186915555624138,
+      "learning_rate": 5.668516443845047e-06,
+      "loss": 0.4443,
+      "step": 958
+    },
+    {
+      "epoch": 3.884556962025316,
+      "grad_norm": 0.21691222183606465,
+      "learning_rate": 5.6289844148227225e-06,
+      "loss": 0.4526,
+      "step": 959
+    },
+    {
+      "epoch": 3.8886075949367087,
+      "grad_norm": 0.6672467948135697,
+      "learning_rate": 5.5895681279842615e-06,
+      "loss": 0.4689,
+      "step": 960
+    },
+    {
+      "epoch": 3.8926582278481012,
+      "grad_norm": 0.21932889999108723,
+      "learning_rate": 5.550267900783019e-06,
+      "loss": 0.4424,
+      "step": 961
+    },
+    {
+      "epoch": 3.8967088607594937,
+      "grad_norm": 0.21877734876099386,
+      "learning_rate": 5.511084049737623e-06,
+      "loss": 0.4512,
+      "step": 962
+    },
+    {
+      "epoch": 3.9007594936708863,
+      "grad_norm": 0.20416259815124008,
+      "learning_rate": 5.4720168904294215e-06,
+      "loss": 0.439,
+      "step": 963
+    },
+    {
+      "epoch": 3.9048101265822783,
+      "grad_norm": 0.21205476807990029,
+      "learning_rate": 5.433066737499948e-06,
+      "loss": 0.445,
+      "step": 964
+    },
+    {
+      "epoch": 3.908860759493671,
+      "grad_norm": 0.21110709235906988,
+      "learning_rate": 5.394233904648376e-06,
+      "loss": 0.4414,
+      "step": 965
+    },
+    {
+      "epoch": 3.9129113924050634,
+      "grad_norm": 0.2163977363997883,
+      "learning_rate": 5.355518704628997e-06,
+      "loss": 0.4317,
+      "step": 966
+    },
+    {
+      "epoch": 3.9169620253164554,
+      "grad_norm": 0.2144141126544641,
+      "learning_rate": 5.316921449248731e-06,
+      "loss": 0.4373,
+      "step": 967
+    },
+    {
+      "epoch": 3.921012658227848,
+      "grad_norm": 0.20139142686791472,
+      "learning_rate": 5.278442449364538e-06,
+      "loss": 0.4544,
+      "step": 968
+    },
+    {
+      "epoch": 3.9250632911392405,
+      "grad_norm": 0.20064564900998758,
+      "learning_rate": 5.240082014881016e-06,
+      "loss": 0.4401,
+      "step": 969
+    },
+    {
+      "epoch": 3.929113924050633,
+      "grad_norm": 0.20403369069177404,
+      "learning_rate": 5.201840454747822e-06,
+      "loss": 0.4368,
+      "step": 970
+    },
+    {
+      "epoch": 3.9331645569620255,
+      "grad_norm": 0.2128976668612367,
+      "learning_rate": 5.163718076957223e-06,
+      "loss": 0.4279,
+      "step": 971
+    },
+    {
+      "epoch": 3.9372151898734176,
+      "grad_norm": 0.1978395021015446,
+      "learning_rate": 5.125715188541609e-06,
+      "loss": 0.439,
+      "step": 972
+    },
+    {
+      "epoch": 3.94126582278481,
+      "grad_norm": 0.2003192164245282,
+      "learning_rate": 5.087832095571021e-06,
+      "loss": 0.4409,
+      "step": 973
+    },
+    {
+      "epoch": 3.9453164556962026,
+      "grad_norm": 0.19382960475631986,
+      "learning_rate": 5.0500691031506766e-06,
+      "loss": 0.4446,
+      "step": 974
+    },
+    {
+      "epoch": 3.9493670886075947,
+      "grad_norm": 0.20660973786993383,
+      "learning_rate": 5.01242651541854e-06,
+      "loss": 0.4422,
+      "step": 975
+    },
+    {
+      "epoch": 3.953417721518987,
+      "grad_norm": 0.21560699953908694,
+      "learning_rate": 4.974904635542815e-06,
+      "loss": 0.4377,
+      "step": 976
+    },
+    {
+      "epoch": 3.9574683544303797,
+      "grad_norm": 0.20299838144352902,
+      "learning_rate": 4.937503765719582e-06,
+      "loss": 0.4426,
+      "step": 977
+    },
+    {
+      "epoch": 3.961518987341772,
+      "grad_norm": 0.20379522658039817,
+      "learning_rate": 4.900224207170299e-06,
+      "loss": 0.4367,
+      "step": 978
+    },
+    {
+      "epoch": 3.9655696202531647,
+      "grad_norm": 0.20749969073608496,
+      "learning_rate": 4.8630662601394065e-06,
+      "loss": 0.439,
+      "step": 979
+    },
+    {
+      "epoch": 3.969620253164557,
+      "grad_norm": 0.20949675599536077,
+      "learning_rate": 4.8260302238918995e-06,
+      "loss": 0.4483,
+      "step": 980
+    },
+    {
+      "epoch": 3.9736708860759493,
+      "grad_norm": 0.2016351629807236,
+      "learning_rate": 4.789116396710924e-06,
+      "loss": 0.4362,
+      "step": 981
+    },
+    {
+      "epoch": 3.977721518987342,
+      "grad_norm": 0.19445217219638622,
+      "learning_rate": 4.752325075895368e-06,
+      "loss": 0.4467,
+      "step": 982
+    },
+    {
+      "epoch": 3.981772151898734,
+      "grad_norm": 0.2077463312340491,
+      "learning_rate": 4.715656557757473e-06,
+      "loss": 0.4423,
+      "step": 983
+    },
+    {
+      "epoch": 3.9858227848101264,
+      "grad_norm": 0.2106093404492608,
+      "learning_rate": 4.679111137620442e-06,
+      "loss": 0.445,
+      "step": 984
+    },
+    {
+      "epoch": 3.989873417721519,
+      "grad_norm": 5.117194238958409,
+      "learning_rate": 4.6426891098160585e-06,
+      "loss": 0.6588,
+      "step": 985
+    },
+    {
+      "epoch": 3.9939240506329114,
+      "grad_norm": 0.2008440914894743,
+      "learning_rate": 4.6063907676823474e-06,
+      "loss": 0.4548,
+      "step": 986
+    },
+    {
+      "epoch": 3.997974683544304,
+      "grad_norm": 0.2047926899621533,
+      "learning_rate": 4.570216403561141e-06,
+      "loss": 0.4522,
+      "step": 987
+    },
+    {
+      "epoch": 4.01620253164557,
+      "grad_norm": 0.3260460763054243,
+      "learning_rate": 4.534166308795815e-06,
+      "loss": 0.4219,
+      "step": 988
+    },
+    {
+      "epoch": 4.020253164556962,
+      "grad_norm": 0.30173133745466063,
+      "learning_rate": 4.498240773728859e-06,
+      "loss": 0.401,
+      "step": 989
+    },
+    {
+      "epoch": 4.024303797468354,
+      "grad_norm": 0.23193971436086427,
+      "learning_rate": 4.462440087699609e-06,
+      "loss": 0.4053,
+      "step": 990
+    },
+    {
+      "epoch": 4.028354430379747,
+      "grad_norm": 0.34270752895071893,
+      "learning_rate": 4.426764539041861e-06,
+      "loss": 0.3959,
+      "step": 991
+    },
+    {
+      "epoch": 4.032405063291139,
+      "grad_norm": 0.37680117484142106,
+      "learning_rate": 4.391214415081582e-06,
+      "loss": 0.4083,
+      "step": 992
+    },
+    {
+      "epoch": 4.036455696202531,
+      "grad_norm": 0.25183626907209405,
+      "learning_rate": 4.355790002134579e-06,
+      "loss": 0.3912,
+      "step": 993
+    },
+    {
+      "epoch": 4.040506329113924,
+      "grad_norm": 0.2993714059514536,
+      "learning_rate": 4.320491585504207e-06,
+      "loss": 0.4061,
+      "step": 994
+    },
+    {
+      "epoch": 4.044556962025316,
+      "grad_norm": 0.32791068066207724,
+      "learning_rate": 4.2853194494790615e-06,
+      "loss": 0.403,
+      "step": 995
+    },
+    {
+      "epoch": 4.048607594936709,
+      "grad_norm": 0.2699269240063089,
+      "learning_rate": 4.250273877330691e-06,
+      "loss": 0.4107,
+      "step": 996
+    },
+    {
+      "epoch": 4.052658227848101,
+      "grad_norm": 0.23869130014983705,
+      "learning_rate": 4.215355151311313e-06,
+      "loss": 0.3944,
+      "step": 997
+    },
+    {
+      "epoch": 4.056708860759493,
+      "grad_norm": 0.25120075702117245,
+      "learning_rate": 4.180563552651542e-06,
+      "loss": 0.3909,
+      "step": 998
+    },
+    {
+      "epoch": 4.060759493670886,
+      "grad_norm": 0.29857486204828204,
+      "learning_rate": 4.145899361558147e-06,
+      "loss": 0.4175,
+      "step": 999
+    },
+    {
+      "epoch": 4.0648101265822785,
+      "grad_norm": 0.25501430470270187,
+      "learning_rate": 4.111362857211738e-06,
+      "loss": 0.413,
+      "step": 1000
+    },
+    {
+      "epoch": 4.0688607594936705,
+      "grad_norm": 0.23291756317386667,
+      "learning_rate": 4.076954317764592e-06,
+      "loss": 0.4018,
+      "step": 1001
+    },
+    {
+      "epoch": 4.0729113924050635,
+      "grad_norm": 0.2492743053369655,
+      "learning_rate": 4.042674020338335e-06,
+      "loss": 0.3987,
+      "step": 1002
+    },
+    {
+      "epoch": 4.076962025316456,
+      "grad_norm": 0.23365767125560172,
+      "learning_rate": 4.0085222410217835e-06,
+      "loss": 0.4108,
+      "step": 1003
+    },
+    {
+      "epoch": 4.0810126582278485,
+      "grad_norm": 0.23690048365018232,
+      "learning_rate": 3.974499254868674e-06,
+      "loss": 0.3985,
+      "step": 1004
+    },
+    {
+      "epoch": 4.085063291139241,
+      "grad_norm": 0.22548731325554242,
+      "learning_rate": 3.940605335895451e-06,
+      "loss": 0.41,
+      "step": 1005
+    },
+    {
+      "epoch": 4.089113924050633,
+      "grad_norm": 0.2360839612636157,
+      "learning_rate": 3.90684075707908e-06,
+      "loss": 0.403,
+      "step": 1006
+    },
+    {
+      "epoch": 4.093164556962026,
+      "grad_norm": 0.2268961171533964,
+      "learning_rate": 3.8732057903548505e-06,
+      "loss": 0.4094,
+      "step": 1007
+    },
+    {
+      "epoch": 4.097215189873418,
+      "grad_norm": 0.24492269906716393,
+      "learning_rate": 3.8397007066141375e-06,
+      "loss": 0.4042,
+      "step": 1008
+    },
+    {
+      "epoch": 4.10126582278481,
+      "grad_norm": 0.23052575044742193,
+      "learning_rate": 3.806325775702304e-06,
+      "loss": 0.4051,
+      "step": 1009
+    },
+    {
+      "epoch": 4.105316455696203,
+      "grad_norm": 0.21676914308820144,
+      "learning_rate": 3.773081266416434e-06,
+      "loss": 0.4087,
+      "step": 1010
+    },
+    {
+      "epoch": 4.109367088607595,
+      "grad_norm": 0.21573472975726823,
+      "learning_rate": 3.739967446503245e-06,
+      "loss": 0.408,
+      "step": 1011
+    },
+    {
+      "epoch": 4.113417721518988,
+      "grad_norm": 0.2137718158603819,
+      "learning_rate": 3.706984582656894e-06,
+      "loss": 0.4106,
+      "step": 1012
+    },
+    {
+      "epoch": 4.11746835443038,
+      "grad_norm": 0.22371445222724046,
+      "learning_rate": 3.6741329405168237e-06,
+      "loss": 0.4118,
+      "step": 1013
+    },
+    {
+      "epoch": 4.121518987341772,
+      "grad_norm": 0.2185639778042743,
+      "learning_rate": 3.641412784665648e-06,
+      "loss": 0.4115,
+      "step": 1014
+    },
+    {
+      "epoch": 4.125569620253165,
+      "grad_norm": 0.20547407227545605,
+      "learning_rate": 3.608824378627005e-06,
+      "loss": 0.4127,
+      "step": 1015
+    },
+    {
+      "epoch": 4.129620253164557,
+      "grad_norm": 0.2059423168293545,
+      "learning_rate": 3.5763679848634337e-06,
+      "loss": 0.4044,
+      "step": 1016
+    },
+    {
+      "epoch": 4.133670886075949,
+      "grad_norm": 0.21905853461315428,
+      "learning_rate": 3.544043864774269e-06,
+      "loss": 0.3939,
+      "step": 1017
+    },
+    {
+      "epoch": 4.137721518987342,
+      "grad_norm": 0.21424354472867468,
+      "learning_rate": 3.5118522786935282e-06,
+      "loss": 0.3963,
+      "step": 1018
+    },
+    {
+      "epoch": 4.141772151898734,
+      "grad_norm": 0.22588593384630537,
+      "learning_rate": 3.479793485887819e-06,
+      "loss": 0.4075,
+      "step": 1019
+    },
+    {
+      "epoch": 4.145822784810127,
+      "grad_norm": 0.20328577472722054,
+      "learning_rate": 3.4478677445542653e-06,
+      "loss": 0.4107,
+      "step": 1020
+    },
+    {
+      "epoch": 4.149873417721519,
+      "grad_norm": 0.21449633843233804,
+      "learning_rate": 3.4160753118183767e-06,
+      "loss": 0.4113,
+      "step": 1021
+    },
+    {
+      "epoch": 4.153924050632911,
+      "grad_norm": 0.22065285498747253,
+      "learning_rate": 3.3844164437320527e-06,
+      "loss": 0.3966,
+      "step": 1022
+    },
+    {
+      "epoch": 4.157974683544304,
+      "grad_norm": 0.20506989523458946,
+      "learning_rate": 3.3528913952714558e-06,
+      "loss": 0.4115,
+      "step": 1023
+    },
+    {
+      "epoch": 4.162025316455696,
+      "grad_norm": 0.20455782878883097,
+      "learning_rate": 3.321500420335e-06,
+      "loss": 0.4123,
+      "step": 1024
+    },
+    {
+      "epoch": 4.166075949367088,
+      "grad_norm": 0.21212125906068993,
+      "learning_rate": 3.290243771741275e-06,
+      "loss": 0.4192,
+      "step": 1025
+    },
+    {
+      "epoch": 4.170126582278481,
+      "grad_norm": 0.21879472691309754,
+      "learning_rate": 3.2591217012270325e-06,
+      "loss": 0.413,
+      "step": 1026
+    },
+    {
+      "epoch": 4.174177215189873,
+      "grad_norm": 0.2039381789442902,
+      "learning_rate": 3.228134459445149e-06,
+      "loss": 0.4005,
+      "step": 1027
+    },
+    {
+      "epoch": 4.178227848101266,
+      "grad_norm": 0.20575882347728885,
+      "learning_rate": 3.1972822959626205e-06,
+      "loss": 0.4087,
+      "step": 1028
+    },
+    {
+      "epoch": 4.182278481012658,
+      "grad_norm": 0.20827476967425498,
+      "learning_rate": 3.166565459258513e-06,
+      "loss": 0.402,
+      "step": 1029
+    },
+    {
+      "epoch": 4.18632911392405,
+      "grad_norm": 0.20271459282080817,
+      "learning_rate": 3.1359841967220193e-06,
+      "loss": 0.4138,
+      "step": 1030
+    },
+    {
+      "epoch": 4.190379746835443,
+      "grad_norm": 0.19905274900476874,
+      "learning_rate": 3.105538754650419e-06,
+      "loss": 0.3984,
+      "step": 1031
+    },
+    {
+      "epoch": 4.194430379746835,
+      "grad_norm": 0.20402033194503427,
+      "learning_rate": 3.07522937824712e-06,
+      "loss": 0.4056,
+      "step": 1032
+    },
+    {
+      "epoch": 4.1984810126582275,
+      "grad_norm": 0.21131547108784288,
+      "learning_rate": 3.0450563116196697e-06,
+      "loss": 0.3953,
+      "step": 1033
+    },
+    {
+      "epoch": 4.2025316455696204,
+      "grad_norm": 0.2157405194363184,
+      "learning_rate": 3.0150197977778008e-06,
+      "loss": 0.4037,
+      "step": 1034
+    },
+    {
+      "epoch": 4.2065822784810125,
+      "grad_norm": 0.203287722556781,
+      "learning_rate": 2.985120078631465e-06,
+      "loss": 0.4103,
+      "step": 1035
+    },
+    {
+      "epoch": 4.2106329113924055,
+      "grad_norm": 0.2081725322385596,
+      "learning_rate": 2.9553573949888893e-06,
+      "loss": 0.4058,
+      "step": 1036
+    },
+    {
+      "epoch": 4.2146835443037975,
+      "grad_norm": 0.19956097920606938,
+      "learning_rate": 2.9257319865546384e-06,
+      "loss": 0.4061,
+      "step": 1037
+    },
+    {
+      "epoch": 4.21873417721519,
+      "grad_norm": 0.20619960715305166,
+      "learning_rate": 2.896244091927678e-06,
+      "loss": 0.4138,
+      "step": 1038
+    },
+    {
+      "epoch": 4.222784810126583,
+      "grad_norm": 0.2066320631715566,
+      "learning_rate": 2.8668939485994584e-06,
+      "loss": 0.401,
+      "step": 1039
+    },
+    {
+      "epoch": 4.226835443037975,
+      "grad_norm": 0.20783934785893607,
+      "learning_rate": 2.837681792951994e-06,
+      "loss": 0.4025,
+      "step": 1040
+    },
+    {
+      "epoch": 4.230886075949367,
+      "grad_norm": 0.20221498467053545,
+      "learning_rate": 2.808607860255981e-06,
+      "loss": 0.4218,
+      "step": 1041
+    },
+    {
+      "epoch": 4.23493670886076,
+      "grad_norm": 0.2043560077620594,
+      "learning_rate": 2.7796723846688634e-06,
+      "loss": 0.4091,
+      "step": 1042
+    },
+    {
+      "epoch": 4.238987341772152,
+      "grad_norm": 0.20994873632947458,
+      "learning_rate": 2.7508755992329937e-06,
+      "loss": 0.3937,
+      "step": 1043
+    },
+    {
+      "epoch": 4.243037974683545,
+      "grad_norm": 0.2034347453247582,
+      "learning_rate": 2.722217735873718e-06,
+      "loss": 0.3998,
+      "step": 1044
+    },
+    {
+      "epoch": 4.247088607594937,
+      "grad_norm": 0.21466440167481066,
+      "learning_rate": 2.6936990253975315e-06,
+      "loss": 0.4013,
+      "step": 1045
+    },
+    {
+      "epoch": 4.251139240506329,
+      "grad_norm": 0.20257916299139916,
+      "learning_rate": 2.665319697490205e-06,
+      "loss": 0.4111,
+      "step": 1046
+    },
+    {
+      "epoch": 4.255189873417722,
+      "grad_norm": 0.20992686249381115,
+      "learning_rate": 2.637079980714945e-06,
+      "loss": 0.4066,
+      "step": 1047
+    },
+    {
+      "epoch": 4.259240506329114,
+      "grad_norm": 0.21455068290035592,
+      "learning_rate": 2.6089801025105453e-06,
+      "loss": 0.3871,
+      "step": 1048
+    },
+    {
+      "epoch": 4.263291139240506,
+      "grad_norm": 0.21046096689439234,
+      "learning_rate": 2.581020289189571e-06,
+      "loss": 0.3983,
+      "step": 1049
+    },
+    {
+      "epoch": 4.267341772151899,
+      "grad_norm": 0.2040377295865848,
+      "learning_rate": 2.553200765936501e-06,
+      "loss": 0.3862,
+      "step": 1050
+    },
+    {
+      "epoch": 4.271392405063291,
+      "grad_norm": 0.20058596076259055,
+      "learning_rate": 2.525521756805962e-06,
+      "loss": 0.4055,
+      "step": 1051
+    },
+    {
+      "epoch": 4.275443037974684,
+      "grad_norm": 0.19775842295914395,
+      "learning_rate": 2.497983484720885e-06,
+      "loss": 0.4026,
+      "step": 1052
+    },
+    {
+      "epoch": 4.279493670886076,
+      "grad_norm": 0.21519319947639992,
+      "learning_rate": 2.470586171470728e-06,
+      "loss": 0.4016,
+      "step": 1053
+    },
+    {
+      "epoch": 4.283544303797468,
+      "grad_norm": 0.20353414307519294,
+      "learning_rate": 2.4433300377096836e-06,
+      "loss": 0.3955,
+      "step": 1054
+    },
+    {
+      "epoch": 4.287594936708861,
+      "grad_norm": 0.19762381543554358,
+      "learning_rate": 2.4162153029549073e-06,
+      "loss": 0.4033,
+      "step": 1055
+    },
+    {
+      "epoch": 4.291645569620253,
+      "grad_norm": 0.18868791147519212,
+      "learning_rate": 2.3892421855847458e-06,
+      "loss": 0.4104,
+      "step": 1056
+    },
+    {
+      "epoch": 4.295696202531645,
+      "grad_norm": 0.19303041888747915,
+      "learning_rate": 2.362410902836978e-06,
+      "loss": 0.4023,
+      "step": 1057
+    },
+    {
+      "epoch": 4.299746835443038,
+      "grad_norm": 0.20945574746989568,
+      "learning_rate": 2.3357216708070653e-06,
+      "loss": 0.3957,
+      "step": 1058
+    },
+    {
+      "epoch": 4.30379746835443,
+      "grad_norm": 0.19440679290268587,
+      "learning_rate": 2.309174704446411e-06,
+      "loss": 0.4037,
+      "step": 1059
+    },
+    {
+      "epoch": 4.307848101265823,
+      "grad_norm": 0.19799146711717355,
+      "learning_rate": 2.2827702175606437e-06,
+      "loss": 0.4018,
+      "step": 1060
+    },
+    {
+      "epoch": 4.311898734177215,
+      "grad_norm": 0.2063848506911879,
+      "learning_rate": 2.256508422807855e-06,
+      "loss": 0.4162,
+      "step": 1061
+    },
+    {
+      "epoch": 4.315949367088607,
+      "grad_norm": 0.19629185202273822,
+      "learning_rate": 2.230389531696946e-06,
+      "loss": 0.415,
+      "step": 1062
+    },
+    {
+      "epoch": 4.32,
+      "grad_norm": 0.20175182140789008,
+      "learning_rate": 2.204413754585857e-06,
+      "loss": 0.4171,
+      "step": 1063
+    },
+    {
+      "epoch": 4.324050632911392,
+      "grad_norm": 0.1924922948286396,
+      "learning_rate": 2.1785813006799406e-06,
+      "loss": 0.4025,
+      "step": 1064
+    },
+    {
+      "epoch": 4.328101265822784,
+      "grad_norm": 0.1951594382589462,
+      "learning_rate": 2.1528923780302224e-06,
+      "loss": 0.4215,
+      "step": 1065
+    },
+    {
+      "epoch": 4.332151898734177,
+      "grad_norm": 0.2081339177726093,
+      "learning_rate": 2.127347193531757e-06,
+      "loss": 0.3992,
+      "step": 1066
+    },
+    {
+      "epoch": 4.3362025316455695,
+      "grad_norm": 0.1983110975694333,
+      "learning_rate": 2.101945952921942e-06,
+      "loss": 0.4127,
+      "step": 1067
+    },
+    {
+      "epoch": 4.340253164556962,
+      "grad_norm": 0.19899124035251442,
+      "learning_rate": 2.0766888607788906e-06,
+      "loss": 0.4062,
+      "step": 1068
+    },
+    {
+      "epoch": 4.3443037974683545,
+      "grad_norm": 0.2023518331536991,
+      "learning_rate": 2.0515761205197337e-06,
+      "loss": 0.3985,
+      "step": 1069
+    },
+    {
+      "epoch": 4.348354430379747,
+      "grad_norm": 3.25126865909323,
+      "learning_rate": 2.0266079343990453e-06,
+      "loss": 0.4128,
+      "step": 1070
+    },
+    {
+      "epoch": 4.3524050632911395,
+      "grad_norm": 0.19417366806592068,
+      "learning_rate": 2.0017845035071494e-06,
+      "loss": 0.4046,
+      "step": 1071
+    },
+    {
+      "epoch": 4.356455696202532,
+      "grad_norm": 0.20972781450325015,
+      "learning_rate": 1.9771060277685537e-06,
+      "loss": 0.404,
+      "step": 1072
+    },
+    {
+      "epoch": 4.360506329113924,
+      "grad_norm": 0.20339994981911935,
+      "learning_rate": 1.95257270594031e-06,
+      "loss": 0.4046,
+      "step": 1073
+    },
+    {
+      "epoch": 4.364556962025317,
+      "grad_norm": 0.19564001031627093,
+      "learning_rate": 1.9281847356104188e-06,
+      "loss": 0.4206,
+      "step": 1074
+    },
+    {
+      "epoch": 4.368607594936709,
+      "grad_norm": 0.19734135230767794,
+      "learning_rate": 1.9039423131962365e-06,
+      "loss": 0.3978,
+      "step": 1075
+    },
+    {
+      "epoch": 4.372658227848102,
+      "grad_norm": 0.19477108829079315,
+      "learning_rate": 1.8798456339429027e-06,
+      "loss": 0.4016,
+      "step": 1076
+    },
+    {
+      "epoch": 4.376708860759494,
+      "grad_norm": 0.19393063139367642,
+      "learning_rate": 1.8558948919217612e-06,
+      "loss": 0.4069,
+      "step": 1077
+    },
+    {
+      "epoch": 4.380759493670886,
+      "grad_norm": 0.1984346796155869,
+      "learning_rate": 1.8320902800287954e-06,
+      "loss": 0.4072,
+      "step": 1078
+    },
+    {
+      "epoch": 4.384810126582279,
+      "grad_norm": 0.1940699647159675,
+      "learning_rate": 1.8084319899830726e-06,
+      "loss": 0.3925,
+      "step": 1079
+    },
+    {
+      "epoch": 4.388860759493671,
+      "grad_norm": 0.3876379618699237,
+      "learning_rate": 1.7849202123252097e-06,
+      "loss": 0.4106,
+      "step": 1080
+    },
+    {
+      "epoch": 4.392911392405063,
+      "grad_norm": 0.19466299436061024,
+      "learning_rate": 1.7615551364158401e-06,
+      "loss": 0.4084,
+      "step": 1081
+    },
+    {
+      "epoch": 4.396962025316456,
+      "grad_norm": 0.18697644014442083,
+      "learning_rate": 1.738336950434061e-06,
+      "loss": 0.4012,
+      "step": 1082
+    },
+    {
+      "epoch": 4.401012658227848,
+      "grad_norm": 0.19210516497130795,
+      "learning_rate": 1.715265841375957e-06,
+      "loss": 0.4119,
+      "step": 1083
+    },
+    {
+      "epoch": 4.405063291139241,
+      "grad_norm": 0.19532915876196374,
+      "learning_rate": 1.6923419950530684e-06,
+      "loss": 0.3981,
+      "step": 1084
+    },
+    {
+      "epoch": 4.409113924050633,
+      "grad_norm": 0.20453689781174117,
+      "learning_rate": 1.6695655960909008e-06,
+      "loss": 0.403,
+      "step": 1085
+    },
+    {
+      "epoch": 4.413164556962025,
+      "grad_norm": 0.20029560395436316,
+      "learning_rate": 1.646936827927441e-06,
+      "loss": 0.3898,
+      "step": 1086
+    },
+    {
+      "epoch": 4.417215189873418,
+      "grad_norm": 0.20627689093029647,
+      "learning_rate": 1.6244558728116766e-06,
+      "loss": 0.3986,
+      "step": 1087
+    },
+    {
+      "epoch": 4.42126582278481,
+      "grad_norm": 0.19312254314442506,
+      "learning_rate": 1.6021229118021265e-06,
+      "loss": 0.4116,
+      "step": 1088
+    },
+    {
+      "epoch": 4.425316455696202,
+      "grad_norm": 0.20488240132618318,
+      "learning_rate": 1.5799381247653967e-06,
+      "loss": 0.4235,
+      "step": 1089
+    },
+    {
+      "epoch": 4.429367088607595,
+      "grad_norm": 0.19817551380696036,
+      "learning_rate": 1.5579016903747013e-06,
+      "loss": 0.4015,
+      "step": 1090
+    },
+    {
+      "epoch": 4.433417721518987,
+      "grad_norm": 0.20128092876564874,
+      "learning_rate": 1.5360137861084656e-06,
+      "loss": 0.3949,
+      "step": 1091
+    },
+    {
+      "epoch": 4.43746835443038,
+      "grad_norm": 0.19182031718352724,
+      "learning_rate": 1.5142745882488475e-06,
+      "loss": 0.4086,
+      "step": 1092
+    },
+    {
+      "epoch": 4.441518987341772,
+      "grad_norm": 0.19359158426784126,
+      "learning_rate": 1.4926842718803691e-06,
+      "loss": 0.4033,
+      "step": 1093
+    },
+    {
+      "epoch": 4.445569620253164,
+      "grad_norm": 0.19784044271308682,
+      "learning_rate": 1.4712430108884657e-06,
+      "loss": 0.3877,
+      "step": 1094
+    },
+    {
+      "epoch": 4.449620253164557,
+      "grad_norm": 0.19391005269990197,
+      "learning_rate": 1.4499509779581078e-06,
+      "loss": 0.4129,
+      "step": 1095
+    },
+    {
+      "epoch": 4.453670886075949,
+      "grad_norm": 0.19089177762161066,
+      "learning_rate": 1.4288083445723988e-06,
+      "loss": 0.4108,
+      "step": 1096
+    },
+    {
+      "epoch": 4.457721518987341,
+      "grad_norm": 0.20652122352406793,
+      "learning_rate": 1.4078152810112045e-06,
+      "loss": 0.3898,
+      "step": 1097
+    },
+    {
+      "epoch": 4.461772151898734,
+      "grad_norm": 0.19178629715815382,
+      "learning_rate": 1.3869719563497697e-06,
+      "loss": 0.4095,
+      "step": 1098
+    },
+    {
+      "epoch": 4.465822784810126,
+      "grad_norm": 3.792973972668285,
+      "learning_rate": 1.3662785384573663e-06,
+      "loss": 0.4411,
+      "step": 1099
+    },
+    {
+      "epoch": 4.469873417721519,
+      "grad_norm": 0.19686849144375046,
+      "learning_rate": 1.3457351939959383e-06,
+      "loss": 0.4052,
+      "step": 1100
+    },
+    {
+      "epoch": 4.473924050632911,
+      "grad_norm": 0.18918831750631715,
+      "learning_rate": 1.3253420884187551e-06,
+      "loss": 0.413,
+      "step": 1101
+    },
+    {
+      "epoch": 4.4779746835443035,
+      "grad_norm": 0.19240588086574215,
+      "learning_rate": 1.3050993859690953e-06,
+      "loss": 0.4032,
+      "step": 1102
+    },
+    {
+      "epoch": 4.4820253164556965,
+      "grad_norm": 0.19200558516302144,
+      "learning_rate": 1.2850072496788869e-06,
+      "loss": 0.4018,
+      "step": 1103
+    },
+    {
+      "epoch": 4.4860759493670885,
+      "grad_norm": 0.18857031901805335,
+      "learning_rate": 1.2650658413674434e-06,
+      "loss": 0.4089,
+      "step": 1104
+    },
+    {
+      "epoch": 4.490126582278481,
+      "grad_norm": 0.2029345279502508,
+      "learning_rate": 1.2452753216401226e-06,
+      "loss": 0.4002,
+      "step": 1105
+    },
+    {
+      "epoch": 4.494177215189874,
+      "grad_norm": 0.19585751586201852,
+      "learning_rate": 1.2256358498870503e-06,
+      "loss": 0.4055,
+      "step": 1106
+    },
+    {
+      "epoch": 4.498227848101266,
+      "grad_norm": 0.1991543157373561,
+      "learning_rate": 1.2061475842818337e-06,
+      "loss": 0.4114,
+      "step": 1107
+    },
+    {
+      "epoch": 4.502278481012659,
+      "grad_norm": 0.1898354295868109,
+      "learning_rate": 1.1868106817802816e-06,
+      "loss": 0.3923,
+      "step": 1108
+    },
+    {
+      "epoch": 4.506329113924051,
+      "grad_norm": 0.20028695659845924,
+      "learning_rate": 1.1676252981191482e-06,
+      "loss": 0.3851,
+      "step": 1109
+    },
+    {
+      "epoch": 4.510379746835443,
+      "grad_norm": 0.198572311167935,
+      "learning_rate": 1.1485915878148823e-06,
+      "loss": 0.4008,
+      "step": 1110
+    },
+    {
+      "epoch": 4.514430379746836,
+      "grad_norm": 0.19647375310786533,
+      "learning_rate": 1.1297097041623584e-06,
+      "loss": 0.4123,
+      "step": 1111
+    },
+    {
+      "epoch": 4.518481012658228,
+      "grad_norm": 0.19406549683240998,
+      "learning_rate": 1.1109797992336847e-06,
+      "loss": 0.4067,
+      "step": 1112
+    },
+    {
+      "epoch": 4.52253164556962,
+      "grad_norm": 0.19239953698799764,
+      "learning_rate": 1.092402023876933e-06,
+      "loss": 0.3859,
+      "step": 1113
+    },
+    {
+      "epoch": 4.526582278481013,
+      "grad_norm": 0.19450056763960294,
+      "learning_rate": 1.0739765277149527e-06,
+      "loss": 0.3887,
+      "step": 1114
+    },
+    {
+      "epoch": 4.530632911392405,
+      "grad_norm": 0.198284546888975,
+      "learning_rate": 1.0557034591441596e-06,
+      "loss": 0.3983,
+      "step": 1115
+    },
+    {
+      "epoch": 4.534683544303798,
+      "grad_norm": 0.2039075890824686,
+      "learning_rate": 1.0375829653333324e-06,
+      "loss": 0.4139,
+      "step": 1116
+    },
+    {
+      "epoch": 4.53873417721519,
+      "grad_norm": 12.934563843155917,
+      "learning_rate": 1.0196151922224385e-06,
+      "loss": 0.4564,
+      "step": 1117
+    },
+    {
+      "epoch": 4.542784810126582,
+      "grad_norm": 0.19092360161368369,
+      "learning_rate": 1.0018002845214526e-06,
+      "loss": 0.4106,
+      "step": 1118
+    },
+    {
+      "epoch": 4.546835443037975,
+      "grad_norm": 0.19139878249017706,
+      "learning_rate": 9.841383857091947e-07,
+      "loss": 0.425,
+      "step": 1119
+    },
+    {
+      "epoch": 4.550886075949367,
+      "grad_norm": 0.20257938434943779,
+      "learning_rate": 9.666296380321616e-07,
+      "loss": 0.3905,
+      "step": 1120
+    },
+    {
+      "epoch": 4.55493670886076,
+      "grad_norm": 0.19565688718832835,
+      "learning_rate": 9.492741825034124e-07,
+      "loss": 0.4095,
+      "step": 1121
+    },
+    {
+      "epoch": 4.558987341772152,
+      "grad_norm": 0.19698438348613598,
+      "learning_rate": 9.320721589013892e-07,
+      "loss": 0.4023,
+      "step": 1122
+    },
+    {
+      "epoch": 4.563037974683544,
+      "grad_norm": 0.2564518421414806,
+      "learning_rate": 9.150237057688339e-07,
+      "loss": 0.4149,
+      "step": 1123
+    },
+    {
+      "epoch": 4.567088607594937,
+      "grad_norm": 0.19802512572015757,
+      "learning_rate": 8.981289604116328e-07,
+      "loss": 0.4088,
+      "step": 1124
+    },
+    {
+      "epoch": 4.571139240506329,
+      "grad_norm": 0.187256633668446,
+      "learning_rate": 8.813880588977542e-07,
+      "loss": 0.3954,
+      "step": 1125
+    },
+    {
+      "epoch": 4.575189873417721,
+      "grad_norm": 0.19195717247193955,
+      "learning_rate": 8.648011360561126e-07,
+      "loss": 0.3995,
+      "step": 1126
+    },
+    {
+      "epoch": 4.579240506329114,
+      "grad_norm": 0.1969024203217451,
+      "learning_rate": 8.483683254755037e-07,
+      "loss": 0.4038,
+      "step": 1127
+    },
+    {
+      "epoch": 4.583291139240506,
+      "grad_norm": 0.20229429192568105,
+      "learning_rate": 8.320897595035227e-07,
+      "loss": 0.3941,
+      "step": 1128
+    },
+    {
+      "epoch": 4.587341772151898,
+      "grad_norm": 0.1863210023112725,
+      "learning_rate": 8.159655692455093e-07,
+      "loss": 0.4066,
+      "step": 1129
+    },
+    {
+      "epoch": 4.591392405063291,
+      "grad_norm": 0.19415672219878802,
+      "learning_rate": 7.999958845634648e-07,
+      "loss": 0.4055,
+      "step": 1130
+    },
+    {
+      "epoch": 4.595443037974683,
+      "grad_norm": 0.2008561944033465,
+      "learning_rate": 7.841808340750478e-07,
+      "loss": 0.3973,
+      "step": 1131
+    },
+    {
+      "epoch": 4.599493670886076,
+      "grad_norm": 0.19261939035811929,
+      "learning_rate": 7.685205451524869e-07,
+      "loss": 0.4089,
+      "step": 1132
+    },
+    {
+      "epoch": 4.603544303797468,
+      "grad_norm": 0.19121181442403765,
+      "learning_rate": 7.530151439216027e-07,
+      "loss": 0.4069,
+      "step": 1133
+    },
+    {
+      "epoch": 4.6075949367088604,
+      "grad_norm": 0.20120687357232353,
+      "learning_rate": 7.376647552607675e-07,
+      "loss": 0.3951,
+      "step": 1134
+    },
+    {
+      "epoch": 4.611645569620253,
+      "grad_norm": 0.18688541424345556,
+      "learning_rate": 7.224695027998963e-07,
+      "loss": 0.4142,
+      "step": 1135
+    },
+    {
+      "epoch": 4.6156962025316455,
+      "grad_norm": 0.1946291587075732,
+      "learning_rate": 7.07429508919466e-07,
+      "loss": 0.3936,
+      "step": 1136
+    },
+    {
+      "epoch": 4.619746835443038,
+      "grad_norm": 0.1937014449193155,
+      "learning_rate": 6.925448947495206e-07,
+      "loss": 0.4006,
+      "step": 1137
+    },
+    {
+      "epoch": 4.6237974683544305,
+      "grad_norm": 0.1915520582234905,
+      "learning_rate": 6.778157801686936e-07,
+      "loss": 0.4051,
+      "step": 1138
+    },
+    {
+      "epoch": 4.627848101265823,
+      "grad_norm": 0.19103489429250678,
+      "learning_rate": 6.632422838032515e-07,
+      "loss": 0.4004,
+      "step": 1139
+    },
+    {
+      "epoch": 4.6318987341772155,
+      "grad_norm": 0.20141531722761263,
+      "learning_rate": 6.488245230261281e-07,
+      "loss": 0.3922,
+      "step": 1140
+    },
+    {
+      "epoch": 4.635949367088608,
+      "grad_norm": 0.19469744820763427,
+      "learning_rate": 6.345626139559868e-07,
+      "loss": 0.408,
+      "step": 1141
+    },
+    {
+      "epoch": 4.64,
+      "grad_norm": 0.18980725628321693,
+      "learning_rate": 6.204566714562866e-07,
+      "loss": 0.3947,
+      "step": 1142
+    },
+    {
+      "epoch": 4.644050632911393,
+      "grad_norm": 0.19331179100698032,
+      "learning_rate": 6.06506809134344e-07,
+      "loss": 0.3986,
+      "step": 1143
+    },
+    {
+      "epoch": 4.648101265822785,
+      "grad_norm": 0.18672037998301164,
+      "learning_rate": 5.927131393404373e-07,
+      "loss": 0.4099,
+      "step": 1144
+    },
+    {
+      "epoch": 4.652151898734177,
+      "grad_norm": 0.19742196886847296,
+      "learning_rate": 5.790757731668817e-07,
+      "loss": 0.4018,
+      "step": 1145
+    },
+    {
+      "epoch": 4.65620253164557,
+      "grad_norm": 0.20315619104068341,
+      "learning_rate": 5.655948204471507e-07,
+      "loss": 0.4014,
+      "step": 1146
+    },
+    {
+      "epoch": 4.660253164556962,
+      "grad_norm": 0.1891400530428415,
+      "learning_rate": 5.522703897549875e-07,
+      "loss": 0.4105,
+      "step": 1147
+    },
+    {
+      "epoch": 4.664303797468355,
+      "grad_norm": 0.20391041665712495,
+      "learning_rate": 5.391025884035239e-07,
+      "loss": 0.4046,
+      "step": 1148
+    },
+    {
+      "epoch": 4.668354430379747,
+      "grad_norm": 0.3280910157525473,
+      "learning_rate": 5.260915224444207e-07,
+      "loss": 0.3943,
+      "step": 1149
+    },
+    {
+      "epoch": 4.672405063291139,
+      "grad_norm": 0.1918892189565304,
+      "learning_rate": 5.132372966670129e-07,
+      "loss": 0.4,
+      "step": 1150
+    },
+    {
+      "epoch": 4.676455696202532,
+      "grad_norm": 0.18893564104716462,
+      "learning_rate": 5.005400145974704e-07,
+      "loss": 0.4054,
+      "step": 1151
+    },
+    {
+      "epoch": 4.680506329113924,
+      "grad_norm": 0.19635329089246006,
+      "learning_rate": 4.879997784979562e-07,
+      "loss": 0.3991,
+      "step": 1152
+    },
+    {
+      "epoch": 4.684556962025317,
+      "grad_norm": 0.191235754865051,
+      "learning_rate": 4.7561668936580984e-07,
+      "loss": 0.4122,
+      "step": 1153
+    },
+    {
+      "epoch": 4.688607594936709,
+      "grad_norm": 0.19362094506593577,
+      "learning_rate": 4.6339084693272306e-07,
+      "loss": 0.3999,
+      "step": 1154
+    },
+    {
+      "epoch": 4.692658227848101,
+      "grad_norm": 0.19092248737044187,
+      "learning_rate": 4.5132234966395847e-07,
+      "loss": 0.4048,
+      "step": 1155
+    },
+    {
+      "epoch": 4.696708860759494,
+      "grad_norm": 0.18786510685780655,
+      "learning_rate": 4.3941129475752795e-07,
+      "loss": 0.423,
+      "step": 1156
+    },
+    {
+      "epoch": 4.700759493670886,
+      "grad_norm": 0.19163445235242726,
+      "learning_rate": 4.27657778143431e-07,
+      "loss": 0.3952,
+      "step": 1157
+    },
+    {
+      "epoch": 4.704810126582278,
+      "grad_norm": 0.20601349126780893,
+      "learning_rate": 4.1606189448287757e-07,
+      "loss": 0.396,
+      "step": 1158
+    },
+    {
+      "epoch": 4.708860759493671,
+      "grad_norm": 0.19492298405083658,
+      "learning_rate": 4.046237371675177e-07,
+      "loss": 0.393,
+      "step": 1159
+    },
+    {
+      "epoch": 4.712911392405063,
+      "grad_norm": 0.19303296214152027,
+      "learning_rate": 3.9334339831869963e-07,
+      "loss": 0.4084,
+      "step": 1160
+    },
+    {
+      "epoch": 4.716962025316455,
+      "grad_norm": 0.19709917681704056,
+      "learning_rate": 3.8222096878671955e-07,
+      "loss": 0.4046,
+      "step": 1161
+    },
+    {
+      "epoch": 4.721012658227848,
+      "grad_norm": 0.18432746747430415,
+      "learning_rate": 3.7125653815009545e-07,
+      "loss": 0.4017,
+      "step": 1162
+    },
+    {
+      "epoch": 4.72506329113924,
+      "grad_norm": 0.18802137450859366,
+      "learning_rate": 3.6045019471484974e-07,
+      "loss": 0.4013,
+      "step": 1163
+    },
+    {
+      "epoch": 4.729113924050633,
+      "grad_norm": 0.1970561696911561,
+      "learning_rate": 3.498020255137813e-07,
+      "loss": 0.3965,
+      "step": 1164
+    },
+    {
+      "epoch": 4.733164556962025,
+      "grad_norm": 0.191018853626761,
+      "learning_rate": 3.393121163057811e-07,
+      "loss": 0.4056,
+      "step": 1165
+    },
+    {
+      "epoch": 4.737215189873417,
+      "grad_norm": 0.20083701125257147,
+      "learning_rate": 3.289805515751399e-07,
+      "loss": 0.4146,
+      "step": 1166
+    },
+    {
+      "epoch": 4.74126582278481,
+      "grad_norm": 0.19163701787143897,
+      "learning_rate": 3.188074145308573e-07,
+      "loss": 0.4,
+      "step": 1167
+    },
+    {
+      "epoch": 4.745316455696202,
+      "grad_norm": 0.19504001219269196,
+      "learning_rate": 3.087927871059804e-07,
+      "loss": 0.4139,
+      "step": 1168
+    },
+    {
+      "epoch": 4.749367088607595,
+      "grad_norm": 0.19769763834742507,
+      "learning_rate": 2.989367499569418e-07,
+      "loss": 0.3838,
+      "step": 1169
+    },
+    {
+      "epoch": 4.7534177215189874,
+      "grad_norm": 0.17982644644799395,
+      "learning_rate": 2.8923938246290917e-07,
+      "loss": 0.4117,
+      "step": 1170
+    },
+    {
+      "epoch": 4.7574683544303795,
+      "grad_norm": 0.18478616592177705,
+      "learning_rate": 2.7970076272514804e-07,
+      "loss": 0.3968,
+      "step": 1171
+    },
+    {
+      "epoch": 4.7615189873417725,
+      "grad_norm": 0.1973770728119646,
+      "learning_rate": 2.703209675663887e-07,
+      "loss": 0.3986,
+      "step": 1172
+    },
+    {
+      "epoch": 4.7655696202531646,
+      "grad_norm": 0.18690634693944766,
+      "learning_rate": 2.6110007253021374e-07,
+      "loss": 0.4007,
+      "step": 1173
+    },
+    {
+      "epoch": 4.769620253164557,
+      "grad_norm": 0.18725918406152725,
+      "learning_rate": 2.520381518804471e-07,
+      "loss": 0.4146,
+      "step": 1174
+    },
+    {
+      "epoch": 4.77367088607595,
+      "grad_norm": 0.1911937863122524,
+      "learning_rate": 2.4313527860054585e-07,
+      "loss": 0.4267,
+      "step": 1175
+    },
+    {
+      "epoch": 4.777721518987342,
+      "grad_norm": 0.18935192670622128,
+      "learning_rate": 2.343915243930317e-07,
+      "loss": 0.4105,
+      "step": 1176
+    },
+    {
+      "epoch": 4.781772151898734,
+      "grad_norm": 0.1872854482602375,
+      "learning_rate": 2.2580695967889367e-07,
+      "loss": 0.4052,
+      "step": 1177
+    },
+    {
+      "epoch": 4.785822784810127,
+      "grad_norm": 0.2957687804968097,
+      "learning_rate": 2.1738165359704189e-07,
+      "loss": 0.3931,
+      "step": 1178
+    },
+    {
+      "epoch": 4.789873417721519,
+      "grad_norm": 0.18376697601265027,
+      "learning_rate": 2.0911567400373257e-07,
+      "loss": 0.4102,
+      "step": 1179
+    },
+    {
+      "epoch": 4.793924050632912,
+      "grad_norm": 0.194282311896596,
+      "learning_rate": 2.0100908747202607e-07,
+      "loss": 0.4086,
+      "step": 1180
+    },
+    {
+      "epoch": 4.797974683544304,
+      "grad_norm": 0.1836596664741118,
+      "learning_rate": 1.9306195929125638e-07,
+      "loss": 0.3979,
+      "step": 1181
+    },
+    {
+      "epoch": 4.802025316455696,
+      "grad_norm": 0.19038038294307646,
+      "learning_rate": 1.8527435346650247e-07,
+      "loss": 0.3984,
+      "step": 1182
+    },
+    {
+      "epoch": 4.806075949367089,
+      "grad_norm": 0.1889161953323751,
+      "learning_rate": 1.7764633271807108e-07,
+      "loss": 0.4088,
+      "step": 1183
+    },
+    {
+      "epoch": 4.810126582278481,
+      "grad_norm": 0.19734667785854407,
+      "learning_rate": 1.7017795848099262e-07,
+      "loss": 0.3886,
+      "step": 1184
+    },
+    {
+      "epoch": 4.814177215189874,
+      "grad_norm": 0.18691484760096852,
+      "learning_rate": 1.6286929090452596e-07,
+      "loss": 0.4153,
+      "step": 1185
+    },
+    {
+      "epoch": 4.818227848101266,
+      "grad_norm": 0.19270386232295753,
+      "learning_rate": 1.557203888516745e-07,
+      "loss": 0.4047,
+      "step": 1186
+    },
+    {
+      "epoch": 4.822278481012658,
+      "grad_norm": 0.226935817178927,
+      "learning_rate": 1.487313098987131e-07,
+      "loss": 0.4126,
+      "step": 1187
+    },
+    {
+      "epoch": 4.826329113924051,
+      "grad_norm": 0.18418238325676692,
+      "learning_rate": 1.4190211033472402e-07,
+      "loss": 0.4039,
+      "step": 1188
+    },
+    {
+      "epoch": 4.830379746835443,
+      "grad_norm": 0.18908156704647866,
+      "learning_rate": 1.3523284516113955e-07,
+      "loss": 0.3962,
+      "step": 1189
+    },
+    {
+      "epoch": 4.834430379746835,
+      "grad_norm": 0.1892402096142589,
+      "learning_rate": 1.2872356809130682e-07,
+      "loss": 0.4018,
+      "step": 1190
+    },
+    {
+      "epoch": 4.838481012658228,
+      "grad_norm": 0.18740554273669613,
+      "learning_rate": 1.2237433155004807e-07,
+      "loss": 0.4042,
+      "step": 1191
+    },
+    {
+      "epoch": 4.84253164556962,
+      "grad_norm": 0.1842092081590548,
+      "learning_rate": 1.1618518667323886e-07,
+      "loss": 0.4043,
+      "step": 1192
+    },
+    {
+      "epoch": 4.846582278481012,
+      "grad_norm": 0.18606984052998185,
+      "learning_rate": 1.1015618330740385e-07,
+      "loss": 0.406,
+      "step": 1193
+    },
+    {
+      "epoch": 4.850632911392405,
+      "grad_norm": 0.7272203882170395,
+      "learning_rate": 1.042873700093061e-07,
+      "loss": 0.3816,
+      "step": 1194
+    },
+    {
+      "epoch": 4.854683544303797,
+      "grad_norm": 0.19549959673004239,
+      "learning_rate": 9.857879404556291e-08,
+      "loss": 0.4058,
+      "step": 1195
+    },
+    {
+      "epoch": 4.85873417721519,
+      "grad_norm": 0.19109421121055906,
+      "learning_rate": 9.303050139225722e-08,
+      "loss": 0.4,
+      "step": 1196
+    },
+    {
+      "epoch": 4.862784810126582,
+      "grad_norm": 0.19140428328526202,
+      "learning_rate": 8.76425367345779e-08,
+      "loss": 0.4061,
+      "step": 1197
+    },
+    {
+      "epoch": 4.866835443037974,
+      "grad_norm": 0.1956149766911906,
+      "learning_rate": 8.241494346644897e-08,
+      "loss": 0.3852,
+      "step": 1198
+    },
+    {
+      "epoch": 4.870886075949367,
+      "grad_norm": 0.19294474947855356,
+      "learning_rate": 7.734776369019204e-08,
+      "loss": 0.404,
+      "step": 1199
+    },
+    {
+      "epoch": 4.874936708860759,
+      "grad_norm": 0.19853681853680752,
+      "learning_rate": 7.244103821617332e-08,
+      "loss": 0.4369,
+      "step": 1200
+    },
+    {
+      "epoch": 4.878987341772152,
+      "grad_norm": 0.1948971483286671,
+      "learning_rate": 6.769480656248606e-08,
+      "loss": 0.4017,
+      "step": 1201
+    },
+    {
+      "epoch": 4.883037974683544,
+      "grad_norm": 0.1885730763440475,
+      "learning_rate": 6.310910695462635e-08,
+      "loss": 0.4041,
+      "step": 1202
+    },
+    {
+      "epoch": 4.8870886075949365,
+      "grad_norm": 0.1875581221601517,
+      "learning_rate": 5.8683976325191185e-08,
+      "loss": 0.3954,
+      "step": 1203
+    },
+    {
+      "epoch": 4.891139240506329,
+      "grad_norm": 0.18435323811932117,
+      "learning_rate": 5.4419450313571984e-08,
+      "loss": 0.411,
+      "step": 1204
+    },
+    {
+      "epoch": 4.8951898734177215,
+      "grad_norm": 1.2341128008972526,
+      "learning_rate": 5.031556326567488e-08,
+      "loss": 0.5608,
+      "step": 1205
+    },
+    {
+      "epoch": 4.899240506329114,
+      "grad_norm": 0.19177833242141606,
+      "learning_rate": 4.637234823364312e-08,
+      "loss": 0.399,
+      "step": 1206
+    },
+    {
+      "epoch": 4.9032911392405065,
+      "grad_norm": 0.18856342942744986,
+      "learning_rate": 4.258983697558838e-08,
+      "loss": 0.3995,
+      "step": 1207
+    },
+    {
+      "epoch": 4.907341772151899,
+      "grad_norm": 0.18736478857931582,
+      "learning_rate": 3.896805995533548e-08,
+      "loss": 0.3961,
+      "step": 1208
+    },
+    {
+      "epoch": 4.911392405063291,
+      "grad_norm": 0.19167130658543072,
+      "learning_rate": 3.550704634218028e-08,
+      "loss": 0.41,
+      "step": 1209
+    },
+    {
+      "epoch": 4.915443037974684,
+      "grad_norm": 0.18959871697577604,
+      "learning_rate": 3.2206824010647676e-08,
+      "loss": 0.4162,
+      "step": 1210
+    },
+    {
+      "epoch": 4.919493670886076,
+      "grad_norm": 0.18772927164510364,
+      "learning_rate": 2.9067419540278476e-08,
+      "loss": 0.4131,
+      "step": 1211
+    },
+    {
+      "epoch": 4.923544303797469,
+      "grad_norm": 0.19283063662545166,
+      "learning_rate": 2.6088858215400638e-08,
+      "loss": 0.4068,
+      "step": 1212
+    },
+    {
+      "epoch": 4.927594936708861,
+      "grad_norm": 0.1866926130432702,
+      "learning_rate": 2.3271164024940564e-08,
+      "loss": 0.4132,
+      "step": 1213
+    },
+    {
+      "epoch": 4.931645569620253,
+      "grad_norm": 0.18789483288924966,
+      "learning_rate": 2.061435966221881e-08,
+      "loss": 0.419,
+      "step": 1214
+    },
+    {
+      "epoch": 4.935696202531646,
+      "grad_norm": 0.19366336294275366,
+      "learning_rate": 1.811846652477245e-08,
+      "loss": 0.3914,
+      "step": 1215
+    },
+    {
+      "epoch": 4.939746835443038,
+      "grad_norm": 0.1888517241477089,
+      "learning_rate": 1.5783504714184106e-08,
+      "loss": 0.402,
+      "step": 1216
+    },
+    {
+      "epoch": 4.943797468354431,
+      "grad_norm": 0.1914430875562289,
+      "learning_rate": 1.360949303591097e-08,
+      "loss": 0.3946,
+      "step": 1217
+    },
+    {
+      "epoch": 4.947848101265823,
+      "grad_norm": 0.18906378325533787,
+      "learning_rate": 1.1596448999144916e-08,
+      "loss": 0.4021,
+      "step": 1218
+    },
+    {
+      "epoch": 4.951898734177215,
+      "grad_norm": 0.18634079709137633,
+      "learning_rate": 9.744388816668172e-09,
+      "loss": 0.3995,
+      "step": 1219
+    },
+    {
+      "epoch": 4.955949367088608,
+      "grad_norm": 0.19651512313040898,
+      "learning_rate": 8.05332740472009e-09,
+      "loss": 0.3968,
+      "step": 1220
+    },
+    {
+      "epoch": 4.96,
+      "grad_norm": 0.2060212140728209,
+      "learning_rate": 6.523278382872811e-09,
+      "loss": 0.4146,
+      "step": 1221
+    },
+    {
+      "epoch": 4.964050632911392,
+      "grad_norm": 0.19194723960872073,
+      "learning_rate": 5.15425407393133e-09,
+      "loss": 0.3851,
+      "step": 1222
+    },
+    {
+      "epoch": 4.968101265822785,
+      "grad_norm": 0.184150488674795,
+      "learning_rate": 3.94626550383137e-09,
+      "loss": 0.4042,
+      "step": 1223
+    },
+    {
+      "epoch": 4.972151898734177,
+      "grad_norm": 0.1986942305622378,
+      "learning_rate": 2.899322401546112e-09,
+      "loss": 0.4011,
+      "step": 1224
+    },
+    {
+      "epoch": 4.976202531645569,
+      "grad_norm": 0.19381145183754,
+      "learning_rate": 2.013433199010706e-09,
+      "loss": 0.3958,
+      "step": 1225
+    },
+    {
+      "epoch": 4.980253164556962,
+      "grad_norm": 0.1893363279397581,
+      "learning_rate": 1.2886050310556563e-09,
+      "loss": 0.4034,
+      "step": 1226
+    },
+    {
+      "epoch": 4.984303797468354,
+      "grad_norm": 0.19539201209810797,
+      "learning_rate": 7.248437353468695e-10,
+      "loss": 0.4072,
+      "step": 1227
+    },
+    {
+      "epoch": 4.988354430379747,
+      "grad_norm": 0.1883101586696984,
+      "learning_rate": 3.221538523412449e-10,
+      "loss": 0.3922,
+      "step": 1228
+    },
+    {
+      "epoch": 4.992405063291139,
+      "grad_norm": 0.1915695814208868,
+      "learning_rate": 8.053862524670663e-11,
+      "loss": 0.3982,
+      "step": 1229
+    },
+    {
+      "epoch": 4.996455696202531,
+      "grad_norm": 0.1936169879704622,
+      "learning_rate": 0.0,
+      "loss": 0.3994,
+      "step": 1230
+    },
+    {
+      "epoch": 4.996455696202531,
+      "step": 1230,
+      "total_flos": 6.214196546092204e+18,
+      "train_loss": 0.08002444532344012,
+      "train_runtime": 11259.7517,
+      "train_samples_per_second": 14.032,
+      "train_steps_per_second": 0.109
+    }
+  ],
+  "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": 6.214196546092204e+18,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}