diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,3927 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.998875140607424,
+  "eval_steps": 500,
+  "global_step": 555,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.008998875140607425,
+      "grad_norm": 6.405796769298009,
+      "learning_rate": 1.4285714285714286e-06,
+      "loss": 1.0356,
+      "step": 1
+    },
+    {
+      "epoch": 0.01799775028121485,
+      "grad_norm": 6.457071617226858,
+      "learning_rate": 2.8571428571428573e-06,
+      "loss": 1.0356,
+      "step": 2
+    },
+    {
+      "epoch": 0.02699662542182227,
+      "grad_norm": 6.306094210313564,
+      "learning_rate": 4.2857142857142855e-06,
+      "loss": 1.0346,
+      "step": 3
+    },
+    {
+      "epoch": 0.0359955005624297,
+      "grad_norm": 4.578449546704467,
+      "learning_rate": 5.7142857142857145e-06,
+      "loss": 0.9945,
+      "step": 4
+    },
+    {
+      "epoch": 0.04499437570303712,
+      "grad_norm": 2.7068662098456318,
+      "learning_rate": 7.1428571428571436e-06,
+      "loss": 0.9512,
+      "step": 5
+    },
+    {
+      "epoch": 0.05399325084364454,
+      "grad_norm": 2.4781853702406553,
+      "learning_rate": 8.571428571428571e-06,
+      "loss": 0.9632,
+      "step": 6
+    },
+    {
+      "epoch": 0.06299212598425197,
+      "grad_norm": 4.778781193622344,
+      "learning_rate": 1e-05,
+      "loss": 0.9478,
+      "step": 7
+    },
+    {
+      "epoch": 0.0719910011248594,
+      "grad_norm": 5.10213288638439,
+      "learning_rate": 1.1428571428571429e-05,
+      "loss": 0.9316,
+      "step": 8
+    },
+    {
+      "epoch": 0.08098987626546682,
+      "grad_norm": 5.371042207893317,
+      "learning_rate": 1.2857142857142859e-05,
+      "loss": 0.9182,
+      "step": 9
+    },
+    {
+      "epoch": 0.08998875140607424,
+      "grad_norm": 4.422548992141668,
+      "learning_rate": 1.4285714285714287e-05,
+      "loss": 0.9052,
+      "step": 10
+    },
+    {
+      "epoch": 0.09898762654668167,
+      "grad_norm": 2.908057739308309,
+      "learning_rate": 1.5714285714285715e-05,
+      "loss": 0.8738,
+      "step": 11
+    },
+    {
+      "epoch": 0.10798650168728909,
+      "grad_norm": 2.081438174306188,
+      "learning_rate": 1.7142857142857142e-05,
+      "loss": 0.8448,
+      "step": 12
+    },
+    {
+      "epoch": 0.11698537682789652,
+      "grad_norm": 1.8625844126627613,
+      "learning_rate": 1.8571428571428575e-05,
+      "loss": 0.8214,
+      "step": 13
+    },
+    {
+      "epoch": 0.12598425196850394,
+      "grad_norm": 1.4150826691986598,
+      "learning_rate": 2e-05,
+      "loss": 0.8118,
+      "step": 14
+    },
+    {
+      "epoch": 0.13498312710911137,
+      "grad_norm": 1.3470926001768453,
+      "learning_rate": 2.1428571428571428e-05,
+      "loss": 0.7967,
+      "step": 15
+    },
+    {
+      "epoch": 0.1439820022497188,
+      "grad_norm": 1.318138672473746,
+      "learning_rate": 2.2857142857142858e-05,
+      "loss": 0.7817,
+      "step": 16
+    },
+    {
+      "epoch": 0.1529808773903262,
+      "grad_norm": 1.132045894481426,
+      "learning_rate": 2.4285714285714285e-05,
+      "loss": 0.7702,
+      "step": 17
+    },
+    {
+      "epoch": 0.16197975253093364,
+      "grad_norm": 1.1216235493173696,
+      "learning_rate": 2.5714285714285718e-05,
+      "loss": 0.7625,
+      "step": 18
+    },
+    {
+      "epoch": 0.17097862767154107,
+      "grad_norm": 1.2949162303101338,
+      "learning_rate": 2.7142857142857148e-05,
+      "loss": 0.7552,
+      "step": 19
+    },
+    {
+      "epoch": 0.17997750281214847,
+      "grad_norm": 1.0018485261634034,
+      "learning_rate": 2.8571428571428574e-05,
+      "loss": 0.748,
+      "step": 20
+    },
+    {
+      "epoch": 0.1889763779527559,
+      "grad_norm": 0.9642142358632533,
+      "learning_rate": 3.0000000000000004e-05,
+      "loss": 0.7319,
+      "step": 21
+    },
+    {
+      "epoch": 0.19797525309336333,
+      "grad_norm": 1.3162346092752937,
+      "learning_rate": 3.142857142857143e-05,
+      "loss": 0.7376,
+      "step": 22
+    },
+    {
+      "epoch": 0.20697412823397077,
+      "grad_norm": 1.2462908815710223,
+      "learning_rate": 3.285714285714286e-05,
+      "loss": 0.7225,
+      "step": 23
+    },
+    {
+      "epoch": 0.21597300337457817,
+      "grad_norm": 1.1288729992725366,
+      "learning_rate": 3.4285714285714284e-05,
+      "loss": 0.7232,
+      "step": 24
+    },
+    {
+      "epoch": 0.2249718785151856,
+      "grad_norm": 1.2886081610849456,
+      "learning_rate": 3.571428571428572e-05,
+      "loss": 0.7235,
+      "step": 25
+    },
+    {
+      "epoch": 0.23397075365579303,
+      "grad_norm": 1.377579395059948,
+      "learning_rate": 3.714285714285715e-05,
+      "loss": 0.7136,
+      "step": 26
+    },
+    {
+      "epoch": 0.24296962879640044,
+      "grad_norm": 1.0739406693404967,
+      "learning_rate": 3.857142857142858e-05,
+      "loss": 0.7138,
+      "step": 27
+    },
+    {
+      "epoch": 0.25196850393700787,
+      "grad_norm": 1.2160719105328928,
+      "learning_rate": 4e-05,
+      "loss": 0.7106,
+      "step": 28
+    },
+    {
+      "epoch": 0.2609673790776153,
+      "grad_norm": 1.3411926888307302,
+      "learning_rate": 4.1428571428571437e-05,
+      "loss": 0.7056,
+      "step": 29
+    },
+    {
+      "epoch": 0.26996625421822273,
+      "grad_norm": 0.8344443792471369,
+      "learning_rate": 4.2857142857142856e-05,
+      "loss": 0.6906,
+      "step": 30
+    },
+    {
+      "epoch": 0.27896512935883017,
+      "grad_norm": 1.6378121927259697,
+      "learning_rate": 4.428571428571429e-05,
+      "loss": 0.7022,
+      "step": 31
+    },
+    {
+      "epoch": 0.2879640044994376,
+      "grad_norm": 1.312672534168789,
+      "learning_rate": 4.5714285714285716e-05,
+      "loss": 0.6925,
+      "step": 32
+    },
+    {
+      "epoch": 0.296962879640045,
+      "grad_norm": 1.1997985746020943,
+      "learning_rate": 4.714285714285715e-05,
+      "loss": 0.6894,
+      "step": 33
+    },
+    {
+      "epoch": 0.3059617547806524,
+      "grad_norm": 1.3374680408068491,
+      "learning_rate": 4.857142857142857e-05,
+      "loss": 0.6924,
+      "step": 34
+    },
+    {
+      "epoch": 0.31496062992125984,
+      "grad_norm": 0.9866632538970782,
+      "learning_rate": 5e-05,
+      "loss": 0.6856,
+      "step": 35
+    },
+    {
+      "epoch": 0.32395950506186727,
+      "grad_norm": 1.7617767807118945,
+      "learning_rate": 5.1428571428571436e-05,
+      "loss": 0.6882,
+      "step": 36
+    },
+    {
+      "epoch": 0.3329583802024747,
+      "grad_norm": 1.2162422657164105,
+      "learning_rate": 5.285714285714286e-05,
+      "loss": 0.6827,
+      "step": 37
+    },
+    {
+      "epoch": 0.34195725534308213,
+      "grad_norm": 1.4381817884560075,
+      "learning_rate": 5.4285714285714295e-05,
+      "loss": 0.6804,
+      "step": 38
+    },
+    {
+      "epoch": 0.35095613048368957,
+      "grad_norm": 1.1994316767887803,
+      "learning_rate": 5.5714285714285715e-05,
+      "loss": 0.675,
+      "step": 39
+    },
+    {
+      "epoch": 0.35995500562429694,
+      "grad_norm": 1.3859110222701834,
+      "learning_rate": 5.714285714285715e-05,
+      "loss": 0.6714,
+      "step": 40
+    },
+    {
+      "epoch": 0.3689538807649044,
+      "grad_norm": 1.0456569888416478,
+      "learning_rate": 5.8571428571428575e-05,
+      "loss": 0.666,
+      "step": 41
+    },
+    {
+      "epoch": 0.3779527559055118,
+      "grad_norm": 1.550645312960814,
+      "learning_rate": 6.000000000000001e-05,
+      "loss": 0.6716,
+      "step": 42
+    },
+    {
+      "epoch": 0.38695163104611924,
+      "grad_norm": 1.7520944490642492,
+      "learning_rate": 6.142857142857143e-05,
+      "loss": 0.6668,
+      "step": 43
+    },
+    {
+      "epoch": 0.39595050618672667,
+      "grad_norm": 1.220392434494151,
+      "learning_rate": 6.285714285714286e-05,
+      "loss": 0.6706,
+      "step": 44
+    },
+    {
+      "epoch": 0.4049493813273341,
+      "grad_norm": 1.5640678258841063,
+      "learning_rate": 6.428571428571429e-05,
+      "loss": 0.6666,
+      "step": 45
+    },
+    {
+      "epoch": 0.41394825646794153,
+      "grad_norm": 1.0339470330368512,
+      "learning_rate": 6.571428571428571e-05,
+      "loss": 0.6611,
+      "step": 46
+    },
+    {
+      "epoch": 0.4229471316085489,
+      "grad_norm": 1.5630693263603568,
+      "learning_rate": 6.714285714285715e-05,
+      "loss": 0.6654,
+      "step": 47
+    },
+    {
+      "epoch": 0.43194600674915634,
+      "grad_norm": 1.4317212538586248,
+      "learning_rate": 6.857142857142857e-05,
+      "loss": 0.666,
+      "step": 48
+    },
+    {
+      "epoch": 0.4409448818897638,
+      "grad_norm": 1.2652376704046324,
+      "learning_rate": 7.000000000000001e-05,
+      "loss": 0.6525,
+      "step": 49
+    },
+    {
+      "epoch": 0.4499437570303712,
+      "grad_norm": 1.9389193690579092,
+      "learning_rate": 7.142857142857143e-05,
+      "loss": 0.6706,
+      "step": 50
+    },
+    {
+      "epoch": 0.45894263217097864,
+      "grad_norm": 1.2095199143374322,
+      "learning_rate": 7.285714285714286e-05,
+      "loss": 0.6578,
+      "step": 51
+    },
+    {
+      "epoch": 0.46794150731158607,
+      "grad_norm": 2.1530205341425943,
+      "learning_rate": 7.42857142857143e-05,
+      "loss": 0.6659,
+      "step": 52
+    },
+    {
+      "epoch": 0.4769403824521935,
+      "grad_norm": 1.3498447393418538,
+      "learning_rate": 7.571428571428571e-05,
+      "loss": 0.6555,
+      "step": 53
+    },
+    {
+      "epoch": 0.4859392575928009,
+      "grad_norm": 1.7344034955388106,
+      "learning_rate": 7.714285714285715e-05,
+      "loss": 0.6655,
+      "step": 54
+    },
+    {
+      "epoch": 0.4949381327334083,
+      "grad_norm": 2.165550267732371,
+      "learning_rate": 7.857142857142858e-05,
+      "loss": 0.6593,
+      "step": 55
+    },
+    {
+      "epoch": 0.5039370078740157,
+      "grad_norm": 1.6288021163787398,
+      "learning_rate": 8e-05,
+      "loss": 0.6541,
+      "step": 56
+    },
+    {
+      "epoch": 0.5129358830146231,
+      "grad_norm": 1.6669527952524703,
+      "learning_rate": 7.999920726649282e-05,
+      "loss": 0.6421,
+      "step": 57
+    },
+    {
+      "epoch": 0.5219347581552306,
+      "grad_norm": 1.2703954706550948,
+      "learning_rate": 7.999682909739257e-05,
+      "loss": 0.6416,
+      "step": 58
+    },
+    {
+      "epoch": 0.530933633295838,
+      "grad_norm": 1.6224554752837215,
+      "learning_rate": 7.999286558696199e-05,
+      "loss": 0.6427,
+      "step": 59
+    },
+    {
+      "epoch": 0.5399325084364455,
+      "grad_norm": 1.3278739806750215,
+      "learning_rate": 7.998731689230145e-05,
+      "loss": 0.6535,
+      "step": 60
+    },
+    {
+      "epoch": 0.5489313835770528,
+      "grad_norm": 1.9995409784489284,
+      "learning_rate": 7.998018323334275e-05,
+      "loss": 0.6496,
+      "step": 61
+    },
+    {
+      "epoch": 0.5579302587176603,
+      "grad_norm": 1.5356642072108388,
+      "learning_rate": 7.997146489284042e-05,
+      "loss": 0.6437,
+      "step": 62
+    },
+    {
+      "epoch": 0.5669291338582677,
+      "grad_norm": 1.256225547258712,
+      "learning_rate": 7.996116221636049e-05,
+      "loss": 0.6478,
+      "step": 63
+    },
+    {
+      "epoch": 0.5759280089988752,
+      "grad_norm": 1.7177343714374778,
+      "learning_rate": 7.994927561226682e-05,
+      "loss": 0.641,
+      "step": 64
+    },
+    {
+      "epoch": 0.5849268841394826,
+      "grad_norm": 1.8288469591109016,
+      "learning_rate": 7.993580555170486e-05,
+      "loss": 0.6483,
+      "step": 65
+    },
+    {
+      "epoch": 0.59392575928009,
+      "grad_norm": 0.9174977020863746,
+      "learning_rate": 7.992075256858302e-05,
+      "loss": 0.639,
+      "step": 66
+    },
+    {
+      "epoch": 0.6029246344206974,
+      "grad_norm": 2.3068233658832846,
+      "learning_rate": 7.990411725955153e-05,
+      "loss": 0.6469,
+      "step": 67
+    },
+    {
+      "epoch": 0.6119235095613048,
+      "grad_norm": 1.331580694798539,
+      "learning_rate": 7.98859002839787e-05,
+      "loss": 0.6403,
+      "step": 68
+    },
+    {
+      "epoch": 0.6209223847019123,
+      "grad_norm": 1.9556860563217953,
+      "learning_rate": 7.986610236392491e-05,
+      "loss": 0.6436,
+      "step": 69
+    },
+    {
+      "epoch": 0.6299212598425197,
+      "grad_norm": 1.4030777265618954,
+      "learning_rate": 7.984472428411388e-05,
+      "loss": 0.6541,
+      "step": 70
+    },
+    {
+      "epoch": 0.6389201349831272,
+      "grad_norm": 1.1120271583327745,
+      "learning_rate": 7.98217668919016e-05,
+      "loss": 0.6408,
+      "step": 71
+    },
+    {
+      "epoch": 0.6479190101237345,
+      "grad_norm": 1.7515091640927072,
+      "learning_rate": 7.97972310972428e-05,
+      "loss": 0.6462,
+      "step": 72
+    },
+    {
+      "epoch": 0.6569178852643419,
+      "grad_norm": 1.2154666389278914,
+      "learning_rate": 7.977111787265479e-05,
+      "loss": 0.6348,
+      "step": 73
+    },
+    {
+      "epoch": 0.6659167604049494,
+      "grad_norm": 1.1948666759848607,
+      "learning_rate": 7.9743428253179e-05,
+      "loss": 0.6398,
+      "step": 74
+    },
+    {
+      "epoch": 0.6749156355455568,
+      "grad_norm": 1.4465768500981744,
+      "learning_rate": 7.971416333633984e-05,
+      "loss": 0.633,
+      "step": 75
+    },
+    {
+      "epoch": 0.6839145106861643,
+      "grad_norm": 0.7079358166882018,
+      "learning_rate": 7.968332428210136e-05,
+      "loss": 0.6344,
+      "step": 76
+    },
+    {
+      "epoch": 0.6929133858267716,
+      "grad_norm": 1.2354926269354385,
+      "learning_rate": 7.965091231282114e-05,
+      "loss": 0.6316,
+      "step": 77
+    },
+    {
+      "epoch": 0.7019122609673791,
+      "grad_norm": 0.8980787882545499,
+      "learning_rate": 7.961692871320186e-05,
+      "loss": 0.6384,
+      "step": 78
+    },
+    {
+      "epoch": 0.7109111361079865,
+      "grad_norm": 1.327053450288205,
+      "learning_rate": 7.958137483024044e-05,
+      "loss": 0.6348,
+      "step": 79
+    },
+    {
+      "epoch": 0.7199100112485939,
+      "grad_norm": 1.0872221047887007,
+      "learning_rate": 7.95442520731746e-05,
+      "loss": 0.6319,
+      "step": 80
+    },
+    {
+      "epoch": 0.7289088863892014,
+      "grad_norm": 1.0406772088732106,
+      "learning_rate": 7.9505561913427e-05,
+      "loss": 0.6255,
+      "step": 81
+    },
+    {
+      "epoch": 0.7379077615298087,
+      "grad_norm": 1.3043796341854046,
+      "learning_rate": 7.946530588454695e-05,
+      "loss": 0.6324,
+      "step": 82
+    },
+    {
+      "epoch": 0.7469066366704162,
+      "grad_norm": 1.0289090852613956,
+      "learning_rate": 7.942348558214958e-05,
+      "loss": 0.6326,
+      "step": 83
+    },
+    {
+      "epoch": 0.7559055118110236,
+      "grad_norm": 1.309994146626142,
+      "learning_rate": 7.938010266385268e-05,
+      "loss": 0.6362,
+      "step": 84
+    },
+    {
+      "epoch": 0.7649043869516311,
+      "grad_norm": 0.9949679280573976,
+      "learning_rate": 7.933515884921086e-05,
+      "loss": 0.6274,
+      "step": 85
+    },
+    {
+      "epoch": 0.7739032620922385,
+      "grad_norm": 1.3529279590081953,
+      "learning_rate": 7.928865591964751e-05,
+      "loss": 0.6356,
+      "step": 86
+    },
+    {
+      "epoch": 0.7829021372328459,
+      "grad_norm": 1.5394293742781517,
+      "learning_rate": 7.924059571838419e-05,
+      "loss": 0.6239,
+      "step": 87
+    },
+    {
+      "epoch": 0.7919010123734533,
+      "grad_norm": 0.8230315586506134,
+      "learning_rate": 7.919098015036746e-05,
+      "loss": 0.6254,
+      "step": 88
+    },
+    {
+      "epoch": 0.8008998875140607,
+      "grad_norm": 2.5412889042558646,
+      "learning_rate": 7.91398111821935e-05,
+      "loss": 0.6386,
+      "step": 89
+    },
+    {
+      "epoch": 0.8098987626546682,
+      "grad_norm": 1.6534609255338006,
+      "learning_rate": 7.908709084203006e-05,
+      "loss": 0.6433,
+      "step": 90
+    },
+    {
+      "epoch": 0.8188976377952756,
+      "grad_norm": 2.115550172588045,
+      "learning_rate": 7.903282121953619e-05,
+      "loss": 0.6274,
+      "step": 91
+    },
+    {
+      "epoch": 0.8278965129358831,
+      "grad_norm": 2.21745507570489,
+      "learning_rate": 7.897700446577928e-05,
+      "loss": 0.6395,
+      "step": 92
+    },
+    {
+      "epoch": 0.8368953880764904,
+      "grad_norm": 1.1348223149712242,
+      "learning_rate": 7.891964279314988e-05,
+      "loss": 0.6304,
+      "step": 93
+    },
+    {
+      "epoch": 0.8458942632170978,
+      "grad_norm": 1.1093790619140609,
+      "learning_rate": 7.886073847527397e-05,
+      "loss": 0.634,
+      "step": 94
+    },
+    {
+      "epoch": 0.8548931383577053,
+      "grad_norm": 1.0247299216217371,
+      "learning_rate": 7.88002938469229e-05,
+      "loss": 0.6187,
+      "step": 95
+    },
+    {
+      "epoch": 0.8638920134983127,
+      "grad_norm": 0.9690359150703725,
+      "learning_rate": 7.873831130392077e-05,
+      "loss": 0.6224,
+      "step": 96
+    },
+    {
+      "epoch": 0.8728908886389202,
+      "grad_norm": 0.8479503109980445,
+      "learning_rate": 7.867479330304951e-05,
+      "loss": 0.6129,
+      "step": 97
+    },
+    {
+      "epoch": 0.8818897637795275,
+      "grad_norm": 0.7677468105242354,
+      "learning_rate": 7.860974236195151e-05,
+      "loss": 0.6133,
+      "step": 98
+    },
+    {
+      "epoch": 0.890888638920135,
+      "grad_norm": 1.0066107483900497,
+      "learning_rate": 7.85431610590298e-05,
+      "loss": 0.6212,
+      "step": 99
+    },
+    {
+      "epoch": 0.8998875140607424,
+      "grad_norm": 0.8385258369213318,
+      "learning_rate": 7.847505203334588e-05,
+      "loss": 0.6162,
+      "step": 100
+    },
+    {
+      "epoch": 0.9088863892013498,
+      "grad_norm": 0.8097143309117704,
+      "learning_rate": 7.840541798451506e-05,
+      "loss": 0.6105,
+      "step": 101
+    },
+    {
+      "epoch": 0.9178852643419573,
+      "grad_norm": 0.6217223076576118,
+      "learning_rate": 7.833426167259955e-05,
+      "loss": 0.6091,
+      "step": 102
+    },
+    {
+      "epoch": 0.9268841394825647,
+      "grad_norm": 0.8171222793279067,
+      "learning_rate": 7.826158591799898e-05,
+      "loss": 0.6124,
+      "step": 103
+    },
+    {
+      "epoch": 0.9358830146231721,
+      "grad_norm": 0.825619425206217,
+      "learning_rate": 7.818739360133863e-05,
+      "loss": 0.6093,
+      "step": 104
+    },
+    {
+      "epoch": 0.9448818897637795,
+      "grad_norm": 0.9916352114638577,
+      "learning_rate": 7.811168766335531e-05,
+      "loss": 0.6152,
+      "step": 105
+    },
+    {
+      "epoch": 0.953880764904387,
+      "grad_norm": 1.0337582772307539,
+      "learning_rate": 7.803447110478067e-05,
+      "loss": 0.6183,
+      "step": 106
+    },
+    {
+      "epoch": 0.9628796400449944,
+      "grad_norm": 0.9165715880682846,
+      "learning_rate": 7.795574698622237e-05,
+      "loss": 0.6176,
+      "step": 107
+    },
+    {
+      "epoch": 0.9718785151856018,
+      "grad_norm": 0.6530649555546069,
+      "learning_rate": 7.787551842804276e-05,
+      "loss": 0.6099,
+      "step": 108
+    },
+    {
+      "epoch": 0.9808773903262092,
+      "grad_norm": 0.7668074713592107,
+      "learning_rate": 7.779378861023516e-05,
+      "loss": 0.6063,
+      "step": 109
+    },
+    {
+      "epoch": 0.9898762654668166,
+      "grad_norm": 0.715515594924498,
+      "learning_rate": 7.77105607722978e-05,
+      "loss": 0.6088,
+      "step": 110
+    },
+    {
+      "epoch": 0.9988751406074241,
+      "grad_norm": 0.7194843087763119,
+      "learning_rate": 7.762583821310548e-05,
+      "loss": 0.6027,
+      "step": 111
+    },
+    {
+      "epoch": 1.0078740157480315,
+      "grad_norm": 1.3373684763348388,
+      "learning_rate": 7.753962429077881e-05,
+      "loss": 1.109,
+      "step": 112
+    },
+    {
+      "epoch": 1.0168728908886389,
+      "grad_norm": 1.2665705644041705,
+      "learning_rate": 7.7451922422551e-05,
+      "loss": 0.5896,
+      "step": 113
+    },
+    {
+      "epoch": 1.0258717660292462,
+      "grad_norm": 0.7146916740896777,
+      "learning_rate": 7.736273608463253e-05,
+      "loss": 0.5776,
+      "step": 114
+    },
+    {
+      "epoch": 1.0348706411698538,
+      "grad_norm": 0.8865840936771311,
+      "learning_rate": 7.727206881207334e-05,
+      "loss": 0.5907,
+      "step": 115
+    },
+    {
+      "epoch": 1.0438695163104612,
+      "grad_norm": 0.7277897994412549,
+      "learning_rate": 7.717992419862268e-05,
+      "loss": 0.5873,
+      "step": 116
+    },
+    {
+      "epoch": 1.0528683914510686,
+      "grad_norm": 0.6536923705270321,
+      "learning_rate": 7.708630589658667e-05,
+      "loss": 0.5835,
+      "step": 117
+    },
+    {
+      "epoch": 1.061867266591676,
+      "grad_norm": 0.7162373408409907,
+      "learning_rate": 7.699121761668355e-05,
+      "loss": 0.5794,
+      "step": 118
+    },
+    {
+      "epoch": 1.0708661417322836,
+      "grad_norm": 1.0254372509038927,
+      "learning_rate": 7.689466312789661e-05,
+      "loss": 0.582,
+      "step": 119
+    },
+    {
+      "epoch": 1.079865016872891,
+      "grad_norm": 1.1137703756989226,
+      "learning_rate": 7.679664625732478e-05,
+      "loss": 0.5907,
+      "step": 120
+    },
+    {
+      "epoch": 1.0888638920134983,
+      "grad_norm": 0.5744577610457379,
+      "learning_rate": 7.669717089003094e-05,
+      "loss": 0.5742,
+      "step": 121
+    },
+    {
+      "epoch": 1.0978627671541057,
+      "grad_norm": 0.6451796147865981,
+      "learning_rate": 7.659624096888792e-05,
+      "loss": 0.5831,
+      "step": 122
+    },
+    {
+      "epoch": 1.106861642294713,
+      "grad_norm": 0.6276129538814031,
+      "learning_rate": 7.649386049442223e-05,
+      "loss": 0.5827,
+      "step": 123
+    },
+    {
+      "epoch": 1.1158605174353207,
+      "grad_norm": 0.7718645523573185,
+      "learning_rate": 7.639003352465551e-05,
+      "loss": 0.5745,
+      "step": 124
+    },
+    {
+      "epoch": 1.124859392575928,
+      "grad_norm": 0.9206192827571739,
+      "learning_rate": 7.628476417494368e-05,
+      "loss": 0.5839,
+      "step": 125
+    },
+    {
+      "epoch": 1.1338582677165354,
+      "grad_norm": 0.698663737317266,
+      "learning_rate": 7.617805661781374e-05,
+      "loss": 0.5754,
+      "step": 126
+    },
+    {
+      "epoch": 1.1428571428571428,
+      "grad_norm": 0.5240414441327959,
+      "learning_rate": 7.60699150827985e-05,
+      "loss": 0.5772,
+      "step": 127
+    },
+    {
+      "epoch": 1.1518560179977504,
+      "grad_norm": 0.5843162509749283,
+      "learning_rate": 7.596034385626888e-05,
+      "loss": 0.5727,
+      "step": 128
+    },
+    {
+      "epoch": 1.1608548931383578,
+      "grad_norm": 0.7391156549586344,
+      "learning_rate": 7.584934728126403e-05,
+      "loss": 0.5736,
+      "step": 129
+    },
+    {
+      "epoch": 1.1698537682789651,
+      "grad_norm": 0.6964429639639632,
+      "learning_rate": 7.573692975731914e-05,
+      "loss": 0.5677,
+      "step": 130
+    },
+    {
+      "epoch": 1.1788526434195725,
+      "grad_norm": 0.5118585657523064,
+      "learning_rate": 7.562309574029112e-05,
+      "loss": 0.5796,
+      "step": 131
+    },
+    {
+      "epoch": 1.18785151856018,
+      "grad_norm": 0.3947628953294446,
+      "learning_rate": 7.550784974218195e-05,
+      "loss": 0.5669,
+      "step": 132
+    },
+    {
+      "epoch": 1.1968503937007875,
+      "grad_norm": 0.37395524397542534,
+      "learning_rate": 7.539119633095983e-05,
+      "loss": 0.5641,
+      "step": 133
+    },
+    {
+      "epoch": 1.2058492688413949,
+      "grad_norm": 0.37361515124611633,
+      "learning_rate": 7.527314013037815e-05,
+      "loss": 0.5763,
+      "step": 134
+    },
+    {
+      "epoch": 1.2148481439820022,
+      "grad_norm": 0.4839280159646048,
+      "learning_rate": 7.515368581979224e-05,
+      "loss": 0.5703,
+      "step": 135
+    },
+    {
+      "epoch": 1.2238470191226096,
+      "grad_norm": 0.6174733585746707,
+      "learning_rate": 7.503283813397379e-05,
+      "loss": 0.5703,
+      "step": 136
+    },
+    {
+      "epoch": 1.232845894263217,
+      "grad_norm": 0.7319038998136339,
+      "learning_rate": 7.491060186292331e-05,
+      "loss": 0.5715,
+      "step": 137
+    },
+    {
+      "epoch": 1.2418447694038246,
+      "grad_norm": 0.7864687770694577,
+      "learning_rate": 7.478698185168019e-05,
+      "loss": 0.5714,
+      "step": 138
+    },
+    {
+      "epoch": 1.250843644544432,
+      "grad_norm": 0.7693307388456105,
+      "learning_rate": 7.466198300013066e-05,
+      "loss": 0.5761,
+      "step": 139
+    },
+    {
+      "epoch": 1.2598425196850394,
+      "grad_norm": 0.7329449732496208,
+      "learning_rate": 7.453561026281366e-05,
+      "loss": 0.5697,
+      "step": 140
+    },
+    {
+      "epoch": 1.2688413948256467,
+      "grad_norm": 0.7219015663079241,
+      "learning_rate": 7.440786864872433e-05,
+      "loss": 0.5725,
+      "step": 141
+    },
+    {
+      "epoch": 1.277840269966254,
+      "grad_norm": 0.7161519080047792,
+      "learning_rate": 7.427876322111558e-05,
+      "loss": 0.5737,
+      "step": 142
+    },
+    {
+      "epoch": 1.2868391451068617,
+      "grad_norm": 0.7985855229347674,
+      "learning_rate": 7.414829909729727e-05,
+      "loss": 0.5724,
+      "step": 143
+    },
+    {
+      "epoch": 1.295838020247469,
+      "grad_norm": 0.8867813272578328,
+      "learning_rate": 7.40164814484336e-05,
+      "loss": 0.5725,
+      "step": 144
+    },
+    {
+      "epoch": 1.3048368953880765,
+      "grad_norm": 0.8113361068419839,
+      "learning_rate": 7.388331549933787e-05,
+      "loss": 0.5743,
+      "step": 145
+    },
+    {
+      "epoch": 1.3138357705286838,
+      "grad_norm": 0.5476904687412559,
+      "learning_rate": 7.37488065282656e-05,
+      "loss": 0.5658,
+      "step": 146
+    },
+    {
+      "epoch": 1.3228346456692912,
+      "grad_norm": 0.45189065402105816,
+      "learning_rate": 7.361295986670522e-05,
+      "loss": 0.5708,
+      "step": 147
+    },
+    {
+      "epoch": 1.3318335208098988,
+      "grad_norm": 0.751469386998812,
+      "learning_rate": 7.347578089916672e-05,
+      "loss": 0.5629,
+      "step": 148
+    },
+    {
+      "epoch": 1.3408323959505062,
+      "grad_norm": 0.7162670873895776,
+      "learning_rate": 7.333727506296831e-05,
+      "loss": 0.5711,
+      "step": 149
+    },
+    {
+      "epoch": 1.3498312710911136,
+      "grad_norm": 0.5796332830646129,
+      "learning_rate": 7.319744784802087e-05,
+      "loss": 0.5701,
+      "step": 150
+    },
+    {
+      "epoch": 1.3588301462317212,
+      "grad_norm": 0.5806837398044375,
+      "learning_rate": 7.305630479661033e-05,
+      "loss": 0.568,
+      "step": 151
+    },
+    {
+      "epoch": 1.3678290213723285,
+      "grad_norm": 0.36656011961132856,
+      "learning_rate": 7.291385150317796e-05,
+      "loss": 0.5704,
+      "step": 152
+    },
+    {
+      "epoch": 1.376827896512936,
+      "grad_norm": 0.4854553631469011,
+      "learning_rate": 7.277009361409874e-05,
+      "loss": 0.5626,
+      "step": 153
+    },
+    {
+      "epoch": 1.3858267716535433,
+      "grad_norm": 0.7060268694492198,
+      "learning_rate": 7.262503682745744e-05,
+      "loss": 0.5682,
+      "step": 154
+    },
+    {
+      "epoch": 1.3948256467941507,
+      "grad_norm": 0.6290468975159552,
+      "learning_rate": 7.247868689282283e-05,
+      "loss": 0.5652,
+      "step": 155
+    },
+    {
+      "epoch": 1.4038245219347583,
+      "grad_norm": 0.5124920327952127,
+      "learning_rate": 7.233104961101974e-05,
+      "loss": 0.5702,
+      "step": 156
+    },
+    {
+      "epoch": 1.4128233970753656,
+      "grad_norm": 0.435666421319146,
+      "learning_rate": 7.21821308338992e-05,
+      "loss": 0.5656,
+      "step": 157
+    },
+    {
+      "epoch": 1.421822272215973,
+      "grad_norm": 0.31789150940846095,
+      "learning_rate": 7.203193646410642e-05,
+      "loss": 0.5625,
+      "step": 158
+    },
+    {
+      "epoch": 1.4308211473565804,
+      "grad_norm": 0.44522630815692826,
+      "learning_rate": 7.188047245484686e-05,
+      "loss": 0.5622,
+      "step": 159
+    },
+    {
+      "epoch": 1.4398200224971878,
+      "grad_norm": 0.5340918126651806,
+      "learning_rate": 7.172774480965033e-05,
+      "loss": 0.5663,
+      "step": 160
+    },
+    {
+      "epoch": 1.4488188976377954,
+      "grad_norm": 0.5126590833097094,
+      "learning_rate": 7.157375958213288e-05,
+      "loss": 0.5604,
+      "step": 161
+    },
+    {
+      "epoch": 1.4578177727784027,
+      "grad_norm": 0.5574211540917228,
+      "learning_rate": 7.141852287575701e-05,
+      "loss": 0.5644,
+      "step": 162
+    },
+    {
+      "epoch": 1.4668166479190101,
+      "grad_norm": 0.49894825274633764,
+      "learning_rate": 7.126204084358963e-05,
+      "loss": 0.5543,
+      "step": 163
+    },
+    {
+      "epoch": 1.4758155230596175,
+      "grad_norm": 0.383714620666216,
+      "learning_rate": 7.110431968805825e-05,
+      "loss": 0.5667,
+      "step": 164
+    },
+    {
+      "epoch": 1.4848143982002249,
+      "grad_norm": 0.3679127273669197,
+      "learning_rate": 7.094536566070514e-05,
+      "loss": 0.5649,
+      "step": 165
+    },
+    {
+      "epoch": 1.4938132733408325,
+      "grad_norm": 0.41052750978866387,
+      "learning_rate": 7.078518506193945e-05,
+      "loss": 0.5597,
+      "step": 166
+    },
+    {
+      "epoch": 1.5028121484814398,
+      "grad_norm": 0.4161268377444414,
+      "learning_rate": 7.062378424078758e-05,
+      "loss": 0.5683,
+      "step": 167
+    },
+    {
+      "epoch": 1.5118110236220472,
+      "grad_norm": 0.4143309696590428,
+      "learning_rate": 7.046116959464149e-05,
+      "loss": 0.5627,
+      "step": 168
+    },
+    {
+      "epoch": 1.5208098987626548,
+      "grad_norm": 0.47622613149080295,
+      "learning_rate": 7.02973475690051e-05,
+      "loss": 0.5619,
+      "step": 169
+    },
+    {
+      "epoch": 1.529808773903262,
+      "grad_norm": 0.5903824719429629,
+      "learning_rate": 7.013232465723888e-05,
+      "loss": 0.5685,
+      "step": 170
+    },
+    {
+      "epoch": 1.5388076490438696,
+      "grad_norm": 0.6410524866967787,
+      "learning_rate": 6.996610740030237e-05,
+      "loss": 0.5649,
+      "step": 171
+    },
+    {
+      "epoch": 1.547806524184477,
+      "grad_norm": 0.5120196634127667,
+      "learning_rate": 6.979870238649506e-05,
+      "loss": 0.5641,
+      "step": 172
+    },
+    {
+      "epoch": 1.5568053993250843,
+      "grad_norm": 0.46968763374354905,
+      "learning_rate": 6.963011625119514e-05,
+      "loss": 0.5607,
+      "step": 173
+    },
+    {
+      "epoch": 1.565804274465692,
+      "grad_norm": 0.7346438690156033,
+      "learning_rate": 6.94603556765965e-05,
+      "loss": 0.5632,
+      "step": 174
+    },
+    {
+      "epoch": 1.574803149606299,
+      "grad_norm": 0.9133676726822815,
+      "learning_rate": 6.928942739144394e-05,
+      "loss": 0.5561,
+      "step": 175
+    },
+    {
+      "epoch": 1.5838020247469067,
+      "grad_norm": 0.9699546226599123,
+      "learning_rate": 6.911733817076638e-05,
+      "loss": 0.5684,
+      "step": 176
+    },
+    {
+      "epoch": 1.592800899887514,
+      "grad_norm": 0.8546409930461957,
+      "learning_rate": 6.894409483560845e-05,
+      "loss": 0.5636,
+      "step": 177
+    },
+    {
+      "epoch": 1.6017997750281214,
+      "grad_norm": 0.5767270121981363,
+      "learning_rate": 6.876970425275993e-05,
+      "loss": 0.5636,
+      "step": 178
+    },
+    {
+      "epoch": 1.610798650168729,
+      "grad_norm": 0.4603868453512875,
+      "learning_rate": 6.859417333448376e-05,
+      "loss": 0.5657,
+      "step": 179
+    },
+    {
+      "epoch": 1.6197975253093362,
+      "grad_norm": 0.4657088093589826,
+      "learning_rate": 6.841750903824196e-05,
+      "loss": 0.553,
+      "step": 180
+    },
+    {
+      "epoch": 1.6287964004499438,
+      "grad_norm": 0.4245830377495125,
+      "learning_rate": 6.823971836641988e-05,
+      "loss": 0.553,
+      "step": 181
+    },
+    {
+      "epoch": 1.6377952755905512,
+      "grad_norm": 0.38313124117842345,
+      "learning_rate": 6.806080836604868e-05,
+      "loss": 0.5606,
+      "step": 182
+    },
+    {
+      "epoch": 1.6467941507311585,
+      "grad_norm": 0.3951074598265005,
+      "learning_rate": 6.788078612852596e-05,
+      "loss": 0.5611,
+      "step": 183
+    },
+    {
+      "epoch": 1.6557930258717661,
+      "grad_norm": 0.3324519467665917,
+      "learning_rate": 6.769965878933468e-05,
+      "loss": 0.5589,
+      "step": 184
+    },
+    {
+      "epoch": 1.6647919010123733,
+      "grad_norm": 0.30525105111986356,
+      "learning_rate": 6.751743352776041e-05,
+      "loss": 0.5513,
+      "step": 185
+    },
+    {
+      "epoch": 1.6737907761529809,
+      "grad_norm": 0.38317669421888395,
+      "learning_rate": 6.733411756660668e-05,
+      "loss": 0.561,
+      "step": 186
+    },
+    {
+      "epoch": 1.6827896512935883,
+      "grad_norm": 0.36452922899831897,
+      "learning_rate": 6.714971817190872e-05,
+      "loss": 0.5565,
+      "step": 187
+    },
+    {
+      "epoch": 1.6917885264341956,
+      "grad_norm": 0.45451136606739156,
+      "learning_rate": 6.696424265264549e-05,
+      "loss": 0.5554,
+      "step": 188
+    },
+    {
+      "epoch": 1.7007874015748032,
+      "grad_norm": 0.5850188409579592,
+      "learning_rate": 6.677769836044991e-05,
+      "loss": 0.5603,
+      "step": 189
+    },
+    {
+      "epoch": 1.7097862767154106,
+      "grad_norm": 0.6094742834131825,
+      "learning_rate": 6.659009268931756e-05,
+      "loss": 0.5574,
+      "step": 190
+    },
+    {
+      "epoch": 1.718785151856018,
+      "grad_norm": 0.6284162963180803,
+      "learning_rate": 6.64014330753135e-05,
+      "loss": 0.5507,
+      "step": 191
+    },
+    {
+      "epoch": 1.7277840269966256,
+      "grad_norm": 0.6446183014432966,
+      "learning_rate": 6.621172699627761e-05,
+      "loss": 0.5551,
+      "step": 192
+    },
+    {
+      "epoch": 1.7367829021372327,
+      "grad_norm": 0.6604713611773748,
+      "learning_rate": 6.602098197152817e-05,
+      "loss": 0.555,
+      "step": 193
+    },
+    {
+      "epoch": 1.7457817772778403,
+      "grad_norm": 0.7414347856141247,
+      "learning_rate": 6.582920556156378e-05,
+      "loss": 0.5631,
+      "step": 194
+    },
+    {
+      "epoch": 1.7547806524184477,
+      "grad_norm": 0.7773047456674285,
+      "learning_rate": 6.563640536776375e-05,
+      "loss": 0.56,
+      "step": 195
+    },
+    {
+      "epoch": 1.763779527559055,
+      "grad_norm": 0.6417760152496987,
+      "learning_rate": 6.544258903208679e-05,
+      "loss": 0.5518,
+      "step": 196
+    },
+    {
+      "epoch": 1.7727784026996627,
+      "grad_norm": 0.3607864424657309,
+      "learning_rate": 6.524776423676806e-05,
+      "loss": 0.55,
+      "step": 197
+    },
+    {
+      "epoch": 1.7817772778402698,
+      "grad_norm": 0.4683312923869542,
+      "learning_rate": 6.505193870401472e-05,
+      "loss": 0.5579,
+      "step": 198
+    },
+    {
+      "epoch": 1.7907761529808774,
+      "grad_norm": 0.6344422826239488,
+      "learning_rate": 6.485512019569986e-05,
+      "loss": 0.5628,
+      "step": 199
+    },
+    {
+      "epoch": 1.7997750281214848,
+      "grad_norm": 0.48250024105262346,
+      "learning_rate": 6.465731651305475e-05,
+      "loss": 0.5561,
+      "step": 200
+    },
+    {
+      "epoch": 1.8087739032620922,
+      "grad_norm": 0.4011620009633319,
+      "learning_rate": 6.445853549635982e-05,
+      "loss": 0.5589,
+      "step": 201
+    },
+    {
+      "epoch": 1.8177727784026998,
+      "grad_norm": 0.46120746083720787,
+      "learning_rate": 6.425878502463363e-05,
+      "loss": 0.5464,
+      "step": 202
+    },
+    {
+      "epoch": 1.826771653543307,
+      "grad_norm": 0.3366919397807797,
+      "learning_rate": 6.405807301532082e-05,
+      "loss": 0.5583,
+      "step": 203
+    },
+    {
+      "epoch": 1.8357705286839145,
+      "grad_norm": 0.3775528944092533,
+      "learning_rate": 6.38564074239781e-05,
+      "loss": 0.5515,
+      "step": 204
+    },
+    {
+      "epoch": 1.844769403824522,
+      "grad_norm": 0.4646001384485239,
+      "learning_rate": 6.365379624395911e-05,
+      "loss": 0.5545,
+      "step": 205
+    },
+    {
+      "epoch": 1.8537682789651293,
+      "grad_norm": 0.3173359890548618,
+      "learning_rate": 6.345024750609735e-05,
+      "loss": 0.5511,
+      "step": 206
+    },
+    {
+      "epoch": 1.862767154105737,
+      "grad_norm": 0.30771535160757224,
+      "learning_rate": 6.324576927838811e-05,
+      "loss": 0.5569,
+      "step": 207
+    },
+    {
+      "epoch": 1.871766029246344,
+      "grad_norm": 0.39138826318539643,
+      "learning_rate": 6.30403696656685e-05,
+      "loss": 0.5507,
+      "step": 208
+    },
+    {
+      "epoch": 1.8807649043869517,
+      "grad_norm": 0.39402910206893743,
+      "learning_rate": 6.28340568092963e-05,
+      "loss": 0.551,
+      "step": 209
+    },
+    {
+      "epoch": 1.889763779527559,
+      "grad_norm": 0.42067586027851417,
+      "learning_rate": 6.26268388868272e-05,
+      "loss": 0.5552,
+      "step": 210
+    },
+    {
+      "epoch": 1.8987626546681664,
+      "grad_norm": 0.39427161933244775,
+      "learning_rate": 6.241872411169075e-05,
+      "loss": 0.5515,
+      "step": 211
+    },
+    {
+      "epoch": 1.907761529808774,
+      "grad_norm": 0.2940525255889463,
+      "learning_rate": 6.220972073286469e-05,
+      "loss": 0.5452,
+      "step": 212
+    },
+    {
+      "epoch": 1.9167604049493814,
+      "grad_norm": 0.34048373966582285,
+      "learning_rate": 6.199983703454813e-05,
+      "loss": 0.5509,
+      "step": 213
+    },
+    {
+      "epoch": 1.9257592800899888,
+      "grad_norm": 0.299832377853422,
+      "learning_rate": 6.178908133583306e-05,
+      "loss": 0.5456,
+      "step": 214
+    },
+    {
+      "epoch": 1.9347581552305961,
+      "grad_norm": 0.2392599130105496,
+      "learning_rate": 6.157746199037473e-05,
+      "loss": 0.5525,
+      "step": 215
+    },
+    {
+      "epoch": 1.9437570303712035,
+      "grad_norm": 0.2884892083080534,
+      "learning_rate": 6.136498738606038e-05,
+      "loss": 0.5494,
+      "step": 216
+    },
+    {
+      "epoch": 1.952755905511811,
+      "grad_norm": 0.25915586035077326,
+      "learning_rate": 6.115166594467696e-05,
+      "loss": 0.5578,
+      "step": 217
+    },
+    {
+      "epoch": 1.9617547806524185,
+      "grad_norm": 0.2404861202851113,
+      "learning_rate": 6.093750612157719e-05,
+      "loss": 0.5501,
+      "step": 218
+    },
+    {
+      "epoch": 1.9707536557930259,
+      "grad_norm": 0.27079067912862353,
+      "learning_rate": 6.0722516405344436e-05,
+      "loss": 0.5544,
+      "step": 219
+    },
+    {
+      "epoch": 1.9797525309336335,
+      "grad_norm": 0.27569696549284556,
+      "learning_rate": 6.050670531745629e-05,
+      "loss": 0.5436,
+      "step": 220
+    },
+    {
+      "epoch": 1.9887514060742406,
+      "grad_norm": 0.39262044690390413,
+      "learning_rate": 6.0290081411946785e-05,
+      "loss": 0.5589,
+      "step": 221
+    },
+    {
+      "epoch": 1.9977502812148482,
+      "grad_norm": 0.3759638408126938,
+      "learning_rate": 6.007265327506734e-05,
+      "loss": 0.5489,
+      "step": 222
+    },
+    {
+      "epoch": 2.0067491563554554,
+      "grad_norm": 0.6721068777680985,
+      "learning_rate": 5.985442952494643e-05,
+      "loss": 0.9973,
+      "step": 223
+    },
+    {
+      "epoch": 2.015748031496063,
+      "grad_norm": 1.2532035078989137,
+      "learning_rate": 5.9635418811248e-05,
+      "loss": 0.5251,
+      "step": 224
+    },
+    {
+      "epoch": 2.0247469066366706,
+      "grad_norm": 1.3965146746879655,
+      "learning_rate": 5.941562981482859e-05,
+      "loss": 0.5222,
+      "step": 225
+    },
+    {
+      "epoch": 2.0337457817772777,
+      "grad_norm": 0.47181130829523693,
+      "learning_rate": 5.9195071247393325e-05,
+      "loss": 0.5175,
+      "step": 226
+    },
+    {
+      "epoch": 2.0427446569178853,
+      "grad_norm": 1.2945631540598115,
+      "learning_rate": 5.897375185115052e-05,
+      "loss": 0.5219,
+      "step": 227
+    },
+    {
+      "epoch": 2.0517435320584925,
+      "grad_norm": 0.753799335003198,
+      "learning_rate": 5.8751680398465244e-05,
+      "loss": 0.5248,
+      "step": 228
+    },
+    {
+      "epoch": 2.0607424071991,
+      "grad_norm": 0.7482718095479951,
+      "learning_rate": 5.8528865691511564e-05,
+      "loss": 0.531,
+      "step": 229
+    },
+    {
+      "epoch": 2.0697412823397077,
+      "grad_norm": 0.9072745588806806,
+      "learning_rate": 5.83053165619237e-05,
+      "loss": 0.5185,
+      "step": 230
+    },
+    {
+      "epoch": 2.078740157480315,
+      "grad_norm": 0.7594832390100079,
+      "learning_rate": 5.808104187044592e-05,
+      "loss": 0.5238,
+      "step": 231
+    },
+    {
+      "epoch": 2.0877390326209224,
+      "grad_norm": 0.7018923770640809,
+      "learning_rate": 5.785605050658134e-05,
+      "loss": 0.5116,
+      "step": 232
+    },
+    {
+      "epoch": 2.09673790776153,
+      "grad_norm": 0.6538828939619331,
+      "learning_rate": 5.7630351388239654e-05,
+      "loss": 0.5188,
+      "step": 233
+    },
+    {
+      "epoch": 2.105736782902137,
+      "grad_norm": 0.5009696448905507,
+      "learning_rate": 5.7403953461383515e-05,
+      "loss": 0.5189,
+      "step": 234
+    },
+    {
+      "epoch": 2.1147356580427448,
+      "grad_norm": 0.5418468893978772,
+      "learning_rate": 5.717686569967406e-05,
+      "loss": 0.513,
+      "step": 235
+    },
+    {
+      "epoch": 2.123734533183352,
+      "grad_norm": 0.5552312277008503,
+      "learning_rate": 5.694909710411517e-05,
+      "loss": 0.5136,
+      "step": 236
+    },
+    {
+      "epoch": 2.1327334083239595,
+      "grad_norm": 0.4618715663984918,
+      "learning_rate": 5.672065670269674e-05,
+      "loss": 0.5226,
+      "step": 237
+    },
+    {
+      "epoch": 2.141732283464567,
+      "grad_norm": 0.4634338024463565,
+      "learning_rate": 5.649155355003677e-05,
+      "loss": 0.517,
+      "step": 238
+    },
+    {
+      "epoch": 2.1507311586051743,
+      "grad_norm": 0.35434206430397164,
+      "learning_rate": 5.6261796727022575e-05,
+      "loss": 0.5166,
+      "step": 239
+    },
+    {
+      "epoch": 2.159730033745782,
+      "grad_norm": 0.46283864431852784,
+      "learning_rate": 5.603139534045075e-05,
+      "loss": 0.5112,
+      "step": 240
+    },
+    {
+      "epoch": 2.168728908886389,
+      "grad_norm": 0.42575715836926703,
+      "learning_rate": 5.5800358522666254e-05,
+      "loss": 0.5078,
+      "step": 241
+    },
+    {
+      "epoch": 2.1777277840269966,
+      "grad_norm": 0.33272119313147885,
+      "learning_rate": 5.556869543120043e-05,
+      "loss": 0.5194,
+      "step": 242
+    },
+    {
+      "epoch": 2.1867266591676042,
+      "grad_norm": 0.3307000934681126,
+      "learning_rate": 5.533641524840805e-05,
+      "loss": 0.5177,
+      "step": 243
+    },
+    {
+      "epoch": 2.1957255343082114,
+      "grad_norm": 0.3028678329115125,
+      "learning_rate": 5.5103527181103266e-05,
+      "loss": 0.5169,
+      "step": 244
+    },
+    {
+      "epoch": 2.204724409448819,
+      "grad_norm": 0.3208549750095541,
+      "learning_rate": 5.4870040460194834e-05,
+      "loss": 0.5166,
+      "step": 245
+    },
+    {
+      "epoch": 2.213723284589426,
+      "grad_norm": 0.24593852068559713,
+      "learning_rate": 5.463596434032011e-05,
+      "loss": 0.5162,
+      "step": 246
+    },
+    {
+      "epoch": 2.2227221597300337,
+      "grad_norm": 0.26893184750686033,
+      "learning_rate": 5.440130809947824e-05,
+      "loss": 0.5211,
+      "step": 247
+    },
+    {
+      "epoch": 2.2317210348706413,
+      "grad_norm": 0.2453740819753266,
+      "learning_rate": 5.4166081038662484e-05,
+      "loss": 0.5114,
+      "step": 248
+    },
+    {
+      "epoch": 2.2407199100112485,
+      "grad_norm": 0.22939493074869363,
+      "learning_rate": 5.3930292481491494e-05,
+      "loss": 0.5101,
+      "step": 249
+    },
+    {
+      "epoch": 2.249718785151856,
+      "grad_norm": 0.25733357689003183,
+      "learning_rate": 5.3693951773839736e-05,
+      "loss": 0.5111,
+      "step": 250
+    },
+    {
+      "epoch": 2.2587176602924632,
+      "grad_norm": 0.2252436543398025,
+      "learning_rate": 5.345706828346715e-05,
+      "loss": 0.516,
+      "step": 251
+    },
+    {
+      "epoch": 2.267716535433071,
+      "grad_norm": 0.28121431385301127,
+      "learning_rate": 5.3219651399647716e-05,
+      "loss": 0.5168,
+      "step": 252
+    },
+    {
+      "epoch": 2.2767154105736784,
+      "grad_norm": 0.2450809438398232,
+      "learning_rate": 5.298171053279739e-05,
+      "loss": 0.5089,
+      "step": 253
+    },
+    {
+      "epoch": 2.2857142857142856,
+      "grad_norm": 0.22424110667284097,
+      "learning_rate": 5.274325511410106e-05,
+      "loss": 0.5165,
+      "step": 254
+    },
+    {
+      "epoch": 2.294713160854893,
+      "grad_norm": 0.2360828175959895,
+      "learning_rate": 5.250429459513876e-05,
+      "loss": 0.5148,
+      "step": 255
+    },
+    {
+      "epoch": 2.303712035995501,
+      "grad_norm": 0.2188304429809361,
+      "learning_rate": 5.226483844751099e-05,
+      "loss": 0.5098,
+      "step": 256
+    },
+    {
+      "epoch": 2.312710911136108,
+      "grad_norm": 0.2334512539619716,
+      "learning_rate": 5.202489616246333e-05,
+      "loss": 0.5158,
+      "step": 257
+    },
+    {
+      "epoch": 2.3217097862767155,
+      "grad_norm": 0.19345732584763242,
+      "learning_rate": 5.178447725051026e-05,
+      "loss": 0.5137,
+      "step": 258
+    },
+    {
+      "epoch": 2.3307086614173227,
+      "grad_norm": 0.22726718301152563,
+      "learning_rate": 5.154359124105814e-05,
+      "loss": 0.5149,
+      "step": 259
+    },
+    {
+      "epoch": 2.3397075365579303,
+      "grad_norm": 0.2149304217989204,
+      "learning_rate": 5.1302247682027494e-05,
+      "loss": 0.5122,
+      "step": 260
+    },
+    {
+      "epoch": 2.3487064116985374,
+      "grad_norm": 0.2574574491521757,
+      "learning_rate": 5.106045613947466e-05,
+      "loss": 0.5106,
+      "step": 261
+    },
+    {
+      "epoch": 2.357705286839145,
+      "grad_norm": 0.2244044383603563,
+      "learning_rate": 5.08182261972125e-05,
+      "loss": 0.5098,
+      "step": 262
+    },
+    {
+      "epoch": 2.3667041619797526,
+      "grad_norm": 0.1722686877882969,
+      "learning_rate": 5.057556745643058e-05,
+      "loss": 0.5152,
+      "step": 263
+    },
+    {
+      "epoch": 2.37570303712036,
+      "grad_norm": 0.20786422486929515,
+      "learning_rate": 5.033248953531466e-05,
+      "loss": 0.5096,
+      "step": 264
+    },
+    {
+      "epoch": 2.3847019122609674,
+      "grad_norm": 0.20518803463398436,
+      "learning_rate": 5.0089002068665376e-05,
+      "loss": 0.5108,
+      "step": 265
+    },
+    {
+      "epoch": 2.393700787401575,
+      "grad_norm": 0.19783852278781122,
+      "learning_rate": 4.9845114707516395e-05,
+      "loss": 0.5169,
+      "step": 266
+    },
+    {
+      "epoch": 2.402699662542182,
+      "grad_norm": 0.15894601748576243,
+      "learning_rate": 4.9600837118751874e-05,
+      "loss": 0.5077,
+      "step": 267
+    },
+    {
+      "epoch": 2.4116985376827897,
+      "grad_norm": 0.19443345762052888,
+      "learning_rate": 4.93561789847233e-05,
+      "loss": 0.5211,
+      "step": 268
+    },
+    {
+      "epoch": 2.420697412823397,
+      "grad_norm": 0.19638270288569096,
+      "learning_rate": 4.91111500028657e-05,
+      "loss": 0.5098,
+      "step": 269
+    },
+    {
+      "epoch": 2.4296962879640045,
+      "grad_norm": 0.17816319122872262,
+      "learning_rate": 4.886575988531329e-05,
+      "loss": 0.5175,
+      "step": 270
+    },
+    {
+      "epoch": 2.438695163104612,
+      "grad_norm": 0.17651511328007938,
+      "learning_rate": 4.862001835851449e-05,
+      "loss": 0.5136,
+      "step": 271
+    },
+    {
+      "epoch": 2.4476940382452193,
+      "grad_norm": 0.1840295184610878,
+      "learning_rate": 4.837393516284642e-05,
+      "loss": 0.5099,
+      "step": 272
+    },
+    {
+      "epoch": 2.456692913385827,
+      "grad_norm": 0.1914170149726011,
+      "learning_rate": 4.8127520052228815e-05,
+      "loss": 0.5165,
+      "step": 273
+    },
+    {
+      "epoch": 2.465691788526434,
+      "grad_norm": 0.184680647588443,
+      "learning_rate": 4.788078279373744e-05,
+      "loss": 0.5084,
+      "step": 274
+    },
+    {
+      "epoch": 2.4746906636670416,
+      "grad_norm": 0.22287775595578857,
+      "learning_rate": 4.763373316721687e-05,
+      "loss": 0.5134,
+      "step": 275
+    },
+    {
+      "epoch": 2.483689538807649,
+      "grad_norm": 0.25180987790543646,
+      "learning_rate": 4.7386380964892984e-05,
+      "loss": 0.5137,
+      "step": 276
+    },
+    {
+      "epoch": 2.4926884139482564,
+      "grad_norm": 0.2645098272392606,
+      "learning_rate": 4.713873599098471e-05,
+      "loss": 0.5128,
+      "step": 277
+    },
+    {
+      "epoch": 2.501687289088864,
+      "grad_norm": 0.16671664940075334,
+      "learning_rate": 4.689080806131547e-05,
+      "loss": 0.5114,
+      "step": 278
+    },
+    {
+      "epoch": 2.5106861642294716,
+      "grad_norm": 0.22458637707427415,
+      "learning_rate": 4.664260700292416e-05,
+      "loss": 0.5095,
+      "step": 279
+    },
+    {
+      "epoch": 2.5196850393700787,
+      "grad_norm": 0.20330358237599677,
+      "learning_rate": 4.639414265367554e-05,
+      "loss": 0.509,
+      "step": 280
+    },
+    {
+      "epoch": 2.5286839145106863,
+      "grad_norm": 0.19076267160273108,
+      "learning_rate": 4.614542486187033e-05,
+      "loss": 0.5135,
+      "step": 281
+    },
+    {
+      "epoch": 2.5376827896512935,
+      "grad_norm": 0.19390672620858349,
+      "learning_rate": 4.589646348585494e-05,
+      "loss": 0.5129,
+      "step": 282
+    },
+    {
+      "epoch": 2.546681664791901,
+      "grad_norm": 0.17431590192861887,
+      "learning_rate": 4.564726839363059e-05,
+      "loss": 0.5119,
+      "step": 283
+    },
+    {
+      "epoch": 2.555680539932508,
+      "grad_norm": 0.19796573953993535,
+      "learning_rate": 4.539784946246225e-05,
+      "loss": 0.5226,
+      "step": 284
+    },
+    {
+      "epoch": 2.564679415073116,
+      "grad_norm": 0.21806700674738302,
+      "learning_rate": 4.5148216578487134e-05,
+      "loss": 0.5205,
+      "step": 285
+    },
+    {
+      "epoch": 2.5736782902137234,
+      "grad_norm": 0.21026084289689406,
+      "learning_rate": 4.4898379636322815e-05,
+      "loss": 0.5078,
+      "step": 286
+    },
+    {
+      "epoch": 2.5826771653543306,
+      "grad_norm": 0.18917285015389398,
+      "learning_rate": 4.4648348538675064e-05,
+      "loss": 0.5146,
+      "step": 287
+    },
+    {
+      "epoch": 2.591676040494938,
+      "grad_norm": 0.17876905148977648,
+      "learning_rate": 4.4398133195945326e-05,
+      "loss": 0.5124,
+      "step": 288
+    },
+    {
+      "epoch": 2.6006749156355458,
+      "grad_norm": 0.19991148697525993,
+      "learning_rate": 4.414774352583791e-05,
+      "loss": 0.5141,
+      "step": 289
+    },
+    {
+      "epoch": 2.609673790776153,
+      "grad_norm": 0.1718371307146163,
+      "learning_rate": 4.3897189452966895e-05,
+      "loss": 0.5126,
+      "step": 290
+    },
+    {
+      "epoch": 2.6186726659167605,
+      "grad_norm": 0.1685835159809121,
+      "learning_rate": 4.364648090846271e-05,
+      "loss": 0.5133,
+      "step": 291
+    },
+    {
+      "epoch": 2.6276715410573677,
+      "grad_norm": 0.14950354028807825,
+      "learning_rate": 4.339562782957857e-05,
+      "loss": 0.5083,
+      "step": 292
+    },
+    {
+      "epoch": 2.6366704161979753,
+      "grad_norm": 0.17804715853771555,
+      "learning_rate": 4.314464015929649e-05,
+      "loss": 0.5194,
+      "step": 293
+    },
+    {
+      "epoch": 2.6456692913385824,
+      "grad_norm": 0.17707266681302694,
+      "learning_rate": 4.28935278459333e-05,
+      "loss": 0.5128,
+      "step": 294
+    },
+    {
+      "epoch": 2.65466816647919,
+      "grad_norm": 0.15273475864497393,
+      "learning_rate": 4.264230084274624e-05,
+      "loss": 0.5148,
+      "step": 295
+    },
+    {
+      "epoch": 2.6636670416197976,
+      "grad_norm": 0.19292123693836521,
+      "learning_rate": 4.239096910753846e-05,
+      "loss": 0.5097,
+      "step": 296
+    },
+    {
+      "epoch": 2.6726659167604048,
+      "grad_norm": 0.19491285573674288,
+      "learning_rate": 4.213954260226438e-05,
+      "loss": 0.5108,
+      "step": 297
+    },
+    {
+      "epoch": 2.6816647919010124,
+      "grad_norm": 0.16377738874278228,
+      "learning_rate": 4.188803129263476e-05,
+      "loss": 0.5084,
+      "step": 298
+    },
+    {
+      "epoch": 2.69066366704162,
+      "grad_norm": 0.17707808785911283,
+      "learning_rate": 4.163644514772172e-05,
+      "loss": 0.5078,
+      "step": 299
+    },
+    {
+      "epoch": 2.699662542182227,
+      "grad_norm": 0.16129691634220186,
+      "learning_rate": 4.1384794139563614e-05,
+      "loss": 0.5177,
+      "step": 300
+    },
+    {
+      "epoch": 2.7086614173228347,
+      "grad_norm": 0.19221250489172526,
+      "learning_rate": 4.113308824276977e-05,
+      "loss": 0.5149,
+      "step": 301
+    },
+    {
+      "epoch": 2.7176602924634423,
+      "grad_norm": 0.17286893566543513,
+      "learning_rate": 4.0881337434125086e-05,
+      "loss": 0.5088,
+      "step": 302
+    },
+    {
+      "epoch": 2.7266591676040495,
+      "grad_norm": 0.1755459663317774,
+      "learning_rate": 4.0629551692194634e-05,
+      "loss": 0.509,
+      "step": 303
+    },
+    {
+      "epoch": 2.735658042744657,
+      "grad_norm": 0.1791201340425473,
+      "learning_rate": 4.037774099692815e-05,
+      "loss": 0.5143,
+      "step": 304
+    },
+    {
+      "epoch": 2.7446569178852642,
+      "grad_norm": 0.18143794043338418,
+      "learning_rate": 4.0125915329264396e-05,
+      "loss": 0.5123,
+      "step": 305
+    },
+    {
+      "epoch": 2.753655793025872,
+      "grad_norm": 0.15111984385935012,
+      "learning_rate": 3.9874084670735624e-05,
+      "loss": 0.5132,
+      "step": 306
+    },
+    {
+      "epoch": 2.762654668166479,
+      "grad_norm": 0.17406268964374014,
+      "learning_rate": 3.962225900307187e-05,
+      "loss": 0.5088,
+      "step": 307
+    },
+    {
+      "epoch": 2.7716535433070866,
+      "grad_norm": 0.1845218510232061,
+      "learning_rate": 3.937044830780537e-05,
+      "loss": 0.5115,
+      "step": 308
+    },
+    {
+      "epoch": 2.780652418447694,
+      "grad_norm": 0.14408452277861603,
+      "learning_rate": 3.9118662565874934e-05,
+      "loss": 0.5088,
+      "step": 309
+    },
+    {
+      "epoch": 2.7896512935883013,
+      "grad_norm": 0.2121818690921875,
+      "learning_rate": 3.886691175723025e-05,
+      "loss": 0.515,
+      "step": 310
+    },
+    {
+      "epoch": 2.798650168728909,
+      "grad_norm": 0.17415478680558893,
+      "learning_rate": 3.8615205860436406e-05,
+      "loss": 0.5102,
+      "step": 311
+    },
+    {
+      "epoch": 2.8076490438695165,
+      "grad_norm": 0.17710673827275958,
+      "learning_rate": 3.83635548522783e-05,
+      "loss": 0.5142,
+      "step": 312
+    },
+    {
+      "epoch": 2.8166479190101237,
+      "grad_norm": 0.2131327323874231,
+      "learning_rate": 3.811196870736526e-05,
+      "loss": 0.5143,
+      "step": 313
+    },
+    {
+      "epoch": 2.8256467941507313,
+      "grad_norm": 0.17135728004223627,
+      "learning_rate": 3.786045739773564e-05,
+      "loss": 0.5066,
+      "step": 314
+    },
+    {
+      "epoch": 2.8346456692913384,
+      "grad_norm": 0.1787641234702102,
+      "learning_rate": 3.7609030892461554e-05,
+      "loss": 0.5077,
+      "step": 315
+    },
+    {
+      "epoch": 2.843644544431946,
+      "grad_norm": 0.1737180877569213,
+      "learning_rate": 3.735769915725378e-05,
+      "loss": 0.5104,
+      "step": 316
+    },
+    {
+      "epoch": 2.852643419572553,
+      "grad_norm": 0.1759592145643783,
+      "learning_rate": 3.710647215406672e-05,
+      "loss": 0.5076,
+      "step": 317
+    },
+    {
+      "epoch": 2.861642294713161,
+      "grad_norm": 0.1674590421194775,
+      "learning_rate": 3.6855359840703525e-05,
+      "loss": 0.5117,
+      "step": 318
+    },
+    {
+      "epoch": 2.8706411698537684,
+      "grad_norm": 0.19195727087026487,
+      "learning_rate": 3.660437217042145e-05,
+      "loss": 0.515,
+      "step": 319
+    },
+    {
+      "epoch": 2.8796400449943755,
+      "grad_norm": 0.2079526174400504,
+      "learning_rate": 3.63535190915373e-05,
+      "loss": 0.5114,
+      "step": 320
+    },
+    {
+      "epoch": 2.888638920134983,
+      "grad_norm": 0.16261938739039505,
+      "learning_rate": 3.610281054703311e-05,
+      "loss": 0.512,
+      "step": 321
+    },
+    {
+      "epoch": 2.8976377952755907,
+      "grad_norm": 0.17301946358427484,
+      "learning_rate": 3.58522564741621e-05,
+      "loss": 0.513,
+      "step": 322
+    },
+    {
+      "epoch": 2.906636670416198,
+      "grad_norm": 0.15809619671547212,
+      "learning_rate": 3.560186680405469e-05,
+      "loss": 0.5092,
+      "step": 323
+    },
+    {
+      "epoch": 2.9156355455568055,
+      "grad_norm": 0.18005047534402024,
+      "learning_rate": 3.535165146132494e-05,
+      "loss": 0.5143,
+      "step": 324
+    },
+    {
+      "epoch": 2.924634420697413,
+      "grad_norm": 0.17055434106531733,
+      "learning_rate": 3.51016203636772e-05,
+      "loss": 0.5044,
+      "step": 325
+    },
+    {
+      "epoch": 2.9336332958380202,
+      "grad_norm": 0.2091344061636353,
+      "learning_rate": 3.485178342151287e-05,
+      "loss": 0.5151,
+      "step": 326
+    },
+    {
+      "epoch": 2.942632170978628,
+      "grad_norm": 0.17776622234467257,
+      "learning_rate": 3.460215053753776e-05,
+      "loss": 0.506,
+      "step": 327
+    },
+    {
+      "epoch": 2.951631046119235,
+      "grad_norm": 0.2020120609501547,
+      "learning_rate": 3.435273160636942e-05,
+      "loss": 0.5106,
+      "step": 328
+    },
+    {
+      "epoch": 2.9606299212598426,
+      "grad_norm": 0.17580656405878176,
+      "learning_rate": 3.410353651414507e-05,
+      "loss": 0.509,
+      "step": 329
+    },
+    {
+      "epoch": 2.9696287964004497,
+      "grad_norm": 0.1874074410148767,
+      "learning_rate": 3.385457513812968e-05,
+      "loss": 0.5119,
+      "step": 330
+    },
+    {
+      "epoch": 2.9786276715410573,
+      "grad_norm": 0.18596738197007068,
+      "learning_rate": 3.360585734632448e-05,
+      "loss": 0.5105,
+      "step": 331
+    },
+    {
+      "epoch": 2.987626546681665,
+      "grad_norm": 0.18012608783890774,
+      "learning_rate": 3.3357392997075854e-05,
+      "loss": 0.51,
+      "step": 332
+    },
+    {
+      "epoch": 2.996625421822272,
+      "grad_norm": 0.17695247800719235,
+      "learning_rate": 3.3109191938684535e-05,
+      "loss": 0.5096,
+      "step": 333
+    },
+    {
+      "epoch": 3.0056242969628797,
+      "grad_norm": 0.4041448073125774,
+      "learning_rate": 3.28612640090153e-05,
+      "loss": 0.9217,
+      "step": 334
+    },
+    {
+      "epoch": 3.014623172103487,
+      "grad_norm": 0.40335779907517166,
+      "learning_rate": 3.261361903510703e-05,
+      "loss": 0.4749,
+      "step": 335
+    },
+    {
+      "epoch": 3.0236220472440944,
+      "grad_norm": 0.23571838314313354,
+      "learning_rate": 3.2366266832783145e-05,
+      "loss": 0.4795,
+      "step": 336
+    },
+    {
+      "epoch": 3.032620922384702,
+      "grad_norm": 0.34962469353680325,
+      "learning_rate": 3.211921720626258e-05,
+      "loss": 0.4772,
+      "step": 337
+    },
+    {
+      "epoch": 3.041619797525309,
+      "grad_norm": 0.3250568724209024,
+      "learning_rate": 3.187247994777119e-05,
+      "loss": 0.4756,
+      "step": 338
+    },
+    {
+      "epoch": 3.050618672665917,
+      "grad_norm": 0.27683092123819425,
+      "learning_rate": 3.1626064837153596e-05,
+      "loss": 0.478,
+      "step": 339
+    },
+    {
+      "epoch": 3.0596175478065244,
+      "grad_norm": 0.29128324637433173,
+      "learning_rate": 3.1379981641485524e-05,
+      "loss": 0.4767,
+      "step": 340
+    },
+    {
+      "epoch": 3.0686164229471316,
+      "grad_norm": 0.2783998825962506,
+      "learning_rate": 3.113424011468672e-05,
+      "loss": 0.4758,
+      "step": 341
+    },
+    {
+      "epoch": 3.077615298087739,
+      "grad_norm": 0.25380987160914664,
+      "learning_rate": 3.0888849997134316e-05,
+      "loss": 0.4663,
+      "step": 342
+    },
+    {
+      "epoch": 3.0866141732283463,
+      "grad_norm": 0.23893069831482153,
+      "learning_rate": 3.064382101527671e-05,
+      "loss": 0.4788,
+      "step": 343
+    },
+    {
+      "epoch": 3.095613048368954,
+      "grad_norm": 0.2546443581638444,
+      "learning_rate": 3.039916288124814e-05,
+      "loss": 0.4794,
+      "step": 344
+    },
+    {
+      "epoch": 3.1046119235095615,
+      "grad_norm": 0.2193178093849395,
+      "learning_rate": 3.015488529248362e-05,
+      "loss": 0.479,
+      "step": 345
+    },
+    {
+      "epoch": 3.1136107986501687,
+      "grad_norm": 0.23966451017854798,
+      "learning_rate": 2.9910997931334637e-05,
+      "loss": 0.4754,
+      "step": 346
+    },
+    {
+      "epoch": 3.1226096737907763,
+      "grad_norm": 0.2258415034497983,
+      "learning_rate": 2.9667510464685345e-05,
+      "loss": 0.4825,
+      "step": 347
+    },
+    {
+      "epoch": 3.1316085489313834,
+      "grad_norm": 0.23731192604698906,
+      "learning_rate": 2.9424432543569428e-05,
+      "loss": 0.4723,
+      "step": 348
+    },
+    {
+      "epoch": 3.140607424071991,
+      "grad_norm": 0.17268744214169082,
+      "learning_rate": 2.918177380278752e-05,
+      "loss": 0.4763,
+      "step": 349
+    },
+    {
+      "epoch": 3.1496062992125986,
+      "grad_norm": 0.23327367459895917,
+      "learning_rate": 2.893954386052535e-05,
+      "loss": 0.4786,
+      "step": 350
+    },
+    {
+      "epoch": 3.1586051743532058,
+      "grad_norm": 0.1956862643734178,
+      "learning_rate": 2.8697752317972513e-05,
+      "loss": 0.4765,
+      "step": 351
+    },
+    {
+      "epoch": 3.1676040494938134,
+      "grad_norm": 0.18627973371325526,
+      "learning_rate": 2.845640875894188e-05,
+      "loss": 0.4745,
+      "step": 352
+    },
+    {
+      "epoch": 3.1766029246344205,
+      "grad_norm": 0.2105328504046573,
+      "learning_rate": 2.8215522749489742e-05,
+      "loss": 0.4804,
+      "step": 353
+    },
+    {
+      "epoch": 3.185601799775028,
+      "grad_norm": 0.16272715621667633,
+      "learning_rate": 2.7975103837536672e-05,
+      "loss": 0.472,
+      "step": 354
+    },
+    {
+      "epoch": 3.1946006749156357,
+      "grad_norm": 0.21097547189014748,
+      "learning_rate": 2.7735161552489022e-05,
+      "loss": 0.4749,
+      "step": 355
+    },
+    {
+      "epoch": 3.203599550056243,
+      "grad_norm": 0.17765983026821935,
+      "learning_rate": 2.749570540486125e-05,
+      "loss": 0.4774,
+      "step": 356
+    },
+    {
+      "epoch": 3.2125984251968505,
+      "grad_norm": 0.18185508796151498,
+      "learning_rate": 2.7256744885898942e-05,
+      "loss": 0.4769,
+      "step": 357
+    },
+    {
+      "epoch": 3.2215973003374576,
+      "grad_norm": 0.17545115382309487,
+      "learning_rate": 2.7018289467202623e-05,
+      "loss": 0.4707,
+      "step": 358
+    },
+    {
+      "epoch": 3.230596175478065,
+      "grad_norm": 0.16118530505673961,
+      "learning_rate": 2.6780348600352284e-05,
+      "loss": 0.4751,
+      "step": 359
+    },
+    {
+      "epoch": 3.239595050618673,
+      "grad_norm": 0.17928864347147178,
+      "learning_rate": 2.6542931716532856e-05,
+      "loss": 0.4807,
+      "step": 360
+    },
+    {
+      "epoch": 3.24859392575928,
+      "grad_norm": 0.16478061360464155,
+      "learning_rate": 2.630604822616027e-05,
+      "loss": 0.4736,
+      "step": 361
+    },
+    {
+      "epoch": 3.2575928008998876,
+      "grad_norm": 0.1478559482793346,
+      "learning_rate": 2.6069707518508523e-05,
+      "loss": 0.4781,
+      "step": 362
+    },
+    {
+      "epoch": 3.2665916760404947,
+      "grad_norm": 0.16844157556458098,
+      "learning_rate": 2.583391896133753e-05,
+      "loss": 0.4765,
+      "step": 363
+    },
+    {
+      "epoch": 3.2755905511811023,
+      "grad_norm": 0.1337175278664685,
+      "learning_rate": 2.5598691900521778e-05,
+      "loss": 0.4741,
+      "step": 364
+    },
+    {
+      "epoch": 3.28458942632171,
+      "grad_norm": 0.13894808378571294,
+      "learning_rate": 2.5364035659679914e-05,
+      "loss": 0.4762,
+      "step": 365
+    },
+    {
+      "epoch": 3.293588301462317,
+      "grad_norm": 0.14038211286961325,
+      "learning_rate": 2.512995953980518e-05,
+      "loss": 0.4785,
+      "step": 366
+    },
+    {
+      "epoch": 3.3025871766029247,
+      "grad_norm": 0.13577552755343492,
+      "learning_rate": 2.4896472818896743e-05,
+      "loss": 0.4798,
+      "step": 367
+    },
+    {
+      "epoch": 3.3115860517435323,
+      "grad_norm": 0.14414082887532978,
+      "learning_rate": 2.4663584751591977e-05,
+      "loss": 0.4784,
+      "step": 368
+    },
+    {
+      "epoch": 3.3205849268841394,
+      "grad_norm": 0.1499138881871864,
+      "learning_rate": 2.443130456879958e-05,
+      "loss": 0.475,
+      "step": 369
+    },
+    {
+      "epoch": 3.329583802024747,
+      "grad_norm": 0.15378837222105818,
+      "learning_rate": 2.4199641477333766e-05,
+      "loss": 0.4695,
+      "step": 370
+    },
+    {
+      "epoch": 3.338582677165354,
+      "grad_norm": 0.15500949142269607,
+      "learning_rate": 2.3968604659549266e-05,
+      "loss": 0.4808,
+      "step": 371
+    },
+    {
+      "epoch": 3.3475815523059618,
+      "grad_norm": 0.13510134534596135,
+      "learning_rate": 2.3738203272977446e-05,
+      "loss": 0.4762,
+      "step": 372
+    },
+    {
+      "epoch": 3.3565804274465694,
+      "grad_norm": 0.1580364610829008,
+      "learning_rate": 2.350844644996325e-05,
+      "loss": 0.4756,
+      "step": 373
+    },
+    {
+      "epoch": 3.3655793025871765,
+      "grad_norm": 0.1447596119360514,
+      "learning_rate": 2.3279343297303293e-05,
+      "loss": 0.476,
+      "step": 374
+    },
+    {
+      "epoch": 3.374578177727784,
+      "grad_norm": 0.13783465555399632,
+      "learning_rate": 2.305090289588485e-05,
+      "loss": 0.478,
+      "step": 375
+    },
+    {
+      "epoch": 3.3835770528683913,
+      "grad_norm": 0.1422413809528974,
+      "learning_rate": 2.2823134300325948e-05,
+      "loss": 0.4717,
+      "step": 376
+    },
+    {
+      "epoch": 3.392575928008999,
+      "grad_norm": 0.12982390766734286,
+      "learning_rate": 2.25960465386165e-05,
+      "loss": 0.4787,
+      "step": 377
+    },
+    {
+      "epoch": 3.4015748031496065,
+      "grad_norm": 0.1445554728575935,
+      "learning_rate": 2.2369648611760352e-05,
+      "loss": 0.4728,
+      "step": 378
+    },
+    {
+      "epoch": 3.4105736782902136,
+      "grad_norm": 0.14091047744757867,
+      "learning_rate": 2.2143949493418654e-05,
+      "loss": 0.4778,
+      "step": 379
+    },
+    {
+      "epoch": 3.4195725534308212,
+      "grad_norm": 0.15001963385136857,
+      "learning_rate": 2.1918958129554106e-05,
+      "loss": 0.4764,
+      "step": 380
+    },
+    {
+      "epoch": 3.4285714285714284,
+      "grad_norm": 0.13768741985271635,
+      "learning_rate": 2.1694683438076317e-05,
+      "loss": 0.471,
+      "step": 381
+    },
+    {
+      "epoch": 3.437570303712036,
+      "grad_norm": 0.14948373435356904,
+      "learning_rate": 2.147113430848844e-05,
+      "loss": 0.4697,
+      "step": 382
+    },
+    {
+      "epoch": 3.4465691788526436,
+      "grad_norm": 0.13361989178999917,
+      "learning_rate": 2.1248319601534772e-05,
+      "loss": 0.4707,
+      "step": 383
+    },
+    {
+      "epoch": 3.4555680539932507,
+      "grad_norm": 0.13151795053856452,
+      "learning_rate": 2.102624814884949e-05,
+      "loss": 0.4747,
+      "step": 384
+    },
+    {
+      "epoch": 3.4645669291338583,
+      "grad_norm": 0.14805117724302444,
+      "learning_rate": 2.080492875260668e-05,
+      "loss": 0.4765,
+      "step": 385
+    },
+    {
+      "epoch": 3.4735658042744655,
+      "grad_norm": 0.15118784969552582,
+      "learning_rate": 2.0584370185171418e-05,
+      "loss": 0.4727,
+      "step": 386
+    },
+    {
+      "epoch": 3.482564679415073,
+      "grad_norm": 0.14047665231367965,
+      "learning_rate": 2.0364581188752012e-05,
+      "loss": 0.4819,
+      "step": 387
+    },
+    {
+      "epoch": 3.4915635545556807,
+      "grad_norm": 0.15664210880968063,
+      "learning_rate": 2.014557047505357e-05,
+      "loss": 0.466,
+      "step": 388
+    },
+    {
+      "epoch": 3.500562429696288,
+      "grad_norm": 0.13241547005518312,
+      "learning_rate": 1.992734672493267e-05,
+      "loss": 0.4772,
+      "step": 389
+    },
+    {
+      "epoch": 3.5095613048368954,
+      "grad_norm": 0.15495785626518518,
+      "learning_rate": 1.970991858805322e-05,
+      "loss": 0.4768,
+      "step": 390
+    },
+    {
+      "epoch": 3.518560179977503,
+      "grad_norm": 0.1451652460628155,
+      "learning_rate": 1.9493294682543715e-05,
+      "loss": 0.4752,
+      "step": 391
+    },
+    {
+      "epoch": 3.52755905511811,
+      "grad_norm": 0.15292105555983718,
+      "learning_rate": 1.927748359465558e-05,
+      "loss": 0.4768,
+      "step": 392
+    },
+    {
+      "epoch": 3.536557930258718,
+      "grad_norm": 0.1457418788935837,
+      "learning_rate": 1.9062493878422823e-05,
+      "loss": 0.4768,
+      "step": 393
+    },
+    {
+      "epoch": 3.545556805399325,
+      "grad_norm": 0.13975287470844083,
+      "learning_rate": 1.884833405532304e-05,
+      "loss": 0.4757,
+      "step": 394
+    },
+    {
+      "epoch": 3.5545556805399325,
+      "grad_norm": 0.14640008673656155,
+      "learning_rate": 1.863501261393963e-05,
+      "loss": 0.4813,
+      "step": 395
+    },
+    {
+      "epoch": 3.5635545556805397,
+      "grad_norm": 0.12836792930570248,
+      "learning_rate": 1.8422538009625285e-05,
+      "loss": 0.4738,
+      "step": 396
+    },
+    {
+      "epoch": 3.5725534308211473,
+      "grad_norm": 0.13526217288902667,
+      "learning_rate": 1.8210918664166945e-05,
+      "loss": 0.4762,
+      "step": 397
+    },
+    {
+      "epoch": 3.581552305961755,
+      "grad_norm": 0.1420752062696319,
+      "learning_rate": 1.8000162965451884e-05,
+      "loss": 0.4803,
+      "step": 398
+    },
+    {
+      "epoch": 3.590551181102362,
+      "grad_norm": 0.1279945537635078,
+      "learning_rate": 1.7790279267135317e-05,
+      "loss": 0.4829,
+      "step": 399
+    },
+    {
+      "epoch": 3.5995500562429696,
+      "grad_norm": 0.13720312932073742,
+      "learning_rate": 1.758127588830928e-05,
+      "loss": 0.4756,
+      "step": 400
+    },
+    {
+      "epoch": 3.6085489313835772,
+      "grad_norm": 0.11571164848869406,
+      "learning_rate": 1.737316111317281e-05,
+      "loss": 0.4788,
+      "step": 401
+    },
+    {
+      "epoch": 3.6175478065241844,
+      "grad_norm": 0.1267184023160666,
+      "learning_rate": 1.716594319070371e-05,
+      "loss": 0.4672,
+      "step": 402
+    },
+    {
+      "epoch": 3.626546681664792,
+      "grad_norm": 0.12093756371303144,
+      "learning_rate": 1.695963033433151e-05,
+      "loss": 0.4767,
+      "step": 403
+    },
+    {
+      "epoch": 3.6355455568053996,
+      "grad_norm": 0.1281599915828755,
+      "learning_rate": 1.6754230721611896e-05,
+      "loss": 0.4762,
+      "step": 404
+    },
+    {
+      "epoch": 3.6445444319460067,
+      "grad_norm": 0.12299228584682197,
+      "learning_rate": 1.654975249390265e-05,
+      "loss": 0.477,
+      "step": 405
+    },
+    {
+      "epoch": 3.653543307086614,
+      "grad_norm": 0.13855705326136183,
+      "learning_rate": 1.634620375604091e-05,
+      "loss": 0.4791,
+      "step": 406
+    },
+    {
+      "epoch": 3.6625421822272215,
+      "grad_norm": 0.12445550210040791,
+      "learning_rate": 1.6143592576021897e-05,
+      "loss": 0.4746,
+      "step": 407
+    },
+    {
+      "epoch": 3.671541057367829,
+      "grad_norm": 0.1328319119905009,
+      "learning_rate": 1.594192698467919e-05,
+      "loss": 0.4753,
+      "step": 408
+    },
+    {
+      "epoch": 3.6805399325084363,
+      "grad_norm": 0.12084524067527391,
+      "learning_rate": 1.574121497536638e-05,
+      "loss": 0.4778,
+      "step": 409
+    },
+    {
+      "epoch": 3.689538807649044,
+      "grad_norm": 0.12320850922380423,
+      "learning_rate": 1.5541464503640195e-05,
+      "loss": 0.4718,
+      "step": 410
+    },
+    {
+      "epoch": 3.6985376827896514,
+      "grad_norm": 0.11554496711321419,
+      "learning_rate": 1.534268348694524e-05,
+      "loss": 0.4734,
+      "step": 411
+    },
+    {
+      "epoch": 3.7075365579302586,
+      "grad_norm": 0.11815716191127555,
+      "learning_rate": 1.5144879804300163e-05,
+      "loss": 0.4701,
+      "step": 412
+    },
+    {
+      "epoch": 3.716535433070866,
+      "grad_norm": 0.12388808731535488,
+      "learning_rate": 1.4948061295985286e-05,
+      "loss": 0.4786,
+      "step": 413
+    },
+    {
+      "epoch": 3.725534308211474,
+      "grad_norm": 0.11077052559901988,
+      "learning_rate": 1.4752235763231944e-05,
+      "loss": 0.4716,
+      "step": 414
+    },
+    {
+      "epoch": 3.734533183352081,
+      "grad_norm": 0.11829062763031385,
+      "learning_rate": 1.4557410967913219e-05,
+      "loss": 0.4819,
+      "step": 415
+    },
+    {
+      "epoch": 3.7435320584926886,
+      "grad_norm": 0.12116279675489595,
+      "learning_rate": 1.4363594632236249e-05,
+      "loss": 0.4746,
+      "step": 416
+    },
+    {
+      "epoch": 3.7525309336332957,
+      "grad_norm": 0.11924444626526026,
+      "learning_rate": 1.4170794438436236e-05,
+      "loss": 0.4826,
+      "step": 417
+    },
+    {
+      "epoch": 3.7615298087739033,
+      "grad_norm": 0.1117894145977932,
+      "learning_rate": 1.3979018028471858e-05,
+      "loss": 0.4719,
+      "step": 418
+    },
+    {
+      "epoch": 3.7705286839145105,
+      "grad_norm": 0.11651706087134892,
+      "learning_rate": 1.3788273003722404e-05,
+      "loss": 0.4789,
+      "step": 419
+    },
+    {
+      "epoch": 3.779527559055118,
+      "grad_norm": 0.11055442276894334,
+      "learning_rate": 1.3598566924686511e-05,
+      "loss": 0.4796,
+      "step": 420
+    },
+    {
+      "epoch": 3.7885264341957257,
+      "grad_norm": 0.10345352334778583,
+      "learning_rate": 1.3409907310682462e-05,
+      "loss": 0.4709,
+      "step": 421
+    },
+    {
+      "epoch": 3.797525309336333,
+      "grad_norm": 0.10232180540617608,
+      "learning_rate": 1.3222301639550099e-05,
+      "loss": 0.4766,
+      "step": 422
+    },
+    {
+      "epoch": 3.8065241844769404,
+      "grad_norm": 0.10715279829216998,
+      "learning_rate": 1.3035757347354526e-05,
+      "loss": 0.4776,
+      "step": 423
+    },
+    {
+      "epoch": 3.815523059617548,
+      "grad_norm": 0.1047852036739672,
+      "learning_rate": 1.2850281828091298e-05,
+      "loss": 0.4854,
+      "step": 424
+    },
+    {
+      "epoch": 3.824521934758155,
+      "grad_norm": 0.10690185941320834,
+      "learning_rate": 1.2665882433393338e-05,
+      "loss": 0.4729,
+      "step": 425
+    },
+    {
+      "epoch": 3.8335208098987628,
+      "grad_norm": 0.11484140024382984,
+      "learning_rate": 1.24825664722396e-05,
+      "loss": 0.475,
+      "step": 426
+    },
+    {
+      "epoch": 3.84251968503937,
+      "grad_norm": 0.11568039124788146,
+      "learning_rate": 1.2300341210665336e-05,
+      "loss": 0.4758,
+      "step": 427
+    },
+    {
+      "epoch": 3.8515185601799775,
+      "grad_norm": 0.10120504659325706,
+      "learning_rate": 1.211921387147406e-05,
+      "loss": 0.4732,
+      "step": 428
+    },
+    {
+      "epoch": 3.8605174353205847,
+      "grad_norm": 0.1194598344922678,
+      "learning_rate": 1.1939191633951328e-05,
+      "loss": 0.4768,
+      "step": 429
+    },
+    {
+      "epoch": 3.8695163104611923,
+      "grad_norm": 0.1090408000804906,
+      "learning_rate": 1.1760281633580136e-05,
+      "loss": 0.4741,
+      "step": 430
+    },
+    {
+      "epoch": 3.8785151856018,
+      "grad_norm": 0.10733618962308203,
+      "learning_rate": 1.1582490961758057e-05,
+      "loss": 0.479,
+      "step": 431
+    },
+    {
+      "epoch": 3.887514060742407,
+      "grad_norm": 0.11031251576967621,
+      "learning_rate": 1.1405826665516253e-05,
+      "loss": 0.4768,
+      "step": 432
+    },
+    {
+      "epoch": 3.8965129358830146,
+      "grad_norm": 0.11348215054291673,
+      "learning_rate": 1.1230295747240092e-05,
+      "loss": 0.4714,
+      "step": 433
+    },
+    {
+      "epoch": 3.905511811023622,
+      "grad_norm": 0.10993620934878778,
+      "learning_rate": 1.1055905164391567e-05,
+      "loss": 0.4785,
+      "step": 434
+    },
+    {
+      "epoch": 3.9145106861642294,
+      "grad_norm": 0.10346360621624287,
+      "learning_rate": 1.0882661829233619e-05,
+      "loss": 0.4789,
+      "step": 435
+    },
+    {
+      "epoch": 3.923509561304837,
+      "grad_norm": 0.10965908377773818,
+      "learning_rate": 1.071057260855608e-05,
+      "loss": 0.4759,
+      "step": 436
+    },
+    {
+      "epoch": 3.9325084364454446,
+      "grad_norm": 0.10838534658950197,
+      "learning_rate": 1.0539644323403514e-05,
+      "loss": 0.4732,
+      "step": 437
+    },
+    {
+      "epoch": 3.9415073115860517,
+      "grad_norm": 0.10164569710162663,
+      "learning_rate": 1.0369883748804868e-05,
+      "loss": 0.4777,
+      "step": 438
+    },
+    {
+      "epoch": 3.9505061867266593,
+      "grad_norm": 0.11037796996055305,
+      "learning_rate": 1.0201297613504946e-05,
+      "loss": 0.4765,
+      "step": 439
+    },
+    {
+      "epoch": 3.9595050618672665,
+      "grad_norm": 0.10427806979539742,
+      "learning_rate": 1.0033892599697638e-05,
+      "loss": 0.4724,
+      "step": 440
+    },
+    {
+      "epoch": 3.968503937007874,
+      "grad_norm": 0.10553954203749775,
+      "learning_rate": 9.86767534276114e-06,
+      "loss": 0.4827,
+      "step": 441
+    },
+    {
+      "epoch": 3.9775028121484812,
+      "grad_norm": 0.10616804220182864,
+      "learning_rate": 9.702652430994917e-06,
+      "loss": 0.4752,
+      "step": 442
+    },
+    {
+      "epoch": 3.986501687289089,
+      "grad_norm": 0.10240539280603339,
+      "learning_rate": 9.538830405358523e-06,
+      "loss": 0.4754,
+      "step": 443
+    },
+    {
+      "epoch": 3.9955005624296964,
+      "grad_norm": 0.10491697645612037,
+      "learning_rate": 9.376215759212423e-06,
+      "loss": 0.477,
+      "step": 444
+    },
+    {
+      "epoch": 4.008998875140607,
+      "grad_norm": 0.17334062662555766,
+      "learning_rate": 9.214814938060561e-06,
+      "loss": 0.4526,
+      "step": 445
+    },
+    {
+      "epoch": 4.017997750281215,
+      "grad_norm": 0.13435271730849074,
+      "learning_rate": 9.054634339294867e-06,
+      "loss": 0.4497,
+      "step": 446
+    },
+    {
+      "epoch": 4.026996625421822,
+      "grad_norm": 0.11872819062021366,
+      "learning_rate": 8.895680311941745e-06,
+      "loss": 0.4455,
+      "step": 447
+    },
+    {
+      "epoch": 4.0359955005624295,
+      "grad_norm": 0.1319008455279487,
+      "learning_rate": 8.737959156410385e-06,
+      "loss": 0.4527,
+      "step": 448
+    },
+    {
+      "epoch": 4.0449943757030375,
+      "grad_norm": 0.14206834817022773,
+      "learning_rate": 8.581477124243002e-06,
+      "loss": 0.451,
+      "step": 449
+    },
+    {
+      "epoch": 4.053993250843645,
+      "grad_norm": 0.14857898517649198,
+      "learning_rate": 8.426240417867121e-06,
+      "loss": 0.4511,
+      "step": 450
+    },
+    {
+      "epoch": 4.062992125984252,
+      "grad_norm": 0.12827796758216584,
+      "learning_rate": 8.272255190349678e-06,
+      "loss": 0.4517,
+      "step": 451
+    },
+    {
+      "epoch": 4.071991001124859,
+      "grad_norm": 0.12515412380900934,
+      "learning_rate": 8.119527545153137e-06,
+      "loss": 0.4514,
+      "step": 452
+    },
+    {
+      "epoch": 4.080989876265467,
+      "grad_norm": 0.13907555587381168,
+      "learning_rate": 7.968063535893588e-06,
+      "loss": 0.4465,
+      "step": 453
+    },
+    {
+      "epoch": 4.089988751406074,
+      "grad_norm": 0.1388856297974169,
+      "learning_rate": 7.817869166100812e-06,
+      "loss": 0.4485,
+      "step": 454
+    },
+    {
+      "epoch": 4.098987626546681,
+      "grad_norm": 0.12576078763204848,
+      "learning_rate": 7.668950388980261e-06,
+      "loss": 0.4506,
+      "step": 455
+    },
+    {
+      "epoch": 4.107986501687289,
+      "grad_norm": 0.12043647761005791,
+      "learning_rate": 7.521313107177182e-06,
+      "loss": 0.4507,
+      "step": 456
+    },
+    {
+      "epoch": 4.116985376827897,
+      "grad_norm": 0.12541943358764632,
+      "learning_rate": 7.374963172542564e-06,
+      "loss": 0.4524,
+      "step": 457
+    },
+    {
+      "epoch": 4.125984251968504,
+      "grad_norm": 0.12465374418622191,
+      "learning_rate": 7.229906385901264e-06,
+      "loss": 0.4463,
+      "step": 458
+    },
+    {
+      "epoch": 4.134983127109112,
+      "grad_norm": 0.1172582061392934,
+      "learning_rate": 7.086148496822054e-06,
+      "loss": 0.4541,
+      "step": 459
+    },
+    {
+      "epoch": 4.143982002249719,
+      "grad_norm": 0.11434456228260406,
+      "learning_rate": 6.943695203389689e-06,
+      "loss": 0.4522,
+      "step": 460
+    },
+    {
+      "epoch": 4.152980877390326,
+      "grad_norm": 0.126721520706509,
+      "learning_rate": 6.802552151979132e-06,
+      "loss": 0.4496,
+      "step": 461
+    },
+    {
+      "epoch": 4.161979752530933,
+      "grad_norm": 0.11357636546475693,
+      "learning_rate": 6.662724937031697e-06,
+      "loss": 0.4481,
+      "step": 462
+    },
+    {
+      "epoch": 4.170978627671541,
+      "grad_norm": 0.10812363979662522,
+      "learning_rate": 6.524219100833291e-06,
+      "loss": 0.45,
+      "step": 463
+    },
+    {
+      "epoch": 4.179977502812148,
+      "grad_norm": 0.10736860028431898,
+      "learning_rate": 6.387040133294786e-06,
+      "loss": 0.4556,
+      "step": 464
+    },
+    {
+      "epoch": 4.188976377952756,
+      "grad_norm": 0.11484073744710532,
+      "learning_rate": 6.2511934717343955e-06,
+      "loss": 0.4585,
+      "step": 465
+    },
+    {
+      "epoch": 4.197975253093364,
+      "grad_norm": 0.1116733301395841,
+      "learning_rate": 6.116684500662127e-06,
+      "loss": 0.4509,
+      "step": 466
+    },
+    {
+      "epoch": 4.206974128233971,
+      "grad_norm": 0.10184483968685969,
+      "learning_rate": 5.983518551566403e-06,
+      "loss": 0.4465,
+      "step": 467
+    },
+    {
+      "epoch": 4.215973003374578,
+      "grad_norm": 0.10648831010048188,
+      "learning_rate": 5.8517009027027285e-06,
+      "loss": 0.4529,
+      "step": 468
+    },
+    {
+      "epoch": 4.224971878515186,
+      "grad_norm": 0.11078629147369119,
+      "learning_rate": 5.72123677888444e-06,
+      "loss": 0.4504,
+      "step": 469
+    },
+    {
+      "epoch": 4.233970753655793,
+      "grad_norm": 0.10270056261316057,
+      "learning_rate": 5.592131351275671e-06,
+      "loss": 0.4488,
+      "step": 470
+    },
+    {
+      "epoch": 4.2429696287964,
+      "grad_norm": 0.0996620878043424,
+      "learning_rate": 5.464389737186348e-06,
+      "loss": 0.4479,
+      "step": 471
+    },
+    {
+      "epoch": 4.251968503937007,
+      "grad_norm": 0.10061707844415065,
+      "learning_rate": 5.338016999869351e-06,
+      "loss": 0.4483,
+      "step": 472
+    },
+    {
+      "epoch": 4.2609673790776155,
+      "grad_norm": 0.10888494466945672,
+      "learning_rate": 5.213018148319835e-06,
+      "loss": 0.4517,
+      "step": 473
+    },
+    {
+      "epoch": 4.269966254218223,
+      "grad_norm": 0.10212521131683365,
+      "learning_rate": 5.089398137076704e-06,
+      "loss": 0.4506,
+      "step": 474
+    },
+    {
+      "epoch": 4.27896512935883,
+      "grad_norm": 0.10500341792664687,
+      "learning_rate": 4.967161866026229e-06,
+      "loss": 0.451,
+      "step": 475
+    },
+    {
+      "epoch": 4.287964004499438,
+      "grad_norm": 0.10466244170953495,
+      "learning_rate": 4.846314180207774e-06,
+      "loss": 0.4473,
+      "step": 476
+    },
+    {
+      "epoch": 4.296962879640045,
+      "grad_norm": 0.09722711168374878,
+      "learning_rate": 4.726859869621847e-06,
+      "loss": 0.4481,
+      "step": 477
+    },
+    {
+      "epoch": 4.305961754780652,
+      "grad_norm": 0.10204450180802838,
+      "learning_rate": 4.608803669040187e-06,
+      "loss": 0.4454,
+      "step": 478
+    },
+    {
+      "epoch": 4.31496062992126,
+      "grad_norm": 0.0995565612941236,
+      "learning_rate": 4.492150257818066e-06,
+      "loss": 0.4498,
+      "step": 479
+    },
+    {
+      "epoch": 4.323959505061867,
+      "grad_norm": 0.10534766800687224,
+      "learning_rate": 4.376904259708892e-06,
+      "loss": 0.4528,
+      "step": 480
+    },
+    {
+      "epoch": 4.3329583802024745,
+      "grad_norm": 0.09462904906757555,
+      "learning_rate": 4.263070242680866e-06,
+      "loss": 0.447,
+      "step": 481
+    },
+    {
+      "epoch": 4.3419572553430825,
+      "grad_norm": 0.09711700163085293,
+      "learning_rate": 4.1506527187359765e-06,
+      "loss": 0.4552,
+      "step": 482
+    },
+    {
+      "epoch": 4.35095613048369,
+      "grad_norm": 0.09726591017318084,
+      "learning_rate": 4.039656143731128e-06,
+      "loss": 0.449,
+      "step": 483
+    },
+    {
+      "epoch": 4.359955005624297,
+      "grad_norm": 0.10398933972434006,
+      "learning_rate": 3.930084917201508e-06,
+      "loss": 0.4541,
+      "step": 484
+    },
+    {
+      "epoch": 4.368953880764904,
+      "grad_norm": 0.09827974212844832,
+      "learning_rate": 3.821943382186275e-06,
+      "loss": 0.4472,
+      "step": 485
+    },
+    {
+      "epoch": 4.377952755905512,
+      "grad_norm": 0.09502494293413481,
+      "learning_rate": 3.715235825056338e-06,
+      "loss": 0.4541,
+      "step": 486
+    },
+    {
+      "epoch": 4.386951631046119,
+      "grad_norm": 0.09221995391540017,
+      "learning_rate": 3.609966475344493e-06,
+      "loss": 0.4507,
+      "step": 487
+    },
+    {
+      "epoch": 4.395950506186726,
+      "grad_norm": 0.09866352346696798,
+      "learning_rate": 3.506139505577779e-06,
+      "loss": 0.4538,
+      "step": 488
+    },
+    {
+      "epoch": 4.404949381327334,
+      "grad_norm": 0.09418594379926497,
+      "learning_rate": 3.4037590311121015e-06,
+      "loss": 0.4547,
+      "step": 489
+    },
+    {
+      "epoch": 4.4139482564679415,
+      "grad_norm": 0.09399762424115403,
+      "learning_rate": 3.302829109969072e-06,
+      "loss": 0.4516,
+      "step": 490
+    },
+    {
+      "epoch": 4.422947131608549,
+      "grad_norm": 0.09056395808047757,
+      "learning_rate": 3.2033537426752236e-06,
+      "loss": 0.4556,
+      "step": 491
+    },
+    {
+      "epoch": 4.431946006749157,
+      "grad_norm": 0.09212248672508358,
+      "learning_rate": 3.1053368721033974e-06,
+      "loss": 0.4566,
+      "step": 492
+    },
+    {
+      "epoch": 4.440944881889764,
+      "grad_norm": 0.0899827038953071,
+      "learning_rate": 3.0087823833164596e-06,
+      "loss": 0.4455,
+      "step": 493
+    },
+    {
+      "epoch": 4.449943757030371,
+      "grad_norm": 0.09311934934388971,
+      "learning_rate": 2.9136941034133424e-06,
+      "loss": 0.4469,
+      "step": 494
+    },
+    {
+      "epoch": 4.458942632170979,
+      "grad_norm": 0.0942657183382691,
+      "learning_rate": 2.8200758013773313e-06,
+      "loss": 0.4501,
+      "step": 495
+    },
+    {
+      "epoch": 4.467941507311586,
+      "grad_norm": 0.09095763881082566,
+      "learning_rate": 2.7279311879266645e-06,
+      "loss": 0.4522,
+      "step": 496
+    },
+    {
+      "epoch": 4.476940382452193,
+      "grad_norm": 0.09128227123247035,
+      "learning_rate": 2.637263915367476e-06,
+      "loss": 0.4539,
+      "step": 497
+    },
+    {
+      "epoch": 4.4859392575928005,
+      "grad_norm": 0.0896975324322128,
+      "learning_rate": 2.5480775774490195e-06,
+      "loss": 0.4505,
+      "step": 498
+    },
+    {
+      "epoch": 4.494938132733409,
+      "grad_norm": 0.09373464127242032,
+      "learning_rate": 2.4603757092212057e-06,
+      "loss": 0.4506,
+      "step": 499
+    },
+    {
+      "epoch": 4.503937007874016,
+      "grad_norm": 0.09045392446753084,
+      "learning_rate": 2.374161786894513e-06,
+      "loss": 0.4535,
+      "step": 500
+    },
+    {
+      "epoch": 4.512935883014623,
+      "grad_norm": 0.09322805283679514,
+      "learning_rate": 2.2894392277022125e-06,
+      "loss": 0.4511,
+      "step": 501
+    },
+    {
+      "epoch": 4.521934758155231,
+      "grad_norm": 0.09493164625872488,
+      "learning_rate": 2.206211389764854e-06,
+      "loss": 0.4531,
+      "step": 502
+    },
+    {
+      "epoch": 4.530933633295838,
+      "grad_norm": 0.09141196271877726,
+      "learning_rate": 2.124481571957242e-06,
+      "loss": 0.4537,
+      "step": 503
+    },
+    {
+      "epoch": 4.539932508436445,
+      "grad_norm": 0.0875486494434062,
+      "learning_rate": 2.0442530137776374e-06,
+      "loss": 0.4431,
+      "step": 504
+    },
+    {
+      "epoch": 4.548931383577052,
+      "grad_norm": 0.09177205535576874,
+      "learning_rate": 1.9655288952193442e-06,
+      "loss": 0.4567,
+      "step": 505
+    },
+    {
+      "epoch": 4.55793025871766,
+      "grad_norm": 0.08907941923419668,
+      "learning_rate": 1.8883123366446955e-06,
+      "loss": 0.4507,
+      "step": 506
+    },
+    {
+      "epoch": 4.566929133858268,
+      "grad_norm": 0.09083887942548946,
+      "learning_rate": 1.8126063986613652e-06,
+      "loss": 0.45,
+      "step": 507
+    },
+    {
+      "epoch": 4.575928008998876,
+      "grad_norm": 0.08937321221433324,
+      "learning_rate": 1.7384140820010253e-06,
+      "loss": 0.4489,
+      "step": 508
+    },
+    {
+      "epoch": 4.584926884139483,
+      "grad_norm": 0.08801698232366557,
+      "learning_rate": 1.6657383274004545e-06,
+      "loss": 0.4534,
+      "step": 509
+    },
+    {
+      "epoch": 4.59392575928009,
+      "grad_norm": 0.08808939037763655,
+      "learning_rate": 1.5945820154849512e-06,
+      "loss": 0.4469,
+      "step": 510
+    },
+    {
+      "epoch": 4.602924634420697,
+      "grad_norm": 0.0903921066592697,
+      "learning_rate": 1.524947966654131e-06,
+      "loss": 0.4535,
+      "step": 511
+    },
+    {
+      "epoch": 4.611923509561305,
+      "grad_norm": 0.0899536812400529,
+      "learning_rate": 1.4568389409702e-06,
+      "loss": 0.4535,
+      "step": 512
+    },
+    {
+      "epoch": 4.620922384701912,
+      "grad_norm": 0.08721249506811954,
+      "learning_rate": 1.390257638048489e-06,
+      "loss": 0.4552,
+      "step": 513
+    },
+    {
+      "epoch": 4.6299212598425195,
+      "grad_norm": 0.08874618721675122,
+      "learning_rate": 1.3252066969504874e-06,
+      "loss": 0.4502,
+      "step": 514
+    },
+    {
+      "epoch": 4.6389201349831275,
+      "grad_norm": 0.09036518689753495,
+      "learning_rate": 1.261688696079233e-06,
+      "loss": 0.451,
+      "step": 515
+    },
+    {
+      "epoch": 4.647919010123735,
+      "grad_norm": 0.08929772055357187,
+      "learning_rate": 1.1997061530771004e-06,
+      "loss": 0.4513,
+      "step": 516
+    },
+    {
+      "epoch": 4.656917885264342,
+      "grad_norm": 0.08770289069775795,
+      "learning_rate": 1.1392615247260275e-06,
+      "loss": 0.4509,
+      "step": 517
+    },
+    {
+      "epoch": 4.665916760404949,
+      "grad_norm": 0.0861779113751004,
+      "learning_rate": 1.080357206850131e-06,
+      "loss": 0.4497,
+      "step": 518
+    },
+    {
+      "epoch": 4.674915635545557,
+      "grad_norm": 0.09018663786133461,
+      "learning_rate": 1.0229955342207254e-06,
+      "loss": 0.4556,
+      "step": 519
+    },
+    {
+      "epoch": 4.683914510686164,
+      "grad_norm": 0.08621786512649365,
+      "learning_rate": 9.67178780463809e-07,
+      "loss": 0.4537,
+      "step": 520
+    },
+    {
+      "epoch": 4.692913385826771,
+      "grad_norm": 0.08431669753032205,
+      "learning_rate": 9.129091579699412e-07,
+      "loss": 0.4534,
+      "step": 521
+    },
+    {
+      "epoch": 4.701912260967379,
+      "grad_norm": 0.08666217491427604,
+      "learning_rate": 8.601888178065177e-07,
+      "loss": 0.4501,
+      "step": 522
+    },
+    {
+      "epoch": 4.7109111361079865,
+      "grad_norm": 0.08654873368758471,
+      "learning_rate": 8.090198496325485e-07,
+      "loss": 0.4506,
+      "step": 523
+    },
+    {
+      "epoch": 4.719910011248594,
+      "grad_norm": 0.08652439676904433,
+      "learning_rate": 7.594042816158187e-07,
+      "loss": 0.4529,
+      "step": 524
+    },
+    {
+      "epoch": 4.728908886389202,
+      "grad_norm": 0.08831640949919531,
+      "learning_rate": 7.113440803524896e-07,
+      "loss": 0.4531,
+      "step": 525
+    },
+    {
+      "epoch": 4.737907761529809,
+      "grad_norm": 0.0866028677777177,
+      "learning_rate": 6.648411507891528e-07,
+      "loss": 0.4522,
+      "step": 526
+    },
+    {
+      "epoch": 4.746906636670416,
+      "grad_norm": 0.08731496032068242,
+      "learning_rate": 6.198973361473349e-07,
+      "loss": 0.4495,
+      "step": 527
+    },
+    {
+      "epoch": 4.755905511811024,
+      "grad_norm": 0.08758993950549104,
+      "learning_rate": 5.765144178504222e-07,
+      "loss": 0.4526,
+      "step": 528
+    },
+    {
+      "epoch": 4.764904386951631,
+      "grad_norm": 0.08831218479369252,
+      "learning_rate": 5.34694115453065e-07,
+      "loss": 0.4532,
+      "step": 529
+    },
+    {
+      "epoch": 4.773903262092238,
+      "grad_norm": 0.0860726407842809,
+      "learning_rate": 4.944380865730125e-07,
+      "loss": 0.4529,
+      "step": 530
+    },
+    {
+      "epoch": 4.7829021372328455,
+      "grad_norm": 0.0840695127911042,
+      "learning_rate": 4.5574792682541167e-07,
+      "loss": 0.4502,
+      "step": 531
+    },
+    {
+      "epoch": 4.791901012373454,
+      "grad_norm": 0.0827762260768475,
+      "learning_rate": 4.186251697595678e-07,
+      "loss": 0.4471,
+      "step": 532
+    },
+    {
+      "epoch": 4.800899887514061,
+      "grad_norm": 0.08825259092439537,
+      "learning_rate": 3.83071286798149e-07,
+      "loss": 0.4434,
+      "step": 533
+    },
+    {
+      "epoch": 4.809898762654668,
+      "grad_norm": 0.08855814892569033,
+      "learning_rate": 3.4908768717887286e-07,
+      "loss": 0.45,
+      "step": 534
+    },
+    {
+      "epoch": 4.818897637795276,
+      "grad_norm": 0.08776896957164552,
+      "learning_rate": 3.1667571789864015e-07,
+      "loss": 0.4539,
+      "step": 535
+    },
+    {
+      "epoch": 4.827896512935883,
+      "grad_norm": 0.0848390478817017,
+      "learning_rate": 2.858366636601639e-07,
+      "loss": 0.451,
+      "step": 536
+    },
+    {
+      "epoch": 4.83689538807649,
+      "grad_norm": 0.0832820856667789,
+      "learning_rate": 2.5657174682101936e-07,
+      "loss": 0.4501,
+      "step": 537
+    },
+    {
+      "epoch": 4.845894263217097,
+      "grad_norm": 0.08506972205806178,
+      "learning_rate": 2.288821273452113e-07,
+      "loss": 0.4495,
+      "step": 538
+    },
+    {
+      "epoch": 4.854893138357705,
+      "grad_norm": 0.08489253043277004,
+      "learning_rate": 2.027689027572066e-07,
+      "loss": 0.4527,
+      "step": 539
+    },
+    {
+      "epoch": 4.863892013498313,
+      "grad_norm": 0.08426467385333178,
+      "learning_rate": 1.7823310809840456e-07,
+      "loss": 0.4464,
+      "step": 540
+    },
+    {
+      "epoch": 4.872890888638921,
+      "grad_norm": 0.08767870830184761,
+      "learning_rate": 1.55275715886134e-07,
+      "loss": 0.4519,
+      "step": 541
+    },
+    {
+      "epoch": 4.881889763779528,
+      "grad_norm": 0.08723387825162499,
+      "learning_rate": 1.3389763607509765e-07,
+      "loss": 0.455,
+      "step": 542
+    },
+    {
+      "epoch": 4.890888638920135,
+      "grad_norm": 0.08523501703956923,
+      "learning_rate": 1.1409971602130754e-07,
+      "loss": 0.4502,
+      "step": 543
+    },
+    {
+      "epoch": 4.899887514060742,
+      "grad_norm": 0.08495566625522424,
+      "learning_rate": 9.588274044848523e-08,
+      "loss": 0.4541,
+      "step": 544
+    },
+    {
+      "epoch": 4.90888638920135,
+      "grad_norm": 0.08415284728722193,
+      "learning_rate": 7.924743141698888e-08,
+      "loss": 0.4534,
+      "step": 545
+    },
+    {
+      "epoch": 4.917885264341957,
+      "grad_norm": 0.0899650650008716,
+      "learning_rate": 6.419444829515175e-08,
+      "loss": 0.4514,
+      "step": 546
+    },
+    {
+      "epoch": 4.926884139482564,
+      "grad_norm": 0.08441613482621164,
+      "learning_rate": 5.072438773318755e-08,
+      "loss": 0.4483,
+      "step": 547
+    },
+    {
+      "epoch": 4.9358830146231725,
+      "grad_norm": 0.08443189586686374,
+      "learning_rate": 3.8837783639507076e-08,
+      "loss": 0.453,
+      "step": 548
+    },
+    {
+      "epoch": 4.94488188976378,
+      "grad_norm": 0.08693922903585893,
+      "learning_rate": 2.8535107159584076e-08,
+      "loss": 0.4519,
+      "step": 549
+    },
+    {
+      "epoch": 4.953880764904387,
+      "grad_norm": 0.0869995258137961,
+      "learning_rate": 1.9816766657254626e-08,
+      "loss": 0.4498,
+      "step": 550
+    },
+    {
+      "epoch": 4.962879640044994,
+      "grad_norm": 0.08843911191962896,
+      "learning_rate": 1.268310769855674e-08,
+      "loss": 0.4495,
+      "step": 551
+    },
+    {
+      "epoch": 4.971878515185602,
+      "grad_norm": 0.08550795511559632,
+      "learning_rate": 7.134413038012433e-09,
+      "loss": 0.4487,
+      "step": 552
+    },
+    {
+      "epoch": 4.980877390326209,
+      "grad_norm": 0.08541041297147574,
+      "learning_rate": 3.170902607432247e-09,
+      "loss": 0.4502,
+      "step": 553
+    },
+    {
+      "epoch": 4.989876265466816,
+      "grad_norm": 0.08600009481079209,
+      "learning_rate": 7.92733507188892e-10,
+      "loss": 0.4553,
+      "step": 554
+    },
+    {
+      "epoch": 4.998875140607424,
+      "grad_norm": 0.0856629080592209,
+      "learning_rate": 0.0,
+      "loss": 0.4591,
+      "step": 555
+    },
+    {
+      "epoch": 4.998875140607424,
+      "step": 555,
+      "total_flos": 2.8638155373837025e+18,
+      "train_loss": 0.09022162760700192,
+      "train_runtime": 11331.6206,
+      "train_samples_per_second": 25.101,
+      "train_steps_per_second": 0.049
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 555,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 2.8638155373837025e+18,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}