diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,14021 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.2293314986813439,
+  "eval_steps": 500,
+  "global_step": 2000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.0,
+      "learning_rate": 0.0,
+      "loss": 6.2779,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.0,
+      "learning_rate": 0.0,
+      "loss": 6.1581,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 14.976716136383175,
+      "learning_rate": 3.816793893129771e-06,
+      "loss": 6.2063,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 14.397488353328699,
+      "learning_rate": 7.633587786259541e-06,
+      "loss": 6.2989,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 14.397488353328699,
+      "learning_rate": 7.633587786259541e-06,
+      "loss": 6.2908,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 18.89007945034403,
+      "learning_rate": 1.1450381679389314e-05,
+      "loss": 6.0954,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 18.89007945034403,
+      "learning_rate": 1.1450381679389314e-05,
+      "loss": 6.2841,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 11.409346480686976,
+      "learning_rate": 1.5267175572519083e-05,
+      "loss": 6.3522,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 28.346056360110595,
+      "learning_rate": 1.9083969465648855e-05,
+      "loss": 6.2796,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 15.852481128362015,
+      "learning_rate": 2.2900763358778628e-05,
+      "loss": 6.2723,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 15.852481128362015,
+      "learning_rate": 2.2900763358778628e-05,
+      "loss": 5.9952,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 48.40977193887413,
+      "learning_rate": 2.6717557251908397e-05,
+      "loss": 6.1167,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 9.259901239279529,
+      "learning_rate": 3.0534351145038166e-05,
+      "loss": 6.2158,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 11.429600547441563,
+      "learning_rate": 3.435114503816794e-05,
+      "loss": 6.1169,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 11.429600547441563,
+      "learning_rate": 3.435114503816794e-05,
+      "loss": 6.1389,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 15.704367677180992,
+      "learning_rate": 3.816793893129771e-05,
+      "loss": 6.0987,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 33.68032910168496,
+      "learning_rate": 4.198473282442748e-05,
+      "loss": 6.0965,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 11.896102639886562,
+      "learning_rate": 4.5801526717557256e-05,
+      "loss": 5.6258,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 15.481580862690418,
+      "learning_rate": 4.9618320610687025e-05,
+      "loss": 5.9343,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 12.662713159465534,
+      "learning_rate": 5.3435114503816794e-05,
+      "loss": 5.7158,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 15.649889306348113,
+      "learning_rate": 5.725190839694656e-05,
+      "loss": 5.8075,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 65.94567816487748,
+      "learning_rate": 6.106870229007633e-05,
+      "loss": 5.6104,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 13.500161461978536,
+      "learning_rate": 6.488549618320611e-05,
+      "loss": 5.7245,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 39.63275193406311,
+      "learning_rate": 6.870229007633588e-05,
+      "loss": 5.4938,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 25.79160793928747,
+      "learning_rate": 7.251908396946565e-05,
+      "loss": 5.8096,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 8.833397229553675,
+      "learning_rate": 7.633587786259542e-05,
+      "loss": 5.6985,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 7.395053164765045,
+      "learning_rate": 8.015267175572518e-05,
+      "loss": 5.6925,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 41.89909059331258,
+      "learning_rate": 8.396946564885496e-05,
+      "loss": 5.4813,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 10.966138043160779,
+      "learning_rate": 8.778625954198472e-05,
+      "loss": 5.6698,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 12.7283315374098,
+      "learning_rate": 9.160305343511451e-05,
+      "loss": 5.6617,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 11.675016207985045,
+      "learning_rate": 9.541984732824429e-05,
+      "loss": 5.2054,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 9.324595112650963,
+      "learning_rate": 9.923664122137405e-05,
+      "loss": 5.2573,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 5.462088289687755,
+      "learning_rate": 0.00010305343511450383,
+      "loss": 5.3966,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 28.258431587421967,
+      "learning_rate": 0.00010687022900763359,
+      "loss": 5.5193,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 8.847354229930158,
+      "learning_rate": 0.00011068702290076336,
+      "loss": 5.1725,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 5.897059143557035,
+      "learning_rate": 0.00011450381679389313,
+      "loss": 5.1518,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 5.112023703007317,
+      "learning_rate": 0.0001183206106870229,
+      "loss": 5.1591,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 5.073706203205232,
+      "learning_rate": 0.00012213740458015266,
+      "loss": 5.0398,
+      "step": 38
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 10.989327102728486,
+      "learning_rate": 0.00012595419847328244,
+      "loss": 5.112,
+      "step": 39
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 8.494888378939828,
+      "learning_rate": 0.00012977099236641222,
+      "loss": 5.2983,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 6.029727471334503,
+      "learning_rate": 0.000133587786259542,
+      "loss": 5.1413,
+      "step": 41
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 10.580341148041809,
+      "learning_rate": 0.00013740458015267177,
+      "loss": 5.3161,
+      "step": 42
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 7.861892307202481,
+      "learning_rate": 0.00014122137404580154,
+      "loss": 5.0632,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.032287490494244,
+      "learning_rate": 0.0001450381679389313,
+      "loss": 5.0655,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.226651239598461,
+      "learning_rate": 0.00014885496183206107,
+      "loss": 5.2151,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.2839008800377965,
+      "learning_rate": 0.00015267175572519084,
+      "loss": 5.2635,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 16.208329066599074,
+      "learning_rate": 0.00015648854961832062,
+      "loss": 5.2023,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 13.549653768489149,
+      "learning_rate": 0.00016030534351145037,
+      "loss": 5.2272,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.811721203312905,
+      "learning_rate": 0.00016412213740458014,
+      "loss": 4.9626,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 6.572309930980864,
+      "learning_rate": 0.00016793893129770992,
+      "loss": 5.0291,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.17916044761434,
+      "learning_rate": 0.0001717557251908397,
+      "loss": 5.1589,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 7.194489680135254,
+      "learning_rate": 0.00017557251908396944,
+      "loss": 5.2516,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 10.55005706873481,
+      "learning_rate": 0.00017938931297709925,
+      "loss": 5.0086,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.6321119276045897,
+      "learning_rate": 0.00018320610687022902,
+      "loss": 5.0754,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 7.810860089915512,
+      "learning_rate": 0.0001870229007633588,
+      "loss": 4.9908,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.940902186870088,
+      "learning_rate": 0.00019083969465648857,
+      "loss": 4.9833,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.611002381269936,
+      "learning_rate": 0.00019465648854961832,
+      "loss": 5.0436,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 20.054002059058206,
+      "learning_rate": 0.0001984732824427481,
+      "loss": 5.0632,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.911527035099265,
+      "learning_rate": 0.00020229007633587788,
+      "loss": 4.6758,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 11.37665907428031,
+      "learning_rate": 0.00020610687022900765,
+      "loss": 5.1242,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.704198024162129,
+      "learning_rate": 0.0002099236641221374,
+      "loss": 5.1245,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.7871548009528184,
+      "learning_rate": 0.00021374045801526718,
+      "loss": 4.891,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.136501595775406,
+      "learning_rate": 0.00021755725190839695,
+      "loss": 4.7869,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.568374478750777,
+      "learning_rate": 0.00022137404580152673,
+      "loss": 4.8957,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.697055076084128,
+      "learning_rate": 0.00022519083969465648,
+      "loss": 4.5216,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.569269081902782,
+      "learning_rate": 0.00022900763358778625,
+      "loss": 4.9846,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 6.716615480986148,
+      "learning_rate": 0.00023282442748091603,
+      "loss": 4.7787,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.3323716057495862,
+      "learning_rate": 0.0002366412213740458,
+      "loss": 4.8606,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 9.383012479834845,
+      "learning_rate": 0.00024045801526717558,
+      "loss": 4.7711,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.032737783937775,
+      "learning_rate": 0.00024427480916030533,
+      "loss": 4.6477,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.309847236686972,
+      "learning_rate": 0.00024809160305343513,
+      "loss": 4.7642,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.573701930155391,
+      "learning_rate": 0.0002519083969465649,
+      "loss": 4.8587,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 8.322277059585852,
+      "learning_rate": 0.00025572519083969463,
+      "loss": 4.5204,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.665817232441201,
+      "learning_rate": 0.00025954198473282443,
+      "loss": 4.7847,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 139.03167086367495,
+      "learning_rate": 0.0002633587786259542,
+      "loss": 4.6664,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 17.6803224637041,
+      "learning_rate": 0.000267175572519084,
+      "loss": 5.0634,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 22.524418929291922,
+      "learning_rate": 0.00027099236641221373,
+      "loss": 5.5552,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 6.950517134284506,
+      "learning_rate": 0.00027480916030534353,
+      "loss": 5.3287,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 15.379633137654181,
+      "learning_rate": 0.0002786259541984733,
+      "loss": 5.2802,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 20.268995253997705,
+      "learning_rate": 0.0002824427480916031,
+      "loss": 5.0836,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 29.259856808399377,
+      "learning_rate": 0.0002862595419847328,
+      "loss": 5.2734,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 9.348142659709751,
+      "learning_rate": 0.0002900763358778626,
+      "loss": 5.0488,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 14.07726663498388,
+      "learning_rate": 0.0002938931297709924,
+      "loss": 5.0828,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 38.94377862083665,
+      "learning_rate": 0.00029770992366412214,
+      "loss": 5.059,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.189460782597058,
+      "learning_rate": 0.00030152671755725194,
+      "loss": 4.9703,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 11.055190070541768,
+      "learning_rate": 0.0003053435114503817,
+      "loss": 4.7264,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 9.82024718477547,
+      "learning_rate": 0.0003091603053435115,
+      "loss": 5.0345,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.978344495248824,
+      "learning_rate": 0.00031297709923664124,
+      "loss": 4.7281,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.101628294492602,
+      "learning_rate": 0.000316793893129771,
+      "loss": 4.6132,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.145155139352889,
+      "learning_rate": 0.00032061068702290074,
+      "loss": 4.8203,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.541789314626976,
+      "learning_rate": 0.00032442748091603054,
+      "loss": 4.9099,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 22.161052158363848,
+      "learning_rate": 0.0003282442748091603,
+      "loss": 4.9713,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.7331702462659,
+      "learning_rate": 0.0003320610687022901,
+      "loss": 4.8406,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.288023385461357,
+      "learning_rate": 0.00033587786259541984,
+      "loss": 4.7399,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 10.254748606389127,
+      "learning_rate": 0.00033969465648854964,
+      "loss": 4.6848,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 9.771604247240735,
+      "learning_rate": 0.0003435114503816794,
+      "loss": 4.7795,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.695161580609848,
+      "learning_rate": 0.0003473282442748092,
+      "loss": 4.5722,
+      "step": 97
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.980790618859492,
+      "learning_rate": 0.0003511450381679389,
+      "loss": 4.6186,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 22.240432294680442,
+      "learning_rate": 0.0003549618320610687,
+      "loss": 4.825,
+      "step": 99
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 8.239680655788304,
+      "learning_rate": 0.0003587786259541985,
+      "loss": 4.6153,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 7.551497396505139,
+      "learning_rate": 0.00036259541984732824,
+      "loss": 4.6794,
+      "step": 101
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 7.914826041030228,
+      "learning_rate": 0.00036641221374045805,
+      "loss": 4.6659,
+      "step": 102
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.029399027607916,
+      "learning_rate": 0.0003702290076335878,
+      "loss": 4.5387,
+      "step": 103
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.660324171777086,
+      "learning_rate": 0.0003740458015267176,
+      "loss": 4.5091,
+      "step": 104
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 7.040733663725759,
+      "learning_rate": 0.00037786259541984735,
+      "loss": 4.7529,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.8592159543896996,
+      "learning_rate": 0.00038167938931297715,
+      "loss": 4.7159,
+      "step": 106
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.6294828525314395,
+      "learning_rate": 0.00038549618320610684,
+      "loss": 4.7294,
+      "step": 107
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.979668727923852,
+      "learning_rate": 0.00038931297709923665,
+      "loss": 4.3714,
+      "step": 108
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.581717881378127,
+      "learning_rate": 0.0003931297709923664,
+      "loss": 4.6298,
+      "step": 109
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.448372221263443,
+      "learning_rate": 0.0003969465648854962,
+      "loss": 4.7571,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.2948860235700668,
+      "learning_rate": 0.00040076335877862595,
+      "loss": 4.3805,
+      "step": 111
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.9831330136880223,
+      "learning_rate": 0.00040458015267175575,
+      "loss": 4.5136,
+      "step": 112
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.6498001940669886,
+      "learning_rate": 0.0004083969465648855,
+      "loss": 4.49,
+      "step": 113
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.3364447260683896,
+      "learning_rate": 0.0004122137404580153,
+      "loss": 4.5943,
+      "step": 114
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.991466654429088,
+      "learning_rate": 0.00041603053435114505,
+      "loss": 4.52,
+      "step": 115
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.082429278038854,
+      "learning_rate": 0.0004198473282442748,
+      "loss": 4.4996,
+      "step": 116
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 52.230357543862944,
+      "learning_rate": 0.00042366412213740455,
+      "loss": 4.7485,
+      "step": 117
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.21748080112374,
+      "learning_rate": 0.00042748091603053435,
+      "loss": 4.4279,
+      "step": 118
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.91623677075748,
+      "learning_rate": 0.00043129770992366415,
+      "loss": 4.3156,
+      "step": 119
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.450892416973839,
+      "learning_rate": 0.0004351145038167939,
+      "loss": 4.3349,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.104235347067771,
+      "learning_rate": 0.0004389312977099237,
+      "loss": 4.7161,
+      "step": 121
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.0993732703424324,
+      "learning_rate": 0.00044274809160305345,
+      "loss": 4.4699,
+      "step": 122
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.960992832423532,
+      "learning_rate": 0.00044656488549618326,
+      "loss": 4.3197,
+      "step": 123
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.212027498483674,
+      "learning_rate": 0.00045038167938931295,
+      "loss": 4.5707,
+      "step": 124
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.4869173074265962,
+      "learning_rate": 0.00045419847328244275,
+      "loss": 4.5839,
+      "step": 125
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.181475099579808,
+      "learning_rate": 0.0004580152671755725,
+      "loss": 4.5064,
+      "step": 126
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.1942828551524878,
+      "learning_rate": 0.0004618320610687023,
+      "loss": 4.3979,
+      "step": 127
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 6.115485717946443,
+      "learning_rate": 0.00046564885496183206,
+      "loss": 4.3856,
+      "step": 128
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.141524556275388,
+      "learning_rate": 0.00046946564885496186,
+      "loss": 4.5029,
+      "step": 129
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.382205404063897,
+      "learning_rate": 0.0004732824427480916,
+      "loss": 4.3509,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 5.5162684169269705,
+      "learning_rate": 0.0004770992366412214,
+      "loss": 4.4551,
+      "step": 131
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.9351584500250985,
+      "learning_rate": 0.00048091603053435116,
+      "loss": 4.5373,
+      "step": 132
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.9141791947655433,
+      "learning_rate": 0.0004847328244274809,
+      "loss": 4.29,
+      "step": 133
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.124128141029574,
+      "learning_rate": 0.0004885496183206107,
+      "loss": 4.2364,
+      "step": 134
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.9783730396723103,
+      "learning_rate": 0.0004923664122137404,
+      "loss": 4.6155,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.07325353745457,
+      "learning_rate": 0.0004961832061068703,
+      "loss": 4.5372,
+      "step": 136
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.8313079291733185,
+      "learning_rate": 0.0005,
+      "loss": 4.3088,
+      "step": 137
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.152865710451047,
+      "learning_rate": 0.0005038167938931298,
+      "loss": 4.6045,
+      "step": 138
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.3967945768183245,
+      "learning_rate": 0.0005076335877862596,
+      "loss": 4.2817,
+      "step": 139
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.4644266643742863,
+      "learning_rate": 0.0005114503816793893,
+      "loss": 4.4599,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.544944406113682,
+      "learning_rate": 0.0005152671755725191,
+      "loss": 4.1953,
+      "step": 141
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 6.150666133802115,
+      "learning_rate": 0.0005190839694656489,
+      "loss": 4.2241,
+      "step": 142
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.705915106052184,
+      "learning_rate": 0.0005229007633587787,
+      "loss": 4.2867,
+      "step": 143
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.1087313750685692,
+      "learning_rate": 0.0005267175572519084,
+      "loss": 4.2412,
+      "step": 144
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.289305669092866,
+      "learning_rate": 0.0005305343511450382,
+      "loss": 4.3098,
+      "step": 145
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.4885574223852382,
+      "learning_rate": 0.000534351145038168,
+      "loss": 4.4438,
+      "step": 146
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.087139734866523,
+      "learning_rate": 0.0005381679389312977,
+      "loss": 4.2856,
+      "step": 147
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.6985085546222614,
+      "learning_rate": 0.0005419847328244275,
+      "loss": 4.4142,
+      "step": 148
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.727728121506411,
+      "learning_rate": 0.0005458015267175572,
+      "loss": 4.4135,
+      "step": 149
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.948736840465955,
+      "learning_rate": 0.0005496183206106871,
+      "loss": 4.5285,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.4424225307169785,
+      "learning_rate": 0.0005534351145038168,
+      "loss": 4.0518,
+      "step": 151
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.2461109523508633,
+      "learning_rate": 0.0005572519083969466,
+      "loss": 4.3069,
+      "step": 152
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.0770259323732496,
+      "learning_rate": 0.0005610687022900763,
+      "loss": 4.3466,
+      "step": 153
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.5353662806779629,
+      "learning_rate": 0.0005648854961832062,
+      "loss": 4.2371,
+      "step": 154
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.0783920823522064,
+      "learning_rate": 0.0005687022900763359,
+      "loss": 4.44,
+      "step": 155
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.106928793234778,
+      "learning_rate": 0.0005725190839694656,
+      "loss": 4.3789,
+      "step": 156
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.3997577002285153,
+      "learning_rate": 0.0005763358778625954,
+      "loss": 4.2723,
+      "step": 157
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 125.74215330630898,
+      "learning_rate": 0.0005801526717557252,
+      "loss": 4.5233,
+      "step": 158
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.0252204462249193,
+      "learning_rate": 0.000583969465648855,
+      "loss": 4.3092,
+      "step": 159
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.0938463664760625,
+      "learning_rate": 0.0005877862595419848,
+      "loss": 4.4223,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 5.469689179908854,
+      "learning_rate": 0.0005916030534351145,
+      "loss": 4.4815,
+      "step": 161
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 16.71766273367972,
+      "learning_rate": 0.0005954198473282443,
+      "loss": 4.3158,
+      "step": 162
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 9.235621807932343,
+      "learning_rate": 0.0005992366412213741,
+      "loss": 4.5763,
+      "step": 163
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 5.078315302431695,
+      "learning_rate": 0.0006030534351145039,
+      "loss": 4.6196,
+      "step": 164
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.1877889855812995,
+      "learning_rate": 0.0006068702290076335,
+      "loss": 4.427,
+      "step": 165
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 29.50152144195882,
+      "learning_rate": 0.0006106870229007634,
+      "loss": 4.3172,
+      "step": 166
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.7986073777991605,
+      "learning_rate": 0.0006145038167938931,
+      "loss": 4.487,
+      "step": 167
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.3206921118180057,
+      "learning_rate": 0.000618320610687023,
+      "loss": 4.5846,
+      "step": 168
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.9789890885059767,
+      "learning_rate": 0.0006221374045801526,
+      "loss": 4.4977,
+      "step": 169
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.7208789856933224,
+      "learning_rate": 0.0006259541984732825,
+      "loss": 4.3776,
+      "step": 170
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.529932794834876,
+      "learning_rate": 0.0006297709923664122,
+      "loss": 4.5379,
+      "step": 171
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 5.069807212846386,
+      "learning_rate": 0.000633587786259542,
+      "loss": 4.538,
+      "step": 172
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.843360393508488,
+      "learning_rate": 0.0006374045801526717,
+      "loss": 4.4068,
+      "step": 173
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.979015124661183,
+      "learning_rate": 0.0006412213740458015,
+      "loss": 4.3093,
+      "step": 174
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 23.118008024134554,
+      "learning_rate": 0.0006450381679389313,
+      "loss": 4.5523,
+      "step": 175
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.676702098627031,
+      "learning_rate": 0.0006488549618320611,
+      "loss": 4.1649,
+      "step": 176
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 5.48001509237802,
+      "learning_rate": 0.0006526717557251909,
+      "loss": 4.1914,
+      "step": 177
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.164314292544173,
+      "learning_rate": 0.0006564885496183206,
+      "loss": 4.5253,
+      "step": 178
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.797441546838409,
+      "learning_rate": 0.0006603053435114504,
+      "loss": 4.5076,
+      "step": 179
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.0967654547103542,
+      "learning_rate": 0.0006641221374045802,
+      "loss": 4.3244,
+      "step": 180
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.6853708103680716,
+      "learning_rate": 0.0006679389312977099,
+      "loss": 4.2549,
+      "step": 181
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.5984768315593856,
+      "learning_rate": 0.0006717557251908397,
+      "loss": 4.3358,
+      "step": 182
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.1474846376523535,
+      "learning_rate": 0.0006755725190839694,
+      "loss": 4.3137,
+      "step": 183
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.7440657356867424,
+      "learning_rate": 0.0006793893129770993,
+      "loss": 4.6206,
+      "step": 184
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.7374322470508698,
+      "learning_rate": 0.000683206106870229,
+      "loss": 4.0924,
+      "step": 185
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.552721310519864,
+      "learning_rate": 0.0006870229007633588,
+      "loss": 4.2106,
+      "step": 186
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.639952888552088,
+      "learning_rate": 0.0006908396946564885,
+      "loss": 4.4707,
+      "step": 187
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.4401389809364398,
+      "learning_rate": 0.0006946564885496184,
+      "loss": 4.4782,
+      "step": 188
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.693920501101157,
+      "learning_rate": 0.0006984732824427481,
+      "loss": 4.3477,
+      "step": 189
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.820752534444456,
+      "learning_rate": 0.0007022900763358778,
+      "loss": 4.3333,
+      "step": 190
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.454411472460561,
+      "learning_rate": 0.0007061068702290076,
+      "loss": 4.3562,
+      "step": 191
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.339729032032386,
+      "learning_rate": 0.0007099236641221374,
+      "loss": 4.3534,
+      "step": 192
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.7533030617578518,
+      "learning_rate": 0.0007137404580152672,
+      "loss": 4.1145,
+      "step": 193
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.5959634267068015,
+      "learning_rate": 0.000717557251908397,
+      "loss": 4.503,
+      "step": 194
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.8956991079901098,
+      "learning_rate": 0.0007213740458015267,
+      "loss": 4.552,
+      "step": 195
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.4517139263532806,
+      "learning_rate": 0.0007251908396946565,
+      "loss": 4.5581,
+      "step": 196
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.5816501245363008,
+      "learning_rate": 0.0007290076335877863,
+      "loss": 4.1946,
+      "step": 197
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.2445169239722784,
+      "learning_rate": 0.0007328244274809161,
+      "loss": 4.2591,
+      "step": 198
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.2714761168683029,
+      "learning_rate": 0.0007366412213740457,
+      "loss": 4.4627,
+      "step": 199
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.5303754696273073,
+      "learning_rate": 0.0007404580152671756,
+      "loss": 4.4048,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.2085041225393867,
+      "learning_rate": 0.0007442748091603053,
+      "loss": 4.2643,
+      "step": 201
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.9112308799012625,
+      "learning_rate": 0.0007480916030534352,
+      "loss": 4.2295,
+      "step": 202
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.5908188587207266,
+      "learning_rate": 0.0007519083969465648,
+      "loss": 4.2813,
+      "step": 203
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.1982689595146219,
+      "learning_rate": 0.0007557251908396947,
+      "loss": 4.247,
+      "step": 204
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.3881318187911207,
+      "learning_rate": 0.0007595419847328244,
+      "loss": 4.3874,
+      "step": 205
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.3609424524329823,
+      "learning_rate": 0.0007633587786259543,
+      "loss": 4.4435,
+      "step": 206
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.1937703208585233,
+      "learning_rate": 0.0007671755725190839,
+      "loss": 4.5193,
+      "step": 207
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.4920614705406754,
+      "learning_rate": 0.0007709923664122137,
+      "loss": 4.3129,
+      "step": 208
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 7.240631868415255,
+      "learning_rate": 0.0007748091603053435,
+      "loss": 4.3302,
+      "step": 209
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.653358300173251,
+      "learning_rate": 0.0007786259541984733,
+      "loss": 4.3083,
+      "step": 210
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.6839357956548613,
+      "learning_rate": 0.000782442748091603,
+      "loss": 4.3758,
+      "step": 211
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.3938372295633423,
+      "learning_rate": 0.0007862595419847328,
+      "loss": 4.4719,
+      "step": 212
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.324803896504514,
+      "learning_rate": 0.0007900763358778626,
+      "loss": 4.3297,
+      "step": 213
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.2204173773929294,
+      "learning_rate": 0.0007938931297709924,
+      "loss": 4.389,
+      "step": 214
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 11.811043448698852,
+      "learning_rate": 0.0007977099236641223,
+      "loss": 4.3733,
+      "step": 215
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.0040900817656526,
+      "learning_rate": 0.0008015267175572519,
+      "loss": 4.0515,
+      "step": 216
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.9331803662548388,
+      "learning_rate": 0.0008053435114503816,
+      "loss": 4.4991,
+      "step": 217
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.4537028726656207,
+      "learning_rate": 0.0008091603053435115,
+      "loss": 4.2445,
+      "step": 218
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 5.53463013377651,
+      "learning_rate": 0.0008129770992366412,
+      "loss": 4.1712,
+      "step": 219
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.5106546596987123,
+      "learning_rate": 0.000816793893129771,
+      "loss": 4.3001,
+      "step": 220
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.9745634275850867,
+      "learning_rate": 0.0008206106870229007,
+      "loss": 4.2108,
+      "step": 221
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.268432076903743,
+      "learning_rate": 0.0008244274809160306,
+      "loss": 4.5328,
+      "step": 222
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.1903152686894933,
+      "learning_rate": 0.0008282442748091604,
+      "loss": 4.2968,
+      "step": 223
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.9380468409358829,
+      "learning_rate": 0.0008320610687022901,
+      "loss": 4.4417,
+      "step": 224
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.3963413799534377,
+      "learning_rate": 0.0008358778625954198,
+      "loss": 4.2871,
+      "step": 225
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.43582208783276,
+      "learning_rate": 0.0008396946564885496,
+      "loss": 4.2819,
+      "step": 226
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.5940462815400278,
+      "learning_rate": 0.0008435114503816795,
+      "loss": 4.4052,
+      "step": 227
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.2051802193213408,
+      "learning_rate": 0.0008473282442748091,
+      "loss": 4.2222,
+      "step": 228
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.0333850802718665,
+      "learning_rate": 0.000851145038167939,
+      "loss": 4.4192,
+      "step": 229
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.400417782835079,
+      "learning_rate": 0.0008549618320610687,
+      "loss": 4.4592,
+      "step": 230
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.7577865572295235,
+      "learning_rate": 0.0008587786259541986,
+      "loss": 4.2213,
+      "step": 231
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.4174422384212568,
+      "learning_rate": 0.0008625954198473283,
+      "loss": 4.4664,
+      "step": 232
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.055376126295404,
+      "learning_rate": 0.0008664122137404581,
+      "loss": 4.2514,
+      "step": 233
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.504414708093824,
+      "learning_rate": 0.0008702290076335878,
+      "loss": 4.0844,
+      "step": 234
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.4343526319105533,
+      "learning_rate": 0.0008740458015267176,
+      "loss": 4.2079,
+      "step": 235
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.2562820442089322,
+      "learning_rate": 0.0008778625954198474,
+      "loss": 4.4373,
+      "step": 236
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.7457012916108225,
+      "learning_rate": 0.000881679389312977,
+      "loss": 4.1786,
+      "step": 237
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1790063308907726,
+      "learning_rate": 0.0008854961832061069,
+      "loss": 4.2874,
+      "step": 238
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.2991863547672804,
+      "learning_rate": 0.0008893129770992367,
+      "loss": 4.1354,
+      "step": 239
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.5286489750714036,
+      "learning_rate": 0.0008931297709923665,
+      "loss": 4.3006,
+      "step": 240
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.558510226836997,
+      "learning_rate": 0.0008969465648854962,
+      "loss": 4.0331,
+      "step": 241
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.4903547154898784,
+      "learning_rate": 0.0009007633587786259,
+      "loss": 4.1535,
+      "step": 242
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.2714238683862837,
+      "learning_rate": 0.0009045801526717558,
+      "loss": 4.1045,
+      "step": 243
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 8.463507600954246,
+      "learning_rate": 0.0009083969465648855,
+      "loss": 4.2937,
+      "step": 244
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.3496143257230424,
+      "learning_rate": 0.0009122137404580153,
+      "loss": 4.2393,
+      "step": 245
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.0867375064306093,
+      "learning_rate": 0.000916030534351145,
+      "loss": 4.2803,
+      "step": 246
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.6739263840471548,
+      "learning_rate": 0.0009198473282442749,
+      "loss": 4.147,
+      "step": 247
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.627907574888252,
+      "learning_rate": 0.0009236641221374046,
+      "loss": 4.1545,
+      "step": 248
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.6288505381343623,
+      "learning_rate": 0.0009274809160305345,
+      "loss": 4.2188,
+      "step": 249
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.5053658945789288,
+      "learning_rate": 0.0009312977099236641,
+      "loss": 4.2284,
+      "step": 250
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.9751834803714665,
+      "learning_rate": 0.0009351145038167939,
+      "loss": 4.3051,
+      "step": 251
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.635500123712402,
+      "learning_rate": 0.0009389312977099237,
+      "loss": 4.1021,
+      "step": 252
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.18029868298015,
+      "learning_rate": 0.0009427480916030535,
+      "loss": 4.323,
+      "step": 253
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.294919077766256,
+      "learning_rate": 0.0009465648854961832,
+      "loss": 4.2792,
+      "step": 254
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.4364356766600967,
+      "learning_rate": 0.000950381679389313,
+      "loss": 4.4129,
+      "step": 255
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 6.557641748511485,
+      "learning_rate": 0.0009541984732824428,
+      "loss": 4.529,
+      "step": 256
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 8.226298705783996,
+      "learning_rate": 0.0009580152671755726,
+      "loss": 3.8596,
+      "step": 257
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.8150668785318773,
+      "learning_rate": 0.0009618320610687023,
+      "loss": 4.2955,
+      "step": 258
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.4614750127674727,
+      "learning_rate": 0.0009656488549618321,
+      "loss": 4.1755,
+      "step": 259
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.7379610237004675,
+      "learning_rate": 0.0009694656488549618,
+      "loss": 3.9348,
+      "step": 260
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.2592173885528375,
+      "learning_rate": 0.0009732824427480917,
+      "loss": 4.3136,
+      "step": 261
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.211981835728531,
+      "learning_rate": 0.0009770992366412213,
+      "loss": 4.2098,
+      "step": 262
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.577569854133963,
+      "learning_rate": 0.0009809160305343512,
+      "loss": 4.2233,
+      "step": 263
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.9361841564074918,
+      "learning_rate": 0.0009847328244274808,
+      "loss": 4.3255,
+      "step": 264
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.0544551880751634,
+      "learning_rate": 0.0009885496183206107,
+      "loss": 4.26,
+      "step": 265
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.6958226014336415,
+      "learning_rate": 0.0009923664122137405,
+      "loss": 4.2677,
+      "step": 266
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.7074348821947696,
+      "learning_rate": 0.0009961832061068704,
+      "loss": 3.9851,
+      "step": 267
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.299170018256827,
+      "learning_rate": 0.001,
+      "loss": 4.26,
+      "step": 268
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.814824816036798,
+      "learning_rate": 0.0009999999655172654,
+      "loss": 4.0256,
+      "step": 269
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.8008135241767156,
+      "learning_rate": 0.0009999998620690664,
+      "loss": 4.2543,
+      "step": 270
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.301365760272003,
+      "learning_rate": 0.0009999996896554175,
+      "loss": 4.0231,
+      "step": 271
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.6300346529640273,
+      "learning_rate": 0.0009999994482763422,
+      "loss": 4.1036,
+      "step": 272
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 8.685751962780564,
+      "learning_rate": 0.0009999991379318737,
+      "loss": 4.1403,
+      "step": 273
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.2498080971780685,
+      "learning_rate": 0.000999998758622055,
+      "loss": 4.2029,
+      "step": 274
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.8047538088429735,
+      "learning_rate": 0.0009999983103469385,
+      "loss": 4.2558,
+      "step": 275
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.937065425249134,
+      "learning_rate": 0.0009999977931065857,
+      "loss": 4.1651,
+      "step": 276
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.437196394731803,
+      "learning_rate": 0.0009999972069010686,
+      "loss": 3.9668,
+      "step": 277
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.8630403951444725,
+      "learning_rate": 0.0009999965517304673,
+      "loss": 4.2141,
+      "step": 278
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.4378229128006432,
+      "learning_rate": 0.0009999958275948725,
+      "loss": 4.1225,
+      "step": 279
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 6.68608702787061,
+      "learning_rate": 0.0009999950344943842,
+      "loss": 4.2859,
+      "step": 280
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.067793589494028,
+      "learning_rate": 0.0009999941724291115,
+      "loss": 4.1785,
+      "step": 281
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.3218467147143906,
+      "learning_rate": 0.0009999932413991737,
+      "loss": 4.1513,
+      "step": 282
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.232099383205669,
+      "learning_rate": 0.0009999922414046986,
+      "loss": 4.2677,
+      "step": 283
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.184228211822442,
+      "learning_rate": 0.0009999911724458248,
+      "loss": 4.2871,
+      "step": 284
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 19.99359159829405,
+      "learning_rate": 0.0009999900345226994,
+      "loss": 4.2853,
+      "step": 285
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.7875740824340312,
+      "learning_rate": 0.0009999888276354795,
+      "loss": 4.2122,
+      "step": 286
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 6.093503870880048,
+      "learning_rate": 0.0009999875517843315,
+      "loss": 4.1125,
+      "step": 287
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.6980520670634527,
+      "learning_rate": 0.0009999862069694312,
+      "loss": 4.2265,
+      "step": 288
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.2798750609827148,
+      "learning_rate": 0.0009999847931909645,
+      "loss": 4.2363,
+      "step": 289
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.5566968015684757,
+      "learning_rate": 0.000999983310449126,
+      "loss": 4.0132,
+      "step": 290
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 19.18546091614537,
+      "learning_rate": 0.0009999817587441203,
+      "loss": 4.061,
+      "step": 291
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.8829242069222285,
+      "learning_rate": 0.0009999801380761615,
+      "loss": 4.1262,
+      "step": 292
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.6842149025296167,
+      "learning_rate": 0.0009999784484454734,
+      "loss": 4.4524,
+      "step": 293
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.859868024080571,
+      "learning_rate": 0.0009999766898522884,
+      "loss": 4.1414,
+      "step": 294
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.8045651847204816,
+      "learning_rate": 0.0009999748622968496,
+      "loss": 4.1495,
+      "step": 295
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.124044628417353,
+      "learning_rate": 0.000999972965779409,
+      "loss": 4.1097,
+      "step": 296
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.5011576239743278,
+      "learning_rate": 0.000999971000300228,
+      "loss": 4.1823,
+      "step": 297
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.01347098774198,
+      "learning_rate": 0.000999968965859578,
+      "loss": 4.1435,
+      "step": 298
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.5913183739162184,
+      "learning_rate": 0.0009999668624577395,
+      "loss": 4.3186,
+      "step": 299
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.47856588583951,
+      "learning_rate": 0.0009999646900950023,
+      "loss": 4.3772,
+      "step": 300
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.908248562541376,
+      "learning_rate": 0.0009999624487716666,
+      "loss": 4.193,
+      "step": 301
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.9294532257161214,
+      "learning_rate": 0.000999960138488041,
+      "loss": 4.2368,
+      "step": 302
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.7995978756885962,
+      "learning_rate": 0.0009999577592444443,
+      "loss": 4.1151,
+      "step": 303
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.0729673487442666,
+      "learning_rate": 0.000999955311041205,
+      "loss": 4.0308,
+      "step": 304
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.4546240183668138,
+      "learning_rate": 0.0009999527938786606,
+      "loss": 4.3905,
+      "step": 305
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.7263021428362864,
+      "learning_rate": 0.0009999502077571581,
+      "loss": 4.0221,
+      "step": 306
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.164973600653351,
+      "learning_rate": 0.0009999475526770545,
+      "loss": 4.3111,
+      "step": 307
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.0563555618686613,
+      "learning_rate": 0.0009999448286387158,
+      "loss": 4.0792,
+      "step": 308
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.2742839474490744,
+      "learning_rate": 0.0009999420356425178,
+      "loss": 4.4126,
+      "step": 309
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.6548903017718362,
+      "learning_rate": 0.0009999391736888457,
+      "loss": 4.1291,
+      "step": 310
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.4706621032242517,
+      "learning_rate": 0.0009999362427780942,
+      "loss": 4.0174,
+      "step": 311
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.2522898936448108,
+      "learning_rate": 0.0009999332429106679,
+      "loss": 4.0531,
+      "step": 312
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.758527838442011,
+      "learning_rate": 0.00099993017408698,
+      "loss": 4.0504,
+      "step": 313
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.926369759156153,
+      "learning_rate": 0.0009999270363074547,
+      "loss": 4.1235,
+      "step": 314
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.4216792867629067,
+      "learning_rate": 0.0009999238295725237,
+      "loss": 4.1285,
+      "step": 315
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.1110654819036574,
+      "learning_rate": 0.00099992055388263,
+      "loss": 4.2889,
+      "step": 316
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.699020600438759,
+      "learning_rate": 0.0009999172092382252,
+      "loss": 4.2314,
+      "step": 317
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.779461678736526,
+      "learning_rate": 0.0009999137956397707,
+      "loss": 4.0981,
+      "step": 318
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1840521271759301,
+      "learning_rate": 0.0009999103130877373,
+      "loss": 4.2072,
+      "step": 319
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.6987405064690515,
+      "learning_rate": 0.0009999067615826054,
+      "loss": 4.1763,
+      "step": 320
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.4758379470902467,
+      "learning_rate": 0.000999903141124865,
+      "loss": 3.9832,
+      "step": 321
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 40.39742878916515,
+      "learning_rate": 0.000999899451715015,
+      "loss": 4.0352,
+      "step": 322
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.4712620565356995,
+      "learning_rate": 0.0009998956933535649,
+      "loss": 4.2948,
+      "step": 323
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.0661716179642924,
+      "learning_rate": 0.0009998918660410324,
+      "loss": 4.1999,
+      "step": 324
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.6555152768866663,
+      "learning_rate": 0.000999887969777946,
+      "loss": 4.4547,
+      "step": 325
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 14.09646828820387,
+      "learning_rate": 0.000999884004564843,
+      "loss": 4.6558,
+      "step": 326
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 5.657304255504427,
+      "learning_rate": 0.00099987997040227,
+      "loss": 4.2387,
+      "step": 327
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.3774895165908796,
+      "learning_rate": 0.0009998758672907838,
+      "loss": 4.3581,
+      "step": 328
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.1797131810201,
+      "learning_rate": 0.0009998716952309501,
+      "loss": 4.4357,
+      "step": 329
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 5.725734880635553,
+      "learning_rate": 0.0009998674542233445,
+      "loss": 4.1463,
+      "step": 330
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.3915923885287005,
+      "learning_rate": 0.000999863144268552,
+      "loss": 4.4046,
+      "step": 331
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 23.666879056714148,
+      "learning_rate": 0.000999858765367167,
+      "loss": 4.213,
+      "step": 332
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.58069587940314,
+      "learning_rate": 0.0009998543175197936,
+      "loss": 4.1942,
+      "step": 333
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.6186294395438663,
+      "learning_rate": 0.000999849800727045,
+      "loss": 4.3216,
+      "step": 334
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 5.074442349441321,
+      "learning_rate": 0.0009998452149895445,
+      "loss": 4.1695,
+      "step": 335
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.361950031079178,
+      "learning_rate": 0.0009998405603079243,
+      "loss": 4.3722,
+      "step": 336
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.778983463763034,
+      "learning_rate": 0.0009998358366828269,
+      "loss": 4.3397,
+      "step": 337
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.250987163401557,
+      "learning_rate": 0.0009998310441149034,
+      "loss": 4.3531,
+      "step": 338
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.3896684163928483,
+      "learning_rate": 0.000999826182604815,
+      "loss": 4.3145,
+      "step": 339
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.2348250676899448,
+      "learning_rate": 0.0009998212521532325,
+      "loss": 4.2225,
+      "step": 340
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.0328905315690773,
+      "learning_rate": 0.0009998162527608354,
+      "loss": 4.2112,
+      "step": 341
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 49.319649477815474,
+      "learning_rate": 0.0009998111844283137,
+      "loss": 4.4247,
+      "step": 342
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.8499403591502945,
+      "learning_rate": 0.0009998060471563665,
+      "loss": 4.4534,
+      "step": 343
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.9487614905504818,
+      "learning_rate": 0.0009998008409457023,
+      "loss": 4.4077,
+      "step": 344
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.010154781846001,
+      "learning_rate": 0.000999795565797039,
+      "loss": 4.1284,
+      "step": 345
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.087743461932623,
+      "learning_rate": 0.0009997902217111045,
+      "loss": 4.1874,
+      "step": 346
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.2264590678729124,
+      "learning_rate": 0.0009997848086886357,
+      "loss": 4.4414,
+      "step": 347
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.6353900773523025,
+      "learning_rate": 0.0009997793267303792,
+      "loss": 4.3114,
+      "step": 348
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.388842554980238,
+      "learning_rate": 0.0009997737758370914,
+      "loss": 4.5055,
+      "step": 349
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 6.371677050558135,
+      "learning_rate": 0.0009997681560095378,
+      "loss": 4.3519,
+      "step": 350
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.808691177664785,
+      "learning_rate": 0.0009997624672484933,
+      "loss": 4.2633,
+      "step": 351
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.5466615306023854,
+      "learning_rate": 0.0009997567095547432,
+      "loss": 4.335,
+      "step": 352
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 7.315933330068135,
+      "learning_rate": 0.000999750882929081,
+      "loss": 4.4478,
+      "step": 353
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.944573607809598,
+      "learning_rate": 0.0009997449873723105,
+      "loss": 4.3793,
+      "step": 354
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 6.500319948924547,
+      "learning_rate": 0.000999739022885245,
+      "loss": 4.4322,
+      "step": 355
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1455197638185235,
+      "learning_rate": 0.0009997329894687072,
+      "loss": 4.278,
+      "step": 356
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 9.105573634207436,
+      "learning_rate": 0.0009997268871235296,
+      "loss": 4.2941,
+      "step": 357
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 5.130138375210888,
+      "learning_rate": 0.0009997207158505533,
+      "loss": 4.321,
+      "step": 358
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.0490440876383507,
+      "learning_rate": 0.0009997144756506298,
+      "loss": 4.3293,
+      "step": 359
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.5517935437634631,
+      "learning_rate": 0.00099970816652462,
+      "loss": 4.1769,
+      "step": 360
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.8913769648293703,
+      "learning_rate": 0.0009997017884733938,
+      "loss": 4.1776,
+      "step": 361
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.15488805085044,
+      "learning_rate": 0.000999695341497831,
+      "loss": 3.973,
+      "step": 362
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.4561573957234442,
+      "learning_rate": 0.0009996888255988207,
+      "loss": 4.1945,
+      "step": 363
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.21447529449004,
+      "learning_rate": 0.0009996822407772623,
+      "loss": 4.1921,
+      "step": 364
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.312013471518438,
+      "learning_rate": 0.0009996755870340633,
+      "loss": 4.1811,
+      "step": 365
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1717353129345247,
+      "learning_rate": 0.0009996688643701419,
+      "loss": 4.3077,
+      "step": 366
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.5318520759135914,
+      "learning_rate": 0.0009996620727864252,
+      "loss": 4.2601,
+      "step": 367
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1480331730448254,
+      "learning_rate": 0.00099965521228385,
+      "loss": 4.0155,
+      "step": 368
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.5368186700592597,
+      "learning_rate": 0.0009996482828633624,
+      "loss": 4.4481,
+      "step": 369
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.3222058672091872,
+      "learning_rate": 0.0009996412845259183,
+      "loss": 4.2202,
+      "step": 370
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.6202185673482237,
+      "learning_rate": 0.0009996342172724833,
+      "loss": 4.2343,
+      "step": 371
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.2894933539204734,
+      "learning_rate": 0.0009996270811040318,
+      "loss": 3.9833,
+      "step": 372
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.6855540418044326,
+      "learning_rate": 0.0009996198760215483,
+      "loss": 3.9795,
+      "step": 373
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1924223767278768,
+      "learning_rate": 0.0009996126020260262,
+      "loss": 4.2868,
+      "step": 374
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.9729971906461921,
+      "learning_rate": 0.0009996052591184695,
+      "loss": 4.2176,
+      "step": 375
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.8480767564801504,
+      "learning_rate": 0.0009995978472998905,
+      "loss": 4.2889,
+      "step": 376
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0565771798226582,
+      "learning_rate": 0.0009995903665713118,
+      "loss": 4.1568,
+      "step": 377
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.360456416142257,
+      "learning_rate": 0.000999582816933765,
+      "loss": 4.0579,
+      "step": 378
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.3195693821900203,
+      "learning_rate": 0.0009995751983882914,
+      "loss": 4.1984,
+      "step": 379
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.08569068216271,
+      "learning_rate": 0.000999567510935942,
+      "loss": 4.3036,
+      "step": 380
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.3444103142098056,
+      "learning_rate": 0.0009995597545777771,
+      "loss": 4.2186,
+      "step": 381
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.5426138852906555,
+      "learning_rate": 0.0009995519293148666,
+      "loss": 4.1794,
+      "step": 382
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.496592540452236,
+      "learning_rate": 0.0009995440351482897,
+      "loss": 4.2067,
+      "step": 383
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.2735700045022074,
+      "learning_rate": 0.0009995360720791353,
+      "loss": 4.2875,
+      "step": 384
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.565558198597502,
+      "learning_rate": 0.000999528040108502,
+      "loss": 4.1659,
+      "step": 385
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.5383763017486842,
+      "learning_rate": 0.0009995199392374972,
+      "loss": 4.3205,
+      "step": 386
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0953126095411645,
+      "learning_rate": 0.0009995117694672386,
+      "loss": 4.2132,
+      "step": 387
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.9375458858753798,
+      "learning_rate": 0.000999503530798853,
+      "loss": 4.191,
+      "step": 388
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.2772614397000868,
+      "learning_rate": 0.0009994952232334766,
+      "loss": 4.271,
+      "step": 389
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.7061787526984669,
+      "learning_rate": 0.0009994868467722556,
+      "loss": 4.0973,
+      "step": 390
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.5312397789414427,
+      "learning_rate": 0.0009994784014163449,
+      "loss": 3.992,
+      "step": 391
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0706364696174504,
+      "learning_rate": 0.0009994698871669098,
+      "loss": 4.2831,
+      "step": 392
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.865631611652127,
+      "learning_rate": 0.0009994613040251246,
+      "loss": 4.2814,
+      "step": 393
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.6606265690727582,
+      "learning_rate": 0.000999452651992173,
+      "loss": 4.3857,
+      "step": 394
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.2448059068201522,
+      "learning_rate": 0.0009994439310692486,
+      "loss": 3.8996,
+      "step": 395
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.2984123516923842,
+      "learning_rate": 0.0009994351412575542,
+      "loss": 4.0641,
+      "step": 396
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.335713094373029,
+      "learning_rate": 0.000999426282558302,
+      "loss": 4.2992,
+      "step": 397
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.05883423286947,
+      "learning_rate": 0.000999417354972714,
+      "loss": 4.0104,
+      "step": 398
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.5888049575668968,
+      "learning_rate": 0.000999408358502022,
+      "loss": 4.1403,
+      "step": 399
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.4650361222185284,
+      "learning_rate": 0.0009993992931474661,
+      "loss": 4.1965,
+      "step": 400
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1101048971432796,
+      "learning_rate": 0.0009993901589102974,
+      "loss": 4.2827,
+      "step": 401
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.9167226703746987,
+      "learning_rate": 0.0009993809557917754,
+      "loss": 4.1344,
+      "step": 402
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.141850988985642,
+      "learning_rate": 0.0009993716837931696,
+      "loss": 4.2622,
+      "step": 403
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.2495698459200755,
+      "learning_rate": 0.000999362342915759,
+      "loss": 4.2708,
+      "step": 404
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1061939151991775,
+      "learning_rate": 0.0009993529331608318,
+      "loss": 4.179,
+      "step": 405
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0820094796277944,
+      "learning_rate": 0.0009993434545296862,
+      "loss": 3.9815,
+      "step": 406
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.2330433334635982,
+      "learning_rate": 0.0009993339070236292,
+      "loss": 4.053,
+      "step": 407
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.2112113674769647,
+      "learning_rate": 0.000999324290643978,
+      "loss": 3.8885,
+      "step": 408
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.2601834908232417,
+      "learning_rate": 0.0009993146053920588,
+      "loss": 3.8749,
+      "step": 409
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0771956521575363,
+      "learning_rate": 0.0009993048512692078,
+      "loss": 3.8832,
+      "step": 410
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.1380153674100617,
+      "learning_rate": 0.00099929502827677,
+      "loss": 4.1943,
+      "step": 411
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.311143180266052,
+      "learning_rate": 0.0009992851364161006,
+      "loss": 4.1327,
+      "step": 412
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.2367140965361143,
+      "learning_rate": 0.0009992751756885637,
+      "loss": 4.1799,
+      "step": 413
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.6106443097675247,
+      "learning_rate": 0.0009992651460955335,
+      "loss": 4.2873,
+      "step": 414
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.0876301309960565,
+      "learning_rate": 0.0009992550476383931,
+      "loss": 4.0479,
+      "step": 415
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.2051395887921463,
+      "learning_rate": 0.0009992448803185356,
+      "loss": 4.1903,
+      "step": 416
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.5932434720450677,
+      "learning_rate": 0.0009992346441373633,
+      "loss": 3.919,
+      "step": 417
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.4427534603119214,
+      "learning_rate": 0.0009992243390962883,
+      "loss": 4.1708,
+      "step": 418
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.385736890036496,
+      "learning_rate": 0.0009992139651967319,
+      "loss": 4.144,
+      "step": 419
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.079967188258638,
+      "learning_rate": 0.0009992035224401245,
+      "loss": 4.1111,
+      "step": 420
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.2856489197098147,
+      "learning_rate": 0.0009991930108279074,
+      "loss": 4.1762,
+      "step": 421
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.3584985734107513,
+      "learning_rate": 0.0009991824303615293,
+      "loss": 4.1227,
+      "step": 422
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1962832701569566,
+      "learning_rate": 0.0009991717810424506,
+      "loss": 4.1031,
+      "step": 423
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.2879293150575224,
+      "learning_rate": 0.0009991610628721397,
+      "loss": 4.3398,
+      "step": 424
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.9798353155821227,
+      "learning_rate": 0.000999150275852075,
+      "loss": 4.0816,
+      "step": 425
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1679414045514513,
+      "learning_rate": 0.0009991394199837444,
+      "loss": 4.1019,
+      "step": 426
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1310909720079843,
+      "learning_rate": 0.0009991284952686455,
+      "loss": 3.9163,
+      "step": 427
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.2169637793504569,
+      "learning_rate": 0.0009991175017082848,
+      "loss": 4.0028,
+      "step": 428
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1690667052571782,
+      "learning_rate": 0.0009991064393041786,
+      "loss": 4.1311,
+      "step": 429
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0864768016068007,
+      "learning_rate": 0.0009990953080578533,
+      "loss": 4.1679,
+      "step": 430
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.9532566683053302,
+      "learning_rate": 0.0009990841079708435,
+      "loss": 4.0866,
+      "step": 431
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0329163094124467,
+      "learning_rate": 0.0009990728390446946,
+      "loss": 3.8579,
+      "step": 432
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.2544594261947024,
+      "learning_rate": 0.0009990615012809608,
+      "loss": 3.9562,
+      "step": 433
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1763266600931532,
+      "learning_rate": 0.0009990500946812058,
+      "loss": 4.3207,
+      "step": 434
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.068722478985381,
+      "learning_rate": 0.000999038619247003,
+      "loss": 4.0708,
+      "step": 435
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 9.278565266188739,
+      "learning_rate": 0.0009990270749799352,
+      "loss": 4.1883,
+      "step": 436
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0817615488588326,
+      "learning_rate": 0.0009990154618815948,
+      "loss": 3.9218,
+      "step": 437
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.2426238347080267,
+      "learning_rate": 0.0009990037799535833,
+      "loss": 4.4039,
+      "step": 438
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0993328383491558,
+      "learning_rate": 0.0009989920291975124,
+      "loss": 4.1526,
+      "step": 439
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.2563924336520977,
+      "learning_rate": 0.0009989802096150029,
+      "loss": 4.2703,
+      "step": 440
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.8318851823902311,
+      "learning_rate": 0.0009989683212076848,
+      "loss": 3.9792,
+      "step": 441
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.027318592959613,
+      "learning_rate": 0.0009989563639771978,
+      "loss": 4.1156,
+      "step": 442
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.551374002379629,
+      "learning_rate": 0.0009989443379251916,
+      "loss": 4.2062,
+      "step": 443
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.663527277241027,
+      "learning_rate": 0.0009989322430533245,
+      "loss": 4.2456,
+      "step": 444
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1294039872013137,
+      "learning_rate": 0.0009989200793632652,
+      "loss": 4.0084,
+      "step": 445
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.991541396545525,
+      "learning_rate": 0.0009989078468566912,
+      "loss": 4.0946,
+      "step": 446
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.5884350569850685,
+      "learning_rate": 0.0009988955455352898,
+      "loss": 4.227,
+      "step": 447
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0780095775459435,
+      "learning_rate": 0.0009988831754007576,
+      "loss": 4.112,
+      "step": 448
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0436200832180618,
+      "learning_rate": 0.000998870736454801,
+      "loss": 3.9062,
+      "step": 449
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.2323208628469333,
+      "learning_rate": 0.0009988582286991356,
+      "loss": 4.1227,
+      "step": 450
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.092529117596904,
+      "learning_rate": 0.0009988456521354868,
+      "loss": 4.0009,
+      "step": 451
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.778822307425622,
+      "learning_rate": 0.000998833006765589,
+      "loss": 3.9905,
+      "step": 452
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0428166764829823,
+      "learning_rate": 0.0009988202925911864,
+      "loss": 4.155,
+      "step": 453
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.0777766526303507,
+      "learning_rate": 0.000998807509614033,
+      "loss": 4.1294,
+      "step": 454
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.3560606439742944,
+      "learning_rate": 0.0009987946578358918,
+      "loss": 4.192,
+      "step": 455
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1686812519653973,
+      "learning_rate": 0.0009987817372585355,
+      "loss": 3.9685,
+      "step": 456
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.199886267327284,
+      "learning_rate": 0.000998768747883746,
+      "loss": 3.9751,
+      "step": 457
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.28051981801247,
+      "learning_rate": 0.0009987556897133151,
+      "loss": 4.1387,
+      "step": 458
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.398752201967532,
+      "learning_rate": 0.0009987425627490441,
+      "loss": 4.3152,
+      "step": 459
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0358707654903616,
+      "learning_rate": 0.0009987293669927436,
+      "loss": 4.0209,
+      "step": 460
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.5618766801048318,
+      "learning_rate": 0.0009987161024462333,
+      "loss": 4.2922,
+      "step": 461
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1293077120946913,
+      "learning_rate": 0.0009987027691113432,
+      "loss": 4.1711,
+      "step": 462
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1834193535955804,
+      "learning_rate": 0.0009986893669899123,
+      "loss": 3.9076,
+      "step": 463
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0382649056081357,
+      "learning_rate": 0.0009986758960837889,
+      "loss": 4.1549,
+      "step": 464
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.2187286862291253,
+      "learning_rate": 0.0009986623563948314,
+      "loss": 4.1721,
+      "step": 465
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.8609658883974771,
+      "learning_rate": 0.000998648747924907,
+      "loss": 4.1522,
+      "step": 466
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1155638497574973,
+      "learning_rate": 0.0009986350706758934,
+      "loss": 4.1651,
+      "step": 467
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.3738184830032705,
+      "learning_rate": 0.0009986213246496762,
+      "loss": 4.2313,
+      "step": 468
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0951433298423616,
+      "learning_rate": 0.000998607509848152,
+      "loss": 4.0457,
+      "step": 469
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.066652661314097,
+      "learning_rate": 0.0009985936262732263,
+      "loss": 4.1577,
+      "step": 470
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1376056134457013,
+      "learning_rate": 0.0009985796739268138,
+      "loss": 4.1039,
+      "step": 471
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.3271850520417687,
+      "learning_rate": 0.000998565652810839,
+      "loss": 3.7839,
+      "step": 472
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.2475621503457206,
+      "learning_rate": 0.000998551562927236,
+      "loss": 4.1462,
+      "step": 473
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.6045387137904503,
+      "learning_rate": 0.000998537404277948,
+      "loss": 3.9835,
+      "step": 474
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.7006939747549183,
+      "learning_rate": 0.0009985231768649284,
+      "loss": 4.1179,
+      "step": 475
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.6887332154261994,
+      "learning_rate": 0.000998508880690139,
+      "loss": 4.1517,
+      "step": 476
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1778855649437638,
+      "learning_rate": 0.000998494515755552,
+      "loss": 4.0779,
+      "step": 477
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.446096024452147,
+      "learning_rate": 0.0009984800820631488,
+      "loss": 4.1715,
+      "step": 478
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.4455122321620089,
+      "learning_rate": 0.0009984655796149201,
+      "loss": 4.1734,
+      "step": 479
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0806047789948092,
+      "learning_rate": 0.0009984510084128661,
+      "loss": 4.0459,
+      "step": 480
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.3034820783296346,
+      "learning_rate": 0.0009984363684589972,
+      "loss": 4.0169,
+      "step": 481
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.343110208424966,
+      "learning_rate": 0.0009984216597553322,
+      "loss": 4.0863,
+      "step": 482
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0345473584785687,
+      "learning_rate": 0.0009984068823039,
+      "loss": 4.037,
+      "step": 483
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.2596976228501233,
+      "learning_rate": 0.0009983920361067388,
+      "loss": 4.0488,
+      "step": 484
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.6754874814706047,
+      "learning_rate": 0.0009983771211658965,
+      "loss": 4.0807,
+      "step": 485
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.1297953873784135,
+      "learning_rate": 0.0009983621374834303,
+      "loss": 4.0427,
+      "step": 486
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.2101768363629144,
+      "learning_rate": 0.0009983470850614068,
+      "loss": 3.9335,
+      "step": 487
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.468270571069635,
+      "learning_rate": 0.0009983319639019024,
+      "loss": 3.9995,
+      "step": 488
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.1587324164941648,
+      "learning_rate": 0.0009983167740070025,
+      "loss": 4.1912,
+      "step": 489
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.7808245295106495,
+      "learning_rate": 0.0009983015153788026,
+      "loss": 3.9911,
+      "step": 490
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.192055817645341,
+      "learning_rate": 0.000998286188019407,
+      "loss": 4.0281,
+      "step": 491
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.2574076509866403,
+      "learning_rate": 0.00099827079193093,
+      "loss": 4.0027,
+      "step": 492
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.4192671064936926,
+      "learning_rate": 0.0009982553271154953,
+      "loss": 3.9539,
+      "step": 493
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0134641904966863,
+      "learning_rate": 0.0009982397935752356,
+      "loss": 3.9714,
+      "step": 494
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 11.683648604597867,
+      "learning_rate": 0.0009982241913122937,
+      "loss": 3.9447,
+      "step": 495
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.783985761581361,
+      "learning_rate": 0.000998208520328822,
+      "loss": 4.3434,
+      "step": 496
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.2507224995234405,
+      "learning_rate": 0.0009981927806269812,
+      "loss": 3.9493,
+      "step": 497
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.7577153994572337,
+      "learning_rate": 0.0009981769722089428,
+      "loss": 4.2464,
+      "step": 498
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.4431069203273459,
+      "learning_rate": 0.0009981610950768873,
+      "loss": 3.9504,
+      "step": 499
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0900601757684483,
+      "learning_rate": 0.0009981451492330046,
+      "loss": 4.1113,
+      "step": 500
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 9.492716403446362,
+      "learning_rate": 0.000998129134679494,
+      "loss": 4.1127,
+      "step": 501
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0761124032807439,
+      "learning_rate": 0.0009981130514185646,
+      "loss": 4.0216,
+      "step": 502
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.2934283885560995,
+      "learning_rate": 0.0009980968994524344,
+      "loss": 4.0085,
+      "step": 503
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.2722402185340702,
+      "learning_rate": 0.0009980806787833316,
+      "loss": 4.2799,
+      "step": 504
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.7145797987384368,
+      "learning_rate": 0.0009980643894134935,
+      "loss": 4.1183,
+      "step": 505
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.39746703327607,
+      "learning_rate": 0.000998048031345167,
+      "loss": 3.9937,
+      "step": 506
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.879708885508142,
+      "learning_rate": 0.0009980316045806082,
+      "loss": 3.9038,
+      "step": 507
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.171093034680459,
+      "learning_rate": 0.0009980151091220826,
+      "loss": 4.0122,
+      "step": 508
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.563034483084031,
+      "learning_rate": 0.000997998544971866,
+      "loss": 4.302,
+      "step": 509
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.2457009152199578,
+      "learning_rate": 0.0009979819121322426,
+      "loss": 3.9838,
+      "step": 510
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.3956653274257755,
+      "learning_rate": 0.000997965210605507,
+      "loss": 4.1505,
+      "step": 511
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.4552619496380232,
+      "learning_rate": 0.0009979484403939626,
+      "loss": 3.8988,
+      "step": 512
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.3036964736196672,
+      "learning_rate": 0.0009979316014999226,
+      "loss": 4.0173,
+      "step": 513
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.5853377677123257,
+      "learning_rate": 0.0009979146939257098,
+      "loss": 4.041,
+      "step": 514
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.1099636757797788,
+      "learning_rate": 0.000997897717673656,
+      "loss": 4.0047,
+      "step": 515
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.1249837924978534,
+      "learning_rate": 0.0009978806727461028,
+      "loss": 3.9929,
+      "step": 516
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.1190538814048168,
+      "learning_rate": 0.000997863559145401,
+      "loss": 4.154,
+      "step": 517
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.2202539139607076,
+      "learning_rate": 0.0009978463768739118,
+      "loss": 4.1019,
+      "step": 518
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.1941719420893548,
+      "learning_rate": 0.0009978291259340045,
+      "loss": 3.8154,
+      "step": 519
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.1001779694858749,
+      "learning_rate": 0.0009978118063280587,
+      "loss": 4.1668,
+      "step": 520
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0526018043421055,
+      "learning_rate": 0.0009977944180584637,
+      "loss": 3.841,
+      "step": 521
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.1309280667773838,
+      "learning_rate": 0.0009977769611276173,
+      "loss": 3.9242,
+      "step": 522
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.209338166494763,
+      "learning_rate": 0.0009977594355379275,
+      "loss": 3.8905,
+      "step": 523
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.317492702298252,
+      "learning_rate": 0.000997741841291812,
+      "loss": 4.0329,
+      "step": 524
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.1830053375198593,
+      "learning_rate": 0.000997724178391697,
+      "loss": 4.2554,
+      "step": 525
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.116456297721812,
+      "learning_rate": 0.0009977064468400193,
+      "loss": 4.0597,
+      "step": 526
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.167414354613278,
+      "learning_rate": 0.0009976886466392244,
+      "loss": 4.0593,
+      "step": 527
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9948125581199422,
+      "learning_rate": 0.0009976707777917676,
+      "loss": 3.9882,
+      "step": 528
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.2243898779337137,
+      "learning_rate": 0.0009976528403001133,
+      "loss": 3.8375,
+      "step": 529
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.1084328206525864,
+      "learning_rate": 0.0009976348341667358,
+      "loss": 4.1369,
+      "step": 530
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0970530849024587,
+      "learning_rate": 0.0009976167593941188,
+      "loss": 4.1603,
+      "step": 531
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0875265358749018,
+      "learning_rate": 0.000997598615984755,
+      "loss": 4.1142,
+      "step": 532
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.1963104200399926,
+      "learning_rate": 0.0009975804039411475,
+      "loss": 4.1495,
+      "step": 533
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9534352857631794,
+      "learning_rate": 0.0009975621232658082,
+      "loss": 4.092,
+      "step": 534
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9980043728935059,
+      "learning_rate": 0.000997543773961258,
+      "loss": 4.0697,
+      "step": 535
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.3228467679356763,
+      "learning_rate": 0.0009975253560300283,
+      "loss": 3.8682,
+      "step": 536
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.2793617773650656,
+      "learning_rate": 0.0009975068694746596,
+      "loss": 4.1459,
+      "step": 537
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.2773022144338195,
+      "learning_rate": 0.0009974883142977015,
+      "loss": 4.3009,
+      "step": 538
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.2144853722947357,
+      "learning_rate": 0.0009974696905017135,
+      "loss": 3.9673,
+      "step": 539
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0462135449876753,
+      "learning_rate": 0.0009974509980892642,
+      "loss": 3.9808,
+      "step": 540
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.1931487705312933,
+      "learning_rate": 0.0009974322370629321,
+      "loss": 4.1655,
+      "step": 541
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.047704443564962,
+      "learning_rate": 0.000997413407425305,
+      "loss": 3.9892,
+      "step": 542
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 5.431836044959927,
+      "learning_rate": 0.0009973945091789796,
+      "loss": 4.0737,
+      "step": 543
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.3574136419934706,
+      "learning_rate": 0.000997375542326563,
+      "loss": 4.0258,
+      "step": 544
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.2484324174567876,
+      "learning_rate": 0.0009973565068706711,
+      "loss": 3.8645,
+      "step": 545
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0670779726939208,
+      "learning_rate": 0.0009973374028139296,
+      "loss": 4.0809,
+      "step": 546
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.00789163371799,
+      "learning_rate": 0.0009973182301589736,
+      "loss": 4.071,
+      "step": 547
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0338300245477874,
+      "learning_rate": 0.0009972989889084473,
+      "loss": 3.9504,
+      "step": 548
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.6691050999369597,
+      "learning_rate": 0.000997279679065005,
+      "loss": 4.0306,
+      "step": 549
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0251625987564845,
+      "learning_rate": 0.0009972603006313098,
+      "loss": 4.2332,
+      "step": 550
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.2845964351464307,
+      "learning_rate": 0.000997240853610035,
+      "loss": 4.0558,
+      "step": 551
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.4575420998231525,
+      "learning_rate": 0.0009972213380038627,
+      "loss": 3.9813,
+      "step": 552
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0498509990573848,
+      "learning_rate": 0.0009972017538154845,
+      "loss": 4.1315,
+      "step": 553
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.2435502290262965,
+      "learning_rate": 0.000997182101047602,
+      "loss": 3.9056,
+      "step": 554
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.1316660489182702,
+      "learning_rate": 0.0009971623797029258,
+      "loss": 3.9719,
+      "step": 555
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.2339845970692411,
+      "learning_rate": 0.0009971425897841765,
+      "loss": 3.8693,
+      "step": 556
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.1917792142387893,
+      "learning_rate": 0.0009971227312940826,
+      "loss": 4.1151,
+      "step": 557
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0801720642026404,
+      "learning_rate": 0.0009971028042353844,
+      "loss": 3.9647,
+      "step": 558
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.1047783844652286,
+      "learning_rate": 0.00099708280861083,
+      "loss": 4.0665,
+      "step": 559
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0425391662554808,
+      "learning_rate": 0.0009970627444231776,
+      "loss": 4.0256,
+      "step": 560
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.5294558586923575,
+      "learning_rate": 0.000997042611675194,
+      "loss": 3.7925,
+      "step": 561
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0730985523561904,
+      "learning_rate": 0.0009970224103696568,
+      "loss": 4.0216,
+      "step": 562
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0970009523396473,
+      "learning_rate": 0.0009970021405093523,
+      "loss": 4.1078,
+      "step": 563
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9911459993513541,
+      "learning_rate": 0.0009969818020970761,
+      "loss": 3.9093,
+      "step": 564
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.2011683868949352,
+      "learning_rate": 0.0009969613951356338,
+      "loss": 4.0084,
+      "step": 565
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9013135092696802,
+      "learning_rate": 0.0009969409196278398,
+      "loss": 3.9724,
+      "step": 566
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 13.208014656328935,
+      "learning_rate": 0.0009969203755765186,
+      "loss": 4.1441,
+      "step": 567
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0969621868469979,
+      "learning_rate": 0.0009968997629845038,
+      "loss": 4.076,
+      "step": 568
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.437276737778198,
+      "learning_rate": 0.0009968790818546383,
+      "loss": 3.9697,
+      "step": 569
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.2006767674664782,
+      "learning_rate": 0.000996858332189775,
+      "loss": 3.9736,
+      "step": 570
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 4.83730972230385,
+      "learning_rate": 0.0009968375139927756,
+      "loss": 3.9491,
+      "step": 571
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.4139022174024958,
+      "learning_rate": 0.000996816627266512,
+      "loss": 4.2448,
+      "step": 572
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0852405502649718,
+      "learning_rate": 0.0009967956720138647,
+      "loss": 3.9133,
+      "step": 573
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.4677284806849238,
+      "learning_rate": 0.0009967746482377243,
+      "loss": 4.0837,
+      "step": 574
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.5182628271121645,
+      "learning_rate": 0.0009967535559409905,
+      "loss": 4.0463,
+      "step": 575
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.3119732062847278,
+      "learning_rate": 0.0009967323951265725,
+      "loss": 3.9247,
+      "step": 576
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.5829729238076788,
+      "learning_rate": 0.0009967111657973892,
+      "loss": 4.0435,
+      "step": 577
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9757103539690971,
+      "learning_rate": 0.000996689867956369,
+      "loss": 4.0654,
+      "step": 578
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0122109034985007,
+      "learning_rate": 0.0009966685016064491,
+      "loss": 4.2211,
+      "step": 579
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0693864205815913,
+      "learning_rate": 0.0009966470667505767,
+      "loss": 4.1024,
+      "step": 580
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0789930424642413,
+      "learning_rate": 0.0009966255633917086,
+      "loss": 4.0756,
+      "step": 581
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.1594346922702778,
+      "learning_rate": 0.0009966039915328105,
+      "loss": 4.1535,
+      "step": 582
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.1167767671618678,
+      "learning_rate": 0.0009965823511768578,
+      "loss": 4.0344,
+      "step": 583
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.220196100134562,
+      "learning_rate": 0.0009965606423268355,
+      "loss": 4.1537,
+      "step": 584
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0584694460874193,
+      "learning_rate": 0.000996538864985738,
+      "loss": 4.1343,
+      "step": 585
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0159307461891351,
+      "learning_rate": 0.0009965170191565688,
+      "loss": 4.0014,
+      "step": 586
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.979051403666123,
+      "learning_rate": 0.0009964951048423414,
+      "loss": 4.0529,
+      "step": 587
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0737042238713586,
+      "learning_rate": 0.0009964731220460784,
+      "loss": 4.1109,
+      "step": 588
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0976324714400292,
+      "learning_rate": 0.000996451070770812,
+      "loss": 4.1789,
+      "step": 589
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.177141319134397,
+      "learning_rate": 0.0009964289510195831,
+      "loss": 4.1089,
+      "step": 590
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 5.845499650279221,
+      "learning_rate": 0.0009964067627954436,
+      "loss": 4.0542,
+      "step": 591
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0327410606400718,
+      "learning_rate": 0.0009963845061014534,
+      "loss": 4.1311,
+      "step": 592
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.8872697504861227,
+      "learning_rate": 0.0009963621809406826,
+      "loss": 4.2138,
+      "step": 593
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9715735278119713,
+      "learning_rate": 0.0009963397873162107,
+      "loss": 3.8796,
+      "step": 594
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.396484083580813,
+      "learning_rate": 0.0009963173252311257,
+      "loss": 4.3613,
+      "step": 595
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.087102400979372,
+      "learning_rate": 0.0009962947946885268,
+      "loss": 4.0034,
+      "step": 596
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.243137341425614,
+      "learning_rate": 0.000996272195691521,
+      "loss": 4.0134,
+      "step": 597
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.164944204791553,
+      "learning_rate": 0.0009962495282432255,
+      "loss": 4.0279,
+      "step": 598
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.1922008678038478,
+      "learning_rate": 0.0009962267923467672,
+      "loss": 4.0981,
+      "step": 599
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.2708796676811687,
+      "learning_rate": 0.0009962039880052817,
+      "loss": 4.2194,
+      "step": 600
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0841555708928168,
+      "learning_rate": 0.0009961811152219148,
+      "loss": 4.1596,
+      "step": 601
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.2385443973044648,
+      "learning_rate": 0.0009961581739998209,
+      "loss": 4.0312,
+      "step": 602
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.2028767241875324,
+      "learning_rate": 0.0009961351643421646,
+      "loss": 4.1067,
+      "step": 603
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.2310479343752814,
+      "learning_rate": 0.0009961120862521195,
+      "loss": 4.15,
+      "step": 604
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.5314113521550154,
+      "learning_rate": 0.000996088939732869,
+      "loss": 3.9311,
+      "step": 605
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.1454211356184008,
+      "learning_rate": 0.0009960657247876056,
+      "loss": 4.2223,
+      "step": 606
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.1993726476533393,
+      "learning_rate": 0.000996042441419531,
+      "loss": 4.0683,
+      "step": 607
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0608622751507033,
+      "learning_rate": 0.0009960190896318572,
+      "loss": 3.9369,
+      "step": 608
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0521679654159657,
+      "learning_rate": 0.0009959956694278052,
+      "loss": 3.9274,
+      "step": 609
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.1003977104744176,
+      "learning_rate": 0.000995972180810605,
+      "loss": 3.7949,
+      "step": 610
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.4170906972367623,
+      "learning_rate": 0.0009959486237834964,
+      "loss": 3.7741,
+      "step": 611
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0458262622318437,
+      "learning_rate": 0.0009959249983497289,
+      "loss": 4.153,
+      "step": 612
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.2039826291625173,
+      "learning_rate": 0.0009959013045125612,
+      "loss": 3.9622,
+      "step": 613
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.1281695800105496,
+      "learning_rate": 0.000995877542275261,
+      "loss": 3.9142,
+      "step": 614
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.2592342179278442,
+      "learning_rate": 0.0009958537116411064,
+      "loss": 4.007,
+      "step": 615
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.2370153859047168,
+      "learning_rate": 0.000995829812613384,
+      "loss": 4.1878,
+      "step": 616
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.1557899550570732,
+      "learning_rate": 0.0009958058451953902,
+      "loss": 4.0741,
+      "step": 617
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9950810165134163,
+      "learning_rate": 0.0009957818093904313,
+      "loss": 4.1322,
+      "step": 618
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0622000307870776,
+      "learning_rate": 0.000995757705201822,
+      "loss": 4.0737,
+      "step": 619
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9522213326536202,
+      "learning_rate": 0.0009957335326328874,
+      "loss": 4.1483,
+      "step": 620
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.872545612406632,
+      "learning_rate": 0.0009957092916869613,
+      "loss": 4.0376,
+      "step": 621
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.3212288739460372,
+      "learning_rate": 0.0009956849823673877,
+      "loss": 3.8471,
+      "step": 622
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.8741241742111597,
+      "learning_rate": 0.0009956606046775192,
+      "loss": 4.0802,
+      "step": 623
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9697166520873234,
+      "learning_rate": 0.0009956361586207186,
+      "loss": 4.0258,
+      "step": 624
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.5366076789215608,
+      "learning_rate": 0.0009956116442003575,
+      "loss": 4.0789,
+      "step": 625
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.1351325188396468,
+      "learning_rate": 0.0009955870614198174,
+      "loss": 3.8349,
+      "step": 626
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9904423078435894,
+      "learning_rate": 0.000995562410282489,
+      "loss": 4.1877,
+      "step": 627
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.2214548363493867,
+      "learning_rate": 0.0009955376907917722,
+      "loss": 3.9226,
+      "step": 628
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.4872237285653587,
+      "learning_rate": 0.0009955129029510768,
+      "loss": 4.1086,
+      "step": 629
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0586558181679757,
+      "learning_rate": 0.0009954880467638219,
+      "loss": 4.1311,
+      "step": 630
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.1166873306787115,
+      "learning_rate": 0.0009954631222334356,
+      "loss": 4.1628,
+      "step": 631
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.1020912519355728,
+      "learning_rate": 0.0009954381293633561,
+      "loss": 4.1827,
+      "step": 632
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.541200983213323,
+      "learning_rate": 0.0009954130681570305,
+      "loss": 4.0049,
+      "step": 633
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.4224241887794318,
+      "learning_rate": 0.0009953879386179157,
+      "loss": 4.2396,
+      "step": 634
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9968253551830029,
+      "learning_rate": 0.0009953627407494777,
+      "loss": 4.1226,
+      "step": 635
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.4185501347974874,
+      "learning_rate": 0.000995337474555192,
+      "loss": 3.8936,
+      "step": 636
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0898506044349183,
+      "learning_rate": 0.0009953121400385438,
+      "loss": 3.8433,
+      "step": 637
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.5626121980531213,
+      "learning_rate": 0.0009952867372030273,
+      "loss": 4.0309,
+      "step": 638
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9483500985414226,
+      "learning_rate": 0.0009952612660521466,
+      "loss": 4.0238,
+      "step": 639
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0794690646472656,
+      "learning_rate": 0.0009952357265894146,
+      "loss": 3.9044,
+      "step": 640
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9944819636930577,
+      "learning_rate": 0.000995210118818354,
+      "loss": 4.1289,
+      "step": 641
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9706289386802778,
+      "learning_rate": 0.0009951844427424973,
+      "loss": 4.0482,
+      "step": 642
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9929649298053834,
+      "learning_rate": 0.0009951586983653858,
+      "loss": 3.9995,
+      "step": 643
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9302516483805908,
+      "learning_rate": 0.0009951328856905703,
+      "loss": 3.9215,
+      "step": 644
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.062409809140414,
+      "learning_rate": 0.0009951070047216116,
+      "loss": 4.0465,
+      "step": 645
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0190850036087398,
+      "learning_rate": 0.000995081055462079,
+      "loss": 4.0241,
+      "step": 646
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.2025402598501818,
+      "learning_rate": 0.0009950550379155519,
+      "loss": 3.843,
+      "step": 647
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0838895111910496,
+      "learning_rate": 0.000995028952085619,
+      "loss": 3.9391,
+      "step": 648
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9156855323769093,
+      "learning_rate": 0.0009950027979758781,
+      "loss": 3.8727,
+      "step": 649
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9831840876750831,
+      "learning_rate": 0.0009949765755899369,
+      "loss": 4.0466,
+      "step": 650
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0144260482607197,
+      "learning_rate": 0.0009949502849314123,
+      "loss": 4.0569,
+      "step": 651
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.087532418616258,
+      "learning_rate": 0.0009949239260039304,
+      "loss": 4.0124,
+      "step": 652
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.1115011584707322,
+      "learning_rate": 0.0009948974988111272,
+      "loss": 3.9083,
+      "step": 653
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.1754413854281331,
+      "learning_rate": 0.0009948710033566475,
+      "loss": 3.7598,
+      "step": 654
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9073616922253441,
+      "learning_rate": 0.000994844439644146,
+      "loss": 3.8556,
+      "step": 655
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9770191942150841,
+      "learning_rate": 0.0009948178076772867,
+      "loss": 3.9657,
+      "step": 656
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.2098902115510097,
+      "learning_rate": 0.0009947911074597428,
+      "loss": 4.0797,
+      "step": 657
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.1240122584040244,
+      "learning_rate": 0.0009947643389951973,
+      "loss": 4.1461,
+      "step": 658
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.2067032998807046,
+      "learning_rate": 0.0009947375022873422,
+      "loss": 4.0355,
+      "step": 659
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9114366762460284,
+      "learning_rate": 0.0009947105973398794,
+      "loss": 3.8396,
+      "step": 660
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.1253322947399755,
+      "learning_rate": 0.0009946836241565195,
+      "loss": 4.3915,
+      "step": 661
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.021944561425325,
+      "learning_rate": 0.0009946565827409833,
+      "loss": 3.902,
+      "step": 662
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9158393543874208,
+      "learning_rate": 0.0009946294730970005,
+      "loss": 4.0877,
+      "step": 663
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.2119808024846095,
+      "learning_rate": 0.0009946022952283106,
+      "loss": 3.8803,
+      "step": 664
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.8761944842006684,
+      "learning_rate": 0.0009945750491386616,
+      "loss": 4.056,
+      "step": 665
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 19.48656722070216,
+      "learning_rate": 0.0009945477348318123,
+      "loss": 4.2876,
+      "step": 666
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0325658117466887,
+      "learning_rate": 0.00099452035231153,
+      "loss": 4.1135,
+      "step": 667
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.9440027707655119,
+      "learning_rate": 0.0009944929015815913,
+      "loss": 4.0217,
+      "step": 668
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.1902740653392259,
+      "learning_rate": 0.0009944653826457828,
+      "loss": 3.8958,
+      "step": 669
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9534347488300721,
+      "learning_rate": 0.0009944377955079004,
+      "loss": 3.937,
+      "step": 670
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.2427934301429755,
+      "learning_rate": 0.0009944101401717486,
+      "loss": 3.9359,
+      "step": 671
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.1803128820308535,
+      "learning_rate": 0.0009943824166411424,
+      "loss": 3.9248,
+      "step": 672
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.3226917163657204,
+      "learning_rate": 0.0009943546249199056,
+      "loss": 4.0167,
+      "step": 673
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.2392777021034116,
+      "learning_rate": 0.0009943267650118716,
+      "loss": 3.7435,
+      "step": 674
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.6423405371266244,
+      "learning_rate": 0.0009942988369208829,
+      "loss": 4.0427,
+      "step": 675
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.012721279386241,
+      "learning_rate": 0.000994270840650792,
+      "loss": 3.8385,
+      "step": 676
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.1603970960331422,
+      "learning_rate": 0.0009942427762054604,
+      "loss": 3.9409,
+      "step": 677
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.3868886939442144,
+      "learning_rate": 0.0009942146435887589,
+      "loss": 3.9335,
+      "step": 678
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.107754114523732,
+      "learning_rate": 0.0009941864428045677,
+      "loss": 4.2012,
+      "step": 679
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.1108239019460056,
+      "learning_rate": 0.0009941581738567768,
+      "loss": 4.1193,
+      "step": 680
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.4730349915863195,
+      "learning_rate": 0.0009941298367492854,
+      "loss": 4.1845,
+      "step": 681
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.2935560769252694,
+      "learning_rate": 0.0009941014314860021,
+      "loss": 4.1548,
+      "step": 682
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.1690456014057595,
+      "learning_rate": 0.0009940729580708448,
+      "loss": 4.022,
+      "step": 683
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.1350933927702227,
+      "learning_rate": 0.0009940444165077408,
+      "loss": 4.0981,
+      "step": 684
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0663624502353592,
+      "learning_rate": 0.0009940158068006267,
+      "loss": 4.159,
+      "step": 685
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.1967570939312862,
+      "learning_rate": 0.0009939871289534488,
+      "loss": 4.0572,
+      "step": 686
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.4308361240039056,
+      "learning_rate": 0.0009939583829701628,
+      "loss": 3.9628,
+      "step": 687
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0749087936397916,
+      "learning_rate": 0.0009939295688547337,
+      "loss": 4.0917,
+      "step": 688
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9961854543075069,
+      "learning_rate": 0.0009939006866111356,
+      "loss": 4.3358,
+      "step": 689
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0662450286553637,
+      "learning_rate": 0.0009938717362433524,
+      "loss": 4.0191,
+      "step": 690
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.203978675004886,
+      "learning_rate": 0.0009938427177553773,
+      "loss": 4.1932,
+      "step": 691
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0547630034650501,
+      "learning_rate": 0.0009938136311512127,
+      "loss": 4.0764,
+      "step": 692
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0074051321598327,
+      "learning_rate": 0.0009937844764348707,
+      "loss": 3.8708,
+      "step": 693
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.998779577972688,
+      "learning_rate": 0.0009937552536103727,
+      "loss": 3.9777,
+      "step": 694
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9559575919471303,
+      "learning_rate": 0.000993725962681749,
+      "loss": 4.0575,
+      "step": 695
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.5793672294254124,
+      "learning_rate": 0.0009936966036530402,
+      "loss": 3.7767,
+      "step": 696
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0000291986542595,
+      "learning_rate": 0.0009936671765282956,
+      "loss": 4.2494,
+      "step": 697
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9365793776478671,
+      "learning_rate": 0.0009936376813115741,
+      "loss": 3.6897,
+      "step": 698
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9531200207053346,
+      "learning_rate": 0.000993608118006944,
+      "loss": 4.0084,
+      "step": 699
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9648629400206552,
+      "learning_rate": 0.0009935784866184833,
+      "loss": 3.9867,
+      "step": 700
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.1217945452813942,
+      "learning_rate": 0.0009935487871502787,
+      "loss": 4.0233,
+      "step": 701
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 5.468742717439602,
+      "learning_rate": 0.0009935190196064267,
+      "loss": 3.8694,
+      "step": 702
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0970464348668187,
+      "learning_rate": 0.0009934891839910333,
+      "loss": 4.1408,
+      "step": 703
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0527563665166895,
+      "learning_rate": 0.0009934592803082138,
+      "loss": 3.9224,
+      "step": 704
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.1788564709369937,
+      "learning_rate": 0.0009934293085620929,
+      "loss": 4.25,
+      "step": 705
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.1177080766521004,
+      "learning_rate": 0.0009933992687568044,
+      "loss": 3.7747,
+      "step": 706
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.331533023082532,
+      "learning_rate": 0.0009933691608964917,
+      "loss": 4.0996,
+      "step": 707
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9932438632422246,
+      "learning_rate": 0.0009933389849853078,
+      "loss": 3.931,
+      "step": 708
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9473595200284471,
+      "learning_rate": 0.0009933087410274148,
+      "loss": 3.9864,
+      "step": 709
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.1266013054058719,
+      "learning_rate": 0.0009932784290269843,
+      "loss": 4.1737,
+      "step": 710
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9884842226835872,
+      "learning_rate": 0.0009932480489881974,
+      "loss": 4.0466,
+      "step": 711
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9530817273057067,
+      "learning_rate": 0.0009932176009152442,
+      "loss": 4.1339,
+      "step": 712
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.1501946911028074,
+      "learning_rate": 0.0009931870848123245,
+      "loss": 4.1296,
+      "step": 713
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.4350472895321407,
+      "learning_rate": 0.0009931565006836476,
+      "loss": 3.9715,
+      "step": 714
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.18946767117426,
+      "learning_rate": 0.0009931258485334315,
+      "loss": 4.0962,
+      "step": 715
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.1231827461701815,
+      "learning_rate": 0.0009930951283659048,
+      "loss": 3.7519,
+      "step": 716
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9354466592494857,
+      "learning_rate": 0.0009930643401853043,
+      "loss": 3.8916,
+      "step": 717
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9972892020163304,
+      "learning_rate": 0.0009930334839958765,
+      "loss": 3.9692,
+      "step": 718
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9218413712056626,
+      "learning_rate": 0.000993002559801878,
+      "loss": 3.898,
+      "step": 719
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0490325173709558,
+      "learning_rate": 0.0009929715676075736,
+      "loss": 3.8237,
+      "step": 720
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.049671714073017,
+      "learning_rate": 0.0009929405074172383,
+      "loss": 4.3424,
+      "step": 721
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.081740112093551,
+      "learning_rate": 0.0009929093792351567,
+      "loss": 3.8379,
+      "step": 722
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9795891942566914,
+      "learning_rate": 0.0009928781830656215,
+      "loss": 4.1246,
+      "step": 723
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.3036431583133856,
+      "learning_rate": 0.0009928469189129363,
+      "loss": 4.11,
+      "step": 724
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.4459492283264697,
+      "learning_rate": 0.0009928155867814131,
+      "loss": 4.0217,
+      "step": 725
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.1151386796548914,
+      "learning_rate": 0.0009927841866753735,
+      "loss": 4.0514,
+      "step": 726
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0899364326406684,
+      "learning_rate": 0.000992752718599149,
+      "loss": 4.2129,
+      "step": 727
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.065228216471037,
+      "learning_rate": 0.0009927211825570793,
+      "loss": 3.9278,
+      "step": 728
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0817986510380675,
+      "learning_rate": 0.000992689578553515,
+      "loss": 3.9406,
+      "step": 729
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0862220484169132,
+      "learning_rate": 0.0009926579065928144,
+      "loss": 3.846,
+      "step": 730
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0620328173436737,
+      "learning_rate": 0.000992626166679347,
+      "loss": 4.0841,
+      "step": 731
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0124476830654119,
+      "learning_rate": 0.0009925943588174897,
+      "loss": 3.8932,
+      "step": 732
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.890905917969885,
+      "learning_rate": 0.0009925624830116305,
+      "loss": 4.2086,
+      "step": 733
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.1394942261731105,
+      "learning_rate": 0.000992530539266166,
+      "loss": 4.0573,
+      "step": 734
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 17.58134690738853,
+      "learning_rate": 0.0009924985275855018,
+      "loss": 4.1003,
+      "step": 735
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9102741055274967,
+      "learning_rate": 0.000992466447974054,
+      "loss": 3.7821,
+      "step": 736
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9754834362355617,
+      "learning_rate": 0.0009924343004362466,
+      "loss": 3.8204,
+      "step": 737
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.187679547843071,
+      "learning_rate": 0.0009924020849765142,
+      "loss": 3.8235,
+      "step": 738
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.2989011173430143,
+      "learning_rate": 0.0009923698015993003,
+      "loss": 3.6622,
+      "step": 739
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0182550696581316,
+      "learning_rate": 0.0009923374503090577,
+      "loss": 4.1227,
+      "step": 740
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.398123225700433,
+      "learning_rate": 0.0009923050311102487,
+      "loss": 3.9437,
+      "step": 741
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.9306596644099066,
+      "learning_rate": 0.0009922725440073446,
+      "loss": 4.097,
+      "step": 742
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9930199840085645,
+      "learning_rate": 0.0009922399890048268,
+      "loss": 3.7801,
+      "step": 743
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.0933343536839801,
+      "learning_rate": 0.0009922073661071855,
+      "loss": 3.9452,
+      "step": 744
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.0066584553571114,
+      "learning_rate": 0.0009921746753189203,
+      "loss": 4.0686,
+      "step": 745
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.1074190292369288,
+      "learning_rate": 0.0009921419166445404,
+      "loss": 3.9655,
+      "step": 746
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9432259455621148,
+      "learning_rate": 0.0009921090900885641,
+      "loss": 4.0372,
+      "step": 747
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.017674450039497,
+      "learning_rate": 0.0009920761956555193,
+      "loss": 3.9217,
+      "step": 748
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.0280470835173998,
+      "learning_rate": 0.0009920432333499433,
+      "loss": 4.0152,
+      "step": 749
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9823285755893133,
+      "learning_rate": 0.0009920102031763822,
+      "loss": 4.0291,
+      "step": 750
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9669013726727463,
+      "learning_rate": 0.0009919771051393922,
+      "loss": 3.8503,
+      "step": 751
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.537456166035882,
+      "learning_rate": 0.0009919439392435385,
+      "loss": 3.769,
+      "step": 752
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.8548767732033391,
+      "learning_rate": 0.0009919107054933956,
+      "loss": 4.0237,
+      "step": 753
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.8811001801027458,
+      "learning_rate": 0.0009918774038935477,
+      "loss": 4.1467,
+      "step": 754
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.2427744326603332,
+      "learning_rate": 0.000991844034448588,
+      "loss": 3.8224,
+      "step": 755
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9975893449290646,
+      "learning_rate": 0.000991810597163119,
+      "loss": 3.8876,
+      "step": 756
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.1122379622875842,
+      "learning_rate": 0.000991777092041753,
+      "loss": 3.7849,
+      "step": 757
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.8907202185325103,
+      "learning_rate": 0.0009917435190891111,
+      "loss": 3.9553,
+      "step": 758
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.6546105708760155,
+      "learning_rate": 0.0009917098783098243,
+      "loss": 3.9838,
+      "step": 759
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.0196671000600643,
+      "learning_rate": 0.0009916761697085327,
+      "loss": 4.1361,
+      "step": 760
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.3762598045752321,
+      "learning_rate": 0.0009916423932898857,
+      "loss": 3.9431,
+      "step": 761
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.2241709652745174,
+      "learning_rate": 0.0009916085490585423,
+      "loss": 4.0906,
+      "step": 762
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.5344877136963833,
+      "learning_rate": 0.0009915746370191701,
+      "loss": 3.9918,
+      "step": 763
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.8689955149528441,
+      "learning_rate": 0.0009915406571764471,
+      "loss": 4.0184,
+      "step": 764
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.2959543445298283,
+      "learning_rate": 0.0009915066095350603,
+      "loss": 4.1736,
+      "step": 765
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.1204533833763546,
+      "learning_rate": 0.0009914724940997053,
+      "loss": 4.0835,
+      "step": 766
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9640682617579192,
+      "learning_rate": 0.0009914383108750883,
+      "loss": 3.8902,
+      "step": 767
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.166510744809395,
+      "learning_rate": 0.000991404059865924,
+      "loss": 4.0781,
+      "step": 768
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.1577045215436794,
+      "learning_rate": 0.0009913697410769366,
+      "loss": 3.883,
+      "step": 769
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.122449767602365,
+      "learning_rate": 0.0009913353545128597,
+      "loss": 3.9835,
+      "step": 770
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.1571686982537643,
+      "learning_rate": 0.0009913009001784364,
+      "loss": 4.1395,
+      "step": 771
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.040446710303204,
+      "learning_rate": 0.0009912663780784188,
+      "loss": 4.0916,
+      "step": 772
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.7034025429253001,
+      "learning_rate": 0.000991231788217569,
+      "loss": 3.8143,
+      "step": 773
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.9180330986454783,
+      "learning_rate": 0.0009911971306006575,
+      "loss": 4.0155,
+      "step": 774
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 6.111673262997343,
+      "learning_rate": 0.000991162405232465,
+      "loss": 3.9434,
+      "step": 775
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.374875056672268,
+      "learning_rate": 0.0009911276121177812,
+      "loss": 4.2326,
+      "step": 776
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.0396137237423868,
+      "learning_rate": 0.0009910927512614051,
+      "loss": 4.0597,
+      "step": 777
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.0137671887451662,
+      "learning_rate": 0.000991057822668145,
+      "loss": 3.9569,
+      "step": 778
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.1563368676764412,
+      "learning_rate": 0.0009910228263428186,
+      "loss": 3.9306,
+      "step": 779
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.0598592427689284,
+      "learning_rate": 0.000990987762290253,
+      "loss": 3.8386,
+      "step": 780
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.577903073121024,
+      "learning_rate": 0.0009909526305152848,
+      "loss": 4.1691,
+      "step": 781
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.9241809607397247,
+      "learning_rate": 0.0009909174310227596,
+      "loss": 3.8345,
+      "step": 782
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9315987122865419,
+      "learning_rate": 0.0009908821638175325,
+      "loss": 3.9253,
+      "step": 783
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.0251002931386355,
+      "learning_rate": 0.000990846828904468,
+      "loss": 4.1566,
+      "step": 784
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.0199818235234839,
+      "learning_rate": 0.0009908114262884397,
+      "loss": 4.1043,
+      "step": 785
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.1156558138385453,
+      "learning_rate": 0.0009907759559743311,
+      "loss": 3.9011,
+      "step": 786
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.3564312470349347,
+      "learning_rate": 0.0009907404179670342,
+      "loss": 3.9485,
+      "step": 787
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9091196002896215,
+      "learning_rate": 0.000990704812271451,
+      "loss": 4.1194,
+      "step": 788
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9502055692126391,
+      "learning_rate": 0.0009906691388924928,
+      "loss": 4.2732,
+      "step": 789
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.035813467791336,
+      "learning_rate": 0.0009906333978350799,
+      "loss": 3.8516,
+      "step": 790
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9775579581560876,
+      "learning_rate": 0.000990597589104142,
+      "loss": 4.0115,
+      "step": 791
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9738646568445928,
+      "learning_rate": 0.0009905617127046182,
+      "loss": 3.8153,
+      "step": 792
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9013381831949332,
+      "learning_rate": 0.0009905257686414573,
+      "loss": 3.9992,
+      "step": 793
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.015169838689599,
+      "learning_rate": 0.0009904897569196168,
+      "loss": 4.1797,
+      "step": 794
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9274410229356569,
+      "learning_rate": 0.0009904536775440641,
+      "loss": 3.8443,
+      "step": 795
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.944539838615447,
+      "learning_rate": 0.0009904175305197752,
+      "loss": 3.9231,
+      "step": 796
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.8035238524617749,
+      "learning_rate": 0.0009903813158517363,
+      "loss": 3.8491,
+      "step": 797
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.05305129000883,
+      "learning_rate": 0.0009903450335449423,
+      "loss": 4.0726,
+      "step": 798
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.1211758829609977,
+      "learning_rate": 0.0009903086836043978,
+      "loss": 4.1236,
+      "step": 799
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.8295000288987961,
+      "learning_rate": 0.0009902722660351166,
+      "loss": 3.949,
+      "step": 800
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9436444990787903,
+      "learning_rate": 0.0009902357808421218,
+      "loss": 3.79,
+      "step": 801
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.032163990231982,
+      "learning_rate": 0.0009901992280304456,
+      "loss": 4.0251,
+      "step": 802
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9752542368413061,
+      "learning_rate": 0.00099016260760513,
+      "loss": 3.9477,
+      "step": 803
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.0107518548941548,
+      "learning_rate": 0.000990125919571226,
+      "loss": 3.9448,
+      "step": 804
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.1096800317514424,
+      "learning_rate": 0.000990089163933794,
+      "loss": 3.8131,
+      "step": 805
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.6176905104762076,
+      "learning_rate": 0.000990052340697904,
+      "loss": 4.0832,
+      "step": 806
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.134213600726209,
+      "learning_rate": 0.0009900154498686349,
+      "loss": 4.2693,
+      "step": 807
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.850471581902836,
+      "learning_rate": 0.0009899784914510748,
+      "loss": 3.5897,
+      "step": 808
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.285536995743178,
+      "learning_rate": 0.0009899414654503216,
+      "loss": 4.0772,
+      "step": 809
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9542309977895963,
+      "learning_rate": 0.0009899043718714826,
+      "loss": 3.7273,
+      "step": 810
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.0242560575682236,
+      "learning_rate": 0.0009898672107196739,
+      "loss": 4.1553,
+      "step": 811
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.0283671739658398,
+      "learning_rate": 0.000989829982000021,
+      "loss": 4.0134,
+      "step": 812
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9105800453033387,
+      "learning_rate": 0.000989792685717659,
+      "loss": 3.6721,
+      "step": 813
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.0205731976565942,
+      "learning_rate": 0.0009897553218777327,
+      "loss": 4.1639,
+      "step": 814
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9840457866716322,
+      "learning_rate": 0.000989717890485395,
+      "loss": 3.9441,
+      "step": 815
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9845232997449745,
+      "learning_rate": 0.0009896803915458094,
+      "loss": 4.0387,
+      "step": 816
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.0526905024878224,
+      "learning_rate": 0.0009896428250641479,
+      "loss": 3.8775,
+      "step": 817
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9491552905833132,
+      "learning_rate": 0.000989605191045592,
+      "loss": 3.9813,
+      "step": 818
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9222606919245979,
+      "learning_rate": 0.0009895674894953327,
+      "loss": 3.8994,
+      "step": 819
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.8609930429913497,
+      "learning_rate": 0.0009895297204185706,
+      "loss": 3.946,
+      "step": 820
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9271486263768111,
+      "learning_rate": 0.0009894918838205145,
+      "loss": 4.0584,
+      "step": 821
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.1179652988962563,
+      "learning_rate": 0.0009894539797063837,
+      "loss": 3.9282,
+      "step": 822
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.0673829988589105,
+      "learning_rate": 0.0009894160080814061,
+      "loss": 4.0913,
+      "step": 823
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.1792253913818123,
+      "learning_rate": 0.0009893779689508194,
+      "loss": 4.0481,
+      "step": 824
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.557280297610356,
+      "learning_rate": 0.0009893398623198703,
+      "loss": 3.9037,
+      "step": 825
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.0303765353183911,
+      "learning_rate": 0.0009893016881938148,
+      "loss": 4.2995,
+      "step": 826
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.0662828684673717,
+      "learning_rate": 0.0009892634465779185,
+      "loss": 3.8875,
+      "step": 827
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9013754159793762,
+      "learning_rate": 0.000989225137477456,
+      "loss": 3.9274,
+      "step": 828
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.0670664254850324,
+      "learning_rate": 0.000989186760897711,
+      "loss": 3.9905,
+      "step": 829
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.4531558653745784,
+      "learning_rate": 0.0009891483168439773,
+      "loss": 4.0812,
+      "step": 830
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.0121252031629329,
+      "learning_rate": 0.000989109805321557,
+      "loss": 4.0183,
+      "step": 831
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.0401444028810551,
+      "learning_rate": 0.0009890712263357626,
+      "loss": 4.0018,
+      "step": 832
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9562074112219895,
+      "learning_rate": 0.000989032579891915,
+      "loss": 4.0212,
+      "step": 833
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.0790867286609265,
+      "learning_rate": 0.000988993865995345,
+      "loss": 4.0656,
+      "step": 834
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9600375726341359,
+      "learning_rate": 0.000988955084651392,
+      "loss": 4.0144,
+      "step": 835
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.0000204253716836,
+      "learning_rate": 0.0009889162358654056,
+      "loss": 3.9309,
+      "step": 836
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.861165516547268,
+      "learning_rate": 0.000988877319642744,
+      "loss": 4.0203,
+      "step": 837
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.016795584224561,
+      "learning_rate": 0.000988838335988775,
+      "loss": 3.9911,
+      "step": 838
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.3210403599451914,
+      "learning_rate": 0.0009887992849088754,
+      "loss": 4.1468,
+      "step": 839
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9810663948603184,
+      "learning_rate": 0.000988760166408432,
+      "loss": 4.037,
+      "step": 840
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9140102356291326,
+      "learning_rate": 0.0009887209804928404,
+      "loss": 3.9914,
+      "step": 841
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.5225827808760473,
+      "learning_rate": 0.0009886817271675052,
+      "loss": 3.8276,
+      "step": 842
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.86655829226022,
+      "learning_rate": 0.000988642406437841,
+      "loss": 4.0931,
+      "step": 843
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.8915003777704951,
+      "learning_rate": 0.0009886030183092712,
+      "loss": 4.063,
+      "step": 844
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.001193182588885,
+      "learning_rate": 0.0009885635627872285,
+      "loss": 4.1955,
+      "step": 845
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.8749579509123556,
+      "learning_rate": 0.0009885240398771554,
+      "loss": 3.8727,
+      "step": 846
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9174336282191572,
+      "learning_rate": 0.0009884844495845029,
+      "loss": 3.7108,
+      "step": 847
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9767350147353064,
+      "learning_rate": 0.000988444791914732,
+      "loss": 3.8084,
+      "step": 848
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9490028396447484,
+      "learning_rate": 0.0009884050668733126,
+      "loss": 3.7333,
+      "step": 849
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9128354970329658,
+      "learning_rate": 0.0009883652744657244,
+      "loss": 3.8161,
+      "step": 850
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.4251674169878141,
+      "learning_rate": 0.0009883254146974554,
+      "loss": 3.869,
+      "step": 851
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.2676229440569307,
+      "learning_rate": 0.0009882854875740037,
+      "loss": 3.8149,
+      "step": 852
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.0518737930407933,
+      "learning_rate": 0.0009882454931008768,
+      "loss": 3.945,
+      "step": 853
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.8419933816407473,
+      "learning_rate": 0.0009882054312835907,
+      "loss": 3.9068,
+      "step": 854
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.0270560328038532,
+      "learning_rate": 0.0009881653021276715,
+      "loss": 4.0392,
+      "step": 855
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9956358241626961,
+      "learning_rate": 0.0009881251056386541,
+      "loss": 3.9998,
+      "step": 856
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.0648682651479946,
+      "learning_rate": 0.000988084841822083,
+      "loss": 3.9471,
+      "step": 857
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.212187276856189,
+      "learning_rate": 0.0009880445106835117,
+      "loss": 3.9788,
+      "step": 858
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.8813828334403914,
+      "learning_rate": 0.000988004112228503,
+      "loss": 3.8079,
+      "step": 859
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9946388133677568,
+      "learning_rate": 0.0009879636464626294,
+      "loss": 4.0274,
+      "step": 860
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9672115346545923,
+      "learning_rate": 0.0009879231133914721,
+      "loss": 3.8767,
+      "step": 861
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.0569634420713294,
+      "learning_rate": 0.000987882513020622,
+      "loss": 4.0523,
+      "step": 862
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.211133236457914,
+      "learning_rate": 0.000987841845355679,
+      "loss": 3.9364,
+      "step": 863
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.934838587017001,
+      "learning_rate": 0.0009878011104022526,
+      "loss": 4.1468,
+      "step": 864
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.132954560078843,
+      "learning_rate": 0.0009877603081659614,
+      "loss": 4.1508,
+      "step": 865
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.140751159603247,
+      "learning_rate": 0.0009877194386524334,
+      "loss": 3.9636,
+      "step": 866
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.0031415043052765,
+      "learning_rate": 0.0009876785018673054,
+      "loss": 3.7413,
+      "step": 867
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.3087147606184937,
+      "learning_rate": 0.0009876374978162242,
+      "loss": 3.9689,
+      "step": 868
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.969628456875773,
+      "learning_rate": 0.0009875964265048452,
+      "loss": 4.0432,
+      "step": 869
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.0141809866142528,
+      "learning_rate": 0.0009875552879388336,
+      "loss": 3.78,
+      "step": 870
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.853587675727436,
+      "learning_rate": 0.000987514082123864,
+      "loss": 4.0411,
+      "step": 871
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9371715538706589,
+      "learning_rate": 0.0009874728090656193,
+      "loss": 3.8285,
+      "step": 872
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.983714183927287,
+      "learning_rate": 0.0009874314687697927,
+      "loss": 4.0886,
+      "step": 873
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.8491315684346135,
+      "learning_rate": 0.0009873900612420866,
+      "loss": 3.7661,
+      "step": 874
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.303370392847567,
+      "learning_rate": 0.0009873485864882116,
+      "loss": 4.0109,
+      "step": 875
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9843773132718185,
+      "learning_rate": 0.000987307044513889,
+      "loss": 3.9235,
+      "step": 876
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.2089768001883556,
+      "learning_rate": 0.0009872654353248486,
+      "loss": 3.9614,
+      "step": 877
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9459274126677216,
+      "learning_rate": 0.0009872237589268295,
+      "loss": 3.9068,
+      "step": 878
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9471254680257389,
+      "learning_rate": 0.00098718201532558,
+      "loss": 3.9324,
+      "step": 879
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.11106568823311,
+      "learning_rate": 0.0009871402045268582,
+      "loss": 3.9746,
+      "step": 880
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.1287428425278907,
+      "learning_rate": 0.000987098326536431,
+      "loss": 3.9029,
+      "step": 881
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9783137258621587,
+      "learning_rate": 0.0009870563813600744,
+      "loss": 3.8805,
+      "step": 882
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9780148690051447,
+      "learning_rate": 0.0009870143690035743,
+      "loss": 4.0786,
+      "step": 883
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.068655865522098,
+      "learning_rate": 0.0009869722894727251,
+      "loss": 3.8926,
+      "step": 884
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.048614788764977,
+      "learning_rate": 0.0009869301427733314,
+      "loss": 3.87,
+      "step": 885
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.4846155192388608,
+      "learning_rate": 0.000986887928911206,
+      "loss": 4.071,
+      "step": 886
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.3897928309650462,
+      "learning_rate": 0.0009868456478921719,
+      "loss": 3.9234,
+      "step": 887
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.5798116263426913,
+      "learning_rate": 0.0009868032997220608,
+      "loss": 4.013,
+      "step": 888
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9463328064283754,
+      "learning_rate": 0.0009867608844067136,
+      "loss": 3.9455,
+      "step": 889
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.5295475704882031,
+      "learning_rate": 0.000986718401951981,
+      "loss": 3.954,
+      "step": 890
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9136178718754765,
+      "learning_rate": 0.0009866758523637228,
+      "loss": 3.9869,
+      "step": 891
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.0605046744467053,
+      "learning_rate": 0.0009866332356478075,
+      "loss": 3.8271,
+      "step": 892
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.1290090070732615,
+      "learning_rate": 0.000986590551810113,
+      "loss": 3.8631,
+      "step": 893
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.1679820681090167,
+      "learning_rate": 0.0009865478008565275,
+      "loss": 4.0724,
+      "step": 894
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.0264440385505964,
+      "learning_rate": 0.0009865049827929475,
+      "loss": 3.9268,
+      "step": 895
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.1053938338315987,
+      "learning_rate": 0.0009864620976252785,
+      "loss": 4.199,
+      "step": 896
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9976409799636035,
+      "learning_rate": 0.000986419145359436,
+      "loss": 4.0217,
+      "step": 897
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.0350905792400589,
+      "learning_rate": 0.0009863761260013443,
+      "loss": 4.0266,
+      "step": 898
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9475569607656157,
+      "learning_rate": 0.0009863330395569374,
+      "loss": 3.9545,
+      "step": 899
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9279763736236232,
+      "learning_rate": 0.000986289886032158,
+      "loss": 3.9865,
+      "step": 900
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9382359932198319,
+      "learning_rate": 0.0009862466654329582,
+      "loss": 4.0322,
+      "step": 901
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.009630499341495,
+      "learning_rate": 0.0009862033777652997,
+      "loss": 4.0391,
+      "step": 902
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.1326568399400498,
+      "learning_rate": 0.000986160023035153,
+      "loss": 3.9343,
+      "step": 903
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.0673938861130499,
+      "learning_rate": 0.0009861166012484982,
+      "loss": 3.8234,
+      "step": 904
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9578065327885132,
+      "learning_rate": 0.0009860731124113247,
+      "loss": 3.7129,
+      "step": 905
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.446209034150686,
+      "learning_rate": 0.0009860295565296306,
+      "loss": 4.0171,
+      "step": 906
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9279686496506361,
+      "learning_rate": 0.000985985933609424,
+      "loss": 3.7854,
+      "step": 907
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.0134078190994902,
+      "learning_rate": 0.0009859422436567212,
+      "loss": 3.9741,
+      "step": 908
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.0887634824285362,
+      "learning_rate": 0.000985898486677549,
+      "loss": 3.9212,
+      "step": 909
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.0777712494328513,
+      "learning_rate": 0.0009858546626779425,
+      "loss": 3.9345,
+      "step": 910
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 11.825139921093331,
+      "learning_rate": 0.0009858107716639464,
+      "loss": 4.0774,
+      "step": 911
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9989047995884904,
+      "learning_rate": 0.000985766813641615,
+      "loss": 3.9242,
+      "step": 912
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.049722250373986,
+      "learning_rate": 0.0009857227886170112,
+      "loss": 3.8436,
+      "step": 913
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.5195700925841094,
+      "learning_rate": 0.0009856786965962074,
+      "loss": 3.83,
+      "step": 914
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.126797217352382,
+      "learning_rate": 0.0009856345375852853,
+      "loss": 4.0606,
+      "step": 915
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.0172499410184732,
+      "learning_rate": 0.0009855903115903357,
+      "loss": 4.0294,
+      "step": 916
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9889958979682758,
+      "learning_rate": 0.0009855460186174588,
+      "loss": 4.099,
+      "step": 917
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.138362331823372,
+      "learning_rate": 0.000985501658672764,
+      "loss": 3.8066,
+      "step": 918
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9452156896068032,
+      "learning_rate": 0.0009854572317623698,
+      "loss": 3.8519,
+      "step": 919
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9880588692204164,
+      "learning_rate": 0.0009854127378924043,
+      "loss": 4.004,
+      "step": 920
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.4311088715173952,
+      "learning_rate": 0.0009853681770690043,
+      "loss": 3.9666,
+      "step": 921
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.1277321117183687,
+      "learning_rate": 0.0009853235492983164,
+      "loss": 3.8508,
+      "step": 922
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.2975300754835295,
+      "learning_rate": 0.000985278854586496,
+      "loss": 3.7989,
+      "step": 923
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.173906149678658,
+      "learning_rate": 0.0009852340929397076,
+      "loss": 3.8196,
+      "step": 924
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9721975277088422,
+      "learning_rate": 0.0009851892643641257,
+      "loss": 3.8182,
+      "step": 925
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.088509826807722,
+      "learning_rate": 0.000985144368865933,
+      "loss": 4.011,
+      "step": 926
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9749600605983302,
+      "learning_rate": 0.0009850994064513226,
+      "loss": 4.0598,
+      "step": 927
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9711442662062042,
+      "learning_rate": 0.000985054377126496,
+      "loss": 3.8963,
+      "step": 928
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9947141307376325,
+      "learning_rate": 0.0009850092808976639,
+      "loss": 3.9325,
+      "step": 929
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.927100842140584,
+      "learning_rate": 0.0009849641177710467,
+      "loss": 4.2408,
+      "step": 930
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.0760633188767201,
+      "learning_rate": 0.0009849188877528736,
+      "loss": 4.2261,
+      "step": 931
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.0243260241102548,
+      "learning_rate": 0.0009848735908493834,
+      "loss": 4.01,
+      "step": 932
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.2803498807092963,
+      "learning_rate": 0.0009848282270668238,
+      "loss": 4.0896,
+      "step": 933
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9634594293415681,
+      "learning_rate": 0.000984782796411452,
+      "loss": 3.7973,
+      "step": 934
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.0590276488895651,
+      "learning_rate": 0.0009847372988895343,
+      "loss": 3.8941,
+      "step": 935
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9590614112339397,
+      "learning_rate": 0.000984691734507346,
+      "loss": 3.9082,
+      "step": 936
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9621083171471031,
+      "learning_rate": 0.0009846461032711723,
+      "loss": 3.6889,
+      "step": 937
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.1535105086169881,
+      "learning_rate": 0.0009846004051873066,
+      "loss": 3.9091,
+      "step": 938
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9910555244318018,
+      "learning_rate": 0.0009845546402620523,
+      "loss": 3.9293,
+      "step": 939
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.1437731125629356,
+      "learning_rate": 0.0009845088085017218,
+      "loss": 4.0819,
+      "step": 940
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.0169184615801394,
+      "learning_rate": 0.000984462909912637,
+      "loss": 4.0344,
+      "step": 941
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.2667335475860932,
+      "learning_rate": 0.0009844169445011282,
+      "loss": 4.0599,
+      "step": 942
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.96515354446607,
+      "learning_rate": 0.0009843709122735358,
+      "loss": 3.95,
+      "step": 943
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.0180271021484424,
+      "learning_rate": 0.000984324813236209,
+      "loss": 3.8148,
+      "step": 944
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.124646891645496,
+      "learning_rate": 0.0009842786473955062,
+      "loss": 3.9023,
+      "step": 945
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.1022123103648478,
+      "learning_rate": 0.0009842324147577954,
+      "loss": 3.7699,
+      "step": 946
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8641990303714225,
+      "learning_rate": 0.0009841861153294534,
+      "loss": 4.0165,
+      "step": 947
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.1137558023121101,
+      "learning_rate": 0.000984139749116866,
+      "loss": 3.8662,
+      "step": 948
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.0550813438030957,
+      "learning_rate": 0.0009840933161264288,
+      "loss": 4.0374,
+      "step": 949
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.2801259017187434,
+      "learning_rate": 0.0009840468163645462,
+      "loss": 3.968,
+      "step": 950
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.0581138918386634,
+      "learning_rate": 0.0009840002498376322,
+      "loss": 4.2772,
+      "step": 951
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.468404368530255,
+      "learning_rate": 0.0009839536165521094,
+      "loss": 3.8075,
+      "step": 952
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.0579489195975014,
+      "learning_rate": 0.0009839069165144103,
+      "loss": 4.0129,
+      "step": 953
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.1827342868024504,
+      "learning_rate": 0.0009838601497309763,
+      "loss": 3.8264,
+      "step": 954
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9053441946748603,
+      "learning_rate": 0.0009838133162082578,
+      "loss": 3.8842,
+      "step": 955
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9006873716267609,
+      "learning_rate": 0.0009837664159527146,
+      "loss": 3.9033,
+      "step": 956
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.0432290926220478,
+      "learning_rate": 0.0009837194489708157,
+      "loss": 4.0429,
+      "step": 957
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.594864037385885,
+      "learning_rate": 0.0009836724152690395,
+      "loss": 3.9817,
+      "step": 958
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.0057062435935376,
+      "learning_rate": 0.0009836253148538731,
+      "loss": 3.9606,
+      "step": 959
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.0305832027166706,
+      "learning_rate": 0.0009835781477318133,
+      "loss": 4.0169,
+      "step": 960
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.0705518913802718,
+      "learning_rate": 0.000983530913909366,
+      "loss": 4.1068,
+      "step": 961
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9899523926793627,
+      "learning_rate": 0.0009834836133930458,
+      "loss": 4.0017,
+      "step": 962
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.1864832209769205,
+      "learning_rate": 0.0009834362461893773,
+      "loss": 3.9282,
+      "step": 963
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.1234925733797614,
+      "learning_rate": 0.0009833888123048937,
+      "loss": 3.9176,
+      "step": 964
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.685584816776114,
+      "learning_rate": 0.0009833413117461378,
+      "loss": 3.8626,
+      "step": 965
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.0112552710642482,
+      "learning_rate": 0.0009832937445196613,
+      "loss": 3.7683,
+      "step": 966
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9565318131956579,
+      "learning_rate": 0.000983246110632025,
+      "loss": 4.139,
+      "step": 967
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.3075776421652565,
+      "learning_rate": 0.0009831984100897994,
+      "loss": 3.9563,
+      "step": 968
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8845577896104776,
+      "learning_rate": 0.0009831506428995636,
+      "loss": 4.0439,
+      "step": 969
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.894457684118974,
+      "learning_rate": 0.0009831028090679064,
+      "loss": 3.8352,
+      "step": 970
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.1072904092835272,
+      "learning_rate": 0.0009830549086014254,
+      "loss": 3.9061,
+      "step": 971
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.0069951502626817,
+      "learning_rate": 0.0009830069415067276,
+      "loss": 3.7084,
+      "step": 972
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9255986894818868,
+      "learning_rate": 0.0009829589077904293,
+      "loss": 3.8614,
+      "step": 973
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.6985012498781358,
+      "learning_rate": 0.0009829108074591556,
+      "loss": 3.7893,
+      "step": 974
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.2817526819866831,
+      "learning_rate": 0.0009828626405195412,
+      "loss": 3.571,
+      "step": 975
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.89370450580645,
+      "learning_rate": 0.0009828144069782296,
+      "loss": 4.0162,
+      "step": 976
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8702613444039932,
+      "learning_rate": 0.0009827661068418738,
+      "loss": 3.9715,
+      "step": 977
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8421932244146796,
+      "learning_rate": 0.0009827177401171361,
+      "loss": 3.9282,
+      "step": 978
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.2138194095477044,
+      "learning_rate": 0.0009826693068106876,
+      "loss": 3.8928,
+      "step": 979
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.5299901798749944,
+      "learning_rate": 0.0009826208069292086,
+      "loss": 3.907,
+      "step": 980
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.871226334841303,
+      "learning_rate": 0.000982572240479389,
+      "loss": 3.9077,
+      "step": 981
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.922406486616021,
+      "learning_rate": 0.0009825236074679274,
+      "loss": 3.9062,
+      "step": 982
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.0249492350825813,
+      "learning_rate": 0.0009824749079015318,
+      "loss": 4.019,
+      "step": 983
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.0086605391722083,
+      "learning_rate": 0.0009824261417869197,
+      "loss": 3.7347,
+      "step": 984
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9512075031329892,
+      "learning_rate": 0.000982377309130817,
+      "loss": 3.9904,
+      "step": 985
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8919191046963494,
+      "learning_rate": 0.0009823284099399596,
+      "loss": 3.9638,
+      "step": 986
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9256174428816256,
+      "learning_rate": 0.000982279444221092,
+      "loss": 4.0487,
+      "step": 987
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9344695530765355,
+      "learning_rate": 0.0009822304119809682,
+      "loss": 3.8153,
+      "step": 988
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.1618597474833134,
+      "learning_rate": 0.0009821813132263513,
+      "loss": 3.7621,
+      "step": 989
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.1694020693267522,
+      "learning_rate": 0.0009821321479640134,
+      "loss": 4.0705,
+      "step": 990
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.1910474258879573,
+      "learning_rate": 0.0009820829162007357,
+      "loss": 3.8143,
+      "step": 991
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.0254334914806764,
+      "learning_rate": 0.0009820336179433091,
+      "loss": 4.0099,
+      "step": 992
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9579173060420042,
+      "learning_rate": 0.0009819842531985337,
+      "loss": 3.977,
+      "step": 993
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9358655906582398,
+      "learning_rate": 0.0009819348219732176,
+      "loss": 3.9295,
+      "step": 994
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.444047997788228,
+      "learning_rate": 0.0009818853242741796,
+      "loss": 3.9272,
+      "step": 995
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.0017835733805611,
+      "learning_rate": 0.0009818357601082467,
+      "loss": 3.715,
+      "step": 996
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9954102067885197,
+      "learning_rate": 0.0009817861294822551,
+      "loss": 3.9233,
+      "step": 997
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.153248118935177,
+      "learning_rate": 0.0009817364324030506,
+      "loss": 3.8396,
+      "step": 998
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8938412800416489,
+      "learning_rate": 0.0009816866688774882,
+      "loss": 4.0166,
+      "step": 999
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.0703556677998551,
+      "learning_rate": 0.0009816368389124314,
+      "loss": 4.1067,
+      "step": 1000
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8779486440666361,
+      "learning_rate": 0.0009815869425147537,
+      "loss": 3.8926,
+      "step": 1001
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8402222763064704,
+      "learning_rate": 0.0009815369796913373,
+      "loss": 4.0144,
+      "step": 1002
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8720916938147475,
+      "learning_rate": 0.0009814869504490731,
+      "loss": 3.9013,
+      "step": 1003
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.0503146728479664,
+      "learning_rate": 0.0009814368547948623,
+      "loss": 3.917,
+      "step": 1004
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.2664811563250036,
+      "learning_rate": 0.0009813866927356142,
+      "loss": 4.0945,
+      "step": 1005
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.850409310378976,
+      "learning_rate": 0.000981336464278248,
+      "loss": 3.9544,
+      "step": 1006
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.1562403798935703,
+      "learning_rate": 0.0009812861694296917,
+      "loss": 4.1595,
+      "step": 1007
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8611576242078582,
+      "learning_rate": 0.0009812358081968825,
+      "loss": 3.816,
+      "step": 1008
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9601032419503073,
+      "learning_rate": 0.0009811853805867668,
+      "loss": 3.8064,
+      "step": 1009
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.002170866931476,
+      "learning_rate": 0.0009811348866063,
+      "loss": 3.9235,
+      "step": 1010
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.0656947634857703,
+      "learning_rate": 0.0009810843262624467,
+      "loss": 4.181,
+      "step": 1011
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.0753461893624978,
+      "learning_rate": 0.000981033699562181,
+      "loss": 4.2012,
+      "step": 1012
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.1098913276962474,
+      "learning_rate": 0.0009809830065124858,
+      "loss": 4.0372,
+      "step": 1013
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8169105624953324,
+      "learning_rate": 0.0009809322471203534,
+      "loss": 3.8732,
+      "step": 1014
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4528576114590877,
+      "learning_rate": 0.0009808814213927847,
+      "loss": 3.9142,
+      "step": 1015
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9777098750022875,
+      "learning_rate": 0.0009808305293367904,
+      "loss": 4.0211,
+      "step": 1016
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9782073638625153,
+      "learning_rate": 0.00098077957095939,
+      "loss": 3.9494,
+      "step": 1017
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 8.065703197536335,
+      "learning_rate": 0.0009807285462676122,
+      "loss": 3.9032,
+      "step": 1018
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9595601135376,
+      "learning_rate": 0.0009806774552684953,
+      "loss": 3.9296,
+      "step": 1019
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.0802493092494738,
+      "learning_rate": 0.0009806262979690857,
+      "loss": 3.7423,
+      "step": 1020
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9486785369449944,
+      "learning_rate": 0.00098057507437644,
+      "loss": 3.7127,
+      "step": 1021
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.0014242351339855,
+      "learning_rate": 0.0009805237844976234,
+      "loss": 3.8458,
+      "step": 1022
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.0327791793212355,
+      "learning_rate": 0.00098047242833971,
+      "loss": 3.922,
+      "step": 1023
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.2100402984317076,
+      "learning_rate": 0.0009804210059097841,
+      "loss": 3.9106,
+      "step": 1024
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8817661392560533,
+      "learning_rate": 0.0009803695172149382,
+      "loss": 3.8888,
+      "step": 1025
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8793720280109008,
+      "learning_rate": 0.0009803179622622738,
+      "loss": 3.8335,
+      "step": 1026
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9229476187394094,
+      "learning_rate": 0.0009802663410589023,
+      "loss": 4.0084,
+      "step": 1027
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.90576892869129,
+      "learning_rate": 0.0009802146536119437,
+      "loss": 3.8512,
+      "step": 1028
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8997648987761988,
+      "learning_rate": 0.0009801628999285274,
+      "loss": 3.8298,
+      "step": 1029
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8960727440752199,
+      "learning_rate": 0.000980111080015792,
+      "loss": 4.0569,
+      "step": 1030
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.1329530184939047,
+      "learning_rate": 0.0009800591938808846,
+      "loss": 3.6638,
+      "step": 1031
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.0287127995449656,
+      "learning_rate": 0.0009800072415309623,
+      "loss": 4.0796,
+      "step": 1032
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9757130951595525,
+      "learning_rate": 0.0009799552229731907,
+      "loss": 3.6638,
+      "step": 1033
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.1210693178755677,
+      "learning_rate": 0.0009799031382147448,
+      "loss": 4.0427,
+      "step": 1034
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.0815215580001027,
+      "learning_rate": 0.000979850987262809,
+      "loss": 3.8554,
+      "step": 1035
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9233147763416208,
+      "learning_rate": 0.0009797987701245761,
+      "loss": 3.7957,
+      "step": 1036
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.849507997381606,
+      "learning_rate": 0.0009797464868072487,
+      "loss": 3.9477,
+      "step": 1037
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.053247892875251,
+      "learning_rate": 0.0009796941373180384,
+      "loss": 4.1077,
+      "step": 1038
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.0125863073977779,
+      "learning_rate": 0.0009796417216641653,
+      "loss": 4.0688,
+      "step": 1039
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9355725066955353,
+      "learning_rate": 0.00097958923985286,
+      "loss": 3.8537,
+      "step": 1040
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.0102238144367948,
+      "learning_rate": 0.0009795366918913604,
+      "loss": 3.9841,
+      "step": 1041
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.286312791698429,
+      "learning_rate": 0.0009794840777869152,
+      "loss": 3.9493,
+      "step": 1042
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.0009944917807516,
+      "learning_rate": 0.0009794313975467813,
+      "loss": 4.1359,
+      "step": 1043
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.7019179369903037,
+      "learning_rate": 0.0009793786511782248,
+      "loss": 3.8608,
+      "step": 1044
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.1226239568174194,
+      "learning_rate": 0.000979325838688521,
+      "loss": 3.9338,
+      "step": 1045
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.04998013546436,
+      "learning_rate": 0.000979272960084955,
+      "loss": 4.0046,
+      "step": 1046
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 23.127785065075507,
+      "learning_rate": 0.0009792200153748195,
+      "loss": 3.8931,
+      "step": 1047
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.0233491385564062,
+      "learning_rate": 0.0009791670045654177,
+      "loss": 3.9848,
+      "step": 1048
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9708739678339066,
+      "learning_rate": 0.0009791139276640614,
+      "loss": 3.7089,
+      "step": 1049
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.3131747514785732,
+      "learning_rate": 0.0009790607846780718,
+      "loss": 3.9716,
+      "step": 1050
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9130761288766097,
+      "learning_rate": 0.0009790075756147783,
+      "loss": 3.9495,
+      "step": 1051
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.002724071290709,
+      "learning_rate": 0.0009789543004815207,
+      "loss": 3.888,
+      "step": 1052
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.141595559541945,
+      "learning_rate": 0.000978900959285647,
+      "loss": 4.1737,
+      "step": 1053
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 5.397410353361243,
+      "learning_rate": 0.0009788475520345146,
+      "loss": 4.1982,
+      "step": 1054
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.0184457212853797,
+      "learning_rate": 0.0009787940787354902,
+      "loss": 3.9478,
+      "step": 1055
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.0252821127272502,
+      "learning_rate": 0.000978740539395949,
+      "loss": 4.2953,
+      "step": 1056
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.1329847435647367,
+      "learning_rate": 0.0009786869340232761,
+      "loss": 4.1711,
+      "step": 1057
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.4394819911694596,
+      "learning_rate": 0.0009786332626248655,
+      "loss": 3.8764,
+      "step": 1058
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.0609581740236735,
+      "learning_rate": 0.0009785795252081199,
+      "loss": 4.0235,
+      "step": 1059
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.1760910160225522,
+      "learning_rate": 0.000978525721780451,
+      "loss": 4.0556,
+      "step": 1060
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.0769567091621046,
+      "learning_rate": 0.0009784718523492804,
+      "loss": 3.8768,
+      "step": 1061
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.967710877572883,
+      "learning_rate": 0.0009784179169220384,
+      "loss": 4.1386,
+      "step": 1062
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.4635240714726403,
+      "learning_rate": 0.0009783639155061643,
+      "loss": 3.7415,
+      "step": 1063
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.3595409786574135,
+      "learning_rate": 0.0009783098481091063,
+      "loss": 4.0034,
+      "step": 1064
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.019660819496443,
+      "learning_rate": 0.0009782557147383225,
+      "loss": 3.8662,
+      "step": 1065
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9728337803487399,
+      "learning_rate": 0.0009782015154012789,
+      "loss": 3.9876,
+      "step": 1066
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.9812589155217257,
+      "learning_rate": 0.0009781472501054517,
+      "loss": 3.894,
+      "step": 1067
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9911645329615667,
+      "learning_rate": 0.0009780929188583256,
+      "loss": 3.9516,
+      "step": 1068
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9923902928553388,
+      "learning_rate": 0.000978038521667395,
+      "loss": 3.9346,
+      "step": 1069
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.3905833246413395,
+      "learning_rate": 0.000977984058540162,
+      "loss": 3.9289,
+      "step": 1070
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9675031408698997,
+      "learning_rate": 0.0009779295294841397,
+      "loss": 3.8922,
+      "step": 1071
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9389966918489403,
+      "learning_rate": 0.0009778749345068487,
+      "loss": 3.9755,
+      "step": 1072
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.405637937238413,
+      "learning_rate": 0.00097782027361582,
+      "loss": 3.9464,
+      "step": 1073
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.073691417837058,
+      "learning_rate": 0.0009777655468185924,
+      "loss": 3.8502,
+      "step": 1074
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.0378858820685037,
+      "learning_rate": 0.0009777107541227147,
+      "loss": 3.9764,
+      "step": 1075
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9890756635492401,
+      "learning_rate": 0.0009776558955357443,
+      "loss": 3.9754,
+      "step": 1076
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8296281014599762,
+      "learning_rate": 0.0009776009710652483,
+      "loss": 3.8537,
+      "step": 1077
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.7290123304984844,
+      "learning_rate": 0.0009775459807188022,
+      "loss": 3.8663,
+      "step": 1078
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9817978258209319,
+      "learning_rate": 0.0009774909245039909,
+      "loss": 3.7691,
+      "step": 1079
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8339342488787881,
+      "learning_rate": 0.0009774358024284082,
+      "loss": 3.9371,
+      "step": 1080
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 17.442307799141258,
+      "learning_rate": 0.0009773806144996575,
+      "loss": 3.8432,
+      "step": 1081
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9967067697136237,
+      "learning_rate": 0.0009773253607253507,
+      "loss": 3.9011,
+      "step": 1082
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.0017458744696415,
+      "learning_rate": 0.000977270041113109,
+      "loss": 3.7896,
+      "step": 1083
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9002236822287353,
+      "learning_rate": 0.0009772146556705629,
+      "loss": 3.9173,
+      "step": 1084
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9977173807841384,
+      "learning_rate": 0.0009771592044053512,
+      "loss": 3.9252,
+      "step": 1085
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.0660577587375237,
+      "learning_rate": 0.000977103687325123,
+      "loss": 4.1303,
+      "step": 1086
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.0083702956163125,
+      "learning_rate": 0.0009770481044375356,
+      "loss": 3.7057,
+      "step": 1087
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.212960703024689,
+      "learning_rate": 0.0009769924557502553,
+      "loss": 4.0127,
+      "step": 1088
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.1826640503187567,
+      "learning_rate": 0.0009769367412709585,
+      "loss": 4.0358,
+      "step": 1089
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.1891628075347187,
+      "learning_rate": 0.0009768809610073291,
+      "loss": 3.8099,
+      "step": 1090
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.1849385218204642,
+      "learning_rate": 0.0009768251149670614,
+      "loss": 3.9771,
+      "step": 1091
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.0960542877740704,
+      "learning_rate": 0.000976769203157858,
+      "loss": 4.009,
+      "step": 1092
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.1200359925239947,
+      "learning_rate": 0.0009767132255874315,
+      "loss": 4.0382,
+      "step": 1093
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.739728366117808,
+      "learning_rate": 0.0009766571822635022,
+      "loss": 3.9109,
+      "step": 1094
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.1876864859921121,
+      "learning_rate": 0.0009766010731938007,
+      "loss": 4.0763,
+      "step": 1095
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9969784528544836,
+      "learning_rate": 0.0009765448983860658,
+      "loss": 3.9442,
+      "step": 1096
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9787198916271599,
+      "learning_rate": 0.0009764886578480461,
+      "loss": 3.8233,
+      "step": 1097
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.0066562113136273,
+      "learning_rate": 0.0009764323515874986,
+      "loss": 4.0883,
+      "step": 1098
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9424351528688372,
+      "learning_rate": 0.00097637597961219,
+      "loss": 4.1467,
+      "step": 1099
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9308179235296616,
+      "learning_rate": 0.0009763195419298955,
+      "loss": 3.905,
+      "step": 1100
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8350404954052347,
+      "learning_rate": 0.0009762630385483997,
+      "loss": 3.9141,
+      "step": 1101
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9793223355773603,
+      "learning_rate": 0.000976206469475496,
+      "loss": 3.7877,
+      "step": 1102
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9135308758085584,
+      "learning_rate": 0.0009761498347189872,
+      "loss": 4.0504,
+      "step": 1103
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.079484610301383,
+      "learning_rate": 0.000976093134286685,
+      "loss": 4.0459,
+      "step": 1104
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.978483905972476,
+      "learning_rate": 0.0009760363681864102,
+      "loss": 3.9595,
+      "step": 1105
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8603675204627352,
+      "learning_rate": 0.0009759795364259923,
+      "loss": 3.8813,
+      "step": 1106
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.318098850626356,
+      "learning_rate": 0.0009759226390132704,
+      "loss": 3.911,
+      "step": 1107
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.7861355971294172,
+      "learning_rate": 0.0009758656759560923,
+      "loss": 3.796,
+      "step": 1108
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.819559094728256,
+      "learning_rate": 0.0009758086472623151,
+      "loss": 3.8339,
+      "step": 1109
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9790363418652527,
+      "learning_rate": 0.0009757515529398047,
+      "loss": 3.7797,
+      "step": 1110
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8292137228396096,
+      "learning_rate": 0.0009756943929964363,
+      "loss": 3.8246,
+      "step": 1111
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.4099819499835757,
+      "learning_rate": 0.0009756371674400939,
+      "loss": 3.8785,
+      "step": 1112
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.196210265983327,
+      "learning_rate": 0.0009755798762786707,
+      "loss": 3.9509,
+      "step": 1113
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.5932713830996734,
+      "learning_rate": 0.0009755225195200689,
+      "loss": 3.7561,
+      "step": 1114
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8865499541842148,
+      "learning_rate": 0.0009754650971722,
+      "loss": 4.0031,
+      "step": 1115
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.050241254972174,
+      "learning_rate": 0.000975407609242984,
+      "loss": 3.8365,
+      "step": 1116
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.979100013549394,
+      "learning_rate": 0.0009753500557403504,
+      "loss": 3.924,
+      "step": 1117
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9226737305817291,
+      "learning_rate": 0.0009752924366722376,
+      "loss": 3.9107,
+      "step": 1118
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.2910312605123768,
+      "learning_rate": 0.0009752347520465931,
+      "loss": 4.0942,
+      "step": 1119
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.0363270903051176,
+      "learning_rate": 0.0009751770018713734,
+      "loss": 4.0032,
+      "step": 1120
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.0710260221815096,
+      "learning_rate": 0.0009751191861545439,
+      "loss": 3.9134,
+      "step": 1121
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9590511023952412,
+      "learning_rate": 0.0009750613049040792,
+      "loss": 4.1202,
+      "step": 1122
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.0740443034445635,
+      "learning_rate": 0.0009750033581279632,
+      "loss": 3.8447,
+      "step": 1123
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.2479679222789966,
+      "learning_rate": 0.0009749453458341882,
+      "loss": 4.0468,
+      "step": 1124
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.0708737222341365,
+      "learning_rate": 0.000974887268030756,
+      "loss": 4.0188,
+      "step": 1125
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9324155150206787,
+      "learning_rate": 0.0009748291247256774,
+      "loss": 3.9363,
+      "step": 1126
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.0026845219515736,
+      "learning_rate": 0.000974770915926972,
+      "loss": 4.0482,
+      "step": 1127
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.2178892594430921,
+      "learning_rate": 0.0009747126416426688,
+      "loss": 3.9245,
+      "step": 1128
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9349787746654444,
+      "learning_rate": 0.0009746543018808057,
+      "loss": 3.8602,
+      "step": 1129
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.3786632824107816,
+      "learning_rate": 0.000974595896649429,
+      "loss": 3.952,
+      "step": 1130
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9194498173676063,
+      "learning_rate": 0.0009745374259565953,
+      "loss": 3.855,
+      "step": 1131
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.0000839234551935,
+      "learning_rate": 0.0009744788898103691,
+      "loss": 3.7253,
+      "step": 1132
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8219108142358327,
+      "learning_rate": 0.0009744202882188245,
+      "loss": 3.8803,
+      "step": 1133
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.6275263292561466,
+      "learning_rate": 0.0009743616211900443,
+      "loss": 4.3225,
+      "step": 1134
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8770319765623361,
+      "learning_rate": 0.0009743028887321206,
+      "loss": 4.0887,
+      "step": 1135
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9949940276746041,
+      "learning_rate": 0.0009742440908531545,
+      "loss": 3.8521,
+      "step": 1136
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.1078975560683888,
+      "learning_rate": 0.0009741852275612559,
+      "loss": 3.926,
+      "step": 1137
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9353909686477906,
+      "learning_rate": 0.0009741262988645441,
+      "loss": 3.5989,
+      "step": 1138
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.0736536842874682,
+      "learning_rate": 0.000974067304771147,
+      "loss": 4.1232,
+      "step": 1139
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9971858344764342,
+      "learning_rate": 0.0009740082452892017,
+      "loss": 3.8146,
+      "step": 1140
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8552780525329229,
+      "learning_rate": 0.0009739491204268545,
+      "loss": 3.7645,
+      "step": 1141
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.3475018234506315,
+      "learning_rate": 0.0009738899301922602,
+      "loss": 3.8519,
+      "step": 1142
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9430418954860429,
+      "learning_rate": 0.0009738306745935833,
+      "loss": 3.6373,
+      "step": 1143
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.0411807247329004,
+      "learning_rate": 0.0009737713536389969,
+      "loss": 4.0995,
+      "step": 1144
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.7056517923553942,
+      "learning_rate": 0.0009737119673366832,
+      "loss": 3.9293,
+      "step": 1145
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.4654860798799498,
+      "learning_rate": 0.0009736525156948333,
+      "loss": 4.0413,
+      "step": 1146
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.1876264805564414,
+      "learning_rate": 0.0009735929987216476,
+      "loss": 3.7652,
+      "step": 1147
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9931410616179552,
+      "learning_rate": 0.0009735334164253351,
+      "loss": 3.9004,
+      "step": 1148
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.3799693589094666,
+      "learning_rate": 0.0009734737688141142,
+      "loss": 3.9107,
+      "step": 1149
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.0810662208730164,
+      "learning_rate": 0.0009734140558962123,
+      "loss": 3.8137,
+      "step": 1150
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.0751209991460242,
+      "learning_rate": 0.0009733542776798653,
+      "loss": 3.9201,
+      "step": 1151
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9391578210044854,
+      "learning_rate": 0.0009732944341733188,
+      "loss": 3.8752,
+      "step": 1152
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9103164985905934,
+      "learning_rate": 0.0009732345253848267,
+      "loss": 3.9437,
+      "step": 1153
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8805930826421503,
+      "learning_rate": 0.0009731745513226526,
+      "loss": 3.8616,
+      "step": 1154
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9862488819729535,
+      "learning_rate": 0.0009731145119950686,
+      "loss": 3.8947,
+      "step": 1155
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8701133860437691,
+      "learning_rate": 0.0009730544074103562,
+      "loss": 3.9924,
+      "step": 1156
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8806260045882947,
+      "learning_rate": 0.0009729942375768055,
+      "loss": 4.0764,
+      "step": 1157
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.0750024293515343,
+      "learning_rate": 0.0009729340025027158,
+      "loss": 4.0736,
+      "step": 1158
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9539820319101168,
+      "learning_rate": 0.0009728737021963954,
+      "loss": 4.1145,
+      "step": 1159
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8400434721530677,
+      "learning_rate": 0.0009728133366661615,
+      "loss": 3.7788,
+      "step": 1160
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.0448476305721772,
+      "learning_rate": 0.0009727529059203406,
+      "loss": 3.8515,
+      "step": 1161
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8076328172221366,
+      "learning_rate": 0.0009726924099672676,
+      "loss": 3.7931,
+      "step": 1162
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8910235966377034,
+      "learning_rate": 0.0009726318488152872,
+      "loss": 3.8467,
+      "step": 1163
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8571276406922334,
+      "learning_rate": 0.0009725712224727523,
+      "loss": 3.9025,
+      "step": 1164
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9590413218400093,
+      "learning_rate": 0.0009725105309480253,
+      "loss": 3.8987,
+      "step": 1165
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.0992033490170205,
+      "learning_rate": 0.0009724497742494776,
+      "loss": 3.9042,
+      "step": 1166
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9293464628060762,
+      "learning_rate": 0.000972388952385489,
+      "loss": 3.9285,
+      "step": 1167
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8336888188418624,
+      "learning_rate": 0.000972328065364449,
+      "loss": 3.7819,
+      "step": 1168
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.862776407006224,
+      "learning_rate": 0.0009722671131947559,
+      "loss": 3.73,
+      "step": 1169
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9177773761208528,
+      "learning_rate": 0.0009722060958848168,
+      "loss": 3.7145,
+      "step": 1170
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9323202380596227,
+      "learning_rate": 0.0009721450134430478,
+      "loss": 3.6659,
+      "step": 1171
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.7276673438564107,
+      "learning_rate": 0.000972083865877874,
+      "loss": 4.0897,
+      "step": 1172
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.101331670613349,
+      "learning_rate": 0.0009720226531977296,
+      "loss": 4.0631,
+      "step": 1173
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.585525437709138,
+      "learning_rate": 0.0009719613754110578,
+      "loss": 3.9867,
+      "step": 1174
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8583964109428411,
+      "learning_rate": 0.0009719000325263109,
+      "loss": 3.7867,
+      "step": 1175
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9082085823031396,
+      "learning_rate": 0.0009718386245519495,
+      "loss": 4.0505,
+      "step": 1176
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.948182118660867,
+      "learning_rate": 0.0009717771514964439,
+      "loss": 3.8383,
+      "step": 1177
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8985462306153706,
+      "learning_rate": 0.0009717156133682734,
+      "loss": 3.884,
+      "step": 1178
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9231041336272474,
+      "learning_rate": 0.0009716540101759255,
+      "loss": 3.9573,
+      "step": 1179
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.2811277116731867,
+      "learning_rate": 0.0009715923419278976,
+      "loss": 4.1143,
+      "step": 1180
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.3713842833463334,
+      "learning_rate": 0.0009715306086326954,
+      "loss": 3.9203,
+      "step": 1181
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.0078768577373445,
+      "learning_rate": 0.0009714688102988339,
+      "loss": 3.746,
+      "step": 1182
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9918324555669916,
+      "learning_rate": 0.000971406946934837,
+      "loss": 3.9872,
+      "step": 1183
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8379789885252091,
+      "learning_rate": 0.0009713450185492378,
+      "loss": 3.8235,
+      "step": 1184
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9684206558888561,
+      "learning_rate": 0.0009712830251505778,
+      "loss": 4.0387,
+      "step": 1185
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.0559634744904582,
+      "learning_rate": 0.0009712209667474079,
+      "loss": 3.8792,
+      "step": 1186
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9237859787622809,
+      "learning_rate": 0.0009711588433482881,
+      "loss": 4.0836,
+      "step": 1187
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.0723381152955602,
+      "learning_rate": 0.0009710966549617868,
+      "loss": 3.9188,
+      "step": 1188
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.0918463392391256,
+      "learning_rate": 0.0009710344015964819,
+      "loss": 3.9639,
+      "step": 1189
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9265418349746243,
+      "learning_rate": 0.00097097208326096,
+      "loss": 4.0069,
+      "step": 1190
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.5178891156216168,
+      "learning_rate": 0.000970909699963817,
+      "loss": 3.7814,
+      "step": 1191
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.6779770924143507,
+      "learning_rate": 0.0009708472517136569,
+      "loss": 4.1803,
+      "step": 1192
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.2165605602695155,
+      "learning_rate": 0.0009707847385190938,
+      "loss": 3.9068,
+      "step": 1193
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.0190319722154504,
+      "learning_rate": 0.00097072216038875,
+      "loss": 3.7683,
+      "step": 1194
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 5.1837821332477105,
+      "learning_rate": 0.000970659517331257,
+      "loss": 3.7351,
+      "step": 1195
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9766148641113799,
+      "learning_rate": 0.000970596809355255,
+      "loss": 3.9429,
+      "step": 1196
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8260206247312476,
+      "learning_rate": 0.0009705340364693935,
+      "loss": 3.7627,
+      "step": 1197
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8411051019628619,
+      "learning_rate": 0.0009704711986823311,
+      "loss": 3.7795,
+      "step": 1198
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9496432667931038,
+      "learning_rate": 0.0009704082960027348,
+      "loss": 3.7983,
+      "step": 1199
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9915788351529774,
+      "learning_rate": 0.0009703453284392807,
+      "loss": 4.319,
+      "step": 1200
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9717986654886581,
+      "learning_rate": 0.0009702822960006544,
+      "loss": 4.0243,
+      "step": 1201
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8185385468634446,
+      "learning_rate": 0.0009702191986955494,
+      "loss": 3.8957,
+      "step": 1202
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.0586923077649357,
+      "learning_rate": 0.0009701560365326694,
+      "loss": 3.8052,
+      "step": 1203
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.1035780592267228,
+      "learning_rate": 0.0009700928095207259,
+      "loss": 3.97,
+      "step": 1204
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.024785938354129,
+      "learning_rate": 0.0009700295176684404,
+      "loss": 3.8009,
+      "step": 1205
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8752527924479723,
+      "learning_rate": 0.0009699661609845425,
+      "loss": 3.9964,
+      "step": 1206
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9247902479294354,
+      "learning_rate": 0.0009699027394777708,
+      "loss": 3.8331,
+      "step": 1207
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.8210148630626366,
+      "learning_rate": 0.0009698392531568736,
+      "loss": 3.9239,
+      "step": 1208
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.1764690578291919,
+      "learning_rate": 0.0009697757020306072,
+      "loss": 3.9713,
+      "step": 1209
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9953746671649932,
+      "learning_rate": 0.0009697120861077375,
+      "loss": 3.9377,
+      "step": 1210
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8694677857898438,
+      "learning_rate": 0.0009696484053970391,
+      "loss": 3.7729,
+      "step": 1211
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.903146042590022,
+      "learning_rate": 0.0009695846599072955,
+      "loss": 3.9141,
+      "step": 1212
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8669656628328067,
+      "learning_rate": 0.0009695208496472991,
+      "loss": 3.8157,
+      "step": 1213
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.1188933199264146,
+      "learning_rate": 0.0009694569746258514,
+      "loss": 3.9868,
+      "step": 1214
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.3689524870133447,
+      "learning_rate": 0.0009693930348517628,
+      "loss": 3.8533,
+      "step": 1215
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8243949113960238,
+      "learning_rate": 0.0009693290303338524,
+      "loss": 3.9983,
+      "step": 1216
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.0370525853140824,
+      "learning_rate": 0.0009692649610809485,
+      "loss": 3.7288,
+      "step": 1217
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9213611034514192,
+      "learning_rate": 0.0009692008271018883,
+      "loss": 3.8335,
+      "step": 1218
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.599298364329189,
+      "learning_rate": 0.0009691366284055176,
+      "loss": 3.941,
+      "step": 1219
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.6653181362401195,
+      "learning_rate": 0.0009690723650006917,
+      "loss": 3.9675,
+      "step": 1220
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9312186132161909,
+      "learning_rate": 0.0009690080368962744,
+      "loss": 3.774,
+      "step": 1221
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9692639306437891,
+      "learning_rate": 0.0009689436441011384,
+      "loss": 3.8665,
+      "step": 1222
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8273415757242735,
+      "learning_rate": 0.0009688791866241657,
+      "loss": 3.9077,
+      "step": 1223
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.0663875762438544,
+      "learning_rate": 0.0009688146644742468,
+      "loss": 3.8998,
+      "step": 1224
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9616301414674919,
+      "learning_rate": 0.0009687500776602813,
+      "loss": 3.8716,
+      "step": 1225
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.0740893948761774,
+      "learning_rate": 0.0009686854261911779,
+      "loss": 3.9514,
+      "step": 1226
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.1757782467496745,
+      "learning_rate": 0.0009686207100758538,
+      "loss": 3.9485,
+      "step": 1227
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8496873265597074,
+      "learning_rate": 0.0009685559293232355,
+      "loss": 3.7576,
+      "step": 1228
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.9374119557694822,
+      "learning_rate": 0.0009684910839422582,
+      "loss": 3.9243,
+      "step": 1229
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9619536208307495,
+      "learning_rate": 0.0009684261739418663,
+      "loss": 4.0992,
+      "step": 1230
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.1671513566718728,
+      "learning_rate": 0.0009683611993310127,
+      "loss": 3.9332,
+      "step": 1231
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9778001991457772,
+      "learning_rate": 0.0009682961601186593,
+      "loss": 3.5797,
+      "step": 1232
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.0265196995344394,
+      "learning_rate": 0.000968231056313777,
+      "loss": 3.9148,
+      "step": 1233
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9059116270277968,
+      "learning_rate": 0.0009681658879253461,
+      "loss": 3.9402,
+      "step": 1234
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9069721940576991,
+      "learning_rate": 0.0009681006549623548,
+      "loss": 4.0391,
+      "step": 1235
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.9526955685459364,
+      "learning_rate": 0.000968035357433801,
+      "loss": 3.8473,
+      "step": 1236
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.5774996971858504,
+      "learning_rate": 0.0009679699953486913,
+      "loss": 3.9154,
+      "step": 1237
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.873144291779001,
+      "learning_rate": 0.0009679045687160411,
+      "loss": 3.9218,
+      "step": 1238
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9817585569336963,
+      "learning_rate": 0.0009678390775448745,
+      "loss": 3.9995,
+      "step": 1239
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8812904909526799,
+      "learning_rate": 0.0009677735218442252,
+      "loss": 3.8041,
+      "step": 1240
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9133073847331136,
+      "learning_rate": 0.0009677079016231349,
+      "loss": 3.6559,
+      "step": 1241
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8747530026434839,
+      "learning_rate": 0.000967642216890655,
+      "loss": 3.8392,
+      "step": 1242
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.7411111083745054,
+      "learning_rate": 0.0009675764676558454,
+      "loss": 3.9429,
+      "step": 1243
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.7965217929417112,
+      "learning_rate": 0.000967510653927775,
+      "loss": 3.6893,
+      "step": 1244
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8904969095504146,
+      "learning_rate": 0.0009674447757155213,
+      "loss": 4.0174,
+      "step": 1245
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.0354428267514901,
+      "learning_rate": 0.0009673788330281709,
+      "loss": 4.0429,
+      "step": 1246
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.0165140078196517,
+      "learning_rate": 0.0009673128258748199,
+      "loss": 4.0743,
+      "step": 1247
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9627060747892144,
+      "learning_rate": 0.0009672467542645722,
+      "loss": 3.6733,
+      "step": 1248
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.7952932516583989,
+      "learning_rate": 0.0009671806182065414,
+      "loss": 3.7791,
+      "step": 1249
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9093254163507255,
+      "learning_rate": 0.0009671144177098494,
+      "loss": 3.8856,
+      "step": 1250
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.877167926426142,
+      "learning_rate": 0.0009670481527836276,
+      "loss": 4.0906,
+      "step": 1251
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9528383678188778,
+      "learning_rate": 0.000966981823437016,
+      "loss": 3.855,
+      "step": 1252
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9803766757838255,
+      "learning_rate": 0.0009669154296791632,
+      "loss": 3.867,
+      "step": 1253
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9565626384520503,
+      "learning_rate": 0.000966848971519227,
+      "loss": 4.0007,
+      "step": 1254
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9049454020266207,
+      "learning_rate": 0.0009667824489663743,
+      "loss": 3.8138,
+      "step": 1255
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9131884616341465,
+      "learning_rate": 0.0009667158620297803,
+      "loss": 4.0401,
+      "step": 1256
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8536423883013678,
+      "learning_rate": 0.0009666492107186296,
+      "loss": 3.8983,
+      "step": 1257
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.832657604313244,
+      "learning_rate": 0.0009665824950421155,
+      "loss": 3.7254,
+      "step": 1258
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9824151155200311,
+      "learning_rate": 0.00096651571500944,
+      "loss": 3.9107,
+      "step": 1259
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8276177811256961,
+      "learning_rate": 0.0009664488706298142,
+      "loss": 3.831,
+      "step": 1260
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.7468232282004362,
+      "learning_rate": 0.0009663819619124581,
+      "loss": 3.7489,
+      "step": 1261
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9434729693255196,
+      "learning_rate": 0.0009663149888666003,
+      "loss": 3.842,
+      "step": 1262
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.1658334038404314,
+      "learning_rate": 0.0009662479515014786,
+      "loss": 3.9475,
+      "step": 1263
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.7075301083317621,
+      "learning_rate": 0.0009661808498263396,
+      "loss": 3.7535,
+      "step": 1264
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8225387168517201,
+      "learning_rate": 0.0009661136838504385,
+      "loss": 4.0481,
+      "step": 1265
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.928875318025512,
+      "learning_rate": 0.0009660464535830395,
+      "loss": 3.8676,
+      "step": 1266
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8002942613710375,
+      "learning_rate": 0.0009659791590334162,
+      "loss": 4.0203,
+      "step": 1267
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8346160161789712,
+      "learning_rate": 0.00096591180021085,
+      "loss": 3.9455,
+      "step": 1268
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8933388746841424,
+      "learning_rate": 0.0009658443771246322,
+      "loss": 3.9002,
+      "step": 1269
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9007159497871604,
+      "learning_rate": 0.0009657768897840623,
+      "loss": 3.713,
+      "step": 1270
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.97180683670039,
+      "learning_rate": 0.000965709338198449,
+      "loss": 4.2397,
+      "step": 1271
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8870994127514898,
+      "learning_rate": 0.0009656417223771097,
+      "loss": 3.9728,
+      "step": 1272
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8368788807463914,
+      "learning_rate": 0.0009655740423293708,
+      "loss": 3.8196,
+      "step": 1273
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.014448825042988,
+      "learning_rate": 0.0009655062980645673,
+      "loss": 4.0431,
+      "step": 1274
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9436348863244459,
+      "learning_rate": 0.0009654384895920434,
+      "loss": 3.8817,
+      "step": 1275
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8354934620718959,
+      "learning_rate": 0.0009653706169211519,
+      "loss": 3.6851,
+      "step": 1276
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8973061654584843,
+      "learning_rate": 0.0009653026800612545,
+      "loss": 3.8962,
+      "step": 1277
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8312593493267536,
+      "learning_rate": 0.0009652346790217221,
+      "loss": 3.776,
+      "step": 1278
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8589715391081593,
+      "learning_rate": 0.0009651666138119337,
+      "loss": 3.7952,
+      "step": 1279
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8005301083109356,
+      "learning_rate": 0.000965098484441278,
+      "loss": 3.743,
+      "step": 1280
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.755995690059652,
+      "learning_rate": 0.0009650302909191517,
+      "loss": 3.6772,
+      "step": 1281
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.1373651802508393,
+      "learning_rate": 0.0009649620332549613,
+      "loss": 3.8801,
+      "step": 1282
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9040321464447719,
+      "learning_rate": 0.0009648937114581212,
+      "loss": 3.7437,
+      "step": 1283
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8630069007659117,
+      "learning_rate": 0.0009648253255380554,
+      "loss": 3.8694,
+      "step": 1284
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9710698534622276,
+      "learning_rate": 0.0009647568755041963,
+      "loss": 3.7386,
+      "step": 1285
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8432617763973004,
+      "learning_rate": 0.0009646883613659851,
+      "loss": 3.7495,
+      "step": 1286
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8465614012657681,
+      "learning_rate": 0.0009646197831328725,
+      "loss": 4.0602,
+      "step": 1287
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9064616818639412,
+      "learning_rate": 0.0009645511408143171,
+      "loss": 3.9505,
+      "step": 1288
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8084622098369357,
+      "learning_rate": 0.0009644824344197872,
+      "loss": 3.7778,
+      "step": 1289
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8689209397891293,
+      "learning_rate": 0.0009644136639587591,
+      "loss": 4.0088,
+      "step": 1290
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9916119943830137,
+      "learning_rate": 0.0009643448294407186,
+      "loss": 3.8514,
+      "step": 1291
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8307468634746445,
+      "learning_rate": 0.0009642759308751601,
+      "loss": 3.788,
+      "step": 1292
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.1219192852661597,
+      "learning_rate": 0.0009642069682715868,
+      "loss": 4.0463,
+      "step": 1293
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8824334248037283,
+      "learning_rate": 0.0009641379416395109,
+      "loss": 3.9621,
+      "step": 1294
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9422907916618047,
+      "learning_rate": 0.0009640688509884532,
+      "loss": 3.9566,
+      "step": 1295
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.892132150044263,
+      "learning_rate": 0.0009639996963279435,
+      "loss": 4.1511,
+      "step": 1296
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.047357009921113,
+      "learning_rate": 0.0009639304776675204,
+      "loss": 3.78,
+      "step": 1297
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9773680715175178,
+      "learning_rate": 0.0009638611950167311,
+      "loss": 3.9136,
+      "step": 1298
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.1464280036974428,
+      "learning_rate": 0.000963791848385132,
+      "loss": 3.9273,
+      "step": 1299
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.3674130327612986,
+      "learning_rate": 0.000963722437782288,
+      "loss": 3.9309,
+      "step": 1300
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8721458001584989,
+      "learning_rate": 0.0009636529632177732,
+      "loss": 3.9968,
+      "step": 1301
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.895984417247757,
+      "learning_rate": 0.0009635834247011701,
+      "loss": 3.8964,
+      "step": 1302
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9600763332124445,
+      "learning_rate": 0.0009635138222420703,
+      "loss": 4.0515,
+      "step": 1303
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8154250062168833,
+      "learning_rate": 0.000963444155850074,
+      "loss": 3.74,
+      "step": 1304
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.1605719827918148,
+      "learning_rate": 0.0009633744255347905,
+      "loss": 3.8061,
+      "step": 1305
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.7590006555552549,
+      "learning_rate": 0.0009633046313058378,
+      "loss": 3.7027,
+      "step": 1306
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.7169831921807566,
+      "learning_rate": 0.0009632347731728424,
+      "loss": 3.7103,
+      "step": 1307
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8770826866495424,
+      "learning_rate": 0.0009631648511454402,
+      "loss": 3.9115,
+      "step": 1308
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.0442792768535483,
+      "learning_rate": 0.0009630948652332756,
+      "loss": 3.9245,
+      "step": 1309
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9231483862403553,
+      "learning_rate": 0.0009630248154460017,
+      "loss": 3.8329,
+      "step": 1310
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.0125368751432804,
+      "learning_rate": 0.0009629547017932805,
+      "loss": 3.8609,
+      "step": 1311
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9052641190427552,
+      "learning_rate": 0.0009628845242847829,
+      "loss": 3.6307,
+      "step": 1312
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9328776451401802,
+      "learning_rate": 0.0009628142829301886,
+      "loss": 3.8112,
+      "step": 1313
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.1176516795359284,
+      "learning_rate": 0.000962743977739186,
+      "loss": 3.848,
+      "step": 1314
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8663653568120429,
+      "learning_rate": 0.0009626736087214724,
+      "loss": 3.9584,
+      "step": 1315
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8757822750567124,
+      "learning_rate": 0.0009626031758867538,
+      "loss": 3.7025,
+      "step": 1316
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9151154299429095,
+      "learning_rate": 0.0009625326792447451,
+      "loss": 4.0714,
+      "step": 1317
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8472711838486595,
+      "learning_rate": 0.00096246211880517,
+      "loss": 3.7862,
+      "step": 1318
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8389505833941808,
+      "learning_rate": 0.000962391494577761,
+      "loss": 3.6895,
+      "step": 1319
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8896074400237217,
+      "learning_rate": 0.0009623208065722592,
+      "loss": 3.9731,
+      "step": 1320
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9099393134619518,
+      "learning_rate": 0.0009622500547984147,
+      "loss": 3.911,
+      "step": 1321
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.027851962307429,
+      "learning_rate": 0.0009621792392659867,
+      "loss": 3.8772,
+      "step": 1322
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8588708996292934,
+      "learning_rate": 0.0009621083599847424,
+      "loss": 3.8149,
+      "step": 1323
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.7471017004927911,
+      "learning_rate": 0.0009620374169644583,
+      "loss": 3.9552,
+      "step": 1324
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9548901110839529,
+      "learning_rate": 0.0009619664102149201,
+      "loss": 3.9511,
+      "step": 1325
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9729491403642043,
+      "learning_rate": 0.0009618953397459211,
+      "loss": 3.7585,
+      "step": 1326
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.142541884602134,
+      "learning_rate": 0.0009618242055672648,
+      "loss": 3.8907,
+      "step": 1327
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.837472054918629,
+      "learning_rate": 0.0009617530076887624,
+      "loss": 3.9658,
+      "step": 1328
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8418913839647469,
+      "learning_rate": 0.0009616817461202345,
+      "loss": 3.8776,
+      "step": 1329
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9682140941402299,
+      "learning_rate": 0.0009616104208715101,
+      "loss": 3.7543,
+      "step": 1330
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8351318604275011,
+      "learning_rate": 0.0009615390319524272,
+      "loss": 3.8855,
+      "step": 1331
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.7895729419228837,
+      "learning_rate": 0.0009614675793728327,
+      "loss": 3.9336,
+      "step": 1332
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.0467883841600463,
+      "learning_rate": 0.0009613960631425818,
+      "loss": 3.7066,
+      "step": 1333
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.1515093810252361,
+      "learning_rate": 0.000961324483271539,
+      "loss": 3.8395,
+      "step": 1334
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9157140218290246,
+      "learning_rate": 0.0009612528397695777,
+      "loss": 3.834,
+      "step": 1335
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.7889100256223686,
+      "learning_rate": 0.0009611811326465791,
+      "loss": 3.8557,
+      "step": 1336
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9009750925686082,
+      "learning_rate": 0.0009611093619124344,
+      "loss": 3.8886,
+      "step": 1337
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.926422369415665,
+      "learning_rate": 0.0009610375275770427,
+      "loss": 3.9171,
+      "step": 1338
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8994071324066222,
+      "learning_rate": 0.0009609656296503121,
+      "loss": 3.7646,
+      "step": 1339
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8856720474014015,
+      "learning_rate": 0.0009608936681421599,
+      "loss": 3.8905,
+      "step": 1340
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9304596087192142,
+      "learning_rate": 0.0009608216430625114,
+      "loss": 4.1114,
+      "step": 1341
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8869970667136994,
+      "learning_rate": 0.0009607495544213014,
+      "loss": 3.8797,
+      "step": 1342
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.932280576761245,
+      "learning_rate": 0.000960677402228473,
+      "loss": 3.8959,
+      "step": 1343
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8621201819116923,
+      "learning_rate": 0.0009606051864939785,
+      "loss": 3.6971,
+      "step": 1344
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8049991112830159,
+      "learning_rate": 0.0009605329072277782,
+      "loss": 3.7375,
+      "step": 1345
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.0974918642394877,
+      "learning_rate": 0.000960460564439842,
+      "loss": 3.8616,
+      "step": 1346
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.6641828320860899,
+      "learning_rate": 0.0009603881581401482,
+      "loss": 3.8839,
+      "step": 1347
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.0861884983737076,
+      "learning_rate": 0.0009603156883386836,
+      "loss": 3.9676,
+      "step": 1348
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.0244211583302405,
+      "learning_rate": 0.0009602431550454442,
+      "loss": 3.8066,
+      "step": 1349
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8563337985715797,
+      "learning_rate": 0.0009601705582704348,
+      "loss": 3.869,
+      "step": 1350
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.198118512910373,
+      "learning_rate": 0.0009600978980236683,
+      "loss": 3.9266,
+      "step": 1351
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.848668418666416,
+      "learning_rate": 0.0009600251743151672,
+      "loss": 3.9361,
+      "step": 1352
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.7900897168160965,
+      "learning_rate": 0.0009599523871549621,
+      "loss": 3.8051,
+      "step": 1353
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.8324642499920368,
+      "learning_rate": 0.0009598795365530928,
+      "loss": 3.7465,
+      "step": 1354
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.8412680673522067,
+      "learning_rate": 0.0009598066225196074,
+      "loss": 3.9093,
+      "step": 1355
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9728173637673643,
+      "learning_rate": 0.0009597336450645633,
+      "loss": 3.7059,
+      "step": 1356
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.876193662132438,
+      "learning_rate": 0.000959660604198026,
+      "loss": 3.9082,
+      "step": 1357
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.7655315804375767,
+      "learning_rate": 0.0009595874999300703,
+      "loss": 3.9181,
+      "step": 1358
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.853681014310914,
+      "learning_rate": 0.0009595143322707795,
+      "loss": 3.9391,
+      "step": 1359
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.7964059368770235,
+      "learning_rate": 0.0009594411012302459,
+      "loss": 3.5117,
+      "step": 1360
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.8580626302316371,
+      "learning_rate": 0.0009593678068185701,
+      "loss": 3.6968,
+      "step": 1361
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.087579312729565,
+      "learning_rate": 0.0009592944490458614,
+      "loss": 3.8156,
+      "step": 1362
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.1299281768825828,
+      "learning_rate": 0.0009592210279222386,
+      "loss": 4.2495,
+      "step": 1363
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9640799539214135,
+      "learning_rate": 0.0009591475434578286,
+      "loss": 4.0634,
+      "step": 1364
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.8668561059440153,
+      "learning_rate": 0.0009590739956627671,
+      "loss": 3.753,
+      "step": 1365
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.7885281899151589,
+      "learning_rate": 0.0009590003845471987,
+      "loss": 3.7043,
+      "step": 1366
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9261394854150734,
+      "learning_rate": 0.0009589267101212764,
+      "loss": 3.6236,
+      "step": 1367
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9429537580988815,
+      "learning_rate": 0.0009588529723951625,
+      "loss": 3.86,
+      "step": 1368
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.8598241321518687,
+      "learning_rate": 0.0009587791713790276,
+      "loss": 3.9132,
+      "step": 1369
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.860429607143447,
+      "learning_rate": 0.0009587053070830512,
+      "loss": 3.8315,
+      "step": 1370
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9738170039946603,
+      "learning_rate": 0.0009586313795174213,
+      "loss": 3.8584,
+      "step": 1371
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9401436272043237,
+      "learning_rate": 0.0009585573886923349,
+      "loss": 4.0028,
+      "step": 1372
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.8186962896664751,
+      "learning_rate": 0.0009584833346179977,
+      "loss": 3.9204,
+      "step": 1373
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.2556251541233945,
+      "learning_rate": 0.000958409217304624,
+      "loss": 3.8679,
+      "step": 1374
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.7697209464479603,
+      "learning_rate": 0.0009583350367624366,
+      "loss": 3.7894,
+      "step": 1375
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 9.507549994913516,
+      "learning_rate": 0.0009582607930016678,
+      "loss": 3.9594,
+      "step": 1376
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9532645638863354,
+      "learning_rate": 0.0009581864860325577,
+      "loss": 3.865,
+      "step": 1377
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9623082091869931,
+      "learning_rate": 0.0009581121158653558,
+      "loss": 4.1883,
+      "step": 1378
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9040667625633498,
+      "learning_rate": 0.0009580376825103199,
+      "loss": 3.776,
+      "step": 1379
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.0746365056026415,
+      "learning_rate": 0.0009579631859777167,
+      "loss": 4.1083,
+      "step": 1380
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9437937263199113,
+      "learning_rate": 0.0009578886262778214,
+      "loss": 3.7122,
+      "step": 1381
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.116488444653678,
+      "learning_rate": 0.0009578140034209185,
+      "loss": 3.7763,
+      "step": 1382
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9142975430367627,
+      "learning_rate": 0.0009577393174173004,
+      "loss": 3.8347,
+      "step": 1383
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9263977272223054,
+      "learning_rate": 0.0009576645682772689,
+      "loss": 3.8793,
+      "step": 1384
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.8145280008171305,
+      "learning_rate": 0.0009575897560111339,
+      "loss": 3.7852,
+      "step": 1385
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.8546542710397201,
+      "learning_rate": 0.0009575148806292146,
+      "loss": 3.7826,
+      "step": 1386
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.7778925084507263,
+      "learning_rate": 0.0009574399421418388,
+      "loss": 3.8225,
+      "step": 1387
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.8519461698431795,
+      "learning_rate": 0.0009573649405593422,
+      "loss": 3.8641,
+      "step": 1388
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9856506255186301,
+      "learning_rate": 0.0009572898758920704,
+      "loss": 3.6021,
+      "step": 1389
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.8880775624844572,
+      "learning_rate": 0.000957214748150377,
+      "loss": 3.901,
+      "step": 1390
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.88782666091247,
+      "learning_rate": 0.0009571395573446242,
+      "loss": 3.8895,
+      "step": 1391
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.8207024024884569,
+      "learning_rate": 0.0009570643034851835,
+      "loss": 3.7878,
+      "step": 1392
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.864460529218865,
+      "learning_rate": 0.0009569889865824345,
+      "loss": 4.1187,
+      "step": 1393
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.895396743529384,
+      "learning_rate": 0.0009569136066467659,
+      "loss": 3.904,
+      "step": 1394
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.8227959276072406,
+      "learning_rate": 0.0009568381636885747,
+      "loss": 3.8694,
+      "step": 1395
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9679815361079878,
+      "learning_rate": 0.0009567626577182671,
+      "loss": 3.8404,
+      "step": 1396
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.8648004126631317,
+      "learning_rate": 0.0009566870887462573,
+      "loss": 3.9228,
+      "step": 1397
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.899389022610693,
+      "learning_rate": 0.0009566114567829691,
+      "loss": 3.6404,
+      "step": 1398
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.8959860816301616,
+      "learning_rate": 0.0009565357618388342,
+      "loss": 3.7719,
+      "step": 1399
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.094924359055759,
+      "learning_rate": 0.0009564600039242932,
+      "loss": 3.7904,
+      "step": 1400
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.8446670399193721,
+      "learning_rate": 0.0009563841830497957,
+      "loss": 3.9743,
+      "step": 1401
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.8958998969059287,
+      "learning_rate": 0.0009563082992257996,
+      "loss": 3.9753,
+      "step": 1402
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9084923371191541,
+      "learning_rate": 0.0009562323524627716,
+      "loss": 3.8617,
+      "step": 1403
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9136328753715062,
+      "learning_rate": 0.0009561563427711872,
+      "loss": 3.9676,
+      "step": 1404
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 4.891810412391087,
+      "learning_rate": 0.0009560802701615304,
+      "loss": 3.8785,
+      "step": 1405
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.7557373806901146,
+      "learning_rate": 0.0009560041346442941,
+      "loss": 3.7352,
+      "step": 1406
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.8049526231273595,
+      "learning_rate": 0.0009559279362299796,
+      "loss": 3.7288,
+      "step": 1407
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.7348808034774211,
+      "learning_rate": 0.0009558516749290971,
+      "loss": 3.8559,
+      "step": 1408
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.8832127608214954,
+      "learning_rate": 0.0009557753507521653,
+      "loss": 3.6959,
+      "step": 1409
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.1720270642285266,
+      "learning_rate": 0.0009556989637097118,
+      "loss": 3.7822,
+      "step": 1410
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.8337517554933296,
+      "learning_rate": 0.0009556225138122727,
+      "loss": 3.9447,
+      "step": 1411
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9227469288336163,
+      "learning_rate": 0.0009555460010703927,
+      "loss": 3.828,
+      "step": 1412
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.8816760042677255,
+      "learning_rate": 0.0009554694254946252,
+      "loss": 3.743,
+      "step": 1413
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.8362884977986006,
+      "learning_rate": 0.0009553927870955327,
+      "loss": 3.6962,
+      "step": 1414
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.8967808253010532,
+      "learning_rate": 0.0009553160858836858,
+      "loss": 3.7658,
+      "step": 1415
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9110474357833716,
+      "learning_rate": 0.000955239321869664,
+      "loss": 3.8064,
+      "step": 1416
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.8845506805266313,
+      "learning_rate": 0.0009551624950640552,
+      "loss": 3.6623,
+      "step": 1417
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.8192694046625639,
+      "learning_rate": 0.0009550856054774566,
+      "loss": 3.8059,
+      "step": 1418
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.1541944222905989,
+      "learning_rate": 0.0009550086531204733,
+      "loss": 3.9599,
+      "step": 1419
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.0772825961084247,
+      "learning_rate": 0.0009549316380037196,
+      "loss": 3.7892,
+      "step": 1420
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9651778558174329,
+      "learning_rate": 0.0009548545601378183,
+      "loss": 4.0461,
+      "step": 1421
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.149649039566938,
+      "learning_rate": 0.0009547774195334007,
+      "loss": 3.9114,
+      "step": 1422
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.8375106345677006,
+      "learning_rate": 0.0009547002162011069,
+      "loss": 3.9692,
+      "step": 1423
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.8539130174313645,
+      "learning_rate": 0.0009546229501515856,
+      "loss": 3.7928,
+      "step": 1424
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.8292832259062058,
+      "learning_rate": 0.0009545456213954944,
+      "loss": 3.8811,
+      "step": 1425
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9122768600395526,
+      "learning_rate": 0.000954468229943499,
+      "loss": 3.9504,
+      "step": 1426
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.8427279401605012,
+      "learning_rate": 0.0009543907758062742,
+      "loss": 3.9653,
+      "step": 1427
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9718384380069693,
+      "learning_rate": 0.0009543132589945034,
+      "loss": 3.8199,
+      "step": 1428
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.388479385024393,
+      "learning_rate": 0.0009542356795188786,
+      "loss": 3.6234,
+      "step": 1429
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.1135206294121325,
+      "learning_rate": 0.0009541580373901002,
+      "loss": 3.6418,
+      "step": 1430
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.0600352743282997,
+      "learning_rate": 0.0009540803326188777,
+      "loss": 3.6644,
+      "step": 1431
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.8970042100748042,
+      "learning_rate": 0.0009540025652159288,
+      "loss": 3.8966,
+      "step": 1432
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.076806394421253,
+      "learning_rate": 0.0009539247351919802,
+      "loss": 4.1181,
+      "step": 1433
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9505175053075758,
+      "learning_rate": 0.0009538468425577669,
+      "loss": 3.9308,
+      "step": 1434
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.8641169394011997,
+      "learning_rate": 0.0009537688873240327,
+      "loss": 4.007,
+      "step": 1435
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9268405539347068,
+      "learning_rate": 0.0009536908695015303,
+      "loss": 3.9137,
+      "step": 1436
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.8592229175334658,
+      "learning_rate": 0.0009536127891010205,
+      "loss": 3.9452,
+      "step": 1437
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 100.03254640056228,
+      "learning_rate": 0.000953534646133273,
+      "loss": 3.7717,
+      "step": 1438
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.097820931106454,
+      "learning_rate": 0.0009534564406090664,
+      "loss": 3.696,
+      "step": 1439
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.9316812179864764,
+      "learning_rate": 0.0009533781725391872,
+      "loss": 3.914,
+      "step": 1440
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.8547042079933125,
+      "learning_rate": 0.0009532998419344316,
+      "loss": 3.985,
+      "step": 1441
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.524241479048602,
+      "learning_rate": 0.0009532214488056032,
+      "loss": 3.9608,
+      "step": 1442
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.3782703551829998,
+      "learning_rate": 0.0009531429931635154,
+      "loss": 3.9929,
+      "step": 1443
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.139360157938609,
+      "learning_rate": 0.0009530644750189892,
+      "loss": 4.2135,
+      "step": 1444
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.165247443753805,
+      "learning_rate": 0.0009529858943828548,
+      "loss": 3.8277,
+      "step": 1445
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.2065276610782492,
+      "learning_rate": 0.0009529072512659512,
+      "loss": 3.8855,
+      "step": 1446
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.461429428752609,
+      "learning_rate": 0.0009528285456791253,
+      "loss": 4.1549,
+      "step": 1447
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.2448386635813538,
+      "learning_rate": 0.0009527497776332334,
+      "loss": 4.1072,
+      "step": 1448
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.6769539626506422,
+      "learning_rate": 0.0009526709471391397,
+      "loss": 4.1297,
+      "step": 1449
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.2687245786765955,
+      "learning_rate": 0.0009525920542077176,
+      "loss": 4.1963,
+      "step": 1450
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.4414397569336557,
+      "learning_rate": 0.0009525130988498489,
+      "loss": 3.8882,
+      "step": 1451
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.5153911338165331,
+      "learning_rate": 0.0009524340810764237,
+      "loss": 4.0018,
+      "step": 1452
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 4.436535378393621,
+      "learning_rate": 0.0009523550008983413,
+      "loss": 4.0123,
+      "step": 1453
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.318419462297184,
+      "learning_rate": 0.0009522758583265092,
+      "loss": 3.9551,
+      "step": 1454
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.221249356173666,
+      "learning_rate": 0.0009521966533718436,
+      "loss": 4.0276,
+      "step": 1455
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 5.687136138333517,
+      "learning_rate": 0.0009521173860452695,
+      "loss": 3.8629,
+      "step": 1456
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.346346554399234,
+      "learning_rate": 0.0009520380563577198,
+      "loss": 3.9149,
+      "step": 1457
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 4.086625604690991,
+      "learning_rate": 0.000951958664320137,
+      "loss": 4.0661,
+      "step": 1458
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.4721430727683573,
+      "learning_rate": 0.0009518792099434717,
+      "loss": 4.1447,
+      "step": 1459
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.2976886361092492,
+      "learning_rate": 0.0009517996932386827,
+      "loss": 4.0543,
+      "step": 1460
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.6697103624943264,
+      "learning_rate": 0.0009517201142167385,
+      "loss": 3.989,
+      "step": 1461
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 4.0215592848270525,
+      "learning_rate": 0.0009516404728886148,
+      "loss": 4.16,
+      "step": 1462
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.4697379135773903,
+      "learning_rate": 0.000951560769265297,
+      "loss": 3.8362,
+      "step": 1463
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.1524336185141049,
+      "learning_rate": 0.0009514810033577786,
+      "loss": 3.7538,
+      "step": 1464
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.0493847440526527,
+      "learning_rate": 0.0009514011751770618,
+      "loss": 3.9116,
+      "step": 1465
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.23190628404012,
+      "learning_rate": 0.0009513212847341573,
+      "loss": 4.0073,
+      "step": 1466
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.6248181854318813,
+      "learning_rate": 0.0009512413320400847,
+      "loss": 3.8487,
+      "step": 1467
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.2971342296501958,
+      "learning_rate": 0.0009511613171058717,
+      "loss": 3.9764,
+      "step": 1468
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.445263290007701,
+      "learning_rate": 0.000951081239942555,
+      "loss": 3.9668,
+      "step": 1469
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.4571797461658378,
+      "learning_rate": 0.0009510011005611796,
+      "loss": 3.9339,
+      "step": 1470
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.2450298431296896,
+      "learning_rate": 0.0009509208989727992,
+      "loss": 4.057,
+      "step": 1471
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.237546372047046,
+      "learning_rate": 0.000950840635188476,
+      "loss": 4.0864,
+      "step": 1472
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.4929855017389642,
+      "learning_rate": 0.0009507603092192812,
+      "loss": 3.9271,
+      "step": 1473
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.3189996569272737,
+      "learning_rate": 0.000950679921076294,
+      "loss": 4.1884,
+      "step": 1474
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.99328213896486,
+      "learning_rate": 0.0009505994707706023,
+      "loss": 3.9981,
+      "step": 1475
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.013478153612559,
+      "learning_rate": 0.000950518958313303,
+      "loss": 4.1527,
+      "step": 1476
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.9889837202721596,
+      "learning_rate": 0.0009504383837155008,
+      "loss": 3.9588,
+      "step": 1477
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.190983343959375,
+      "learning_rate": 0.0009503577469883098,
+      "loss": 3.8343,
+      "step": 1478
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.1477902413587535,
+      "learning_rate": 0.0009502770481428525,
+      "loss": 3.9696,
+      "step": 1479
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.1100575455524337,
+      "learning_rate": 0.000950196287190259,
+      "loss": 3.9837,
+      "step": 1480
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.3399869539582643,
+      "learning_rate": 0.0009501154641416695,
+      "loss": 4.1498,
+      "step": 1481
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.0409958644089108,
+      "learning_rate": 0.0009500345790082317,
+      "loss": 3.9849,
+      "step": 1482
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.2192002708950527,
+      "learning_rate": 0.0009499536318011019,
+      "loss": 3.8925,
+      "step": 1483
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.1164133077716576,
+      "learning_rate": 0.0009498726225314458,
+      "loss": 4.1419,
+      "step": 1484
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.9516042613942988,
+      "learning_rate": 0.0009497915512104367,
+      "loss": 4.1061,
+      "step": 1485
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.0761513596116465,
+      "learning_rate": 0.0009497104178492568,
+      "loss": 3.9046,
+      "step": 1486
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.1571756675128302,
+      "learning_rate": 0.0009496292224590973,
+      "loss": 3.9149,
+      "step": 1487
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.2972725958310365,
+      "learning_rate": 0.000949547965051157,
+      "loss": 4.124,
+      "step": 1488
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.872410383831955,
+      "learning_rate": 0.0009494666456366441,
+      "loss": 3.762,
+      "step": 1489
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.8477867956326395,
+      "learning_rate": 0.0009493852642267751,
+      "loss": 4.1682,
+      "step": 1490
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.0949012112852705,
+      "learning_rate": 0.0009493038208327749,
+      "loss": 3.8855,
+      "step": 1491
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.4820274398500546,
+      "learning_rate": 0.0009492223154658773,
+      "loss": 4.0535,
+      "step": 1492
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.0338407045086,
+      "learning_rate": 0.0009491407481373241,
+      "loss": 3.9,
+      "step": 1493
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.2365497846020905,
+      "learning_rate": 0.0009490591188583661,
+      "loss": 3.9303,
+      "step": 1494
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.016761854506582,
+      "learning_rate": 0.0009489774276402625,
+      "loss": 3.6776,
+      "step": 1495
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.2383822971701957,
+      "learning_rate": 0.0009488956744942811,
+      "loss": 3.9083,
+      "step": 1496
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.3734595514182755,
+      "learning_rate": 0.0009488138594316982,
+      "loss": 3.8552,
+      "step": 1497
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.1644271060663924,
+      "learning_rate": 0.0009487319824637983,
+      "loss": 3.9592,
+      "step": 1498
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.0977849033455462,
+      "learning_rate": 0.0009486500436018752,
+      "loss": 4.2405,
+      "step": 1499
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.2055573402943967,
+      "learning_rate": 0.0009485680428572308,
+      "loss": 3.9396,
+      "step": 1500
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.1374716930789182,
+      "learning_rate": 0.0009484859802411751,
+      "loss": 3.8888,
+      "step": 1501
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.296678328768621,
+      "learning_rate": 0.0009484038557650274,
+      "loss": 3.9994,
+      "step": 1502
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.055013626944651,
+      "learning_rate": 0.0009483216694401152,
+      "loss": 3.8271,
+      "step": 1503
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.0186429480108605,
+      "learning_rate": 0.0009482394212777745,
+      "loss": 3.9758,
+      "step": 1504
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.06816243495957,
+      "learning_rate": 0.0009481571112893498,
+      "loss": 4.2682,
+      "step": 1505
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.2746938476499259,
+      "learning_rate": 0.0009480747394861944,
+      "loss": 3.8056,
+      "step": 1506
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.3143444691467647,
+      "learning_rate": 0.0009479923058796695,
+      "loss": 3.9706,
+      "step": 1507
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 9.560587365045166,
+      "learning_rate": 0.0009479098104811459,
+      "loss": 4.1207,
+      "step": 1508
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.9078763675290515,
+      "learning_rate": 0.0009478272533020016,
+      "loss": 3.8487,
+      "step": 1509
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.4529298475723176,
+      "learning_rate": 0.0009477446343536241,
+      "loss": 4.0019,
+      "step": 1510
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.0733672459692003,
+      "learning_rate": 0.0009476619536474091,
+      "loss": 3.9246,
+      "step": 1511
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.9352742201273678,
+      "learning_rate": 0.0009475792111947607,
+      "loss": 4.078,
+      "step": 1512
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.265932280169156,
+      "learning_rate": 0.0009474964070070919,
+      "loss": 4.1204,
+      "step": 1513
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.023218237649011,
+      "learning_rate": 0.0009474135410958239,
+      "loss": 3.9499,
+      "step": 1514
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.3532599547223871,
+      "learning_rate": 0.0009473306134723862,
+      "loss": 3.9292,
+      "step": 1515
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.2315722301854204,
+      "learning_rate": 0.0009472476241482173,
+      "loss": 3.9577,
+      "step": 1516
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 24.04739000407346,
+      "learning_rate": 0.000947164573134764,
+      "loss": 4.0181,
+      "step": 1517
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.9936336517979658,
+      "learning_rate": 0.0009470814604434816,
+      "loss": 3.9791,
+      "step": 1518
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.217222741629316,
+      "learning_rate": 0.000946998286085834,
+      "loss": 3.852,
+      "step": 1519
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.51117778184176,
+      "learning_rate": 0.0009469150500732932,
+      "loss": 4.096,
+      "step": 1520
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.5809599260014615,
+      "learning_rate": 0.0009468317524173402,
+      "loss": 4.4122,
+      "step": 1521
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.5046749860915247,
+      "learning_rate": 0.0009467483931294644,
+      "loss": 4.1132,
+      "step": 1522
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.3334451624224986,
+      "learning_rate": 0.0009466649722211635,
+      "loss": 4.2068,
+      "step": 1523
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.2237324082068999,
+      "learning_rate": 0.000946581489703944,
+      "loss": 4.1176,
+      "step": 1524
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.219661009942738,
+      "learning_rate": 0.0009464979455893205,
+      "loss": 4.1237,
+      "step": 1525
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.146067514462624,
+      "learning_rate": 0.0009464143398888166,
+      "loss": 4.1596,
+      "step": 1526
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.9523317630879572,
+      "learning_rate": 0.0009463306726139638,
+      "loss": 3.9609,
+      "step": 1527
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.7650553678302636,
+      "learning_rate": 0.0009462469437763026,
+      "loss": 3.9253,
+      "step": 1528
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.2136665087837992,
+      "learning_rate": 0.0009461631533873818,
+      "loss": 4.1455,
+      "step": 1529
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.9704508737243392,
+      "learning_rate": 0.0009460793014587585,
+      "loss": 4.0206,
+      "step": 1530
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.9117749646748721,
+      "learning_rate": 0.0009459953880019987,
+      "loss": 4.2635,
+      "step": 1531
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.9277265592837262,
+      "learning_rate": 0.0009459114130286766,
+      "loss": 4.0095,
+      "step": 1532
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.9304485127432531,
+      "learning_rate": 0.0009458273765503749,
+      "loss": 3.9747,
+      "step": 1533
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.8501982683726252,
+      "learning_rate": 0.0009457432785786848,
+      "loss": 4.0774,
+      "step": 1534
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.9991426301048467,
+      "learning_rate": 0.0009456591191252061,
+      "loss": 3.8427,
+      "step": 1535
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.9804655560531712,
+      "learning_rate": 0.0009455748982015468,
+      "loss": 4.2069,
+      "step": 1536
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.5784437476664313,
+      "learning_rate": 0.0009454906158193239,
+      "loss": 4.0708,
+      "step": 1537
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.193697252983465,
+      "learning_rate": 0.0009454062719901624,
+      "loss": 3.789,
+      "step": 1538
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.8229813174629523,
+      "learning_rate": 0.0009453218667256958,
+      "loss": 4.0608,
+      "step": 1539
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 9.674128229081756,
+      "learning_rate": 0.0009452374000375664,
+      "loss": 3.8153,
+      "step": 1540
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.1564343287849521,
+      "learning_rate": 0.0009451528719374245,
+      "loss": 4.0524,
+      "step": 1541
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.9500107507142022,
+      "learning_rate": 0.0009450682824369294,
+      "loss": 4.3137,
+      "step": 1542
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.3586169269801127,
+      "learning_rate": 0.0009449836315477485,
+      "loss": 4.1483,
+      "step": 1543
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.193557500630441,
+      "learning_rate": 0.0009448989192815578,
+      "loss": 3.9929,
+      "step": 1544
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.9576633356336297,
+      "learning_rate": 0.0009448141456500416,
+      "loss": 3.7209,
+      "step": 1545
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.9564482486986808,
+      "learning_rate": 0.0009447293106648931,
+      "loss": 3.919,
+      "step": 1546
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.8284744779787477,
+      "learning_rate": 0.0009446444143378134,
+      "loss": 3.8823,
+      "step": 1547
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.0445801764427596,
+      "learning_rate": 0.0009445594566805126,
+      "loss": 4.0477,
+      "step": 1548
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.8123353106551443,
+      "learning_rate": 0.0009444744377047088,
+      "loss": 3.8422,
+      "step": 1549
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.0680378005644848,
+      "learning_rate": 0.0009443893574221286,
+      "loss": 3.9865,
+      "step": 1550
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.810638326995603,
+      "learning_rate": 0.0009443042158445074,
+      "loss": 3.9771,
+      "step": 1551
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.022334654841791,
+      "learning_rate": 0.000944219012983589,
+      "loss": 3.9238,
+      "step": 1552
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.8558674804976343,
+      "learning_rate": 0.0009441337488511252,
+      "loss": 4.0984,
+      "step": 1553
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.1785105958799997,
+      "learning_rate": 0.0009440484234588766,
+      "loss": 4.0524,
+      "step": 1554
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.7951909862614366,
+      "learning_rate": 0.0009439630368186125,
+      "loss": 3.8537,
+      "step": 1555
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.879575418996428,
+      "learning_rate": 0.0009438775889421102,
+      "loss": 4.021,
+      "step": 1556
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.9481200475103162,
+      "learning_rate": 0.0009437920798411554,
+      "loss": 4.2121,
+      "step": 1557
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.7959816321982579,
+      "learning_rate": 0.0009437065095275429,
+      "loss": 3.8938,
+      "step": 1558
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.8096393262236952,
+      "learning_rate": 0.0009436208780130751,
+      "loss": 3.9656,
+      "step": 1559
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 5.167593975881696,
+      "learning_rate": 0.0009435351853095633,
+      "loss": 4.2262,
+      "step": 1560
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.972610552078697,
+      "learning_rate": 0.0009434494314288273,
+      "loss": 3.8205,
+      "step": 1561
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.1002194924816127,
+      "learning_rate": 0.0009433636163826951,
+      "loss": 3.8968,
+      "step": 1562
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.0283139853241532,
+      "learning_rate": 0.0009432777401830033,
+      "loss": 4.0238,
+      "step": 1563
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 4.313783840843981,
+      "learning_rate": 0.0009431918028415969,
+      "loss": 4.0874,
+      "step": 1564
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.7397377527852067,
+      "learning_rate": 0.0009431058043703293,
+      "loss": 3.674,
+      "step": 1565
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.1820265790251154,
+      "learning_rate": 0.0009430197447810625,
+      "loss": 4.056,
+      "step": 1566
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.8199724368057919,
+      "learning_rate": 0.0009429336240856662,
+      "loss": 4.1426,
+      "step": 1567
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.7419548672209354,
+      "learning_rate": 0.00094284744229602,
+      "loss": 3.9271,
+      "step": 1568
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.8815043237422977,
+      "learning_rate": 0.0009427611994240104,
+      "loss": 4.0304,
+      "step": 1569
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.7904006837456864,
+      "learning_rate": 0.0009426748954815332,
+      "loss": 4.0589,
+      "step": 1570
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.8735374751290628,
+      "learning_rate": 0.0009425885304804922,
+      "loss": 3.7514,
+      "step": 1571
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.9342411924816054,
+      "learning_rate": 0.0009425021044328,
+      "loss": 3.9183,
+      "step": 1572
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.8715894308471137,
+      "learning_rate": 0.0009424156173503772,
+      "loss": 4.1754,
+      "step": 1573
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.7973900423560782,
+      "learning_rate": 0.0009423290692451534,
+      "loss": 4.226,
+      "step": 1574
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.76011519970482,
+      "learning_rate": 0.000942242460129066,
+      "loss": 4.0764,
+      "step": 1575
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.9107290131367709,
+      "learning_rate": 0.0009421557900140612,
+      "loss": 4.1168,
+      "step": 1576
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.849593226471349,
+      "learning_rate": 0.0009420690589120932,
+      "loss": 3.9129,
+      "step": 1577
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.0000657201654148,
+      "learning_rate": 0.0009419822668351255,
+      "loss": 3.5998,
+      "step": 1578
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.7279339698657166,
+      "learning_rate": 0.0009418954137951288,
+      "loss": 3.7973,
+      "step": 1579
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.8413751506774984,
+      "learning_rate": 0.000941808499804083,
+      "loss": 4.0201,
+      "step": 1580
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.8181680616477367,
+      "learning_rate": 0.0009417215248739764,
+      "loss": 4.0575,
+      "step": 1581
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.1747989231598868,
+      "learning_rate": 0.0009416344890168054,
+      "loss": 3.9116,
+      "step": 1582
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.421598318262801,
+      "learning_rate": 0.000941547392244575,
+      "loss": 4.0196,
+      "step": 1583
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.7547175061390377,
+      "learning_rate": 0.0009414602345692984,
+      "loss": 3.7012,
+      "step": 1584
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.5826797581345966,
+      "learning_rate": 0.0009413730160029974,
+      "loss": 3.8621,
+      "step": 1585
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.1070260331650765,
+      "learning_rate": 0.0009412857365577023,
+      "loss": 4.1316,
+      "step": 1586
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.3883932107405876,
+      "learning_rate": 0.0009411983962454515,
+      "loss": 3.9567,
+      "step": 1587
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.936034317123037,
+      "learning_rate": 0.0009411109950782919,
+      "loss": 3.6744,
+      "step": 1588
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.9210368930577055,
+      "learning_rate": 0.0009410235330682788,
+      "loss": 3.7847,
+      "step": 1589
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.7362103916198269,
+      "learning_rate": 0.0009409360102274761,
+      "loss": 3.7751,
+      "step": 1590
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.8055415788013954,
+      "learning_rate": 0.0009408484265679558,
+      "loss": 3.8055,
+      "step": 1591
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.8571256899522368,
+      "learning_rate": 0.0009407607821017983,
+      "loss": 3.8728,
+      "step": 1592
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.8064263379105193,
+      "learning_rate": 0.0009406730768410927,
+      "loss": 3.8357,
+      "step": 1593
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.8946668171330434,
+      "learning_rate": 0.0009405853107979361,
+      "loss": 3.8633,
+      "step": 1594
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.9536201270201781,
+      "learning_rate": 0.0009404974839844341,
+      "loss": 3.9501,
+      "step": 1595
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.8312478199505631,
+      "learning_rate": 0.0009404095964127008,
+      "loss": 3.8984,
+      "step": 1596
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.078526366735139,
+      "learning_rate": 0.0009403216480948589,
+      "loss": 4.0352,
+      "step": 1597
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.3225514410492496,
+      "learning_rate": 0.0009402336390430388,
+      "loss": 3.8265,
+      "step": 1598
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.8197474651758992,
+      "learning_rate": 0.0009401455692693798,
+      "loss": 3.9119,
+      "step": 1599
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.92057023464428,
+      "learning_rate": 0.0009400574387860294,
+      "loss": 4.0379,
+      "step": 1600
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.8132215355699012,
+      "learning_rate": 0.0009399692476051436,
+      "loss": 3.7861,
+      "step": 1601
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.833548101742673,
+      "learning_rate": 0.0009398809957388868,
+      "loss": 3.8447,
+      "step": 1602
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.8332069686852757,
+      "learning_rate": 0.0009397926831994314,
+      "loss": 4.0532,
+      "step": 1603
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.9185271456748192,
+      "learning_rate": 0.0009397043099989587,
+      "loss": 3.8971,
+      "step": 1604
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.9425320342615834,
+      "learning_rate": 0.0009396158761496577,
+      "loss": 3.9502,
+      "step": 1605
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.404888821930948,
+      "learning_rate": 0.0009395273816637267,
+      "loss": 4.0818,
+      "step": 1606
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.8844218704775557,
+      "learning_rate": 0.0009394388265533713,
+      "loss": 4.0767,
+      "step": 1607
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.8528465410415034,
+      "learning_rate": 0.0009393502108308064,
+      "loss": 3.9177,
+      "step": 1608
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.7899627795532184,
+      "learning_rate": 0.0009392615345082547,
+      "loss": 3.9613,
+      "step": 1609
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.9213105646111837,
+      "learning_rate": 0.0009391727975979474,
+      "loss": 4.0821,
+      "step": 1610
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.8460076098727417,
+      "learning_rate": 0.0009390840001121239,
+      "loss": 3.8884,
+      "step": 1611
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.9467796437566335,
+      "learning_rate": 0.0009389951420630325,
+      "loss": 3.9064,
+      "step": 1612
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.8273534642521473,
+      "learning_rate": 0.0009389062234629292,
+      "loss": 3.8386,
+      "step": 1613
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.8297470549250323,
+      "learning_rate": 0.0009388172443240788,
+      "loss": 3.7848,
+      "step": 1614
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.7988434522509076,
+      "learning_rate": 0.0009387282046587539,
+      "loss": 3.7489,
+      "step": 1615
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.7861922341286257,
+      "learning_rate": 0.0009386391044792363,
+      "loss": 3.9704,
+      "step": 1616
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.6713201587247575,
+      "learning_rate": 0.0009385499437978153,
+      "loss": 3.8542,
+      "step": 1617
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.7783170485905523,
+      "learning_rate": 0.0009384607226267891,
+      "loss": 4.139,
+      "step": 1618
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.7969057870169202,
+      "learning_rate": 0.0009383714409784643,
+      "loss": 3.6903,
+      "step": 1619
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.8276175065589113,
+      "learning_rate": 0.000938282098865155,
+      "loss": 3.9059,
+      "step": 1620
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.8181316850078265,
+      "learning_rate": 0.0009381926962991847,
+      "loss": 3.9939,
+      "step": 1621
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.7002785078678915,
+      "learning_rate": 0.0009381032332928847,
+      "loss": 3.7722,
+      "step": 1622
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.9223497190758715,
+      "learning_rate": 0.0009380137098585946,
+      "loss": 4.1211,
+      "step": 1623
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.7529287177545222,
+      "learning_rate": 0.0009379241260086626,
+      "loss": 3.9051,
+      "step": 1624
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.8076963284551374,
+      "learning_rate": 0.0009378344817554449,
+      "loss": 3.8456,
+      "step": 1625
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.7686064022008927,
+      "learning_rate": 0.0009377447771113065,
+      "loss": 3.8874,
+      "step": 1626
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.7493614990883468,
+      "learning_rate": 0.0009376550120886203,
+      "loss": 3.9097,
+      "step": 1627
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.8736190721445515,
+      "learning_rate": 0.0009375651866997674,
+      "loss": 3.7398,
+      "step": 1628
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.8490897050280988,
+      "learning_rate": 0.0009374753009571379,
+      "loss": 3.8893,
+      "step": 1629
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.7491797176064796,
+      "learning_rate": 0.0009373853548731297,
+      "loss": 3.8386,
+      "step": 1630
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.1272079851857126,
+      "learning_rate": 0.000937295348460149,
+      "loss": 4.0181,
+      "step": 1631
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.7333622769505543,
+      "learning_rate": 0.0009372052817306106,
+      "loss": 3.8798,
+      "step": 1632
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.9629933054130179,
+      "learning_rate": 0.0009371151546969376,
+      "loss": 3.8187,
+      "step": 1633
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.753777036520204,
+      "learning_rate": 0.0009370249673715611,
+      "loss": 3.9808,
+      "step": 1634
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.9179589990776899,
+      "learning_rate": 0.0009369347197669207,
+      "loss": 3.9414,
+      "step": 1635
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.8206318918357529,
+      "learning_rate": 0.0009368444118954646,
+      "loss": 3.9135,
+      "step": 1636
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.9581877896806545,
+      "learning_rate": 0.0009367540437696489,
+      "loss": 4.0041,
+      "step": 1637
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.0497348716078656,
+      "learning_rate": 0.0009366636154019381,
+      "loss": 4.0523,
+      "step": 1638
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.8960927057628982,
+      "learning_rate": 0.0009365731268048052,
+      "loss": 4.068,
+      "step": 1639
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.952542194081992,
+      "learning_rate": 0.0009364825779907311,
+      "loss": 3.9961,
+      "step": 1640
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.9146892852264598,
+      "learning_rate": 0.0009363919689722056,
+      "loss": 3.8575,
+      "step": 1641
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.9114499760732222,
+      "learning_rate": 0.0009363012997617264,
+      "loss": 3.8914,
+      "step": 1642
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.829714448292901,
+      "learning_rate": 0.0009362105703717994,
+      "loss": 3.9217,
+      "step": 1643
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.8082979094520754,
+      "learning_rate": 0.0009361197808149393,
+      "loss": 3.9194,
+      "step": 1644
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.0337575180932346,
+      "learning_rate": 0.0009360289311036688,
+      "loss": 3.8969,
+      "step": 1645
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.4726775104047347,
+      "learning_rate": 0.0009359380212505184,
+      "loss": 3.8682,
+      "step": 1646
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.9743521122422942,
+      "learning_rate": 0.0009358470512680278,
+      "loss": 3.8105,
+      "step": 1647
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.090128331060611,
+      "learning_rate": 0.0009357560211687445,
+      "loss": 3.9929,
+      "step": 1648
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.9535019585618488,
+      "learning_rate": 0.0009356649309652243,
+      "loss": 4.0109,
+      "step": 1649
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 6.797417280233116,
+      "learning_rate": 0.0009355737806700315,
+      "loss": 3.8693,
+      "step": 1650
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.8577025814415281,
+      "learning_rate": 0.0009354825702957383,
+      "loss": 3.9917,
+      "step": 1651
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.8604061744666929,
+      "learning_rate": 0.0009353912998549259,
+      "loss": 3.7423,
+      "step": 1652
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.762141008956544,
+      "learning_rate": 0.0009352999693601827,
+      "loss": 3.7144,
+      "step": 1653
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.7781953832937091,
+      "learning_rate": 0.0009352085788241064,
+      "loss": 3.9071,
+      "step": 1654
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.9058494095421323,
+      "learning_rate": 0.0009351171282593026,
+      "loss": 3.9115,
+      "step": 1655
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.7667131026365983,
+      "learning_rate": 0.0009350256176783847,
+      "loss": 3.7106,
+      "step": 1656
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.780151354407763,
+      "learning_rate": 0.0009349340470939753,
+      "loss": 3.9903,
+      "step": 1657
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.8538957671923801,
+      "learning_rate": 0.0009348424165187049,
+      "loss": 3.9689,
+      "step": 1658
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.0701679972706033,
+      "learning_rate": 0.0009347507259652119,
+      "loss": 3.9973,
+      "step": 1659
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.8249055808082802,
+      "learning_rate": 0.0009346589754461433,
+      "loss": 4.0428,
+      "step": 1660
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.9436817844502606,
+      "learning_rate": 0.0009345671649741545,
+      "loss": 3.9746,
+      "step": 1661
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.0399213961447786,
+      "learning_rate": 0.0009344752945619089,
+      "loss": 3.6444,
+      "step": 1662
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.8929596272267862,
+      "learning_rate": 0.0009343833642220781,
+      "loss": 3.9582,
+      "step": 1663
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.8093334227261171,
+      "learning_rate": 0.0009342913739673424,
+      "loss": 4.0904,
+      "step": 1664
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.9219805722206545,
+      "learning_rate": 0.00093419932381039,
+      "loss": 4.1267,
+      "step": 1665
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.7532023837495229,
+      "learning_rate": 0.0009341072137639175,
+      "loss": 3.8564,
+      "step": 1666
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.7779195152595106,
+      "learning_rate": 0.0009340150438406296,
+      "loss": 3.9052,
+      "step": 1667
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.7457270737522609,
+      "learning_rate": 0.0009339228140532396,
+      "loss": 3.815,
+      "step": 1668
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.7647594144786466,
+      "learning_rate": 0.0009338305244144687,
+      "loss": 4.0877,
+      "step": 1669
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.7021595546337704,
+      "learning_rate": 0.0009337381749370463,
+      "loss": 4.1293,
+      "step": 1670
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.906568690032215,
+      "learning_rate": 0.0009336457656337108,
+      "loss": 4.0644,
+      "step": 1671
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.7012265097192134,
+      "learning_rate": 0.0009335532965172079,
+      "loss": 4.1072,
+      "step": 1672
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.7732403071715452,
+      "learning_rate": 0.0009334607676002919,
+      "loss": 3.9331,
+      "step": 1673
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.7923006863552206,
+      "learning_rate": 0.0009333681788957256,
+      "loss": 3.6242,
+      "step": 1674
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.8032684546739858,
+      "learning_rate": 0.0009332755304162798,
+      "loss": 3.7891,
+      "step": 1675
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.7222815984055483,
+      "learning_rate": 0.0009331828221747335,
+      "loss": 3.7836,
+      "step": 1676
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.9917444213289307,
+      "learning_rate": 0.0009330900541838741,
+      "loss": 4.0377,
+      "step": 1677
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.8171278402653075,
+      "learning_rate": 0.0009329972264564972,
+      "loss": 4.1606,
+      "step": 1678
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.7672392564253656,
+      "learning_rate": 0.0009329043390054066,
+      "loss": 3.9218,
+      "step": 1679
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.7584724642661171,
+      "learning_rate": 0.0009328113918434142,
+      "loss": 4.0897,
+      "step": 1680
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.7724129285953324,
+      "learning_rate": 0.0009327183849833406,
+      "loss": 4.0097,
+      "step": 1681
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.7507744463502709,
+      "learning_rate": 0.0009326253184380141,
+      "loss": 3.821,
+      "step": 1682
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.4669691118757309,
+      "learning_rate": 0.0009325321922202716,
+      "loss": 3.9893,
+      "step": 1683
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.7938763496320073,
+      "learning_rate": 0.0009324390063429578,
+      "loss": 3.9173,
+      "step": 1684
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.7643137269748055,
+      "learning_rate": 0.0009323457608189263,
+      "loss": 3.8917,
+      "step": 1685
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.8344994935118849,
+      "learning_rate": 0.0009322524556610384,
+      "loss": 3.8858,
+      "step": 1686
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.7922495896139837,
+      "learning_rate": 0.0009321590908821635,
+      "loss": 3.855,
+      "step": 1687
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.7318636723197567,
+      "learning_rate": 0.0009320656664951797,
+      "loss": 3.7846,
+      "step": 1688
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.8048885033748879,
+      "learning_rate": 0.0009319721825129734,
+      "loss": 3.9843,
+      "step": 1689
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.6788426578606397,
+      "learning_rate": 0.0009318786389484383,
+      "loss": 3.8998,
+      "step": 1690
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.1904636374596727,
+      "learning_rate": 0.0009317850358144778,
+      "loss": 3.844,
+      "step": 1691
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.8615193138836781,
+      "learning_rate": 0.0009316913731240018,
+      "loss": 3.9595,
+      "step": 1692
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.8742236947980859,
+      "learning_rate": 0.0009315976508899298,
+      "loss": 4.0222,
+      "step": 1693
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.8087282051723653,
+      "learning_rate": 0.0009315038691251887,
+      "loss": 4.2426,
+      "step": 1694
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.7926203782658983,
+      "learning_rate": 0.0009314100278427143,
+      "loss": 4.0534,
+      "step": 1695
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.8359905977572799,
+      "learning_rate": 0.0009313161270554498,
+      "loss": 3.8151,
+      "step": 1696
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 11.604295267328535,
+      "learning_rate": 0.0009312221667763472,
+      "loss": 3.8582,
+      "step": 1697
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.8537436856084206,
+      "learning_rate": 0.0009311281470183667,
+      "loss": 3.9817,
+      "step": 1698
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.8062037709247555,
+      "learning_rate": 0.0009310340677944762,
+      "loss": 3.629,
+      "step": 1699
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.8393436748602189,
+      "learning_rate": 0.0009309399291176524,
+      "loss": 3.7647,
+      "step": 1700
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.8116359873529164,
+      "learning_rate": 0.0009308457310008798,
+      "loss": 3.7836,
+      "step": 1701
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7517490744134788,
+      "learning_rate": 0.0009307514734571514,
+      "loss": 3.6888,
+      "step": 1702
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.8253786834923303,
+      "learning_rate": 0.0009306571564994679,
+      "loss": 3.8984,
+      "step": 1703
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.8015326217350007,
+      "learning_rate": 0.000930562780140839,
+      "loss": 4.1322,
+      "step": 1704
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7858679107150502,
+      "learning_rate": 0.0009304683443942816,
+      "loss": 3.8469,
+      "step": 1705
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.8218888911618002,
+      "learning_rate": 0.0009303738492728216,
+      "loss": 4.2129,
+      "step": 1706
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.843649693780184,
+      "learning_rate": 0.000930279294789493,
+      "loss": 4.0119,
+      "step": 1707
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7257506556096036,
+      "learning_rate": 0.0009301846809573373,
+      "loss": 3.9074,
+      "step": 1708
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.9542468769680683,
+      "learning_rate": 0.000930090007789405,
+      "loss": 3.986,
+      "step": 1709
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7997410615586994,
+      "learning_rate": 0.0009299952752987544,
+      "loss": 4.0993,
+      "step": 1710
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.8987492700026134,
+      "learning_rate": 0.0009299004834984519,
+      "loss": 3.948,
+      "step": 1711
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.8415929862194048,
+      "learning_rate": 0.0009298056324015724,
+      "loss": 4.0483,
+      "step": 1712
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.8261529669197057,
+      "learning_rate": 0.0009297107220211988,
+      "loss": 3.7812,
+      "step": 1713
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.9515629041107165,
+      "learning_rate": 0.0009296157523704223,
+      "loss": 3.9395,
+      "step": 1714
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7455512850288687,
+      "learning_rate": 0.0009295207234623418,
+      "loss": 3.802,
+      "step": 1715
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7662096731200948,
+      "learning_rate": 0.000929425635310065,
+      "loss": 3.8961,
+      "step": 1716
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.855663163393645,
+      "learning_rate": 0.0009293304879267073,
+      "loss": 4.1121,
+      "step": 1717
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.8799383891984581,
+      "learning_rate": 0.0009292352813253926,
+      "loss": 3.7898,
+      "step": 1718
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7783464561415505,
+      "learning_rate": 0.0009291400155192528,
+      "loss": 3.9871,
+      "step": 1719
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.9576272483913795,
+      "learning_rate": 0.0009290446905214281,
+      "loss": 3.9688,
+      "step": 1720
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7261910084645548,
+      "learning_rate": 0.0009289493063450666,
+      "loss": 3.7884,
+      "step": 1721
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7791356841865663,
+      "learning_rate": 0.0009288538630033247,
+      "loss": 3.8566,
+      "step": 1722
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.8966443865678572,
+      "learning_rate": 0.0009287583605093674,
+      "loss": 3.8291,
+      "step": 1723
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7594350827599745,
+      "learning_rate": 0.000928662798876367,
+      "loss": 3.878,
+      "step": 1724
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.6998314455870827,
+      "learning_rate": 0.0009285671781175045,
+      "loss": 3.7937,
+      "step": 1725
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7538798918095796,
+      "learning_rate": 0.000928471498245969,
+      "loss": 3.9407,
+      "step": 1726
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.8949359104559965,
+      "learning_rate": 0.0009283757592749577,
+      "loss": 3.6857,
+      "step": 1727
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.997048592744118,
+      "learning_rate": 0.0009282799612176762,
+      "loss": 3.8139,
+      "step": 1728
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7096870755998815,
+      "learning_rate": 0.0009281841040873376,
+      "loss": 3.9418,
+      "step": 1729
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.4256473461416777,
+      "learning_rate": 0.0009280881878971637,
+      "loss": 4.0744,
+      "step": 1730
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7891220120543624,
+      "learning_rate": 0.0009279922126603846,
+      "loss": 3.7633,
+      "step": 1731
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.8120797148426995,
+      "learning_rate": 0.000927896178390238,
+      "loss": 3.8918,
+      "step": 1732
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.8351171858637474,
+      "learning_rate": 0.0009278000850999699,
+      "loss": 3.7812,
+      "step": 1733
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.8344408713837536,
+      "learning_rate": 0.0009277039328028347,
+      "loss": 3.7804,
+      "step": 1734
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.931313187750291,
+      "learning_rate": 0.0009276077215120949,
+      "loss": 3.9961,
+      "step": 1735
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.1476245919951218,
+      "learning_rate": 0.0009275114512410208,
+      "loss": 4.0585,
+      "step": 1736
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7535830476151243,
+      "learning_rate": 0.000927415122002891,
+      "loss": 3.8156,
+      "step": 1737
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.9351259409107812,
+      "learning_rate": 0.0009273187338109925,
+      "loss": 3.8141,
+      "step": 1738
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.9441794320389927,
+      "learning_rate": 0.0009272222866786201,
+      "loss": 3.8083,
+      "step": 1739
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.1663459174316086,
+      "learning_rate": 0.0009271257806190769,
+      "loss": 3.766,
+      "step": 1740
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7962866347647475,
+      "learning_rate": 0.0009270292156456738,
+      "loss": 3.7229,
+      "step": 1741
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 11.616583877491955,
+      "learning_rate": 0.0009269325917717306,
+      "loss": 3.9633,
+      "step": 1742
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.0458903472976837,
+      "learning_rate": 0.0009268359090105743,
+      "loss": 3.9312,
+      "step": 1743
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7778326497556163,
+      "learning_rate": 0.0009267391673755405,
+      "loss": 3.8247,
+      "step": 1744
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.9819921840205279,
+      "learning_rate": 0.0009266423668799731,
+      "loss": 3.8534,
+      "step": 1745
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.8671874484727672,
+      "learning_rate": 0.0009265455075372237,
+      "loss": 3.8799,
+      "step": 1746
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.8489935037793109,
+      "learning_rate": 0.0009264485893606523,
+      "loss": 4.106,
+      "step": 1747
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.971229627090724,
+      "learning_rate": 0.0009263516123636267,
+      "loss": 3.9358,
+      "step": 1748
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.8045513352178814,
+      "learning_rate": 0.0009262545765595232,
+      "loss": 4.1579,
+      "step": 1749
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.065910225156871,
+      "learning_rate": 0.000926157481961726,
+      "loss": 3.8547,
+      "step": 1750
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.9454332248180906,
+      "learning_rate": 0.0009260603285836276,
+      "loss": 3.8257,
+      "step": 1751
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.9348987501065008,
+      "learning_rate": 0.0009259631164386282,
+      "loss": 3.9301,
+      "step": 1752
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.8073237945709126,
+      "learning_rate": 0.0009258658455401365,
+      "loss": 4.0541,
+      "step": 1753
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.846591835454308,
+      "learning_rate": 0.0009257685159015692,
+      "loss": 3.927,
+      "step": 1754
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.8831930649897933,
+      "learning_rate": 0.0009256711275363509,
+      "loss": 4.0913,
+      "step": 1755
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.8146969547146948,
+      "learning_rate": 0.0009255736804579147,
+      "loss": 4.2306,
+      "step": 1756
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.859883479624131,
+      "learning_rate": 0.0009254761746797013,
+      "loss": 3.9352,
+      "step": 1757
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.791658449959038,
+      "learning_rate": 0.0009253786102151602,
+      "loss": 3.9185,
+      "step": 1758
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7992737629348875,
+      "learning_rate": 0.0009252809870777481,
+      "loss": 3.9275,
+      "step": 1759
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.823155707534069,
+      "learning_rate": 0.0009251833052809304,
+      "loss": 3.857,
+      "step": 1760
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7476570237676693,
+      "learning_rate": 0.0009250855648381805,
+      "loss": 4.0202,
+      "step": 1761
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.8925801737834225,
+      "learning_rate": 0.0009249877657629799,
+      "loss": 3.9271,
+      "step": 1762
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.8735460533612577,
+      "learning_rate": 0.0009248899080688178,
+      "loss": 3.9404,
+      "step": 1763
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.8394179098760687,
+      "learning_rate": 0.0009247919917691923,
+      "loss": 3.8906,
+      "step": 1764
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.6953944212479741,
+      "learning_rate": 0.0009246940168776086,
+      "loss": 3.8831,
+      "step": 1765
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7875327213678065,
+      "learning_rate": 0.0009245959834075807,
+      "loss": 3.9056,
+      "step": 1766
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7431720907168309,
+      "learning_rate": 0.0009244978913726304,
+      "loss": 3.8732,
+      "step": 1767
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7919507758347588,
+      "learning_rate": 0.0009243997407862878,
+      "loss": 3.7194,
+      "step": 1768
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7550298797389087,
+      "learning_rate": 0.0009243015316620906,
+      "loss": 3.9597,
+      "step": 1769
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7688099175732305,
+      "learning_rate": 0.0009242032640135852,
+      "loss": 3.839,
+      "step": 1770
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7923743287095547,
+      "learning_rate": 0.0009241049378543254,
+      "loss": 4.0319,
+      "step": 1771
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.8016773094046883,
+      "learning_rate": 0.0009240065531978736,
+      "loss": 3.9313,
+      "step": 1772
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7245499745674537,
+      "learning_rate": 0.0009239081100578002,
+      "loss": 4.0616,
+      "step": 1773
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7349379964966979,
+      "learning_rate": 0.0009238096084476832,
+      "loss": 3.8669,
+      "step": 1774
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.8883865150224716,
+      "learning_rate": 0.0009237110483811096,
+      "loss": 3.927,
+      "step": 1775
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.6844550126018027,
+      "learning_rate": 0.0009236124298716734,
+      "loss": 3.8708,
+      "step": 1776
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.8989022935128861,
+      "learning_rate": 0.0009235137529329772,
+      "loss": 4.0256,
+      "step": 1777
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7477214716622635,
+      "learning_rate": 0.0009234150175786318,
+      "loss": 4.0495,
+      "step": 1778
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.8517445964292621,
+      "learning_rate": 0.0009233162238222556,
+      "loss": 3.8835,
+      "step": 1779
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7899104988648538,
+      "learning_rate": 0.0009232173716774757,
+      "loss": 3.8527,
+      "step": 1780
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7808154590149765,
+      "learning_rate": 0.0009231184611579265,
+      "loss": 4.074,
+      "step": 1781
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.753827120218923,
+      "learning_rate": 0.000923019492277251,
+      "loss": 3.9111,
+      "step": 1782
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7511817397495992,
+      "learning_rate": 0.0009229204650491001,
+      "loss": 3.7422,
+      "step": 1783
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7311636094782424,
+      "learning_rate": 0.0009228213794871325,
+      "loss": 3.9425,
+      "step": 1784
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.8130288451187192,
+      "learning_rate": 0.0009227222356050154,
+      "loss": 3.9246,
+      "step": 1785
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.6859527839557056,
+      "learning_rate": 0.0009226230334164236,
+      "loss": 3.9071,
+      "step": 1786
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.6554642927013951,
+      "learning_rate": 0.0009225237729350403,
+      "loss": 3.8776,
+      "step": 1787
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.6799951737970095,
+      "learning_rate": 0.0009224244541745566,
+      "loss": 3.8753,
+      "step": 1788
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.6646520844313295,
+      "learning_rate": 0.0009223250771486717,
+      "loss": 4.1116,
+      "step": 1789
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.7719803771986256,
+      "learning_rate": 0.0009222256418710923,
+      "loss": 3.8908,
+      "step": 1790
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.6406067536316352,
+      "learning_rate": 0.0009221261483555343,
+      "loss": 3.7226,
+      "step": 1791
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.6580578428348078,
+      "learning_rate": 0.0009220265966157205,
+      "loss": 3.9598,
+      "step": 1792
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.8482858376155865,
+      "learning_rate": 0.0009219269866653823,
+      "loss": 4.0739,
+      "step": 1793
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.6123555971166764,
+      "learning_rate": 0.0009218273185182588,
+      "loss": 3.7107,
+      "step": 1794
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.7444375156775016,
+      "learning_rate": 0.0009217275921880976,
+      "loss": 3.9803,
+      "step": 1795
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.7530992421856789,
+      "learning_rate": 0.000921627807688654,
+      "loss": 4.1292,
+      "step": 1796
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.8191054366305114,
+      "learning_rate": 0.0009215279650336911,
+      "loss": 4.0176,
+      "step": 1797
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.8438937131357908,
+      "learning_rate": 0.0009214280642369806,
+      "loss": 3.9014,
+      "step": 1798
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.7271331352492184,
+      "learning_rate": 0.0009213281053123018,
+      "loss": 3.7445,
+      "step": 1799
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.8140841300279414,
+      "learning_rate": 0.000921228088273442,
+      "loss": 3.8131,
+      "step": 1800
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.8031924890203358,
+      "learning_rate": 0.0009211280131341968,
+      "loss": 3.8962,
+      "step": 1801
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.8874699153381481,
+      "learning_rate": 0.0009210278799083695,
+      "loss": 4.0392,
+      "step": 1802
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.7355808677346425,
+      "learning_rate": 0.000920927688609772,
+      "loss": 3.9477,
+      "step": 1803
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.8743984453870507,
+      "learning_rate": 0.0009208274392522231,
+      "loss": 3.736,
+      "step": 1804
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.8879938107849425,
+      "learning_rate": 0.0009207271318495509,
+      "loss": 3.8183,
+      "step": 1805
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.7651803721305752,
+      "learning_rate": 0.0009206267664155906,
+      "loss": 3.8609,
+      "step": 1806
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.0002119888770291,
+      "learning_rate": 0.0009205263429641857,
+      "loss": 3.9593,
+      "step": 1807
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.755914410267963,
+      "learning_rate": 0.0009204258615091879,
+      "loss": 4.0576,
+      "step": 1808
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.707192874180788,
+      "learning_rate": 0.0009203253220644564,
+      "loss": 3.9897,
+      "step": 1809
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.902495213379042,
+      "learning_rate": 0.0009202247246438589,
+      "loss": 3.7554,
+      "step": 1810
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.8281844281527484,
+      "learning_rate": 0.0009201240692612708,
+      "loss": 4.0296,
+      "step": 1811
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.7687804425058603,
+      "learning_rate": 0.0009200233559305758,
+      "loss": 3.6559,
+      "step": 1812
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.8466753724426532,
+      "learning_rate": 0.0009199225846656649,
+      "loss": 3.9031,
+      "step": 1813
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.7624971298739351,
+      "learning_rate": 0.0009198217554804382,
+      "loss": 3.8011,
+      "step": 1814
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.014555041063583,
+      "learning_rate": 0.0009197208683888028,
+      "loss": 3.7543,
+      "step": 1815
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.9639278815623971,
+      "learning_rate": 0.0009196199234046741,
+      "loss": 3.9144,
+      "step": 1816
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.8192233948572364,
+      "learning_rate": 0.0009195189205419757,
+      "loss": 3.635,
+      "step": 1817
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.672990505689812,
+      "learning_rate": 0.000919417859814639,
+      "loss": 4.1342,
+      "step": 1818
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.1506955777088501,
+      "learning_rate": 0.0009193167412366034,
+      "loss": 4.0161,
+      "step": 1819
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.8265074984790751,
+      "learning_rate": 0.0009192155648218162,
+      "loss": 3.742,
+      "step": 1820
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.7634441509013512,
+      "learning_rate": 0.0009191143305842329,
+      "loss": 3.7862,
+      "step": 1821
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.8291031773683417,
+      "learning_rate": 0.0009190130385378166,
+      "loss": 3.816,
+      "step": 1822
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.8230051077879045,
+      "learning_rate": 0.0009189116886965388,
+      "loss": 3.7654,
+      "step": 1823
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.7790386666004395,
+      "learning_rate": 0.0009188102810743788,
+      "loss": 3.9439,
+      "step": 1824
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.9363530618420868,
+      "learning_rate": 0.0009187088156853236,
+      "loss": 3.7313,
+      "step": 1825
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.7081668138081271,
+      "learning_rate": 0.0009186072925433689,
+      "loss": 3.955,
+      "step": 1826
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.7781707783032774,
+      "learning_rate": 0.0009185057116625172,
+      "loss": 3.8381,
+      "step": 1827
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.0191013808768887,
+      "learning_rate": 0.0009184040730567803,
+      "loss": 3.9728,
+      "step": 1828
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.7789452398353326,
+      "learning_rate": 0.0009183023767401769,
+      "loss": 3.7614,
+      "step": 1829
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.737485223435407,
+      "learning_rate": 0.0009182006227267343,
+      "loss": 3.7757,
+      "step": 1830
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.2679289133418514,
+      "learning_rate": 0.0009180988110304873,
+      "loss": 3.8905,
+      "step": 1831
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.8429420251021547,
+      "learning_rate": 0.000917996941665479,
+      "loss": 3.8348,
+      "step": 1832
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.8267997301456189,
+      "learning_rate": 0.0009178950146457606,
+      "loss": 3.9278,
+      "step": 1833
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.7778305424837477,
+      "learning_rate": 0.0009177930299853903,
+      "loss": 3.7844,
+      "step": 1834
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.8438975589576946,
+      "learning_rate": 0.0009176909876984356,
+      "loss": 3.8029,
+      "step": 1835
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.7887038280353068,
+      "learning_rate": 0.0009175888877989712,
+      "loss": 3.9373,
+      "step": 1836
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.7917287022730356,
+      "learning_rate": 0.0009174867303010795,
+      "loss": 3.8645,
+      "step": 1837
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.7858372896969633,
+      "learning_rate": 0.0009173845152188516,
+      "loss": 3.7449,
+      "step": 1838
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.919925507440622,
+      "learning_rate": 0.0009172822425663855,
+      "loss": 3.9729,
+      "step": 1839
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.7658943378631716,
+      "learning_rate": 0.0009171799123577886,
+      "loss": 3.9106,
+      "step": 1840
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.9375314599172345,
+      "learning_rate": 0.0009170775246071747,
+      "loss": 3.7064,
+      "step": 1841
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.7705771052998489,
+      "learning_rate": 0.0009169750793286667,
+      "loss": 4.1107,
+      "step": 1842
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.8236536063075164,
+      "learning_rate": 0.0009168725765363946,
+      "loss": 3.8327,
+      "step": 1843
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.7604607176306432,
+      "learning_rate": 0.0009167700162444969,
+      "loss": 4.1545,
+      "step": 1844
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.8230602815712921,
+      "learning_rate": 0.0009166673984671198,
+      "loss": 4.0707,
+      "step": 1845
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.8425585918564236,
+      "learning_rate": 0.0009165647232184176,
+      "loss": 3.6911,
+      "step": 1846
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.9760367661105567,
+      "learning_rate": 0.0009164619905125522,
+      "loss": 4.1045,
+      "step": 1847
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.3801627427716208,
+      "learning_rate": 0.0009163592003636936,
+      "loss": 4.0525,
+      "step": 1848
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.7512200452746538,
+      "learning_rate": 0.00091625635278602,
+      "loss": 3.7542,
+      "step": 1849
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.7317553373794693,
+      "learning_rate": 0.000916153447793717,
+      "loss": 3.9897,
+      "step": 1850
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.6494220629829268,
+      "learning_rate": 0.0009160504854009786,
+      "loss": 3.8179,
+      "step": 1851
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.8998951049503917,
+      "learning_rate": 0.0009159474656220063,
+      "loss": 3.7069,
+      "step": 1852
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.6875317935421839,
+      "learning_rate": 0.0009158443884710097,
+      "loss": 3.8136,
+      "step": 1853
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.7862866529427058,
+      "learning_rate": 0.0009157412539622065,
+      "loss": 3.9661,
+      "step": 1854
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.7279264963532441,
+      "learning_rate": 0.0009156380621098221,
+      "loss": 3.823,
+      "step": 1855
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.8142735591548157,
+      "learning_rate": 0.0009155348129280898,
+      "loss": 3.9211,
+      "step": 1856
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.5653489373235878,
+      "learning_rate": 0.000915431506431251,
+      "loss": 3.9058,
+      "step": 1857
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.7764229853632041,
+      "learning_rate": 0.0009153281426335547,
+      "loss": 3.8099,
+      "step": 1858
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.8359349117264595,
+      "learning_rate": 0.0009152247215492577,
+      "loss": 3.635,
+      "step": 1859
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.8042832959697725,
+      "learning_rate": 0.0009151212431926256,
+      "loss": 3.7794,
+      "step": 1860
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.782330829803909,
+      "learning_rate": 0.0009150177075779308,
+      "loss": 3.875,
+      "step": 1861
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.7089863621982626,
+      "learning_rate": 0.0009149141147194542,
+      "loss": 3.7641,
+      "step": 1862
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.8236834522705743,
+      "learning_rate": 0.0009148104646314844,
+      "loss": 4.1159,
+      "step": 1863
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.8510699317920725,
+      "learning_rate": 0.000914706757328318,
+      "loss": 4.0558,
+      "step": 1864
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.7765692534307267,
+      "learning_rate": 0.0009146029928242596,
+      "loss": 3.7697,
+      "step": 1865
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.8373970432684154,
+      "learning_rate": 0.0009144991711336214,
+      "loss": 4.0329,
+      "step": 1866
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.7889179335701646,
+      "learning_rate": 0.0009143952922707235,
+      "loss": 4.0346,
+      "step": 1867
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.745680691083741,
+      "learning_rate": 0.0009142913562498942,
+      "loss": 3.9779,
+      "step": 1868
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.715853609255922,
+      "learning_rate": 0.0009141873630854694,
+      "loss": 3.6749,
+      "step": 1869
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.8374924245709989,
+      "learning_rate": 0.0009140833127917929,
+      "loss": 3.9144,
+      "step": 1870
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.8150094207887062,
+      "learning_rate": 0.0009139792053832166,
+      "loss": 3.7356,
+      "step": 1871
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.8486350829809888,
+      "learning_rate": 0.0009138750408741001,
+      "loss": 3.776,
+      "step": 1872
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.7168042142083179,
+      "learning_rate": 0.000913770819278811,
+      "loss": 4.0069,
+      "step": 1873
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.737257814021183,
+      "learning_rate": 0.0009136665406117244,
+      "loss": 3.9646,
+      "step": 1874
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.8023341444635669,
+      "learning_rate": 0.0009135622048872238,
+      "loss": 4.1729,
+      "step": 1875
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.787586192629909,
+      "learning_rate": 0.0009134578121197002,
+      "loss": 3.9172,
+      "step": 1876
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.8086846678599074,
+      "learning_rate": 0.0009133533623235526,
+      "loss": 3.8854,
+      "step": 1877
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.7654224177481054,
+      "learning_rate": 0.000913248855513188,
+      "loss": 3.8118,
+      "step": 1878
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.6966105734241713,
+      "learning_rate": 0.0009131442917030211,
+      "loss": 3.8454,
+      "step": 1879
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.7511299460136852,
+      "learning_rate": 0.0009130396709074741,
+      "loss": 3.7579,
+      "step": 1880
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.7658076535013472,
+      "learning_rate": 0.0009129349931409781,
+      "loss": 3.8432,
+      "step": 1881
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.845521630657328,
+      "learning_rate": 0.0009128302584179708,
+      "loss": 3.9815,
+      "step": 1882
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.8057288971208282,
+      "learning_rate": 0.0009127254667528988,
+      "loss": 3.7148,
+      "step": 1883
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.8479880670693027,
+      "learning_rate": 0.0009126206181602158,
+      "loss": 3.637,
+      "step": 1884
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.8928818442290861,
+      "learning_rate": 0.0009125157126543838,
+      "loss": 4.0144,
+      "step": 1885
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.8250737958305291,
+      "learning_rate": 0.0009124107502498725,
+      "loss": 3.8882,
+      "step": 1886
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.9283298052785008,
+      "learning_rate": 0.0009123057309611595,
+      "loss": 3.8937,
+      "step": 1887
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.9361109070855631,
+      "learning_rate": 0.0009122006548027302,
+      "loss": 3.9103,
+      "step": 1888
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.7157477299426674,
+      "learning_rate": 0.0009120955217890778,
+      "loss": 3.8233,
+      "step": 1889
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.7367734720067374,
+      "learning_rate": 0.0009119903319347034,
+      "loss": 3.8003,
+      "step": 1890
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.803302277594201,
+      "learning_rate": 0.000911885085254116,
+      "loss": 4.0575,
+      "step": 1891
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.8384415192309669,
+      "learning_rate": 0.0009117797817618323,
+      "loss": 3.6837,
+      "step": 1892
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.7765752248559238,
+      "learning_rate": 0.000911674421472377,
+      "loss": 3.7868,
+      "step": 1893
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.783679942864075,
+      "learning_rate": 0.0009115690044002824,
+      "loss": 4.0815,
+      "step": 1894
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.9174847503966219,
+      "learning_rate": 0.0009114635305600889,
+      "loss": 3.7492,
+      "step": 1895
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.8343904465590879,
+      "learning_rate": 0.0009113579999663447,
+      "loss": 3.8396,
+      "step": 1896
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.6606211945574318,
+      "learning_rate": 0.0009112524126336054,
+      "loss": 3.7131,
+      "step": 1897
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.7892498117175678,
+      "learning_rate": 0.0009111467685764351,
+      "loss": 3.6507,
+      "step": 1898
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.6738211710550357,
+      "learning_rate": 0.0009110410678094051,
+      "loss": 3.7384,
+      "step": 1899
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.734039570032989,
+      "learning_rate": 0.0009109353103470951,
+      "loss": 3.9633,
+      "step": 1900
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.7367034375505778,
+      "learning_rate": 0.0009108294962040921,
+      "loss": 3.9019,
+      "step": 1901
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.8155576192936432,
+      "learning_rate": 0.0009107236253949912,
+      "loss": 3.7207,
+      "step": 1902
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.8191568115814969,
+      "learning_rate": 0.0009106176979343955,
+      "loss": 3.7114,
+      "step": 1903
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.7659376802400928,
+      "learning_rate": 0.0009105117138369151,
+      "loss": 3.9048,
+      "step": 1904
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.8645358968946236,
+      "learning_rate": 0.0009104056731171691,
+      "loss": 3.764,
+      "step": 1905
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.743512245489565,
+      "learning_rate": 0.0009102995757897834,
+      "loss": 4.0159,
+      "step": 1906
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.9874288868372059,
+      "learning_rate": 0.0009101934218693923,
+      "loss": 3.8555,
+      "step": 1907
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.7417169356225424,
+      "learning_rate": 0.0009100872113706375,
+      "loss": 3.757,
+      "step": 1908
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.8688799585931287,
+      "learning_rate": 0.0009099809443081691,
+      "loss": 3.9607,
+      "step": 1909
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.7855815392661997,
+      "learning_rate": 0.0009098746206966443,
+      "loss": 3.8604,
+      "step": 1910
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.779294703214083,
+      "learning_rate": 0.0009097682405507285,
+      "loss": 3.836,
+      "step": 1911
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.802981965605641,
+      "learning_rate": 0.0009096618038850948,
+      "loss": 3.7794,
+      "step": 1912
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.7755808893316486,
+      "learning_rate": 0.0009095553107144241,
+      "loss": 3.9996,
+      "step": 1913
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.7509275435967226,
+      "learning_rate": 0.0009094487610534052,
+      "loss": 3.8933,
+      "step": 1914
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.9782908684647244,
+      "learning_rate": 0.0009093421549167343,
+      "loss": 3.9325,
+      "step": 1915
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.785092811355111,
+      "learning_rate": 0.0009092354923191161,
+      "loss": 3.9434,
+      "step": 1916
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.7332001658406608,
+      "learning_rate": 0.0009091287732752624,
+      "loss": 3.7755,
+      "step": 1917
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.8036973770978606,
+      "learning_rate": 0.0009090219977998933,
+      "loss": 3.9152,
+      "step": 1918
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.6927973935071611,
+      "learning_rate": 0.000908915165907736,
+      "loss": 3.831,
+      "step": 1919
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.8270807484327787,
+      "learning_rate": 0.0009088082776135263,
+      "loss": 3.9398,
+      "step": 1920
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.7404107686238394,
+      "learning_rate": 0.0009087013329320073,
+      "loss": 3.9327,
+      "step": 1921
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.7606837600028491,
+      "learning_rate": 0.0009085943318779301,
+      "loss": 3.929,
+      "step": 1922
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.8069176759296681,
+      "learning_rate": 0.0009084872744660532,
+      "loss": 3.9459,
+      "step": 1923
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.1093902626127157,
+      "learning_rate": 0.0009083801607111433,
+      "loss": 3.5617,
+      "step": 1924
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.791581672317756,
+      "learning_rate": 0.0009082729906279746,
+      "loss": 3.9925,
+      "step": 1925
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.7517541397884347,
+      "learning_rate": 0.0009081657642313292,
+      "loss": 3.7571,
+      "step": 1926
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.7825258256250016,
+      "learning_rate": 0.0009080584815359972,
+      "loss": 3.9193,
+      "step": 1927
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.1595234216895824,
+      "learning_rate": 0.0009079511425567759,
+      "loss": 3.892,
+      "step": 1928
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.7670731979033678,
+      "learning_rate": 0.0009078437473084706,
+      "loss": 4.027,
+      "step": 1929
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.8607211981699647,
+      "learning_rate": 0.0009077362958058946,
+      "loss": 3.7733,
+      "step": 1930
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.8049263088180011,
+      "learning_rate": 0.0009076287880638689,
+      "loss": 3.6674,
+      "step": 1931
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.691509536228359,
+      "learning_rate": 0.0009075212240972218,
+      "loss": 3.8831,
+      "step": 1932
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.8216622557736523,
+      "learning_rate": 0.00090741360392079,
+      "loss": 4.0133,
+      "step": 1933
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.8288696875682553,
+      "learning_rate": 0.0009073059275494176,
+      "loss": 3.9177,
+      "step": 1934
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.9203377439830267,
+      "learning_rate": 0.0009071981949979564,
+      "loss": 3.8384,
+      "step": 1935
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.7781585509007852,
+      "learning_rate": 0.000907090406281266,
+      "loss": 3.9732,
+      "step": 1936
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.8280677722696165,
+      "learning_rate": 0.000906982561414214,
+      "loss": 3.634,
+      "step": 1937
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.7518900611477576,
+      "learning_rate": 0.0009068746604116755,
+      "loss": 3.8711,
+      "step": 1938
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.8106404644925548,
+      "learning_rate": 0.0009067667032885334,
+      "loss": 3.8588,
+      "step": 1939
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.6361541551754019,
+      "learning_rate": 0.0009066586900596781,
+      "loss": 3.8098,
+      "step": 1940
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.7432764438217282,
+      "learning_rate": 0.000906550620740008,
+      "loss": 3.8003,
+      "step": 1941
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.732384818892767,
+      "learning_rate": 0.0009064424953444296,
+      "loss": 3.7279,
+      "step": 1942
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.9980519538446884,
+      "learning_rate": 0.0009063343138878563,
+      "loss": 4.0684,
+      "step": 1943
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.632460493196114,
+      "learning_rate": 0.0009062260763852099,
+      "loss": 3.6478,
+      "step": 1944
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.7642968508702586,
+      "learning_rate": 0.0009061177828514198,
+      "loss": 3.7716,
+      "step": 1945
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.7378431925689061,
+      "learning_rate": 0.0009060094333014226,
+      "loss": 3.8119,
+      "step": 1946
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.8674244382184695,
+      "learning_rate": 0.0009059010277501634,
+      "loss": 3.8387,
+      "step": 1947
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.7395142721521389,
+      "learning_rate": 0.0009057925662125946,
+      "loss": 3.7808,
+      "step": 1948
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.6009771416139514,
+      "learning_rate": 0.0009056840487036764,
+      "loss": 3.7449,
+      "step": 1949
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.203273373696835,
+      "learning_rate": 0.0009055754752383768,
+      "loss": 3.8459,
+      "step": 1950
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.8843068352665313,
+      "learning_rate": 0.0009054668458316713,
+      "loss": 3.7464,
+      "step": 1951
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.800926332755142,
+      "learning_rate": 0.0009053581604985433,
+      "loss": 3.918,
+      "step": 1952
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.7685167081390272,
+      "learning_rate": 0.000905249419253984,
+      "loss": 3.7961,
+      "step": 1953
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.9226753810285627,
+      "learning_rate": 0.0009051406221129919,
+      "loss": 3.777,
+      "step": 1954
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.8547811644391046,
+      "learning_rate": 0.0009050317690905737,
+      "loss": 3.7778,
+      "step": 1955
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.2227429562391292,
+      "learning_rate": 0.0009049228602017437,
+      "loss": 3.6934,
+      "step": 1956
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.7837950862346422,
+      "learning_rate": 0.0009048138954615235,
+      "loss": 3.8584,
+      "step": 1957
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.7528380751923922,
+      "learning_rate": 0.0009047048748849429,
+      "loss": 4.0075,
+      "step": 1958
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.8334144698450073,
+      "learning_rate": 0.0009045957984870393,
+      "loss": 3.8519,
+      "step": 1959
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.0811523992115364,
+      "learning_rate": 0.0009044866662828575,
+      "loss": 3.6694,
+      "step": 1960
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.8495816965140921,
+      "learning_rate": 0.0009043774782874503,
+      "loss": 3.9996,
+      "step": 1961
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.7749261423145792,
+      "learning_rate": 0.0009042682345158781,
+      "loss": 3.9862,
+      "step": 1962
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.9227475897478702,
+      "learning_rate": 0.0009041589349832091,
+      "loss": 3.9152,
+      "step": 1963
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.8475081027530467,
+      "learning_rate": 0.000904049579704519,
+      "loss": 3.7206,
+      "step": 1964
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.9566101619415467,
+      "learning_rate": 0.0009039401686948912,
+      "loss": 4.044,
+      "step": 1965
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.8892129339208293,
+      "learning_rate": 0.0009038307019694169,
+      "loss": 3.7331,
+      "step": 1966
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.7641767318056124,
+      "learning_rate": 0.000903721179543195,
+      "loss": 3.7638,
+      "step": 1967
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.2407977318541443,
+      "learning_rate": 0.0009036116014313321,
+      "loss": 3.8328,
+      "step": 1968
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.9155852515990093,
+      "learning_rate": 0.0009035019676489422,
+      "loss": 3.7025,
+      "step": 1969
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.7659430900868343,
+      "learning_rate": 0.0009033922782111473,
+      "loss": 3.7129,
+      "step": 1970
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.9472145739973923,
+      "learning_rate": 0.000903282533133077,
+      "loss": 3.9758,
+      "step": 1971
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.7617990564909911,
+      "learning_rate": 0.0009031727324298686,
+      "loss": 3.6597,
+      "step": 1972
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.7335432871610481,
+      "learning_rate": 0.0009030628761166668,
+      "loss": 3.845,
+      "step": 1973
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.8064227453612388,
+      "learning_rate": 0.0009029529642086245,
+      "loss": 3.9377,
+      "step": 1974
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.7066322481379579,
+      "learning_rate": 0.0009028429967209015,
+      "loss": 3.9424,
+      "step": 1975
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.9572870170320034,
+      "learning_rate": 0.0009027329736686663,
+      "loss": 3.8628,
+      "step": 1976
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.8082987502478605,
+      "learning_rate": 0.000902622895067094,
+      "loss": 3.7577,
+      "step": 1977
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.7520252941106546,
+      "learning_rate": 0.000902512760931368,
+      "loss": 3.8899,
+      "step": 1978
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.8268251566722512,
+      "learning_rate": 0.0009024025712766792,
+      "loss": 3.9555,
+      "step": 1979
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.2371682447529315,
+      "learning_rate": 0.0009022923261182264,
+      "loss": 3.9017,
+      "step": 1980
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.525743532097485,
+      "learning_rate": 0.0009021820254712153,
+      "loss": 3.9388,
+      "step": 1981
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.8420085655139568,
+      "learning_rate": 0.0009020716693508602,
+      "loss": 3.889,
+      "step": 1982
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.7732562003767723,
+      "learning_rate": 0.0009019612577723826,
+      "loss": 3.643,
+      "step": 1983
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.3540444843049353,
+      "learning_rate": 0.0009018507907510114,
+      "loss": 3.7167,
+      "step": 1984
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.8681951724928355,
+      "learning_rate": 0.0009017402683019838,
+      "loss": 4.0191,
+      "step": 1985
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.7473612536305064,
+      "learning_rate": 0.0009016296904405439,
+      "loss": 3.6434,
+      "step": 1986
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.8371563551217536,
+      "learning_rate": 0.0009015190571819438,
+      "loss": 3.8937,
+      "step": 1987
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.8281837540136497,
+      "learning_rate": 0.0009014083685414437,
+      "loss": 3.9874,
+      "step": 1988
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.7216865191352599,
+      "learning_rate": 0.0009012976245343106,
+      "loss": 3.8798,
+      "step": 1989
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.7310562616815615,
+      "learning_rate": 0.0009011868251758195,
+      "loss": 4.0126,
+      "step": 1990
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.7863301358877134,
+      "learning_rate": 0.0009010759704812533,
+      "loss": 3.994,
+      "step": 1991
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.9997000058176198,
+      "learning_rate": 0.0009009650604659023,
+      "loss": 3.7062,
+      "step": 1992
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.5939330143181982,
+      "learning_rate": 0.0009008540951450641,
+      "loss": 3.8502,
+      "step": 1993
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.9004997480496979,
+      "learning_rate": 0.0009007430745340446,
+      "loss": 3.938,
+      "step": 1994
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.7469871916243367,
+      "learning_rate": 0.0009006319986481567,
+      "loss": 4.0888,
+      "step": 1995
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.7956750844256011,
+      "learning_rate": 0.0009005208675027215,
+      "loss": 3.7604,
+      "step": 1996
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.9616605125875768,
+      "learning_rate": 0.000900409681113067,
+      "loss": 3.9209,
+      "step": 1997
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.9479515174477374,
+      "learning_rate": 0.0009002984394945298,
+      "loss": 3.9356,
+      "step": 1998
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.7796120932780577,
+      "learning_rate": 0.0009001871426624528,
+      "loss": 3.9046,
+      "step": 1999
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.7378291584580816,
+      "learning_rate": 0.0009000757906321882,
+      "loss": 3.6397,
+      "step": 2000
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 8721,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 1000,
+  "total_flos": 6542275092480.0,
+  "train_batch_size": 32,
+  "trial_name": null,
+  "trial_params": null
+}