diff --git "a/checkpoint-632/trainer_state.json" "b/checkpoint-632/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-632/trainer_state.json"
@@ -0,0 +1,4457 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.6000474721101353,
+  "eval_steps": 500,
+  "global_step": 632,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0009494422027059103,
+      "grad_norm": 5.364119052886963,
+      "learning_rate": 6.329113924050633e-07,
+      "loss": 2.9616,
+      "step": 1
+    },
+    {
+      "epoch": 0.0018988844054118206,
+      "grad_norm": 5.426063060760498,
+      "learning_rate": 1.2658227848101265e-06,
+      "loss": 2.9664,
+      "step": 2
+    },
+    {
+      "epoch": 0.0028483266081177306,
+      "grad_norm": 5.507386684417725,
+      "learning_rate": 1.8987341772151901e-06,
+      "loss": 2.96,
+      "step": 3
+    },
+    {
+      "epoch": 0.0037977688108236413,
+      "grad_norm": 5.467552185058594,
+      "learning_rate": 2.531645569620253e-06,
+      "loss": 2.975,
+      "step": 4
+    },
+    {
+      "epoch": 0.004747211013529551,
+      "grad_norm": 5.386384963989258,
+      "learning_rate": 3.1645569620253167e-06,
+      "loss": 2.9473,
+      "step": 5
+    },
+    {
+      "epoch": 0.005696653216235461,
+      "grad_norm": 4.9166951179504395,
+      "learning_rate": 3.7974683544303802e-06,
+      "loss": 2.8597,
+      "step": 6
+    },
+    {
+      "epoch": 0.006646095418941372,
+      "grad_norm": 5.469020843505859,
+      "learning_rate": 4.430379746835443e-06,
+      "loss": 2.9633,
+      "step": 7
+    },
+    {
+      "epoch": 0.0075955376216472826,
+      "grad_norm": 5.380453586578369,
+      "learning_rate": 5.063291139240506e-06,
+      "loss": 2.8931,
+      "step": 8
+    },
+    {
+      "epoch": 0.008544979824353193,
+      "grad_norm": 4.922253131866455,
+      "learning_rate": 5.69620253164557e-06,
+      "loss": 2.7452,
+      "step": 9
+    },
+    {
+      "epoch": 0.009494422027059102,
+      "grad_norm": 5.517508029937744,
+      "learning_rate": 6.329113924050633e-06,
+      "loss": 2.8326,
+      "step": 10
+    },
+    {
+      "epoch": 0.010443864229765013,
+      "grad_norm": 5.810976982116699,
+      "learning_rate": 6.9620253164556965e-06,
+      "loss": 2.7854,
+      "step": 11
+    },
+    {
+      "epoch": 0.011393306432470923,
+      "grad_norm": 5.690661430358887,
+      "learning_rate": 7.5949367088607605e-06,
+      "loss": 2.7069,
+      "step": 12
+    },
+    {
+      "epoch": 0.012342748635176834,
+      "grad_norm": 5.994122505187988,
+      "learning_rate": 8.227848101265822e-06,
+      "loss": 2.5705,
+      "step": 13
+    },
+    {
+      "epoch": 0.013292190837882745,
+      "grad_norm": 5.86803674697876,
+      "learning_rate": 8.860759493670886e-06,
+      "loss": 2.4461,
+      "step": 14
+    },
+    {
+      "epoch": 0.014241633040588654,
+      "grad_norm": 5.448781490325928,
+      "learning_rate": 9.49367088607595e-06,
+      "loss": 2.2408,
+      "step": 15
+    },
+    {
+      "epoch": 0.015191075243294565,
+      "grad_norm": 6.4004902839660645,
+      "learning_rate": 1.0126582278481012e-05,
+      "loss": 2.1205,
+      "step": 16
+    },
+    {
+      "epoch": 0.016140517446000476,
+      "grad_norm": 6.970590591430664,
+      "learning_rate": 1.0759493670886076e-05,
+      "loss": 1.9474,
+      "step": 17
+    },
+    {
+      "epoch": 0.017089959648706386,
+      "grad_norm": 7.423785209655762,
+      "learning_rate": 1.139240506329114e-05,
+      "loss": 1.7348,
+      "step": 18
+    },
+    {
+      "epoch": 0.018039401851412295,
+      "grad_norm": 7.429481029510498,
+      "learning_rate": 1.2025316455696203e-05,
+      "loss": 1.4835,
+      "step": 19
+    },
+    {
+      "epoch": 0.018988844054118204,
+      "grad_norm": 6.7193284034729,
+      "learning_rate": 1.2658227848101267e-05,
+      "loss": 1.2373,
+      "step": 20
+    },
+    {
+      "epoch": 0.019938286256824117,
+      "grad_norm": 4.46099853515625,
+      "learning_rate": 1.3291139240506329e-05,
+      "loss": 1.1095,
+      "step": 21
+    },
+    {
+      "epoch": 0.020887728459530026,
+      "grad_norm": 3.001573085784912,
+      "learning_rate": 1.3924050632911393e-05,
+      "loss": 0.8642,
+      "step": 22
+    },
+    {
+      "epoch": 0.021837170662235936,
+      "grad_norm": 2.197000026702881,
+      "learning_rate": 1.4556962025316457e-05,
+      "loss": 0.7734,
+      "step": 23
+    },
+    {
+      "epoch": 0.022786612864941845,
+      "grad_norm": 1.8113943338394165,
+      "learning_rate": 1.5189873417721521e-05,
+      "loss": 0.7341,
+      "step": 24
+    },
+    {
+      "epoch": 0.023736055067647758,
+      "grad_norm": 1.7461305856704712,
+      "learning_rate": 1.5822784810126583e-05,
+      "loss": 0.6743,
+      "step": 25
+    },
+    {
+      "epoch": 0.024685497270353667,
+      "grad_norm": 1.3315849304199219,
+      "learning_rate": 1.6455696202531644e-05,
+      "loss": 0.5975,
+      "step": 26
+    },
+    {
+      "epoch": 0.025634939473059577,
+      "grad_norm": 0.726314127445221,
+      "learning_rate": 1.7088607594936708e-05,
+      "loss": 0.5659,
+      "step": 27
+    },
+    {
+      "epoch": 0.02658438167576549,
+      "grad_norm": 0.6269010901451111,
+      "learning_rate": 1.7721518987341772e-05,
+      "loss": 0.5226,
+      "step": 28
+    },
+    {
+      "epoch": 0.0275338238784714,
+      "grad_norm": 0.5819966197013855,
+      "learning_rate": 1.8354430379746836e-05,
+      "loss": 0.6132,
+      "step": 29
+    },
+    {
+      "epoch": 0.028483266081177308,
+      "grad_norm": 0.6247850060462952,
+      "learning_rate": 1.89873417721519e-05,
+      "loss": 0.5,
+      "step": 30
+    },
+    {
+      "epoch": 0.029432708283883217,
+      "grad_norm": 0.702621579170227,
+      "learning_rate": 1.962025316455696e-05,
+      "loss": 0.4958,
+      "step": 31
+    },
+    {
+      "epoch": 0.03038215048658913,
+      "grad_norm": 0.6045309901237488,
+      "learning_rate": 2.0253164556962025e-05,
+      "loss": 0.4405,
+      "step": 32
+    },
+    {
+      "epoch": 0.031331592689295036,
+      "grad_norm": 0.5436626076698303,
+      "learning_rate": 2.088607594936709e-05,
+      "loss": 0.5607,
+      "step": 33
+    },
+    {
+      "epoch": 0.03228103489200095,
+      "grad_norm": 0.43146297335624695,
+      "learning_rate": 2.1518987341772153e-05,
+      "loss": 0.3987,
+      "step": 34
+    },
+    {
+      "epoch": 0.03323047709470686,
+      "grad_norm": 0.5124548673629761,
+      "learning_rate": 2.2151898734177217e-05,
+      "loss": 0.5084,
+      "step": 35
+    },
+    {
+      "epoch": 0.03417991929741277,
+      "grad_norm": 0.4466649293899536,
+      "learning_rate": 2.278481012658228e-05,
+      "loss": 0.3761,
+      "step": 36
+    },
+    {
+      "epoch": 0.03512936150011868,
+      "grad_norm": 0.41221529245376587,
+      "learning_rate": 2.341772151898734e-05,
+      "loss": 0.3859,
+      "step": 37
+    },
+    {
+      "epoch": 0.03607880370282459,
+      "grad_norm": 0.3802257180213928,
+      "learning_rate": 2.4050632911392405e-05,
+      "loss": 0.3447,
+      "step": 38
+    },
+    {
+      "epoch": 0.0370282459055305,
+      "grad_norm": 0.47727710008621216,
+      "learning_rate": 2.468354430379747e-05,
+      "loss": 0.3914,
+      "step": 39
+    },
+    {
+      "epoch": 0.03797768810823641,
+      "grad_norm": 0.41048529744148254,
+      "learning_rate": 2.5316455696202533e-05,
+      "loss": 0.2988,
+      "step": 40
+    },
+    {
+      "epoch": 0.038927130310942325,
+      "grad_norm": 0.5019667744636536,
+      "learning_rate": 2.5949367088607597e-05,
+      "loss": 0.2938,
+      "step": 41
+    },
+    {
+      "epoch": 0.039876572513648234,
+      "grad_norm": 0.42121732234954834,
+      "learning_rate": 2.6582278481012658e-05,
+      "loss": 0.2579,
+      "step": 42
+    },
+    {
+      "epoch": 0.04082601471635414,
+      "grad_norm": 0.4193897247314453,
+      "learning_rate": 2.7215189873417722e-05,
+      "loss": 0.3262,
+      "step": 43
+    },
+    {
+      "epoch": 0.04177545691906005,
+      "grad_norm": 0.2978931665420532,
+      "learning_rate": 2.7848101265822786e-05,
+      "loss": 0.2365,
+      "step": 44
+    },
+    {
+      "epoch": 0.04272489912176596,
+      "grad_norm": 0.34771448373794556,
+      "learning_rate": 2.848101265822785e-05,
+      "loss": 0.2364,
+      "step": 45
+    },
+    {
+      "epoch": 0.04367434132447187,
+      "grad_norm": 0.3881576955318451,
+      "learning_rate": 2.9113924050632914e-05,
+      "loss": 0.286,
+      "step": 46
+    },
+    {
+      "epoch": 0.04462378352717778,
+      "grad_norm": 0.33863797783851624,
+      "learning_rate": 2.9746835443037974e-05,
+      "loss": 0.2739,
+      "step": 47
+    },
+    {
+      "epoch": 0.04557322572988369,
+      "grad_norm": 0.2894616723060608,
+      "learning_rate": 3.0379746835443042e-05,
+      "loss": 0.2587,
+      "step": 48
+    },
+    {
+      "epoch": 0.046522667932589606,
+      "grad_norm": 0.22292694449424744,
+      "learning_rate": 3.10126582278481e-05,
+      "loss": 0.1861,
+      "step": 49
+    },
+    {
+      "epoch": 0.047472110135295516,
+      "grad_norm": 0.21907460689544678,
+      "learning_rate": 3.1645569620253167e-05,
+      "loss": 0.1755,
+      "step": 50
+    },
+    {
+      "epoch": 0.048421552338001425,
+      "grad_norm": 0.29593944549560547,
+      "learning_rate": 3.227848101265823e-05,
+      "loss": 0.1856,
+      "step": 51
+    },
+    {
+      "epoch": 0.049370994540707334,
+      "grad_norm": 0.23055657744407654,
+      "learning_rate": 3.291139240506329e-05,
+      "loss": 0.2102,
+      "step": 52
+    },
+    {
+      "epoch": 0.050320436743413244,
+      "grad_norm": 0.18929323554039001,
+      "learning_rate": 3.354430379746836e-05,
+      "loss": 0.1909,
+      "step": 53
+    },
+    {
+      "epoch": 0.05126987894611915,
+      "grad_norm": 0.15004883706569672,
+      "learning_rate": 3.4177215189873416e-05,
+      "loss": 0.1619,
+      "step": 54
+    },
+    {
+      "epoch": 0.05221932114882506,
+      "grad_norm": 0.15621644258499146,
+      "learning_rate": 3.4810126582278487e-05,
+      "loss": 0.1759,
+      "step": 55
+    },
+    {
+      "epoch": 0.05316876335153098,
+      "grad_norm": 0.16266578435897827,
+      "learning_rate": 3.5443037974683544e-05,
+      "loss": 0.1657,
+      "step": 56
+    },
+    {
+      "epoch": 0.05411820555423689,
+      "grad_norm": 0.14417718350887299,
+      "learning_rate": 3.607594936708861e-05,
+      "loss": 0.1698,
+      "step": 57
+    },
+    {
+      "epoch": 0.0550676477569428,
+      "grad_norm": 0.21402889490127563,
+      "learning_rate": 3.670886075949367e-05,
+      "loss": 0.2185,
+      "step": 58
+    },
+    {
+      "epoch": 0.05601708995964871,
+      "grad_norm": 0.1997889280319214,
+      "learning_rate": 3.7341772151898736e-05,
+      "loss": 0.2143,
+      "step": 59
+    },
+    {
+      "epoch": 0.056966532162354616,
+      "grad_norm": 0.13755086064338684,
+      "learning_rate": 3.79746835443038e-05,
+      "loss": 0.1677,
+      "step": 60
+    },
+    {
+      "epoch": 0.057915974365060525,
+      "grad_norm": 0.19304363429546356,
+      "learning_rate": 3.8607594936708864e-05,
+      "loss": 0.2113,
+      "step": 61
+    },
+    {
+      "epoch": 0.058865416567766435,
+      "grad_norm": 0.14066031575202942,
+      "learning_rate": 3.924050632911392e-05,
+      "loss": 0.1612,
+      "step": 62
+    },
+    {
+      "epoch": 0.059814858770472344,
+      "grad_norm": 0.13375213742256165,
+      "learning_rate": 3.987341772151899e-05,
+      "loss": 0.164,
+      "step": 63
+    },
+    {
+      "epoch": 0.06076430097317826,
+      "grad_norm": 0.15216922760009766,
+      "learning_rate": 4.050632911392405e-05,
+      "loss": 0.16,
+      "step": 64
+    },
+    {
+      "epoch": 0.06171374317588417,
+      "grad_norm": 0.16130389273166656,
+      "learning_rate": 4.113924050632912e-05,
+      "loss": 0.1957,
+      "step": 65
+    },
+    {
+      "epoch": 0.06266318537859007,
+      "grad_norm": 0.1791229248046875,
+      "learning_rate": 4.177215189873418e-05,
+      "loss": 0.1993,
+      "step": 66
+    },
+    {
+      "epoch": 0.06361262758129599,
+      "grad_norm": 0.11038907617330551,
+      "learning_rate": 4.240506329113924e-05,
+      "loss": 0.1517,
+      "step": 67
+    },
+    {
+      "epoch": 0.0645620697840019,
+      "grad_norm": 0.13327902555465698,
+      "learning_rate": 4.3037974683544305e-05,
+      "loss": 0.1501,
+      "step": 68
+    },
+    {
+      "epoch": 0.06551151198670781,
+      "grad_norm": 0.13731731474399567,
+      "learning_rate": 4.367088607594937e-05,
+      "loss": 0.1596,
+      "step": 69
+    },
+    {
+      "epoch": 0.06646095418941372,
+      "grad_norm": 0.13924308121204376,
+      "learning_rate": 4.430379746835443e-05,
+      "loss": 0.152,
+      "step": 70
+    },
+    {
+      "epoch": 0.06741039639211963,
+      "grad_norm": 0.1482289433479309,
+      "learning_rate": 4.49367088607595e-05,
+      "loss": 0.1536,
+      "step": 71
+    },
+    {
+      "epoch": 0.06835983859482554,
+      "grad_norm": 0.10759364813566208,
+      "learning_rate": 4.556962025316456e-05,
+      "loss": 0.1543,
+      "step": 72
+    },
+    {
+      "epoch": 0.06930928079753144,
+      "grad_norm": 0.12899678945541382,
+      "learning_rate": 4.6202531645569625e-05,
+      "loss": 0.165,
+      "step": 73
+    },
+    {
+      "epoch": 0.07025872300023736,
+      "grad_norm": 0.11689919233322144,
+      "learning_rate": 4.683544303797468e-05,
+      "loss": 0.1564,
+      "step": 74
+    },
+    {
+      "epoch": 0.07120816520294328,
+      "grad_norm": 0.12697139382362366,
+      "learning_rate": 4.7468354430379746e-05,
+      "loss": 0.162,
+      "step": 75
+    },
+    {
+      "epoch": 0.07215760740564918,
+      "grad_norm": 0.12069376558065414,
+      "learning_rate": 4.810126582278481e-05,
+      "loss": 0.1467,
+      "step": 76
+    },
+    {
+      "epoch": 0.0731070496083551,
+      "grad_norm": 0.10199815034866333,
+      "learning_rate": 4.8734177215189874e-05,
+      "loss": 0.1528,
+      "step": 77
+    },
+    {
+      "epoch": 0.074056491811061,
+      "grad_norm": 0.1142750009894371,
+      "learning_rate": 4.936708860759494e-05,
+      "loss": 0.1574,
+      "step": 78
+    },
+    {
+      "epoch": 0.07500593401376691,
+      "grad_norm": 0.11019093543291092,
+      "learning_rate": 5e-05,
+      "loss": 0.1512,
+      "step": 79
+    },
+    {
+      "epoch": 0.07595537621647282,
+      "grad_norm": 0.09426973015069962,
+      "learning_rate": 5.0632911392405066e-05,
+      "loss": 0.1481,
+      "step": 80
+    },
+    {
+      "epoch": 0.07690481841917873,
+      "grad_norm": 0.09757663309574127,
+      "learning_rate": 5.1265822784810124e-05,
+      "loss": 0.1484,
+      "step": 81
+    },
+    {
+      "epoch": 0.07785426062188465,
+      "grad_norm": 0.10646392405033112,
+      "learning_rate": 5.1898734177215194e-05,
+      "loss": 0.1549,
+      "step": 82
+    },
+    {
+      "epoch": 0.07880370282459055,
+      "grad_norm": 0.12109784036874771,
+      "learning_rate": 5.253164556962026e-05,
+      "loss": 0.1448,
+      "step": 83
+    },
+    {
+      "epoch": 0.07975314502729647,
+      "grad_norm": 0.12039211392402649,
+      "learning_rate": 5.3164556962025316e-05,
+      "loss": 0.1538,
+      "step": 84
+    },
+    {
+      "epoch": 0.08070258723000237,
+      "grad_norm": 0.16873961687088013,
+      "learning_rate": 5.379746835443038e-05,
+      "loss": 0.1971,
+      "step": 85
+    },
+    {
+      "epoch": 0.08165202943270829,
+      "grad_norm": 0.12140022218227386,
+      "learning_rate": 5.4430379746835444e-05,
+      "loss": 0.1497,
+      "step": 86
+    },
+    {
+      "epoch": 0.08260147163541419,
+      "grad_norm": 0.14637599885463715,
+      "learning_rate": 5.5063291139240514e-05,
+      "loss": 0.1958,
+      "step": 87
+    },
+    {
+      "epoch": 0.0835509138381201,
+      "grad_norm": 0.1141396313905716,
+      "learning_rate": 5.569620253164557e-05,
+      "loss": 0.1457,
+      "step": 88
+    },
+    {
+      "epoch": 0.08450035604082601,
+      "grad_norm": 0.2128390371799469,
+      "learning_rate": 5.6329113924050636e-05,
+      "loss": 0.2339,
+      "step": 89
+    },
+    {
+      "epoch": 0.08544979824353192,
+      "grad_norm": 0.18838858604431152,
+      "learning_rate": 5.69620253164557e-05,
+      "loss": 0.2029,
+      "step": 90
+    },
+    {
+      "epoch": 0.08639924044623784,
+      "grad_norm": 0.19592566788196564,
+      "learning_rate": 5.759493670886076e-05,
+      "loss": 0.2276,
+      "step": 91
+    },
+    {
+      "epoch": 0.08734868264894374,
+      "grad_norm": 0.14753012359142303,
+      "learning_rate": 5.822784810126583e-05,
+      "loss": 0.1916,
+      "step": 92
+    },
+    {
+      "epoch": 0.08829812485164966,
+      "grad_norm": 0.1494351178407669,
+      "learning_rate": 5.886075949367089e-05,
+      "loss": 0.1913,
+      "step": 93
+    },
+    {
+      "epoch": 0.08924756705435556,
+      "grad_norm": 0.1173478439450264,
+      "learning_rate": 5.949367088607595e-05,
+      "loss": 0.1438,
+      "step": 94
+    },
+    {
+      "epoch": 0.09019700925706148,
+      "grad_norm": 0.12023188918828964,
+      "learning_rate": 6.012658227848101e-05,
+      "loss": 0.1516,
+      "step": 95
+    },
+    {
+      "epoch": 0.09114645145976738,
+      "grad_norm": 0.1275833547115326,
+      "learning_rate": 6.0759493670886084e-05,
+      "loss": 0.1492,
+      "step": 96
+    },
+    {
+      "epoch": 0.0920958936624733,
+      "grad_norm": 0.1360282599925995,
+      "learning_rate": 6.139240506329115e-05,
+      "loss": 0.1507,
+      "step": 97
+    },
+    {
+      "epoch": 0.09304533586517921,
+      "grad_norm": 0.1586841195821762,
+      "learning_rate": 6.20253164556962e-05,
+      "loss": 0.1956,
+      "step": 98
+    },
+    {
+      "epoch": 0.09399477806788512,
+      "grad_norm": 0.14281995594501495,
+      "learning_rate": 6.265822784810128e-05,
+      "loss": 0.1774,
+      "step": 99
+    },
+    {
+      "epoch": 0.09494422027059103,
+      "grad_norm": 0.12553077936172485,
+      "learning_rate": 6.329113924050633e-05,
+      "loss": 0.148,
+      "step": 100
+    },
+    {
+      "epoch": 0.09589366247329693,
+      "grad_norm": 0.1117570698261261,
+      "learning_rate": 6.392405063291139e-05,
+      "loss": 0.16,
+      "step": 101
+    },
+    {
+      "epoch": 0.09684310467600285,
+      "grad_norm": 0.13955281674861908,
+      "learning_rate": 6.455696202531646e-05,
+      "loss": 0.1464,
+      "step": 102
+    },
+    {
+      "epoch": 0.09779254687870875,
+      "grad_norm": 0.10990285873413086,
+      "learning_rate": 6.518987341772153e-05,
+      "loss": 0.147,
+      "step": 103
+    },
+    {
+      "epoch": 0.09874198908141467,
+      "grad_norm": 0.10545991361141205,
+      "learning_rate": 6.582278481012658e-05,
+      "loss": 0.1436,
+      "step": 104
+    },
+    {
+      "epoch": 0.09969143128412059,
+      "grad_norm": 0.1717437207698822,
+      "learning_rate": 6.645569620253165e-05,
+      "loss": 0.2278,
+      "step": 105
+    },
+    {
+      "epoch": 0.10064087348682649,
+      "grad_norm": 0.10950994491577148,
+      "learning_rate": 6.708860759493672e-05,
+      "loss": 0.1493,
+      "step": 106
+    },
+    {
+      "epoch": 0.1015903156895324,
+      "grad_norm": 0.11200258880853653,
+      "learning_rate": 6.772151898734177e-05,
+      "loss": 0.1536,
+      "step": 107
+    },
+    {
+      "epoch": 0.1025397578922383,
+      "grad_norm": 0.10955105721950531,
+      "learning_rate": 6.835443037974683e-05,
+      "loss": 0.1483,
+      "step": 108
+    },
+    {
+      "epoch": 0.10348920009494422,
+      "grad_norm": 0.11920775473117828,
+      "learning_rate": 6.89873417721519e-05,
+      "loss": 0.1492,
+      "step": 109
+    },
+    {
+      "epoch": 0.10443864229765012,
+      "grad_norm": 0.1390092819929123,
+      "learning_rate": 6.962025316455697e-05,
+      "loss": 0.1849,
+      "step": 110
+    },
+    {
+      "epoch": 0.10538808450035604,
+      "grad_norm": 0.1363140493631363,
+      "learning_rate": 7.025316455696203e-05,
+      "loss": 0.1849,
+      "step": 111
+    },
+    {
+      "epoch": 0.10633752670306196,
+      "grad_norm": 0.09190025180578232,
+      "learning_rate": 7.088607594936709e-05,
+      "loss": 0.1587,
+      "step": 112
+    },
+    {
+      "epoch": 0.10728696890576786,
+      "grad_norm": 0.09020426124334335,
+      "learning_rate": 7.151898734177216e-05,
+      "loss": 0.1377,
+      "step": 113
+    },
+    {
+      "epoch": 0.10823641110847378,
+      "grad_norm": 0.10544883459806442,
+      "learning_rate": 7.215189873417722e-05,
+      "loss": 0.1516,
+      "step": 114
+    },
+    {
+      "epoch": 0.10918585331117968,
+      "grad_norm": 0.12401281297206879,
+      "learning_rate": 7.278481012658229e-05,
+      "loss": 0.154,
+      "step": 115
+    },
+    {
+      "epoch": 0.1101352955138856,
+      "grad_norm": 0.1008707657456398,
+      "learning_rate": 7.341772151898734e-05,
+      "loss": 0.1448,
+      "step": 116
+    },
+    {
+      "epoch": 0.1110847377165915,
+      "grad_norm": 0.10302747040987015,
+      "learning_rate": 7.40506329113924e-05,
+      "loss": 0.1451,
+      "step": 117
+    },
+    {
+      "epoch": 0.11203417991929741,
+      "grad_norm": 0.12748293578624725,
+      "learning_rate": 7.468354430379747e-05,
+      "loss": 0.1829,
+      "step": 118
+    },
+    {
+      "epoch": 0.11298362212200333,
+      "grad_norm": 0.10413361340761185,
+      "learning_rate": 7.531645569620254e-05,
+      "loss": 0.1371,
+      "step": 119
+    },
+    {
+      "epoch": 0.11393306432470923,
+      "grad_norm": 0.1243433803319931,
+      "learning_rate": 7.59493670886076e-05,
+      "loss": 0.1409,
+      "step": 120
+    },
+    {
+      "epoch": 0.11488250652741515,
+      "grad_norm": 0.11630933731794357,
+      "learning_rate": 7.658227848101266e-05,
+      "loss": 0.1372,
+      "step": 121
+    },
+    {
+      "epoch": 0.11583194873012105,
+      "grad_norm": 0.17981529235839844,
+      "learning_rate": 7.721518987341773e-05,
+      "loss": 0.2257,
+      "step": 122
+    },
+    {
+      "epoch": 0.11678139093282697,
+      "grad_norm": 0.14063452184200287,
+      "learning_rate": 7.78481012658228e-05,
+      "loss": 0.1841,
+      "step": 123
+    },
+    {
+      "epoch": 0.11773083313553287,
+      "grad_norm": 0.1264188438653946,
+      "learning_rate": 7.848101265822784e-05,
+      "loss": 0.1471,
+      "step": 124
+    },
+    {
+      "epoch": 0.11868027533823879,
+      "grad_norm": 0.12827955186367035,
+      "learning_rate": 7.911392405063291e-05,
+      "loss": 0.1493,
+      "step": 125
+    },
+    {
+      "epoch": 0.11962971754094469,
+      "grad_norm": 0.09800329059362411,
+      "learning_rate": 7.974683544303798e-05,
+      "loss": 0.1414,
+      "step": 126
+    },
+    {
+      "epoch": 0.1205791597436506,
+      "grad_norm": 0.09902197122573853,
+      "learning_rate": 8.037974683544304e-05,
+      "loss": 0.1462,
+      "step": 127
+    },
+    {
+      "epoch": 0.12152860194635652,
+      "grad_norm": 0.09450504928827286,
+      "learning_rate": 8.10126582278481e-05,
+      "loss": 0.1484,
+      "step": 128
+    },
+    {
+      "epoch": 0.12247804414906242,
+      "grad_norm": 0.11012883484363556,
+      "learning_rate": 8.164556962025317e-05,
+      "loss": 0.1437,
+      "step": 129
+    },
+    {
+      "epoch": 0.12342748635176834,
+      "grad_norm": 0.11717642843723297,
+      "learning_rate": 8.227848101265824e-05,
+      "loss": 0.1478,
+      "step": 130
+    },
+    {
+      "epoch": 0.12437692855447424,
+      "grad_norm": 0.08754123747348785,
+      "learning_rate": 8.29113924050633e-05,
+      "loss": 0.1408,
+      "step": 131
+    },
+    {
+      "epoch": 0.12532637075718014,
+      "grad_norm": 0.10017862170934677,
+      "learning_rate": 8.354430379746835e-05,
+      "loss": 0.1476,
+      "step": 132
+    },
+    {
+      "epoch": 0.12627581295988607,
+      "grad_norm": 0.08994068205356598,
+      "learning_rate": 8.417721518987342e-05,
+      "loss": 0.1478,
+      "step": 133
+    },
+    {
+      "epoch": 0.12722525516259198,
+      "grad_norm": 0.09894968569278717,
+      "learning_rate": 8.481012658227848e-05,
+      "loss": 0.1309,
+      "step": 134
+    },
+    {
+      "epoch": 0.12817469736529788,
+      "grad_norm": 0.10028701275587082,
+      "learning_rate": 8.544303797468355e-05,
+      "loss": 0.1433,
+      "step": 135
+    },
+    {
+      "epoch": 0.1291241395680038,
+      "grad_norm": 0.0897536426782608,
+      "learning_rate": 8.607594936708861e-05,
+      "loss": 0.1459,
+      "step": 136
+    },
+    {
+      "epoch": 0.1300735817707097,
+      "grad_norm": 0.10435349494218826,
+      "learning_rate": 8.670886075949367e-05,
+      "loss": 0.1434,
+      "step": 137
+    },
+    {
+      "epoch": 0.13102302397341561,
+      "grad_norm": 0.11718117445707321,
+      "learning_rate": 8.734177215189874e-05,
+      "loss": 0.1509,
+      "step": 138
+    },
+    {
+      "epoch": 0.13197246617612152,
+      "grad_norm": 0.14426474273204803,
+      "learning_rate": 8.797468354430381e-05,
+      "loss": 0.1373,
+      "step": 139
+    },
+    {
+      "epoch": 0.13292190837882745,
+      "grad_norm": 0.13101965188980103,
+      "learning_rate": 8.860759493670887e-05,
+      "loss": 0.1358,
+      "step": 140
+    },
+    {
+      "epoch": 0.13387135058153335,
+      "grad_norm": 0.11235956102609634,
+      "learning_rate": 8.924050632911392e-05,
+      "loss": 0.1394,
+      "step": 141
+    },
+    {
+      "epoch": 0.13482079278423925,
+      "grad_norm": 0.11327100545167923,
+      "learning_rate": 8.9873417721519e-05,
+      "loss": 0.1443,
+      "step": 142
+    },
+    {
+      "epoch": 0.13577023498694518,
+      "grad_norm": 0.10912016034126282,
+      "learning_rate": 9.050632911392407e-05,
+      "loss": 0.1698,
+      "step": 143
+    },
+    {
+      "epoch": 0.13671967718965108,
+      "grad_norm": 0.16535617411136627,
+      "learning_rate": 9.113924050632912e-05,
+      "loss": 0.2255,
+      "step": 144
+    },
+    {
+      "epoch": 0.137669119392357,
+      "grad_norm": 0.10184327512979507,
+      "learning_rate": 9.177215189873418e-05,
+      "loss": 0.1371,
+      "step": 145
+    },
+    {
+      "epoch": 0.1386185615950629,
+      "grad_norm": 0.10998040437698364,
+      "learning_rate": 9.240506329113925e-05,
+      "loss": 0.1794,
+      "step": 146
+    },
+    {
+      "epoch": 0.13956800379776882,
+      "grad_norm": 0.08974044770002365,
+      "learning_rate": 9.303797468354431e-05,
+      "loss": 0.144,
+      "step": 147
+    },
+    {
+      "epoch": 0.14051744600047472,
+      "grad_norm": 0.12724193930625916,
+      "learning_rate": 9.367088607594936e-05,
+      "loss": 0.1794,
+      "step": 148
+    },
+    {
+      "epoch": 0.14146688820318062,
+      "grad_norm": 0.1079091802239418,
+      "learning_rate": 9.430379746835444e-05,
+      "loss": 0.1399,
+      "step": 149
+    },
+    {
+      "epoch": 0.14241633040588655,
+      "grad_norm": 0.09480807185173035,
+      "learning_rate": 9.493670886075949e-05,
+      "loss": 0.1395,
+      "step": 150
+    },
+    {
+      "epoch": 0.14336577260859246,
+      "grad_norm": 0.08620745688676834,
+      "learning_rate": 9.556962025316456e-05,
+      "loss": 0.1415,
+      "step": 151
+    },
+    {
+      "epoch": 0.14431521481129836,
+      "grad_norm": 0.10517002642154694,
+      "learning_rate": 9.620253164556962e-05,
+      "loss": 0.1723,
+      "step": 152
+    },
+    {
+      "epoch": 0.14526465701400426,
+      "grad_norm": 0.0956311896443367,
+      "learning_rate": 9.683544303797469e-05,
+      "loss": 0.1515,
+      "step": 153
+    },
+    {
+      "epoch": 0.1462140992167102,
+      "grad_norm": 0.08050324022769928,
+      "learning_rate": 9.746835443037975e-05,
+      "loss": 0.1322,
+      "step": 154
+    },
+    {
+      "epoch": 0.1471635414194161,
+      "grad_norm": 0.0853201299905777,
+      "learning_rate": 9.810126582278482e-05,
+      "loss": 0.142,
+      "step": 155
+    },
+    {
+      "epoch": 0.148112983622122,
+      "grad_norm": 0.09991180151700974,
+      "learning_rate": 9.873417721518988e-05,
+      "loss": 0.1348,
+      "step": 156
+    },
+    {
+      "epoch": 0.14906242582482793,
+      "grad_norm": 0.08640603721141815,
+      "learning_rate": 9.936708860759493e-05,
+      "loss": 0.1397,
+      "step": 157
+    },
+    {
+      "epoch": 0.15001186802753383,
+      "grad_norm": 0.09057717025279999,
+      "learning_rate": 0.0001,
+      "loss": 0.1381,
+      "step": 158
+    },
+    {
+      "epoch": 0.15096131023023973,
+      "grad_norm": 0.09916041046380997,
+      "learning_rate": 0.00010063291139240508,
+      "loss": 0.1509,
+      "step": 159
+    },
+    {
+      "epoch": 0.15191075243294563,
+      "grad_norm": 0.09434045851230621,
+      "learning_rate": 0.00010126582278481013,
+      "loss": 0.1388,
+      "step": 160
+    },
+    {
+      "epoch": 0.15286019463565156,
+      "grad_norm": 0.1273377537727356,
+      "learning_rate": 0.0001018987341772152,
+      "loss": 0.1401,
+      "step": 161
+    },
+    {
+      "epoch": 0.15380963683835747,
+      "grad_norm": 0.1297912299633026,
+      "learning_rate": 0.00010253164556962025,
+      "loss": 0.1852,
+      "step": 162
+    },
+    {
+      "epoch": 0.15475907904106337,
+      "grad_norm": 0.1151595488190651,
+      "learning_rate": 0.00010316455696202532,
+      "loss": 0.1848,
+      "step": 163
+    },
+    {
+      "epoch": 0.1557085212437693,
+      "grad_norm": 0.13381290435791016,
+      "learning_rate": 0.00010379746835443039,
+      "loss": 0.1438,
+      "step": 164
+    },
+    {
+      "epoch": 0.1566579634464752,
+      "grad_norm": 0.07880119979381561,
+      "learning_rate": 0.00010443037974683545,
+      "loss": 0.1327,
+      "step": 165
+    },
+    {
+      "epoch": 0.1576074056491811,
+      "grad_norm": 0.0843740776181221,
+      "learning_rate": 0.00010506329113924052,
+      "loss": 0.1398,
+      "step": 166
+    },
+    {
+      "epoch": 0.158556847851887,
+      "grad_norm": 0.0981813594698906,
+      "learning_rate": 0.00010569620253164559,
+      "loss": 0.1409,
+      "step": 167
+    },
+    {
+      "epoch": 0.15950629005459294,
+      "grad_norm": 0.10005304962396622,
+      "learning_rate": 0.00010632911392405063,
+      "loss": 0.1783,
+      "step": 168
+    },
+    {
+      "epoch": 0.16045573225729884,
+      "grad_norm": 0.08365727961063385,
+      "learning_rate": 0.00010696202531645569,
+      "loss": 0.1275,
+      "step": 169
+    },
+    {
+      "epoch": 0.16140517446000474,
+      "grad_norm": 0.1017635315656662,
+      "learning_rate": 0.00010759493670886076,
+      "loss": 0.1792,
+      "step": 170
+    },
+    {
+      "epoch": 0.16235461666271067,
+      "grad_norm": 0.07007888704538345,
+      "learning_rate": 0.00010822784810126583,
+      "loss": 0.1473,
+      "step": 171
+    },
+    {
+      "epoch": 0.16330405886541657,
+      "grad_norm": 0.07718679308891296,
+      "learning_rate": 0.00010886075949367089,
+      "loss": 0.1396,
+      "step": 172
+    },
+    {
+      "epoch": 0.16425350106812248,
+      "grad_norm": 0.07228100299835205,
+      "learning_rate": 0.00010949367088607596,
+      "loss": 0.1398,
+      "step": 173
+    },
+    {
+      "epoch": 0.16520294327082838,
+      "grad_norm": 0.07955378293991089,
+      "learning_rate": 0.00011012658227848103,
+      "loss": 0.1402,
+      "step": 174
+    },
+    {
+      "epoch": 0.1661523854735343,
+      "grad_norm": 0.0816427692770958,
+      "learning_rate": 0.00011075949367088607,
+      "loss": 0.1345,
+      "step": 175
+    },
+    {
+      "epoch": 0.1671018276762402,
+      "grad_norm": 0.07641757279634476,
+      "learning_rate": 0.00011139240506329114,
+      "loss": 0.1373,
+      "step": 176
+    },
+    {
+      "epoch": 0.1680512698789461,
+      "grad_norm": 0.07354450225830078,
+      "learning_rate": 0.0001120253164556962,
+      "loss": 0.1394,
+      "step": 177
+    },
+    {
+      "epoch": 0.16900071208165202,
+      "grad_norm": 0.08322398364543915,
+      "learning_rate": 0.00011265822784810127,
+      "loss": 0.138,
+      "step": 178
+    },
+    {
+      "epoch": 0.16995015428435795,
+      "grad_norm": 0.13528607785701752,
+      "learning_rate": 0.00011329113924050634,
+      "loss": 0.2188,
+      "step": 179
+    },
+    {
+      "epoch": 0.17089959648706385,
+      "grad_norm": 0.10803692042827606,
+      "learning_rate": 0.0001139240506329114,
+      "loss": 0.1782,
+      "step": 180
+    },
+    {
+      "epoch": 0.17184903868976975,
+      "grad_norm": 0.08404573053121567,
+      "learning_rate": 0.00011455696202531647,
+      "loss": 0.1394,
+      "step": 181
+    },
+    {
+      "epoch": 0.17279848089247568,
+      "grad_norm": 0.12790893018245697,
+      "learning_rate": 0.00011518987341772151,
+      "loss": 0.2157,
+      "step": 182
+    },
+    {
+      "epoch": 0.17374792309518158,
+      "grad_norm": 0.09879907220602036,
+      "learning_rate": 0.00011582278481012658,
+      "loss": 0.1693,
+      "step": 183
+    },
+    {
+      "epoch": 0.17469736529788749,
+      "grad_norm": 0.08092228323221207,
+      "learning_rate": 0.00011645569620253166,
+      "loss": 0.136,
+      "step": 184
+    },
+    {
+      "epoch": 0.1756468075005934,
+      "grad_norm": 0.07660632580518723,
+      "learning_rate": 0.00011708860759493671,
+      "loss": 0.1332,
+      "step": 185
+    },
+    {
+      "epoch": 0.17659624970329932,
+      "grad_norm": 0.07474201172590256,
+      "learning_rate": 0.00011772151898734178,
+      "loss": 0.1301,
+      "step": 186
+    },
+    {
+      "epoch": 0.17754569190600522,
+      "grad_norm": 0.09162931889295578,
+      "learning_rate": 0.00011835443037974685,
+      "loss": 0.1407,
+      "step": 187
+    },
+    {
+      "epoch": 0.17849513410871112,
+      "grad_norm": 0.08646775782108307,
+      "learning_rate": 0.0001189873417721519,
+      "loss": 0.139,
+      "step": 188
+    },
+    {
+      "epoch": 0.17944457631141705,
+      "grad_norm": 0.0759253203868866,
+      "learning_rate": 0.00011962025316455696,
+      "loss": 0.1342,
+      "step": 189
+    },
+    {
+      "epoch": 0.18039401851412296,
+      "grad_norm": 0.08292865008115768,
+      "learning_rate": 0.00012025316455696203,
+      "loss": 0.1389,
+      "step": 190
+    },
+    {
+      "epoch": 0.18134346071682886,
+      "grad_norm": 0.12379574030637741,
+      "learning_rate": 0.0001208860759493671,
+      "loss": 0.1795,
+      "step": 191
+    },
+    {
+      "epoch": 0.18229290291953476,
+      "grad_norm": 0.10240278393030167,
+      "learning_rate": 0.00012151898734177217,
+      "loss": 0.1721,
+      "step": 192
+    },
+    {
+      "epoch": 0.1832423451222407,
+      "grad_norm": 0.09666036069393158,
+      "learning_rate": 0.00012215189873417722,
+      "loss": 0.1783,
+      "step": 193
+    },
+    {
+      "epoch": 0.1841917873249466,
+      "grad_norm": 0.08314768224954605,
+      "learning_rate": 0.0001227848101265823,
+      "loss": 0.1429,
+      "step": 194
+    },
+    {
+      "epoch": 0.1851412295276525,
+      "grad_norm": 0.07590368390083313,
+      "learning_rate": 0.00012341772151898734,
+      "loss": 0.1393,
+      "step": 195
+    },
+    {
+      "epoch": 0.18609067173035843,
+      "grad_norm": 0.10585250705480576,
+      "learning_rate": 0.0001240506329113924,
+      "loss": 0.2155,
+      "step": 196
+    },
+    {
+      "epoch": 0.18704011393306433,
+      "grad_norm": 0.06995555013418198,
+      "learning_rate": 0.00012468354430379748,
+      "loss": 0.1374,
+      "step": 197
+    },
+    {
+      "epoch": 0.18798955613577023,
+      "grad_norm": 0.07370735704898834,
+      "learning_rate": 0.00012531645569620255,
+      "loss": 0.1367,
+      "step": 198
+    },
+    {
+      "epoch": 0.18893899833847613,
+      "grad_norm": 0.07194443792104721,
+      "learning_rate": 0.0001259493670886076,
+      "loss": 0.1437,
+      "step": 199
+    },
+    {
+      "epoch": 0.18988844054118206,
+      "grad_norm": 0.06982647627592087,
+      "learning_rate": 0.00012658227848101267,
+      "loss": 0.1358,
+      "step": 200
+    },
+    {
+      "epoch": 0.19083788274388797,
+      "grad_norm": 0.06538347154855728,
+      "learning_rate": 0.0001272151898734177,
+      "loss": 0.1354,
+      "step": 201
+    },
+    {
+      "epoch": 0.19178732494659387,
+      "grad_norm": 0.07789324969053268,
+      "learning_rate": 0.00012784810126582278,
+      "loss": 0.178,
+      "step": 202
+    },
+    {
+      "epoch": 0.1927367671492998,
+      "grad_norm": 0.07376820594072342,
+      "learning_rate": 0.00012848101265822785,
+      "loss": 0.1621,
+      "step": 203
+    },
+    {
+      "epoch": 0.1936862093520057,
+      "grad_norm": 0.0720745250582695,
+      "learning_rate": 0.00012911392405063292,
+      "loss": 0.132,
+      "step": 204
+    },
+    {
+      "epoch": 0.1946356515547116,
+      "grad_norm": 0.06211116537451744,
+      "learning_rate": 0.000129746835443038,
+      "loss": 0.1387,
+      "step": 205
+    },
+    {
+      "epoch": 0.1955850937574175,
+      "grad_norm": 0.06701771914958954,
+      "learning_rate": 0.00013037974683544306,
+      "loss": 0.14,
+      "step": 206
+    },
+    {
+      "epoch": 0.19653453596012344,
+      "grad_norm": 0.07692532986402512,
+      "learning_rate": 0.0001310126582278481,
+      "loss": 0.1322,
+      "step": 207
+    },
+    {
+      "epoch": 0.19748397816282934,
+      "grad_norm": 0.07763269543647766,
+      "learning_rate": 0.00013164556962025315,
+      "loss": 0.1393,
+      "step": 208
+    },
+    {
+      "epoch": 0.19843342036553524,
+      "grad_norm": 0.08769022673368454,
+      "learning_rate": 0.00013227848101265822,
+      "loss": 0.1489,
+      "step": 209
+    },
+    {
+      "epoch": 0.19938286256824117,
+      "grad_norm": 0.08881859481334686,
+      "learning_rate": 0.0001329113924050633,
+      "loss": 0.1765,
+      "step": 210
+    },
+    {
+      "epoch": 0.20033230477094707,
+      "grad_norm": 0.06811822950839996,
+      "learning_rate": 0.00013354430379746836,
+      "loss": 0.1332,
+      "step": 211
+    },
+    {
+      "epoch": 0.20128174697365298,
+      "grad_norm": 0.06390922516584396,
+      "learning_rate": 0.00013417721518987343,
+      "loss": 0.1343,
+      "step": 212
+    },
+    {
+      "epoch": 0.20223118917635888,
+      "grad_norm": 0.06630406528711319,
+      "learning_rate": 0.0001348101265822785,
+      "loss": 0.1329,
+      "step": 213
+    },
+    {
+      "epoch": 0.2031806313790648,
+      "grad_norm": 0.0730772465467453,
+      "learning_rate": 0.00013544303797468355,
+      "loss": 0.1354,
+      "step": 214
+    },
+    {
+      "epoch": 0.2041300735817707,
+      "grad_norm": 0.06487323343753815,
+      "learning_rate": 0.00013607594936708862,
+      "loss": 0.1297,
+      "step": 215
+    },
+    {
+      "epoch": 0.2050795157844766,
+      "grad_norm": 0.06967955082654953,
+      "learning_rate": 0.00013670886075949366,
+      "loss": 0.1398,
+      "step": 216
+    },
+    {
+      "epoch": 0.20602895798718254,
+      "grad_norm": 0.08531820774078369,
+      "learning_rate": 0.00013734177215189873,
+      "loss": 0.1336,
+      "step": 217
+    },
+    {
+      "epoch": 0.20697840018988845,
+      "grad_norm": 0.0757659375667572,
+      "learning_rate": 0.0001379746835443038,
+      "loss": 0.1606,
+      "step": 218
+    },
+    {
+      "epoch": 0.20792784239259435,
+      "grad_norm": 0.060206469148397446,
+      "learning_rate": 0.00013860759493670888,
+      "loss": 0.1337,
+      "step": 219
+    },
+    {
+      "epoch": 0.20887728459530025,
+      "grad_norm": 0.07996556162834167,
+      "learning_rate": 0.00013924050632911395,
+      "loss": 0.1308,
+      "step": 220
+    },
+    {
+      "epoch": 0.20982672679800618,
+      "grad_norm": 0.06206861138343811,
+      "learning_rate": 0.000139873417721519,
+      "loss": 0.1347,
+      "step": 221
+    },
+    {
+      "epoch": 0.21077616900071208,
+      "grad_norm": 0.08736416697502136,
+      "learning_rate": 0.00014050632911392406,
+      "loss": 0.1768,
+      "step": 222
+    },
+    {
+      "epoch": 0.21172561120341798,
+      "grad_norm": 0.06427916139364243,
+      "learning_rate": 0.00014113924050632913,
+      "loss": 0.1374,
+      "step": 223
+    },
+    {
+      "epoch": 0.21267505340612392,
+      "grad_norm": 0.10996536910533905,
+      "learning_rate": 0.00014177215189873418,
+      "loss": 0.222,
+      "step": 224
+    },
+    {
+      "epoch": 0.21362449560882982,
+      "grad_norm": 0.08439125120639801,
+      "learning_rate": 0.00014240506329113925,
+      "loss": 0.1854,
+      "step": 225
+    },
+    {
+      "epoch": 0.21457393781153572,
+      "grad_norm": 0.06892693787813187,
+      "learning_rate": 0.00014303797468354432,
+      "loss": 0.139,
+      "step": 226
+    },
+    {
+      "epoch": 0.21552338001424162,
+      "grad_norm": 0.08241122961044312,
+      "learning_rate": 0.0001436708860759494,
+      "loss": 0.173,
+      "step": 227
+    },
+    {
+      "epoch": 0.21647282221694755,
+      "grad_norm": 0.07911046594381332,
+      "learning_rate": 0.00014430379746835443,
+      "loss": 0.1418,
+      "step": 228
+    },
+    {
+      "epoch": 0.21742226441965345,
+      "grad_norm": 0.06346064805984497,
+      "learning_rate": 0.0001449367088607595,
+      "loss": 0.1406,
+      "step": 229
+    },
+    {
+      "epoch": 0.21837170662235936,
+      "grad_norm": 0.060393668711185455,
+      "learning_rate": 0.00014556962025316457,
+      "loss": 0.1417,
+      "step": 230
+    },
+    {
+      "epoch": 0.2193211488250653,
+      "grad_norm": 0.05912507325410843,
+      "learning_rate": 0.00014620253164556962,
+      "loss": 0.1298,
+      "step": 231
+    },
+    {
+      "epoch": 0.2202705910277712,
+      "grad_norm": 0.07730337232351303,
+      "learning_rate": 0.0001468354430379747,
+      "loss": 0.1769,
+      "step": 232
+    },
+    {
+      "epoch": 0.2212200332304771,
+      "grad_norm": 0.07612381875514984,
+      "learning_rate": 0.00014746835443037976,
+      "loss": 0.1338,
+      "step": 233
+    },
+    {
+      "epoch": 0.222169475433183,
+      "grad_norm": 0.055311791598796844,
+      "learning_rate": 0.0001481012658227848,
+      "loss": 0.1313,
+      "step": 234
+    },
+    {
+      "epoch": 0.22311891763588892,
+      "grad_norm": 0.08492033183574677,
+      "learning_rate": 0.00014873417721518987,
+      "loss": 0.1367,
+      "step": 235
+    },
+    {
+      "epoch": 0.22406835983859483,
+      "grad_norm": 0.07133237272500992,
+      "learning_rate": 0.00014936708860759494,
+      "loss": 0.1308,
+      "step": 236
+    },
+    {
+      "epoch": 0.22501780204130073,
+      "grad_norm": 0.07148605585098267,
+      "learning_rate": 0.00015000000000000001,
+      "loss": 0.133,
+      "step": 237
+    },
+    {
+      "epoch": 0.22596724424400666,
+      "grad_norm": 0.06900472939014435,
+      "learning_rate": 0.00015063291139240508,
+      "loss": 0.138,
+      "step": 238
+    },
+    {
+      "epoch": 0.22691668644671256,
+      "grad_norm": 0.062325432896614075,
+      "learning_rate": 0.00015126582278481013,
+      "loss": 0.1338,
+      "step": 239
+    },
+    {
+      "epoch": 0.22786612864941846,
+      "grad_norm": 0.06719667464494705,
+      "learning_rate": 0.0001518987341772152,
+      "loss": 0.1316,
+      "step": 240
+    },
+    {
+      "epoch": 0.22881557085212437,
+      "grad_norm": 0.07456009089946747,
+      "learning_rate": 0.00015253164556962024,
+      "loss": 0.1412,
+      "step": 241
+    },
+    {
+      "epoch": 0.2297650130548303,
+      "grad_norm": 0.05619575083255768,
+      "learning_rate": 0.00015316455696202531,
+      "loss": 0.1342,
+      "step": 242
+    },
+    {
+      "epoch": 0.2307144552575362,
+      "grad_norm": 0.06157098710536957,
+      "learning_rate": 0.00015379746835443038,
+      "loss": 0.1329,
+      "step": 243
+    },
+    {
+      "epoch": 0.2316638974602421,
+      "grad_norm": 0.06759827584028244,
+      "learning_rate": 0.00015443037974683546,
+      "loss": 0.1411,
+      "step": 244
+    },
+    {
+      "epoch": 0.232613339662948,
+      "grad_norm": 0.06892479956150055,
+      "learning_rate": 0.00015506329113924053,
+      "loss": 0.1484,
+      "step": 245
+    },
+    {
+      "epoch": 0.23356278186565393,
+      "grad_norm": 0.08536699414253235,
+      "learning_rate": 0.0001556962025316456,
+      "loss": 0.1855,
+      "step": 246
+    },
+    {
+      "epoch": 0.23451222406835984,
+      "grad_norm": 0.06800314784049988,
+      "learning_rate": 0.00015632911392405064,
+      "loss": 0.1379,
+      "step": 247
+    },
+    {
+      "epoch": 0.23546166627106574,
+      "grad_norm": 0.0625622496008873,
+      "learning_rate": 0.00015696202531645568,
+      "loss": 0.1344,
+      "step": 248
+    },
+    {
+      "epoch": 0.23641110847377167,
+      "grad_norm": 0.06030593812465668,
+      "learning_rate": 0.00015759493670886075,
+      "loss": 0.1254,
+      "step": 249
+    },
+    {
+      "epoch": 0.23736055067647757,
+      "grad_norm": 0.06694353371858597,
+      "learning_rate": 0.00015822784810126583,
+      "loss": 0.1413,
+      "step": 250
+    },
+    {
+      "epoch": 0.23830999287918347,
+      "grad_norm": 0.06594134122133255,
+      "learning_rate": 0.0001588607594936709,
+      "loss": 0.1394,
+      "step": 251
+    },
+    {
+      "epoch": 0.23925943508188938,
+      "grad_norm": 0.09062930941581726,
+      "learning_rate": 0.00015949367088607597,
+      "loss": 0.1883,
+      "step": 252
+    },
+    {
+      "epoch": 0.2402088772845953,
+      "grad_norm": 0.06029089167714119,
+      "learning_rate": 0.00016012658227848104,
+      "loss": 0.1271,
+      "step": 253
+    },
+    {
+      "epoch": 0.2411583194873012,
+      "grad_norm": 0.08471622318029404,
+      "learning_rate": 0.00016075949367088608,
+      "loss": 0.172,
+      "step": 254
+    },
+    {
+      "epoch": 0.2421077616900071,
+      "grad_norm": 0.061710160225629807,
+      "learning_rate": 0.00016139240506329115,
+      "loss": 0.1348,
+      "step": 255
+    },
+    {
+      "epoch": 0.24305720389271304,
+      "grad_norm": 0.0812671035528183,
+      "learning_rate": 0.0001620253164556962,
+      "loss": 0.1312,
+      "step": 256
+    },
+    {
+      "epoch": 0.24400664609541894,
+      "grad_norm": 0.06917005032300949,
+      "learning_rate": 0.00016265822784810127,
+      "loss": 0.1464,
+      "step": 257
+    },
+    {
+      "epoch": 0.24495608829812485,
+      "grad_norm": 0.0905887708067894,
+      "learning_rate": 0.00016329113924050634,
+      "loss": 0.1759,
+      "step": 258
+    },
+    {
+      "epoch": 0.24590553050083075,
+      "grad_norm": 0.05976787209510803,
+      "learning_rate": 0.0001639240506329114,
+      "loss": 0.1404,
+      "step": 259
+    },
+    {
+      "epoch": 0.24685497270353668,
+      "grad_norm": 0.07545675337314606,
+      "learning_rate": 0.00016455696202531648,
+      "loss": 0.1322,
+      "step": 260
+    },
+    {
+      "epoch": 0.24780441490624258,
+      "grad_norm": 0.07035024464130402,
+      "learning_rate": 0.00016518987341772152,
+      "loss": 0.1378,
+      "step": 261
+    },
+    {
+      "epoch": 0.24875385710894848,
+      "grad_norm": 0.07665737718343735,
+      "learning_rate": 0.0001658227848101266,
+      "loss": 0.1827,
+      "step": 262
+    },
+    {
+      "epoch": 0.24970329931165441,
+      "grad_norm": 0.06619013845920563,
+      "learning_rate": 0.00016645569620253166,
+      "loss": 0.1284,
+      "step": 263
+    },
+    {
+      "epoch": 0.2506527415143603,
+      "grad_norm": 0.0647001713514328,
+      "learning_rate": 0.0001670886075949367,
+      "loss": 0.133,
+      "step": 264
+    },
+    {
+      "epoch": 0.2516021837170662,
+      "grad_norm": 0.060702718794345856,
+      "learning_rate": 0.00016772151898734178,
+      "loss": 0.1335,
+      "step": 265
+    },
+    {
+      "epoch": 0.25255162591977215,
+      "grad_norm": 0.0508468896150589,
+      "learning_rate": 0.00016835443037974685,
+      "loss": 0.1333,
+      "step": 266
+    },
+    {
+      "epoch": 0.253501068122478,
+      "grad_norm": 0.09877864271402359,
+      "learning_rate": 0.0001689873417721519,
+      "loss": 0.2031,
+      "step": 267
+    },
+    {
+      "epoch": 0.25445051032518395,
+      "grad_norm": 0.06673337519168854,
+      "learning_rate": 0.00016962025316455696,
+      "loss": 0.1356,
+      "step": 268
+    },
+    {
+      "epoch": 0.2553999525278899,
+      "grad_norm": 0.10604165494441986,
+      "learning_rate": 0.00017025316455696204,
+      "loss": 0.2517,
+      "step": 269
+    },
+    {
+      "epoch": 0.25634939473059576,
+      "grad_norm": 0.07689858227968216,
+      "learning_rate": 0.0001708860759493671,
+      "loss": 0.1761,
+      "step": 270
+    },
+    {
+      "epoch": 0.2572988369333017,
+      "grad_norm": 0.05482449755072594,
+      "learning_rate": 0.00017151898734177218,
+      "loss": 0.131,
+      "step": 271
+    },
+    {
+      "epoch": 0.2582482791360076,
+      "grad_norm": 0.08622145652770996,
+      "learning_rate": 0.00017215189873417722,
+      "loss": 0.1335,
+      "step": 272
+    },
+    {
+      "epoch": 0.2591977213387135,
+      "grad_norm": 0.0748213455080986,
+      "learning_rate": 0.0001727848101265823,
+      "loss": 0.176,
+      "step": 273
+    },
+    {
+      "epoch": 0.2601471635414194,
+      "grad_norm": 0.06163305416703224,
+      "learning_rate": 0.00017341772151898733,
+      "loss": 0.1381,
+      "step": 274
+    },
+    {
+      "epoch": 0.26109660574412535,
+      "grad_norm": 0.06141841039061546,
+      "learning_rate": 0.0001740506329113924,
+      "loss": 0.1353,
+      "step": 275
+    },
+    {
+      "epoch": 0.26204604794683123,
+      "grad_norm": 0.07326913625001907,
+      "learning_rate": 0.00017468354430379748,
+      "loss": 0.1441,
+      "step": 276
+    },
+    {
+      "epoch": 0.26299549014953716,
+      "grad_norm": 0.05951124057173729,
+      "learning_rate": 0.00017531645569620255,
+      "loss": 0.1321,
+      "step": 277
+    },
+    {
+      "epoch": 0.26394493235224303,
+      "grad_norm": 0.08364073932170868,
+      "learning_rate": 0.00017594936708860762,
+      "loss": 0.187,
+      "step": 278
+    },
+    {
+      "epoch": 0.26489437455494896,
+      "grad_norm": 0.05849132314324379,
+      "learning_rate": 0.00017658227848101266,
+      "loss": 0.1393,
+      "step": 279
+    },
+    {
+      "epoch": 0.2658438167576549,
+      "grad_norm": 0.05452360957860947,
+      "learning_rate": 0.00017721518987341773,
+      "loss": 0.1342,
+      "step": 280
+    },
+    {
+      "epoch": 0.26679325896036077,
+      "grad_norm": 0.04878188297152519,
+      "learning_rate": 0.00017784810126582278,
+      "loss": 0.1445,
+      "step": 281
+    },
+    {
+      "epoch": 0.2677427011630667,
+      "grad_norm": 0.06066753342747688,
+      "learning_rate": 0.00017848101265822785,
+      "loss": 0.1423,
+      "step": 282
+    },
+    {
+      "epoch": 0.26869214336577263,
+      "grad_norm": 0.04918207973241806,
+      "learning_rate": 0.00017911392405063292,
+      "loss": 0.1316,
+      "step": 283
+    },
+    {
+      "epoch": 0.2696415855684785,
+      "grad_norm": 0.05103525519371033,
+      "learning_rate": 0.000179746835443038,
+      "loss": 0.1313,
+      "step": 284
+    },
+    {
+      "epoch": 0.27059102777118443,
+      "grad_norm": 0.05667628347873688,
+      "learning_rate": 0.00018037974683544306,
+      "loss": 0.1434,
+      "step": 285
+    },
+    {
+      "epoch": 0.27154046997389036,
+      "grad_norm": 0.06226016581058502,
+      "learning_rate": 0.00018101265822784813,
+      "loss": 0.1357,
+      "step": 286
+    },
+    {
+      "epoch": 0.27248991217659624,
+      "grad_norm": 0.04695293679833412,
+      "learning_rate": 0.00018164556962025317,
+      "loss": 0.1314,
+      "step": 287
+    },
+    {
+      "epoch": 0.27343935437930217,
+      "grad_norm": 0.05762844532728195,
+      "learning_rate": 0.00018227848101265824,
+      "loss": 0.1349,
+      "step": 288
+    },
+    {
+      "epoch": 0.27438879658200804,
+      "grad_norm": 0.05454534292221069,
+      "learning_rate": 0.0001829113924050633,
+      "loss": 0.1432,
+      "step": 289
+    },
+    {
+      "epoch": 0.275338238784714,
+      "grad_norm": 0.050270579755306244,
+      "learning_rate": 0.00018354430379746836,
+      "loss": 0.1272,
+      "step": 290
+    },
+    {
+      "epoch": 0.2762876809874199,
+      "grad_norm": 0.0688452497124672,
+      "learning_rate": 0.00018417721518987343,
+      "loss": 0.1708,
+      "step": 291
+    },
+    {
+      "epoch": 0.2772371231901258,
+      "grad_norm": 0.06213200092315674,
+      "learning_rate": 0.0001848101265822785,
+      "loss": 0.1674,
+      "step": 292
+    },
+    {
+      "epoch": 0.2781865653928317,
+      "grad_norm": 0.059717319905757904,
+      "learning_rate": 0.00018544303797468354,
+      "loss": 0.169,
+      "step": 293
+    },
+    {
+      "epoch": 0.27913600759553764,
+      "grad_norm": 0.06223325803875923,
+      "learning_rate": 0.00018607594936708861,
+      "loss": 0.1369,
+      "step": 294
+    },
+    {
+      "epoch": 0.2800854497982435,
+      "grad_norm": 0.053163208067417145,
+      "learning_rate": 0.00018670886075949369,
+      "loss": 0.133,
+      "step": 295
+    },
+    {
+      "epoch": 0.28103489200094944,
+      "grad_norm": 0.06647945195436478,
+      "learning_rate": 0.00018734177215189873,
+      "loss": 0.1438,
+      "step": 296
+    },
+    {
+      "epoch": 0.2819843342036554,
+      "grad_norm": 0.0588272288441658,
+      "learning_rate": 0.0001879746835443038,
+      "loss": 0.1338,
+      "step": 297
+    },
+    {
+      "epoch": 0.28293377640636125,
+      "grad_norm": 0.05841274932026863,
+      "learning_rate": 0.00018860759493670887,
+      "loss": 0.1329,
+      "step": 298
+    },
+    {
+      "epoch": 0.2838832186090672,
+      "grad_norm": 0.09033369272947311,
+      "learning_rate": 0.00018924050632911394,
+      "loss": 0.1747,
+      "step": 299
+    },
+    {
+      "epoch": 0.2848326608117731,
+      "grad_norm": 0.052215326577425,
+      "learning_rate": 0.00018987341772151899,
+      "loss": 0.1296,
+      "step": 300
+    },
+    {
+      "epoch": 0.285782103014479,
+      "grad_norm": 0.05880101025104523,
+      "learning_rate": 0.00019050632911392406,
+      "loss": 0.1287,
+      "step": 301
+    },
+    {
+      "epoch": 0.2867315452171849,
+      "grad_norm": 0.0691700354218483,
+      "learning_rate": 0.00019113924050632913,
+      "loss": 0.1676,
+      "step": 302
+    },
+    {
+      "epoch": 0.2876809874198908,
+      "grad_norm": 0.057025909423828125,
+      "learning_rate": 0.0001917721518987342,
+      "loss": 0.1346,
+      "step": 303
+    },
+    {
+      "epoch": 0.2886304296225967,
+      "grad_norm": 0.04936329275369644,
+      "learning_rate": 0.00019240506329113924,
+      "loss": 0.1354,
+      "step": 304
+    },
+    {
+      "epoch": 0.28957987182530265,
+      "grad_norm": 0.0680055245757103,
+      "learning_rate": 0.0001930379746835443,
+      "loss": 0.1344,
+      "step": 305
+    },
+    {
+      "epoch": 0.2905293140280085,
+      "grad_norm": 0.07374466210603714,
+      "learning_rate": 0.00019367088607594938,
+      "loss": 0.1428,
+      "step": 306
+    },
+    {
+      "epoch": 0.29147875623071445,
+      "grad_norm": 0.061204761266708374,
+      "learning_rate": 0.00019430379746835443,
+      "loss": 0.1246,
+      "step": 307
+    },
+    {
+      "epoch": 0.2924281984334204,
+      "grad_norm": 0.053467705845832825,
+      "learning_rate": 0.0001949367088607595,
+      "loss": 0.1342,
+      "step": 308
+    },
+    {
+      "epoch": 0.29337764063612626,
+      "grad_norm": 0.057525087147951126,
+      "learning_rate": 0.00019556962025316457,
+      "loss": 0.1377,
+      "step": 309
+    },
+    {
+      "epoch": 0.2943270828388322,
+      "grad_norm": 0.07857844978570938,
+      "learning_rate": 0.00019620253164556964,
+      "loss": 0.2076,
+      "step": 310
+    },
+    {
+      "epoch": 0.2952765250415381,
+      "grad_norm": 0.05250545218586922,
+      "learning_rate": 0.0001968354430379747,
+      "loss": 0.1432,
+      "step": 311
+    },
+    {
+      "epoch": 0.296225967244244,
+      "grad_norm": 0.07495012134313583,
+      "learning_rate": 0.00019746835443037975,
+      "loss": 0.1766,
+      "step": 312
+    },
+    {
+      "epoch": 0.2971754094469499,
+      "grad_norm": 0.04692578688263893,
+      "learning_rate": 0.0001981012658227848,
+      "loss": 0.1408,
+      "step": 313
+    },
+    {
+      "epoch": 0.29812485164965585,
+      "grad_norm": 0.055666085332632065,
+      "learning_rate": 0.00019873417721518987,
+      "loss": 0.1391,
+      "step": 314
+    },
+    {
+      "epoch": 0.29907429385236173,
+      "grad_norm": 0.050465911626815796,
+      "learning_rate": 0.00019936708860759494,
+      "loss": 0.1415,
+      "step": 315
+    },
+    {
+      "epoch": 0.30002373605506766,
+      "grad_norm": 0.051260240375995636,
+      "learning_rate": 0.0002,
+      "loss": 0.1423,
+      "step": 316
+    },
+    {
+      "epoch": 0.30097317825777353,
+      "grad_norm": 0.0503215529024601,
+      "learning_rate": 0.000199999938945738,
+      "loss": 0.1348,
+      "step": 317
+    },
+    {
+      "epoch": 0.30192262046047946,
+      "grad_norm": 0.04917483776807785,
+      "learning_rate": 0.0001999997557830265,
+      "loss": 0.1342,
+      "step": 318
+    },
+    {
+      "epoch": 0.3028720626631854,
+      "grad_norm": 0.06354209035634995,
+      "learning_rate": 0.00019999945051208916,
+      "loss": 0.1365,
+      "step": 319
+    },
+    {
+      "epoch": 0.30382150486589127,
+      "grad_norm": 0.04878314957022667,
+      "learning_rate": 0.0001999990231332988,
+      "loss": 0.13,
+      "step": 320
+    },
+    {
+      "epoch": 0.3047709470685972,
+      "grad_norm": 0.07046223431825638,
+      "learning_rate": 0.0001999984736471772,
+      "loss": 0.1394,
+      "step": 321
+    },
+    {
+      "epoch": 0.30572038927130313,
+      "grad_norm": 0.04456232488155365,
+      "learning_rate": 0.00019999780205439538,
+      "loss": 0.1278,
+      "step": 322
+    },
+    {
+      "epoch": 0.306669831474009,
+      "grad_norm": 0.06280628591775894,
+      "learning_rate": 0.00019999700835577342,
+      "loss": 0.1715,
+      "step": 323
+    },
+    {
+      "epoch": 0.30761927367671493,
+      "grad_norm": 0.07462131977081299,
+      "learning_rate": 0.00019999609255228046,
+      "loss": 0.1772,
+      "step": 324
+    },
+    {
+      "epoch": 0.30856871587942086,
+      "grad_norm": 0.059642352163791656,
+      "learning_rate": 0.00019999505464503482,
+      "loss": 0.1294,
+      "step": 325
+    },
+    {
+      "epoch": 0.30951815808212674,
+      "grad_norm": 0.06458820402622223,
+      "learning_rate": 0.00019999389463530383,
+      "loss": 0.173,
+      "step": 326
+    },
+    {
+      "epoch": 0.31046760028483267,
+      "grad_norm": 0.05901939421892166,
+      "learning_rate": 0.00019999261252450396,
+      "loss": 0.1419,
+      "step": 327
+    },
+    {
+      "epoch": 0.3114170424875386,
+      "grad_norm": 0.055540215224027634,
+      "learning_rate": 0.00019999120831420083,
+      "loss": 0.1314,
+      "step": 328
+    },
+    {
+      "epoch": 0.3123664846902445,
+      "grad_norm": 0.0546739287674427,
+      "learning_rate": 0.00019998968200610903,
+      "loss": 0.1354,
+      "step": 329
+    },
+    {
+      "epoch": 0.3133159268929504,
+      "grad_norm": 0.0689477026462555,
+      "learning_rate": 0.00019998803360209234,
+      "loss": 0.132,
+      "step": 330
+    },
+    {
+      "epoch": 0.3142653690956563,
+      "grad_norm": 0.05279696360230446,
+      "learning_rate": 0.00019998626310416365,
+      "loss": 0.1424,
+      "step": 331
+    },
+    {
+      "epoch": 0.3152148112983622,
+      "grad_norm": 0.055384278297424316,
+      "learning_rate": 0.00019998437051448482,
+      "loss": 0.141,
+      "step": 332
+    },
+    {
+      "epoch": 0.31616425350106814,
+      "grad_norm": 0.04636182263493538,
+      "learning_rate": 0.0001999823558353669,
+      "loss": 0.1414,
+      "step": 333
+    },
+    {
+      "epoch": 0.317113695703774,
+      "grad_norm": 0.04795726016163826,
+      "learning_rate": 0.00019998021906926993,
+      "loss": 0.1255,
+      "step": 334
+    },
+    {
+      "epoch": 0.31806313790647994,
+      "grad_norm": 0.05326540395617485,
+      "learning_rate": 0.00019997796021880318,
+      "loss": 0.1309,
+      "step": 335
+    },
+    {
+      "epoch": 0.3190125801091859,
+      "grad_norm": 0.0684736892580986,
+      "learning_rate": 0.00019997557928672484,
+      "loss": 0.1825,
+      "step": 336
+    },
+    {
+      "epoch": 0.31996202231189175,
+      "grad_norm": 0.042282164096832275,
+      "learning_rate": 0.0001999730762759422,
+      "loss": 0.12,
+      "step": 337
+    },
+    {
+      "epoch": 0.3209114645145977,
+      "grad_norm": 0.05297423154115677,
+      "learning_rate": 0.00019997045118951175,
+      "loss": 0.1309,
+      "step": 338
+    },
+    {
+      "epoch": 0.3218609067173036,
+      "grad_norm": 0.080621138215065,
+      "learning_rate": 0.00019996770403063883,
+      "loss": 0.2134,
+      "step": 339
+    },
+    {
+      "epoch": 0.3228103489200095,
+      "grad_norm": 0.05552308261394501,
+      "learning_rate": 0.00019996483480267803,
+      "loss": 0.1361,
+      "step": 340
+    },
+    {
+      "epoch": 0.3237597911227154,
+      "grad_norm": 0.05070111155509949,
+      "learning_rate": 0.00019996184350913287,
+      "loss": 0.1314,
+      "step": 341
+    },
+    {
+      "epoch": 0.32470923332542134,
+      "grad_norm": 0.04412266984581947,
+      "learning_rate": 0.00019995873015365601,
+      "loss": 0.1299,
+      "step": 342
+    },
+    {
+      "epoch": 0.3256586755281272,
+      "grad_norm": 0.0445338599383831,
+      "learning_rate": 0.00019995549474004917,
+      "loss": 0.1313,
+      "step": 343
+    },
+    {
+      "epoch": 0.32660811773083315,
+      "grad_norm": 0.08224980533123016,
+      "learning_rate": 0.000199952137272263,
+      "loss": 0.1844,
+      "step": 344
+    },
+    {
+      "epoch": 0.327557559933539,
+      "grad_norm": 0.04331446811556816,
+      "learning_rate": 0.0001999486577543972,
+      "loss": 0.133,
+      "step": 345
+    },
+    {
+      "epoch": 0.32850700213624495,
+      "grad_norm": 0.049314577132463455,
+      "learning_rate": 0.00019994505619070068,
+      "loss": 0.1351,
+      "step": 346
+    },
+    {
+      "epoch": 0.3294564443389509,
+      "grad_norm": 0.0697011798620224,
+      "learning_rate": 0.00019994133258557117,
+      "loss": 0.1709,
+      "step": 347
+    },
+    {
+      "epoch": 0.33040588654165676,
+      "grad_norm": 0.0510990135371685,
+      "learning_rate": 0.00019993748694355557,
+      "loss": 0.1365,
+      "step": 348
+    },
+    {
+      "epoch": 0.3313553287443627,
+      "grad_norm": 0.05100785568356514,
+      "learning_rate": 0.00019993351926934967,
+      "loss": 0.1302,
+      "step": 349
+    },
+    {
+      "epoch": 0.3323047709470686,
+      "grad_norm": 0.08001980185508728,
+      "learning_rate": 0.00019992942956779838,
+      "loss": 0.1736,
+      "step": 350
+    },
+    {
+      "epoch": 0.3332542131497745,
+      "grad_norm": 0.05298507958650589,
+      "learning_rate": 0.00019992521784389559,
+      "loss": 0.159,
+      "step": 351
+    },
+    {
+      "epoch": 0.3342036553524804,
+      "grad_norm": 0.04655485600233078,
+      "learning_rate": 0.00019992088410278414,
+      "loss": 0.1401,
+      "step": 352
+    },
+    {
+      "epoch": 0.33515309755518635,
+      "grad_norm": 0.047509439289569855,
+      "learning_rate": 0.00019991642834975594,
+      "loss": 0.1369,
+      "step": 353
+    },
+    {
+      "epoch": 0.3361025397578922,
+      "grad_norm": 0.046006906777620316,
+      "learning_rate": 0.0001999118505902518,
+      "loss": 0.1384,
+      "step": 354
+    },
+    {
+      "epoch": 0.33705198196059816,
+      "grad_norm": 0.07522892951965332,
+      "learning_rate": 0.00019990715082986155,
+      "loss": 0.2254,
+      "step": 355
+    },
+    {
+      "epoch": 0.33800142416330403,
+      "grad_norm": 0.048646144568920135,
+      "learning_rate": 0.00019990232907432404,
+      "loss": 0.1355,
+      "step": 356
+    },
+    {
+      "epoch": 0.33895086636600996,
+      "grad_norm": 0.03941798582673073,
+      "learning_rate": 0.000199897385329527,
+      "loss": 0.1242,
+      "step": 357
+    },
+    {
+      "epoch": 0.3399003085687159,
+      "grad_norm": 0.04582727700471878,
+      "learning_rate": 0.0001998923196015072,
+      "loss": 0.1347,
+      "step": 358
+    },
+    {
+      "epoch": 0.34084975077142177,
+      "grad_norm": 0.05890033766627312,
+      "learning_rate": 0.00019988713189645027,
+      "loss": 0.1356,
+      "step": 359
+    },
+    {
+      "epoch": 0.3417991929741277,
+      "grad_norm": 0.050398606806993484,
+      "learning_rate": 0.00019988182222069093,
+      "loss": 0.1379,
+      "step": 360
+    },
+    {
+      "epoch": 0.3427486351768336,
+      "grad_norm": 0.053657352924346924,
+      "learning_rate": 0.00019987639058071267,
+      "loss": 0.1417,
+      "step": 361
+    },
+    {
+      "epoch": 0.3436980773795395,
+      "grad_norm": 0.04928993433713913,
+      "learning_rate": 0.00019987083698314804,
+      "loss": 0.1269,
+      "step": 362
+    },
+    {
+      "epoch": 0.34464751958224543,
+      "grad_norm": 0.04932550713419914,
+      "learning_rate": 0.0001998651614347784,
+      "loss": 0.1429,
+      "step": 363
+    },
+    {
+      "epoch": 0.34559696178495136,
+      "grad_norm": 0.0531768873333931,
+      "learning_rate": 0.00019985936394253413,
+      "loss": 0.1367,
+      "step": 364
+    },
+    {
+      "epoch": 0.34654640398765724,
+      "grad_norm": 0.05342009291052818,
+      "learning_rate": 0.00019985344451349443,
+      "loss": 0.1365,
+      "step": 365
+    },
+    {
+      "epoch": 0.34749584619036317,
+      "grad_norm": 0.04960772022604942,
+      "learning_rate": 0.00019984740315488742,
+      "loss": 0.133,
+      "step": 366
+    },
+    {
+      "epoch": 0.3484452883930691,
+      "grad_norm": 0.04490765556693077,
+      "learning_rate": 0.00019984123987409013,
+      "loss": 0.1347,
+      "step": 367
+    },
+    {
+      "epoch": 0.34939473059577497,
+      "grad_norm": 0.05546121671795845,
+      "learning_rate": 0.0001998349546786285,
+      "loss": 0.169,
+      "step": 368
+    },
+    {
+      "epoch": 0.3503441727984809,
+      "grad_norm": 0.04962169751524925,
+      "learning_rate": 0.0001998285475761772,
+      "loss": 0.1325,
+      "step": 369
+    },
+    {
+      "epoch": 0.3512936150011868,
+      "grad_norm": 0.0451858825981617,
+      "learning_rate": 0.00019982201857455988,
+      "loss": 0.1291,
+      "step": 370
+    },
+    {
+      "epoch": 0.3522430572038927,
+      "grad_norm": 0.07738906145095825,
+      "learning_rate": 0.00019981536768174903,
+      "loss": 0.1841,
+      "step": 371
+    },
+    {
+      "epoch": 0.35319249940659864,
+      "grad_norm": 0.05104148015379906,
+      "learning_rate": 0.000199808594905866,
+      "loss": 0.1375,
+      "step": 372
+    },
+    {
+      "epoch": 0.3541419416093045,
+      "grad_norm": 0.04850155860185623,
+      "learning_rate": 0.00019980170025518082,
+      "loss": 0.1335,
+      "step": 373
+    },
+    {
+      "epoch": 0.35509138381201044,
+      "grad_norm": 0.050271324813365936,
+      "learning_rate": 0.00019979468373811248,
+      "loss": 0.1394,
+      "step": 374
+    },
+    {
+      "epoch": 0.35604082601471637,
+      "grad_norm": 0.050799645483493805,
+      "learning_rate": 0.0001997875453632288,
+      "loss": 0.135,
+      "step": 375
+    },
+    {
+      "epoch": 0.35699026821742225,
+      "grad_norm": 0.05703526735305786,
+      "learning_rate": 0.00019978028513924627,
+      "loss": 0.1371,
+      "step": 376
+    },
+    {
+      "epoch": 0.3579397104201282,
+      "grad_norm": 0.06665853410959244,
+      "learning_rate": 0.00019977290307503028,
+      "loss": 0.1837,
+      "step": 377
+    },
+    {
+      "epoch": 0.3588891526228341,
+      "grad_norm": 0.04639972746372223,
+      "learning_rate": 0.000199765399179595,
+      "loss": 0.1315,
+      "step": 378
+    },
+    {
+      "epoch": 0.35983859482554,
+      "grad_norm": 0.07625308632850647,
+      "learning_rate": 0.00019975777346210326,
+      "loss": 0.2064,
+      "step": 379
+    },
+    {
+      "epoch": 0.3607880370282459,
+      "grad_norm": 0.048770248889923096,
+      "learning_rate": 0.00019975002593186674,
+      "loss": 0.1363,
+      "step": 380
+    },
+    {
+      "epoch": 0.36173747923095184,
+      "grad_norm": 0.04932136833667755,
+      "learning_rate": 0.00019974215659834582,
+      "loss": 0.1374,
+      "step": 381
+    },
+    {
+      "epoch": 0.3626869214336577,
+      "grad_norm": 0.03848756104707718,
+      "learning_rate": 0.00019973416547114964,
+      "loss": 0.1333,
+      "step": 382
+    },
+    {
+      "epoch": 0.36363636363636365,
+      "grad_norm": 0.04468891769647598,
+      "learning_rate": 0.00019972605256003605,
+      "loss": 0.129,
+      "step": 383
+    },
+    {
+      "epoch": 0.3645858058390695,
+      "grad_norm": 0.048413511365652084,
+      "learning_rate": 0.0001997178178749116,
+      "loss": 0.1314,
+      "step": 384
+    },
+    {
+      "epoch": 0.36553524804177545,
+      "grad_norm": 0.045054856687784195,
+      "learning_rate": 0.00019970946142583155,
+      "loss": 0.1323,
+      "step": 385
+    },
+    {
+      "epoch": 0.3664846902444814,
+      "grad_norm": 0.05541200935840607,
+      "learning_rate": 0.00019970098322299982,
+      "loss": 0.1342,
+      "step": 386
+    },
+    {
+      "epoch": 0.36743413244718726,
+      "grad_norm": 0.06861472874879837,
+      "learning_rate": 0.00019969238327676906,
+      "loss": 0.1347,
+      "step": 387
+    },
+    {
+      "epoch": 0.3683835746498932,
+      "grad_norm": 0.043996453285217285,
+      "learning_rate": 0.00019968366159764047,
+      "loss": 0.132,
+      "step": 388
+    },
+    {
+      "epoch": 0.3693330168525991,
+      "grad_norm": 0.06562239676713943,
+      "learning_rate": 0.000199674818196264,
+      "loss": 0.1759,
+      "step": 389
+    },
+    {
+      "epoch": 0.370282459055305,
+      "grad_norm": 0.04714899882674217,
+      "learning_rate": 0.00019966585308343822,
+      "loss": 0.1274,
+      "step": 390
+    },
+    {
+      "epoch": 0.3712319012580109,
+      "grad_norm": 0.04736959934234619,
+      "learning_rate": 0.00019965676627011026,
+      "loss": 0.1265,
+      "step": 391
+    },
+    {
+      "epoch": 0.37218134346071685,
+      "grad_norm": 0.056829433888196945,
+      "learning_rate": 0.0001996475577673759,
+      "loss": 0.1402,
+      "step": 392
+    },
+    {
+      "epoch": 0.3731307856634227,
+      "grad_norm": 0.0426231250166893,
+      "learning_rate": 0.00019963822758647953,
+      "loss": 0.1364,
+      "step": 393
+    },
+    {
+      "epoch": 0.37408022786612866,
+      "grad_norm": 0.07376877963542938,
+      "learning_rate": 0.00019962877573881404,
+      "loss": 0.2042,
+      "step": 394
+    },
+    {
+      "epoch": 0.3750296700688346,
+      "grad_norm": 0.043273668736219406,
+      "learning_rate": 0.00019961920223592104,
+      "loss": 0.132,
+      "step": 395
+    },
+    {
+      "epoch": 0.37597911227154046,
+      "grad_norm": 0.044406965374946594,
+      "learning_rate": 0.00019960950708949052,
+      "loss": 0.1344,
+      "step": 396
+    },
+    {
+      "epoch": 0.3769285544742464,
+      "grad_norm": 0.040342606604099274,
+      "learning_rate": 0.00019959969031136106,
+      "loss": 0.1214,
+      "step": 397
+    },
+    {
+      "epoch": 0.37787799667695227,
+      "grad_norm": 0.05118388682603836,
+      "learning_rate": 0.00019958975191351983,
+      "loss": 0.14,
+      "step": 398
+    },
+    {
+      "epoch": 0.3788274388796582,
+      "grad_norm": 0.045876793563365936,
+      "learning_rate": 0.00019957969190810245,
+      "loss": 0.1335,
+      "step": 399
+    },
+    {
+      "epoch": 0.3797768810823641,
+      "grad_norm": 0.0645332932472229,
+      "learning_rate": 0.00019956951030739308,
+      "loss": 0.1702,
+      "step": 400
+    },
+    {
+      "epoch": 0.38072632328507,
+      "grad_norm": 0.05039132386445999,
+      "learning_rate": 0.00019955920712382423,
+      "loss": 0.136,
+      "step": 401
+    },
+    {
+      "epoch": 0.38167576548777593,
+      "grad_norm": 0.052004653960466385,
+      "learning_rate": 0.00019954878236997704,
+      "loss": 0.1386,
+      "step": 402
+    },
+    {
+      "epoch": 0.38262520769048186,
+      "grad_norm": 0.05021458491683006,
+      "learning_rate": 0.00019953823605858105,
+      "loss": 0.1378,
+      "step": 403
+    },
+    {
+      "epoch": 0.38357464989318774,
+      "grad_norm": 0.058653559535741806,
+      "learning_rate": 0.0001995275682025141,
+      "loss": 0.1437,
+      "step": 404
+    },
+    {
+      "epoch": 0.38452409209589367,
+      "grad_norm": 0.04466673359274864,
+      "learning_rate": 0.00019951677881480264,
+      "loss": 0.1334,
+      "step": 405
+    },
+    {
+      "epoch": 0.3854735342985996,
+      "grad_norm": 0.06119415909051895,
+      "learning_rate": 0.00019950586790862138,
+      "loss": 0.1296,
+      "step": 406
+    },
+    {
+      "epoch": 0.38642297650130547,
+      "grad_norm": 0.04749077931046486,
+      "learning_rate": 0.0001994948354972935,
+      "loss": 0.1341,
+      "step": 407
+    },
+    {
+      "epoch": 0.3873724187040114,
+      "grad_norm": 0.037752799689769745,
+      "learning_rate": 0.00019948368159429053,
+      "loss": 0.134,
+      "step": 408
+    },
+    {
+      "epoch": 0.38832186090671733,
+      "grad_norm": 0.08903038501739502,
+      "learning_rate": 0.00019947240621323226,
+      "loss": 0.2155,
+      "step": 409
+    },
+    {
+      "epoch": 0.3892713031094232,
+      "grad_norm": 0.03878140076994896,
+      "learning_rate": 0.00019946100936788698,
+      "loss": 0.1176,
+      "step": 410
+    },
+    {
+      "epoch": 0.39022074531212914,
+      "grad_norm": 0.04927309602499008,
+      "learning_rate": 0.00019944949107217113,
+      "loss": 0.1344,
+      "step": 411
+    },
+    {
+      "epoch": 0.391170187514835,
+      "grad_norm": 0.04933890327811241,
+      "learning_rate": 0.00019943785134014962,
+      "loss": 0.1315,
+      "step": 412
+    },
+    {
+      "epoch": 0.39211962971754094,
+      "grad_norm": 0.06702516227960587,
+      "learning_rate": 0.0001994260901860355,
+      "loss": 0.1826,
+      "step": 413
+    },
+    {
+      "epoch": 0.39306907192024687,
+      "grad_norm": 0.048132237046957016,
+      "learning_rate": 0.00019941420762419014,
+      "loss": 0.1436,
+      "step": 414
+    },
+    {
+      "epoch": 0.39401851412295275,
+      "grad_norm": 0.07756894826889038,
+      "learning_rate": 0.00019940220366912318,
+      "loss": 0.2162,
+      "step": 415
+    },
+    {
+      "epoch": 0.3949679563256587,
+      "grad_norm": 0.04789011925458908,
+      "learning_rate": 0.00019939007833549242,
+      "loss": 0.1295,
+      "step": 416
+    },
+    {
+      "epoch": 0.3959173985283646,
+      "grad_norm": 0.04369444027543068,
+      "learning_rate": 0.000199377831638104,
+      "loss": 0.1322,
+      "step": 417
+    },
+    {
+      "epoch": 0.3968668407310705,
+      "grad_norm": 0.05376122146844864,
+      "learning_rate": 0.00019936546359191216,
+      "loss": 0.1743,
+      "step": 418
+    },
+    {
+      "epoch": 0.3978162829337764,
+      "grad_norm": 0.045930229127407074,
+      "learning_rate": 0.0001993529742120193,
+      "loss": 0.1336,
+      "step": 419
+    },
+    {
+      "epoch": 0.39876572513648234,
+      "grad_norm": 0.039980966597795486,
+      "learning_rate": 0.00019934036351367606,
+      "loss": 0.1349,
+      "step": 420
+    },
+    {
+      "epoch": 0.3997151673391882,
+      "grad_norm": 0.03797341510653496,
+      "learning_rate": 0.00019932763151228115,
+      "loss": 0.1256,
+      "step": 421
+    },
+    {
+      "epoch": 0.40066460954189415,
+      "grad_norm": 0.04779914394021034,
+      "learning_rate": 0.00019931477822338146,
+      "loss": 0.1411,
+      "step": 422
+    },
+    {
+      "epoch": 0.4016140517446,
+      "grad_norm": 0.040458668023347855,
+      "learning_rate": 0.00019930180366267193,
+      "loss": 0.126,
+      "step": 423
+    },
+    {
+      "epoch": 0.40256349394730595,
+      "grad_norm": 0.04114462807774544,
+      "learning_rate": 0.0001992887078459956,
+      "loss": 0.127,
+      "step": 424
+    },
+    {
+      "epoch": 0.4035129361500119,
+      "grad_norm": 0.048119012266397476,
+      "learning_rate": 0.00019927549078934358,
+      "loss": 0.1346,
+      "step": 425
+    },
+    {
+      "epoch": 0.40446237835271776,
+      "grad_norm": 0.0545562319457531,
+      "learning_rate": 0.00019926215250885504,
+      "loss": 0.1387,
+      "step": 426
+    },
+    {
+      "epoch": 0.4054118205554237,
+      "grad_norm": 0.052092909812927246,
+      "learning_rate": 0.00019924869302081715,
+      "loss": 0.1389,
+      "step": 427
+    },
+    {
+      "epoch": 0.4063612627581296,
+      "grad_norm": 0.03847799077630043,
+      "learning_rate": 0.0001992351123416651,
+      "loss": 0.1234,
+      "step": 428
+    },
+    {
+      "epoch": 0.4073107049608355,
+      "grad_norm": 0.0436912477016449,
+      "learning_rate": 0.000199221410487982,
+      "loss": 0.1362,
+      "step": 429
+    },
+    {
+      "epoch": 0.4082601471635414,
+      "grad_norm": 0.04420888423919678,
+      "learning_rate": 0.00019920758747649908,
+      "loss": 0.1243,
+      "step": 430
+    },
+    {
+      "epoch": 0.40920958936624735,
+      "grad_norm": 0.037297070026397705,
+      "learning_rate": 0.00019919364332409535,
+      "loss": 0.1331,
+      "step": 431
+    },
+    {
+      "epoch": 0.4101590315689532,
+      "grad_norm": 0.03854360058903694,
+      "learning_rate": 0.00019917957804779782,
+      "loss": 0.1266,
+      "step": 432
+    },
+    {
+      "epoch": 0.41110847377165916,
+      "grad_norm": 0.04071418195962906,
+      "learning_rate": 0.00019916539166478137,
+      "loss": 0.1292,
+      "step": 433
+    },
+    {
+      "epoch": 0.4120579159743651,
+      "grad_norm": 0.04560808837413788,
+      "learning_rate": 0.00019915108419236882,
+      "loss": 0.1381,
+      "step": 434
+    },
+    {
+      "epoch": 0.41300735817707096,
+      "grad_norm": 0.06313233822584152,
+      "learning_rate": 0.00019913665564803078,
+      "loss": 0.2031,
+      "step": 435
+    },
+    {
+      "epoch": 0.4139568003797769,
+      "grad_norm": 0.04507524147629738,
+      "learning_rate": 0.00019912210604938578,
+      "loss": 0.1277,
+      "step": 436
+    },
+    {
+      "epoch": 0.41490624258248276,
+      "grad_norm": 0.05048058554530144,
+      "learning_rate": 0.00019910743541420007,
+      "loss": 0.1315,
+      "step": 437
+    },
+    {
+      "epoch": 0.4158556847851887,
+      "grad_norm": 0.04872648045420647,
+      "learning_rate": 0.0001990926437603878,
+      "loss": 0.1292,
+      "step": 438
+    },
+    {
+      "epoch": 0.4168051269878946,
+      "grad_norm": 0.04400710016489029,
+      "learning_rate": 0.00019907773110601075,
+      "loss": 0.1236,
+      "step": 439
+    },
+    {
+      "epoch": 0.4177545691906005,
+      "grad_norm": 0.051591627299785614,
+      "learning_rate": 0.00019906269746927863,
+      "loss": 0.1358,
+      "step": 440
+    },
+    {
+      "epoch": 0.41870401139330643,
+      "grad_norm": 0.04288725182414055,
+      "learning_rate": 0.00019904754286854877,
+      "loss": 0.126,
+      "step": 441
+    },
+    {
+      "epoch": 0.41965345359601236,
+      "grad_norm": 0.04984726384282112,
+      "learning_rate": 0.00019903226732232622,
+      "loss": 0.1326,
+      "step": 442
+    },
+    {
+      "epoch": 0.42060289579871823,
+      "grad_norm": 0.041585132479667664,
+      "learning_rate": 0.00019901687084926373,
+      "loss": 0.136,
+      "step": 443
+    },
+    {
+      "epoch": 0.42155233800142417,
+      "grad_norm": 0.05849035084247589,
+      "learning_rate": 0.0001990013534681617,
+      "loss": 0.1727,
+      "step": 444
+    },
+    {
+      "epoch": 0.4225017802041301,
+      "grad_norm": 0.043387994170188904,
+      "learning_rate": 0.00019898571519796817,
+      "loss": 0.1393,
+      "step": 445
+    },
+    {
+      "epoch": 0.42345122240683597,
+      "grad_norm": 0.05867496132850647,
+      "learning_rate": 0.0001989699560577788,
+      "loss": 0.1664,
+      "step": 446
+    },
+    {
+      "epoch": 0.4244006646095419,
+      "grad_norm": 0.07019232958555222,
+      "learning_rate": 0.00019895407606683685,
+      "loss": 0.1653,
+      "step": 447
+    },
+    {
+      "epoch": 0.42535010681224783,
+      "grad_norm": 0.04676515609025955,
+      "learning_rate": 0.00019893807524453314,
+      "loss": 0.1368,
+      "step": 448
+    },
+    {
+      "epoch": 0.4262995490149537,
+      "grad_norm": 0.06640240550041199,
+      "learning_rate": 0.00019892195361040607,
+      "loss": 0.2089,
+      "step": 449
+    },
+    {
+      "epoch": 0.42724899121765963,
+      "grad_norm": 0.044658735394477844,
+      "learning_rate": 0.00019890571118414148,
+      "loss": 0.1298,
+      "step": 450
+    },
+    {
+      "epoch": 0.4281984334203655,
+      "grad_norm": 0.04810122773051262,
+      "learning_rate": 0.00019888934798557278,
+      "loss": 0.1288,
+      "step": 451
+    },
+    {
+      "epoch": 0.42914787562307144,
+      "grad_norm": 0.0425436794757843,
+      "learning_rate": 0.0001988728640346808,
+      "loss": 0.1354,
+      "step": 452
+    },
+    {
+      "epoch": 0.43009731782577737,
+      "grad_norm": 0.04513363912701607,
+      "learning_rate": 0.0001988562593515939,
+      "loss": 0.1346,
+      "step": 453
+    },
+    {
+      "epoch": 0.43104676002848324,
+      "grad_norm": 0.052022870630025864,
+      "learning_rate": 0.0001988395339565878,
+      "loss": 0.1302,
+      "step": 454
+    },
+    {
+      "epoch": 0.4319962022311892,
+      "grad_norm": 0.04852641373872757,
+      "learning_rate": 0.0001988226878700856,
+      "loss": 0.1388,
+      "step": 455
+    },
+    {
+      "epoch": 0.4329456444338951,
+      "grad_norm": 0.04990584775805473,
+      "learning_rate": 0.00019880572111265785,
+      "loss": 0.1552,
+      "step": 456
+    },
+    {
+      "epoch": 0.433895086636601,
+      "grad_norm": 0.052271679043769836,
+      "learning_rate": 0.00019878863370502238,
+      "loss": 0.1404,
+      "step": 457
+    },
+    {
+      "epoch": 0.4348445288393069,
+      "grad_norm": 0.04795520752668381,
+      "learning_rate": 0.00019877142566804436,
+      "loss": 0.1341,
+      "step": 458
+    },
+    {
+      "epoch": 0.43579397104201284,
+      "grad_norm": 0.048165664076805115,
+      "learning_rate": 0.00019875409702273632,
+      "loss": 0.1343,
+      "step": 459
+    },
+    {
+      "epoch": 0.4367434132447187,
+      "grad_norm": 0.04213611036539078,
+      "learning_rate": 0.000198736647790258,
+      "loss": 0.1369,
+      "step": 460
+    },
+    {
+      "epoch": 0.43769285544742464,
+      "grad_norm": 0.05819966271519661,
+      "learning_rate": 0.00019871907799191632,
+      "loss": 0.1615,
+      "step": 461
+    },
+    {
+      "epoch": 0.4386422976501306,
+      "grad_norm": 0.057378821074962616,
+      "learning_rate": 0.00019870138764916558,
+      "loss": 0.175,
+      "step": 462
+    },
+    {
+      "epoch": 0.43959173985283645,
+      "grad_norm": 0.0432853177189827,
+      "learning_rate": 0.00019868357678360724,
+      "loss": 0.1371,
+      "step": 463
+    },
+    {
+      "epoch": 0.4405411820555424,
+      "grad_norm": 0.03890872746706009,
+      "learning_rate": 0.0001986656454169898,
+      "loss": 0.1332,
+      "step": 464
+    },
+    {
+      "epoch": 0.44149062425824825,
+      "grad_norm": 0.04006613418459892,
+      "learning_rate": 0.00019864759357120896,
+      "loss": 0.1342,
+      "step": 465
+    },
+    {
+      "epoch": 0.4424400664609542,
+      "grad_norm": 0.049053166061639786,
+      "learning_rate": 0.00019862942126830767,
+      "loss": 0.1756,
+      "step": 466
+    },
+    {
+      "epoch": 0.4433895086636601,
+      "grad_norm": 0.03966079652309418,
+      "learning_rate": 0.00019861112853047577,
+      "loss": 0.1303,
+      "step": 467
+    },
+    {
+      "epoch": 0.444338950866366,
+      "grad_norm": 0.04506433755159378,
+      "learning_rate": 0.0001985927153800503,
+      "loss": 0.136,
+      "step": 468
+    },
+    {
+      "epoch": 0.4452883930690719,
+      "grad_norm": 0.04392915591597557,
+      "learning_rate": 0.00019857418183951526,
+      "loss": 0.1397,
+      "step": 469
+    },
+    {
+      "epoch": 0.44623783527177785,
+      "grad_norm": 0.038007620722055435,
+      "learning_rate": 0.0001985555279315017,
+      "loss": 0.1246,
+      "step": 470
+    },
+    {
+      "epoch": 0.4471872774744837,
+      "grad_norm": 0.048948097974061966,
+      "learning_rate": 0.00019853675367878764,
+      "loss": 0.1329,
+      "step": 471
+    },
+    {
+      "epoch": 0.44813671967718965,
+      "grad_norm": 0.04174380376935005,
+      "learning_rate": 0.00019851785910429806,
+      "loss": 0.13,
+      "step": 472
+    },
+    {
+      "epoch": 0.4490861618798956,
+      "grad_norm": 0.048575468361377716,
+      "learning_rate": 0.00019849884423110478,
+      "loss": 0.1385,
+      "step": 473
+    },
+    {
+      "epoch": 0.45003560408260146,
+      "grad_norm": 0.05167670175433159,
+      "learning_rate": 0.00019847970908242664,
+      "loss": 0.1684,
+      "step": 474
+    },
+    {
+      "epoch": 0.4509850462853074,
+      "grad_norm": 0.06849198788404465,
+      "learning_rate": 0.00019846045368162923,
+      "loss": 0.1795,
+      "step": 475
+    },
+    {
+      "epoch": 0.4519344884880133,
+      "grad_norm": 0.044273603707551956,
+      "learning_rate": 0.0001984410780522251,
+      "loss": 0.1246,
+      "step": 476
+    },
+    {
+      "epoch": 0.4528839306907192,
+      "grad_norm": 0.048194363713264465,
+      "learning_rate": 0.00019842158221787353,
+      "loss": 0.1366,
+      "step": 477
+    },
+    {
+      "epoch": 0.4538333728934251,
+      "grad_norm": 0.033906418830156326,
+      "learning_rate": 0.00019840196620238057,
+      "loss": 0.1235,
+      "step": 478
+    },
+    {
+      "epoch": 0.454782815096131,
+      "grad_norm": 0.043933141976594925,
+      "learning_rate": 0.00019838223002969905,
+      "loss": 0.1195,
+      "step": 479
+    },
+    {
+      "epoch": 0.45573225729883693,
+      "grad_norm": 0.056823644787073135,
+      "learning_rate": 0.00019836237372392854,
+      "loss": 0.1757,
+      "step": 480
+    },
+    {
+      "epoch": 0.45668169950154286,
+      "grad_norm": 0.07587820291519165,
+      "learning_rate": 0.00019834239730931526,
+      "loss": 0.1784,
+      "step": 481
+    },
+    {
+      "epoch": 0.45763114170424873,
+      "grad_norm": 0.04008018970489502,
+      "learning_rate": 0.0001983223008102521,
+      "loss": 0.1306,
+      "step": 482
+    },
+    {
+      "epoch": 0.45858058390695466,
+      "grad_norm": 0.05180038511753082,
+      "learning_rate": 0.00019830208425127867,
+      "loss": 0.1485,
+      "step": 483
+    },
+    {
+      "epoch": 0.4595300261096606,
+      "grad_norm": 0.0691617876291275,
+      "learning_rate": 0.00019828174765708104,
+      "loss": 0.1249,
+      "step": 484
+    },
+    {
+      "epoch": 0.46047946831236647,
+      "grad_norm": 0.0565367266535759,
+      "learning_rate": 0.00019826129105249195,
+      "loss": 0.1744,
+      "step": 485
+    },
+    {
+      "epoch": 0.4614289105150724,
+      "grad_norm": 0.044927019625902176,
+      "learning_rate": 0.00019824071446249072,
+      "loss": 0.1341,
+      "step": 486
+    },
+    {
+      "epoch": 0.46237835271777833,
+      "grad_norm": 0.04481721669435501,
+      "learning_rate": 0.00019822001791220298,
+      "loss": 0.1354,
+      "step": 487
+    },
+    {
+      "epoch": 0.4633277949204842,
+      "grad_norm": 0.05233500525355339,
+      "learning_rate": 0.0001981992014269011,
+      "loss": 0.1501,
+      "step": 488
+    },
+    {
+      "epoch": 0.46427723712319013,
+      "grad_norm": 0.044350553303956985,
+      "learning_rate": 0.00019817826503200372,
+      "loss": 0.1335,
+      "step": 489
+    },
+    {
+      "epoch": 0.465226679325896,
+      "grad_norm": 0.03551819548010826,
+      "learning_rate": 0.000198157208753076,
+      "loss": 0.1322,
+      "step": 490
+    },
+    {
+      "epoch": 0.46617612152860194,
+      "grad_norm": 0.04409592226147652,
+      "learning_rate": 0.00019813603261582943,
+      "loss": 0.1561,
+      "step": 491
+    },
+    {
+      "epoch": 0.46712556373130787,
+      "grad_norm": 0.04842127487063408,
+      "learning_rate": 0.0001981147366461219,
+      "loss": 0.1296,
+      "step": 492
+    },
+    {
+      "epoch": 0.46807500593401374,
+      "grad_norm": 0.04349881038069725,
+      "learning_rate": 0.00019809332086995757,
+      "loss": 0.1319,
+      "step": 493
+    },
+    {
+      "epoch": 0.4690244481367197,
+      "grad_norm": 0.04413028433918953,
+      "learning_rate": 0.00019807178531348698,
+      "loss": 0.1321,
+      "step": 494
+    },
+    {
+      "epoch": 0.4699738903394256,
+      "grad_norm": 0.03972313553094864,
+      "learning_rate": 0.00019805013000300683,
+      "loss": 0.1358,
+      "step": 495
+    },
+    {
+      "epoch": 0.4709233325421315,
+      "grad_norm": 0.052269116044044495,
+      "learning_rate": 0.00019802835496496012,
+      "loss": 0.1389,
+      "step": 496
+    },
+    {
+      "epoch": 0.4718727747448374,
+      "grad_norm": 0.0379653237760067,
+      "learning_rate": 0.00019800646022593603,
+      "loss": 0.1283,
+      "step": 497
+    },
+    {
+      "epoch": 0.47282221694754334,
+      "grad_norm": 0.04370688647031784,
+      "learning_rate": 0.0001979844458126699,
+      "loss": 0.1278,
+      "step": 498
+    },
+    {
+      "epoch": 0.4737716591502492,
+      "grad_norm": 0.03912369906902313,
+      "learning_rate": 0.0001979623117520432,
+      "loss": 0.1257,
+      "step": 499
+    },
+    {
+      "epoch": 0.47472110135295514,
+      "grad_norm": 0.039594005793333054,
+      "learning_rate": 0.00019794005807108352,
+      "loss": 0.1375,
+      "step": 500
+    },
+    {
+      "epoch": 0.4756705435556611,
+      "grad_norm": 0.03889892250299454,
+      "learning_rate": 0.00019791768479696448,
+      "loss": 0.13,
+      "step": 501
+    },
+    {
+      "epoch": 0.47661998575836695,
+      "grad_norm": 0.03966660797595978,
+      "learning_rate": 0.00019789519195700578,
+      "loss": 0.1268,
+      "step": 502
+    },
+    {
+      "epoch": 0.4775694279610729,
+      "grad_norm": 0.04501716047525406,
+      "learning_rate": 0.00019787257957867306,
+      "loss": 0.1423,
+      "step": 503
+    },
+    {
+      "epoch": 0.47851887016377875,
+      "grad_norm": 0.06255436688661575,
+      "learning_rate": 0.000197849847689578,
+      "loss": 0.1799,
+      "step": 504
+    },
+    {
+      "epoch": 0.4794683123664847,
+      "grad_norm": 0.050308458507061005,
+      "learning_rate": 0.00019782699631747813,
+      "loss": 0.1733,
+      "step": 505
+    },
+    {
+      "epoch": 0.4804177545691906,
+      "grad_norm": 0.0357963964343071,
+      "learning_rate": 0.00019780402549027698,
+      "loss": 0.1268,
+      "step": 506
+    },
+    {
+      "epoch": 0.4813671967718965,
+      "grad_norm": 0.03651968017220497,
+      "learning_rate": 0.00019778093523602384,
+      "loss": 0.1267,
+      "step": 507
+    },
+    {
+      "epoch": 0.4823166389746024,
+      "grad_norm": 0.043042074888944626,
+      "learning_rate": 0.0001977577255829139,
+      "loss": 0.1256,
+      "step": 508
+    },
+    {
+      "epoch": 0.48326608117730835,
+      "grad_norm": 0.07031014561653137,
+      "learning_rate": 0.00019773439655928815,
+      "loss": 0.1796,
+      "step": 509
+    },
+    {
+      "epoch": 0.4842155233800142,
+      "grad_norm": 0.04429268836975098,
+      "learning_rate": 0.00019771094819363326,
+      "loss": 0.1298,
+      "step": 510
+    },
+    {
+      "epoch": 0.48516496558272015,
+      "grad_norm": 0.0373898483812809,
+      "learning_rate": 0.00019768738051458172,
+      "loss": 0.1232,
+      "step": 511
+    },
+    {
+      "epoch": 0.4861144077854261,
+      "grad_norm": 0.05853155627846718,
+      "learning_rate": 0.00019766369355091166,
+      "loss": 0.1694,
+      "step": 512
+    },
+    {
+      "epoch": 0.48706384998813196,
+      "grad_norm": 0.05050895735621452,
+      "learning_rate": 0.00019763988733154686,
+      "loss": 0.1665,
+      "step": 513
+    },
+    {
+      "epoch": 0.4880132921908379,
+      "grad_norm": 0.04074448347091675,
+      "learning_rate": 0.0001976159618855568,
+      "loss": 0.1336,
+      "step": 514
+    },
+    {
+      "epoch": 0.4889627343935438,
+      "grad_norm": 0.03826110064983368,
+      "learning_rate": 0.00019759191724215644,
+      "loss": 0.132,
+      "step": 515
+    },
+    {
+      "epoch": 0.4899121765962497,
+      "grad_norm": 0.04392875358462334,
+      "learning_rate": 0.0001975677534307064,
+      "loss": 0.1204,
+      "step": 516
+    },
+    {
+      "epoch": 0.4908616187989556,
+      "grad_norm": 0.04615531116724014,
+      "learning_rate": 0.0001975434704807127,
+      "loss": 0.1358,
+      "step": 517
+    },
+    {
+      "epoch": 0.4918110610016615,
+      "grad_norm": 0.053060565143823624,
+      "learning_rate": 0.00019751906842182688,
+      "loss": 0.1299,
+      "step": 518
+    },
+    {
+      "epoch": 0.49276050320436743,
+      "grad_norm": 0.04905511438846588,
+      "learning_rate": 0.00019749454728384594,
+      "loss": 0.1284,
+      "step": 519
+    },
+    {
+      "epoch": 0.49370994540707336,
+      "grad_norm": 0.04257996007800102,
+      "learning_rate": 0.00019746990709671234,
+      "loss": 0.1353,
+      "step": 520
+    },
+    {
+      "epoch": 0.49465938760977923,
+      "grad_norm": 0.05581909418106079,
+      "learning_rate": 0.0001974451478905138,
+      "loss": 0.1594,
+      "step": 521
+    },
+    {
+      "epoch": 0.49560882981248516,
+      "grad_norm": 0.04603990167379379,
+      "learning_rate": 0.00019742026969548338,
+      "loss": 0.1383,
+      "step": 522
+    },
+    {
+      "epoch": 0.4965582720151911,
+      "grad_norm": 0.058511972427368164,
+      "learning_rate": 0.00019739527254199958,
+      "loss": 0.1725,
+      "step": 523
+    },
+    {
+      "epoch": 0.49750771421789697,
+      "grad_norm": 0.03875808045268059,
+      "learning_rate": 0.000197370156460586,
+      "loss": 0.1405,
+      "step": 524
+    },
+    {
+      "epoch": 0.4984571564206029,
+      "grad_norm": 0.040860000997781754,
+      "learning_rate": 0.00019734492148191151,
+      "loss": 0.139,
+      "step": 525
+    },
+    {
+      "epoch": 0.49940659862330883,
+      "grad_norm": 0.06110459193587303,
+      "learning_rate": 0.00019731956763679014,
+      "loss": 0.223,
+      "step": 526
+    },
+    {
+      "epoch": 0.5003560408260147,
+      "grad_norm": 0.05238598585128784,
+      "learning_rate": 0.00019729409495618117,
+      "loss": 0.1681,
+      "step": 527
+    },
+    {
+      "epoch": 0.5013054830287206,
+      "grad_norm": 0.05180145800113678,
+      "learning_rate": 0.00019726850347118885,
+      "loss": 0.1743,
+      "step": 528
+    },
+    {
+      "epoch": 0.5022549252314266,
+      "grad_norm": 0.05066410079598427,
+      "learning_rate": 0.00019724279321306262,
+      "loss": 0.1634,
+      "step": 529
+    },
+    {
+      "epoch": 0.5032043674341324,
+      "grad_norm": 0.06856084614992142,
+      "learning_rate": 0.00019721696421319684,
+      "loss": 0.1685,
+      "step": 530
+    },
+    {
+      "epoch": 0.5041538096368383,
+      "grad_norm": 0.045972324907779694,
+      "learning_rate": 0.00019719101650313096,
+      "loss": 0.1245,
+      "step": 531
+    },
+    {
+      "epoch": 0.5051032518395443,
+      "grad_norm": 0.04522623121738434,
+      "learning_rate": 0.00019716495011454934,
+      "loss": 0.1367,
+      "step": 532
+    },
+    {
+      "epoch": 0.5060526940422502,
+      "grad_norm": 0.0780516117811203,
+      "learning_rate": 0.00019713876507928126,
+      "loss": 0.1351,
+      "step": 533
+    },
+    {
+      "epoch": 0.507002136244956,
+      "grad_norm": 0.04264210909605026,
+      "learning_rate": 0.00019711246142930088,
+      "loss": 0.1312,
+      "step": 534
+    },
+    {
+      "epoch": 0.507951578447662,
+      "grad_norm": 0.059501100331544876,
+      "learning_rate": 0.00019708603919672718,
+      "loss": 0.1698,
+      "step": 535
+    },
+    {
+      "epoch": 0.5089010206503679,
+      "grad_norm": 0.060105033218860626,
+      "learning_rate": 0.00019705949841382396,
+      "loss": 0.1303,
+      "step": 536
+    },
+    {
+      "epoch": 0.5098504628530738,
+      "grad_norm": 0.04733967408537865,
+      "learning_rate": 0.00019703283911299982,
+      "loss": 0.1245,
+      "step": 537
+    },
+    {
+      "epoch": 0.5107999050557798,
+      "grad_norm": 0.04254663735628128,
+      "learning_rate": 0.00019700606132680798,
+      "loss": 0.1343,
+      "step": 538
+    },
+    {
+      "epoch": 0.5117493472584856,
+      "grad_norm": 0.06302463263273239,
+      "learning_rate": 0.00019697916508794645,
+      "loss": 0.1831,
+      "step": 539
+    },
+    {
+      "epoch": 0.5126987894611915,
+      "grad_norm": 0.05301344394683838,
+      "learning_rate": 0.0001969521504292578,
+      "loss": 0.1316,
+      "step": 540
+    },
+    {
+      "epoch": 0.5136482316638975,
+      "grad_norm": 0.04151083528995514,
+      "learning_rate": 0.00019692501738372922,
+      "loss": 0.1335,
+      "step": 541
+    },
+    {
+      "epoch": 0.5145976738666034,
+      "grad_norm": 0.05647062510251999,
+      "learning_rate": 0.00019689776598449257,
+      "loss": 0.1688,
+      "step": 542
+    },
+    {
+      "epoch": 0.5155471160693093,
+      "grad_norm": 0.037060294300317764,
+      "learning_rate": 0.000196870396264824,
+      "loss": 0.1339,
+      "step": 543
+    },
+    {
+      "epoch": 0.5164965582720152,
+      "grad_norm": 0.04036247730255127,
+      "learning_rate": 0.0001968429082581443,
+      "loss": 0.1361,
+      "step": 544
+    },
+    {
+      "epoch": 0.5174460004747211,
+      "grad_norm": 0.040889665484428406,
+      "learning_rate": 0.00019681530199801875,
+      "loss": 0.1356,
+      "step": 545
+    },
+    {
+      "epoch": 0.518395442677427,
+      "grad_norm": 0.0538480207324028,
+      "learning_rate": 0.00019678757751815686,
+      "loss": 0.1689,
+      "step": 546
+    },
+    {
+      "epoch": 0.519344884880133,
+      "grad_norm": 0.04074794426560402,
+      "learning_rate": 0.0001967597348524126,
+      "loss": 0.1329,
+      "step": 547
+    },
+    {
+      "epoch": 0.5202943270828388,
+      "grad_norm": 0.03896891698241234,
+      "learning_rate": 0.00019673177403478428,
+      "loss": 0.1356,
+      "step": 548
+    },
+    {
+      "epoch": 0.5212437692855447,
+      "grad_norm": 0.04619259387254715,
+      "learning_rate": 0.00019670369509941442,
+      "loss": 0.163,
+      "step": 549
+    },
+    {
+      "epoch": 0.5221932114882507,
+      "grad_norm": 0.035968657582998276,
+      "learning_rate": 0.00019667549808058976,
+      "loss": 0.1242,
+      "step": 550
+    },
+    {
+      "epoch": 0.5231426536909566,
+      "grad_norm": 0.04564007744193077,
+      "learning_rate": 0.0001966471830127413,
+      "loss": 0.1364,
+      "step": 551
+    },
+    {
+      "epoch": 0.5240920958936625,
+      "grad_norm": 0.03991610184311867,
+      "learning_rate": 0.00019661874993044415,
+      "loss": 0.1312,
+      "step": 552
+    },
+    {
+      "epoch": 0.5250415380963683,
+      "grad_norm": 0.037240512669086456,
+      "learning_rate": 0.00019659019886841752,
+      "loss": 0.1279,
+      "step": 553
+    },
+    {
+      "epoch": 0.5259909802990743,
+      "grad_norm": 0.06598762422800064,
+      "learning_rate": 0.00019656152986152468,
+      "loss": 0.2165,
+      "step": 554
+    },
+    {
+      "epoch": 0.5269404225017802,
+      "grad_norm": 0.03867746889591217,
+      "learning_rate": 0.00019653274294477292,
+      "loss": 0.1233,
+      "step": 555
+    },
+    {
+      "epoch": 0.5278898647044861,
+      "grad_norm": 0.051915477961301804,
+      "learning_rate": 0.00019650383815331357,
+      "loss": 0.168,
+      "step": 556
+    },
+    {
+      "epoch": 0.528839306907192,
+      "grad_norm": 0.054896485060453415,
+      "learning_rate": 0.00019647481552244182,
+      "loss": 0.1678,
+      "step": 557
+    },
+    {
+      "epoch": 0.5297887491098979,
+      "grad_norm": 0.05439051240682602,
+      "learning_rate": 0.00019644567508759675,
+      "loss": 0.1607,
+      "step": 558
+    },
+    {
+      "epoch": 0.5307381913126038,
+      "grad_norm": 0.03601578250527382,
+      "learning_rate": 0.00019641641688436135,
+      "loss": 0.1271,
+      "step": 559
+    },
+    {
+      "epoch": 0.5316876335153098,
+      "grad_norm": 0.06025104597210884,
+      "learning_rate": 0.00019638704094846236,
+      "loss": 0.176,
+      "step": 560
+    },
+    {
+      "epoch": 0.5326370757180157,
+      "grad_norm": 0.04126368835568428,
+      "learning_rate": 0.00019635754731577032,
+      "loss": 0.1319,
+      "step": 561
+    },
+    {
+      "epoch": 0.5335865179207215,
+      "grad_norm": 0.05305393040180206,
+      "learning_rate": 0.00019632793602229943,
+      "loss": 0.1699,
+      "step": 562
+    },
+    {
+      "epoch": 0.5345359601234275,
+      "grad_norm": 0.03538331016898155,
+      "learning_rate": 0.00019629820710420764,
+      "loss": 0.124,
+      "step": 563
+    },
+    {
+      "epoch": 0.5354854023261334,
+      "grad_norm": 0.05861300975084305,
+      "learning_rate": 0.0001962683605977965,
+      "loss": 0.1688,
+      "step": 564
+    },
+    {
+      "epoch": 0.5364348445288393,
+      "grad_norm": 0.040226079523563385,
+      "learning_rate": 0.0001962383965395111,
+      "loss": 0.1334,
+      "step": 565
+    },
+    {
+      "epoch": 0.5373842867315453,
+      "grad_norm": 0.035788875073194504,
+      "learning_rate": 0.00019620831496594017,
+      "loss": 0.1281,
+      "step": 566
+    },
+    {
+      "epoch": 0.5383337289342511,
+      "grad_norm": 0.0334162712097168,
+      "learning_rate": 0.0001961781159138158,
+      "loss": 0.1317,
+      "step": 567
+    },
+    {
+      "epoch": 0.539283171136957,
+      "grad_norm": 0.03352081775665283,
+      "learning_rate": 0.00019614779942001364,
+      "loss": 0.1334,
+      "step": 568
+    },
+    {
+      "epoch": 0.540232613339663,
+      "grad_norm": 0.03684060648083687,
+      "learning_rate": 0.00019611736552155274,
+      "loss": 0.1349,
+      "step": 569
+    },
+    {
+      "epoch": 0.5411820555423689,
+      "grad_norm": 0.03640671446919441,
+      "learning_rate": 0.00019608681425559542,
+      "loss": 0.1278,
+      "step": 570
+    },
+    {
+      "epoch": 0.5421314977450747,
+      "grad_norm": 0.04167250171303749,
+      "learning_rate": 0.00019605614565944748,
+      "loss": 0.1384,
+      "step": 571
+    },
+    {
+      "epoch": 0.5430809399477807,
+      "grad_norm": 0.0416824147105217,
+      "learning_rate": 0.00019602535977055778,
+      "loss": 0.1319,
+      "step": 572
+    },
+    {
+      "epoch": 0.5440303821504866,
+      "grad_norm": 0.03897137567400932,
+      "learning_rate": 0.00019599445662651861,
+      "loss": 0.1389,
+      "step": 573
+    },
+    {
+      "epoch": 0.5449798243531925,
+      "grad_norm": 0.03894896060228348,
+      "learning_rate": 0.00019596343626506526,
+      "loss": 0.1341,
+      "step": 574
+    },
+    {
+      "epoch": 0.5459292665558985,
+      "grad_norm": 0.04211690276861191,
+      "learning_rate": 0.00019593229872407627,
+      "loss": 0.1377,
+      "step": 575
+    },
+    {
+      "epoch": 0.5468787087586043,
+      "grad_norm": 0.04308454692363739,
+      "learning_rate": 0.00019590104404157327,
+      "loss": 0.1268,
+      "step": 576
+    },
+    {
+      "epoch": 0.5478281509613102,
+      "grad_norm": 0.0525001622736454,
+      "learning_rate": 0.00019586967225572086,
+      "loss": 0.1775,
+      "step": 577
+    },
+    {
+      "epoch": 0.5487775931640161,
+      "grad_norm": 0.056315965950489044,
+      "learning_rate": 0.00019583818340482664,
+      "loss": 0.1688,
+      "step": 578
+    },
+    {
+      "epoch": 0.5497270353667221,
+      "grad_norm": 0.03801283985376358,
+      "learning_rate": 0.0001958065775273412,
+      "loss": 0.1309,
+      "step": 579
+    },
+    {
+      "epoch": 0.550676477569428,
+      "grad_norm": 0.03738854080438614,
+      "learning_rate": 0.00019577485466185804,
+      "loss": 0.137,
+      "step": 580
+    },
+    {
+      "epoch": 0.5516259197721338,
+      "grad_norm": 0.03772661089897156,
+      "learning_rate": 0.0001957430148471134,
+      "loss": 0.1276,
+      "step": 581
+    },
+    {
+      "epoch": 0.5525753619748398,
+      "grad_norm": 0.039842378348112106,
+      "learning_rate": 0.00019571105812198652,
+      "loss": 0.1329,
+      "step": 582
+    },
+    {
+      "epoch": 0.5535248041775457,
+      "grad_norm": 0.033689334988594055,
+      "learning_rate": 0.0001956789845254992,
+      "loss": 0.1265,
+      "step": 583
+    },
+    {
+      "epoch": 0.5544742463802516,
+      "grad_norm": 0.046588387340307236,
+      "learning_rate": 0.00019564679409681608,
+      "loss": 0.1645,
+      "step": 584
+    },
+    {
+      "epoch": 0.5554236885829575,
+      "grad_norm": 0.03861064463853836,
+      "learning_rate": 0.0001956144868752444,
+      "loss": 0.1267,
+      "step": 585
+    },
+    {
+      "epoch": 0.5563731307856634,
+      "grad_norm": 0.03467525169253349,
+      "learning_rate": 0.000195582062900234,
+      "loss": 0.1299,
+      "step": 586
+    },
+    {
+      "epoch": 0.5573225729883693,
+      "grad_norm": 0.03659389913082123,
+      "learning_rate": 0.0001955495222113774,
+      "loss": 0.1286,
+      "step": 587
+    },
+    {
+      "epoch": 0.5582720151910753,
+      "grad_norm": 0.03826770931482315,
+      "learning_rate": 0.0001955168648484095,
+      "loss": 0.1313,
+      "step": 588
+    },
+    {
+      "epoch": 0.5592214573937812,
+      "grad_norm": 0.038110729306936264,
+      "learning_rate": 0.00019548409085120772,
+      "loss": 0.137,
+      "step": 589
+    },
+    {
+      "epoch": 0.560170899596487,
+      "grad_norm": 0.03989555314183235,
+      "learning_rate": 0.0001954512002597919,
+      "loss": 0.132,
+      "step": 590
+    },
+    {
+      "epoch": 0.561120341799193,
+      "grad_norm": 0.05395180359482765,
+      "learning_rate": 0.00019541819311432427,
+      "loss": 0.1401,
+      "step": 591
+    },
+    {
+      "epoch": 0.5620697840018989,
+      "grad_norm": 0.05007918179035187,
+      "learning_rate": 0.00019538506945510938,
+      "loss": 0.1584,
+      "step": 592
+    },
+    {
+      "epoch": 0.5630192262046048,
+      "grad_norm": 0.047849785536527634,
+      "learning_rate": 0.00019535182932259404,
+      "loss": 0.1265,
+      "step": 593
+    },
+    {
+      "epoch": 0.5639686684073107,
+      "grad_norm": 0.04303041473031044,
+      "learning_rate": 0.00019531847275736726,
+      "loss": 0.1245,
+      "step": 594
+    },
+    {
+      "epoch": 0.5649181106100166,
+      "grad_norm": 0.04128289222717285,
+      "learning_rate": 0.00019528499980016025,
+      "loss": 0.1317,
+      "step": 595
+    },
+    {
+      "epoch": 0.5658675528127225,
+      "grad_norm": 0.04311414808034897,
+      "learning_rate": 0.00019525141049184637,
+      "loss": 0.1364,
+      "step": 596
+    },
+    {
+      "epoch": 0.5668169950154285,
+      "grad_norm": 0.03765838220715523,
+      "learning_rate": 0.00019521770487344103,
+      "loss": 0.1268,
+      "step": 597
+    },
+    {
+      "epoch": 0.5677664372181344,
+      "grad_norm": 0.03674585744738579,
+      "learning_rate": 0.00019518388298610164,
+      "loss": 0.1297,
+      "step": 598
+    },
+    {
+      "epoch": 0.5687158794208402,
+      "grad_norm": 0.036937762051820755,
+      "learning_rate": 0.0001951499448711276,
+      "loss": 0.1303,
+      "step": 599
+    },
+    {
+      "epoch": 0.5696653216235462,
+      "grad_norm": 0.03748161345720291,
+      "learning_rate": 0.0001951158905699603,
+      "loss": 0.1328,
+      "step": 600
+    },
+    {
+      "epoch": 0.5706147638262521,
+      "grad_norm": 0.04011257737874985,
+      "learning_rate": 0.00019508172012418283,
+      "loss": 0.1346,
+      "step": 601
+    },
+    {
+      "epoch": 0.571564206028958,
+      "grad_norm": 0.03853931650519371,
+      "learning_rate": 0.00019504743357552035,
+      "loss": 0.1279,
+      "step": 602
+    },
+    {
+      "epoch": 0.572513648231664,
+      "grad_norm": 0.03750459849834442,
+      "learning_rate": 0.0001950130309658396,
+      "loss": 0.1227,
+      "step": 603
+    },
+    {
+      "epoch": 0.5734630904343698,
+      "grad_norm": 0.05542079731822014,
+      "learning_rate": 0.00019497851233714908,
+      "loss": 0.1647,
+      "step": 604
+    },
+    {
+      "epoch": 0.5744125326370757,
+      "grad_norm": 0.04472218081355095,
+      "learning_rate": 0.00019494387773159898,
+      "loss": 0.1416,
+      "step": 605
+    },
+    {
+      "epoch": 0.5753619748397816,
+      "grad_norm": 0.052323974668979645,
+      "learning_rate": 0.00019490912719148114,
+      "loss": 0.1367,
+      "step": 606
+    },
+    {
+      "epoch": 0.5763114170424876,
+      "grad_norm": 0.037580832839012146,
+      "learning_rate": 0.00019487426075922893,
+      "loss": 0.131,
+      "step": 607
+    },
+    {
+      "epoch": 0.5772608592451934,
+      "grad_norm": 0.03929577395319939,
+      "learning_rate": 0.0001948392784774172,
+      "loss": 0.128,
+      "step": 608
+    },
+    {
+      "epoch": 0.5782103014478993,
+      "grad_norm": 0.03706606104969978,
+      "learning_rate": 0.0001948041803887623,
+      "loss": 0.1316,
+      "step": 609
+    },
+    {
+      "epoch": 0.5791597436506053,
+      "grad_norm": 0.038938358426094055,
+      "learning_rate": 0.00019476896653612203,
+      "loss": 0.1275,
+      "step": 610
+    },
+    {
+      "epoch": 0.5801091858533112,
+      "grad_norm": 0.04818068817257881,
+      "learning_rate": 0.00019473363696249546,
+      "loss": 0.1662,
+      "step": 611
+    },
+    {
+      "epoch": 0.581058628056017,
+      "grad_norm": 0.03735940158367157,
+      "learning_rate": 0.00019469819171102304,
+      "loss": 0.1361,
+      "step": 612
+    },
+    {
+      "epoch": 0.582008070258723,
+      "grad_norm": 0.03568827733397484,
+      "learning_rate": 0.00019466263082498645,
+      "loss": 0.1216,
+      "step": 613
+    },
+    {
+      "epoch": 0.5829575124614289,
+      "grad_norm": 0.03913251310586929,
+      "learning_rate": 0.0001946269543478085,
+      "loss": 0.1321,
+      "step": 614
+    },
+    {
+      "epoch": 0.5839069546641348,
+      "grad_norm": 0.062009479850530624,
+      "learning_rate": 0.0001945911623230533,
+      "loss": 0.1778,
+      "step": 615
+    },
+    {
+      "epoch": 0.5848563968668408,
+      "grad_norm": 0.039088111370801926,
+      "learning_rate": 0.0001945552547944259,
+      "loss": 0.1352,
+      "step": 616
+    },
+    {
+      "epoch": 0.5858058390695466,
+      "grad_norm": 0.041976600885391235,
+      "learning_rate": 0.0001945192318057725,
+      "loss": 0.1394,
+      "step": 617
+    },
+    {
+      "epoch": 0.5867552812722525,
+      "grad_norm": 0.03723563253879547,
+      "learning_rate": 0.00019448309340108018,
+      "loss": 0.1246,
+      "step": 618
+    },
+    {
+      "epoch": 0.5877047234749585,
+      "grad_norm": 0.0382399819791317,
+      "learning_rate": 0.00019444683962447707,
+      "loss": 0.1232,
+      "step": 619
+    },
+    {
+      "epoch": 0.5886541656776644,
+      "grad_norm": 0.03758077695965767,
+      "learning_rate": 0.0001944104705202321,
+      "loss": 0.1417,
+      "step": 620
+    },
+    {
+      "epoch": 0.5896036078803703,
+      "grad_norm": 0.034823786467313766,
+      "learning_rate": 0.000194373986132755,
+      "loss": 0.1304,
+      "step": 621
+    },
+    {
+      "epoch": 0.5905530500830762,
+      "grad_norm": 0.03755120187997818,
+      "learning_rate": 0.00019433738650659641,
+      "loss": 0.133,
+      "step": 622
+    },
+    {
+      "epoch": 0.5915024922857821,
+      "grad_norm": 0.03759913146495819,
+      "learning_rate": 0.00019430067168644754,
+      "loss": 0.1222,
+      "step": 623
+    },
+    {
+      "epoch": 0.592451934488488,
+      "grad_norm": 0.06232694163918495,
+      "learning_rate": 0.0001942638417171403,
+      "loss": 0.1778,
+      "step": 624
+    },
+    {
+      "epoch": 0.593401376691194,
+      "grad_norm": 0.05642306059598923,
+      "learning_rate": 0.00019422689664364725,
+      "loss": 0.1706,
+      "step": 625
+    },
+    {
+      "epoch": 0.5943508188938998,
+      "grad_norm": 0.0827709287405014,
+      "learning_rate": 0.00019418983651108148,
+      "loss": 0.2371,
+      "step": 626
+    },
+    {
+      "epoch": 0.5953002610966057,
+      "grad_norm": 0.03614366054534912,
+      "learning_rate": 0.00019415266136469652,
+      "loss": 0.1225,
+      "step": 627
+    },
+    {
+      "epoch": 0.5962497032993117,
+      "grad_norm": 0.042416494339704514,
+      "learning_rate": 0.00019411537124988643,
+      "loss": 0.1239,
+      "step": 628
+    },
+    {
+      "epoch": 0.5971991455020176,
+      "grad_norm": 0.037246908992528915,
+      "learning_rate": 0.00019407796621218566,
+      "loss": 0.1292,
+      "step": 629
+    },
+    {
+      "epoch": 0.5981485877047235,
+      "grad_norm": 0.05374092981219292,
+      "learning_rate": 0.00019404044629726887,
+      "loss": 0.1782,
+      "step": 630
+    },
+    {
+      "epoch": 0.5990980299074293,
+      "grad_norm": 0.052854426205158234,
+      "learning_rate": 0.00019400281155095112,
+      "loss": 0.1711,
+      "step": 631
+    },
+    {
+      "epoch": 0.6000474721101353,
+      "grad_norm": 0.038800131529569626,
+      "learning_rate": 0.00019396506201918765,
+      "loss": 0.1285,
+      "step": 632
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 3159,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 158,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 3.667125138649252e+17,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}