diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,7371 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.0013828289519605221,
+  "eval_steps": 500,
+  "global_step": 1050,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.7606893775878203,
+      "learning_rate": 1.9987137664166173e-07,
+      "loss": 1.2661,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6309869956147725,
+      "learning_rate": 3.9974275328332345e-07,
+      "loss": 1.0203,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6659301879012498,
+      "learning_rate": 5.996141299249851e-07,
+      "loss": 1.0215,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6033700558639792,
+      "learning_rate": 7.994855065666469e-07,
+      "loss": 1.2324,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.418739799852574,
+      "learning_rate": 9.993568832083086e-07,
+      "loss": 1.2197,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.029142019551676,
+      "learning_rate": 1.1992282598499703e-06,
+      "loss": 1.2627,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.531594763110282,
+      "learning_rate": 1.3990996364916321e-06,
+      "loss": 0.9839,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6305178347457068,
+      "learning_rate": 1.5989710131332938e-06,
+      "loss": 0.9658,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6595041228378495,
+      "learning_rate": 1.7988423897749555e-06,
+      "loss": 1.04,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.7047877463175605,
+      "learning_rate": 1.998713766416617e-06,
+      "loss": 1.2061,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5867683977617453,
+      "learning_rate": 2.198585143058279e-06,
+      "loss": 1.127,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3194737224788275,
+      "learning_rate": 2.3984565196999405e-06,
+      "loss": 0.9258,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4926511833003497,
+      "learning_rate": 2.5983278963416026e-06,
+      "loss": 0.9429,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6183620002609274,
+      "learning_rate": 2.7981992729832643e-06,
+      "loss": 0.8943,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.366962654153374,
+      "learning_rate": 2.998070649624926e-06,
+      "loss": 1.2217,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.9560812587904448,
+      "learning_rate": 3.1979420262665876e-06,
+      "loss": 1.0327,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.009134888726646,
+      "learning_rate": 3.3978134029082497e-06,
+      "loss": 1.1025,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6811120668007795,
+      "learning_rate": 3.597684779549911e-06,
+      "loss": 1.0371,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.2278619554787675,
+      "learning_rate": 3.7975561561915727e-06,
+      "loss": 1.1865,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3724203185974215,
+      "learning_rate": 3.997427532833234e-06,
+      "loss": 1.1035,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3157186551862456,
+      "learning_rate": 4.1972989094748964e-06,
+      "loss": 0.8867,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.776941111977726,
+      "learning_rate": 4.397170286116558e-06,
+      "loss": 1.0046,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.343637284275504,
+      "learning_rate": 4.59704166275822e-06,
+      "loss": 0.9487,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5295891610352097,
+      "learning_rate": 4.796913039399881e-06,
+      "loss": 1.0127,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.399758338506859,
+      "learning_rate": 4.996784416041543e-06,
+      "loss": 1.1382,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3290809781108288,
+      "learning_rate": 5.196655792683205e-06,
+      "loss": 1.1323,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3679284404085665,
+      "learning_rate": 5.396527169324867e-06,
+      "loss": 0.8953,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6115584885046332,
+      "learning_rate": 5.5963985459665286e-06,
+      "loss": 1.1387,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6211991241401886,
+      "learning_rate": 5.79626992260819e-06,
+      "loss": 1.1016,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.640652538250095,
+      "learning_rate": 5.996141299249852e-06,
+      "loss": 1.0269,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.824652871340495,
+      "learning_rate": 6.196012675891513e-06,
+      "loss": 1.2095,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4110736346128432,
+      "learning_rate": 6.395884052533175e-06,
+      "loss": 1.0413,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.250216751379655,
+      "learning_rate": 6.595755429174837e-06,
+      "loss": 1.1128,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.537239893179174,
+      "learning_rate": 6.7956268058164995e-06,
+      "loss": 1.124,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1626932680819158,
+      "learning_rate": 6.99549818245816e-06,
+      "loss": 1.0034,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.289856498558217,
+      "learning_rate": 7.195369559099822e-06,
+      "loss": 0.762,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.389701236410772,
+      "learning_rate": 7.395240935741484e-06,
+      "loss": 0.748,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.388511122615389,
+      "learning_rate": 7.595112312383145e-06,
+      "loss": 1.0537,
+      "step": 38
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.713931803558457,
+      "learning_rate": 7.794983689024808e-06,
+      "loss": 0.9602,
+      "step": 39
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6122543354824959,
+      "learning_rate": 7.994855065666469e-06,
+      "loss": 1.2021,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.365800993728332,
+      "learning_rate": 8.19472644230813e-06,
+      "loss": 0.9084,
+      "step": 41
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6727039741930945,
+      "learning_rate": 8.394597818949793e-06,
+      "loss": 1.0947,
+      "step": 42
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5138380390147157,
+      "learning_rate": 8.594469195591453e-06,
+      "loss": 0.8064,
+      "step": 43
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.7588432850123146,
+      "learning_rate": 8.794340572233115e-06,
+      "loss": 1.1538,
+      "step": 44
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2531249086755734,
+      "learning_rate": 8.994211948874777e-06,
+      "loss": 0.8459,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5562325472791454,
+      "learning_rate": 9.19408332551644e-06,
+      "loss": 1.1045,
+      "step": 46
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3422113081182725,
+      "learning_rate": 9.3939547021581e-06,
+      "loss": 0.894,
+      "step": 47
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3460385993734032,
+      "learning_rate": 9.593826078799762e-06,
+      "loss": 1.0391,
+      "step": 48
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6914894136418361,
+      "learning_rate": 9.793697455441424e-06,
+      "loss": 1.0205,
+      "step": 49
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.4353475594186835,
+      "learning_rate": 9.993568832083086e-06,
+      "loss": 0.8682,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5196039726683364,
+      "learning_rate": 1.0193440208724748e-05,
+      "loss": 0.8628,
+      "step": 51
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3375554242638321,
+      "learning_rate": 1.039331158536641e-05,
+      "loss": 0.8696,
+      "step": 52
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.185575330917505,
+      "learning_rate": 1.0593182962008073e-05,
+      "loss": 0.7842,
+      "step": 53
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6165352948220826,
+      "learning_rate": 1.0793054338649735e-05,
+      "loss": 0.8988,
+      "step": 54
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2629515114338927,
+      "learning_rate": 1.0992925715291395e-05,
+      "loss": 0.8145,
+      "step": 55
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.61186899066469,
+      "learning_rate": 1.1192797091933057e-05,
+      "loss": 1.0251,
+      "step": 56
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.2511658297260166,
+      "learning_rate": 1.1392668468574718e-05,
+      "loss": 1.2676,
+      "step": 57
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.199622766487703,
+      "learning_rate": 1.159253984521638e-05,
+      "loss": 0.8975,
+      "step": 58
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4470571349753039,
+      "learning_rate": 1.1792411221858042e-05,
+      "loss": 1.1226,
+      "step": 59
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6373096937876388,
+      "learning_rate": 1.1992282598499704e-05,
+      "loss": 0.9121,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3382912585880213,
+      "learning_rate": 1.2192153975141364e-05,
+      "loss": 1.0303,
+      "step": 61
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3112658647535718,
+      "learning_rate": 1.2392025351783026e-05,
+      "loss": 0.8279,
+      "step": 62
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2790544352963376,
+      "learning_rate": 1.2591896728424688e-05,
+      "loss": 0.8994,
+      "step": 63
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5075888986771395,
+      "learning_rate": 1.279176810506635e-05,
+      "loss": 1.1196,
+      "step": 64
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4346198377896118,
+      "learning_rate": 1.2991639481708013e-05,
+      "loss": 1.0732,
+      "step": 65
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.126938104357545,
+      "learning_rate": 1.3191510858349675e-05,
+      "loss": 1.0679,
+      "step": 66
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.674625451642783,
+      "learning_rate": 1.3391382234991337e-05,
+      "loss": 0.9331,
+      "step": 67
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3584407358282393,
+      "learning_rate": 1.3591253611632999e-05,
+      "loss": 1.1221,
+      "step": 68
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3745639282935145,
+      "learning_rate": 1.3791124988274658e-05,
+      "loss": 1.1638,
+      "step": 69
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.2901302620432893,
+      "learning_rate": 1.399099636491632e-05,
+      "loss": 1.0566,
+      "step": 70
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4115272261169016,
+      "learning_rate": 1.4190867741557982e-05,
+      "loss": 1.0718,
+      "step": 71
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.405597238567573,
+      "learning_rate": 1.4390739118199644e-05,
+      "loss": 1.0337,
+      "step": 72
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3864118209515417,
+      "learning_rate": 1.4590610494841306e-05,
+      "loss": 1.043,
+      "step": 73
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2755328766501015,
+      "learning_rate": 1.4790481871482968e-05,
+      "loss": 0.9449,
+      "step": 74
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.319458318380841,
+      "learning_rate": 1.499035324812463e-05,
+      "loss": 0.9866,
+      "step": 75
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.7407324492405107,
+      "learning_rate": 1.519022462476629e-05,
+      "loss": 1.1025,
+      "step": 76
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.8555178344658927,
+      "learning_rate": 1.5390096001407954e-05,
+      "loss": 1.0867,
+      "step": 77
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3711862560050792,
+      "learning_rate": 1.5589967378049616e-05,
+      "loss": 0.9534,
+      "step": 78
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.267290173756033,
+      "learning_rate": 1.578983875469128e-05,
+      "loss": 0.8936,
+      "step": 79
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4311663374123942,
+      "learning_rate": 1.5989710131332937e-05,
+      "loss": 1.0107,
+      "step": 80
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.9429771818157184,
+      "learning_rate": 1.61895815079746e-05,
+      "loss": 1.1055,
+      "step": 81
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.63894618288027,
+      "learning_rate": 1.638945288461626e-05,
+      "loss": 1.0952,
+      "step": 82
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5621150496261256,
+      "learning_rate": 1.6589324261257924e-05,
+      "loss": 1.0947,
+      "step": 83
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2928258811178885,
+      "learning_rate": 1.6789195637899586e-05,
+      "loss": 0.9937,
+      "step": 84
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4040869713795998,
+      "learning_rate": 1.6989067014541248e-05,
+      "loss": 0.8042,
+      "step": 85
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.9867943860716022,
+      "learning_rate": 1.7188938391182907e-05,
+      "loss": 1.1855,
+      "step": 86
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.7542284561232562,
+      "learning_rate": 1.738880976782457e-05,
+      "loss": 1.1851,
+      "step": 87
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2353359779309703,
+      "learning_rate": 1.758868114446623e-05,
+      "loss": 0.9365,
+      "step": 88
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5886895116160513,
+      "learning_rate": 1.7788552521107893e-05,
+      "loss": 1.1655,
+      "step": 89
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3167177278943167,
+      "learning_rate": 1.7988423897749555e-05,
+      "loss": 0.8616,
+      "step": 90
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2334640618366841,
+      "learning_rate": 1.8188295274391217e-05,
+      "loss": 0.8926,
+      "step": 91
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3694971248990673,
+      "learning_rate": 1.838816665103288e-05,
+      "loss": 0.8657,
+      "step": 92
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.536450876336402,
+      "learning_rate": 1.858803802767454e-05,
+      "loss": 0.8989,
+      "step": 93
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.503486475633459,
+      "learning_rate": 1.87879094043162e-05,
+      "loss": 0.9666,
+      "step": 94
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4799371333559792,
+      "learning_rate": 1.8987780780957862e-05,
+      "loss": 0.908,
+      "step": 95
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.445103068256348,
+      "learning_rate": 1.9187652157599524e-05,
+      "loss": 1.1133,
+      "step": 96
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3727616384190364,
+      "learning_rate": 1.9387523534241186e-05,
+      "loss": 0.9873,
+      "step": 97
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4973721532746032,
+      "learning_rate": 1.9587394910882848e-05,
+      "loss": 1.1631,
+      "step": 98
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4844608482833286,
+      "learning_rate": 1.978726628752451e-05,
+      "loss": 0.9077,
+      "step": 99
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3523284858357845,
+      "learning_rate": 1.9987137664166172e-05,
+      "loss": 1.1802,
+      "step": 100
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5967706963503001,
+      "learning_rate": 1.9987137664080615e-05,
+      "loss": 1.0609,
+      "step": 101
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3559830978684864,
+      "learning_rate": 1.9987137663823937e-05,
+      "loss": 0.8206,
+      "step": 102
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.7266245006631533,
+      "learning_rate": 1.9987137663396145e-05,
+      "loss": 1.1592,
+      "step": 103
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.853960504393059,
+      "learning_rate": 1.998713766279724e-05,
+      "loss": 1.1533,
+      "step": 104
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2286744128831855,
+      "learning_rate": 1.998713766202721e-05,
+      "loss": 0.8247,
+      "step": 105
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3893239653802958,
+      "learning_rate": 1.9987137661086066e-05,
+      "loss": 0.9492,
+      "step": 106
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3692356136528112,
+      "learning_rate": 1.998713765997381e-05,
+      "loss": 1.0894,
+      "step": 107
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.7511733753505103,
+      "learning_rate": 1.998713765869043e-05,
+      "loss": 1.2705,
+      "step": 108
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3086898085225258,
+      "learning_rate": 1.9987137657235935e-05,
+      "loss": 0.9897,
+      "step": 109
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.42803459298684,
+      "learning_rate": 1.9987137655610323e-05,
+      "loss": 0.7253,
+      "step": 110
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.30859589078358,
+      "learning_rate": 1.9987137653813597e-05,
+      "loss": 1.002,
+      "step": 111
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.219937455184391,
+      "learning_rate": 1.998713765184575e-05,
+      "loss": 0.7993,
+      "step": 112
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4863576397365221,
+      "learning_rate": 1.9987137649706788e-05,
+      "loss": 0.9639,
+      "step": 113
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.583284837415097,
+      "learning_rate": 1.998713764739671e-05,
+      "loss": 1.332,
+      "step": 114
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3680241667776767,
+      "learning_rate": 1.9987137644915512e-05,
+      "loss": 0.9436,
+      "step": 115
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4681322137713064,
+      "learning_rate": 1.9987137642263198e-05,
+      "loss": 0.9321,
+      "step": 116
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4511203679715612,
+      "learning_rate": 1.998713763943977e-05,
+      "loss": 0.9893,
+      "step": 117
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.372370763714992,
+      "learning_rate": 1.9987137636445224e-05,
+      "loss": 1.04,
+      "step": 118
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2612725759249404,
+      "learning_rate": 1.998713763327956e-05,
+      "loss": 1.0459,
+      "step": 119
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5887129977859773,
+      "learning_rate": 1.998713762994278e-05,
+      "loss": 1.0454,
+      "step": 120
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3537302438803718,
+      "learning_rate": 1.9987137626434883e-05,
+      "loss": 0.8994,
+      "step": 121
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5585427335645314,
+      "learning_rate": 1.9987137622755867e-05,
+      "loss": 1.0,
+      "step": 122
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.347974170281702,
+      "learning_rate": 1.9987137618905733e-05,
+      "loss": 0.9209,
+      "step": 123
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4807293745540115,
+      "learning_rate": 1.9987137614884484e-05,
+      "loss": 1.0269,
+      "step": 124
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6090058579276014,
+      "learning_rate": 1.9987137610692122e-05,
+      "loss": 0.9746,
+      "step": 125
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6827583098829078,
+      "learning_rate": 1.9987137606328635e-05,
+      "loss": 1.1555,
+      "step": 126
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.32453176574229,
+      "learning_rate": 1.9987137601794034e-05,
+      "loss": 0.8347,
+      "step": 127
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5403616596967007,
+      "learning_rate": 1.998713759708832e-05,
+      "loss": 1.2529,
+      "step": 128
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3944282814286415,
+      "learning_rate": 1.9987137592211486e-05,
+      "loss": 1.0327,
+      "step": 129
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2533564327695044,
+      "learning_rate": 1.9987137587163535e-05,
+      "loss": 0.8794,
+      "step": 130
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6117121198465103,
+      "learning_rate": 1.9987137581944467e-05,
+      "loss": 0.938,
+      "step": 131
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3219173541847693,
+      "learning_rate": 1.9987137576554285e-05,
+      "loss": 1.0723,
+      "step": 132
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.34345512703341,
+      "learning_rate": 1.9987137570992982e-05,
+      "loss": 1.1235,
+      "step": 133
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.8319389791779082,
+      "learning_rate": 1.9987137565260565e-05,
+      "loss": 1.1768,
+      "step": 134
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.284328251936961,
+      "learning_rate": 1.998713755935703e-05,
+      "loss": 0.8037,
+      "step": 135
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6430563939145162,
+      "learning_rate": 1.9987137553282377e-05,
+      "loss": 0.9888,
+      "step": 136
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.643652384365159,
+      "learning_rate": 1.9987137547036607e-05,
+      "loss": 1.1528,
+      "step": 137
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5986467569403668,
+      "learning_rate": 1.998713754061972e-05,
+      "loss": 1.1282,
+      "step": 138
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2897843622760647,
+      "learning_rate": 1.9987137534031717e-05,
+      "loss": 0.8491,
+      "step": 139
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5515178785409631,
+      "learning_rate": 1.9987137527272597e-05,
+      "loss": 0.9893,
+      "step": 140
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5441821771382438,
+      "learning_rate": 1.998713752034236e-05,
+      "loss": 1.1855,
+      "step": 141
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2874668302244616,
+      "learning_rate": 1.9987137513241005e-05,
+      "loss": 0.8123,
+      "step": 142
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4621717166069157,
+      "learning_rate": 1.9987137505968533e-05,
+      "loss": 0.936,
+      "step": 143
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5017293656723831,
+      "learning_rate": 1.9987137498524946e-05,
+      "loss": 1.0967,
+      "step": 144
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4607500578230534,
+      "learning_rate": 1.9987137490910242e-05,
+      "loss": 1.0801,
+      "step": 145
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6647483594284533,
+      "learning_rate": 1.998713748312442e-05,
+      "loss": 1.0786,
+      "step": 146
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.471892469419148,
+      "learning_rate": 1.998713747516748e-05,
+      "loss": 1.0103,
+      "step": 147
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4000494846045373,
+      "learning_rate": 1.9987137467039424e-05,
+      "loss": 1.1025,
+      "step": 148
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3810649109155702,
+      "learning_rate": 1.9987137458740252e-05,
+      "loss": 1.0601,
+      "step": 149
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.291241601508599,
+      "learning_rate": 1.998713745026996e-05,
+      "loss": 0.9727,
+      "step": 150
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.567335048434997,
+      "learning_rate": 1.9987137441628554e-05,
+      "loss": 1.1538,
+      "step": 151
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.8288608602426044,
+      "learning_rate": 1.998713743281603e-05,
+      "loss": 1.1812,
+      "step": 152
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2820451176629666,
+      "learning_rate": 1.9987137423832388e-05,
+      "loss": 0.99,
+      "step": 153
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4646354018497445,
+      "learning_rate": 1.9987137414677632e-05,
+      "loss": 0.9419,
+      "step": 154
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5551528617571317,
+      "learning_rate": 1.998713740535176e-05,
+      "loss": 1.0073,
+      "step": 155
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4728851545201214,
+      "learning_rate": 1.9987137395854765e-05,
+      "loss": 1.0078,
+      "step": 156
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.088527371808708,
+      "learning_rate": 1.9987137386186656e-05,
+      "loss": 1.2344,
+      "step": 157
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1853336852694984,
+      "learning_rate": 1.998713737634743e-05,
+      "loss": 0.9736,
+      "step": 158
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3366185287280403,
+      "learning_rate": 1.998713736633709e-05,
+      "loss": 1.3555,
+      "step": 159
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2789196593945422,
+      "learning_rate": 1.9987137356155628e-05,
+      "loss": 0.9136,
+      "step": 160
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6349893323451148,
+      "learning_rate": 1.998713734580305e-05,
+      "loss": 0.9644,
+      "step": 161
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3782827031908806,
+      "learning_rate": 1.9987137335279356e-05,
+      "loss": 1.04,
+      "step": 162
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5935483318715946,
+      "learning_rate": 1.998713732458455e-05,
+      "loss": 1.1729,
+      "step": 163
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3679427758261218,
+      "learning_rate": 1.998713731371862e-05,
+      "loss": 0.9417,
+      "step": 164
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.296487359632256,
+      "learning_rate": 1.9987137302681574e-05,
+      "loss": 0.908,
+      "step": 165
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.344633831962935,
+      "learning_rate": 1.9987137291473414e-05,
+      "loss": 0.9961,
+      "step": 166
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2687331287784427,
+      "learning_rate": 1.9987137280094136e-05,
+      "loss": 0.7136,
+      "step": 167
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6592947126890205,
+      "learning_rate": 1.9987137268543737e-05,
+      "loss": 0.9565,
+      "step": 168
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4035285046668897,
+      "learning_rate": 1.9987137256822228e-05,
+      "loss": 1.2236,
+      "step": 169
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.5359744028731024,
+      "learning_rate": 1.99871372449296e-05,
+      "loss": 1.3691,
+      "step": 170
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2878914339886232,
+      "learning_rate": 1.9987137232865852e-05,
+      "loss": 1.0078,
+      "step": 171
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.516899674021975,
+      "learning_rate": 1.998713722063099e-05,
+      "loss": 0.8313,
+      "step": 172
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4427677330522974,
+      "learning_rate": 1.998713720822501e-05,
+      "loss": 0.9763,
+      "step": 173
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.605005870463726,
+      "learning_rate": 1.9987137195647912e-05,
+      "loss": 1.0288,
+      "step": 174
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3037056797793092,
+      "learning_rate": 1.9987137182899697e-05,
+      "loss": 1.0386,
+      "step": 175
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 6.41374817239471,
+      "learning_rate": 1.9987137169980368e-05,
+      "loss": 1.0155,
+      "step": 176
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5914228315805112,
+      "learning_rate": 1.9987137156889918e-05,
+      "loss": 0.9578,
+      "step": 177
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.552733069845167,
+      "learning_rate": 1.9987137143628353e-05,
+      "loss": 1.0044,
+      "step": 178
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.181120955265809,
+      "learning_rate": 1.9987137130195674e-05,
+      "loss": 0.8853,
+      "step": 179
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6993552668310692,
+      "learning_rate": 1.9987137116591874e-05,
+      "loss": 1.2163,
+      "step": 180
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5882948470014477,
+      "learning_rate": 1.9987137102816957e-05,
+      "loss": 1.0244,
+      "step": 181
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1614984525357406,
+      "learning_rate": 1.9987137088870926e-05,
+      "loss": 0.9224,
+      "step": 182
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2248082341717805,
+      "learning_rate": 1.9987137074753776e-05,
+      "loss": 1.0205,
+      "step": 183
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2218550717028729,
+      "learning_rate": 1.9987137060465506e-05,
+      "loss": 0.7527,
+      "step": 184
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4586790855711262,
+      "learning_rate": 1.9987137046006125e-05,
+      "loss": 1.0479,
+      "step": 185
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.373406961251245,
+      "learning_rate": 1.9987137031375623e-05,
+      "loss": 0.9517,
+      "step": 186
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5436809200012414,
+      "learning_rate": 1.9987137016574004e-05,
+      "loss": 1.1509,
+      "step": 187
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.3507387866732805,
+      "learning_rate": 1.998713700160127e-05,
+      "loss": 1.0176,
+      "step": 188
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 5.297258661379631,
+      "learning_rate": 1.9987136986457422e-05,
+      "loss": 1.4824,
+      "step": 189
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2368830057880673,
+      "learning_rate": 1.998713697114245e-05,
+      "loss": 1.1924,
+      "step": 190
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1857806859735769,
+      "learning_rate": 1.9987136955656367e-05,
+      "loss": 1.0742,
+      "step": 191
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.7825860483980114,
+      "learning_rate": 1.9987136939999162e-05,
+      "loss": 1.0195,
+      "step": 192
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2280524856086843,
+      "learning_rate": 1.9987136924170844e-05,
+      "loss": 0.7825,
+      "step": 193
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6792452917630727,
+      "learning_rate": 1.9987136908171408e-05,
+      "loss": 1.5215,
+      "step": 194
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4562428159086307,
+      "learning_rate": 1.9987136892000855e-05,
+      "loss": 0.9019,
+      "step": 195
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3504191065955236,
+      "learning_rate": 1.9987136875659184e-05,
+      "loss": 1.0806,
+      "step": 196
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2588221129110082,
+      "learning_rate": 1.9987136859146398e-05,
+      "loss": 0.9414,
+      "step": 197
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4294441724818887,
+      "learning_rate": 1.9987136842462492e-05,
+      "loss": 0.877,
+      "step": 198
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3326717414183664,
+      "learning_rate": 1.9987136825607475e-05,
+      "loss": 1.0249,
+      "step": 199
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4654617023390064,
+      "learning_rate": 1.9987136808581333e-05,
+      "loss": 1.3115,
+      "step": 200
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3611227616743549,
+      "learning_rate": 1.998713679138408e-05,
+      "loss": 1.0933,
+      "step": 201
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.218928495198492,
+      "learning_rate": 1.9987136774015708e-05,
+      "loss": 1.0298,
+      "step": 202
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4398903253963204,
+      "learning_rate": 1.998713675647622e-05,
+      "loss": 1.0562,
+      "step": 203
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.164106176184216,
+      "learning_rate": 1.998713673876561e-05,
+      "loss": 0.7361,
+      "step": 204
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1984851096592528,
+      "learning_rate": 1.998713672088389e-05,
+      "loss": 0.9363,
+      "step": 205
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6200642821426983,
+      "learning_rate": 1.9987136702831052e-05,
+      "loss": 1.0679,
+      "step": 206
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4575069356658439,
+      "learning_rate": 1.9987136684607094e-05,
+      "loss": 0.9756,
+      "step": 207
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4889251037166171,
+      "learning_rate": 1.9987136666212022e-05,
+      "loss": 1.1917,
+      "step": 208
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3879905778499126,
+      "learning_rate": 1.998713664764583e-05,
+      "loss": 1.2432,
+      "step": 209
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2198236577991115,
+      "learning_rate": 1.998713662890852e-05,
+      "loss": 0.9595,
+      "step": 210
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4453703997587537,
+      "learning_rate": 1.9987136610000097e-05,
+      "loss": 1.1616,
+      "step": 211
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1224739537265145,
+      "learning_rate": 1.9987136590920558e-05,
+      "loss": 0.7456,
+      "step": 212
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.7481158876914644,
+      "learning_rate": 1.9987136571669897e-05,
+      "loss": 1.2329,
+      "step": 213
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2297082397573267,
+      "learning_rate": 1.9987136552248123e-05,
+      "loss": 0.8488,
+      "step": 214
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1761797866377828,
+      "learning_rate": 1.998713653265523e-05,
+      "loss": 1.0037,
+      "step": 215
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.362201295709729,
+      "learning_rate": 1.998713651289122e-05,
+      "loss": 0.9146,
+      "step": 216
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2537215147272014,
+      "learning_rate": 1.9987136492956095e-05,
+      "loss": 1.2041,
+      "step": 217
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.448224325449314,
+      "learning_rate": 1.998713647284985e-05,
+      "loss": 1.0142,
+      "step": 218
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.410545401095706,
+      "learning_rate": 1.998713645257249e-05,
+      "loss": 1.1021,
+      "step": 219
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2530084646171964,
+      "learning_rate": 1.9987136432124015e-05,
+      "loss": 1.0127,
+      "step": 220
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5647764026701685,
+      "learning_rate": 1.9987136411504417e-05,
+      "loss": 1.218,
+      "step": 221
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4965770130166682,
+      "learning_rate": 1.998713639071371e-05,
+      "loss": 1.2778,
+      "step": 222
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2998445821439122,
+      "learning_rate": 1.998713636975188e-05,
+      "loss": 1.1587,
+      "step": 223
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.423425855704122,
+      "learning_rate": 1.9987136348618936e-05,
+      "loss": 1.0908,
+      "step": 224
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2130956307599563,
+      "learning_rate": 1.9987136327314875e-05,
+      "loss": 0.8555,
+      "step": 225
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2287258823391218,
+      "learning_rate": 1.9987136305839696e-05,
+      "loss": 1.0327,
+      "step": 226
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.7651201049529341,
+      "learning_rate": 1.99871362841934e-05,
+      "loss": 0.8445,
+      "step": 227
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6722209964563237,
+      "learning_rate": 1.9987136262375986e-05,
+      "loss": 0.9731,
+      "step": 228
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.497507760433936,
+      "learning_rate": 1.9987136240387455e-05,
+      "loss": 1.0557,
+      "step": 229
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2785662013142645,
+      "learning_rate": 1.998713621822781e-05,
+      "loss": 1.0759,
+      "step": 230
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1870064462648997,
+      "learning_rate": 1.9987136195897046e-05,
+      "loss": 0.8004,
+      "step": 231
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3741907859676556,
+      "learning_rate": 1.9987136173395168e-05,
+      "loss": 0.9941,
+      "step": 232
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3119498416802988,
+      "learning_rate": 1.998713615072217e-05,
+      "loss": 0.8967,
+      "step": 233
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2002045000188675,
+      "learning_rate": 1.9987136127878054e-05,
+      "loss": 0.8645,
+      "step": 234
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.9354233841904378,
+      "learning_rate": 1.9987136104862823e-05,
+      "loss": 1.3108,
+      "step": 235
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1685749126802762,
+      "learning_rate": 1.9987136081676476e-05,
+      "loss": 1.0063,
+      "step": 236
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4541965758445865,
+      "learning_rate": 1.998713605831901e-05,
+      "loss": 1.2368,
+      "step": 237
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4966472667521429,
+      "learning_rate": 1.9987136034790427e-05,
+      "loss": 1.1562,
+      "step": 238
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1252135497967393,
+      "learning_rate": 1.998713601109073e-05,
+      "loss": 0.9272,
+      "step": 239
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2862879631290096,
+      "learning_rate": 1.9987135987219915e-05,
+      "loss": 1.0952,
+      "step": 240
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1803493032860841,
+      "learning_rate": 1.998713596317798e-05,
+      "loss": 0.8796,
+      "step": 241
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3732131271314898,
+      "learning_rate": 1.998713593896493e-05,
+      "loss": 0.8728,
+      "step": 242
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.843332340327588,
+      "learning_rate": 1.998713591458076e-05,
+      "loss": 1.1235,
+      "step": 243
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4092784484812428,
+      "learning_rate": 1.998713589002548e-05,
+      "loss": 1.1475,
+      "step": 244
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4532488698359423,
+      "learning_rate": 1.9987135865299077e-05,
+      "loss": 1.2202,
+      "step": 245
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2482460591341684,
+      "learning_rate": 1.9987135840401563e-05,
+      "loss": 1.0107,
+      "step": 246
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.0546269280965994,
+      "learning_rate": 1.9987135815332925e-05,
+      "loss": 0.8613,
+      "step": 247
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4859790366992953,
+      "learning_rate": 1.9987135790093176e-05,
+      "loss": 0.9561,
+      "step": 248
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6812543790965855,
+      "learning_rate": 1.9987135764682306e-05,
+      "loss": 1.1309,
+      "step": 249
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6401873050121658,
+      "learning_rate": 1.9987135739100322e-05,
+      "loss": 1.0386,
+      "step": 250
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6207494898719814,
+      "learning_rate": 1.998713571334722e-05,
+      "loss": 1.1196,
+      "step": 251
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4317063216348882,
+      "learning_rate": 1.9987135687423e-05,
+      "loss": 1.4102,
+      "step": 252
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.113637873767119,
+      "learning_rate": 1.9987135661327665e-05,
+      "loss": 1.0486,
+      "step": 253
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2295469676351236,
+      "learning_rate": 1.998713563506121e-05,
+      "loss": 1.126,
+      "step": 254
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.267949501027812,
+      "learning_rate": 1.998713560862364e-05,
+      "loss": 0.9463,
+      "step": 255
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2672233849276275,
+      "learning_rate": 1.9987135582014952e-05,
+      "loss": 1.0049,
+      "step": 256
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.9002193104919143,
+      "learning_rate": 1.998713555523515e-05,
+      "loss": 1.1787,
+      "step": 257
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3370982252943828,
+      "learning_rate": 1.9987135528284227e-05,
+      "loss": 0.8774,
+      "step": 258
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4656245477671628,
+      "learning_rate": 1.9987135501162188e-05,
+      "loss": 1.0776,
+      "step": 259
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.460249468510235,
+      "learning_rate": 1.9987135473869034e-05,
+      "loss": 1.0869,
+      "step": 260
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3532936167330853,
+      "learning_rate": 1.9987135446404764e-05,
+      "loss": 1.042,
+      "step": 261
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.6509903838487965,
+      "learning_rate": 1.9987135418769375e-05,
+      "loss": 1.0762,
+      "step": 262
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2325162775848921,
+      "learning_rate": 1.998713539096287e-05,
+      "loss": 1.0479,
+      "step": 263
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3464994026321577,
+      "learning_rate": 1.9987135362985246e-05,
+      "loss": 1.3032,
+      "step": 264
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1800045572613145,
+      "learning_rate": 1.9987135334836508e-05,
+      "loss": 0.9053,
+      "step": 265
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.250291122867789,
+      "learning_rate": 1.998713530651665e-05,
+      "loss": 1.0007,
+      "step": 266
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1946241104427617,
+      "learning_rate": 1.998713527802568e-05,
+      "loss": 0.9241,
+      "step": 267
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.487375779935634,
+      "learning_rate": 1.998713524936359e-05,
+      "loss": 0.9678,
+      "step": 268
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.840819017323815,
+      "learning_rate": 1.998713522053038e-05,
+      "loss": 1.229,
+      "step": 269
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2631792998121114,
+      "learning_rate": 1.9987135191526058e-05,
+      "loss": 1.0269,
+      "step": 270
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3140498048438995,
+      "learning_rate": 1.9987135162350615e-05,
+      "loss": 0.9121,
+      "step": 271
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.170500088619602,
+      "learning_rate": 1.9987135133004057e-05,
+      "loss": 0.9747,
+      "step": 272
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3861599938193538,
+      "learning_rate": 1.9987135103486382e-05,
+      "loss": 1.1235,
+      "step": 273
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6092609253981107,
+      "learning_rate": 1.998713507379759e-05,
+      "loss": 1.1162,
+      "step": 274
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6361200152494701,
+      "learning_rate": 1.9987135043937686e-05,
+      "loss": 1.1748,
+      "step": 275
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2067700939919823,
+      "learning_rate": 1.9987135013906658e-05,
+      "loss": 0.9553,
+      "step": 276
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.436517794537693,
+      "learning_rate": 1.9987134983704515e-05,
+      "loss": 1.2119,
+      "step": 277
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.1511939659113466,
+      "learning_rate": 1.9987134953331256e-05,
+      "loss": 1.2773,
+      "step": 278
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.8390793558934753,
+      "learning_rate": 1.9987134922786878e-05,
+      "loss": 1.0054,
+      "step": 279
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5686166440659264,
+      "learning_rate": 1.9987134892071387e-05,
+      "loss": 1.127,
+      "step": 280
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2083798870896632,
+      "learning_rate": 1.9987134861184774e-05,
+      "loss": 0.9434,
+      "step": 281
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4122657953055318,
+      "learning_rate": 1.9987134830127047e-05,
+      "loss": 1.2188,
+      "step": 282
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1350635656126322,
+      "learning_rate": 1.9987134798898206e-05,
+      "loss": 1.0483,
+      "step": 283
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4298312422403636,
+      "learning_rate": 1.9987134767498244e-05,
+      "loss": 1.1311,
+      "step": 284
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2904350602823484,
+      "learning_rate": 1.9987134735927165e-05,
+      "loss": 0.9795,
+      "step": 285
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.1420862264365685,
+      "learning_rate": 1.998713470418497e-05,
+      "loss": 1.1885,
+      "step": 286
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.392412472752939,
+      "learning_rate": 1.998713467227166e-05,
+      "loss": 0.9011,
+      "step": 287
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1268209449298847,
+      "learning_rate": 1.998713464018723e-05,
+      "loss": 0.901,
+      "step": 288
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2330088237901513,
+      "learning_rate": 1.9987134607931684e-05,
+      "loss": 0.9854,
+      "step": 289
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3189372286837642,
+      "learning_rate": 1.998713457550502e-05,
+      "loss": 1.2178,
+      "step": 290
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6026963848880358,
+      "learning_rate": 1.998713454290724e-05,
+      "loss": 1.2588,
+      "step": 291
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6531557653553364,
+      "learning_rate": 1.9987134510138346e-05,
+      "loss": 1.1782,
+      "step": 292
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3121425959477593,
+      "learning_rate": 1.9987134477198332e-05,
+      "loss": 1.0249,
+      "step": 293
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4198084475037995,
+      "learning_rate": 1.99871344440872e-05,
+      "loss": 1.082,
+      "step": 294
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4350899105282737,
+      "learning_rate": 1.9987134410804955e-05,
+      "loss": 1.0654,
+      "step": 295
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3712130763629988,
+      "learning_rate": 1.9987134377351592e-05,
+      "loss": 1.2212,
+      "step": 296
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5887759510144823,
+      "learning_rate": 1.9987134343727112e-05,
+      "loss": 1.2979,
+      "step": 297
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4478677791694348,
+      "learning_rate": 1.9987134309931514e-05,
+      "loss": 1.1001,
+      "step": 298
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2576373434428822,
+      "learning_rate": 1.99871342759648e-05,
+      "loss": 1.0718,
+      "step": 299
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4256205429014164,
+      "learning_rate": 1.9987134241826968e-05,
+      "loss": 1.085,
+      "step": 300
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4271245611690806,
+      "learning_rate": 1.9987134207518017e-05,
+      "loss": 1.0908,
+      "step": 301
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5800194275542863,
+      "learning_rate": 1.9987134173037952e-05,
+      "loss": 1.0366,
+      "step": 302
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2984763680159521,
+      "learning_rate": 1.998713413838677e-05,
+      "loss": 0.7585,
+      "step": 303
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3092491216855093,
+      "learning_rate": 1.998713410356447e-05,
+      "loss": 0.9385,
+      "step": 304
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4435233099191391,
+      "learning_rate": 1.9987134068571054e-05,
+      "loss": 1.1387,
+      "step": 305
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2024302272421041,
+      "learning_rate": 1.998713403340652e-05,
+      "loss": 1.0972,
+      "step": 306
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3563925373749564,
+      "learning_rate": 1.9987133998070872e-05,
+      "loss": 1.3511,
+      "step": 307
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.112606071387441,
+      "learning_rate": 1.99871339625641e-05,
+      "loss": 0.8452,
+      "step": 308
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.8223862521201437,
+      "learning_rate": 1.998713392688622e-05,
+      "loss": 1.2734,
+      "step": 309
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4561634087092594,
+      "learning_rate": 1.998713389103722e-05,
+      "loss": 0.9316,
+      "step": 310
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4436867307861319,
+      "learning_rate": 1.9987133855017104e-05,
+      "loss": 0.8496,
+      "step": 311
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.493544755186546,
+      "learning_rate": 1.998713381882587e-05,
+      "loss": 1.0193,
+      "step": 312
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3876574280927871,
+      "learning_rate": 1.9987133782463518e-05,
+      "loss": 1.1562,
+      "step": 313
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4680882850366128,
+      "learning_rate": 1.998713374593005e-05,
+      "loss": 1.041,
+      "step": 314
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.111288220807091,
+      "learning_rate": 1.998713370922546e-05,
+      "loss": 0.7397,
+      "step": 315
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.300394315705773,
+      "learning_rate": 1.9987133672349763e-05,
+      "loss": 1.2212,
+      "step": 316
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4298144007995033,
+      "learning_rate": 1.998713363530294e-05,
+      "loss": 0.8408,
+      "step": 317
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.411120351980561,
+      "learning_rate": 1.9987133598085005e-05,
+      "loss": 1.0322,
+      "step": 318
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.390748179262935,
+      "learning_rate": 1.9987133560695954e-05,
+      "loss": 1.1216,
+      "step": 319
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6878184971652517,
+      "learning_rate": 1.9987133523135782e-05,
+      "loss": 1.623,
+      "step": 320
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3016979206613386,
+      "learning_rate": 1.9987133485404496e-05,
+      "loss": 1.0786,
+      "step": 321
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2869525634453167,
+      "learning_rate": 1.9987133447502093e-05,
+      "loss": 1.0962,
+      "step": 322
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1379173225914037,
+      "learning_rate": 1.998713340942857e-05,
+      "loss": 0.947,
+      "step": 323
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4545278033355944,
+      "learning_rate": 1.9987133371183936e-05,
+      "loss": 0.8896,
+      "step": 324
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4069010604959604,
+      "learning_rate": 1.998713333276818e-05,
+      "loss": 0.9976,
+      "step": 325
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.7156357033539855,
+      "learning_rate": 1.998713329418131e-05,
+      "loss": 0.9531,
+      "step": 326
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.173119812081541,
+      "learning_rate": 1.998713325542332e-05,
+      "loss": 1.2153,
+      "step": 327
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4500621387715866,
+      "learning_rate": 1.998713321649422e-05,
+      "loss": 1.1772,
+      "step": 328
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4510778137363773,
+      "learning_rate": 1.9987133177393996e-05,
+      "loss": 0.9912,
+      "step": 329
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6420584819429325,
+      "learning_rate": 1.9987133138122655e-05,
+      "loss": 1.1052,
+      "step": 330
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5966324265511227,
+      "learning_rate": 1.99871330986802e-05,
+      "loss": 1.2773,
+      "step": 331
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.501765721886076,
+      "learning_rate": 1.9987133059066627e-05,
+      "loss": 1.1851,
+      "step": 332
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.151430343435899,
+      "learning_rate": 1.9987133019281937e-05,
+      "loss": 0.8174,
+      "step": 333
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2471635584969445,
+      "learning_rate": 1.9987132979326132e-05,
+      "loss": 0.9841,
+      "step": 334
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2371023926509273,
+      "learning_rate": 1.9987132939199207e-05,
+      "loss": 1.2236,
+      "step": 335
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.278805564128117,
+      "learning_rate": 1.998713289890117e-05,
+      "loss": 1.0291,
+      "step": 336
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.7398313318913377,
+      "learning_rate": 1.998713285843201e-05,
+      "loss": 1.3423,
+      "step": 337
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4052499394046083,
+      "learning_rate": 1.9987132817791738e-05,
+      "loss": 0.8772,
+      "step": 338
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1912293349692202,
+      "learning_rate": 1.998713277698035e-05,
+      "loss": 1.0208,
+      "step": 339
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3666748393597512,
+      "learning_rate": 1.9987132735997842e-05,
+      "loss": 0.8625,
+      "step": 340
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5465332578529514,
+      "learning_rate": 1.9987132694844218e-05,
+      "loss": 0.8997,
+      "step": 341
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5089420969365752,
+      "learning_rate": 1.9987132653519476e-05,
+      "loss": 1.0049,
+      "step": 342
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.349159309213207,
+      "learning_rate": 1.9987132612023617e-05,
+      "loss": 0.9966,
+      "step": 343
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3449943015534462,
+      "learning_rate": 1.998713257035664e-05,
+      "loss": 1.1069,
+      "step": 344
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.39728502916487,
+      "learning_rate": 1.998713252851855e-05,
+      "loss": 1.0352,
+      "step": 345
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2075446558527767,
+      "learning_rate": 1.998713248650934e-05,
+      "loss": 0.9761,
+      "step": 346
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.293865895263314,
+      "learning_rate": 1.9987132444329013e-05,
+      "loss": 1.0396,
+      "step": 347
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5222763722867265,
+      "learning_rate": 1.9987132401977572e-05,
+      "loss": 1.2549,
+      "step": 348
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2457845178606959,
+      "learning_rate": 1.9987132359455014e-05,
+      "loss": 1.0806,
+      "step": 349
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1105058306215057,
+      "learning_rate": 1.9987132316761335e-05,
+      "loss": 0.6942,
+      "step": 350
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2236023676099563,
+      "learning_rate": 1.998713227389654e-05,
+      "loss": 1.1216,
+      "step": 351
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4477721858036554,
+      "learning_rate": 1.998713223086063e-05,
+      "loss": 1.2402,
+      "step": 352
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5136750251983349,
+      "learning_rate": 1.99871321876536e-05,
+      "loss": 1.2197,
+      "step": 353
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.399526502373988,
+      "learning_rate": 1.9987132144275462e-05,
+      "loss": 1.0806,
+      "step": 354
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2940325552841134,
+      "learning_rate": 1.9987132100726198e-05,
+      "loss": 1.0332,
+      "step": 355
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4742300998430757,
+      "learning_rate": 1.9987132057005824e-05,
+      "loss": 1.0398,
+      "step": 356
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5627445792465893,
+      "learning_rate": 1.9987132013114328e-05,
+      "loss": 1.4336,
+      "step": 357
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.648836638955749,
+      "learning_rate": 1.9987131969051715e-05,
+      "loss": 1.2664,
+      "step": 358
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4582596987345042,
+      "learning_rate": 1.9987131924817987e-05,
+      "loss": 1.0312,
+      "step": 359
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.345015484391034,
+      "learning_rate": 1.9987131880413142e-05,
+      "loss": 1.0854,
+      "step": 360
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2583835798088336,
+      "learning_rate": 1.9987131835837183e-05,
+      "loss": 1.1523,
+      "step": 361
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2674516752913751,
+      "learning_rate": 1.9987131791090103e-05,
+      "loss": 1.1841,
+      "step": 362
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1174681517737213,
+      "learning_rate": 1.9987131746171905e-05,
+      "loss": 0.8179,
+      "step": 363
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.233300816047143,
+      "learning_rate": 1.9987131701082593e-05,
+      "loss": 0.9304,
+      "step": 364
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4512182051192815,
+      "learning_rate": 1.9987131655822163e-05,
+      "loss": 1.1294,
+      "step": 365
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3345825730693905,
+      "learning_rate": 1.9987131610390616e-05,
+      "loss": 0.8857,
+      "step": 366
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3641540881124432,
+      "learning_rate": 1.9987131564787954e-05,
+      "loss": 1.1587,
+      "step": 367
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1132844088325526,
+      "learning_rate": 1.9987131519014172e-05,
+      "loss": 0.9668,
+      "step": 368
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.5349697520541237,
+      "learning_rate": 1.9987131473069275e-05,
+      "loss": 1.0449,
+      "step": 369
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6139902399659052,
+      "learning_rate": 1.998713142695326e-05,
+      "loss": 1.2153,
+      "step": 370
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2517687204986532,
+      "learning_rate": 1.9987131380666132e-05,
+      "loss": 1.1138,
+      "step": 371
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.598555482032179,
+      "learning_rate": 1.9987131334207883e-05,
+      "loss": 1.0562,
+      "step": 372
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5807465788517505,
+      "learning_rate": 1.998713128757852e-05,
+      "loss": 1.0562,
+      "step": 373
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3574912513184039,
+      "learning_rate": 1.9987131240778038e-05,
+      "loss": 1.1392,
+      "step": 374
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4411701949854177,
+      "learning_rate": 1.9987131193806443e-05,
+      "loss": 1.2773,
+      "step": 375
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3052203523651282,
+      "learning_rate": 1.9987131146663726e-05,
+      "loss": 1.1646,
+      "step": 376
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.469237226471248,
+      "learning_rate": 1.9987131099349892e-05,
+      "loss": 1.2939,
+      "step": 377
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.372562588930507,
+      "learning_rate": 1.9987131051864947e-05,
+      "loss": 1.1694,
+      "step": 378
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2335358193573602,
+      "learning_rate": 1.9987131004208878e-05,
+      "loss": 1.031,
+      "step": 379
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3829394352298936,
+      "learning_rate": 1.9987130956381695e-05,
+      "loss": 1.3291,
+      "step": 380
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.336219579352166,
+      "learning_rate": 1.9987130908383397e-05,
+      "loss": 1.1357,
+      "step": 381
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.229865371267842,
+      "learning_rate": 1.9987130860213982e-05,
+      "loss": 0.8726,
+      "step": 382
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5392005974637575,
+      "learning_rate": 1.998713081187345e-05,
+      "loss": 1.2695,
+      "step": 383
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.37197573732938,
+      "learning_rate": 1.99871307633618e-05,
+      "loss": 0.8086,
+      "step": 384
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3784664414437018,
+      "learning_rate": 1.998713071467903e-05,
+      "loss": 0.7888,
+      "step": 385
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.7513404889250386,
+      "learning_rate": 1.9987130665825145e-05,
+      "loss": 1.1362,
+      "step": 386
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2736921933340037,
+      "learning_rate": 1.998713061680015e-05,
+      "loss": 1.1714,
+      "step": 387
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2041976407741213,
+      "learning_rate": 1.998713056760403e-05,
+      "loss": 0.8787,
+      "step": 388
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6177069525417012,
+      "learning_rate": 1.9987130518236796e-05,
+      "loss": 1.0649,
+      "step": 389
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4721639199778487,
+      "learning_rate": 1.9987130468698443e-05,
+      "loss": 1.1335,
+      "step": 390
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3756040633400872,
+      "learning_rate": 1.9987130418988977e-05,
+      "loss": 1.0305,
+      "step": 391
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.537226554931287,
+      "learning_rate": 1.9987130369108392e-05,
+      "loss": 1.1743,
+      "step": 392
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1174255864331706,
+      "learning_rate": 1.998713031905669e-05,
+      "loss": 0.8643,
+      "step": 393
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2935066602437382,
+      "learning_rate": 1.998713026883387e-05,
+      "loss": 1.0122,
+      "step": 394
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3161875828998775,
+      "learning_rate": 1.9987130218439937e-05,
+      "loss": 1.1963,
+      "step": 395
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1725545311713894,
+      "learning_rate": 1.9987130167874882e-05,
+      "loss": 0.8955,
+      "step": 396
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3587935947278473,
+      "learning_rate": 1.9987130117138716e-05,
+      "loss": 1.3047,
+      "step": 397
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6236889024946715,
+      "learning_rate": 1.998713006623143e-05,
+      "loss": 1.3059,
+      "step": 398
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3764629817326224,
+      "learning_rate": 1.998713001515303e-05,
+      "loss": 1.0251,
+      "step": 399
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4647699362912994,
+      "learning_rate": 1.9987129963903507e-05,
+      "loss": 1.0669,
+      "step": 400
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4657227993932738,
+      "learning_rate": 1.9987129912482875e-05,
+      "loss": 1.0771,
+      "step": 401
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1475934410696251,
+      "learning_rate": 1.9987129860891118e-05,
+      "loss": 0.7487,
+      "step": 402
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1185488109724004,
+      "learning_rate": 1.998712980912825e-05,
+      "loss": 0.8333,
+      "step": 403
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1478134847377641,
+      "learning_rate": 1.998712975719426e-05,
+      "loss": 0.8425,
+      "step": 404
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5884043483062686,
+      "learning_rate": 1.998712970508916e-05,
+      "loss": 1.2261,
+      "step": 405
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1914829417074941,
+      "learning_rate": 1.9987129652812937e-05,
+      "loss": 0.9338,
+      "step": 406
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3301683302888327,
+      "learning_rate": 1.9987129600365603e-05,
+      "loss": 1.0498,
+      "step": 407
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6187372567989755,
+      "learning_rate": 1.9987129547747147e-05,
+      "loss": 1.2808,
+      "step": 408
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.387618211057973,
+      "learning_rate": 1.9987129494957577e-05,
+      "loss": 0.9478,
+      "step": 409
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6757831706856345,
+      "learning_rate": 1.998712944199689e-05,
+      "loss": 1.2085,
+      "step": 410
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1978801500587295,
+      "learning_rate": 1.9987129388865084e-05,
+      "loss": 0.9832,
+      "step": 411
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3673435667062035,
+      "learning_rate": 1.998712933556216e-05,
+      "loss": 1.2881,
+      "step": 412
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5788008498507276,
+      "learning_rate": 1.9987129282088124e-05,
+      "loss": 1.1064,
+      "step": 413
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6632483554810742,
+      "learning_rate": 1.998712922844297e-05,
+      "loss": 0.9927,
+      "step": 414
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2649303460828523,
+      "learning_rate": 1.9987129174626697e-05,
+      "loss": 0.7739,
+      "step": 415
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4122309335646195,
+      "learning_rate": 1.9987129120639308e-05,
+      "loss": 1.3481,
+      "step": 416
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.065638171251623,
+      "learning_rate": 1.9987129066480804e-05,
+      "loss": 0.8473,
+      "step": 417
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1715535549535618,
+      "learning_rate": 1.9987129012151182e-05,
+      "loss": 1.03,
+      "step": 418
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3741055527085377,
+      "learning_rate": 1.9987128957650443e-05,
+      "loss": 1.1533,
+      "step": 419
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.552872408123689,
+      "learning_rate": 1.9987128902978586e-05,
+      "loss": 1.0898,
+      "step": 420
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1991418571547852,
+      "learning_rate": 1.9987128848135612e-05,
+      "loss": 0.9165,
+      "step": 421
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2962942086813531,
+      "learning_rate": 1.9987128793121523e-05,
+      "loss": 0.959,
+      "step": 422
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3885154153107566,
+      "learning_rate": 1.9987128737936317e-05,
+      "loss": 1.0127,
+      "step": 423
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4445720528616126,
+      "learning_rate": 1.9987128682579993e-05,
+      "loss": 1.1777,
+      "step": 424
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3674045935341204,
+      "learning_rate": 1.9987128627052552e-05,
+      "loss": 1.0908,
+      "step": 425
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.532860667864957,
+      "learning_rate": 1.9987128571353996e-05,
+      "loss": 1.2407,
+      "step": 426
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5194662124326712,
+      "learning_rate": 1.9987128515484323e-05,
+      "loss": 1.0776,
+      "step": 427
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4947986226025227,
+      "learning_rate": 1.9987128459443532e-05,
+      "loss": 1.1694,
+      "step": 428
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.341074031154639,
+      "learning_rate": 1.9987128403231624e-05,
+      "loss": 1.1943,
+      "step": 429
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2326182165188424,
+      "learning_rate": 1.9987128346848598e-05,
+      "loss": 1.1953,
+      "step": 430
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5086673823304502,
+      "learning_rate": 1.998712829029446e-05,
+      "loss": 1.479,
+      "step": 431
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.412940787009459,
+      "learning_rate": 1.9987128233569203e-05,
+      "loss": 1.0776,
+      "step": 432
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3074923762724662,
+      "learning_rate": 1.9987128176672828e-05,
+      "loss": 0.964,
+      "step": 433
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1959743786526598,
+      "learning_rate": 1.9987128119605335e-05,
+      "loss": 1.1096,
+      "step": 434
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.118259510441688,
+      "learning_rate": 1.9987128062366728e-05,
+      "loss": 1.2041,
+      "step": 435
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.249154806498558,
+      "learning_rate": 1.9987128004957003e-05,
+      "loss": 1.1475,
+      "step": 436
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4764401824181401,
+      "learning_rate": 1.998712794737616e-05,
+      "loss": 1.2129,
+      "step": 437
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2829897629142872,
+      "learning_rate": 1.99871278896242e-05,
+      "loss": 0.8232,
+      "step": 438
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1711389900615377,
+      "learning_rate": 1.9987127831701127e-05,
+      "loss": 0.8652,
+      "step": 439
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5817636522572431,
+      "learning_rate": 1.9987127773606935e-05,
+      "loss": 1.1501,
+      "step": 440
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.425003215719242,
+      "learning_rate": 1.9987127715341626e-05,
+      "loss": 0.8148,
+      "step": 441
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.209871721403762,
+      "learning_rate": 1.99871276569052e-05,
+      "loss": 1.032,
+      "step": 442
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3594030618511614,
+      "learning_rate": 1.9987127598297658e-05,
+      "loss": 1.063,
+      "step": 443
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5489763571385993,
+      "learning_rate": 1.9987127539519e-05,
+      "loss": 1.1472,
+      "step": 444
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.558108452603105,
+      "learning_rate": 1.9987127480569222e-05,
+      "loss": 1.1392,
+      "step": 445
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6835636569916044,
+      "learning_rate": 1.998712742144833e-05,
+      "loss": 0.8538,
+      "step": 446
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4465448023335057,
+      "learning_rate": 1.9987127362156317e-05,
+      "loss": 1.1118,
+      "step": 447
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3945247959874505,
+      "learning_rate": 1.998712730269319e-05,
+      "loss": 1.0474,
+      "step": 448
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3995660671355106,
+      "learning_rate": 1.9987127243058948e-05,
+      "loss": 1.2715,
+      "step": 449
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2017161555070637,
+      "learning_rate": 1.9987127183253587e-05,
+      "loss": 0.8717,
+      "step": 450
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6241209146557911,
+      "learning_rate": 1.9987127123277112e-05,
+      "loss": 1.1621,
+      "step": 451
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.37528867292231,
+      "learning_rate": 1.998712706312952e-05,
+      "loss": 1.3516,
+      "step": 452
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4002238963111293,
+      "learning_rate": 1.9987127002810805e-05,
+      "loss": 1.4434,
+      "step": 453
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3386979611751777,
+      "learning_rate": 1.9987126942320977e-05,
+      "loss": 1.1279,
+      "step": 454
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2331553360793606,
+      "learning_rate": 1.9987126881660035e-05,
+      "loss": 1.3242,
+      "step": 455
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3176095596304842,
+      "learning_rate": 1.9987126820827975e-05,
+      "loss": 1.0254,
+      "step": 456
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.7361831175388596,
+      "learning_rate": 1.9987126759824798e-05,
+      "loss": 1.4082,
+      "step": 457
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3695028699250806,
+      "learning_rate": 1.9987126698650503e-05,
+      "loss": 1.0952,
+      "step": 458
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3807934174708962,
+      "learning_rate": 1.998712663730509e-05,
+      "loss": 1.2754,
+      "step": 459
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.405644943618242,
+      "learning_rate": 1.998712657578856e-05,
+      "loss": 1.0391,
+      "step": 460
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3024471029131435,
+      "learning_rate": 1.9987126514100916e-05,
+      "loss": 0.9492,
+      "step": 461
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3651728145489033,
+      "learning_rate": 1.9987126452242158e-05,
+      "loss": 1.1826,
+      "step": 462
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.762135796760041,
+      "learning_rate": 1.9987126390212278e-05,
+      "loss": 0.9407,
+      "step": 463
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4363094043315714,
+      "learning_rate": 1.9987126328011284e-05,
+      "loss": 1.1528,
+      "step": 464
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6129940784154089,
+      "learning_rate": 1.998712626563917e-05,
+      "loss": 1.1328,
+      "step": 465
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3819253953209747,
+      "learning_rate": 1.9987126203095944e-05,
+      "loss": 1.1875,
+      "step": 466
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6519036755362349,
+      "learning_rate": 1.9987126140381598e-05,
+      "loss": 0.9451,
+      "step": 467
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3511190844697576,
+      "learning_rate": 1.9987126077496137e-05,
+      "loss": 1.0403,
+      "step": 468
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.564720711939377,
+      "learning_rate": 1.9987126014439555e-05,
+      "loss": 0.9751,
+      "step": 469
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4255618409946305,
+      "learning_rate": 1.998712595121186e-05,
+      "loss": 1.3662,
+      "step": 470
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3824766365408712,
+      "learning_rate": 1.998712588781305e-05,
+      "loss": 1.231,
+      "step": 471
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1348217213302187,
+      "learning_rate": 1.9987125824243118e-05,
+      "loss": 0.8664,
+      "step": 472
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.319183950754767,
+      "learning_rate": 1.9987125760502073e-05,
+      "loss": 1.0464,
+      "step": 473
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5993886942126265,
+      "learning_rate": 1.998712569658991e-05,
+      "loss": 1.1465,
+      "step": 474
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.307771703523588,
+      "learning_rate": 1.998712563250663e-05,
+      "loss": 0.9648,
+      "step": 475
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3273010839797708,
+      "learning_rate": 1.9987125568252235e-05,
+      "loss": 1.0803,
+      "step": 476
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2931202687342411,
+      "learning_rate": 1.998712550382672e-05,
+      "loss": 1.0278,
+      "step": 477
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5231253279668793,
+      "learning_rate": 1.998712543923009e-05,
+      "loss": 1.0728,
+      "step": 478
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.571230643675912,
+      "learning_rate": 1.9987125374462345e-05,
+      "loss": 0.9575,
+      "step": 479
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.9704596906935322,
+      "learning_rate": 1.998712530952348e-05,
+      "loss": 0.5698,
+      "step": 480
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.336413962186845,
+      "learning_rate": 1.9987125244413497e-05,
+      "loss": 0.9678,
+      "step": 481
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4168362889935309,
+      "learning_rate": 1.99871251791324e-05,
+      "loss": 1.1187,
+      "step": 482
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4037238850121838,
+      "learning_rate": 1.998712511368019e-05,
+      "loss": 1.1353,
+      "step": 483
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3262065719987022,
+      "learning_rate": 1.9987125048056857e-05,
+      "loss": 0.9097,
+      "step": 484
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.671160061312432,
+      "learning_rate": 1.998712498226241e-05,
+      "loss": 0.9546,
+      "step": 485
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1816509151798846,
+      "learning_rate": 1.9987124916296847e-05,
+      "loss": 0.8204,
+      "step": 486
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.987327481825663,
+      "learning_rate": 1.998712485016017e-05,
+      "loss": 1.0908,
+      "step": 487
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3044140009402685,
+      "learning_rate": 1.998712478385237e-05,
+      "loss": 0.6975,
+      "step": 488
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3681160961295311,
+      "learning_rate": 1.9987124717373456e-05,
+      "loss": 1.0376,
+      "step": 489
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1662470653370265,
+      "learning_rate": 1.9987124650723425e-05,
+      "loss": 0.8999,
+      "step": 490
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.955964732470391,
+      "learning_rate": 1.9987124583902277e-05,
+      "loss": 0.936,
+      "step": 491
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3811996882324795,
+      "learning_rate": 1.998712451691001e-05,
+      "loss": 0.9722,
+      "step": 492
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2879772820098143,
+      "learning_rate": 1.998712444974663e-05,
+      "loss": 1.0381,
+      "step": 493
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5150881730913257,
+      "learning_rate": 1.9987124382412133e-05,
+      "loss": 1.2769,
+      "step": 494
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3184129944246596,
+      "learning_rate": 1.9987124314906517e-05,
+      "loss": 0.9736,
+      "step": 495
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.200679225223346,
+      "learning_rate": 1.9987124247229784e-05,
+      "loss": 1.127,
+      "step": 496
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5320774878683545,
+      "learning_rate": 1.9987124179381937e-05,
+      "loss": 0.9507,
+      "step": 497
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6060628047173415,
+      "learning_rate": 1.9987124111362972e-05,
+      "loss": 1.2412,
+      "step": 498
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.258856393509351,
+      "learning_rate": 1.9987124043172893e-05,
+      "loss": 1.1118,
+      "step": 499
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3975783111522668,
+      "learning_rate": 1.9987123974811693e-05,
+      "loss": 1.1433,
+      "step": 500
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3950471751805775,
+      "learning_rate": 1.9987123906279378e-05,
+      "loss": 1.0288,
+      "step": 501
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2512397340397223,
+      "learning_rate": 1.9987123837575946e-05,
+      "loss": 1.0239,
+      "step": 502
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.263164436068245,
+      "learning_rate": 1.9987123768701397e-05,
+      "loss": 0.8936,
+      "step": 503
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4037686390373896,
+      "learning_rate": 1.9987123699655733e-05,
+      "loss": 1.1235,
+      "step": 504
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3659744985191171,
+      "learning_rate": 1.998712363043895e-05,
+      "loss": 1.0747,
+      "step": 505
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.376366153357756,
+      "learning_rate": 1.998712356105105e-05,
+      "loss": 1.1763,
+      "step": 506
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4641260448821078,
+      "learning_rate": 1.9987123491492037e-05,
+      "loss": 1.0679,
+      "step": 507
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.031643705641492,
+      "learning_rate": 1.9987123421761903e-05,
+      "loss": 1.0664,
+      "step": 508
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5926351200671756,
+      "learning_rate": 1.9987123351860654e-05,
+      "loss": 0.8276,
+      "step": 509
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4654828521183403,
+      "learning_rate": 1.998712328178829e-05,
+      "loss": 0.9824,
+      "step": 510
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2682795060942258,
+      "learning_rate": 1.9987123211544805e-05,
+      "loss": 1.0742,
+      "step": 511
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.369318103034824,
+      "learning_rate": 1.9987123141130207e-05,
+      "loss": 0.8789,
+      "step": 512
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6292519328381165,
+      "learning_rate": 1.9987123070544492e-05,
+      "loss": 1.2373,
+      "step": 513
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3376980608169589,
+      "learning_rate": 1.998712299978766e-05,
+      "loss": 1.1094,
+      "step": 514
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2716911839179506,
+      "learning_rate": 1.998712292885971e-05,
+      "loss": 1.1191,
+      "step": 515
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.339136053533414,
+      "learning_rate": 1.998712285776064e-05,
+      "loss": 1.1194,
+      "step": 516
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3899463755054369,
+      "learning_rate": 1.998712278649046e-05,
+      "loss": 0.9385,
+      "step": 517
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6223871058376773,
+      "learning_rate": 1.9987122715049162e-05,
+      "loss": 1.0879,
+      "step": 518
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4453309753886716,
+      "learning_rate": 1.9987122643436744e-05,
+      "loss": 1.1309,
+      "step": 519
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.493237394295358,
+      "learning_rate": 1.998712257165321e-05,
+      "loss": 0.9314,
+      "step": 520
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2732503086421525,
+      "learning_rate": 1.9987122499698563e-05,
+      "loss": 0.9292,
+      "step": 521
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6236682716810176,
+      "learning_rate": 1.9987122427572796e-05,
+      "loss": 1.1748,
+      "step": 522
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.515768299997108,
+      "learning_rate": 1.9987122355275912e-05,
+      "loss": 1.3193,
+      "step": 523
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3479966770266996,
+      "learning_rate": 1.9987122282807913e-05,
+      "loss": 1.0522,
+      "step": 524
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.403937748814087,
+      "learning_rate": 1.9987122210168797e-05,
+      "loss": 1.1699,
+      "step": 525
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5415329058930827,
+      "learning_rate": 1.9987122137358563e-05,
+      "loss": 1.1724,
+      "step": 526
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1302019754866746,
+      "learning_rate": 1.9987122064377215e-05,
+      "loss": 0.8884,
+      "step": 527
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4066187904822227,
+      "learning_rate": 1.9987121991224746e-05,
+      "loss": 1.1982,
+      "step": 528
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.752745586280342,
+      "learning_rate": 1.9987121917901163e-05,
+      "loss": 0.9292,
+      "step": 529
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2235450317364251,
+      "learning_rate": 1.9987121844406465e-05,
+      "loss": 0.718,
+      "step": 530
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.058199295585321,
+      "learning_rate": 1.9987121770740647e-05,
+      "loss": 1.1069,
+      "step": 531
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1476317712758355,
+      "learning_rate": 1.9987121696903714e-05,
+      "loss": 0.8225,
+      "step": 532
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.14079473337686,
+      "learning_rate": 1.9987121622895664e-05,
+      "loss": 1.1982,
+      "step": 533
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1450672652130476,
+      "learning_rate": 1.9987121548716496e-05,
+      "loss": 0.9846,
+      "step": 534
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4299076932178725,
+      "learning_rate": 1.9987121474366214e-05,
+      "loss": 1.1406,
+      "step": 535
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3329198563730262,
+      "learning_rate": 1.9987121399844814e-05,
+      "loss": 1.0049,
+      "step": 536
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.604790091647731,
+      "learning_rate": 1.9987121325152297e-05,
+      "loss": 1.3701,
+      "step": 537
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.434635293318551,
+      "learning_rate": 1.9987121250288666e-05,
+      "loss": 1.082,
+      "step": 538
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.666714055659191,
+      "learning_rate": 1.9987121175253913e-05,
+      "loss": 1.085,
+      "step": 539
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4226224003912102,
+      "learning_rate": 1.9987121100048047e-05,
+      "loss": 1.2051,
+      "step": 540
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2511933828953308,
+      "learning_rate": 1.9987121024671062e-05,
+      "loss": 1.2402,
+      "step": 541
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3084831532671841,
+      "learning_rate": 1.9987120949122964e-05,
+      "loss": 0.9443,
+      "step": 542
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4584988727166361,
+      "learning_rate": 1.9987120873403744e-05,
+      "loss": 1.3955,
+      "step": 543
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.428342760040857,
+      "learning_rate": 1.998712079751341e-05,
+      "loss": 1.3101,
+      "step": 544
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3158574077988072,
+      "learning_rate": 1.998712072145196e-05,
+      "loss": 0.7957,
+      "step": 545
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4295350707463081,
+      "learning_rate": 1.9987120645219394e-05,
+      "loss": 0.8933,
+      "step": 546
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.165025965192516,
+      "learning_rate": 1.998712056881571e-05,
+      "loss": 0.8276,
+      "step": 547
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4528078122718755,
+      "learning_rate": 1.998712049224091e-05,
+      "loss": 0.8083,
+      "step": 548
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5781879223658353,
+      "learning_rate": 1.9987120415494992e-05,
+      "loss": 1.2329,
+      "step": 549
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5789662753194544,
+      "learning_rate": 1.998712033857796e-05,
+      "loss": 1.1785,
+      "step": 550
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.569737874885361,
+      "learning_rate": 1.998712026148981e-05,
+      "loss": 1.0239,
+      "step": 551
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6544237684923906,
+      "learning_rate": 1.9987120184230542e-05,
+      "loss": 1.2188,
+      "step": 552
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4897981857749656,
+      "learning_rate": 1.9987120106800157e-05,
+      "loss": 0.9932,
+      "step": 553
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.389021474020001,
+      "learning_rate": 1.998712002919866e-05,
+      "loss": 0.999,
+      "step": 554
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1405223120882084,
+      "learning_rate": 1.998711995142604e-05,
+      "loss": 0.7424,
+      "step": 555
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.185691108288321,
+      "learning_rate": 1.998711987348231e-05,
+      "loss": 0.7996,
+      "step": 556
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.43576535833388,
+      "learning_rate": 1.998711979536746e-05,
+      "loss": 0.9016,
+      "step": 557
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.441801435679889,
+      "learning_rate": 1.998711971708149e-05,
+      "loss": 1.2871,
+      "step": 558
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2522551697395645,
+      "learning_rate": 1.9987119638624406e-05,
+      "loss": 1.0081,
+      "step": 559
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.5242727686522075,
+      "learning_rate": 1.9987119559996205e-05,
+      "loss": 1.0684,
+      "step": 560
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1584184263154855,
+      "learning_rate": 1.998711948119689e-05,
+      "loss": 0.937,
+      "step": 561
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3845975744292038,
+      "learning_rate": 1.9987119402226455e-05,
+      "loss": 0.9844,
+      "step": 562
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4928494727667099,
+      "learning_rate": 1.9987119323084904e-05,
+      "loss": 0.978,
+      "step": 563
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3153438186089024,
+      "learning_rate": 1.998711924377224e-05,
+      "loss": 1.1406,
+      "step": 564
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.182589769242443,
+      "learning_rate": 1.9987119164288456e-05,
+      "loss": 0.9453,
+      "step": 565
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6190086099255498,
+      "learning_rate": 1.9987119084633556e-05,
+      "loss": 1.0068,
+      "step": 566
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.7278427279790267,
+      "learning_rate": 1.9987119004807534e-05,
+      "loss": 1.1943,
+      "step": 567
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5206561399031653,
+      "learning_rate": 1.9987118924810406e-05,
+      "loss": 1.135,
+      "step": 568
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2452719917553423,
+      "learning_rate": 1.9987118844642152e-05,
+      "loss": 0.9482,
+      "step": 569
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.467003941047643,
+      "learning_rate": 1.9987118764302785e-05,
+      "loss": 1.1509,
+      "step": 570
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4054390582372336,
+      "learning_rate": 1.9987118683792303e-05,
+      "loss": 1.198,
+      "step": 571
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.272502719898727,
+      "learning_rate": 1.9987118603110704e-05,
+      "loss": 0.9441,
+      "step": 572
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.7036276959120753,
+      "learning_rate": 1.9987118522257987e-05,
+      "loss": 0.9404,
+      "step": 573
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4008403571998713,
+      "learning_rate": 1.9987118441234153e-05,
+      "loss": 1.1782,
+      "step": 574
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3780338023625531,
+      "learning_rate": 1.99871183600392e-05,
+      "loss": 0.9756,
+      "step": 575
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5752758329543866,
+      "learning_rate": 1.9987118278673135e-05,
+      "loss": 1.0825,
+      "step": 576
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3406921458824255,
+      "learning_rate": 1.998711819713595e-05,
+      "loss": 1.1577,
+      "step": 577
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3907261500915051,
+      "learning_rate": 1.9987118115427653e-05,
+      "loss": 1.1401,
+      "step": 578
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.524183980524673,
+      "learning_rate": 1.9987118033548237e-05,
+      "loss": 1.0981,
+      "step": 579
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2482801049313539,
+      "learning_rate": 1.9987117951497704e-05,
+      "loss": 0.9141,
+      "step": 580
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2064956908957294,
+      "learning_rate": 1.9987117869276053e-05,
+      "loss": 0.9875,
+      "step": 581
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1643412211540758,
+      "learning_rate": 1.9987117786883285e-05,
+      "loss": 0.9698,
+      "step": 582
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4993791884967647,
+      "learning_rate": 1.9987117704319402e-05,
+      "loss": 1.248,
+      "step": 583
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3948389563355794,
+      "learning_rate": 1.9987117621584402e-05,
+      "loss": 1.0063,
+      "step": 584
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.330724259558462,
+      "learning_rate": 1.9987117538678287e-05,
+      "loss": 1.1738,
+      "step": 585
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4120188323945273,
+      "learning_rate": 1.9987117455601052e-05,
+      "loss": 0.8906,
+      "step": 586
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4829841924379208,
+      "learning_rate": 1.9987117372352702e-05,
+      "loss": 1.3521,
+      "step": 587
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2671887191973559,
+      "learning_rate": 1.998711728893324e-05,
+      "loss": 1.269,
+      "step": 588
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2368140930322822,
+      "learning_rate": 1.9987117205342657e-05,
+      "loss": 0.9272,
+      "step": 589
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4778222255826394,
+      "learning_rate": 1.9987117121580955e-05,
+      "loss": 1.0173,
+      "step": 590
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.264914371999989,
+      "learning_rate": 1.9987117037648138e-05,
+      "loss": 1.0161,
+      "step": 591
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4229804970912034,
+      "learning_rate": 1.9987116953544204e-05,
+      "loss": 1.0698,
+      "step": 592
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2352070961414836,
+      "learning_rate": 1.9987116869269155e-05,
+      "loss": 0.9082,
+      "step": 593
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4084381880842094,
+      "learning_rate": 1.998711678482299e-05,
+      "loss": 0.9497,
+      "step": 594
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2205508205767823,
+      "learning_rate": 1.998711670020571e-05,
+      "loss": 0.9209,
+      "step": 595
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4641196941005041,
+      "learning_rate": 1.9987116615417308e-05,
+      "loss": 1.0869,
+      "step": 596
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3167521760834573,
+      "learning_rate": 1.998711653045779e-05,
+      "loss": 0.9243,
+      "step": 597
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2122620752244149,
+      "learning_rate": 1.998711644532716e-05,
+      "loss": 0.9941,
+      "step": 598
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3182841413718736,
+      "learning_rate": 1.9987116360025412e-05,
+      "loss": 0.9146,
+      "step": 599
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.9865642906948461,
+      "learning_rate": 1.9987116274552544e-05,
+      "loss": 0.677,
+      "step": 600
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.0519971654875537,
+      "learning_rate": 1.9987116188908562e-05,
+      "loss": 0.8759,
+      "step": 601
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3976590851125434,
+      "learning_rate": 1.9987116103093462e-05,
+      "loss": 0.9541,
+      "step": 602
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3524131964223884,
+      "learning_rate": 1.998711601710725e-05,
+      "loss": 0.8649,
+      "step": 603
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.393339786119136,
+      "learning_rate": 1.9987115930949916e-05,
+      "loss": 1.063,
+      "step": 604
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.20624365928446,
+      "learning_rate": 1.998711584462147e-05,
+      "loss": 0.8467,
+      "step": 605
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2868621542439067,
+      "learning_rate": 1.9987115758121903e-05,
+      "loss": 1.0874,
+      "step": 606
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3762024043668264,
+      "learning_rate": 1.998711567145122e-05,
+      "loss": 1.1128,
+      "step": 607
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2717329916066964,
+      "learning_rate": 1.9987115584609423e-05,
+      "loss": 0.979,
+      "step": 608
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.9060732024706921,
+      "learning_rate": 1.9987115497596506e-05,
+      "loss": 1.1963,
+      "step": 609
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.850594504948011,
+      "learning_rate": 1.9987115410412476e-05,
+      "loss": 1.2773,
+      "step": 610
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.082675342763078,
+      "learning_rate": 1.9987115323057324e-05,
+      "loss": 0.7998,
+      "step": 611
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4798614947487947,
+      "learning_rate": 1.998711523553106e-05,
+      "loss": 0.9824,
+      "step": 612
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4786380127577596,
+      "learning_rate": 1.9987115147833678e-05,
+      "loss": 1.0532,
+      "step": 613
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2027341034242498,
+      "learning_rate": 1.9987115059965184e-05,
+      "loss": 0.7175,
+      "step": 614
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1420401064515893,
+      "learning_rate": 1.998711497192557e-05,
+      "loss": 0.9194,
+      "step": 615
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4165370638785542,
+      "learning_rate": 1.9987114883714836e-05,
+      "loss": 1.3896,
+      "step": 616
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3367857443041973,
+      "learning_rate": 1.998711479533299e-05,
+      "loss": 1.0693,
+      "step": 617
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5473374725433258,
+      "learning_rate": 1.9987114706780024e-05,
+      "loss": 1.0574,
+      "step": 618
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4585661384408657,
+      "learning_rate": 1.9987114618055942e-05,
+      "loss": 1.1694,
+      "step": 619
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2822642614977586,
+      "learning_rate": 1.9987114529160746e-05,
+      "loss": 0.9783,
+      "step": 620
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5816217339862666,
+      "learning_rate": 1.9987114440094432e-05,
+      "loss": 1.0952,
+      "step": 621
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2257193321796762,
+      "learning_rate": 1.9987114350857e-05,
+      "loss": 1.1079,
+      "step": 622
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.309632347509409,
+      "learning_rate": 1.9987114261448454e-05,
+      "loss": 1.1812,
+      "step": 623
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6966196110363756,
+      "learning_rate": 1.998711417186879e-05,
+      "loss": 1.1802,
+      "step": 624
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2052937566651365,
+      "learning_rate": 1.998711408211801e-05,
+      "loss": 1.0134,
+      "step": 625
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4199972644296006,
+      "learning_rate": 1.9987113992196115e-05,
+      "loss": 1.0642,
+      "step": 626
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.9872892625699888,
+      "learning_rate": 1.99871139021031e-05,
+      "loss": 0.6287,
+      "step": 627
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1968713207536923,
+      "learning_rate": 1.998711381183897e-05,
+      "loss": 1.3048,
+      "step": 628
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.336713777284274,
+      "learning_rate": 1.9987113721403724e-05,
+      "loss": 1.1768,
+      "step": 629
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.0040876172768192,
+      "learning_rate": 1.998711363079736e-05,
+      "loss": 0.6316,
+      "step": 630
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.394970180985794,
+      "learning_rate": 1.998711354001988e-05,
+      "loss": 0.9873,
+      "step": 631
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.8922649238553313,
+      "learning_rate": 1.9987113449071284e-05,
+      "loss": 1.021,
+      "step": 632
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.378808858312023,
+      "learning_rate": 1.9987113357951573e-05,
+      "loss": 1.1714,
+      "step": 633
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.412743137330125,
+      "learning_rate": 1.9987113266660743e-05,
+      "loss": 1.1133,
+      "step": 634
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2676975564281785,
+      "learning_rate": 1.9987113175198797e-05,
+      "loss": 1.0298,
+      "step": 635
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5424696489312668,
+      "learning_rate": 1.9987113083565736e-05,
+      "loss": 0.9683,
+      "step": 636
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3260628791704618,
+      "learning_rate": 1.9987112991761554e-05,
+      "loss": 1.1099,
+      "step": 637
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.266884918399898,
+      "learning_rate": 1.998711289978626e-05,
+      "loss": 0.9512,
+      "step": 638
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5270561864642218,
+      "learning_rate": 1.9987112807639847e-05,
+      "loss": 1.064,
+      "step": 639
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4400064758314115,
+      "learning_rate": 1.9987112715322316e-05,
+      "loss": 1.1011,
+      "step": 640
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6518365609138084,
+      "learning_rate": 1.9987112622833673e-05,
+      "loss": 1.145,
+      "step": 641
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.452207870652463,
+      "learning_rate": 1.998711253017391e-05,
+      "loss": 0.9692,
+      "step": 642
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.326383368773011,
+      "learning_rate": 1.9987112437343033e-05,
+      "loss": 1.2935,
+      "step": 643
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3355349553288423,
+      "learning_rate": 1.9987112344341038e-05,
+      "loss": 1.2222,
+      "step": 644
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.122738472430234,
+      "learning_rate": 1.9987112251167925e-05,
+      "loss": 0.7256,
+      "step": 645
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3147483369476194,
+      "learning_rate": 1.99871121578237e-05,
+      "loss": 0.8979,
+      "step": 646
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3035519163608602,
+      "learning_rate": 1.9987112064308354e-05,
+      "loss": 0.9814,
+      "step": 647
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.057259104012112,
+      "learning_rate": 1.9987111970621895e-05,
+      "loss": 0.9218,
+      "step": 648
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4210116102105392,
+      "learning_rate": 1.9987111876764316e-05,
+      "loss": 1.4053,
+      "step": 649
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2613940692201504,
+      "learning_rate": 1.9987111782735622e-05,
+      "loss": 0.7996,
+      "step": 650
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.322573379471133,
+      "learning_rate": 1.998711168853581e-05,
+      "loss": 1.1382,
+      "step": 651
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.305896541467082,
+      "learning_rate": 1.9987111594164885e-05,
+      "loss": 1.1143,
+      "step": 652
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3382142949321556,
+      "learning_rate": 1.998711149962284e-05,
+      "loss": 1.0535,
+      "step": 653
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4852670498793803,
+      "learning_rate": 1.998711140490968e-05,
+      "loss": 1.3887,
+      "step": 654
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6490804855525398,
+      "learning_rate": 1.9987111310025406e-05,
+      "loss": 1.1665,
+      "step": 655
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4282243255866311,
+      "learning_rate": 1.998711121497001e-05,
+      "loss": 1.373,
+      "step": 656
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4489768924068471,
+      "learning_rate": 1.99871111197435e-05,
+      "loss": 1.3115,
+      "step": 657
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4947159201894757,
+      "learning_rate": 1.9987111024345876e-05,
+      "loss": 1.0076,
+      "step": 658
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.8907842017367373,
+      "learning_rate": 1.9987110928777134e-05,
+      "loss": 1.085,
+      "step": 659
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3443146783338529,
+      "learning_rate": 1.9987110833037274e-05,
+      "loss": 0.8779,
+      "step": 660
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3128185793838723,
+      "learning_rate": 1.9987110737126297e-05,
+      "loss": 0.8906,
+      "step": 661
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3565536247056234,
+      "learning_rate": 1.9987110641044205e-05,
+      "loss": 1.1406,
+      "step": 662
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3247892788004012,
+      "learning_rate": 1.9987110544790996e-05,
+      "loss": 1.0415,
+      "step": 663
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.545456052463225,
+      "learning_rate": 1.9987110448366673e-05,
+      "loss": 1.2065,
+      "step": 664
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3501599040521126,
+      "learning_rate": 1.998711035177123e-05,
+      "loss": 1.0681,
+      "step": 665
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3808037775007893,
+      "learning_rate": 1.998711025500467e-05,
+      "loss": 1.063,
+      "step": 666
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6253041202891707,
+      "learning_rate": 1.9987110158066995e-05,
+      "loss": 1.1885,
+      "step": 667
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4007815104121037,
+      "learning_rate": 1.9987110060958204e-05,
+      "loss": 1.1724,
+      "step": 668
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4629386211512707,
+      "learning_rate": 1.9987109963678297e-05,
+      "loss": 0.9282,
+      "step": 669
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.021802163517637,
+      "learning_rate": 1.998710986622727e-05,
+      "loss": 0.9714,
+      "step": 670
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5436687185534275,
+      "learning_rate": 1.998710976860513e-05,
+      "loss": 1.1479,
+      "step": 671
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.160774913713551,
+      "learning_rate": 1.9987109670811878e-05,
+      "loss": 0.9717,
+      "step": 672
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.230560493076619,
+      "learning_rate": 1.9987109572847503e-05,
+      "loss": 1.0293,
+      "step": 673
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3510419250974308,
+      "learning_rate": 1.998710947471201e-05,
+      "loss": 1.2441,
+      "step": 674
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6390329538306043,
+      "learning_rate": 1.9987109376405404e-05,
+      "loss": 0.9307,
+      "step": 675
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2157854067529505,
+      "learning_rate": 1.998710927792768e-05,
+      "loss": 0.9744,
+      "step": 676
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3413725824857745,
+      "learning_rate": 1.998710917927884e-05,
+      "loss": 0.9907,
+      "step": 677
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1392438181712974,
+      "learning_rate": 1.9987109080458886e-05,
+      "loss": 0.842,
+      "step": 678
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.7910797207510936,
+      "learning_rate": 1.9987108981467812e-05,
+      "loss": 1.0349,
+      "step": 679
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.7146485376146172,
+      "learning_rate": 1.9987108882305624e-05,
+      "loss": 1.249,
+      "step": 680
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4949146534725626,
+      "learning_rate": 1.998710878297232e-05,
+      "loss": 0.9182,
+      "step": 681
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2618244705550679,
+      "learning_rate": 1.99871086834679e-05,
+      "loss": 0.9727,
+      "step": 682
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3765949622146711,
+      "learning_rate": 1.998710858379236e-05,
+      "loss": 1.2476,
+      "step": 683
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.602171470423512,
+      "learning_rate": 1.9987108483945704e-05,
+      "loss": 1.0188,
+      "step": 684
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3913645117452391,
+      "learning_rate": 1.998710838392793e-05,
+      "loss": 1.2559,
+      "step": 685
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.264691184952601,
+      "learning_rate": 1.998710828373904e-05,
+      "loss": 1.141,
+      "step": 686
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.173459825163688,
+      "learning_rate": 1.9987108183379037e-05,
+      "loss": 1.103,
+      "step": 687
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.097596027717826,
+      "learning_rate": 1.9987108082847915e-05,
+      "loss": 0.8381,
+      "step": 688
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2034304342604811,
+      "learning_rate": 1.998710798214568e-05,
+      "loss": 0.9883,
+      "step": 689
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.7652333593228435,
+      "learning_rate": 1.9987107881272326e-05,
+      "loss": 1.1753,
+      "step": 690
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5108886332139433,
+      "learning_rate": 1.9987107780227855e-05,
+      "loss": 0.9995,
+      "step": 691
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3480552193342785,
+      "learning_rate": 1.998710767901227e-05,
+      "loss": 1.1069,
+      "step": 692
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.7728059963495015,
+      "learning_rate": 1.9987107577625566e-05,
+      "loss": 1.21,
+      "step": 693
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.0796451632070045,
+      "learning_rate": 1.9987107476067746e-05,
+      "loss": 0.7893,
+      "step": 694
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3219168131101844,
+      "learning_rate": 1.998710737433881e-05,
+      "loss": 1.0444,
+      "step": 695
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2528828279885713,
+      "learning_rate": 1.998710727243876e-05,
+      "loss": 0.9204,
+      "step": 696
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.86128579058543,
+      "learning_rate": 1.998710717036759e-05,
+      "loss": 1.0723,
+      "step": 697
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3041871076369294,
+      "learning_rate": 1.9987107068125304e-05,
+      "loss": 0.8728,
+      "step": 698
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3466484392619607,
+      "learning_rate": 1.9987106965711903e-05,
+      "loss": 1.1587,
+      "step": 699
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.288085289666488,
+      "learning_rate": 1.9987106863127383e-05,
+      "loss": 0.8987,
+      "step": 700
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5536798633510678,
+      "learning_rate": 1.998710676037175e-05,
+      "loss": 1.0635,
+      "step": 701
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5278129149740884,
+      "learning_rate": 1.9987106657445e-05,
+      "loss": 1.1748,
+      "step": 702
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.451362196916674,
+      "learning_rate": 1.998710655434713e-05,
+      "loss": 1.0684,
+      "step": 703
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4024265870516706,
+      "learning_rate": 1.9987106451078143e-05,
+      "loss": 1.0586,
+      "step": 704
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.207255271689983,
+      "learning_rate": 1.9987106347638046e-05,
+      "loss": 0.8878,
+      "step": 705
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.290699376140706,
+      "learning_rate": 1.9987106244026828e-05,
+      "loss": 1.0669,
+      "step": 706
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.59509430083373,
+      "learning_rate": 1.9987106140244495e-05,
+      "loss": 0.8689,
+      "step": 707
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1957916595871216,
+      "learning_rate": 1.9987106036291046e-05,
+      "loss": 1.083,
+      "step": 708
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.72105579031331,
+      "learning_rate": 1.9987105932166478e-05,
+      "loss": 1.1636,
+      "step": 709
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.382731462248507,
+      "learning_rate": 1.9987105827870796e-05,
+      "loss": 1.062,
+      "step": 710
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.551610998527054,
+      "learning_rate": 1.9987105723403997e-05,
+      "loss": 1.0952,
+      "step": 711
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3185997863841104,
+      "learning_rate": 1.998710561876608e-05,
+      "loss": 1.0718,
+      "step": 712
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2069513979673576,
+      "learning_rate": 1.9987105513957052e-05,
+      "loss": 1.0271,
+      "step": 713
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2577603252322975,
+      "learning_rate": 1.99871054089769e-05,
+      "loss": 1.2554,
+      "step": 714
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2935456432433567,
+      "learning_rate": 1.9987105303825637e-05,
+      "loss": 0.9082,
+      "step": 715
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4179547942344395,
+      "learning_rate": 1.9987105198503256e-05,
+      "loss": 1.0073,
+      "step": 716
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.328943112785479,
+      "learning_rate": 1.9987105093009758e-05,
+      "loss": 1.123,
+      "step": 717
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4098949665827412,
+      "learning_rate": 1.9987104987345142e-05,
+      "loss": 1.0293,
+      "step": 718
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4214344390123663,
+      "learning_rate": 1.9987104881509412e-05,
+      "loss": 1.0229,
+      "step": 719
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3416758209216337,
+      "learning_rate": 1.9987104775502565e-05,
+      "loss": 1.1436,
+      "step": 720
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4237725308505191,
+      "learning_rate": 1.9987104669324603e-05,
+      "loss": 1.1929,
+      "step": 721
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3687761504579778,
+      "learning_rate": 1.9987104562975523e-05,
+      "loss": 1.3022,
+      "step": 722
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3765289735554194,
+      "learning_rate": 1.9987104456455326e-05,
+      "loss": 1.1396,
+      "step": 723
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.541360137296102,
+      "learning_rate": 1.9987104349764015e-05,
+      "loss": 1.1387,
+      "step": 724
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3964313630399163,
+      "learning_rate": 1.9987104242901586e-05,
+      "loss": 1.2661,
+      "step": 725
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3132344870153054,
+      "learning_rate": 1.998710413586804e-05,
+      "loss": 0.803,
+      "step": 726
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.0713750757758809,
+      "learning_rate": 1.998710402866338e-05,
+      "loss": 0.9272,
+      "step": 727
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.367866304004786,
+      "learning_rate": 1.99871039212876e-05,
+      "loss": 1.0552,
+      "step": 728
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5459352200940286,
+      "learning_rate": 1.9987103813740706e-05,
+      "loss": 0.958,
+      "step": 729
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5067830895423084,
+      "learning_rate": 1.9987103706022696e-05,
+      "loss": 1.1226,
+      "step": 730
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6479653500367886,
+      "learning_rate": 1.9987103598133568e-05,
+      "loss": 1.2471,
+      "step": 731
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6666981773576608,
+      "learning_rate": 1.9987103490073326e-05,
+      "loss": 1.4688,
+      "step": 732
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.428583938680096,
+      "learning_rate": 1.9987103381841963e-05,
+      "loss": 1.0205,
+      "step": 733
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2731246563962941,
+      "learning_rate": 1.9987103273439486e-05,
+      "loss": 0.8911,
+      "step": 734
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4243107986209032,
+      "learning_rate": 1.9987103164865895e-05,
+      "loss": 1.03,
+      "step": 735
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3989436762598895,
+      "learning_rate": 1.9987103056121186e-05,
+      "loss": 1.417,
+      "step": 736
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4000947818006868,
+      "learning_rate": 1.9987102947205363e-05,
+      "loss": 1.1895,
+      "step": 737
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6918059618998056,
+      "learning_rate": 1.998710283811842e-05,
+      "loss": 1.3525,
+      "step": 738
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3003928031227066,
+      "learning_rate": 1.9987102728860358e-05,
+      "loss": 1.4688,
+      "step": 739
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.141722373345842,
+      "learning_rate": 1.9987102619431185e-05,
+      "loss": 0.9658,
+      "step": 740
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.573134283039164,
+      "learning_rate": 1.9987102509830895e-05,
+      "loss": 1.2236,
+      "step": 741
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4701910051828575,
+      "learning_rate": 1.9987102400059488e-05,
+      "loss": 1.0308,
+      "step": 742
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6167478839228668,
+      "learning_rate": 1.9987102290116963e-05,
+      "loss": 1.2148,
+      "step": 743
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.571759596241835,
+      "learning_rate": 1.9987102180003323e-05,
+      "loss": 0.9634,
+      "step": 744
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3890374798232403,
+      "learning_rate": 1.9987102069718566e-05,
+      "loss": 1.0757,
+      "step": 745
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.504401266695203,
+      "learning_rate": 1.9987101959262695e-05,
+      "loss": 1.2969,
+      "step": 746
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.7050286948958047,
+      "learning_rate": 1.9987101848635703e-05,
+      "loss": 1.1855,
+      "step": 747
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.3302342633535655,
+      "learning_rate": 1.99871017378376e-05,
+      "loss": 1.0029,
+      "step": 748
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1726586836777555,
+      "learning_rate": 1.9987101626868376e-05,
+      "loss": 0.8875,
+      "step": 749
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1309768536675715,
+      "learning_rate": 1.9987101515728038e-05,
+      "loss": 0.776,
+      "step": 750
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.420652681576941,
+      "learning_rate": 1.9987101404416585e-05,
+      "loss": 0.9873,
+      "step": 751
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.386190437365037,
+      "learning_rate": 1.9987101292934015e-05,
+      "loss": 1.1187,
+      "step": 752
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3566620161334535,
+      "learning_rate": 1.9987101181280328e-05,
+      "loss": 0.8821,
+      "step": 753
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.9408317345230028,
+      "learning_rate": 1.9987101069455522e-05,
+      "loss": 1.0693,
+      "step": 754
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.56971805385568,
+      "learning_rate": 1.9987100957459603e-05,
+      "loss": 1.1768,
+      "step": 755
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4630337939215117,
+      "learning_rate": 1.9987100845292566e-05,
+      "loss": 0.908,
+      "step": 756
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1405404987126921,
+      "learning_rate": 1.998710073295441e-05,
+      "loss": 0.9304,
+      "step": 757
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1854335975890458,
+      "learning_rate": 1.9987100620445146e-05,
+      "loss": 0.9138,
+      "step": 758
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1961755566368726,
+      "learning_rate": 1.998710050776476e-05,
+      "loss": 0.9028,
+      "step": 759
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.816168788031518,
+      "learning_rate": 1.998710039491326e-05,
+      "loss": 1.0928,
+      "step": 760
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3122649209352832,
+      "learning_rate": 1.9987100281890637e-05,
+      "loss": 0.7943,
+      "step": 761
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3824356772581814,
+      "learning_rate": 1.9987100168696905e-05,
+      "loss": 0.9326,
+      "step": 762
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.9970254832414989,
+      "learning_rate": 1.9987100055332055e-05,
+      "loss": 1.2017,
+      "step": 763
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.0152023227107345,
+      "learning_rate": 1.9987099941796087e-05,
+      "loss": 1.0283,
+      "step": 764
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2097719551391337,
+      "learning_rate": 1.9987099828089006e-05,
+      "loss": 0.7932,
+      "step": 765
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.315002915919447,
+      "learning_rate": 1.9987099714210806e-05,
+      "loss": 0.7952,
+      "step": 766
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5626288551604641,
+      "learning_rate": 1.998709960016149e-05,
+      "loss": 0.9338,
+      "step": 767
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3844585643201008,
+      "learning_rate": 1.9987099485941058e-05,
+      "loss": 0.9922,
+      "step": 768
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3022254052537359,
+      "learning_rate": 1.998709937154951e-05,
+      "loss": 0.9419,
+      "step": 769
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.337784481425636,
+      "learning_rate": 1.9987099256986843e-05,
+      "loss": 0.9673,
+      "step": 770
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.8603991725166038,
+      "learning_rate": 1.9987099142253062e-05,
+      "loss": 0.9731,
+      "step": 771
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5100064301511766,
+      "learning_rate": 1.9987099027348164e-05,
+      "loss": 0.8608,
+      "step": 772
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3202460007031802,
+      "learning_rate": 1.998709891227215e-05,
+      "loss": 0.8213,
+      "step": 773
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5315124909541973,
+      "learning_rate": 1.998709879702502e-05,
+      "loss": 1.0015,
+      "step": 774
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.211235649567147,
+      "learning_rate": 1.9987098681606774e-05,
+      "loss": 0.998,
+      "step": 775
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4279629671760472,
+      "learning_rate": 1.9987098566017412e-05,
+      "loss": 1.0015,
+      "step": 776
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6089934850528844,
+      "learning_rate": 1.9987098450256933e-05,
+      "loss": 0.9775,
+      "step": 777
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6054628098278185,
+      "learning_rate": 1.9987098334325336e-05,
+      "loss": 0.8777,
+      "step": 778
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6315309158421591,
+      "learning_rate": 1.9987098218222625e-05,
+      "loss": 1.4512,
+      "step": 779
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.7004193854534184,
+      "learning_rate": 1.99870981019488e-05,
+      "loss": 1.063,
+      "step": 780
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5688582181649218,
+      "learning_rate": 1.9987097985503853e-05,
+      "loss": 1.1738,
+      "step": 781
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.328985227360653,
+      "learning_rate": 1.9987097868887792e-05,
+      "loss": 1.1372,
+      "step": 782
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2720145952628958,
+      "learning_rate": 1.9987097752100614e-05,
+      "loss": 1.0957,
+      "step": 783
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.255102187400032,
+      "learning_rate": 1.998709763514232e-05,
+      "loss": 1.136,
+      "step": 784
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.9222227572781523,
+      "learning_rate": 1.998709751801291e-05,
+      "loss": 1.124,
+      "step": 785
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.2078018868549645,
+      "learning_rate": 1.9987097400712387e-05,
+      "loss": 0.8569,
+      "step": 786
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2309476374653963,
+      "learning_rate": 1.9987097283240745e-05,
+      "loss": 0.8304,
+      "step": 787
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3484528771647524,
+      "learning_rate": 1.9987097165597986e-05,
+      "loss": 0.8293,
+      "step": 788
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4499400521086596,
+      "learning_rate": 1.9987097047784116e-05,
+      "loss": 0.9702,
+      "step": 789
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5028887588847124,
+      "learning_rate": 1.998709692979912e-05,
+      "loss": 1.0757,
+      "step": 790
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.285102773556734,
+      "learning_rate": 1.9987096811643012e-05,
+      "loss": 1.1536,
+      "step": 791
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.0089337521213861,
+      "learning_rate": 1.9987096693315793e-05,
+      "loss": 0.6887,
+      "step": 792
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.480111699997287,
+      "learning_rate": 1.9987096574817452e-05,
+      "loss": 1.3062,
+      "step": 793
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4345337491085886,
+      "learning_rate": 1.9987096456147997e-05,
+      "loss": 1.332,
+      "step": 794
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.134974659036074,
+      "learning_rate": 1.9987096337307425e-05,
+      "loss": 0.8276,
+      "step": 795
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4965801991985115,
+      "learning_rate": 1.9987096218295735e-05,
+      "loss": 1.0957,
+      "step": 796
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.222256726201776,
+      "learning_rate": 1.998709609911293e-05,
+      "loss": 1.0752,
+      "step": 797
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.7575671215453084,
+      "learning_rate": 1.998709597975901e-05,
+      "loss": 1.0083,
+      "step": 798
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4720840759160299,
+      "learning_rate": 1.9987095860233973e-05,
+      "loss": 1.1504,
+      "step": 799
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2665481967904304,
+      "learning_rate": 1.998709574053782e-05,
+      "loss": 1.0571,
+      "step": 800
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4942725029882085,
+      "learning_rate": 1.998709562067055e-05,
+      "loss": 1.0281,
+      "step": 801
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4656890464729986,
+      "learning_rate": 1.9987095500632163e-05,
+      "loss": 1.1528,
+      "step": 802
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3554836129464756,
+      "learning_rate": 1.9987095380422663e-05,
+      "loss": 0.843,
+      "step": 803
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.894606369217556,
+      "learning_rate": 1.9987095260042046e-05,
+      "loss": 1.0195,
+      "step": 804
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1050881313546503,
+      "learning_rate": 1.9987095139490307e-05,
+      "loss": 1.0964,
+      "step": 805
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3231647539180627,
+      "learning_rate": 1.998709501876746e-05,
+      "loss": 0.9297,
+      "step": 806
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5019149477785743,
+      "learning_rate": 1.9987094897873492e-05,
+      "loss": 1.2041,
+      "step": 807
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.415103255369782,
+      "learning_rate": 1.9987094776808408e-05,
+      "loss": 1.1363,
+      "step": 808
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4292438423220264,
+      "learning_rate": 1.998709465557221e-05,
+      "loss": 1.0215,
+      "step": 809
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3631496857826377,
+      "learning_rate": 1.998709453416489e-05,
+      "loss": 1.1367,
+      "step": 810
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5958988438532098,
+      "learning_rate": 1.998709441258646e-05,
+      "loss": 0.8403,
+      "step": 811
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2404648934547142,
+      "learning_rate": 1.9987094290836908e-05,
+      "loss": 0.9897,
+      "step": 812
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2901676411337786,
+      "learning_rate": 1.9987094168916246e-05,
+      "loss": 1.1948,
+      "step": 813
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5824231827880457,
+      "learning_rate": 1.9987094046824467e-05,
+      "loss": 1.1421,
+      "step": 814
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.34019990419172,
+      "learning_rate": 1.998709392456157e-05,
+      "loss": 0.9907,
+      "step": 815
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3852494313352994,
+      "learning_rate": 1.9987093802127558e-05,
+      "loss": 1.3809,
+      "step": 816
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.086899781881123,
+      "learning_rate": 1.9987093679522425e-05,
+      "loss": 1.0034,
+      "step": 817
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2579824676353277,
+      "learning_rate": 1.9987093556746182e-05,
+      "loss": 1.2119,
+      "step": 818
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.561370594018747,
+      "learning_rate": 1.9987093433798818e-05,
+      "loss": 0.8748,
+      "step": 819
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5032734121545135,
+      "learning_rate": 1.998709331068034e-05,
+      "loss": 1.1152,
+      "step": 820
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3814552672437526,
+      "learning_rate": 1.9987093187390747e-05,
+      "loss": 1.1843,
+      "step": 821
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3504626258500267,
+      "learning_rate": 1.9987093063930037e-05,
+      "loss": 0.9717,
+      "step": 822
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2147077380026143,
+      "learning_rate": 1.9987092940298212e-05,
+      "loss": 1.0317,
+      "step": 823
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3930592608611219,
+      "learning_rate": 1.9987092816495267e-05,
+      "loss": 1.1606,
+      "step": 824
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.442497935830793,
+      "learning_rate": 1.998709269252121e-05,
+      "loss": 1.2671,
+      "step": 825
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4213934281982663,
+      "learning_rate": 1.9987092568376033e-05,
+      "loss": 1.2305,
+      "step": 826
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3760625894916132,
+      "learning_rate": 1.9987092444059742e-05,
+      "loss": 1.05,
+      "step": 827
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3814721804788586,
+      "learning_rate": 1.9987092319572333e-05,
+      "loss": 1.0588,
+      "step": 828
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.351643157492701,
+      "learning_rate": 1.998709219491381e-05,
+      "loss": 0.9736,
+      "step": 829
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4654418538069118,
+      "learning_rate": 1.998709207008417e-05,
+      "loss": 0.9937,
+      "step": 830
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5263068931029544,
+      "learning_rate": 1.9987091945083414e-05,
+      "loss": 0.9033,
+      "step": 831
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.7842237177919213,
+      "learning_rate": 1.998709181991154e-05,
+      "loss": 1.1143,
+      "step": 832
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2578003213423168,
+      "learning_rate": 1.9987091694568554e-05,
+      "loss": 1.0737,
+      "step": 833
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.290251582253559,
+      "learning_rate": 1.9987091569054446e-05,
+      "loss": 1.106,
+      "step": 834
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5820525744849467,
+      "learning_rate": 1.9987091443369228e-05,
+      "loss": 0.9775,
+      "step": 835
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.034803715080408,
+      "learning_rate": 1.9987091317512888e-05,
+      "loss": 1.5298,
+      "step": 836
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.370775017275762,
+      "learning_rate": 1.9987091191485434e-05,
+      "loss": 1.0808,
+      "step": 837
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.203426372879798,
+      "learning_rate": 1.9987091065286866e-05,
+      "loss": 0.8484,
+      "step": 838
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.472281491787967,
+      "learning_rate": 1.998709093891718e-05,
+      "loss": 1.2417,
+      "step": 839
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.35629893447563,
+      "learning_rate": 1.998709081237638e-05,
+      "loss": 1.0454,
+      "step": 840
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3418428504763331,
+      "learning_rate": 1.998709068566446e-05,
+      "loss": 1.2407,
+      "step": 841
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.597578349930757,
+      "learning_rate": 1.9987090558781428e-05,
+      "loss": 1.1807,
+      "step": 842
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2311407282794153,
+      "learning_rate": 1.9987090431727275e-05,
+      "loss": 0.8896,
+      "step": 843
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4178008513448201,
+      "learning_rate": 1.998709030450201e-05,
+      "loss": 0.8955,
+      "step": 844
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5618513668328484,
+      "learning_rate": 1.9987090177105627e-05,
+      "loss": 1.2349,
+      "step": 845
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4270898118069286,
+      "learning_rate": 1.998709004953813e-05,
+      "loss": 1.1602,
+      "step": 846
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.351973763225415,
+      "learning_rate": 1.9987089921799512e-05,
+      "loss": 0.9619,
+      "step": 847
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3401014341565456,
+      "learning_rate": 1.9987089793889782e-05,
+      "loss": 1.0742,
+      "step": 848
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6887019786317796,
+      "learning_rate": 1.9987089665808934e-05,
+      "loss": 1.2412,
+      "step": 849
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.088422231200263,
+      "learning_rate": 1.998708953755697e-05,
+      "loss": 1.1689,
+      "step": 850
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4060000727342827,
+      "learning_rate": 1.9987089409133892e-05,
+      "loss": 1.0312,
+      "step": 851
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.468620943933195,
+      "learning_rate": 1.9987089280539694e-05,
+      "loss": 1.3384,
+      "step": 852
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2095174523175067,
+      "learning_rate": 1.9987089151774383e-05,
+      "loss": 0.9453,
+      "step": 853
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5398079088472956,
+      "learning_rate": 1.9987089022837957e-05,
+      "loss": 1.0903,
+      "step": 854
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3921664834710665,
+      "learning_rate": 1.998708889373041e-05,
+      "loss": 1.1382,
+      "step": 855
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3964311496221613,
+      "learning_rate": 1.998708876445175e-05,
+      "loss": 1.1187,
+      "step": 856
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.466906669154562,
+      "learning_rate": 1.9987088635001974e-05,
+      "loss": 1.1182,
+      "step": 857
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5142837567458134,
+      "learning_rate": 1.9987088505381084e-05,
+      "loss": 1.1387,
+      "step": 858
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1370781399189034,
+      "learning_rate": 1.9987088375589074e-05,
+      "loss": 0.8085,
+      "step": 859
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4665875037821643,
+      "learning_rate": 1.998708824562595e-05,
+      "loss": 0.7821,
+      "step": 860
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3791529621161744,
+      "learning_rate": 1.998708811549171e-05,
+      "loss": 1.2798,
+      "step": 861
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5197005385997149,
+      "learning_rate": 1.998708798518635e-05,
+      "loss": 0.9878,
+      "step": 862
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3488411821826511,
+      "learning_rate": 1.9987087854709876e-05,
+      "loss": 1.1792,
+      "step": 863
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2142359939782157,
+      "learning_rate": 1.9987087724062288e-05,
+      "loss": 0.8457,
+      "step": 864
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3530979585478393,
+      "learning_rate": 1.9987087593243582e-05,
+      "loss": 1.3516,
+      "step": 865
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6397092897666035,
+      "learning_rate": 1.998708746225376e-05,
+      "loss": 1.0549,
+      "step": 866
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.6138026881916416,
+      "learning_rate": 1.998708733109282e-05,
+      "loss": 1.1567,
+      "step": 867
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3154816139354335,
+      "learning_rate": 1.998708719976077e-05,
+      "loss": 1.0815,
+      "step": 868
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.173674105220784,
+      "learning_rate": 1.99870870682576e-05,
+      "loss": 0.9167,
+      "step": 869
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3137354258882203,
+      "learning_rate": 1.9987086936583316e-05,
+      "loss": 1.1519,
+      "step": 870
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5185882859024293,
+      "learning_rate": 1.998708680473791e-05,
+      "loss": 1.2812,
+      "step": 871
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3709848909554136,
+      "learning_rate": 1.9987086672721392e-05,
+      "loss": 1.0547,
+      "step": 872
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3325545997240518,
+      "learning_rate": 1.998708654053376e-05,
+      "loss": 0.9304,
+      "step": 873
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.386636049195486,
+      "learning_rate": 1.998708640817501e-05,
+      "loss": 1.0137,
+      "step": 874
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2007196831292133,
+      "learning_rate": 1.9987086275645143e-05,
+      "loss": 0.9956,
+      "step": 875
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.311727250908302,
+      "learning_rate": 1.998708614294416e-05,
+      "loss": 1.1835,
+      "step": 876
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1115493956386744,
+      "learning_rate": 1.998708601007206e-05,
+      "loss": 0.9232,
+      "step": 877
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.172070143664195,
+      "learning_rate": 1.9987085877028846e-05,
+      "loss": 0.9355,
+      "step": 878
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.447338025216525,
+      "learning_rate": 1.9987085743814515e-05,
+      "loss": 1.1665,
+      "step": 879
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1706545259954808,
+      "learning_rate": 1.998708561042907e-05,
+      "loss": 0.8203,
+      "step": 880
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3880558069151838,
+      "learning_rate": 1.9987085476872505e-05,
+      "loss": 1.0071,
+      "step": 881
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1667808692531616,
+      "learning_rate": 1.9987085343144827e-05,
+      "loss": 1.0264,
+      "step": 882
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5148075691612235,
+      "learning_rate": 1.998708520924603e-05,
+      "loss": 1.2673,
+      "step": 883
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.386801789712205,
+      "learning_rate": 1.998708507517612e-05,
+      "loss": 1.312,
+      "step": 884
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.368463672667285,
+      "learning_rate": 1.9987084940935094e-05,
+      "loss": 0.8694,
+      "step": 885
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.182389858834444,
+      "learning_rate": 1.998708480652295e-05,
+      "loss": 0.8215,
+      "step": 886
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.0986447481990533,
+      "learning_rate": 1.998708467193969e-05,
+      "loss": 0.7944,
+      "step": 887
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3674717634866946,
+      "learning_rate": 1.9987084537185317e-05,
+      "loss": 1.3647,
+      "step": 888
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3206806008314564,
+      "learning_rate": 1.9987084402259823e-05,
+      "loss": 0.8439,
+      "step": 889
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3219431000611161,
+      "learning_rate": 1.9987084267163218e-05,
+      "loss": 0.9321,
+      "step": 890
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4836284968850555,
+      "learning_rate": 1.998708413189549e-05,
+      "loss": 0.9932,
+      "step": 891
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3668394681187164,
+      "learning_rate": 1.998708399645665e-05,
+      "loss": 0.9336,
+      "step": 892
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.320958763729333,
+      "learning_rate": 1.9987083860846697e-05,
+      "loss": 1.0522,
+      "step": 893
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2960251470969435,
+      "learning_rate": 1.9987083725065628e-05,
+      "loss": 1.0601,
+      "step": 894
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2879063363197292,
+      "learning_rate": 1.9987083589113438e-05,
+      "loss": 1.1995,
+      "step": 895
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4173808074301995,
+      "learning_rate": 1.9987083452990134e-05,
+      "loss": 1.1328,
+      "step": 896
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.647653329918102,
+      "learning_rate": 1.9987083316695716e-05,
+      "loss": 0.8931,
+      "step": 897
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5179972223349427,
+      "learning_rate": 1.998708318023018e-05,
+      "loss": 1.127,
+      "step": 898
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.076814712191796,
+      "learning_rate": 1.9987083043593527e-05,
+      "loss": 0.7616,
+      "step": 899
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.276015979147599,
+      "learning_rate": 1.9987082906785756e-05,
+      "loss": 0.9219,
+      "step": 900
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1857722412281888,
+      "learning_rate": 1.9987082769806874e-05,
+      "loss": 0.9678,
+      "step": 901
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3479789900288521,
+      "learning_rate": 1.998708263265687e-05,
+      "loss": 0.9089,
+      "step": 902
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3195777970681604,
+      "learning_rate": 1.9987082495335758e-05,
+      "loss": 0.9951,
+      "step": 903
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3581419866546747,
+      "learning_rate": 1.9987082357843524e-05,
+      "loss": 1.123,
+      "step": 904
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5727848309304446,
+      "learning_rate": 1.9987082220180175e-05,
+      "loss": 0.9819,
+      "step": 905
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3205555799285698,
+      "learning_rate": 1.9987082082345715e-05,
+      "loss": 0.9795,
+      "step": 906
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.317048230393812,
+      "learning_rate": 1.998708194434013e-05,
+      "loss": 1.1035,
+      "step": 907
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.431071710934343,
+      "learning_rate": 1.9987081806163437e-05,
+      "loss": 1.1475,
+      "step": 908
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1609133937381462,
+      "learning_rate": 1.9987081667815624e-05,
+      "loss": 0.814,
+      "step": 909
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.384707730781913,
+      "learning_rate": 1.9987081529296698e-05,
+      "loss": 0.8794,
+      "step": 910
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5367797342988712,
+      "learning_rate": 1.9987081390606654e-05,
+      "loss": 1.0771,
+      "step": 911
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2577939713339525,
+      "learning_rate": 1.9987081251745492e-05,
+      "loss": 0.9978,
+      "step": 912
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2155792855343357,
+      "learning_rate": 1.9987081112713216e-05,
+      "loss": 0.8826,
+      "step": 913
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.194446673994916,
+      "learning_rate": 1.9987080973509822e-05,
+      "loss": 0.9719,
+      "step": 914
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.7847968142400237,
+      "learning_rate": 1.9987080834135315e-05,
+      "loss": 1.1602,
+      "step": 915
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.358919089148,
+      "learning_rate": 1.9987080694589693e-05,
+      "loss": 0.963,
+      "step": 916
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6045487266790097,
+      "learning_rate": 1.998708055487295e-05,
+      "loss": 1.2173,
+      "step": 917
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2325892508575802,
+      "learning_rate": 1.9987080414985096e-05,
+      "loss": 1.2583,
+      "step": 918
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3760215259187882,
+      "learning_rate": 1.998708027492612e-05,
+      "loss": 1.0178,
+      "step": 919
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2209938130452385,
+      "learning_rate": 1.9987080134696036e-05,
+      "loss": 0.9644,
+      "step": 920
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4837495791298865,
+      "learning_rate": 1.9987079994294832e-05,
+      "loss": 1.2671,
+      "step": 921
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1712147696620374,
+      "learning_rate": 1.998707985372251e-05,
+      "loss": 0.8811,
+      "step": 922
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.0794699752010404,
+      "learning_rate": 1.9987079712979073e-05,
+      "loss": 0.7729,
+      "step": 923
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3445345783801999,
+      "learning_rate": 1.998707957206452e-05,
+      "loss": 1.0173,
+      "step": 924
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.7472072524194444,
+      "learning_rate": 1.9987079430978854e-05,
+      "loss": 0.9331,
+      "step": 925
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.543046703059527,
+      "learning_rate": 1.998707928972207e-05,
+      "loss": 1.0269,
+      "step": 926
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.324626128513467,
+      "learning_rate": 1.9987079148294168e-05,
+      "loss": 1.1631,
+      "step": 927
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1854381228596318,
+      "learning_rate": 1.9987079006695155e-05,
+      "loss": 1.1416,
+      "step": 928
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2027428255344925,
+      "learning_rate": 1.998707886492502e-05,
+      "loss": 0.8867,
+      "step": 929
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2220711572567051,
+      "learning_rate": 1.9987078722983773e-05,
+      "loss": 0.9329,
+      "step": 930
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1448460168869423,
+      "learning_rate": 1.998707858087141e-05,
+      "loss": 0.9451,
+      "step": 931
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4958335709992259,
+      "learning_rate": 1.9987078438587928e-05,
+      "loss": 1.4434,
+      "step": 932
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.346344992669711,
+      "learning_rate": 1.9987078296133334e-05,
+      "loss": 1.0913,
+      "step": 933
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3363698516281248,
+      "learning_rate": 1.9987078153507622e-05,
+      "loss": 0.8296,
+      "step": 934
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3589696170085601,
+      "learning_rate": 1.9987078010710793e-05,
+      "loss": 0.9023,
+      "step": 935
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4344616167764417,
+      "learning_rate": 1.998707786774285e-05,
+      "loss": 0.979,
+      "step": 936
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2713608923513668,
+      "learning_rate": 1.9987077724603792e-05,
+      "loss": 0.9534,
+      "step": 937
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2713932409015922,
+      "learning_rate": 1.9987077581293617e-05,
+      "loss": 0.8833,
+      "step": 938
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.503750245561207,
+      "learning_rate": 1.9987077437812324e-05,
+      "loss": 1.0559,
+      "step": 939
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3532982413592054,
+      "learning_rate": 1.9987077294159917e-05,
+      "loss": 1.062,
+      "step": 940
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1098027546360176,
+      "learning_rate": 1.9987077150336392e-05,
+      "loss": 0.9761,
+      "step": 941
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5020020951918047,
+      "learning_rate": 1.9987077006341753e-05,
+      "loss": 1.1499,
+      "step": 942
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.728181451078964,
+      "learning_rate": 1.9987076862175997e-05,
+      "loss": 1.1675,
+      "step": 943
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.9714985386850048,
+      "learning_rate": 1.998707671783913e-05,
+      "loss": 0.9001,
+      "step": 944
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4464476380613716,
+      "learning_rate": 1.9987076573331138e-05,
+      "loss": 1.0903,
+      "step": 945
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4606544914092194,
+      "learning_rate": 1.9987076428652036e-05,
+      "loss": 0.9028,
+      "step": 946
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.272069231053658,
+      "learning_rate": 1.998707628380182e-05,
+      "loss": 1.0962,
+      "step": 947
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.7230595745828539,
+      "learning_rate": 1.998707613878048e-05,
+      "loss": 0.8521,
+      "step": 948
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.300503812878925,
+      "learning_rate": 1.998707599358803e-05,
+      "loss": 1.1948,
+      "step": 949
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.5942669040621786,
+      "learning_rate": 1.9987075848224467e-05,
+      "loss": 1.3125,
+      "step": 950
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.0863721238505213,
+      "learning_rate": 1.9987075702689784e-05,
+      "loss": 0.6089,
+      "step": 951
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2076364129769672,
+      "learning_rate": 1.9987075556983986e-05,
+      "loss": 0.9429,
+      "step": 952
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3794536422622228,
+      "learning_rate": 1.998707541110707e-05,
+      "loss": 1.1157,
+      "step": 953
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4282720677884286,
+      "learning_rate": 1.998707526505904e-05,
+      "loss": 1.3086,
+      "step": 954
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.911294734373929,
+      "learning_rate": 1.9987075118839893e-05,
+      "loss": 1.1006,
+      "step": 955
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.0824802169577883,
+      "learning_rate": 1.9987074972449632e-05,
+      "loss": 0.8625,
+      "step": 956
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.149159314909156,
+      "learning_rate": 1.9987074825888253e-05,
+      "loss": 0.8418,
+      "step": 957
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.0282677965801343,
+      "learning_rate": 1.998707467915576e-05,
+      "loss": 0.7828,
+      "step": 958
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3897086564056225,
+      "learning_rate": 1.998707453225215e-05,
+      "loss": 0.9727,
+      "step": 959
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.021249538178589,
+      "learning_rate": 1.9987074385177424e-05,
+      "loss": 1.3066,
+      "step": 960
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.201796543738665,
+      "learning_rate": 1.998707423793158e-05,
+      "loss": 1.0884,
+      "step": 961
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1668547864653698,
+      "learning_rate": 1.9987074090514624e-05,
+      "loss": 0.8582,
+      "step": 962
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5007328786208534,
+      "learning_rate": 1.9987073942926553e-05,
+      "loss": 1.4854,
+      "step": 963
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3901716629208065,
+      "learning_rate": 1.998707379516736e-05,
+      "loss": 1.1118,
+      "step": 964
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.8653776583137223,
+      "learning_rate": 1.9987073647237055e-05,
+      "loss": 1.3545,
+      "step": 965
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.56213130415653,
+      "learning_rate": 1.9987073499135635e-05,
+      "loss": 1.1133,
+      "step": 966
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3848714640083555,
+      "learning_rate": 1.9987073350863097e-05,
+      "loss": 1.0474,
+      "step": 967
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.373882359661735,
+      "learning_rate": 1.9987073202419444e-05,
+      "loss": 0.8799,
+      "step": 968
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1127207633441214,
+      "learning_rate": 1.9987073053804678e-05,
+      "loss": 0.7136,
+      "step": 969
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5057635839648527,
+      "learning_rate": 1.9987072905018794e-05,
+      "loss": 1.0786,
+      "step": 970
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4512359481775232,
+      "learning_rate": 1.9987072756061792e-05,
+      "loss": 1.0767,
+      "step": 971
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2935991853539208,
+      "learning_rate": 1.9987072606933676e-05,
+      "loss": 0.9043,
+      "step": 972
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5209018831176055,
+      "learning_rate": 1.9987072457634446e-05,
+      "loss": 0.9844,
+      "step": 973
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3792282031219223,
+      "learning_rate": 1.99870723081641e-05,
+      "loss": 1.064,
+      "step": 974
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.5484618877504626,
+      "learning_rate": 1.9987072158522633e-05,
+      "loss": 1.2065,
+      "step": 975
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2008882513378307,
+      "learning_rate": 1.9987072008710054e-05,
+      "loss": 1.0464,
+      "step": 976
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.8986969444793178,
+      "learning_rate": 1.998707185872636e-05,
+      "loss": 1.1172,
+      "step": 977
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3350536702115297,
+      "learning_rate": 1.9987071708571546e-05,
+      "loss": 0.8992,
+      "step": 978
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5173070167115277,
+      "learning_rate": 1.998707155824562e-05,
+      "loss": 1.2266,
+      "step": 979
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4124116903735762,
+      "learning_rate": 1.9987071407748577e-05,
+      "loss": 1.1138,
+      "step": 980
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3352253225817725,
+      "learning_rate": 1.998707125708042e-05,
+      "loss": 1.0955,
+      "step": 981
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4522892999727808,
+      "learning_rate": 1.9987071106241145e-05,
+      "loss": 1.2671,
+      "step": 982
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5706098284521814,
+      "learning_rate": 1.9987070955230752e-05,
+      "loss": 0.9414,
+      "step": 983
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3449523780503634,
+      "learning_rate": 1.9987070804049246e-05,
+      "loss": 0.8872,
+      "step": 984
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5850751815679072,
+      "learning_rate": 1.9987070652696625e-05,
+      "loss": 0.9785,
+      "step": 985
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.165208239874497,
+      "learning_rate": 1.9987070501172886e-05,
+      "loss": 0.7468,
+      "step": 986
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2309327234745444,
+      "learning_rate": 1.9987070349478034e-05,
+      "loss": 1.0417,
+      "step": 987
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3223209341362328,
+      "learning_rate": 1.9987070197612063e-05,
+      "loss": 1.2329,
+      "step": 988
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4816674686396816,
+      "learning_rate": 1.998707004557498e-05,
+      "loss": 1.1089,
+      "step": 989
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5255133155784253,
+      "learning_rate": 1.9987069893366777e-05,
+      "loss": 1.3491,
+      "step": 990
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3819343666633588,
+      "learning_rate": 1.998706974098746e-05,
+      "loss": 1.3428,
+      "step": 991
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2381302413559898,
+      "learning_rate": 1.9987069588437026e-05,
+      "loss": 1.2075,
+      "step": 992
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.461217761569255,
+      "learning_rate": 1.998706943571548e-05,
+      "loss": 1.1909,
+      "step": 993
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1254364862336295,
+      "learning_rate": 1.9987069282822813e-05,
+      "loss": 0.6924,
+      "step": 994
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6500349214499657,
+      "learning_rate": 1.9987069129759033e-05,
+      "loss": 1.0767,
+      "step": 995
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4546834320765407,
+      "learning_rate": 1.998706897652414e-05,
+      "loss": 1.0977,
+      "step": 996
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.252559425786882,
+      "learning_rate": 1.9987068823118127e-05,
+      "loss": 1.0679,
+      "step": 997
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5876222878761754,
+      "learning_rate": 1.9987068669540998e-05,
+      "loss": 1.2803,
+      "step": 998
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2116287227515572,
+      "learning_rate": 1.9987068515792754e-05,
+      "loss": 1.1375,
+      "step": 999
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.240425683833313,
+      "learning_rate": 1.9987068361873396e-05,
+      "loss": 1.0547,
+      "step": 1000
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2933436654028845,
+      "learning_rate": 1.998706820778292e-05,
+      "loss": 0.7144,
+      "step": 1001
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1963323593089625,
+      "learning_rate": 1.998706805352133e-05,
+      "loss": 0.8813,
+      "step": 1002
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4549715348700485,
+      "learning_rate": 1.9987067899088624e-05,
+      "loss": 1.0391,
+      "step": 1003
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2333546054569298,
+      "learning_rate": 1.99870677444848e-05,
+      "loss": 0.7906,
+      "step": 1004
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.52465044299091,
+      "learning_rate": 1.9987067589709864e-05,
+      "loss": 1.0142,
+      "step": 1005
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4930032627302572,
+      "learning_rate": 1.9987067434763808e-05,
+      "loss": 0.9639,
+      "step": 1006
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4642384003096232,
+      "learning_rate": 1.998706727964664e-05,
+      "loss": 1.0327,
+      "step": 1007
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5414143519084702,
+      "learning_rate": 1.9987067124358355e-05,
+      "loss": 1.3159,
+      "step": 1008
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4039205967547408,
+      "learning_rate": 1.9987066968898953e-05,
+      "loss": 1.2461,
+      "step": 1009
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2414071372949058,
+      "learning_rate": 1.998706681326844e-05,
+      "loss": 0.9956,
+      "step": 1010
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2983472808819996,
+      "learning_rate": 1.9987066657466805e-05,
+      "loss": 0.8813,
+      "step": 1011
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.761422278247473,
+      "learning_rate": 1.9987066501494056e-05,
+      "loss": 0.9802,
+      "step": 1012
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.479746619778649,
+      "learning_rate": 1.9987066345350194e-05,
+      "loss": 1.1113,
+      "step": 1013
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2514074508122057,
+      "learning_rate": 1.9987066189035213e-05,
+      "loss": 0.8638,
+      "step": 1014
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3598011708072335,
+      "learning_rate": 1.9987066032549115e-05,
+      "loss": 0.8608,
+      "step": 1015
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.549200140783105,
+      "learning_rate": 1.9987065875891903e-05,
+      "loss": 0.9316,
+      "step": 1016
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2272549757974123,
+      "learning_rate": 1.9987065719063577e-05,
+      "loss": 1.0112,
+      "step": 1017
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3991878350621656,
+      "learning_rate": 1.9987065562064136e-05,
+      "loss": 1.0027,
+      "step": 1018
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2909559273587736,
+      "learning_rate": 1.9987065404893578e-05,
+      "loss": 1.146,
+      "step": 1019
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1112358175602057,
+      "learning_rate": 1.9987065247551902e-05,
+      "loss": 1.0652,
+      "step": 1020
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5003957226425053,
+      "learning_rate": 1.9987065090039112e-05,
+      "loss": 1.0273,
+      "step": 1021
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5152362442890153,
+      "learning_rate": 1.9987064932355208e-05,
+      "loss": 1.1184,
+      "step": 1022
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3607195639177156,
+      "learning_rate": 1.9987064774500183e-05,
+      "loss": 0.9226,
+      "step": 1023
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.698345231653125,
+      "learning_rate": 1.9987064616474047e-05,
+      "loss": 1.1123,
+      "step": 1024
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5695758818930627,
+      "learning_rate": 1.9987064458276793e-05,
+      "loss": 1.1885,
+      "step": 1025
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4696271184692014,
+      "learning_rate": 1.9987064299908426e-05,
+      "loss": 0.9551,
+      "step": 1026
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6599879280478664,
+      "learning_rate": 1.998706414136894e-05,
+      "loss": 1.1338,
+      "step": 1027
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.491199345971443,
+      "learning_rate": 1.998706398265834e-05,
+      "loss": 0.9976,
+      "step": 1028
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.414433214689594,
+      "learning_rate": 1.9987063823776624e-05,
+      "loss": 1.0718,
+      "step": 1029
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4469076066431696,
+      "learning_rate": 1.9987063664723792e-05,
+      "loss": 0.9719,
+      "step": 1030
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4665110140626512,
+      "learning_rate": 1.9987063505499847e-05,
+      "loss": 1.0483,
+      "step": 1031
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4347099096051572,
+      "learning_rate": 1.9987063346104783e-05,
+      "loss": 1.165,
+      "step": 1032
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.878622307210149,
+      "learning_rate": 1.9987063186538606e-05,
+      "loss": 1.187,
+      "step": 1033
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1710938456855313,
+      "learning_rate": 1.998706302680131e-05,
+      "loss": 0.9302,
+      "step": 1034
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2041144822812797,
+      "learning_rate": 1.99870628668929e-05,
+      "loss": 0.9492,
+      "step": 1035
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2667927946906612,
+      "learning_rate": 1.9987062706813375e-05,
+      "loss": 1.1992,
+      "step": 1036
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3668022702948182,
+      "learning_rate": 1.9987062546562734e-05,
+      "loss": 1.1658,
+      "step": 1037
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.162993945528312,
+      "learning_rate": 1.9987062386140975e-05,
+      "loss": 0.8176,
+      "step": 1038
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3421640573925686,
+      "learning_rate": 1.9987062225548102e-05,
+      "loss": 1.137,
+      "step": 1039
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.8028209330731337,
+      "learning_rate": 1.9987062064784115e-05,
+      "loss": 1.0303,
+      "step": 1040
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2577400897489976,
+      "learning_rate": 1.998706190384901e-05,
+      "loss": 0.9785,
+      "step": 1041
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2057695430094857,
+      "learning_rate": 1.998706174274279e-05,
+      "loss": 0.8578,
+      "step": 1042
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.209358564161647,
+      "learning_rate": 1.9987061581465456e-05,
+      "loss": 0.9175,
+      "step": 1043
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.0412541844351448,
+      "learning_rate": 1.9987061420017005e-05,
+      "loss": 0.7444,
+      "step": 1044
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.255398774110792,
+      "learning_rate": 1.9987061258397437e-05,
+      "loss": 0.8672,
+      "step": 1045
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.0389995699304273,
+      "learning_rate": 1.9987061096606754e-05,
+      "loss": 0.8448,
+      "step": 1046
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3324595459170354,
+      "learning_rate": 1.9987060934644958e-05,
+      "loss": 0.8772,
+      "step": 1047
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6740032447181894,
+      "learning_rate": 1.9987060772512043e-05,
+      "loss": 1.3564,
+      "step": 1048
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.299509245601464,
+      "learning_rate": 1.9987060610208015e-05,
+      "loss": 1.1079,
+      "step": 1049
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.182862663465126,
+      "learning_rate": 1.998706044773287e-05,
+      "loss": 0.946,
+      "step": 1050
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 759313,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 1050,
+  "total_flos": 59260426661888.0,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}