diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,4662 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.9977298524404086,
+  "eval_steps": 500,
+  "global_step": 660,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.003026863412788498,
+      "grad_norm": 2.8440141200803324,
+      "learning_rate": 3.0303030303030305e-07,
+      "loss": 0.6325,
+      "step": 1
+    },
+    {
+      "epoch": 0.006053726825576996,
+      "grad_norm": 2.8197101153579824,
+      "learning_rate": 6.060606060606061e-07,
+      "loss": 0.6482,
+      "step": 2
+    },
+    {
+      "epoch": 0.009080590238365494,
+      "grad_norm": 2.9497013049149947,
+      "learning_rate": 9.090909090909091e-07,
+      "loss": 0.6348,
+      "step": 3
+    },
+    {
+      "epoch": 0.012107453651153992,
+      "grad_norm": 2.63569786667829,
+      "learning_rate": 1.2121212121212122e-06,
+      "loss": 0.6223,
+      "step": 4
+    },
+    {
+      "epoch": 0.01513431706394249,
+      "grad_norm": 2.694784573108838,
+      "learning_rate": 1.5151515151515152e-06,
+      "loss": 0.6409,
+      "step": 5
+    },
+    {
+      "epoch": 0.018161180476730987,
+      "grad_norm": 2.494604106886768,
+      "learning_rate": 1.8181818181818183e-06,
+      "loss": 0.5927,
+      "step": 6
+    },
+    {
+      "epoch": 0.021188043889519486,
+      "grad_norm": 1.8209373201851167,
+      "learning_rate": 2.1212121212121216e-06,
+      "loss": 0.6106,
+      "step": 7
+    },
+    {
+      "epoch": 0.024214907302307985,
+      "grad_norm": 1.6568855210038849,
+      "learning_rate": 2.4242424242424244e-06,
+      "loss": 0.5891,
+      "step": 8
+    },
+    {
+      "epoch": 0.02724177071509648,
+      "grad_norm": 1.695508761777406,
+      "learning_rate": 2.7272727272727272e-06,
+      "loss": 0.5909,
+      "step": 9
+    },
+    {
+      "epoch": 0.03026863412788498,
+      "grad_norm": 1.6561921636696004,
+      "learning_rate": 3.0303030303030305e-06,
+      "loss": 0.5571,
+      "step": 10
+    },
+    {
+      "epoch": 0.03329549754067348,
+      "grad_norm": 1.7731535498181017,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 0.5695,
+      "step": 11
+    },
+    {
+      "epoch": 0.036322360953461974,
+      "grad_norm": 1.312548464396275,
+      "learning_rate": 3.6363636363636366e-06,
+      "loss": 0.5256,
+      "step": 12
+    },
+    {
+      "epoch": 0.03934922436625047,
+      "grad_norm": 1.4287433366329747,
+      "learning_rate": 3.93939393939394e-06,
+      "loss": 0.5212,
+      "step": 13
+    },
+    {
+      "epoch": 0.04237608777903897,
+      "grad_norm": 1.3724331750656265,
+      "learning_rate": 4.242424242424243e-06,
+      "loss": 0.5151,
+      "step": 14
+    },
+    {
+      "epoch": 0.04540295119182747,
+      "grad_norm": 1.242390321458998,
+      "learning_rate": 4.5454545454545455e-06,
+      "loss": 0.5093,
+      "step": 15
+    },
+    {
+      "epoch": 0.04842981460461597,
+      "grad_norm": 1.026970784992291,
+      "learning_rate": 4.848484848484849e-06,
+      "loss": 0.5078,
+      "step": 16
+    },
+    {
+      "epoch": 0.051456678017404466,
+      "grad_norm": 1.07571958425287,
+      "learning_rate": 5.151515151515152e-06,
+      "loss": 0.497,
+      "step": 17
+    },
+    {
+      "epoch": 0.05448354143019296,
+      "grad_norm": 1.0952624155315898,
+      "learning_rate": 5.4545454545454545e-06,
+      "loss": 0.4804,
+      "step": 18
+    },
+    {
+      "epoch": 0.057510404842981463,
+      "grad_norm": 1.1037842980829813,
+      "learning_rate": 5.7575757575757586e-06,
+      "loss": 0.4815,
+      "step": 19
+    },
+    {
+      "epoch": 0.06053726825576996,
+      "grad_norm": 0.9883731147115445,
+      "learning_rate": 6.060606060606061e-06,
+      "loss": 0.4862,
+      "step": 20
+    },
+    {
+      "epoch": 0.06356413166855845,
+      "grad_norm": 0.8371892867266533,
+      "learning_rate": 6.363636363636364e-06,
+      "loss": 0.4358,
+      "step": 21
+    },
+    {
+      "epoch": 0.06659099508134696,
+      "grad_norm": 0.8675072616622022,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 0.4515,
+      "step": 22
+    },
+    {
+      "epoch": 0.06961785849413545,
+      "grad_norm": 0.9126482594724041,
+      "learning_rate": 6.969696969696971e-06,
+      "loss": 0.4653,
+      "step": 23
+    },
+    {
+      "epoch": 0.07264472190692395,
+      "grad_norm": 0.8421757107885606,
+      "learning_rate": 7.272727272727273e-06,
+      "loss": 0.4515,
+      "step": 24
+    },
+    {
+      "epoch": 0.07567158531971245,
+      "grad_norm": 0.8603747512522966,
+      "learning_rate": 7.5757575757575764e-06,
+      "loss": 0.4554,
+      "step": 25
+    },
+    {
+      "epoch": 0.07869844873250094,
+      "grad_norm": 0.7906903714616359,
+      "learning_rate": 7.87878787878788e-06,
+      "loss": 0.4553,
+      "step": 26
+    },
+    {
+      "epoch": 0.08172531214528944,
+      "grad_norm": 0.7977972982910061,
+      "learning_rate": 8.181818181818183e-06,
+      "loss": 0.4537,
+      "step": 27
+    },
+    {
+      "epoch": 0.08475217555807794,
+      "grad_norm": 0.7516390456913151,
+      "learning_rate": 8.484848484848486e-06,
+      "loss": 0.4176,
+      "step": 28
+    },
+    {
+      "epoch": 0.08777903897086645,
+      "grad_norm": 0.7606592759890292,
+      "learning_rate": 8.787878787878788e-06,
+      "loss": 0.4189,
+      "step": 29
+    },
+    {
+      "epoch": 0.09080590238365494,
+      "grad_norm": 0.7828097878660688,
+      "learning_rate": 9.090909090909091e-06,
+      "loss": 0.4369,
+      "step": 30
+    },
+    {
+      "epoch": 0.09383276579644344,
+      "grad_norm": 0.7322577325586848,
+      "learning_rate": 9.393939393939396e-06,
+      "loss": 0.4183,
+      "step": 31
+    },
+    {
+      "epoch": 0.09685962920923194,
+      "grad_norm": 0.7445364717404914,
+      "learning_rate": 9.696969696969698e-06,
+      "loss": 0.4215,
+      "step": 32
+    },
+    {
+      "epoch": 0.09988649262202043,
+      "grad_norm": 0.7664311387693546,
+      "learning_rate": 1e-05,
+      "loss": 0.4356,
+      "step": 33
+    },
+    {
+      "epoch": 0.10291335603480893,
+      "grad_norm": 0.7845333759717925,
+      "learning_rate": 1.0303030303030304e-05,
+      "loss": 0.4287,
+      "step": 34
+    },
+    {
+      "epoch": 0.10594021944759743,
+      "grad_norm": 0.7993059806394502,
+      "learning_rate": 1.0606060606060606e-05,
+      "loss": 0.4279,
+      "step": 35
+    },
+    {
+      "epoch": 0.10896708286038592,
+      "grad_norm": 0.7609561272234476,
+      "learning_rate": 1.0909090909090909e-05,
+      "loss": 0.4275,
+      "step": 36
+    },
+    {
+      "epoch": 0.11199394627317442,
+      "grad_norm": 0.7841627250943042,
+      "learning_rate": 1.1212121212121212e-05,
+      "loss": 0.4431,
+      "step": 37
+    },
+    {
+      "epoch": 0.11502080968596293,
+      "grad_norm": 0.7608483797195187,
+      "learning_rate": 1.1515151515151517e-05,
+      "loss": 0.4152,
+      "step": 38
+    },
+    {
+      "epoch": 0.11804767309875142,
+      "grad_norm": 0.7649533027267864,
+      "learning_rate": 1.181818181818182e-05,
+      "loss": 0.4081,
+      "step": 39
+    },
+    {
+      "epoch": 0.12107453651153992,
+      "grad_norm": 0.7430996120799133,
+      "learning_rate": 1.2121212121212122e-05,
+      "loss": 0.4136,
+      "step": 40
+    },
+    {
+      "epoch": 0.12410139992432842,
+      "grad_norm": 0.7656884678646703,
+      "learning_rate": 1.2424242424242425e-05,
+      "loss": 0.4206,
+      "step": 41
+    },
+    {
+      "epoch": 0.1271282633371169,
+      "grad_norm": 0.7746462433810486,
+      "learning_rate": 1.2727272727272728e-05,
+      "loss": 0.3929,
+      "step": 42
+    },
+    {
+      "epoch": 0.1301551267499054,
+      "grad_norm": 0.7734022810425616,
+      "learning_rate": 1.3030303030303032e-05,
+      "loss": 0.4077,
+      "step": 43
+    },
+    {
+      "epoch": 0.13318199016269391,
+      "grad_norm": 0.8075872959996047,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.4299,
+      "step": 44
+    },
+    {
+      "epoch": 0.1362088535754824,
+      "grad_norm": 0.7719129828942466,
+      "learning_rate": 1.3636363636363637e-05,
+      "loss": 0.4025,
+      "step": 45
+    },
+    {
+      "epoch": 0.1392357169882709,
+      "grad_norm": 0.7909230272587213,
+      "learning_rate": 1.3939393939393942e-05,
+      "loss": 0.4168,
+      "step": 46
+    },
+    {
+      "epoch": 0.1422625804010594,
+      "grad_norm": 0.7429143017725387,
+      "learning_rate": 1.4242424242424245e-05,
+      "loss": 0.3759,
+      "step": 47
+    },
+    {
+      "epoch": 0.1452894438138479,
+      "grad_norm": 0.7057883797163559,
+      "learning_rate": 1.4545454545454546e-05,
+      "loss": 0.3989,
+      "step": 48
+    },
+    {
+      "epoch": 0.14831630722663638,
+      "grad_norm": 0.768347342916585,
+      "learning_rate": 1.484848484848485e-05,
+      "loss": 0.4037,
+      "step": 49
+    },
+    {
+      "epoch": 0.1513431706394249,
+      "grad_norm": 0.7592054197827858,
+      "learning_rate": 1.5151515151515153e-05,
+      "loss": 0.3966,
+      "step": 50
+    },
+    {
+      "epoch": 0.1543700340522134,
+      "grad_norm": 0.7281269982340662,
+      "learning_rate": 1.5454545454545454e-05,
+      "loss": 0.3878,
+      "step": 51
+    },
+    {
+      "epoch": 0.15739689746500188,
+      "grad_norm": 0.8025911686752701,
+      "learning_rate": 1.575757575757576e-05,
+      "loss": 0.3798,
+      "step": 52
+    },
+    {
+      "epoch": 0.1604237608777904,
+      "grad_norm": 0.6740681847134368,
+      "learning_rate": 1.606060606060606e-05,
+      "loss": 0.3588,
+      "step": 53
+    },
+    {
+      "epoch": 0.16345062429057888,
+      "grad_norm": 0.8246093622607957,
+      "learning_rate": 1.6363636363636366e-05,
+      "loss": 0.3949,
+      "step": 54
+    },
+    {
+      "epoch": 0.16647748770336737,
+      "grad_norm": 0.7707492554897742,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 0.3997,
+      "step": 55
+    },
+    {
+      "epoch": 0.1695043511161559,
+      "grad_norm": 0.8351860678788088,
+      "learning_rate": 1.6969696969696972e-05,
+      "loss": 0.3942,
+      "step": 56
+    },
+    {
+      "epoch": 0.17253121452894438,
+      "grad_norm": 0.7329142118770092,
+      "learning_rate": 1.7272727272727274e-05,
+      "loss": 0.3842,
+      "step": 57
+    },
+    {
+      "epoch": 0.1755580779417329,
+      "grad_norm": 0.7655234817416188,
+      "learning_rate": 1.7575757575757576e-05,
+      "loss": 0.3938,
+      "step": 58
+    },
+    {
+      "epoch": 0.17858494135452138,
+      "grad_norm": 0.731419503039697,
+      "learning_rate": 1.787878787878788e-05,
+      "loss": 0.3883,
+      "step": 59
+    },
+    {
+      "epoch": 0.18161180476730987,
+      "grad_norm": 0.7273059370079011,
+      "learning_rate": 1.8181818181818182e-05,
+      "loss": 0.3733,
+      "step": 60
+    },
+    {
+      "epoch": 0.1846386681800984,
+      "grad_norm": 0.7394189814297651,
+      "learning_rate": 1.8484848484848487e-05,
+      "loss": 0.3768,
+      "step": 61
+    },
+    {
+      "epoch": 0.18766553159288688,
+      "grad_norm": 0.7899342586993907,
+      "learning_rate": 1.8787878787878792e-05,
+      "loss": 0.3854,
+      "step": 62
+    },
+    {
+      "epoch": 0.19069239500567536,
+      "grad_norm": 1.136797663433098,
+      "learning_rate": 1.9090909090909094e-05,
+      "loss": 0.3882,
+      "step": 63
+    },
+    {
+      "epoch": 0.19371925841846388,
+      "grad_norm": 0.9424071936052238,
+      "learning_rate": 1.9393939393939395e-05,
+      "loss": 0.399,
+      "step": 64
+    },
+    {
+      "epoch": 0.19674612183125237,
+      "grad_norm": 3.7712667189550984,
+      "learning_rate": 1.96969696969697e-05,
+      "loss": 0.3826,
+      "step": 65
+    },
+    {
+      "epoch": 0.19977298524404086,
+      "grad_norm": 1.350922478532856,
+      "learning_rate": 2e-05,
+      "loss": 0.4104,
+      "step": 66
+    },
+    {
+      "epoch": 0.20279984865682937,
+      "grad_norm": 0.9432201708837927,
+      "learning_rate": 1.9999860139251737e-05,
+      "loss": 0.387,
+      "step": 67
+    },
+    {
+      "epoch": 0.20582671206961786,
+      "grad_norm": 0.9541254938496068,
+      "learning_rate": 1.9999440560919153e-05,
+      "loss": 0.3889,
+      "step": 68
+    },
+    {
+      "epoch": 0.20885357548240635,
+      "grad_norm": 0.8754189705653768,
+      "learning_rate": 1.9998741276738753e-05,
+      "loss": 0.3737,
+      "step": 69
+    },
+    {
+      "epoch": 0.21188043889519487,
+      "grad_norm": 0.8424129302523434,
+      "learning_rate": 1.999776230627102e-05,
+      "loss": 0.3722,
+      "step": 70
+    },
+    {
+      "epoch": 0.21490730230798336,
+      "grad_norm": 1.0295840592542231,
+      "learning_rate": 1.9996503676899863e-05,
+      "loss": 0.3777,
+      "step": 71
+    },
+    {
+      "epoch": 0.21793416572077184,
+      "grad_norm": 0.8508980138593863,
+      "learning_rate": 1.9994965423831853e-05,
+      "loss": 0.3926,
+      "step": 72
+    },
+    {
+      "epoch": 0.22096102913356036,
+      "grad_norm": 0.8397928821166675,
+      "learning_rate": 1.9993147590095232e-05,
+      "loss": 0.3833,
+      "step": 73
+    },
+    {
+      "epoch": 0.22398789254634885,
+      "grad_norm": 0.937076924582728,
+      "learning_rate": 1.999105022653872e-05,
+      "loss": 0.3795,
+      "step": 74
+    },
+    {
+      "epoch": 0.22701475595913734,
+      "grad_norm": 0.7942863082661523,
+      "learning_rate": 1.9988673391830082e-05,
+      "loss": 0.3944,
+      "step": 75
+    },
+    {
+      "epoch": 0.23004161937192585,
+      "grad_norm": 0.8546848500850027,
+      "learning_rate": 1.9986017152454497e-05,
+      "loss": 0.3655,
+      "step": 76
+    },
+    {
+      "epoch": 0.23306848278471434,
+      "grad_norm": 0.8673680144635848,
+      "learning_rate": 1.9983081582712684e-05,
+      "loss": 0.407,
+      "step": 77
+    },
+    {
+      "epoch": 0.23609534619750283,
+      "grad_norm": 0.8022083921918615,
+      "learning_rate": 1.9979866764718846e-05,
+      "loss": 0.3825,
+      "step": 78
+    },
+    {
+      "epoch": 0.23912220961029135,
+      "grad_norm": 0.7744682596791275,
+      "learning_rate": 1.997637278839835e-05,
+      "loss": 0.357,
+      "step": 79
+    },
+    {
+      "epoch": 0.24214907302307984,
+      "grad_norm": 0.7642616278292101,
+      "learning_rate": 1.9972599751485225e-05,
+      "loss": 0.3679,
+      "step": 80
+    },
+    {
+      "epoch": 0.24517593643586832,
+      "grad_norm": 0.7775339154628217,
+      "learning_rate": 1.9968547759519426e-05,
+      "loss": 0.3964,
+      "step": 81
+    },
+    {
+      "epoch": 0.24820279984865684,
+      "grad_norm": 0.8246839522111197,
+      "learning_rate": 1.9964216925843876e-05,
+      "loss": 0.3786,
+      "step": 82
+    },
+    {
+      "epoch": 0.2512296632614453,
+      "grad_norm": 0.7200279581420893,
+      "learning_rate": 1.9959607371601303e-05,
+      "loss": 0.3851,
+      "step": 83
+    },
+    {
+      "epoch": 0.2542565266742338,
+      "grad_norm": 0.9024237141869264,
+      "learning_rate": 1.9954719225730847e-05,
+      "loss": 0.3694,
+      "step": 84
+    },
+    {
+      "epoch": 0.25728339008702233,
+      "grad_norm": 0.7149360298554913,
+      "learning_rate": 1.994955262496446e-05,
+      "loss": 0.3663,
+      "step": 85
+    },
+    {
+      "epoch": 0.2603102534998108,
+      "grad_norm": 0.7327618434035029,
+      "learning_rate": 1.9944107713823068e-05,
+      "loss": 0.3667,
+      "step": 86
+    },
+    {
+      "epoch": 0.2633371169125993,
+      "grad_norm": 0.8033923212091437,
+      "learning_rate": 1.9938384644612542e-05,
+      "loss": 0.3876,
+      "step": 87
+    },
+    {
+      "epoch": 0.26636398032538783,
+      "grad_norm": 0.6949084182967816,
+      "learning_rate": 1.9932383577419432e-05,
+      "loss": 0.3892,
+      "step": 88
+    },
+    {
+      "epoch": 0.2693908437381763,
+      "grad_norm": 0.7152833496550212,
+      "learning_rate": 1.9926104680106484e-05,
+      "loss": 0.3736,
+      "step": 89
+    },
+    {
+      "epoch": 0.2724177071509648,
+      "grad_norm": 0.6900281323660108,
+      "learning_rate": 1.9919548128307954e-05,
+      "loss": 0.3855,
+      "step": 90
+    },
+    {
+      "epoch": 0.2754445705637533,
+      "grad_norm": 0.6626391070223204,
+      "learning_rate": 1.9912714105424694e-05,
+      "loss": 0.3577,
+      "step": 91
+    },
+    {
+      "epoch": 0.2784714339765418,
+      "grad_norm": 0.7536070183579019,
+      "learning_rate": 1.990560280261901e-05,
+      "loss": 0.3918,
+      "step": 92
+    },
+    {
+      "epoch": 0.2814982973893303,
+      "grad_norm": 0.7726096391248967,
+      "learning_rate": 1.989821441880933e-05,
+      "loss": 0.3734,
+      "step": 93
+    },
+    {
+      "epoch": 0.2845251608021188,
+      "grad_norm": 0.7208965613609746,
+      "learning_rate": 1.9890549160664633e-05,
+      "loss": 0.3805,
+      "step": 94
+    },
+    {
+      "epoch": 0.2875520242149073,
+      "grad_norm": 0.6751682688556743,
+      "learning_rate": 1.9882607242598663e-05,
+      "loss": 0.3739,
+      "step": 95
+    },
+    {
+      "epoch": 0.2905788876276958,
+      "grad_norm": 0.7424875342223862,
+      "learning_rate": 1.9874388886763944e-05,
+      "loss": 0.3738,
+      "step": 96
+    },
+    {
+      "epoch": 0.2936057510404843,
+      "grad_norm": 0.7586039907595037,
+      "learning_rate": 1.9865894323045558e-05,
+      "loss": 0.3587,
+      "step": 97
+    },
+    {
+      "epoch": 0.29663261445327277,
+      "grad_norm": 0.685527220168254,
+      "learning_rate": 1.9857123789054707e-05,
+      "loss": 0.3749,
+      "step": 98
+    },
+    {
+      "epoch": 0.2996594778660613,
+      "grad_norm": 0.7687791133048779,
+      "learning_rate": 1.9848077530122083e-05,
+      "loss": 0.3808,
+      "step": 99
+    },
+    {
+      "epoch": 0.3026863412788498,
+      "grad_norm": 0.7486489976287071,
+      "learning_rate": 1.9838755799290993e-05,
+      "loss": 0.387,
+      "step": 100
+    },
+    {
+      "epoch": 0.30571320469163826,
+      "grad_norm": 0.7403020404834152,
+      "learning_rate": 1.9829158857310288e-05,
+      "loss": 0.3916,
+      "step": 101
+    },
+    {
+      "epoch": 0.3087400681044268,
+      "grad_norm": 0.7839784162225644,
+      "learning_rate": 1.9819286972627066e-05,
+      "loss": 0.3826,
+      "step": 102
+    },
+    {
+      "epoch": 0.3117669315172153,
+      "grad_norm": 0.7500889107030478,
+      "learning_rate": 1.9809140421379168e-05,
+      "loss": 0.3873,
+      "step": 103
+    },
+    {
+      "epoch": 0.31479379493000376,
+      "grad_norm": 0.7139038248772349,
+      "learning_rate": 1.979871948738743e-05,
+      "loss": 0.3714,
+      "step": 104
+    },
+    {
+      "epoch": 0.3178206583427923,
+      "grad_norm": 0.7850572232770251,
+      "learning_rate": 1.978802446214779e-05,
+      "loss": 0.3671,
+      "step": 105
+    },
+    {
+      "epoch": 0.3208475217555808,
+      "grad_norm": 0.7492434831159727,
+      "learning_rate": 1.9777055644823087e-05,
+      "loss": 0.3729,
+      "step": 106
+    },
+    {
+      "epoch": 0.32387438516836925,
+      "grad_norm": 0.7526520291494687,
+      "learning_rate": 1.9765813342234726e-05,
+      "loss": 0.357,
+      "step": 107
+    },
+    {
+      "epoch": 0.32690124858115777,
+      "grad_norm": 0.7084835662113623,
+      "learning_rate": 1.9754297868854075e-05,
+      "loss": 0.3716,
+      "step": 108
+    },
+    {
+      "epoch": 0.3299281119939463,
+      "grad_norm": 0.7288411152674839,
+      "learning_rate": 1.9742509546793673e-05,
+      "loss": 0.3739,
+      "step": 109
+    },
+    {
+      "epoch": 0.33295497540673474,
+      "grad_norm": 0.7107541259553729,
+      "learning_rate": 1.973044870579824e-05,
+      "loss": 0.3574,
+      "step": 110
+    },
+    {
+      "epoch": 0.33598183881952326,
+      "grad_norm": 0.6754595805121915,
+      "learning_rate": 1.9718115683235418e-05,
+      "loss": 0.3452,
+      "step": 111
+    },
+    {
+      "epoch": 0.3390087022323118,
+      "grad_norm": 1.00835787615974,
+      "learning_rate": 1.970551082408636e-05,
+      "loss": 0.3513,
+      "step": 112
+    },
+    {
+      "epoch": 0.3420355656451003,
+      "grad_norm": 0.7635308216754979,
+      "learning_rate": 1.969263448093608e-05,
+      "loss": 0.3584,
+      "step": 113
+    },
+    {
+      "epoch": 0.34506242905788875,
+      "grad_norm": 0.7687080953265656,
+      "learning_rate": 1.9679487013963566e-05,
+      "loss": 0.3853,
+      "step": 114
+    },
+    {
+      "epoch": 0.34808929247067727,
+      "grad_norm": 0.755529041845172,
+      "learning_rate": 1.9666068790931733e-05,
+      "loss": 0.3703,
+      "step": 115
+    },
+    {
+      "epoch": 0.3511161558834658,
+      "grad_norm": 0.6541501600357843,
+      "learning_rate": 1.9652380187177128e-05,
+      "loss": 0.3606,
+      "step": 116
+    },
+    {
+      "epoch": 0.35414301929625425,
+      "grad_norm": 0.6938651304629346,
+      "learning_rate": 1.9638421585599422e-05,
+      "loss": 0.369,
+      "step": 117
+    },
+    {
+      "epoch": 0.35716988270904276,
+      "grad_norm": 0.7723593971123168,
+      "learning_rate": 1.9624193376650708e-05,
+      "loss": 0.3602,
+      "step": 118
+    },
+    {
+      "epoch": 0.3601967461218313,
+      "grad_norm": 0.7139567709868522,
+      "learning_rate": 1.960969595832457e-05,
+      "loss": 0.3635,
+      "step": 119
+    },
+    {
+      "epoch": 0.36322360953461974,
+      "grad_norm": 0.77688878421537,
+      "learning_rate": 1.9594929736144978e-05,
+      "loss": 0.3547,
+      "step": 120
+    },
+    {
+      "epoch": 0.36625047294740826,
+      "grad_norm": 0.6757012649248691,
+      "learning_rate": 1.957989512315489e-05,
+      "loss": 0.3502,
+      "step": 121
+    },
+    {
+      "epoch": 0.3692773363601968,
+      "grad_norm": 0.7381821915922165,
+      "learning_rate": 1.956459253990476e-05,
+      "loss": 0.3569,
+      "step": 122
+    },
+    {
+      "epoch": 0.37230419977298523,
+      "grad_norm": 0.6760236779642173,
+      "learning_rate": 1.9549022414440738e-05,
+      "loss": 0.362,
+      "step": 123
+    },
+    {
+      "epoch": 0.37533106318577375,
+      "grad_norm": 0.7080537276107784,
+      "learning_rate": 1.9533185182292705e-05,
+      "loss": 0.365,
+      "step": 124
+    },
+    {
+      "epoch": 0.37835792659856227,
+      "grad_norm": 0.7133076040886531,
+      "learning_rate": 1.9517081286462082e-05,
+      "loss": 0.3559,
+      "step": 125
+    },
+    {
+      "epoch": 0.3813847900113507,
+      "grad_norm": 0.9828399631650351,
+      "learning_rate": 1.9500711177409456e-05,
+      "loss": 0.3539,
+      "step": 126
+    },
+    {
+      "epoch": 0.38441165342413924,
+      "grad_norm": 0.7084013435861097,
+      "learning_rate": 1.9484075313041968e-05,
+      "loss": 0.3721,
+      "step": 127
+    },
+    {
+      "epoch": 0.38743851683692776,
+      "grad_norm": 0.7628533725910128,
+      "learning_rate": 1.9467174158700507e-05,
+      "loss": 0.3772,
+      "step": 128
+    },
+    {
+      "epoch": 0.3904653802497162,
+      "grad_norm": 0.6762585784431314,
+      "learning_rate": 1.9450008187146685e-05,
+      "loss": 0.3663,
+      "step": 129
+    },
+    {
+      "epoch": 0.39349224366250474,
+      "grad_norm": 0.7158087505553148,
+      "learning_rate": 1.9432577878549635e-05,
+      "loss": 0.3742,
+      "step": 130
+    },
+    {
+      "epoch": 0.39651910707529325,
+      "grad_norm": 0.6947024555074877,
+      "learning_rate": 1.9414883720472557e-05,
+      "loss": 0.3595,
+      "step": 131
+    },
+    {
+      "epoch": 0.3995459704880817,
+      "grad_norm": 0.6969163264860893,
+      "learning_rate": 1.9396926207859085e-05,
+      "loss": 0.3512,
+      "step": 132
+    },
+    {
+      "epoch": 0.40257283390087023,
+      "grad_norm": 0.7294961902944351,
+      "learning_rate": 1.937870584301945e-05,
+      "loss": 0.3883,
+      "step": 133
+    },
+    {
+      "epoch": 0.40559969731365875,
+      "grad_norm": 0.6457701710987579,
+      "learning_rate": 1.9360223135616423e-05,
+      "loss": 0.3608,
+      "step": 134
+    },
+    {
+      "epoch": 0.4086265607264472,
+      "grad_norm": 0.7173994437601927,
+      "learning_rate": 1.9341478602651068e-05,
+      "loss": 0.3806,
+      "step": 135
+    },
+    {
+      "epoch": 0.4116534241392357,
+      "grad_norm": 0.6392998311676709,
+      "learning_rate": 1.932247276844826e-05,
+      "loss": 0.3541,
+      "step": 136
+    },
+    {
+      "epoch": 0.41468028755202424,
+      "grad_norm": 0.7512793841273344,
+      "learning_rate": 1.9303206164642037e-05,
+      "loss": 0.3792,
+      "step": 137
+    },
+    {
+      "epoch": 0.4177071509648127,
+      "grad_norm": 0.6565735904038665,
+      "learning_rate": 1.9283679330160726e-05,
+      "loss": 0.3397,
+      "step": 138
+    },
+    {
+      "epoch": 0.4207340143776012,
+      "grad_norm": 0.6707820206684717,
+      "learning_rate": 1.9263892811211865e-05,
+      "loss": 0.3584,
+      "step": 139
+    },
+    {
+      "epoch": 0.42376087779038973,
+      "grad_norm": 0.6534323861529194,
+      "learning_rate": 1.9243847161266924e-05,
+      "loss": 0.3567,
+      "step": 140
+    },
+    {
+      "epoch": 0.4267877412031782,
+      "grad_norm": 0.6795229574390856,
+      "learning_rate": 1.9223542941045817e-05,
+      "loss": 0.3541,
+      "step": 141
+    },
+    {
+      "epoch": 0.4298146046159667,
+      "grad_norm": 0.7208831286917592,
+      "learning_rate": 1.920298071850123e-05,
+      "loss": 0.3628,
+      "step": 142
+    },
+    {
+      "epoch": 0.4328414680287552,
+      "grad_norm": 0.7840018801148547,
+      "learning_rate": 1.9182161068802742e-05,
+      "loss": 0.3791,
+      "step": 143
+    },
+    {
+      "epoch": 0.4358683314415437,
+      "grad_norm": 0.7043953933630419,
+      "learning_rate": 1.9161084574320696e-05,
+      "loss": 0.3721,
+      "step": 144
+    },
+    {
+      "epoch": 0.4388951948543322,
+      "grad_norm": 0.6870885617493193,
+      "learning_rate": 1.913975182460996e-05,
+      "loss": 0.3624,
+      "step": 145
+    },
+    {
+      "epoch": 0.4419220582671207,
+      "grad_norm": 0.6951354895708466,
+      "learning_rate": 1.9118163416393392e-05,
+      "loss": 0.3552,
+      "step": 146
+    },
+    {
+      "epoch": 0.4449489216799092,
+      "grad_norm": 0.7966286451749266,
+      "learning_rate": 1.9096319953545186e-05,
+      "loss": 0.3818,
+      "step": 147
+    },
+    {
+      "epoch": 0.4479757850926977,
+      "grad_norm": 0.6977666801701438,
+      "learning_rate": 1.9074222047073945e-05,
+      "loss": 0.3533,
+      "step": 148
+    },
+    {
+      "epoch": 0.4510026485054862,
+      "grad_norm": 0.7095198961686515,
+      "learning_rate": 1.9051870315105626e-05,
+      "loss": 0.3612,
+      "step": 149
+    },
+    {
+      "epoch": 0.4540295119182747,
+      "grad_norm": 0.7802656650324079,
+      "learning_rate": 1.9029265382866216e-05,
+      "loss": 0.3565,
+      "step": 150
+    },
+    {
+      "epoch": 0.4570563753310632,
+      "grad_norm": 0.656414739226516,
+      "learning_rate": 1.9006407882664256e-05,
+      "loss": 0.3517,
+      "step": 151
+    },
+    {
+      "epoch": 0.4600832387438517,
+      "grad_norm": 0.6995682946672621,
+      "learning_rate": 1.8983298453873172e-05,
+      "loss": 0.3406,
+      "step": 152
+    },
+    {
+      "epoch": 0.46311010215664017,
+      "grad_norm": 0.6563258105268044,
+      "learning_rate": 1.895993774291336e-05,
+      "loss": 0.3449,
+      "step": 153
+    },
+    {
+      "epoch": 0.4661369655694287,
+      "grad_norm": 0.6630233739007957,
+      "learning_rate": 1.8936326403234125e-05,
+      "loss": 0.3464,
+      "step": 154
+    },
+    {
+      "epoch": 0.4691638289822172,
+      "grad_norm": 0.7041491781830824,
+      "learning_rate": 1.891246509529539e-05,
+      "loss": 0.3516,
+      "step": 155
+    },
+    {
+      "epoch": 0.47219069239500566,
+      "grad_norm": 0.6603080900864151,
+      "learning_rate": 1.8888354486549238e-05,
+      "loss": 0.3815,
+      "step": 156
+    },
+    {
+      "epoch": 0.4752175558077942,
+      "grad_norm": 0.7043196215498162,
+      "learning_rate": 1.886399525142122e-05,
+      "loss": 0.3859,
+      "step": 157
+    },
+    {
+      "epoch": 0.4782444192205827,
+      "grad_norm": 0.6903608871053512,
+      "learning_rate": 1.8839388071291506e-05,
+      "loss": 0.349,
+      "step": 158
+    },
+    {
+      "epoch": 0.48127128263337116,
+      "grad_norm": 0.6448952983732636,
+      "learning_rate": 1.881453363447582e-05,
+      "loss": 0.3337,
+      "step": 159
+    },
+    {
+      "epoch": 0.48429814604615967,
+      "grad_norm": 0.6463437537683724,
+      "learning_rate": 1.8789432636206197e-05,
+      "loss": 0.3624,
+      "step": 160
+    },
+    {
+      "epoch": 0.4873250094589482,
+      "grad_norm": 0.6263452253889749,
+      "learning_rate": 1.8764085778611507e-05,
+      "loss": 0.3366,
+      "step": 161
+    },
+    {
+      "epoch": 0.49035187287173665,
+      "grad_norm": 0.6649171326434136,
+      "learning_rate": 1.873849377069785e-05,
+      "loss": 0.3587,
+      "step": 162
+    },
+    {
+      "epoch": 0.49337873628452517,
+      "grad_norm": 0.7018969432536324,
+      "learning_rate": 1.87126573283287e-05,
+      "loss": 0.3646,
+      "step": 163
+    },
+    {
+      "epoch": 0.4964055996973137,
+      "grad_norm": 0.6818381246082317,
+      "learning_rate": 1.8686577174204887e-05,
+      "loss": 0.3826,
+      "step": 164
+    },
+    {
+      "epoch": 0.49943246311010214,
+      "grad_norm": 0.6305593485236309,
+      "learning_rate": 1.866025403784439e-05,
+      "loss": 0.3449,
+      "step": 165
+    },
+    {
+      "epoch": 0.5024593265228906,
+      "grad_norm": 5.358298166222743,
+      "learning_rate": 1.863368865556191e-05,
+      "loss": 0.375,
+      "step": 166
+    },
+    {
+      "epoch": 0.5054861899356792,
+      "grad_norm": 7.990017346980466,
+      "learning_rate": 1.8606881770448305e-05,
+      "loss": 0.4149,
+      "step": 167
+    },
+    {
+      "epoch": 0.5085130533484676,
+      "grad_norm": 2.6076807860160462,
+      "learning_rate": 1.8579834132349773e-05,
+      "loss": 0.3957,
+      "step": 168
+    },
+    {
+      "epoch": 0.5115399167612561,
+      "grad_norm": 1.0340709453456605,
+      "learning_rate": 1.8552546497846893e-05,
+      "loss": 0.3602,
+      "step": 169
+    },
+    {
+      "epoch": 0.5145667801740447,
+      "grad_norm": 2.461722928196783,
+      "learning_rate": 1.8525019630233463e-05,
+      "loss": 0.3676,
+      "step": 170
+    },
+    {
+      "epoch": 0.5175936435868331,
+      "grad_norm": 0.7732375799800538,
+      "learning_rate": 1.8497254299495147e-05,
+      "loss": 0.3696,
+      "step": 171
+    },
+    {
+      "epoch": 0.5206205069996216,
+      "grad_norm": 0.7562228050393626,
+      "learning_rate": 1.8469251282287925e-05,
+      "loss": 0.3636,
+      "step": 172
+    },
+    {
+      "epoch": 0.5236473704124102,
+      "grad_norm": 0.6426620621943433,
+      "learning_rate": 1.8441011361916387e-05,
+      "loss": 0.3565,
+      "step": 173
+    },
+    {
+      "epoch": 0.5266742338251986,
+      "grad_norm": 0.7509850658991156,
+      "learning_rate": 1.8412535328311813e-05,
+      "loss": 0.3549,
+      "step": 174
+    },
+    {
+      "epoch": 0.5297010972379871,
+      "grad_norm": 0.6692445640150556,
+      "learning_rate": 1.8383823978010077e-05,
+      "loss": 0.3624,
+      "step": 175
+    },
+    {
+      "epoch": 0.5327279606507757,
+      "grad_norm": 0.6977832041612454,
+      "learning_rate": 1.8354878114129368e-05,
+      "loss": 0.342,
+      "step": 176
+    },
+    {
+      "epoch": 0.5357548240635641,
+      "grad_norm": 0.7123708617567439,
+      "learning_rate": 1.8325698546347714e-05,
+      "loss": 0.3605,
+      "step": 177
+    },
+    {
+      "epoch": 0.5387816874763526,
+      "grad_norm": 0.866682366444865,
+      "learning_rate": 1.8296286090880362e-05,
+      "loss": 0.3454,
+      "step": 178
+    },
+    {
+      "epoch": 0.5418085508891411,
+      "grad_norm": 0.763836729800482,
+      "learning_rate": 1.8266641570456915e-05,
+      "loss": 0.3536,
+      "step": 179
+    },
+    {
+      "epoch": 0.5448354143019296,
+      "grad_norm": 0.6384798659884225,
+      "learning_rate": 1.8236765814298328e-05,
+      "loss": 0.3485,
+      "step": 180
+    },
+    {
+      "epoch": 0.5478622777147181,
+      "grad_norm": 0.6832113292151631,
+      "learning_rate": 1.820665965809373e-05,
+      "loss": 0.3414,
+      "step": 181
+    },
+    {
+      "epoch": 0.5508891411275066,
+      "grad_norm": 0.6819756440114341,
+      "learning_rate": 1.8176323943977034e-05,
+      "loss": 0.3551,
+      "step": 182
+    },
+    {
+      "epoch": 0.5539160045402951,
+      "grad_norm": 0.6673615593713749,
+      "learning_rate": 1.814575952050336e-05,
+      "loss": 0.3433,
+      "step": 183
+    },
+    {
+      "epoch": 0.5569428679530836,
+      "grad_norm": 0.7097405136951915,
+      "learning_rate": 1.8114967242625342e-05,
+      "loss": 0.3377,
+      "step": 184
+    },
+    {
+      "epoch": 0.5599697313658721,
+      "grad_norm": 0.7167454748509154,
+      "learning_rate": 1.808394797166919e-05,
+      "loss": 0.3545,
+      "step": 185
+    },
+    {
+      "epoch": 0.5629965947786606,
+      "grad_norm": 0.6765738560776606,
+      "learning_rate": 1.8052702575310588e-05,
+      "loss": 0.3573,
+      "step": 186
+    },
+    {
+      "epoch": 0.5660234581914491,
+      "grad_norm": 0.7565399926034315,
+      "learning_rate": 1.802123192755044e-05,
+      "loss": 0.3669,
+      "step": 187
+    },
+    {
+      "epoch": 0.5690503216042376,
+      "grad_norm": 0.6735307190958191,
+      "learning_rate": 1.7989536908690413e-05,
+      "loss": 0.3465,
+      "step": 188
+    },
+    {
+      "epoch": 0.5720771850170261,
+      "grad_norm": 0.6382861983908663,
+      "learning_rate": 1.7957618405308323e-05,
+      "loss": 0.3519,
+      "step": 189
+    },
+    {
+      "epoch": 0.5751040484298146,
+      "grad_norm": 0.6227843215548685,
+      "learning_rate": 1.792547731023332e-05,
+      "loss": 0.3246,
+      "step": 190
+    },
+    {
+      "epoch": 0.5781309118426031,
+      "grad_norm": 0.7133117252653911,
+      "learning_rate": 1.789311452252092e-05,
+      "loss": 0.3276,
+      "step": 191
+    },
+    {
+      "epoch": 0.5811577752553916,
+      "grad_norm": 0.644521702614739,
+      "learning_rate": 1.7860530947427878e-05,
+      "loss": 0.3402,
+      "step": 192
+    },
+    {
+      "epoch": 0.58418463866818,
+      "grad_norm": 0.7354079207423627,
+      "learning_rate": 1.782772749638682e-05,
+      "loss": 0.3473,
+      "step": 193
+    },
+    {
+      "epoch": 0.5872115020809686,
+      "grad_norm": 0.6573056129447126,
+      "learning_rate": 1.779470508698079e-05,
+      "loss": 0.3527,
+      "step": 194
+    },
+    {
+      "epoch": 0.5902383654937571,
+      "grad_norm": 0.6441042625320013,
+      "learning_rate": 1.776146464291757e-05,
+      "loss": 0.3423,
+      "step": 195
+    },
+    {
+      "epoch": 0.5932652289065455,
+      "grad_norm": 0.7171848091562617,
+      "learning_rate": 1.772800709400383e-05,
+      "loss": 0.3576,
+      "step": 196
+    },
+    {
+      "epoch": 0.5962920923193341,
+      "grad_norm": 0.6902976639058958,
+      "learning_rate": 1.7694333376119144e-05,
+      "loss": 0.3536,
+      "step": 197
+    },
+    {
+      "epoch": 0.5993189557321226,
+      "grad_norm": 0.6383242498591908,
+      "learning_rate": 1.766044443118978e-05,
+      "loss": 0.3261,
+      "step": 198
+    },
+    {
+      "epoch": 0.602345819144911,
+      "grad_norm": 0.6431197917283268,
+      "learning_rate": 1.762634120716238e-05,
+      "loss": 0.3256,
+      "step": 199
+    },
+    {
+      "epoch": 0.6053726825576996,
+      "grad_norm": 0.6729373472138573,
+      "learning_rate": 1.7592024657977432e-05,
+      "loss": 0.3537,
+      "step": 200
+    },
+    {
+      "epoch": 0.6083995459704881,
+      "grad_norm": 0.6633374948326238,
+      "learning_rate": 1.7557495743542586e-05,
+      "loss": 0.3485,
+      "step": 201
+    },
+    {
+      "epoch": 0.6114264093832765,
+      "grad_norm": 0.6395015806445717,
+      "learning_rate": 1.75227554297058e-05,
+      "loss": 0.3465,
+      "step": 202
+    },
+    {
+      "epoch": 0.6144532727960651,
+      "grad_norm": 0.642498687839001,
+      "learning_rate": 1.7487804688228327e-05,
+      "loss": 0.3654,
+      "step": 203
+    },
+    {
+      "epoch": 0.6174801362088536,
+      "grad_norm": 0.6433445120508655,
+      "learning_rate": 1.745264449675755e-05,
+      "loss": 0.3465,
+      "step": 204
+    },
+    {
+      "epoch": 0.620506999621642,
+      "grad_norm": 0.6259962543483338,
+      "learning_rate": 1.7417275838799596e-05,
+      "loss": 0.3486,
+      "step": 205
+    },
+    {
+      "epoch": 0.6235338630344306,
+      "grad_norm": 0.6442492322603779,
+      "learning_rate": 1.7381699703691866e-05,
+      "loss": 0.3484,
+      "step": 206
+    },
+    {
+      "epoch": 0.626560726447219,
+      "grad_norm": 0.6540114337394606,
+      "learning_rate": 1.734591708657533e-05,
+      "loss": 0.3525,
+      "step": 207
+    },
+    {
+      "epoch": 0.6295875898600075,
+      "grad_norm": 0.6516087731084556,
+      "learning_rate": 1.730992898836672e-05,
+      "loss": 0.3411,
+      "step": 208
+    },
+    {
+      "epoch": 0.6326144532727961,
+      "grad_norm": 0.6468985554146921,
+      "learning_rate": 1.7273736415730488e-05,
+      "loss": 0.3464,
+      "step": 209
+    },
+    {
+      "epoch": 0.6356413166855845,
+      "grad_norm": 0.6521613726875738,
+      "learning_rate": 1.72373403810507e-05,
+      "loss": 0.3382,
+      "step": 210
+    },
+    {
+      "epoch": 0.638668180098373,
+      "grad_norm": 0.6762678976694868,
+      "learning_rate": 1.720074190240269e-05,
+      "loss": 0.3626,
+      "step": 211
+    },
+    {
+      "epoch": 0.6416950435111616,
+      "grad_norm": 0.6079528221165867,
+      "learning_rate": 1.7163942003524574e-05,
+      "loss": 0.327,
+      "step": 212
+    },
+    {
+      "epoch": 0.64472190692395,
+      "grad_norm": 0.6242437225979809,
+      "learning_rate": 1.7126941713788633e-05,
+      "loss": 0.368,
+      "step": 213
+    },
+    {
+      "epoch": 0.6477487703367385,
+      "grad_norm": 0.6903975275505906,
+      "learning_rate": 1.70897420681725e-05,
+      "loss": 0.3415,
+      "step": 214
+    },
+    {
+      "epoch": 0.6507756337495271,
+      "grad_norm": 0.590203004828331,
+      "learning_rate": 1.7052344107230244e-05,
+      "loss": 0.3234,
+      "step": 215
+    },
+    {
+      "epoch": 0.6538024971623155,
+      "grad_norm": 0.6677697789314384,
+      "learning_rate": 1.7014748877063212e-05,
+      "loss": 0.3533,
+      "step": 216
+    },
+    {
+      "epoch": 0.656829360575104,
+      "grad_norm": 0.6670963352609979,
+      "learning_rate": 1.697695742929082e-05,
+      "loss": 0.3436,
+      "step": 217
+    },
+    {
+      "epoch": 0.6598562239878926,
+      "grad_norm": 0.6575625917378536,
+      "learning_rate": 1.693897082102109e-05,
+      "loss": 0.3472,
+      "step": 218
+    },
+    {
+      "epoch": 0.662883087400681,
+      "grad_norm": 0.5999593554665122,
+      "learning_rate": 1.6900790114821122e-05,
+      "loss": 0.3236,
+      "step": 219
+    },
+    {
+      "epoch": 0.6659099508134695,
+      "grad_norm": 0.6115779159757535,
+      "learning_rate": 1.686241637868734e-05,
+      "loss": 0.3365,
+      "step": 220
+    },
+    {
+      "epoch": 0.6689368142262581,
+      "grad_norm": 0.6444993917865307,
+      "learning_rate": 1.682385068601563e-05,
+      "loss": 0.3527,
+      "step": 221
+    },
+    {
+      "epoch": 0.6719636776390465,
+      "grad_norm": 0.6471351819890215,
+      "learning_rate": 1.6785094115571323e-05,
+      "loss": 0.3485,
+      "step": 222
+    },
+    {
+      "epoch": 0.6749905410518351,
+      "grad_norm": 0.6896050144079396,
+      "learning_rate": 1.674614775145901e-05,
+      "loss": 0.3504,
+      "step": 223
+    },
+    {
+      "epoch": 0.6780174044646236,
+      "grad_norm": 0.5976668968865958,
+      "learning_rate": 1.670701268309221e-05,
+      "loss": 0.3356,
+      "step": 224
+    },
+    {
+      "epoch": 0.681044267877412,
+      "grad_norm": 0.677779034753431,
+      "learning_rate": 1.666769000516292e-05,
+      "loss": 0.357,
+      "step": 225
+    },
+    {
+      "epoch": 0.6840711312902006,
+      "grad_norm": 0.64961250817071,
+      "learning_rate": 1.6628180817610963e-05,
+      "loss": 0.3447,
+      "step": 226
+    },
+    {
+      "epoch": 0.687097994702989,
+      "grad_norm": 0.5904212084223892,
+      "learning_rate": 1.658848622559325e-05,
+      "loss": 0.3348,
+      "step": 227
+    },
+    {
+      "epoch": 0.6901248581157775,
+      "grad_norm": 0.6021772246149271,
+      "learning_rate": 1.6548607339452853e-05,
+      "loss": 0.33,
+      "step": 228
+    },
+    {
+      "epoch": 0.6931517215285661,
+      "grad_norm": 0.6411129466686579,
+      "learning_rate": 1.6508545274687936e-05,
+      "loss": 0.3492,
+      "step": 229
+    },
+    {
+      "epoch": 0.6961785849413545,
+      "grad_norm": 0.6018246198139819,
+      "learning_rate": 1.6468301151920576e-05,
+      "loss": 0.3298,
+      "step": 230
+    },
+    {
+      "epoch": 0.699205448354143,
+      "grad_norm": 0.6221084767843331,
+      "learning_rate": 1.6427876096865394e-05,
+      "loss": 0.3311,
+      "step": 231
+    },
+    {
+      "epoch": 0.7022323117669316,
+      "grad_norm": 0.6368438744190789,
+      "learning_rate": 1.6387271240298082e-05,
+      "loss": 0.3399,
+      "step": 232
+    },
+    {
+      "epoch": 0.70525917517972,
+      "grad_norm": 0.6625039128766806,
+      "learning_rate": 1.6346487718023762e-05,
+      "loss": 0.3411,
+      "step": 233
+    },
+    {
+      "epoch": 0.7082860385925085,
+      "grad_norm": 0.6438120259103706,
+      "learning_rate": 1.6305526670845225e-05,
+      "loss": 0.3405,
+      "step": 234
+    },
+    {
+      "epoch": 0.7113129020052971,
+      "grad_norm": 0.6438242455382884,
+      "learning_rate": 1.6264389244531015e-05,
+      "loss": 0.3503,
+      "step": 235
+    },
+    {
+      "epoch": 0.7143397654180855,
+      "grad_norm": 0.6180274363890628,
+      "learning_rate": 1.6223076589783368e-05,
+      "loss": 0.35,
+      "step": 236
+    },
+    {
+      "epoch": 0.717366628830874,
+      "grad_norm": 0.6580117790166639,
+      "learning_rate": 1.6181589862206053e-05,
+      "loss": 0.3434,
+      "step": 237
+    },
+    {
+      "epoch": 0.7203934922436626,
+      "grad_norm": 0.6448417581379895,
+      "learning_rate": 1.613993022227202e-05,
+      "loss": 0.3441,
+      "step": 238
+    },
+    {
+      "epoch": 0.723420355656451,
+      "grad_norm": 0.6006923261650329,
+      "learning_rate": 1.6098098835290955e-05,
+      "loss": 0.3306,
+      "step": 239
+    },
+    {
+      "epoch": 0.7264472190692395,
+      "grad_norm": 0.6811013675430798,
+      "learning_rate": 1.6056096871376667e-05,
+      "loss": 0.3516,
+      "step": 240
+    },
+    {
+      "epoch": 0.729474082482028,
+      "grad_norm": 0.7168704377751873,
+      "learning_rate": 1.6013925505414386e-05,
+      "loss": 0.319,
+      "step": 241
+    },
+    {
+      "epoch": 0.7325009458948165,
+      "grad_norm": 0.6625804053728521,
+      "learning_rate": 1.5971585917027864e-05,
+      "loss": 0.3491,
+      "step": 242
+    },
+    {
+      "epoch": 0.735527809307605,
+      "grad_norm": 0.6070280940408769,
+      "learning_rate": 1.5929079290546408e-05,
+      "loss": 0.338,
+      "step": 243
+    },
+    {
+      "epoch": 0.7385546727203935,
+      "grad_norm": 0.6356704326247297,
+      "learning_rate": 1.5886406814971728e-05,
+      "loss": 0.3462,
+      "step": 244
+    },
+    {
+      "epoch": 0.741581536133182,
+      "grad_norm": 0.6081232942154914,
+      "learning_rate": 1.584356968394471e-05,
+      "loss": 0.3455,
+      "step": 245
+    },
+    {
+      "epoch": 0.7446083995459705,
+      "grad_norm": 0.6273621150510207,
+      "learning_rate": 1.5800569095711983e-05,
+      "loss": 0.3441,
+      "step": 246
+    },
+    {
+      "epoch": 0.747635262958759,
+      "grad_norm": 0.6075741999138976,
+      "learning_rate": 1.575740625309244e-05,
+      "loss": 0.3485,
+      "step": 247
+    },
+    {
+      "epoch": 0.7506621263715475,
+      "grad_norm": 0.6292049188325937,
+      "learning_rate": 1.5714082363443576e-05,
+      "loss": 0.349,
+      "step": 248
+    },
+    {
+      "epoch": 0.753688989784336,
+      "grad_norm": 0.6776329239537239,
+      "learning_rate": 1.5670598638627707e-05,
+      "loss": 0.3243,
+      "step": 249
+    },
+    {
+      "epoch": 0.7567158531971245,
+      "grad_norm": 0.6286049068663724,
+      "learning_rate": 1.5626956294978103e-05,
+      "loss": 0.3541,
+      "step": 250
+    },
+    {
+      "epoch": 0.759742716609913,
+      "grad_norm": 0.625336274524146,
+      "learning_rate": 1.5583156553264923e-05,
+      "loss": 0.331,
+      "step": 251
+    },
+    {
+      "epoch": 0.7627695800227015,
+      "grad_norm": 0.6282444513640331,
+      "learning_rate": 1.5539200638661106e-05,
+      "loss": 0.3237,
+      "step": 252
+    },
+    {
+      "epoch": 0.76579644343549,
+      "grad_norm": 0.657265913947973,
+      "learning_rate": 1.5495089780708062e-05,
+      "loss": 0.3558,
+      "step": 253
+    },
+    {
+      "epoch": 0.7688233068482785,
+      "grad_norm": 0.6718505469583875,
+      "learning_rate": 1.5450825213281317e-05,
+      "loss": 0.3298,
+      "step": 254
+    },
+    {
+      "epoch": 0.771850170261067,
+      "grad_norm": 0.6384288833859875,
+      "learning_rate": 1.5406408174555978e-05,
+      "loss": 0.3402,
+      "step": 255
+    },
+    {
+      "epoch": 0.7748770336738555,
+      "grad_norm": 0.6167079934155094,
+      "learning_rate": 1.5361839906972095e-05,
+      "loss": 0.332,
+      "step": 256
+    },
+    {
+      "epoch": 0.777903897086644,
+      "grad_norm": 0.5856521701270695,
+      "learning_rate": 1.531712165719992e-05,
+      "loss": 0.3268,
+      "step": 257
+    },
+    {
+      "epoch": 0.7809307604994324,
+      "grad_norm": 0.6315333440433661,
+      "learning_rate": 1.5272254676105026e-05,
+      "loss": 0.3339,
+      "step": 258
+    },
+    {
+      "epoch": 0.783957623912221,
+      "grad_norm": 0.6151852008505125,
+      "learning_rate": 1.5227240218713326e-05,
+      "loss": 0.3454,
+      "step": 259
+    },
+    {
+      "epoch": 0.7869844873250095,
+      "grad_norm": 0.6525177657122342,
+      "learning_rate": 1.5182079544175957e-05,
+      "loss": 0.3426,
+      "step": 260
+    },
+    {
+      "epoch": 0.7900113507377979,
+      "grad_norm": 2.0601118715003124,
+      "learning_rate": 1.5136773915734067e-05,
+      "loss": 0.3301,
+      "step": 261
+    },
+    {
+      "epoch": 0.7930382141505865,
+      "grad_norm": 0.6148405292834966,
+      "learning_rate": 1.5091324600683472e-05,
+      "loss": 0.3359,
+      "step": 262
+    },
+    {
+      "epoch": 0.796065077563375,
+      "grad_norm": 0.606554846455503,
+      "learning_rate": 1.5045732870339213e-05,
+      "loss": 0.3388,
+      "step": 263
+    },
+    {
+      "epoch": 0.7990919409761634,
+      "grad_norm": 0.6424196147504349,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.3203,
+      "step": 264
+    },
+    {
+      "epoch": 0.802118804388952,
+      "grad_norm": 0.7594411942376794,
+      "learning_rate": 1.4954127268912525e-05,
+      "loss": 0.3328,
+      "step": 265
+    },
+    {
+      "epoch": 0.8051456678017405,
+      "grad_norm": 0.594345133861949,
+      "learning_rate": 1.4908115960235683e-05,
+      "loss": 0.3232,
+      "step": 266
+    },
+    {
+      "epoch": 0.8081725312145289,
+      "grad_norm": 0.629145266898277,
+      "learning_rate": 1.4861967361004687e-05,
+      "loss": 0.3199,
+      "step": 267
+    },
+    {
+      "epoch": 0.8111993946273175,
+      "grad_norm": 0.6723694037426021,
+      "learning_rate": 1.4815682762095065e-05,
+      "loss": 0.3272,
+      "step": 268
+    },
+    {
+      "epoch": 0.814226258040106,
+      "grad_norm": 0.6081773003391021,
+      "learning_rate": 1.476926345818654e-05,
+      "loss": 0.3337,
+      "step": 269
+    },
+    {
+      "epoch": 0.8172531214528944,
+      "grad_norm": 0.6596995086118688,
+      "learning_rate": 1.472271074772683e-05,
+      "loss": 0.33,
+      "step": 270
+    },
+    {
+      "epoch": 0.820279984865683,
+      "grad_norm": 0.6639287479626809,
+      "learning_rate": 1.4676025932895315e-05,
+      "loss": 0.3375,
+      "step": 271
+    },
+    {
+      "epoch": 0.8233068482784714,
+      "grad_norm": 0.6350072169841312,
+      "learning_rate": 1.4629210319566626e-05,
+      "loss": 0.3391,
+      "step": 272
+    },
+    {
+      "epoch": 0.8263337116912599,
+      "grad_norm": 0.6648707919319814,
+      "learning_rate": 1.4582265217274105e-05,
+      "loss": 0.3311,
+      "step": 273
+    },
+    {
+      "epoch": 0.8293605751040485,
+      "grad_norm": 0.6171746682682803,
+      "learning_rate": 1.4535191939173179e-05,
+      "loss": 0.32,
+      "step": 274
+    },
+    {
+      "epoch": 0.8323874385168369,
+      "grad_norm": 0.6492823972913485,
+      "learning_rate": 1.4487991802004625e-05,
+      "loss": 0.3367,
+      "step": 275
+    },
+    {
+      "epoch": 0.8354143019296254,
+      "grad_norm": 0.707768687864805,
+      "learning_rate": 1.4440666126057743e-05,
+      "loss": 0.3355,
+      "step": 276
+    },
+    {
+      "epoch": 0.838441165342414,
+      "grad_norm": 0.6716001416412137,
+      "learning_rate": 1.4393216235133427e-05,
+      "loss": 0.3282,
+      "step": 277
+    },
+    {
+      "epoch": 0.8414680287552024,
+      "grad_norm": 0.6952828704049803,
+      "learning_rate": 1.4345643456507126e-05,
+      "loss": 0.339,
+      "step": 278
+    },
+    {
+      "epoch": 0.8444948921679909,
+      "grad_norm": 0.6578169862792245,
+      "learning_rate": 1.4297949120891718e-05,
+      "loss": 0.3495,
+      "step": 279
+    },
+    {
+      "epoch": 0.8475217555807795,
+      "grad_norm": 0.6836523479616898,
+      "learning_rate": 1.4250134562400301e-05,
+      "loss": 0.3281,
+      "step": 280
+    },
+    {
+      "epoch": 0.8505486189935679,
+      "grad_norm": 0.6698763755960151,
+      "learning_rate": 1.4202201118508863e-05,
+      "loss": 0.3328,
+      "step": 281
+    },
+    {
+      "epoch": 0.8535754824063564,
+      "grad_norm": 0.6259093741559557,
+      "learning_rate": 1.4154150130018867e-05,
+      "loss": 0.3269,
+      "step": 282
+    },
+    {
+      "epoch": 0.856602345819145,
+      "grad_norm": 0.685182965238572,
+      "learning_rate": 1.4105982941019751e-05,
+      "loss": 0.3322,
+      "step": 283
+    },
+    {
+      "epoch": 0.8596292092319334,
+      "grad_norm": 0.6998080935948778,
+      "learning_rate": 1.405770089885134e-05,
+      "loss": 0.3397,
+      "step": 284
+    },
+    {
+      "epoch": 0.8626560726447219,
+      "grad_norm": 0.5960276092687048,
+      "learning_rate": 1.4009305354066138e-05,
+      "loss": 0.3237,
+      "step": 285
+    },
+    {
+      "epoch": 0.8656829360575105,
+      "grad_norm": 0.6626436083226441,
+      "learning_rate": 1.396079766039157e-05,
+      "loss": 0.3189,
+      "step": 286
+    },
+    {
+      "epoch": 0.8687097994702989,
+      "grad_norm": 0.6103017111852415,
+      "learning_rate": 1.39121791746921e-05,
+      "loss": 0.333,
+      "step": 287
+    },
+    {
+      "epoch": 0.8717366628830874,
+      "grad_norm": 0.6017406546850014,
+      "learning_rate": 1.3863451256931286e-05,
+      "loss": 0.3237,
+      "step": 288
+    },
+    {
+      "epoch": 0.874763526295876,
+      "grad_norm": 0.5710358909353168,
+      "learning_rate": 1.381461527013374e-05,
+      "loss": 0.3068,
+      "step": 289
+    },
+    {
+      "epoch": 0.8777903897086644,
+      "grad_norm": 0.6425357205467291,
+      "learning_rate": 1.3765672580346986e-05,
+      "loss": 0.3313,
+      "step": 290
+    },
+    {
+      "epoch": 0.8808172531214529,
+      "grad_norm": 0.6320879925844012,
+      "learning_rate": 1.3716624556603275e-05,
+      "loss": 0.3355,
+      "step": 291
+    },
+    {
+      "epoch": 0.8838441165342414,
+      "grad_norm": 0.629323259224594,
+      "learning_rate": 1.3667472570881264e-05,
+      "loss": 0.3184,
+      "step": 292
+    },
+    {
+      "epoch": 0.8868709799470299,
+      "grad_norm": 0.6347394350406389,
+      "learning_rate": 1.361821799806765e-05,
+      "loss": 0.3425,
+      "step": 293
+    },
+    {
+      "epoch": 0.8898978433598184,
+      "grad_norm": 0.5821466399153497,
+      "learning_rate": 1.356886221591872e-05,
+      "loss": 0.3206,
+      "step": 294
+    },
+    {
+      "epoch": 0.8929247067726069,
+      "grad_norm": 0.680615693134758,
+      "learning_rate": 1.3519406605021797e-05,
+      "loss": 0.3379,
+      "step": 295
+    },
+    {
+      "epoch": 0.8959515701853954,
+      "grad_norm": 0.5954627355314841,
+      "learning_rate": 1.3469852548756626e-05,
+      "loss": 0.3191,
+      "step": 296
+    },
+    {
+      "epoch": 0.8989784335981839,
+      "grad_norm": 0.5781234816777454,
+      "learning_rate": 1.342020143325669e-05,
+      "loss": 0.3251,
+      "step": 297
+    },
+    {
+      "epoch": 0.9020052970109724,
+      "grad_norm": 0.5963819952771768,
+      "learning_rate": 1.3370454647370418e-05,
+      "loss": 0.3255,
+      "step": 298
+    },
+    {
+      "epoch": 0.9050321604237609,
+      "grad_norm": 0.603037499891043,
+      "learning_rate": 1.3320613582622354e-05,
+      "loss": 0.3232,
+      "step": 299
+    },
+    {
+      "epoch": 0.9080590238365494,
+      "grad_norm": 0.6396937747313373,
+      "learning_rate": 1.3270679633174219e-05,
+      "loss": 0.3313,
+      "step": 300
+    },
+    {
+      "epoch": 0.9110858872493379,
+      "grad_norm": 0.6340697453770731,
+      "learning_rate": 1.3220654195785917e-05,
+      "loss": 0.3292,
+      "step": 301
+    },
+    {
+      "epoch": 0.9141127506621264,
+      "grad_norm": 0.6601671806023984,
+      "learning_rate": 1.3170538669776469e-05,
+      "loss": 0.3216,
+      "step": 302
+    },
+    {
+      "epoch": 0.9171396140749148,
+      "grad_norm": 0.5956375098950856,
+      "learning_rate": 1.3120334456984871e-05,
+      "loss": 0.3061,
+      "step": 303
+    },
+    {
+      "epoch": 0.9201664774877034,
+      "grad_norm": 0.595106620690898,
+      "learning_rate": 1.3070042961730878e-05,
+      "loss": 0.318,
+      "step": 304
+    },
+    {
+      "epoch": 0.9231933409004919,
+      "grad_norm": 0.6128775382195482,
+      "learning_rate": 1.3019665590775717e-05,
+      "loss": 0.3356,
+      "step": 305
+    },
+    {
+      "epoch": 0.9262202043132803,
+      "grad_norm": 0.6442034237161239,
+      "learning_rate": 1.296920375328275e-05,
+      "loss": 0.3207,
+      "step": 306
+    },
+    {
+      "epoch": 0.9292470677260689,
+      "grad_norm": 0.5986326243545633,
+      "learning_rate": 1.2918658860778046e-05,
+      "loss": 0.3265,
+      "step": 307
+    },
+    {
+      "epoch": 0.9322739311388574,
+      "grad_norm": 0.6031097070579334,
+      "learning_rate": 1.2868032327110904e-05,
+      "loss": 0.3172,
+      "step": 308
+    },
+    {
+      "epoch": 0.9353007945516458,
+      "grad_norm": 0.5750713572766266,
+      "learning_rate": 1.2817325568414299e-05,
+      "loss": 0.3139,
+      "step": 309
+    },
+    {
+      "epoch": 0.9383276579644344,
+      "grad_norm": 0.5954487529156475,
+      "learning_rate": 1.2766540003065272e-05,
+      "loss": 0.3398,
+      "step": 310
+    },
+    {
+      "epoch": 0.9413545213772229,
+      "grad_norm": 0.6538536406473328,
+      "learning_rate": 1.2715677051645259e-05,
+      "loss": 0.3185,
+      "step": 311
+    },
+    {
+      "epoch": 0.9443813847900113,
+      "grad_norm": 0.616864572506109,
+      "learning_rate": 1.266473813690035e-05,
+      "loss": 0.3211,
+      "step": 312
+    },
+    {
+      "epoch": 0.9474082482027999,
+      "grad_norm": 0.5816201717490193,
+      "learning_rate": 1.2613724683701491e-05,
+      "loss": 0.3218,
+      "step": 313
+    },
+    {
+      "epoch": 0.9504351116155884,
+      "grad_norm": 0.5947876312165618,
+      "learning_rate": 1.2562638119004627e-05,
+      "loss": 0.3198,
+      "step": 314
+    },
+    {
+      "epoch": 0.9534619750283768,
+      "grad_norm": 0.6440220254668659,
+      "learning_rate": 1.2511479871810792e-05,
+      "loss": 0.3318,
+      "step": 315
+    },
+    {
+      "epoch": 0.9564888384411654,
+      "grad_norm": 0.6197719734293635,
+      "learning_rate": 1.2460251373126136e-05,
+      "loss": 0.3256,
+      "step": 316
+    },
+    {
+      "epoch": 0.9595157018539539,
+      "grad_norm": 0.5738872482746382,
+      "learning_rate": 1.2408954055921884e-05,
+      "loss": 0.3172,
+      "step": 317
+    },
+    {
+      "epoch": 0.9625425652667423,
+      "grad_norm": 0.6110187036001771,
+      "learning_rate": 1.2357589355094275e-05,
+      "loss": 0.3433,
+      "step": 318
+    },
+    {
+      "epoch": 0.9655694286795309,
+      "grad_norm": 0.5656219895128842,
+      "learning_rate": 1.2306158707424402e-05,
+      "loss": 0.3366,
+      "step": 319
+    },
+    {
+      "epoch": 0.9685962920923193,
+      "grad_norm": 0.6109170055438233,
+      "learning_rate": 1.2254663551538047e-05,
+      "loss": 0.32,
+      "step": 320
+    },
+    {
+      "epoch": 0.9716231555051078,
+      "grad_norm": 0.5931757539379091,
+      "learning_rate": 1.2203105327865407e-05,
+      "loss": 0.342,
+      "step": 321
+    },
+    {
+      "epoch": 0.9746500189178964,
+      "grad_norm": 0.5966887324193206,
+      "learning_rate": 1.215148547860084e-05,
+      "loss": 0.3201,
+      "step": 322
+    },
+    {
+      "epoch": 0.9776768823306848,
+      "grad_norm": 0.591797249538581,
+      "learning_rate": 1.2099805447662485e-05,
+      "loss": 0.3261,
+      "step": 323
+    },
+    {
+      "epoch": 0.9807037457434733,
+      "grad_norm": 0.610160648958815,
+      "learning_rate": 1.2048066680651908e-05,
+      "loss": 0.3477,
+      "step": 324
+    },
+    {
+      "epoch": 0.9837306091562619,
+      "grad_norm": 0.5833666072992763,
+      "learning_rate": 1.1996270624813642e-05,
+      "loss": 0.3223,
+      "step": 325
+    },
+    {
+      "epoch": 0.9867574725690503,
+      "grad_norm": 0.5774192865557437,
+      "learning_rate": 1.194441872899471e-05,
+      "loss": 0.3078,
+      "step": 326
+    },
+    {
+      "epoch": 0.9897843359818388,
+      "grad_norm": 0.6237299181052061,
+      "learning_rate": 1.1892512443604103e-05,
+      "loss": 0.3335,
+      "step": 327
+    },
+    {
+      "epoch": 0.9928111993946274,
+      "grad_norm": 0.6104295420390287,
+      "learning_rate": 1.1840553220572204e-05,
+      "loss": 0.3374,
+      "step": 328
+    },
+    {
+      "epoch": 0.9958380628074158,
+      "grad_norm": 0.5776365546306047,
+      "learning_rate": 1.1788542513310178e-05,
+      "loss": 0.338,
+      "step": 329
+    },
+    {
+      "epoch": 0.9988649262202043,
+      "grad_norm": 0.5785406820693397,
+      "learning_rate": 1.1736481776669307e-05,
+      "loss": 0.3044,
+      "step": 330
+    },
+    {
+      "epoch": 1.0018917896329929,
+      "grad_norm": 0.632622001627882,
+      "learning_rate": 1.1684372466900306e-05,
+      "loss": 0.2764,
+      "step": 331
+    },
+    {
+      "epoch": 1.0049186530457812,
+      "grad_norm": 0.7186535316373412,
+      "learning_rate": 1.1632216041612595e-05,
+      "loss": 0.2374,
+      "step": 332
+    },
+    {
+      "epoch": 1.0079455164585698,
+      "grad_norm": 0.6329342128257043,
+      "learning_rate": 1.15800139597335e-05,
+      "loss": 0.2238,
+      "step": 333
+    },
+    {
+      "epoch": 1.0109723798713584,
+      "grad_norm": 0.6181301921849538,
+      "learning_rate": 1.1527767681467472e-05,
+      "loss": 0.2448,
+      "step": 334
+    },
+    {
+      "epoch": 1.0139992432841467,
+      "grad_norm": 0.6833195509010448,
+      "learning_rate": 1.1475478668255223e-05,
+      "loss": 0.2365,
+      "step": 335
+    },
+    {
+      "epoch": 1.0170261066969353,
+      "grad_norm": 0.719463280143904,
+      "learning_rate": 1.1423148382732854e-05,
+      "loss": 0.2267,
+      "step": 336
+    },
+    {
+      "epoch": 1.0200529701097238,
+      "grad_norm": 0.7055311937993489,
+      "learning_rate": 1.1370778288690947e-05,
+      "loss": 0.2266,
+      "step": 337
+    },
+    {
+      "epoch": 1.0230798335225122,
+      "grad_norm": 0.6659076647698453,
+      "learning_rate": 1.1318369851033604e-05,
+      "loss": 0.231,
+      "step": 338
+    },
+    {
+      "epoch": 1.0261066969353008,
+      "grad_norm": 0.639651633733867,
+      "learning_rate": 1.1265924535737494e-05,
+      "loss": 0.2149,
+      "step": 339
+    },
+    {
+      "epoch": 1.0291335603480893,
+      "grad_norm": 0.6620609296766656,
+      "learning_rate": 1.121344380981082e-05,
+      "loss": 0.2185,
+      "step": 340
+    },
+    {
+      "epoch": 1.0321604237608777,
+      "grad_norm": 0.6651276576514856,
+      "learning_rate": 1.1160929141252303e-05,
+      "loss": 0.2278,
+      "step": 341
+    },
+    {
+      "epoch": 1.0351872871736663,
+      "grad_norm": 0.6353643443259348,
+      "learning_rate": 1.1108381999010111e-05,
+      "loss": 0.2254,
+      "step": 342
+    },
+    {
+      "epoch": 1.0382141505864548,
+      "grad_norm": 0.6388142941339991,
+      "learning_rate": 1.1055803852940772e-05,
+      "loss": 0.2259,
+      "step": 343
+    },
+    {
+      "epoch": 1.0412410139992432,
+      "grad_norm": 0.6195195418616724,
+      "learning_rate": 1.1003196173768051e-05,
+      "loss": 0.2275,
+      "step": 344
+    },
+    {
+      "epoch": 1.0442678774120318,
+      "grad_norm": 0.6667543345358281,
+      "learning_rate": 1.0950560433041825e-05,
+      "loss": 0.2282,
+      "step": 345
+    },
+    {
+      "epoch": 1.0472947408248203,
+      "grad_norm": 0.607262534132557,
+      "learning_rate": 1.0897898103096917e-05,
+      "loss": 0.217,
+      "step": 346
+    },
+    {
+      "epoch": 1.0503216042376087,
+      "grad_norm": 0.6449449031721443,
+      "learning_rate": 1.0845210657011893e-05,
+      "loss": 0.226,
+      "step": 347
+    },
+    {
+      "epoch": 1.0533484676503972,
+      "grad_norm": 0.6367999965959172,
+      "learning_rate": 1.0792499568567885e-05,
+      "loss": 0.2183,
+      "step": 348
+    },
+    {
+      "epoch": 1.0563753310631858,
+      "grad_norm": 0.6222423903243186,
+      "learning_rate": 1.0739766312207344e-05,
+      "loss": 0.2195,
+      "step": 349
+    },
+    {
+      "epoch": 1.0594021944759742,
+      "grad_norm": 0.6212450088702329,
+      "learning_rate": 1.068701236299281e-05,
+      "loss": 0.2186,
+      "step": 350
+    },
+    {
+      "epoch": 1.0624290578887627,
+      "grad_norm": 0.612150003968409,
+      "learning_rate": 1.0634239196565646e-05,
+      "loss": 0.2203,
+      "step": 351
+    },
+    {
+      "epoch": 1.0654559213015513,
+      "grad_norm": 0.6194218540590405,
+      "learning_rate": 1.0581448289104759e-05,
+      "loss": 0.2175,
+      "step": 352
+    },
+    {
+      "epoch": 1.0684827847143397,
+      "grad_norm": 0.6458278272718033,
+      "learning_rate": 1.0528641117285315e-05,
+      "loss": 0.2236,
+      "step": 353
+    },
+    {
+      "epoch": 1.0715096481271282,
+      "grad_norm": 0.6147953294881713,
+      "learning_rate": 1.0475819158237426e-05,
+      "loss": 0.2078,
+      "step": 354
+    },
+    {
+      "epoch": 1.0745365115399168,
+      "grad_norm": 0.6276120471148431,
+      "learning_rate": 1.0422983889504831e-05,
+      "loss": 0.2233,
+      "step": 355
+    },
+    {
+      "epoch": 1.0775633749527052,
+      "grad_norm": 0.6762936899362387,
+      "learning_rate": 1.0370136789003582e-05,
+      "loss": 0.2202,
+      "step": 356
+    },
+    {
+      "epoch": 1.0805902383654937,
+      "grad_norm": 0.6185871388612753,
+      "learning_rate": 1.031727933498068e-05,
+      "loss": 0.2256,
+      "step": 357
+    },
+    {
+      "epoch": 1.0836171017782823,
+      "grad_norm": 0.6013566266274644,
+      "learning_rate": 1.0264413005972736e-05,
+      "loss": 0.2104,
+      "step": 358
+    },
+    {
+      "epoch": 1.0866439651910706,
+      "grad_norm": 0.617470201874533,
+      "learning_rate": 1.0211539280764617e-05,
+      "loss": 0.2214,
+      "step": 359
+    },
+    {
+      "epoch": 1.0896708286038592,
+      "grad_norm": 0.6341822551734001,
+      "learning_rate": 1.015865963834808e-05,
+      "loss": 0.2311,
+      "step": 360
+    },
+    {
+      "epoch": 1.0926976920166478,
+      "grad_norm": 0.5916702110200754,
+      "learning_rate": 1.0105775557880398e-05,
+      "loss": 0.218,
+      "step": 361
+    },
+    {
+      "epoch": 1.0957245554294364,
+      "grad_norm": 0.6469301349778502,
+      "learning_rate": 1.0052888518642978e-05,
+      "loss": 0.2226,
+      "step": 362
+    },
+    {
+      "epoch": 1.0987514188422247,
+      "grad_norm": 0.6314773957505443,
+      "learning_rate": 1e-05,
+      "loss": 0.2196,
+      "step": 363
+    },
+    {
+      "epoch": 1.1017782822550133,
+      "grad_norm": 0.6218914142376414,
+      "learning_rate": 9.947111481357023e-06,
+      "loss": 0.2109,
+      "step": 364
+    },
+    {
+      "epoch": 1.1048051456678016,
+      "grad_norm": 0.6343787108826822,
+      "learning_rate": 9.894224442119606e-06,
+      "loss": 0.2249,
+      "step": 365
+    },
+    {
+      "epoch": 1.1078320090805902,
+      "grad_norm": 0.609809344209566,
+      "learning_rate": 9.841340361651921e-06,
+      "loss": 0.2166,
+      "step": 366
+    },
+    {
+      "epoch": 1.1108588724933788,
+      "grad_norm": 0.6473140697520069,
+      "learning_rate": 9.788460719235386e-06,
+      "loss": 0.2232,
+      "step": 367
+    },
+    {
+      "epoch": 1.1138857359061674,
+      "grad_norm": 0.5748472611307098,
+      "learning_rate": 9.735586994027267e-06,
+      "loss": 0.2099,
+      "step": 368
+    },
+    {
+      "epoch": 1.1169125993189557,
+      "grad_norm": 0.6316430849130864,
+      "learning_rate": 9.682720665019325e-06,
+      "loss": 0.2144,
+      "step": 369
+    },
+    {
+      "epoch": 1.1199394627317443,
+      "grad_norm": 0.6296007505286115,
+      "learning_rate": 9.62986321099642e-06,
+      "loss": 0.2208,
+      "step": 370
+    },
+    {
+      "epoch": 1.1229663261445326,
+      "grad_norm": 0.6823858600011482,
+      "learning_rate": 9.57701611049517e-06,
+      "loss": 0.2242,
+      "step": 371
+    },
+    {
+      "epoch": 1.1259931895573212,
+      "grad_norm": 0.6650894034096742,
+      "learning_rate": 9.524180841762577e-06,
+      "loss": 0.212,
+      "step": 372
+    },
+    {
+      "epoch": 1.1290200529701098,
+      "grad_norm": 0.6336561626195444,
+      "learning_rate": 9.471358882714687e-06,
+      "loss": 0.2196,
+      "step": 373
+    },
+    {
+      "epoch": 1.1320469163828983,
+      "grad_norm": 0.6379191893873614,
+      "learning_rate": 9.418551710895243e-06,
+      "loss": 0.2162,
+      "step": 374
+    },
+    {
+      "epoch": 1.1350737797956867,
+      "grad_norm": 0.6542484207305792,
+      "learning_rate": 9.365760803434356e-06,
+      "loss": 0.2217,
+      "step": 375
+    },
+    {
+      "epoch": 1.1381006432084753,
+      "grad_norm": 0.6373650182223146,
+      "learning_rate": 9.312987637007191e-06,
+      "loss": 0.2219,
+      "step": 376
+    },
+    {
+      "epoch": 1.1411275066212636,
+      "grad_norm": 0.6257665216598209,
+      "learning_rate": 9.260233687792657e-06,
+      "loss": 0.2162,
+      "step": 377
+    },
+    {
+      "epoch": 1.1441543700340522,
+      "grad_norm": 0.6609828970438415,
+      "learning_rate": 9.207500431432115e-06,
+      "loss": 0.2292,
+      "step": 378
+    },
+    {
+      "epoch": 1.1471812334468408,
+      "grad_norm": 0.6017357468261275,
+      "learning_rate": 9.154789342988108e-06,
+      "loss": 0.2184,
+      "step": 379
+    },
+    {
+      "epoch": 1.1502080968596293,
+      "grad_norm": 0.6323834438254539,
+      "learning_rate": 9.102101896903084e-06,
+      "loss": 0.2185,
+      "step": 380
+    },
+    {
+      "epoch": 1.1532349602724177,
+      "grad_norm": 0.660070921182074,
+      "learning_rate": 9.049439566958176e-06,
+      "loss": 0.2274,
+      "step": 381
+    },
+    {
+      "epoch": 1.1562618236852062,
+      "grad_norm": 0.6084704841010108,
+      "learning_rate": 8.99680382623195e-06,
+      "loss": 0.2121,
+      "step": 382
+    },
+    {
+      "epoch": 1.1592886870979946,
+      "grad_norm": 0.6186544372739292,
+      "learning_rate": 8.944196147059233e-06,
+      "loss": 0.2162,
+      "step": 383
+    },
+    {
+      "epoch": 1.1623155505107832,
+      "grad_norm": 0.6230296751611448,
+      "learning_rate": 8.89161800098989e-06,
+      "loss": 0.2153,
+      "step": 384
+    },
+    {
+      "epoch": 1.1653424139235717,
+      "grad_norm": 0.626041423363818,
+      "learning_rate": 8.839070858747697e-06,
+      "loss": 0.2161,
+      "step": 385
+    },
+    {
+      "epoch": 1.1683692773363603,
+      "grad_norm": 0.652915069580727,
+      "learning_rate": 8.786556190189183e-06,
+      "loss": 0.2212,
+      "step": 386
+    },
+    {
+      "epoch": 1.1713961407491487,
+      "grad_norm": 0.6554779366999796,
+      "learning_rate": 8.734075464262507e-06,
+      "loss": 0.2167,
+      "step": 387
+    },
+    {
+      "epoch": 1.1744230041619372,
+      "grad_norm": 0.6260055940986169,
+      "learning_rate": 8.681630148966397e-06,
+      "loss": 0.2087,
+      "step": 388
+    },
+    {
+      "epoch": 1.1774498675747256,
+      "grad_norm": 0.6477810307466224,
+      "learning_rate": 8.629221711309056e-06,
+      "loss": 0.2122,
+      "step": 389
+    },
+    {
+      "epoch": 1.1804767309875142,
+      "grad_norm": 0.6669101250263473,
+      "learning_rate": 8.576851617267151e-06,
+      "loss": 0.2283,
+      "step": 390
+    },
+    {
+      "epoch": 1.1835035944003027,
+      "grad_norm": 0.6085419246382261,
+      "learning_rate": 8.52452133174478e-06,
+      "loss": 0.2048,
+      "step": 391
+    },
+    {
+      "epoch": 1.1865304578130913,
+      "grad_norm": 0.6527792323115998,
+      "learning_rate": 8.472232318532531e-06,
+      "loss": 0.2255,
+      "step": 392
+    },
+    {
+      "epoch": 1.1895573212258796,
+      "grad_norm": 0.6187593856551973,
+      "learning_rate": 8.419986040266502e-06,
+      "loss": 0.2159,
+      "step": 393
+    },
+    {
+      "epoch": 1.1925841846386682,
+      "grad_norm": 0.6307064556353604,
+      "learning_rate": 8.367783958387407e-06,
+      "loss": 0.2134,
+      "step": 394
+    },
+    {
+      "epoch": 1.1956110480514566,
+      "grad_norm": 0.6640289996584182,
+      "learning_rate": 8.315627533099697e-06,
+      "loss": 0.2261,
+      "step": 395
+    },
+    {
+      "epoch": 1.1986379114642451,
+      "grad_norm": 0.6349221985704178,
+      "learning_rate": 8.263518223330698e-06,
+      "loss": 0.2223,
+      "step": 396
+    },
+    {
+      "epoch": 1.2016647748770337,
+      "grad_norm": 0.6554209162411619,
+      "learning_rate": 8.211457486689829e-06,
+      "loss": 0.2171,
+      "step": 397
+    },
+    {
+      "epoch": 1.2046916382898223,
+      "grad_norm": 0.67547725613166,
+      "learning_rate": 8.159446779427798e-06,
+      "loss": 0.2246,
+      "step": 398
+    },
+    {
+      "epoch": 1.2077185017026106,
+      "grad_norm": 0.6304773870775762,
+      "learning_rate": 8.107487556395902e-06,
+      "loss": 0.2157,
+      "step": 399
+    },
+    {
+      "epoch": 1.2107453651153992,
+      "grad_norm": 0.6070999733657653,
+      "learning_rate": 8.055581271005292e-06,
+      "loss": 0.2143,
+      "step": 400
+    },
+    {
+      "epoch": 1.2137722285281876,
+      "grad_norm": 0.7298232509159093,
+      "learning_rate": 8.00372937518636e-06,
+      "loss": 0.2246,
+      "step": 401
+    },
+    {
+      "epoch": 1.2167990919409761,
+      "grad_norm": 0.6118800553731263,
+      "learning_rate": 7.951933319348095e-06,
+      "loss": 0.2227,
+      "step": 402
+    },
+    {
+      "epoch": 1.2198259553537647,
+      "grad_norm": 0.5878921291474919,
+      "learning_rate": 7.900194552337516e-06,
+      "loss": 0.2056,
+      "step": 403
+    },
+    {
+      "epoch": 1.2228528187665533,
+      "grad_norm": 0.645364333349695,
+      "learning_rate": 7.848514521399167e-06,
+      "loss": 0.2291,
+      "step": 404
+    },
+    {
+      "epoch": 1.2258796821793416,
+      "grad_norm": 0.6740697249365711,
+      "learning_rate": 7.796894672134594e-06,
+      "loss": 0.2252,
+      "step": 405
+    },
+    {
+      "epoch": 1.2289065455921302,
+      "grad_norm": 0.6413698711962775,
+      "learning_rate": 7.745336448461958e-06,
+      "loss": 0.2259,
+      "step": 406
+    },
+    {
+      "epoch": 1.2319334090049185,
+      "grad_norm": 0.5749866175676727,
+      "learning_rate": 7.6938412925756e-06,
+      "loss": 0.2195,
+      "step": 407
+    },
+    {
+      "epoch": 1.2349602724177071,
+      "grad_norm": 0.6023656670496965,
+      "learning_rate": 7.642410644905726e-06,
+      "loss": 0.2124,
+      "step": 408
+    },
+    {
+      "epoch": 1.2379871358304957,
+      "grad_norm": 0.6237982462823463,
+      "learning_rate": 7.591045944078119e-06,
+      "loss": 0.2264,
+      "step": 409
+    },
+    {
+      "epoch": 1.2410139992432843,
+      "grad_norm": 0.7284751927557074,
+      "learning_rate": 7.539748626873866e-06,
+      "loss": 0.2182,
+      "step": 410
+    },
+    {
+      "epoch": 1.2440408626560726,
+      "grad_norm": 0.6365512378807111,
+      "learning_rate": 7.488520128189209e-06,
+      "loss": 0.2163,
+      "step": 411
+    },
+    {
+      "epoch": 1.2470677260688612,
+      "grad_norm": 0.6223553048159864,
+      "learning_rate": 7.4373618809953755e-06,
+      "loss": 0.2099,
+      "step": 412
+    },
+    {
+      "epoch": 1.2500945894816495,
+      "grad_norm": 0.6748262691345155,
+      "learning_rate": 7.386275316298513e-06,
+      "loss": 0.2275,
+      "step": 413
+    },
+    {
+      "epoch": 1.253121452894438,
+      "grad_norm": 0.6208295989154898,
+      "learning_rate": 7.335261863099652e-06,
+      "loss": 0.2163,
+      "step": 414
+    },
+    {
+      "epoch": 1.2561483163072267,
+      "grad_norm": 0.6159567618877604,
+      "learning_rate": 7.2843229483547405e-06,
+      "loss": 0.2106,
+      "step": 415
+    },
+    {
+      "epoch": 1.2591751797200152,
+      "grad_norm": 0.8859672316119711,
+      "learning_rate": 7.233459996934731e-06,
+      "loss": 0.2186,
+      "step": 416
+    },
+    {
+      "epoch": 1.2622020431328036,
+      "grad_norm": 0.6001126743845364,
+      "learning_rate": 7.182674431585703e-06,
+      "loss": 0.2124,
+      "step": 417
+    },
+    {
+      "epoch": 1.2652289065455922,
+      "grad_norm": 0.6245209813888175,
+      "learning_rate": 7.131967672889101e-06,
+      "loss": 0.2141,
+      "step": 418
+    },
+    {
+      "epoch": 1.2682557699583805,
+      "grad_norm": 0.6383126764704137,
+      "learning_rate": 7.081341139221955e-06,
+      "loss": 0.2218,
+      "step": 419
+    },
+    {
+      "epoch": 1.271282633371169,
+      "grad_norm": 0.6275243226642444,
+      "learning_rate": 7.0307962467172555e-06,
+      "loss": 0.2213,
+      "step": 420
+    },
+    {
+      "epoch": 1.2743094967839577,
+      "grad_norm": 0.6313847562368273,
+      "learning_rate": 6.9803344092242855e-06,
+      "loss": 0.2253,
+      "step": 421
+    },
+    {
+      "epoch": 1.2773363601967462,
+      "grad_norm": 0.6665406782893395,
+      "learning_rate": 6.929957038269123e-06,
+      "loss": 0.2165,
+      "step": 422
+    },
+    {
+      "epoch": 1.2803632236095346,
+      "grad_norm": 0.5862543856389217,
+      "learning_rate": 6.87966554301513e-06,
+      "loss": 0.2091,
+      "step": 423
+    },
+    {
+      "epoch": 1.2833900870223232,
+      "grad_norm": 0.6175480764499578,
+      "learning_rate": 6.8294613302235325e-06,
+      "loss": 0.2081,
+      "step": 424
+    },
+    {
+      "epoch": 1.2864169504351115,
+      "grad_norm": 0.5891620566570527,
+      "learning_rate": 6.779345804214088e-06,
+      "loss": 0.2163,
+      "step": 425
+    },
+    {
+      "epoch": 1.2894438138479,
+      "grad_norm": 0.5923080371579886,
+      "learning_rate": 6.729320366825785e-06,
+      "loss": 0.2187,
+      "step": 426
+    },
+    {
+      "epoch": 1.2924706772606886,
+      "grad_norm": 0.6360752767932615,
+      "learning_rate": 6.679386417377649e-06,
+      "loss": 0.2145,
+      "step": 427
+    },
+    {
+      "epoch": 1.2954975406734772,
+      "grad_norm": 0.6078055357184483,
+      "learning_rate": 6.629545352629583e-06,
+      "loss": 0.2065,
+      "step": 428
+    },
+    {
+      "epoch": 1.2985244040862656,
+      "grad_norm": 0.6343943648459572,
+      "learning_rate": 6.579798566743314e-06,
+      "loss": 0.2164,
+      "step": 429
+    },
+    {
+      "epoch": 1.3015512674990541,
+      "grad_norm": 0.5992770301648469,
+      "learning_rate": 6.530147451243377e-06,
+      "loss": 0.2211,
+      "step": 430
+    },
+    {
+      "epoch": 1.3045781309118425,
+      "grad_norm": 0.6652323333689157,
+      "learning_rate": 6.480593394978208e-06,
+      "loss": 0.2242,
+      "step": 431
+    },
+    {
+      "epoch": 1.307604994324631,
+      "grad_norm": 0.6482914904565523,
+      "learning_rate": 6.431137784081283e-06,
+      "loss": 0.2326,
+      "step": 432
+    },
+    {
+      "epoch": 1.3106318577374196,
+      "grad_norm": 0.5967757506853021,
+      "learning_rate": 6.381782001932352e-06,
+      "loss": 0.2123,
+      "step": 433
+    },
+    {
+      "epoch": 1.3136587211502082,
+      "grad_norm": 0.6130778348760161,
+      "learning_rate": 6.33252742911874e-06,
+      "loss": 0.2185,
+      "step": 434
+    },
+    {
+      "epoch": 1.3166855845629966,
+      "grad_norm": 0.630484767286309,
+      "learning_rate": 6.283375443396726e-06,
+      "loss": 0.2119,
+      "step": 435
+    },
+    {
+      "epoch": 1.3197124479757851,
+      "grad_norm": 0.640417560021759,
+      "learning_rate": 6.234327419653013e-06,
+      "loss": 0.2147,
+      "step": 436
+    },
+    {
+      "epoch": 1.3227393113885735,
+      "grad_norm": 0.879849245907959,
+      "learning_rate": 6.185384729866264e-06,
+      "loss": 0.2136,
+      "step": 437
+    },
+    {
+      "epoch": 1.325766174801362,
+      "grad_norm": 0.630318366001794,
+      "learning_rate": 6.136548743068713e-06,
+      "loss": 0.2194,
+      "step": 438
+    },
+    {
+      "epoch": 1.3287930382141506,
+      "grad_norm": 0.6534112700404463,
+      "learning_rate": 6.087820825307904e-06,
+      "loss": 0.2167,
+      "step": 439
+    },
+    {
+      "epoch": 1.3318199016269392,
+      "grad_norm": 0.6569838023197502,
+      "learning_rate": 6.039202339608432e-06,
+      "loss": 0.2234,
+      "step": 440
+    },
+    {
+      "epoch": 1.3348467650397275,
+      "grad_norm": 0.6161266440561077,
+      "learning_rate": 5.990694645933866e-06,
+      "loss": 0.2098,
+      "step": 441
+    },
+    {
+      "epoch": 1.3378736284525161,
+      "grad_norm": 0.6397601609686502,
+      "learning_rate": 5.9422991011486635e-06,
+      "loss": 0.2179,
+      "step": 442
+    },
+    {
+      "epoch": 1.3409004918653045,
+      "grad_norm": 0.6187714419658753,
+      "learning_rate": 5.894017058980249e-06,
+      "loss": 0.2157,
+      "step": 443
+    },
+    {
+      "epoch": 1.343927355278093,
+      "grad_norm": 0.665492117203472,
+      "learning_rate": 5.845849869981137e-06,
+      "loss": 0.2086,
+      "step": 444
+    },
+    {
+      "epoch": 1.3469542186908816,
+      "grad_norm": 0.6083563646079609,
+      "learning_rate": 5.797798881491138e-06,
+      "loss": 0.209,
+      "step": 445
+    },
+    {
+      "epoch": 1.3499810821036702,
+      "grad_norm": 0.5973834262456549,
+      "learning_rate": 5.749865437599703e-06,
+      "loss": 0.2155,
+      "step": 446
+    },
+    {
+      "epoch": 1.3530079455164585,
+      "grad_norm": 0.6334294877323696,
+      "learning_rate": 5.702050879108284e-06,
+      "loss": 0.21,
+      "step": 447
+    },
+    {
+      "epoch": 1.356034808929247,
+      "grad_norm": 0.6359272501252871,
+      "learning_rate": 5.654356543492883e-06,
+      "loss": 0.2043,
+      "step": 448
+    },
+    {
+      "epoch": 1.3590616723420355,
+      "grad_norm": 0.634374849207576,
+      "learning_rate": 5.606783764866576e-06,
+      "loss": 0.2211,
+      "step": 449
+    },
+    {
+      "epoch": 1.362088535754824,
+      "grad_norm": 0.5808908471624168,
+      "learning_rate": 5.559333873942259e-06,
+      "loss": 0.208,
+      "step": 450
+    },
+    {
+      "epoch": 1.3651153991676126,
+      "grad_norm": 0.6453211463051013,
+      "learning_rate": 5.512008197995379e-06,
+      "loss": 0.2218,
+      "step": 451
+    },
+    {
+      "epoch": 1.3681422625804012,
+      "grad_norm": 0.6447629961780341,
+      "learning_rate": 5.464808060826825e-06,
+      "loss": 0.2191,
+      "step": 452
+    },
+    {
+      "epoch": 1.3711691259931895,
+      "grad_norm": 0.6270632561984243,
+      "learning_rate": 5.417734782725896e-06,
+      "loss": 0.2221,
+      "step": 453
+    },
+    {
+      "epoch": 1.374195989405978,
+      "grad_norm": 0.6595164453949044,
+      "learning_rate": 5.370789680433376e-06,
+      "loss": 0.2267,
+      "step": 454
+    },
+    {
+      "epoch": 1.3772228528187664,
+      "grad_norm": 0.6097864488572073,
+      "learning_rate": 5.323974067104687e-06,
+      "loss": 0.215,
+      "step": 455
+    },
+    {
+      "epoch": 1.380249716231555,
+      "grad_norm": 0.6061420193409621,
+      "learning_rate": 5.277289252273175e-06,
+      "loss": 0.2112,
+      "step": 456
+    },
+    {
+      "epoch": 1.3832765796443436,
+      "grad_norm": 0.5792360108013481,
+      "learning_rate": 5.230736541813463e-06,
+      "loss": 0.2001,
+      "step": 457
+    },
+    {
+      "epoch": 1.3863034430571322,
+      "grad_norm": 0.5989426189945712,
+      "learning_rate": 5.184317237904939e-06,
+      "loss": 0.21,
+      "step": 458
+    },
+    {
+      "epoch": 1.3893303064699205,
+      "grad_norm": 0.633762427328295,
+      "learning_rate": 5.138032638995315e-06,
+      "loss": 0.2143,
+      "step": 459
+    },
+    {
+      "epoch": 1.392357169882709,
+      "grad_norm": 0.5663439779108818,
+      "learning_rate": 5.091884039764321e-06,
+      "loss": 0.2,
+      "step": 460
+    },
+    {
+      "epoch": 1.3953840332954974,
+      "grad_norm": 0.5909360418710811,
+      "learning_rate": 5.045872731087479e-06,
+      "loss": 0.2079,
+      "step": 461
+    },
+    {
+      "epoch": 1.398410896708286,
+      "grad_norm": 0.6374813706451291,
+      "learning_rate": 5.000000000000003e-06,
+      "loss": 0.2262,
+      "step": 462
+    },
+    {
+      "epoch": 1.4014377601210746,
+      "grad_norm": 0.5933315174862764,
+      "learning_rate": 4.954267129660789e-06,
+      "loss": 0.2042,
+      "step": 463
+    },
+    {
+      "epoch": 1.4044646235338631,
+      "grad_norm": 0.6009534692431658,
+      "learning_rate": 4.908675399316534e-06,
+      "loss": 0.2089,
+      "step": 464
+    },
+    {
+      "epoch": 1.4074914869466515,
+      "grad_norm": 0.5734827229522611,
+      "learning_rate": 4.863226084265939e-06,
+      "loss": 0.21,
+      "step": 465
+    },
+    {
+      "epoch": 1.41051835035944,
+      "grad_norm": 0.6317791609703491,
+      "learning_rate": 4.817920455824045e-06,
+      "loss": 0.2259,
+      "step": 466
+    },
+    {
+      "epoch": 1.4135452137722284,
+      "grad_norm": 0.6389425798672173,
+      "learning_rate": 4.772759781286679e-06,
+      "loss": 0.2197,
+      "step": 467
+    },
+    {
+      "epoch": 1.416572077185017,
+      "grad_norm": 0.5858669826761062,
+      "learning_rate": 4.727745323894976e-06,
+      "loss": 0.2085,
+      "step": 468
+    },
+    {
+      "epoch": 1.4195989405978056,
+      "grad_norm": 0.5472083177649537,
+      "learning_rate": 4.682878342800087e-06,
+      "loss": 0.1987,
+      "step": 469
+    },
+    {
+      "epoch": 1.4226258040105941,
+      "grad_norm": 0.5829240525304423,
+      "learning_rate": 4.638160093027908e-06,
+      "loss": 0.2083,
+      "step": 470
+    },
+    {
+      "epoch": 1.4256526674233825,
+      "grad_norm": 0.6159972666886693,
+      "learning_rate": 4.593591825444028e-06,
+      "loss": 0.2113,
+      "step": 471
+    },
+    {
+      "epoch": 1.428679530836171,
+      "grad_norm": 0.6109940422898528,
+      "learning_rate": 4.549174786718684e-06,
+      "loss": 0.2138,
+      "step": 472
+    },
+    {
+      "epoch": 1.4317063942489594,
+      "grad_norm": 0.622514409033188,
+      "learning_rate": 4.504910219291941e-06,
+      "loss": 0.2093,
+      "step": 473
+    },
+    {
+      "epoch": 1.434733257661748,
+      "grad_norm": 0.5703829713266653,
+      "learning_rate": 4.460799361338898e-06,
+      "loss": 0.2045,
+      "step": 474
+    },
+    {
+      "epoch": 1.4377601210745365,
+      "grad_norm": 0.634239481695123,
+      "learning_rate": 4.416843446735077e-06,
+      "loss": 0.2165,
+      "step": 475
+    },
+    {
+      "epoch": 1.4407869844873251,
+      "grad_norm": 0.6364801272990533,
+      "learning_rate": 4.373043705021899e-06,
+      "loss": 0.2132,
+      "step": 476
+    },
+    {
+      "epoch": 1.4438138479001135,
+      "grad_norm": 0.5960094045766494,
+      "learning_rate": 4.3294013613722944e-06,
+      "loss": 0.2117,
+      "step": 477
+    },
+    {
+      "epoch": 1.446840711312902,
+      "grad_norm": 0.6432363605525505,
+      "learning_rate": 4.2859176365564294e-06,
+      "loss": 0.2108,
+      "step": 478
+    },
+    {
+      "epoch": 1.4498675747256904,
+      "grad_norm": 0.6010139698019782,
+      "learning_rate": 4.2425937469075626e-06,
+      "loss": 0.2151,
+      "step": 479
+    },
+    {
+      "epoch": 1.452894438138479,
+      "grad_norm": 0.5929542365805702,
+      "learning_rate": 4.19943090428802e-06,
+      "loss": 0.2146,
+      "step": 480
+    },
+    {
+      "epoch": 1.4559213015512675,
+      "grad_norm": 0.6385375938797935,
+      "learning_rate": 4.1564303160552935e-06,
+      "loss": 0.2167,
+      "step": 481
+    },
+    {
+      "epoch": 1.458948164964056,
+      "grad_norm": 0.5781634332755634,
+      "learning_rate": 4.113593185028273e-06,
+      "loss": 0.1998,
+      "step": 482
+    },
+    {
+      "epoch": 1.4619750283768445,
+      "grad_norm": 0.5909222500316325,
+      "learning_rate": 4.070920709453597e-06,
+      "loss": 0.2153,
+      "step": 483
+    },
+    {
+      "epoch": 1.465001891789633,
+      "grad_norm": 0.6031188924458815,
+      "learning_rate": 4.028414082972141e-06,
+      "loss": 0.2172,
+      "step": 484
+    },
+    {
+      "epoch": 1.4680287552024214,
+      "grad_norm": 0.5890373884080109,
+      "learning_rate": 3.986074494585619e-06,
+      "loss": 0.2045,
+      "step": 485
+    },
+    {
+      "epoch": 1.47105561861521,
+      "grad_norm": 0.6276567166102123,
+      "learning_rate": 3.943903128623336e-06,
+      "loss": 0.2085,
+      "step": 486
+    },
+    {
+      "epoch": 1.4740824820279985,
+      "grad_norm": 0.6408744678823586,
+      "learning_rate": 3.9019011647090465e-06,
+      "loss": 0.2162,
+      "step": 487
+    },
+    {
+      "epoch": 1.477109345440787,
+      "grad_norm": 0.6310990927420499,
+      "learning_rate": 3.860069777727983e-06,
+      "loss": 0.2156,
+      "step": 488
+    },
+    {
+      "epoch": 1.4801362088535754,
+      "grad_norm": 0.5994673729631368,
+      "learning_rate": 3.818410137793947e-06,
+      "loss": 0.2057,
+      "step": 489
+    },
+    {
+      "epoch": 1.483163072266364,
+      "grad_norm": 0.6257810650502967,
+      "learning_rate": 3.7769234102166365e-06,
+      "loss": 0.2135,
+      "step": 490
+    },
+    {
+      "epoch": 1.4861899356791524,
+      "grad_norm": 0.6363796668404582,
+      "learning_rate": 3.735610755468988e-06,
+      "loss": 0.2109,
+      "step": 491
+    },
+    {
+      "epoch": 1.489216799091941,
+      "grad_norm": 0.6307577562828248,
+      "learning_rate": 3.6944733291547784e-06,
+      "loss": 0.2047,
+      "step": 492
+    },
+    {
+      "epoch": 1.4922436625047295,
+      "grad_norm": 0.6321682857446494,
+      "learning_rate": 3.653512281976238e-06,
+      "loss": 0.2156,
+      "step": 493
+    },
+    {
+      "epoch": 1.495270525917518,
+      "grad_norm": 0.6471324515788089,
+      "learning_rate": 3.612728759701919e-06,
+      "loss": 0.2181,
+      "step": 494
+    },
+    {
+      "epoch": 1.4982973893303064,
+      "grad_norm": 0.6008831097668246,
+      "learning_rate": 3.5721239031346067e-06,
+      "loss": 0.2167,
+      "step": 495
+    },
+    {
+      "epoch": 1.501324252743095,
+      "grad_norm": 0.612154069380997,
+      "learning_rate": 3.5316988480794255e-06,
+      "loss": 0.1963,
+      "step": 496
+    },
+    {
+      "epoch": 1.5043511161558833,
+      "grad_norm": 0.660532802741499,
+      "learning_rate": 3.4914547253120655e-06,
+      "loss": 0.2122,
+      "step": 497
+    },
+    {
+      "epoch": 1.507377979568672,
+      "grad_norm": 0.5837573349390066,
+      "learning_rate": 3.4513926605471504e-06,
+      "loss": 0.1943,
+      "step": 498
+    },
+    {
+      "epoch": 1.5104048429814605,
+      "grad_norm": 0.6243373552042033,
+      "learning_rate": 3.4115137744067516e-06,
+      "loss": 0.2149,
+      "step": 499
+    },
+    {
+      "epoch": 1.513431706394249,
+      "grad_norm": 0.6123443982882361,
+      "learning_rate": 3.37181918238904e-06,
+      "loss": 0.2183,
+      "step": 500
+    },
+    {
+      "epoch": 1.5164585698070374,
+      "grad_norm": 0.595278172088759,
+      "learning_rate": 3.3323099948370853e-06,
+      "loss": 0.2049,
+      "step": 501
+    },
+    {
+      "epoch": 1.519485433219826,
+      "grad_norm": 0.5997382845622334,
+      "learning_rate": 3.292987316907792e-06,
+      "loss": 0.2043,
+      "step": 502
+    },
+    {
+      "epoch": 1.5225122966326143,
+      "grad_norm": 0.5953543829179927,
+      "learning_rate": 3.253852248540994e-06,
+      "loss": 0.2135,
+      "step": 503
+    },
+    {
+      "epoch": 1.525539160045403,
+      "grad_norm": 0.6513002085255232,
+      "learning_rate": 3.2149058844286796e-06,
+      "loss": 0.2182,
+      "step": 504
+    },
+    {
+      "epoch": 1.5285660234581915,
+      "grad_norm": 0.6033294685777496,
+      "learning_rate": 3.1761493139843734e-06,
+      "loss": 0.2058,
+      "step": 505
+    },
+    {
+      "epoch": 1.53159288687098,
+      "grad_norm": 0.5643099983712151,
+      "learning_rate": 3.1375836213126653e-06,
+      "loss": 0.2035,
+      "step": 506
+    },
+    {
+      "epoch": 1.5346197502837684,
+      "grad_norm": 0.6211580795036665,
+      "learning_rate": 3.099209885178882e-06,
+      "loss": 0.2157,
+      "step": 507
+    },
+    {
+      "epoch": 1.537646613696557,
+      "grad_norm": 0.5868472013643298,
+      "learning_rate": 3.0610291789789094e-06,
+      "loss": 0.2002,
+      "step": 508
+    },
+    {
+      "epoch": 1.5406734771093453,
+      "grad_norm": 0.5780562693591113,
+      "learning_rate": 3.023042570709185e-06,
+      "loss": 0.1997,
+      "step": 509
+    },
+    {
+      "epoch": 1.543700340522134,
+      "grad_norm": 0.6137737194095165,
+      "learning_rate": 2.9852511229367862e-06,
+      "loss": 0.1998,
+      "step": 510
+    },
+    {
+      "epoch": 1.5467272039349225,
+      "grad_norm": 0.6102220042354304,
+      "learning_rate": 2.9476558927697605e-06,
+      "loss": 0.2004,
+      "step": 511
+    },
+    {
+      "epoch": 1.549754067347711,
+      "grad_norm": 0.6366583771438068,
+      "learning_rate": 2.9102579318274994e-06,
+      "loss": 0.1985,
+      "step": 512
+    },
+    {
+      "epoch": 1.5527809307604994,
+      "grad_norm": 0.6080240108336545,
+      "learning_rate": 2.8730582862113743e-06,
+      "loss": 0.2034,
+      "step": 513
+    },
+    {
+      "epoch": 1.555807794173288,
+      "grad_norm": 0.6666262431691008,
+      "learning_rate": 2.8360579964754277e-06,
+      "loss": 0.2201,
+      "step": 514
+    },
+    {
+      "epoch": 1.5588346575860763,
+      "grad_norm": 0.6360139401247478,
+      "learning_rate": 2.7992580975973136e-06,
+      "loss": 0.2063,
+      "step": 515
+    },
+    {
+      "epoch": 1.5618615209988649,
+      "grad_norm": 0.6701996649234718,
+      "learning_rate": 2.7626596189492983e-06,
+      "loss": 0.2169,
+      "step": 516
+    },
+    {
+      "epoch": 1.5648883844116535,
+      "grad_norm": 0.6615426716305579,
+      "learning_rate": 2.726263584269513e-06,
+      "loss": 0.209,
+      "step": 517
+    },
+    {
+      "epoch": 1.567915247824442,
+      "grad_norm": 0.6398639227020508,
+      "learning_rate": 2.690071011633284e-06,
+      "loss": 0.2138,
+      "step": 518
+    },
+    {
+      "epoch": 1.5709421112372304,
+      "grad_norm": 0.6221144917583517,
+      "learning_rate": 2.6540829134246683e-06,
+      "loss": 0.2162,
+      "step": 519
+    },
+    {
+      "epoch": 1.573968974650019,
+      "grad_norm": 0.5938391896611732,
+      "learning_rate": 2.618300296308135e-06,
+      "loss": 0.2096,
+      "step": 520
+    },
+    {
+      "epoch": 1.5769958380628073,
+      "grad_norm": 0.576453997079251,
+      "learning_rate": 2.582724161200405e-06,
+      "loss": 0.2054,
+      "step": 521
+    },
+    {
+      "epoch": 1.5800227014755959,
+      "grad_norm": 0.6775795311296927,
+      "learning_rate": 2.5473555032424534e-06,
+      "loss": 0.216,
+      "step": 522
+    },
+    {
+      "epoch": 1.5830495648883844,
+      "grad_norm": 0.6342076995377148,
+      "learning_rate": 2.5121953117716744e-06,
+      "loss": 0.1985,
+      "step": 523
+    },
+    {
+      "epoch": 1.586076428301173,
+      "grad_norm": 0.639354001743106,
+      "learning_rate": 2.477244570294206e-06,
+      "loss": 0.2293,
+      "step": 524
+    },
+    {
+      "epoch": 1.5891032917139614,
+      "grad_norm": 0.635272389526875,
+      "learning_rate": 2.4425042564574186e-06,
+      "loss": 0.214,
+      "step": 525
+    },
+    {
+      "epoch": 1.59213015512675,
+      "grad_norm": 0.5933378765217413,
+      "learning_rate": 2.4079753420225694e-06,
+      "loss": 0.2092,
+      "step": 526
+    },
+    {
+      "epoch": 1.5951570185395383,
+      "grad_norm": 0.634801017529104,
+      "learning_rate": 2.3736587928376197e-06,
+      "loss": 0.2094,
+      "step": 527
+    },
+    {
+      "epoch": 1.5981838819523269,
+      "grad_norm": 0.613549927685821,
+      "learning_rate": 2.339555568810221e-06,
+      "loss": 0.2016,
+      "step": 528
+    },
+    {
+      "epoch": 1.6012107453651154,
+      "grad_norm": 0.6550688860964111,
+      "learning_rate": 2.305666623880858e-06,
+      "loss": 0.2165,
+      "step": 529
+    },
+    {
+      "epoch": 1.604237608777904,
+      "grad_norm": 0.6055231898950925,
+      "learning_rate": 2.27199290599617e-06,
+      "loss": 0.2067,
+      "step": 530
+    },
+    {
+      "epoch": 1.6072644721906924,
+      "grad_norm": 0.6744807855364295,
+      "learning_rate": 2.2385353570824308e-06,
+      "loss": 0.215,
+      "step": 531
+    },
+    {
+      "epoch": 1.610291335603481,
+      "grad_norm": 0.6666427950126947,
+      "learning_rate": 2.2052949130192136e-06,
+      "loss": 0.2163,
+      "step": 532
+    },
+    {
+      "epoch": 1.6133181990162693,
+      "grad_norm": 0.5732938006169825,
+      "learning_rate": 2.172272503613183e-06,
+      "loss": 0.1935,
+      "step": 533
+    },
+    {
+      "epoch": 1.6163450624290578,
+      "grad_norm": 0.6869210108593815,
+      "learning_rate": 2.1394690525721275e-06,
+      "loss": 0.2207,
+      "step": 534
+    },
+    {
+      "epoch": 1.6193719258418464,
+      "grad_norm": 0.6048773194396152,
+      "learning_rate": 2.1068854774790783e-06,
+      "loss": 0.2085,
+      "step": 535
+    },
+    {
+      "epoch": 1.622398789254635,
+      "grad_norm": 0.6442339930934871,
+      "learning_rate": 2.0745226897666858e-06,
+      "loss": 0.2082,
+      "step": 536
+    },
+    {
+      "epoch": 1.6254256526674233,
+      "grad_norm": 0.6102637623940661,
+      "learning_rate": 2.0423815946916783e-06,
+      "loss": 0.2039,
+      "step": 537
+    },
+    {
+      "epoch": 1.628452516080212,
+      "grad_norm": 0.617039348177892,
+      "learning_rate": 2.010463091309587e-06,
+      "loss": 0.2073,
+      "step": 538
+    },
+    {
+      "epoch": 1.6314793794930003,
+      "grad_norm": 0.5996746429602261,
+      "learning_rate": 1.9787680724495617e-06,
+      "loss": 0.2074,
+      "step": 539
+    },
+    {
+      "epoch": 1.6345062429057888,
+      "grad_norm": 0.5762361879811994,
+      "learning_rate": 1.947297424689414e-06,
+      "loss": 0.2062,
+      "step": 540
+    },
+    {
+      "epoch": 1.6375331063185774,
+      "grad_norm": 0.5766804929625047,
+      "learning_rate": 1.9160520283308115e-06,
+      "loss": 0.2045,
+      "step": 541
+    },
+    {
+      "epoch": 1.640559969731366,
+      "grad_norm": 0.6236267504413758,
+      "learning_rate": 1.8850327573746584e-06,
+      "loss": 0.2138,
+      "step": 542
+    },
+    {
+      "epoch": 1.6435868331441543,
+      "grad_norm": 0.6000173573638069,
+      "learning_rate": 1.854240479496643e-06,
+      "loss": 0.2061,
+      "step": 543
+    },
+    {
+      "epoch": 1.646613696556943,
+      "grad_norm": 0.6315427983778534,
+      "learning_rate": 1.8236760560229715e-06,
+      "loss": 0.2196,
+      "step": 544
+    },
+    {
+      "epoch": 1.6496405599697312,
+      "grad_norm": 0.5729840191367517,
+      "learning_rate": 1.7933403419062689e-06,
+      "loss": 0.1996,
+      "step": 545
+    },
+    {
+      "epoch": 1.6526674233825198,
+      "grad_norm": 0.6077565912593988,
+      "learning_rate": 1.7632341857016733e-06,
+      "loss": 0.2039,
+      "step": 546
+    },
+    {
+      "epoch": 1.6556942867953084,
+      "grad_norm": 0.6053281663293922,
+      "learning_rate": 1.7333584295430894e-06,
+      "loss": 0.2161,
+      "step": 547
+    },
+    {
+      "epoch": 1.658721150208097,
+      "grad_norm": 0.5777645251848361,
+      "learning_rate": 1.7037139091196396e-06,
+      "loss": 0.1987,
+      "step": 548
+    },
+    {
+      "epoch": 1.6617480136208853,
+      "grad_norm": 0.670792755703484,
+      "learning_rate": 1.6743014536522872e-06,
+      "loss": 0.2141,
+      "step": 549
+    },
+    {
+      "epoch": 1.6647748770336739,
+      "grad_norm": 0.6346567439018682,
+      "learning_rate": 1.6451218858706374e-06,
+      "loss": 0.2146,
+      "step": 550
+    },
+    {
+      "epoch": 1.6678017404464622,
+      "grad_norm": 0.6036394029366919,
+      "learning_rate": 1.616176021989926e-06,
+      "loss": 0.2014,
+      "step": 551
+    },
+    {
+      "epoch": 1.6708286038592508,
+      "grad_norm": 0.5980070253481372,
+      "learning_rate": 1.587464671688187e-06,
+      "loss": 0.2063,
+      "step": 552
+    },
+    {
+      "epoch": 1.6738554672720394,
+      "grad_norm": 0.6094750555328841,
+      "learning_rate": 1.558988638083616e-06,
+      "loss": 0.2057,
+      "step": 553
+    },
+    {
+      "epoch": 1.676882330684828,
+      "grad_norm": 0.6335747363844814,
+      "learning_rate": 1.5307487177120773e-06,
+      "loss": 0.2172,
+      "step": 554
+    },
+    {
+      "epoch": 1.6799091940976163,
+      "grad_norm": 0.6213945696513632,
+      "learning_rate": 1.5027457005048573e-06,
+      "loss": 0.2194,
+      "step": 555
+    },
+    {
+      "epoch": 1.6829360575104049,
+      "grad_norm": 0.6355218877808894,
+      "learning_rate": 1.4749803697665366e-06,
+      "loss": 0.2135,
+      "step": 556
+    },
+    {
+      "epoch": 1.6859629209231932,
+      "grad_norm": 0.5929644274617915,
+      "learning_rate": 1.4474535021531099e-06,
+      "loss": 0.2123,
+      "step": 557
+    },
+    {
+      "epoch": 1.6889897843359818,
+      "grad_norm": 0.6040263654664385,
+      "learning_rate": 1.4201658676502294e-06,
+      "loss": 0.2042,
+      "step": 558
+    },
+    {
+      "epoch": 1.6920166477487704,
+      "grad_norm": 0.6166211783787556,
+      "learning_rate": 1.3931182295516965e-06,
+      "loss": 0.2094,
+      "step": 559
+    },
+    {
+      "epoch": 1.695043511161559,
+      "grad_norm": 0.5921617569024974,
+      "learning_rate": 1.3663113444380905e-06,
+      "loss": 0.2004,
+      "step": 560
+    },
+    {
+      "epoch": 1.6980703745743473,
+      "grad_norm": 0.6110475895364043,
+      "learning_rate": 1.339745962155613e-06,
+      "loss": 0.2137,
+      "step": 561
+    },
+    {
+      "epoch": 1.7010972379871359,
+      "grad_norm": 0.5795636437533754,
+      "learning_rate": 1.3134228257951142e-06,
+      "loss": 0.2069,
+      "step": 562
+    },
+    {
+      "epoch": 1.7041241013999242,
+      "grad_norm": 0.5913111670561687,
+      "learning_rate": 1.2873426716713012e-06,
+      "loss": 0.2026,
+      "step": 563
+    },
+    {
+      "epoch": 1.7071509648127128,
+      "grad_norm": 0.6141655247104946,
+      "learning_rate": 1.2615062293021508e-06,
+      "loss": 0.2046,
+      "step": 564
+    },
+    {
+      "epoch": 1.7101778282255014,
+      "grad_norm": 0.6122094891939365,
+      "learning_rate": 1.2359142213884933e-06,
+      "loss": 0.2108,
+      "step": 565
+    },
+    {
+      "epoch": 1.71320469163829,
+      "grad_norm": 0.5880250702701763,
+      "learning_rate": 1.2105673637938054e-06,
+      "loss": 0.2061,
+      "step": 566
+    },
+    {
+      "epoch": 1.7162315550510783,
+      "grad_norm": 0.644802029777015,
+      "learning_rate": 1.1854663655241804e-06,
+      "loss": 0.2092,
+      "step": 567
+    },
+    {
+      "epoch": 1.7192584184638668,
+      "grad_norm": 0.5933224978042897,
+      "learning_rate": 1.1606119287084982e-06,
+      "loss": 0.2018,
+      "step": 568
+    },
+    {
+      "epoch": 1.7222852818766552,
+      "grad_norm": 0.6075772552462014,
+      "learning_rate": 1.136004748578785e-06,
+      "loss": 0.2059,
+      "step": 569
+    },
+    {
+      "epoch": 1.7253121452894438,
+      "grad_norm": 0.6206025213226827,
+      "learning_rate": 1.1116455134507665e-06,
+      "loss": 0.216,
+      "step": 570
+    },
+    {
+      "epoch": 1.7283390087022323,
+      "grad_norm": 0.6058021093887288,
+      "learning_rate": 1.0875349047046113e-06,
+      "loss": 0.2004,
+      "step": 571
+    },
+    {
+      "epoch": 1.731365872115021,
+      "grad_norm": 0.625312425810076,
+      "learning_rate": 1.0636735967658785e-06,
+      "loss": 0.2111,
+      "step": 572
+    },
+    {
+      "epoch": 1.7343927355278093,
+      "grad_norm": 0.6240446307599206,
+      "learning_rate": 1.0400622570866426e-06,
+      "loss": 0.2072,
+      "step": 573
+    },
+    {
+      "epoch": 1.7374195989405978,
+      "grad_norm": 0.5745951338884003,
+      "learning_rate": 1.0167015461268303e-06,
+      "loss": 0.2012,
+      "step": 574
+    },
+    {
+      "epoch": 1.7404464623533862,
+      "grad_norm": 0.5893090334077228,
+      "learning_rate": 9.935921173357444e-07,
+      "loss": 0.2034,
+      "step": 575
+    },
+    {
+      "epoch": 1.7434733257661748,
+      "grad_norm": 0.5849350990457602,
+      "learning_rate": 9.707346171337895e-07,
+      "loss": 0.1994,
+      "step": 576
+    },
+    {
+      "epoch": 1.7465001891789633,
+      "grad_norm": 0.5900050093321753,
+      "learning_rate": 9.481296848943744e-07,
+      "loss": 0.2008,
+      "step": 577
+    },
+    {
+      "epoch": 1.749527052591752,
+      "grad_norm": 0.596520061463495,
+      "learning_rate": 9.257779529260558e-07,
+      "loss": 0.196,
+      "step": 578
+    },
+    {
+      "epoch": 1.7525539160045402,
+      "grad_norm": 0.6368089885606508,
+      "learning_rate": 9.036800464548157e-07,
+      "loss": 0.2205,
+      "step": 579
+    },
+    {
+      "epoch": 1.7555807794173288,
+      "grad_norm": 0.6130352673108144,
+      "learning_rate": 8.818365836066101e-07,
+      "loss": 0.2029,
+      "step": 580
+    },
+    {
+      "epoch": 1.7586076428301172,
+      "grad_norm": 0.6100753262929383,
+      "learning_rate": 8.602481753900427e-07,
+      "loss": 0.2122,
+      "step": 581
+    },
+    {
+      "epoch": 1.7616345062429057,
+      "grad_norm": 0.6107786951452405,
+      "learning_rate": 8.389154256793042e-07,
+      "loss": 0.2114,
+      "step": 582
+    },
+    {
+      "epoch": 1.7646613696556943,
+      "grad_norm": 0.5741030719214796,
+      "learning_rate": 8.178389311972612e-07,
+      "loss": 0.2014,
+      "step": 583
+    },
+    {
+      "epoch": 1.7676882330684829,
+      "grad_norm": 0.5763837222640515,
+      "learning_rate": 7.970192814987676e-07,
+      "loss": 0.1972,
+      "step": 584
+    },
+    {
+      "epoch": 1.7707150964812712,
+      "grad_norm": 0.5869325520805518,
+      "learning_rate": 7.764570589541876e-07,
+      "loss": 0.2052,
+      "step": 585
+    },
+    {
+      "epoch": 1.7737419598940598,
+      "grad_norm": 0.5869445159167366,
+      "learning_rate": 7.561528387330797e-07,
+      "loss": 0.1953,
+      "step": 586
+    },
+    {
+      "epoch": 1.7767688233068482,
+      "grad_norm": 0.5879475383420326,
+      "learning_rate": 7.361071887881376e-07,
+      "loss": 0.1974,
+      "step": 587
+    },
+    {
+      "epoch": 1.7797956867196367,
+      "grad_norm": 0.5824738421859162,
+      "learning_rate": 7.163206698392744e-07,
+      "loss": 0.2026,
+      "step": 588
+    },
+    {
+      "epoch": 1.7828225501324253,
+      "grad_norm": 0.5814748278065084,
+      "learning_rate": 6.96793835357964e-07,
+      "loss": 0.1973,
+      "step": 589
+    },
+    {
+      "epoch": 1.7858494135452139,
+      "grad_norm": 0.6059991877260993,
+      "learning_rate": 6.775272315517423e-07,
+      "loss": 0.2112,
+      "step": 590
+    },
+    {
+      "epoch": 1.7888762769580022,
+      "grad_norm": 0.5939172412365742,
+      "learning_rate": 6.585213973489335e-07,
+      "loss": 0.2089,
+      "step": 591
+    },
+    {
+      "epoch": 1.7919031403707908,
+      "grad_norm": 0.6426668357169981,
+      "learning_rate": 6.397768643835755e-07,
+      "loss": 0.2159,
+      "step": 592
+    },
+    {
+      "epoch": 1.7949300037835791,
+      "grad_norm": 0.5890635703258731,
+      "learning_rate": 6.212941569805508e-07,
+      "loss": 0.202,
+      "step": 593
+    },
+    {
+      "epoch": 1.7979568671963677,
+      "grad_norm": 0.5900350624698916,
+      "learning_rate": 6.030737921409169e-07,
+      "loss": 0.1967,
+      "step": 594
+    },
+    {
+      "epoch": 1.8009837306091563,
+      "grad_norm": 0.6133680048354371,
+      "learning_rate": 5.851162795274445e-07,
+      "loss": 0.2152,
+      "step": 595
+    },
+    {
+      "epoch": 1.8040105940219449,
+      "grad_norm": 0.6252876382778517,
+      "learning_rate": 5.674221214503639e-07,
+      "loss": 0.2157,
+      "step": 596
+    },
+    {
+      "epoch": 1.8070374574347332,
+      "grad_norm": 0.5921376196093544,
+      "learning_rate": 5.499918128533155e-07,
+      "loss": 0.1996,
+      "step": 597
+    },
+    {
+      "epoch": 1.8100643208475218,
+      "grad_norm": 0.6035205775752059,
+      "learning_rate": 5.328258412994958e-07,
+      "loss": 0.2016,
+      "step": 598
+    },
+    {
+      "epoch": 1.8130911842603101,
+      "grad_norm": 0.6107857188805569,
+      "learning_rate": 5.159246869580348e-07,
+      "loss": 0.2121,
+      "step": 599
+    },
+    {
+      "epoch": 1.8161180476730987,
+      "grad_norm": 0.629504837680152,
+      "learning_rate": 4.992888225905467e-07,
+      "loss": 0.2097,
+      "step": 600
+    },
+    {
+      "epoch": 1.8191449110858873,
+      "grad_norm": 0.6180268901785609,
+      "learning_rate": 4.829187135379221e-07,
+      "loss": 0.2148,
+      "step": 601
+    },
+    {
+      "epoch": 1.8221717744986758,
+      "grad_norm": 0.5962090599473324,
+      "learning_rate": 4.6681481770729844e-07,
+      "loss": 0.1997,
+      "step": 602
+    },
+    {
+      "epoch": 1.8251986379114642,
+      "grad_norm": 0.5624363097348722,
+      "learning_rate": 4.509775855592613e-07,
+      "loss": 0.2095,
+      "step": 603
+    },
+    {
+      "epoch": 1.8282255013242528,
+      "grad_norm": 0.595388708993232,
+      "learning_rate": 4.354074600952407e-07,
+      "loss": 0.1962,
+      "step": 604
+    },
+    {
+      "epoch": 1.8312523647370411,
+      "grad_norm": 0.576884201221483,
+      "learning_rate": 4.2010487684511105e-07,
+      "loss": 0.1988,
+      "step": 605
+    },
+    {
+      "epoch": 1.8342792281498297,
+      "grad_norm": 0.5882149891593514,
+      "learning_rate": 4.0507026385502747e-07,
+      "loss": 0.1977,
+      "step": 606
+    },
+    {
+      "epoch": 1.8373060915626183,
+      "grad_norm": 0.6118041726885657,
+      "learning_rate": 3.9030404167542777e-07,
+      "loss": 0.2017,
+      "step": 607
+    },
+    {
+      "epoch": 1.8403329549754068,
+      "grad_norm": 0.6063629697580617,
+      "learning_rate": 3.7580662334929517e-07,
+      "loss": 0.1986,
+      "step": 608
+    },
+    {
+      "epoch": 1.8433598183881952,
+      "grad_norm": 0.5639852137423518,
+      "learning_rate": 3.615784144005796e-07,
+      "loss": 0.1957,
+      "step": 609
+    },
+    {
+      "epoch": 1.8463866818009838,
+      "grad_norm": 0.5874234200187757,
+      "learning_rate": 3.476198128228736e-07,
+      "loss": 0.2,
+      "step": 610
+    },
+    {
+      "epoch": 1.849413545213772,
+      "grad_norm": 0.5690860795643277,
+      "learning_rate": 3.339312090682689e-07,
+      "loss": 0.195,
+      "step": 611
+    },
+    {
+      "epoch": 1.8524404086265607,
+      "grad_norm": 0.574629611710639,
+      "learning_rate": 3.2051298603643754e-07,
+      "loss": 0.2082,
+      "step": 612
+    },
+    {
+      "epoch": 1.8554672720393492,
+      "grad_norm": 0.628760829657872,
+      "learning_rate": 3.0736551906392354e-07,
+      "loss": 0.2164,
+      "step": 613
+    },
+    {
+      "epoch": 1.8584941354521378,
+      "grad_norm": 0.6108390153983083,
+      "learning_rate": 2.9448917591363923e-07,
+      "loss": 0.2046,
+      "step": 614
+    },
+    {
+      "epoch": 1.8615209988649262,
+      "grad_norm": 0.5732909690515572,
+      "learning_rate": 2.818843167645835e-07,
+      "loss": 0.2084,
+      "step": 615
+    },
+    {
+      "epoch": 1.8645478622777147,
+      "grad_norm": 0.6246006616107485,
+      "learning_rate": 2.6955129420176193e-07,
+      "loss": 0.212,
+      "step": 616
+    },
+    {
+      "epoch": 1.867574725690503,
+      "grad_norm": 0.5784429760575157,
+      "learning_rate": 2.5749045320632824e-07,
+      "loss": 0.2062,
+      "step": 617
+    },
+    {
+      "epoch": 1.8706015891032917,
+      "grad_norm": 0.6047491468244323,
+      "learning_rate": 2.4570213114592957e-07,
+      "loss": 0.1994,
+      "step": 618
+    },
+    {
+      "epoch": 1.8736284525160802,
+      "grad_norm": 0.5744051008957742,
+      "learning_rate": 2.3418665776527738e-07,
+      "loss": 0.1985,
+      "step": 619
+    },
+    {
+      "epoch": 1.8766553159288688,
+      "grad_norm": 0.6070000023909466,
+      "learning_rate": 2.2294435517691504e-07,
+      "loss": 0.2006,
+      "step": 620
+    },
+    {
+      "epoch": 1.8796821793416572,
+      "grad_norm": 0.5656284648850038,
+      "learning_rate": 2.119755378522137e-07,
+      "loss": 0.1979,
+      "step": 621
+    },
+    {
+      "epoch": 1.8827090427544457,
+      "grad_norm": 0.6073188639432301,
+      "learning_rate": 2.0128051261257165e-07,
+      "loss": 0.2146,
+      "step": 622
+    },
+    {
+      "epoch": 1.885735906167234,
+      "grad_norm": 0.5761872161430275,
+      "learning_rate": 1.908595786208367e-07,
+      "loss": 0.2057,
+      "step": 623
+    },
+    {
+      "epoch": 1.8887627695800226,
+      "grad_norm": 0.5873650831121161,
+      "learning_rate": 1.8071302737293294e-07,
+      "loss": 0.2118,
+      "step": 624
+    },
+    {
+      "epoch": 1.8917896329928112,
+      "grad_norm": 0.6067593337449608,
+      "learning_rate": 1.7084114268971275e-07,
+      "loss": 0.2085,
+      "step": 625
+    },
+    {
+      "epoch": 1.8948164964055998,
+      "grad_norm": 0.6343981264358917,
+      "learning_rate": 1.612442007090076e-07,
+      "loss": 0.2119,
+      "step": 626
+    },
+    {
+      "epoch": 1.8978433598183881,
+      "grad_norm": 0.6224948216597962,
+      "learning_rate": 1.519224698779198e-07,
+      "loss": 0.2113,
+      "step": 627
+    },
+    {
+      "epoch": 1.9008702232311767,
+      "grad_norm": 0.6111312319905354,
+      "learning_rate": 1.4287621094529524e-07,
+      "loss": 0.2088,
+      "step": 628
+    },
+    {
+      "epoch": 1.903897086643965,
+      "grad_norm": 0.5764196204629675,
+      "learning_rate": 1.3410567695444576e-07,
+      "loss": 0.1983,
+      "step": 629
+    },
+    {
+      "epoch": 1.9069239500567536,
+      "grad_norm": 0.5961296574291337,
+      "learning_rate": 1.2561111323605714e-07,
+      "loss": 0.2059,
+      "step": 630
+    },
+    {
+      "epoch": 1.9099508134695422,
+      "grad_norm": 0.6068625811691762,
+      "learning_rate": 1.1739275740134004e-07,
+      "loss": 0.2057,
+      "step": 631
+    },
+    {
+      "epoch": 1.9129776768823308,
+      "grad_norm": 0.5993042320432096,
+      "learning_rate": 1.0945083933537104e-07,
+      "loss": 0.2039,
+      "step": 632
+    },
+    {
+      "epoch": 1.9160045402951191,
+      "grad_norm": 0.6387504804647781,
+      "learning_rate": 1.0178558119067316e-07,
+      "loss": 0.221,
+      "step": 633
+    },
+    {
+      "epoch": 1.9190314037079077,
+      "grad_norm": 0.5681921721141407,
+      "learning_rate": 9.439719738099318e-08,
+      "loss": 0.192,
+      "step": 634
+    },
+    {
+      "epoch": 1.922058267120696,
+      "grad_norm": 0.6047463401251821,
+      "learning_rate": 8.728589457530857e-08,
+      "loss": 0.2013,
+      "step": 635
+    },
+    {
+      "epoch": 1.9250851305334846,
+      "grad_norm": 0.5783425166578283,
+      "learning_rate": 8.04518716920466e-08,
+      "loss": 0.1973,
+      "step": 636
+    },
+    {
+      "epoch": 1.9281119939462732,
+      "grad_norm": 0.5430380101677842,
+      "learning_rate": 7.389531989351773e-08,
+      "loss": 0.1937,
+      "step": 637
+    },
+    {
+      "epoch": 1.9311388573590618,
+      "grad_norm": 0.6167999118950785,
+      "learning_rate": 6.761642258056977e-08,
+      "loss": 0.2157,
+      "step": 638
+    },
+    {
+      "epoch": 1.9341657207718501,
+      "grad_norm": 0.6019257406342596,
+      "learning_rate": 6.161535538745877e-08,
+      "loss": 0.2017,
+      "step": 639
+    },
+    {
+      "epoch": 1.9371925841846387,
+      "grad_norm": 0.5714231788407736,
+      "learning_rate": 5.5892286176932875e-08,
+      "loss": 0.1948,
+      "step": 640
+    },
+    {
+      "epoch": 1.940219447597427,
+      "grad_norm": 0.5842245838860273,
+      "learning_rate": 5.044737503554165e-08,
+      "loss": 0.2017,
+      "step": 641
+    },
+    {
+      "epoch": 1.9432463110102156,
+      "grad_norm": 0.5822154323575391,
+      "learning_rate": 4.528077426915412e-08,
+      "loss": 0.1957,
+      "step": 642
+    },
+    {
+      "epoch": 1.9462731744230042,
+      "grad_norm": 0.5908234581426911,
+      "learning_rate": 4.0392628398699954e-08,
+      "loss": 0.2035,
+      "step": 643
+    },
+    {
+      "epoch": 1.9493000378357928,
+      "grad_norm": 0.5988708459672253,
+      "learning_rate": 3.578307415612714e-08,
+      "loss": 0.2083,
+      "step": 644
+    },
+    {
+      "epoch": 1.952326901248581,
+      "grad_norm": 0.5797130368239053,
+      "learning_rate": 3.1452240480577265e-08,
+      "loss": 0.2043,
+      "step": 645
+    },
+    {
+      "epoch": 1.9553537646613697,
+      "grad_norm": 0.5627873509668126,
+      "learning_rate": 2.7400248514776184e-08,
+      "loss": 0.2022,
+      "step": 646
+    },
+    {
+      "epoch": 1.958380628074158,
+      "grad_norm": 0.615955080897471,
+      "learning_rate": 2.3627211601651157e-08,
+      "loss": 0.2194,
+      "step": 647
+    },
+    {
+      "epoch": 1.9614074914869466,
+      "grad_norm": 0.5730235525716271,
+      "learning_rate": 2.013323528115674e-08,
+      "loss": 0.2026,
+      "step": 648
+    },
+    {
+      "epoch": 1.9644343548997352,
+      "grad_norm": 0.5935165715416985,
+      "learning_rate": 1.6918417287318245e-08,
+      "loss": 0.2001,
+      "step": 649
+    },
+    {
+      "epoch": 1.9674612183125237,
+      "grad_norm": 0.5627318421892451,
+      "learning_rate": 1.3982847545507271e-08,
+      "loss": 0.1974,
+      "step": 650
+    },
+    {
+      "epoch": 1.970488081725312,
+      "grad_norm": 0.5884315952389003,
+      "learning_rate": 1.1326608169920373e-08,
+      "loss": 0.1996,
+      "step": 651
+    },
+    {
+      "epoch": 1.9735149451381007,
+      "grad_norm": 0.6042671500832755,
+      "learning_rate": 8.949773461282008e-09,
+      "loss": 0.206,
+      "step": 652
+    },
+    {
+      "epoch": 1.976541808550889,
+      "grad_norm": 0.5970847174220899,
+      "learning_rate": 6.8524099047695415e-09,
+      "loss": 0.199,
+      "step": 653
+    },
+    {
+      "epoch": 1.9795686719636776,
+      "grad_norm": 0.5778936302284102,
+      "learning_rate": 5.034576168149175e-09,
+      "loss": 0.1928,
+      "step": 654
+    },
+    {
+      "epoch": 1.9825955353764662,
+      "grad_norm": 0.5715405367701467,
+      "learning_rate": 3.4963231001383657e-09,
+      "loss": 0.2047,
+      "step": 655
+    },
+    {
+      "epoch": 1.9856223987892547,
+      "grad_norm": 0.6024335813971338,
+      "learning_rate": 2.237693728981416e-09,
+      "loss": 0.2118,
+      "step": 656
+    },
+    {
+      "epoch": 1.988649262202043,
+      "grad_norm": 0.6439076789899352,
+      "learning_rate": 1.2587232612493172e-09,
+      "loss": 0.2183,
+      "step": 657
+    },
+    {
+      "epoch": 1.9916761256148316,
+      "grad_norm": 0.6272109183910406,
+      "learning_rate": 5.594390808494332e-10,
+      "loss": 0.2015,
+      "step": 658
+    },
+    {
+      "epoch": 1.99470298902762,
+      "grad_norm": 0.5967013837844725,
+      "learning_rate": 1.3986074826388697e-10,
+      "loss": 0.2071,
+      "step": 659
+    },
+    {
+      "epoch": 1.9977298524404086,
+      "grad_norm": 0.591545841659203,
+      "learning_rate": 0.0,
+      "loss": 0.2065,
+      "step": 660
+    },
+    {
+      "epoch": 1.9977298524404086,
+      "step": 660,
+      "total_flos": 123862839459840.0,
+      "train_loss": 0.29154424624461117,
+      "train_runtime": 2690.8911,
+      "train_samples_per_second": 31.42,
+      "train_steps_per_second": 0.245
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 660,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "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": 123862839459840.0,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}