diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,8500 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.997622444127437,
+  "eval_steps": 50,
+  "global_step": 1182,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.002536059597400539,
+      "grad_norm": 11.653282165527344,
+      "learning_rate": 1.680672268907563e-07,
+      "loss": 0.8056,
+      "step": 1
+    },
+    {
+      "epoch": 0.005072119194801078,
+      "grad_norm": 12.802275657653809,
+      "learning_rate": 3.361344537815126e-07,
+      "loss": 0.8145,
+      "step": 2
+    },
+    {
+      "epoch": 0.007608178792201617,
+      "grad_norm": 12.750452041625977,
+      "learning_rate": 5.042016806722689e-07,
+      "loss": 0.803,
+      "step": 3
+    },
+    {
+      "epoch": 0.010144238389602156,
+      "grad_norm": 13.883561134338379,
+      "learning_rate": 6.722689075630252e-07,
+      "loss": 0.8006,
+      "step": 4
+    },
+    {
+      "epoch": 0.012680297987002695,
+      "grad_norm": 12.906583786010742,
+      "learning_rate": 8.403361344537816e-07,
+      "loss": 0.8117,
+      "step": 5
+    },
+    {
+      "epoch": 0.015216357584403234,
+      "grad_norm": 11.863070487976074,
+      "learning_rate": 1.0084033613445378e-06,
+      "loss": 0.8119,
+      "step": 6
+    },
+    {
+      "epoch": 0.017752417181803773,
+      "grad_norm": 10.837064743041992,
+      "learning_rate": 1.1764705882352942e-06,
+      "loss": 0.8135,
+      "step": 7
+    },
+    {
+      "epoch": 0.020288476779204312,
+      "grad_norm": 9.839982986450195,
+      "learning_rate": 1.3445378151260504e-06,
+      "loss": 0.8255,
+      "step": 8
+    },
+    {
+      "epoch": 0.02282453637660485,
+      "grad_norm": 9.438532829284668,
+      "learning_rate": 1.5126050420168068e-06,
+      "loss": 0.804,
+      "step": 9
+    },
+    {
+      "epoch": 0.02536059597400539,
+      "grad_norm": 9.155749320983887,
+      "learning_rate": 1.6806722689075632e-06,
+      "loss": 0.7807,
+      "step": 10
+    },
+    {
+      "epoch": 0.02789665557140593,
+      "grad_norm": 9.168850898742676,
+      "learning_rate": 1.8487394957983196e-06,
+      "loss": 0.7859,
+      "step": 11
+    },
+    {
+      "epoch": 0.030432715168806468,
+      "grad_norm": 9.751949310302734,
+      "learning_rate": 2.0168067226890756e-06,
+      "loss": 0.7879,
+      "step": 12
+    },
+    {
+      "epoch": 0.03296877476620701,
+      "grad_norm": 8.917466163635254,
+      "learning_rate": 2.184873949579832e-06,
+      "loss": 0.7855,
+      "step": 13
+    },
+    {
+      "epoch": 0.035504834363607546,
+      "grad_norm": 8.31714916229248,
+      "learning_rate": 2.3529411764705885e-06,
+      "loss": 0.766,
+      "step": 14
+    },
+    {
+      "epoch": 0.038040893961008085,
+      "grad_norm": 7.515654563903809,
+      "learning_rate": 2.521008403361345e-06,
+      "loss": 0.7651,
+      "step": 15
+    },
+    {
+      "epoch": 0.040576953558408624,
+      "grad_norm": 6.223039150238037,
+      "learning_rate": 2.689075630252101e-06,
+      "loss": 0.7663,
+      "step": 16
+    },
+    {
+      "epoch": 0.04311301315580916,
+      "grad_norm": 6.400880336761475,
+      "learning_rate": 2.8571428571428573e-06,
+      "loss": 0.7875,
+      "step": 17
+    },
+    {
+      "epoch": 0.0456490727532097,
+      "grad_norm": 7.137135028839111,
+      "learning_rate": 3.0252100840336137e-06,
+      "loss": 0.7452,
+      "step": 18
+    },
+    {
+      "epoch": 0.04818513235061024,
+      "grad_norm": 7.165759563446045,
+      "learning_rate": 3.1932773109243696e-06,
+      "loss": 0.7739,
+      "step": 19
+    },
+    {
+      "epoch": 0.05072119194801078,
+      "grad_norm": 7.375353813171387,
+      "learning_rate": 3.3613445378151265e-06,
+      "loss": 0.7337,
+      "step": 20
+    },
+    {
+      "epoch": 0.05325725154541132,
+      "grad_norm": 5.602230072021484,
+      "learning_rate": 3.529411764705883e-06,
+      "loss": 0.7512,
+      "step": 21
+    },
+    {
+      "epoch": 0.05579331114281186,
+      "grad_norm": 5.487696170806885,
+      "learning_rate": 3.6974789915966393e-06,
+      "loss": 0.7517,
+      "step": 22
+    },
+    {
+      "epoch": 0.0583293707402124,
+      "grad_norm": 5.460212707519531,
+      "learning_rate": 3.865546218487396e-06,
+      "loss": 0.7431,
+      "step": 23
+    },
+    {
+      "epoch": 0.060865430337612936,
+      "grad_norm": 5.078911304473877,
+      "learning_rate": 4.033613445378151e-06,
+      "loss": 0.7189,
+      "step": 24
+    },
+    {
+      "epoch": 0.06340148993501347,
+      "grad_norm": 4.588583469390869,
+      "learning_rate": 4.201680672268908e-06,
+      "loss": 0.7387,
+      "step": 25
+    },
+    {
+      "epoch": 0.06593754953241401,
+      "grad_norm": 4.4840593338012695,
+      "learning_rate": 4.369747899159664e-06,
+      "loss": 0.746,
+      "step": 26
+    },
+    {
+      "epoch": 0.06847360912981455,
+      "grad_norm": 5.0837249755859375,
+      "learning_rate": 4.5378151260504205e-06,
+      "loss": 0.7648,
+      "step": 27
+    },
+    {
+      "epoch": 0.07100966872721509,
+      "grad_norm": 4.900995254516602,
+      "learning_rate": 4.705882352941177e-06,
+      "loss": 0.7173,
+      "step": 28
+    },
+    {
+      "epoch": 0.07354572832461563,
+      "grad_norm": 4.927539348602295,
+      "learning_rate": 4.873949579831933e-06,
+      "loss": 0.7336,
+      "step": 29
+    },
+    {
+      "epoch": 0.07608178792201617,
+      "grad_norm": 5.527489185333252,
+      "learning_rate": 5.04201680672269e-06,
+      "loss": 0.7336,
+      "step": 30
+    },
+    {
+      "epoch": 0.07861784751941671,
+      "grad_norm": 7.812469959259033,
+      "learning_rate": 5.210084033613446e-06,
+      "loss": 0.7495,
+      "step": 31
+    },
+    {
+      "epoch": 0.08115390711681725,
+      "grad_norm": 4.948735237121582,
+      "learning_rate": 5.378151260504202e-06,
+      "loss": 0.7212,
+      "step": 32
+    },
+    {
+      "epoch": 0.08368996671421779,
+      "grad_norm": 3.99983811378479,
+      "learning_rate": 5.546218487394959e-06,
+      "loss": 0.7289,
+      "step": 33
+    },
+    {
+      "epoch": 0.08622602631161833,
+      "grad_norm": 4.154791831970215,
+      "learning_rate": 5.7142857142857145e-06,
+      "loss": 0.709,
+      "step": 34
+    },
+    {
+      "epoch": 0.08876208590901886,
+      "grad_norm": 4.3912739753723145,
+      "learning_rate": 5.882352941176471e-06,
+      "loss": 0.7366,
+      "step": 35
+    },
+    {
+      "epoch": 0.0912981455064194,
+      "grad_norm": 6.254603862762451,
+      "learning_rate": 6.050420168067227e-06,
+      "loss": 0.7402,
+      "step": 36
+    },
+    {
+      "epoch": 0.09383420510381994,
+      "grad_norm": 3.847163200378418,
+      "learning_rate": 6.218487394957984e-06,
+      "loss": 0.7107,
+      "step": 37
+    },
+    {
+      "epoch": 0.09637026470122048,
+      "grad_norm": 5.022808074951172,
+      "learning_rate": 6.386554621848739e-06,
+      "loss": 0.7229,
+      "step": 38
+    },
+    {
+      "epoch": 0.09890632429862102,
+      "grad_norm": 4.145415782928467,
+      "learning_rate": 6.5546218487394966e-06,
+      "loss": 0.7056,
+      "step": 39
+    },
+    {
+      "epoch": 0.10144238389602156,
+      "grad_norm": 4.076045513153076,
+      "learning_rate": 6.722689075630253e-06,
+      "loss": 0.7206,
+      "step": 40
+    },
+    {
+      "epoch": 0.1039784434934221,
+      "grad_norm": 3.7817814350128174,
+      "learning_rate": 6.8907563025210085e-06,
+      "loss": 0.714,
+      "step": 41
+    },
+    {
+      "epoch": 0.10651450309082264,
+      "grad_norm": 3.7016332149505615,
+      "learning_rate": 7.058823529411766e-06,
+      "loss": 0.7161,
+      "step": 42
+    },
+    {
+      "epoch": 0.10905056268822318,
+      "grad_norm": 4.031497955322266,
+      "learning_rate": 7.226890756302521e-06,
+      "loss": 0.7269,
+      "step": 43
+    },
+    {
+      "epoch": 0.11158662228562372,
+      "grad_norm": 4.782078266143799,
+      "learning_rate": 7.394957983193279e-06,
+      "loss": 0.7326,
+      "step": 44
+    },
+    {
+      "epoch": 0.11412268188302425,
+      "grad_norm": 4.8900065422058105,
+      "learning_rate": 7.563025210084034e-06,
+      "loss": 0.7149,
+      "step": 45
+    },
+    {
+      "epoch": 0.1166587414804248,
+      "grad_norm": 4.996344089508057,
+      "learning_rate": 7.731092436974791e-06,
+      "loss": 0.6888,
+      "step": 46
+    },
+    {
+      "epoch": 0.11919480107782533,
+      "grad_norm": 4.4249396324157715,
+      "learning_rate": 7.899159663865547e-06,
+      "loss": 0.72,
+      "step": 47
+    },
+    {
+      "epoch": 0.12173086067522587,
+      "grad_norm": 3.6222777366638184,
+      "learning_rate": 8.067226890756303e-06,
+      "loss": 0.7381,
+      "step": 48
+    },
+    {
+      "epoch": 0.12426692027262641,
+      "grad_norm": 3.631304979324341,
+      "learning_rate": 8.23529411764706e-06,
+      "loss": 0.6974,
+      "step": 49
+    },
+    {
+      "epoch": 0.12680297987002695,
+      "grad_norm": 3.6000781059265137,
+      "learning_rate": 8.403361344537815e-06,
+      "loss": 0.6991,
+      "step": 50
+    },
+    {
+      "epoch": 0.12680297987002695,
+      "eval_loss": NaN,
+      "eval_runtime": 232.2001,
+      "eval_samples_per_second": 35.81,
+      "eval_steps_per_second": 4.479,
+      "step": 50
+    },
+    {
+      "epoch": 0.12933903946742747,
+      "grad_norm": 3.4208874702453613,
+      "learning_rate": 8.571428571428571e-06,
+      "loss": 0.726,
+      "step": 51
+    },
+    {
+      "epoch": 0.13187509906482803,
+      "grad_norm": 3.458233594894409,
+      "learning_rate": 8.739495798319328e-06,
+      "loss": 0.7188,
+      "step": 52
+    },
+    {
+      "epoch": 0.13441115866222855,
+      "grad_norm": 3.3983688354492188,
+      "learning_rate": 8.907563025210085e-06,
+      "loss": 0.7008,
+      "step": 53
+    },
+    {
+      "epoch": 0.1369472182596291,
+      "grad_norm": 3.452065944671631,
+      "learning_rate": 9.075630252100841e-06,
+      "loss": 0.7077,
+      "step": 54
+    },
+    {
+      "epoch": 0.13948327785702963,
+      "grad_norm": 3.2041687965393066,
+      "learning_rate": 9.243697478991598e-06,
+      "loss": 0.7191,
+      "step": 55
+    },
+    {
+      "epoch": 0.14201933745443018,
+      "grad_norm": 3.414273977279663,
+      "learning_rate": 9.411764705882354e-06,
+      "loss": 0.7138,
+      "step": 56
+    },
+    {
+      "epoch": 0.1445553970518307,
+      "grad_norm": 3.3661060333251953,
+      "learning_rate": 9.579831932773111e-06,
+      "loss": 0.7116,
+      "step": 57
+    },
+    {
+      "epoch": 0.14709145664923126,
+      "grad_norm": 3.3029797077178955,
+      "learning_rate": 9.747899159663867e-06,
+      "loss": 0.7025,
+      "step": 58
+    },
+    {
+      "epoch": 0.1496275162466318,
+      "grad_norm": 3.7844340801239014,
+      "learning_rate": 9.915966386554622e-06,
+      "loss": 0.703,
+      "step": 59
+    },
+    {
+      "epoch": 0.15216357584403234,
+      "grad_norm": 3.9964404106140137,
+      "learning_rate": 1.008403361344538e-05,
+      "loss": 0.7117,
+      "step": 60
+    },
+    {
+      "epoch": 0.15469963544143286,
+      "grad_norm": 3.33013916015625,
+      "learning_rate": 1.0252100840336135e-05,
+      "loss": 0.6834,
+      "step": 61
+    },
+    {
+      "epoch": 0.15723569503883342,
+      "grad_norm": 3.5664029121398926,
+      "learning_rate": 1.0420168067226892e-05,
+      "loss": 0.7115,
+      "step": 62
+    },
+    {
+      "epoch": 0.15977175463623394,
+      "grad_norm": 4.009420871734619,
+      "learning_rate": 1.0588235294117648e-05,
+      "loss": 0.6972,
+      "step": 63
+    },
+    {
+      "epoch": 0.1623078142336345,
+      "grad_norm": 5.011332035064697,
+      "learning_rate": 1.0756302521008403e-05,
+      "loss": 0.6925,
+      "step": 64
+    },
+    {
+      "epoch": 0.16484387383103502,
+      "grad_norm": 3.8215227127075195,
+      "learning_rate": 1.0924369747899159e-05,
+      "loss": 0.7115,
+      "step": 65
+    },
+    {
+      "epoch": 0.16737993342843557,
+      "grad_norm": 4.597329139709473,
+      "learning_rate": 1.1092436974789918e-05,
+      "loss": 0.7223,
+      "step": 66
+    },
+    {
+      "epoch": 0.1699159930258361,
+      "grad_norm": 5.164331436157227,
+      "learning_rate": 1.1260504201680673e-05,
+      "loss": 0.7082,
+      "step": 67
+    },
+    {
+      "epoch": 0.17245205262323665,
+      "grad_norm": 4.405480861663818,
+      "learning_rate": 1.1428571428571429e-05,
+      "loss": 0.7056,
+      "step": 68
+    },
+    {
+      "epoch": 0.17498811222063718,
+      "grad_norm": 3.8048319816589355,
+      "learning_rate": 1.1596638655462186e-05,
+      "loss": 0.7068,
+      "step": 69
+    },
+    {
+      "epoch": 0.17752417181803773,
+      "grad_norm": 4.435548782348633,
+      "learning_rate": 1.1764705882352942e-05,
+      "loss": 0.6969,
+      "step": 70
+    },
+    {
+      "epoch": 0.18006023141543825,
+      "grad_norm": 4.299891471862793,
+      "learning_rate": 1.1932773109243699e-05,
+      "loss": 0.7358,
+      "step": 71
+    },
+    {
+      "epoch": 0.1825962910128388,
+      "grad_norm": 3.5945470333099365,
+      "learning_rate": 1.2100840336134455e-05,
+      "loss": 0.7301,
+      "step": 72
+    },
+    {
+      "epoch": 0.18513235061023933,
+      "grad_norm": 3.5449764728546143,
+      "learning_rate": 1.2268907563025212e-05,
+      "loss": 0.7005,
+      "step": 73
+    },
+    {
+      "epoch": 0.18766841020763989,
+      "grad_norm": 3.3907058238983154,
+      "learning_rate": 1.2436974789915967e-05,
+      "loss": 0.7144,
+      "step": 74
+    },
+    {
+      "epoch": 0.1902044698050404,
+      "grad_norm": 3.4866995811462402,
+      "learning_rate": 1.2605042016806723e-05,
+      "loss": 0.7169,
+      "step": 75
+    },
+    {
+      "epoch": 0.19274052940244096,
+      "grad_norm": 3.2452280521392822,
+      "learning_rate": 1.2773109243697479e-05,
+      "loss": 0.7134,
+      "step": 76
+    },
+    {
+      "epoch": 0.1952765889998415,
+      "grad_norm": 3.4761757850646973,
+      "learning_rate": 1.2941176470588238e-05,
+      "loss": 0.7193,
+      "step": 77
+    },
+    {
+      "epoch": 0.19781264859724204,
+      "grad_norm": 3.327453851699829,
+      "learning_rate": 1.3109243697478993e-05,
+      "loss": 0.714,
+      "step": 78
+    },
+    {
+      "epoch": 0.20034870819464257,
+      "grad_norm": 3.3360109329223633,
+      "learning_rate": 1.3277310924369749e-05,
+      "loss": 0.7215,
+      "step": 79
+    },
+    {
+      "epoch": 0.20288476779204312,
+      "grad_norm": 3.1527507305145264,
+      "learning_rate": 1.3445378151260506e-05,
+      "loss": 0.6998,
+      "step": 80
+    },
+    {
+      "epoch": 0.20542082738944364,
+      "grad_norm": 3.2852983474731445,
+      "learning_rate": 1.3613445378151261e-05,
+      "loss": 0.709,
+      "step": 81
+    },
+    {
+      "epoch": 0.2079568869868442,
+      "grad_norm": 3.1540558338165283,
+      "learning_rate": 1.3781512605042017e-05,
+      "loss": 0.7073,
+      "step": 82
+    },
+    {
+      "epoch": 0.21049294658424472,
+      "grad_norm": 3.7169458866119385,
+      "learning_rate": 1.3949579831932774e-05,
+      "loss": 0.7147,
+      "step": 83
+    },
+    {
+      "epoch": 0.21302900618164528,
+      "grad_norm": 3.803159236907959,
+      "learning_rate": 1.4117647058823532e-05,
+      "loss": 0.7111,
+      "step": 84
+    },
+    {
+      "epoch": 0.2155650657790458,
+      "grad_norm": 4.325738906860352,
+      "learning_rate": 1.4285714285714287e-05,
+      "loss": 0.685,
+      "step": 85
+    },
+    {
+      "epoch": 0.21810112537644635,
+      "grad_norm": 5.129235744476318,
+      "learning_rate": 1.4453781512605043e-05,
+      "loss": 0.7056,
+      "step": 86
+    },
+    {
+      "epoch": 0.22063718497384688,
+      "grad_norm": 4.4139018058776855,
+      "learning_rate": 1.4621848739495798e-05,
+      "loss": 0.7208,
+      "step": 87
+    },
+    {
+      "epoch": 0.22317324457124743,
+      "grad_norm": 3.9154489040374756,
+      "learning_rate": 1.4789915966386557e-05,
+      "loss": 0.7103,
+      "step": 88
+    },
+    {
+      "epoch": 0.22570930416864796,
+      "grad_norm": 3.9222123622894287,
+      "learning_rate": 1.4957983193277313e-05,
+      "loss": 0.7058,
+      "step": 89
+    },
+    {
+      "epoch": 0.2282453637660485,
+      "grad_norm": 3.79371976852417,
+      "learning_rate": 1.5126050420168068e-05,
+      "loss": 0.7068,
+      "step": 90
+    },
+    {
+      "epoch": 0.23078142336344903,
+      "grad_norm": 4.0589494705200195,
+      "learning_rate": 1.5294117647058822e-05,
+      "loss": 0.7105,
+      "step": 91
+    },
+    {
+      "epoch": 0.2333174829608496,
+      "grad_norm": 3.5458948612213135,
+      "learning_rate": 1.5462184873949583e-05,
+      "loss": 0.7054,
+      "step": 92
+    },
+    {
+      "epoch": 0.2358535425582501,
+      "grad_norm": 3.477810859680176,
+      "learning_rate": 1.5630252100840337e-05,
+      "loss": 0.729,
+      "step": 93
+    },
+    {
+      "epoch": 0.23838960215565067,
+      "grad_norm": 3.700502872467041,
+      "learning_rate": 1.5798319327731094e-05,
+      "loss": 0.7027,
+      "step": 94
+    },
+    {
+      "epoch": 0.2409256617530512,
+      "grad_norm": 3.3739285469055176,
+      "learning_rate": 1.596638655462185e-05,
+      "loss": 0.7106,
+      "step": 95
+    },
+    {
+      "epoch": 0.24346172135045174,
+      "grad_norm": 3.449275493621826,
+      "learning_rate": 1.6134453781512605e-05,
+      "loss": 0.6852,
+      "step": 96
+    },
+    {
+      "epoch": 0.24599778094785227,
+      "grad_norm": 3.827040672302246,
+      "learning_rate": 1.6302521008403362e-05,
+      "loss": 0.7047,
+      "step": 97
+    },
+    {
+      "epoch": 0.24853384054525282,
+      "grad_norm": 3.4870052337646484,
+      "learning_rate": 1.647058823529412e-05,
+      "loss": 0.6909,
+      "step": 98
+    },
+    {
+      "epoch": 0.25106990014265335,
+      "grad_norm": 3.3494443893432617,
+      "learning_rate": 1.6638655462184877e-05,
+      "loss": 0.7115,
+      "step": 99
+    },
+    {
+      "epoch": 0.2536059597400539,
+      "grad_norm": 3.547975778579712,
+      "learning_rate": 1.680672268907563e-05,
+      "loss": 0.7057,
+      "step": 100
+    },
+    {
+      "epoch": 0.2536059597400539,
+      "eval_loss": NaN,
+      "eval_runtime": 206.4023,
+      "eval_samples_per_second": 40.285,
+      "eval_steps_per_second": 5.039,
+      "step": 100
+    },
+    {
+      "epoch": 0.25614201933745445,
+      "grad_norm": 4.909302711486816,
+      "learning_rate": 1.6974789915966388e-05,
+      "loss": 0.705,
+      "step": 101
+    },
+    {
+      "epoch": 0.25867807893485495,
+      "grad_norm": 5.4081597328186035,
+      "learning_rate": 1.7142857142857142e-05,
+      "loss": 0.7326,
+      "step": 102
+    },
+    {
+      "epoch": 0.2612141385322555,
+      "grad_norm": 3.250433921813965,
+      "learning_rate": 1.7310924369747902e-05,
+      "loss": 0.7102,
+      "step": 103
+    },
+    {
+      "epoch": 0.26375019812965605,
+      "grad_norm": 7.251071453094482,
+      "learning_rate": 1.7478991596638656e-05,
+      "loss": 0.7296,
+      "step": 104
+    },
+    {
+      "epoch": 0.2662862577270566,
+      "grad_norm": 3.968695878982544,
+      "learning_rate": 1.7647058823529414e-05,
+      "loss": 0.7343,
+      "step": 105
+    },
+    {
+      "epoch": 0.2688223173244571,
+      "grad_norm": 8.821077346801758,
+      "learning_rate": 1.781512605042017e-05,
+      "loss": 0.7092,
+      "step": 106
+    },
+    {
+      "epoch": 0.27135837692185766,
+      "grad_norm": 7.964052200317383,
+      "learning_rate": 1.7983193277310925e-05,
+      "loss": 0.7259,
+      "step": 107
+    },
+    {
+      "epoch": 0.2738944365192582,
+      "grad_norm": 5.083383083343506,
+      "learning_rate": 1.8151260504201682e-05,
+      "loss": 0.7145,
+      "step": 108
+    },
+    {
+      "epoch": 0.27643049611665876,
+      "grad_norm": 4.314772605895996,
+      "learning_rate": 1.831932773109244e-05,
+      "loss": 0.7166,
+      "step": 109
+    },
+    {
+      "epoch": 0.27896655571405926,
+      "grad_norm": 4.2335333824157715,
+      "learning_rate": 1.8487394957983196e-05,
+      "loss": 0.7182,
+      "step": 110
+    },
+    {
+      "epoch": 0.2815026153114598,
+      "grad_norm": 3.759748697280884,
+      "learning_rate": 1.865546218487395e-05,
+      "loss": 0.7344,
+      "step": 111
+    },
+    {
+      "epoch": 0.28403867490886037,
+      "grad_norm": 3.97318696975708,
+      "learning_rate": 1.8823529411764708e-05,
+      "loss": 0.7199,
+      "step": 112
+    },
+    {
+      "epoch": 0.2865747345062609,
+      "grad_norm": 3.738929510116577,
+      "learning_rate": 1.899159663865546e-05,
+      "loss": 0.7346,
+      "step": 113
+    },
+    {
+      "epoch": 0.2891107941036614,
+      "grad_norm": 3.3690345287323,
+      "learning_rate": 1.9159663865546222e-05,
+      "loss": 0.7337,
+      "step": 114
+    },
+    {
+      "epoch": 0.29164685370106197,
+      "grad_norm": 3.4982593059539795,
+      "learning_rate": 1.9327731092436976e-05,
+      "loss": 0.7121,
+      "step": 115
+    },
+    {
+      "epoch": 0.2941829132984625,
+      "grad_norm": 3.4797208309173584,
+      "learning_rate": 1.9495798319327733e-05,
+      "loss": 0.7193,
+      "step": 116
+    },
+    {
+      "epoch": 0.2967189728958631,
+      "grad_norm": 3.3013737201690674,
+      "learning_rate": 1.9663865546218487e-05,
+      "loss": 0.7145,
+      "step": 117
+    },
+    {
+      "epoch": 0.2992550324932636,
+      "grad_norm": 3.1991143226623535,
+      "learning_rate": 1.9831932773109244e-05,
+      "loss": 0.7191,
+      "step": 118
+    },
+    {
+      "epoch": 0.3017910920906641,
+      "grad_norm": 2.994180917739868,
+      "learning_rate": 2e-05,
+      "loss": 0.7153,
+      "step": 119
+    },
+    {
+      "epoch": 0.3043271516880647,
+      "grad_norm": 3.2345073223114014,
+      "learning_rate": 1.9999956328017772e-05,
+      "loss": 0.702,
+      "step": 120
+    },
+    {
+      "epoch": 0.30686321128546523,
+      "grad_norm": 3.233721971511841,
+      "learning_rate": 1.9999825312452537e-05,
+      "loss": 0.7001,
+      "step": 121
+    },
+    {
+      "epoch": 0.30939927088286573,
+      "grad_norm": 3.098071575164795,
+      "learning_rate": 1.9999606954448635e-05,
+      "loss": 0.6975,
+      "step": 122
+    },
+    {
+      "epoch": 0.3119353304802663,
+      "grad_norm": 2.9595870971679688,
+      "learning_rate": 1.9999301255913288e-05,
+      "loss": 0.6998,
+      "step": 123
+    },
+    {
+      "epoch": 0.31447139007766683,
+      "grad_norm": 2.9322428703308105,
+      "learning_rate": 1.9998908219516596e-05,
+      "loss": 0.7082,
+      "step": 124
+    },
+    {
+      "epoch": 0.3170074496750674,
+      "grad_norm": 3.709397792816162,
+      "learning_rate": 1.9998427848691487e-05,
+      "loss": 0.7045,
+      "step": 125
+    },
+    {
+      "epoch": 0.3195435092724679,
+      "grad_norm": 3.301387310028076,
+      "learning_rate": 1.9997860147633717e-05,
+      "loss": 0.717,
+      "step": 126
+    },
+    {
+      "epoch": 0.32207956886986844,
+      "grad_norm": 4.1625075340271,
+      "learning_rate": 1.999720512130181e-05,
+      "loss": 0.6957,
+      "step": 127
+    },
+    {
+      "epoch": 0.324615628467269,
+      "grad_norm": 4.067363262176514,
+      "learning_rate": 1.999646277541702e-05,
+      "loss": 0.7147,
+      "step": 128
+    },
+    {
+      "epoch": 0.32715168806466954,
+      "grad_norm": 4.256079196929932,
+      "learning_rate": 1.99956331164633e-05,
+      "loss": 0.7155,
+      "step": 129
+    },
+    {
+      "epoch": 0.32968774766207004,
+      "grad_norm": 3.9020440578460693,
+      "learning_rate": 1.9994716151687213e-05,
+      "loss": 0.7298,
+      "step": 130
+    },
+    {
+      "epoch": 0.3322238072594706,
+      "grad_norm": 3.5549631118774414,
+      "learning_rate": 1.9993711889097892e-05,
+      "loss": 0.7039,
+      "step": 131
+    },
+    {
+      "epoch": 0.33475986685687115,
+      "grad_norm": 3.6972148418426514,
+      "learning_rate": 1.9992620337466976e-05,
+      "loss": 0.7223,
+      "step": 132
+    },
+    {
+      "epoch": 0.3372959264542717,
+      "grad_norm": 3.4242022037506104,
+      "learning_rate": 1.999144150632849e-05,
+      "loss": 0.7061,
+      "step": 133
+    },
+    {
+      "epoch": 0.3398319860516722,
+      "grad_norm": 3.457040309906006,
+      "learning_rate": 1.9990175405978835e-05,
+      "loss": 0.7321,
+      "step": 134
+    },
+    {
+      "epoch": 0.34236804564907275,
+      "grad_norm": 3.4427542686462402,
+      "learning_rate": 1.9988822047476618e-05,
+      "loss": 0.7151,
+      "step": 135
+    },
+    {
+      "epoch": 0.3449041052464733,
+      "grad_norm": 3.4381825923919678,
+      "learning_rate": 1.998738144264261e-05,
+      "loss": 0.7257,
+      "step": 136
+    },
+    {
+      "epoch": 0.34744016484387386,
+      "grad_norm": 3.6148738861083984,
+      "learning_rate": 1.998585360405963e-05,
+      "loss": 0.7283,
+      "step": 137
+    },
+    {
+      "epoch": 0.34997622444127435,
+      "grad_norm": 3.6557974815368652,
+      "learning_rate": 1.9984238545072422e-05,
+      "loss": 0.7194,
+      "step": 138
+    },
+    {
+      "epoch": 0.3525122840386749,
+      "grad_norm": 3.391035795211792,
+      "learning_rate": 1.9982536279787558e-05,
+      "loss": 0.7302,
+      "step": 139
+    },
+    {
+      "epoch": 0.35504834363607546,
+      "grad_norm": 3.433739185333252,
+      "learning_rate": 1.9980746823073288e-05,
+      "loss": 0.7463,
+      "step": 140
+    },
+    {
+      "epoch": 0.357584403233476,
+      "grad_norm": 3.35868501663208,
+      "learning_rate": 1.9978870190559442e-05,
+      "loss": 0.7184,
+      "step": 141
+    },
+    {
+      "epoch": 0.3601204628308765,
+      "grad_norm": 3.231398105621338,
+      "learning_rate": 1.9976906398637274e-05,
+      "loss": 0.7122,
+      "step": 142
+    },
+    {
+      "epoch": 0.36265652242827706,
+      "grad_norm": 3.266012191772461,
+      "learning_rate": 1.9974855464459317e-05,
+      "loss": 0.716,
+      "step": 143
+    },
+    {
+      "epoch": 0.3651925820256776,
+      "grad_norm": 3.420013666152954,
+      "learning_rate": 1.9972717405939245e-05,
+      "loss": 0.7114,
+      "step": 144
+    },
+    {
+      "epoch": 0.36772864162307817,
+      "grad_norm": 3.199929714202881,
+      "learning_rate": 1.997049224175171e-05,
+      "loss": 0.7473,
+      "step": 145
+    },
+    {
+      "epoch": 0.37026470122047866,
+      "grad_norm": 3.529853343963623,
+      "learning_rate": 1.9968179991332175e-05,
+      "loss": 0.716,
+      "step": 146
+    },
+    {
+      "epoch": 0.3728007608178792,
+      "grad_norm": 8.057876586914062,
+      "learning_rate": 1.9965780674876757e-05,
+      "loss": 0.7137,
+      "step": 147
+    },
+    {
+      "epoch": 0.37533682041527977,
+      "grad_norm": 8.219863891601562,
+      "learning_rate": 1.996329431334203e-05,
+      "loss": 0.7232,
+      "step": 148
+    },
+    {
+      "epoch": 0.3778728800126803,
+      "grad_norm": 6.105635166168213,
+      "learning_rate": 1.9960720928444867e-05,
+      "loss": 0.7315,
+      "step": 149
+    },
+    {
+      "epoch": 0.3804089396100808,
+      "grad_norm": 7.312891006469727,
+      "learning_rate": 1.995806054266223e-05,
+      "loss": 0.7213,
+      "step": 150
+    },
+    {
+      "epoch": 0.3804089396100808,
+      "eval_loss": NaN,
+      "eval_runtime": 207.0038,
+      "eval_samples_per_second": 40.168,
+      "eval_steps_per_second": 5.024,
+      "step": 150
+    },
+    {
+      "epoch": 0.3829449992074814,
+      "grad_norm": 6.768436908721924,
+      "learning_rate": 1.9955313179230983e-05,
+      "loss": 0.7335,
+      "step": 151
+    },
+    {
+      "epoch": 0.3854810588048819,
+      "grad_norm": 4.340691566467285,
+      "learning_rate": 1.9952478862147687e-05,
+      "loss": 0.727,
+      "step": 152
+    },
+    {
+      "epoch": 0.3880171184022825,
+      "grad_norm": 5.009334564208984,
+      "learning_rate": 1.9949557616168393e-05,
+      "loss": 0.7275,
+      "step": 153
+    },
+    {
+      "epoch": 0.390553177999683,
+      "grad_norm": 4.268547058105469,
+      "learning_rate": 1.994654946680842e-05,
+      "loss": 0.7416,
+      "step": 154
+    },
+    {
+      "epoch": 0.39308923759708353,
+      "grad_norm": 3.9493772983551025,
+      "learning_rate": 1.9943454440342137e-05,
+      "loss": 0.7286,
+      "step": 155
+    },
+    {
+      "epoch": 0.3956252971944841,
+      "grad_norm": 4.0655975341796875,
+      "learning_rate": 1.994027256380273e-05,
+      "loss": 0.7202,
+      "step": 156
+    },
+    {
+      "epoch": 0.39816135679188464,
+      "grad_norm": 3.8038692474365234,
+      "learning_rate": 1.993700386498198e-05,
+      "loss": 0.7473,
+      "step": 157
+    },
+    {
+      "epoch": 0.40069741638928513,
+      "grad_norm": 4.228780269622803,
+      "learning_rate": 1.9933648372429985e-05,
+      "loss": 0.7234,
+      "step": 158
+    },
+    {
+      "epoch": 0.4032334759866857,
+      "grad_norm": 3.7967047691345215,
+      "learning_rate": 1.9930206115454952e-05,
+      "loss": 0.7332,
+      "step": 159
+    },
+    {
+      "epoch": 0.40576953558408624,
+      "grad_norm": 3.569568395614624,
+      "learning_rate": 1.9926677124122925e-05,
+      "loss": 0.7306,
+      "step": 160
+    },
+    {
+      "epoch": 0.4083055951814868,
+      "grad_norm": 4.667778968811035,
+      "learning_rate": 1.9923061429257503e-05,
+      "loss": 0.7162,
+      "step": 161
+    },
+    {
+      "epoch": 0.4108416547788873,
+      "grad_norm": 3.978611946105957,
+      "learning_rate": 1.9919359062439602e-05,
+      "loss": 0.7055,
+      "step": 162
+    },
+    {
+      "epoch": 0.41337771437628784,
+      "grad_norm": 3.470977306365967,
+      "learning_rate": 1.9915570056007168e-05,
+      "loss": 0.7193,
+      "step": 163
+    },
+    {
+      "epoch": 0.4159137739736884,
+      "grad_norm": 3.408731698989868,
+      "learning_rate": 1.9911694443054876e-05,
+      "loss": 0.7363,
+      "step": 164
+    },
+    {
+      "epoch": 0.41844983357108895,
+      "grad_norm": 3.778426170349121,
+      "learning_rate": 1.990773225743387e-05,
+      "loss": 0.7612,
+      "step": 165
+    },
+    {
+      "epoch": 0.42098589316848944,
+      "grad_norm": 3.317411184310913,
+      "learning_rate": 1.990368353375145e-05,
+      "loss": 0.723,
+      "step": 166
+    },
+    {
+      "epoch": 0.42352195276589,
+      "grad_norm": 3.471238374710083,
+      "learning_rate": 1.9899548307370777e-05,
+      "loss": 0.7146,
+      "step": 167
+    },
+    {
+      "epoch": 0.42605801236329055,
+      "grad_norm": 9.388542175292969,
+      "learning_rate": 1.9895326614410548e-05,
+      "loss": 0.7356,
+      "step": 168
+    },
+    {
+      "epoch": 0.4285940719606911,
+      "grad_norm": 6.8452067375183105,
+      "learning_rate": 1.9891018491744712e-05,
+      "loss": 0.7529,
+      "step": 169
+    },
+    {
+      "epoch": 0.4311301315580916,
+      "grad_norm": 9.223915100097656,
+      "learning_rate": 1.9886623977002116e-05,
+      "loss": 0.7511,
+      "step": 170
+    },
+    {
+      "epoch": 0.43366619115549215,
+      "grad_norm": 8.058502197265625,
+      "learning_rate": 1.98821431085662e-05,
+      "loss": 0.7249,
+      "step": 171
+    },
+    {
+      "epoch": 0.4362022507528927,
+      "grad_norm": 5.087131977081299,
+      "learning_rate": 1.987757592557464e-05,
+      "loss": 0.7254,
+      "step": 172
+    },
+    {
+      "epoch": 0.43873831035029326,
+      "grad_norm": 5.34420108795166,
+      "learning_rate": 1.987292246791902e-05,
+      "loss": 0.73,
+      "step": 173
+    },
+    {
+      "epoch": 0.44127436994769376,
+      "grad_norm": 5.787833213806152,
+      "learning_rate": 1.9868182776244488e-05,
+      "loss": 0.7303,
+      "step": 174
+    },
+    {
+      "epoch": 0.4438104295450943,
+      "grad_norm": 4.302624702453613,
+      "learning_rate": 1.9863356891949393e-05,
+      "loss": 0.7229,
+      "step": 175
+    },
+    {
+      "epoch": 0.44634648914249486,
+      "grad_norm": 4.179818630218506,
+      "learning_rate": 1.9858444857184918e-05,
+      "loss": 0.7378,
+      "step": 176
+    },
+    {
+      "epoch": 0.4488825487398954,
+      "grad_norm": 4.3689188957214355,
+      "learning_rate": 1.985344671485472e-05,
+      "loss": 0.7166,
+      "step": 177
+    },
+    {
+      "epoch": 0.4514186083372959,
+      "grad_norm": 3.953857421875,
+      "learning_rate": 1.984836250861456e-05,
+      "loss": 0.7445,
+      "step": 178
+    },
+    {
+      "epoch": 0.45395466793469647,
+      "grad_norm": 3.9543066024780273,
+      "learning_rate": 1.984319228287191e-05,
+      "loss": 0.7399,
+      "step": 179
+    },
+    {
+      "epoch": 0.456490727532097,
+      "grad_norm": 3.5704877376556396,
+      "learning_rate": 1.983793608278557e-05,
+      "loss": 0.7244,
+      "step": 180
+    },
+    {
+      "epoch": 0.4590267871294975,
+      "grad_norm": 3.776400566101074,
+      "learning_rate": 1.983259395426527e-05,
+      "loss": 0.7462,
+      "step": 181
+    },
+    {
+      "epoch": 0.46156284672689807,
+      "grad_norm": 3.885683059692383,
+      "learning_rate": 1.982716594397129e-05,
+      "loss": 0.7526,
+      "step": 182
+    },
+    {
+      "epoch": 0.4640989063242986,
+      "grad_norm": 3.5768251419067383,
+      "learning_rate": 1.9821652099314016e-05,
+      "loss": 0.7364,
+      "step": 183
+    },
+    {
+      "epoch": 0.4666349659216992,
+      "grad_norm": 3.583132266998291,
+      "learning_rate": 1.9816052468453554e-05,
+      "loss": 0.7545,
+      "step": 184
+    },
+    {
+      "epoch": 0.46917102551909967,
+      "grad_norm": 3.498037576675415,
+      "learning_rate": 1.98103671002993e-05,
+      "loss": 0.7522,
+      "step": 185
+    },
+    {
+      "epoch": 0.4717070851165002,
+      "grad_norm": 3.925713062286377,
+      "learning_rate": 1.9804596044509517e-05,
+      "loss": 0.7477,
+      "step": 186
+    },
+    {
+      "epoch": 0.4742431447139008,
+      "grad_norm": 3.772189140319824,
+      "learning_rate": 1.979873935149089e-05,
+      "loss": 0.7674,
+      "step": 187
+    },
+    {
+      "epoch": 0.47677920431130133,
+      "grad_norm": 3.5768580436706543,
+      "learning_rate": 1.97927970723981e-05,
+      "loss": 0.7376,
+      "step": 188
+    },
+    {
+      "epoch": 0.4793152639087018,
+      "grad_norm": 4.358255863189697,
+      "learning_rate": 1.978676925913337e-05,
+      "loss": 0.7443,
+      "step": 189
+    },
+    {
+      "epoch": 0.4818513235061024,
+      "grad_norm": 3.811522960662842,
+      "learning_rate": 1.9780655964346006e-05,
+      "loss": 0.7474,
+      "step": 190
+    },
+    {
+      "epoch": 0.48438738310350293,
+      "grad_norm": 7.138657569885254,
+      "learning_rate": 1.9774457241431947e-05,
+      "loss": 0.7355,
+      "step": 191
+    },
+    {
+      "epoch": 0.4869234427009035,
+      "grad_norm": 5.5900092124938965,
+      "learning_rate": 1.9768173144533305e-05,
+      "loss": 0.7293,
+      "step": 192
+    },
+    {
+      "epoch": 0.489459502298304,
+      "grad_norm": 4.383650302886963,
+      "learning_rate": 1.976180372853787e-05,
+      "loss": 0.7401,
+      "step": 193
+    },
+    {
+      "epoch": 0.49199556189570454,
+      "grad_norm": 4.673737525939941,
+      "learning_rate": 1.9755349049078642e-05,
+      "loss": 0.7298,
+      "step": 194
+    },
+    {
+      "epoch": 0.4945316214931051,
+      "grad_norm": 4.192466735839844,
+      "learning_rate": 1.974880916253335e-05,
+      "loss": 0.7367,
+      "step": 195
+    },
+    {
+      "epoch": 0.49706768109050564,
+      "grad_norm": 4.307677745819092,
+      "learning_rate": 1.9742184126023962e-05,
+      "loss": 0.7567,
+      "step": 196
+    },
+    {
+      "epoch": 0.49960374068790614,
+      "grad_norm": 4.165250778198242,
+      "learning_rate": 1.9735473997416173e-05,
+      "loss": 0.7409,
+      "step": 197
+    },
+    {
+      "epoch": 0.5021398002853067,
+      "grad_norm": 3.3816635608673096,
+      "learning_rate": 1.97286788353189e-05,
+      "loss": 0.7259,
+      "step": 198
+    },
+    {
+      "epoch": 0.5046758598827072,
+      "grad_norm": 3.783017635345459,
+      "learning_rate": 1.972179869908379e-05,
+      "loss": 0.7366,
+      "step": 199
+    },
+    {
+      "epoch": 0.5072119194801078,
+      "grad_norm": 3.870734691619873,
+      "learning_rate": 1.9714833648804676e-05,
+      "loss": 0.7463,
+      "step": 200
+    },
+    {
+      "epoch": 0.5072119194801078,
+      "eval_loss": NaN,
+      "eval_runtime": 206.6586,
+      "eval_samples_per_second": 40.235,
+      "eval_steps_per_second": 5.032,
+      "step": 200
+    },
+    {
+      "epoch": 0.5097479790775084,
+      "grad_norm": 3.7613582611083984,
+      "learning_rate": 1.9707783745317065e-05,
+      "loss": 0.7292,
+      "step": 201
+    },
+    {
+      "epoch": 0.5122840386749089,
+      "grad_norm": 3.52028226852417,
+      "learning_rate": 1.970064905019762e-05,
+      "loss": 0.7413,
+      "step": 202
+    },
+    {
+      "epoch": 0.5148200982723093,
+      "grad_norm": 3.310667037963867,
+      "learning_rate": 1.9693429625763584e-05,
+      "loss": 0.7551,
+      "step": 203
+    },
+    {
+      "epoch": 0.5173561578697099,
+      "grad_norm": 3.2779347896575928,
+      "learning_rate": 1.9686125535072284e-05,
+      "loss": 0.7584,
+      "step": 204
+    },
+    {
+      "epoch": 0.5198922174671105,
+      "grad_norm": 3.358673334121704,
+      "learning_rate": 1.9678736841920535e-05,
+      "loss": 0.7216,
+      "step": 205
+    },
+    {
+      "epoch": 0.522428277064511,
+      "grad_norm": 3.3773913383483887,
+      "learning_rate": 1.9671263610844117e-05,
+      "loss": 0.743,
+      "step": 206
+    },
+    {
+      "epoch": 0.5249643366619116,
+      "grad_norm": 3.1578938961029053,
+      "learning_rate": 1.966370590711719e-05,
+      "loss": 0.7238,
+      "step": 207
+    },
+    {
+      "epoch": 0.5275003962593121,
+      "grad_norm": 4.021521091461182,
+      "learning_rate": 1.9656063796751735e-05,
+      "loss": 0.745,
+      "step": 208
+    },
+    {
+      "epoch": 0.5300364558567127,
+      "grad_norm": 3.6202096939086914,
+      "learning_rate": 1.964833734649697e-05,
+      "loss": 0.7434,
+      "step": 209
+    },
+    {
+      "epoch": 0.5325725154541132,
+      "grad_norm": 4.109938621520996,
+      "learning_rate": 1.9640526623838787e-05,
+      "loss": 0.7264,
+      "step": 210
+    },
+    {
+      "epoch": 0.5351085750515137,
+      "grad_norm": 4.244260787963867,
+      "learning_rate": 1.9632631696999123e-05,
+      "loss": 0.753,
+      "step": 211
+    },
+    {
+      "epoch": 0.5376446346489142,
+      "grad_norm": 3.5682835578918457,
+      "learning_rate": 1.9624652634935403e-05,
+      "loss": 0.7294,
+      "step": 212
+    },
+    {
+      "epoch": 0.5401806942463148,
+      "grad_norm": 4.292395114898682,
+      "learning_rate": 1.9616589507339916e-05,
+      "loss": 0.7241,
+      "step": 213
+    },
+    {
+      "epoch": 0.5427167538437153,
+      "grad_norm": 4.509610652923584,
+      "learning_rate": 1.9608442384639218e-05,
+      "loss": 0.7241,
+      "step": 214
+    },
+    {
+      "epoch": 0.5452528134411159,
+      "grad_norm": 4.151599884033203,
+      "learning_rate": 1.9600211337993508e-05,
+      "loss": 0.729,
+      "step": 215
+    },
+    {
+      "epoch": 0.5477888730385164,
+      "grad_norm": 3.8853487968444824,
+      "learning_rate": 1.9591896439296007e-05,
+      "loss": 0.7141,
+      "step": 216
+    },
+    {
+      "epoch": 0.550324932635917,
+      "grad_norm": 3.62223219871521,
+      "learning_rate": 1.958349776117234e-05,
+      "loss": 0.7504,
+      "step": 217
+    },
+    {
+      "epoch": 0.5528609922333175,
+      "grad_norm": 3.9218106269836426,
+      "learning_rate": 1.957501537697989e-05,
+      "loss": 0.7484,
+      "step": 218
+    },
+    {
+      "epoch": 0.555397051830718,
+      "grad_norm": 3.451983690261841,
+      "learning_rate": 1.9566449360807166e-05,
+      "loss": 0.7422,
+      "step": 219
+    },
+    {
+      "epoch": 0.5579331114281185,
+      "grad_norm": 3.7707138061523438,
+      "learning_rate": 1.9557799787473146e-05,
+      "loss": 0.7263,
+      "step": 220
+    },
+    {
+      "epoch": 0.5604691710255191,
+      "grad_norm": 3.6123878955841064,
+      "learning_rate": 1.9549066732526635e-05,
+      "loss": 0.7369,
+      "step": 221
+    },
+    {
+      "epoch": 0.5630052306229196,
+      "grad_norm": 3.314817428588867,
+      "learning_rate": 1.9540250272245595e-05,
+      "loss": 0.7259,
+      "step": 222
+    },
+    {
+      "epoch": 0.5655412902203202,
+      "grad_norm": 3.369518995285034,
+      "learning_rate": 1.9531350483636484e-05,
+      "loss": 0.7534,
+      "step": 223
+    },
+    {
+      "epoch": 0.5680773498177207,
+      "grad_norm": 3.4133052825927734,
+      "learning_rate": 1.9522367444433587e-05,
+      "loss": 0.7787,
+      "step": 224
+    },
+    {
+      "epoch": 0.5706134094151213,
+      "grad_norm": 3.40303373336792,
+      "learning_rate": 1.9513301233098332e-05,
+      "loss": 0.733,
+      "step": 225
+    },
+    {
+      "epoch": 0.5731494690125218,
+      "grad_norm": 3.5303635597229004,
+      "learning_rate": 1.9504151928818594e-05,
+      "loss": 0.7262,
+      "step": 226
+    },
+    {
+      "epoch": 0.5756855286099223,
+      "grad_norm": 3.4674553871154785,
+      "learning_rate": 1.9494919611508037e-05,
+      "loss": 0.7331,
+      "step": 227
+    },
+    {
+      "epoch": 0.5782215882073228,
+      "grad_norm": 3.3834900856018066,
+      "learning_rate": 1.948560436180537e-05,
+      "loss": 0.7316,
+      "step": 228
+    },
+    {
+      "epoch": 0.5807576478047234,
+      "grad_norm": 3.387651205062866,
+      "learning_rate": 1.9476206261073675e-05,
+      "loss": 0.7588,
+      "step": 229
+    },
+    {
+      "epoch": 0.5832937074021239,
+      "grad_norm": 3.5845932960510254,
+      "learning_rate": 1.94667253913997e-05,
+      "loss": 0.7528,
+      "step": 230
+    },
+    {
+      "epoch": 0.5858297669995245,
+      "grad_norm": 5.973608016967773,
+      "learning_rate": 1.9457161835593114e-05,
+      "loss": 0.743,
+      "step": 231
+    },
+    {
+      "epoch": 0.588365826596925,
+      "grad_norm": 5.038896560668945,
+      "learning_rate": 1.9447515677185803e-05,
+      "loss": 0.7554,
+      "step": 232
+    },
+    {
+      "epoch": 0.5909018861943256,
+      "grad_norm": 4.64965295791626,
+      "learning_rate": 1.9437787000431143e-05,
+      "loss": 0.7585,
+      "step": 233
+    },
+    {
+      "epoch": 0.5934379457917262,
+      "grad_norm": 4.600735664367676,
+      "learning_rate": 1.9427975890303252e-05,
+      "loss": 0.7324,
+      "step": 234
+    },
+    {
+      "epoch": 0.5959740053891266,
+      "grad_norm": 4.252320766448975,
+      "learning_rate": 1.9418082432496253e-05,
+      "loss": 0.7383,
+      "step": 235
+    },
+    {
+      "epoch": 0.5985100649865271,
+      "grad_norm": 4.28120756149292,
+      "learning_rate": 1.940810671342353e-05,
+      "loss": 0.7463,
+      "step": 236
+    },
+    {
+      "epoch": 0.6010461245839277,
+      "grad_norm": 3.9514646530151367,
+      "learning_rate": 1.939804882021697e-05,
+      "loss": 0.7576,
+      "step": 237
+    },
+    {
+      "epoch": 0.6035821841813283,
+      "grad_norm": 3.741912603378296,
+      "learning_rate": 1.9387908840726197e-05,
+      "loss": 0.7651,
+      "step": 238
+    },
+    {
+      "epoch": 0.6061182437787288,
+      "grad_norm": 3.95212721824646,
+      "learning_rate": 1.9377686863517812e-05,
+      "loss": 0.7601,
+      "step": 239
+    },
+    {
+      "epoch": 0.6086543033761294,
+      "grad_norm": 3.4727978706359863,
+      "learning_rate": 1.936738297787462e-05,
+      "loss": 0.7363,
+      "step": 240
+    },
+    {
+      "epoch": 0.6111903629735299,
+      "grad_norm": 3.4831690788269043,
+      "learning_rate": 1.935699727379484e-05,
+      "loss": 0.7781,
+      "step": 241
+    },
+    {
+      "epoch": 0.6137264225709305,
+      "grad_norm": 3.496211528778076,
+      "learning_rate": 1.9346529841991327e-05,
+      "loss": 0.7633,
+      "step": 242
+    },
+    {
+      "epoch": 0.6162624821683309,
+      "grad_norm": 3.73543119430542,
+      "learning_rate": 1.933598077389079e-05,
+      "loss": 0.7393,
+      "step": 243
+    },
+    {
+      "epoch": 0.6187985417657315,
+      "grad_norm": 3.5055994987487793,
+      "learning_rate": 1.9325350161632958e-05,
+      "loss": 0.7476,
+      "step": 244
+    },
+    {
+      "epoch": 0.621334601363132,
+      "grad_norm": 3.2391014099121094,
+      "learning_rate": 1.931463809806982e-05,
+      "loss": 0.7485,
+      "step": 245
+    },
+    {
+      "epoch": 0.6238706609605326,
+      "grad_norm": 3.499049186706543,
+      "learning_rate": 1.9303844676764783e-05,
+      "loss": 0.7472,
+      "step": 246
+    },
+    {
+      "epoch": 0.6264067205579331,
+      "grad_norm": 3.155977249145508,
+      "learning_rate": 1.9292969991991874e-05,
+      "loss": 0.7523,
+      "step": 247
+    },
+    {
+      "epoch": 0.6289427801553337,
+      "grad_norm": 3.4776058197021484,
+      "learning_rate": 1.92820141387349e-05,
+      "loss": 0.7319,
+      "step": 248
+    },
+    {
+      "epoch": 0.6314788397527342,
+      "grad_norm": 3.1562836170196533,
+      "learning_rate": 1.9270977212686618e-05,
+      "loss": 0.755,
+      "step": 249
+    },
+    {
+      "epoch": 0.6340148993501348,
+      "grad_norm": 3.3259477615356445,
+      "learning_rate": 1.9259859310247924e-05,
+      "loss": 0.7549,
+      "step": 250
+    },
+    {
+      "epoch": 0.6340148993501348,
+      "eval_loss": NaN,
+      "eval_runtime": 206.3529,
+      "eval_samples_per_second": 40.295,
+      "eval_steps_per_second": 5.04,
+      "step": 250
+    },
+    {
+      "epoch": 0.6365509589475352,
+      "grad_norm": 4.16396951675415,
+      "learning_rate": 1.9248660528526985e-05,
+      "loss": 0.7479,
+      "step": 251
+    },
+    {
+      "epoch": 0.6390870185449358,
+      "grad_norm": 5.292008876800537,
+      "learning_rate": 1.9237380965338393e-05,
+      "loss": 0.7458,
+      "step": 252
+    },
+    {
+      "epoch": 0.6416230781423363,
+      "grad_norm": 3.914210319519043,
+      "learning_rate": 1.9226020719202333e-05,
+      "loss": 0.7427,
+      "step": 253
+    },
+    {
+      "epoch": 0.6441591377397369,
+      "grad_norm": 4.041393756866455,
+      "learning_rate": 1.92145798893437e-05,
+      "loss": 0.7466,
+      "step": 254
+    },
+    {
+      "epoch": 0.6466951973371374,
+      "grad_norm": 3.9303200244903564,
+      "learning_rate": 1.9203058575691226e-05,
+      "loss": 0.7726,
+      "step": 255
+    },
+    {
+      "epoch": 0.649231256934538,
+      "grad_norm": 4.203293800354004,
+      "learning_rate": 1.9191456878876643e-05,
+      "loss": 0.7301,
+      "step": 256
+    },
+    {
+      "epoch": 0.6517673165319385,
+      "grad_norm": 3.7889516353607178,
+      "learning_rate": 1.9179774900233763e-05,
+      "loss": 0.7518,
+      "step": 257
+    },
+    {
+      "epoch": 0.6543033761293391,
+      "grad_norm": 4.092428207397461,
+      "learning_rate": 1.9168012741797624e-05,
+      "loss": 0.7376,
+      "step": 258
+    },
+    {
+      "epoch": 0.6568394357267395,
+      "grad_norm": 3.5720880031585693,
+      "learning_rate": 1.915617050630358e-05,
+      "loss": 0.7455,
+      "step": 259
+    },
+    {
+      "epoch": 0.6593754953241401,
+      "grad_norm": 3.904362440109253,
+      "learning_rate": 1.914424829718641e-05,
+      "loss": 0.7415,
+      "step": 260
+    },
+    {
+      "epoch": 0.6619115549215406,
+      "grad_norm": 3.481815814971924,
+      "learning_rate": 1.9132246218579412e-05,
+      "loss": 0.7463,
+      "step": 261
+    },
+    {
+      "epoch": 0.6644476145189412,
+      "grad_norm": 3.5754215717315674,
+      "learning_rate": 1.9120164375313504e-05,
+      "loss": 0.7426,
+      "step": 262
+    },
+    {
+      "epoch": 0.6669836741163417,
+      "grad_norm": 3.496607542037964,
+      "learning_rate": 1.9108002872916288e-05,
+      "loss": 0.7357,
+      "step": 263
+    },
+    {
+      "epoch": 0.6695197337137423,
+      "grad_norm": 3.690295457839966,
+      "learning_rate": 1.9095761817611157e-05,
+      "loss": 0.7469,
+      "step": 264
+    },
+    {
+      "epoch": 0.6720557933111428,
+      "grad_norm": 3.593576192855835,
+      "learning_rate": 1.9083441316316333e-05,
+      "loss": 0.7475,
+      "step": 265
+    },
+    {
+      "epoch": 0.6745918529085434,
+      "grad_norm": 3.6301517486572266,
+      "learning_rate": 1.9071041476643962e-05,
+      "loss": 0.7527,
+      "step": 266
+    },
+    {
+      "epoch": 0.6771279125059438,
+      "grad_norm": 3.5619664192199707,
+      "learning_rate": 1.9058562406899158e-05,
+      "loss": 0.7439,
+      "step": 267
+    },
+    {
+      "epoch": 0.6796639721033444,
+      "grad_norm": 3.7166190147399902,
+      "learning_rate": 1.9046004216079065e-05,
+      "loss": 0.727,
+      "step": 268
+    },
+    {
+      "epoch": 0.682200031700745,
+      "grad_norm": 3.420332670211792,
+      "learning_rate": 1.90333670138719e-05,
+      "loss": 0.7651,
+      "step": 269
+    },
+    {
+      "epoch": 0.6847360912981455,
+      "grad_norm": 3.198678970336914,
+      "learning_rate": 1.9020650910655993e-05,
+      "loss": 0.7584,
+      "step": 270
+    },
+    {
+      "epoch": 0.687272150895546,
+      "grad_norm": 3.469536066055298,
+      "learning_rate": 1.9007856017498834e-05,
+      "loss": 0.758,
+      "step": 271
+    },
+    {
+      "epoch": 0.6898082104929466,
+      "grad_norm": 3.1529855728149414,
+      "learning_rate": 1.8994982446156097e-05,
+      "loss": 0.7498,
+      "step": 272
+    },
+    {
+      "epoch": 0.6923442700903472,
+      "grad_norm": 6.391468524932861,
+      "learning_rate": 1.898203030907065e-05,
+      "loss": 0.7253,
+      "step": 273
+    },
+    {
+      "epoch": 0.6948803296877477,
+      "grad_norm": 4.930320739746094,
+      "learning_rate": 1.89689997193716e-05,
+      "loss": 0.7695,
+      "step": 274
+    },
+    {
+      "epoch": 0.6974163892851482,
+      "grad_norm": 5.144972801208496,
+      "learning_rate": 1.8955890790873276e-05,
+      "loss": 0.7513,
+      "step": 275
+    },
+    {
+      "epoch": 0.6999524488825487,
+      "grad_norm": 4.549212455749512,
+      "learning_rate": 1.894270363807426e-05,
+      "loss": 0.7429,
+      "step": 276
+    },
+    {
+      "epoch": 0.7024885084799493,
+      "grad_norm": 4.320601463317871,
+      "learning_rate": 1.8929438376156377e-05,
+      "loss": 0.7601,
+      "step": 277
+    },
+    {
+      "epoch": 0.7050245680773498,
+      "grad_norm": 4.369263648986816,
+      "learning_rate": 1.8916095120983677e-05,
+      "loss": 0.7608,
+      "step": 278
+    },
+    {
+      "epoch": 0.7075606276747504,
+      "grad_norm": 4.001751899719238,
+      "learning_rate": 1.8902673989101444e-05,
+      "loss": 0.7721,
+      "step": 279
+    },
+    {
+      "epoch": 0.7100966872721509,
+      "grad_norm": 3.7000808715820312,
+      "learning_rate": 1.888917509773516e-05,
+      "loss": 0.7387,
+      "step": 280
+    },
+    {
+      "epoch": 0.7126327468695515,
+      "grad_norm": 3.694392204284668,
+      "learning_rate": 1.88755985647895e-05,
+      "loss": 0.7516,
+      "step": 281
+    },
+    {
+      "epoch": 0.715168806466952,
+      "grad_norm": 3.418475866317749,
+      "learning_rate": 1.886194450884728e-05,
+      "loss": 0.7523,
+      "step": 282
+    },
+    {
+      "epoch": 0.7177048660643525,
+      "grad_norm": 3.687631845474243,
+      "learning_rate": 1.8848213049168446e-05,
+      "loss": 0.7668,
+      "step": 283
+    },
+    {
+      "epoch": 0.720240925661753,
+      "grad_norm": 3.446545362472534,
+      "learning_rate": 1.8834404305689e-05,
+      "loss": 0.7864,
+      "step": 284
+    },
+    {
+      "epoch": 0.7227769852591536,
+      "grad_norm": 4.012823104858398,
+      "learning_rate": 1.8820518399019986e-05,
+      "loss": 0.7638,
+      "step": 285
+    },
+    {
+      "epoch": 0.7253130448565541,
+      "grad_norm": 3.6890742778778076,
+      "learning_rate": 1.880655545044642e-05,
+      "loss": 0.774,
+      "step": 286
+    },
+    {
+      "epoch": 0.7278491044539547,
+      "grad_norm": 3.477963447570801,
+      "learning_rate": 1.879251558192623e-05,
+      "loss": 0.7448,
+      "step": 287
+    },
+    {
+      "epoch": 0.7303851640513552,
+      "grad_norm": 3.656797170639038,
+      "learning_rate": 1.877839891608919e-05,
+      "loss": 0.7502,
+      "step": 288
+    },
+    {
+      "epoch": 0.7329212236487558,
+      "grad_norm": 3.661099672317505,
+      "learning_rate": 1.876420557623586e-05,
+      "loss": 0.7582,
+      "step": 289
+    },
+    {
+      "epoch": 0.7354572832461563,
+      "grad_norm": 3.435574531555176,
+      "learning_rate": 1.8749935686336496e-05,
+      "loss": 0.7684,
+      "step": 290
+    },
+    {
+      "epoch": 0.7379933428435568,
+      "grad_norm": 3.4438607692718506,
+      "learning_rate": 1.8735589371029973e-05,
+      "loss": 0.7526,
+      "step": 291
+    },
+    {
+      "epoch": 0.7405294024409573,
+      "grad_norm": 3.577694892883301,
+      "learning_rate": 1.8721166755622696e-05,
+      "loss": 0.7441,
+      "step": 292
+    },
+    {
+      "epoch": 0.7430654620383579,
+      "grad_norm": 3.602250337600708,
+      "learning_rate": 1.8706667966087505e-05,
+      "loss": 0.7572,
+      "step": 293
+    },
+    {
+      "epoch": 0.7456015216357584,
+      "grad_norm": 3.9699854850769043,
+      "learning_rate": 1.869209312906258e-05,
+      "loss": 0.7728,
+      "step": 294
+    },
+    {
+      "epoch": 0.748137581233159,
+      "grad_norm": 3.718890428543091,
+      "learning_rate": 1.8677442371850323e-05,
+      "loss": 0.7555,
+      "step": 295
+    },
+    {
+      "epoch": 0.7506736408305595,
+      "grad_norm": 3.8095226287841797,
+      "learning_rate": 1.8662715822416257e-05,
+      "loss": 0.7572,
+      "step": 296
+    },
+    {
+      "epoch": 0.7532097004279601,
+      "grad_norm": 4.708703517913818,
+      "learning_rate": 1.8647913609387897e-05,
+      "loss": 0.7373,
+      "step": 297
+    },
+    {
+      "epoch": 0.7557457600253606,
+      "grad_norm": 4.488431453704834,
+      "learning_rate": 1.863303586205365e-05,
+      "loss": 0.7804,
+      "step": 298
+    },
+    {
+      "epoch": 0.7582818196227611,
+      "grad_norm": 4.450040817260742,
+      "learning_rate": 1.8618082710361648e-05,
+      "loss": 0.7603,
+      "step": 299
+    },
+    {
+      "epoch": 0.7608178792201616,
+      "grad_norm": 4.4052734375,
+      "learning_rate": 1.8603054284918657e-05,
+      "loss": 0.7564,
+      "step": 300
+    },
+    {
+      "epoch": 0.7608178792201616,
+      "eval_loss": NaN,
+      "eval_runtime": 206.8311,
+      "eval_samples_per_second": 40.202,
+      "eval_steps_per_second": 5.028,
+      "step": 300
+    },
+    {
+      "epoch": 0.7633539388175622,
+      "grad_norm": 4.103545665740967,
+      "learning_rate": 1.8587950716988892e-05,
+      "loss": 0.7869,
+      "step": 301
+    },
+    {
+      "epoch": 0.7658899984149627,
+      "grad_norm": 3.7855234146118164,
+      "learning_rate": 1.8572772138492915e-05,
+      "loss": 0.7823,
+      "step": 302
+    },
+    {
+      "epoch": 0.7684260580123633,
+      "grad_norm": 3.637251615524292,
+      "learning_rate": 1.8557518682006442e-05,
+      "loss": 0.7565,
+      "step": 303
+    },
+    {
+      "epoch": 0.7709621176097639,
+      "grad_norm": 3.495508909225464,
+      "learning_rate": 1.8542190480759207e-05,
+      "loss": 0.7562,
+      "step": 304
+    },
+    {
+      "epoch": 0.7734981772071644,
+      "grad_norm": 3.9170336723327637,
+      "learning_rate": 1.8526787668633797e-05,
+      "loss": 0.7453,
+      "step": 305
+    },
+    {
+      "epoch": 0.776034236804565,
+      "grad_norm": 3.5141119956970215,
+      "learning_rate": 1.851131038016448e-05,
+      "loss": 0.7679,
+      "step": 306
+    },
+    {
+      "epoch": 0.7785702964019654,
+      "grad_norm": 3.705007791519165,
+      "learning_rate": 1.8495758750536035e-05,
+      "loss": 0.7268,
+      "step": 307
+    },
+    {
+      "epoch": 0.781106355999366,
+      "grad_norm": 3.47196364402771,
+      "learning_rate": 1.8480132915582555e-05,
+      "loss": 0.7561,
+      "step": 308
+    },
+    {
+      "epoch": 0.7836424155967665,
+      "grad_norm": 3.6013100147247314,
+      "learning_rate": 1.8464433011786273e-05,
+      "loss": 0.7589,
+      "step": 309
+    },
+    {
+      "epoch": 0.7861784751941671,
+      "grad_norm": 3.549534320831299,
+      "learning_rate": 1.8448659176276385e-05,
+      "loss": 0.7882,
+      "step": 310
+    },
+    {
+      "epoch": 0.7887145347915676,
+      "grad_norm": 3.5890965461730957,
+      "learning_rate": 1.8432811546827812e-05,
+      "loss": 0.722,
+      "step": 311
+    },
+    {
+      "epoch": 0.7912505943889682,
+      "grad_norm": 3.7022740840911865,
+      "learning_rate": 1.8416890261860038e-05,
+      "loss": 0.7478,
+      "step": 312
+    },
+    {
+      "epoch": 0.7937866539863687,
+      "grad_norm": 4.593228816986084,
+      "learning_rate": 1.840089546043588e-05,
+      "loss": 0.7716,
+      "step": 313
+    },
+    {
+      "epoch": 0.7963227135837693,
+      "grad_norm": 4.206724166870117,
+      "learning_rate": 1.838482728226027e-05,
+      "loss": 0.7691,
+      "step": 314
+    },
+    {
+      "epoch": 0.7988587731811697,
+      "grad_norm": 4.59312629699707,
+      "learning_rate": 1.8368685867679048e-05,
+      "loss": 0.763,
+      "step": 315
+    },
+    {
+      "epoch": 0.8013948327785703,
+      "grad_norm": 4.171518325805664,
+      "learning_rate": 1.835247135767773e-05,
+      "loss": 0.7389,
+      "step": 316
+    },
+    {
+      "epoch": 0.8039308923759708,
+      "grad_norm": 3.9752421379089355,
+      "learning_rate": 1.833618389388027e-05,
+      "loss": 0.7665,
+      "step": 317
+    },
+    {
+      "epoch": 0.8064669519733714,
+      "grad_norm": 3.7727441787719727,
+      "learning_rate": 1.8319823618547836e-05,
+      "loss": 0.751,
+      "step": 318
+    },
+    {
+      "epoch": 0.8090030115707719,
+      "grad_norm": 3.8601176738739014,
+      "learning_rate": 1.8303390674577562e-05,
+      "loss": 0.7676,
+      "step": 319
+    },
+    {
+      "epoch": 0.8115390711681725,
+      "grad_norm": 3.843137502670288,
+      "learning_rate": 1.828688520550129e-05,
+      "loss": 0.7598,
+      "step": 320
+    },
+    {
+      "epoch": 0.814075130765573,
+      "grad_norm": 3.8456783294677734,
+      "learning_rate": 1.8270307355484342e-05,
+      "loss": 0.7425,
+      "step": 321
+    },
+    {
+      "epoch": 0.8166111903629736,
+      "grad_norm": 3.703662633895874,
+      "learning_rate": 1.8253657269324212e-05,
+      "loss": 0.7756,
+      "step": 322
+    },
+    {
+      "epoch": 0.819147249960374,
+      "grad_norm": 3.534630537033081,
+      "learning_rate": 1.823693509244937e-05,
+      "loss": 0.7328,
+      "step": 323
+    },
+    {
+      "epoch": 0.8216833095577746,
+      "grad_norm": 3.6060171127319336,
+      "learning_rate": 1.8220140970917934e-05,
+      "loss": 0.7482,
+      "step": 324
+    },
+    {
+      "epoch": 0.8242193691551751,
+      "grad_norm": 3.857970714569092,
+      "learning_rate": 1.820327505141642e-05,
+      "loss": 0.7442,
+      "step": 325
+    },
+    {
+      "epoch": 0.8267554287525757,
+      "grad_norm": 3.495211601257324,
+      "learning_rate": 1.8186337481258452e-05,
+      "loss": 0.7713,
+      "step": 326
+    },
+    {
+      "epoch": 0.8292914883499762,
+      "grad_norm": 3.427976369857788,
+      "learning_rate": 1.8169328408383484e-05,
+      "loss": 0.7712,
+      "step": 327
+    },
+    {
+      "epoch": 0.8318275479473768,
+      "grad_norm": 3.5397775173187256,
+      "learning_rate": 1.81522479813555e-05,
+      "loss": 0.7539,
+      "step": 328
+    },
+    {
+      "epoch": 0.8343636075447773,
+      "grad_norm": 3.4796149730682373,
+      "learning_rate": 1.8135096349361725e-05,
+      "loss": 0.7707,
+      "step": 329
+    },
+    {
+      "epoch": 0.8368996671421779,
+      "grad_norm": 3.389249086380005,
+      "learning_rate": 1.8117873662211315e-05,
+      "loss": 0.7227,
+      "step": 330
+    },
+    {
+      "epoch": 0.8394357267395783,
+      "grad_norm": 3.5652847290039062,
+      "learning_rate": 1.810058007033404e-05,
+      "loss": 0.763,
+      "step": 331
+    },
+    {
+      "epoch": 0.8419717863369789,
+      "grad_norm": 3.2756998538970947,
+      "learning_rate": 1.808321572477899e-05,
+      "loss": 0.7757,
+      "step": 332
+    },
+    {
+      "epoch": 0.8445078459343794,
+      "grad_norm": 3.320709705352783,
+      "learning_rate": 1.8065780777213242e-05,
+      "loss": 0.7588,
+      "step": 333
+    },
+    {
+      "epoch": 0.84704390553178,
+      "grad_norm": 5.3580193519592285,
+      "learning_rate": 1.8048275379920545e-05,
+      "loss": 0.7991,
+      "step": 334
+    },
+    {
+      "epoch": 0.8495799651291805,
+      "grad_norm": 4.250069618225098,
+      "learning_rate": 1.8030699685799976e-05,
+      "loss": 0.7518,
+      "step": 335
+    },
+    {
+      "epoch": 0.8521160247265811,
+      "grad_norm": 4.51885461807251,
+      "learning_rate": 1.8013053848364612e-05,
+      "loss": 0.7415,
+      "step": 336
+    },
+    {
+      "epoch": 0.8546520843239817,
+      "grad_norm": 4.98030948638916,
+      "learning_rate": 1.79953380217402e-05,
+      "loss": 0.7638,
+      "step": 337
+    },
+    {
+      "epoch": 0.8571881439213822,
+      "grad_norm": 4.827258110046387,
+      "learning_rate": 1.7977552360663787e-05,
+      "loss": 0.7626,
+      "step": 338
+    },
+    {
+      "epoch": 0.8597242035187826,
+      "grad_norm": 5.554749488830566,
+      "learning_rate": 1.7959697020482387e-05,
+      "loss": 0.7599,
+      "step": 339
+    },
+    {
+      "epoch": 0.8622602631161832,
+      "grad_norm": 4.254626274108887,
+      "learning_rate": 1.7941772157151627e-05,
+      "loss": 0.7604,
+      "step": 340
+    },
+    {
+      "epoch": 0.8647963227135838,
+      "grad_norm": 5.148410320281982,
+      "learning_rate": 1.792377792723436e-05,
+      "loss": 0.778,
+      "step": 341
+    },
+    {
+      "epoch": 0.8673323823109843,
+      "grad_norm": 4.256086826324463,
+      "learning_rate": 1.790571448789933e-05,
+      "loss": 0.746,
+      "step": 342
+    },
+    {
+      "epoch": 0.8698684419083849,
+      "grad_norm": 4.854703903198242,
+      "learning_rate": 1.7887581996919784e-05,
+      "loss": 0.751,
+      "step": 343
+    },
+    {
+      "epoch": 0.8724045015057854,
+      "grad_norm": 3.7882001399993896,
+      "learning_rate": 1.786938061267207e-05,
+      "loss": 0.7545,
+      "step": 344
+    },
+    {
+      "epoch": 0.874940561103186,
+      "grad_norm": 4.779317855834961,
+      "learning_rate": 1.785111049413431e-05,
+      "loss": 0.7638,
+      "step": 345
+    },
+    {
+      "epoch": 0.8774766207005865,
+      "grad_norm": 4.138965129852295,
+      "learning_rate": 1.783277180088495e-05,
+      "loss": 0.7691,
+      "step": 346
+    },
+    {
+      "epoch": 0.880012680297987,
+      "grad_norm": 4.164398193359375,
+      "learning_rate": 1.7814364693101414e-05,
+      "loss": 0.7743,
+      "step": 347
+    },
+    {
+      "epoch": 0.8825487398953875,
+      "grad_norm": 3.9533638954162598,
+      "learning_rate": 1.7795889331558677e-05,
+      "loss": 0.7305,
+      "step": 348
+    },
+    {
+      "epoch": 0.8850847994927881,
+      "grad_norm": 4.169763565063477,
+      "learning_rate": 1.7777345877627872e-05,
+      "loss": 0.7497,
+      "step": 349
+    },
+    {
+      "epoch": 0.8876208590901886,
+      "grad_norm": 3.5670506954193115,
+      "learning_rate": 1.7758734493274878e-05,
+      "loss": 0.7733,
+      "step": 350
+    },
+    {
+      "epoch": 0.8876208590901886,
+      "eval_loss": NaN,
+      "eval_runtime": 206.4641,
+      "eval_samples_per_second": 40.273,
+      "eval_steps_per_second": 5.037,
+      "step": 350
+    },
+    {
+      "epoch": 0.8901569186875892,
+      "grad_norm": 3.6017632484436035,
+      "learning_rate": 1.77400553410589e-05,
+      "loss": 0.738,
+      "step": 351
+    },
+    {
+      "epoch": 0.8926929782849897,
+      "grad_norm": 3.649099826812744,
+      "learning_rate": 1.7721308584131062e-05,
+      "loss": 0.7729,
+      "step": 352
+    },
+    {
+      "epoch": 0.8952290378823903,
+      "grad_norm": 3.6460583209991455,
+      "learning_rate": 1.7702494386232972e-05,
+      "loss": 0.7586,
+      "step": 353
+    },
+    {
+      "epoch": 0.8977650974797908,
+      "grad_norm": 4.857143402099609,
+      "learning_rate": 1.768361291169529e-05,
+      "loss": 0.7574,
+      "step": 354
+    },
+    {
+      "epoch": 0.9003011570771913,
+      "grad_norm": 4.302799224853516,
+      "learning_rate": 1.7664664325436308e-05,
+      "loss": 0.7619,
+      "step": 355
+    },
+    {
+      "epoch": 0.9028372166745918,
+      "grad_norm": 7.7203826904296875,
+      "learning_rate": 1.7645648792960475e-05,
+      "loss": 0.7666,
+      "step": 356
+    },
+    {
+      "epoch": 0.9053732762719924,
+      "grad_norm": 7.272395133972168,
+      "learning_rate": 1.7626566480357008e-05,
+      "loss": 0.7632,
+      "step": 357
+    },
+    {
+      "epoch": 0.9079093358693929,
+      "grad_norm": 5.671116828918457,
+      "learning_rate": 1.760741755429838e-05,
+      "loss": 0.7806,
+      "step": 358
+    },
+    {
+      "epoch": 0.9104453954667935,
+      "grad_norm": 6.177848815917969,
+      "learning_rate": 1.7588202182038904e-05,
+      "loss": 0.7809,
+      "step": 359
+    },
+    {
+      "epoch": 0.912981455064194,
+      "grad_norm": 4.729391574859619,
+      "learning_rate": 1.7568920531413264e-05,
+      "loss": 0.7589,
+      "step": 360
+    },
+    {
+      "epoch": 0.9155175146615946,
+      "grad_norm": 5.8260698318481445,
+      "learning_rate": 1.754957277083503e-05,
+      "loss": 0.7596,
+      "step": 361
+    },
+    {
+      "epoch": 0.918053574258995,
+      "grad_norm": 4.793227195739746,
+      "learning_rate": 1.753015906929522e-05,
+      "loss": 0.7575,
+      "step": 362
+    },
+    {
+      "epoch": 0.9205896338563956,
+      "grad_norm": 4.630208969116211,
+      "learning_rate": 1.7510679596360808e-05,
+      "loss": 0.7561,
+      "step": 363
+    },
+    {
+      "epoch": 0.9231256934537961,
+      "grad_norm": 4.497549057006836,
+      "learning_rate": 1.749113452217322e-05,
+      "loss": 0.7662,
+      "step": 364
+    },
+    {
+      "epoch": 0.9256617530511967,
+      "grad_norm": 4.097873687744141,
+      "learning_rate": 1.7471524017446886e-05,
+      "loss": 0.7629,
+      "step": 365
+    },
+    {
+      "epoch": 0.9281978126485972,
+      "grad_norm": 4.151478290557861,
+      "learning_rate": 1.745184825346773e-05,
+      "loss": 0.7695,
+      "step": 366
+    },
+    {
+      "epoch": 0.9307338722459978,
+      "grad_norm": 3.8862032890319824,
+      "learning_rate": 1.7432107402091678e-05,
+      "loss": 0.7819,
+      "step": 367
+    },
+    {
+      "epoch": 0.9332699318433983,
+      "grad_norm": 4.575004577636719,
+      "learning_rate": 1.741230163574315e-05,
+      "loss": 0.7846,
+      "step": 368
+    },
+    {
+      "epoch": 0.9358059914407989,
+      "grad_norm": 3.8442764282226562,
+      "learning_rate": 1.7392431127413557e-05,
+      "loss": 0.7512,
+      "step": 369
+    },
+    {
+      "epoch": 0.9383420510381993,
+      "grad_norm": 3.7705078125,
+      "learning_rate": 1.73724960506598e-05,
+      "loss": 0.7579,
+      "step": 370
+    },
+    {
+      "epoch": 0.9408781106355999,
+      "grad_norm": 3.802544355392456,
+      "learning_rate": 1.735249657960274e-05,
+      "loss": 0.7795,
+      "step": 371
+    },
+    {
+      "epoch": 0.9434141702330004,
+      "grad_norm": 3.7860090732574463,
+      "learning_rate": 1.7332432888925694e-05,
+      "loss": 0.7546,
+      "step": 372
+    },
+    {
+      "epoch": 0.945950229830401,
+      "grad_norm": 3.4378321170806885,
+      "learning_rate": 1.731230515387288e-05,
+      "loss": 0.7447,
+      "step": 373
+    },
+    {
+      "epoch": 0.9484862894278016,
+      "grad_norm": 3.5739309787750244,
+      "learning_rate": 1.7292113550247926e-05,
+      "loss": 0.7858,
+      "step": 374
+    },
+    {
+      "epoch": 0.9510223490252021,
+      "grad_norm": 3.7891266345977783,
+      "learning_rate": 1.7271858254412293e-05,
+      "loss": 0.7754,
+      "step": 375
+    },
+    {
+      "epoch": 0.9535584086226027,
+      "grad_norm": 4.067729473114014,
+      "learning_rate": 1.725153944328377e-05,
+      "loss": 0.7549,
+      "step": 376
+    },
+    {
+      "epoch": 0.9560944682200032,
+      "grad_norm": 3.7381367683410645,
+      "learning_rate": 1.7231157294334907e-05,
+      "loss": 0.7694,
+      "step": 377
+    },
+    {
+      "epoch": 0.9586305278174037,
+      "grad_norm": 4.68396520614624,
+      "learning_rate": 1.7210711985591472e-05,
+      "loss": 0.7743,
+      "step": 378
+    },
+    {
+      "epoch": 0.9611665874148042,
+      "grad_norm": 3.919663667678833,
+      "learning_rate": 1.7190203695630905e-05,
+      "loss": 0.7454,
+      "step": 379
+    },
+    {
+      "epoch": 0.9637026470122048,
+      "grad_norm": 4.522794246673584,
+      "learning_rate": 1.716963260358073e-05,
+      "loss": 0.7654,
+      "step": 380
+    },
+    {
+      "epoch": 0.9662387066096053,
+      "grad_norm": 5.512786388397217,
+      "learning_rate": 1.714899888911703e-05,
+      "loss": 0.7758,
+      "step": 381
+    },
+    {
+      "epoch": 0.9687747662070059,
+      "grad_norm": 6.227116107940674,
+      "learning_rate": 1.712830273246284e-05,
+      "loss": 0.7679,
+      "step": 382
+    },
+    {
+      "epoch": 0.9713108258044064,
+      "grad_norm": 3.994640350341797,
+      "learning_rate": 1.71075443143866e-05,
+      "loss": 0.7569,
+      "step": 383
+    },
+    {
+      "epoch": 0.973846885401807,
+      "grad_norm": 5.445749759674072,
+      "learning_rate": 1.708672381620056e-05,
+      "loss": 0.7789,
+      "step": 384
+    },
+    {
+      "epoch": 0.9763829449992075,
+      "grad_norm": 4.122302532196045,
+      "learning_rate": 1.7065841419759212e-05,
+      "loss": 0.7745,
+      "step": 385
+    },
+    {
+      "epoch": 0.978919004596608,
+      "grad_norm": 4.450962543487549,
+      "learning_rate": 1.704489730745768e-05,
+      "loss": 0.7599,
+      "step": 386
+    },
+    {
+      "epoch": 0.9814550641940085,
+      "grad_norm": 4.200963973999023,
+      "learning_rate": 1.702389166223015e-05,
+      "loss": 0.7936,
+      "step": 387
+    },
+    {
+      "epoch": 0.9839911237914091,
+      "grad_norm": 3.901350259780884,
+      "learning_rate": 1.7002824667548243e-05,
+      "loss": 0.7652,
+      "step": 388
+    },
+    {
+      "epoch": 0.9865271833888096,
+      "grad_norm": 3.8245253562927246,
+      "learning_rate": 1.6981696507419457e-05,
+      "loss": 0.7848,
+      "step": 389
+    },
+    {
+      "epoch": 0.9890632429862102,
+      "grad_norm": 4.071739673614502,
+      "learning_rate": 1.6960507366385508e-05,
+      "loss": 0.7636,
+      "step": 390
+    },
+    {
+      "epoch": 0.9915993025836107,
+      "grad_norm": 3.932081937789917,
+      "learning_rate": 1.6939257429520756e-05,
+      "loss": 0.7848,
+      "step": 391
+    },
+    {
+      "epoch": 0.9941353621810113,
+      "grad_norm": 4.051665782928467,
+      "learning_rate": 1.691794688243058e-05,
+      "loss": 0.7463,
+      "step": 392
+    },
+    {
+      "epoch": 0.9966714217784118,
+      "grad_norm": 3.5884952545166016,
+      "learning_rate": 1.689657591124974e-05,
+      "loss": 0.7833,
+      "step": 393
+    },
+    {
+      "epoch": 0.9992074813758123,
+      "grad_norm": 3.639853000640869,
+      "learning_rate": 1.6875144702640776e-05,
+      "loss": 0.7559,
+      "step": 394
+    },
+    {
+      "epoch": 1.001743540973213,
+      "grad_norm": 3.44031023979187,
+      "learning_rate": 1.6853653443792354e-05,
+      "loss": 0.7374,
+      "step": 395
+    },
+    {
+      "epoch": 1.0042796005706134,
+      "grad_norm": 3.99192214012146,
+      "learning_rate": 1.6832102322417653e-05,
+      "loss": 0.7667,
+      "step": 396
+    },
+    {
+      "epoch": 1.006815660168014,
+      "grad_norm": 3.743447780609131,
+      "learning_rate": 1.6810491526752713e-05,
+      "loss": 0.7685,
+      "step": 397
+    },
+    {
+      "epoch": 1.0093517197654145,
+      "grad_norm": 4.088460445404053,
+      "learning_rate": 1.6788821245554792e-05,
+      "loss": 0.7446,
+      "step": 398
+    },
+    {
+      "epoch": 1.011887779362815,
+      "grad_norm": 4.043125629425049,
+      "learning_rate": 1.6767091668100706e-05,
+      "loss": 0.7451,
+      "step": 399
+    },
+    {
+      "epoch": 1.0144238389602156,
+      "grad_norm": 6.229911804199219,
+      "learning_rate": 1.674530298418521e-05,
+      "loss": 0.7588,
+      "step": 400
+    },
+    {
+      "epoch": 1.0144238389602156,
+      "eval_loss": NaN,
+      "eval_runtime": 206.7042,
+      "eval_samples_per_second": 40.227,
+      "eval_steps_per_second": 5.031,
+      "step": 400
+    },
+    {
+      "epoch": 1.016959898557616,
+      "grad_norm": 4.685118198394775,
+      "learning_rate": 1.67234553841193e-05,
+      "loss": 0.749,
+      "step": 401
+    },
+    {
+      "epoch": 1.0194959581550167,
+      "grad_norm": 4.485588073730469,
+      "learning_rate": 1.6701549058728586e-05,
+      "loss": 0.7798,
+      "step": 402
+    },
+    {
+      "epoch": 1.0220320177524171,
+      "grad_norm": 4.3955888748168945,
+      "learning_rate": 1.6679584199351594e-05,
+      "loss": 0.7533,
+      "step": 403
+    },
+    {
+      "epoch": 1.0245680773498178,
+      "grad_norm": 4.288748741149902,
+      "learning_rate": 1.6657560997838105e-05,
+      "loss": 0.744,
+      "step": 404
+    },
+    {
+      "epoch": 1.0271041369472182,
+      "grad_norm": 3.8942112922668457,
+      "learning_rate": 1.6635479646547505e-05,
+      "loss": 0.734,
+      "step": 405
+    },
+    {
+      "epoch": 1.0296401965446187,
+      "grad_norm": 4.0398688316345215,
+      "learning_rate": 1.6613340338347065e-05,
+      "loss": 0.7521,
+      "step": 406
+    },
+    {
+      "epoch": 1.0321762561420194,
+      "grad_norm": 3.8805272579193115,
+      "learning_rate": 1.6591143266610277e-05,
+      "loss": 0.7363,
+      "step": 407
+    },
+    {
+      "epoch": 1.0347123157394198,
+      "grad_norm": 3.7131450176239014,
+      "learning_rate": 1.6568888625215168e-05,
+      "loss": 0.7523,
+      "step": 408
+    },
+    {
+      "epoch": 1.0372483753368205,
+      "grad_norm": 3.6574320793151855,
+      "learning_rate": 1.65465766085426e-05,
+      "loss": 0.7534,
+      "step": 409
+    },
+    {
+      "epoch": 1.039784434934221,
+      "grad_norm": 4.02537202835083,
+      "learning_rate": 1.652420741147457e-05,
+      "loss": 0.7604,
+      "step": 410
+    },
+    {
+      "epoch": 1.0423204945316216,
+      "grad_norm": 3.5677833557128906,
+      "learning_rate": 1.6501781229392517e-05,
+      "loss": 0.7587,
+      "step": 411
+    },
+    {
+      "epoch": 1.044856554129022,
+      "grad_norm": 3.8453805446624756,
+      "learning_rate": 1.6479298258175603e-05,
+      "loss": 0.7862,
+      "step": 412
+    },
+    {
+      "epoch": 1.0473926137264227,
+      "grad_norm": 3.581329584121704,
+      "learning_rate": 1.6456758694199008e-05,
+      "loss": 0.7921,
+      "step": 413
+    },
+    {
+      "epoch": 1.0499286733238231,
+      "grad_norm": 3.6574368476867676,
+      "learning_rate": 1.6434162734332227e-05,
+      "loss": 0.7932,
+      "step": 414
+    },
+    {
+      "epoch": 1.0524647329212236,
+      "grad_norm": 3.598440408706665,
+      "learning_rate": 1.641151057593733e-05,
+      "loss": 0.7453,
+      "step": 415
+    },
+    {
+      "epoch": 1.0550007925186242,
+      "grad_norm": 3.7227725982666016,
+      "learning_rate": 1.6388802416867245e-05,
+      "loss": 0.7972,
+      "step": 416
+    },
+    {
+      "epoch": 1.0575368521160247,
+      "grad_norm": 5.265769004821777,
+      "learning_rate": 1.636603845546404e-05,
+      "loss": 0.7631,
+      "step": 417
+    },
+    {
+      "epoch": 1.0600729117134253,
+      "grad_norm": 4.70854377746582,
+      "learning_rate": 1.6343218890557172e-05,
+      "loss": 0.7727,
+      "step": 418
+    },
+    {
+      "epoch": 1.0626089713108258,
+      "grad_norm": 4.911214351654053,
+      "learning_rate": 1.6320343921461774e-05,
+      "loss": 0.7696,
+      "step": 419
+    },
+    {
+      "epoch": 1.0651450309082264,
+      "grad_norm": 4.542303562164307,
+      "learning_rate": 1.6297413747976892e-05,
+      "loss": 0.7602,
+      "step": 420
+    },
+    {
+      "epoch": 1.0676810905056269,
+      "grad_norm": 4.8770365715026855,
+      "learning_rate": 1.6274428570383755e-05,
+      "loss": 0.758,
+      "step": 421
+    },
+    {
+      "epoch": 1.0702171501030273,
+      "grad_norm": 4.914615154266357,
+      "learning_rate": 1.6251388589444013e-05,
+      "loss": 0.7438,
+      "step": 422
+    },
+    {
+      "epoch": 1.072753209700428,
+      "grad_norm": 9.583602905273438,
+      "learning_rate": 1.6228294006397994e-05,
+      "loss": 0.7587,
+      "step": 423
+    },
+    {
+      "epoch": 1.0752892692978284,
+      "grad_norm": 6.488173007965088,
+      "learning_rate": 1.6205145022962946e-05,
+      "loss": 0.7806,
+      "step": 424
+    },
+    {
+      "epoch": 1.077825328895229,
+      "grad_norm": 5.551193714141846,
+      "learning_rate": 1.618194184133126e-05,
+      "loss": 0.7495,
+      "step": 425
+    },
+    {
+      "epoch": 1.0803613884926295,
+      "grad_norm": 4.765059471130371,
+      "learning_rate": 1.615868466416873e-05,
+      "loss": 0.7742,
+      "step": 426
+    },
+    {
+      "epoch": 1.0828974480900302,
+      "grad_norm": 4.457275867462158,
+      "learning_rate": 1.613537369461276e-05,
+      "loss": 0.7782,
+      "step": 427
+    },
+    {
+      "epoch": 1.0854335076874306,
+      "grad_norm": 4.418558120727539,
+      "learning_rate": 1.61120091362706e-05,
+      "loss": 0.7528,
+      "step": 428
+    },
+    {
+      "epoch": 1.0879695672848313,
+      "grad_norm": 3.9728074073791504,
+      "learning_rate": 1.608859119321756e-05,
+      "loss": 0.7718,
+      "step": 429
+    },
+    {
+      "epoch": 1.0905056268822317,
+      "grad_norm": 3.781358003616333,
+      "learning_rate": 1.6065120069995246e-05,
+      "loss": 0.7534,
+      "step": 430
+    },
+    {
+      "epoch": 1.0930416864796322,
+      "grad_norm": 3.913799524307251,
+      "learning_rate": 1.6041595971609747e-05,
+      "loss": 0.7516,
+      "step": 431
+    },
+    {
+      "epoch": 1.0955777460770328,
+      "grad_norm": 3.8430593013763428,
+      "learning_rate": 1.601801910352987e-05,
+      "loss": 0.7498,
+      "step": 432
+    },
+    {
+      "epoch": 1.0981138056744333,
+      "grad_norm": 3.853971242904663,
+      "learning_rate": 1.5994389671685323e-05,
+      "loss": 0.7626,
+      "step": 433
+    },
+    {
+      "epoch": 1.100649865271834,
+      "grad_norm": 3.683640241622925,
+      "learning_rate": 1.5970707882464934e-05,
+      "loss": 0.7697,
+      "step": 434
+    },
+    {
+      "epoch": 1.1031859248692344,
+      "grad_norm": 3.7113516330718994,
+      "learning_rate": 1.594697394271484e-05,
+      "loss": 0.7695,
+      "step": 435
+    },
+    {
+      "epoch": 1.105721984466635,
+      "grad_norm": 3.3953802585601807,
+      "learning_rate": 1.5923188059736678e-05,
+      "loss": 0.7755,
+      "step": 436
+    },
+    {
+      "epoch": 1.1082580440640355,
+      "grad_norm": 3.871687173843384,
+      "learning_rate": 1.589935044128578e-05,
+      "loss": 0.7587,
+      "step": 437
+    },
+    {
+      "epoch": 1.1107941036614362,
+      "grad_norm": 3.934541702270508,
+      "learning_rate": 1.5875461295569356e-05,
+      "loss": 0.7669,
+      "step": 438
+    },
+    {
+      "epoch": 1.1133301632588366,
+      "grad_norm": 3.9485020637512207,
+      "learning_rate": 1.5851520831244674e-05,
+      "loss": 0.7666,
+      "step": 439
+    },
+    {
+      "epoch": 1.115866222856237,
+      "grad_norm": 5.383018493652344,
+      "learning_rate": 1.582752925741724e-05,
+      "loss": 0.7603,
+      "step": 440
+    },
+    {
+      "epoch": 1.1184022824536377,
+      "grad_norm": 4.171992301940918,
+      "learning_rate": 1.5803486783638976e-05,
+      "loss": 0.7647,
+      "step": 441
+    },
+    {
+      "epoch": 1.1209383420510382,
+      "grad_norm": 4.432652950286865,
+      "learning_rate": 1.5779393619906373e-05,
+      "loss": 0.7774,
+      "step": 442
+    },
+    {
+      "epoch": 1.1234744016484388,
+      "grad_norm": 4.361434459686279,
+      "learning_rate": 1.575524997665868e-05,
+      "loss": 0.7781,
+      "step": 443
+    },
+    {
+      "epoch": 1.1260104612458393,
+      "grad_norm": 3.9876654148101807,
+      "learning_rate": 1.573105606477605e-05,
+      "loss": 0.7794,
+      "step": 444
+    },
+    {
+      "epoch": 1.1285465208432397,
+      "grad_norm": 4.282565593719482,
+      "learning_rate": 1.5706812095577696e-05,
+      "loss": 0.779,
+      "step": 445
+    },
+    {
+      "epoch": 1.1310825804406404,
+      "grad_norm": 3.9868757724761963,
+      "learning_rate": 1.5682518280820055e-05,
+      "loss": 0.7713,
+      "step": 446
+    },
+    {
+      "epoch": 1.1336186400380408,
+      "grad_norm": 4.086343288421631,
+      "learning_rate": 1.565817483269494e-05,
+      "loss": 0.7502,
+      "step": 447
+    },
+    {
+      "epoch": 1.1361546996354415,
+      "grad_norm": 3.90673828125,
+      "learning_rate": 1.563378196382768e-05,
+      "loss": 0.7851,
+      "step": 448
+    },
+    {
+      "epoch": 1.138690759232842,
+      "grad_norm": 4.063551425933838,
+      "learning_rate": 1.5609339887275262e-05,
+      "loss": 0.7524,
+      "step": 449
+    },
+    {
+      "epoch": 1.1412268188302426,
+      "grad_norm": 3.974640369415283,
+      "learning_rate": 1.5584848816524462e-05,
+      "loss": 0.7618,
+      "step": 450
+    },
+    {
+      "epoch": 1.1412268188302426,
+      "eval_loss": NaN,
+      "eval_runtime": 206.8241,
+      "eval_samples_per_second": 40.203,
+      "eval_steps_per_second": 5.028,
+      "step": 450
+    },
+    {
+      "epoch": 1.143762878427643,
+      "grad_norm": 3.8022267818450928,
+      "learning_rate": 1.5560308965490013e-05,
+      "loss": 0.7592,
+      "step": 451
+    },
+    {
+      "epoch": 1.1462989380250437,
+      "grad_norm": 4.060011386871338,
+      "learning_rate": 1.55357205485127e-05,
+      "loss": 0.7659,
+      "step": 452
+    },
+    {
+      "epoch": 1.1488349976224441,
+      "grad_norm": 3.8058009147644043,
+      "learning_rate": 1.5511083780357507e-05,
+      "loss": 0.759,
+      "step": 453
+    },
+    {
+      "epoch": 1.1513710572198446,
+      "grad_norm": 3.945983409881592,
+      "learning_rate": 1.5486398876211724e-05,
+      "loss": 0.7562,
+      "step": 454
+    },
+    {
+      "epoch": 1.1539071168172452,
+      "grad_norm": 3.7832720279693604,
+      "learning_rate": 1.5461666051683103e-05,
+      "loss": 0.7494,
+      "step": 455
+    },
+    {
+      "epoch": 1.1564431764146457,
+      "grad_norm": 3.485811710357666,
+      "learning_rate": 1.5436885522797927e-05,
+      "loss": 0.7674,
+      "step": 456
+    },
+    {
+      "epoch": 1.1589792360120463,
+      "grad_norm": 3.889207124710083,
+      "learning_rate": 1.5412057505999167e-05,
+      "loss": 0.7599,
+      "step": 457
+    },
+    {
+      "epoch": 1.1615152956094468,
+      "grad_norm": 3.859790563583374,
+      "learning_rate": 1.5387182218144565e-05,
+      "loss": 0.7942,
+      "step": 458
+    },
+    {
+      "epoch": 1.1640513552068474,
+      "grad_norm": 5.167479038238525,
+      "learning_rate": 1.5362259876504745e-05,
+      "loss": 0.7737,
+      "step": 459
+    },
+    {
+      "epoch": 1.1665874148042479,
+      "grad_norm": 4.347815036773682,
+      "learning_rate": 1.5337290698761316e-05,
+      "loss": 0.7638,
+      "step": 460
+    },
+    {
+      "epoch": 1.1691234744016485,
+      "grad_norm": 6.943878650665283,
+      "learning_rate": 1.531227490300498e-05,
+      "loss": 0.7754,
+      "step": 461
+    },
+    {
+      "epoch": 1.171659533999049,
+      "grad_norm": 6.679975509643555,
+      "learning_rate": 1.528721270773361e-05,
+      "loss": 0.759,
+      "step": 462
+    },
+    {
+      "epoch": 1.1741955935964494,
+      "grad_norm": 4.526600360870361,
+      "learning_rate": 1.526210433185036e-05,
+      "loss": 0.7824,
+      "step": 463
+    },
+    {
+      "epoch": 1.17673165319385,
+      "grad_norm": 4.113241195678711,
+      "learning_rate": 1.5236949994661736e-05,
+      "loss": 0.7556,
+      "step": 464
+    },
+    {
+      "epoch": 1.1792677127912505,
+      "grad_norm": 4.551198482513428,
+      "learning_rate": 1.5211749915875693e-05,
+      "loss": 0.7644,
+      "step": 465
+    },
+    {
+      "epoch": 1.1818037723886512,
+      "grad_norm": 4.393762588500977,
+      "learning_rate": 1.5186504315599711e-05,
+      "loss": 0.7661,
+      "step": 466
+    },
+    {
+      "epoch": 1.1843398319860516,
+      "grad_norm": 3.9809296131134033,
+      "learning_rate": 1.5161213414338867e-05,
+      "loss": 0.7718,
+      "step": 467
+    },
+    {
+      "epoch": 1.1868758915834523,
+      "grad_norm": 4.403090476989746,
+      "learning_rate": 1.5135877432993923e-05,
+      "loss": 0.7614,
+      "step": 468
+    },
+    {
+      "epoch": 1.1894119511808527,
+      "grad_norm": 4.471353054046631,
+      "learning_rate": 1.5110496592859381e-05,
+      "loss": 0.7916,
+      "step": 469
+    },
+    {
+      "epoch": 1.1919480107782534,
+      "grad_norm": 3.77409029006958,
+      "learning_rate": 1.5085071115621564e-05,
+      "loss": 0.7803,
+      "step": 470
+    },
+    {
+      "epoch": 1.1944840703756539,
+      "grad_norm": 4.09141206741333,
+      "learning_rate": 1.5059601223356666e-05,
+      "loss": 0.7693,
+      "step": 471
+    },
+    {
+      "epoch": 1.1970201299730543,
+      "grad_norm": 4.290613651275635,
+      "learning_rate": 1.503408713852883e-05,
+      "loss": 0.7662,
+      "step": 472
+    },
+    {
+      "epoch": 1.199556189570455,
+      "grad_norm": 3.730660915374756,
+      "learning_rate": 1.5008529083988182e-05,
+      "loss": 0.7702,
+      "step": 473
+    },
+    {
+      "epoch": 1.2020922491678554,
+      "grad_norm": 3.85732364654541,
+      "learning_rate": 1.4982927282968907e-05,
+      "loss": 0.7741,
+      "step": 474
+    },
+    {
+      "epoch": 1.204628308765256,
+      "grad_norm": 3.743196725845337,
+      "learning_rate": 1.4957281959087278e-05,
+      "loss": 0.7709,
+      "step": 475
+    },
+    {
+      "epoch": 1.2071643683626565,
+      "grad_norm": 3.6812498569488525,
+      "learning_rate": 1.493159333633973e-05,
+      "loss": 0.7396,
+      "step": 476
+    },
+    {
+      "epoch": 1.209700427960057,
+      "grad_norm": 3.738018035888672,
+      "learning_rate": 1.4905861639100869e-05,
+      "loss": 0.7489,
+      "step": 477
+    },
+    {
+      "epoch": 1.2122364875574576,
+      "grad_norm": 3.8504762649536133,
+      "learning_rate": 1.4880087092121546e-05,
+      "loss": 0.7752,
+      "step": 478
+    },
+    {
+      "epoch": 1.214772547154858,
+      "grad_norm": 3.9431211948394775,
+      "learning_rate": 1.4854269920526872e-05,
+      "loss": 0.7704,
+      "step": 479
+    },
+    {
+      "epoch": 1.2173086067522587,
+      "grad_norm": 3.850952386856079,
+      "learning_rate": 1.4828410349814257e-05,
+      "loss": 0.7689,
+      "step": 480
+    },
+    {
+      "epoch": 1.2198446663496592,
+      "grad_norm": 7.408934593200684,
+      "learning_rate": 1.4802508605851447e-05,
+      "loss": 0.7627,
+      "step": 481
+    },
+    {
+      "epoch": 1.2223807259470598,
+      "grad_norm": 6.095399856567383,
+      "learning_rate": 1.4776564914874537e-05,
+      "loss": 0.7579,
+      "step": 482
+    },
+    {
+      "epoch": 1.2249167855444603,
+      "grad_norm": 5.658483028411865,
+      "learning_rate": 1.4750579503486011e-05,
+      "loss": 0.7719,
+      "step": 483
+    },
+    {
+      "epoch": 1.227452845141861,
+      "grad_norm": 6.234167098999023,
+      "learning_rate": 1.4724552598652756e-05,
+      "loss": 0.7709,
+      "step": 484
+    },
+    {
+      "epoch": 1.2299889047392614,
+      "grad_norm": 4.473567008972168,
+      "learning_rate": 1.4698484427704079e-05,
+      "loss": 0.7863,
+      "step": 485
+    },
+    {
+      "epoch": 1.2325249643366618,
+      "grad_norm": 7.316669940948486,
+      "learning_rate": 1.4672375218329714e-05,
+      "loss": 0.7751,
+      "step": 486
+    },
+    {
+      "epoch": 1.2350610239340625,
+      "grad_norm": 5.022395610809326,
+      "learning_rate": 1.4646225198577847e-05,
+      "loss": 0.7815,
+      "step": 487
+    },
+    {
+      "epoch": 1.237597083531463,
+      "grad_norm": 5.747352600097656,
+      "learning_rate": 1.4620034596853122e-05,
+      "loss": 0.7762,
+      "step": 488
+    },
+    {
+      "epoch": 1.2401331431288636,
+      "grad_norm": 5.750426769256592,
+      "learning_rate": 1.4593803641914633e-05,
+      "loss": 0.788,
+      "step": 489
+    },
+    {
+      "epoch": 1.242669202726264,
+      "grad_norm": 4.795891284942627,
+      "learning_rate": 1.4567532562873942e-05,
+      "loss": 0.7856,
+      "step": 490
+    },
+    {
+      "epoch": 1.2452052623236647,
+      "grad_norm": 4.847960472106934,
+      "learning_rate": 1.454122158919307e-05,
+      "loss": 0.7596,
+      "step": 491
+    },
+    {
+      "epoch": 1.2477413219210651,
+      "grad_norm": 4.716196060180664,
+      "learning_rate": 1.4514870950682492e-05,
+      "loss": 0.7601,
+      "step": 492
+    },
+    {
+      "epoch": 1.2502773815184658,
+      "grad_norm": 3.933077812194824,
+      "learning_rate": 1.4488480877499127e-05,
+      "loss": 0.764,
+      "step": 493
+    },
+    {
+      "epoch": 1.2528134411158662,
+      "grad_norm": 4.576488018035889,
+      "learning_rate": 1.4462051600144338e-05,
+      "loss": 0.785,
+      "step": 494
+    },
+    {
+      "epoch": 1.2553495007132667,
+      "grad_norm": 4.397135257720947,
+      "learning_rate": 1.4435583349461918e-05,
+      "loss": 0.7731,
+      "step": 495
+    },
+    {
+      "epoch": 1.2578855603106673,
+      "grad_norm": 4.1071553230285645,
+      "learning_rate": 1.4409076356636051e-05,
+      "loss": 0.7836,
+      "step": 496
+    },
+    {
+      "epoch": 1.2604216199080678,
+      "grad_norm": 4.0747599601745605,
+      "learning_rate": 1.4382530853189325e-05,
+      "loss": 0.7565,
+      "step": 497
+    },
+    {
+      "epoch": 1.2629576795054684,
+      "grad_norm": 4.071841716766357,
+      "learning_rate": 1.4355947070980696e-05,
+      "loss": 0.7865,
+      "step": 498
+    },
+    {
+      "epoch": 1.2654937391028689,
+      "grad_norm": 3.949681520462036,
+      "learning_rate": 1.432932524220345e-05,
+      "loss": 0.7334,
+      "step": 499
+    },
+    {
+      "epoch": 1.2680297987002693,
+      "grad_norm": 4.420541286468506,
+      "learning_rate": 1.4302665599383196e-05,
+      "loss": 0.779,
+      "step": 500
+    },
+    {
+      "epoch": 1.2680297987002693,
+      "eval_loss": NaN,
+      "eval_runtime": 206.7956,
+      "eval_samples_per_second": 40.209,
+      "eval_steps_per_second": 5.029,
+      "step": 500
+    },
+    {
+      "epoch": 1.27056585829767,
+      "grad_norm": 3.622514486312866,
+      "learning_rate": 1.4275968375375825e-05,
+      "loss": 0.7537,
+      "step": 501
+    },
+    {
+      "epoch": 1.2731019178950707,
+      "grad_norm": 5.610208034515381,
+      "learning_rate": 1.4249233803365474e-05,
+      "loss": 0.7708,
+      "step": 502
+    },
+    {
+      "epoch": 1.275637977492471,
+      "grad_norm": 4.894537448883057,
+      "learning_rate": 1.4222462116862492e-05,
+      "loss": 0.7685,
+      "step": 503
+    },
+    {
+      "epoch": 1.2781740370898715,
+      "grad_norm": 5.148238658905029,
+      "learning_rate": 1.4195653549701406e-05,
+      "loss": 0.7702,
+      "step": 504
+    },
+    {
+      "epoch": 1.2807100966872722,
+      "grad_norm": 4.645820617675781,
+      "learning_rate": 1.4168808336038869e-05,
+      "loss": 0.7687,
+      "step": 505
+    },
+    {
+      "epoch": 1.2832461562846726,
+      "grad_norm": 5.605941295623779,
+      "learning_rate": 1.4141926710351617e-05,
+      "loss": 0.7623,
+      "step": 506
+    },
+    {
+      "epoch": 1.2857822158820733,
+      "grad_norm": 5.204614639282227,
+      "learning_rate": 1.411500890743443e-05,
+      "loss": 0.7778,
+      "step": 507
+    },
+    {
+      "epoch": 1.2883182754794738,
+      "grad_norm": 5.3651123046875,
+      "learning_rate": 1.4088055162398066e-05,
+      "loss": 0.7713,
+      "step": 508
+    },
+    {
+      "epoch": 1.2908543350768742,
+      "grad_norm": 4.416318416595459,
+      "learning_rate": 1.4061065710667223e-05,
+      "loss": 0.7558,
+      "step": 509
+    },
+    {
+      "epoch": 1.2933903946742749,
+      "grad_norm": 4.6826372146606445,
+      "learning_rate": 1.4034040787978472e-05,
+      "loss": 0.7586,
+      "step": 510
+    },
+    {
+      "epoch": 1.2959264542716755,
+      "grad_norm": 4.509591579437256,
+      "learning_rate": 1.40069806303782e-05,
+      "loss": 0.7881,
+      "step": 511
+    },
+    {
+      "epoch": 1.298462513869076,
+      "grad_norm": 3.8483219146728516,
+      "learning_rate": 1.397988547422055e-05,
+      "loss": 0.7438,
+      "step": 512
+    },
+    {
+      "epoch": 1.3009985734664764,
+      "grad_norm": 3.8421945571899414,
+      "learning_rate": 1.3952755556165363e-05,
+      "loss": 0.7635,
+      "step": 513
+    },
+    {
+      "epoch": 1.303534633063877,
+      "grad_norm": 3.7505085468292236,
+      "learning_rate": 1.3925591113176093e-05,
+      "loss": 0.7547,
+      "step": 514
+    },
+    {
+      "epoch": 1.3060706926612775,
+      "grad_norm": 3.987705707550049,
+      "learning_rate": 1.3898392382517759e-05,
+      "loss": 0.779,
+      "step": 515
+    },
+    {
+      "epoch": 1.3086067522586782,
+      "grad_norm": 3.8721580505371094,
+      "learning_rate": 1.3871159601754853e-05,
+      "loss": 0.7425,
+      "step": 516
+    },
+    {
+      "epoch": 1.3111428118560786,
+      "grad_norm": 3.712494373321533,
+      "learning_rate": 1.3843893008749283e-05,
+      "loss": 0.7676,
+      "step": 517
+    },
+    {
+      "epoch": 1.313678871453479,
+      "grad_norm": 3.6847803592681885,
+      "learning_rate": 1.3816592841658278e-05,
+      "loss": 0.7496,
+      "step": 518
+    },
+    {
+      "epoch": 1.3162149310508797,
+      "grad_norm": 3.620512008666992,
+      "learning_rate": 1.3789259338932324e-05,
+      "loss": 0.7524,
+      "step": 519
+    },
+    {
+      "epoch": 1.3187509906482802,
+      "grad_norm": 3.88765287399292,
+      "learning_rate": 1.3761892739313065e-05,
+      "loss": 0.7625,
+      "step": 520
+    },
+    {
+      "epoch": 1.3212870502456808,
+      "grad_norm": 4.04611349105835,
+      "learning_rate": 1.3734493281831236e-05,
+      "loss": 0.7672,
+      "step": 521
+    },
+    {
+      "epoch": 1.3238231098430813,
+      "grad_norm": 3.500861167907715,
+      "learning_rate": 1.3707061205804562e-05,
+      "loss": 0.7527,
+      "step": 522
+    },
+    {
+      "epoch": 1.326359169440482,
+      "grad_norm": 4.245331764221191,
+      "learning_rate": 1.3679596750835667e-05,
+      "loss": 0.7687,
+      "step": 523
+    },
+    {
+      "epoch": 1.3288952290378824,
+      "grad_norm": 4.601229190826416,
+      "learning_rate": 1.3652100156809989e-05,
+      "loss": 0.7476,
+      "step": 524
+    },
+    {
+      "epoch": 1.331431288635283,
+      "grad_norm": 4.409153461456299,
+      "learning_rate": 1.3624571663893683e-05,
+      "loss": 0.7623,
+      "step": 525
+    },
+    {
+      "epoch": 1.3339673482326835,
+      "grad_norm": 4.51855993270874,
+      "learning_rate": 1.3597011512531517e-05,
+      "loss": 0.7666,
+      "step": 526
+    },
+    {
+      "epoch": 1.336503407830084,
+      "grad_norm": 4.2342753410339355,
+      "learning_rate": 1.3569419943444783e-05,
+      "loss": 0.7683,
+      "step": 527
+    },
+    {
+      "epoch": 1.3390394674274846,
+      "grad_norm": 4.348354816436768,
+      "learning_rate": 1.3541797197629183e-05,
+      "loss": 0.7599,
+      "step": 528
+    },
+    {
+      "epoch": 1.341575527024885,
+      "grad_norm": 4.881316661834717,
+      "learning_rate": 1.3514143516352727e-05,
+      "loss": 0.7726,
+      "step": 529
+    },
+    {
+      "epoch": 1.3441115866222857,
+      "grad_norm": 4.846878528594971,
+      "learning_rate": 1.348645914115363e-05,
+      "loss": 0.7851,
+      "step": 530
+    },
+    {
+      "epoch": 1.3466476462196861,
+      "grad_norm": 4.4096293449401855,
+      "learning_rate": 1.3458744313838205e-05,
+      "loss": 0.7871,
+      "step": 531
+    },
+    {
+      "epoch": 1.3491837058170866,
+      "grad_norm": 4.435279846191406,
+      "learning_rate": 1.343099927647874e-05,
+      "loss": 0.7638,
+      "step": 532
+    },
+    {
+      "epoch": 1.3517197654144872,
+      "grad_norm": 3.9706711769104004,
+      "learning_rate": 1.3403224271411386e-05,
+      "loss": 0.7881,
+      "step": 533
+    },
+    {
+      "epoch": 1.354255825011888,
+      "grad_norm": 3.9394490718841553,
+      "learning_rate": 1.3375419541234054e-05,
+      "loss": 0.7602,
+      "step": 534
+    },
+    {
+      "epoch": 1.3567918846092883,
+      "grad_norm": 4.002357482910156,
+      "learning_rate": 1.3347585328804278e-05,
+      "loss": 0.7725,
+      "step": 535
+    },
+    {
+      "epoch": 1.3593279442066888,
+      "grad_norm": 3.9288394451141357,
+      "learning_rate": 1.3319721877237103e-05,
+      "loss": 0.7975,
+      "step": 536
+    },
+    {
+      "epoch": 1.3618640038040895,
+      "grad_norm": 3.7613887786865234,
+      "learning_rate": 1.3291829429902965e-05,
+      "loss": 0.7584,
+      "step": 537
+    },
+    {
+      "epoch": 1.36440006340149,
+      "grad_norm": 4.194827079772949,
+      "learning_rate": 1.3263908230425552e-05,
+      "loss": 0.7935,
+      "step": 538
+    },
+    {
+      "epoch": 1.3669361229988906,
+      "grad_norm": 3.715162515640259,
+      "learning_rate": 1.3235958522679693e-05,
+      "loss": 0.7919,
+      "step": 539
+    },
+    {
+      "epoch": 1.369472182596291,
+      "grad_norm": 3.9799184799194336,
+      "learning_rate": 1.3207980550789217e-05,
+      "loss": 0.767,
+      "step": 540
+    },
+    {
+      "epoch": 1.3720082421936914,
+      "grad_norm": 3.725858449935913,
+      "learning_rate": 1.3179974559124822e-05,
+      "loss": 0.7534,
+      "step": 541
+    },
+    {
+      "epoch": 1.374544301791092,
+      "grad_norm": 4.065157413482666,
+      "learning_rate": 1.3151940792301937e-05,
+      "loss": 0.7694,
+      "step": 542
+    },
+    {
+      "epoch": 1.3770803613884925,
+      "grad_norm": 3.9822163581848145,
+      "learning_rate": 1.3123879495178597e-05,
+      "loss": 0.7802,
+      "step": 543
+    },
+    {
+      "epoch": 1.3796164209858932,
+      "grad_norm": 4.991753101348877,
+      "learning_rate": 1.3095790912853301e-05,
+      "loss": 0.7728,
+      "step": 544
+    },
+    {
+      "epoch": 1.3821524805832937,
+      "grad_norm": 4.176737308502197,
+      "learning_rate": 1.306767529066286e-05,
+      "loss": 0.7711,
+      "step": 545
+    },
+    {
+      "epoch": 1.3846885401806943,
+      "grad_norm": 4.375234603881836,
+      "learning_rate": 1.303953287418026e-05,
+      "loss": 0.7644,
+      "step": 546
+    },
+    {
+      "epoch": 1.3872245997780948,
+      "grad_norm": 4.565071105957031,
+      "learning_rate": 1.3011363909212529e-05,
+      "loss": 0.7876,
+      "step": 547
+    },
+    {
+      "epoch": 1.3897606593754954,
+      "grad_norm": 5.081917762756348,
+      "learning_rate": 1.2983168641798574e-05,
+      "loss": 0.7878,
+      "step": 548
+    },
+    {
+      "epoch": 1.3922967189728959,
+      "grad_norm": 4.294445991516113,
+      "learning_rate": 1.2954947318207035e-05,
+      "loss": 0.7803,
+      "step": 549
+    },
+    {
+      "epoch": 1.3948327785702963,
+      "grad_norm": 5.077268600463867,
+      "learning_rate": 1.2926700184934144e-05,
+      "loss": 0.7479,
+      "step": 550
+    },
+    {
+      "epoch": 1.3948327785702963,
+      "eval_loss": NaN,
+      "eval_runtime": 207.3657,
+      "eval_samples_per_second": 40.098,
+      "eval_steps_per_second": 5.015,
+      "step": 550
+    },
+    {
+      "epoch": 1.397368838167697,
+      "grad_norm": 4.5675129890441895,
+      "learning_rate": 1.289842748870156e-05,
+      "loss": 0.78,
+      "step": 551
+    },
+    {
+      "epoch": 1.3999048977650974,
+      "grad_norm": 4.379223346710205,
+      "learning_rate": 1.287012947645422e-05,
+      "loss": 0.7727,
+      "step": 552
+    },
+    {
+      "epoch": 1.402440957362498,
+      "grad_norm": 4.593344211578369,
+      "learning_rate": 1.2841806395358182e-05,
+      "loss": 0.8058,
+      "step": 553
+    },
+    {
+      "epoch": 1.4049770169598985,
+      "grad_norm": 4.072418212890625,
+      "learning_rate": 1.2813458492798467e-05,
+      "loss": 0.7889,
+      "step": 554
+    },
+    {
+      "epoch": 1.407513076557299,
+      "grad_norm": 4.03213357925415,
+      "learning_rate": 1.2785086016376891e-05,
+      "loss": 0.7777,
+      "step": 555
+    },
+    {
+      "epoch": 1.4100491361546996,
+      "grad_norm": 4.503752708435059,
+      "learning_rate": 1.2756689213909912e-05,
+      "loss": 0.7769,
+      "step": 556
+    },
+    {
+      "epoch": 1.4125851957521003,
+      "grad_norm": 3.815361261367798,
+      "learning_rate": 1.2728268333426463e-05,
+      "loss": 0.7833,
+      "step": 557
+    },
+    {
+      "epoch": 1.4151212553495007,
+      "grad_norm": 4.12267541885376,
+      "learning_rate": 1.2699823623165776e-05,
+      "loss": 0.8036,
+      "step": 558
+    },
+    {
+      "epoch": 1.4176573149469012,
+      "grad_norm": 3.888458728790283,
+      "learning_rate": 1.2671355331575237e-05,
+      "loss": 0.7872,
+      "step": 559
+    },
+    {
+      "epoch": 1.4201933745443018,
+      "grad_norm": 3.5564417839050293,
+      "learning_rate": 1.2642863707308179e-05,
+      "loss": 0.7538,
+      "step": 560
+    },
+    {
+      "epoch": 1.4227294341417023,
+      "grad_norm": 4.016533374786377,
+      "learning_rate": 1.2614348999221752e-05,
+      "loss": 0.7436,
+      "step": 561
+    },
+    {
+      "epoch": 1.425265493739103,
+      "grad_norm": 3.6715681552886963,
+      "learning_rate": 1.2585811456374719e-05,
+      "loss": 0.7702,
+      "step": 562
+    },
+    {
+      "epoch": 1.4278015533365034,
+      "grad_norm": 3.9228549003601074,
+      "learning_rate": 1.2557251328025291e-05,
+      "loss": 0.7586,
+      "step": 563
+    },
+    {
+      "epoch": 1.4303376129339038,
+      "grad_norm": 3.6572742462158203,
+      "learning_rate": 1.2528668863628952e-05,
+      "loss": 0.7762,
+      "step": 564
+    },
+    {
+      "epoch": 1.4328736725313045,
+      "grad_norm": 4.534142017364502,
+      "learning_rate": 1.250006431283628e-05,
+      "loss": 0.7535,
+      "step": 565
+    },
+    {
+      "epoch": 1.4354097321287052,
+      "grad_norm": 6.687479496002197,
+      "learning_rate": 1.247143792549076e-05,
+      "loss": 0.7731,
+      "step": 566
+    },
+    {
+      "epoch": 1.4379457917261056,
+      "grad_norm": 6.1511335372924805,
+      "learning_rate": 1.2442789951626606e-05,
+      "loss": 0.7676,
+      "step": 567
+    },
+    {
+      "epoch": 1.440481851323506,
+      "grad_norm": 4.822118759155273,
+      "learning_rate": 1.2414120641466582e-05,
+      "loss": 0.767,
+      "step": 568
+    },
+    {
+      "epoch": 1.4430179109209067,
+      "grad_norm": 5.115494251251221,
+      "learning_rate": 1.238543024541981e-05,
+      "loss": 0.7355,
+      "step": 569
+    },
+    {
+      "epoch": 1.4455539705183071,
+      "grad_norm": 5.49458646774292,
+      "learning_rate": 1.2356719014079577e-05,
+      "loss": 0.7757,
+      "step": 570
+    },
+    {
+      "epoch": 1.4480900301157078,
+      "grad_norm": 4.576963901519775,
+      "learning_rate": 1.2327987198221165e-05,
+      "loss": 0.7979,
+      "step": 571
+    },
+    {
+      "epoch": 1.4506260897131082,
+      "grad_norm": 5.3251190185546875,
+      "learning_rate": 1.2299235048799645e-05,
+      "loss": 0.7726,
+      "step": 572
+    },
+    {
+      "epoch": 1.4531621493105087,
+      "grad_norm": 4.509469032287598,
+      "learning_rate": 1.2270462816947685e-05,
+      "loss": 0.7749,
+      "step": 573
+    },
+    {
+      "epoch": 1.4556982089079094,
+      "grad_norm": 5.10255765914917,
+      "learning_rate": 1.2241670753973369e-05,
+      "loss": 0.7687,
+      "step": 574
+    },
+    {
+      "epoch": 1.4582342685053098,
+      "grad_norm": 4.80985689163208,
+      "learning_rate": 1.2212859111357984e-05,
+      "loss": 0.7855,
+      "step": 575
+    },
+    {
+      "epoch": 1.4607703281027105,
+      "grad_norm": 4.553503036499023,
+      "learning_rate": 1.2184028140753846e-05,
+      "loss": 0.7978,
+      "step": 576
+    },
+    {
+      "epoch": 1.463306387700111,
+      "grad_norm": 4.262579917907715,
+      "learning_rate": 1.2155178093982075e-05,
+      "loss": 0.7631,
+      "step": 577
+    },
+    {
+      "epoch": 1.4658424472975116,
+      "grad_norm": 4.628722667694092,
+      "learning_rate": 1.212630922303042e-05,
+      "loss": 0.7683,
+      "step": 578
+    },
+    {
+      "epoch": 1.468378506894912,
+      "grad_norm": 3.8812975883483887,
+      "learning_rate": 1.2097421780051043e-05,
+      "loss": 0.7783,
+      "step": 579
+    },
+    {
+      "epoch": 1.4709145664923127,
+      "grad_norm": 4.333192348480225,
+      "learning_rate": 1.206851601735833e-05,
+      "loss": 0.7896,
+      "step": 580
+    },
+    {
+      "epoch": 1.4734506260897131,
+      "grad_norm": 4.1532301902771,
+      "learning_rate": 1.2039592187426665e-05,
+      "loss": 0.7657,
+      "step": 581
+    },
+    {
+      "epoch": 1.4759866856871136,
+      "grad_norm": 4.598367691040039,
+      "learning_rate": 1.2010650542888249e-05,
+      "loss": 0.7774,
+      "step": 582
+    },
+    {
+      "epoch": 1.4785227452845142,
+      "grad_norm": 3.8422224521636963,
+      "learning_rate": 1.1981691336530877e-05,
+      "loss": 0.7662,
+      "step": 583
+    },
+    {
+      "epoch": 1.4810588048819147,
+      "grad_norm": 4.708389759063721,
+      "learning_rate": 1.1952714821295736e-05,
+      "loss": 0.7953,
+      "step": 584
+    },
+    {
+      "epoch": 1.4835948644793153,
+      "grad_norm": 4.034779071807861,
+      "learning_rate": 1.1923721250275202e-05,
+      "loss": 0.7723,
+      "step": 585
+    },
+    {
+      "epoch": 1.4861309240767158,
+      "grad_norm": 5.730340480804443,
+      "learning_rate": 1.1894710876710617e-05,
+      "loss": 0.7935,
+      "step": 586
+    },
+    {
+      "epoch": 1.4886669836741162,
+      "grad_norm": 4.635005474090576,
+      "learning_rate": 1.1865683953990089e-05,
+      "loss": 0.7753,
+      "step": 587
+    },
+    {
+      "epoch": 1.4912030432715169,
+      "grad_norm": 4.392043590545654,
+      "learning_rate": 1.183664073564626e-05,
+      "loss": 0.7734,
+      "step": 588
+    },
+    {
+      "epoch": 1.4937391028689175,
+      "grad_norm": 4.556002616882324,
+      "learning_rate": 1.1807581475354119e-05,
+      "loss": 0.766,
+      "step": 589
+    },
+    {
+      "epoch": 1.496275162466318,
+      "grad_norm": 4.405285358428955,
+      "learning_rate": 1.1778506426928762e-05,
+      "loss": 0.752,
+      "step": 590
+    },
+    {
+      "epoch": 1.4988112220637184,
+      "grad_norm": 4.8693037033081055,
+      "learning_rate": 1.1749415844323193e-05,
+      "loss": 0.7662,
+      "step": 591
+    },
+    {
+      "epoch": 1.501347281661119,
+      "grad_norm": 4.855468273162842,
+      "learning_rate": 1.1720309981626093e-05,
+      "loss": 0.7917,
+      "step": 592
+    },
+    {
+      "epoch": 1.5038833412585195,
+      "grad_norm": 4.176314353942871,
+      "learning_rate": 1.16911890930596e-05,
+      "loss": 0.7833,
+      "step": 593
+    },
+    {
+      "epoch": 1.5064194008559202,
+      "grad_norm": 4.653109073638916,
+      "learning_rate": 1.1662053432977104e-05,
+      "loss": 0.7673,
+      "step": 594
+    },
+    {
+      "epoch": 1.5089554604533206,
+      "grad_norm": 4.294135570526123,
+      "learning_rate": 1.1632903255861015e-05,
+      "loss": 0.7574,
+      "step": 595
+    },
+    {
+      "epoch": 1.511491520050721,
+      "grad_norm": 4.264169692993164,
+      "learning_rate": 1.1603738816320531e-05,
+      "loss": 0.7775,
+      "step": 596
+    },
+    {
+      "epoch": 1.5140275796481217,
+      "grad_norm": 4.763448715209961,
+      "learning_rate": 1.1574560369089428e-05,
+      "loss": 0.7778,
+      "step": 597
+    },
+    {
+      "epoch": 1.5165636392455224,
+      "grad_norm": 4.264140605926514,
+      "learning_rate": 1.1545368169023831e-05,
+      "loss": 0.7498,
+      "step": 598
+    },
+    {
+      "epoch": 1.5190996988429228,
+      "grad_norm": 4.322105884552002,
+      "learning_rate": 1.1516162471099994e-05,
+      "loss": 0.7705,
+      "step": 599
+    },
+    {
+      "epoch": 1.5216357584403233,
+      "grad_norm": 4.198470592498779,
+      "learning_rate": 1.1486943530412063e-05,
+      "loss": 0.7957,
+      "step": 600
+    },
+    {
+      "epoch": 1.5216357584403233,
+      "eval_loss": NaN,
+      "eval_runtime": 206.594,
+      "eval_samples_per_second": 40.248,
+      "eval_steps_per_second": 5.034,
+      "step": 600
+    },
+    {
+      "epoch": 1.5241718180377237,
+      "grad_norm": 4.315439224243164,
+      "learning_rate": 1.145771160216984e-05,
+      "loss": 0.7765,
+      "step": 601
+    },
+    {
+      "epoch": 1.5267078776351244,
+      "grad_norm": 4.259639263153076,
+      "learning_rate": 1.1428466941696578e-05,
+      "loss": 0.7338,
+      "step": 602
+    },
+    {
+      "epoch": 1.529243937232525,
+      "grad_norm": 4.439230442047119,
+      "learning_rate": 1.139920980442674e-05,
+      "loss": 0.7615,
+      "step": 603
+    },
+    {
+      "epoch": 1.5317799968299255,
+      "grad_norm": 4.219053268432617,
+      "learning_rate": 1.136994044590376e-05,
+      "loss": 0.7774,
+      "step": 604
+    },
+    {
+      "epoch": 1.534316056427326,
+      "grad_norm": 4.197740077972412,
+      "learning_rate": 1.134065912177782e-05,
+      "loss": 0.7758,
+      "step": 605
+    },
+    {
+      "epoch": 1.5368521160247266,
+      "grad_norm": 5.02440881729126,
+      "learning_rate": 1.131136608780361e-05,
+      "loss": 0.7788,
+      "step": 606
+    },
+    {
+      "epoch": 1.5393881756221273,
+      "grad_norm": 4.4782867431640625,
+      "learning_rate": 1.1282061599838101e-05,
+      "loss": 0.7774,
+      "step": 607
+    },
+    {
+      "epoch": 1.5419242352195277,
+      "grad_norm": 5.940610885620117,
+      "learning_rate": 1.1252745913838315e-05,
+      "loss": 0.7669,
+      "step": 608
+    },
+    {
+      "epoch": 1.5444602948169281,
+      "grad_norm": 4.91011381149292,
+      "learning_rate": 1.122341928585907e-05,
+      "loss": 0.7916,
+      "step": 609
+    },
+    {
+      "epoch": 1.5469963544143286,
+      "grad_norm": 4.792825698852539,
+      "learning_rate": 1.1194081972050765e-05,
+      "loss": 0.7635,
+      "step": 610
+    },
+    {
+      "epoch": 1.5495324140117293,
+      "grad_norm": 5.109334945678711,
+      "learning_rate": 1.1164734228657127e-05,
+      "loss": 0.7681,
+      "step": 611
+    },
+    {
+      "epoch": 1.55206847360913,
+      "grad_norm": 6.440131187438965,
+      "learning_rate": 1.113537631201298e-05,
+      "loss": 0.7698,
+      "step": 612
+    },
+    {
+      "epoch": 1.5546045332065304,
+      "grad_norm": 6.067620277404785,
+      "learning_rate": 1.110600847854201e-05,
+      "loss": 0.783,
+      "step": 613
+    },
+    {
+      "epoch": 1.5571405928039308,
+      "grad_norm": 4.841549873352051,
+      "learning_rate": 1.107663098475452e-05,
+      "loss": 0.7806,
+      "step": 614
+    },
+    {
+      "epoch": 1.5596766524013315,
+      "grad_norm": 5.2280988693237305,
+      "learning_rate": 1.104724408724518e-05,
+      "loss": 0.7716,
+      "step": 615
+    },
+    {
+      "epoch": 1.5622127119987321,
+      "grad_norm": 4.952966690063477,
+      "learning_rate": 1.1017848042690804e-05,
+      "loss": 0.7702,
+      "step": 616
+    },
+    {
+      "epoch": 1.5647487715961326,
+      "grad_norm": 4.623702526092529,
+      "learning_rate": 1.0988443107848103e-05,
+      "loss": 0.7658,
+      "step": 617
+    },
+    {
+      "epoch": 1.567284831193533,
+      "grad_norm": 4.3320183753967285,
+      "learning_rate": 1.0959029539551436e-05,
+      "loss": 0.7808,
+      "step": 618
+    },
+    {
+      "epoch": 1.5698208907909335,
+      "grad_norm": 4.16457986831665,
+      "learning_rate": 1.0929607594710567e-05,
+      "loss": 0.7786,
+      "step": 619
+    },
+    {
+      "epoch": 1.5723569503883341,
+      "grad_norm": 4.45630407333374,
+      "learning_rate": 1.0900177530308427e-05,
+      "loss": 0.7843,
+      "step": 620
+    },
+    {
+      "epoch": 1.5748930099857348,
+      "grad_norm": 4.01480770111084,
+      "learning_rate": 1.0870739603398866e-05,
+      "loss": 0.7711,
+      "step": 621
+    },
+    {
+      "epoch": 1.5774290695831352,
+      "grad_norm": 4.143961429595947,
+      "learning_rate": 1.0841294071104407e-05,
+      "loss": 0.7685,
+      "step": 622
+    },
+    {
+      "epoch": 1.5799651291805357,
+      "grad_norm": 3.7771260738372803,
+      "learning_rate": 1.0811841190614003e-05,
+      "loss": 0.7565,
+      "step": 623
+    },
+    {
+      "epoch": 1.5825011887779363,
+      "grad_norm": 3.722019910812378,
+      "learning_rate": 1.0782381219180788e-05,
+      "loss": 0.784,
+      "step": 624
+    },
+    {
+      "epoch": 1.5850372483753368,
+      "grad_norm": 4.088432312011719,
+      "learning_rate": 1.0752914414119836e-05,
+      "loss": 0.7712,
+      "step": 625
+    },
+    {
+      "epoch": 1.5875733079727374,
+      "grad_norm": 4.5803985595703125,
+      "learning_rate": 1.07234410328059e-05,
+      "loss": 0.7514,
+      "step": 626
+    },
+    {
+      "epoch": 1.5901093675701379,
+      "grad_norm": 4.218702793121338,
+      "learning_rate": 1.0693961332671178e-05,
+      "loss": 0.7671,
+      "step": 627
+    },
+    {
+      "epoch": 1.5926454271675383,
+      "grad_norm": 4.5559163093566895,
+      "learning_rate": 1.066447557120306e-05,
+      "loss": 0.7898,
+      "step": 628
+    },
+    {
+      "epoch": 1.595181486764939,
+      "grad_norm": 5.258092880249023,
+      "learning_rate": 1.0634984005941875e-05,
+      "loss": 0.7966,
+      "step": 629
+    },
+    {
+      "epoch": 1.5977175463623396,
+      "grad_norm": 5.0127105712890625,
+      "learning_rate": 1.0605486894478644e-05,
+      "loss": 0.7682,
+      "step": 630
+    },
+    {
+      "epoch": 1.60025360595974,
+      "grad_norm": 4.945410251617432,
+      "learning_rate": 1.0575984494452833e-05,
+      "loss": 0.7556,
+      "step": 631
+    },
+    {
+      "epoch": 1.6027896655571405,
+      "grad_norm": 4.277896881103516,
+      "learning_rate": 1.0546477063550102e-05,
+      "loss": 0.7696,
+      "step": 632
+    },
+    {
+      "epoch": 1.605325725154541,
+      "grad_norm": 4.382696628570557,
+      "learning_rate": 1.0516964859500052e-05,
+      "loss": 0.7691,
+      "step": 633
+    },
+    {
+      "epoch": 1.6078617847519416,
+      "grad_norm": 4.495989799499512,
+      "learning_rate": 1.0487448140073967e-05,
+      "loss": 0.7751,
+      "step": 634
+    },
+    {
+      "epoch": 1.6103978443493423,
+      "grad_norm": 4.560750484466553,
+      "learning_rate": 1.0457927163082581e-05,
+      "loss": 0.7608,
+      "step": 635
+    },
+    {
+      "epoch": 1.6129339039467427,
+      "grad_norm": 3.86458683013916,
+      "learning_rate": 1.0428402186373812e-05,
+      "loss": 0.7863,
+      "step": 636
+    },
+    {
+      "epoch": 1.6154699635441432,
+      "grad_norm": 4.609869956970215,
+      "learning_rate": 1.0398873467830506e-05,
+      "loss": 0.7731,
+      "step": 637
+    },
+    {
+      "epoch": 1.6180060231415438,
+      "grad_norm": 4.20988130569458,
+      "learning_rate": 1.03693412653682e-05,
+      "loss": 0.7914,
+      "step": 638
+    },
+    {
+      "epoch": 1.6205420827389445,
+      "grad_norm": 4.387001991271973,
+      "learning_rate": 1.033980583693286e-05,
+      "loss": 0.7554,
+      "step": 639
+    },
+    {
+      "epoch": 1.623078142336345,
+      "grad_norm": 4.377809047698975,
+      "learning_rate": 1.0310267440498622e-05,
+      "loss": 0.7968,
+      "step": 640
+    },
+    {
+      "epoch": 1.6256142019337454,
+      "grad_norm": 5.118124485015869,
+      "learning_rate": 1.0280726334065558e-05,
+      "loss": 0.7924,
+      "step": 641
+    },
+    {
+      "epoch": 1.6281502615311458,
+      "grad_norm": 4.3000688552856445,
+      "learning_rate": 1.0251182775657403e-05,
+      "loss": 0.7731,
+      "step": 642
+    },
+    {
+      "epoch": 1.6306863211285465,
+      "grad_norm": 4.691357135772705,
+      "learning_rate": 1.02216370233193e-05,
+      "loss": 0.7711,
+      "step": 643
+    },
+    {
+      "epoch": 1.6332223807259472,
+      "grad_norm": 4.082294464111328,
+      "learning_rate": 1.0192089335115569e-05,
+      "loss": 0.7856,
+      "step": 644
+    },
+    {
+      "epoch": 1.6357584403233476,
+      "grad_norm": 4.61932373046875,
+      "learning_rate": 1.0162539969127432e-05,
+      "loss": 0.772,
+      "step": 645
+    },
+    {
+      "epoch": 1.638294499920748,
+      "grad_norm": 3.823538064956665,
+      "learning_rate": 1.013298918345077e-05,
+      "loss": 0.7636,
+      "step": 646
+    },
+    {
+      "epoch": 1.6408305595181487,
+      "grad_norm": 4.515269756317139,
+      "learning_rate": 1.0103437236193852e-05,
+      "loss": 0.7715,
+      "step": 647
+    },
+    {
+      "epoch": 1.6433666191155494,
+      "grad_norm": 4.864955902099609,
+      "learning_rate": 1.0073884385475107e-05,
+      "loss": 0.7546,
+      "step": 648
+    },
+    {
+      "epoch": 1.6459026787129498,
+      "grad_norm": 5.148191928863525,
+      "learning_rate": 1.0044330889420848e-05,
+      "loss": 0.787,
+      "step": 649
+    },
+    {
+      "epoch": 1.6484387383103503,
+      "grad_norm": 7.2982354164123535,
+      "learning_rate": 1.0014777006163028e-05,
+      "loss": 0.7877,
+      "step": 650
+    },
+    {
+      "epoch": 1.6484387383103503,
+      "eval_loss": NaN,
+      "eval_runtime": 206.9341,
+      "eval_samples_per_second": 40.182,
+      "eval_steps_per_second": 5.026,
+      "step": 650
+    },
+    {
+      "epoch": 1.6509747979077507,
+      "grad_norm": 7.431939125061035,
+      "learning_rate": 9.985222993836977e-06,
+      "loss": 0.7817,
+      "step": 651
+    },
+    {
+      "epoch": 1.6535108575051514,
+      "grad_norm": 6.551576614379883,
+      "learning_rate": 9.955669110579154e-06,
+      "loss": 0.774,
+      "step": 652
+    },
+    {
+      "epoch": 1.656046917102552,
+      "grad_norm": 7.8915228843688965,
+      "learning_rate": 9.926115614524894e-06,
+      "loss": 0.8103,
+      "step": 653
+    },
+    {
+      "epoch": 1.6585829766999525,
+      "grad_norm": 5.262277126312256,
+      "learning_rate": 9.896562763806152e-06,
+      "loss": 0.7629,
+      "step": 654
+    },
+    {
+      "epoch": 1.661119036297353,
+      "grad_norm": 6.142648696899414,
+      "learning_rate": 9.867010816549234e-06,
+      "loss": 0.755,
+      "step": 655
+    },
+    {
+      "epoch": 1.6636550958947536,
+      "grad_norm": 5.427359104156494,
+      "learning_rate": 9.837460030872571e-06,
+      "loss": 0.7973,
+      "step": 656
+    },
+    {
+      "epoch": 1.666191155492154,
+      "grad_norm": 5.329654216766357,
+      "learning_rate": 9.807910664884433e-06,
+      "loss": 0.786,
+      "step": 657
+    },
+    {
+      "epoch": 1.6687272150895547,
+      "grad_norm": 5.370080947875977,
+      "learning_rate": 9.778362976680703e-06,
+      "loss": 0.7451,
+      "step": 658
+    },
+    {
+      "epoch": 1.6712632746869551,
+      "grad_norm": 4.519922256469727,
+      "learning_rate": 9.748817224342604e-06,
+      "loss": 0.7914,
+      "step": 659
+    },
+    {
+      "epoch": 1.6737993342843556,
+      "grad_norm": 5.708288192749023,
+      "learning_rate": 9.719273665934442e-06,
+      "loss": 0.762,
+      "step": 660
+    },
+    {
+      "epoch": 1.6763353938817562,
+      "grad_norm": 4.4073872566223145,
+      "learning_rate": 9.68973255950138e-06,
+      "loss": 0.7749,
+      "step": 661
+    },
+    {
+      "epoch": 1.678871453479157,
+      "grad_norm": 5.624901294708252,
+      "learning_rate": 9.660194163067142e-06,
+      "loss": 0.7603,
+      "step": 662
+    },
+    {
+      "epoch": 1.6814075130765573,
+      "grad_norm": 5.436640739440918,
+      "learning_rate": 9.630658734631803e-06,
+      "loss": 0.7811,
+      "step": 663
+    },
+    {
+      "epoch": 1.6839435726739578,
+      "grad_norm": 5.290567398071289,
+      "learning_rate": 9.601126532169498e-06,
+      "loss": 0.7768,
+      "step": 664
+    },
+    {
+      "epoch": 1.6864796322713582,
+      "grad_norm": 5.254617214202881,
+      "learning_rate": 9.571597813626192e-06,
+      "loss": 0.767,
+      "step": 665
+    },
+    {
+      "epoch": 1.6890156918687589,
+      "grad_norm": 4.919734001159668,
+      "learning_rate": 9.54207283691742e-06,
+      "loss": 0.7822,
+      "step": 666
+    },
+    {
+      "epoch": 1.6915517514661595,
+      "grad_norm": 4.24241304397583,
+      "learning_rate": 9.512551859926038e-06,
+      "loss": 0.7681,
+      "step": 667
+    },
+    {
+      "epoch": 1.69408781106356,
+      "grad_norm": 8.8985013961792,
+      "learning_rate": 9.48303514049995e-06,
+      "loss": 0.7912,
+      "step": 668
+    },
+    {
+      "epoch": 1.6966238706609604,
+      "grad_norm": 8.58611011505127,
+      "learning_rate": 9.453522936449901e-06,
+      "loss": 0.7751,
+      "step": 669
+    },
+    {
+      "epoch": 1.699159930258361,
+      "grad_norm": 3.757329225540161,
+      "learning_rate": 9.424015505547165e-06,
+      "loss": 0.7738,
+      "step": 670
+    },
+    {
+      "epoch": 1.7016959898557618,
+      "grad_norm": 9.10895824432373,
+      "learning_rate": 9.394513105521359e-06,
+      "loss": 0.7627,
+      "step": 671
+    },
+    {
+      "epoch": 1.7042320494531622,
+      "grad_norm": 7.489527225494385,
+      "learning_rate": 9.36501599405813e-06,
+      "loss": 0.7682,
+      "step": 672
+    },
+    {
+      "epoch": 1.7067681090505626,
+      "grad_norm": 5.360051155090332,
+      "learning_rate": 9.335524428796941e-06,
+      "loss": 0.7727,
+      "step": 673
+    },
+    {
+      "epoch": 1.709304168647963,
+      "grad_norm": 6.477458953857422,
+      "learning_rate": 9.306038667328825e-06,
+      "loss": 0.7542,
+      "step": 674
+    },
+    {
+      "epoch": 1.7118402282453637,
+      "grad_norm": 5.043211936950684,
+      "learning_rate": 9.2765589671941e-06,
+      "loss": 0.7984,
+      "step": 675
+    },
+    {
+      "epoch": 1.7143762878427644,
+      "grad_norm": 5.121224403381348,
+      "learning_rate": 9.247085585880166e-06,
+      "loss": 0.7712,
+      "step": 676
+    },
+    {
+      "epoch": 1.7169123474401649,
+      "grad_norm": 5.69315767288208,
+      "learning_rate": 9.217618780819217e-06,
+      "loss": 0.7663,
+      "step": 677
+    },
+    {
+      "epoch": 1.7194484070375653,
+      "grad_norm": 4.174893379211426,
+      "learning_rate": 9.188158809386e-06,
+      "loss": 0.7664,
+      "step": 678
+    },
+    {
+      "epoch": 1.721984466634966,
+      "grad_norm": 5.404911518096924,
+      "learning_rate": 9.158705928895596e-06,
+      "loss": 0.7512,
+      "step": 679
+    },
+    {
+      "epoch": 1.7245205262323666,
+      "grad_norm": 4.873772621154785,
+      "learning_rate": 9.129260396601137e-06,
+      "loss": 0.7469,
+      "step": 680
+    },
+    {
+      "epoch": 1.727056585829767,
+      "grad_norm": 4.571645736694336,
+      "learning_rate": 9.099822469691576e-06,
+      "loss": 0.7858,
+      "step": 681
+    },
+    {
+      "epoch": 1.7295926454271675,
+      "grad_norm": 4.775193691253662,
+      "learning_rate": 9.070392405289436e-06,
+      "loss": 0.7759,
+      "step": 682
+    },
+    {
+      "epoch": 1.732128705024568,
+      "grad_norm": 4.750755786895752,
+      "learning_rate": 9.040970460448565e-06,
+      "loss": 0.7762,
+      "step": 683
+    },
+    {
+      "epoch": 1.7346647646219686,
+      "grad_norm": 4.8016252517700195,
+      "learning_rate": 9.011556892151899e-06,
+      "loss": 0.7637,
+      "step": 684
+    },
+    {
+      "epoch": 1.7372008242193693,
+      "grad_norm": 4.042902946472168,
+      "learning_rate": 8.982151957309196e-06,
+      "loss": 0.7916,
+      "step": 685
+    },
+    {
+      "epoch": 1.7397368838167697,
+      "grad_norm": 4.747760772705078,
+      "learning_rate": 8.952755912754825e-06,
+      "loss": 0.7603,
+      "step": 686
+    },
+    {
+      "epoch": 1.7422729434141702,
+      "grad_norm": 4.7758259773254395,
+      "learning_rate": 8.923369015245487e-06,
+      "loss": 0.7674,
+      "step": 687
+    },
+    {
+      "epoch": 1.7448090030115708,
+      "grad_norm": 4.502845764160156,
+      "learning_rate": 8.893991521457989e-06,
+      "loss": 0.7715,
+      "step": 688
+    },
+    {
+      "epoch": 1.7473450626089713,
+      "grad_norm": 5.1893815994262695,
+      "learning_rate": 8.864623687987022e-06,
+      "loss": 0.7826,
+      "step": 689
+    },
+    {
+      "epoch": 1.749881122206372,
+      "grad_norm": 5.01784610748291,
+      "learning_rate": 8.835265771342878e-06,
+      "loss": 0.7888,
+      "step": 690
+    },
+    {
+      "epoch": 1.7524171818037724,
+      "grad_norm": 4.870006561279297,
+      "learning_rate": 8.805918027949237e-06,
+      "loss": 0.7704,
+      "step": 691
+    },
+    {
+      "epoch": 1.7549532414011728,
+      "grad_norm": 4.26132345199585,
+      "learning_rate": 8.776580714140934e-06,
+      "loss": 0.7983,
+      "step": 692
+    },
+    {
+      "epoch": 1.7574893009985735,
+      "grad_norm": 5.457138538360596,
+      "learning_rate": 8.747254086161686e-06,
+      "loss": 0.7766,
+      "step": 693
+    },
+    {
+      "epoch": 1.7600253605959741,
+      "grad_norm": 4.7016682624816895,
+      "learning_rate": 8.717938400161902e-06,
+      "loss": 0.7881,
+      "step": 694
+    },
+    {
+      "epoch": 1.7625614201933746,
+      "grad_norm": 4.873697280883789,
+      "learning_rate": 8.688633912196395e-06,
+      "loss": 0.7767,
+      "step": 695
+    },
+    {
+      "epoch": 1.765097479790775,
+      "grad_norm": 5.145116806030273,
+      "learning_rate": 8.659340878222182e-06,
+      "loss": 0.801,
+      "step": 696
+    },
+    {
+      "epoch": 1.7676335393881755,
+      "grad_norm": 4.3196330070495605,
+      "learning_rate": 8.630059554096243e-06,
+      "loss": 0.7737,
+      "step": 697
+    },
+    {
+      "epoch": 1.7701695989855761,
+      "grad_norm": 4.999388217926025,
+      "learning_rate": 8.60079019557326e-06,
+      "loss": 0.7742,
+      "step": 698
+    },
+    {
+      "epoch": 1.7727056585829768,
+      "grad_norm": 4.529718399047852,
+      "learning_rate": 8.571533058303424e-06,
+      "loss": 0.8007,
+      "step": 699
+    },
+    {
+      "epoch": 1.7752417181803772,
+      "grad_norm": 4.1417388916015625,
+      "learning_rate": 8.542288397830166e-06,
+      "loss": 0.7655,
+      "step": 700
+    },
+    {
+      "epoch": 1.7752417181803772,
+      "eval_loss": NaN,
+      "eval_runtime": 207.4533,
+      "eval_samples_per_second": 40.081,
+      "eval_steps_per_second": 5.013,
+      "step": 700
+    },
+    {
+      "epoch": 1.7777777777777777,
+      "grad_norm": 4.700773239135742,
+      "learning_rate": 8.513056469587942e-06,
+      "loss": 0.7783,
+      "step": 701
+    },
+    {
+      "epoch": 1.7803138373751783,
+      "grad_norm": 4.4216389656066895,
+      "learning_rate": 8.483837528900008e-06,
+      "loss": 0.7911,
+      "step": 702
+    },
+    {
+      "epoch": 1.782849896972579,
+      "grad_norm": 4.117276191711426,
+      "learning_rate": 8.454631830976167e-06,
+      "loss": 0.774,
+      "step": 703
+    },
+    {
+      "epoch": 1.7853859565699794,
+      "grad_norm": 4.172786235809326,
+      "learning_rate": 8.425439630910575e-06,
+      "loss": 0.7773,
+      "step": 704
+    },
+    {
+      "epoch": 1.78792201616738,
+      "grad_norm": 5.070644855499268,
+      "learning_rate": 8.396261183679475e-06,
+      "loss": 0.7856,
+      "step": 705
+    },
+    {
+      "epoch": 1.7904580757647803,
+      "grad_norm": 4.331361770629883,
+      "learning_rate": 8.367096744138985e-06,
+      "loss": 0.7781,
+      "step": 706
+    },
+    {
+      "epoch": 1.792994135362181,
+      "grad_norm": 4.1829071044921875,
+      "learning_rate": 8.337946567022897e-06,
+      "loss": 0.7792,
+      "step": 707
+    },
+    {
+      "epoch": 1.7955301949595817,
+      "grad_norm": 4.379103183746338,
+      "learning_rate": 8.3088109069404e-06,
+      "loss": 0.7881,
+      "step": 708
+    },
+    {
+      "epoch": 1.798066254556982,
+      "grad_norm": 4.556863784790039,
+      "learning_rate": 8.279690018373912e-06,
+      "loss": 0.8164,
+      "step": 709
+    },
+    {
+      "epoch": 1.8006023141543825,
+      "grad_norm": 4.11606502532959,
+      "learning_rate": 8.250584155676809e-06,
+      "loss": 0.7684,
+      "step": 710
+    },
+    {
+      "epoch": 1.8031383737517832,
+      "grad_norm": 4.723000526428223,
+      "learning_rate": 8.221493573071238e-06,
+      "loss": 0.7792,
+      "step": 711
+    },
+    {
+      "epoch": 1.8056744333491839,
+      "grad_norm": 4.217459678649902,
+      "learning_rate": 8.192418524645884e-06,
+      "loss": 0.7793,
+      "step": 712
+    },
+    {
+      "epoch": 1.8082104929465843,
+      "grad_norm": 8.686624526977539,
+      "learning_rate": 8.163359264353745e-06,
+      "loss": 0.7862,
+      "step": 713
+    },
+    {
+      "epoch": 1.8107465525439848,
+      "grad_norm": 4.927021026611328,
+      "learning_rate": 8.134316046009915e-06,
+      "loss": 0.7555,
+      "step": 714
+    },
+    {
+      "epoch": 1.8132826121413852,
+      "grad_norm": 7.500072002410889,
+      "learning_rate": 8.105289123289384e-06,
+      "loss": 0.7688,
+      "step": 715
+    },
+    {
+      "epoch": 1.8158186717387859,
+      "grad_norm": 6.166264533996582,
+      "learning_rate": 8.076278749724797e-06,
+      "loss": 0.7882,
+      "step": 716
+    },
+    {
+      "epoch": 1.8183547313361865,
+      "grad_norm": 4.8428263664245605,
+      "learning_rate": 8.047285178704266e-06,
+      "loss": 0.7562,
+      "step": 717
+    },
+    {
+      "epoch": 1.820890790933587,
+      "grad_norm": 6.607320785522461,
+      "learning_rate": 8.018308663469128e-06,
+      "loss": 0.7855,
+      "step": 718
+    },
+    {
+      "epoch": 1.8234268505309874,
+      "grad_norm": 4.755509853363037,
+      "learning_rate": 7.989349457111754e-06,
+      "loss": 0.7797,
+      "step": 719
+    },
+    {
+      "epoch": 1.825962910128388,
+      "grad_norm": 4.9010515213012695,
+      "learning_rate": 7.960407812573338e-06,
+      "loss": 0.7722,
+      "step": 720
+    },
+    {
+      "epoch": 1.8284989697257885,
+      "grad_norm": 5.554948806762695,
+      "learning_rate": 7.93148398264167e-06,
+      "loss": 0.7609,
+      "step": 721
+    },
+    {
+      "epoch": 1.8310350293231892,
+      "grad_norm": 4.208399295806885,
+      "learning_rate": 7.902578219948958e-06,
+      "loss": 0.7903,
+      "step": 722
+    },
+    {
+      "epoch": 1.8335710889205896,
+      "grad_norm": 4.844840049743652,
+      "learning_rate": 7.873690776969583e-06,
+      "loss": 0.7646,
+      "step": 723
+    },
+    {
+      "epoch": 1.83610714851799,
+      "grad_norm": 4.84792947769165,
+      "learning_rate": 7.844821906017929e-06,
+      "loss": 0.7582,
+      "step": 724
+    },
+    {
+      "epoch": 1.8386432081153907,
+      "grad_norm": 4.329719066619873,
+      "learning_rate": 7.815971859246158e-06,
+      "loss": 0.8061,
+      "step": 725
+    },
+    {
+      "epoch": 1.8411792677127914,
+      "grad_norm": 4.478653430938721,
+      "learning_rate": 7.787140888642018e-06,
+      "loss": 0.7984,
+      "step": 726
+    },
+    {
+      "epoch": 1.8437153273101918,
+      "grad_norm": 4.262007713317871,
+      "learning_rate": 7.758329246026633e-06,
+      "loss": 0.7657,
+      "step": 727
+    },
+    {
+      "epoch": 1.8462513869075923,
+      "grad_norm": 4.718945026397705,
+      "learning_rate": 7.729537183052318e-06,
+      "loss": 0.7431,
+      "step": 728
+    },
+    {
+      "epoch": 1.8487874465049927,
+      "grad_norm": 4.219466686248779,
+      "learning_rate": 7.700764951200357e-06,
+      "loss": 0.7745,
+      "step": 729
+    },
+    {
+      "epoch": 1.8513235061023934,
+      "grad_norm": 3.950295925140381,
+      "learning_rate": 7.672012801778838e-06,
+      "loss": 0.7736,
+      "step": 730
+    },
+    {
+      "epoch": 1.853859565699794,
+      "grad_norm": 4.399406433105469,
+      "learning_rate": 7.643280985920423e-06,
+      "loss": 0.7827,
+      "step": 731
+    },
+    {
+      "epoch": 1.8563956252971945,
+      "grad_norm": 3.9940927028656006,
+      "learning_rate": 7.614569754580195e-06,
+      "loss": 0.8065,
+      "step": 732
+    },
+    {
+      "epoch": 1.858931684894595,
+      "grad_norm": 4.296957015991211,
+      "learning_rate": 7.585879358533422e-06,
+      "loss": 0.7829,
+      "step": 733
+    },
+    {
+      "epoch": 1.8614677444919956,
+      "grad_norm": 6.42236328125,
+      "learning_rate": 7.557210048373395e-06,
+      "loss": 0.783,
+      "step": 734
+    },
+    {
+      "epoch": 1.8640038040893963,
+      "grad_norm": 5.146311283111572,
+      "learning_rate": 7.528562074509244e-06,
+      "loss": 0.7513,
+      "step": 735
+    },
+    {
+      "epoch": 1.8665398636867967,
+      "grad_norm": 6.278244972229004,
+      "learning_rate": 7.499935687163724e-06,
+      "loss": 0.7536,
+      "step": 736
+    },
+    {
+      "epoch": 1.8690759232841971,
+      "grad_norm": 7.04921817779541,
+      "learning_rate": 7.471331136371051e-06,
+      "loss": 0.776,
+      "step": 737
+    },
+    {
+      "epoch": 1.8716119828815976,
+      "grad_norm": 5.505727767944336,
+      "learning_rate": 7.442748671974712e-06,
+      "loss": 0.7661,
+      "step": 738
+    },
+    {
+      "epoch": 1.8741480424789982,
+      "grad_norm": 6.636948108673096,
+      "learning_rate": 7.4141885436252845e-06,
+      "loss": 0.7818,
+      "step": 739
+    },
+    {
+      "epoch": 1.876684102076399,
+      "grad_norm": 7.15848445892334,
+      "learning_rate": 7.38565100077825e-06,
+      "loss": 0.7732,
+      "step": 740
+    },
+    {
+      "epoch": 1.8792201616737993,
+      "grad_norm": 4.574160099029541,
+      "learning_rate": 7.3571362926918245e-06,
+      "loss": 0.7749,
+      "step": 741
+    },
+    {
+      "epoch": 1.8817562212711998,
+      "grad_norm": 5.277390956878662,
+      "learning_rate": 7.328644668424766e-06,
+      "loss": 0.7661,
+      "step": 742
+    },
+    {
+      "epoch": 1.8842922808686005,
+      "grad_norm": 5.560710906982422,
+      "learning_rate": 7.300176376834225e-06,
+      "loss": 0.7744,
+      "step": 743
+    },
+    {
+      "epoch": 1.886828340466001,
+      "grad_norm": 3.971520185470581,
+      "learning_rate": 7.271731666573538e-06,
+      "loss": 0.7582,
+      "step": 744
+    },
+    {
+      "epoch": 1.8893644000634016,
+      "grad_norm": 6.022958278656006,
+      "learning_rate": 7.24331078609009e-06,
+      "loss": 0.7953,
+      "step": 745
+    },
+    {
+      "epoch": 1.891900459660802,
+      "grad_norm": 5.177083492279053,
+      "learning_rate": 7.214913983623113e-06,
+      "loss": 0.7811,
+      "step": 746
+    },
+    {
+      "epoch": 1.8944365192582024,
+      "grad_norm": 4.187081813812256,
+      "learning_rate": 7.186541507201535e-06,
+      "loss": 0.7764,
+      "step": 747
+    },
+    {
+      "epoch": 1.896972578855603,
+      "grad_norm": 4.742189884185791,
+      "learning_rate": 7.158193604641821e-06,
+      "loss": 0.7806,
+      "step": 748
+    },
+    {
+      "epoch": 1.8995086384530038,
+      "grad_norm": 4.245124340057373,
+      "learning_rate": 7.129870523545781e-06,
+      "loss": 0.7604,
+      "step": 749
+    },
+    {
+      "epoch": 1.9020446980504042,
+      "grad_norm": 4.04760217666626,
+      "learning_rate": 7.101572511298443e-06,
+      "loss": 0.7826,
+      "step": 750
+    },
+    {
+      "epoch": 1.9020446980504042,
+      "eval_loss": NaN,
+      "eval_runtime": 206.4955,
+      "eval_samples_per_second": 40.267,
+      "eval_steps_per_second": 5.036,
+      "step": 750
+    },
+    {
+      "epoch": 1.9045807576478047,
+      "grad_norm": 4.356484413146973,
+      "learning_rate": 7.07329981506586e-06,
+      "loss": 0.7727,
+      "step": 751
+    },
+    {
+      "epoch": 1.907116817245205,
+      "grad_norm": 4.629985332489014,
+      "learning_rate": 7.045052681792968e-06,
+      "loss": 0.7594,
+      "step": 752
+    },
+    {
+      "epoch": 1.9096528768426058,
+      "grad_norm": 4.18834924697876,
+      "learning_rate": 7.0168313582014294e-06,
+      "loss": 0.7632,
+      "step": 753
+    },
+    {
+      "epoch": 1.9121889364400064,
+      "grad_norm": 4.285244464874268,
+      "learning_rate": 6.988636090787473e-06,
+      "loss": 0.7815,
+      "step": 754
+    },
+    {
+      "epoch": 1.9147249960374069,
+      "grad_norm": 4.316561222076416,
+      "learning_rate": 6.960467125819741e-06,
+      "loss": 0.7851,
+      "step": 755
+    },
+    {
+      "epoch": 1.9172610556348073,
+      "grad_norm": 5.78687858581543,
+      "learning_rate": 6.932324709337146e-06,
+      "loss": 0.7992,
+      "step": 756
+    },
+    {
+      "epoch": 1.919797115232208,
+      "grad_norm": 5.057698726654053,
+      "learning_rate": 6.9042090871467005e-06,
+      "loss": 0.7716,
+      "step": 757
+    },
+    {
+      "epoch": 1.9223331748296086,
+      "grad_norm": 4.4669928550720215,
+      "learning_rate": 6.876120504821404e-06,
+      "loss": 0.7654,
+      "step": 758
+    },
+    {
+      "epoch": 1.924869234427009,
+      "grad_norm": 5.319240093231201,
+      "learning_rate": 6.848059207698069e-06,
+      "loss": 0.7461,
+      "step": 759
+    },
+    {
+      "epoch": 1.9274052940244095,
+      "grad_norm": 4.797286510467529,
+      "learning_rate": 6.820025440875182e-06,
+      "loss": 0.7848,
+      "step": 760
+    },
+    {
+      "epoch": 1.92994135362181,
+      "grad_norm": 4.8291015625,
+      "learning_rate": 6.792019449210787e-06,
+      "loss": 0.7579,
+      "step": 761
+    },
+    {
+      "epoch": 1.9324774132192106,
+      "grad_norm": 4.524660587310791,
+      "learning_rate": 6.764041477320307e-06,
+      "loss": 0.7614,
+      "step": 762
+    },
+    {
+      "epoch": 1.9350134728166113,
+      "grad_norm": 4.612803936004639,
+      "learning_rate": 6.736091769574451e-06,
+      "loss": 0.7865,
+      "step": 763
+    },
+    {
+      "epoch": 1.9375495324140117,
+      "grad_norm": 4.612975120544434,
+      "learning_rate": 6.708170570097041e-06,
+      "loss": 0.762,
+      "step": 764
+    },
+    {
+      "epoch": 1.9400855920114122,
+      "grad_norm": 4.094751834869385,
+      "learning_rate": 6.6802781227629e-06,
+      "loss": 0.7777,
+      "step": 765
+    },
+    {
+      "epoch": 1.9426216516088128,
+      "grad_norm": 4.402510166168213,
+      "learning_rate": 6.652414671195726e-06,
+      "loss": 0.7786,
+      "step": 766
+    },
+    {
+      "epoch": 1.9451577112062135,
+      "grad_norm": 4.165380477905273,
+      "learning_rate": 6.624580458765949e-06,
+      "loss": 0.792,
+      "step": 767
+    },
+    {
+      "epoch": 1.947693770803614,
+      "grad_norm": 4.46851110458374,
+      "learning_rate": 6.596775728588616e-06,
+      "loss": 0.7779,
+      "step": 768
+    },
+    {
+      "epoch": 1.9502298304010144,
+      "grad_norm": 4.09420919418335,
+      "learning_rate": 6.569000723521266e-06,
+      "loss": 0.7596,
+      "step": 769
+    },
+    {
+      "epoch": 1.9527658899984148,
+      "grad_norm": 4.088312149047852,
+      "learning_rate": 6.541255686161796e-06,
+      "loss": 0.7808,
+      "step": 770
+    },
+    {
+      "epoch": 1.9553019495958155,
+      "grad_norm": 4.014118671417236,
+      "learning_rate": 6.5135408588463724e-06,
+      "loss": 0.782,
+      "step": 771
+    },
+    {
+      "epoch": 1.9578380091932162,
+      "grad_norm": 3.733555793762207,
+      "learning_rate": 6.4858564836472744e-06,
+      "loss": 0.7908,
+      "step": 772
+    },
+    {
+      "epoch": 1.9603740687906166,
+      "grad_norm": 4.4494709968566895,
+      "learning_rate": 6.458202802370821e-06,
+      "loss": 0.7577,
+      "step": 773
+    },
+    {
+      "epoch": 1.962910128388017,
+      "grad_norm": 4.325969696044922,
+      "learning_rate": 6.43058005655522e-06,
+      "loss": 0.7952,
+      "step": 774
+    },
+    {
+      "epoch": 1.9654461879854177,
+      "grad_norm": 5.431465148925781,
+      "learning_rate": 6.4029884874684835e-06,
+      "loss": 0.7833,
+      "step": 775
+    },
+    {
+      "epoch": 1.9679822475828181,
+      "grad_norm": 5.4291181564331055,
+      "learning_rate": 6.37542833610632e-06,
+      "loss": 0.801,
+      "step": 776
+    },
+    {
+      "epoch": 1.9705183071802188,
+      "grad_norm": 4.234609603881836,
+      "learning_rate": 6.347899843190012e-06,
+      "loss": 0.7989,
+      "step": 777
+    },
+    {
+      "epoch": 1.9730543667776193,
+      "grad_norm": 4.783573627471924,
+      "learning_rate": 6.320403249164335e-06,
+      "loss": 0.7722,
+      "step": 778
+    },
+    {
+      "epoch": 1.9755904263750197,
+      "grad_norm": 5.321080207824707,
+      "learning_rate": 6.292938794195442e-06,
+      "loss": 0.765,
+      "step": 779
+    },
+    {
+      "epoch": 1.9781264859724204,
+      "grad_norm": 4.841993808746338,
+      "learning_rate": 6.265506718168765e-06,
+      "loss": 0.7924,
+      "step": 780
+    },
+    {
+      "epoch": 1.980662545569821,
+      "grad_norm": 5.188013076782227,
+      "learning_rate": 6.238107260686939e-06,
+      "loss": 0.795,
+      "step": 781
+    },
+    {
+      "epoch": 1.9831986051672215,
+      "grad_norm": 4.518028736114502,
+      "learning_rate": 6.21074066106768e-06,
+      "loss": 0.7666,
+      "step": 782
+    },
+    {
+      "epoch": 1.985734664764622,
+      "grad_norm": 4.010153293609619,
+      "learning_rate": 6.1834071583417235e-06,
+      "loss": 0.7519,
+      "step": 783
+    },
+    {
+      "epoch": 1.9882707243620223,
+      "grad_norm": 4.495395183563232,
+      "learning_rate": 6.156106991250721e-06,
+      "loss": 0.7809,
+      "step": 784
+    },
+    {
+      "epoch": 1.990806783959423,
+      "grad_norm": 4.426692008972168,
+      "learning_rate": 6.128840398245147e-06,
+      "loss": 0.7829,
+      "step": 785
+    },
+    {
+      "epoch": 1.9933428435568237,
+      "grad_norm": 3.9877264499664307,
+      "learning_rate": 6.101607617482244e-06,
+      "loss": 0.7804,
+      "step": 786
+    },
+    {
+      "epoch": 1.9958789031542241,
+      "grad_norm": 4.457921028137207,
+      "learning_rate": 6.074408886823911e-06,
+      "loss": 0.7627,
+      "step": 787
+    },
+    {
+      "epoch": 1.9984149627516246,
+      "grad_norm": 4.236499786376953,
+      "learning_rate": 6.047244443834639e-06,
+      "loss": 0.7767,
+      "step": 788
+    },
+    {
+      "epoch": 2.000951022349025,
+      "grad_norm": 4.1788330078125,
+      "learning_rate": 6.020114525779453e-06,
+      "loss": 0.782,
+      "step": 789
+    },
+    {
+      "epoch": 2.003487081946426,
+      "grad_norm": 4.259964942932129,
+      "learning_rate": 5.993019369621802e-06,
+      "loss": 0.7523,
+      "step": 790
+    },
+    {
+      "epoch": 2.0060231415438263,
+      "grad_norm": 4.062742710113525,
+      "learning_rate": 5.965959212021531e-06,
+      "loss": 0.7573,
+      "step": 791
+    },
+    {
+      "epoch": 2.0085592011412268,
+      "grad_norm": 4.349952220916748,
+      "learning_rate": 5.938934289332781e-06,
+      "loss": 0.7566,
+      "step": 792
+    },
+    {
+      "epoch": 2.011095260738627,
+      "grad_norm": 4.47599458694458,
+      "learning_rate": 5.911944837601935e-06,
+      "loss": 0.772,
+      "step": 793
+    },
+    {
+      "epoch": 2.013631320336028,
+      "grad_norm": 4.20798921585083,
+      "learning_rate": 5.884991092565574e-06,
+      "loss": 0.8088,
+      "step": 794
+    },
+    {
+      "epoch": 2.0161673799334285,
+      "grad_norm": 5.474221706390381,
+      "learning_rate": 5.858073289648382e-06,
+      "loss": 0.7503,
+      "step": 795
+    },
+    {
+      "epoch": 2.018703439530829,
+      "grad_norm": 5.710548400878906,
+      "learning_rate": 5.831191663961132e-06,
+      "loss": 0.7591,
+      "step": 796
+    },
+    {
+      "epoch": 2.0212394991282294,
+      "grad_norm": 5.0567216873168945,
+      "learning_rate": 5.804346450298599e-06,
+      "loss": 0.7462,
+      "step": 797
+    },
+    {
+      "epoch": 2.02377555872563,
+      "grad_norm": 6.114500045776367,
+      "learning_rate": 5.777537883137507e-06,
+      "loss": 0.7874,
+      "step": 798
+    },
+    {
+      "epoch": 2.0263116183230307,
+      "grad_norm": 4.301583766937256,
+      "learning_rate": 5.750766196634532e-06,
+      "loss": 0.7825,
+      "step": 799
+    },
+    {
+      "epoch": 2.028847677920431,
+      "grad_norm": 6.107265949249268,
+      "learning_rate": 5.724031624624175e-06,
+      "loss": 0.777,
+      "step": 800
+    },
+    {
+      "epoch": 2.028847677920431,
+      "eval_loss": NaN,
+      "eval_runtime": 206.9019,
+      "eval_samples_per_second": 40.188,
+      "eval_steps_per_second": 5.027,
+      "step": 800
+    },
+    {
+      "epoch": 2.0313837375178316,
+      "grad_norm": 5.1263017654418945,
+      "learning_rate": 5.697334400616808e-06,
+      "loss": 0.7751,
+      "step": 801
+    },
+    {
+      "epoch": 2.033919797115232,
+      "grad_norm": 6.007981300354004,
+      "learning_rate": 5.6706747577965544e-06,
+      "loss": 0.7629,
+      "step": 802
+    },
+    {
+      "epoch": 2.036455856712633,
+      "grad_norm": 7.525304317474365,
+      "learning_rate": 5.644052929019308e-06,
+      "loss": 0.7783,
+      "step": 803
+    },
+    {
+      "epoch": 2.0389919163100334,
+      "grad_norm": 5.56154727935791,
+      "learning_rate": 5.617469146810677e-06,
+      "loss": 0.7935,
+      "step": 804
+    },
+    {
+      "epoch": 2.041527975907434,
+      "grad_norm": 6.130625247955322,
+      "learning_rate": 5.590923643363953e-06,
+      "loss": 0.755,
+      "step": 805
+    },
+    {
+      "epoch": 2.0440640355048343,
+      "grad_norm": 5.978578567504883,
+      "learning_rate": 5.564416650538086e-06,
+      "loss": 0.7853,
+      "step": 806
+    },
+    {
+      "epoch": 2.0466000951022347,
+      "grad_norm": 5.146383762359619,
+      "learning_rate": 5.5379483998556615e-06,
+      "loss": 0.7612,
+      "step": 807
+    },
+    {
+      "epoch": 2.0491361546996356,
+      "grad_norm": 5.653777122497559,
+      "learning_rate": 5.5115191225008745e-06,
+      "loss": 0.7741,
+      "step": 808
+    },
+    {
+      "epoch": 2.051672214297036,
+      "grad_norm": 4.839360237121582,
+      "learning_rate": 5.48512904931751e-06,
+      "loss": 0.7628,
+      "step": 809
+    },
+    {
+      "epoch": 2.0542082738944365,
+      "grad_norm": 4.505990505218506,
+      "learning_rate": 5.458778410806933e-06,
+      "loss": 0.7951,
+      "step": 810
+    },
+    {
+      "epoch": 2.056744333491837,
+      "grad_norm": 4.722829341888428,
+      "learning_rate": 5.432467437126057e-06,
+      "loss": 0.771,
+      "step": 811
+    },
+    {
+      "epoch": 2.0592803930892374,
+      "grad_norm": 3.994661569595337,
+      "learning_rate": 5.406196358085371e-06,
+      "loss": 0.7596,
+      "step": 812
+    },
+    {
+      "epoch": 2.0618164526866383,
+      "grad_norm": 4.28508996963501,
+      "learning_rate": 5.379965403146881e-06,
+      "loss": 0.7867,
+      "step": 813
+    },
+    {
+      "epoch": 2.0643525122840387,
+      "grad_norm": 4.682456016540527,
+      "learning_rate": 5.353774801422158e-06,
+      "loss": 0.7838,
+      "step": 814
+    },
+    {
+      "epoch": 2.066888571881439,
+      "grad_norm": 4.162257194519043,
+      "learning_rate": 5.327624781670293e-06,
+      "loss": 0.7858,
+      "step": 815
+    },
+    {
+      "epoch": 2.0694246314788396,
+      "grad_norm": 3.9294047355651855,
+      "learning_rate": 5.301515572295926e-06,
+      "loss": 0.7613,
+      "step": 816
+    },
+    {
+      "epoch": 2.0719606910762405,
+      "grad_norm": 4.095000743865967,
+      "learning_rate": 5.275447401347247e-06,
+      "loss": 0.7693,
+      "step": 817
+    },
+    {
+      "epoch": 2.074496750673641,
+      "grad_norm": 4.737927436828613,
+      "learning_rate": 5.2494204965139905e-06,
+      "loss": 0.7555,
+      "step": 818
+    },
+    {
+      "epoch": 2.0770328102710414,
+      "grad_norm": 4.214750289916992,
+      "learning_rate": 5.223435085125466e-06,
+      "loss": 0.7692,
+      "step": 819
+    },
+    {
+      "epoch": 2.079568869868442,
+      "grad_norm": 5.1862006187438965,
+      "learning_rate": 5.197491394148556e-06,
+      "loss": 0.7745,
+      "step": 820
+    },
+    {
+      "epoch": 2.0821049294658422,
+      "grad_norm": 4.2943644523620605,
+      "learning_rate": 5.171589650185744e-06,
+      "loss": 0.7901,
+      "step": 821
+    },
+    {
+      "epoch": 2.084640989063243,
+      "grad_norm": 4.665522575378418,
+      "learning_rate": 5.145730079473128e-06,
+      "loss": 0.7694,
+      "step": 822
+    },
+    {
+      "epoch": 2.0871770486606436,
+      "grad_norm": 4.158400058746338,
+      "learning_rate": 5.119912907878455e-06,
+      "loss": 0.7719,
+      "step": 823
+    },
+    {
+      "epoch": 2.089713108258044,
+      "grad_norm": 4.519285202026367,
+      "learning_rate": 5.0941383608991326e-06,
+      "loss": 0.7718,
+      "step": 824
+    },
+    {
+      "epoch": 2.0922491678554445,
+      "grad_norm": 4.45874547958374,
+      "learning_rate": 5.068406663660277e-06,
+      "loss": 0.7724,
+      "step": 825
+    },
+    {
+      "epoch": 2.0947852274528453,
+      "grad_norm": 4.540599346160889,
+      "learning_rate": 5.042718040912722e-06,
+      "loss": 0.7566,
+      "step": 826
+    },
+    {
+      "epoch": 2.097321287050246,
+      "grad_norm": 4.259554862976074,
+      "learning_rate": 5.0170727170310995e-06,
+      "loss": 0.7656,
+      "step": 827
+    },
+    {
+      "epoch": 2.0998573466476462,
+      "grad_norm": 4.207907199859619,
+      "learning_rate": 4.9914709160118216e-06,
+      "loss": 0.7588,
+      "step": 828
+    },
+    {
+      "epoch": 2.1023934062450467,
+      "grad_norm": 4.362234115600586,
+      "learning_rate": 4.965912861471174e-06,
+      "loss": 0.7696,
+      "step": 829
+    },
+    {
+      "epoch": 2.104929465842447,
+      "grad_norm": 3.852293014526367,
+      "learning_rate": 4.940398776643336e-06,
+      "loss": 0.774,
+      "step": 830
+    },
+    {
+      "epoch": 2.107465525439848,
+      "grad_norm": 3.9607598781585693,
+      "learning_rate": 4.914928884378439e-06,
+      "loss": 0.7783,
+      "step": 831
+    },
+    {
+      "epoch": 2.1100015850372484,
+      "grad_norm": 4.0023016929626465,
+      "learning_rate": 4.889503407140622e-06,
+      "loss": 0.7529,
+      "step": 832
+    },
+    {
+      "epoch": 2.112537644634649,
+      "grad_norm": 4.012899875640869,
+      "learning_rate": 4.86412256700608e-06,
+      "loss": 0.7768,
+      "step": 833
+    },
+    {
+      "epoch": 2.1150737042320493,
+      "grad_norm": 3.976773738861084,
+      "learning_rate": 4.838786585661135e-06,
+      "loss": 0.7852,
+      "step": 834
+    },
+    {
+      "epoch": 2.11760976382945,
+      "grad_norm": 3.8476829528808594,
+      "learning_rate": 4.8134956844002914e-06,
+      "loss": 0.7799,
+      "step": 835
+    },
+    {
+      "epoch": 2.1201458234268507,
+      "grad_norm": 4.0478620529174805,
+      "learning_rate": 4.788250084124307e-06,
+      "loss": 0.7534,
+      "step": 836
+    },
+    {
+      "epoch": 2.122681883024251,
+      "grad_norm": 5.393147945404053,
+      "learning_rate": 4.763050005338264e-06,
+      "loss": 0.7848,
+      "step": 837
+    },
+    {
+      "epoch": 2.1252179426216515,
+      "grad_norm": 6.066077709197998,
+      "learning_rate": 4.737895668149644e-06,
+      "loss": 0.7586,
+      "step": 838
+    },
+    {
+      "epoch": 2.127754002219052,
+      "grad_norm": 4.707492828369141,
+      "learning_rate": 4.712787292266391e-06,
+      "loss": 0.7791,
+      "step": 839
+    },
+    {
+      "epoch": 2.130290061816453,
+      "grad_norm": 4.887709140777588,
+      "learning_rate": 4.687725096995026e-06,
+      "loss": 0.7703,
+      "step": 840
+    },
+    {
+      "epoch": 2.1328261214138533,
+      "grad_norm": 4.583090305328369,
+      "learning_rate": 4.662709301238686e-06,
+      "loss": 0.7526,
+      "step": 841
+    },
+    {
+      "epoch": 2.1353621810112537,
+      "grad_norm": 5.005523204803467,
+      "learning_rate": 4.6377401234952615e-06,
+      "loss": 0.7887,
+      "step": 842
+    },
+    {
+      "epoch": 2.137898240608654,
+      "grad_norm": 5.198577404022217,
+      "learning_rate": 4.612817781855439e-06,
+      "loss": 0.7777,
+      "step": 843
+    },
+    {
+      "epoch": 2.1404343002060546,
+      "grad_norm": 4.368279457092285,
+      "learning_rate": 4.587942494000835e-06,
+      "loss": 0.7704,
+      "step": 844
+    },
+    {
+      "epoch": 2.1429703598034555,
+      "grad_norm": 4.274848937988281,
+      "learning_rate": 4.563114477202076e-06,
+      "loss": 0.7718,
+      "step": 845
+    },
+    {
+      "epoch": 2.145506419400856,
+      "grad_norm": 4.577635288238525,
+      "learning_rate": 4.538333948316902e-06,
+      "loss": 0.7741,
+      "step": 846
+    },
+    {
+      "epoch": 2.1480424789982564,
+      "grad_norm": 4.26487922668457,
+      "learning_rate": 4.513601123788278e-06,
+      "loss": 0.7831,
+      "step": 847
+    },
+    {
+      "epoch": 2.150578538595657,
+      "grad_norm": 4.352411270141602,
+      "learning_rate": 4.488916219642498e-06,
+      "loss": 0.7614,
+      "step": 848
+    },
+    {
+      "epoch": 2.1531145981930577,
+      "grad_norm": 4.089112758636475,
+      "learning_rate": 4.464279451487301e-06,
+      "loss": 0.7718,
+      "step": 849
+    },
+    {
+      "epoch": 2.155650657790458,
+      "grad_norm": 4.205545902252197,
+      "learning_rate": 4.439691034509986e-06,
+      "loss": 0.7652,
+      "step": 850
+    },
+    {
+      "epoch": 2.155650657790458,
+      "eval_loss": NaN,
+      "eval_runtime": 207.3078,
+      "eval_samples_per_second": 40.109,
+      "eval_steps_per_second": 5.017,
+      "step": 850
+    },
+    {
+      "epoch": 2.1581867173878586,
+      "grad_norm": 4.343478202819824,
+      "learning_rate": 4.415151183475542e-06,
+      "loss": 0.7821,
+      "step": 851
+    },
+    {
+      "epoch": 2.160722776985259,
+      "grad_norm": 4.295416355133057,
+      "learning_rate": 4.390660112724742e-06,
+      "loss": 0.7875,
+      "step": 852
+    },
+    {
+      "epoch": 2.1632588365826595,
+      "grad_norm": 4.089602470397949,
+      "learning_rate": 4.366218036172324e-06,
+      "loss": 0.7694,
+      "step": 853
+    },
+    {
+      "epoch": 2.1657948961800604,
+      "grad_norm": 4.0583696365356445,
+      "learning_rate": 4.341825167305058e-06,
+      "loss": 0.7702,
+      "step": 854
+    },
+    {
+      "epoch": 2.168330955777461,
+      "grad_norm": 4.053401470184326,
+      "learning_rate": 4.317481719179949e-06,
+      "loss": 0.7592,
+      "step": 855
+    },
+    {
+      "epoch": 2.1708670153748613,
+      "grad_norm": 3.817272424697876,
+      "learning_rate": 4.2931879044223094e-06,
+      "loss": 0.7755,
+      "step": 856
+    },
+    {
+      "epoch": 2.1734030749722617,
+      "grad_norm": 4.260085105895996,
+      "learning_rate": 4.2689439352239535e-06,
+      "loss": 0.778,
+      "step": 857
+    },
+    {
+      "epoch": 2.1759391345696626,
+      "grad_norm": 4.195194244384766,
+      "learning_rate": 4.2447500233413205e-06,
+      "loss": 0.7769,
+      "step": 858
+    },
+    {
+      "epoch": 2.178475194167063,
+      "grad_norm": 4.542140960693359,
+      "learning_rate": 4.220606380093628e-06,
+      "loss": 0.7791,
+      "step": 859
+    },
+    {
+      "epoch": 2.1810112537644635,
+      "grad_norm": 5.197386264801025,
+      "learning_rate": 4.196513216361027e-06,
+      "loss": 0.7626,
+      "step": 860
+    },
+    {
+      "epoch": 2.183547313361864,
+      "grad_norm": 5.448459625244141,
+      "learning_rate": 4.1724707425827615e-06,
+      "loss": 0.7683,
+      "step": 861
+    },
+    {
+      "epoch": 2.1860833729592644,
+      "grad_norm": 4.393735408782959,
+      "learning_rate": 4.14847916875533e-06,
+      "loss": 0.7402,
+      "step": 862
+    },
+    {
+      "epoch": 2.1886194325566652,
+      "grad_norm": 5.02966833114624,
+      "learning_rate": 4.1245387044306475e-06,
+      "loss": 0.7678,
+      "step": 863
+    },
+    {
+      "epoch": 2.1911554921540657,
+      "grad_norm": 4.1165971755981445,
+      "learning_rate": 4.100649558714221e-06,
+      "loss": 0.788,
+      "step": 864
+    },
+    {
+      "epoch": 2.193691551751466,
+      "grad_norm": 4.929838180541992,
+      "learning_rate": 4.076811940263322e-06,
+      "loss": 0.7563,
+      "step": 865
+    },
+    {
+      "epoch": 2.1962276113488666,
+      "grad_norm": 4.211369514465332,
+      "learning_rate": 4.053026057285163e-06,
+      "loss": 0.7435,
+      "step": 866
+    },
+    {
+      "epoch": 2.198763670946267,
+      "grad_norm": 4.34557580947876,
+      "learning_rate": 4.029292117535064e-06,
+      "loss": 0.7687,
+      "step": 867
+    },
+    {
+      "epoch": 2.201299730543668,
+      "grad_norm": 4.37393045425415,
+      "learning_rate": 4.00561032831468e-06,
+      "loss": 0.7647,
+      "step": 868
+    },
+    {
+      "epoch": 2.2038357901410683,
+      "grad_norm": 4.262077808380127,
+      "learning_rate": 3.98198089647013e-06,
+      "loss": 0.7568,
+      "step": 869
+    },
+    {
+      "epoch": 2.206371849738469,
+      "grad_norm": 4.286789417266846,
+      "learning_rate": 3.958404028390255e-06,
+      "loss": 0.785,
+      "step": 870
+    },
+    {
+      "epoch": 2.2089079093358692,
+      "grad_norm": 4.210546493530273,
+      "learning_rate": 3.934879930004758e-06,
+      "loss": 0.7605,
+      "step": 871
+    },
+    {
+      "epoch": 2.21144396893327,
+      "grad_norm": 3.9751172065734863,
+      "learning_rate": 3.911408806782442e-06,
+      "loss": 0.7405,
+      "step": 872
+    },
+    {
+      "epoch": 2.2139800285306706,
+      "grad_norm": 3.995595932006836,
+      "learning_rate": 3.887990863729404e-06,
+      "loss": 0.7532,
+      "step": 873
+    },
+    {
+      "epoch": 2.216516088128071,
+      "grad_norm": 3.7829172611236572,
+      "learning_rate": 3.8646263053872425e-06,
+      "loss": 0.7843,
+      "step": 874
+    },
+    {
+      "epoch": 2.2190521477254714,
+      "grad_norm": 4.247334003448486,
+      "learning_rate": 3.841315335831271e-06,
+      "loss": 0.769,
+      "step": 875
+    },
+    {
+      "epoch": 2.2215882073228723,
+      "grad_norm": 4.032822132110596,
+      "learning_rate": 3.81805815866874e-06,
+      "loss": 0.7565,
+      "step": 876
+    },
+    {
+      "epoch": 2.2241242669202728,
+      "grad_norm": 3.9699594974517822,
+      "learning_rate": 3.7948549770370568e-06,
+      "loss": 0.7582,
+      "step": 877
+    },
+    {
+      "epoch": 2.226660326517673,
+      "grad_norm": 4.53628396987915,
+      "learning_rate": 3.771705993602006e-06,
+      "loss": 0.789,
+      "step": 878
+    },
+    {
+      "epoch": 2.2291963861150736,
+      "grad_norm": 5.106379508972168,
+      "learning_rate": 3.7486114105559912e-06,
+      "loss": 0.7959,
+      "step": 879
+    },
+    {
+      "epoch": 2.231732445712474,
+      "grad_norm": 4.434255123138428,
+      "learning_rate": 3.725571429616246e-06,
+      "loss": 0.7564,
+      "step": 880
+    },
+    {
+      "epoch": 2.234268505309875,
+      "grad_norm": 5.324062347412109,
+      "learning_rate": 3.7025862520231113e-06,
+      "loss": 0.7505,
+      "step": 881
+    },
+    {
+      "epoch": 2.2368045649072754,
+      "grad_norm": 4.399535655975342,
+      "learning_rate": 3.6796560785382274e-06,
+      "loss": 0.7516,
+      "step": 882
+    },
+    {
+      "epoch": 2.239340624504676,
+      "grad_norm": 4.95442008972168,
+      "learning_rate": 3.656781109442833e-06,
+      "loss": 0.7416,
+      "step": 883
+    },
+    {
+      "epoch": 2.2418766841020763,
+      "grad_norm": 4.9247612953186035,
+      "learning_rate": 3.6339615445359666e-06,
+      "loss": 0.7678,
+      "step": 884
+    },
+    {
+      "epoch": 2.2444127436994767,
+      "grad_norm": 5.491630554199219,
+      "learning_rate": 3.6111975831327593e-06,
+      "loss": 0.7668,
+      "step": 885
+    },
+    {
+      "epoch": 2.2469488032968776,
+      "grad_norm": 7.566525936126709,
+      "learning_rate": 3.5884894240626743e-06,
+      "loss": 0.7735,
+      "step": 886
+    },
+    {
+      "epoch": 2.249484862894278,
+      "grad_norm": 5.582590103149414,
+      "learning_rate": 3.565837265667775e-06,
+      "loss": 0.7824,
+      "step": 887
+    },
+    {
+      "epoch": 2.2520209224916785,
+      "grad_norm": 5.3473029136657715,
+      "learning_rate": 3.543241305800993e-06,
+      "loss": 0.7658,
+      "step": 888
+    },
+    {
+      "epoch": 2.254556982089079,
+      "grad_norm": 6.214839458465576,
+      "learning_rate": 3.5207017418244005e-06,
+      "loss": 0.8003,
+      "step": 889
+    },
+    {
+      "epoch": 2.2570930416864794,
+      "grad_norm": 4.573577880859375,
+      "learning_rate": 3.4982187706074856e-06,
+      "loss": 0.7712,
+      "step": 890
+    },
+    {
+      "epoch": 2.2596291012838803,
+      "grad_norm": 4.793715476989746,
+      "learning_rate": 3.4757925885254294e-06,
+      "loss": 0.7512,
+      "step": 891
+    },
+    {
+      "epoch": 2.2621651608812807,
+      "grad_norm": 5.82136344909668,
+      "learning_rate": 3.4534233914574013e-06,
+      "loss": 0.7483,
+      "step": 892
+    },
+    {
+      "epoch": 2.264701220478681,
+      "grad_norm": 4.513565540313721,
+      "learning_rate": 3.4311113747848324e-06,
+      "loss": 0.7525,
+      "step": 893
+    },
+    {
+      "epoch": 2.2672372800760816,
+      "grad_norm": 4.095034122467041,
+      "learning_rate": 3.4088567333897272e-06,
+      "loss": 0.7621,
+      "step": 894
+    },
+    {
+      "epoch": 2.2697733396734825,
+      "grad_norm": 5.215566635131836,
+      "learning_rate": 3.386659661652937e-06,
+      "loss": 0.7738,
+      "step": 895
+    },
+    {
+      "epoch": 2.272309399270883,
+      "grad_norm": 4.385331153869629,
+      "learning_rate": 3.3645203534524983e-06,
+      "loss": 0.7569,
+      "step": 896
+    },
+    {
+      "epoch": 2.2748454588682834,
+      "grad_norm": 4.473877906799316,
+      "learning_rate": 3.342439002161897e-06,
+      "loss": 0.7486,
+      "step": 897
+    },
+    {
+      "epoch": 2.277381518465684,
+      "grad_norm": 5.167495250701904,
+      "learning_rate": 3.320415800648411e-06,
+      "loss": 0.7697,
+      "step": 898
+    },
+    {
+      "epoch": 2.2799175780630847,
+      "grad_norm": 4.346419811248779,
+      "learning_rate": 3.298450941271416e-06,
+      "loss": 0.7665,
+      "step": 899
+    },
+    {
+      "epoch": 2.282453637660485,
+      "grad_norm": 5.885275840759277,
+      "learning_rate": 3.276544615880699e-06,
+      "loss": 0.7799,
+      "step": 900
+    },
+    {
+      "epoch": 2.282453637660485,
+      "eval_loss": NaN,
+      "eval_runtime": 207.2979,
+      "eval_samples_per_second": 40.111,
+      "eval_steps_per_second": 5.017,
+      "step": 900
+    },
+    {
+      "epoch": 2.2849896972578856,
+      "grad_norm": 5.007364749908447,
+      "learning_rate": 3.2546970158147917e-06,
+      "loss": 0.7486,
+      "step": 901
+    },
+    {
+      "epoch": 2.287525756855286,
+      "grad_norm": 4.034124851226807,
+      "learning_rate": 3.232908331899298e-06,
+      "loss": 0.7558,
+      "step": 902
+    },
+    {
+      "epoch": 2.2900618164526865,
+      "grad_norm": 4.541777610778809,
+      "learning_rate": 3.211178754445212e-06,
+      "loss": 0.7831,
+      "step": 903
+    },
+    {
+      "epoch": 2.2925978760500874,
+      "grad_norm": 4.531301975250244,
+      "learning_rate": 3.189508473247289e-06,
+      "loss": 0.7594,
+      "step": 904
+    },
+    {
+      "epoch": 2.295133935647488,
+      "grad_norm": 4.7508368492126465,
+      "learning_rate": 3.1678976775823454e-06,
+      "loss": 0.7568,
+      "step": 905
+    },
+    {
+      "epoch": 2.2976699952448882,
+      "grad_norm": 4.086254596710205,
+      "learning_rate": 3.1463465562076456e-06,
+      "loss": 0.7718,
+      "step": 906
+    },
+    {
+      "epoch": 2.3002060548422887,
+      "grad_norm": 4.427765846252441,
+      "learning_rate": 3.1248552973592282e-06,
+      "loss": 0.7599,
+      "step": 907
+    },
+    {
+      "epoch": 2.302742114439689,
+      "grad_norm": 4.405720233917236,
+      "learning_rate": 3.10342408875026e-06,
+      "loss": 0.7652,
+      "step": 908
+    },
+    {
+      "epoch": 2.30527817403709,
+      "grad_norm": 4.072112560272217,
+      "learning_rate": 3.0820531175694236e-06,
+      "loss": 0.7605,
+      "step": 909
+    },
+    {
+      "epoch": 2.3078142336344905,
+      "grad_norm": 4.144969940185547,
+      "learning_rate": 3.0607425704792426e-06,
+      "loss": 0.7542,
+      "step": 910
+    },
+    {
+      "epoch": 2.310350293231891,
+      "grad_norm": 4.27333402633667,
+      "learning_rate": 3.039492633614496e-06,
+      "loss": 0.7964,
+      "step": 911
+    },
+    {
+      "epoch": 2.3128863528292913,
+      "grad_norm": 3.909830331802368,
+      "learning_rate": 3.018303492580548e-06,
+      "loss": 0.7936,
+      "step": 912
+    },
+    {
+      "epoch": 2.315422412426692,
+      "grad_norm": 4.339663982391357,
+      "learning_rate": 2.997175332451757e-06,
+      "loss": 0.7822,
+      "step": 913
+    },
+    {
+      "epoch": 2.3179584720240927,
+      "grad_norm": 3.849867582321167,
+      "learning_rate": 2.976108337769854e-06,
+      "loss": 0.7868,
+      "step": 914
+    },
+    {
+      "epoch": 2.320494531621493,
+      "grad_norm": 4.143746852874756,
+      "learning_rate": 2.95510269254232e-06,
+      "loss": 0.7496,
+      "step": 915
+    },
+    {
+      "epoch": 2.3230305912188935,
+      "grad_norm": 4.088156223297119,
+      "learning_rate": 2.934158580240789e-06,
+      "loss": 0.783,
+      "step": 916
+    },
+    {
+      "epoch": 2.325566650816294,
+      "grad_norm": 3.79500675201416,
+      "learning_rate": 2.9132761837994426e-06,
+      "loss": 0.7609,
+      "step": 917
+    },
+    {
+      "epoch": 2.328102710413695,
+      "grad_norm": 3.894336223602295,
+      "learning_rate": 2.892455685613402e-06,
+      "loss": 0.7434,
+      "step": 918
+    },
+    {
+      "epoch": 2.3306387700110953,
+      "grad_norm": 3.9885644912719727,
+      "learning_rate": 2.8716972675371645e-06,
+      "loss": 0.7546,
+      "step": 919
+    },
+    {
+      "epoch": 2.3331748296084958,
+      "grad_norm": 4.468260288238525,
+      "learning_rate": 2.851001110882975e-06,
+      "loss": 0.7528,
+      "step": 920
+    },
+    {
+      "epoch": 2.335710889205896,
+      "grad_norm": 4.394624710083008,
+      "learning_rate": 2.830367396419269e-06,
+      "loss": 0.7835,
+      "step": 921
+    },
+    {
+      "epoch": 2.338246948803297,
+      "grad_norm": 4.524784564971924,
+      "learning_rate": 2.809796304369099e-06,
+      "loss": 0.7822,
+      "step": 922
+    },
+    {
+      "epoch": 2.3407830084006975,
+      "grad_norm": 4.426603317260742,
+      "learning_rate": 2.7892880144085253e-06,
+      "loss": 0.777,
+      "step": 923
+    },
+    {
+      "epoch": 2.343319067998098,
+      "grad_norm": 4.747066974639893,
+      "learning_rate": 2.768842705665097e-06,
+      "loss": 0.7792,
+      "step": 924
+    },
+    {
+      "epoch": 2.3458551275954984,
+      "grad_norm": 4.15431022644043,
+      "learning_rate": 2.748460556716234e-06,
+      "loss": 0.7746,
+      "step": 925
+    },
+    {
+      "epoch": 2.348391187192899,
+      "grad_norm": 4.427835941314697,
+      "learning_rate": 2.7281417455877103e-06,
+      "loss": 0.7863,
+      "step": 926
+    },
+    {
+      "epoch": 2.3509272467902997,
+      "grad_norm": 4.1503520011901855,
+      "learning_rate": 2.7078864497520774e-06,
+      "loss": 0.7492,
+      "step": 927
+    },
+    {
+      "epoch": 2.3534633063877,
+      "grad_norm": 4.640717506408691,
+      "learning_rate": 2.68769484612712e-06,
+      "loss": 0.7928,
+      "step": 928
+    },
+    {
+      "epoch": 2.3559993659851006,
+      "grad_norm": 4.538097381591797,
+      "learning_rate": 2.6675671110743084e-06,
+      "loss": 0.7493,
+      "step": 929
+    },
+    {
+      "epoch": 2.358535425582501,
+      "grad_norm": 4.215808391571045,
+      "learning_rate": 2.647503420397263e-06,
+      "loss": 0.7547,
+      "step": 930
+    },
+    {
+      "epoch": 2.3610714851799015,
+      "grad_norm": 4.320316314697266,
+      "learning_rate": 2.627503949340202e-06,
+      "loss": 0.7907,
+      "step": 931
+    },
+    {
+      "epoch": 2.3636075447773024,
+      "grad_norm": 4.114699363708496,
+      "learning_rate": 2.607568872586449e-06,
+      "loss": 0.7579,
+      "step": 932
+    },
+    {
+      "epoch": 2.366143604374703,
+      "grad_norm": 4.327573299407959,
+      "learning_rate": 2.587698364256854e-06,
+      "loss": 0.7552,
+      "step": 933
+    },
+    {
+      "epoch": 2.3686796639721033,
+      "grad_norm": 4.364943027496338,
+      "learning_rate": 2.5678925979083257e-06,
+      "loss": 0.7708,
+      "step": 934
+    },
+    {
+      "epoch": 2.3712157235695037,
+      "grad_norm": 3.9766640663146973,
+      "learning_rate": 2.5481517465322726e-06,
+      "loss": 0.7452,
+      "step": 935
+    },
+    {
+      "epoch": 2.3737517831669046,
+      "grad_norm": 4.603160381317139,
+      "learning_rate": 2.528475982553118e-06,
+      "loss": 0.7529,
+      "step": 936
+    },
+    {
+      "epoch": 2.376287842764305,
+      "grad_norm": 4.066822052001953,
+      "learning_rate": 2.508865477826784e-06,
+      "loss": 0.794,
+      "step": 937
+    },
+    {
+      "epoch": 2.3788239023617055,
+      "grad_norm": 3.7265427112579346,
+      "learning_rate": 2.4893204036391927e-06,
+      "loss": 0.7597,
+      "step": 938
+    },
+    {
+      "epoch": 2.381359961959106,
+      "grad_norm": 4.10790491104126,
+      "learning_rate": 2.469840930704779e-06,
+      "loss": 0.7756,
+      "step": 939
+    },
+    {
+      "epoch": 2.383896021556507,
+      "grad_norm": 4.320775032043457,
+      "learning_rate": 2.4504272291649722e-06,
+      "loss": 0.7634,
+      "step": 940
+    },
+    {
+      "epoch": 2.3864320811539073,
+      "grad_norm": 4.10369873046875,
+      "learning_rate": 2.431079468586741e-06,
+      "loss": 0.775,
+      "step": 941
+    },
+    {
+      "epoch": 2.3889681407513077,
+      "grad_norm": 4.050859451293945,
+      "learning_rate": 2.411797817961097e-06,
+      "loss": 0.7625,
+      "step": 942
+    },
+    {
+      "epoch": 2.391504200348708,
+      "grad_norm": 4.410302639007568,
+      "learning_rate": 2.3925824457016244e-06,
+      "loss": 0.7935,
+      "step": 943
+    },
+    {
+      "epoch": 2.3940402599461086,
+      "grad_norm": 4.687882900238037,
+      "learning_rate": 2.3734335196429926e-06,
+      "loss": 0.7792,
+      "step": 944
+    },
+    {
+      "epoch": 2.3965763195435095,
+      "grad_norm": 4.416008949279785,
+      "learning_rate": 2.354351207039526e-06,
+      "loss": 0.776,
+      "step": 945
+    },
+    {
+      "epoch": 2.39911237914091,
+      "grad_norm": 4.8968825340271,
+      "learning_rate": 2.3353356745636944e-06,
+      "loss": 0.775,
+      "step": 946
+    },
+    {
+      "epoch": 2.4016484387383104,
+      "grad_norm": 4.886624336242676,
+      "learning_rate": 2.3163870883047113e-06,
+      "loss": 0.7894,
+      "step": 947
+    },
+    {
+      "epoch": 2.404184498335711,
+      "grad_norm": 4.6147141456604,
+      "learning_rate": 2.2975056137670315e-06,
+      "loss": 0.7685,
+      "step": 948
+    },
+    {
+      "epoch": 2.4067205579331112,
+      "grad_norm": 5.081536769866943,
+      "learning_rate": 2.2786914158689412e-06,
+      "loss": 0.7694,
+      "step": 949
+    },
+    {
+      "epoch": 2.409256617530512,
+      "grad_norm": 4.840873718261719,
+      "learning_rate": 2.2599446589411033e-06,
+      "loss": 0.7779,
+      "step": 950
+    },
+    {
+      "epoch": 2.409256617530512,
+      "eval_loss": NaN,
+      "eval_runtime": 206.7579,
+      "eval_samples_per_second": 40.216,
+      "eval_steps_per_second": 5.03,
+      "step": 950
+    },
+    {
+      "epoch": 2.4117926771279126,
+      "grad_norm": 4.529221057891846,
+      "learning_rate": 2.2412655067251268e-06,
+      "loss": 0.8008,
+      "step": 951
+    },
+    {
+      "epoch": 2.414328736725313,
+      "grad_norm": 4.9453206062316895,
+      "learning_rate": 2.222654122372131e-06,
+      "loss": 0.7606,
+      "step": 952
+    },
+    {
+      "epoch": 2.4168647963227134,
+      "grad_norm": 4.322136402130127,
+      "learning_rate": 2.204110668441326e-06,
+      "loss": 0.7674,
+      "step": 953
+    },
+    {
+      "epoch": 2.419400855920114,
+      "grad_norm": 4.078553199768066,
+      "learning_rate": 2.185635306898589e-06,
+      "loss": 0.7788,
+      "step": 954
+    },
+    {
+      "epoch": 2.4219369155175148,
+      "grad_norm": 4.5291314125061035,
+      "learning_rate": 2.1672281991150534e-06,
+      "loss": 0.7796,
+      "step": 955
+    },
+    {
+      "epoch": 2.424472975114915,
+      "grad_norm": 4.254805088043213,
+      "learning_rate": 2.1488895058656946e-06,
+      "loss": 0.7744,
+      "step": 956
+    },
+    {
+      "epoch": 2.4270090347123157,
+      "grad_norm": 4.069855690002441,
+      "learning_rate": 2.13061938732793e-06,
+      "loss": 0.7981,
+      "step": 957
+    },
+    {
+      "epoch": 2.429545094309716,
+      "grad_norm": 3.9820396900177,
+      "learning_rate": 2.112418003080222e-06,
+      "loss": 0.7393,
+      "step": 958
+    },
+    {
+      "epoch": 2.432081153907117,
+      "grad_norm": 4.322441577911377,
+      "learning_rate": 2.0942855121006677e-06,
+      "loss": 0.7737,
+      "step": 959
+    },
+    {
+      "epoch": 2.4346172135045174,
+      "grad_norm": 3.9814891815185547,
+      "learning_rate": 2.0762220727656425e-06,
+      "loss": 0.7801,
+      "step": 960
+    },
+    {
+      "epoch": 2.437153273101918,
+      "grad_norm": 4.416321277618408,
+      "learning_rate": 2.058227842848376e-06,
+      "loss": 0.7957,
+      "step": 961
+    },
+    {
+      "epoch": 2.4396893326993183,
+      "grad_norm": 4.082301616668701,
+      "learning_rate": 2.0403029795176156e-06,
+      "loss": 0.7702,
+      "step": 962
+    },
+    {
+      "epoch": 2.442225392296719,
+      "grad_norm": 3.9780466556549072,
+      "learning_rate": 2.022447639336218e-06,
+      "loss": 0.7421,
+      "step": 963
+    },
+    {
+      "epoch": 2.4447614518941196,
+      "grad_norm": 4.323615074157715,
+      "learning_rate": 2.004661978259804e-06,
+      "loss": 0.7442,
+      "step": 964
+    },
+    {
+      "epoch": 2.44729751149152,
+      "grad_norm": 4.2946600914001465,
+      "learning_rate": 1.9869461516353893e-06,
+      "loss": 0.7528,
+      "step": 965
+    },
+    {
+      "epoch": 2.4498335710889205,
+      "grad_norm": 4.678606033325195,
+      "learning_rate": 1.9693003142000277e-06,
+      "loss": 0.7594,
+      "step": 966
+    },
+    {
+      "epoch": 2.452369630686321,
+      "grad_norm": 4.818963527679443,
+      "learning_rate": 1.951724620079457e-06,
+      "loss": 0.7373,
+      "step": 967
+    },
+    {
+      "epoch": 2.454905690283722,
+      "grad_norm": 4.596181392669678,
+      "learning_rate": 1.9342192227867595e-06,
+      "loss": 0.7584,
+      "step": 968
+    },
+    {
+      "epoch": 2.4574417498811223,
+      "grad_norm": 4.812653064727783,
+      "learning_rate": 1.916784275221013e-06,
+      "loss": 0.7502,
+      "step": 969
+    },
+    {
+      "epoch": 2.4599778094785227,
+      "grad_norm": 5.208719253540039,
+      "learning_rate": 1.8994199296659633e-06,
+      "loss": 0.7833,
+      "step": 970
+    },
+    {
+      "epoch": 2.462513869075923,
+      "grad_norm": 4.855763912200928,
+      "learning_rate": 1.882126337788691e-06,
+      "loss": 0.7709,
+      "step": 971
+    },
+    {
+      "epoch": 2.4650499286733236,
+      "grad_norm": 4.337666034698486,
+      "learning_rate": 1.8649036506382745e-06,
+      "loss": 0.7439,
+      "step": 972
+    },
+    {
+      "epoch": 2.4675859882707245,
+      "grad_norm": 5.401799201965332,
+      "learning_rate": 1.847752018644503e-06,
+      "loss": 0.7605,
+      "step": 973
+    },
+    {
+      "epoch": 2.470122047868125,
+      "grad_norm": 4.645407676696777,
+      "learning_rate": 1.8306715916165185e-06,
+      "loss": 0.7686,
+      "step": 974
+    },
+    {
+      "epoch": 2.4726581074655254,
+      "grad_norm": 4.161717414855957,
+      "learning_rate": 1.8136625187415535e-06,
+      "loss": 0.7555,
+      "step": 975
+    },
+    {
+      "epoch": 2.475194167062926,
+      "grad_norm": 4.439455032348633,
+      "learning_rate": 1.7967249485835847e-06,
+      "loss": 0.7886,
+      "step": 976
+    },
+    {
+      "epoch": 2.4777302266603263,
+      "grad_norm": 4.506514549255371,
+      "learning_rate": 1.7798590290820684e-06,
+      "loss": 0.7779,
+      "step": 977
+    },
+    {
+      "epoch": 2.480266286257727,
+      "grad_norm": 4.490094184875488,
+      "learning_rate": 1.763064907550631e-06,
+      "loss": 0.7617,
+      "step": 978
+    },
+    {
+      "epoch": 2.4828023458551276,
+      "grad_norm": 4.33901309967041,
+      "learning_rate": 1.7463427306757886e-06,
+      "loss": 0.7519,
+      "step": 979
+    },
+    {
+      "epoch": 2.485338405452528,
+      "grad_norm": 4.4128899574279785,
+      "learning_rate": 1.7296926445156636e-06,
+      "loss": 0.7365,
+      "step": 980
+    },
+    {
+      "epoch": 2.4878744650499285,
+      "grad_norm": 4.4159321784973145,
+      "learning_rate": 1.7131147944987091e-06,
+      "loss": 0.7715,
+      "step": 981
+    },
+    {
+      "epoch": 2.4904105246473294,
+      "grad_norm": 4.302286624908447,
+      "learning_rate": 1.696609325422439e-06,
+      "loss": 0.7536,
+      "step": 982
+    },
+    {
+      "epoch": 2.49294658424473,
+      "grad_norm": 4.743297100067139,
+      "learning_rate": 1.6801763814521644e-06,
+      "loss": 0.7681,
+      "step": 983
+    },
+    {
+      "epoch": 2.4954826438421303,
+      "grad_norm": 4.709976673126221,
+      "learning_rate": 1.6638161061197322e-06,
+      "loss": 0.7688,
+      "step": 984
+    },
+    {
+      "epoch": 2.4980187034395307,
+      "grad_norm": 4.287345886230469,
+      "learning_rate": 1.6475286423222736e-06,
+      "loss": 0.7628,
+      "step": 985
+    },
+    {
+      "epoch": 2.5005547630369316,
+      "grad_norm": 5.02631950378418,
+      "learning_rate": 1.6313141323209558e-06,
+      "loss": 0.7665,
+      "step": 986
+    },
+    {
+      "epoch": 2.503090822634332,
+      "grad_norm": 4.6641974449157715,
+      "learning_rate": 1.6151727177397315e-06,
+      "loss": 0.7862,
+      "step": 987
+    },
+    {
+      "epoch": 2.5056268822317325,
+      "grad_norm": 4.26420259475708,
+      "learning_rate": 1.5991045395641236e-06,
+      "loss": 0.785,
+      "step": 988
+    },
+    {
+      "epoch": 2.508162941829133,
+      "grad_norm": 4.711597919464111,
+      "learning_rate": 1.583109738139964e-06,
+      "loss": 0.7657,
+      "step": 989
+    },
+    {
+      "epoch": 2.5106990014265333,
+      "grad_norm": 4.841354846954346,
+      "learning_rate": 1.5671884531721903e-06,
+      "loss": 0.7813,
+      "step": 990
+    },
+    {
+      "epoch": 2.5132350610239342,
+      "grad_norm": 4.445319175720215,
+      "learning_rate": 1.5513408237236194e-06,
+      "loss": 0.7431,
+      "step": 991
+    },
+    {
+      "epoch": 2.5157711206213347,
+      "grad_norm": 4.24920129776001,
+      "learning_rate": 1.535566988213728e-06,
+      "loss": 0.7564,
+      "step": 992
+    },
+    {
+      "epoch": 2.518307180218735,
+      "grad_norm": 4.309378147125244,
+      "learning_rate": 1.519867084417449e-06,
+      "loss": 0.7661,
+      "step": 993
+    },
+    {
+      "epoch": 2.5208432398161356,
+      "grad_norm": 4.118340015411377,
+      "learning_rate": 1.504241249463968e-06,
+      "loss": 0.7851,
+      "step": 994
+    },
+    {
+      "epoch": 2.523379299413536,
+      "grad_norm": 4.363431453704834,
+      "learning_rate": 1.4886896198355205e-06,
+      "loss": 0.7556,
+      "step": 995
+    },
+    {
+      "epoch": 2.525915359010937,
+      "grad_norm": 4.172931671142578,
+      "learning_rate": 1.473212331366205e-06,
+      "loss": 0.7393,
+      "step": 996
+    },
+    {
+      "epoch": 2.5284514186083373,
+      "grad_norm": 4.247659683227539,
+      "learning_rate": 1.457809519240796e-06,
+      "loss": 0.7837,
+      "step": 997
+    },
+    {
+      "epoch": 2.5309874782057378,
+      "grad_norm": 4.314168453216553,
+      "learning_rate": 1.4424813179935604e-06,
+      "loss": 0.7713,
+      "step": 998
+    },
+    {
+      "epoch": 2.533523537803138,
+      "grad_norm": 3.864086389541626,
+      "learning_rate": 1.4272278615070877e-06,
+      "loss": 0.7709,
+      "step": 999
+    },
+    {
+      "epoch": 2.5360595974005387,
+      "grad_norm": 4.122450828552246,
+      "learning_rate": 1.4120492830111054e-06,
+      "loss": 0.749,
+      "step": 1000
+    },
+    {
+      "epoch": 2.5360595974005387,
+      "eval_loss": NaN,
+      "eval_runtime": 206.9865,
+      "eval_samples_per_second": 40.172,
+      "eval_steps_per_second": 5.024,
+      "step": 1000
+    },
+    {
+      "epoch": 2.5385956569979395,
+      "grad_norm": 4.103104591369629,
+      "learning_rate": 1.396945715081347e-06,
+      "loss": 0.7697,
+      "step": 1001
+    },
+    {
+      "epoch": 2.54113171659534,
+      "grad_norm": 4.2063164710998535,
+      "learning_rate": 1.3819172896383515e-06,
+      "loss": 0.7621,
+      "step": 1002
+    },
+    {
+      "epoch": 2.5436677761927404,
+      "grad_norm": 4.657142162322998,
+      "learning_rate": 1.366964137946355e-06,
+      "loss": 0.7675,
+      "step": 1003
+    },
+    {
+      "epoch": 2.5462038357901413,
+      "grad_norm": 4.5668158531188965,
+      "learning_rate": 1.352086390612105e-06,
+      "loss": 0.7609,
+      "step": 1004
+    },
+    {
+      "epoch": 2.5487398953875418,
+      "grad_norm": 4.0702362060546875,
+      "learning_rate": 1.3372841775837476e-06,
+      "loss": 0.7852,
+      "step": 1005
+    },
+    {
+      "epoch": 2.551275954984942,
+      "grad_norm": 4.430232524871826,
+      "learning_rate": 1.32255762814968e-06,
+      "loss": 0.7845,
+      "step": 1006
+    },
+    {
+      "epoch": 2.5538120145823426,
+      "grad_norm": 4.624642848968506,
+      "learning_rate": 1.307906870937422e-06,
+      "loss": 0.7875,
+      "step": 1007
+    },
+    {
+      "epoch": 2.556348074179743,
+      "grad_norm": 4.245471000671387,
+      "learning_rate": 1.2933320339124955e-06,
+      "loss": 0.7517,
+      "step": 1008
+    },
+    {
+      "epoch": 2.558884133777144,
+      "grad_norm": 3.987699031829834,
+      "learning_rate": 1.2788332443773066e-06,
+      "loss": 0.7679,
+      "step": 1009
+    },
+    {
+      "epoch": 2.5614201933745444,
+      "grad_norm": 4.822307586669922,
+      "learning_rate": 1.2644106289700297e-06,
+      "loss": 0.7868,
+      "step": 1010
+    },
+    {
+      "epoch": 2.563956252971945,
+      "grad_norm": 4.65307092666626,
+      "learning_rate": 1.250064313663506e-06,
+      "loss": 0.7712,
+      "step": 1011
+    },
+    {
+      "epoch": 2.5664923125693453,
+      "grad_norm": 4.200237274169922,
+      "learning_rate": 1.2357944237641439e-06,
+      "loss": 0.7696,
+      "step": 1012
+    },
+    {
+      "epoch": 2.5690283721667457,
+      "grad_norm": 4.255523681640625,
+      "learning_rate": 1.2216010839108117e-06,
+      "loss": 0.7771,
+      "step": 1013
+    },
+    {
+      "epoch": 2.5715644317641466,
+      "grad_norm": 4.196678638458252,
+      "learning_rate": 1.2074844180737743e-06,
+      "loss": 0.7856,
+      "step": 1014
+    },
+    {
+      "epoch": 2.574100491361547,
+      "grad_norm": 4.187068939208984,
+      "learning_rate": 1.1934445495535818e-06,
+      "loss": 0.781,
+      "step": 1015
+    },
+    {
+      "epoch": 2.5766365509589475,
+      "grad_norm": 4.118671894073486,
+      "learning_rate": 1.179481600980017e-06,
+      "loss": 0.7937,
+      "step": 1016
+    },
+    {
+      "epoch": 2.579172610556348,
+      "grad_norm": 4.127950191497803,
+      "learning_rate": 1.1655956943110046e-06,
+      "loss": 0.7692,
+      "step": 1017
+    },
+    {
+      "epoch": 2.5817086701537484,
+      "grad_norm": 4.368701934814453,
+      "learning_rate": 1.1517869508315583e-06,
+      "loss": 0.7714,
+      "step": 1018
+    },
+    {
+      "epoch": 2.5842447297511493,
+      "grad_norm": 4.254402160644531,
+      "learning_rate": 1.1380554911527209e-06,
+      "loss": 0.7619,
+      "step": 1019
+    },
+    {
+      "epoch": 2.5867807893485497,
+      "grad_norm": 3.9647600650787354,
+      "learning_rate": 1.1244014352105025e-06,
+      "loss": 0.7724,
+      "step": 1020
+    },
+    {
+      "epoch": 2.58931684894595,
+      "grad_norm": 3.9789211750030518,
+      "learning_rate": 1.1108249022648421e-06,
+      "loss": 0.7602,
+      "step": 1021
+    },
+    {
+      "epoch": 2.591852908543351,
+      "grad_norm": 4.251387596130371,
+      "learning_rate": 1.0973260108985594e-06,
+      "loss": 0.7304,
+      "step": 1022
+    },
+    {
+      "epoch": 2.594388968140751,
+      "grad_norm": 4.478714942932129,
+      "learning_rate": 1.0839048790163254e-06,
+      "loss": 0.7804,
+      "step": 1023
+    },
+    {
+      "epoch": 2.596925027738152,
+      "grad_norm": 4.948723316192627,
+      "learning_rate": 1.0705616238436245e-06,
+      "loss": 0.7684,
+      "step": 1024
+    },
+    {
+      "epoch": 2.5994610873355524,
+      "grad_norm": 4.321564674377441,
+      "learning_rate": 1.0572963619257393e-06,
+      "loss": 0.7611,
+      "step": 1025
+    },
+    {
+      "epoch": 2.601997146932953,
+      "grad_norm": 4.178552150726318,
+      "learning_rate": 1.0441092091267247e-06,
+      "loss": 0.7495,
+      "step": 1026
+    },
+    {
+      "epoch": 2.6045332065303537,
+      "grad_norm": 4.253757953643799,
+      "learning_rate": 1.0310002806284037e-06,
+      "loss": 0.7507,
+      "step": 1027
+    },
+    {
+      "epoch": 2.607069266127754,
+      "grad_norm": 4.590270519256592,
+      "learning_rate": 1.0179696909293501e-06,
+      "loss": 0.7762,
+      "step": 1028
+    },
+    {
+      "epoch": 2.6096053257251546,
+      "grad_norm": 4.320034503936768,
+      "learning_rate": 1.0050175538439054e-06,
+      "loss": 0.7483,
+      "step": 1029
+    },
+    {
+      "epoch": 2.612141385322555,
+      "grad_norm": 3.9760520458221436,
+      "learning_rate": 9.921439825011647e-07,
+      "loss": 0.7562,
+      "step": 1030
+    },
+    {
+      "epoch": 2.6146774449199555,
+      "grad_norm": 4.410188674926758,
+      "learning_rate": 9.793490893440095e-07,
+      "loss": 0.7724,
+      "step": 1031
+    },
+    {
+      "epoch": 2.6172135045173563,
+      "grad_norm": 4.840196132659912,
+      "learning_rate": 9.666329861281042e-07,
+      "loss": 0.7704,
+      "step": 1032
+    },
+    {
+      "epoch": 2.619749564114757,
+      "grad_norm": 4.912466526031494,
+      "learning_rate": 9.53995783920938e-07,
+      "loss": 0.7781,
+      "step": 1033
+    },
+    {
+      "epoch": 2.6222856237121572,
+      "grad_norm": 4.116244316101074,
+      "learning_rate": 9.414375931008446e-07,
+      "loss": 0.7691,
+      "step": 1034
+    },
+    {
+      "epoch": 2.6248216833095577,
+      "grad_norm": 4.244197368621826,
+      "learning_rate": 9.289585233560406e-07,
+      "loss": 0.7795,
+      "step": 1035
+    },
+    {
+      "epoch": 2.627357742906958,
+      "grad_norm": 4.277320861816406,
+      "learning_rate": 9.165586836836682e-07,
+      "loss": 0.743,
+      "step": 1036
+    },
+    {
+      "epoch": 2.629893802504359,
+      "grad_norm": 4.302981853485107,
+      "learning_rate": 9.042381823888446e-07,
+      "loss": 0.779,
+      "step": 1037
+    },
+    {
+      "epoch": 2.6324298621017594,
+      "grad_norm": 4.026116847991943,
+      "learning_rate": 8.919971270837114e-07,
+      "loss": 0.7543,
+      "step": 1038
+    },
+    {
+      "epoch": 2.63496592169916,
+      "grad_norm": 4.270394802093506,
+      "learning_rate": 8.79835624686498e-07,
+      "loss": 0.7687,
+      "step": 1039
+    },
+    {
+      "epoch": 2.6375019812965603,
+      "grad_norm": 3.911893606185913,
+      "learning_rate": 8.677537814205905e-07,
+      "loss": 0.7524,
+      "step": 1040
+    },
+    {
+      "epoch": 2.6400380408939608,
+      "grad_norm": 3.993382453918457,
+      "learning_rate": 8.55751702813592e-07,
+      "loss": 0.7554,
+      "step": 1041
+    },
+    {
+      "epoch": 2.6425741004913617,
+      "grad_norm": 4.202291965484619,
+      "learning_rate": 8.438294936964231e-07,
+      "loss": 0.7735,
+      "step": 1042
+    },
+    {
+      "epoch": 2.645110160088762,
+      "grad_norm": 4.204040050506592,
+      "learning_rate": 8.319872582023769e-07,
+      "loss": 0.7493,
+      "step": 1043
+    },
+    {
+      "epoch": 2.6476462196861625,
+      "grad_norm": 3.9854254722595215,
+      "learning_rate": 8.202250997662398e-07,
+      "loss": 0.7841,
+      "step": 1044
+    },
+    {
+      "epoch": 2.6501822792835634,
+      "grad_norm": 4.154744625091553,
+      "learning_rate": 8.085431211233607e-07,
+      "loss": 0.7536,
+      "step": 1045
+    },
+    {
+      "epoch": 2.652718338880964,
+      "grad_norm": 3.8959550857543945,
+      "learning_rate": 7.969414243087769e-07,
+      "loss": 0.7738,
+      "step": 1046
+    },
+    {
+      "epoch": 2.6552543984783643,
+      "grad_norm": 4.237845420837402,
+      "learning_rate": 7.854201106563053e-07,
+      "loss": 0.7819,
+      "step": 1047
+    },
+    {
+      "epoch": 2.6577904580757648,
+      "grad_norm": 4.1497087478637695,
+      "learning_rate": 7.739792807976676e-07,
+      "loss": 0.7579,
+      "step": 1048
+    },
+    {
+      "epoch": 2.660326517673165,
+      "grad_norm": 4.020683288574219,
+      "learning_rate": 7.626190346616069e-07,
+      "loss": 0.7528,
+      "step": 1049
+    },
+    {
+      "epoch": 2.662862577270566,
+      "grad_norm": 4.269307613372803,
+      "learning_rate": 7.513394714730182e-07,
+      "loss": 0.7617,
+      "step": 1050
+    },
+    {
+      "epoch": 2.662862577270566,
+      "eval_loss": NaN,
+      "eval_runtime": 206.651,
+      "eval_samples_per_second": 40.237,
+      "eval_steps_per_second": 5.033,
+      "step": 1050
+    },
+    {
+      "epoch": 2.6653986368679665,
+      "grad_norm": 4.087046146392822,
+      "learning_rate": 7.401406897520769e-07,
+      "loss": 0.7457,
+      "step": 1051
+    },
+    {
+      "epoch": 2.667934696465367,
+      "grad_norm": 4.389514923095703,
+      "learning_rate": 7.290227873133826e-07,
+      "loss": 0.7632,
+      "step": 1052
+    },
+    {
+      "epoch": 2.6704707560627674,
+      "grad_norm": 4.557702541351318,
+      "learning_rate": 7.179858612651036e-07,
+      "loss": 0.7554,
+      "step": 1053
+    },
+    {
+      "epoch": 2.673006815660168,
+      "grad_norm": 4.396777629852295,
+      "learning_rate": 7.070300080081261e-07,
+      "loss": 0.7493,
+      "step": 1054
+    },
+    {
+      "epoch": 2.6755428752575687,
+      "grad_norm": 4.202752590179443,
+      "learning_rate": 6.96155323235217e-07,
+      "loss": 0.7628,
+      "step": 1055
+    },
+    {
+      "epoch": 2.678078934854969,
+      "grad_norm": 4.209216117858887,
+      "learning_rate": 6.853619019301816e-07,
+      "loss": 0.7599,
+      "step": 1056
+    },
+    {
+      "epoch": 2.6806149944523696,
+      "grad_norm": 4.431794166564941,
+      "learning_rate": 6.746498383670463e-07,
+      "loss": 0.7794,
+      "step": 1057
+    },
+    {
+      "epoch": 2.68315105404977,
+      "grad_norm": 4.154015064239502,
+      "learning_rate": 6.64019226109216e-07,
+      "loss": 0.7645,
+      "step": 1058
+    },
+    {
+      "epoch": 2.6856871136471705,
+      "grad_norm": 4.195700168609619,
+      "learning_rate": 6.534701580086733e-07,
+      "loss": 0.764,
+      "step": 1059
+    },
+    {
+      "epoch": 2.6882231732445714,
+      "grad_norm": 4.079827308654785,
+      "learning_rate": 6.43002726205163e-07,
+      "loss": 0.7505,
+      "step": 1060
+    },
+    {
+      "epoch": 2.690759232841972,
+      "grad_norm": 4.238046646118164,
+      "learning_rate": 6.326170221253825e-07,
+      "loss": 0.7709,
+      "step": 1061
+    },
+    {
+      "epoch": 2.6932952924393723,
+      "grad_norm": 4.267265796661377,
+      "learning_rate": 6.2231313648219e-07,
+      "loss": 0.7473,
+      "step": 1062
+    },
+    {
+      "epoch": 2.6958313520367727,
+      "grad_norm": 4.235876560211182,
+      "learning_rate": 6.120911592738055e-07,
+      "loss": 0.7512,
+      "step": 1063
+    },
+    {
+      "epoch": 2.698367411634173,
+      "grad_norm": 4.081003189086914,
+      "learning_rate": 6.019511797830335e-07,
+      "loss": 0.7327,
+      "step": 1064
+    },
+    {
+      "epoch": 2.700903471231574,
+      "grad_norm": 3.9032557010650635,
+      "learning_rate": 5.918932865764726e-07,
+      "loss": 0.7562,
+      "step": 1065
+    },
+    {
+      "epoch": 2.7034395308289745,
+      "grad_norm": 3.8793997764587402,
+      "learning_rate": 5.819175675037492e-07,
+      "loss": 0.7558,
+      "step": 1066
+    },
+    {
+      "epoch": 2.705975590426375,
+      "grad_norm": 4.520419120788574,
+      "learning_rate": 5.720241096967504e-07,
+      "loss": 0.7629,
+      "step": 1067
+    },
+    {
+      "epoch": 2.708511650023776,
+      "grad_norm": 4.16015625,
+      "learning_rate": 5.622129995688586e-07,
+      "loss": 0.7887,
+      "step": 1068
+    },
+    {
+      "epoch": 2.7110477096211762,
+      "grad_norm": 3.9902470111846924,
+      "learning_rate": 5.524843228141963e-07,
+      "loss": 0.7883,
+      "step": 1069
+    },
+    {
+      "epoch": 2.7135837692185767,
+      "grad_norm": 3.864551067352295,
+      "learning_rate": 5.428381644068881e-07,
+      "loss": 0.7532,
+      "step": 1070
+    },
+    {
+      "epoch": 2.716119828815977,
+      "grad_norm": 4.239519119262695,
+      "learning_rate": 5.332746086002993e-07,
+      "loss": 0.7425,
+      "step": 1071
+    },
+    {
+      "epoch": 2.7186558884133776,
+      "grad_norm": 4.569217681884766,
+      "learning_rate": 5.237937389263248e-07,
+      "loss": 0.76,
+      "step": 1072
+    },
+    {
+      "epoch": 2.7211919480107785,
+      "grad_norm": 4.2440409660339355,
+      "learning_rate": 5.143956381946347e-07,
+      "loss": 0.7857,
+      "step": 1073
+    },
+    {
+      "epoch": 2.723728007608179,
+      "grad_norm": 4.227081298828125,
+      "learning_rate": 5.050803884919664e-07,
+      "loss": 0.7623,
+      "step": 1074
+    },
+    {
+      "epoch": 2.7262640672055793,
+      "grad_norm": 4.057560920715332,
+      "learning_rate": 4.958480711814062e-07,
+      "loss": 0.7763,
+      "step": 1075
+    },
+    {
+      "epoch": 2.72880012680298,
+      "grad_norm": 4.183563232421875,
+      "learning_rate": 4.866987669016709e-07,
+      "loss": 0.7624,
+      "step": 1076
+    },
+    {
+      "epoch": 2.7313361864003802,
+      "grad_norm": 4.229283332824707,
+      "learning_rate": 4.77632555566414e-07,
+      "loss": 0.7696,
+      "step": 1077
+    },
+    {
+      "epoch": 2.733872245997781,
+      "grad_norm": 4.2020182609558105,
+      "learning_rate": 4.6864951636351964e-07,
+      "loss": 0.7764,
+      "step": 1078
+    },
+    {
+      "epoch": 2.7364083055951816,
+      "grad_norm": 3.9811360836029053,
+      "learning_rate": 4.597497277544083e-07,
+      "loss": 0.7908,
+      "step": 1079
+    },
+    {
+      "epoch": 2.738944365192582,
+      "grad_norm": 4.074171543121338,
+      "learning_rate": 4.50933267473368e-07,
+      "loss": 0.7708,
+      "step": 1080
+    },
+    {
+      "epoch": 2.7414804247899824,
+      "grad_norm": 4.1424479484558105,
+      "learning_rate": 4.422002125268565e-07,
+      "loss": 0.797,
+      "step": 1081
+    },
+    {
+      "epoch": 2.744016484387383,
+      "grad_norm": 3.9790263175964355,
+      "learning_rate": 4.335506391928357e-07,
+      "loss": 0.7622,
+      "step": 1082
+    },
+    {
+      "epoch": 2.7465525439847838,
+      "grad_norm": 3.9956624507904053,
+      "learning_rate": 4.2498462302011155e-07,
+      "loss": 0.749,
+      "step": 1083
+    },
+    {
+      "epoch": 2.749088603582184,
+      "grad_norm": 3.9971725940704346,
+      "learning_rate": 4.1650223882766074e-07,
+      "loss": 0.7701,
+      "step": 1084
+    },
+    {
+      "epoch": 2.7516246631795847,
+      "grad_norm": 4.045487880706787,
+      "learning_rate": 4.0810356070399513e-07,
+      "loss": 0.7822,
+      "step": 1085
+    },
+    {
+      "epoch": 2.754160722776985,
+      "grad_norm": 4.2575364112854,
+      "learning_rate": 3.99788662006495e-07,
+      "loss": 0.7534,
+      "step": 1086
+    },
+    {
+      "epoch": 2.7566967823743855,
+      "grad_norm": 4.3285112380981445,
+      "learning_rate": 3.9155761536078365e-07,
+      "loss": 0.7902,
+      "step": 1087
+    },
+    {
+      "epoch": 2.7592328419717864,
+      "grad_norm": 4.146034240722656,
+      "learning_rate": 3.8341049266008457e-07,
+      "loss": 0.7565,
+      "step": 1088
+    },
+    {
+      "epoch": 2.761768901569187,
+      "grad_norm": 4.155364990234375,
+      "learning_rate": 3.753473650645978e-07,
+      "loss": 0.7726,
+      "step": 1089
+    },
+    {
+      "epoch": 2.7643049611665873,
+      "grad_norm": 4.021719455718994,
+      "learning_rate": 3.6736830300087787e-07,
+      "loss": 0.7676,
+      "step": 1090
+    },
+    {
+      "epoch": 2.766841020763988,
+      "grad_norm": 4.42950963973999,
+      "learning_rate": 3.5947337616121456e-07,
+      "loss": 0.7784,
+      "step": 1091
+    },
+    {
+      "epoch": 2.7693770803613886,
+      "grad_norm": 4.054685592651367,
+      "learning_rate": 3.516626535030276e-07,
+      "loss": 0.7598,
+      "step": 1092
+    },
+    {
+      "epoch": 2.771913139958789,
+      "grad_norm": 3.8842220306396484,
+      "learning_rate": 3.4393620324826847e-07,
+      "loss": 0.7624,
+      "step": 1093
+    },
+    {
+      "epoch": 2.7744491995561895,
+      "grad_norm": 3.734255075454712,
+      "learning_rate": 3.362940928828129e-07,
+      "loss": 0.7726,
+      "step": 1094
+    },
+    {
+      "epoch": 2.77698525915359,
+      "grad_norm": 4.042450904846191,
+      "learning_rate": 3.2873638915588593e-07,
+      "loss": 0.7872,
+      "step": 1095
+    },
+    {
+      "epoch": 2.779521318750991,
+      "grad_norm": 4.343115329742432,
+      "learning_rate": 3.2126315807946786e-07,
+      "loss": 0.7596,
+      "step": 1096
+    },
+    {
+      "epoch": 2.7820573783483913,
+      "grad_norm": 4.098153591156006,
+      "learning_rate": 3.138744649277181e-07,
+      "loss": 0.7641,
+      "step": 1097
+    },
+    {
+      "epoch": 2.7845934379457917,
+      "grad_norm": 3.9757375717163086,
+      "learning_rate": 3.065703742364168e-07,
+      "loss": 0.7574,
+      "step": 1098
+    },
+    {
+      "epoch": 2.787129497543192,
+      "grad_norm": 4.160877227783203,
+      "learning_rate": 2.9935094980238276e-07,
+      "loss": 0.7696,
+      "step": 1099
+    },
+    {
+      "epoch": 2.7896655571405926,
+      "grad_norm": 3.812747001647949,
+      "learning_rate": 2.922162546829355e-07,
+      "loss": 0.7784,
+      "step": 1100
+    },
+    {
+      "epoch": 2.7896655571405926,
+      "eval_loss": NaN,
+      "eval_runtime": 206.9857,
+      "eval_samples_per_second": 40.172,
+      "eval_steps_per_second": 5.025,
+      "step": 1100
+    },
+    {
+      "epoch": 2.7922016167379935,
+      "grad_norm": 3.691394805908203,
+      "learning_rate": 2.8516635119532845e-07,
+      "loss": 0.7503,
+      "step": 1101
+    },
+    {
+      "epoch": 2.794737676335394,
+      "grad_norm": 33.4482421875,
+      "learning_rate": 2.7820130091621325e-07,
+      "loss": 0.7734,
+      "step": 1102
+    },
+    {
+      "epoch": 2.7972737359327944,
+      "grad_norm": 4.202887535095215,
+      "learning_rate": 2.71321164681102e-07,
+      "loss": 0.7644,
+      "step": 1103
+    },
+    {
+      "epoch": 2.799809795530195,
+      "grad_norm": 4.044505596160889,
+      "learning_rate": 2.6452600258383144e-07,
+      "loss": 0.7785,
+      "step": 1104
+    },
+    {
+      "epoch": 2.8023458551275953,
+      "grad_norm": 4.082708358764648,
+      "learning_rate": 2.578158739760395e-07,
+      "loss": 0.77,
+      "step": 1105
+    },
+    {
+      "epoch": 2.804881914724996,
+      "grad_norm": 3.879438877105713,
+      "learning_rate": 2.511908374666516e-07,
+      "loss": 0.7674,
+      "step": 1106
+    },
+    {
+      "epoch": 2.8074179743223966,
+      "grad_norm": 3.7508392333984375,
+      "learning_rate": 2.446509509213624e-07,
+      "loss": 0.7645,
+      "step": 1107
+    },
+    {
+      "epoch": 2.809954033919797,
+      "grad_norm": 3.896162748336792,
+      "learning_rate": 2.3819627146213443e-07,
+      "loss": 0.7809,
+      "step": 1108
+    },
+    {
+      "epoch": 2.812490093517198,
+      "grad_norm": 4.181882381439209,
+      "learning_rate": 2.3182685546669692e-07,
+      "loss": 0.7542,
+      "step": 1109
+    },
+    {
+      "epoch": 2.815026153114598,
+      "grad_norm": 3.9994637966156006,
+      "learning_rate": 2.2554275856805253e-07,
+      "loss": 0.7595,
+      "step": 1110
+    },
+    {
+      "epoch": 2.817562212711999,
+      "grad_norm": 3.9829647541046143,
+      "learning_rate": 2.1934403565399666e-07,
+      "loss": 0.773,
+      "step": 1111
+    },
+    {
+      "epoch": 2.8200982723093992,
+      "grad_norm": 4.087826251983643,
+      "learning_rate": 2.132307408666312e-07,
+      "loss": 0.774,
+      "step": 1112
+    },
+    {
+      "epoch": 2.8226343319067997,
+      "grad_norm": 3.939371109008789,
+      "learning_rate": 2.0720292760189942e-07,
+      "loss": 0.7727,
+      "step": 1113
+    },
+    {
+      "epoch": 2.8251703915042006,
+      "grad_norm": 4.170963764190674,
+      "learning_rate": 2.0126064850910954e-07,
+      "loss": 0.7873,
+      "step": 1114
+    },
+    {
+      "epoch": 2.827706451101601,
+      "grad_norm": 3.9816012382507324,
+      "learning_rate": 1.9540395549048297e-07,
+      "loss": 0.783,
+      "step": 1115
+    },
+    {
+      "epoch": 2.8302425106990015,
+      "grad_norm": 4.0752997398376465,
+      "learning_rate": 1.896328997006991e-07,
+      "loss": 0.7742,
+      "step": 1116
+    },
+    {
+      "epoch": 2.832778570296402,
+      "grad_norm": 3.971782922744751,
+      "learning_rate": 1.8394753154644784e-07,
+      "loss": 0.7417,
+      "step": 1117
+    },
+    {
+      "epoch": 2.8353146298938023,
+      "grad_norm": 3.952244281768799,
+      "learning_rate": 1.7834790068598674e-07,
+      "loss": 0.7758,
+      "step": 1118
+    },
+    {
+      "epoch": 2.8378506894912032,
+      "grad_norm": 3.9631309509277344,
+      "learning_rate": 1.728340560287134e-07,
+      "loss": 0.7785,
+      "step": 1119
+    },
+    {
+      "epoch": 2.8403867490886037,
+      "grad_norm": 3.875110387802124,
+      "learning_rate": 1.674060457347304e-07,
+      "loss": 0.7633,
+      "step": 1120
+    },
+    {
+      "epoch": 2.842922808686004,
+      "grad_norm": 3.884068012237549,
+      "learning_rate": 1.620639172144356e-07,
+      "loss": 0.7594,
+      "step": 1121
+    },
+    {
+      "epoch": 2.8454588682834046,
+      "grad_norm": 3.9125704765319824,
+      "learning_rate": 1.568077171280935e-07,
+      "loss": 0.759,
+      "step": 1122
+    },
+    {
+      "epoch": 2.847994927880805,
+      "grad_norm": 3.95465087890625,
+      "learning_rate": 1.5163749138544237e-07,
+      "loss": 0.7853,
+      "step": 1123
+    },
+    {
+      "epoch": 2.850530987478206,
+      "grad_norm": 3.9252095222473145,
+      "learning_rate": 1.465532851452811e-07,
+      "loss": 0.7748,
+      "step": 1124
+    },
+    {
+      "epoch": 2.8530670470756063,
+      "grad_norm": 3.891159772872925,
+      "learning_rate": 1.415551428150852e-07,
+      "loss": 0.7876,
+      "step": 1125
+    },
+    {
+      "epoch": 2.8556031066730068,
+      "grad_norm": 3.771494150161743,
+      "learning_rate": 1.366431080506081e-07,
+      "loss": 0.7611,
+      "step": 1126
+    },
+    {
+      "epoch": 2.858139166270407,
+      "grad_norm": 4.0067009925842285,
+      "learning_rate": 1.3181722375551152e-07,
+      "loss": 0.7627,
+      "step": 1127
+    },
+    {
+      "epoch": 2.8606752258678076,
+      "grad_norm": 3.869405746459961,
+      "learning_rate": 1.270775320809814e-07,
+      "loss": 0.7413,
+      "step": 1128
+    },
+    {
+      "epoch": 2.8632112854652085,
+      "grad_norm": 4.035890102386475,
+      "learning_rate": 1.2242407442536352e-07,
+      "loss": 0.7732,
+      "step": 1129
+    },
+    {
+      "epoch": 2.865747345062609,
+      "grad_norm": 3.938883066177368,
+      "learning_rate": 1.1785689143380186e-07,
+      "loss": 0.7386,
+      "step": 1130
+    },
+    {
+      "epoch": 2.8682834046600094,
+      "grad_norm": 4.010357856750488,
+      "learning_rate": 1.1337602299788308e-07,
+      "loss": 0.7695,
+      "step": 1131
+    },
+    {
+      "epoch": 2.8708194642574103,
+      "grad_norm": 3.9605135917663574,
+      "learning_rate": 1.089815082552892e-07,
+      "loss": 0.7576,
+      "step": 1132
+    },
+    {
+      "epoch": 2.8733555238548107,
+      "grad_norm": 4.002065181732178,
+      "learning_rate": 1.0467338558945328e-07,
+      "loss": 0.7544,
+      "step": 1133
+    },
+    {
+      "epoch": 2.875891583452211,
+      "grad_norm": 4.0115838050842285,
+      "learning_rate": 1.0045169262922761e-07,
+      "loss": 0.7839,
+      "step": 1134
+    },
+    {
+      "epoch": 2.8784276430496116,
+      "grad_norm": 3.8898210525512695,
+      "learning_rate": 9.631646624855162e-08,
+      "loss": 0.7546,
+      "step": 1135
+    },
+    {
+      "epoch": 2.880963702647012,
+      "grad_norm": 4.145599842071533,
+      "learning_rate": 9.226774256613225e-08,
+      "loss": 0.7445,
+      "step": 1136
+    },
+    {
+      "epoch": 2.883499762244413,
+      "grad_norm": 3.827849864959717,
+      "learning_rate": 8.830555694512632e-08,
+      "loss": 0.7729,
+      "step": 1137
+    },
+    {
+      "epoch": 2.8860358218418134,
+      "grad_norm": 3.8289098739624023,
+      "learning_rate": 8.442994399283532e-08,
+      "loss": 0.782,
+      "step": 1138
+    },
+    {
+      "epoch": 2.888571881439214,
+      "grad_norm": 3.8252921104431152,
+      "learning_rate": 8.064093756039782e-08,
+      "loss": 0.7613,
+      "step": 1139
+    },
+    {
+      "epoch": 2.8911079410366143,
+      "grad_norm": 3.922124147415161,
+      "learning_rate": 7.693857074249966e-08,
+      "loss": 0.7442,
+      "step": 1140
+    },
+    {
+      "epoch": 2.8936440006340147,
+      "grad_norm": 3.9693527221679688,
+      "learning_rate": 7.332287587707877e-08,
+      "loss": 0.7661,
+      "step": 1141
+    },
+    {
+      "epoch": 2.8961800602314156,
+      "grad_norm": 3.9805407524108887,
+      "learning_rate": 6.979388454504744e-08,
+      "loss": 0.7579,
+      "step": 1142
+    },
+    {
+      "epoch": 2.898716119828816,
+      "grad_norm": 3.8455755710601807,
+      "learning_rate": 6.635162757001712e-08,
+      "loss": 0.7647,
+      "step": 1143
+    },
+    {
+      "epoch": 2.9012521794262165,
+      "grad_norm": 3.7146923542022705,
+      "learning_rate": 6.299613501802304e-08,
+      "loss": 0.7737,
+      "step": 1144
+    },
+    {
+      "epoch": 2.903788239023617,
+      "grad_norm": 3.733086109161377,
+      "learning_rate": 5.972743619726883e-08,
+      "loss": 0.7709,
+      "step": 1145
+    },
+    {
+      "epoch": 2.9063242986210174,
+      "grad_norm": 3.933823347091675,
+      "learning_rate": 5.654555965786457e-08,
+      "loss": 0.7697,
+      "step": 1146
+    },
+    {
+      "epoch": 2.9088603582184183,
+      "grad_norm": 3.916609048843384,
+      "learning_rate": 5.3450533191581336e-08,
+      "loss": 0.7797,
+      "step": 1147
+    },
+    {
+      "epoch": 2.9113964178158187,
+      "grad_norm": 3.7848339080810547,
+      "learning_rate": 5.0442383831608155e-08,
+      "loss": 0.7703,
+      "step": 1148
+    },
+    {
+      "epoch": 2.913932477413219,
+      "grad_norm": 3.7588272094726562,
+      "learning_rate": 4.752113785231438e-08,
+      "loss": 0.7469,
+      "step": 1149
+    },
+    {
+      "epoch": 2.9164685370106196,
+      "grad_norm": 3.9002275466918945,
+      "learning_rate": 4.4686820769018755e-08,
+      "loss": 0.797,
+      "step": 1150
+    },
+    {
+      "epoch": 2.9164685370106196,
+      "eval_loss": NaN,
+      "eval_runtime": 206.4346,
+      "eval_samples_per_second": 40.279,
+      "eval_steps_per_second": 5.038,
+      "step": 1150
+    },
+    {
+      "epoch": 2.91900459660802,
+      "grad_norm": 3.942675828933716,
+      "learning_rate": 4.1939457337771824e-08,
+      "loss": 0.7743,
+      "step": 1151
+    },
+    {
+      "epoch": 2.921540656205421,
+      "grad_norm": 3.9600582122802734,
+      "learning_rate": 3.9279071555134994e-08,
+      "loss": 0.7655,
+      "step": 1152
+    },
+    {
+      "epoch": 2.9240767158028214,
+      "grad_norm": 3.982541561126709,
+      "learning_rate": 3.670568665797181e-08,
+      "loss": 0.7512,
+      "step": 1153
+    },
+    {
+      "epoch": 2.926612775400222,
+      "grad_norm": 3.8570940494537354,
+      "learning_rate": 3.421932512324699e-08,
+      "loss": 0.7526,
+      "step": 1154
+    },
+    {
+      "epoch": 2.9291488349976227,
+      "grad_norm": 3.814025402069092,
+      "learning_rate": 3.182000866782664e-08,
+      "loss": 0.7679,
+      "step": 1155
+    },
+    {
+      "epoch": 2.931684894595023,
+      "grad_norm": 3.9061741828918457,
+      "learning_rate": 2.9507758248292774e-08,
+      "loss": 0.7924,
+      "step": 1156
+    },
+    {
+      "epoch": 2.9342209541924236,
+      "grad_norm": 3.8834214210510254,
+      "learning_rate": 2.728259406075684e-08,
+      "loss": 0.7852,
+      "step": 1157
+    },
+    {
+      "epoch": 2.936757013789824,
+      "grad_norm": 4.070812225341797,
+      "learning_rate": 2.51445355406843e-08,
+      "loss": 0.7676,
+      "step": 1158
+    },
+    {
+      "epoch": 2.9392930733872245,
+      "grad_norm": 4.090744495391846,
+      "learning_rate": 2.3093601362728092e-08,
+      "loss": 0.7541,
+      "step": 1159
+    },
+    {
+      "epoch": 2.9418291329846253,
+      "grad_norm": 3.853752613067627,
+      "learning_rate": 2.1129809440558757e-08,
+      "loss": 0.7531,
+      "step": 1160
+    },
+    {
+      "epoch": 2.944365192582026,
+      "grad_norm": 3.790019989013672,
+      "learning_rate": 1.925317692671347e-08,
+      "loss": 0.759,
+      "step": 1161
+    },
+    {
+      "epoch": 2.9469012521794262,
+      "grad_norm": 3.9019384384155273,
+      "learning_rate": 1.7463720212445022e-08,
+      "loss": 0.7591,
+      "step": 1162
+    },
+    {
+      "epoch": 2.9494373117768267,
+      "grad_norm": 3.9709768295288086,
+      "learning_rate": 1.5761454927577526e-08,
+      "loss": 0.774,
+      "step": 1163
+    },
+    {
+      "epoch": 2.951973371374227,
+      "grad_norm": 3.91044282913208,
+      "learning_rate": 1.4146395940370927e-08,
+      "loss": 0.7434,
+      "step": 1164
+    },
+    {
+      "epoch": 2.954509430971628,
+      "grad_norm": 3.8549511432647705,
+      "learning_rate": 1.2618557357391147e-08,
+      "loss": 0.7847,
+      "step": 1165
+    },
+    {
+      "epoch": 2.9570454905690284,
+      "grad_norm": 3.7837891578674316,
+      "learning_rate": 1.1177952523385715e-08,
+      "loss": 0.7809,
+      "step": 1166
+    },
+    {
+      "epoch": 2.959581550166429,
+      "grad_norm": 4.0853447914123535,
+      "learning_rate": 9.82459402116831e-09,
+      "loss": 0.7617,
+      "step": 1167
+    },
+    {
+      "epoch": 2.9621176097638293,
+      "grad_norm": 3.8751251697540283,
+      "learning_rate": 8.558493671507739e-09,
+      "loss": 0.7668,
+      "step": 1168
+    },
+    {
+      "epoch": 2.9646536693612298,
+      "grad_norm": 3.9072928428649902,
+      "learning_rate": 7.37966253302802e-09,
+      "loss": 0.7699,
+      "step": 1169
+    },
+    {
+      "epoch": 2.9671897289586306,
+      "grad_norm": 3.995150089263916,
+      "learning_rate": 6.288110902107347e-09,
+      "loss": 0.7687,
+      "step": 1170
+    },
+    {
+      "epoch": 2.969725788556031,
+      "grad_norm": 3.7405359745025635,
+      "learning_rate": 5.283848312789275e-09,
+      "loss": 0.7691,
+      "step": 1171
+    },
+    {
+      "epoch": 2.9722618481534315,
+      "grad_norm": 3.91318941116333,
+      "learning_rate": 4.366883536702782e-09,
+      "loss": 0.7712,
+      "step": 1172
+    },
+    {
+      "epoch": 2.9747979077508324,
+      "grad_norm": 3.920619487762451,
+      "learning_rate": 3.5372245829801142e-09,
+      "loss": 0.7742,
+      "step": 1173
+    },
+    {
+      "epoch": 2.9773339673482324,
+      "grad_norm": 3.744755983352661,
+      "learning_rate": 2.794878698193504e-09,
+      "loss": 0.7621,
+      "step": 1174
+    },
+    {
+      "epoch": 2.9798700269456333,
+      "grad_norm": 4.027815341949463,
+      "learning_rate": 2.1398523662841118e-09,
+      "loss": 0.7718,
+      "step": 1175
+    },
+    {
+      "epoch": 2.9824060865430337,
+      "grad_norm": 3.725977659225464,
+      "learning_rate": 1.5721513085131811e-09,
+      "loss": 0.7392,
+      "step": 1176
+    },
+    {
+      "epoch": 2.984942146140434,
+      "grad_norm": 4.051129341125488,
+      "learning_rate": 1.0917804834065237e-09,
+      "loss": 0.7499,
+      "step": 1177
+    },
+    {
+      "epoch": 2.987478205737835,
+      "grad_norm": 3.8135509490966797,
+      "learning_rate": 6.987440867134432e-10,
+      "loss": 0.7814,
+      "step": 1178
+    },
+    {
+      "epoch": 2.9900142653352355,
+      "grad_norm": 4.0044755935668945,
+      "learning_rate": 3.930455513689868e-10,
+      "loss": 0.7588,
+      "step": 1179
+    },
+    {
+      "epoch": 2.992550324932636,
+      "grad_norm": 3.93560791015625,
+      "learning_rate": 1.7468754746507998e-10,
+      "loss": 0.753,
+      "step": 1180
+    },
+    {
+      "epoch": 2.9950863845300364,
+      "grad_norm": 3.733422040939331,
+      "learning_rate": 4.367198222832159e-11,
+      "loss": 0.7567,
+      "step": 1181
+    },
+    {
+      "epoch": 2.997622444127437,
+      "grad_norm": 3.872838020324707,
+      "learning_rate": 0.0,
+      "loss": 0.7427,
+      "step": 1182
+    },
+    {
+      "epoch": 2.997622444127437,
+      "step": 1182,
+      "total_flos": 5.2878430857678815e+19,
+      "train_loss": 0.7602980379424926,
+      "train_runtime": 103647.3691,
+      "train_samples_per_second": 5.843,
+      "train_steps_per_second": 0.011
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 1182,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 5.2878430857678815e+19,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}