diff --git "a/1400/trainer_state.json" "b/1400/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/1400/trainer_state.json"
@@ -0,0 +1,9821 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.5500910746812386,
+  "eval_steps": 500,
+  "global_step": 1400,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.46431864432643544,
+      "learning_rate": 1.2121212121212122e-06,
+      "loss": 1.4151,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.42969176658539837,
+      "learning_rate": 2.4242424242424244e-06,
+      "loss": 1.3729,
+      "step": 2
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.5004307936270223,
+      "learning_rate": 3.636363636363636e-06,
+      "loss": 1.3989,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.43666634920041486,
+      "learning_rate": 4.848484848484849e-06,
+      "loss": 1.3363,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.4691114419353825,
+      "learning_rate": 6.060606060606061e-06,
+      "loss": 1.4293,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.4277596061377729,
+      "learning_rate": 7.272727272727272e-06,
+      "loss": 1.4343,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.4238339229382504,
+      "learning_rate": 8.484848484848486e-06,
+      "loss": 1.4462,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.40992048534183273,
+      "learning_rate": 9.696969696969698e-06,
+      "loss": 1.2756,
+      "step": 8
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.37885700313540693,
+      "learning_rate": 1.0909090909090909e-05,
+      "loss": 1.3464,
+      "step": 9
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.36066141170123023,
+      "learning_rate": 1.2121212121212122e-05,
+      "loss": 1.3419,
+      "step": 10
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.35617169386863406,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 1.3533,
+      "step": 11
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.3040188564782602,
+      "learning_rate": 1.4545454545454545e-05,
+      "loss": 1.2395,
+      "step": 12
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.31038319439216566,
+      "learning_rate": 1.5757575757575756e-05,
+      "loss": 1.3082,
+      "step": 13
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.26683768372135835,
+      "learning_rate": 1.6969696969696972e-05,
+      "loss": 1.3063,
+      "step": 14
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.3652323682563078,
+      "learning_rate": 1.8181818181818182e-05,
+      "loss": 1.3045,
+      "step": 15
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.23559121485457843,
+      "learning_rate": 1.9393939393939395e-05,
+      "loss": 1.2366,
+      "step": 16
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.2342299313020104,
+      "learning_rate": 2.0606060606060608e-05,
+      "loss": 1.2831,
+      "step": 17
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.2202931700357255,
+      "learning_rate": 2.1818181818181818e-05,
+      "loss": 1.3064,
+      "step": 18
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.2097660599292375,
+      "learning_rate": 2.3030303030303034e-05,
+      "loss": 1.2376,
+      "step": 19
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.2356785314652122,
+      "learning_rate": 2.4242424242424244e-05,
+      "loss": 1.2802,
+      "step": 20
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.24639302530564244,
+      "learning_rate": 2.5454545454545454e-05,
+      "loss": 1.3016,
+      "step": 21
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.24373126133228787,
+      "learning_rate": 2.6666666666666667e-05,
+      "loss": 1.3407,
+      "step": 22
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.24488805144123432,
+      "learning_rate": 2.7878787878787883e-05,
+      "loss": 1.3325,
+      "step": 23
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.2653033507571198,
+      "learning_rate": 2.909090909090909e-05,
+      "loss": 1.2811,
+      "step": 24
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.2841724819336817,
+      "learning_rate": 3.0303030303030306e-05,
+      "loss": 1.2837,
+      "step": 25
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.2183883020111492,
+      "learning_rate": 3.151515151515151e-05,
+      "loss": 1.2472,
+      "step": 26
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.2137995163762026,
+      "learning_rate": 3.272727272727273e-05,
+      "loss": 1.2854,
+      "step": 27
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.19499006223503876,
+      "learning_rate": 3.3939393939393945e-05,
+      "loss": 1.3018,
+      "step": 28
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.17367919355340256,
+      "learning_rate": 3.515151515151515e-05,
+      "loss": 1.2824,
+      "step": 29
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.18326045693683557,
+      "learning_rate": 3.6363636363636364e-05,
+      "loss": 1.2192,
+      "step": 30
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.17474388188066411,
+      "learning_rate": 3.757575757575758e-05,
+      "loss": 1.2078,
+      "step": 31
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.17856970178098716,
+      "learning_rate": 3.878787878787879e-05,
+      "loss": 1.2683,
+      "step": 32
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.18617589704298348,
+      "learning_rate": 4e-05,
+      "loss": 1.2265,
+      "step": 33
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.17653209733215317,
+      "learning_rate": 4.1212121212121216e-05,
+      "loss": 1.319,
+      "step": 34
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.1722921367585233,
+      "learning_rate": 4.242424242424243e-05,
+      "loss": 1.2117,
+      "step": 35
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.176642606378719,
+      "learning_rate": 4.3636363636363636e-05,
+      "loss": 1.2512,
+      "step": 36
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.16696442324691066,
+      "learning_rate": 4.484848484848485e-05,
+      "loss": 1.2637,
+      "step": 37
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.17035384059517106,
+      "learning_rate": 4.606060606060607e-05,
+      "loss": 1.2699,
+      "step": 38
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.15545801881444482,
+      "learning_rate": 4.7272727272727275e-05,
+      "loss": 1.2939,
+      "step": 39
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.17111439344347512,
+      "learning_rate": 4.848484848484849e-05,
+      "loss": 1.3033,
+      "step": 40
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.16994151343455458,
+      "learning_rate": 4.9696969696969694e-05,
+      "loss": 1.2603,
+      "step": 41
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.15929214926453447,
+      "learning_rate": 5.090909090909091e-05,
+      "loss": 1.2626,
+      "step": 42
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.16761261516238699,
+      "learning_rate": 5.212121212121213e-05,
+      "loss": 1.296,
+      "step": 43
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.15754700542426123,
+      "learning_rate": 5.333333333333333e-05,
+      "loss": 1.278,
+      "step": 44
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.15522526683877644,
+      "learning_rate": 5.4545454545454546e-05,
+      "loss": 1.2355,
+      "step": 45
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.1577929926930023,
+      "learning_rate": 5.5757575757575766e-05,
+      "loss": 1.2879,
+      "step": 46
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.31075066632858317,
+      "learning_rate": 5.696969696969697e-05,
+      "loss": 1.2202,
+      "step": 47
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.1663780653395111,
+      "learning_rate": 5.818181818181818e-05,
+      "loss": 1.2319,
+      "step": 48
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.16049499655883026,
+      "learning_rate": 5.93939393939394e-05,
+      "loss": 1.2801,
+      "step": 49
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.14515773124436285,
+      "learning_rate": 6.060606060606061e-05,
+      "loss": 1.2588,
+      "step": 50
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.14653064850325623,
+      "learning_rate": 6.181818181818182e-05,
+      "loss": 1.2677,
+      "step": 51
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.17193239746689878,
+      "learning_rate": 6.303030303030302e-05,
+      "loss": 1.2742,
+      "step": 52
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.1967020450342533,
+      "learning_rate": 6.424242424242424e-05,
+      "loss": 1.1545,
+      "step": 53
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.16247531997247225,
+      "learning_rate": 6.545454545454546e-05,
+      "loss": 1.222,
+      "step": 54
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.14990706377244528,
+      "learning_rate": 6.666666666666667e-05,
+      "loss": 1.2103,
+      "step": 55
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.1412817445239095,
+      "learning_rate": 6.787878787878789e-05,
+      "loss": 1.2169,
+      "step": 56
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.14575971073482757,
+      "learning_rate": 6.90909090909091e-05,
+      "loss": 1.2751,
+      "step": 57
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.13714747569950891,
+      "learning_rate": 7.03030303030303e-05,
+      "loss": 1.2508,
+      "step": 58
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.14334695156859903,
+      "learning_rate": 7.151515151515152e-05,
+      "loss": 1.2721,
+      "step": 59
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.1456824177522916,
+      "learning_rate": 7.272727272727273e-05,
+      "loss": 1.2649,
+      "step": 60
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.15030318240210044,
+      "learning_rate": 7.393939393939395e-05,
+      "loss": 1.2167,
+      "step": 61
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.1651326066719482,
+      "learning_rate": 7.515151515151515e-05,
+      "loss": 1.3126,
+      "step": 62
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.1408250406479118,
+      "learning_rate": 7.636363636363637e-05,
+      "loss": 1.2891,
+      "step": 63
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.21501384376905694,
+      "learning_rate": 7.757575757575758e-05,
+      "loss": 1.3019,
+      "step": 64
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.1365168726167339,
+      "learning_rate": 7.878787878787879e-05,
+      "loss": 1.2498,
+      "step": 65
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.1431463689660936,
+      "learning_rate": 8e-05,
+      "loss": 1.2793,
+      "step": 66
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.13689045214286194,
+      "learning_rate": 8.121212121212121e-05,
+      "loss": 1.2295,
+      "step": 67
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.13483608710081227,
+      "learning_rate": 8.242424242424243e-05,
+      "loss": 1.2258,
+      "step": 68
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.13707618564415613,
+      "learning_rate": 8.363636363636364e-05,
+      "loss": 1.2252,
+      "step": 69
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.13780236215967515,
+      "learning_rate": 8.484848484848486e-05,
+      "loss": 1.2565,
+      "step": 70
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.14036805493494423,
+      "learning_rate": 8.606060606060606e-05,
+      "loss": 1.3023,
+      "step": 71
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.12776919439147982,
+      "learning_rate": 8.727272727272727e-05,
+      "loss": 1.2292,
+      "step": 72
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.1289941815481437,
+      "learning_rate": 8.848484848484849e-05,
+      "loss": 1.2191,
+      "step": 73
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.13943952294847306,
+      "learning_rate": 8.96969696969697e-05,
+      "loss": 1.2915,
+      "step": 74
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.1493528502117281,
+      "learning_rate": 9.090909090909092e-05,
+      "loss": 1.2797,
+      "step": 75
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.1252401242451818,
+      "learning_rate": 9.212121212121214e-05,
+      "loss": 1.2552,
+      "step": 76
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.13969800467546992,
+      "learning_rate": 9.333333333333334e-05,
+      "loss": 1.3147,
+      "step": 77
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.1277258491470434,
+      "learning_rate": 9.454545454545455e-05,
+      "loss": 1.2089,
+      "step": 78
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.133041369314817,
+      "learning_rate": 9.575757575757576e-05,
+      "loss": 1.2761,
+      "step": 79
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.14564572037181842,
+      "learning_rate": 9.696969696969698e-05,
+      "loss": 1.1901,
+      "step": 80
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.13666505656492195,
+      "learning_rate": 9.818181818181818e-05,
+      "loss": 1.2615,
+      "step": 81
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.135007805210003,
+      "learning_rate": 9.939393939393939e-05,
+      "loss": 1.2669,
+      "step": 82
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.17287563365884975,
+      "learning_rate": 0.00010060606060606062,
+      "loss": 1.2669,
+      "step": 83
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.12934306326048103,
+      "learning_rate": 0.00010181818181818181,
+      "loss": 1.1979,
+      "step": 84
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.13517436169178096,
+      "learning_rate": 0.00010303030303030303,
+      "loss": 1.2226,
+      "step": 85
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.12105351159271568,
+      "learning_rate": 0.00010424242424242425,
+      "loss": 1.1172,
+      "step": 86
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.1281676431775383,
+      "learning_rate": 0.00010545454545454545,
+      "loss": 1.2046,
+      "step": 87
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.11730963057933333,
+      "learning_rate": 0.00010666666666666667,
+      "loss": 1.1883,
+      "step": 88
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.12655235108503246,
+      "learning_rate": 0.00010787878787878789,
+      "loss": 1.1331,
+      "step": 89
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.13047560307970027,
+      "learning_rate": 0.00010909090909090909,
+      "loss": 1.2731,
+      "step": 90
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.12193522973752649,
+      "learning_rate": 0.00011030303030303031,
+      "loss": 1.2161,
+      "step": 91
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.12804360300116346,
+      "learning_rate": 0.00011151515151515153,
+      "loss": 1.3062,
+      "step": 92
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.15991741754516206,
+      "learning_rate": 0.00011272727272727272,
+      "loss": 1.239,
+      "step": 93
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.15140182244454561,
+      "learning_rate": 0.00011393939393939394,
+      "loss": 1.2349,
+      "step": 94
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.12320241076263434,
+      "learning_rate": 0.00011515151515151516,
+      "loss": 1.2875,
+      "step": 95
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.13235998458230466,
+      "learning_rate": 0.00011636363636363636,
+      "loss": 1.2218,
+      "step": 96
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.11783688734798668,
+      "learning_rate": 0.00011757575757575758,
+      "loss": 1.1864,
+      "step": 97
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.3151933420750235,
+      "learning_rate": 0.0001187878787878788,
+      "loss": 1.3023,
+      "step": 98
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.12665632567219295,
+      "learning_rate": 0.00012,
+      "loss": 1.2249,
+      "step": 99
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.1228886740460738,
+      "learning_rate": 0.00012121212121212122,
+      "loss": 1.2517,
+      "step": 100
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.11892005244989344,
+      "learning_rate": 0.00012242424242424243,
+      "loss": 1.2586,
+      "step": 101
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.1232340827222201,
+      "learning_rate": 0.00012363636363636364,
+      "loss": 1.3217,
+      "step": 102
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.13837226869323116,
+      "learning_rate": 0.00012484848484848487,
+      "loss": 1.2693,
+      "step": 103
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.12068217991774362,
+      "learning_rate": 0.00012606060606060605,
+      "loss": 1.2623,
+      "step": 104
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.16779277284606545,
+      "learning_rate": 0.00012727272727272728,
+      "loss": 1.2415,
+      "step": 105
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.13396891539963085,
+      "learning_rate": 0.0001284848484848485,
+      "loss": 1.2313,
+      "step": 106
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.12457104490772812,
+      "learning_rate": 0.0001296969696969697,
+      "loss": 1.1758,
+      "step": 107
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.12676816816563452,
+      "learning_rate": 0.00013090909090909093,
+      "loss": 1.2478,
+      "step": 108
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.11973639622066906,
+      "learning_rate": 0.00013212121212121213,
+      "loss": 1.2335,
+      "step": 109
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.1330159646034068,
+      "learning_rate": 0.00013333333333333334,
+      "loss": 1.26,
+      "step": 110
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.1298003025338099,
+      "learning_rate": 0.00013454545454545455,
+      "loss": 1.1907,
+      "step": 111
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.1226154813287666,
+      "learning_rate": 0.00013575757575757578,
+      "loss": 1.1807,
+      "step": 112
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.12533753244302145,
+      "learning_rate": 0.00013696969696969696,
+      "loss": 1.2098,
+      "step": 113
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.12673266503840944,
+      "learning_rate": 0.0001381818181818182,
+      "loss": 1.2265,
+      "step": 114
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.1299039569361384,
+      "learning_rate": 0.0001393939393939394,
+      "loss": 1.2534,
+      "step": 115
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.13023496663090803,
+      "learning_rate": 0.0001406060606060606,
+      "loss": 1.2453,
+      "step": 116
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.12001793500864573,
+      "learning_rate": 0.00014181818181818184,
+      "loss": 1.1608,
+      "step": 117
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.14561862193041028,
+      "learning_rate": 0.00014303030303030304,
+      "loss": 1.2233,
+      "step": 118
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.12636130876430832,
+      "learning_rate": 0.00014424242424242425,
+      "loss": 1.2833,
+      "step": 119
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.189556849271166,
+      "learning_rate": 0.00014545454545454546,
+      "loss": 1.3105,
+      "step": 120
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.12409073764495662,
+      "learning_rate": 0.00014666666666666666,
+      "loss": 1.1534,
+      "step": 121
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.12149212466969316,
+      "learning_rate": 0.0001478787878787879,
+      "loss": 1.3039,
+      "step": 122
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.12147336887953522,
+      "learning_rate": 0.0001490909090909091,
+      "loss": 1.326,
+      "step": 123
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.1176585016163167,
+      "learning_rate": 0.0001503030303030303,
+      "loss": 1.191,
+      "step": 124
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.2066428974234372,
+      "learning_rate": 0.00015151515151515152,
+      "loss": 1.3054,
+      "step": 125
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.29582724255710047,
+      "learning_rate": 0.00015272727272727275,
+      "loss": 1.2032,
+      "step": 126
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.13084381204119358,
+      "learning_rate": 0.00015393939393939393,
+      "loss": 1.2289,
+      "step": 127
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.1294157600411397,
+      "learning_rate": 0.00015515151515151516,
+      "loss": 1.2561,
+      "step": 128
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.14039614543447027,
+      "learning_rate": 0.00015636363636363637,
+      "loss": 1.243,
+      "step": 129
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.19939984917282128,
+      "learning_rate": 0.00015757575757575757,
+      "loss": 1.1286,
+      "step": 130
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.14402764349968203,
+      "learning_rate": 0.0001587878787878788,
+      "loss": 1.1959,
+      "step": 131
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.13970978861500938,
+      "learning_rate": 0.00016,
+      "loss": 1.1814,
+      "step": 132
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.14539538472563127,
+      "learning_rate": 0.00016121212121212122,
+      "loss": 1.2317,
+      "step": 133
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.13456425455391557,
+      "learning_rate": 0.00016242424242424243,
+      "loss": 1.2239,
+      "step": 134
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.1314997837157779,
+      "learning_rate": 0.00016363636363636366,
+      "loss": 1.1986,
+      "step": 135
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.14046946525591422,
+      "learning_rate": 0.00016484848484848487,
+      "loss": 1.2238,
+      "step": 136
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.6095538041505763,
+      "learning_rate": 0.00016606060606060607,
+      "loss": 1.2332,
+      "step": 137
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.17707289712054367,
+      "learning_rate": 0.00016727272727272728,
+      "loss": 1.2401,
+      "step": 138
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.19335172179099247,
+      "learning_rate": 0.00016848484848484848,
+      "loss": 1.2361,
+      "step": 139
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.13725591818701255,
+      "learning_rate": 0.00016969696969696972,
+      "loss": 1.193,
+      "step": 140
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.15535575462507384,
+      "learning_rate": 0.0001709090909090909,
+      "loss": 1.2769,
+      "step": 141
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.14909436560898923,
+      "learning_rate": 0.00017212121212121213,
+      "loss": 1.2602,
+      "step": 142
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.15054368082407957,
+      "learning_rate": 0.00017333333333333334,
+      "loss": 1.2607,
+      "step": 143
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.13386897838741724,
+      "learning_rate": 0.00017454545454545454,
+      "loss": 1.168,
+      "step": 144
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.13567889528730145,
+      "learning_rate": 0.00017575757575757578,
+      "loss": 1.1984,
+      "step": 145
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.13994382298003089,
+      "learning_rate": 0.00017696969696969698,
+      "loss": 1.2795,
+      "step": 146
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.13941573210713187,
+      "learning_rate": 0.0001781818181818182,
+      "loss": 1.2303,
+      "step": 147
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.18302605925485763,
+      "learning_rate": 0.0001793939393939394,
+      "loss": 1.2696,
+      "step": 148
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.1547402223275396,
+      "learning_rate": 0.00018060606060606063,
+      "loss": 1.1276,
+      "step": 149
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.19947594494850646,
+      "learning_rate": 0.00018181818181818183,
+      "loss": 1.271,
+      "step": 150
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.1517101450465788,
+      "learning_rate": 0.00018303030303030304,
+      "loss": 1.2193,
+      "step": 151
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.19251063116857103,
+      "learning_rate": 0.00018424242424242427,
+      "loss": 1.2703,
+      "step": 152
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.16789099560498666,
+      "learning_rate": 0.00018545454545454545,
+      "loss": 1.2244,
+      "step": 153
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.14907376557922342,
+      "learning_rate": 0.0001866666666666667,
+      "loss": 1.264,
+      "step": 154
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.14276598263036905,
+      "learning_rate": 0.0001878787878787879,
+      "loss": 1.2545,
+      "step": 155
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.14526753816999002,
+      "learning_rate": 0.0001890909090909091,
+      "loss": 1.2912,
+      "step": 156
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.1627048894660859,
+      "learning_rate": 0.0001903030303030303,
+      "loss": 1.2573,
+      "step": 157
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.16405036632332695,
+      "learning_rate": 0.0001915151515151515,
+      "loss": 1.2359,
+      "step": 158
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.14533427219788658,
+      "learning_rate": 0.00019272727272727274,
+      "loss": 1.1718,
+      "step": 159
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.13802382666732702,
+      "learning_rate": 0.00019393939393939395,
+      "loss": 1.2297,
+      "step": 160
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.15620193618511755,
+      "learning_rate": 0.00019515151515151516,
+      "loss": 1.2287,
+      "step": 161
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.1401696295700075,
+      "learning_rate": 0.00019636363636363636,
+      "loss": 1.2231,
+      "step": 162
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.15816133304035035,
+      "learning_rate": 0.0001975757575757576,
+      "loss": 1.2804,
+      "step": 163
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.14626275180535692,
+      "learning_rate": 0.00019878787878787878,
+      "loss": 1.2115,
+      "step": 164
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.13100680398305042,
+      "learning_rate": 0.0002,
+      "loss": 1.2524,
+      "step": 165
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.14849458896148926,
+      "learning_rate": 0.00019999977531546566,
+      "loss": 1.2161,
+      "step": 166
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.13628125499037252,
+      "learning_rate": 0.0001999991012628722,
+      "loss": 1.2452,
+      "step": 167
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.18617698759086793,
+      "learning_rate": 0.00019999797784524866,
+      "loss": 1.2197,
+      "step": 168
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.14416004826313944,
+      "learning_rate": 0.00019999640506764336,
+      "loss": 1.2796,
+      "step": 169
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.13807081386834757,
+      "learning_rate": 0.0001999943829371238,
+      "loss": 1.2732,
+      "step": 170
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.16526927436841996,
+      "learning_rate": 0.0001999919114627769,
+      "loss": 1.3016,
+      "step": 171
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.14479672734919855,
+      "learning_rate": 0.0001999889906557086,
+      "loss": 1.3106,
+      "step": 172
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.13829284006072087,
+      "learning_rate": 0.00019998562052904418,
+      "loss": 1.3355,
+      "step": 173
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.13484630104616105,
+      "learning_rate": 0.0001999818010979279,
+      "loss": 1.1928,
+      "step": 174
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.14972770674556948,
+      "learning_rate": 0.00019997753237952317,
+      "loss": 1.2559,
+      "step": 175
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.13378525020528342,
+      "learning_rate": 0.00019997281439301218,
+      "loss": 1.2673,
+      "step": 176
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.13242998699125438,
+      "learning_rate": 0.00019996764715959618,
+      "loss": 1.2272,
+      "step": 177
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.12938881004364342,
+      "learning_rate": 0.00019996203070249516,
+      "loss": 1.2035,
+      "step": 178
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.13388032350164566,
+      "learning_rate": 0.00019995596504694763,
+      "loss": 1.2642,
+      "step": 179
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.13893372222140873,
+      "learning_rate": 0.00019994945022021082,
+      "loss": 1.2235,
+      "step": 180
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.14131710715500717,
+      "learning_rate": 0.00019994248625156038,
+      "loss": 1.1095,
+      "step": 181
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.13448100369103572,
+      "learning_rate": 0.0001999350731722902,
+      "loss": 1.1879,
+      "step": 182
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.13862444003216381,
+      "learning_rate": 0.00019992721101571236,
+      "loss": 1.2227,
+      "step": 183
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.13506115547921224,
+      "learning_rate": 0.00019991889981715698,
+      "loss": 1.2833,
+      "step": 184
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.13174857502600473,
+      "learning_rate": 0.00019991013961397197,
+      "loss": 1.2394,
+      "step": 185
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.1290276308949748,
+      "learning_rate": 0.00019990093044552304,
+      "loss": 1.2659,
+      "step": 186
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.1388159912078538,
+      "learning_rate": 0.0001998912723531933,
+      "loss": 1.3052,
+      "step": 187
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.1256806205303357,
+      "learning_rate": 0.00019988116538038325,
+      "loss": 1.2031,
+      "step": 188
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.13256850855084143,
+      "learning_rate": 0.00019987060957251047,
+      "loss": 1.211,
+      "step": 189
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.13197363789890235,
+      "learning_rate": 0.0001998596049770095,
+      "loss": 1.2256,
+      "step": 190
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.13277364593883098,
+      "learning_rate": 0.00019984815164333163,
+      "loss": 1.2174,
+      "step": 191
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.13838072824574454,
+      "learning_rate": 0.00019983624962294458,
+      "loss": 1.3128,
+      "step": 192
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.13524759737199996,
+      "learning_rate": 0.0001998238989693323,
+      "loss": 1.1806,
+      "step": 193
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.12669987683723832,
+      "learning_rate": 0.0001998110997379949,
+      "loss": 1.2171,
+      "step": 194
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.1461834612451898,
+      "learning_rate": 0.00019979785198644806,
+      "loss": 1.2231,
+      "step": 195
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.13265793664862735,
+      "learning_rate": 0.0001997841557742232,
+      "loss": 1.1718,
+      "step": 196
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.12842971557690963,
+      "learning_rate": 0.00019977001116286674,
+      "loss": 1.2758,
+      "step": 197
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.12188365921206967,
+      "learning_rate": 0.00019975541821594026,
+      "loss": 1.2457,
+      "step": 198
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.12679949330022622,
+      "learning_rate": 0.00019974037699901993,
+      "loss": 1.1825,
+      "step": 199
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.12949746150357985,
+      "learning_rate": 0.00019972488757969635,
+      "loss": 1.2666,
+      "step": 200
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.1363496149379173,
+      "learning_rate": 0.00019970895002757413,
+      "loss": 1.2031,
+      "step": 201
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.14218340110669314,
+      "learning_rate": 0.0001996925644142717,
+      "loss": 1.3073,
+      "step": 202
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.14234535389443218,
+      "learning_rate": 0.00019967573081342103,
+      "loss": 1.2444,
+      "step": 203
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.12866113026310516,
+      "learning_rate": 0.000199658449300667,
+      "loss": 1.2257,
+      "step": 204
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.1324053366295965,
+      "learning_rate": 0.00019964071995366744,
+      "loss": 1.2374,
+      "step": 205
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.12906841330218152,
+      "learning_rate": 0.00019962254285209254,
+      "loss": 1.2334,
+      "step": 206
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.13620873131846425,
+      "learning_rate": 0.00019960391807762463,
+      "loss": 1.242,
+      "step": 207
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.14877366842835116,
+      "learning_rate": 0.00019958484571395757,
+      "loss": 1.1772,
+      "step": 208
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.13914108740445985,
+      "learning_rate": 0.00019956532584679675,
+      "loss": 1.2734,
+      "step": 209
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.13198394930310692,
+      "learning_rate": 0.00019954535856385837,
+      "loss": 1.1728,
+      "step": 210
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.3807736597404611,
+      "learning_rate": 0.0001995249439548693,
+      "loss": 1.2089,
+      "step": 211
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.1682550557564819,
+      "learning_rate": 0.00019950408211156636,
+      "loss": 1.2423,
+      "step": 212
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.2102196862007261,
+      "learning_rate": 0.0001994827731276963,
+      "loss": 1.2096,
+      "step": 213
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.154346739470422,
+      "learning_rate": 0.00019946101709901514,
+      "loss": 1.2847,
+      "step": 214
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.16416668358293746,
+      "learning_rate": 0.0001994388141232876,
+      "loss": 1.2503,
+      "step": 215
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.13134349458231093,
+      "learning_rate": 0.0001994161643002871,
+      "loss": 1.1231,
+      "step": 216
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.15083246389185287,
+      "learning_rate": 0.00019939306773179497,
+      "loss": 1.1614,
+      "step": 217
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.1742387260929692,
+      "learning_rate": 0.00019936952452159995,
+      "loss": 1.3568,
+      "step": 218
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.18146911432436974,
+      "learning_rate": 0.00019934553477549794,
+      "loss": 1.2686,
+      "step": 219
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.1393593447949332,
+      "learning_rate": 0.00019932109860129154,
+      "loss": 1.1141,
+      "step": 220
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.14856124153987935,
+      "learning_rate": 0.00019929621610878927,
+      "loss": 1.234,
+      "step": 221
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.14820851831477327,
+      "learning_rate": 0.0001992708874098054,
+      "loss": 1.2069,
+      "step": 222
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.17893142790958147,
+      "learning_rate": 0.00019924511261815926,
+      "loss": 1.1278,
+      "step": 223
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.14573658703265605,
+      "learning_rate": 0.00019921889184967476,
+      "loss": 1.2292,
+      "step": 224
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.15282321197574994,
+      "learning_rate": 0.00019919222522217996,
+      "loss": 1.2482,
+      "step": 225
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.16342112084119492,
+      "learning_rate": 0.00019916511285550642,
+      "loss": 1.2172,
+      "step": 226
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.1475889153814455,
+      "learning_rate": 0.00019913755487148876,
+      "loss": 1.1747,
+      "step": 227
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.163738064491857,
+      "learning_rate": 0.00019910955139396396,
+      "loss": 1.3007,
+      "step": 228
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.14427856196022704,
+      "learning_rate": 0.00019908110254877106,
+      "loss": 1.2464,
+      "step": 229
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.20204742660246344,
+      "learning_rate": 0.00019905220846375032,
+      "loss": 1.2515,
+      "step": 230
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.15134144918251685,
+      "learning_rate": 0.0001990228692687429,
+      "loss": 1.1786,
+      "step": 231
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.1636590177812163,
+      "learning_rate": 0.00019899308509558998,
+      "loss": 1.1974,
+      "step": 232
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.15552319776955892,
+      "learning_rate": 0.00019896285607813244,
+      "loss": 1.2308,
+      "step": 233
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.17104898009833774,
+      "learning_rate": 0.00019893218235221015,
+      "loss": 1.2828,
+      "step": 234
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.16387378763964267,
+      "learning_rate": 0.00019890106405566138,
+      "loss": 1.2779,
+      "step": 235
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.14622126798612248,
+      "learning_rate": 0.00019886950132832207,
+      "loss": 1.2894,
+      "step": 236
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.16619841547518147,
+      "learning_rate": 0.0001988374943120254,
+      "loss": 1.2133,
+      "step": 237
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.12664832399697545,
+      "learning_rate": 0.00019880504315060096,
+      "loss": 1.1807,
+      "step": 238
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.2015108381613456,
+      "learning_rate": 0.00019877214798987426,
+      "loss": 1.1876,
+      "step": 239
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.14468620723711506,
+      "learning_rate": 0.00019873880897766598,
+      "loss": 1.1883,
+      "step": 240
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.1549018650770757,
+      "learning_rate": 0.00019870502626379127,
+      "loss": 1.2896,
+      "step": 241
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.1492917963684983,
+      "learning_rate": 0.0001986708000000593,
+      "loss": 1.2102,
+      "step": 242
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.178606606459489,
+      "learning_rate": 0.00019863613034027224,
+      "loss": 1.2292,
+      "step": 243
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.206170239681528,
+      "learning_rate": 0.00019860101744022485,
+      "loss": 1.2666,
+      "step": 244
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.13741043007948167,
+      "learning_rate": 0.0001985654614577036,
+      "loss": 1.2022,
+      "step": 245
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.1595080658199459,
+      "learning_rate": 0.0001985294625524861,
+      "loss": 1.1203,
+      "step": 246
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.13929705183853777,
+      "learning_rate": 0.00019849302088634034,
+      "loss": 1.1505,
+      "step": 247
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.14045247607912964,
+      "learning_rate": 0.00019845613662302383,
+      "loss": 1.1897,
+      "step": 248
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.15002651347444407,
+      "learning_rate": 0.00019841880992828306,
+      "loss": 1.2133,
+      "step": 249
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.1567929487810952,
+      "learning_rate": 0.00019838104096985267,
+      "loss": 1.129,
+      "step": 250
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.15240634543877116,
+      "learning_rate": 0.00019834282991745464,
+      "loss": 1.1995,
+      "step": 251
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.151807679821367,
+      "learning_rate": 0.00019830417694279766,
+      "loss": 1.25,
+      "step": 252
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.1648599156208311,
+      "learning_rate": 0.0001982650822195762,
+      "loss": 1.2511,
+      "step": 253
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.15363401233808713,
+      "learning_rate": 0.00019822554592346993,
+      "loss": 1.1794,
+      "step": 254
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.1569644350778875,
+      "learning_rate": 0.00019818556823214268,
+      "loss": 1.2033,
+      "step": 255
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.15996552747294254,
+      "learning_rate": 0.0001981451493252418,
+      "loss": 1.2809,
+      "step": 256
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.15863104885072635,
+      "learning_rate": 0.0001981042893843974,
+      "loss": 1.1667,
+      "step": 257
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.2887466971861171,
+      "learning_rate": 0.0001980629885932214,
+      "loss": 1.1915,
+      "step": 258
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.15233015979193984,
+      "learning_rate": 0.00019802124713730681,
+      "loss": 1.1734,
+      "step": 259
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.18207884538436447,
+      "learning_rate": 0.00019797906520422677,
+      "loss": 1.2575,
+      "step": 260
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.17323546756038308,
+      "learning_rate": 0.0001979364429835339,
+      "loss": 1.1704,
+      "step": 261
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.14592153602263633,
+      "learning_rate": 0.00019789338066675922,
+      "loss": 1.192,
+      "step": 262
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.19250697792287097,
+      "learning_rate": 0.0001978498784474115,
+      "loss": 1.2779,
+      "step": 263
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.1429107680887097,
+      "learning_rate": 0.0001978059365209762,
+      "loss": 1.2529,
+      "step": 264
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.48514081074992116,
+      "learning_rate": 0.00019776155508491482,
+      "loss": 1.1917,
+      "step": 265
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.1534376167748161,
+      "learning_rate": 0.0001977167343386638,
+      "loss": 1.2384,
+      "step": 266
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.16744875760032166,
+      "learning_rate": 0.00019767147448363366,
+      "loss": 1.1744,
+      "step": 267
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.29195538170738244,
+      "learning_rate": 0.00019762577572320824,
+      "loss": 1.1418,
+      "step": 268
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.1820804717651353,
+      "learning_rate": 0.00019757963826274357,
+      "loss": 1.2815,
+      "step": 269
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.17522345110441973,
+      "learning_rate": 0.00019753306230956718,
+      "loss": 1.2363,
+      "step": 270
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.16354388270886613,
+      "learning_rate": 0.000197486048072977,
+      "loss": 1.2845,
+      "step": 271
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.17590082756401024,
+      "learning_rate": 0.0001974385957642404,
+      "loss": 1.192,
+      "step": 272
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.17345720403188775,
+      "learning_rate": 0.00019739070559659347,
+      "loss": 1.2068,
+      "step": 273
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.16070434867766506,
+      "learning_rate": 0.00019734237778523976,
+      "loss": 1.189,
+      "step": 274
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.18983443066710415,
+      "learning_rate": 0.0001972936125473495,
+      "loss": 1.2223,
+      "step": 275
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.15724400187981355,
+      "learning_rate": 0.00019724441010205863,
+      "loss": 1.2292,
+      "step": 276
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.14570729442956004,
+      "learning_rate": 0.00019719477067046766,
+      "loss": 1.1421,
+      "step": 277
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.1559242881177266,
+      "learning_rate": 0.00019714469447564088,
+      "loss": 1.2598,
+      "step": 278
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.16621830243096108,
+      "learning_rate": 0.0001970941817426052,
+      "loss": 1.3038,
+      "step": 279
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 4.675483994100576,
+      "learning_rate": 0.00019704323269834927,
+      "loss": 1.2298,
+      "step": 280
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.2769699381619058,
+      "learning_rate": 0.00019699184757182225,
+      "loss": 1.2566,
+      "step": 281
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.20189839889100783,
+      "learning_rate": 0.00019694002659393305,
+      "loss": 1.3181,
+      "step": 282
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.19497107359413876,
+      "learning_rate": 0.00019688776999754912,
+      "loss": 1.1502,
+      "step": 283
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.1982266815755412,
+      "learning_rate": 0.00019683507801749545,
+      "loss": 1.2053,
+      "step": 284
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.1924340950322314,
+      "learning_rate": 0.00019678195089055346,
+      "loss": 1.2149,
+      "step": 285
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.1725322346446431,
+      "learning_rate": 0.00019672838885546008,
+      "loss": 1.2553,
+      "step": 286
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.2535488743520272,
+      "learning_rate": 0.00019667439215290648,
+      "loss": 1.2576,
+      "step": 287
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.37837586860064026,
+      "learning_rate": 0.00019661996102553718,
+      "loss": 1.1815,
+      "step": 288
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.17520419597901843,
+      "learning_rate": 0.00019656509571794878,
+      "loss": 1.1932,
+      "step": 289
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.17056234784450633,
+      "learning_rate": 0.00019650979647668906,
+      "loss": 1.163,
+      "step": 290
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.18272246580207432,
+      "learning_rate": 0.00019645406355025565,
+      "loss": 1.1887,
+      "step": 291
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.17889037954429915,
+      "learning_rate": 0.00019639789718909508,
+      "loss": 1.2126,
+      "step": 292
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.23993734971101424,
+      "learning_rate": 0.00019634129764560168,
+      "loss": 1.2485,
+      "step": 293
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.1847578318208199,
+      "learning_rate": 0.00019628426517411625,
+      "loss": 1.2549,
+      "step": 294
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.23185098827091005,
+      "learning_rate": 0.00019622680003092503,
+      "loss": 1.1599,
+      "step": 295
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.220638044092583,
+      "learning_rate": 0.00019616890247425866,
+      "loss": 1.2281,
+      "step": 296
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.2303439219825616,
+      "learning_rate": 0.00019611057276429085,
+      "loss": 1.2208,
+      "step": 297
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.1744807302230573,
+      "learning_rate": 0.00019605181116313724,
+      "loss": 1.2303,
+      "step": 298
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.17510946821872422,
+      "learning_rate": 0.0001959926179348543,
+      "loss": 1.2385,
+      "step": 299
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.2218474349751746,
+      "learning_rate": 0.00019593299334543808,
+      "loss": 1.2153,
+      "step": 300
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.1742070481516402,
+      "learning_rate": 0.00019587293766282308,
+      "loss": 1.1628,
+      "step": 301
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.15250311715180823,
+      "learning_rate": 0.00019581245115688094,
+      "loss": 1.1632,
+      "step": 302
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.1744397677094501,
+      "learning_rate": 0.0001957515340994193,
+      "loss": 1.254,
+      "step": 303
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.1686772182789891,
+      "learning_rate": 0.00019569018676418053,
+      "loss": 1.2169,
+      "step": 304
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.16404966161017623,
+      "learning_rate": 0.00019562840942684067,
+      "loss": 1.2221,
+      "step": 305
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.16052011449463713,
+      "learning_rate": 0.00019556620236500793,
+      "loss": 1.2045,
+      "step": 306
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.16343251390831215,
+      "learning_rate": 0.0001955035658582216,
+      "loss": 1.2289,
+      "step": 307
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.14387162360389305,
+      "learning_rate": 0.00019544050018795075,
+      "loss": 1.1365,
+      "step": 308
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.15304461439740238,
+      "learning_rate": 0.00019537700563759304,
+      "loss": 1.1931,
+      "step": 309
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.17059958050065627,
+      "learning_rate": 0.00019531308249247327,
+      "loss": 1.2166,
+      "step": 310
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.17633385530926995,
+      "learning_rate": 0.00019524873103984235,
+      "loss": 1.2604,
+      "step": 311
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.17855814403303746,
+      "learning_rate": 0.00019518395156887576,
+      "loss": 1.1615,
+      "step": 312
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.19823982444256988,
+      "learning_rate": 0.00019511874437067243,
+      "loss": 1.2153,
+      "step": 313
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.1570784627362585,
+      "learning_rate": 0.0001950531097382533,
+      "loss": 1.2788,
+      "step": 314
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.2183125402112695,
+      "learning_rate": 0.00019498704796656018,
+      "loss": 1.2966,
+      "step": 315
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.18173933276147194,
+      "learning_rate": 0.00019492055935245418,
+      "loss": 1.2978,
+      "step": 316
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.17483116680914407,
+      "learning_rate": 0.00019485364419471454,
+      "loss": 1.258,
+      "step": 317
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.15490767356815494,
+      "learning_rate": 0.0001947863027940374,
+      "loss": 1.2088,
+      "step": 318
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.14703966491934156,
+      "learning_rate": 0.00019471853545303405,
+      "loss": 1.2355,
+      "step": 319
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.14386689086661608,
+      "learning_rate": 0.00019465034247623003,
+      "loss": 1.2583,
+      "step": 320
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.18818904376313625,
+      "learning_rate": 0.00019458172417006347,
+      "loss": 1.2181,
+      "step": 321
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.17393313719202513,
+      "learning_rate": 0.00019451268084288385,
+      "loss": 1.3453,
+      "step": 322
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.14706823379985753,
+      "learning_rate": 0.00019444321280495043,
+      "loss": 1.2234,
+      "step": 323
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.15282014755252687,
+      "learning_rate": 0.00019437332036843118,
+      "loss": 1.1262,
+      "step": 324
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.1618727884326225,
+      "learning_rate": 0.00019430300384740105,
+      "loss": 1.3136,
+      "step": 325
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.16090758705378874,
+      "learning_rate": 0.00019423226355784077,
+      "loss": 1.2055,
+      "step": 326
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.15241156801091013,
+      "learning_rate": 0.00019416109981763526,
+      "loss": 1.2678,
+      "step": 327
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.14216697909809062,
+      "learning_rate": 0.0001940895129465724,
+      "loss": 1.2841,
+      "step": 328
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.15790232415414485,
+      "learning_rate": 0.00019401750326634144,
+      "loss": 1.3119,
+      "step": 329
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.13322691961062616,
+      "learning_rate": 0.0001939450711005316,
+      "loss": 1.1293,
+      "step": 330
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.14075018938835404,
+      "learning_rate": 0.00019387221677463062,
+      "loss": 1.2176,
+      "step": 331
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.21565975459393052,
+      "learning_rate": 0.00019379894061602335,
+      "loss": 1.1723,
+      "step": 332
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.17967631394222838,
+      "learning_rate": 0.00019372524295399013,
+      "loss": 1.239,
+      "step": 333
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.21187969201978435,
+      "learning_rate": 0.0001936511241197055,
+      "loss": 1.2207,
+      "step": 334
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.16967789022974608,
+      "learning_rate": 0.00019357658444623654,
+      "loss": 1.2478,
+      "step": 335
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.14810621660374448,
+      "learning_rate": 0.0001935016242685415,
+      "loss": 1.1223,
+      "step": 336
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.1489106421847434,
+      "learning_rate": 0.00019342624392346824,
+      "loss": 1.1592,
+      "step": 337
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.17625176068748855,
+      "learning_rate": 0.0001933504437497527,
+      "loss": 1.2145,
+      "step": 338
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.17250255512763446,
+      "learning_rate": 0.00019327422408801744,
+      "loss": 1.2504,
+      "step": 339
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.16079375745566896,
+      "learning_rate": 0.00019319758528077,
+      "loss": 1.1795,
+      "step": 340
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.15454466809245995,
+      "learning_rate": 0.0001931205276724015,
+      "loss": 1.2123,
+      "step": 341
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.7021323604447972,
+      "learning_rate": 0.000193043051609185,
+      "loss": 1.2239,
+      "step": 342
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.1572764339385847,
+      "learning_rate": 0.00019296515743927399,
+      "loss": 1.2516,
+      "step": 343
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.2136637778252246,
+      "learning_rate": 0.00019288684551270073,
+      "loss": 1.2321,
+      "step": 344
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.4546540454773654,
+      "learning_rate": 0.00019280811618137484,
+      "loss": 1.18,
+      "step": 345
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.9809832576786297,
+      "learning_rate": 0.00019272896979908154,
+      "loss": 1.2081,
+      "step": 346
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.5246256133291822,
+      "learning_rate": 0.00019264940672148018,
+      "loss": 1.2722,
+      "step": 347
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.24941717134878091,
+      "learning_rate": 0.00019256942730610268,
+      "loss": 1.2352,
+      "step": 348
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.3356068462072784,
+      "learning_rate": 0.00019248903191235176,
+      "loss": 1.2225,
+      "step": 349
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.19535845221880543,
+      "learning_rate": 0.00019240822090149944,
+      "loss": 1.1669,
+      "step": 350
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.22306941566416597,
+      "learning_rate": 0.00019232699463668542,
+      "loss": 1.2281,
+      "step": 351
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.2700134013989352,
+      "learning_rate": 0.00019224535348291542,
+      "loss": 1.1939,
+      "step": 352
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.24406908935562743,
+      "learning_rate": 0.00019216329780705953,
+      "loss": 1.1839,
+      "step": 353
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.20465183000217488,
+      "learning_rate": 0.00019208082797785055,
+      "loss": 1.2277,
+      "step": 354
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.21324820828129784,
+      "learning_rate": 0.00019199794436588243,
+      "loss": 1.2072,
+      "step": 355
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.1780562512431263,
+      "learning_rate": 0.00019191464734360844,
+      "loss": 1.2082,
+      "step": 356
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.16547971467615655,
+      "learning_rate": 0.00019183093728533966,
+      "loss": 1.1978,
+      "step": 357
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.22904664933247196,
+      "learning_rate": 0.00019174681456724318,
+      "loss": 1.1562,
+      "step": 358
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.1737397860007602,
+      "learning_rate": 0.00019166227956734052,
+      "loss": 1.2383,
+      "step": 359
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.1589465455917568,
+      "learning_rate": 0.00019157733266550575,
+      "loss": 1.2158,
+      "step": 360
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.16253126221999709,
+      "learning_rate": 0.00019149197424346405,
+      "loss": 1.1952,
+      "step": 361
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.22436676243032663,
+      "learning_rate": 0.00019140620468478968,
+      "loss": 1.2315,
+      "step": 362
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.19291682612950423,
+      "learning_rate": 0.00019132002437490458,
+      "loss": 1.2283,
+      "step": 363
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.1519191258459668,
+      "learning_rate": 0.00019123343370107637,
+      "loss": 1.1151,
+      "step": 364
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.17179909633547025,
+      "learning_rate": 0.00019114643305241676,
+      "loss": 1.1576,
+      "step": 365
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.17992599023321432,
+      "learning_rate": 0.00019105902281987976,
+      "loss": 1.2592,
+      "step": 366
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.17714099390314453,
+      "learning_rate": 0.00019097120339625994,
+      "loss": 1.2578,
+      "step": 367
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.2455577642687935,
+      "learning_rate": 0.00019088297517619055,
+      "loss": 1.2361,
+      "step": 368
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.18398518628783986,
+      "learning_rate": 0.00019079433855614201,
+      "loss": 1.1906,
+      "step": 369
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.18944067022821645,
+      "learning_rate": 0.00019070529393441985,
+      "loss": 1.237,
+      "step": 370
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.17639967519781063,
+      "learning_rate": 0.00019061584171116303,
+      "loss": 1.1841,
+      "step": 371
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.15947129998283005,
+      "learning_rate": 0.00019052598228834217,
+      "loss": 1.1722,
+      "step": 372
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.1693354353719105,
+      "learning_rate": 0.00019043571606975777,
+      "loss": 1.2204,
+      "step": 373
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.16236190451963983,
+      "learning_rate": 0.00019034504346103823,
+      "loss": 1.1778,
+      "step": 374
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.17702370729269964,
+      "learning_rate": 0.00019025396486963827,
+      "loss": 1.2065,
+      "step": 375
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.19388150596154238,
+      "learning_rate": 0.00019016248070483687,
+      "loss": 1.2942,
+      "step": 376
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.16152000400319103,
+      "learning_rate": 0.0001900705913777356,
+      "loss": 1.1784,
+      "step": 377
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.1545267913996029,
+      "learning_rate": 0.00018997829730125663,
+      "loss": 1.1829,
+      "step": 378
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.15421727704318197,
+      "learning_rate": 0.000189885598890141,
+      "loss": 1.177,
+      "step": 379
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.1624966073814206,
+      "learning_rate": 0.00018979249656094673,
+      "loss": 1.2439,
+      "step": 380
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.9490737312904575,
+      "learning_rate": 0.00018969899073204686,
+      "loss": 1.2085,
+      "step": 381
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.8982903208613089,
+      "learning_rate": 0.00018960508182362768,
+      "loss": 1.2347,
+      "step": 382
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.3771428474797688,
+      "learning_rate": 0.00018951077025768678,
+      "loss": 1.2546,
+      "step": 383
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.4776152950069111,
+      "learning_rate": 0.00018941605645803115,
+      "loss": 1.2904,
+      "step": 384
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.18786943849618057,
+      "learning_rate": 0.00018932094085027533,
+      "loss": 1.2122,
+      "step": 385
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.6297025984167128,
+      "learning_rate": 0.0001892254238618394,
+      "loss": 1.171,
+      "step": 386
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.20382660707264952,
+      "learning_rate": 0.0001891295059219472,
+      "loss": 1.1874,
+      "step": 387
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.3580819775908755,
+      "learning_rate": 0.00018903318746162429,
+      "loss": 1.1531,
+      "step": 388
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.43619056173016185,
+      "learning_rate": 0.00018893646891369602,
+      "loss": 1.2289,
+      "step": 389
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.29385240705823723,
+      "learning_rate": 0.0001888393507127856,
+      "loss": 1.2073,
+      "step": 390
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.3136086850525623,
+      "learning_rate": 0.00018874183329531223,
+      "loss": 1.1898,
+      "step": 391
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.2307767217662562,
+      "learning_rate": 0.000188643917099489,
+      "loss": 1.207,
+      "step": 392
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.18703654518135468,
+      "learning_rate": 0.000188545602565321,
+      "loss": 1.1688,
+      "step": 393
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.4809351333934126,
+      "learning_rate": 0.00018844689013460336,
+      "loss": 1.2519,
+      "step": 394
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.40370101428544464,
+      "learning_rate": 0.0001883477802509192,
+      "loss": 1.2411,
+      "step": 395
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.2858848636432859,
+      "learning_rate": 0.00018824827335963765,
+      "loss": 1.194,
+      "step": 396
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.32195602638999565,
+      "learning_rate": 0.000188148369907912,
+      "loss": 1.0988,
+      "step": 397
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.23790306908901832,
+      "learning_rate": 0.00018804807034467733,
+      "loss": 1.2237,
+      "step": 398
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.20126988767112128,
+      "learning_rate": 0.0001879473751206489,
+      "loss": 1.2731,
+      "step": 399
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.3336380339194037,
+      "learning_rate": 0.00018784628468831996,
+      "loss": 1.2369,
+      "step": 400
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.5054330893305989,
+      "learning_rate": 0.0001877447995019596,
+      "loss": 1.2443,
+      "step": 401
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.2297866279715136,
+      "learning_rate": 0.0001876429200176108,
+      "loss": 1.2376,
+      "step": 402
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.39350567174184636,
+      "learning_rate": 0.00018754064669308858,
+      "loss": 1.2126,
+      "step": 403
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.2025361091435325,
+      "learning_rate": 0.00018743797998797753,
+      "loss": 1.2224,
+      "step": 404
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.31824903419753814,
+      "learning_rate": 0.00018733492036363005,
+      "loss": 1.2942,
+      "step": 405
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.4642066748643017,
+      "learning_rate": 0.00018723146828316428,
+      "loss": 1.2515,
+      "step": 406
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.7833055646295342,
+      "learning_rate": 0.00018712762421146183,
+      "loss": 1.2207,
+      "step": 407
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.2810249021786599,
+      "learning_rate": 0.00018702338861516587,
+      "loss": 1.2755,
+      "step": 408
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.460995724241333,
+      "learning_rate": 0.0001869187619626789,
+      "loss": 1.2856,
+      "step": 409
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.49139203044984286,
+      "learning_rate": 0.00018681374472416073,
+      "loss": 1.2392,
+      "step": 410
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.691604613969173,
+      "learning_rate": 0.0001867083373715264,
+      "loss": 1.2992,
+      "step": 411
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.8014112047318501,
+      "learning_rate": 0.00018660254037844388,
+      "loss": 1.2683,
+      "step": 412
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.31614342841331383,
+      "learning_rate": 0.00018649635422033215,
+      "loss": 1.2356,
+      "step": 413
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.2559855196513244,
+      "learning_rate": 0.000186389779374359,
+      "loss": 1.2053,
+      "step": 414
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.6613999986014714,
+      "learning_rate": 0.0001862828163194388,
+      "loss": 1.2568,
+      "step": 415
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.27190082167109786,
+      "learning_rate": 0.0001861754655362304,
+      "loss": 1.1288,
+      "step": 416
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.43819582203066043,
+      "learning_rate": 0.00018606772750713504,
+      "loss": 1.1758,
+      "step": 417
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.41738497400383384,
+      "learning_rate": 0.0001859596027162941,
+      "loss": 1.2993,
+      "step": 418
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.2595142634740817,
+      "learning_rate": 0.000185851091649587,
+      "loss": 1.269,
+      "step": 419
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.2795314201020271,
+      "learning_rate": 0.00018574219479462878,
+      "loss": 1.1915,
+      "step": 420
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.2502992494749938,
+      "learning_rate": 0.00018563291264076835,
+      "loss": 1.2157,
+      "step": 421
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.27422512335538374,
+      "learning_rate": 0.00018552324567908585,
+      "loss": 1.2541,
+      "step": 422
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.3360989016060905,
+      "learning_rate": 0.00018541319440239066,
+      "loss": 1.2666,
+      "step": 423
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.9441131913572127,
+      "learning_rate": 0.00018530275930521924,
+      "loss": 1.2924,
+      "step": 424
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.3772373301771213,
+      "learning_rate": 0.00018519194088383273,
+      "loss": 1.1952,
+      "step": 425
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.22091753616251295,
+      "learning_rate": 0.0001850807396362148,
+      "loss": 1.1858,
+      "step": 426
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.21423504993321807,
+      "learning_rate": 0.00018496915606206951,
+      "loss": 1.2245,
+      "step": 427
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.5238946238105926,
+      "learning_rate": 0.00018485719066281892,
+      "loss": 1.2351,
+      "step": 428
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.3037858949309141,
+      "learning_rate": 0.0001847448439416009,
+      "loss": 1.1669,
+      "step": 429
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.21553286799952254,
+      "learning_rate": 0.00018463211640326686,
+      "loss": 1.1454,
+      "step": 430
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.202875547805464,
+      "learning_rate": 0.0001845190085543795,
+      "loss": 1.188,
+      "step": 431
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.24385408620619278,
+      "learning_rate": 0.00018440552090321047,
+      "loss": 1.2307,
+      "step": 432
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.23793944272430378,
+      "learning_rate": 0.0001842916539597382,
+      "loss": 1.2253,
+      "step": 433
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.17062488448810784,
+      "learning_rate": 0.0001841774082356455,
+      "loss": 1.2681,
+      "step": 434
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.20003742001916064,
+      "learning_rate": 0.00018406278424431736,
+      "loss": 1.2428,
+      "step": 435
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.2696052831337752,
+      "learning_rate": 0.0001839477825008385,
+      "loss": 1.2945,
+      "step": 436
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.23302960820538443,
+      "learning_rate": 0.00018383240352199117,
+      "loss": 1.1718,
+      "step": 437
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.38187833239777536,
+      "learning_rate": 0.00018371664782625287,
+      "loss": 1.2311,
+      "step": 438
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.4052561772533732,
+      "learning_rate": 0.00018360051593379383,
+      "loss": 1.1639,
+      "step": 439
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.23379763821020377,
+      "learning_rate": 0.0001834840083664749,
+      "loss": 1.1809,
+      "step": 440
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.2368414607613928,
+      "learning_rate": 0.00018336712564784503,
+      "loss": 1.2357,
+      "step": 441
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.20230633988510938,
+      "learning_rate": 0.000183249868303139,
+      "loss": 1.1851,
+      "step": 442
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.170513157244292,
+      "learning_rate": 0.00018313223685927505,
+      "loss": 1.205,
+      "step": 443
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.18082295035256266,
+      "learning_rate": 0.0001830142318448525,
+      "loss": 1.2305,
+      "step": 444
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.18286299264146286,
+      "learning_rate": 0.00018289585379014942,
+      "loss": 1.23,
+      "step": 445
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.17868104103482751,
+      "learning_rate": 0.00018277710322712012,
+      "loss": 1.2894,
+      "step": 446
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.1820411127336495,
+      "learning_rate": 0.00018265798068939294,
+      "loss": 1.2395,
+      "step": 447
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.1738237541783663,
+      "learning_rate": 0.0001825384867122677,
+      "loss": 1.1576,
+      "step": 448
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.15693445967795147,
+      "learning_rate": 0.0001824186218327134,
+      "loss": 1.0809,
+      "step": 449
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.18509145652208978,
+      "learning_rate": 0.00018229838658936564,
+      "loss": 1.2717,
+      "step": 450
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.14702488366564262,
+      "learning_rate": 0.0001821777815225245,
+      "loss": 1.2236,
+      "step": 451
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.1828399354418095,
+      "learning_rate": 0.00018205680717415187,
+      "loss": 1.2565,
+      "step": 452
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.17460984182013486,
+      "learning_rate": 0.00018193546408786898,
+      "loss": 1.2474,
+      "step": 453
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.2001623109673152,
+      "learning_rate": 0.00018181375280895416,
+      "loss": 1.2544,
+      "step": 454
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.17228631742863837,
+      "learning_rate": 0.00018169167388434025,
+      "loss": 1.1851,
+      "step": 455
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.1644862232819482,
+      "learning_rate": 0.00018156922786261216,
+      "loss": 1.1817,
+      "step": 456
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.19775186397477057,
+      "learning_rate": 0.00018144641529400446,
+      "loss": 1.257,
+      "step": 457
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.1626281991220394,
+      "learning_rate": 0.00018132323673039885,
+      "loss": 1.2277,
+      "step": 458
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.16158256707311264,
+      "learning_rate": 0.00018119969272532166,
+      "loss": 1.1624,
+      "step": 459
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.17705809207051687,
+      "learning_rate": 0.00018107578383394146,
+      "loss": 1.2421,
+      "step": 460
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.17639060401882287,
+      "learning_rate": 0.00018095151061306645,
+      "loss": 1.285,
+      "step": 461
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.16918796486576196,
+      "learning_rate": 0.00018082687362114212,
+      "loss": 1.2606,
+      "step": 462
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.15968377185965665,
+      "learning_rate": 0.0001807018734182485,
+      "loss": 1.194,
+      "step": 463
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.17537027967397978,
+      "learning_rate": 0.00018057651056609784,
+      "loss": 1.1594,
+      "step": 464
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.15753665403127565,
+      "learning_rate": 0.00018045078562803203,
+      "loss": 1.1382,
+      "step": 465
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.17121200763916436,
+      "learning_rate": 0.00018032469916902003,
+      "loss": 1.2286,
+      "step": 466
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.19120510133331003,
+      "learning_rate": 0.00018019825175565542,
+      "loss": 1.2835,
+      "step": 467
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.1671735980123817,
+      "learning_rate": 0.0001800714439561538,
+      "loss": 1.2201,
+      "step": 468
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.1579098534969056,
+      "learning_rate": 0.00017994427634035015,
+      "loss": 1.2156,
+      "step": 469
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.1746075421158512,
+      "learning_rate": 0.00017981674947969636,
+      "loss": 1.2049,
+      "step": 470
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.16878182886737042,
+      "learning_rate": 0.00017968886394725874,
+      "loss": 1.2204,
+      "step": 471
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.16725956538286493,
+      "learning_rate": 0.00017956062031771535,
+      "loss": 1.2091,
+      "step": 472
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.18877845951705005,
+      "learning_rate": 0.00017943201916735335,
+      "loss": 1.241,
+      "step": 473
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.180337447476004,
+      "learning_rate": 0.00017930306107406653,
+      "loss": 1.2253,
+      "step": 474
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.16688572366717752,
+      "learning_rate": 0.0001791737466173527,
+      "loss": 1.239,
+      "step": 475
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.15385917621135983,
+      "learning_rate": 0.00017904407637831099,
+      "loss": 1.2476,
+      "step": 476
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.17725645269055587,
+      "learning_rate": 0.00017891405093963938,
+      "loss": 1.2599,
+      "step": 477
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.14758551718901028,
+      "learning_rate": 0.00017878367088563195,
+      "loss": 1.2249,
+      "step": 478
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.15216962408661316,
+      "learning_rate": 0.00017865293680217637,
+      "loss": 1.2346,
+      "step": 479
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.16679282848599514,
+      "learning_rate": 0.00017852184927675112,
+      "loss": 1.2443,
+      "step": 480
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.16723562739069214,
+      "learning_rate": 0.00017839040889842305,
+      "loss": 1.224,
+      "step": 481
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.15922276239929914,
+      "learning_rate": 0.00017825861625784455,
+      "loss": 1.2739,
+      "step": 482
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.1510107938469514,
+      "learning_rate": 0.00017812647194725094,
+      "loss": 1.1764,
+      "step": 483
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.16446999054333494,
+      "learning_rate": 0.00017799397656045792,
+      "loss": 1.2498,
+      "step": 484
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.18566301651865832,
+      "learning_rate": 0.00017786113069285874,
+      "loss": 1.232,
+      "step": 485
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.20592971655306183,
+      "learning_rate": 0.00017772793494142167,
+      "loss": 1.1586,
+      "step": 486
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.1581947714375729,
+      "learning_rate": 0.00017759438990468725,
+      "loss": 1.2502,
+      "step": 487
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.15466760695169174,
+      "learning_rate": 0.00017746049618276545,
+      "loss": 1.1605,
+      "step": 488
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.16041506222444918,
+      "learning_rate": 0.00017732625437733335,
+      "loss": 1.2778,
+      "step": 489
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.17168109661676773,
+      "learning_rate": 0.0001771916650916321,
+      "loss": 1.262,
+      "step": 490
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.1788973186498254,
+      "learning_rate": 0.00017705672893046425,
+      "loss": 1.2111,
+      "step": 491
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.1759644359346382,
+      "learning_rate": 0.00017692144650019125,
+      "loss": 1.2546,
+      "step": 492
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.15710749736088767,
+      "learning_rate": 0.0001767858184087304,
+      "loss": 1.2487,
+      "step": 493
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.1648235522911144,
+      "learning_rate": 0.00017664984526555248,
+      "loss": 1.2469,
+      "step": 494
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.15452607969890703,
+      "learning_rate": 0.0001765135276816787,
+      "loss": 1.1855,
+      "step": 495
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.1837695597880219,
+      "learning_rate": 0.00017637686626967812,
+      "loss": 1.2185,
+      "step": 496
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.15861390725762364,
+      "learning_rate": 0.00017623986164366486,
+      "loss": 1.2056,
+      "step": 497
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.1663260460966887,
+      "learning_rate": 0.00017610251441929533,
+      "loss": 1.1242,
+      "step": 498
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.1803309720529981,
+      "learning_rate": 0.00017596482521376546,
+      "loss": 1.2938,
+      "step": 499
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.14909085011764342,
+      "learning_rate": 0.00017582679464580797,
+      "loss": 1.1953,
+      "step": 500
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.15779022242482527,
+      "learning_rate": 0.00017568842333568952,
+      "loss": 1.2792,
+      "step": 501
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.1553327313967345,
+      "learning_rate": 0.00017554971190520798,
+      "loss": 1.2286,
+      "step": 502
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.16363964666273684,
+      "learning_rate": 0.00017541066097768963,
+      "loss": 1.2753,
+      "step": 503
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.16668099163659675,
+      "learning_rate": 0.00017527127117798635,
+      "loss": 1.185,
+      "step": 504
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.13957568397594883,
+      "learning_rate": 0.0001751315431324727,
+      "loss": 1.143,
+      "step": 505
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.1553111736740035,
+      "learning_rate": 0.00017499147746904335,
+      "loss": 1.2492,
+      "step": 506
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.1691517335818193,
+      "learning_rate": 0.00017485107481711012,
+      "loss": 1.2619,
+      "step": 507
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.15480883994395986,
+      "learning_rate": 0.00017471033580759903,
+      "loss": 1.2396,
+      "step": 508
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.1451690143792058,
+      "learning_rate": 0.00017456926107294765,
+      "loss": 1.1732,
+      "step": 509
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.1524398957482947,
+      "learning_rate": 0.00017442785124710227,
+      "loss": 1.2083,
+      "step": 510
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.16790264977550012,
+      "learning_rate": 0.0001742861069655148,
+      "loss": 1.2201,
+      "step": 511
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.1529847047636337,
+      "learning_rate": 0.0001741440288651403,
+      "loss": 1.243,
+      "step": 512
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.1485875402374676,
+      "learning_rate": 0.00017400161758443375,
+      "loss": 1.2053,
+      "step": 513
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.16950094279079617,
+      "learning_rate": 0.00017385887376334742,
+      "loss": 1.1944,
+      "step": 514
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.15289337084330445,
+      "learning_rate": 0.00017371579804332789,
+      "loss": 1.2503,
+      "step": 515
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.15337063655317973,
+      "learning_rate": 0.00017357239106731317,
+      "loss": 1.3092,
+      "step": 516
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.1458937961897621,
+      "learning_rate": 0.00017342865347972988,
+      "loss": 1.2244,
+      "step": 517
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.19897118610161338,
+      "learning_rate": 0.00017328458592649027,
+      "loss": 1.2238,
+      "step": 518
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.15850805264911003,
+      "learning_rate": 0.00017314018905498931,
+      "loss": 1.195,
+      "step": 519
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.14445183074519347,
+      "learning_rate": 0.00017299546351410197,
+      "loss": 1.1974,
+      "step": 520
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.18180731722745677,
+      "learning_rate": 0.00017285040995418,
+      "loss": 1.2107,
+      "step": 521
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.14943874953193587,
+      "learning_rate": 0.00017270502902704926,
+      "loss": 1.1843,
+      "step": 522
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.15767466790910512,
+      "learning_rate": 0.00017255932138600665,
+      "loss": 1.1409,
+      "step": 523
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.16402921378654775,
+      "learning_rate": 0.00017241328768581726,
+      "loss": 1.2135,
+      "step": 524
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.15526246786505485,
+      "learning_rate": 0.00017226692858271134,
+      "loss": 1.2255,
+      "step": 525
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.16608155892622348,
+      "learning_rate": 0.00017212024473438147,
+      "loss": 1.2691,
+      "step": 526
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.14913271520144072,
+      "learning_rate": 0.00017197323679997943,
+      "loss": 1.1574,
+      "step": 527
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.1471910610421707,
+      "learning_rate": 0.00017182590544011347,
+      "loss": 1.2774,
+      "step": 528
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.1417464185073962,
+      "learning_rate": 0.00017167825131684513,
+      "loss": 1.2446,
+      "step": 529
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.1610488125634495,
+      "learning_rate": 0.0001715302750936864,
+      "loss": 1.2862,
+      "step": 530
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.20227974555123074,
+      "learning_rate": 0.00017138197743559654,
+      "loss": 1.207,
+      "step": 531
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.1355502559749413,
+      "learning_rate": 0.00017123335900897946,
+      "loss": 1.1019,
+      "step": 532
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.1559423167028215,
+      "learning_rate": 0.00017108442048168038,
+      "loss": 1.2549,
+      "step": 533
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.15898973818185586,
+      "learning_rate": 0.00017093516252298296,
+      "loss": 1.2705,
+      "step": 534
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.15169569998999652,
+      "learning_rate": 0.00017078558580360632,
+      "loss": 1.2454,
+      "step": 535
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.15976111665597925,
+      "learning_rate": 0.00017063569099570196,
+      "loss": 1.2585,
+      "step": 536
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.14488877221999352,
+      "learning_rate": 0.00017048547877285077,
+      "loss": 1.2169,
+      "step": 537
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.14919533098974924,
+      "learning_rate": 0.00017033494981006002,
+      "loss": 1.2358,
+      "step": 538
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.15251746717084805,
+      "learning_rate": 0.00017018410478376032,
+      "loss": 1.2241,
+      "step": 539
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.1456060482002663,
+      "learning_rate": 0.00017003294437180255,
+      "loss": 1.2298,
+      "step": 540
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.17048886778787248,
+      "learning_rate": 0.00016988146925345484,
+      "loss": 1.2707,
+      "step": 541
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.15304381059310815,
+      "learning_rate": 0.00016972968010939954,
+      "loss": 1.1498,
+      "step": 542
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.16590055969071696,
+      "learning_rate": 0.0001695775776217301,
+      "loss": 1.2481,
+      "step": 543
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.14299575837437278,
+      "learning_rate": 0.00016942516247394807,
+      "loss": 1.2058,
+      "step": 544
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.14275107775859475,
+      "learning_rate": 0.00016927243535095997,
+      "loss": 1.2178,
+      "step": 545
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.1554250137491414,
+      "learning_rate": 0.0001691193969390742,
+      "loss": 1.1197,
+      "step": 546
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.16958418467021688,
+      "learning_rate": 0.0001689660479259981,
+      "loss": 1.1768,
+      "step": 547
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.1546216583314497,
+      "learning_rate": 0.00016881238900083473,
+      "loss": 1.1741,
+      "step": 548
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.15287056494787424,
+      "learning_rate": 0.0001686584208540797,
+      "loss": 1.2328,
+      "step": 549
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.1419329373337611,
+      "learning_rate": 0.0001685041441776183,
+      "loss": 1.1743,
+      "step": 550
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.15662192296485464,
+      "learning_rate": 0.00016834955966472213,
+      "loss": 1.1861,
+      "step": 551
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.16304778894798697,
+      "learning_rate": 0.00016819466801004621,
+      "loss": 1.2045,
+      "step": 552
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.15690090424895087,
+      "learning_rate": 0.00016803946990962576,
+      "loss": 1.1553,
+      "step": 553
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.15227523196027068,
+      "learning_rate": 0.000167883966060873,
+      "loss": 1.126,
+      "step": 554
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.15442134859978873,
+      "learning_rate": 0.00016772815716257412,
+      "loss": 1.169,
+      "step": 555
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.16235578615890994,
+      "learning_rate": 0.00016757204391488613,
+      "loss": 1.198,
+      "step": 556
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.16631365293435893,
+      "learning_rate": 0.00016741562701933367,
+      "loss": 1.2191,
+      "step": 557
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.15668252106453312,
+      "learning_rate": 0.0001672589071788059,
+      "loss": 1.1094,
+      "step": 558
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.14883987789397163,
+      "learning_rate": 0.00016710188509755329,
+      "loss": 1.0653,
+      "step": 559
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.14723412204861966,
+      "learning_rate": 0.00016694456148118452,
+      "loss": 1.0632,
+      "step": 560
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.2694273773199372,
+      "learning_rate": 0.00016678693703666325,
+      "loss": 1.1577,
+      "step": 561
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.3010203762394184,
+      "learning_rate": 0.00016662901247230502,
+      "loss": 1.1049,
+      "step": 562
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.16960159731526514,
+      "learning_rate": 0.0001664707884977739,
+      "loss": 1.2185,
+      "step": 563
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.1593257551859938,
+      "learning_rate": 0.00016631226582407952,
+      "loss": 1.1085,
+      "step": 564
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.1695114549114674,
+      "learning_rate": 0.00016615344516357378,
+      "loss": 1.1863,
+      "step": 565
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.17116727204552032,
+      "learning_rate": 0.00016599432722994755,
+      "loss": 1.0921,
+      "step": 566
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.16265948588203852,
+      "learning_rate": 0.00016583491273822765,
+      "loss": 1.1905,
+      "step": 567
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.16500105151611763,
+      "learning_rate": 0.00016567520240477344,
+      "loss": 1.1477,
+      "step": 568
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.15445707774433456,
+      "learning_rate": 0.00016551519694727381,
+      "loss": 1.1247,
+      "step": 569
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.17740657850033337,
+      "learning_rate": 0.0001653548970847438,
+      "loss": 1.0164,
+      "step": 570
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.16152198390194278,
+      "learning_rate": 0.0001651943035375214,
+      "loss": 1.1779,
+      "step": 571
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.16299638453443582,
+      "learning_rate": 0.00016503341702726426,
+      "loss": 1.1847,
+      "step": 572
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.1486770667934901,
+      "learning_rate": 0.00016487223827694672,
+      "loss": 1.1002,
+      "step": 573
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.15907088594322072,
+      "learning_rate": 0.00016471076801085615,
+      "loss": 1.1127,
+      "step": 574
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.15491780431797444,
+      "learning_rate": 0.00016454900695458998,
+      "loss": 1.1196,
+      "step": 575
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.16584586610927665,
+      "learning_rate": 0.00016438695583505242,
+      "loss": 1.1441,
+      "step": 576
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.15197831968756778,
+      "learning_rate": 0.00016422461538045103,
+      "loss": 1.0532,
+      "step": 577
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.15669296934233234,
+      "learning_rate": 0.00016406198632029357,
+      "loss": 1.1753,
+      "step": 578
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.1523894998655591,
+      "learning_rate": 0.0001638990693853848,
+      "loss": 1.1339,
+      "step": 579
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.1570057583413597,
+      "learning_rate": 0.000163735865307823,
+      "loss": 1.1117,
+      "step": 580
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.1637341350288717,
+      "learning_rate": 0.00016357237482099684,
+      "loss": 1.1369,
+      "step": 581
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.15703259344496112,
+      "learning_rate": 0.0001634085986595819,
+      "loss": 1.1448,
+      "step": 582
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.1672273979359662,
+      "learning_rate": 0.00016324453755953773,
+      "loss": 1.1972,
+      "step": 583
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.16032943590057747,
+      "learning_rate": 0.0001630801922581041,
+      "loss": 1.1811,
+      "step": 584
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.15414179865764022,
+      "learning_rate": 0.00016291556349379795,
+      "loss": 1.1126,
+      "step": 585
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.16133591336152153,
+      "learning_rate": 0.00016275065200641004,
+      "loss": 1.0664,
+      "step": 586
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.17496271909044792,
+      "learning_rate": 0.00016258545853700158,
+      "loss": 1.2042,
+      "step": 587
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.16189829224809252,
+      "learning_rate": 0.00016241998382790095,
+      "loss": 1.1896,
+      "step": 588
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.1508460775909727,
+      "learning_rate": 0.00016225422862270027,
+      "loss": 1.1205,
+      "step": 589
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.16848110855261456,
+      "learning_rate": 0.00016208819366625218,
+      "loss": 1.1132,
+      "step": 590
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.15996856299765072,
+      "learning_rate": 0.00016192187970466644,
+      "loss": 1.1022,
+      "step": 591
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.15697088450293906,
+      "learning_rate": 0.0001617552874853065,
+      "loss": 1.0483,
+      "step": 592
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.16338270251123885,
+      "learning_rate": 0.0001615884177567863,
+      "loss": 1.1049,
+      "step": 593
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.1605597226615145,
+      "learning_rate": 0.0001614212712689668,
+      "loss": 1.1319,
+      "step": 594
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.1694169270615921,
+      "learning_rate": 0.00016125384877295257,
+      "loss": 1.2143,
+      "step": 595
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.16306866904399633,
+      "learning_rate": 0.00016108615102108855,
+      "loss": 1.164,
+      "step": 596
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.1646758516051161,
+      "learning_rate": 0.00016091817876695655,
+      "loss": 1.2176,
+      "step": 597
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.16334394129553162,
+      "learning_rate": 0.00016074993276537198,
+      "loss": 1.1526,
+      "step": 598
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.16103491072240317,
+      "learning_rate": 0.00016058141377238026,
+      "loss": 1.1022,
+      "step": 599
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.16029258404100036,
+      "learning_rate": 0.00016041262254525362,
+      "loss": 1.1508,
+      "step": 600
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.15686217802450533,
+      "learning_rate": 0.00016024355984248768,
+      "loss": 1.1471,
+      "step": 601
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.1520812354157866,
+      "learning_rate": 0.0001600742264237979,
+      "loss": 1.0994,
+      "step": 602
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.161954690104593,
+      "learning_rate": 0.0001599046230501163,
+      "loss": 1.1475,
+      "step": 603
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.16470046818858164,
+      "learning_rate": 0.00015973475048358795,
+      "loss": 1.1425,
+      "step": 604
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.17783426542152342,
+      "learning_rate": 0.00015956460948756765,
+      "loss": 1.1933,
+      "step": 605
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.20834095672163827,
+      "learning_rate": 0.0001593942008266164,
+      "loss": 1.1545,
+      "step": 606
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.1601922371369916,
+      "learning_rate": 0.00015922352526649803,
+      "loss": 1.1626,
+      "step": 607
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.17590795577463478,
+      "learning_rate": 0.00015905258357417569,
+      "loss": 1.1428,
+      "step": 608
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.16438364538914646,
+      "learning_rate": 0.00015888137651780845,
+      "loss": 1.1237,
+      "step": 609
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.1532404311365829,
+      "learning_rate": 0.00015870990486674792,
+      "loss": 1.1292,
+      "step": 610
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.1658092976448754,
+      "learning_rate": 0.0001585381693915346,
+      "loss": 1.2051,
+      "step": 611
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.17009266166893572,
+      "learning_rate": 0.00015836617086389468,
+      "loss": 1.1574,
+      "step": 612
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.17117032466097273,
+      "learning_rate": 0.00015819391005673626,
+      "loss": 1.1715,
+      "step": 613
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.16766945870038483,
+      "learning_rate": 0.00015802138774414622,
+      "loss": 1.1322,
+      "step": 614
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.16603674100502688,
+      "learning_rate": 0.00015784860470138633,
+      "loss": 1.1682,
+      "step": 615
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.1538296997751752,
+      "learning_rate": 0.00015767556170489025,
+      "loss": 1.0979,
+      "step": 616
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.16355862351540293,
+      "learning_rate": 0.00015750225953225968,
+      "loss": 1.0859,
+      "step": 617
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.17100029950170442,
+      "learning_rate": 0.00015732869896226094,
+      "loss": 1.1513,
+      "step": 618
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.16981502045992727,
+      "learning_rate": 0.0001571548807748215,
+      "loss": 1.1665,
+      "step": 619
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.15981257682597005,
+      "learning_rate": 0.00015698080575102661,
+      "loss": 1.0862,
+      "step": 620
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.17114664647857297,
+      "learning_rate": 0.00015680647467311557,
+      "loss": 1.1883,
+      "step": 621
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.16818704328900685,
+      "learning_rate": 0.00015663188832447833,
+      "loss": 1.1345,
+      "step": 622
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.16716958487866437,
+      "learning_rate": 0.0001564570474896519,
+      "loss": 1.1613,
+      "step": 623
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.16425076739154273,
+      "learning_rate": 0.00015628195295431697,
+      "loss": 1.1153,
+      "step": 624
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.1668267227341853,
+      "learning_rate": 0.0001561066055052941,
+      "loss": 1.1796,
+      "step": 625
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.15748311738845247,
+      "learning_rate": 0.00015593100593054064,
+      "loss": 1.125,
+      "step": 626
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.1578229320002536,
+      "learning_rate": 0.00015575515501914668,
+      "loss": 1.1751,
+      "step": 627
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.17990453419427754,
+      "learning_rate": 0.0001555790535613318,
+      "loss": 1.2107,
+      "step": 628
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.17170087096955436,
+      "learning_rate": 0.0001554027023484416,
+      "loss": 1.1704,
+      "step": 629
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.1641233552033525,
+      "learning_rate": 0.00015522610217294375,
+      "loss": 1.1399,
+      "step": 630
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.22816800313793714,
+      "learning_rate": 0.00015504925382842487,
+      "loss": 1.1374,
+      "step": 631
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.1699514304776241,
+      "learning_rate": 0.00015487215810958675,
+      "loss": 1.1668,
+      "step": 632
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.16580705156733896,
+      "learning_rate": 0.00015469481581224272,
+      "loss": 1.2467,
+      "step": 633
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.1621295978537637,
+      "learning_rate": 0.0001545172277333142,
+      "loss": 1.095,
+      "step": 634
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.15661235975760687,
+      "learning_rate": 0.00015433939467082713,
+      "loss": 1.1437,
+      "step": 635
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.15913894864881573,
+      "learning_rate": 0.00015416131742390827,
+      "loss": 1.1682,
+      "step": 636
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.17078546916822446,
+      "learning_rate": 0.00015398299679278172,
+      "loss": 1.1673,
+      "step": 637
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.17593093163481005,
+      "learning_rate": 0.00015380443357876518,
+      "loss": 1.1588,
+      "step": 638
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.16563306408223222,
+      "learning_rate": 0.00015362562858426654,
+      "loss": 1.1542,
+      "step": 639
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.15824017459526013,
+      "learning_rate": 0.0001534465826127801,
+      "loss": 1.136,
+      "step": 640
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.1555748015065671,
+      "learning_rate": 0.00015326729646888314,
+      "loss": 1.1744,
+      "step": 641
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.15054650982383877,
+      "learning_rate": 0.0001530877709582321,
+      "loss": 1.0996,
+      "step": 642
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.1732483344367302,
+      "learning_rate": 0.00015290800688755907,
+      "loss": 1.1375,
+      "step": 643
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.17086276533942882,
+      "learning_rate": 0.0001527280050646682,
+      "loss": 1.1399,
+      "step": 644
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.14791498583408913,
+      "learning_rate": 0.00015254776629843205,
+      "loss": 1.0112,
+      "step": 645
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.16492159023612152,
+      "learning_rate": 0.00015236729139878782,
+      "loss": 1.1179,
+      "step": 646
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.15550107018855622,
+      "learning_rate": 0.0001521865811767339,
+      "loss": 1.0872,
+      "step": 647
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.17086993477527834,
+      "learning_rate": 0.00015200563644432612,
+      "loss": 1.1747,
+      "step": 648
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.16832629109129904,
+      "learning_rate": 0.0001518244580146742,
+      "loss": 1.2623,
+      "step": 649
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.15555098188988337,
+      "learning_rate": 0.00015164304670193792,
+      "loss": 1.1172,
+      "step": 650
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.16504655864531245,
+      "learning_rate": 0.00015146140332132358,
+      "loss": 1.1615,
+      "step": 651
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.15904615747067738,
+      "learning_rate": 0.00015127952868908043,
+      "loss": 1.1841,
+      "step": 652
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.1664190253819244,
+      "learning_rate": 0.00015109742362249672,
+      "loss": 1.2258,
+      "step": 653
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.17622098883404838,
+      "learning_rate": 0.00015091508893989633,
+      "loss": 1.2141,
+      "step": 654
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.17253106779139057,
+      "learning_rate": 0.00015073252546063493,
+      "loss": 1.1289,
+      "step": 655
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.15924181512457228,
+      "learning_rate": 0.0001505497340050963,
+      "loss": 1.1836,
+      "step": 656
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.1755114666296529,
+      "learning_rate": 0.00015036671539468878,
+      "loss": 1.2109,
+      "step": 657
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.1623211151951732,
+      "learning_rate": 0.00015018347045184132,
+      "loss": 1.1922,
+      "step": 658
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.16895842955233992,
+      "learning_rate": 0.00015000000000000001,
+      "loss": 1.2402,
+      "step": 659
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.17126264186496687,
+      "learning_rate": 0.00014981630486362435,
+      "loss": 1.2327,
+      "step": 660
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.19632534974860108,
+      "learning_rate": 0.00014963238586818345,
+      "loss": 1.1586,
+      "step": 661
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.16708414744148634,
+      "learning_rate": 0.00014944824384015236,
+      "loss": 1.1766,
+      "step": 662
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.1800283830073218,
+      "learning_rate": 0.00014926387960700842,
+      "loss": 1.0902,
+      "step": 663
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.17570100482291343,
+      "learning_rate": 0.0001490792939972275,
+      "loss": 1.1875,
+      "step": 664
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.15904497800793038,
+      "learning_rate": 0.0001488944878402802,
+      "loss": 1.083,
+      "step": 665
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.1629549994998532,
+      "learning_rate": 0.00014870946196662822,
+      "loss": 1.1505,
+      "step": 666
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.17495490666041913,
+      "learning_rate": 0.00014852421720772062,
+      "loss": 1.1107,
+      "step": 667
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.1814722009088628,
+      "learning_rate": 0.00014833875439599004,
+      "loss": 1.2089,
+      "step": 668
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.16312442272141373,
+      "learning_rate": 0.00014815307436484898,
+      "loss": 1.1518,
+      "step": 669
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.18878089936272002,
+      "learning_rate": 0.00014796717794868607,
+      "loss": 1.1806,
+      "step": 670
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.171347321491223,
+      "learning_rate": 0.00014778106598286234,
+      "loss": 1.3189,
+      "step": 671
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.15561777538677532,
+      "learning_rate": 0.00014759473930370736,
+      "loss": 1.1071,
+      "step": 672
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.16095434527629326,
+      "learning_rate": 0.0001474081987485156,
+      "loss": 1.173,
+      "step": 673
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.1594519070011647,
+      "learning_rate": 0.00014722144515554264,
+      "loss": 1.1099,
+      "step": 674
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.15853021638823037,
+      "learning_rate": 0.00014703447936400134,
+      "loss": 1.1563,
+      "step": 675
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.21942200368949344,
+      "learning_rate": 0.00014684730221405814,
+      "loss": 1.1226,
+      "step": 676
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.16328115550493413,
+      "learning_rate": 0.00014665991454682924,
+      "loss": 1.1878,
+      "step": 677
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.16551970389762746,
+      "learning_rate": 0.00014647231720437686,
+      "loss": 1.055,
+      "step": 678
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.17275962400330097,
+      "learning_rate": 0.00014628451102970547,
+      "loss": 1.1302,
+      "step": 679
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.17269943862734122,
+      "learning_rate": 0.00014609649686675785,
+      "loss": 1.1635,
+      "step": 680
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.18932844828433326,
+      "learning_rate": 0.00014590827556041158,
+      "loss": 1.0954,
+      "step": 681
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.1670913427771278,
+      "learning_rate": 0.00014571984795647494,
+      "loss": 1.1336,
+      "step": 682
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.1727053222121231,
+      "learning_rate": 0.00014553121490168332,
+      "loss": 1.1477,
+      "step": 683
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.20281162440108022,
+      "learning_rate": 0.00014534237724369534,
+      "loss": 1.1364,
+      "step": 684
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.17595782723751183,
+      "learning_rate": 0.00014515333583108896,
+      "loss": 1.1977,
+      "step": 685
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.16755958800968615,
+      "learning_rate": 0.00014496409151335785,
+      "loss": 1.1315,
+      "step": 686
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.17908885139779734,
+      "learning_rate": 0.00014477464514090743,
+      "loss": 1.1851,
+      "step": 687
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.16047769884625865,
+      "learning_rate": 0.00014458499756505116,
+      "loss": 1.1137,
+      "step": 688
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.18864957014640948,
+      "learning_rate": 0.00014439514963800648,
+      "loss": 1.1962,
+      "step": 689
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.16074020271372416,
+      "learning_rate": 0.00014420510221289137,
+      "loss": 1.1448,
+      "step": 690
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.17166433855606958,
+      "learning_rate": 0.00014401485614372008,
+      "loss": 1.0487,
+      "step": 691
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.1773682331611409,
+      "learning_rate": 0.0001438244122853996,
+      "loss": 1.16,
+      "step": 692
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.16596601643110168,
+      "learning_rate": 0.00014363377149372584,
+      "loss": 1.0843,
+      "step": 693
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.16748936526973612,
+      "learning_rate": 0.0001434429346253794,
+      "loss": 1.1619,
+      "step": 694
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.17600274748860967,
+      "learning_rate": 0.00014325190253792222,
+      "loss": 1.151,
+      "step": 695
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.1721019830304793,
+      "learning_rate": 0.0001430606760897934,
+      "loss": 1.0609,
+      "step": 696
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.172361054646693,
+      "learning_rate": 0.00014286925614030542,
+      "loss": 1.1777,
+      "step": 697
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.1938123204770535,
+      "learning_rate": 0.00014267764354964038,
+      "loss": 1.1078,
+      "step": 698
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.171152731950173,
+      "learning_rate": 0.00014248583917884594,
+      "loss": 1.2344,
+      "step": 699
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.176915280903416,
+      "learning_rate": 0.00014229384388983167,
+      "loss": 1.1443,
+      "step": 700
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.16620426221482948,
+      "learning_rate": 0.00014210165854536494,
+      "loss": 1.1635,
+      "step": 701
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.17368168932295722,
+      "learning_rate": 0.0001419092840090673,
+      "loss": 1.2076,
+      "step": 702
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.16396003568599715,
+      "learning_rate": 0.0001417167211454104,
+      "loss": 1.1798,
+      "step": 703
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.17474861300401356,
+      "learning_rate": 0.0001415239708197122,
+      "loss": 1.1951,
+      "step": 704
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.17687469132536782,
+      "learning_rate": 0.00014133103389813302,
+      "loss": 1.1863,
+      "step": 705
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.1729350163866909,
+      "learning_rate": 0.0001411379112476717,
+      "loss": 1.1791,
+      "step": 706
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.16183372577765218,
+      "learning_rate": 0.0001409446037361617,
+      "loss": 1.11,
+      "step": 707
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.17468439327379603,
+      "learning_rate": 0.0001407511122322672,
+      "loss": 1.1459,
+      "step": 708
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.1746393388604993,
+      "learning_rate": 0.00014055743760547917,
+      "loss": 1.1782,
+      "step": 709
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.1709730768079813,
+      "learning_rate": 0.00014036358072611147,
+      "loss": 1.1729,
+      "step": 710
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.16338226575160056,
+      "learning_rate": 0.00014016954246529696,
+      "loss": 1.1637,
+      "step": 711
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.16522796461612693,
+      "learning_rate": 0.00013997532369498355,
+      "loss": 1.1786,
+      "step": 712
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.16443017398263532,
+      "learning_rate": 0.0001397809252879303,
+      "loss": 1.1885,
+      "step": 713
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.16678395716368374,
+      "learning_rate": 0.0001395863481177036,
+      "loss": 1.1576,
+      "step": 714
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.15796662849679724,
+      "learning_rate": 0.000139391593058673,
+      "loss": 1.1719,
+      "step": 715
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.1605406960709786,
+      "learning_rate": 0.00013919666098600753,
+      "loss": 1.1442,
+      "step": 716
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.16720350808668913,
+      "learning_rate": 0.00013900155277567157,
+      "loss": 1.1231,
+      "step": 717
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.16025549611436293,
+      "learning_rate": 0.00013880626930442113,
+      "loss": 1.0959,
+      "step": 718
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.1682327263659866,
+      "learning_rate": 0.00013861081144979974,
+      "loss": 1.1816,
+      "step": 719
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.16843691606816277,
+      "learning_rate": 0.00013841518009013445,
+      "loss": 1.1607,
+      "step": 720
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.17753480625595375,
+      "learning_rate": 0.0001382193761045322,
+      "loss": 1.2085,
+      "step": 721
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.175444570487372,
+      "learning_rate": 0.0001380234003728754,
+      "loss": 1.1883,
+      "step": 722
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.17844909537485765,
+      "learning_rate": 0.00013782725377581848,
+      "loss": 1.2548,
+      "step": 723
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.17827531614400435,
+      "learning_rate": 0.00013763093719478358,
+      "loss": 1.1753,
+      "step": 724
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.16568782268956592,
+      "learning_rate": 0.00013743445151195657,
+      "loss": 1.1269,
+      "step": 725
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.16229148230338714,
+      "learning_rate": 0.00013723779761028347,
+      "loss": 1.1233,
+      "step": 726
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.18071555500117806,
+      "learning_rate": 0.000137040976373466,
+      "loss": 1.1691,
+      "step": 727
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.16528568438217509,
+      "learning_rate": 0.000136843988685958,
+      "loss": 1.1044,
+      "step": 728
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.15621961073605914,
+      "learning_rate": 0.00013664683543296112,
+      "loss": 1.0585,
+      "step": 729
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.1821012365890871,
+      "learning_rate": 0.00013644951750042114,
+      "loss": 1.1061,
+      "step": 730
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.18832728793079637,
+      "learning_rate": 0.00013625203577502382,
+      "loss": 1.2088,
+      "step": 731
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.17821269011741103,
+      "learning_rate": 0.00013605439114419094,
+      "loss": 1.1076,
+      "step": 732
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.15816928920099374,
+      "learning_rate": 0.00013585658449607633,
+      "loss": 1.057,
+      "step": 733
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.1634582673036166,
+      "learning_rate": 0.00013565861671956187,
+      "loss": 1.2206,
+      "step": 734
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.16402056158577089,
+      "learning_rate": 0.00013546048870425356,
+      "loss": 1.0809,
+      "step": 735
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.18674575425072043,
+      "learning_rate": 0.0001352622013404774,
+      "loss": 1.0979,
+      "step": 736
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.16768128529630819,
+      "learning_rate": 0.00013506375551927547,
+      "loss": 1.1813,
+      "step": 737
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.17842009039615817,
+      "learning_rate": 0.00013486515213240188,
+      "loss": 1.1767,
+      "step": 738
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.16460609654310757,
+      "learning_rate": 0.0001346663920723188,
+      "loss": 1.1711,
+      "step": 739
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.16130836204900895,
+      "learning_rate": 0.00013446747623219255,
+      "loss": 1.1506,
+      "step": 740
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.19357518099915463,
+      "learning_rate": 0.00013426840550588933,
+      "loss": 1.1454,
+      "step": 741
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.16818483070815926,
+      "learning_rate": 0.0001340691807879714,
+      "loss": 1.105,
+      "step": 742
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.17634247767232533,
+      "learning_rate": 0.00013386980297369307,
+      "loss": 1.1173,
+      "step": 743
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.17275588243641285,
+      "learning_rate": 0.0001336702729589965,
+      "loss": 1.1614,
+      "step": 744
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.18630211892037335,
+      "learning_rate": 0.00013347059164050794,
+      "loss": 1.0822,
+      "step": 745
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.1610612417401003,
+      "learning_rate": 0.0001332707599155334,
+      "loss": 1.11,
+      "step": 746
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.1710900937487699,
+      "learning_rate": 0.00013307077868205487,
+      "loss": 1.1556,
+      "step": 747
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.16789088046962233,
+      "learning_rate": 0.00013287064883872612,
+      "loss": 1.166,
+      "step": 748
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.17152792861267624,
+      "learning_rate": 0.00013267037128486883,
+      "loss": 1.2076,
+      "step": 749
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.16987939988427406,
+      "learning_rate": 0.00013246994692046836,
+      "loss": 1.1892,
+      "step": 750
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.17050807620104472,
+      "learning_rate": 0.00013226937664616976,
+      "loss": 1.1405,
+      "step": 751
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.1619859173580362,
+      "learning_rate": 0.00013206866136327388,
+      "loss": 1.15,
+      "step": 752
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.1693102604231561,
+      "learning_rate": 0.00013186780197373306,
+      "loss": 1.1108,
+      "step": 753
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.15751817301555188,
+      "learning_rate": 0.00013166679938014726,
+      "loss": 1.1075,
+      "step": 754
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.17344055609711165,
+      "learning_rate": 0.00013146565448576004,
+      "loss": 1.2239,
+      "step": 755
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.16419033762795351,
+      "learning_rate": 0.00013126436819445422,
+      "loss": 1.2146,
+      "step": 756
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.17241446484871606,
+      "learning_rate": 0.00013106294141074825,
+      "loss": 1.1808,
+      "step": 757
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.16299230805525922,
+      "learning_rate": 0.0001308613750397917,
+      "loss": 1.1639,
+      "step": 758
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.15993384139183944,
+      "learning_rate": 0.00013065966998736155,
+      "loss": 1.0664,
+      "step": 759
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.17951699730636844,
+      "learning_rate": 0.00013045782715985792,
+      "loss": 1.1565,
+      "step": 760
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.16420615622643212,
+      "learning_rate": 0.0001302558474643,
+      "loss": 1.1675,
+      "step": 761
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.1718558627273243,
+      "learning_rate": 0.0001300537318083221,
+      "loss": 1.1758,
+      "step": 762
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.16040004880583997,
+      "learning_rate": 0.00012985148110016947,
+      "loss": 1.1286,
+      "step": 763
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.1644924005506398,
+      "learning_rate": 0.0001296490962486942,
+      "loss": 1.0901,
+      "step": 764
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.17508769995003778,
+      "learning_rate": 0.00012944657816335123,
+      "loss": 1.158,
+      "step": 765
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.16636235063611218,
+      "learning_rate": 0.0001292439277541942,
+      "loss": 1.1506,
+      "step": 766
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.1651752039403403,
+      "learning_rate": 0.00012904114593187136,
+      "loss": 1.1167,
+      "step": 767
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.1727449182027404,
+      "learning_rate": 0.0001288382336076215,
+      "loss": 1.1208,
+      "step": 768
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.1678888505155555,
+      "learning_rate": 0.00012863519169326984,
+      "loss": 1.1164,
+      "step": 769
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.1717914946350261,
+      "learning_rate": 0.0001284320211012239,
+      "loss": 1.174,
+      "step": 770
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.15833038102003086,
+      "learning_rate": 0.00012822872274446958,
+      "loss": 1.0557,
+      "step": 771
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.16693919668597118,
+      "learning_rate": 0.00012802529753656668,
+      "loss": 1.2202,
+      "step": 772
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.1608034670310254,
+      "learning_rate": 0.0001278217463916453,
+      "loss": 1.0761,
+      "step": 773
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.16522372158185394,
+      "learning_rate": 0.0001276180702244012,
+      "loss": 1.114,
+      "step": 774
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.15699403016449,
+      "learning_rate": 0.00012741426995009213,
+      "loss": 1.0487,
+      "step": 775
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.1724966872571711,
+      "learning_rate": 0.00012721034648453353,
+      "loss": 1.2198,
+      "step": 776
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.16612867386952304,
+      "learning_rate": 0.00012700630074409427,
+      "loss": 1.1481,
+      "step": 777
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.1755115998846381,
+      "learning_rate": 0.0001268021336456929,
+      "loss": 1.1295,
+      "step": 778
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.16610612624318807,
+      "learning_rate": 0.00012659784610679318,
+      "loss": 1.0735,
+      "step": 779
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.17210665028457767,
+      "learning_rate": 0.0001263934390454001,
+      "loss": 1.128,
+      "step": 780
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.17115180879996908,
+      "learning_rate": 0.00012618891338005573,
+      "loss": 1.2114,
+      "step": 781
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.1708121456807184,
+      "learning_rate": 0.0001259842700298352,
+      "loss": 1.2037,
+      "step": 782
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.16413369503599784,
+      "learning_rate": 0.00012577950991434248,
+      "loss": 1.115,
+      "step": 783
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.16575779512688402,
+      "learning_rate": 0.0001255746339537061,
+      "loss": 1.13,
+      "step": 784
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.16787768159350958,
+      "learning_rate": 0.00012536964306857526,
+      "loss": 1.1187,
+      "step": 785
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.161257935533138,
+      "learning_rate": 0.00012516453818011566,
+      "loss": 1.1272,
+      "step": 786
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.17163817923820932,
+      "learning_rate": 0.00012495932021000517,
+      "loss": 1.1371,
+      "step": 787
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.16451801686420228,
+      "learning_rate": 0.0001247539900804299,
+      "loss": 1.1234,
+      "step": 788
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.4897545674498291,
+      "learning_rate": 0.00012454854871407994,
+      "loss": 1.1276,
+      "step": 789
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.1674813107615878,
+      "learning_rate": 0.00012434299703414524,
+      "loss": 1.0717,
+      "step": 790
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.16889265391095676,
+      "learning_rate": 0.0001241373359643114,
+      "loss": 1.1094,
+      "step": 791
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.17807200243744,
+      "learning_rate": 0.0001239315664287558,
+      "loss": 1.2072,
+      "step": 792
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.1821416845887793,
+      "learning_rate": 0.00012372568935214298,
+      "loss": 1.1574,
+      "step": 793
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.16787645348154986,
+      "learning_rate": 0.00012351970565962085,
+      "loss": 1.0884,
+      "step": 794
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.17767787321450104,
+      "learning_rate": 0.00012331361627681645,
+      "loss": 1.1561,
+      "step": 795
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.16990542040230855,
+      "learning_rate": 0.00012310742212983167,
+      "loss": 1.1454,
+      "step": 796
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.17040623149046621,
+      "learning_rate": 0.00012290112414523928,
+      "loss": 1.1143,
+      "step": 797
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.17383642491755713,
+      "learning_rate": 0.00012269472325007858,
+      "loss": 1.1502,
+      "step": 798
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.1670271756098212,
+      "learning_rate": 0.00012248822037185138,
+      "loss": 1.1872,
+      "step": 799
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.17338597290651167,
+      "learning_rate": 0.0001222816164385177,
+      "loss": 1.201,
+      "step": 800
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.16342647338148947,
+      "learning_rate": 0.00012207491237849172,
+      "loss": 1.0994,
+      "step": 801
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.15847559756714308,
+      "learning_rate": 0.0001218681091206376,
+      "loss": 1.1272,
+      "step": 802
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.17295023549456498,
+      "learning_rate": 0.00012166120759426514,
+      "loss": 1.1026,
+      "step": 803
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.17574482838436642,
+      "learning_rate": 0.00012145420872912585,
+      "loss": 1.1785,
+      "step": 804
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.1636907763577917,
+      "learning_rate": 0.0001212471134554086,
+      "loss": 1.125,
+      "step": 805
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.1683476239550862,
+      "learning_rate": 0.00012103992270373547,
+      "loss": 1.1196,
+      "step": 806
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.1634846923426851,
+      "learning_rate": 0.00012083263740515765,
+      "loss": 1.1666,
+      "step": 807
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.20362360607016913,
+      "learning_rate": 0.00012062525849115107,
+      "loss": 1.2166,
+      "step": 808
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.15520795046759722,
+      "learning_rate": 0.00012041778689361254,
+      "loss": 1.1046,
+      "step": 809
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.17498352958051774,
+      "learning_rate": 0.00012021022354485514,
+      "loss": 1.1331,
+      "step": 810
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.17105423893572264,
+      "learning_rate": 0.00012000256937760445,
+      "loss": 1.0799,
+      "step": 811
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.16837214432904796,
+      "learning_rate": 0.00011979482532499401,
+      "loss": 1.1856,
+      "step": 812
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.1717737899447442,
+      "learning_rate": 0.00011958699232056134,
+      "loss": 1.1738,
+      "step": 813
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.16901679352588425,
+      "learning_rate": 0.0001193790712982437,
+      "loss": 1.1361,
+      "step": 814
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.39457461047161113,
+      "learning_rate": 0.00011917106319237386,
+      "loss": 1.1095,
+      "step": 815
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.1689082831860728,
+      "learning_rate": 0.00011896296893767587,
+      "loss": 1.1518,
+      "step": 816
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.17381008421938537,
+      "learning_rate": 0.00011875478946926093,
+      "loss": 1.1251,
+      "step": 817
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.1723002103848165,
+      "learning_rate": 0.00011854652572262323,
+      "loss": 1.1128,
+      "step": 818
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.1799457776202214,
+      "learning_rate": 0.00011833817863363564,
+      "loss": 1.1702,
+      "step": 819
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.18319688210909812,
+      "learning_rate": 0.00011812974913854545,
+      "loss": 1.1537,
+      "step": 820
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.1716604871697964,
+      "learning_rate": 0.0001179212381739704,
+      "loss": 1.1567,
+      "step": 821
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.17098997576829758,
+      "learning_rate": 0.00011771264667689427,
+      "loss": 1.1309,
+      "step": 822
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.2607304915983766,
+      "learning_rate": 0.00011750397558466273,
+      "loss": 1.1151,
+      "step": 823
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.1869030292628108,
+      "learning_rate": 0.00011729522583497912,
+      "loss": 1.0764,
+      "step": 824
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.1969233515350324,
+      "learning_rate": 0.00011708639836590023,
+      "loss": 1.0986,
+      "step": 825
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.17794486028770468,
+      "learning_rate": 0.00011687749411583213,
+      "loss": 1.1327,
+      "step": 826
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.19408346665780069,
+      "learning_rate": 0.00011666851402352588,
+      "loss": 1.0768,
+      "step": 827
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.5122187922492092,
+      "learning_rate": 0.00011645945902807341,
+      "loss": 1.1223,
+      "step": 828
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.18948615786386766,
+      "learning_rate": 0.00011625033006890315,
+      "loss": 1.1452,
+      "step": 829
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.16917117240636045,
+      "learning_rate": 0.00011604112808577603,
+      "loss": 1.131,
+      "step": 830
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.177460873056157,
+      "learning_rate": 0.00011583185401878101,
+      "loss": 1.1691,
+      "step": 831
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.2727518607538819,
+      "learning_rate": 0.00011562250880833104,
+      "loss": 1.185,
+      "step": 832
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.1722336921590558,
+      "learning_rate": 0.00011541309339515869,
+      "loss": 1.2202,
+      "step": 833
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.16731429410912582,
+      "learning_rate": 0.00011520360872031209,
+      "loss": 1.1114,
+      "step": 834
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.1747423744661194,
+      "learning_rate": 0.00011499405572515059,
+      "loss": 1.0929,
+      "step": 835
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.18705677818301408,
+      "learning_rate": 0.00011478443535134049,
+      "loss": 1.101,
+      "step": 836
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.18090585152937122,
+      "learning_rate": 0.00011457474854085096,
+      "loss": 1.2376,
+      "step": 837
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.1572625515956593,
+      "learning_rate": 0.00011436499623594963,
+      "loss": 1.1039,
+      "step": 838
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.18868561701551964,
+      "learning_rate": 0.00011415517937919846,
+      "loss": 1.1091,
+      "step": 839
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.3765201515989829,
+      "learning_rate": 0.00011394529891344958,
+      "loss": 1.1904,
+      "step": 840
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.17659930390592027,
+      "learning_rate": 0.00011373535578184082,
+      "loss": 1.1767,
+      "step": 841
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.16250915193456436,
+      "learning_rate": 0.00011352535092779173,
+      "loss": 1.0396,
+      "step": 842
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.17679862621844297,
+      "learning_rate": 0.00011331528529499909,
+      "loss": 1.0899,
+      "step": 843
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.17303502214887417,
+      "learning_rate": 0.00011310515982743293,
+      "loss": 1.1623,
+      "step": 844
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.17241019588300063,
+      "learning_rate": 0.00011289497546933212,
+      "loss": 1.1761,
+      "step": 845
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.19940689264510733,
+      "learning_rate": 0.00011268473316520007,
+      "loss": 1.186,
+      "step": 846
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.17205111928537736,
+      "learning_rate": 0.00011247443385980078,
+      "loss": 1.1669,
+      "step": 847
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.168610859884427,
+      "learning_rate": 0.0001122640784981542,
+      "loss": 1.1223,
+      "step": 848
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.1571679218925349,
+      "learning_rate": 0.0001120536680255323,
+      "loss": 1.1017,
+      "step": 849
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.1731635445936976,
+      "learning_rate": 0.00011184320338745467,
+      "loss": 1.1549,
+      "step": 850
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.16986390394426032,
+      "learning_rate": 0.00011163268552968423,
+      "loss": 1.1561,
+      "step": 851
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.17807860518694296,
+      "learning_rate": 0.00011142211539822318,
+      "loss": 1.1191,
+      "step": 852
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.1848822566355218,
+      "learning_rate": 0.0001112114939393085,
+      "loss": 1.1265,
+      "step": 853
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.16154933469362054,
+      "learning_rate": 0.00011100082209940795,
+      "loss": 1.1696,
+      "step": 854
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.17539593545795423,
+      "learning_rate": 0.00011079010082521557,
+      "loss": 1.2098,
+      "step": 855
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.16955097983529596,
+      "learning_rate": 0.00011057933106364758,
+      "loss": 1.134,
+      "step": 856
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.1627286025891699,
+      "learning_rate": 0.00011036851376183812,
+      "loss": 1.0968,
+      "step": 857
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.18241009216885773,
+      "learning_rate": 0.0001101576498671349,
+      "loss": 1.0784,
+      "step": 858
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.1602213810951013,
+      "learning_rate": 0.00010994674032709513,
+      "loss": 1.0272,
+      "step": 859
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.17710273355839293,
+      "learning_rate": 0.00010973578608948094,
+      "loss": 1.1807,
+      "step": 860
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.16846043025211901,
+      "learning_rate": 0.00010952478810225548,
+      "loss": 1.1237,
+      "step": 861
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.17578602240771765,
+      "learning_rate": 0.00010931374731357841,
+      "loss": 1.1305,
+      "step": 862
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.1741114211076835,
+      "learning_rate": 0.0001091026646718018,
+      "loss": 1.0663,
+      "step": 863
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.17419627851243186,
+      "learning_rate": 0.0001088915411254657,
+      "loss": 1.0864,
+      "step": 864
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.1941495171655647,
+      "learning_rate": 0.00010868037762329404,
+      "loss": 1.1471,
+      "step": 865
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.16828482820770493,
+      "learning_rate": 0.0001084691751141903,
+      "loss": 1.1386,
+      "step": 866
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.1812212801001379,
+      "learning_rate": 0.00010825793454723325,
+      "loss": 1.193,
+      "step": 867
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.31177412914918357,
+      "learning_rate": 0.00010804665687167262,
+      "loss": 1.1416,
+      "step": 868
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.16307908317649347,
+      "learning_rate": 0.00010783534303692493,
+      "loss": 1.0742,
+      "step": 869
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.17264040918528867,
+      "learning_rate": 0.00010762399399256917,
+      "loss": 1.1196,
+      "step": 870
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.16662882952717148,
+      "learning_rate": 0.00010741261068834265,
+      "loss": 1.1095,
+      "step": 871
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.16909860632274898,
+      "learning_rate": 0.00010720119407413647,
+      "loss": 1.0978,
+      "step": 872
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.1841227749300958,
+      "learning_rate": 0.00010698974509999158,
+      "loss": 1.1249,
+      "step": 873
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.163410587005958,
+      "learning_rate": 0.00010677826471609422,
+      "loss": 1.1027,
+      "step": 874
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.16473194468337188,
+      "learning_rate": 0.00010656675387277182,
+      "loss": 1.0452,
+      "step": 875
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.16465304101505895,
+      "learning_rate": 0.00010635521352048872,
+      "loss": 1.0975,
+      "step": 876
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.1648559678877552,
+      "learning_rate": 0.00010614364460984176,
+      "loss": 1.1865,
+      "step": 877
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.17182083260240222,
+      "learning_rate": 0.00010593204809155628,
+      "loss": 1.1666,
+      "step": 878
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.262043699845629,
+      "learning_rate": 0.00010572042491648149,
+      "loss": 1.1705,
+      "step": 879
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.1642146793634745,
+      "learning_rate": 0.00010550877603558655,
+      "loss": 1.0943,
+      "step": 880
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.1740460470258003,
+      "learning_rate": 0.00010529710239995605,
+      "loss": 1.0492,
+      "step": 881
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.16424504391420278,
+      "learning_rate": 0.0001050854049607858,
+      "loss": 1.0877,
+      "step": 882
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.1754464752434157,
+      "learning_rate": 0.00010487368466937866,
+      "loss": 1.2004,
+      "step": 883
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.17019599469065266,
+      "learning_rate": 0.00010466194247714008,
+      "loss": 1.1736,
+      "step": 884
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.16453531456339282,
+      "learning_rate": 0.00010445017933557404,
+      "loss": 1.0682,
+      "step": 885
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.16576841634828685,
+      "learning_rate": 0.00010423839619627853,
+      "loss": 1.1239,
+      "step": 886
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.17959933112877494,
+      "learning_rate": 0.00010402659401094152,
+      "loss": 1.1741,
+      "step": 887
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.1764446278189948,
+      "learning_rate": 0.00010381477373133652,
+      "loss": 1.1465,
+      "step": 888
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.1704634166756464,
+      "learning_rate": 0.0001036029363093183,
+      "loss": 1.2092,
+      "step": 889
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.16952721202785456,
+      "learning_rate": 0.00010339108269681874,
+      "loss": 1.1478,
+      "step": 890
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.1787903145977452,
+      "learning_rate": 0.00010317921384584244,
+      "loss": 1.1818,
+      "step": 891
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.1623533169508836,
+      "learning_rate": 0.00010296733070846252,
+      "loss": 1.1352,
+      "step": 892
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.17651784025914782,
+      "learning_rate": 0.00010275543423681621,
+      "loss": 1.175,
+      "step": 893
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.16473007072288753,
+      "learning_rate": 0.00010254352538310075,
+      "loss": 1.1837,
+      "step": 894
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.16909571695790346,
+      "learning_rate": 0.00010233160509956894,
+      "loss": 1.1435,
+      "step": 895
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.1904744004900901,
+      "learning_rate": 0.000102119674338525,
+      "loss": 1.1549,
+      "step": 896
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.17373949711988154,
+      "learning_rate": 0.00010190773405232024,
+      "loss": 1.202,
+      "step": 897
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.18192103377408994,
+      "learning_rate": 0.00010169578519334873,
+      "loss": 1.1851,
+      "step": 898
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.18229237608440976,
+      "learning_rate": 0.0001014838287140431,
+      "loss": 1.1233,
+      "step": 899
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.16940278596266165,
+      "learning_rate": 0.00010127186556687019,
+      "loss": 1.169,
+      "step": 900
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.18262078712459354,
+      "learning_rate": 0.00010105989670432681,
+      "loss": 1.199,
+      "step": 901
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.16747513400418446,
+      "learning_rate": 0.00010084792307893552,
+      "loss": 1.0666,
+      "step": 902
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.1746037115339502,
+      "learning_rate": 0.00010063594564324012,
+      "loss": 1.1885,
+      "step": 903
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.1732153359208303,
+      "learning_rate": 0.00010042396534980176,
+      "loss": 1.1155,
+      "step": 904
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.18911672921606115,
+      "learning_rate": 0.00010021198315119424,
+      "loss": 1.1495,
+      "step": 905
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.1810967486335703,
+      "learning_rate": 0.0001,
+      "loss": 1.2386,
+      "step": 906
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.16208600336701465,
+      "learning_rate": 9.978801684880578e-05,
+      "loss": 1.1016,
+      "step": 907
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.17108242412955507,
+      "learning_rate": 9.957603465019826e-05,
+      "loss": 1.1659,
+      "step": 908
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.17777524975299183,
+      "learning_rate": 9.93640543567599e-05,
+      "loss": 1.1877,
+      "step": 909
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.32271912031116656,
+      "learning_rate": 9.91520769210645e-05,
+      "loss": 1.1254,
+      "step": 910
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.16598806172835462,
+      "learning_rate": 9.894010329567323e-05,
+      "loss": 1.187,
+      "step": 911
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.17032736087667158,
+      "learning_rate": 9.872813443312984e-05,
+      "loss": 1.2073,
+      "step": 912
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.16171590357915658,
+      "learning_rate": 9.851617128595694e-05,
+      "loss": 1.0839,
+      "step": 913
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.22200097454760384,
+      "learning_rate": 9.830421480665128e-05,
+      "loss": 1.1299,
+      "step": 914
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.18159512381331983,
+      "learning_rate": 9.809226594767978e-05,
+      "loss": 1.1991,
+      "step": 915
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.1729262170321024,
+      "learning_rate": 9.788032566147505e-05,
+      "loss": 1.0962,
+      "step": 916
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.17432638214872634,
+      "learning_rate": 9.766839490043108e-05,
+      "loss": 1.1669,
+      "step": 917
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.16088749813770933,
+      "learning_rate": 9.745647461689931e-05,
+      "loss": 1.0548,
+      "step": 918
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.1741820101835776,
+      "learning_rate": 9.724456576318381e-05,
+      "loss": 1.1986,
+      "step": 919
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.1749247436598961,
+      "learning_rate": 9.70326692915375e-05,
+      "loss": 1.1474,
+      "step": 920
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.1637038092120215,
+      "learning_rate": 9.682078615415754e-05,
+      "loss": 1.1231,
+      "step": 921
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.18807058316399383,
+      "learning_rate": 9.660891730318128e-05,
+      "loss": 1.2045,
+      "step": 922
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.1748352594360658,
+      "learning_rate": 9.639706369068171e-05,
+      "loss": 1.1689,
+      "step": 923
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.1716288593635444,
+      "learning_rate": 9.61852262686635e-05,
+      "loss": 1.1572,
+      "step": 924
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.18271775085545605,
+      "learning_rate": 9.597340598905852e-05,
+      "loss": 1.21,
+      "step": 925
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.1717878271479738,
+      "learning_rate": 9.576160380372149e-05,
+      "loss": 1.0988,
+      "step": 926
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.1680182822271049,
+      "learning_rate": 9.5549820664426e-05,
+      "loss": 1.0968,
+      "step": 927
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.1742620341628087,
+      "learning_rate": 9.533805752285993e-05,
+      "loss": 1.1635,
+      "step": 928
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.16338762068272702,
+      "learning_rate": 9.512631533062138e-05,
+      "loss": 0.9968,
+      "step": 929
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.17102172627425277,
+      "learning_rate": 9.491459503921421e-05,
+      "loss": 1.1481,
+      "step": 930
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.166342175891061,
+      "learning_rate": 9.470289760004398e-05,
+      "loss": 1.0721,
+      "step": 931
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.2215993724024064,
+      "learning_rate": 9.449122396441345e-05,
+      "loss": 1.1877,
+      "step": 932
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.1659659894828161,
+      "learning_rate": 9.427957508351852e-05,
+      "loss": 1.1691,
+      "step": 933
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.1656871284817402,
+      "learning_rate": 9.406795190844376e-05,
+      "loss": 1.06,
+      "step": 934
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.1586845482565193,
+      "learning_rate": 9.385635539015825e-05,
+      "loss": 1.1,
+      "step": 935
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.16465420669616035,
+      "learning_rate": 9.364478647951133e-05,
+      "loss": 1.1403,
+      "step": 936
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.15668641868741373,
+      "learning_rate": 9.343324612722819e-05,
+      "loss": 1.1075,
+      "step": 937
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.16611169088286057,
+      "learning_rate": 9.32217352839058e-05,
+      "loss": 1.1256,
+      "step": 938
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.17301783500328655,
+      "learning_rate": 9.301025490000841e-05,
+      "loss": 1.1797,
+      "step": 939
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.17015053210523748,
+      "learning_rate": 9.279880592586354e-05,
+      "loss": 1.1087,
+      "step": 940
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.17913490315482883,
+      "learning_rate": 9.25873893116574e-05,
+      "loss": 1.1673,
+      "step": 941
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.18716341523653843,
+      "learning_rate": 9.237600600743085e-05,
+      "loss": 1.1448,
+      "step": 942
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.16020067287870843,
+      "learning_rate": 9.216465696307512e-05,
+      "loss": 1.1152,
+      "step": 943
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.18082661292174337,
+      "learning_rate": 9.195334312832742e-05,
+      "loss": 1.1923,
+      "step": 944
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.16932315705059348,
+      "learning_rate": 9.174206545276677e-05,
+      "loss": 1.2025,
+      "step": 945
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.1773723182281052,
+      "learning_rate": 9.153082488580968e-05,
+      "loss": 1.1583,
+      "step": 946
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.1650084811263765,
+      "learning_rate": 9.131962237670598e-05,
+      "loss": 1.1456,
+      "step": 947
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.17296637806848497,
+      "learning_rate": 9.11084588745343e-05,
+      "loss": 1.1188,
+      "step": 948
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.18441757049220042,
+      "learning_rate": 9.089733532819824e-05,
+      "loss": 1.1057,
+      "step": 949
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.16479278966433117,
+      "learning_rate": 9.068625268642161e-05,
+      "loss": 0.9686,
+      "step": 950
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.1721522611977686,
+      "learning_rate": 9.047521189774455e-05,
+      "loss": 1.1593,
+      "step": 951
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.17693809369137542,
+      "learning_rate": 9.026421391051907e-05,
+      "loss": 1.141,
+      "step": 952
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.16748786371543328,
+      "learning_rate": 9.005325967290488e-05,
+      "loss": 1.0954,
+      "step": 953
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.16738864271495277,
+      "learning_rate": 8.984235013286511e-05,
+      "loss": 1.1954,
+      "step": 954
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.17608800377093817,
+      "learning_rate": 8.963148623816191e-05,
+      "loss": 1.2054,
+      "step": 955
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.1572056318124515,
+      "learning_rate": 8.942066893635246e-05,
+      "loss": 1.062,
+      "step": 956
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.17321331852374994,
+      "learning_rate": 8.920989917478447e-05,
+      "loss": 1.1774,
+      "step": 957
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.24022668199221645,
+      "learning_rate": 8.899917790059208e-05,
+      "loss": 1.1422,
+      "step": 958
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.17078130002322278,
+      "learning_rate": 8.878850606069153e-05,
+      "loss": 1.102,
+      "step": 959
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.18816676453808523,
+      "learning_rate": 8.857788460177686e-05,
+      "loss": 1.1228,
+      "step": 960
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.17061330738089536,
+      "learning_rate": 8.836731447031581e-05,
+      "loss": 1.1664,
+      "step": 961
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.15744686305416977,
+      "learning_rate": 8.815679661254537e-05,
+      "loss": 1.0526,
+      "step": 962
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.17140964720406338,
+      "learning_rate": 8.79463319744677e-05,
+      "loss": 1.1015,
+      "step": 963
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.177839632569439,
+      "learning_rate": 8.77359215018458e-05,
+      "loss": 1.179,
+      "step": 964
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.18976111884283367,
+      "learning_rate": 8.752556614019923e-05,
+      "loss": 1.1901,
+      "step": 965
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.17757617341847579,
+      "learning_rate": 8.731526683479992e-05,
+      "loss": 1.1195,
+      "step": 966
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.18025608157061349,
+      "learning_rate": 8.710502453066792e-05,
+      "loss": 1.1681,
+      "step": 967
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.17991548690786335,
+      "learning_rate": 8.689484017256711e-05,
+      "loss": 1.0382,
+      "step": 968
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.18377930665665893,
+      "learning_rate": 8.668471470500095e-05,
+      "loss": 1.2159,
+      "step": 969
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.16785777204268718,
+      "learning_rate": 8.647464907220832e-05,
+      "loss": 1.1001,
+      "step": 970
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.1718542715910871,
+      "learning_rate": 8.626464421815919e-05,
+      "loss": 1.1362,
+      "step": 971
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.17567452985919224,
+      "learning_rate": 8.605470108655045e-05,
+      "loss": 1.0701,
+      "step": 972
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.16645014392512258,
+      "learning_rate": 8.584482062080154e-05,
+      "loss": 1.1224,
+      "step": 973
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.1683034297078507,
+      "learning_rate": 8.563500376405041e-05,
+      "loss": 1.1833,
+      "step": 974
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.17027006999646288,
+      "learning_rate": 8.542525145914905e-05,
+      "loss": 1.2105,
+      "step": 975
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.16661484142693264,
+      "learning_rate": 8.521556464865954e-05,
+      "loss": 1.1345,
+      "step": 976
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.16535108955142738,
+      "learning_rate": 8.500594427484946e-05,
+      "loss": 1.0801,
+      "step": 977
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.1619767851731314,
+      "learning_rate": 8.479639127968792e-05,
+      "loss": 1.1398,
+      "step": 978
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.16148830590153632,
+      "learning_rate": 8.458690660484134e-05,
+      "loss": 1.1714,
+      "step": 979
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.1718526387843433,
+      "learning_rate": 8.4377491191669e-05,
+      "loss": 1.1196,
+      "step": 980
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.1589791532312397,
+      "learning_rate": 8.4168145981219e-05,
+      "loss": 1.0975,
+      "step": 981
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.18228960673193118,
+      "learning_rate": 8.395887191422397e-05,
+      "loss": 1.0911,
+      "step": 982
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.17051635436251017,
+      "learning_rate": 8.374966993109687e-05,
+      "loss": 1.1156,
+      "step": 983
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.1734092864786433,
+      "learning_rate": 8.35405409719266e-05,
+      "loss": 1.1387,
+      "step": 984
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.17960160313198717,
+      "learning_rate": 8.333148597647414e-05,
+      "loss": 1.1255,
+      "step": 985
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.1667460148360989,
+      "learning_rate": 8.312250588416791e-05,
+      "loss": 1.0877,
+      "step": 986
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.17227304205526361,
+      "learning_rate": 8.291360163409978e-05,
+      "loss": 1.1643,
+      "step": 987
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.16172787345643957,
+      "learning_rate": 8.270477416502091e-05,
+      "loss": 1.0813,
+      "step": 988
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.17717380607042416,
+      "learning_rate": 8.249602441533726e-05,
+      "loss": 1.1653,
+      "step": 989
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.17646325594155327,
+      "learning_rate": 8.228735332310575e-05,
+      "loss": 1.1671,
+      "step": 990
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.16889550666185796,
+      "learning_rate": 8.207876182602958e-05,
+      "loss": 1.1068,
+      "step": 991
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.1668114849825541,
+      "learning_rate": 8.187025086145458e-05,
+      "loss": 1.1467,
+      "step": 992
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.1721245962364379,
+      "learning_rate": 8.16618213663644e-05,
+      "loss": 1.1377,
+      "step": 993
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.1929335873892253,
+      "learning_rate": 8.145347427737678e-05,
+      "loss": 1.119,
+      "step": 994
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.16623130316399012,
+      "learning_rate": 8.12452105307391e-05,
+      "loss": 1.0687,
+      "step": 995
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.18085333131496864,
+      "learning_rate": 8.103703106232416e-05,
+      "loss": 1.1517,
+      "step": 996
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.17906886829733978,
+      "learning_rate": 8.082893680762619e-05,
+      "loss": 1.1325,
+      "step": 997
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.16359826556853824,
+      "learning_rate": 8.062092870175628e-05,
+      "loss": 1.0353,
+      "step": 998
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.1626009671933035,
+      "learning_rate": 8.041300767943867e-05,
+      "loss": 1.1598,
+      "step": 999
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.16643269286545453,
+      "learning_rate": 8.0205174675006e-05,
+      "loss": 1.1226,
+      "step": 1000
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.1699654548245084,
+      "learning_rate": 7.999743062239557e-05,
+      "loss": 1.0999,
+      "step": 1001
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.1827266429905637,
+      "learning_rate": 7.978977645514487e-05,
+      "loss": 1.2791,
+      "step": 1002
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.44434388351574605,
+      "learning_rate": 7.958221310638749e-05,
+      "loss": 1.1693,
+      "step": 1003
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.17634755268047692,
+      "learning_rate": 7.937474150884897e-05,
+      "loss": 1.0845,
+      "step": 1004
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.177726814790075,
+      "learning_rate": 7.916736259484239e-05,
+      "loss": 1.1849,
+      "step": 1005
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.16479013332848777,
+      "learning_rate": 7.896007729626457e-05,
+      "loss": 1.1296,
+      "step": 1006
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.1607507116543152,
+      "learning_rate": 7.875288654459144e-05,
+      "loss": 1.0829,
+      "step": 1007
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.16387244898588724,
+      "learning_rate": 7.854579127087417e-05,
+      "loss": 1.1604,
+      "step": 1008
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.163742117997135,
+      "learning_rate": 7.833879240573487e-05,
+      "loss": 1.0757,
+      "step": 1009
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.16638334168591146,
+      "learning_rate": 7.813189087936243e-05,
+      "loss": 1.1637,
+      "step": 1010
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.16818864513814694,
+      "learning_rate": 7.792508762150833e-05,
+      "loss": 1.1564,
+      "step": 1011
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.17901829622442805,
+      "learning_rate": 7.771838356148232e-05,
+      "loss": 1.1841,
+      "step": 1012
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.16614583867770816,
+      "learning_rate": 7.751177962814866e-05,
+      "loss": 1.1255,
+      "step": 1013
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.19013974114118293,
+      "learning_rate": 7.730527674992143e-05,
+      "loss": 1.1488,
+      "step": 1014
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.17442438513245137,
+      "learning_rate": 7.709887585476075e-05,
+      "loss": 1.1066,
+      "step": 1015
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.17701832779209928,
+      "learning_rate": 7.689257787016834e-05,
+      "loss": 1.1755,
+      "step": 1016
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.16936912919421254,
+      "learning_rate": 7.668638372318359e-05,
+      "loss": 1.1057,
+      "step": 1017
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.19112857435989614,
+      "learning_rate": 7.648029434037915e-05,
+      "loss": 1.1412,
+      "step": 1018
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.16103962948152212,
+      "learning_rate": 7.627431064785706e-05,
+      "loss": 1.0929,
+      "step": 1019
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.16677548612612422,
+      "learning_rate": 7.606843357124426e-05,
+      "loss": 1.2046,
+      "step": 1020
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.16233738083605964,
+      "learning_rate": 7.58626640356886e-05,
+      "loss": 1.0678,
+      "step": 1021
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.16602043351095722,
+      "learning_rate": 7.565700296585483e-05,
+      "loss": 1.1216,
+      "step": 1022
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.16627151579102536,
+      "learning_rate": 7.54514512859201e-05,
+      "loss": 1.0911,
+      "step": 1023
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.20281510390834787,
+      "learning_rate": 7.524600991957012e-05,
+      "loss": 1.1722,
+      "step": 1024
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.16069933741616732,
+      "learning_rate": 7.504067978999484e-05,
+      "loss": 1.0839,
+      "step": 1025
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.18107514785611215,
+      "learning_rate": 7.483546181988436e-05,
+      "loss": 1.0717,
+      "step": 1026
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.1721242495617235,
+      "learning_rate": 7.463035693142473e-05,
+      "loss": 1.1599,
+      "step": 1027
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.17918262771658913,
+      "learning_rate": 7.442536604629395e-05,
+      "loss": 1.089,
+      "step": 1028
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.17784491718923554,
+      "learning_rate": 7.422049008565757e-05,
+      "loss": 1.176,
+      "step": 1029
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.15630505382249996,
+      "learning_rate": 7.401572997016479e-05,
+      "loss": 1.039,
+      "step": 1030
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.17070239450442767,
+      "learning_rate": 7.381108661994429e-05,
+      "loss": 1.1059,
+      "step": 1031
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.16222624124278331,
+      "learning_rate": 7.360656095459995e-05,
+      "loss": 1.024,
+      "step": 1032
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.17125506848213806,
+      "learning_rate": 7.340215389320687e-05,
+      "loss": 1.1567,
+      "step": 1033
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.17959953833242476,
+      "learning_rate": 7.31978663543071e-05,
+      "loss": 1.141,
+      "step": 1034
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.17010870239393844,
+      "learning_rate": 7.299369925590574e-05,
+      "loss": 1.1046,
+      "step": 1035
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.15998424148316903,
+      "learning_rate": 7.278965351546648e-05,
+      "loss": 1.0627,
+      "step": 1036
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.1710612967988063,
+      "learning_rate": 7.258573004990788e-05,
+      "loss": 1.1486,
+      "step": 1037
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.1607397255850716,
+      "learning_rate": 7.238192977559884e-05,
+      "loss": 1.0751,
+      "step": 1038
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.17530360487610264,
+      "learning_rate": 7.217825360835473e-05,
+      "loss": 1.1281,
+      "step": 1039
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.17374375355320706,
+      "learning_rate": 7.197470246343333e-05,
+      "loss": 1.1493,
+      "step": 1040
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.15987413285182245,
+      "learning_rate": 7.177127725553045e-05,
+      "loss": 1.1446,
+      "step": 1041
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.1789355664165954,
+      "learning_rate": 7.156797889877613e-05,
+      "loss": 1.2304,
+      "step": 1042
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.17282586483463735,
+      "learning_rate": 7.136480830673019e-05,
+      "loss": 1.1909,
+      "step": 1043
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.159213004708279,
+      "learning_rate": 7.116176639237852e-05,
+      "loss": 1.0624,
+      "step": 1044
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.16197929506166975,
+      "learning_rate": 7.095885406812866e-05,
+      "loss": 1.1283,
+      "step": 1045
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.16495179889101508,
+      "learning_rate": 7.075607224580581e-05,
+      "loss": 1.1378,
+      "step": 1046
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.1662731898113741,
+      "learning_rate": 7.05534218366488e-05,
+      "loss": 1.0542,
+      "step": 1047
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.1650210560996418,
+      "learning_rate": 7.035090375130581e-05,
+      "loss": 1.1326,
+      "step": 1048
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.1708049460231556,
+      "learning_rate": 7.014851889983057e-05,
+      "loss": 1.137,
+      "step": 1049
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.17558230008855352,
+      "learning_rate": 6.994626819167789e-05,
+      "loss": 1.1359,
+      "step": 1050
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.16645826974884173,
+      "learning_rate": 6.974415253570003e-05,
+      "loss": 1.1742,
+      "step": 1051
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.17184188151368943,
+      "learning_rate": 6.954217284014211e-05,
+      "loss": 1.1953,
+      "step": 1052
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.17068592296841717,
+      "learning_rate": 6.934033001263847e-05,
+      "loss": 1.1301,
+      "step": 1053
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.16302700472895656,
+      "learning_rate": 6.913862496020831e-05,
+      "loss": 1.1232,
+      "step": 1054
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.17057720351205383,
+      "learning_rate": 6.893705858925178e-05,
+      "loss": 1.1247,
+      "step": 1055
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.1582652133644645,
+      "learning_rate": 6.873563180554583e-05,
+      "loss": 1.1203,
+      "step": 1056
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.16895339487465857,
+      "learning_rate": 6.853434551424e-05,
+      "loss": 1.1846,
+      "step": 1057
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.1726993947504367,
+      "learning_rate": 6.833320061985277e-05,
+      "loss": 1.1963,
+      "step": 1058
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.16055788377798388,
+      "learning_rate": 6.813219802626698e-05,
+      "loss": 1.1439,
+      "step": 1059
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.17265920805658908,
+      "learning_rate": 6.793133863672616e-05,
+      "loss": 1.192,
+      "step": 1060
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.1833601588530699,
+      "learning_rate": 6.773062335383024e-05,
+      "loss": 1.1128,
+      "step": 1061
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.1717674054389487,
+      "learning_rate": 6.753005307953167e-05,
+      "loss": 1.134,
+      "step": 1062
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.18207071739270747,
+      "learning_rate": 6.73296287151312e-05,
+      "loss": 1.1723,
+      "step": 1063
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.1624856029472637,
+      "learning_rate": 6.712935116127389e-05,
+      "loss": 1.1144,
+      "step": 1064
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.19915390833072513,
+      "learning_rate": 6.692922131794517e-05,
+      "loss": 1.1072,
+      "step": 1065
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.1722567762020263,
+      "learning_rate": 6.672924008446662e-05,
+      "loss": 1.1425,
+      "step": 1066
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.15963511981578907,
+      "learning_rate": 6.652940835949208e-05,
+      "loss": 1.1211,
+      "step": 1067
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.16578458195408932,
+      "learning_rate": 6.632972704100349e-05,
+      "loss": 1.1244,
+      "step": 1068
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.17110384656595054,
+      "learning_rate": 6.613019702630694e-05,
+      "loss": 1.1795,
+      "step": 1069
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.16990207952122507,
+      "learning_rate": 6.593081921202859e-05,
+      "loss": 1.0817,
+      "step": 1070
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.16467375370533938,
+      "learning_rate": 6.57315944941107e-05,
+      "loss": 1.0977,
+      "step": 1071
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.17152460282733728,
+      "learning_rate": 6.553252376780748e-05,
+      "loss": 1.1868,
+      "step": 1072
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.18132610186641904,
+      "learning_rate": 6.533360792768122e-05,
+      "loss": 1.2005,
+      "step": 1073
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.16698085910075294,
+      "learning_rate": 6.513484786759818e-05,
+      "loss": 1.1362,
+      "step": 1074
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.16504199012773973,
+      "learning_rate": 6.493624448072457e-05,
+      "loss": 1.1303,
+      "step": 1075
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.1779596734054924,
+      "learning_rate": 6.473779865952263e-05,
+      "loss": 1.1894,
+      "step": 1076
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.16364715074236819,
+      "learning_rate": 6.453951129574644e-05,
+      "loss": 1.1443,
+      "step": 1077
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.16251550362999542,
+      "learning_rate": 6.434138328043815e-05,
+      "loss": 1.1238,
+      "step": 1078
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.1666990603079008,
+      "learning_rate": 6.414341550392368e-05,
+      "loss": 1.0909,
+      "step": 1079
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.15779118156939148,
+      "learning_rate": 6.39456088558091e-05,
+      "loss": 1.1109,
+      "step": 1080
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.1677767995314692,
+      "learning_rate": 6.374796422497621e-05,
+      "loss": 1.2103,
+      "step": 1081
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.17165919802575422,
+      "learning_rate": 6.355048249957886e-05,
+      "loss": 1.1879,
+      "step": 1082
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.1918026223221136,
+      "learning_rate": 6.33531645670389e-05,
+      "loss": 1.1733,
+      "step": 1083
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.1787527260843118,
+      "learning_rate": 6.3156011314042e-05,
+      "loss": 1.1338,
+      "step": 1084
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.1694260433762424,
+      "learning_rate": 6.2959023626534e-05,
+      "loss": 1.1537,
+      "step": 1085
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.18311039386845188,
+      "learning_rate": 6.276220238971652e-05,
+      "loss": 1.1976,
+      "step": 1086
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.17113962947413622,
+      "learning_rate": 6.256554848804343e-05,
+      "loss": 1.0813,
+      "step": 1087
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.1664054538940142,
+      "learning_rate": 6.236906280521646e-05,
+      "loss": 1.1514,
+      "step": 1088
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.16227842164425618,
+      "learning_rate": 6.217274622418153e-05,
+      "loss": 1.0597,
+      "step": 1089
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.17755371081391708,
+      "learning_rate": 6.197659962712461e-05,
+      "loss": 1.19,
+      "step": 1090
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.15844523213403552,
+      "learning_rate": 6.178062389546784e-05,
+      "loss": 1.0587,
+      "step": 1091
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.16903107223771263,
+      "learning_rate": 6.158481990986557e-05,
+      "loss": 1.1339,
+      "step": 1092
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.17002303254909312,
+      "learning_rate": 6.138918855020028e-05,
+      "loss": 1.1158,
+      "step": 1093
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.16613843815618992,
+      "learning_rate": 6.11937306955789e-05,
+      "loss": 1.1894,
+      "step": 1094
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.17111783959067672,
+      "learning_rate": 6.099844722432843e-05,
+      "loss": 1.1974,
+      "step": 1095
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.1670020953276936,
+      "learning_rate": 6.080333901399251e-05,
+      "loss": 1.1018,
+      "step": 1096
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.18080514034863332,
+      "learning_rate": 6.060840694132701e-05,
+      "loss": 1.1845,
+      "step": 1097
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.18525347317947505,
+      "learning_rate": 6.0413651882296406e-05,
+      "loss": 1.0796,
+      "step": 1098
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.20400403778138837,
+      "learning_rate": 6.021907471206971e-05,
+      "loss": 1.0117,
+      "step": 1099
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.1641784925803925,
+      "learning_rate": 6.002467630501646e-05,
+      "loss": 1.097,
+      "step": 1100
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 0.16623722195850973,
+      "learning_rate": 5.983045753470308e-05,
+      "loss": 1.0141,
+      "step": 1101
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 0.1810578331051459,
+      "learning_rate": 5.9636419273888546e-05,
+      "loss": 1.0277,
+      "step": 1102
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 0.18889563641964546,
+      "learning_rate": 5.944256239452085e-05,
+      "loss": 0.9148,
+      "step": 1103
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 0.18087970961891994,
+      "learning_rate": 5.924888776773281e-05,
+      "loss": 1.0529,
+      "step": 1104
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 0.1733509400885039,
+      "learning_rate": 5.9055396263838315e-05,
+      "loss": 0.9925,
+      "step": 1105
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 0.17285204625514128,
+      "learning_rate": 5.886208875232833e-05,
+      "loss": 1.0022,
+      "step": 1106
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 0.17977965852266828,
+      "learning_rate": 5.8668966101867005e-05,
+      "loss": 1.0279,
+      "step": 1107
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 0.1805768384069582,
+      "learning_rate": 5.847602918028785e-05,
+      "loss": 0.9688,
+      "step": 1108
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 0.19307882052881373,
+      "learning_rate": 5.82832788545896e-05,
+      "loss": 1.0361,
+      "step": 1109
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 0.18038343158251127,
+      "learning_rate": 5.809071599093272e-05,
+      "loss": 0.9618,
+      "step": 1110
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 0.18081635249066572,
+      "learning_rate": 5.789834145463506e-05,
+      "loss": 0.9634,
+      "step": 1111
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 0.2132256408954024,
+      "learning_rate": 5.7706156110168384e-05,
+      "loss": 0.9917,
+      "step": 1112
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 0.20894695338897498,
+      "learning_rate": 5.751416082115408e-05,
+      "loss": 1.0608,
+      "step": 1113
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 0.18561065850228853,
+      "learning_rate": 5.732235645035964e-05,
+      "loss": 1.0319,
+      "step": 1114
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 0.19972592759433763,
+      "learning_rate": 5.713074385969457e-05,
+      "loss": 1.068,
+      "step": 1115
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 0.18875304906041382,
+      "learning_rate": 5.6939323910206645e-05,
+      "loss": 1.0706,
+      "step": 1116
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 0.20051381946623245,
+      "learning_rate": 5.6748097462077796e-05,
+      "loss": 1.0879,
+      "step": 1117
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 0.18028693825056266,
+      "learning_rate": 5.65570653746206e-05,
+      "loss": 0.9736,
+      "step": 1118
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 0.18371629750739465,
+      "learning_rate": 5.63662285062742e-05,
+      "loss": 1.0459,
+      "step": 1119
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 0.18620839376911252,
+      "learning_rate": 5.6175587714600385e-05,
+      "loss": 1.0564,
+      "step": 1120
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 0.1761497636850895,
+      "learning_rate": 5.598514385627996e-05,
+      "loss": 0.9928,
+      "step": 1121
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 0.18717899836165747,
+      "learning_rate": 5.579489778710867e-05,
+      "loss": 0.9903,
+      "step": 1122
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 0.1792885931651523,
+      "learning_rate": 5.56048503619935e-05,
+      "loss": 0.9889,
+      "step": 1123
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 0.18624892968399204,
+      "learning_rate": 5.541500243494888e-05,
+      "loss": 1.0291,
+      "step": 1124
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 0.17837324324146603,
+      "learning_rate": 5.522535485909257e-05,
+      "loss": 1.0182,
+      "step": 1125
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 0.18979627533160237,
+      "learning_rate": 5.5035908486642195e-05,
+      "loss": 0.9828,
+      "step": 1126
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 0.19273836099434338,
+      "learning_rate": 5.484666416891109e-05,
+      "loss": 1.0496,
+      "step": 1127
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 0.19543417583922992,
+      "learning_rate": 5.4657622756304704e-05,
+      "loss": 0.9901,
+      "step": 1128
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 0.19508414506527536,
+      "learning_rate": 5.4468785098316674e-05,
+      "loss": 1.0326,
+      "step": 1129
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 0.185374829651796,
+      "learning_rate": 5.428015204352508e-05,
+      "loss": 0.9779,
+      "step": 1130
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 0.19699548343809967,
+      "learning_rate": 5.409172443958843e-05,
+      "loss": 1.0246,
+      "step": 1131
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 0.20300075118398447,
+      "learning_rate": 5.3903503133242136e-05,
+      "loss": 0.969,
+      "step": 1132
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 0.19110544018613584,
+      "learning_rate": 5.371548897029457e-05,
+      "loss": 1.0747,
+      "step": 1133
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 0.18550194883382987,
+      "learning_rate": 5.3527682795623146e-05,
+      "loss": 1.0157,
+      "step": 1134
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 0.18331490400990383,
+      "learning_rate": 5.334008545317082e-05,
+      "loss": 0.9947,
+      "step": 1135
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 0.18751165571321418,
+      "learning_rate": 5.31526977859419e-05,
+      "loss": 1.0502,
+      "step": 1136
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 0.19211604448762654,
+      "learning_rate": 5.296552063599868e-05,
+      "loss": 1.0145,
+      "step": 1137
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 0.186297345787599,
+      "learning_rate": 5.277855484445735e-05,
+      "loss": 1.0537,
+      "step": 1138
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 0.18982650964521386,
+      "learning_rate": 5.259180125148442e-05,
+      "loss": 0.9838,
+      "step": 1139
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 0.19925867583848977,
+      "learning_rate": 5.240526069629265e-05,
+      "loss": 1.0231,
+      "step": 1140
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 0.1884322287285855,
+      "learning_rate": 5.22189340171377e-05,
+      "loss": 0.9296,
+      "step": 1141
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 0.1796241132232387,
+      "learning_rate": 5.203282205131395e-05,
+      "loss": 0.9681,
+      "step": 1142
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 0.1839238478548194,
+      "learning_rate": 5.1846925635151045e-05,
+      "loss": 0.9101,
+      "step": 1143
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 0.19345190226496187,
+      "learning_rate": 5.166124560401002e-05,
+      "loss": 0.9651,
+      "step": 1144
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 0.18755030741172238,
+      "learning_rate": 5.1475782792279426e-05,
+      "loss": 1.0174,
+      "step": 1145
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 0.18631640479941738,
+      "learning_rate": 5.129053803337181e-05,
+      "loss": 1.035,
+      "step": 1146
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 0.1975877349106955,
+      "learning_rate": 5.1105512159719805e-05,
+      "loss": 0.9989,
+      "step": 1147
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 0.18636484306751205,
+      "learning_rate": 5.092070600277252e-05,
+      "loss": 1.0426,
+      "step": 1148
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 0.18611358587003984,
+      "learning_rate": 5.073612039299157e-05,
+      "loss": 0.9949,
+      "step": 1149
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 0.1910181776639786,
+      "learning_rate": 5.0551756159847666e-05,
+      "loss": 0.9716,
+      "step": 1150
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 0.1896157384615499,
+      "learning_rate": 5.036761413181659e-05,
+      "loss": 1.0205,
+      "step": 1151
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 0.18470372913964078,
+      "learning_rate": 5.0183695136375664e-05,
+      "loss": 0.9938,
+      "step": 1152
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 0.19182549114492065,
+      "learning_rate": 5.000000000000002e-05,
+      "loss": 1.0717,
+      "step": 1153
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 0.20299654057754923,
+      "learning_rate": 4.9816529548158706e-05,
+      "loss": 1.0932,
+      "step": 1154
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 0.18753899816476238,
+      "learning_rate": 4.963328460531127e-05,
+      "loss": 1.0785,
+      "step": 1155
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 0.1912398459667478,
+      "learning_rate": 4.9450265994903664e-05,
+      "loss": 1.0354,
+      "step": 1156
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 0.1970798090720647,
+      "learning_rate": 4.9267474539365086e-05,
+      "loss": 1.0394,
+      "step": 1157
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 0.19438353474466769,
+      "learning_rate": 4.908491106010368e-05,
+      "loss": 1.0563,
+      "step": 1158
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 0.18583224152081138,
+      "learning_rate": 4.8902576377503316e-05,
+      "loss": 1.0375,
+      "step": 1159
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 0.21067204344598053,
+      "learning_rate": 4.87204713109196e-05,
+      "loss": 1.0275,
+      "step": 1160
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 0.19271054378491778,
+      "learning_rate": 4.8538596678676406e-05,
+      "loss": 1.0241,
+      "step": 1161
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 0.20698647247820803,
+      "learning_rate": 4.8356953298062115e-05,
+      "loss": 1.0183,
+      "step": 1162
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 0.19147199804367737,
+      "learning_rate": 4.817554198532581e-05,
+      "loss": 1.0464,
+      "step": 1163
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 0.18948541542489386,
+      "learning_rate": 4.79943635556739e-05,
+      "loss": 0.942,
+      "step": 1164
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 0.20063322256200955,
+      "learning_rate": 4.7813418823266146e-05,
+      "loss": 1.0207,
+      "step": 1165
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 0.20277313248067322,
+      "learning_rate": 4.763270860121222e-05,
+      "loss": 1.1196,
+      "step": 1166
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 0.18916247962333935,
+      "learning_rate": 4.745223370156797e-05,
+      "loss": 1.0635,
+      "step": 1167
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 0.20770456045992683,
+      "learning_rate": 4.727199493533181e-05,
+      "loss": 1.0641,
+      "step": 1168
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 0.1928796702717179,
+      "learning_rate": 4.709199311244098e-05,
+      "loss": 1.0173,
+      "step": 1169
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 0.19067624131226646,
+      "learning_rate": 4.691222904176791e-05,
+      "loss": 1.0739,
+      "step": 1170
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 0.18794968941934675,
+      "learning_rate": 4.6732703531116874e-05,
+      "loss": 0.9853,
+      "step": 1171
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 0.18128537252239216,
+      "learning_rate": 4.6553417387219886e-05,
+      "loss": 0.9791,
+      "step": 1172
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 0.19969579280811037,
+      "learning_rate": 4.6374371415733496e-05,
+      "loss": 1.0328,
+      "step": 1173
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 0.17771947900276572,
+      "learning_rate": 4.619556642123484e-05,
+      "loss": 0.9977,
+      "step": 1174
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 0.18877925953999256,
+      "learning_rate": 4.601700320721829e-05,
+      "loss": 1.0163,
+      "step": 1175
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 0.1976777570336078,
+      "learning_rate": 4.583868257609171e-05,
+      "loss": 0.9831,
+      "step": 1176
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 0.1926309732820373,
+      "learning_rate": 4.566060532917288e-05,
+      "loss": 1.0369,
+      "step": 1177
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 0.1981154480792855,
+      "learning_rate": 4.5482772266685844e-05,
+      "loss": 1.0265,
+      "step": 1178
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 0.18961229347269684,
+      "learning_rate": 4.530518418775733e-05,
+      "loss": 0.978,
+      "step": 1179
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 0.1902416426203216,
+      "learning_rate": 4.512784189041328e-05,
+      "loss": 1.0476,
+      "step": 1180
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 0.1963783030038318,
+      "learning_rate": 4.495074617157513e-05,
+      "loss": 1.1034,
+      "step": 1181
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 0.1976594259307984,
+      "learning_rate": 4.477389782705628e-05,
+      "loss": 1.0411,
+      "step": 1182
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 0.19599497931317425,
+      "learning_rate": 4.459729765155842e-05,
+      "loss": 0.9831,
+      "step": 1183
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 0.18768317847907803,
+      "learning_rate": 4.4420946438668164e-05,
+      "loss": 0.9667,
+      "step": 1184
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 0.1893955423774859,
+      "learning_rate": 4.424484498085335e-05,
+      "loss": 1.0176,
+      "step": 1185
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 0.185214177313968,
+      "learning_rate": 4.4068994069459376e-05,
+      "loss": 0.9672,
+      "step": 1186
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 0.2001674222762866,
+      "learning_rate": 4.389339449470592e-05,
+      "loss": 1.0872,
+      "step": 1187
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 0.18927674242494813,
+      "learning_rate": 4.371804704568309e-05,
+      "loss": 1.0305,
+      "step": 1188
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 0.189797567081393,
+      "learning_rate": 4.354295251034811e-05,
+      "loss": 0.9866,
+      "step": 1189
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 0.19621874043278917,
+      "learning_rate": 4.336811167552167e-05,
+      "loss": 0.9989,
+      "step": 1190
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 0.1961781136176054,
+      "learning_rate": 4.3193525326884435e-05,
+      "loss": 1.1191,
+      "step": 1191
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 0.19610651673820242,
+      "learning_rate": 4.301919424897338e-05,
+      "loss": 1.0448,
+      "step": 1192
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 0.19393858136213526,
+      "learning_rate": 4.284511922517853e-05,
+      "loss": 1.021,
+      "step": 1193
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 0.19611938956585503,
+      "learning_rate": 4.267130103773911e-05,
+      "loss": 1.003,
+      "step": 1194
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 0.19521183418608554,
+      "learning_rate": 4.249774046774034e-05,
+      "loss": 0.9956,
+      "step": 1195
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 0.1903820141620171,
+      "learning_rate": 4.232443829510977e-05,
+      "loss": 0.988,
+      "step": 1196
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 0.19917471096571515,
+      "learning_rate": 4.215139529861367e-05,
+      "loss": 1.0294,
+      "step": 1197
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 0.1989120051737976,
+      "learning_rate": 4.19786122558538e-05,
+      "loss": 0.9865,
+      "step": 1198
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 0.1878933904328363,
+      "learning_rate": 4.1806089943263706e-05,
+      "loss": 0.9914,
+      "step": 1199
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 0.1889686688150144,
+      "learning_rate": 4.163382913610533e-05,
+      "loss": 0.9183,
+      "step": 1200
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 0.19599798287453252,
+      "learning_rate": 4.146183060846538e-05,
+      "loss": 0.9809,
+      "step": 1201
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 0.19159410980792513,
+      "learning_rate": 4.129009513325212e-05,
+      "loss": 1.0689,
+      "step": 1202
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 0.1916180919600845,
+      "learning_rate": 4.111862348219158e-05,
+      "loss": 1.0453,
+      "step": 1203
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 0.18629810473389968,
+      "learning_rate": 4.094741642582434e-05,
+      "loss": 0.9991,
+      "step": 1204
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 0.18983910244997687,
+      "learning_rate": 4.077647473350201e-05,
+      "loss": 1.0355,
+      "step": 1205
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 0.19604918383133388,
+      "learning_rate": 4.060579917338362e-05,
+      "loss": 0.9918,
+      "step": 1206
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 0.19914658125329224,
+      "learning_rate": 4.043539051243239e-05,
+      "loss": 1.0083,
+      "step": 1207
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 0.19984032461233703,
+      "learning_rate": 4.026524951641204e-05,
+      "loss": 1.0402,
+      "step": 1208
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 0.19713889646881289,
+      "learning_rate": 4.009537694988372e-05,
+      "loss": 1.0814,
+      "step": 1209
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 0.18984773417984263,
+      "learning_rate": 3.99257735762021e-05,
+      "loss": 0.9727,
+      "step": 1210
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 0.1969234745130071,
+      "learning_rate": 3.975644015751234e-05,
+      "loss": 1.1246,
+      "step": 1211
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 0.20095976880290764,
+      "learning_rate": 3.958737745474638e-05,
+      "loss": 1.0264,
+      "step": 1212
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 0.19573419897255526,
+      "learning_rate": 3.9418586227619746e-05,
+      "loss": 0.9699,
+      "step": 1213
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 0.20794714498925052,
+      "learning_rate": 3.9250067234628054e-05,
+      "loss": 1.0871,
+      "step": 1214
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 0.200309362571014,
+      "learning_rate": 3.9081821233043436e-05,
+      "loss": 1.0869,
+      "step": 1215
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 0.21221715688656814,
+      "learning_rate": 3.891384897891148e-05,
+      "loss": 1.0829,
+      "step": 1216
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 0.1895941245282512,
+      "learning_rate": 3.874615122704746e-05,
+      "loss": 0.9934,
+      "step": 1217
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 0.20282251657546532,
+      "learning_rate": 3.857872873103322e-05,
+      "loss": 1.0394,
+      "step": 1218
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 0.20092601921573158,
+      "learning_rate": 3.8411582243213694e-05,
+      "loss": 1.0626,
+      "step": 1219
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 0.1967616729990936,
+      "learning_rate": 3.824471251469353e-05,
+      "loss": 1.1027,
+      "step": 1220
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 0.1974168096006322,
+      "learning_rate": 3.807812029533362e-05,
+      "loss": 1.0285,
+      "step": 1221
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 0.2005755550749436,
+      "learning_rate": 3.791180633374785e-05,
+      "loss": 1.0579,
+      "step": 1222
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 0.22982259456971435,
+      "learning_rate": 3.7745771377299755e-05,
+      "loss": 0.9251,
+      "step": 1223
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 0.19486627427023995,
+      "learning_rate": 3.758001617209906e-05,
+      "loss": 0.9991,
+      "step": 1224
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 0.19268654100626367,
+      "learning_rate": 3.7414541462998446e-05,
+      "loss": 1.0843,
+      "step": 1225
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 0.19142859148030442,
+      "learning_rate": 3.724934799358998e-05,
+      "loss": 0.9734,
+      "step": 1226
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 0.2078245745909079,
+      "learning_rate": 3.708443650620206e-05,
+      "loss": 1.079,
+      "step": 1227
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 0.18957789857941185,
+      "learning_rate": 3.691980774189591e-05,
+      "loss": 0.9865,
+      "step": 1228
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 0.2022379994557547,
+      "learning_rate": 3.675546244046228e-05,
+      "loss": 1.0147,
+      "step": 1229
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 0.20484113736244272,
+      "learning_rate": 3.6591401340418116e-05,
+      "loss": 1.0491,
+      "step": 1230
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 0.19932572874171609,
+      "learning_rate": 3.642762517900322e-05,
+      "loss": 1.0187,
+      "step": 1231
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 0.19513525838609894,
+      "learning_rate": 3.626413469217702e-05,
+      "loss": 1.0346,
+      "step": 1232
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 0.18447148945806335,
+      "learning_rate": 3.6100930614615205e-05,
+      "loss": 0.9713,
+      "step": 1233
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 0.19615772374760426,
+      "learning_rate": 3.593801367970645e-05,
+      "loss": 1.0279,
+      "step": 1234
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 0.195942402850121,
+      "learning_rate": 3.5775384619549e-05,
+      "loss": 1.044,
+      "step": 1235
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 0.20033402806459327,
+      "learning_rate": 3.561304416494762e-05,
+      "loss": 0.9634,
+      "step": 1236
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 0.20133608619277885,
+      "learning_rate": 3.5450993045409996e-05,
+      "loss": 0.9424,
+      "step": 1237
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 0.1976979767858137,
+      "learning_rate": 3.5289231989143865e-05,
+      "loss": 1.0329,
+      "step": 1238
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 0.1866564132240131,
+      "learning_rate": 3.512776172305331e-05,
+      "loss": 0.9951,
+      "step": 1239
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 0.20643351098366144,
+      "learning_rate": 3.496658297273574e-05,
+      "loss": 1.0768,
+      "step": 1240
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 0.2016711813980471,
+      "learning_rate": 3.4805696462478634e-05,
+      "loss": 1.0351,
+      "step": 1241
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 0.19967214131051508,
+      "learning_rate": 3.46451029152562e-05,
+      "loss": 1.0102,
+      "step": 1242
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 0.1983504742901789,
+      "learning_rate": 3.448480305272619e-05,
+      "loss": 1.0268,
+      "step": 1243
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 0.19682722769626732,
+      "learning_rate": 3.4324797595226565e-05,
+      "loss": 1.0053,
+      "step": 1244
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 0.20088167609003055,
+      "learning_rate": 3.41650872617724e-05,
+      "loss": 0.9733,
+      "step": 1245
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 0.20335900600853737,
+      "learning_rate": 3.400567277005247e-05,
+      "loss": 0.9861,
+      "step": 1246
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 0.20886916546979215,
+      "learning_rate": 3.3846554836426234e-05,
+      "loss": 1.0473,
+      "step": 1247
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 0.2010663342671541,
+      "learning_rate": 3.36877341759205e-05,
+      "loss": 1.0358,
+      "step": 1248
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 0.21062432438070422,
+      "learning_rate": 3.3529211502226123e-05,
+      "loss": 1.0166,
+      "step": 1249
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 0.2208916600042823,
+      "learning_rate": 3.337098752769503e-05,
+      "loss": 1.1117,
+      "step": 1250
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 0.18314118494949805,
+      "learning_rate": 3.321306296333673e-05,
+      "loss": 0.9577,
+      "step": 1251
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 0.2165551432625507,
+      "learning_rate": 3.3055438518815486e-05,
+      "loss": 1.0629,
+      "step": 1252
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 0.1919444697411042,
+      "learning_rate": 3.289811490244671e-05,
+      "loss": 1.0157,
+      "step": 1253
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 0.19605731642039462,
+      "learning_rate": 3.274109282119413e-05,
+      "loss": 1.0183,
+      "step": 1254
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 0.1978411361540464,
+      "learning_rate": 3.258437298066634e-05,
+      "loss": 1.0333,
+      "step": 1255
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 0.1954273247181233,
+      "learning_rate": 3.242795608511388e-05,
+      "loss": 1.0204,
+      "step": 1256
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 0.19539709125757848,
+      "learning_rate": 3.227184283742591e-05,
+      "loss": 1.0393,
+      "step": 1257
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 0.19509958919620607,
+      "learning_rate": 3.2116033939127024e-05,
+      "loss": 1.0703,
+      "step": 1258
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 0.19145181109714973,
+      "learning_rate": 3.196053009037428e-05,
+      "loss": 0.963,
+      "step": 1259
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 0.19905167727180642,
+      "learning_rate": 3.180533198995379e-05,
+      "loss": 1.109,
+      "step": 1260
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 0.20242244581363775,
+      "learning_rate": 3.165044033527789e-05,
+      "loss": 1.0477,
+      "step": 1261
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 0.19788917203746156,
+      "learning_rate": 3.1495855822381715e-05,
+      "loss": 1.0331,
+      "step": 1262
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 0.20117465106402488,
+      "learning_rate": 3.134157914592032e-05,
+      "loss": 1.0634,
+      "step": 1263
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 0.20012651544737803,
+      "learning_rate": 3.1187610999165316e-05,
+      "loss": 1.0525,
+      "step": 1264
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 0.2043849789970364,
+      "learning_rate": 3.1033952074001884e-05,
+      "loss": 1.0009,
+      "step": 1265
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 0.20436413566516082,
+      "learning_rate": 3.088060306092582e-05,
+      "loss": 1.0194,
+      "step": 1266
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 0.2471012507273153,
+      "learning_rate": 3.072756464904006e-05,
+      "loss": 1.0174,
+      "step": 1267
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 0.19900543237914783,
+      "learning_rate": 3.057483752605196e-05,
+      "loss": 1.1252,
+      "step": 1268
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 0.21526226315725566,
+      "learning_rate": 3.042242237826991e-05,
+      "loss": 1.0376,
+      "step": 1269
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 0.20806176621387384,
+      "learning_rate": 3.0270319890600462e-05,
+      "loss": 1.0543,
+      "step": 1270
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 0.2067694678280775,
+      "learning_rate": 3.0118530746545148e-05,
+      "loss": 1.0625,
+      "step": 1271
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 0.1984329234985004,
+      "learning_rate": 2.9967055628197472e-05,
+      "loss": 1.0445,
+      "step": 1272
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 0.2051174512399526,
+      "learning_rate": 2.9815895216239732e-05,
+      "loss": 1.0023,
+      "step": 1273
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 0.19190337658134426,
+      "learning_rate": 2.9665050189940015e-05,
+      "loss": 1.0081,
+      "step": 1274
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 0.19812292614431837,
+      "learning_rate": 2.951452122714926e-05,
+      "loss": 1.0347,
+      "step": 1275
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 0.20217031978231875,
+      "learning_rate": 2.9364309004298053e-05,
+      "loss": 1.0683,
+      "step": 1276
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 0.2002443376802822,
+      "learning_rate": 2.9214414196393704e-05,
+      "loss": 1.0534,
+      "step": 1277
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 0.19168165480576083,
+      "learning_rate": 2.9064837477017048e-05,
+      "loss": 1.0077,
+      "step": 1278
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 0.19459332024136894,
+      "learning_rate": 2.8915579518319624e-05,
+      "loss": 1.041,
+      "step": 1279
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 0.20329713707478517,
+      "learning_rate": 2.876664099102053e-05,
+      "loss": 1.0385,
+      "step": 1280
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 0.1969232285759602,
+      "learning_rate": 2.861802256440348e-05,
+      "loss": 1.018,
+      "step": 1281
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 0.21865970166706636,
+      "learning_rate": 2.8469724906313678e-05,
+      "loss": 1.0325,
+      "step": 1282
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 0.2045555091720532,
+      "learning_rate": 2.8321748683154893e-05,
+      "loss": 1.0077,
+      "step": 1283
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 0.19858767150496162,
+      "learning_rate": 2.8174094559886534e-05,
+      "loss": 1.0477,
+      "step": 1284
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 0.1979848674300343,
+      "learning_rate": 2.8026763200020555e-05,
+      "loss": 1.0647,
+      "step": 1285
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 0.18999100747223083,
+      "learning_rate": 2.7879755265618555e-05,
+      "loss": 0.9991,
+      "step": 1286
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 0.19318555274581345,
+      "learning_rate": 2.773307141728867e-05,
+      "loss": 1.0689,
+      "step": 1287
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 0.2001048343990683,
+      "learning_rate": 2.7586712314182773e-05,
+      "loss": 0.9894,
+      "step": 1288
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 0.1941010794439273,
+      "learning_rate": 2.7440678613993332e-05,
+      "loss": 1.0117,
+      "step": 1289
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 0.20543809678008965,
+      "learning_rate": 2.729497097295075e-05,
+      "loss": 0.9712,
+      "step": 1290
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 0.19570648696674608,
+      "learning_rate": 2.7149590045820028e-05,
+      "loss": 1.0332,
+      "step": 1291
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 0.20214826838793598,
+      "learning_rate": 2.7004536485898045e-05,
+      "loss": 1.024,
+      "step": 1292
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 0.2115211012217786,
+      "learning_rate": 2.685981094501069e-05,
+      "loss": 0.9931,
+      "step": 1293
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 0.21220137438280212,
+      "learning_rate": 2.6715414073509748e-05,
+      "loss": 1.0435,
+      "step": 1294
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 0.20609154009368855,
+      "learning_rate": 2.6571346520270147e-05,
+      "loss": 1.1253,
+      "step": 1295
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 0.1923052032802701,
+      "learning_rate": 2.6427608932686843e-05,
+      "loss": 0.9451,
+      "step": 1296
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 0.19730527167467898,
+      "learning_rate": 2.6284201956672137e-05,
+      "loss": 1.0458,
+      "step": 1297
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 0.19989767746964465,
+      "learning_rate": 2.614112623665259e-05,
+      "loss": 0.9961,
+      "step": 1298
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 0.20743521472072698,
+      "learning_rate": 2.599838241556626e-05,
+      "loss": 0.986,
+      "step": 1299
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 0.211185028570926,
+      "learning_rate": 2.5855971134859736e-05,
+      "loss": 1.0033,
+      "step": 1300
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 0.18980333720726505,
+      "learning_rate": 2.5713893034485214e-05,
+      "loss": 1.0542,
+      "step": 1301
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 0.19527708839800334,
+      "learning_rate": 2.5572148752897795e-05,
+      "loss": 0.9458,
+      "step": 1302
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 0.19737801451718315,
+      "learning_rate": 2.5430738927052344e-05,
+      "loss": 1.0229,
+      "step": 1303
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 0.19944968925099527,
+      "learning_rate": 2.5289664192400997e-05,
+      "loss": 1.0187,
+      "step": 1304
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 0.19850167089007587,
+      "learning_rate": 2.514892518288988e-05,
+      "loss": 1.0301,
+      "step": 1305
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 0.20869345185658936,
+      "learning_rate": 2.5008522530956646e-05,
+      "loss": 1.0882,
+      "step": 1306
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 0.192429822864874,
+      "learning_rate": 2.4868456867527313e-05,
+      "loss": 1.0768,
+      "step": 1307
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 0.20041667374784053,
+      "learning_rate": 2.472872882201368e-05,
+      "loss": 0.9564,
+      "step": 1308
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 0.19278196420916055,
+      "learning_rate": 2.4589339022310386e-05,
+      "loss": 1.0012,
+      "step": 1309
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 0.20632547530350015,
+      "learning_rate": 2.445028809479203e-05,
+      "loss": 1.0691,
+      "step": 1310
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 0.20026721503616646,
+      "learning_rate": 2.431157666431052e-05,
+      "loss": 1.0614,
+      "step": 1311
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 0.19489852973676156,
+      "learning_rate": 2.4173205354192063e-05,
+      "loss": 0.9797,
+      "step": 1312
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 0.19651304043270207,
+      "learning_rate": 2.403517478623456e-05,
+      "loss": 1.0636,
+      "step": 1313
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 0.20753886611474595,
+      "learning_rate": 2.3897485580704682e-05,
+      "loss": 1.0488,
+      "step": 1314
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 0.1902807052625808,
+      "learning_rate": 2.376013835633517e-05,
+      "loss": 1.0122,
+      "step": 1315
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 0.2128547880840383,
+      "learning_rate": 2.3623133730321922e-05,
+      "loss": 1.0662,
+      "step": 1316
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 0.19178946945027847,
+      "learning_rate": 2.3486472318321307e-05,
+      "loss": 0.999,
+      "step": 1317
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 0.19790833247794626,
+      "learning_rate": 2.3350154734447537e-05,
+      "loss": 1.0543,
+      "step": 1318
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 0.19615861400513737,
+      "learning_rate": 2.32141815912696e-05,
+      "loss": 1.0323,
+      "step": 1319
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 0.19111434134530353,
+      "learning_rate": 2.3078553499808797e-05,
+      "loss": 1.0206,
+      "step": 1320
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 0.1967812553288128,
+      "learning_rate": 2.2943271069535755e-05,
+      "loss": 1.0213,
+      "step": 1321
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 0.19874170003799302,
+      "learning_rate": 2.2808334908367914e-05,
+      "loss": 0.9631,
+      "step": 1322
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 0.23014562765430222,
+      "learning_rate": 2.267374562266662e-05,
+      "loss": 1.0333,
+      "step": 1323
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 0.21310855355754507,
+      "learning_rate": 2.2539503817234553e-05,
+      "loss": 1.0808,
+      "step": 1324
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 0.19972397701820568,
+      "learning_rate": 2.240561009531281e-05,
+      "loss": 1.0527,
+      "step": 1325
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 0.1927849389481322,
+      "learning_rate": 2.227206505857834e-05,
+      "loss": 1.0776,
+      "step": 1326
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 0.19740116004042996,
+      "learning_rate": 2.2138869307141263e-05,
+      "loss": 1.014,
+      "step": 1327
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 0.20289387092281236,
+      "learning_rate": 2.2006023439542088e-05,
+      "loss": 1.0103,
+      "step": 1328
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 0.20394382008794595,
+      "learning_rate": 2.1873528052749092e-05,
+      "loss": 0.9351,
+      "step": 1329
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 0.20548016222534313,
+      "learning_rate": 2.1741383742155474e-05,
+      "loss": 1.0066,
+      "step": 1330
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 0.19563695871816103,
+      "learning_rate": 2.1609591101576942e-05,
+      "loss": 1.0477,
+      "step": 1331
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 0.21242314187055802,
+      "learning_rate": 2.1478150723248857e-05,
+      "loss": 0.9558,
+      "step": 1332
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 0.21115961226521227,
+      "learning_rate": 2.1347063197823647e-05,
+      "loss": 1.013,
+      "step": 1333
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 0.1994263855293853,
+      "learning_rate": 2.1216329114368084e-05,
+      "loss": 1.0287,
+      "step": 1334
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 0.20246037782306023,
+      "learning_rate": 2.1085949060360654e-05,
+      "loss": 1.0011,
+      "step": 1335
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 0.19513564076453144,
+      "learning_rate": 2.0955923621689034e-05,
+      "loss": 0.9725,
+      "step": 1336
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 0.22329723996513792,
+      "learning_rate": 2.0826253382647333e-05,
+      "loss": 1.1098,
+      "step": 1337
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 0.20377464973870935,
+      "learning_rate": 2.0696938925933506e-05,
+      "loss": 1.0675,
+      "step": 1338
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 0.19102737130076528,
+      "learning_rate": 2.056798083264667e-05,
+      "loss": 0.9492,
+      "step": 1339
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 0.2073014798883011,
+      "learning_rate": 2.043937968228469e-05,
+      "loss": 1.0387,
+      "step": 1340
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 0.24290986242591792,
+      "learning_rate": 2.0311136052741277e-05,
+      "loss": 1.0558,
+      "step": 1341
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 0.2085266713259981,
+      "learning_rate": 2.0183250520303652e-05,
+      "loss": 1.0638,
+      "step": 1342
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 0.1955217978555781,
+      "learning_rate": 2.0055723659649904e-05,
+      "loss": 1.0096,
+      "step": 1343
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 0.19927692668612149,
+      "learning_rate": 1.9928556043846214e-05,
+      "loss": 1.0701,
+      "step": 1344
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 0.19744727496920214,
+      "learning_rate": 1.9801748244344585e-05,
+      "loss": 1.0438,
+      "step": 1345
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 0.2055358202234184,
+      "learning_rate": 1.967530083097996e-05,
+      "loss": 1.0575,
+      "step": 1346
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 0.2022241159482209,
+      "learning_rate": 1.9549214371968004e-05,
+      "loss": 1.0226,
+      "step": 1347
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 0.19649509124937034,
+      "learning_rate": 1.9423489433902186e-05,
+      "loss": 1.0202,
+      "step": 1348
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 0.20600851551211816,
+      "learning_rate": 1.929812658175154e-05,
+      "loss": 1.021,
+      "step": 1349
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 0.19294116399601477,
+      "learning_rate": 1.9173126378857907e-05,
+      "loss": 1.0213,
+      "step": 1350
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 0.1938777663276966,
+      "learning_rate": 1.9048489386933543e-05,
+      "loss": 0.9898,
+      "step": 1351
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 0.19185115735103692,
+      "learning_rate": 1.892421616605857e-05,
+      "loss": 0.9944,
+      "step": 1352
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 0.19670853913434655,
+      "learning_rate": 1.8800307274678364e-05,
+      "loss": 0.9919,
+      "step": 1353
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 0.20990234935789304,
+      "learning_rate": 1.8676763269601194e-05,
+      "loss": 0.9881,
+      "step": 1354
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 0.19572956406722788,
+      "learning_rate": 1.8553584705995562e-05,
+      "loss": 1.0294,
+      "step": 1355
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 0.19113731201482373,
+      "learning_rate": 1.8430772137387853e-05,
+      "loss": 1.0309,
+      "step": 1356
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 0.20326050779588162,
+      "learning_rate": 1.8308326115659757e-05,
+      "loss": 1.0583,
+      "step": 1357
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 0.20127668715102875,
+      "learning_rate": 1.8186247191045856e-05,
+      "loss": 1.0456,
+      "step": 1358
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 0.20570924872407959,
+      "learning_rate": 1.806453591213103e-05,
+      "loss": 1.0397,
+      "step": 1359
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 0.19884839319020708,
+      "learning_rate": 1.794319282584813e-05,
+      "loss": 1.008,
+      "step": 1360
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 0.22176454676380103,
+      "learning_rate": 1.7822218477475494e-05,
+      "loss": 0.9774,
+      "step": 1361
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 0.20018826586404945,
+      "learning_rate": 1.7701613410634365e-05,
+      "loss": 1.0615,
+      "step": 1362
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 0.20389686562227474,
+      "learning_rate": 1.7581378167286656e-05,
+      "loss": 1.043,
+      "step": 1363
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 0.19726813509954272,
+      "learning_rate": 1.7461513287732313e-05,
+      "loss": 1.0322,
+      "step": 1364
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 0.1951457034588497,
+      "learning_rate": 1.734201931060706e-05,
+      "loss": 1.0342,
+      "step": 1365
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 0.2013088268621123,
+      "learning_rate": 1.722289677287987e-05,
+      "loss": 0.9935,
+      "step": 1366
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 0.20117633179429092,
+      "learning_rate": 1.710414620985059e-05,
+      "loss": 1.0012,
+      "step": 1367
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 0.2033234102340342,
+      "learning_rate": 1.6985768155147496e-05,
+      "loss": 0.9818,
+      "step": 1368
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 0.20104360454517034,
+      "learning_rate": 1.686776314072497e-05,
+      "loss": 1.0999,
+      "step": 1369
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 0.20940487689123094,
+      "learning_rate": 1.6750131696861015e-05,
+      "loss": 1.0285,
+      "step": 1370
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 0.21680835461628276,
+      "learning_rate": 1.663287435215498e-05,
+      "loss": 0.9571,
+      "step": 1371
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 0.19827472641956026,
+      "learning_rate": 1.6515991633525118e-05,
+      "loss": 1.0499,
+      "step": 1372
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 0.1942134691900752,
+      "learning_rate": 1.6399484066206183e-05,
+      "loss": 1.0287,
+      "step": 1373
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 0.1971846855407925,
+      "learning_rate": 1.6283352173747145e-05,
+      "loss": 1.0295,
+      "step": 1374
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 0.2004843637777262,
+      "learning_rate": 1.6167596478008817e-05,
+      "loss": 0.9847,
+      "step": 1375
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 0.19534562714480122,
+      "learning_rate": 1.6052217499161515e-05,
+      "loss": 0.9995,
+      "step": 1376
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 0.20788157295613702,
+      "learning_rate": 1.5937215755682665e-05,
+      "loss": 1.0213,
+      "step": 1377
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 0.19455795086744224,
+      "learning_rate": 1.582259176435451e-05,
+      "loss": 1.032,
+      "step": 1378
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 0.20159782581581032,
+      "learning_rate": 1.5708346040261813e-05,
+      "loss": 1.0669,
+      "step": 1379
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 0.20640526620902436,
+      "learning_rate": 1.5594479096789537e-05,
+      "loss": 0.9989,
+      "step": 1380
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 0.19589322284099392,
+      "learning_rate": 1.5480991445620542e-05,
+      "loss": 1.0104,
+      "step": 1381
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 0.2036185439151077,
+      "learning_rate": 1.5367883596733155e-05,
+      "loss": 1.0446,
+      "step": 1382
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 0.18952222603214305,
+      "learning_rate": 1.5255156058399122e-05,
+      "loss": 0.9417,
+      "step": 1383
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 0.19652022738548006,
+      "learning_rate": 1.5142809337181063e-05,
+      "loss": 1.0583,
+      "step": 1384
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 0.20045206927797848,
+      "learning_rate": 1.5030843937930483e-05,
+      "loss": 1.0201,
+      "step": 1385
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 0.2055930699724937,
+      "learning_rate": 1.4919260363785215e-05,
+      "loss": 1.0243,
+      "step": 1386
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 0.2162787120662479,
+      "learning_rate": 1.4808059116167305e-05,
+      "loss": 1.0907,
+      "step": 1387
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 0.1977866282036809,
+      "learning_rate": 1.4697240694780767e-05,
+      "loss": 0.986,
+      "step": 1388
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 0.20811708085703706,
+      "learning_rate": 1.4586805597609331e-05,
+      "loss": 1.0941,
+      "step": 1389
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 0.20740381984269454,
+      "learning_rate": 1.4476754320914188e-05,
+      "loss": 1.0413,
+      "step": 1390
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 0.20638961593596633,
+      "learning_rate": 1.4367087359231668e-05,
+      "loss": 1.0453,
+      "step": 1391
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 0.20334283004311088,
+      "learning_rate": 1.4257805205371234e-05,
+      "loss": 1.0005,
+      "step": 1392
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 0.20299293540353064,
+      "learning_rate": 1.4148908350413049e-05,
+      "loss": 1.0365,
+      "step": 1393
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 0.19832697349630915,
+      "learning_rate": 1.4040397283705897e-05,
+      "loss": 1.0487,
+      "step": 1394
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 0.21129996076963695,
+      "learning_rate": 1.3932272492864984e-05,
+      "loss": 1.0439,
+      "step": 1395
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 0.19706789983721826,
+      "learning_rate": 1.3824534463769633e-05,
+      "loss": 1.045,
+      "step": 1396
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 0.19765871458568013,
+      "learning_rate": 1.3717183680561252e-05,
+      "loss": 0.9727,
+      "step": 1397
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 0.20274254850720416,
+      "learning_rate": 1.3610220625641002e-05,
+      "loss": 1.003,
+      "step": 1398
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 0.2027777493566929,
+      "learning_rate": 1.350364577966785e-05,
+      "loss": 1.0572,
+      "step": 1399
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 0.19979864676290385,
+      "learning_rate": 1.339745962155613e-05,
+      "loss": 1.0418,
+      "step": 1400
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 1647,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 50,
+  "total_flos": 1.2994679641473024e+16,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}