diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,8652 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.988354430379747,
+  "eval_steps": 500,
+  "global_step": 1230,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.004050632911392405,
+      "grad_norm": 13.070682558154115,
+      "learning_rate": 3.2520325203252037e-07,
+      "loss": 1.5366,
+      "step": 1
+    },
+    {
+      "epoch": 0.00810126582278481,
+      "grad_norm": 12.816684960493069,
+      "learning_rate": 6.504065040650407e-07,
+      "loss": 1.5337,
+      "step": 2
+    },
+    {
+      "epoch": 0.012151898734177215,
+      "grad_norm": 12.624671584922462,
+      "learning_rate": 9.75609756097561e-07,
+      "loss": 1.4877,
+      "step": 3
+    },
+    {
+      "epoch": 0.01620253164556962,
+      "grad_norm": 12.717031129866578,
+      "learning_rate": 1.3008130081300815e-06,
+      "loss": 1.5098,
+      "step": 4
+    },
+    {
+      "epoch": 0.020253164556962026,
+      "grad_norm": 12.06970188461979,
+      "learning_rate": 1.6260162601626018e-06,
+      "loss": 1.4757,
+      "step": 5
+    },
+    {
+      "epoch": 0.02430379746835443,
+      "grad_norm": 11.529520664165974,
+      "learning_rate": 1.951219512195122e-06,
+      "loss": 1.4691,
+      "step": 6
+    },
+    {
+      "epoch": 0.028354430379746835,
+      "grad_norm": 8.869204703818426,
+      "learning_rate": 2.2764227642276426e-06,
+      "loss": 1.3796,
+      "step": 7
+    },
+    {
+      "epoch": 0.03240506329113924,
+      "grad_norm": 8.186682220066986,
+      "learning_rate": 2.601626016260163e-06,
+      "loss": 1.3409,
+      "step": 8
+    },
+    {
+      "epoch": 0.03645569620253165,
+      "grad_norm": 5.4819713123976985,
+      "learning_rate": 2.926829268292683e-06,
+      "loss": 1.2654,
+      "step": 9
+    },
+    {
+      "epoch": 0.04050632911392405,
+      "grad_norm": 4.8359842631532945,
+      "learning_rate": 3.2520325203252037e-06,
+      "loss": 1.2504,
+      "step": 10
+    },
+    {
+      "epoch": 0.044556962025316456,
+      "grad_norm": 4.60864063197578,
+      "learning_rate": 3.577235772357724e-06,
+      "loss": 1.232,
+      "step": 11
+    },
+    {
+      "epoch": 0.04860759493670886,
+      "grad_norm": 5.126998477104743,
+      "learning_rate": 3.902439024390244e-06,
+      "loss": 1.181,
+      "step": 12
+    },
+    {
+      "epoch": 0.052658227848101265,
+      "grad_norm": 4.5022622142291855,
+      "learning_rate": 4.227642276422765e-06,
+      "loss": 1.1315,
+      "step": 13
+    },
+    {
+      "epoch": 0.05670886075949367,
+      "grad_norm": 4.461719676639331,
+      "learning_rate": 4.552845528455285e-06,
+      "loss": 1.1285,
+      "step": 14
+    },
+    {
+      "epoch": 0.060759493670886074,
+      "grad_norm": 3.775003764293768,
+      "learning_rate": 4.8780487804878055e-06,
+      "loss": 1.1062,
+      "step": 15
+    },
+    {
+      "epoch": 0.06481012658227848,
+      "grad_norm": 3.4315515795813285,
+      "learning_rate": 5.203252032520326e-06,
+      "loss": 1.048,
+      "step": 16
+    },
+    {
+      "epoch": 0.06886075949367089,
+      "grad_norm": 3.5353044346476934,
+      "learning_rate": 5.528455284552846e-06,
+      "loss": 1.0239,
+      "step": 17
+    },
+    {
+      "epoch": 0.0729113924050633,
+      "grad_norm": 3.141403491481378,
+      "learning_rate": 5.853658536585366e-06,
+      "loss": 0.9933,
+      "step": 18
+    },
+    {
+      "epoch": 0.0769620253164557,
+      "grad_norm": 2.5771614063119026,
+      "learning_rate": 6.178861788617887e-06,
+      "loss": 0.9881,
+      "step": 19
+    },
+    {
+      "epoch": 0.0810126582278481,
+      "grad_norm": 1.905623653237893,
+      "learning_rate": 6.504065040650407e-06,
+      "loss": 0.9678,
+      "step": 20
+    },
+    {
+      "epoch": 0.08506329113924051,
+      "grad_norm": 1.9420441972597766,
+      "learning_rate": 6.829268292682928e-06,
+      "loss": 0.956,
+      "step": 21
+    },
+    {
+      "epoch": 0.08911392405063291,
+      "grad_norm": 2.1730161360481617,
+      "learning_rate": 7.154471544715448e-06,
+      "loss": 0.9552,
+      "step": 22
+    },
+    {
+      "epoch": 0.09316455696202532,
+      "grad_norm": 1.9381123982513848,
+      "learning_rate": 7.4796747967479676e-06,
+      "loss": 0.9276,
+      "step": 23
+    },
+    {
+      "epoch": 0.09721518987341772,
+      "grad_norm": 1.5656833933565186,
+      "learning_rate": 7.804878048780489e-06,
+      "loss": 0.9343,
+      "step": 24
+    },
+    {
+      "epoch": 0.10126582278481013,
+      "grad_norm": 1.4936157082371473,
+      "learning_rate": 8.130081300813009e-06,
+      "loss": 0.9096,
+      "step": 25
+    },
+    {
+      "epoch": 0.10531645569620253,
+      "grad_norm": 1.415812472502614,
+      "learning_rate": 8.45528455284553e-06,
+      "loss": 0.8855,
+      "step": 26
+    },
+    {
+      "epoch": 0.10936708860759493,
+      "grad_norm": 1.310765261928352,
+      "learning_rate": 8.78048780487805e-06,
+      "loss": 0.8839,
+      "step": 27
+    },
+    {
+      "epoch": 0.11341772151898734,
+      "grad_norm": 1.4110932176258717,
+      "learning_rate": 9.10569105691057e-06,
+      "loss": 0.8788,
+      "step": 28
+    },
+    {
+      "epoch": 0.11746835443037974,
+      "grad_norm": 1.4146512774423003,
+      "learning_rate": 9.43089430894309e-06,
+      "loss": 0.8801,
+      "step": 29
+    },
+    {
+      "epoch": 0.12151898734177215,
+      "grad_norm": 1.2354342076393547,
+      "learning_rate": 9.756097560975611e-06,
+      "loss": 0.8712,
+      "step": 30
+    },
+    {
+      "epoch": 0.12556962025316457,
+      "grad_norm": 1.1262657524983688,
+      "learning_rate": 1.008130081300813e-05,
+      "loss": 0.885,
+      "step": 31
+    },
+    {
+      "epoch": 0.12962025316455697,
+      "grad_norm": 1.1191282797927915,
+      "learning_rate": 1.0406504065040652e-05,
+      "loss": 0.8555,
+      "step": 32
+    },
+    {
+      "epoch": 0.13367088607594937,
+      "grad_norm": 1.08744264564874,
+      "learning_rate": 1.0731707317073172e-05,
+      "loss": 0.8612,
+      "step": 33
+    },
+    {
+      "epoch": 0.13772151898734178,
+      "grad_norm": 1.1181128094525983,
+      "learning_rate": 1.1056910569105692e-05,
+      "loss": 0.8432,
+      "step": 34
+    },
+    {
+      "epoch": 0.14177215189873418,
+      "grad_norm": 1.1031631594442757,
+      "learning_rate": 1.1382113821138213e-05,
+      "loss": 0.844,
+      "step": 35
+    },
+    {
+      "epoch": 0.1458227848101266,
+      "grad_norm": 1.2474710915154827,
+      "learning_rate": 1.1707317073170731e-05,
+      "loss": 0.8214,
+      "step": 36
+    },
+    {
+      "epoch": 0.149873417721519,
+      "grad_norm": 1.1648395992962133,
+      "learning_rate": 1.2032520325203254e-05,
+      "loss": 0.823,
+      "step": 37
+    },
+    {
+      "epoch": 0.1539240506329114,
+      "grad_norm": 1.1740140061771136,
+      "learning_rate": 1.2357723577235774e-05,
+      "loss": 0.8393,
+      "step": 38
+    },
+    {
+      "epoch": 0.1579746835443038,
+      "grad_norm": 1.135685173301494,
+      "learning_rate": 1.2682926829268294e-05,
+      "loss": 0.8261,
+      "step": 39
+    },
+    {
+      "epoch": 0.1620253164556962,
+      "grad_norm": 1.1957926505352108,
+      "learning_rate": 1.3008130081300815e-05,
+      "loss": 0.8117,
+      "step": 40
+    },
+    {
+      "epoch": 0.1660759493670886,
+      "grad_norm": 1.1019597771207195,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.828,
+      "step": 41
+    },
+    {
+      "epoch": 0.17012658227848101,
+      "grad_norm": 1.1887882624212374,
+      "learning_rate": 1.3658536585365855e-05,
+      "loss": 0.8192,
+      "step": 42
+    },
+    {
+      "epoch": 0.17417721518987342,
+      "grad_norm": 1.1874018015492218,
+      "learning_rate": 1.3983739837398376e-05,
+      "loss": 0.8289,
+      "step": 43
+    },
+    {
+      "epoch": 0.17822784810126582,
+      "grad_norm": 1.167971248228732,
+      "learning_rate": 1.4308943089430896e-05,
+      "loss": 0.8226,
+      "step": 44
+    },
+    {
+      "epoch": 0.18227848101265823,
+      "grad_norm": 2.6066048878576136,
+      "learning_rate": 1.4634146341463415e-05,
+      "loss": 0.8042,
+      "step": 45
+    },
+    {
+      "epoch": 0.18632911392405063,
+      "grad_norm": 1.3375584181481408,
+      "learning_rate": 1.4959349593495935e-05,
+      "loss": 0.814,
+      "step": 46
+    },
+    {
+      "epoch": 0.19037974683544304,
+      "grad_norm": 1.0504220817120329,
+      "learning_rate": 1.528455284552846e-05,
+      "loss": 0.8195,
+      "step": 47
+    },
+    {
+      "epoch": 0.19443037974683544,
+      "grad_norm": 1.4639777248652057,
+      "learning_rate": 1.5609756097560978e-05,
+      "loss": 0.7971,
+      "step": 48
+    },
+    {
+      "epoch": 0.19848101265822785,
+      "grad_norm": 1.1854744342337808,
+      "learning_rate": 1.5934959349593496e-05,
+      "loss": 0.7933,
+      "step": 49
+    },
+    {
+      "epoch": 0.20253164556962025,
+      "grad_norm": 1.1081996025173924,
+      "learning_rate": 1.6260162601626018e-05,
+      "loss": 0.7972,
+      "step": 50
+    },
+    {
+      "epoch": 0.20658227848101265,
+      "grad_norm": 1.3017250710494046,
+      "learning_rate": 1.6585365853658537e-05,
+      "loss": 0.7936,
+      "step": 51
+    },
+    {
+      "epoch": 0.21063291139240506,
+      "grad_norm": 1.250327011189933,
+      "learning_rate": 1.691056910569106e-05,
+      "loss": 0.7808,
+      "step": 52
+    },
+    {
+      "epoch": 0.21468354430379746,
+      "grad_norm": 1.027278321693512,
+      "learning_rate": 1.7235772357723578e-05,
+      "loss": 0.7916,
+      "step": 53
+    },
+    {
+      "epoch": 0.21873417721518987,
+      "grad_norm": 1.1146830711541784,
+      "learning_rate": 1.75609756097561e-05,
+      "loss": 0.7938,
+      "step": 54
+    },
+    {
+      "epoch": 0.22278481012658227,
+      "grad_norm": 1.246476840296528,
+      "learning_rate": 1.788617886178862e-05,
+      "loss": 0.7878,
+      "step": 55
+    },
+    {
+      "epoch": 0.22683544303797468,
+      "grad_norm": 1.1815971147492443,
+      "learning_rate": 1.821138211382114e-05,
+      "loss": 0.7862,
+      "step": 56
+    },
+    {
+      "epoch": 0.23088607594936708,
+      "grad_norm": 1.1487957128483568,
+      "learning_rate": 1.8536585365853663e-05,
+      "loss": 0.7882,
+      "step": 57
+    },
+    {
+      "epoch": 0.23493670886075949,
+      "grad_norm": 1.3251299081557577,
+      "learning_rate": 1.886178861788618e-05,
+      "loss": 0.7953,
+      "step": 58
+    },
+    {
+      "epoch": 0.2389873417721519,
+      "grad_norm": 1.134601476121093,
+      "learning_rate": 1.91869918699187e-05,
+      "loss": 0.7756,
+      "step": 59
+    },
+    {
+      "epoch": 0.2430379746835443,
+      "grad_norm": 1.2967707053856998,
+      "learning_rate": 1.9512195121951222e-05,
+      "loss": 0.7592,
+      "step": 60
+    },
+    {
+      "epoch": 0.2470886075949367,
+      "grad_norm": 1.2952709473533852,
+      "learning_rate": 1.983739837398374e-05,
+      "loss": 0.7996,
+      "step": 61
+    },
+    {
+      "epoch": 0.25113924050632913,
+      "grad_norm": 1.3650157037085144,
+      "learning_rate": 2.016260162601626e-05,
+      "loss": 0.7844,
+      "step": 62
+    },
+    {
+      "epoch": 0.25518987341772154,
+      "grad_norm": 1.3192783462202553,
+      "learning_rate": 2.048780487804878e-05,
+      "loss": 0.7796,
+      "step": 63
+    },
+    {
+      "epoch": 0.25924050632911394,
+      "grad_norm": 1.504849859915205,
+      "learning_rate": 2.0813008130081303e-05,
+      "loss": 0.7857,
+      "step": 64
+    },
+    {
+      "epoch": 0.26329113924050634,
+      "grad_norm": 1.3706369755347956,
+      "learning_rate": 2.1138211382113822e-05,
+      "loss": 0.7613,
+      "step": 65
+    },
+    {
+      "epoch": 0.26734177215189875,
+      "grad_norm": 1.4685105126945526,
+      "learning_rate": 2.1463414634146344e-05,
+      "loss": 0.7632,
+      "step": 66
+    },
+    {
+      "epoch": 0.27139240506329115,
+      "grad_norm": 1.2727136666928813,
+      "learning_rate": 2.1788617886178863e-05,
+      "loss": 0.7821,
+      "step": 67
+    },
+    {
+      "epoch": 0.27544303797468356,
+      "grad_norm": 1.189753951208195,
+      "learning_rate": 2.2113821138211385e-05,
+      "loss": 0.7628,
+      "step": 68
+    },
+    {
+      "epoch": 0.27949367088607596,
+      "grad_norm": 1.260957492760979,
+      "learning_rate": 2.2439024390243907e-05,
+      "loss": 0.7744,
+      "step": 69
+    },
+    {
+      "epoch": 0.28354430379746837,
+      "grad_norm": 1.2762177414321179,
+      "learning_rate": 2.2764227642276426e-05,
+      "loss": 0.769,
+      "step": 70
+    },
+    {
+      "epoch": 0.28759493670886077,
+      "grad_norm": 1.4182927759466595,
+      "learning_rate": 2.3089430894308948e-05,
+      "loss": 0.7692,
+      "step": 71
+    },
+    {
+      "epoch": 0.2916455696202532,
+      "grad_norm": 0.9832372254062136,
+      "learning_rate": 2.3414634146341463e-05,
+      "loss": 0.7735,
+      "step": 72
+    },
+    {
+      "epoch": 0.2956962025316456,
+      "grad_norm": 1.3346290555121902,
+      "learning_rate": 2.3739837398373985e-05,
+      "loss": 0.7759,
+      "step": 73
+    },
+    {
+      "epoch": 0.299746835443038,
+      "grad_norm": 1.112344734538355,
+      "learning_rate": 2.4065040650406507e-05,
+      "loss": 0.7805,
+      "step": 74
+    },
+    {
+      "epoch": 0.3037974683544304,
+      "grad_norm": 1.1583510329088598,
+      "learning_rate": 2.4390243902439026e-05,
+      "loss": 0.773,
+      "step": 75
+    },
+    {
+      "epoch": 0.3078481012658228,
+      "grad_norm": 1.3496176984077568,
+      "learning_rate": 2.4715447154471548e-05,
+      "loss": 0.7698,
+      "step": 76
+    },
+    {
+      "epoch": 0.3118987341772152,
+      "grad_norm": 1.0931378472917403,
+      "learning_rate": 2.5040650406504066e-05,
+      "loss": 0.7535,
+      "step": 77
+    },
+    {
+      "epoch": 0.3159493670886076,
+      "grad_norm": 1.541722242722326,
+      "learning_rate": 2.536585365853659e-05,
+      "loss": 0.7595,
+      "step": 78
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.153720636751802,
+      "learning_rate": 2.569105691056911e-05,
+      "loss": 0.7488,
+      "step": 79
+    },
+    {
+      "epoch": 0.3240506329113924,
+      "grad_norm": 1.2694179982055285,
+      "learning_rate": 2.601626016260163e-05,
+      "loss": 0.7687,
+      "step": 80
+    },
+    {
+      "epoch": 0.3281012658227848,
+      "grad_norm": 1.3294190302322715,
+      "learning_rate": 2.634146341463415e-05,
+      "loss": 0.7535,
+      "step": 81
+    },
+    {
+      "epoch": 0.3321518987341772,
+      "grad_norm": 1.2629392378956876,
+      "learning_rate": 2.6666666666666667e-05,
+      "loss": 0.7655,
+      "step": 82
+    },
+    {
+      "epoch": 0.3362025316455696,
+      "grad_norm": 1.2590443563103841,
+      "learning_rate": 2.699186991869919e-05,
+      "loss": 0.74,
+      "step": 83
+    },
+    {
+      "epoch": 0.34025316455696203,
+      "grad_norm": 1.1371536092721817,
+      "learning_rate": 2.731707317073171e-05,
+      "loss": 0.7632,
+      "step": 84
+    },
+    {
+      "epoch": 0.34430379746835443,
+      "grad_norm": 1.1738001885704685,
+      "learning_rate": 2.764227642276423e-05,
+      "loss": 0.7492,
+      "step": 85
+    },
+    {
+      "epoch": 0.34835443037974684,
+      "grad_norm": 1.2879593030625143,
+      "learning_rate": 2.796747967479675e-05,
+      "loss": 0.7542,
+      "step": 86
+    },
+    {
+      "epoch": 0.35240506329113924,
+      "grad_norm": 1.3186915896577849,
+      "learning_rate": 2.829268292682927e-05,
+      "loss": 0.7573,
+      "step": 87
+    },
+    {
+      "epoch": 0.35645569620253165,
+      "grad_norm": 1.1746609878487781,
+      "learning_rate": 2.8617886178861792e-05,
+      "loss": 0.7703,
+      "step": 88
+    },
+    {
+      "epoch": 0.36050632911392405,
+      "grad_norm": 1.270366319533098,
+      "learning_rate": 2.8943089430894314e-05,
+      "loss": 0.7583,
+      "step": 89
+    },
+    {
+      "epoch": 0.36455696202531646,
+      "grad_norm": 1.2812382539911065,
+      "learning_rate": 2.926829268292683e-05,
+      "loss": 0.7873,
+      "step": 90
+    },
+    {
+      "epoch": 0.36860759493670886,
+      "grad_norm": 1.18314872223529,
+      "learning_rate": 2.959349593495935e-05,
+      "loss": 0.7479,
+      "step": 91
+    },
+    {
+      "epoch": 0.37265822784810126,
+      "grad_norm": 1.143640632242186,
+      "learning_rate": 2.991869918699187e-05,
+      "loss": 0.7431,
+      "step": 92
+    },
+    {
+      "epoch": 0.37670886075949367,
+      "grad_norm": 1.3017799627472584,
+      "learning_rate": 3.0243902439024392e-05,
+      "loss": 0.7608,
+      "step": 93
+    },
+    {
+      "epoch": 0.3807594936708861,
+      "grad_norm": 1.1363574613404113,
+      "learning_rate": 3.056910569105692e-05,
+      "loss": 0.7605,
+      "step": 94
+    },
+    {
+      "epoch": 0.3848101265822785,
+      "grad_norm": 1.3244416427851384,
+      "learning_rate": 3.089430894308943e-05,
+      "loss": 0.7625,
+      "step": 95
+    },
+    {
+      "epoch": 0.3888607594936709,
+      "grad_norm": 1.417842490360146,
+      "learning_rate": 3.1219512195121955e-05,
+      "loss": 0.7527,
+      "step": 96
+    },
+    {
+      "epoch": 0.3929113924050633,
+      "grad_norm": 1.0825059736868283,
+      "learning_rate": 3.154471544715447e-05,
+      "loss": 0.7737,
+      "step": 97
+    },
+    {
+      "epoch": 0.3969620253164557,
+      "grad_norm": 1.6036968768124127,
+      "learning_rate": 3.186991869918699e-05,
+      "loss": 0.752,
+      "step": 98
+    },
+    {
+      "epoch": 0.4010126582278481,
+      "grad_norm": 1.2166964138234508,
+      "learning_rate": 3.2195121951219514e-05,
+      "loss": 0.7597,
+      "step": 99
+    },
+    {
+      "epoch": 0.4050632911392405,
+      "grad_norm": 1.2471638345272789,
+      "learning_rate": 3.2520325203252037e-05,
+      "loss": 0.7457,
+      "step": 100
+    },
+    {
+      "epoch": 0.4091139240506329,
+      "grad_norm": 1.3248870657651672,
+      "learning_rate": 3.284552845528456e-05,
+      "loss": 0.7665,
+      "step": 101
+    },
+    {
+      "epoch": 0.4131645569620253,
+      "grad_norm": 1.1116584363895345,
+      "learning_rate": 3.3170731707317074e-05,
+      "loss": 0.7464,
+      "step": 102
+    },
+    {
+      "epoch": 0.4172151898734177,
+      "grad_norm": 1.2966472864377694,
+      "learning_rate": 3.3495934959349596e-05,
+      "loss": 0.753,
+      "step": 103
+    },
+    {
+      "epoch": 0.4212658227848101,
+      "grad_norm": 1.173190802680476,
+      "learning_rate": 3.382113821138212e-05,
+      "loss": 0.7437,
+      "step": 104
+    },
+    {
+      "epoch": 0.4253164556962025,
+      "grad_norm": 1.4834863985913838,
+      "learning_rate": 3.414634146341463e-05,
+      "loss": 0.7545,
+      "step": 105
+    },
+    {
+      "epoch": 0.4293670886075949,
+      "grad_norm": 1.2422460802109472,
+      "learning_rate": 3.4471544715447155e-05,
+      "loss": 0.7597,
+      "step": 106
+    },
+    {
+      "epoch": 0.43341772151898733,
+      "grad_norm": 1.50857442556596,
+      "learning_rate": 3.479674796747968e-05,
+      "loss": 0.7653,
+      "step": 107
+    },
+    {
+      "epoch": 0.43746835443037974,
+      "grad_norm": 1.153395446403412,
+      "learning_rate": 3.51219512195122e-05,
+      "loss": 0.7516,
+      "step": 108
+    },
+    {
+      "epoch": 0.44151898734177214,
+      "grad_norm": 1.309790923345341,
+      "learning_rate": 3.544715447154472e-05,
+      "loss": 0.7499,
+      "step": 109
+    },
+    {
+      "epoch": 0.44556962025316454,
+      "grad_norm": 1.0517033240242326,
+      "learning_rate": 3.577235772357724e-05,
+      "loss": 0.7511,
+      "step": 110
+    },
+    {
+      "epoch": 0.44962025316455695,
+      "grad_norm": 1.1886292849996805,
+      "learning_rate": 3.609756097560976e-05,
+      "loss": 0.7424,
+      "step": 111
+    },
+    {
+      "epoch": 0.45367088607594935,
+      "grad_norm": 1.263059782031441,
+      "learning_rate": 3.642276422764228e-05,
+      "loss": 0.7574,
+      "step": 112
+    },
+    {
+      "epoch": 0.45772151898734176,
+      "grad_norm": 1.1943560154898587,
+      "learning_rate": 3.67479674796748e-05,
+      "loss": 0.7471,
+      "step": 113
+    },
+    {
+      "epoch": 0.46177215189873416,
+      "grad_norm": 1.1010684215547855,
+      "learning_rate": 3.7073170731707325e-05,
+      "loss": 0.7473,
+      "step": 114
+    },
+    {
+      "epoch": 0.46582278481012657,
+      "grad_norm": 1.117543616488177,
+      "learning_rate": 3.739837398373984e-05,
+      "loss": 0.7563,
+      "step": 115
+    },
+    {
+      "epoch": 0.46987341772151897,
+      "grad_norm": 1.1099683704280694,
+      "learning_rate": 3.772357723577236e-05,
+      "loss": 0.7398,
+      "step": 116
+    },
+    {
+      "epoch": 0.4739240506329114,
+      "grad_norm": 0.9444333197513489,
+      "learning_rate": 3.804878048780488e-05,
+      "loss": 0.7467,
+      "step": 117
+    },
+    {
+      "epoch": 0.4779746835443038,
+      "grad_norm": 1.0202777224861608,
+      "learning_rate": 3.83739837398374e-05,
+      "loss": 0.7352,
+      "step": 118
+    },
+    {
+      "epoch": 0.4820253164556962,
+      "grad_norm": 1.175107052377322,
+      "learning_rate": 3.869918699186992e-05,
+      "loss": 0.7471,
+      "step": 119
+    },
+    {
+      "epoch": 0.4860759493670886,
+      "grad_norm": 1.1525339494023825,
+      "learning_rate": 3.9024390243902444e-05,
+      "loss": 0.7484,
+      "step": 120
+    },
+    {
+      "epoch": 0.490126582278481,
+      "grad_norm": 1.2828159299983246,
+      "learning_rate": 3.9349593495934966e-05,
+      "loss": 0.7458,
+      "step": 121
+    },
+    {
+      "epoch": 0.4941772151898734,
+      "grad_norm": 1.074803672842246,
+      "learning_rate": 3.967479674796748e-05,
+      "loss": 0.7434,
+      "step": 122
+    },
+    {
+      "epoch": 0.4982278481012658,
+      "grad_norm": 1.1750604281216221,
+      "learning_rate": 4e-05,
+      "loss": 0.7503,
+      "step": 123
+    },
+    {
+      "epoch": 0.5022784810126583,
+      "grad_norm": 0.964046343863995,
+      "learning_rate": 3.999991946137476e-05,
+      "loss": 0.7613,
+      "step": 124
+    },
+    {
+      "epoch": 0.5063291139240507,
+      "grad_norm": 1.1752143602271032,
+      "learning_rate": 3.999967784614766e-05,
+      "loss": 0.737,
+      "step": 125
+    },
+    {
+      "epoch": 0.5103797468354431,
+      "grad_norm": 1.0908991756431567,
+      "learning_rate": 3.9999275156264656e-05,
+      "loss": 0.7427,
+      "step": 126
+    },
+    {
+      "epoch": 0.5144303797468355,
+      "grad_norm": 1.0488256031784267,
+      "learning_rate": 3.999871139496895e-05,
+      "loss": 0.7353,
+      "step": 127
+    },
+    {
+      "epoch": 0.5184810126582279,
+      "grad_norm": 1.0482625231158456,
+      "learning_rate": 3.9997986566800995e-05,
+      "loss": 0.7525,
+      "step": 128
+    },
+    {
+      "epoch": 0.5225316455696203,
+      "grad_norm": 0.9871826387719858,
+      "learning_rate": 3.999710067759846e-05,
+      "loss": 0.7554,
+      "step": 129
+    },
+    {
+      "epoch": 0.5265822784810127,
+      "grad_norm": 1.2402491861511533,
+      "learning_rate": 3.999605373449617e-05,
+      "loss": 0.7521,
+      "step": 130
+    },
+    {
+      "epoch": 0.5306329113924051,
+      "grad_norm": 0.998326886292833,
+      "learning_rate": 3.9994845745926075e-05,
+      "loss": 0.7503,
+      "step": 131
+    },
+    {
+      "epoch": 0.5346835443037975,
+      "grad_norm": 1.0112831605768453,
+      "learning_rate": 3.999347672161713e-05,
+      "loss": 0.7484,
+      "step": 132
+    },
+    {
+      "epoch": 0.5387341772151899,
+      "grad_norm": 1.0504606589667682,
+      "learning_rate": 3.999194667259528e-05,
+      "loss": 0.7654,
+      "step": 133
+    },
+    {
+      "epoch": 0.5427848101265823,
+      "grad_norm": 0.9029713427033789,
+      "learning_rate": 3.999025561118334e-05,
+      "loss": 0.7333,
+      "step": 134
+    },
+    {
+      "epoch": 0.5468354430379747,
+      "grad_norm": 1.0375579798933683,
+      "learning_rate": 3.998840355100086e-05,
+      "loss": 0.7528,
+      "step": 135
+    },
+    {
+      "epoch": 0.5508860759493671,
+      "grad_norm": 0.995897574359677,
+      "learning_rate": 3.998639050696409e-05,
+      "loss": 0.7304,
+      "step": 136
+    },
+    {
+      "epoch": 0.5549367088607595,
+      "grad_norm": 1.3925818252202051,
+      "learning_rate": 3.998421649528582e-05,
+      "loss": 0.7301,
+      "step": 137
+    },
+    {
+      "epoch": 0.5589873417721519,
+      "grad_norm": 0.9408429428433862,
+      "learning_rate": 3.9981881533475234e-05,
+      "loss": 0.7413,
+      "step": 138
+    },
+    {
+      "epoch": 0.5630379746835443,
+      "grad_norm": 1.4171376200869321,
+      "learning_rate": 3.997938564033779e-05,
+      "loss": 0.7374,
+      "step": 139
+    },
+    {
+      "epoch": 0.5670886075949367,
+      "grad_norm": 0.9664335143738435,
+      "learning_rate": 3.9976728835975064e-05,
+      "loss": 0.7572,
+      "step": 140
+    },
+    {
+      "epoch": 0.5711392405063291,
+      "grad_norm": 1.4022345271747627,
+      "learning_rate": 3.9973911141784605e-05,
+      "loss": 0.7385,
+      "step": 141
+    },
+    {
+      "epoch": 0.5751898734177215,
+      "grad_norm": 1.0362251944571663,
+      "learning_rate": 3.997093258045973e-05,
+      "loss": 0.7591,
+      "step": 142
+    },
+    {
+      "epoch": 0.579240506329114,
+      "grad_norm": 1.4385306298078189,
+      "learning_rate": 3.996779317598936e-05,
+      "loss": 0.7493,
+      "step": 143
+    },
+    {
+      "epoch": 0.5832911392405064,
+      "grad_norm": 1.2282994678517773,
+      "learning_rate": 3.996449295365782e-05,
+      "loss": 0.735,
+      "step": 144
+    },
+    {
+      "epoch": 0.5873417721518988,
+      "grad_norm": 1.0013469607188992,
+      "learning_rate": 3.996103194004467e-05,
+      "loss": 0.7415,
+      "step": 145
+    },
+    {
+      "epoch": 0.5913924050632912,
+      "grad_norm": 1.261504952524951,
+      "learning_rate": 3.995741016302441e-05,
+      "loss": 0.7652,
+      "step": 146
+    },
+    {
+      "epoch": 0.5954430379746836,
+      "grad_norm": 0.9362124987435154,
+      "learning_rate": 3.9953627651766364e-05,
+      "loss": 0.7379,
+      "step": 147
+    },
+    {
+      "epoch": 0.599493670886076,
+      "grad_norm": 1.1206663870309257,
+      "learning_rate": 3.9949684436734325e-05,
+      "loss": 0.7473,
+      "step": 148
+    },
+    {
+      "epoch": 0.6035443037974684,
+      "grad_norm": 1.1148588779096054,
+      "learning_rate": 3.994558054968643e-05,
+      "loss": 0.7415,
+      "step": 149
+    },
+    {
+      "epoch": 0.6075949367088608,
+      "grad_norm": 0.9588564242521396,
+      "learning_rate": 3.994131602367481e-05,
+      "loss": 0.753,
+      "step": 150
+    },
+    {
+      "epoch": 0.6116455696202532,
+      "grad_norm": 1.1975223473442032,
+      "learning_rate": 3.9936890893045376e-05,
+      "loss": 0.7528,
+      "step": 151
+    },
+    {
+      "epoch": 0.6156962025316456,
+      "grad_norm": 0.9705495827110426,
+      "learning_rate": 3.993230519343752e-05,
+      "loss": 0.7441,
+      "step": 152
+    },
+    {
+      "epoch": 0.619746835443038,
+      "grad_norm": 0.9815197893252781,
+      "learning_rate": 3.992755896178383e-05,
+      "loss": 0.7334,
+      "step": 153
+    },
+    {
+      "epoch": 0.6237974683544304,
+      "grad_norm": 1.0069248656010863,
+      "learning_rate": 3.992265223630981e-05,
+      "loss": 0.7392,
+      "step": 154
+    },
+    {
+      "epoch": 0.6278481012658228,
+      "grad_norm": 0.990204228374541,
+      "learning_rate": 3.991758505653355e-05,
+      "loss": 0.7428,
+      "step": 155
+    },
+    {
+      "epoch": 0.6318987341772152,
+      "grad_norm": 1.0878804528772386,
+      "learning_rate": 3.991235746326543e-05,
+      "loss": 0.7514,
+      "step": 156
+    },
+    {
+      "epoch": 0.6359493670886076,
+      "grad_norm": 0.9557688556133223,
+      "learning_rate": 3.9906969498607745e-05,
+      "loss": 0.7359,
+      "step": 157
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.9586680666198227,
+      "learning_rate": 3.990142120595444e-05,
+      "loss": 0.738,
+      "step": 158
+    },
+    {
+      "epoch": 0.6440506329113924,
+      "grad_norm": 0.9898017858640755,
+      "learning_rate": 3.98957126299907e-05,
+      "loss": 0.7326,
+      "step": 159
+    },
+    {
+      "epoch": 0.6481012658227848,
+      "grad_norm": 0.9734655129946332,
+      "learning_rate": 3.9889843816692596e-05,
+      "loss": 0.7364,
+      "step": 160
+    },
+    {
+      "epoch": 0.6521518987341772,
+      "grad_norm": 0.9283838181990174,
+      "learning_rate": 3.9883814813326766e-05,
+      "loss": 0.7296,
+      "step": 161
+    },
+    {
+      "epoch": 0.6562025316455696,
+      "grad_norm": 1.0077950068281587,
+      "learning_rate": 3.9877625668449956e-05,
+      "loss": 0.7332,
+      "step": 162
+    },
+    {
+      "epoch": 0.660253164556962,
+      "grad_norm": 0.9164230845877345,
+      "learning_rate": 3.98712764319087e-05,
+      "loss": 0.733,
+      "step": 163
+    },
+    {
+      "epoch": 0.6643037974683544,
+      "grad_norm": 0.9047807166493744,
+      "learning_rate": 3.9864767154838864e-05,
+      "loss": 0.7255,
+      "step": 164
+    },
+    {
+      "epoch": 0.6683544303797468,
+      "grad_norm": 0.9166787881039281,
+      "learning_rate": 3.9858097889665277e-05,
+      "loss": 0.7315,
+      "step": 165
+    },
+    {
+      "epoch": 0.6724050632911392,
+      "grad_norm": 0.9154349578090769,
+      "learning_rate": 3.985126869010129e-05,
+      "loss": 0.7313,
+      "step": 166
+    },
+    {
+      "epoch": 0.6764556962025317,
+      "grad_norm": 0.866973971766779,
+      "learning_rate": 3.984427961114833e-05,
+      "loss": 0.7441,
+      "step": 167
+    },
+    {
+      "epoch": 0.6805063291139241,
+      "grad_norm": 0.8975141095692059,
+      "learning_rate": 3.9837130709095475e-05,
+      "loss": 0.7378,
+      "step": 168
+    },
+    {
+      "epoch": 0.6845569620253165,
+      "grad_norm": 0.8381899674559047,
+      "learning_rate": 3.982982204151901e-05,
+      "loss": 0.7278,
+      "step": 169
+    },
+    {
+      "epoch": 0.6886075949367089,
+      "grad_norm": 0.9194761067707109,
+      "learning_rate": 3.982235366728193e-05,
+      "loss": 0.7201,
+      "step": 170
+    },
+    {
+      "epoch": 0.6926582278481013,
+      "grad_norm": 0.7646228806213905,
+      "learning_rate": 3.9814725646533505e-05,
+      "loss": 0.7213,
+      "step": 171
+    },
+    {
+      "epoch": 0.6967088607594937,
+      "grad_norm": 0.875048191568092,
+      "learning_rate": 3.9806938040708746e-05,
+      "loss": 0.7401,
+      "step": 172
+    },
+    {
+      "epoch": 0.7007594936708861,
+      "grad_norm": 0.7983934203402991,
+      "learning_rate": 3.9798990912527976e-05,
+      "loss": 0.7299,
+      "step": 173
+    },
+    {
+      "epoch": 0.7048101265822785,
+      "grad_norm": 0.7599241965667751,
+      "learning_rate": 3.979088432599627e-05,
+      "loss": 0.7376,
+      "step": 174
+    },
+    {
+      "epoch": 0.7088607594936709,
+      "grad_norm": 0.7791469018232992,
+      "learning_rate": 3.9782618346402964e-05,
+      "loss": 0.7114,
+      "step": 175
+    },
+    {
+      "epoch": 0.7129113924050633,
+      "grad_norm": 0.8522073724177315,
+      "learning_rate": 3.977419304032111e-05,
+      "loss": 0.735,
+      "step": 176
+    },
+    {
+      "epoch": 0.7169620253164557,
+      "grad_norm": 0.8046725650190303,
+      "learning_rate": 3.976560847560697e-05,
+      "loss": 0.7126,
+      "step": 177
+    },
+    {
+      "epoch": 0.7210126582278481,
+      "grad_norm": 0.860358265323839,
+      "learning_rate": 3.9756864721399456e-05,
+      "loss": 0.7408,
+      "step": 178
+    },
+    {
+      "epoch": 0.7250632911392405,
+      "grad_norm": 0.8739260894194718,
+      "learning_rate": 3.974796184811956e-05,
+      "loss": 0.7468,
+      "step": 179
+    },
+    {
+      "epoch": 0.7291139240506329,
+      "grad_norm": 0.8516117442549455,
+      "learning_rate": 3.973889992746979e-05,
+      "loss": 0.7241,
+      "step": 180
+    },
+    {
+      "epoch": 0.7331645569620253,
+      "grad_norm": 0.8432607913322947,
+      "learning_rate": 3.972967903243361e-05,
+      "loss": 0.7332,
+      "step": 181
+    },
+    {
+      "epoch": 0.7372151898734177,
+      "grad_norm": 0.8053955215859024,
+      "learning_rate": 3.972029923727486e-05,
+      "loss": 0.7204,
+      "step": 182
+    },
+    {
+      "epoch": 0.7412658227848101,
+      "grad_norm": 0.8544588905323212,
+      "learning_rate": 3.971076061753709e-05,
+      "loss": 0.7145,
+      "step": 183
+    },
+    {
+      "epoch": 0.7453164556962025,
+      "grad_norm": 0.8351905612201952,
+      "learning_rate": 3.9701063250043066e-05,
+      "loss": 0.7202,
+      "step": 184
+    },
+    {
+      "epoch": 0.7493670886075949,
+      "grad_norm": 0.8176599379003234,
+      "learning_rate": 3.969120721289402e-05,
+      "loss": 0.7213,
+      "step": 185
+    },
+    {
+      "epoch": 0.7534177215189873,
+      "grad_norm": 0.7810201489879313,
+      "learning_rate": 3.9681192585469146e-05,
+      "loss": 0.7071,
+      "step": 186
+    },
+    {
+      "epoch": 0.7574683544303797,
+      "grad_norm": 0.9133931989405283,
+      "learning_rate": 3.9671019448424865e-05,
+      "loss": 0.737,
+      "step": 187
+    },
+    {
+      "epoch": 0.7615189873417721,
+      "grad_norm": 0.8163205576374468,
+      "learning_rate": 3.966068788369422e-05,
+      "loss": 0.7321,
+      "step": 188
+    },
+    {
+      "epoch": 0.7655696202531646,
+      "grad_norm": 0.91876838907761,
+      "learning_rate": 3.965019797448622e-05,
+      "loss": 0.7093,
+      "step": 189
+    },
+    {
+      "epoch": 0.769620253164557,
+      "grad_norm": 0.8822737129589666,
+      "learning_rate": 3.963954980528515e-05,
+      "loss": 0.7258,
+      "step": 190
+    },
+    {
+      "epoch": 0.7736708860759494,
+      "grad_norm": 0.9639372645768006,
+      "learning_rate": 3.9628743461849905e-05,
+      "loss": 0.7274,
+      "step": 191
+    },
+    {
+      "epoch": 0.7777215189873418,
+      "grad_norm": 0.8656795760784538,
+      "learning_rate": 3.961777903121329e-05,
+      "loss": 0.7174,
+      "step": 192
+    },
+    {
+      "epoch": 0.7817721518987342,
+      "grad_norm": 0.8774214517467834,
+      "learning_rate": 3.960665660168131e-05,
+      "loss": 0.7213,
+      "step": 193
+    },
+    {
+      "epoch": 0.7858227848101266,
+      "grad_norm": 0.8951631602423663,
+      "learning_rate": 3.9595376262832485e-05,
+      "loss": 0.7056,
+      "step": 194
+    },
+    {
+      "epoch": 0.789873417721519,
+      "grad_norm": 0.9793444233453383,
+      "learning_rate": 3.9583938105517127e-05,
+      "loss": 0.7324,
+      "step": 195
+    },
+    {
+      "epoch": 0.7939240506329114,
+      "grad_norm": 0.8682734472877827,
+      "learning_rate": 3.957234222185657e-05,
+      "loss": 0.7169,
+      "step": 196
+    },
+    {
+      "epoch": 0.7979746835443038,
+      "grad_norm": 0.9215145515296794,
+      "learning_rate": 3.9560588705242474e-05,
+      "loss": 0.7204,
+      "step": 197
+    },
+    {
+      "epoch": 0.8020253164556962,
+      "grad_norm": 0.8496556947116541,
+      "learning_rate": 3.954867765033605e-05,
+      "loss": 0.7196,
+      "step": 198
+    },
+    {
+      "epoch": 0.8060759493670886,
+      "grad_norm": 0.8213490985668107,
+      "learning_rate": 3.953660915306728e-05,
+      "loss": 0.7294,
+      "step": 199
+    },
+    {
+      "epoch": 0.810126582278481,
+      "grad_norm": 0.7847541074800203,
+      "learning_rate": 3.952438331063419e-05,
+      "loss": 0.7161,
+      "step": 200
+    },
+    {
+      "epoch": 0.8141772151898734,
+      "grad_norm": 0.9992823927279778,
+      "learning_rate": 3.951200022150205e-05,
+      "loss": 0.7124,
+      "step": 201
+    },
+    {
+      "epoch": 0.8182278481012658,
+      "grad_norm": 0.7565152579859121,
+      "learning_rate": 3.949945998540253e-05,
+      "loss": 0.7083,
+      "step": 202
+    },
+    {
+      "epoch": 0.8222784810126582,
+      "grad_norm": 1.0164386503225726,
+      "learning_rate": 3.9486762703332993e-05,
+      "loss": 0.7091,
+      "step": 203
+    },
+    {
+      "epoch": 0.8263291139240506,
+      "grad_norm": 0.7248127671105634,
+      "learning_rate": 3.947390847755559e-05,
+      "loss": 0.707,
+      "step": 204
+    },
+    {
+      "epoch": 0.830379746835443,
+      "grad_norm": 0.9029637098863897,
+      "learning_rate": 3.946089741159648e-05,
+      "loss": 0.7295,
+      "step": 205
+    },
+    {
+      "epoch": 0.8344303797468354,
+      "grad_norm": 0.7474758900767335,
+      "learning_rate": 3.944772961024501e-05,
+      "loss": 0.7035,
+      "step": 206
+    },
+    {
+      "epoch": 0.8384810126582278,
+      "grad_norm": 0.9102336621101458,
+      "learning_rate": 3.943440517955285e-05,
+      "loss": 0.7196,
+      "step": 207
+    },
+    {
+      "epoch": 0.8425316455696202,
+      "grad_norm": 0.7612244518273419,
+      "learning_rate": 3.9420924226833126e-05,
+      "loss": 0.7153,
+      "step": 208
+    },
+    {
+      "epoch": 0.8465822784810126,
+      "grad_norm": 0.8224181050579885,
+      "learning_rate": 3.9407286860659566e-05,
+      "loss": 0.7032,
+      "step": 209
+    },
+    {
+      "epoch": 0.850632911392405,
+      "grad_norm": 0.8532838885585351,
+      "learning_rate": 3.9393493190865657e-05,
+      "loss": 0.7289,
+      "step": 210
+    },
+    {
+      "epoch": 0.8546835443037974,
+      "grad_norm": 0.7828622363129715,
+      "learning_rate": 3.937954332854371e-05,
+      "loss": 0.711,
+      "step": 211
+    },
+    {
+      "epoch": 0.8587341772151899,
+      "grad_norm": 0.7995668200093173,
+      "learning_rate": 3.9365437386044016e-05,
+      "loss": 0.7175,
+      "step": 212
+    },
+    {
+      "epoch": 0.8627848101265823,
+      "grad_norm": 0.7550117420651218,
+      "learning_rate": 3.935117547697387e-05,
+      "loss": 0.701,
+      "step": 213
+    },
+    {
+      "epoch": 0.8668354430379747,
+      "grad_norm": 0.7816015157038998,
+      "learning_rate": 3.933675771619675e-05,
+      "loss": 0.6991,
+      "step": 214
+    },
+    {
+      "epoch": 0.8708860759493671,
+      "grad_norm": 0.8145969864655296,
+      "learning_rate": 3.932218421983131e-05,
+      "loss": 0.6986,
+      "step": 215
+    },
+    {
+      "epoch": 0.8749367088607595,
+      "grad_norm": 0.8101295104564256,
+      "learning_rate": 3.9307455105250484e-05,
+      "loss": 0.7115,
+      "step": 216
+    },
+    {
+      "epoch": 0.8789873417721519,
+      "grad_norm": 0.8325782514101805,
+      "learning_rate": 3.929257049108054e-05,
+      "loss": 0.71,
+      "step": 217
+    },
+    {
+      "epoch": 0.8830379746835443,
+      "grad_norm": 0.819547044949777,
+      "learning_rate": 3.927753049720011e-05,
+      "loss": 0.7207,
+      "step": 218
+    },
+    {
+      "epoch": 0.8870886075949367,
+      "grad_norm": 0.8373240222301674,
+      "learning_rate": 3.9262335244739234e-05,
+      "loss": 0.7013,
+      "step": 219
+    },
+    {
+      "epoch": 0.8911392405063291,
+      "grad_norm": 0.8370412201471923,
+      "learning_rate": 3.92469848560784e-05,
+      "loss": 0.7018,
+      "step": 220
+    },
+    {
+      "epoch": 0.8951898734177215,
+      "grad_norm": 0.7543577945502784,
+      "learning_rate": 3.923147945484751e-05,
+      "loss": 0.701,
+      "step": 221
+    },
+    {
+      "epoch": 0.8992405063291139,
+      "grad_norm": 0.8035502997864228,
+      "learning_rate": 3.9215819165924956e-05,
+      "loss": 0.7116,
+      "step": 222
+    },
+    {
+      "epoch": 0.9032911392405063,
+      "grad_norm": 0.7755775348917496,
+      "learning_rate": 3.920000411543654e-05,
+      "loss": 0.7032,
+      "step": 223
+    },
+    {
+      "epoch": 0.9073417721518987,
+      "grad_norm": 0.8465613358033643,
+      "learning_rate": 3.9184034430754495e-05,
+      "loss": 0.7138,
+      "step": 224
+    },
+    {
+      "epoch": 0.9113924050632911,
+      "grad_norm": 0.7639069476845027,
+      "learning_rate": 3.916791024049648e-05,
+      "loss": 0.7001,
+      "step": 225
+    },
+    {
+      "epoch": 0.9154430379746835,
+      "grad_norm": 0.7741767743784045,
+      "learning_rate": 3.91516316745245e-05,
+      "loss": 0.6911,
+      "step": 226
+    },
+    {
+      "epoch": 0.9194936708860759,
+      "grad_norm": 0.78777081146722,
+      "learning_rate": 3.913519886394389e-05,
+      "loss": 0.7171,
+      "step": 227
+    },
+    {
+      "epoch": 0.9235443037974683,
+      "grad_norm": 0.7456497313421884,
+      "learning_rate": 3.911861194110225e-05,
+      "loss": 0.7122,
+      "step": 228
+    },
+    {
+      "epoch": 0.9275949367088607,
+      "grad_norm": 0.7760853371915731,
+      "learning_rate": 3.910187103958837e-05,
+      "loss": 0.7216,
+      "step": 229
+    },
+    {
+      "epoch": 0.9316455696202531,
+      "grad_norm": 0.7757834703658523,
+      "learning_rate": 3.908497629423117e-05,
+      "loss": 0.7074,
+      "step": 230
+    },
+    {
+      "epoch": 0.9356962025316455,
+      "grad_norm": 0.7555198490732327,
+      "learning_rate": 3.9067927841098614e-05,
+      "loss": 0.7114,
+      "step": 231
+    },
+    {
+      "epoch": 0.9397468354430379,
+      "grad_norm": 0.7386878505690627,
+      "learning_rate": 3.9050725817496594e-05,
+      "loss": 0.6838,
+      "step": 232
+    },
+    {
+      "epoch": 0.9437974683544303,
+      "grad_norm": 0.7635614807784474,
+      "learning_rate": 3.9033370361967844e-05,
+      "loss": 0.701,
+      "step": 233
+    },
+    {
+      "epoch": 0.9478481012658228,
+      "grad_norm": 0.8200299449808857,
+      "learning_rate": 3.901586161429081e-05,
+      "loss": 0.7077,
+      "step": 234
+    },
+    {
+      "epoch": 0.9518987341772152,
+      "grad_norm": 0.8501022649191765,
+      "learning_rate": 3.8998199715478545e-05,
+      "loss": 0.699,
+      "step": 235
+    },
+    {
+      "epoch": 0.9559493670886076,
+      "grad_norm": 0.7696824560729838,
+      "learning_rate": 3.8980384807777564e-05,
+      "loss": 0.6865,
+      "step": 236
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.8088582563914647,
+      "learning_rate": 3.896241703466667e-05,
+      "loss": 0.7102,
+      "step": 237
+    },
+    {
+      "epoch": 0.9640506329113924,
+      "grad_norm": 0.7280651079774261,
+      "learning_rate": 3.894429654085585e-05,
+      "loss": 0.6927,
+      "step": 238
+    },
+    {
+      "epoch": 0.9681012658227848,
+      "grad_norm": 0.8392000889067409,
+      "learning_rate": 3.892602347228505e-05,
+      "loss": 0.6952,
+      "step": 239
+    },
+    {
+      "epoch": 0.9721518987341772,
+      "grad_norm": 0.7500667433778668,
+      "learning_rate": 3.890759797612307e-05,
+      "loss": 0.7028,
+      "step": 240
+    },
+    {
+      "epoch": 0.9762025316455696,
+      "grad_norm": 0.8422773774313594,
+      "learning_rate": 3.888902020076632e-05,
+      "loss": 0.727,
+      "step": 241
+    },
+    {
+      "epoch": 0.980253164556962,
+      "grad_norm": 0.7684032068538085,
+      "learning_rate": 3.887029029583764e-05,
+      "loss": 0.698,
+      "step": 242
+    },
+    {
+      "epoch": 0.9843037974683544,
+      "grad_norm": 0.7555068544699503,
+      "learning_rate": 3.8851408412185125e-05,
+      "loss": 0.6929,
+      "step": 243
+    },
+    {
+      "epoch": 0.9883544303797468,
+      "grad_norm": 0.819435799484883,
+      "learning_rate": 3.8832374701880855e-05,
+      "loss": 0.7108,
+      "step": 244
+    },
+    {
+      "epoch": 0.9924050632911392,
+      "grad_norm": 0.7170280925509722,
+      "learning_rate": 3.881318931821972e-05,
+      "loss": 0.6692,
+      "step": 245
+    },
+    {
+      "epoch": 0.9964556962025316,
+      "grad_norm": 0.7809498422803793,
+      "learning_rate": 3.879385241571817e-05,
+      "loss": 0.6898,
+      "step": 246
+    },
+    {
+      "epoch": 1.0020253164556963,
+      "grad_norm": 0.7404655515283329,
+      "learning_rate": 3.8774364150112955e-05,
+      "loss": 0.6218,
+      "step": 247
+    },
+    {
+      "epoch": 1.0060759493670886,
+      "grad_norm": 0.8057955865152169,
+      "learning_rate": 3.8754724678359884e-05,
+      "loss": 0.5234,
+      "step": 248
+    },
+    {
+      "epoch": 1.010126582278481,
+      "grad_norm": 0.7485528408023238,
+      "learning_rate": 3.873493415863256e-05,
+      "loss": 0.5625,
+      "step": 249
+    },
+    {
+      "epoch": 1.0141772151898734,
+      "grad_norm": 0.8715575443837718,
+      "learning_rate": 3.871499275032111e-05,
+      "loss": 0.5277,
+      "step": 250
+    },
+    {
+      "epoch": 1.0182278481012659,
+      "grad_norm": 0.8410506454487814,
+      "learning_rate": 3.869490061403091e-05,
+      "loss": 0.5384,
+      "step": 251
+    },
+    {
+      "epoch": 1.0222784810126582,
+      "grad_norm": 0.7893545545375371,
+      "learning_rate": 3.867465791158124e-05,
+      "loss": 0.4953,
+      "step": 252
+    },
+    {
+      "epoch": 1.0263291139240507,
+      "grad_norm": 0.9674532880077107,
+      "learning_rate": 3.865426480600407e-05,
+      "loss": 0.5184,
+      "step": 253
+    },
+    {
+      "epoch": 1.030379746835443,
+      "grad_norm": 0.7192073985562774,
+      "learning_rate": 3.863372146154264e-05,
+      "loss": 0.5277,
+      "step": 254
+    },
+    {
+      "epoch": 1.0344303797468355,
+      "grad_norm": 0.9118943324115049,
+      "learning_rate": 3.861302804365024e-05,
+      "loss": 0.5321,
+      "step": 255
+    },
+    {
+      "epoch": 1.0384810126582278,
+      "grad_norm": 0.918914383542382,
+      "learning_rate": 3.85921847189888e-05,
+      "loss": 0.5479,
+      "step": 256
+    },
+    {
+      "epoch": 1.0425316455696203,
+      "grad_norm": 0.9845939805042,
+      "learning_rate": 3.85711916554276e-05,
+      "loss": 0.5155,
+      "step": 257
+    },
+    {
+      "epoch": 1.0465822784810126,
+      "grad_norm": 0.8129187032410854,
+      "learning_rate": 3.85500490220419e-05,
+      "loss": 0.5282,
+      "step": 258
+    },
+    {
+      "epoch": 1.0506329113924051,
+      "grad_norm": 0.805977894523452,
+      "learning_rate": 3.852875698911154e-05,
+      "loss": 0.546,
+      "step": 259
+    },
+    {
+      "epoch": 1.0546835443037974,
+      "grad_norm": 0.7977439411192497,
+      "learning_rate": 3.850731572811963e-05,
+      "loss": 0.5461,
+      "step": 260
+    },
+    {
+      "epoch": 1.05873417721519,
+      "grad_norm": 0.6950513249391592,
+      "learning_rate": 3.848572541175116e-05,
+      "loss": 0.5229,
+      "step": 261
+    },
+    {
+      "epoch": 1.0627848101265822,
+      "grad_norm": 0.8062489513643872,
+      "learning_rate": 3.846398621389154e-05,
+      "loss": 0.5379,
+      "step": 262
+    },
+    {
+      "epoch": 1.0668354430379747,
+      "grad_norm": 0.8081716543051349,
+      "learning_rate": 3.84420983096253e-05,
+      "loss": 0.5331,
+      "step": 263
+    },
+    {
+      "epoch": 1.070886075949367,
+      "grad_norm": 0.7249537220310828,
+      "learning_rate": 3.8420061875234606e-05,
+      "loss": 0.5192,
+      "step": 264
+    },
+    {
+      "epoch": 1.0749367088607595,
+      "grad_norm": 0.7315106073454433,
+      "learning_rate": 3.839787708819787e-05,
+      "loss": 0.5338,
+      "step": 265
+    },
+    {
+      "epoch": 1.0789873417721518,
+      "grad_norm": 0.7150373097381064,
+      "learning_rate": 3.8375544127188325e-05,
+      "loss": 0.5234,
+      "step": 266
+    },
+    {
+      "epoch": 1.0830379746835443,
+      "grad_norm": 0.6934565796816425,
+      "learning_rate": 3.8353063172072564e-05,
+      "loss": 0.535,
+      "step": 267
+    },
+    {
+      "epoch": 1.0870886075949366,
+      "grad_norm": 0.6997759413155197,
+      "learning_rate": 3.8330434403909105e-05,
+      "loss": 0.512,
+      "step": 268
+    },
+    {
+      "epoch": 1.0911392405063292,
+      "grad_norm": 0.7423645168138938,
+      "learning_rate": 3.8307658004946934e-05,
+      "loss": 0.5488,
+      "step": 269
+    },
+    {
+      "epoch": 1.0951898734177214,
+      "grad_norm": 0.7986571551557017,
+      "learning_rate": 3.8284734158624046e-05,
+      "loss": 0.5021,
+      "step": 270
+    },
+    {
+      "epoch": 1.099240506329114,
+      "grad_norm": 0.7524722979881758,
+      "learning_rate": 3.826166304956594e-05,
+      "loss": 0.5263,
+      "step": 271
+    },
+    {
+      "epoch": 1.1032911392405063,
+      "grad_norm": 0.8482189791378912,
+      "learning_rate": 3.8238444863584164e-05,
+      "loss": 0.5317,
+      "step": 272
+    },
+    {
+      "epoch": 1.1073417721518988,
+      "grad_norm": 0.7422307635212442,
+      "learning_rate": 3.821507978767479e-05,
+      "loss": 0.5205,
+      "step": 273
+    },
+    {
+      "epoch": 1.111392405063291,
+      "grad_norm": 0.7788951426841969,
+      "learning_rate": 3.819156801001693e-05,
+      "loss": 0.5113,
+      "step": 274
+    },
+    {
+      "epoch": 1.1154430379746836,
+      "grad_norm": 0.748315007166111,
+      "learning_rate": 3.816790971997121e-05,
+      "loss": 0.5254,
+      "step": 275
+    },
+    {
+      "epoch": 1.1194936708860759,
+      "grad_norm": 0.7178244384497735,
+      "learning_rate": 3.8144105108078246e-05,
+      "loss": 0.5312,
+      "step": 276
+    },
+    {
+      "epoch": 1.1235443037974684,
+      "grad_norm": 0.9382327389524957,
+      "learning_rate": 3.81201543660571e-05,
+      "loss": 0.5316,
+      "step": 277
+    },
+    {
+      "epoch": 1.1275949367088607,
+      "grad_norm": 0.7571409024191443,
+      "learning_rate": 3.809605768680377e-05,
+      "loss": 0.5191,
+      "step": 278
+    },
+    {
+      "epoch": 1.1316455696202532,
+      "grad_norm": 0.7799605830068794,
+      "learning_rate": 3.807181526438958e-05,
+      "loss": 0.5181,
+      "step": 279
+    },
+    {
+      "epoch": 1.1356962025316455,
+      "grad_norm": 0.7733810846154847,
+      "learning_rate": 3.8047427294059697e-05,
+      "loss": 0.5487,
+      "step": 280
+    },
+    {
+      "epoch": 1.139746835443038,
+      "grad_norm": 0.6891615367826055,
+      "learning_rate": 3.802289397223145e-05,
+      "loss": 0.5135,
+      "step": 281
+    },
+    {
+      "epoch": 1.1437974683544303,
+      "grad_norm": 0.7667150685334513,
+      "learning_rate": 3.7998215496492854e-05,
+      "loss": 0.5409,
+      "step": 282
+    },
+    {
+      "epoch": 1.1478481012658228,
+      "grad_norm": 0.7123277578676773,
+      "learning_rate": 3.797339206560096e-05,
+      "loss": 0.5326,
+      "step": 283
+    },
+    {
+      "epoch": 1.1518987341772151,
+      "grad_norm": 0.694437875538524,
+      "learning_rate": 3.794842387948027e-05,
+      "loss": 0.5149,
+      "step": 284
+    },
+    {
+      "epoch": 1.1559493670886076,
+      "grad_norm": 0.7272468841603782,
+      "learning_rate": 3.7923311139221114e-05,
+      "loss": 0.5099,
+      "step": 285
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.7214482876106814,
+      "learning_rate": 3.7898054047078054e-05,
+      "loss": 0.53,
+      "step": 286
+    },
+    {
+      "epoch": 1.1640506329113924,
+      "grad_norm": 0.7509220151467625,
+      "learning_rate": 3.787265280646825e-05,
+      "loss": 0.5412,
+      "step": 287
+    },
+    {
+      "epoch": 1.1681012658227847,
+      "grad_norm": 0.7103431483032837,
+      "learning_rate": 3.7847107621969786e-05,
+      "loss": 0.5278,
+      "step": 288
+    },
+    {
+      "epoch": 1.1721518987341772,
+      "grad_norm": 0.7082030946456621,
+      "learning_rate": 3.7821418699320064e-05,
+      "loss": 0.5096,
+      "step": 289
+    },
+    {
+      "epoch": 1.1762025316455695,
+      "grad_norm": 0.7569723675761619,
+      "learning_rate": 3.7795586245414145e-05,
+      "loss": 0.5092,
+      "step": 290
+    },
+    {
+      "epoch": 1.180253164556962,
+      "grad_norm": 0.8325054947220218,
+      "learning_rate": 3.776961046830306e-05,
+      "loss": 0.53,
+      "step": 291
+    },
+    {
+      "epoch": 1.1843037974683543,
+      "grad_norm": 0.8088161999728385,
+      "learning_rate": 3.774349157719215e-05,
+      "loss": 0.5307,
+      "step": 292
+    },
+    {
+      "epoch": 1.1883544303797469,
+      "grad_norm": 0.8049123643132198,
+      "learning_rate": 3.7717229782439365e-05,
+      "loss": 0.5245,
+      "step": 293
+    },
+    {
+      "epoch": 1.1924050632911392,
+      "grad_norm": 0.7824210733842739,
+      "learning_rate": 3.769082529555359e-05,
+      "loss": 0.5372,
+      "step": 294
+    },
+    {
+      "epoch": 1.1964556962025317,
+      "grad_norm": 0.7679218040400987,
+      "learning_rate": 3.766427832919294e-05,
+      "loss": 0.545,
+      "step": 295
+    },
+    {
+      "epoch": 1.200506329113924,
+      "grad_norm": 0.9808285536647569,
+      "learning_rate": 3.7637589097163024e-05,
+      "loss": 0.5356,
+      "step": 296
+    },
+    {
+      "epoch": 1.2045569620253165,
+      "grad_norm": 0.7198903787827498,
+      "learning_rate": 3.761075781441526e-05,
+      "loss": 0.5478,
+      "step": 297
+    },
+    {
+      "epoch": 1.2086075949367088,
+      "grad_norm": 0.9022674875389265,
+      "learning_rate": 3.75837846970451e-05,
+      "loss": 0.5193,
+      "step": 298
+    },
+    {
+      "epoch": 1.2126582278481013,
+      "grad_norm": 0.8747857144105797,
+      "learning_rate": 3.755666996229032e-05,
+      "loss": 0.5323,
+      "step": 299
+    },
+    {
+      "epoch": 1.2167088607594936,
+      "grad_norm": 0.7959170848696048,
+      "learning_rate": 3.752941382852927e-05,
+      "loss": 0.5436,
+      "step": 300
+    },
+    {
+      "epoch": 1.220759493670886,
+      "grad_norm": 0.827401771748068,
+      "learning_rate": 3.7502016515279115e-05,
+      "loss": 0.5142,
+      "step": 301
+    },
+    {
+      "epoch": 1.2248101265822784,
+      "grad_norm": 0.7720493359062818,
+      "learning_rate": 3.7474478243194043e-05,
+      "loss": 0.5357,
+      "step": 302
+    },
+    {
+      "epoch": 1.228860759493671,
+      "grad_norm": 0.7866420485448031,
+      "learning_rate": 3.744679923406351e-05,
+      "loss": 0.5355,
+      "step": 303
+    },
+    {
+      "epoch": 1.2329113924050632,
+      "grad_norm": 0.7206358618384165,
+      "learning_rate": 3.741897971081043e-05,
+      "loss": 0.5328,
+      "step": 304
+    },
+    {
+      "epoch": 1.2369620253164557,
+      "grad_norm": 0.7709194022258278,
+      "learning_rate": 3.739101989748946e-05,
+      "loss": 0.532,
+      "step": 305
+    },
+    {
+      "epoch": 1.241012658227848,
+      "grad_norm": 0.7014059633502272,
+      "learning_rate": 3.7362920019285066e-05,
+      "loss": 0.5202,
+      "step": 306
+    },
+    {
+      "epoch": 1.2450632911392405,
+      "grad_norm": 0.800200097018151,
+      "learning_rate": 3.73346803025098e-05,
+      "loss": 0.5286,
+      "step": 307
+    },
+    {
+      "epoch": 1.2491139240506328,
+      "grad_norm": 0.7096095464738837,
+      "learning_rate": 3.730630097460247e-05,
+      "loss": 0.5243,
+      "step": 308
+    },
+    {
+      "epoch": 1.2531645569620253,
+      "grad_norm": 0.7306519564008936,
+      "learning_rate": 3.727778226412628e-05,
+      "loss": 0.5419,
+      "step": 309
+    },
+    {
+      "epoch": 1.2572151898734178,
+      "grad_norm": 0.660471977033729,
+      "learning_rate": 3.7249124400767006e-05,
+      "loss": 0.5202,
+      "step": 310
+    },
+    {
+      "epoch": 1.2612658227848101,
+      "grad_norm": 0.6801180199485616,
+      "learning_rate": 3.722032761533114e-05,
+      "loss": 0.5054,
+      "step": 311
+    },
+    {
+      "epoch": 1.2653164556962024,
+      "grad_norm": 0.6993904375101487,
+      "learning_rate": 3.719139213974403e-05,
+      "loss": 0.5384,
+      "step": 312
+    },
+    {
+      "epoch": 1.269367088607595,
+      "grad_norm": 0.6780235760377035,
+      "learning_rate": 3.7162318207048006e-05,
+      "loss": 0.5217,
+      "step": 313
+    },
+    {
+      "epoch": 1.2734177215189875,
+      "grad_norm": 0.702790212870474,
+      "learning_rate": 3.713310605140055e-05,
+      "loss": 0.5472,
+      "step": 314
+    },
+    {
+      "epoch": 1.2774683544303798,
+      "grad_norm": 0.6653195031906711,
+      "learning_rate": 3.710375590807233e-05,
+      "loss": 0.5405,
+      "step": 315
+    },
+    {
+      "epoch": 1.281518987341772,
+      "grad_norm": 0.7359972753002769,
+      "learning_rate": 3.7074268013445365e-05,
+      "loss": 0.5553,
+      "step": 316
+    },
+    {
+      "epoch": 1.2855696202531646,
+      "grad_norm": 0.7000096129582465,
+      "learning_rate": 3.7044642605011114e-05,
+      "loss": 0.5364,
+      "step": 317
+    },
+    {
+      "epoch": 1.289620253164557,
+      "grad_norm": 0.7508830851425659,
+      "learning_rate": 3.701487992136854e-05,
+      "loss": 0.5154,
+      "step": 318
+    },
+    {
+      "epoch": 1.2936708860759494,
+      "grad_norm": 0.7390988842810752,
+      "learning_rate": 3.69849802022222e-05,
+      "loss": 0.5212,
+      "step": 319
+    },
+    {
+      "epoch": 1.2977215189873417,
+      "grad_norm": 0.804500009730465,
+      "learning_rate": 3.6954943688380334e-05,
+      "loss": 0.522,
+      "step": 320
+    },
+    {
+      "epoch": 1.3017721518987342,
+      "grad_norm": 0.7587696176035242,
+      "learning_rate": 3.692477062175289e-05,
+      "loss": 0.539,
+      "step": 321
+    },
+    {
+      "epoch": 1.3058227848101267,
+      "grad_norm": 0.8234156774099743,
+      "learning_rate": 3.689446124534958e-05,
+      "loss": 0.5279,
+      "step": 322
+    },
+    {
+      "epoch": 1.309873417721519,
+      "grad_norm": 0.8580244375922627,
+      "learning_rate": 3.686401580327799e-05,
+      "loss": 0.5214,
+      "step": 323
+    },
+    {
+      "epoch": 1.3139240506329113,
+      "grad_norm": 0.7915890681887378,
+      "learning_rate": 3.683343454074149e-05,
+      "loss": 0.531,
+      "step": 324
+    },
+    {
+      "epoch": 1.3179746835443038,
+      "grad_norm": 0.8547107243679443,
+      "learning_rate": 3.6802717704037386e-05,
+      "loss": 0.537,
+      "step": 325
+    },
+    {
+      "epoch": 1.3220253164556963,
+      "grad_norm": 0.7484823729741003,
+      "learning_rate": 3.6771865540554855e-05,
+      "loss": 0.5107,
+      "step": 326
+    },
+    {
+      "epoch": 1.3260759493670886,
+      "grad_norm": 0.7503125446745016,
+      "learning_rate": 3.674087829877297e-05,
+      "loss": 0.5409,
+      "step": 327
+    },
+    {
+      "epoch": 1.330126582278481,
+      "grad_norm": 0.843416614556142,
+      "learning_rate": 3.6709756228258735e-05,
+      "loss": 0.521,
+      "step": 328
+    },
+    {
+      "epoch": 1.3341772151898734,
+      "grad_norm": 0.6577744513350768,
+      "learning_rate": 3.667849957966501e-05,
+      "loss": 0.5228,
+      "step": 329
+    },
+    {
+      "epoch": 1.338227848101266,
+      "grad_norm": 0.8835530052772871,
+      "learning_rate": 3.6647108604728546e-05,
+      "loss": 0.5388,
+      "step": 330
+    },
+    {
+      "epoch": 1.3422784810126582,
+      "grad_norm": 0.6368919763682271,
+      "learning_rate": 3.661558355626795e-05,
+      "loss": 0.5311,
+      "step": 331
+    },
+    {
+      "epoch": 1.3463291139240505,
+      "grad_norm": 0.7162512285611687,
+      "learning_rate": 3.658392468818163e-05,
+      "loss": 0.5282,
+      "step": 332
+    },
+    {
+      "epoch": 1.350379746835443,
+      "grad_norm": 0.6718622134321749,
+      "learning_rate": 3.655213225544574e-05,
+      "loss": 0.5316,
+      "step": 333
+    },
+    {
+      "epoch": 1.3544303797468356,
+      "grad_norm": 0.7250706775740503,
+      "learning_rate": 3.652020651411218e-05,
+      "loss": 0.5411,
+      "step": 334
+    },
+    {
+      "epoch": 1.3584810126582278,
+      "grad_norm": 0.7478876938819983,
+      "learning_rate": 3.6488147721306474e-05,
+      "loss": 0.5367,
+      "step": 335
+    },
+    {
+      "epoch": 1.3625316455696201,
+      "grad_norm": 0.6852879864467744,
+      "learning_rate": 3.645595613522574e-05,
+      "loss": 0.5392,
+      "step": 336
+    },
+    {
+      "epoch": 1.3665822784810127,
+      "grad_norm": 0.6721499188959996,
+      "learning_rate": 3.642363201513657e-05,
+      "loss": 0.5308,
+      "step": 337
+    },
+    {
+      "epoch": 1.3706329113924052,
+      "grad_norm": 0.7573140746267256,
+      "learning_rate": 3.6391175621373006e-05,
+      "loss": 0.5179,
+      "step": 338
+    },
+    {
+      "epoch": 1.3746835443037975,
+      "grad_norm": 0.6750364938556693,
+      "learning_rate": 3.6358587215334355e-05,
+      "loss": 0.5256,
+      "step": 339
+    },
+    {
+      "epoch": 1.3787341772151898,
+      "grad_norm": 0.6854265908885088,
+      "learning_rate": 3.632586705948318e-05,
+      "loss": 0.5369,
+      "step": 340
+    },
+    {
+      "epoch": 1.3827848101265823,
+      "grad_norm": 0.711498394960962,
+      "learning_rate": 3.629301541734311e-05,
+      "loss": 0.5173,
+      "step": 341
+    },
+    {
+      "epoch": 1.3868354430379748,
+      "grad_norm": 0.7522170790417387,
+      "learning_rate": 3.626003255349676e-05,
+      "loss": 0.5399,
+      "step": 342
+    },
+    {
+      "epoch": 1.390886075949367,
+      "grad_norm": 0.7074639265261163,
+      "learning_rate": 3.622691873358357e-05,
+      "loss": 0.5448,
+      "step": 343
+    },
+    {
+      "epoch": 1.3949367088607594,
+      "grad_norm": 0.7144564318166584,
+      "learning_rate": 3.61936742242977e-05,
+      "loss": 0.5257,
+      "step": 344
+    },
+    {
+      "epoch": 1.398987341772152,
+      "grad_norm": 0.7268486687524479,
+      "learning_rate": 3.6160299293385864e-05,
+      "loss": 0.5383,
+      "step": 345
+    },
+    {
+      "epoch": 1.4030379746835444,
+      "grad_norm": 0.6775452050677643,
+      "learning_rate": 3.612679420964516e-05,
+      "loss": 0.5398,
+      "step": 346
+    },
+    {
+      "epoch": 1.4070886075949367,
+      "grad_norm": 0.7318793542713288,
+      "learning_rate": 3.609315924292092e-05,
+      "loss": 0.5359,
+      "step": 347
+    },
+    {
+      "epoch": 1.411139240506329,
+      "grad_norm": 0.6859756782825144,
+      "learning_rate": 3.6059394664104554e-05,
+      "loss": 0.5296,
+      "step": 348
+    },
+    {
+      "epoch": 1.4151898734177215,
+      "grad_norm": 0.68414095552362,
+      "learning_rate": 3.602550074513133e-05,
+      "loss": 0.5324,
+      "step": 349
+    },
+    {
+      "epoch": 1.419240506329114,
+      "grad_norm": 0.613057025982154,
+      "learning_rate": 3.599147775897822e-05,
+      "loss": 0.5406,
+      "step": 350
+    },
+    {
+      "epoch": 1.4232911392405063,
+      "grad_norm": 0.6377083472046945,
+      "learning_rate": 3.595732597966167e-05,
+      "loss": 0.5397,
+      "step": 351
+    },
+    {
+      "epoch": 1.4273417721518986,
+      "grad_norm": 0.665236602524987,
+      "learning_rate": 3.592304568223542e-05,
+      "loss": 0.5216,
+      "step": 352
+    },
+    {
+      "epoch": 1.4313924050632911,
+      "grad_norm": 0.6606080358590196,
+      "learning_rate": 3.588863714278826e-05,
+      "loss": 0.5355,
+      "step": 353
+    },
+    {
+      "epoch": 1.4354430379746836,
+      "grad_norm": 0.6904466341019709,
+      "learning_rate": 3.585410063844186e-05,
+      "loss": 0.5127,
+      "step": 354
+    },
+    {
+      "epoch": 1.439493670886076,
+      "grad_norm": 0.6895591238474661,
+      "learning_rate": 3.581943644734846e-05,
+      "loss": 0.5224,
+      "step": 355
+    },
+    {
+      "epoch": 1.4435443037974682,
+      "grad_norm": 0.7273692590207156,
+      "learning_rate": 3.578464484868869e-05,
+      "loss": 0.504,
+      "step": 356
+    },
+    {
+      "epoch": 1.4475949367088607,
+      "grad_norm": 0.6429519993949224,
+      "learning_rate": 3.5749726122669316e-05,
+      "loss": 0.5271,
+      "step": 357
+    },
+    {
+      "epoch": 1.4516455696202533,
+      "grad_norm": 0.7135283509143735,
+      "learning_rate": 3.5714680550520943e-05,
+      "loss": 0.5514,
+      "step": 358
+    },
+    {
+      "epoch": 1.4556962025316456,
+      "grad_norm": 0.6683876398910585,
+      "learning_rate": 3.5679508414495794e-05,
+      "loss": 0.5228,
+      "step": 359
+    },
+    {
+      "epoch": 1.4597468354430378,
+      "grad_norm": 0.7155928592764508,
+      "learning_rate": 3.564420999786543e-05,
+      "loss": 0.5413,
+      "step": 360
+    },
+    {
+      "epoch": 1.4637974683544304,
+      "grad_norm": 0.6738401474956277,
+      "learning_rate": 3.560878558491842e-05,
+      "loss": 0.5114,
+      "step": 361
+    },
+    {
+      "epoch": 1.4678481012658229,
+      "grad_norm": 0.732536191887441,
+      "learning_rate": 3.5573235460958145e-05,
+      "loss": 0.5409,
+      "step": 362
+    },
+    {
+      "epoch": 1.4718987341772152,
+      "grad_norm": 0.6558760792477842,
+      "learning_rate": 3.553755991230039e-05,
+      "loss": 0.5381,
+      "step": 363
+    },
+    {
+      "epoch": 1.4759493670886075,
+      "grad_norm": 0.7478628551178501,
+      "learning_rate": 3.5501759226271144e-05,
+      "loss": 0.5314,
+      "step": 364
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.6228172876238911,
+      "learning_rate": 3.546583369120419e-05,
+      "loss": 0.5249,
+      "step": 365
+    },
+    {
+      "epoch": 1.4840506329113925,
+      "grad_norm": 0.6707993880359097,
+      "learning_rate": 3.5429783596438864e-05,
+      "loss": 0.5323,
+      "step": 366
+    },
+    {
+      "epoch": 1.4881012658227848,
+      "grad_norm": 0.655482865300226,
+      "learning_rate": 3.539360923231766e-05,
+      "loss": 0.5449,
+      "step": 367
+    },
+    {
+      "epoch": 1.492151898734177,
+      "grad_norm": 0.6244418525101414,
+      "learning_rate": 3.535731089018394e-05,
+      "loss": 0.5264,
+      "step": 368
+    },
+    {
+      "epoch": 1.4962025316455696,
+      "grad_norm": 0.6469200862308522,
+      "learning_rate": 3.532088886237956e-05,
+      "loss": 0.5281,
+      "step": 369
+    },
+    {
+      "epoch": 1.5002531645569621,
+      "grad_norm": 0.6699153401940771,
+      "learning_rate": 3.528434344224253e-05,
+      "loss": 0.5217,
+      "step": 370
+    },
+    {
+      "epoch": 1.5043037974683544,
+      "grad_norm": 0.6891353571106934,
+      "learning_rate": 3.524767492410464e-05,
+      "loss": 0.5368,
+      "step": 371
+    },
+    {
+      "epoch": 1.5083544303797467,
+      "grad_norm": 0.6670125836747715,
+      "learning_rate": 3.521088360328908e-05,
+      "loss": 0.5332,
+      "step": 372
+    },
+    {
+      "epoch": 1.5124050632911392,
+      "grad_norm": 0.7057240377315893,
+      "learning_rate": 3.517396977610811e-05,
+      "loss": 0.5208,
+      "step": 373
+    },
+    {
+      "epoch": 1.5164556962025317,
+      "grad_norm": 0.7327796161863389,
+      "learning_rate": 3.5136933739860595e-05,
+      "loss": 0.5468,
+      "step": 374
+    },
+    {
+      "epoch": 1.520506329113924,
+      "grad_norm": 0.6753369184184024,
+      "learning_rate": 3.509977579282971e-05,
+      "loss": 0.5399,
+      "step": 375
+    },
+    {
+      "epoch": 1.5245569620253163,
+      "grad_norm": 0.7478144960743418,
+      "learning_rate": 3.5062496234280424e-05,
+      "loss": 0.5195,
+      "step": 376
+    },
+    {
+      "epoch": 1.5286075949367088,
+      "grad_norm": 0.6737933609533807,
+      "learning_rate": 3.502509536445719e-05,
+      "loss": 0.5303,
+      "step": 377
+    },
+    {
+      "epoch": 1.5326582278481014,
+      "grad_norm": 0.7185535637993247,
+      "learning_rate": 3.498757348458147e-05,
+      "loss": 0.5436,
+      "step": 378
+    },
+    {
+      "epoch": 1.5367088607594936,
+      "grad_norm": 0.7183583417925914,
+      "learning_rate": 3.4949930896849324e-05,
+      "loss": 0.5278,
+      "step": 379
+    },
+    {
+      "epoch": 1.540759493670886,
+      "grad_norm": 0.6928189876192659,
+      "learning_rate": 3.491216790442899e-05,
+      "loss": 0.5294,
+      "step": 380
+    },
+    {
+      "epoch": 1.5448101265822785,
+      "grad_norm": 0.6862300775380987,
+      "learning_rate": 3.487428481145839e-05,
+      "loss": 0.5263,
+      "step": 381
+    },
+    {
+      "epoch": 1.548860759493671,
+      "grad_norm": 0.6315972715511865,
+      "learning_rate": 3.483628192304278e-05,
+      "loss": 0.5143,
+      "step": 382
+    },
+    {
+      "epoch": 1.5529113924050633,
+      "grad_norm": 0.7331343116357,
+      "learning_rate": 3.479815954525219e-05,
+      "loss": 0.5337,
+      "step": 383
+    },
+    {
+      "epoch": 1.5569620253164556,
+      "grad_norm": 0.6106010610684125,
+      "learning_rate": 3.475991798511899e-05,
+      "loss": 0.5393,
+      "step": 384
+    },
+    {
+      "epoch": 1.561012658227848,
+      "grad_norm": 0.6086694962930539,
+      "learning_rate": 3.4721557550635464e-05,
+      "loss": 0.5237,
+      "step": 385
+    },
+    {
+      "epoch": 1.5650632911392406,
+      "grad_norm": 0.7382887643865306,
+      "learning_rate": 3.468307855075128e-05,
+      "loss": 0.5275,
+      "step": 386
+    },
+    {
+      "epoch": 1.5691139240506329,
+      "grad_norm": 0.631018387270069,
+      "learning_rate": 3.4644481295371005e-05,
+      "loss": 0.5359,
+      "step": 387
+    },
+    {
+      "epoch": 1.5731645569620252,
+      "grad_norm": 0.693616832669688,
+      "learning_rate": 3.460576609535163e-05,
+      "loss": 0.538,
+      "step": 388
+    },
+    {
+      "epoch": 1.5772151898734177,
+      "grad_norm": 0.7301350162734437,
+      "learning_rate": 3.456693326250006e-05,
+      "loss": 0.5419,
+      "step": 389
+    },
+    {
+      "epoch": 1.5812658227848102,
+      "grad_norm": 0.6801657484337921,
+      "learning_rate": 3.452798310957058e-05,
+      "loss": 0.5253,
+      "step": 390
+    },
+    {
+      "epoch": 1.5853164556962025,
+      "grad_norm": 0.6558640845767176,
+      "learning_rate": 3.4488915950262386e-05,
+      "loss": 0.5425,
+      "step": 391
+    },
+    {
+      "epoch": 1.5893670886075948,
+      "grad_norm": 0.71843457418095,
+      "learning_rate": 3.4449732099216985e-05,
+      "loss": 0.5262,
+      "step": 392
+    },
+    {
+      "epoch": 1.5934177215189873,
+      "grad_norm": 0.6662167206859232,
+      "learning_rate": 3.441043187201574e-05,
+      "loss": 0.5472,
+      "step": 393
+    },
+    {
+      "epoch": 1.5974683544303798,
+      "grad_norm": 0.6789139188291285,
+      "learning_rate": 3.437101558517728e-05,
+      "loss": 0.5104,
+      "step": 394
+    },
+    {
+      "epoch": 1.6015189873417721,
+      "grad_norm": 0.7249146290949106,
+      "learning_rate": 3.433148355615496e-05,
+      "loss": 0.5178,
+      "step": 395
+    },
+    {
+      "epoch": 1.6055696202531644,
+      "grad_norm": 0.6481751750827071,
+      "learning_rate": 3.4291836103334294e-05,
+      "loss": 0.5287,
+      "step": 396
+    },
+    {
+      "epoch": 1.609620253164557,
+      "grad_norm": 0.6213912783890521,
+      "learning_rate": 3.425207354603043e-05,
+      "loss": 0.5405,
+      "step": 397
+    },
+    {
+      "epoch": 1.6136708860759494,
+      "grad_norm": 0.6676637884919397,
+      "learning_rate": 3.421219620448553e-05,
+      "loss": 0.5302,
+      "step": 398
+    },
+    {
+      "epoch": 1.6177215189873417,
+      "grad_norm": 0.6601772043849077,
+      "learning_rate": 3.417220439986623e-05,
+      "loss": 0.5061,
+      "step": 399
+    },
+    {
+      "epoch": 1.621772151898734,
+      "grad_norm": 0.6683258248732366,
+      "learning_rate": 3.4132098454261024e-05,
+      "loss": 0.5228,
+      "step": 400
+    },
+    {
+      "epoch": 1.6258227848101265,
+      "grad_norm": 0.7845457634576957,
+      "learning_rate": 3.4091878690677676e-05,
+      "loss": 0.5363,
+      "step": 401
+    },
+    {
+      "epoch": 1.629873417721519,
+      "grad_norm": 0.6631395186343568,
+      "learning_rate": 3.405154543304065e-05,
+      "loss": 0.529,
+      "step": 402
+    },
+    {
+      "epoch": 1.6339240506329114,
+      "grad_norm": 0.6590175789882661,
+      "learning_rate": 3.401109900618843e-05,
+      "loss": 0.5276,
+      "step": 403
+    },
+    {
+      "epoch": 1.6379746835443036,
+      "grad_norm": 0.804021938597579,
+      "learning_rate": 3.3970539735870996e-05,
+      "loss": 0.5381,
+      "step": 404
+    },
+    {
+      "epoch": 1.6420253164556962,
+      "grad_norm": 0.7043686654659039,
+      "learning_rate": 3.392986794874714e-05,
+      "loss": 0.5182,
+      "step": 405
+    },
+    {
+      "epoch": 1.6460759493670887,
+      "grad_norm": 0.7808279582005543,
+      "learning_rate": 3.388908397238184e-05,
+      "loss": 0.5365,
+      "step": 406
+    },
+    {
+      "epoch": 1.650126582278481,
+      "grad_norm": 0.6871277890105988,
+      "learning_rate": 3.384818813524362e-05,
+      "loss": 0.557,
+      "step": 407
+    },
+    {
+      "epoch": 1.6541772151898733,
+      "grad_norm": 0.6398068078438307,
+      "learning_rate": 3.380718076670195e-05,
+      "loss": 0.5349,
+      "step": 408
+    },
+    {
+      "epoch": 1.6582278481012658,
+      "grad_norm": 0.7123870815908797,
+      "learning_rate": 3.376606219702454e-05,
+      "loss": 0.5212,
+      "step": 409
+    },
+    {
+      "epoch": 1.6622784810126583,
+      "grad_norm": 0.6458659466192169,
+      "learning_rate": 3.372483275737468e-05,
+      "loss": 0.5183,
+      "step": 410
+    },
+    {
+      "epoch": 1.6663291139240506,
+      "grad_norm": 0.6865141733159028,
+      "learning_rate": 3.368349277980861e-05,
+      "loss": 0.5312,
+      "step": 411
+    },
+    {
+      "epoch": 1.6703797468354429,
+      "grad_norm": 0.7020996148531212,
+      "learning_rate": 3.3642042597272844e-05,
+      "loss": 0.5229,
+      "step": 412
+    },
+    {
+      "epoch": 1.6744303797468354,
+      "grad_norm": 0.7013373527235036,
+      "learning_rate": 3.360048254360144e-05,
+      "loss": 0.5145,
+      "step": 413
+    },
+    {
+      "epoch": 1.678481012658228,
+      "grad_norm": 0.6807397783547273,
+      "learning_rate": 3.355881295351336e-05,
+      "loss": 0.5225,
+      "step": 414
+    },
+    {
+      "epoch": 1.6825316455696202,
+      "grad_norm": 0.7815850262758718,
+      "learning_rate": 3.351703416260975e-05,
+      "loss": 0.5256,
+      "step": 415
+    },
+    {
+      "epoch": 1.6865822784810125,
+      "grad_norm": 0.6092038448657008,
+      "learning_rate": 3.347514650737126e-05,
+      "loss": 0.5099,
+      "step": 416
+    },
+    {
+      "epoch": 1.690632911392405,
+      "grad_norm": 0.7951297689860264,
+      "learning_rate": 3.3433150325155295e-05,
+      "loss": 0.5337,
+      "step": 417
+    },
+    {
+      "epoch": 1.6946835443037975,
+      "grad_norm": 0.656051867473319,
+      "learning_rate": 3.339104595419334e-05,
+      "loss": 0.5381,
+      "step": 418
+    },
+    {
+      "epoch": 1.6987341772151898,
+      "grad_norm": 0.6760921969646709,
+      "learning_rate": 3.3348833733588204e-05,
+      "loss": 0.5214,
+      "step": 419
+    },
+    {
+      "epoch": 1.7027848101265821,
+      "grad_norm": 0.6599622191231304,
+      "learning_rate": 3.3306514003311305e-05,
+      "loss": 0.5098,
+      "step": 420
+    },
+    {
+      "epoch": 1.7068354430379746,
+      "grad_norm": 0.7085663992394153,
+      "learning_rate": 3.326408710419996e-05,
+      "loss": 0.5191,
+      "step": 421
+    },
+    {
+      "epoch": 1.7108860759493671,
+      "grad_norm": 0.6477426790384693,
+      "learning_rate": 3.322155337795454e-05,
+      "loss": 0.5204,
+      "step": 422
+    },
+    {
+      "epoch": 1.7149367088607594,
+      "grad_norm": 0.6418552217118475,
+      "learning_rate": 3.317891316713587e-05,
+      "loss": 0.5377,
+      "step": 423
+    },
+    {
+      "epoch": 1.7189873417721517,
+      "grad_norm": 0.6154455568315832,
+      "learning_rate": 3.313616681516231e-05,
+      "loss": 0.5313,
+      "step": 424
+    },
+    {
+      "epoch": 1.7230379746835442,
+      "grad_norm": 0.6721894153940952,
+      "learning_rate": 3.309331466630713e-05,
+      "loss": 0.5257,
+      "step": 425
+    },
+    {
+      "epoch": 1.7270886075949368,
+      "grad_norm": 0.6448637485464261,
+      "learning_rate": 3.305035706569563e-05,
+      "loss": 0.5195,
+      "step": 426
+    },
+    {
+      "epoch": 1.7311392405063293,
+      "grad_norm": 0.6684171391989575,
+      "learning_rate": 3.3007294359302433e-05,
+      "loss": 0.5311,
+      "step": 427
+    },
+    {
+      "epoch": 1.7351898734177216,
+      "grad_norm": 0.627461674012358,
+      "learning_rate": 3.296412689394864e-05,
+      "loss": 0.5133,
+      "step": 428
+    },
+    {
+      "epoch": 1.7392405063291139,
+      "grad_norm": 0.6526464159922001,
+      "learning_rate": 3.292085501729909e-05,
+      "loss": 0.5205,
+      "step": 429
+    },
+    {
+      "epoch": 1.7432911392405064,
+      "grad_norm": 0.6720406247790295,
+      "learning_rate": 3.2877479077859534e-05,
+      "loss": 0.5447,
+      "step": 430
+    },
+    {
+      "epoch": 1.747341772151899,
+      "grad_norm": 0.7009519355771469,
+      "learning_rate": 3.283399942497381e-05,
+      "loss": 0.5293,
+      "step": 431
+    },
+    {
+      "epoch": 1.7513924050632912,
+      "grad_norm": 0.6497805963682699,
+      "learning_rate": 3.279041640882108e-05,
+      "loss": 0.5127,
+      "step": 432
+    },
+    {
+      "epoch": 1.7554430379746835,
+      "grad_norm": 0.6203796903522117,
+      "learning_rate": 3.2746730380412964e-05,
+      "loss": 0.5043,
+      "step": 433
+    },
+    {
+      "epoch": 1.759493670886076,
+      "grad_norm": 0.6956945341240407,
+      "learning_rate": 3.2702941691590726e-05,
+      "loss": 0.525,
+      "step": 434
+    },
+    {
+      "epoch": 1.7635443037974685,
+      "grad_norm": 0.6543199034850877,
+      "learning_rate": 3.265905069502244e-05,
+      "loss": 0.519,
+      "step": 435
+    },
+    {
+      "epoch": 1.7675949367088608,
+      "grad_norm": 0.6490235066717059,
+      "learning_rate": 3.261505774420016e-05,
+      "loss": 0.5049,
+      "step": 436
+    },
+    {
+      "epoch": 1.771645569620253,
+      "grad_norm": 0.647501841942791,
+      "learning_rate": 3.257096319343707e-05,
+      "loss": 0.5331,
+      "step": 437
+    },
+    {
+      "epoch": 1.7756962025316456,
+      "grad_norm": 0.6246481028662274,
+      "learning_rate": 3.2526767397864614e-05,
+      "loss": 0.5272,
+      "step": 438
+    },
+    {
+      "epoch": 1.7797468354430381,
+      "grad_norm": 0.6628180692979595,
+      "learning_rate": 3.248247071342966e-05,
+      "loss": 0.5355,
+      "step": 439
+    },
+    {
+      "epoch": 1.7837974683544304,
+      "grad_norm": 0.6612407131411471,
+      "learning_rate": 3.243807349689161e-05,
+      "loss": 0.526,
+      "step": 440
+    },
+    {
+      "epoch": 1.7878481012658227,
+      "grad_norm": 0.6672038785950684,
+      "learning_rate": 3.2393576105819544e-05,
+      "loss": 0.5377,
+      "step": 441
+    },
+    {
+      "epoch": 1.7918987341772152,
+      "grad_norm": 0.6884691351694165,
+      "learning_rate": 3.2348978898589333e-05,
+      "loss": 0.5388,
+      "step": 442
+    },
+    {
+      "epoch": 1.7959493670886078,
+      "grad_norm": 0.6719270472611492,
+      "learning_rate": 3.230428223438075e-05,
+      "loss": 0.5003,
+      "step": 443
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.6480729999295298,
+      "learning_rate": 3.225948647317459e-05,
+      "loss": 0.5089,
+      "step": 444
+    },
+    {
+      "epoch": 1.8040506329113923,
+      "grad_norm": 0.6304743201218572,
+      "learning_rate": 3.2214591975749745e-05,
+      "loss": 0.5113,
+      "step": 445
+    },
+    {
+      "epoch": 1.8081012658227849,
+      "grad_norm": 0.6958008052523298,
+      "learning_rate": 3.216959910368034e-05,
+      "loss": 0.5212,
+      "step": 446
+    },
+    {
+      "epoch": 1.8121518987341774,
+      "grad_norm": 0.6211879441167298,
+      "learning_rate": 3.212450821933277e-05,
+      "loss": 0.5281,
+      "step": 447
+    },
+    {
+      "epoch": 1.8162025316455697,
+      "grad_norm": 0.6561004129241005,
+      "learning_rate": 3.207931968586281e-05,
+      "loss": 0.525,
+      "step": 448
+    },
+    {
+      "epoch": 1.820253164556962,
+      "grad_norm": 0.6881798305303279,
+      "learning_rate": 3.203403386721272e-05,
+      "loss": 0.5296,
+      "step": 449
+    },
+    {
+      "epoch": 1.8243037974683545,
+      "grad_norm": 0.6215500615501516,
+      "learning_rate": 3.1988651128108245e-05,
+      "loss": 0.5042,
+      "step": 450
+    },
+    {
+      "epoch": 1.828354430379747,
+      "grad_norm": 0.6849482447180655,
+      "learning_rate": 3.194317183405573e-05,
+      "loss": 0.5202,
+      "step": 451
+    },
+    {
+      "epoch": 1.8324050632911393,
+      "grad_norm": 0.6681449774813805,
+      "learning_rate": 3.189759635133914e-05,
+      "loss": 0.5234,
+      "step": 452
+    },
+    {
+      "epoch": 1.8364556962025316,
+      "grad_norm": 0.6133007901631723,
+      "learning_rate": 3.185192504701718e-05,
+      "loss": 0.5152,
+      "step": 453
+    },
+    {
+      "epoch": 1.840506329113924,
+      "grad_norm": 0.6484936950792196,
+      "learning_rate": 3.1806158288920234e-05,
+      "loss": 0.5278,
+      "step": 454
+    },
+    {
+      "epoch": 1.8445569620253166,
+      "grad_norm": 0.6702182339406958,
+      "learning_rate": 3.1760296445647477e-05,
+      "loss": 0.5227,
+      "step": 455
+    },
+    {
+      "epoch": 1.848607594936709,
+      "grad_norm": 0.6253174450278614,
+      "learning_rate": 3.1714339886563896e-05,
+      "loss": 0.5219,
+      "step": 456
+    },
+    {
+      "epoch": 1.8526582278481012,
+      "grad_norm": 0.6518332856826008,
+      "learning_rate": 3.166828898179731e-05,
+      "loss": 0.5313,
+      "step": 457
+    },
+    {
+      "epoch": 1.8567088607594937,
+      "grad_norm": 0.6106613630478455,
+      "learning_rate": 3.162214410223536e-05,
+      "loss": 0.5298,
+      "step": 458
+    },
+    {
+      "epoch": 1.8607594936708862,
+      "grad_norm": 0.6000448994210339,
+      "learning_rate": 3.157590561952257e-05,
+      "loss": 0.49,
+      "step": 459
+    },
+    {
+      "epoch": 1.8648101265822785,
+      "grad_norm": 0.6396193264535383,
+      "learning_rate": 3.152957390605732e-05,
+      "loss": 0.5421,
+      "step": 460
+    },
+    {
+      "epoch": 1.8688607594936708,
+      "grad_norm": 0.6397720653724658,
+      "learning_rate": 3.148314933498886e-05,
+      "loss": 0.5155,
+      "step": 461
+    },
+    {
+      "epoch": 1.8729113924050633,
+      "grad_norm": 0.6582960119924623,
+      "learning_rate": 3.143663228021431e-05,
+      "loss": 0.5278,
+      "step": 462
+    },
+    {
+      "epoch": 1.8769620253164558,
+      "grad_norm": 0.7086723077692496,
+      "learning_rate": 3.1390023116375624e-05,
+      "loss": 0.5411,
+      "step": 463
+    },
+    {
+      "epoch": 1.8810126582278481,
+      "grad_norm": 0.6080637396236079,
+      "learning_rate": 3.134332221885661e-05,
+      "loss": 0.5188,
+      "step": 464
+    },
+    {
+      "epoch": 1.8850632911392404,
+      "grad_norm": 0.640837574327225,
+      "learning_rate": 3.129652996377987e-05,
+      "loss": 0.5291,
+      "step": 465
+    },
+    {
+      "epoch": 1.889113924050633,
+      "grad_norm": 0.6233050562531265,
+      "learning_rate": 3.12496467280038e-05,
+      "loss": 0.5089,
+      "step": 466
+    },
+    {
+      "epoch": 1.8931645569620255,
+      "grad_norm": 0.6353880133634127,
+      "learning_rate": 3.120267288911952e-05,
+      "loss": 0.5123,
+      "step": 467
+    },
+    {
+      "epoch": 1.8972151898734178,
+      "grad_norm": 0.6051787988009817,
+      "learning_rate": 3.11556088254479e-05,
+      "loss": 0.536,
+      "step": 468
+    },
+    {
+      "epoch": 1.90126582278481,
+      "grad_norm": 0.6226987956013583,
+      "learning_rate": 3.11084549160364e-05,
+      "loss": 0.4967,
+      "step": 469
+    },
+    {
+      "epoch": 1.9053164556962026,
+      "grad_norm": 0.6187262500717698,
+      "learning_rate": 3.106121154065615e-05,
+      "loss": 0.5158,
+      "step": 470
+    },
+    {
+      "epoch": 1.909367088607595,
+      "grad_norm": 0.6718049137861567,
+      "learning_rate": 3.1013879079798805e-05,
+      "loss": 0.5124,
+      "step": 471
+    },
+    {
+      "epoch": 1.9134177215189874,
+      "grad_norm": 0.6047820575185009,
+      "learning_rate": 3.096645791467348e-05,
+      "loss": 0.5091,
+      "step": 472
+    },
+    {
+      "epoch": 1.9174683544303797,
+      "grad_norm": 0.6494743558779025,
+      "learning_rate": 3.091894842720373e-05,
+      "loss": 0.529,
+      "step": 473
+    },
+    {
+      "epoch": 1.9215189873417722,
+      "grad_norm": 0.5955449619180364,
+      "learning_rate": 3.0871351000024425e-05,
+      "loss": 0.5271,
+      "step": 474
+    },
+    {
+      "epoch": 1.9255696202531647,
+      "grad_norm": 0.6350441105265189,
+      "learning_rate": 3.0823666016478716e-05,
+      "loss": 0.5149,
+      "step": 475
+    },
+    {
+      "epoch": 1.929620253164557,
+      "grad_norm": 0.6191621105969239,
+      "learning_rate": 3.0775893860614896e-05,
+      "loss": 0.5326,
+      "step": 476
+    },
+    {
+      "epoch": 1.9336708860759493,
+      "grad_norm": 0.6031896307296335,
+      "learning_rate": 3.0728034917183336e-05,
+      "loss": 0.5162,
+      "step": 477
+    },
+    {
+      "epoch": 1.9377215189873418,
+      "grad_norm": 0.6158415586646282,
+      "learning_rate": 3.06800895716334e-05,
+      "loss": 0.5165,
+      "step": 478
+    },
+    {
+      "epoch": 1.9417721518987343,
+      "grad_norm": 0.6194958868585683,
+      "learning_rate": 3.063205821011029e-05,
+      "loss": 0.5157,
+      "step": 479
+    },
+    {
+      "epoch": 1.9458227848101266,
+      "grad_norm": 0.6254917796571432,
+      "learning_rate": 3.0583941219452016e-05,
+      "loss": 0.5115,
+      "step": 480
+    },
+    {
+      "epoch": 1.949873417721519,
+      "grad_norm": 0.6529870830368293,
+      "learning_rate": 3.053573898718618e-05,
+      "loss": 0.5064,
+      "step": 481
+    },
+    {
+      "epoch": 1.9539240506329114,
+      "grad_norm": 0.6513339635176223,
+      "learning_rate": 3.0487451901526956e-05,
+      "loss": 0.5363,
+      "step": 482
+    },
+    {
+      "epoch": 1.957974683544304,
+      "grad_norm": 0.6084180044365278,
+      "learning_rate": 3.0439080351371875e-05,
+      "loss": 0.5183,
+      "step": 483
+    },
+    {
+      "epoch": 1.9620253164556962,
+      "grad_norm": 0.5980112472276721,
+      "learning_rate": 3.0390624726298764e-05,
+      "loss": 0.5257,
+      "step": 484
+    },
+    {
+      "epoch": 1.9660759493670885,
+      "grad_norm": 0.6588997206328401,
+      "learning_rate": 3.034208541656255e-05,
+      "loss": 0.5321,
+      "step": 485
+    },
+    {
+      "epoch": 1.970126582278481,
+      "grad_norm": 0.6433086344222481,
+      "learning_rate": 3.029346281309218e-05,
+      "loss": 0.5155,
+      "step": 486
+    },
+    {
+      "epoch": 1.9741772151898735,
+      "grad_norm": 0.5991004735063122,
+      "learning_rate": 3.0244757307487415e-05,
+      "loss": 0.531,
+      "step": 487
+    },
+    {
+      "epoch": 1.9782278481012658,
+      "grad_norm": 0.6094758307536475,
+      "learning_rate": 3.019596929201569e-05,
+      "loss": 0.5279,
+      "step": 488
+    },
+    {
+      "epoch": 1.9822784810126581,
+      "grad_norm": 0.5971377268301424,
+      "learning_rate": 3.0147099159608985e-05,
+      "loss": 0.5148,
+      "step": 489
+    },
+    {
+      "epoch": 1.9863291139240506,
+      "grad_norm": 0.6280801772916395,
+      "learning_rate": 3.0098147303860616e-05,
+      "loss": 0.5095,
+      "step": 490
+    },
+    {
+      "epoch": 1.9903797468354432,
+      "grad_norm": 0.6002200462831045,
+      "learning_rate": 3.0049114119022117e-05,
+      "loss": 0.5224,
+      "step": 491
+    },
+    {
+      "epoch": 1.9944303797468355,
+      "grad_norm": 0.64598386219167,
+      "learning_rate": 3.0000000000000004e-05,
+      "loss": 0.5215,
+      "step": 492
+    },
+    {
+      "epoch": 1.9984810126582278,
+      "grad_norm": 0.607169976688393,
+      "learning_rate": 2.995080534235264e-05,
+      "loss": 0.5048,
+      "step": 493
+    },
+    {
+      "epoch": 2.0040506329113925,
+      "grad_norm": 0.9403998964508724,
+      "learning_rate": 2.9901530542287044e-05,
+      "loss": 0.3335,
+      "step": 494
+    },
+    {
+      "epoch": 2.008101265822785,
+      "grad_norm": 0.6121347400868989,
+      "learning_rate": 2.9852175996655676e-05,
+      "loss": 0.3117,
+      "step": 495
+    },
+    {
+      "epoch": 2.012151898734177,
+      "grad_norm": 1.369755779521615,
+      "learning_rate": 2.980274210295326e-05,
+      "loss": 0.3298,
+      "step": 496
+    },
+    {
+      "epoch": 2.0162025316455696,
+      "grad_norm": 0.9141450003607747,
+      "learning_rate": 2.9753229259313578e-05,
+      "loss": 0.3308,
+      "step": 497
+    },
+    {
+      "epoch": 2.020253164556962,
+      "grad_norm": 0.5942578047062798,
+      "learning_rate": 2.9703637864506274e-05,
+      "loss": 0.3154,
+      "step": 498
+    },
+    {
+      "epoch": 2.0243037974683546,
+      "grad_norm": 0.6555146007062196,
+      "learning_rate": 2.965396831793362e-05,
+      "loss": 0.3147,
+      "step": 499
+    },
+    {
+      "epoch": 2.0283544303797467,
+      "grad_norm": 0.6306509868241879,
+      "learning_rate": 2.9604221019627316e-05,
+      "loss": 0.3158,
+      "step": 500
+    },
+    {
+      "epoch": 2.0324050632911392,
+      "grad_norm": 0.6081298799931152,
+      "learning_rate": 2.955439637024526e-05,
+      "loss": 0.3103,
+      "step": 501
+    },
+    {
+      "epoch": 2.0364556962025318,
+      "grad_norm": 0.6348090562675244,
+      "learning_rate": 2.9504494771068334e-05,
+      "loss": 0.3221,
+      "step": 502
+    },
+    {
+      "epoch": 2.0405063291139243,
+      "grad_norm": 0.6436762747808167,
+      "learning_rate": 2.9454516623997156e-05,
+      "loss": 0.3294,
+      "step": 503
+    },
+    {
+      "epoch": 2.0445569620253163,
+      "grad_norm": 0.661662547369622,
+      "learning_rate": 2.9404462331548847e-05,
+      "loss": 0.3279,
+      "step": 504
+    },
+    {
+      "epoch": 2.048607594936709,
+      "grad_norm": 0.6543847135091998,
+      "learning_rate": 2.93543322968538e-05,
+      "loss": 0.3086,
+      "step": 505
+    },
+    {
+      "epoch": 2.0526582278481014,
+      "grad_norm": 0.6001626399709417,
+      "learning_rate": 2.9304126923652428e-05,
+      "loss": 0.3184,
+      "step": 506
+    },
+    {
+      "epoch": 2.056708860759494,
+      "grad_norm": 0.6451304466220742,
+      "learning_rate": 2.9253846616291896e-05,
+      "loss": 0.3223,
+      "step": 507
+    },
+    {
+      "epoch": 2.060759493670886,
+      "grad_norm": 0.6119951019250512,
+      "learning_rate": 2.9203491779722896e-05,
+      "loss": 0.3328,
+      "step": 508
+    },
+    {
+      "epoch": 2.0648101265822785,
+      "grad_norm": 0.6022780956580829,
+      "learning_rate": 2.9153062819496357e-05,
+      "loss": 0.3017,
+      "step": 509
+    },
+    {
+      "epoch": 2.068860759493671,
+      "grad_norm": 0.5920853260839704,
+      "learning_rate": 2.9102560141760178e-05,
+      "loss": 0.3286,
+      "step": 510
+    },
+    {
+      "epoch": 2.0729113924050635,
+      "grad_norm": 0.6243584413249846,
+      "learning_rate": 2.9051984153256004e-05,
+      "loss": 0.3224,
+      "step": 511
+    },
+    {
+      "epoch": 2.0769620253164556,
+      "grad_norm": 0.5955698611027065,
+      "learning_rate": 2.900133526131588e-05,
+      "loss": 0.3084,
+      "step": 512
+    },
+    {
+      "epoch": 2.081012658227848,
+      "grad_norm": 0.6401173316814759,
+      "learning_rate": 2.8950613873859025e-05,
+      "loss": 0.2989,
+      "step": 513
+    },
+    {
+      "epoch": 2.0850632911392406,
+      "grad_norm": 0.6295660698130163,
+      "learning_rate": 2.8899820399388515e-05,
+      "loss": 0.3355,
+      "step": 514
+    },
+    {
+      "epoch": 2.089113924050633,
+      "grad_norm": 0.6431761752056766,
+      "learning_rate": 2.8848955246988012e-05,
+      "loss": 0.3132,
+      "step": 515
+    },
+    {
+      "epoch": 2.093164556962025,
+      "grad_norm": 0.643888111656521,
+      "learning_rate": 2.879801882631847e-05,
+      "loss": 0.3215,
+      "step": 516
+    },
+    {
+      "epoch": 2.0972151898734177,
+      "grad_norm": 0.6117008902759047,
+      "learning_rate": 2.8747011547614808e-05,
+      "loss": 0.3132,
+      "step": 517
+    },
+    {
+      "epoch": 2.1012658227848102,
+      "grad_norm": 0.6496242598041817,
+      "learning_rate": 2.8695933821682635e-05,
+      "loss": 0.3452,
+      "step": 518
+    },
+    {
+      "epoch": 2.1053164556962027,
+      "grad_norm": 0.6200996030006051,
+      "learning_rate": 2.864478605989494e-05,
+      "loss": 0.3129,
+      "step": 519
+    },
+    {
+      "epoch": 2.109367088607595,
+      "grad_norm": 0.6108134520105635,
+      "learning_rate": 2.8593568674188765e-05,
+      "loss": 0.3228,
+      "step": 520
+    },
+    {
+      "epoch": 2.1134177215189873,
+      "grad_norm": 0.6368630442254329,
+      "learning_rate": 2.8542282077061892e-05,
+      "loss": 0.3101,
+      "step": 521
+    },
+    {
+      "epoch": 2.11746835443038,
+      "grad_norm": 0.5932056264507057,
+      "learning_rate": 2.8490926681569523e-05,
+      "loss": 0.3226,
+      "step": 522
+    },
+    {
+      "epoch": 2.1215189873417724,
+      "grad_norm": 0.6095498627914768,
+      "learning_rate": 2.8439502901320956e-05,
+      "loss": 0.2977,
+      "step": 523
+    },
+    {
+      "epoch": 2.1255696202531644,
+      "grad_norm": 0.6323549055376018,
+      "learning_rate": 2.8388011150476237e-05,
+      "loss": 0.3114,
+      "step": 524
+    },
+    {
+      "epoch": 2.129620253164557,
+      "grad_norm": 0.5826437284226451,
+      "learning_rate": 2.8336451843742866e-05,
+      "loss": 0.3053,
+      "step": 525
+    },
+    {
+      "epoch": 2.1336708860759495,
+      "grad_norm": 0.6044663856299161,
+      "learning_rate": 2.8284825396372387e-05,
+      "loss": 0.3044,
+      "step": 526
+    },
+    {
+      "epoch": 2.137721518987342,
+      "grad_norm": 0.5847065204714459,
+      "learning_rate": 2.8233132224157132e-05,
+      "loss": 0.3004,
+      "step": 527
+    },
+    {
+      "epoch": 2.141772151898734,
+      "grad_norm": 0.5947317961363195,
+      "learning_rate": 2.8181372743426805e-05,
+      "loss": 0.2903,
+      "step": 528
+    },
+    {
+      "epoch": 2.1458227848101266,
+      "grad_norm": 0.6039047244405518,
+      "learning_rate": 2.8129547371045128e-05,
+      "loss": 0.3345,
+      "step": 529
+    },
+    {
+      "epoch": 2.149873417721519,
+      "grad_norm": 0.595742065073521,
+      "learning_rate": 2.8077656524406534e-05,
+      "loss": 0.3183,
+      "step": 530
+    },
+    {
+      "epoch": 2.1539240506329116,
+      "grad_norm": 0.59419276802551,
+      "learning_rate": 2.802570062143278e-05,
+      "loss": 0.3194,
+      "step": 531
+    },
+    {
+      "epoch": 2.1579746835443037,
+      "grad_norm": 0.588556342599953,
+      "learning_rate": 2.7973680080569555e-05,
+      "loss": 0.3207,
+      "step": 532
+    },
+    {
+      "epoch": 2.162025316455696,
+      "grad_norm": 0.5986626267908666,
+      "learning_rate": 2.792159532078314e-05,
+      "loss": 0.3145,
+      "step": 533
+    },
+    {
+      "epoch": 2.1660759493670887,
+      "grad_norm": 0.6109019910926544,
+      "learning_rate": 2.7869446761557033e-05,
+      "loss": 0.3172,
+      "step": 534
+    },
+    {
+      "epoch": 2.170126582278481,
+      "grad_norm": 0.6191474508010307,
+      "learning_rate": 2.781723482288857e-05,
+      "loss": 0.3161,
+      "step": 535
+    },
+    {
+      "epoch": 2.1741772151898733,
+      "grad_norm": 0.621625659144699,
+      "learning_rate": 2.7764959925285517e-05,
+      "loss": 0.3127,
+      "step": 536
+    },
+    {
+      "epoch": 2.178227848101266,
+      "grad_norm": 0.6169547221615751,
+      "learning_rate": 2.771262248976272e-05,
+      "loss": 0.3026,
+      "step": 537
+    },
+    {
+      "epoch": 2.1822784810126583,
+      "grad_norm": 0.5959262488160192,
+      "learning_rate": 2.7660222937838677e-05,
+      "loss": 0.3233,
+      "step": 538
+    },
+    {
+      "epoch": 2.186329113924051,
+      "grad_norm": 0.621587940391302,
+      "learning_rate": 2.7607761691532186e-05,
+      "loss": 0.3123,
+      "step": 539
+    },
+    {
+      "epoch": 2.190379746835443,
+      "grad_norm": 0.6410790988989044,
+      "learning_rate": 2.7555239173358916e-05,
+      "loss": 0.3137,
+      "step": 540
+    },
+    {
+      "epoch": 2.1944303797468354,
+      "grad_norm": 0.5933857603318469,
+      "learning_rate": 2.7502655806328e-05,
+      "loss": 0.3085,
+      "step": 541
+    },
+    {
+      "epoch": 2.198481012658228,
+      "grad_norm": 0.6222242419207609,
+      "learning_rate": 2.7450012013938648e-05,
+      "loss": 0.3196,
+      "step": 542
+    },
+    {
+      "epoch": 2.2025316455696204,
+      "grad_norm": 0.6013610715388936,
+      "learning_rate": 2.739730822017673e-05,
+      "loss": 0.3216,
+      "step": 543
+    },
+    {
+      "epoch": 2.2065822784810125,
+      "grad_norm": 0.5875321466348733,
+      "learning_rate": 2.7344544849511355e-05,
+      "loss": 0.305,
+      "step": 544
+    },
+    {
+      "epoch": 2.210632911392405,
+      "grad_norm": 0.5807187131624661,
+      "learning_rate": 2.7291722326891456e-05,
+      "loss": 0.3238,
+      "step": 545
+    },
+    {
+      "epoch": 2.2146835443037975,
+      "grad_norm": 0.5947923116772702,
+      "learning_rate": 2.723884107774236e-05,
+      "loss": 0.3191,
+      "step": 546
+    },
+    {
+      "epoch": 2.21873417721519,
+      "grad_norm": 0.5646865797520015,
+      "learning_rate": 2.718590152796239e-05,
+      "loss": 0.3183,
+      "step": 547
+    },
+    {
+      "epoch": 2.222784810126582,
+      "grad_norm": 0.5497471928410257,
+      "learning_rate": 2.71329041039194e-05,
+      "loss": 0.3253,
+      "step": 548
+    },
+    {
+      "epoch": 2.2268354430379746,
+      "grad_norm": 0.5593996663287893,
+      "learning_rate": 2.7079849232447357e-05,
+      "loss": 0.3216,
+      "step": 549
+    },
+    {
+      "epoch": 2.230886075949367,
+      "grad_norm": 0.6001319859954903,
+      "learning_rate": 2.7026737340842895e-05,
+      "loss": 0.332,
+      "step": 550
+    },
+    {
+      "epoch": 2.2349367088607597,
+      "grad_norm": 0.5707339058909686,
+      "learning_rate": 2.697356885686189e-05,
+      "loss": 0.3173,
+      "step": 551
+    },
+    {
+      "epoch": 2.2389873417721518,
+      "grad_norm": 0.5822892409273688,
+      "learning_rate": 2.6920344208716014e-05,
+      "loss": 0.3199,
+      "step": 552
+    },
+    {
+      "epoch": 2.2430379746835443,
+      "grad_norm": 0.5890623392833432,
+      "learning_rate": 2.6867063825069252e-05,
+      "loss": 0.328,
+      "step": 553
+    },
+    {
+      "epoch": 2.247088607594937,
+      "grad_norm": 0.6033696106566171,
+      "learning_rate": 2.6813728135034494e-05,
+      "loss": 0.3156,
+      "step": 554
+    },
+    {
+      "epoch": 2.2511392405063293,
+      "grad_norm": 0.6294181745413604,
+      "learning_rate": 2.6760337568170056e-05,
+      "loss": 0.3082,
+      "step": 555
+    },
+    {
+      "epoch": 2.2551898734177214,
+      "grad_norm": 0.6215018515253593,
+      "learning_rate": 2.6706892554476226e-05,
+      "loss": 0.302,
+      "step": 556
+    },
+    {
+      "epoch": 2.259240506329114,
+      "grad_norm": 0.6111252621800833,
+      "learning_rate": 2.6653393524391795e-05,
+      "loss": 0.3327,
+      "step": 557
+    },
+    {
+      "epoch": 2.2632911392405064,
+      "grad_norm": 0.5767087138082719,
+      "learning_rate": 2.6599840908790592e-05,
+      "loss": 0.3115,
+      "step": 558
+    },
+    {
+      "epoch": 2.267341772151899,
+      "grad_norm": 0.6319051339163412,
+      "learning_rate": 2.6546235138978028e-05,
+      "loss": 0.3086,
+      "step": 559
+    },
+    {
+      "epoch": 2.271392405063291,
+      "grad_norm": 0.5785364320444443,
+      "learning_rate": 2.6492576646687597e-05,
+      "loss": 0.3158,
+      "step": 560
+    },
+    {
+      "epoch": 2.2754430379746835,
+      "grad_norm": 0.5945508162017812,
+      "learning_rate": 2.6438865864077425e-05,
+      "loss": 0.3012,
+      "step": 561
+    },
+    {
+      "epoch": 2.279493670886076,
+      "grad_norm": 0.633674267734089,
+      "learning_rate": 2.6385103223726766e-05,
+      "loss": 0.3088,
+      "step": 562
+    },
+    {
+      "epoch": 2.2835443037974685,
+      "grad_norm": 0.578306564165435,
+      "learning_rate": 2.6331289158632537e-05,
+      "loss": 0.3036,
+      "step": 563
+    },
+    {
+      "epoch": 2.2875949367088606,
+      "grad_norm": 0.591940975205045,
+      "learning_rate": 2.6277424102205817e-05,
+      "loss": 0.3059,
+      "step": 564
+    },
+    {
+      "epoch": 2.291645569620253,
+      "grad_norm": 0.5768117177463489,
+      "learning_rate": 2.6223508488268374e-05,
+      "loss": 0.3273,
+      "step": 565
+    },
+    {
+      "epoch": 2.2956962025316456,
+      "grad_norm": 0.5764105261420913,
+      "learning_rate": 2.6169542751049148e-05,
+      "loss": 0.3156,
+      "step": 566
+    },
+    {
+      "epoch": 2.299746835443038,
+      "grad_norm": 0.5925099393127067,
+      "learning_rate": 2.6115527325180754e-05,
+      "loss": 0.3132,
+      "step": 567
+    },
+    {
+      "epoch": 2.3037974683544302,
+      "grad_norm": 0.62761436203238,
+      "learning_rate": 2.606146264569603e-05,
+      "loss": 0.3138,
+      "step": 568
+    },
+    {
+      "epoch": 2.3078481012658227,
+      "grad_norm": 0.6055851359631572,
+      "learning_rate": 2.6007349148024447e-05,
+      "loss": 0.3146,
+      "step": 569
+    },
+    {
+      "epoch": 2.3118987341772153,
+      "grad_norm": 0.588276541274497,
+      "learning_rate": 2.5953187267988694e-05,
+      "loss": 0.3267,
+      "step": 570
+    },
+    {
+      "epoch": 2.3159493670886078,
+      "grad_norm": 0.5657592685842114,
+      "learning_rate": 2.5898977441801097e-05,
+      "loss": 0.3321,
+      "step": 571
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 0.5734331374314922,
+      "learning_rate": 2.584472010606015e-05,
+      "loss": 0.3191,
+      "step": 572
+    },
+    {
+      "epoch": 2.3240506329113924,
+      "grad_norm": 0.5933800060183994,
+      "learning_rate": 2.5790415697746976e-05,
+      "loss": 0.3159,
+      "step": 573
+    },
+    {
+      "epoch": 2.328101265822785,
+      "grad_norm": 0.5608034354386431,
+      "learning_rate": 2.5736064654221808e-05,
+      "loss": 0.3114,
+      "step": 574
+    },
+    {
+      "epoch": 2.3321518987341774,
+      "grad_norm": 0.5798016163577461,
+      "learning_rate": 2.568166741322048e-05,
+      "loss": 0.3248,
+      "step": 575
+    },
+    {
+      "epoch": 2.3362025316455695,
+      "grad_norm": 0.5973116506578668,
+      "learning_rate": 2.56272244128509e-05,
+      "loss": 0.3218,
+      "step": 576
+    },
+    {
+      "epoch": 2.340253164556962,
+      "grad_norm": 0.5755943083573903,
+      "learning_rate": 2.55727360915895e-05,
+      "loss": 0.3181,
+      "step": 577
+    },
+    {
+      "epoch": 2.3443037974683545,
+      "grad_norm": 0.5723575348096773,
+      "learning_rate": 2.5518202888277734e-05,
+      "loss": 0.3228,
+      "step": 578
+    },
+    {
+      "epoch": 2.348354430379747,
+      "grad_norm": 0.6356742073347886,
+      "learning_rate": 2.5463625242118523e-05,
+      "loss": 0.3196,
+      "step": 579
+    },
+    {
+      "epoch": 2.352405063291139,
+      "grad_norm": 0.6353956924459588,
+      "learning_rate": 2.5409003592672723e-05,
+      "loss": 0.3542,
+      "step": 580
+    },
+    {
+      "epoch": 2.3564556962025316,
+      "grad_norm": 0.6164130319917572,
+      "learning_rate": 2.535433837985559e-05,
+      "loss": 0.3144,
+      "step": 581
+    },
+    {
+      "epoch": 2.360506329113924,
+      "grad_norm": 0.6265122924025903,
+      "learning_rate": 2.529963004393324e-05,
+      "loss": 0.3107,
+      "step": 582
+    },
+    {
+      "epoch": 2.3645569620253166,
+      "grad_norm": 0.5862418269780816,
+      "learning_rate": 2.524487902551908e-05,
+      "loss": 0.3267,
+      "step": 583
+    },
+    {
+      "epoch": 2.3686075949367087,
+      "grad_norm": 0.601853621725247,
+      "learning_rate": 2.519008576557029e-05,
+      "loss": 0.3069,
+      "step": 584
+    },
+    {
+      "epoch": 2.372658227848101,
+      "grad_norm": 0.5983991672167693,
+      "learning_rate": 2.5135250705384254e-05,
+      "loss": 0.3235,
+      "step": 585
+    },
+    {
+      "epoch": 2.3767088607594937,
+      "grad_norm": 0.626165640372555,
+      "learning_rate": 2.5080374286595007e-05,
+      "loss": 0.3327,
+      "step": 586
+    },
+    {
+      "epoch": 2.3807594936708862,
+      "grad_norm": 0.577616105272544,
+      "learning_rate": 2.5025456951169677e-05,
+      "loss": 0.3297,
+      "step": 587
+    },
+    {
+      "epoch": 2.3848101265822783,
+      "grad_norm": 0.6036565702450788,
+      "learning_rate": 2.4970499141404942e-05,
+      "loss": 0.3134,
+      "step": 588
+    },
+    {
+      "epoch": 2.388860759493671,
+      "grad_norm": 0.5931580954861955,
+      "learning_rate": 2.491550129992345e-05,
+      "loss": 0.3525,
+      "step": 589
+    },
+    {
+      "epoch": 2.3929113924050633,
+      "grad_norm": 0.5911245242170052,
+      "learning_rate": 2.486046386967024e-05,
+      "loss": 0.3263,
+      "step": 590
+    },
+    {
+      "epoch": 2.396962025316456,
+      "grad_norm": 0.5909777528713461,
+      "learning_rate": 2.4805387293909214e-05,
+      "loss": 0.3357,
+      "step": 591
+    },
+    {
+      "epoch": 2.401012658227848,
+      "grad_norm": 0.5861310570952096,
+      "learning_rate": 2.4750272016219552e-05,
+      "loss": 0.3338,
+      "step": 592
+    },
+    {
+      "epoch": 2.4050632911392404,
+      "grad_norm": 0.5903176505223747,
+      "learning_rate": 2.4695118480492114e-05,
+      "loss": 0.3179,
+      "step": 593
+    },
+    {
+      "epoch": 2.409113924050633,
+      "grad_norm": 0.5805406912117017,
+      "learning_rate": 2.4639927130925898e-05,
+      "loss": 0.3334,
+      "step": 594
+    },
+    {
+      "epoch": 2.4131645569620255,
+      "grad_norm": 0.5637813039281818,
+      "learning_rate": 2.458469841202444e-05,
+      "loss": 0.356,
+      "step": 595
+    },
+    {
+      "epoch": 2.4172151898734175,
+      "grad_norm": 0.6045182928419369,
+      "learning_rate": 2.452943276859226e-05,
+      "loss": 0.3177,
+      "step": 596
+    },
+    {
+      "epoch": 2.42126582278481,
+      "grad_norm": 0.6043162929754949,
+      "learning_rate": 2.447413064573125e-05,
+      "loss": 0.3253,
+      "step": 597
+    },
+    {
+      "epoch": 2.4253164556962026,
+      "grad_norm": 0.597067353339536,
+      "learning_rate": 2.4418792488837095e-05,
+      "loss": 0.3272,
+      "step": 598
+    },
+    {
+      "epoch": 2.429367088607595,
+      "grad_norm": 0.5760374218156048,
+      "learning_rate": 2.4363418743595713e-05,
+      "loss": 0.3162,
+      "step": 599
+    },
+    {
+      "epoch": 2.433417721518987,
+      "grad_norm": 0.6601122893925263,
+      "learning_rate": 2.430800985597963e-05,
+      "loss": 0.3147,
+      "step": 600
+    },
+    {
+      "epoch": 2.4374683544303797,
+      "grad_norm": 0.564544209503751,
+      "learning_rate": 2.4252566272244415e-05,
+      "loss": 0.3206,
+      "step": 601
+    },
+    {
+      "epoch": 2.441518987341772,
+      "grad_norm": 0.6371393104934018,
+      "learning_rate": 2.4197088438925063e-05,
+      "loss": 0.328,
+      "step": 602
+    },
+    {
+      "epoch": 2.4455696202531647,
+      "grad_norm": 0.6029597989483182,
+      "learning_rate": 2.4141576802832417e-05,
+      "loss": 0.3134,
+      "step": 603
+    },
+    {
+      "epoch": 2.449620253164557,
+      "grad_norm": 0.594069971561629,
+      "learning_rate": 2.408603181104957e-05,
+      "loss": 0.328,
+      "step": 604
+    },
+    {
+      "epoch": 2.4536708860759493,
+      "grad_norm": 0.5971516058402228,
+      "learning_rate": 2.4030453910928245e-05,
+      "loss": 0.3179,
+      "step": 605
+    },
+    {
+      "epoch": 2.457721518987342,
+      "grad_norm": 0.5764610121314291,
+      "learning_rate": 2.397484355008521e-05,
+      "loss": 0.3169,
+      "step": 606
+    },
+    {
+      "epoch": 2.4617721518987343,
+      "grad_norm": 0.6146982592031123,
+      "learning_rate": 2.3919201176398662e-05,
+      "loss": 0.3138,
+      "step": 607
+    },
+    {
+      "epoch": 2.4658227848101264,
+      "grad_norm": 0.6099924875445413,
+      "learning_rate": 2.3863527238004633e-05,
+      "loss": 0.2995,
+      "step": 608
+    },
+    {
+      "epoch": 2.469873417721519,
+      "grad_norm": 0.5892296334111441,
+      "learning_rate": 2.380782218329337e-05,
+      "loss": 0.3217,
+      "step": 609
+    },
+    {
+      "epoch": 2.4739240506329114,
+      "grad_norm": 0.6219860556840322,
+      "learning_rate": 2.3752086460905725e-05,
+      "loss": 0.317,
+      "step": 610
+    },
+    {
+      "epoch": 2.477974683544304,
+      "grad_norm": 0.5731041306536542,
+      "learning_rate": 2.3696320519729544e-05,
+      "loss": 0.3235,
+      "step": 611
+    },
+    {
+      "epoch": 2.482025316455696,
+      "grad_norm": 0.6155629860953881,
+      "learning_rate": 2.3640524808896045e-05,
+      "loss": 0.3298,
+      "step": 612
+    },
+    {
+      "epoch": 2.4860759493670885,
+      "grad_norm": 0.6148321219448996,
+      "learning_rate": 2.3584699777776222e-05,
+      "loss": 0.3061,
+      "step": 613
+    },
+    {
+      "epoch": 2.490126582278481,
+      "grad_norm": 0.5655634447998682,
+      "learning_rate": 2.3528845875977195e-05,
+      "loss": 0.3046,
+      "step": 614
+    },
+    {
+      "epoch": 2.4941772151898736,
+      "grad_norm": 0.5819307837356728,
+      "learning_rate": 2.3472963553338614e-05,
+      "loss": 0.3327,
+      "step": 615
+    },
+    {
+      "epoch": 2.4982278481012656,
+      "grad_norm": 0.5918834988459256,
+      "learning_rate": 2.341705325992901e-05,
+      "loss": 0.3163,
+      "step": 616
+    },
+    {
+      "epoch": 2.502278481012658,
+      "grad_norm": 0.6178544472842665,
+      "learning_rate": 2.336111544604222e-05,
+      "loss": 0.3147,
+      "step": 617
+    },
+    {
+      "epoch": 2.5063291139240507,
+      "grad_norm": 0.5767750277316077,
+      "learning_rate": 2.33051505621937e-05,
+      "loss": 0.3316,
+      "step": 618
+    },
+    {
+      "epoch": 2.510379746835443,
+      "grad_norm": 0.5729152322864887,
+      "learning_rate": 2.324915905911693e-05,
+      "loss": 0.3255,
+      "step": 619
+    },
+    {
+      "epoch": 2.5144303797468357,
+      "grad_norm": 0.5782390643543696,
+      "learning_rate": 2.319314138775977e-05,
+      "loss": 0.3313,
+      "step": 620
+    },
+    {
+      "epoch": 2.5184810126582278,
+      "grad_norm": 0.6180136198009902,
+      "learning_rate": 2.3137097999280856e-05,
+      "loss": 0.3303,
+      "step": 621
+    },
+    {
+      "epoch": 2.5225316455696203,
+      "grad_norm": 0.6062605747413887,
+      "learning_rate": 2.308102934504593e-05,
+      "loss": 0.3356,
+      "step": 622
+    },
+    {
+      "epoch": 2.526582278481013,
+      "grad_norm": 0.5654733218208525,
+      "learning_rate": 2.3024935876624222e-05,
+      "loss": 0.3296,
+      "step": 623
+    },
+    {
+      "epoch": 2.530632911392405,
+      "grad_norm": 0.6026171776246693,
+      "learning_rate": 2.2968818045784813e-05,
+      "loss": 0.2889,
+      "step": 624
+    },
+    {
+      "epoch": 2.5346835443037974,
+      "grad_norm": 0.5713426048373026,
+      "learning_rate": 2.2912676304493006e-05,
+      "loss": 0.3407,
+      "step": 625
+    },
+    {
+      "epoch": 2.53873417721519,
+      "grad_norm": 0.5817670148578413,
+      "learning_rate": 2.2856511104906668e-05,
+      "loss": 0.3328,
+      "step": 626
+    },
+    {
+      "epoch": 2.5427848101265824,
+      "grad_norm": 0.5556513023590391,
+      "learning_rate": 2.2800322899372586e-05,
+      "loss": 0.3193,
+      "step": 627
+    },
+    {
+      "epoch": 2.546835443037975,
+      "grad_norm": 0.6094947147029373,
+      "learning_rate": 2.2744112140422844e-05,
+      "loss": 0.3031,
+      "step": 628
+    },
+    {
+      "epoch": 2.550886075949367,
+      "grad_norm": 0.6187556378816954,
+      "learning_rate": 2.2687879280771177e-05,
+      "loss": 0.3324,
+      "step": 629
+    },
+    {
+      "epoch": 2.5549367088607595,
+      "grad_norm": 0.578343581984409,
+      "learning_rate": 2.26316247733093e-05,
+      "loss": 0.3159,
+      "step": 630
+    },
+    {
+      "epoch": 2.558987341772152,
+      "grad_norm": 0.6155955059223527,
+      "learning_rate": 2.257534907110328e-05,
+      "loss": 0.3011,
+      "step": 631
+    },
+    {
+      "epoch": 2.563037974683544,
+      "grad_norm": 0.5795629358942243,
+      "learning_rate": 2.2519052627389882e-05,
+      "loss": 0.3078,
+      "step": 632
+    },
+    {
+      "epoch": 2.5670886075949366,
+      "grad_norm": 0.5674527144394165,
+      "learning_rate": 2.246273589557294e-05,
+      "loss": 0.3161,
+      "step": 633
+    },
+    {
+      "epoch": 2.571139240506329,
+      "grad_norm": 0.5588357821430461,
+      "learning_rate": 2.240639932921966e-05,
+      "loss": 0.3346,
+      "step": 634
+    },
+    {
+      "epoch": 2.5751898734177217,
+      "grad_norm": 0.5444232203136201,
+      "learning_rate": 2.2350043382056995e-05,
+      "loss": 0.3095,
+      "step": 635
+    },
+    {
+      "epoch": 2.579240506329114,
+      "grad_norm": 0.573094339905033,
+      "learning_rate": 2.2293668507968015e-05,
+      "loss": 0.327,
+      "step": 636
+    },
+    {
+      "epoch": 2.5832911392405062,
+      "grad_norm": 0.5641807843472327,
+      "learning_rate": 2.2237275160988186e-05,
+      "loss": 0.3188,
+      "step": 637
+    },
+    {
+      "epoch": 2.5873417721518988,
+      "grad_norm": 0.5731316098538122,
+      "learning_rate": 2.2180863795301787e-05,
+      "loss": 0.3203,
+      "step": 638
+    },
+    {
+      "epoch": 2.5913924050632913,
+      "grad_norm": 0.6008689579124836,
+      "learning_rate": 2.212443486523819e-05,
+      "loss": 0.3038,
+      "step": 639
+    },
+    {
+      "epoch": 2.5954430379746833,
+      "grad_norm": 0.5930253947920351,
+      "learning_rate": 2.2067988825268243e-05,
+      "loss": 0.3331,
+      "step": 640
+    },
+    {
+      "epoch": 2.599493670886076,
+      "grad_norm": 0.6234375229361323,
+      "learning_rate": 2.2011526130000596e-05,
+      "loss": 0.3278,
+      "step": 641
+    },
+    {
+      "epoch": 2.6035443037974684,
+      "grad_norm": 0.5726925904398659,
+      "learning_rate": 2.1955047234178038e-05,
+      "loss": 0.3265,
+      "step": 642
+    },
+    {
+      "epoch": 2.607594936708861,
+      "grad_norm": 0.5602802447740196,
+      "learning_rate": 2.1898552592673825e-05,
+      "loss": 0.2997,
+      "step": 643
+    },
+    {
+      "epoch": 2.6116455696202534,
+      "grad_norm": 0.6250768776339883,
+      "learning_rate": 2.184204266048803e-05,
+      "loss": 0.3101,
+      "step": 644
+    },
+    {
+      "epoch": 2.6156962025316455,
+      "grad_norm": 0.5732980555618912,
+      "learning_rate": 2.1785517892743887e-05,
+      "loss": 0.3214,
+      "step": 645
+    },
+    {
+      "epoch": 2.619746835443038,
+      "grad_norm": 0.5680801273512105,
+      "learning_rate": 2.17289787446841e-05,
+      "loss": 0.3251,
+      "step": 646
+    },
+    {
+      "epoch": 2.6237974683544305,
+      "grad_norm": 0.6410330049386856,
+      "learning_rate": 2.1672425671667198e-05,
+      "loss": 0.3156,
+      "step": 647
+    },
+    {
+      "epoch": 2.6278481012658226,
+      "grad_norm": 0.5720985436707079,
+      "learning_rate": 2.161585912916385e-05,
+      "loss": 0.3239,
+      "step": 648
+    },
+    {
+      "epoch": 2.631898734177215,
+      "grad_norm": 0.6046249812341012,
+      "learning_rate": 2.1559279572753214e-05,
+      "loss": 0.3105,
+      "step": 649
+    },
+    {
+      "epoch": 2.6359493670886076,
+      "grad_norm": 0.6876269057353852,
+      "learning_rate": 2.1502687458119268e-05,
+      "loss": 0.3299,
+      "step": 650
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 0.58244204054477,
+      "learning_rate": 2.1446083241047116e-05,
+      "loss": 0.3317,
+      "step": 651
+    },
+    {
+      "epoch": 2.6440506329113926,
+      "grad_norm": 0.5983256216436481,
+      "learning_rate": 2.1389467377419333e-05,
+      "loss": 0.3201,
+      "step": 652
+    },
+    {
+      "epoch": 2.6481012658227847,
+      "grad_norm": 0.7469663683045384,
+      "learning_rate": 2.133284032321232e-05,
+      "loss": 0.3177,
+      "step": 653
+    },
+    {
+      "epoch": 2.6521518987341772,
+      "grad_norm": 0.5798429434187768,
+      "learning_rate": 2.1276202534492566e-05,
+      "loss": 0.3245,
+      "step": 654
+    },
+    {
+      "epoch": 2.6562025316455697,
+      "grad_norm": 0.5942437606602983,
+      "learning_rate": 2.121955446741306e-05,
+      "loss": 0.3253,
+      "step": 655
+    },
+    {
+      "epoch": 2.660253164556962,
+      "grad_norm": 0.7023373901425931,
+      "learning_rate": 2.1162896578209517e-05,
+      "loss": 0.3175,
+      "step": 656
+    },
+    {
+      "epoch": 2.6643037974683543,
+      "grad_norm": 0.5896360714913341,
+      "learning_rate": 2.1106229323196813e-05,
+      "loss": 0.3226,
+      "step": 657
+    },
+    {
+      "epoch": 2.668354430379747,
+      "grad_norm": 0.5868391745742445,
+      "learning_rate": 2.1049553158765214e-05,
+      "loss": 0.3065,
+      "step": 658
+    },
+    {
+      "epoch": 2.6724050632911394,
+      "grad_norm": 0.6304207166625573,
+      "learning_rate": 2.0992868541376764e-05,
+      "loss": 0.3211,
+      "step": 659
+    },
+    {
+      "epoch": 2.676455696202532,
+      "grad_norm": 0.5681706371931647,
+      "learning_rate": 2.093617592756158e-05,
+      "loss": 0.3195,
+      "step": 660
+    },
+    {
+      "epoch": 2.680506329113924,
+      "grad_norm": 0.5797899035511879,
+      "learning_rate": 2.0879475773914167e-05,
+      "loss": 0.3259,
+      "step": 661
+    },
+    {
+      "epoch": 2.6845569620253165,
+      "grad_norm": 0.5822364707118522,
+      "learning_rate": 2.082276853708978e-05,
+      "loss": 0.3068,
+      "step": 662
+    },
+    {
+      "epoch": 2.688607594936709,
+      "grad_norm": 0.5393363957464694,
+      "learning_rate": 2.076605467380071e-05,
+      "loss": 0.3162,
+      "step": 663
+    },
+    {
+      "epoch": 2.692658227848101,
+      "grad_norm": 0.545175776819639,
+      "learning_rate": 2.0709334640812613e-05,
+      "loss": 0.3144,
+      "step": 664
+    },
+    {
+      "epoch": 2.6967088607594936,
+      "grad_norm": 0.5712002741743452,
+      "learning_rate": 2.0652608894940824e-05,
+      "loss": 0.3236,
+      "step": 665
+    },
+    {
+      "epoch": 2.700759493670886,
+      "grad_norm": 0.5965312052859405,
+      "learning_rate": 2.0595877893046722e-05,
+      "loss": 0.3176,
+      "step": 666
+    },
+    {
+      "epoch": 2.7048101265822786,
+      "grad_norm": 0.5424265994855902,
+      "learning_rate": 2.0539142092033985e-05,
+      "loss": 0.3435,
+      "step": 667
+    },
+    {
+      "epoch": 2.708860759493671,
+      "grad_norm": 0.5877662523916398,
+      "learning_rate": 2.048240194884496e-05,
+      "loss": 0.3293,
+      "step": 668
+    },
+    {
+      "epoch": 2.712911392405063,
+      "grad_norm": 0.5698340760986162,
+      "learning_rate": 2.042565792045695e-05,
+      "loss": 0.3205,
+      "step": 669
+    },
+    {
+      "epoch": 2.7169620253164557,
+      "grad_norm": 0.545263680138819,
+      "learning_rate": 2.036891046387857e-05,
+      "loss": 0.3303,
+      "step": 670
+    },
+    {
+      "epoch": 2.721012658227848,
+      "grad_norm": 0.5638305008547335,
+      "learning_rate": 2.0312160036146036e-05,
+      "loss": 0.3057,
+      "step": 671
+    },
+    {
+      "epoch": 2.7250632911392403,
+      "grad_norm": 0.5668569570372243,
+      "learning_rate": 2.025540709431948e-05,
+      "loss": 0.3147,
+      "step": 672
+    },
+    {
+      "epoch": 2.729113924050633,
+      "grad_norm": 0.5508904346281702,
+      "learning_rate": 2.0198652095479298e-05,
+      "loss": 0.3223,
+      "step": 673
+    },
+    {
+      "epoch": 2.7331645569620253,
+      "grad_norm": 0.5336877954656891,
+      "learning_rate": 2.014189549672245e-05,
+      "loss": 0.3293,
+      "step": 674
+    },
+    {
+      "epoch": 2.737215189873418,
+      "grad_norm": 0.5560032262837084,
+      "learning_rate": 2.0085137755158776e-05,
+      "loss": 0.3148,
+      "step": 675
+    },
+    {
+      "epoch": 2.7412658227848103,
+      "grad_norm": 0.5608281169880271,
+      "learning_rate": 2.0028379327907327e-05,
+      "loss": 0.3176,
+      "step": 676
+    },
+    {
+      "epoch": 2.7453164556962024,
+      "grad_norm": 0.5676638838709893,
+      "learning_rate": 1.9971620672092676e-05,
+      "loss": 0.32,
+      "step": 677
+    },
+    {
+      "epoch": 2.749367088607595,
+      "grad_norm": 0.5774226354604267,
+      "learning_rate": 1.991486224484123e-05,
+      "loss": 0.3248,
+      "step": 678
+    },
+    {
+      "epoch": 2.7534177215189874,
+      "grad_norm": 0.6128810310596289,
+      "learning_rate": 1.985810450327756e-05,
+      "loss": 0.3328,
+      "step": 679
+    },
+    {
+      "epoch": 2.7574683544303795,
+      "grad_norm": 0.5344446723744176,
+      "learning_rate": 1.9801347904520706e-05,
+      "loss": 0.3171,
+      "step": 680
+    },
+    {
+      "epoch": 2.761518987341772,
+      "grad_norm": 0.5659425213920813,
+      "learning_rate": 1.974459290568053e-05,
+      "loss": 0.3093,
+      "step": 681
+    },
+    {
+      "epoch": 2.7655696202531646,
+      "grad_norm": 0.604726986785336,
+      "learning_rate": 1.968783996385397e-05,
+      "loss": 0.3424,
+      "step": 682
+    },
+    {
+      "epoch": 2.769620253164557,
+      "grad_norm": 0.5792237438997209,
+      "learning_rate": 1.963108953612143e-05,
+      "loss": 0.3378,
+      "step": 683
+    },
+    {
+      "epoch": 2.7736708860759496,
+      "grad_norm": 0.5716112289505682,
+      "learning_rate": 1.9574342079543056e-05,
+      "loss": 0.3067,
+      "step": 684
+    },
+    {
+      "epoch": 2.7777215189873417,
+      "grad_norm": 0.5750432265938701,
+      "learning_rate": 1.9517598051155046e-05,
+      "loss": 0.3288,
+      "step": 685
+    },
+    {
+      "epoch": 2.781772151898734,
+      "grad_norm": 0.5571719782940517,
+      "learning_rate": 1.9460857907966025e-05,
+      "loss": 0.3294,
+      "step": 686
+    },
+    {
+      "epoch": 2.7858227848101267,
+      "grad_norm": 0.5502747244121644,
+      "learning_rate": 1.9404122106953285e-05,
+      "loss": 0.3101,
+      "step": 687
+    },
+    {
+      "epoch": 2.7898734177215188,
+      "grad_norm": 0.5771528734138521,
+      "learning_rate": 1.9347391105059176e-05,
+      "loss": 0.3146,
+      "step": 688
+    },
+    {
+      "epoch": 2.7939240506329113,
+      "grad_norm": 0.5888819251558312,
+      "learning_rate": 1.92906653591874e-05,
+      "loss": 0.3088,
+      "step": 689
+    },
+    {
+      "epoch": 2.797974683544304,
+      "grad_norm": 0.5640400138612696,
+      "learning_rate": 1.9233945326199295e-05,
+      "loss": 0.3253,
+      "step": 690
+    },
+    {
+      "epoch": 2.8020253164556963,
+      "grad_norm": 0.600356511361642,
+      "learning_rate": 1.917723146291022e-05,
+      "loss": 0.3142,
+      "step": 691
+    },
+    {
+      "epoch": 2.806075949367089,
+      "grad_norm": 0.5497878080364649,
+      "learning_rate": 1.912052422608584e-05,
+      "loss": 0.3272,
+      "step": 692
+    },
+    {
+      "epoch": 2.810126582278481,
+      "grad_norm": 0.5595041072083419,
+      "learning_rate": 1.9063824072438428e-05,
+      "loss": 0.3115,
+      "step": 693
+    },
+    {
+      "epoch": 2.8141772151898734,
+      "grad_norm": 0.5900088178713231,
+      "learning_rate": 1.9007131458623246e-05,
+      "loss": 0.3253,
+      "step": 694
+    },
+    {
+      "epoch": 2.818227848101266,
+      "grad_norm": 0.5796333157577446,
+      "learning_rate": 1.895044684123479e-05,
+      "loss": 0.3133,
+      "step": 695
+    },
+    {
+      "epoch": 2.822278481012658,
+      "grad_norm": 0.5616689837875714,
+      "learning_rate": 1.8893770676803194e-05,
+      "loss": 0.3195,
+      "step": 696
+    },
+    {
+      "epoch": 2.8263291139240505,
+      "grad_norm": 0.5985354601681668,
+      "learning_rate": 1.8837103421790486e-05,
+      "loss": 0.3136,
+      "step": 697
+    },
+    {
+      "epoch": 2.830379746835443,
+      "grad_norm": 0.5726945269013398,
+      "learning_rate": 1.8780445532586952e-05,
+      "loss": 0.3083,
+      "step": 698
+    },
+    {
+      "epoch": 2.8344303797468355,
+      "grad_norm": 0.5666330938974999,
+      "learning_rate": 1.872379746550743e-05,
+      "loss": 0.3076,
+      "step": 699
+    },
+    {
+      "epoch": 2.838481012658228,
+      "grad_norm": 0.5802466492858106,
+      "learning_rate": 1.866715967678769e-05,
+      "loss": 0.3231,
+      "step": 700
+    },
+    {
+      "epoch": 2.84253164556962,
+      "grad_norm": 0.5717436792294017,
+      "learning_rate": 1.861053262258067e-05,
+      "loss": 0.3213,
+      "step": 701
+    },
+    {
+      "epoch": 2.8465822784810126,
+      "grad_norm": 0.5662256008792681,
+      "learning_rate": 1.8553916758952897e-05,
+      "loss": 0.327,
+      "step": 702
+    },
+    {
+      "epoch": 2.850632911392405,
+      "grad_norm": 0.5620343797787994,
+      "learning_rate": 1.8497312541880735e-05,
+      "loss": 0.3273,
+      "step": 703
+    },
+    {
+      "epoch": 2.8546835443037972,
+      "grad_norm": 0.5705388996905202,
+      "learning_rate": 1.8440720427246786e-05,
+      "loss": 0.3214,
+      "step": 704
+    },
+    {
+      "epoch": 2.8587341772151897,
+      "grad_norm": 0.5406223172485588,
+      "learning_rate": 1.8384140870836157e-05,
+      "loss": 0.3222,
+      "step": 705
+    },
+    {
+      "epoch": 2.8627848101265823,
+      "grad_norm": 0.5568917547246185,
+      "learning_rate": 1.8327574328332806e-05,
+      "loss": 0.3333,
+      "step": 706
+    },
+    {
+      "epoch": 2.8668354430379748,
+      "grad_norm": 0.5733994241993376,
+      "learning_rate": 1.8271021255315906e-05,
+      "loss": 0.3103,
+      "step": 707
+    },
+    {
+      "epoch": 2.8708860759493673,
+      "grad_norm": 0.5225031343913608,
+      "learning_rate": 1.8214482107256117e-05,
+      "loss": 0.303,
+      "step": 708
+    },
+    {
+      "epoch": 2.8749367088607594,
+      "grad_norm": 0.5553532937600009,
+      "learning_rate": 1.8157957339511968e-05,
+      "loss": 0.3188,
+      "step": 709
+    },
+    {
+      "epoch": 2.878987341772152,
+      "grad_norm": 0.5576356304453806,
+      "learning_rate": 1.8101447407326182e-05,
+      "loss": 0.3262,
+      "step": 710
+    },
+    {
+      "epoch": 2.8830379746835444,
+      "grad_norm": 0.5688538928214123,
+      "learning_rate": 1.8044952765821966e-05,
+      "loss": 0.3156,
+      "step": 711
+    },
+    {
+      "epoch": 2.8870886075949365,
+      "grad_norm": 0.5519266590181565,
+      "learning_rate": 1.7988473869999407e-05,
+      "loss": 0.3091,
+      "step": 712
+    },
+    {
+      "epoch": 2.891139240506329,
+      "grad_norm": 0.532735186685207,
+      "learning_rate": 1.7932011174731764e-05,
+      "loss": 0.3181,
+      "step": 713
+    },
+    {
+      "epoch": 2.8951898734177215,
+      "grad_norm": 0.5513340913347524,
+      "learning_rate": 1.7875565134761817e-05,
+      "loss": 0.2956,
+      "step": 714
+    },
+    {
+      "epoch": 2.899240506329114,
+      "grad_norm": 0.5731651586409731,
+      "learning_rate": 1.7819136204698226e-05,
+      "loss": 0.3307,
+      "step": 715
+    },
+    {
+      "epoch": 2.9032911392405065,
+      "grad_norm": 0.5724520932703475,
+      "learning_rate": 1.776272483901182e-05,
+      "loss": 0.3148,
+      "step": 716
+    },
+    {
+      "epoch": 2.9073417721518986,
+      "grad_norm": 0.5603203952453423,
+      "learning_rate": 1.7706331492031995e-05,
+      "loss": 0.317,
+      "step": 717
+    },
+    {
+      "epoch": 2.911392405063291,
+      "grad_norm": 0.5646628796166276,
+      "learning_rate": 1.764995661794301e-05,
+      "loss": 0.3338,
+      "step": 718
+    },
+    {
+      "epoch": 2.9154430379746836,
+      "grad_norm": 0.5595948681587484,
+      "learning_rate": 1.759360067078035e-05,
+      "loss": 0.3132,
+      "step": 719
+    },
+    {
+      "epoch": 2.9194936708860757,
+      "grad_norm": 0.5624696807037703,
+      "learning_rate": 1.7537264104427064e-05,
+      "loss": 0.3239,
+      "step": 720
+    },
+    {
+      "epoch": 2.923544303797468,
+      "grad_norm": 0.5470968255641537,
+      "learning_rate": 1.748094737261012e-05,
+      "loss": 0.3271,
+      "step": 721
+    },
+    {
+      "epoch": 2.9275949367088607,
+      "grad_norm": 0.5635202121231361,
+      "learning_rate": 1.7424650928896726e-05,
+      "loss": 0.3213,
+      "step": 722
+    },
+    {
+      "epoch": 2.9316455696202532,
+      "grad_norm": 0.5397991686517396,
+      "learning_rate": 1.7368375226690712e-05,
+      "loss": 0.3181,
+      "step": 723
+    },
+    {
+      "epoch": 2.9356962025316458,
+      "grad_norm": 0.5476745898354921,
+      "learning_rate": 1.731212071922883e-05,
+      "loss": 0.3171,
+      "step": 724
+    },
+    {
+      "epoch": 2.939746835443038,
+      "grad_norm": 0.580958812285453,
+      "learning_rate": 1.7255887859577156e-05,
+      "loss": 0.3221,
+      "step": 725
+    },
+    {
+      "epoch": 2.9437974683544303,
+      "grad_norm": 0.5713054453350138,
+      "learning_rate": 1.7199677100627427e-05,
+      "loss": 0.3087,
+      "step": 726
+    },
+    {
+      "epoch": 2.947848101265823,
+      "grad_norm": 0.589905334934796,
+      "learning_rate": 1.7143488895093343e-05,
+      "loss": 0.293,
+      "step": 727
+    },
+    {
+      "epoch": 2.951898734177215,
+      "grad_norm": 0.5954948204656471,
+      "learning_rate": 1.7087323695506994e-05,
+      "loss": 0.3283,
+      "step": 728
+    },
+    {
+      "epoch": 2.9559493670886074,
+      "grad_norm": 0.5515860467983891,
+      "learning_rate": 1.7031181954215194e-05,
+      "loss": 0.3001,
+      "step": 729
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 0.5498937262528399,
+      "learning_rate": 1.6975064123375788e-05,
+      "loss": 0.3204,
+      "step": 730
+    },
+    {
+      "epoch": 2.9640506329113925,
+      "grad_norm": 0.5710628271863746,
+      "learning_rate": 1.6918970654954084e-05,
+      "loss": 0.3039,
+      "step": 731
+    },
+    {
+      "epoch": 2.968101265822785,
+      "grad_norm": 0.5395682632726876,
+      "learning_rate": 1.686290200071915e-05,
+      "loss": 0.3153,
+      "step": 732
+    },
+    {
+      "epoch": 2.972151898734177,
+      "grad_norm": 0.5376996573338679,
+      "learning_rate": 1.6806858612240234e-05,
+      "loss": 0.3132,
+      "step": 733
+    },
+    {
+      "epoch": 2.9762025316455696,
+      "grad_norm": 0.5582550560130555,
+      "learning_rate": 1.6750840940883078e-05,
+      "loss": 0.3227,
+      "step": 734
+    },
+    {
+      "epoch": 2.980253164556962,
+      "grad_norm": 0.5088943888548881,
+      "learning_rate": 1.6694849437806305e-05,
+      "loss": 0.3289,
+      "step": 735
+    },
+    {
+      "epoch": 2.984303797468354,
+      "grad_norm": 0.5717784138016254,
+      "learning_rate": 1.663888455395778e-05,
+      "loss": 0.3249,
+      "step": 736
+    },
+    {
+      "epoch": 2.9883544303797467,
+      "grad_norm": 0.5398450075203803,
+      "learning_rate": 1.6582946740070995e-05,
+      "loss": 0.307,
+      "step": 737
+    },
+    {
+      "epoch": 2.992405063291139,
+      "grad_norm": 0.5472258751686856,
+      "learning_rate": 1.6527036446661396e-05,
+      "loss": 0.3049,
+      "step": 738
+    },
+    {
+      "epoch": 2.9964556962025317,
+      "grad_norm": 0.5400793084609854,
+      "learning_rate": 1.6471154124022818e-05,
+      "loss": 0.3167,
+      "step": 739
+    },
+    {
+      "epoch": 3.002025316455696,
+      "grad_norm": 0.5731646738012371,
+      "learning_rate": 1.6415300222223788e-05,
+      "loss": 0.2436,
+      "step": 740
+    },
+    {
+      "epoch": 3.0060759493670886,
+      "grad_norm": 0.6006123767767062,
+      "learning_rate": 1.6359475191103958e-05,
+      "loss": 0.185,
+      "step": 741
+    },
+    {
+      "epoch": 3.010126582278481,
+      "grad_norm": 0.5018098346523897,
+      "learning_rate": 1.6303679480270466e-05,
+      "loss": 0.1811,
+      "step": 742
+    },
+    {
+      "epoch": 3.0141772151898736,
+      "grad_norm": 0.4738191402210632,
+      "learning_rate": 1.624791353909428e-05,
+      "loss": 0.1765,
+      "step": 743
+    },
+    {
+      "epoch": 3.0182278481012657,
+      "grad_norm": 0.643383317440617,
+      "learning_rate": 1.619217781670663e-05,
+      "loss": 0.1845,
+      "step": 744
+    },
+    {
+      "epoch": 3.022278481012658,
+      "grad_norm": 0.7784249364541905,
+      "learning_rate": 1.6136472761995373e-05,
+      "loss": 0.1991,
+      "step": 745
+    },
+    {
+      "epoch": 3.0263291139240507,
+      "grad_norm": 0.6017537813198908,
+      "learning_rate": 1.608079882360134e-05,
+      "loss": 0.1732,
+      "step": 746
+    },
+    {
+      "epoch": 3.030379746835443,
+      "grad_norm": 0.49255142415631376,
+      "learning_rate": 1.60251564499148e-05,
+      "loss": 0.1934,
+      "step": 747
+    },
+    {
+      "epoch": 3.0344303797468353,
+      "grad_norm": 0.515399233739333,
+      "learning_rate": 1.596954608907176e-05,
+      "loss": 0.1835,
+      "step": 748
+    },
+    {
+      "epoch": 3.038481012658228,
+      "grad_norm": 0.5987044953896159,
+      "learning_rate": 1.591396818895043e-05,
+      "loss": 0.1907,
+      "step": 749
+    },
+    {
+      "epoch": 3.0425316455696203,
+      "grad_norm": 0.5499365240466416,
+      "learning_rate": 1.585842319716759e-05,
+      "loss": 0.185,
+      "step": 750
+    },
+    {
+      "epoch": 3.046582278481013,
+      "grad_norm": 0.47761008828608165,
+      "learning_rate": 1.5802911561074944e-05,
+      "loss": 0.1502,
+      "step": 751
+    },
+    {
+      "epoch": 3.050632911392405,
+      "grad_norm": 0.481756795436437,
+      "learning_rate": 1.5747433727755595e-05,
+      "loss": 0.1887,
+      "step": 752
+    },
+    {
+      "epoch": 3.0546835443037974,
+      "grad_norm": 0.5392674051686914,
+      "learning_rate": 1.5691990144020376e-05,
+      "loss": 0.1792,
+      "step": 753
+    },
+    {
+      "epoch": 3.05873417721519,
+      "grad_norm": 0.5554068347593998,
+      "learning_rate": 1.5636581256404297e-05,
+      "loss": 0.1752,
+      "step": 754
+    },
+    {
+      "epoch": 3.0627848101265824,
+      "grad_norm": 0.5293084332872574,
+      "learning_rate": 1.558120751116291e-05,
+      "loss": 0.1705,
+      "step": 755
+    },
+    {
+      "epoch": 3.0668354430379745,
+      "grad_norm": 0.5205731348480668,
+      "learning_rate": 1.552586935426876e-05,
+      "loss": 0.1693,
+      "step": 756
+    },
+    {
+      "epoch": 3.070886075949367,
+      "grad_norm": 0.4867919230614548,
+      "learning_rate": 1.547056723140774e-05,
+      "loss": 0.1658,
+      "step": 757
+    },
+    {
+      "epoch": 3.0749367088607595,
+      "grad_norm": 0.4930478338734456,
+      "learning_rate": 1.5415301587975565e-05,
+      "loss": 0.1908,
+      "step": 758
+    },
+    {
+      "epoch": 3.078987341772152,
+      "grad_norm": 0.4781061664877187,
+      "learning_rate": 1.536007286907411e-05,
+      "loss": 0.1554,
+      "step": 759
+    },
+    {
+      "epoch": 3.083037974683544,
+      "grad_norm": 0.4790115436796962,
+      "learning_rate": 1.5304881519507896e-05,
+      "loss": 0.1738,
+      "step": 760
+    },
+    {
+      "epoch": 3.0870886075949366,
+      "grad_norm": 0.48116014554573056,
+      "learning_rate": 1.5249727983780453e-05,
+      "loss": 0.1864,
+      "step": 761
+    },
+    {
+      "epoch": 3.091139240506329,
+      "grad_norm": 0.4688711015124774,
+      "learning_rate": 1.5194612706090786e-05,
+      "loss": 0.1502,
+      "step": 762
+    },
+    {
+      "epoch": 3.0951898734177217,
+      "grad_norm": 0.48485580241748727,
+      "learning_rate": 1.5139536130329771e-05,
+      "loss": 0.1837,
+      "step": 763
+    },
+    {
+      "epoch": 3.0992405063291137,
+      "grad_norm": 0.4724908014564415,
+      "learning_rate": 1.508449870007656e-05,
+      "loss": 0.1637,
+      "step": 764
+    },
+    {
+      "epoch": 3.1032911392405063,
+      "grad_norm": 0.49639112244219097,
+      "learning_rate": 1.5029500858595056e-05,
+      "loss": 0.1674,
+      "step": 765
+    },
+    {
+      "epoch": 3.1073417721518988,
+      "grad_norm": 0.4831733396524702,
+      "learning_rate": 1.4974543048830328e-05,
+      "loss": 0.1777,
+      "step": 766
+    },
+    {
+      "epoch": 3.1113924050632913,
+      "grad_norm": 0.4783009629344156,
+      "learning_rate": 1.4919625713405e-05,
+      "loss": 0.1739,
+      "step": 767
+    },
+    {
+      "epoch": 3.1154430379746834,
+      "grad_norm": 0.4783845631909948,
+      "learning_rate": 1.4864749294615756e-05,
+      "loss": 0.1767,
+      "step": 768
+    },
+    {
+      "epoch": 3.119493670886076,
+      "grad_norm": 0.4668876194816775,
+      "learning_rate": 1.4809914234429716e-05,
+      "loss": 0.1638,
+      "step": 769
+    },
+    {
+      "epoch": 3.1235443037974684,
+      "grad_norm": 0.46251372963794973,
+      "learning_rate": 1.4755120974480923e-05,
+      "loss": 0.1688,
+      "step": 770
+    },
+    {
+      "epoch": 3.127594936708861,
+      "grad_norm": 0.47866157363392525,
+      "learning_rate": 1.4700369956066771e-05,
+      "loss": 0.1871,
+      "step": 771
+    },
+    {
+      "epoch": 3.131645569620253,
+      "grad_norm": 0.4989084010113452,
+      "learning_rate": 1.4645661620144413e-05,
+      "loss": 0.1648,
+      "step": 772
+    },
+    {
+      "epoch": 3.1356962025316455,
+      "grad_norm": 0.47052710853954127,
+      "learning_rate": 1.4590996407327284e-05,
+      "loss": 0.1806,
+      "step": 773
+    },
+    {
+      "epoch": 3.139746835443038,
+      "grad_norm": 0.4821167582636259,
+      "learning_rate": 1.4536374757881487e-05,
+      "loss": 0.1885,
+      "step": 774
+    },
+    {
+      "epoch": 3.1437974683544305,
+      "grad_norm": 0.45912292187891607,
+      "learning_rate": 1.4481797111722271e-05,
+      "loss": 0.171,
+      "step": 775
+    },
+    {
+      "epoch": 3.1478481012658226,
+      "grad_norm": 0.46710809187031876,
+      "learning_rate": 1.4427263908410507e-05,
+      "loss": 0.1772,
+      "step": 776
+    },
+    {
+      "epoch": 3.151898734177215,
+      "grad_norm": 0.4858364549433246,
+      "learning_rate": 1.4372775587149108e-05,
+      "loss": 0.1698,
+      "step": 777
+    },
+    {
+      "epoch": 3.1559493670886076,
+      "grad_norm": 0.4881606337453914,
+      "learning_rate": 1.4318332586779522e-05,
+      "loss": 0.1782,
+      "step": 778
+    },
+    {
+      "epoch": 3.16,
+      "grad_norm": 0.4696933834284012,
+      "learning_rate": 1.4263935345778202e-05,
+      "loss": 0.1754,
+      "step": 779
+    },
+    {
+      "epoch": 3.164050632911392,
+      "grad_norm": 0.47518120746888404,
+      "learning_rate": 1.420958430225303e-05,
+      "loss": 0.1888,
+      "step": 780
+    },
+    {
+      "epoch": 3.1681012658227847,
+      "grad_norm": 0.46457595060502804,
+      "learning_rate": 1.415527989393985e-05,
+      "loss": 0.1765,
+      "step": 781
+    },
+    {
+      "epoch": 3.1721518987341772,
+      "grad_norm": 0.48615729230367516,
+      "learning_rate": 1.410102255819891e-05,
+      "loss": 0.1764,
+      "step": 782
+    },
+    {
+      "epoch": 3.1762025316455698,
+      "grad_norm": 0.48348399203946085,
+      "learning_rate": 1.404681273201131e-05,
+      "loss": 0.1666,
+      "step": 783
+    },
+    {
+      "epoch": 3.180253164556962,
+      "grad_norm": 0.46810602139970603,
+      "learning_rate": 1.399265085197556e-05,
+      "loss": 0.1725,
+      "step": 784
+    },
+    {
+      "epoch": 3.1843037974683543,
+      "grad_norm": 0.4636457920572497,
+      "learning_rate": 1.393853735430398e-05,
+      "loss": 0.1862,
+      "step": 785
+    },
+    {
+      "epoch": 3.188354430379747,
+      "grad_norm": 0.47027734014540257,
+      "learning_rate": 1.3884472674819246e-05,
+      "loss": 0.1702,
+      "step": 786
+    },
+    {
+      "epoch": 3.1924050632911394,
+      "grad_norm": 0.47364174264611847,
+      "learning_rate": 1.3830457248950864e-05,
+      "loss": 0.1758,
+      "step": 787
+    },
+    {
+      "epoch": 3.1964556962025314,
+      "grad_norm": 0.47529527967953367,
+      "learning_rate": 1.377649151173163e-05,
+      "loss": 0.1778,
+      "step": 788
+    },
+    {
+      "epoch": 3.200506329113924,
+      "grad_norm": 0.47142635115450265,
+      "learning_rate": 1.3722575897794181e-05,
+      "loss": 0.1809,
+      "step": 789
+    },
+    {
+      "epoch": 3.2045569620253165,
+      "grad_norm": 0.46997430272972923,
+      "learning_rate": 1.3668710841367472e-05,
+      "loss": 0.1773,
+      "step": 790
+    },
+    {
+      "epoch": 3.208607594936709,
+      "grad_norm": 0.4491274126086909,
+      "learning_rate": 1.361489677627324e-05,
+      "loss": 0.178,
+      "step": 791
+    },
+    {
+      "epoch": 3.212658227848101,
+      "grad_norm": 0.47492287265144123,
+      "learning_rate": 1.3561134135922585e-05,
+      "loss": 0.1505,
+      "step": 792
+    },
+    {
+      "epoch": 3.2167088607594936,
+      "grad_norm": 0.4598692951983188,
+      "learning_rate": 1.350742335331241e-05,
+      "loss": 0.172,
+      "step": 793
+    },
+    {
+      "epoch": 3.220759493670886,
+      "grad_norm": 0.47604127702969534,
+      "learning_rate": 1.345376486102198e-05,
+      "loss": 0.1745,
+      "step": 794
+    },
+    {
+      "epoch": 3.2248101265822786,
+      "grad_norm": 0.4872714227638417,
+      "learning_rate": 1.3400159091209414e-05,
+      "loss": 0.1883,
+      "step": 795
+    },
+    {
+      "epoch": 3.2288607594936707,
+      "grad_norm": 0.46794217420253337,
+      "learning_rate": 1.3346606475608216e-05,
+      "loss": 0.1575,
+      "step": 796
+    },
+    {
+      "epoch": 3.232911392405063,
+      "grad_norm": 0.46872276089525255,
+      "learning_rate": 1.3293107445523781e-05,
+      "loss": 0.1696,
+      "step": 797
+    },
+    {
+      "epoch": 3.2369620253164557,
+      "grad_norm": 0.4856172394537645,
+      "learning_rate": 1.3239662431829949e-05,
+      "loss": 0.1749,
+      "step": 798
+    },
+    {
+      "epoch": 3.2410126582278482,
+      "grad_norm": 0.4935960254383048,
+      "learning_rate": 1.3186271864965509e-05,
+      "loss": 0.1643,
+      "step": 799
+    },
+    {
+      "epoch": 3.2450632911392403,
+      "grad_norm": 0.49129633218035323,
+      "learning_rate": 1.3132936174930756e-05,
+      "loss": 0.1862,
+      "step": 800
+    },
+    {
+      "epoch": 3.249113924050633,
+      "grad_norm": 0.475736850319681,
+      "learning_rate": 1.3079655791283995e-05,
+      "loss": 0.1756,
+      "step": 801
+    },
+    {
+      "epoch": 3.2531645569620253,
+      "grad_norm": 0.478801883042352,
+      "learning_rate": 1.3026431143138108e-05,
+      "loss": 0.1767,
+      "step": 802
+    },
+    {
+      "epoch": 3.257215189873418,
+      "grad_norm": 0.4740522359940149,
+      "learning_rate": 1.2973262659157114e-05,
+      "loss": 0.1725,
+      "step": 803
+    },
+    {
+      "epoch": 3.26126582278481,
+      "grad_norm": 0.4783190370480576,
+      "learning_rate": 1.2920150767552651e-05,
+      "loss": 0.1709,
+      "step": 804
+    },
+    {
+      "epoch": 3.2653164556962024,
+      "grad_norm": 0.4567239850318125,
+      "learning_rate": 1.2867095896080607e-05,
+      "loss": 0.1739,
+      "step": 805
+    },
+    {
+      "epoch": 3.269367088607595,
+      "grad_norm": 0.4764303124610351,
+      "learning_rate": 1.2814098472037612e-05,
+      "loss": 0.1603,
+      "step": 806
+    },
+    {
+      "epoch": 3.2734177215189875,
+      "grad_norm": 0.4442874894207537,
+      "learning_rate": 1.276115892225764e-05,
+      "loss": 0.1648,
+      "step": 807
+    },
+    {
+      "epoch": 3.27746835443038,
+      "grad_norm": 0.460102138024477,
+      "learning_rate": 1.2708277673108555e-05,
+      "loss": 0.1726,
+      "step": 808
+    },
+    {
+      "epoch": 3.281518987341772,
+      "grad_norm": 0.46318300713152155,
+      "learning_rate": 1.2655455150488649e-05,
+      "loss": 0.1664,
+      "step": 809
+    },
+    {
+      "epoch": 3.2855696202531646,
+      "grad_norm": 0.4671965842978333,
+      "learning_rate": 1.2602691779823272e-05,
+      "loss": 0.178,
+      "step": 810
+    },
+    {
+      "epoch": 3.289620253164557,
+      "grad_norm": 0.4697468209365764,
+      "learning_rate": 1.2549987986061355e-05,
+      "loss": 0.181,
+      "step": 811
+    },
+    {
+      "epoch": 3.293670886075949,
+      "grad_norm": 0.45606467184862703,
+      "learning_rate": 1.2497344193672005e-05,
+      "loss": 0.1783,
+      "step": 812
+    },
+    {
+      "epoch": 3.2977215189873417,
+      "grad_norm": 0.46211178360702443,
+      "learning_rate": 1.2444760826641092e-05,
+      "loss": 0.1832,
+      "step": 813
+    },
+    {
+      "epoch": 3.301772151898734,
+      "grad_norm": 0.45649306970090026,
+      "learning_rate": 1.2392238308467817e-05,
+      "loss": 0.1701,
+      "step": 814
+    },
+    {
+      "epoch": 3.3058227848101267,
+      "grad_norm": 0.4559266403516016,
+      "learning_rate": 1.2339777062161326e-05,
+      "loss": 0.1615,
+      "step": 815
+    },
+    {
+      "epoch": 3.309873417721519,
+      "grad_norm": 0.4565480816788766,
+      "learning_rate": 1.2287377510237293e-05,
+      "loss": 0.1817,
+      "step": 816
+    },
+    {
+      "epoch": 3.3139240506329113,
+      "grad_norm": 0.4611346882955347,
+      "learning_rate": 1.2235040074714488e-05,
+      "loss": 0.1783,
+      "step": 817
+    },
+    {
+      "epoch": 3.317974683544304,
+      "grad_norm": 0.46039222415104436,
+      "learning_rate": 1.2182765177111434e-05,
+      "loss": 0.1824,
+      "step": 818
+    },
+    {
+      "epoch": 3.3220253164556963,
+      "grad_norm": 0.4609945063676585,
+      "learning_rate": 1.213055323844297e-05,
+      "loss": 0.179,
+      "step": 819
+    },
+    {
+      "epoch": 3.3260759493670884,
+      "grad_norm": 0.47739813096437467,
+      "learning_rate": 1.2078404679216864e-05,
+      "loss": 0.1723,
+      "step": 820
+    },
+    {
+      "epoch": 3.330126582278481,
+      "grad_norm": 0.46812402377547574,
+      "learning_rate": 1.2026319919430458e-05,
+      "loss": 0.1594,
+      "step": 821
+    },
+    {
+      "epoch": 3.3341772151898734,
+      "grad_norm": 0.4525455097481518,
+      "learning_rate": 1.1974299378567227e-05,
+      "loss": 0.1723,
+      "step": 822
+    },
+    {
+      "epoch": 3.338227848101266,
+      "grad_norm": 0.4744495248821913,
+      "learning_rate": 1.1922343475593462e-05,
+      "loss": 0.1617,
+      "step": 823
+    },
+    {
+      "epoch": 3.3422784810126585,
+      "grad_norm": 0.44746106714406964,
+      "learning_rate": 1.187045262895488e-05,
+      "loss": 0.1784,
+      "step": 824
+    },
+    {
+      "epoch": 3.3463291139240505,
+      "grad_norm": 0.4669653384852627,
+      "learning_rate": 1.1818627256573203e-05,
+      "loss": 0.1574,
+      "step": 825
+    },
+    {
+      "epoch": 3.350379746835443,
+      "grad_norm": 0.4658668966231434,
+      "learning_rate": 1.1766867775842864e-05,
+      "loss": 0.1874,
+      "step": 826
+    },
+    {
+      "epoch": 3.3544303797468356,
+      "grad_norm": 0.47120188961563564,
+      "learning_rate": 1.1715174603627615e-05,
+      "loss": 0.1795,
+      "step": 827
+    },
+    {
+      "epoch": 3.3584810126582276,
+      "grad_norm": 0.46422331019045027,
+      "learning_rate": 1.1663548156257147e-05,
+      "loss": 0.1836,
+      "step": 828
+    },
+    {
+      "epoch": 3.36253164556962,
+      "grad_norm": 0.4528248223833948,
+      "learning_rate": 1.161198884952377e-05,
+      "loss": 0.1537,
+      "step": 829
+    },
+    {
+      "epoch": 3.3665822784810127,
+      "grad_norm": 0.4761337839343475,
+      "learning_rate": 1.1560497098679056e-05,
+      "loss": 0.1745,
+      "step": 830
+    },
+    {
+      "epoch": 3.370632911392405,
+      "grad_norm": 0.47779473321462806,
+      "learning_rate": 1.1509073318430479e-05,
+      "loss": 0.1558,
+      "step": 831
+    },
+    {
+      "epoch": 3.3746835443037977,
+      "grad_norm": 0.4715025112320968,
+      "learning_rate": 1.1457717922938116e-05,
+      "loss": 0.1771,
+      "step": 832
+    },
+    {
+      "epoch": 3.3787341772151898,
+      "grad_norm": 0.4492626613783732,
+      "learning_rate": 1.1406431325811233e-05,
+      "loss": 0.177,
+      "step": 833
+    },
+    {
+      "epoch": 3.3827848101265823,
+      "grad_norm": 0.4706593956347404,
+      "learning_rate": 1.135521394010506e-05,
+      "loss": 0.1724,
+      "step": 834
+    },
+    {
+      "epoch": 3.386835443037975,
+      "grad_norm": 0.4898506232139487,
+      "learning_rate": 1.1304066178317367e-05,
+      "loss": 0.1744,
+      "step": 835
+    },
+    {
+      "epoch": 3.390886075949367,
+      "grad_norm": 0.451757238497467,
+      "learning_rate": 1.1252988452385199e-05,
+      "loss": 0.1862,
+      "step": 836
+    },
+    {
+      "epoch": 3.3949367088607594,
+      "grad_norm": 0.4511378209736183,
+      "learning_rate": 1.1201981173681536e-05,
+      "loss": 0.1765,
+      "step": 837
+    },
+    {
+      "epoch": 3.398987341772152,
+      "grad_norm": 0.4677567769621468,
+      "learning_rate": 1.1151044753011991e-05,
+      "loss": 0.1709,
+      "step": 838
+    },
+    {
+      "epoch": 3.4030379746835444,
+      "grad_norm": 0.4541809940207104,
+      "learning_rate": 1.1100179600611491e-05,
+      "loss": 0.1679,
+      "step": 839
+    },
+    {
+      "epoch": 3.407088607594937,
+      "grad_norm": 0.45308574379622496,
+      "learning_rate": 1.1049386126140985e-05,
+      "loss": 0.1713,
+      "step": 840
+    },
+    {
+      "epoch": 3.411139240506329,
+      "grad_norm": 0.4595244204416368,
+      "learning_rate": 1.0998664738684128e-05,
+      "loss": 0.1662,
+      "step": 841
+    },
+    {
+      "epoch": 3.4151898734177215,
+      "grad_norm": 0.4499126182965053,
+      "learning_rate": 1.0948015846744e-05,
+      "loss": 0.1815,
+      "step": 842
+    },
+    {
+      "epoch": 3.419240506329114,
+      "grad_norm": 0.4560425228429156,
+      "learning_rate": 1.0897439858239832e-05,
+      "loss": 0.1653,
+      "step": 843
+    },
+    {
+      "epoch": 3.423291139240506,
+      "grad_norm": 0.4601254431409881,
+      "learning_rate": 1.0846937180503652e-05,
+      "loss": 0.1801,
+      "step": 844
+    },
+    {
+      "epoch": 3.4273417721518986,
+      "grad_norm": 0.46953213501450625,
+      "learning_rate": 1.0796508220277117e-05,
+      "loss": 0.1641,
+      "step": 845
+    },
+    {
+      "epoch": 3.431392405063291,
+      "grad_norm": 0.46667889055684986,
+      "learning_rate": 1.0746153383708107e-05,
+      "loss": 0.186,
+      "step": 846
+    },
+    {
+      "epoch": 3.4354430379746836,
+      "grad_norm": 0.46661006840467917,
+      "learning_rate": 1.0695873076347579e-05,
+      "loss": 0.1615,
+      "step": 847
+    },
+    {
+      "epoch": 3.439493670886076,
+      "grad_norm": 0.46528674600345327,
+      "learning_rate": 1.0645667703146205e-05,
+      "loss": 0.1573,
+      "step": 848
+    },
+    {
+      "epoch": 3.4435443037974682,
+      "grad_norm": 0.461062822783689,
+      "learning_rate": 1.0595537668451161e-05,
+      "loss": 0.1697,
+      "step": 849
+    },
+    {
+      "epoch": 3.4475949367088607,
+      "grad_norm": 0.47038645095998327,
+      "learning_rate": 1.0545483376002854e-05,
+      "loss": 0.1642,
+      "step": 850
+    },
+    {
+      "epoch": 3.4516455696202533,
+      "grad_norm": 0.46080780000276933,
+      "learning_rate": 1.0495505228931676e-05,
+      "loss": 0.1735,
+      "step": 851
+    },
+    {
+      "epoch": 3.4556962025316453,
+      "grad_norm": 0.4399539498067828,
+      "learning_rate": 1.044560362975474e-05,
+      "loss": 0.1726,
+      "step": 852
+    },
+    {
+      "epoch": 3.459746835443038,
+      "grad_norm": 0.45142587864197625,
+      "learning_rate": 1.0395778980372695e-05,
+      "loss": 0.1654,
+      "step": 853
+    },
+    {
+      "epoch": 3.4637974683544304,
+      "grad_norm": 0.4659484821319064,
+      "learning_rate": 1.0346031682066381e-05,
+      "loss": 0.1799,
+      "step": 854
+    },
+    {
+      "epoch": 3.467848101265823,
+      "grad_norm": 0.4627353099608884,
+      "learning_rate": 1.0296362135493724e-05,
+      "loss": 0.1581,
+      "step": 855
+    },
+    {
+      "epoch": 3.4718987341772154,
+      "grad_norm": 0.457953768505968,
+      "learning_rate": 1.0246770740686422e-05,
+      "loss": 0.1596,
+      "step": 856
+    },
+    {
+      "epoch": 3.4759493670886075,
+      "grad_norm": 0.45976297829060253,
+      "learning_rate": 1.0197257897046743e-05,
+      "loss": 0.1825,
+      "step": 857
+    },
+    {
+      "epoch": 3.48,
+      "grad_norm": 0.46273448350579716,
+      "learning_rate": 1.014782400334433e-05,
+      "loss": 0.1785,
+      "step": 858
+    },
+    {
+      "epoch": 3.4840506329113925,
+      "grad_norm": 0.4557327634007069,
+      "learning_rate": 1.009846945771296e-05,
+      "loss": 0.1821,
+      "step": 859
+    },
+    {
+      "epoch": 3.4881012658227846,
+      "grad_norm": 0.4312815074687807,
+      "learning_rate": 1.0049194657647363e-05,
+      "loss": 0.1731,
+      "step": 860
+    },
+    {
+      "epoch": 3.492151898734177,
+      "grad_norm": 0.4659134775633044,
+      "learning_rate": 1.0000000000000006e-05,
+      "loss": 0.1447,
+      "step": 861
+    },
+    {
+      "epoch": 3.4962025316455696,
+      "grad_norm": 0.438964372622972,
+      "learning_rate": 9.950885880977891e-06,
+      "loss": 0.1651,
+      "step": 862
+    },
+    {
+      "epoch": 3.500253164556962,
+      "grad_norm": 0.43417759926278576,
+      "learning_rate": 9.901852696139382e-06,
+      "loss": 0.1483,
+      "step": 863
+    },
+    {
+      "epoch": 3.5043037974683546,
+      "grad_norm": 0.4501578156112754,
+      "learning_rate": 9.852900840391027e-06,
+      "loss": 0.1676,
+      "step": 864
+    },
+    {
+      "epoch": 3.5083544303797467,
+      "grad_norm": 0.4627156145023752,
+      "learning_rate": 9.804030707984313e-06,
+      "loss": 0.1682,
+      "step": 865
+    },
+    {
+      "epoch": 3.512405063291139,
+      "grad_norm": 0.4477855614096106,
+      "learning_rate": 9.755242692512599e-06,
+      "loss": 0.1675,
+      "step": 866
+    },
+    {
+      "epoch": 3.5164556962025317,
+      "grad_norm": 0.4555742614352992,
+      "learning_rate": 9.70653718690782e-06,
+      "loss": 0.1782,
+      "step": 867
+    },
+    {
+      "epoch": 3.520506329113924,
+      "grad_norm": 0.4530677627327635,
+      "learning_rate": 9.657914583437454e-06,
+      "loss": 0.1668,
+      "step": 868
+    },
+    {
+      "epoch": 3.5245569620253163,
+      "grad_norm": 0.45080137228112893,
+      "learning_rate": 9.609375273701246e-06,
+      "loss": 0.1825,
+      "step": 869
+    },
+    {
+      "epoch": 3.528607594936709,
+      "grad_norm": 0.45709518737282195,
+      "learning_rate": 9.560919648628133e-06,
+      "loss": 0.1629,
+      "step": 870
+    },
+    {
+      "epoch": 3.5326582278481014,
+      "grad_norm": 0.4489061762411965,
+      "learning_rate": 9.512548098473047e-06,
+      "loss": 0.171,
+      "step": 871
+    },
+    {
+      "epoch": 3.536708860759494,
+      "grad_norm": 0.45890548100608847,
+      "learning_rate": 9.464261012813825e-06,
+      "loss": 0.1761,
+      "step": 872
+    },
+    {
+      "epoch": 3.540759493670886,
+      "grad_norm": 0.4679056354580481,
+      "learning_rate": 9.416058780547987e-06,
+      "loss": 0.1741,
+      "step": 873
+    },
+    {
+      "epoch": 3.5448101265822785,
+      "grad_norm": 0.45286989520585924,
+      "learning_rate": 9.367941789889714e-06,
+      "loss": 0.1812,
+      "step": 874
+    },
+    {
+      "epoch": 3.548860759493671,
+      "grad_norm": 0.4430247467477408,
+      "learning_rate": 9.319910428366607e-06,
+      "loss": 0.1593,
+      "step": 875
+    },
+    {
+      "epoch": 3.552911392405063,
+      "grad_norm": 0.4664700368308305,
+      "learning_rate": 9.271965082816667e-06,
+      "loss": 0.1699,
+      "step": 876
+    },
+    {
+      "epoch": 3.5569620253164556,
+      "grad_norm": 0.4644136258031747,
+      "learning_rate": 9.224106139385111e-06,
+      "loss": 0.1627,
+      "step": 877
+    },
+    {
+      "epoch": 3.561012658227848,
+      "grad_norm": 0.45437175073449365,
+      "learning_rate": 9.176333983521291e-06,
+      "loss": 0.1796,
+      "step": 878
+    },
+    {
+      "epoch": 3.5650632911392406,
+      "grad_norm": 0.4459515007434466,
+      "learning_rate": 9.12864899997558e-06,
+      "loss": 0.1808,
+      "step": 879
+    },
+    {
+      "epoch": 3.569113924050633,
+      "grad_norm": 0.44891552005190893,
+      "learning_rate": 9.08105157279628e-06,
+      "loss": 0.1624,
+      "step": 880
+    },
+    {
+      "epoch": 3.573164556962025,
+      "grad_norm": 0.46260460177370344,
+      "learning_rate": 9.03354208532653e-06,
+      "loss": 0.162,
+      "step": 881
+    },
+    {
+      "epoch": 3.5772151898734177,
+      "grad_norm": 0.44945487263804007,
+      "learning_rate": 8.986120920201205e-06,
+      "loss": 0.1727,
+      "step": 882
+    },
+    {
+      "epoch": 3.58126582278481,
+      "grad_norm": 0.44682101601132135,
+      "learning_rate": 8.938788459343852e-06,
+      "loss": 0.1798,
+      "step": 883
+    },
+    {
+      "epoch": 3.5853164556962023,
+      "grad_norm": 0.4302041449774112,
+      "learning_rate": 8.8915450839636e-06,
+      "loss": 0.1645,
+      "step": 884
+    },
+    {
+      "epoch": 3.589367088607595,
+      "grad_norm": 0.4509857531379134,
+      "learning_rate": 8.844391174552116e-06,
+      "loss": 0.1743,
+      "step": 885
+    },
+    {
+      "epoch": 3.5934177215189873,
+      "grad_norm": 0.4637909228891772,
+      "learning_rate": 8.797327110880479e-06,
+      "loss": 0.1613,
+      "step": 886
+    },
+    {
+      "epoch": 3.59746835443038,
+      "grad_norm": 0.44449811993939325,
+      "learning_rate": 8.750353271996206e-06,
+      "loss": 0.1485,
+      "step": 887
+    },
+    {
+      "epoch": 3.6015189873417723,
+      "grad_norm": 0.44919949805470444,
+      "learning_rate": 8.703470036220132e-06,
+      "loss": 0.1702,
+      "step": 888
+    },
+    {
+      "epoch": 3.6055696202531644,
+      "grad_norm": 0.45086223989799723,
+      "learning_rate": 8.656677781143394e-06,
+      "loss": 0.1784,
+      "step": 889
+    },
+    {
+      "epoch": 3.609620253164557,
+      "grad_norm": 0.46286641994706995,
+      "learning_rate": 8.609976883624377e-06,
+      "loss": 0.1881,
+      "step": 890
+    },
+    {
+      "epoch": 3.6136708860759494,
+      "grad_norm": 0.45581230076916185,
+      "learning_rate": 8.563367719785698e-06,
+      "loss": 0.1704,
+      "step": 891
+    },
+    {
+      "epoch": 3.6177215189873415,
+      "grad_norm": 0.4531377784773999,
+      "learning_rate": 8.516850665011138e-06,
+      "loss": 0.1569,
+      "step": 892
+    },
+    {
+      "epoch": 3.621772151898734,
+      "grad_norm": 0.441039711173958,
+      "learning_rate": 8.47042609394269e-06,
+      "loss": 0.1596,
+      "step": 893
+    },
+    {
+      "epoch": 3.6258227848101265,
+      "grad_norm": 0.4437609701719831,
+      "learning_rate": 8.424094380477432e-06,
+      "loss": 0.1696,
+      "step": 894
+    },
+    {
+      "epoch": 3.629873417721519,
+      "grad_norm": 0.45968741862492085,
+      "learning_rate": 8.37785589776465e-06,
+      "loss": 0.1577,
+      "step": 895
+    },
+    {
+      "epoch": 3.6339240506329116,
+      "grad_norm": 0.4557864317211521,
+      "learning_rate": 8.331711018202694e-06,
+      "loss": 0.1834,
+      "step": 896
+    },
+    {
+      "epoch": 3.6379746835443036,
+      "grad_norm": 0.4465099994199235,
+      "learning_rate": 8.285660113436104e-06,
+      "loss": 0.1708,
+      "step": 897
+    },
+    {
+      "epoch": 3.642025316455696,
+      "grad_norm": 0.444485634484314,
+      "learning_rate": 8.239703554352527e-06,
+      "loss": 0.1603,
+      "step": 898
+    },
+    {
+      "epoch": 3.6460759493670887,
+      "grad_norm": 0.4451722032119556,
+      "learning_rate": 8.193841711079775e-06,
+      "loss": 0.168,
+      "step": 899
+    },
+    {
+      "epoch": 3.6501265822784807,
+      "grad_norm": 0.4416930767103303,
+      "learning_rate": 8.148074952982828e-06,
+      "loss": 0.1795,
+      "step": 900
+    },
+    {
+      "epoch": 3.6541772151898733,
+      "grad_norm": 0.4494716564829912,
+      "learning_rate": 8.102403648660859e-06,
+      "loss": 0.1595,
+      "step": 901
+    },
+    {
+      "epoch": 3.6582278481012658,
+      "grad_norm": 0.4435898690473248,
+      "learning_rate": 8.056828165944282e-06,
+      "loss": 0.1739,
+      "step": 902
+    },
+    {
+      "epoch": 3.6622784810126583,
+      "grad_norm": 0.45942289980172635,
+      "learning_rate": 8.011348871891762e-06,
+      "loss": 0.1568,
+      "step": 903
+    },
+    {
+      "epoch": 3.666329113924051,
+      "grad_norm": 0.4521986913805659,
+      "learning_rate": 7.965966132787287e-06,
+      "loss": 0.1648,
+      "step": 904
+    },
+    {
+      "epoch": 3.670379746835443,
+      "grad_norm": 0.460508446665727,
+      "learning_rate": 7.920680314137189e-06,
+      "loss": 0.1903,
+      "step": 905
+    },
+    {
+      "epoch": 3.6744303797468354,
+      "grad_norm": 0.44451054793420675,
+      "learning_rate": 7.875491780667246e-06,
+      "loss": 0.1786,
+      "step": 906
+    },
+    {
+      "epoch": 3.678481012658228,
+      "grad_norm": 0.4464915925632161,
+      "learning_rate": 7.830400896319667e-06,
+      "loss": 0.1591,
+      "step": 907
+    },
+    {
+      "epoch": 3.68253164556962,
+      "grad_norm": 0.4399562210245223,
+      "learning_rate": 7.785408024250259e-06,
+      "loss": 0.168,
+      "step": 908
+    },
+    {
+      "epoch": 3.6865822784810125,
+      "grad_norm": 0.4520238203322593,
+      "learning_rate": 7.74051352682542e-06,
+      "loss": 0.1664,
+      "step": 909
+    },
+    {
+      "epoch": 3.690632911392405,
+      "grad_norm": 0.4337419850460006,
+      "learning_rate": 7.695717765619257e-06,
+      "loss": 0.1754,
+      "step": 910
+    },
+    {
+      "epoch": 3.6946835443037975,
+      "grad_norm": 0.4541417405899359,
+      "learning_rate": 7.651021101410673e-06,
+      "loss": 0.1645,
+      "step": 911
+    },
+    {
+      "epoch": 3.69873417721519,
+      "grad_norm": 0.4438469767043674,
+      "learning_rate": 7.606423894180464e-06,
+      "loss": 0.1536,
+      "step": 912
+    },
+    {
+      "epoch": 3.702784810126582,
+      "grad_norm": 0.4549679911160796,
+      "learning_rate": 7.56192650310839e-06,
+      "loss": 0.1569,
+      "step": 913
+    },
+    {
+      "epoch": 3.7068354430379746,
+      "grad_norm": 0.44901287672291995,
+      "learning_rate": 7.517529286570349e-06,
+      "loss": 0.1527,
+      "step": 914
+    },
+    {
+      "epoch": 3.710886075949367,
+      "grad_norm": 0.45209522815765113,
+      "learning_rate": 7.473232602135387e-06,
+      "loss": 0.1674,
+      "step": 915
+    },
+    {
+      "epoch": 3.714936708860759,
+      "grad_norm": 0.4460518362415555,
+      "learning_rate": 7.429036806562935e-06,
+      "loss": 0.165,
+      "step": 916
+    },
+    {
+      "epoch": 3.7189873417721517,
+      "grad_norm": 0.4580874394721424,
+      "learning_rate": 7.3849422557998455e-06,
+      "loss": 0.176,
+      "step": 917
+    },
+    {
+      "epoch": 3.7230379746835442,
+      "grad_norm": 0.4473393761518577,
+      "learning_rate": 7.340949304977567e-06,
+      "loss": 0.1791,
+      "step": 918
+    },
+    {
+      "epoch": 3.7270886075949368,
+      "grad_norm": 0.4353803125422697,
+      "learning_rate": 7.297058308409282e-06,
+      "loss": 0.1779,
+      "step": 919
+    },
+    {
+      "epoch": 3.7311392405063293,
+      "grad_norm": 0.4266762576894105,
+      "learning_rate": 7.25326961958704e-06,
+      "loss": 0.1595,
+      "step": 920
+    },
+    {
+      "epoch": 3.7351898734177214,
+      "grad_norm": 0.4240785495488553,
+      "learning_rate": 7.209583591178921e-06,
+      "loss": 0.15,
+      "step": 921
+    },
+    {
+      "epoch": 3.739240506329114,
+      "grad_norm": 0.4618746642686346,
+      "learning_rate": 7.1660005750261925e-06,
+      "loss": 0.1751,
+      "step": 922
+    },
+    {
+      "epoch": 3.7432911392405064,
+      "grad_norm": 0.43705669917507584,
+      "learning_rate": 7.1225209221404765e-06,
+      "loss": 0.1543,
+      "step": 923
+    },
+    {
+      "epoch": 3.747341772151899,
+      "grad_norm": 0.4413889345083343,
+      "learning_rate": 7.079144982700909e-06,
+      "loss": 0.1833,
+      "step": 924
+    },
+    {
+      "epoch": 3.7513924050632914,
+      "grad_norm": 0.44138245926080694,
+      "learning_rate": 7.0358731060513695e-06,
+      "loss": 0.1697,
+      "step": 925
+    },
+    {
+      "epoch": 3.7554430379746835,
+      "grad_norm": 0.4358583024715019,
+      "learning_rate": 6.99270564069757e-06,
+      "loss": 0.1627,
+      "step": 926
+    },
+    {
+      "epoch": 3.759493670886076,
+      "grad_norm": 0.44199270558841913,
+      "learning_rate": 6.949642934304375e-06,
+      "loss": 0.1722,
+      "step": 927
+    },
+    {
+      "epoch": 3.7635443037974685,
+      "grad_norm": 0.4420194404681457,
+      "learning_rate": 6.906685333692871e-06,
+      "loss": 0.1667,
+      "step": 928
+    },
+    {
+      "epoch": 3.7675949367088606,
+      "grad_norm": 0.43277513408626395,
+      "learning_rate": 6.86383318483769e-06,
+      "loss": 0.1762,
+      "step": 929
+    },
+    {
+      "epoch": 3.771645569620253,
+      "grad_norm": 0.46545405094009057,
+      "learning_rate": 6.821086832864139e-06,
+      "loss": 0.1512,
+      "step": 930
+    },
+    {
+      "epoch": 3.7756962025316456,
+      "grad_norm": 0.4566793047431844,
+      "learning_rate": 6.77844662204546e-06,
+      "loss": 0.1852,
+      "step": 931
+    },
+    {
+      "epoch": 3.779746835443038,
+      "grad_norm": 0.4465189461919435,
+      "learning_rate": 6.7359128958000455e-06,
+      "loss": 0.1543,
+      "step": 932
+    },
+    {
+      "epoch": 3.7837974683544306,
+      "grad_norm": 0.4472862029306214,
+      "learning_rate": 6.693485996688695e-06,
+      "loss": 0.1714,
+      "step": 933
+    },
+    {
+      "epoch": 3.7878481012658227,
+      "grad_norm": 0.43482245535390757,
+      "learning_rate": 6.651166266411801e-06,
+      "loss": 0.1626,
+      "step": 934
+    },
+    {
+      "epoch": 3.7918987341772152,
+      "grad_norm": 0.45061386461404557,
+      "learning_rate": 6.6089540458066725e-06,
+      "loss": 0.1691,
+      "step": 935
+    },
+    {
+      "epoch": 3.7959493670886078,
+      "grad_norm": 0.4556978554787606,
+      "learning_rate": 6.566849674844711e-06,
+      "loss": 0.1727,
+      "step": 936
+    },
+    {
+      "epoch": 3.8,
+      "grad_norm": 0.45621777270414987,
+      "learning_rate": 6.524853492628747e-06,
+      "loss": 0.1439,
+      "step": 937
+    },
+    {
+      "epoch": 3.8040506329113923,
+      "grad_norm": 0.43592086727145707,
+      "learning_rate": 6.4829658373902536e-06,
+      "loss": 0.1899,
+      "step": 938
+    },
+    {
+      "epoch": 3.808101265822785,
+      "grad_norm": 0.4341235379622057,
+      "learning_rate": 6.441187046486648e-06,
+      "loss": 0.1646,
+      "step": 939
+    },
+    {
+      "epoch": 3.8121518987341774,
+      "grad_norm": 0.4401539656047739,
+      "learning_rate": 6.399517456398567e-06,
+      "loss": 0.1596,
+      "step": 940
+    },
+    {
+      "epoch": 3.81620253164557,
+      "grad_norm": 0.439686877749814,
+      "learning_rate": 6.357957402727164e-06,
+      "loss": 0.1708,
+      "step": 941
+    },
+    {
+      "epoch": 3.820253164556962,
+      "grad_norm": 0.44113340004329726,
+      "learning_rate": 6.316507220191395e-06,
+      "loss": 0.1569,
+      "step": 942
+    },
+    {
+      "epoch": 3.8243037974683545,
+      "grad_norm": 0.727674800140053,
+      "learning_rate": 6.275167242625331e-06,
+      "loss": 0.1601,
+      "step": 943
+    },
+    {
+      "epoch": 3.828354430379747,
+      "grad_norm": 0.43779586968116,
+      "learning_rate": 6.233937802975471e-06,
+      "loss": 0.1652,
+      "step": 944
+    },
+    {
+      "epoch": 3.832405063291139,
+      "grad_norm": 0.4328430398145829,
+      "learning_rate": 6.192819233298046e-06,
+      "loss": 0.1593,
+      "step": 945
+    },
+    {
+      "epoch": 3.8364556962025316,
+      "grad_norm": 0.4387413000241011,
+      "learning_rate": 6.151811864756383e-06,
+      "loss": 0.1715,
+      "step": 946
+    },
+    {
+      "epoch": 3.840506329113924,
+      "grad_norm": 0.4340317460137217,
+      "learning_rate": 6.1109160276181655e-06,
+      "loss": 0.1517,
+      "step": 947
+    },
+    {
+      "epoch": 3.8445569620253166,
+      "grad_norm": 0.4484354406848648,
+      "learning_rate": 6.070132051252868e-06,
+      "loss": 0.1637,
+      "step": 948
+    },
+    {
+      "epoch": 3.848607594936709,
+      "grad_norm": 0.4383435532150945,
+      "learning_rate": 6.0294602641290034e-06,
+      "loss": 0.1605,
+      "step": 949
+    },
+    {
+      "epoch": 3.852658227848101,
+      "grad_norm": 0.45477837334484916,
+      "learning_rate": 5.988900993811575e-06,
+      "loss": 0.1745,
+      "step": 950
+    },
+    {
+      "epoch": 3.8567088607594937,
+      "grad_norm": 0.45608659826152476,
+      "learning_rate": 5.948454566959363e-06,
+      "loss": 0.1604,
+      "step": 951
+    },
+    {
+      "epoch": 3.8607594936708862,
+      "grad_norm": 0.4364088938097114,
+      "learning_rate": 5.908121309322328e-06,
+      "loss": 0.15,
+      "step": 952
+    },
+    {
+      "epoch": 3.8648101265822783,
+      "grad_norm": 0.43305993753504235,
+      "learning_rate": 5.867901545738976e-06,
+      "loss": 0.158,
+      "step": 953
+    },
+    {
+      "epoch": 3.868860759493671,
+      "grad_norm": 0.4341950521999476,
+      "learning_rate": 5.827795600133774e-06,
+      "loss": 0.1598,
+      "step": 954
+    },
+    {
+      "epoch": 3.8729113924050633,
+      "grad_norm": 0.4368222335389369,
+      "learning_rate": 5.787803795514466e-06,
+      "loss": 0.1812,
+      "step": 955
+    },
+    {
+      "epoch": 3.876962025316456,
+      "grad_norm": 0.43840430095527977,
+      "learning_rate": 5.747926453969576e-06,
+      "loss": 0.1713,
+      "step": 956
+    },
+    {
+      "epoch": 3.8810126582278484,
+      "grad_norm": 0.4377652089656497,
+      "learning_rate": 5.708163896665708e-06,
+      "loss": 0.1664,
+      "step": 957
+    },
+    {
+      "epoch": 3.8850632911392404,
+      "grad_norm": 0.43322712257026247,
+      "learning_rate": 5.668516443845047e-06,
+      "loss": 0.157,
+      "step": 958
+    },
+    {
+      "epoch": 3.889113924050633,
+      "grad_norm": 0.4282140224907731,
+      "learning_rate": 5.6289844148227225e-06,
+      "loss": 0.1708,
+      "step": 959
+    },
+    {
+      "epoch": 3.8931645569620255,
+      "grad_norm": 0.450386774633833,
+      "learning_rate": 5.5895681279842615e-06,
+      "loss": 0.179,
+      "step": 960
+    },
+    {
+      "epoch": 3.8972151898734175,
+      "grad_norm": 0.42776547117901154,
+      "learning_rate": 5.550267900783019e-06,
+      "loss": 0.1666,
+      "step": 961
+    },
+    {
+      "epoch": 3.90126582278481,
+      "grad_norm": 0.4409751814892039,
+      "learning_rate": 5.511084049737623e-06,
+      "loss": 0.1565,
+      "step": 962
+    },
+    {
+      "epoch": 3.9053164556962026,
+      "grad_norm": 0.4436827772324616,
+      "learning_rate": 5.4720168904294215e-06,
+      "loss": 0.1431,
+      "step": 963
+    },
+    {
+      "epoch": 3.909367088607595,
+      "grad_norm": 0.4485690665361972,
+      "learning_rate": 5.433066737499948e-06,
+      "loss": 0.1672,
+      "step": 964
+    },
+    {
+      "epoch": 3.9134177215189876,
+      "grad_norm": 0.4399151392459079,
+      "learning_rate": 5.394233904648376e-06,
+      "loss": 0.1782,
+      "step": 965
+    },
+    {
+      "epoch": 3.9174683544303797,
+      "grad_norm": 0.4414122837875361,
+      "learning_rate": 5.355518704628997e-06,
+      "loss": 0.1687,
+      "step": 966
+    },
+    {
+      "epoch": 3.921518987341772,
+      "grad_norm": 0.44165647465351154,
+      "learning_rate": 5.316921449248731e-06,
+      "loss": 0.168,
+      "step": 967
+    },
+    {
+      "epoch": 3.9255696202531647,
+      "grad_norm": 0.44685737041931434,
+      "learning_rate": 5.278442449364538e-06,
+      "loss": 0.1788,
+      "step": 968
+    },
+    {
+      "epoch": 3.9296202531645568,
+      "grad_norm": 0.43996063432500854,
+      "learning_rate": 5.240082014881016e-06,
+      "loss": 0.17,
+      "step": 969
+    },
+    {
+      "epoch": 3.9336708860759493,
+      "grad_norm": 0.4329358627064166,
+      "learning_rate": 5.201840454747822e-06,
+      "loss": 0.159,
+      "step": 970
+    },
+    {
+      "epoch": 3.937721518987342,
+      "grad_norm": 0.4427176891875773,
+      "learning_rate": 5.163718076957223e-06,
+      "loss": 0.1452,
+      "step": 971
+    },
+    {
+      "epoch": 3.9417721518987343,
+      "grad_norm": 0.43896230448158713,
+      "learning_rate": 5.125715188541609e-06,
+      "loss": 0.1682,
+      "step": 972
+    },
+    {
+      "epoch": 3.945822784810127,
+      "grad_norm": 0.44053256989145456,
+      "learning_rate": 5.087832095571021e-06,
+      "loss": 0.1713,
+      "step": 973
+    },
+    {
+      "epoch": 3.949873417721519,
+      "grad_norm": 0.42639859844921046,
+      "learning_rate": 5.0500691031506766e-06,
+      "loss": 0.1583,
+      "step": 974
+    },
+    {
+      "epoch": 3.9539240506329114,
+      "grad_norm": 0.4362464413568351,
+      "learning_rate": 5.01242651541854e-06,
+      "loss": 0.1724,
+      "step": 975
+    },
+    {
+      "epoch": 3.957974683544304,
+      "grad_norm": 0.43209837469965784,
+      "learning_rate": 4.974904635542815e-06,
+      "loss": 0.1724,
+      "step": 976
+    },
+    {
+      "epoch": 3.962025316455696,
+      "grad_norm": 0.4362308844926615,
+      "learning_rate": 4.937503765719582e-06,
+      "loss": 0.1557,
+      "step": 977
+    },
+    {
+      "epoch": 3.9660759493670885,
+      "grad_norm": 0.4531678303728902,
+      "learning_rate": 4.900224207170299e-06,
+      "loss": 0.1697,
+      "step": 978
+    },
+    {
+      "epoch": 3.970126582278481,
+      "grad_norm": 0.4427321530856479,
+      "learning_rate": 4.8630662601394065e-06,
+      "loss": 0.1558,
+      "step": 979
+    },
+    {
+      "epoch": 3.9741772151898735,
+      "grad_norm": 0.4329912722597872,
+      "learning_rate": 4.8260302238918995e-06,
+      "loss": 0.1723,
+      "step": 980
+    },
+    {
+      "epoch": 3.978227848101266,
+      "grad_norm": 0.43316473210312695,
+      "learning_rate": 4.789116396710924e-06,
+      "loss": 0.1661,
+      "step": 981
+    },
+    {
+      "epoch": 3.982278481012658,
+      "grad_norm": 0.4366797852493454,
+      "learning_rate": 4.752325075895368e-06,
+      "loss": 0.1602,
+      "step": 982
+    },
+    {
+      "epoch": 3.9863291139240506,
+      "grad_norm": 0.4530967672816123,
+      "learning_rate": 4.715656557757473e-06,
+      "loss": 0.1528,
+      "step": 983
+    },
+    {
+      "epoch": 3.990379746835443,
+      "grad_norm": 0.4405588893762934,
+      "learning_rate": 4.679111137620442e-06,
+      "loss": 0.1556,
+      "step": 984
+    },
+    {
+      "epoch": 3.9944303797468352,
+      "grad_norm": 0.44648149463819436,
+      "learning_rate": 4.6426891098160585e-06,
+      "loss": 0.1744,
+      "step": 985
+    },
+    {
+      "epoch": 3.9984810126582278,
+      "grad_norm": 0.4359937378606629,
+      "learning_rate": 4.6063907676823474e-06,
+      "loss": 0.1919,
+      "step": 986
+    },
+    {
+      "epoch": 4.004050632911392,
+      "grad_norm": 0.4948465161896855,
+      "learning_rate": 4.570216403561141e-06,
+      "loss": 0.1229,
+      "step": 987
+    },
+    {
+      "epoch": 4.008101265822785,
+      "grad_norm": 0.42849905773702746,
+      "learning_rate": 4.534166308795815e-06,
+      "loss": 0.1015,
+      "step": 988
+    },
+    {
+      "epoch": 4.012151898734177,
+      "grad_norm": 0.4156276842213543,
+      "learning_rate": 4.498240773728859e-06,
+      "loss": 0.105,
+      "step": 989
+    },
+    {
+      "epoch": 4.01620253164557,
+      "grad_norm": 0.36679034153844736,
+      "learning_rate": 4.462440087699609e-06,
+      "loss": 0.0978,
+      "step": 990
+    },
+    {
+      "epoch": 4.020253164556962,
+      "grad_norm": 0.3340864006285961,
+      "learning_rate": 4.426764539041861e-06,
+      "loss": 0.1072,
+      "step": 991
+    },
+    {
+      "epoch": 4.024303797468354,
+      "grad_norm": 0.32100060183401286,
+      "learning_rate": 4.391214415081582e-06,
+      "loss": 0.0941,
+      "step": 992
+    },
+    {
+      "epoch": 4.028354430379747,
+      "grad_norm": 0.3434641562554867,
+      "learning_rate": 4.355790002134579e-06,
+      "loss": 0.0986,
+      "step": 993
+    },
+    {
+      "epoch": 4.032405063291139,
+      "grad_norm": 0.4327757576839274,
+      "learning_rate": 4.320491585504207e-06,
+      "loss": 0.1114,
+      "step": 994
+    },
+    {
+      "epoch": 4.036455696202531,
+      "grad_norm": 0.46760489435698865,
+      "learning_rate": 4.2853194494790615e-06,
+      "loss": 0.1133,
+      "step": 995
+    },
+    {
+      "epoch": 4.040506329113924,
+      "grad_norm": 0.43128488001803955,
+      "learning_rate": 4.250273877330691e-06,
+      "loss": 0.1024,
+      "step": 996
+    },
+    {
+      "epoch": 4.044556962025316,
+      "grad_norm": 0.43559590826376887,
+      "learning_rate": 4.215355151311313e-06,
+      "loss": 0.106,
+      "step": 997
+    },
+    {
+      "epoch": 4.048607594936709,
+      "grad_norm": 0.423171521568045,
+      "learning_rate": 4.180563552651542e-06,
+      "loss": 0.108,
+      "step": 998
+    },
+    {
+      "epoch": 4.052658227848101,
+      "grad_norm": 0.36648312953563567,
+      "learning_rate": 4.145899361558147e-06,
+      "loss": 0.1007,
+      "step": 999
+    },
+    {
+      "epoch": 4.056708860759493,
+      "grad_norm": 0.3392403719107097,
+      "learning_rate": 4.111362857211738e-06,
+      "loss": 0.106,
+      "step": 1000
+    },
+    {
+      "epoch": 4.060759493670886,
+      "grad_norm": 0.3456802420100745,
+      "learning_rate": 4.076954317764592e-06,
+      "loss": 0.0943,
+      "step": 1001
+    },
+    {
+      "epoch": 4.0648101265822785,
+      "grad_norm": 0.3674116655462302,
+      "learning_rate": 4.042674020338335e-06,
+      "loss": 0.1024,
+      "step": 1002
+    },
+    {
+      "epoch": 4.0688607594936705,
+      "grad_norm": 0.3926831655958058,
+      "learning_rate": 4.0085222410217835e-06,
+      "loss": 0.1199,
+      "step": 1003
+    },
+    {
+      "epoch": 4.0729113924050635,
+      "grad_norm": 0.388267365877086,
+      "learning_rate": 3.974499254868674e-06,
+      "loss": 0.1049,
+      "step": 1004
+    },
+    {
+      "epoch": 4.076962025316456,
+      "grad_norm": 0.3385847191120139,
+      "learning_rate": 3.940605335895451e-06,
+      "loss": 0.1,
+      "step": 1005
+    },
+    {
+      "epoch": 4.0810126582278485,
+      "grad_norm": 0.32041627889315627,
+      "learning_rate": 3.90684075707908e-06,
+      "loss": 0.108,
+      "step": 1006
+    },
+    {
+      "epoch": 4.085063291139241,
+      "grad_norm": 0.33559936580466876,
+      "learning_rate": 3.8732057903548505e-06,
+      "loss": 0.1044,
+      "step": 1007
+    },
+    {
+      "epoch": 4.089113924050633,
+      "grad_norm": 0.3554107166893919,
+      "learning_rate": 3.8397007066141375e-06,
+      "loss": 0.1206,
+      "step": 1008
+    },
+    {
+      "epoch": 4.093164556962026,
+      "grad_norm": 0.3215324178530859,
+      "learning_rate": 3.806325775702304e-06,
+      "loss": 0.0816,
+      "step": 1009
+    },
+    {
+      "epoch": 4.097215189873418,
+      "grad_norm": 0.34817704501848934,
+      "learning_rate": 3.773081266416434e-06,
+      "loss": 0.1093,
+      "step": 1010
+    },
+    {
+      "epoch": 4.10126582278481,
+      "grad_norm": 0.35782961889007764,
+      "learning_rate": 3.739967446503245e-06,
+      "loss": 0.1113,
+      "step": 1011
+    },
+    {
+      "epoch": 4.105316455696203,
+      "grad_norm": 0.35449378029469825,
+      "learning_rate": 3.706984582656894e-06,
+      "loss": 0.1094,
+      "step": 1012
+    },
+    {
+      "epoch": 4.109367088607595,
+      "grad_norm": 0.363393676560073,
+      "learning_rate": 3.6741329405168237e-06,
+      "loss": 0.0934,
+      "step": 1013
+    },
+    {
+      "epoch": 4.113417721518988,
+      "grad_norm": 0.34526561450404797,
+      "learning_rate": 3.641412784665648e-06,
+      "loss": 0.111,
+      "step": 1014
+    },
+    {
+      "epoch": 4.11746835443038,
+      "grad_norm": 0.32782635739981364,
+      "learning_rate": 3.608824378627005e-06,
+      "loss": 0.0924,
+      "step": 1015
+    },
+    {
+      "epoch": 4.121518987341772,
+      "grad_norm": 0.32972395253626813,
+      "learning_rate": 3.5763679848634337e-06,
+      "loss": 0.0834,
+      "step": 1016
+    },
+    {
+      "epoch": 4.125569620253165,
+      "grad_norm": 0.3411781825174774,
+      "learning_rate": 3.544043864774269e-06,
+      "loss": 0.0998,
+      "step": 1017
+    },
+    {
+      "epoch": 4.129620253164557,
+      "grad_norm": 0.3394710240280812,
+      "learning_rate": 3.5118522786935282e-06,
+      "loss": 0.0813,
+      "step": 1018
+    },
+    {
+      "epoch": 4.133670886075949,
+      "grad_norm": 0.34196425465672703,
+      "learning_rate": 3.479793485887819e-06,
+      "loss": 0.0932,
+      "step": 1019
+    },
+    {
+      "epoch": 4.137721518987342,
+      "grad_norm": 0.3538708515515628,
+      "learning_rate": 3.4478677445542653e-06,
+      "loss": 0.0953,
+      "step": 1020
+    },
+    {
+      "epoch": 4.141772151898734,
+      "grad_norm": 0.33553836967244355,
+      "learning_rate": 3.4160753118183767e-06,
+      "loss": 0.0889,
+      "step": 1021
+    },
+    {
+      "epoch": 4.145822784810127,
+      "grad_norm": 0.3509012046970835,
+      "learning_rate": 3.3844164437320527e-06,
+      "loss": 0.1134,
+      "step": 1022
+    },
+    {
+      "epoch": 4.149873417721519,
+      "grad_norm": 0.33522234135324436,
+      "learning_rate": 3.3528913952714558e-06,
+      "loss": 0.0854,
+      "step": 1023
+    },
+    {
+      "epoch": 4.153924050632911,
+      "grad_norm": 0.3458908747498122,
+      "learning_rate": 3.321500420335e-06,
+      "loss": 0.1009,
+      "step": 1024
+    },
+    {
+      "epoch": 4.157974683544304,
+      "grad_norm": 0.34657178227273683,
+      "learning_rate": 3.290243771741275e-06,
+      "loss": 0.0945,
+      "step": 1025
+    },
+    {
+      "epoch": 4.162025316455696,
+      "grad_norm": 0.3598392956995071,
+      "learning_rate": 3.2591217012270325e-06,
+      "loss": 0.0928,
+      "step": 1026
+    },
+    {
+      "epoch": 4.166075949367088,
+      "grad_norm": 0.33459198523127587,
+      "learning_rate": 3.228134459445149e-06,
+      "loss": 0.0973,
+      "step": 1027
+    },
+    {
+      "epoch": 4.170126582278481,
+      "grad_norm": 0.3359518215094223,
+      "learning_rate": 3.1972822959626205e-06,
+      "loss": 0.0974,
+      "step": 1028
+    },
+    {
+      "epoch": 4.174177215189873,
+      "grad_norm": 0.358139473300798,
+      "learning_rate": 3.166565459258513e-06,
+      "loss": 0.1051,
+      "step": 1029
+    },
+    {
+      "epoch": 4.178227848101266,
+      "grad_norm": 0.3478046798491538,
+      "learning_rate": 3.1359841967220193e-06,
+      "loss": 0.0976,
+      "step": 1030
+    },
+    {
+      "epoch": 4.182278481012658,
+      "grad_norm": 0.32762863231320827,
+      "learning_rate": 3.105538754650419e-06,
+      "loss": 0.0839,
+      "step": 1031
+    },
+    {
+      "epoch": 4.18632911392405,
+      "grad_norm": 0.3299139508134279,
+      "learning_rate": 3.07522937824712e-06,
+      "loss": 0.0918,
+      "step": 1032
+    },
+    {
+      "epoch": 4.190379746835443,
+      "grad_norm": 0.3302173138919579,
+      "learning_rate": 3.0450563116196697e-06,
+      "loss": 0.0825,
+      "step": 1033
+    },
+    {
+      "epoch": 4.194430379746835,
+      "grad_norm": 0.3325794259560467,
+      "learning_rate": 3.0150197977778008e-06,
+      "loss": 0.1033,
+      "step": 1034
+    },
+    {
+      "epoch": 4.1984810126582275,
+      "grad_norm": 0.3192600011133069,
+      "learning_rate": 2.985120078631465e-06,
+      "loss": 0.1069,
+      "step": 1035
+    },
+    {
+      "epoch": 4.2025316455696204,
+      "grad_norm": 0.34865391000958446,
+      "learning_rate": 2.9553573949888893e-06,
+      "loss": 0.107,
+      "step": 1036
+    },
+    {
+      "epoch": 4.2065822784810125,
+      "grad_norm": 0.3331037020813621,
+      "learning_rate": 2.9257319865546384e-06,
+      "loss": 0.0998,
+      "step": 1037
+    },
+    {
+      "epoch": 4.2106329113924055,
+      "grad_norm": 0.32966167318532935,
+      "learning_rate": 2.896244091927678e-06,
+      "loss": 0.0973,
+      "step": 1038
+    },
+    {
+      "epoch": 4.2146835443037975,
+      "grad_norm": 0.33915282879230996,
+      "learning_rate": 2.8668939485994584e-06,
+      "loss": 0.0896,
+      "step": 1039
+    },
+    {
+      "epoch": 4.21873417721519,
+      "grad_norm": 0.3474364729549145,
+      "learning_rate": 2.837681792951994e-06,
+      "loss": 0.0914,
+      "step": 1040
+    },
+    {
+      "epoch": 4.222784810126583,
+      "grad_norm": 0.3445631849952997,
+      "learning_rate": 2.808607860255981e-06,
+      "loss": 0.0986,
+      "step": 1041
+    },
+    {
+      "epoch": 4.226835443037975,
+      "grad_norm": 0.33613073218877326,
+      "learning_rate": 2.7796723846688634e-06,
+      "loss": 0.1057,
+      "step": 1042
+    },
+    {
+      "epoch": 4.230886075949367,
+      "grad_norm": 0.3533049071454075,
+      "learning_rate": 2.7508755992329937e-06,
+      "loss": 0.1049,
+      "step": 1043
+    },
+    {
+      "epoch": 4.23493670886076,
+      "grad_norm": 0.35246335588209365,
+      "learning_rate": 2.722217735873718e-06,
+      "loss": 0.112,
+      "step": 1044
+    },
+    {
+      "epoch": 4.238987341772152,
+      "grad_norm": 0.35151742074480996,
+      "learning_rate": 2.6936990253975315e-06,
+      "loss": 0.084,
+      "step": 1045
+    },
+    {
+      "epoch": 4.243037974683545,
+      "grad_norm": 0.3598340223399337,
+      "learning_rate": 2.665319697490205e-06,
+      "loss": 0.095,
+      "step": 1046
+    },
+    {
+      "epoch": 4.247088607594937,
+      "grad_norm": 0.3349862442290465,
+      "learning_rate": 2.637079980714945e-06,
+      "loss": 0.0904,
+      "step": 1047
+    },
+    {
+      "epoch": 4.251139240506329,
+      "grad_norm": 0.3260708651480878,
+      "learning_rate": 2.6089801025105453e-06,
+      "loss": 0.081,
+      "step": 1048
+    },
+    {
+      "epoch": 4.255189873417722,
+      "grad_norm": 0.338047686632206,
+      "learning_rate": 2.581020289189571e-06,
+      "loss": 0.0965,
+      "step": 1049
+    },
+    {
+      "epoch": 4.259240506329114,
+      "grad_norm": 0.33934135096078155,
+      "learning_rate": 2.553200765936501e-06,
+      "loss": 0.1123,
+      "step": 1050
+    },
+    {
+      "epoch": 4.263291139240506,
+      "grad_norm": 0.3476865802574614,
+      "learning_rate": 2.525521756805962e-06,
+      "loss": 0.1004,
+      "step": 1051
+    },
+    {
+      "epoch": 4.267341772151899,
+      "grad_norm": 0.31982491323490575,
+      "learning_rate": 2.497983484720885e-06,
+      "loss": 0.0843,
+      "step": 1052
+    },
+    {
+      "epoch": 4.271392405063291,
+      "grad_norm": 0.32980115861387144,
+      "learning_rate": 2.470586171470728e-06,
+      "loss": 0.0899,
+      "step": 1053
+    },
+    {
+      "epoch": 4.275443037974684,
+      "grad_norm": 0.3174617893229616,
+      "learning_rate": 2.4433300377096836e-06,
+      "loss": 0.092,
+      "step": 1054
+    },
+    {
+      "epoch": 4.279493670886076,
+      "grad_norm": 0.31671715641178927,
+      "learning_rate": 2.4162153029549073e-06,
+      "loss": 0.0844,
+      "step": 1055
+    },
+    {
+      "epoch": 4.283544303797468,
+      "grad_norm": 0.32918152695151326,
+      "learning_rate": 2.3892421855847458e-06,
+      "loss": 0.0893,
+      "step": 1056
+    },
+    {
+      "epoch": 4.287594936708861,
+      "grad_norm": 0.33251975975936887,
+      "learning_rate": 2.362410902836978e-06,
+      "loss": 0.0912,
+      "step": 1057
+    },
+    {
+      "epoch": 4.291645569620253,
+      "grad_norm": 0.3266450704602441,
+      "learning_rate": 2.3357216708070653e-06,
+      "loss": 0.1035,
+      "step": 1058
+    },
+    {
+      "epoch": 4.295696202531645,
+      "grad_norm": 0.32572068037951224,
+      "learning_rate": 2.309174704446411e-06,
+      "loss": 0.0735,
+      "step": 1059
+    },
+    {
+      "epoch": 4.299746835443038,
+      "grad_norm": 0.32826001987370446,
+      "learning_rate": 2.2827702175606437e-06,
+      "loss": 0.107,
+      "step": 1060
+    },
+    {
+      "epoch": 4.30379746835443,
+      "grad_norm": 0.3375241938203144,
+      "learning_rate": 2.256508422807855e-06,
+      "loss": 0.1039,
+      "step": 1061
+    },
+    {
+      "epoch": 4.307848101265823,
+      "grad_norm": 0.34359080344887494,
+      "learning_rate": 2.230389531696946e-06,
+      "loss": 0.0927,
+      "step": 1062
+    },
+    {
+      "epoch": 4.311898734177215,
+      "grad_norm": 0.34050616753285223,
+      "learning_rate": 2.204413754585857e-06,
+      "loss": 0.087,
+      "step": 1063
+    },
+    {
+      "epoch": 4.315949367088607,
+      "grad_norm": 0.32422400800496265,
+      "learning_rate": 2.1785813006799406e-06,
+      "loss": 0.1012,
+      "step": 1064
+    },
+    {
+      "epoch": 4.32,
+      "grad_norm": 0.3327458650398495,
+      "learning_rate": 2.1528923780302224e-06,
+      "loss": 0.0786,
+      "step": 1065
+    },
+    {
+      "epoch": 4.324050632911392,
+      "grad_norm": 0.31802963759728314,
+      "learning_rate": 2.127347193531757e-06,
+      "loss": 0.1003,
+      "step": 1066
+    },
+    {
+      "epoch": 4.328101265822784,
+      "grad_norm": 0.32973999173821267,
+      "learning_rate": 2.101945952921942e-06,
+      "loss": 0.0868,
+      "step": 1067
+    },
+    {
+      "epoch": 4.332151898734177,
+      "grad_norm": 0.3288733586431568,
+      "learning_rate": 2.0766888607788906e-06,
+      "loss": 0.0984,
+      "step": 1068
+    },
+    {
+      "epoch": 4.3362025316455695,
+      "grad_norm": 0.3093571980280102,
+      "learning_rate": 2.0515761205197337e-06,
+      "loss": 0.0916,
+      "step": 1069
+    },
+    {
+      "epoch": 4.340253164556962,
+      "grad_norm": 0.33308309407551095,
+      "learning_rate": 2.0266079343990453e-06,
+      "loss": 0.1044,
+      "step": 1070
+    },
+    {
+      "epoch": 4.3443037974683545,
+      "grad_norm": 0.3324145958343055,
+      "learning_rate": 2.0017845035071494e-06,
+      "loss": 0.1018,
+      "step": 1071
+    },
+    {
+      "epoch": 4.348354430379747,
+      "grad_norm": 0.3242293760965934,
+      "learning_rate": 1.9771060277685537e-06,
+      "loss": 0.0914,
+      "step": 1072
+    },
+    {
+      "epoch": 4.3524050632911395,
+      "grad_norm": 0.3380439446662605,
+      "learning_rate": 1.95257270594031e-06,
+      "loss": 0.0854,
+      "step": 1073
+    },
+    {
+      "epoch": 4.356455696202532,
+      "grad_norm": 0.3428754182995044,
+      "learning_rate": 1.9281847356104188e-06,
+      "loss": 0.1119,
+      "step": 1074
+    },
+    {
+      "epoch": 4.360506329113924,
+      "grad_norm": 0.32950077516995924,
+      "learning_rate": 1.9039423131962365e-06,
+      "loss": 0.1007,
+      "step": 1075
+    },
+    {
+      "epoch": 4.364556962025317,
+      "grad_norm": 0.3265196879898116,
+      "learning_rate": 1.8798456339429027e-06,
+      "loss": 0.1048,
+      "step": 1076
+    },
+    {
+      "epoch": 4.368607594936709,
+      "grad_norm": 0.341231301545443,
+      "learning_rate": 1.8558948919217612e-06,
+      "loss": 0.0954,
+      "step": 1077
+    },
+    {
+      "epoch": 4.372658227848102,
+      "grad_norm": 0.34717473884045674,
+      "learning_rate": 1.8320902800287954e-06,
+      "loss": 0.1098,
+      "step": 1078
+    },
+    {
+      "epoch": 4.376708860759494,
+      "grad_norm": 0.3329429165211134,
+      "learning_rate": 1.8084319899830726e-06,
+      "loss": 0.1021,
+      "step": 1079
+    },
+    {
+      "epoch": 4.380759493670886,
+      "grad_norm": 0.3392905860681266,
+      "learning_rate": 1.7849202123252097e-06,
+      "loss": 0.0876,
+      "step": 1080
+    },
+    {
+      "epoch": 4.384810126582279,
+      "grad_norm": 0.3236836533095692,
+      "learning_rate": 1.7615551364158401e-06,
+      "loss": 0.1054,
+      "step": 1081
+    },
+    {
+      "epoch": 4.388860759493671,
+      "grad_norm": 0.3164553226539558,
+      "learning_rate": 1.738336950434061e-06,
+      "loss": 0.0977,
+      "step": 1082
+    },
+    {
+      "epoch": 4.392911392405063,
+      "grad_norm": 0.3375402055418742,
+      "learning_rate": 1.715265841375957e-06,
+      "loss": 0.0986,
+      "step": 1083
+    },
+    {
+      "epoch": 4.396962025316456,
+      "grad_norm": 0.3287307228988444,
+      "learning_rate": 1.6923419950530684e-06,
+      "loss": 0.096,
+      "step": 1084
+    },
+    {
+      "epoch": 4.401012658227848,
+      "grad_norm": 0.3388915441577605,
+      "learning_rate": 1.6695655960909008e-06,
+      "loss": 0.0916,
+      "step": 1085
+    },
+    {
+      "epoch": 4.405063291139241,
+      "grad_norm": 0.35428120714275607,
+      "learning_rate": 1.646936827927441e-06,
+      "loss": 0.0971,
+      "step": 1086
+    },
+    {
+      "epoch": 4.409113924050633,
+      "grad_norm": 0.3154212366764666,
+      "learning_rate": 1.6244558728116766e-06,
+      "loss": 0.0876,
+      "step": 1087
+    },
+    {
+      "epoch": 4.413164556962025,
+      "grad_norm": 0.3272193841808593,
+      "learning_rate": 1.6021229118021265e-06,
+      "loss": 0.1025,
+      "step": 1088
+    },
+    {
+      "epoch": 4.417215189873418,
+      "grad_norm": 0.346989187809871,
+      "learning_rate": 1.5799381247653967e-06,
+      "loss": 0.1031,
+      "step": 1089
+    },
+    {
+      "epoch": 4.42126582278481,
+      "grad_norm": 0.34216706245320705,
+      "learning_rate": 1.5579016903747013e-06,
+      "loss": 0.0929,
+      "step": 1090
+    },
+    {
+      "epoch": 4.425316455696202,
+      "grad_norm": 0.3455672892317423,
+      "learning_rate": 1.5360137861084656e-06,
+      "loss": 0.0873,
+      "step": 1091
+    },
+    {
+      "epoch": 4.429367088607595,
+      "grad_norm": 0.33852498702354755,
+      "learning_rate": 1.5142745882488475e-06,
+      "loss": 0.0818,
+      "step": 1092
+    },
+    {
+      "epoch": 4.433417721518987,
+      "grad_norm": 0.34684604206314334,
+      "learning_rate": 1.4926842718803691e-06,
+      "loss": 0.0861,
+      "step": 1093
+    },
+    {
+      "epoch": 4.43746835443038,
+      "grad_norm": 0.35086204013303657,
+      "learning_rate": 1.4712430108884657e-06,
+      "loss": 0.0976,
+      "step": 1094
+    },
+    {
+      "epoch": 4.441518987341772,
+      "grad_norm": 0.33651040070233573,
+      "learning_rate": 1.4499509779581078e-06,
+      "loss": 0.1007,
+      "step": 1095
+    },
+    {
+      "epoch": 4.445569620253164,
+      "grad_norm": 0.32586021434465545,
+      "learning_rate": 1.4288083445723988e-06,
+      "loss": 0.1018,
+      "step": 1096
+    },
+    {
+      "epoch": 4.449620253164557,
+      "grad_norm": 0.3241842694626627,
+      "learning_rate": 1.4078152810112045e-06,
+      "loss": 0.094,
+      "step": 1097
+    },
+    {
+      "epoch": 4.453670886075949,
+      "grad_norm": 0.3437928695410118,
+      "learning_rate": 1.3869719563497697e-06,
+      "loss": 0.0941,
+      "step": 1098
+    },
+    {
+      "epoch": 4.457721518987341,
+      "grad_norm": 0.32481217051219213,
+      "learning_rate": 1.3662785384573663e-06,
+      "loss": 0.1006,
+      "step": 1099
+    },
+    {
+      "epoch": 4.461772151898734,
+      "grad_norm": 0.32886868828872473,
+      "learning_rate": 1.3457351939959383e-06,
+      "loss": 0.0849,
+      "step": 1100
+    },
+    {
+      "epoch": 4.465822784810126,
+      "grad_norm": 0.3444734378255048,
+      "learning_rate": 1.3253420884187551e-06,
+      "loss": 0.1031,
+      "step": 1101
+    },
+    {
+      "epoch": 4.469873417721519,
+      "grad_norm": 0.3315804446472597,
+      "learning_rate": 1.3050993859690953e-06,
+      "loss": 0.1019,
+      "step": 1102
+    },
+    {
+      "epoch": 4.473924050632911,
+      "grad_norm": 0.3134078323513888,
+      "learning_rate": 1.2850072496788869e-06,
+      "loss": 0.0873,
+      "step": 1103
+    },
+    {
+      "epoch": 4.4779746835443035,
+      "grad_norm": 0.3391662297139124,
+      "learning_rate": 1.2650658413674434e-06,
+      "loss": 0.1012,
+      "step": 1104
+    },
+    {
+      "epoch": 4.4820253164556965,
+      "grad_norm": 0.3117156066966001,
+      "learning_rate": 1.2452753216401226e-06,
+      "loss": 0.0871,
+      "step": 1105
+    },
+    {
+      "epoch": 4.4860759493670885,
+      "grad_norm": 0.348922485788536,
+      "learning_rate": 1.2256358498870503e-06,
+      "loss": 0.0964,
+      "step": 1106
+    },
+    {
+      "epoch": 4.490126582278481,
+      "grad_norm": 0.31733814889399814,
+      "learning_rate": 1.2061475842818337e-06,
+      "loss": 0.0865,
+      "step": 1107
+    },
+    {
+      "epoch": 4.494177215189874,
+      "grad_norm": 0.31514686219641336,
+      "learning_rate": 1.1868106817802816e-06,
+      "loss": 0.084,
+      "step": 1108
+    },
+    {
+      "epoch": 4.498227848101266,
+      "grad_norm": 0.32940168797138775,
+      "learning_rate": 1.1676252981191482e-06,
+      "loss": 0.0873,
+      "step": 1109
+    },
+    {
+      "epoch": 4.502278481012659,
+      "grad_norm": 0.334314573396825,
+      "learning_rate": 1.1485915878148823e-06,
+      "loss": 0.1062,
+      "step": 1110
+    },
+    {
+      "epoch": 4.506329113924051,
+      "grad_norm": 0.3234243311837868,
+      "learning_rate": 1.1297097041623584e-06,
+      "loss": 0.0949,
+      "step": 1111
+    },
+    {
+      "epoch": 4.510379746835443,
+      "grad_norm": 0.3359962801478974,
+      "learning_rate": 1.1109797992336847e-06,
+      "loss": 0.0809,
+      "step": 1112
+    },
+    {
+      "epoch": 4.514430379746836,
+      "grad_norm": 0.3208459110279401,
+      "learning_rate": 1.092402023876933e-06,
+      "loss": 0.093,
+      "step": 1113
+    },
+    {
+      "epoch": 4.518481012658228,
+      "grad_norm": 0.32566058872378106,
+      "learning_rate": 1.0739765277149527e-06,
+      "loss": 0.0813,
+      "step": 1114
+    },
+    {
+      "epoch": 4.52253164556962,
+      "grad_norm": 0.32951616242192716,
+      "learning_rate": 1.0557034591441596e-06,
+      "loss": 0.0881,
+      "step": 1115
+    },
+    {
+      "epoch": 4.526582278481013,
+      "grad_norm": 0.33071298840834373,
+      "learning_rate": 1.0375829653333324e-06,
+      "loss": 0.0951,
+      "step": 1116
+    },
+    {
+      "epoch": 4.530632911392405,
+      "grad_norm": 0.3434640918871925,
+      "learning_rate": 1.0196151922224385e-06,
+      "loss": 0.1004,
+      "step": 1117
+    },
+    {
+      "epoch": 4.534683544303798,
+      "grad_norm": 0.3388733787519661,
+      "learning_rate": 1.0018002845214526e-06,
+      "loss": 0.1041,
+      "step": 1118
+    },
+    {
+      "epoch": 4.53873417721519,
+      "grad_norm": 0.3218282024645901,
+      "learning_rate": 9.841383857091947e-07,
+      "loss": 0.082,
+      "step": 1119
+    },
+    {
+      "epoch": 4.542784810126582,
+      "grad_norm": 0.33886281186657385,
+      "learning_rate": 9.666296380321616e-07,
+      "loss": 0.1148,
+      "step": 1120
+    },
+    {
+      "epoch": 4.546835443037975,
+      "grad_norm": 0.34794977485326584,
+      "learning_rate": 9.492741825034124e-07,
+      "loss": 0.0811,
+      "step": 1121
+    },
+    {
+      "epoch": 4.550886075949367,
+      "grad_norm": 0.3202442303145058,
+      "learning_rate": 9.320721589013892e-07,
+      "loss": 0.0971,
+      "step": 1122
+    },
+    {
+      "epoch": 4.55493670886076,
+      "grad_norm": 0.32611060259451086,
+      "learning_rate": 9.150237057688339e-07,
+      "loss": 0.0834,
+      "step": 1123
+    },
+    {
+      "epoch": 4.558987341772152,
+      "grad_norm": 0.3339594431688712,
+      "learning_rate": 8.981289604116328e-07,
+      "loss": 0.103,
+      "step": 1124
+    },
+    {
+      "epoch": 4.563037974683544,
+      "grad_norm": 0.34216158239334715,
+      "learning_rate": 8.813880588977542e-07,
+      "loss": 0.0983,
+      "step": 1125
+    },
+    {
+      "epoch": 4.567088607594937,
+      "grad_norm": 0.3408880849330529,
+      "learning_rate": 8.648011360561126e-07,
+      "loss": 0.0949,
+      "step": 1126
+    },
+    {
+      "epoch": 4.571139240506329,
+      "grad_norm": 0.3221304824184791,
+      "learning_rate": 8.483683254755037e-07,
+      "loss": 0.0946,
+      "step": 1127
+    },
+    {
+      "epoch": 4.575189873417721,
+      "grad_norm": 0.35567589804936495,
+      "learning_rate": 8.320897595035227e-07,
+      "loss": 0.1046,
+      "step": 1128
+    },
+    {
+      "epoch": 4.579240506329114,
+      "grad_norm": 0.3309841716298761,
+      "learning_rate": 8.159655692455093e-07,
+      "loss": 0.0909,
+      "step": 1129
+    },
+    {
+      "epoch": 4.583291139240506,
+      "grad_norm": 0.3336583749522916,
+      "learning_rate": 7.999958845634648e-07,
+      "loss": 0.0882,
+      "step": 1130
+    },
+    {
+      "epoch": 4.587341772151898,
+      "grad_norm": 0.332668304405888,
+      "learning_rate": 7.841808340750478e-07,
+      "loss": 0.098,
+      "step": 1131
+    },
+    {
+      "epoch": 4.591392405063291,
+      "grad_norm": 0.3338701431514458,
+      "learning_rate": 7.685205451524869e-07,
+      "loss": 0.0963,
+      "step": 1132
+    },
+    {
+      "epoch": 4.595443037974683,
+      "grad_norm": 0.31320254657305957,
+      "learning_rate": 7.530151439216027e-07,
+      "loss": 0.0842,
+      "step": 1133
+    },
+    {
+      "epoch": 4.599493670886076,
+      "grad_norm": 0.34273683349645306,
+      "learning_rate": 7.376647552607675e-07,
+      "loss": 0.0852,
+      "step": 1134
+    },
+    {
+      "epoch": 4.603544303797468,
+      "grad_norm": 0.3292167633350391,
+      "learning_rate": 7.224695027998963e-07,
+      "loss": 0.0999,
+      "step": 1135
+    },
+    {
+      "epoch": 4.6075949367088604,
+      "grad_norm": 0.33580065177470975,
+      "learning_rate": 7.07429508919466e-07,
+      "loss": 0.0883,
+      "step": 1136
+    },
+    {
+      "epoch": 4.611645569620253,
+      "grad_norm": 0.3395146073895252,
+      "learning_rate": 6.925448947495206e-07,
+      "loss": 0.1085,
+      "step": 1137
+    },
+    {
+      "epoch": 4.6156962025316455,
+      "grad_norm": 0.3314542155009118,
+      "learning_rate": 6.778157801686936e-07,
+      "loss": 0.09,
+      "step": 1138
+    },
+    {
+      "epoch": 4.619746835443038,
+      "grad_norm": 0.33860737396352086,
+      "learning_rate": 6.632422838032515e-07,
+      "loss": 0.1127,
+      "step": 1139
+    },
+    {
+      "epoch": 4.6237974683544305,
+      "grad_norm": 0.3226039073521735,
+      "learning_rate": 6.488245230261281e-07,
+      "loss": 0.086,
+      "step": 1140
+    },
+    {
+      "epoch": 4.627848101265823,
+      "grad_norm": 0.3404358957685745,
+      "learning_rate": 6.345626139559868e-07,
+      "loss": 0.0989,
+      "step": 1141
+    },
+    {
+      "epoch": 4.6318987341772155,
+      "grad_norm": 0.3425688459295036,
+      "learning_rate": 6.204566714562866e-07,
+      "loss": 0.0824,
+      "step": 1142
+    },
+    {
+      "epoch": 4.635949367088608,
+      "grad_norm": 0.32784667061380024,
+      "learning_rate": 6.06506809134344e-07,
+      "loss": 0.0974,
+      "step": 1143
+    },
+    {
+      "epoch": 4.64,
+      "grad_norm": 0.33605713880439797,
+      "learning_rate": 5.927131393404373e-07,
+      "loss": 0.0941,
+      "step": 1144
+    },
+    {
+      "epoch": 4.644050632911393,
+      "grad_norm": 0.3305098023210367,
+      "learning_rate": 5.790757731668817e-07,
+      "loss": 0.1035,
+      "step": 1145
+    },
+    {
+      "epoch": 4.648101265822785,
+      "grad_norm": 0.32933403459516786,
+      "learning_rate": 5.655948204471507e-07,
+      "loss": 0.0967,
+      "step": 1146
+    },
+    {
+      "epoch": 4.652151898734177,
+      "grad_norm": 0.33728998922382725,
+      "learning_rate": 5.522703897549875e-07,
+      "loss": 0.107,
+      "step": 1147
+    },
+    {
+      "epoch": 4.65620253164557,
+      "grad_norm": 0.31591539744136254,
+      "learning_rate": 5.391025884035239e-07,
+      "loss": 0.0781,
+      "step": 1148
+    },
+    {
+      "epoch": 4.660253164556962,
+      "grad_norm": 0.33545163815683293,
+      "learning_rate": 5.260915224444207e-07,
+      "loss": 0.0902,
+      "step": 1149
+    },
+    {
+      "epoch": 4.664303797468355,
+      "grad_norm": 0.341794243591244,
+      "learning_rate": 5.132372966670129e-07,
+      "loss": 0.0957,
+      "step": 1150
+    },
+    {
+      "epoch": 4.668354430379747,
+      "grad_norm": 0.33209441618582763,
+      "learning_rate": 5.005400145974704e-07,
+      "loss": 0.0869,
+      "step": 1151
+    },
+    {
+      "epoch": 4.672405063291139,
+      "grad_norm": 0.34086256269711546,
+      "learning_rate": 4.879997784979562e-07,
+      "loss": 0.0913,
+      "step": 1152
+    },
+    {
+      "epoch": 4.676455696202532,
+      "grad_norm": 0.33919509924400015,
+      "learning_rate": 4.7561668936580984e-07,
+      "loss": 0.1004,
+      "step": 1153
+    },
+    {
+      "epoch": 4.680506329113924,
+      "grad_norm": 0.33210768536503654,
+      "learning_rate": 4.6339084693272306e-07,
+      "loss": 0.1017,
+      "step": 1154
+    },
+    {
+      "epoch": 4.684556962025317,
+      "grad_norm": 0.32963893108893566,
+      "learning_rate": 4.5132234966395847e-07,
+      "loss": 0.106,
+      "step": 1155
+    },
+    {
+      "epoch": 4.688607594936709,
+      "grad_norm": 0.34479180599502884,
+      "learning_rate": 4.3941129475752795e-07,
+      "loss": 0.1023,
+      "step": 1156
+    },
+    {
+      "epoch": 4.692658227848101,
+      "grad_norm": 0.351460995245885,
+      "learning_rate": 4.27657778143431e-07,
+      "loss": 0.0991,
+      "step": 1157
+    },
+    {
+      "epoch": 4.696708860759494,
+      "grad_norm": 0.3336743244584053,
+      "learning_rate": 4.1606189448287757e-07,
+      "loss": 0.0849,
+      "step": 1158
+    },
+    {
+      "epoch": 4.700759493670886,
+      "grad_norm": 0.3263468117367759,
+      "learning_rate": 4.046237371675177e-07,
+      "loss": 0.086,
+      "step": 1159
+    },
+    {
+      "epoch": 4.704810126582278,
+      "grad_norm": 0.33666230597967084,
+      "learning_rate": 3.9334339831869963e-07,
+      "loss": 0.0925,
+      "step": 1160
+    },
+    {
+      "epoch": 4.708860759493671,
+      "grad_norm": 0.3284306251957539,
+      "learning_rate": 3.8222096878671955e-07,
+      "loss": 0.0871,
+      "step": 1161
+    },
+    {
+      "epoch": 4.712911392405063,
+      "grad_norm": 0.32301256205536116,
+      "learning_rate": 3.7125653815009545e-07,
+      "loss": 0.0846,
+      "step": 1162
+    },
+    {
+      "epoch": 4.716962025316455,
+      "grad_norm": 0.3351224823567399,
+      "learning_rate": 3.6045019471484974e-07,
+      "loss": 0.1071,
+      "step": 1163
+    },
+    {
+      "epoch": 4.721012658227848,
+      "grad_norm": 0.33860923005826593,
+      "learning_rate": 3.498020255137813e-07,
+      "loss": 0.104,
+      "step": 1164
+    },
+    {
+      "epoch": 4.72506329113924,
+      "grad_norm": 0.33744926167770223,
+      "learning_rate": 3.393121163057811e-07,
+      "loss": 0.1071,
+      "step": 1165
+    },
+    {
+      "epoch": 4.729113924050633,
+      "grad_norm": 0.3287451521187486,
+      "learning_rate": 3.289805515751399e-07,
+      "loss": 0.096,
+      "step": 1166
+    },
+    {
+      "epoch": 4.733164556962025,
+      "grad_norm": 0.3382383074936323,
+      "learning_rate": 3.188074145308573e-07,
+      "loss": 0.1028,
+      "step": 1167
+    },
+    {
+      "epoch": 4.737215189873417,
+      "grad_norm": 0.3256678231857013,
+      "learning_rate": 3.087927871059804e-07,
+      "loss": 0.0796,
+      "step": 1168
+    },
+    {
+      "epoch": 4.74126582278481,
+      "grad_norm": 0.34202882369148085,
+      "learning_rate": 2.989367499569418e-07,
+      "loss": 0.0884,
+      "step": 1169
+    },
+    {
+      "epoch": 4.745316455696202,
+      "grad_norm": 0.3368144543878746,
+      "learning_rate": 2.8923938246290917e-07,
+      "loss": 0.1142,
+      "step": 1170
+    },
+    {
+      "epoch": 4.749367088607595,
+      "grad_norm": 0.32431702292142545,
+      "learning_rate": 2.7970076272514804e-07,
+      "loss": 0.0915,
+      "step": 1171
+    },
+    {
+      "epoch": 4.7534177215189874,
+      "grad_norm": 0.31639639603422604,
+      "learning_rate": 2.703209675663887e-07,
+      "loss": 0.0933,
+      "step": 1172
+    },
+    {
+      "epoch": 4.7574683544303795,
+      "grad_norm": 0.3351232185120638,
+      "learning_rate": 2.6110007253021374e-07,
+      "loss": 0.0984,
+      "step": 1173
+    },
+    {
+      "epoch": 4.7615189873417725,
+      "grad_norm": 0.3409540418268868,
+      "learning_rate": 2.520381518804471e-07,
+      "loss": 0.0994,
+      "step": 1174
+    },
+    {
+      "epoch": 4.7655696202531646,
+      "grad_norm": 0.323882384979777,
+      "learning_rate": 2.4313527860054585e-07,
+      "loss": 0.0867,
+      "step": 1175
+    },
+    {
+      "epoch": 4.769620253164557,
+      "grad_norm": 0.33219328676187626,
+      "learning_rate": 2.343915243930317e-07,
+      "loss": 0.0993,
+      "step": 1176
+    },
+    {
+      "epoch": 4.77367088607595,
+      "grad_norm": 0.3213989180535203,
+      "learning_rate": 2.2580695967889367e-07,
+      "loss": 0.0899,
+      "step": 1177
+    },
+    {
+      "epoch": 4.777721518987342,
+      "grad_norm": 0.33029582700189,
+      "learning_rate": 2.1738165359704189e-07,
+      "loss": 0.1044,
+      "step": 1178
+    },
+    {
+      "epoch": 4.781772151898734,
+      "grad_norm": 0.3290389037566337,
+      "learning_rate": 2.0911567400373257e-07,
+      "loss": 0.081,
+      "step": 1179
+    },
+    {
+      "epoch": 4.785822784810127,
+      "grad_norm": 0.33287742667683157,
+      "learning_rate": 2.0100908747202607e-07,
+      "loss": 0.1082,
+      "step": 1180
+    },
+    {
+      "epoch": 4.789873417721519,
+      "grad_norm": 0.3225253076885149,
+      "learning_rate": 1.9306195929125638e-07,
+      "loss": 0.0864,
+      "step": 1181
+    },
+    {
+      "epoch": 4.793924050632912,
+      "grad_norm": 0.3255570609712025,
+      "learning_rate": 1.8527435346650247e-07,
+      "loss": 0.091,
+      "step": 1182
+    },
+    {
+      "epoch": 4.797974683544304,
+      "grad_norm": 0.34718379871651034,
+      "learning_rate": 1.7764633271807108e-07,
+      "loss": 0.0903,
+      "step": 1183
+    },
+    {
+      "epoch": 4.802025316455696,
+      "grad_norm": 0.32857781272036457,
+      "learning_rate": 1.7017795848099262e-07,
+      "loss": 0.0841,
+      "step": 1184
+    },
+    {
+      "epoch": 4.806075949367089,
+      "grad_norm": 0.3342851889557658,
+      "learning_rate": 1.6286929090452596e-07,
+      "loss": 0.1055,
+      "step": 1185
+    },
+    {
+      "epoch": 4.810126582278481,
+      "grad_norm": 0.3233290229297408,
+      "learning_rate": 1.557203888516745e-07,
+      "loss": 0.0856,
+      "step": 1186
+    },
+    {
+      "epoch": 4.814177215189874,
+      "grad_norm": 0.332274290109675,
+      "learning_rate": 1.487313098987131e-07,
+      "loss": 0.0822,
+      "step": 1187
+    },
+    {
+      "epoch": 4.818227848101266,
+      "grad_norm": 0.33380444436374496,
+      "learning_rate": 1.4190211033472402e-07,
+      "loss": 0.0905,
+      "step": 1188
+    },
+    {
+      "epoch": 4.822278481012658,
+      "grad_norm": 0.33769431320095494,
+      "learning_rate": 1.3523284516113955e-07,
+      "loss": 0.1121,
+      "step": 1189
+    },
+    {
+      "epoch": 4.826329113924051,
+      "grad_norm": 0.3298516301932957,
+      "learning_rate": 1.2872356809130682e-07,
+      "loss": 0.0975,
+      "step": 1190
+    },
+    {
+      "epoch": 4.830379746835443,
+      "grad_norm": 0.33281842040767484,
+      "learning_rate": 1.2237433155004807e-07,
+      "loss": 0.1037,
+      "step": 1191
+    },
+    {
+      "epoch": 4.834430379746835,
+      "grad_norm": 0.344179006730819,
+      "learning_rate": 1.1618518667323886e-07,
+      "loss": 0.12,
+      "step": 1192
+    },
+    {
+      "epoch": 4.838481012658228,
+      "grad_norm": 0.3272852496800166,
+      "learning_rate": 1.1015618330740385e-07,
+      "loss": 0.0953,
+      "step": 1193
+    },
+    {
+      "epoch": 4.84253164556962,
+      "grad_norm": 0.34421640708794504,
+      "learning_rate": 1.042873700093061e-07,
+      "loss": 0.106,
+      "step": 1194
+    },
+    {
+      "epoch": 4.846582278481012,
+      "grad_norm": 0.3338011301504633,
+      "learning_rate": 9.857879404556291e-08,
+      "loss": 0.1138,
+      "step": 1195
+    },
+    {
+      "epoch": 4.850632911392405,
+      "grad_norm": 0.323365963216708,
+      "learning_rate": 9.303050139225722e-08,
+      "loss": 0.089,
+      "step": 1196
+    },
+    {
+      "epoch": 4.854683544303797,
+      "grad_norm": 0.3219238217868187,
+      "learning_rate": 8.76425367345779e-08,
+      "loss": 0.0944,
+      "step": 1197
+    },
+    {
+      "epoch": 4.85873417721519,
+      "grad_norm": 0.32289593213388335,
+      "learning_rate": 8.241494346644897e-08,
+      "loss": 0.0883,
+      "step": 1198
+    },
+    {
+      "epoch": 4.862784810126582,
+      "grad_norm": 0.32339360637885484,
+      "learning_rate": 7.734776369019204e-08,
+      "loss": 0.0836,
+      "step": 1199
+    },
+    {
+      "epoch": 4.866835443037974,
+      "grad_norm": 0.3256445148170869,
+      "learning_rate": 7.244103821617332e-08,
+      "loss": 0.0999,
+      "step": 1200
+    },
+    {
+      "epoch": 4.870886075949367,
+      "grad_norm": 0.35272463373942803,
+      "learning_rate": 6.769480656248606e-08,
+      "loss": 0.1026,
+      "step": 1201
+    },
+    {
+      "epoch": 4.874936708860759,
+      "grad_norm": 0.35415527948759895,
+      "learning_rate": 6.310910695462635e-08,
+      "loss": 0.1166,
+      "step": 1202
+    },
+    {
+      "epoch": 4.878987341772152,
+      "grad_norm": 0.34211283697446165,
+      "learning_rate": 5.8683976325191185e-08,
+      "loss": 0.1044,
+      "step": 1203
+    },
+    {
+      "epoch": 4.883037974683544,
+      "grad_norm": 0.3352777191754061,
+      "learning_rate": 5.4419450313571984e-08,
+      "loss": 0.0863,
+      "step": 1204
+    },
+    {
+      "epoch": 4.8870886075949365,
+      "grad_norm": 0.34314853949588003,
+      "learning_rate": 5.031556326567488e-08,
+      "loss": 0.0865,
+      "step": 1205
+    },
+    {
+      "epoch": 4.891139240506329,
+      "grad_norm": 0.3410681033146302,
+      "learning_rate": 4.637234823364312e-08,
+      "loss": 0.1024,
+      "step": 1206
+    },
+    {
+      "epoch": 4.8951898734177215,
+      "grad_norm": 0.32746225307266047,
+      "learning_rate": 4.258983697558838e-08,
+      "loss": 0.0947,
+      "step": 1207
+    },
+    {
+      "epoch": 4.899240506329114,
+      "grad_norm": 0.31852748431188416,
+      "learning_rate": 3.896805995533548e-08,
+      "loss": 0.0895,
+      "step": 1208
+    },
+    {
+      "epoch": 4.9032911392405065,
+      "grad_norm": 0.3318794308133162,
+      "learning_rate": 3.550704634218028e-08,
+      "loss": 0.0989,
+      "step": 1209
+    },
+    {
+      "epoch": 4.907341772151899,
+      "grad_norm": 0.31564295132277326,
+      "learning_rate": 3.2206824010647676e-08,
+      "loss": 0.0911,
+      "step": 1210
+    },
+    {
+      "epoch": 4.911392405063291,
+      "grad_norm": 0.3363334597403827,
+      "learning_rate": 2.9067419540278476e-08,
+      "loss": 0.0947,
+      "step": 1211
+    },
+    {
+      "epoch": 4.915443037974684,
+      "grad_norm": 0.32852928856664537,
+      "learning_rate": 2.6088858215400638e-08,
+      "loss": 0.0901,
+      "step": 1212
+    },
+    {
+      "epoch": 4.919493670886076,
+      "grad_norm": 0.312018763715005,
+      "learning_rate": 2.3271164024940564e-08,
+      "loss": 0.0862,
+      "step": 1213
+    },
+    {
+      "epoch": 4.923544303797469,
+      "grad_norm": 0.34218314836750235,
+      "learning_rate": 2.061435966221881e-08,
+      "loss": 0.0886,
+      "step": 1214
+    },
+    {
+      "epoch": 4.927594936708861,
+      "grad_norm": 0.3497304105154166,
+      "learning_rate": 1.811846652477245e-08,
+      "loss": 0.1088,
+      "step": 1215
+    },
+    {
+      "epoch": 4.931645569620253,
+      "grad_norm": 0.32028079796213693,
+      "learning_rate": 1.5783504714184106e-08,
+      "loss": 0.0774,
+      "step": 1216
+    },
+    {
+      "epoch": 4.935696202531646,
+      "grad_norm": 0.32826248835520144,
+      "learning_rate": 1.360949303591097e-08,
+      "loss": 0.0848,
+      "step": 1217
+    },
+    {
+      "epoch": 4.939746835443038,
+      "grad_norm": 0.33200845692607633,
+      "learning_rate": 1.1596448999144916e-08,
+      "loss": 0.0974,
+      "step": 1218
+    },
+    {
+      "epoch": 4.943797468354431,
+      "grad_norm": 0.3144297201415711,
+      "learning_rate": 9.744388816668172e-09,
+      "loss": 0.0748,
+      "step": 1219
+    },
+    {
+      "epoch": 4.947848101265823,
+      "grad_norm": 0.33328866503387145,
+      "learning_rate": 8.05332740472009e-09,
+      "loss": 0.1145,
+      "step": 1220
+    },
+    {
+      "epoch": 4.951898734177215,
+      "grad_norm": 0.34343477337089134,
+      "learning_rate": 6.523278382872811e-09,
+      "loss": 0.1074,
+      "step": 1221
+    },
+    {
+      "epoch": 4.955949367088608,
+      "grad_norm": 0.3338856501884481,
+      "learning_rate": 5.15425407393133e-09,
+      "loss": 0.1165,
+      "step": 1222
+    },
+    {
+      "epoch": 4.96,
+      "grad_norm": 0.33346866617676013,
+      "learning_rate": 3.94626550383137e-09,
+      "loss": 0.0837,
+      "step": 1223
+    },
+    {
+      "epoch": 4.964050632911392,
+      "grad_norm": 0.31158056931853234,
+      "learning_rate": 2.899322401546112e-09,
+      "loss": 0.0838,
+      "step": 1224
+    },
+    {
+      "epoch": 4.968101265822785,
+      "grad_norm": 0.325850683427699,
+      "learning_rate": 2.013433199010706e-09,
+      "loss": 0.0919,
+      "step": 1225
+    },
+    {
+      "epoch": 4.972151898734177,
+      "grad_norm": 0.34259808039216744,
+      "learning_rate": 1.2886050310556563e-09,
+      "loss": 0.0973,
+      "step": 1226
+    },
+    {
+      "epoch": 4.976202531645569,
+      "grad_norm": 0.31660378437022424,
+      "learning_rate": 7.248437353468695e-10,
+      "loss": 0.0806,
+      "step": 1227
+    },
+    {
+      "epoch": 4.980253164556962,
+      "grad_norm": 0.33402545112363785,
+      "learning_rate": 3.221538523412449e-10,
+      "loss": 0.1035,
+      "step": 1228
+    },
+    {
+      "epoch": 4.984303797468354,
+      "grad_norm": 0.32626292472025725,
+      "learning_rate": 8.053862524670663e-11,
+      "loss": 0.0902,
+      "step": 1229
+    },
+    {
+      "epoch": 4.988354430379747,
+      "grad_norm": 0.3392134858258251,
+      "learning_rate": 0.0,
+      "loss": 0.0991,
+      "step": 1230
+    },
+    {
+      "epoch": 4.988354430379747,
+      "step": 1230,
+      "total_flos": 7.127260343566336e+17,
+      "train_loss": 0.3819762018213912,
+      "train_runtime": 23901.5525,
+      "train_samples_per_second": 6.61,
+      "train_steps_per_second": 0.051
+    }
+  ],
+  "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": 7.127260343566336e+17,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}