diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,7649 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 360,
+  "global_step": 1085,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "grad_norm": 16.129365409595223,
+      "learning_rate": 3.0303030303030305e-07,
+      "loss": 0.9764,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 23.837321667388085,
+      "learning_rate": 6.060606060606061e-07,
+      "loss": 1.1089,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 21.22641145653194,
+      "learning_rate": 9.090909090909091e-07,
+      "loss": 1.146,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 19.43838955090012,
+      "learning_rate": 1.2121212121212122e-06,
+      "loss": 0.9844,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 21.119259211109828,
+      "learning_rate": 1.5151515151515152e-06,
+      "loss": 1.111,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 13.239645949682986,
+      "learning_rate": 1.8181818181818183e-06,
+      "loss": 1.0758,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 16.634917711449923,
+      "learning_rate": 2.1212121212121216e-06,
+      "loss": 1.0476,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 11.236126323459082,
+      "learning_rate": 2.4242424242424244e-06,
+      "loss": 0.9484,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 16.700714724312387,
+      "learning_rate": 2.7272727272727272e-06,
+      "loss": 0.946,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 8.92070999051616,
+      "learning_rate": 3.0303030303030305e-06,
+      "loss": 0.9667,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 8.284153090916707,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 0.8781,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 6.132327487551557,
+      "learning_rate": 3.6363636363636366e-06,
+      "loss": 0.8894,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 6.078573420798111,
+      "learning_rate": 3.93939393939394e-06,
+      "loss": 0.9164,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.8212192573632233,
+      "learning_rate": 4.242424242424243e-06,
+      "loss": 0.8519,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 6.167377075103833,
+      "learning_rate": 4.5454545454545455e-06,
+      "loss": 0.9036,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.6918572965840766,
+      "learning_rate": 4.848484848484849e-06,
+      "loss": 0.8936,
+      "step": 16
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.3749857940241776,
+      "learning_rate": 5.151515151515152e-06,
+      "loss": 0.8125,
+      "step": 17
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 8.242810694053402,
+      "learning_rate": 5.4545454545454545e-06,
+      "loss": 0.8036,
+      "step": 18
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.410736818533579,
+      "learning_rate": 5.7575757575757586e-06,
+      "loss": 0.8736,
+      "step": 19
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.4969703990640535,
+      "learning_rate": 6.060606060606061e-06,
+      "loss": 0.8694,
+      "step": 20
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.0162345460801148,
+      "learning_rate": 6.363636363636364e-06,
+      "loss": 0.8277,
+      "step": 21
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.7821235760367586,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 0.7463,
+      "step": 22
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.984372802199261,
+      "learning_rate": 6.969696969696971e-06,
+      "loss": 0.7771,
+      "step": 23
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.1781858812979578,
+      "learning_rate": 7.272727272727273e-06,
+      "loss": 0.8202,
+      "step": 24
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.4244788380973845,
+      "learning_rate": 7.5757575757575764e-06,
+      "loss": 0.7329,
+      "step": 25
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.323382123238373,
+      "learning_rate": 7.87878787878788e-06,
+      "loss": 0.8099,
+      "step": 26
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.438565494153211,
+      "learning_rate": 8.181818181818183e-06,
+      "loss": 0.6891,
+      "step": 27
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 6.928726222746485,
+      "learning_rate": 8.484848484848486e-06,
+      "loss": 0.7683,
+      "step": 28
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.696562865037612,
+      "learning_rate": 8.787878787878788e-06,
+      "loss": 0.8103,
+      "step": 29
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.0766589777704834,
+      "learning_rate": 9.090909090909091e-06,
+      "loss": 0.7722,
+      "step": 30
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.168856586909847,
+      "learning_rate": 9.393939393939396e-06,
+      "loss": 0.8968,
+      "step": 31
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.2461676550878007,
+      "learning_rate": 9.696969696969698e-06,
+      "loss": 0.808,
+      "step": 32
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.453728879072183,
+      "learning_rate": 1e-05,
+      "loss": 0.8598,
+      "step": 33
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.9113959644302687,
+      "learning_rate": 9.999977704975616e-06,
+      "loss": 0.7454,
+      "step": 34
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.4099367075523803,
+      "learning_rate": 9.999910820101292e-06,
+      "loss": 0.7574,
+      "step": 35
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.270300156314859,
+      "learning_rate": 9.999799345973507e-06,
+      "loss": 0.7398,
+      "step": 36
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.358981799608668,
+      "learning_rate": 9.99964328358639e-06,
+      "loss": 0.7587,
+      "step": 37
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.1653542146382745,
+      "learning_rate": 9.999442634331703e-06,
+      "loss": 0.6518,
+      "step": 38
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.738717692770909,
+      "learning_rate": 9.999197399998842e-06,
+      "loss": 0.7824,
+      "step": 39
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.000494320016642,
+      "learning_rate": 9.998907582774808e-06,
+      "loss": 0.8417,
+      "step": 40
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.452509426088607,
+      "learning_rate": 9.998573185244192e-06,
+      "loss": 0.8029,
+      "step": 41
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.7926851998623228,
+      "learning_rate": 9.998194210389158e-06,
+      "loss": 0.7651,
+      "step": 42
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.026512006063561,
+      "learning_rate": 9.997770661589404e-06,
+      "loss": 0.7327,
+      "step": 43
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.4209663160371155,
+      "learning_rate": 9.997302542622144e-06,
+      "loss": 0.6456,
+      "step": 44
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.641518350587522,
+      "learning_rate": 9.996789857662068e-06,
+      "loss": 0.6919,
+      "step": 45
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 5.599607788159595,
+      "learning_rate": 9.996232611281305e-06,
+      "loss": 0.6614,
+      "step": 46
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 5.715328472466144,
+      "learning_rate": 9.995630808449384e-06,
+      "loss": 0.6826,
+      "step": 47
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.203030177637588,
+      "learning_rate": 9.994984454533185e-06,
+      "loss": 0.6869,
+      "step": 48
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 5.760101944558736,
+      "learning_rate": 9.994293555296905e-06,
+      "loss": 0.605,
+      "step": 49
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 4.853347031830067,
+      "learning_rate": 9.993558116901984e-06,
+      "loss": 0.6368,
+      "step": 50
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 5.1151345822300724,
+      "learning_rate": 9.992778145907073e-06,
+      "loss": 0.6057,
+      "step": 51
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 5.406684758008133,
+      "learning_rate": 9.99195364926796e-06,
+      "loss": 0.6529,
+      "step": 52
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 4.619474763840996,
+      "learning_rate": 9.991084634337512e-06,
+      "loss": 0.6274,
+      "step": 53
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 6.581535039528409,
+      "learning_rate": 9.990171108865615e-06,
+      "loss": 0.5609,
+      "step": 54
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 4.37519555779229,
+      "learning_rate": 9.989213080999097e-06,
+      "loss": 0.5885,
+      "step": 55
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 4.831452306017058,
+      "learning_rate": 9.98821055928166e-06,
+      "loss": 0.6207,
+      "step": 56
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 4.856668685797109,
+      "learning_rate": 9.987163552653802e-06,
+      "loss": 0.5988,
+      "step": 57
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.795835211461076,
+      "learning_rate": 9.986072070452739e-06,
+      "loss": 0.5497,
+      "step": 58
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.5271814227815894,
+      "learning_rate": 9.984936122412319e-06,
+      "loss": 0.5599,
+      "step": 59
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.0934898475040646,
+      "learning_rate": 9.98375571866294e-06,
+      "loss": 0.4616,
+      "step": 60
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.7916913155015934,
+      "learning_rate": 9.982530869731452e-06,
+      "loss": 0.5413,
+      "step": 61
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.56088982742967,
+      "learning_rate": 9.981261586541068e-06,
+      "loss": 0.4671,
+      "step": 62
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.168195444967521,
+      "learning_rate": 9.979947880411274e-06,
+      "loss": 0.5476,
+      "step": 63
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.6048313787874067,
+      "learning_rate": 9.978589763057707e-06,
+      "loss": 0.5249,
+      "step": 64
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.3963291553988126,
+      "learning_rate": 9.977187246592077e-06,
+      "loss": 0.4986,
+      "step": 65
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.3557715502121024,
+      "learning_rate": 9.975740343522033e-06,
+      "loss": 0.5556,
+      "step": 66
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.257861259561017,
+      "learning_rate": 9.974249066751078e-06,
+      "loss": 0.4784,
+      "step": 67
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.3375531169232153,
+      "learning_rate": 9.972713429578427e-06,
+      "loss": 0.3906,
+      "step": 68
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.559723628495009,
+      "learning_rate": 9.971133445698908e-06,
+      "loss": 0.4879,
+      "step": 69
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.194903952303971,
+      "learning_rate": 9.969509129202836e-06,
+      "loss": 0.4214,
+      "step": 70
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.3275730613175782,
+      "learning_rate": 9.96784049457588e-06,
+      "loss": 0.4623,
+      "step": 71
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.5635212966654914,
+      "learning_rate": 9.966127556698937e-06,
+      "loss": 0.5575,
+      "step": 72
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.0097194021841425,
+      "learning_rate": 9.964370330848005e-06,
+      "loss": 0.6115,
+      "step": 73
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.816462672408414,
+      "learning_rate": 9.962568832694046e-06,
+      "loss": 0.534,
+      "step": 74
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.3189782455504777,
+      "learning_rate": 9.960723078302831e-06,
+      "loss": 0.4456,
+      "step": 75
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.0074509816127737,
+      "learning_rate": 9.958833084134821e-06,
+      "loss": 0.4709,
+      "step": 76
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.5611868066341175,
+      "learning_rate": 9.956898867045e-06,
+      "loss": 0.5881,
+      "step": 77
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.700986720821575,
+      "learning_rate": 9.954920444282733e-06,
+      "loss": 0.573,
+      "step": 78
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.4159123698956884,
+      "learning_rate": 9.952897833491617e-06,
+      "loss": 0.5419,
+      "step": 79
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.1081028350711035,
+      "learning_rate": 9.950831052709315e-06,
+      "loss": 0.4993,
+      "step": 80
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.6391399727281084,
+      "learning_rate": 9.948720120367395e-06,
+      "loss": 0.4646,
+      "step": 81
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.2824993832357268,
+      "learning_rate": 9.946565055291175e-06,
+      "loss": 0.4608,
+      "step": 82
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.0836295676643863,
+      "learning_rate": 9.944365876699545e-06,
+      "loss": 0.5675,
+      "step": 83
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.6895916676660425,
+      "learning_rate": 9.942122604204801e-06,
+      "loss": 0.5574,
+      "step": 84
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.674274085195954,
+      "learning_rate": 9.939835257812468e-06,
+      "loss": 0.516,
+      "step": 85
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.5758744418706985,
+      "learning_rate": 9.937503857921126e-06,
+      "loss": 0.4949,
+      "step": 86
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.11345842522007,
+      "learning_rate": 9.93512842532222e-06,
+      "loss": 0.4908,
+      "step": 87
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.070822431047042,
+      "learning_rate": 9.932708981199884e-06,
+      "loss": 0.411,
+      "step": 88
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.4626185376360046,
+      "learning_rate": 9.93024554713074e-06,
+      "loss": 0.4968,
+      "step": 89
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.3899527875013096,
+      "learning_rate": 9.92773814508372e-06,
+      "loss": 0.6052,
+      "step": 90
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.0242500276011803,
+      "learning_rate": 9.925186797419859e-06,
+      "loss": 0.4088,
+      "step": 91
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.254960261668724,
+      "learning_rate": 9.9225915268921e-06,
+      "loss": 0.5032,
+      "step": 92
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.121594536508012,
+      "learning_rate": 9.919952356645093e-06,
+      "loss": 0.4736,
+      "step": 93
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.591853805964485,
+      "learning_rate": 9.917269310214981e-06,
+      "loss": 0.5118,
+      "step": 94
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.975273286527196,
+      "learning_rate": 9.9145424115292e-06,
+      "loss": 0.4194,
+      "step": 95
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.0207999126195797,
+      "learning_rate": 9.911771684906257e-06,
+      "loss": 0.481,
+      "step": 96
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.223652296880912,
+      "learning_rate": 9.908957155055522e-06,
+      "loss": 0.4881,
+      "step": 97
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.145057143621607,
+      "learning_rate": 9.906098847077e-06,
+      "loss": 0.4256,
+      "step": 98
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.1968182998546792,
+      "learning_rate": 9.903196786461106e-06,
+      "loss": 0.4447,
+      "step": 99
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.488996647778079,
+      "learning_rate": 9.900250999088447e-06,
+      "loss": 0.4832,
+      "step": 100
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.7075122583499844,
+      "learning_rate": 9.897261511229584e-06,
+      "loss": 0.5647,
+      "step": 101
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.249140261199728,
+      "learning_rate": 9.894228349544796e-06,
+      "loss": 0.6393,
+      "step": 102
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.138039351395474,
+      "learning_rate": 9.891151541083853e-06,
+      "loss": 0.4877,
+      "step": 103
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.2281784162372515,
+      "learning_rate": 9.888031113285758e-06,
+      "loss": 0.471,
+      "step": 104
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.320283473080627,
+      "learning_rate": 9.88486709397852e-06,
+      "loss": 0.533,
+      "step": 105
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.3375483348489654,
+      "learning_rate": 9.881659511378893e-06,
+      "loss": 0.4609,
+      "step": 106
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.9407782549862607,
+      "learning_rate": 9.87840839409213e-06,
+      "loss": 0.4836,
+      "step": 107
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.233860400341602,
+      "learning_rate": 9.875113771111726e-06,
+      "loss": 0.4909,
+      "step": 108
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.843385330799683,
+      "learning_rate": 9.871775671819162e-06,
+      "loss": 0.4387,
+      "step": 109
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.1589676091186156,
+      "learning_rate": 9.868394125983639e-06,
+      "loss": 0.486,
+      "step": 110
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.040342418056587,
+      "learning_rate": 9.864969163761818e-06,
+      "loss": 0.4788,
+      "step": 111
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.428882692637533,
+      "learning_rate": 9.861500815697541e-06,
+      "loss": 0.5592,
+      "step": 112
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.2835414884007244,
+      "learning_rate": 9.857989112721574e-06,
+      "loss": 0.4444,
+      "step": 113
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.112223960038215,
+      "learning_rate": 9.854434086151318e-06,
+      "loss": 0.4274,
+      "step": 114
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.0412310191062737,
+      "learning_rate": 9.850835767690532e-06,
+      "loss": 0.4685,
+      "step": 115
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.4647322646697654,
+      "learning_rate": 9.847194189429057e-06,
+      "loss": 0.5397,
+      "step": 116
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.7943751391033103,
+      "learning_rate": 9.843509383842524e-06,
+      "loss": 0.4868,
+      "step": 117
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.8448536703371594,
+      "learning_rate": 9.839781383792065e-06,
+      "loss": 0.47,
+      "step": 118
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.912968908916307,
+      "learning_rate": 9.836010222524018e-06,
+      "loss": 0.456,
+      "step": 119
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.1780775293684287,
+      "learning_rate": 9.832195933669639e-06,
+      "loss": 0.4071,
+      "step": 120
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.8028208875094607,
+      "learning_rate": 9.828338551244794e-06,
+      "loss": 0.4125,
+      "step": 121
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.130779457730023,
+      "learning_rate": 9.824438109649654e-06,
+      "loss": 0.5437,
+      "step": 122
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.0007324967598823,
+      "learning_rate": 9.820494643668397e-06,
+      "loss": 0.4136,
+      "step": 123
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.179925551173916,
+      "learning_rate": 9.81650818846889e-06,
+      "loss": 0.3808,
+      "step": 124
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.1852746498130355,
+      "learning_rate": 9.812478779602382e-06,
+      "loss": 0.5346,
+      "step": 125
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.138463505709442,
+      "learning_rate": 9.808406453003177e-06,
+      "loss": 0.4781,
+      "step": 126
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.6209448601330558,
+      "learning_rate": 9.804291244988324e-06,
+      "loss": 0.3635,
+      "step": 127
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.130274052940016,
+      "learning_rate": 9.80013319225729e-06,
+      "loss": 0.477,
+      "step": 128
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.386102342816235,
+      "learning_rate": 9.79593233189163e-06,
+      "loss": 0.475,
+      "step": 129
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.245518577211273,
+      "learning_rate": 9.791688701354656e-06,
+      "loss": 0.4893,
+      "step": 130
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.0216906989251298,
+      "learning_rate": 9.787402338491107e-06,
+      "loss": 0.5008,
+      "step": 131
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.0992634451137686,
+      "learning_rate": 9.78307328152681e-06,
+      "loss": 0.491,
+      "step": 132
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.182492396879651,
+      "learning_rate": 9.778701569068336e-06,
+      "loss": 0.4291,
+      "step": 133
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.322495194500256,
+      "learning_rate": 9.77428724010266e-06,
+      "loss": 0.4719,
+      "step": 134
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.0990527106566095,
+      "learning_rate": 9.76983033399681e-06,
+      "loss": 0.545,
+      "step": 135
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.197449587105137,
+      "learning_rate": 9.765330890497518e-06,
+      "loss": 0.4402,
+      "step": 136
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.1444974057383197,
+      "learning_rate": 9.760788949730866e-06,
+      "loss": 0.4596,
+      "step": 137
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.0082778523434954,
+      "learning_rate": 9.756204552201926e-06,
+      "loss": 0.5243,
+      "step": 138
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.3012157115740797,
+      "learning_rate": 9.751577738794397e-06,
+      "loss": 0.466,
+      "step": 139
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.027306320186225,
+      "learning_rate": 9.746908550770252e-06,
+      "loss": 0.4119,
+      "step": 140
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.2945854233184275,
+      "learning_rate": 9.74219702976935e-06,
+      "loss": 0.5636,
+      "step": 141
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.8562471392647044,
+      "learning_rate": 9.737443217809083e-06,
+      "loss": 0.4643,
+      "step": 142
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.1780745804692625,
+      "learning_rate": 9.732647157283994e-06,
+      "loss": 0.4851,
+      "step": 143
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.987075025431446,
+      "learning_rate": 9.727808890965396e-06,
+      "loss": 0.528,
+      "step": 144
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.9653726410307981,
+      "learning_rate": 9.722928462000995e-06,
+      "loss": 0.4628,
+      "step": 145
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.0367912532832206,
+      "learning_rate": 9.718005913914504e-06,
+      "loss": 0.4967,
+      "step": 146
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.071421018578479,
+      "learning_rate": 9.713041290605255e-06,
+      "loss": 0.4603,
+      "step": 147
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.8900063373062326,
+      "learning_rate": 9.708034636347807e-06,
+      "loss": 0.4436,
+      "step": 148
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.0076364938112627,
+      "learning_rate": 9.702985995791554e-06,
+      "loss": 0.4846,
+      "step": 149
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.6019702541050833,
+      "learning_rate": 9.69789541396032e-06,
+      "loss": 0.5282,
+      "step": 150
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.825342999834098,
+      "learning_rate": 9.69276293625196e-06,
+      "loss": 0.4162,
+      "step": 151
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.7815368946800145,
+      "learning_rate": 9.687588608437963e-06,
+      "loss": 0.5089,
+      "step": 152
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.9315012959714708,
+      "learning_rate": 9.682372476663038e-06,
+      "loss": 0.3959,
+      "step": 153
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.2305953049147274,
+      "learning_rate": 9.677114587444695e-06,
+      "loss": 0.4668,
+      "step": 154
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.741881922031329,
+      "learning_rate": 9.671814987672844e-06,
+      "loss": 0.4607,
+      "step": 155
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.3110309083466585,
+      "learning_rate": 9.666473724609364e-06,
+      "loss": 0.5638,
+      "step": 156
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.0682482956606467,
+      "learning_rate": 9.661090845887693e-06,
+      "loss": 0.4686,
+      "step": 157
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.6193571588443016,
+      "learning_rate": 9.655666399512399e-06,
+      "loss": 0.5315,
+      "step": 158
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.9797453669645546,
+      "learning_rate": 9.650200433858742e-06,
+      "loss": 0.4896,
+      "step": 159
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.4216613612020654,
+      "learning_rate": 9.64469299767226e-06,
+      "loss": 0.3741,
+      "step": 160
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.9693170147881809,
+      "learning_rate": 9.639144140068324e-06,
+      "loss": 0.4272,
+      "step": 161
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.2194466246097613,
+      "learning_rate": 9.633553910531697e-06,
+      "loss": 0.4457,
+      "step": 162
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.201593460499194,
+      "learning_rate": 9.627922358916103e-06,
+      "loss": 0.4102,
+      "step": 163
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.6525412697622233,
+      "learning_rate": 9.622249535443773e-06,
+      "loss": 0.3761,
+      "step": 164
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.2227663902993156,
+      "learning_rate": 9.616535490705003e-06,
+      "loss": 0.4669,
+      "step": 165
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.7035693121877094,
+      "learning_rate": 9.610780275657699e-06,
+      "loss": 0.5599,
+      "step": 166
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.011593619843652,
+      "learning_rate": 9.604983941626924e-06,
+      "loss": 0.4761,
+      "step": 167
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.6713907964698778,
+      "learning_rate": 9.599146540304445e-06,
+      "loss": 0.5458,
+      "step": 168
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.0578837694185688,
+      "learning_rate": 9.59326812374826e-06,
+      "loss": 0.5034,
+      "step": 169
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.009264530645865,
+      "learning_rate": 9.587348744382146e-06,
+      "loss": 0.4794,
+      "step": 170
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.8448925507784202,
+      "learning_rate": 9.581388454995188e-06,
+      "loss": 0.5835,
+      "step": 171
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.7924002279972235,
+      "learning_rate": 9.575387308741301e-06,
+      "loss": 0.5,
+      "step": 172
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.1290643907506412,
+      "learning_rate": 9.569345359138771e-06,
+      "loss": 0.3709,
+      "step": 173
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.218893686302514,
+      "learning_rate": 9.56326266006976e-06,
+      "loss": 0.5468,
+      "step": 174
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.2371127166937597,
+      "learning_rate": 9.557139265779839e-06,
+      "loss": 0.5076,
+      "step": 175
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.2443513250129365,
+      "learning_rate": 9.550975230877495e-06,
+      "loss": 0.3882,
+      "step": 176
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.080847037819759,
+      "learning_rate": 9.544770610333656e-06,
+      "loss": 0.5072,
+      "step": 177
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.063334530889104,
+      "learning_rate": 9.538525459481185e-06,
+      "loss": 0.5129,
+      "step": 178
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.8429299549087172,
+      "learning_rate": 9.5322398340144e-06,
+      "loss": 0.5222,
+      "step": 179
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.5812772666454338,
+      "learning_rate": 9.52591378998857e-06,
+      "loss": 0.5519,
+      "step": 180
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.058285902666481,
+      "learning_rate": 9.519547383819417e-06,
+      "loss": 0.4551,
+      "step": 181
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.8818304058002688,
+      "learning_rate": 9.513140672282613e-06,
+      "loss": 0.4273,
+      "step": 182
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.9209762314258565,
+      "learning_rate": 9.506693712513276e-06,
+      "loss": 0.5037,
+      "step": 183
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.0870128777895767,
+      "learning_rate": 9.500206562005451e-06,
+      "loss": 0.5833,
+      "step": 184
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.2623949551946128,
+      "learning_rate": 9.493679278611616e-06,
+      "loss": 0.5183,
+      "step": 185
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.333429280532828,
+      "learning_rate": 9.487111920542144e-06,
+      "loss": 0.4874,
+      "step": 186
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.815562707725,
+      "learning_rate": 9.480504546364799e-06,
+      "loss": 0.4692,
+      "step": 187
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.7660766038392852,
+      "learning_rate": 9.47385721500421e-06,
+      "loss": 0.5067,
+      "step": 188
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.9680194643495112,
+      "learning_rate": 9.467169985741338e-06,
+      "loss": 0.4563,
+      "step": 189
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.9920282177444568,
+      "learning_rate": 9.460442918212965e-06,
+      "loss": 0.4578,
+      "step": 190
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.241206997432981,
+      "learning_rate": 9.453676072411142e-06,
+      "loss": 0.4639,
+      "step": 191
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.311099672661341,
+      "learning_rate": 9.446869508682666e-06,
+      "loss": 0.5097,
+      "step": 192
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.7661459736458465,
+      "learning_rate": 9.440023287728536e-06,
+      "loss": 0.5023,
+      "step": 193
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.305876852824869,
+      "learning_rate": 9.433137470603424e-06,
+      "loss": 0.5889,
+      "step": 194
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.7528835048955134,
+      "learning_rate": 9.426212118715107e-06,
+      "loss": 0.4261,
+      "step": 195
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.9984846584908784,
+      "learning_rate": 9.419247293823947e-06,
+      "loss": 0.4611,
+      "step": 196
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.2838282643195535,
+      "learning_rate": 9.412243058042316e-06,
+      "loss": 0.5117,
+      "step": 197
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.9369138203254217,
+      "learning_rate": 9.405199473834058e-06,
+      "loss": 0.4893,
+      "step": 198
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.7502665191769333,
+      "learning_rate": 9.398116604013926e-06,
+      "loss": 0.4881,
+      "step": 199
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.8704327603370965,
+      "learning_rate": 9.390994511747021e-06,
+      "loss": 0.4546,
+      "step": 200
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.17508798618509,
+      "learning_rate": 9.383833260548234e-06,
+      "loss": 0.4726,
+      "step": 201
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.2849265007425497,
+      "learning_rate": 9.37663291428167e-06,
+      "loss": 0.4642,
+      "step": 202
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.9841051903862645,
+      "learning_rate": 9.36939353716009e-06,
+      "loss": 0.4622,
+      "step": 203
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.257470318496205,
+      "learning_rate": 9.362115193744329e-06,
+      "loss": 0.4282,
+      "step": 204
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.392014920409948,
+      "learning_rate": 9.35479794894272e-06,
+      "loss": 0.4345,
+      "step": 205
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.8052120410615011,
+      "learning_rate": 9.347441868010531e-06,
+      "loss": 0.4247,
+      "step": 206
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.7695741916190524,
+      "learning_rate": 9.340047016549359e-06,
+      "loss": 0.5194,
+      "step": 207
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.8450154842082274,
+      "learning_rate": 9.332613460506563e-06,
+      "loss": 0.4131,
+      "step": 208
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.8662503542524322,
+      "learning_rate": 9.325141266174666e-06,
+      "loss": 0.4947,
+      "step": 209
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 4.295232461263357,
+      "learning_rate": 9.317630500190774e-06,
+      "loss": 0.5474,
+      "step": 210
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.7297698392367244,
+      "learning_rate": 9.310081229535968e-06,
+      "loss": 0.4592,
+      "step": 211
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.9626826054816364,
+      "learning_rate": 9.302493521534718e-06,
+      "loss": 0.4943,
+      "step": 212
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.7282048031834156,
+      "learning_rate": 9.294867443854278e-06,
+      "loss": 0.4225,
+      "step": 213
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.8569486568399847,
+      "learning_rate": 9.287203064504084e-06,
+      "loss": 0.4129,
+      "step": 214
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.7618256251989388,
+      "learning_rate": 9.279500451835146e-06,
+      "loss": 0.4748,
+      "step": 215
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.3797244120825227,
+      "learning_rate": 9.271759674539437e-06,
+      "loss": 0.5891,
+      "step": 216
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.072223692212461,
+      "learning_rate": 9.263980801649286e-06,
+      "loss": 0.5031,
+      "step": 217
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.050876876872574,
+      "learning_rate": 9.256163902536756e-06,
+      "loss": 0.4236,
+      "step": 218
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.1198932578024876,
+      "learning_rate": 9.248309046913032e-06,
+      "loss": 0.5258,
+      "step": 219
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.146695653453272,
+      "learning_rate": 9.24041630482779e-06,
+      "loss": 0.4794,
+      "step": 220
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.7199343616506693,
+      "learning_rate": 9.232485746668583e-06,
+      "loss": 0.3485,
+      "step": 221
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.286003830086287,
+      "learning_rate": 9.224517443160205e-06,
+      "loss": 0.526,
+      "step": 222
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.7514678642956694,
+      "learning_rate": 9.216511465364066e-06,
+      "loss": 0.5042,
+      "step": 223
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.210433946519186,
+      "learning_rate": 9.208467884677553e-06,
+      "loss": 0.5777,
+      "step": 224
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.8071033204951692,
+      "learning_rate": 9.200386772833394e-06,
+      "loss": 0.4398,
+      "step": 225
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.0356702912458835,
+      "learning_rate": 9.192268201899028e-06,
+      "loss": 0.4432,
+      "step": 226
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.798415887689644,
+      "learning_rate": 9.184112244275949e-06,
+      "loss": 0.4139,
+      "step": 227
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.9228155662900879,
+      "learning_rate": 9.175918972699063e-06,
+      "loss": 0.4344,
+      "step": 228
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.0327204788445496,
+      "learning_rate": 9.167688460236049e-06,
+      "loss": 0.4207,
+      "step": 229
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.5559510625922655,
+      "learning_rate": 9.159420780286699e-06,
+      "loss": 0.3875,
+      "step": 230
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.7831249479009101,
+      "learning_rate": 9.151116006582259e-06,
+      "loss": 0.5297,
+      "step": 231
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.1056676282274136,
+      "learning_rate": 9.142774213184785e-06,
+      "loss": 0.4287,
+      "step": 232
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.7222296800330472,
+      "learning_rate": 9.134395474486471e-06,
+      "loss": 0.494,
+      "step": 233
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.9376342013761003,
+      "learning_rate": 9.12597986520899e-06,
+      "loss": 0.4693,
+      "step": 234
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.4953747502341712,
+      "learning_rate": 9.117527460402826e-06,
+      "loss": 0.4201,
+      "step": 235
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.718097073507747,
+      "learning_rate": 9.109038335446612e-06,
+      "loss": 0.4611,
+      "step": 236
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.8055791887833874,
+      "learning_rate": 9.100512566046445e-06,
+      "loss": 0.5568,
+      "step": 237
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.8902275123309493,
+      "learning_rate": 9.091950228235218e-06,
+      "loss": 0.5016,
+      "step": 238
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.7798148032487224,
+      "learning_rate": 9.083351398371944e-06,
+      "loss": 0.3797,
+      "step": 239
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.9541566954758047,
+      "learning_rate": 9.074716153141074e-06,
+      "loss": 0.4734,
+      "step": 240
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.8373280148490558,
+      "learning_rate": 9.066044569551806e-06,
+      "loss": 0.546,
+      "step": 241
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.439997049046741,
+      "learning_rate": 9.057336724937409e-06,
+      "loss": 0.4715,
+      "step": 242
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.0915262974626607,
+      "learning_rate": 9.048592696954525e-06,
+      "loss": 0.4606,
+      "step": 243
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.8438692284406528,
+      "learning_rate": 9.039812563582481e-06,
+      "loss": 0.4807,
+      "step": 244
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.238913866496149,
+      "learning_rate": 9.030996403122593e-06,
+      "loss": 0.4993,
+      "step": 245
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.727513240522199,
+      "learning_rate": 9.022144294197465e-06,
+      "loss": 0.4872,
+      "step": 246
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.8561137203921283,
+      "learning_rate": 9.013256315750291e-06,
+      "loss": 0.4521,
+      "step": 247
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.9418328007205319,
+      "learning_rate": 9.00433254704415e-06,
+      "loss": 0.4693,
+      "step": 248
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.1530206936727425,
+      "learning_rate": 8.995373067661297e-06,
+      "loss": 0.4198,
+      "step": 249
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.750946114305235,
+      "learning_rate": 8.986377957502459e-06,
+      "loss": 0.5011,
+      "step": 250
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.8983504775289417,
+      "learning_rate": 8.977347296786113e-06,
+      "loss": 0.4639,
+      "step": 251
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.466069503803939,
+      "learning_rate": 8.968281166047782e-06,
+      "loss": 0.509,
+      "step": 252
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.8749710619294966,
+      "learning_rate": 8.959179646139307e-06,
+      "loss": 0.4688,
+      "step": 253
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.5434884020163335,
+      "learning_rate": 8.95004281822813e-06,
+      "loss": 0.4903,
+      "step": 254
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.3537492058556992,
+      "learning_rate": 8.940870763796575e-06,
+      "loss": 0.4347,
+      "step": 255
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.046709463687296,
+      "learning_rate": 8.93166356464111e-06,
+      "loss": 0.4785,
+      "step": 256
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.8894291264166485,
+      "learning_rate": 8.922421302871628e-06,
+      "loss": 0.5522,
+      "step": 257
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.8650094987487535,
+      "learning_rate": 8.91314406091071e-06,
+      "loss": 0.4594,
+      "step": 258
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.189918791153672,
+      "learning_rate": 8.903831921492889e-06,
+      "loss": 0.4493,
+      "step": 259
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.0404519379156523,
+      "learning_rate": 8.894484967663917e-06,
+      "loss": 0.448,
+      "step": 260
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.800269060842294,
+      "learning_rate": 8.885103282780017e-06,
+      "loss": 0.4584,
+      "step": 261
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.006281433815957,
+      "learning_rate": 8.875686950507148e-06,
+      "loss": 0.5058,
+      "step": 262
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.8369002219910917,
+      "learning_rate": 8.866236054820252e-06,
+      "loss": 0.4434,
+      "step": 263
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.224177043935375,
+      "learning_rate": 8.85675068000251e-06,
+      "loss": 0.5084,
+      "step": 264
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.869776027369733,
+      "learning_rate": 8.847230910644586e-06,
+      "loss": 0.3914,
+      "step": 265
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.5520635751992689,
+      "learning_rate": 8.837676831643878e-06,
+      "loss": 0.3972,
+      "step": 266
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.6861043321143212,
+      "learning_rate": 8.828088528203754e-06,
+      "loss": 0.4617,
+      "step": 267
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.7990345859922723,
+      "learning_rate": 8.818466085832797e-06,
+      "loss": 0.4946,
+      "step": 268
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.6043477816641816,
+      "learning_rate": 8.808809590344043e-06,
+      "loss": 0.4647,
+      "step": 269
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.5615958854636416,
+      "learning_rate": 8.799119127854211e-06,
+      "loss": 0.3511,
+      "step": 270
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.2697929179032403,
+      "learning_rate": 8.789394784782945e-06,
+      "loss": 0.4179,
+      "step": 271
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.5996581325163564,
+      "learning_rate": 8.779636647852028e-06,
+      "loss": 0.4466,
+      "step": 272
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.8067082124141383,
+      "learning_rate": 8.76984480408462e-06,
+      "loss": 0.4125,
+      "step": 273
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.1459929653678342,
+      "learning_rate": 8.760019340804478e-06,
+      "loss": 0.5071,
+      "step": 274
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.2872319731726547,
+      "learning_rate": 8.750160345635183e-06,
+      "loss": 0.4584,
+      "step": 275
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.7407350134278134,
+      "learning_rate": 8.740267906499347e-06,
+      "loss": 0.4991,
+      "step": 276
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.609979578565858,
+      "learning_rate": 8.73034211161784e-06,
+      "loss": 0.5205,
+      "step": 277
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.200677272547663,
+      "learning_rate": 8.720383049508998e-06,
+      "loss": 0.5487,
+      "step": 278
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.6432820372618948,
+      "learning_rate": 8.710390808987834e-06,
+      "loss": 0.4042,
+      "step": 279
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.7156721753241313,
+      "learning_rate": 8.700365479165246e-06,
+      "loss": 0.3996,
+      "step": 280
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.9082538060806855,
+      "learning_rate": 8.690307149447222e-06,
+      "loss": 0.4812,
+      "step": 281
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.63902982998481,
+      "learning_rate": 8.680215909534044e-06,
+      "loss": 0.446,
+      "step": 282
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.587339989963004,
+      "learning_rate": 8.67009184941949e-06,
+      "loss": 0.3901,
+      "step": 283
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.6319010966753653,
+      "learning_rate": 8.659935059390029e-06,
+      "loss": 0.4277,
+      "step": 284
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.7250865003847689,
+      "learning_rate": 8.64974563002401e-06,
+      "loss": 0.3561,
+      "step": 285
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.7173033701439209,
+      "learning_rate": 8.639523652190863e-06,
+      "loss": 0.3946,
+      "step": 286
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.0533966512825748,
+      "learning_rate": 8.62926921705029e-06,
+      "loss": 0.4255,
+      "step": 287
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.9264061742598322,
+      "learning_rate": 8.618982416051437e-06,
+      "loss": 0.4631,
+      "step": 288
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.0607639449740534,
+      "learning_rate": 8.608663340932105e-06,
+      "loss": 0.4806,
+      "step": 289
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.9707326291467142,
+      "learning_rate": 8.598312083717897e-06,
+      "loss": 0.4838,
+      "step": 290
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.8760820163547305,
+      "learning_rate": 8.587928736721432e-06,
+      "loss": 0.5108,
+      "step": 291
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.7862721737223892,
+      "learning_rate": 8.577513392541496e-06,
+      "loss": 0.3962,
+      "step": 292
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.9882280047781893,
+      "learning_rate": 8.567066144062232e-06,
+      "loss": 0.5171,
+      "step": 293
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.0639143960233897,
+      "learning_rate": 8.556587084452305e-06,
+      "loss": 0.5001,
+      "step": 294
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.8782269179129345,
+      "learning_rate": 8.546076307164068e-06,
+      "loss": 0.4568,
+      "step": 295
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.6079291594479586,
+      "learning_rate": 8.535533905932739e-06,
+      "loss": 0.4282,
+      "step": 296
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.0434150812546203,
+      "learning_rate": 8.524959974775551e-06,
+      "loss": 0.4884,
+      "step": 297
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.9071711134863052,
+      "learning_rate": 8.514354607990927e-06,
+      "loss": 0.4573,
+      "step": 298
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.8196098537969094,
+      "learning_rate": 8.503717900157632e-06,
+      "loss": 0.4788,
+      "step": 299
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.9075527663018237,
+      "learning_rate": 8.49304994613393e-06,
+      "loss": 0.5259,
+      "step": 300
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.6611505782230385,
+      "learning_rate": 8.482350841056739e-06,
+      "loss": 0.4273,
+      "step": 301
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.8747309724134273,
+      "learning_rate": 8.47162068034078e-06,
+      "loss": 0.4702,
+      "step": 302
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.9210086107236377,
+      "learning_rate": 8.460859559677735e-06,
+      "loss": 0.4309,
+      "step": 303
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.7431646203338342,
+      "learning_rate": 8.45006757503538e-06,
+      "loss": 0.4613,
+      "step": 304
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.5052628264537502,
+      "learning_rate": 8.43924482265674e-06,
+      "loss": 0.3804,
+      "step": 305
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.9158138263968971,
+      "learning_rate": 8.428391399059228e-06,
+      "loss": 0.5001,
+      "step": 306
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.480815467739981,
+      "learning_rate": 8.41750740103378e-06,
+      "loss": 0.5295,
+      "step": 307
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.003610118605416,
+      "learning_rate": 8.406592925643995e-06,
+      "loss": 0.4291,
+      "step": 308
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.8511717607117526,
+      "learning_rate": 8.395648070225273e-06,
+      "loss": 0.4799,
+      "step": 309
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.5701507442392661,
+      "learning_rate": 8.384672932383942e-06,
+      "loss": 0.4526,
+      "step": 310
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.5948064665545478,
+      "learning_rate": 8.373667609996388e-06,
+      "loss": 0.3488,
+      "step": 311
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.7172848929373001,
+      "learning_rate": 8.362632201208182e-06,
+      "loss": 0.5191,
+      "step": 312
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.861421603347419,
+      "learning_rate": 8.351566804433207e-06,
+      "loss": 0.4473,
+      "step": 313
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.8015561005409069,
+      "learning_rate": 8.340471518352781e-06,
+      "loss": 0.4527,
+      "step": 314
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.1556575012116967,
+      "learning_rate": 8.329346441914774e-06,
+      "loss": 0.5484,
+      "step": 315
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.0076914952446954,
+      "learning_rate": 8.318191674332724e-06,
+      "loss": 0.4091,
+      "step": 316
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.867497503681997,
+      "learning_rate": 8.307007315084958e-06,
+      "loss": 0.4939,
+      "step": 317
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.1262537928980025,
+      "learning_rate": 8.295793463913701e-06,
+      "loss": 0.5834,
+      "step": 318
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.7637175217775476,
+      "learning_rate": 8.284550220824186e-06,
+      "loss": 0.4526,
+      "step": 319
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.648339289988361,
+      "learning_rate": 8.273277686083768e-06,
+      "loss": 0.3449,
+      "step": 320
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.3992238626915308,
+      "learning_rate": 8.261975960221016e-06,
+      "loss": 0.4495,
+      "step": 321
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.9018619117764128,
+      "learning_rate": 8.250645144024837e-06,
+      "loss": 0.3955,
+      "step": 322
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.9522016280115226,
+      "learning_rate": 8.23928533854356e-06,
+      "loss": 0.5687,
+      "step": 323
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.8534830913654115,
+      "learning_rate": 8.227896645084037e-06,
+      "loss": 0.4745,
+      "step": 324
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.9964611419849239,
+      "learning_rate": 8.216479165210748e-06,
+      "loss": 0.4019,
+      "step": 325
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.068111314185152,
+      "learning_rate": 8.205033000744895e-06,
+      "loss": 0.5143,
+      "step": 326
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.5590845818504766,
+      "learning_rate": 8.193558253763479e-06,
+      "loss": 0.3693,
+      "step": 327
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.186509835454746,
+      "learning_rate": 8.182055026598408e-06,
+      "loss": 0.416,
+      "step": 328
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.5844801019590957,
+      "learning_rate": 8.170523421835573e-06,
+      "loss": 0.3545,
+      "step": 329
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.8575577226281546,
+      "learning_rate": 8.158963542313937e-06,
+      "loss": 0.4761,
+      "step": 330
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.7966713390914204,
+      "learning_rate": 8.14737549112462e-06,
+      "loss": 0.3864,
+      "step": 331
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.8380592128499222,
+      "learning_rate": 8.135759371609977e-06,
+      "loss": 0.4845,
+      "step": 332
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.0455274179399447,
+      "learning_rate": 8.12411528736267e-06,
+      "loss": 0.5534,
+      "step": 333
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.7294266023913523,
+      "learning_rate": 8.112443342224761e-06,
+      "loss": 0.4138,
+      "step": 334
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.0058011955497626,
+      "learning_rate": 8.100743640286768e-06,
+      "loss": 0.4999,
+      "step": 335
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.651553365904595,
+      "learning_rate": 8.089016285886747e-06,
+      "loss": 0.4881,
+      "step": 336
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.773227004412968,
+      "learning_rate": 8.077261383609363e-06,
+      "loss": 0.4598,
+      "step": 337
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.8790139684478144,
+      "learning_rate": 8.065479038284942e-06,
+      "loss": 0.4663,
+      "step": 338
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.919118851556668,
+      "learning_rate": 8.05366935498856e-06,
+      "loss": 0.4419,
+      "step": 339
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.964652133274893,
+      "learning_rate": 8.041832439039085e-06,
+      "loss": 0.4673,
+      "step": 340
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.7974809835670555,
+      "learning_rate": 8.02996839599825e-06,
+      "loss": 0.4727,
+      "step": 341
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.8553171988918498,
+      "learning_rate": 8.018077331669707e-06,
+      "loss": 0.4163,
+      "step": 342
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.9389312069845352,
+      "learning_rate": 8.006159352098082e-06,
+      "loss": 0.4334,
+      "step": 343
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.6346055321688229,
+      "learning_rate": 7.994214563568036e-06,
+      "loss": 0.4002,
+      "step": 344
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.599227362292106,
+      "learning_rate": 7.982243072603306e-06,
+      "loss": 0.4076,
+      "step": 345
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.7368614934087998,
+      "learning_rate": 7.970244985965767e-06,
+      "loss": 0.5056,
+      "step": 346
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.591309812293656,
+      "learning_rate": 7.958220410654474e-06,
+      "loss": 0.3454,
+      "step": 347
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.9143496936902584,
+      "learning_rate": 7.946169453904706e-06,
+      "loss": 0.4637,
+      "step": 348
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.9965493738056397,
+      "learning_rate": 7.93409222318701e-06,
+      "loss": 0.4425,
+      "step": 349
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.5549050181218969,
+      "learning_rate": 7.921988826206252e-06,
+      "loss": 0.4003,
+      "step": 350
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.0576285352992456,
+      "learning_rate": 7.909859370900642e-06,
+      "loss": 0.4399,
+      "step": 351
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.8356947778398593,
+      "learning_rate": 7.89770396544078e-06,
+      "loss": 0.487,
+      "step": 352
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.8974007514954563,
+      "learning_rate": 7.88552271822869e-06,
+      "loss": 0.4701,
+      "step": 353
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.3226326302090365,
+      "learning_rate": 7.873315737896856e-06,
+      "loss": 0.4546,
+      "step": 354
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.9787830221538376,
+      "learning_rate": 7.861083133307248e-06,
+      "loss": 0.5219,
+      "step": 355
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.8025367170940212,
+      "learning_rate": 7.84882501355035e-06,
+      "loss": 0.5475,
+      "step": 356
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.9086914362763832,
+      "learning_rate": 7.836541487944194e-06,
+      "loss": 0.5392,
+      "step": 357
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.0101191131901044,
+      "learning_rate": 7.824232666033382e-06,
+      "loss": 0.4568,
+      "step": 358
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.6340523028668645,
+      "learning_rate": 7.81189865758811e-06,
+      "loss": 0.4964,
+      "step": 359
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.2283404114277596,
+      "learning_rate": 7.79953957260318e-06,
+      "loss": 0.5183,
+      "step": 360
+    },
+    {
+      "epoch": 0.33,
+      "eval_loss": 0.4910432696342468,
+      "eval_runtime": 594.926,
+      "eval_samples_per_second": 4.555,
+      "eval_steps_per_second": 0.286,
+      "step": 360
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.123698396493111,
+      "learning_rate": 7.78715552129704e-06,
+      "loss": 0.4642,
+      "step": 361
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.9212714053414477,
+      "learning_rate": 7.774746614110775e-06,
+      "loss": 0.4842,
+      "step": 362
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.7372173704119815,
+      "learning_rate": 7.762312961707142e-06,
+      "loss": 0.4649,
+      "step": 363
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.077334722606827,
+      "learning_rate": 7.749854674969572e-06,
+      "loss": 0.4999,
+      "step": 364
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.970176276378124,
+      "learning_rate": 7.737371865001191e-06,
+      "loss": 0.4424,
+      "step": 365
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.989426659221798,
+      "learning_rate": 7.72486464312382e-06,
+      "loss": 0.4116,
+      "step": 366
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.9514397016000429,
+      "learning_rate": 7.712333120876983e-06,
+      "loss": 0.4875,
+      "step": 367
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.6795737086731901,
+      "learning_rate": 7.699777410016919e-06,
+      "loss": 0.4681,
+      "step": 368
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.7615296207530442,
+      "learning_rate": 7.68719762251558e-06,
+      "loss": 0.4944,
+      "step": 369
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.107061451712905,
+      "learning_rate": 7.674593870559635e-06,
+      "loss": 0.4358,
+      "step": 370
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.6660235962385783,
+      "learning_rate": 7.661966266549463e-06,
+      "loss": 0.4869,
+      "step": 371
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.9951185485725758,
+      "learning_rate": 7.649314923098164e-06,
+      "loss": 0.4729,
+      "step": 372
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.9058474675851362,
+      "learning_rate": 7.636639953030541e-06,
+      "loss": 0.4598,
+      "step": 373
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.88590824247099,
+      "learning_rate": 7.623941469382099e-06,
+      "loss": 0.3653,
+      "step": 374
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.6804952330361298,
+      "learning_rate": 7.6112195853980405e-06,
+      "loss": 0.4879,
+      "step": 375
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.822142130579958,
+      "learning_rate": 7.598474414532252e-06,
+      "loss": 0.4422,
+      "step": 376
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.3799371329130317,
+      "learning_rate": 7.585706070446288e-06,
+      "loss": 0.4564,
+      "step": 377
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.770691949972335,
+      "learning_rate": 7.572914667008371e-06,
+      "loss": 0.4476,
+      "step": 378
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.1595484976714605,
+      "learning_rate": 7.560100318292356e-06,
+      "loss": 0.3401,
+      "step": 379
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.5977458132598084,
+      "learning_rate": 7.5472631385767325e-06,
+      "loss": 0.3351,
+      "step": 380
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.7747990571416914,
+      "learning_rate": 7.5344032423435955e-06,
+      "loss": 0.3917,
+      "step": 381
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.8120059216772098,
+      "learning_rate": 7.521520744277624e-06,
+      "loss": 0.3841,
+      "step": 382
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.6871158947809795,
+      "learning_rate": 7.508615759265059e-06,
+      "loss": 0.4564,
+      "step": 383
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.6166435032638933,
+      "learning_rate": 7.495688402392687e-06,
+      "loss": 0.4139,
+      "step": 384
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.9846131718538027,
+      "learning_rate": 7.4827387889468e-06,
+      "loss": 0.5126,
+      "step": 385
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.5748881218512112,
+      "learning_rate": 7.469767034412176e-06,
+      "loss": 0.3554,
+      "step": 386
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.8364294972416482,
+      "learning_rate": 7.456773254471053e-06,
+      "loss": 0.4038,
+      "step": 387
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.5067889157844772,
+      "learning_rate": 7.443757565002081e-06,
+      "loss": 0.4199,
+      "step": 388
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.7982314184205184,
+      "learning_rate": 7.4307200820793116e-06,
+      "loss": 0.5696,
+      "step": 389
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.8564950286038326,
+      "learning_rate": 7.417660921971141e-06,
+      "loss": 0.4031,
+      "step": 390
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.7423105060855828,
+      "learning_rate": 7.404580201139287e-06,
+      "loss": 0.364,
+      "step": 391
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.7521782559338535,
+      "learning_rate": 7.391478036237747e-06,
+      "loss": 0.4695,
+      "step": 392
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.7234339577319666,
+      "learning_rate": 7.378354544111754e-06,
+      "loss": 0.3918,
+      "step": 393
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.8347981133237585,
+      "learning_rate": 7.3652098417967385e-06,
+      "loss": 0.4166,
+      "step": 394
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.9133644782594936,
+      "learning_rate": 7.352044046517286e-06,
+      "loss": 0.3654,
+      "step": 395
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.8683261855992503,
+      "learning_rate": 7.338857275686084e-06,
+      "loss": 0.4664,
+      "step": 396
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.7767493778615355,
+      "learning_rate": 7.325649646902887e-06,
+      "loss": 0.5069,
+      "step": 397
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.5207608762075067,
+      "learning_rate": 7.312421277953455e-06,
+      "loss": 0.4519,
+      "step": 398
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.558289509724299,
+      "learning_rate": 7.2991722868085116e-06,
+      "loss": 0.4433,
+      "step": 399
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.6758395841551788,
+      "learning_rate": 7.285902791622689e-06,
+      "loss": 0.4029,
+      "step": 400
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.7967561396799412,
+      "learning_rate": 7.2726129107334756e-06,
+      "loss": 0.3314,
+      "step": 401
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.721033476704588,
+      "learning_rate": 7.259302762660158e-06,
+      "loss": 0.4638,
+      "step": 402
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.296407197054947,
+      "learning_rate": 7.245972466102767e-06,
+      "loss": 0.4548,
+      "step": 403
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.795981307210963,
+      "learning_rate": 7.2326221399410154e-06,
+      "loss": 0.4076,
+      "step": 404
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.8949401637558156,
+      "learning_rate": 7.219251903233245e-06,
+      "loss": 0.4093,
+      "step": 405
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.73684976879178,
+      "learning_rate": 7.205861875215356e-06,
+      "loss": 0.3819,
+      "step": 406
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.7843482784746565,
+      "learning_rate": 7.192452175299749e-06,
+      "loss": 0.3852,
+      "step": 407
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.515979867371627,
+      "learning_rate": 7.179022923074258e-06,
+      "loss": 0.3516,
+      "step": 408
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.6687889356941592,
+      "learning_rate": 7.165574238301085e-06,
+      "loss": 0.5291,
+      "step": 409
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.318380976106726,
+      "learning_rate": 7.152106240915735e-06,
+      "loss": 0.5291,
+      "step": 410
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.6326669832522336,
+      "learning_rate": 7.138619051025936e-06,
+      "loss": 0.3583,
+      "step": 411
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.8202220733844416,
+      "learning_rate": 7.125112788910581e-06,
+      "loss": 0.4199,
+      "step": 412
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.6587100069818803,
+      "learning_rate": 7.1115875750186484e-06,
+      "loss": 0.4062,
+      "step": 413
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.0834531329912767,
+      "learning_rate": 7.098043529968124e-06,
+      "loss": 0.4267,
+      "step": 414
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.9123822757520978,
+      "learning_rate": 7.084480774544937e-06,
+      "loss": 0.5463,
+      "step": 415
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.5790511832425012,
+      "learning_rate": 7.0708994297018725e-06,
+      "loss": 0.4292,
+      "step": 416
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.609041808151019,
+      "learning_rate": 7.057299616557494e-06,
+      "loss": 0.4687,
+      "step": 417
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.0643914858474877,
+      "learning_rate": 7.0436814563950685e-06,
+      "loss": 0.4086,
+      "step": 418
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.6647393613676398,
+      "learning_rate": 7.030045070661484e-06,
+      "loss": 0.4021,
+      "step": 419
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.8392635961659551,
+      "learning_rate": 7.016390580966157e-06,
+      "loss": 0.4092,
+      "step": 420
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.5961711152340379,
+      "learning_rate": 7.0027181090799635e-06,
+      "loss": 0.4444,
+      "step": 421
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.9508635361928839,
+      "learning_rate": 6.989027776934138e-06,
+      "loss": 0.4403,
+      "step": 422
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.6913630397691195,
+      "learning_rate": 6.975319706619197e-06,
+      "loss": 0.3893,
+      "step": 423
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.7460218203007862,
+      "learning_rate": 6.9615940203838485e-06,
+      "loss": 0.4421,
+      "step": 424
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.733368109080223,
+      "learning_rate": 6.947850840633892e-06,
+      "loss": 0.4468,
+      "step": 425
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.8378497057178176,
+      "learning_rate": 6.93409028993114e-06,
+      "loss": 0.4311,
+      "step": 426
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.52179250865843,
+      "learning_rate": 6.92031249099232e-06,
+      "loss": 0.3415,
+      "step": 427
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.6773803379432295,
+      "learning_rate": 6.906517566687974e-06,
+      "loss": 0.4516,
+      "step": 428
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.6271440674538975,
+      "learning_rate": 6.8927056400413735e-06,
+      "loss": 0.3851,
+      "step": 429
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.703735768216468,
+      "learning_rate": 6.878876834227414e-06,
+      "loss": 0.4994,
+      "step": 430
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.7864526474999918,
+      "learning_rate": 6.8650312725715205e-06,
+      "loss": 0.462,
+      "step": 431
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.7484281417044363,
+      "learning_rate": 6.85116907854855e-06,
+      "loss": 0.4013,
+      "step": 432
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.878619986993286,
+      "learning_rate": 6.8372903757816785e-06,
+      "loss": 0.3441,
+      "step": 433
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.7713170367925475,
+      "learning_rate": 6.823395288041315e-06,
+      "loss": 0.3744,
+      "step": 434
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.6816840368792083,
+      "learning_rate": 6.809483939243991e-06,
+      "loss": 0.426,
+      "step": 435
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.6874406652983192,
+      "learning_rate": 6.795556453451247e-06,
+      "loss": 0.4653,
+      "step": 436
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.8669154711798979,
+      "learning_rate": 6.781612954868538e-06,
+      "loss": 0.4462,
+      "step": 437
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.6552135176054872,
+      "learning_rate": 6.767653567844121e-06,
+      "loss": 0.406,
+      "step": 438
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.86645573903988,
+      "learning_rate": 6.7536784168679435e-06,
+      "loss": 0.397,
+      "step": 439
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.790952614026057,
+      "learning_rate": 6.739687626570542e-06,
+      "loss": 0.2901,
+      "step": 440
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.9158821336195229,
+      "learning_rate": 6.725681321721916e-06,
+      "loss": 0.4875,
+      "step": 441
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.716712729279155,
+      "learning_rate": 6.711659627230431e-06,
+      "loss": 0.4671,
+      "step": 442
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.358168373370002,
+      "learning_rate": 6.697622668141699e-06,
+      "loss": 0.4856,
+      "step": 443
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.5652289238354264,
+      "learning_rate": 6.683570569637452e-06,
+      "loss": 0.3989,
+      "step": 444
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.2252972642864846,
+      "learning_rate": 6.669503457034445e-06,
+      "loss": 0.4992,
+      "step": 445
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.3986600415958652,
+      "learning_rate": 6.6554214557833245e-06,
+      "loss": 0.4299,
+      "step": 446
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.7075847740067474,
+      "learning_rate": 6.641324691467514e-06,
+      "loss": 0.4358,
+      "step": 447
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.389187659662136,
+      "learning_rate": 6.627213289802098e-06,
+      "loss": 0.3526,
+      "step": 448
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.756357024413377,
+      "learning_rate": 6.613087376632691e-06,
+      "loss": 0.4007,
+      "step": 449
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.0007846755874965,
+      "learning_rate": 6.5989470779343265e-06,
+      "loss": 0.5168,
+      "step": 450
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.5519088968464894,
+      "learning_rate": 6.584792519810326e-06,
+      "loss": 0.3959,
+      "step": 451
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.095052602192115,
+      "learning_rate": 6.570623828491175e-06,
+      "loss": 0.4732,
+      "step": 452
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.5882275503599168,
+      "learning_rate": 6.556441130333403e-06,
+      "loss": 0.4358,
+      "step": 453
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.6100778415480488,
+      "learning_rate": 6.542244551818451e-06,
+      "loss": 0.3534,
+      "step": 454
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.7528487768698109,
+      "learning_rate": 6.528034219551542e-06,
+      "loss": 0.4225,
+      "step": 455
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.115929646962235,
+      "learning_rate": 6.513810260260559e-06,
+      "loss": 0.4989,
+      "step": 456
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.843848673295388,
+      "learning_rate": 6.499572800794911e-06,
+      "loss": 0.5289,
+      "step": 457
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.5327924766590826,
+      "learning_rate": 6.485321968124398e-06,
+      "loss": 0.4567,
+      "step": 458
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.4479520006001745,
+      "learning_rate": 6.471057889338088e-06,
+      "loss": 0.4135,
+      "step": 459
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.5507352265630177,
+      "learning_rate": 6.456780691643172e-06,
+      "loss": 0.4116,
+      "step": 460
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.4570245443001153,
+      "learning_rate": 6.442490502363838e-06,
+      "loss": 0.3086,
+      "step": 461
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.8396731671196835,
+      "learning_rate": 6.428187448940136e-06,
+      "loss": 0.3928,
+      "step": 462
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.7001256773742297,
+      "learning_rate": 6.413871658926834e-06,
+      "loss": 0.3558,
+      "step": 463
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.0514844538438757,
+      "learning_rate": 6.399543259992288e-06,
+      "loss": 0.5153,
+      "step": 464
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.7729510151345593,
+      "learning_rate": 6.385202379917297e-06,
+      "loss": 0.3732,
+      "step": 465
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.75896661140076,
+      "learning_rate": 6.370849146593973e-06,
+      "loss": 0.4428,
+      "step": 466
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.278481398853675,
+      "learning_rate": 6.356483688024589e-06,
+      "loss": 0.499,
+      "step": 467
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.9928357122955027,
+      "learning_rate": 6.342106132320442e-06,
+      "loss": 0.5037,
+      "step": 468
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.8000026678751535,
+      "learning_rate": 6.327716607700719e-06,
+      "loss": 0.4435,
+      "step": 469
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.8559089886724838,
+      "learning_rate": 6.313315242491338e-06,
+      "loss": 0.438,
+      "step": 470
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.0025851627973745,
+      "learning_rate": 6.298902165123815e-06,
+      "loss": 0.3831,
+      "step": 471
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.5664343523094808,
+      "learning_rate": 6.284477504134117e-06,
+      "loss": 0.3618,
+      "step": 472
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.019729314852934,
+      "learning_rate": 6.2700413881615045e-06,
+      "loss": 0.4521,
+      "step": 473
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.5101081475388585,
+      "learning_rate": 6.255593945947407e-06,
+      "loss": 0.423,
+      "step": 474
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.564874785306883,
+      "learning_rate": 6.241135306334254e-06,
+      "loss": 0.3541,
+      "step": 475
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.8365721139692799,
+      "learning_rate": 6.226665598264331e-06,
+      "loss": 0.4385,
+      "step": 476
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.811282555709379,
+      "learning_rate": 6.21218495077864e-06,
+      "loss": 0.432,
+      "step": 477
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.7131707973857517,
+      "learning_rate": 6.197693493015735e-06,
+      "loss": 0.4757,
+      "step": 478
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.6746091553025404,
+      "learning_rate": 6.183191354210577e-06,
+      "loss": 0.3876,
+      "step": 479
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.6680270115475315,
+      "learning_rate": 6.168678663693383e-06,
+      "loss": 0.5057,
+      "step": 480
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.783038683063815,
+      "learning_rate": 6.154155550888467e-06,
+      "loss": 0.4514,
+      "step": 481
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.7422076759380254,
+      "learning_rate": 6.139622145313089e-06,
+      "loss": 0.431,
+      "step": 482
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.9480782360670714,
+      "learning_rate": 6.125078576576305e-06,
+      "loss": 0.4041,
+      "step": 483
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.601477220996701,
+      "learning_rate": 6.110524974377802e-06,
+      "loss": 0.3845,
+      "step": 484
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.1706407426281413,
+      "learning_rate": 6.095961468506744e-06,
+      "loss": 0.4361,
+      "step": 485
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.3929696780713714,
+      "learning_rate": 6.081388188840623e-06,
+      "loss": 0.3073,
+      "step": 486
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.0401261029546944,
+      "learning_rate": 6.066805265344084e-06,
+      "loss": 0.4436,
+      "step": 487
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.9148504474098886,
+      "learning_rate": 6.052212828067787e-06,
+      "loss": 0.4593,
+      "step": 488
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.7635819099442833,
+      "learning_rate": 6.037611007147224e-06,
+      "loss": 0.5092,
+      "step": 489
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.4802426475721033,
+      "learning_rate": 6.0229999328015786e-06,
+      "loss": 0.4024,
+      "step": 490
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.4685599882750913,
+      "learning_rate": 6.008379735332556e-06,
+      "loss": 0.3088,
+      "step": 491
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.748612085026246,
+      "learning_rate": 5.993750545123217e-06,
+      "loss": 0.4406,
+      "step": 492
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.772166754683321,
+      "learning_rate": 5.979112492636824e-06,
+      "loss": 0.4471,
+      "step": 493
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.7325520509519283,
+      "learning_rate": 5.964465708415673e-06,
+      "loss": 0.3231,
+      "step": 494
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.6030922828091225,
+      "learning_rate": 5.949810323079927e-06,
+      "loss": 0.4312,
+      "step": 495
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.1985222157973876,
+      "learning_rate": 5.935146467326456e-06,
+      "loss": 0.4116,
+      "step": 496
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.7954988435751924,
+      "learning_rate": 5.920474271927668e-06,
+      "loss": 0.3573,
+      "step": 497
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.7965610417403968,
+      "learning_rate": 5.905793867730344e-06,
+      "loss": 0.4705,
+      "step": 498
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.4845270873323246,
+      "learning_rate": 5.891105385654474e-06,
+      "loss": 0.3947,
+      "step": 499
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.6494063913341859,
+      "learning_rate": 5.876408956692084e-06,
+      "loss": 0.3885,
+      "step": 500
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.61982882759216,
+      "learning_rate": 5.861704711906068e-06,
+      "loss": 0.3829,
+      "step": 501
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.7422732722633554,
+      "learning_rate": 5.846992782429028e-06,
+      "loss": 0.4317,
+      "step": 502
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.8793061898705192,
+      "learning_rate": 5.832273299462092e-06,
+      "loss": 0.4528,
+      "step": 503
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.461398015349738,
+      "learning_rate": 5.8175463942737544e-06,
+      "loss": 0.3867,
+      "step": 504
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.8164032576204665,
+      "learning_rate": 5.8028121981987e-06,
+      "loss": 0.4303,
+      "step": 505
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.681083362899287,
+      "learning_rate": 5.7880708426366296e-06,
+      "loss": 0.4281,
+      "step": 506
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.6403401791373944,
+      "learning_rate": 5.773322459051099e-06,
+      "loss": 0.4122,
+      "step": 507
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.7904392830657831,
+      "learning_rate": 5.758567178968336e-06,
+      "loss": 0.5003,
+      "step": 508
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.6816836281671608,
+      "learning_rate": 5.7438051339760715e-06,
+      "loss": 0.4503,
+      "step": 509
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.4737121910087774,
+      "learning_rate": 5.729036455722369e-06,
+      "loss": 0.3735,
+      "step": 510
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.1113107777546505,
+      "learning_rate": 5.714261275914442e-06,
+      "loss": 0.4567,
+      "step": 511
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.437714366596938,
+      "learning_rate": 5.6994797263174906e-06,
+      "loss": 0.4383,
+      "step": 512
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.480294742385032,
+      "learning_rate": 5.684691938753517e-06,
+      "loss": 0.3716,
+      "step": 513
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.009485638198127,
+      "learning_rate": 5.669898045100156e-06,
+      "loss": 0.3101,
+      "step": 514
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.9276848419323365,
+      "learning_rate": 5.655098177289496e-06,
+      "loss": 0.4296,
+      "step": 515
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.89516458805726,
+      "learning_rate": 5.6402924673068994e-06,
+      "loss": 0.4179,
+      "step": 516
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.5877840417481992,
+      "learning_rate": 5.625481047189835e-06,
+      "loss": 0.4067,
+      "step": 517
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.5771628202151042,
+      "learning_rate": 5.610664049026691e-06,
+      "loss": 0.4776,
+      "step": 518
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.0167337793713376,
+      "learning_rate": 5.595841604955601e-06,
+      "loss": 0.5034,
+      "step": 519
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.6304179843092055,
+      "learning_rate": 5.581013847163267e-06,
+      "loss": 0.3563,
+      "step": 520
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.6878087948379197,
+      "learning_rate": 5.566180907883777e-06,
+      "loss": 0.4238,
+      "step": 521
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.5753762648228329,
+      "learning_rate": 5.551342919397429e-06,
+      "loss": 0.3613,
+      "step": 522
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.854366092949144,
+      "learning_rate": 5.536500014029548e-06,
+      "loss": 0.3682,
+      "step": 523
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.7481329815152795,
+      "learning_rate": 5.521652324149307e-06,
+      "loss": 0.4116,
+      "step": 524
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.734446987948073,
+      "learning_rate": 5.506799982168553e-06,
+      "loss": 0.4758,
+      "step": 525
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.9732915897763694,
+      "learning_rate": 5.491943120540616e-06,
+      "loss": 0.3322,
+      "step": 526
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.754698004300824,
+      "learning_rate": 5.47708187175913e-06,
+      "loss": 0.4162,
+      "step": 527
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.5763085121064524,
+      "learning_rate": 5.4622163683568584e-06,
+      "loss": 0.4129,
+      "step": 528
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.7488374814897032,
+      "learning_rate": 5.447346742904508e-06,
+      "loss": 0.3756,
+      "step": 529
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.9252531289816768,
+      "learning_rate": 5.432473128009538e-06,
+      "loss": 0.3684,
+      "step": 530
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.6637183599770184,
+      "learning_rate": 5.417595656314997e-06,
+      "loss": 0.4825,
+      "step": 531
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.592669717139386,
+      "learning_rate": 5.402714460498318e-06,
+      "loss": 0.3897,
+      "step": 532
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.1055466691732154,
+      "learning_rate": 5.387829673270152e-06,
+      "loss": 0.3825,
+      "step": 533
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.7506770604252968,
+      "learning_rate": 5.3729414273731784e-06,
+      "loss": 0.4271,
+      "step": 534
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.3962443967517109,
+      "learning_rate": 5.358049855580917e-06,
+      "loss": 0.394,
+      "step": 535
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.7444969274386137,
+      "learning_rate": 5.343155090696551e-06,
+      "loss": 0.5035,
+      "step": 536
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.6890656706103564,
+      "learning_rate": 5.328257265551742e-06,
+      "loss": 0.4379,
+      "step": 537
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.589046069503504,
+      "learning_rate": 5.313356513005433e-06,
+      "loss": 0.4579,
+      "step": 538
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.2484378057944006,
+      "learning_rate": 5.298452965942687e-06,
+      "loss": 0.3371,
+      "step": 539
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.871499749767781,
+      "learning_rate": 5.28354675727348e-06,
+      "loss": 0.4392,
+      "step": 540
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.8492888640142464,
+      "learning_rate": 5.268638019931524e-06,
+      "loss": 0.4509,
+      "step": 541
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.6893664867467961,
+      "learning_rate": 5.25372688687309e-06,
+      "loss": 0.4249,
+      "step": 542
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.138165869330146,
+      "learning_rate": 5.2388134910758015e-06,
+      "loss": 0.5334,
+      "step": 543
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.7083901767536924,
+      "learning_rate": 5.223897965537469e-06,
+      "loss": 0.3961,
+      "step": 544
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.590896869491855,
+      "learning_rate": 5.208980443274899e-06,
+      "loss": 0.395,
+      "step": 545
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.5600514572209423,
+      "learning_rate": 5.1940610573226955e-06,
+      "loss": 0.3988,
+      "step": 546
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.7810548008575084,
+      "learning_rate": 5.179139940732091e-06,
+      "loss": 0.4167,
+      "step": 547
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.7561585348208417,
+      "learning_rate": 5.1642172265697475e-06,
+      "loss": 0.4578,
+      "step": 548
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.5113933802960846,
+      "learning_rate": 5.149293047916576e-06,
+      "loss": 0.3575,
+      "step": 549
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.4740769283673236,
+      "learning_rate": 5.134367537866546e-06,
+      "loss": 0.3786,
+      "step": 550
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.7964337445061287,
+      "learning_rate": 5.119440829525504e-06,
+      "loss": 0.436,
+      "step": 551
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.474822981146484,
+      "learning_rate": 5.104513056009978e-06,
+      "loss": 0.3078,
+      "step": 552
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.0652463208640217,
+      "learning_rate": 5.089584350446001e-06,
+      "loss": 0.3805,
+      "step": 553
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.061096697608077,
+      "learning_rate": 5.074654845967912e-06,
+      "loss": 0.4306,
+      "step": 554
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.4104876880344734,
+      "learning_rate": 5.059724675717177e-06,
+      "loss": 0.366,
+      "step": 555
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.5017780767437552,
+      "learning_rate": 5.044793972841203e-06,
+      "loss": 0.3364,
+      "step": 556
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.5589038243980964,
+      "learning_rate": 5.029862870492142e-06,
+      "loss": 0.4908,
+      "step": 557
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.8861893390087554,
+      "learning_rate": 5.0149315018257104e-06,
+      "loss": 0.5019,
+      "step": 558
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.5951533316204654,
+      "learning_rate": 5e-06,
+      "loss": 0.3338,
+      "step": 559
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.7630862605909332,
+      "learning_rate": 4.98506849817429e-06,
+      "loss": 0.4317,
+      "step": 560
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.1599141486049276,
+      "learning_rate": 4.97013712950786e-06,
+      "loss": 0.5757,
+      "step": 561
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.5004945405458627,
+      "learning_rate": 4.955206027158798e-06,
+      "loss": 0.4114,
+      "step": 562
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.524002260752503,
+      "learning_rate": 4.940275324282824e-06,
+      "loss": 0.4123,
+      "step": 563
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.7188355830824718,
+      "learning_rate": 4.925345154032092e-06,
+      "loss": 0.4532,
+      "step": 564
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.769161229661791,
+      "learning_rate": 4.910415649554001e-06,
+      "loss": 0.4467,
+      "step": 565
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.7121837803660678,
+      "learning_rate": 4.895486943990023e-06,
+      "loss": 0.4278,
+      "step": 566
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.3657353013619622,
+      "learning_rate": 4.8805591704745e-06,
+      "loss": 0.2821,
+      "step": 567
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.7857169814598675,
+      "learning_rate": 4.865632462133456e-06,
+      "loss": 0.4362,
+      "step": 568
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.7123413107341325,
+      "learning_rate": 4.850706952083426e-06,
+      "loss": 0.4296,
+      "step": 569
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.66979921252736,
+      "learning_rate": 4.835782773430255e-06,
+      "loss": 0.3368,
+      "step": 570
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.7608078910887999,
+      "learning_rate": 4.8208600592679105e-06,
+      "loss": 0.4087,
+      "step": 571
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.8553210769178123,
+      "learning_rate": 4.805938942677305e-06,
+      "loss": 0.4332,
+      "step": 572
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.7077102226161969,
+      "learning_rate": 4.791019556725104e-06,
+      "loss": 0.4783,
+      "step": 573
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.567173583407488,
+      "learning_rate": 4.776102034462533e-06,
+      "loss": 0.4127,
+      "step": 574
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.71170113748187,
+      "learning_rate": 4.7611865089242e-06,
+      "loss": 0.3824,
+      "step": 575
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.7817717043669978,
+      "learning_rate": 4.746273113126912e-06,
+      "loss": 0.3953,
+      "step": 576
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.6662767108236813,
+      "learning_rate": 4.7313619800684765e-06,
+      "loss": 0.4033,
+      "step": 577
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.8754863396071395,
+      "learning_rate": 4.7164532427265206e-06,
+      "loss": 0.4289,
+      "step": 578
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.5528783522233447,
+      "learning_rate": 4.701547034057314e-06,
+      "loss": 0.392,
+      "step": 579
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.4926378728821097,
+      "learning_rate": 4.686643486994568e-06,
+      "loss": 0.4548,
+      "step": 580
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.7303457881353712,
+      "learning_rate": 4.67174273444826e-06,
+      "loss": 0.3818,
+      "step": 581
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.6238622098504738,
+      "learning_rate": 4.656844909303449e-06,
+      "loss": 0.4397,
+      "step": 582
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.4101186566517188,
+      "learning_rate": 4.641950144419085e-06,
+      "loss": 0.2988,
+      "step": 583
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.1784078440486314,
+      "learning_rate": 4.627058572626823e-06,
+      "loss": 0.3701,
+      "step": 584
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.7501312534120357,
+      "learning_rate": 4.612170326729849e-06,
+      "loss": 0.4129,
+      "step": 585
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.48698910341266,
+      "learning_rate": 4.597285539501684e-06,
+      "loss": 0.4013,
+      "step": 586
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.6417757683273082,
+      "learning_rate": 4.5824043436850054e-06,
+      "loss": 0.3804,
+      "step": 587
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.5281429537537168,
+      "learning_rate": 4.567526871990462e-06,
+      "loss": 0.3938,
+      "step": 588
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.5988936524719048,
+      "learning_rate": 4.552653257095495e-06,
+      "loss": 0.368,
+      "step": 589
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.258896440894041,
+      "learning_rate": 4.537783631643143e-06,
+      "loss": 0.5043,
+      "step": 590
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.7919794024010391,
+      "learning_rate": 4.522918128240871e-06,
+      "loss": 0.3857,
+      "step": 591
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.873037865036066,
+      "learning_rate": 4.508056879459387e-06,
+      "loss": 0.3815,
+      "step": 592
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.3923805891179883,
+      "learning_rate": 4.493200017831448e-06,
+      "loss": 0.3562,
+      "step": 593
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.7449443105770417,
+      "learning_rate": 4.478347675850692e-06,
+      "loss": 0.3864,
+      "step": 594
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.5506965557164598,
+      "learning_rate": 4.463499985970455e-06,
+      "loss": 0.3961,
+      "step": 595
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.7471922729491982,
+      "learning_rate": 4.448657080602573e-06,
+      "loss": 0.4226,
+      "step": 596
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.3267013207814808,
+      "learning_rate": 4.433819092116223e-06,
+      "loss": 0.3554,
+      "step": 597
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.8265104161059438,
+      "learning_rate": 4.418986152836735e-06,
+      "loss": 0.3621,
+      "step": 598
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.6099837359624083,
+      "learning_rate": 4.404158395044401e-06,
+      "loss": 0.3619,
+      "step": 599
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.6571840549052268,
+      "learning_rate": 4.38933595097331e-06,
+      "loss": 0.4452,
+      "step": 600
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.951655679068924,
+      "learning_rate": 4.374518952810167e-06,
+      "loss": 0.4506,
+      "step": 601
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.0878776186312247,
+      "learning_rate": 4.359707532693102e-06,
+      "loss": 0.4417,
+      "step": 602
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.082656505070747,
+      "learning_rate": 4.344901822710505e-06,
+      "loss": 0.5075,
+      "step": 603
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.9874784900014248,
+      "learning_rate": 4.330101954899847e-06,
+      "loss": 0.4667,
+      "step": 604
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.414272287045042,
+      "learning_rate": 4.3153080612464835e-06,
+      "loss": 0.3666,
+      "step": 605
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.1885859077717607,
+      "learning_rate": 4.300520273682511e-06,
+      "loss": 0.4392,
+      "step": 606
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.0519927072192363,
+      "learning_rate": 4.28573872408556e-06,
+      "loss": 0.3745,
+      "step": 607
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.956594925966504,
+      "learning_rate": 4.270963544277633e-06,
+      "loss": 0.3878,
+      "step": 608
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.7636709472530088,
+      "learning_rate": 4.256194866023929e-06,
+      "loss": 0.4661,
+      "step": 609
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.9901540818140182,
+      "learning_rate": 4.241432821031665e-06,
+      "loss": 0.3897,
+      "step": 610
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.764906598642035,
+      "learning_rate": 4.226677540948902e-06,
+      "loss": 0.3544,
+      "step": 611
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.7932823295954536,
+      "learning_rate": 4.211929157363372e-06,
+      "loss": 0.4192,
+      "step": 612
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.4835930955379155,
+      "learning_rate": 4.197187801801301e-06,
+      "loss": 0.3727,
+      "step": 613
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.7336104464617754,
+      "learning_rate": 4.182453605726246e-06,
+      "loss": 0.4964,
+      "step": 614
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.8329880443284419,
+      "learning_rate": 4.167726700537909e-06,
+      "loss": 0.3721,
+      "step": 615
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.013138218420455,
+      "learning_rate": 4.153007217570973e-06,
+      "loss": 0.4649,
+      "step": 616
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.8427860818655313,
+      "learning_rate": 4.138295288093935e-06,
+      "loss": 0.374,
+      "step": 617
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.6825250019411562,
+      "learning_rate": 4.123591043307918e-06,
+      "loss": 0.4285,
+      "step": 618
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.4196800310040327,
+      "learning_rate": 4.1088946143455275e-06,
+      "loss": 0.4058,
+      "step": 619
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.9422054561550897,
+      "learning_rate": 4.094206132269658e-06,
+      "loss": 0.388,
+      "step": 620
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.7459635361160109,
+      "learning_rate": 4.079525728072335e-06,
+      "loss": 0.4262,
+      "step": 621
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.5984050785760113,
+      "learning_rate": 4.0648535326735464e-06,
+      "loss": 0.3966,
+      "step": 622
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.7310491470959026,
+      "learning_rate": 4.050189676920075e-06,
+      "loss": 0.3646,
+      "step": 623
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.660090655393537,
+      "learning_rate": 4.035534291584328e-06,
+      "loss": 0.3155,
+      "step": 624
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.6484172315966699,
+      "learning_rate": 4.020887507363177e-06,
+      "loss": 0.3591,
+      "step": 625
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.0685621080461996,
+      "learning_rate": 4.006249454876785e-06,
+      "loss": 0.4811,
+      "step": 626
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.8125872688192208,
+      "learning_rate": 3.991620264667446e-06,
+      "loss": 0.4088,
+      "step": 627
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.9152975230896734,
+      "learning_rate": 3.977000067198422e-06,
+      "loss": 0.3687,
+      "step": 628
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.007047221638238,
+      "learning_rate": 3.962388992852778e-06,
+      "loss": 0.3888,
+      "step": 629
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.884062017423324,
+      "learning_rate": 3.947787171932215e-06,
+      "loss": 0.3866,
+      "step": 630
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.813504934485305,
+      "learning_rate": 3.933194734655916e-06,
+      "loss": 0.4673,
+      "step": 631
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.7247817556858769,
+      "learning_rate": 3.918611811159379e-06,
+      "loss": 0.4553,
+      "step": 632
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.9109345821437327,
+      "learning_rate": 3.904038531493257e-06,
+      "loss": 0.4128,
+      "step": 633
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.7269043143853864,
+      "learning_rate": 3.889475025622199e-06,
+      "loss": 0.4578,
+      "step": 634
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.8714889385967017,
+      "learning_rate": 3.874921423423697e-06,
+      "loss": 0.463,
+      "step": 635
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.8601128891856535,
+      "learning_rate": 3.860377854686913e-06,
+      "loss": 0.3878,
+      "step": 636
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.588587861996327,
+      "learning_rate": 3.845844449111535e-06,
+      "loss": 0.3896,
+      "step": 637
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.6959027640047006,
+      "learning_rate": 3.8313213363066195e-06,
+      "loss": 0.4031,
+      "step": 638
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.0487519369918483,
+      "learning_rate": 3.816808645789425e-06,
+      "loss": 0.5667,
+      "step": 639
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.5283004570346033,
+      "learning_rate": 3.802306506984266e-06,
+      "loss": 0.3764,
+      "step": 640
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.5944350505584564,
+      "learning_rate": 3.7878150492213617e-06,
+      "loss": 0.398,
+      "step": 641
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.9997700960820088,
+      "learning_rate": 3.7733344017356702e-06,
+      "loss": 0.4948,
+      "step": 642
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.7499616339572772,
+      "learning_rate": 3.758864693665748e-06,
+      "loss": 0.3559,
+      "step": 643
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.5788625250459556,
+      "learning_rate": 3.744406054052594e-06,
+      "loss": 0.2929,
+      "step": 644
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.9465951183623658,
+      "learning_rate": 3.7299586118384967e-06,
+      "loss": 0.4659,
+      "step": 645
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.935063815996509,
+      "learning_rate": 3.715522495865885e-06,
+      "loss": 0.4169,
+      "step": 646
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.7069776267296644,
+      "learning_rate": 3.701097834876185e-06,
+      "loss": 0.4085,
+      "step": 647
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.7694847688907462,
+      "learning_rate": 3.6866847575086626e-06,
+      "loss": 0.3837,
+      "step": 648
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.6799663024285176,
+      "learning_rate": 3.6722833922992824e-06,
+      "loss": 0.4804,
+      "step": 649
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.7949733183074061,
+      "learning_rate": 3.6578938676795576e-06,
+      "loss": 0.4979,
+      "step": 650
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.5398405813591576,
+      "learning_rate": 3.6435163119754134e-06,
+      "loss": 0.3302,
+      "step": 651
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.69848168155202,
+      "learning_rate": 3.6291508534060293e-06,
+      "loss": 0.4512,
+      "step": 652
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.0085078247743695,
+      "learning_rate": 3.614797620082703e-06,
+      "loss": 0.4211,
+      "step": 653
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.8349562513355984,
+      "learning_rate": 3.600456740007714e-06,
+      "loss": 0.3772,
+      "step": 654
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.5313049832419303,
+      "learning_rate": 3.586128341073167e-06,
+      "loss": 0.3375,
+      "step": 655
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.601512381086004,
+      "learning_rate": 3.5718125510598646e-06,
+      "loss": 0.3185,
+      "step": 656
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.7021346399312534,
+      "learning_rate": 3.557509497636163e-06,
+      "loss": 0.3741,
+      "step": 657
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.812220332081257,
+      "learning_rate": 3.54321930835683e-06,
+      "loss": 0.4198,
+      "step": 658
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.9490992488522427,
+      "learning_rate": 3.5289421106619125e-06,
+      "loss": 0.3687,
+      "step": 659
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.593260907761964,
+      "learning_rate": 3.5146780318756025e-06,
+      "loss": 0.4384,
+      "step": 660
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.54237996978636,
+      "learning_rate": 3.500427199205091e-06,
+      "loss": 0.3594,
+      "step": 661
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.7268772903478766,
+      "learning_rate": 3.486189739739442e-06,
+      "loss": 0.4085,
+      "step": 662
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.8912939127949164,
+      "learning_rate": 3.4719657804484607e-06,
+      "loss": 0.467,
+      "step": 663
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.8970124049976202,
+      "learning_rate": 3.4577554481815513e-06,
+      "loss": 0.4586,
+      "step": 664
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.9433147628840528,
+      "learning_rate": 3.4435588696665977e-06,
+      "loss": 0.3936,
+      "step": 665
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.7797627873045827,
+      "learning_rate": 3.429376171508827e-06,
+      "loss": 0.4969,
+      "step": 666
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.9262430414234508,
+      "learning_rate": 3.415207480189676e-06,
+      "loss": 0.3561,
+      "step": 667
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.2861119104303473,
+      "learning_rate": 3.401052922065675e-06,
+      "loss": 0.3552,
+      "step": 668
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.632481046878911,
+      "learning_rate": 3.386912623367311e-06,
+      "loss": 0.3602,
+      "step": 669
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.9372946810317673,
+      "learning_rate": 3.3727867101979037e-06,
+      "loss": 0.4401,
+      "step": 670
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.5715805933345222,
+      "learning_rate": 3.3586753085324863e-06,
+      "loss": 0.348,
+      "step": 671
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.2178424888350166,
+      "learning_rate": 3.344578544216678e-06,
+      "loss": 0.4701,
+      "step": 672
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.57523250263701,
+      "learning_rate": 3.3304965429655567e-06,
+      "loss": 0.4093,
+      "step": 673
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.4714423430336083,
+      "learning_rate": 3.3164294303625487e-06,
+      "loss": 0.3806,
+      "step": 674
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.774836779247577,
+      "learning_rate": 3.302377331858302e-06,
+      "loss": 0.4084,
+      "step": 675
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.8798583661138206,
+      "learning_rate": 3.2883403727695695e-06,
+      "loss": 0.3936,
+      "step": 676
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.9543516934159417,
+      "learning_rate": 3.2743186782780855e-06,
+      "loss": 0.3679,
+      "step": 677
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.9366343183808823,
+      "learning_rate": 3.2603123734294608e-06,
+      "loss": 0.3919,
+      "step": 678
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.9434599159853194,
+      "learning_rate": 3.246321583132058e-06,
+      "loss": 0.4365,
+      "step": 679
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.992091467377912,
+      "learning_rate": 3.2323464321558806e-06,
+      "loss": 0.3731,
+      "step": 680
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.6956161944817005,
+      "learning_rate": 3.2183870451314624e-06,
+      "loss": 0.3644,
+      "step": 681
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.792655894061527,
+      "learning_rate": 3.2044435465487545e-06,
+      "loss": 0.432,
+      "step": 682
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.7565960934993627,
+      "learning_rate": 3.190516060756009e-06,
+      "loss": 0.3551,
+      "step": 683
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.0251586478522685,
+      "learning_rate": 3.1766047119586846e-06,
+      "loss": 0.4069,
+      "step": 684
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.4722903621029912,
+      "learning_rate": 3.162709624218324e-06,
+      "loss": 0.3428,
+      "step": 685
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.36831955527062,
+      "learning_rate": 3.148830921451452e-06,
+      "loss": 0.4013,
+      "step": 686
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.83177751191809,
+      "learning_rate": 3.1349687274284803e-06,
+      "loss": 0.3712,
+      "step": 687
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.375489215381062,
+      "learning_rate": 3.121123165772588e-06,
+      "loss": 0.2576,
+      "step": 688
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.7824334878491648,
+      "learning_rate": 3.107294359958628e-06,
+      "loss": 0.3375,
+      "step": 689
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.754892268091059,
+      "learning_rate": 3.0934824333120273e-06,
+      "loss": 0.3291,
+      "step": 690
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.8208162050718975,
+      "learning_rate": 3.0796875090076824e-06,
+      "loss": 0.3929,
+      "step": 691
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.7351771828031815,
+      "learning_rate": 3.065909710068861e-06,
+      "loss": 0.4085,
+      "step": 692
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.6384512419654458,
+      "learning_rate": 3.052149159366109e-06,
+      "loss": 0.3737,
+      "step": 693
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.484119659153863,
+      "learning_rate": 3.0384059796161536e-06,
+      "loss": 0.3597,
+      "step": 694
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.7353506217902939,
+      "learning_rate": 3.024680293380804e-06,
+      "loss": 0.3506,
+      "step": 695
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.6917973155179618,
+      "learning_rate": 3.0109722230658634e-06,
+      "loss": 0.384,
+      "step": 696
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.746371203974182,
+      "learning_rate": 2.99728189092004e-06,
+      "loss": 0.3238,
+      "step": 697
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.7133582169680055,
+      "learning_rate": 2.983609419033845e-06,
+      "loss": 0.3962,
+      "step": 698
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.432909539517969,
+      "learning_rate": 2.969954929338518e-06,
+      "loss": 0.2525,
+      "step": 699
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.4916482205359396,
+      "learning_rate": 2.9563185436049323e-06,
+      "loss": 0.3378,
+      "step": 700
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.8166936097543536,
+      "learning_rate": 2.942700383442508e-06,
+      "loss": 0.4769,
+      "step": 701
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.358690049804268,
+      "learning_rate": 2.929100570298129e-06,
+      "loss": 0.3305,
+      "step": 702
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.4731450654852187,
+      "learning_rate": 2.915519225455065e-06,
+      "loss": 0.3507,
+      "step": 703
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.6254972480142047,
+      "learning_rate": 2.901956470031877e-06,
+      "loss": 0.3509,
+      "step": 704
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.5825345498709213,
+      "learning_rate": 2.888412424981353e-06,
+      "loss": 0.4081,
+      "step": 705
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.5897743608822972,
+      "learning_rate": 2.87488721108942e-06,
+      "loss": 0.3497,
+      "step": 706
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.4379237508967013,
+      "learning_rate": 2.8613809489740665e-06,
+      "loss": 0.2969,
+      "step": 707
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.800012440266111,
+      "learning_rate": 2.847893759084267e-06,
+      "loss": 0.4193,
+      "step": 708
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.9042276012953723,
+      "learning_rate": 2.8344257616989145e-06,
+      "loss": 0.3966,
+      "step": 709
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.8274995596459567,
+      "learning_rate": 2.820977076925744e-06,
+      "loss": 0.3928,
+      "step": 710
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.7082205838669184,
+      "learning_rate": 2.807547824700252e-06,
+      "loss": 0.3248,
+      "step": 711
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.142315177939501,
+      "learning_rate": 2.7941381247846455e-06,
+      "loss": 0.4237,
+      "step": 712
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.658641685481167,
+      "learning_rate": 2.780748096766758e-06,
+      "loss": 0.3241,
+      "step": 713
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.374221636644457,
+      "learning_rate": 2.7673778600589862e-06,
+      "loss": 0.3503,
+      "step": 714
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.683744855360216,
+      "learning_rate": 2.7540275338972345e-06,
+      "loss": 0.4029,
+      "step": 715
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.7431637682180983,
+      "learning_rate": 2.7406972373398443e-06,
+      "loss": 0.3403,
+      "step": 716
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.6389313240780434,
+      "learning_rate": 2.7273870892665257e-06,
+      "loss": 0.3558,
+      "step": 717
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.9951930342344457,
+      "learning_rate": 2.7140972083773125e-06,
+      "loss": 0.4519,
+      "step": 718
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.4977291642632802,
+      "learning_rate": 2.700827713191492e-06,
+      "loss": 0.3244,
+      "step": 719
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.9719128206467786,
+      "learning_rate": 2.6875787220465466e-06,
+      "loss": 0.4308,
+      "step": 720
+    },
+    {
+      "epoch": 0.66,
+      "eval_loss": 0.4720384180545807,
+      "eval_runtime": 531.3135,
+      "eval_samples_per_second": 5.101,
+      "eval_steps_per_second": 0.32,
+      "step": 720
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.9044872002249398,
+      "learning_rate": 2.6743503530971136e-06,
+      "loss": 0.3943,
+      "step": 721
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.7774415480268881,
+      "learning_rate": 2.6611427243139166e-06,
+      "loss": 0.3772,
+      "step": 722
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.3385371503925785,
+      "learning_rate": 2.647955953482717e-06,
+      "loss": 0.3347,
+      "step": 723
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.5686275838880652,
+      "learning_rate": 2.6347901582032627e-06,
+      "loss": 0.3345,
+      "step": 724
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.5884350530337958,
+      "learning_rate": 2.6216454558882486e-06,
+      "loss": 0.3385,
+      "step": 725
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.2760933804201096,
+      "learning_rate": 2.6085219637622545e-06,
+      "loss": 0.4525,
+      "step": 726
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.8598949852430244,
+      "learning_rate": 2.595419798860713e-06,
+      "loss": 0.4161,
+      "step": 727
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.614789920339542,
+      "learning_rate": 2.5823390780288604e-06,
+      "loss": 0.387,
+      "step": 728
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.6825637810384086,
+      "learning_rate": 2.5692799179206905e-06,
+      "loss": 0.3596,
+      "step": 729
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.5818504183911322,
+      "learning_rate": 2.5562424349979198e-06,
+      "loss": 0.3139,
+      "step": 730
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.9512239910567963,
+      "learning_rate": 2.5432267455289505e-06,
+      "loss": 0.4087,
+      "step": 731
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.4346478577174004,
+      "learning_rate": 2.5302329655878243e-06,
+      "loss": 0.3677,
+      "step": 732
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.6197744535752765,
+      "learning_rate": 2.5172612110532013e-06,
+      "loss": 0.3781,
+      "step": 733
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.0219491074323597,
+      "learning_rate": 2.5043115976073166e-06,
+      "loss": 0.497,
+      "step": 734
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.7400391427698392,
+      "learning_rate": 2.491384240734943e-06,
+      "loss": 0.3739,
+      "step": 735
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.6890186070354904,
+      "learning_rate": 2.478479255722378e-06,
+      "loss": 0.3927,
+      "step": 736
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.5819118237338632,
+      "learning_rate": 2.4655967576564066e-06,
+      "loss": 0.4005,
+      "step": 737
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.844066756069027,
+      "learning_rate": 2.4527368614232683e-06,
+      "loss": 0.4126,
+      "step": 738
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.947388676610224,
+      "learning_rate": 2.4398996817076458e-06,
+      "loss": 0.4356,
+      "step": 739
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.64465771062421,
+      "learning_rate": 2.4270853329916305e-06,
+      "loss": 0.3574,
+      "step": 740
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.7772779084636083,
+      "learning_rate": 2.414293929553713e-06,
+      "loss": 0.357,
+      "step": 741
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.559427759832433,
+      "learning_rate": 2.4015255854677487e-06,
+      "loss": 0.3445,
+      "step": 742
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.3649092735699666,
+      "learning_rate": 2.388780414601959e-06,
+      "loss": 0.3573,
+      "step": 743
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.7111213233065832,
+      "learning_rate": 2.3760585306179013e-06,
+      "loss": 0.3266,
+      "step": 744
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.6863130092798397,
+      "learning_rate": 2.3633600469694608e-06,
+      "loss": 0.4125,
+      "step": 745
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.2514447035959724,
+      "learning_rate": 2.3506850769018363e-06,
+      "loss": 0.3657,
+      "step": 746
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.4130571629239113,
+      "learning_rate": 2.3380337334505374e-06,
+      "loss": 0.3493,
+      "step": 747
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.4412927726602656,
+      "learning_rate": 2.3254061294403664e-06,
+      "loss": 0.2888,
+      "step": 748
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.6984121235842002,
+      "learning_rate": 2.3128023774844196e-06,
+      "loss": 0.288,
+      "step": 749
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.9026809021111732,
+      "learning_rate": 2.300222589983084e-06,
+      "loss": 0.4988,
+      "step": 750
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.7493875713221276,
+      "learning_rate": 2.2876668791230196e-06,
+      "loss": 0.371,
+      "step": 751
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.2417503401806425,
+      "learning_rate": 2.2751353568761814e-06,
+      "loss": 0.4444,
+      "step": 752
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.7447814422023238,
+      "learning_rate": 2.2626281349988106e-06,
+      "loss": 0.385,
+      "step": 753
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.379646006684634,
+      "learning_rate": 2.2501453250304282e-06,
+      "loss": 0.2819,
+      "step": 754
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.700160730403637,
+      "learning_rate": 2.2376870382928608e-06,
+      "loss": 0.3979,
+      "step": 755
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.4562605513649785,
+      "learning_rate": 2.2252533858892278e-06,
+      "loss": 0.3582,
+      "step": 756
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.8929935611713011,
+      "learning_rate": 2.212844478702962e-06,
+      "loss": 0.4928,
+      "step": 757
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.8464390400413067,
+      "learning_rate": 2.200460427396819e-06,
+      "loss": 0.3976,
+      "step": 758
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.5935828621108423,
+      "learning_rate": 2.1881013424118925e-06,
+      "loss": 0.318,
+      "step": 759
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.641770016765562,
+      "learning_rate": 2.175767333966618e-06,
+      "loss": 0.3763,
+      "step": 760
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.9210613798451146,
+      "learning_rate": 2.1634585120558076e-06,
+      "loss": 0.3928,
+      "step": 761
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.7068856262802559,
+      "learning_rate": 2.1511749864496533e-06,
+      "loss": 0.4222,
+      "step": 762
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.794733425784197,
+      "learning_rate": 2.138916866692754e-06,
+      "loss": 0.5005,
+      "step": 763
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.6616396879522488,
+      "learning_rate": 2.1266842621031434e-06,
+      "loss": 0.3553,
+      "step": 764
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.8785861775392818,
+      "learning_rate": 2.1144772817713106e-06,
+      "loss": 0.4732,
+      "step": 765
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.5125823752054015,
+      "learning_rate": 2.1022960345592226e-06,
+      "loss": 0.3431,
+      "step": 766
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.9035924379439653,
+      "learning_rate": 2.09014062909936e-06,
+      "loss": 0.3238,
+      "step": 767
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.2467574366537604,
+      "learning_rate": 2.0780111737937495e-06,
+      "loss": 0.5255,
+      "step": 768
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.6757524552649274,
+      "learning_rate": 2.0659077768129897e-06,
+      "loss": 0.3495,
+      "step": 769
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.8178528858642133,
+      "learning_rate": 2.0538305460952944e-06,
+      "loss": 0.3654,
+      "step": 770
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.9355937874790798,
+      "learning_rate": 2.041779589345527e-06,
+      "loss": 0.4529,
+      "step": 771
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.521367156724969,
+      "learning_rate": 2.029755014034234e-06,
+      "loss": 0.301,
+      "step": 772
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.826744504300451,
+      "learning_rate": 2.0177569273966944e-06,
+      "loss": 0.3665,
+      "step": 773
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.7149354203694684,
+      "learning_rate": 2.0057854364319644e-06,
+      "loss": 0.3673,
+      "step": 774
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.4674644295610002,
+      "learning_rate": 1.9938406479019186e-06,
+      "loss": 0.3403,
+      "step": 775
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.5070539943420556,
+      "learning_rate": 1.9819226683302934e-06,
+      "loss": 0.3795,
+      "step": 776
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.4784387385558468,
+      "learning_rate": 1.9700316040017513e-06,
+      "loss": 0.4026,
+      "step": 777
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.5868950806618929,
+      "learning_rate": 1.9581675609609176e-06,
+      "loss": 0.3108,
+      "step": 778
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.592305830159606,
+      "learning_rate": 1.9463306450114417e-06,
+      "loss": 0.3549,
+      "step": 779
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.655256097625306,
+      "learning_rate": 1.934520961715058e-06,
+      "loss": 0.4225,
+      "step": 780
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.5809519234724267,
+      "learning_rate": 1.922738616390639e-06,
+      "loss": 0.3112,
+      "step": 781
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.9427582694211518,
+      "learning_rate": 1.910983714113253e-06,
+      "loss": 0.421,
+      "step": 782
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.4899484568922368,
+      "learning_rate": 1.8992563597132324e-06,
+      "loss": 0.3075,
+      "step": 783
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.6154594194570981,
+      "learning_rate": 1.8875566577752403e-06,
+      "loss": 0.3521,
+      "step": 784
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.3640422359449944,
+      "learning_rate": 1.8758847126373304e-06,
+      "loss": 0.3433,
+      "step": 785
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.8143589393862407,
+      "learning_rate": 1.8642406283900238e-06,
+      "loss": 0.3839,
+      "step": 786
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.5446087202063494,
+      "learning_rate": 1.8526245088753802e-06,
+      "loss": 0.2688,
+      "step": 787
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.6333832866166789,
+      "learning_rate": 1.8410364576860645e-06,
+      "loss": 0.3158,
+      "step": 788
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.7966540146020227,
+      "learning_rate": 1.8294765781644285e-06,
+      "loss": 0.411,
+      "step": 789
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.4442265911590433,
+      "learning_rate": 1.8179449734015946e-06,
+      "loss": 0.44,
+      "step": 790
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.0035737445307986,
+      "learning_rate": 1.8064417462365225e-06,
+      "loss": 0.386,
+      "step": 791
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.8535330938944725,
+      "learning_rate": 1.7949669992551056e-06,
+      "loss": 0.4503,
+      "step": 792
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.4981921904993794,
+      "learning_rate": 1.7835208347892536e-06,
+      "loss": 0.3836,
+      "step": 793
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.9171239926769839,
+      "learning_rate": 1.7721033549159656e-06,
+      "loss": 0.3972,
+      "step": 794
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.5637043438891536,
+      "learning_rate": 1.7607146614564418e-06,
+      "loss": 0.3937,
+      "step": 795
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.6447122243659613,
+      "learning_rate": 1.7493548559751638e-06,
+      "loss": 0.3466,
+      "step": 796
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.0092846561255793,
+      "learning_rate": 1.7380240397789838e-06,
+      "loss": 0.4356,
+      "step": 797
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.5957076368272978,
+      "learning_rate": 1.7267223139162343e-06,
+      "loss": 0.351,
+      "step": 798
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.8330485102111689,
+      "learning_rate": 1.7154497791758156e-06,
+      "loss": 0.383,
+      "step": 799
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.7558593725919067,
+      "learning_rate": 1.704206536086301e-06,
+      "loss": 0.4745,
+      "step": 800
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.8130146446382733,
+      "learning_rate": 1.692992684915043e-06,
+      "loss": 0.3835,
+      "step": 801
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.8881982094164256,
+      "learning_rate": 1.681808325667278e-06,
+      "loss": 0.4934,
+      "step": 802
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.5165095033729978,
+      "learning_rate": 1.6706535580852268e-06,
+      "loss": 0.3863,
+      "step": 803
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.774357680731744,
+      "learning_rate": 1.6595284816472196e-06,
+      "loss": 0.3748,
+      "step": 804
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.767710277007722,
+      "learning_rate": 1.6484331955667948e-06,
+      "loss": 0.3393,
+      "step": 805
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.8448330342374666,
+      "learning_rate": 1.6373677987918196e-06,
+      "loss": 0.4351,
+      "step": 806
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.85389177211945,
+      "learning_rate": 1.6263323900036126e-06,
+      "loss": 0.3646,
+      "step": 807
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.4324306759275551,
+      "learning_rate": 1.615327067616057e-06,
+      "loss": 0.2679,
+      "step": 808
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.542343302777527,
+      "learning_rate": 1.6043519297747284e-06,
+      "loss": 0.4538,
+      "step": 809
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.4664118852248411,
+      "learning_rate": 1.5934070743560066e-06,
+      "loss": 0.2817,
+      "step": 810
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.3646102425906645,
+      "learning_rate": 1.5824925989662216e-06,
+      "loss": 0.3436,
+      "step": 811
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.726338790957311,
+      "learning_rate": 1.5716086009407739e-06,
+      "loss": 0.4175,
+      "step": 812
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.7959210588836272,
+      "learning_rate": 1.56075517734326e-06,
+      "loss": 0.3795,
+      "step": 813
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.7311804415302914,
+      "learning_rate": 1.549932424964622e-06,
+      "loss": 0.4281,
+      "step": 814
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.8431199283255997,
+      "learning_rate": 1.5391404403222676e-06,
+      "loss": 0.3554,
+      "step": 815
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.6029303411198748,
+      "learning_rate": 1.528379319659221e-06,
+      "loss": 0.399,
+      "step": 816
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.041057354884482,
+      "learning_rate": 1.517649158943263e-06,
+      "loss": 0.3892,
+      "step": 817
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.200908102889503,
+      "learning_rate": 1.5069500538660714e-06,
+      "loss": 0.3841,
+      "step": 818
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.5822240990320229,
+      "learning_rate": 1.4962820998423683e-06,
+      "loss": 0.3568,
+      "step": 819
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.1371703772175765,
+      "learning_rate": 1.4856453920090742e-06,
+      "loss": 0.4127,
+      "step": 820
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.474872780748052,
+      "learning_rate": 1.4750400252244511e-06,
+      "loss": 0.3988,
+      "step": 821
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.7498309567196768,
+      "learning_rate": 1.4644660940672628e-06,
+      "loss": 0.3476,
+      "step": 822
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.8541907978098617,
+      "learning_rate": 1.4539236928359319e-06,
+      "loss": 0.434,
+      "step": 823
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.942758452810042,
+      "learning_rate": 1.4434129155476961e-06,
+      "loss": 0.364,
+      "step": 824
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.5720835349530153,
+      "learning_rate": 1.4329338559377692e-06,
+      "loss": 0.335,
+      "step": 825
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.5728894698285647,
+      "learning_rate": 1.4224866074585052e-06,
+      "loss": 0.4565,
+      "step": 826
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.7418556788519788,
+      "learning_rate": 1.412071263278571e-06,
+      "loss": 0.4165,
+      "step": 827
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.304963198709215,
+      "learning_rate": 1.4016879162821046e-06,
+      "loss": 0.4437,
+      "step": 828
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.6243364242282008,
+      "learning_rate": 1.3913366590678967e-06,
+      "loss": 0.465,
+      "step": 829
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.95589187270623,
+      "learning_rate": 1.381017583948563e-06,
+      "loss": 0.4478,
+      "step": 830
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.5997871441952878,
+      "learning_rate": 1.370730782949713e-06,
+      "loss": 0.3457,
+      "step": 831
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.8336248474819126,
+      "learning_rate": 1.3604763478091375e-06,
+      "loss": 0.3157,
+      "step": 832
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.7356602754382147,
+      "learning_rate": 1.3502543699759918e-06,
+      "loss": 0.3872,
+      "step": 833
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.546716942025288,
+      "learning_rate": 1.340064940609972e-06,
+      "loss": 0.3602,
+      "step": 834
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.8094889043425166,
+      "learning_rate": 1.3299081505805088e-06,
+      "loss": 0.3067,
+      "step": 835
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.6451477176671583,
+      "learning_rate": 1.319784090465958e-06,
+      "loss": 0.2976,
+      "step": 836
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.9811172173472407,
+      "learning_rate": 1.3096928505527812e-06,
+      "loss": 0.4255,
+      "step": 837
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.082775157955807,
+      "learning_rate": 1.2996345208347566e-06,
+      "loss": 0.4362,
+      "step": 838
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.8098736692974182,
+      "learning_rate": 1.2896091910121667e-06,
+      "loss": 0.3936,
+      "step": 839
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.9547737166972166,
+      "learning_rate": 1.2796169504910028e-06,
+      "loss": 0.381,
+      "step": 840
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.6251866261565813,
+      "learning_rate": 1.2696578883821614e-06,
+      "loss": 0.4156,
+      "step": 841
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.8001878571293004,
+      "learning_rate": 1.2597320935006541e-06,
+      "loss": 0.3925,
+      "step": 842
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.61611282004936,
+      "learning_rate": 1.2498396543648196e-06,
+      "loss": 0.3515,
+      "step": 843
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.7985809369413097,
+      "learning_rate": 1.2399806591955228e-06,
+      "loss": 0.3292,
+      "step": 844
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.5964590321964889,
+      "learning_rate": 1.2301551959153812e-06,
+      "loss": 0.3877,
+      "step": 845
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.5734877825215414,
+      "learning_rate": 1.2203633521479735e-06,
+      "loss": 0.4054,
+      "step": 846
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.7916360058254706,
+      "learning_rate": 1.210605215217056e-06,
+      "loss": 0.4528,
+      "step": 847
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.033617142829962,
+      "learning_rate": 1.2008808721457882e-06,
+      "loss": 0.4323,
+      "step": 848
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.7334991655343743,
+      "learning_rate": 1.191190409655959e-06,
+      "loss": 0.4516,
+      "step": 849
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.0862219093692516,
+      "learning_rate": 1.181533914167205e-06,
+      "loss": 0.3788,
+      "step": 850
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.7005854621422525,
+      "learning_rate": 1.1719114717962476e-06,
+      "loss": 0.3615,
+      "step": 851
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.9223633799154827,
+      "learning_rate": 1.1623231683561249e-06,
+      "loss": 0.4191,
+      "step": 852
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.5355028370021262,
+      "learning_rate": 1.1527690893554154e-06,
+      "loss": 0.4084,
+      "step": 853
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.3409107385851025,
+      "learning_rate": 1.143249319997491e-06,
+      "loss": 0.3211,
+      "step": 854
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.5436127305633105,
+      "learning_rate": 1.1337639451797493e-06,
+      "loss": 0.3983,
+      "step": 855
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.5214360728755154,
+      "learning_rate": 1.1243130494928533e-06,
+      "loss": 0.3446,
+      "step": 856
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.5727956777063972,
+      "learning_rate": 1.1148967172199848e-06,
+      "loss": 0.3665,
+      "step": 857
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.6801572434918612,
+      "learning_rate": 1.1055150323360852e-06,
+      "loss": 0.3841,
+      "step": 858
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.7269406372047564,
+      "learning_rate": 1.0961680785071117e-06,
+      "loss": 0.4971,
+      "step": 859
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.661213058624383,
+      "learning_rate": 1.0868559390892903e-06,
+      "loss": 0.3375,
+      "step": 860
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.7080284683248386,
+      "learning_rate": 1.0775786971283725e-06,
+      "loss": 0.351,
+      "step": 861
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.4422313394158195,
+      "learning_rate": 1.06833643535889e-06,
+      "loss": 0.398,
+      "step": 862
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.5369162182430969,
+      "learning_rate": 1.0591292362034255e-06,
+      "loss": 0.4384,
+      "step": 863
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.4444746013578487,
+      "learning_rate": 1.0499571817718707e-06,
+      "loss": 0.2999,
+      "step": 864
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.997456484534967,
+      "learning_rate": 1.0408203538606948e-06,
+      "loss": 0.3597,
+      "step": 865
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.6400301763903629,
+      "learning_rate": 1.0317188339522188e-06,
+      "loss": 0.3436,
+      "step": 866
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.6663652100357638,
+      "learning_rate": 1.0226527032138877e-06,
+      "loss": 0.3255,
+      "step": 867
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.4660952325217926,
+      "learning_rate": 1.0136220424975434e-06,
+      "loss": 0.3129,
+      "step": 868
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.068851578412713,
+      "learning_rate": 1.0046269323387037e-06,
+      "loss": 0.3851,
+      "step": 869
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.4622200769565,
+      "learning_rate": 9.956674529558518e-07,
+      "loss": 0.3648,
+      "step": 870
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.6996589952235117,
+      "learning_rate": 9.867436842497103e-07,
+      "loss": 0.473,
+      "step": 871
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.6621520206328781,
+      "learning_rate": 9.778557058025357e-07,
+      "loss": 0.3517,
+      "step": 872
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.3571771711435245,
+      "learning_rate": 9.69003596877408e-07,
+      "loss": 0.3822,
+      "step": 873
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.868120941242638,
+      "learning_rate": 9.601874364175206e-07,
+      "loss": 0.3294,
+      "step": 874
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.6696207498752886,
+      "learning_rate": 9.514073030454763e-07,
+      "loss": 0.3205,
+      "step": 875
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.0728764939030246,
+      "learning_rate": 9.426632750625919e-07,
+      "loss": 0.4558,
+      "step": 876
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.880333599994576,
+      "learning_rate": 9.339554304481952e-07,
+      "loss": 0.4021,
+      "step": 877
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.9734543015644022,
+      "learning_rate": 9.252838468589265e-07,
+      "loss": 0.3466,
+      "step": 878
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.3057676108168481,
+      "learning_rate": 9.166486016280562e-07,
+      "loss": 0.3686,
+      "step": 879
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.6983493502731448,
+      "learning_rate": 9.080497717647841e-07,
+      "loss": 0.4109,
+      "step": 880
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.6293001419023172,
+      "learning_rate": 8.994874339535569e-07,
+      "loss": 0.288,
+      "step": 881
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.4963241842091834,
+      "learning_rate": 8.909616645533886e-07,
+      "loss": 0.2871,
+      "step": 882
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.4822389606951585,
+      "learning_rate": 8.824725395971745e-07,
+      "loss": 0.3439,
+      "step": 883
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.7177199805665169,
+      "learning_rate": 8.740201347910133e-07,
+      "loss": 0.3232,
+      "step": 884
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 1.5014572884433388,
+      "learning_rate": 8.656045255135314e-07,
+      "loss": 0.3401,
+      "step": 885
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 1.5310901313894647,
+      "learning_rate": 8.572257868152173e-07,
+      "loss": 0.3108,
+      "step": 886
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 1.8246210773553824,
+      "learning_rate": 8.488839934177424e-07,
+      "loss": 0.3941,
+      "step": 887
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.074647338585339,
+      "learning_rate": 8.405792197133023e-07,
+      "loss": 0.3768,
+      "step": 888
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 1.6233824464176518,
+      "learning_rate": 8.323115397639514e-07,
+      "loss": 0.3695,
+      "step": 889
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 1.9626326451926888,
+      "learning_rate": 8.240810273009381e-07,
+      "loss": 0.3497,
+      "step": 890
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 1.72705310438082,
+      "learning_rate": 8.15887755724053e-07,
+      "loss": 0.4354,
+      "step": 891
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.2193068673729988,
+      "learning_rate": 8.077317981009731e-07,
+      "loss": 0.4535,
+      "step": 892
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 1.412893314116447,
+      "learning_rate": 7.996132271666062e-07,
+      "loss": 0.3713,
+      "step": 893
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 1.439263478672628,
+      "learning_rate": 7.915321153224487e-07,
+      "loss": 0.2963,
+      "step": 894
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 1.742179158576329,
+      "learning_rate": 7.834885346359361e-07,
+      "loss": 0.3215,
+      "step": 895
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.5193201275472255,
+      "learning_rate": 7.754825568397955e-07,
+      "loss": 0.3238,
+      "step": 896
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.9407844197719575,
+      "learning_rate": 7.675142533314173e-07,
+      "loss": 0.2893,
+      "step": 897
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.7124951515190545,
+      "learning_rate": 7.595836951722107e-07,
+      "loss": 0.3607,
+      "step": 898
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.6206596440963452,
+      "learning_rate": 7.516909530869687e-07,
+      "loss": 0.3338,
+      "step": 899
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.8379118207091338,
+      "learning_rate": 7.438360974632442e-07,
+      "loss": 0.4015,
+      "step": 900
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.0108442161255153,
+      "learning_rate": 7.360191983507154e-07,
+      "loss": 0.3681,
+      "step": 901
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.5768823402693721,
+      "learning_rate": 7.282403254605636e-07,
+      "loss": 0.318,
+      "step": 902
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.8501393420883374,
+      "learning_rate": 7.204995481648547e-07,
+      "loss": 0.3474,
+      "step": 903
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.8000174106515199,
+      "learning_rate": 7.127969354959164e-07,
+      "loss": 0.3711,
+      "step": 904
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.627552766175395,
+      "learning_rate": 7.051325561457217e-07,
+      "loss": 0.3521,
+      "step": 905
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 1.7703400679821946,
+      "learning_rate": 6.975064784652829e-07,
+      "loss": 0.2869,
+      "step": 906
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 1.9555640812703932,
+      "learning_rate": 6.899187704640326e-07,
+      "loss": 0.4446,
+      "step": 907
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.3485869295221904,
+      "learning_rate": 6.823694998092272e-07,
+      "loss": 0.3514,
+      "step": 908
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 1.6365626256545733,
+      "learning_rate": 6.748587338253338e-07,
+      "loss": 0.3263,
+      "step": 909
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 1.8406884572978746,
+      "learning_rate": 6.673865394934376e-07,
+      "loss": 0.3831,
+      "step": 910
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 1.8441311582841047,
+      "learning_rate": 6.59952983450643e-07,
+      "loss": 0.3672,
+      "step": 911
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 1.7850511624040863,
+      "learning_rate": 6.525581319894703e-07,
+      "loss": 0.344,
+      "step": 912
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 1.8426852257973754,
+      "learning_rate": 6.452020510572799e-07,
+      "loss": 0.3709,
+      "step": 913
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 1.8933625260612084,
+      "learning_rate": 6.378848062556742e-07,
+      "loss": 0.3625,
+      "step": 914
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 1.4364799820160647,
+      "learning_rate": 6.306064628399111e-07,
+      "loss": 0.3232,
+      "step": 915
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 1.8362485371909967,
+      "learning_rate": 6.23367085718331e-07,
+      "loss": 0.3444,
+      "step": 916
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.9245874517833783,
+      "learning_rate": 6.161667394517684e-07,
+      "loss": 0.3453,
+      "step": 917
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.082501618335563,
+      "learning_rate": 6.0900548825298e-07,
+      "loss": 0.3834,
+      "step": 918
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.6265615291277613,
+      "learning_rate": 6.018833959860753e-07,
+      "loss": 0.3871,
+      "step": 919
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.5527250089612845,
+      "learning_rate": 5.948005261659434e-07,
+      "loss": 0.2978,
+      "step": 920
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.7448719082955784,
+      "learning_rate": 5.877569419576851e-07,
+      "loss": 0.2905,
+      "step": 921
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.5352526453200945,
+      "learning_rate": 5.807527061760543e-07,
+      "loss": 0.3547,
+      "step": 922
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.010400248545671,
+      "learning_rate": 5.73787881284893e-07,
+      "loss": 0.3434,
+      "step": 923
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.08981264322849,
+      "learning_rate": 5.668625293965774e-07,
+      "loss": 0.3871,
+      "step": 924
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.544640538751419,
+      "learning_rate": 5.599767122714628e-07,
+      "loss": 0.3843,
+      "step": 925
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.50324978941004,
+      "learning_rate": 5.531304913173357e-07,
+      "loss": 0.3062,
+      "step": 926
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.7101075222047506,
+      "learning_rate": 5.463239275888599e-07,
+      "loss": 0.3505,
+      "step": 927
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.0135553262312933,
+      "learning_rate": 5.395570817870361e-07,
+      "loss": 0.4003,
+      "step": 928
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 1.64919292631168,
+      "learning_rate": 5.328300142586628e-07,
+      "loss": 0.4257,
+      "step": 929
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 1.497483142992759,
+      "learning_rate": 5.261427849957928e-07,
+      "loss": 0.3368,
+      "step": 930
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 1.578537761793323,
+      "learning_rate": 5.194954536352021e-07,
+      "loss": 0.3491,
+      "step": 931
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.144696531083044,
+      "learning_rate": 5.128880794578573e-07,
+      "loss": 0.4249,
+      "step": 932
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.216467711281768,
+      "learning_rate": 5.063207213883858e-07,
+      "loss": 0.4191,
+      "step": 933
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 1.513179703987502,
+      "learning_rate": 4.99793437994549e-07,
+      "loss": 0.2815,
+      "step": 934
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.0210242568879364,
+      "learning_rate": 4.933062874867267e-07,
+      "loss": 0.5077,
+      "step": 935
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 1.5808310904304832,
+      "learning_rate": 4.868593277173878e-07,
+      "loss": 0.3387,
+      "step": 936
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 1.9352852569545673,
+      "learning_rate": 4.804526161805834e-07,
+      "loss": 0.4165,
+      "step": 937
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 1.4953454574921947,
+      "learning_rate": 4.7408621001143076e-07,
+      "loss": 0.3195,
+      "step": 938
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 1.7650338603997144,
+      "learning_rate": 4.6776016598560125e-07,
+      "loss": 0.2577,
+      "step": 939
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 1.5257201063214694,
+      "learning_rate": 4.614745405188159e-07,
+      "loss": 0.4102,
+      "step": 940
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 1.5975131026855944,
+      "learning_rate": 4.5522938966634514e-07,
+      "loss": 0.2455,
+      "step": 941
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 1.7214214108104902,
+      "learning_rate": 4.490247691225058e-07,
+      "loss": 0.3679,
+      "step": 942
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 1.9125664855083362,
+      "learning_rate": 4.428607342201635e-07,
+      "loss": 0.4053,
+      "step": 943
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.23497833527835,
+      "learning_rate": 4.3673733993024103e-07,
+      "loss": 0.4223,
+      "step": 944
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 1.7278794126523196,
+      "learning_rate": 4.3065464086123055e-07,
+      "loss": 0.3194,
+      "step": 945
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 1.3338372611487148,
+      "learning_rate": 4.2461269125869896e-07,
+      "loss": 0.3989,
+      "step": 946
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 1.5824737042344048,
+      "learning_rate": 4.1861154500481285e-07,
+      "loss": 0.3773,
+      "step": 947
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 1.5501936663824547,
+      "learning_rate": 4.1265125561785466e-07,
+      "loss": 0.3073,
+      "step": 948
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 1.6744791716497447,
+      "learning_rate": 4.067318762517419e-07,
+      "loss": 0.3962,
+      "step": 949
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 1.6176701058731642,
+      "learning_rate": 4.0085345969555656e-07,
+      "loss": 0.2961,
+      "step": 950
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 1.668829643612793,
+      "learning_rate": 3.9501605837307613e-07,
+      "loss": 0.3983,
+      "step": 951
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 1.6415929342466784,
+      "learning_rate": 3.8921972434230183e-07,
+      "loss": 0.3354,
+      "step": 952
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 1.9107002582221606,
+      "learning_rate": 3.8346450929499735e-07,
+      "loss": 0.2923,
+      "step": 953
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 1.5837382493413166,
+      "learning_rate": 3.7775046455622853e-07,
+      "loss": 0.3572,
+      "step": 954
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 1.9578171945494112,
+      "learning_rate": 3.7207764108389835e-07,
+      "loss": 0.4438,
+      "step": 955
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.012312836551855,
+      "learning_rate": 3.664460894683036e-07,
+      "loss": 0.3921,
+      "step": 956
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 1.573634527392608,
+      "learning_rate": 3.6085585993167804e-07,
+      "loss": 0.3537,
+      "step": 957
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 1.87219129280588,
+      "learning_rate": 3.553070023277405e-07,
+      "loss": 0.394,
+      "step": 958
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 1.908976251642539,
+      "learning_rate": 3.4979956614125956e-07,
+      "loss": 0.3465,
+      "step": 959
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 1.7447296830368977,
+      "learning_rate": 3.4433360048760357e-07,
+      "loss": 0.4396,
+      "step": 960
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 1.4938127648980004,
+      "learning_rate": 3.389091541123074e-07,
+      "loss": 0.3833,
+      "step": 961
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 1.6511435214606442,
+      "learning_rate": 3.3352627539063707e-07,
+      "loss": 0.2598,
+      "step": 962
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 1.9044840155234906,
+      "learning_rate": 3.2818501232715794e-07,
+      "loss": 0.4915,
+      "step": 963
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 1.6336211800573899,
+      "learning_rate": 3.2288541255530546e-07,
+      "loss": 0.3413,
+      "step": 964
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 1.352176707831654,
+      "learning_rate": 3.1762752333696303e-07,
+      "loss": 0.368,
+      "step": 965
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 1.515583060896109,
+      "learning_rate": 3.124113915620375e-07,
+      "loss": 0.3465,
+      "step": 966
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 1.454044534368895,
+      "learning_rate": 3.072370637480415e-07,
+      "loss": 0.2625,
+      "step": 967
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 1.767987541706428,
+      "learning_rate": 3.0210458603968264e-07,
+      "loss": 0.3992,
+      "step": 968
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 1.7273964826434631,
+      "learning_rate": 2.970140042084474e-07,
+      "loss": 0.3708,
+      "step": 969
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 1.647723427115109,
+      "learning_rate": 2.919653636521935e-07,
+      "loss": 0.3525,
+      "step": 970
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 1.8493344960971922,
+      "learning_rate": 2.8695870939474626e-07,
+      "loss": 0.3229,
+      "step": 971
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 1.794125649715343,
+      "learning_rate": 2.8199408608549696e-07,
+      "loss": 0.3222,
+      "step": 972
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 1.9721188133776415,
+      "learning_rate": 2.7707153799900685e-07,
+      "loss": 0.3836,
+      "step": 973
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 1.6417411860470836,
+      "learning_rate": 2.721911090346052e-07,
+      "loss": 0.3736,
+      "step": 974
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 1.938690974748052,
+      "learning_rate": 2.673528427160066e-07,
+      "loss": 0.3205,
+      "step": 975
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 1.7687245607064443,
+      "learning_rate": 2.6255678219091754e-07,
+      "loss": 0.368,
+      "step": 976
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 1.9023761904635534,
+      "learning_rate": 2.578029702306506e-07,
+      "loss": 0.3177,
+      "step": 977
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 1.4839406363342937,
+      "learning_rate": 2.530914492297487e-07,
+      "loss": 0.3611,
+      "step": 978
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.3385861771964014,
+      "learning_rate": 2.484222612056025e-07,
+      "loss": 0.3316,
+      "step": 979
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 1.9746739789044454,
+      "learning_rate": 2.437954477980753e-07,
+      "loss": 0.3969,
+      "step": 980
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 1.6887922769065924,
+      "learning_rate": 2.392110502691353e-07,
+      "loss": 0.3968,
+      "step": 981
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.175353517627243,
+      "learning_rate": 2.3466910950248334e-07,
+      "loss": 0.3524,
+      "step": 982
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 1.9590405768224393,
+      "learning_rate": 2.3016966600319158e-07,
+      "loss": 0.4014,
+      "step": 983
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 1.7539702537980744,
+      "learning_rate": 2.2571275989734075e-07,
+      "loss": 0.3558,
+      "step": 984
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 1.8410012610197661,
+      "learning_rate": 2.212984309316646e-07,
+      "loss": 0.4289,
+      "step": 985
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 1.74770502941488,
+      "learning_rate": 2.1692671847319046e-07,
+      "loss": 0.4004,
+      "step": 986
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 1.444442338435704,
+      "learning_rate": 2.125976615088926e-07,
+      "loss": 0.3453,
+      "step": 987
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 1.8660604720500005,
+      "learning_rate": 2.083112986453445e-07,
+      "loss": 0.3741,
+      "step": 988
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 1.6568713310219376,
+      "learning_rate": 2.040676681083703e-07,
+      "loss": 0.4042,
+      "step": 989
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 1.387457724388811,
+      "learning_rate": 1.998668077427096e-07,
+      "loss": 0.2919,
+      "step": 990
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 1.3853710037974833,
+      "learning_rate": 1.9570875501167651e-07,
+      "loss": 0.2827,
+      "step": 991
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 1.5185303113021482,
+      "learning_rate": 1.9159354699682496e-07,
+      "loss": 0.3902,
+      "step": 992
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.9015197180656451,
+      "learning_rate": 1.8752122039762011e-07,
+      "loss": 0.3062,
+      "step": 993
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.281701595570607,
+      "learning_rate": 1.8349181153111074e-07,
+      "loss": 0.4513,
+      "step": 994
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.5802871368090137,
+      "learning_rate": 1.7950535633160404e-07,
+      "loss": 0.3702,
+      "step": 995
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.5604086315052326,
+      "learning_rate": 1.7556189035034642e-07,
+      "loss": 0.3407,
+      "step": 996
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.4540871479317765,
+      "learning_rate": 1.7166144875520762e-07,
+      "loss": 0.2945,
+      "step": 997
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.6711052975721497,
+      "learning_rate": 1.6780406633036094e-07,
+      "loss": 0.3555,
+      "step": 998
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.4918015703526908,
+      "learning_rate": 1.6398977747598245e-07,
+      "loss": 0.3745,
+      "step": 999
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.523355764764963,
+      "learning_rate": 1.6021861620793666e-07,
+      "loss": 0.2958,
+      "step": 1000
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.933124025444002,
+      "learning_rate": 1.564906161574764e-07,
+      "loss": 0.3181,
+      "step": 1001
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.5606282051722882,
+      "learning_rate": 1.5280581057094346e-07,
+      "loss": 0.2937,
+      "step": 1002
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.9267494105876177,
+      "learning_rate": 1.4916423230946885e-07,
+      "loss": 0.4209,
+      "step": 1003
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 1.547866759100206,
+      "learning_rate": 1.4556591384868368e-07,
+      "loss": 0.3583,
+      "step": 1004
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.162739247908966,
+      "learning_rate": 1.4201088727842648e-07,
+      "loss": 0.4472,
+      "step": 1005
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.074387460144302,
+      "learning_rate": 1.3849918430245911e-07,
+      "loss": 0.4204,
+      "step": 1006
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 1.75405906319246,
+      "learning_rate": 1.3503083623818413e-07,
+      "loss": 0.3286,
+      "step": 1007
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.208827466327546,
+      "learning_rate": 1.316058740163617e-07,
+      "loss": 0.4257,
+      "step": 1008
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 1.9131618964245363,
+      "learning_rate": 1.282243281808393e-07,
+      "loss": 0.3653,
+      "step": 1009
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 1.6704670693024548,
+      "learning_rate": 1.2488622888827517e-07,
+      "loss": 0.4076,
+      "step": 1010
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 1.7005183654466214,
+      "learning_rate": 1.2159160590787145e-07,
+      "loss": 0.4183,
+      "step": 1011
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 1.7590126284604837,
+      "learning_rate": 1.1834048862110815e-07,
+      "loss": 0.3497,
+      "step": 1012
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 1.8495558159166359,
+      "learning_rate": 1.1513290602148175e-07,
+      "loss": 0.3989,
+      "step": 1013
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 1.5491493514285797,
+      "learning_rate": 1.1196888671424378e-07,
+      "loss": 0.4034,
+      "step": 1014
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 1.696992836044335,
+      "learning_rate": 1.0884845891614926e-07,
+      "loss": 0.3906,
+      "step": 1015
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 1.6104580839191522,
+      "learning_rate": 1.0577165045520532e-07,
+      "loss": 0.343,
+      "step": 1016
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 1.7381991868528064,
+      "learning_rate": 1.0273848877041804e-07,
+      "loss": 0.4373,
+      "step": 1017
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.0710854401837224,
+      "learning_rate": 9.974900091155425e-08,
+      "loss": 0.406,
+      "step": 1018
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.594545766782277,
+      "learning_rate": 9.680321353889577e-08,
+      "loss": 0.3871,
+      "step": 1019
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 1.9064797609695343,
+      "learning_rate": 9.390115292300161e-08,
+      "loss": 0.4445,
+      "step": 1020
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 1.8137268442191112,
+      "learning_rate": 9.10428449444778e-08,
+      "loss": 0.3649,
+      "step": 1021
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 1.632472777376261,
+      "learning_rate": 8.822831509374297e-08,
+      "loss": 0.3455,
+      "step": 1022
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 1.7487886157086556,
+      "learning_rate": 8.545758847080144e-08,
+      "loss": 0.3152,
+      "step": 1023
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 1.4612562835731007,
+      "learning_rate": 8.273068978501997e-08,
+      "loss": 0.3597,
+      "step": 1024
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 1.7201161847179576,
+      "learning_rate": 8.004764335490856e-08,
+      "loss": 0.3958,
+      "step": 1025
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 1.8017582089862216,
+      "learning_rate": 7.74084731079e-08,
+      "loss": 0.3913,
+      "step": 1026
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 1.3234352702837793,
+      "learning_rate": 7.481320258014124e-08,
+      "loss": 0.3034,
+      "step": 1027
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 1.3724778893397676,
+      "learning_rate": 7.226185491628069e-08,
+      "loss": 0.2954,
+      "step": 1028
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 1.6969184913162358,
+      "learning_rate": 6.975445286926064e-08,
+      "loss": 0.3897,
+      "step": 1029
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 1.9317477089546637,
+      "learning_rate": 6.729101880011746e-08,
+      "loss": 0.377,
+      "step": 1030
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 1.554878777893729,
+      "learning_rate": 6.48715746777806e-08,
+      "loss": 0.4087,
+      "step": 1031
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 1.7534864024515837,
+      "learning_rate": 6.2496142078875e-08,
+      "loss": 0.4967,
+      "step": 1032
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.4804685848961765,
+      "learning_rate": 6.016474218753287e-08,
+      "loss": 0.4207,
+      "step": 1033
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 1.612223174759301,
+      "learning_rate": 5.787739579520113e-08,
+      "loss": 0.4037,
+      "step": 1034
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 1.4804110785714588,
+      "learning_rate": 5.5634123300457585e-08,
+      "loss": 0.2889,
+      "step": 1035
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 1.8113911538397998,
+      "learning_rate": 5.343494470882671e-08,
+      "loss": 0.362,
+      "step": 1036
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.7167779730798833,
+      "learning_rate": 5.127987963260583e-08,
+      "loss": 0.3482,
+      "step": 1037
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.4150092955119626,
+      "learning_rate": 4.916894729068644e-08,
+      "loss": 0.2178,
+      "step": 1038
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.633077629302677,
+      "learning_rate": 4.7102166508383173e-08,
+      "loss": 0.3825,
+      "step": 1039
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.425048677014493,
+      "learning_rate": 4.5079555717267855e-08,
+      "loss": 0.3305,
+      "step": 1040
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.9174655190388756,
+      "learning_rate": 4.31011329550024e-08,
+      "loss": 0.3396,
+      "step": 1041
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.3843933529043337,
+      "learning_rate": 4.1166915865180625e-08,
+      "loss": 0.2631,
+      "step": 1042
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.4653796486037136,
+      "learning_rate": 3.927692169716946e-08,
+      "loss": 0.3227,
+      "step": 1043
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.635515966199329,
+      "learning_rate": 3.743116730595575e-08,
+      "loss": 0.3638,
+      "step": 1044
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.0640494844558477,
+      "learning_rate": 3.562966915199473e-08,
+      "loss": 0.363,
+      "step": 1045
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.5455021482175313,
+      "learning_rate": 3.387244330106454e-08,
+      "loss": 0.3479,
+      "step": 1046
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.8431655117744339,
+      "learning_rate": 3.2159505424122495e-08,
+      "loss": 0.328,
+      "step": 1047
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 1.4359979387464212,
+      "learning_rate": 3.049087079716462e-08,
+      "loss": 0.3124,
+      "step": 1048
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 1.680594322782353,
+      "learning_rate": 2.8866554301091863e-08,
+      "loss": 0.4445,
+      "step": 1049
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 1.6887895844276681,
+      "learning_rate": 2.7286570421574677e-08,
+      "loss": 0.3209,
+      "step": 1050
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 1.7841936111238237,
+      "learning_rate": 2.575093324892364e-08,
+      "loss": 0.3856,
+      "step": 1051
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 1.7816368074530717,
+      "learning_rate": 2.42596564779668e-08,
+      "loss": 0.4697,
+      "step": 1052
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 1.5805844941387572,
+      "learning_rate": 2.281275340792477e-08,
+      "loss": 0.368,
+      "step": 1053
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 1.637820628172062,
+      "learning_rate": 2.141023694229305e-08,
+      "loss": 0.3718,
+      "step": 1054
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 1.417147920836824,
+      "learning_rate": 2.0052119588727105e-08,
+      "loss": 0.3546,
+      "step": 1055
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 1.989312613003882,
+      "learning_rate": 1.873841345893135e-08,
+      "loss": 0.4529,
+      "step": 1056
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 1.537277887843072,
+      "learning_rate": 1.746913026854924e-08,
+      "loss": 0.3583,
+      "step": 1057
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 1.8191477288789326,
+      "learning_rate": 1.6244281337060575e-08,
+      "loss": 0.3171,
+      "step": 1058
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 1.827847238198048,
+      "learning_rate": 1.506387758768102e-08,
+      "loss": 0.3793,
+      "step": 1059
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 1.7778694252802711,
+      "learning_rate": 1.3927929547261632e-08,
+      "loss": 0.3753,
+      "step": 1060
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 1.543634711451779,
+      "learning_rate": 1.2836447346198932e-08,
+      "loss": 0.338,
+      "step": 1061
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.16029561176065,
+      "learning_rate": 1.1789440718341094e-08,
+      "loss": 0.3576,
+      "step": 1062
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.2006580842983734,
+      "learning_rate": 1.0786919000903562e-08,
+      "loss": 0.3864,
+      "step": 1063
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 1.7044121149329359,
+      "learning_rate": 9.828891134385787e-09,
+      "loss": 0.344,
+      "step": 1064
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 1.4977033409076637,
+      "learning_rate": 8.915365662488517e-09,
+      "loss": 0.3146,
+      "step": 1065
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 1.4691930868903902,
+      "learning_rate": 8.046350732041075e-09,
+      "loss": 0.3099,
+      "step": 1066
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 1.5513156194536561,
+      "learning_rate": 7.221854092926972e-09,
+      "loss": 0.341,
+      "step": 1067
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 1.703611443801933,
+      "learning_rate": 6.441883098015633e-09,
+      "loss": 0.3734,
+      "step": 1068
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.1506089060675455,
+      "learning_rate": 5.706444703096337e-09,
+      "loss": 0.3859,
+      "step": 1069
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 1.621345533853751,
+      "learning_rate": 5.015545466814931e-09,
+      "loss": 0.405,
+      "step": 1070
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 1.8186009796771267,
+      "learning_rate": 4.369191550617768e-09,
+      "loss": 0.3964,
+      "step": 1071
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 1.8157548418762393,
+      "learning_rate": 3.7673887186961965e-09,
+      "loss": 0.4728,
+      "step": 1072
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.070715007431859,
+      "learning_rate": 3.210142337932709e-09,
+      "loss": 0.464,
+      "step": 1073
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 1.6191854446760734,
+      "learning_rate": 2.6974573778565383e-09,
+      "loss": 0.3648,
+      "step": 1074
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 1.63053598233112,
+      "learning_rate": 2.229338410597026e-09,
+      "loss": 0.4364,
+      "step": 1075
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 1.5220833342814482,
+      "learning_rate": 1.8057896108436557e-09,
+      "loss": 0.4131,
+      "step": 1076
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 1.9238625872095938,
+      "learning_rate": 1.4268147558088585e-09,
+      "loss": 0.4702,
+      "step": 1077
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 1.6227133478707174,
+      "learning_rate": 1.0924172251941533e-09,
+      "loss": 0.3781,
+      "step": 1078
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 1.5844245279524536,
+      "learning_rate": 8.026000011596147e-10,
+      "loss": 0.367,
+      "step": 1079
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.8755295726944123,
+      "learning_rate": 5.573656682977824e-10,
+      "loss": 0.3954,
+      "step": 1080
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 0.4678705930709839,
+      "eval_runtime": 533.0785,
+      "eval_samples_per_second": 5.084,
+      "eval_steps_per_second": 0.319,
+      "step": 1080
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.95867908444184,
+      "learning_rate": 3.567164136120127e-10,
+      "loss": 0.3425,
+      "step": 1081
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.9895722241127884,
+      "learning_rate": 2.0065402649371845e-10,
+      "loss": 0.3204,
+      "step": 1082
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.5770678955265505,
+      "learning_rate": 8.917989870849131e-11,
+      "loss": 0.3511,
+      "step": 1083
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.7159701303303903,
+      "learning_rate": 2.2295024383889307e-11,
+      "loss": 0.2967,
+      "step": 1084
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.9974547783910455,
+      "learning_rate": 0.0,
+      "loss": 0.3133,
+      "step": 1085
+    },
+    {
+      "epoch": 1.0,
+      "step": 1085,
+      "total_flos": 2242775013654528.0,
+      "train_loss": 0.4361060568951242,
+      "train_runtime": 100289.4425,
+      "train_samples_per_second": 0.346,
+      "train_steps_per_second": 0.011
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 1085,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 100.0,
+  "total_flos": 2242775013654528.0,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}