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": 4.982278481012658,
+  "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": 6.075022189168995,
+      "learning_rate": 3.2520325203252037e-07,
+      "loss": 0.7857,
+      "step": 1
+    },
+    {
+      "epoch": 0.00810126582278481,
+      "grad_norm": 6.2086765778276005,
+      "learning_rate": 6.504065040650407e-07,
+      "loss": 0.8111,
+      "step": 2
+    },
+    {
+      "epoch": 0.012151898734177215,
+      "grad_norm": 6.066264251847621,
+      "learning_rate": 9.75609756097561e-07,
+      "loss": 0.804,
+      "step": 3
+    },
+    {
+      "epoch": 0.01620253164556962,
+      "grad_norm": 6.179507223169048,
+      "learning_rate": 1.3008130081300815e-06,
+      "loss": 0.8111,
+      "step": 4
+    },
+    {
+      "epoch": 0.020253164556962026,
+      "grad_norm": 5.647456779236567,
+      "learning_rate": 1.6260162601626018e-06,
+      "loss": 0.7916,
+      "step": 5
+    },
+    {
+      "epoch": 0.02430379746835443,
+      "grad_norm": 5.551606575085782,
+      "learning_rate": 1.951219512195122e-06,
+      "loss": 0.7952,
+      "step": 6
+    },
+    {
+      "epoch": 0.028354430379746835,
+      "grad_norm": 4.280280744019211,
+      "learning_rate": 2.2764227642276426e-06,
+      "loss": 0.7634,
+      "step": 7
+    },
+    {
+      "epoch": 0.03240506329113924,
+      "grad_norm": 3.821245780860169,
+      "learning_rate": 2.601626016260163e-06,
+      "loss": 0.7453,
+      "step": 8
+    },
+    {
+      "epoch": 0.03645569620253165,
+      "grad_norm": 2.235432119847156,
+      "learning_rate": 2.926829268292683e-06,
+      "loss": 0.7163,
+      "step": 9
+    },
+    {
+      "epoch": 0.04050632911392405,
+      "grad_norm": 2.025994994341747,
+      "learning_rate": 3.2520325203252037e-06,
+      "loss": 0.7191,
+      "step": 10
+    },
+    {
+      "epoch": 0.044556962025316456,
+      "grad_norm": 1.9320683684899431,
+      "learning_rate": 3.577235772357724e-06,
+      "loss": 0.7278,
+      "step": 11
+    },
+    {
+      "epoch": 0.04860759493670886,
+      "grad_norm": 3.709349993613935,
+      "learning_rate": 3.902439024390244e-06,
+      "loss": 0.7184,
+      "step": 12
+    },
+    {
+      "epoch": 0.052658227848101265,
+      "grad_norm": 4.097189450071972,
+      "learning_rate": 4.227642276422765e-06,
+      "loss": 0.7247,
+      "step": 13
+    },
+    {
+      "epoch": 0.05670886075949367,
+      "grad_norm": 4.1512966301912515,
+      "learning_rate": 4.552845528455285e-06,
+      "loss": 0.7087,
+      "step": 14
+    },
+    {
+      "epoch": 0.060759493670886074,
+      "grad_norm": 3.9434829699612433,
+      "learning_rate": 4.8780487804878055e-06,
+      "loss": 0.7013,
+      "step": 15
+    },
+    {
+      "epoch": 0.06481012658227848,
+      "grad_norm": 3.0373354272515583,
+      "learning_rate": 5.203252032520326e-06,
+      "loss": 0.6637,
+      "step": 16
+    },
+    {
+      "epoch": 0.06886075949367089,
+      "grad_norm": 2.7859037306919485,
+      "learning_rate": 5.528455284552846e-06,
+      "loss": 0.6587,
+      "step": 17
+    },
+    {
+      "epoch": 0.0729113924050633,
+      "grad_norm": 2.110026448963815,
+      "learning_rate": 5.853658536585366e-06,
+      "loss": 0.6462,
+      "step": 18
+    },
+    {
+      "epoch": 0.0769620253164557,
+      "grad_norm": 1.3476808069399706,
+      "learning_rate": 6.178861788617887e-06,
+      "loss": 0.6148,
+      "step": 19
+    },
+    {
+      "epoch": 0.0810126582278481,
+      "grad_norm": 1.2011538280758425,
+      "learning_rate": 6.504065040650407e-06,
+      "loss": 0.6218,
+      "step": 20
+    },
+    {
+      "epoch": 0.08506329113924051,
+      "grad_norm": 1.2567050219596014,
+      "learning_rate": 6.829268292682928e-06,
+      "loss": 0.6333,
+      "step": 21
+    },
+    {
+      "epoch": 0.08911392405063291,
+      "grad_norm": 1.2335957703178186,
+      "learning_rate": 7.154471544715448e-06,
+      "loss": 0.584,
+      "step": 22
+    },
+    {
+      "epoch": 0.09316455696202532,
+      "grad_norm": 1.0181960797572749,
+      "learning_rate": 7.4796747967479676e-06,
+      "loss": 0.5824,
+      "step": 23
+    },
+    {
+      "epoch": 0.09721518987341772,
+      "grad_norm": 0.7819637624084433,
+      "learning_rate": 7.804878048780489e-06,
+      "loss": 0.5761,
+      "step": 24
+    },
+    {
+      "epoch": 0.10126582278481013,
+      "grad_norm": 0.9007444998746347,
+      "learning_rate": 8.130081300813009e-06,
+      "loss": 0.5588,
+      "step": 25
+    },
+    {
+      "epoch": 0.10531645569620253,
+      "grad_norm": 0.8383493119011176,
+      "learning_rate": 8.45528455284553e-06,
+      "loss": 0.5698,
+      "step": 26
+    },
+    {
+      "epoch": 0.10936708860759493,
+      "grad_norm": 0.6719854992155601,
+      "learning_rate": 8.78048780487805e-06,
+      "loss": 0.5647,
+      "step": 27
+    },
+    {
+      "epoch": 0.11341772151898734,
+      "grad_norm": 0.7758098953475778,
+      "learning_rate": 9.10569105691057e-06,
+      "loss": 0.5412,
+      "step": 28
+    },
+    {
+      "epoch": 0.11746835443037974,
+      "grad_norm": 0.6959295059262672,
+      "learning_rate": 9.43089430894309e-06,
+      "loss": 0.5659,
+      "step": 29
+    },
+    {
+      "epoch": 0.12151898734177215,
+      "grad_norm": 0.4903376528505027,
+      "learning_rate": 9.756097560975611e-06,
+      "loss": 0.5434,
+      "step": 30
+    },
+    {
+      "epoch": 0.12556962025316457,
+      "grad_norm": 0.6723067876451173,
+      "learning_rate": 1.008130081300813e-05,
+      "loss": 0.5516,
+      "step": 31
+    },
+    {
+      "epoch": 0.12962025316455697,
+      "grad_norm": 0.67924043997143,
+      "learning_rate": 1.0406504065040652e-05,
+      "loss": 0.5429,
+      "step": 32
+    },
+    {
+      "epoch": 0.13367088607594937,
+      "grad_norm": 0.5074279217921603,
+      "learning_rate": 1.0731707317073172e-05,
+      "loss": 0.5281,
+      "step": 33
+    },
+    {
+      "epoch": 0.13772151898734178,
+      "grad_norm": 0.5148372811547323,
+      "learning_rate": 1.1056910569105692e-05,
+      "loss": 0.5457,
+      "step": 34
+    },
+    {
+      "epoch": 0.14177215189873418,
+      "grad_norm": 0.6000032661877054,
+      "learning_rate": 1.1382113821138213e-05,
+      "loss": 0.5488,
+      "step": 35
+    },
+    {
+      "epoch": 0.1458227848101266,
+      "grad_norm": 0.4851855170437134,
+      "learning_rate": 1.1707317073170731e-05,
+      "loss": 0.5215,
+      "step": 36
+    },
+    {
+      "epoch": 0.149873417721519,
+      "grad_norm": 0.4968209544050209,
+      "learning_rate": 1.2032520325203254e-05,
+      "loss": 0.5238,
+      "step": 37
+    },
+    {
+      "epoch": 0.1539240506329114,
+      "grad_norm": 0.47905262542564386,
+      "learning_rate": 1.2357723577235774e-05,
+      "loss": 0.5081,
+      "step": 38
+    },
+    {
+      "epoch": 0.1579746835443038,
+      "grad_norm": 0.4196316641108655,
+      "learning_rate": 1.2682926829268294e-05,
+      "loss": 0.5182,
+      "step": 39
+    },
+    {
+      "epoch": 0.1620253164556962,
+      "grad_norm": 0.4990331164828479,
+      "learning_rate": 1.3008130081300815e-05,
+      "loss": 0.5292,
+      "step": 40
+    },
+    {
+      "epoch": 0.1660759493670886,
+      "grad_norm": 0.47306508530829766,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.5101,
+      "step": 41
+    },
+    {
+      "epoch": 0.17012658227848101,
+      "grad_norm": 0.42709891098479524,
+      "learning_rate": 1.3658536585365855e-05,
+      "loss": 0.5078,
+      "step": 42
+    },
+    {
+      "epoch": 0.17417721518987342,
+      "grad_norm": 0.5008387679055025,
+      "learning_rate": 1.3983739837398376e-05,
+      "loss": 0.4982,
+      "step": 43
+    },
+    {
+      "epoch": 0.17822784810126582,
+      "grad_norm": 0.44143138175128727,
+      "learning_rate": 1.4308943089430896e-05,
+      "loss": 0.5131,
+      "step": 44
+    },
+    {
+      "epoch": 0.18227848101265823,
+      "grad_norm": 0.5286444157249438,
+      "learning_rate": 1.4634146341463415e-05,
+      "loss": 0.5079,
+      "step": 45
+    },
+    {
+      "epoch": 0.18632911392405063,
+      "grad_norm": 0.434520893255903,
+      "learning_rate": 1.4959349593495935e-05,
+      "loss": 0.513,
+      "step": 46
+    },
+    {
+      "epoch": 0.19037974683544304,
+      "grad_norm": 0.515357302031543,
+      "learning_rate": 1.528455284552846e-05,
+      "loss": 0.5131,
+      "step": 47
+    },
+    {
+      "epoch": 0.19443037974683544,
+      "grad_norm": 0.4444312605090478,
+      "learning_rate": 1.5609756097560978e-05,
+      "loss": 0.5106,
+      "step": 48
+    },
+    {
+      "epoch": 0.19848101265822785,
+      "grad_norm": 0.4530215718084883,
+      "learning_rate": 1.5934959349593496e-05,
+      "loss": 0.5223,
+      "step": 49
+    },
+    {
+      "epoch": 0.20253164556962025,
+      "grad_norm": 0.4226575728904452,
+      "learning_rate": 1.6260162601626018e-05,
+      "loss": 0.5052,
+      "step": 50
+    },
+    {
+      "epoch": 0.20658227848101265,
+      "grad_norm": 0.37872888027537716,
+      "learning_rate": 1.6585365853658537e-05,
+      "loss": 0.5081,
+      "step": 51
+    },
+    {
+      "epoch": 0.21063291139240506,
+      "grad_norm": 0.4936483773471764,
+      "learning_rate": 1.691056910569106e-05,
+      "loss": 0.4969,
+      "step": 52
+    },
+    {
+      "epoch": 0.21468354430379746,
+      "grad_norm": 0.36278453209010936,
+      "learning_rate": 1.7235772357723578e-05,
+      "loss": 0.4999,
+      "step": 53
+    },
+    {
+      "epoch": 0.21873417721518987,
+      "grad_norm": 0.38780198498767404,
+      "learning_rate": 1.75609756097561e-05,
+      "loss": 0.487,
+      "step": 54
+    },
+    {
+      "epoch": 0.22278481012658227,
+      "grad_norm": 0.5010851988662626,
+      "learning_rate": 1.788617886178862e-05,
+      "loss": 0.5059,
+      "step": 55
+    },
+    {
+      "epoch": 0.22683544303797468,
+      "grad_norm": 0.4689659617108022,
+      "learning_rate": 1.821138211382114e-05,
+      "loss": 0.488,
+      "step": 56
+    },
+    {
+      "epoch": 0.23088607594936708,
+      "grad_norm": 0.5583412157132028,
+      "learning_rate": 1.8536585365853663e-05,
+      "loss": 0.4935,
+      "step": 57
+    },
+    {
+      "epoch": 0.23493670886075949,
+      "grad_norm": 0.5633301253513429,
+      "learning_rate": 1.886178861788618e-05,
+      "loss": 0.4807,
+      "step": 58
+    },
+    {
+      "epoch": 0.2389873417721519,
+      "grad_norm": 0.6084422344496034,
+      "learning_rate": 1.91869918699187e-05,
+      "loss": 0.4865,
+      "step": 59
+    },
+    {
+      "epoch": 0.2430379746835443,
+      "grad_norm": 0.6162376964346101,
+      "learning_rate": 1.9512195121951222e-05,
+      "loss": 0.4839,
+      "step": 60
+    },
+    {
+      "epoch": 0.2470886075949367,
+      "grad_norm": 0.5976829993463736,
+      "learning_rate": 1.983739837398374e-05,
+      "loss": 0.485,
+      "step": 61
+    },
+    {
+      "epoch": 0.25113924050632913,
+      "grad_norm": 0.5448367765971719,
+      "learning_rate": 2.016260162601626e-05,
+      "loss": 0.4773,
+      "step": 62
+    },
+    {
+      "epoch": 0.25518987341772154,
+      "grad_norm": 0.5631205323638745,
+      "learning_rate": 2.048780487804878e-05,
+      "loss": 0.4816,
+      "step": 63
+    },
+    {
+      "epoch": 0.25924050632911394,
+      "grad_norm": 0.7959572859295161,
+      "learning_rate": 2.0813008130081303e-05,
+      "loss": 0.4827,
+      "step": 64
+    },
+    {
+      "epoch": 0.26329113924050634,
+      "grad_norm": 0.5289391909823953,
+      "learning_rate": 2.1138211382113822e-05,
+      "loss": 0.4673,
+      "step": 65
+    },
+    {
+      "epoch": 0.26734177215189875,
+      "grad_norm": 0.5331104283020521,
+      "learning_rate": 2.1463414634146344e-05,
+      "loss": 0.4906,
+      "step": 66
+    },
+    {
+      "epoch": 0.27139240506329115,
+      "grad_norm": 0.6749067292132194,
+      "learning_rate": 2.1788617886178863e-05,
+      "loss": 0.4839,
+      "step": 67
+    },
+    {
+      "epoch": 0.27544303797468356,
+      "grad_norm": 0.803685256878661,
+      "learning_rate": 2.2113821138211385e-05,
+      "loss": 0.4815,
+      "step": 68
+    },
+    {
+      "epoch": 0.27949367088607596,
+      "grad_norm": 0.6143011748328481,
+      "learning_rate": 2.2439024390243907e-05,
+      "loss": 0.4895,
+      "step": 69
+    },
+    {
+      "epoch": 0.28354430379746837,
+      "grad_norm": 0.8043124705893278,
+      "learning_rate": 2.2764227642276426e-05,
+      "loss": 0.4805,
+      "step": 70
+    },
+    {
+      "epoch": 0.28759493670886077,
+      "grad_norm": 0.876816454010533,
+      "learning_rate": 2.3089430894308948e-05,
+      "loss": 0.4745,
+      "step": 71
+    },
+    {
+      "epoch": 0.2916455696202532,
+      "grad_norm": 0.6095994776003223,
+      "learning_rate": 2.3414634146341463e-05,
+      "loss": 0.4893,
+      "step": 72
+    },
+    {
+      "epoch": 0.2956962025316456,
+      "grad_norm": 0.5833709656595453,
+      "learning_rate": 2.3739837398373985e-05,
+      "loss": 0.4706,
+      "step": 73
+    },
+    {
+      "epoch": 0.299746835443038,
+      "grad_norm": 0.7680105750341497,
+      "learning_rate": 2.4065040650406507e-05,
+      "loss": 0.4581,
+      "step": 74
+    },
+    {
+      "epoch": 0.3037974683544304,
+      "grad_norm": 0.7053186265284164,
+      "learning_rate": 2.4390243902439026e-05,
+      "loss": 0.475,
+      "step": 75
+    },
+    {
+      "epoch": 0.3078481012658228,
+      "grad_norm": 0.689911869122022,
+      "learning_rate": 2.4715447154471548e-05,
+      "loss": 0.4881,
+      "step": 76
+    },
+    {
+      "epoch": 0.3118987341772152,
+      "grad_norm": 0.4669855247786175,
+      "learning_rate": 2.5040650406504066e-05,
+      "loss": 0.4738,
+      "step": 77
+    },
+    {
+      "epoch": 0.3159493670886076,
+      "grad_norm": 0.7909596016302085,
+      "learning_rate": 2.536585365853659e-05,
+      "loss": 0.4784,
+      "step": 78
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.8407640920780911,
+      "learning_rate": 2.569105691056911e-05,
+      "loss": 0.4715,
+      "step": 79
+    },
+    {
+      "epoch": 0.3240506329113924,
+      "grad_norm": 0.7463720045618459,
+      "learning_rate": 2.601626016260163e-05,
+      "loss": 0.4742,
+      "step": 80
+    },
+    {
+      "epoch": 0.3281012658227848,
+      "grad_norm": 0.6819813585726631,
+      "learning_rate": 2.634146341463415e-05,
+      "loss": 0.468,
+      "step": 81
+    },
+    {
+      "epoch": 0.3321518987341772,
+      "grad_norm": 0.7600420585930052,
+      "learning_rate": 2.6666666666666667e-05,
+      "loss": 0.464,
+      "step": 82
+    },
+    {
+      "epoch": 0.3362025316455696,
+      "grad_norm": 1.0135395050705016,
+      "learning_rate": 2.699186991869919e-05,
+      "loss": 0.4741,
+      "step": 83
+    },
+    {
+      "epoch": 0.34025316455696203,
+      "grad_norm": 1.0008765205437586,
+      "learning_rate": 2.731707317073171e-05,
+      "loss": 0.4601,
+      "step": 84
+    },
+    {
+      "epoch": 0.34430379746835443,
+      "grad_norm": 1.0632739041829378,
+      "learning_rate": 2.764227642276423e-05,
+      "loss": 0.4608,
+      "step": 85
+    },
+    {
+      "epoch": 0.34835443037974684,
+      "grad_norm": 1.0659747353790856,
+      "learning_rate": 2.796747967479675e-05,
+      "loss": 0.4631,
+      "step": 86
+    },
+    {
+      "epoch": 0.35240506329113924,
+      "grad_norm": 0.807306778001417,
+      "learning_rate": 2.829268292682927e-05,
+      "loss": 0.4792,
+      "step": 87
+    },
+    {
+      "epoch": 0.35645569620253165,
+      "grad_norm": 0.7173975116801105,
+      "learning_rate": 2.8617886178861792e-05,
+      "loss": 0.4809,
+      "step": 88
+    },
+    {
+      "epoch": 0.36050632911392405,
+      "grad_norm": 0.7445911552237261,
+      "learning_rate": 2.8943089430894314e-05,
+      "loss": 0.468,
+      "step": 89
+    },
+    {
+      "epoch": 0.36455696202531646,
+      "grad_norm": 0.7876433765660337,
+      "learning_rate": 2.926829268292683e-05,
+      "loss": 0.4656,
+      "step": 90
+    },
+    {
+      "epoch": 0.36860759493670886,
+      "grad_norm": 0.7331404705300154,
+      "learning_rate": 2.959349593495935e-05,
+      "loss": 0.4724,
+      "step": 91
+    },
+    {
+      "epoch": 0.37265822784810126,
+      "grad_norm": 0.6861741579764106,
+      "learning_rate": 2.991869918699187e-05,
+      "loss": 0.4617,
+      "step": 92
+    },
+    {
+      "epoch": 0.37670886075949367,
+      "grad_norm": 0.5470125127721497,
+      "learning_rate": 3.0243902439024392e-05,
+      "loss": 0.459,
+      "step": 93
+    },
+    {
+      "epoch": 0.3807594936708861,
+      "grad_norm": 0.6116205951781973,
+      "learning_rate": 3.056910569105692e-05,
+      "loss": 0.4523,
+      "step": 94
+    },
+    {
+      "epoch": 0.3848101265822785,
+      "grad_norm": 0.5209001357679004,
+      "learning_rate": 3.089430894308943e-05,
+      "loss": 0.4731,
+      "step": 95
+    },
+    {
+      "epoch": 0.3888607594936709,
+      "grad_norm": 0.6433871240545289,
+      "learning_rate": 3.1219512195121955e-05,
+      "loss": 0.4663,
+      "step": 96
+    },
+    {
+      "epoch": 0.3929113924050633,
+      "grad_norm": 0.7977691683216891,
+      "learning_rate": 3.154471544715447e-05,
+      "loss": 0.457,
+      "step": 97
+    },
+    {
+      "epoch": 0.3969620253164557,
+      "grad_norm": 0.9640276626741505,
+      "learning_rate": 3.186991869918699e-05,
+      "loss": 0.4634,
+      "step": 98
+    },
+    {
+      "epoch": 0.4010126582278481,
+      "grad_norm": 1.0215172679696565,
+      "learning_rate": 3.2195121951219514e-05,
+      "loss": 0.4616,
+      "step": 99
+    },
+    {
+      "epoch": 0.4050632911392405,
+      "grad_norm": 0.802662033046199,
+      "learning_rate": 3.2520325203252037e-05,
+      "loss": 0.4629,
+      "step": 100
+    },
+    {
+      "epoch": 0.4091139240506329,
+      "grad_norm": 0.6845933921131884,
+      "learning_rate": 3.284552845528456e-05,
+      "loss": 0.4523,
+      "step": 101
+    },
+    {
+      "epoch": 0.4131645569620253,
+      "grad_norm": 0.7914191182901353,
+      "learning_rate": 3.3170731707317074e-05,
+      "loss": 0.4659,
+      "step": 102
+    },
+    {
+      "epoch": 0.4172151898734177,
+      "grad_norm": 0.9561174622826782,
+      "learning_rate": 3.3495934959349596e-05,
+      "loss": 0.4594,
+      "step": 103
+    },
+    {
+      "epoch": 0.4212658227848101,
+      "grad_norm": 0.8191259440208818,
+      "learning_rate": 3.382113821138212e-05,
+      "loss": 0.4604,
+      "step": 104
+    },
+    {
+      "epoch": 0.4253164556962025,
+      "grad_norm": 0.8630470703123507,
+      "learning_rate": 3.414634146341463e-05,
+      "loss": 0.4636,
+      "step": 105
+    },
+    {
+      "epoch": 0.4293670886075949,
+      "grad_norm": 0.9059217293305548,
+      "learning_rate": 3.4471544715447155e-05,
+      "loss": 0.461,
+      "step": 106
+    },
+    {
+      "epoch": 0.43341772151898733,
+      "grad_norm": 1.0786934547130973,
+      "learning_rate": 3.479674796747968e-05,
+      "loss": 0.4613,
+      "step": 107
+    },
+    {
+      "epoch": 0.43746835443037974,
+      "grad_norm": 0.9153265051062183,
+      "learning_rate": 3.51219512195122e-05,
+      "loss": 0.4617,
+      "step": 108
+    },
+    {
+      "epoch": 0.44151898734177214,
+      "grad_norm": 1.0873742727511382,
+      "learning_rate": 3.544715447154472e-05,
+      "loss": 0.4648,
+      "step": 109
+    },
+    {
+      "epoch": 0.44556962025316454,
+      "grad_norm": 0.9793158152273732,
+      "learning_rate": 3.577235772357724e-05,
+      "loss": 0.4519,
+      "step": 110
+    },
+    {
+      "epoch": 0.44962025316455695,
+      "grad_norm": 0.5245875968974063,
+      "learning_rate": 3.609756097560976e-05,
+      "loss": 0.4595,
+      "step": 111
+    },
+    {
+      "epoch": 0.45367088607594935,
+      "grad_norm": 1.0162869609507255,
+      "learning_rate": 3.642276422764228e-05,
+      "loss": 0.4638,
+      "step": 112
+    },
+    {
+      "epoch": 0.45772151898734176,
+      "grad_norm": 1.1944824869660204,
+      "learning_rate": 3.67479674796748e-05,
+      "loss": 0.461,
+      "step": 113
+    },
+    {
+      "epoch": 0.46177215189873416,
+      "grad_norm": 0.7626247954683802,
+      "learning_rate": 3.7073170731707325e-05,
+      "loss": 0.4465,
+      "step": 114
+    },
+    {
+      "epoch": 0.46582278481012657,
+      "grad_norm": 1.1424395235787947,
+      "learning_rate": 3.739837398373984e-05,
+      "loss": 0.4732,
+      "step": 115
+    },
+    {
+      "epoch": 0.46987341772151897,
+      "grad_norm": 0.9944993647334079,
+      "learning_rate": 3.772357723577236e-05,
+      "loss": 0.4625,
+      "step": 116
+    },
+    {
+      "epoch": 0.4739240506329114,
+      "grad_norm": 1.1458352955391256,
+      "learning_rate": 3.804878048780488e-05,
+      "loss": 0.456,
+      "step": 117
+    },
+    {
+      "epoch": 0.4779746835443038,
+      "grad_norm": 0.7832924821015952,
+      "learning_rate": 3.83739837398374e-05,
+      "loss": 0.45,
+      "step": 118
+    },
+    {
+      "epoch": 0.4820253164556962,
+      "grad_norm": 1.121473080782614,
+      "learning_rate": 3.869918699186992e-05,
+      "loss": 0.4604,
+      "step": 119
+    },
+    {
+      "epoch": 0.4860759493670886,
+      "grad_norm": 0.7413167929000939,
+      "learning_rate": 3.9024390243902444e-05,
+      "loss": 0.4494,
+      "step": 120
+    },
+    {
+      "epoch": 0.490126582278481,
+      "grad_norm": 1.204755868686124,
+      "learning_rate": 3.9349593495934966e-05,
+      "loss": 0.4509,
+      "step": 121
+    },
+    {
+      "epoch": 0.4941772151898734,
+      "grad_norm": 0.5485774592099574,
+      "learning_rate": 3.967479674796748e-05,
+      "loss": 0.4448,
+      "step": 122
+    },
+    {
+      "epoch": 0.4982278481012658,
+      "grad_norm": 1.1525577136768412,
+      "learning_rate": 4e-05,
+      "loss": 0.4702,
+      "step": 123
+    },
+    {
+      "epoch": 0.5022784810126583,
+      "grad_norm": 0.6163604663899699,
+      "learning_rate": 3.999991946137476e-05,
+      "loss": 0.4394,
+      "step": 124
+    },
+    {
+      "epoch": 0.5063291139240507,
+      "grad_norm": 0.8262732444479524,
+      "learning_rate": 3.999967784614766e-05,
+      "loss": 0.4517,
+      "step": 125
+    },
+    {
+      "epoch": 0.5103797468354431,
+      "grad_norm": 0.9010458487035942,
+      "learning_rate": 3.9999275156264656e-05,
+      "loss": 0.4507,
+      "step": 126
+    },
+    {
+      "epoch": 0.5144303797468355,
+      "grad_norm": 0.6370298937860062,
+      "learning_rate": 3.999871139496895e-05,
+      "loss": 0.4457,
+      "step": 127
+    },
+    {
+      "epoch": 0.5184810126582279,
+      "grad_norm": 0.7449299194262946,
+      "learning_rate": 3.9997986566800995e-05,
+      "loss": 0.4509,
+      "step": 128
+    },
+    {
+      "epoch": 0.5225316455696203,
+      "grad_norm": 0.5311158505789502,
+      "learning_rate": 3.999710067759846e-05,
+      "loss": 0.4379,
+      "step": 129
+    },
+    {
+      "epoch": 0.5265822784810127,
+      "grad_norm": 0.8179410465595235,
+      "learning_rate": 3.999605373449617e-05,
+      "loss": 0.4541,
+      "step": 130
+    },
+    {
+      "epoch": 0.5306329113924051,
+      "grad_norm": 0.6773485680435168,
+      "learning_rate": 3.9994845745926075e-05,
+      "loss": 0.4506,
+      "step": 131
+    },
+    {
+      "epoch": 0.5346835443037975,
+      "grad_norm": 0.6648548247184675,
+      "learning_rate": 3.999347672161713e-05,
+      "loss": 0.4606,
+      "step": 132
+    },
+    {
+      "epoch": 0.5387341772151899,
+      "grad_norm": 0.813340025702157,
+      "learning_rate": 3.999194667259528e-05,
+      "loss": 0.464,
+      "step": 133
+    },
+    {
+      "epoch": 0.5427848101265823,
+      "grad_norm": 0.613531858354214,
+      "learning_rate": 3.999025561118334e-05,
+      "loss": 0.4646,
+      "step": 134
+    },
+    {
+      "epoch": 0.5468354430379747,
+      "grad_norm": 0.666876637328891,
+      "learning_rate": 3.998840355100086e-05,
+      "loss": 0.4518,
+      "step": 135
+    },
+    {
+      "epoch": 0.5508860759493671,
+      "grad_norm": 0.6436840727391604,
+      "learning_rate": 3.998639050696409e-05,
+      "loss": 0.4429,
+      "step": 136
+    },
+    {
+      "epoch": 0.5549367088607595,
+      "grad_norm": 0.602775289661446,
+      "learning_rate": 3.998421649528582e-05,
+      "loss": 0.4615,
+      "step": 137
+    },
+    {
+      "epoch": 0.5589873417721519,
+      "grad_norm": 0.548785520848227,
+      "learning_rate": 3.9981881533475234e-05,
+      "loss": 0.4527,
+      "step": 138
+    },
+    {
+      "epoch": 0.5630379746835443,
+      "grad_norm": 0.5692275206288651,
+      "learning_rate": 3.997938564033779e-05,
+      "loss": 0.4613,
+      "step": 139
+    },
+    {
+      "epoch": 0.5670886075949367,
+      "grad_norm": 0.728781095156738,
+      "learning_rate": 3.9976728835975064e-05,
+      "loss": 0.4547,
+      "step": 140
+    },
+    {
+      "epoch": 0.5711392405063291,
+      "grad_norm": 0.8004419045265093,
+      "learning_rate": 3.9973911141784605e-05,
+      "loss": 0.4554,
+      "step": 141
+    },
+    {
+      "epoch": 0.5751898734177215,
+      "grad_norm": 0.6193374648541174,
+      "learning_rate": 3.997093258045973e-05,
+      "loss": 0.4457,
+      "step": 142
+    },
+    {
+      "epoch": 0.579240506329114,
+      "grad_norm": 0.6284220052427476,
+      "learning_rate": 3.996779317598936e-05,
+      "loss": 0.44,
+      "step": 143
+    },
+    {
+      "epoch": 0.5832911392405064,
+      "grad_norm": 0.6184767831916409,
+      "learning_rate": 3.996449295365782e-05,
+      "loss": 0.4516,
+      "step": 144
+    },
+    {
+      "epoch": 0.5873417721518988,
+      "grad_norm": 0.595315385533298,
+      "learning_rate": 3.996103194004467e-05,
+      "loss": 0.4457,
+      "step": 145
+    },
+    {
+      "epoch": 0.5913924050632912,
+      "grad_norm": 0.6974209999759629,
+      "learning_rate": 3.995741016302441e-05,
+      "loss": 0.4434,
+      "step": 146
+    },
+    {
+      "epoch": 0.5954430379746836,
+      "grad_norm": 0.5817176577783566,
+      "learning_rate": 3.9953627651766364e-05,
+      "loss": 0.4387,
+      "step": 147
+    },
+    {
+      "epoch": 0.599493670886076,
+      "grad_norm": 0.7392081555050031,
+      "learning_rate": 3.9949684436734325e-05,
+      "loss": 0.4626,
+      "step": 148
+    },
+    {
+      "epoch": 0.6035443037974684,
+      "grad_norm": 0.690960856238991,
+      "learning_rate": 3.994558054968643e-05,
+      "loss": 0.4546,
+      "step": 149
+    },
+    {
+      "epoch": 0.6075949367088608,
+      "grad_norm": 0.6733543828926696,
+      "learning_rate": 3.994131602367481e-05,
+      "loss": 0.4416,
+      "step": 150
+    },
+    {
+      "epoch": 0.6116455696202532,
+      "grad_norm": 0.7631669375333339,
+      "learning_rate": 3.9936890893045376e-05,
+      "loss": 0.4413,
+      "step": 151
+    },
+    {
+      "epoch": 0.6156962025316456,
+      "grad_norm": 0.6178961875750846,
+      "learning_rate": 3.993230519343752e-05,
+      "loss": 0.441,
+      "step": 152
+    },
+    {
+      "epoch": 0.619746835443038,
+      "grad_norm": 0.6842638191128397,
+      "learning_rate": 3.992755896178383e-05,
+      "loss": 0.4627,
+      "step": 153
+    },
+    {
+      "epoch": 0.6237974683544304,
+      "grad_norm": 0.9287754231320403,
+      "learning_rate": 3.992265223630981e-05,
+      "loss": 0.4476,
+      "step": 154
+    },
+    {
+      "epoch": 0.6278481012658228,
+      "grad_norm": 0.6791638988820639,
+      "learning_rate": 3.991758505653355e-05,
+      "loss": 0.4434,
+      "step": 155
+    },
+    {
+      "epoch": 0.6318987341772152,
+      "grad_norm": 0.6781921652978049,
+      "learning_rate": 3.991235746326543e-05,
+      "loss": 0.4474,
+      "step": 156
+    },
+    {
+      "epoch": 0.6359493670886076,
+      "grad_norm": 0.7533971297281138,
+      "learning_rate": 3.9906969498607745e-05,
+      "loss": 0.4602,
+      "step": 157
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.49430671276627275,
+      "learning_rate": 3.990142120595444e-05,
+      "loss": 0.4384,
+      "step": 158
+    },
+    {
+      "epoch": 0.6440506329113924,
+      "grad_norm": 0.5533636674902206,
+      "learning_rate": 3.98957126299907e-05,
+      "loss": 0.4372,
+      "step": 159
+    },
+    {
+      "epoch": 0.6481012658227848,
+      "grad_norm": 0.5253616684971629,
+      "learning_rate": 3.9889843816692596e-05,
+      "loss": 0.4338,
+      "step": 160
+    },
+    {
+      "epoch": 0.6521518987341772,
+      "grad_norm": 0.5925734609341349,
+      "learning_rate": 3.9883814813326766e-05,
+      "loss": 0.435,
+      "step": 161
+    },
+    {
+      "epoch": 0.6562025316455696,
+      "grad_norm": 0.5658199409263178,
+      "learning_rate": 3.9877625668449956e-05,
+      "loss": 0.4621,
+      "step": 162
+    },
+    {
+      "epoch": 0.660253164556962,
+      "grad_norm": 0.4784078951569835,
+      "learning_rate": 3.98712764319087e-05,
+      "loss": 0.4377,
+      "step": 163
+    },
+    {
+      "epoch": 0.6643037974683544,
+      "grad_norm": 0.5196850655712263,
+      "learning_rate": 3.9864767154838864e-05,
+      "loss": 0.441,
+      "step": 164
+    },
+    {
+      "epoch": 0.6683544303797468,
+      "grad_norm": 0.5838811480199775,
+      "learning_rate": 3.9858097889665277e-05,
+      "loss": 0.4536,
+      "step": 165
+    },
+    {
+      "epoch": 0.6724050632911392,
+      "grad_norm": 0.49826245261004853,
+      "learning_rate": 3.985126869010129e-05,
+      "loss": 0.4417,
+      "step": 166
+    },
+    {
+      "epoch": 0.6764556962025317,
+      "grad_norm": 0.6278030307336518,
+      "learning_rate": 3.984427961114833e-05,
+      "loss": 0.4334,
+      "step": 167
+    },
+    {
+      "epoch": 0.6805063291139241,
+      "grad_norm": 0.5441333245929146,
+      "learning_rate": 3.9837130709095475e-05,
+      "loss": 0.4423,
+      "step": 168
+    },
+    {
+      "epoch": 0.6845569620253165,
+      "grad_norm": 0.4918019941220083,
+      "learning_rate": 3.982982204151901e-05,
+      "loss": 0.4522,
+      "step": 169
+    },
+    {
+      "epoch": 0.6886075949367089,
+      "grad_norm": 0.47021555511118224,
+      "learning_rate": 3.982235366728193e-05,
+      "loss": 0.4534,
+      "step": 170
+    },
+    {
+      "epoch": 0.6926582278481013,
+      "grad_norm": 0.4934022293088361,
+      "learning_rate": 3.9814725646533505e-05,
+      "loss": 0.449,
+      "step": 171
+    },
+    {
+      "epoch": 0.6967088607594937,
+      "grad_norm": 0.572989623326767,
+      "learning_rate": 3.9806938040708746e-05,
+      "loss": 0.4435,
+      "step": 172
+    },
+    {
+      "epoch": 0.7007594936708861,
+      "grad_norm": 0.44530540556924214,
+      "learning_rate": 3.9798990912527976e-05,
+      "loss": 0.4416,
+      "step": 173
+    },
+    {
+      "epoch": 0.7048101265822785,
+      "grad_norm": 0.4267564703176547,
+      "learning_rate": 3.979088432599627e-05,
+      "loss": 0.4525,
+      "step": 174
+    },
+    {
+      "epoch": 0.7088607594936709,
+      "grad_norm": 0.4557328704945752,
+      "learning_rate": 3.9782618346402964e-05,
+      "loss": 0.4476,
+      "step": 175
+    },
+    {
+      "epoch": 0.7129113924050633,
+      "grad_norm": 0.3890979193541277,
+      "learning_rate": 3.977419304032111e-05,
+      "loss": 0.4472,
+      "step": 176
+    },
+    {
+      "epoch": 0.7169620253164557,
+      "grad_norm": 0.44409544044017873,
+      "learning_rate": 3.976560847560697e-05,
+      "loss": 0.4496,
+      "step": 177
+    },
+    {
+      "epoch": 0.7210126582278481,
+      "grad_norm": 0.35817552090902616,
+      "learning_rate": 3.9756864721399456e-05,
+      "loss": 0.4304,
+      "step": 178
+    },
+    {
+      "epoch": 0.7250632911392405,
+      "grad_norm": 0.37135481052801783,
+      "learning_rate": 3.974796184811956e-05,
+      "loss": 0.4301,
+      "step": 179
+    },
+    {
+      "epoch": 0.7291139240506329,
+      "grad_norm": 0.5311316497172024,
+      "learning_rate": 3.973889992746979e-05,
+      "loss": 0.4335,
+      "step": 180
+    },
+    {
+      "epoch": 0.7331645569620253,
+      "grad_norm": 0.5547137903021998,
+      "learning_rate": 3.972967903243361e-05,
+      "loss": 0.4509,
+      "step": 181
+    },
+    {
+      "epoch": 0.7372151898734177,
+      "grad_norm": 0.5448385036830042,
+      "learning_rate": 3.972029923727486e-05,
+      "loss": 0.4399,
+      "step": 182
+    },
+    {
+      "epoch": 0.7412658227848101,
+      "grad_norm": 0.5730390403023856,
+      "learning_rate": 3.971076061753709e-05,
+      "loss": 0.44,
+      "step": 183
+    },
+    {
+      "epoch": 0.7453164556962025,
+      "grad_norm": 0.47024493387916044,
+      "learning_rate": 3.9701063250043066e-05,
+      "loss": 0.4335,
+      "step": 184
+    },
+    {
+      "epoch": 0.7493670886075949,
+      "grad_norm": 0.4876338801831159,
+      "learning_rate": 3.969120721289402e-05,
+      "loss": 0.445,
+      "step": 185
+    },
+    {
+      "epoch": 0.7534177215189873,
+      "grad_norm": 0.5303716277526078,
+      "learning_rate": 3.9681192585469146e-05,
+      "loss": 0.4322,
+      "step": 186
+    },
+    {
+      "epoch": 0.7574683544303797,
+      "grad_norm": 0.49705061208950635,
+      "learning_rate": 3.9671019448424865e-05,
+      "loss": 0.4489,
+      "step": 187
+    },
+    {
+      "epoch": 0.7615189873417721,
+      "grad_norm": 0.5352085236985481,
+      "learning_rate": 3.966068788369422e-05,
+      "loss": 0.4291,
+      "step": 188
+    },
+    {
+      "epoch": 0.7655696202531646,
+      "grad_norm": 0.4891214568612076,
+      "learning_rate": 3.965019797448622e-05,
+      "loss": 0.4383,
+      "step": 189
+    },
+    {
+      "epoch": 0.769620253164557,
+      "grad_norm": 0.5748138439442975,
+      "learning_rate": 3.963954980528515e-05,
+      "loss": 0.4293,
+      "step": 190
+    },
+    {
+      "epoch": 0.7736708860759494,
+      "grad_norm": 0.6882976301770074,
+      "learning_rate": 3.9628743461849905e-05,
+      "loss": 0.432,
+      "step": 191
+    },
+    {
+      "epoch": 0.7777215189873418,
+      "grad_norm": 0.7806943323818509,
+      "learning_rate": 3.961777903121329e-05,
+      "loss": 0.443,
+      "step": 192
+    },
+    {
+      "epoch": 0.7817721518987342,
+      "grad_norm": 0.6880300726849585,
+      "learning_rate": 3.960665660168131e-05,
+      "loss": 0.4254,
+      "step": 193
+    },
+    {
+      "epoch": 0.7858227848101266,
+      "grad_norm": 0.6627176166244353,
+      "learning_rate": 3.9595376262832485e-05,
+      "loss": 0.4362,
+      "step": 194
+    },
+    {
+      "epoch": 0.789873417721519,
+      "grad_norm": 0.4684754230880586,
+      "learning_rate": 3.9583938105517127e-05,
+      "loss": 0.4351,
+      "step": 195
+    },
+    {
+      "epoch": 0.7939240506329114,
+      "grad_norm": 0.697837017397665,
+      "learning_rate": 3.957234222185657e-05,
+      "loss": 0.4322,
+      "step": 196
+    },
+    {
+      "epoch": 0.7979746835443038,
+      "grad_norm": 0.8468087100136369,
+      "learning_rate": 3.9560588705242474e-05,
+      "loss": 0.4356,
+      "step": 197
+    },
+    {
+      "epoch": 0.8020253164556962,
+      "grad_norm": 0.6768397913904604,
+      "learning_rate": 3.954867765033605e-05,
+      "loss": 0.4514,
+      "step": 198
+    },
+    {
+      "epoch": 0.8060759493670886,
+      "grad_norm": 0.4343277270296152,
+      "learning_rate": 3.953660915306728e-05,
+      "loss": 0.4493,
+      "step": 199
+    },
+    {
+      "epoch": 0.810126582278481,
+      "grad_norm": 0.6544632865240597,
+      "learning_rate": 3.952438331063419e-05,
+      "loss": 0.4444,
+      "step": 200
+    },
+    {
+      "epoch": 0.8141772151898734,
+      "grad_norm": 0.7357169343142697,
+      "learning_rate": 3.951200022150205e-05,
+      "loss": 0.4416,
+      "step": 201
+    },
+    {
+      "epoch": 0.8182278481012658,
+      "grad_norm": 0.6608233311641407,
+      "learning_rate": 3.949945998540253e-05,
+      "loss": 0.4311,
+      "step": 202
+    },
+    {
+      "epoch": 0.8222784810126582,
+      "grad_norm": 0.433406732774848,
+      "learning_rate": 3.9486762703332993e-05,
+      "loss": 0.4275,
+      "step": 203
+    },
+    {
+      "epoch": 0.8263291139240506,
+      "grad_norm": 0.574359085170714,
+      "learning_rate": 3.947390847755559e-05,
+      "loss": 0.4327,
+      "step": 204
+    },
+    {
+      "epoch": 0.830379746835443,
+      "grad_norm": 0.5885411836868626,
+      "learning_rate": 3.946089741159648e-05,
+      "loss": 0.4338,
+      "step": 205
+    },
+    {
+      "epoch": 0.8344303797468354,
+      "grad_norm": 0.5279934074509568,
+      "learning_rate": 3.944772961024501e-05,
+      "loss": 0.433,
+      "step": 206
+    },
+    {
+      "epoch": 0.8384810126582278,
+      "grad_norm": 0.43933228959587,
+      "learning_rate": 3.943440517955285e-05,
+      "loss": 0.4346,
+      "step": 207
+    },
+    {
+      "epoch": 0.8425316455696202,
+      "grad_norm": 0.5358246951979698,
+      "learning_rate": 3.9420924226833126e-05,
+      "loss": 0.4259,
+      "step": 208
+    },
+    {
+      "epoch": 0.8465822784810126,
+      "grad_norm": 0.4730612206238966,
+      "learning_rate": 3.9407286860659566e-05,
+      "loss": 0.4473,
+      "step": 209
+    },
+    {
+      "epoch": 0.850632911392405,
+      "grad_norm": 0.5059978180558838,
+      "learning_rate": 3.9393493190865657e-05,
+      "loss": 0.4267,
+      "step": 210
+    },
+    {
+      "epoch": 0.8546835443037974,
+      "grad_norm": 0.4352484384023733,
+      "learning_rate": 3.937954332854371e-05,
+      "loss": 0.4356,
+      "step": 211
+    },
+    {
+      "epoch": 0.8587341772151899,
+      "grad_norm": 0.4051353662775338,
+      "learning_rate": 3.9365437386044016e-05,
+      "loss": 0.4372,
+      "step": 212
+    },
+    {
+      "epoch": 0.8627848101265823,
+      "grad_norm": 0.4423966426612205,
+      "learning_rate": 3.935117547697387e-05,
+      "loss": 0.4329,
+      "step": 213
+    },
+    {
+      "epoch": 0.8668354430379747,
+      "grad_norm": 0.41562957136203643,
+      "learning_rate": 3.933675771619675e-05,
+      "loss": 0.441,
+      "step": 214
+    },
+    {
+      "epoch": 0.8708860759493671,
+      "grad_norm": 0.45752003808372294,
+      "learning_rate": 3.932218421983131e-05,
+      "loss": 0.4434,
+      "step": 215
+    },
+    {
+      "epoch": 0.8749367088607595,
+      "grad_norm": 0.45608371737288117,
+      "learning_rate": 3.9307455105250484e-05,
+      "loss": 0.4149,
+      "step": 216
+    },
+    {
+      "epoch": 0.8789873417721519,
+      "grad_norm": 0.5499621229994818,
+      "learning_rate": 3.929257049108054e-05,
+      "loss": 0.4485,
+      "step": 217
+    },
+    {
+      "epoch": 0.8830379746835443,
+      "grad_norm": 0.5562946040013638,
+      "learning_rate": 3.927753049720011e-05,
+      "loss": 0.4366,
+      "step": 218
+    },
+    {
+      "epoch": 0.8870886075949367,
+      "grad_norm": 0.447872473908021,
+      "learning_rate": 3.9262335244739234e-05,
+      "loss": 0.4505,
+      "step": 219
+    },
+    {
+      "epoch": 0.8911392405063291,
+      "grad_norm": 0.340452152081178,
+      "learning_rate": 3.92469848560784e-05,
+      "loss": 0.4233,
+      "step": 220
+    },
+    {
+      "epoch": 0.8951898734177215,
+      "grad_norm": 0.425475717212266,
+      "learning_rate": 3.923147945484751e-05,
+      "loss": 0.4306,
+      "step": 221
+    },
+    {
+      "epoch": 0.8992405063291139,
+      "grad_norm": 0.5042619296085626,
+      "learning_rate": 3.9215819165924956e-05,
+      "loss": 0.4284,
+      "step": 222
+    },
+    {
+      "epoch": 0.9032911392405063,
+      "grad_norm": 0.42530764996283527,
+      "learning_rate": 3.920000411543654e-05,
+      "loss": 0.4142,
+      "step": 223
+    },
+    {
+      "epoch": 0.9073417721518987,
+      "grad_norm": 0.4101270688763705,
+      "learning_rate": 3.9184034430754495e-05,
+      "loss": 0.4351,
+      "step": 224
+    },
+    {
+      "epoch": 0.9113924050632911,
+      "grad_norm": 0.49612510579696373,
+      "learning_rate": 3.916791024049648e-05,
+      "loss": 0.4349,
+      "step": 225
+    },
+    {
+      "epoch": 0.9154430379746835,
+      "grad_norm": 0.4378191750494089,
+      "learning_rate": 3.91516316745245e-05,
+      "loss": 0.4302,
+      "step": 226
+    },
+    {
+      "epoch": 0.9194936708860759,
+      "grad_norm": 0.465002690881198,
+      "learning_rate": 3.913519886394389e-05,
+      "loss": 0.4139,
+      "step": 227
+    },
+    {
+      "epoch": 0.9235443037974683,
+      "grad_norm": 0.5330927432168837,
+      "learning_rate": 3.911861194110225e-05,
+      "loss": 0.4335,
+      "step": 228
+    },
+    {
+      "epoch": 0.9275949367088607,
+      "grad_norm": 0.46852503667488293,
+      "learning_rate": 3.910187103958837e-05,
+      "loss": 0.4405,
+      "step": 229
+    },
+    {
+      "epoch": 0.9316455696202531,
+      "grad_norm": 0.7451388353219125,
+      "learning_rate": 3.908497629423117e-05,
+      "loss": 0.4348,
+      "step": 230
+    },
+    {
+      "epoch": 0.9356962025316455,
+      "grad_norm": 0.5811759504902709,
+      "learning_rate": 3.9067927841098614e-05,
+      "loss": 0.4236,
+      "step": 231
+    },
+    {
+      "epoch": 0.9397468354430379,
+      "grad_norm": 0.5276316901799204,
+      "learning_rate": 3.9050725817496594e-05,
+      "loss": 0.4408,
+      "step": 232
+    },
+    {
+      "epoch": 0.9437974683544303,
+      "grad_norm": 0.9195084087667525,
+      "learning_rate": 3.9033370361967844e-05,
+      "loss": 0.4323,
+      "step": 233
+    },
+    {
+      "epoch": 0.9478481012658228,
+      "grad_norm": 0.5231606113605961,
+      "learning_rate": 3.901586161429081e-05,
+      "loss": 0.4299,
+      "step": 234
+    },
+    {
+      "epoch": 0.9518987341772152,
+      "grad_norm": 0.7809649790270857,
+      "learning_rate": 3.8998199715478545e-05,
+      "loss": 0.4287,
+      "step": 235
+    },
+    {
+      "epoch": 0.9559493670886076,
+      "grad_norm": 0.5064453671965894,
+      "learning_rate": 3.8980384807777564e-05,
+      "loss": 0.4459,
+      "step": 236
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.5887753448981267,
+      "learning_rate": 3.896241703466667e-05,
+      "loss": 0.4458,
+      "step": 237
+    },
+    {
+      "epoch": 0.9640506329113924,
+      "grad_norm": 0.5589882344160915,
+      "learning_rate": 3.894429654085585e-05,
+      "loss": 0.439,
+      "step": 238
+    },
+    {
+      "epoch": 0.9681012658227848,
+      "grad_norm": 0.5007184532119229,
+      "learning_rate": 3.892602347228505e-05,
+      "loss": 0.4346,
+      "step": 239
+    },
+    {
+      "epoch": 0.9721518987341772,
+      "grad_norm": 0.43452908477873575,
+      "learning_rate": 3.890759797612307e-05,
+      "loss": 0.4262,
+      "step": 240
+    },
+    {
+      "epoch": 0.9762025316455696,
+      "grad_norm": 0.4707734985854168,
+      "learning_rate": 3.888902020076632e-05,
+      "loss": 0.4391,
+      "step": 241
+    },
+    {
+      "epoch": 0.980253164556962,
+      "grad_norm": 0.5092716017907045,
+      "learning_rate": 3.887029029583764e-05,
+      "loss": 0.4361,
+      "step": 242
+    },
+    {
+      "epoch": 0.9843037974683544,
+      "grad_norm": 0.48277807201559775,
+      "learning_rate": 3.8851408412185125e-05,
+      "loss": 0.4223,
+      "step": 243
+    },
+    {
+      "epoch": 0.9883544303797468,
+      "grad_norm": 0.4319716345727951,
+      "learning_rate": 3.8832374701880855e-05,
+      "loss": 0.4356,
+      "step": 244
+    },
+    {
+      "epoch": 0.9924050632911392,
+      "grad_norm": 0.36445533552238885,
+      "learning_rate": 3.881318931821972e-05,
+      "loss": 0.428,
+      "step": 245
+    },
+    {
+      "epoch": 0.9964556962025316,
+      "grad_norm": 0.45837182185855746,
+      "learning_rate": 3.879385241571817e-05,
+      "loss": 0.4115,
+      "step": 246
+    },
+    {
+      "epoch": 1.000506329113924,
+      "grad_norm": 0.42846713717171087,
+      "learning_rate": 3.8774364150112955e-05,
+      "loss": 0.417,
+      "step": 247
+    },
+    {
+      "epoch": 1.0045569620253165,
+      "grad_norm": 0.4475160130693605,
+      "learning_rate": 3.8754724678359884e-05,
+      "loss": 0.4088,
+      "step": 248
+    },
+    {
+      "epoch": 1.0086075949367088,
+      "grad_norm": 0.5940227208665126,
+      "learning_rate": 3.873493415863256e-05,
+      "loss": 0.4026,
+      "step": 249
+    },
+    {
+      "epoch": 1.0126582278481013,
+      "grad_norm": 0.5984809580575989,
+      "learning_rate": 3.871499275032111e-05,
+      "loss": 0.391,
+      "step": 250
+    },
+    {
+      "epoch": 1.0167088607594936,
+      "grad_norm": 0.6455531442080304,
+      "learning_rate": 3.869490061403091e-05,
+      "loss": 0.3989,
+      "step": 251
+    },
+    {
+      "epoch": 1.0207594936708861,
+      "grad_norm": 0.5830108701651686,
+      "learning_rate": 3.867465791158124e-05,
+      "loss": 0.4025,
+      "step": 252
+    },
+    {
+      "epoch": 1.0248101265822784,
+      "grad_norm": 0.492185634155446,
+      "learning_rate": 3.865426480600407e-05,
+      "loss": 0.3817,
+      "step": 253
+    },
+    {
+      "epoch": 1.028860759493671,
+      "grad_norm": 0.5630901998110869,
+      "learning_rate": 3.863372146154264e-05,
+      "loss": 0.3947,
+      "step": 254
+    },
+    {
+      "epoch": 1.0329113924050632,
+      "grad_norm": 0.5738707214476045,
+      "learning_rate": 3.861302804365024e-05,
+      "loss": 0.3894,
+      "step": 255
+    },
+    {
+      "epoch": 1.0369620253164558,
+      "grad_norm": 0.5096629511966352,
+      "learning_rate": 3.85921847189888e-05,
+      "loss": 0.399,
+      "step": 256
+    },
+    {
+      "epoch": 1.041012658227848,
+      "grad_norm": 0.5277544857558533,
+      "learning_rate": 3.85711916554276e-05,
+      "loss": 0.3961,
+      "step": 257
+    },
+    {
+      "epoch": 1.0450632911392406,
+      "grad_norm": 0.6010517948552411,
+      "learning_rate": 3.85500490220419e-05,
+      "loss": 0.4013,
+      "step": 258
+    },
+    {
+      "epoch": 1.0491139240506329,
+      "grad_norm": 0.5437842678117706,
+      "learning_rate": 3.852875698911154e-05,
+      "loss": 0.4018,
+      "step": 259
+    },
+    {
+      "epoch": 1.0531645569620254,
+      "grad_norm": 0.4869754931899248,
+      "learning_rate": 3.850731572811963e-05,
+      "loss": 0.39,
+      "step": 260
+    },
+    {
+      "epoch": 1.0572151898734177,
+      "grad_norm": 0.43596926830493293,
+      "learning_rate": 3.848572541175116e-05,
+      "loss": 0.3935,
+      "step": 261
+    },
+    {
+      "epoch": 1.0612658227848102,
+      "grad_norm": 0.5360513519539389,
+      "learning_rate": 3.846398621389154e-05,
+      "loss": 0.3983,
+      "step": 262
+    },
+    {
+      "epoch": 1.0653164556962025,
+      "grad_norm": 0.5419099925766745,
+      "learning_rate": 3.84420983096253e-05,
+      "loss": 0.4016,
+      "step": 263
+    },
+    {
+      "epoch": 1.069367088607595,
+      "grad_norm": 0.361264290878164,
+      "learning_rate": 3.8420061875234606e-05,
+      "loss": 0.3982,
+      "step": 264
+    },
+    {
+      "epoch": 1.0734177215189873,
+      "grad_norm": 0.5043585555376391,
+      "learning_rate": 3.839787708819787e-05,
+      "loss": 0.3983,
+      "step": 265
+    },
+    {
+      "epoch": 1.0774683544303798,
+      "grad_norm": 0.47161825464823015,
+      "learning_rate": 3.8375544127188325e-05,
+      "loss": 0.4145,
+      "step": 266
+    },
+    {
+      "epoch": 1.081518987341772,
+      "grad_norm": 0.4270391331689771,
+      "learning_rate": 3.8353063172072564e-05,
+      "loss": 0.3911,
+      "step": 267
+    },
+    {
+      "epoch": 1.0855696202531646,
+      "grad_norm": 0.47905940013690723,
+      "learning_rate": 3.8330434403909105e-05,
+      "loss": 0.3883,
+      "step": 268
+    },
+    {
+      "epoch": 1.089620253164557,
+      "grad_norm": 0.4373718790641677,
+      "learning_rate": 3.8307658004946934e-05,
+      "loss": 0.4065,
+      "step": 269
+    },
+    {
+      "epoch": 1.0936708860759494,
+      "grad_norm": 0.4465192343102378,
+      "learning_rate": 3.8284734158624046e-05,
+      "loss": 0.4003,
+      "step": 270
+    },
+    {
+      "epoch": 1.0977215189873417,
+      "grad_norm": 0.38632458039364553,
+      "learning_rate": 3.826166304956594e-05,
+      "loss": 0.3912,
+      "step": 271
+    },
+    {
+      "epoch": 1.1017721518987342,
+      "grad_norm": 0.48464805437641706,
+      "learning_rate": 3.8238444863584164e-05,
+      "loss": 0.4028,
+      "step": 272
+    },
+    {
+      "epoch": 1.1058227848101265,
+      "grad_norm": 0.5256830006632051,
+      "learning_rate": 3.821507978767479e-05,
+      "loss": 0.3911,
+      "step": 273
+    },
+    {
+      "epoch": 1.109873417721519,
+      "grad_norm": 0.386820027199745,
+      "learning_rate": 3.819156801001693e-05,
+      "loss": 0.3886,
+      "step": 274
+    },
+    {
+      "epoch": 1.1139240506329113,
+      "grad_norm": 0.37162224846775394,
+      "learning_rate": 3.816790971997121e-05,
+      "loss": 0.3754,
+      "step": 275
+    },
+    {
+      "epoch": 1.1179746835443038,
+      "grad_norm": 0.552140139971997,
+      "learning_rate": 3.8144105108078246e-05,
+      "loss": 0.3976,
+      "step": 276
+    },
+    {
+      "epoch": 1.1220253164556961,
+      "grad_norm": 0.48144142019180364,
+      "learning_rate": 3.81201543660571e-05,
+      "loss": 0.3964,
+      "step": 277
+    },
+    {
+      "epoch": 1.1260759493670887,
+      "grad_norm": 0.4075361500553016,
+      "learning_rate": 3.809605768680377e-05,
+      "loss": 0.393,
+      "step": 278
+    },
+    {
+      "epoch": 1.130126582278481,
+      "grad_norm": 0.40716277469199785,
+      "learning_rate": 3.807181526438958e-05,
+      "loss": 0.3943,
+      "step": 279
+    },
+    {
+      "epoch": 1.1341772151898735,
+      "grad_norm": 0.4845807288582223,
+      "learning_rate": 3.8047427294059697e-05,
+      "loss": 0.3872,
+      "step": 280
+    },
+    {
+      "epoch": 1.1382278481012658,
+      "grad_norm": 0.46033742402787553,
+      "learning_rate": 3.802289397223145e-05,
+      "loss": 0.4035,
+      "step": 281
+    },
+    {
+      "epoch": 1.1422784810126583,
+      "grad_norm": 0.47130214096300227,
+      "learning_rate": 3.7998215496492854e-05,
+      "loss": 0.3768,
+      "step": 282
+    },
+    {
+      "epoch": 1.1463291139240506,
+      "grad_norm": 0.525857484293531,
+      "learning_rate": 3.797339206560096e-05,
+      "loss": 0.395,
+      "step": 283
+    },
+    {
+      "epoch": 1.150379746835443,
+      "grad_norm": 0.36396194003969784,
+      "learning_rate": 3.794842387948027e-05,
+      "loss": 0.3849,
+      "step": 284
+    },
+    {
+      "epoch": 1.1544303797468354,
+      "grad_norm": 0.4588995748295555,
+      "learning_rate": 3.7923311139221114e-05,
+      "loss": 0.3833,
+      "step": 285
+    },
+    {
+      "epoch": 1.158481012658228,
+      "grad_norm": 0.3815038224280723,
+      "learning_rate": 3.7898054047078054e-05,
+      "loss": 0.3877,
+      "step": 286
+    },
+    {
+      "epoch": 1.1625316455696202,
+      "grad_norm": 0.3792786182390277,
+      "learning_rate": 3.787265280646825e-05,
+      "loss": 0.3875,
+      "step": 287
+    },
+    {
+      "epoch": 1.1665822784810127,
+      "grad_norm": 0.3463294601751543,
+      "learning_rate": 3.7847107621969786e-05,
+      "loss": 0.3824,
+      "step": 288
+    },
+    {
+      "epoch": 1.170632911392405,
+      "grad_norm": 0.3533948704501378,
+      "learning_rate": 3.7821418699320064e-05,
+      "loss": 0.4027,
+      "step": 289
+    },
+    {
+      "epoch": 1.1746835443037975,
+      "grad_norm": 0.383282579294443,
+      "learning_rate": 3.7795586245414145e-05,
+      "loss": 0.3829,
+      "step": 290
+    },
+    {
+      "epoch": 1.1787341772151898,
+      "grad_norm": 0.4581540129948383,
+      "learning_rate": 3.776961046830306e-05,
+      "loss": 0.3884,
+      "step": 291
+    },
+    {
+      "epoch": 1.1827848101265823,
+      "grad_norm": 0.31533948192061,
+      "learning_rate": 3.774349157719215e-05,
+      "loss": 0.3804,
+      "step": 292
+    },
+    {
+      "epoch": 1.1868354430379746,
+      "grad_norm": 0.45931504918110055,
+      "learning_rate": 3.7717229782439365e-05,
+      "loss": 0.3784,
+      "step": 293
+    },
+    {
+      "epoch": 1.1908860759493671,
+      "grad_norm": 0.385529161554429,
+      "learning_rate": 3.769082529555359e-05,
+      "loss": 0.3915,
+      "step": 294
+    },
+    {
+      "epoch": 1.1949367088607594,
+      "grad_norm": 0.5018597815210667,
+      "learning_rate": 3.766427832919294e-05,
+      "loss": 0.3994,
+      "step": 295
+    },
+    {
+      "epoch": 1.198987341772152,
+      "grad_norm": 0.39512673782390434,
+      "learning_rate": 3.7637589097163024e-05,
+      "loss": 0.3837,
+      "step": 296
+    },
+    {
+      "epoch": 1.2030379746835442,
+      "grad_norm": 0.47429372405601045,
+      "learning_rate": 3.761075781441526e-05,
+      "loss": 0.3932,
+      "step": 297
+    },
+    {
+      "epoch": 1.2070886075949367,
+      "grad_norm": 0.5020604723040009,
+      "learning_rate": 3.75837846970451e-05,
+      "loss": 0.4013,
+      "step": 298
+    },
+    {
+      "epoch": 1.211139240506329,
+      "grad_norm": 0.36772609445549226,
+      "learning_rate": 3.755666996229032e-05,
+      "loss": 0.3953,
+      "step": 299
+    },
+    {
+      "epoch": 1.2151898734177216,
+      "grad_norm": 0.5107517143831877,
+      "learning_rate": 3.752941382852927e-05,
+      "loss": 0.3949,
+      "step": 300
+    },
+    {
+      "epoch": 1.2192405063291138,
+      "grad_norm": 0.4558400282616621,
+      "learning_rate": 3.7502016515279115e-05,
+      "loss": 0.3903,
+      "step": 301
+    },
+    {
+      "epoch": 1.2232911392405064,
+      "grad_norm": 0.3705457410208397,
+      "learning_rate": 3.7474478243194043e-05,
+      "loss": 0.3932,
+      "step": 302
+    },
+    {
+      "epoch": 1.2273417721518987,
+      "grad_norm": 0.4705876925186712,
+      "learning_rate": 3.744679923406351e-05,
+      "loss": 0.3832,
+      "step": 303
+    },
+    {
+      "epoch": 1.2313924050632912,
+      "grad_norm": 0.4804883669070078,
+      "learning_rate": 3.741897971081043e-05,
+      "loss": 0.3991,
+      "step": 304
+    },
+    {
+      "epoch": 1.2354430379746835,
+      "grad_norm": 0.46520999685601383,
+      "learning_rate": 3.739101989748946e-05,
+      "loss": 0.3954,
+      "step": 305
+    },
+    {
+      "epoch": 1.239493670886076,
+      "grad_norm": 0.47325550939824357,
+      "learning_rate": 3.7362920019285066e-05,
+      "loss": 0.3889,
+      "step": 306
+    },
+    {
+      "epoch": 1.2435443037974683,
+      "grad_norm": 0.4849233515700871,
+      "learning_rate": 3.73346803025098e-05,
+      "loss": 0.3949,
+      "step": 307
+    },
+    {
+      "epoch": 1.2475949367088608,
+      "grad_norm": 0.464517746924604,
+      "learning_rate": 3.730630097460247e-05,
+      "loss": 0.4053,
+      "step": 308
+    },
+    {
+      "epoch": 1.251645569620253,
+      "grad_norm": 0.45802271406769574,
+      "learning_rate": 3.727778226412628e-05,
+      "loss": 0.3881,
+      "step": 309
+    },
+    {
+      "epoch": 1.2556962025316456,
+      "grad_norm": 0.4432164196221639,
+      "learning_rate": 3.7249124400767006e-05,
+      "loss": 0.394,
+      "step": 310
+    },
+    {
+      "epoch": 1.259746835443038,
+      "grad_norm": 0.5708930864850308,
+      "learning_rate": 3.722032761533114e-05,
+      "loss": 0.3968,
+      "step": 311
+    },
+    {
+      "epoch": 1.2637974683544304,
+      "grad_norm": 0.5130021744021385,
+      "learning_rate": 3.719139213974403e-05,
+      "loss": 0.4121,
+      "step": 312
+    },
+    {
+      "epoch": 1.2678481012658227,
+      "grad_norm": 0.31551363249863335,
+      "learning_rate": 3.7162318207048006e-05,
+      "loss": 0.3857,
+      "step": 313
+    },
+    {
+      "epoch": 1.2718987341772152,
+      "grad_norm": 0.5475930391432549,
+      "learning_rate": 3.713310605140055e-05,
+      "loss": 0.3984,
+      "step": 314
+    },
+    {
+      "epoch": 1.2759493670886077,
+      "grad_norm": 0.47601310300117133,
+      "learning_rate": 3.710375590807233e-05,
+      "loss": 0.3769,
+      "step": 315
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.33704554013660254,
+      "learning_rate": 3.7074268013445365e-05,
+      "loss": 0.3941,
+      "step": 316
+    },
+    {
+      "epoch": 1.2840506329113923,
+      "grad_norm": 0.41852272360394643,
+      "learning_rate": 3.7044642605011114e-05,
+      "loss": 0.3911,
+      "step": 317
+    },
+    {
+      "epoch": 1.2881012658227848,
+      "grad_norm": 0.38233786933627495,
+      "learning_rate": 3.701487992136854e-05,
+      "loss": 0.3896,
+      "step": 318
+    },
+    {
+      "epoch": 1.2921518987341774,
+      "grad_norm": 0.42233299838615085,
+      "learning_rate": 3.69849802022222e-05,
+      "loss": 0.3898,
+      "step": 319
+    },
+    {
+      "epoch": 1.2962025316455696,
+      "grad_norm": 0.4161038967515966,
+      "learning_rate": 3.6954943688380334e-05,
+      "loss": 0.3917,
+      "step": 320
+    },
+    {
+      "epoch": 1.300253164556962,
+      "grad_norm": 0.4648766752611507,
+      "learning_rate": 3.692477062175289e-05,
+      "loss": 0.3846,
+      "step": 321
+    },
+    {
+      "epoch": 1.3043037974683545,
+      "grad_norm": 0.4552268243131953,
+      "learning_rate": 3.689446124534958e-05,
+      "loss": 0.3969,
+      "step": 322
+    },
+    {
+      "epoch": 1.308354430379747,
+      "grad_norm": 0.32470285688438244,
+      "learning_rate": 3.686401580327799e-05,
+      "loss": 0.4024,
+      "step": 323
+    },
+    {
+      "epoch": 1.3124050632911393,
+      "grad_norm": 0.4962223664303966,
+      "learning_rate": 3.683343454074149e-05,
+      "loss": 0.3961,
+      "step": 324
+    },
+    {
+      "epoch": 1.3164556962025316,
+      "grad_norm": 0.410393736975107,
+      "learning_rate": 3.6802717704037386e-05,
+      "loss": 0.3941,
+      "step": 325
+    },
+    {
+      "epoch": 1.320506329113924,
+      "grad_norm": 0.4129386128714801,
+      "learning_rate": 3.6771865540554855e-05,
+      "loss": 0.3891,
+      "step": 326
+    },
+    {
+      "epoch": 1.3245569620253166,
+      "grad_norm": 0.3566631320933079,
+      "learning_rate": 3.674087829877297e-05,
+      "loss": 0.4105,
+      "step": 327
+    },
+    {
+      "epoch": 1.3286075949367089,
+      "grad_norm": 0.43111152172537187,
+      "learning_rate": 3.6709756228258735e-05,
+      "loss": 0.3931,
+      "step": 328
+    },
+    {
+      "epoch": 1.3326582278481012,
+      "grad_norm": 0.3555052781250808,
+      "learning_rate": 3.667849957966501e-05,
+      "loss": 0.3875,
+      "step": 329
+    },
+    {
+      "epoch": 1.3367088607594937,
+      "grad_norm": 0.3774229543111491,
+      "learning_rate": 3.6647108604728546e-05,
+      "loss": 0.3947,
+      "step": 330
+    },
+    {
+      "epoch": 1.3407594936708862,
+      "grad_norm": 0.5286882945104647,
+      "learning_rate": 3.661558355626795e-05,
+      "loss": 0.3864,
+      "step": 331
+    },
+    {
+      "epoch": 1.3448101265822785,
+      "grad_norm": 0.46547590467815364,
+      "learning_rate": 3.658392468818163e-05,
+      "loss": 0.4023,
+      "step": 332
+    },
+    {
+      "epoch": 1.3488607594936708,
+      "grad_norm": 0.33145392002591817,
+      "learning_rate": 3.655213225544574e-05,
+      "loss": 0.3911,
+      "step": 333
+    },
+    {
+      "epoch": 1.3529113924050633,
+      "grad_norm": 0.4934182361287748,
+      "learning_rate": 3.652020651411218e-05,
+      "loss": 0.3804,
+      "step": 334
+    },
+    {
+      "epoch": 1.3569620253164558,
+      "grad_norm": 0.38413713398172217,
+      "learning_rate": 3.6488147721306474e-05,
+      "loss": 0.4007,
+      "step": 335
+    },
+    {
+      "epoch": 1.3610126582278481,
+      "grad_norm": 0.424462068735566,
+      "learning_rate": 3.645595613522574e-05,
+      "loss": 0.3814,
+      "step": 336
+    },
+    {
+      "epoch": 1.3650632911392404,
+      "grad_norm": 0.3700388423680204,
+      "learning_rate": 3.642363201513657e-05,
+      "loss": 0.4015,
+      "step": 337
+    },
+    {
+      "epoch": 1.369113924050633,
+      "grad_norm": 0.3679921989043293,
+      "learning_rate": 3.6391175621373006e-05,
+      "loss": 0.3971,
+      "step": 338
+    },
+    {
+      "epoch": 1.3731645569620254,
+      "grad_norm": 0.4080762017550274,
+      "learning_rate": 3.6358587215334355e-05,
+      "loss": 0.3805,
+      "step": 339
+    },
+    {
+      "epoch": 1.3772151898734177,
+      "grad_norm": 0.42724774716092023,
+      "learning_rate": 3.632586705948318e-05,
+      "loss": 0.3768,
+      "step": 340
+    },
+    {
+      "epoch": 1.38126582278481,
+      "grad_norm": 0.35293570211547437,
+      "learning_rate": 3.629301541734311e-05,
+      "loss": 0.3969,
+      "step": 341
+    },
+    {
+      "epoch": 1.3853164556962025,
+      "grad_norm": 0.5131863284623499,
+      "learning_rate": 3.626003255349676e-05,
+      "loss": 0.3989,
+      "step": 342
+    },
+    {
+      "epoch": 1.389367088607595,
+      "grad_norm": 0.4943336116136505,
+      "learning_rate": 3.622691873358357e-05,
+      "loss": 0.3902,
+      "step": 343
+    },
+    {
+      "epoch": 1.3934177215189874,
+      "grad_norm": 0.403898080722612,
+      "learning_rate": 3.61936742242977e-05,
+      "loss": 0.3903,
+      "step": 344
+    },
+    {
+      "epoch": 1.3974683544303796,
+      "grad_norm": 0.5058606436343387,
+      "learning_rate": 3.6160299293385864e-05,
+      "loss": 0.389,
+      "step": 345
+    },
+    {
+      "epoch": 1.4015189873417722,
+      "grad_norm": 0.5275135133232849,
+      "learning_rate": 3.612679420964516e-05,
+      "loss": 0.3953,
+      "step": 346
+    },
+    {
+      "epoch": 1.4055696202531647,
+      "grad_norm": 0.39559404411190097,
+      "learning_rate": 3.609315924292092e-05,
+      "loss": 0.3825,
+      "step": 347
+    },
+    {
+      "epoch": 1.409620253164557,
+      "grad_norm": 0.48168357917572874,
+      "learning_rate": 3.6059394664104554e-05,
+      "loss": 0.3919,
+      "step": 348
+    },
+    {
+      "epoch": 1.4136708860759493,
+      "grad_norm": 0.6116413531171212,
+      "learning_rate": 3.602550074513133e-05,
+      "loss": 0.3825,
+      "step": 349
+    },
+    {
+      "epoch": 1.4177215189873418,
+      "grad_norm": 0.4034948676112011,
+      "learning_rate": 3.599147775897822e-05,
+      "loss": 0.3914,
+      "step": 350
+    },
+    {
+      "epoch": 1.4217721518987343,
+      "grad_norm": 0.47732984596073313,
+      "learning_rate": 3.595732597966167e-05,
+      "loss": 0.4034,
+      "step": 351
+    },
+    {
+      "epoch": 1.4258227848101266,
+      "grad_norm": 0.5790683093689306,
+      "learning_rate": 3.592304568223542e-05,
+      "loss": 0.397,
+      "step": 352
+    },
+    {
+      "epoch": 1.4298734177215189,
+      "grad_norm": 0.4100561278675857,
+      "learning_rate": 3.588863714278826e-05,
+      "loss": 0.3996,
+      "step": 353
+    },
+    {
+      "epoch": 1.4339240506329114,
+      "grad_norm": 0.45528423329483847,
+      "learning_rate": 3.585410063844186e-05,
+      "loss": 0.3993,
+      "step": 354
+    },
+    {
+      "epoch": 1.437974683544304,
+      "grad_norm": 0.4531170013362468,
+      "learning_rate": 3.581943644734846e-05,
+      "loss": 0.3998,
+      "step": 355
+    },
+    {
+      "epoch": 1.4420253164556962,
+      "grad_norm": 0.43254118756618454,
+      "learning_rate": 3.578464484868869e-05,
+      "loss": 0.3879,
+      "step": 356
+    },
+    {
+      "epoch": 1.4460759493670885,
+      "grad_norm": 0.462966060287851,
+      "learning_rate": 3.5749726122669316e-05,
+      "loss": 0.3872,
+      "step": 357
+    },
+    {
+      "epoch": 1.450126582278481,
+      "grad_norm": 0.49479938428544495,
+      "learning_rate": 3.5714680550520943e-05,
+      "loss": 0.3812,
+      "step": 358
+    },
+    {
+      "epoch": 1.4541772151898735,
+      "grad_norm": 0.41131248518175395,
+      "learning_rate": 3.5679508414495794e-05,
+      "loss": 0.3862,
+      "step": 359
+    },
+    {
+      "epoch": 1.4582278481012658,
+      "grad_norm": 0.40088472484440124,
+      "learning_rate": 3.564420999786543e-05,
+      "loss": 0.3873,
+      "step": 360
+    },
+    {
+      "epoch": 1.4622784810126581,
+      "grad_norm": 0.4419076229265041,
+      "learning_rate": 3.560878558491842e-05,
+      "loss": 0.3881,
+      "step": 361
+    },
+    {
+      "epoch": 1.4663291139240506,
+      "grad_norm": 0.3782785629216795,
+      "learning_rate": 3.5573235460958145e-05,
+      "loss": 0.3977,
+      "step": 362
+    },
+    {
+      "epoch": 1.4703797468354431,
+      "grad_norm": 0.3754304701333025,
+      "learning_rate": 3.553755991230039e-05,
+      "loss": 0.3946,
+      "step": 363
+    },
+    {
+      "epoch": 1.4744303797468354,
+      "grad_norm": 0.4219496824951547,
+      "learning_rate": 3.5501759226271144e-05,
+      "loss": 0.3855,
+      "step": 364
+    },
+    {
+      "epoch": 1.4784810126582277,
+      "grad_norm": 0.4237688470118234,
+      "learning_rate": 3.546583369120419e-05,
+      "loss": 0.3951,
+      "step": 365
+    },
+    {
+      "epoch": 1.4825316455696202,
+      "grad_norm": 0.4949327452205551,
+      "learning_rate": 3.5429783596438864e-05,
+      "loss": 0.3884,
+      "step": 366
+    },
+    {
+      "epoch": 1.4865822784810128,
+      "grad_norm": 0.41497001619903423,
+      "learning_rate": 3.539360923231766e-05,
+      "loss": 0.3961,
+      "step": 367
+    },
+    {
+      "epoch": 1.490632911392405,
+      "grad_norm": 0.3891344295753064,
+      "learning_rate": 3.535731089018394e-05,
+      "loss": 0.3904,
+      "step": 368
+    },
+    {
+      "epoch": 1.4946835443037974,
+      "grad_norm": 0.5369057071522283,
+      "learning_rate": 3.532088886237956e-05,
+      "loss": 0.3968,
+      "step": 369
+    },
+    {
+      "epoch": 1.4987341772151899,
+      "grad_norm": 0.4418549830304682,
+      "learning_rate": 3.528434344224253e-05,
+      "loss": 0.406,
+      "step": 370
+    },
+    {
+      "epoch": 1.5027848101265824,
+      "grad_norm": 0.32733633690475794,
+      "learning_rate": 3.524767492410464e-05,
+      "loss": 0.3772,
+      "step": 371
+    },
+    {
+      "epoch": 1.5068354430379747,
+      "grad_norm": 0.4128505789362775,
+      "learning_rate": 3.521088360328908e-05,
+      "loss": 0.3923,
+      "step": 372
+    },
+    {
+      "epoch": 1.510886075949367,
+      "grad_norm": 0.3826952924058261,
+      "learning_rate": 3.517396977610811e-05,
+      "loss": 0.4069,
+      "step": 373
+    },
+    {
+      "epoch": 1.5149367088607595,
+      "grad_norm": 0.43069549514916966,
+      "learning_rate": 3.5136933739860595e-05,
+      "loss": 0.4029,
+      "step": 374
+    },
+    {
+      "epoch": 1.518987341772152,
+      "grad_norm": 0.3222015587958947,
+      "learning_rate": 3.509977579282971e-05,
+      "loss": 0.3889,
+      "step": 375
+    },
+    {
+      "epoch": 1.5230379746835443,
+      "grad_norm": 0.39336776887655156,
+      "learning_rate": 3.5062496234280424e-05,
+      "loss": 0.3999,
+      "step": 376
+    },
+    {
+      "epoch": 1.5270886075949366,
+      "grad_norm": 0.3572909974827801,
+      "learning_rate": 3.502509536445719e-05,
+      "loss": 0.4112,
+      "step": 377
+    },
+    {
+      "epoch": 1.531139240506329,
+      "grad_norm": 0.3575133457450437,
+      "learning_rate": 3.498757348458147e-05,
+      "loss": 0.3917,
+      "step": 378
+    },
+    {
+      "epoch": 1.5351898734177216,
+      "grad_norm": 0.4239357375979974,
+      "learning_rate": 3.4949930896849324e-05,
+      "loss": 0.3968,
+      "step": 379
+    },
+    {
+      "epoch": 1.539240506329114,
+      "grad_norm": 0.3172158629589699,
+      "learning_rate": 3.491216790442899e-05,
+      "loss": 0.3991,
+      "step": 380
+    },
+    {
+      "epoch": 1.5432911392405062,
+      "grad_norm": 0.4134242983446708,
+      "learning_rate": 3.487428481145839e-05,
+      "loss": 0.3974,
+      "step": 381
+    },
+    {
+      "epoch": 1.5473417721518987,
+      "grad_norm": 0.37136408152156763,
+      "learning_rate": 3.483628192304278e-05,
+      "loss": 0.3833,
+      "step": 382
+    },
+    {
+      "epoch": 1.5513924050632912,
+      "grad_norm": 0.3074440168554623,
+      "learning_rate": 3.479815954525219e-05,
+      "loss": 0.3806,
+      "step": 383
+    },
+    {
+      "epoch": 1.5554430379746835,
+      "grad_norm": 0.37635919677487084,
+      "learning_rate": 3.475991798511899e-05,
+      "loss": 0.3857,
+      "step": 384
+    },
+    {
+      "epoch": 1.5594936708860758,
+      "grad_norm": 0.35729156009717794,
+      "learning_rate": 3.4721557550635464e-05,
+      "loss": 0.3959,
+      "step": 385
+    },
+    {
+      "epoch": 1.5635443037974683,
+      "grad_norm": 0.3773256729441813,
+      "learning_rate": 3.468307855075128e-05,
+      "loss": 0.3912,
+      "step": 386
+    },
+    {
+      "epoch": 1.5675949367088609,
+      "grad_norm": 0.32107533339623334,
+      "learning_rate": 3.4644481295371005e-05,
+      "loss": 0.3944,
+      "step": 387
+    },
+    {
+      "epoch": 1.5716455696202531,
+      "grad_norm": 0.3134503287235794,
+      "learning_rate": 3.460576609535163e-05,
+      "loss": 0.3915,
+      "step": 388
+    },
+    {
+      "epoch": 1.5756962025316454,
+      "grad_norm": 0.3477855016707046,
+      "learning_rate": 3.456693326250006e-05,
+      "loss": 0.3886,
+      "step": 389
+    },
+    {
+      "epoch": 1.579746835443038,
+      "grad_norm": 0.3171095344442211,
+      "learning_rate": 3.452798310957058e-05,
+      "loss": 0.3935,
+      "step": 390
+    },
+    {
+      "epoch": 1.5837974683544305,
+      "grad_norm": 0.38986051114949494,
+      "learning_rate": 3.4488915950262386e-05,
+      "loss": 0.3965,
+      "step": 391
+    },
+    {
+      "epoch": 1.5878481012658228,
+      "grad_norm": 0.37357739858250044,
+      "learning_rate": 3.4449732099216985e-05,
+      "loss": 0.4017,
+      "step": 392
+    },
+    {
+      "epoch": 1.591898734177215,
+      "grad_norm": 0.36686665416708264,
+      "learning_rate": 3.441043187201574e-05,
+      "loss": 0.3906,
+      "step": 393
+    },
+    {
+      "epoch": 1.5959493670886076,
+      "grad_norm": 0.3144000746958555,
+      "learning_rate": 3.437101558517728e-05,
+      "loss": 0.3919,
+      "step": 394
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.378567933973124,
+      "learning_rate": 3.433148355615496e-05,
+      "loss": 0.3932,
+      "step": 395
+    },
+    {
+      "epoch": 1.6040506329113924,
+      "grad_norm": 0.33708159432329543,
+      "learning_rate": 3.4291836103334294e-05,
+      "loss": 0.3977,
+      "step": 396
+    },
+    {
+      "epoch": 1.6081012658227847,
+      "grad_norm": 0.367320742911276,
+      "learning_rate": 3.425207354603043e-05,
+      "loss": 0.3889,
+      "step": 397
+    },
+    {
+      "epoch": 1.6121518987341772,
+      "grad_norm": 0.392432521878777,
+      "learning_rate": 3.421219620448553e-05,
+      "loss": 0.3905,
+      "step": 398
+    },
+    {
+      "epoch": 1.6162025316455697,
+      "grad_norm": 0.3286947227092628,
+      "learning_rate": 3.417220439986623e-05,
+      "loss": 0.3861,
+      "step": 399
+    },
+    {
+      "epoch": 1.620253164556962,
+      "grad_norm": 0.37978680959680966,
+      "learning_rate": 3.4132098454261024e-05,
+      "loss": 0.3981,
+      "step": 400
+    },
+    {
+      "epoch": 1.6243037974683543,
+      "grad_norm": 0.38303287328575897,
+      "learning_rate": 3.4091878690677676e-05,
+      "loss": 0.4032,
+      "step": 401
+    },
+    {
+      "epoch": 1.6283544303797468,
+      "grad_norm": 0.41179520986844215,
+      "learning_rate": 3.405154543304065e-05,
+      "loss": 0.3862,
+      "step": 402
+    },
+    {
+      "epoch": 1.6324050632911393,
+      "grad_norm": 0.3343212891421193,
+      "learning_rate": 3.401109900618843e-05,
+      "loss": 0.3771,
+      "step": 403
+    },
+    {
+      "epoch": 1.6364556962025316,
+      "grad_norm": 0.3452057585781342,
+      "learning_rate": 3.3970539735870996e-05,
+      "loss": 0.3945,
+      "step": 404
+    },
+    {
+      "epoch": 1.640506329113924,
+      "grad_norm": 0.32843136824170777,
+      "learning_rate": 3.392986794874714e-05,
+      "loss": 0.3822,
+      "step": 405
+    },
+    {
+      "epoch": 1.6445569620253164,
+      "grad_norm": 0.30503115708072204,
+      "learning_rate": 3.388908397238184e-05,
+      "loss": 0.3859,
+      "step": 406
+    },
+    {
+      "epoch": 1.648607594936709,
+      "grad_norm": 0.33326958355758224,
+      "learning_rate": 3.384818813524362e-05,
+      "loss": 0.3826,
+      "step": 407
+    },
+    {
+      "epoch": 1.6526582278481012,
+      "grad_norm": 0.3242960399196334,
+      "learning_rate": 3.380718076670195e-05,
+      "loss": 0.3884,
+      "step": 408
+    },
+    {
+      "epoch": 1.6567088607594935,
+      "grad_norm": 0.2938558211267508,
+      "learning_rate": 3.376606219702454e-05,
+      "loss": 0.3871,
+      "step": 409
+    },
+    {
+      "epoch": 1.660759493670886,
+      "grad_norm": 0.33549416776850904,
+      "learning_rate": 3.372483275737468e-05,
+      "loss": 0.3867,
+      "step": 410
+    },
+    {
+      "epoch": 1.6648101265822786,
+      "grad_norm": 0.27224849982673116,
+      "learning_rate": 3.368349277980861e-05,
+      "loss": 0.3817,
+      "step": 411
+    },
+    {
+      "epoch": 1.6688607594936709,
+      "grad_norm": 0.4226553101005686,
+      "learning_rate": 3.3642042597272844e-05,
+      "loss": 0.3835,
+      "step": 412
+    },
+    {
+      "epoch": 1.6729113924050631,
+      "grad_norm": 0.37628227271109255,
+      "learning_rate": 3.360048254360144e-05,
+      "loss": 0.3847,
+      "step": 413
+    },
+    {
+      "epoch": 1.6769620253164557,
+      "grad_norm": 0.3672933104606073,
+      "learning_rate": 3.355881295351336e-05,
+      "loss": 0.3781,
+      "step": 414
+    },
+    {
+      "epoch": 1.6810126582278482,
+      "grad_norm": 0.3745020649719186,
+      "learning_rate": 3.351703416260975e-05,
+      "loss": 0.4017,
+      "step": 415
+    },
+    {
+      "epoch": 1.6850632911392405,
+      "grad_norm": 0.3354249856065732,
+      "learning_rate": 3.347514650737126e-05,
+      "loss": 0.394,
+      "step": 416
+    },
+    {
+      "epoch": 1.6891139240506328,
+      "grad_norm": 0.33620281953567777,
+      "learning_rate": 3.3433150325155295e-05,
+      "loss": 0.391,
+      "step": 417
+    },
+    {
+      "epoch": 1.6931645569620253,
+      "grad_norm": 0.37314764766387576,
+      "learning_rate": 3.339104595419334e-05,
+      "loss": 0.3922,
+      "step": 418
+    },
+    {
+      "epoch": 1.6972151898734178,
+      "grad_norm": 0.4004379732486349,
+      "learning_rate": 3.3348833733588204e-05,
+      "loss": 0.3888,
+      "step": 419
+    },
+    {
+      "epoch": 1.70126582278481,
+      "grad_norm": 0.4316542034496058,
+      "learning_rate": 3.3306514003311305e-05,
+      "loss": 0.405,
+      "step": 420
+    },
+    {
+      "epoch": 1.7053164556962024,
+      "grad_norm": 0.34437913454306374,
+      "learning_rate": 3.326408710419996e-05,
+      "loss": 0.3891,
+      "step": 421
+    },
+    {
+      "epoch": 1.709367088607595,
+      "grad_norm": 0.4137436339659764,
+      "learning_rate": 3.322155337795454e-05,
+      "loss": 0.4006,
+      "step": 422
+    },
+    {
+      "epoch": 1.7134177215189874,
+      "grad_norm": 0.4736723484803527,
+      "learning_rate": 3.317891316713587e-05,
+      "loss": 0.3834,
+      "step": 423
+    },
+    {
+      "epoch": 1.7174683544303797,
+      "grad_norm": 0.4191699185442718,
+      "learning_rate": 3.313616681516231e-05,
+      "loss": 0.4033,
+      "step": 424
+    },
+    {
+      "epoch": 1.721518987341772,
+      "grad_norm": 0.3907786928235513,
+      "learning_rate": 3.309331466630713e-05,
+      "loss": 0.389,
+      "step": 425
+    },
+    {
+      "epoch": 1.7255696202531645,
+      "grad_norm": 0.49966173395205915,
+      "learning_rate": 3.305035706569563e-05,
+      "loss": 0.3933,
+      "step": 426
+    },
+    {
+      "epoch": 1.729620253164557,
+      "grad_norm": 0.3992013800642112,
+      "learning_rate": 3.3007294359302433e-05,
+      "loss": 0.3844,
+      "step": 427
+    },
+    {
+      "epoch": 1.7336708860759493,
+      "grad_norm": 0.47842084166903664,
+      "learning_rate": 3.296412689394864e-05,
+      "loss": 0.3977,
+      "step": 428
+    },
+    {
+      "epoch": 1.7377215189873416,
+      "grad_norm": 0.5629374407450002,
+      "learning_rate": 3.292085501729909e-05,
+      "loss": 0.3952,
+      "step": 429
+    },
+    {
+      "epoch": 1.7417721518987341,
+      "grad_norm": 0.33685438271297347,
+      "learning_rate": 3.2877479077859534e-05,
+      "loss": 0.3835,
+      "step": 430
+    },
+    {
+      "epoch": 1.7458227848101266,
+      "grad_norm": 0.37840280797971043,
+      "learning_rate": 3.283399942497381e-05,
+      "loss": 0.3932,
+      "step": 431
+    },
+    {
+      "epoch": 1.749873417721519,
+      "grad_norm": 0.39744193074890893,
+      "learning_rate": 3.279041640882108e-05,
+      "loss": 0.3756,
+      "step": 432
+    },
+    {
+      "epoch": 1.7539240506329112,
+      "grad_norm": 0.38917574102069297,
+      "learning_rate": 3.2746730380412964e-05,
+      "loss": 0.3873,
+      "step": 433
+    },
+    {
+      "epoch": 1.7579746835443038,
+      "grad_norm": 0.2965067325036941,
+      "learning_rate": 3.2702941691590726e-05,
+      "loss": 0.3873,
+      "step": 434
+    },
+    {
+      "epoch": 1.7620253164556963,
+      "grad_norm": 0.4171549893360577,
+      "learning_rate": 3.265905069502244e-05,
+      "loss": 0.395,
+      "step": 435
+    },
+    {
+      "epoch": 1.7660759493670886,
+      "grad_norm": 0.3537429225186739,
+      "learning_rate": 3.261505774420016e-05,
+      "loss": 0.4049,
+      "step": 436
+    },
+    {
+      "epoch": 1.7701265822784809,
+      "grad_norm": 0.34978341160987386,
+      "learning_rate": 3.257096319343707e-05,
+      "loss": 0.3757,
+      "step": 437
+    },
+    {
+      "epoch": 1.7741772151898734,
+      "grad_norm": 0.33592563108049805,
+      "learning_rate": 3.2526767397864614e-05,
+      "loss": 0.382,
+      "step": 438
+    },
+    {
+      "epoch": 1.7782278481012659,
+      "grad_norm": 0.3784576556073765,
+      "learning_rate": 3.248247071342966e-05,
+      "loss": 0.3865,
+      "step": 439
+    },
+    {
+      "epoch": 1.7822784810126582,
+      "grad_norm": 0.3539196070110637,
+      "learning_rate": 3.243807349689161e-05,
+      "loss": 0.389,
+      "step": 440
+    },
+    {
+      "epoch": 1.7863291139240505,
+      "grad_norm": 0.3251058407891909,
+      "learning_rate": 3.2393576105819544e-05,
+      "loss": 0.3806,
+      "step": 441
+    },
+    {
+      "epoch": 1.790379746835443,
+      "grad_norm": 0.31453790177027224,
+      "learning_rate": 3.2348978898589333e-05,
+      "loss": 0.3911,
+      "step": 442
+    },
+    {
+      "epoch": 1.7944303797468355,
+      "grad_norm": 0.28374391945905564,
+      "learning_rate": 3.230428223438075e-05,
+      "loss": 0.3872,
+      "step": 443
+    },
+    {
+      "epoch": 1.7984810126582278,
+      "grad_norm": 0.3225605455187976,
+      "learning_rate": 3.225948647317459e-05,
+      "loss": 0.4026,
+      "step": 444
+    },
+    {
+      "epoch": 1.80253164556962,
+      "grad_norm": 0.30465514970901214,
+      "learning_rate": 3.2214591975749745e-05,
+      "loss": 0.3919,
+      "step": 445
+    },
+    {
+      "epoch": 1.8065822784810126,
+      "grad_norm": 0.32098203513606355,
+      "learning_rate": 3.216959910368034e-05,
+      "loss": 0.391,
+      "step": 446
+    },
+    {
+      "epoch": 1.8106329113924051,
+      "grad_norm": 0.36746925709393086,
+      "learning_rate": 3.212450821933277e-05,
+      "loss": 0.3863,
+      "step": 447
+    },
+    {
+      "epoch": 1.8146835443037974,
+      "grad_norm": 0.28781719580040677,
+      "learning_rate": 3.207931968586281e-05,
+      "loss": 0.3891,
+      "step": 448
+    },
+    {
+      "epoch": 1.81873417721519,
+      "grad_norm": 0.33179306879111725,
+      "learning_rate": 3.203403386721272e-05,
+      "loss": 0.3649,
+      "step": 449
+    },
+    {
+      "epoch": 1.8227848101265822,
+      "grad_norm": 0.35681802288314035,
+      "learning_rate": 3.1988651128108245e-05,
+      "loss": 0.3825,
+      "step": 450
+    },
+    {
+      "epoch": 1.8268354430379747,
+      "grad_norm": 0.3055277147743676,
+      "learning_rate": 3.194317183405573e-05,
+      "loss": 0.3819,
+      "step": 451
+    },
+    {
+      "epoch": 1.8308860759493673,
+      "grad_norm": 0.339251060533746,
+      "learning_rate": 3.189759635133914e-05,
+      "loss": 0.3916,
+      "step": 452
+    },
+    {
+      "epoch": 1.8349367088607595,
+      "grad_norm": 0.38804153832091026,
+      "learning_rate": 3.185192504701718e-05,
+      "loss": 0.3877,
+      "step": 453
+    },
+    {
+      "epoch": 1.8389873417721518,
+      "grad_norm": 0.2887661818958886,
+      "learning_rate": 3.1806158288920234e-05,
+      "loss": 0.4015,
+      "step": 454
+    },
+    {
+      "epoch": 1.8430379746835444,
+      "grad_norm": 0.32622294568928434,
+      "learning_rate": 3.1760296445647477e-05,
+      "loss": 0.3869,
+      "step": 455
+    },
+    {
+      "epoch": 1.8470886075949369,
+      "grad_norm": 0.33276947063791035,
+      "learning_rate": 3.1714339886563896e-05,
+      "loss": 0.3812,
+      "step": 456
+    },
+    {
+      "epoch": 1.8511392405063292,
+      "grad_norm": 0.32960070236673655,
+      "learning_rate": 3.166828898179731e-05,
+      "loss": 0.3948,
+      "step": 457
+    },
+    {
+      "epoch": 1.8551898734177215,
+      "grad_norm": 0.322029632229707,
+      "learning_rate": 3.162214410223536e-05,
+      "loss": 0.3852,
+      "step": 458
+    },
+    {
+      "epoch": 1.859240506329114,
+      "grad_norm": 0.3310387175908109,
+      "learning_rate": 3.157590561952257e-05,
+      "loss": 0.3939,
+      "step": 459
+    },
+    {
+      "epoch": 1.8632911392405065,
+      "grad_norm": 0.36890429271771463,
+      "learning_rate": 3.152957390605732e-05,
+      "loss": 0.3851,
+      "step": 460
+    },
+    {
+      "epoch": 1.8673417721518988,
+      "grad_norm": 0.32662772230543996,
+      "learning_rate": 3.148314933498886e-05,
+      "loss": 0.3857,
+      "step": 461
+    },
+    {
+      "epoch": 1.871392405063291,
+      "grad_norm": 0.3389531058013578,
+      "learning_rate": 3.143663228021431e-05,
+      "loss": 0.3841,
+      "step": 462
+    },
+    {
+      "epoch": 1.8754430379746836,
+      "grad_norm": 0.3384211205964113,
+      "learning_rate": 3.1390023116375624e-05,
+      "loss": 0.3764,
+      "step": 463
+    },
+    {
+      "epoch": 1.879493670886076,
+      "grad_norm": 0.3829318014823403,
+      "learning_rate": 3.134332221885661e-05,
+      "loss": 0.3908,
+      "step": 464
+    },
+    {
+      "epoch": 1.8835443037974684,
+      "grad_norm": 0.36041143079641613,
+      "learning_rate": 3.129652996377987e-05,
+      "loss": 0.3906,
+      "step": 465
+    },
+    {
+      "epoch": 1.8875949367088607,
+      "grad_norm": 0.3879161991912513,
+      "learning_rate": 3.12496467280038e-05,
+      "loss": 0.3754,
+      "step": 466
+    },
+    {
+      "epoch": 1.8916455696202532,
+      "grad_norm": 0.3492339178126296,
+      "learning_rate": 3.120267288911952e-05,
+      "loss": 0.3898,
+      "step": 467
+    },
+    {
+      "epoch": 1.8956962025316457,
+      "grad_norm": 0.36302393600810434,
+      "learning_rate": 3.11556088254479e-05,
+      "loss": 0.3839,
+      "step": 468
+    },
+    {
+      "epoch": 1.899746835443038,
+      "grad_norm": 0.3177524901287015,
+      "learning_rate": 3.11084549160364e-05,
+      "loss": 0.3809,
+      "step": 469
+    },
+    {
+      "epoch": 1.9037974683544303,
+      "grad_norm": 0.33895535355886575,
+      "learning_rate": 3.106121154065615e-05,
+      "loss": 0.3912,
+      "step": 470
+    },
+    {
+      "epoch": 1.9078481012658228,
+      "grad_norm": 0.38797636081472986,
+      "learning_rate": 3.1013879079798805e-05,
+      "loss": 0.389,
+      "step": 471
+    },
+    {
+      "epoch": 1.9118987341772153,
+      "grad_norm": 0.37010595733153906,
+      "learning_rate": 3.096645791467348e-05,
+      "loss": 0.3926,
+      "step": 472
+    },
+    {
+      "epoch": 1.9159493670886076,
+      "grad_norm": 0.33656588398869663,
+      "learning_rate": 3.091894842720373e-05,
+      "loss": 0.4024,
+      "step": 473
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.36130584743167243,
+      "learning_rate": 3.0871351000024425e-05,
+      "loss": 0.3989,
+      "step": 474
+    },
+    {
+      "epoch": 1.9240506329113924,
+      "grad_norm": 0.3982406389844217,
+      "learning_rate": 3.0823666016478716e-05,
+      "loss": 0.3969,
+      "step": 475
+    },
+    {
+      "epoch": 1.928101265822785,
+      "grad_norm": 0.3861971188473323,
+      "learning_rate": 3.0775893860614896e-05,
+      "loss": 0.3956,
+      "step": 476
+    },
+    {
+      "epoch": 1.9321518987341773,
+      "grad_norm": 0.31855864313338383,
+      "learning_rate": 3.0728034917183336e-05,
+      "loss": 0.3992,
+      "step": 477
+    },
+    {
+      "epoch": 1.9362025316455695,
+      "grad_norm": 0.39743253656460126,
+      "learning_rate": 3.06800895716334e-05,
+      "loss": 0.3864,
+      "step": 478
+    },
+    {
+      "epoch": 1.940253164556962,
+      "grad_norm": 0.3685798944798544,
+      "learning_rate": 3.063205821011029e-05,
+      "loss": 0.398,
+      "step": 479
+    },
+    {
+      "epoch": 1.9443037974683546,
+      "grad_norm": 0.365123768766173,
+      "learning_rate": 3.0583941219452016e-05,
+      "loss": 0.407,
+      "step": 480
+    },
+    {
+      "epoch": 1.9483544303797469,
+      "grad_norm": 0.4974807150397006,
+      "learning_rate": 3.053573898718618e-05,
+      "loss": 0.4059,
+      "step": 481
+    },
+    {
+      "epoch": 1.9524050632911392,
+      "grad_norm": 0.3810903016092579,
+      "learning_rate": 3.0487451901526956e-05,
+      "loss": 0.389,
+      "step": 482
+    },
+    {
+      "epoch": 1.9564556962025317,
+      "grad_norm": 0.35099937983283536,
+      "learning_rate": 3.0439080351371875e-05,
+      "loss": 0.3785,
+      "step": 483
+    },
+    {
+      "epoch": 1.9605063291139242,
+      "grad_norm": 0.4067980582219428,
+      "learning_rate": 3.0390624726298764e-05,
+      "loss": 0.3881,
+      "step": 484
+    },
+    {
+      "epoch": 1.9645569620253165,
+      "grad_norm": 0.29921430997414966,
+      "learning_rate": 3.034208541656255e-05,
+      "loss": 0.3836,
+      "step": 485
+    },
+    {
+      "epoch": 1.9686075949367088,
+      "grad_norm": 0.39919296297458734,
+      "learning_rate": 3.029346281309218e-05,
+      "loss": 0.3787,
+      "step": 486
+    },
+    {
+      "epoch": 1.9726582278481013,
+      "grad_norm": 0.31766087506064133,
+      "learning_rate": 3.0244757307487415e-05,
+      "loss": 0.3873,
+      "step": 487
+    },
+    {
+      "epoch": 1.9767088607594938,
+      "grad_norm": 0.34005673077629617,
+      "learning_rate": 3.019596929201569e-05,
+      "loss": 0.3913,
+      "step": 488
+    },
+    {
+      "epoch": 1.980759493670886,
+      "grad_norm": 0.38014616689099634,
+      "learning_rate": 3.0147099159608985e-05,
+      "loss": 0.3878,
+      "step": 489
+    },
+    {
+      "epoch": 1.9848101265822784,
+      "grad_norm": 0.35279861201378376,
+      "learning_rate": 3.0098147303860616e-05,
+      "loss": 0.3843,
+      "step": 490
+    },
+    {
+      "epoch": 1.988860759493671,
+      "grad_norm": 0.32964799163914915,
+      "learning_rate": 3.0049114119022117e-05,
+      "loss": 0.3903,
+      "step": 491
+    },
+    {
+      "epoch": 1.9929113924050634,
+      "grad_norm": 0.47293593750748913,
+      "learning_rate": 3.0000000000000004e-05,
+      "loss": 0.4063,
+      "step": 492
+    },
+    {
+      "epoch": 1.9969620253164557,
+      "grad_norm": 0.2994380552037168,
+      "learning_rate": 2.995080534235264e-05,
+      "loss": 0.3783,
+      "step": 493
+    },
+    {
+      "epoch": 2.001012658227848,
+      "grad_norm": 0.40139599536710713,
+      "learning_rate": 2.9901530542287044e-05,
+      "loss": 0.3702,
+      "step": 494
+    },
+    {
+      "epoch": 2.0050632911392405,
+      "grad_norm": 0.3462983194841245,
+      "learning_rate": 2.9852175996655676e-05,
+      "loss": 0.3385,
+      "step": 495
+    },
+    {
+      "epoch": 2.009113924050633,
+      "grad_norm": 0.49239514921585714,
+      "learning_rate": 2.980274210295326e-05,
+      "loss": 0.3341,
+      "step": 496
+    },
+    {
+      "epoch": 2.013164556962025,
+      "grad_norm": 0.4061349280505586,
+      "learning_rate": 2.9753229259313578e-05,
+      "loss": 0.3239,
+      "step": 497
+    },
+    {
+      "epoch": 2.0172151898734176,
+      "grad_norm": 0.40165466185864174,
+      "learning_rate": 2.9703637864506274e-05,
+      "loss": 0.3375,
+      "step": 498
+    },
+    {
+      "epoch": 2.02126582278481,
+      "grad_norm": 0.40269766883992,
+      "learning_rate": 2.965396831793362e-05,
+      "loss": 0.3334,
+      "step": 499
+    },
+    {
+      "epoch": 2.0253164556962027,
+      "grad_norm": 0.42099870717692744,
+      "learning_rate": 2.9604221019627316e-05,
+      "loss": 0.3249,
+      "step": 500
+    },
+    {
+      "epoch": 2.0293670886075947,
+      "grad_norm": 0.4785889412573864,
+      "learning_rate": 2.955439637024526e-05,
+      "loss": 0.3397,
+      "step": 501
+    },
+    {
+      "epoch": 2.0334177215189873,
+      "grad_norm": 0.41336942431048906,
+      "learning_rate": 2.9504494771068334e-05,
+      "loss": 0.3309,
+      "step": 502
+    },
+    {
+      "epoch": 2.0374683544303798,
+      "grad_norm": 0.34683929707171995,
+      "learning_rate": 2.9454516623997156e-05,
+      "loss": 0.329,
+      "step": 503
+    },
+    {
+      "epoch": 2.0415189873417723,
+      "grad_norm": 0.47706454649544777,
+      "learning_rate": 2.9404462331548847e-05,
+      "loss": 0.3323,
+      "step": 504
+    },
+    {
+      "epoch": 2.0455696202531644,
+      "grad_norm": 0.34152256462988434,
+      "learning_rate": 2.93543322968538e-05,
+      "loss": 0.3385,
+      "step": 505
+    },
+    {
+      "epoch": 2.049620253164557,
+      "grad_norm": 0.4275592767121899,
+      "learning_rate": 2.9304126923652428e-05,
+      "loss": 0.3367,
+      "step": 506
+    },
+    {
+      "epoch": 2.0536708860759494,
+      "grad_norm": 0.4022043966916375,
+      "learning_rate": 2.9253846616291896e-05,
+      "loss": 0.3415,
+      "step": 507
+    },
+    {
+      "epoch": 2.057721518987342,
+      "grad_norm": 0.4372336240202637,
+      "learning_rate": 2.9203491779722896e-05,
+      "loss": 0.3336,
+      "step": 508
+    },
+    {
+      "epoch": 2.061772151898734,
+      "grad_norm": 0.32949428773902995,
+      "learning_rate": 2.9153062819496357e-05,
+      "loss": 0.3403,
+      "step": 509
+    },
+    {
+      "epoch": 2.0658227848101265,
+      "grad_norm": 0.4432902664443495,
+      "learning_rate": 2.9102560141760178e-05,
+      "loss": 0.3389,
+      "step": 510
+    },
+    {
+      "epoch": 2.069873417721519,
+      "grad_norm": 0.37055115327309146,
+      "learning_rate": 2.9051984153256004e-05,
+      "loss": 0.3407,
+      "step": 511
+    },
+    {
+      "epoch": 2.0739240506329115,
+      "grad_norm": 0.29672731907301264,
+      "learning_rate": 2.900133526131588e-05,
+      "loss": 0.3436,
+      "step": 512
+    },
+    {
+      "epoch": 2.0779746835443036,
+      "grad_norm": 0.3970805539688809,
+      "learning_rate": 2.8950613873859025e-05,
+      "loss": 0.3504,
+      "step": 513
+    },
+    {
+      "epoch": 2.082025316455696,
+      "grad_norm": 0.28550747805396903,
+      "learning_rate": 2.8899820399388515e-05,
+      "loss": 0.3344,
+      "step": 514
+    },
+    {
+      "epoch": 2.0860759493670886,
+      "grad_norm": 0.3528870029023396,
+      "learning_rate": 2.8848955246988012e-05,
+      "loss": 0.337,
+      "step": 515
+    },
+    {
+      "epoch": 2.090126582278481,
+      "grad_norm": 0.30003341808774936,
+      "learning_rate": 2.879801882631847e-05,
+      "loss": 0.3316,
+      "step": 516
+    },
+    {
+      "epoch": 2.094177215189873,
+      "grad_norm": 0.3723905703909138,
+      "learning_rate": 2.8747011547614808e-05,
+      "loss": 0.3427,
+      "step": 517
+    },
+    {
+      "epoch": 2.0982278481012657,
+      "grad_norm": 0.3271953855719539,
+      "learning_rate": 2.8695933821682635e-05,
+      "loss": 0.3381,
+      "step": 518
+    },
+    {
+      "epoch": 2.1022784810126582,
+      "grad_norm": 0.2779472221661318,
+      "learning_rate": 2.864478605989494e-05,
+      "loss": 0.336,
+      "step": 519
+    },
+    {
+      "epoch": 2.1063291139240508,
+      "grad_norm": 0.32439753733279963,
+      "learning_rate": 2.8593568674188765e-05,
+      "loss": 0.3411,
+      "step": 520
+    },
+    {
+      "epoch": 2.110379746835443,
+      "grad_norm": 0.3328654608927568,
+      "learning_rate": 2.8542282077061892e-05,
+      "loss": 0.327,
+      "step": 521
+    },
+    {
+      "epoch": 2.1144303797468353,
+      "grad_norm": 0.32298119306536005,
+      "learning_rate": 2.8490926681569523e-05,
+      "loss": 0.3319,
+      "step": 522
+    },
+    {
+      "epoch": 2.118481012658228,
+      "grad_norm": 0.3062886716712015,
+      "learning_rate": 2.8439502901320956e-05,
+      "loss": 0.3337,
+      "step": 523
+    },
+    {
+      "epoch": 2.1225316455696204,
+      "grad_norm": 0.35833014342418973,
+      "learning_rate": 2.8388011150476237e-05,
+      "loss": 0.3438,
+      "step": 524
+    },
+    {
+      "epoch": 2.1265822784810124,
+      "grad_norm": 0.34587478568115365,
+      "learning_rate": 2.8336451843742866e-05,
+      "loss": 0.343,
+      "step": 525
+    },
+    {
+      "epoch": 2.130632911392405,
+      "grad_norm": 0.2924332581053405,
+      "learning_rate": 2.8284825396372387e-05,
+      "loss": 0.3404,
+      "step": 526
+    },
+    {
+      "epoch": 2.1346835443037975,
+      "grad_norm": 0.295756865404568,
+      "learning_rate": 2.8233132224157132e-05,
+      "loss": 0.3338,
+      "step": 527
+    },
+    {
+      "epoch": 2.13873417721519,
+      "grad_norm": 0.2667430338808131,
+      "learning_rate": 2.8181372743426805e-05,
+      "loss": 0.3333,
+      "step": 528
+    },
+    {
+      "epoch": 2.1427848101265825,
+      "grad_norm": 0.3122511697442158,
+      "learning_rate": 2.8129547371045128e-05,
+      "loss": 0.3307,
+      "step": 529
+    },
+    {
+      "epoch": 2.1468354430379746,
+      "grad_norm": 0.3205337881563142,
+      "learning_rate": 2.8077656524406534e-05,
+      "loss": 0.3356,
+      "step": 530
+    },
+    {
+      "epoch": 2.150886075949367,
+      "grad_norm": 0.28654513999919146,
+      "learning_rate": 2.802570062143278e-05,
+      "loss": 0.3481,
+      "step": 531
+    },
+    {
+      "epoch": 2.1549367088607596,
+      "grad_norm": 0.28655562991050215,
+      "learning_rate": 2.7973680080569555e-05,
+      "loss": 0.3383,
+      "step": 532
+    },
+    {
+      "epoch": 2.1589873417721517,
+      "grad_norm": 0.2889811494360431,
+      "learning_rate": 2.792159532078314e-05,
+      "loss": 0.3426,
+      "step": 533
+    },
+    {
+      "epoch": 2.163037974683544,
+      "grad_norm": 0.2981519650697927,
+      "learning_rate": 2.7869446761557033e-05,
+      "loss": 0.3303,
+      "step": 534
+    },
+    {
+      "epoch": 2.1670886075949367,
+      "grad_norm": 0.2806460478248113,
+      "learning_rate": 2.781723482288857e-05,
+      "loss": 0.3378,
+      "step": 535
+    },
+    {
+      "epoch": 2.1711392405063292,
+      "grad_norm": 0.27306522492311325,
+      "learning_rate": 2.7764959925285517e-05,
+      "loss": 0.3458,
+      "step": 536
+    },
+    {
+      "epoch": 2.1751898734177217,
+      "grad_norm": 0.2680084006268731,
+      "learning_rate": 2.771262248976272e-05,
+      "loss": 0.3251,
+      "step": 537
+    },
+    {
+      "epoch": 2.179240506329114,
+      "grad_norm": 0.29230561280385176,
+      "learning_rate": 2.7660222937838677e-05,
+      "loss": 0.3372,
+      "step": 538
+    },
+    {
+      "epoch": 2.1832911392405063,
+      "grad_norm": 0.28136462733822876,
+      "learning_rate": 2.7607761691532186e-05,
+      "loss": 0.3456,
+      "step": 539
+    },
+    {
+      "epoch": 2.187341772151899,
+      "grad_norm": 0.28240864423107864,
+      "learning_rate": 2.7555239173358916e-05,
+      "loss": 0.3293,
+      "step": 540
+    },
+    {
+      "epoch": 2.191392405063291,
+      "grad_norm": 0.2710200235497235,
+      "learning_rate": 2.7502655806328e-05,
+      "loss": 0.3239,
+      "step": 541
+    },
+    {
+      "epoch": 2.1954430379746834,
+      "grad_norm": 0.3192797303539684,
+      "learning_rate": 2.7450012013938648e-05,
+      "loss": 0.3409,
+      "step": 542
+    },
+    {
+      "epoch": 2.199493670886076,
+      "grad_norm": 0.33842374372876405,
+      "learning_rate": 2.739730822017673e-05,
+      "loss": 0.326,
+      "step": 543
+    },
+    {
+      "epoch": 2.2035443037974685,
+      "grad_norm": 0.2748625260669312,
+      "learning_rate": 2.7344544849511355e-05,
+      "loss": 0.3309,
+      "step": 544
+    },
+    {
+      "epoch": 2.207594936708861,
+      "grad_norm": 0.3721191952606732,
+      "learning_rate": 2.7291722326891456e-05,
+      "loss": 0.3322,
+      "step": 545
+    },
+    {
+      "epoch": 2.211645569620253,
+      "grad_norm": 0.3495827428275391,
+      "learning_rate": 2.723884107774236e-05,
+      "loss": 0.337,
+      "step": 546
+    },
+    {
+      "epoch": 2.2156962025316456,
+      "grad_norm": 0.2764986635932342,
+      "learning_rate": 2.718590152796239e-05,
+      "loss": 0.351,
+      "step": 547
+    },
+    {
+      "epoch": 2.219746835443038,
+      "grad_norm": 0.4564264176204414,
+      "learning_rate": 2.71329041039194e-05,
+      "loss": 0.3381,
+      "step": 548
+    },
+    {
+      "epoch": 2.22379746835443,
+      "grad_norm": 0.34545074244535084,
+      "learning_rate": 2.7079849232447357e-05,
+      "loss": 0.3388,
+      "step": 549
+    },
+    {
+      "epoch": 2.2278481012658227,
+      "grad_norm": 0.2933779446615015,
+      "learning_rate": 2.7026737340842895e-05,
+      "loss": 0.3367,
+      "step": 550
+    },
+    {
+      "epoch": 2.231898734177215,
+      "grad_norm": 0.359216584355082,
+      "learning_rate": 2.697356885686189e-05,
+      "loss": 0.3414,
+      "step": 551
+    },
+    {
+      "epoch": 2.2359493670886077,
+      "grad_norm": 0.32001628092449186,
+      "learning_rate": 2.6920344208716014e-05,
+      "loss": 0.3462,
+      "step": 552
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 0.3039976058570966,
+      "learning_rate": 2.6867063825069252e-05,
+      "loss": 0.3316,
+      "step": 553
+    },
+    {
+      "epoch": 2.2440506329113923,
+      "grad_norm": 0.3267917340443827,
+      "learning_rate": 2.6813728135034494e-05,
+      "loss": 0.338,
+      "step": 554
+    },
+    {
+      "epoch": 2.248101265822785,
+      "grad_norm": 0.3021052085277106,
+      "learning_rate": 2.6760337568170056e-05,
+      "loss": 0.3439,
+      "step": 555
+    },
+    {
+      "epoch": 2.2521518987341773,
+      "grad_norm": 0.3812429359662877,
+      "learning_rate": 2.6706892554476226e-05,
+      "loss": 0.3385,
+      "step": 556
+    },
+    {
+      "epoch": 2.2562025316455694,
+      "grad_norm": 0.2783653866105728,
+      "learning_rate": 2.6653393524391795e-05,
+      "loss": 0.3368,
+      "step": 557
+    },
+    {
+      "epoch": 2.260253164556962,
+      "grad_norm": 0.32538018302620003,
+      "learning_rate": 2.6599840908790592e-05,
+      "loss": 0.3397,
+      "step": 558
+    },
+    {
+      "epoch": 2.2643037974683544,
+      "grad_norm": 0.310806731818039,
+      "learning_rate": 2.6546235138978028e-05,
+      "loss": 0.3327,
+      "step": 559
+    },
+    {
+      "epoch": 2.268354430379747,
+      "grad_norm": 0.3185911005470215,
+      "learning_rate": 2.6492576646687597e-05,
+      "loss": 0.3517,
+      "step": 560
+    },
+    {
+      "epoch": 2.2724050632911394,
+      "grad_norm": 0.3548514016008944,
+      "learning_rate": 2.6438865864077425e-05,
+      "loss": 0.3358,
+      "step": 561
+    },
+    {
+      "epoch": 2.2764556962025315,
+      "grad_norm": 0.2677134142282245,
+      "learning_rate": 2.6385103223726766e-05,
+      "loss": 0.3383,
+      "step": 562
+    },
+    {
+      "epoch": 2.280506329113924,
+      "grad_norm": 0.3617818083309844,
+      "learning_rate": 2.6331289158632537e-05,
+      "loss": 0.3321,
+      "step": 563
+    },
+    {
+      "epoch": 2.2845569620253166,
+      "grad_norm": 0.3239127169072989,
+      "learning_rate": 2.6277424102205817e-05,
+      "loss": 0.3472,
+      "step": 564
+    },
+    {
+      "epoch": 2.2886075949367086,
+      "grad_norm": 0.27312888656574963,
+      "learning_rate": 2.6223508488268374e-05,
+      "loss": 0.3417,
+      "step": 565
+    },
+    {
+      "epoch": 2.292658227848101,
+      "grad_norm": 0.2862672347899802,
+      "learning_rate": 2.6169542751049148e-05,
+      "loss": 0.3366,
+      "step": 566
+    },
+    {
+      "epoch": 2.2967088607594937,
+      "grad_norm": 0.30777139534199743,
+      "learning_rate": 2.6115527325180754e-05,
+      "loss": 0.3384,
+      "step": 567
+    },
+    {
+      "epoch": 2.300759493670886,
+      "grad_norm": 0.2798749226745631,
+      "learning_rate": 2.606146264569603e-05,
+      "loss": 0.3467,
+      "step": 568
+    },
+    {
+      "epoch": 2.3048101265822787,
+      "grad_norm": 0.3198051825354587,
+      "learning_rate": 2.6007349148024447e-05,
+      "loss": 0.3435,
+      "step": 569
+    },
+    {
+      "epoch": 2.3088607594936708,
+      "grad_norm": 0.2798493129421067,
+      "learning_rate": 2.5953187267988694e-05,
+      "loss": 0.3365,
+      "step": 570
+    },
+    {
+      "epoch": 2.3129113924050633,
+      "grad_norm": 0.29764097847703286,
+      "learning_rate": 2.5898977441801097e-05,
+      "loss": 0.3316,
+      "step": 571
+    },
+    {
+      "epoch": 2.316962025316456,
+      "grad_norm": 0.2904666109462977,
+      "learning_rate": 2.584472010606015e-05,
+      "loss": 0.3358,
+      "step": 572
+    },
+    {
+      "epoch": 2.321012658227848,
+      "grad_norm": 0.27408781109535624,
+      "learning_rate": 2.5790415697746976e-05,
+      "loss": 0.3458,
+      "step": 573
+    },
+    {
+      "epoch": 2.3250632911392404,
+      "grad_norm": 0.3224708872307838,
+      "learning_rate": 2.5736064654221808e-05,
+      "loss": 0.3352,
+      "step": 574
+    },
+    {
+      "epoch": 2.329113924050633,
+      "grad_norm": 0.2552201636824158,
+      "learning_rate": 2.568166741322048e-05,
+      "loss": 0.342,
+      "step": 575
+    },
+    {
+      "epoch": 2.3331645569620254,
+      "grad_norm": 0.2766767593461259,
+      "learning_rate": 2.56272244128509e-05,
+      "loss": 0.3352,
+      "step": 576
+    },
+    {
+      "epoch": 2.337215189873418,
+      "grad_norm": 0.24530650479696034,
+      "learning_rate": 2.55727360915895e-05,
+      "loss": 0.3391,
+      "step": 577
+    },
+    {
+      "epoch": 2.34126582278481,
+      "grad_norm": 0.2602890612933093,
+      "learning_rate": 2.5518202888277734e-05,
+      "loss": 0.3282,
+      "step": 578
+    },
+    {
+      "epoch": 2.3453164556962025,
+      "grad_norm": 0.2737989858440301,
+      "learning_rate": 2.5463625242118523e-05,
+      "loss": 0.3506,
+      "step": 579
+    },
+    {
+      "epoch": 2.349367088607595,
+      "grad_norm": 0.25615478194831826,
+      "learning_rate": 2.5409003592672723e-05,
+      "loss": 0.3319,
+      "step": 580
+    },
+    {
+      "epoch": 2.353417721518987,
+      "grad_norm": 0.2803957108454961,
+      "learning_rate": 2.535433837985559e-05,
+      "loss": 0.33,
+      "step": 581
+    },
+    {
+      "epoch": 2.3574683544303796,
+      "grad_norm": 0.24753843568899034,
+      "learning_rate": 2.529963004393324e-05,
+      "loss": 0.3425,
+      "step": 582
+    },
+    {
+      "epoch": 2.361518987341772,
+      "grad_norm": 0.3243928036223699,
+      "learning_rate": 2.524487902551908e-05,
+      "loss": 0.3402,
+      "step": 583
+    },
+    {
+      "epoch": 2.3655696202531646,
+      "grad_norm": 0.3286263642832382,
+      "learning_rate": 2.519008576557029e-05,
+      "loss": 0.3439,
+      "step": 584
+    },
+    {
+      "epoch": 2.369620253164557,
+      "grad_norm": 0.316848745686048,
+      "learning_rate": 2.5135250705384254e-05,
+      "loss": 0.3455,
+      "step": 585
+    },
+    {
+      "epoch": 2.3736708860759492,
+      "grad_norm": 0.3010537969695793,
+      "learning_rate": 2.5080374286595007e-05,
+      "loss": 0.3295,
+      "step": 586
+    },
+    {
+      "epoch": 2.3777215189873417,
+      "grad_norm": 0.30443602153705956,
+      "learning_rate": 2.5025456951169677e-05,
+      "loss": 0.34,
+      "step": 587
+    },
+    {
+      "epoch": 2.3817721518987343,
+      "grad_norm": 0.27197630084762325,
+      "learning_rate": 2.4970499141404942e-05,
+      "loss": 0.3385,
+      "step": 588
+    },
+    {
+      "epoch": 2.3858227848101268,
+      "grad_norm": 0.302709335099644,
+      "learning_rate": 2.491550129992345e-05,
+      "loss": 0.3406,
+      "step": 589
+    },
+    {
+      "epoch": 2.389873417721519,
+      "grad_norm": 0.2693756461042658,
+      "learning_rate": 2.486046386967024e-05,
+      "loss": 0.3349,
+      "step": 590
+    },
+    {
+      "epoch": 2.3939240506329114,
+      "grad_norm": 0.29669938452301253,
+      "learning_rate": 2.4805387293909214e-05,
+      "loss": 0.3338,
+      "step": 591
+    },
+    {
+      "epoch": 2.397974683544304,
+      "grad_norm": 0.24292251769442577,
+      "learning_rate": 2.4750272016219552e-05,
+      "loss": 0.343,
+      "step": 592
+    },
+    {
+      "epoch": 2.4020253164556964,
+      "grad_norm": 0.36204568884489774,
+      "learning_rate": 2.4695118480492114e-05,
+      "loss": 0.3366,
+      "step": 593
+    },
+    {
+      "epoch": 2.4060759493670885,
+      "grad_norm": 0.25730494072454113,
+      "learning_rate": 2.4639927130925898e-05,
+      "loss": 0.348,
+      "step": 594
+    },
+    {
+      "epoch": 2.410126582278481,
+      "grad_norm": 0.31024310824734136,
+      "learning_rate": 2.458469841202444e-05,
+      "loss": 0.3409,
+      "step": 595
+    },
+    {
+      "epoch": 2.4141772151898735,
+      "grad_norm": 0.27210499953456907,
+      "learning_rate": 2.452943276859226e-05,
+      "loss": 0.3414,
+      "step": 596
+    },
+    {
+      "epoch": 2.418227848101266,
+      "grad_norm": 0.30396653583798416,
+      "learning_rate": 2.447413064573125e-05,
+      "loss": 0.3416,
+      "step": 597
+    },
+    {
+      "epoch": 2.422278481012658,
+      "grad_norm": 0.2764823561870851,
+      "learning_rate": 2.4418792488837095e-05,
+      "loss": 0.33,
+      "step": 598
+    },
+    {
+      "epoch": 2.4263291139240506,
+      "grad_norm": 0.2825010727549695,
+      "learning_rate": 2.4363418743595713e-05,
+      "loss": 0.3533,
+      "step": 599
+    },
+    {
+      "epoch": 2.430379746835443,
+      "grad_norm": 0.2694142371145247,
+      "learning_rate": 2.430800985597963e-05,
+      "loss": 0.3295,
+      "step": 600
+    },
+    {
+      "epoch": 2.4344303797468356,
+      "grad_norm": 0.2574976431562949,
+      "learning_rate": 2.4252566272244415e-05,
+      "loss": 0.3401,
+      "step": 601
+    },
+    {
+      "epoch": 2.4384810126582277,
+      "grad_norm": 0.2745673703602058,
+      "learning_rate": 2.4197088438925063e-05,
+      "loss": 0.3382,
+      "step": 602
+    },
+    {
+      "epoch": 2.44253164556962,
+      "grad_norm": 0.29928293483164264,
+      "learning_rate": 2.4141576802832417e-05,
+      "loss": 0.3272,
+      "step": 603
+    },
+    {
+      "epoch": 2.4465822784810127,
+      "grad_norm": 0.24174254700527195,
+      "learning_rate": 2.408603181104957e-05,
+      "loss": 0.345,
+      "step": 604
+    },
+    {
+      "epoch": 2.4506329113924052,
+      "grad_norm": 0.2926600573513234,
+      "learning_rate": 2.4030453910928245e-05,
+      "loss": 0.3509,
+      "step": 605
+    },
+    {
+      "epoch": 2.4546835443037973,
+      "grad_norm": 0.30316324832918956,
+      "learning_rate": 2.397484355008521e-05,
+      "loss": 0.3395,
+      "step": 606
+    },
+    {
+      "epoch": 2.45873417721519,
+      "grad_norm": 0.2787361870632683,
+      "learning_rate": 2.3919201176398662e-05,
+      "loss": 0.3378,
+      "step": 607
+    },
+    {
+      "epoch": 2.4627848101265823,
+      "grad_norm": 0.3001193218197824,
+      "learning_rate": 2.3863527238004633e-05,
+      "loss": 0.3323,
+      "step": 608
+    },
+    {
+      "epoch": 2.466835443037975,
+      "grad_norm": 0.34011667547342694,
+      "learning_rate": 2.380782218329337e-05,
+      "loss": 0.3442,
+      "step": 609
+    },
+    {
+      "epoch": 2.470886075949367,
+      "grad_norm": 0.25021596665255613,
+      "learning_rate": 2.3752086460905725e-05,
+      "loss": 0.351,
+      "step": 610
+    },
+    {
+      "epoch": 2.4749367088607594,
+      "grad_norm": 0.3479056326913941,
+      "learning_rate": 2.3696320519729544e-05,
+      "loss": 0.3362,
+      "step": 611
+    },
+    {
+      "epoch": 2.478987341772152,
+      "grad_norm": 0.27485830299882336,
+      "learning_rate": 2.3640524808896045e-05,
+      "loss": 0.3408,
+      "step": 612
+    },
+    {
+      "epoch": 2.4830379746835445,
+      "grad_norm": 0.32614972314196006,
+      "learning_rate": 2.3584699777776222e-05,
+      "loss": 0.3361,
+      "step": 613
+    },
+    {
+      "epoch": 2.4870886075949366,
+      "grad_norm": 0.29410357983003543,
+      "learning_rate": 2.3528845875977195e-05,
+      "loss": 0.3353,
+      "step": 614
+    },
+    {
+      "epoch": 2.491139240506329,
+      "grad_norm": 0.29940404554581634,
+      "learning_rate": 2.3472963553338614e-05,
+      "loss": 0.3343,
+      "step": 615
+    },
+    {
+      "epoch": 2.4951898734177216,
+      "grad_norm": 0.26112655194660317,
+      "learning_rate": 2.341705325992901e-05,
+      "loss": 0.3422,
+      "step": 616
+    },
+    {
+      "epoch": 2.499240506329114,
+      "grad_norm": 0.3119134823599642,
+      "learning_rate": 2.336111544604222e-05,
+      "loss": 0.333,
+      "step": 617
+    },
+    {
+      "epoch": 2.503291139240506,
+      "grad_norm": 0.29324116293762553,
+      "learning_rate": 2.33051505621937e-05,
+      "loss": 0.3396,
+      "step": 618
+    },
+    {
+      "epoch": 2.5073417721518987,
+      "grad_norm": 0.27156334859212683,
+      "learning_rate": 2.324915905911693e-05,
+      "loss": 0.3318,
+      "step": 619
+    },
+    {
+      "epoch": 2.511392405063291,
+      "grad_norm": 0.3229463044188078,
+      "learning_rate": 2.319314138775977e-05,
+      "loss": 0.3408,
+      "step": 620
+    },
+    {
+      "epoch": 2.5154430379746833,
+      "grad_norm": 0.30524357748857384,
+      "learning_rate": 2.3137097999280856e-05,
+      "loss": 0.3271,
+      "step": 621
+    },
+    {
+      "epoch": 2.519493670886076,
+      "grad_norm": 0.2812495759188102,
+      "learning_rate": 2.308102934504593e-05,
+      "loss": 0.3372,
+      "step": 622
+    },
+    {
+      "epoch": 2.5235443037974683,
+      "grad_norm": 0.31829649592308124,
+      "learning_rate": 2.3024935876624222e-05,
+      "loss": 0.3451,
+      "step": 623
+    },
+    {
+      "epoch": 2.527594936708861,
+      "grad_norm": 0.2687551742211805,
+      "learning_rate": 2.2968818045784813e-05,
+      "loss": 0.3326,
+      "step": 624
+    },
+    {
+      "epoch": 2.5316455696202533,
+      "grad_norm": 0.3028323740044882,
+      "learning_rate": 2.2912676304493006e-05,
+      "loss": 0.3413,
+      "step": 625
+    },
+    {
+      "epoch": 2.5356962025316454,
+      "grad_norm": 0.3055849136664519,
+      "learning_rate": 2.2856511104906668e-05,
+      "loss": 0.349,
+      "step": 626
+    },
+    {
+      "epoch": 2.539746835443038,
+      "grad_norm": 0.35621919670930813,
+      "learning_rate": 2.2800322899372586e-05,
+      "loss": 0.342,
+      "step": 627
+    },
+    {
+      "epoch": 2.5437974683544304,
+      "grad_norm": 0.29329492903927684,
+      "learning_rate": 2.2744112140422844e-05,
+      "loss": 0.3418,
+      "step": 628
+    },
+    {
+      "epoch": 2.547848101265823,
+      "grad_norm": 0.3300507018173765,
+      "learning_rate": 2.2687879280771177e-05,
+      "loss": 0.3308,
+      "step": 629
+    },
+    {
+      "epoch": 2.5518987341772155,
+      "grad_norm": 0.27560475420439445,
+      "learning_rate": 2.26316247733093e-05,
+      "loss": 0.3433,
+      "step": 630
+    },
+    {
+      "epoch": 2.5559493670886075,
+      "grad_norm": 0.34417431948781896,
+      "learning_rate": 2.257534907110328e-05,
+      "loss": 0.3391,
+      "step": 631
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 0.2607661474724853,
+      "learning_rate": 2.2519052627389882e-05,
+      "loss": 0.336,
+      "step": 632
+    },
+    {
+      "epoch": 2.5640506329113926,
+      "grad_norm": 0.30619681095820483,
+      "learning_rate": 2.246273589557294e-05,
+      "loss": 0.3554,
+      "step": 633
+    },
+    {
+      "epoch": 2.5681012658227846,
+      "grad_norm": 0.2722608749012109,
+      "learning_rate": 2.240639932921966e-05,
+      "loss": 0.3426,
+      "step": 634
+    },
+    {
+      "epoch": 2.572151898734177,
+      "grad_norm": 0.26830149318617935,
+      "learning_rate": 2.2350043382056995e-05,
+      "loss": 0.3423,
+      "step": 635
+    },
+    {
+      "epoch": 2.5762025316455697,
+      "grad_norm": 0.25844651807902025,
+      "learning_rate": 2.2293668507968015e-05,
+      "loss": 0.3289,
+      "step": 636
+    },
+    {
+      "epoch": 2.580253164556962,
+      "grad_norm": 0.279007800429231,
+      "learning_rate": 2.2237275160988186e-05,
+      "loss": 0.333,
+      "step": 637
+    },
+    {
+      "epoch": 2.5843037974683547,
+      "grad_norm": 0.29994086924478264,
+      "learning_rate": 2.2180863795301787e-05,
+      "loss": 0.3334,
+      "step": 638
+    },
+    {
+      "epoch": 2.5883544303797468,
+      "grad_norm": 0.2568834037084406,
+      "learning_rate": 2.212443486523819e-05,
+      "loss": 0.3361,
+      "step": 639
+    },
+    {
+      "epoch": 2.5924050632911393,
+      "grad_norm": 0.3230173542713307,
+      "learning_rate": 2.2067988825268243e-05,
+      "loss": 0.3426,
+      "step": 640
+    },
+    {
+      "epoch": 2.596455696202532,
+      "grad_norm": 0.2441388518770024,
+      "learning_rate": 2.2011526130000596e-05,
+      "loss": 0.3395,
+      "step": 641
+    },
+    {
+      "epoch": 2.600506329113924,
+      "grad_norm": 0.2983750912562916,
+      "learning_rate": 2.1955047234178038e-05,
+      "loss": 0.3462,
+      "step": 642
+    },
+    {
+      "epoch": 2.6045569620253164,
+      "grad_norm": 0.23701581808537073,
+      "learning_rate": 2.1898552592673825e-05,
+      "loss": 0.3455,
+      "step": 643
+    },
+    {
+      "epoch": 2.608607594936709,
+      "grad_norm": 0.281290854787871,
+      "learning_rate": 2.184204266048803e-05,
+      "loss": 0.3389,
+      "step": 644
+    },
+    {
+      "epoch": 2.6126582278481014,
+      "grad_norm": 0.26077973649596836,
+      "learning_rate": 2.1785517892743887e-05,
+      "loss": 0.3361,
+      "step": 645
+    },
+    {
+      "epoch": 2.616708860759494,
+      "grad_norm": 0.28039978698563034,
+      "learning_rate": 2.17289787446841e-05,
+      "loss": 0.3445,
+      "step": 646
+    },
+    {
+      "epoch": 2.620759493670886,
+      "grad_norm": 0.2658942435317639,
+      "learning_rate": 2.1672425671667198e-05,
+      "loss": 0.3339,
+      "step": 647
+    },
+    {
+      "epoch": 2.6248101265822785,
+      "grad_norm": 0.25116588358131436,
+      "learning_rate": 2.161585912916385e-05,
+      "loss": 0.3381,
+      "step": 648
+    },
+    {
+      "epoch": 2.628860759493671,
+      "grad_norm": 0.24844206594292026,
+      "learning_rate": 2.1559279572753214e-05,
+      "loss": 0.3345,
+      "step": 649
+    },
+    {
+      "epoch": 2.632911392405063,
+      "grad_norm": 0.23848208813039332,
+      "learning_rate": 2.1502687458119268e-05,
+      "loss": 0.3303,
+      "step": 650
+    },
+    {
+      "epoch": 2.6369620253164556,
+      "grad_norm": 0.231256496385481,
+      "learning_rate": 2.1446083241047116e-05,
+      "loss": 0.3364,
+      "step": 651
+    },
+    {
+      "epoch": 2.641012658227848,
+      "grad_norm": 0.2608190463865017,
+      "learning_rate": 2.1389467377419333e-05,
+      "loss": 0.3414,
+      "step": 652
+    },
+    {
+      "epoch": 2.6450632911392407,
+      "grad_norm": 0.2716137155718485,
+      "learning_rate": 2.133284032321232e-05,
+      "loss": 0.3499,
+      "step": 653
+    },
+    {
+      "epoch": 2.649113924050633,
+      "grad_norm": 0.2572787877999268,
+      "learning_rate": 2.1276202534492566e-05,
+      "loss": 0.3416,
+      "step": 654
+    },
+    {
+      "epoch": 2.6531645569620252,
+      "grad_norm": 0.2694700807200122,
+      "learning_rate": 2.121955446741306e-05,
+      "loss": 0.3342,
+      "step": 655
+    },
+    {
+      "epoch": 2.6572151898734178,
+      "grad_norm": 0.29463888058635335,
+      "learning_rate": 2.1162896578209517e-05,
+      "loss": 0.3394,
+      "step": 656
+    },
+    {
+      "epoch": 2.6612658227848103,
+      "grad_norm": 0.2582608468325674,
+      "learning_rate": 2.1106229323196813e-05,
+      "loss": 0.3357,
+      "step": 657
+    },
+    {
+      "epoch": 2.6653164556962023,
+      "grad_norm": 0.2980882307689955,
+      "learning_rate": 2.1049553158765214e-05,
+      "loss": 0.3331,
+      "step": 658
+    },
+    {
+      "epoch": 2.669367088607595,
+      "grad_norm": 0.26381170191146036,
+      "learning_rate": 2.0992868541376764e-05,
+      "loss": 0.3418,
+      "step": 659
+    },
+    {
+      "epoch": 2.6734177215189874,
+      "grad_norm": 0.28738467160275366,
+      "learning_rate": 2.093617592756158e-05,
+      "loss": 0.3387,
+      "step": 660
+    },
+    {
+      "epoch": 2.67746835443038,
+      "grad_norm": 0.27934304049273534,
+      "learning_rate": 2.0879475773914167e-05,
+      "loss": 0.3484,
+      "step": 661
+    },
+    {
+      "epoch": 2.6815189873417724,
+      "grad_norm": 0.3368104275639727,
+      "learning_rate": 2.082276853708978e-05,
+      "loss": 0.3397,
+      "step": 662
+    },
+    {
+      "epoch": 2.6855696202531645,
+      "grad_norm": 0.2794212213313693,
+      "learning_rate": 2.076605467380071e-05,
+      "loss": 0.3406,
+      "step": 663
+    },
+    {
+      "epoch": 2.689620253164557,
+      "grad_norm": 0.3490346583454692,
+      "learning_rate": 2.0709334640812613e-05,
+      "loss": 0.3375,
+      "step": 664
+    },
+    {
+      "epoch": 2.6936708860759495,
+      "grad_norm": 0.2916431039198685,
+      "learning_rate": 2.0652608894940824e-05,
+      "loss": 0.339,
+      "step": 665
+    },
+    {
+      "epoch": 2.6977215189873416,
+      "grad_norm": 0.32519791914787033,
+      "learning_rate": 2.0595877893046722e-05,
+      "loss": 0.3394,
+      "step": 666
+    },
+    {
+      "epoch": 2.701772151898734,
+      "grad_norm": 0.2784660125261922,
+      "learning_rate": 2.0539142092033985e-05,
+      "loss": 0.3405,
+      "step": 667
+    },
+    {
+      "epoch": 2.7058227848101266,
+      "grad_norm": 0.283822686495062,
+      "learning_rate": 2.048240194884496e-05,
+      "loss": 0.3359,
+      "step": 668
+    },
+    {
+      "epoch": 2.709873417721519,
+      "grad_norm": 0.29539376929405176,
+      "learning_rate": 2.042565792045695e-05,
+      "loss": 0.3409,
+      "step": 669
+    },
+    {
+      "epoch": 2.7139240506329116,
+      "grad_norm": 0.32249614921428665,
+      "learning_rate": 2.036891046387857e-05,
+      "loss": 0.3449,
+      "step": 670
+    },
+    {
+      "epoch": 2.7179746835443037,
+      "grad_norm": 0.2962355360211899,
+      "learning_rate": 2.0312160036146036e-05,
+      "loss": 0.3364,
+      "step": 671
+    },
+    {
+      "epoch": 2.7220253164556962,
+      "grad_norm": 0.27064933540680786,
+      "learning_rate": 2.025540709431948e-05,
+      "loss": 0.3297,
+      "step": 672
+    },
+    {
+      "epoch": 2.7260759493670887,
+      "grad_norm": 0.28998384904186497,
+      "learning_rate": 2.0198652095479298e-05,
+      "loss": 0.336,
+      "step": 673
+    },
+    {
+      "epoch": 2.730126582278481,
+      "grad_norm": 0.2677764306239443,
+      "learning_rate": 2.014189549672245e-05,
+      "loss": 0.3445,
+      "step": 674
+    },
+    {
+      "epoch": 2.7341772151898733,
+      "grad_norm": 0.2640198192058773,
+      "learning_rate": 2.0085137755158776e-05,
+      "loss": 0.3368,
+      "step": 675
+    },
+    {
+      "epoch": 2.738227848101266,
+      "grad_norm": 0.2725199607672248,
+      "learning_rate": 2.0028379327907327e-05,
+      "loss": 0.3464,
+      "step": 676
+    },
+    {
+      "epoch": 2.7422784810126584,
+      "grad_norm": 0.24082059704059708,
+      "learning_rate": 1.9971620672092676e-05,
+      "loss": 0.3309,
+      "step": 677
+    },
+    {
+      "epoch": 2.746329113924051,
+      "grad_norm": 0.24337450898247912,
+      "learning_rate": 1.991486224484123e-05,
+      "loss": 0.3453,
+      "step": 678
+    },
+    {
+      "epoch": 2.750379746835443,
+      "grad_norm": 0.23745878024675257,
+      "learning_rate": 1.985810450327756e-05,
+      "loss": 0.3379,
+      "step": 679
+    },
+    {
+      "epoch": 2.7544303797468355,
+      "grad_norm": 0.2986765588412128,
+      "learning_rate": 1.9801347904520706e-05,
+      "loss": 0.3429,
+      "step": 680
+    },
+    {
+      "epoch": 2.758481012658228,
+      "grad_norm": 0.26983617781398794,
+      "learning_rate": 1.974459290568053e-05,
+      "loss": 0.3332,
+      "step": 681
+    },
+    {
+      "epoch": 2.76253164556962,
+      "grad_norm": 0.28087520296812113,
+      "learning_rate": 1.968783996385397e-05,
+      "loss": 0.3354,
+      "step": 682
+    },
+    {
+      "epoch": 2.7665822784810126,
+      "grad_norm": 0.28094444362051596,
+      "learning_rate": 1.963108953612143e-05,
+      "loss": 0.3368,
+      "step": 683
+    },
+    {
+      "epoch": 2.770632911392405,
+      "grad_norm": 0.26928633788855794,
+      "learning_rate": 1.9574342079543056e-05,
+      "loss": 0.3377,
+      "step": 684
+    },
+    {
+      "epoch": 2.7746835443037976,
+      "grad_norm": 0.28265341483375733,
+      "learning_rate": 1.9517598051155046e-05,
+      "loss": 0.3345,
+      "step": 685
+    },
+    {
+      "epoch": 2.77873417721519,
+      "grad_norm": 0.2751234367201597,
+      "learning_rate": 1.9460857907966025e-05,
+      "loss": 0.3519,
+      "step": 686
+    },
+    {
+      "epoch": 2.782784810126582,
+      "grad_norm": 0.27429447282026326,
+      "learning_rate": 1.9404122106953285e-05,
+      "loss": 0.335,
+      "step": 687
+    },
+    {
+      "epoch": 2.7868354430379747,
+      "grad_norm": 0.2633383845549836,
+      "learning_rate": 1.9347391105059176e-05,
+      "loss": 0.336,
+      "step": 688
+    },
+    {
+      "epoch": 2.790886075949367,
+      "grad_norm": 0.2454279790049147,
+      "learning_rate": 1.92906653591874e-05,
+      "loss": 0.34,
+      "step": 689
+    },
+    {
+      "epoch": 2.7949367088607593,
+      "grad_norm": 0.26640593372065074,
+      "learning_rate": 1.9233945326199295e-05,
+      "loss": 0.3404,
+      "step": 690
+    },
+    {
+      "epoch": 2.798987341772152,
+      "grad_norm": 0.23017579309885458,
+      "learning_rate": 1.917723146291022e-05,
+      "loss": 0.3238,
+      "step": 691
+    },
+    {
+      "epoch": 2.8030379746835443,
+      "grad_norm": 0.26920026286186716,
+      "learning_rate": 1.912052422608584e-05,
+      "loss": 0.3323,
+      "step": 692
+    },
+    {
+      "epoch": 2.807088607594937,
+      "grad_norm": 0.22425066519556303,
+      "learning_rate": 1.9063824072438428e-05,
+      "loss": 0.3347,
+      "step": 693
+    },
+    {
+      "epoch": 2.8111392405063294,
+      "grad_norm": 0.28373282051758286,
+      "learning_rate": 1.9007131458623246e-05,
+      "loss": 0.3428,
+      "step": 694
+    },
+    {
+      "epoch": 2.8151898734177214,
+      "grad_norm": 0.23596604949378816,
+      "learning_rate": 1.895044684123479e-05,
+      "loss": 0.3423,
+      "step": 695
+    },
+    {
+      "epoch": 2.819240506329114,
+      "grad_norm": 0.27662559921035695,
+      "learning_rate": 1.8893770676803194e-05,
+      "loss": 0.3357,
+      "step": 696
+    },
+    {
+      "epoch": 2.8232911392405065,
+      "grad_norm": 0.2403925618203336,
+      "learning_rate": 1.8837103421790486e-05,
+      "loss": 0.3335,
+      "step": 697
+    },
+    {
+      "epoch": 2.8273417721518985,
+      "grad_norm": 0.2802989258962142,
+      "learning_rate": 1.8780445532586952e-05,
+      "loss": 0.3443,
+      "step": 698
+    },
+    {
+      "epoch": 2.831392405063291,
+      "grad_norm": 0.2417955332858613,
+      "learning_rate": 1.872379746550743e-05,
+      "loss": 0.3379,
+      "step": 699
+    },
+    {
+      "epoch": 2.8354430379746836,
+      "grad_norm": 0.28411461395004634,
+      "learning_rate": 1.866715967678769e-05,
+      "loss": 0.3308,
+      "step": 700
+    },
+    {
+      "epoch": 2.839493670886076,
+      "grad_norm": 0.25870417278689084,
+      "learning_rate": 1.861053262258067e-05,
+      "loss": 0.3351,
+      "step": 701
+    },
+    {
+      "epoch": 2.8435443037974686,
+      "grad_norm": 0.3140060410866941,
+      "learning_rate": 1.8553916758952897e-05,
+      "loss": 0.349,
+      "step": 702
+    },
+    {
+      "epoch": 2.8475949367088607,
+      "grad_norm": 0.24764532303627787,
+      "learning_rate": 1.8497312541880735e-05,
+      "loss": 0.3334,
+      "step": 703
+    },
+    {
+      "epoch": 2.851645569620253,
+      "grad_norm": 0.26935771596668423,
+      "learning_rate": 1.8440720427246786e-05,
+      "loss": 0.3413,
+      "step": 704
+    },
+    {
+      "epoch": 2.8556962025316457,
+      "grad_norm": 0.24626613227234728,
+      "learning_rate": 1.8384140870836157e-05,
+      "loss": 0.3404,
+      "step": 705
+    },
+    {
+      "epoch": 2.8597468354430378,
+      "grad_norm": 0.23967942945938292,
+      "learning_rate": 1.8327574328332806e-05,
+      "loss": 0.341,
+      "step": 706
+    },
+    {
+      "epoch": 2.8637974683544303,
+      "grad_norm": 0.23860690703230908,
+      "learning_rate": 1.8271021255315906e-05,
+      "loss": 0.3358,
+      "step": 707
+    },
+    {
+      "epoch": 2.867848101265823,
+      "grad_norm": 0.26571246357580325,
+      "learning_rate": 1.8214482107256117e-05,
+      "loss": 0.3458,
+      "step": 708
+    },
+    {
+      "epoch": 2.8718987341772153,
+      "grad_norm": 0.2452982665266372,
+      "learning_rate": 1.8157957339511968e-05,
+      "loss": 0.3395,
+      "step": 709
+    },
+    {
+      "epoch": 2.875949367088608,
+      "grad_norm": 0.24973461385109538,
+      "learning_rate": 1.8101447407326182e-05,
+      "loss": 0.3337,
+      "step": 710
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 0.2567081230984209,
+      "learning_rate": 1.8044952765821966e-05,
+      "loss": 0.3352,
+      "step": 711
+    },
+    {
+      "epoch": 2.8840506329113924,
+      "grad_norm": 0.24789623649666598,
+      "learning_rate": 1.7988473869999407e-05,
+      "loss": 0.3379,
+      "step": 712
+    },
+    {
+      "epoch": 2.888101265822785,
+      "grad_norm": 0.2870159837859169,
+      "learning_rate": 1.7932011174731764e-05,
+      "loss": 0.3423,
+      "step": 713
+    },
+    {
+      "epoch": 2.892151898734177,
+      "grad_norm": 0.26306810075567333,
+      "learning_rate": 1.7875565134761817e-05,
+      "loss": 0.3449,
+      "step": 714
+    },
+    {
+      "epoch": 2.8962025316455695,
+      "grad_norm": 0.25875040593847953,
+      "learning_rate": 1.7819136204698226e-05,
+      "loss": 0.3422,
+      "step": 715
+    },
+    {
+      "epoch": 2.900253164556962,
+      "grad_norm": 0.2596251868666502,
+      "learning_rate": 1.776272483901182e-05,
+      "loss": 0.334,
+      "step": 716
+    },
+    {
+      "epoch": 2.9043037974683545,
+      "grad_norm": 0.2810596343318251,
+      "learning_rate": 1.7706331492031995e-05,
+      "loss": 0.3379,
+      "step": 717
+    },
+    {
+      "epoch": 2.908354430379747,
+      "grad_norm": 0.2361445294906021,
+      "learning_rate": 1.764995661794301e-05,
+      "loss": 0.3503,
+      "step": 718
+    },
+    {
+      "epoch": 2.912405063291139,
+      "grad_norm": 0.2535883738190092,
+      "learning_rate": 1.759360067078035e-05,
+      "loss": 0.3366,
+      "step": 719
+    },
+    {
+      "epoch": 2.9164556962025316,
+      "grad_norm": 0.2681428427890256,
+      "learning_rate": 1.7537264104427064e-05,
+      "loss": 0.336,
+      "step": 720
+    },
+    {
+      "epoch": 2.920506329113924,
+      "grad_norm": 0.2643742505882568,
+      "learning_rate": 1.748094737261012e-05,
+      "loss": 0.3337,
+      "step": 721
+    },
+    {
+      "epoch": 2.9245569620253162,
+      "grad_norm": 0.26424545117546,
+      "learning_rate": 1.7424650928896726e-05,
+      "loss": 0.3443,
+      "step": 722
+    },
+    {
+      "epoch": 2.9286075949367087,
+      "grad_norm": 0.2476670828388002,
+      "learning_rate": 1.7368375226690712e-05,
+      "loss": 0.3329,
+      "step": 723
+    },
+    {
+      "epoch": 2.9326582278481013,
+      "grad_norm": 0.2711123634055774,
+      "learning_rate": 1.731212071922883e-05,
+      "loss": 0.3381,
+      "step": 724
+    },
+    {
+      "epoch": 2.9367088607594938,
+      "grad_norm": 0.21999327816079553,
+      "learning_rate": 1.7255887859577156e-05,
+      "loss": 0.3402,
+      "step": 725
+    },
+    {
+      "epoch": 2.9407594936708863,
+      "grad_norm": 0.24218534994131333,
+      "learning_rate": 1.7199677100627427e-05,
+      "loss": 0.3357,
+      "step": 726
+    },
+    {
+      "epoch": 2.9448101265822784,
+      "grad_norm": 0.2334442171142268,
+      "learning_rate": 1.7143488895093343e-05,
+      "loss": 0.3289,
+      "step": 727
+    },
+    {
+      "epoch": 2.948860759493671,
+      "grad_norm": 0.2336503058194854,
+      "learning_rate": 1.7087323695506994e-05,
+      "loss": 0.3415,
+      "step": 728
+    },
+    {
+      "epoch": 2.9529113924050634,
+      "grad_norm": 0.24724704791580987,
+      "learning_rate": 1.7031181954215194e-05,
+      "loss": 0.3338,
+      "step": 729
+    },
+    {
+      "epoch": 2.9569620253164555,
+      "grad_norm": 0.22889965122101946,
+      "learning_rate": 1.6975064123375788e-05,
+      "loss": 0.3349,
+      "step": 730
+    },
+    {
+      "epoch": 2.961012658227848,
+      "grad_norm": 0.250714315761411,
+      "learning_rate": 1.6918970654954084e-05,
+      "loss": 0.3393,
+      "step": 731
+    },
+    {
+      "epoch": 2.9650632911392405,
+      "grad_norm": 0.22504027204355165,
+      "learning_rate": 1.686290200071915e-05,
+      "loss": 0.3402,
+      "step": 732
+    },
+    {
+      "epoch": 2.969113924050633,
+      "grad_norm": 0.24361310187384624,
+      "learning_rate": 1.6806858612240234e-05,
+      "loss": 0.3446,
+      "step": 733
+    },
+    {
+      "epoch": 2.9731645569620255,
+      "grad_norm": 0.23021658587678506,
+      "learning_rate": 1.6750840940883078e-05,
+      "loss": 0.3282,
+      "step": 734
+    },
+    {
+      "epoch": 2.9772151898734176,
+      "grad_norm": 0.23468671765401133,
+      "learning_rate": 1.6694849437806305e-05,
+      "loss": 0.3298,
+      "step": 735
+    },
+    {
+      "epoch": 2.98126582278481,
+      "grad_norm": 0.23805420958323242,
+      "learning_rate": 1.663888455395778e-05,
+      "loss": 0.325,
+      "step": 736
+    },
+    {
+      "epoch": 2.9853164556962026,
+      "grad_norm": 0.25703663804353466,
+      "learning_rate": 1.6582946740070995e-05,
+      "loss": 0.3395,
+      "step": 737
+    },
+    {
+      "epoch": 2.9893670886075947,
+      "grad_norm": 0.24167764354249896,
+      "learning_rate": 1.6527036446661396e-05,
+      "loss": 0.3297,
+      "step": 738
+    },
+    {
+      "epoch": 2.993417721518987,
+      "grad_norm": 0.2785025213279554,
+      "learning_rate": 1.6471154124022818e-05,
+      "loss": 0.3386,
+      "step": 739
+    },
+    {
+      "epoch": 2.9974683544303797,
+      "grad_norm": 0.23108538535316767,
+      "learning_rate": 1.6415300222223788e-05,
+      "loss": 0.3232,
+      "step": 740
+    },
+    {
+      "epoch": 3.0015189873417722,
+      "grad_norm": 0.29139163333347856,
+      "learning_rate": 1.6359475191103958e-05,
+      "loss": 0.3216,
+      "step": 741
+    },
+    {
+      "epoch": 3.0055696202531648,
+      "grad_norm": 0.27178407853350495,
+      "learning_rate": 1.6303679480270466e-05,
+      "loss": 0.2925,
+      "step": 742
+    },
+    {
+      "epoch": 3.009620253164557,
+      "grad_norm": 0.6467308701810187,
+      "learning_rate": 1.624791353909428e-05,
+      "loss": 0.2877,
+      "step": 743
+    },
+    {
+      "epoch": 3.0136708860759494,
+      "grad_norm": 0.43162597727279867,
+      "learning_rate": 1.619217781670663e-05,
+      "loss": 0.287,
+      "step": 744
+    },
+    {
+      "epoch": 3.017721518987342,
+      "grad_norm": 0.36569748798116525,
+      "learning_rate": 1.6136472761995373e-05,
+      "loss": 0.2884,
+      "step": 745
+    },
+    {
+      "epoch": 3.0217721518987344,
+      "grad_norm": 0.36161946883252327,
+      "learning_rate": 1.608079882360134e-05,
+      "loss": 0.294,
+      "step": 746
+    },
+    {
+      "epoch": 3.0258227848101265,
+      "grad_norm": 0.3201941983542823,
+      "learning_rate": 1.60251564499148e-05,
+      "loss": 0.2792,
+      "step": 747
+    },
+    {
+      "epoch": 3.029873417721519,
+      "grad_norm": 0.3640930461242147,
+      "learning_rate": 1.596954608907176e-05,
+      "loss": 0.2892,
+      "step": 748
+    },
+    {
+      "epoch": 3.0339240506329115,
+      "grad_norm": 0.35772329716944057,
+      "learning_rate": 1.591396818895043e-05,
+      "loss": 0.2873,
+      "step": 749
+    },
+    {
+      "epoch": 3.037974683544304,
+      "grad_norm": 0.33620275615478656,
+      "learning_rate": 1.585842319716759e-05,
+      "loss": 0.2778,
+      "step": 750
+    },
+    {
+      "epoch": 3.042025316455696,
+      "grad_norm": 0.3296089499518773,
+      "learning_rate": 1.5802911561074944e-05,
+      "loss": 0.2821,
+      "step": 751
+    },
+    {
+      "epoch": 3.0460759493670886,
+      "grad_norm": 0.3011618465598423,
+      "learning_rate": 1.5747433727755595e-05,
+      "loss": 0.2914,
+      "step": 752
+    },
+    {
+      "epoch": 3.050126582278481,
+      "grad_norm": 0.3359336491654183,
+      "learning_rate": 1.5691990144020376e-05,
+      "loss": 0.2881,
+      "step": 753
+    },
+    {
+      "epoch": 3.0541772151898736,
+      "grad_norm": 0.2899772463878424,
+      "learning_rate": 1.5636581256404297e-05,
+      "loss": 0.2805,
+      "step": 754
+    },
+    {
+      "epoch": 3.0582278481012657,
+      "grad_norm": 0.32992933847716505,
+      "learning_rate": 1.558120751116291e-05,
+      "loss": 0.2847,
+      "step": 755
+    },
+    {
+      "epoch": 3.062278481012658,
+      "grad_norm": 0.2708213775888268,
+      "learning_rate": 1.552586935426876e-05,
+      "loss": 0.2863,
+      "step": 756
+    },
+    {
+      "epoch": 3.0663291139240507,
+      "grad_norm": 0.31590064590721223,
+      "learning_rate": 1.547056723140774e-05,
+      "loss": 0.2807,
+      "step": 757
+    },
+    {
+      "epoch": 3.0703797468354432,
+      "grad_norm": 0.27069128474758775,
+      "learning_rate": 1.5415301587975565e-05,
+      "loss": 0.2839,
+      "step": 758
+    },
+    {
+      "epoch": 3.0744303797468353,
+      "grad_norm": 0.27327795145546424,
+      "learning_rate": 1.536007286907411e-05,
+      "loss": 0.2923,
+      "step": 759
+    },
+    {
+      "epoch": 3.078481012658228,
+      "grad_norm": 0.2773333809819723,
+      "learning_rate": 1.5304881519507896e-05,
+      "loss": 0.2912,
+      "step": 760
+    },
+    {
+      "epoch": 3.0825316455696203,
+      "grad_norm": 0.2907474550811488,
+      "learning_rate": 1.5249727983780453e-05,
+      "loss": 0.2843,
+      "step": 761
+    },
+    {
+      "epoch": 3.086582278481013,
+      "grad_norm": 0.2870132594098572,
+      "learning_rate": 1.5194612706090786e-05,
+      "loss": 0.2891,
+      "step": 762
+    },
+    {
+      "epoch": 3.090632911392405,
+      "grad_norm": 0.2664999164418968,
+      "learning_rate": 1.5139536130329771e-05,
+      "loss": 0.2962,
+      "step": 763
+    },
+    {
+      "epoch": 3.0946835443037974,
+      "grad_norm": 0.2707291408319488,
+      "learning_rate": 1.508449870007656e-05,
+      "loss": 0.2911,
+      "step": 764
+    },
+    {
+      "epoch": 3.09873417721519,
+      "grad_norm": 0.2563692463823429,
+      "learning_rate": 1.5029500858595056e-05,
+      "loss": 0.2893,
+      "step": 765
+    },
+    {
+      "epoch": 3.1027848101265825,
+      "grad_norm": 0.27967294790162284,
+      "learning_rate": 1.4974543048830328e-05,
+      "loss": 0.2848,
+      "step": 766
+    },
+    {
+      "epoch": 3.1068354430379745,
+      "grad_norm": 0.23105133891619267,
+      "learning_rate": 1.4919625713405e-05,
+      "loss": 0.2934,
+      "step": 767
+    },
+    {
+      "epoch": 3.110886075949367,
+      "grad_norm": 0.2666880630769176,
+      "learning_rate": 1.4864749294615756e-05,
+      "loss": 0.2865,
+      "step": 768
+    },
+    {
+      "epoch": 3.1149367088607596,
+      "grad_norm": 0.23679364664257702,
+      "learning_rate": 1.4809914234429716e-05,
+      "loss": 0.2867,
+      "step": 769
+    },
+    {
+      "epoch": 3.118987341772152,
+      "grad_norm": 0.2297862922950535,
+      "learning_rate": 1.4755120974480923e-05,
+      "loss": 0.2803,
+      "step": 770
+    },
+    {
+      "epoch": 3.123037974683544,
+      "grad_norm": 0.27317464987133044,
+      "learning_rate": 1.4700369956066771e-05,
+      "loss": 0.2863,
+      "step": 771
+    },
+    {
+      "epoch": 3.1270886075949367,
+      "grad_norm": 0.2627951678901216,
+      "learning_rate": 1.4645661620144413e-05,
+      "loss": 0.292,
+      "step": 772
+    },
+    {
+      "epoch": 3.131139240506329,
+      "grad_norm": 0.24939950062725152,
+      "learning_rate": 1.4590996407327284e-05,
+      "loss": 0.2961,
+      "step": 773
+    },
+    {
+      "epoch": 3.1351898734177217,
+      "grad_norm": 0.21580847796472283,
+      "learning_rate": 1.4536374757881487e-05,
+      "loss": 0.2841,
+      "step": 774
+    },
+    {
+      "epoch": 3.1392405063291138,
+      "grad_norm": 0.24979887704200043,
+      "learning_rate": 1.4481797111722271e-05,
+      "loss": 0.2826,
+      "step": 775
+    },
+    {
+      "epoch": 3.1432911392405063,
+      "grad_norm": 0.2442545428856667,
+      "learning_rate": 1.4427263908410507e-05,
+      "loss": 0.2912,
+      "step": 776
+    },
+    {
+      "epoch": 3.147341772151899,
+      "grad_norm": 0.25934014434732144,
+      "learning_rate": 1.4372775587149108e-05,
+      "loss": 0.2822,
+      "step": 777
+    },
+    {
+      "epoch": 3.1513924050632913,
+      "grad_norm": 0.24987757313801345,
+      "learning_rate": 1.4318332586779522e-05,
+      "loss": 0.2976,
+      "step": 778
+    },
+    {
+      "epoch": 3.1554430379746834,
+      "grad_norm": 0.2678994665789973,
+      "learning_rate": 1.4263935345778202e-05,
+      "loss": 0.2925,
+      "step": 779
+    },
+    {
+      "epoch": 3.159493670886076,
+      "grad_norm": 0.23995350099421514,
+      "learning_rate": 1.420958430225303e-05,
+      "loss": 0.2775,
+      "step": 780
+    },
+    {
+      "epoch": 3.1635443037974684,
+      "grad_norm": 0.22165223254839372,
+      "learning_rate": 1.415527989393985e-05,
+      "loss": 0.2847,
+      "step": 781
+    },
+    {
+      "epoch": 3.167594936708861,
+      "grad_norm": 0.25889148569546694,
+      "learning_rate": 1.410102255819891e-05,
+      "loss": 0.2846,
+      "step": 782
+    },
+    {
+      "epoch": 3.171645569620253,
+      "grad_norm": 0.2399490375403679,
+      "learning_rate": 1.404681273201131e-05,
+      "loss": 0.2884,
+      "step": 783
+    },
+    {
+      "epoch": 3.1756962025316455,
+      "grad_norm": 0.24948759763067918,
+      "learning_rate": 1.399265085197556e-05,
+      "loss": 0.2925,
+      "step": 784
+    },
+    {
+      "epoch": 3.179746835443038,
+      "grad_norm": 0.24109988982453162,
+      "learning_rate": 1.393853735430398e-05,
+      "loss": 0.2801,
+      "step": 785
+    },
+    {
+      "epoch": 3.1837974683544306,
+      "grad_norm": 0.23155794989539136,
+      "learning_rate": 1.3884472674819246e-05,
+      "loss": 0.2855,
+      "step": 786
+    },
+    {
+      "epoch": 3.1878481012658226,
+      "grad_norm": 0.23421820671219792,
+      "learning_rate": 1.3830457248950864e-05,
+      "loss": 0.2896,
+      "step": 787
+    },
+    {
+      "epoch": 3.191898734177215,
+      "grad_norm": 0.21331803454878973,
+      "learning_rate": 1.377649151173163e-05,
+      "loss": 0.2865,
+      "step": 788
+    },
+    {
+      "epoch": 3.1959493670886077,
+      "grad_norm": 0.2164888083548083,
+      "learning_rate": 1.3722575897794181e-05,
+      "loss": 0.2726,
+      "step": 789
+    },
+    {
+      "epoch": 3.2,
+      "grad_norm": 0.23335132804271697,
+      "learning_rate": 1.3668710841367472e-05,
+      "loss": 0.2848,
+      "step": 790
+    },
+    {
+      "epoch": 3.2040506329113922,
+      "grad_norm": 0.20829418220970852,
+      "learning_rate": 1.361489677627324e-05,
+      "loss": 0.2964,
+      "step": 791
+    },
+    {
+      "epoch": 3.2081012658227848,
+      "grad_norm": 0.23414537709480518,
+      "learning_rate": 1.3561134135922585e-05,
+      "loss": 0.2934,
+      "step": 792
+    },
+    {
+      "epoch": 3.2121518987341773,
+      "grad_norm": 0.2176746272363139,
+      "learning_rate": 1.350742335331241e-05,
+      "loss": 0.3042,
+      "step": 793
+    },
+    {
+      "epoch": 3.21620253164557,
+      "grad_norm": 0.22267002128216265,
+      "learning_rate": 1.345376486102198e-05,
+      "loss": 0.2876,
+      "step": 794
+    },
+    {
+      "epoch": 3.220253164556962,
+      "grad_norm": 0.2204769125552546,
+      "learning_rate": 1.3400159091209414e-05,
+      "loss": 0.2838,
+      "step": 795
+    },
+    {
+      "epoch": 3.2243037974683544,
+      "grad_norm": 0.1988424456421814,
+      "learning_rate": 1.3346606475608216e-05,
+      "loss": 0.2777,
+      "step": 796
+    },
+    {
+      "epoch": 3.228354430379747,
+      "grad_norm": 0.23854789660752368,
+      "learning_rate": 1.3293107445523781e-05,
+      "loss": 0.2985,
+      "step": 797
+    },
+    {
+      "epoch": 3.2324050632911394,
+      "grad_norm": 0.21164866444243996,
+      "learning_rate": 1.3239662431829949e-05,
+      "loss": 0.2883,
+      "step": 798
+    },
+    {
+      "epoch": 3.2364556962025315,
+      "grad_norm": 0.2150061686028223,
+      "learning_rate": 1.3186271864965509e-05,
+      "loss": 0.2954,
+      "step": 799
+    },
+    {
+      "epoch": 3.240506329113924,
+      "grad_norm": 0.24649483052112459,
+      "learning_rate": 1.3132936174930756e-05,
+      "loss": 0.2932,
+      "step": 800
+    },
+    {
+      "epoch": 3.2445569620253165,
+      "grad_norm": 0.2214953205836067,
+      "learning_rate": 1.3079655791283995e-05,
+      "loss": 0.2878,
+      "step": 801
+    },
+    {
+      "epoch": 3.248607594936709,
+      "grad_norm": 0.2513385019884068,
+      "learning_rate": 1.3026431143138108e-05,
+      "loss": 0.2882,
+      "step": 802
+    },
+    {
+      "epoch": 3.252658227848101,
+      "grad_norm": 0.21397454765750795,
+      "learning_rate": 1.2973262659157114e-05,
+      "loss": 0.2866,
+      "step": 803
+    },
+    {
+      "epoch": 3.2567088607594936,
+      "grad_norm": 0.23987327454644225,
+      "learning_rate": 1.2920150767552651e-05,
+      "loss": 0.2874,
+      "step": 804
+    },
+    {
+      "epoch": 3.260759493670886,
+      "grad_norm": 0.2114678875386145,
+      "learning_rate": 1.2867095896080607e-05,
+      "loss": 0.2875,
+      "step": 805
+    },
+    {
+      "epoch": 3.2648101265822786,
+      "grad_norm": 0.21248819429477062,
+      "learning_rate": 1.2814098472037612e-05,
+      "loss": 0.2805,
+      "step": 806
+    },
+    {
+      "epoch": 3.2688607594936707,
+      "grad_norm": 0.20731994435435652,
+      "learning_rate": 1.276115892225764e-05,
+      "loss": 0.2868,
+      "step": 807
+    },
+    {
+      "epoch": 3.2729113924050632,
+      "grad_norm": 0.2160797315386458,
+      "learning_rate": 1.2708277673108555e-05,
+      "loss": 0.2927,
+      "step": 808
+    },
+    {
+      "epoch": 3.2769620253164558,
+      "grad_norm": 0.204917384705678,
+      "learning_rate": 1.2655455150488649e-05,
+      "loss": 0.2866,
+      "step": 809
+    },
+    {
+      "epoch": 3.2810126582278483,
+      "grad_norm": 0.22323604997333582,
+      "learning_rate": 1.2602691779823272e-05,
+      "loss": 0.2946,
+      "step": 810
+    },
+    {
+      "epoch": 3.2850632911392403,
+      "grad_norm": 0.24196361319227186,
+      "learning_rate": 1.2549987986061355e-05,
+      "loss": 0.2861,
+      "step": 811
+    },
+    {
+      "epoch": 3.289113924050633,
+      "grad_norm": 0.21774631755080523,
+      "learning_rate": 1.2497344193672005e-05,
+      "loss": 0.2996,
+      "step": 812
+    },
+    {
+      "epoch": 3.2931645569620254,
+      "grad_norm": 0.23926786938407224,
+      "learning_rate": 1.2444760826641092e-05,
+      "loss": 0.2931,
+      "step": 813
+    },
+    {
+      "epoch": 3.297215189873418,
+      "grad_norm": 0.219562369402471,
+      "learning_rate": 1.2392238308467817e-05,
+      "loss": 0.2929,
+      "step": 814
+    },
+    {
+      "epoch": 3.30126582278481,
+      "grad_norm": 0.20304422746421327,
+      "learning_rate": 1.2339777062161326e-05,
+      "loss": 0.2878,
+      "step": 815
+    },
+    {
+      "epoch": 3.3053164556962025,
+      "grad_norm": 0.20717568693085944,
+      "learning_rate": 1.2287377510237293e-05,
+      "loss": 0.2932,
+      "step": 816
+    },
+    {
+      "epoch": 3.309367088607595,
+      "grad_norm": 0.19992246480105957,
+      "learning_rate": 1.2235040074714488e-05,
+      "loss": 0.287,
+      "step": 817
+    },
+    {
+      "epoch": 3.3134177215189875,
+      "grad_norm": 0.2021391148370685,
+      "learning_rate": 1.2182765177111434e-05,
+      "loss": 0.2886,
+      "step": 818
+    },
+    {
+      "epoch": 3.3174683544303796,
+      "grad_norm": 0.1920961813040169,
+      "learning_rate": 1.213055323844297e-05,
+      "loss": 0.286,
+      "step": 819
+    },
+    {
+      "epoch": 3.321518987341772,
+      "grad_norm": 0.22141642225655164,
+      "learning_rate": 1.2078404679216864e-05,
+      "loss": 0.2777,
+      "step": 820
+    },
+    {
+      "epoch": 3.3255696202531646,
+      "grad_norm": 0.20425615800691138,
+      "learning_rate": 1.2026319919430458e-05,
+      "loss": 0.2905,
+      "step": 821
+    },
+    {
+      "epoch": 3.329620253164557,
+      "grad_norm": 0.20899243103472706,
+      "learning_rate": 1.1974299378567227e-05,
+      "loss": 0.2935,
+      "step": 822
+    },
+    {
+      "epoch": 3.333670886075949,
+      "grad_norm": 0.20789366033547146,
+      "learning_rate": 1.1922343475593462e-05,
+      "loss": 0.2712,
+      "step": 823
+    },
+    {
+      "epoch": 3.3377215189873417,
+      "grad_norm": 0.22568646930699734,
+      "learning_rate": 1.187045262895488e-05,
+      "loss": 0.2776,
+      "step": 824
+    },
+    {
+      "epoch": 3.3417721518987342,
+      "grad_norm": 0.2125484657327162,
+      "learning_rate": 1.1818627256573203e-05,
+      "loss": 0.2951,
+      "step": 825
+    },
+    {
+      "epoch": 3.3458227848101267,
+      "grad_norm": 0.1974562616173397,
+      "learning_rate": 1.1766867775842864e-05,
+      "loss": 0.2908,
+      "step": 826
+    },
+    {
+      "epoch": 3.349873417721519,
+      "grad_norm": 0.22797808940806027,
+      "learning_rate": 1.1715174603627615e-05,
+      "loss": 0.2701,
+      "step": 827
+    },
+    {
+      "epoch": 3.3539240506329113,
+      "grad_norm": 0.21614535549554612,
+      "learning_rate": 1.1663548156257147e-05,
+      "loss": 0.2932,
+      "step": 828
+    },
+    {
+      "epoch": 3.357974683544304,
+      "grad_norm": 0.1923485837170381,
+      "learning_rate": 1.161198884952377e-05,
+      "loss": 0.2876,
+      "step": 829
+    },
+    {
+      "epoch": 3.3620253164556964,
+      "grad_norm": 0.2100017719297909,
+      "learning_rate": 1.1560497098679056e-05,
+      "loss": 0.2884,
+      "step": 830
+    },
+    {
+      "epoch": 3.3660759493670884,
+      "grad_norm": 0.20411682685283355,
+      "learning_rate": 1.1509073318430479e-05,
+      "loss": 0.2842,
+      "step": 831
+    },
+    {
+      "epoch": 3.370126582278481,
+      "grad_norm": 0.2293387069952781,
+      "learning_rate": 1.1457717922938116e-05,
+      "loss": 0.2937,
+      "step": 832
+    },
+    {
+      "epoch": 3.3741772151898735,
+      "grad_norm": 0.21702790023292826,
+      "learning_rate": 1.1406431325811233e-05,
+      "loss": 0.2845,
+      "step": 833
+    },
+    {
+      "epoch": 3.378227848101266,
+      "grad_norm": 0.21193513520266027,
+      "learning_rate": 1.135521394010506e-05,
+      "loss": 0.2792,
+      "step": 834
+    },
+    {
+      "epoch": 3.382278481012658,
+      "grad_norm": 0.20640711978675622,
+      "learning_rate": 1.1304066178317367e-05,
+      "loss": 0.2814,
+      "step": 835
+    },
+    {
+      "epoch": 3.3863291139240506,
+      "grad_norm": 0.22192354573598064,
+      "learning_rate": 1.1252988452385199e-05,
+      "loss": 0.295,
+      "step": 836
+    },
+    {
+      "epoch": 3.390379746835443,
+      "grad_norm": 0.20112866681643712,
+      "learning_rate": 1.1201981173681536e-05,
+      "loss": 0.2862,
+      "step": 837
+    },
+    {
+      "epoch": 3.3944303797468356,
+      "grad_norm": 0.20442912533624277,
+      "learning_rate": 1.1151044753011991e-05,
+      "loss": 0.2843,
+      "step": 838
+    },
+    {
+      "epoch": 3.3984810126582277,
+      "grad_norm": 0.1890480924986652,
+      "learning_rate": 1.1100179600611491e-05,
+      "loss": 0.2794,
+      "step": 839
+    },
+    {
+      "epoch": 3.40253164556962,
+      "grad_norm": 0.20022571466924236,
+      "learning_rate": 1.1049386126140985e-05,
+      "loss": 0.2911,
+      "step": 840
+    },
+    {
+      "epoch": 3.4065822784810127,
+      "grad_norm": 0.2158270307865968,
+      "learning_rate": 1.0998664738684128e-05,
+      "loss": 0.2832,
+      "step": 841
+    },
+    {
+      "epoch": 3.410632911392405,
+      "grad_norm": 0.2175879437400424,
+      "learning_rate": 1.0948015846744e-05,
+      "loss": 0.2934,
+      "step": 842
+    },
+    {
+      "epoch": 3.4146835443037973,
+      "grad_norm": 0.1874746643141621,
+      "learning_rate": 1.0897439858239832e-05,
+      "loss": 0.2838,
+      "step": 843
+    },
+    {
+      "epoch": 3.41873417721519,
+      "grad_norm": 0.1873305757425555,
+      "learning_rate": 1.0846937180503652e-05,
+      "loss": 0.2843,
+      "step": 844
+    },
+    {
+      "epoch": 3.4227848101265823,
+      "grad_norm": 0.20870206631035448,
+      "learning_rate": 1.0796508220277117e-05,
+      "loss": 0.2798,
+      "step": 845
+    },
+    {
+      "epoch": 3.426835443037975,
+      "grad_norm": 0.21100092524707215,
+      "learning_rate": 1.0746153383708107e-05,
+      "loss": 0.2966,
+      "step": 846
+    },
+    {
+      "epoch": 3.430886075949367,
+      "grad_norm": 0.20460975177476892,
+      "learning_rate": 1.0695873076347579e-05,
+      "loss": 0.2913,
+      "step": 847
+    },
+    {
+      "epoch": 3.4349367088607594,
+      "grad_norm": 0.22138428218856823,
+      "learning_rate": 1.0645667703146205e-05,
+      "loss": 0.288,
+      "step": 848
+    },
+    {
+      "epoch": 3.438987341772152,
+      "grad_norm": 0.22449197499253778,
+      "learning_rate": 1.0595537668451161e-05,
+      "loss": 0.2911,
+      "step": 849
+    },
+    {
+      "epoch": 3.4430379746835444,
+      "grad_norm": 0.24837991156045872,
+      "learning_rate": 1.0545483376002854e-05,
+      "loss": 0.2845,
+      "step": 850
+    },
+    {
+      "epoch": 3.4470886075949365,
+      "grad_norm": 0.23198748148413162,
+      "learning_rate": 1.0495505228931676e-05,
+      "loss": 0.2899,
+      "step": 851
+    },
+    {
+      "epoch": 3.451139240506329,
+      "grad_norm": 0.2280099440852967,
+      "learning_rate": 1.044560362975474e-05,
+      "loss": 0.2734,
+      "step": 852
+    },
+    {
+      "epoch": 3.4551898734177215,
+      "grad_norm": 0.21544896576957115,
+      "learning_rate": 1.0395778980372695e-05,
+      "loss": 0.2886,
+      "step": 853
+    },
+    {
+      "epoch": 3.459240506329114,
+      "grad_norm": 0.2575843390234529,
+      "learning_rate": 1.0346031682066381e-05,
+      "loss": 0.2772,
+      "step": 854
+    },
+    {
+      "epoch": 3.463291139240506,
+      "grad_norm": 0.1988591481973347,
+      "learning_rate": 1.0296362135493724e-05,
+      "loss": 0.2804,
+      "step": 855
+    },
+    {
+      "epoch": 3.4673417721518986,
+      "grad_norm": 0.2197270639467392,
+      "learning_rate": 1.0246770740686422e-05,
+      "loss": 0.2852,
+      "step": 856
+    },
+    {
+      "epoch": 3.471392405063291,
+      "grad_norm": 0.22697873178082958,
+      "learning_rate": 1.0197257897046743e-05,
+      "loss": 0.276,
+      "step": 857
+    },
+    {
+      "epoch": 3.4754430379746837,
+      "grad_norm": 0.2075252152907557,
+      "learning_rate": 1.014782400334433e-05,
+      "loss": 0.2802,
+      "step": 858
+    },
+    {
+      "epoch": 3.479493670886076,
+      "grad_norm": 0.21384384118373037,
+      "learning_rate": 1.009846945771296e-05,
+      "loss": 0.2905,
+      "step": 859
+    },
+    {
+      "epoch": 3.4835443037974683,
+      "grad_norm": 0.20302553472673937,
+      "learning_rate": 1.0049194657647363e-05,
+      "loss": 0.2837,
+      "step": 860
+    },
+    {
+      "epoch": 3.487594936708861,
+      "grad_norm": 0.21604239496866418,
+      "learning_rate": 1.0000000000000006e-05,
+      "loss": 0.2894,
+      "step": 861
+    },
+    {
+      "epoch": 3.4916455696202533,
+      "grad_norm": 0.20281097098870102,
+      "learning_rate": 9.950885880977891e-06,
+      "loss": 0.2901,
+      "step": 862
+    },
+    {
+      "epoch": 3.4956962025316454,
+      "grad_norm": 0.22238302294511939,
+      "learning_rate": 9.901852696139382e-06,
+      "loss": 0.2839,
+      "step": 863
+    },
+    {
+      "epoch": 3.499746835443038,
+      "grad_norm": 0.2145843496436782,
+      "learning_rate": 9.852900840391027e-06,
+      "loss": 0.2903,
+      "step": 864
+    },
+    {
+      "epoch": 3.5037974683544304,
+      "grad_norm": 0.21287049264040034,
+      "learning_rate": 9.804030707984313e-06,
+      "loss": 0.2956,
+      "step": 865
+    },
+    {
+      "epoch": 3.507848101265823,
+      "grad_norm": 0.20225493073130907,
+      "learning_rate": 9.755242692512599e-06,
+      "loss": 0.2827,
+      "step": 866
+    },
+    {
+      "epoch": 3.5118987341772154,
+      "grad_norm": 0.22209946469975445,
+      "learning_rate": 9.70653718690782e-06,
+      "loss": 0.2897,
+      "step": 867
+    },
+    {
+      "epoch": 3.5159493670886075,
+      "grad_norm": 0.19852029445189523,
+      "learning_rate": 9.657914583437454e-06,
+      "loss": 0.2845,
+      "step": 868
+    },
+    {
+      "epoch": 3.52,
+      "grad_norm": 0.20706582903765908,
+      "learning_rate": 9.609375273701246e-06,
+      "loss": 0.2836,
+      "step": 869
+    },
+    {
+      "epoch": 3.5240506329113925,
+      "grad_norm": 0.1936995312024811,
+      "learning_rate": 9.560919648628133e-06,
+      "loss": 0.2858,
+      "step": 870
+    },
+    {
+      "epoch": 3.5281012658227846,
+      "grad_norm": 0.1882415113282504,
+      "learning_rate": 9.512548098473047e-06,
+      "loss": 0.2862,
+      "step": 871
+    },
+    {
+      "epoch": 3.532151898734177,
+      "grad_norm": 0.2008268073197883,
+      "learning_rate": 9.464261012813825e-06,
+      "loss": 0.2842,
+      "step": 872
+    },
+    {
+      "epoch": 3.5362025316455696,
+      "grad_norm": 0.20481160084149458,
+      "learning_rate": 9.416058780547987e-06,
+      "loss": 0.2834,
+      "step": 873
+    },
+    {
+      "epoch": 3.540253164556962,
+      "grad_norm": 0.20072761642104803,
+      "learning_rate": 9.367941789889714e-06,
+      "loss": 0.2974,
+      "step": 874
+    },
+    {
+      "epoch": 3.5443037974683547,
+      "grad_norm": 0.21108337955934747,
+      "learning_rate": 9.319910428366607e-06,
+      "loss": 0.2798,
+      "step": 875
+    },
+    {
+      "epoch": 3.5483544303797467,
+      "grad_norm": 0.19976229576230065,
+      "learning_rate": 9.271965082816667e-06,
+      "loss": 0.2876,
+      "step": 876
+    },
+    {
+      "epoch": 3.5524050632911393,
+      "grad_norm": 0.22247653168595152,
+      "learning_rate": 9.224106139385111e-06,
+      "loss": 0.2847,
+      "step": 877
+    },
+    {
+      "epoch": 3.5564556962025318,
+      "grad_norm": 0.21211245676647425,
+      "learning_rate": 9.176333983521291e-06,
+      "loss": 0.2965,
+      "step": 878
+    },
+    {
+      "epoch": 3.560506329113924,
+      "grad_norm": 0.19718256249484264,
+      "learning_rate": 9.12864899997558e-06,
+      "loss": 0.2969,
+      "step": 879
+    },
+    {
+      "epoch": 3.5645569620253164,
+      "grad_norm": 0.21722184691199534,
+      "learning_rate": 9.08105157279628e-06,
+      "loss": 0.2875,
+      "step": 880
+    },
+    {
+      "epoch": 3.568607594936709,
+      "grad_norm": 0.21797068556016855,
+      "learning_rate": 9.03354208532653e-06,
+      "loss": 0.2944,
+      "step": 881
+    },
+    {
+      "epoch": 3.5726582278481014,
+      "grad_norm": 0.19312699182823007,
+      "learning_rate": 8.986120920201205e-06,
+      "loss": 0.2812,
+      "step": 882
+    },
+    {
+      "epoch": 3.576708860759494,
+      "grad_norm": 0.20749636858225792,
+      "learning_rate": 8.938788459343852e-06,
+      "loss": 0.2815,
+      "step": 883
+    },
+    {
+      "epoch": 3.580759493670886,
+      "grad_norm": 0.19806588776111292,
+      "learning_rate": 8.8915450839636e-06,
+      "loss": 0.2855,
+      "step": 884
+    },
+    {
+      "epoch": 3.5848101265822785,
+      "grad_norm": 0.19246340817984178,
+      "learning_rate": 8.844391174552116e-06,
+      "loss": 0.2803,
+      "step": 885
+    },
+    {
+      "epoch": 3.588860759493671,
+      "grad_norm": 0.19731222160624567,
+      "learning_rate": 8.797327110880479e-06,
+      "loss": 0.2861,
+      "step": 886
+    },
+    {
+      "epoch": 3.592911392405063,
+      "grad_norm": 0.2127932660508,
+      "learning_rate": 8.750353271996206e-06,
+      "loss": 0.2891,
+      "step": 887
+    },
+    {
+      "epoch": 3.5969620253164556,
+      "grad_norm": 0.19701944230542026,
+      "learning_rate": 8.703470036220132e-06,
+      "loss": 0.2836,
+      "step": 888
+    },
+    {
+      "epoch": 3.601012658227848,
+      "grad_norm": 0.19653289172246158,
+      "learning_rate": 8.656677781143394e-06,
+      "loss": 0.2979,
+      "step": 889
+    },
+    {
+      "epoch": 3.6050632911392406,
+      "grad_norm": 0.21468455432878147,
+      "learning_rate": 8.609976883624377e-06,
+      "loss": 0.2938,
+      "step": 890
+    },
+    {
+      "epoch": 3.609113924050633,
+      "grad_norm": 0.1901333404947332,
+      "learning_rate": 8.563367719785698e-06,
+      "loss": 0.2862,
+      "step": 891
+    },
+    {
+      "epoch": 3.613164556962025,
+      "grad_norm": 0.20111365319272487,
+      "learning_rate": 8.516850665011138e-06,
+      "loss": 0.2876,
+      "step": 892
+    },
+    {
+      "epoch": 3.6172151898734177,
+      "grad_norm": 0.21475291429817356,
+      "learning_rate": 8.47042609394269e-06,
+      "loss": 0.2883,
+      "step": 893
+    },
+    {
+      "epoch": 3.6212658227848102,
+      "grad_norm": 0.19489609810408057,
+      "learning_rate": 8.424094380477432e-06,
+      "loss": 0.281,
+      "step": 894
+    },
+    {
+      "epoch": 3.6253164556962023,
+      "grad_norm": 0.21390260241308087,
+      "learning_rate": 8.37785589776465e-06,
+      "loss": 0.2919,
+      "step": 895
+    },
+    {
+      "epoch": 3.629367088607595,
+      "grad_norm": 0.2198000772503074,
+      "learning_rate": 8.331711018202694e-06,
+      "loss": 0.2869,
+      "step": 896
+    },
+    {
+      "epoch": 3.6334177215189873,
+      "grad_norm": 0.20061402122764105,
+      "learning_rate": 8.285660113436104e-06,
+      "loss": 0.2832,
+      "step": 897
+    },
+    {
+      "epoch": 3.63746835443038,
+      "grad_norm": 0.2097354818602275,
+      "learning_rate": 8.239703554352527e-06,
+      "loss": 0.294,
+      "step": 898
+    },
+    {
+      "epoch": 3.6415189873417724,
+      "grad_norm": 0.18827044684311525,
+      "learning_rate": 8.193841711079775e-06,
+      "loss": 0.2868,
+      "step": 899
+    },
+    {
+      "epoch": 3.6455696202531644,
+      "grad_norm": 0.1904833352191653,
+      "learning_rate": 8.148074952982828e-06,
+      "loss": 0.292,
+      "step": 900
+    },
+    {
+      "epoch": 3.649620253164557,
+      "grad_norm": 0.21709309070796495,
+      "learning_rate": 8.102403648660859e-06,
+      "loss": 0.279,
+      "step": 901
+    },
+    {
+      "epoch": 3.6536708860759495,
+      "grad_norm": 0.19989430957448667,
+      "learning_rate": 8.056828165944282e-06,
+      "loss": 0.2921,
+      "step": 902
+    },
+    {
+      "epoch": 3.6577215189873415,
+      "grad_norm": 0.18311013873712442,
+      "learning_rate": 8.011348871891762e-06,
+      "loss": 0.2863,
+      "step": 903
+    },
+    {
+      "epoch": 3.661772151898734,
+      "grad_norm": 0.20105107809619982,
+      "learning_rate": 7.965966132787287e-06,
+      "loss": 0.288,
+      "step": 904
+    },
+    {
+      "epoch": 3.6658227848101266,
+      "grad_norm": 0.2053133549997552,
+      "learning_rate": 7.920680314137189e-06,
+      "loss": 0.2867,
+      "step": 905
+    },
+    {
+      "epoch": 3.669873417721519,
+      "grad_norm": 0.20102575746773482,
+      "learning_rate": 7.875491780667246e-06,
+      "loss": 0.2899,
+      "step": 906
+    },
+    {
+      "epoch": 3.6739240506329116,
+      "grad_norm": 0.21386802301427754,
+      "learning_rate": 7.830400896319667e-06,
+      "loss": 0.2869,
+      "step": 907
+    },
+    {
+      "epoch": 3.6779746835443037,
+      "grad_norm": 0.20636145682346843,
+      "learning_rate": 7.785408024250259e-06,
+      "loss": 0.2867,
+      "step": 908
+    },
+    {
+      "epoch": 3.682025316455696,
+      "grad_norm": 0.19234325608174394,
+      "learning_rate": 7.74051352682542e-06,
+      "loss": 0.2877,
+      "step": 909
+    },
+    {
+      "epoch": 3.6860759493670887,
+      "grad_norm": 0.19737294720216655,
+      "learning_rate": 7.695717765619257e-06,
+      "loss": 0.2884,
+      "step": 910
+    },
+    {
+      "epoch": 3.690126582278481,
+      "grad_norm": 0.19404286765636175,
+      "learning_rate": 7.651021101410673e-06,
+      "loss": 0.2855,
+      "step": 911
+    },
+    {
+      "epoch": 3.6941772151898733,
+      "grad_norm": 0.20004870684654893,
+      "learning_rate": 7.606423894180464e-06,
+      "loss": 0.2968,
+      "step": 912
+    },
+    {
+      "epoch": 3.698227848101266,
+      "grad_norm": 0.19520879088025883,
+      "learning_rate": 7.56192650310839e-06,
+      "loss": 0.2897,
+      "step": 913
+    },
+    {
+      "epoch": 3.7022784810126583,
+      "grad_norm": 0.2018357608113645,
+      "learning_rate": 7.517529286570349e-06,
+      "loss": 0.2786,
+      "step": 914
+    },
+    {
+      "epoch": 3.706329113924051,
+      "grad_norm": 0.20066773106884014,
+      "learning_rate": 7.473232602135387e-06,
+      "loss": 0.2832,
+      "step": 915
+    },
+    {
+      "epoch": 3.710379746835443,
+      "grad_norm": 0.19715793797498712,
+      "learning_rate": 7.429036806562935e-06,
+      "loss": 0.2756,
+      "step": 916
+    },
+    {
+      "epoch": 3.7144303797468354,
+      "grad_norm": 0.20278712432869672,
+      "learning_rate": 7.3849422557998455e-06,
+      "loss": 0.2871,
+      "step": 917
+    },
+    {
+      "epoch": 3.718481012658228,
+      "grad_norm": 0.2089609482347749,
+      "learning_rate": 7.340949304977567e-06,
+      "loss": 0.2937,
+      "step": 918
+    },
+    {
+      "epoch": 3.72253164556962,
+      "grad_norm": 0.19122982698056584,
+      "learning_rate": 7.297058308409282e-06,
+      "loss": 0.2947,
+      "step": 919
+    },
+    {
+      "epoch": 3.7265822784810125,
+      "grad_norm": 0.20745302684056235,
+      "learning_rate": 7.25326961958704e-06,
+      "loss": 0.281,
+      "step": 920
+    },
+    {
+      "epoch": 3.730632911392405,
+      "grad_norm": 0.20459447406864384,
+      "learning_rate": 7.209583591178921e-06,
+      "loss": 0.2902,
+      "step": 921
+    },
+    {
+      "epoch": 3.7346835443037976,
+      "grad_norm": 0.19858233397747974,
+      "learning_rate": 7.1660005750261925e-06,
+      "loss": 0.2964,
+      "step": 922
+    },
+    {
+      "epoch": 3.73873417721519,
+      "grad_norm": 0.18371801745591865,
+      "learning_rate": 7.1225209221404765e-06,
+      "loss": 0.284,
+      "step": 923
+    },
+    {
+      "epoch": 3.742784810126582,
+      "grad_norm": 0.19468656197017672,
+      "learning_rate": 7.079144982700909e-06,
+      "loss": 0.2888,
+      "step": 924
+    },
+    {
+      "epoch": 3.7468354430379747,
+      "grad_norm": 0.19969291299950873,
+      "learning_rate": 7.0358731060513695e-06,
+      "loss": 0.2718,
+      "step": 925
+    },
+    {
+      "epoch": 3.750886075949367,
+      "grad_norm": 0.19491813874995495,
+      "learning_rate": 6.99270564069757e-06,
+      "loss": 0.2768,
+      "step": 926
+    },
+    {
+      "epoch": 3.7549367088607593,
+      "grad_norm": 0.20287188243459622,
+      "learning_rate": 6.949642934304375e-06,
+      "loss": 0.2793,
+      "step": 927
+    },
+    {
+      "epoch": 3.7589873417721518,
+      "grad_norm": 0.19440067320820234,
+      "learning_rate": 6.906685333692871e-06,
+      "loss": 0.2976,
+      "step": 928
+    },
+    {
+      "epoch": 3.7630379746835443,
+      "grad_norm": 0.18331979301269158,
+      "learning_rate": 6.86383318483769e-06,
+      "loss": 0.2785,
+      "step": 929
+    },
+    {
+      "epoch": 3.767088607594937,
+      "grad_norm": 0.1836947910395293,
+      "learning_rate": 6.821086832864139e-06,
+      "loss": 0.2908,
+      "step": 930
+    },
+    {
+      "epoch": 3.7711392405063293,
+      "grad_norm": 0.19553835939866884,
+      "learning_rate": 6.77844662204546e-06,
+      "loss": 0.2821,
+      "step": 931
+    },
+    {
+      "epoch": 3.7751898734177214,
+      "grad_norm": 0.19098099240846114,
+      "learning_rate": 6.7359128958000455e-06,
+      "loss": 0.2829,
+      "step": 932
+    },
+    {
+      "epoch": 3.779240506329114,
+      "grad_norm": 0.18621343979146357,
+      "learning_rate": 6.693485996688695e-06,
+      "loss": 0.2763,
+      "step": 933
+    },
+    {
+      "epoch": 3.7832911392405064,
+      "grad_norm": 0.18256640973286,
+      "learning_rate": 6.651166266411801e-06,
+      "loss": 0.293,
+      "step": 934
+    },
+    {
+      "epoch": 3.7873417721518985,
+      "grad_norm": 0.18275592170162497,
+      "learning_rate": 6.6089540458066725e-06,
+      "loss": 0.2854,
+      "step": 935
+    },
+    {
+      "epoch": 3.791392405063291,
+      "grad_norm": 0.19935181954775735,
+      "learning_rate": 6.566849674844711e-06,
+      "loss": 0.283,
+      "step": 936
+    },
+    {
+      "epoch": 3.7954430379746835,
+      "grad_norm": 0.18921118664933856,
+      "learning_rate": 6.524853492628747e-06,
+      "loss": 0.2878,
+      "step": 937
+    },
+    {
+      "epoch": 3.799493670886076,
+      "grad_norm": 0.18750155753070802,
+      "learning_rate": 6.4829658373902536e-06,
+      "loss": 0.2792,
+      "step": 938
+    },
+    {
+      "epoch": 3.8035443037974686,
+      "grad_norm": 0.1849010169040959,
+      "learning_rate": 6.441187046486648e-06,
+      "loss": 0.2959,
+      "step": 939
+    },
+    {
+      "epoch": 3.8075949367088606,
+      "grad_norm": 0.19667534145208243,
+      "learning_rate": 6.399517456398567e-06,
+      "loss": 0.294,
+      "step": 940
+    },
+    {
+      "epoch": 3.811645569620253,
+      "grad_norm": 0.18995832619892009,
+      "learning_rate": 6.357957402727164e-06,
+      "loss": 0.2812,
+      "step": 941
+    },
+    {
+      "epoch": 3.8156962025316457,
+      "grad_norm": 0.18249748809680816,
+      "learning_rate": 6.316507220191395e-06,
+      "loss": 0.2903,
+      "step": 942
+    },
+    {
+      "epoch": 3.8197468354430377,
+      "grad_norm": 0.18565067952256553,
+      "learning_rate": 6.275167242625331e-06,
+      "loss": 0.285,
+      "step": 943
+    },
+    {
+      "epoch": 3.8237974683544302,
+      "grad_norm": 0.18232144594490224,
+      "learning_rate": 6.233937802975471e-06,
+      "loss": 0.2811,
+      "step": 944
+    },
+    {
+      "epoch": 3.8278481012658228,
+      "grad_norm": 0.18168224094827404,
+      "learning_rate": 6.192819233298046e-06,
+      "loss": 0.2736,
+      "step": 945
+    },
+    {
+      "epoch": 3.8318987341772153,
+      "grad_norm": 0.19073649932936845,
+      "learning_rate": 6.151811864756383e-06,
+      "loss": 0.2747,
+      "step": 946
+    },
+    {
+      "epoch": 3.835949367088608,
+      "grad_norm": 0.19471677994473252,
+      "learning_rate": 6.1109160276181655e-06,
+      "loss": 0.2862,
+      "step": 947
+    },
+    {
+      "epoch": 3.84,
+      "grad_norm": 0.1920677030667708,
+      "learning_rate": 6.070132051252868e-06,
+      "loss": 0.2878,
+      "step": 948
+    },
+    {
+      "epoch": 3.8440506329113924,
+      "grad_norm": 0.18599014607869174,
+      "learning_rate": 6.0294602641290034e-06,
+      "loss": 0.2836,
+      "step": 949
+    },
+    {
+      "epoch": 3.848101265822785,
+      "grad_norm": 0.1922572474775001,
+      "learning_rate": 5.988900993811575e-06,
+      "loss": 0.2916,
+      "step": 950
+    },
+    {
+      "epoch": 3.852151898734177,
+      "grad_norm": 0.20006782271175866,
+      "learning_rate": 5.948454566959363e-06,
+      "loss": 0.2931,
+      "step": 951
+    },
+    {
+      "epoch": 3.8562025316455695,
+      "grad_norm": 0.194205118442489,
+      "learning_rate": 5.908121309322328e-06,
+      "loss": 0.2891,
+      "step": 952
+    },
+    {
+      "epoch": 3.860253164556962,
+      "grad_norm": 0.18766943053814916,
+      "learning_rate": 5.867901545738976e-06,
+      "loss": 0.282,
+      "step": 953
+    },
+    {
+      "epoch": 3.8643037974683545,
+      "grad_norm": 0.1928828068774487,
+      "learning_rate": 5.827795600133774e-06,
+      "loss": 0.2839,
+      "step": 954
+    },
+    {
+      "epoch": 3.868354430379747,
+      "grad_norm": 0.1832513311117474,
+      "learning_rate": 5.787803795514466e-06,
+      "loss": 0.287,
+      "step": 955
+    },
+    {
+      "epoch": 3.872405063291139,
+      "grad_norm": 0.20206380833746704,
+      "learning_rate": 5.747926453969576e-06,
+      "loss": 0.2935,
+      "step": 956
+    },
+    {
+      "epoch": 3.8764556962025316,
+      "grad_norm": 0.18391288409744858,
+      "learning_rate": 5.708163896665708e-06,
+      "loss": 0.2845,
+      "step": 957
+    },
+    {
+      "epoch": 3.880506329113924,
+      "grad_norm": 0.20250440596573424,
+      "learning_rate": 5.668516443845047e-06,
+      "loss": 0.2804,
+      "step": 958
+    },
+    {
+      "epoch": 3.884556962025316,
+      "grad_norm": 0.18400503111848354,
+      "learning_rate": 5.6289844148227225e-06,
+      "loss": 0.286,
+      "step": 959
+    },
+    {
+      "epoch": 3.8886075949367087,
+      "grad_norm": 0.19279335800558178,
+      "learning_rate": 5.5895681279842615e-06,
+      "loss": 0.2963,
+      "step": 960
+    },
+    {
+      "epoch": 3.8926582278481012,
+      "grad_norm": 0.1911355756587035,
+      "learning_rate": 5.550267900783019e-06,
+      "loss": 0.2911,
+      "step": 961
+    },
+    {
+      "epoch": 3.8967088607594937,
+      "grad_norm": 0.18627172812702592,
+      "learning_rate": 5.511084049737623e-06,
+      "loss": 0.2882,
+      "step": 962
+    },
+    {
+      "epoch": 3.9007594936708863,
+      "grad_norm": 0.19594212803913766,
+      "learning_rate": 5.4720168904294215e-06,
+      "loss": 0.278,
+      "step": 963
+    },
+    {
+      "epoch": 3.9048101265822783,
+      "grad_norm": 0.19916518980341885,
+      "learning_rate": 5.433066737499948e-06,
+      "loss": 0.2965,
+      "step": 964
+    },
+    {
+      "epoch": 3.908860759493671,
+      "grad_norm": 0.2044302116861937,
+      "learning_rate": 5.394233904648376e-06,
+      "loss": 0.2934,
+      "step": 965
+    },
+    {
+      "epoch": 3.9129113924050634,
+      "grad_norm": 0.20286905376301975,
+      "learning_rate": 5.355518704628997e-06,
+      "loss": 0.2902,
+      "step": 966
+    },
+    {
+      "epoch": 3.9169620253164554,
+      "grad_norm": 0.17693954167503392,
+      "learning_rate": 5.316921449248731e-06,
+      "loss": 0.2737,
+      "step": 967
+    },
+    {
+      "epoch": 3.921012658227848,
+      "grad_norm": 0.1911978977825356,
+      "learning_rate": 5.278442449364538e-06,
+      "loss": 0.2868,
+      "step": 968
+    },
+    {
+      "epoch": 3.9250632911392405,
+      "grad_norm": 0.18830742585504048,
+      "learning_rate": 5.240082014881016e-06,
+      "loss": 0.2787,
+      "step": 969
+    },
+    {
+      "epoch": 3.929113924050633,
+      "grad_norm": 0.1878996818657122,
+      "learning_rate": 5.201840454747822e-06,
+      "loss": 0.286,
+      "step": 970
+    },
+    {
+      "epoch": 3.9331645569620255,
+      "grad_norm": 0.17179216946683623,
+      "learning_rate": 5.163718076957223e-06,
+      "loss": 0.2793,
+      "step": 971
+    },
+    {
+      "epoch": 3.9372151898734176,
+      "grad_norm": 0.17796474901406226,
+      "learning_rate": 5.125715188541609e-06,
+      "loss": 0.2834,
+      "step": 972
+    },
+    {
+      "epoch": 3.94126582278481,
+      "grad_norm": 0.18470197822815546,
+      "learning_rate": 5.087832095571021e-06,
+      "loss": 0.2923,
+      "step": 973
+    },
+    {
+      "epoch": 3.9453164556962026,
+      "grad_norm": 0.18595561154685533,
+      "learning_rate": 5.0500691031506766e-06,
+      "loss": 0.2777,
+      "step": 974
+    },
+    {
+      "epoch": 3.9493670886075947,
+      "grad_norm": 0.1997617625139002,
+      "learning_rate": 5.01242651541854e-06,
+      "loss": 0.2885,
+      "step": 975
+    },
+    {
+      "epoch": 3.953417721518987,
+      "grad_norm": 0.18070794835922735,
+      "learning_rate": 4.974904635542815e-06,
+      "loss": 0.2861,
+      "step": 976
+    },
+    {
+      "epoch": 3.9574683544303797,
+      "grad_norm": 0.17368031363690042,
+      "learning_rate": 4.937503765719582e-06,
+      "loss": 0.2879,
+      "step": 977
+    },
+    {
+      "epoch": 3.961518987341772,
+      "grad_norm": 0.18834305588323838,
+      "learning_rate": 4.900224207170299e-06,
+      "loss": 0.2824,
+      "step": 978
+    },
+    {
+      "epoch": 3.9655696202531647,
+      "grad_norm": 0.20120986719114786,
+      "learning_rate": 4.8630662601394065e-06,
+      "loss": 0.2823,
+      "step": 979
+    },
+    {
+      "epoch": 3.969620253164557,
+      "grad_norm": 0.19204434756657618,
+      "learning_rate": 4.8260302238918995e-06,
+      "loss": 0.2835,
+      "step": 980
+    },
+    {
+      "epoch": 3.9736708860759493,
+      "grad_norm": 0.178360716304687,
+      "learning_rate": 4.789116396710924e-06,
+      "loss": 0.2826,
+      "step": 981
+    },
+    {
+      "epoch": 3.977721518987342,
+      "grad_norm": 0.19149806338598108,
+      "learning_rate": 4.752325075895368e-06,
+      "loss": 0.2865,
+      "step": 982
+    },
+    {
+      "epoch": 3.981772151898734,
+      "grad_norm": 0.1881040725297849,
+      "learning_rate": 4.715656557757473e-06,
+      "loss": 0.2855,
+      "step": 983
+    },
+    {
+      "epoch": 3.9858227848101264,
+      "grad_norm": 0.19286988002795308,
+      "learning_rate": 4.679111137620442e-06,
+      "loss": 0.2857,
+      "step": 984
+    },
+    {
+      "epoch": 3.989873417721519,
+      "grad_norm": 0.17897801247564354,
+      "learning_rate": 4.6426891098160585e-06,
+      "loss": 0.2886,
+      "step": 985
+    },
+    {
+      "epoch": 3.9939240506329114,
+      "grad_norm": 0.17423859711689654,
+      "learning_rate": 4.6063907676823474e-06,
+      "loss": 0.2837,
+      "step": 986
+    },
+    {
+      "epoch": 3.997974683544304,
+      "grad_norm": 0.2032173833025341,
+      "learning_rate": 4.570216403561141e-06,
+      "loss": 0.2803,
+      "step": 987
+    },
+    {
+      "epoch": 4.002025316455696,
+      "grad_norm": 0.2263153915378573,
+      "learning_rate": 4.534166308795815e-06,
+      "loss": 0.2611,
+      "step": 988
+    },
+    {
+      "epoch": 4.006075949367089,
+      "grad_norm": 0.31584858763744017,
+      "learning_rate": 4.498240773728859e-06,
+      "loss": 0.263,
+      "step": 989
+    },
+    {
+      "epoch": 4.010126582278481,
+      "grad_norm": 0.23351836320939232,
+      "learning_rate": 4.462440087699609e-06,
+      "loss": 0.2523,
+      "step": 990
+    },
+    {
+      "epoch": 4.014177215189873,
+      "grad_norm": 0.3243120856448389,
+      "learning_rate": 4.426764539041861e-06,
+      "loss": 0.2512,
+      "step": 991
+    },
+    {
+      "epoch": 4.018227848101266,
+      "grad_norm": 0.3713878317863449,
+      "learning_rate": 4.391214415081582e-06,
+      "loss": 0.2443,
+      "step": 992
+    },
+    {
+      "epoch": 4.022278481012658,
+      "grad_norm": 0.23770467606063517,
+      "learning_rate": 4.355790002134579e-06,
+      "loss": 0.256,
+      "step": 993
+    },
+    {
+      "epoch": 4.02632911392405,
+      "grad_norm": 0.2935540933312601,
+      "learning_rate": 4.320491585504207e-06,
+      "loss": 0.2502,
+      "step": 994
+    },
+    {
+      "epoch": 4.030379746835443,
+      "grad_norm": 0.31406677710609227,
+      "learning_rate": 4.2853194494790615e-06,
+      "loss": 0.2572,
+      "step": 995
+    },
+    {
+      "epoch": 4.034430379746835,
+      "grad_norm": 0.26832976435469164,
+      "learning_rate": 4.250273877330691e-06,
+      "loss": 0.241,
+      "step": 996
+    },
+    {
+      "epoch": 4.038481012658228,
+      "grad_norm": 0.20641178947567068,
+      "learning_rate": 4.215355151311313e-06,
+      "loss": 0.2537,
+      "step": 997
+    },
+    {
+      "epoch": 4.04253164556962,
+      "grad_norm": 0.23429478203740672,
+      "learning_rate": 4.180563552651542e-06,
+      "loss": 0.2415,
+      "step": 998
+    },
+    {
+      "epoch": 4.046582278481012,
+      "grad_norm": 0.3085602921001155,
+      "learning_rate": 4.145899361558147e-06,
+      "loss": 0.2553,
+      "step": 999
+    },
+    {
+      "epoch": 4.050632911392405,
+      "grad_norm": 0.23262258913433967,
+      "learning_rate": 4.111362857211738e-06,
+      "loss": 0.2477,
+      "step": 1000
+    },
+    {
+      "epoch": 4.054683544303797,
+      "grad_norm": 0.21776482708293568,
+      "learning_rate": 4.076954317764592e-06,
+      "loss": 0.2504,
+      "step": 1001
+    },
+    {
+      "epoch": 4.0587341772151895,
+      "grad_norm": 0.20471831275099187,
+      "learning_rate": 4.042674020338335e-06,
+      "loss": 0.2462,
+      "step": 1002
+    },
+    {
+      "epoch": 4.062784810126582,
+      "grad_norm": 0.25047089766161834,
+      "learning_rate": 4.0085222410217835e-06,
+      "loss": 0.2583,
+      "step": 1003
+    },
+    {
+      "epoch": 4.0668354430379745,
+      "grad_norm": 0.21842245657289022,
+      "learning_rate": 3.974499254868674e-06,
+      "loss": 0.2526,
+      "step": 1004
+    },
+    {
+      "epoch": 4.0708860759493675,
+      "grad_norm": 0.23891187446673975,
+      "learning_rate": 3.940605335895451e-06,
+      "loss": 0.2527,
+      "step": 1005
+    },
+    {
+      "epoch": 4.0749367088607595,
+      "grad_norm": 0.2219146055227406,
+      "learning_rate": 3.90684075707908e-06,
+      "loss": 0.2624,
+      "step": 1006
+    },
+    {
+      "epoch": 4.078987341772152,
+      "grad_norm": 0.21086918415838868,
+      "learning_rate": 3.8732057903548505e-06,
+      "loss": 0.2539,
+      "step": 1007
+    },
+    {
+      "epoch": 4.083037974683545,
+      "grad_norm": 0.19395577025700936,
+      "learning_rate": 3.8397007066141375e-06,
+      "loss": 0.2427,
+      "step": 1008
+    },
+    {
+      "epoch": 4.087088607594937,
+      "grad_norm": 0.19464142097816337,
+      "learning_rate": 3.806325775702304e-06,
+      "loss": 0.2531,
+      "step": 1009
+    },
+    {
+      "epoch": 4.091139240506329,
+      "grad_norm": 0.18906252075844904,
+      "learning_rate": 3.773081266416434e-06,
+      "loss": 0.2527,
+      "step": 1010
+    },
+    {
+      "epoch": 4.095189873417722,
+      "grad_norm": 0.18832727891730447,
+      "learning_rate": 3.739967446503245e-06,
+      "loss": 0.2455,
+      "step": 1011
+    },
+    {
+      "epoch": 4.099240506329114,
+      "grad_norm": 0.19348478101696823,
+      "learning_rate": 3.706984582656894e-06,
+      "loss": 0.2622,
+      "step": 1012
+    },
+    {
+      "epoch": 4.103291139240507,
+      "grad_norm": 0.19478795094269585,
+      "learning_rate": 3.6741329405168237e-06,
+      "loss": 0.2551,
+      "step": 1013
+    },
+    {
+      "epoch": 4.107341772151899,
+      "grad_norm": 0.17754263244380156,
+      "learning_rate": 3.641412784665648e-06,
+      "loss": 0.243,
+      "step": 1014
+    },
+    {
+      "epoch": 4.111392405063291,
+      "grad_norm": 0.18691557224892097,
+      "learning_rate": 3.608824378627005e-06,
+      "loss": 0.2461,
+      "step": 1015
+    },
+    {
+      "epoch": 4.115443037974684,
+      "grad_norm": 0.18660537435565525,
+      "learning_rate": 3.5763679848634337e-06,
+      "loss": 0.2536,
+      "step": 1016
+    },
+    {
+      "epoch": 4.119493670886076,
+      "grad_norm": 0.17725074436387336,
+      "learning_rate": 3.544043864774269e-06,
+      "loss": 0.2455,
+      "step": 1017
+    },
+    {
+      "epoch": 4.123544303797468,
+      "grad_norm": 0.20052456724718365,
+      "learning_rate": 3.5118522786935282e-06,
+      "loss": 0.2505,
+      "step": 1018
+    },
+    {
+      "epoch": 4.127594936708861,
+      "grad_norm": 0.18009749404105796,
+      "learning_rate": 3.479793485887819e-06,
+      "loss": 0.2472,
+      "step": 1019
+    },
+    {
+      "epoch": 4.131645569620253,
+      "grad_norm": 0.17931565214519846,
+      "learning_rate": 3.4478677445542653e-06,
+      "loss": 0.259,
+      "step": 1020
+    },
+    {
+      "epoch": 4.135696202531646,
+      "grad_norm": 0.17920424298155543,
+      "learning_rate": 3.4160753118183767e-06,
+      "loss": 0.2594,
+      "step": 1021
+    },
+    {
+      "epoch": 4.139746835443038,
+      "grad_norm": 0.17840127680037474,
+      "learning_rate": 3.3844164437320527e-06,
+      "loss": 0.2407,
+      "step": 1022
+    },
+    {
+      "epoch": 4.14379746835443,
+      "grad_norm": 0.1812170831811252,
+      "learning_rate": 3.3528913952714558e-06,
+      "loss": 0.2535,
+      "step": 1023
+    },
+    {
+      "epoch": 4.147848101265823,
+      "grad_norm": 0.17547999450319718,
+      "learning_rate": 3.321500420335e-06,
+      "loss": 0.2519,
+      "step": 1024
+    },
+    {
+      "epoch": 4.151898734177215,
+      "grad_norm": 0.17977655286074598,
+      "learning_rate": 3.290243771741275e-06,
+      "loss": 0.2405,
+      "step": 1025
+    },
+    {
+      "epoch": 4.155949367088607,
+      "grad_norm": 0.17573789067624582,
+      "learning_rate": 3.2591217012270325e-06,
+      "loss": 0.2457,
+      "step": 1026
+    },
+    {
+      "epoch": 4.16,
+      "grad_norm": 0.18347038381469324,
+      "learning_rate": 3.228134459445149e-06,
+      "loss": 0.2531,
+      "step": 1027
+    },
+    {
+      "epoch": 4.164050632911392,
+      "grad_norm": 0.17371461814521033,
+      "learning_rate": 3.1972822959626205e-06,
+      "loss": 0.24,
+      "step": 1028
+    },
+    {
+      "epoch": 4.168101265822785,
+      "grad_norm": 0.18080274874643049,
+      "learning_rate": 3.166565459258513e-06,
+      "loss": 0.2427,
+      "step": 1029
+    },
+    {
+      "epoch": 4.172151898734177,
+      "grad_norm": 0.1782615514517165,
+      "learning_rate": 3.1359841967220193e-06,
+      "loss": 0.2566,
+      "step": 1030
+    },
+    {
+      "epoch": 4.176202531645569,
+      "grad_norm": 0.1757368072718625,
+      "learning_rate": 3.105538754650419e-06,
+      "loss": 0.2483,
+      "step": 1031
+    },
+    {
+      "epoch": 4.180253164556962,
+      "grad_norm": 0.17516592649488116,
+      "learning_rate": 3.07522937824712e-06,
+      "loss": 0.2448,
+      "step": 1032
+    },
+    {
+      "epoch": 4.184303797468354,
+      "grad_norm": 0.1788317882667316,
+      "learning_rate": 3.0450563116196697e-06,
+      "loss": 0.2434,
+      "step": 1033
+    },
+    {
+      "epoch": 4.188354430379746,
+      "grad_norm": 0.16851405632701066,
+      "learning_rate": 3.0150197977778008e-06,
+      "loss": 0.2525,
+      "step": 1034
+    },
+    {
+      "epoch": 4.192405063291139,
+      "grad_norm": 0.17794258130467738,
+      "learning_rate": 2.985120078631465e-06,
+      "loss": 0.2535,
+      "step": 1035
+    },
+    {
+      "epoch": 4.1964556962025314,
+      "grad_norm": 0.1755874444295416,
+      "learning_rate": 2.9553573949888893e-06,
+      "loss": 0.2446,
+      "step": 1036
+    },
+    {
+      "epoch": 4.200506329113924,
+      "grad_norm": 0.17445950627173526,
+      "learning_rate": 2.9257319865546384e-06,
+      "loss": 0.2563,
+      "step": 1037
+    },
+    {
+      "epoch": 4.2045569620253165,
+      "grad_norm": 0.1736325227893343,
+      "learning_rate": 2.896244091927678e-06,
+      "loss": 0.2534,
+      "step": 1038
+    },
+    {
+      "epoch": 4.2086075949367086,
+      "grad_norm": 0.16756205655473083,
+      "learning_rate": 2.8668939485994584e-06,
+      "loss": 0.2503,
+      "step": 1039
+    },
+    {
+      "epoch": 4.2126582278481015,
+      "grad_norm": 0.17550267072150952,
+      "learning_rate": 2.837681792951994e-06,
+      "loss": 0.2422,
+      "step": 1040
+    },
+    {
+      "epoch": 4.216708860759494,
+      "grad_norm": 0.1679711741883475,
+      "learning_rate": 2.808607860255981e-06,
+      "loss": 0.251,
+      "step": 1041
+    },
+    {
+      "epoch": 4.220759493670886,
+      "grad_norm": 0.17734826337598078,
+      "learning_rate": 2.7796723846688634e-06,
+      "loss": 0.2487,
+      "step": 1042
+    },
+    {
+      "epoch": 4.224810126582279,
+      "grad_norm": 0.1748836987034128,
+      "learning_rate": 2.7508755992329937e-06,
+      "loss": 0.252,
+      "step": 1043
+    },
+    {
+      "epoch": 4.228860759493671,
+      "grad_norm": 0.17178645686366356,
+      "learning_rate": 2.722217735873718e-06,
+      "loss": 0.2394,
+      "step": 1044
+    },
+    {
+      "epoch": 4.232911392405064,
+      "grad_norm": 0.1702143051865863,
+      "learning_rate": 2.6936990253975315e-06,
+      "loss": 0.2441,
+      "step": 1045
+    },
+    {
+      "epoch": 4.236962025316456,
+      "grad_norm": 0.17677639524151428,
+      "learning_rate": 2.665319697490205e-06,
+      "loss": 0.2624,
+      "step": 1046
+    },
+    {
+      "epoch": 4.241012658227848,
+      "grad_norm": 0.16989287266247702,
+      "learning_rate": 2.637079980714945e-06,
+      "loss": 0.2406,
+      "step": 1047
+    },
+    {
+      "epoch": 4.245063291139241,
+      "grad_norm": 0.1741486865182564,
+      "learning_rate": 2.6089801025105453e-06,
+      "loss": 0.2445,
+      "step": 1048
+    },
+    {
+      "epoch": 4.249113924050633,
+      "grad_norm": 0.16684030122550814,
+      "learning_rate": 2.581020289189571e-06,
+      "loss": 0.2436,
+      "step": 1049
+    },
+    {
+      "epoch": 4.253164556962025,
+      "grad_norm": 0.1721115191752493,
+      "learning_rate": 2.553200765936501e-06,
+      "loss": 0.2394,
+      "step": 1050
+    },
+    {
+      "epoch": 4.257215189873418,
+      "grad_norm": 0.17416724956943397,
+      "learning_rate": 2.525521756805962e-06,
+      "loss": 0.2483,
+      "step": 1051
+    },
+    {
+      "epoch": 4.26126582278481,
+      "grad_norm": 0.1710610465214727,
+      "learning_rate": 2.497983484720885e-06,
+      "loss": 0.2501,
+      "step": 1052
+    },
+    {
+      "epoch": 4.265316455696203,
+      "grad_norm": 0.17087839577889113,
+      "learning_rate": 2.470586171470728e-06,
+      "loss": 0.2494,
+      "step": 1053
+    },
+    {
+      "epoch": 4.269367088607595,
+      "grad_norm": 0.17132857369057308,
+      "learning_rate": 2.4433300377096836e-06,
+      "loss": 0.2508,
+      "step": 1054
+    },
+    {
+      "epoch": 4.273417721518987,
+      "grad_norm": 0.17485738291098674,
+      "learning_rate": 2.4162153029549073e-06,
+      "loss": 0.2557,
+      "step": 1055
+    },
+    {
+      "epoch": 4.27746835443038,
+      "grad_norm": 0.17256987359381804,
+      "learning_rate": 2.3892421855847458e-06,
+      "loss": 0.2532,
+      "step": 1056
+    },
+    {
+      "epoch": 4.281518987341772,
+      "grad_norm": 0.1740955994225746,
+      "learning_rate": 2.362410902836978e-06,
+      "loss": 0.249,
+      "step": 1057
+    },
+    {
+      "epoch": 4.285569620253165,
+      "grad_norm": 0.17058381567370362,
+      "learning_rate": 2.3357216708070653e-06,
+      "loss": 0.2533,
+      "step": 1058
+    },
+    {
+      "epoch": 4.289620253164557,
+      "grad_norm": 0.17109943485495982,
+      "learning_rate": 2.309174704446411e-06,
+      "loss": 0.247,
+      "step": 1059
+    },
+    {
+      "epoch": 4.293670886075949,
+      "grad_norm": 0.1806830559958293,
+      "learning_rate": 2.2827702175606437e-06,
+      "loss": 0.246,
+      "step": 1060
+    },
+    {
+      "epoch": 4.297721518987342,
+      "grad_norm": 0.1725895703643416,
+      "learning_rate": 2.256508422807855e-06,
+      "loss": 0.2537,
+      "step": 1061
+    },
+    {
+      "epoch": 4.301772151898734,
+      "grad_norm": 0.17211252500508203,
+      "learning_rate": 2.230389531696946e-06,
+      "loss": 0.2513,
+      "step": 1062
+    },
+    {
+      "epoch": 4.305822784810126,
+      "grad_norm": 0.16844048491708458,
+      "learning_rate": 2.204413754585857e-06,
+      "loss": 0.2454,
+      "step": 1063
+    },
+    {
+      "epoch": 4.309873417721519,
+      "grad_norm": 0.17404970491305088,
+      "learning_rate": 2.1785813006799406e-06,
+      "loss": 0.243,
+      "step": 1064
+    },
+    {
+      "epoch": 4.313924050632911,
+      "grad_norm": 0.16815793543000698,
+      "learning_rate": 2.1528923780302224e-06,
+      "loss": 0.2477,
+      "step": 1065
+    },
+    {
+      "epoch": 4.317974683544303,
+      "grad_norm": 0.17430018406292555,
+      "learning_rate": 2.127347193531757e-06,
+      "loss": 0.2479,
+      "step": 1066
+    },
+    {
+      "epoch": 4.322025316455696,
+      "grad_norm": 0.18369666576355873,
+      "learning_rate": 2.101945952921942e-06,
+      "loss": 0.2513,
+      "step": 1067
+    },
+    {
+      "epoch": 4.326075949367088,
+      "grad_norm": 0.16817270858830236,
+      "learning_rate": 2.0766888607788906e-06,
+      "loss": 0.2556,
+      "step": 1068
+    },
+    {
+      "epoch": 4.330126582278481,
+      "grad_norm": 0.17050523433957276,
+      "learning_rate": 2.0515761205197337e-06,
+      "loss": 0.2449,
+      "step": 1069
+    },
+    {
+      "epoch": 4.334177215189873,
+      "grad_norm": 0.17012827316534923,
+      "learning_rate": 2.0266079343990453e-06,
+      "loss": 0.2529,
+      "step": 1070
+    },
+    {
+      "epoch": 4.3382278481012655,
+      "grad_norm": 0.16943554627043486,
+      "learning_rate": 2.0017845035071494e-06,
+      "loss": 0.2436,
+      "step": 1071
+    },
+    {
+      "epoch": 4.3422784810126585,
+      "grad_norm": 0.17369694798154178,
+      "learning_rate": 1.9771060277685537e-06,
+      "loss": 0.246,
+      "step": 1072
+    },
+    {
+      "epoch": 4.3463291139240505,
+      "grad_norm": 0.1691729020133195,
+      "learning_rate": 1.95257270594031e-06,
+      "loss": 0.2502,
+      "step": 1073
+    },
+    {
+      "epoch": 4.3503797468354435,
+      "grad_norm": 0.1751773192428585,
+      "learning_rate": 1.9281847356104188e-06,
+      "loss": 0.2605,
+      "step": 1074
+    },
+    {
+      "epoch": 4.3544303797468356,
+      "grad_norm": 0.1737913343607389,
+      "learning_rate": 1.9039423131962365e-06,
+      "loss": 0.2507,
+      "step": 1075
+    },
+    {
+      "epoch": 4.358481012658228,
+      "grad_norm": 0.1686141668011873,
+      "learning_rate": 1.8798456339429027e-06,
+      "loss": 0.2444,
+      "step": 1076
+    },
+    {
+      "epoch": 4.362531645569621,
+      "grad_norm": 0.1738378884259171,
+      "learning_rate": 1.8558948919217612e-06,
+      "loss": 0.2583,
+      "step": 1077
+    },
+    {
+      "epoch": 4.366582278481013,
+      "grad_norm": 0.17309076185558958,
+      "learning_rate": 1.8320902800287954e-06,
+      "loss": 0.2493,
+      "step": 1078
+    },
+    {
+      "epoch": 4.370632911392405,
+      "grad_norm": 0.1722717949926062,
+      "learning_rate": 1.8084319899830726e-06,
+      "loss": 0.2499,
+      "step": 1079
+    },
+    {
+      "epoch": 4.374683544303798,
+      "grad_norm": 0.1655092775122385,
+      "learning_rate": 1.7849202123252097e-06,
+      "loss": 0.2549,
+      "step": 1080
+    },
+    {
+      "epoch": 4.37873417721519,
+      "grad_norm": 0.17441698353205282,
+      "learning_rate": 1.7615551364158401e-06,
+      "loss": 0.2534,
+      "step": 1081
+    },
+    {
+      "epoch": 4.382784810126582,
+      "grad_norm": 0.16784647037187556,
+      "learning_rate": 1.738336950434061e-06,
+      "loss": 0.2544,
+      "step": 1082
+    },
+    {
+      "epoch": 4.386835443037975,
+      "grad_norm": 0.1721801600785089,
+      "learning_rate": 1.715265841375957e-06,
+      "loss": 0.25,
+      "step": 1083
+    },
+    {
+      "epoch": 4.390886075949367,
+      "grad_norm": 0.17392834976103722,
+      "learning_rate": 1.6923419950530684e-06,
+      "loss": 0.2522,
+      "step": 1084
+    },
+    {
+      "epoch": 4.39493670886076,
+      "grad_norm": 0.16558903631857938,
+      "learning_rate": 1.6695655960909008e-06,
+      "loss": 0.2596,
+      "step": 1085
+    },
+    {
+      "epoch": 4.398987341772152,
+      "grad_norm": 0.1657825708678759,
+      "learning_rate": 1.646936827927441e-06,
+      "loss": 0.2513,
+      "step": 1086
+    },
+    {
+      "epoch": 4.403037974683544,
+      "grad_norm": 0.1729835376538812,
+      "learning_rate": 1.6244558728116766e-06,
+      "loss": 0.2569,
+      "step": 1087
+    },
+    {
+      "epoch": 4.407088607594937,
+      "grad_norm": 0.17113816274857319,
+      "learning_rate": 1.6021229118021265e-06,
+      "loss": 0.2504,
+      "step": 1088
+    },
+    {
+      "epoch": 4.411139240506329,
+      "grad_norm": 0.16543320288567653,
+      "learning_rate": 1.5799381247653967e-06,
+      "loss": 0.249,
+      "step": 1089
+    },
+    {
+      "epoch": 4.415189873417722,
+      "grad_norm": 0.17263224702433796,
+      "learning_rate": 1.5579016903747013e-06,
+      "loss": 0.2428,
+      "step": 1090
+    },
+    {
+      "epoch": 4.419240506329114,
+      "grad_norm": 0.17912003703236856,
+      "learning_rate": 1.5360137861084656e-06,
+      "loss": 0.2476,
+      "step": 1091
+    },
+    {
+      "epoch": 4.423291139240506,
+      "grad_norm": 0.1695803814493009,
+      "learning_rate": 1.5142745882488475e-06,
+      "loss": 0.2521,
+      "step": 1092
+    },
+    {
+      "epoch": 4.427341772151899,
+      "grad_norm": 0.16945201316461697,
+      "learning_rate": 1.4926842718803691e-06,
+      "loss": 0.2527,
+      "step": 1093
+    },
+    {
+      "epoch": 4.431392405063291,
+      "grad_norm": 0.17721525134215205,
+      "learning_rate": 1.4712430108884657e-06,
+      "loss": 0.2531,
+      "step": 1094
+    },
+    {
+      "epoch": 4.435443037974683,
+      "grad_norm": 0.16719727888585548,
+      "learning_rate": 1.4499509779581078e-06,
+      "loss": 0.2561,
+      "step": 1095
+    },
+    {
+      "epoch": 4.439493670886076,
+      "grad_norm": 0.16963753672759757,
+      "learning_rate": 1.4288083445723988e-06,
+      "loss": 0.247,
+      "step": 1096
+    },
+    {
+      "epoch": 4.443544303797468,
+      "grad_norm": 0.16680066563387674,
+      "learning_rate": 1.4078152810112045e-06,
+      "loss": 0.2456,
+      "step": 1097
+    },
+    {
+      "epoch": 4.44759493670886,
+      "grad_norm": 0.1696786044538861,
+      "learning_rate": 1.3869719563497697e-06,
+      "loss": 0.252,
+      "step": 1098
+    },
+    {
+      "epoch": 4.451645569620253,
+      "grad_norm": 0.1776657586508514,
+      "learning_rate": 1.3662785384573663e-06,
+      "loss": 0.2604,
+      "step": 1099
+    },
+    {
+      "epoch": 4.455696202531645,
+      "grad_norm": 0.17277990912479196,
+      "learning_rate": 1.3457351939959383e-06,
+      "loss": 0.2569,
+      "step": 1100
+    },
+    {
+      "epoch": 4.459746835443038,
+      "grad_norm": 0.16771029816258448,
+      "learning_rate": 1.3253420884187551e-06,
+      "loss": 0.255,
+      "step": 1101
+    },
+    {
+      "epoch": 4.46379746835443,
+      "grad_norm": 0.16222252828049355,
+      "learning_rate": 1.3050993859690953e-06,
+      "loss": 0.2393,
+      "step": 1102
+    },
+    {
+      "epoch": 4.467848101265822,
+      "grad_norm": 0.17658549742707605,
+      "learning_rate": 1.2850072496788869e-06,
+      "loss": 0.248,
+      "step": 1103
+    },
+    {
+      "epoch": 4.471898734177215,
+      "grad_norm": 0.17129487232365112,
+      "learning_rate": 1.2650658413674434e-06,
+      "loss": 0.2483,
+      "step": 1104
+    },
+    {
+      "epoch": 4.4759493670886075,
+      "grad_norm": 0.1670801282827011,
+      "learning_rate": 1.2452753216401226e-06,
+      "loss": 0.2486,
+      "step": 1105
+    },
+    {
+      "epoch": 4.48,
+      "grad_norm": 0.16985283700182455,
+      "learning_rate": 1.2256358498870503e-06,
+      "loss": 0.2452,
+      "step": 1106
+    },
+    {
+      "epoch": 4.4840506329113925,
+      "grad_norm": 0.16793956168654506,
+      "learning_rate": 1.2061475842818337e-06,
+      "loss": 0.2413,
+      "step": 1107
+    },
+    {
+      "epoch": 4.488101265822785,
+      "grad_norm": 0.17342921031311534,
+      "learning_rate": 1.1868106817802816e-06,
+      "loss": 0.254,
+      "step": 1108
+    },
+    {
+      "epoch": 4.4921518987341775,
+      "grad_norm": 0.17135084627915032,
+      "learning_rate": 1.1676252981191482e-06,
+      "loss": 0.2544,
+      "step": 1109
+    },
+    {
+      "epoch": 4.49620253164557,
+      "grad_norm": 0.16627612371510261,
+      "learning_rate": 1.1485915878148823e-06,
+      "loss": 0.2603,
+      "step": 1110
+    },
+    {
+      "epoch": 4.500253164556962,
+      "grad_norm": 0.17407581938264088,
+      "learning_rate": 1.1297097041623584e-06,
+      "loss": 0.2462,
+      "step": 1111
+    },
+    {
+      "epoch": 4.504303797468355,
+      "grad_norm": 0.16116005494356683,
+      "learning_rate": 1.1109797992336847e-06,
+      "loss": 0.2442,
+      "step": 1112
+    },
+    {
+      "epoch": 4.508354430379747,
+      "grad_norm": 0.16751254661097376,
+      "learning_rate": 1.092402023876933e-06,
+      "loss": 0.2571,
+      "step": 1113
+    },
+    {
+      "epoch": 4.512405063291139,
+      "grad_norm": 0.1658412390766363,
+      "learning_rate": 1.0739765277149527e-06,
+      "loss": 0.2402,
+      "step": 1114
+    },
+    {
+      "epoch": 4.516455696202532,
+      "grad_norm": 0.16954621827092656,
+      "learning_rate": 1.0557034591441596e-06,
+      "loss": 0.2564,
+      "step": 1115
+    },
+    {
+      "epoch": 4.520506329113924,
+      "grad_norm": 0.1632470904122153,
+      "learning_rate": 1.0375829653333324e-06,
+      "loss": 0.251,
+      "step": 1116
+    },
+    {
+      "epoch": 4.524556962025317,
+      "grad_norm": 0.16362910871881894,
+      "learning_rate": 1.0196151922224385e-06,
+      "loss": 0.2473,
+      "step": 1117
+    },
+    {
+      "epoch": 4.528607594936709,
+      "grad_norm": 0.16423474825556725,
+      "learning_rate": 1.0018002845214526e-06,
+      "loss": 0.2348,
+      "step": 1118
+    },
+    {
+      "epoch": 4.532658227848101,
+      "grad_norm": 0.18478025233580314,
+      "learning_rate": 9.841383857091947e-07,
+      "loss": 0.262,
+      "step": 1119
+    },
+    {
+      "epoch": 4.536708860759494,
+      "grad_norm": 0.16893210627197114,
+      "learning_rate": 9.666296380321616e-07,
+      "loss": 0.2406,
+      "step": 1120
+    },
+    {
+      "epoch": 4.540759493670886,
+      "grad_norm": 0.1742477115459927,
+      "learning_rate": 9.492741825034124e-07,
+      "loss": 0.255,
+      "step": 1121
+    },
+    {
+      "epoch": 4.544810126582279,
+      "grad_norm": 0.1677994278302905,
+      "learning_rate": 9.320721589013892e-07,
+      "loss": 0.2514,
+      "step": 1122
+    },
+    {
+      "epoch": 4.548860759493671,
+      "grad_norm": 0.16606363152549716,
+      "learning_rate": 9.150237057688339e-07,
+      "loss": 0.25,
+      "step": 1123
+    },
+    {
+      "epoch": 4.552911392405063,
+      "grad_norm": 0.18135893807626008,
+      "learning_rate": 8.981289604116328e-07,
+      "loss": 0.2559,
+      "step": 1124
+    },
+    {
+      "epoch": 4.556962025316456,
+      "grad_norm": 0.16749402657225154,
+      "learning_rate": 8.813880588977542e-07,
+      "loss": 0.2448,
+      "step": 1125
+    },
+    {
+      "epoch": 4.561012658227848,
+      "grad_norm": 0.17144527027539155,
+      "learning_rate": 8.648011360561126e-07,
+      "loss": 0.2563,
+      "step": 1126
+    },
+    {
+      "epoch": 4.56506329113924,
+      "grad_norm": 0.16313920792116593,
+      "learning_rate": 8.483683254755037e-07,
+      "loss": 0.2538,
+      "step": 1127
+    },
+    {
+      "epoch": 4.569113924050633,
+      "grad_norm": 0.16645601337084526,
+      "learning_rate": 8.320897595035227e-07,
+      "loss": 0.2425,
+      "step": 1128
+    },
+    {
+      "epoch": 4.573164556962025,
+      "grad_norm": 0.16069493411630026,
+      "learning_rate": 8.159655692455093e-07,
+      "loss": 0.2587,
+      "step": 1129
+    },
+    {
+      "epoch": 4.577215189873417,
+      "grad_norm": 0.17225133632002051,
+      "learning_rate": 7.999958845634648e-07,
+      "loss": 0.248,
+      "step": 1130
+    },
+    {
+      "epoch": 4.58126582278481,
+      "grad_norm": 0.16760416287670166,
+      "learning_rate": 7.841808340750478e-07,
+      "loss": 0.2515,
+      "step": 1131
+    },
+    {
+      "epoch": 4.585316455696202,
+      "grad_norm": 0.16873830616922394,
+      "learning_rate": 7.685205451524869e-07,
+      "loss": 0.2558,
+      "step": 1132
+    },
+    {
+      "epoch": 4.589367088607595,
+      "grad_norm": 0.1672301403181721,
+      "learning_rate": 7.530151439216027e-07,
+      "loss": 0.2574,
+      "step": 1133
+    },
+    {
+      "epoch": 4.593417721518987,
+      "grad_norm": 0.169940515266237,
+      "learning_rate": 7.376647552607675e-07,
+      "loss": 0.2527,
+      "step": 1134
+    },
+    {
+      "epoch": 4.597468354430379,
+      "grad_norm": 0.16989201957826824,
+      "learning_rate": 7.224695027998963e-07,
+      "loss": 0.243,
+      "step": 1135
+    },
+    {
+      "epoch": 4.601518987341772,
+      "grad_norm": 0.17577718142674498,
+      "learning_rate": 7.07429508919466e-07,
+      "loss": 0.2587,
+      "step": 1136
+    },
+    {
+      "epoch": 4.605569620253164,
+      "grad_norm": 0.16852819915217243,
+      "learning_rate": 6.925448947495206e-07,
+      "loss": 0.25,
+      "step": 1137
+    },
+    {
+      "epoch": 4.609620253164557,
+      "grad_norm": 0.16596392814771777,
+      "learning_rate": 6.778157801686936e-07,
+      "loss": 0.2515,
+      "step": 1138
+    },
+    {
+      "epoch": 4.613670886075949,
+      "grad_norm": 0.16307939338977837,
+      "learning_rate": 6.632422838032515e-07,
+      "loss": 0.255,
+      "step": 1139
+    },
+    {
+      "epoch": 4.6177215189873415,
+      "grad_norm": 0.16608076304551733,
+      "learning_rate": 6.488245230261281e-07,
+      "loss": 0.2496,
+      "step": 1140
+    },
+    {
+      "epoch": 4.6217721518987345,
+      "grad_norm": 0.16634009454075038,
+      "learning_rate": 6.345626139559868e-07,
+      "loss": 0.2494,
+      "step": 1141
+    },
+    {
+      "epoch": 4.6258227848101265,
+      "grad_norm": 0.1665720394938799,
+      "learning_rate": 6.204566714562866e-07,
+      "loss": 0.246,
+      "step": 1142
+    },
+    {
+      "epoch": 4.629873417721519,
+      "grad_norm": 0.16578369624744313,
+      "learning_rate": 6.06506809134344e-07,
+      "loss": 0.2445,
+      "step": 1143
+    },
+    {
+      "epoch": 4.633924050632912,
+      "grad_norm": 0.17239972270813506,
+      "learning_rate": 5.927131393404373e-07,
+      "loss": 0.2438,
+      "step": 1144
+    },
+    {
+      "epoch": 4.637974683544304,
+      "grad_norm": 0.17147582766920183,
+      "learning_rate": 5.790757731668817e-07,
+      "loss": 0.2516,
+      "step": 1145
+    },
+    {
+      "epoch": 4.642025316455696,
+      "grad_norm": 0.16173680971600768,
+      "learning_rate": 5.655948204471507e-07,
+      "loss": 0.2478,
+      "step": 1146
+    },
+    {
+      "epoch": 4.646075949367089,
+      "grad_norm": 0.16633364142418425,
+      "learning_rate": 5.522703897549875e-07,
+      "loss": 0.2521,
+      "step": 1147
+    },
+    {
+      "epoch": 4.650126582278481,
+      "grad_norm": 0.17296822664858055,
+      "learning_rate": 5.391025884035239e-07,
+      "loss": 0.2618,
+      "step": 1148
+    },
+    {
+      "epoch": 4.654177215189874,
+      "grad_norm": 0.1673143454059018,
+      "learning_rate": 5.260915224444207e-07,
+      "loss": 0.2424,
+      "step": 1149
+    },
+    {
+      "epoch": 4.658227848101266,
+      "grad_norm": 0.16009663615480502,
+      "learning_rate": 5.132372966670129e-07,
+      "loss": 0.2511,
+      "step": 1150
+    },
+    {
+      "epoch": 4.662278481012658,
+      "grad_norm": 0.16529830436498294,
+      "learning_rate": 5.005400145974704e-07,
+      "loss": 0.2549,
+      "step": 1151
+    },
+    {
+      "epoch": 4.666329113924051,
+      "grad_norm": 0.16530722315382787,
+      "learning_rate": 4.879997784979562e-07,
+      "loss": 0.2427,
+      "step": 1152
+    },
+    {
+      "epoch": 4.670379746835443,
+      "grad_norm": 0.16615300846815545,
+      "learning_rate": 4.7561668936580984e-07,
+      "loss": 0.2553,
+      "step": 1153
+    },
+    {
+      "epoch": 4.674430379746836,
+      "grad_norm": 0.1688395763670177,
+      "learning_rate": 4.6339084693272306e-07,
+      "loss": 0.2467,
+      "step": 1154
+    },
+    {
+      "epoch": 4.678481012658228,
+      "grad_norm": 0.17042758356862503,
+      "learning_rate": 4.5132234966395847e-07,
+      "loss": 0.2505,
+      "step": 1155
+    },
+    {
+      "epoch": 4.68253164556962,
+      "grad_norm": 0.17182685741614245,
+      "learning_rate": 4.3941129475752795e-07,
+      "loss": 0.2502,
+      "step": 1156
+    },
+    {
+      "epoch": 4.686582278481013,
+      "grad_norm": 0.16312043217450334,
+      "learning_rate": 4.27657778143431e-07,
+      "loss": 0.2491,
+      "step": 1157
+    },
+    {
+      "epoch": 4.690632911392405,
+      "grad_norm": 0.16178869102882665,
+      "learning_rate": 4.1606189448287757e-07,
+      "loss": 0.2493,
+      "step": 1158
+    },
+    {
+      "epoch": 4.694683544303797,
+      "grad_norm": 0.16354310853579188,
+      "learning_rate": 4.046237371675177e-07,
+      "loss": 0.2492,
+      "step": 1159
+    },
+    {
+      "epoch": 4.69873417721519,
+      "grad_norm": 0.1622168025960343,
+      "learning_rate": 3.9334339831869963e-07,
+      "loss": 0.2461,
+      "step": 1160
+    },
+    {
+      "epoch": 4.702784810126582,
+      "grad_norm": 0.16153872967330254,
+      "learning_rate": 3.8222096878671955e-07,
+      "loss": 0.2413,
+      "step": 1161
+    },
+    {
+      "epoch": 4.706835443037974,
+      "grad_norm": 0.17650341529482033,
+      "learning_rate": 3.7125653815009545e-07,
+      "loss": 0.2582,
+      "step": 1162
+    },
+    {
+      "epoch": 4.710886075949367,
+      "grad_norm": 0.1644904512135916,
+      "learning_rate": 3.6045019471484974e-07,
+      "loss": 0.2466,
+      "step": 1163
+    },
+    {
+      "epoch": 4.714936708860759,
+      "grad_norm": 0.15761316697304179,
+      "learning_rate": 3.498020255137813e-07,
+      "loss": 0.2597,
+      "step": 1164
+    },
+    {
+      "epoch": 4.718987341772152,
+      "grad_norm": 0.1656561359925105,
+      "learning_rate": 3.393121163057811e-07,
+      "loss": 0.2516,
+      "step": 1165
+    },
+    {
+      "epoch": 4.723037974683544,
+      "grad_norm": 0.16324288399656797,
+      "learning_rate": 3.289805515751399e-07,
+      "loss": 0.2455,
+      "step": 1166
+    },
+    {
+      "epoch": 4.727088607594936,
+      "grad_norm": 0.16161038697618624,
+      "learning_rate": 3.188074145308573e-07,
+      "loss": 0.2536,
+      "step": 1167
+    },
+    {
+      "epoch": 4.731139240506329,
+      "grad_norm": 0.16043300980767636,
+      "learning_rate": 3.087927871059804e-07,
+      "loss": 0.2445,
+      "step": 1168
+    },
+    {
+      "epoch": 4.735189873417721,
+      "grad_norm": 0.16935571156832005,
+      "learning_rate": 2.989367499569418e-07,
+      "loss": 0.253,
+      "step": 1169
+    },
+    {
+      "epoch": 4.739240506329114,
+      "grad_norm": 0.1651959570947618,
+      "learning_rate": 2.8923938246290917e-07,
+      "loss": 0.2451,
+      "step": 1170
+    },
+    {
+      "epoch": 4.743291139240506,
+      "grad_norm": 0.17527231000836851,
+      "learning_rate": 2.7970076272514804e-07,
+      "loss": 0.254,
+      "step": 1171
+    },
+    {
+      "epoch": 4.7473417721518985,
+      "grad_norm": 0.16978492994359695,
+      "learning_rate": 2.703209675663887e-07,
+      "loss": 0.2469,
+      "step": 1172
+    },
+    {
+      "epoch": 4.751392405063291,
+      "grad_norm": 0.16871258427268412,
+      "learning_rate": 2.6110007253021374e-07,
+      "loss": 0.2495,
+      "step": 1173
+    },
+    {
+      "epoch": 4.7554430379746835,
+      "grad_norm": 0.1624947855061142,
+      "learning_rate": 2.520381518804471e-07,
+      "loss": 0.2533,
+      "step": 1174
+    },
+    {
+      "epoch": 4.759493670886076,
+      "grad_norm": 0.16090789876396913,
+      "learning_rate": 2.4313527860054585e-07,
+      "loss": 0.2506,
+      "step": 1175
+    },
+    {
+      "epoch": 4.7635443037974685,
+      "grad_norm": 0.16696961878883163,
+      "learning_rate": 2.343915243930317e-07,
+      "loss": 0.2527,
+      "step": 1176
+    },
+    {
+      "epoch": 4.767594936708861,
+      "grad_norm": 0.16068057613216374,
+      "learning_rate": 2.2580695967889367e-07,
+      "loss": 0.2426,
+      "step": 1177
+    },
+    {
+      "epoch": 4.7716455696202535,
+      "grad_norm": 0.16057015524131257,
+      "learning_rate": 2.1738165359704189e-07,
+      "loss": 0.2439,
+      "step": 1178
+    },
+    {
+      "epoch": 4.775696202531646,
+      "grad_norm": 0.16292556187362223,
+      "learning_rate": 2.0911567400373257e-07,
+      "loss": 0.242,
+      "step": 1179
+    },
+    {
+      "epoch": 4.779746835443038,
+      "grad_norm": 0.16363304186308875,
+      "learning_rate": 2.0100908747202607e-07,
+      "loss": 0.2368,
+      "step": 1180
+    },
+    {
+      "epoch": 4.783797468354431,
+      "grad_norm": 0.16186143370876993,
+      "learning_rate": 1.9306195929125638e-07,
+      "loss": 0.2533,
+      "step": 1181
+    },
+    {
+      "epoch": 4.787848101265823,
+      "grad_norm": 0.17052727952369665,
+      "learning_rate": 1.8527435346650247e-07,
+      "loss": 0.2488,
+      "step": 1182
+    },
+    {
+      "epoch": 4.791898734177215,
+      "grad_norm": 0.16926012973854657,
+      "learning_rate": 1.7764633271807108e-07,
+      "loss": 0.2573,
+      "step": 1183
+    },
+    {
+      "epoch": 4.795949367088608,
+      "grad_norm": 0.16381856510572526,
+      "learning_rate": 1.7017795848099262e-07,
+      "loss": 0.2525,
+      "step": 1184
+    },
+    {
+      "epoch": 4.8,
+      "grad_norm": 0.16381511380849378,
+      "learning_rate": 1.6286929090452596e-07,
+      "loss": 0.2479,
+      "step": 1185
+    },
+    {
+      "epoch": 4.804050632911393,
+      "grad_norm": 0.16417805919120684,
+      "learning_rate": 1.557203888516745e-07,
+      "loss": 0.2491,
+      "step": 1186
+    },
+    {
+      "epoch": 4.808101265822785,
+      "grad_norm": 0.1658752667942242,
+      "learning_rate": 1.487313098987131e-07,
+      "loss": 0.2487,
+      "step": 1187
+    },
+    {
+      "epoch": 4.812151898734177,
+      "grad_norm": 0.16213069399405833,
+      "learning_rate": 1.4190211033472402e-07,
+      "loss": 0.2444,
+      "step": 1188
+    },
+    {
+      "epoch": 4.81620253164557,
+      "grad_norm": 0.1642797086222475,
+      "learning_rate": 1.3523284516113955e-07,
+      "loss": 0.2508,
+      "step": 1189
+    },
+    {
+      "epoch": 4.820253164556962,
+      "grad_norm": 0.15885744637580262,
+      "learning_rate": 1.2872356809130682e-07,
+      "loss": 0.2527,
+      "step": 1190
+    },
+    {
+      "epoch": 4.824303797468355,
+      "grad_norm": 0.16447143808442186,
+      "learning_rate": 1.2237433155004807e-07,
+      "loss": 0.2556,
+      "step": 1191
+    },
+    {
+      "epoch": 4.828354430379747,
+      "grad_norm": 0.1617548028314596,
+      "learning_rate": 1.1618518667323886e-07,
+      "loss": 0.2441,
+      "step": 1192
+    },
+    {
+      "epoch": 4.832405063291139,
+      "grad_norm": 0.16680417461486297,
+      "learning_rate": 1.1015618330740385e-07,
+      "loss": 0.252,
+      "step": 1193
+    },
+    {
+      "epoch": 4.836455696202532,
+      "grad_norm": 0.15860488742991688,
+      "learning_rate": 1.042873700093061e-07,
+      "loss": 0.2526,
+      "step": 1194
+    },
+    {
+      "epoch": 4.840506329113924,
+      "grad_norm": 0.16187302379000765,
+      "learning_rate": 9.857879404556291e-08,
+      "loss": 0.2439,
+      "step": 1195
+    },
+    {
+      "epoch": 4.844556962025316,
+      "grad_norm": 0.16015289337984162,
+      "learning_rate": 9.303050139225722e-08,
+      "loss": 0.2565,
+      "step": 1196
+    },
+    {
+      "epoch": 4.848607594936709,
+      "grad_norm": 0.16402211307751735,
+      "learning_rate": 8.76425367345779e-08,
+      "loss": 0.2406,
+      "step": 1197
+    },
+    {
+      "epoch": 4.852658227848101,
+      "grad_norm": 0.1689567494026619,
+      "learning_rate": 8.241494346644897e-08,
+      "loss": 0.2497,
+      "step": 1198
+    },
+    {
+      "epoch": 4.856708860759493,
+      "grad_norm": 0.17005203479370717,
+      "learning_rate": 7.734776369019204e-08,
+      "loss": 0.2503,
+      "step": 1199
+    },
+    {
+      "epoch": 4.860759493670886,
+      "grad_norm": 0.16829078841105324,
+      "learning_rate": 7.244103821617332e-08,
+      "loss": 0.2524,
+      "step": 1200
+    },
+    {
+      "epoch": 4.864810126582278,
+      "grad_norm": 0.16065778672264655,
+      "learning_rate": 6.769480656248606e-08,
+      "loss": 0.249,
+      "step": 1201
+    },
+    {
+      "epoch": 4.868860759493671,
+      "grad_norm": 0.16396774950030898,
+      "learning_rate": 6.310910695462635e-08,
+      "loss": 0.2467,
+      "step": 1202
+    },
+    {
+      "epoch": 4.872911392405063,
+      "grad_norm": 0.16124162602397485,
+      "learning_rate": 5.8683976325191185e-08,
+      "loss": 0.2512,
+      "step": 1203
+    },
+    {
+      "epoch": 4.876962025316455,
+      "grad_norm": 0.16280180398718755,
+      "learning_rate": 5.4419450313571984e-08,
+      "loss": 0.2481,
+      "step": 1204
+    },
+    {
+      "epoch": 4.881012658227848,
+      "grad_norm": 0.1696148953710051,
+      "learning_rate": 5.031556326567488e-08,
+      "loss": 0.2545,
+      "step": 1205
+    },
+    {
+      "epoch": 4.88506329113924,
+      "grad_norm": 0.16264883019165102,
+      "learning_rate": 4.637234823364312e-08,
+      "loss": 0.2534,
+      "step": 1206
+    },
+    {
+      "epoch": 4.889113924050633,
+      "grad_norm": 0.16714089000613558,
+      "learning_rate": 4.258983697558838e-08,
+      "loss": 0.2439,
+      "step": 1207
+    },
+    {
+      "epoch": 4.8931645569620255,
+      "grad_norm": 0.15896261658015548,
+      "learning_rate": 3.896805995533548e-08,
+      "loss": 0.2504,
+      "step": 1208
+    },
+    {
+      "epoch": 4.8972151898734175,
+      "grad_norm": 0.16092498355856125,
+      "learning_rate": 3.550704634218028e-08,
+      "loss": 0.2502,
+      "step": 1209
+    },
+    {
+      "epoch": 4.9012658227848105,
+      "grad_norm": 0.16648693933693343,
+      "learning_rate": 3.2206824010647676e-08,
+      "loss": 0.2463,
+      "step": 1210
+    },
+    {
+      "epoch": 4.905316455696203,
+      "grad_norm": 0.1743769388965327,
+      "learning_rate": 2.9067419540278476e-08,
+      "loss": 0.2546,
+      "step": 1211
+    },
+    {
+      "epoch": 4.909367088607595,
+      "grad_norm": 0.16895793698275374,
+      "learning_rate": 2.6088858215400638e-08,
+      "loss": 0.246,
+      "step": 1212
+    },
+    {
+      "epoch": 4.913417721518988,
+      "grad_norm": 0.1606651857922036,
+      "learning_rate": 2.3271164024940564e-08,
+      "loss": 0.2635,
+      "step": 1213
+    },
+    {
+      "epoch": 4.91746835443038,
+      "grad_norm": 0.16466944462055805,
+      "learning_rate": 2.061435966221881e-08,
+      "loss": 0.252,
+      "step": 1214
+    },
+    {
+      "epoch": 4.921518987341772,
+      "grad_norm": 0.1630952493069387,
+      "learning_rate": 1.811846652477245e-08,
+      "loss": 0.254,
+      "step": 1215
+    },
+    {
+      "epoch": 4.925569620253165,
+      "grad_norm": 0.16221163899736368,
+      "learning_rate": 1.5783504714184106e-08,
+      "loss": 0.255,
+      "step": 1216
+    },
+    {
+      "epoch": 4.929620253164557,
+      "grad_norm": 0.15784690445664412,
+      "learning_rate": 1.360949303591097e-08,
+      "loss": 0.2524,
+      "step": 1217
+    },
+    {
+      "epoch": 4.93367088607595,
+      "grad_norm": 0.1644776921781106,
+      "learning_rate": 1.1596448999144916e-08,
+      "loss": 0.243,
+      "step": 1218
+    },
+    {
+      "epoch": 4.937721518987342,
+      "grad_norm": 0.16128076716103582,
+      "learning_rate": 9.744388816668172e-09,
+      "loss": 0.2516,
+      "step": 1219
+    },
+    {
+      "epoch": 4.941772151898734,
+      "grad_norm": 0.16441348852303092,
+      "learning_rate": 8.05332740472009e-09,
+      "loss": 0.2511,
+      "step": 1220
+    },
+    {
+      "epoch": 4.945822784810127,
+      "grad_norm": 0.16632024981491902,
+      "learning_rate": 6.523278382872811e-09,
+      "loss": 0.2559,
+      "step": 1221
+    },
+    {
+      "epoch": 4.949873417721519,
+      "grad_norm": 0.16280370904142583,
+      "learning_rate": 5.15425407393133e-09,
+      "loss": 0.2539,
+      "step": 1222
+    },
+    {
+      "epoch": 4.953924050632912,
+      "grad_norm": 0.16430094751834598,
+      "learning_rate": 3.94626550383137e-09,
+      "loss": 0.2499,
+      "step": 1223
+    },
+    {
+      "epoch": 4.957974683544304,
+      "grad_norm": 0.15998484803505467,
+      "learning_rate": 2.899322401546112e-09,
+      "loss": 0.2466,
+      "step": 1224
+    },
+    {
+      "epoch": 4.962025316455696,
+      "grad_norm": 0.1608182745734581,
+      "learning_rate": 2.013433199010706e-09,
+      "loss": 0.2535,
+      "step": 1225
+    },
+    {
+      "epoch": 4.966075949367089,
+      "grad_norm": 0.1631259514016836,
+      "learning_rate": 1.2886050310556563e-09,
+      "loss": 0.24,
+      "step": 1226
+    },
+    {
+      "epoch": 4.970126582278481,
+      "grad_norm": 0.16719363888732816,
+      "learning_rate": 7.248437353468695e-10,
+      "loss": 0.245,
+      "step": 1227
+    },
+    {
+      "epoch": 4.974177215189873,
+      "grad_norm": 0.163453024355554,
+      "learning_rate": 3.221538523412449e-10,
+      "loss": 0.2511,
+      "step": 1228
+    },
+    {
+      "epoch": 4.978227848101266,
+      "grad_norm": 0.16054266738468684,
+      "learning_rate": 8.053862524670663e-11,
+      "loss": 0.2449,
+      "step": 1229
+    },
+    {
+      "epoch": 4.982278481012658,
+      "grad_norm": 0.1646875034388065,
+      "learning_rate": 0.0,
+      "loss": 0.2517,
+      "step": 1230
+    },
+    {
+      "epoch": 4.982278481012658,
+      "step": 1230,
+      "total_flos": 6.179014728805941e+18,
+      "train_loss": 0.35052244286469325,
+      "train_runtime": 55060.6716,
+      "train_samples_per_second": 2.87,
+      "train_steps_per_second": 0.022
+    }
+  ],
+  "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": 6.179014728805941e+18,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}