diff --git "a/checkpoint-690/trainer_state.json" "b/checkpoint-690/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-690/trainer_state.json"
@@ -0,0 +1,4863 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 5.994236311239193,
+  "eval_steps": 500,
+  "global_step": 690,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.008645533141210375,
+      "grad_norm": 34.72848892211914,
+      "learning_rate": 5.0000000000000004e-08,
+      "loss": 2.477,
+      "step": 1
+    },
+    {
+      "epoch": 0.01729106628242075,
+      "grad_norm": 33.733909606933594,
+      "learning_rate": 1.0000000000000001e-07,
+      "loss": 2.4134,
+      "step": 2
+    },
+    {
+      "epoch": 0.025936599423631124,
+      "grad_norm": 34.543819427490234,
+      "learning_rate": 1.5000000000000002e-07,
+      "loss": 2.4467,
+      "step": 3
+    },
+    {
+      "epoch": 0.0345821325648415,
+      "grad_norm": 35.37831115722656,
+      "learning_rate": 2.0000000000000002e-07,
+      "loss": 2.4801,
+      "step": 4
+    },
+    {
+      "epoch": 0.043227665706051875,
+      "grad_norm": 33.97856140136719,
+      "learning_rate": 2.5000000000000004e-07,
+      "loss": 2.4422,
+      "step": 5
+    },
+    {
+      "epoch": 0.05187319884726225,
+      "grad_norm": 34.11160659790039,
+      "learning_rate": 3.0000000000000004e-07,
+      "loss": 2.4003,
+      "step": 6
+    },
+    {
+      "epoch": 0.06051873198847262,
+      "grad_norm": 34.086463928222656,
+      "learning_rate": 3.5000000000000004e-07,
+      "loss": 2.4211,
+      "step": 7
+    },
+    {
+      "epoch": 0.069164265129683,
+      "grad_norm": 33.96665573120117,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 2.4054,
+      "step": 8
+    },
+    {
+      "epoch": 0.07780979827089338,
+      "grad_norm": 34.871307373046875,
+      "learning_rate": 4.5000000000000003e-07,
+      "loss": 2.4077,
+      "step": 9
+    },
+    {
+      "epoch": 0.08645533141210375,
+      "grad_norm": 33.91160583496094,
+      "learning_rate": 5.000000000000001e-07,
+      "loss": 2.3598,
+      "step": 10
+    },
+    {
+      "epoch": 0.09510086455331412,
+      "grad_norm": 33.40217971801758,
+      "learning_rate": 5.5e-07,
+      "loss": 2.3209,
+      "step": 11
+    },
+    {
+      "epoch": 0.1037463976945245,
+      "grad_norm": 33.771121978759766,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 2.3178,
+      "step": 12
+    },
+    {
+      "epoch": 0.11239193083573487,
+      "grad_norm": 33.577476501464844,
+      "learning_rate": 6.5e-07,
+      "loss": 2.3107,
+      "step": 13
+    },
+    {
+      "epoch": 0.12103746397694524,
+      "grad_norm": 31.330514907836914,
+      "learning_rate": 7.000000000000001e-07,
+      "loss": 2.095,
+      "step": 14
+    },
+    {
+      "epoch": 0.12968299711815562,
+      "grad_norm": 33.60646057128906,
+      "learning_rate": 7.5e-07,
+      "loss": 2.1663,
+      "step": 15
+    },
+    {
+      "epoch": 0.138328530259366,
+      "grad_norm": 31.96607208251953,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 1.9986,
+      "step": 16
+    },
+    {
+      "epoch": 0.14697406340057637,
+      "grad_norm": 32.33183288574219,
+      "learning_rate": 8.500000000000001e-07,
+      "loss": 1.9401,
+      "step": 17
+    },
+    {
+      "epoch": 0.15561959654178675,
+      "grad_norm": 33.50197219848633,
+      "learning_rate": 9.000000000000001e-07,
+      "loss": 1.8529,
+      "step": 18
+    },
+    {
+      "epoch": 0.1642651296829971,
+      "grad_norm": 33.55098342895508,
+      "learning_rate": 9.500000000000001e-07,
+      "loss": 1.7087,
+      "step": 19
+    },
+    {
+      "epoch": 0.1729106628242075,
+      "grad_norm": 33.90129852294922,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 1.6324,
+      "step": 20
+    },
+    {
+      "epoch": 0.18155619596541786,
+      "grad_norm": 33.738037109375,
+      "learning_rate": 1.0500000000000001e-06,
+      "loss": 1.4896,
+      "step": 21
+    },
+    {
+      "epoch": 0.19020172910662825,
+      "grad_norm": 33.6258659362793,
+      "learning_rate": 1.1e-06,
+      "loss": 1.3664,
+      "step": 22
+    },
+    {
+      "epoch": 0.1988472622478386,
+      "grad_norm": 30.822349548339844,
+      "learning_rate": 1.1500000000000002e-06,
+      "loss": 1.2193,
+      "step": 23
+    },
+    {
+      "epoch": 0.207492795389049,
+      "grad_norm": 29.608501434326172,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 1.074,
+      "step": 24
+    },
+    {
+      "epoch": 0.21613832853025935,
+      "grad_norm": 27.651105880737305,
+      "learning_rate": 1.25e-06,
+      "loss": 0.938,
+      "step": 25
+    },
+    {
+      "epoch": 0.22478386167146974,
+      "grad_norm": 29.479637145996094,
+      "learning_rate": 1.3e-06,
+      "loss": 0.7728,
+      "step": 26
+    },
+    {
+      "epoch": 0.2334293948126801,
+      "grad_norm": 29.068634033203125,
+      "learning_rate": 1.3500000000000002e-06,
+      "loss": 0.6051,
+      "step": 27
+    },
+    {
+      "epoch": 0.2420749279538905,
+      "grad_norm": 24.850099563598633,
+      "learning_rate": 1.4000000000000001e-06,
+      "loss": 0.4463,
+      "step": 28
+    },
+    {
+      "epoch": 0.2507204610951009,
+      "grad_norm": 22.095216751098633,
+      "learning_rate": 1.45e-06,
+      "loss": 0.3489,
+      "step": 29
+    },
+    {
+      "epoch": 0.25936599423631124,
+      "grad_norm": 19.491201400756836,
+      "learning_rate": 1.5e-06,
+      "loss": 0.261,
+      "step": 30
+    },
+    {
+      "epoch": 0.2680115273775216,
+      "grad_norm": 14.492341041564941,
+      "learning_rate": 1.5500000000000002e-06,
+      "loss": 0.1707,
+      "step": 31
+    },
+    {
+      "epoch": 0.276657060518732,
+      "grad_norm": 6.020577907562256,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 0.1151,
+      "step": 32
+    },
+    {
+      "epoch": 0.28530259365994237,
+      "grad_norm": 2.90791916847229,
+      "learning_rate": 1.6500000000000003e-06,
+      "loss": 0.0976,
+      "step": 33
+    },
+    {
+      "epoch": 0.29394812680115273,
+      "grad_norm": 2.637803554534912,
+      "learning_rate": 1.7000000000000002e-06,
+      "loss": 0.0911,
+      "step": 34
+    },
+    {
+      "epoch": 0.3025936599423631,
+      "grad_norm": 1.804861068725586,
+      "learning_rate": 1.75e-06,
+      "loss": 0.0834,
+      "step": 35
+    },
+    {
+      "epoch": 0.3112391930835735,
+      "grad_norm": 2.049024820327759,
+      "learning_rate": 1.8000000000000001e-06,
+      "loss": 0.0842,
+      "step": 36
+    },
+    {
+      "epoch": 0.31988472622478387,
+      "grad_norm": 1.3263498544692993,
+      "learning_rate": 1.85e-06,
+      "loss": 0.0744,
+      "step": 37
+    },
+    {
+      "epoch": 0.3285302593659942,
+      "grad_norm": 1.7187089920043945,
+      "learning_rate": 1.9000000000000002e-06,
+      "loss": 0.0783,
+      "step": 38
+    },
+    {
+      "epoch": 0.3371757925072046,
+      "grad_norm": 1.3925131559371948,
+      "learning_rate": 1.9500000000000004e-06,
+      "loss": 0.073,
+      "step": 39
+    },
+    {
+      "epoch": 0.345821325648415,
+      "grad_norm": 1.2181739807128906,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 0.0749,
+      "step": 40
+    },
+    {
+      "epoch": 0.35446685878962536,
+      "grad_norm": 1.0519245862960815,
+      "learning_rate": 2.05e-06,
+      "loss": 0.0692,
+      "step": 41
+    },
+    {
+      "epoch": 0.3631123919308357,
+      "grad_norm": 0.9188923835754395,
+      "learning_rate": 2.1000000000000002e-06,
+      "loss": 0.0726,
+      "step": 42
+    },
+    {
+      "epoch": 0.37175792507204614,
+      "grad_norm": 0.8273228406906128,
+      "learning_rate": 2.15e-06,
+      "loss": 0.0651,
+      "step": 43
+    },
+    {
+      "epoch": 0.3804034582132565,
+      "grad_norm": 0.9098994135856628,
+      "learning_rate": 2.2e-06,
+      "loss": 0.066,
+      "step": 44
+    },
+    {
+      "epoch": 0.38904899135446686,
+      "grad_norm": 0.8456838726997375,
+      "learning_rate": 2.25e-06,
+      "loss": 0.0646,
+      "step": 45
+    },
+    {
+      "epoch": 0.3976945244956772,
+      "grad_norm": 0.8240940570831299,
+      "learning_rate": 2.3000000000000004e-06,
+      "loss": 0.0604,
+      "step": 46
+    },
+    {
+      "epoch": 0.40634005763688763,
+      "grad_norm": 1.111759901046753,
+      "learning_rate": 2.35e-06,
+      "loss": 0.0589,
+      "step": 47
+    },
+    {
+      "epoch": 0.414985590778098,
+      "grad_norm": 0.9933035373687744,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 0.0613,
+      "step": 48
+    },
+    {
+      "epoch": 0.42363112391930835,
+      "grad_norm": 0.7491716742515564,
+      "learning_rate": 2.4500000000000003e-06,
+      "loss": 0.0583,
+      "step": 49
+    },
+    {
+      "epoch": 0.4322766570605187,
+      "grad_norm": 0.9089523553848267,
+      "learning_rate": 2.5e-06,
+      "loss": 0.0558,
+      "step": 50
+    },
+    {
+      "epoch": 0.4409221902017291,
+      "grad_norm": 0.7088611721992493,
+      "learning_rate": 2.55e-06,
+      "loss": 0.0553,
+      "step": 51
+    },
+    {
+      "epoch": 0.4495677233429395,
+      "grad_norm": 0.7892571091651917,
+      "learning_rate": 2.6e-06,
+      "loss": 0.064,
+      "step": 52
+    },
+    {
+      "epoch": 0.45821325648414984,
+      "grad_norm": 0.9248467087745667,
+      "learning_rate": 2.6500000000000005e-06,
+      "loss": 0.0653,
+      "step": 53
+    },
+    {
+      "epoch": 0.4668587896253602,
+      "grad_norm": 0.7224969863891602,
+      "learning_rate": 2.7000000000000004e-06,
+      "loss": 0.0611,
+      "step": 54
+    },
+    {
+      "epoch": 0.4755043227665706,
+      "grad_norm": 0.8231533765792847,
+      "learning_rate": 2.7500000000000004e-06,
+      "loss": 0.0617,
+      "step": 55
+    },
+    {
+      "epoch": 0.484149855907781,
+      "grad_norm": 0.7306967973709106,
+      "learning_rate": 2.8000000000000003e-06,
+      "loss": 0.0571,
+      "step": 56
+    },
+    {
+      "epoch": 0.49279538904899134,
+      "grad_norm": 0.8415323495864868,
+      "learning_rate": 2.85e-06,
+      "loss": 0.0548,
+      "step": 57
+    },
+    {
+      "epoch": 0.5014409221902018,
+      "grad_norm": 1.5560295581817627,
+      "learning_rate": 2.9e-06,
+      "loss": 0.0672,
+      "step": 58
+    },
+    {
+      "epoch": 0.5100864553314121,
+      "grad_norm": 0.9170955419540405,
+      "learning_rate": 2.95e-06,
+      "loss": 0.068,
+      "step": 59
+    },
+    {
+      "epoch": 0.5187319884726225,
+      "grad_norm": 0.6508005857467651,
+      "learning_rate": 3e-06,
+      "loss": 0.057,
+      "step": 60
+    },
+    {
+      "epoch": 0.5273775216138329,
+      "grad_norm": 0.8307355642318726,
+      "learning_rate": 3.05e-06,
+      "loss": 0.0598,
+      "step": 61
+    },
+    {
+      "epoch": 0.5360230547550432,
+      "grad_norm": 1.11078679561615,
+      "learning_rate": 3.1000000000000004e-06,
+      "loss": 0.0575,
+      "step": 62
+    },
+    {
+      "epoch": 0.5446685878962536,
+      "grad_norm": 1.0765758752822876,
+      "learning_rate": 3.1500000000000003e-06,
+      "loss": 0.0615,
+      "step": 63
+    },
+    {
+      "epoch": 0.553314121037464,
+      "grad_norm": 0.8381508588790894,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 0.0606,
+      "step": 64
+    },
+    {
+      "epoch": 0.5619596541786743,
+      "grad_norm": 1.007628321647644,
+      "learning_rate": 3.2500000000000002e-06,
+      "loss": 0.0572,
+      "step": 65
+    },
+    {
+      "epoch": 0.5706051873198847,
+      "grad_norm": 0.7254197597503662,
+      "learning_rate": 3.3000000000000006e-06,
+      "loss": 0.0543,
+      "step": 66
+    },
+    {
+      "epoch": 0.579250720461095,
+      "grad_norm": 0.5906903147697449,
+      "learning_rate": 3.3500000000000005e-06,
+      "loss": 0.0557,
+      "step": 67
+    },
+    {
+      "epoch": 0.5878962536023055,
+      "grad_norm": 0.6791537404060364,
+      "learning_rate": 3.4000000000000005e-06,
+      "loss": 0.0516,
+      "step": 68
+    },
+    {
+      "epoch": 0.5965417867435159,
+      "grad_norm": 0.6390945315361023,
+      "learning_rate": 3.45e-06,
+      "loss": 0.0529,
+      "step": 69
+    },
+    {
+      "epoch": 0.6051873198847262,
+      "grad_norm": 0.61552494764328,
+      "learning_rate": 3.5e-06,
+      "loss": 0.0554,
+      "step": 70
+    },
+    {
+      "epoch": 0.6138328530259366,
+      "grad_norm": 0.6545206308364868,
+      "learning_rate": 3.5500000000000003e-06,
+      "loss": 0.0545,
+      "step": 71
+    },
+    {
+      "epoch": 0.622478386167147,
+      "grad_norm": 0.9062793254852295,
+      "learning_rate": 3.6000000000000003e-06,
+      "loss": 0.0559,
+      "step": 72
+    },
+    {
+      "epoch": 0.6311239193083573,
+      "grad_norm": 0.9642562866210938,
+      "learning_rate": 3.65e-06,
+      "loss": 0.0485,
+      "step": 73
+    },
+    {
+      "epoch": 0.6397694524495677,
+      "grad_norm": 0.7286660075187683,
+      "learning_rate": 3.7e-06,
+      "loss": 0.0564,
+      "step": 74
+    },
+    {
+      "epoch": 0.6484149855907781,
+      "grad_norm": 0.6356053948402405,
+      "learning_rate": 3.7500000000000005e-06,
+      "loss": 0.053,
+      "step": 75
+    },
+    {
+      "epoch": 0.6570605187319885,
+      "grad_norm": 0.706794261932373,
+      "learning_rate": 3.8000000000000005e-06,
+      "loss": 0.0531,
+      "step": 76
+    },
+    {
+      "epoch": 0.6657060518731989,
+      "grad_norm": 0.6616448163986206,
+      "learning_rate": 3.85e-06,
+      "loss": 0.0504,
+      "step": 77
+    },
+    {
+      "epoch": 0.6743515850144092,
+      "grad_norm": 0.7465748190879822,
+      "learning_rate": 3.900000000000001e-06,
+      "loss": 0.0495,
+      "step": 78
+    },
+    {
+      "epoch": 0.6829971181556196,
+      "grad_norm": 0.8153467774391174,
+      "learning_rate": 3.95e-06,
+      "loss": 0.0495,
+      "step": 79
+    },
+    {
+      "epoch": 0.69164265129683,
+      "grad_norm": 0.7728897333145142,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.0434,
+      "step": 80
+    },
+    {
+      "epoch": 0.7002881844380403,
+      "grad_norm": 0.6173391938209534,
+      "learning_rate": 4.05e-06,
+      "loss": 0.0432,
+      "step": 81
+    },
+    {
+      "epoch": 0.7089337175792507,
+      "grad_norm": 0.7128047943115234,
+      "learning_rate": 4.1e-06,
+      "loss": 0.0512,
+      "step": 82
+    },
+    {
+      "epoch": 0.7175792507204611,
+      "grad_norm": 0.6098653674125671,
+      "learning_rate": 4.15e-06,
+      "loss": 0.0415,
+      "step": 83
+    },
+    {
+      "epoch": 0.7262247838616714,
+      "grad_norm": 0.7464293241500854,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 0.0453,
+      "step": 84
+    },
+    {
+      "epoch": 0.7348703170028819,
+      "grad_norm": 0.8350300788879395,
+      "learning_rate": 4.25e-06,
+      "loss": 0.0463,
+      "step": 85
+    },
+    {
+      "epoch": 0.7435158501440923,
+      "grad_norm": 0.7880110740661621,
+      "learning_rate": 4.3e-06,
+      "loss": 0.0426,
+      "step": 86
+    },
+    {
+      "epoch": 0.7521613832853026,
+      "grad_norm": 0.77886962890625,
+      "learning_rate": 4.350000000000001e-06,
+      "loss": 0.0476,
+      "step": 87
+    },
+    {
+      "epoch": 0.760806916426513,
+      "grad_norm": 1.0404386520385742,
+      "learning_rate": 4.4e-06,
+      "loss": 0.0501,
+      "step": 88
+    },
+    {
+      "epoch": 0.7694524495677233,
+      "grad_norm": 0.5827208757400513,
+      "learning_rate": 4.450000000000001e-06,
+      "loss": 0.0396,
+      "step": 89
+    },
+    {
+      "epoch": 0.7780979827089337,
+      "grad_norm": 0.5928618907928467,
+      "learning_rate": 4.5e-06,
+      "loss": 0.0438,
+      "step": 90
+    },
+    {
+      "epoch": 0.7867435158501441,
+      "grad_norm": 0.5311946272850037,
+      "learning_rate": 4.5500000000000005e-06,
+      "loss": 0.0388,
+      "step": 91
+    },
+    {
+      "epoch": 0.7953890489913544,
+      "grad_norm": 0.7609073519706726,
+      "learning_rate": 4.600000000000001e-06,
+      "loss": 0.0403,
+      "step": 92
+    },
+    {
+      "epoch": 0.8040345821325648,
+      "grad_norm": 0.6055853962898254,
+      "learning_rate": 4.65e-06,
+      "loss": 0.0402,
+      "step": 93
+    },
+    {
+      "epoch": 0.8126801152737753,
+      "grad_norm": 0.8020023703575134,
+      "learning_rate": 4.7e-06,
+      "loss": 0.0415,
+      "step": 94
+    },
+    {
+      "epoch": 0.8213256484149856,
+      "grad_norm": 0.9083772301673889,
+      "learning_rate": 4.75e-06,
+      "loss": 0.0428,
+      "step": 95
+    },
+    {
+      "epoch": 0.829971181556196,
+      "grad_norm": 0.6658433079719543,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 0.0381,
+      "step": 96
+    },
+    {
+      "epoch": 0.8386167146974063,
+      "grad_norm": 0.925826907157898,
+      "learning_rate": 4.85e-06,
+      "loss": 0.0465,
+      "step": 97
+    },
+    {
+      "epoch": 0.8472622478386167,
+      "grad_norm": 0.5956787467002869,
+      "learning_rate": 4.9000000000000005e-06,
+      "loss": 0.0429,
+      "step": 98
+    },
+    {
+      "epoch": 0.8559077809798271,
+      "grad_norm": 0.8485273718833923,
+      "learning_rate": 4.95e-06,
+      "loss": 0.0465,
+      "step": 99
+    },
+    {
+      "epoch": 0.8645533141210374,
+      "grad_norm": 0.7477124333381653,
+      "learning_rate": 5e-06,
+      "loss": 0.0407,
+      "step": 100
+    },
+    {
+      "epoch": 0.8731988472622478,
+      "grad_norm": 0.624204695224762,
+      "learning_rate": 4.999964559102694e-06,
+      "loss": 0.0433,
+      "step": 101
+    },
+    {
+      "epoch": 0.8818443804034583,
+      "grad_norm": 0.6230912804603577,
+      "learning_rate": 4.999858237415621e-06,
+      "loss": 0.0364,
+      "step": 102
+    },
+    {
+      "epoch": 0.8904899135446686,
+      "grad_norm": 0.9286472201347351,
+      "learning_rate": 4.999681037953289e-06,
+      "loss": 0.0385,
+      "step": 103
+    },
+    {
+      "epoch": 0.899135446685879,
+      "grad_norm": 1.0528490543365479,
+      "learning_rate": 4.999432965739786e-06,
+      "loss": 0.0406,
+      "step": 104
+    },
+    {
+      "epoch": 0.9077809798270894,
+      "grad_norm": 1.1465263366699219,
+      "learning_rate": 4.999114027808632e-06,
+      "loss": 0.0526,
+      "step": 105
+    },
+    {
+      "epoch": 0.9164265129682997,
+      "grad_norm": 0.7173194885253906,
+      "learning_rate": 4.998724233202585e-06,
+      "loss": 0.05,
+      "step": 106
+    },
+    {
+      "epoch": 0.9250720461095101,
+      "grad_norm": 0.7150623798370361,
+      "learning_rate": 4.998263592973382e-06,
+      "loss": 0.0392,
+      "step": 107
+    },
+    {
+      "epoch": 0.9337175792507204,
+      "grad_norm": 0.6749414801597595,
+      "learning_rate": 4.9977321201814235e-06,
+      "loss": 0.0381,
+      "step": 108
+    },
+    {
+      "epoch": 0.9423631123919308,
+      "grad_norm": 0.7057478427886963,
+      "learning_rate": 4.997129829895409e-06,
+      "loss": 0.0427,
+      "step": 109
+    },
+    {
+      "epoch": 0.9510086455331412,
+      "grad_norm": 0.8370860815048218,
+      "learning_rate": 4.996456739191905e-06,
+      "loss": 0.0353,
+      "step": 110
+    },
+    {
+      "epoch": 0.9596541786743515,
+      "grad_norm": 0.8583172559738159,
+      "learning_rate": 4.995712867154863e-06,
+      "loss": 0.0366,
+      "step": 111
+    },
+    {
+      "epoch": 0.968299711815562,
+      "grad_norm": 0.9564568400382996,
+      "learning_rate": 4.994898234875075e-06,
+      "loss": 0.0463,
+      "step": 112
+    },
+    {
+      "epoch": 0.9769452449567724,
+      "grad_norm": 0.7943828105926514,
+      "learning_rate": 4.9940128654495826e-06,
+      "loss": 0.0366,
+      "step": 113
+    },
+    {
+      "epoch": 0.9855907780979827,
+      "grad_norm": 0.9387117624282837,
+      "learning_rate": 4.9930567839810125e-06,
+      "loss": 0.0433,
+      "step": 114
+    },
+    {
+      "epoch": 0.9942363112391931,
+      "grad_norm": 0.5982036590576172,
+      "learning_rate": 4.992030017576876e-06,
+      "loss": 0.0311,
+      "step": 115
+    },
+    {
+      "epoch": 1.0086455331412103,
+      "grad_norm": 1.6264456510543823,
+      "learning_rate": 4.990932595348788e-06,
+      "loss": 0.079,
+      "step": 116
+    },
+    {
+      "epoch": 1.0172910662824208,
+      "grad_norm": 0.5183172225952148,
+      "learning_rate": 4.989764548411654e-06,
+      "loss": 0.0268,
+      "step": 117
+    },
+    {
+      "epoch": 1.0259365994236311,
+      "grad_norm": 0.5135197043418884,
+      "learning_rate": 4.988525909882779e-06,
+      "loss": 0.0281,
+      "step": 118
+    },
+    {
+      "epoch": 1.0345821325648414,
+      "grad_norm": 0.6001684665679932,
+      "learning_rate": 4.987216714880929e-06,
+      "loss": 0.0312,
+      "step": 119
+    },
+    {
+      "epoch": 1.043227665706052,
+      "grad_norm": 0.5347486138343811,
+      "learning_rate": 4.9858370005253435e-06,
+      "loss": 0.0264,
+      "step": 120
+    },
+    {
+      "epoch": 1.0518731988472623,
+      "grad_norm": 0.8435088396072388,
+      "learning_rate": 4.9843868059346725e-06,
+      "loss": 0.0296,
+      "step": 121
+    },
+    {
+      "epoch": 1.0605187319884726,
+      "grad_norm": 0.8622809052467346,
+      "learning_rate": 4.982866172225876e-06,
+      "loss": 0.0277,
+      "step": 122
+    },
+    {
+      "epoch": 1.069164265129683,
+      "grad_norm": 0.8440551161766052,
+      "learning_rate": 4.981275142513049e-06,
+      "loss": 0.0301,
+      "step": 123
+    },
+    {
+      "epoch": 1.0778097982708934,
+      "grad_norm": 0.6853605508804321,
+      "learning_rate": 4.979613761906212e-06,
+      "loss": 0.0264,
+      "step": 124
+    },
+    {
+      "epoch": 1.0864553314121037,
+      "grad_norm": 0.7620036602020264,
+      "learning_rate": 4.977882077510018e-06,
+      "loss": 0.0281,
+      "step": 125
+    },
+    {
+      "epoch": 1.0951008645533142,
+      "grad_norm": 0.9126653075218201,
+      "learning_rate": 4.9760801384224274e-06,
+      "loss": 0.0312,
+      "step": 126
+    },
+    {
+      "epoch": 1.1037463976945245,
+      "grad_norm": 0.6158509850502014,
+      "learning_rate": 4.97420799573331e-06,
+      "loss": 0.0263,
+      "step": 127
+    },
+    {
+      "epoch": 1.1123919308357348,
+      "grad_norm": 0.7442693114280701,
+      "learning_rate": 4.972265702523001e-06,
+      "loss": 0.0251,
+      "step": 128
+    },
+    {
+      "epoch": 1.1210374639769451,
+      "grad_norm": 0.5755389928817749,
+      "learning_rate": 4.970253313860788e-06,
+      "loss": 0.0276,
+      "step": 129
+    },
+    {
+      "epoch": 1.1296829971181557,
+      "grad_norm": 0.6565670967102051,
+      "learning_rate": 4.968170886803361e-06,
+      "loss": 0.029,
+      "step": 130
+    },
+    {
+      "epoch": 1.138328530259366,
+      "grad_norm": 0.6405811309814453,
+      "learning_rate": 4.966018480393189e-06,
+      "loss": 0.0293,
+      "step": 131
+    },
+    {
+      "epoch": 1.1469740634005763,
+      "grad_norm": 0.5925298929214478,
+      "learning_rate": 4.9637961556568405e-06,
+      "loss": 0.0264,
+      "step": 132
+    },
+    {
+      "epoch": 1.1556195965417868,
+      "grad_norm": 0.5921182632446289,
+      "learning_rate": 4.961503975603263e-06,
+      "loss": 0.0305,
+      "step": 133
+    },
+    {
+      "epoch": 1.1642651296829971,
+      "grad_norm": 0.6314800977706909,
+      "learning_rate": 4.959142005221991e-06,
+      "loss": 0.0242,
+      "step": 134
+    },
+    {
+      "epoch": 1.1729106628242074,
+      "grad_norm": 0.6051074862480164,
+      "learning_rate": 4.956710311481303e-06,
+      "loss": 0.0303,
+      "step": 135
+    },
+    {
+      "epoch": 1.181556195965418,
+      "grad_norm": 0.7119901776313782,
+      "learning_rate": 4.954208963326327e-06,
+      "loss": 0.023,
+      "step": 136
+    },
+    {
+      "epoch": 1.1902017291066282,
+      "grad_norm": 1.087274432182312,
+      "learning_rate": 4.951638031677081e-06,
+      "loss": 0.0222,
+      "step": 137
+    },
+    {
+      "epoch": 1.1988472622478386,
+      "grad_norm": 0.5598759651184082,
+      "learning_rate": 4.948997589426463e-06,
+      "loss": 0.0255,
+      "step": 138
+    },
+    {
+      "epoch": 1.207492795389049,
+      "grad_norm": 0.7533854246139526,
+      "learning_rate": 4.94628771143819e-06,
+      "loss": 0.0245,
+      "step": 139
+    },
+    {
+      "epoch": 1.2161383285302594,
+      "grad_norm": 0.7722812294960022,
+      "learning_rate": 4.943508474544667e-06,
+      "loss": 0.0237,
+      "step": 140
+    },
+    {
+      "epoch": 1.2247838616714697,
+      "grad_norm": 0.6604969501495361,
+      "learning_rate": 4.940659957544813e-06,
+      "loss": 0.02,
+      "step": 141
+    },
+    {
+      "epoch": 1.23342939481268,
+      "grad_norm": 0.5789263248443604,
+      "learning_rate": 4.937742241201826e-06,
+      "loss": 0.0263,
+      "step": 142
+    },
+    {
+      "epoch": 1.2420749279538905,
+      "grad_norm": 1.1343308687210083,
+      "learning_rate": 4.934755408240896e-06,
+      "loss": 0.0291,
+      "step": 143
+    },
+    {
+      "epoch": 1.2507204610951008,
+      "grad_norm": 0.7230944037437439,
+      "learning_rate": 4.931699543346854e-06,
+      "loss": 0.0245,
+      "step": 144
+    },
+    {
+      "epoch": 1.2593659942363113,
+      "grad_norm": 0.591327428817749,
+      "learning_rate": 4.928574733161775e-06,
+      "loss": 0.0241,
+      "step": 145
+    },
+    {
+      "epoch": 1.2680115273775217,
+      "grad_norm": 0.6604726910591125,
+      "learning_rate": 4.925381066282522e-06,
+      "loss": 0.0242,
+      "step": 146
+    },
+    {
+      "epoch": 1.276657060518732,
+      "grad_norm": 0.5772287845611572,
+      "learning_rate": 4.922118633258229e-06,
+      "loss": 0.0272,
+      "step": 147
+    },
+    {
+      "epoch": 1.2853025936599423,
+      "grad_norm": 0.7148783802986145,
+      "learning_rate": 4.918787526587739e-06,
+      "loss": 0.0196,
+      "step": 148
+    },
+    {
+      "epoch": 1.2939481268011528,
+      "grad_norm": 0.660778284072876,
+      "learning_rate": 4.9153878407169815e-06,
+      "loss": 0.0202,
+      "step": 149
+    },
+    {
+      "epoch": 1.302593659942363,
+      "grad_norm": 0.6156355738639832,
+      "learning_rate": 4.911919672036291e-06,
+      "loss": 0.0297,
+      "step": 150
+    },
+    {
+      "epoch": 1.3112391930835736,
+      "grad_norm": 0.6828760504722595,
+      "learning_rate": 4.908383118877672e-06,
+      "loss": 0.0235,
+      "step": 151
+    },
+    {
+      "epoch": 1.319884726224784,
+      "grad_norm": 0.7847033143043518,
+      "learning_rate": 4.904778281512022e-06,
+      "loss": 0.0278,
+      "step": 152
+    },
+    {
+      "epoch": 1.3285302593659942,
+      "grad_norm": 0.9113219976425171,
+      "learning_rate": 4.901105262146275e-06,
+      "loss": 0.0226,
+      "step": 153
+    },
+    {
+      "epoch": 1.3371757925072045,
+      "grad_norm": 0.5406802892684937,
+      "learning_rate": 4.897364164920515e-06,
+      "loss": 0.0205,
+      "step": 154
+    },
+    {
+      "epoch": 1.345821325648415,
+      "grad_norm": 0.7523823976516724,
+      "learning_rate": 4.8935550959050135e-06,
+      "loss": 0.0209,
+      "step": 155
+    },
+    {
+      "epoch": 1.3544668587896254,
+      "grad_norm": 0.7060834169387817,
+      "learning_rate": 4.889678163097233e-06,
+      "loss": 0.0217,
+      "step": 156
+    },
+    {
+      "epoch": 1.3631123919308357,
+      "grad_norm": 0.758028507232666,
+      "learning_rate": 4.885733476418752e-06,
+      "loss": 0.02,
+      "step": 157
+    },
+    {
+      "epoch": 1.3717579250720462,
+      "grad_norm": 0.7745693325996399,
+      "learning_rate": 4.8817211477121615e-06,
+      "loss": 0.0197,
+      "step": 158
+    },
+    {
+      "epoch": 1.3804034582132565,
+      "grad_norm": 0.665055513381958,
+      "learning_rate": 4.8776412907378845e-06,
+      "loss": 0.019,
+      "step": 159
+    },
+    {
+      "epoch": 1.3890489913544668,
+      "grad_norm": 0.6661068797111511,
+      "learning_rate": 4.8734940211709535e-06,
+      "loss": 0.0196,
+      "step": 160
+    },
+    {
+      "epoch": 1.397694524495677,
+      "grad_norm": 0.9675925374031067,
+      "learning_rate": 4.8692794565977335e-06,
+      "loss": 0.0171,
+      "step": 161
+    },
+    {
+      "epoch": 1.4063400576368876,
+      "grad_norm": 0.6929494738578796,
+      "learning_rate": 4.864997716512584e-06,
+      "loss": 0.0198,
+      "step": 162
+    },
+    {
+      "epoch": 1.414985590778098,
+      "grad_norm": 0.6929437518119812,
+      "learning_rate": 4.8606489223144744e-06,
+      "loss": 0.0152,
+      "step": 163
+    },
+    {
+      "epoch": 1.4236311239193085,
+      "grad_norm": 0.7960386276245117,
+      "learning_rate": 4.8562331973035396e-06,
+      "loss": 0.0177,
+      "step": 164
+    },
+    {
+      "epoch": 1.4322766570605188,
+      "grad_norm": 0.8672429919242859,
+      "learning_rate": 4.851750666677583e-06,
+      "loss": 0.0215,
+      "step": 165
+    },
+    {
+      "epoch": 1.440922190201729,
+      "grad_norm": 0.7653059959411621,
+      "learning_rate": 4.847201457528533e-06,
+      "loss": 0.0172,
+      "step": 166
+    },
+    {
+      "epoch": 1.4495677233429394,
+      "grad_norm": 0.8027440905570984,
+      "learning_rate": 4.842585698838832e-06,
+      "loss": 0.0145,
+      "step": 167
+    },
+    {
+      "epoch": 1.45821325648415,
+      "grad_norm": 0.6481142044067383,
+      "learning_rate": 4.837903521477784e-06,
+      "loss": 0.0184,
+      "step": 168
+    },
+    {
+      "epoch": 1.4668587896253602,
+      "grad_norm": 0.8014233112335205,
+      "learning_rate": 4.833155058197842e-06,
+      "loss": 0.0207,
+      "step": 169
+    },
+    {
+      "epoch": 1.4755043227665707,
+      "grad_norm": 0.719078779220581,
+      "learning_rate": 4.828340443630847e-06,
+      "loss": 0.0184,
+      "step": 170
+    },
+    {
+      "epoch": 1.484149855907781,
+      "grad_norm": 0.6776456832885742,
+      "learning_rate": 4.823459814284205e-06,
+      "loss": 0.0186,
+      "step": 171
+    },
+    {
+      "epoch": 1.4927953890489913,
+      "grad_norm": 0.8054779171943665,
+      "learning_rate": 4.818513308537025e-06,
+      "loss": 0.0179,
+      "step": 172
+    },
+    {
+      "epoch": 1.5014409221902016,
+      "grad_norm": 0.8131511211395264,
+      "learning_rate": 4.813501066636188e-06,
+      "loss": 0.0177,
+      "step": 173
+    },
+    {
+      "epoch": 1.510086455331412,
+      "grad_norm": 0.7842754125595093,
+      "learning_rate": 4.808423230692374e-06,
+      "loss": 0.0182,
+      "step": 174
+    },
+    {
+      "epoch": 1.5187319884726225,
+      "grad_norm": 0.6084232330322266,
+      "learning_rate": 4.8032799446760326e-06,
+      "loss": 0.014,
+      "step": 175
+    },
+    {
+      "epoch": 1.527377521613833,
+      "grad_norm": 0.7110248804092407,
+      "learning_rate": 4.798071354413302e-06,
+      "loss": 0.0164,
+      "step": 176
+    },
+    {
+      "epoch": 1.5360230547550433,
+      "grad_norm": 0.707081139087677,
+      "learning_rate": 4.792797607581872e-06,
+      "loss": 0.0137,
+      "step": 177
+    },
+    {
+      "epoch": 1.5446685878962536,
+      "grad_norm": 0.6509385704994202,
+      "learning_rate": 4.787458853706798e-06,
+      "loss": 0.0145,
+      "step": 178
+    },
+    {
+      "epoch": 1.553314121037464,
+      "grad_norm": 0.7796662449836731,
+      "learning_rate": 4.7820552441562625e-06,
+      "loss": 0.0169,
+      "step": 179
+    },
+    {
+      "epoch": 1.5619596541786742,
+      "grad_norm": 0.7351200580596924,
+      "learning_rate": 4.7765869321372835e-06,
+      "loss": 0.0152,
+      "step": 180
+    },
+    {
+      "epoch": 1.5706051873198847,
+      "grad_norm": 0.9273463487625122,
+      "learning_rate": 4.771054072691367e-06,
+      "loss": 0.0181,
+      "step": 181
+    },
+    {
+      "epoch": 1.579250720461095,
+      "grad_norm": 0.7243021130561829,
+      "learning_rate": 4.7654568226901165e-06,
+      "loss": 0.0148,
+      "step": 182
+    },
+    {
+      "epoch": 1.5878962536023056,
+      "grad_norm": 0.729568600654602,
+      "learning_rate": 4.759795340830782e-06,
+      "loss": 0.0145,
+      "step": 183
+    },
+    {
+      "epoch": 1.5965417867435159,
+      "grad_norm": 0.5814985632896423,
+      "learning_rate": 4.754069787631761e-06,
+      "loss": 0.0128,
+      "step": 184
+    },
+    {
+      "epoch": 1.6051873198847262,
+      "grad_norm": 0.49074554443359375,
+      "learning_rate": 4.7482803254280485e-06,
+      "loss": 0.0125,
+      "step": 185
+    },
+    {
+      "epoch": 1.6138328530259365,
+      "grad_norm": 0.5640458464622498,
+      "learning_rate": 4.742427118366632e-06,
+      "loss": 0.0155,
+      "step": 186
+    },
+    {
+      "epoch": 1.622478386167147,
+      "grad_norm": 0.574414849281311,
+      "learning_rate": 4.736510332401841e-06,
+      "loss": 0.0164,
+      "step": 187
+    },
+    {
+      "epoch": 1.6311239193083573,
+      "grad_norm": 0.5837032198905945,
+      "learning_rate": 4.730530135290638e-06,
+      "loss": 0.0129,
+      "step": 188
+    },
+    {
+      "epoch": 1.6397694524495678,
+      "grad_norm": 0.5181459188461304,
+      "learning_rate": 4.724486696587862e-06,
+      "loss": 0.016,
+      "step": 189
+    },
+    {
+      "epoch": 1.6484149855907781,
+      "grad_norm": 0.990842342376709,
+      "learning_rate": 4.718380187641429e-06,
+      "loss": 0.0173,
+      "step": 190
+    },
+    {
+      "epoch": 1.6570605187319885,
+      "grad_norm": 0.6795041561126709,
+      "learning_rate": 4.712210781587463e-06,
+      "loss": 0.013,
+      "step": 191
+    },
+    {
+      "epoch": 1.6657060518731988,
+      "grad_norm": 0.7275810837745667,
+      "learning_rate": 4.705978653345392e-06,
+      "loss": 0.0142,
+      "step": 192
+    },
+    {
+      "epoch": 1.674351585014409,
+      "grad_norm": 0.575529932975769,
+      "learning_rate": 4.699683979612991e-06,
+      "loss": 0.0128,
+      "step": 193
+    },
+    {
+      "epoch": 1.6829971181556196,
+      "grad_norm": 0.8860746622085571,
+      "learning_rate": 4.693326938861367e-06,
+      "loss": 0.0115,
+      "step": 194
+    },
+    {
+      "epoch": 1.6916426512968301,
+      "grad_norm": 0.985464870929718,
+      "learning_rate": 4.686907711329903e-06,
+      "loss": 0.0146,
+      "step": 195
+    },
+    {
+      "epoch": 1.7002881844380404,
+      "grad_norm": 0.5690702199935913,
+      "learning_rate": 4.680426479021147e-06,
+      "loss": 0.0133,
+      "step": 196
+    },
+    {
+      "epoch": 1.7089337175792507,
+      "grad_norm": 0.536115825176239,
+      "learning_rate": 4.67388342569565e-06,
+      "loss": 0.0082,
+      "step": 197
+    },
+    {
+      "epoch": 1.717579250720461,
+      "grad_norm": 0.9312114119529724,
+      "learning_rate": 4.667278736866755e-06,
+      "loss": 0.0127,
+      "step": 198
+    },
+    {
+      "epoch": 1.7262247838616713,
+      "grad_norm": 0.8902615904808044,
+      "learning_rate": 4.660612599795343e-06,
+      "loss": 0.0121,
+      "step": 199
+    },
+    {
+      "epoch": 1.7348703170028819,
+      "grad_norm": 0.8274016380310059,
+      "learning_rate": 4.653885203484516e-06,
+      "loss": 0.0142,
+      "step": 200
+    },
+    {
+      "epoch": 1.7435158501440924,
+      "grad_norm": 1.00035560131073,
+      "learning_rate": 4.647096738674243e-06,
+      "loss": 0.0104,
+      "step": 201
+    },
+    {
+      "epoch": 1.7521613832853027,
+      "grad_norm": 0.5850755572319031,
+      "learning_rate": 4.640247397835953e-06,
+      "loss": 0.0083,
+      "step": 202
+    },
+    {
+      "epoch": 1.760806916426513,
+      "grad_norm": 0.588136613368988,
+      "learning_rate": 4.633337375167074e-06,
+      "loss": 0.0108,
+      "step": 203
+    },
+    {
+      "epoch": 1.7694524495677233,
+      "grad_norm": 0.6305558681488037,
+      "learning_rate": 4.626366866585528e-06,
+      "loss": 0.0103,
+      "step": 204
+    },
+    {
+      "epoch": 1.7780979827089336,
+      "grad_norm": 0.46658825874328613,
+      "learning_rate": 4.619336069724177e-06,
+      "loss": 0.0103,
+      "step": 205
+    },
+    {
+      "epoch": 1.7867435158501441,
+      "grad_norm": 2.3057351112365723,
+      "learning_rate": 4.612245183925225e-06,
+      "loss": 0.0123,
+      "step": 206
+    },
+    {
+      "epoch": 1.7953890489913544,
+      "grad_norm": 0.6322522163391113,
+      "learning_rate": 4.605094410234551e-06,
+      "loss": 0.0123,
+      "step": 207
+    },
+    {
+      "epoch": 1.804034582132565,
+      "grad_norm": 1.1089496612548828,
+      "learning_rate": 4.597883951396027e-06,
+      "loss": 0.0149,
+      "step": 208
+    },
+    {
+      "epoch": 1.8126801152737753,
+      "grad_norm": 0.7867231965065002,
+      "learning_rate": 4.590614011845758e-06,
+      "loss": 0.01,
+      "step": 209
+    },
+    {
+      "epoch": 1.8213256484149856,
+      "grad_norm": 0.8597843647003174,
+      "learning_rate": 4.583284797706288e-06,
+      "loss": 0.0121,
+      "step": 210
+    },
+    {
+      "epoch": 1.8299711815561959,
+      "grad_norm": 0.6839237809181213,
+      "learning_rate": 4.575896516780757e-06,
+      "loss": 0.0108,
+      "step": 211
+    },
+    {
+      "epoch": 1.8386167146974062,
+      "grad_norm": 0.9036144614219666,
+      "learning_rate": 4.568449378547011e-06,
+      "loss": 0.009,
+      "step": 212
+    },
+    {
+      "epoch": 1.8472622478386167,
+      "grad_norm": 0.439732164144516,
+      "learning_rate": 4.560943594151657e-06,
+      "loss": 0.0097,
+      "step": 213
+    },
+    {
+      "epoch": 1.8559077809798272,
+      "grad_norm": 0.5021887421607971,
+      "learning_rate": 4.553379376404085e-06,
+      "loss": 0.0087,
+      "step": 214
+    },
+    {
+      "epoch": 1.8645533141210375,
+      "grad_norm": 0.36055275797843933,
+      "learning_rate": 4.5457569397704226e-06,
+      "loss": 0.0079,
+      "step": 215
+    },
+    {
+      "epoch": 1.8731988472622478,
+      "grad_norm": 0.5071420669555664,
+      "learning_rate": 4.538076500367469e-06,
+      "loss": 0.0065,
+      "step": 216
+    },
+    {
+      "epoch": 1.8818443804034581,
+      "grad_norm": 0.5510254502296448,
+      "learning_rate": 4.530338275956553e-06,
+      "loss": 0.0099,
+      "step": 217
+    },
+    {
+      "epoch": 1.8904899135446684,
+      "grad_norm": 0.8489411473274231,
+      "learning_rate": 4.522542485937369e-06,
+      "loss": 0.0068,
+      "step": 218
+    },
+    {
+      "epoch": 1.899135446685879,
+      "grad_norm": 0.48011505603790283,
+      "learning_rate": 4.514689351341751e-06,
+      "loss": 0.007,
+      "step": 219
+    },
+    {
+      "epoch": 1.9077809798270895,
+      "grad_norm": 0.7145248055458069,
+      "learning_rate": 4.506779094827409e-06,
+      "loss": 0.0085,
+      "step": 220
+    },
+    {
+      "epoch": 1.9164265129682998,
+      "grad_norm": 0.7936932444572449,
+      "learning_rate": 4.498811940671615e-06,
+      "loss": 0.0086,
+      "step": 221
+    },
+    {
+      "epoch": 1.92507204610951,
+      "grad_norm": 0.7309126257896423,
+      "learning_rate": 4.49078811476484e-06,
+      "loss": 0.0093,
+      "step": 222
+    },
+    {
+      "epoch": 1.9337175792507204,
+      "grad_norm": 0.6176413893699646,
+      "learning_rate": 4.482707844604359e-06,
+      "loss": 0.0095,
+      "step": 223
+    },
+    {
+      "epoch": 1.9423631123919307,
+      "grad_norm": 0.995867133140564,
+      "learning_rate": 4.474571359287791e-06,
+      "loss": 0.0061,
+      "step": 224
+    },
+    {
+      "epoch": 1.9510086455331412,
+      "grad_norm": 0.47976699471473694,
+      "learning_rate": 4.466378889506607e-06,
+      "loss": 0.0054,
+      "step": 225
+    },
+    {
+      "epoch": 1.9596541786743515,
+      "grad_norm": 0.740679144859314,
+      "learning_rate": 4.458130667539592e-06,
+      "loss": 0.0097,
+      "step": 226
+    },
+    {
+      "epoch": 1.968299711815562,
+      "grad_norm": 0.565301775932312,
+      "learning_rate": 4.449826927246257e-06,
+      "loss": 0.008,
+      "step": 227
+    },
+    {
+      "epoch": 1.9769452449567724,
+      "grad_norm": 0.6384701728820801,
+      "learning_rate": 4.441467904060207e-06,
+      "loss": 0.0053,
+      "step": 228
+    },
+    {
+      "epoch": 1.9855907780979827,
+      "grad_norm": 0.8729382157325745,
+      "learning_rate": 4.4330538349824684e-06,
+      "loss": 0.0092,
+      "step": 229
+    },
+    {
+      "epoch": 1.994236311239193,
+      "grad_norm": 0.6458576321601868,
+      "learning_rate": 4.424584958574766e-06,
+      "loss": 0.0082,
+      "step": 230
+    },
+    {
+      "epoch": 2.0086455331412103,
+      "grad_norm": 2.1104297637939453,
+      "learning_rate": 4.4160615149527646e-06,
+      "loss": 0.0172,
+      "step": 231
+    },
+    {
+      "epoch": 2.0172910662824206,
+      "grad_norm": 0.8500051498413086,
+      "learning_rate": 4.407483745779256e-06,
+      "loss": 0.0075,
+      "step": 232
+    },
+    {
+      "epoch": 2.025936599423631,
+      "grad_norm": 0.8504526615142822,
+      "learning_rate": 4.39885189425731e-06,
+      "loss": 0.0063,
+      "step": 233
+    },
+    {
+      "epoch": 2.0345821325648417,
+      "grad_norm": 0.5642857551574707,
+      "learning_rate": 4.3901662051233755e-06,
+      "loss": 0.0046,
+      "step": 234
+    },
+    {
+      "epoch": 2.043227665706052,
+      "grad_norm": 1.6137267351150513,
+      "learning_rate": 4.381426924640346e-06,
+      "loss": 0.0111,
+      "step": 235
+    },
+    {
+      "epoch": 2.0518731988472623,
+      "grad_norm": 0.6515895128250122,
+      "learning_rate": 4.372634300590578e-06,
+      "loss": 0.0054,
+      "step": 236
+    },
+    {
+      "epoch": 2.0605187319884726,
+      "grad_norm": 0.42837992310523987,
+      "learning_rate": 4.363788582268857e-06,
+      "loss": 0.0064,
+      "step": 237
+    },
+    {
+      "epoch": 2.069164265129683,
+      "grad_norm": 0.6755346059799194,
+      "learning_rate": 4.35489002047534e-06,
+      "loss": 0.0066,
+      "step": 238
+    },
+    {
+      "epoch": 2.077809798270893,
+      "grad_norm": 0.5027966499328613,
+      "learning_rate": 4.345938867508439e-06,
+      "loss": 0.0053,
+      "step": 239
+    },
+    {
+      "epoch": 2.086455331412104,
+      "grad_norm": 0.44720327854156494,
+      "learning_rate": 4.336935377157668e-06,
+      "loss": 0.0053,
+      "step": 240
+    },
+    {
+      "epoch": 2.0951008645533142,
+      "grad_norm": 0.9336118102073669,
+      "learning_rate": 4.32787980469645e-06,
+      "loss": 0.0071,
+      "step": 241
+    },
+    {
+      "epoch": 2.1037463976945245,
+      "grad_norm": 0.6465862393379211,
+      "learning_rate": 4.318772406874873e-06,
+      "loss": 0.004,
+      "step": 242
+    },
+    {
+      "epoch": 2.112391930835735,
+      "grad_norm": 0.4339396357536316,
+      "learning_rate": 4.309613441912421e-06,
+      "loss": 0.0062,
+      "step": 243
+    },
+    {
+      "epoch": 2.121037463976945,
+      "grad_norm": 0.805067777633667,
+      "learning_rate": 4.30040316949064e-06,
+      "loss": 0.0052,
+      "step": 244
+    },
+    {
+      "epoch": 2.1296829971181555,
+      "grad_norm": 0.5986116528511047,
+      "learning_rate": 4.291141850745788e-06,
+      "loss": 0.004,
+      "step": 245
+    },
+    {
+      "epoch": 2.138328530259366,
+      "grad_norm": 0.33404719829559326,
+      "learning_rate": 4.281829748261422e-06,
+      "loss": 0.004,
+      "step": 246
+    },
+    {
+      "epoch": 2.1469740634005765,
+      "grad_norm": 0.3907548189163208,
+      "learning_rate": 4.272467126060954e-06,
+      "loss": 0.0057,
+      "step": 247
+    },
+    {
+      "epoch": 2.155619596541787,
+      "grad_norm": 0.4643802046775818,
+      "learning_rate": 4.263054249600172e-06,
+      "loss": 0.0045,
+      "step": 248
+    },
+    {
+      "epoch": 2.164265129682997,
+      "grad_norm": 0.46979233622550964,
+      "learning_rate": 4.253591385759705e-06,
+      "loss": 0.0045,
+      "step": 249
+    },
+    {
+      "epoch": 2.1729106628242074,
+      "grad_norm": 0.4389543831348419,
+      "learning_rate": 4.244078802837462e-06,
+      "loss": 0.005,
+      "step": 250
+    },
+    {
+      "epoch": 2.1815561959654177,
+      "grad_norm": 0.33877941966056824,
+      "learning_rate": 4.234516770541023e-06,
+      "loss": 0.0036,
+      "step": 251
+    },
+    {
+      "epoch": 2.1902017291066285,
+      "grad_norm": 0.41794854402542114,
+      "learning_rate": 4.224905559979991e-06,
+      "loss": 0.0049,
+      "step": 252
+    },
+    {
+      "epoch": 2.1988472622478388,
+      "grad_norm": 0.471942275762558,
+      "learning_rate": 4.215245443658307e-06,
+      "loss": 0.004,
+      "step": 253
+    },
+    {
+      "epoch": 2.207492795389049,
+      "grad_norm": 0.5759933590888977,
+      "learning_rate": 4.205536695466524e-06,
+      "loss": 0.0037,
+      "step": 254
+    },
+    {
+      "epoch": 2.2161383285302594,
+      "grad_norm": 0.4565725326538086,
+      "learning_rate": 4.1957795906740404e-06,
+      "loss": 0.0041,
+      "step": 255
+    },
+    {
+      "epoch": 2.2247838616714697,
+      "grad_norm": 0.5055815577507019,
+      "learning_rate": 4.1859744059212945e-06,
+      "loss": 0.0058,
+      "step": 256
+    },
+    {
+      "epoch": 2.23342939481268,
+      "grad_norm": 0.5749806761741638,
+      "learning_rate": 4.176121419211924e-06,
+      "loss": 0.0061,
+      "step": 257
+    },
+    {
+      "epoch": 2.2420749279538903,
+      "grad_norm": 1.0351425409317017,
+      "learning_rate": 4.16622090990488e-06,
+      "loss": 0.0046,
+      "step": 258
+    },
+    {
+      "epoch": 2.250720461095101,
+      "grad_norm": 1.2486116886138916,
+      "learning_rate": 4.15627315870651e-06,
+      "loss": 0.0065,
+      "step": 259
+    },
+    {
+      "epoch": 2.2593659942363113,
+      "grad_norm": 0.6449837684631348,
+      "learning_rate": 4.146278447662597e-06,
+      "loss": 0.0055,
+      "step": 260
+    },
+    {
+      "epoch": 2.2680115273775217,
+      "grad_norm": 0.7886547446250916,
+      "learning_rate": 4.136237060150363e-06,
+      "loss": 0.0083,
+      "step": 261
+    },
+    {
+      "epoch": 2.276657060518732,
+      "grad_norm": 0.797860324382782,
+      "learning_rate": 4.126149280870434e-06,
+      "loss": 0.0043,
+      "step": 262
+    },
+    {
+      "epoch": 2.2853025936599423,
+      "grad_norm": 0.8402999639511108,
+      "learning_rate": 4.116015395838772e-06,
+      "loss": 0.0044,
+      "step": 263
+    },
+    {
+      "epoch": 2.2939481268011526,
+      "grad_norm": 0.4193873107433319,
+      "learning_rate": 4.105835692378557e-06,
+      "loss": 0.0028,
+      "step": 264
+    },
+    {
+      "epoch": 2.302593659942363,
+      "grad_norm": 0.4523886740207672,
+      "learning_rate": 4.095610459112051e-06,
+      "loss": 0.005,
+      "step": 265
+    },
+    {
+      "epoch": 2.3112391930835736,
+      "grad_norm": 0.45222657918930054,
+      "learning_rate": 4.0853399859524066e-06,
+      "loss": 0.0027,
+      "step": 266
+    },
+    {
+      "epoch": 2.319884726224784,
+      "grad_norm": 0.5312837362289429,
+      "learning_rate": 4.075024564095452e-06,
+      "loss": 0.0072,
+      "step": 267
+    },
+    {
+      "epoch": 2.3285302593659942,
+      "grad_norm": 0.5837632417678833,
+      "learning_rate": 4.064664486011433e-06,
+      "loss": 0.0056,
+      "step": 268
+    },
+    {
+      "epoch": 2.3371757925072045,
+      "grad_norm": 0.44099336862564087,
+      "learning_rate": 4.05426004543672e-06,
+      "loss": 0.0052,
+      "step": 269
+    },
+    {
+      "epoch": 2.345821325648415,
+      "grad_norm": 0.4609047472476959,
+      "learning_rate": 4.04381153736548e-06,
+      "loss": 0.0041,
+      "step": 270
+    },
+    {
+      "epoch": 2.3544668587896256,
+      "grad_norm": 0.4224575161933899,
+      "learning_rate": 4.033319258041316e-06,
+      "loss": 0.0056,
+      "step": 271
+    },
+    {
+      "epoch": 2.363112391930836,
+      "grad_norm": 0.45308536291122437,
+      "learning_rate": 4.022783504948862e-06,
+      "loss": 0.0037,
+      "step": 272
+    },
+    {
+      "epoch": 2.371757925072046,
+      "grad_norm": 0.29924383759498596,
+      "learning_rate": 4.012204576805352e-06,
+      "loss": 0.0037,
+      "step": 273
+    },
+    {
+      "epoch": 2.3804034582132565,
+      "grad_norm": 0.33644095063209534,
+      "learning_rate": 4.001582773552153e-06,
+      "loss": 0.003,
+      "step": 274
+    },
+    {
+      "epoch": 2.389048991354467,
+      "grad_norm": 0.553075909614563,
+      "learning_rate": 3.990918396346254e-06,
+      "loss": 0.0024,
+      "step": 275
+    },
+    {
+      "epoch": 2.397694524495677,
+      "grad_norm": 0.7815229892730713,
+      "learning_rate": 3.9802117475517335e-06,
+      "loss": 0.0024,
+      "step": 276
+    },
+    {
+      "epoch": 2.4063400576368874,
+      "grad_norm": 0.512950599193573,
+      "learning_rate": 3.969463130731183e-06,
+      "loss": 0.0037,
+      "step": 277
+    },
+    {
+      "epoch": 2.414985590778098,
+      "grad_norm": 0.45993947982788086,
+      "learning_rate": 3.958672850637104e-06,
+      "loss": 0.0029,
+      "step": 278
+    },
+    {
+      "epoch": 2.4236311239193085,
+      "grad_norm": 0.9428088068962097,
+      "learning_rate": 3.947841213203262e-06,
+      "loss": 0.0029,
+      "step": 279
+    },
+    {
+      "epoch": 2.4322766570605188,
+      "grad_norm": 0.4953707158565521,
+      "learning_rate": 3.936968525536018e-06,
+      "loss": 0.0029,
+      "step": 280
+    },
+    {
+      "epoch": 2.440922190201729,
+      "grad_norm": 0.3896157443523407,
+      "learning_rate": 3.926055095905616e-06,
+      "loss": 0.003,
+      "step": 281
+    },
+    {
+      "epoch": 2.4495677233429394,
+      "grad_norm": 0.5497294664382935,
+      "learning_rate": 3.9151012337374495e-06,
+      "loss": 0.004,
+      "step": 282
+    },
+    {
+      "epoch": 2.4582132564841497,
+      "grad_norm": 0.31193655729293823,
+      "learning_rate": 3.9041072496032805e-06,
+      "loss": 0.0025,
+      "step": 283
+    },
+    {
+      "epoch": 2.46685878962536,
+      "grad_norm": 0.6194396615028381,
+      "learning_rate": 3.893073455212438e-06,
+      "loss": 0.0035,
+      "step": 284
+    },
+    {
+      "epoch": 2.4755043227665707,
+      "grad_norm": 0.5500622987747192,
+      "learning_rate": 3.882000163402984e-06,
+      "loss": 0.0021,
+      "step": 285
+    },
+    {
+      "epoch": 2.484149855907781,
+      "grad_norm": 0.3981945812702179,
+      "learning_rate": 3.870887688132834e-06,
+      "loss": 0.0046,
+      "step": 286
+    },
+    {
+      "epoch": 2.4927953890489913,
+      "grad_norm": 0.6703558564186096,
+      "learning_rate": 3.859736344470866e-06,
+      "loss": 0.0032,
+      "step": 287
+    },
+    {
+      "epoch": 2.5014409221902016,
+      "grad_norm": 0.6285901665687561,
+      "learning_rate": 3.8485464485879785e-06,
+      "loss": 0.0055,
+      "step": 288
+    },
+    {
+      "epoch": 2.510086455331412,
+      "grad_norm": 0.45444604754447937,
+      "learning_rate": 3.837318317748134e-06,
+      "loss": 0.0029,
+      "step": 289
+    },
+    {
+      "epoch": 2.5187319884726227,
+      "grad_norm": 0.5909140706062317,
+      "learning_rate": 3.826052270299356e-06,
+      "loss": 0.0018,
+      "step": 290
+    },
+    {
+      "epoch": 2.527377521613833,
+      "grad_norm": 0.28365251421928406,
+      "learning_rate": 3.814748625664711e-06,
+      "loss": 0.0016,
+      "step": 291
+    },
+    {
+      "epoch": 2.5360230547550433,
+      "grad_norm": 0.6033309102058411,
+      "learning_rate": 3.8034077043332463e-06,
+      "loss": 0.0026,
+      "step": 292
+    },
+    {
+      "epoch": 2.5446685878962536,
+      "grad_norm": 0.5435200929641724,
+      "learning_rate": 3.7920298278509028e-06,
+      "loss": 0.003,
+      "step": 293
+    },
+    {
+      "epoch": 2.553314121037464,
+      "grad_norm": 0.4895404875278473,
+      "learning_rate": 3.7806153188114027e-06,
+      "loss": 0.0028,
+      "step": 294
+    },
+    {
+      "epoch": 2.561959654178674,
+      "grad_norm": 0.37084832787513733,
+      "learning_rate": 3.7691645008471e-06,
+      "loss": 0.0019,
+      "step": 295
+    },
+    {
+      "epoch": 2.5706051873198845,
+      "grad_norm": 0.35942429304122925,
+      "learning_rate": 3.7576776986198064e-06,
+      "loss": 0.004,
+      "step": 296
+    },
+    {
+      "epoch": 2.5792507204610953,
+      "grad_norm": 0.9116867780685425,
+      "learning_rate": 3.7461552378115833e-06,
+      "loss": 0.0032,
+      "step": 297
+    },
+    {
+      "epoch": 2.5878962536023056,
+      "grad_norm": 0.4446312487125397,
+      "learning_rate": 3.734597445115511e-06,
+      "loss": 0.0016,
+      "step": 298
+    },
+    {
+      "epoch": 2.596541786743516,
+      "grad_norm": 0.2611502408981323,
+      "learning_rate": 3.7230046482264256e-06,
+      "loss": 0.0031,
+      "step": 299
+    },
+    {
+      "epoch": 2.605187319884726,
+      "grad_norm": 0.5151038765907288,
+      "learning_rate": 3.711377175831626e-06,
+      "loss": 0.0028,
+      "step": 300
+    },
+    {
+      "epoch": 2.6138328530259365,
+      "grad_norm": 0.3680933713912964,
+      "learning_rate": 3.6997153576015552e-06,
+      "loss": 0.0008,
+      "step": 301
+    },
+    {
+      "epoch": 2.6224783861671472,
+      "grad_norm": 0.42117324471473694,
+      "learning_rate": 3.6880195241804567e-06,
+      "loss": 0.0037,
+      "step": 302
+    },
+    {
+      "epoch": 2.631123919308357,
+      "grad_norm": 0.35740363597869873,
+      "learning_rate": 3.676290007176994e-06,
+      "loss": 0.0015,
+      "step": 303
+    },
+    {
+      "epoch": 2.639769452449568,
+      "grad_norm": 0.5276333093643188,
+      "learning_rate": 3.6645271391548542e-06,
+      "loss": 0.0043,
+      "step": 304
+    },
+    {
+      "epoch": 2.648414985590778,
+      "grad_norm": 0.6532303094863892,
+      "learning_rate": 3.652731253623315e-06,
+      "loss": 0.0023,
+      "step": 305
+    },
+    {
+      "epoch": 2.6570605187319885,
+      "grad_norm": 0.2517545521259308,
+      "learning_rate": 3.6409026850277908e-06,
+      "loss": 0.001,
+      "step": 306
+    },
+    {
+      "epoch": 2.6657060518731988,
+      "grad_norm": 0.4791451096534729,
+      "learning_rate": 3.6290417687403485e-06,
+      "loss": 0.0025,
+      "step": 307
+    },
+    {
+      "epoch": 2.674351585014409,
+      "grad_norm": 0.665262758731842,
+      "learning_rate": 3.617148841050202e-06,
+      "loss": 0.0022,
+      "step": 308
+    },
+    {
+      "epoch": 2.68299711815562,
+      "grad_norm": 0.40816545486450195,
+      "learning_rate": 3.6052242391541746e-06,
+      "loss": 0.0033,
+      "step": 309
+    },
+    {
+      "epoch": 2.69164265129683,
+      "grad_norm": 0.3500838577747345,
+      "learning_rate": 3.593268301147139e-06,
+      "loss": 0.0024,
+      "step": 310
+    },
+    {
+      "epoch": 2.7002881844380404,
+      "grad_norm": 0.8193225264549255,
+      "learning_rate": 3.5812813660124313e-06,
+      "loss": 0.0031,
+      "step": 311
+    },
+    {
+      "epoch": 2.7089337175792507,
+      "grad_norm": 0.3912574350833893,
+      "learning_rate": 3.5692637736122427e-06,
+      "loss": 0.0016,
+      "step": 312
+    },
+    {
+      "epoch": 2.717579250720461,
+      "grad_norm": 0.46311694383621216,
+      "learning_rate": 3.5572158646779787e-06,
+      "loss": 0.002,
+      "step": 313
+    },
+    {
+      "epoch": 2.7262247838616713,
+      "grad_norm": 0.5172969698905945,
+      "learning_rate": 3.5451379808006014e-06,
+      "loss": 0.0019,
+      "step": 314
+    },
+    {
+      "epoch": 2.7348703170028816,
+      "grad_norm": 0.2887652814388275,
+      "learning_rate": 3.5330304644209456e-06,
+      "loss": 0.0023,
+      "step": 315
+    },
+    {
+      "epoch": 2.7435158501440924,
+      "grad_norm": 0.6363840699195862,
+      "learning_rate": 3.520893658820007e-06,
+      "loss": 0.0047,
+      "step": 316
+    },
+    {
+      "epoch": 2.7521613832853027,
+      "grad_norm": 0.8680986762046814,
+      "learning_rate": 3.50872790810921e-06,
+      "loss": 0.0017,
+      "step": 317
+    },
+    {
+      "epoch": 2.760806916426513,
+      "grad_norm": 0.2707628607749939,
+      "learning_rate": 3.4965335572206516e-06,
+      "loss": 0.0019,
+      "step": 318
+    },
+    {
+      "epoch": 2.7694524495677233,
+      "grad_norm": 0.5891854166984558,
+      "learning_rate": 3.484310951897323e-06,
+      "loss": 0.0025,
+      "step": 319
+    },
+    {
+      "epoch": 2.7780979827089336,
+      "grad_norm": 0.3361755609512329,
+      "learning_rate": 3.4720604386833024e-06,
+      "loss": 0.0016,
+      "step": 320
+    },
+    {
+      "epoch": 2.7867435158501443,
+      "grad_norm": 0.2326732724905014,
+      "learning_rate": 3.459782364913935e-06,
+      "loss": 0.0049,
+      "step": 321
+    },
+    {
+      "epoch": 2.795389048991354,
+      "grad_norm": 0.7111459970474243,
+      "learning_rate": 3.447477078705983e-06,
+      "loss": 0.0023,
+      "step": 322
+    },
+    {
+      "epoch": 2.804034582132565,
+      "grad_norm": 0.5887550711631775,
+      "learning_rate": 3.4351449289477543e-06,
+      "loss": 0.0029,
+      "step": 323
+    },
+    {
+      "epoch": 2.8126801152737753,
+      "grad_norm": 0.6490927934646606,
+      "learning_rate": 3.4227862652892106e-06,
+      "loss": 0.0017,
+      "step": 324
+    },
+    {
+      "epoch": 2.8213256484149856,
+      "grad_norm": 0.2970047891139984,
+      "learning_rate": 3.410401438132056e-06,
+      "loss": 0.0031,
+      "step": 325
+    },
+    {
+      "epoch": 2.829971181556196,
+      "grad_norm": 0.9381114840507507,
+      "learning_rate": 3.3979907986197996e-06,
+      "loss": 0.0034,
+      "step": 326
+    },
+    {
+      "epoch": 2.838616714697406,
+      "grad_norm": 0.34777429699897766,
+      "learning_rate": 3.385554698627803e-06,
+      "loss": 0.0029,
+      "step": 327
+    },
+    {
+      "epoch": 2.847262247838617,
+      "grad_norm": 0.5431174635887146,
+      "learning_rate": 3.3730934907532997e-06,
+      "loss": 0.0012,
+      "step": 328
+    },
+    {
+      "epoch": 2.8559077809798272,
+      "grad_norm": 0.22815977036952972,
+      "learning_rate": 3.3606075283054005e-06,
+      "loss": 0.0011,
+      "step": 329
+    },
+    {
+      "epoch": 2.8645533141210375,
+      "grad_norm": 0.2716272473335266,
+      "learning_rate": 3.3480971652950757e-06,
+      "loss": 0.0015,
+      "step": 330
+    },
+    {
+      "epoch": 2.873198847262248,
+      "grad_norm": 0.42019474506378174,
+      "learning_rate": 3.3355627564251185e-06,
+      "loss": 0.0026,
+      "step": 331
+    },
+    {
+      "epoch": 2.881844380403458,
+      "grad_norm": 0.41707322001457214,
+      "learning_rate": 3.3230046570800866e-06,
+      "loss": 0.0017,
+      "step": 332
+    },
+    {
+      "epoch": 2.8904899135446684,
+      "grad_norm": 0.23598358035087585,
+      "learning_rate": 3.3104232233162272e-06,
+      "loss": 0.0008,
+      "step": 333
+    },
+    {
+      "epoch": 2.8991354466858787,
+      "grad_norm": 0.53864985704422,
+      "learning_rate": 3.2978188118513814e-06,
+      "loss": 0.0017,
+      "step": 334
+    },
+    {
+      "epoch": 2.9077809798270895,
+      "grad_norm": 0.2139906883239746,
+      "learning_rate": 3.2851917800548726e-06,
+      "loss": 0.0025,
+      "step": 335
+    },
+    {
+      "epoch": 2.9164265129683,
+      "grad_norm": 0.49649330973625183,
+      "learning_rate": 3.272542485937369e-06,
+      "loss": 0.0016,
+      "step": 336
+    },
+    {
+      "epoch": 2.92507204610951,
+      "grad_norm": 0.1875542253255844,
+      "learning_rate": 3.259871288140738e-06,
+      "loss": 0.0015,
+      "step": 337
+    },
+    {
+      "epoch": 2.9337175792507204,
+      "grad_norm": 0.4745478928089142,
+      "learning_rate": 3.247178545927876e-06,
+      "loss": 0.0023,
+      "step": 338
+    },
+    {
+      "epoch": 2.9423631123919307,
+      "grad_norm": 0.5308757424354553,
+      "learning_rate": 3.234464619172522e-06,
+      "loss": 0.0026,
+      "step": 339
+    },
+    {
+      "epoch": 2.9510086455331415,
+      "grad_norm": 0.3783188760280609,
+      "learning_rate": 3.221729868349053e-06,
+      "loss": 0.0007,
+      "step": 340
+    },
+    {
+      "epoch": 2.9596541786743513,
+      "grad_norm": 0.2250707745552063,
+      "learning_rate": 3.208974654522266e-06,
+      "loss": 0.0006,
+      "step": 341
+    },
+    {
+      "epoch": 2.968299711815562,
+      "grad_norm": 0.2039024382829666,
+      "learning_rate": 3.1961993393371405e-06,
+      "loss": 0.001,
+      "step": 342
+    },
+    {
+      "epoch": 2.9769452449567724,
+      "grad_norm": 0.14253279566764832,
+      "learning_rate": 3.183404285008582e-06,
+      "loss": 0.0041,
+      "step": 343
+    },
+    {
+      "epoch": 2.9855907780979827,
+      "grad_norm": 0.45035484433174133,
+      "learning_rate": 3.1705898543111576e-06,
+      "loss": 0.0007,
+      "step": 344
+    },
+    {
+      "epoch": 2.994236311239193,
+      "grad_norm": 0.11087123304605484,
+      "learning_rate": 3.157756410568803e-06,
+      "loss": 0.0016,
+      "step": 345
+    },
+    {
+      "epoch": 3.0086455331412103,
+      "grad_norm": 0.2526646554470062,
+      "learning_rate": 3.14490431764453e-06,
+      "loss": 0.0041,
+      "step": 346
+    },
+    {
+      "epoch": 3.0172910662824206,
+      "grad_norm": 0.4000411927700043,
+      "learning_rate": 3.132033939930101e-06,
+      "loss": 0.0015,
+      "step": 347
+    },
+    {
+      "epoch": 3.025936599423631,
+      "grad_norm": 0.30819934606552124,
+      "learning_rate": 3.1191456423357047e-06,
+      "loss": 0.001,
+      "step": 348
+    },
+    {
+      "epoch": 3.0345821325648417,
+      "grad_norm": 0.17910632491111755,
+      "learning_rate": 3.106239790279606e-06,
+      "loss": 0.0009,
+      "step": 349
+    },
+    {
+      "epoch": 3.043227665706052,
+      "grad_norm": 0.2142518013715744,
+      "learning_rate": 3.093316749677788e-06,
+      "loss": 0.0016,
+      "step": 350
+    },
+    {
+      "epoch": 3.0518731988472623,
+      "grad_norm": 0.17407968640327454,
+      "learning_rate": 3.0803768869335726e-06,
+      "loss": 0.0012,
+      "step": 351
+    },
+    {
+      "epoch": 3.0605187319884726,
+      "grad_norm": 0.420547753572464,
+      "learning_rate": 3.0674205689272378e-06,
+      "loss": 0.0012,
+      "step": 352
+    },
+    {
+      "epoch": 3.069164265129683,
+      "grad_norm": 0.298476904630661,
+      "learning_rate": 3.054448163005613e-06,
+      "loss": 0.0016,
+      "step": 353
+    },
+    {
+      "epoch": 3.077809798270893,
+      "grad_norm": 0.34958237409591675,
+      "learning_rate": 3.041460036971664e-06,
+      "loss": 0.001,
+      "step": 354
+    },
+    {
+      "epoch": 3.086455331412104,
+      "grad_norm": 0.34075653553009033,
+      "learning_rate": 3.028456559074061e-06,
+      "loss": 0.0012,
+      "step": 355
+    },
+    {
+      "epoch": 3.0951008645533142,
+      "grad_norm": 0.2592026889324188,
+      "learning_rate": 3.0154380979967456e-06,
+      "loss": 0.0008,
+      "step": 356
+    },
+    {
+      "epoch": 3.1037463976945245,
+      "grad_norm": 0.22481408715248108,
+      "learning_rate": 3.0024050228484713e-06,
+      "loss": 0.001,
+      "step": 357
+    },
+    {
+      "epoch": 3.112391930835735,
+      "grad_norm": 0.29716426134109497,
+      "learning_rate": 2.9893577031523403e-06,
+      "loss": 0.0014,
+      "step": 358
+    },
+    {
+      "epoch": 3.121037463976945,
+      "grad_norm": 0.10284000635147095,
+      "learning_rate": 2.976296508835326e-06,
+      "loss": 0.0003,
+      "step": 359
+    },
+    {
+      "epoch": 3.1296829971181555,
+      "grad_norm": 0.9767228364944458,
+      "learning_rate": 2.963221810217786e-06,
+      "loss": 0.0016,
+      "step": 360
+    },
+    {
+      "epoch": 3.138328530259366,
+      "grad_norm": 0.08798839896917343,
+      "learning_rate": 2.9501339780029614e-06,
+      "loss": 0.0006,
+      "step": 361
+    },
+    {
+      "epoch": 3.1469740634005765,
+      "grad_norm": 0.4243233799934387,
+      "learning_rate": 2.937033383266466e-06,
+      "loss": 0.0026,
+      "step": 362
+    },
+    {
+      "epoch": 3.155619596541787,
+      "grad_norm": 0.06578671187162399,
+      "learning_rate": 2.923920397445766e-06,
+      "loss": 0.0004,
+      "step": 363
+    },
+    {
+      "epoch": 3.164265129682997,
+      "grad_norm": 0.4194351136684418,
+      "learning_rate": 2.910795392329649e-06,
+      "loss": 0.0012,
+      "step": 364
+    },
+    {
+      "epoch": 3.1729106628242074,
+      "grad_norm": 0.18414922058582306,
+      "learning_rate": 2.8976587400476804e-06,
+      "loss": 0.0007,
+      "step": 365
+    },
+    {
+      "epoch": 3.1815561959654177,
+      "grad_norm": 0.3256891071796417,
+      "learning_rate": 2.884510813059657e-06,
+      "loss": 0.0009,
+      "step": 366
+    },
+    {
+      "epoch": 3.1902017291066285,
+      "grad_norm": 0.48699238896369934,
+      "learning_rate": 2.871351984145042e-06,
+      "loss": 0.0011,
+      "step": 367
+    },
+    {
+      "epoch": 3.1988472622478388,
+      "grad_norm": 0.12711212038993835,
+      "learning_rate": 2.8581826263923993e-06,
+      "loss": 0.0004,
+      "step": 368
+    },
+    {
+      "epoch": 3.207492795389049,
+      "grad_norm": 0.2759152054786682,
+      "learning_rate": 2.8450031131888147e-06,
+      "loss": 0.0008,
+      "step": 369
+    },
+    {
+      "epoch": 3.2161383285302594,
+      "grad_norm": 0.17881423234939575,
+      "learning_rate": 2.8318138182093053e-06,
+      "loss": 0.0005,
+      "step": 370
+    },
+    {
+      "epoch": 3.2247838616714697,
+      "grad_norm": 0.09530279785394669,
+      "learning_rate": 2.8186151154062314e-06,
+      "loss": 0.0005,
+      "step": 371
+    },
+    {
+      "epoch": 3.23342939481268,
+      "grad_norm": 0.29938796162605286,
+      "learning_rate": 2.8054073789986884e-06,
+      "loss": 0.0007,
+      "step": 372
+    },
+    {
+      "epoch": 3.2420749279538903,
+      "grad_norm": 0.137968972325325,
+      "learning_rate": 2.792190983461902e-06,
+      "loss": 0.0005,
+      "step": 373
+    },
+    {
+      "epoch": 3.250720461095101,
+      "grad_norm": 0.13374479115009308,
+      "learning_rate": 2.7789663035166035e-06,
+      "loss": 0.0002,
+      "step": 374
+    },
+    {
+      "epoch": 3.2593659942363113,
+      "grad_norm": 0.5121517777442932,
+      "learning_rate": 2.7657337141184137e-06,
+      "loss": 0.0007,
+      "step": 375
+    },
+    {
+      "epoch": 3.2680115273775217,
+      "grad_norm": 0.34778279066085815,
+      "learning_rate": 2.7524935904472056e-06,
+      "loss": 0.0006,
+      "step": 376
+    },
+    {
+      "epoch": 3.276657060518732,
+      "grad_norm": 0.2674623131752014,
+      "learning_rate": 2.73924630789647e-06,
+      "loss": 0.0005,
+      "step": 377
+    },
+    {
+      "epoch": 3.2853025936599423,
+      "grad_norm": 0.20955437421798706,
+      "learning_rate": 2.7259922420626705e-06,
+      "loss": 0.0006,
+      "step": 378
+    },
+    {
+      "epoch": 3.2939481268011526,
+      "grad_norm": 0.29623010754585266,
+      "learning_rate": 2.7127317687345973e-06,
+      "loss": 0.0018,
+      "step": 379
+    },
+    {
+      "epoch": 3.302593659942363,
+      "grad_norm": 0.2673618197441101,
+      "learning_rate": 2.699465263882708e-06,
+      "loss": 0.0016,
+      "step": 380
+    },
+    {
+      "epoch": 3.3112391930835736,
+      "grad_norm": 0.061357282102108,
+      "learning_rate": 2.686193103648472e-06,
+      "loss": 0.0004,
+      "step": 381
+    },
+    {
+      "epoch": 3.319884726224784,
+      "grad_norm": 0.1466841846704483,
+      "learning_rate": 2.672915664333704e-06,
+      "loss": 0.0007,
+      "step": 382
+    },
+    {
+      "epoch": 3.3285302593659942,
+      "grad_norm": 0.10404661297798157,
+      "learning_rate": 2.6596333223898934e-06,
+      "loss": 0.0003,
+      "step": 383
+    },
+    {
+      "epoch": 3.3371757925072045,
+      "grad_norm": 0.1709136813879013,
+      "learning_rate": 2.6463464544075344e-06,
+      "loss": 0.0002,
+      "step": 384
+    },
+    {
+      "epoch": 3.345821325648415,
+      "grad_norm": 0.08438851684331894,
+      "learning_rate": 2.6330554371054466e-06,
+      "loss": 0.0003,
+      "step": 385
+    },
+    {
+      "epoch": 3.3544668587896256,
+      "grad_norm": 0.15904630720615387,
+      "learning_rate": 2.6197606473200924e-06,
+      "loss": 0.001,
+      "step": 386
+    },
+    {
+      "epoch": 3.363112391930836,
+      "grad_norm": 0.1417740285396576,
+      "learning_rate": 2.6064624619948966e-06,
+      "loss": 0.0003,
+      "step": 387
+    },
+    {
+      "epoch": 3.371757925072046,
+      "grad_norm": 0.8411481380462646,
+      "learning_rate": 2.593161258169554e-06,
+      "loss": 0.0016,
+      "step": 388
+    },
+    {
+      "epoch": 3.3804034582132565,
+      "grad_norm": 0.13281837105751038,
+      "learning_rate": 2.579857412969345e-06,
+      "loss": 0.0006,
+      "step": 389
+    },
+    {
+      "epoch": 3.389048991354467,
+      "grad_norm": 0.06628571450710297,
+      "learning_rate": 2.5665513035944373e-06,
+      "loss": 0.0003,
+      "step": 390
+    },
+    {
+      "epoch": 3.397694524495677,
+      "grad_norm": 0.2239917516708374,
+      "learning_rate": 2.5532433073091967e-06,
+      "loss": 0.0005,
+      "step": 391
+    },
+    {
+      "epoch": 3.4063400576368874,
+      "grad_norm": 0.1752793937921524,
+      "learning_rate": 2.539933801431487e-06,
+      "loss": 0.0005,
+      "step": 392
+    },
+    {
+      "epoch": 3.414985590778098,
+      "grad_norm": 0.0841381773352623,
+      "learning_rate": 2.5266231633219733e-06,
+      "loss": 0.0004,
+      "step": 393
+    },
+    {
+      "epoch": 3.4236311239193085,
+      "grad_norm": 0.0805194228887558,
+      "learning_rate": 2.513311770373421e-06,
+      "loss": 0.0002,
+      "step": 394
+    },
+    {
+      "epoch": 3.4322766570605188,
+      "grad_norm": 0.07594503462314606,
+      "learning_rate": 2.5e-06,
+      "loss": 0.0002,
+      "step": 395
+    },
+    {
+      "epoch": 3.440922190201729,
+      "grad_norm": 0.10026411712169647,
+      "learning_rate": 2.4866882296265797e-06,
+      "loss": 0.0003,
+      "step": 396
+    },
+    {
+      "epoch": 3.4495677233429394,
+      "grad_norm": 0.05829321965575218,
+      "learning_rate": 2.473376836678028e-06,
+      "loss": 0.0003,
+      "step": 397
+    },
+    {
+      "epoch": 3.4582132564841497,
+      "grad_norm": 0.4893040657043457,
+      "learning_rate": 2.4600661985685132e-06,
+      "loss": 0.0014,
+      "step": 398
+    },
+    {
+      "epoch": 3.46685878962536,
+      "grad_norm": 0.06461616605520248,
+      "learning_rate": 2.446756692690804e-06,
+      "loss": 0.0002,
+      "step": 399
+    },
+    {
+      "epoch": 3.4755043227665707,
+      "grad_norm": 0.06487792730331421,
+      "learning_rate": 2.4334486964055635e-06,
+      "loss": 0.0002,
+      "step": 400
+    },
+    {
+      "epoch": 3.484149855907781,
+      "grad_norm": 0.08846238255500793,
+      "learning_rate": 2.4201425870306566e-06,
+      "loss": 0.0005,
+      "step": 401
+    },
+    {
+      "epoch": 3.4927953890489913,
+      "grad_norm": 0.5399437546730042,
+      "learning_rate": 2.406838741830446e-06,
+      "loss": 0.001,
+      "step": 402
+    },
+    {
+      "epoch": 3.5014409221902016,
+      "grad_norm": 0.07543563842773438,
+      "learning_rate": 2.393537538005104e-06,
+      "loss": 0.0004,
+      "step": 403
+    },
+    {
+      "epoch": 3.510086455331412,
+      "grad_norm": 0.09962673485279083,
+      "learning_rate": 2.380239352679908e-06,
+      "loss": 0.0006,
+      "step": 404
+    },
+    {
+      "epoch": 3.5187319884726227,
+      "grad_norm": 0.09634942561388016,
+      "learning_rate": 2.3669445628945543e-06,
+      "loss": 0.0005,
+      "step": 405
+    },
+    {
+      "epoch": 3.527377521613833,
+      "grad_norm": 0.07776723802089691,
+      "learning_rate": 2.3536535455924656e-06,
+      "loss": 0.0002,
+      "step": 406
+    },
+    {
+      "epoch": 3.5360230547550433,
+      "grad_norm": 0.26890671253204346,
+      "learning_rate": 2.340366677610107e-06,
+      "loss": 0.0002,
+      "step": 407
+    },
+    {
+      "epoch": 3.5446685878962536,
+      "grad_norm": 0.0821603462100029,
+      "learning_rate": 2.327084335666297e-06,
+      "loss": 0.0001,
+      "step": 408
+    },
+    {
+      "epoch": 3.553314121037464,
+      "grad_norm": 0.20579983294010162,
+      "learning_rate": 2.313806896351529e-06,
+      "loss": 0.0003,
+      "step": 409
+    },
+    {
+      "epoch": 3.561959654178674,
+      "grad_norm": 0.024787653237581253,
+      "learning_rate": 2.300534736117292e-06,
+      "loss": 0.0001,
+      "step": 410
+    },
+    {
+      "epoch": 3.5706051873198845,
+      "grad_norm": 0.04598158970475197,
+      "learning_rate": 2.2872682312654035e-06,
+      "loss": 0.0001,
+      "step": 411
+    },
+    {
+      "epoch": 3.5792507204610953,
+      "grad_norm": 0.05497601255774498,
+      "learning_rate": 2.2740077579373303e-06,
+      "loss": 0.0003,
+      "step": 412
+    },
+    {
+      "epoch": 3.5878962536023056,
+      "grad_norm": 0.058321401476860046,
+      "learning_rate": 2.2607536921035313e-06,
+      "loss": 0.0002,
+      "step": 413
+    },
+    {
+      "epoch": 3.596541786743516,
+      "grad_norm": 0.024839522317051888,
+      "learning_rate": 2.247506409552795e-06,
+      "loss": 0.0001,
+      "step": 414
+    },
+    {
+      "epoch": 3.605187319884726,
+      "grad_norm": 1.3820059299468994,
+      "learning_rate": 2.234266285881587e-06,
+      "loss": 0.0014,
+      "step": 415
+    },
+    {
+      "epoch": 3.6138328530259365,
+      "grad_norm": 0.03965742886066437,
+      "learning_rate": 2.221033696483397e-06,
+      "loss": 0.0001,
+      "step": 416
+    },
+    {
+      "epoch": 3.6224783861671472,
+      "grad_norm": 0.033283866941928864,
+      "learning_rate": 2.2078090165380992e-06,
+      "loss": 0.0001,
+      "step": 417
+    },
+    {
+      "epoch": 3.631123919308357,
+      "grad_norm": 0.015545632690191269,
+      "learning_rate": 2.194592621001311e-06,
+      "loss": 0.0001,
+      "step": 418
+    },
+    {
+      "epoch": 3.639769452449568,
+      "grad_norm": 0.03403123468160629,
+      "learning_rate": 2.1813848845937695e-06,
+      "loss": 0.0001,
+      "step": 419
+    },
+    {
+      "epoch": 3.648414985590778,
+      "grad_norm": 0.04488883912563324,
+      "learning_rate": 2.1681861817906955e-06,
+      "loss": 0.0001,
+      "step": 420
+    },
+    {
+      "epoch": 3.6570605187319885,
+      "grad_norm": 0.22106269001960754,
+      "learning_rate": 2.1549968868111866e-06,
+      "loss": 0.0003,
+      "step": 421
+    },
+    {
+      "epoch": 3.6657060518731988,
+      "grad_norm": 0.12951771914958954,
+      "learning_rate": 2.141817373607601e-06,
+      "loss": 0.0003,
+      "step": 422
+    },
+    {
+      "epoch": 3.674351585014409,
+      "grad_norm": 0.17822900414466858,
+      "learning_rate": 2.1286480158549583e-06,
+      "loss": 0.0004,
+      "step": 423
+    },
+    {
+      "epoch": 3.68299711815562,
+      "grad_norm": 0.10862100124359131,
+      "learning_rate": 2.1154891869403436e-06,
+      "loss": 0.0002,
+      "step": 424
+    },
+    {
+      "epoch": 3.69164265129683,
+      "grad_norm": 0.09040237963199615,
+      "learning_rate": 2.1023412599523204e-06,
+      "loss": 0.0003,
+      "step": 425
+    },
+    {
+      "epoch": 3.7002881844380404,
+      "grad_norm": 0.14486011862754822,
+      "learning_rate": 2.089204607670352e-06,
+      "loss": 0.0003,
+      "step": 426
+    },
+    {
+      "epoch": 3.7089337175792507,
+      "grad_norm": 0.058098066598176956,
+      "learning_rate": 2.0760796025542342e-06,
+      "loss": 0.0002,
+      "step": 427
+    },
+    {
+      "epoch": 3.717579250720461,
+      "grad_norm": 0.13733893632888794,
+      "learning_rate": 2.0629666167335344e-06,
+      "loss": 0.0002,
+      "step": 428
+    },
+    {
+      "epoch": 3.7262247838616713,
+      "grad_norm": 0.12006496638059616,
+      "learning_rate": 2.0498660219970395e-06,
+      "loss": 0.0007,
+      "step": 429
+    },
+    {
+      "epoch": 3.7348703170028816,
+      "grad_norm": 0.07560478895902634,
+      "learning_rate": 2.0367781897822147e-06,
+      "loss": 0.0001,
+      "step": 430
+    },
+    {
+      "epoch": 3.7435158501440924,
+      "grad_norm": 0.021039756014943123,
+      "learning_rate": 2.0237034911646745e-06,
+      "loss": 0.0001,
+      "step": 431
+    },
+    {
+      "epoch": 3.7521613832853027,
+      "grad_norm": 0.010244839824736118,
+      "learning_rate": 2.0106422968476606e-06,
+      "loss": 0.0,
+      "step": 432
+    },
+    {
+      "epoch": 3.760806916426513,
+      "grad_norm": 0.10909762978553772,
+      "learning_rate": 1.9975949771515296e-06,
+      "loss": 0.0001,
+      "step": 433
+    },
+    {
+      "epoch": 3.7694524495677233,
+      "grad_norm": 0.05506197735667229,
+      "learning_rate": 1.9845619020032552e-06,
+      "loss": 0.0002,
+      "step": 434
+    },
+    {
+      "epoch": 3.7780979827089336,
+      "grad_norm": 0.19995303452014923,
+      "learning_rate": 1.9715434409259393e-06,
+      "loss": 0.0011,
+      "step": 435
+    },
+    {
+      "epoch": 3.7867435158501443,
+      "grad_norm": 0.009247799403965473,
+      "learning_rate": 1.958539963028337e-06,
+      "loss": 0.0,
+      "step": 436
+    },
+    {
+      "epoch": 3.795389048991354,
+      "grad_norm": 0.015599964186549187,
+      "learning_rate": 1.9455518369943873e-06,
+      "loss": 0.0001,
+      "step": 437
+    },
+    {
+      "epoch": 3.804034582132565,
+      "grad_norm": 0.15373478829860687,
+      "learning_rate": 1.9325794310727626e-06,
+      "loss": 0.0006,
+      "step": 438
+    },
+    {
+      "epoch": 3.8126801152737753,
+      "grad_norm": 0.05203329771757126,
+      "learning_rate": 1.9196231130664282e-06,
+      "loss": 0.0002,
+      "step": 439
+    },
+    {
+      "epoch": 3.8213256484149856,
+      "grad_norm": 0.08206374943256378,
+      "learning_rate": 1.906683250322213e-06,
+      "loss": 0.0003,
+      "step": 440
+    },
+    {
+      "epoch": 3.829971181556196,
+      "grad_norm": 0.06621237844228745,
+      "learning_rate": 1.8937602097203945e-06,
+      "loss": 0.0001,
+      "step": 441
+    },
+    {
+      "epoch": 3.838616714697406,
+      "grad_norm": 0.3289477527141571,
+      "learning_rate": 1.8808543576642966e-06,
+      "loss": 0.0016,
+      "step": 442
+    },
+    {
+      "epoch": 3.847262247838617,
+      "grad_norm": 0.14183855056762695,
+      "learning_rate": 1.8679660600698996e-06,
+      "loss": 0.0002,
+      "step": 443
+    },
+    {
+      "epoch": 3.8559077809798272,
+      "grad_norm": 0.43348291516304016,
+      "learning_rate": 1.8550956823554708e-06,
+      "loss": 0.0004,
+      "step": 444
+    },
+    {
+      "epoch": 3.8645533141210375,
+      "grad_norm": 0.3135347068309784,
+      "learning_rate": 1.8422435894311973e-06,
+      "loss": 0.0022,
+      "step": 445
+    },
+    {
+      "epoch": 3.873198847262248,
+      "grad_norm": 0.04297763854265213,
+      "learning_rate": 1.8294101456888433e-06,
+      "loss": 0.0002,
+      "step": 446
+    },
+    {
+      "epoch": 3.881844380403458,
+      "grad_norm": 0.03995918110013008,
+      "learning_rate": 1.8165957149914182e-06,
+      "loss": 0.0001,
+      "step": 447
+    },
+    {
+      "epoch": 3.8904899135446684,
+      "grad_norm": 0.05494571849703789,
+      "learning_rate": 1.8038006606628599e-06,
+      "loss": 0.0002,
+      "step": 448
+    },
+    {
+      "epoch": 3.8991354466858787,
+      "grad_norm": 0.00816399697214365,
+      "learning_rate": 1.7910253454777346e-06,
+      "loss": 0.0,
+      "step": 449
+    },
+    {
+      "epoch": 3.9077809798270895,
+      "grad_norm": 0.012747962959110737,
+      "learning_rate": 1.7782701316509482e-06,
+      "loss": 0.0001,
+      "step": 450
+    },
+    {
+      "epoch": 3.9164265129683,
+      "grad_norm": 0.055127181112766266,
+      "learning_rate": 1.7655353808274795e-06,
+      "loss": 0.0001,
+      "step": 451
+    },
+    {
+      "epoch": 3.92507204610951,
+      "grad_norm": 0.33472880721092224,
+      "learning_rate": 1.752821454072124e-06,
+      "loss": 0.0006,
+      "step": 452
+    },
+    {
+      "epoch": 3.9337175792507204,
+      "grad_norm": 0.06558925658464432,
+      "learning_rate": 1.7401287118592626e-06,
+      "loss": 0.0002,
+      "step": 453
+    },
+    {
+      "epoch": 3.9423631123919307,
+      "grad_norm": 0.06597518920898438,
+      "learning_rate": 1.7274575140626318e-06,
+      "loss": 0.0003,
+      "step": 454
+    },
+    {
+      "epoch": 3.9510086455331415,
+      "grad_norm": 0.06961897015571594,
+      "learning_rate": 1.7148082199451288e-06,
+      "loss": 0.0002,
+      "step": 455
+    },
+    {
+      "epoch": 3.9596541786743513,
+      "grad_norm": 0.07268885523080826,
+      "learning_rate": 1.7021811881486186e-06,
+      "loss": 0.0004,
+      "step": 456
+    },
+    {
+      "epoch": 3.968299711815562,
+      "grad_norm": 0.05247066915035248,
+      "learning_rate": 1.6895767766837734e-06,
+      "loss": 0.0001,
+      "step": 457
+    },
+    {
+      "epoch": 3.9769452449567724,
+      "grad_norm": 0.12619982659816742,
+      "learning_rate": 1.6769953429199142e-06,
+      "loss": 0.0005,
+      "step": 458
+    },
+    {
+      "epoch": 3.9855907780979827,
+      "grad_norm": 0.043070387095212936,
+      "learning_rate": 1.6644372435748823e-06,
+      "loss": 0.0002,
+      "step": 459
+    },
+    {
+      "epoch": 3.994236311239193,
+      "grad_norm": 0.029151560738682747,
+      "learning_rate": 1.6519028347049242e-06,
+      "loss": 0.0,
+      "step": 460
+    },
+    {
+      "epoch": 4.008645533141211,
+      "grad_norm": 0.25047677755355835,
+      "learning_rate": 1.6393924716946003e-06,
+      "loss": 0.0005,
+      "step": 461
+    },
+    {
+      "epoch": 4.017291066282421,
+      "grad_norm": 0.03903050720691681,
+      "learning_rate": 1.626906509246701e-06,
+      "loss": 0.0001,
+      "step": 462
+    },
+    {
+      "epoch": 4.025936599423631,
+      "grad_norm": 0.06342215090990067,
+      "learning_rate": 1.6144453013721978e-06,
+      "loss": 0.0002,
+      "step": 463
+    },
+    {
+      "epoch": 4.034582132564841,
+      "grad_norm": 0.020777089521288872,
+      "learning_rate": 1.6020092013802002e-06,
+      "loss": 0.0,
+      "step": 464
+    },
+    {
+      "epoch": 4.043227665706052,
+      "grad_norm": 0.01243128813803196,
+      "learning_rate": 1.5895985618679447e-06,
+      "loss": 0.0001,
+      "step": 465
+    },
+    {
+      "epoch": 4.051873198847262,
+      "grad_norm": 0.016826143488287926,
+      "learning_rate": 1.5772137347107902e-06,
+      "loss": 0.001,
+      "step": 466
+    },
+    {
+      "epoch": 4.060518731988473,
+      "grad_norm": 0.25355905294418335,
+      "learning_rate": 1.5648550710522467e-06,
+      "loss": 0.0001,
+      "step": 467
+    },
+    {
+      "epoch": 4.069164265129683,
+      "grad_norm": 0.0483023077249527,
+      "learning_rate": 1.5525229212940168e-06,
+      "loss": 0.0002,
+      "step": 468
+    },
+    {
+      "epoch": 4.077809798270893,
+      "grad_norm": 0.04395914822816849,
+      "learning_rate": 1.5402176350860653e-06,
+      "loss": 0.0002,
+      "step": 469
+    },
+    {
+      "epoch": 4.086455331412104,
+      "grad_norm": 0.026249030604958534,
+      "learning_rate": 1.5279395613166987e-06,
+      "loss": 0.0001,
+      "step": 470
+    },
+    {
+      "epoch": 4.095100864553314,
+      "grad_norm": 0.019938549026846886,
+      "learning_rate": 1.5156890481026786e-06,
+      "loss": 0.0001,
+      "step": 471
+    },
+    {
+      "epoch": 4.1037463976945245,
+      "grad_norm": 0.01866944693028927,
+      "learning_rate": 1.5034664427793486e-06,
+      "loss": 0.0,
+      "step": 472
+    },
+    {
+      "epoch": 4.112391930835735,
+      "grad_norm": 0.013607538305222988,
+      "learning_rate": 1.4912720918907906e-06,
+      "loss": 0.0,
+      "step": 473
+    },
+    {
+      "epoch": 4.121037463976945,
+      "grad_norm": 0.06787893176078796,
+      "learning_rate": 1.4791063411799938e-06,
+      "loss": 0.0003,
+      "step": 474
+    },
+    {
+      "epoch": 4.129682997118156,
+      "grad_norm": 0.019246675074100494,
+      "learning_rate": 1.4669695355790552e-06,
+      "loss": 0.0,
+      "step": 475
+    },
+    {
+      "epoch": 4.138328530259366,
+      "grad_norm": 0.0765075534582138,
+      "learning_rate": 1.4548620191994e-06,
+      "loss": 0.0003,
+      "step": 476
+    },
+    {
+      "epoch": 4.1469740634005765,
+      "grad_norm": 0.03727831318974495,
+      "learning_rate": 1.4427841353220224e-06,
+      "loss": 0.0002,
+      "step": 477
+    },
+    {
+      "epoch": 4.155619596541786,
+      "grad_norm": 0.0503949299454689,
+      "learning_rate": 1.4307362263877584e-06,
+      "loss": 0.0002,
+      "step": 478
+    },
+    {
+      "epoch": 4.164265129682997,
+      "grad_norm": 0.03264373913407326,
+      "learning_rate": 1.4187186339875697e-06,
+      "loss": 0.0002,
+      "step": 479
+    },
+    {
+      "epoch": 4.172910662824208,
+      "grad_norm": 0.061017073690891266,
+      "learning_rate": 1.4067316988528618e-06,
+      "loss": 0.0,
+      "step": 480
+    },
+    {
+      "epoch": 4.181556195965418,
+      "grad_norm": 0.06418503820896149,
+      "learning_rate": 1.3947757608458263e-06,
+      "loss": 0.0002,
+      "step": 481
+    },
+    {
+      "epoch": 4.1902017291066285,
+      "grad_norm": 0.021499959751963615,
+      "learning_rate": 1.3828511589497977e-06,
+      "loss": 0.0,
+      "step": 482
+    },
+    {
+      "epoch": 4.198847262247838,
+      "grad_norm": 0.008911274373531342,
+      "learning_rate": 1.3709582312596519e-06,
+      "loss": 0.0,
+      "step": 483
+    },
+    {
+      "epoch": 4.207492795389049,
+      "grad_norm": 0.007865059189498425,
+      "learning_rate": 1.3590973149722103e-06,
+      "loss": 0.0,
+      "step": 484
+    },
+    {
+      "epoch": 4.216138328530259,
+      "grad_norm": 0.0180173609405756,
+      "learning_rate": 1.347268746376685e-06,
+      "loss": 0.0,
+      "step": 485
+    },
+    {
+      "epoch": 4.22478386167147,
+      "grad_norm": 0.0172914769500494,
+      "learning_rate": 1.3354728608451462e-06,
+      "loss": 0.0,
+      "step": 486
+    },
+    {
+      "epoch": 4.23342939481268,
+      "grad_norm": 0.015868855640292168,
+      "learning_rate": 1.3237099928230066e-06,
+      "loss": 0.0,
+      "step": 487
+    },
+    {
+      "epoch": 4.24207492795389,
+      "grad_norm": 0.02281869202852249,
+      "learning_rate": 1.3119804758195443e-06,
+      "loss": 0.0001,
+      "step": 488
+    },
+    {
+      "epoch": 4.250720461095101,
+      "grad_norm": 0.06229427456855774,
+      "learning_rate": 1.300284642398445e-06,
+      "loss": 0.0001,
+      "step": 489
+    },
+    {
+      "epoch": 4.259365994236311,
+      "grad_norm": 0.04283831641077995,
+      "learning_rate": 1.288622824168375e-06,
+      "loss": 0.0001,
+      "step": 490
+    },
+    {
+      "epoch": 4.268011527377522,
+      "grad_norm": 0.023561948910355568,
+      "learning_rate": 1.276995351773575e-06,
+      "loss": 0.0001,
+      "step": 491
+    },
+    {
+      "epoch": 4.276657060518732,
+      "grad_norm": 0.04599250480532646,
+      "learning_rate": 1.26540255488449e-06,
+      "loss": 0.0001,
+      "step": 492
+    },
+    {
+      "epoch": 4.285302593659942,
+      "grad_norm": 0.03455409035086632,
+      "learning_rate": 1.2538447621884173e-06,
+      "loss": 0.0001,
+      "step": 493
+    },
+    {
+      "epoch": 4.293948126801153,
+      "grad_norm": 0.06476529687643051,
+      "learning_rate": 1.2423223013801946e-06,
+      "loss": 0.0002,
+      "step": 494
+    },
+    {
+      "epoch": 4.302593659942363,
+      "grad_norm": 0.007448709569871426,
+      "learning_rate": 1.230835499152901e-06,
+      "loss": 0.0001,
+      "step": 495
+    },
+    {
+      "epoch": 4.311239193083574,
+      "grad_norm": 0.024221738800406456,
+      "learning_rate": 1.219384681188598e-06,
+      "loss": 0.0,
+      "step": 496
+    },
+    {
+      "epoch": 4.3198847262247835,
+      "grad_norm": 0.013653815723955631,
+      "learning_rate": 1.207970172149098e-06,
+      "loss": 0.0,
+      "step": 497
+    },
+    {
+      "epoch": 4.328530259365994,
+      "grad_norm": 0.0163737740367651,
+      "learning_rate": 1.1965922956667536e-06,
+      "loss": 0.0001,
+      "step": 498
+    },
+    {
+      "epoch": 4.337175792507205,
+      "grad_norm": 0.04039165377616882,
+      "learning_rate": 1.1852513743352886e-06,
+      "loss": 0.0001,
+      "step": 499
+    },
+    {
+      "epoch": 4.345821325648415,
+      "grad_norm": 0.01361683290451765,
+      "learning_rate": 1.1739477297006442e-06,
+      "loss": 0.0001,
+      "step": 500
+    },
+    {
+      "epoch": 4.354466858789626,
+      "grad_norm": 0.05603167042136192,
+      "learning_rate": 1.1626816822518662e-06,
+      "loss": 0.0002,
+      "step": 501
+    },
+    {
+      "epoch": 4.363112391930835,
+      "grad_norm": 0.1649046540260315,
+      "learning_rate": 1.1514535514120217e-06,
+      "loss": 0.0003,
+      "step": 502
+    },
+    {
+      "epoch": 4.371757925072046,
+      "grad_norm": 0.06988149136304855,
+      "learning_rate": 1.140263655529135e-06,
+      "loss": 0.0001,
+      "step": 503
+    },
+    {
+      "epoch": 4.380403458213257,
+      "grad_norm": 0.01356775127351284,
+      "learning_rate": 1.1291123118671665e-06,
+      "loss": 0.0,
+      "step": 504
+    },
+    {
+      "epoch": 4.389048991354467,
+      "grad_norm": 0.011344455182552338,
+      "learning_rate": 1.1179998365970174e-06,
+      "loss": 0.0001,
+      "step": 505
+    },
+    {
+      "epoch": 4.3976945244956775,
+      "grad_norm": 0.015661202371120453,
+      "learning_rate": 1.1069265447875618e-06,
+      "loss": 0.0,
+      "step": 506
+    },
+    {
+      "epoch": 4.406340057636887,
+      "grad_norm": 0.13560903072357178,
+      "learning_rate": 1.0958927503967207e-06,
+      "loss": 0.0006,
+      "step": 507
+    },
+    {
+      "epoch": 4.414985590778098,
+      "grad_norm": 0.13171304762363434,
+      "learning_rate": 1.0848987662625516e-06,
+      "loss": 0.0,
+      "step": 508
+    },
+    {
+      "epoch": 4.423631123919308,
+      "grad_norm": 0.021776214241981506,
+      "learning_rate": 1.073944904094385e-06,
+      "loss": 0.0,
+      "step": 509
+    },
+    {
+      "epoch": 4.432276657060519,
+      "grad_norm": 0.004369642119854689,
+      "learning_rate": 1.0630314744639831e-06,
+      "loss": 0.0,
+      "step": 510
+    },
+    {
+      "epoch": 4.4409221902017295,
+      "grad_norm": 0.05059368908405304,
+      "learning_rate": 1.052158786796739e-06,
+      "loss": 0.0001,
+      "step": 511
+    },
+    {
+      "epoch": 4.449567723342939,
+      "grad_norm": 0.009102565236389637,
+      "learning_rate": 1.0413271493628964e-06,
+      "loss": 0.0,
+      "step": 512
+    },
+    {
+      "epoch": 4.45821325648415,
+      "grad_norm": 0.013081284239888191,
+      "learning_rate": 1.0305368692688175e-06,
+      "loss": 0.0,
+      "step": 513
+    },
+    {
+      "epoch": 4.46685878962536,
+      "grad_norm": 0.07242248952388763,
+      "learning_rate": 1.019788252448267e-06,
+      "loss": 0.0003,
+      "step": 514
+    },
+    {
+      "epoch": 4.475504322766571,
+      "grad_norm": 0.006698797456920147,
+      "learning_rate": 1.0090816036537462e-06,
+      "loss": 0.0,
+      "step": 515
+    },
+    {
+      "epoch": 4.484149855907781,
+      "grad_norm": 0.02764168381690979,
+      "learning_rate": 9.984172264478476e-07,
+      "loss": 0.0001,
+      "step": 516
+    },
+    {
+      "epoch": 4.492795389048991,
+      "grad_norm": 0.006561846937984228,
+      "learning_rate": 9.877954231946485e-07,
+      "loss": 0.0,
+      "step": 517
+    },
+    {
+      "epoch": 4.501440922190202,
+      "grad_norm": 0.02488504722714424,
+      "learning_rate": 9.772164950511387e-07,
+      "loss": 0.0001,
+      "step": 518
+    },
+    {
+      "epoch": 4.510086455331412,
+      "grad_norm": 0.01242034137248993,
+      "learning_rate": 9.66680741958685e-07,
+      "loss": 0.0001,
+      "step": 519
+    },
+    {
+      "epoch": 4.518731988472623,
+      "grad_norm": 0.16501882672309875,
+      "learning_rate": 9.561884626345206e-07,
+      "loss": 0.0002,
+      "step": 520
+    },
+    {
+      "epoch": 4.527377521613833,
+      "grad_norm": 0.0346071794629097,
+      "learning_rate": 9.457399545632815e-07,
+      "loss": 0.0001,
+      "step": 521
+    },
+    {
+      "epoch": 4.536023054755043,
+      "grad_norm": 0.010261784307658672,
+      "learning_rate": 9.353355139885672e-07,
+      "loss": 0.0,
+      "step": 522
+    },
+    {
+      "epoch": 4.544668587896254,
+      "grad_norm": 0.017157072201371193,
+      "learning_rate": 9.249754359045485e-07,
+      "loss": 0.0004,
+      "step": 523
+    },
+    {
+      "epoch": 4.553314121037464,
+      "grad_norm": 0.09536877274513245,
+      "learning_rate": 9.146600140475945e-07,
+      "loss": 0.0001,
+      "step": 524
+    },
+    {
+      "epoch": 4.561959654178675,
+      "grad_norm": 0.03881097957491875,
+      "learning_rate": 9.043895408879505e-07,
+      "loss": 0.0001,
+      "step": 525
+    },
+    {
+      "epoch": 4.5706051873198845,
+      "grad_norm": 0.003924703225493431,
+      "learning_rate": 8.941643076214438e-07,
+      "loss": 0.0,
+      "step": 526
+    },
+    {
+      "epoch": 4.579250720461095,
+      "grad_norm": 0.010769468732178211,
+      "learning_rate": 8.839846041612296e-07,
+      "loss": 0.0001,
+      "step": 527
+    },
+    {
+      "epoch": 4.587896253602305,
+      "grad_norm": 0.03893817961215973,
+      "learning_rate": 8.738507191295658e-07,
+      "loss": 0.0,
+      "step": 528
+    },
+    {
+      "epoch": 4.596541786743516,
+      "grad_norm": 0.010444357059895992,
+      "learning_rate": 8.637629398496378e-07,
+      "loss": 0.0001,
+      "step": 529
+    },
+    {
+      "epoch": 4.605187319884726,
+      "grad_norm": 0.018974248319864273,
+      "learning_rate": 8.537215523374037e-07,
+      "loss": 0.0001,
+      "step": 530
+    },
+    {
+      "epoch": 4.6138328530259365,
+      "grad_norm": 0.04680433124303818,
+      "learning_rate": 8.437268412934898e-07,
+      "loss": 0.0001,
+      "step": 531
+    },
+    {
+      "epoch": 4.622478386167147,
+      "grad_norm": 0.02349163219332695,
+      "learning_rate": 8.337790900951201e-07,
+      "loss": 0.0,
+      "step": 532
+    },
+    {
+      "epoch": 4.631123919308357,
+      "grad_norm": 0.019928090274333954,
+      "learning_rate": 8.238785807880767e-07,
+      "loss": 0.0001,
+      "step": 533
+    },
+    {
+      "epoch": 4.639769452449568,
+      "grad_norm": 0.009950011037290096,
+      "learning_rate": 8.140255940787059e-07,
+      "loss": 0.0001,
+      "step": 534
+    },
+    {
+      "epoch": 4.648414985590778,
+      "grad_norm": 0.034434814006090164,
+      "learning_rate": 8.042204093259598e-07,
+      "loss": 0.0,
+      "step": 535
+    },
+    {
+      "epoch": 4.6570605187319885,
+      "grad_norm": 0.008066104725003242,
+      "learning_rate": 7.944633045334762e-07,
+      "loss": 0.0,
+      "step": 536
+    },
+    {
+      "epoch": 4.665706051873199,
+      "grad_norm": 0.0360458604991436,
+      "learning_rate": 7.847545563416936e-07,
+      "loss": 0.0001,
+      "step": 537
+    },
+    {
+      "epoch": 4.674351585014409,
+      "grad_norm": 0.006509549915790558,
+      "learning_rate": 7.750944400200103e-07,
+      "loss": 0.0001,
+      "step": 538
+    },
+    {
+      "epoch": 4.68299711815562,
+      "grad_norm": 0.11465685069561005,
+      "learning_rate": 7.654832294589776e-07,
+      "loss": 0.0,
+      "step": 539
+    },
+    {
+      "epoch": 4.69164265129683,
+      "grad_norm": 0.03756583854556084,
+      "learning_rate": 7.559211971625385e-07,
+      "loss": 0.0001,
+      "step": 540
+    },
+    {
+      "epoch": 4.70028818443804,
+      "grad_norm": 0.01114535704255104,
+      "learning_rate": 7.46408614240296e-07,
+      "loss": 0.0,
+      "step": 541
+    },
+    {
+      "epoch": 4.708933717579251,
+      "grad_norm": 0.021791907027363777,
+      "learning_rate": 7.369457503998287e-07,
+      "loss": 0.0,
+      "step": 542
+    },
+    {
+      "epoch": 4.717579250720461,
+      "grad_norm": 0.012248961254954338,
+      "learning_rate": 7.275328739390466e-07,
+      "loss": 0.0,
+      "step": 543
+    },
+    {
+      "epoch": 4.726224783861672,
+      "grad_norm": 0.006273179780691862,
+      "learning_rate": 7.181702517385789e-07,
+      "loss": 0.0,
+      "step": 544
+    },
+    {
+      "epoch": 4.734870317002882,
+      "grad_norm": 0.021544231101870537,
+      "learning_rate": 7.088581492542121e-07,
+      "loss": 0.0001,
+      "step": 545
+    },
+    {
+      "epoch": 4.743515850144092,
+      "grad_norm": 0.016482336446642876,
+      "learning_rate": 6.995968305093604e-07,
+      "loss": 0.0001,
+      "step": 546
+    },
+    {
+      "epoch": 4.752161383285302,
+      "grad_norm": 0.035197075456380844,
+      "learning_rate": 6.903865580875796e-07,
+      "loss": 0.0,
+      "step": 547
+    },
+    {
+      "epoch": 4.760806916426513,
+      "grad_norm": 0.03350463882088661,
+      "learning_rate": 6.81227593125127e-07,
+      "loss": 0.0001,
+      "step": 548
+    },
+    {
+      "epoch": 4.769452449567723,
+      "grad_norm": 0.025157498195767403,
+      "learning_rate": 6.721201953035511e-07,
+      "loss": 0.0001,
+      "step": 549
+    },
+    {
+      "epoch": 4.778097982708934,
+      "grad_norm": 0.013705157674849033,
+      "learning_rate": 6.630646228423324e-07,
+      "loss": 0.0,
+      "step": 550
+    },
+    {
+      "epoch": 4.786743515850144,
+      "grad_norm": 0.05199694633483887,
+      "learning_rate": 6.54061132491561e-07,
+      "loss": 0.0001,
+      "step": 551
+    },
+    {
+      "epoch": 4.795389048991354,
+      "grad_norm": 0.021124713122844696,
+      "learning_rate": 6.451099795246604e-07,
+      "loss": 0.0001,
+      "step": 552
+    },
+    {
+      "epoch": 4.804034582132565,
+      "grad_norm": 0.011140077374875546,
+      "learning_rate": 6.362114177311438e-07,
+      "loss": 0.0,
+      "step": 553
+    },
+    {
+      "epoch": 4.812680115273775,
+      "grad_norm": 0.01086009107530117,
+      "learning_rate": 6.273656994094232e-07,
+      "loss": 0.0001,
+      "step": 554
+    },
+    {
+      "epoch": 4.821325648414986,
+      "grad_norm": 0.029642069712281227,
+      "learning_rate": 6.185730753596538e-07,
+      "loss": 0.0,
+      "step": 555
+    },
+    {
+      "epoch": 4.829971181556196,
+      "grad_norm": 0.014636816456913948,
+      "learning_rate": 6.098337948766256e-07,
+      "loss": 0.0,
+      "step": 556
+    },
+    {
+      "epoch": 4.838616714697406,
+      "grad_norm": 0.0111089451238513,
+      "learning_rate": 6.011481057426916e-07,
+      "loss": 0.0,
+      "step": 557
+    },
+    {
+      "epoch": 4.847262247838617,
+      "grad_norm": 0.008355767466127872,
+      "learning_rate": 5.925162542207441e-07,
+      "loss": 0.0001,
+      "step": 558
+    },
+    {
+      "epoch": 4.855907780979827,
+      "grad_norm": 0.020910680294036865,
+      "learning_rate": 5.839384850472359e-07,
+      "loss": 0.0,
+      "step": 559
+    },
+    {
+      "epoch": 4.8645533141210375,
+      "grad_norm": 0.023057645186781883,
+      "learning_rate": 5.754150414252341e-07,
+      "loss": 0.0,
+      "step": 560
+    },
+    {
+      "epoch": 4.873198847262248,
+      "grad_norm": 0.04212991148233414,
+      "learning_rate": 5.669461650175326e-07,
+      "loss": 0.0001,
+      "step": 561
+    },
+    {
+      "epoch": 4.881844380403458,
+      "grad_norm": 0.004758734721690416,
+      "learning_rate": 5.585320959397935e-07,
+      "loss": 0.0,
+      "step": 562
+    },
+    {
+      "epoch": 4.890489913544669,
+      "grad_norm": 0.006756670773029327,
+      "learning_rate": 5.501730727537436e-07,
+      "loss": 0.0,
+      "step": 563
+    },
+    {
+      "epoch": 4.899135446685879,
+      "grad_norm": 0.014876470901072025,
+      "learning_rate": 5.418693324604082e-07,
+      "loss": 0.0,
+      "step": 564
+    },
+    {
+      "epoch": 4.9077809798270895,
+      "grad_norm": 0.008136936463415623,
+      "learning_rate": 5.336211104933939e-07,
+      "loss": 0.0,
+      "step": 565
+    },
+    {
+      "epoch": 4.916426512968299,
+      "grad_norm": 0.006519688293337822,
+      "learning_rate": 5.254286407122103e-07,
+      "loss": 0.0,
+      "step": 566
+    },
+    {
+      "epoch": 4.92507204610951,
+      "grad_norm": 0.016903480514883995,
+      "learning_rate": 5.172921553956417e-07,
+      "loss": 0.0,
+      "step": 567
+    },
+    {
+      "epoch": 4.93371757925072,
+      "grad_norm": 0.003450587159022689,
+      "learning_rate": 5.092118852351599e-07,
+      "loss": 0.0,
+      "step": 568
+    },
+    {
+      "epoch": 4.942363112391931,
+      "grad_norm": 0.028615275397896767,
+      "learning_rate": 5.01188059328386e-07,
+      "loss": 0.0001,
+      "step": 569
+    },
+    {
+      "epoch": 4.9510086455331415,
+      "grad_norm": 0.011685383506119251,
+      "learning_rate": 4.932209051725915e-07,
+      "loss": 0.0,
+      "step": 570
+    },
+    {
+      "epoch": 4.959654178674351,
+      "grad_norm": 0.006484777666628361,
+      "learning_rate": 4.853106486582499e-07,
+      "loss": 0.0,
+      "step": 571
+    },
+    {
+      "epoch": 4.968299711815562,
+      "grad_norm": 0.006807015277445316,
+      "learning_rate": 4.774575140626317e-07,
+      "loss": 0.0,
+      "step": 572
+    },
+    {
+      "epoch": 4.976945244956772,
+      "grad_norm": 0.005784816108644009,
+      "learning_rate": 4.6966172404344754e-07,
+      "loss": 0.0,
+      "step": 573
+    },
+    {
+      "epoch": 4.985590778097983,
+      "grad_norm": 0.0060491194017231464,
+      "learning_rate": 4.619234996325314e-07,
+      "loss": 0.0001,
+      "step": 574
+    },
+    {
+      "epoch": 4.994236311239193,
+      "grad_norm": 0.08785240352153778,
+      "learning_rate": 4.542430602295775e-07,
+      "loss": 0.0003,
+      "step": 575
+    },
+    {
+      "epoch": 5.008645533141211,
+      "grad_norm": 0.11832298338413239,
+      "learning_rate": 4.4662062359591584e-07,
+      "loss": 0.0001,
+      "step": 576
+    },
+    {
+      "epoch": 5.017291066282421,
+      "grad_norm": 0.006518571171909571,
+      "learning_rate": 4.3905640584834293e-07,
+      "loss": 0.0,
+      "step": 577
+    },
+    {
+      "epoch": 5.025936599423631,
+      "grad_norm": 0.008394552394747734,
+      "learning_rate": 4.3155062145298994e-07,
+      "loss": 0.0,
+      "step": 578
+    },
+    {
+      "epoch": 5.034582132564841,
+      "grad_norm": 0.014321188442409039,
+      "learning_rate": 4.241034832192434e-07,
+      "loss": 0.0,
+      "step": 579
+    },
+    {
+      "epoch": 5.043227665706052,
+      "grad_norm": 0.01109258271753788,
+      "learning_rate": 4.167152022937124e-07,
+      "loss": 0.0,
+      "step": 580
+    },
+    {
+      "epoch": 5.051873198847262,
+      "grad_norm": 0.00782827753573656,
+      "learning_rate": 4.0938598815424226e-07,
+      "loss": 0.0,
+      "step": 581
+    },
+    {
+      "epoch": 5.060518731988473,
+      "grad_norm": 0.006567042786628008,
+      "learning_rate": 4.0211604860397294e-07,
+      "loss": 0.0001,
+      "step": 582
+    },
+    {
+      "epoch": 5.069164265129683,
+      "grad_norm": 0.032781753689050674,
+      "learning_rate": 3.9490558976544967e-07,
+      "loss": 0.0,
+      "step": 583
+    },
+    {
+      "epoch": 5.077809798270893,
+      "grad_norm": 0.006290372461080551,
+      "learning_rate": 3.877548160747768e-07,
+      "loss": 0.0001,
+      "step": 584
+    },
+    {
+      "epoch": 5.086455331412104,
+      "grad_norm": 0.05169885233044624,
+      "learning_rate": 3.8066393027582276e-07,
+      "loss": 0.0001,
+      "step": 585
+    },
+    {
+      "epoch": 5.095100864553314,
+      "grad_norm": 0.024971339851617813,
+      "learning_rate": 3.7363313341447326e-07,
+      "loss": 0.0001,
+      "step": 586
+    },
+    {
+      "epoch": 5.1037463976945245,
+      "grad_norm": 0.007651512045413256,
+      "learning_rate": 3.666626248329272e-07,
+      "loss": 0.0,
+      "step": 587
+    },
+    {
+      "epoch": 5.112391930835735,
+      "grad_norm": 0.014207379892468452,
+      "learning_rate": 3.5975260216404714e-07,
+      "loss": 0.0,
+      "step": 588
+    },
+    {
+      "epoch": 5.121037463976945,
+      "grad_norm": 0.01968054287135601,
+      "learning_rate": 3.529032613257574e-07,
+      "loss": 0.0,
+      "step": 589
+    },
+    {
+      "epoch": 5.129682997118156,
+      "grad_norm": 0.015984253957867622,
+      "learning_rate": 3.4611479651548456e-07,
+      "loss": 0.0,
+      "step": 590
+    },
+    {
+      "epoch": 5.138328530259366,
+      "grad_norm": 0.008436279371380806,
+      "learning_rate": 3.393874002046577e-07,
+      "loss": 0.0001,
+      "step": 591
+    },
+    {
+      "epoch": 5.1469740634005765,
+      "grad_norm": 0.04552963003516197,
+      "learning_rate": 3.327212631332452e-07,
+      "loss": 0.0,
+      "step": 592
+    },
+    {
+      "epoch": 5.155619596541786,
+      "grad_norm": 0.015880795195698738,
+      "learning_rate": 3.2611657430435013e-07,
+      "loss": 0.0001,
+      "step": 593
+    },
+    {
+      "epoch": 5.164265129682997,
+      "grad_norm": 0.03732914850115776,
+      "learning_rate": 3.195735209788528e-07,
+      "loss": 0.0,
+      "step": 594
+    },
+    {
+      "epoch": 5.172910662824208,
+      "grad_norm": 0.007819831371307373,
+      "learning_rate": 3.130922886700968e-07,
+      "loss": 0.0,
+      "step": 595
+    },
+    {
+      "epoch": 5.181556195965418,
+      "grad_norm": 0.014321624301373959,
+      "learning_rate": 3.066730611386337e-07,
+      "loss": 0.0,
+      "step": 596
+    },
+    {
+      "epoch": 5.1902017291066285,
+      "grad_norm": 0.008290286175906658,
+      "learning_rate": 3.003160203870095e-07,
+      "loss": 0.0,
+      "step": 597
+    },
+    {
+      "epoch": 5.198847262247838,
+      "grad_norm": 0.006545263342559338,
+      "learning_rate": 2.940213466546085e-07,
+      "loss": 0.0,
+      "step": 598
+    },
+    {
+      "epoch": 5.207492795389049,
+      "grad_norm": 0.006481291726231575,
+      "learning_rate": 2.8778921841253774e-07,
+      "loss": 0.0002,
+      "step": 599
+    },
+    {
+      "epoch": 5.216138328530259,
+      "grad_norm": 0.07197843492031097,
+      "learning_rate": 2.8161981235857145e-07,
+      "loss": 0.0,
+      "step": 600
+    },
+    {
+      "epoch": 5.22478386167147,
+      "grad_norm": 0.014453639276325703,
+      "learning_rate": 2.7551330341213794e-07,
+      "loss": 0.0,
+      "step": 601
+    },
+    {
+      "epoch": 5.23342939481268,
+      "grad_norm": 0.006451854016631842,
+      "learning_rate": 2.6946986470936354e-07,
+      "loss": 0.0,
+      "step": 602
+    },
+    {
+      "epoch": 5.24207492795389,
+      "grad_norm": 0.009162579663097858,
+      "learning_rate": 2.634896675981599e-07,
+      "loss": 0.0001,
+      "step": 603
+    },
+    {
+      "epoch": 5.250720461095101,
+      "grad_norm": 0.029183892533183098,
+      "learning_rate": 2.5757288163336806e-07,
+      "loss": 0.0001,
+      "step": 604
+    },
+    {
+      "epoch": 5.259365994236311,
+      "grad_norm": 0.06650669872760773,
+      "learning_rate": 2.517196745719522e-07,
+      "loss": 0.0001,
+      "step": 605
+    },
+    {
+      "epoch": 5.268011527377522,
+      "grad_norm": 0.04278871789574623,
+      "learning_rate": 2.4593021236823916e-07,
+      "loss": 0.0,
+      "step": 606
+    },
+    {
+      "epoch": 5.276657060518732,
+      "grad_norm": 0.005027587525546551,
+      "learning_rate": 2.4020465916921866e-07,
+      "loss": 0.0,
+      "step": 607
+    },
+    {
+      "epoch": 5.285302593659942,
+      "grad_norm": 0.012285275384783745,
+      "learning_rate": 2.3454317730988414e-07,
+      "loss": 0.0,
+      "step": 608
+    },
+    {
+      "epoch": 5.293948126801153,
+      "grad_norm": 0.012882730923593044,
+      "learning_rate": 2.2894592730863336e-07,
+      "loss": 0.0,
+      "step": 609
+    },
+    {
+      "epoch": 5.302593659942363,
+      "grad_norm": 0.026212921366095543,
+      "learning_rate": 2.2341306786271695e-07,
+      "loss": 0.0001,
+      "step": 610
+    },
+    {
+      "epoch": 5.311239193083574,
+      "grad_norm": 0.007668660953640938,
+      "learning_rate": 2.1794475584373724e-07,
+      "loss": 0.0,
+      "step": 611
+    },
+    {
+      "epoch": 5.3198847262247835,
+      "grad_norm": 0.007228130474686623,
+      "learning_rate": 2.1254114629320228e-07,
+      "loss": 0.0,
+      "step": 612
+    },
+    {
+      "epoch": 5.328530259365994,
+      "grad_norm": 0.007553818169981241,
+      "learning_rate": 2.072023924181285e-07,
+      "loss": 0.0,
+      "step": 613
+    },
+    {
+      "epoch": 5.337175792507205,
+      "grad_norm": 0.028273316100239754,
+      "learning_rate": 2.019286455866981e-07,
+      "loss": 0.0001,
+      "step": 614
+    },
+    {
+      "epoch": 5.345821325648415,
+      "grad_norm": 0.017403678968548775,
+      "learning_rate": 1.967200553239676e-07,
+      "loss": 0.0,
+      "step": 615
+    },
+    {
+      "epoch": 5.354466858789626,
+      "grad_norm": 0.05251264199614525,
+      "learning_rate": 1.9157676930762702e-07,
+      "loss": 0.0,
+      "step": 616
+    },
+    {
+      "epoch": 5.363112391930835,
+      "grad_norm": 0.0052774338982999325,
+      "learning_rate": 1.8649893336381315e-07,
+      "loss": 0.0001,
+      "step": 617
+    },
+    {
+      "epoch": 5.371757925072046,
+      "grad_norm": 0.050105899572372437,
+      "learning_rate": 1.8148669146297564e-07,
+      "loss": 0.0,
+      "step": 618
+    },
+    {
+      "epoch": 5.380403458213257,
+      "grad_norm": 0.010968392714858055,
+      "learning_rate": 1.7654018571579557e-07,
+      "loss": 0.0,
+      "step": 619
+    },
+    {
+      "epoch": 5.389048991354467,
+      "grad_norm": 0.006572465877979994,
+      "learning_rate": 1.7165955636915395e-07,
+      "loss": 0.0,
+      "step": 620
+    },
+    {
+      "epoch": 5.3976945244956775,
+      "grad_norm": 0.015997696667909622,
+      "learning_rate": 1.6684494180215837e-07,
+      "loss": 0.0,
+      "step": 621
+    },
+    {
+      "epoch": 5.406340057636887,
+      "grad_norm": 0.01475190743803978,
+      "learning_rate": 1.620964785222162e-07,
+      "loss": 0.0,
+      "step": 622
+    },
+    {
+      "epoch": 5.414985590778098,
+      "grad_norm": 0.006957387086004019,
+      "learning_rate": 1.5741430116116812e-07,
+      "loss": 0.0,
+      "step": 623
+    },
+    {
+      "epoch": 5.423631123919308,
+      "grad_norm": 0.027358924970030785,
+      "learning_rate": 1.5279854247146703e-07,
+      "loss": 0.0,
+      "step": 624
+    },
+    {
+      "epoch": 5.432276657060519,
+      "grad_norm": 0.013768823817372322,
+      "learning_rate": 1.4824933332241693e-07,
+      "loss": 0.0,
+      "step": 625
+    },
+    {
+      "epoch": 5.4409221902017295,
+      "grad_norm": 0.010743286460638046,
+      "learning_rate": 1.4376680269646087e-07,
+      "loss": 0.0,
+      "step": 626
+    },
+    {
+      "epoch": 5.449567723342939,
+      "grad_norm": 0.007050006650388241,
+      "learning_rate": 1.3935107768552558e-07,
+      "loss": 0.0,
+      "step": 627
+    },
+    {
+      "epoch": 5.45821325648415,
+      "grad_norm": 0.012325974181294441,
+      "learning_rate": 1.3500228348741595e-07,
+      "loss": 0.0,
+      "step": 628
+    },
+    {
+      "epoch": 5.46685878962536,
+      "grad_norm": 0.023326944559812546,
+      "learning_rate": 1.307205434022671e-07,
+      "loss": 0.0,
+      "step": 629
+    },
+    {
+      "epoch": 5.475504322766571,
+      "grad_norm": 0.018656721338629723,
+      "learning_rate": 1.265059788290468e-07,
+      "loss": 0.0,
+      "step": 630
+    },
+    {
+      "epoch": 5.484149855907781,
+      "grad_norm": 0.017813757061958313,
+      "learning_rate": 1.223587092621162e-07,
+      "loss": 0.0,
+      "step": 631
+    },
+    {
+      "epoch": 5.492795389048991,
+      "grad_norm": 0.014615735039114952,
+      "learning_rate": 1.1827885228783865e-07,
+      "loss": 0.0,
+      "step": 632
+    },
+    {
+      "epoch": 5.501440922190202,
+      "grad_norm": 0.010695324279367924,
+      "learning_rate": 1.1426652358124817e-07,
+      "loss": 0.0,
+      "step": 633
+    },
+    {
+      "epoch": 5.510086455331412,
+      "grad_norm": 0.007640014868229628,
+      "learning_rate": 1.1032183690276754e-07,
+      "loss": 0.0,
+      "step": 634
+    },
+    {
+      "epoch": 5.518731988472623,
+      "grad_norm": 0.020597506314516068,
+      "learning_rate": 1.0644490409498636e-07,
+      "loss": 0.0001,
+      "step": 635
+    },
+    {
+      "epoch": 5.527377521613833,
+      "grad_norm": 0.004687067586928606,
+      "learning_rate": 1.0263583507948593e-07,
+      "loss": 0.0,
+      "step": 636
+    },
+    {
+      "epoch": 5.536023054755043,
+      "grad_norm": 0.007369566708803177,
+      "learning_rate": 9.889473785372555e-08,
+      "loss": 0.0,
+      "step": 637
+    },
+    {
+      "epoch": 5.544668587896254,
+      "grad_norm": 0.014998351223766804,
+      "learning_rate": 9.522171848797917e-08,
+      "loss": 0.0,
+      "step": 638
+    },
+    {
+      "epoch": 5.553314121037464,
+      "grad_norm": 0.005501555744558573,
+      "learning_rate": 9.161688112232836e-08,
+      "loss": 0.0,
+      "step": 639
+    },
+    {
+      "epoch": 5.561959654178675,
+      "grad_norm": 0.029769135639071465,
+      "learning_rate": 8.808032796371019e-08,
+      "loss": 0.0001,
+      "step": 640
+    },
+    {
+      "epoch": 5.5706051873198845,
+      "grad_norm": 0.01651173271238804,
+      "learning_rate": 8.461215928301819e-08,
+      "loss": 0.0,
+      "step": 641
+    },
+    {
+      "epoch": 5.579250720461095,
+      "grad_norm": 0.0046575190499424934,
+      "learning_rate": 8.121247341226074e-08,
+      "loss": 0.0,
+      "step": 642
+    },
+    {
+      "epoch": 5.587896253602305,
+      "grad_norm": 0.008450312539935112,
+      "learning_rate": 7.78813667417716e-08,
+      "loss": 0.0,
+      "step": 643
+    },
+    {
+      "epoch": 5.596541786743516,
+      "grad_norm": 0.007621787954121828,
+      "learning_rate": 7.46189337174788e-08,
+      "loss": 0.0,
+      "step": 644
+    },
+    {
+      "epoch": 5.605187319884726,
+      "grad_norm": 0.0086066210642457,
+      "learning_rate": 7.142526683822537e-08,
+      "loss": 0.0,
+      "step": 645
+    },
+    {
+      "epoch": 5.6138328530259365,
+      "grad_norm": 0.012834986671805382,
+      "learning_rate": 6.830045665314672e-08,
+      "loss": 0.0,
+      "step": 646
+    },
+    {
+      "epoch": 5.622478386167147,
+      "grad_norm": 0.005586333107203245,
+      "learning_rate": 6.524459175910464e-08,
+      "loss": 0.0,
+      "step": 647
+    },
+    {
+      "epoch": 5.631123919308357,
+      "grad_norm": 0.009992795065045357,
+      "learning_rate": 6.225775879817431e-08,
+      "loss": 0.0,
+      "step": 648
+    },
+    {
+      "epoch": 5.639769452449568,
+      "grad_norm": 0.006056077778339386,
+      "learning_rate": 5.934004245518793e-08,
+      "loss": 0.0,
+      "step": 649
+    },
+    {
+      "epoch": 5.648414985590778,
+      "grad_norm": 0.008581247180700302,
+      "learning_rate": 5.649152545533332e-08,
+      "loss": 0.0002,
+      "step": 650
+    },
+    {
+      "epoch": 5.6570605187319885,
+      "grad_norm": 0.05463126301765442,
+      "learning_rate": 5.371228856180993e-08,
+      "loss": 0.0,
+      "step": 651
+    },
+    {
+      "epoch": 5.665706051873199,
+      "grad_norm": 0.009801640175282955,
+      "learning_rate": 5.100241057353683e-08,
+      "loss": 0.0,
+      "step": 652
+    },
+    {
+      "epoch": 5.674351585014409,
+      "grad_norm": 0.008849680423736572,
+      "learning_rate": 4.8361968322920005e-08,
+      "loss": 0.0,
+      "step": 653
+    },
+    {
+      "epoch": 5.68299711815562,
+      "grad_norm": 0.008909969590604305,
+      "learning_rate": 4.579103667367385e-08,
+      "loss": 0.0001,
+      "step": 654
+    },
+    {
+      "epoch": 5.69164265129683,
+      "grad_norm": 0.7292889952659607,
+      "learning_rate": 4.328968851869758e-08,
+      "loss": 0.0,
+      "step": 655
+    },
+    {
+      "epoch": 5.70028818443804,
+      "grad_norm": 0.0066415779292583466,
+      "learning_rate": 4.0857994778009945e-08,
+      "loss": 0.0001,
+      "step": 656
+    },
+    {
+      "epoch": 5.708933717579251,
+      "grad_norm": 0.03848518431186676,
+      "learning_rate": 3.84960243967375e-08,
+      "loss": 0.0002,
+      "step": 657
+    },
+    {
+      "epoch": 5.717579250720461,
+      "grad_norm": 0.055817726999521255,
+      "learning_rate": 3.620384434316005e-08,
+      "loss": 0.0001,
+      "step": 658
+    },
+    {
+      "epoch": 5.726224783861672,
+      "grad_norm": 0.008442546240985394,
+      "learning_rate": 3.398151960681162e-08,
+      "loss": 0.0001,
+      "step": 659
+    },
+    {
+      "epoch": 5.734870317002882,
+      "grad_norm": 0.03930019959807396,
+      "learning_rate": 3.182911319663862e-08,
+      "loss": 0.0,
+      "step": 660
+    },
+    {
+      "epoch": 5.743515850144092,
+      "grad_norm": 0.012743629515171051,
+      "learning_rate": 2.9746686139212612e-08,
+      "loss": 0.0,
+      "step": 661
+    },
+    {
+      "epoch": 5.752161383285302,
+      "grad_norm": 0.008391858078539371,
+      "learning_rate": 2.7734297477000626e-08,
+      "loss": 0.0002,
+      "step": 662
+    },
+    {
+      "epoch": 5.760806916426513,
+      "grad_norm": 0.06344736367464066,
+      "learning_rate": 2.5792004266690095e-08,
+      "loss": 0.0,
+      "step": 663
+    },
+    {
+      "epoch": 5.769452449567723,
+      "grad_norm": 0.00729775195941329,
+      "learning_rate": 2.3919861577572924e-08,
+      "loss": 0.0,
+      "step": 664
+    },
+    {
+      "epoch": 5.778097982708934,
+      "grad_norm": 0.01558664534240961,
+      "learning_rate": 2.211792248998229e-08,
+      "loss": 0.0001,
+      "step": 665
+    },
+    {
+      "epoch": 5.786743515850144,
+      "grad_norm": 0.016456814482808113,
+      "learning_rate": 2.038623809378859e-08,
+      "loss": 0.0,
+      "step": 666
+    },
+    {
+      "epoch": 5.795389048991354,
+      "grad_norm": 0.008453302085399628,
+      "learning_rate": 1.872485748695113e-08,
+      "loss": 0.0,
+      "step": 667
+    },
+    {
+      "epoch": 5.804034582132565,
+      "grad_norm": 0.004515933804214001,
+      "learning_rate": 1.7133827774125368e-08,
+      "loss": 0.0001,
+      "step": 668
+    },
+    {
+      "epoch": 5.812680115273775,
+      "grad_norm": 0.02883203886449337,
+      "learning_rate": 1.5613194065327854e-08,
+      "loss": 0.0001,
+      "step": 669
+    },
+    {
+      "epoch": 5.821325648414986,
+      "grad_norm": 0.02576618641614914,
+      "learning_rate": 1.4162999474657268e-08,
+      "loss": 0.0,
+      "step": 670
+    },
+    {
+      "epoch": 5.829971181556196,
+      "grad_norm": 0.006904300302267075,
+      "learning_rate": 1.2783285119071232e-08,
+      "loss": 0.0,
+      "step": 671
+    },
+    {
+      "epoch": 5.838616714697406,
+      "grad_norm": 0.015726493671536446,
+      "learning_rate": 1.1474090117221948e-08,
+      "loss": 0.0001,
+      "step": 672
+    },
+    {
+      "epoch": 5.847262247838617,
+      "grad_norm": 0.032056134194135666,
+      "learning_rate": 1.023545158834599e-08,
+      "loss": 0.0,
+      "step": 673
+    },
+    {
+      "epoch": 5.855907780979827,
+      "grad_norm": 0.0056428308598697186,
+      "learning_rate": 9.067404651211808e-09,
+      "loss": 0.0001,
+      "step": 674
+    },
+    {
+      "epoch": 5.8645533141210375,
+      "grad_norm": 0.037972234189510345,
+      "learning_rate": 7.96998242312469e-09,
+      "loss": 0.0001,
+      "step": 675
+    },
+    {
+      "epoch": 5.873198847262248,
+      "grad_norm": 0.018561307340860367,
+      "learning_rate": 6.943216018987509e-09,
+      "loss": 0.0,
+      "step": 676
+    },
+    {
+      "epoch": 5.881844380403458,
+      "grad_norm": 0.014271295629441738,
+      "learning_rate": 5.987134550418106e-09,
+      "loss": 0.0,
+      "step": 677
+    },
+    {
+      "epoch": 5.890489913544669,
+      "grad_norm": 0.006300410255789757,
+      "learning_rate": 5.101765124925218e-09,
+      "loss": 0.0001,
+      "step": 678
+    },
+    {
+      "epoch": 5.899135446685879,
+      "grad_norm": 0.03615653142333031,
+      "learning_rate": 4.287132845137709e-09,
+      "loss": 0.0,
+      "step": 679
+    },
+    {
+      "epoch": 5.9077809798270895,
+      "grad_norm": 0.00793964322656393,
+      "learning_rate": 3.5432608080951392e-09,
+      "loss": 0.0,
+      "step": 680
+    },
+    {
+      "epoch": 5.916426512968299,
+      "grad_norm": 0.011542227119207382,
+      "learning_rate": 2.870170104591341e-09,
+      "loss": 0.0001,
+      "step": 681
+    },
+    {
+      "epoch": 5.92507204610951,
+      "grad_norm": 0.04817141219973564,
+      "learning_rate": 2.2678798185771235e-09,
+      "loss": 0.0,
+      "step": 682
+    },
+    {
+      "epoch": 5.93371757925072,
+      "grad_norm": 0.006957252975553274,
+      "learning_rate": 1.7364070266193135e-09,
+      "loss": 0.0001,
+      "step": 683
+    },
+    {
+      "epoch": 5.942363112391931,
+      "grad_norm": 0.028225049376487732,
+      "learning_rate": 1.2757667974155896e-09,
+      "loss": 0.0002,
+      "step": 684
+    },
+    {
+      "epoch": 5.9510086455331415,
+      "grad_norm": 0.060419417917728424,
+      "learning_rate": 8.859721913684339e-10,
+      "loss": 0.0,
+      "step": 685
+    },
+    {
+      "epoch": 5.959654178674351,
+      "grad_norm": 0.005427508614957333,
+      "learning_rate": 5.670342602148715e-10,
+      "loss": 0.0001,
+      "step": 686
+    },
+    {
+      "epoch": 5.968299711815562,
+      "grad_norm": 0.031979430466890335,
+      "learning_rate": 3.1896204671144627e-10,
+      "loss": 0.0001,
+      "step": 687
+    },
+    {
+      "epoch": 5.976945244956772,
+      "grad_norm": 0.015397738665342331,
+      "learning_rate": 1.4176258437970103e-10,
+      "loss": 0.0001,
+      "step": 688
+    },
+    {
+      "epoch": 5.985590778097983,
+      "grad_norm": 0.03583737090229988,
+      "learning_rate": 3.544089730633804e-11,
+      "loss": 0.0001,
+      "step": 689
+    },
+    {
+      "epoch": 5.994236311239193,
+      "grad_norm": 0.044873710721731186,
+      "learning_rate": 0.0,
+      "loss": 0.0,
+      "step": 690
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 690,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 6,
+  "save_steps": 115,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.747175381925888e+18,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}