diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,9492 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.9991668980838657,
+  "eval_steps": 500,
+  "global_step": 1350,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0022216051096917524,
+      "grad_norm": 0.038190219551324844,
+      "learning_rate": 2.2222222222222223e-05,
+      "loss": 9.2923,
+      "step": 1
+    },
+    {
+      "epoch": 0.004443210219383505,
+      "grad_norm": 0.031047649681568146,
+      "learning_rate": 4.4444444444444447e-05,
+      "loss": 9.4912,
+      "step": 2
+    },
+    {
+      "epoch": 0.006664815329075257,
+      "grad_norm": 0.03203844651579857,
+      "learning_rate": 6.666666666666667e-05,
+      "loss": 9.4908,
+      "step": 3
+    },
+    {
+      "epoch": 0.00888642043876701,
+      "grad_norm": 0.030807552859187126,
+      "learning_rate": 8.888888888888889e-05,
+      "loss": 9.5519,
+      "step": 4
+    },
+    {
+      "epoch": 0.011108025548458762,
+      "grad_norm": 0.03309250250458717,
+      "learning_rate": 0.0001111111111111111,
+      "loss": 9.2202,
+      "step": 5
+    },
+    {
+      "epoch": 0.013329630658150514,
+      "grad_norm": 0.031062453985214233,
+      "learning_rate": 0.00013333333333333334,
+      "loss": 9.2113,
+      "step": 6
+    },
+    {
+      "epoch": 0.015551235767842266,
+      "grad_norm": 0.03397192060947418,
+      "learning_rate": 0.00015555555555555556,
+      "loss": 9.5323,
+      "step": 7
+    },
+    {
+      "epoch": 0.01777284087753402,
+      "grad_norm": 0.03161627799272537,
+      "learning_rate": 0.00017777777777777779,
+      "loss": 9.3455,
+      "step": 8
+    },
+    {
+      "epoch": 0.01999444598722577,
+      "grad_norm": 0.03578317165374756,
+      "learning_rate": 0.0002,
+      "loss": 9.3865,
+      "step": 9
+    },
+    {
+      "epoch": 0.022216051096917523,
+      "grad_norm": 0.03146255016326904,
+      "learning_rate": 0.0002222222222222222,
+      "loss": 9.453,
+      "step": 10
+    },
+    {
+      "epoch": 0.024437656206609277,
+      "grad_norm": 0.0328507199883461,
+      "learning_rate": 0.0002444444444444445,
+      "loss": 9.3986,
+      "step": 11
+    },
+    {
+      "epoch": 0.026659261316301027,
+      "grad_norm": 0.03335467725992203,
+      "learning_rate": 0.0002666666666666667,
+      "loss": 9.4296,
+      "step": 12
+    },
+    {
+      "epoch": 0.02888086642599278,
+      "grad_norm": 0.03611304983496666,
+      "learning_rate": 0.0002888888888888889,
+      "loss": 9.3082,
+      "step": 13
+    },
+    {
+      "epoch": 0.03110247153568453,
+      "grad_norm": 0.03576128929853439,
+      "learning_rate": 0.0003111111111111111,
+      "loss": 9.2974,
+      "step": 14
+    },
+    {
+      "epoch": 0.03332407664537628,
+      "grad_norm": 0.03287427872419357,
+      "learning_rate": 0.0003333333333333333,
+      "loss": 9.3691,
+      "step": 15
+    },
+    {
+      "epoch": 0.03554568175506804,
+      "grad_norm": 0.033554255962371826,
+      "learning_rate": 0.00035555555555555557,
+      "loss": 9.4406,
+      "step": 16
+    },
+    {
+      "epoch": 0.03776728686475979,
+      "grad_norm": 0.03429733216762543,
+      "learning_rate": 0.00037777777777777777,
+      "loss": 9.138,
+      "step": 17
+    },
+    {
+      "epoch": 0.03998889197445154,
+      "grad_norm": 0.03480910137295723,
+      "learning_rate": 0.0004,
+      "loss": 9.1428,
+      "step": 18
+    },
+    {
+      "epoch": 0.042210497084143296,
+      "grad_norm": 0.03632334619760513,
+      "learning_rate": 0.00042222222222222227,
+      "loss": 9.3811,
+      "step": 19
+    },
+    {
+      "epoch": 0.044432102193835046,
+      "grad_norm": 0.03147323802113533,
+      "learning_rate": 0.0004444444444444444,
+      "loss": 9.4037,
+      "step": 20
+    },
+    {
+      "epoch": 0.0466537073035268,
+      "grad_norm": 0.03463975712656975,
+      "learning_rate": 0.00046666666666666666,
+      "loss": 9.278,
+      "step": 21
+    },
+    {
+      "epoch": 0.048875312413218554,
+      "grad_norm": 0.029492104426026344,
+      "learning_rate": 0.000488888888888889,
+      "loss": 9.2992,
+      "step": 22
+    },
+    {
+      "epoch": 0.051096917522910304,
+      "grad_norm": 0.03341001272201538,
+      "learning_rate": 0.0005111111111111112,
+      "loss": 9.1653,
+      "step": 23
+    },
+    {
+      "epoch": 0.053318522632602054,
+      "grad_norm": 0.03173983842134476,
+      "learning_rate": 0.0005333333333333334,
+      "loss": 9.1349,
+      "step": 24
+    },
+    {
+      "epoch": 0.055540127742293804,
+      "grad_norm": 0.03498266637325287,
+      "learning_rate": 0.0005555555555555556,
+      "loss": 9.0536,
+      "step": 25
+    },
+    {
+      "epoch": 0.05776173285198556,
+      "grad_norm": 0.03197816014289856,
+      "learning_rate": 0.0005777777777777778,
+      "loss": 9.3777,
+      "step": 26
+    },
+    {
+      "epoch": 0.05998333796167731,
+      "grad_norm": 0.03408670425415039,
+      "learning_rate": 0.0006000000000000001,
+      "loss": 9.1597,
+      "step": 27
+    },
+    {
+      "epoch": 0.06220494307136906,
+      "grad_norm": 0.03266999125480652,
+      "learning_rate": 0.0006222222222222223,
+      "loss": 9.6004,
+      "step": 28
+    },
+    {
+      "epoch": 0.06442654818106082,
+      "grad_norm": 0.0343904010951519,
+      "learning_rate": 0.0006444444444444444,
+      "loss": 9.692,
+      "step": 29
+    },
+    {
+      "epoch": 0.06664815329075256,
+      "grad_norm": 0.031083853915333748,
+      "learning_rate": 0.0006666666666666666,
+      "loss": 9.3645,
+      "step": 30
+    },
+    {
+      "epoch": 0.06886975840044432,
+      "grad_norm": 0.03298530727624893,
+      "learning_rate": 0.000688888888888889,
+      "loss": 9.2381,
+      "step": 31
+    },
+    {
+      "epoch": 0.07109136351013608,
+      "grad_norm": 0.03705119341611862,
+      "learning_rate": 0.0007111111111111111,
+      "loss": 9.1475,
+      "step": 32
+    },
+    {
+      "epoch": 0.07331296861982782,
+      "grad_norm": 0.03243013098835945,
+      "learning_rate": 0.0007333333333333333,
+      "loss": 9.3959,
+      "step": 33
+    },
+    {
+      "epoch": 0.07553457372951958,
+      "grad_norm": 0.033184703439474106,
+      "learning_rate": 0.0007555555555555555,
+      "loss": 9.4742,
+      "step": 34
+    },
+    {
+      "epoch": 0.07775617883921133,
+      "grad_norm": 0.03434516862034798,
+      "learning_rate": 0.0007777777777777777,
+      "loss": 9.4363,
+      "step": 35
+    },
+    {
+      "epoch": 0.07997778394890308,
+      "grad_norm": 0.03398028388619423,
+      "learning_rate": 0.0008,
+      "loss": 9.2335,
+      "step": 36
+    },
+    {
+      "epoch": 0.08219938905859484,
+      "grad_norm": 0.03263188153505325,
+      "learning_rate": 0.0008222222222222222,
+      "loss": 9.1648,
+      "step": 37
+    },
+    {
+      "epoch": 0.08442099416828659,
+      "grad_norm": 0.03378288820385933,
+      "learning_rate": 0.0008444444444444445,
+      "loss": 9.3229,
+      "step": 38
+    },
+    {
+      "epoch": 0.08664259927797834,
+      "grad_norm": 0.03087441436946392,
+      "learning_rate": 0.0008666666666666666,
+      "loss": 9.4068,
+      "step": 39
+    },
+    {
+      "epoch": 0.08886420438767009,
+      "grad_norm": 0.03767932578921318,
+      "learning_rate": 0.0008888888888888888,
+      "loss": 9.0222,
+      "step": 40
+    },
+    {
+      "epoch": 0.09108580949736185,
+      "grad_norm": 0.03177729249000549,
+      "learning_rate": 0.0009111111111111111,
+      "loss": 9.2663,
+      "step": 41
+    },
+    {
+      "epoch": 0.0933074146070536,
+      "grad_norm": 0.03294915333390236,
+      "learning_rate": 0.0009333333333333333,
+      "loss": 9.4539,
+      "step": 42
+    },
+    {
+      "epoch": 0.09552901971674535,
+      "grad_norm": 0.03155413269996643,
+      "learning_rate": 0.0009555555555555556,
+      "loss": 9.0169,
+      "step": 43
+    },
+    {
+      "epoch": 0.09775062482643711,
+      "grad_norm": 0.035774391144514084,
+      "learning_rate": 0.000977777777777778,
+      "loss": 9.2882,
+      "step": 44
+    },
+    {
+      "epoch": 0.09997222993612885,
+      "grad_norm": 0.03097563050687313,
+      "learning_rate": 0.001,
+      "loss": 9.4435,
+      "step": 45
+    },
+    {
+      "epoch": 0.10219383504582061,
+      "grad_norm": 0.026782240718603134,
+      "learning_rate": 0.0010222222222222223,
+      "loss": 9.4043,
+      "step": 46
+    },
+    {
+      "epoch": 0.10441544015551235,
+      "grad_norm": 0.032297562807798386,
+      "learning_rate": 0.0010444444444444444,
+      "loss": 9.4282,
+      "step": 47
+    },
+    {
+      "epoch": 0.10663704526520411,
+      "grad_norm": 0.03288343921303749,
+      "learning_rate": 0.0010666666666666667,
+      "loss": 9.4966,
+      "step": 48
+    },
+    {
+      "epoch": 0.10885865037489587,
+      "grad_norm": 0.02954668551683426,
+      "learning_rate": 0.001088888888888889,
+      "loss": 9.4091,
+      "step": 49
+    },
+    {
+      "epoch": 0.11108025548458761,
+      "grad_norm": 0.03393658623099327,
+      "learning_rate": 0.0011111111111111111,
+      "loss": 9.0923,
+      "step": 50
+    },
+    {
+      "epoch": 0.11330186059427937,
+      "grad_norm": 0.031363073736429214,
+      "learning_rate": 0.0011333333333333334,
+      "loss": 9.1016,
+      "step": 51
+    },
+    {
+      "epoch": 0.11552346570397112,
+      "grad_norm": 0.03455385938286781,
+      "learning_rate": 0.0011555555555555555,
+      "loss": 9.257,
+      "step": 52
+    },
+    {
+      "epoch": 0.11774507081366287,
+      "grad_norm": 0.0350220650434494,
+      "learning_rate": 0.0011777777777777778,
+      "loss": 9.1766,
+      "step": 53
+    },
+    {
+      "epoch": 0.11996667592335462,
+      "grad_norm": 0.032912105321884155,
+      "learning_rate": 0.0012000000000000001,
+      "loss": 9.184,
+      "step": 54
+    },
+    {
+      "epoch": 0.12218828103304638,
+      "grad_norm": 0.03227195516228676,
+      "learning_rate": 0.0012222222222222222,
+      "loss": 9.3788,
+      "step": 55
+    },
+    {
+      "epoch": 0.12440988614273812,
+      "grad_norm": 0.034647729247808456,
+      "learning_rate": 0.0012444444444444445,
+      "loss": 9.059,
+      "step": 56
+    },
+    {
+      "epoch": 0.12663149125242987,
+      "grad_norm": 0.02899823896586895,
+      "learning_rate": 0.0012666666666666666,
+      "loss": 9.0815,
+      "step": 57
+    },
+    {
+      "epoch": 0.12885309636212164,
+      "grad_norm": 0.03136138617992401,
+      "learning_rate": 0.001288888888888889,
+      "loss": 9.1401,
+      "step": 58
+    },
+    {
+      "epoch": 0.13107470147181338,
+      "grad_norm": 0.034191861748695374,
+      "learning_rate": 0.0013111111111111112,
+      "loss": 8.9662,
+      "step": 59
+    },
+    {
+      "epoch": 0.13329630658150513,
+      "grad_norm": 0.02805282361805439,
+      "learning_rate": 0.0013333333333333333,
+      "loss": 9.2464,
+      "step": 60
+    },
+    {
+      "epoch": 0.1355179116911969,
+      "grad_norm": 0.03340882435441017,
+      "learning_rate": 0.0013555555555555556,
+      "loss": 9.0857,
+      "step": 61
+    },
+    {
+      "epoch": 0.13773951680088864,
+      "grad_norm": 0.03609136864542961,
+      "learning_rate": 0.001377777777777778,
+      "loss": 9.304,
+      "step": 62
+    },
+    {
+      "epoch": 0.13996112191058038,
+      "grad_norm": 0.03209686651825905,
+      "learning_rate": 0.0014,
+      "loss": 9.1027,
+      "step": 63
+    },
+    {
+      "epoch": 0.14218272702027215,
+      "grad_norm": 0.031370747834444046,
+      "learning_rate": 0.0014222222222222223,
+      "loss": 9.1994,
+      "step": 64
+    },
+    {
+      "epoch": 0.1444043321299639,
+      "grad_norm": 0.03196916729211807,
+      "learning_rate": 0.0014444444444444444,
+      "loss": 9.0403,
+      "step": 65
+    },
+    {
+      "epoch": 0.14662593723965564,
+      "grad_norm": 0.029793815687298775,
+      "learning_rate": 0.0014666666666666667,
+      "loss": 9.4128,
+      "step": 66
+    },
+    {
+      "epoch": 0.1488475423493474,
+      "grad_norm": 0.030722036957740784,
+      "learning_rate": 0.001488888888888889,
+      "loss": 9.4157,
+      "step": 67
+    },
+    {
+      "epoch": 0.15106914745903915,
+      "grad_norm": 0.0261401180177927,
+      "learning_rate": 0.001511111111111111,
+      "loss": 9.1977,
+      "step": 68
+    },
+    {
+      "epoch": 0.1532907525687309,
+      "grad_norm": 0.02957984060049057,
+      "learning_rate": 0.0015333333333333332,
+      "loss": 9.1687,
+      "step": 69
+    },
+    {
+      "epoch": 0.15551235767842267,
+      "grad_norm": 0.03383724391460419,
+      "learning_rate": 0.0015555555555555555,
+      "loss": 8.9243,
+      "step": 70
+    },
+    {
+      "epoch": 0.1577339627881144,
+      "grad_norm": 0.02876967564225197,
+      "learning_rate": 0.0015777777777777778,
+      "loss": 9.0331,
+      "step": 71
+    },
+    {
+      "epoch": 0.15995556789780616,
+      "grad_norm": 0.030894216150045395,
+      "learning_rate": 0.0016,
+      "loss": 9.1879,
+      "step": 72
+    },
+    {
+      "epoch": 0.16217717300749793,
+      "grad_norm": 0.030421720817685127,
+      "learning_rate": 0.0016222222222222222,
+      "loss": 9.222,
+      "step": 73
+    },
+    {
+      "epoch": 0.16439877811718967,
+      "grad_norm": 0.03096602112054825,
+      "learning_rate": 0.0016444444444444445,
+      "loss": 8.9741,
+      "step": 74
+    },
+    {
+      "epoch": 0.1666203832268814,
+      "grad_norm": 0.03013846091926098,
+      "learning_rate": 0.0016666666666666668,
+      "loss": 9.2672,
+      "step": 75
+    },
+    {
+      "epoch": 0.16884198833657318,
+      "grad_norm": 0.03172660619020462,
+      "learning_rate": 0.001688888888888889,
+      "loss": 9.1982,
+      "step": 76
+    },
+    {
+      "epoch": 0.17106359344626493,
+      "grad_norm": 0.03142988681793213,
+      "learning_rate": 0.0017111111111111114,
+      "loss": 8.9948,
+      "step": 77
+    },
+    {
+      "epoch": 0.17328519855595667,
+      "grad_norm": 0.029608210548758507,
+      "learning_rate": 0.0017333333333333333,
+      "loss": 9.5021,
+      "step": 78
+    },
+    {
+      "epoch": 0.17550680366564844,
+      "grad_norm": 0.03226481378078461,
+      "learning_rate": 0.0017555555555555556,
+      "loss": 8.8922,
+      "step": 79
+    },
+    {
+      "epoch": 0.17772840877534019,
+      "grad_norm": 0.029186666011810303,
+      "learning_rate": 0.0017777777777777776,
+      "loss": 9.2245,
+      "step": 80
+    },
+    {
+      "epoch": 0.17995001388503193,
+      "grad_norm": 0.028802979737520218,
+      "learning_rate": 0.0018,
+      "loss": 9.1839,
+      "step": 81
+    },
+    {
+      "epoch": 0.1821716189947237,
+      "grad_norm": 0.028176436200737953,
+      "learning_rate": 0.0018222222222222223,
+      "loss": 9.1026,
+      "step": 82
+    },
+    {
+      "epoch": 0.18439322410441544,
+      "grad_norm": 0.031202396377921104,
+      "learning_rate": 0.0018444444444444446,
+      "loss": 9.0409,
+      "step": 83
+    },
+    {
+      "epoch": 0.1866148292141072,
+      "grad_norm": 0.032177042216062546,
+      "learning_rate": 0.0018666666666666666,
+      "loss": 8.844,
+      "step": 84
+    },
+    {
+      "epoch": 0.18883643432379896,
+      "grad_norm": 0.03255876526236534,
+      "learning_rate": 0.001888888888888889,
+      "loss": 8.8978,
+      "step": 85
+    },
+    {
+      "epoch": 0.1910580394334907,
+      "grad_norm": 0.03154279664158821,
+      "learning_rate": 0.0019111111111111113,
+      "loss": 9.0468,
+      "step": 86
+    },
+    {
+      "epoch": 0.19327964454318244,
+      "grad_norm": 0.02897936850786209,
+      "learning_rate": 0.0019333333333333336,
+      "loss": 9.0149,
+      "step": 87
+    },
+    {
+      "epoch": 0.19550124965287421,
+      "grad_norm": 0.030778463929891586,
+      "learning_rate": 0.001955555555555556,
+      "loss": 9.07,
+      "step": 88
+    },
+    {
+      "epoch": 0.19772285476256596,
+      "grad_norm": 0.031090715900063515,
+      "learning_rate": 0.0019777777777777775,
+      "loss": 9.0561,
+      "step": 89
+    },
+    {
+      "epoch": 0.1999444598722577,
+      "grad_norm": 0.03048640303313732,
+      "learning_rate": 0.002,
+      "loss": 8.8544,
+      "step": 90
+    },
+    {
+      "epoch": 0.20216606498194944,
+      "grad_norm": 0.028690746054053307,
+      "learning_rate": 0.002022222222222222,
+      "loss": 9.2081,
+      "step": 91
+    },
+    {
+      "epoch": 0.20438767009164122,
+      "grad_norm": 0.032125819474458694,
+      "learning_rate": 0.0020444444444444447,
+      "loss": 8.9882,
+      "step": 92
+    },
+    {
+      "epoch": 0.20660927520133296,
+      "grad_norm": 0.03342963382601738,
+      "learning_rate": 0.0020666666666666667,
+      "loss": 9.0407,
+      "step": 93
+    },
+    {
+      "epoch": 0.2088308803110247,
+      "grad_norm": 0.03291715681552887,
+      "learning_rate": 0.002088888888888889,
+      "loss": 8.9794,
+      "step": 94
+    },
+    {
+      "epoch": 0.21105248542071647,
+      "grad_norm": 0.02747849002480507,
+      "learning_rate": 0.0021111111111111113,
+      "loss": 9.042,
+      "step": 95
+    },
+    {
+      "epoch": 0.21327409053040822,
+      "grad_norm": 0.03284389525651932,
+      "learning_rate": 0.0021333333333333334,
+      "loss": 8.9185,
+      "step": 96
+    },
+    {
+      "epoch": 0.21549569564009996,
+      "grad_norm": 0.030258171260356903,
+      "learning_rate": 0.0021555555555555555,
+      "loss": 9.0191,
+      "step": 97
+    },
+    {
+      "epoch": 0.21771730074979173,
+      "grad_norm": 0.03098541684448719,
+      "learning_rate": 0.002177777777777778,
+      "loss": 9.0654,
+      "step": 98
+    },
+    {
+      "epoch": 0.21993890585948347,
+      "grad_norm": 0.03234381601214409,
+      "learning_rate": 0.0021999999999999997,
+      "loss": 8.8876,
+      "step": 99
+    },
+    {
+      "epoch": 0.22216051096917522,
+      "grad_norm": 0.029549606144428253,
+      "learning_rate": 0.0022222222222222222,
+      "loss": 9.0951,
+      "step": 100
+    },
+    {
+      "epoch": 0.224382116078867,
+      "grad_norm": 0.028349105268716812,
+      "learning_rate": 0.0022444444444444443,
+      "loss": 8.9689,
+      "step": 101
+    },
+    {
+      "epoch": 0.22660372118855873,
+      "grad_norm": 0.03373165801167488,
+      "learning_rate": 0.002266666666666667,
+      "loss": 9.0545,
+      "step": 102
+    },
+    {
+      "epoch": 0.22882532629825048,
+      "grad_norm": 0.03023650497198105,
+      "learning_rate": 0.002288888888888889,
+      "loss": 8.7262,
+      "step": 103
+    },
+    {
+      "epoch": 0.23104693140794225,
+      "grad_norm": 0.03220418840646744,
+      "learning_rate": 0.002311111111111111,
+      "loss": 8.8946,
+      "step": 104
+    },
+    {
+      "epoch": 0.233268536517634,
+      "grad_norm": 0.03183241933584213,
+      "learning_rate": 0.0023333333333333335,
+      "loss": 8.7574,
+      "step": 105
+    },
+    {
+      "epoch": 0.23549014162732573,
+      "grad_norm": 0.02865174598991871,
+      "learning_rate": 0.0023555555555555556,
+      "loss": 9.2181,
+      "step": 106
+    },
+    {
+      "epoch": 0.2377117467370175,
+      "grad_norm": 0.02678099274635315,
+      "learning_rate": 0.002377777777777778,
+      "loss": 9.0188,
+      "step": 107
+    },
+    {
+      "epoch": 0.23993335184670925,
+      "grad_norm": 0.028721222653985023,
+      "learning_rate": 0.0024000000000000002,
+      "loss": 8.9342,
+      "step": 108
+    },
+    {
+      "epoch": 0.242154956956401,
+      "grad_norm": 0.02757999859750271,
+      "learning_rate": 0.0024222222222222223,
+      "loss": 8.9704,
+      "step": 109
+    },
+    {
+      "epoch": 0.24437656206609276,
+      "grad_norm": 0.025576164945960045,
+      "learning_rate": 0.0024444444444444444,
+      "loss": 9.0426,
+      "step": 110
+    },
+    {
+      "epoch": 0.2465981671757845,
+      "grad_norm": 0.029820214956998825,
+      "learning_rate": 0.0024666666666666665,
+      "loss": 8.938,
+      "step": 111
+    },
+    {
+      "epoch": 0.24881977228547625,
+      "grad_norm": 0.030512187629938126,
+      "learning_rate": 0.002488888888888889,
+      "loss": 8.7319,
+      "step": 112
+    },
+    {
+      "epoch": 0.251041377395168,
+      "grad_norm": 0.028842419385910034,
+      "learning_rate": 0.002511111111111111,
+      "loss": 8.9827,
+      "step": 113
+    },
+    {
+      "epoch": 0.25326298250485973,
+      "grad_norm": 0.029324037954211235,
+      "learning_rate": 0.002533333333333333,
+      "loss": 8.6722,
+      "step": 114
+    },
+    {
+      "epoch": 0.2554845876145515,
+      "grad_norm": 0.028672564774751663,
+      "learning_rate": 0.0025555555555555557,
+      "loss": 8.8596,
+      "step": 115
+    },
+    {
+      "epoch": 0.2577061927242433,
+      "grad_norm": 0.027796434238553047,
+      "learning_rate": 0.002577777777777778,
+      "loss": 8.9966,
+      "step": 116
+    },
+    {
+      "epoch": 0.259927797833935,
+      "grad_norm": 0.028437964618206024,
+      "learning_rate": 0.0026000000000000003,
+      "loss": 8.81,
+      "step": 117
+    },
+    {
+      "epoch": 0.26214940294362676,
+      "grad_norm": 0.030968384817242622,
+      "learning_rate": 0.0026222222222222224,
+      "loss": 8.9205,
+      "step": 118
+    },
+    {
+      "epoch": 0.26437100805331853,
+      "grad_norm": 0.029907088726758957,
+      "learning_rate": 0.0026444444444444445,
+      "loss": 8.7998,
+      "step": 119
+    },
+    {
+      "epoch": 0.26659261316301025,
+      "grad_norm": 0.03147348761558533,
+      "learning_rate": 0.0026666666666666666,
+      "loss": 8.4368,
+      "step": 120
+    },
+    {
+      "epoch": 0.268814218272702,
+      "grad_norm": 0.029168879613280296,
+      "learning_rate": 0.0026888888888888887,
+      "loss": 8.8344,
+      "step": 121
+    },
+    {
+      "epoch": 0.2710358233823938,
+      "grad_norm": 0.028275858610868454,
+      "learning_rate": 0.002711111111111111,
+      "loss": 8.8366,
+      "step": 122
+    },
+    {
+      "epoch": 0.2732574284920855,
+      "grad_norm": 0.028297165408730507,
+      "learning_rate": 0.0027333333333333333,
+      "loss": 8.7836,
+      "step": 123
+    },
+    {
+      "epoch": 0.2754790336017773,
+      "grad_norm": 0.02639463171362877,
+      "learning_rate": 0.002755555555555556,
+      "loss": 8.9441,
+      "step": 124
+    },
+    {
+      "epoch": 0.27770063871146905,
+      "grad_norm": 0.02884727157652378,
+      "learning_rate": 0.002777777777777778,
+      "loss": 8.6444,
+      "step": 125
+    },
+    {
+      "epoch": 0.27992224382116077,
+      "grad_norm": 0.028951751068234444,
+      "learning_rate": 0.0028,
+      "loss": 8.6928,
+      "step": 126
+    },
+    {
+      "epoch": 0.28214384893085254,
+      "grad_norm": 0.03159439191222191,
+      "learning_rate": 0.0028222222222222225,
+      "loss": 8.5061,
+      "step": 127
+    },
+    {
+      "epoch": 0.2843654540405443,
+      "grad_norm": 0.03017883561551571,
+      "learning_rate": 0.0028444444444444446,
+      "loss": 8.7759,
+      "step": 128
+    },
+    {
+      "epoch": 0.286587059150236,
+      "grad_norm": 0.02969433180987835,
+      "learning_rate": 0.0028666666666666667,
+      "loss": 8.8132,
+      "step": 129
+    },
+    {
+      "epoch": 0.2888086642599278,
+      "grad_norm": 0.03073931485414505,
+      "learning_rate": 0.0028888888888888888,
+      "loss": 8.6928,
+      "step": 130
+    },
+    {
+      "epoch": 0.29103026936961957,
+      "grad_norm": 0.026906149461865425,
+      "learning_rate": 0.002911111111111111,
+      "loss": 8.6811,
+      "step": 131
+    },
+    {
+      "epoch": 0.2932518744793113,
+      "grad_norm": 0.028735142201185226,
+      "learning_rate": 0.0029333333333333334,
+      "loss": 8.8406,
+      "step": 132
+    },
+    {
+      "epoch": 0.29547347958900305,
+      "grad_norm": 0.029568219557404518,
+      "learning_rate": 0.0029555555555555555,
+      "loss": 8.6118,
+      "step": 133
+    },
+    {
+      "epoch": 0.2976950846986948,
+      "grad_norm": 0.02788039855659008,
+      "learning_rate": 0.002977777777777778,
+      "loss": 8.736,
+      "step": 134
+    },
+    {
+      "epoch": 0.29991668980838654,
+      "grad_norm": 0.033334843814373016,
+      "learning_rate": 0.003,
+      "loss": 8.5657,
+      "step": 135
+    },
+    {
+      "epoch": 0.3021382949180783,
+      "grad_norm": 0.030152609571814537,
+      "learning_rate": 0.0029975308641975312,
+      "loss": 8.6157,
+      "step": 136
+    },
+    {
+      "epoch": 0.3043599000277701,
+      "grad_norm": 0.02878383919596672,
+      "learning_rate": 0.0029950617283950615,
+      "loss": 8.6853,
+      "step": 137
+    },
+    {
+      "epoch": 0.3065815051374618,
+      "grad_norm": 0.028065672144293785,
+      "learning_rate": 0.0029925925925925927,
+      "loss": 8.6246,
+      "step": 138
+    },
+    {
+      "epoch": 0.30880311024715357,
+      "grad_norm": 0.027555333450436592,
+      "learning_rate": 0.0029901234567901234,
+      "loss": 8.808,
+      "step": 139
+    },
+    {
+      "epoch": 0.31102471535684534,
+      "grad_norm": 0.027237268164753914,
+      "learning_rate": 0.0029876543209876546,
+      "loss": 8.6793,
+      "step": 140
+    },
+    {
+      "epoch": 0.31324632046653705,
+      "grad_norm": 0.028404848650097847,
+      "learning_rate": 0.002985185185185185,
+      "loss": 8.5969,
+      "step": 141
+    },
+    {
+      "epoch": 0.3154679255762288,
+      "grad_norm": 0.028260594233870506,
+      "learning_rate": 0.002982716049382716,
+      "loss": 8.604,
+      "step": 142
+    },
+    {
+      "epoch": 0.3176895306859206,
+      "grad_norm": 0.02859456278383732,
+      "learning_rate": 0.0029802469135802472,
+      "loss": 8.3553,
+      "step": 143
+    },
+    {
+      "epoch": 0.3199111357956123,
+      "grad_norm": 0.02776123769581318,
+      "learning_rate": 0.002977777777777778,
+      "loss": 8.7692,
+      "step": 144
+    },
+    {
+      "epoch": 0.3221327409053041,
+      "grad_norm": 0.03206472471356392,
+      "learning_rate": 0.0029753086419753087,
+      "loss": 8.5463,
+      "step": 145
+    },
+    {
+      "epoch": 0.32435434601499585,
+      "grad_norm": 0.027482545003294945,
+      "learning_rate": 0.0029728395061728394,
+      "loss": 8.3237,
+      "step": 146
+    },
+    {
+      "epoch": 0.32657595112468757,
+      "grad_norm": 0.028557900339365005,
+      "learning_rate": 0.0029703703703703706,
+      "loss": 8.4529,
+      "step": 147
+    },
+    {
+      "epoch": 0.32879755623437934,
+      "grad_norm": 0.027708956971764565,
+      "learning_rate": 0.0029679012345679013,
+      "loss": 8.6945,
+      "step": 148
+    },
+    {
+      "epoch": 0.3310191613440711,
+      "grad_norm": 0.02987179346382618,
+      "learning_rate": 0.002965432098765432,
+      "loss": 8.4202,
+      "step": 149
+    },
+    {
+      "epoch": 0.3332407664537628,
+      "grad_norm": 0.027143696323037148,
+      "learning_rate": 0.002962962962962963,
+      "loss": 8.7476,
+      "step": 150
+    },
+    {
+      "epoch": 0.3354623715634546,
+      "grad_norm": 0.030200520530343056,
+      "learning_rate": 0.002960493827160494,
+      "loss": 8.4821,
+      "step": 151
+    },
+    {
+      "epoch": 0.33768397667314637,
+      "grad_norm": 0.03403831645846367,
+      "learning_rate": 0.0029580246913580247,
+      "loss": 8.5051,
+      "step": 152
+    },
+    {
+      "epoch": 0.3399055817828381,
+      "grad_norm": 0.02961106039583683,
+      "learning_rate": 0.0029555555555555555,
+      "loss": 8.4996,
+      "step": 153
+    },
+    {
+      "epoch": 0.34212718689252986,
+      "grad_norm": 0.02923700399696827,
+      "learning_rate": 0.0029530864197530866,
+      "loss": 8.4532,
+      "step": 154
+    },
+    {
+      "epoch": 0.3443487920022216,
+      "grad_norm": 0.027527211233973503,
+      "learning_rate": 0.0029506172839506174,
+      "loss": 8.4647,
+      "step": 155
+    },
+    {
+      "epoch": 0.34657039711191334,
+      "grad_norm": 0.028258679434657097,
+      "learning_rate": 0.002948148148148148,
+      "loss": 8.3038,
+      "step": 156
+    },
+    {
+      "epoch": 0.3487920022216051,
+      "grad_norm": 0.03036580979824066,
+      "learning_rate": 0.002945679012345679,
+      "loss": 8.4641,
+      "step": 157
+    },
+    {
+      "epoch": 0.3510136073312969,
+      "grad_norm": 0.030313914641737938,
+      "learning_rate": 0.00294320987654321,
+      "loss": 8.5855,
+      "step": 158
+    },
+    {
+      "epoch": 0.3532352124409886,
+      "grad_norm": 0.02866736240684986,
+      "learning_rate": 0.0029407407407407407,
+      "loss": 8.4413,
+      "step": 159
+    },
+    {
+      "epoch": 0.35545681755068037,
+      "grad_norm": 0.032261501997709274,
+      "learning_rate": 0.0029382716049382715,
+      "loss": 8.3972,
+      "step": 160
+    },
+    {
+      "epoch": 0.35767842266037214,
+      "grad_norm": 0.028271513059735298,
+      "learning_rate": 0.0029358024691358026,
+      "loss": 8.5005,
+      "step": 161
+    },
+    {
+      "epoch": 0.35990002777006386,
+      "grad_norm": 0.03222686052322388,
+      "learning_rate": 0.0029333333333333334,
+      "loss": 8.3212,
+      "step": 162
+    },
+    {
+      "epoch": 0.36212163287975563,
+      "grad_norm": 0.02958948165178299,
+      "learning_rate": 0.002930864197530864,
+      "loss": 8.1741,
+      "step": 163
+    },
+    {
+      "epoch": 0.3643432379894474,
+      "grad_norm": 0.030546046793460846,
+      "learning_rate": 0.002928395061728395,
+      "loss": 8.2262,
+      "step": 164
+    },
+    {
+      "epoch": 0.3665648430991391,
+      "grad_norm": 0.029829349368810654,
+      "learning_rate": 0.002925925925925926,
+      "loss": 8.3214,
+      "step": 165
+    },
+    {
+      "epoch": 0.3687864482088309,
+      "grad_norm": 0.02913397178053856,
+      "learning_rate": 0.002923456790123457,
+      "loss": 8.4382,
+      "step": 166
+    },
+    {
+      "epoch": 0.37100805331852266,
+      "grad_norm": 0.02896488830447197,
+      "learning_rate": 0.0029209876543209875,
+      "loss": 8.4278,
+      "step": 167
+    },
+    {
+      "epoch": 0.3732296584282144,
+      "grad_norm": 0.03006349317729473,
+      "learning_rate": 0.0029185185185185186,
+      "loss": 8.1418,
+      "step": 168
+    },
+    {
+      "epoch": 0.37545126353790614,
+      "grad_norm": 0.02750089392066002,
+      "learning_rate": 0.0029160493827160494,
+      "loss": 8.3711,
+      "step": 169
+    },
+    {
+      "epoch": 0.3776728686475979,
+      "grad_norm": 0.029414812102913857,
+      "learning_rate": 0.0029135802469135805,
+      "loss": 8.5652,
+      "step": 170
+    },
+    {
+      "epoch": 0.37989447375728963,
+      "grad_norm": 0.02833523415029049,
+      "learning_rate": 0.002911111111111111,
+      "loss": 8.2343,
+      "step": 171
+    },
+    {
+      "epoch": 0.3821160788669814,
+      "grad_norm": 0.030764391645789146,
+      "learning_rate": 0.002908641975308642,
+      "loss": 8.4708,
+      "step": 172
+    },
+    {
+      "epoch": 0.3843376839766732,
+      "grad_norm": 0.031116491183638573,
+      "learning_rate": 0.002906172839506173,
+      "loss": 8.3074,
+      "step": 173
+    },
+    {
+      "epoch": 0.3865592890863649,
+      "grad_norm": 0.029229944571852684,
+      "learning_rate": 0.002903703703703704,
+      "loss": 8.288,
+      "step": 174
+    },
+    {
+      "epoch": 0.38878089419605666,
+      "grad_norm": 0.02907656691968441,
+      "learning_rate": 0.0029012345679012346,
+      "loss": 8.4092,
+      "step": 175
+    },
+    {
+      "epoch": 0.39100249930574843,
+      "grad_norm": 0.02829660475254059,
+      "learning_rate": 0.0028987654320987654,
+      "loss": 8.3569,
+      "step": 176
+    },
+    {
+      "epoch": 0.39322410441544015,
+      "grad_norm": 0.029642395675182343,
+      "learning_rate": 0.0028962962962962966,
+      "loss": 8.3601,
+      "step": 177
+    },
+    {
+      "epoch": 0.3954457095251319,
+      "grad_norm": 0.029279954731464386,
+      "learning_rate": 0.0028938271604938273,
+      "loss": 8.1301,
+      "step": 178
+    },
+    {
+      "epoch": 0.3976673146348237,
+      "grad_norm": 0.0306511539965868,
+      "learning_rate": 0.002891358024691358,
+      "loss": 8.1096,
+      "step": 179
+    },
+    {
+      "epoch": 0.3998889197445154,
+      "grad_norm": 0.02933737076818943,
+      "learning_rate": 0.0028888888888888888,
+      "loss": 8.2866,
+      "step": 180
+    },
+    {
+      "epoch": 0.4021105248542072,
+      "grad_norm": 0.030354322865605354,
+      "learning_rate": 0.00288641975308642,
+      "loss": 8.4484,
+      "step": 181
+    },
+    {
+      "epoch": 0.4043321299638989,
+      "grad_norm": 0.027711838483810425,
+      "learning_rate": 0.0028839506172839507,
+      "loss": 8.2382,
+      "step": 182
+    },
+    {
+      "epoch": 0.40655373507359066,
+      "grad_norm": 0.02723979391157627,
+      "learning_rate": 0.0028814814814814814,
+      "loss": 8.2111,
+      "step": 183
+    },
+    {
+      "epoch": 0.40877534018328243,
+      "grad_norm": 0.028976181522011757,
+      "learning_rate": 0.0028790123456790126,
+      "loss": 8.0978,
+      "step": 184
+    },
+    {
+      "epoch": 0.41099694529297415,
+      "grad_norm": 0.028827300295233727,
+      "learning_rate": 0.0028765432098765433,
+      "loss": 8.1841,
+      "step": 185
+    },
+    {
+      "epoch": 0.4132185504026659,
+      "grad_norm": 0.033451732248067856,
+      "learning_rate": 0.002874074074074074,
+      "loss": 7.8206,
+      "step": 186
+    },
+    {
+      "epoch": 0.4154401555123577,
+      "grad_norm": 0.02918138913810253,
+      "learning_rate": 0.0028716049382716048,
+      "loss": 8.1682,
+      "step": 187
+    },
+    {
+      "epoch": 0.4176617606220494,
+      "grad_norm": 0.0293852798640728,
+      "learning_rate": 0.002869135802469136,
+      "loss": 8.1494,
+      "step": 188
+    },
+    {
+      "epoch": 0.4198833657317412,
+      "grad_norm": 0.027722623199224472,
+      "learning_rate": 0.0028666666666666667,
+      "loss": 8.0988,
+      "step": 189
+    },
+    {
+      "epoch": 0.42210497084143295,
+      "grad_norm": 0.027863891795277596,
+      "learning_rate": 0.0028641975308641974,
+      "loss": 8.1322,
+      "step": 190
+    },
+    {
+      "epoch": 0.42432657595112466,
+      "grad_norm": 0.030969025567173958,
+      "learning_rate": 0.0028617283950617286,
+      "loss": 8.0404,
+      "step": 191
+    },
+    {
+      "epoch": 0.42654818106081643,
+      "grad_norm": 0.028908630833029747,
+      "learning_rate": 0.0028592592592592593,
+      "loss": 7.8183,
+      "step": 192
+    },
+    {
+      "epoch": 0.4287697861705082,
+      "grad_norm": 0.030768541619181633,
+      "learning_rate": 0.00285679012345679,
+      "loss": 7.775,
+      "step": 193
+    },
+    {
+      "epoch": 0.4309913912801999,
+      "grad_norm": 0.029368046671152115,
+      "learning_rate": 0.0028543209876543208,
+      "loss": 7.8377,
+      "step": 194
+    },
+    {
+      "epoch": 0.4332129963898917,
+      "grad_norm": 0.03016546182334423,
+      "learning_rate": 0.002851851851851852,
+      "loss": 7.8806,
+      "step": 195
+    },
+    {
+      "epoch": 0.43543460149958346,
+      "grad_norm": 0.025432731956243515,
+      "learning_rate": 0.002849382716049383,
+      "loss": 8.0247,
+      "step": 196
+    },
+    {
+      "epoch": 0.4376562066092752,
+      "grad_norm": 0.02614755742251873,
+      "learning_rate": 0.0028469135802469134,
+      "loss": 7.7943,
+      "step": 197
+    },
+    {
+      "epoch": 0.43987781171896695,
+      "grad_norm": 0.02812175638973713,
+      "learning_rate": 0.0028444444444444446,
+      "loss": 8.1555,
+      "step": 198
+    },
+    {
+      "epoch": 0.4420994168286587,
+      "grad_norm": 0.02803466096520424,
+      "learning_rate": 0.0028419753086419753,
+      "loss": 8.277,
+      "step": 199
+    },
+    {
+      "epoch": 0.44432102193835044,
+      "grad_norm": 0.02616594173014164,
+      "learning_rate": 0.0028395061728395065,
+      "loss": 7.9641,
+      "step": 200
+    },
+    {
+      "epoch": 0.4465426270480422,
+      "grad_norm": 0.03199322149157524,
+      "learning_rate": 0.002837037037037037,
+      "loss": 7.8528,
+      "step": 201
+    },
+    {
+      "epoch": 0.448764232157734,
+      "grad_norm": 0.026955347508192062,
+      "learning_rate": 0.002834567901234568,
+      "loss": 7.8566,
+      "step": 202
+    },
+    {
+      "epoch": 0.4509858372674257,
+      "grad_norm": 0.028074899688363075,
+      "learning_rate": 0.002832098765432099,
+      "loss": 7.7715,
+      "step": 203
+    },
+    {
+      "epoch": 0.45320744237711746,
+      "grad_norm": 0.030562324449419975,
+      "learning_rate": 0.00282962962962963,
+      "loss": 7.7334,
+      "step": 204
+    },
+    {
+      "epoch": 0.45542904748680924,
+      "grad_norm": 0.03391079977154732,
+      "learning_rate": 0.0028271604938271606,
+      "loss": 7.4609,
+      "step": 205
+    },
+    {
+      "epoch": 0.45765065259650095,
+      "grad_norm": 0.025406498461961746,
+      "learning_rate": 0.0028246913580246913,
+      "loss": 7.9336,
+      "step": 206
+    },
+    {
+      "epoch": 0.4598722577061927,
+      "grad_norm": 0.025080909952521324,
+      "learning_rate": 0.0028222222222222225,
+      "loss": 7.9637,
+      "step": 207
+    },
+    {
+      "epoch": 0.4620938628158845,
+      "grad_norm": 0.025412667542696,
+      "learning_rate": 0.0028197530864197532,
+      "loss": 7.9139,
+      "step": 208
+    },
+    {
+      "epoch": 0.4643154679255762,
+      "grad_norm": 0.02715262398123741,
+      "learning_rate": 0.002817283950617284,
+      "loss": 7.8741,
+      "step": 209
+    },
+    {
+      "epoch": 0.466537073035268,
+      "grad_norm": 0.0327037051320076,
+      "learning_rate": 0.0028148148148148147,
+      "loss": 7.9813,
+      "step": 210
+    },
+    {
+      "epoch": 0.46875867814495975,
+      "grad_norm": 0.02404477261006832,
+      "learning_rate": 0.002812345679012346,
+      "loss": 7.6723,
+      "step": 211
+    },
+    {
+      "epoch": 0.47098028325465147,
+      "grad_norm": 0.02602028287947178,
+      "learning_rate": 0.0028098765432098766,
+      "loss": 7.8431,
+      "step": 212
+    },
+    {
+      "epoch": 0.47320188836434324,
+      "grad_norm": 0.02408405765891075,
+      "learning_rate": 0.0028074074074074073,
+      "loss": 7.8601,
+      "step": 213
+    },
+    {
+      "epoch": 0.475423493474035,
+      "grad_norm": 0.029246551916003227,
+      "learning_rate": 0.0028049382716049385,
+      "loss": 7.5496,
+      "step": 214
+    },
+    {
+      "epoch": 0.4776450985837267,
+      "grad_norm": 0.028270237147808075,
+      "learning_rate": 0.0028024691358024692,
+      "loss": 7.9834,
+      "step": 215
+    },
+    {
+      "epoch": 0.4798667036934185,
+      "grad_norm": 0.025235634297132492,
+      "learning_rate": 0.0028,
+      "loss": 7.7878,
+      "step": 216
+    },
+    {
+      "epoch": 0.48208830880311027,
+      "grad_norm": 0.030682511627674103,
+      "learning_rate": 0.0027975308641975307,
+      "loss": 7.8148,
+      "step": 217
+    },
+    {
+      "epoch": 0.484309913912802,
+      "grad_norm": 0.023467013612389565,
+      "learning_rate": 0.002795061728395062,
+      "loss": 7.7851,
+      "step": 218
+    },
+    {
+      "epoch": 0.48653151902249375,
+      "grad_norm": 0.02427959442138672,
+      "learning_rate": 0.0027925925925925926,
+      "loss": 7.5556,
+      "step": 219
+    },
+    {
+      "epoch": 0.4887531241321855,
+      "grad_norm": 0.026607820764183998,
+      "learning_rate": 0.0027901234567901233,
+      "loss": 7.7181,
+      "step": 220
+    },
+    {
+      "epoch": 0.49097472924187724,
+      "grad_norm": 0.024195892736315727,
+      "learning_rate": 0.0027876543209876545,
+      "loss": 7.7836,
+      "step": 221
+    },
+    {
+      "epoch": 0.493196334351569,
+      "grad_norm": 0.023083994165062904,
+      "learning_rate": 0.0027851851851851852,
+      "loss": 7.9689,
+      "step": 222
+    },
+    {
+      "epoch": 0.4954179394612608,
+      "grad_norm": 0.03038424253463745,
+      "learning_rate": 0.002782716049382716,
+      "loss": 7.5597,
+      "step": 223
+    },
+    {
+      "epoch": 0.4976395445709525,
+      "grad_norm": 0.02380535379052162,
+      "learning_rate": 0.0027802469135802467,
+      "loss": 7.772,
+      "step": 224
+    },
+    {
+      "epoch": 0.49986114968064427,
+      "grad_norm": 0.028999164700508118,
+      "learning_rate": 0.002777777777777778,
+      "loss": 7.5047,
+      "step": 225
+    },
+    {
+      "epoch": 0.502082754790336,
+      "grad_norm": 0.02723139151930809,
+      "learning_rate": 0.002775308641975309,
+      "loss": 7.4959,
+      "step": 226
+    },
+    {
+      "epoch": 0.5043043599000278,
+      "grad_norm": 0.02604815550148487,
+      "learning_rate": 0.0027728395061728394,
+      "loss": 7.7955,
+      "step": 227
+    },
+    {
+      "epoch": 0.5065259650097195,
+      "grad_norm": 0.024884553626179695,
+      "learning_rate": 0.0027703703703703705,
+      "loss": 7.829,
+      "step": 228
+    },
+    {
+      "epoch": 0.5087475701194113,
+      "grad_norm": 0.02377072162926197,
+      "learning_rate": 0.0027679012345679013,
+      "loss": 7.9101,
+      "step": 229
+    },
+    {
+      "epoch": 0.510969175229103,
+      "grad_norm": 0.023848138749599457,
+      "learning_rate": 0.0027654320987654324,
+      "loss": 7.8452,
+      "step": 230
+    },
+    {
+      "epoch": 0.5131907803387947,
+      "grad_norm": 0.026330411434173584,
+      "learning_rate": 0.0027629629629629627,
+      "loss": 7.5427,
+      "step": 231
+    },
+    {
+      "epoch": 0.5154123854484866,
+      "grad_norm": 0.028968099504709244,
+      "learning_rate": 0.002760493827160494,
+      "loss": 7.5541,
+      "step": 232
+    },
+    {
+      "epoch": 0.5176339905581783,
+      "grad_norm": 0.02400917559862137,
+      "learning_rate": 0.0027580246913580246,
+      "loss": 7.7601,
+      "step": 233
+    },
+    {
+      "epoch": 0.51985559566787,
+      "grad_norm": 0.026197567582130432,
+      "learning_rate": 0.002755555555555556,
+      "loss": 7.6337,
+      "step": 234
+    },
+    {
+      "epoch": 0.5220772007775618,
+      "grad_norm": 0.02702515758574009,
+      "learning_rate": 0.0027530864197530865,
+      "loss": 7.7061,
+      "step": 235
+    },
+    {
+      "epoch": 0.5242988058872535,
+      "grad_norm": 0.02299593575298786,
+      "learning_rate": 0.0027506172839506173,
+      "loss": 7.7556,
+      "step": 236
+    },
+    {
+      "epoch": 0.5265204109969452,
+      "grad_norm": 0.023501381278038025,
+      "learning_rate": 0.0027481481481481484,
+      "loss": 7.6893,
+      "step": 237
+    },
+    {
+      "epoch": 0.5287420161066371,
+      "grad_norm": 0.02764345146715641,
+      "learning_rate": 0.002745679012345679,
+      "loss": 7.517,
+      "step": 238
+    },
+    {
+      "epoch": 0.5309636212163288,
+      "grad_norm": 0.028914740309119225,
+      "learning_rate": 0.00274320987654321,
+      "loss": 7.4575,
+      "step": 239
+    },
+    {
+      "epoch": 0.5331852263260205,
+      "grad_norm": 0.02865949086844921,
+      "learning_rate": 0.0027407407407407406,
+      "loss": 7.7941,
+      "step": 240
+    },
+    {
+      "epoch": 0.5354068314357123,
+      "grad_norm": 0.026122935116291046,
+      "learning_rate": 0.002738271604938272,
+      "loss": 7.4522,
+      "step": 241
+    },
+    {
+      "epoch": 0.537628436545404,
+      "grad_norm": 0.021501965820789337,
+      "learning_rate": 0.0027358024691358025,
+      "loss": 7.4257,
+      "step": 242
+    },
+    {
+      "epoch": 0.5398500416550958,
+      "grad_norm": 0.03300165757536888,
+      "learning_rate": 0.0027333333333333333,
+      "loss": 7.6302,
+      "step": 243
+    },
+    {
+      "epoch": 0.5420716467647876,
+      "grad_norm": 0.027942990884184837,
+      "learning_rate": 0.0027308641975308644,
+      "loss": 7.4413,
+      "step": 244
+    },
+    {
+      "epoch": 0.5442932518744793,
+      "grad_norm": 0.02506766840815544,
+      "learning_rate": 0.002728395061728395,
+      "loss": 7.4413,
+      "step": 245
+    },
+    {
+      "epoch": 0.546514856984171,
+      "grad_norm": 0.023775186389684677,
+      "learning_rate": 0.002725925925925926,
+      "loss": 7.6108,
+      "step": 246
+    },
+    {
+      "epoch": 0.5487364620938628,
+      "grad_norm": 0.025394640862941742,
+      "learning_rate": 0.0027234567901234566,
+      "loss": 7.5574,
+      "step": 247
+    },
+    {
+      "epoch": 0.5509580672035546,
+      "grad_norm": 0.023591671139001846,
+      "learning_rate": 0.002720987654320988,
+      "loss": 7.8029,
+      "step": 248
+    },
+    {
+      "epoch": 0.5531796723132463,
+      "grad_norm": 0.025905868038535118,
+      "learning_rate": 0.0027185185185185185,
+      "loss": 7.2404,
+      "step": 249
+    },
+    {
+      "epoch": 0.5554012774229381,
+      "grad_norm": 0.02958277426660061,
+      "learning_rate": 0.0027160493827160493,
+      "loss": 7.1893,
+      "step": 250
+    },
+    {
+      "epoch": 0.5576228825326298,
+      "grad_norm": 0.028149964287877083,
+      "learning_rate": 0.0027135802469135805,
+      "loss": 7.2321,
+      "step": 251
+    },
+    {
+      "epoch": 0.5598444876423215,
+      "grad_norm": 0.024707626551389694,
+      "learning_rate": 0.002711111111111111,
+      "loss": 7.573,
+      "step": 252
+    },
+    {
+      "epoch": 0.5620660927520134,
+      "grad_norm": 0.028305284678936005,
+      "learning_rate": 0.002708641975308642,
+      "loss": 7.2663,
+      "step": 253
+    },
+    {
+      "epoch": 0.5642876978617051,
+      "grad_norm": 0.024400560185313225,
+      "learning_rate": 0.0027061728395061727,
+      "loss": 7.4774,
+      "step": 254
+    },
+    {
+      "epoch": 0.5665093029713968,
+      "grad_norm": 0.021780114620923996,
+      "learning_rate": 0.002703703703703704,
+      "loss": 7.5634,
+      "step": 255
+    },
+    {
+      "epoch": 0.5687309080810886,
+      "grad_norm": 0.02674521692097187,
+      "learning_rate": 0.002701234567901235,
+      "loss": 7.5755,
+      "step": 256
+    },
+    {
+      "epoch": 0.5709525131907803,
+      "grad_norm": 0.024626342579722404,
+      "learning_rate": 0.0026987654320987653,
+      "loss": 7.8217,
+      "step": 257
+    },
+    {
+      "epoch": 0.573174118300472,
+      "grad_norm": 0.022464681416749954,
+      "learning_rate": 0.0026962962962962965,
+      "loss": 7.5361,
+      "step": 258
+    },
+    {
+      "epoch": 0.5753957234101639,
+      "grad_norm": 0.024344492703676224,
+      "learning_rate": 0.002693827160493827,
+      "loss": 7.6709,
+      "step": 259
+    },
+    {
+      "epoch": 0.5776173285198556,
+      "grad_norm": 0.02072468213737011,
+      "learning_rate": 0.0026913580246913584,
+      "loss": 7.4544,
+      "step": 260
+    },
+    {
+      "epoch": 0.5798389336295473,
+      "grad_norm": 0.025870997458696365,
+      "learning_rate": 0.0026888888888888887,
+      "loss": 7.3973,
+      "step": 261
+    },
+    {
+      "epoch": 0.5820605387392391,
+      "grad_norm": 0.023839730769395828,
+      "learning_rate": 0.00268641975308642,
+      "loss": 7.5558,
+      "step": 262
+    },
+    {
+      "epoch": 0.5842821438489308,
+      "grad_norm": 0.024624641984701157,
+      "learning_rate": 0.0026839506172839506,
+      "loss": 7.4092,
+      "step": 263
+    },
+    {
+      "epoch": 0.5865037489586226,
+      "grad_norm": 0.02999279834330082,
+      "learning_rate": 0.0026814814814814817,
+      "loss": 7.3372,
+      "step": 264
+    },
+    {
+      "epoch": 0.5887253540683144,
+      "grad_norm": 0.025074807927012444,
+      "learning_rate": 0.0026790123456790125,
+      "loss": 7.3204,
+      "step": 265
+    },
+    {
+      "epoch": 0.5909469591780061,
+      "grad_norm": 0.02607882395386696,
+      "learning_rate": 0.002676543209876543,
+      "loss": 7.5993,
+      "step": 266
+    },
+    {
+      "epoch": 0.5931685642876978,
+      "grad_norm": 0.024859610944986343,
+      "learning_rate": 0.0026740740740740744,
+      "loss": 7.3766,
+      "step": 267
+    },
+    {
+      "epoch": 0.5953901693973896,
+      "grad_norm": 0.03427242487668991,
+      "learning_rate": 0.0026716049382716047,
+      "loss": 7.3104,
+      "step": 268
+    },
+    {
+      "epoch": 0.5976117745070814,
+      "grad_norm": 0.024289652705192566,
+      "learning_rate": 0.002669135802469136,
+      "loss": 7.5856,
+      "step": 269
+    },
+    {
+      "epoch": 0.5998333796167731,
+      "grad_norm": 0.023779133334755898,
+      "learning_rate": 0.0026666666666666666,
+      "loss": 7.6531,
+      "step": 270
+    },
+    {
+      "epoch": 0.6020549847264649,
+      "grad_norm": 0.02560480311512947,
+      "learning_rate": 0.0026641975308641977,
+      "loss": 7.39,
+      "step": 271
+    },
+    {
+      "epoch": 0.6042765898361566,
+      "grad_norm": 0.023350968956947327,
+      "learning_rate": 0.0026617283950617285,
+      "loss": 7.4113,
+      "step": 272
+    },
+    {
+      "epoch": 0.6064981949458483,
+      "grad_norm": 0.031989067792892456,
+      "learning_rate": 0.002659259259259259,
+      "loss": 7.405,
+      "step": 273
+    },
+    {
+      "epoch": 0.6087198000555402,
+      "grad_norm": 0.03666763752698898,
+      "learning_rate": 0.0026567901234567904,
+      "loss": 7.1569,
+      "step": 274
+    },
+    {
+      "epoch": 0.6109414051652319,
+      "grad_norm": 0.024998236447572708,
+      "learning_rate": 0.002654320987654321,
+      "loss": 7.6451,
+      "step": 275
+    },
+    {
+      "epoch": 0.6131630102749236,
+      "grad_norm": 0.029696516692638397,
+      "learning_rate": 0.002651851851851852,
+      "loss": 7.2943,
+      "step": 276
+    },
+    {
+      "epoch": 0.6153846153846154,
+      "grad_norm": 0.02424914389848709,
+      "learning_rate": 0.0026493827160493826,
+      "loss": 7.4713,
+      "step": 277
+    },
+    {
+      "epoch": 0.6176062204943071,
+      "grad_norm": 0.026769736781716347,
+      "learning_rate": 0.0026469135802469138,
+      "loss": 7.4151,
+      "step": 278
+    },
+    {
+      "epoch": 0.6198278256039988,
+      "grad_norm": 0.027783267199993134,
+      "learning_rate": 0.0026444444444444445,
+      "loss": 7.232,
+      "step": 279
+    },
+    {
+      "epoch": 0.6220494307136907,
+      "grad_norm": 0.02091244049370289,
+      "learning_rate": 0.0026419753086419752,
+      "loss": 7.3426,
+      "step": 280
+    },
+    {
+      "epoch": 0.6242710358233824,
+      "grad_norm": 0.027589626610279083,
+      "learning_rate": 0.0026395061728395064,
+      "loss": 7.4943,
+      "step": 281
+    },
+    {
+      "epoch": 0.6264926409330741,
+      "grad_norm": 0.026247195899486542,
+      "learning_rate": 0.002637037037037037,
+      "loss": 7.6639,
+      "step": 282
+    },
+    {
+      "epoch": 0.6287142460427659,
+      "grad_norm": 0.027282902970910072,
+      "learning_rate": 0.002634567901234568,
+      "loss": 7.7374,
+      "step": 283
+    },
+    {
+      "epoch": 0.6309358511524576,
+      "grad_norm": 0.025696996599435806,
+      "learning_rate": 0.0026320987654320986,
+      "loss": 7.4841,
+      "step": 284
+    },
+    {
+      "epoch": 0.6331574562621494,
+      "grad_norm": 0.025370385497808456,
+      "learning_rate": 0.0026296296296296298,
+      "loss": 7.306,
+      "step": 285
+    },
+    {
+      "epoch": 0.6353790613718412,
+      "grad_norm": 0.024677366018295288,
+      "learning_rate": 0.0026271604938271605,
+      "loss": 7.4392,
+      "step": 286
+    },
+    {
+      "epoch": 0.6376006664815329,
+      "grad_norm": 0.02663363516330719,
+      "learning_rate": 0.0026246913580246912,
+      "loss": 7.4561,
+      "step": 287
+    },
+    {
+      "epoch": 0.6398222715912246,
+      "grad_norm": 0.025270260870456696,
+      "learning_rate": 0.0026222222222222224,
+      "loss": 7.591,
+      "step": 288
+    },
+    {
+      "epoch": 0.6420438767009164,
+      "grad_norm": 0.026884667575359344,
+      "learning_rate": 0.002619753086419753,
+      "loss": 7.4559,
+      "step": 289
+    },
+    {
+      "epoch": 0.6442654818106082,
+      "grad_norm": 0.02411939948797226,
+      "learning_rate": 0.002617283950617284,
+      "loss": 7.6314,
+      "step": 290
+    },
+    {
+      "epoch": 0.6464870869202999,
+      "grad_norm": 0.026495875790715218,
+      "learning_rate": 0.0026148148148148146,
+      "loss": 7.141,
+      "step": 291
+    },
+    {
+      "epoch": 0.6487086920299917,
+      "grad_norm": 0.02428489178419113,
+      "learning_rate": 0.0026123456790123458,
+      "loss": 7.1076,
+      "step": 292
+    },
+    {
+      "epoch": 0.6509302971396834,
+      "grad_norm": 0.026380060240626335,
+      "learning_rate": 0.0026098765432098765,
+      "loss": 7.1815,
+      "step": 293
+    },
+    {
+      "epoch": 0.6531519022493751,
+      "grad_norm": 0.024545544758439064,
+      "learning_rate": 0.0026074074074074072,
+      "loss": 7.2299,
+      "step": 294
+    },
+    {
+      "epoch": 0.655373507359067,
+      "grad_norm": 0.030594579875469208,
+      "learning_rate": 0.0026049382716049384,
+      "loss": 7.1528,
+      "step": 295
+    },
+    {
+      "epoch": 0.6575951124687587,
+      "grad_norm": 0.023866072297096252,
+      "learning_rate": 0.002602469135802469,
+      "loss": 7.5857,
+      "step": 296
+    },
+    {
+      "epoch": 0.6598167175784504,
+      "grad_norm": 0.02427583374083042,
+      "learning_rate": 0.0026000000000000003,
+      "loss": 7.0576,
+      "step": 297
+    },
+    {
+      "epoch": 0.6620383226881422,
+      "grad_norm": 0.02977169305086136,
+      "learning_rate": 0.0025975308641975306,
+      "loss": 7.1765,
+      "step": 298
+    },
+    {
+      "epoch": 0.6642599277978339,
+      "grad_norm": 0.02559109963476658,
+      "learning_rate": 0.002595061728395062,
+      "loss": 7.3946,
+      "step": 299
+    },
+    {
+      "epoch": 0.6664815329075257,
+      "grad_norm": 0.024115154519677162,
+      "learning_rate": 0.0025925925925925925,
+      "loss": 7.5401,
+      "step": 300
+    },
+    {
+      "epoch": 0.6687031380172175,
+      "grad_norm": 0.02193550579249859,
+      "learning_rate": 0.0025901234567901237,
+      "loss": 7.3353,
+      "step": 301
+    },
+    {
+      "epoch": 0.6709247431269092,
+      "grad_norm": 0.02353992499411106,
+      "learning_rate": 0.0025876543209876544,
+      "loss": 7.3974,
+      "step": 302
+    },
+    {
+      "epoch": 0.6731463482366009,
+      "grad_norm": 0.020420249551534653,
+      "learning_rate": 0.002585185185185185,
+      "loss": 7.5565,
+      "step": 303
+    },
+    {
+      "epoch": 0.6753679533462927,
+      "grad_norm": 0.02567903883755207,
+      "learning_rate": 0.0025827160493827163,
+      "loss": 7.1631,
+      "step": 304
+    },
+    {
+      "epoch": 0.6775895584559845,
+      "grad_norm": 0.02823437750339508,
+      "learning_rate": 0.002580246913580247,
+      "loss": 7.5935,
+      "step": 305
+    },
+    {
+      "epoch": 0.6798111635656762,
+      "grad_norm": 0.022785518318414688,
+      "learning_rate": 0.002577777777777778,
+      "loss": 7.2874,
+      "step": 306
+    },
+    {
+      "epoch": 0.682032768675368,
+      "grad_norm": 0.03670334070920944,
+      "learning_rate": 0.0025753086419753085,
+      "loss": 7.0896,
+      "step": 307
+    },
+    {
+      "epoch": 0.6842543737850597,
+      "grad_norm": 0.022983603179454803,
+      "learning_rate": 0.0025728395061728397,
+      "loss": 7.4564,
+      "step": 308
+    },
+    {
+      "epoch": 0.6864759788947514,
+      "grad_norm": 0.026292193681001663,
+      "learning_rate": 0.0025703703703703704,
+      "loss": 7.1879,
+      "step": 309
+    },
+    {
+      "epoch": 0.6886975840044433,
+      "grad_norm": 0.025101250037550926,
+      "learning_rate": 0.002567901234567901,
+      "loss": 7.1207,
+      "step": 310
+    },
+    {
+      "epoch": 0.690919189114135,
+      "grad_norm": 0.027889365330338478,
+      "learning_rate": 0.0025654320987654323,
+      "loss": 7.3256,
+      "step": 311
+    },
+    {
+      "epoch": 0.6931407942238267,
+      "grad_norm": 0.029532425105571747,
+      "learning_rate": 0.002562962962962963,
+      "loss": 7.3423,
+      "step": 312
+    },
+    {
+      "epoch": 0.6953623993335185,
+      "grad_norm": 0.02462322823703289,
+      "learning_rate": 0.002560493827160494,
+      "loss": 7.1061,
+      "step": 313
+    },
+    {
+      "epoch": 0.6975840044432102,
+      "grad_norm": 0.024397345259785652,
+      "learning_rate": 0.0025580246913580245,
+      "loss": 7.3286,
+      "step": 314
+    },
+    {
+      "epoch": 0.6998056095529019,
+      "grad_norm": 0.02319549024105072,
+      "learning_rate": 0.0025555555555555557,
+      "loss": 7.496,
+      "step": 315
+    },
+    {
+      "epoch": 0.7020272146625938,
+      "grad_norm": 0.024352865293622017,
+      "learning_rate": 0.0025530864197530864,
+      "loss": 7.4917,
+      "step": 316
+    },
+    {
+      "epoch": 0.7042488197722855,
+      "grad_norm": 0.022795524448156357,
+      "learning_rate": 0.002550617283950617,
+      "loss": 7.351,
+      "step": 317
+    },
+    {
+      "epoch": 0.7064704248819772,
+      "grad_norm": 0.02231544256210327,
+      "learning_rate": 0.0025481481481481483,
+      "loss": 7.2621,
+      "step": 318
+    },
+    {
+      "epoch": 0.708692029991669,
+      "grad_norm": 0.02476155012845993,
+      "learning_rate": 0.002545679012345679,
+      "loss": 6.8912,
+      "step": 319
+    },
+    {
+      "epoch": 0.7109136351013607,
+      "grad_norm": 0.02561456710100174,
+      "learning_rate": 0.00254320987654321,
+      "loss": 7.024,
+      "step": 320
+    },
+    {
+      "epoch": 0.7131352402110525,
+      "grad_norm": 0.02208608016371727,
+      "learning_rate": 0.0025407407407407405,
+      "loss": 7.2996,
+      "step": 321
+    },
+    {
+      "epoch": 0.7153568453207443,
+      "grad_norm": 0.025326820090413094,
+      "learning_rate": 0.0025382716049382717,
+      "loss": 7.242,
+      "step": 322
+    },
+    {
+      "epoch": 0.717578450430436,
+      "grad_norm": 0.028136832639575005,
+      "learning_rate": 0.0025358024691358024,
+      "loss": 7.2752,
+      "step": 323
+    },
+    {
+      "epoch": 0.7198000555401277,
+      "grad_norm": 0.025537578389048576,
+      "learning_rate": 0.002533333333333333,
+      "loss": 7.3984,
+      "step": 324
+    },
+    {
+      "epoch": 0.7220216606498195,
+      "grad_norm": 0.026090625673532486,
+      "learning_rate": 0.0025308641975308644,
+      "loss": 7.1918,
+      "step": 325
+    },
+    {
+      "epoch": 0.7242432657595113,
+      "grad_norm": 0.02365339919924736,
+      "learning_rate": 0.002528395061728395,
+      "loss": 7.2199,
+      "step": 326
+    },
+    {
+      "epoch": 0.726464870869203,
+      "grad_norm": 0.02548445574939251,
+      "learning_rate": 0.0025259259259259263,
+      "loss": 7.4444,
+      "step": 327
+    },
+    {
+      "epoch": 0.7286864759788948,
+      "grad_norm": 0.02948014996945858,
+      "learning_rate": 0.0025234567901234566,
+      "loss": 7.1497,
+      "step": 328
+    },
+    {
+      "epoch": 0.7309080810885865,
+      "grad_norm": 0.024226009845733643,
+      "learning_rate": 0.0025209876543209877,
+      "loss": 7.2391,
+      "step": 329
+    },
+    {
+      "epoch": 0.7331296861982782,
+      "grad_norm": 0.026054972782731056,
+      "learning_rate": 0.0025185185185185185,
+      "loss": 7.4225,
+      "step": 330
+    },
+    {
+      "epoch": 0.7353512913079701,
+      "grad_norm": 0.022704575210809708,
+      "learning_rate": 0.0025160493827160496,
+      "loss": 7.0227,
+      "step": 331
+    },
+    {
+      "epoch": 0.7375728964176618,
+      "grad_norm": 0.027680471539497375,
+      "learning_rate": 0.0025135802469135804,
+      "loss": 7.1835,
+      "step": 332
+    },
+    {
+      "epoch": 0.7397945015273535,
+      "grad_norm": 0.023738441988825798,
+      "learning_rate": 0.002511111111111111,
+      "loss": 7.3861,
+      "step": 333
+    },
+    {
+      "epoch": 0.7420161066370453,
+      "grad_norm": 0.022443674504756927,
+      "learning_rate": 0.0025086419753086423,
+      "loss": 7.3538,
+      "step": 334
+    },
+    {
+      "epoch": 0.744237711746737,
+      "grad_norm": 0.024583682417869568,
+      "learning_rate": 0.002506172839506173,
+      "loss": 7.4094,
+      "step": 335
+    },
+    {
+      "epoch": 0.7464593168564287,
+      "grad_norm": 0.021103166043758392,
+      "learning_rate": 0.0025037037037037037,
+      "loss": 7.2646,
+      "step": 336
+    },
+    {
+      "epoch": 0.7486809219661206,
+      "grad_norm": 0.023894960060715675,
+      "learning_rate": 0.0025012345679012345,
+      "loss": 7.4297,
+      "step": 337
+    },
+    {
+      "epoch": 0.7509025270758123,
+      "grad_norm": 0.022599199786782265,
+      "learning_rate": 0.0024987654320987656,
+      "loss": 7.1094,
+      "step": 338
+    },
+    {
+      "epoch": 0.753124132185504,
+      "grad_norm": 0.025741979479789734,
+      "learning_rate": 0.0024962962962962964,
+      "loss": 7.2174,
+      "step": 339
+    },
+    {
+      "epoch": 0.7553457372951958,
+      "grad_norm": 0.027009744197130203,
+      "learning_rate": 0.002493827160493827,
+      "loss": 7.3833,
+      "step": 340
+    },
+    {
+      "epoch": 0.7575673424048875,
+      "grad_norm": 0.026970168575644493,
+      "learning_rate": 0.0024913580246913583,
+      "loss": 7.4427,
+      "step": 341
+    },
+    {
+      "epoch": 0.7597889475145793,
+      "grad_norm": 0.02861645445227623,
+      "learning_rate": 0.002488888888888889,
+      "loss": 6.9496,
+      "step": 342
+    },
+    {
+      "epoch": 0.7620105526242711,
+      "grad_norm": 0.02145262062549591,
+      "learning_rate": 0.0024864197530864197,
+      "loss": 7.3852,
+      "step": 343
+    },
+    {
+      "epoch": 0.7642321577339628,
+      "grad_norm": 0.023498961701989174,
+      "learning_rate": 0.0024839506172839505,
+      "loss": 7.0481,
+      "step": 344
+    },
+    {
+      "epoch": 0.7664537628436545,
+      "grad_norm": 0.026649903506040573,
+      "learning_rate": 0.0024814814814814816,
+      "loss": 7.3784,
+      "step": 345
+    },
+    {
+      "epoch": 0.7686753679533463,
+      "grad_norm": 0.026928873732686043,
+      "learning_rate": 0.0024790123456790124,
+      "loss": 7.384,
+      "step": 346
+    },
+    {
+      "epoch": 0.7708969730630381,
+      "grad_norm": 0.025973590090870857,
+      "learning_rate": 0.002476543209876543,
+      "loss": 7.076,
+      "step": 347
+    },
+    {
+      "epoch": 0.7731185781727298,
+      "grad_norm": 0.02380303107202053,
+      "learning_rate": 0.0024740740740740743,
+      "loss": 7.1654,
+      "step": 348
+    },
+    {
+      "epoch": 0.7753401832824216,
+      "grad_norm": 0.027677161619067192,
+      "learning_rate": 0.002471604938271605,
+      "loss": 7.0076,
+      "step": 349
+    },
+    {
+      "epoch": 0.7775617883921133,
+      "grad_norm": 0.022956712171435356,
+      "learning_rate": 0.0024691358024691358,
+      "loss": 7.2777,
+      "step": 350
+    },
+    {
+      "epoch": 0.779783393501805,
+      "grad_norm": 0.023897536098957062,
+      "learning_rate": 0.0024666666666666665,
+      "loss": 7.0587,
+      "step": 351
+    },
+    {
+      "epoch": 0.7820049986114969,
+      "grad_norm": 0.026128023862838745,
+      "learning_rate": 0.0024641975308641977,
+      "loss": 7.5733,
+      "step": 352
+    },
+    {
+      "epoch": 0.7842266037211886,
+      "grad_norm": 0.019262492656707764,
+      "learning_rate": 0.0024617283950617284,
+      "loss": 7.1129,
+      "step": 353
+    },
+    {
+      "epoch": 0.7864482088308803,
+      "grad_norm": 0.026841487735509872,
+      "learning_rate": 0.002459259259259259,
+      "loss": 6.9328,
+      "step": 354
+    },
+    {
+      "epoch": 0.7886698139405721,
+      "grad_norm": 0.027686692774295807,
+      "learning_rate": 0.0024567901234567903,
+      "loss": 7.1355,
+      "step": 355
+    },
+    {
+      "epoch": 0.7908914190502638,
+      "grad_norm": 0.026043646037578583,
+      "learning_rate": 0.002454320987654321,
+      "loss": 7.3791,
+      "step": 356
+    },
+    {
+      "epoch": 0.7931130241599555,
+      "grad_norm": 0.024341454729437828,
+      "learning_rate": 0.002451851851851852,
+      "loss": 6.9768,
+      "step": 357
+    },
+    {
+      "epoch": 0.7953346292696474,
+      "grad_norm": 0.028582463040947914,
+      "learning_rate": 0.0024493827160493825,
+      "loss": 7.1255,
+      "step": 358
+    },
+    {
+      "epoch": 0.7975562343793391,
+      "grad_norm": 0.01925045996904373,
+      "learning_rate": 0.0024469135802469137,
+      "loss": 7.3265,
+      "step": 359
+    },
+    {
+      "epoch": 0.7997778394890308,
+      "grad_norm": 0.025576984509825706,
+      "learning_rate": 0.0024444444444444444,
+      "loss": 7.2449,
+      "step": 360
+    },
+    {
+      "epoch": 0.8019994445987225,
+      "grad_norm": 0.025266211479902267,
+      "learning_rate": 0.0024419753086419756,
+      "loss": 7.2763,
+      "step": 361
+    },
+    {
+      "epoch": 0.8042210497084143,
+      "grad_norm": 0.02819187007844448,
+      "learning_rate": 0.0024395061728395063,
+      "loss": 7.0403,
+      "step": 362
+    },
+    {
+      "epoch": 0.8064426548181061,
+      "grad_norm": 0.021835366263985634,
+      "learning_rate": 0.002437037037037037,
+      "loss": 6.9576,
+      "step": 363
+    },
+    {
+      "epoch": 0.8086642599277978,
+      "grad_norm": 0.028835922479629517,
+      "learning_rate": 0.002434567901234568,
+      "loss": 7.2849,
+      "step": 364
+    },
+    {
+      "epoch": 0.8108858650374896,
+      "grad_norm": 0.029858190566301346,
+      "learning_rate": 0.002432098765432099,
+      "loss": 7.4981,
+      "step": 365
+    },
+    {
+      "epoch": 0.8131074701471813,
+      "grad_norm": 0.031198391690850258,
+      "learning_rate": 0.0024296296296296297,
+      "loss": 7.5295,
+      "step": 366
+    },
+    {
+      "epoch": 0.815329075256873,
+      "grad_norm": 0.022797249257564545,
+      "learning_rate": 0.0024271604938271604,
+      "loss": 6.9192,
+      "step": 367
+    },
+    {
+      "epoch": 0.8175506803665649,
+      "grad_norm": 0.024386614561080933,
+      "learning_rate": 0.0024246913580246916,
+      "loss": 7.2783,
+      "step": 368
+    },
+    {
+      "epoch": 0.8197722854762566,
+      "grad_norm": 0.02686323970556259,
+      "learning_rate": 0.0024222222222222223,
+      "loss": 7.1663,
+      "step": 369
+    },
+    {
+      "epoch": 0.8219938905859483,
+      "grad_norm": 0.023222779855132103,
+      "learning_rate": 0.002419753086419753,
+      "loss": 7.0728,
+      "step": 370
+    },
+    {
+      "epoch": 0.8242154956956401,
+      "grad_norm": 0.024168046191334724,
+      "learning_rate": 0.002417283950617284,
+      "loss": 7.1315,
+      "step": 371
+    },
+    {
+      "epoch": 0.8264371008053318,
+      "grad_norm": 0.022769639268517494,
+      "learning_rate": 0.002414814814814815,
+      "loss": 7.0673,
+      "step": 372
+    },
+    {
+      "epoch": 0.8286587059150236,
+      "grad_norm": 0.024920092895627022,
+      "learning_rate": 0.0024123456790123457,
+      "loss": 7.2212,
+      "step": 373
+    },
+    {
+      "epoch": 0.8308803110247154,
+      "grad_norm": 0.030025171115994453,
+      "learning_rate": 0.0024098765432098764,
+      "loss": 7.4588,
+      "step": 374
+    },
+    {
+      "epoch": 0.8331019161344071,
+      "grad_norm": 0.03373974934220314,
+      "learning_rate": 0.0024074074074074076,
+      "loss": 7.0058,
+      "step": 375
+    },
+    {
+      "epoch": 0.8353235212440988,
+      "grad_norm": 0.022572096437215805,
+      "learning_rate": 0.0024049382716049383,
+      "loss": 7.1412,
+      "step": 376
+    },
+    {
+      "epoch": 0.8375451263537906,
+      "grad_norm": 0.026795724406838417,
+      "learning_rate": 0.002402469135802469,
+      "loss": 7.0259,
+      "step": 377
+    },
+    {
+      "epoch": 0.8397667314634824,
+      "grad_norm": 0.026618717238307,
+      "learning_rate": 0.0024000000000000002,
+      "loss": 7.1129,
+      "step": 378
+    },
+    {
+      "epoch": 0.8419883365731741,
+      "grad_norm": 0.02441958524286747,
+      "learning_rate": 0.002397530864197531,
+      "loss": 7.0767,
+      "step": 379
+    },
+    {
+      "epoch": 0.8442099416828659,
+      "grad_norm": 0.02320600487291813,
+      "learning_rate": 0.0023950617283950617,
+      "loss": 7.0922,
+      "step": 380
+    },
+    {
+      "epoch": 0.8464315467925576,
+      "grad_norm": 0.027361933141946793,
+      "learning_rate": 0.0023925925925925924,
+      "loss": 7.0389,
+      "step": 381
+    },
+    {
+      "epoch": 0.8486531519022493,
+      "grad_norm": 0.028075607493519783,
+      "learning_rate": 0.0023901234567901236,
+      "loss": 7.0829,
+      "step": 382
+    },
+    {
+      "epoch": 0.8508747570119412,
+      "grad_norm": 0.027039077132940292,
+      "learning_rate": 0.0023876543209876543,
+      "loss": 7.0892,
+      "step": 383
+    },
+    {
+      "epoch": 0.8530963621216329,
+      "grad_norm": 0.027716996148228645,
+      "learning_rate": 0.002385185185185185,
+      "loss": 7.0117,
+      "step": 384
+    },
+    {
+      "epoch": 0.8553179672313246,
+      "grad_norm": 0.024904819205403328,
+      "learning_rate": 0.0023827160493827162,
+      "loss": 7.2925,
+      "step": 385
+    },
+    {
+      "epoch": 0.8575395723410164,
+      "grad_norm": 0.023340992629528046,
+      "learning_rate": 0.002380246913580247,
+      "loss": 7.0719,
+      "step": 386
+    },
+    {
+      "epoch": 0.8597611774507081,
+      "grad_norm": 0.025253724306821823,
+      "learning_rate": 0.002377777777777778,
+      "loss": 7.2108,
+      "step": 387
+    },
+    {
+      "epoch": 0.8619827825603998,
+      "grad_norm": 0.031119728460907936,
+      "learning_rate": 0.0023753086419753084,
+      "loss": 7.3954,
+      "step": 388
+    },
+    {
+      "epoch": 0.8642043876700917,
+      "grad_norm": 0.02352498285472393,
+      "learning_rate": 0.0023728395061728396,
+      "loss": 7.0734,
+      "step": 389
+    },
+    {
+      "epoch": 0.8664259927797834,
+      "grad_norm": 0.024964703246951103,
+      "learning_rate": 0.0023703703703703703,
+      "loss": 6.9242,
+      "step": 390
+    },
+    {
+      "epoch": 0.8686475978894751,
+      "grad_norm": 0.03068098984658718,
+      "learning_rate": 0.0023679012345679015,
+      "loss": 6.8725,
+      "step": 391
+    },
+    {
+      "epoch": 0.8708692029991669,
+      "grad_norm": 0.025133926421403885,
+      "learning_rate": 0.0023654320987654322,
+      "loss": 6.9236,
+      "step": 392
+    },
+    {
+      "epoch": 0.8730908081088586,
+      "grad_norm": 0.02985168993473053,
+      "learning_rate": 0.002362962962962963,
+      "loss": 7.032,
+      "step": 393
+    },
+    {
+      "epoch": 0.8753124132185504,
+      "grad_norm": 0.023576263338327408,
+      "learning_rate": 0.002360493827160494,
+      "loss": 7.4937,
+      "step": 394
+    },
+    {
+      "epoch": 0.8775340183282422,
+      "grad_norm": 0.027415264397859573,
+      "learning_rate": 0.002358024691358025,
+      "loss": 6.9758,
+      "step": 395
+    },
+    {
+      "epoch": 0.8797556234379339,
+      "grad_norm": 0.02160666324198246,
+      "learning_rate": 0.0023555555555555556,
+      "loss": 7.1107,
+      "step": 396
+    },
+    {
+      "epoch": 0.8819772285476256,
+      "grad_norm": 0.026980215683579445,
+      "learning_rate": 0.0023530864197530863,
+      "loss": 7.2185,
+      "step": 397
+    },
+    {
+      "epoch": 0.8841988336573174,
+      "grad_norm": 0.03549497202038765,
+      "learning_rate": 0.0023506172839506175,
+      "loss": 6.7697,
+      "step": 398
+    },
+    {
+      "epoch": 0.8864204387670092,
+      "grad_norm": 0.024073494598269463,
+      "learning_rate": 0.002348148148148148,
+      "loss": 7.0492,
+      "step": 399
+    },
+    {
+      "epoch": 0.8886420438767009,
+      "grad_norm": 0.023343253880739212,
+      "learning_rate": 0.002345679012345679,
+      "loss": 7.0478,
+      "step": 400
+    },
+    {
+      "epoch": 0.8908636489863927,
+      "grad_norm": 0.02530418150126934,
+      "learning_rate": 0.00234320987654321,
+      "loss": 7.1246,
+      "step": 401
+    },
+    {
+      "epoch": 0.8930852540960844,
+      "grad_norm": 0.02425829879939556,
+      "learning_rate": 0.002340740740740741,
+      "loss": 6.9963,
+      "step": 402
+    },
+    {
+      "epoch": 0.8953068592057761,
+      "grad_norm": 0.023169947788119316,
+      "learning_rate": 0.0023382716049382716,
+      "loss": 7.0213,
+      "step": 403
+    },
+    {
+      "epoch": 0.897528464315468,
+      "grad_norm": 0.031261272728443146,
+      "learning_rate": 0.0023358024691358024,
+      "loss": 6.9636,
+      "step": 404
+    },
+    {
+      "epoch": 0.8997500694251597,
+      "grad_norm": 0.02489764802157879,
+      "learning_rate": 0.0023333333333333335,
+      "loss": 7.0127,
+      "step": 405
+    },
+    {
+      "epoch": 0.9019716745348514,
+      "grad_norm": 0.024346090853214264,
+      "learning_rate": 0.0023308641975308643,
+      "loss": 7.0773,
+      "step": 406
+    },
+    {
+      "epoch": 0.9041932796445432,
+      "grad_norm": 0.028360968455672264,
+      "learning_rate": 0.002328395061728395,
+      "loss": 7.146,
+      "step": 407
+    },
+    {
+      "epoch": 0.9064148847542349,
+      "grad_norm": 0.02553647756576538,
+      "learning_rate": 0.002325925925925926,
+      "loss": 7.1414,
+      "step": 408
+    },
+    {
+      "epoch": 0.9086364898639266,
+      "grad_norm": 0.02944980002939701,
+      "learning_rate": 0.002323456790123457,
+      "loss": 7.0134,
+      "step": 409
+    },
+    {
+      "epoch": 0.9108580949736185,
+      "grad_norm": 0.022414475679397583,
+      "learning_rate": 0.0023209876543209876,
+      "loss": 7.5138,
+      "step": 410
+    },
+    {
+      "epoch": 0.9130797000833102,
+      "grad_norm": 0.028593771159648895,
+      "learning_rate": 0.0023185185185185184,
+      "loss": 6.9529,
+      "step": 411
+    },
+    {
+      "epoch": 0.9153013051930019,
+      "grad_norm": 0.02755245380103588,
+      "learning_rate": 0.0023160493827160495,
+      "loss": 7.0148,
+      "step": 412
+    },
+    {
+      "epoch": 0.9175229103026937,
+      "grad_norm": 0.029654711484909058,
+      "learning_rate": 0.0023135802469135803,
+      "loss": 7.0715,
+      "step": 413
+    },
+    {
+      "epoch": 0.9197445154123854,
+      "grad_norm": 0.029837142676115036,
+      "learning_rate": 0.002311111111111111,
+      "loss": 7.2309,
+      "step": 414
+    },
+    {
+      "epoch": 0.9219661205220772,
+      "grad_norm": 0.03259390965104103,
+      "learning_rate": 0.002308641975308642,
+      "loss": 6.6208,
+      "step": 415
+    },
+    {
+      "epoch": 0.924187725631769,
+      "grad_norm": 0.028287826105952263,
+      "learning_rate": 0.002306172839506173,
+      "loss": 7.0942,
+      "step": 416
+    },
+    {
+      "epoch": 0.9264093307414607,
+      "grad_norm": 0.024518249556422234,
+      "learning_rate": 0.0023037037037037036,
+      "loss": 6.9903,
+      "step": 417
+    },
+    {
+      "epoch": 0.9286309358511524,
+      "grad_norm": 0.020759275183081627,
+      "learning_rate": 0.0023012345679012344,
+      "loss": 6.9734,
+      "step": 418
+    },
+    {
+      "epoch": 0.9308525409608442,
+      "grad_norm": 0.025013025850057602,
+      "learning_rate": 0.0022987654320987655,
+      "loss": 7.0148,
+      "step": 419
+    },
+    {
+      "epoch": 0.933074146070536,
+      "grad_norm": 0.029127171263098717,
+      "learning_rate": 0.0022962962962962963,
+      "loss": 7.2263,
+      "step": 420
+    },
+    {
+      "epoch": 0.9352957511802277,
+      "grad_norm": 0.023335881531238556,
+      "learning_rate": 0.002293827160493827,
+      "loss": 7.0932,
+      "step": 421
+    },
+    {
+      "epoch": 0.9375173562899195,
+      "grad_norm": 0.02313905581831932,
+      "learning_rate": 0.002291358024691358,
+      "loss": 7.1537,
+      "step": 422
+    },
+    {
+      "epoch": 0.9397389613996112,
+      "grad_norm": 0.03614825755357742,
+      "learning_rate": 0.002288888888888889,
+      "loss": 6.7949,
+      "step": 423
+    },
+    {
+      "epoch": 0.9419605665093029,
+      "grad_norm": 0.026687975972890854,
+      "learning_rate": 0.00228641975308642,
+      "loss": 6.8291,
+      "step": 424
+    },
+    {
+      "epoch": 0.9441821716189948,
+      "grad_norm": 0.020853884518146515,
+      "learning_rate": 0.0022839506172839504,
+      "loss": 7.058,
+      "step": 425
+    },
+    {
+      "epoch": 0.9464037767286865,
+      "grad_norm": 0.030094467103481293,
+      "learning_rate": 0.0022814814814814816,
+      "loss": 6.9071,
+      "step": 426
+    },
+    {
+      "epoch": 0.9486253818383782,
+      "grad_norm": 0.027994267642498016,
+      "learning_rate": 0.0022790123456790123,
+      "loss": 7.0707,
+      "step": 427
+    },
+    {
+      "epoch": 0.95084698694807,
+      "grad_norm": 0.03092566318809986,
+      "learning_rate": 0.0022765432098765435,
+      "loss": 6.7857,
+      "step": 428
+    },
+    {
+      "epoch": 0.9530685920577617,
+      "grad_norm": 0.02169840969145298,
+      "learning_rate": 0.0022740740740740738,
+      "loss": 6.9255,
+      "step": 429
+    },
+    {
+      "epoch": 0.9552901971674534,
+      "grad_norm": 0.02525992877781391,
+      "learning_rate": 0.002271604938271605,
+      "loss": 6.9151,
+      "step": 430
+    },
+    {
+      "epoch": 0.9575118022771453,
+      "grad_norm": 0.023625439032912254,
+      "learning_rate": 0.002269135802469136,
+      "loss": 6.9183,
+      "step": 431
+    },
+    {
+      "epoch": 0.959733407386837,
+      "grad_norm": 0.024517634883522987,
+      "learning_rate": 0.002266666666666667,
+      "loss": 6.9958,
+      "step": 432
+    },
+    {
+      "epoch": 0.9619550124965287,
+      "grad_norm": 0.02542831376194954,
+      "learning_rate": 0.0022641975308641976,
+      "loss": 6.7201,
+      "step": 433
+    },
+    {
+      "epoch": 0.9641766176062205,
+      "grad_norm": 0.032386329025030136,
+      "learning_rate": 0.0022617283950617283,
+      "loss": 6.7065,
+      "step": 434
+    },
+    {
+      "epoch": 0.9663982227159122,
+      "grad_norm": 0.021518751978874207,
+      "learning_rate": 0.0022592592592592595,
+      "loss": 7.1132,
+      "step": 435
+    },
+    {
+      "epoch": 0.968619827825604,
+      "grad_norm": 0.028782367706298828,
+      "learning_rate": 0.00225679012345679,
+      "loss": 6.9016,
+      "step": 436
+    },
+    {
+      "epoch": 0.9708414329352958,
+      "grad_norm": 0.025123797357082367,
+      "learning_rate": 0.002254320987654321,
+      "loss": 6.72,
+      "step": 437
+    },
+    {
+      "epoch": 0.9730630380449875,
+      "grad_norm": 0.03904864937067032,
+      "learning_rate": 0.002251851851851852,
+      "loss": 7.0102,
+      "step": 438
+    },
+    {
+      "epoch": 0.9752846431546792,
+      "grad_norm": 0.024356381967663765,
+      "learning_rate": 0.002249382716049383,
+      "loss": 7.2729,
+      "step": 439
+    },
+    {
+      "epoch": 0.977506248264371,
+      "grad_norm": 0.03238334879279137,
+      "learning_rate": 0.0022469135802469136,
+      "loss": 7.051,
+      "step": 440
+    },
+    {
+      "epoch": 0.9797278533740628,
+      "grad_norm": 0.023336123675107956,
+      "learning_rate": 0.0022444444444444443,
+      "loss": 7.4224,
+      "step": 441
+    },
+    {
+      "epoch": 0.9819494584837545,
+      "grad_norm": 0.03011210262775421,
+      "learning_rate": 0.0022419753086419755,
+      "loss": 7.0443,
+      "step": 442
+    },
+    {
+      "epoch": 0.9841710635934463,
+      "grad_norm": 0.028482751920819283,
+      "learning_rate": 0.002239506172839506,
+      "loss": 6.6453,
+      "step": 443
+    },
+    {
+      "epoch": 0.986392668703138,
+      "grad_norm": 0.03540631756186485,
+      "learning_rate": 0.002237037037037037,
+      "loss": 6.6587,
+      "step": 444
+    },
+    {
+      "epoch": 0.9886142738128297,
+      "grad_norm": 0.03502856567502022,
+      "learning_rate": 0.002234567901234568,
+      "loss": 6.5522,
+      "step": 445
+    },
+    {
+      "epoch": 0.9908358789225216,
+      "grad_norm": 0.03989952802658081,
+      "learning_rate": 0.002232098765432099,
+      "loss": 6.9773,
+      "step": 446
+    },
+    {
+      "epoch": 0.9930574840322133,
+      "grad_norm": 0.027936214581131935,
+      "learning_rate": 0.0022296296296296296,
+      "loss": 6.8403,
+      "step": 447
+    },
+    {
+      "epoch": 0.995279089141905,
+      "grad_norm": 0.021183231845498085,
+      "learning_rate": 0.0022271604938271603,
+      "loss": 6.913,
+      "step": 448
+    },
+    {
+      "epoch": 0.9975006942515968,
+      "grad_norm": 0.02642223611474037,
+      "learning_rate": 0.0022246913580246915,
+      "loss": 6.8466,
+      "step": 449
+    },
+    {
+      "epoch": 0.9997222993612885,
+      "grad_norm": 0.033271968364715576,
+      "learning_rate": 0.0022222222222222222,
+      "loss": 6.9681,
+      "step": 450
+    },
+    {
+      "epoch": 1.0019439044709804,
+      "grad_norm": 0.02616877481341362,
+      "learning_rate": 0.002219753086419753,
+      "loss": 6.7205,
+      "step": 451
+    },
+    {
+      "epoch": 1.004165509580672,
+      "grad_norm": 0.03265068680047989,
+      "learning_rate": 0.002217283950617284,
+      "loss": 6.8105,
+      "step": 452
+    },
+    {
+      "epoch": 1.0063871146903638,
+      "grad_norm": 0.034761156886816025,
+      "learning_rate": 0.002214814814814815,
+      "loss": 6.9132,
+      "step": 453
+    },
+    {
+      "epoch": 1.0086087198000555,
+      "grad_norm": 0.02546847052872181,
+      "learning_rate": 0.002212345679012346,
+      "loss": 7.0172,
+      "step": 454
+    },
+    {
+      "epoch": 1.0108303249097472,
+      "grad_norm": 0.027320275083184242,
+      "learning_rate": 0.0022098765432098763,
+      "loss": 6.7983,
+      "step": 455
+    },
+    {
+      "epoch": 1.013051930019439,
+      "grad_norm": 0.03206401690840721,
+      "learning_rate": 0.0022074074074074075,
+      "loss": 7.0936,
+      "step": 456
+    },
+    {
+      "epoch": 1.0152735351291309,
+      "grad_norm": 0.026150790974497795,
+      "learning_rate": 0.0022049382716049382,
+      "loss": 6.9636,
+      "step": 457
+    },
+    {
+      "epoch": 1.0174951402388226,
+      "grad_norm": 0.023814262822270393,
+      "learning_rate": 0.0022024691358024694,
+      "loss": 6.9101,
+      "step": 458
+    },
+    {
+      "epoch": 1.0197167453485143,
+      "grad_norm": 0.0383935272693634,
+      "learning_rate": 0.0021999999999999997,
+      "loss": 6.364,
+      "step": 459
+    },
+    {
+      "epoch": 1.021938350458206,
+      "grad_norm": 0.03840316832065582,
+      "learning_rate": 0.002197530864197531,
+      "loss": 6.7273,
+      "step": 460
+    },
+    {
+      "epoch": 1.0241599555678977,
+      "grad_norm": 0.025719953700900078,
+      "learning_rate": 0.002195061728395062,
+      "loss": 6.8634,
+      "step": 461
+    },
+    {
+      "epoch": 1.0263815606775895,
+      "grad_norm": 0.024488503113389015,
+      "learning_rate": 0.0021925925925925928,
+      "loss": 6.9608,
+      "step": 462
+    },
+    {
+      "epoch": 1.0286031657872814,
+      "grad_norm": 0.03528802841901779,
+      "learning_rate": 0.0021901234567901235,
+      "loss": 6.6742,
+      "step": 463
+    },
+    {
+      "epoch": 1.030824770896973,
+      "grad_norm": 0.0282592736184597,
+      "learning_rate": 0.0021876543209876542,
+      "loss": 7.0349,
+      "step": 464
+    },
+    {
+      "epoch": 1.0330463760066648,
+      "grad_norm": 0.02441062405705452,
+      "learning_rate": 0.0021851851851851854,
+      "loss": 7.0426,
+      "step": 465
+    },
+    {
+      "epoch": 1.0352679811163565,
+      "grad_norm": 0.027456697076559067,
+      "learning_rate": 0.002182716049382716,
+      "loss": 6.7109,
+      "step": 466
+    },
+    {
+      "epoch": 1.0374895862260483,
+      "grad_norm": 0.026548922061920166,
+      "learning_rate": 0.002180246913580247,
+      "loss": 6.8414,
+      "step": 467
+    },
+    {
+      "epoch": 1.03971119133574,
+      "grad_norm": 0.0322716198861599,
+      "learning_rate": 0.002177777777777778,
+      "loss": 6.9097,
+      "step": 468
+    },
+    {
+      "epoch": 1.041932796445432,
+      "grad_norm": 0.02603553794324398,
+      "learning_rate": 0.0021753086419753088,
+      "loss": 6.8715,
+      "step": 469
+    },
+    {
+      "epoch": 1.0441544015551236,
+      "grad_norm": 0.02610725536942482,
+      "learning_rate": 0.0021728395061728395,
+      "loss": 6.9104,
+      "step": 470
+    },
+    {
+      "epoch": 1.0463760066648153,
+      "grad_norm": 0.028794175013899803,
+      "learning_rate": 0.0021703703703703702,
+      "loss": 6.8365,
+      "step": 471
+    },
+    {
+      "epoch": 1.048597611774507,
+      "grad_norm": 0.031022299081087112,
+      "learning_rate": 0.0021679012345679014,
+      "loss": 7.0016,
+      "step": 472
+    },
+    {
+      "epoch": 1.0508192168841988,
+      "grad_norm": 0.037397563457489014,
+      "learning_rate": 0.002165432098765432,
+      "loss": 6.8337,
+      "step": 473
+    },
+    {
+      "epoch": 1.0530408219938905,
+      "grad_norm": 0.03376161679625511,
+      "learning_rate": 0.002162962962962963,
+      "loss": 6.9936,
+      "step": 474
+    },
+    {
+      "epoch": 1.0552624271035824,
+      "grad_norm": 0.025874005630612373,
+      "learning_rate": 0.002160493827160494,
+      "loss": 6.9863,
+      "step": 475
+    },
+    {
+      "epoch": 1.0574840322132741,
+      "grad_norm": 0.023572055622935295,
+      "learning_rate": 0.002158024691358025,
+      "loss": 6.7167,
+      "step": 476
+    },
+    {
+      "epoch": 1.0597056373229659,
+      "grad_norm": 0.032695479691028595,
+      "learning_rate": 0.0021555555555555555,
+      "loss": 6.6609,
+      "step": 477
+    },
+    {
+      "epoch": 1.0619272424326576,
+      "grad_norm": 0.035048022866249084,
+      "learning_rate": 0.0021530864197530863,
+      "loss": 6.4713,
+      "step": 478
+    },
+    {
+      "epoch": 1.0641488475423493,
+      "grad_norm": 0.031220491975545883,
+      "learning_rate": 0.0021506172839506174,
+      "loss": 6.4922,
+      "step": 479
+    },
+    {
+      "epoch": 1.066370452652041,
+      "grad_norm": 0.025978129357099533,
+      "learning_rate": 0.002148148148148148,
+      "loss": 6.8746,
+      "step": 480
+    },
+    {
+      "epoch": 1.068592057761733,
+      "grad_norm": 0.03034123219549656,
+      "learning_rate": 0.002145679012345679,
+      "loss": 6.8272,
+      "step": 481
+    },
+    {
+      "epoch": 1.0708136628714247,
+      "grad_norm": 0.030644550919532776,
+      "learning_rate": 0.0021432098765432096,
+      "loss": 6.6396,
+      "step": 482
+    },
+    {
+      "epoch": 1.0730352679811164,
+      "grad_norm": 0.03000597283244133,
+      "learning_rate": 0.002140740740740741,
+      "loss": 6.7608,
+      "step": 483
+    },
+    {
+      "epoch": 1.075256873090808,
+      "grad_norm": 0.029728814959526062,
+      "learning_rate": 0.002138271604938272,
+      "loss": 7.0496,
+      "step": 484
+    },
+    {
+      "epoch": 1.0774784782004998,
+      "grad_norm": 0.025527026504278183,
+      "learning_rate": 0.0021358024691358023,
+      "loss": 6.93,
+      "step": 485
+    },
+    {
+      "epoch": 1.0797000833101915,
+      "grad_norm": 0.02411215379834175,
+      "learning_rate": 0.0021333333333333334,
+      "loss": 6.7498,
+      "step": 486
+    },
+    {
+      "epoch": 1.0819216884198835,
+      "grad_norm": 0.03165782615542412,
+      "learning_rate": 0.002130864197530864,
+      "loss": 7.0893,
+      "step": 487
+    },
+    {
+      "epoch": 1.0841432935295752,
+      "grad_norm": 0.03456574305891991,
+      "learning_rate": 0.0021283950617283953,
+      "loss": 7.2461,
+      "step": 488
+    },
+    {
+      "epoch": 1.0863648986392669,
+      "grad_norm": 0.031305279582738876,
+      "learning_rate": 0.0021259259259259256,
+      "loss": 6.8805,
+      "step": 489
+    },
+    {
+      "epoch": 1.0885865037489586,
+      "grad_norm": 0.03053886443376541,
+      "learning_rate": 0.002123456790123457,
+      "loss": 6.8608,
+      "step": 490
+    },
+    {
+      "epoch": 1.0908081088586503,
+      "grad_norm": 0.025559354573488235,
+      "learning_rate": 0.002120987654320988,
+      "loss": 6.8154,
+      "step": 491
+    },
+    {
+      "epoch": 1.093029713968342,
+      "grad_norm": 0.023908918723464012,
+      "learning_rate": 0.0021185185185185187,
+      "loss": 6.5367,
+      "step": 492
+    },
+    {
+      "epoch": 1.095251319078034,
+      "grad_norm": 0.02523021586239338,
+      "learning_rate": 0.0021160493827160494,
+      "loss": 6.8869,
+      "step": 493
+    },
+    {
+      "epoch": 1.0974729241877257,
+      "grad_norm": 0.02901410311460495,
+      "learning_rate": 0.00211358024691358,
+      "loss": 7.1647,
+      "step": 494
+    },
+    {
+      "epoch": 1.0996945292974174,
+      "grad_norm": 0.026036538183689117,
+      "learning_rate": 0.0021111111111111113,
+      "loss": 6.6957,
+      "step": 495
+    },
+    {
+      "epoch": 1.1019161344071091,
+      "grad_norm": 0.022965164855122566,
+      "learning_rate": 0.002108641975308642,
+      "loss": 6.9348,
+      "step": 496
+    },
+    {
+      "epoch": 1.1041377395168008,
+      "grad_norm": 0.026766669005155563,
+      "learning_rate": 0.002106172839506173,
+      "loss": 6.8628,
+      "step": 497
+    },
+    {
+      "epoch": 1.1063593446264925,
+      "grad_norm": 0.026180027052760124,
+      "learning_rate": 0.002103703703703704,
+      "loss": 6.6114,
+      "step": 498
+    },
+    {
+      "epoch": 1.1085809497361845,
+      "grad_norm": 0.02789599448442459,
+      "learning_rate": 0.0021012345679012347,
+      "loss": 7.2988,
+      "step": 499
+    },
+    {
+      "epoch": 1.1108025548458762,
+      "grad_norm": 0.024918923154473305,
+      "learning_rate": 0.0020987654320987655,
+      "loss": 6.6549,
+      "step": 500
+    },
+    {
+      "epoch": 1.113024159955568,
+      "grad_norm": 0.0366026945412159,
+      "learning_rate": 0.002096296296296296,
+      "loss": 6.6426,
+      "step": 501
+    },
+    {
+      "epoch": 1.1152457650652596,
+      "grad_norm": 0.024064958095550537,
+      "learning_rate": 0.0020938271604938274,
+      "loss": 6.952,
+      "step": 502
+    },
+    {
+      "epoch": 1.1174673701749513,
+      "grad_norm": 0.03428691625595093,
+      "learning_rate": 0.002091358024691358,
+      "loss": 6.7374,
+      "step": 503
+    },
+    {
+      "epoch": 1.119688975284643,
+      "grad_norm": 0.025278417393565178,
+      "learning_rate": 0.002088888888888889,
+      "loss": 6.9648,
+      "step": 504
+    },
+    {
+      "epoch": 1.121910580394335,
+      "grad_norm": 0.030725261196494102,
+      "learning_rate": 0.00208641975308642,
+      "loss": 6.7106,
+      "step": 505
+    },
+    {
+      "epoch": 1.1241321855040267,
+      "grad_norm": 0.030535412952303886,
+      "learning_rate": 0.0020839506172839507,
+      "loss": 6.6493,
+      "step": 506
+    },
+    {
+      "epoch": 1.1263537906137184,
+      "grad_norm": 0.03326031193137169,
+      "learning_rate": 0.0020814814814814815,
+      "loss": 6.9396,
+      "step": 507
+    },
+    {
+      "epoch": 1.1285753957234101,
+      "grad_norm": 0.02815030701458454,
+      "learning_rate": 0.002079012345679012,
+      "loss": 6.6862,
+      "step": 508
+    },
+    {
+      "epoch": 1.1307970008331019,
+      "grad_norm": 0.031352583318948746,
+      "learning_rate": 0.0020765432098765434,
+      "loss": 6.9931,
+      "step": 509
+    },
+    {
+      "epoch": 1.1330186059427936,
+      "grad_norm": 0.03211772069334984,
+      "learning_rate": 0.002074074074074074,
+      "loss": 6.9284,
+      "step": 510
+    },
+    {
+      "epoch": 1.1352402110524855,
+      "grad_norm": 0.035757821053266525,
+      "learning_rate": 0.002071604938271605,
+      "loss": 6.7358,
+      "step": 511
+    },
+    {
+      "epoch": 1.1374618161621772,
+      "grad_norm": 0.03229344263672829,
+      "learning_rate": 0.0020691358024691356,
+      "loss": 6.5755,
+      "step": 512
+    },
+    {
+      "epoch": 1.139683421271869,
+      "grad_norm": 0.03723408281803131,
+      "learning_rate": 0.0020666666666666667,
+      "loss": 6.4818,
+      "step": 513
+    },
+    {
+      "epoch": 1.1419050263815607,
+      "grad_norm": 0.0280930045992136,
+      "learning_rate": 0.002064197530864198,
+      "loss": 6.6271,
+      "step": 514
+    },
+    {
+      "epoch": 1.1441266314912524,
+      "grad_norm": 0.0272973645478487,
+      "learning_rate": 0.002061728395061728,
+      "loss": 6.9401,
+      "step": 515
+    },
+    {
+      "epoch": 1.146348236600944,
+      "grad_norm": 0.028022082522511482,
+      "learning_rate": 0.0020592592592592594,
+      "loss": 6.3293,
+      "step": 516
+    },
+    {
+      "epoch": 1.148569841710636,
+      "grad_norm": 0.02875290811061859,
+      "learning_rate": 0.00205679012345679,
+      "loss": 7.0726,
+      "step": 517
+    },
+    {
+      "epoch": 1.1507914468203277,
+      "grad_norm": 0.033188607543706894,
+      "learning_rate": 0.0020543209876543213,
+      "loss": 6.6995,
+      "step": 518
+    },
+    {
+      "epoch": 1.1530130519300195,
+      "grad_norm": 0.03337842598557472,
+      "learning_rate": 0.0020518518518518516,
+      "loss": 6.6899,
+      "step": 519
+    },
+    {
+      "epoch": 1.1552346570397112,
+      "grad_norm": 0.03756226226687431,
+      "learning_rate": 0.0020493827160493827,
+      "loss": 6.6454,
+      "step": 520
+    },
+    {
+      "epoch": 1.157456262149403,
+      "grad_norm": 0.029184823855757713,
+      "learning_rate": 0.002046913580246914,
+      "loss": 6.6364,
+      "step": 521
+    },
+    {
+      "epoch": 1.1596778672590946,
+      "grad_norm": 0.030157530680298805,
+      "learning_rate": 0.0020444444444444447,
+      "loss": 6.9319,
+      "step": 522
+    },
+    {
+      "epoch": 1.1618994723687863,
+      "grad_norm": 0.030866265296936035,
+      "learning_rate": 0.0020419753086419754,
+      "loss": 6.7,
+      "step": 523
+    },
+    {
+      "epoch": 1.1641210774784783,
+      "grad_norm": 0.03673654794692993,
+      "learning_rate": 0.002039506172839506,
+      "loss": 6.4016,
+      "step": 524
+    },
+    {
+      "epoch": 1.16634268258817,
+      "grad_norm": 0.026699114590883255,
+      "learning_rate": 0.0020370370370370373,
+      "loss": 6.8644,
+      "step": 525
+    },
+    {
+      "epoch": 1.1685642876978617,
+      "grad_norm": 0.031055321916937828,
+      "learning_rate": 0.002034567901234568,
+      "loss": 6.9568,
+      "step": 526
+    },
+    {
+      "epoch": 1.1707858928075534,
+      "grad_norm": 0.03368706256151199,
+      "learning_rate": 0.0020320987654320988,
+      "loss": 6.4072,
+      "step": 527
+    },
+    {
+      "epoch": 1.1730074979172451,
+      "grad_norm": 0.03287887200713158,
+      "learning_rate": 0.00202962962962963,
+      "loss": 6.7167,
+      "step": 528
+    },
+    {
+      "epoch": 1.175229103026937,
+      "grad_norm": 0.028194939717650414,
+      "learning_rate": 0.0020271604938271607,
+      "loss": 6.9782,
+      "step": 529
+    },
+    {
+      "epoch": 1.1774507081366288,
+      "grad_norm": 0.035647619515657425,
+      "learning_rate": 0.0020246913580246914,
+      "loss": 7.0751,
+      "step": 530
+    },
+    {
+      "epoch": 1.1796723132463205,
+      "grad_norm": 0.03429765999317169,
+      "learning_rate": 0.002022222222222222,
+      "loss": 6.7846,
+      "step": 531
+    },
+    {
+      "epoch": 1.1818939183560122,
+      "grad_norm": 0.03038814477622509,
+      "learning_rate": 0.0020197530864197533,
+      "loss": 6.7832,
+      "step": 532
+    },
+    {
+      "epoch": 1.184115523465704,
+      "grad_norm": 0.03291501849889755,
+      "learning_rate": 0.002017283950617284,
+      "loss": 6.756,
+      "step": 533
+    },
+    {
+      "epoch": 1.1863371285753956,
+      "grad_norm": 0.03274846449494362,
+      "learning_rate": 0.0020148148148148148,
+      "loss": 6.5424,
+      "step": 534
+    },
+    {
+      "epoch": 1.1885587336850874,
+      "grad_norm": 0.029521383345127106,
+      "learning_rate": 0.002012345679012346,
+      "loss": 6.5659,
+      "step": 535
+    },
+    {
+      "epoch": 1.1907803387947793,
+      "grad_norm": 0.033934336155653,
+      "learning_rate": 0.0020098765432098767,
+      "loss": 6.4736,
+      "step": 536
+    },
+    {
+      "epoch": 1.193001943904471,
+      "grad_norm": 0.03130839020013809,
+      "learning_rate": 0.0020074074074074074,
+      "loss": 6.7473,
+      "step": 537
+    },
+    {
+      "epoch": 1.1952235490141627,
+      "grad_norm": 0.025503745302557945,
+      "learning_rate": 0.002004938271604938,
+      "loss": 6.7036,
+      "step": 538
+    },
+    {
+      "epoch": 1.1974451541238544,
+      "grad_norm": 0.03354375436902046,
+      "learning_rate": 0.0020024691358024693,
+      "loss": 6.7916,
+      "step": 539
+    },
+    {
+      "epoch": 1.1996667592335462,
+      "grad_norm": 0.03416256606578827,
+      "learning_rate": 0.002,
+      "loss": 6.8635,
+      "step": 540
+    },
+    {
+      "epoch": 1.201888364343238,
+      "grad_norm": 0.029446998611092567,
+      "learning_rate": 0.0019975308641975308,
+      "loss": 7.029,
+      "step": 541
+    },
+    {
+      "epoch": 1.2041099694529298,
+      "grad_norm": 0.03176094591617584,
+      "learning_rate": 0.0019950617283950615,
+      "loss": 6.5223,
+      "step": 542
+    },
+    {
+      "epoch": 1.2063315745626215,
+      "grad_norm": 0.026768583804368973,
+      "learning_rate": 0.0019925925925925927,
+      "loss": 6.6736,
+      "step": 543
+    },
+    {
+      "epoch": 1.2085531796723132,
+      "grad_norm": 0.030084427446126938,
+      "learning_rate": 0.001990123456790124,
+      "loss": 7.0851,
+      "step": 544
+    },
+    {
+      "epoch": 1.210774784782005,
+      "grad_norm": 0.03175819665193558,
+      "learning_rate": 0.001987654320987654,
+      "loss": 7.0009,
+      "step": 545
+    },
+    {
+      "epoch": 1.2129963898916967,
+      "grad_norm": 0.033195335417985916,
+      "learning_rate": 0.0019851851851851853,
+      "loss": 6.6627,
+      "step": 546
+    },
+    {
+      "epoch": 1.2152179950013884,
+      "grad_norm": 0.027749711647629738,
+      "learning_rate": 0.001982716049382716,
+      "loss": 6.6973,
+      "step": 547
+    },
+    {
+      "epoch": 1.2174396001110803,
+      "grad_norm": 0.027580155059695244,
+      "learning_rate": 0.001980246913580247,
+      "loss": 6.7475,
+      "step": 548
+    },
+    {
+      "epoch": 1.219661205220772,
+      "grad_norm": 0.029093164950609207,
+      "learning_rate": 0.0019777777777777775,
+      "loss": 6.711,
+      "step": 549
+    },
+    {
+      "epoch": 1.2218828103304638,
+      "grad_norm": 0.02825808897614479,
+      "learning_rate": 0.0019753086419753087,
+      "loss": 6.4686,
+      "step": 550
+    },
+    {
+      "epoch": 1.2241044154401555,
+      "grad_norm": 0.045339666306972504,
+      "learning_rate": 0.00197283950617284,
+      "loss": 6.4988,
+      "step": 551
+    },
+    {
+      "epoch": 1.2263260205498472,
+      "grad_norm": 0.034293051809072495,
+      "learning_rate": 0.00197037037037037,
+      "loss": 6.3252,
+      "step": 552
+    },
+    {
+      "epoch": 1.2285476256595391,
+      "grad_norm": 0.030633745715022087,
+      "learning_rate": 0.0019679012345679013,
+      "loss": 6.7252,
+      "step": 553
+    },
+    {
+      "epoch": 1.2307692307692308,
+      "grad_norm": 0.039270538836717606,
+      "learning_rate": 0.001965432098765432,
+      "loss": 6.9372,
+      "step": 554
+    },
+    {
+      "epoch": 1.2329908358789226,
+      "grad_norm": 0.030756397172808647,
+      "learning_rate": 0.0019629629629629632,
+      "loss": 6.5979,
+      "step": 555
+    },
+    {
+      "epoch": 1.2352124409886143,
+      "grad_norm": 0.033471401780843735,
+      "learning_rate": 0.0019604938271604935,
+      "loss": 6.9454,
+      "step": 556
+    },
+    {
+      "epoch": 1.237434046098306,
+      "grad_norm": 0.026848113164305687,
+      "learning_rate": 0.0019580246913580247,
+      "loss": 6.7686,
+      "step": 557
+    },
+    {
+      "epoch": 1.2396556512079977,
+      "grad_norm": 0.034598518162965775,
+      "learning_rate": 0.001955555555555556,
+      "loss": 6.8229,
+      "step": 558
+    },
+    {
+      "epoch": 1.2418772563176894,
+      "grad_norm": 0.029663242399692535,
+      "learning_rate": 0.0019530864197530864,
+      "loss": 6.5045,
+      "step": 559
+    },
+    {
+      "epoch": 1.2440988614273814,
+      "grad_norm": 0.031031301245093346,
+      "learning_rate": 0.0019506172839506173,
+      "loss": 6.8129,
+      "step": 560
+    },
+    {
+      "epoch": 1.246320466537073,
+      "grad_norm": 0.03552239388227463,
+      "learning_rate": 0.001948148148148148,
+      "loss": 6.8661,
+      "step": 561
+    },
+    {
+      "epoch": 1.2485420716467648,
+      "grad_norm": 0.03027603216469288,
+      "learning_rate": 0.001945679012345679,
+      "loss": 6.6394,
+      "step": 562
+    },
+    {
+      "epoch": 1.2507636767564565,
+      "grad_norm": 0.05250061675906181,
+      "learning_rate": 0.0019432098765432098,
+      "loss": 6.4261,
+      "step": 563
+    },
+    {
+      "epoch": 1.2529852818661482,
+      "grad_norm": 0.04094979166984558,
+      "learning_rate": 0.0019407407407407407,
+      "loss": 6.6154,
+      "step": 564
+    },
+    {
+      "epoch": 1.2552068869758402,
+      "grad_norm": 0.03538697585463524,
+      "learning_rate": 0.0019382716049382714,
+      "loss": 6.7878,
+      "step": 565
+    },
+    {
+      "epoch": 1.2574284920855319,
+      "grad_norm": 0.030879583209753036,
+      "learning_rate": 0.0019358024691358024,
+      "loss": 6.6464,
+      "step": 566
+    },
+    {
+      "epoch": 1.2596500971952236,
+      "grad_norm": 0.04202036187052727,
+      "learning_rate": 0.0019333333333333336,
+      "loss": 6.6937,
+      "step": 567
+    },
+    {
+      "epoch": 1.2618717023049153,
+      "grad_norm": 0.037544719874858856,
+      "learning_rate": 0.001930864197530864,
+      "loss": 6.6652,
+      "step": 568
+    },
+    {
+      "epoch": 1.264093307414607,
+      "grad_norm": 0.02640129253268242,
+      "learning_rate": 0.0019283950617283952,
+      "loss": 6.5707,
+      "step": 569
+    },
+    {
+      "epoch": 1.2663149125242987,
+      "grad_norm": 0.030197612941265106,
+      "learning_rate": 0.0019259259259259258,
+      "loss": 6.3537,
+      "step": 570
+    },
+    {
+      "epoch": 1.2685365176339904,
+      "grad_norm": 0.03666810691356659,
+      "learning_rate": 0.001923456790123457,
+      "loss": 6.9591,
+      "step": 571
+    },
+    {
+      "epoch": 1.2707581227436824,
+      "grad_norm": 0.026158703491091728,
+      "learning_rate": 0.0019209876543209875,
+      "loss": 6.5475,
+      "step": 572
+    },
+    {
+      "epoch": 1.272979727853374,
+      "grad_norm": 0.030652476474642754,
+      "learning_rate": 0.0019185185185185186,
+      "loss": 6.647,
+      "step": 573
+    },
+    {
+      "epoch": 1.2752013329630658,
+      "grad_norm": 0.029845140874385834,
+      "learning_rate": 0.0019160493827160496,
+      "loss": 6.6944,
+      "step": 574
+    },
+    {
+      "epoch": 1.2774229380727575,
+      "grad_norm": 0.04585261270403862,
+      "learning_rate": 0.0019135802469135803,
+      "loss": 6.9121,
+      "step": 575
+    },
+    {
+      "epoch": 1.2796445431824492,
+      "grad_norm": 0.02669621631503105,
+      "learning_rate": 0.0019111111111111113,
+      "loss": 6.6599,
+      "step": 576
+    },
+    {
+      "epoch": 1.2818661482921412,
+      "grad_norm": 0.0333004929125309,
+      "learning_rate": 0.001908641975308642,
+      "loss": 6.6859,
+      "step": 577
+    },
+    {
+      "epoch": 1.284087753401833,
+      "grad_norm": 0.0356912836432457,
+      "learning_rate": 0.001906172839506173,
+      "loss": 6.4453,
+      "step": 578
+    },
+    {
+      "epoch": 1.2863093585115246,
+      "grad_norm": 0.029126396402716637,
+      "learning_rate": 0.0019037037037037037,
+      "loss": 6.3332,
+      "step": 579
+    },
+    {
+      "epoch": 1.2885309636212163,
+      "grad_norm": 0.030171919614076614,
+      "learning_rate": 0.0019012345679012346,
+      "loss": 6.5991,
+      "step": 580
+    },
+    {
+      "epoch": 1.290752568730908,
+      "grad_norm": 0.03224395960569382,
+      "learning_rate": 0.0018987654320987656,
+      "loss": 6.5085,
+      "step": 581
+    },
+    {
+      "epoch": 1.2929741738405998,
+      "grad_norm": 0.027340635657310486,
+      "learning_rate": 0.0018962962962962963,
+      "loss": 6.4404,
+      "step": 582
+    },
+    {
+      "epoch": 1.2951957789502915,
+      "grad_norm": 0.03136056661605835,
+      "learning_rate": 0.0018938271604938273,
+      "loss": 6.3854,
+      "step": 583
+    },
+    {
+      "epoch": 1.2974173840599834,
+      "grad_norm": 0.03169075772166252,
+      "learning_rate": 0.001891358024691358,
+      "loss": 6.6688,
+      "step": 584
+    },
+    {
+      "epoch": 1.2996389891696751,
+      "grad_norm": 0.03269726783037186,
+      "learning_rate": 0.001888888888888889,
+      "loss": 6.7879,
+      "step": 585
+    },
+    {
+      "epoch": 1.3018605942793668,
+      "grad_norm": 0.03851224109530449,
+      "learning_rate": 0.0018864197530864197,
+      "loss": 7.0064,
+      "step": 586
+    },
+    {
+      "epoch": 1.3040821993890586,
+      "grad_norm": 0.034009456634521484,
+      "learning_rate": 0.0018839506172839506,
+      "loss": 6.6696,
+      "step": 587
+    },
+    {
+      "epoch": 1.3063038044987503,
+      "grad_norm": 0.03421778231859207,
+      "learning_rate": 0.0018814814814814816,
+      "loss": 6.7469,
+      "step": 588
+    },
+    {
+      "epoch": 1.3085254096084422,
+      "grad_norm": 0.0405525378882885,
+      "learning_rate": 0.0018790123456790123,
+      "loss": 6.3569,
+      "step": 589
+    },
+    {
+      "epoch": 1.310747014718134,
+      "grad_norm": 0.04027966782450676,
+      "learning_rate": 0.0018765432098765433,
+      "loss": 6.66,
+      "step": 590
+    },
+    {
+      "epoch": 1.3129686198278256,
+      "grad_norm": 0.028766606003046036,
+      "learning_rate": 0.001874074074074074,
+      "loss": 6.5273,
+      "step": 591
+    },
+    {
+      "epoch": 1.3151902249375174,
+      "grad_norm": 0.03105996735394001,
+      "learning_rate": 0.001871604938271605,
+      "loss": 6.8601,
+      "step": 592
+    },
+    {
+      "epoch": 1.317411830047209,
+      "grad_norm": 0.028445599600672722,
+      "learning_rate": 0.0018691358024691357,
+      "loss": 6.6357,
+      "step": 593
+    },
+    {
+      "epoch": 1.3196334351569008,
+      "grad_norm": 0.042267780750989914,
+      "learning_rate": 0.0018666666666666666,
+      "loss": 6.6521,
+      "step": 594
+    },
+    {
+      "epoch": 1.3218550402665925,
+      "grad_norm": 0.029775841161608696,
+      "learning_rate": 0.0018641975308641974,
+      "loss": 6.5442,
+      "step": 595
+    },
+    {
+      "epoch": 1.3240766453762844,
+      "grad_norm": 0.033110231161117554,
+      "learning_rate": 0.0018617283950617283,
+      "loss": 6.5038,
+      "step": 596
+    },
+    {
+      "epoch": 1.3262982504859762,
+      "grad_norm": 0.026784425601363182,
+      "learning_rate": 0.0018592592592592595,
+      "loss": 6.6861,
+      "step": 597
+    },
+    {
+      "epoch": 1.3285198555956679,
+      "grad_norm": 0.030177338048815727,
+      "learning_rate": 0.00185679012345679,
+      "loss": 6.9846,
+      "step": 598
+    },
+    {
+      "epoch": 1.3307414607053596,
+      "grad_norm": 0.0336572602391243,
+      "learning_rate": 0.0018543209876543212,
+      "loss": 6.7044,
+      "step": 599
+    },
+    {
+      "epoch": 1.3329630658150513,
+      "grad_norm": 0.030566373839974403,
+      "learning_rate": 0.0018518518518518517,
+      "loss": 6.628,
+      "step": 600
+    },
+    {
+      "epoch": 1.3351846709247432,
+      "grad_norm": 0.03165126219391823,
+      "learning_rate": 0.0018493827160493829,
+      "loss": 6.7926,
+      "step": 601
+    },
+    {
+      "epoch": 1.337406276034435,
+      "grad_norm": 0.03844063729047775,
+      "learning_rate": 0.0018469135802469134,
+      "loss": 7.1151,
+      "step": 602
+    },
+    {
+      "epoch": 1.3396278811441267,
+      "grad_norm": 0.026797592639923096,
+      "learning_rate": 0.0018444444444444446,
+      "loss": 6.739,
+      "step": 603
+    },
+    {
+      "epoch": 1.3418494862538184,
+      "grad_norm": 0.04878336936235428,
+      "learning_rate": 0.0018419753086419755,
+      "loss": 6.7186,
+      "step": 604
+    },
+    {
+      "epoch": 1.34407109136351,
+      "grad_norm": 0.03213360160589218,
+      "learning_rate": 0.0018395061728395062,
+      "loss": 6.7677,
+      "step": 605
+    },
+    {
+      "epoch": 1.3462926964732018,
+      "grad_norm": 0.03563079237937927,
+      "learning_rate": 0.0018370370370370372,
+      "loss": 6.8938,
+      "step": 606
+    },
+    {
+      "epoch": 1.3485143015828935,
+      "grad_norm": 0.0332307294011116,
+      "learning_rate": 0.001834567901234568,
+      "loss": 6.4853,
+      "step": 607
+    },
+    {
+      "epoch": 1.3507359066925855,
+      "grad_norm": 0.03405038267374039,
+      "learning_rate": 0.0018320987654320989,
+      "loss": 6.7939,
+      "step": 608
+    },
+    {
+      "epoch": 1.3529575118022772,
+      "grad_norm": 0.031904082745313644,
+      "learning_rate": 0.0018296296296296296,
+      "loss": 6.9467,
+      "step": 609
+    },
+    {
+      "epoch": 1.355179116911969,
+      "grad_norm": 0.03952391445636749,
+      "learning_rate": 0.0018271604938271606,
+      "loss": 6.6978,
+      "step": 610
+    },
+    {
+      "epoch": 1.3574007220216606,
+      "grad_norm": 0.04087958112359047,
+      "learning_rate": 0.0018246913580246915,
+      "loss": 6.4224,
+      "step": 611
+    },
+    {
+      "epoch": 1.3596223271313523,
+      "grad_norm": 0.03634248301386833,
+      "learning_rate": 0.0018222222222222223,
+      "loss": 6.6674,
+      "step": 612
+    },
+    {
+      "epoch": 1.3618439322410443,
+      "grad_norm": 0.03480825945734978,
+      "learning_rate": 0.0018197530864197532,
+      "loss": 6.8653,
+      "step": 613
+    },
+    {
+      "epoch": 1.364065537350736,
+      "grad_norm": 0.02999613992869854,
+      "learning_rate": 0.001817283950617284,
+      "loss": 6.4436,
+      "step": 614
+    },
+    {
+      "epoch": 1.3662871424604277,
+      "grad_norm": 0.03151191398501396,
+      "learning_rate": 0.001814814814814815,
+      "loss": 6.5541,
+      "step": 615
+    },
+    {
+      "epoch": 1.3685087475701194,
+      "grad_norm": 0.03740501031279564,
+      "learning_rate": 0.0018123456790123456,
+      "loss": 6.7676,
+      "step": 616
+    },
+    {
+      "epoch": 1.3707303526798111,
+      "grad_norm": 0.0368155799806118,
+      "learning_rate": 0.0018098765432098766,
+      "loss": 6.5592,
+      "step": 617
+    },
+    {
+      "epoch": 1.3729519577895029,
+      "grad_norm": 0.04287653788924217,
+      "learning_rate": 0.0018074074074074075,
+      "loss": 6.3789,
+      "step": 618
+    },
+    {
+      "epoch": 1.3751735628991946,
+      "grad_norm": 0.052219755947589874,
+      "learning_rate": 0.0018049382716049383,
+      "loss": 6.2714,
+      "step": 619
+    },
+    {
+      "epoch": 1.3773951680088863,
+      "grad_norm": 0.0451362282037735,
+      "learning_rate": 0.0018024691358024692,
+      "loss": 6.5485,
+      "step": 620
+    },
+    {
+      "epoch": 1.3796167731185782,
+      "grad_norm": 0.03573337197303772,
+      "learning_rate": 0.0018,
+      "loss": 6.7343,
+      "step": 621
+    },
+    {
+      "epoch": 1.38183837822827,
+      "grad_norm": 0.03500332310795784,
+      "learning_rate": 0.001797530864197531,
+      "loss": 6.5455,
+      "step": 622
+    },
+    {
+      "epoch": 1.3840599833379617,
+      "grad_norm": 0.0386490635573864,
+      "learning_rate": 0.0017950617283950616,
+      "loss": 6.5688,
+      "step": 623
+    },
+    {
+      "epoch": 1.3862815884476534,
+      "grad_norm": 0.03757933899760246,
+      "learning_rate": 0.0017925925925925926,
+      "loss": 6.8372,
+      "step": 624
+    },
+    {
+      "epoch": 1.3885031935573453,
+      "grad_norm": 0.030975468456745148,
+      "learning_rate": 0.0017901234567901233,
+      "loss": 6.6744,
+      "step": 625
+    },
+    {
+      "epoch": 1.390724798667037,
+      "grad_norm": 0.03750336542725563,
+      "learning_rate": 0.0017876543209876543,
+      "loss": 6.529,
+      "step": 626
+    },
+    {
+      "epoch": 1.3929464037767287,
+      "grad_norm": 0.04257288947701454,
+      "learning_rate": 0.0017851851851851854,
+      "loss": 6.4571,
+      "step": 627
+    },
+    {
+      "epoch": 1.3951680088864205,
+      "grad_norm": 0.03363563492894173,
+      "learning_rate": 0.001782716049382716,
+      "loss": 6.5041,
+      "step": 628
+    },
+    {
+      "epoch": 1.3973896139961122,
+      "grad_norm": 0.03184138238430023,
+      "learning_rate": 0.0017802469135802471,
+      "loss": 6.4177,
+      "step": 629
+    },
+    {
+      "epoch": 1.3996112191058039,
+      "grad_norm": 0.030448872596025467,
+      "learning_rate": 0.0017777777777777776,
+      "loss": 6.7623,
+      "step": 630
+    },
+    {
+      "epoch": 1.4018328242154956,
+      "grad_norm": 0.03545617684721947,
+      "learning_rate": 0.0017753086419753088,
+      "loss": 6.7921,
+      "step": 631
+    },
+    {
+      "epoch": 1.4040544293251873,
+      "grad_norm": 0.04112512245774269,
+      "learning_rate": 0.0017728395061728393,
+      "loss": 6.7947,
+      "step": 632
+    },
+    {
+      "epoch": 1.4062760344348793,
+      "grad_norm": 0.039878182113170624,
+      "learning_rate": 0.0017703703703703705,
+      "loss": 6.6468,
+      "step": 633
+    },
+    {
+      "epoch": 1.408497639544571,
+      "grad_norm": 0.045425985008478165,
+      "learning_rate": 0.0017679012345679015,
+      "loss": 6.5777,
+      "step": 634
+    },
+    {
+      "epoch": 1.4107192446542627,
+      "grad_norm": 0.039729323238134384,
+      "learning_rate": 0.0017654320987654322,
+      "loss": 6.5231,
+      "step": 635
+    },
+    {
+      "epoch": 1.4129408497639544,
+      "grad_norm": 0.03206026554107666,
+      "learning_rate": 0.0017629629629629631,
+      "loss": 6.5691,
+      "step": 636
+    },
+    {
+      "epoch": 1.4151624548736463,
+      "grad_norm": 0.02712268941104412,
+      "learning_rate": 0.0017604938271604939,
+      "loss": 6.6543,
+      "step": 637
+    },
+    {
+      "epoch": 1.417384059983338,
+      "grad_norm": 0.04199135676026344,
+      "learning_rate": 0.0017580246913580248,
+      "loss": 6.7355,
+      "step": 638
+    },
+    {
+      "epoch": 1.4196056650930298,
+      "grad_norm": 0.042644452303647995,
+      "learning_rate": 0.0017555555555555556,
+      "loss": 6.4943,
+      "step": 639
+    },
+    {
+      "epoch": 1.4218272702027215,
+      "grad_norm": 0.03070429153740406,
+      "learning_rate": 0.0017530864197530865,
+      "loss": 6.4935,
+      "step": 640
+    },
+    {
+      "epoch": 1.4240488753124132,
+      "grad_norm": 0.04355219006538391,
+      "learning_rate": 0.0017506172839506175,
+      "loss": 6.7335,
+      "step": 641
+    },
+    {
+      "epoch": 1.426270480422105,
+      "grad_norm": 0.043914977461099625,
+      "learning_rate": 0.0017481481481481482,
+      "loss": 6.6347,
+      "step": 642
+    },
+    {
+      "epoch": 1.4284920855317966,
+      "grad_norm": 0.03733412176370621,
+      "learning_rate": 0.0017456790123456791,
+      "loss": 6.4863,
+      "step": 643
+    },
+    {
+      "epoch": 1.4307136906414883,
+      "grad_norm": 0.04044484347105026,
+      "learning_rate": 0.0017432098765432099,
+      "loss": 6.205,
+      "step": 644
+    },
+    {
+      "epoch": 1.4329352957511803,
+      "grad_norm": 0.04359281435608864,
+      "learning_rate": 0.0017407407407407408,
+      "loss": 6.4957,
+      "step": 645
+    },
+    {
+      "epoch": 1.435156900860872,
+      "grad_norm": 0.04081505164504051,
+      "learning_rate": 0.0017382716049382716,
+      "loss": 6.6605,
+      "step": 646
+    },
+    {
+      "epoch": 1.4373785059705637,
+      "grad_norm": 0.052760712802410126,
+      "learning_rate": 0.0017358024691358025,
+      "loss": 6.4727,
+      "step": 647
+    },
+    {
+      "epoch": 1.4396001110802554,
+      "grad_norm": 0.03821408748626709,
+      "learning_rate": 0.0017333333333333333,
+      "loss": 6.3824,
+      "step": 648
+    },
+    {
+      "epoch": 1.4418217161899474,
+      "grad_norm": 0.03151766210794449,
+      "learning_rate": 0.0017308641975308642,
+      "loss": 6.5262,
+      "step": 649
+    },
+    {
+      "epoch": 1.444043321299639,
+      "grad_norm": 0.030120082199573517,
+      "learning_rate": 0.0017283950617283952,
+      "loss": 6.5886,
+      "step": 650
+    },
+    {
+      "epoch": 1.4462649264093308,
+      "grad_norm": 0.04470925033092499,
+      "learning_rate": 0.0017259259259259259,
+      "loss": 6.3653,
+      "step": 651
+    },
+    {
+      "epoch": 1.4484865315190225,
+      "grad_norm": 0.033646512776613235,
+      "learning_rate": 0.0017234567901234568,
+      "loss": 6.4888,
+      "step": 652
+    },
+    {
+      "epoch": 1.4507081366287142,
+      "grad_norm": 0.04000632092356682,
+      "learning_rate": 0.0017209876543209876,
+      "loss": 6.6951,
+      "step": 653
+    },
+    {
+      "epoch": 1.452929741738406,
+      "grad_norm": 0.044797301292419434,
+      "learning_rate": 0.0017185185185185185,
+      "loss": 6.8074,
+      "step": 654
+    },
+    {
+      "epoch": 1.4551513468480977,
+      "grad_norm": 0.03085845522582531,
+      "learning_rate": 0.0017160493827160493,
+      "loss": 6.5297,
+      "step": 655
+    },
+    {
+      "epoch": 1.4573729519577894,
+      "grad_norm": 0.04692323878407478,
+      "learning_rate": 0.0017135802469135802,
+      "loss": 6.4301,
+      "step": 656
+    },
+    {
+      "epoch": 1.4595945570674813,
+      "grad_norm": 0.04330454021692276,
+      "learning_rate": 0.0017111111111111114,
+      "loss": 6.294,
+      "step": 657
+    },
+    {
+      "epoch": 1.461816162177173,
+      "grad_norm": 0.03762800619006157,
+      "learning_rate": 0.001708641975308642,
+      "loss": 6.6351,
+      "step": 658
+    },
+    {
+      "epoch": 1.4640377672868647,
+      "grad_norm": 0.041210997849702835,
+      "learning_rate": 0.001706172839506173,
+      "loss": 6.0957,
+      "step": 659
+    },
+    {
+      "epoch": 1.4662593723965565,
+      "grad_norm": 0.0367109514772892,
+      "learning_rate": 0.0017037037037037036,
+      "loss": 6.6901,
+      "step": 660
+    },
+    {
+      "epoch": 1.4684809775062484,
+      "grad_norm": 0.03937892988324165,
+      "learning_rate": 0.0017012345679012348,
+      "loss": 6.6719,
+      "step": 661
+    },
+    {
+      "epoch": 1.4707025826159401,
+      "grad_norm": 0.04125594347715378,
+      "learning_rate": 0.0016987654320987653,
+      "loss": 6.5182,
+      "step": 662
+    },
+    {
+      "epoch": 1.4729241877256318,
+      "grad_norm": 0.038301244378089905,
+      "learning_rate": 0.0016962962962962964,
+      "loss": 6.7307,
+      "step": 663
+    },
+    {
+      "epoch": 1.4751457928353235,
+      "grad_norm": 0.032814789563417435,
+      "learning_rate": 0.0016938271604938274,
+      "loss": 6.7457,
+      "step": 664
+    },
+    {
+      "epoch": 1.4773673979450153,
+      "grad_norm": 0.032927006483078,
+      "learning_rate": 0.0016913580246913581,
+      "loss": 6.2871,
+      "step": 665
+    },
+    {
+      "epoch": 1.479589003054707,
+      "grad_norm": 0.05104907602071762,
+      "learning_rate": 0.001688888888888889,
+      "loss": 6.4206,
+      "step": 666
+    },
+    {
+      "epoch": 1.4818106081643987,
+      "grad_norm": 0.03182735666632652,
+      "learning_rate": 0.0016864197530864198,
+      "loss": 6.3451,
+      "step": 667
+    },
+    {
+      "epoch": 1.4840322132740904,
+      "grad_norm": 0.031046222895383835,
+      "learning_rate": 0.0016839506172839508,
+      "loss": 6.5484,
+      "step": 668
+    },
+    {
+      "epoch": 1.4862538183837823,
+      "grad_norm": 0.03595875948667526,
+      "learning_rate": 0.0016814814814814813,
+      "loss": 6.7068,
+      "step": 669
+    },
+    {
+      "epoch": 1.488475423493474,
+      "grad_norm": 0.03241979330778122,
+      "learning_rate": 0.0016790123456790125,
+      "loss": 6.7495,
+      "step": 670
+    },
+    {
+      "epoch": 1.4906970286031658,
+      "grad_norm": 0.04917046055197716,
+      "learning_rate": 0.0016765432098765434,
+      "loss": 6.8188,
+      "step": 671
+    },
+    {
+      "epoch": 1.4929186337128575,
+      "grad_norm": 0.041194915771484375,
+      "learning_rate": 0.0016740740740740741,
+      "loss": 6.6899,
+      "step": 672
+    },
+    {
+      "epoch": 1.4951402388225494,
+      "grad_norm": 0.040103353559970856,
+      "learning_rate": 0.001671604938271605,
+      "loss": 6.5504,
+      "step": 673
+    },
+    {
+      "epoch": 1.4973618439322411,
+      "grad_norm": 0.04008682444691658,
+      "learning_rate": 0.0016691358024691358,
+      "loss": 6.4534,
+      "step": 674
+    },
+    {
+      "epoch": 1.4995834490419329,
+      "grad_norm": 0.02651919610798359,
+      "learning_rate": 0.0016666666666666668,
+      "loss": 6.7499,
+      "step": 675
+    },
+    {
+      "epoch": 1.5018050541516246,
+      "grad_norm": 0.03353941813111305,
+      "learning_rate": 0.0016641975308641975,
+      "loss": 6.5624,
+      "step": 676
+    },
+    {
+      "epoch": 1.5040266592613163,
+      "grad_norm": 0.030263148248195648,
+      "learning_rate": 0.0016617283950617285,
+      "loss": 6.8292,
+      "step": 677
+    },
+    {
+      "epoch": 1.506248264371008,
+      "grad_norm": 0.0390033982694149,
+      "learning_rate": 0.0016592592592592592,
+      "loss": 6.4628,
+      "step": 678
+    },
+    {
+      "epoch": 1.5084698694806997,
+      "grad_norm": 0.040038738399744034,
+      "learning_rate": 0.0016567901234567901,
+      "loss": 6.6102,
+      "step": 679
+    },
+    {
+      "epoch": 1.5106914745903914,
+      "grad_norm": 0.04442286118865013,
+      "learning_rate": 0.001654320987654321,
+      "loss": 6.3741,
+      "step": 680
+    },
+    {
+      "epoch": 1.5129130797000832,
+      "grad_norm": 0.05280248448252678,
+      "learning_rate": 0.0016518518518518518,
+      "loss": 6.1835,
+      "step": 681
+    },
+    {
+      "epoch": 1.515134684809775,
+      "grad_norm": 0.03858866170048714,
+      "learning_rate": 0.0016493827160493828,
+      "loss": 6.5735,
+      "step": 682
+    },
+    {
+      "epoch": 1.5173562899194668,
+      "grad_norm": 0.04009943827986717,
+      "learning_rate": 0.0016469135802469135,
+      "loss": 6.3924,
+      "step": 683
+    },
+    {
+      "epoch": 1.5195778950291585,
+      "grad_norm": 0.05272547900676727,
+      "learning_rate": 0.0016444444444444445,
+      "loss": 6.4248,
+      "step": 684
+    },
+    {
+      "epoch": 1.5217995001388505,
+      "grad_norm": 0.03855543211102486,
+      "learning_rate": 0.0016419753086419752,
+      "loss": 6.4265,
+      "step": 685
+    },
+    {
+      "epoch": 1.5240211052485422,
+      "grad_norm": 0.0413784384727478,
+      "learning_rate": 0.0016395061728395062,
+      "loss": 6.6763,
+      "step": 686
+    },
+    {
+      "epoch": 1.526242710358234,
+      "grad_norm": 0.034833308309316635,
+      "learning_rate": 0.001637037037037037,
+      "loss": 6.5616,
+      "step": 687
+    },
+    {
+      "epoch": 1.5284643154679256,
+      "grad_norm": 0.03764263913035393,
+      "learning_rate": 0.0016345679012345678,
+      "loss": 6.8207,
+      "step": 688
+    },
+    {
+      "epoch": 1.5306859205776173,
+      "grad_norm": 0.0339258648455143,
+      "learning_rate": 0.0016320987654320988,
+      "loss": 6.4341,
+      "step": 689
+    },
+    {
+      "epoch": 1.532907525687309,
+      "grad_norm": 0.04417251795530319,
+      "learning_rate": 0.0016296296296296295,
+      "loss": 6.5555,
+      "step": 690
+    },
+    {
+      "epoch": 1.5351291307970008,
+      "grad_norm": 0.0337689109146595,
+      "learning_rate": 0.0016271604938271605,
+      "loss": 6.3568,
+      "step": 691
+    },
+    {
+      "epoch": 1.5373507359066925,
+      "grad_norm": 0.039920464158058167,
+      "learning_rate": 0.0016246913580246912,
+      "loss": 6.4116,
+      "step": 692
+    },
+    {
+      "epoch": 1.5395723410163842,
+      "grad_norm": 0.042915038764476776,
+      "learning_rate": 0.0016222222222222222,
+      "loss": 6.4515,
+      "step": 693
+    },
+    {
+      "epoch": 1.5417939461260761,
+      "grad_norm": 0.03201558440923691,
+      "learning_rate": 0.0016197530864197533,
+      "loss": 6.6761,
+      "step": 694
+    },
+    {
+      "epoch": 1.5440155512357678,
+      "grad_norm": 0.03667712211608887,
+      "learning_rate": 0.0016172839506172839,
+      "loss": 6.4992,
+      "step": 695
+    },
+    {
+      "epoch": 1.5462371563454596,
+      "grad_norm": 0.042801886796951294,
+      "learning_rate": 0.001614814814814815,
+      "loss": 6.2258,
+      "step": 696
+    },
+    {
+      "epoch": 1.5484587614551515,
+      "grad_norm": 0.0630866214632988,
+      "learning_rate": 0.0016123456790123455,
+      "loss": 6.8541,
+      "step": 697
+    },
+    {
+      "epoch": 1.5506803665648432,
+      "grad_norm": 0.03677830100059509,
+      "learning_rate": 0.0016098765432098767,
+      "loss": 6.3493,
+      "step": 698
+    },
+    {
+      "epoch": 1.552901971674535,
+      "grad_norm": 0.060171615332365036,
+      "learning_rate": 0.0016074074074074072,
+      "loss": 6.1666,
+      "step": 699
+    },
+    {
+      "epoch": 1.5551235767842266,
+      "grad_norm": 0.035057131201028824,
+      "learning_rate": 0.0016049382716049384,
+      "loss": 6.2471,
+      "step": 700
+    },
+    {
+      "epoch": 1.5573451818939184,
+      "grad_norm": 0.03420359641313553,
+      "learning_rate": 0.0016024691358024693,
+      "loss": 6.7564,
+      "step": 701
+    },
+    {
+      "epoch": 1.55956678700361,
+      "grad_norm": 0.03487620875239372,
+      "learning_rate": 0.0016,
+      "loss": 6.2373,
+      "step": 702
+    },
+    {
+      "epoch": 1.5617883921133018,
+      "grad_norm": 0.03929250314831734,
+      "learning_rate": 0.001597530864197531,
+      "loss": 6.2193,
+      "step": 703
+    },
+    {
+      "epoch": 1.5640099972229935,
+      "grad_norm": 0.0515265055000782,
+      "learning_rate": 0.0015950617283950618,
+      "loss": 6.533,
+      "step": 704
+    },
+    {
+      "epoch": 1.5662316023326852,
+      "grad_norm": 0.0337432399392128,
+      "learning_rate": 0.0015925925925925927,
+      "loss": 6.5107,
+      "step": 705
+    },
+    {
+      "epoch": 1.5684532074423772,
+      "grad_norm": 0.06337834894657135,
+      "learning_rate": 0.0015901234567901234,
+      "loss": 6.6714,
+      "step": 706
+    },
+    {
+      "epoch": 1.5706748125520689,
+      "grad_norm": 0.03698400408029556,
+      "learning_rate": 0.0015876543209876544,
+      "loss": 6.6214,
+      "step": 707
+    },
+    {
+      "epoch": 1.5728964176617606,
+      "grad_norm": 0.045376867055892944,
+      "learning_rate": 0.0015851851851851851,
+      "loss": 6.403,
+      "step": 708
+    },
+    {
+      "epoch": 1.5751180227714525,
+      "grad_norm": 0.04952572286128998,
+      "learning_rate": 0.001582716049382716,
+      "loss": 6.156,
+      "step": 709
+    },
+    {
+      "epoch": 1.5773396278811442,
+      "grad_norm": 0.04453912749886513,
+      "learning_rate": 0.001580246913580247,
+      "loss": 6.7879,
+      "step": 710
+    },
+    {
+      "epoch": 1.579561232990836,
+      "grad_norm": 0.04479203373193741,
+      "learning_rate": 0.0015777777777777778,
+      "loss": 6.4831,
+      "step": 711
+    },
+    {
+      "epoch": 1.5817828381005277,
+      "grad_norm": 0.03946154564619064,
+      "learning_rate": 0.0015753086419753087,
+      "loss": 6.4438,
+      "step": 712
+    },
+    {
+      "epoch": 1.5840044432102194,
+      "grad_norm": 0.029172010719776154,
+      "learning_rate": 0.0015728395061728395,
+      "loss": 6.5927,
+      "step": 713
+    },
+    {
+      "epoch": 1.586226048319911,
+      "grad_norm": 0.03784365952014923,
+      "learning_rate": 0.0015703703703703704,
+      "loss": 6.6141,
+      "step": 714
+    },
+    {
+      "epoch": 1.5884476534296028,
+      "grad_norm": 0.05160333588719368,
+      "learning_rate": 0.0015679012345679011,
+      "loss": 6.2963,
+      "step": 715
+    },
+    {
+      "epoch": 1.5906692585392945,
+      "grad_norm": 0.0242854543030262,
+      "learning_rate": 0.001565432098765432,
+      "loss": 6.5513,
+      "step": 716
+    },
+    {
+      "epoch": 1.5928908636489862,
+      "grad_norm": 0.03806491941213608,
+      "learning_rate": 0.001562962962962963,
+      "loss": 6.5976,
+      "step": 717
+    },
+    {
+      "epoch": 1.5951124687586782,
+      "grad_norm": 0.037991028279066086,
+      "learning_rate": 0.0015604938271604938,
+      "loss": 6.7611,
+      "step": 718
+    },
+    {
+      "epoch": 1.59733407386837,
+      "grad_norm": 0.04018974304199219,
+      "learning_rate": 0.0015580246913580247,
+      "loss": 6.587,
+      "step": 719
+    },
+    {
+      "epoch": 1.5995556789780616,
+      "grad_norm": 0.034009311348199844,
+      "learning_rate": 0.0015555555555555555,
+      "loss": 6.6634,
+      "step": 720
+    },
+    {
+      "epoch": 1.6017772840877536,
+      "grad_norm": 0.05159347131848335,
+      "learning_rate": 0.0015530864197530864,
+      "loss": 6.187,
+      "step": 721
+    },
+    {
+      "epoch": 1.6039988891974453,
+      "grad_norm": 0.03494350239634514,
+      "learning_rate": 0.0015506172839506172,
+      "loss": 6.5968,
+      "step": 722
+    },
+    {
+      "epoch": 1.606220494307137,
+      "grad_norm": 0.03169875964522362,
+      "learning_rate": 0.001548148148148148,
+      "loss": 6.3776,
+      "step": 723
+    },
+    {
+      "epoch": 1.6084420994168287,
+      "grad_norm": 0.04075143113732338,
+      "learning_rate": 0.0015456790123456793,
+      "loss": 6.7278,
+      "step": 724
+    },
+    {
+      "epoch": 1.6106637045265204,
+      "grad_norm": 0.05059502273797989,
+      "learning_rate": 0.0015432098765432098,
+      "loss": 6.5432,
+      "step": 725
+    },
+    {
+      "epoch": 1.6128853096362121,
+      "grad_norm": 0.03574662283062935,
+      "learning_rate": 0.001540740740740741,
+      "loss": 6.6231,
+      "step": 726
+    },
+    {
+      "epoch": 1.6151069147459038,
+      "grad_norm": 0.03380226716399193,
+      "learning_rate": 0.0015382716049382715,
+      "loss": 6.7766,
+      "step": 727
+    },
+    {
+      "epoch": 1.6173285198555956,
+      "grad_norm": 0.032838184386491776,
+      "learning_rate": 0.0015358024691358026,
+      "loss": 6.3362,
+      "step": 728
+    },
+    {
+      "epoch": 1.6195501249652873,
+      "grad_norm": 0.03813943266868591,
+      "learning_rate": 0.0015333333333333332,
+      "loss": 6.6814,
+      "step": 729
+    },
+    {
+      "epoch": 1.6217717300749792,
+      "grad_norm": 0.0440312996506691,
+      "learning_rate": 0.0015308641975308643,
+      "loss": 6.5525,
+      "step": 730
+    },
+    {
+      "epoch": 1.623993335184671,
+      "grad_norm": 0.03285110741853714,
+      "learning_rate": 0.0015283950617283948,
+      "loss": 6.5697,
+      "step": 731
+    },
+    {
+      "epoch": 1.6262149402943626,
+      "grad_norm": 0.02949458360671997,
+      "learning_rate": 0.001525925925925926,
+      "loss": 6.6347,
+      "step": 732
+    },
+    {
+      "epoch": 1.6284365454040546,
+      "grad_norm": 0.042863499373197556,
+      "learning_rate": 0.001523456790123457,
+      "loss": 6.7917,
+      "step": 733
+    },
+    {
+      "epoch": 1.6306581505137463,
+      "grad_norm": 0.04673711955547333,
+      "learning_rate": 0.0015209876543209877,
+      "loss": 6.7026,
+      "step": 734
+    },
+    {
+      "epoch": 1.632879755623438,
+      "grad_norm": 0.03455612435936928,
+      "learning_rate": 0.0015185185185185187,
+      "loss": 6.2779,
+      "step": 735
+    },
+    {
+      "epoch": 1.6351013607331297,
+      "grad_norm": 0.04289144277572632,
+      "learning_rate": 0.0015160493827160494,
+      "loss": 6.3884,
+      "step": 736
+    },
+    {
+      "epoch": 1.6373229658428214,
+      "grad_norm": 0.03550789877772331,
+      "learning_rate": 0.0015135802469135803,
+      "loss": 6.523,
+      "step": 737
+    },
+    {
+      "epoch": 1.6395445709525132,
+      "grad_norm": 0.03848389908671379,
+      "learning_rate": 0.001511111111111111,
+      "loss": 6.7456,
+      "step": 738
+    },
+    {
+      "epoch": 1.6417661760622049,
+      "grad_norm": 0.041664306074380875,
+      "learning_rate": 0.001508641975308642,
+      "loss": 6.542,
+      "step": 739
+    },
+    {
+      "epoch": 1.6439877811718966,
+      "grad_norm": 0.03599809482693672,
+      "learning_rate": 0.001506172839506173,
+      "loss": 6.7232,
+      "step": 740
+    },
+    {
+      "epoch": 1.6462093862815883,
+      "grad_norm": 0.03411336615681648,
+      "learning_rate": 0.0015037037037037037,
+      "loss": 6.3224,
+      "step": 741
+    },
+    {
+      "epoch": 1.6484309913912802,
+      "grad_norm": 0.0341917909681797,
+      "learning_rate": 0.0015012345679012347,
+      "loss": 6.4651,
+      "step": 742
+    },
+    {
+      "epoch": 1.650652596500972,
+      "grad_norm": 0.035019487142562866,
+      "learning_rate": 0.0014987654320987656,
+      "loss": 6.4225,
+      "step": 743
+    },
+    {
+      "epoch": 1.6528742016106637,
+      "grad_norm": 0.04106474295258522,
+      "learning_rate": 0.0014962962962962963,
+      "loss": 6.5229,
+      "step": 744
+    },
+    {
+      "epoch": 1.6550958067203556,
+      "grad_norm": 0.048563841730356216,
+      "learning_rate": 0.0014938271604938273,
+      "loss": 6.6798,
+      "step": 745
+    },
+    {
+      "epoch": 1.6573174118300473,
+      "grad_norm": 0.03529489040374756,
+      "learning_rate": 0.001491358024691358,
+      "loss": 6.5415,
+      "step": 746
+    },
+    {
+      "epoch": 1.659539016939739,
+      "grad_norm": 0.03518693521618843,
+      "learning_rate": 0.001488888888888889,
+      "loss": 6.443,
+      "step": 747
+    },
+    {
+      "epoch": 1.6617606220494308,
+      "grad_norm": 0.032215364277362823,
+      "learning_rate": 0.0014864197530864197,
+      "loss": 6.5331,
+      "step": 748
+    },
+    {
+      "epoch": 1.6639822271591225,
+      "grad_norm": 0.04053748399019241,
+      "learning_rate": 0.0014839506172839507,
+      "loss": 6.5905,
+      "step": 749
+    },
+    {
+      "epoch": 1.6662038322688142,
+      "grad_norm": 0.05106385052204132,
+      "learning_rate": 0.0014814814814814814,
+      "loss": 6.2633,
+      "step": 750
+    },
+    {
+      "epoch": 1.668425437378506,
+      "grad_norm": 0.033272698521614075,
+      "learning_rate": 0.0014790123456790124,
+      "loss": 6.3453,
+      "step": 751
+    },
+    {
+      "epoch": 1.6706470424881976,
+      "grad_norm": 0.046030763536691666,
+      "learning_rate": 0.0014765432098765433,
+      "loss": 6.7877,
+      "step": 752
+    },
+    {
+      "epoch": 1.6728686475978893,
+      "grad_norm": 0.06251224130392075,
+      "learning_rate": 0.001474074074074074,
+      "loss": 6.3076,
+      "step": 753
+    },
+    {
+      "epoch": 1.6750902527075813,
+      "grad_norm": 0.0467204675078392,
+      "learning_rate": 0.001471604938271605,
+      "loss": 6.3643,
+      "step": 754
+    },
+    {
+      "epoch": 1.677311857817273,
+      "grad_norm": 0.03680356591939926,
+      "learning_rate": 0.0014691358024691357,
+      "loss": 6.3828,
+      "step": 755
+    },
+    {
+      "epoch": 1.6795334629269647,
+      "grad_norm": 0.03661566972732544,
+      "learning_rate": 0.0014666666666666667,
+      "loss": 6.2534,
+      "step": 756
+    },
+    {
+      "epoch": 1.6817550680366566,
+      "grad_norm": 0.042951829731464386,
+      "learning_rate": 0.0014641975308641974,
+      "loss": 6.6749,
+      "step": 757
+    },
+    {
+      "epoch": 1.6839766731463484,
+      "grad_norm": 0.04375871270895004,
+      "learning_rate": 0.0014617283950617286,
+      "loss": 6.3219,
+      "step": 758
+    },
+    {
+      "epoch": 1.68619827825604,
+      "grad_norm": 0.039236899465322495,
+      "learning_rate": 0.0014592592592592593,
+      "loss": 6.6573,
+      "step": 759
+    },
+    {
+      "epoch": 1.6884198833657318,
+      "grad_norm": 0.03778542950749397,
+      "learning_rate": 0.0014567901234567903,
+      "loss": 6.6269,
+      "step": 760
+    },
+    {
+      "epoch": 1.6906414884754235,
+      "grad_norm": 0.03874615207314491,
+      "learning_rate": 0.001454320987654321,
+      "loss": 6.1051,
+      "step": 761
+    },
+    {
+      "epoch": 1.6928630935851152,
+      "grad_norm": 0.045976828783750534,
+      "learning_rate": 0.001451851851851852,
+      "loss": 6.3818,
+      "step": 762
+    },
+    {
+      "epoch": 1.695084698694807,
+      "grad_norm": 0.05158616229891777,
+      "learning_rate": 0.0014493827160493827,
+      "loss": 6.282,
+      "step": 763
+    },
+    {
+      "epoch": 1.6973063038044987,
+      "grad_norm": 0.03779592737555504,
+      "learning_rate": 0.0014469135802469136,
+      "loss": 6.557,
+      "step": 764
+    },
+    {
+      "epoch": 1.6995279089141904,
+      "grad_norm": 0.03591408580541611,
+      "learning_rate": 0.0014444444444444444,
+      "loss": 6.3211,
+      "step": 765
+    },
+    {
+      "epoch": 1.7017495140238823,
+      "grad_norm": 0.052788786590099335,
+      "learning_rate": 0.0014419753086419753,
+      "loss": 6.7388,
+      "step": 766
+    },
+    {
+      "epoch": 1.703971119133574,
+      "grad_norm": 0.04353979602456093,
+      "learning_rate": 0.0014395061728395063,
+      "loss": 6.1932,
+      "step": 767
+    },
+    {
+      "epoch": 1.7061927242432657,
+      "grad_norm": 0.05018811672925949,
+      "learning_rate": 0.001437037037037037,
+      "loss": 6.5017,
+      "step": 768
+    },
+    {
+      "epoch": 1.7084143293529577,
+      "grad_norm": 0.0431215800344944,
+      "learning_rate": 0.001434567901234568,
+      "loss": 6.0638,
+      "step": 769
+    },
+    {
+      "epoch": 1.7106359344626494,
+      "grad_norm": 0.0426732562482357,
+      "learning_rate": 0.0014320987654320987,
+      "loss": 6.41,
+      "step": 770
+    },
+    {
+      "epoch": 1.712857539572341,
+      "grad_norm": 0.03792653605341911,
+      "learning_rate": 0.0014296296296296297,
+      "loss": 6.56,
+      "step": 771
+    },
+    {
+      "epoch": 1.7150791446820328,
+      "grad_norm": 0.036381419748067856,
+      "learning_rate": 0.0014271604938271604,
+      "loss": 6.4816,
+      "step": 772
+    },
+    {
+      "epoch": 1.7173007497917245,
+      "grad_norm": 0.040225084871053696,
+      "learning_rate": 0.0014246913580246916,
+      "loss": 6.4904,
+      "step": 773
+    },
+    {
+      "epoch": 1.7195223549014163,
+      "grad_norm": 0.03945305570960045,
+      "learning_rate": 0.0014222222222222223,
+      "loss": 6.4095,
+      "step": 774
+    },
+    {
+      "epoch": 1.721743960011108,
+      "grad_norm": 0.04760852828621864,
+      "learning_rate": 0.0014197530864197532,
+      "loss": 6.697,
+      "step": 775
+    },
+    {
+      "epoch": 1.7239655651207997,
+      "grad_norm": 0.04400087893009186,
+      "learning_rate": 0.001417283950617284,
+      "loss": 6.2095,
+      "step": 776
+    },
+    {
+      "epoch": 1.7261871702304914,
+      "grad_norm": 0.053322967141866684,
+      "learning_rate": 0.001414814814814815,
+      "loss": 6.4843,
+      "step": 777
+    },
+    {
+      "epoch": 1.7284087753401833,
+      "grad_norm": 0.03650697320699692,
+      "learning_rate": 0.0014123456790123457,
+      "loss": 6.3268,
+      "step": 778
+    },
+    {
+      "epoch": 1.730630380449875,
+      "grad_norm": 0.04410003870725632,
+      "learning_rate": 0.0014098765432098766,
+      "loss": 6.7052,
+      "step": 779
+    },
+    {
+      "epoch": 1.7328519855595668,
+      "grad_norm": 0.04782085120677948,
+      "learning_rate": 0.0014074074074074073,
+      "loss": 6.0185,
+      "step": 780
+    },
+    {
+      "epoch": 1.7350735906692587,
+      "grad_norm": 0.04607386514544487,
+      "learning_rate": 0.0014049382716049383,
+      "loss": 6.3951,
+      "step": 781
+    },
+    {
+      "epoch": 1.7372951957789504,
+      "grad_norm": 0.04139106720685959,
+      "learning_rate": 0.0014024691358024693,
+      "loss": 6.4367,
+      "step": 782
+    },
+    {
+      "epoch": 1.7395168008886421,
+      "grad_norm": 0.03845690190792084,
+      "learning_rate": 0.0014,
+      "loss": 6.3108,
+      "step": 783
+    },
+    {
+      "epoch": 1.7417384059983338,
+      "grad_norm": 0.039490532130002975,
+      "learning_rate": 0.001397530864197531,
+      "loss": 6.232,
+      "step": 784
+    },
+    {
+      "epoch": 1.7439600111080256,
+      "grad_norm": 0.0364510640501976,
+      "learning_rate": 0.0013950617283950617,
+      "loss": 6.4885,
+      "step": 785
+    },
+    {
+      "epoch": 1.7461816162177173,
+      "grad_norm": 0.04298551380634308,
+      "learning_rate": 0.0013925925925925926,
+      "loss": 6.31,
+      "step": 786
+    },
+    {
+      "epoch": 1.748403221327409,
+      "grad_norm": 0.0361141711473465,
+      "learning_rate": 0.0013901234567901234,
+      "loss": 6.3128,
+      "step": 787
+    },
+    {
+      "epoch": 1.7506248264371007,
+      "grad_norm": 0.04445536434650421,
+      "learning_rate": 0.0013876543209876545,
+      "loss": 6.321,
+      "step": 788
+    },
+    {
+      "epoch": 1.7528464315467924,
+      "grad_norm": 0.03602416813373566,
+      "learning_rate": 0.0013851851851851853,
+      "loss": 6.3506,
+      "step": 789
+    },
+    {
+      "epoch": 1.7550680366564844,
+      "grad_norm": 0.04584746062755585,
+      "learning_rate": 0.0013827160493827162,
+      "loss": 6.3949,
+      "step": 790
+    },
+    {
+      "epoch": 1.757289641766176,
+      "grad_norm": 0.04659420996904373,
+      "learning_rate": 0.001380246913580247,
+      "loss": 6.3049,
+      "step": 791
+    },
+    {
+      "epoch": 1.7595112468758678,
+      "grad_norm": 0.057917628437280655,
+      "learning_rate": 0.001377777777777778,
+      "loss": 6.8404,
+      "step": 792
+    },
+    {
+      "epoch": 1.7617328519855595,
+      "grad_norm": 0.03402859717607498,
+      "learning_rate": 0.0013753086419753086,
+      "loss": 6.4307,
+      "step": 793
+    },
+    {
+      "epoch": 1.7639544570952514,
+      "grad_norm": 0.04686408489942551,
+      "learning_rate": 0.0013728395061728396,
+      "loss": 6.5087,
+      "step": 794
+    },
+    {
+      "epoch": 1.7661760622049432,
+      "grad_norm": 0.045328594744205475,
+      "learning_rate": 0.0013703703703703703,
+      "loss": 6.5259,
+      "step": 795
+    },
+    {
+      "epoch": 1.7683976673146349,
+      "grad_norm": 0.04060142859816551,
+      "learning_rate": 0.0013679012345679013,
+      "loss": 6.6074,
+      "step": 796
+    },
+    {
+      "epoch": 1.7706192724243266,
+      "grad_norm": 0.04000634327530861,
+      "learning_rate": 0.0013654320987654322,
+      "loss": 6.3505,
+      "step": 797
+    },
+    {
+      "epoch": 1.7728408775340183,
+      "grad_norm": 0.038890015333890915,
+      "learning_rate": 0.001362962962962963,
+      "loss": 6.6029,
+      "step": 798
+    },
+    {
+      "epoch": 1.77506248264371,
+      "grad_norm": 0.03620298206806183,
+      "learning_rate": 0.001360493827160494,
+      "loss": 6.4328,
+      "step": 799
+    },
+    {
+      "epoch": 1.7772840877534017,
+      "grad_norm": 0.03766550123691559,
+      "learning_rate": 0.0013580246913580246,
+      "loss": 6.6665,
+      "step": 800
+    },
+    {
+      "epoch": 1.7795056928630935,
+      "grad_norm": 0.03726218268275261,
+      "learning_rate": 0.0013555555555555556,
+      "loss": 6.4154,
+      "step": 801
+    },
+    {
+      "epoch": 1.7817272979727852,
+      "grad_norm": 0.07254994660615921,
+      "learning_rate": 0.0013530864197530863,
+      "loss": 6.3644,
+      "step": 802
+    },
+    {
+      "epoch": 1.783948903082477,
+      "grad_norm": 0.03856687247753143,
+      "learning_rate": 0.0013506172839506175,
+      "loss": 6.5204,
+      "step": 803
+    },
+    {
+      "epoch": 1.7861705081921688,
+      "grad_norm": 0.03715340420603752,
+      "learning_rate": 0.0013481481481481482,
+      "loss": 6.4817,
+      "step": 804
+    },
+    {
+      "epoch": 1.7883921133018605,
+      "grad_norm": 0.05337759479880333,
+      "learning_rate": 0.0013456790123456792,
+      "loss": 6.2069,
+      "step": 805
+    },
+    {
+      "epoch": 1.7906137184115525,
+      "grad_norm": 0.037285178899765015,
+      "learning_rate": 0.00134320987654321,
+      "loss": 6.5046,
+      "step": 806
+    },
+    {
+      "epoch": 1.7928353235212442,
+      "grad_norm": 0.045639440417289734,
+      "learning_rate": 0.0013407407407407409,
+      "loss": 6.5314,
+      "step": 807
+    },
+    {
+      "epoch": 1.795056928630936,
+      "grad_norm": 0.05591326579451561,
+      "learning_rate": 0.0013382716049382716,
+      "loss": 6.3526,
+      "step": 808
+    },
+    {
+      "epoch": 1.7972785337406276,
+      "grad_norm": 0.03941037505865097,
+      "learning_rate": 0.0013358024691358023,
+      "loss": 6.3686,
+      "step": 809
+    },
+    {
+      "epoch": 1.7995001388503193,
+      "grad_norm": 0.03840603679418564,
+      "learning_rate": 0.0013333333333333333,
+      "loss": 6.3138,
+      "step": 810
+    },
+    {
+      "epoch": 1.801721743960011,
+      "grad_norm": 0.03633950650691986,
+      "learning_rate": 0.0013308641975308642,
+      "loss": 6.5286,
+      "step": 811
+    },
+    {
+      "epoch": 1.8039433490697028,
+      "grad_norm": 0.04522920399904251,
+      "learning_rate": 0.0013283950617283952,
+      "loss": 6.394,
+      "step": 812
+    },
+    {
+      "epoch": 1.8061649541793945,
+      "grad_norm": 0.047142960131168365,
+      "learning_rate": 0.001325925925925926,
+      "loss": 6.3963,
+      "step": 813
+    },
+    {
+      "epoch": 1.8083865592890862,
+      "grad_norm": 0.04085370525717735,
+      "learning_rate": 0.0013234567901234569,
+      "loss": 6.1235,
+      "step": 814
+    },
+    {
+      "epoch": 1.8106081643987781,
+      "grad_norm": 0.047318898141384125,
+      "learning_rate": 0.0013209876543209876,
+      "loss": 6.4397,
+      "step": 815
+    },
+    {
+      "epoch": 1.8128297695084699,
+      "grad_norm": 0.038756296038627625,
+      "learning_rate": 0.0013185185185185186,
+      "loss": 6.4171,
+      "step": 816
+    },
+    {
+      "epoch": 1.8150513746181616,
+      "grad_norm": 0.043222397565841675,
+      "learning_rate": 0.0013160493827160493,
+      "loss": 6.269,
+      "step": 817
+    },
+    {
+      "epoch": 1.8172729797278535,
+      "grad_norm": 0.037330929189920425,
+      "learning_rate": 0.0013135802469135802,
+      "loss": 6.289,
+      "step": 818
+    },
+    {
+      "epoch": 1.8194945848375452,
+      "grad_norm": 0.0425407811999321,
+      "learning_rate": 0.0013111111111111112,
+      "loss": 6.3536,
+      "step": 819
+    },
+    {
+      "epoch": 1.821716189947237,
+      "grad_norm": 0.049727726727724075,
+      "learning_rate": 0.001308641975308642,
+      "loss": 6.5171,
+      "step": 820
+    },
+    {
+      "epoch": 1.8239377950569287,
+      "grad_norm": 0.036017488688230515,
+      "learning_rate": 0.0013061728395061729,
+      "loss": 6.3858,
+      "step": 821
+    },
+    {
+      "epoch": 1.8261594001666204,
+      "grad_norm": 0.06953127682209015,
+      "learning_rate": 0.0013037037037037036,
+      "loss": 6.5516,
+      "step": 822
+    },
+    {
+      "epoch": 1.828381005276312,
+      "grad_norm": 0.042069535702466965,
+      "learning_rate": 0.0013012345679012346,
+      "loss": 6.507,
+      "step": 823
+    },
+    {
+      "epoch": 1.8306026103860038,
+      "grad_norm": 0.037850432097911835,
+      "learning_rate": 0.0012987654320987653,
+      "loss": 6.4339,
+      "step": 824
+    },
+    {
+      "epoch": 1.8328242154956955,
+      "grad_norm": 0.03076726384460926,
+      "learning_rate": 0.0012962962962962963,
+      "loss": 6.4881,
+      "step": 825
+    },
+    {
+      "epoch": 1.8350458206053872,
+      "grad_norm": 0.044283296912908554,
+      "learning_rate": 0.0012938271604938272,
+      "loss": 6.2005,
+      "step": 826
+    },
+    {
+      "epoch": 1.8372674257150792,
+      "grad_norm": 0.05264724791049957,
+      "learning_rate": 0.0012913580246913582,
+      "loss": 6.5486,
+      "step": 827
+    },
+    {
+      "epoch": 1.8394890308247709,
+      "grad_norm": 0.061552777886390686,
+      "learning_rate": 0.001288888888888889,
+      "loss": 6.0343,
+      "step": 828
+    },
+    {
+      "epoch": 1.8417106359344626,
+      "grad_norm": 0.05705179274082184,
+      "learning_rate": 0.0012864197530864198,
+      "loss": 6.2431,
+      "step": 829
+    },
+    {
+      "epoch": 1.8439322410441545,
+      "grad_norm": 0.05785943940281868,
+      "learning_rate": 0.0012839506172839506,
+      "loss": 6.2922,
+      "step": 830
+    },
+    {
+      "epoch": 1.8461538461538463,
+      "grad_norm": 0.04676644876599312,
+      "learning_rate": 0.0012814814814814815,
+      "loss": 6.308,
+      "step": 831
+    },
+    {
+      "epoch": 1.848375451263538,
+      "grad_norm": 0.05422269180417061,
+      "learning_rate": 0.0012790123456790123,
+      "loss": 6.2359,
+      "step": 832
+    },
+    {
+      "epoch": 1.8505970563732297,
+      "grad_norm": 0.055780939757823944,
+      "learning_rate": 0.0012765432098765432,
+      "loss": 6.1558,
+      "step": 833
+    },
+    {
+      "epoch": 1.8528186614829214,
+      "grad_norm": 0.04008074477314949,
+      "learning_rate": 0.0012740740740740742,
+      "loss": 6.4584,
+      "step": 834
+    },
+    {
+      "epoch": 1.8550402665926131,
+      "grad_norm": 0.03681843355298042,
+      "learning_rate": 0.001271604938271605,
+      "loss": 6.1506,
+      "step": 835
+    },
+    {
+      "epoch": 1.8572618717023048,
+      "grad_norm": 0.036749836057424545,
+      "learning_rate": 0.0012691358024691359,
+      "loss": 6.231,
+      "step": 836
+    },
+    {
+      "epoch": 1.8594834768119965,
+      "grad_norm": 0.06679121404886246,
+      "learning_rate": 0.0012666666666666666,
+      "loss": 6.3374,
+      "step": 837
+    },
+    {
+      "epoch": 1.8617050819216883,
+      "grad_norm": 0.04006166383624077,
+      "learning_rate": 0.0012641975308641975,
+      "loss": 6.3086,
+      "step": 838
+    },
+    {
+      "epoch": 1.8639266870313802,
+      "grad_norm": 0.035358864814043045,
+      "learning_rate": 0.0012617283950617283,
+      "loss": 6.5338,
+      "step": 839
+    },
+    {
+      "epoch": 1.866148292141072,
+      "grad_norm": 0.040647611021995544,
+      "learning_rate": 0.0012592592592592592,
+      "loss": 6.2605,
+      "step": 840
+    },
+    {
+      "epoch": 1.8683698972507636,
+      "grad_norm": 0.03877075016498566,
+      "learning_rate": 0.0012567901234567902,
+      "loss": 6.6015,
+      "step": 841
+    },
+    {
+      "epoch": 1.8705915023604556,
+      "grad_norm": 0.0363762229681015,
+      "learning_rate": 0.0012543209876543211,
+      "loss": 6.1976,
+      "step": 842
+    },
+    {
+      "epoch": 1.8728131074701473,
+      "grad_norm": 0.039430152624845505,
+      "learning_rate": 0.0012518518518518519,
+      "loss": 6.3574,
+      "step": 843
+    },
+    {
+      "epoch": 1.875034712579839,
+      "grad_norm": 0.04425748437643051,
+      "learning_rate": 0.0012493827160493828,
+      "loss": 6.217,
+      "step": 844
+    },
+    {
+      "epoch": 1.8772563176895307,
+      "grad_norm": 0.03652188554406166,
+      "learning_rate": 0.0012469135802469136,
+      "loss": 6.3168,
+      "step": 845
+    },
+    {
+      "epoch": 1.8794779227992224,
+      "grad_norm": 0.03891198709607124,
+      "learning_rate": 0.0012444444444444445,
+      "loss": 6.6005,
+      "step": 846
+    },
+    {
+      "epoch": 1.8816995279089141,
+      "grad_norm": 0.04919458553195,
+      "learning_rate": 0.0012419753086419752,
+      "loss": 6.4197,
+      "step": 847
+    },
+    {
+      "epoch": 1.8839211330186059,
+      "grad_norm": 0.04319242388010025,
+      "learning_rate": 0.0012395061728395062,
+      "loss": 6.325,
+      "step": 848
+    },
+    {
+      "epoch": 1.8861427381282976,
+      "grad_norm": 0.045118287205696106,
+      "learning_rate": 0.0012370370370370371,
+      "loss": 6.203,
+      "step": 849
+    },
+    {
+      "epoch": 1.8883643432379893,
+      "grad_norm": 0.04654219374060631,
+      "learning_rate": 0.0012345679012345679,
+      "loss": 6.3819,
+      "step": 850
+    },
+    {
+      "epoch": 1.8905859483476812,
+      "grad_norm": 0.03855947032570839,
+      "learning_rate": 0.0012320987654320988,
+      "loss": 6.4401,
+      "step": 851
+    },
+    {
+      "epoch": 1.892807553457373,
+      "grad_norm": 0.04785769432783127,
+      "learning_rate": 0.0012296296296296296,
+      "loss": 6.1072,
+      "step": 852
+    },
+    {
+      "epoch": 1.8950291585670647,
+      "grad_norm": 0.048641055822372437,
+      "learning_rate": 0.0012271604938271605,
+      "loss": 6.2974,
+      "step": 853
+    },
+    {
+      "epoch": 1.8972507636767566,
+      "grad_norm": 0.05832189321517944,
+      "learning_rate": 0.0012246913580246912,
+      "loss": 6.7048,
+      "step": 854
+    },
+    {
+      "epoch": 1.8994723687864483,
+      "grad_norm": 0.05154256522655487,
+      "learning_rate": 0.0012222222222222222,
+      "loss": 6.5057,
+      "step": 855
+    },
+    {
+      "epoch": 1.90169397389614,
+      "grad_norm": 0.052275024354457855,
+      "learning_rate": 0.0012197530864197532,
+      "loss": 6.3259,
+      "step": 856
+    },
+    {
+      "epoch": 1.9039155790058317,
+      "grad_norm": 0.04063504561781883,
+      "learning_rate": 0.001217283950617284,
+      "loss": 6.2659,
+      "step": 857
+    },
+    {
+      "epoch": 1.9061371841155235,
+      "grad_norm": 0.04190412908792496,
+      "learning_rate": 0.0012148148148148148,
+      "loss": 6.3509,
+      "step": 858
+    },
+    {
+      "epoch": 1.9083587892252152,
+      "grad_norm": 0.04152333736419678,
+      "learning_rate": 0.0012123456790123458,
+      "loss": 6.1813,
+      "step": 859
+    },
+    {
+      "epoch": 1.910580394334907,
+      "grad_norm": 0.04355795681476593,
+      "learning_rate": 0.0012098765432098765,
+      "loss": 6.3199,
+      "step": 860
+    },
+    {
+      "epoch": 1.9128019994445986,
+      "grad_norm": 0.04959927871823311,
+      "learning_rate": 0.0012074074074074075,
+      "loss": 6.4291,
+      "step": 861
+    },
+    {
+      "epoch": 1.9150236045542903,
+      "grad_norm": 0.057414937764406204,
+      "learning_rate": 0.0012049382716049382,
+      "loss": 6.1151,
+      "step": 862
+    },
+    {
+      "epoch": 1.9172452096639823,
+      "grad_norm": 0.045283444225788116,
+      "learning_rate": 0.0012024691358024692,
+      "loss": 6.2453,
+      "step": 863
+    },
+    {
+      "epoch": 1.919466814773674,
+      "grad_norm": 0.0428093746304512,
+      "learning_rate": 0.0012000000000000001,
+      "loss": 6.2606,
+      "step": 864
+    },
+    {
+      "epoch": 1.9216884198833657,
+      "grad_norm": 0.03790568187832832,
+      "learning_rate": 0.0011975308641975308,
+      "loss": 6.2255,
+      "step": 865
+    },
+    {
+      "epoch": 1.9239100249930576,
+      "grad_norm": 0.05110872909426689,
+      "learning_rate": 0.0011950617283950618,
+      "loss": 6.1768,
+      "step": 866
+    },
+    {
+      "epoch": 1.9261316301027493,
+      "grad_norm": 0.04739182069897652,
+      "learning_rate": 0.0011925925925925925,
+      "loss": 6.5636,
+      "step": 867
+    },
+    {
+      "epoch": 1.928353235212441,
+      "grad_norm": 0.05451798439025879,
+      "learning_rate": 0.0011901234567901235,
+      "loss": 6.4633,
+      "step": 868
+    },
+    {
+      "epoch": 1.9305748403221328,
+      "grad_norm": 0.04096847027540207,
+      "learning_rate": 0.0011876543209876542,
+      "loss": 6.6023,
+      "step": 869
+    },
+    {
+      "epoch": 1.9327964454318245,
+      "grad_norm": 0.04547788202762604,
+      "learning_rate": 0.0011851851851851852,
+      "loss": 6.2564,
+      "step": 870
+    },
+    {
+      "epoch": 1.9350180505415162,
+      "grad_norm": 0.044414080679416656,
+      "learning_rate": 0.0011827160493827161,
+      "loss": 6.2996,
+      "step": 871
+    },
+    {
+      "epoch": 1.937239655651208,
+      "grad_norm": 0.058865927159786224,
+      "learning_rate": 0.001180246913580247,
+      "loss": 6.7942,
+      "step": 872
+    },
+    {
+      "epoch": 1.9394612607608996,
+      "grad_norm": 0.04335283115506172,
+      "learning_rate": 0.0011777777777777778,
+      "loss": 6.1104,
+      "step": 873
+    },
+    {
+      "epoch": 1.9416828658705914,
+      "grad_norm": 0.07237697392702103,
+      "learning_rate": 0.0011753086419753088,
+      "loss": 6.2427,
+      "step": 874
+    },
+    {
+      "epoch": 1.9439044709802833,
+      "grad_norm": 0.03939530625939369,
+      "learning_rate": 0.0011728395061728395,
+      "loss": 6.1578,
+      "step": 875
+    },
+    {
+      "epoch": 1.946126076089975,
+      "grad_norm": 0.05042354017496109,
+      "learning_rate": 0.0011703703703703704,
+      "loss": 6.5269,
+      "step": 876
+    },
+    {
+      "epoch": 1.9483476811996667,
+      "grad_norm": 0.04062674194574356,
+      "learning_rate": 0.0011679012345679012,
+      "loss": 6.4539,
+      "step": 877
+    },
+    {
+      "epoch": 1.9505692863093587,
+      "grad_norm": 0.06741116940975189,
+      "learning_rate": 0.0011654320987654321,
+      "loss": 6.3244,
+      "step": 878
+    },
+    {
+      "epoch": 1.9527908914190504,
+      "grad_norm": 0.04551272839307785,
+      "learning_rate": 0.001162962962962963,
+      "loss": 6.5239,
+      "step": 879
+    },
+    {
+      "epoch": 1.955012496528742,
+      "grad_norm": 0.040041789412498474,
+      "learning_rate": 0.0011604938271604938,
+      "loss": 6.2119,
+      "step": 880
+    },
+    {
+      "epoch": 1.9572341016384338,
+      "grad_norm": 0.04255872964859009,
+      "learning_rate": 0.0011580246913580248,
+      "loss": 6.3607,
+      "step": 881
+    },
+    {
+      "epoch": 1.9594557067481255,
+      "grad_norm": 0.04652669280767441,
+      "learning_rate": 0.0011555555555555555,
+      "loss": 6.5182,
+      "step": 882
+    },
+    {
+      "epoch": 1.9616773118578172,
+      "grad_norm": 0.04208851605653763,
+      "learning_rate": 0.0011530864197530865,
+      "loss": 6.2325,
+      "step": 883
+    },
+    {
+      "epoch": 1.963898916967509,
+      "grad_norm": 0.0562903955578804,
+      "learning_rate": 0.0011506172839506172,
+      "loss": 6.1653,
+      "step": 884
+    },
+    {
+      "epoch": 1.9661205220772007,
+      "grad_norm": 0.03999092057347298,
+      "learning_rate": 0.0011481481481481481,
+      "loss": 6.2752,
+      "step": 885
+    },
+    {
+      "epoch": 1.9683421271868924,
+      "grad_norm": 0.057688936591148376,
+      "learning_rate": 0.001145679012345679,
+      "loss": 6.5369,
+      "step": 886
+    },
+    {
+      "epoch": 1.9705637322965843,
+      "grad_norm": 0.04506329819560051,
+      "learning_rate": 0.00114320987654321,
+      "loss": 6.2929,
+      "step": 887
+    },
+    {
+      "epoch": 1.972785337406276,
+      "grad_norm": 0.04111102595925331,
+      "learning_rate": 0.0011407407407407408,
+      "loss": 6.4603,
+      "step": 888
+    },
+    {
+      "epoch": 1.9750069425159678,
+      "grad_norm": 0.05236571282148361,
+      "learning_rate": 0.0011382716049382717,
+      "loss": 6.4297,
+      "step": 889
+    },
+    {
+      "epoch": 1.9772285476256597,
+      "grad_norm": 0.04919304698705673,
+      "learning_rate": 0.0011358024691358025,
+      "loss": 6.0283,
+      "step": 890
+    },
+    {
+      "epoch": 1.9794501527353514,
+      "grad_norm": 0.04021111875772476,
+      "learning_rate": 0.0011333333333333334,
+      "loss": 6.054,
+      "step": 891
+    },
+    {
+      "epoch": 1.9816717578450431,
+      "grad_norm": 0.04439817741513252,
+      "learning_rate": 0.0011308641975308641,
+      "loss": 6.4379,
+      "step": 892
+    },
+    {
+      "epoch": 1.9838933629547348,
+      "grad_norm": 0.04219279810786247,
+      "learning_rate": 0.001128395061728395,
+      "loss": 6.6574,
+      "step": 893
+    },
+    {
+      "epoch": 1.9861149680644266,
+      "grad_norm": 0.04125041142106056,
+      "learning_rate": 0.001125925925925926,
+      "loss": 6.5064,
+      "step": 894
+    },
+    {
+      "epoch": 1.9883365731741183,
+      "grad_norm": 0.05733527988195419,
+      "learning_rate": 0.0011234567901234568,
+      "loss": 6.363,
+      "step": 895
+    },
+    {
+      "epoch": 1.99055817828381,
+      "grad_norm": 0.05712810903787613,
+      "learning_rate": 0.0011209876543209877,
+      "loss": 6.4407,
+      "step": 896
+    },
+    {
+      "epoch": 1.9927797833935017,
+      "grad_norm": 0.04266197234392166,
+      "learning_rate": 0.0011185185185185185,
+      "loss": 6.287,
+      "step": 897
+    },
+    {
+      "epoch": 1.9950013885031934,
+      "grad_norm": 0.04644905403256416,
+      "learning_rate": 0.0011160493827160494,
+      "loss": 6.4703,
+      "step": 898
+    },
+    {
+      "epoch": 1.9972229936128854,
+      "grad_norm": 0.046037863940000534,
+      "learning_rate": 0.0011135802469135802,
+      "loss": 6.1607,
+      "step": 899
+    },
+    {
+      "epoch": 1.999444598722577,
+      "grad_norm": 0.04576002061367035,
+      "learning_rate": 0.0011111111111111111,
+      "loss": 6.3653,
+      "step": 900
+    },
+    {
+      "epoch": 2.001666203832269,
+      "grad_norm": 0.044221434742212296,
+      "learning_rate": 0.001108641975308642,
+      "loss": 6.5826,
+      "step": 901
+    },
+    {
+      "epoch": 2.0038878089419607,
+      "grad_norm": 0.056456565856933594,
+      "learning_rate": 0.001106172839506173,
+      "loss": 5.9399,
+      "step": 902
+    },
+    {
+      "epoch": 2.0061094140516524,
+      "grad_norm": 0.04235842078924179,
+      "learning_rate": 0.0011037037037037037,
+      "loss": 6.1734,
+      "step": 903
+    },
+    {
+      "epoch": 2.008331019161344,
+      "grad_norm": 0.03594612330198288,
+      "learning_rate": 0.0011012345679012347,
+      "loss": 6.425,
+      "step": 904
+    },
+    {
+      "epoch": 2.010552624271036,
+      "grad_norm": 0.043131452053785324,
+      "learning_rate": 0.0010987654320987654,
+      "loss": 5.9612,
+      "step": 905
+    },
+    {
+      "epoch": 2.0127742293807276,
+      "grad_norm": 0.05280978977680206,
+      "learning_rate": 0.0010962962962962964,
+      "loss": 6.551,
+      "step": 906
+    },
+    {
+      "epoch": 2.0149958344904193,
+      "grad_norm": 0.0525561161339283,
+      "learning_rate": 0.0010938271604938271,
+      "loss": 6.4781,
+      "step": 907
+    },
+    {
+      "epoch": 2.017217439600111,
+      "grad_norm": 0.03620578348636627,
+      "learning_rate": 0.001091358024691358,
+      "loss": 6.3529,
+      "step": 908
+    },
+    {
+      "epoch": 2.0194390447098027,
+      "grad_norm": 0.04835907369852066,
+      "learning_rate": 0.001088888888888889,
+      "loss": 6.2281,
+      "step": 909
+    },
+    {
+      "epoch": 2.0216606498194944,
+      "grad_norm": 0.047538936138153076,
+      "learning_rate": 0.0010864197530864198,
+      "loss": 6.5803,
+      "step": 910
+    },
+    {
+      "epoch": 2.023882254929186,
+      "grad_norm": 0.04271189868450165,
+      "learning_rate": 0.0010839506172839507,
+      "loss": 6.3975,
+      "step": 911
+    },
+    {
+      "epoch": 2.026103860038878,
+      "grad_norm": 0.05404246225953102,
+      "learning_rate": 0.0010814814814814814,
+      "loss": 6.5456,
+      "step": 912
+    },
+    {
+      "epoch": 2.02832546514857,
+      "grad_norm": 0.040102358907461166,
+      "learning_rate": 0.0010790123456790124,
+      "loss": 6.4139,
+      "step": 913
+    },
+    {
+      "epoch": 2.0305470702582618,
+      "grad_norm": 0.060644637793302536,
+      "learning_rate": 0.0010765432098765431,
+      "loss": 6.1298,
+      "step": 914
+    },
+    {
+      "epoch": 2.0327686753679535,
+      "grad_norm": 0.06346485018730164,
+      "learning_rate": 0.001074074074074074,
+      "loss": 6.0192,
+      "step": 915
+    },
+    {
+      "epoch": 2.034990280477645,
+      "grad_norm": 0.033958740532398224,
+      "learning_rate": 0.0010716049382716048,
+      "loss": 6.2263,
+      "step": 916
+    },
+    {
+      "epoch": 2.037211885587337,
+      "grad_norm": 0.05422692373394966,
+      "learning_rate": 0.001069135802469136,
+      "loss": 6.4275,
+      "step": 917
+    },
+    {
+      "epoch": 2.0394334906970286,
+      "grad_norm": 0.04352685436606407,
+      "learning_rate": 0.0010666666666666667,
+      "loss": 6.7603,
+      "step": 918
+    },
+    {
+      "epoch": 2.0416550958067203,
+      "grad_norm": 0.03349572792649269,
+      "learning_rate": 0.0010641975308641977,
+      "loss": 6.2464,
+      "step": 919
+    },
+    {
+      "epoch": 2.043876700916412,
+      "grad_norm": 0.05073563754558563,
+      "learning_rate": 0.0010617283950617284,
+      "loss": 6.1693,
+      "step": 920
+    },
+    {
+      "epoch": 2.0460983060261038,
+      "grad_norm": 0.04028453305363655,
+      "learning_rate": 0.0010592592592592594,
+      "loss": 6.2452,
+      "step": 921
+    },
+    {
+      "epoch": 2.0483199111357955,
+      "grad_norm": 0.06080702692270279,
+      "learning_rate": 0.00105679012345679,
+      "loss": 6.134,
+      "step": 922
+    },
+    {
+      "epoch": 2.050541516245487,
+      "grad_norm": 0.0440647229552269,
+      "learning_rate": 0.001054320987654321,
+      "loss": 6.2069,
+      "step": 923
+    },
+    {
+      "epoch": 2.052763121355179,
+      "grad_norm": 0.05307413637638092,
+      "learning_rate": 0.001051851851851852,
+      "loss": 6.6085,
+      "step": 924
+    },
+    {
+      "epoch": 2.054984726464871,
+      "grad_norm": 0.04364542290568352,
+      "learning_rate": 0.0010493827160493827,
+      "loss": 6.4156,
+      "step": 925
+    },
+    {
+      "epoch": 2.057206331574563,
+      "grad_norm": 0.04296353831887245,
+      "learning_rate": 0.0010469135802469137,
+      "loss": 6.5389,
+      "step": 926
+    },
+    {
+      "epoch": 2.0594279366842545,
+      "grad_norm": 0.05563860014081001,
+      "learning_rate": 0.0010444444444444444,
+      "loss": 6.43,
+      "step": 927
+    },
+    {
+      "epoch": 2.061649541793946,
+      "grad_norm": 0.050656117498874664,
+      "learning_rate": 0.0010419753086419754,
+      "loss": 6.3053,
+      "step": 928
+    },
+    {
+      "epoch": 2.063871146903638,
+      "grad_norm": 0.06056036800146103,
+      "learning_rate": 0.001039506172839506,
+      "loss": 6.343,
+      "step": 929
+    },
+    {
+      "epoch": 2.0660927520133296,
+      "grad_norm": 0.04997817054390907,
+      "learning_rate": 0.001037037037037037,
+      "loss": 6.4957,
+      "step": 930
+    },
+    {
+      "epoch": 2.0683143571230214,
+      "grad_norm": 0.052574366331100464,
+      "learning_rate": 0.0010345679012345678,
+      "loss": 6.5022,
+      "step": 931
+    },
+    {
+      "epoch": 2.070535962232713,
+      "grad_norm": 0.052750907838344574,
+      "learning_rate": 0.001032098765432099,
+      "loss": 6.3314,
+      "step": 932
+    },
+    {
+      "epoch": 2.072757567342405,
+      "grad_norm": 0.052910882979631424,
+      "learning_rate": 0.0010296296296296297,
+      "loss": 6.3761,
+      "step": 933
+    },
+    {
+      "epoch": 2.0749791724520965,
+      "grad_norm": 0.04710180684924126,
+      "learning_rate": 0.0010271604938271606,
+      "loss": 6.1257,
+      "step": 934
+    },
+    {
+      "epoch": 2.0772007775617882,
+      "grad_norm": 0.043091725558042526,
+      "learning_rate": 0.0010246913580246914,
+      "loss": 6.4062,
+      "step": 935
+    },
+    {
+      "epoch": 2.07942238267148,
+      "grad_norm": 0.054223112761974335,
+      "learning_rate": 0.0010222222222222223,
+      "loss": 6.3952,
+      "step": 936
+    },
+    {
+      "epoch": 2.081643987781172,
+      "grad_norm": 0.07009416073560715,
+      "learning_rate": 0.001019753086419753,
+      "loss": 6.2675,
+      "step": 937
+    },
+    {
+      "epoch": 2.083865592890864,
+      "grad_norm": 0.05066051706671715,
+      "learning_rate": 0.001017283950617284,
+      "loss": 6.2899,
+      "step": 938
+    },
+    {
+      "epoch": 2.0860871980005555,
+      "grad_norm": 0.05297587439417839,
+      "learning_rate": 0.001014814814814815,
+      "loss": 6.3116,
+      "step": 939
+    },
+    {
+      "epoch": 2.0883088031102472,
+      "grad_norm": 0.04367813467979431,
+      "learning_rate": 0.0010123456790123457,
+      "loss": 6.1994,
+      "step": 940
+    },
+    {
+      "epoch": 2.090530408219939,
+      "grad_norm": 0.06020025908946991,
+      "learning_rate": 0.0010098765432098766,
+      "loss": 6.1024,
+      "step": 941
+    },
+    {
+      "epoch": 2.0927520133296307,
+      "grad_norm": 0.038126129657030106,
+      "learning_rate": 0.0010074074074074074,
+      "loss": 6.2977,
+      "step": 942
+    },
+    {
+      "epoch": 2.0949736184393224,
+      "grad_norm": 0.08368539810180664,
+      "learning_rate": 0.0010049382716049383,
+      "loss": 5.8574,
+      "step": 943
+    },
+    {
+      "epoch": 2.097195223549014,
+      "grad_norm": 0.05177859216928482,
+      "learning_rate": 0.001002469135802469,
+      "loss": 6.6817,
+      "step": 944
+    },
+    {
+      "epoch": 2.099416828658706,
+      "grad_norm": 0.04672938957810402,
+      "learning_rate": 0.001,
+      "loss": 6.1247,
+      "step": 945
+    },
+    {
+      "epoch": 2.1016384337683975,
+      "grad_norm": 0.046738337725400925,
+      "learning_rate": 0.0009975308641975308,
+      "loss": 6.2191,
+      "step": 946
+    },
+    {
+      "epoch": 2.1038600388780893,
+      "grad_norm": 0.050877295434474945,
+      "learning_rate": 0.000995061728395062,
+      "loss": 6.3428,
+      "step": 947
+    },
+    {
+      "epoch": 2.106081643987781,
+      "grad_norm": 0.04899909347295761,
+      "learning_rate": 0.0009925925925925927,
+      "loss": 6.0717,
+      "step": 948
+    },
+    {
+      "epoch": 2.108303249097473,
+      "grad_norm": 0.046309296041727066,
+      "learning_rate": 0.0009901234567901234,
+      "loss": 6.096,
+      "step": 949
+    },
+    {
+      "epoch": 2.110524854207165,
+      "grad_norm": 0.047153934836387634,
+      "learning_rate": 0.0009876543209876543,
+      "loss": 6.5243,
+      "step": 950
+    },
+    {
+      "epoch": 2.1127464593168566,
+      "grad_norm": 0.04886336624622345,
+      "learning_rate": 0.000985185185185185,
+      "loss": 6.3339,
+      "step": 951
+    },
+    {
+      "epoch": 2.1149680644265483,
+      "grad_norm": 0.04061562195420265,
+      "learning_rate": 0.000982716049382716,
+      "loss": 6.3668,
+      "step": 952
+    },
+    {
+      "epoch": 2.11718966953624,
+      "grad_norm": 0.04561516270041466,
+      "learning_rate": 0.0009802469135802468,
+      "loss": 6.3071,
+      "step": 953
+    },
+    {
+      "epoch": 2.1194112746459317,
+      "grad_norm": 0.05609046295285225,
+      "learning_rate": 0.000977777777777778,
+      "loss": 6.4806,
+      "step": 954
+    },
+    {
+      "epoch": 2.1216328797556234,
+      "grad_norm": 0.04938310757279396,
+      "learning_rate": 0.0009753086419753087,
+      "loss": 6.5487,
+      "step": 955
+    },
+    {
+      "epoch": 2.123854484865315,
+      "grad_norm": 0.045022930949926376,
+      "learning_rate": 0.0009728395061728395,
+      "loss": 6.1813,
+      "step": 956
+    },
+    {
+      "epoch": 2.126076089975007,
+      "grad_norm": 0.04955757409334183,
+      "learning_rate": 0.0009703703703703704,
+      "loss": 6.2323,
+      "step": 957
+    },
+    {
+      "epoch": 2.1282976950846986,
+      "grad_norm": 0.04581235349178314,
+      "learning_rate": 0.0009679012345679012,
+      "loss": 6.4931,
+      "step": 958
+    },
+    {
+      "epoch": 2.1305193001943903,
+      "grad_norm": 0.05137951672077179,
+      "learning_rate": 0.000965432098765432,
+      "loss": 6.5015,
+      "step": 959
+    },
+    {
+      "epoch": 2.132740905304082,
+      "grad_norm": 0.051154669374227524,
+      "learning_rate": 0.0009629629629629629,
+      "loss": 6.1415,
+      "step": 960
+    },
+    {
+      "epoch": 2.134962510413774,
+      "grad_norm": 0.050333019345998764,
+      "learning_rate": 0.0009604938271604937,
+      "loss": 6.2002,
+      "step": 961
+    },
+    {
+      "epoch": 2.137184115523466,
+      "grad_norm": 0.062799833714962,
+      "learning_rate": 0.0009580246913580248,
+      "loss": 6.1475,
+      "step": 962
+    },
+    {
+      "epoch": 2.1394057206331576,
+      "grad_norm": 0.07070323079824448,
+      "learning_rate": 0.0009555555555555556,
+      "loss": 6.0727,
+      "step": 963
+    },
+    {
+      "epoch": 2.1416273257428493,
+      "grad_norm": 0.06419878453016281,
+      "learning_rate": 0.0009530864197530865,
+      "loss": 6.4612,
+      "step": 964
+    },
+    {
+      "epoch": 2.143848930852541,
+      "grad_norm": 0.04130703583359718,
+      "learning_rate": 0.0009506172839506173,
+      "loss": 6.0991,
+      "step": 965
+    },
+    {
+      "epoch": 2.1460705359622327,
+      "grad_norm": 0.05469075217843056,
+      "learning_rate": 0.0009481481481481482,
+      "loss": 6.2976,
+      "step": 966
+    },
+    {
+      "epoch": 2.1482921410719245,
+      "grad_norm": 0.04793629050254822,
+      "learning_rate": 0.000945679012345679,
+      "loss": 6.1802,
+      "step": 967
+    },
+    {
+      "epoch": 2.150513746181616,
+      "grad_norm": 0.04094194620847702,
+      "learning_rate": 0.0009432098765432098,
+      "loss": 6.2516,
+      "step": 968
+    },
+    {
+      "epoch": 2.152735351291308,
+      "grad_norm": 0.04848310351371765,
+      "learning_rate": 0.0009407407407407408,
+      "loss": 6.255,
+      "step": 969
+    },
+    {
+      "epoch": 2.1549569564009996,
+      "grad_norm": 0.04366128891706467,
+      "learning_rate": 0.0009382716049382716,
+      "loss": 6.0069,
+      "step": 970
+    },
+    {
+      "epoch": 2.1571785615106913,
+      "grad_norm": 0.05942326784133911,
+      "learning_rate": 0.0009358024691358025,
+      "loss": 6.2808,
+      "step": 971
+    },
+    {
+      "epoch": 2.159400166620383,
+      "grad_norm": 0.05622553825378418,
+      "learning_rate": 0.0009333333333333333,
+      "loss": 6.013,
+      "step": 972
+    },
+    {
+      "epoch": 2.1616217717300747,
+      "grad_norm": 0.04568716883659363,
+      "learning_rate": 0.0009308641975308642,
+      "loss": 6.283,
+      "step": 973
+    },
+    {
+      "epoch": 2.163843376839767,
+      "grad_norm": 0.0689627081155777,
+      "learning_rate": 0.000928395061728395,
+      "loss": 6.3248,
+      "step": 974
+    },
+    {
+      "epoch": 2.1660649819494586,
+      "grad_norm": 0.04783041030168533,
+      "learning_rate": 0.0009259259259259259,
+      "loss": 6.2849,
+      "step": 975
+    },
+    {
+      "epoch": 2.1682865870591503,
+      "grad_norm": 0.04585827887058258,
+      "learning_rate": 0.0009234567901234567,
+      "loss": 6.3175,
+      "step": 976
+    },
+    {
+      "epoch": 2.170508192168842,
+      "grad_norm": 0.04715028405189514,
+      "learning_rate": 0.0009209876543209878,
+      "loss": 6.2875,
+      "step": 977
+    },
+    {
+      "epoch": 2.1727297972785338,
+      "grad_norm": 0.04998511075973511,
+      "learning_rate": 0.0009185185185185186,
+      "loss": 6.3223,
+      "step": 978
+    },
+    {
+      "epoch": 2.1749514023882255,
+      "grad_norm": 0.05318531021475792,
+      "learning_rate": 0.0009160493827160494,
+      "loss": 6.2933,
+      "step": 979
+    },
+    {
+      "epoch": 2.177173007497917,
+      "grad_norm": 0.041119400411844254,
+      "learning_rate": 0.0009135802469135803,
+      "loss": 6.1904,
+      "step": 980
+    },
+    {
+      "epoch": 2.179394612607609,
+      "grad_norm": 0.044577039778232574,
+      "learning_rate": 0.0009111111111111111,
+      "loss": 6.4205,
+      "step": 981
+    },
+    {
+      "epoch": 2.1816162177173006,
+      "grad_norm": 0.05574436113238335,
+      "learning_rate": 0.000908641975308642,
+      "loss": 6.0254,
+      "step": 982
+    },
+    {
+      "epoch": 2.1838378228269923,
+      "grad_norm": 0.044923435896635056,
+      "learning_rate": 0.0009061728395061728,
+      "loss": 6.1393,
+      "step": 983
+    },
+    {
+      "epoch": 2.186059427936684,
+      "grad_norm": 0.03584975376725197,
+      "learning_rate": 0.0009037037037037038,
+      "loss": 6.4361,
+      "step": 984
+    },
+    {
+      "epoch": 2.1882810330463762,
+      "grad_norm": 0.05221906676888466,
+      "learning_rate": 0.0009012345679012346,
+      "loss": 6.4745,
+      "step": 985
+    },
+    {
+      "epoch": 2.190502638156068,
+      "grad_norm": 0.04038187488913536,
+      "learning_rate": 0.0008987654320987655,
+      "loss": 6.2099,
+      "step": 986
+    },
+    {
+      "epoch": 2.1927242432657597,
+      "grad_norm": 0.042402077466249466,
+      "learning_rate": 0.0008962962962962963,
+      "loss": 6.445,
+      "step": 987
+    },
+    {
+      "epoch": 2.1949458483754514,
+      "grad_norm": 0.03708990290760994,
+      "learning_rate": 0.0008938271604938271,
+      "loss": 5.9841,
+      "step": 988
+    },
+    {
+      "epoch": 2.197167453485143,
+      "grad_norm": 0.05086686089634895,
+      "learning_rate": 0.000891358024691358,
+      "loss": 6.0279,
+      "step": 989
+    },
+    {
+      "epoch": 2.199389058594835,
+      "grad_norm": 0.05341973528265953,
+      "learning_rate": 0.0008888888888888888,
+      "loss": 6.0115,
+      "step": 990
+    },
+    {
+      "epoch": 2.2016106637045265,
+      "grad_norm": 0.048969414085149765,
+      "learning_rate": 0.0008864197530864197,
+      "loss": 6.1645,
+      "step": 991
+    },
+    {
+      "epoch": 2.2038322688142182,
+      "grad_norm": 0.044248126447200775,
+      "learning_rate": 0.0008839506172839507,
+      "loss": 6.2637,
+      "step": 992
+    },
+    {
+      "epoch": 2.20605387392391,
+      "grad_norm": 0.04010443016886711,
+      "learning_rate": 0.0008814814814814816,
+      "loss": 6.3516,
+      "step": 993
+    },
+    {
+      "epoch": 2.2082754790336017,
+      "grad_norm": 0.05198604613542557,
+      "learning_rate": 0.0008790123456790124,
+      "loss": 6.3876,
+      "step": 994
+    },
+    {
+      "epoch": 2.2104970841432934,
+      "grad_norm": 0.05000235140323639,
+      "learning_rate": 0.0008765432098765433,
+      "loss": 6.152,
+      "step": 995
+    },
+    {
+      "epoch": 2.212718689252985,
+      "grad_norm": 0.05713628605008125,
+      "learning_rate": 0.0008740740740740741,
+      "loss": 6.3673,
+      "step": 996
+    },
+    {
+      "epoch": 2.214940294362677,
+      "grad_norm": 0.057237010449171066,
+      "learning_rate": 0.0008716049382716049,
+      "loss": 6.1046,
+      "step": 997
+    },
+    {
+      "epoch": 2.217161899472369,
+      "grad_norm": 0.0695032924413681,
+      "learning_rate": 0.0008691358024691358,
+      "loss": 6.2222,
+      "step": 998
+    },
+    {
+      "epoch": 2.2193835045820607,
+      "grad_norm": 0.050379615277051926,
+      "learning_rate": 0.0008666666666666666,
+      "loss": 6.1442,
+      "step": 999
+    },
+    {
+      "epoch": 2.2216051096917524,
+      "grad_norm": 0.040284428745508194,
+      "learning_rate": 0.0008641975308641976,
+      "loss": 6.3371,
+      "step": 1000
+    },
+    {
+      "epoch": 2.223826714801444,
+      "grad_norm": 0.04303325340151787,
+      "learning_rate": 0.0008617283950617284,
+      "loss": 6.5961,
+      "step": 1001
+    },
+    {
+      "epoch": 2.226048319911136,
+      "grad_norm": 0.0598965659737587,
+      "learning_rate": 0.0008592592592592593,
+      "loss": 6.4752,
+      "step": 1002
+    },
+    {
+      "epoch": 2.2282699250208275,
+      "grad_norm": 0.04657476767897606,
+      "learning_rate": 0.0008567901234567901,
+      "loss": 6.354,
+      "step": 1003
+    },
+    {
+      "epoch": 2.2304915301305193,
+      "grad_norm": 0.057953160256147385,
+      "learning_rate": 0.000854320987654321,
+      "loss": 6.1963,
+      "step": 1004
+    },
+    {
+      "epoch": 2.232713135240211,
+      "grad_norm": 0.038973864167928696,
+      "learning_rate": 0.0008518518518518518,
+      "loss": 6.0581,
+      "step": 1005
+    },
+    {
+      "epoch": 2.2349347403499027,
+      "grad_norm": 0.042866986244916916,
+      "learning_rate": 0.0008493827160493826,
+      "loss": 6.3811,
+      "step": 1006
+    },
+    {
+      "epoch": 2.2371563454595944,
+      "grad_norm": 0.0404992550611496,
+      "learning_rate": 0.0008469135802469137,
+      "loss": 6.3086,
+      "step": 1007
+    },
+    {
+      "epoch": 2.239377950569286,
+      "grad_norm": 0.046063970774412155,
+      "learning_rate": 0.0008444444444444445,
+      "loss": 6.1674,
+      "step": 1008
+    },
+    {
+      "epoch": 2.2415995556789783,
+      "grad_norm": 0.0643259584903717,
+      "learning_rate": 0.0008419753086419754,
+      "loss": 6.086,
+      "step": 1009
+    },
+    {
+      "epoch": 2.24382116078867,
+      "grad_norm": 0.0459156259894371,
+      "learning_rate": 0.0008395061728395062,
+      "loss": 6.0461,
+      "step": 1010
+    },
+    {
+      "epoch": 2.2460427658983617,
+      "grad_norm": 0.05444741249084473,
+      "learning_rate": 0.0008370370370370371,
+      "loss": 6.0138,
+      "step": 1011
+    },
+    {
+      "epoch": 2.2482643710080534,
+      "grad_norm": 0.044266391545534134,
+      "learning_rate": 0.0008345679012345679,
+      "loss": 6.1491,
+      "step": 1012
+    },
+    {
+      "epoch": 2.250485976117745,
+      "grad_norm": 0.05736247077584267,
+      "learning_rate": 0.0008320987654320988,
+      "loss": 6.235,
+      "step": 1013
+    },
+    {
+      "epoch": 2.252707581227437,
+      "grad_norm": 0.05814267694950104,
+      "learning_rate": 0.0008296296296296296,
+      "loss": 6.1927,
+      "step": 1014
+    },
+    {
+      "epoch": 2.2549291863371286,
+      "grad_norm": 0.05598462000489235,
+      "learning_rate": 0.0008271604938271605,
+      "loss": 6.0999,
+      "step": 1015
+    },
+    {
+      "epoch": 2.2571507914468203,
+      "grad_norm": 0.04414599761366844,
+      "learning_rate": 0.0008246913580246914,
+      "loss": 6.555,
+      "step": 1016
+    },
+    {
+      "epoch": 2.259372396556512,
+      "grad_norm": 0.04954313486814499,
+      "learning_rate": 0.0008222222222222222,
+      "loss": 6.0939,
+      "step": 1017
+    },
+    {
+      "epoch": 2.2615940016662037,
+      "grad_norm": 0.062405653297901154,
+      "learning_rate": 0.0008197530864197531,
+      "loss": 6.2513,
+      "step": 1018
+    },
+    {
+      "epoch": 2.2638156067758954,
+      "grad_norm": 0.046896327286958694,
+      "learning_rate": 0.0008172839506172839,
+      "loss": 6.4348,
+      "step": 1019
+    },
+    {
+      "epoch": 2.266037211885587,
+      "grad_norm": 0.04528604820370674,
+      "learning_rate": 0.0008148148148148148,
+      "loss": 5.8828,
+      "step": 1020
+    },
+    {
+      "epoch": 2.268258816995279,
+      "grad_norm": 0.05379809811711311,
+      "learning_rate": 0.0008123456790123456,
+      "loss": 6.2505,
+      "step": 1021
+    },
+    {
+      "epoch": 2.270480422104971,
+      "grad_norm": 0.04056351259350777,
+      "learning_rate": 0.0008098765432098767,
+      "loss": 6.1893,
+      "step": 1022
+    },
+    {
+      "epoch": 2.2727020272146627,
+      "grad_norm": 0.045778386294841766,
+      "learning_rate": 0.0008074074074074075,
+      "loss": 6.0922,
+      "step": 1023
+    },
+    {
+      "epoch": 2.2749236323243545,
+      "grad_norm": 0.04002056270837784,
+      "learning_rate": 0.0008049382716049384,
+      "loss": 6.0907,
+      "step": 1024
+    },
+    {
+      "epoch": 2.277145237434046,
+      "grad_norm": 0.04901435226202011,
+      "learning_rate": 0.0008024691358024692,
+      "loss": 6.2603,
+      "step": 1025
+    },
+    {
+      "epoch": 2.279366842543738,
+      "grad_norm": 0.04776861518621445,
+      "learning_rate": 0.0008,
+      "loss": 6.2154,
+      "step": 1026
+    },
+    {
+      "epoch": 2.2815884476534296,
+      "grad_norm": 0.07206933945417404,
+      "learning_rate": 0.0007975308641975309,
+      "loss": 6.2986,
+      "step": 1027
+    },
+    {
+      "epoch": 2.2838100527631213,
+      "grad_norm": 0.058153215795755386,
+      "learning_rate": 0.0007950617283950617,
+      "loss": 6.5325,
+      "step": 1028
+    },
+    {
+      "epoch": 2.286031657872813,
+      "grad_norm": 0.05404128134250641,
+      "learning_rate": 0.0007925925925925926,
+      "loss": 6.2201,
+      "step": 1029
+    },
+    {
+      "epoch": 2.2882532629825048,
+      "grad_norm": 0.04366792365908623,
+      "learning_rate": 0.0007901234567901235,
+      "loss": 6.4792,
+      "step": 1030
+    },
+    {
+      "epoch": 2.2904748680921965,
+      "grad_norm": 0.050734743475914,
+      "learning_rate": 0.0007876543209876544,
+      "loss": 6.3662,
+      "step": 1031
+    },
+    {
+      "epoch": 2.292696473201888,
+      "grad_norm": 0.04845214635133743,
+      "learning_rate": 0.0007851851851851852,
+      "loss": 6.1045,
+      "step": 1032
+    },
+    {
+      "epoch": 2.2949180783115803,
+      "grad_norm": 0.059091925621032715,
+      "learning_rate": 0.000782716049382716,
+      "loss": 6.2868,
+      "step": 1033
+    },
+    {
+      "epoch": 2.297139683421272,
+      "grad_norm": 0.07321962714195251,
+      "learning_rate": 0.0007802469135802469,
+      "loss": 6.4283,
+      "step": 1034
+    },
+    {
+      "epoch": 2.2993612885309638,
+      "grad_norm": 0.042170073837041855,
+      "learning_rate": 0.0007777777777777777,
+      "loss": 6.3964,
+      "step": 1035
+    },
+    {
+      "epoch": 2.3015828936406555,
+      "grad_norm": 0.04940120503306389,
+      "learning_rate": 0.0007753086419753086,
+      "loss": 5.9966,
+      "step": 1036
+    },
+    {
+      "epoch": 2.303804498750347,
+      "grad_norm": 0.05433443561196327,
+      "learning_rate": 0.0007728395061728396,
+      "loss": 6.167,
+      "step": 1037
+    },
+    {
+      "epoch": 2.306026103860039,
+      "grad_norm": 0.050197504460811615,
+      "learning_rate": 0.0007703703703703705,
+      "loss": 6.211,
+      "step": 1038
+    },
+    {
+      "epoch": 2.3082477089697306,
+      "grad_norm": 0.04551539942622185,
+      "learning_rate": 0.0007679012345679013,
+      "loss": 6.23,
+      "step": 1039
+    },
+    {
+      "epoch": 2.3104693140794224,
+      "grad_norm": 0.04176529496908188,
+      "learning_rate": 0.0007654320987654322,
+      "loss": 6.4212,
+      "step": 1040
+    },
+    {
+      "epoch": 2.312690919189114,
+      "grad_norm": 0.04640606790781021,
+      "learning_rate": 0.000762962962962963,
+      "loss": 6.3226,
+      "step": 1041
+    },
+    {
+      "epoch": 2.314912524298806,
+      "grad_norm": 0.05249840021133423,
+      "learning_rate": 0.0007604938271604939,
+      "loss": 6.4028,
+      "step": 1042
+    },
+    {
+      "epoch": 2.3171341294084975,
+      "grad_norm": 0.05913037806749344,
+      "learning_rate": 0.0007580246913580247,
+      "loss": 6.0982,
+      "step": 1043
+    },
+    {
+      "epoch": 2.319355734518189,
+      "grad_norm": 0.043238360434770584,
+      "learning_rate": 0.0007555555555555555,
+      "loss": 6.2909,
+      "step": 1044
+    },
+    {
+      "epoch": 2.321577339627881,
+      "grad_norm": 0.06719733029603958,
+      "learning_rate": 0.0007530864197530865,
+      "loss": 6.4409,
+      "step": 1045
+    },
+    {
+      "epoch": 2.3237989447375726,
+      "grad_norm": 0.04194079339504242,
+      "learning_rate": 0.0007506172839506173,
+      "loss": 6.2922,
+      "step": 1046
+    },
+    {
+      "epoch": 2.326020549847265,
+      "grad_norm": 0.05690063163638115,
+      "learning_rate": 0.0007481481481481482,
+      "loss": 5.9464,
+      "step": 1047
+    },
+    {
+      "epoch": 2.3282421549569565,
+      "grad_norm": 0.04586060345172882,
+      "learning_rate": 0.000745679012345679,
+      "loss": 6.215,
+      "step": 1048
+    },
+    {
+      "epoch": 2.3304637600666482,
+      "grad_norm": 0.04375745728611946,
+      "learning_rate": 0.0007432098765432099,
+      "loss": 6.1578,
+      "step": 1049
+    },
+    {
+      "epoch": 2.33268536517634,
+      "grad_norm": 0.04396922513842583,
+      "learning_rate": 0.0007407407407407407,
+      "loss": 6.3354,
+      "step": 1050
+    },
+    {
+      "epoch": 2.3349069702860317,
+      "grad_norm": 0.053402066230773926,
+      "learning_rate": 0.0007382716049382717,
+      "loss": 5.8338,
+      "step": 1051
+    },
+    {
+      "epoch": 2.3371285753957234,
+      "grad_norm": 0.04780908674001694,
+      "learning_rate": 0.0007358024691358025,
+      "loss": 6.1654,
+      "step": 1052
+    },
+    {
+      "epoch": 2.339350180505415,
+      "grad_norm": 0.047660406678915024,
+      "learning_rate": 0.0007333333333333333,
+      "loss": 6.2184,
+      "step": 1053
+    },
+    {
+      "epoch": 2.341571785615107,
+      "grad_norm": 0.04346200451254845,
+      "learning_rate": 0.0007308641975308643,
+      "loss": 6.1987,
+      "step": 1054
+    },
+    {
+      "epoch": 2.3437933907247985,
+      "grad_norm": 0.03536480292677879,
+      "learning_rate": 0.0007283950617283951,
+      "loss": 6.2667,
+      "step": 1055
+    },
+    {
+      "epoch": 2.3460149958344902,
+      "grad_norm": 0.05556477606296539,
+      "learning_rate": 0.000725925925925926,
+      "loss": 5.865,
+      "step": 1056
+    },
+    {
+      "epoch": 2.3482366009441824,
+      "grad_norm": 0.05735522508621216,
+      "learning_rate": 0.0007234567901234568,
+      "loss": 6.2176,
+      "step": 1057
+    },
+    {
+      "epoch": 2.350458206053874,
+      "grad_norm": 0.06092414632439613,
+      "learning_rate": 0.0007209876543209877,
+      "loss": 6.1767,
+      "step": 1058
+    },
+    {
+      "epoch": 2.352679811163566,
+      "grad_norm": 0.05797049030661583,
+      "learning_rate": 0.0007185185185185185,
+      "loss": 6.1351,
+      "step": 1059
+    },
+    {
+      "epoch": 2.3549014162732576,
+      "grad_norm": 0.06134674698114395,
+      "learning_rate": 0.0007160493827160494,
+      "loss": 5.8104,
+      "step": 1060
+    },
+    {
+      "epoch": 2.3571230213829493,
+      "grad_norm": 0.056714169681072235,
+      "learning_rate": 0.0007135802469135802,
+      "loss": 6.2394,
+      "step": 1061
+    },
+    {
+      "epoch": 2.359344626492641,
+      "grad_norm": 0.035053446888923645,
+      "learning_rate": 0.0007111111111111111,
+      "loss": 6.4877,
+      "step": 1062
+    },
+    {
+      "epoch": 2.3615662316023327,
+      "grad_norm": 0.04958026111125946,
+      "learning_rate": 0.000708641975308642,
+      "loss": 6.1811,
+      "step": 1063
+    },
+    {
+      "epoch": 2.3637878367120244,
+      "grad_norm": 0.06432318687438965,
+      "learning_rate": 0.0007061728395061728,
+      "loss": 6.1777,
+      "step": 1064
+    },
+    {
+      "epoch": 2.366009441821716,
+      "grad_norm": 0.05245295539498329,
+      "learning_rate": 0.0007037037037037037,
+      "loss": 6.2577,
+      "step": 1065
+    },
+    {
+      "epoch": 2.368231046931408,
+      "grad_norm": 0.049988944083452225,
+      "learning_rate": 0.0007012345679012346,
+      "loss": 6.2057,
+      "step": 1066
+    },
+    {
+      "epoch": 2.3704526520410996,
+      "grad_norm": 0.05296039208769798,
+      "learning_rate": 0.0006987654320987655,
+      "loss": 6.1838,
+      "step": 1067
+    },
+    {
+      "epoch": 2.3726742571507913,
+      "grad_norm": 0.05535007640719414,
+      "learning_rate": 0.0006962962962962963,
+      "loss": 6.2703,
+      "step": 1068
+    },
+    {
+      "epoch": 2.374895862260483,
+      "grad_norm": 0.05080622807145119,
+      "learning_rate": 0.0006938271604938273,
+      "loss": 6.0225,
+      "step": 1069
+    },
+    {
+      "epoch": 2.3771174673701747,
+      "grad_norm": 0.06902575492858887,
+      "learning_rate": 0.0006913580246913581,
+      "loss": 6.3811,
+      "step": 1070
+    },
+    {
+      "epoch": 2.379339072479867,
+      "grad_norm": 0.05062783136963844,
+      "learning_rate": 0.000688888888888889,
+      "loss": 6.1324,
+      "step": 1071
+    },
+    {
+      "epoch": 2.3815606775895586,
+      "grad_norm": 0.04968835413455963,
+      "learning_rate": 0.0006864197530864198,
+      "loss": 6.2514,
+      "step": 1072
+    },
+    {
+      "epoch": 2.3837822826992503,
+      "grad_norm": 0.05717412009835243,
+      "learning_rate": 0.0006839506172839506,
+      "loss": 6.2891,
+      "step": 1073
+    },
+    {
+      "epoch": 2.386003887808942,
+      "grad_norm": 0.050053175538778305,
+      "learning_rate": 0.0006814814814814815,
+      "loss": 5.7885,
+      "step": 1074
+    },
+    {
+      "epoch": 2.3882254929186337,
+      "grad_norm": 0.06874313950538635,
+      "learning_rate": 0.0006790123456790123,
+      "loss": 6.6177,
+      "step": 1075
+    },
+    {
+      "epoch": 2.3904470980283254,
+      "grad_norm": 0.055651504546403885,
+      "learning_rate": 0.0006765432098765432,
+      "loss": 6.1885,
+      "step": 1076
+    },
+    {
+      "epoch": 2.392668703138017,
+      "grad_norm": 0.04400521516799927,
+      "learning_rate": 0.0006740740740740741,
+      "loss": 6.1594,
+      "step": 1077
+    },
+    {
+      "epoch": 2.394890308247709,
+      "grad_norm": 0.054639481008052826,
+      "learning_rate": 0.000671604938271605,
+      "loss": 6.3318,
+      "step": 1078
+    },
+    {
+      "epoch": 2.3971119133574006,
+      "grad_norm": 0.04897744581103325,
+      "learning_rate": 0.0006691358024691358,
+      "loss": 6.2182,
+      "step": 1079
+    },
+    {
+      "epoch": 2.3993335184670923,
+      "grad_norm": 0.05204634368419647,
+      "learning_rate": 0.0006666666666666666,
+      "loss": 6.3016,
+      "step": 1080
+    },
+    {
+      "epoch": 2.4015551235767845,
+      "grad_norm": 0.07360327988862991,
+      "learning_rate": 0.0006641975308641976,
+      "loss": 6.3416,
+      "step": 1081
+    },
+    {
+      "epoch": 2.403776728686476,
+      "grad_norm": 0.057785168290138245,
+      "learning_rate": 0.0006617283950617284,
+      "loss": 5.7694,
+      "step": 1082
+    },
+    {
+      "epoch": 2.405998333796168,
+      "grad_norm": 0.0587155781686306,
+      "learning_rate": 0.0006592592592592593,
+      "loss": 6.0952,
+      "step": 1083
+    },
+    {
+      "epoch": 2.4082199389058596,
+      "grad_norm": 0.04847797378897667,
+      "learning_rate": 0.0006567901234567901,
+      "loss": 6.158,
+      "step": 1084
+    },
+    {
+      "epoch": 2.4104415440155513,
+      "grad_norm": 0.06603430211544037,
+      "learning_rate": 0.000654320987654321,
+      "loss": 6.328,
+      "step": 1085
+    },
+    {
+      "epoch": 2.412663149125243,
+      "grad_norm": 0.0714314728975296,
+      "learning_rate": 0.0006518518518518518,
+      "loss": 6.2416,
+      "step": 1086
+    },
+    {
+      "epoch": 2.4148847542349348,
+      "grad_norm": 0.06255636364221573,
+      "learning_rate": 0.0006493827160493827,
+      "loss": 6.4735,
+      "step": 1087
+    },
+    {
+      "epoch": 2.4171063593446265,
+      "grad_norm": 0.04875761270523071,
+      "learning_rate": 0.0006469135802469136,
+      "loss": 6.5874,
+      "step": 1088
+    },
+    {
+      "epoch": 2.419327964454318,
+      "grad_norm": 0.044291410595178604,
+      "learning_rate": 0.0006444444444444444,
+      "loss": 6.0333,
+      "step": 1089
+    },
+    {
+      "epoch": 2.42154956956401,
+      "grad_norm": 0.05041727423667908,
+      "learning_rate": 0.0006419753086419753,
+      "loss": 6.0858,
+      "step": 1090
+    },
+    {
+      "epoch": 2.4237711746737016,
+      "grad_norm": 0.05274490267038345,
+      "learning_rate": 0.0006395061728395061,
+      "loss": 6.3765,
+      "step": 1091
+    },
+    {
+      "epoch": 2.4259927797833933,
+      "grad_norm": 0.06735141575336456,
+      "learning_rate": 0.0006370370370370371,
+      "loss": 6.4237,
+      "step": 1092
+    },
+    {
+      "epoch": 2.428214384893085,
+      "grad_norm": 0.046824995428323746,
+      "learning_rate": 0.0006345679012345679,
+      "loss": 6.1497,
+      "step": 1093
+    },
+    {
+      "epoch": 2.4304359900027768,
+      "grad_norm": 0.04495004564523697,
+      "learning_rate": 0.0006320987654320988,
+      "loss": 6.0391,
+      "step": 1094
+    },
+    {
+      "epoch": 2.432657595112469,
+      "grad_norm": 0.05551285296678543,
+      "learning_rate": 0.0006296296296296296,
+      "loss": 6.2817,
+      "step": 1095
+    },
+    {
+      "epoch": 2.4348792002221606,
+      "grad_norm": 0.04686591401696205,
+      "learning_rate": 0.0006271604938271606,
+      "loss": 6.2083,
+      "step": 1096
+    },
+    {
+      "epoch": 2.4371008053318524,
+      "grad_norm": 0.048113562166690826,
+      "learning_rate": 0.0006246913580246914,
+      "loss": 6.4244,
+      "step": 1097
+    },
+    {
+      "epoch": 2.439322410441544,
+      "grad_norm": 0.0601482093334198,
+      "learning_rate": 0.0006222222222222223,
+      "loss": 6.1859,
+      "step": 1098
+    },
+    {
+      "epoch": 2.441544015551236,
+      "grad_norm": 0.05677533149719238,
+      "learning_rate": 0.0006197530864197531,
+      "loss": 6.34,
+      "step": 1099
+    },
+    {
+      "epoch": 2.4437656206609275,
+      "grad_norm": 0.05106941983103752,
+      "learning_rate": 0.0006172839506172839,
+      "loss": 6.0394,
+      "step": 1100
+    },
+    {
+      "epoch": 2.4459872257706192,
+      "grad_norm": 0.05489387735724449,
+      "learning_rate": 0.0006148148148148148,
+      "loss": 6.3246,
+      "step": 1101
+    },
+    {
+      "epoch": 2.448208830880311,
+      "grad_norm": 0.06443698704242706,
+      "learning_rate": 0.0006123456790123456,
+      "loss": 6.4187,
+      "step": 1102
+    },
+    {
+      "epoch": 2.4504304359900027,
+      "grad_norm": 0.045124612748622894,
+      "learning_rate": 0.0006098765432098766,
+      "loss": 6.3125,
+      "step": 1103
+    },
+    {
+      "epoch": 2.4526520410996944,
+      "grad_norm": 0.04628342017531395,
+      "learning_rate": 0.0006074074074074074,
+      "loss": 6.2699,
+      "step": 1104
+    },
+    {
+      "epoch": 2.4548736462093865,
+      "grad_norm": 0.049948979169130325,
+      "learning_rate": 0.0006049382716049383,
+      "loss": 6.1297,
+      "step": 1105
+    },
+    {
+      "epoch": 2.4570952513190782,
+      "grad_norm": 0.05172669515013695,
+      "learning_rate": 0.0006024691358024691,
+      "loss": 5.8902,
+      "step": 1106
+    },
+    {
+      "epoch": 2.45931685642877,
+      "grad_norm": 0.05094560608267784,
+      "learning_rate": 0.0006000000000000001,
+      "loss": 6.1132,
+      "step": 1107
+    },
+    {
+      "epoch": 2.4615384615384617,
+      "grad_norm": 0.05464263632893562,
+      "learning_rate": 0.0005975308641975309,
+      "loss": 6.4178,
+      "step": 1108
+    },
+    {
+      "epoch": 2.4637600666481534,
+      "grad_norm": 0.049421846866607666,
+      "learning_rate": 0.0005950617283950617,
+      "loss": 6.0326,
+      "step": 1109
+    },
+    {
+      "epoch": 2.465981671757845,
+      "grad_norm": 0.0449303537607193,
+      "learning_rate": 0.0005925925925925926,
+      "loss": 6.3347,
+      "step": 1110
+    },
+    {
+      "epoch": 2.468203276867537,
+      "grad_norm": 0.055606018751859665,
+      "learning_rate": 0.0005901234567901235,
+      "loss": 6.0466,
+      "step": 1111
+    },
+    {
+      "epoch": 2.4704248819772285,
+      "grad_norm": 0.0669509619474411,
+      "learning_rate": 0.0005876543209876544,
+      "loss": 6.3029,
+      "step": 1112
+    },
+    {
+      "epoch": 2.4726464870869203,
+      "grad_norm": 0.05385357514023781,
+      "learning_rate": 0.0005851851851851852,
+      "loss": 6.2174,
+      "step": 1113
+    },
+    {
+      "epoch": 2.474868092196612,
+      "grad_norm": 0.06624482572078705,
+      "learning_rate": 0.0005827160493827161,
+      "loss": 6.2003,
+      "step": 1114
+    },
+    {
+      "epoch": 2.4770896973063037,
+      "grad_norm": 0.05494081974029541,
+      "learning_rate": 0.0005802469135802469,
+      "loss": 6.0242,
+      "step": 1115
+    },
+    {
+      "epoch": 2.4793113024159954,
+      "grad_norm": 0.041538678109645844,
+      "learning_rate": 0.0005777777777777778,
+      "loss": 6.1636,
+      "step": 1116
+    },
+    {
+      "epoch": 2.481532907525687,
+      "grad_norm": 0.06696430593729019,
+      "learning_rate": 0.0005753086419753086,
+      "loss": 5.9784,
+      "step": 1117
+    },
+    {
+      "epoch": 2.483754512635379,
+      "grad_norm": 0.04873541742563248,
+      "learning_rate": 0.0005728395061728395,
+      "loss": 6.0676,
+      "step": 1118
+    },
+    {
+      "epoch": 2.485976117745071,
+      "grad_norm": 0.044331517070531845,
+      "learning_rate": 0.0005703703703703704,
+      "loss": 6.3874,
+      "step": 1119
+    },
+    {
+      "epoch": 2.4881977228547627,
+      "grad_norm": 0.04742440581321716,
+      "learning_rate": 0.0005679012345679012,
+      "loss": 6.4032,
+      "step": 1120
+    },
+    {
+      "epoch": 2.4904193279644544,
+      "grad_norm": 0.0625811517238617,
+      "learning_rate": 0.0005654320987654321,
+      "loss": 6.4212,
+      "step": 1121
+    },
+    {
+      "epoch": 2.492640933074146,
+      "grad_norm": 0.05020739510655403,
+      "learning_rate": 0.000562962962962963,
+      "loss": 6.404,
+      "step": 1122
+    },
+    {
+      "epoch": 2.494862538183838,
+      "grad_norm": 0.05465688928961754,
+      "learning_rate": 0.0005604938271604939,
+      "loss": 6.5443,
+      "step": 1123
+    },
+    {
+      "epoch": 2.4970841432935296,
+      "grad_norm": 0.043207861483097076,
+      "learning_rate": 0.0005580246913580247,
+      "loss": 6.2504,
+      "step": 1124
+    },
+    {
+      "epoch": 2.4993057484032213,
+      "grad_norm": 0.06038907542824745,
+      "learning_rate": 0.0005555555555555556,
+      "loss": 6.0477,
+      "step": 1125
+    },
+    {
+      "epoch": 2.501527353512913,
+      "grad_norm": 0.041071973741054535,
+      "learning_rate": 0.0005530864197530865,
+      "loss": 6.1362,
+      "step": 1126
+    },
+    {
+      "epoch": 2.5037489586226047,
+      "grad_norm": 0.05664040148258209,
+      "learning_rate": 0.0005506172839506173,
+      "loss": 6.1586,
+      "step": 1127
+    },
+    {
+      "epoch": 2.5059705637322964,
+      "grad_norm": 0.055108409374952316,
+      "learning_rate": 0.0005481481481481482,
+      "loss": 5.8115,
+      "step": 1128
+    },
+    {
+      "epoch": 2.5081921688419886,
+      "grad_norm": 0.04850602149963379,
+      "learning_rate": 0.000545679012345679,
+      "loss": 5.8996,
+      "step": 1129
+    },
+    {
+      "epoch": 2.5104137739516803,
+      "grad_norm": 0.05887668579816818,
+      "learning_rate": 0.0005432098765432099,
+      "loss": 6.0543,
+      "step": 1130
+    },
+    {
+      "epoch": 2.512635379061372,
+      "grad_norm": 0.06887295842170715,
+      "learning_rate": 0.0005407407407407407,
+      "loss": 5.9565,
+      "step": 1131
+    },
+    {
+      "epoch": 2.5148569841710637,
+      "grad_norm": 0.04778408631682396,
+      "learning_rate": 0.0005382716049382716,
+      "loss": 6.0406,
+      "step": 1132
+    },
+    {
+      "epoch": 2.5170785892807555,
+      "grad_norm": 0.04520206153392792,
+      "learning_rate": 0.0005358024691358024,
+      "loss": 6.0049,
+      "step": 1133
+    },
+    {
+      "epoch": 2.519300194390447,
+      "grad_norm": 0.04291908070445061,
+      "learning_rate": 0.0005333333333333334,
+      "loss": 6.1445,
+      "step": 1134
+    },
+    {
+      "epoch": 2.521521799500139,
+      "grad_norm": 0.06293677538633347,
+      "learning_rate": 0.0005308641975308642,
+      "loss": 6.2649,
+      "step": 1135
+    },
+    {
+      "epoch": 2.5237434046098306,
+      "grad_norm": 0.050695933401584625,
+      "learning_rate": 0.000528395061728395,
+      "loss": 6.3058,
+      "step": 1136
+    },
+    {
+      "epoch": 2.5259650097195223,
+      "grad_norm": 0.05170504003763199,
+      "learning_rate": 0.000525925925925926,
+      "loss": 5.7667,
+      "step": 1137
+    },
+    {
+      "epoch": 2.528186614829214,
+      "grad_norm": 0.04742743819952011,
+      "learning_rate": 0.0005234567901234568,
+      "loss": 6.3765,
+      "step": 1138
+    },
+    {
+      "epoch": 2.5304082199389057,
+      "grad_norm": 0.08276710659265518,
+      "learning_rate": 0.0005209876543209877,
+      "loss": 6.7101,
+      "step": 1139
+    },
+    {
+      "epoch": 2.5326298250485975,
+      "grad_norm": 0.045468445867300034,
+      "learning_rate": 0.0005185185185185185,
+      "loss": 5.9224,
+      "step": 1140
+    },
+    {
+      "epoch": 2.534851430158289,
+      "grad_norm": 0.05360748618841171,
+      "learning_rate": 0.0005160493827160495,
+      "loss": 6.417,
+      "step": 1141
+    },
+    {
+      "epoch": 2.537073035267981,
+      "grad_norm": 0.053457144647836685,
+      "learning_rate": 0.0005135802469135803,
+      "loss": 6.3672,
+      "step": 1142
+    },
+    {
+      "epoch": 2.5392946403776726,
+      "grad_norm": 0.04614447429776192,
+      "learning_rate": 0.0005111111111111112,
+      "loss": 6.5498,
+      "step": 1143
+    },
+    {
+      "epoch": 2.5415162454873648,
+      "grad_norm": 0.047121625393629074,
+      "learning_rate": 0.000508641975308642,
+      "loss": 6.047,
+      "step": 1144
+    },
+    {
+      "epoch": 2.5437378505970565,
+      "grad_norm": 0.045867182314395905,
+      "learning_rate": 0.0005061728395061728,
+      "loss": 6.2337,
+      "step": 1145
+    },
+    {
+      "epoch": 2.545959455706748,
+      "grad_norm": 0.07289320230484009,
+      "learning_rate": 0.0005037037037037037,
+      "loss": 6.4153,
+      "step": 1146
+    },
+    {
+      "epoch": 2.54818106081644,
+      "grad_norm": 0.05743289366364479,
+      "learning_rate": 0.0005012345679012345,
+      "loss": 5.8916,
+      "step": 1147
+    },
+    {
+      "epoch": 2.5504026659261316,
+      "grad_norm": 0.03891833871603012,
+      "learning_rate": 0.0004987654320987654,
+      "loss": 6.2259,
+      "step": 1148
+    },
+    {
+      "epoch": 2.5526242710358233,
+      "grad_norm": 0.058750517666339874,
+      "learning_rate": 0.0004962962962962963,
+      "loss": 6.1005,
+      "step": 1149
+    },
+    {
+      "epoch": 2.554845876145515,
+      "grad_norm": 0.05141089856624603,
+      "learning_rate": 0.0004938271604938272,
+      "loss": 6.1119,
+      "step": 1150
+    },
+    {
+      "epoch": 2.5570674812552068,
+      "grad_norm": 0.05462400242686272,
+      "learning_rate": 0.000491358024691358,
+      "loss": 6.1334,
+      "step": 1151
+    },
+    {
+      "epoch": 2.5592890863648985,
+      "grad_norm": 0.046415410935878754,
+      "learning_rate": 0.000488888888888889,
+      "loss": 6.2168,
+      "step": 1152
+    },
+    {
+      "epoch": 2.5615106914745907,
+      "grad_norm": 0.03819238767027855,
+      "learning_rate": 0.00048641975308641976,
+      "loss": 6.2081,
+      "step": 1153
+    },
+    {
+      "epoch": 2.5637322965842824,
+      "grad_norm": 0.047222115099430084,
+      "learning_rate": 0.0004839506172839506,
+      "loss": 6.1387,
+      "step": 1154
+    },
+    {
+      "epoch": 2.565953901693974,
+      "grad_norm": 0.07276540994644165,
+      "learning_rate": 0.00048148148148148144,
+      "loss": 6.107,
+      "step": 1155
+    },
+    {
+      "epoch": 2.568175506803666,
+      "grad_norm": 0.05205007642507553,
+      "learning_rate": 0.0004790123456790124,
+      "loss": 6.2721,
+      "step": 1156
+    },
+    {
+      "epoch": 2.5703971119133575,
+      "grad_norm": 0.051688313484191895,
+      "learning_rate": 0.00047654320987654324,
+      "loss": 6.3645,
+      "step": 1157
+    },
+    {
+      "epoch": 2.5726187170230492,
+      "grad_norm": 0.07061296701431274,
+      "learning_rate": 0.0004740740740740741,
+      "loss": 6.1971,
+      "step": 1158
+    },
+    {
+      "epoch": 2.574840322132741,
+      "grad_norm": 0.05680854991078377,
+      "learning_rate": 0.0004716049382716049,
+      "loss": 5.8928,
+      "step": 1159
+    },
+    {
+      "epoch": 2.5770619272424327,
+      "grad_norm": 0.04242394119501114,
+      "learning_rate": 0.0004691358024691358,
+      "loss": 6.3864,
+      "step": 1160
+    },
+    {
+      "epoch": 2.5792835323521244,
+      "grad_norm": 0.04473596066236496,
+      "learning_rate": 0.00046666666666666666,
+      "loss": 6.3061,
+      "step": 1161
+    },
+    {
+      "epoch": 2.581505137461816,
+      "grad_norm": 0.0491853766143322,
+      "learning_rate": 0.0004641975308641975,
+      "loss": 6.3205,
+      "step": 1162
+    },
+    {
+      "epoch": 2.583726742571508,
+      "grad_norm": 0.058161403983831406,
+      "learning_rate": 0.00046172839506172835,
+      "loss": 6.3608,
+      "step": 1163
+    },
+    {
+      "epoch": 2.5859483476811995,
+      "grad_norm": 0.06768645346164703,
+      "learning_rate": 0.0004592592592592593,
+      "loss": 6.3228,
+      "step": 1164
+    },
+    {
+      "epoch": 2.5881699527908912,
+      "grad_norm": 0.037545450031757355,
+      "learning_rate": 0.00045679012345679014,
+      "loss": 6.3419,
+      "step": 1165
+    },
+    {
+      "epoch": 2.590391557900583,
+      "grad_norm": 0.050215255469083786,
+      "learning_rate": 0.000454320987654321,
+      "loss": 6.2376,
+      "step": 1166
+    },
+    {
+      "epoch": 2.5926131630102747,
+      "grad_norm": 0.060462694615125656,
+      "learning_rate": 0.0004518518518518519,
+      "loss": 6.5205,
+      "step": 1167
+    },
+    {
+      "epoch": 2.594834768119967,
+      "grad_norm": 0.05351869389414787,
+      "learning_rate": 0.0004493827160493827,
+      "loss": 6.4749,
+      "step": 1168
+    },
+    {
+      "epoch": 2.5970563732296585,
+      "grad_norm": 0.042427029460668564,
+      "learning_rate": 0.00044691358024691357,
+      "loss": 6.2745,
+      "step": 1169
+    },
+    {
+      "epoch": 2.5992779783393503,
+      "grad_norm": 0.050901032984256744,
+      "learning_rate": 0.0004444444444444444,
+      "loss": 6.0122,
+      "step": 1170
+    },
+    {
+      "epoch": 2.601499583449042,
+      "grad_norm": 0.054089177399873734,
+      "learning_rate": 0.00044197530864197536,
+      "loss": 5.9396,
+      "step": 1171
+    },
+    {
+      "epoch": 2.6037211885587337,
+      "grad_norm": 0.06038270145654678,
+      "learning_rate": 0.0004395061728395062,
+      "loss": 6.383,
+      "step": 1172
+    },
+    {
+      "epoch": 2.6059427936684254,
+      "grad_norm": 0.06470409035682678,
+      "learning_rate": 0.00043703703703703705,
+      "loss": 6.0012,
+      "step": 1173
+    },
+    {
+      "epoch": 2.608164398778117,
+      "grad_norm": 0.047819335013628006,
+      "learning_rate": 0.0004345679012345679,
+      "loss": 6.057,
+      "step": 1174
+    },
+    {
+      "epoch": 2.610386003887809,
+      "grad_norm": 0.05305195972323418,
+      "learning_rate": 0.0004320987654320988,
+      "loss": 6.0522,
+      "step": 1175
+    },
+    {
+      "epoch": 2.6126076089975006,
+      "grad_norm": 0.044150613248348236,
+      "learning_rate": 0.00042962962962962963,
+      "loss": 6.1186,
+      "step": 1176
+    },
+    {
+      "epoch": 2.6148292141071927,
+      "grad_norm": 0.055792711675167084,
+      "learning_rate": 0.0004271604938271605,
+      "loss": 5.635,
+      "step": 1177
+    },
+    {
+      "epoch": 2.6170508192168844,
+      "grad_norm": 0.06640937924385071,
+      "learning_rate": 0.0004246913580246913,
+      "loss": 5.8405,
+      "step": 1178
+    },
+    {
+      "epoch": 2.619272424326576,
+      "grad_norm": 0.06097469478845596,
+      "learning_rate": 0.00042222222222222227,
+      "loss": 6.2364,
+      "step": 1179
+    },
+    {
+      "epoch": 2.621494029436268,
+      "grad_norm": 0.05301012843847275,
+      "learning_rate": 0.0004197530864197531,
+      "loss": 6.3979,
+      "step": 1180
+    },
+    {
+      "epoch": 2.6237156345459596,
+      "grad_norm": 0.0558147132396698,
+      "learning_rate": 0.00041728395061728396,
+      "loss": 5.9983,
+      "step": 1181
+    },
+    {
+      "epoch": 2.6259372396556513,
+      "grad_norm": 0.07527412474155426,
+      "learning_rate": 0.0004148148148148148,
+      "loss": 6.277,
+      "step": 1182
+    },
+    {
+      "epoch": 2.628158844765343,
+      "grad_norm": 0.04627913609147072,
+      "learning_rate": 0.0004123456790123457,
+      "loss": 6.4959,
+      "step": 1183
+    },
+    {
+      "epoch": 2.6303804498750347,
+      "grad_norm": 0.06572852283716202,
+      "learning_rate": 0.00040987654320987654,
+      "loss": 5.9408,
+      "step": 1184
+    },
+    {
+      "epoch": 2.6326020549847264,
+      "grad_norm": 0.04756429046392441,
+      "learning_rate": 0.0004074074074074074,
+      "loss": 6.1319,
+      "step": 1185
+    },
+    {
+      "epoch": 2.634823660094418,
+      "grad_norm": 0.047943782061338425,
+      "learning_rate": 0.00040493827160493833,
+      "loss": 6.2338,
+      "step": 1186
+    },
+    {
+      "epoch": 2.63704526520411,
+      "grad_norm": 0.053217049688100815,
+      "learning_rate": 0.0004024691358024692,
+      "loss": 6.3978,
+      "step": 1187
+    },
+    {
+      "epoch": 2.6392668703138016,
+      "grad_norm": 0.041944731026887894,
+      "learning_rate": 0.0004,
+      "loss": 6.0719,
+      "step": 1188
+    },
+    {
+      "epoch": 2.6414884754234933,
+      "grad_norm": 0.08210073411464691,
+      "learning_rate": 0.00039753086419753086,
+      "loss": 5.7097,
+      "step": 1189
+    },
+    {
+      "epoch": 2.643710080533185,
+      "grad_norm": 0.05313239246606827,
+      "learning_rate": 0.00039506172839506176,
+      "loss": 6.1562,
+      "step": 1190
+    },
+    {
+      "epoch": 2.6459316856428767,
+      "grad_norm": 0.06471510976552963,
+      "learning_rate": 0.0003925925925925926,
+      "loss": 6.4404,
+      "step": 1191
+    },
+    {
+      "epoch": 2.648153290752569,
+      "grad_norm": 0.06585077941417694,
+      "learning_rate": 0.00039012345679012345,
+      "loss": 6.4337,
+      "step": 1192
+    },
+    {
+      "epoch": 2.6503748958622606,
+      "grad_norm": 0.041349560022354126,
+      "learning_rate": 0.0003876543209876543,
+      "loss": 6.3703,
+      "step": 1193
+    },
+    {
+      "epoch": 2.6525965009719523,
+      "grad_norm": 0.07425452023744583,
+      "learning_rate": 0.00038518518518518524,
+      "loss": 6.292,
+      "step": 1194
+    },
+    {
+      "epoch": 2.654818106081644,
+      "grad_norm": 0.04713748022913933,
+      "learning_rate": 0.0003827160493827161,
+      "loss": 6.2272,
+      "step": 1195
+    },
+    {
+      "epoch": 2.6570397111913358,
+      "grad_norm": 0.05057023838162422,
+      "learning_rate": 0.0003802469135802469,
+      "loss": 6.3826,
+      "step": 1196
+    },
+    {
+      "epoch": 2.6592613163010275,
+      "grad_norm": 0.0660553053021431,
+      "learning_rate": 0.00037777777777777777,
+      "loss": 6.2613,
+      "step": 1197
+    },
+    {
+      "epoch": 2.661482921410719,
+      "grad_norm": 0.05463806912302971,
+      "learning_rate": 0.00037530864197530867,
+      "loss": 6.2697,
+      "step": 1198
+    },
+    {
+      "epoch": 2.663704526520411,
+      "grad_norm": 0.06074529141187668,
+      "learning_rate": 0.0003728395061728395,
+      "loss": 6.2092,
+      "step": 1199
+    },
+    {
+      "epoch": 2.6659261316301026,
+      "grad_norm": 0.06366332620382309,
+      "learning_rate": 0.00037037037037037035,
+      "loss": 5.9921,
+      "step": 1200
+    },
+    {
+      "epoch": 2.6681477367397948,
+      "grad_norm": 0.08365251123905182,
+      "learning_rate": 0.00036790123456790125,
+      "loss": 6.4445,
+      "step": 1201
+    },
+    {
+      "epoch": 2.6703693418494865,
+      "grad_norm": 0.06650342047214508,
+      "learning_rate": 0.00036543209876543215,
+      "loss": 5.7579,
+      "step": 1202
+    },
+    {
+      "epoch": 2.672590946959178,
+      "grad_norm": 0.06269630044698715,
+      "learning_rate": 0.000362962962962963,
+      "loss": 5.8058,
+      "step": 1203
+    },
+    {
+      "epoch": 2.67481255206887,
+      "grad_norm": 0.04522446542978287,
+      "learning_rate": 0.00036049382716049383,
+      "loss": 6.4158,
+      "step": 1204
+    },
+    {
+      "epoch": 2.6770341571785616,
+      "grad_norm": 0.072049580514431,
+      "learning_rate": 0.0003580246913580247,
+      "loss": 6.2609,
+      "step": 1205
+    },
+    {
+      "epoch": 2.6792557622882534,
+      "grad_norm": 0.056158773601055145,
+      "learning_rate": 0.00035555555555555557,
+      "loss": 6.102,
+      "step": 1206
+    },
+    {
+      "epoch": 2.681477367397945,
+      "grad_norm": 0.06047451123595238,
+      "learning_rate": 0.0003530864197530864,
+      "loss": 6.308,
+      "step": 1207
+    },
+    {
+      "epoch": 2.683698972507637,
+      "grad_norm": 0.04580730199813843,
+      "learning_rate": 0.0003506172839506173,
+      "loss": 6.2663,
+      "step": 1208
+    },
+    {
+      "epoch": 2.6859205776173285,
+      "grad_norm": 0.06703020632266998,
+      "learning_rate": 0.00034814814814814816,
+      "loss": 6.215,
+      "step": 1209
+    },
+    {
+      "epoch": 2.68814218272702,
+      "grad_norm": 0.05186597257852554,
+      "learning_rate": 0.00034567901234567905,
+      "loss": 6.3966,
+      "step": 1210
+    },
+    {
+      "epoch": 2.690363787836712,
+      "grad_norm": 0.04586606100201607,
+      "learning_rate": 0.0003432098765432099,
+      "loss": 6.0347,
+      "step": 1211
+    },
+    {
+      "epoch": 2.6925853929464036,
+      "grad_norm": 0.04665886610746384,
+      "learning_rate": 0.00034074074074074074,
+      "loss": 6.3227,
+      "step": 1212
+    },
+    {
+      "epoch": 2.6948069980560954,
+      "grad_norm": 0.05063016712665558,
+      "learning_rate": 0.0003382716049382716,
+      "loss": 6.0182,
+      "step": 1213
+    },
+    {
+      "epoch": 2.697028603165787,
+      "grad_norm": 0.0436723567545414,
+      "learning_rate": 0.0003358024691358025,
+      "loss": 6.2065,
+      "step": 1214
+    },
+    {
+      "epoch": 2.699250208275479,
+      "grad_norm": 0.06806289404630661,
+      "learning_rate": 0.0003333333333333333,
+      "loss": 6.2903,
+      "step": 1215
+    },
+    {
+      "epoch": 2.701471813385171,
+      "grad_norm": 0.04538436979055405,
+      "learning_rate": 0.0003308641975308642,
+      "loss": 6.2279,
+      "step": 1216
+    },
+    {
+      "epoch": 2.7036934184948627,
+      "grad_norm": 0.0578451082110405,
+      "learning_rate": 0.00032839506172839506,
+      "loss": 5.9165,
+      "step": 1217
+    },
+    {
+      "epoch": 2.7059150236045544,
+      "grad_norm": 0.06210317835211754,
+      "learning_rate": 0.0003259259259259259,
+      "loss": 5.9868,
+      "step": 1218
+    },
+    {
+      "epoch": 2.708136628714246,
+      "grad_norm": 0.054704975336790085,
+      "learning_rate": 0.0003234567901234568,
+      "loss": 6.291,
+      "step": 1219
+    },
+    {
+      "epoch": 2.710358233823938,
+      "grad_norm": 0.05349496006965637,
+      "learning_rate": 0.00032098765432098765,
+      "loss": 6.2186,
+      "step": 1220
+    },
+    {
+      "epoch": 2.7125798389336295,
+      "grad_norm": 0.044549427926540375,
+      "learning_rate": 0.00031851851851851854,
+      "loss": 6.162,
+      "step": 1221
+    },
+    {
+      "epoch": 2.7148014440433212,
+      "grad_norm": 0.058624979108572006,
+      "learning_rate": 0.0003160493827160494,
+      "loss": 5.9312,
+      "step": 1222
+    },
+    {
+      "epoch": 2.717023049153013,
+      "grad_norm": 0.045808497816324234,
+      "learning_rate": 0.0003135802469135803,
+      "loss": 6.0983,
+      "step": 1223
+    },
+    {
+      "epoch": 2.7192446542627047,
+      "grad_norm": 0.0645824670791626,
+      "learning_rate": 0.0003111111111111111,
+      "loss": 6.2318,
+      "step": 1224
+    },
+    {
+      "epoch": 2.7214662593723964,
+      "grad_norm": 0.06571119278669357,
+      "learning_rate": 0.00030864197530864197,
+      "loss": 6.1587,
+      "step": 1225
+    },
+    {
+      "epoch": 2.7236878644820885,
+      "grad_norm": 0.04960980638861656,
+      "learning_rate": 0.0003061728395061728,
+      "loss": 6.3242,
+      "step": 1226
+    },
+    {
+      "epoch": 2.7259094695917803,
+      "grad_norm": 0.05776514858007431,
+      "learning_rate": 0.0003037037037037037,
+      "loss": 6.3115,
+      "step": 1227
+    },
+    {
+      "epoch": 2.728131074701472,
+      "grad_norm": 0.07400120794773102,
+      "learning_rate": 0.00030123456790123455,
+      "loss": 6.3894,
+      "step": 1228
+    },
+    {
+      "epoch": 2.7303526798111637,
+      "grad_norm": 0.04943540692329407,
+      "learning_rate": 0.00029876543209876545,
+      "loss": 6.0005,
+      "step": 1229
+    },
+    {
+      "epoch": 2.7325742849208554,
+      "grad_norm": 0.07010313123464584,
+      "learning_rate": 0.0002962962962962963,
+      "loss": 5.8268,
+      "step": 1230
+    },
+    {
+      "epoch": 2.734795890030547,
+      "grad_norm": 0.047465983778238297,
+      "learning_rate": 0.0002938271604938272,
+      "loss": 6.2481,
+      "step": 1231
+    },
+    {
+      "epoch": 2.737017495140239,
+      "grad_norm": 0.06186727434396744,
+      "learning_rate": 0.00029135802469135803,
+      "loss": 5.8877,
+      "step": 1232
+    },
+    {
+      "epoch": 2.7392391002499306,
+      "grad_norm": 0.061240870505571365,
+      "learning_rate": 0.0002888888888888889,
+      "loss": 5.8979,
+      "step": 1233
+    },
+    {
+      "epoch": 2.7414607053596223,
+      "grad_norm": 0.05173368752002716,
+      "learning_rate": 0.00028641975308641977,
+      "loss": 6.3599,
+      "step": 1234
+    },
+    {
+      "epoch": 2.743682310469314,
+      "grad_norm": 0.07248469442129135,
+      "learning_rate": 0.0002839506172839506,
+      "loss": 6.0223,
+      "step": 1235
+    },
+    {
+      "epoch": 2.7459039155790057,
+      "grad_norm": 0.05211493372917175,
+      "learning_rate": 0.0002814814814814815,
+      "loss": 6.2403,
+      "step": 1236
+    },
+    {
+      "epoch": 2.7481255206886974,
+      "grad_norm": 0.04681209474802017,
+      "learning_rate": 0.00027901234567901236,
+      "loss": 6.3467,
+      "step": 1237
+    },
+    {
+      "epoch": 2.750347125798389,
+      "grad_norm": 0.052145108580589294,
+      "learning_rate": 0.00027654320987654325,
+      "loss": 6.0161,
+      "step": 1238
+    },
+    {
+      "epoch": 2.752568730908081,
+      "grad_norm": 0.05447794869542122,
+      "learning_rate": 0.0002740740740740741,
+      "loss": 6.2148,
+      "step": 1239
+    },
+    {
+      "epoch": 2.7547903360177726,
+      "grad_norm": 0.04743650555610657,
+      "learning_rate": 0.00027160493827160494,
+      "loss": 6.3513,
+      "step": 1240
+    },
+    {
+      "epoch": 2.7570119411274647,
+      "grad_norm": 0.0546528585255146,
+      "learning_rate": 0.0002691358024691358,
+      "loss": 6.1256,
+      "step": 1241
+    },
+    {
+      "epoch": 2.7592335462371564,
+      "grad_norm": 0.07402806729078293,
+      "learning_rate": 0.0002666666666666667,
+      "loss": 6.2216,
+      "step": 1242
+    },
+    {
+      "epoch": 2.761455151346848,
+      "grad_norm": 0.044279150664806366,
+      "learning_rate": 0.0002641975308641975,
+      "loss": 6.1238,
+      "step": 1243
+    },
+    {
+      "epoch": 2.76367675645654,
+      "grad_norm": 0.05609931796789169,
+      "learning_rate": 0.0002617283950617284,
+      "loss": 6.1818,
+      "step": 1244
+    },
+    {
+      "epoch": 2.7658983615662316,
+      "grad_norm": 0.04170677438378334,
+      "learning_rate": 0.00025925925925925926,
+      "loss": 6.288,
+      "step": 1245
+    },
+    {
+      "epoch": 2.7681199666759233,
+      "grad_norm": 0.05834333971142769,
+      "learning_rate": 0.00025679012345679016,
+      "loss": 5.7777,
+      "step": 1246
+    },
+    {
+      "epoch": 2.770341571785615,
+      "grad_norm": 0.061063870787620544,
+      "learning_rate": 0.000254320987654321,
+      "loss": 6.0799,
+      "step": 1247
+    },
+    {
+      "epoch": 2.7725631768953067,
+      "grad_norm": 0.0801524817943573,
+      "learning_rate": 0.00025185185185185185,
+      "loss": 6.1619,
+      "step": 1248
+    },
+    {
+      "epoch": 2.7747847820049985,
+      "grad_norm": 0.04955747723579407,
+      "learning_rate": 0.0002493827160493827,
+      "loss": 6.2996,
+      "step": 1249
+    },
+    {
+      "epoch": 2.7770063871146906,
+      "grad_norm": 0.05656527727842331,
+      "learning_rate": 0.0002469135802469136,
+      "loss": 6.453,
+      "step": 1250
+    },
+    {
+      "epoch": 2.7792279922243823,
+      "grad_norm": 0.04463182017207146,
+      "learning_rate": 0.0002444444444444445,
+      "loss": 6.2475,
+      "step": 1251
+    },
+    {
+      "epoch": 2.781449597334074,
+      "grad_norm": 0.0526060089468956,
+      "learning_rate": 0.0002419753086419753,
+      "loss": 6.4083,
+      "step": 1252
+    },
+    {
+      "epoch": 2.7836712024437658,
+      "grad_norm": 0.05947474390268326,
+      "learning_rate": 0.0002395061728395062,
+      "loss": 6.0038,
+      "step": 1253
+    },
+    {
+      "epoch": 2.7858928075534575,
+      "grad_norm": 0.05380906164646149,
+      "learning_rate": 0.00023703703703703704,
+      "loss": 6.1766,
+      "step": 1254
+    },
+    {
+      "epoch": 2.788114412663149,
+      "grad_norm": 0.04500656947493553,
+      "learning_rate": 0.0002345679012345679,
+      "loss": 5.9869,
+      "step": 1255
+    },
+    {
+      "epoch": 2.790336017772841,
+      "grad_norm": 0.045927878469228745,
+      "learning_rate": 0.00023209876543209875,
+      "loss": 6.1664,
+      "step": 1256
+    },
+    {
+      "epoch": 2.7925576228825326,
+      "grad_norm": 0.05554772913455963,
+      "learning_rate": 0.00022962962962962965,
+      "loss": 6.1204,
+      "step": 1257
+    },
+    {
+      "epoch": 2.7947792279922243,
+      "grad_norm": 0.05797259137034416,
+      "learning_rate": 0.0002271604938271605,
+      "loss": 6.5318,
+      "step": 1258
+    },
+    {
+      "epoch": 2.797000833101916,
+      "grad_norm": 0.045507803559303284,
+      "learning_rate": 0.00022469135802469136,
+      "loss": 6.1506,
+      "step": 1259
+    },
+    {
+      "epoch": 2.7992224382116078,
+      "grad_norm": 0.04965315759181976,
+      "learning_rate": 0.0002222222222222222,
+      "loss": 6.3176,
+      "step": 1260
+    },
+    {
+      "epoch": 2.8014440433212995,
+      "grad_norm": 0.05060115456581116,
+      "learning_rate": 0.0002197530864197531,
+      "loss": 6.3351,
+      "step": 1261
+    },
+    {
+      "epoch": 2.803665648430991,
+      "grad_norm": 0.07024865597486496,
+      "learning_rate": 0.00021728395061728395,
+      "loss": 6.3026,
+      "step": 1262
+    },
+    {
+      "epoch": 2.805887253540683,
+      "grad_norm": 0.05440425127744675,
+      "learning_rate": 0.00021481481481481482,
+      "loss": 5.9646,
+      "step": 1263
+    },
+    {
+      "epoch": 2.8081088586503746,
+      "grad_norm": 0.04848971962928772,
+      "learning_rate": 0.00021234567901234566,
+      "loss": 6.4208,
+      "step": 1264
+    },
+    {
+      "epoch": 2.810330463760067,
+      "grad_norm": 0.07514321804046631,
+      "learning_rate": 0.00020987654320987656,
+      "loss": 6.272,
+      "step": 1265
+    },
+    {
+      "epoch": 2.8125520688697585,
+      "grad_norm": 0.07213981449604034,
+      "learning_rate": 0.0002074074074074074,
+      "loss": 6.2225,
+      "step": 1266
+    },
+    {
+      "epoch": 2.81477367397945,
+      "grad_norm": 0.047682832926511765,
+      "learning_rate": 0.00020493827160493827,
+      "loss": 6.4061,
+      "step": 1267
+    },
+    {
+      "epoch": 2.816995279089142,
+      "grad_norm": 0.04737213999032974,
+      "learning_rate": 0.00020246913580246917,
+      "loss": 6.0628,
+      "step": 1268
+    },
+    {
+      "epoch": 2.8192168841988337,
+      "grad_norm": 0.04613003507256508,
+      "learning_rate": 0.0002,
+      "loss": 6.1632,
+      "step": 1269
+    },
+    {
+      "epoch": 2.8214384893085254,
+      "grad_norm": 0.05441945791244507,
+      "learning_rate": 0.00019753086419753088,
+      "loss": 5.8471,
+      "step": 1270
+    },
+    {
+      "epoch": 2.823660094418217,
+      "grad_norm": 0.04635797068476677,
+      "learning_rate": 0.00019506172839506172,
+      "loss": 6.4254,
+      "step": 1271
+    },
+    {
+      "epoch": 2.825881699527909,
+      "grad_norm": 0.04944876953959465,
+      "learning_rate": 0.00019259259259259262,
+      "loss": 6.2107,
+      "step": 1272
+    },
+    {
+      "epoch": 2.8281033046376005,
+      "grad_norm": 0.052042677998542786,
+      "learning_rate": 0.00019012345679012346,
+      "loss": 6.0596,
+      "step": 1273
+    },
+    {
+      "epoch": 2.8303249097472927,
+      "grad_norm": 0.04883561655879021,
+      "learning_rate": 0.00018765432098765433,
+      "loss": 6.0991,
+      "step": 1274
+    },
+    {
+      "epoch": 2.8325465148569844,
+      "grad_norm": 0.06335494667291641,
+      "learning_rate": 0.00018518518518518518,
+      "loss": 6.2702,
+      "step": 1275
+    },
+    {
+      "epoch": 2.834768119966676,
+      "grad_norm": 0.06728754192590714,
+      "learning_rate": 0.00018271604938271607,
+      "loss": 6.2134,
+      "step": 1276
+    },
+    {
+      "epoch": 2.836989725076368,
+      "grad_norm": 0.04811491444706917,
+      "learning_rate": 0.00018024691358024692,
+      "loss": 6.1782,
+      "step": 1277
+    },
+    {
+      "epoch": 2.8392113301860595,
+      "grad_norm": 0.05488584190607071,
+      "learning_rate": 0.00017777777777777779,
+      "loss": 6.236,
+      "step": 1278
+    },
+    {
+      "epoch": 2.8414329352957512,
+      "grad_norm": 0.061347976326942444,
+      "learning_rate": 0.00017530864197530866,
+      "loss": 6.5296,
+      "step": 1279
+    },
+    {
+      "epoch": 2.843654540405443,
+      "grad_norm": 0.04860737919807434,
+      "learning_rate": 0.00017283950617283953,
+      "loss": 6.2929,
+      "step": 1280
+    },
+    {
+      "epoch": 2.8458761455151347,
+      "grad_norm": 0.03956132382154465,
+      "learning_rate": 0.00017037037037037037,
+      "loss": 6.3365,
+      "step": 1281
+    },
+    {
+      "epoch": 2.8480977506248264,
+      "grad_norm": 0.060977958142757416,
+      "learning_rate": 0.00016790123456790124,
+      "loss": 6.6201,
+      "step": 1282
+    },
+    {
+      "epoch": 2.850319355734518,
+      "grad_norm": 0.05982637405395508,
+      "learning_rate": 0.0001654320987654321,
+      "loss": 6.1179,
+      "step": 1283
+    },
+    {
+      "epoch": 2.85254096084421,
+      "grad_norm": 0.06356212496757507,
+      "learning_rate": 0.00016296296296296295,
+      "loss": 6.1473,
+      "step": 1284
+    },
+    {
+      "epoch": 2.8547625659539015,
+      "grad_norm": 0.05018434673547745,
+      "learning_rate": 0.00016049382716049382,
+      "loss": 6.4789,
+      "step": 1285
+    },
+    {
+      "epoch": 2.8569841710635933,
+      "grad_norm": 0.09307318180799484,
+      "learning_rate": 0.0001580246913580247,
+      "loss": 6.3637,
+      "step": 1286
+    },
+    {
+      "epoch": 2.859205776173285,
+      "grad_norm": 0.06669056415557861,
+      "learning_rate": 0.00015555555555555556,
+      "loss": 6.0945,
+      "step": 1287
+    },
+    {
+      "epoch": 2.8614273812829767,
+      "grad_norm": 0.053704340010881424,
+      "learning_rate": 0.0001530864197530864,
+      "loss": 6.3972,
+      "step": 1288
+    },
+    {
+      "epoch": 2.863648986392669,
+      "grad_norm": 0.04556475579738617,
+      "learning_rate": 0.00015061728395061728,
+      "loss": 6.3623,
+      "step": 1289
+    },
+    {
+      "epoch": 2.8658705915023606,
+      "grad_norm": 0.07496027648448944,
+      "learning_rate": 0.00014814814814814815,
+      "loss": 6.4401,
+      "step": 1290
+    },
+    {
+      "epoch": 2.8680921966120523,
+      "grad_norm": 0.04833129793405533,
+      "learning_rate": 0.00014567901234567902,
+      "loss": 6.4533,
+      "step": 1291
+    },
+    {
+      "epoch": 2.870313801721744,
+      "grad_norm": 0.04120290279388428,
+      "learning_rate": 0.00014320987654320989,
+      "loss": 6.2905,
+      "step": 1292
+    },
+    {
+      "epoch": 2.8725354068314357,
+      "grad_norm": 0.05503775551915169,
+      "learning_rate": 0.00014074074074074076,
+      "loss": 6.2986,
+      "step": 1293
+    },
+    {
+      "epoch": 2.8747570119411274,
+      "grad_norm": 0.0667162612080574,
+      "learning_rate": 0.00013827160493827163,
+      "loss": 6.4405,
+      "step": 1294
+    },
+    {
+      "epoch": 2.876978617050819,
+      "grad_norm": 0.05970746651291847,
+      "learning_rate": 0.00013580246913580247,
+      "loss": 5.9265,
+      "step": 1295
+    },
+    {
+      "epoch": 2.879200222160511,
+      "grad_norm": 0.04566876217722893,
+      "learning_rate": 0.00013333333333333334,
+      "loss": 6.1844,
+      "step": 1296
+    },
+    {
+      "epoch": 2.8814218272702026,
+      "grad_norm": 0.05335947498679161,
+      "learning_rate": 0.0001308641975308642,
+      "loss": 6.1903,
+      "step": 1297
+    },
+    {
+      "epoch": 2.8836434323798947,
+      "grad_norm": 0.0448589064180851,
+      "learning_rate": 0.00012839506172839508,
+      "loss": 6.2494,
+      "step": 1298
+    },
+    {
+      "epoch": 2.8858650374895864,
+      "grad_norm": 0.054932381957769394,
+      "learning_rate": 0.00012592592592592592,
+      "loss": 5.9922,
+      "step": 1299
+    },
+    {
+      "epoch": 2.888086642599278,
+      "grad_norm": 0.05590014159679413,
+      "learning_rate": 0.0001234567901234568,
+      "loss": 5.867,
+      "step": 1300
+    },
+    {
+      "epoch": 2.89030824770897,
+      "grad_norm": 0.052572041749954224,
+      "learning_rate": 0.00012098765432098765,
+      "loss": 6.4757,
+      "step": 1301
+    },
+    {
+      "epoch": 2.8925298528186616,
+      "grad_norm": 0.06353537738323212,
+      "learning_rate": 0.00011851851851851852,
+      "loss": 6.2653,
+      "step": 1302
+    },
+    {
+      "epoch": 2.8947514579283533,
+      "grad_norm": 0.05200917646288872,
+      "learning_rate": 0.00011604938271604938,
+      "loss": 5.9033,
+      "step": 1303
+    },
+    {
+      "epoch": 2.896973063038045,
+      "grad_norm": 0.04855872318148613,
+      "learning_rate": 0.00011358024691358025,
+      "loss": 6.437,
+      "step": 1304
+    },
+    {
+      "epoch": 2.8991946681477367,
+      "grad_norm": 0.07043344527482986,
+      "learning_rate": 0.0001111111111111111,
+      "loss": 6.3508,
+      "step": 1305
+    },
+    {
+      "epoch": 2.9014162732574285,
+      "grad_norm": 0.05130558833479881,
+      "learning_rate": 0.00010864197530864197,
+      "loss": 6.1261,
+      "step": 1306
+    },
+    {
+      "epoch": 2.90363787836712,
+      "grad_norm": 0.049986936151981354,
+      "learning_rate": 0.00010617283950617283,
+      "loss": 6.1206,
+      "step": 1307
+    },
+    {
+      "epoch": 2.905859483476812,
+      "grad_norm": 0.054929640144109726,
+      "learning_rate": 0.0001037037037037037,
+      "loss": 6.3947,
+      "step": 1308
+    },
+    {
+      "epoch": 2.9080810885865036,
+      "grad_norm": 0.06551669538021088,
+      "learning_rate": 0.00010123456790123458,
+      "loss": 5.8056,
+      "step": 1309
+    },
+    {
+      "epoch": 2.9103026936961953,
+      "grad_norm": 0.061959158629179,
+      "learning_rate": 9.876543209876544e-05,
+      "loss": 5.952,
+      "step": 1310
+    },
+    {
+      "epoch": 2.912524298805887,
+      "grad_norm": 0.048053912818431854,
+      "learning_rate": 9.629629629629631e-05,
+      "loss": 6.1003,
+      "step": 1311
+    },
+    {
+      "epoch": 2.9147459039155788,
+      "grad_norm": 0.04767164960503578,
+      "learning_rate": 9.382716049382717e-05,
+      "loss": 6.6017,
+      "step": 1312
+    },
+    {
+      "epoch": 2.916967509025271,
+      "grad_norm": 0.0531221441924572,
+      "learning_rate": 9.135802469135804e-05,
+      "loss": 6.2595,
+      "step": 1313
+    },
+    {
+      "epoch": 2.9191891141349626,
+      "grad_norm": 0.048378121107816696,
+      "learning_rate": 8.888888888888889e-05,
+      "loss": 6.1151,
+      "step": 1314
+    },
+    {
+      "epoch": 2.9214107192446543,
+      "grad_norm": 0.07125979661941528,
+      "learning_rate": 8.641975308641976e-05,
+      "loss": 5.9638,
+      "step": 1315
+    },
+    {
+      "epoch": 2.923632324354346,
+      "grad_norm": 0.05735112354159355,
+      "learning_rate": 8.395061728395062e-05,
+      "loss": 6.006,
+      "step": 1316
+    },
+    {
+      "epoch": 2.9258539294640378,
+      "grad_norm": 0.03954094275832176,
+      "learning_rate": 8.148148148148148e-05,
+      "loss": 6.5584,
+      "step": 1317
+    },
+    {
+      "epoch": 2.9280755345737295,
+      "grad_norm": 0.06284288316965103,
+      "learning_rate": 7.901234567901235e-05,
+      "loss": 6.2671,
+      "step": 1318
+    },
+    {
+      "epoch": 2.930297139683421,
+      "grad_norm": 0.054538823664188385,
+      "learning_rate": 7.65432098765432e-05,
+      "loss": 6.0642,
+      "step": 1319
+    },
+    {
+      "epoch": 2.932518744793113,
+      "grad_norm": 0.04820949211716652,
+      "learning_rate": 7.407407407407407e-05,
+      "loss": 6.2689,
+      "step": 1320
+    },
+    {
+      "epoch": 2.9347403499028046,
+      "grad_norm": 0.04879571497440338,
+      "learning_rate": 7.160493827160494e-05,
+      "loss": 6.4478,
+      "step": 1321
+    },
+    {
+      "epoch": 2.936961955012497,
+      "grad_norm": 0.06558883935213089,
+      "learning_rate": 6.913580246913581e-05,
+      "loss": 6.3775,
+      "step": 1322
+    },
+    {
+      "epoch": 2.9391835601221885,
+      "grad_norm": 0.05510873347520828,
+      "learning_rate": 6.666666666666667e-05,
+      "loss": 6.3437,
+      "step": 1323
+    },
+    {
+      "epoch": 2.9414051652318802,
+      "grad_norm": 0.057643014937639236,
+      "learning_rate": 6.419753086419754e-05,
+      "loss": 6.5145,
+      "step": 1324
+    },
+    {
+      "epoch": 2.943626770341572,
+      "grad_norm": 0.05520697683095932,
+      "learning_rate": 6.17283950617284e-05,
+      "loss": 6.2416,
+      "step": 1325
+    },
+    {
+      "epoch": 2.9458483754512637,
+      "grad_norm": 0.07084875553846359,
+      "learning_rate": 5.925925925925926e-05,
+      "loss": 6.2991,
+      "step": 1326
+    },
+    {
+      "epoch": 2.9480699805609554,
+      "grad_norm": 0.05468638986349106,
+      "learning_rate": 5.679012345679012e-05,
+      "loss": 5.9266,
+      "step": 1327
+    },
+    {
+      "epoch": 2.950291585670647,
+      "grad_norm": 0.044095635414123535,
+      "learning_rate": 5.4320987654320986e-05,
+      "loss": 6.2646,
+      "step": 1328
+    },
+    {
+      "epoch": 2.952513190780339,
+      "grad_norm": 0.058495812118053436,
+      "learning_rate": 5.185185185185185e-05,
+      "loss": 6.2563,
+      "step": 1329
+    },
+    {
+      "epoch": 2.9547347958900305,
+      "grad_norm": 0.052444517612457275,
+      "learning_rate": 4.938271604938272e-05,
+      "loss": 6.0427,
+      "step": 1330
+    },
+    {
+      "epoch": 2.9569564009997222,
+      "grad_norm": 0.06517951190471649,
+      "learning_rate": 4.691358024691358e-05,
+      "loss": 6.1779,
+      "step": 1331
+    },
+    {
+      "epoch": 2.959178006109414,
+      "grad_norm": 0.05335681512951851,
+      "learning_rate": 4.4444444444444447e-05,
+      "loss": 6.2289,
+      "step": 1332
+    },
+    {
+      "epoch": 2.9613996112191057,
+      "grad_norm": 0.053911663591861725,
+      "learning_rate": 4.197530864197531e-05,
+      "loss": 6.4103,
+      "step": 1333
+    },
+    {
+      "epoch": 2.9636212163287974,
+      "grad_norm": 0.048319973051548004,
+      "learning_rate": 3.950617283950617e-05,
+      "loss": 6.3186,
+      "step": 1334
+    },
+    {
+      "epoch": 2.965842821438489,
+      "grad_norm": 0.062335867434740067,
+      "learning_rate": 3.7037037037037037e-05,
+      "loss": 6.2393,
+      "step": 1335
+    },
+    {
+      "epoch": 2.968064426548181,
+      "grad_norm": 0.06816807389259338,
+      "learning_rate": 3.456790123456791e-05,
+      "loss": 6.4822,
+      "step": 1336
+    },
+    {
+      "epoch": 2.970286031657873,
+      "grad_norm": 0.06443248689174652,
+      "learning_rate": 3.209876543209877e-05,
+      "loss": 6.0114,
+      "step": 1337
+    },
+    {
+      "epoch": 2.9725076367675647,
+      "grad_norm": 0.04631428420543671,
+      "learning_rate": 2.962962962962963e-05,
+      "loss": 6.2148,
+      "step": 1338
+    },
+    {
+      "epoch": 2.9747292418772564,
+      "grad_norm": 0.06791841238737106,
+      "learning_rate": 2.7160493827160493e-05,
+      "loss": 6.2634,
+      "step": 1339
+    },
+    {
+      "epoch": 2.976950846986948,
+      "grad_norm": 0.050836220383644104,
+      "learning_rate": 2.469135802469136e-05,
+      "loss": 6.3086,
+      "step": 1340
+    },
+    {
+      "epoch": 2.97917245209664,
+      "grad_norm": 0.04386338219046593,
+      "learning_rate": 2.2222222222222223e-05,
+      "loss": 6.2438,
+      "step": 1341
+    },
+    {
+      "epoch": 2.9813940572063315,
+      "grad_norm": 0.04103785753250122,
+      "learning_rate": 1.9753086419753087e-05,
+      "loss": 6.0759,
+      "step": 1342
+    },
+    {
+      "epoch": 2.9836156623160233,
+      "grad_norm": 0.05627121403813362,
+      "learning_rate": 1.7283950617283953e-05,
+      "loss": 5.9595,
+      "step": 1343
+    },
+    {
+      "epoch": 2.985837267425715,
+      "grad_norm": 0.04390830174088478,
+      "learning_rate": 1.4814814814814815e-05,
+      "loss": 6.2224,
+      "step": 1344
+    },
+    {
+      "epoch": 2.9880588725354067,
+      "grad_norm": 0.0480441190302372,
+      "learning_rate": 1.234567901234568e-05,
+      "loss": 6.4803,
+      "step": 1345
+    },
+    {
+      "epoch": 2.990280477645099,
+      "grad_norm": 0.0458701029419899,
+      "learning_rate": 9.876543209876543e-06,
+      "loss": 6.2263,
+      "step": 1346
+    },
+    {
+      "epoch": 2.9925020827547906,
+      "grad_norm": 0.04264097660779953,
+      "learning_rate": 7.4074074074074075e-06,
+      "loss": 6.0607,
+      "step": 1347
+    },
+    {
+      "epoch": 2.9947236878644823,
+      "grad_norm": 0.04319873824715614,
+      "learning_rate": 4.938271604938272e-06,
+      "loss": 6.2425,
+      "step": 1348
+    },
+    {
+      "epoch": 2.996945292974174,
+      "grad_norm": 0.04556776583194733,
+      "learning_rate": 2.469135802469136e-06,
+      "loss": 6.0128,
+      "step": 1349
+    },
+    {
+      "epoch": 2.9991668980838657,
+      "grad_norm": 0.04006155580282211,
+      "learning_rate": 0.0,
+      "loss": 6.1653,
+      "step": 1350
+    },
+    {
+      "epoch": 2.9991668980838657,
+      "step": 1350,
+      "total_flos": 6.93375979052073e+16,
+      "train_loss": 6.92749930664345,
+      "train_runtime": 1433.2577,
+      "train_samples_per_second": 15.075,
+      "train_steps_per_second": 0.942
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 1350,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 6.93375979052073e+16,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}